diff --git a/runs/t0/t0_S3_lr0.0003/checkpoint-864/chat_template.jinja b/runs/t0/t0_S3_lr0.0003/checkpoint-864/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..699ff8df401fe4788525e9c1f9b86a99eadd6230
--- /dev/null
+++ b/runs/t0/t0_S3_lr0.0003/checkpoint-864/chat_template.jinja
@@ -0,0 +1,85 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0].role == 'system' %}
+        {{- messages[0].content + '\n\n' }}
+    {%- endif %}
+    {{- "# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0].content + '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
+{%- for message in messages[::-1] %}
+    {%- set index = (messages|length - 1) - loop.index0 %}
+    {%- if ns.multi_step_tool and message.role == "user" and not(message.content.startswith('<tool_response>') and message.content.endswith('</tool_response>')) %}
+        {%- set ns.multi_step_tool = false %}
+        {%- set ns.last_query_index = index %}
+    {%- endif %}
+{%- endfor %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {%- set content = message.content %}
+        {%- set reasoning_content = '' %}
+        {%- if message.reasoning_content is defined and message.reasoning_content is not none %}
+            {%- set reasoning_content = message.reasoning_content %}
+        {%- else %}
+            {%- if '</think>' in message.content %}
+                {%- set content = message.content.split('</think>')[-1].lstrip('\n') %}
+                {%- set reasoning_content = message.content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
+            {%- endif %}
+        {%- endif %}
+        {%- if loop.index0 > ns.last_query_index %}
+            {%- if loop.last or (not loop.last and reasoning_content) %}
+                {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content.strip('\n') + '\n</think>\n\n' + content.lstrip('\n') }}
+            {%- else %}
+                {{- '<|im_start|>' + message.role + '\n' + content }}
+            {%- endif %}
+        {%- else %}
+            {{- '<|im_start|>' + message.role + '\n' + content }}
+        {%- endif %}
+        {%- if message.tool_calls %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if (loop.first and content) or (not loop.first) %}
+                    {{- '\n' }}
+                {%- endif %}
+                {%- if tool_call.function %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {{- '<tool_call>\n{"name": "' }}
+                {{- tool_call.name }}
+                {{- '", "arguments": ' }}
+                {%- if tool_call.arguments is string %}
+                    {{- tool_call.arguments }}
+                {%- else %}
+                    {{- tool_call.arguments | tojson }}
+                {%- endif %}
+                {{- '}\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.first or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+    {%- if enable_thinking is defined and enable_thinking is false %}
+        {{- '<think>\n\n</think>\n\n' }}
+    {%- endif %}
+{%- endif %}
\ No newline at end of file
diff --git a/runs/t0/t0_S3_lr0.0003/checkpoint-864/config.json b/runs/t0/t0_S3_lr0.0003/checkpoint-864/config.json
new file mode 100644
index 0000000000000000000000000000000000000000..edf37b33d02b391ecdf2ca887c98eec60d9c7686
--- /dev/null
+++ b/runs/t0/t0_S3_lr0.0003/checkpoint-864/config.json
@@ -0,0 +1,32 @@
+{
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": null,
+  "dtype": "float32",
+  "eos_token_id": 151645,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 768,
+  "initializer_range": 0.02,
+  "intermediate_size": 2048,
+  "max_position_embeddings": 2048,
+  "mlp_bias": false,
+  "model_type": "llama",
+  "num_attention_heads": 6,
+  "num_hidden_layers": 20,
+  "num_key_value_heads": 6,
+  "pad_token_id": 151645,
+  "pretraining_tp": 1,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 10000.0,
+    "rope_type": "default"
+  },
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": false,
+  "vocab_size": 151671
+}
diff --git a/runs/t0/t0_S3_lr0.0003/checkpoint-864/generation_config.json b/runs/t0/t0_S3_lr0.0003/checkpoint-864/generation_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..8c4242f32d40df256857f87264f646130f684e33
--- /dev/null
+++ b/runs/t0/t0_S3_lr0.0003/checkpoint-864/generation_config.json
@@ -0,0 +1,11 @@
+{
+  "_from_model_config": true,
+  "eos_token_id": [
+    151645
+  ],
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "transformers_version": "5.15.0",
+  "use_cache": true
+}
diff --git a/runs/t0/t0_S3_lr0.0003/checkpoint-864/tokenizer_config.json b/runs/t0/t0_S3_lr0.0003/checkpoint-864/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..839a3743177a3ed77e1ffbae67520e8500ca0532
--- /dev/null
+++ b/runs/t0/t0_S3_lr0.0003/checkpoint-864/tokenizer_config.json
@@ -0,0 +1,19 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|l2r_pred|>",
+    "<|r2l_pred|>"
+  ],
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 131072,
+  "pad_token": "<|im_end|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}
diff --git a/runs/t0/t0_S3_lr0.0003/checkpoint-864/trainer_state.json b/runs/t0/t0_S3_lr0.0003/checkpoint-864/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..52fee71673b1634c5c0463fa0d4057355ef825a1
--- /dev/null
+++ b/runs/t0/t0_S3_lr0.0003/checkpoint-864/trainer_state.json
@@ -0,0 +1,739 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 12.0,
+  "eval_steps": 72,
+  "global_step": 864,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.013961605584642234,
+      "grad_norm": 2.5511720180511475,
+      "learning_rate": 0.0,
+      "loss": 12.086503982543945,
+      "step": 1
+    },
+    {
+      "epoch": 0.13961605584642234,
+      "grad_norm": 2.48952317237854,
+      "learning_rate": 2.6999999999999996e-05,
+      "loss": 11.761650933159721,
+      "step": 10
+    },
+    {
+      "epoch": 0.2792321116928447,
+      "grad_norm": 1.994715690612793,
+      "learning_rate": 5.6999999999999996e-05,
+      "loss": 10.831330871582031,
+      "step": 20
+    },
+    {
+      "epoch": 0.418848167539267,
+      "grad_norm": 1.869005560874939,
+      "learning_rate": 8.699999999999999e-05,
+      "loss": 10.210093688964843,
+      "step": 30
+    },
+    {
+      "epoch": 0.5584642233856894,
+      "grad_norm": 1.5961251258850098,
+      "learning_rate": 0.000117,
+      "loss": 9.370370483398437,
+      "step": 40
+    },
+    {
+      "epoch": 0.6980802792321117,
+      "grad_norm": 1.033909797668457,
+      "learning_rate": 0.000147,
+      "loss": 8.43478012084961,
+      "step": 50
+    },
+    {
+      "epoch": 0.837696335078534,
+      "grad_norm": 0.7044252157211304,
+      "learning_rate": 0.00017699999999999997,
+      "loss": 7.778045654296875,
+      "step": 60
+    },
+    {
+      "epoch": 0.9773123909249564,
+      "grad_norm": 0.4867978096008301,
+      "learning_rate": 0.00020699999999999996,
+      "loss": 7.52720947265625,
+      "step": 70
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 7.445277214050293,
+      "eval_runtime": 28.509,
+      "eval_samples_per_second": 85.622,
+      "eval_steps_per_second": 1.368,
+      "step": 72
+    },
+    {
+      "epoch": 1.111692844677138,
+      "grad_norm": 0.6972123384475708,
+      "learning_rate": 0.000237,
+      "loss": 7.379924011230469,
+      "step": 80
+    },
+    {
+      "epoch": 1.2513089005235603,
+      "grad_norm": 0.6060797572135925,
+      "learning_rate": 0.000267,
+      "loss": 7.171945190429687,
+      "step": 90
+    },
+    {
+      "epoch": 1.3909249563699826,
+      "grad_norm": 0.8712506890296936,
+      "learning_rate": 0.00029699999999999996,
+      "loss": 7.012107849121094,
+      "step": 100
+    },
+    {
+      "epoch": 1.530541012216405,
+      "grad_norm": 0.5367537140846252,
+      "learning_rate": 0.0003,
+      "loss": 6.8590141296386715,
+      "step": 110
+    },
+    {
+      "epoch": 1.6701570680628273,
+      "grad_norm": 0.5023939609527588,
+      "learning_rate": 0.0003,
+      "loss": 6.714488220214844,
+      "step": 120
+    },
+    {
+      "epoch": 1.8097731239092496,
+      "grad_norm": 0.7802119255065918,
+      "learning_rate": 0.0003,
+      "loss": 6.580387878417969,
+      "step": 130
+    },
+    {
+      "epoch": 1.949389179755672,
+      "grad_norm": 1.0640617609024048,
+      "learning_rate": 0.0003,
+      "loss": 6.485602569580078,
+      "step": 140
+    },
+    {
+      "epoch": 2.0,
+      "eval_loss": 6.399908542633057,
+      "eval_runtime": 4.1035,
+      "eval_samples_per_second": 594.865,
+      "eval_steps_per_second": 9.504,
+      "step": 144
+    },
+    {
+      "epoch": 2.0837696335078535,
+      "grad_norm": 0.6536300778388977,
+      "learning_rate": 0.0003,
+      "loss": 6.3652080535888675,
+      "step": 150
+    },
+    {
+      "epoch": 2.223385689354276,
+      "grad_norm": 0.861472487449646,
+      "learning_rate": 0.0003,
+      "loss": 6.283423233032226,
+      "step": 160
+    },
+    {
+      "epoch": 2.363001745200698,
+      "grad_norm": 0.48216626048088074,
+      "learning_rate": 0.0003,
+      "loss": 6.193089294433594,
+      "step": 170
+    },
+    {
+      "epoch": 2.5026178010471205,
+      "grad_norm": 1.4223753213882446,
+      "learning_rate": 0.0003,
+      "loss": 6.143649291992188,
+      "step": 180
+    },
+    {
+      "epoch": 2.642233856893543,
+      "grad_norm": 0.5395786762237549,
+      "learning_rate": 0.0003,
+      "loss": 6.06384506225586,
+      "step": 190
+    },
+    {
+      "epoch": 2.781849912739965,
+      "grad_norm": 1.3055977821350098,
+      "learning_rate": 0.0003,
+      "loss": 5.971584320068359,
+      "step": 200
+    },
+    {
+      "epoch": 2.9214659685863875,
+      "grad_norm": 0.7041425704956055,
+      "learning_rate": 0.0003,
+      "loss": 5.910428619384765,
+      "step": 210
+    },
+    {
+      "epoch": 3.0,
+      "eval_loss": 5.867927551269531,
+      "eval_runtime": 4.1161,
+      "eval_samples_per_second": 593.031,
+      "eval_steps_per_second": 9.475,
+      "step": 216
+    },
+    {
+      "epoch": 3.055846422338569,
+      "grad_norm": 1.0085630416870117,
+      "learning_rate": 0.0003,
+      "loss": 5.828436279296875,
+      "step": 220
+    },
+    {
+      "epoch": 3.195462478184991,
+      "grad_norm": 0.7435064315795898,
+      "learning_rate": 0.0003,
+      "loss": 5.782387161254883,
+      "step": 230
+    },
+    {
+      "epoch": 3.3350785340314135,
+      "grad_norm": 0.4216879606246948,
+      "learning_rate": 0.0003,
+      "loss": 5.703854751586914,
+      "step": 240
+    },
+    {
+      "epoch": 3.474694589877836,
+      "grad_norm": 0.734950065612793,
+      "learning_rate": 0.0003,
+      "loss": 5.653765487670898,
+      "step": 250
+    },
+    {
+      "epoch": 3.614310645724258,
+      "grad_norm": 0.7231405973434448,
+      "learning_rate": 0.0003,
+      "loss": 5.640114593505859,
+      "step": 260
+    },
+    {
+      "epoch": 3.7539267015706805,
+      "grad_norm": 0.43395647406578064,
+      "learning_rate": 0.0003,
+      "loss": 5.542626953125,
+      "step": 270
+    },
+    {
+      "epoch": 3.893542757417103,
+      "grad_norm": 0.6494223475456238,
+      "learning_rate": 0.0003,
+      "loss": 5.490487670898437,
+      "step": 280
+    },
+    {
+      "epoch": 4.0,
+      "eval_loss": 5.469155788421631,
+      "eval_runtime": 4.1041,
+      "eval_samples_per_second": 594.77,
+      "eval_steps_per_second": 9.503,
+      "step": 288
+    },
+    {
+      "epoch": 4.027923211169284,
+      "grad_norm": 0.40678805112838745,
+      "learning_rate": 0.0003,
+      "loss": 5.421612548828125,
+      "step": 290
+    },
+    {
+      "epoch": 4.167539267015707,
+      "grad_norm": 0.948244035243988,
+      "learning_rate": 0.0003,
+      "loss": 5.366499328613282,
+      "step": 300
+    },
+    {
+      "epoch": 4.307155322862129,
+      "grad_norm": 0.7340747714042664,
+      "learning_rate": 0.0003,
+      "loss": 5.337801742553711,
+      "step": 310
+    },
+    {
+      "epoch": 4.446771378708552,
+      "grad_norm": 0.5626649856567383,
+      "learning_rate": 0.0003,
+      "loss": 5.294442749023437,
+      "step": 320
+    },
+    {
+      "epoch": 4.5863874345549736,
+      "grad_norm": 0.6366444230079651,
+      "learning_rate": 0.0003,
+      "loss": 5.244319915771484,
+      "step": 330
+    },
+    {
+      "epoch": 4.726003490401396,
+      "grad_norm": 0.48262909054756165,
+      "learning_rate": 0.0003,
+      "loss": 5.197024154663086,
+      "step": 340
+    },
+    {
+      "epoch": 4.865619546247818,
+      "grad_norm": 0.5314050912857056,
+      "learning_rate": 0.0003,
+      "loss": 5.135079193115234,
+      "step": 350
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 1.139448881149292,
+      "learning_rate": 0.0003,
+      "loss": 5.122449493408203,
+      "step": 360
+    },
+    {
+      "epoch": 5.0,
+      "eval_loss": 5.215811252593994,
+      "eval_runtime": 4.0976,
+      "eval_samples_per_second": 595.708,
+      "eval_steps_per_second": 9.518,
+      "step": 360
+    },
+    {
+      "epoch": 5.139616055846423,
+      "grad_norm": 1.0207905769348145,
+      "learning_rate": 0.0003,
+      "loss": 5.087126922607422,
+      "step": 370
+    },
+    {
+      "epoch": 5.279232111692845,
+      "grad_norm": 0.45449236035346985,
+      "learning_rate": 0.0003,
+      "loss": 5.048321914672852,
+      "step": 380
+    },
+    {
+      "epoch": 5.418848167539267,
+      "grad_norm": 0.411428838968277,
+      "learning_rate": 0.0003,
+      "loss": 4.9702400207519535,
+      "step": 390
+    },
+    {
+      "epoch": 5.558464223385689,
+      "grad_norm": 0.77912837266922,
+      "learning_rate": 0.0003,
+      "loss": 4.928448486328125,
+      "step": 400
+    },
+    {
+      "epoch": 5.698080279232112,
+      "grad_norm": 0.6806897521018982,
+      "learning_rate": 0.0003,
+      "loss": 4.942894744873047,
+      "step": 410
+    },
+    {
+      "epoch": 5.837696335078534,
+      "grad_norm": 0.6517318487167358,
+      "learning_rate": 0.0003,
+      "loss": 4.901599884033203,
+      "step": 420
+    },
+    {
+      "epoch": 5.977312390924956,
+      "grad_norm": 0.33704403042793274,
+      "learning_rate": 0.0003,
+      "loss": 4.843351364135742,
+      "step": 430
+    },
+    {
+      "epoch": 6.0,
+      "eval_loss": 4.90324068069458,
+      "eval_runtime": 4.1113,
+      "eval_samples_per_second": 593.724,
+      "eval_steps_per_second": 9.486,
+      "step": 432
+    },
+    {
+      "epoch": 6.111692844677138,
+      "grad_norm": 0.4563899040222168,
+      "learning_rate": 0.0003,
+      "loss": 4.772877883911133,
+      "step": 440
+    },
+    {
+      "epoch": 6.2513089005235605,
+      "grad_norm": 0.8754525780677795,
+      "learning_rate": 0.0003,
+      "loss": 4.744700241088867,
+      "step": 450
+    },
+    {
+      "epoch": 6.390924956369982,
+      "grad_norm": 0.958769679069519,
+      "learning_rate": 0.0003,
+      "loss": 4.739905548095703,
+      "step": 460
+    },
+    {
+      "epoch": 6.530541012216405,
+      "grad_norm": 0.4947556257247925,
+      "learning_rate": 0.0003,
+      "loss": 4.723302841186523,
+      "step": 470
+    },
+    {
+      "epoch": 6.670157068062827,
+      "grad_norm": 0.32547932863235474,
+      "learning_rate": 0.0003,
+      "loss": 4.660966873168945,
+      "step": 480
+    },
+    {
+      "epoch": 6.80977312390925,
+      "grad_norm": 0.3814781606197357,
+      "learning_rate": 0.0003,
+      "loss": 4.627614593505859,
+      "step": 490
+    },
+    {
+      "epoch": 6.949389179755672,
+      "grad_norm": 0.8448502421379089,
+      "learning_rate": 0.0003,
+      "loss": 4.592163848876953,
+      "step": 500
+    },
+    {
+      "epoch": 7.0,
+      "eval_loss": 4.711507797241211,
+      "eval_runtime": 4.1149,
+      "eval_samples_per_second": 593.208,
+      "eval_steps_per_second": 9.478,
+      "step": 504
+    },
+    {
+      "epoch": 7.0837696335078535,
+      "grad_norm": 0.7609332203865051,
+      "learning_rate": 0.0003,
+      "loss": 4.566059875488281,
+      "step": 510
+    },
+    {
+      "epoch": 7.223385689354275,
+      "grad_norm": 0.6690367460250854,
+      "learning_rate": 0.0003,
+      "loss": 4.540019989013672,
+      "step": 520
+    },
+    {
+      "epoch": 7.363001745200698,
+      "grad_norm": 0.7389764785766602,
+      "learning_rate": 0.0003,
+      "loss": 4.510246276855469,
+      "step": 530
+    },
+    {
+      "epoch": 7.50261780104712,
+      "grad_norm": 0.3664371371269226,
+      "learning_rate": 0.0003,
+      "loss": 4.489366149902343,
+      "step": 540
+    },
+    {
+      "epoch": 7.642233856893543,
+      "grad_norm": 0.35561150312423706,
+      "learning_rate": 0.0003,
+      "loss": 4.4513092041015625,
+      "step": 550
+    },
+    {
+      "epoch": 7.781849912739965,
+      "grad_norm": 0.3935790956020355,
+      "learning_rate": 0.0003,
+      "loss": 4.422682189941407,
+      "step": 560
+    },
+    {
+      "epoch": 7.9214659685863875,
+      "grad_norm": 0.46314138174057007,
+      "learning_rate": 0.0003,
+      "loss": 4.4081573486328125,
+      "step": 570
+    },
+    {
+      "epoch": 8.0,
+      "eval_loss": 4.537301540374756,
+      "eval_runtime": 4.0993,
+      "eval_samples_per_second": 595.464,
+      "eval_steps_per_second": 9.514,
+      "step": 576
+    },
+    {
+      "epoch": 8.055846422338568,
+      "grad_norm": 0.7809078097343445,
+      "learning_rate": 0.0003,
+      "loss": 4.366108703613281,
+      "step": 580
+    },
+    {
+      "epoch": 8.195462478184991,
+      "grad_norm": 0.8947575092315674,
+      "learning_rate": 0.0003,
+      "loss": 4.3550163269042965,
+      "step": 590
+    },
+    {
+      "epoch": 8.335078534031414,
+      "grad_norm": 0.7821335196495056,
+      "learning_rate": 0.0003,
+      "loss": 4.352532196044922,
+      "step": 600
+    },
+    {
+      "epoch": 8.474694589877837,
+      "grad_norm": 0.37841781973838806,
+      "learning_rate": 0.0003,
+      "loss": 4.314500427246093,
+      "step": 610
+    },
+    {
+      "epoch": 8.614310645724258,
+      "grad_norm": 0.35384002327919006,
+      "learning_rate": 0.0003,
+      "loss": 4.2818855285644535,
+      "step": 620
+    },
+    {
+      "epoch": 8.75392670157068,
+      "grad_norm": 0.32511261105537415,
+      "learning_rate": 0.0003,
+      "loss": 4.258856964111328,
+      "step": 630
+    },
+    {
+      "epoch": 8.893542757417103,
+      "grad_norm": 0.5073819756507874,
+      "learning_rate": 0.0003,
+      "loss": 4.258666610717773,
+      "step": 640
+    },
+    {
+      "epoch": 9.0,
+      "eval_loss": 4.416695594787598,
+      "eval_runtime": 4.1007,
+      "eval_samples_per_second": 595.27,
+      "eval_steps_per_second": 9.511,
+      "step": 648
+    },
+    {
+      "epoch": 9.027923211169284,
+      "grad_norm": 0.5689071416854858,
+      "learning_rate": 0.0003,
+      "loss": 4.231290817260742,
+      "step": 650
+    },
+    {
+      "epoch": 9.167539267015707,
+      "grad_norm": 0.9789859056472778,
+      "learning_rate": 0.0003,
+      "loss": 4.191323852539062,
+      "step": 660
+    },
+    {
+      "epoch": 9.30715532286213,
+      "grad_norm": 0.8916944265365601,
+      "learning_rate": 0.0003,
+      "loss": 4.21032943725586,
+      "step": 670
+    },
+    {
+      "epoch": 9.44677137870855,
+      "grad_norm": 0.5173555612564087,
+      "learning_rate": 0.0003,
+      "loss": 4.184156036376953,
+      "step": 680
+    },
+    {
+      "epoch": 9.586387434554974,
+      "grad_norm": 0.34253525733947754,
+      "learning_rate": 0.0003,
+      "loss": 4.148382568359375,
+      "step": 690
+    },
+    {
+      "epoch": 9.726003490401396,
+      "grad_norm": 0.42179620265960693,
+      "learning_rate": 0.0003,
+      "loss": 4.139811706542969,
+      "step": 700
+    },
+    {
+      "epoch": 9.86561954624782,
+      "grad_norm": 0.504753053188324,
+      "learning_rate": 0.0003,
+      "loss": 4.118508529663086,
+      "step": 710
+    },
+    {
+      "epoch": 10.0,
+      "grad_norm": 0.5721378922462463,
+      "learning_rate": 0.0003,
+      "loss": 4.104060363769531,
+      "step": 720
+    },
+    {
+      "epoch": 10.0,
+      "eval_loss": 4.33306360244751,
+      "eval_runtime": 4.1147,
+      "eval_samples_per_second": 593.236,
+      "eval_steps_per_second": 9.478,
+      "step": 720
+    },
+    {
+      "epoch": 10.139616055846423,
+      "grad_norm": 0.841262698173523,
+      "learning_rate": 0.0003,
+      "loss": 4.05061149597168,
+      "step": 730
+    },
+    {
+      "epoch": 10.279232111692846,
+      "grad_norm": 0.8539153933525085,
+      "learning_rate": 0.0003,
+      "loss": 4.062906265258789,
+      "step": 740
+    },
+    {
+      "epoch": 10.418848167539267,
+      "grad_norm": 0.5237502455711365,
+      "learning_rate": 0.0003,
+      "loss": 4.045460510253906,
+      "step": 750
+    },
+    {
+      "epoch": 10.55846422338569,
+      "grad_norm": 0.38326072692871094,
+      "learning_rate": 0.0003,
+      "loss": 4.043621444702149,
+      "step": 760
+    },
+    {
+      "epoch": 10.698080279232112,
+      "grad_norm": 0.4053055942058563,
+      "learning_rate": 0.0003,
+      "loss": 4.020166015625,
+      "step": 770
+    },
+    {
+      "epoch": 10.837696335078533,
+      "grad_norm": 0.46810540556907654,
+      "learning_rate": 0.0003,
+      "loss": 4.012229919433594,
+      "step": 780
+    },
+    {
+      "epoch": 10.977312390924956,
+      "grad_norm": 0.45282575488090515,
+      "learning_rate": 0.0003,
+      "loss": 3.9989597320556642,
+      "step": 790
+    },
+    {
+      "epoch": 11.0,
+      "eval_loss": 4.255781650543213,
+      "eval_runtime": 4.0982,
+      "eval_samples_per_second": 595.63,
+      "eval_steps_per_second": 9.516,
+      "step": 792
+    },
+    {
+      "epoch": 11.111692844677139,
+      "grad_norm": 0.812175989151001,
+      "learning_rate": 0.0003,
+      "loss": 3.9566085815429686,
+      "step": 800
+    },
+    {
+      "epoch": 11.25130890052356,
+      "grad_norm": 0.7589244842529297,
+      "learning_rate": 0.0003,
+      "loss": 3.9253917694091798,
+      "step": 810
+    },
+    {
+      "epoch": 11.390924956369982,
+      "grad_norm": 0.6504644751548767,
+      "learning_rate": 0.0003,
+      "loss": 3.927887725830078,
+      "step": 820
+    },
+    {
+      "epoch": 11.530541012216405,
+      "grad_norm": 0.47677385807037354,
+      "learning_rate": 0.0003,
+      "loss": 3.9283016204833983,
+      "step": 830
+    },
+    {
+      "epoch": 11.670157068062828,
+      "grad_norm": 0.4261974096298218,
+      "learning_rate": 0.0003,
+      "loss": 3.9170894622802734,
+      "step": 840
+    },
+    {
+      "epoch": 11.809773123909249,
+      "grad_norm": 0.437930166721344,
+      "learning_rate": 0.0003,
+      "loss": 3.906512451171875,
+      "step": 850
+    },
+    {
+      "epoch": 11.949389179755672,
+      "grad_norm": 0.3658443093299866,
+      "learning_rate": 0.0003,
+      "loss": 3.906256103515625,
+      "step": 860
+    },
+    {
+      "epoch": 12.0,
+      "eval_loss": 4.190188407897949,
+      "eval_runtime": 4.0969,
+      "eval_samples_per_second": 595.819,
+      "eval_steps_per_second": 9.519,
+      "step": 864
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 864,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 12,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": false,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 7.656447700673495e+17,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/runs/t0/t0_S3_lr0.0003/train.log b/runs/t0/t0_S3_lr0.0003/train.log
new file mode 100644
index 0000000000000000000000000000000000000000..a7f55e89fb8d16f9b6e1930f715d202780e96d44
--- /dev/null
+++ b/runs/t0/t0_S3_lr0.0003/train.log
@@ -0,0 +1,1599 @@
+W0817 19:32:41.033000 2215445 torch/distributed/run.py:851] 
+W0817 19:32:41.033000 2215445 torch/distributed/run.py:851] *****************************************
+W0817 19:32:41.033000 2215445 torch/distributed/run.py:851] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+W0817 19:32:41.033000 2215445 torch/distributed/run.py:851] *****************************************
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+total params: 258072576
+[dataset] loaded packed cache: 36614 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+total params: 258072576
+[dataset] loaded packed cache: 36614 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+total params: 258072576
+[dataset] loaded packed cache: 36614 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+total params: 258072576
+[dataset] loaded packed cache: 36614 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+total params: 258072576
+[dataset] loaded packed cache: 36614 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+total params: 258072576
+[dataset] loaded packed cache: 36614 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+total params: 258072576
+[dataset] loaded packed cache: 36614 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+total params: 258072576
+[dataset] loaded packed cache: 36614 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+[transformers] ***** Running training *****
+[transformers]   Num examples = 36,614
+[transformers]   Num Epochs = 12
+[transformers]   Num update steps per epoch = 72
+[transformers]   Instantaneous batch size per device = 8
+[transformers]   Total train batch size (w. parallel, distributed & accumulation) = 512
+[transformers]   Gradient Accumulation steps = 8
+[transformers]   Total optimization steps = 864
+[transformers]   Number of trainable parameters = 258,072,576
+wandb: WARNING Changes to your `wandb` environment variables will be ignored because your `wandb` session has already started. For more information on how to modify your settings with `wandb.init()` arguments, please refer to https://wandb.me/wandb-init.
+wandb: setting up run lwba0vwf
+wandb: Tracking run with wandb version 0.28.2
+wandb: Run data is saved locally in /home/ubuntu/workspace/data-augmentations-for-pretraining/wandb/run-20260817_193302-lwba0vwf
+wandb: Run `wandb offline` to turn off syncing.
+wandb: Syncing run t0_S3_lr0.0003
+wandb: ⭐️ View project at https://wandb.ai/slowrun-wankers/data-aug-pretraining
+wandb: 🚀 View run at https://wandb.ai/slowrun-wankers/data-aug-pretraining/runs/lwba0vwf
+
  0%|          | 0/864 [00:00<?, ?it/s]wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+[rank1]:W0817 19:33:25.620000 2215568 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+[rank4]:W0817 19:33:25.783000 2215571 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+[rank5]:W0817 19:33:25.996000 2215572 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+[rank2]:W0817 19:33:26.358000 2215569 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+[rank6]:W0817 19:33:26.383000 2215573 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+[rank7]:W0817 19:33:26.593000 2215574 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+[rank3]:W0817 19:33:27.012000 2215570 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+[rank0]:W0817 19:33:27.369000 2215567 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+
  0%|          | 1/864 [00:59<14:19:48, 59.78s/it]
                                                  
{'loss': '12.09', 'grad_norm': '2.551', 'learning_rate': '0', 'epoch': '0.01396'}
+
  0%|          | 1/864 [00:59<14:19:48, 59.78s/it]
  0%|          | 2/864 [01:00<6:01:29, 25.16s/it] 
  0%|          | 3/864 [01:01<3:22:02, 14.08s/it]
  0%|          | 4/864 [01:02<2:07:10,  8.87s/it]
  1%|          | 5/864 [01:03<1:25:49,  5.99s/it]
  1%|          | 6/864 [01:04<1:00:55,  4.26s/it]
  1%|          | 7/864 [01:05<45:07,  3.16s/it]  
  1%|          | 8/864 [01:06<34:46,  2.44s/it]
  1%|          | 9/864 [01:06<27:50,  1.95s/it]
  1%|          | 10/864 [01:07<23:08,  1.63s/it]
                                                
{'loss': '11.76', 'grad_norm': '2.49', 'learning_rate': '2.7e-05', 'epoch': '0.1396'}
+
  1%|          | 10/864 [01:07<23:08,  1.63s/it]
  1%|▏         | 11/864 [01:08<19:55,  1.40s/it]
  1%|▏         | 12/864 [01:09<17:41,  1.25s/it]
  2%|▏         | 13/864 [01:10<16:08,  1.14s/it]
  2%|▏         | 14/864 [01:11<15:04,  1.06s/it]
  2%|▏         | 15/864 [01:12<14:18,  1.01s/it]
  2%|▏         | 16/864 [01:13<13:46,  1.03it/s]
  2%|▏         | 17/864 [01:14<13:24,  1.05it/s]
  2%|▏         | 18/864 [01:14<13:07,  1.07it/s]
  2%|▏         | 19/864 [01:15<12:56,  1.09it/s]
  2%|▏         | 20/864 [01:16<12:48,  1.10it/s]
                                                
{'loss': '10.83', 'grad_norm': '1.995', 'learning_rate': '5.7e-05', 'epoch': '0.2792'}
+
  2%|▏         | 20/864 [01:16<12:48,  1.10it/s]
  2%|▏         | 21/864 [01:17<12:41,  1.11it/s]
  3%|▎         | 22/864 [01:18<12:37,  1.11it/s]
  3%|▎         | 23/864 [01:19<12:34,  1.11it/s]
  3%|▎         | 24/864 [01:20<12:31,  1.12it/s]
  3%|▎         | 25/864 [01:21<12:29,  1.12it/s]
  3%|▎         | 26/864 [01:22<12:27,  1.12it/s]
  3%|▎         | 27/864 [01:22<12:26,  1.12it/s]
  3%|▎         | 28/864 [01:23<12:24,  1.12it/s]
  3%|▎         | 29/864 [01:24<12:22,  1.12it/s]
  3%|▎         | 30/864 [01:25<12:21,  1.12it/s]
                                                
{'loss': '10.21', 'grad_norm': '1.869', 'learning_rate': '8.7e-05', 'epoch': '0.4188'}
+
  3%|▎         | 30/864 [01:25<12:21,  1.12it/s]
  4%|▎         | 31/864 [01:26<12:20,  1.12it/s]
  4%|▎         | 32/864 [01:27<12:19,  1.12it/s]
  4%|▍         | 33/864 [01:28<12:19,  1.12it/s]
  4%|▍         | 34/864 [01:29<12:18,  1.12it/s]
  4%|▍         | 35/864 [01:30<12:17,  1.12it/s]
  4%|▍         | 36/864 [01:30<12:16,  1.12it/s]
  4%|▍         | 37/864 [01:31<12:15,  1.13it/s]
  4%|▍         | 38/864 [01:32<12:14,  1.13it/s]
  5%|▍         | 39/864 [01:33<12:13,  1.13it/s]
  5%|▍         | 40/864 [01:34<12:12,  1.13it/s]
                                                
{'loss': '9.37', 'grad_norm': '1.596', 'learning_rate': '0.000117', 'epoch': '0.5585'}
+
  5%|▍         | 40/864 [01:34<12:12,  1.13it/s]
  5%|▍         | 41/864 [01:35<12:11,  1.12it/s]
  5%|▍         | 42/864 [01:36<12:10,  1.12it/s]
  5%|▍         | 43/864 [01:37<12:09,  1.12it/s]
  5%|▌         | 44/864 [01:38<12:09,  1.12it/s]
  5%|▌         | 45/864 [01:38<12:08,  1.12it/s]
  5%|▌         | 46/864 [01:39<12:07,  1.12it/s]
  5%|▌         | 47/864 [01:40<12:06,  1.12it/s]
  6%|▌         | 48/864 [01:41<12:06,  1.12it/s]
  6%|▌         | 49/864 [01:42<12:05,  1.12it/s]
  6%|▌         | 50/864 [01:43<12:04,  1.12it/s]
                                                
{'loss': '8.435', 'grad_norm': '1.034', 'learning_rate': '0.000147', 'epoch': '0.6981'}
+
  6%|▌         | 50/864 [01:43<12:04,  1.12it/s]
  6%|▌         | 51/864 [01:44<12:03,  1.12it/s]
  6%|▌         | 52/864 [01:45<12:03,  1.12it/s]
  6%|▌         | 53/864 [01:46<12:02,  1.12it/s]
  6%|▋         | 54/864 [01:46<12:01,  1.12it/s]
  6%|▋         | 55/864 [01:47<12:01,  1.12it/s]
  6%|▋         | 56/864 [01:48<11:59,  1.12it/s]
  7%|▋         | 57/864 [01:49<11:58,  1.12it/s]
  7%|▋         | 58/864 [01:50<11:57,  1.12it/s]
  7%|▋         | 59/864 [01:51<11:56,  1.12it/s]
  7%|▋         | 60/864 [01:52<11:55,  1.12it/s]
                                                
{'loss': '7.778', 'grad_norm': '0.7044', 'learning_rate': '0.000177', 'epoch': '0.8377'}
+
  7%|▋         | 60/864 [01:52<11:55,  1.12it/s]
  7%|▋         | 61/864 [01:53<11:55,  1.12it/s]
  7%|▋         | 62/864 [01:54<11:54,  1.12it/s]
  7%|▋         | 63/864 [01:55<11:53,  1.12it/s]
  7%|▋         | 64/864 [01:55<11:52,  1.12it/s]
  8%|▊         | 65/864 [01:56<11:51,  1.12it/s]
  8%|▊         | 66/864 [01:57<11:51,  1.12it/s]
  8%|▊         | 67/864 [01:58<11:50,  1.12it/s]
  8%|▊         | 68/864 [01:59<11:49,  1.12it/s]
  8%|▊         | 69/864 [02:00<11:49,  1.12it/s]
  8%|▊         | 70/864 [02:01<11:49,  1.12it/s]
                                                
{'loss': '7.527', 'grad_norm': '0.4868', 'learning_rate': '0.000207', 'epoch': '0.9773'}
+
  8%|▊         | 70/864 [02:01<11:49,  1.12it/s]
  8%|▊         | 71/864 [02:02<11:49,  1.12it/s]
  8%|▊         | 72/864 [02:02<10:29,  1.26it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:02, 16.89it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 12.31it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 11.07it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 10.48it/s]
+
 28%|██▊       | 11/39 [00:01<00:02, 10.19it/s]
+
 33%|███▎      | 13/39 [00:01<00:02,  9.98it/s]
+
 38%|███▊      | 15/39 [00:01<00:02,  9.88it/s]
+
 41%|████      | 16/39 [00:01<00:02,  9.82it/s]
+
 44%|████▎     | 17/39 [00:01<00:02,  9.77it/s]
+
 46%|████▌     | 18/39 [00:01<00:02,  9.75it/s]
+
 49%|████▊     | 19/39 [00:01<00:02,  9.75it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01,  9.70it/s]
+
 54%|█████▍    | 21/39 [00:02<00:01,  9.67it/s]
+
 56%|█████▋    | 22/39 [00:02<00:01,  9.67it/s]
+
 59%|█████▉    | 23/39 [00:02<00:01,  9.69it/s]
+
 62%|██████▏   | 24/39 [00:02<00:01,  9.68it/s]
+
 64%|██████▍   | 25/39 [00:02<00:01,  9.67it/s]
+
 67%|██████▋   | 26/39 [00:02<00:01,  9.67it/s]
+
 69%|██████▉   | 27/39 [00:02<00:01,  9.65it/s]
+
 72%|███████▏  | 28/39 [00:02<00:01,  9.64it/s]
+
 74%|███████▍  | 29/39 [00:02<00:01,  9.63it/s]
+
 77%|███████▋  | 30/39 [00:02<00:00,  9.64it/s]
+
 79%|███████▉  | 31/39 [00:03<00:00,  9.65it/s]
+
 82%|████████▏ | 32/39 [00:03<00:00,  9.65it/s]
+
 85%|████████▍ | 33/39 [00:03<00:00,  9.65it/s]
+
 87%|████████▋ | 34/39 [00:03<00:00,  9.64it/s]
+
 90%|████████▉ | 35/39 [00:03<00:00,  9.65it/s]
+
 92%|█████████▏| 36/39 [00:03<00:00,  9.65it/s]
+
 95%|█████████▍| 37/39 [00:03<00:00,  9.65it/s]
+
 97%|█████████▋| 38/39 [00:03<00:00,  9.65it/s]
+
100%|██████████| 39/39 [00:03<00:00,  9.65it/s]
                                                
+
                                               
{'eval_loss': '7.445', 'eval_runtime': '28.51', 'eval_samples_per_second': '85.62', 'eval_steps_per_second': '1.368', 'epoch': '1'}
+
  8%|▊         | 72/864 [02:31<10:29,  1.26it/s]
+
100%|██████████| 39/39 [00:04<00:00,  9.65it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-72
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-72/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-72/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:01<00:00,  1.05s/it]
Writing model shards: 100%|██████████| 1/1 [00:01<00:00,  1.05s/it]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-72/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-72/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-72/tokenizer_config.json
+[eval] epoch 1.0 step 72 loss 7.4453 (best 7.4453 @ ep 1.0, 0/10000 since best)
+
+
                                               
  8%|▊         | 73/864 [02:35<2:17:08, 10.40s/it]
  9%|▊         | 74/864 [02:36<1:39:23,  7.55s/it]
  9%|▊         | 75/864 [02:37<1:13:00,  5.55s/it]
  9%|▉         | 76/864 [02:38<54:33,  4.15s/it]  
  9%|▉         | 77/864 [02:39<41:39,  3.18s/it]
  9%|▉         | 78/864 [02:40<32:38,  2.49s/it]
  9%|▉         | 79/864 [02:40<26:20,  2.01s/it]
  9%|▉         | 80/864 [02:41<21:55,  1.68s/it]
                                                
{'loss': '7.38', 'grad_norm': '0.6972', 'learning_rate': '0.000237', 'epoch': '1.112'}
+
  9%|▉         | 80/864 [02:41<21:55,  1.68s/it]
  9%|▉         | 81/864 [02:42<18:49,  1.44s/it]
  9%|▉         | 82/864 [02:43<16:39,  1.28s/it]
 10%|▉         | 83/864 [02:44<15:08,  1.16s/it]
 10%|▉         | 84/864 [02:45<14:05,  1.08s/it]
 10%|▉         | 85/864 [02:46<13:21,  1.03s/it]
 10%|▉         | 86/864 [02:47<12:49,  1.01it/s]
 10%|█         | 87/864 [02:48<12:27,  1.04it/s]
 10%|█         | 88/864 [02:48<12:11,  1.06it/s]
 10%|█         | 89/864 [02:49<12:00,  1.08it/s]
 10%|█         | 90/864 [02:50<11:52,  1.09it/s]
                                                
{'loss': '7.172', 'grad_norm': '0.6061', 'learning_rate': '0.000267', 'epoch': '1.251'}
+
 10%|█         | 90/864 [02:50<11:52,  1.09it/s]
 11%|█         | 91/864 [02:51<11:46,  1.09it/s]
 11%|█         | 92/864 [02:52<11:41,  1.10it/s]
 11%|█         | 93/864 [02:53<11:38,  1.10it/s]
 11%|█         | 94/864 [02:54<11:35,  1.11it/s]
 11%|█         | 95/864 [02:55<11:33,  1.11it/s]
 11%|█         | 96/864 [02:56<11:32,  1.11it/s]
 11%|█         | 97/864 [02:57<11:30,  1.11it/s]
 11%|█▏        | 98/864 [02:57<11:28,  1.11it/s]
 11%|█▏        | 99/864 [02:58<11:27,  1.11it/s]
 12%|█▏        | 100/864 [02:59<11:26,  1.11it/s]
                                                 
{'loss': '7.012', 'grad_norm': '0.8713', 'learning_rate': '0.000297', 'epoch': '1.391'}
+
 12%|█▏        | 100/864 [02:59<11:26,  1.11it/s]
 12%|█▏        | 101/864 [03:00<11:26,  1.11it/s]
 12%|█▏        | 102/864 [03:01<11:24,  1.11it/s]
 12%|█▏        | 103/864 [03:02<11:23,  1.11it/s]
 12%|█▏        | 104/864 [03:03<11:22,  1.11it/s]
 12%|█▏        | 105/864 [03:04<11:21,  1.11it/s]
 12%|█▏        | 106/864 [03:05<11:20,  1.11it/s]
 12%|█▏        | 107/864 [03:06<11:19,  1.11it/s]
 12%|█▎        | 108/864 [03:06<11:18,  1.11it/s]
 13%|█▎        | 109/864 [03:07<11:17,  1.11it/s]
 13%|█▎        | 110/864 [03:08<11:17,  1.11it/s]
                                                 
{'loss': '6.859', 'grad_norm': '0.5368', 'learning_rate': '0.0003', 'epoch': '1.531'}
+
 13%|█▎        | 110/864 [03:08<11:17,  1.11it/s]
 13%|█▎        | 111/864 [03:09<11:16,  1.11it/s]
 13%|█▎        | 112/864 [03:10<11:15,  1.11it/s]
 13%|█▎        | 113/864 [03:11<11:13,  1.11it/s]
 13%|█▎        | 114/864 [03:12<11:12,  1.11it/s]
 13%|█▎        | 115/864 [03:13<11:12,  1.11it/s]
 13%|█▎        | 116/864 [03:14<11:11,  1.11it/s]
 14%|█▎        | 117/864 [03:15<11:10,  1.11it/s]
 14%|█▎        | 118/864 [03:15<11:09,  1.11it/s]
 14%|█▍        | 119/864 [03:16<11:08,  1.11it/s]
 14%|█▍        | 120/864 [03:17<11:07,  1.11it/s]
                                                 
{'loss': '6.714', 'grad_norm': '0.5024', 'learning_rate': '0.0003', 'epoch': '1.67'}
+
 14%|█▍        | 120/864 [03:17<11:07,  1.11it/s]
 14%|█▍        | 121/864 [03:18<11:06,  1.11it/s]
 14%|█▍        | 122/864 [03:19<11:06,  1.11it/s]
 14%|█▍        | 123/864 [03:20<11:05,  1.11it/s]
 14%|█▍        | 124/864 [03:21<11:05,  1.11it/s]
 14%|█▍        | 125/864 [03:22<11:04,  1.11it/s]
 15%|█▍        | 126/864 [03:23<11:02,  1.11it/s]
 15%|█▍        | 127/864 [03:23<11:01,  1.11it/s]
 15%|█▍        | 128/864 [03:24<11:02,  1.11it/s]
 15%|█▍        | 129/864 [03:25<11:01,  1.11it/s]
 15%|█▌        | 130/864 [03:26<11:00,  1.11it/s]
                                                 
{'loss': '6.58', 'grad_norm': '0.7802', 'learning_rate': '0.0003', 'epoch': '1.81'}
+
 15%|█▌        | 130/864 [03:26<11:00,  1.11it/s]
 15%|█▌        | 131/864 [03:27<10:59,  1.11it/s]
 15%|█▌        | 132/864 [03:28<10:58,  1.11it/s]
 15%|█▌        | 133/864 [03:29<10:56,  1.11it/s]
 16%|█▌        | 134/864 [03:30<10:55,  1.11it/s]
 16%|█▌        | 135/864 [03:31<10:54,  1.11it/s]
 16%|█▌        | 136/864 [03:32<10:53,  1.11it/s]
 16%|█▌        | 137/864 [03:32<10:52,  1.11it/s]
 16%|█▌        | 138/864 [03:33<10:51,  1.11it/s]
 16%|█▌        | 139/864 [03:34<10:51,  1.11it/s]
 16%|█▌        | 140/864 [03:35<10:50,  1.11it/s]
                                                 
{'loss': '6.486', 'grad_norm': '1.064', 'learning_rate': '0.0003', 'epoch': '1.949'}
+
 16%|█▌        | 140/864 [03:35<10:50,  1.11it/s]
 16%|█▋        | 141/864 [03:36<10:49,  1.11it/s]
 16%|█▋        | 142/864 [03:37<10:49,  1.11it/s]
 17%|█▋        | 143/864 [03:38<10:47,  1.11it/s]
 17%|█▋        | 144/864 [03:38<09:34,  1.25it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:01, 19.65it/s]
+
 10%|█         | 4/39 [00:00<00:02, 11.98it/s]
+
 15%|█▌        | 6/39 [00:00<00:03, 10.92it/s]
+
 21%|██        | 8/39 [00:00<00:03, 10.28it/s]
+
 26%|██▌       | 10/39 [00:00<00:02, 10.13it/s]
+
 31%|███       | 12/39 [00:01<00:02,  9.88it/s]
+
 36%|███▌      | 14/39 [00:01<00:02,  9.89it/s]
+
 41%|████      | 16/39 [00:01<00:02,  9.74it/s]
+
 46%|████▌     | 18/39 [00:01<00:02,  9.78it/s]
+
 49%|████▊     | 19/39 [00:01<00:02,  9.76it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01,  9.64it/s]
+
 54%|█████▍    | 21/39 [00:02<00:01,  9.72it/s]
+
 56%|█████▋    | 22/39 [00:02<00:01,  9.70it/s]
+
 59%|█████▉    | 23/39 [00:02<00:01,  9.70it/s]
+
 62%|██████▏   | 24/39 [00:02<00:01,  9.69it/s]
+
 64%|██████▍   | 25/39 [00:02<00:01,  9.69it/s]
+
 67%|██████▋   | 26/39 [00:02<00:01,  9.68it/s]
+
 69%|██████▉   | 27/39 [00:02<00:01,  9.67it/s]
+
 72%|███████▏  | 28/39 [00:02<00:01,  9.67it/s]
+
 74%|███████▍  | 29/39 [00:02<00:01,  9.67it/s]
+
 77%|███████▋  | 30/39 [00:03<00:00,  9.66it/s]
+
 79%|███████▉  | 31/39 [00:03<00:00,  9.65it/s]
+
 82%|████████▏ | 32/39 [00:03<00:00,  9.66it/s]
+
 85%|████████▍ | 33/39 [00:03<00:00,  9.66it/s]
+
 87%|████████▋ | 34/39 [00:03<00:00,  9.66it/s]
+
 90%|████████▉ | 35/39 [00:03<00:00,  9.66it/s]
+
 92%|█████████▏| 36/39 [00:03<00:00,  9.65it/s]
+
 95%|█████████▍| 37/39 [00:03<00:00,  9.66it/s]
+
 97%|█████████▋| 38/39 [00:03<00:00,  9.66it/s]
+
100%|██████████| 39/39 [00:03<00:00,  9.66it/s]
                                                 
+
                                               
{'eval_loss': '6.4', 'eval_runtime': '4.104', 'eval_samples_per_second': '594.9', 'eval_steps_per_second': '9.504', 'epoch': '2'}
+
 17%|█▋        | 144/864 [03:43<09:34,  1.25it/s]
+
100%|██████████| 39/39 [00:04<00:00,  9.66it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-144
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-144/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-144/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.11it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.11it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-144/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-144/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-144/tokenizer_config.json
+[eval] epoch 2.0 step 144 loss 6.3999 (best 6.3999 @ ep 2.0, 0/10000 since best)
+
+
                                               
 17%|█▋        | 145/864 [03:47<37:08,  3.10s/it]
 17%|█▋        | 146/864 [03:48<29:10,  2.44s/it]
 17%|█▋        | 147/864 [03:49<23:37,  1.98s/it]
 17%|█▋        | 148/864 [03:50<19:43,  1.65s/it]
 17%|█▋        | 149/864 [03:50<16:59,  1.43s/it]
 17%|█▋        | 150/864 [03:51<15:04,  1.27s/it]
                                                 
{'loss': '6.365', 'grad_norm': '0.6536', 'learning_rate': '0.0003', 'epoch': '2.084'}
+
 17%|█▋        | 150/864 [03:51<15:04,  1.27s/it]
 17%|█▋        | 151/864 [03:52<13:44,  1.16s/it]
 18%|█▊        | 152/864 [03:53<12:49,  1.08s/it]
 18%|█▊        | 153/864 [03:54<12:09,  1.03s/it]
 18%|█▊        | 154/864 [03:55<11:40,  1.01it/s]
 18%|█▊        | 155/864 [03:56<11:20,  1.04it/s]
 18%|█▊        | 156/864 [03:57<11:06,  1.06it/s]
 18%|█▊        | 157/864 [03:58<10:56,  1.08it/s]
 18%|█▊        | 158/864 [03:59<10:48,  1.09it/s]
 18%|█▊        | 159/864 [03:59<10:43,  1.10it/s]
 19%|█▊        | 160/864 [04:00<10:38,  1.10it/s]
                                                 
{'loss': '6.283', 'grad_norm': '0.8615', 'learning_rate': '0.0003', 'epoch': '2.223'}
+
 19%|█▊        | 160/864 [04:00<10:38,  1.10it/s]
 19%|█▊        | 161/864 [04:01<10:36,  1.10it/s]
 19%|█▉        | 162/864 [04:02<10:33,  1.11it/s]
 19%|█▉        | 163/864 [04:03<10:31,  1.11it/s]
 19%|█▉        | 164/864 [04:04<10:30,  1.11it/s]
 19%|█▉        | 165/864 [04:05<10:28,  1.11it/s]
 19%|█▉        | 166/864 [04:06<10:27,  1.11it/s]
 19%|█▉        | 167/864 [04:07<10:26,  1.11it/s]
 19%|█▉        | 168/864 [04:08<10:25,  1.11it/s]
 20%|█▉        | 169/864 [04:08<10:24,  1.11it/s]
 20%|█▉        | 170/864 [04:09<10:23,  1.11it/s]
                                                 
{'loss': '6.193', 'grad_norm': '0.4822', 'learning_rate': '0.0003', 'epoch': '2.363'}
+
 20%|█▉        | 170/864 [04:09<10:23,  1.11it/s]
 20%|█▉        | 171/864 [04:10<10:22,  1.11it/s]
 20%|█▉        | 172/864 [04:11<10:21,  1.11it/s]
 20%|██        | 173/864 [04:12<10:20,  1.11it/s]
 20%|██        | 174/864 [04:13<10:19,  1.11it/s]
 20%|██        | 175/864 [04:14<10:18,  1.11it/s]
 20%|██        | 176/864 [04:15<10:17,  1.11it/s]
 20%|██        | 177/864 [04:16<10:16,  1.11it/s]
 21%|██        | 178/864 [04:17<10:16,  1.11it/s]
 21%|██        | 179/864 [04:17<10:15,  1.11it/s]
 21%|██        | 180/864 [04:18<10:16,  1.11it/s]
                                                 
{'loss': '6.144', 'grad_norm': '1.422', 'learning_rate': '0.0003', 'epoch': '2.503'}
+
 21%|██        | 180/864 [04:18<10:16,  1.11it/s]
 21%|██        | 181/864 [04:19<10:14,  1.11it/s]
 21%|██        | 182/864 [04:20<10:13,  1.11it/s]
 21%|██        | 183/864 [04:21<10:12,  1.11it/s]
 21%|██▏       | 184/864 [04:22<10:10,  1.11it/s]
 21%|██▏       | 185/864 [04:23<10:09,  1.11it/s]
 22%|██▏       | 186/864 [04:24<10:09,  1.11it/s]
 22%|██▏       | 187/864 [04:25<10:08,  1.11it/s]
 22%|██▏       | 188/864 [04:26<10:08,  1.11it/s]
 22%|██▏       | 189/864 [04:26<10:09,  1.11it/s]
 22%|██▏       | 190/864 [04:27<10:08,  1.11it/s]
                                                 
{'loss': '6.064', 'grad_norm': '0.5396', 'learning_rate': '0.0003', 'epoch': '2.642'}
+
 22%|██▏       | 190/864 [04:27<10:08,  1.11it/s]
 22%|██▏       | 191/864 [04:28<10:06,  1.11it/s]
 22%|██▏       | 192/864 [04:29<10:04,  1.11it/s]
 22%|██▏       | 193/864 [04:30<10:03,  1.11it/s]
 22%|██▏       | 194/864 [04:31<10:02,  1.11it/s]
 23%|██▎       | 195/864 [04:32<10:01,  1.11it/s]
 23%|██▎       | 196/864 [04:33<10:00,  1.11it/s]
 23%|██▎       | 197/864 [04:34<09:59,  1.11it/s]
 23%|██▎       | 198/864 [04:35<09:58,  1.11it/s]
 23%|██▎       | 199/864 [04:35<09:57,  1.11it/s]
 23%|██▎       | 200/864 [04:36<09:56,  1.11it/s]
                                                 
{'loss': '5.972', 'grad_norm': '1.306', 'learning_rate': '0.0003', 'epoch': '2.782'}
+
 23%|██▎       | 200/864 [04:36<09:56,  1.11it/s]
 23%|██▎       | 201/864 [04:37<09:55,  1.11it/s]
 23%|██▎       | 202/864 [04:38<09:55,  1.11it/s]
 23%|██▎       | 203/864 [04:39<09:54,  1.11it/s]
 24%|██▎       | 204/864 [04:40<09:53,  1.11it/s]
 24%|██▎       | 205/864 [04:41<09:52,  1.11it/s]
 24%|██▍       | 206/864 [04:42<09:51,  1.11it/s]
 24%|██▍       | 207/864 [04:43<09:50,  1.11it/s]
 24%|██▍       | 208/864 [04:44<09:49,  1.11it/s]
 24%|██▍       | 209/864 [04:44<09:48,  1.11it/s]
 24%|██▍       | 210/864 [04:45<09:47,  1.11it/s]
                                                 
{'loss': '5.91', 'grad_norm': '0.7041', 'learning_rate': '0.0003', 'epoch': '2.921'}
+
 24%|██▍       | 210/864 [04:45<09:47,  1.11it/s]
 24%|██▍       | 211/864 [04:46<09:46,  1.11it/s]
 25%|██▍       | 212/864 [04:47<09:45,  1.11it/s]
 25%|██▍       | 213/864 [04:48<09:45,  1.11it/s]
 25%|██▍       | 214/864 [04:49<09:44,  1.11it/s]
 25%|██▍       | 215/864 [04:50<09:43,  1.11it/s]
 25%|██▌       | 216/864 [04:50<08:38,  1.25it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:01, 19.22it/s]
+
 10%|█         | 4/39 [00:00<00:02, 11.91it/s]
+
 15%|█▌        | 6/39 [00:00<00:03, 10.87it/s]
+
 21%|██        | 8/39 [00:00<00:03, 10.26it/s]
+
 26%|██▌       | 10/39 [00:00<00:02, 10.11it/s]
+
 31%|███       | 12/39 [00:01<00:02,  9.84it/s]
+
 36%|███▌      | 14/39 [00:01<00:02,  9.87it/s]
+
 41%|████      | 16/39 [00:01<00:02,  9.70it/s]
+
 46%|████▌     | 18/39 [00:01<00:02,  9.76it/s]
+
 49%|████▊     | 19/39 [00:01<00:02,  9.73it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01,  9.62it/s]
+
 56%|█████▋    | 22/39 [00:02<00:01,  9.72it/s]
+
 59%|█████▉    | 23/39 [00:02<00:01,  9.72it/s]
+
 62%|██████▏   | 24/39 [00:02<00:01,  9.71it/s]
+
 64%|██████▍   | 25/39 [00:02<00:01,  9.69it/s]
+
 67%|██████▋   | 26/39 [00:02<00:01,  9.67it/s]
+
 69%|██████▉   | 27/39 [00:02<00:01,  9.67it/s]
+
 72%|███████▏  | 28/39 [00:02<00:01,  9.66it/s]
+
 74%|███████▍  | 29/39 [00:02<00:01,  9.66it/s]
+
 77%|███████▋  | 30/39 [00:03<00:00,  9.66it/s]
+
 79%|███████▉  | 31/39 [00:03<00:00,  9.66it/s]
+
 82%|████████▏ | 32/39 [00:03<00:00,  9.66it/s]
+
 85%|████████▍ | 33/39 [00:03<00:00,  9.66it/s]
+
 87%|████████▋ | 34/39 [00:03<00:00,  9.66it/s]
+
 90%|████████▉ | 35/39 [00:03<00:00,  9.65it/s]
+
 92%|█████████▏| 36/39 [00:03<00:00,  9.66it/s]
+
 95%|█████████▍| 37/39 [00:03<00:00,  9.65it/s]
+
 97%|█████████▋| 38/39 [00:03<00:00,  9.66it/s]
+
100%|██████████| 39/39 [00:03<00:00,  9.66it/s]
                                                 
+
                                               
{'eval_loss': '5.868', 'eval_runtime': '4.116', 'eval_samples_per_second': '593', 'eval_steps_per_second': '9.475', 'epoch': '3'}
+
 25%|██▌       | 216/864 [04:54<08:38,  1.25it/s]
+
100%|██████████| 39/39 [00:04<00:00,  9.66it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-216
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-216/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-216/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.11it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.11it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-216/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-216/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-216/tokenizer_config.json
+[eval] epoch 3.0 step 216 loss 5.8679 (best 5.8679 @ ep 3.0, 0/10000 since best)
+
+
                                               
 25%|██▌       | 217/864 [04:59<33:38,  3.12s/it]
 25%|██▌       | 218/864 [05:00<26:24,  2.45s/it]
 25%|██▌       | 219/864 [05:01<21:21,  1.99s/it]
 25%|██▌       | 220/864 [05:02<17:49,  1.66s/it]
                                                 
{'loss': '5.828', 'grad_norm': '1.009', 'learning_rate': '0.0003', 'epoch': '3.056'}
+
 25%|██▌       | 220/864 [05:02<17:49,  1.66s/it]
 26%|██▌       | 221/864 [05:03<15:21,  1.43s/it]
 26%|██▌       | 222/864 [05:03<13:37,  1.27s/it]
 26%|██▌       | 223/864 [05:04<12:24,  1.16s/it]
 26%|██▌       | 224/864 [05:05<11:33,  1.08s/it]
 26%|██▌       | 225/864 [05:06<10:57,  1.03s/it]
 26%|██▌       | 226/864 [05:07<10:31,  1.01it/s]
 26%|██▋       | 227/864 [05:08<10:13,  1.04it/s]
 26%|██▋       | 228/864 [05:09<10:00,  1.06it/s]
 27%|██▋       | 229/864 [05:10<09:50,  1.07it/s]
 27%|██▋       | 230/864 [05:11<09:44,  1.09it/s]
                                                 
{'loss': '5.782', 'grad_norm': '0.7435', 'learning_rate': '0.0003', 'epoch': '3.195'}
+
 27%|██▋       | 230/864 [05:11<09:44,  1.09it/s]
 27%|██▋       | 231/864 [05:12<09:39,  1.09it/s]
 27%|██▋       | 232/864 [05:12<09:35,  1.10it/s]
 27%|██▋       | 233/864 [05:13<09:32,  1.10it/s]
 27%|██▋       | 234/864 [05:14<09:29,  1.11it/s]
 27%|██▋       | 235/864 [05:15<09:28,  1.11it/s]
 27%|██▋       | 236/864 [05:16<09:26,  1.11it/s]
 27%|██▋       | 237/864 [05:17<09:25,  1.11it/s]
 28%|██▊       | 238/864 [05:18<09:23,  1.11it/s]
 28%|██▊       | 239/864 [05:19<09:22,  1.11it/s]
 28%|██▊       | 240/864 [05:20<09:21,  1.11it/s]
                                                 
{'loss': '5.704', 'grad_norm': '0.4217', 'learning_rate': '0.0003', 'epoch': '3.335'}
+
 28%|██▊       | 240/864 [05:20<09:21,  1.11it/s]
 28%|██▊       | 241/864 [05:21<09:20,  1.11it/s]
 28%|██▊       | 242/864 [05:21<09:18,  1.11it/s]
 28%|██▊       | 243/864 [05:22<09:18,  1.11it/s]
 28%|██▊       | 244/864 [05:23<09:16,  1.11it/s]
 28%|██▊       | 245/864 [05:24<09:16,  1.11it/s]
 28%|██▊       | 246/864 [05:25<09:15,  1.11it/s]
 29%|██▊       | 247/864 [05:26<09:14,  1.11it/s]
 29%|██▊       | 248/864 [05:27<09:13,  1.11it/s]
 29%|██▉       | 249/864 [05:28<09:12,  1.11it/s]
 29%|██▉       | 250/864 [05:29<09:11,  1.11it/s]
                                                 
{'loss': '5.654', 'grad_norm': '0.735', 'learning_rate': '0.0003', 'epoch': '3.475'}
+
 29%|██▉       | 250/864 [05:29<09:11,  1.11it/s]
 29%|██▉       | 251/864 [05:29<09:10,  1.11it/s]
 29%|██▉       | 252/864 [05:30<09:09,  1.11it/s]
 29%|██▉       | 253/864 [05:31<09:09,  1.11it/s]
 29%|██▉       | 254/864 [05:32<09:08,  1.11it/s]
 30%|██▉       | 255/864 [05:33<09:07,  1.11it/s]
 30%|██▉       | 256/864 [05:34<09:06,  1.11it/s]
 30%|██▉       | 257/864 [05:35<09:05,  1.11it/s]
 30%|██▉       | 258/864 [05:36<09:04,  1.11it/s]
 30%|██▉       | 259/864 [05:37<09:03,  1.11it/s]
 30%|███       | 260/864 [05:38<09:02,  1.11it/s]
                                                 
{'loss': '5.64', 'grad_norm': '0.7231', 'learning_rate': '0.0003', 'epoch': '3.614'}
+
 30%|███       | 260/864 [05:38<09:02,  1.11it/s]
 30%|███       | 261/864 [05:38<09:01,  1.11it/s]
 30%|███       | 262/864 [05:39<09:00,  1.11it/s]
 30%|███       | 263/864 [05:40<09:00,  1.11it/s]
 31%|███       | 264/864 [05:41<08:59,  1.11it/s]
 31%|███       | 265/864 [05:42<08:58,  1.11it/s]
 31%|███       | 266/864 [05:43<08:57,  1.11it/s]
 31%|███       | 267/864 [05:44<08:56,  1.11it/s]
 31%|███       | 268/864 [05:45<08:55,  1.11it/s]
 31%|███       | 269/864 [05:46<08:55,  1.11it/s]
 31%|███▏      | 270/864 [05:47<08:54,  1.11it/s]
                                                 
{'loss': '5.543', 'grad_norm': '0.434', 'learning_rate': '0.0003', 'epoch': '3.754'}
+
 31%|███▏      | 270/864 [05:47<08:54,  1.11it/s]
 31%|███▏      | 271/864 [05:47<08:53,  1.11it/s]
 31%|███▏      | 272/864 [05:48<08:52,  1.11it/s]
 32%|███▏      | 273/864 [05:49<08:51,  1.11it/s]
 32%|███▏      | 274/864 [05:50<08:50,  1.11it/s]
 32%|███▏      | 275/864 [05:51<08:50,  1.11it/s]
 32%|███▏      | 276/864 [05:52<08:49,  1.11it/s]
 32%|███▏      | 277/864 [05:53<08:48,  1.11it/s]
 32%|███▏      | 278/864 [05:54<08:47,  1.11it/s]
 32%|███▏      | 279/864 [05:55<08:46,  1.11it/s]
 32%|███▏      | 280/864 [05:56<08:46,  1.11it/s]
                                                 
{'loss': '5.49', 'grad_norm': '0.6494', 'learning_rate': '0.0003', 'epoch': '3.894'}
+
 32%|███▏      | 280/864 [05:56<08:46,  1.11it/s]
 33%|███▎      | 281/864 [05:56<08:45,  1.11it/s]
 33%|███▎      | 282/864 [05:57<08:44,  1.11it/s]
 33%|███▎      | 283/864 [05:58<08:42,  1.11it/s]
 33%|███▎      | 284/864 [05:59<08:41,  1.11it/s]
 33%|███▎      | 285/864 [06:00<08:40,  1.11it/s]
 33%|███▎      | 286/864 [06:01<08:41,  1.11it/s]
 33%|███▎      | 287/864 [06:02<08:39,  1.11it/s]
 33%|███▎      | 288/864 [06:02<07:40,  1.25it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:02, 14.53it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 11.75it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 10.79it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 10.32it/s]
+
 28%|██▊       | 11/39 [00:01<00:02, 10.08it/s]
+
 33%|███▎      | 13/39 [00:01<00:02,  9.92it/s]
+
 36%|███▌      | 14/39 [00:01<00:02,  9.86it/s]
+
 38%|███▊      | 15/39 [00:01<00:02,  9.81it/s]
+
 41%|████      | 16/39 [00:01<00:02,  9.77it/s]
+
 44%|████▎     | 17/39 [00:01<00:02,  9.73it/s]
+
 46%|████▌     | 18/39 [00:01<00:02,  9.70it/s]
+
 49%|████▊     | 19/39 [00:01<00:02,  9.69it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01,  9.68it/s]
+
 54%|█████▍    | 21/39 [00:02<00:01,  9.66it/s]
+
 56%|█████▋    | 22/39 [00:02<00:01,  9.65it/s]
+
 59%|█████▉    | 23/39 [00:02<00:01,  9.68it/s]
+
 62%|██████▏   | 24/39 [00:02<00:01,  9.67it/s]
+
 64%|██████▍   | 25/39 [00:02<00:01,  9.66it/s]
+
 67%|██████▋   | 26/39 [00:02<00:01,  9.66it/s]
+
 69%|██████▉   | 27/39 [00:02<00:01,  9.65it/s]
+
 72%|███████▏  | 28/39 [00:02<00:01,  9.65it/s]
+
 74%|███████▍  | 29/39 [00:02<00:01,  9.65it/s]
+
 77%|███████▋  | 30/39 [00:03<00:00,  9.66it/s]
+
 79%|███████▉  | 31/39 [00:03<00:00,  9.66it/s]
+
 82%|████████▏ | 32/39 [00:03<00:00,  9.65it/s]
+
 85%|████████▍ | 33/39 [00:03<00:00,  9.66it/s]
+
 87%|████████▋ | 34/39 [00:03<00:00,  9.66it/s]
+
 90%|████████▉ | 35/39 [00:03<00:00,  9.65it/s]
+
 92%|█████████▏| 36/39 [00:03<00:00,  9.65it/s]
+
 95%|█████████▍| 37/39 [00:03<00:00,  9.65it/s]
+
 97%|█████████▋| 38/39 [00:03<00:00,  9.63it/s]
+
100%|██████████| 39/39 [00:03<00:00,  9.64it/s]
                                                 
+
                                               
{'eval_loss': '5.469', 'eval_runtime': '4.104', 'eval_samples_per_second': '594.8', 'eval_steps_per_second': '9.503', 'epoch': '4'}
+
 33%|███▎      | 288/864 [06:07<07:40,  1.25it/s]
+
100%|██████████| 39/39 [00:04<00:00,  9.64it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-288
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-288/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-288/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.12it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.12it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-288/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-288/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-288/tokenizer_config.json
+[eval] epoch 4.0 step 288 loss 5.4692 (best 5.4692 @ ep 4.0, 0/10000 since best)
+
+
                                               
 33%|███▎      | 289/864 [06:11<29:50,  3.11s/it]
 34%|███▎      | 290/864 [06:12<23:25,  2.45s/it]
                                                 
{'loss': '5.422', 'grad_norm': '0.4068', 'learning_rate': '0.0003', 'epoch': '4.028'}
+
 34%|███▎      | 290/864 [06:12<23:25,  2.45s/it]
 34%|███▎      | 291/864 [06:13<18:56,  1.98s/it]
 34%|███▍      | 292/864 [06:14<15:48,  1.66s/it]
 34%|███▍      | 293/864 [06:15<13:36,  1.43s/it]
 34%|███▍      | 294/864 [06:15<12:04,  1.27s/it]
 34%|███▍      | 295/864 [06:16<10:59,  1.16s/it]
 34%|███▍      | 296/864 [06:17<10:14,  1.08s/it]
 34%|███▍      | 297/864 [06:18<09:42,  1.03s/it]
 34%|███▍      | 298/864 [06:19<09:20,  1.01it/s]
 35%|███▍      | 299/864 [06:20<09:04,  1.04it/s]
 35%|███▍      | 300/864 [06:21<08:52,  1.06it/s]
                                                 
{'loss': '5.366', 'grad_norm': '0.9482', 'learning_rate': '0.0003', 'epoch': '4.168'}
+
 35%|███▍      | 300/864 [06:21<08:52,  1.06it/s]
 35%|███▍      | 301/864 [06:22<08:43,  1.07it/s]
 35%|███▍      | 302/864 [06:23<08:38,  1.08it/s]
 35%|███▌      | 303/864 [06:24<08:33,  1.09it/s]
 35%|███▌      | 304/864 [06:24<08:29,  1.10it/s]
 35%|███▌      | 305/864 [06:25<08:27,  1.10it/s]
 35%|███▌      | 306/864 [06:26<08:24,  1.11it/s]
 36%|███▌      | 307/864 [06:27<08:22,  1.11it/s]
 36%|███▌      | 308/864 [06:28<08:21,  1.11it/s]
 36%|███▌      | 309/864 [06:29<08:20,  1.11it/s]
 36%|███▌      | 310/864 [06:30<08:19,  1.11it/s]
                                                 
{'loss': '5.338', 'grad_norm': '0.7341', 'learning_rate': '0.0003', 'epoch': '4.307'}
+
 36%|███▌      | 310/864 [06:30<08:19,  1.11it/s]
 36%|███▌      | 311/864 [06:31<08:17,  1.11it/s]
 36%|███▌      | 312/864 [06:32<08:16,  1.11it/s]
 36%|███▌      | 313/864 [06:33<08:15,  1.11it/s]
 36%|███▋      | 314/864 [06:33<08:14,  1.11it/s]
 36%|███▋      | 315/864 [06:34<08:13,  1.11it/s]
 37%|███▋      | 316/864 [06:35<08:12,  1.11it/s]
 37%|███▋      | 317/864 [06:36<08:12,  1.11it/s]
 37%|███▋      | 318/864 [06:37<08:11,  1.11it/s]
 37%|███▋      | 319/864 [06:38<08:11,  1.11it/s]
 37%|███▋      | 320/864 [06:39<08:09,  1.11it/s]
                                                 
{'loss': '5.294', 'grad_norm': '0.5627', 'learning_rate': '0.0003', 'epoch': '4.447'}
+
 37%|███▋      | 320/864 [06:39<08:09,  1.11it/s]
 37%|███▋      | 321/864 [06:40<08:08,  1.11it/s]
 37%|███▋      | 322/864 [06:41<08:07,  1.11it/s]
 37%|███▋      | 323/864 [06:42<08:06,  1.11it/s]
 38%|███▊      | 324/864 [06:42<08:05,  1.11it/s]
 38%|███▊      | 325/864 [06:43<08:04,  1.11it/s]
 38%|███▊      | 326/864 [06:44<08:03,  1.11it/s]
 38%|███▊      | 327/864 [06:45<08:02,  1.11it/s]
 38%|███▊      | 328/864 [06:46<08:01,  1.11it/s]
 38%|███▊      | 329/864 [06:47<08:00,  1.11it/s]
 38%|███▊      | 330/864 [06:48<07:59,  1.11it/s]
                                                 
{'loss': '5.244', 'grad_norm': '0.6366', 'learning_rate': '0.0003', 'epoch': '4.586'}
+
 38%|███▊      | 330/864 [06:48<07:59,  1.11it/s]
 38%|███▊      | 331/864 [06:49<07:59,  1.11it/s]
 38%|███▊      | 332/864 [06:50<07:58,  1.11it/s]
 39%|███▊      | 333/864 [06:51<07:57,  1.11it/s]
 39%|███▊      | 334/864 [06:51<07:56,  1.11it/s]
 39%|███▉      | 335/864 [06:52<07:55,  1.11it/s]
 39%|███▉      | 336/864 [06:53<07:54,  1.11it/s]
 39%|███▉      | 337/864 [06:54<07:53,  1.11it/s]
 39%|███▉      | 338/864 [06:55<07:53,  1.11it/s]
 39%|███▉      | 339/864 [06:56<07:51,  1.11it/s]
 39%|███▉      | 340/864 [06:57<07:50,  1.11it/s]
                                                 
{'loss': '5.197', 'grad_norm': '0.4826', 'learning_rate': '0.0003', 'epoch': '4.726'}
+
 39%|███▉      | 340/864 [06:57<07:50,  1.11it/s]
 39%|███▉      | 341/864 [06:58<07:49,  1.11it/s]
 40%|███▉      | 342/864 [06:59<07:48,  1.11it/s]
 40%|███▉      | 343/864 [07:00<07:48,  1.11it/s]
 40%|███▉      | 344/864 [07:00<07:47,  1.11it/s]
 40%|███▉      | 345/864 [07:01<07:46,  1.11it/s]
 40%|████      | 346/864 [07:02<07:45,  1.11it/s]
 40%|████      | 347/864 [07:03<07:44,  1.11it/s]
 40%|████      | 348/864 [07:04<07:43,  1.11it/s]
 40%|████      | 349/864 [07:05<07:42,  1.11it/s]
 41%|████      | 350/864 [07:06<07:42,  1.11it/s]
                                                 
{'loss': '5.135', 'grad_norm': '0.5314', 'learning_rate': '0.0003', 'epoch': '4.866'}
+
 41%|████      | 350/864 [07:06<07:42,  1.11it/s]
 41%|████      | 351/864 [07:07<07:41,  1.11it/s]
 41%|████      | 352/864 [07:08<07:40,  1.11it/s]
 41%|████      | 353/864 [07:08<07:39,  1.11it/s]
 41%|████      | 354/864 [07:09<07:38,  1.11it/s]
 41%|████      | 355/864 [07:10<07:37,  1.11it/s]
 41%|████      | 356/864 [07:11<07:37,  1.11it/s]
 41%|████▏     | 357/864 [07:12<07:36,  1.11it/s]
 41%|████▏     | 358/864 [07:13<07:35,  1.11it/s]
 42%|████▏     | 359/864 [07:14<07:34,  1.11it/s]
 42%|████▏     | 360/864 [07:14<06:43,  1.25it/s]
                                                 
{'loss': '5.122', 'grad_norm': '1.139', 'learning_rate': '0.0003', 'epoch': '5'}
+
 42%|████▏     | 360/864 [07:14<06:43,  1.25it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:02, 14.78it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 11.74it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 10.76it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 10.33it/s]
+
 28%|██▊       | 11/39 [00:01<00:02, 10.07it/s]
+
 33%|███▎      | 13/39 [00:01<00:02,  9.94it/s]
+
 36%|███▌      | 14/39 [00:01<00:02,  9.87it/s]
+
 38%|███▊      | 15/39 [00:01<00:02,  9.83it/s]
+
 41%|████      | 16/39 [00:01<00:02,  9.69it/s]
+
 46%|████▌     | 18/39 [00:01<00:02,  9.76it/s]
+
 49%|████▊     | 19/39 [00:01<00:02,  9.74it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01,  9.61it/s]
+
 56%|█████▋    | 22/39 [00:02<00:01,  9.71it/s]
+
 59%|█████▉    | 23/39 [00:02<00:01,  9.71it/s]
+
 62%|██████▏   | 24/39 [00:02<00:01,  9.71it/s]
+
 64%|██████▍   | 25/39 [00:02<00:01,  9.69it/s]
+
 67%|██████▋   | 26/39 [00:02<00:01,  9.68it/s]
+
 69%|██████▉   | 27/39 [00:02<00:01,  9.68it/s]
+
 72%|███████▏  | 28/39 [00:02<00:01,  9.68it/s]
+
 74%|███████▍  | 29/39 [00:02<00:01,  9.67it/s]
+
 77%|███████▋  | 30/39 [00:03<00:00,  9.66it/s]
+
 79%|███████▉  | 31/39 [00:03<00:00,  9.65it/s]
+
 82%|████████▏ | 32/39 [00:03<00:00,  9.65it/s]
+
 85%|████████▍ | 33/39 [00:03<00:00,  9.65it/s]
+
 87%|████████▋ | 34/39 [00:03<00:00,  9.65it/s]
+
 90%|████████▉ | 35/39 [00:03<00:00,  9.64it/s]
+
 92%|█████████▏| 36/39 [00:03<00:00,  9.65it/s]
+
 95%|█████████▍| 37/39 [00:03<00:00,  9.63it/s]
+
 97%|█████████▋| 38/39 [00:03<00:00,  9.63it/s]
+
100%|██████████| 39/39 [00:03<00:00,  9.64it/s]
                                                 
+
                                               
{'eval_loss': '5.216', 'eval_runtime': '4.098', 'eval_samples_per_second': '595.7', 'eval_steps_per_second': '9.518', 'epoch': '5'}
+
 42%|████▏     | 360/864 [07:19<06:43,  1.25it/s]
+
100%|██████████| 39/39 [00:04<00:00,  9.64it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-360
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-360/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-360/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.11it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.11it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-360/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-360/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-360/tokenizer_config.json
+[eval] epoch 5.0 step 360 loss 5.2158 (best 5.2158 @ ep 5.0, 0/10000 since best)
+
+
                                               
 42%|████▏     | 361/864 [07:23<26:05,  3.11s/it]
 42%|████▏     | 362/864 [07:24<20:29,  2.45s/it]
 42%|████▏     | 363/864 [07:25<16:34,  1.99s/it]
 42%|████▏     | 364/864 [07:26<13:49,  1.66s/it]
 42%|████▏     | 365/864 [07:27<11:54,  1.43s/it]
 42%|████▏     | 366/864 [07:27<10:33,  1.27s/it]
 42%|████▏     | 367/864 [07:28<09:36,  1.16s/it]
 43%|████▎     | 368/864 [07:29<08:56,  1.08s/it]
 43%|████▎     | 369/864 [07:30<08:29,  1.03s/it]
 43%|████▎     | 370/864 [07:31<08:08,  1.01it/s]
                                                 
{'loss': '5.087', 'grad_norm': '1.021', 'learning_rate': '0.0003', 'epoch': '5.14'}
+
 43%|████▎     | 370/864 [07:31<08:08,  1.01it/s]
 43%|████▎     | 371/864 [07:32<07:54,  1.04it/s]
 43%|████▎     | 372/864 [07:33<07:44,  1.06it/s]
 43%|████▎     | 373/864 [07:34<07:36,  1.07it/s]
 43%|████▎     | 374/864 [07:35<07:31,  1.08it/s]
 43%|████▎     | 375/864 [07:36<07:27,  1.09it/s]
 44%|████▎     | 376/864 [07:36<07:24,  1.10it/s]
 44%|████▎     | 377/864 [07:37<07:21,  1.10it/s]
 44%|████▍     | 378/864 [07:38<07:20,  1.10it/s]
 44%|████▍     | 379/864 [07:39<07:18,  1.11it/s]
 44%|████▍     | 380/864 [07:40<07:16,  1.11it/s]
                                                 
{'loss': '5.048', 'grad_norm': '0.4545', 'learning_rate': '0.0003', 'epoch': '5.279'}
+
 44%|████▍     | 380/864 [07:40<07:16,  1.11it/s]
 44%|████▍     | 381/864 [07:41<07:15,  1.11it/s]
 44%|████▍     | 382/864 [07:42<07:14,  1.11it/s]
 44%|████▍     | 383/864 [07:43<07:13,  1.11it/s]
 44%|████▍     | 384/864 [07:44<07:12,  1.11it/s]
 45%|████▍     | 385/864 [07:45<07:11,  1.11it/s]
 45%|████▍     | 386/864 [07:45<07:10,  1.11it/s]
 45%|████▍     | 387/864 [07:46<07:09,  1.11it/s]
 45%|████▍     | 388/864 [07:47<07:08,  1.11it/s]
 45%|████▌     | 389/864 [07:48<07:07,  1.11it/s]
 45%|████▌     | 390/864 [07:49<07:06,  1.11it/s]
                                                 
{'loss': '4.97', 'grad_norm': '0.4114', 'learning_rate': '0.0003', 'epoch': '5.419'}
+
 45%|████▌     | 390/864 [07:49<07:06,  1.11it/s]
 45%|████▌     | 391/864 [07:50<07:05,  1.11it/s]
 45%|████▌     | 392/864 [07:51<07:04,  1.11it/s]
 45%|████▌     | 393/864 [07:52<07:03,  1.11it/s]
 46%|████▌     | 394/864 [07:53<07:02,  1.11it/s]
 46%|████▌     | 395/864 [07:54<07:01,  1.11it/s]
 46%|████▌     | 396/864 [07:54<07:00,  1.11it/s]
 46%|████▌     | 397/864 [07:55<06:59,  1.11it/s]
 46%|████▌     | 398/864 [07:56<06:58,  1.11it/s]
 46%|████▌     | 399/864 [07:57<06:57,  1.11it/s]
 46%|████▋     | 400/864 [07:58<06:57,  1.11it/s]
                                                 
{'loss': '4.928', 'grad_norm': '0.7791', 'learning_rate': '0.0003', 'epoch': '5.558'}
+
 46%|████▋     | 400/864 [07:58<06:57,  1.11it/s]
 46%|████▋     | 401/864 [07:59<06:56,  1.11it/s]
 47%|████▋     | 402/864 [08:00<06:55,  1.11it/s]
 47%|████▋     | 403/864 [08:01<06:54,  1.11it/s]
 47%|████▋     | 404/864 [08:02<06:53,  1.11it/s]
 47%|████▋     | 405/864 [08:03<06:52,  1.11it/s]
 47%|████▋     | 406/864 [08:03<06:52,  1.11it/s]
 47%|████▋     | 407/864 [08:04<06:51,  1.11it/s]
 47%|████▋     | 408/864 [08:05<06:50,  1.11it/s]
 47%|████▋     | 409/864 [08:06<06:49,  1.11it/s]
 47%|████▋     | 410/864 [08:07<06:47,  1.11it/s]
                                                 
{'loss': '4.943', 'grad_norm': '0.6807', 'learning_rate': '0.0003', 'epoch': '5.698'}
+
 47%|████▋     | 410/864 [08:07<06:47,  1.11it/s]
 48%|████▊     | 411/864 [08:08<06:47,  1.11it/s]
 48%|████▊     | 412/864 [08:09<06:46,  1.11it/s]
 48%|████▊     | 413/864 [08:10<06:45,  1.11it/s]
 48%|████▊     | 414/864 [08:11<06:44,  1.11it/s]
 48%|████▊     | 415/864 [08:12<06:43,  1.11it/s]
 48%|████▊     | 416/864 [08:12<06:42,  1.11it/s]
 48%|████▊     | 417/864 [08:13<06:41,  1.11it/s]
 48%|████▊     | 418/864 [08:14<06:40,  1.11it/s]
 48%|████▊     | 419/864 [08:15<06:39,  1.11it/s]
 49%|████▊     | 420/864 [08:16<06:38,  1.11it/s]
                                                 
{'loss': '4.902', 'grad_norm': '0.6517', 'learning_rate': '0.0003', 'epoch': '5.838'}
+
 49%|████▊     | 420/864 [08:16<06:38,  1.11it/s]
 49%|████▊     | 421/864 [08:17<06:38,  1.11it/s]
 49%|████▉     | 422/864 [08:18<06:37,  1.11it/s]
 49%|████▉     | 423/864 [08:19<06:36,  1.11it/s]
 49%|████▉     | 424/864 [08:20<06:35,  1.11it/s]
 49%|████▉     | 425/864 [08:21<06:35,  1.11it/s]
 49%|████▉     | 426/864 [08:21<06:34,  1.11it/s]
 49%|████▉     | 427/864 [08:22<06:33,  1.11it/s]
 50%|████▉     | 428/864 [08:23<06:32,  1.11it/s]
 50%|████▉     | 429/864 [08:24<06:31,  1.11it/s]
 50%|████▉     | 430/864 [08:25<06:30,  1.11it/s]
                                                 
{'loss': '4.843', 'grad_norm': '0.337', 'learning_rate': '0.0003', 'epoch': '5.977'}
+
 50%|████▉     | 430/864 [08:25<06:30,  1.11it/s]
 50%|████▉     | 431/864 [08:26<06:29,  1.11it/s]
 50%|█████     | 432/864 [08:27<05:45,  1.25it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:01, 19.79it/s]
+
 10%|█         | 4/39 [00:00<00:02, 12.24it/s]
+
 15%|█▌        | 6/39 [00:00<00:03, 10.89it/s]
+
 21%|██        | 8/39 [00:00<00:02, 10.37it/s]
+
 26%|██▌       | 10/39 [00:00<00:02, 10.10it/s]
+
 31%|███       | 12/39 [00:01<00:02,  9.94it/s]
+
 36%|███▌      | 14/39 [00:01<00:02,  9.84it/s]
+
 41%|████      | 16/39 [00:01<00:02,  9.76it/s]
+
 44%|████▎     | 17/39 [00:01<00:02,  9.71it/s]
+
 46%|████▌     | 18/39 [00:01<00:02,  9.70it/s]
+
 49%|████▊     | 19/39 [00:01<00:02,  9.69it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01,  9.68it/s]
+
 54%|█████▍    | 21/39 [00:02<00:01,  9.67it/s]
+
 56%|█████▋    | 22/39 [00:02<00:01,  9.66it/s]
+
 59%|█████▉    | 23/39 [00:02<00:01,  9.68it/s]
+
 62%|██████▏   | 24/39 [00:02<00:01,  9.67it/s]
+
 64%|██████▍   | 25/39 [00:02<00:01,  9.67it/s]
+
 67%|██████▋   | 26/39 [00:02<00:01,  9.67it/s]
+
 69%|██████▉   | 27/39 [00:02<00:01,  9.65it/s]
+
 72%|███████▏  | 28/39 [00:02<00:01,  9.64it/s]
+
 74%|███████▍  | 29/39 [00:02<00:01,  9.65it/s]
+
 77%|███████▋  | 30/39 [00:03<00:00,  9.66it/s]
+
 79%|███████▉  | 31/39 [00:03<00:00,  9.64it/s]
+
 82%|████████▏ | 32/39 [00:03<00:00,  9.66it/s]
+
 85%|████████▍ | 33/39 [00:03<00:00,  9.65it/s]
+
 87%|████████▋ | 34/39 [00:03<00:00,  9.66it/s]
+
 90%|████████▉ | 35/39 [00:03<00:00,  9.65it/s]
+
 92%|█████████▏| 36/39 [00:03<00:00,  9.66it/s]
+
 95%|█████████▍| 37/39 [00:03<00:00,  9.65it/s]
+
 97%|█████████▋| 38/39 [00:03<00:00,  9.66it/s]
+
100%|██████████| 39/39 [00:03<00:00,  9.66it/s]
                                                 
+
                                               
{'eval_loss': '4.903', 'eval_runtime': '4.111', 'eval_samples_per_second': '593.7', 'eval_steps_per_second': '9.486', 'epoch': '6'}
+
 50%|█████     | 432/864 [08:31<05:45,  1.25it/s]
+
100%|██████████| 39/39 [00:04<00:00,  9.66it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-432
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-432/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-432/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:01<00:00,  1.03s/it]
Writing model shards: 100%|██████████| 1/1 [00:01<00:00,  1.03s/it]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-432/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-432/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-432/tokenizer_config.json
+[eval] epoch 6.0 step 432 loss 4.9032 (best 4.9032 @ ep 6.0, 0/10000 since best)
+
+
                                               
 50%|█████     | 433/864 [08:35<22:34,  3.14s/it]
 50%|█████     | 434/864 [08:36<17:41,  2.47s/it]
 50%|█████     | 435/864 [08:37<14:17,  2.00s/it]
 50%|█████     | 436/864 [08:38<11:53,  1.67s/it]
 51%|█████     | 437/864 [08:39<10:13,  1.44s/it]
 51%|█████     | 438/864 [08:40<09:03,  1.28s/it]
 51%|█████     | 439/864 [08:41<08:14,  1.16s/it]
 51%|█████     | 440/864 [08:41<07:39,  1.08s/it]
                                                 
{'loss': '4.773', 'grad_norm': '0.4564', 'learning_rate': '0.0003', 'epoch': '6.112'}
+
 51%|█████     | 440/864 [08:41<07:39,  1.08s/it]
 51%|█████     | 441/864 [08:42<07:15,  1.03s/it]
 51%|█████     | 442/864 [08:43<06:58,  1.01it/s]
 51%|█████▏    | 443/864 [08:44<06:45,  1.04it/s]
 51%|█████▏    | 444/864 [08:45<06:36,  1.06it/s]
 52%|█████▏    | 445/864 [08:46<06:30,  1.07it/s]
 52%|█████▏    | 446/864 [08:47<06:25,  1.08it/s]
 52%|█████▏    | 447/864 [08:48<06:21,  1.09it/s]
 52%|█████▏    | 448/864 [08:49<06:18,  1.10it/s]
 52%|█████▏    | 449/864 [08:50<06:16,  1.10it/s]
 52%|█████▏    | 450/864 [08:50<06:14,  1.10it/s]
                                                 
{'loss': '4.745', 'grad_norm': '0.8755', 'learning_rate': '0.0003', 'epoch': '6.251'}
+
 52%|█████▏    | 450/864 [08:50<06:14,  1.10it/s]
 52%|█████▏    | 451/864 [08:51<06:13,  1.11it/s]
 52%|█████▏    | 452/864 [08:52<06:11,  1.11it/s]
 52%|█████▏    | 453/864 [08:53<06:10,  1.11it/s]
 53%|█████▎    | 454/864 [08:54<06:09,  1.11it/s]
 53%|█████▎    | 455/864 [08:55<06:08,  1.11it/s]
 53%|█████▎    | 456/864 [08:56<06:07,  1.11it/s]
 53%|█████▎    | 457/864 [08:57<06:06,  1.11it/s]
 53%|█████▎    | 458/864 [08:58<06:05,  1.11it/s]
 53%|█████▎    | 459/864 [08:59<06:04,  1.11it/s]
 53%|█████▎    | 460/864 [08:59<06:03,  1.11it/s]
                                                 
{'loss': '4.74', 'grad_norm': '0.9588', 'learning_rate': '0.0003', 'epoch': '6.391'}
+
 53%|█████▎    | 460/864 [08:59<06:03,  1.11it/s]
 53%|█████▎    | 461/864 [09:00<06:02,  1.11it/s]
 53%|█████▎    | 462/864 [09:01<06:01,  1.11it/s]
 54%|█████▎    | 463/864 [09:02<06:00,  1.11it/s]
 54%|█████▎    | 464/864 [09:03<05:59,  1.11it/s]
 54%|█████▍    | 465/864 [09:04<05:58,  1.11it/s]
 54%|█████▍    | 466/864 [09:05<05:57,  1.11it/s]
 54%|█████▍    | 467/864 [09:06<05:56,  1.11it/s]
 54%|█████▍    | 468/864 [09:07<05:56,  1.11it/s]
 54%|█████▍    | 469/864 [09:08<05:55,  1.11it/s]
 54%|█████▍    | 470/864 [09:08<05:54,  1.11it/s]
                                                 
{'loss': '4.723', 'grad_norm': '0.4948', 'learning_rate': '0.0003', 'epoch': '6.531'}
+
 54%|█████▍    | 470/864 [09:08<05:54,  1.11it/s]
 55%|█████▍    | 471/864 [09:09<05:53,  1.11it/s]
 55%|█████▍    | 472/864 [09:10<05:53,  1.11it/s]
 55%|█████▍    | 473/864 [09:11<05:51,  1.11it/s]
 55%|█████▍    | 474/864 [09:12<05:51,  1.11it/s]
 55%|█████▍    | 475/864 [09:13<05:50,  1.11it/s]
 55%|█████▌    | 476/864 [09:14<05:49,  1.11it/s]
 55%|█████▌    | 477/864 [09:15<05:48,  1.11it/s]
 55%|█████▌    | 478/864 [09:16<05:47,  1.11it/s]
 55%|█████▌    | 479/864 [09:17<05:46,  1.11it/s]
 56%|█████▌    | 480/864 [09:17<05:46,  1.11it/s]
                                                 
{'loss': '4.661', 'grad_norm': '0.3255', 'learning_rate': '0.0003', 'epoch': '6.67'}
+
 56%|█████▌    | 480/864 [09:17<05:46,  1.11it/s]
 56%|█████▌    | 481/864 [09:18<05:45,  1.11it/s]
 56%|█████▌    | 482/864 [09:19<05:44,  1.11it/s]
 56%|█████▌    | 483/864 [09:20<05:42,  1.11it/s]
 56%|█████▌    | 484/864 [09:21<05:41,  1.11it/s]
 56%|█████▌    | 485/864 [09:22<05:41,  1.11it/s]
 56%|█████▋    | 486/864 [09:23<05:40,  1.11it/s]
 56%|█████▋    | 487/864 [09:24<05:39,  1.11it/s]
 56%|█████▋    | 488/864 [09:25<05:38,  1.11it/s]
 57%|█████▋    | 489/864 [09:26<05:37,  1.11it/s]
 57%|█████▋    | 490/864 [09:26<05:36,  1.11it/s]
                                                 
{'loss': '4.628', 'grad_norm': '0.3815', 'learning_rate': '0.0003', 'epoch': '6.81'}
+
 57%|█████▋    | 490/864 [09:26<05:36,  1.11it/s]
 57%|█████▋    | 491/864 [09:27<05:35,  1.11it/s]
 57%|█████▋    | 492/864 [09:28<05:34,  1.11it/s]
 57%|█████▋    | 493/864 [09:29<05:33,  1.11it/s]
 57%|█████▋    | 494/864 [09:30<05:33,  1.11it/s]
 57%|█████▋    | 495/864 [09:31<05:32,  1.11it/s]
 57%|█████▋    | 496/864 [09:32<05:31,  1.11it/s]
 58%|█████▊    | 497/864 [09:33<05:30,  1.11it/s]
 58%|█████▊    | 498/864 [09:34<05:29,  1.11it/s]
 58%|█████▊    | 499/864 [09:35<05:28,  1.11it/s]
 58%|█████▊    | 500/864 [09:35<05:27,  1.11it/s]
                                                 
{'loss': '4.592', 'grad_norm': '0.8449', 'learning_rate': '0.0003', 'epoch': '6.949'}
+
 58%|█████▊    | 500/864 [09:35<05:27,  1.11it/s]
 58%|█████▊    | 501/864 [09:36<05:27,  1.11it/s]
 58%|█████▊    | 502/864 [09:37<05:26,  1.11it/s]
 58%|█████▊    | 503/864 [09:38<05:25,  1.11it/s]
 58%|█████▊    | 504/864 [09:39<04:48,  1.25it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:01, 19.15it/s]
+
 10%|█         | 4/39 [00:00<00:02, 12.13it/s]
+
 15%|█▌        | 6/39 [00:00<00:03, 10.84it/s]
+
 21%|██        | 8/39 [00:00<00:02, 10.34it/s]
+
 26%|██▌       | 10/39 [00:00<00:02, 10.08it/s]
+
 31%|███       | 12/39 [00:01<00:02,  9.93it/s]
+
 36%|███▌      | 14/39 [00:01<00:02,  9.83it/s]
+
 41%|████      | 16/39 [00:01<00:02,  9.78it/s]
+
 44%|████▎     | 17/39 [00:01<00:02,  9.74it/s]
+
 46%|████▌     | 18/39 [00:01<00:02,  9.72it/s]
+
 49%|████▊     | 19/39 [00:01<00:02,  9.72it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01,  9.70it/s]
+
 54%|█████▍    | 21/39 [00:02<00:01,  9.67it/s]
+
 56%|█████▋    | 22/39 [00:02<00:01,  9.66it/s]
+
 59%|█████▉    | 23/39 [00:02<00:01,  9.66it/s]
+
 62%|██████▏   | 24/39 [00:02<00:01,  9.68it/s]
+
 64%|██████▍   | 25/39 [00:02<00:01,  9.67it/s]
+
 67%|██████▋   | 26/39 [00:02<00:01,  9.67it/s]
+
 69%|██████▉   | 27/39 [00:02<00:01,  9.62it/s]
+
 72%|███████▏  | 28/39 [00:02<00:01,  9.63it/s]
+
 74%|███████▍  | 29/39 [00:02<00:01,  9.64it/s]
+
 77%|███████▋  | 30/39 [00:03<00:00,  9.64it/s]
+
 79%|███████▉  | 31/39 [00:03<00:00,  9.65it/s]
+
 82%|████████▏ | 32/39 [00:03<00:00,  9.64it/s]
+
 85%|████████▍ | 33/39 [00:03<00:00,  9.65it/s]
+
 87%|████████▋ | 34/39 [00:03<00:00,  9.65it/s]
+
 90%|████████▉ | 35/39 [00:03<00:00,  9.66it/s]
+
 92%|█████████▏| 36/39 [00:03<00:00,  9.66it/s]
+
 95%|█████████▍| 37/39 [00:03<00:00,  9.66it/s]
+
 97%|█████████▋| 38/39 [00:03<00:00,  9.66it/s]
+
100%|██████████| 39/39 [00:03<00:00,  9.66it/s]
                                                 
+
                                               
{'eval_loss': '4.712', 'eval_runtime': '4.115', 'eval_samples_per_second': '593.2', 'eval_steps_per_second': '9.478', 'epoch': '7'}
+
 58%|█████▊    | 504/864 [09:43<04:48,  1.25it/s]
+
100%|██████████| 39/39 [00:04<00:00,  9.66it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-504
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-504/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-504/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.12it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.12it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-504/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-504/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-504/tokenizer_config.json
+[eval] epoch 7.0 step 504 loss 4.7115 (best 4.7115 @ ep 7.0, 0/10000 since best)
+
+
                                               
 58%|█████▊    | 505/864 [09:47<18:36,  3.11s/it]
 59%|█████▊    | 506/864 [09:48<14:35,  2.45s/it]
 59%|█████▊    | 507/864 [09:49<11:47,  1.98s/it]
 59%|█████▉    | 508/864 [09:50<09:49,  1.66s/it]
 59%|█████▉    | 509/864 [09:51<08:27,  1.43s/it]
 59%|█████▉    | 510/864 [09:52<07:29,  1.27s/it]
                                                 
{'loss': '4.566', 'grad_norm': '0.7609', 'learning_rate': '0.0003', 'epoch': '7.084'}
+
 59%|█████▉    | 510/864 [09:52<07:29,  1.27s/it]
 59%|█████▉    | 511/864 [09:53<06:49,  1.16s/it]
 59%|█████▉    | 512/864 [09:53<06:20,  1.08s/it]
 59%|█████▉    | 513/864 [09:54<06:00,  1.03s/it]
 59%|█████▉    | 514/864 [09:55<05:45,  1.01it/s]
 60%|█████▉    | 515/864 [09:56<05:35,  1.04it/s]
 60%|█████▉    | 516/864 [09:57<05:28,  1.06it/s]
 60%|█████▉    | 517/864 [09:58<05:22,  1.08it/s]
 60%|█████▉    | 518/864 [09:59<05:18,  1.09it/s]
 60%|██████    | 519/864 [10:00<05:15,  1.09it/s]
 60%|██████    | 520/864 [10:01<05:13,  1.10it/s]
                                                 
{'loss': '4.54', 'grad_norm': '0.669', 'learning_rate': '0.0003', 'epoch': '7.223'}
+
 60%|██████    | 520/864 [10:01<05:13,  1.10it/s]
 60%|██████    | 521/864 [10:02<05:10,  1.10it/s]
 60%|██████    | 522/864 [10:02<05:09,  1.11it/s]
 61%|██████    | 523/864 [10:03<05:08,  1.11it/s]
 61%|██████    | 524/864 [10:04<05:06,  1.11it/s]
 61%|██████    | 525/864 [10:05<05:05,  1.11it/s]
 61%|██████    | 526/864 [10:06<05:04,  1.11it/s]
 61%|██████    | 527/864 [10:07<05:03,  1.11it/s]
 61%|██████    | 528/864 [10:08<05:02,  1.11it/s]
 61%|██████    | 529/864 [10:09<05:01,  1.11it/s]
 61%|██████▏   | 530/864 [10:10<05:00,  1.11it/s]
                                                 
{'loss': '4.51', 'grad_norm': '0.739', 'learning_rate': '0.0003', 'epoch': '7.363'}
+
 61%|██████▏   | 530/864 [10:10<05:00,  1.11it/s]
 61%|██████▏   | 531/864 [10:11<04:59,  1.11it/s]
 62%|██████▏   | 532/864 [10:11<04:58,  1.11it/s]
 62%|██████▏   | 533/864 [10:12<04:57,  1.11it/s]
 62%|██████▏   | 534/864 [10:13<04:57,  1.11it/s]
 62%|██████▏   | 535/864 [10:14<04:56,  1.11it/s]
 62%|██████▏   | 536/864 [10:15<04:55,  1.11it/s]
 62%|██████▏   | 537/864 [10:16<04:54,  1.11it/s]
 62%|██████▏   | 538/864 [10:17<04:53,  1.11it/s]
 62%|██████▏   | 539/864 [10:18<04:52,  1.11it/s]
 62%|██████▎   | 540/864 [10:19<04:51,  1.11it/s]
                                                 
{'loss': '4.489', 'grad_norm': '0.3664', 'learning_rate': '0.0003', 'epoch': '7.503'}
+
 62%|██████▎   | 540/864 [10:19<04:51,  1.11it/s]
 63%|██████▎   | 541/864 [10:20<04:50,  1.11it/s]
 63%|██████▎   | 542/864 [10:20<04:50,  1.11it/s]
 63%|██████▎   | 543/864 [10:21<04:48,  1.11it/s]
 63%|██████▎   | 544/864 [10:22<04:47,  1.11it/s]
 63%|██████▎   | 545/864 [10:23<04:47,  1.11it/s]
 63%|██████▎   | 546/864 [10:24<04:46,  1.11it/s]
 63%|██████▎   | 547/864 [10:25<04:45,  1.11it/s]
 63%|██████▎   | 548/864 [10:26<04:44,  1.11it/s]
 64%|██████▎   | 549/864 [10:27<04:43,  1.11it/s]
 64%|██████▎   | 550/864 [10:28<04:42,  1.11it/s]
                                                 
{'loss': '4.451', 'grad_norm': '0.3556', 'learning_rate': '0.0003', 'epoch': '7.642'}
+
 64%|██████▎   | 550/864 [10:28<04:42,  1.11it/s]
 64%|██████▍   | 551/864 [10:29<04:41,  1.11it/s]
 64%|██████▍   | 552/864 [10:29<04:40,  1.11it/s]
 64%|██████▍   | 553/864 [10:30<04:39,  1.11it/s]
 64%|██████▍   | 554/864 [10:31<04:39,  1.11it/s]
 64%|██████▍   | 555/864 [10:32<04:38,  1.11it/s]
 64%|██████▍   | 556/864 [10:33<04:37,  1.11it/s]
 64%|██████▍   | 557/864 [10:34<04:36,  1.11it/s]
 65%|██████▍   | 558/864 [10:35<04:35,  1.11it/s]
 65%|██████▍   | 559/864 [10:36<04:34,  1.11it/s]
 65%|██████▍   | 560/864 [10:37<04:33,  1.11it/s]
                                                 
{'loss': '4.423', 'grad_norm': '0.3936', 'learning_rate': '0.0003', 'epoch': '7.782'}
+
 65%|██████▍   | 560/864 [10:37<04:33,  1.11it/s]
 65%|██████▍   | 561/864 [10:38<04:32,  1.11it/s]
 65%|██████▌   | 562/864 [10:38<04:31,  1.11it/s]
 65%|██████▌   | 563/864 [10:39<04:30,  1.11it/s]
 65%|██████▌   | 564/864 [10:40<04:29,  1.11it/s]
 65%|██████▌   | 565/864 [10:41<04:29,  1.11it/s]
 66%|██████▌   | 566/864 [10:42<04:28,  1.11it/s]
 66%|██████▌   | 567/864 [10:43<04:27,  1.11it/s]
 66%|██████▌   | 568/864 [10:44<04:26,  1.11it/s]
 66%|██████▌   | 569/864 [10:45<04:25,  1.11it/s]
 66%|██████▌   | 570/864 [10:46<04:24,  1.11it/s]
                                                 
{'loss': '4.408', 'grad_norm': '0.4631', 'learning_rate': '0.0003', 'epoch': '7.921'}
+
 66%|██████▌   | 570/864 [10:46<04:24,  1.11it/s]
 66%|██████▌   | 571/864 [10:47<04:23,  1.11it/s]
 66%|██████▌   | 572/864 [10:47<04:22,  1.11it/s]
 66%|██████▋   | 573/864 [10:48<04:21,  1.11it/s]
 66%|██████▋   | 574/864 [10:49<04:21,  1.11it/s]
 67%|██████▋   | 575/864 [10:50<04:20,  1.11it/s]
 67%|██████▋   | 576/864 [10:51<03:50,  1.25it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:01, 19.92it/s]
+
 10%|█         | 4/39 [00:00<00:02, 12.22it/s]
+
 15%|█▌        | 6/39 [00:00<00:03, 10.90it/s]
+
 21%|██        | 8/39 [00:00<00:02, 10.37it/s]
+
 26%|██▌       | 10/39 [00:00<00:02, 10.09it/s]
+
 31%|███       | 12/39 [00:01<00:02,  9.94it/s]
+
 36%|███▌      | 14/39 [00:01<00:02,  9.83it/s]
+
 41%|████      | 16/39 [00:01<00:02,  9.77it/s]
+
 44%|████▎     | 17/39 [00:01<00:02,  9.74it/s]
+
 46%|████▌     | 18/39 [00:01<00:02,  9.72it/s]
+
 49%|████▊     | 19/39 [00:01<00:02,  9.69it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01,  9.70it/s]
+
 54%|█████▍    | 21/39 [00:02<00:01,  9.68it/s]
+
 56%|█████▋    | 22/39 [00:02<00:01,  9.67it/s]
+
 59%|█████▉    | 23/39 [00:02<00:01,  9.68it/s]
+
 62%|██████▏   | 24/39 [00:02<00:01,  9.68it/s]
+
 64%|██████▍   | 25/39 [00:02<00:01,  9.67it/s]
+
 67%|██████▋   | 26/39 [00:02<00:01,  9.66it/s]
+
 69%|██████▉   | 27/39 [00:02<00:01,  9.65it/s]
+
 72%|███████▏  | 28/39 [00:02<00:01,  9.65it/s]
+
 74%|███████▍  | 29/39 [00:02<00:01,  9.65it/s]
+
 77%|███████▋  | 30/39 [00:03<00:00,  9.64it/s]
+
 79%|███████▉  | 31/39 [00:03<00:00,  9.65it/s]
+
 82%|████████▏ | 32/39 [00:03<00:00,  9.65it/s]
+
 85%|████████▍ | 33/39 [00:03<00:00,  9.66it/s]
+
 87%|████████▋ | 34/39 [00:03<00:00,  9.65it/s]
+
 90%|████████▉ | 35/39 [00:03<00:00,  9.64it/s]
+
 92%|█████████▏| 36/39 [00:03<00:00,  9.65it/s]
+
 95%|█████████▍| 37/39 [00:03<00:00,  9.65it/s]
+
 97%|█████████▋| 38/39 [00:03<00:00,  9.65it/s]
+
100%|██████████| 39/39 [00:03<00:00,  9.65it/s]
                                                 
+
                                               
{'eval_loss': '4.537', 'eval_runtime': '4.099', 'eval_samples_per_second': '595.5', 'eval_steps_per_second': '9.514', 'epoch': '8'}
+
 67%|██████▋   | 576/864 [10:55<03:50,  1.25it/s]
+
100%|██████████| 39/39 [00:04<00:00,  9.65it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-576
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-576/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-576/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.11it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.11it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-576/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-576/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-576/tokenizer_config.json
+[eval] epoch 8.0 step 576 loss 4.5373 (best 4.5373 @ ep 8.0, 0/10000 since best)
+
+
                                               
 67%|██████▋   | 577/864 [10:59<14:51,  3.10s/it]
 67%|██████▋   | 578/864 [11:00<11:38,  2.44s/it]
 67%|██████▋   | 579/864 [11:01<09:24,  1.98s/it]
 67%|██████▋   | 580/864 [11:02<07:50,  1.66s/it]
                                                 
{'loss': '4.366', 'grad_norm': '0.7809', 'learning_rate': '0.0003', 'epoch': '8.056'}
+
 67%|██████▋   | 580/864 [11:02<07:50,  1.66s/it]
 67%|██████▋   | 581/864 [11:03<06:44,  1.43s/it]
 67%|██████▋   | 582/864 [11:04<05:58,  1.27s/it]
 67%|██████▋   | 583/864 [11:05<05:25,  1.16s/it]
 68%|██████▊   | 584/864 [11:06<05:02,  1.08s/it]
 68%|██████▊   | 585/864 [11:06<04:46,  1.03s/it]
 68%|██████▊   | 586/864 [11:07<04:34,  1.01it/s]
 68%|██████▊   | 587/864 [11:08<04:26,  1.04it/s]
 68%|██████▊   | 588/864 [11:09<04:20,  1.06it/s]
 68%|██████▊   | 589/864 [11:10<04:15,  1.07it/s]
 68%|██████▊   | 590/864 [11:11<04:12,  1.08it/s]
                                                 
{'loss': '4.355', 'grad_norm': '0.8948', 'learning_rate': '0.0003', 'epoch': '8.195'}
+
 68%|██████▊   | 590/864 [11:11<04:12,  1.08it/s]
 68%|██████▊   | 591/864 [11:12<04:09,  1.09it/s]
 69%|██████▊   | 592/864 [11:13<04:07,  1.10it/s]
 69%|██████▊   | 593/864 [11:14<04:06,  1.10it/s]
 69%|██████▉   | 594/864 [11:15<04:04,  1.10it/s]
 69%|██████▉   | 595/864 [11:15<04:03,  1.11it/s]
 69%|██████▉   | 596/864 [11:16<04:01,  1.11it/s]
 69%|██████▉   | 597/864 [11:17<04:00,  1.11it/s]
 69%|██████▉   | 598/864 [11:18<03:59,  1.11it/s]
 69%|██████▉   | 599/864 [11:19<03:58,  1.11it/s]
 69%|██████▉   | 600/864 [11:20<03:57,  1.11it/s]
                                                 
{'loss': '4.353', 'grad_norm': '0.7821', 'learning_rate': '0.0003', 'epoch': '8.335'}
+
 69%|██████▉   | 600/864 [11:20<03:57,  1.11it/s]
 70%|██████▉   | 601/864 [11:21<03:56,  1.11it/s]
 70%|██████▉   | 602/864 [11:22<03:55,  1.11it/s]
 70%|██████▉   | 603/864 [11:23<03:54,  1.11it/s]
 70%|██████▉   | 604/864 [11:24<03:53,  1.11it/s]
 70%|███████   | 605/864 [11:24<03:52,  1.11it/s]
 70%|███████   | 606/864 [11:25<03:52,  1.11it/s]
 70%|███████   | 607/864 [11:26<03:51,  1.11it/s]
 70%|███████   | 608/864 [11:27<03:50,  1.11it/s]
 70%|███████   | 609/864 [11:28<03:49,  1.11it/s]
 71%|███████   | 610/864 [11:29<03:48,  1.11it/s]
                                                 
{'loss': '4.315', 'grad_norm': '0.3784', 'learning_rate': '0.0003', 'epoch': '8.475'}
+
 71%|███████   | 610/864 [11:29<03:48,  1.11it/s]
 71%|███████   | 611/864 [11:30<03:47,  1.11it/s]
 71%|███████   | 612/864 [11:31<03:46,  1.11it/s]
 71%|███████   | 613/864 [11:32<03:45,  1.11it/s]
 71%|███████   | 614/864 [11:33<03:45,  1.11it/s]
 71%|███████   | 615/864 [11:33<03:44,  1.11it/s]
 71%|███████▏  | 616/864 [11:34<03:43,  1.11it/s]
 71%|███████▏  | 617/864 [11:35<03:42,  1.11it/s]
 72%|███████▏  | 618/864 [11:36<03:41,  1.11it/s]
 72%|███████▏  | 619/864 [11:37<03:40,  1.11it/s]
 72%|███████▏  | 620/864 [11:38<03:39,  1.11it/s]
                                                 
{'loss': '4.282', 'grad_norm': '0.3538', 'learning_rate': '0.0003', 'epoch': '8.614'}
+
 72%|███████▏  | 620/864 [11:38<03:39,  1.11it/s]
 72%|███████▏  | 621/864 [11:39<03:38,  1.11it/s]
 72%|███████▏  | 622/864 [11:40<03:37,  1.11it/s]
 72%|███████▏  | 623/864 [11:41<03:36,  1.11it/s]
 72%|███████▏  | 624/864 [11:42<03:36,  1.11it/s]
 72%|███████▏  | 625/864 [11:42<03:35,  1.11it/s]
 72%|███████▏  | 626/864 [11:43<03:34,  1.11it/s]
 73%|███████▎  | 627/864 [11:44<03:33,  1.11it/s]
 73%|███████▎  | 628/864 [11:45<03:32,  1.11it/s]
 73%|███████▎  | 629/864 [11:46<03:31,  1.11it/s]
 73%|███████▎  | 630/864 [11:47<03:30,  1.11it/s]
                                                 
{'loss': '4.259', 'grad_norm': '0.3251', 'learning_rate': '0.0003', 'epoch': '8.754'}
+
 73%|███████▎  | 630/864 [11:47<03:30,  1.11it/s]
 73%|███████▎  | 631/864 [11:48<03:29,  1.11it/s]
 73%|███████▎  | 632/864 [11:49<03:28,  1.11it/s]
 73%|███████▎  | 633/864 [11:50<03:27,  1.11it/s]
 73%|███████▎  | 634/864 [11:51<03:27,  1.11it/s]
 73%|███████▎  | 635/864 [11:51<03:26,  1.11it/s]
 74%|███████▎  | 636/864 [11:52<03:25,  1.11it/s]
 74%|███████▎  | 637/864 [11:53<03:24,  1.11it/s]
 74%|███████▍  | 638/864 [11:54<03:23,  1.11it/s]
 74%|███████▍  | 639/864 [11:55<03:22,  1.11it/s]
 74%|███████▍  | 640/864 [11:56<03:21,  1.11it/s]
                                                 
{'loss': '4.259', 'grad_norm': '0.5074', 'learning_rate': '0.0003', 'epoch': '8.894'}
+
 74%|███████▍  | 640/864 [11:56<03:21,  1.11it/s]
 74%|███████▍  | 641/864 [11:57<03:20,  1.11it/s]
 74%|███████▍  | 642/864 [11:58<03:19,  1.11it/s]
 74%|███████▍  | 643/864 [11:59<03:18,  1.11it/s]
 75%|███████▍  | 644/864 [12:00<03:17,  1.11it/s]
 75%|███████▍  | 645/864 [12:00<03:17,  1.11it/s]
 75%|███████▍  | 646/864 [12:01<03:16,  1.11it/s]
 75%|███████▍  | 647/864 [12:02<03:15,  1.11it/s]
 75%|███████▌  | 648/864 [12:03<02:52,  1.25it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:01, 19.61it/s]
+
 10%|█         | 4/39 [00:00<00:02, 12.21it/s]
+
 15%|█▌        | 6/39 [00:00<00:03, 10.87it/s]
+
 21%|██        | 8/39 [00:00<00:02, 10.37it/s]
+
 26%|██▌       | 10/39 [00:00<00:02, 10.08it/s]
+
 31%|███       | 12/39 [00:01<00:02,  9.94it/s]
+
 36%|███▌      | 14/39 [00:01<00:02,  9.83it/s]
+
 41%|████      | 16/39 [00:01<00:02,  9.78it/s]
+
 44%|████▎     | 17/39 [00:01<00:02,  9.73it/s]
+
 46%|████▌     | 18/39 [00:01<00:02,  9.72it/s]
+
 49%|████▊     | 19/39 [00:01<00:02,  9.70it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01,  9.69it/s]
+
 54%|█████▍    | 21/39 [00:02<00:01,  9.66it/s]
+
 56%|█████▋    | 22/39 [00:02<00:01,  9.65it/s]
+
 59%|█████▉    | 23/39 [00:02<00:01,  9.68it/s]
+
 62%|██████▏   | 24/39 [00:02<00:01,  9.66it/s]
+
 64%|██████▍   | 25/39 [00:02<00:01,  9.66it/s]
+
 67%|██████▋   | 26/39 [00:02<00:01,  9.66it/s]
+
 69%|██████▉   | 27/39 [00:02<00:01,  9.66it/s]
+
 72%|███████▏  | 28/39 [00:02<00:01,  9.66it/s]
+
 74%|███████▍  | 29/39 [00:02<00:01,  9.66it/s]
+
 77%|███████▋  | 30/39 [00:03<00:00,  9.65it/s]
+
 79%|███████▉  | 31/39 [00:03<00:00,  9.64it/s]
+
 82%|████████▏ | 32/39 [00:03<00:00,  9.64it/s]
+
 85%|████████▍ | 33/39 [00:03<00:00,  9.64it/s]
+
 87%|████████▋ | 34/39 [00:03<00:00,  9.65it/s]
+
 90%|████████▉ | 35/39 [00:03<00:00,  9.65it/s]
+
 92%|█████████▏| 36/39 [00:03<00:00,  9.65it/s]
+
 95%|█████████▍| 37/39 [00:03<00:00,  9.65it/s]
+
 97%|█████████▋| 38/39 [00:03<00:00,  9.64it/s]
+
100%|██████████| 39/39 [00:03<00:00,  9.64it/s]
                                                 
+
                                               
{'eval_loss': '4.417', 'eval_runtime': '4.101', 'eval_samples_per_second': '595.3', 'eval_steps_per_second': '9.511', 'epoch': '9'}
+
 75%|███████▌  | 648/864 [12:07<02:52,  1.25it/s]
+
100%|██████████| 39/39 [00:04<00:00,  9.64it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-648
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-648/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-648/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.11it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.10it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-648/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-648/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-648/tokenizer_config.json
+[eval] epoch 9.0 step 648 loss 4.4167 (best 4.4167 @ ep 9.0, 0/10000 since best)
+
+
                                               
 75%|███████▌  | 649/864 [12:11<11:06,  3.10s/it]
 75%|███████▌  | 650/864 [12:12<08:42,  2.44s/it]
                                                 
{'loss': '4.231', 'grad_norm': '0.5689', 'learning_rate': '0.0003', 'epoch': '9.028'}
+
 75%|███████▌  | 650/864 [12:12<08:42,  2.44s/it]
 75%|███████▌  | 651/864 [12:13<07:01,  1.98s/it]
 75%|███████▌  | 652/864 [12:14<05:50,  1.65s/it]
 76%|███████▌  | 653/864 [12:15<05:01,  1.43s/it]
 76%|███████▌  | 654/864 [12:16<04:26,  1.27s/it]
 76%|███████▌  | 655/864 [12:17<04:02,  1.16s/it]
 76%|███████▌  | 656/864 [12:18<03:45,  1.08s/it]
 76%|███████▌  | 657/864 [12:18<03:32,  1.03s/it]
 76%|███████▌  | 658/864 [12:19<03:23,  1.01it/s]
 76%|███████▋  | 659/864 [12:20<03:17,  1.04it/s]
 76%|███████▋  | 660/864 [12:21<03:12,  1.06it/s]
                                                 
{'loss': '4.191', 'grad_norm': '0.979', 'learning_rate': '0.0003', 'epoch': '9.168'}
+
 76%|███████▋  | 660/864 [12:21<03:12,  1.06it/s]
 77%|███████▋  | 661/864 [12:22<03:09,  1.07it/s]
 77%|███████▋  | 662/864 [12:23<03:06,  1.08it/s]
 77%|███████▋  | 663/864 [12:24<03:04,  1.09it/s]
 77%|███████▋  | 664/864 [12:25<03:02,  1.10it/s]
 77%|███████▋  | 665/864 [12:26<03:00,  1.10it/s]
 77%|███████▋  | 666/864 [12:27<02:59,  1.11it/s]
 77%|███████▋  | 667/864 [12:27<02:57,  1.11it/s]
 77%|███████▋  | 668/864 [12:28<02:56,  1.11it/s]
 77%|███████▋  | 669/864 [12:29<02:55,  1.11it/s]
 78%|███████▊  | 670/864 [12:30<02:54,  1.11it/s]
                                                 
{'loss': '4.21', 'grad_norm': '0.8917', 'learning_rate': '0.0003', 'epoch': '9.307'}
+
 78%|███████▊  | 670/864 [12:30<02:54,  1.11it/s]
 78%|███████▊  | 671/864 [12:31<02:53,  1.11it/s]
 78%|███████▊  | 672/864 [12:32<02:52,  1.11it/s]
 78%|███████▊  | 673/864 [12:33<02:51,  1.11it/s]
 78%|███████▊  | 674/864 [12:34<02:50,  1.11it/s]
 78%|███████▊  | 675/864 [12:35<02:49,  1.11it/s]
 78%|███████▊  | 676/864 [12:36<02:49,  1.11it/s]
 78%|███████▊  | 677/864 [12:36<02:48,  1.11it/s]
 78%|███████▊  | 678/864 [12:37<02:47,  1.11it/s]
 79%|███████▊  | 679/864 [12:38<02:46,  1.11it/s]
 79%|███████▊  | 680/864 [12:39<02:45,  1.11it/s]
                                                 
{'loss': '4.184', 'grad_norm': '0.5174', 'learning_rate': '0.0003', 'epoch': '9.447'}
+
 79%|███████▊  | 680/864 [12:39<02:45,  1.11it/s]
 79%|███████▉  | 681/864 [12:40<02:44,  1.11it/s]
 79%|███████▉  | 682/864 [12:41<02:43,  1.11it/s]
 79%|███████▉  | 683/864 [12:42<02:42,  1.11it/s]
 79%|███████▉  | 684/864 [12:43<02:41,  1.11it/s]
 79%|███████▉  | 685/864 [12:44<02:41,  1.11it/s]
 79%|███████▉  | 686/864 [12:45<02:40,  1.11it/s]
 80%|███████▉  | 687/864 [12:45<02:39,  1.11it/s]
 80%|███████▉  | 688/864 [12:46<02:38,  1.11it/s]
 80%|███████▉  | 689/864 [12:47<02:37,  1.11it/s]
 80%|███████▉  | 690/864 [12:48<02:36,  1.11it/s]
                                                 
{'loss': '4.148', 'grad_norm': '0.3425', 'learning_rate': '0.0003', 'epoch': '9.586'}
+
 80%|███████▉  | 690/864 [12:48<02:36,  1.11it/s]
 80%|███████▉  | 691/864 [12:49<02:35,  1.11it/s]
 80%|████████  | 692/864 [12:50<02:34,  1.11it/s]
 80%|████████  | 693/864 [12:51<02:33,  1.11it/s]
 80%|████████  | 694/864 [12:52<02:32,  1.11it/s]
 80%|████████  | 695/864 [12:53<02:31,  1.11it/s]
 81%|████████  | 696/864 [12:54<02:31,  1.11it/s]
 81%|████████  | 697/864 [12:54<02:30,  1.11it/s]
 81%|████████  | 698/864 [12:55<02:29,  1.11it/s]
 81%|████████  | 699/864 [12:56<02:28,  1.11it/s]
 81%|████████  | 700/864 [12:57<02:27,  1.11it/s]
                                                 
{'loss': '4.14', 'grad_norm': '0.4218', 'learning_rate': '0.0003', 'epoch': '9.726'}
+
 81%|████████  | 700/864 [12:57<02:27,  1.11it/s]
 81%|████████  | 701/864 [12:58<02:26,  1.11it/s]
 81%|████████▏ | 702/864 [12:59<02:25,  1.11it/s]
 81%|████████▏ | 703/864 [13:00<02:24,  1.11it/s]
 81%|████████▏ | 704/864 [13:01<02:23,  1.11it/s]
 82%|████████▏ | 705/864 [13:02<02:22,  1.11it/s]
 82%|████████▏ | 706/864 [13:03<02:22,  1.11it/s]
 82%|████████▏ | 707/864 [13:03<02:21,  1.11it/s]
 82%|████████▏ | 708/864 [13:04<02:20,  1.11it/s]
 82%|████████▏ | 709/864 [13:05<02:19,  1.11it/s]
 82%|████████▏ | 710/864 [13:06<02:18,  1.11it/s]
                                                 
{'loss': '4.119', 'grad_norm': '0.5048', 'learning_rate': '0.0003', 'epoch': '9.866'}
+
 82%|████████▏ | 710/864 [13:06<02:18,  1.11it/s]
 82%|████████▏ | 711/864 [13:07<02:17,  1.11it/s]
 82%|████████▏ | 712/864 [13:08<02:16,  1.11it/s]
 83%|████████▎ | 713/864 [13:09<02:15,  1.11it/s]
 83%|████████▎ | 714/864 [13:10<02:14,  1.11it/s]
 83%|████████▎ | 715/864 [13:11<02:14,  1.11it/s]
 83%|████████▎ | 716/864 [13:12<02:13,  1.11it/s]
 83%|████████▎ | 717/864 [13:12<02:12,  1.11it/s]
 83%|████████▎ | 718/864 [13:13<02:11,  1.11it/s]
 83%|████████▎ | 719/864 [13:14<02:10,  1.11it/s]
 83%|████████▎ | 720/864 [13:15<01:55,  1.25it/s]
                                                 
{'loss': '4.104', 'grad_norm': '0.5721', 'learning_rate': '0.0003', 'epoch': '10'}
+
 83%|████████▎ | 720/864 [13:15<01:55,  1.25it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:01, 19.49it/s]
+
 10%|█         | 4/39 [00:00<00:02, 11.96it/s]
+
 15%|█▌        | 6/39 [00:00<00:03, 10.90it/s]
+
 21%|██        | 8/39 [00:00<00:03, 10.26it/s]
+
 26%|██▌       | 10/39 [00:00<00:02, 10.12it/s]
+
 31%|███       | 12/39 [00:01<00:02,  9.87it/s]
+
 36%|███▌      | 14/39 [00:01<00:02,  9.86it/s]
+
 41%|████      | 16/39 [00:01<00:02,  9.72it/s]
+
 46%|████▌     | 18/39 [00:01<00:02,  9.76it/s]
+
 49%|████▊     | 19/39 [00:01<00:02,  9.74it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01,  9.62it/s]
+
 56%|█████▋    | 22/39 [00:02<00:01,  9.71it/s]
+
 59%|█████▉    | 23/39 [00:02<00:01,  9.71it/s]
+
 62%|██████▏   | 24/39 [00:02<00:01,  9.70it/s]
+
 64%|██████▍   | 25/39 [00:02<00:01,  9.69it/s]
+
 67%|██████▋   | 26/39 [00:02<00:01,  9.68it/s]
+
 69%|██████▉   | 27/39 [00:02<00:01,  9.67it/s]
+
 72%|███████▏  | 28/39 [00:02<00:01,  9.67it/s]
+
 74%|███████▍  | 29/39 [00:02<00:01,  9.66it/s]
+
 77%|███████▋  | 30/39 [00:03<00:00,  9.66it/s]
+
 79%|███████▉  | 31/39 [00:03<00:00,  9.66it/s]
+
 82%|████████▏ | 32/39 [00:03<00:00,  9.66it/s]
+
 85%|████████▍ | 33/39 [00:03<00:00,  9.65it/s]
+
 87%|████████▋ | 34/39 [00:03<00:00,  9.66it/s]
+
 90%|████████▉ | 35/39 [00:03<00:00,  9.66it/s]
+
 92%|█████████▏| 36/39 [00:03<00:00,  9.66it/s]
+
 95%|█████████▍| 37/39 [00:03<00:00,  9.66it/s]
+
 97%|█████████▋| 38/39 [00:03<00:00,  9.66it/s]
+
100%|██████████| 39/39 [00:03<00:00,  9.66it/s]
                                                 
+
                                               
{'eval_loss': '4.333', 'eval_runtime': '4.115', 'eval_samples_per_second': '593.2', 'eval_steps_per_second': '9.478', 'epoch': '10'}
+
 83%|████████▎ | 720/864 [13:19<01:55,  1.25it/s]
+
100%|██████████| 39/39 [00:04<00:00,  9.66it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-720
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-720/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-720/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.12it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.12it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-720/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-720/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-720/tokenizer_config.json
+[eval] epoch 10.0 step 720 loss 4.3331 (best 4.3331 @ ep 10.0, 0/10000 since best)
+
+
                                               
 83%|████████▎ | 721/864 [13:23<07:23,  3.10s/it]
 84%|████████▎ | 722/864 [13:24<05:46,  2.44s/it]
 84%|████████▎ | 723/864 [13:25<04:39,  1.98s/it]
 84%|████████▍ | 724/864 [13:26<03:51,  1.66s/it]
 84%|████████▍ | 725/864 [13:27<03:18,  1.43s/it]
 84%|████████▍ | 726/864 [13:28<02:55,  1.27s/it]
 84%|████████▍ | 727/864 [13:29<02:38,  1.16s/it]
 84%|████████▍ | 728/864 [13:30<02:26,  1.08s/it]
 84%|████████▍ | 729/864 [13:30<02:18,  1.03s/it]
 84%|████████▍ | 730/864 [13:31<02:12,  1.01it/s]
                                                 
{'loss': '4.051', 'grad_norm': '0.8413', 'learning_rate': '0.0003', 'epoch': '10.14'}
+
 84%|████████▍ | 730/864 [13:31<02:12,  1.01it/s]
 85%|████████▍ | 731/864 [13:32<02:07,  1.04it/s]
 85%|████████▍ | 732/864 [13:33<02:04,  1.06it/s]
 85%|████████▍ | 733/864 [13:34<02:01,  1.08it/s]
 85%|████████▍ | 734/864 [13:35<01:59,  1.09it/s]
 85%|████████▌ | 735/864 [13:36<01:57,  1.09it/s]
 85%|████████▌ | 736/864 [13:37<01:56,  1.10it/s]
 85%|████████▌ | 737/864 [13:38<01:55,  1.10it/s]
 85%|████████▌ | 738/864 [13:39<01:54,  1.10it/s]
 86%|████████▌ | 739/864 [13:39<01:53,  1.11it/s]
 86%|████████▌ | 740/864 [13:40<01:51,  1.11it/s]
                                                 
{'loss': '4.063', 'grad_norm': '0.8539', 'learning_rate': '0.0003', 'epoch': '10.28'}
+
 86%|████████▌ | 740/864 [13:40<01:51,  1.11it/s]
 86%|████████▌ | 741/864 [13:41<01:50,  1.11it/s]
 86%|████████▌ | 742/864 [13:42<01:49,  1.11it/s]
 86%|████████▌ | 743/864 [13:43<01:48,  1.11it/s]
 86%|████████▌ | 744/864 [13:44<01:47,  1.11it/s]
 86%|████████▌ | 745/864 [13:45<01:46,  1.11it/s]
 86%|████████▋ | 746/864 [13:46<01:46,  1.11it/s]
 86%|████████▋ | 747/864 [13:47<01:45,  1.11it/s]
 87%|████████▋ | 748/864 [13:48<01:44,  1.11it/s]
 87%|████████▋ | 749/864 [13:48<01:43,  1.11it/s]
 87%|████████▋ | 750/864 [13:49<01:42,  1.11it/s]
                                                 
{'loss': '4.045', 'grad_norm': '0.5238', 'learning_rate': '0.0003', 'epoch': '10.42'}
+
 87%|████████▋ | 750/864 [13:49<01:42,  1.11it/s]
 87%|████████▋ | 751/864 [13:50<01:41,  1.11it/s]
 87%|████████▋ | 752/864 [13:51<01:40,  1.11it/s]
 87%|████████▋ | 753/864 [13:52<01:39,  1.11it/s]
 87%|████████▋ | 754/864 [13:53<01:38,  1.11it/s]
 87%|████████▋ | 755/864 [13:54<01:38,  1.11it/s]
 88%|████████▊ | 756/864 [13:55<01:37,  1.11it/s]
 88%|████████▊ | 757/864 [13:56<01:36,  1.11it/s]
 88%|████████▊ | 758/864 [13:57<01:35,  1.11it/s]
 88%|████████▊ | 759/864 [13:57<01:34,  1.11it/s]
 88%|████████▊ | 760/864 [13:58<01:33,  1.11it/s]
                                                 
{'loss': '4.044', 'grad_norm': '0.3833', 'learning_rate': '0.0003', 'epoch': '10.56'}
+
 88%|████████▊ | 760/864 [13:58<01:33,  1.11it/s]
 88%|████████▊ | 761/864 [13:59<01:32,  1.11it/s]
 88%|████████▊ | 762/864 [14:00<01:31,  1.11it/s]
 88%|████████▊ | 763/864 [14:01<01:30,  1.11it/s]
 88%|████████▊ | 764/864 [14:02<01:30,  1.11it/s]
 89%|████████▊ | 765/864 [14:03<01:29,  1.11it/s]
 89%|████████▊ | 766/864 [14:04<01:28,  1.11it/s]
 89%|████████▉ | 767/864 [14:05<01:27,  1.11it/s]
 89%|████████▉ | 768/864 [14:06<01:26,  1.11it/s]
 89%|████████▉ | 769/864 [14:06<01:25,  1.11it/s]
 89%|████████▉ | 770/864 [14:07<01:24,  1.11it/s]
                                                 
{'loss': '4.02', 'grad_norm': '0.4053', 'learning_rate': '0.0003', 'epoch': '10.7'}
+
 89%|████████▉ | 770/864 [14:07<01:24,  1.11it/s]
 89%|████████▉ | 771/864 [14:08<01:23,  1.11it/s]
 89%|████████▉ | 772/864 [14:09<01:22,  1.11it/s]
 89%|████████▉ | 773/864 [14:10<01:21,  1.11it/s]
 90%|████████▉ | 774/864 [14:11<01:20,  1.11it/s]
 90%|████████▉ | 775/864 [14:12<01:20,  1.11it/s]
 90%|████████▉ | 776/864 [14:13<01:19,  1.11it/s]
 90%|████████▉ | 777/864 [14:14<01:18,  1.11it/s]
 90%|█████████ | 778/864 [14:15<01:17,  1.11it/s]
 90%|█████████ | 779/864 [14:15<01:16,  1.11it/s]
 90%|█████████ | 780/864 [14:16<01:15,  1.11it/s]
                                                 
{'loss': '4.012', 'grad_norm': '0.4681', 'learning_rate': '0.0003', 'epoch': '10.84'}
+
 90%|█████████ | 780/864 [14:16<01:15,  1.11it/s]
 90%|█████████ | 781/864 [14:17<01:15,  1.11it/s]
 91%|█████████ | 782/864 [14:18<01:14,  1.11it/s]
 91%|█████████ | 783/864 [14:19<01:13,  1.11it/s]
 91%|█████████ | 784/864 [14:20<01:12,  1.11it/s]
 91%|█████████ | 785/864 [14:21<01:11,  1.11it/s]
 91%|█████████ | 786/864 [14:22<01:10,  1.11it/s]
 91%|█████████ | 787/864 [14:23<01:09,  1.11it/s]
 91%|█████████ | 788/864 [14:24<01:08,  1.11it/s]
 91%|█████████▏| 789/864 [14:24<01:07,  1.11it/s]
 91%|█████████▏| 790/864 [14:25<01:06,  1.11it/s]
                                                 
{'loss': '3.999', 'grad_norm': '0.4528', 'learning_rate': '0.0003', 'epoch': '10.98'}
+
 91%|█████████▏| 790/864 [14:25<01:06,  1.11it/s]
 92%|█████████▏| 791/864 [14:26<01:05,  1.11it/s]
 92%|█████████▏| 792/864 [14:27<00:57,  1.25it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:02, 14.91it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 11.57it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 10.82it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 10.26it/s]
+
 28%|██▊       | 11/39 [00:01<00:02, 10.12it/s]
+
 33%|███▎      | 13/39 [00:01<00:02,  9.89it/s]
+
 38%|███▊      | 15/39 [00:01<00:02,  9.89it/s]
+
 41%|████      | 16/39 [00:01<00:02,  9.84it/s]
+
 44%|████▎     | 17/39 [00:01<00:02,  9.70it/s]
+
 49%|████▊     | 19/39 [00:01<00:02,  9.76it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01,  9.73it/s]
+
 54%|█████▍    | 21/39 [00:02<00:01,  9.61it/s]
+
 59%|█████▉    | 23/39 [00:02<00:01,  9.73it/s]
+
 62%|██████▏   | 24/39 [00:02<00:01,  9.71it/s]
+
 64%|██████▍   | 25/39 [00:02<00:01,  9.69it/s]
+
 67%|██████▋   | 26/39 [00:02<00:01,  9.68it/s]
+
 69%|██████▉   | 27/39 [00:02<00:01,  9.67it/s]
+
 72%|███████▏  | 28/39 [00:02<00:01,  9.66it/s]
+
 74%|███████▍  | 29/39 [00:02<00:01,  9.67it/s]
+
 77%|███████▋  | 30/39 [00:03<00:00,  9.67it/s]
+
 79%|███████▉  | 31/39 [00:03<00:00,  9.67it/s]
+
 82%|████████▏ | 32/39 [00:03<00:00,  9.67it/s]
+
 85%|████████▍ | 33/39 [00:03<00:00,  9.66it/s]
+
 87%|████████▋ | 34/39 [00:03<00:00,  9.67it/s]
+
 90%|████████▉ | 35/39 [00:03<00:00,  9.66it/s]
+
 92%|█████████▏| 36/39 [00:03<00:00,  9.65it/s]
+
 95%|█████████▍| 37/39 [00:03<00:00,  9.65it/s]
+
 97%|█████████▋| 38/39 [00:03<00:00,  9.65it/s]
+
100%|██████████| 39/39 [00:03<00:00,  9.65it/s]
                                                 
+
                                               
{'eval_loss': '4.256', 'eval_runtime': '4.098', 'eval_samples_per_second': '595.6', 'eval_steps_per_second': '9.516', 'epoch': '11'}
+
 92%|█████████▏| 792/864 [14:31<00:57,  1.25it/s]
+
100%|██████████| 39/39 [00:04<00:00,  9.65it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-792
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-792/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-792/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:01<00:00,  1.03s/it]
Writing model shards: 100%|██████████| 1/1 [00:01<00:00,  1.03s/it]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-792/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-792/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-792/tokenizer_config.json
+[eval] epoch 11.0 step 792 loss 4.2558 (best 4.2558 @ ep 11.0, 0/10000 since best)
+
+
                                               
 92%|█████████▏| 793/864 [14:35<03:42,  3.14s/it]
 92%|█████████▏| 794/864 [14:36<02:52,  2.47s/it]
 92%|█████████▏| 795/864 [14:37<02:17,  2.00s/it]
 92%|█████████▏| 796/864 [14:38<01:53,  1.67s/it]
 92%|█████████▏| 797/864 [14:39<01:36,  1.44s/it]
 92%|█████████▏| 798/864 [14:40<01:24,  1.28s/it]
 92%|█████████▏| 799/864 [14:41<01:15,  1.16s/it]
 93%|█████████▎| 800/864 [14:42<01:09,  1.08s/it]
                                                 
{'loss': '3.957', 'grad_norm': '0.8122', 'learning_rate': '0.0003', 'epoch': '11.11'}
+
 93%|█████████▎| 800/864 [14:42<01:09,  1.08s/it]
 93%|█████████▎| 801/864 [14:43<01:04,  1.03s/it]
 93%|█████████▎| 802/864 [14:44<01:01,  1.01it/s]
 93%|█████████▎| 803/864 [14:44<00:58,  1.04it/s]
 93%|█████████▎| 804/864 [14:45<00:56,  1.06it/s]
 93%|█████████▎| 805/864 [14:46<00:54,  1.07it/s]
 93%|█████████▎| 806/864 [14:47<00:53,  1.08it/s]
 93%|█████████▎| 807/864 [14:48<00:52,  1.09it/s]
 94%|█████████▎| 808/864 [14:49<00:50,  1.10it/s]
 94%|█████████▎| 809/864 [14:50<00:49,  1.10it/s]
 94%|█████████▍| 810/864 [14:51<00:48,  1.11it/s]
                                                 
{'loss': '3.925', 'grad_norm': '0.7589', 'learning_rate': '0.0003', 'epoch': '11.25'}
+
 94%|█████████▍| 810/864 [14:51<00:48,  1.11it/s]
 94%|█████████▍| 811/864 [14:52<00:47,  1.11it/s]
 94%|█████████▍| 812/864 [14:53<00:46,  1.11it/s]
 94%|█████████▍| 813/864 [14:53<00:46,  1.11it/s]
 94%|█████████▍| 814/864 [14:54<00:45,  1.11it/s]
 94%|█████████▍| 815/864 [14:55<00:44,  1.11it/s]
 94%|█████████▍| 816/864 [14:56<00:43,  1.11it/s]
 95%|█████████▍| 817/864 [14:57<00:42,  1.11it/s]
 95%|█████████▍| 818/864 [14:58<00:41,  1.11it/s]
 95%|█████████▍| 819/864 [14:59<00:40,  1.11it/s]
 95%|█████████▍| 820/864 [15:00<00:39,  1.11it/s]
                                                 
{'loss': '3.928', 'grad_norm': '0.6505', 'learning_rate': '0.0003', 'epoch': '11.39'}
+
 95%|█████████▍| 820/864 [15:00<00:39,  1.11it/s]
 95%|█████████▌| 821/864 [15:01<00:38,  1.11it/s]
 95%|█████████▌| 822/864 [15:02<00:37,  1.11it/s]
 95%|█████████▌| 823/864 [15:02<00:36,  1.11it/s]
 95%|█████████▌| 824/864 [15:03<00:35,  1.11it/s]
 95%|█████████▌| 825/864 [15:04<00:35,  1.11it/s]
 96%|█████████▌| 826/864 [15:05<00:34,  1.11it/s]
 96%|█████████▌| 827/864 [15:06<00:33,  1.11it/s]
 96%|█████████▌| 828/864 [15:07<00:32,  1.11it/s]
 96%|█████████▌| 829/864 [15:08<00:31,  1.11it/s]
 96%|█████████▌| 830/864 [15:09<00:30,  1.11it/s]
                                                 
{'loss': '3.928', 'grad_norm': '0.4768', 'learning_rate': '0.0003', 'epoch': '11.53'}
+
 96%|█████████▌| 830/864 [15:09<00:30,  1.11it/s]
 96%|█████████▌| 831/864 [15:10<00:29,  1.11it/s]
 96%|█████████▋| 832/864 [15:11<00:28,  1.11it/s]
 96%|█████████▋| 833/864 [15:11<00:27,  1.11it/s]
 97%|█████████▋| 834/864 [15:12<00:26,  1.11it/s]
 97%|█████████▋| 835/864 [15:13<00:26,  1.11it/s]
 97%|█████████▋| 836/864 [15:14<00:25,  1.11it/s]
 97%|█████████▋| 837/864 [15:15<00:24,  1.11it/s]
 97%|█████████▋| 838/864 [15:16<00:23,  1.11it/s]
 97%|█████████▋| 839/864 [15:17<00:22,  1.11it/s]
 97%|█████████▋| 840/864 [15:18<00:21,  1.11it/s]
                                                 
{'loss': '3.917', 'grad_norm': '0.4262', 'learning_rate': '0.0003', 'epoch': '11.67'}
+
 97%|█████████▋| 840/864 [15:18<00:21,  1.11it/s]
 97%|█████████▋| 841/864 [15:19<00:20,  1.11it/s]
 97%|█████████▋| 842/864 [15:20<00:19,  1.11it/s]
 98%|█████████▊| 843/864 [15:20<00:18,  1.11it/s]
 98%|█████████▊| 844/864 [15:21<00:17,  1.11it/s]
 98%|█████████▊| 845/864 [15:22<00:17,  1.11it/s]
 98%|█████████▊| 846/864 [15:23<00:16,  1.11it/s]
 98%|█████████▊| 847/864 [15:24<00:15,  1.11it/s]
 98%|█████████▊| 848/864 [15:25<00:14,  1.11it/s]
 98%|█████████▊| 849/864 [15:26<00:13,  1.11it/s]
 98%|█████████▊| 850/864 [15:27<00:12,  1.11it/s]
                                                 
{'loss': '3.907', 'grad_norm': '0.4379', 'learning_rate': '0.0003', 'epoch': '11.81'}
+
 98%|█████████▊| 850/864 [15:27<00:12,  1.11it/s]
 98%|█████████▊| 851/864 [15:28<00:11,  1.11it/s]
 99%|█████████▊| 852/864 [15:28<00:10,  1.11it/s]
 99%|█████████▊| 853/864 [15:29<00:09,  1.11it/s]
 99%|█████████▉| 854/864 [15:30<00:09,  1.11it/s]
 99%|█████████▉| 855/864 [15:31<00:08,  1.11it/s]
 99%|█████████▉| 856/864 [15:32<00:07,  1.11it/s]
 99%|█████████▉| 857/864 [15:33<00:06,  1.11it/s]
 99%|█████████▉| 858/864 [15:34<00:05,  1.11it/s]
 99%|█████████▉| 859/864 [15:35<00:04,  1.11it/s]
100%|█████████▉| 860/864 [15:36<00:03,  1.11it/s]
                                                 
{'loss': '3.906', 'grad_norm': '0.3658', 'learning_rate': '0.0003', 'epoch': '11.95'}
+
100%|█████████▉| 860/864 [15:36<00:03,  1.11it/s]
100%|█████████▉| 861/864 [15:37<00:02,  1.11it/s]
100%|█████████▉| 862/864 [15:37<00:01,  1.11it/s]
100%|█████████▉| 863/864 [15:38<00:00,  1.11it/s]
100%|██████████| 864/864 [15:39<00:00,  1.25it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:02, 14.80it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 11.71it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 10.77it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 10.30it/s]
+
 28%|██▊       | 11/39 [00:01<00:02, 10.08it/s]
+
 33%|███▎      | 13/39 [00:01<00:02,  9.93it/s]
+
 36%|███▌      | 14/39 [00:01<00:02,  9.87it/s]
+
 38%|███▊      | 15/39 [00:01<00:02,  9.82it/s]
+
 41%|████      | 16/39 [00:01<00:02,  9.68it/s]
+
 46%|████▌     | 18/39 [00:01<00:02,  9.75it/s]
+
 49%|████▊     | 19/39 [00:01<00:02,  9.72it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01,  9.60it/s]
+
 56%|█████▋    | 22/39 [00:02<00:01,  9.70it/s]
+
 59%|█████▉    | 23/39 [00:02<00:01,  9.70it/s]
+
 62%|██████▏   | 24/39 [00:02<00:01,  9.69it/s]
+
 64%|██████▍   | 25/39 [00:02<00:01,  9.69it/s]
+
 67%|██████▋   | 26/39 [00:02<00:01,  9.68it/s]
+
 69%|██████▉   | 27/39 [00:02<00:01,  9.67it/s]
+
 72%|███████▏  | 28/39 [00:02<00:01,  9.67it/s]
+
 74%|███████▍  | 29/39 [00:02<00:01,  9.67it/s]
+
 77%|███████▋  | 30/39 [00:03<00:00,  9.67it/s]
+
 79%|███████▉  | 31/39 [00:03<00:00,  9.66it/s]
+
 82%|████████▏ | 32/39 [00:03<00:00,  9.66it/s]
+
 85%|████████▍ | 33/39 [00:03<00:00,  9.66it/s]
+
 87%|████████▋ | 34/39 [00:03<00:00,  9.66it/s]
+
 90%|████████▉ | 35/39 [00:03<00:00,  9.65it/s]
+
 92%|█████████▏| 36/39 [00:03<00:00,  9.66it/s]
+
 95%|█████████▍| 37/39 [00:03<00:00,  9.66it/s]
+
 97%|█████████▋| 38/39 [00:03<00:00,  9.65it/s]
+
100%|██████████| 39/39 [00:03<00:00,  9.65it/s]
                                                 
+
                                               
{'eval_loss': '4.19', 'eval_runtime': '4.097', 'eval_samples_per_second': '595.8', 'eval_steps_per_second': '9.519', 'epoch': '12'}
+
100%|██████████| 864/864 [15:43<00:00,  1.25it/s]
+
100%|██████████| 39/39 [00:04<00:00,  9.65it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-864
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-864/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-864/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.13it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.13it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-864/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-864/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0003/checkpoint-864/tokenizer_config.json
+[eval] epoch 12.0 step 864 loss 4.1902 (best 4.1902 @ ep 12.0, 0/10000 since best)
+
+
                                               [transformers] 
+
+Training completed. Do not forget to share your model on huggingface.co/models =)
+
+
+
                                                 
{'train_runtime': '948.5', 'train_samples_per_second': '463.2', 'train_steps_per_second': '0.911', 'train_loss': '5.308', 'epoch': '12'}
+
100%|██████████| 864/864 [15:46<00:00,  1.25it/s]
100%|██████████| 864/864 [15:46<00:00,  1.10s/it]
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+wandb: 
+wandb: 🚀 View run t0_S3_lr0.0003 at: https://wandb.ai/slowrun-wankers/data-aug-pretraining/runs/lwba0vwf
+wandb: Find logs at: wandb/run-20260817_193302-lwba0vwf/logs
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+[rank0]:[W817 19:49:35.128554285 ProcessGroupNCCL.cpp:1575] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator())
diff --git a/runs/t0/t0_S3_lr0.0006/checkpoint-864/generation_config.json b/runs/t0/t0_S3_lr0.0006/checkpoint-864/generation_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..8c4242f32d40df256857f87264f646130f684e33
--- /dev/null
+++ b/runs/t0/t0_S3_lr0.0006/checkpoint-864/generation_config.json
@@ -0,0 +1,11 @@
+{
+  "_from_model_config": true,
+  "eos_token_id": [
+    151645
+  ],
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "transformers_version": "5.15.0",
+  "use_cache": true
+}
diff --git a/runs/t0/t0_S3_lr0.0006/train.log b/runs/t0/t0_S3_lr0.0006/train.log
new file mode 100644
index 0000000000000000000000000000000000000000..4a0e4fae7b238a3fe8de37718f1814ccd30760e5
--- /dev/null
+++ b/runs/t0/t0_S3_lr0.0006/train.log
@@ -0,0 +1,1603 @@
+W0817 19:49:38.740000 2227006 torch/distributed/run.py:851] 
+W0817 19:49:38.740000 2227006 torch/distributed/run.py:851] *****************************************
+W0817 19:49:38.740000 2227006 torch/distributed/run.py:851] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+W0817 19:49:38.740000 2227006 torch/distributed/run.py:851] *****************************************
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+EOS token: <|im_end|>
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+total params: 258072576
+[dataset] loaded packed cache: 36614 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+total params: 258072576
+[dataset] loaded packed cache: 36614 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+total params: 258072576
+[dataset] loaded packed cache: 36614 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+total params: 258072576
+[dataset] loaded packed cache: 36614 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+total params: 258072576
+[dataset] loaded packed cache: 36614 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+total params: 258072576
+[dataset] loaded packed cache: 36614 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+total params: 258072576
+[dataset] loaded packed cache: 36614 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+total params: 258072576
+[dataset] loaded packed cache: 36614 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+[transformers] ***** Running training *****
+[transformers]   Num examples = 36,614
+[transformers]   Num Epochs = 12
+[transformers]   Num update steps per epoch = 72
+[transformers]   Instantaneous batch size per device = 8
+[transformers]   Total train batch size (w. parallel, distributed & accumulation) = 512
+[transformers]   Gradient Accumulation steps = 8
+[transformers]   Total optimization steps = 864
+[transformers]   Number of trainable parameters = 258,072,576
+wandb: WARNING Changes to your `wandb` environment variables will be ignored because your `wandb` session has already started. For more information on how to modify your settings with `wandb.init()` arguments, please refer to https://wandb.me/wandb-init.
+wandb: setting up run 0fm7mass
+wandb: Tracking run with wandb version 0.28.2
+wandb: Run data is saved locally in /home/ubuntu/workspace/data-augmentations-for-pretraining/wandb/run-20260817_195000-0fm7mass
+wandb: Run `wandb offline` to turn off syncing.
+wandb: Syncing run t0_S3_lr0.0006
+wandb: ⭐️ View project at https://wandb.ai/slowrun-wankers/data-aug-pretraining
+wandb: 🚀 View run at https://wandb.ai/slowrun-wankers/data-aug-pretraining/runs/0fm7mass
+
  0%|          | 0/864 [00:00<?, ?it/s]wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+[rank1]:W0817 19:50:23.456000 2227113 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+[rank3]:W0817 19:50:23.661000 2227115 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+[rank2]:W0817 19:50:23.955000 2227114 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+[rank4]:W0817 19:50:24.204000 2227116 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+[rank6]:W0817 19:50:24.253000 2227118 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+[rank7]:W0817 19:50:24.295000 2227119 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+[rank5]:W0817 19:50:24.375000 2227117 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+[rank0]:W0817 19:50:25.745000 2227112 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+
  0%|          | 1/864 [00:41<10:01:34, 41.82s/it]
                                                  
{'loss': '12.06', 'grad_norm': '2.611', 'learning_rate': '0', 'epoch': '0.01396'}
+
  0%|          | 1/864 [00:41<10:01:34, 41.82s/it]
  0%|          | 2/864 [00:42<4:15:00, 17.75s/it] 
  0%|          | 3/864 [00:43<2:24:13, 10.05s/it]
  0%|          | 4/864 [00:44<1:32:14,  6.44s/it]
  1%|          | 5/864 [00:45<1:03:30,  4.44s/it]
  1%|          | 6/864 [00:46<46:12,  3.23s/it]  
  1%|          | 7/864 [00:47<35:13,  2.47s/it]
  1%|          | 8/864 [00:48<28:02,  1.97s/it]
  1%|          | 9/864 [00:48<23:12,  1.63s/it]
  1%|          | 10/864 [00:49<19:55,  1.40s/it]
                                                
{'loss': '11.51', 'grad_norm': '2.056', 'learning_rate': '5.4e-05', 'epoch': '0.1396'}
+
  1%|          | 10/864 [00:49<19:55,  1.40s/it]
  1%|▏         | 11/864 [00:50<17:41,  1.24s/it]
  1%|▏         | 12/864 [00:51<16:08,  1.14s/it]
  2%|▏         | 13/864 [00:52<15:03,  1.06s/it]
  2%|▏         | 14/864 [00:53<14:18,  1.01s/it]
  2%|▏         | 15/864 [00:54<13:47,  1.03it/s]
  2%|▏         | 16/864 [00:55<13:24,  1.05it/s]
  2%|▏         | 17/864 [00:56<13:08,  1.07it/s]
  2%|▏         | 18/864 [00:56<12:56,  1.09it/s]
  2%|▏         | 19/864 [00:57<12:48,  1.10it/s]
  2%|▏         | 20/864 [00:58<12:41,  1.11it/s]
                                                
{'loss': '10.46', 'grad_norm': '1.888', 'learning_rate': '0.000114', 'epoch': '0.2792'}
+
  2%|▏         | 20/864 [00:58<12:41,  1.11it/s]
  2%|▏         | 21/864 [00:59<12:37,  1.11it/s]
  3%|▎         | 22/864 [01:00<12:34,  1.12it/s]
  3%|▎         | 23/864 [01:01<12:30,  1.12it/s]
  3%|▎         | 24/864 [01:02<12:28,  1.12it/s]
  3%|▎         | 25/864 [01:03<12:25,  1.12it/s]
  3%|▎         | 26/864 [01:04<12:24,  1.13it/s]
  3%|▎         | 27/864 [01:04<12:22,  1.13it/s]
  3%|▎         | 28/864 [01:05<12:21,  1.13it/s]
  3%|▎         | 29/864 [01:06<12:20,  1.13it/s]
  3%|▎         | 30/864 [01:07<12:19,  1.13it/s]
                                                
{'loss': '9.403', 'grad_norm': '1.524', 'learning_rate': '0.000174', 'epoch': '0.4188'}
+
  3%|▎         | 30/864 [01:07<12:19,  1.13it/s]
  4%|▎         | 31/864 [01:08<12:18,  1.13it/s]
  4%|▎         | 32/864 [01:09<12:17,  1.13it/s]
  4%|▍         | 33/864 [01:10<12:16,  1.13it/s]
  4%|▍         | 34/864 [01:11<12:15,  1.13it/s]
  4%|▍         | 35/864 [01:12<12:15,  1.13it/s]
  4%|▍         | 36/864 [01:12<12:14,  1.13it/s]
  4%|▍         | 37/864 [01:13<12:14,  1.13it/s]
  4%|▍         | 38/864 [01:14<12:13,  1.13it/s]
  5%|▍         | 39/864 [01:15<12:12,  1.13it/s]
  5%|▍         | 40/864 [01:16<12:11,  1.13it/s]
                                                
{'loss': '8.225', 'grad_norm': '0.6948', 'learning_rate': '0.000234', 'epoch': '0.5585'}
+
  5%|▍         | 40/864 [01:16<12:11,  1.13it/s]
  5%|▍         | 41/864 [01:17<12:10,  1.13it/s]
  5%|▍         | 42/864 [01:18<12:09,  1.13it/s]
  5%|▍         | 43/864 [01:19<12:09,  1.13it/s]
  5%|▌         | 44/864 [01:20<12:08,  1.13it/s]
  5%|▌         | 45/864 [01:20<12:07,  1.13it/s]
  5%|▌         | 46/864 [01:21<12:06,  1.13it/s]
  5%|▌         | 47/864 [01:22<12:05,  1.13it/s]
  6%|▌         | 48/864 [01:23<12:05,  1.13it/s]
  6%|▌         | 49/864 [01:24<12:04,  1.12it/s]
  6%|▌         | 50/864 [01:25<12:03,  1.13it/s]
                                                
{'loss': '7.628', 'grad_norm': '0.3216', 'learning_rate': '0.000294', 'epoch': '0.6981'}
+
  6%|▌         | 50/864 [01:25<12:03,  1.13it/s]
  6%|▌         | 51/864 [01:26<12:02,  1.12it/s]
  6%|▌         | 52/864 [01:27<12:02,  1.12it/s]
  6%|▌         | 53/864 [01:28<12:01,  1.12it/s]
  6%|▋         | 54/864 [01:28<12:01,  1.12it/s]
  6%|▋         | 55/864 [01:29<11:59,  1.12it/s]
  6%|▋         | 56/864 [01:30<11:58,  1.12it/s]
  7%|▋         | 57/864 [01:31<11:58,  1.12it/s]
  7%|▋         | 58/864 [01:32<11:57,  1.12it/s]
  7%|▋         | 59/864 [01:33<11:57,  1.12it/s]
  7%|▋         | 60/864 [01:34<11:56,  1.12it/s]
                                                
{'loss': '7.488', 'grad_norm': '0.5528', 'learning_rate': '0.000354', 'epoch': '0.8377'}
+
  7%|▋         | 60/864 [01:34<11:56,  1.12it/s]
  7%|▋         | 61/864 [01:35<11:55,  1.12it/s]
  7%|▋         | 62/864 [01:36<11:55,  1.12it/s]
  7%|▋         | 63/864 [01:36<11:54,  1.12it/s]
  7%|▋         | 64/864 [01:37<11:53,  1.12it/s]
  8%|▊         | 65/864 [01:38<11:53,  1.12it/s]
  8%|▊         | 66/864 [01:39<11:52,  1.12it/s]
  8%|▊         | 67/864 [01:40<11:51,  1.12it/s]
  8%|▊         | 68/864 [01:41<11:50,  1.12it/s]
  8%|▊         | 69/864 [01:42<11:49,  1.12it/s]
  8%|▊         | 70/864 [01:43<11:49,  1.12it/s]
                                                
{'loss': '7.324', 'grad_norm': '1.752', 'learning_rate': '0.000414', 'epoch': '0.9773'}
+
  8%|▊         | 70/864 [01:43<11:49,  1.12it/s]
  8%|▊         | 71/864 [01:44<11:48,  1.12it/s]
  8%|▊         | 72/864 [01:44<10:29,  1.26it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 24.05it/s]
+
 15%|█▌        | 6/39 [00:00<00:02, 12.80it/s]
+
 21%|██        | 8/39 [00:00<00:02, 11.46it/s]
+
 26%|██▌       | 10/39 [00:00<00:02, 10.78it/s]
+
 31%|███       | 12/39 [00:01<00:02, 10.39it/s]
+
 36%|███▌      | 14/39 [00:01<00:02, 10.14it/s]
+
 41%|████      | 16/39 [00:01<00:02,  9.97it/s]
+
 46%|████▌     | 18/39 [00:01<00:02,  9.87it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01,  9.79it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01,  9.77it/s]
+
 56%|█████▋    | 22/39 [00:02<00:01,  9.74it/s]
+
 59%|█████▉    | 23/39 [00:02<00:01,  9.74it/s]
+
 62%|██████▏   | 24/39 [00:02<00:01,  9.72it/s]
+
 64%|██████▍   | 25/39 [00:02<00:01,  9.70it/s]
+
 67%|██████▋   | 26/39 [00:02<00:01,  9.69it/s]
+
 69%|██████▉   | 27/39 [00:02<00:01,  9.65it/s]
+
 72%|███████▏  | 28/39 [00:02<00:01,  9.67it/s]
+
 74%|███████▍  | 29/39 [00:02<00:01,  9.66it/s]
+
 77%|███████▋  | 30/39 [00:02<00:00,  9.64it/s]
+
 79%|███████▉  | 31/39 [00:03<00:00,  9.64it/s]
+
 82%|████████▏ | 32/39 [00:03<00:00,  9.64it/s]
+
 85%|████████▍ | 33/39 [00:03<00:00,  9.64it/s]
+
 87%|████████▋ | 34/39 [00:03<00:00,  9.64it/s]
+
 90%|████████▉ | 35/39 [00:03<00:00,  9.66it/s]
+
 92%|█████████▏| 36/39 [00:03<00:00,  9.66it/s]
+
 95%|█████████▍| 37/39 [00:03<00:00,  9.65it/s]
+
 97%|█████████▋| 38/39 [00:03<00:00,  9.66it/s]
+
100%|██████████| 39/39 [00:03<00:00,  9.63it/s]
                                                
+
                                               
{'eval_loss': '7.211', 'eval_runtime': '28.72', 'eval_samples_per_second': '85', 'eval_steps_per_second': '1.358', 'epoch': '1'}
+
  8%|▊         | 72/864 [02:13<10:29,  1.26it/s]
+
100%|██████████| 39/39 [00:03<00:00,  9.63it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-72
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-72/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-72/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:01<00:00,  1.03s/it]
Writing model shards: 100%|██████████| 1/1 [00:01<00:00,  1.03s/it]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-72/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-72/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-72/tokenizer_config.json
+[eval] epoch 1.0 step 72 loss 7.2114 (best 7.2114 @ ep 1.0, 0/10000 since best)
+
+
                                               
  8%|▊         | 73/864 [02:17<2:17:40, 10.44s/it]
  9%|▊         | 74/864 [02:18<1:39:47,  7.58s/it]
  9%|▊         | 75/864 [02:19<1:13:17,  5.57s/it]
  9%|▉         | 76/864 [02:20<54:45,  4.17s/it]  
  9%|▉         | 77/864 [02:21<41:48,  3.19s/it]
  9%|▉         | 78/864 [02:22<32:45,  2.50s/it]
  9%|▉         | 79/864 [02:22<26:25,  2.02s/it]
  9%|▉         | 80/864 [02:23<21:59,  1.68s/it]
                                                
{'loss': '7.13', 'grad_norm': '0.5828', 'learning_rate': '0.000474', 'epoch': '1.112'}
+
  9%|▉         | 80/864 [02:23<21:59,  1.68s/it]
  9%|▉         | 81/864 [02:24<18:53,  1.45s/it]
  9%|▉         | 82/864 [02:25<16:42,  1.28s/it]
 10%|▉         | 83/864 [02:26<15:11,  1.17s/it]
 10%|▉         | 84/864 [02:27<14:07,  1.09s/it]
 10%|▉         | 85/864 [02:28<13:22,  1.03s/it]
 10%|▉         | 86/864 [02:29<12:50,  1.01it/s]
 10%|█         | 87/864 [02:30<12:27,  1.04it/s]
 10%|█         | 88/864 [02:31<12:11,  1.06it/s]
 10%|█         | 89/864 [02:31<11:59,  1.08it/s]
 10%|█         | 90/864 [02:32<11:51,  1.09it/s]
                                                
{'loss': '6.937', 'grad_norm': '0.8481', 'learning_rate': '0.000534', 'epoch': '1.251'}
+
 10%|█         | 90/864 [02:32<11:51,  1.09it/s]
 11%|█         | 91/864 [02:33<11:45,  1.10it/s]
 11%|█         | 92/864 [02:34<11:40,  1.10it/s]
 11%|█         | 93/864 [02:35<11:37,  1.11it/s]
 11%|█         | 94/864 [02:36<11:35,  1.11it/s]
 11%|█         | 95/864 [02:37<11:33,  1.11it/s]
 11%|█         | 96/864 [02:38<11:31,  1.11it/s]
 11%|█         | 97/864 [02:39<11:30,  1.11it/s]
 11%|█▏        | 98/864 [02:40<11:29,  1.11it/s]
 11%|█▏        | 99/864 [02:40<11:27,  1.11it/s]
 12%|█▏        | 100/864 [02:41<11:26,  1.11it/s]
                                                 
{'loss': '6.788', 'grad_norm': '0.6969', 'learning_rate': '0.000594', 'epoch': '1.391'}
+
 12%|█▏        | 100/864 [02:41<11:26,  1.11it/s]
 12%|█▏        | 101/864 [02:42<11:25,  1.11it/s]
 12%|█▏        | 102/864 [02:43<11:24,  1.11it/s]
 12%|█▏        | 103/864 [02:44<11:23,  1.11it/s]
 12%|█▏        | 104/864 [02:45<11:22,  1.11it/s]
 12%|█▏        | 105/864 [02:46<11:21,  1.11it/s]
 12%|█▏        | 106/864 [02:47<11:20,  1.11it/s]
 12%|█▏        | 107/864 [02:48<11:19,  1.11it/s]
 12%|█▎        | 108/864 [02:49<11:17,  1.12it/s]
 13%|█▎        | 109/864 [02:49<11:17,  1.11it/s]
 13%|█▎        | 110/864 [02:50<11:16,  1.12it/s]
                                                 
{'loss': '6.702', 'grad_norm': '0.7814', 'learning_rate': '0.0006', 'epoch': '1.531'}
+
 13%|█▎        | 110/864 [02:50<11:16,  1.12it/s]
 13%|█▎        | 111/864 [02:51<11:15,  1.11it/s]
 13%|█▎        | 112/864 [02:52<11:14,  1.11it/s]
 13%|█▎        | 113/864 [02:53<11:13,  1.12it/s]
 13%|█▎        | 114/864 [02:54<11:12,  1.11it/s]
 13%|█▎        | 115/864 [02:55<11:11,  1.12it/s]
 13%|█▎        | 116/864 [02:56<11:10,  1.12it/s]
 14%|█▎        | 117/864 [02:57<11:09,  1.12it/s]
 14%|█▎        | 118/864 [02:57<11:08,  1.12it/s]
 14%|█▍        | 119/864 [02:58<11:07,  1.12it/s]
 14%|█▍        | 120/864 [02:59<11:06,  1.12it/s]
                                                 
{'loss': '6.547', 'grad_norm': '1.143', 'learning_rate': '0.0006', 'epoch': '1.67'}
+
 14%|█▍        | 120/864 [02:59<11:06,  1.12it/s]
 14%|█▍        | 121/864 [03:00<11:06,  1.12it/s]
 14%|█▍        | 122/864 [03:01<11:05,  1.12it/s]
 14%|█▍        | 123/864 [03:02<11:04,  1.12it/s]
 14%|█▍        | 124/864 [03:03<11:03,  1.12it/s]
 14%|█▍        | 125/864 [03:04<11:02,  1.12it/s]
 15%|█▍        | 126/864 [03:05<11:02,  1.11it/s]
 15%|█▍        | 127/864 [03:06<11:01,  1.11it/s]
 15%|█▍        | 128/864 [03:06<11:00,  1.11it/s]
 15%|█▍        | 129/864 [03:07<10:59,  1.11it/s]
 15%|█▌        | 130/864 [03:08<10:58,  1.12it/s]
                                                 
{'loss': '6.435', 'grad_norm': '0.641', 'learning_rate': '0.0006', 'epoch': '1.81'}
+
 15%|█▌        | 130/864 [03:08<10:58,  1.12it/s]
 15%|█▌        | 131/864 [03:09<10:57,  1.11it/s]
 15%|█▌        | 132/864 [03:10<10:56,  1.11it/s]
 15%|█▌        | 133/864 [03:11<10:55,  1.12it/s]
 16%|█▌        | 134/864 [03:12<10:54,  1.12it/s]
 16%|█▌        | 135/864 [03:13<10:53,  1.12it/s]
 16%|█▌        | 136/864 [03:14<10:52,  1.12it/s]
 16%|█▌        | 137/864 [03:15<10:51,  1.12it/s]
 16%|█▌        | 138/864 [03:15<10:50,  1.12it/s]
 16%|█▌        | 139/864 [03:16<10:49,  1.12it/s]
 16%|█▌        | 140/864 [03:17<10:49,  1.12it/s]
                                                 
{'loss': '6.351', 'grad_norm': '0.375', 'learning_rate': '0.0006', 'epoch': '1.949'}
+
 16%|█▌        | 140/864 [03:17<10:49,  1.12it/s]
 16%|█▋        | 141/864 [03:18<10:48,  1.12it/s]
 16%|█▋        | 142/864 [03:19<10:47,  1.11it/s]
 17%|█▋        | 143/864 [03:20<10:46,  1.12it/s]
 17%|█▋        | 144/864 [03:20<09:34,  1.25it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:01, 19.44it/s]
+
 10%|█         | 4/39 [00:00<00:02, 12.11it/s]
+
 15%|█▌        | 6/39 [00:00<00:03, 10.85it/s]
+
 21%|██        | 8/39 [00:00<00:02, 10.33it/s]
+
 26%|██▌       | 10/39 [00:00<00:02, 10.07it/s]
+
 31%|███       | 12/39 [00:01<00:02,  9.92it/s]
+
 36%|███▌      | 14/39 [00:01<00:02,  9.82it/s]
+
 41%|████      | 16/39 [00:01<00:02,  9.76it/s]
+
 44%|████▎     | 17/39 [00:01<00:02,  9.73it/s]
+
 46%|████▌     | 18/39 [00:01<00:02,  9.72it/s]
+
 49%|████▊     | 19/39 [00:01<00:02,  9.58it/s]
+
 54%|█████▍    | 21/39 [00:02<00:01,  9.71it/s]
+
 56%|█████▋    | 22/39 [00:02<00:01,  9.70it/s]
+
 59%|█████▉    | 23/39 [00:02<00:01,  9.70it/s]
+
 62%|██████▏   | 24/39 [00:02<00:01,  9.69it/s]
+
 64%|██████▍   | 25/39 [00:02<00:01,  9.68it/s]
+
 67%|██████▋   | 26/39 [00:02<00:01,  9.67it/s]
+
 69%|██████▉   | 27/39 [00:02<00:01,  9.67it/s]
+
 72%|███████▏  | 28/39 [00:02<00:01,  9.67it/s]
+
 74%|███████▍  | 29/39 [00:02<00:01,  9.66it/s]
+
 77%|███████▋  | 30/39 [00:03<00:00,  9.66it/s]
+
 79%|███████▉  | 31/39 [00:03<00:00,  9.66it/s]
+
 82%|████████▏ | 32/39 [00:03<00:00,  9.66it/s]
+
 85%|████████▍ | 33/39 [00:03<00:00,  9.66it/s]
+
 87%|████████▋ | 34/39 [00:03<00:00,  9.66it/s]
+
 90%|████████▉ | 35/39 [00:03<00:00,  9.66it/s]
+
 92%|█████████▏| 36/39 [00:03<00:00,  9.66it/s]
+
 95%|█████████▍| 37/39 [00:03<00:00,  9.66it/s]
+
 97%|█████████▋| 38/39 [00:03<00:00,  9.66it/s]
+
100%|██████████| 39/39 [00:03<00:00,  9.66it/s]
                                                 
+
                                               
{'eval_loss': '6.276', 'eval_runtime': '4.105', 'eval_samples_per_second': '594.6', 'eval_steps_per_second': '9.501', 'epoch': '2'}
+
 17%|█▋        | 144/864 [03:25<09:34,  1.25it/s]
+
100%|██████████| 39/39 [00:04<00:00,  9.66it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-144
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-144/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-144/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.13it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.13it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-144/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-144/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-144/tokenizer_config.json
+[eval] epoch 2.0 step 144 loss 6.2764 (best 6.2764 @ ep 2.0, 0/10000 since best)
+
+
                                               
 17%|█▋        | 145/864 [03:29<37:05,  3.10s/it]
 17%|█▋        | 146/864 [03:30<29:08,  2.44s/it]
 17%|█▋        | 147/864 [03:31<23:35,  1.97s/it]
 17%|█▋        | 148/864 [03:32<19:42,  1.65s/it]
 17%|█▋        | 149/864 [03:33<16:58,  1.42s/it]
 17%|█▋        | 150/864 [03:33<15:04,  1.27s/it]
                                                 
{'loss': '6.234', 'grad_norm': '0.5606', 'learning_rate': '0.0006', 'epoch': '2.084'}
+
 17%|█▋        | 150/864 [03:33<15:04,  1.27s/it]
 17%|█▋        | 151/864 [03:34<13:44,  1.16s/it]
 18%|█▊        | 152/864 [03:35<12:47,  1.08s/it]
 18%|█▊        | 153/864 [03:36<12:08,  1.02s/it]
 18%|█▊        | 154/864 [03:37<11:39,  1.01it/s]
 18%|█▊        | 155/864 [03:38<11:19,  1.04it/s]
 18%|█▊        | 156/864 [03:39<11:05,  1.06it/s]
 18%|█▊        | 157/864 [03:40<10:55,  1.08it/s]
 18%|█▊        | 158/864 [03:41<10:48,  1.09it/s]
 18%|█▊        | 159/864 [03:41<10:42,  1.10it/s]
 19%|█▊        | 160/864 [03:42<10:38,  1.10it/s]
                                                 
{'loss': '6.151', 'grad_norm': '0.4197', 'learning_rate': '0.0006', 'epoch': '2.223'}
+
 19%|█▊        | 160/864 [03:42<10:38,  1.10it/s]
 19%|█▊        | 161/864 [03:43<10:35,  1.11it/s]
 19%|█▉        | 162/864 [03:44<10:33,  1.11it/s]
 19%|█▉        | 163/864 [03:45<10:30,  1.11it/s]
 19%|█▉        | 164/864 [03:46<10:29,  1.11it/s]
 19%|█▉        | 165/864 [03:47<10:27,  1.11it/s]
 19%|█▉        | 166/864 [03:48<10:26,  1.11it/s]
 19%|█▉        | 167/864 [03:49<10:25,  1.11it/s]
 19%|█▉        | 168/864 [03:50<10:25,  1.11it/s]
 20%|█▉        | 169/864 [03:50<10:24,  1.11it/s]
 20%|█▉        | 170/864 [03:51<10:22,  1.11it/s]
                                                 
{'loss': '6.06', 'grad_norm': '0.5928', 'learning_rate': '0.0006', 'epoch': '2.363'}
+
 20%|█▉        | 170/864 [03:51<10:22,  1.11it/s]
 20%|█▉        | 171/864 [03:52<10:22,  1.11it/s]
 20%|█▉        | 172/864 [03:53<10:21,  1.11it/s]
 20%|██        | 173/864 [03:54<10:19,  1.11it/s]
 20%|██        | 174/864 [03:55<10:18,  1.12it/s]
 20%|██        | 175/864 [03:56<10:17,  1.12it/s]
 20%|██        | 176/864 [03:57<10:16,  1.12it/s]
 20%|██        | 177/864 [03:58<10:15,  1.12it/s]
 21%|██        | 178/864 [03:59<10:14,  1.12it/s]
 21%|██        | 179/864 [03:59<10:13,  1.12it/s]
 21%|██        | 180/864 [04:00<10:13,  1.11it/s]
                                                 
{'loss': '6.083', 'grad_norm': '0.8861', 'learning_rate': '0.0006', 'epoch': '2.503'}
+
 21%|██        | 180/864 [04:00<10:13,  1.11it/s]
 21%|██        | 181/864 [04:01<10:12,  1.12it/s]
 21%|██        | 182/864 [04:02<10:11,  1.12it/s]
 21%|██        | 183/864 [04:03<10:10,  1.12it/s]
 21%|██▏       | 184/864 [04:04<10:09,  1.12it/s]
 21%|██▏       | 185/864 [04:05<10:09,  1.11it/s]
 22%|██▏       | 186/864 [04:06<10:07,  1.12it/s]
 22%|██▏       | 187/864 [04:07<10:06,  1.12it/s]
 22%|██▏       | 188/864 [04:07<10:06,  1.12it/s]
 22%|██▏       | 189/864 [04:08<10:05,  1.12it/s]
 22%|██▏       | 190/864 [04:09<10:04,  1.12it/s]
                                                 
{'loss': '5.966', 'grad_norm': '0.3844', 'learning_rate': '0.0006', 'epoch': '2.642'}
+
 22%|██▏       | 190/864 [04:09<10:04,  1.12it/s]
 22%|██▏       | 191/864 [04:10<10:03,  1.12it/s]
 22%|██▏       | 192/864 [04:11<10:02,  1.12it/s]
 22%|██▏       | 193/864 [04:12<10:01,  1.12it/s]
 22%|██▏       | 194/864 [04:13<10:01,  1.11it/s]
 23%|██▎       | 195/864 [04:14<10:00,  1.11it/s]
 23%|██▎       | 196/864 [04:15<09:58,  1.12it/s]
 23%|██▎       | 197/864 [04:16<09:57,  1.12it/s]
 23%|██▎       | 198/864 [04:16<09:56,  1.12it/s]
 23%|██▎       | 199/864 [04:17<09:55,  1.12it/s]
 23%|██▎       | 200/864 [04:18<09:55,  1.11it/s]
                                                 
{'loss': '5.871', 'grad_norm': '1.134', 'learning_rate': '0.0006', 'epoch': '2.782'}
+
 23%|██▎       | 200/864 [04:18<09:55,  1.11it/s]
 23%|██▎       | 201/864 [04:19<09:54,  1.11it/s]
 23%|██▎       | 202/864 [04:20<09:53,  1.12it/s]
 23%|██▎       | 203/864 [04:21<09:52,  1.11it/s]
 24%|██▎       | 204/864 [04:22<09:52,  1.11it/s]
 24%|██▎       | 205/864 [04:23<09:51,  1.11it/s]
 24%|██▍       | 206/864 [04:24<09:50,  1.12it/s]
 24%|██▍       | 207/864 [04:25<09:49,  1.12it/s]
 24%|██▍       | 208/864 [04:25<09:48,  1.12it/s]
 24%|██▍       | 209/864 [04:26<09:47,  1.12it/s]
 24%|██▍       | 210/864 [04:27<09:46,  1.12it/s]
                                                 
{'loss': '5.833', 'grad_norm': '0.5506', 'learning_rate': '0.0006', 'epoch': '2.921'}
+
 24%|██▍       | 210/864 [04:27<09:46,  1.12it/s]
 24%|██▍       | 211/864 [04:28<09:45,  1.11it/s]
 25%|██▍       | 212/864 [04:29<09:44,  1.12it/s]
 25%|██▍       | 213/864 [04:30<09:43,  1.12it/s]
 25%|██▍       | 214/864 [04:31<09:42,  1.12it/s]
 25%|██▍       | 215/864 [04:32<09:41,  1.12it/s]
 25%|██▌       | 216/864 [04:32<08:36,  1.26it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:01, 19.24it/s]
+
 10%|█         | 4/39 [00:00<00:02, 12.33it/s]
+
 15%|█▌        | 6/39 [00:00<00:03, 10.82it/s]
+
 21%|██        | 8/39 [00:00<00:02, 10.42it/s]
+
 26%|██▌       | 10/39 [00:00<00:02, 10.04it/s]
+
 31%|███       | 12/39 [00:01<00:02,  9.98it/s]
+
 36%|███▌      | 14/39 [00:01<00:02,  9.80it/s]
+
 41%|████      | 16/39 [00:01<00:02,  9.82it/s]
+
 44%|████▎     | 17/39 [00:01<00:02,  9.79it/s]
+
 46%|████▌     | 18/39 [00:01<00:02,  9.67it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01,  9.74it/s]
+
 54%|█████▍    | 21/39 [00:02<00:01,  9.72it/s]
+
 56%|█████▋    | 22/39 [00:02<00:01,  9.61it/s]
+
 62%|██████▏   | 24/39 [00:02<00:01,  9.73it/s]
+
 64%|██████▍   | 25/39 [00:02<00:01,  9.71it/s]
+
 67%|██████▋   | 26/39 [00:02<00:01,  9.70it/s]
+
 69%|██████▉   | 27/39 [00:02<00:01,  9.70it/s]
+
 72%|███████▏  | 28/39 [00:02<00:01,  9.68it/s]
+
 74%|███████▍  | 29/39 [00:02<00:01,  9.67it/s]
+
 77%|███████▋  | 30/39 [00:02<00:00,  9.68it/s]
+
 79%|███████▉  | 31/39 [00:03<00:00,  9.68it/s]
+
 82%|████████▏ | 32/39 [00:03<00:00,  9.67it/s]
+
 85%|████████▍ | 33/39 [00:03<00:00,  9.67it/s]
+
 87%|████████▋ | 34/39 [00:03<00:00,  9.67it/s]
+
 90%|████████▉ | 35/39 [00:03<00:00,  9.66it/s]
+
 92%|█████████▏| 36/39 [00:03<00:00,  9.66it/s]
+
 95%|█████████▍| 37/39 [00:03<00:00,  9.66it/s]
+
 97%|█████████▋| 38/39 [00:03<00:00,  9.67it/s]
+
100%|██████████| 39/39 [00:03<00:00,  9.66it/s]
                                                 
+
                                               
{'eval_loss': '5.774', 'eval_runtime': '4.095', 'eval_samples_per_second': '596.1', 'eval_steps_per_second': '9.524', 'epoch': '3'}
+
 25%|██▌       | 216/864 [04:36<08:36,  1.26it/s]
+
100%|██████████| 39/39 [00:04<00:00,  9.66it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-216
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-216/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-216/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.14it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.14it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-216/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-216/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-216/tokenizer_config.json
+[eval] epoch 3.0 step 216 loss 5.7738 (best 5.7738 @ ep 3.0, 0/10000 since best)
+
+
                                               
 25%|██▌       | 217/864 [04:41<33:11,  3.08s/it]
 25%|██▌       | 218/864 [04:42<26:05,  2.42s/it]
 25%|██▌       | 219/864 [04:42<21:07,  1.97s/it]
 25%|██▌       | 220/864 [04:43<17:39,  1.64s/it]
                                                 
{'loss': '5.731', 'grad_norm': '0.3969', 'learning_rate': '0.0006', 'epoch': '3.056'}
+
 25%|██▌       | 220/864 [04:43<17:39,  1.64s/it]
 26%|██▌       | 221/864 [04:44<15:13,  1.42s/it]
 26%|██▌       | 222/864 [04:45<13:31,  1.26s/it]
 26%|██▌       | 223/864 [04:46<12:19,  1.15s/it]
 26%|██▌       | 224/864 [04:47<11:29,  1.08s/it]
 26%|██▌       | 225/864 [04:48<10:53,  1.02s/it]
 26%|██▌       | 226/864 [04:49<10:29,  1.01it/s]
 26%|██▋       | 227/864 [04:50<10:11,  1.04it/s]
 26%|██▋       | 228/864 [04:51<09:59,  1.06it/s]
 27%|██▋       | 229/864 [04:51<09:49,  1.08it/s]
 27%|██▋       | 230/864 [04:52<09:42,  1.09it/s]
                                                 
{'loss': '5.671', 'grad_norm': '0.9978', 'learning_rate': '0.0006', 'epoch': '3.195'}
+
 27%|██▋       | 230/864 [04:52<09:42,  1.09it/s]
 27%|██▋       | 231/864 [04:53<09:38,  1.10it/s]
 27%|██▋       | 232/864 [04:54<09:34,  1.10it/s]
 27%|██▋       | 233/864 [04:55<09:31,  1.10it/s]
 27%|██▋       | 234/864 [04:56<09:29,  1.11it/s]
 27%|██▋       | 235/864 [04:57<09:27,  1.11it/s]
 27%|██▋       | 236/864 [04:58<09:26,  1.11it/s]
 27%|██▋       | 237/864 [04:59<09:24,  1.11it/s]
 28%|██▊       | 238/864 [05:00<09:23,  1.11it/s]
 28%|██▊       | 239/864 [05:00<09:22,  1.11it/s]
 28%|██▊       | 240/864 [05:01<09:21,  1.11it/s]
                                                 
{'loss': '5.657', 'grad_norm': '0.3981', 'learning_rate': '0.0006', 'epoch': '3.335'}
+
 28%|██▊       | 240/864 [05:01<09:21,  1.11it/s]
 28%|██▊       | 241/864 [05:02<09:19,  1.11it/s]
 28%|██▊       | 242/864 [05:03<09:18,  1.11it/s]
 28%|██▊       | 243/864 [05:04<09:17,  1.11it/s]
 28%|██▊       | 244/864 [05:05<09:17,  1.11it/s]
 28%|██▊       | 245/864 [05:06<09:16,  1.11it/s]
 28%|██▊       | 246/864 [05:07<09:14,  1.11it/s]
 29%|██▊       | 247/864 [05:08<09:14,  1.11it/s]
 29%|██▊       | 248/864 [05:09<09:13,  1.11it/s]
 29%|██▉       | 249/864 [05:09<09:12,  1.11it/s]
 29%|██▉       | 250/864 [05:10<09:11,  1.11it/s]
                                                 
{'loss': '5.569', 'grad_norm': '0.2279', 'learning_rate': '0.0006', 'epoch': '3.475'}
+
 29%|██▉       | 250/864 [05:10<09:11,  1.11it/s]
 29%|██▉       | 251/864 [05:11<09:10,  1.11it/s]
 29%|██▉       | 252/864 [05:12<09:09,  1.11it/s]
 29%|██▉       | 253/864 [05:13<09:08,  1.11it/s]
 29%|██▉       | 254/864 [05:14<09:07,  1.11it/s]
 30%|██▉       | 255/864 [05:15<09:07,  1.11it/s]
 30%|██▉       | 256/864 [05:16<09:06,  1.11it/s]
 30%|██▉       | 257/864 [05:17<09:05,  1.11it/s]
 30%|██▉       | 258/864 [05:17<09:04,  1.11it/s]
 30%|██▉       | 259/864 [05:18<09:03,  1.11it/s]
 30%|███       | 260/864 [05:19<09:02,  1.11it/s]
                                                 
{'loss': '5.489', 'grad_norm': '0.5184', 'learning_rate': '0.0006', 'epoch': '3.614'}
+
 30%|███       | 260/864 [05:19<09:02,  1.11it/s]
 30%|███       | 261/864 [05:20<09:01,  1.11it/s]
 30%|███       | 262/864 [05:21<09:01,  1.11it/s]
 30%|███       | 263/864 [05:22<09:00,  1.11it/s]
 31%|███       | 264/864 [05:23<08:59,  1.11it/s]
 31%|███       | 265/864 [05:24<08:58,  1.11it/s]
 31%|███       | 266/864 [05:25<08:57,  1.11it/s]
 31%|███       | 267/864 [05:26<08:56,  1.11it/s]
 31%|███       | 268/864 [05:26<08:55,  1.11it/s]
 31%|███       | 269/864 [05:27<08:55,  1.11it/s]
 31%|███▏      | 270/864 [05:28<08:54,  1.11it/s]
                                                 
{'loss': '5.47', 'grad_norm': '0.6049', 'learning_rate': '0.0006', 'epoch': '3.754'}
+
 31%|███▏      | 270/864 [05:28<08:54,  1.11it/s]
 31%|███▏      | 271/864 [05:29<08:53,  1.11it/s]
 31%|███▏      | 272/864 [05:30<08:52,  1.11it/s]
 32%|███▏      | 273/864 [05:31<08:51,  1.11it/s]
 32%|███▏      | 274/864 [05:32<08:49,  1.11it/s]
 32%|███▏      | 275/864 [05:33<08:49,  1.11it/s]
 32%|███▏      | 276/864 [05:34<08:48,  1.11it/s]
 32%|███▏      | 277/864 [05:35<08:47,  1.11it/s]
 32%|███▏      | 278/864 [05:35<08:46,  1.11it/s]
 32%|███▏      | 279/864 [05:36<08:45,  1.11it/s]
 32%|███▏      | 280/864 [05:37<08:44,  1.11it/s]
                                                 
{'loss': '5.39', 'grad_norm': '0.2342', 'learning_rate': '0.0006', 'epoch': '3.894'}
+
 32%|███▏      | 280/864 [05:37<08:44,  1.11it/s]
 33%|███▎      | 281/864 [05:38<08:44,  1.11it/s]
 33%|███▎      | 282/864 [05:39<08:43,  1.11it/s]
 33%|███▎      | 283/864 [05:40<08:42,  1.11it/s]
 33%|███▎      | 284/864 [05:41<08:41,  1.11it/s]
 33%|███▎      | 285/864 [05:42<08:40,  1.11it/s]
 33%|███▎      | 286/864 [05:43<08:39,  1.11it/s]
 33%|███▎      | 287/864 [05:44<08:38,  1.11it/s]
 33%|███▎      | 288/864 [05:44<07:40,  1.25it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:01, 19.23it/s]
+
 10%|█         | 4/39 [00:00<00:02, 12.11it/s]
+
 15%|█▌        | 6/39 [00:00<00:03, 10.85it/s]
+
 21%|██        | 8/39 [00:00<00:03, 10.29it/s]
+
 26%|██▌       | 10/39 [00:00<00:02, 10.03it/s]
+
 31%|███       | 12/39 [00:01<00:02,  9.89it/s]
+
 36%|███▌      | 14/39 [00:01<00:02,  9.80it/s]
+
 38%|███▊      | 15/39 [00:01<00:02,  9.77it/s]
+
 41%|████      | 16/39 [00:01<00:02,  9.74it/s]
+
 44%|████▎     | 17/39 [00:01<00:02,  9.72it/s]
+
 46%|████▌     | 18/39 [00:01<00:02,  9.70it/s]
+
 49%|████▊     | 19/39 [00:01<00:02,  9.68it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01,  9.68it/s]
+
 54%|█████▍    | 21/39 [00:02<00:01,  9.67it/s]
+
 56%|█████▋    | 22/39 [00:02<00:01,  9.66it/s]
+
 59%|█████▉    | 23/39 [00:02<00:01,  9.67it/s]
+
 62%|██████▏   | 24/39 [00:02<00:01,  9.66it/s]
+
 64%|██████▍   | 25/39 [00:02<00:01,  9.67it/s]
+
 67%|██████▋   | 26/39 [00:02<00:01,  9.67it/s]
+
 69%|██████▉   | 27/39 [00:02<00:01,  9.67it/s]
+
 72%|███████▏  | 28/39 [00:02<00:01,  9.66it/s]
+
 74%|███████▍  | 29/39 [00:02<00:01,  9.65it/s]
+
 77%|███████▋  | 30/39 [00:03<00:00,  9.66it/s]
+
 79%|███████▉  | 31/39 [00:03<00:00,  9.67it/s]
+
 82%|████████▏ | 32/39 [00:03<00:00,  9.66it/s]
+
 85%|████████▍ | 33/39 [00:03<00:00,  9.67it/s]
+
 87%|████████▋ | 34/39 [00:03<00:00,  9.66it/s]
+
 90%|████████▉ | 35/39 [00:03<00:00,  9.66it/s]
+
 92%|█████████▏| 36/39 [00:03<00:00,  9.66it/s]
+
 95%|█████████▍| 37/39 [00:03<00:00,  9.66it/s]
+
 97%|█████████▋| 38/39 [00:03<00:00,  9.66it/s]
+
100%|██████████| 39/39 [00:03<00:00,  9.66it/s]
                                                 
+
                                               
{'eval_loss': '5.35', 'eval_runtime': '4.114', 'eval_samples_per_second': '593.3', 'eval_steps_per_second': '9.48', 'epoch': '4'}
+
 33%|███▎      | 288/864 [05:48<07:40,  1.25it/s]
+
100%|██████████| 39/39 [00:04<00:00,  9.66it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-288
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-288/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-288/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.12it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.12it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-288/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-288/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-288/tokenizer_config.json
+[eval] epoch 4.0 step 288 loss 5.3500 (best 5.3500 @ ep 4.0, 0/10000 since best)
+
+
                                               
 33%|███▎      | 289/864 [05:53<29:40,  3.10s/it]
 34%|███▎      | 290/864 [05:53<23:18,  2.44s/it]
                                                 
{'loss': '5.291', 'grad_norm': '0.7218', 'learning_rate': '0.0006', 'epoch': '4.028'}
+
 34%|███▎      | 290/864 [05:53<23:18,  2.44s/it]
 34%|███▎      | 291/864 [05:54<18:51,  1.97s/it]
 34%|███▍      | 292/864 [05:55<15:44,  1.65s/it]
 34%|███▍      | 293/864 [05:56<13:34,  1.43s/it]
 34%|███▍      | 294/864 [05:57<12:02,  1.27s/it]
 34%|███▍      | 295/864 [05:58<10:58,  1.16s/it]
 34%|███▍      | 296/864 [05:59<10:13,  1.08s/it]
 34%|███▍      | 297/864 [06:00<09:41,  1.03s/it]
 34%|███▍      | 298/864 [06:01<09:19,  1.01it/s]
 35%|███▍      | 299/864 [06:02<09:03,  1.04it/s]
 35%|███▍      | 300/864 [06:02<08:51,  1.06it/s]
                                                 
{'loss': '5.302', 'grad_norm': '0.6472', 'learning_rate': '0.0006', 'epoch': '4.168'}
+
 35%|███▍      | 300/864 [06:02<08:51,  1.06it/s]
 35%|███▍      | 301/864 [06:03<08:43,  1.08it/s]
 35%|███▍      | 302/864 [06:04<08:37,  1.09it/s]
 35%|███▌      | 303/864 [06:05<08:32,  1.09it/s]
 35%|███▌      | 304/864 [06:06<08:29,  1.10it/s]
 35%|███▌      | 305/864 [06:07<08:26,  1.10it/s]
 35%|███▌      | 306/864 [06:08<08:24,  1.11it/s]
 36%|███▌      | 307/864 [06:09<08:22,  1.11it/s]
 36%|███▌      | 308/864 [06:10<08:20,  1.11it/s]
 36%|███▌      | 309/864 [06:11<08:19,  1.11it/s]
 36%|███▌      | 310/864 [06:11<08:18,  1.11it/s]
                                                 
{'loss': '5.247', 'grad_norm': '0.2796', 'learning_rate': '0.0006', 'epoch': '4.307'}
+
 36%|███▌      | 310/864 [06:11<08:18,  1.11it/s]
 36%|███▌      | 311/864 [06:12<08:17,  1.11it/s]
 36%|███▌      | 312/864 [06:13<08:16,  1.11it/s]
 36%|███▌      | 313/864 [06:14<08:14,  1.11it/s]
 36%|███▋      | 314/864 [06:15<08:13,  1.11it/s]
 36%|███▋      | 315/864 [06:16<08:12,  1.11it/s]
 37%|███▋      | 316/864 [06:17<08:11,  1.11it/s]
 37%|███▋      | 317/864 [06:18<08:11,  1.11it/s]
 37%|███▋      | 318/864 [06:19<08:10,  1.11it/s]
 37%|███▋      | 319/864 [06:20<08:09,  1.11it/s]
 37%|███▋      | 320/864 [06:20<08:08,  1.11it/s]
                                                 
{'loss': '5.154', 'grad_norm': '0.1917', 'learning_rate': '0.0006', 'epoch': '4.447'}
+
 37%|███▋      | 320/864 [06:20<08:08,  1.11it/s]
 37%|███▋      | 321/864 [06:21<08:07,  1.11it/s]
 37%|███▋      | 322/864 [06:22<08:06,  1.11it/s]
 37%|███▋      | 323/864 [06:23<08:05,  1.12it/s]
 38%|███▊      | 324/864 [06:24<08:04,  1.11it/s]
 38%|███▊      | 325/864 [06:25<08:03,  1.11it/s]
 38%|███▊      | 326/864 [06:26<08:02,  1.11it/s]
 38%|███▊      | 327/864 [06:27<08:01,  1.12it/s]
 38%|███▊      | 328/864 [06:28<08:00,  1.12it/s]
 38%|███▊      | 329/864 [06:28<07:59,  1.11it/s]
 38%|███▊      | 330/864 [06:29<07:58,  1.12it/s]
                                                 
{'loss': '5.114', 'grad_norm': '0.5547', 'learning_rate': '0.0006', 'epoch': '4.586'}
+
 38%|███▊      | 330/864 [06:29<07:58,  1.12it/s]
 38%|███▊      | 331/864 [06:30<07:58,  1.11it/s]
 38%|███▊      | 332/864 [06:31<07:57,  1.11it/s]
 39%|███▊      | 333/864 [06:32<07:56,  1.12it/s]
 39%|███▊      | 334/864 [06:33<07:55,  1.12it/s]
 39%|███▉      | 335/864 [06:34<07:54,  1.12it/s]
 39%|███▉      | 336/864 [06:35<07:53,  1.12it/s]
 39%|███▉      | 337/864 [06:36<07:52,  1.11it/s]
 39%|███▉      | 338/864 [06:37<07:51,  1.11it/s]
 39%|███▉      | 339/864 [06:37<07:50,  1.11it/s]
 39%|███▉      | 340/864 [06:38<07:49,  1.12it/s]
                                                 
{'loss': '5.057', 'grad_norm': '0.7748', 'learning_rate': '0.0006', 'epoch': '4.726'}
+
 39%|███▉      | 340/864 [06:38<07:49,  1.12it/s]
 39%|███▉      | 341/864 [06:39<07:49,  1.11it/s]
 40%|███▉      | 342/864 [06:40<07:48,  1.11it/s]
 40%|███▉      | 343/864 [06:41<07:47,  1.11it/s]
 40%|███▉      | 344/864 [06:42<07:46,  1.11it/s]
 40%|███▉      | 345/864 [06:43<07:45,  1.11it/s]
 40%|████      | 346/864 [06:44<07:44,  1.12it/s]
 40%|████      | 347/864 [06:45<07:43,  1.12it/s]
 40%|████      | 348/864 [06:46<07:42,  1.12it/s]
 40%|████      | 349/864 [06:46<07:41,  1.12it/s]
 41%|████      | 350/864 [06:47<07:40,  1.12it/s]
                                                 
{'loss': '5.059', 'grad_norm': '0.4927', 'learning_rate': '0.0006', 'epoch': '4.866'}
+
 41%|████      | 350/864 [06:47<07:40,  1.12it/s]
 41%|████      | 351/864 [06:48<07:40,  1.11it/s]
 41%|████      | 352/864 [06:49<07:39,  1.11it/s]
 41%|████      | 353/864 [06:50<07:38,  1.12it/s]
 41%|████      | 354/864 [06:51<07:37,  1.11it/s]
 41%|████      | 355/864 [06:52<07:36,  1.11it/s]
 41%|████      | 356/864 [06:53<07:35,  1.12it/s]
 41%|████▏     | 357/864 [06:54<07:34,  1.11it/s]
 41%|████▏     | 358/864 [06:55<07:34,  1.11it/s]
 42%|████▏     | 359/864 [06:55<07:33,  1.11it/s]
 42%|████▏     | 360/864 [06:56<06:42,  1.25it/s]
                                                 
{'loss': '4.985', 'grad_norm': '0.1963', 'learning_rate': '0.0006', 'epoch': '5'}
+
 42%|████▏     | 360/864 [06:56<06:42,  1.25it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:01, 19.66it/s]
+
 10%|█         | 4/39 [00:00<00:02, 12.19it/s]
+
 15%|█▌        | 6/39 [00:00<00:03, 10.87it/s]
+
 21%|██        | 8/39 [00:00<00:02, 10.36it/s]
+
 26%|██▌       | 10/39 [00:00<00:02, 10.08it/s]
+
 31%|███       | 12/39 [00:01<00:02,  9.93it/s]
+
 36%|███▌      | 14/39 [00:01<00:02,  9.82it/s]
+
 41%|████      | 16/39 [00:01<00:02,  9.76it/s]
+
 44%|████▎     | 17/39 [00:01<00:02,  9.73it/s]
+
 46%|████▌     | 18/39 [00:01<00:02,  9.71it/s]
+
 49%|████▊     | 19/39 [00:01<00:02,  9.68it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01,  9.69it/s]
+
 54%|█████▍    | 21/39 [00:02<00:01,  9.68it/s]
+
 56%|█████▋    | 22/39 [00:02<00:01,  9.65it/s]
+
 59%|█████▉    | 23/39 [00:02<00:01,  9.68it/s]
+
 62%|██████▏   | 24/39 [00:02<00:01,  9.67it/s]
+
 64%|██████▍   | 25/39 [00:02<00:01,  9.67it/s]
+
 67%|██████▋   | 26/39 [00:02<00:01,  9.66it/s]
+
 69%|██████▉   | 27/39 [00:02<00:01,  9.67it/s]
+
 72%|███████▏  | 28/39 [00:02<00:01,  9.66it/s]
+
 74%|███████▍  | 29/39 [00:02<00:01,  9.66it/s]
+
 77%|███████▋  | 30/39 [00:03<00:00,  9.65it/s]
+
 79%|███████▉  | 31/39 [00:03<00:00,  9.65it/s]
+
 82%|████████▏ | 32/39 [00:03<00:00,  9.65it/s]
+
 85%|████████▍ | 33/39 [00:03<00:00,  9.65it/s]
+
 87%|████████▋ | 34/39 [00:03<00:00,  9.63it/s]
+
 90%|████████▉ | 35/39 [00:03<00:00,  9.62it/s]
+
 92%|█████████▏| 36/39 [00:03<00:00,  9.65it/s]
+
 95%|█████████▍| 37/39 [00:03<00:00,  9.65it/s]
+
 97%|█████████▋| 38/39 [00:03<00:00,  9.65it/s]
+
100%|██████████| 39/39 [00:03<00:00,  9.65it/s]
                                                 
+
                                               
{'eval_loss': '5.019', 'eval_runtime': '4.102', 'eval_samples_per_second': '595.1', 'eval_steps_per_second': '9.509', 'epoch': '5'}
+
 42%|████▏     | 360/864 [07:00<06:42,  1.25it/s]
+
100%|██████████| 39/39 [00:04<00:00,  9.65it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-360
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-360/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-360/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.13it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.13it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-360/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-360/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-360/tokenizer_config.json
+[eval] epoch 5.0 step 360 loss 5.0192 (best 5.0192 @ ep 5.0, 0/10000 since best)
+
+
                                               
 42%|████▏     | 361/864 [07:04<25:59,  3.10s/it]
 42%|████▏     | 362/864 [07:05<20:24,  2.44s/it]
 42%|████▏     | 363/864 [07:06<16:30,  1.98s/it]
 42%|████▏     | 364/864 [07:07<13:46,  1.65s/it]
 42%|████▏     | 365/864 [07:08<11:51,  1.43s/it]
 42%|████▏     | 366/864 [07:09<10:31,  1.27s/it]
 42%|████▏     | 367/864 [07:10<09:35,  1.16s/it]
 43%|████▎     | 368/864 [07:11<08:55,  1.08s/it]
 43%|████▎     | 369/864 [07:12<08:27,  1.03s/it]
 43%|████▎     | 370/864 [07:13<08:08,  1.01it/s]
                                                 
{'loss': '4.859', 'grad_norm': '0.2485', 'learning_rate': '0.0006', 'epoch': '5.14'}
+
 43%|████▎     | 370/864 [07:13<08:08,  1.01it/s]
 43%|████▎     | 371/864 [07:13<07:54,  1.04it/s]
 43%|████▎     | 372/864 [07:14<07:44,  1.06it/s]
 43%|████▎     | 373/864 [07:15<07:37,  1.07it/s]
 43%|████▎     | 374/864 [07:16<07:31,  1.08it/s]
 43%|████▎     | 375/864 [07:17<07:27,  1.09it/s]
 44%|████▎     | 376/864 [07:18<07:24,  1.10it/s]
 44%|████▎     | 377/864 [07:19<07:22,  1.10it/s]
 44%|████▍     | 378/864 [07:20<07:20,  1.10it/s]
 44%|████▍     | 379/864 [07:21<07:18,  1.11it/s]
 44%|████▍     | 380/864 [07:22<07:17,  1.11it/s]
                                                 
{'loss': '4.838', 'grad_norm': '0.593', 'learning_rate': '0.0006', 'epoch': '5.279'}
+
 44%|████▍     | 380/864 [07:22<07:17,  1.11it/s]
 44%|████▍     | 381/864 [07:22<07:16,  1.11it/s]
 44%|████▍     | 382/864 [07:23<07:15,  1.11it/s]
 44%|████▍     | 383/864 [07:24<07:14,  1.11it/s]
 44%|████▍     | 384/864 [07:25<07:12,  1.11it/s]
 45%|████▍     | 385/864 [07:26<07:11,  1.11it/s]
 45%|████▍     | 386/864 [07:27<07:10,  1.11it/s]
 45%|████▍     | 387/864 [07:28<07:09,  1.11it/s]
 45%|████▍     | 388/864 [07:29<07:08,  1.11it/s]
 45%|████▌     | 389/864 [07:30<07:07,  1.11it/s]
 45%|████▌     | 390/864 [07:31<07:06,  1.11it/s]
                                                 
{'loss': '4.836', 'grad_norm': '0.755', 'learning_rate': '0.0006', 'epoch': '5.419'}
+
 45%|████▌     | 390/864 [07:31<07:06,  1.11it/s]
 45%|████▌     | 391/864 [07:31<07:05,  1.11it/s]
 45%|████▌     | 392/864 [07:32<07:05,  1.11it/s]
 45%|████▌     | 393/864 [07:33<07:03,  1.11it/s]
 46%|████▌     | 394/864 [07:34<07:02,  1.11it/s]
 46%|████▌     | 395/864 [07:35<07:02,  1.11it/s]
 46%|████▌     | 396/864 [07:36<07:01,  1.11it/s]
 46%|████▌     | 397/864 [07:37<07:00,  1.11it/s]
 46%|████▌     | 398/864 [07:38<06:59,  1.11it/s]
 46%|████▌     | 399/864 [07:39<06:58,  1.11it/s]
 46%|████▋     | 400/864 [07:40<06:57,  1.11it/s]
                                                 
{'loss': '4.792', 'grad_norm': '0.3209', 'learning_rate': '0.0006', 'epoch': '5.558'}
+
 46%|████▋     | 400/864 [07:40<06:57,  1.11it/s]
 46%|████▋     | 401/864 [07:40<06:56,  1.11it/s]
 47%|████▋     | 402/864 [07:41<06:55,  1.11it/s]
 47%|████▋     | 403/864 [07:42<06:54,  1.11it/s]
 47%|████▋     | 404/864 [07:43<06:53,  1.11it/s]
 47%|████▋     | 405/864 [07:44<06:53,  1.11it/s]
 47%|████▋     | 406/864 [07:45<06:52,  1.11it/s]
 47%|████▋     | 407/864 [07:46<06:51,  1.11it/s]
 47%|████▋     | 408/864 [07:47<06:50,  1.11it/s]
 47%|████▋     | 409/864 [07:48<06:49,  1.11it/s]
 47%|████▋     | 410/864 [07:49<06:48,  1.11it/s]
                                                 
{'loss': '4.706', 'grad_norm': '0.2011', 'learning_rate': '0.0006', 'epoch': '5.698'}
+
 47%|████▋     | 410/864 [07:49<06:48,  1.11it/s]
 48%|████▊     | 411/864 [07:49<06:48,  1.11it/s]
 48%|████▊     | 412/864 [07:50<06:47,  1.11it/s]
 48%|████▊     | 413/864 [07:51<06:46,  1.11it/s]
 48%|████▊     | 414/864 [07:52<06:45,  1.11it/s]
 48%|████▊     | 415/864 [07:53<06:43,  1.11it/s]
 48%|████▊     | 416/864 [07:54<06:42,  1.11it/s]
 48%|████▊     | 417/864 [07:55<06:42,  1.11it/s]
 48%|████▊     | 418/864 [07:56<06:40,  1.11it/s]
 48%|████▊     | 419/864 [07:57<06:40,  1.11it/s]
 49%|████▊     | 420/864 [07:58<06:38,  1.11it/s]
                                                 
{'loss': '4.655', 'grad_norm': '0.5072', 'learning_rate': '0.0006', 'epoch': '5.838'}
+
 49%|████▊     | 420/864 [07:58<06:38,  1.11it/s]
 49%|████▊     | 421/864 [07:58<06:37,  1.11it/s]
 49%|████▉     | 422/864 [07:59<06:37,  1.11it/s]
 49%|████▉     | 423/864 [08:00<06:36,  1.11it/s]
 49%|████▉     | 424/864 [08:01<06:35,  1.11it/s]
 49%|████▉     | 425/864 [08:02<06:34,  1.11it/s]
 49%|████▉     | 426/864 [08:03<06:33,  1.11it/s]
 49%|████▉     | 427/864 [08:04<06:32,  1.11it/s]
 50%|████▉     | 428/864 [08:05<06:31,  1.11it/s]
 50%|████▉     | 429/864 [08:06<06:31,  1.11it/s]
 50%|████▉     | 430/864 [08:07<06:30,  1.11it/s]
                                                 
{'loss': '4.654', 'grad_norm': '0.5024', 'learning_rate': '0.0006', 'epoch': '5.977'}
+
 50%|████▉     | 430/864 [08:07<06:30,  1.11it/s]
 50%|████▉     | 431/864 [08:07<06:29,  1.11it/s]
 50%|█████     | 432/864 [08:08<05:45,  1.25it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:01, 20.00it/s]
+
 10%|█         | 4/39 [00:00<00:02, 12.17it/s]
+
 15%|█▌        | 6/39 [00:00<00:03, 10.88it/s]
+
 21%|██        | 8/39 [00:00<00:02, 10.33it/s]
+
 26%|██▌       | 10/39 [00:00<00:02, 10.08it/s]
+
 31%|███       | 12/39 [00:01<00:02,  9.92it/s]
+
 36%|███▌      | 14/39 [00:01<00:02,  9.83it/s]
+
 41%|████      | 16/39 [00:01<00:02,  9.74it/s]
+
 44%|████▎     | 17/39 [00:01<00:02,  9.71it/s]
+
 46%|████▌     | 18/39 [00:01<00:02,  9.72it/s]
+
 49%|████▊     | 19/39 [00:01<00:02,  9.68it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01,  9.68it/s]
+
 54%|█████▍    | 21/39 [00:02<00:01,  9.64it/s]
+
 56%|█████▋    | 22/39 [00:02<00:01,  9.66it/s]
+
 59%|█████▉    | 23/39 [00:02<00:01,  9.65it/s]
+
 62%|██████▏   | 24/39 [00:02<00:01,  9.66it/s]
+
 64%|██████▍   | 25/39 [00:02<00:01,  9.66it/s]
+
 67%|██████▋   | 26/39 [00:02<00:01,  9.65it/s]
+
 69%|██████▉   | 27/39 [00:02<00:01,  9.64it/s]
+
 72%|███████▏  | 28/39 [00:02<00:01,  9.65it/s]
+
 74%|███████▍  | 29/39 [00:02<00:01,  9.65it/s]
+
 77%|███████▋  | 30/39 [00:03<00:00,  9.66it/s]
+
 79%|███████▉  | 31/39 [00:03<00:00,  9.65it/s]
+
 82%|████████▏ | 32/39 [00:03<00:00,  9.65it/s]
+
 85%|████████▍ | 33/39 [00:03<00:00,  9.65it/s]
+
 87%|████████▋ | 34/39 [00:03<00:00,  9.65it/s]
+
 90%|████████▉ | 35/39 [00:03<00:00,  9.65it/s]
+
 92%|█████████▏| 36/39 [00:03<00:00,  9.66it/s]
+
 95%|█████████▍| 37/39 [00:03<00:00,  9.66it/s]
+
 97%|█████████▋| 38/39 [00:03<00:00,  9.66it/s]
+
100%|██████████| 39/39 [00:03<00:00,  9.66it/s]
                                                 
+
                                               
{'eval_loss': '4.724', 'eval_runtime': '4.113', 'eval_samples_per_second': '593.5', 'eval_steps_per_second': '9.482', 'epoch': '6'}
+
 50%|█████     | 432/864 [08:12<05:45,  1.25it/s]
+
100%|██████████| 39/39 [00:04<00:00,  9.66it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-432
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-432/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-432/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:01<00:00,  1.03s/it]
Writing model shards: 100%|██████████| 1/1 [00:01<00:00,  1.03s/it]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-432/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-432/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-432/tokenizer_config.json
+[eval] epoch 6.0 step 432 loss 4.7242 (best 4.7242 @ ep 6.0, 0/10000 since best)
+
+
                                               
 50%|█████     | 433/864 [08:17<22:35,  3.15s/it]
 50%|█████     | 434/864 [08:18<17:42,  2.47s/it]
 50%|█████     | 435/864 [08:18<14:17,  2.00s/it]
 50%|█████     | 436/864 [08:19<11:53,  1.67s/it]
 51%|█████     | 437/864 [08:20<10:13,  1.44s/it]
 51%|█████     | 438/864 [08:21<09:03,  1.28s/it]
 51%|█████     | 439/864 [08:22<08:14,  1.16s/it]
 51%|█████     | 440/864 [08:23<07:39,  1.08s/it]
                                                 
{'loss': '4.563', 'grad_norm': '0.2405', 'learning_rate': '0.0006', 'epoch': '6.112'}
+
 51%|█████     | 440/864 [08:23<07:39,  1.08s/it]
 51%|█████     | 441/864 [08:24<07:15,  1.03s/it]
 51%|█████     | 442/864 [08:25<06:58,  1.01it/s]
 51%|█████▏    | 443/864 [08:26<06:45,  1.04it/s]
 51%|█████▏    | 444/864 [08:26<06:36,  1.06it/s]
 52%|█████▏    | 445/864 [08:27<06:30,  1.07it/s]
 52%|█████▏    | 446/864 [08:28<06:25,  1.08it/s]
 52%|█████▏    | 447/864 [08:29<06:23,  1.09it/s]
 52%|█████▏    | 448/864 [08:30<06:19,  1.10it/s]
 52%|█████▏    | 449/864 [08:31<06:17,  1.10it/s]
 52%|█████▏    | 450/864 [08:32<06:15,  1.10it/s]
                                                 
{'loss': '4.496', 'grad_norm': '0.4446', 'learning_rate': '0.0006', 'epoch': '6.251'}
+
 52%|█████▏    | 450/864 [08:32<06:15,  1.10it/s]
 52%|█████▏    | 451/864 [08:33<06:14,  1.10it/s]
 52%|█████▏    | 452/864 [08:34<06:12,  1.11it/s]
 52%|█████▏    | 453/864 [08:35<06:11,  1.11it/s]
 53%|█████▎    | 454/864 [08:36<06:10,  1.11it/s]
 53%|█████▎    | 455/864 [08:36<06:09,  1.11it/s]
 53%|█████▎    | 456/864 [08:37<06:08,  1.11it/s]
 53%|█████▎    | 457/864 [08:38<06:06,  1.11it/s]
 53%|█████▎    | 458/864 [08:39<06:05,  1.11it/s]
 53%|█████▎    | 459/864 [08:40<06:04,  1.11it/s]
 53%|█████▎    | 460/864 [08:41<06:03,  1.11it/s]
                                                 
{'loss': '4.462', 'grad_norm': '0.4215', 'learning_rate': '0.0006', 'epoch': '6.391'}
+
 53%|█████▎    | 460/864 [08:41<06:03,  1.11it/s]
 53%|█████▎    | 461/864 [08:42<06:02,  1.11it/s]
 53%|█████▎    | 462/864 [08:43<06:01,  1.11it/s]
 54%|█████▎    | 463/864 [08:44<06:00,  1.11it/s]
 54%|█████▎    | 464/864 [08:45<05:59,  1.11it/s]
 54%|█████▍    | 465/864 [08:45<05:58,  1.11it/s]
 54%|█████▍    | 466/864 [08:46<05:57,  1.11it/s]
 54%|█████▍    | 467/864 [08:47<05:56,  1.11it/s]
 54%|█████▍    | 468/864 [08:48<05:55,  1.11it/s]
 54%|█████▍    | 469/864 [08:49<05:54,  1.11it/s]
 54%|█████▍    | 470/864 [08:50<05:53,  1.11it/s]
                                                 
{'loss': '4.436', 'grad_norm': '0.393', 'learning_rate': '0.0006', 'epoch': '6.531'}
+
 54%|█████▍    | 470/864 [08:50<05:53,  1.11it/s]
 55%|█████▍    | 471/864 [08:51<05:53,  1.11it/s]
 55%|█████▍    | 472/864 [08:52<05:52,  1.11it/s]
 55%|█████▍    | 473/864 [08:53<05:51,  1.11it/s]
 55%|█████▍    | 474/864 [08:54<05:50,  1.11it/s]
 55%|█████▍    | 475/864 [08:54<05:49,  1.11it/s]
 55%|█████▌    | 476/864 [08:55<05:48,  1.11it/s]
 55%|█████▌    | 477/864 [08:56<05:47,  1.11it/s]
 55%|█████▌    | 478/864 [08:57<05:46,  1.11it/s]
 55%|█████▌    | 479/864 [08:58<05:46,  1.11it/s]
 56%|█████▌    | 480/864 [08:59<05:45,  1.11it/s]
                                                 
{'loss': '4.4', 'grad_norm': '0.4589', 'learning_rate': '0.0006', 'epoch': '6.67'}
+
 56%|█████▌    | 480/864 [08:59<05:45,  1.11it/s]
 56%|█████▌    | 481/864 [09:00<05:44,  1.11it/s]
 56%|█████▌    | 482/864 [09:01<05:43,  1.11it/s]
 56%|█████▌    | 483/864 [09:02<05:42,  1.11it/s]
 56%|█████▌    | 484/864 [09:02<05:41,  1.11it/s]
 56%|█████▌    | 485/864 [09:03<05:40,  1.11it/s]
 56%|█████▋    | 486/864 [09:04<05:39,  1.11it/s]
 56%|█████▋    | 487/864 [09:05<05:38,  1.11it/s]
 56%|█████▋    | 488/864 [09:06<05:37,  1.11it/s]
 57%|█████▋    | 489/864 [09:07<05:37,  1.11it/s]
 57%|█████▋    | 490/864 [09:08<05:36,  1.11it/s]
                                                 
{'loss': '4.42', 'grad_norm': '0.5056', 'learning_rate': '0.0006', 'epoch': '6.81'}
+
 57%|█████▋    | 490/864 [09:08<05:36,  1.11it/s]
 57%|█████▋    | 491/864 [09:09<05:35,  1.11it/s]
 57%|█████▋    | 492/864 [09:10<05:34,  1.11it/s]
 57%|█████▋    | 493/864 [09:11<05:33,  1.11it/s]
 57%|█████▋    | 494/864 [09:11<05:32,  1.11it/s]
 57%|█████▋    | 495/864 [09:12<05:31,  1.11it/s]
 57%|█████▋    | 496/864 [09:13<05:30,  1.11it/s]
 58%|█████▊    | 497/864 [09:14<05:29,  1.11it/s]
 58%|█████▊    | 498/864 [09:15<05:29,  1.11it/s]
 58%|█████▊    | 499/864 [09:16<05:27,  1.11it/s]
 58%|█████▊    | 500/864 [09:17<05:27,  1.11it/s]
                                                 
{'loss': '4.36', 'grad_norm': '0.2333', 'learning_rate': '0.0006', 'epoch': '6.949'}
+
 58%|█████▊    | 500/864 [09:17<05:27,  1.11it/s]
 58%|█████▊    | 501/864 [09:18<05:26,  1.11it/s]
 58%|█████▊    | 502/864 [09:19<05:25,  1.11it/s]
 58%|█████▊    | 503/864 [09:20<05:24,  1.11it/s]
 58%|█████▊    | 504/864 [09:20<04:47,  1.25it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:02, 14.70it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 11.70it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 10.75it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 10.32it/s]
+
 28%|██▊       | 11/39 [00:01<00:02, 10.07it/s]
+
 33%|███▎      | 13/39 [00:01<00:02,  9.92it/s]
+
 36%|███▌      | 14/39 [00:01<00:02,  9.88it/s]
+
 38%|███▊      | 15/39 [00:01<00:02,  9.81it/s]
+
 41%|████      | 16/39 [00:01<00:02,  9.77it/s]
+
 44%|████▎     | 17/39 [00:01<00:02,  9.74it/s]
+
 46%|████▌     | 18/39 [00:01<00:02,  9.73it/s]
+
 49%|████▊     | 19/39 [00:01<00:02,  9.69it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01,  9.68it/s]
+
 54%|█████▍    | 21/39 [00:02<00:01,  9.67it/s]
+
 56%|█████▋    | 22/39 [00:02<00:01,  9.68it/s]
+
 59%|█████▉    | 23/39 [00:02<00:01,  9.67it/s]
+
 62%|██████▏   | 24/39 [00:02<00:01,  9.67it/s]
+
 64%|██████▍   | 25/39 [00:02<00:01,  9.64it/s]
+
 67%|██████▋   | 26/39 [00:02<00:01,  9.66it/s]
+
 69%|██████▉   | 27/39 [00:02<00:01,  9.66it/s]
+
 72%|███████▏  | 28/39 [00:02<00:01,  9.66it/s]
+
 74%|███████▍  | 29/39 [00:02<00:01,  9.65it/s]
+
 77%|███████▋  | 30/39 [00:03<00:00,  9.65it/s]
+
 79%|███████▉  | 31/39 [00:03<00:00,  9.66it/s]
+
 82%|████████▏ | 32/39 [00:03<00:00,  9.66it/s]
+
 85%|████████▍ | 33/39 [00:03<00:00,  9.65it/s]
+
 87%|████████▋ | 34/39 [00:03<00:00,  9.66it/s]
+
 90%|████████▉ | 35/39 [00:03<00:00,  9.64it/s]
+
 92%|█████████▏| 36/39 [00:03<00:00,  9.63it/s]
+
 95%|█████████▍| 37/39 [00:03<00:00,  9.64it/s]
+
 97%|█████████▋| 38/39 [00:03<00:00,  9.64it/s]
+
100%|██████████| 39/39 [00:03<00:00,  9.66it/s]
                                                 
+
                                               
{'eval_loss': '4.463', 'eval_runtime': '4.101', 'eval_samples_per_second': '595.3', 'eval_steps_per_second': '9.511', 'epoch': '7'}
+
 58%|█████▊    | 504/864 [09:24<04:47,  1.25it/s]
+
100%|██████████| 39/39 [00:04<00:00,  9.66it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-504
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-504/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-504/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.13it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.13it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-504/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-504/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-504/tokenizer_config.json
+[eval] epoch 7.0 step 504 loss 4.4628 (best 4.4628 @ ep 7.0, 0/10000 since best)
+
+
                                               
 58%|█████▊    | 505/864 [09:29<18:29,  3.09s/it]
 59%|█████▊    | 506/864 [09:29<14:30,  2.43s/it]
 59%|█████▊    | 507/864 [09:30<11:43,  1.97s/it]
 59%|█████▉    | 508/864 [09:31<09:47,  1.65s/it]
 59%|█████▉    | 509/864 [09:32<08:25,  1.42s/it]
 59%|█████▉    | 510/864 [09:33<07:28,  1.27s/it]
                                                 
{'loss': '4.285', 'grad_norm': '0.4354', 'learning_rate': '0.0006', 'epoch': '7.084'}
+
 59%|█████▉    | 510/864 [09:33<07:28,  1.27s/it]
 59%|█████▉    | 511/864 [09:34<06:48,  1.16s/it]
 59%|█████▉    | 512/864 [09:35<06:19,  1.08s/it]
 59%|█████▉    | 513/864 [09:36<05:59,  1.03s/it]
 59%|█████▉    | 514/864 [09:37<05:45,  1.01it/s]
 60%|█████▉    | 515/864 [09:38<05:35,  1.04it/s]
 60%|█████▉    | 516/864 [09:38<05:27,  1.06it/s]
 60%|█████▉    | 517/864 [09:39<05:22,  1.08it/s]
 60%|█████▉    | 518/864 [09:40<05:18,  1.09it/s]
 60%|██████    | 519/864 [09:41<05:15,  1.09it/s]
 60%|██████    | 520/864 [09:42<05:13,  1.10it/s]
                                                 
{'loss': '4.247', 'grad_norm': '0.361', 'learning_rate': '0.0006', 'epoch': '7.223'}
+
 60%|██████    | 520/864 [09:42<05:13,  1.10it/s]
 60%|██████    | 521/864 [09:43<05:11,  1.10it/s]
 60%|██████    | 522/864 [09:44<05:09,  1.10it/s]
 61%|██████    | 523/864 [09:45<05:08,  1.11it/s]
 61%|██████    | 524/864 [09:46<05:07,  1.11it/s]
 61%|██████    | 525/864 [09:47<05:06,  1.11it/s]
 61%|██████    | 526/864 [09:47<05:04,  1.11it/s]
 61%|██████    | 527/864 [09:48<05:03,  1.11it/s]
 61%|██████    | 528/864 [09:49<05:02,  1.11it/s]
 61%|██████    | 529/864 [09:50<05:01,  1.11it/s]
 61%|██████▏   | 530/864 [09:51<05:00,  1.11it/s]
                                                 
{'loss': '4.243', 'grad_norm': '0.595', 'learning_rate': '0.0006', 'epoch': '7.363'}
+
 61%|██████▏   | 530/864 [09:51<05:00,  1.11it/s]
 61%|██████▏   | 531/864 [09:52<04:59,  1.11it/s]
 62%|██████▏   | 532/864 [09:53<04:58,  1.11it/s]
 62%|██████▏   | 533/864 [09:54<04:57,  1.11it/s]
 62%|██████▏   | 534/864 [09:55<04:57,  1.11it/s]
 62%|██████▏   | 535/864 [09:56<04:56,  1.11it/s]
 62%|██████▏   | 536/864 [09:56<04:55,  1.11it/s]
 62%|██████▏   | 537/864 [09:57<04:53,  1.11it/s]
 62%|██████▏   | 538/864 [09:58<04:53,  1.11it/s]
 62%|██████▏   | 539/864 [09:59<04:52,  1.11it/s]
 62%|██████▎   | 540/864 [10:00<04:51,  1.11it/s]
                                                 
{'loss': '4.254', 'grad_norm': '0.3657', 'learning_rate': '0.0006', 'epoch': '7.503'}
+
 62%|██████▎   | 540/864 [10:00<04:51,  1.11it/s]
 63%|██████▎   | 541/864 [10:01<04:50,  1.11it/s]
 63%|██████▎   | 542/864 [10:02<04:49,  1.11it/s]
 63%|██████▎   | 543/864 [10:03<04:48,  1.11it/s]
 63%|██████▎   | 544/864 [10:04<04:47,  1.11it/s]
 63%|██████▎   | 545/864 [10:05<04:46,  1.11it/s]
 63%|██████▎   | 546/864 [10:05<04:45,  1.11it/s]
 63%|██████▎   | 547/864 [10:06<04:44,  1.11it/s]
 63%|██████▎   | 548/864 [10:07<04:43,  1.11it/s]
 64%|██████▎   | 549/864 [10:08<04:42,  1.11it/s]
 64%|██████▎   | 550/864 [10:09<04:41,  1.11it/s]
                                                 
{'loss': '4.206', 'grad_norm': '0.2189', 'learning_rate': '0.0006', 'epoch': '7.642'}
+
 64%|██████▎   | 550/864 [10:09<04:41,  1.11it/s]
 64%|██████▍   | 551/864 [10:10<04:41,  1.11it/s]
 64%|██████▍   | 552/864 [10:11<04:40,  1.11it/s]
 64%|██████▍   | 553/864 [10:12<04:39,  1.11it/s]
 64%|██████▍   | 554/864 [10:13<04:38,  1.11it/s]
 64%|██████▍   | 555/864 [10:14<04:37,  1.11it/s]
 64%|██████▍   | 556/864 [10:14<04:36,  1.11it/s]
 64%|██████▍   | 557/864 [10:15<04:35,  1.11it/s]
 65%|██████▍   | 558/864 [10:16<04:34,  1.11it/s]
 65%|██████▍   | 559/864 [10:17<04:34,  1.11it/s]
 65%|██████▍   | 560/864 [10:18<04:33,  1.11it/s]
                                                 
{'loss': '4.172', 'grad_norm': '0.2015', 'learning_rate': '0.0006', 'epoch': '7.782'}
+
 65%|██████▍   | 560/864 [10:18<04:33,  1.11it/s]
 65%|██████▍   | 561/864 [10:19<04:32,  1.11it/s]
 65%|██████▌   | 562/864 [10:20<04:31,  1.11it/s]
 65%|██████▌   | 563/864 [10:21<04:30,  1.11it/s]
 65%|██████▌   | 564/864 [10:22<04:29,  1.11it/s]
 65%|██████▌   | 565/864 [10:23<04:28,  1.11it/s]
 66%|██████▌   | 566/864 [10:23<04:27,  1.11it/s]
 66%|██████▌   | 567/864 [10:24<04:26,  1.11it/s]
 66%|██████▌   | 568/864 [10:25<04:25,  1.12it/s]
 66%|██████▌   | 569/864 [10:26<04:24,  1.11it/s]
 66%|██████▌   | 570/864 [10:27<04:23,  1.11it/s]
                                                 
{'loss': '4.147', 'grad_norm': '0.3295', 'learning_rate': '0.0006', 'epoch': '7.921'}
+
 66%|██████▌   | 570/864 [10:27<04:23,  1.11it/s]
 66%|██████▌   | 571/864 [10:28<04:23,  1.11it/s]
 66%|██████▌   | 572/864 [10:29<04:22,  1.11it/s]
 66%|██████▋   | 573/864 [10:30<04:21,  1.11it/s]
 66%|██████▋   | 574/864 [10:31<04:20,  1.11it/s]
 67%|██████▋   | 575/864 [10:31<04:19,  1.11it/s]
 67%|██████▋   | 576/864 [10:32<03:50,  1.25it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:02, 14.72it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 11.66it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 10.77it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 10.31it/s]
+
 28%|██▊       | 11/39 [00:01<00:02, 10.08it/s]
+
 33%|███▎      | 13/39 [00:01<00:02,  9.91it/s]
+
 36%|███▌      | 14/39 [00:01<00:02,  9.88it/s]
+
 38%|███▊      | 15/39 [00:01<00:02,  9.84it/s]
+
 41%|████      | 16/39 [00:01<00:02,  9.78it/s]
+
 44%|████▎     | 17/39 [00:01<00:02,  9.75it/s]
+
 46%|████▌     | 18/39 [00:01<00:02,  9.72it/s]
+
 49%|████▊     | 19/39 [00:01<00:02,  9.72it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01,  9.69it/s]
+
 54%|█████▍    | 21/39 [00:02<00:01,  9.68it/s]
+
 56%|█████▋    | 22/39 [00:02<00:01,  9.67it/s]
+
 59%|█████▉    | 23/39 [00:02<00:01,  9.68it/s]
+
 62%|██████▏   | 24/39 [00:02<00:01,  9.67it/s]
+
 64%|██████▍   | 25/39 [00:02<00:01,  9.67it/s]
+
 67%|██████▋   | 26/39 [00:02<00:01,  9.67it/s]
+
 69%|██████▉   | 27/39 [00:02<00:01,  9.66it/s]
+
 72%|███████▏  | 28/39 [00:02<00:01,  9.67it/s]
+
 74%|███████▍  | 29/39 [00:02<00:01,  9.66it/s]
+
 77%|███████▋  | 30/39 [00:03<00:00,  9.66it/s]
+
 79%|███████▉  | 31/39 [00:03<00:00,  9.66it/s]
+
 82%|████████▏ | 32/39 [00:03<00:00,  9.66it/s]
+
 85%|████████▍ | 33/39 [00:03<00:00,  9.66it/s]
+
 87%|████████▋ | 34/39 [00:03<00:00,  9.66it/s]
+
 90%|████████▉ | 35/39 [00:03<00:00,  9.66it/s]
+
 92%|█████████▏| 36/39 [00:03<00:00,  9.66it/s]
+
 95%|█████████▍| 37/39 [00:03<00:00,  9.66it/s]
+
 97%|█████████▋| 38/39 [00:03<00:00,  9.66it/s]
+
100%|██████████| 39/39 [00:03<00:00,  9.66it/s]
                                                 
+
                                               
{'eval_loss': '4.315', 'eval_runtime': '4.098', 'eval_samples_per_second': '595.7', 'eval_steps_per_second': '9.517', 'epoch': '8'}
+
 67%|██████▋   | 576/864 [10:36<03:50,  1.25it/s]
+
100%|██████████| 39/39 [00:04<00:00,  9.66it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-576
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-576/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-576/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.12it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.12it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-576/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-576/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-576/tokenizer_config.json
+[eval] epoch 8.0 step 576 loss 4.3148 (best 4.3148 @ ep 8.0, 0/10000 since best)
+
+
                                               
 67%|██████▋   | 577/864 [10:41<14:51,  3.11s/it]
 67%|██████▋   | 578/864 [10:41<11:38,  2.44s/it]
 67%|██████▋   | 579/864 [10:42<09:24,  1.98s/it]
 67%|██████▋   | 580/864 [10:43<07:49,  1.65s/it]
                                                 
{'loss': '4.101', 'grad_norm': '0.4736', 'learning_rate': '0.0006', 'epoch': '8.056'}
+
 67%|██████▋   | 580/864 [10:43<07:49,  1.65s/it]
 67%|██████▋   | 581/864 [10:44<06:43,  1.43s/it]
 67%|██████▋   | 582/864 [10:45<05:57,  1.27s/it]
 67%|██████▋   | 583/864 [10:46<05:25,  1.16s/it]
 68%|██████▊   | 584/864 [10:47<05:02,  1.08s/it]
 68%|██████▊   | 585/864 [10:48<04:45,  1.02s/it]
 68%|██████▊   | 586/864 [10:49<04:34,  1.01it/s]
 68%|██████▊   | 587/864 [10:50<04:25,  1.04it/s]
 68%|██████▊   | 588/864 [10:50<04:19,  1.06it/s]
 68%|██████▊   | 589/864 [10:51<04:15,  1.08it/s]
 68%|██████▊   | 590/864 [10:52<04:11,  1.09it/s]
                                                 
{'loss': '4.117', 'grad_norm': '0.7782', 'learning_rate': '0.0006', 'epoch': '8.195'}
+
 68%|██████▊   | 590/864 [10:52<04:11,  1.09it/s]
 68%|██████▊   | 591/864 [10:53<04:09,  1.10it/s]
 69%|██████▊   | 592/864 [10:54<04:06,  1.10it/s]
 69%|██████▊   | 593/864 [10:55<04:05,  1.11it/s]
 69%|██████▉   | 594/864 [10:56<04:03,  1.11it/s]
 69%|██████▉   | 595/864 [10:57<04:02,  1.11it/s]
 69%|██████▉   | 596/864 [10:58<04:01,  1.11it/s]
 69%|██████▉   | 597/864 [10:58<04:00,  1.11it/s]
 69%|██████▉   | 598/864 [10:59<03:59,  1.11it/s]
 69%|██████▉   | 599/864 [11:00<03:57,  1.11it/s]
 69%|██████▉   | 600/864 [11:01<03:56,  1.11it/s]
                                                 
{'loss': '4.237', 'grad_norm': '0.9462', 'learning_rate': '0.0006', 'epoch': '8.335'}
+
 69%|██████▉   | 600/864 [11:01<03:56,  1.11it/s]
 70%|██████▉   | 601/864 [11:02<03:56,  1.11it/s]
 70%|██████▉   | 602/864 [11:03<03:55,  1.11it/s]
 70%|██████▉   | 603/864 [11:04<03:54,  1.11it/s]
 70%|██████▉   | 604/864 [11:05<03:53,  1.11it/s]
 70%|███████   | 605/864 [11:06<03:52,  1.11it/s]
 70%|███████   | 606/864 [11:07<03:51,  1.11it/s]
 70%|███████   | 607/864 [11:07<03:50,  1.11it/s]
 70%|███████   | 608/864 [11:08<03:49,  1.11it/s]
 70%|███████   | 609/864 [11:09<03:48,  1.11it/s]
 71%|███████   | 610/864 [11:10<03:48,  1.11it/s]
                                                 
{'loss': '4.264', 'grad_norm': '0.516', 'learning_rate': '0.0006', 'epoch': '8.475'}
+
 71%|███████   | 610/864 [11:10<03:48,  1.11it/s]
 71%|███████   | 611/864 [11:11<03:47,  1.11it/s]
 71%|███████   | 612/864 [11:12<03:46,  1.11it/s]
 71%|███████   | 613/864 [11:13<03:45,  1.11it/s]
 71%|███████   | 614/864 [11:14<03:44,  1.11it/s]
 71%|███████   | 615/864 [11:15<03:43,  1.11it/s]
 71%|███████▏  | 616/864 [11:16<03:42,  1.11it/s]
 71%|███████▏  | 617/864 [11:16<03:41,  1.11it/s]
 72%|███████▏  | 618/864 [11:17<03:40,  1.11it/s]
 72%|███████▏  | 619/864 [11:18<03:39,  1.11it/s]
 72%|███████▏  | 620/864 [11:19<03:39,  1.11it/s]
                                                 
{'loss': '4.15', 'grad_norm': '0.1853', 'learning_rate': '0.0006', 'epoch': '8.614'}
+
 72%|███████▏  | 620/864 [11:19<03:39,  1.11it/s]
 72%|███████▏  | 621/864 [11:20<03:38,  1.11it/s]
 72%|███████▏  | 622/864 [11:21<03:37,  1.11it/s]
 72%|███████▏  | 623/864 [11:22<03:36,  1.11it/s]
 72%|███████▏  | 624/864 [11:23<03:35,  1.11it/s]
 72%|███████▏  | 625/864 [11:24<03:34,  1.11it/s]
 72%|███████▏  | 626/864 [11:25<03:33,  1.11it/s]
 73%|███████▎  | 627/864 [11:25<03:32,  1.11it/s]
 73%|███████▎  | 628/864 [11:26<03:31,  1.11it/s]
 73%|███████▎  | 629/864 [11:27<03:30,  1.11it/s]
 73%|███████▎  | 630/864 [11:28<03:29,  1.11it/s]
                                                 
{'loss': '4.079', 'grad_norm': '0.1461', 'learning_rate': '0.0006', 'epoch': '8.754'}
+
 73%|███████▎  | 630/864 [11:28<03:29,  1.11it/s]
 73%|███████▎  | 631/864 [11:29<03:29,  1.11it/s]
 73%|███████▎  | 632/864 [11:30<03:28,  1.11it/s]
 73%|███████▎  | 633/864 [11:31<03:27,  1.12it/s]
 73%|███████▎  | 634/864 [11:32<03:26,  1.11it/s]
 73%|███████▎  | 635/864 [11:33<03:25,  1.11it/s]
 74%|███████▎  | 636/864 [11:33<03:24,  1.11it/s]
 74%|███████▎  | 637/864 [11:34<03:23,  1.11it/s]
 74%|███████▍  | 638/864 [11:35<03:22,  1.11it/s]
 74%|███████▍  | 639/864 [11:36<03:21,  1.11it/s]
 74%|███████▍  | 640/864 [11:37<03:21,  1.11it/s]
                                                 
{'loss': '4.053', 'grad_norm': '0.152', 'learning_rate': '0.0006', 'epoch': '8.894'}
+
 74%|███████▍  | 640/864 [11:37<03:21,  1.11it/s]
 74%|███████▍  | 641/864 [11:38<03:20,  1.11it/s]
 74%|███████▍  | 642/864 [11:39<03:19,  1.11it/s]
 74%|███████▍  | 643/864 [11:40<03:18,  1.11it/s]
 75%|███████▍  | 644/864 [11:41<03:17,  1.11it/s]
 75%|███████▍  | 645/864 [11:42<03:16,  1.11it/s]
 75%|███████▍  | 646/864 [11:42<03:15,  1.11it/s]
 75%|███████▍  | 647/864 [11:43<03:14,  1.11it/s]
 75%|███████▌  | 648/864 [11:44<02:52,  1.25it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:01, 19.91it/s]
+
 10%|█         | 4/39 [00:00<00:02, 12.23it/s]
+
 15%|█▌        | 6/39 [00:00<00:03, 10.91it/s]
+
 21%|██        | 8/39 [00:00<00:02, 10.37it/s]
+
 26%|██▌       | 10/39 [00:00<00:02, 10.10it/s]
+
 31%|███       | 12/39 [00:01<00:02,  9.94it/s]
+
 36%|███▌      | 14/39 [00:01<00:02,  9.85it/s]
+
 41%|████      | 16/39 [00:01<00:02,  9.75it/s]
+
 44%|████▎     | 17/39 [00:01<00:02,  9.75it/s]
+
 46%|████▌     | 18/39 [00:01<00:02,  9.74it/s]
+
 49%|████▊     | 19/39 [00:01<00:02,  9.73it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01,  9.69it/s]
+
 54%|█████▍    | 21/39 [00:02<00:01,  9.67it/s]
+
 56%|█████▋    | 22/39 [00:02<00:01,  9.67it/s]
+
 59%|█████▉    | 23/39 [00:02<00:01,  9.68it/s]
+
 62%|██████▏   | 24/39 [00:02<00:01,  9.67it/s]
+
 64%|██████▍   | 25/39 [00:02<00:01,  9.67it/s]
+
 67%|██████▋   | 26/39 [00:02<00:01,  9.67it/s]
+
 69%|██████▉   | 27/39 [00:02<00:01,  9.67it/s]
+
 72%|███████▏  | 28/39 [00:02<00:01,  9.66it/s]
+
 74%|███████▍  | 29/39 [00:02<00:01,  9.67it/s]
+
 77%|███████▋  | 30/39 [00:03<00:00,  9.67it/s]
+
 79%|███████▉  | 31/39 [00:03<00:00,  9.67it/s]
+
 82%|████████▏ | 32/39 [00:03<00:00,  9.66it/s]
+
 85%|████████▍ | 33/39 [00:03<00:00,  9.66it/s]
+
 87%|████████▋ | 34/39 [00:03<00:00,  9.65it/s]
+
 90%|████████▉ | 35/39 [00:03<00:00,  9.65it/s]
+
 92%|█████████▏| 36/39 [00:03<00:00,  9.65it/s]
+
 95%|█████████▍| 37/39 [00:03<00:00,  9.65it/s]
+
 97%|█████████▋| 38/39 [00:03<00:00,  9.66it/s]
+
100%|██████████| 39/39 [00:03<00:00,  9.66it/s]
                                                 
+
                                               
{'eval_loss': '4.224', 'eval_runtime': '4.099', 'eval_samples_per_second': '595.6', 'eval_steps_per_second': '9.515', 'epoch': '9'}
+
 75%|███████▌  | 648/864 [11:48<02:52,  1.25it/s]
+
100%|██████████| 39/39 [00:04<00:00,  9.66it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-648
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-648/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-648/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.13it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.13it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-648/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-648/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-648/tokenizer_config.json
+[eval] epoch 9.0 step 648 loss 4.2243 (best 4.2243 @ ep 9.0, 0/10000 since best)
+
+
                                               
 75%|███████▌  | 649/864 [11:52<11:06,  3.10s/it]
 75%|███████▌  | 650/864 [11:53<08:41,  2.44s/it]
                                                 
{'loss': '4.001', 'grad_norm': '0.179', 'learning_rate': '0.0006', 'epoch': '9.028'}
+
 75%|███████▌  | 650/864 [11:53<08:41,  2.44s/it]
 75%|███████▌  | 651/864 [11:54<07:00,  1.98s/it]
 75%|███████▌  | 652/864 [11:55<05:50,  1.65s/it]
 76%|███████▌  | 653/864 [11:56<05:00,  1.42s/it]
 76%|███████▌  | 654/864 [11:57<04:25,  1.27s/it]
 76%|███████▌  | 655/864 [11:58<04:01,  1.16s/it]
 76%|███████▌  | 656/864 [11:59<03:44,  1.08s/it]
 76%|███████▌  | 657/864 [12:00<03:31,  1.02s/it]
 76%|███████▌  | 658/864 [12:00<03:23,  1.01it/s]
 76%|███████▋  | 659/864 [12:01<03:16,  1.04it/s]
 76%|███████▋  | 660/864 [12:02<03:11,  1.06it/s]
                                                 
{'loss': '3.923', 'grad_norm': '0.1812', 'learning_rate': '0.0006', 'epoch': '9.168'}
+
 76%|███████▋  | 660/864 [12:02<03:11,  1.06it/s]
 77%|███████▋  | 661/864 [12:03<03:08,  1.08it/s]
 77%|███████▋  | 662/864 [12:04<03:05,  1.09it/s]
 77%|███████▋  | 663/864 [12:05<03:03,  1.10it/s]
 77%|███████▋  | 664/864 [12:06<03:01,  1.10it/s]
 77%|███████▋  | 665/864 [12:07<03:00,  1.11it/s]
 77%|███████▋  | 666/864 [12:08<02:58,  1.11it/s]
 77%|███████▋  | 667/864 [12:09<02:57,  1.11it/s]
 77%|███████▋  | 668/864 [12:09<02:56,  1.11it/s]
 77%|███████▋  | 669/864 [12:10<02:55,  1.11it/s]
 78%|███████▊  | 670/864 [12:11<02:54,  1.11it/s]
                                                 
{'loss': '3.92', 'grad_norm': '0.4014', 'learning_rate': '0.0006', 'epoch': '9.307'}
+
 78%|███████▊  | 670/864 [12:11<02:54,  1.11it/s]
 78%|███████▊  | 671/864 [12:12<02:53,  1.11it/s]
 78%|███████▊  | 672/864 [12:13<02:52,  1.11it/s]
 78%|███████▊  | 673/864 [12:14<02:51,  1.11it/s]
 78%|███████▊  | 674/864 [12:15<02:50,  1.11it/s]
 78%|███████▊  | 675/864 [12:16<02:49,  1.11it/s]
 78%|███████▊  | 676/864 [12:17<02:48,  1.11it/s]
 78%|███████▊  | 677/864 [12:18<02:47,  1.12it/s]
 78%|███████▊  | 678/864 [12:18<02:46,  1.11it/s]
 79%|███████▊  | 679/864 [12:19<02:45,  1.11it/s]
 79%|███████▊  | 680/864 [12:20<02:45,  1.11it/s]
                                                 
{'loss': '3.921', 'grad_norm': '0.1859', 'learning_rate': '0.0006', 'epoch': '9.447'}
+
 79%|███████▊  | 680/864 [12:20<02:45,  1.11it/s]
 79%|███████▉  | 681/864 [12:21<02:44,  1.11it/s]
 79%|███████▉  | 682/864 [12:22<02:43,  1.11it/s]
 79%|███████▉  | 683/864 [12:23<02:42,  1.11it/s]
 79%|███████▉  | 684/864 [12:24<02:41,  1.11it/s]
 79%|███████▉  | 685/864 [12:25<02:40,  1.11it/s]
 79%|███████▉  | 686/864 [12:26<02:39,  1.11it/s]
 80%|███████▉  | 687/864 [12:26<02:38,  1.11it/s]
 80%|███████▉  | 688/864 [12:27<02:38,  1.11it/s]
 80%|███████▉  | 689/864 [12:28<02:37,  1.11it/s]
 80%|███████▉  | 690/864 [12:29<02:36,  1.11it/s]
                                                 
{'loss': '3.895', 'grad_norm': '0.178', 'learning_rate': '0.0006', 'epoch': '9.586'}
+
 80%|███████▉  | 690/864 [12:29<02:36,  1.11it/s]
 80%|███████▉  | 691/864 [12:30<02:35,  1.11it/s]
 80%|████████  | 692/864 [12:31<02:34,  1.11it/s]
 80%|████████  | 693/864 [12:32<02:33,  1.11it/s]
 80%|████████  | 694/864 [12:33<02:32,  1.11it/s]
 80%|████████  | 695/864 [12:34<02:31,  1.11it/s]
 81%|████████  | 696/864 [12:35<02:30,  1.11it/s]
 81%|████████  | 697/864 [12:35<02:29,  1.11it/s]
 81%|████████  | 698/864 [12:36<02:29,  1.11it/s]
 81%|████████  | 699/864 [12:37<02:28,  1.11it/s]
 81%|████████  | 700/864 [12:38<02:27,  1.11it/s]
                                                 
{'loss': '3.896', 'grad_norm': '0.2272', 'learning_rate': '0.0006', 'epoch': '9.726'}
+
 81%|████████  | 700/864 [12:38<02:27,  1.11it/s]
 81%|████████  | 701/864 [12:39<02:26,  1.11it/s]
 81%|████████▏ | 702/864 [12:40<02:25,  1.11it/s]
 81%|████████▏ | 703/864 [12:41<02:24,  1.11it/s]
 81%|████████▏ | 704/864 [12:42<02:23,  1.11it/s]
 82%|████████▏ | 705/864 [12:43<02:22,  1.11it/s]
 82%|████████▏ | 706/864 [12:44<02:21,  1.11it/s]
 82%|████████▏ | 707/864 [12:44<02:20,  1.11it/s]
 82%|████████▏ | 708/864 [12:45<02:20,  1.11it/s]
 82%|████████▏ | 709/864 [12:46<02:19,  1.11it/s]
 82%|████████▏ | 710/864 [12:47<02:18,  1.11it/s]
                                                 
{'loss': '3.877', 'grad_norm': '0.1844', 'learning_rate': '0.0006', 'epoch': '9.866'}
+
 82%|████████▏ | 710/864 [12:47<02:18,  1.11it/s]
 82%|████████▏ | 711/864 [12:48<02:17,  1.11it/s]
 82%|████████▏ | 712/864 [12:49<02:16,  1.11it/s]
 83%|████████▎ | 713/864 [12:50<02:15,  1.11it/s]
 83%|████████▎ | 714/864 [12:51<02:14,  1.11it/s]
 83%|████████▎ | 715/864 [12:52<02:13,  1.11it/s]
 83%|████████▎ | 716/864 [12:53<02:12,  1.11it/s]
 83%|████████▎ | 717/864 [12:53<02:11,  1.11it/s]
 83%|████████▎ | 718/864 [12:54<02:11,  1.11it/s]
 83%|████████▎ | 719/864 [12:55<02:10,  1.12it/s]
 83%|████████▎ | 720/864 [12:56<01:54,  1.25it/s]
                                                 
{'loss': '3.859', 'grad_norm': '0.2214', 'learning_rate': '0.0006', 'epoch': '10'}
+
 83%|████████▎ | 720/864 [12:56<01:54,  1.25it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:01, 18.93it/s]
+
 10%|█         | 4/39 [00:00<00:02, 12.16it/s]
+
 15%|█▌        | 6/39 [00:00<00:03, 10.85it/s]
+
 21%|██        | 8/39 [00:00<00:02, 10.34it/s]
+
 26%|██▌       | 10/39 [00:00<00:02, 10.07it/s]
+
 31%|███       | 12/39 [00:01<00:02,  9.92it/s]
+
 36%|███▌      | 14/39 [00:01<00:02,  9.82it/s]
+
 38%|███▊      | 15/39 [00:01<00:02,  9.79it/s]
+
 41%|████      | 16/39 [00:01<00:02,  9.76it/s]
+
 44%|████▎     | 17/39 [00:01<00:02,  9.73it/s]
+
 46%|████▌     | 18/39 [00:01<00:02,  9.71it/s]
+
 49%|████▊     | 19/39 [00:01<00:02,  9.70it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01,  9.68it/s]
+
 54%|█████▍    | 21/39 [00:02<00:01,  9.67it/s]
+
 56%|█████▋    | 22/39 [00:02<00:01,  9.66it/s]
+
 59%|█████▉    | 23/39 [00:02<00:01,  9.68it/s]
+
 62%|██████▏   | 24/39 [00:02<00:01,  9.68it/s]
+
 64%|██████▍   | 25/39 [00:02<00:01,  9.67it/s]
+
 67%|██████▋   | 26/39 [00:02<00:01,  9.67it/s]
+
 69%|██████▉   | 27/39 [00:02<00:01,  9.67it/s]
+
 72%|███████▏  | 28/39 [00:02<00:01,  9.67it/s]
+
 74%|███████▍  | 29/39 [00:02<00:01,  9.67it/s]
+
 77%|███████▋  | 30/39 [00:03<00:00,  9.66it/s]
+
 79%|███████▉  | 31/39 [00:03<00:00,  9.66it/s]
+
 82%|████████▏ | 32/39 [00:03<00:00,  9.66it/s]
+
 85%|████████▍ | 33/39 [00:03<00:00,  9.66it/s]
+
 87%|████████▋ | 34/39 [00:03<00:00,  9.66it/s]
+
 90%|████████▉ | 35/39 [00:03<00:00,  9.66it/s]
+
 92%|█████████▏| 36/39 [00:03<00:00,  9.66it/s]
+
 95%|█████████▍| 37/39 [00:03<00:00,  9.66it/s]
+
 97%|█████████▋| 38/39 [00:03<00:00,  9.66it/s]
+
100%|██████████| 39/39 [00:03<00:00,  9.65it/s]
                                                 
+
                                               
{'eval_loss': '4.13', 'eval_runtime': '4.097', 'eval_samples_per_second': '595.7', 'eval_steps_per_second': '9.518', 'epoch': '10'}
+
 83%|████████▎ | 720/864 [13:00<01:54,  1.25it/s]
+
100%|██████████| 39/39 [00:04<00:00,  9.65it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-720
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-720/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-720/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.12it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.12it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-720/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-720/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-720/tokenizer_config.json
+[eval] epoch 10.0 step 720 loss 4.1301 (best 4.1301 @ ep 10.0, 0/10000 since best)
+
+
                                               
 83%|████████▎ | 721/864 [13:04<07:23,  3.10s/it]
 84%|████████▎ | 722/864 [13:05<05:46,  2.44s/it]
 84%|████████▎ | 723/864 [13:06<04:38,  1.98s/it]
 84%|████████▍ | 724/864 [13:07<03:51,  1.65s/it]
 84%|████████▍ | 725/864 [13:08<03:18,  1.43s/it]
 84%|████████▍ | 726/864 [13:09<02:54,  1.27s/it]
 84%|████████▍ | 727/864 [13:10<02:38,  1.15s/it]
 84%|████████▍ | 728/864 [13:11<02:26,  1.08s/it]
 84%|████████▍ | 729/864 [13:11<02:18,  1.02s/it]
 84%|████████▍ | 730/864 [13:12<02:12,  1.01it/s]
                                                 
{'loss': '3.778', 'grad_norm': '0.4108', 'learning_rate': '0.0006', 'epoch': '10.14'}
+
 84%|████████▍ | 730/864 [13:12<02:12,  1.01it/s]
 85%|████████▍ | 731/864 [13:13<02:07,  1.04it/s]
 85%|████████▍ | 732/864 [13:14<02:04,  1.06it/s]
 85%|████████▍ | 733/864 [13:15<02:01,  1.08it/s]
 85%|████████▍ | 734/864 [13:16<01:59,  1.09it/s]
 85%|████████▌ | 735/864 [13:17<01:57,  1.10it/s]
 85%|████████▌ | 736/864 [13:18<01:56,  1.10it/s]
 85%|████████▌ | 737/864 [13:19<01:54,  1.11it/s]
 85%|████████▌ | 738/864 [13:20<01:53,  1.11it/s]
 86%|████████▌ | 739/864 [13:20<01:52,  1.11it/s]
 86%|████████▌ | 740/864 [13:21<01:51,  1.11it/s]
                                                 
{'loss': '3.811', 'grad_norm': '0.4865', 'learning_rate': '0.0006', 'epoch': '10.28'}
+
 86%|████████▌ | 740/864 [13:21<01:51,  1.11it/s]
 86%|████████▌ | 741/864 [13:22<01:50,  1.11it/s]
 86%|████████▌ | 742/864 [13:23<02:04,  1.02s/it]
 86%|████████▌ | 743/864 [13:24<01:58,  1.02it/s]
 86%|████████▌ | 744/864 [13:25<01:54,  1.05it/s]
 86%|████████▌ | 745/864 [13:26<01:51,  1.07it/s]
 86%|████████▋ | 746/864 [13:27<01:49,  1.08it/s]
 86%|████████▋ | 747/864 [13:28<01:47,  1.09it/s]
 87%|████████▋ | 748/864 [13:29<01:45,  1.10it/s]
 87%|████████▋ | 749/864 [13:30<01:44,  1.10it/s]
 87%|████████▋ | 750/864 [13:31<01:43,  1.11it/s]
                                                 
{'loss': '3.795', 'grad_norm': '0.2312', 'learning_rate': '0.0006', 'epoch': '10.42'}
+
 87%|████████▋ | 750/864 [13:31<01:43,  1.11it/s]
 87%|████████▋ | 751/864 [13:32<01:41,  1.11it/s]
 87%|████████▋ | 752/864 [13:32<01:40,  1.11it/s]
 87%|████████▋ | 753/864 [13:33<01:39,  1.11it/s]
 87%|████████▋ | 754/864 [13:34<01:38,  1.11it/s]
 87%|████████▋ | 755/864 [13:35<01:37,  1.11it/s]
 88%|████████▊ | 756/864 [13:36<01:37,  1.11it/s]
 88%|████████▊ | 757/864 [13:37<01:36,  1.11it/s]
 88%|████████▊ | 758/864 [13:38<01:35,  1.11it/s]
 88%|████████▊ | 759/864 [13:39<01:34,  1.11it/s]
 88%|████████▊ | 760/864 [13:40<01:33,  1.11it/s]
                                                 
{'loss': '3.791', 'grad_norm': '0.2154', 'learning_rate': '0.0006', 'epoch': '10.56'}
+
 88%|████████▊ | 760/864 [13:40<01:33,  1.11it/s]
 88%|████████▊ | 761/864 [13:41<01:32,  1.11it/s]
 88%|████████▊ | 762/864 [13:41<01:31,  1.11it/s]
 88%|████████▊ | 763/864 [13:42<01:30,  1.11it/s]
 88%|████████▊ | 764/864 [13:43<01:29,  1.11it/s]
 89%|████████▊ | 765/864 [13:44<01:28,  1.11it/s]
 89%|████████▊ | 766/864 [13:45<01:27,  1.11it/s]
 89%|████████▉ | 767/864 [13:46<01:27,  1.11it/s]
 89%|████████▉ | 768/864 [13:47<01:26,  1.11it/s]
 89%|████████▉ | 769/864 [13:48<01:25,  1.11it/s]
 89%|████████▉ | 770/864 [13:49<01:24,  1.11it/s]
                                                 
{'loss': '3.771', 'grad_norm': '0.1859', 'learning_rate': '0.0006', 'epoch': '10.7'}
+
 89%|████████▉ | 770/864 [13:49<01:24,  1.11it/s]
 89%|████████▉ | 771/864 [13:50<01:23,  1.11it/s]
 89%|████████▉ | 772/864 [13:50<01:22,  1.11it/s]
 89%|████████▉ | 773/864 [13:51<01:21,  1.11it/s]
 90%|████████▉ | 774/864 [13:52<01:20,  1.11it/s]
 90%|████████▉ | 775/864 [13:53<01:19,  1.11it/s]
 90%|████████▉ | 776/864 [13:54<01:19,  1.11it/s]
 90%|████████▉ | 777/864 [13:55<01:18,  1.11it/s]
 90%|█████████ | 778/864 [13:56<01:17,  1.11it/s]
 90%|█████████ | 779/864 [13:57<01:16,  1.11it/s]
 90%|█████████ | 780/864 [13:58<01:15,  1.11it/s]
                                                 
{'loss': '3.766', 'grad_norm': '0.1816', 'learning_rate': '0.0006', 'epoch': '10.84'}
+
 90%|█████████ | 780/864 [13:58<01:15,  1.11it/s]
 90%|█████████ | 781/864 [13:59<01:14,  1.11it/s]
 91%|█████████ | 782/864 [13:59<01:13,  1.11it/s]
 91%|█████████ | 783/864 [14:00<01:12,  1.11it/s]
 91%|█████████ | 784/864 [14:01<01:11,  1.11it/s]
 91%|█████████ | 785/864 [14:02<01:10,  1.11it/s]
 91%|█████████ | 786/864 [14:03<01:10,  1.11it/s]
 91%|█████████ | 787/864 [14:04<01:09,  1.11it/s]
 91%|█████████ | 788/864 [14:05<01:08,  1.11it/s]
 91%|█████████▏| 789/864 [14:06<01:07,  1.11it/s]
 91%|█████████▏| 790/864 [14:07<01:06,  1.11it/s]
                                                 
{'loss': '3.752', 'grad_norm': '0.2209', 'learning_rate': '0.0006', 'epoch': '10.98'}
+
 91%|█████████▏| 790/864 [14:07<01:06,  1.11it/s]
 92%|█████████▏| 791/864 [14:07<01:05,  1.11it/s]
 92%|█████████▏| 792/864 [14:08<00:57,  1.25it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:02, 18.43it/s]
+
 10%|█         | 4/39 [00:00<00:02, 12.20it/s]
+
 15%|█▌        | 6/39 [00:00<00:03, 10.75it/s]
+
 21%|██        | 8/39 [00:00<00:02, 10.39it/s]
+
 26%|██▌       | 10/39 [00:00<00:02, 10.02it/s]
+
 31%|███       | 12/39 [00:01<00:02,  9.98it/s]
+
 36%|███▌      | 14/39 [00:01<00:02,  9.79it/s]
+
 38%|███▊      | 15/39 [00:01<00:02,  9.76it/s]
+
 44%|████▎     | 17/39 [00:01<00:02,  9.81it/s]
+
 46%|████▌     | 18/39 [00:01<00:02,  9.68it/s]
+
 49%|████▊     | 19/39 [00:01<00:02,  9.67it/s]
+
 54%|█████▍    | 21/39 [00:02<00:01,  9.76it/s]
+
 56%|█████▋    | 22/39 [00:02<00:01,  9.63it/s]
+
 62%|██████▏   | 24/39 [00:02<00:01,  9.74it/s]
+
 64%|██████▍   | 25/39 [00:02<00:01,  9.72it/s]
+
 67%|██████▋   | 26/39 [00:02<00:01,  9.70it/s]
+
 69%|██████▉   | 27/39 [00:02<00:01,  9.69it/s]
+
 72%|███████▏  | 28/39 [00:02<00:01,  9.69it/s]
+
 74%|███████▍  | 29/39 [00:02<00:01,  9.68it/s]
+
 77%|███████▋  | 30/39 [00:03<00:00,  9.67it/s]
+
 79%|███████▉  | 31/39 [00:03<00:00,  9.66it/s]
+
 82%|████████▏ | 32/39 [00:03<00:00,  9.67it/s]
+
 85%|████████▍ | 33/39 [00:03<00:00,  9.67it/s]
+
 87%|████████▋ | 34/39 [00:03<00:00,  9.66it/s]
+
 90%|████████▉ | 35/39 [00:03<00:00,  9.66it/s]
+
 92%|█████████▏| 36/39 [00:03<00:00,  9.66it/s]
+
 95%|█████████▍| 37/39 [00:03<00:00,  9.66it/s]
+
 97%|█████████▋| 38/39 [00:03<00:00,  9.66it/s]
+
100%|██████████| 39/39 [00:03<00:00,  9.65it/s]
                                                 
+
                                               
{'eval_loss': '4.074', 'eval_runtime': '4.106', 'eval_samples_per_second': '594.4', 'eval_steps_per_second': '9.497', 'epoch': '11'}
+
 92%|█████████▏| 792/864 [14:12<00:57,  1.25it/s]
+
100%|██████████| 39/39 [00:04<00:00,  9.65it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-792
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-792/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-792/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:01<00:00,  1.04s/it]
Writing model shards: 100%|██████████| 1/1 [00:01<00:00,  1.05s/it]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-792/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-792/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-792/tokenizer_config.json
+[eval] epoch 11.0 step 792 loss 4.0743 (best 4.0743 @ ep 11.0, 0/10000 since best)
+
+
                                               
 92%|█████████▏| 793/864 [14:17<03:43,  3.15s/it]
 92%|█████████▏| 794/864 [14:18<02:52,  2.47s/it]
 92%|█████████▏| 795/864 [14:18<02:17,  2.00s/it]
 92%|█████████▏| 796/864 [14:19<01:53,  1.67s/it]
 92%|█████████▏| 797/864 [14:20<01:36,  1.44s/it]
 92%|█████████▏| 798/864 [14:21<01:24,  1.28s/it]
 92%|█████████▏| 799/864 [14:22<01:15,  1.16s/it]
 93%|█████████▎| 800/864 [14:23<01:09,  1.08s/it]
                                                 
{'loss': '3.675', 'grad_norm': '0.4186', 'learning_rate': '0.0006', 'epoch': '11.11'}
+
 93%|█████████▎| 800/864 [14:23<01:09,  1.08s/it]
 93%|█████████▎| 801/864 [14:24<01:04,  1.03s/it]
 93%|█████████▎| 802/864 [14:25<01:01,  1.01it/s]
 93%|█████████▎| 803/864 [14:26<00:58,  1.04it/s]
 93%|█████████▎| 804/864 [14:27<00:56,  1.06it/s]
 93%|█████████▎| 805/864 [14:27<00:54,  1.07it/s]
 93%|█████████▎| 806/864 [14:28<00:53,  1.09it/s]
 93%|█████████▎| 807/864 [14:29<00:52,  1.09it/s]
 94%|█████████▎| 808/864 [14:30<00:50,  1.10it/s]
 94%|█████████▎| 809/864 [14:31<00:49,  1.10it/s]
 94%|█████████▍| 810/864 [14:32<00:48,  1.11it/s]
                                                 
{'loss': '3.669', 'grad_norm': '0.5337', 'learning_rate': '0.0006', 'epoch': '11.25'}
+
 94%|█████████▍| 810/864 [14:32<00:48,  1.11it/s]
 94%|█████████▍| 811/864 [14:33<00:47,  1.11it/s]
 94%|█████████▍| 812/864 [14:34<00:46,  1.11it/s]
 94%|█████████▍| 813/864 [14:35<00:45,  1.11it/s]
 94%|█████████▍| 814/864 [14:36<00:45,  1.11it/s]
 94%|█████████▍| 815/864 [14:36<00:44,  1.11it/s]
 94%|█████████▍| 816/864 [14:37<00:43,  1.11it/s]
 95%|█████████▍| 817/864 [14:38<00:42,  1.11it/s]
 95%|█████████▍| 818/864 [14:39<00:41,  1.11it/s]
 95%|█████████▍| 819/864 [14:40<00:40,  1.11it/s]
 95%|█████████▍| 820/864 [14:41<00:39,  1.11it/s]
                                                 
{'loss': '3.686', 'grad_norm': '0.3253', 'learning_rate': '0.0006', 'epoch': '11.39'}
+
 95%|█████████▍| 820/864 [14:41<00:39,  1.11it/s]
 95%|█████████▌| 821/864 [14:42<00:38,  1.11it/s]
 95%|█████████▌| 822/864 [14:43<00:37,  1.11it/s]
 95%|█████████▌| 823/864 [14:44<00:36,  1.11it/s]
 95%|█████████▌| 824/864 [14:45<00:35,  1.11it/s]
 95%|█████████▌| 825/864 [14:45<00:35,  1.11it/s]
 96%|█████████▌| 826/864 [14:46<00:34,  1.11it/s]
 96%|█████████▌| 827/864 [14:47<00:33,  1.11it/s]
 96%|█████████▌| 828/864 [14:48<00:32,  1.11it/s]
 96%|█████████▌| 829/864 [14:49<00:31,  1.11it/s]
 96%|█████████▌| 830/864 [14:50<00:30,  1.11it/s]
                                                 
{'loss': '3.676', 'grad_norm': '0.228', 'learning_rate': '0.0006', 'epoch': '11.53'}
+
 96%|█████████▌| 830/864 [14:50<00:30,  1.11it/s]
 96%|█████████▌| 831/864 [14:51<00:29,  1.11it/s]
 96%|█████████▋| 832/864 [14:52<00:28,  1.11it/s]
 96%|█████████▋| 833/864 [14:53<00:27,  1.11it/s]
 97%|█████████▋| 834/864 [14:54<00:26,  1.11it/s]
 97%|█████████▋| 835/864 [14:54<00:26,  1.11it/s]
 97%|█████████▋| 836/864 [14:55<00:25,  1.11it/s]
 97%|█████████▋| 837/864 [14:56<00:24,  1.11it/s]
 97%|█████████▋| 838/864 [14:57<00:23,  1.11it/s]
 97%|█████████▋| 839/864 [14:58<00:22,  1.11it/s]
 97%|█████████▋| 840/864 [14:59<00:21,  1.11it/s]
                                                 
{'loss': '3.661', 'grad_norm': '0.1938', 'learning_rate': '0.0006', 'epoch': '11.67'}
+
 97%|█████████▋| 840/864 [14:59<00:21,  1.11it/s]
 97%|█████████▋| 841/864 [15:00<00:20,  1.11it/s]
 97%|█████████▋| 842/864 [15:01<00:19,  1.11it/s]
 98%|█████████▊| 843/864 [15:02<00:18,  1.11it/s]
 98%|█████████▊| 844/864 [15:03<00:17,  1.11it/s]
 98%|█████████▊| 845/864 [15:03<00:17,  1.11it/s]
 98%|█████████▊| 846/864 [15:04<00:16,  1.11it/s]
 98%|█████████▊| 847/864 [15:05<00:15,  1.11it/s]
 98%|█████████▊| 848/864 [15:06<00:14,  1.11it/s]
 98%|█████████▊| 849/864 [15:07<00:13,  1.11it/s]
 98%|█████████▊| 850/864 [15:08<00:12,  1.11it/s]
                                                 
{'loss': '3.655', 'grad_norm': '0.1969', 'learning_rate': '0.0006', 'epoch': '11.81'}
+
 98%|█████████▊| 850/864 [15:08<00:12,  1.11it/s]
 98%|█████████▊| 851/864 [15:09<00:11,  1.11it/s]
 99%|█████████▊| 852/864 [15:10<00:10,  1.11it/s]
 99%|█████████▊| 853/864 [15:11<00:09,  1.11it/s]
 99%|█████████▉| 854/864 [15:11<00:08,  1.11it/s]
 99%|█████████▉| 855/864 [15:12<00:08,  1.11it/s]
 99%|█████████▉| 856/864 [15:13<00:07,  1.11it/s]
 99%|█████████▉| 857/864 [15:14<00:06,  1.11it/s]
 99%|█████████▉| 858/864 [15:15<00:05,  1.11it/s]
 99%|█████████▉| 859/864 [15:16<00:04,  1.11it/s]
100%|█████████▉| 860/864 [15:17<00:03,  1.11it/s]
                                                 
{'loss': '3.657', 'grad_norm': '0.1914', 'learning_rate': '0.0006', 'epoch': '11.95'}
+
100%|█████████▉| 860/864 [15:17<00:03,  1.11it/s]
100%|█████████▉| 861/864 [15:18<00:02,  1.11it/s]
100%|█████████▉| 862/864 [15:19<00:01,  1.11it/s]
100%|█████████▉| 863/864 [15:20<00:00,  1.11it/s]
100%|██████████| 864/864 [15:20<00:00,  1.25it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:01, 19.30it/s]
+
 10%|█         | 4/39 [00:00<00:02, 12.15it/s]
+
 15%|█▌        | 6/39 [00:00<00:03, 10.85it/s]
+
 21%|██        | 8/39 [00:00<00:02, 10.35it/s]
+
 26%|██▌       | 10/39 [00:00<00:02, 10.07it/s]
+
 31%|███       | 12/39 [00:01<00:02,  9.93it/s]
+
 36%|███▌      | 14/39 [00:01<00:02,  9.83it/s]
+
 41%|████      | 16/39 [00:01<00:02,  9.78it/s]
+
 44%|████▎     | 17/39 [00:01<00:02,  9.75it/s]
+
 46%|████▌     | 18/39 [00:01<00:02,  9.72it/s]
+
 49%|████▊     | 19/39 [00:01<00:02,  9.70it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01,  9.69it/s]
+
 54%|█████▍    | 21/39 [00:02<00:01,  9.68it/s]
+
 56%|█████▋    | 22/39 [00:02<00:01,  9.67it/s]
+
 59%|█████▉    | 23/39 [00:02<00:01,  9.68it/s]
+
 62%|██████▏   | 24/39 [00:02<00:01,  9.68it/s]
+
 64%|██████▍   | 25/39 [00:02<00:01,  9.67it/s]
+
 67%|██████▋   | 26/39 [00:02<00:01,  9.66it/s]
+
 69%|██████▉   | 27/39 [00:02<00:01,  9.65it/s]
+
 72%|███████▏  | 28/39 [00:02<00:01,  9.65it/s]
+
 74%|███████▍  | 29/39 [00:02<00:01,  9.66it/s]
+
 77%|███████▋  | 30/39 [00:03<00:00,  9.66it/s]
+
 79%|███████▉  | 31/39 [00:03<00:00,  9.66it/s]
+
 82%|████████▏ | 32/39 [00:03<00:00,  9.66it/s]
+
 85%|████████▍ | 33/39 [00:03<00:00,  9.66it/s]
+
 87%|████████▋ | 34/39 [00:03<00:00,  9.66it/s]
+
 90%|████████▉ | 35/39 [00:03<00:00,  9.66it/s]
+
 92%|█████████▏| 36/39 [00:03<00:00,  9.67it/s]
+
 95%|█████████▍| 37/39 [00:03<00:00,  9.66it/s]
+
 97%|█████████▋| 38/39 [00:03<00:00,  9.66it/s]
+
100%|██████████| 39/39 [00:03<00:00,  9.67it/s]
                                                 
+
                                               
{'eval_loss': '4.032', 'eval_runtime': '4.113', 'eval_samples_per_second': '593.4', 'eval_steps_per_second': '9.481', 'epoch': '12'}
+
100%|██████████| 864/864 [15:24<00:00,  1.25it/s]
+
100%|██████████| 39/39 [00:04<00:00,  9.67it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-864
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-864/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-864/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.13it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.13it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-864/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-864/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0006/checkpoint-864/tokenizer_config.json
+[eval] epoch 12.0 step 864 loss 4.0316 (best 4.0316 @ ep 12.0, 0/10000 since best)
+
+
                                               [transformers] 
+
+Training completed. Do not forget to share your model on huggingface.co/models =)
+
+
+
                                                 
{'train_runtime': '929.9', 'train_samples_per_second': '472.5', 'train_steps_per_second': '0.929', 'train_loss': '5.085', 'epoch': '12'}
+
100%|██████████| 864/864 [15:28<00:00,  1.25it/s]
100%|██████████| 864/864 [15:28<00:00,  1.07s/it]
+wandb: 
+wandb: 🚀 View run t0_S3_lr0.0006 at: https://wandb.ai/slowrun-wankers/data-aug-pretraining/runs/0fm7mass
+wandb: Find logs at: wandb/run-20260817_195000-0fm7mass/logs
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+[rank0]:[W817 20:06:13.788700542 ProcessGroupNCCL.cpp:1575] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator())
diff --git a/runs/t0/t0_S3_lr0.0012/checkpoint-864/chat_template.jinja b/runs/t0/t0_S3_lr0.0012/checkpoint-864/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..699ff8df401fe4788525e9c1f9b86a99eadd6230
--- /dev/null
+++ b/runs/t0/t0_S3_lr0.0012/checkpoint-864/chat_template.jinja
@@ -0,0 +1,85 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0].role == 'system' %}
+        {{- messages[0].content + '\n\n' }}
+    {%- endif %}
+    {{- "# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0].content + '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
+{%- for message in messages[::-1] %}
+    {%- set index = (messages|length - 1) - loop.index0 %}
+    {%- if ns.multi_step_tool and message.role == "user" and not(message.content.startswith('<tool_response>') and message.content.endswith('</tool_response>')) %}
+        {%- set ns.multi_step_tool = false %}
+        {%- set ns.last_query_index = index %}
+    {%- endif %}
+{%- endfor %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {%- set content = message.content %}
+        {%- set reasoning_content = '' %}
+        {%- if message.reasoning_content is defined and message.reasoning_content is not none %}
+            {%- set reasoning_content = message.reasoning_content %}
+        {%- else %}
+            {%- if '</think>' in message.content %}
+                {%- set content = message.content.split('</think>')[-1].lstrip('\n') %}
+                {%- set reasoning_content = message.content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
+            {%- endif %}
+        {%- endif %}
+        {%- if loop.index0 > ns.last_query_index %}
+            {%- if loop.last or (not loop.last and reasoning_content) %}
+                {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content.strip('\n') + '\n</think>\n\n' + content.lstrip('\n') }}
+            {%- else %}
+                {{- '<|im_start|>' + message.role + '\n' + content }}
+            {%- endif %}
+        {%- else %}
+            {{- '<|im_start|>' + message.role + '\n' + content }}
+        {%- endif %}
+        {%- if message.tool_calls %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if (loop.first and content) or (not loop.first) %}
+                    {{- '\n' }}
+                {%- endif %}
+                {%- if tool_call.function %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {{- '<tool_call>\n{"name": "' }}
+                {{- tool_call.name }}
+                {{- '", "arguments": ' }}
+                {%- if tool_call.arguments is string %}
+                    {{- tool_call.arguments }}
+                {%- else %}
+                    {{- tool_call.arguments | tojson }}
+                {%- endif %}
+                {{- '}\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.first or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+    {%- if enable_thinking is defined and enable_thinking is false %}
+        {{- '<think>\n\n</think>\n\n' }}
+    {%- endif %}
+{%- endif %}
\ No newline at end of file
diff --git a/runs/t0/t0_S3_lr0.0012/checkpoint-864/config.json b/runs/t0/t0_S3_lr0.0012/checkpoint-864/config.json
new file mode 100644
index 0000000000000000000000000000000000000000..edf37b33d02b391ecdf2ca887c98eec60d9c7686
--- /dev/null
+++ b/runs/t0/t0_S3_lr0.0012/checkpoint-864/config.json
@@ -0,0 +1,32 @@
+{
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": null,
+  "dtype": "float32",
+  "eos_token_id": 151645,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 768,
+  "initializer_range": 0.02,
+  "intermediate_size": 2048,
+  "max_position_embeddings": 2048,
+  "mlp_bias": false,
+  "model_type": "llama",
+  "num_attention_heads": 6,
+  "num_hidden_layers": 20,
+  "num_key_value_heads": 6,
+  "pad_token_id": 151645,
+  "pretraining_tp": 1,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 10000.0,
+    "rope_type": "default"
+  },
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": false,
+  "vocab_size": 151671
+}
diff --git a/runs/t0/t0_S3_lr0.0012/checkpoint-864/generation_config.json b/runs/t0/t0_S3_lr0.0012/checkpoint-864/generation_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..8c4242f32d40df256857f87264f646130f684e33
--- /dev/null
+++ b/runs/t0/t0_S3_lr0.0012/checkpoint-864/generation_config.json
@@ -0,0 +1,11 @@
+{
+  "_from_model_config": true,
+  "eos_token_id": [
+    151645
+  ],
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "transformers_version": "5.15.0",
+  "use_cache": true
+}
diff --git a/runs/t0/t0_S3_lr0.0012/checkpoint-864/tokenizer_config.json b/runs/t0/t0_S3_lr0.0012/checkpoint-864/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..839a3743177a3ed77e1ffbae67520e8500ca0532
--- /dev/null
+++ b/runs/t0/t0_S3_lr0.0012/checkpoint-864/tokenizer_config.json
@@ -0,0 +1,19 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|l2r_pred|>",
+    "<|r2l_pred|>"
+  ],
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 131072,
+  "pad_token": "<|im_end|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}
diff --git a/runs/t0/t0_S3_lr0.0012/checkpoint-864/trainer_state.json b/runs/t0/t0_S3_lr0.0012/checkpoint-864/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..21db5fe63956eea876c2e38570b47235f7475e26
--- /dev/null
+++ b/runs/t0/t0_S3_lr0.0012/checkpoint-864/trainer_state.json
@@ -0,0 +1,739 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 12.0,
+  "eval_steps": 72,
+  "global_step": 864,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.013961605584642234,
+      "grad_norm": 2.625096082687378,
+      "learning_rate": 0.0,
+      "loss": 12.076139450073242,
+      "step": 1
+    },
+    {
+      "epoch": 0.13961605584642234,
+      "grad_norm": 1.9982500076293945,
+      "learning_rate": 0.00010799999999999998,
+      "loss": 11.33876207139757,
+      "step": 10
+    },
+    {
+      "epoch": 0.2792321116928447,
+      "grad_norm": 1.7737500667572021,
+      "learning_rate": 0.00022799999999999999,
+      "loss": 9.99845199584961,
+      "step": 20
+    },
+    {
+      "epoch": 0.418848167539267,
+      "grad_norm": 0.6299463510513306,
+      "learning_rate": 0.00034799999999999995,
+      "loss": 8.386837768554688,
+      "step": 30
+    },
+    {
+      "epoch": 0.5584642233856894,
+      "grad_norm": 0.4135228395462036,
+      "learning_rate": 0.000468,
+      "loss": 7.641131591796875,
+      "step": 40
+    },
+    {
+      "epoch": 0.6980802792321117,
+      "grad_norm": 0.6167853474617004,
+      "learning_rate": 0.000588,
+      "loss": 7.482666015625,
+      "step": 50
+    },
+    {
+      "epoch": 0.837696335078534,
+      "grad_norm": 0.7763964533805847,
+      "learning_rate": 0.0007079999999999999,
+      "loss": 7.347389984130859,
+      "step": 60
+    },
+    {
+      "epoch": 0.9773123909249564,
+      "grad_norm": 0.4353961646556854,
+      "learning_rate": 0.0008279999999999999,
+      "loss": 7.112811279296875,
+      "step": 70
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 6.975900173187256,
+      "eval_runtime": 27.9985,
+      "eval_samples_per_second": 87.183,
+      "eval_steps_per_second": 1.393,
+      "step": 72
+    },
+    {
+      "epoch": 1.111692844677138,
+      "grad_norm": 0.6767817735671997,
+      "learning_rate": 0.000948,
+      "loss": 6.936407470703125,
+      "step": 80
+    },
+    {
+      "epoch": 1.2513089005235603,
+      "grad_norm": 0.5563651919364929,
+      "learning_rate": 0.001068,
+      "loss": 6.7925537109375,
+      "step": 90
+    },
+    {
+      "epoch": 1.3909249563699826,
+      "grad_norm": 0.9051397442817688,
+      "learning_rate": 0.0011879999999999998,
+      "loss": 6.682518768310547,
+      "step": 100
+    },
+    {
+      "epoch": 1.530541012216405,
+      "grad_norm": 0.3846557140350342,
+      "learning_rate": 0.0012,
+      "loss": 6.563172912597656,
+      "step": 110
+    },
+    {
+      "epoch": 1.6701570680628273,
+      "grad_norm": 0.413596510887146,
+      "learning_rate": 0.0012,
+      "loss": 6.416682434082031,
+      "step": 120
+    },
+    {
+      "epoch": 1.8097731239092496,
+      "grad_norm": 0.6242079138755798,
+      "learning_rate": 0.0012,
+      "loss": 6.320872497558594,
+      "step": 130
+    },
+    {
+      "epoch": 1.949389179755672,
+      "grad_norm": 0.4003734290599823,
+      "learning_rate": 0.0012,
+      "loss": 6.239190292358399,
+      "step": 140
+    },
+    {
+      "epoch": 2.0,
+      "eval_loss": 6.174596786499023,
+      "eval_runtime": 4.094,
+      "eval_samples_per_second": 596.236,
+      "eval_steps_per_second": 9.526,
+      "step": 144
+    },
+    {
+      "epoch": 2.0837696335078535,
+      "grad_norm": 0.3320383131504059,
+      "learning_rate": 0.0012,
+      "loss": 6.135129928588867,
+      "step": 150
+    },
+    {
+      "epoch": 2.223385689354276,
+      "grad_norm": 0.4379850924015045,
+      "learning_rate": 0.0012,
+      "loss": 6.063224029541016,
+      "step": 160
+    },
+    {
+      "epoch": 2.363001745200698,
+      "grad_norm": 0.4536810517311096,
+      "learning_rate": 0.0012,
+      "loss": 5.9813385009765625,
+      "step": 170
+    },
+    {
+      "epoch": 2.5026178010471205,
+      "grad_norm": 0.8055464029312134,
+      "learning_rate": 0.0012,
+      "loss": 5.904870986938477,
+      "step": 180
+    },
+    {
+      "epoch": 2.642233856893543,
+      "grad_norm": 0.41387513279914856,
+      "learning_rate": 0.0012,
+      "loss": 5.888151550292969,
+      "step": 190
+    },
+    {
+      "epoch": 2.781849912739965,
+      "grad_norm": 0.30246642231941223,
+      "learning_rate": 0.0012,
+      "loss": 5.781609725952149,
+      "step": 200
+    },
+    {
+      "epoch": 2.9214659685863875,
+      "grad_norm": 0.2841012477874756,
+      "learning_rate": 0.0012,
+      "loss": 5.709811401367188,
+      "step": 210
+    },
+    {
+      "epoch": 3.0,
+      "eval_loss": 5.691598892211914,
+      "eval_runtime": 4.097,
+      "eval_samples_per_second": 595.796,
+      "eval_steps_per_second": 9.519,
+      "step": 216
+    },
+    {
+      "epoch": 3.055846422338569,
+      "grad_norm": 0.38285037875175476,
+      "learning_rate": 0.0012,
+      "loss": 5.628893661499023,
+      "step": 220
+    },
+    {
+      "epoch": 3.195462478184991,
+      "grad_norm": 0.46441373229026794,
+      "learning_rate": 0.0012,
+      "loss": 5.541646575927734,
+      "step": 230
+    },
+    {
+      "epoch": 3.3350785340314135,
+      "grad_norm": 0.5794033408164978,
+      "learning_rate": 0.0012,
+      "loss": 5.48505744934082,
+      "step": 240
+    },
+    {
+      "epoch": 3.474694589877836,
+      "grad_norm": 0.560783326625824,
+      "learning_rate": 0.0012,
+      "loss": 5.4691825866699215,
+      "step": 250
+    },
+    {
+      "epoch": 3.614310645724258,
+      "grad_norm": 0.2417001873254776,
+      "learning_rate": 0.0012,
+      "loss": 5.374152755737304,
+      "step": 260
+    },
+    {
+      "epoch": 3.7539267015706805,
+      "grad_norm": 0.7059624791145325,
+      "learning_rate": 0.0012,
+      "loss": 5.307765197753906,
+      "step": 270
+    },
+    {
+      "epoch": 3.893542757417103,
+      "grad_norm": 0.26051831245422363,
+      "learning_rate": 0.0012,
+      "loss": 5.307967758178711,
+      "step": 280
+    },
+    {
+      "epoch": 4.0,
+      "eval_loss": 5.226171016693115,
+      "eval_runtime": 4.087,
+      "eval_samples_per_second": 597.255,
+      "eval_steps_per_second": 9.542,
+      "step": 288
+    },
+    {
+      "epoch": 4.027923211169284,
+      "grad_norm": 0.20769786834716797,
+      "learning_rate": 0.0012,
+      "loss": 5.167432022094727,
+      "step": 290
+    },
+    {
+      "epoch": 4.167539267015707,
+      "grad_norm": 0.5713754892349243,
+      "learning_rate": 0.0012,
+      "loss": 5.101666259765625,
+      "step": 300
+    },
+    {
+      "epoch": 4.307155322862129,
+      "grad_norm": 0.23590517044067383,
+      "learning_rate": 0.0012,
+      "loss": 5.0780693054199215,
+      "step": 310
+    },
+    {
+      "epoch": 4.446771378708552,
+      "grad_norm": 0.17983147501945496,
+      "learning_rate": 0.0012,
+      "loss": 4.969886016845703,
+      "step": 320
+    },
+    {
+      "epoch": 4.5863874345549736,
+      "grad_norm": 0.6250120997428894,
+      "learning_rate": 0.0012,
+      "loss": 4.961678314208984,
+      "step": 330
+    },
+    {
+      "epoch": 4.726003490401396,
+      "grad_norm": 0.5018834471702576,
+      "learning_rate": 0.0012,
+      "loss": 4.9425300598144535,
+      "step": 340
+    },
+    {
+      "epoch": 4.865619546247818,
+      "grad_norm": 0.1806376427412033,
+      "learning_rate": 0.0012,
+      "loss": 4.830913543701172,
+      "step": 350
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 0.42157575488090515,
+      "learning_rate": 0.0012,
+      "loss": 4.768266296386718,
+      "step": 360
+    },
+    {
+      "epoch": 5.0,
+      "eval_loss": 4.85431432723999,
+      "eval_runtime": 4.0892,
+      "eval_samples_per_second": 596.936,
+      "eval_steps_per_second": 9.537,
+      "step": 360
+    },
+    {
+      "epoch": 5.139616055846423,
+      "grad_norm": 0.4851687252521515,
+      "learning_rate": 0.0012,
+      "loss": 4.7232307434082035,
+      "step": 370
+    },
+    {
+      "epoch": 5.279232111692845,
+      "grad_norm": 0.20989617705345154,
+      "learning_rate": 0.0012,
+      "loss": 4.6745655059814455,
+      "step": 380
+    },
+    {
+      "epoch": 5.418848167539267,
+      "grad_norm": 0.18929427862167358,
+      "learning_rate": 0.0012,
+      "loss": 4.5781105041503904,
+      "step": 390
+    },
+    {
+      "epoch": 5.558464223385689,
+      "grad_norm": 0.5501760244369507,
+      "learning_rate": 0.0012,
+      "loss": 4.528529357910156,
+      "step": 400
+    },
+    {
+      "epoch": 5.698080279232112,
+      "grad_norm": 0.33675816655158997,
+      "learning_rate": 0.0012,
+      "loss": 4.550762176513672,
+      "step": 410
+    },
+    {
+      "epoch": 5.837696335078534,
+      "grad_norm": 0.1552075892686844,
+      "learning_rate": 0.0012,
+      "loss": 4.4934234619140625,
+      "step": 420
+    },
+    {
+      "epoch": 5.977312390924956,
+      "grad_norm": 0.3856748640537262,
+      "learning_rate": 0.0012,
+      "loss": 4.433892822265625,
+      "step": 430
+    },
+    {
+      "epoch": 6.0,
+      "eval_loss": 4.608406066894531,
+      "eval_runtime": 4.0882,
+      "eval_samples_per_second": 597.088,
+      "eval_steps_per_second": 9.54,
+      "step": 432
+    },
+    {
+      "epoch": 6.111692844677138,
+      "grad_norm": 0.4819994270801544,
+      "learning_rate": 0.0012,
+      "loss": 4.424343872070312,
+      "step": 440
+    },
+    {
+      "epoch": 6.2513089005235605,
+      "grad_norm": 0.20761343836784363,
+      "learning_rate": 0.0012,
+      "loss": 4.38978271484375,
+      "step": 450
+    },
+    {
+      "epoch": 6.390924956369982,
+      "grad_norm": 0.15534761548042297,
+      "learning_rate": 0.0012,
+      "loss": 4.303800201416015,
+      "step": 460
+    },
+    {
+      "epoch": 6.530541012216405,
+      "grad_norm": 0.16674354672431946,
+      "learning_rate": 0.0012,
+      "loss": 4.259825134277344,
+      "step": 470
+    },
+    {
+      "epoch": 6.670157068062827,
+      "grad_norm": 0.40141749382019043,
+      "learning_rate": 0.0012,
+      "loss": 4.243598937988281,
+      "step": 480
+    },
+    {
+      "epoch": 6.80977312390925,
+      "grad_norm": 0.16742125153541565,
+      "learning_rate": 0.0012,
+      "loss": 4.259889221191406,
+      "step": 490
+    },
+    {
+      "epoch": 6.949389179755672,
+      "grad_norm": 0.19535250961780548,
+      "learning_rate": 0.0012,
+      "loss": 4.196575927734375,
+      "step": 500
+    },
+    {
+      "epoch": 7.0,
+      "eval_loss": 4.354514122009277,
+      "eval_runtime": 4.0881,
+      "eval_samples_per_second": 597.094,
+      "eval_steps_per_second": 9.54,
+      "step": 504
+    },
+    {
+      "epoch": 7.0837696335078535,
+      "grad_norm": 0.433318555355072,
+      "learning_rate": 0.0012,
+      "loss": 4.1356159210205075,
+      "step": 510
+    },
+    {
+      "epoch": 7.223385689354275,
+      "grad_norm": 0.4595310688018799,
+      "learning_rate": 0.0012,
+      "loss": 4.177924728393554,
+      "step": 520
+    },
+    {
+      "epoch": 7.363001745200698,
+      "grad_norm": 0.18393999338150024,
+      "learning_rate": 0.0012,
+      "loss": 4.169436645507813,
+      "step": 530
+    },
+    {
+      "epoch": 7.50261780104712,
+      "grad_norm": 0.12354253232479095,
+      "learning_rate": 0.0012,
+      "loss": 4.10106086730957,
+      "step": 540
+    },
+    {
+      "epoch": 7.642233856893543,
+      "grad_norm": 0.11428206413984299,
+      "learning_rate": 0.0012,
+      "loss": 4.0617938995361325,
+      "step": 550
+    },
+    {
+      "epoch": 7.781849912739965,
+      "grad_norm": 0.24703778326511383,
+      "learning_rate": 0.0012,
+      "loss": 4.040187072753906,
+      "step": 560
+    },
+    {
+      "epoch": 7.9214659685863875,
+      "grad_norm": 0.1985350400209427,
+      "learning_rate": 0.0012,
+      "loss": 4.034548950195313,
+      "step": 570
+    },
+    {
+      "epoch": 8.0,
+      "eval_loss": 4.230696201324463,
+      "eval_runtime": 4.0967,
+      "eval_samples_per_second": 595.847,
+      "eval_steps_per_second": 9.52,
+      "step": 576
+    },
+    {
+      "epoch": 8.055846422338568,
+      "grad_norm": 0.22547073662281036,
+      "learning_rate": 0.0012,
+      "loss": 3.9681182861328126,
+      "step": 580
+    },
+    {
+      "epoch": 8.195462478184991,
+      "grad_norm": 0.17060305178165436,
+      "learning_rate": 0.0012,
+      "loss": 3.8946533203125,
+      "step": 590
+    },
+    {
+      "epoch": 8.335078534031414,
+      "grad_norm": 0.15548421442508698,
+      "learning_rate": 0.0012,
+      "loss": 3.884337615966797,
+      "step": 600
+    },
+    {
+      "epoch": 8.474694589877837,
+      "grad_norm": 0.16989755630493164,
+      "learning_rate": 0.0012,
+      "loss": 3.867962646484375,
+      "step": 610
+    },
+    {
+      "epoch": 8.614310645724258,
+      "grad_norm": 0.3586825132369995,
+      "learning_rate": 0.0012,
+      "loss": 3.874079132080078,
+      "step": 620
+    },
+    {
+      "epoch": 8.75392670157068,
+      "grad_norm": 0.2342267632484436,
+      "learning_rate": 0.0012,
+      "loss": 3.8905319213867187,
+      "step": 630
+    },
+    {
+      "epoch": 8.893542757417103,
+      "grad_norm": 0.18740172684192657,
+      "learning_rate": 0.0012,
+      "loss": 3.8916366577148436,
+      "step": 640
+    },
+    {
+      "epoch": 9.0,
+      "eval_loss": 4.121691703796387,
+      "eval_runtime": 4.1052,
+      "eval_samples_per_second": 594.617,
+      "eval_steps_per_second": 9.5,
+      "step": 648
+    },
+    {
+      "epoch": 9.027923211169284,
+      "grad_norm": 0.17139847576618195,
+      "learning_rate": 0.0012,
+      "loss": 3.832661819458008,
+      "step": 650
+    },
+    {
+      "epoch": 9.167539267015707,
+      "grad_norm": 0.4771718382835388,
+      "learning_rate": 0.0012,
+      "loss": 3.7312690734863283,
+      "step": 660
+    },
+    {
+      "epoch": 9.30715532286213,
+      "grad_norm": 0.3208262026309967,
+      "learning_rate": 0.0012,
+      "loss": 3.7902557373046877,
+      "step": 670
+    },
+    {
+      "epoch": 9.44677137870855,
+      "grad_norm": 0.269268274307251,
+      "learning_rate": 0.0012,
+      "loss": 3.79544677734375,
+      "step": 680
+    },
+    {
+      "epoch": 9.586387434554974,
+      "grad_norm": 0.16161218285560608,
+      "learning_rate": 0.0012,
+      "loss": 3.7524093627929687,
+      "step": 690
+    },
+    {
+      "epoch": 9.726003490401396,
+      "grad_norm": 0.13752730190753937,
+      "learning_rate": 0.0012,
+      "loss": 3.7424781799316404,
+      "step": 700
+    },
+    {
+      "epoch": 9.86561954624782,
+      "grad_norm": 0.12593114376068115,
+      "learning_rate": 0.0012,
+      "loss": 3.7211395263671876,
+      "step": 710
+    },
+    {
+      "epoch": 10.0,
+      "grad_norm": 0.13294491171836853,
+      "learning_rate": 0.0012,
+      "loss": 3.708177185058594,
+      "step": 720
+    },
+    {
+      "epoch": 10.0,
+      "eval_loss": 4.056910037994385,
+      "eval_runtime": 4.0907,
+      "eval_samples_per_second": 596.717,
+      "eval_steps_per_second": 9.534,
+      "step": 720
+    },
+    {
+      "epoch": 10.139616055846423,
+      "grad_norm": 0.5313000082969666,
+      "learning_rate": 0.0012,
+      "loss": 3.5885597229003907,
+      "step": 730
+    },
+    {
+      "epoch": 10.279232111692846,
+      "grad_norm": 0.37508344650268555,
+      "learning_rate": 0.0012,
+      "loss": 3.6925918579101564,
+      "step": 740
+    },
+    {
+      "epoch": 10.418848167539267,
+      "grad_norm": 0.1845429688692093,
+      "learning_rate": 0.0012,
+      "loss": 3.656546783447266,
+      "step": 750
+    },
+    {
+      "epoch": 10.55846422338569,
+      "grad_norm": 0.13196681439876556,
+      "learning_rate": 0.0012,
+      "loss": 3.6393321990966796,
+      "step": 760
+    },
+    {
+      "epoch": 10.698080279232112,
+      "grad_norm": 0.12762899696826935,
+      "learning_rate": 0.0012,
+      "loss": 3.6187034606933595,
+      "step": 770
+    },
+    {
+      "epoch": 10.837696335078533,
+      "grad_norm": 0.12318549305200577,
+      "learning_rate": 0.0012,
+      "loss": 3.6179458618164064,
+      "step": 780
+    },
+    {
+      "epoch": 10.977312390924956,
+      "grad_norm": 0.12958157062530518,
+      "learning_rate": 0.0012,
+      "loss": 3.6080657958984377,
+      "step": 790
+    },
+    {
+      "epoch": 11.0,
+      "eval_loss": 4.0241265296936035,
+      "eval_runtime": 4.0952,
+      "eval_samples_per_second": 596.068,
+      "eval_steps_per_second": 9.523,
+      "step": 792
+    },
+    {
+      "epoch": 11.111692844677139,
+      "grad_norm": 0.28613024950027466,
+      "learning_rate": 0.0012,
+      "loss": 3.4843353271484374,
+      "step": 800
+    },
+    {
+      "epoch": 11.25130890052356,
+      "grad_norm": 0.3105561137199402,
+      "learning_rate": 0.0012,
+      "loss": 3.458019256591797,
+      "step": 810
+    },
+    {
+      "epoch": 11.390924956369982,
+      "grad_norm": 0.2634027302265167,
+      "learning_rate": 0.0012,
+      "loss": 3.488126754760742,
+      "step": 820
+    },
+    {
+      "epoch": 11.530541012216405,
+      "grad_norm": 0.14770400524139404,
+      "learning_rate": 0.0012,
+      "loss": 3.4990020751953126,
+      "step": 830
+    },
+    {
+      "epoch": 11.670157068062828,
+      "grad_norm": 0.13869911432266235,
+      "learning_rate": 0.0012,
+      "loss": 3.4969722747802736,
+      "step": 840
+    },
+    {
+      "epoch": 11.809773123909249,
+      "grad_norm": 0.12742026150226593,
+      "learning_rate": 0.0012,
+      "loss": 3.502721405029297,
+      "step": 850
+    },
+    {
+      "epoch": 11.949389179755672,
+      "grad_norm": 0.1485251933336258,
+      "learning_rate": 0.0012,
+      "loss": 3.5136146545410156,
+      "step": 860
+    },
+    {
+      "epoch": 12.0,
+      "eval_loss": 4.010306358337402,
+      "eval_runtime": 4.1105,
+      "eval_samples_per_second": 593.846,
+      "eval_steps_per_second": 9.488,
+      "step": 864
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 864,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 12,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": false,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 7.656447700673495e+17,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/runs/t0/t0_S3_lr0.0012/train.log b/runs/t0/t0_S3_lr0.0012/train.log
new file mode 100644
index 0000000000000000000000000000000000000000..8080dbbd32dbb0ed8a1264fa4cd8477d6170ae0d
--- /dev/null
+++ b/runs/t0/t0_S3_lr0.0012/train.log
@@ -0,0 +1,1600 @@
+W0817 20:06:17.499000 2237554 torch/distributed/run.py:851] 
+W0817 20:06:17.499000 2237554 torch/distributed/run.py:851] *****************************************
+W0817 20:06:17.499000 2237554 torch/distributed/run.py:851] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+W0817 20:06:17.499000 2237554 torch/distributed/run.py:851] *****************************************
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+total params: 258072576
+[dataset] loaded packed cache: 36614 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+total params: 258072576
+[dataset] loaded packed cache: 36614 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+total params: 258072576
+total params: 258072576
+[dataset] loaded packed cache: 36614 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[dataset] loaded packed cache: 36614 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+total params: 258072576
+[dataset] loaded packed cache: 36614 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+total params: 258072576
+[dataset] loaded packed cache: 36614 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+total params: 258072576
+[dataset] loaded packed cache: 36614 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+total params: 258072576
+[dataset] loaded packed cache: 36614 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+[transformers] ***** Running training *****
+[transformers]   Num examples = 36,614
+[transformers]   Num Epochs = 12
+[transformers]   Num update steps per epoch = 72
+[transformers]   Instantaneous batch size per device = 8
+[transformers]   Total train batch size (w. parallel, distributed & accumulation) = 512
+[transformers]   Gradient Accumulation steps = 8
+[transformers]   Total optimization steps = 864
+[transformers]   Number of trainable parameters = 258,072,576
+wandb: WARNING Changes to your `wandb` environment variables will be ignored because your `wandb` session has already started. For more information on how to modify your settings with `wandb.init()` arguments, please refer to https://wandb.me/wandb-init.
+wandb: setting up run hver985o
+wandb: Tracking run with wandb version 0.28.2
+wandb: Run data is saved locally in /home/ubuntu/workspace/data-augmentations-for-pretraining/wandb/run-20260817_200638-hver985o
+wandb: Run `wandb offline` to turn off syncing.
+wandb: Syncing run t0_S3_lr0.0012
+wandb: ⭐️ View project at https://wandb.ai/slowrun-wankers/data-aug-pretraining
+wandb: 🚀 View run at https://wandb.ai/slowrun-wankers/data-aug-pretraining/runs/hver985o
+
  0%|          | 0/864 [00:00<?, ?it/s]wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+[rank5]:W0817 20:07:01.415000 2237666 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+[rank6]:W0817 20:07:01.633000 2237667 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+[rank7]:W0817 20:07:01.708000 2237668 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+[rank1]:W0817 20:07:01.783000 2237662 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+[rank2]:W0817 20:07:01.853000 2237663 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+[rank4]:W0817 20:07:02.043000 2237665 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+[rank3]:W0817 20:07:02.096000 2237664 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+[rank0]:W0817 20:07:03.366000 2237661 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+
  0%|          | 1/864 [00:41<9:57:23, 41.53s/it]
                                                 
{'loss': '12.08', 'grad_norm': '2.625', 'learning_rate': '0', 'epoch': '0.01396'}
+
  0%|          | 1/864 [00:41<9:57:23, 41.53s/it]
  0%|          | 2/864 [00:42<4:13:20, 17.63s/it]
  0%|          | 3/864 [00:43<2:23:19,  9.99s/it]
  0%|          | 4/864 [00:44<1:31:40,  6.40s/it]
  1%|          | 5/864 [00:45<1:03:08,  4.41s/it]
  1%|          | 6/864 [00:45<45:57,  3.21s/it]  
  1%|          | 7/864 [00:46<35:01,  2.45s/it]
  1%|          | 8/864 [00:47<27:52,  1.95s/it]
  1%|          | 9/864 [00:48<23:06,  1.62s/it]
  1%|          | 10/864 [00:49<19:50,  1.39s/it]
                                                
{'loss': '11.34', 'grad_norm': '1.998', 'learning_rate': '0.000108', 'epoch': '0.1396'}
+
  1%|          | 10/864 [00:49<19:50,  1.39s/it]
  1%|▏         | 11/864 [00:50<17:37,  1.24s/it]
  1%|▏         | 12/864 [00:51<16:05,  1.13s/it]
  2%|▏         | 13/864 [00:52<15:00,  1.06s/it]
  2%|▏         | 14/864 [00:53<14:15,  1.01s/it]
  2%|▏         | 15/864 [00:53<13:43,  1.03it/s]
  2%|▏         | 16/864 [00:54<13:21,  1.06it/s]
  2%|▏         | 17/864 [00:55<13:05,  1.08it/s]
  2%|▏         | 18/864 [00:56<12:53,  1.09it/s]
  2%|▏         | 19/864 [00:57<12:45,  1.10it/s]
  2%|▏         | 20/864 [00:58<12:38,  1.11it/s]
                                                
{'loss': '9.998', 'grad_norm': '1.774', 'learning_rate': '0.000228', 'epoch': '0.2792'}
+
  2%|▏         | 20/864 [00:58<12:38,  1.11it/s]
  2%|▏         | 21/864 [00:59<12:34,  1.12it/s]
  3%|▎         | 22/864 [01:00<12:30,  1.12it/s]
  3%|▎         | 23/864 [01:01<12:28,  1.12it/s]
  3%|▎         | 24/864 [01:01<12:26,  1.13it/s]
  3%|▎         | 25/864 [01:02<12:24,  1.13it/s]
  3%|▎         | 26/864 [01:03<12:23,  1.13it/s]
  3%|▎         | 27/864 [01:04<12:21,  1.13it/s]
  3%|▎         | 28/864 [01:05<12:20,  1.13it/s]
  3%|▎         | 29/864 [01:06<12:19,  1.13it/s]
  3%|▎         | 30/864 [01:07<12:18,  1.13it/s]
                                                
{'loss': '8.387', 'grad_norm': '0.6299', 'learning_rate': '0.000348', 'epoch': '0.4188'}
+
  3%|▎         | 30/864 [01:07<12:18,  1.13it/s]
  4%|▎         | 31/864 [01:08<12:19,  1.13it/s]
  4%|▎         | 32/864 [01:09<12:18,  1.13it/s]
  4%|▍         | 33/864 [01:09<12:18,  1.13it/s]
  4%|▍         | 34/864 [01:10<12:17,  1.13it/s]
  4%|▍         | 35/864 [01:11<12:17,  1.12it/s]
  4%|▍         | 36/864 [01:12<12:16,  1.12it/s]
  4%|▍         | 37/864 [01:13<12:16,  1.12it/s]
  4%|▍         | 38/864 [01:14<12:15,  1.12it/s]
  5%|▍         | 39/864 [01:15<12:14,  1.12it/s]
  5%|▍         | 40/864 [01:16<12:13,  1.12it/s]
                                                
{'loss': '7.641', 'grad_norm': '0.4135', 'learning_rate': '0.000468', 'epoch': '0.5585'}
+
  5%|▍         | 40/864 [01:16<12:13,  1.12it/s]
  5%|▍         | 41/864 [01:17<12:13,  1.12it/s]
  5%|▍         | 42/864 [01:17<12:12,  1.12it/s]
  5%|▍         | 43/864 [01:18<12:11,  1.12it/s]
  5%|▌         | 44/864 [01:19<12:11,  1.12it/s]
  5%|▌         | 45/864 [01:20<12:10,  1.12it/s]
  5%|▌         | 46/864 [01:21<12:08,  1.12it/s]
  5%|▌         | 47/864 [01:22<12:08,  1.12it/s]
  6%|▌         | 48/864 [01:23<12:07,  1.12it/s]
  6%|▌         | 49/864 [01:24<12:06,  1.12it/s]
  6%|▌         | 50/864 [01:25<12:05,  1.12it/s]
                                                
{'loss': '7.483', 'grad_norm': '0.6168', 'learning_rate': '0.000588', 'epoch': '0.6981'}
+
  6%|▌         | 50/864 [01:25<12:05,  1.12it/s]
  6%|▌         | 51/864 [01:25<12:05,  1.12it/s]
  6%|▌         | 52/864 [01:26<12:04,  1.12it/s]
  6%|▌         | 53/864 [01:27<12:03,  1.12it/s]
  6%|▋         | 54/864 [01:28<12:02,  1.12it/s]
  6%|▋         | 55/864 [01:29<12:02,  1.12it/s]
  6%|▋         | 56/864 [01:30<12:01,  1.12it/s]
  7%|▋         | 57/864 [01:31<12:00,  1.12it/s]
  7%|▋         | 58/864 [01:32<12:00,  1.12it/s]
  7%|▋         | 59/864 [01:33<11:59,  1.12it/s]
  7%|▋         | 60/864 [01:34<11:58,  1.12it/s]
                                                
{'loss': '7.347', 'grad_norm': '0.7764', 'learning_rate': '0.000708', 'epoch': '0.8377'}
+
  7%|▋         | 60/864 [01:34<11:58,  1.12it/s]
  7%|▋         | 61/864 [01:34<11:58,  1.12it/s]
  7%|▋         | 62/864 [01:35<11:57,  1.12it/s]
  7%|▋         | 63/864 [01:36<11:56,  1.12it/s]
  7%|▋         | 64/864 [01:37<11:56,  1.12it/s]
  8%|▊         | 65/864 [01:38<11:54,  1.12it/s]
  8%|▊         | 66/864 [01:39<11:54,  1.12it/s]
  8%|▊         | 67/864 [01:40<11:53,  1.12it/s]
  8%|▊         | 68/864 [01:41<11:52,  1.12it/s]
  8%|▊         | 69/864 [01:42<11:51,  1.12it/s]
  8%|▊         | 70/864 [01:42<11:50,  1.12it/s]
                                                
{'loss': '7.113', 'grad_norm': '0.4354', 'learning_rate': '0.000828', 'epoch': '0.9773'}
+
  8%|▊         | 70/864 [01:42<11:50,  1.12it/s]
  8%|▊         | 71/864 [01:43<11:49,  1.12it/s]
  8%|▊         | 72/864 [01:44<10:30,  1.26it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 24.10it/s]
+
 15%|█▌        | 6/39 [00:00<00:02, 12.79it/s]
+
 21%|██        | 8/39 [00:00<00:02, 11.45it/s]
+
 26%|██▌       | 10/39 [00:00<00:02, 10.77it/s]
+
 31%|███       | 12/39 [00:01<00:02, 10.38it/s]
+
 36%|███▌      | 14/39 [00:01<00:02, 10.14it/s]
+
 41%|████      | 16/39 [00:01<00:02,  9.98it/s]
+
 46%|████▌     | 18/39 [00:01<00:02,  9.88it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01,  9.81it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01,  9.77it/s]
+
 56%|█████▋    | 22/39 [00:02<00:01,  9.76it/s]
+
 59%|█████▉    | 23/39 [00:02<00:01,  9.74it/s]
+
 62%|██████▏   | 24/39 [00:02<00:01,  9.72it/s]
+
 64%|██████▍   | 25/39 [00:02<00:01,  9.71it/s]
+
 67%|██████▋   | 26/39 [00:02<00:01,  9.70it/s]
+
 69%|██████▉   | 27/39 [00:02<00:01,  9.69it/s]
+
 72%|███████▏  | 28/39 [00:02<00:01,  9.68it/s]
+
 74%|███████▍  | 29/39 [00:02<00:01,  9.68it/s]
+
 77%|███████▋  | 30/39 [00:02<00:00,  9.67it/s]
+
 79%|███████▉  | 31/39 [00:03<00:00,  9.67it/s]
+
 82%|████████▏ | 32/39 [00:03<00:00,  9.67it/s]
+
 85%|████████▍ | 33/39 [00:03<00:00,  9.67it/s]
+
 87%|████████▋ | 34/39 [00:03<00:00,  9.67it/s]
+
 90%|████████▉ | 35/39 [00:03<00:00,  9.66it/s]
+
 92%|█████████▏| 36/39 [00:03<00:00,  9.67it/s]
+
 95%|█████████▍| 37/39 [00:03<00:00,  9.66it/s]
+
 97%|█████████▋| 38/39 [00:03<00:00,  9.66it/s]
+
100%|██████████| 39/39 [00:03<00:00,  9.64it/s]
                                                
+
                                               
{'eval_loss': '6.976', 'eval_runtime': '28', 'eval_samples_per_second': '87.18', 'eval_steps_per_second': '1.393', 'epoch': '1'}
+
  8%|▊         | 72/864 [02:12<10:30,  1.26it/s]
+
100%|██████████| 39/39 [00:03<00:00,  9.64it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-72
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-72/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-72/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:01<00:00,  1.01s/it]
Writing model shards: 100%|██████████| 1/1 [00:01<00:00,  1.01s/it]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-72/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-72/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-72/tokenizer_config.json
+[eval] epoch 1.0 step 72 loss 6.9759 (best 6.9759 @ ep 1.0, 0/10000 since best)
+
+
                                               
  8%|▊         | 73/864 [02:16<2:14:48, 10.23s/it]
  9%|▊         | 74/864 [02:17<1:37:46,  7.43s/it]
  9%|▊         | 75/864 [02:18<1:11:53,  5.47s/it]
  9%|▉         | 76/864 [02:19<53:46,  4.09s/it]  
  9%|▉         | 77/864 [02:20<41:07,  3.14s/it]
  9%|▉         | 78/864 [02:21<32:16,  2.46s/it]
  9%|▉         | 79/864 [02:22<26:05,  1.99s/it]
  9%|▉         | 80/864 [02:22<21:45,  1.67s/it]
                                                
{'loss': '6.936', 'grad_norm': '0.6768', 'learning_rate': '0.000948', 'epoch': '1.112'}
+
  9%|▉         | 80/864 [02:22<21:45,  1.67s/it]
  9%|▉         | 81/864 [02:23<18:43,  1.43s/it]
  9%|▉         | 82/864 [02:24<16:35,  1.27s/it]
 10%|▉         | 83/864 [02:25<15:06,  1.16s/it]
 10%|▉         | 84/864 [02:26<14:03,  1.08s/it]
 10%|▉         | 85/864 [02:27<13:19,  1.03s/it]
 10%|▉         | 86/864 [02:28<12:47,  1.01it/s]
 10%|█         | 87/864 [02:29<12:25,  1.04it/s]
 10%|█         | 88/864 [02:30<12:09,  1.06it/s]
 10%|█         | 89/864 [02:30<11:58,  1.08it/s]
 10%|█         | 90/864 [02:31<11:50,  1.09it/s]
                                                
{'loss': '6.793', 'grad_norm': '0.5564', 'learning_rate': '0.001068', 'epoch': '1.251'}
+
 10%|█         | 90/864 [02:31<11:50,  1.09it/s]
 11%|█         | 91/864 [02:32<11:44,  1.10it/s]
 11%|█         | 92/864 [02:33<11:40,  1.10it/s]
 11%|█         | 93/864 [02:34<11:36,  1.11it/s]
 11%|█         | 94/864 [02:35<11:34,  1.11it/s]
 11%|█         | 95/864 [02:36<11:31,  1.11it/s]
 11%|█         | 96/864 [02:37<11:30,  1.11it/s]
 11%|█         | 97/864 [02:38<11:29,  1.11it/s]
 11%|█▏        | 98/864 [02:39<11:28,  1.11it/s]
 11%|█▏        | 99/864 [02:39<11:26,  1.11it/s]
 12%|█▏        | 100/864 [02:40<11:25,  1.11it/s]
                                                 
{'loss': '6.683', 'grad_norm': '0.9051', 'learning_rate': '0.001188', 'epoch': '1.391'}
+
 12%|█▏        | 100/864 [02:40<11:25,  1.11it/s]
 12%|█▏        | 101/864 [02:41<11:24,  1.11it/s]
 12%|█▏        | 102/864 [02:42<11:23,  1.11it/s]
 12%|█▏        | 103/864 [02:43<11:23,  1.11it/s]
 12%|█▏        | 104/864 [02:44<11:22,  1.11it/s]
 12%|█▏        | 105/864 [02:45<11:21,  1.11it/s]
 12%|█▏        | 106/864 [02:46<11:20,  1.11it/s]
 12%|█▏        | 107/864 [02:47<11:19,  1.11it/s]
 12%|█▎        | 108/864 [02:48<11:18,  1.11it/s]
 13%|█▎        | 109/864 [02:48<11:17,  1.11it/s]
 13%|█▎        | 110/864 [02:49<11:16,  1.11it/s]
                                                 
{'loss': '6.563', 'grad_norm': '0.3847', 'learning_rate': '0.0012', 'epoch': '1.531'}
+
 13%|█▎        | 110/864 [02:49<11:16,  1.11it/s]
 13%|█▎        | 111/864 [02:50<11:15,  1.11it/s]
 13%|█▎        | 112/864 [02:51<11:14,  1.11it/s]
 13%|█▎        | 113/864 [02:52<11:13,  1.11it/s]
 13%|█▎        | 114/864 [02:53<11:12,  1.11it/s]
 13%|█▎        | 115/864 [02:54<11:12,  1.11it/s]
 13%|█▎        | 116/864 [02:55<11:10,  1.12it/s]
 14%|█▎        | 117/864 [02:56<11:09,  1.12it/s]
 14%|█▎        | 118/864 [02:57<11:09,  1.11it/s]
 14%|█▍        | 119/864 [02:57<11:08,  1.11it/s]
 14%|█▍        | 120/864 [02:58<11:07,  1.12it/s]
                                                 
{'loss': '6.417', 'grad_norm': '0.4136', 'learning_rate': '0.0012', 'epoch': '1.67'}
+
 14%|█▍        | 120/864 [02:58<11:07,  1.12it/s]
 14%|█▍        | 121/864 [02:59<11:06,  1.11it/s]
 14%|█▍        | 122/864 [03:00<11:05,  1.12it/s]
 14%|█▍        | 123/864 [03:01<11:04,  1.12it/s]
 14%|█▍        | 124/864 [03:02<11:03,  1.12it/s]
 14%|█▍        | 125/864 [03:03<11:03,  1.11it/s]
 15%|█▍        | 126/864 [03:04<11:02,  1.11it/s]
 15%|█▍        | 127/864 [03:05<11:00,  1.12it/s]
 15%|█▍        | 128/864 [03:05<11:00,  1.11it/s]
 15%|█▍        | 129/864 [03:06<10:59,  1.12it/s]
 15%|█▌        | 130/864 [03:07<10:58,  1.12it/s]
                                                 
{'loss': '6.321', 'grad_norm': '0.6242', 'learning_rate': '0.0012', 'epoch': '1.81'}
+
 15%|█▌        | 130/864 [03:07<10:58,  1.12it/s]
 15%|█▌        | 131/864 [03:08<10:57,  1.11it/s]
 15%|█▌        | 132/864 [03:09<10:56,  1.11it/s]
 15%|█▌        | 133/864 [03:10<10:55,  1.11it/s]
 16%|█▌        | 134/864 [03:11<10:54,  1.11it/s]
 16%|█▌        | 135/864 [03:12<10:54,  1.11it/s]
 16%|█▌        | 136/864 [03:13<10:53,  1.11it/s]
 16%|█▌        | 137/864 [03:14<10:52,  1.11it/s]
 16%|█▌        | 138/864 [03:14<10:51,  1.11it/s]
 16%|█▌        | 139/864 [03:15<10:50,  1.11it/s]
 16%|█▌        | 140/864 [03:16<10:49,  1.11it/s]
                                                 
{'loss': '6.239', 'grad_norm': '0.4004', 'learning_rate': '0.0012', 'epoch': '1.949'}
+
 16%|█▌        | 140/864 [03:16<10:49,  1.11it/s]
 16%|█▋        | 141/864 [03:17<10:48,  1.11it/s]
 16%|█▋        | 142/864 [03:18<10:48,  1.11it/s]
 17%|█▋        | 143/864 [03:19<10:46,  1.11it/s]
 17%|█▋        | 144/864 [03:20<09:34,  1.25it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:02, 14.43it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 11.77it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 10.80it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 10.35it/s]
+
 28%|██▊       | 11/39 [00:01<00:02, 10.10it/s]
+
 33%|███▎      | 13/39 [00:01<00:02,  9.95it/s]
+
 36%|███▌      | 14/39 [00:01<00:02,  9.89it/s]
+
 38%|███▊      | 15/39 [00:01<00:02,  9.84it/s]
+
 41%|████      | 16/39 [00:01<00:02,  9.79it/s]
+
 44%|████▎     | 17/39 [00:01<00:02,  9.77it/s]
+
 46%|████▌     | 18/39 [00:01<00:02,  9.74it/s]
+
 49%|████▊     | 19/39 [00:01<00:02,  9.72it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01,  9.68it/s]
+
 54%|█████▍    | 21/39 [00:02<00:01,  9.70it/s]
+
 56%|█████▋    | 22/39 [00:02<00:01,  9.68it/s]
+
 59%|█████▉    | 23/39 [00:02<00:01,  9.69it/s]
+
 62%|██████▏   | 24/39 [00:02<00:01,  9.69it/s]
+
 64%|██████▍   | 25/39 [00:02<00:01,  9.66it/s]
+
 67%|██████▋   | 26/39 [00:02<00:01,  9.68it/s]
+
 69%|██████▉   | 27/39 [00:02<00:01,  9.67it/s]
+
 72%|███████▏  | 28/39 [00:02<00:01,  9.67it/s]
+
 74%|███████▍  | 29/39 [00:02<00:01,  9.67it/s]
+
 77%|███████▋  | 30/39 [00:02<00:00,  9.66it/s]
+
 79%|███████▉  | 31/39 [00:03<00:00,  9.66it/s]
+
 82%|████████▏ | 32/39 [00:03<00:00,  9.65it/s]
+
 85%|████████▍ | 33/39 [00:03<00:00,  9.64it/s]
+
 87%|████████▋ | 34/39 [00:03<00:00,  9.64it/s]
+
 90%|████████▉ | 35/39 [00:03<00:00,  9.64it/s]
+
 92%|█████████▏| 36/39 [00:03<00:00,  9.64it/s]
+
 95%|█████████▍| 37/39 [00:03<00:00,  9.65it/s]
+
 97%|█████████▋| 38/39 [00:03<00:00,  9.65it/s]
+
100%|██████████| 39/39 [00:03<00:00,  9.66it/s]
                                                 
+
                                               
{'eval_loss': '6.175', 'eval_runtime': '4.094', 'eval_samples_per_second': '596.2', 'eval_steps_per_second': '9.526', 'epoch': '2'}
+
 17%|█▋        | 144/864 [03:24<09:34,  1.25it/s]
+
100%|██████████| 39/39 [00:04<00:00,  9.66it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-144
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-144/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-144/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.14it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.14it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-144/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-144/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-144/tokenizer_config.json
+[eval] epoch 2.0 step 144 loss 6.1746 (best 6.1746 @ ep 2.0, 0/10000 since best)
+
+
                                               
 17%|█▋        | 145/864 [03:28<36:54,  3.08s/it]
 17%|█▋        | 146/864 [03:29<29:01,  2.43s/it]
 17%|█▋        | 147/864 [03:30<23:30,  1.97s/it]
 17%|█▋        | 148/864 [03:31<19:38,  1.65s/it]
 17%|█▋        | 149/864 [03:32<16:56,  1.42s/it]
 17%|█▋        | 150/864 [03:32<15:02,  1.26s/it]
                                                 
{'loss': '6.135', 'grad_norm': '0.332', 'learning_rate': '0.0012', 'epoch': '2.084'}
+
 17%|█▋        | 150/864 [03:32<15:02,  1.26s/it]
 17%|█▋        | 151/864 [03:33<13:43,  1.15s/it]
 18%|█▊        | 152/864 [03:34<12:46,  1.08s/it]
 18%|█▊        | 153/864 [03:35<12:07,  1.02s/it]
 18%|█▊        | 154/864 [03:36<11:40,  1.01it/s]
 18%|█▊        | 155/864 [03:37<11:20,  1.04it/s]
 18%|█▊        | 156/864 [03:38<11:06,  1.06it/s]
 18%|█▊        | 157/864 [03:39<10:56,  1.08it/s]
 18%|█▊        | 158/864 [03:40<10:49,  1.09it/s]
 18%|█▊        | 159/864 [03:40<10:43,  1.10it/s]
 19%|█▊        | 160/864 [03:41<10:39,  1.10it/s]
                                                 
{'loss': '6.063', 'grad_norm': '0.438', 'learning_rate': '0.0012', 'epoch': '2.223'}
+
 19%|█▊        | 160/864 [03:41<10:39,  1.10it/s]
 19%|█▊        | 161/864 [03:42<10:36,  1.10it/s]
 19%|█▉        | 162/864 [03:43<10:34,  1.11it/s]
 19%|█▉        | 163/864 [03:44<10:31,  1.11it/s]
 19%|█▉        | 164/864 [03:45<10:30,  1.11it/s]
 19%|█▉        | 165/864 [03:46<10:29,  1.11it/s]
 19%|█▉        | 166/864 [03:47<10:27,  1.11it/s]
 19%|█▉        | 167/864 [03:48<10:26,  1.11it/s]
 19%|█▉        | 168/864 [03:49<10:25,  1.11it/s]
 20%|█▉        | 169/864 [03:49<10:24,  1.11it/s]
 20%|█▉        | 170/864 [03:50<10:23,  1.11it/s]
                                                 
{'loss': '5.981', 'grad_norm': '0.4537', 'learning_rate': '0.0012', 'epoch': '2.363'}
+
 20%|█▉        | 170/864 [03:50<10:23,  1.11it/s]
 20%|█▉        | 171/864 [03:51<10:22,  1.11it/s]
 20%|█▉        | 172/864 [03:52<10:21,  1.11it/s]
 20%|██        | 173/864 [03:53<10:20,  1.11it/s]
 20%|██        | 174/864 [03:54<10:19,  1.11it/s]
 20%|██        | 175/864 [03:55<10:18,  1.11it/s]
 20%|██        | 176/864 [03:56<10:18,  1.11it/s]
 20%|██        | 177/864 [03:57<10:18,  1.11it/s]
 21%|██        | 178/864 [03:58<10:16,  1.11it/s]
 21%|██        | 179/864 [03:58<10:15,  1.11it/s]
 21%|██        | 180/864 [03:59<10:15,  1.11it/s]
                                                 
{'loss': '5.905', 'grad_norm': '0.8055', 'learning_rate': '0.0012', 'epoch': '2.503'}
+
 21%|██        | 180/864 [03:59<10:15,  1.11it/s]
 21%|██        | 181/864 [04:00<10:14,  1.11it/s]
 21%|██        | 182/864 [04:01<10:13,  1.11it/s]
 21%|██        | 183/864 [04:02<10:13,  1.11it/s]
 21%|██▏       | 184/864 [04:03<10:12,  1.11it/s]
 21%|██▏       | 185/864 [04:04<10:10,  1.11it/s]
 22%|██▏       | 186/864 [04:05<10:10,  1.11it/s]
 22%|██▏       | 187/864 [04:06<10:09,  1.11it/s]
 22%|██▏       | 188/864 [04:07<10:08,  1.11it/s]
 22%|██▏       | 189/864 [04:07<10:07,  1.11it/s]
 22%|██▏       | 190/864 [04:08<10:06,  1.11it/s]
                                                 
{'loss': '5.888', 'grad_norm': '0.4139', 'learning_rate': '0.0012', 'epoch': '2.642'}
+
 22%|██▏       | 190/864 [04:08<10:06,  1.11it/s]
 22%|██▏       | 191/864 [04:09<10:05,  1.11it/s]
 22%|██▏       | 192/864 [04:10<10:04,  1.11it/s]
 22%|██▏       | 193/864 [04:11<10:03,  1.11it/s]
 22%|██▏       | 194/864 [04:12<10:02,  1.11it/s]
 23%|██▎       | 195/864 [04:13<10:01,  1.11it/s]
 23%|██▎       | 196/864 [04:14<10:00,  1.11it/s]
 23%|██▎       | 197/864 [04:15<09:59,  1.11it/s]
 23%|██▎       | 198/864 [04:16<09:58,  1.11it/s]
 23%|██▎       | 199/864 [04:16<09:57,  1.11it/s]
 23%|██▎       | 200/864 [04:17<09:56,  1.11it/s]
                                                 
{'loss': '5.782', 'grad_norm': '0.3025', 'learning_rate': '0.0012', 'epoch': '2.782'}
+
 23%|██▎       | 200/864 [04:17<09:56,  1.11it/s]
 23%|██▎       | 201/864 [04:18<09:56,  1.11it/s]
 23%|██▎       | 202/864 [04:19<09:55,  1.11it/s]
 23%|██▎       | 203/864 [04:20<09:54,  1.11it/s]
 24%|██▎       | 204/864 [04:21<09:53,  1.11it/s]
 24%|██▎       | 205/864 [04:22<09:52,  1.11it/s]
 24%|██▍       | 206/864 [04:23<09:51,  1.11it/s]
 24%|██▍       | 207/864 [04:24<09:50,  1.11it/s]
 24%|██▍       | 208/864 [04:25<09:49,  1.11it/s]
 24%|██▍       | 209/864 [04:25<09:48,  1.11it/s]
 24%|██▍       | 210/864 [04:26<09:47,  1.11it/s]
                                                 
{'loss': '5.71', 'grad_norm': '0.2841', 'learning_rate': '0.0012', 'epoch': '2.921'}
+
 24%|██▍       | 210/864 [04:26<09:47,  1.11it/s]
 24%|██▍       | 211/864 [04:27<09:47,  1.11it/s]
 25%|██▍       | 212/864 [04:28<09:46,  1.11it/s]
 25%|██▍       | 213/864 [04:29<09:45,  1.11it/s]
 25%|██▍       | 214/864 [04:30<09:44,  1.11it/s]
 25%|██▍       | 215/864 [04:31<09:43,  1.11it/s]
 25%|██▌       | 216/864 [04:31<08:38,  1.25it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:02, 14.33it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 11.75it/s]
+
 18%|█▊        | 7/39 [00:00<00:03, 10.65it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 10.36it/s]
+
 28%|██▊       | 11/39 [00:01<00:02, 10.02it/s]
+
 33%|███▎      | 13/39 [00:01<00:02,  9.97it/s]
+
 36%|███▌      | 14/39 [00:01<00:02,  9.92it/s]
+
 38%|███▊      | 15/39 [00:01<00:02,  9.75it/s]
+
 44%|████▎     | 17/39 [00:01<00:02,  9.80it/s]
+
 46%|████▌     | 18/39 [00:01<00:02,  9.78it/s]
+
 49%|████▊     | 19/39 [00:01<00:02,  9.63it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01,  9.71it/s]
+
 54%|█████▍    | 21/39 [00:02<00:01,  9.72it/s]
+
 56%|█████▋    | 22/39 [00:02<00:01,  9.71it/s]
+
 59%|█████▉    | 23/39 [00:02<00:01,  9.70it/s]
+
 62%|██████▏   | 24/39 [00:02<00:01,  9.68it/s]
+
 64%|██████▍   | 25/39 [00:02<00:01,  9.68it/s]
+
 67%|██████▋   | 26/39 [00:02<00:01,  9.68it/s]
+
 69%|██████▉   | 27/39 [00:02<00:01,  9.67it/s]
+
 72%|███████▏  | 28/39 [00:02<00:01,  9.67it/s]
+
 74%|███████▍  | 29/39 [00:02<00:01,  9.66it/s]
+
 77%|███████▋  | 30/39 [00:03<00:00,  9.67it/s]
+
 79%|███████▉  | 31/39 [00:03<00:00,  9.66it/s]
+
 82%|████████▏ | 32/39 [00:03<00:00,  9.66it/s]
+
 85%|████████▍ | 33/39 [00:03<00:00,  9.65it/s]
+
 87%|████████▋ | 34/39 [00:03<00:00,  9.66it/s]
+
 90%|████████▉ | 35/39 [00:03<00:00,  9.65it/s]
+
 92%|█████████▏| 36/39 [00:03<00:00,  9.65it/s]
+
 95%|█████████▍| 37/39 [00:03<00:00,  9.65it/s]
+
 97%|█████████▋| 38/39 [00:03<00:00,  9.65it/s]
+
100%|██████████| 39/39 [00:03<00:00,  9.66it/s]
                                                 
+
                                               
{'eval_loss': '5.692', 'eval_runtime': '4.097', 'eval_samples_per_second': '595.8', 'eval_steps_per_second': '9.519', 'epoch': '3'}
+
 25%|██▌       | 216/864 [04:35<08:38,  1.25it/s]
+
100%|██████████| 39/39 [00:04<00:00,  9.66it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-216
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-216/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-216/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.15it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.15it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-216/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-216/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-216/tokenizer_config.json
+[eval] epoch 3.0 step 216 loss 5.6916 (best 5.6916 @ ep 3.0, 0/10000 since best)
+
+
                                               
 25%|██▌       | 217/864 [04:40<33:12,  3.08s/it]
 25%|██▌       | 218/864 [04:41<26:06,  2.42s/it]
 25%|██▌       | 219/864 [04:42<21:08,  1.97s/it]
 25%|██▌       | 220/864 [04:42<17:39,  1.65s/it]
                                                 
{'loss': '5.629', 'grad_norm': '0.3829', 'learning_rate': '0.0012', 'epoch': '3.056'}
+
 25%|██▌       | 220/864 [04:42<17:39,  1.65s/it]
 26%|██▌       | 221/864 [04:43<15:13,  1.42s/it]
 26%|██▌       | 222/864 [04:44<13:31,  1.26s/it]
 26%|██▌       | 223/864 [04:45<12:19,  1.15s/it]
 26%|██▌       | 224/864 [04:46<11:28,  1.08s/it]
 26%|██▌       | 225/864 [04:47<10:53,  1.02s/it]
 26%|██▌       | 226/864 [04:48<10:28,  1.01it/s]
 26%|██▋       | 227/864 [04:49<10:10,  1.04it/s]
 26%|██▋       | 228/864 [04:50<09:58,  1.06it/s]
 27%|██▋       | 229/864 [04:51<09:49,  1.08it/s]
 27%|██▋       | 230/864 [04:51<09:42,  1.09it/s]
                                                 
{'loss': '5.542', 'grad_norm': '0.4644', 'learning_rate': '0.0012', 'epoch': '3.195'}
+
 27%|██▋       | 230/864 [04:51<09:42,  1.09it/s]
 27%|██▋       | 231/864 [04:52<09:37,  1.10it/s]
 27%|██▋       | 232/864 [04:53<09:33,  1.10it/s]
 27%|██▋       | 233/864 [04:54<09:31,  1.10it/s]
 27%|██▋       | 234/864 [04:55<09:28,  1.11it/s]
 27%|██▋       | 235/864 [04:56<09:27,  1.11it/s]
 27%|██▋       | 236/864 [04:57<09:25,  1.11it/s]
 27%|██▋       | 237/864 [04:58<09:24,  1.11it/s]
 28%|██▊       | 238/864 [04:59<09:22,  1.11it/s]
 28%|██▊       | 239/864 [05:00<09:21,  1.11it/s]
 28%|██▊       | 240/864 [05:00<09:20,  1.11it/s]
                                                 
{'loss': '5.485', 'grad_norm': '0.5794', 'learning_rate': '0.0012', 'epoch': '3.335'}
+
 28%|██▊       | 240/864 [05:00<09:20,  1.11it/s]
 28%|██▊       | 241/864 [05:01<09:19,  1.11it/s]
 28%|██▊       | 242/864 [05:02<09:18,  1.11it/s]
 28%|██▊       | 243/864 [05:03<09:18,  1.11it/s]
 28%|██▊       | 244/864 [05:04<09:17,  1.11it/s]
 28%|██▊       | 245/864 [05:05<09:16,  1.11it/s]
 28%|██▊       | 246/864 [05:06<09:14,  1.11it/s]
 29%|██▊       | 247/864 [05:07<09:14,  1.11it/s]
 29%|██▊       | 248/864 [05:08<09:12,  1.11it/s]
 29%|██▉       | 249/864 [05:09<09:11,  1.11it/s]
 29%|██▉       | 250/864 [05:09<09:10,  1.12it/s]
                                                 
{'loss': '5.469', 'grad_norm': '0.5608', 'learning_rate': '0.0012', 'epoch': '3.475'}
+
 29%|██▉       | 250/864 [05:09<09:10,  1.12it/s]
 29%|██▉       | 251/864 [05:10<09:09,  1.11it/s]
 29%|██▉       | 252/864 [05:11<09:08,  1.12it/s]
 29%|██▉       | 253/864 [05:12<09:08,  1.11it/s]
 29%|██▉       | 254/864 [05:13<09:07,  1.11it/s]
 30%|██▉       | 255/864 [05:14<09:06,  1.11it/s]
 30%|██▉       | 256/864 [05:15<09:05,  1.12it/s]
 30%|██▉       | 257/864 [05:16<09:04,  1.11it/s]
 30%|██▉       | 258/864 [05:17<09:03,  1.11it/s]
 30%|██▉       | 259/864 [05:17<09:02,  1.12it/s]
 30%|███       | 260/864 [05:18<09:01,  1.11it/s]
                                                 
{'loss': '5.374', 'grad_norm': '0.2417', 'learning_rate': '0.0012', 'epoch': '3.614'}
+
 30%|███       | 260/864 [05:18<09:01,  1.11it/s]
 30%|███       | 261/864 [05:19<09:01,  1.11it/s]
 30%|███       | 262/864 [05:20<08:59,  1.12it/s]
 30%|███       | 263/864 [05:21<08:58,  1.12it/s]
 31%|███       | 264/864 [05:22<08:58,  1.11it/s]
 31%|███       | 265/864 [05:23<08:57,  1.11it/s]
 31%|███       | 266/864 [05:24<08:56,  1.12it/s]
 31%|███       | 267/864 [05:25<08:56,  1.11it/s]
 31%|███       | 268/864 [05:26<08:55,  1.11it/s]
 31%|███       | 269/864 [05:26<08:54,  1.11it/s]
 31%|███▏      | 270/864 [05:27<08:53,  1.11it/s]
                                                 
{'loss': '5.308', 'grad_norm': '0.706', 'learning_rate': '0.0012', 'epoch': '3.754'}
+
 31%|███▏      | 270/864 [05:27<08:53,  1.11it/s]
 31%|███▏      | 271/864 [05:28<08:52,  1.11it/s]
 31%|███▏      | 272/864 [05:29<08:51,  1.11it/s]
 32%|███▏      | 273/864 [05:30<08:50,  1.11it/s]
 32%|███▏      | 274/864 [05:31<08:49,  1.11it/s]
 32%|███▏      | 275/864 [05:32<08:48,  1.11it/s]
 32%|███▏      | 276/864 [05:33<08:47,  1.11it/s]
 32%|███▏      | 277/864 [05:34<08:47,  1.11it/s]
 32%|███▏      | 278/864 [05:35<08:46,  1.11it/s]
 32%|███▏      | 279/864 [05:35<08:44,  1.11it/s]
 32%|███▏      | 280/864 [05:36<08:44,  1.11it/s]
                                                 
{'loss': '5.308', 'grad_norm': '0.2605', 'learning_rate': '0.0012', 'epoch': '3.894'}
+
 32%|███▏      | 280/864 [05:36<08:44,  1.11it/s]
 33%|███▎      | 281/864 [05:37<08:43,  1.11it/s]
 33%|███▎      | 282/864 [05:38<08:41,  1.12it/s]
 33%|███▎      | 283/864 [05:39<08:41,  1.11it/s]
 33%|███▎      | 284/864 [05:40<08:40,  1.11it/s]
 33%|███▎      | 285/864 [05:41<08:39,  1.11it/s]
 33%|███▎      | 286/864 [05:42<08:38,  1.11it/s]
 33%|███▎      | 287/864 [05:43<08:38,  1.11it/s]
 33%|███▎      | 288/864 [05:43<07:39,  1.25it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:02, 14.55it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 11.81it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 10.70it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 10.37it/s]
+
 28%|██▊       | 11/39 [00:01<00:02, 10.02it/s]
+
 33%|███▎      | 13/39 [00:01<00:02,  9.97it/s]
+
 36%|███▌      | 14/39 [00:01<00:02,  9.92it/s]
+
 38%|███▊      | 15/39 [00:01<00:02,  9.75it/s]
+
 44%|████▎     | 17/39 [00:01<00:02,  9.80it/s]
+
 46%|████▌     | 18/39 [00:01<00:02,  9.78it/s]
+
 49%|████▊     | 19/39 [00:01<00:02,  9.63it/s]
+
 54%|█████▍    | 21/39 [00:02<00:01,  9.72it/s]
+
 56%|█████▋    | 22/39 [00:02<00:01,  9.72it/s]
+
 59%|█████▉    | 23/39 [00:02<00:01,  9.70it/s]
+
 62%|██████▏   | 24/39 [00:02<00:01,  9.69it/s]
+
 64%|██████▍   | 25/39 [00:02<00:01,  9.68it/s]
+
 67%|██████▋   | 26/39 [00:02<00:01,  9.67it/s]
+
 69%|██████▉   | 27/39 [00:02<00:01,  9.66it/s]
+
 72%|███████▏  | 28/39 [00:02<00:01,  9.66it/s]
+
 74%|███████▍  | 29/39 [00:02<00:01,  9.66it/s]
+
 77%|███████▋  | 30/39 [00:03<00:00,  9.66it/s]
+
 79%|███████▉  | 31/39 [00:03<00:00,  9.66it/s]
+
 82%|████████▏ | 32/39 [00:03<00:00,  9.65it/s]
+
 85%|████████▍ | 33/39 [00:03<00:00,  9.66it/s]
+
 87%|████████▋ | 34/39 [00:03<00:00,  9.66it/s]
+
 90%|████████▉ | 35/39 [00:03<00:00,  9.65it/s]
+
 92%|█████████▏| 36/39 [00:03<00:00,  9.66it/s]
+
 95%|█████████▍| 37/39 [00:03<00:00,  9.66it/s]
+
 97%|█████████▋| 38/39 [00:03<00:00,  9.66it/s]
+
100%|██████████| 39/39 [00:03<00:00,  9.66it/s]
                                                 
+
                                               
{'eval_loss': '5.226', 'eval_runtime': '4.087', 'eval_samples_per_second': '597.3', 'eval_steps_per_second': '9.542', 'epoch': '4'}
+
 33%|███▎      | 288/864 [05:47<07:39,  1.25it/s]
+
100%|██████████| 39/39 [00:04<00:00,  9.66it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-288
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-288/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-288/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.14it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.14it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-288/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-288/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-288/tokenizer_config.json
+[eval] epoch 4.0 step 288 loss 5.2262 (best 5.2262 @ ep 4.0, 0/10000 since best)
+
+
                                               
 33%|███▎      | 289/864 [05:52<29:33,  3.08s/it]
 34%|███▎      | 290/864 [05:53<23:13,  2.43s/it]
                                                 
{'loss': '5.167', 'grad_norm': '0.2077', 'learning_rate': '0.0012', 'epoch': '4.028'}
+
 34%|███▎      | 290/864 [05:53<23:13,  2.43s/it]
 34%|███▎      | 291/864 [05:53<18:47,  1.97s/it]
 34%|███▍      | 292/864 [05:54<15:41,  1.65s/it]
 34%|███▍      | 293/864 [05:55<13:31,  1.42s/it]
 34%|███▍      | 294/864 [05:56<12:00,  1.26s/it]
 34%|███▍      | 295/864 [05:57<10:56,  1.15s/it]
 34%|███▍      | 296/864 [05:58<10:11,  1.08s/it]
 34%|███▍      | 297/864 [05:59<09:40,  1.02s/it]
 34%|███▍      | 298/864 [06:00<09:17,  1.01it/s]
 35%|███▍      | 299/864 [06:01<09:01,  1.04it/s]
 35%|███▍      | 300/864 [06:01<08:50,  1.06it/s]
                                                 
{'loss': '5.102', 'grad_norm': '0.5714', 'learning_rate': '0.0012', 'epoch': '4.168'}
+
 35%|███▍      | 300/864 [06:01<08:50,  1.06it/s]
 35%|███▍      | 301/864 [06:02<08:42,  1.08it/s]
 35%|███▍      | 302/864 [06:03<08:36,  1.09it/s]
 35%|███▌      | 303/864 [06:04<08:32,  1.10it/s]
 35%|███▌      | 304/864 [06:05<08:28,  1.10it/s]
 35%|███▌      | 305/864 [06:06<08:26,  1.10it/s]
 35%|███▌      | 306/864 [06:07<08:24,  1.11it/s]
 36%|███▌      | 307/864 [06:08<08:22,  1.11it/s]
 36%|███▌      | 308/864 [06:09<08:21,  1.11it/s]
 36%|███▌      | 309/864 [06:10<08:19,  1.11it/s]
 36%|███▌      | 310/864 [06:10<08:18,  1.11it/s]
                                                 
{'loss': '5.078', 'grad_norm': '0.2359', 'learning_rate': '0.0012', 'epoch': '4.307'}
+
 36%|███▌      | 310/864 [06:10<08:18,  1.11it/s]
 36%|███▌      | 311/864 [06:11<08:17,  1.11it/s]
 36%|███▌      | 312/864 [06:12<08:16,  1.11it/s]
 36%|███▌      | 313/864 [06:13<08:15,  1.11it/s]
 36%|███▋      | 314/864 [06:14<08:13,  1.11it/s]
 36%|███▋      | 315/864 [06:15<08:12,  1.11it/s]
 37%|███▋      | 316/864 [06:16<08:12,  1.11it/s]
 37%|███▋      | 317/864 [06:17<08:10,  1.11it/s]
 37%|███▋      | 318/864 [06:18<08:09,  1.11it/s]
 37%|███▋      | 319/864 [06:19<08:09,  1.11it/s]
 37%|███▋      | 320/864 [06:19<08:08,  1.11it/s]
                                                 
{'loss': '4.97', 'grad_norm': '0.1798', 'learning_rate': '0.0012', 'epoch': '4.447'}
+
 37%|███▋      | 320/864 [06:19<08:08,  1.11it/s]
 37%|███▋      | 321/864 [06:20<08:07,  1.11it/s]
 37%|███▋      | 322/864 [06:21<08:06,  1.11it/s]
 37%|███▋      | 323/864 [06:22<08:05,  1.11it/s]
 38%|███▊      | 324/864 [06:23<08:04,  1.11it/s]
 38%|███▊      | 325/864 [06:24<08:03,  1.11it/s]
 38%|███▊      | 326/864 [06:25<08:02,  1.12it/s]
 38%|███▊      | 327/864 [06:26<08:01,  1.12it/s]
 38%|███▊      | 328/864 [06:27<08:00,  1.12it/s]
 38%|███▊      | 329/864 [06:28<07:59,  1.12it/s]
 38%|███▊      | 330/864 [06:28<07:58,  1.12it/s]
                                                 
{'loss': '4.962', 'grad_norm': '0.625', 'learning_rate': '0.0012', 'epoch': '4.586'}
+
 38%|███▊      | 330/864 [06:28<07:58,  1.12it/s]
 38%|███▊      | 331/864 [06:29<07:58,  1.11it/s]
 38%|███▊      | 332/864 [06:30<07:57,  1.11it/s]
 39%|███▊      | 333/864 [06:31<07:56,  1.11it/s]
 39%|███▊      | 334/864 [06:32<07:55,  1.11it/s]
 39%|███▉      | 335/864 [06:33<07:54,  1.11it/s]
 39%|███▉      | 336/864 [06:34<07:53,  1.11it/s]
 39%|███▉      | 337/864 [06:35<07:53,  1.11it/s]
 39%|███▉      | 338/864 [06:36<07:52,  1.11it/s]
 39%|███▉      | 339/864 [06:36<07:51,  1.11it/s]
 39%|███▉      | 340/864 [06:37<07:50,  1.11it/s]
                                                 
{'loss': '4.943', 'grad_norm': '0.5019', 'learning_rate': '0.0012', 'epoch': '4.726'}
+
 39%|███▉      | 340/864 [06:37<07:50,  1.11it/s]
 39%|███▉      | 341/864 [06:38<07:49,  1.11it/s]
 40%|███▉      | 342/864 [06:39<07:49,  1.11it/s]
 40%|███▉      | 343/864 [06:40<07:48,  1.11it/s]
 40%|███▉      | 344/864 [06:41<07:46,  1.11it/s]
 40%|███▉      | 345/864 [06:42<07:45,  1.11it/s]
 40%|████      | 346/864 [06:43<07:45,  1.11it/s]
 40%|████      | 347/864 [06:44<07:44,  1.11it/s]
 40%|████      | 348/864 [06:45<07:43,  1.11it/s]
 40%|████      | 349/864 [06:45<07:42,  1.11it/s]
 41%|████      | 350/864 [06:46<07:41,  1.11it/s]
                                                 
{'loss': '4.831', 'grad_norm': '0.1806', 'learning_rate': '0.0012', 'epoch': '4.866'}
+
 41%|████      | 350/864 [06:46<07:41,  1.11it/s]
 41%|████      | 351/864 [06:47<07:40,  1.11it/s]
 41%|████      | 352/864 [06:48<07:39,  1.11it/s]
 41%|████      | 353/864 [06:49<07:38,  1.11it/s]
 41%|████      | 354/864 [06:50<07:37,  1.11it/s]
 41%|████      | 355/864 [06:51<07:36,  1.11it/s]
 41%|████      | 356/864 [06:52<07:36,  1.11it/s]
 41%|████▏     | 357/864 [06:53<07:35,  1.11it/s]
 41%|████▏     | 358/864 [06:54<07:34,  1.11it/s]
 42%|████▏     | 359/864 [06:54<07:33,  1.11it/s]
 42%|████▏     | 360/864 [06:55<06:42,  1.25it/s]
                                                 
{'loss': '4.768', 'grad_norm': '0.4216', 'learning_rate': '0.0012', 'epoch': '5'}
+
 42%|████▏     | 360/864 [06:55<06:42,  1.25it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:02, 15.01it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 11.80it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 10.80it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 10.34it/s]
+
 28%|██▊       | 11/39 [00:01<00:02, 10.09it/s]
+
 33%|███▎      | 13/39 [00:01<00:02,  9.94it/s]
+
 36%|███▌      | 14/39 [00:01<00:02,  9.80it/s]
+
 38%|███▊      | 15/39 [00:01<00:02,  9.80it/s]
+
 41%|████      | 16/39 [00:01<00:02,  9.77it/s]
+
 44%|████▎     | 17/39 [00:01<00:02,  9.74it/s]
+
 46%|████▌     | 18/39 [00:01<00:02,  9.60it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01,  9.71it/s]
+
 54%|█████▍    | 21/39 [00:02<00:01,  9.70it/s]
+
 56%|█████▋    | 22/39 [00:02<00:01,  9.58it/s]
+
 62%|██████▏   | 24/39 [00:02<00:01,  9.70it/s]
+
 64%|██████▍   | 25/39 [00:02<00:01,  9.70it/s]
+
 67%|██████▋   | 26/39 [00:02<00:01,  9.68it/s]
+
 69%|██████▉   | 27/39 [00:02<00:01,  9.68it/s]
+
 72%|███████▏  | 28/39 [00:02<00:01,  9.67it/s]
+
 74%|███████▍  | 29/39 [00:02<00:01,  9.67it/s]
+
 77%|███████▋  | 30/39 [00:03<00:00,  9.67it/s]
+
 79%|███████▉  | 31/39 [00:03<00:00,  9.66it/s]
+
 82%|████████▏ | 32/39 [00:03<00:00,  9.67it/s]
+
 85%|████████▍ | 33/39 [00:03<00:00,  9.67it/s]
+
 87%|████████▋ | 34/39 [00:03<00:00,  9.66it/s]
+
 90%|████████▉ | 35/39 [00:03<00:00,  9.66it/s]
+
 92%|█████████▏| 36/39 [00:03<00:00,  9.66it/s]
+
 95%|█████████▍| 37/39 [00:03<00:00,  9.67it/s]
+
 97%|█████████▋| 38/39 [00:03<00:00,  9.67it/s]
+
100%|██████████| 39/39 [00:03<00:00,  9.66it/s]
                                                 
+
                                               
{'eval_loss': '4.854', 'eval_runtime': '4.089', 'eval_samples_per_second': '596.9', 'eval_steps_per_second': '9.537', 'epoch': '5'}
+
 42%|████▏     | 360/864 [06:59<06:42,  1.25it/s]
+
100%|██████████| 39/39 [00:04<00:00,  9.66it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-360
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-360/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-360/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.14it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.14it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-360/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-360/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-360/tokenizer_config.json
+[eval] epoch 5.0 step 360 loss 4.8543 (best 4.8543 @ ep 5.0, 0/10000 since best)
+
+
                                               
 42%|████▏     | 361/864 [07:03<25:56,  3.09s/it]
 42%|████▏     | 362/864 [07:04<20:22,  2.44s/it]
 42%|████▏     | 363/864 [07:05<16:28,  1.97s/it]
 42%|████▏     | 364/864 [07:06<13:45,  1.65s/it]
 42%|████▏     | 365/864 [07:07<11:51,  1.43s/it]
 42%|████▏     | 366/864 [07:08<10:31,  1.27s/it]
 42%|████▏     | 367/864 [07:09<09:34,  1.16s/it]
 43%|████▎     | 368/864 [07:10<08:55,  1.08s/it]
 43%|████▎     | 369/864 [07:11<08:27,  1.02s/it]
 43%|████▎     | 370/864 [07:12<08:07,  1.01it/s]
                                                 
{'loss': '4.723', 'grad_norm': '0.4852', 'learning_rate': '0.0012', 'epoch': '5.14'}
+
 43%|████▎     | 370/864 [07:12<08:07,  1.01it/s]
 43%|████▎     | 371/864 [07:12<07:53,  1.04it/s]
 43%|████▎     | 372/864 [07:13<07:43,  1.06it/s]
 43%|████▎     | 373/864 [07:14<07:36,  1.08it/s]
 43%|████▎     | 374/864 [07:15<07:30,  1.09it/s]
 43%|████▎     | 375/864 [07:16<07:26,  1.10it/s]
 44%|████▎     | 376/864 [07:17<07:23,  1.10it/s]
 44%|████▎     | 377/864 [07:18<07:21,  1.10it/s]
 44%|████▍     | 378/864 [07:19<07:19,  1.11it/s]
 44%|████▍     | 379/864 [07:20<07:17,  1.11it/s]
 44%|████▍     | 380/864 [07:21<07:16,  1.11it/s]
                                                 
{'loss': '4.675', 'grad_norm': '0.2099', 'learning_rate': '0.0012', 'epoch': '5.279'}
+
 44%|████▍     | 380/864 [07:21<07:16,  1.11it/s]
 44%|████▍     | 381/864 [07:21<07:15,  1.11it/s]
 44%|████▍     | 382/864 [07:22<07:13,  1.11it/s]
 44%|████▍     | 383/864 [07:23<07:12,  1.11it/s]
 44%|████▍     | 384/864 [07:24<07:11,  1.11it/s]
 45%|████▍     | 385/864 [07:25<07:10,  1.11it/s]
 45%|████▍     | 386/864 [07:26<07:09,  1.11it/s]
 45%|████▍     | 387/864 [07:27<07:09,  1.11it/s]
 45%|████▍     | 388/864 [07:28<07:07,  1.11it/s]
 45%|████▌     | 389/864 [07:29<07:06,  1.11it/s]
 45%|████▌     | 390/864 [07:30<07:06,  1.11it/s]
                                                 
{'loss': '4.578', 'grad_norm': '0.1893', 'learning_rate': '0.0012', 'epoch': '5.419'}
+
 45%|████▌     | 390/864 [07:30<07:06,  1.11it/s]
 45%|████▌     | 391/864 [07:30<07:05,  1.11it/s]
 45%|████▌     | 392/864 [07:31<07:04,  1.11it/s]
 45%|████▌     | 393/864 [07:32<07:03,  1.11it/s]
 46%|████▌     | 394/864 [07:33<07:02,  1.11it/s]
 46%|████▌     | 395/864 [07:34<07:01,  1.11it/s]
 46%|████▌     | 396/864 [07:35<07:00,  1.11it/s]
 46%|████▌     | 397/864 [07:36<06:59,  1.11it/s]
 46%|████▌     | 398/864 [07:37<06:58,  1.11it/s]
 46%|████▌     | 399/864 [07:38<06:57,  1.11it/s]
 46%|████▋     | 400/864 [07:39<06:56,  1.11it/s]
                                                 
{'loss': '4.529', 'grad_norm': '0.5502', 'learning_rate': '0.0012', 'epoch': '5.558'}
+
 46%|████▋     | 400/864 [07:39<06:56,  1.11it/s]
 46%|████▋     | 401/864 [07:39<06:56,  1.11it/s]
 47%|████▋     | 402/864 [07:40<06:55,  1.11it/s]
 47%|████▋     | 403/864 [07:41<06:54,  1.11it/s]
 47%|████▋     | 404/864 [07:42<06:53,  1.11it/s]
 47%|████▋     | 405/864 [07:43<06:52,  1.11it/s]
 47%|████▋     | 406/864 [07:44<06:51,  1.11it/s]
 47%|████▋     | 407/864 [07:45<06:50,  1.11it/s]
 47%|████▋     | 408/864 [07:46<06:49,  1.11it/s]
 47%|████▋     | 409/864 [07:47<06:48,  1.11it/s]
 47%|████▋     | 410/864 [07:47<06:48,  1.11it/s]
                                                 
{'loss': '4.551', 'grad_norm': '0.3368', 'learning_rate': '0.0012', 'epoch': '5.698'}
+
 47%|████▋     | 410/864 [07:48<06:48,  1.11it/s]
 48%|████▊     | 411/864 [07:48<06:47,  1.11it/s]
 48%|████▊     | 412/864 [07:49<06:46,  1.11it/s]
 48%|████▊     | 413/864 [07:50<06:45,  1.11it/s]
 48%|████▊     | 414/864 [07:51<06:44,  1.11it/s]
 48%|████▊     | 415/864 [07:52<06:43,  1.11it/s]
 48%|████▊     | 416/864 [07:53<06:42,  1.11it/s]
 48%|████▊     | 417/864 [07:54<06:41,  1.11it/s]
 48%|████▊     | 418/864 [07:55<06:40,  1.11it/s]
 48%|████▊     | 419/864 [07:56<06:39,  1.11it/s]
 49%|████▊     | 420/864 [07:56<06:38,  1.11it/s]
                                                 
{'loss': '4.493', 'grad_norm': '0.1552', 'learning_rate': '0.0012', 'epoch': '5.838'}
+
 49%|████▊     | 420/864 [07:56<06:38,  1.11it/s]
 49%|████▊     | 421/864 [07:57<06:37,  1.11it/s]
 49%|████▉     | 422/864 [07:58<06:37,  1.11it/s]
 49%|████▉     | 423/864 [07:59<06:36,  1.11it/s]
 49%|████▉     | 424/864 [08:00<06:35,  1.11it/s]
 49%|████▉     | 425/864 [08:01<06:34,  1.11it/s]
 49%|████▉     | 426/864 [08:02<06:33,  1.11it/s]
 49%|████▉     | 427/864 [08:03<06:32,  1.11it/s]
 50%|████▉     | 428/864 [08:04<06:31,  1.11it/s]
 50%|████▉     | 429/864 [08:05<06:30,  1.11it/s]
 50%|████▉     | 430/864 [08:05<06:30,  1.11it/s]
                                                 
{'loss': '4.434', 'grad_norm': '0.3857', 'learning_rate': '0.0012', 'epoch': '5.977'}
+
 50%|████▉     | 430/864 [08:05<06:30,  1.11it/s]
 50%|████▉     | 431/864 [08:06<06:29,  1.11it/s]
 50%|█████     | 432/864 [08:07<05:45,  1.25it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:02, 14.90it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 11.75it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 10.78it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 10.33it/s]
+
 28%|██▊       | 11/39 [00:01<00:02, 10.08it/s]
+
 33%|███▎      | 13/39 [00:01<00:02,  9.93it/s]
+
 36%|███▌      | 14/39 [00:01<00:02,  9.87it/s]
+
 38%|███▊      | 15/39 [00:01<00:02,  9.82it/s]
+
 41%|████      | 16/39 [00:01<00:02,  9.78it/s]
+
 44%|████▎     | 17/39 [00:01<00:02,  9.74it/s]
+
 46%|████▌     | 18/39 [00:01<00:02,  9.71it/s]
+
 49%|████▊     | 19/39 [00:01<00:02,  9.69it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01,  9.68it/s]
+
 54%|█████▍    | 21/39 [00:02<00:01,  9.67it/s]
+
 56%|█████▋    | 22/39 [00:02<00:01,  9.65it/s]
+
 59%|█████▉    | 23/39 [00:02<00:01,  9.67it/s]
+
 62%|██████▏   | 24/39 [00:02<00:01,  9.67it/s]
+
 64%|██████▍   | 25/39 [00:02<00:01,  9.67it/s]
+
 67%|██████▋   | 26/39 [00:02<00:01,  9.66it/s]
+
 69%|██████▉   | 27/39 [00:02<00:01,  9.66it/s]
+
 72%|███████▏  | 28/39 [00:02<00:01,  9.66it/s]
+
 74%|███████▍  | 29/39 [00:02<00:01,  9.65it/s]
+
 77%|███████▋  | 30/39 [00:03<00:00,  9.65it/s]
+
 79%|███████▉  | 31/39 [00:03<00:00,  9.65it/s]
+
 82%|████████▏ | 32/39 [00:03<00:00,  9.65it/s]
+
 85%|████████▍ | 33/39 [00:03<00:00,  9.65it/s]
+
 87%|████████▋ | 34/39 [00:03<00:00,  9.65it/s]
+
 90%|████████▉ | 35/39 [00:03<00:00,  9.65it/s]
+
 92%|█████████▏| 36/39 [00:03<00:00,  9.66it/s]
+
 95%|█████████▍| 37/39 [00:03<00:00,  9.66it/s]
+
 97%|█████████▋| 38/39 [00:03<00:00,  9.65it/s]
+
100%|██████████| 39/39 [00:03<00:00,  9.65it/s]
                                                 
+
                                               
{'eval_loss': '4.608', 'eval_runtime': '4.088', 'eval_samples_per_second': '597.1', 'eval_steps_per_second': '9.54', 'epoch': '6'}
+
 50%|█████     | 432/864 [08:11<05:45,  1.25it/s]
+
100%|██████████| 39/39 [00:04<00:00,  9.65it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-432
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-432/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-432/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:01<00:00,  1.02s/it]
Writing model shards: 100%|██████████| 1/1 [00:01<00:00,  1.02s/it]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-432/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-432/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-432/tokenizer_config.json
+[eval] epoch 6.0 step 432 loss 4.6084 (best 4.6084 @ ep 6.0, 0/10000 since best)
+
+
                                               
 50%|█████     | 433/864 [08:15<22:26,  3.12s/it]
 50%|█████     | 434/864 [08:16<17:35,  2.46s/it]
 50%|█████     | 435/864 [08:17<14:13,  1.99s/it]
 50%|█████     | 436/864 [08:18<11:51,  1.66s/it]
 51%|█████     | 437/864 [08:19<10:12,  1.44s/it]
 51%|█████     | 438/864 [08:20<09:03,  1.28s/it]
 51%|█████     | 439/864 [08:21<08:14,  1.16s/it]
 51%|█████     | 440/864 [08:22<07:39,  1.08s/it]
                                                 
{'loss': '4.424', 'grad_norm': '0.482', 'learning_rate': '0.0012', 'epoch': '6.112'}
+
 51%|█████     | 440/864 [08:22<07:39,  1.08s/it]
 51%|█████     | 441/864 [08:23<07:14,  1.03s/it]
 51%|█████     | 442/864 [08:24<06:57,  1.01it/s]
 51%|█████▏    | 443/864 [08:24<06:45,  1.04it/s]
 51%|█████▏    | 444/864 [08:25<06:36,  1.06it/s]
 52%|█████▏    | 445/864 [08:26<06:30,  1.07it/s]
 52%|█████▏    | 446/864 [08:27<06:25,  1.08it/s]
 52%|█████▏    | 447/864 [08:28<06:21,  1.09it/s]
 52%|█████▏    | 448/864 [08:29<06:18,  1.10it/s]
 52%|█████▏    | 449/864 [08:30<06:16,  1.10it/s]
 52%|█████▏    | 450/864 [08:31<06:14,  1.10it/s]
                                                 
{'loss': '4.39', 'grad_norm': '0.2076', 'learning_rate': '0.0012', 'epoch': '6.251'}
+
 52%|█████▏    | 450/864 [08:31<06:14,  1.10it/s]
 52%|█████▏    | 451/864 [08:32<06:13,  1.11it/s]
 52%|█████▏    | 452/864 [08:33<06:12,  1.11it/s]
 52%|█████▏    | 453/864 [08:33<06:11,  1.11it/s]
 53%|█████▎    | 454/864 [08:34<06:09,  1.11it/s]
 53%|█████▎    | 455/864 [08:35<06:09,  1.11it/s]
 53%|█████▎    | 456/864 [08:36<06:07,  1.11it/s]
 53%|█████▎    | 457/864 [08:37<06:06,  1.11it/s]
 53%|█████▎    | 458/864 [08:38<06:05,  1.11it/s]
 53%|█████▎    | 459/864 [08:39<06:04,  1.11it/s]
 53%|█████▎    | 460/864 [08:40<06:04,  1.11it/s]
                                                 
{'loss': '4.304', 'grad_norm': '0.1553', 'learning_rate': '0.0012', 'epoch': '6.391'}
+
 53%|█████▎    | 460/864 [08:40<06:04,  1.11it/s]
 53%|█████▎    | 461/864 [08:41<06:03,  1.11it/s]
 53%|█████▎    | 462/864 [08:42<06:02,  1.11it/s]
 54%|█████▎    | 463/864 [08:43<06:01,  1.11it/s]
 54%|█████▎    | 464/864 [08:43<06:00,  1.11it/s]
 54%|█████▍    | 465/864 [08:44<05:59,  1.11it/s]
 54%|█████▍    | 466/864 [08:45<05:58,  1.11it/s]
 54%|█████▍    | 467/864 [08:46<05:57,  1.11it/s]
 54%|█████▍    | 468/864 [08:47<05:56,  1.11it/s]
 54%|█████▍    | 469/864 [08:48<05:55,  1.11it/s]
 54%|█████▍    | 470/864 [08:49<05:54,  1.11it/s]
                                                 
{'loss': '4.26', 'grad_norm': '0.1667', 'learning_rate': '0.0012', 'epoch': '6.531'}
+
 54%|█████▍    | 470/864 [08:49<05:54,  1.11it/s]
 55%|█████▍    | 471/864 [08:50<05:53,  1.11it/s]
 55%|█████▍    | 472/864 [08:51<05:53,  1.11it/s]
 55%|█████▍    | 473/864 [08:52<05:52,  1.11it/s]
 55%|█████▍    | 474/864 [08:52<05:51,  1.11it/s]
 55%|█████▍    | 475/864 [08:53<05:50,  1.11it/s]
 55%|█████▌    | 476/864 [08:54<05:49,  1.11it/s]
 55%|█████▌    | 477/864 [08:55<05:48,  1.11it/s]
 55%|█████▌    | 478/864 [08:56<05:47,  1.11it/s]
 55%|█████▌    | 479/864 [08:57<05:47,  1.11it/s]
 56%|█████▌    | 480/864 [08:58<05:46,  1.11it/s]
                                                 
{'loss': '4.244', 'grad_norm': '0.4014', 'learning_rate': '0.0012', 'epoch': '6.67'}
+
 56%|█████▌    | 480/864 [08:58<05:46,  1.11it/s]
 56%|█████▌    | 481/864 [08:59<05:45,  1.11it/s]
 56%|█████▌    | 482/864 [09:00<05:44,  1.11it/s]
 56%|█████▌    | 483/864 [09:01<05:43,  1.11it/s]
 56%|█████▌    | 484/864 [09:01<05:42,  1.11it/s]
 56%|█████▌    | 485/864 [09:02<05:41,  1.11it/s]
 56%|█████▋    | 486/864 [09:03<05:39,  1.11it/s]
 56%|█████▋    | 487/864 [09:04<05:38,  1.11it/s]
 56%|█████▋    | 488/864 [09:05<05:37,  1.11it/s]
 57%|█████▋    | 489/864 [09:06<05:37,  1.11it/s]
 57%|█████▋    | 490/864 [09:07<05:36,  1.11it/s]
                                                 
{'loss': '4.26', 'grad_norm': '0.1674', 'learning_rate': '0.0012', 'epoch': '6.81'}
+
 57%|█████▋    | 490/864 [09:07<05:36,  1.11it/s]
 57%|█████▋    | 491/864 [09:08<05:35,  1.11it/s]
 57%|█████▋    | 492/864 [09:09<05:34,  1.11it/s]
 57%|█████▋    | 493/864 [09:10<05:34,  1.11it/s]
 57%|█████▋    | 494/864 [09:10<05:33,  1.11it/s]
 57%|█████▋    | 495/864 [09:11<05:32,  1.11it/s]
 57%|█████▋    | 496/864 [09:12<05:31,  1.11it/s]
 58%|█████▊    | 497/864 [09:13<05:30,  1.11it/s]
 58%|█████▊    | 498/864 [09:14<05:29,  1.11it/s]
 58%|█████▊    | 499/864 [09:15<05:28,  1.11it/s]
 58%|█████▊    | 500/864 [09:16<05:27,  1.11it/s]
                                                 
{'loss': '4.197', 'grad_norm': '0.1954', 'learning_rate': '0.0012', 'epoch': '6.949'}
+
 58%|█████▊    | 500/864 [09:16<05:27,  1.11it/s]
 58%|█████▊    | 501/864 [09:17<05:26,  1.11it/s]
 58%|█████▊    | 502/864 [09:18<05:26,  1.11it/s]
 58%|█████▊    | 503/864 [09:19<05:25,  1.11it/s]
 58%|█████▊    | 504/864 [09:19<04:48,  1.25it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:02, 15.32it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 11.91it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 10.85it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 10.36it/s]
+
 28%|██▊       | 11/39 [00:01<00:02, 10.10it/s]
+
 33%|███▎      | 13/39 [00:01<00:02,  9.94it/s]
+
 36%|███▌      | 14/39 [00:01<00:02,  9.89it/s]
+
 38%|███▊      | 15/39 [00:01<00:02,  9.83it/s]
+
 41%|████      | 16/39 [00:01<00:02,  9.79it/s]
+
 44%|████▎     | 17/39 [00:01<00:02,  9.75it/s]
+
 46%|████▌     | 18/39 [00:01<00:02,  9.72it/s]
+
 49%|████▊     | 19/39 [00:01<00:02,  9.70it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01,  9.69it/s]
+
 54%|█████▍    | 21/39 [00:02<00:01,  9.66it/s]
+
 56%|█████▋    | 22/39 [00:02<00:01,  9.67it/s]
+
 59%|█████▉    | 23/39 [00:02<00:01,  9.69it/s]
+
 62%|██████▏   | 24/39 [00:02<00:01,  9.68it/s]
+
 64%|██████▍   | 25/39 [00:02<00:01,  9.67it/s]
+
 67%|██████▋   | 26/39 [00:02<00:01,  9.67it/s]
+
 69%|██████▉   | 27/39 [00:02<00:01,  9.67it/s]
+
 72%|███████▏  | 28/39 [00:02<00:01,  9.66it/s]
+
 74%|███████▍  | 29/39 [00:02<00:01,  9.66it/s]
+
 77%|███████▋  | 30/39 [00:02<00:00,  9.66it/s]
+
 79%|███████▉  | 31/39 [00:03<00:00,  9.66it/s]
+
 82%|████████▏ | 32/39 [00:03<00:00,  9.66it/s]
+
 85%|████████▍ | 33/39 [00:03<00:00,  9.66it/s]
+
 87%|████████▋ | 34/39 [00:03<00:00,  9.65it/s]
+
 90%|████████▉ | 35/39 [00:03<00:00,  9.66it/s]
+
 92%|█████████▏| 36/39 [00:03<00:00,  9.66it/s]
+
 95%|█████████▍| 37/39 [00:03<00:00,  9.65it/s]
+
 97%|█████████▋| 38/39 [00:03<00:00,  9.65it/s]
+
100%|██████████| 39/39 [00:03<00:00,  9.66it/s]
                                                 
+
                                               
{'eval_loss': '4.355', 'eval_runtime': '4.088', 'eval_samples_per_second': '597.1', 'eval_steps_per_second': '9.54', 'epoch': '7'}
+
 58%|█████▊    | 504/864 [09:23<04:48,  1.25it/s]
+
100%|██████████| 39/39 [00:04<00:00,  9.66it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-504
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-504/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-504/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.11it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.10it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-504/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-504/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-504/tokenizer_config.json
+[eval] epoch 7.0 step 504 loss 4.3545 (best 4.3545 @ ep 7.0, 0/10000 since best)
+
+
                                               
 58%|█████▊    | 505/864 [09:28<18:29,  3.09s/it]
 59%|█████▊    | 506/864 [09:28<14:31,  2.43s/it]
 59%|█████▊    | 507/864 [09:29<11:44,  1.97s/it]
 59%|█████▉    | 508/864 [09:30<09:47,  1.65s/it]
 59%|█████▉    | 509/864 [09:31<08:25,  1.43s/it]
 59%|█████▉    | 510/864 [09:32<07:28,  1.27s/it]
                                                 
{'loss': '4.136', 'grad_norm': '0.4333', 'learning_rate': '0.0012', 'epoch': '7.084'}
+
 59%|█████▉    | 510/864 [09:32<07:28,  1.27s/it]
 59%|█████▉    | 511/864 [09:33<06:48,  1.16s/it]
 59%|█████▉    | 512/864 [09:34<06:20,  1.08s/it]
 59%|█████▉    | 513/864 [09:35<06:00,  1.03s/it]
 59%|█████▉    | 514/864 [09:36<05:45,  1.01it/s]
 60%|█████▉    | 515/864 [09:37<05:35,  1.04it/s]
 60%|█████▉    | 516/864 [09:37<05:28,  1.06it/s]
 60%|█████▉    | 517/864 [09:38<05:23,  1.07it/s]
 60%|█████▉    | 518/864 [09:39<05:18,  1.09it/s]
 60%|██████    | 519/864 [09:40<05:15,  1.09it/s]
 60%|██████    | 520/864 [09:41<05:13,  1.10it/s]
                                                 
{'loss': '4.178', 'grad_norm': '0.4595', 'learning_rate': '0.0012', 'epoch': '7.223'}
+
 60%|██████    | 520/864 [09:41<05:13,  1.10it/s]
 60%|██████    | 521/864 [09:42<05:11,  1.10it/s]
 60%|██████    | 522/864 [09:43<05:09,  1.11it/s]
 61%|██████    | 523/864 [09:44<05:08,  1.11it/s]
 61%|██████    | 524/864 [09:45<05:06,  1.11it/s]
 61%|██████    | 525/864 [09:46<05:05,  1.11it/s]
 61%|██████    | 526/864 [09:46<05:04,  1.11it/s]
 61%|██████    | 527/864 [09:47<05:03,  1.11it/s]
 61%|██████    | 528/864 [09:48<05:02,  1.11it/s]
 61%|██████    | 529/864 [09:49<05:01,  1.11it/s]
 61%|██████▏   | 530/864 [09:50<05:00,  1.11it/s]
                                                 
{'loss': '4.169', 'grad_norm': '0.1839', 'learning_rate': '0.0012', 'epoch': '7.363'}
+
 61%|██████▏   | 530/864 [09:50<05:00,  1.11it/s]
 61%|██████▏   | 531/864 [09:51<04:59,  1.11it/s]
 62%|██████▏   | 532/864 [09:52<04:58,  1.11it/s]
 62%|██████▏   | 533/864 [09:53<04:57,  1.11it/s]
 62%|██████▏   | 534/864 [09:54<04:56,  1.11it/s]
 62%|██████▏   | 535/864 [09:55<04:56,  1.11it/s]
 62%|██████▏   | 536/864 [09:55<04:55,  1.11it/s]
 62%|██████▏   | 537/864 [09:56<04:54,  1.11it/s]
 62%|██████▏   | 538/864 [09:57<04:53,  1.11it/s]
 62%|██████▏   | 539/864 [09:58<04:52,  1.11it/s]
 62%|██████▎   | 540/864 [09:59<04:51,  1.11it/s]
                                                 
{'loss': '4.101', 'grad_norm': '0.1235', 'learning_rate': '0.0012', 'epoch': '7.503'}
+
 62%|██████▎   | 540/864 [09:59<04:51,  1.11it/s]
 63%|██████▎   | 541/864 [10:00<04:50,  1.11it/s]
 63%|██████▎   | 542/864 [10:01<04:49,  1.11it/s]
 63%|██████▎   | 543/864 [10:02<04:48,  1.11it/s]
 63%|██████▎   | 544/864 [10:03<04:47,  1.11it/s]
 63%|██████▎   | 545/864 [10:04<04:46,  1.11it/s]
 63%|██████▎   | 546/864 [10:04<04:45,  1.11it/s]
 63%|██████▎   | 547/864 [10:05<04:44,  1.11it/s]
 63%|██████▎   | 548/864 [10:06<04:43,  1.11it/s]
 64%|██████▎   | 549/864 [10:07<04:43,  1.11it/s]
 64%|██████▎   | 550/864 [10:08<04:42,  1.11it/s]
                                                 
{'loss': '4.062', 'grad_norm': '0.1143', 'learning_rate': '0.0012', 'epoch': '7.642'}
+
 64%|██████▎   | 550/864 [10:08<04:42,  1.11it/s]
 64%|██████▍   | 551/864 [10:09<04:41,  1.11it/s]
 64%|██████▍   | 552/864 [10:10<04:40,  1.11it/s]
 64%|██████▍   | 553/864 [10:11<04:39,  1.11it/s]
 64%|██████▍   | 554/864 [10:12<04:38,  1.11it/s]
 64%|██████▍   | 555/864 [10:12<04:37,  1.11it/s]
 64%|██████▍   | 556/864 [10:13<04:36,  1.11it/s]
 64%|██████▍   | 557/864 [10:14<04:36,  1.11it/s]
 65%|██████▍   | 558/864 [10:15<04:35,  1.11it/s]
 65%|██████▍   | 559/864 [10:16<04:34,  1.11it/s]
 65%|██████▍   | 560/864 [10:17<04:33,  1.11it/s]
                                                 
{'loss': '4.04', 'grad_norm': '0.247', 'learning_rate': '0.0012', 'epoch': '7.782'}
+
 65%|██████▍   | 560/864 [10:17<04:33,  1.11it/s]
 65%|██████▍   | 561/864 [10:18<04:32,  1.11it/s]
 65%|██████▌   | 562/864 [10:19<04:31,  1.11it/s]
 65%|██████▌   | 563/864 [10:20<04:30,  1.11it/s]
 65%|██████▌   | 564/864 [10:21<04:29,  1.11it/s]
 65%|██████▌   | 565/864 [10:21<04:28,  1.11it/s]
 66%|██████▌   | 566/864 [10:22<04:27,  1.11it/s]
 66%|██████▌   | 567/864 [10:23<04:27,  1.11it/s]
 66%|██████▌   | 568/864 [10:24<04:26,  1.11it/s]
 66%|██████▌   | 569/864 [10:25<04:25,  1.11it/s]
 66%|██████▌   | 570/864 [10:26<04:24,  1.11it/s]
                                                 
{'loss': '4.035', 'grad_norm': '0.1985', 'learning_rate': '0.0012', 'epoch': '7.921'}
+
 66%|██████▌   | 570/864 [10:26<04:24,  1.11it/s]
 66%|██████▌   | 571/864 [10:27<04:23,  1.11it/s]
 66%|██████▌   | 572/864 [10:28<04:22,  1.11it/s]
 66%|██████▋   | 573/864 [10:29<04:21,  1.11it/s]
 66%|██████▋   | 574/864 [10:30<04:20,  1.11it/s]
 67%|██████▋   | 575/864 [10:30<04:19,  1.11it/s]
 67%|██████▋   | 576/864 [10:31<03:50,  1.25it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:02, 14.90it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 11.58it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 10.82it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 10.25it/s]
+
 28%|██▊       | 11/39 [00:01<00:02, 10.12it/s]
+
 33%|███▎      | 13/39 [00:01<00:02,  9.87it/s]
+
 38%|███▊      | 15/39 [00:01<00:02,  9.87it/s]
+
 41%|████      | 16/39 [00:01<00:02,  9.82it/s]
+
 44%|████▎     | 17/39 [00:01<00:02,  9.69it/s]
+
 49%|████▊     | 19/39 [00:01<00:02,  9.76it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01,  9.73it/s]
+
 54%|█████▍    | 21/39 [00:02<00:01,  9.60it/s]
+
 59%|█████▉    | 23/39 [00:02<00:01,  9.71it/s]
+
 62%|██████▏   | 24/39 [00:02<00:01,  9.71it/s]
+
 64%|██████▍   | 25/39 [00:02<00:01,  9.69it/s]
+
 67%|██████▋   | 26/39 [00:02<00:01,  9.68it/s]
+
 69%|██████▉   | 27/39 [00:02<00:01,  9.68it/s]
+
 72%|███████▏  | 28/39 [00:02<00:01,  9.67it/s]
+
 74%|███████▍  | 29/39 [00:02<00:01,  9.67it/s]
+
 77%|███████▋  | 30/39 [00:03<00:00,  9.67it/s]
+
 79%|███████▉  | 31/39 [00:03<00:00,  9.66it/s]
+
 82%|████████▏ | 32/39 [00:03<00:00,  9.67it/s]
+
 85%|████████▍ | 33/39 [00:03<00:00,  9.66it/s]
+
 87%|████████▋ | 34/39 [00:03<00:00,  9.66it/s]
+
 90%|████████▉ | 35/39 [00:03<00:00,  9.61it/s]
+
 92%|█████████▏| 36/39 [00:03<00:00,  9.67it/s]
+
 95%|█████████▍| 37/39 [00:03<00:00,  9.68it/s]
+
 97%|█████████▋| 38/39 [00:03<00:00,  9.68it/s]
+
100%|██████████| 39/39 [00:03<00:00,  9.67it/s]
                                                 
+
                                               
{'eval_loss': '4.231', 'eval_runtime': '4.097', 'eval_samples_per_second': '595.8', 'eval_steps_per_second': '9.52', 'epoch': '8'}
+
 67%|██████▋   | 576/864 [10:35<03:50,  1.25it/s]
+
100%|██████████| 39/39 [00:04<00:00,  9.67it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-576
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-576/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-576/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.13it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.13it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-576/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-576/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-576/tokenizer_config.json
+[eval] epoch 8.0 step 576 loss 4.2307 (best 4.2307 @ ep 8.0, 0/10000 since best)
+
+
                                               
 67%|██████▋   | 577/864 [10:39<14:46,  3.09s/it]
 67%|██████▋   | 578/864 [10:40<11:35,  2.43s/it]
 67%|██████▋   | 579/864 [10:41<09:21,  1.97s/it]
 67%|██████▋   | 580/864 [10:42<07:48,  1.65s/it]
                                                 
{'loss': '3.968', 'grad_norm': '0.2255', 'learning_rate': '0.0012', 'epoch': '8.056'}
+
 67%|██████▋   | 580/864 [10:42<07:48,  1.65s/it]
 67%|██████▋   | 581/864 [10:43<06:43,  1.42s/it]
 67%|██████▋   | 582/864 [10:44<05:57,  1.27s/it]
 67%|██████▋   | 583/864 [10:45<05:25,  1.16s/it]
 68%|██████▊   | 584/864 [10:46<05:02,  1.08s/it]
 68%|██████▊   | 585/864 [10:47<04:46,  1.03s/it]
 68%|██████▊   | 586/864 [10:48<04:34,  1.01it/s]
 68%|██████▊   | 587/864 [10:48<04:26,  1.04it/s]
 68%|██████▊   | 588/864 [10:49<04:20,  1.06it/s]
 68%|██████▊   | 589/864 [10:50<04:15,  1.07it/s]
 68%|██████▊   | 590/864 [10:51<04:12,  1.09it/s]
                                                 
{'loss': '3.895', 'grad_norm': '0.1706', 'learning_rate': '0.0012', 'epoch': '8.195'}
+
 68%|██████▊   | 590/864 [10:51<04:12,  1.09it/s]
 68%|██████▊   | 591/864 [10:52<04:09,  1.09it/s]
 69%|██████▊   | 592/864 [10:53<04:07,  1.10it/s]
 69%|██████▊   | 593/864 [10:54<04:05,  1.10it/s]
 69%|██████▉   | 594/864 [10:55<04:04,  1.10it/s]
 69%|██████▉   | 595/864 [10:56<04:03,  1.11it/s]
 69%|██████▉   | 596/864 [10:57<04:01,  1.11it/s]
 69%|██████▉   | 597/864 [10:57<04:00,  1.11it/s]
 69%|██████▉   | 598/864 [10:58<03:59,  1.11it/s]
 69%|██████▉   | 599/864 [10:59<03:59,  1.11it/s]
 69%|██████▉   | 600/864 [11:00<03:58,  1.11it/s]
                                                 
{'loss': '3.884', 'grad_norm': '0.1555', 'learning_rate': '0.0012', 'epoch': '8.335'}
+
 69%|██████▉   | 600/864 [11:00<03:58,  1.11it/s]
 70%|██████▉   | 601/864 [11:01<03:57,  1.11it/s]
 70%|██████▉   | 602/864 [11:02<03:56,  1.11it/s]
 70%|██████▉   | 603/864 [11:03<03:55,  1.11it/s]
 70%|██████▉   | 604/864 [11:04<03:54,  1.11it/s]
 70%|███████   | 605/864 [11:05<03:53,  1.11it/s]
 70%|███████   | 606/864 [11:06<03:52,  1.11it/s]
 70%|███████   | 607/864 [11:06<03:51,  1.11it/s]
 70%|███████   | 608/864 [11:07<03:50,  1.11it/s]
 70%|███████   | 609/864 [11:08<03:49,  1.11it/s]
 71%|███████   | 610/864 [11:09<03:48,  1.11it/s]
                                                 
{'loss': '3.868', 'grad_norm': '0.1699', 'learning_rate': '0.0012', 'epoch': '8.475'}
+
 71%|███████   | 610/864 [11:09<03:48,  1.11it/s]
 71%|███████   | 611/864 [11:10<03:47,  1.11it/s]
 71%|███████   | 612/864 [11:11<03:46,  1.11it/s]
 71%|███████   | 613/864 [11:12<03:45,  1.11it/s]
 71%|███████   | 614/864 [11:13<03:44,  1.11it/s]
 71%|███████   | 615/864 [11:14<03:43,  1.11it/s]
 71%|███████▏  | 616/864 [11:15<03:42,  1.11it/s]
 71%|███████▏  | 617/864 [11:15<03:41,  1.11it/s]
 72%|███████▏  | 618/864 [11:16<03:41,  1.11it/s]
 72%|███████▏  | 619/864 [11:17<03:40,  1.11it/s]
 72%|███████▏  | 620/864 [11:18<03:39,  1.11it/s]
                                                 
{'loss': '3.874', 'grad_norm': '0.3587', 'learning_rate': '0.0012', 'epoch': '8.614'}
+
 72%|███████▏  | 620/864 [11:18<03:39,  1.11it/s]
 72%|███████▏  | 621/864 [11:19<03:38,  1.11it/s]
 72%|███████▏  | 622/864 [11:20<03:37,  1.11it/s]
 72%|███████▏  | 623/864 [11:21<03:36,  1.11it/s]
 72%|███████▏  | 624/864 [11:22<03:35,  1.11it/s]
 72%|███████▏  | 625/864 [11:23<03:34,  1.11it/s]
 72%|███████▏  | 626/864 [11:24<03:33,  1.11it/s]
 73%|███████▎  | 627/864 [11:24<03:32,  1.11it/s]
 73%|███████▎  | 628/864 [11:25<03:31,  1.11it/s]
 73%|███████▎  | 629/864 [11:26<03:30,  1.11it/s]
 73%|███████▎  | 630/864 [11:27<03:30,  1.11it/s]
                                                 
{'loss': '3.891', 'grad_norm': '0.2342', 'learning_rate': '0.0012', 'epoch': '8.754'}
+
 73%|███████▎  | 630/864 [11:27<03:30,  1.11it/s]
 73%|███████▎  | 631/864 [11:28<03:29,  1.11it/s]
 73%|███████▎  | 632/864 [11:29<03:28,  1.11it/s]
 73%|███████▎  | 633/864 [11:30<03:27,  1.11it/s]
 73%|███████▎  | 634/864 [11:31<03:26,  1.11it/s]
 73%|███████▎  | 635/864 [11:32<03:25,  1.11it/s]
 74%|███████▎  | 636/864 [11:33<03:25,  1.11it/s]
 74%|███████▎  | 637/864 [11:33<03:24,  1.11it/s]
 74%|███████▍  | 638/864 [11:34<03:23,  1.11it/s]
 74%|███████▍  | 639/864 [11:35<03:22,  1.11it/s]
 74%|███████▍  | 640/864 [11:36<03:21,  1.11it/s]
                                                 
{'loss': '3.892', 'grad_norm': '0.1874', 'learning_rate': '0.0012', 'epoch': '8.894'}
+
 74%|███████▍  | 640/864 [11:36<03:21,  1.11it/s]
 74%|███████▍  | 641/864 [11:37<03:20,  1.11it/s]
 74%|███████▍  | 642/864 [11:38<03:19,  1.11it/s]
 74%|███████▍  | 643/864 [11:39<03:18,  1.11it/s]
 75%|███████▍  | 644/864 [11:40<03:17,  1.11it/s]
 75%|███████▍  | 645/864 [11:41<03:16,  1.11it/s]
 75%|███████▍  | 646/864 [11:42<03:16,  1.11it/s]
 75%|███████▍  | 647/864 [11:42<03:15,  1.11it/s]
 75%|███████▌  | 648/864 [11:43<02:52,  1.25it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:01, 19.64it/s]
+
 10%|█         | 4/39 [00:00<00:02, 12.19it/s]
+
 15%|█▌        | 6/39 [00:00<00:03, 10.88it/s]
+
 21%|██        | 8/39 [00:00<00:02, 10.35it/s]
+
 26%|██▌       | 10/39 [00:00<00:02, 10.09it/s]
+
 31%|███       | 12/39 [00:01<00:02,  9.93it/s]
+
 36%|███▌      | 14/39 [00:01<00:02,  9.84it/s]
+
 41%|████      | 16/39 [00:01<00:02,  9.77it/s]
+
 44%|████▎     | 17/39 [00:01<00:02,  9.74it/s]
+
 46%|████▌     | 18/39 [00:01<00:02,  9.73it/s]
+
 49%|████▊     | 19/39 [00:01<00:02,  9.71it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01,  9.69it/s]
+
 54%|█████▍    | 21/39 [00:02<00:01,  9.67it/s]
+
 56%|█████▋    | 22/39 [00:02<00:01,  9.67it/s]
+
 59%|█████▉    | 23/39 [00:02<00:01,  9.68it/s]
+
 62%|██████▏   | 24/39 [00:02<00:01,  9.67it/s]
+
 64%|██████▍   | 25/39 [00:02<00:01,  9.66it/s]
+
 67%|██████▋   | 26/39 [00:02<00:01,  9.66it/s]
+
 69%|██████▉   | 27/39 [00:02<00:01,  9.66it/s]
+
 72%|███████▏  | 28/39 [00:02<00:01,  9.66it/s]
+
 74%|███████▍  | 29/39 [00:02<00:01,  9.66it/s]
+
 77%|███████▋  | 30/39 [00:03<00:00,  9.66it/s]
+
 79%|███████▉  | 31/39 [00:03<00:00,  9.66it/s]
+
 82%|████████▏ | 32/39 [00:03<00:00,  9.65it/s]
+
 85%|████████▍ | 33/39 [00:03<00:00,  9.65it/s]
+
 87%|████████▋ | 34/39 [00:03<00:00,  9.65it/s]
+
 90%|████████▉ | 35/39 [00:03<00:00,  9.65it/s]
+
 92%|█████████▏| 36/39 [00:03<00:00,  9.65it/s]
+
 95%|█████████▍| 37/39 [00:03<00:00,  9.65it/s]
+
 97%|█████████▋| 38/39 [00:03<00:00,  9.65it/s]
+
100%|██████████| 39/39 [00:03<00:00,  9.65it/s]
                                                 
+
                                               
{'eval_loss': '4.122', 'eval_runtime': '4.105', 'eval_samples_per_second': '594.6', 'eval_steps_per_second': '9.5', 'epoch': '9'}
+
 75%|███████▌  | 648/864 [11:47<02:52,  1.25it/s]
+
100%|██████████| 39/39 [00:04<00:00,  9.65it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-648
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-648/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-648/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.12it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.12it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-648/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-648/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-648/tokenizer_config.json
+[eval] epoch 9.0 step 648 loss 4.1217 (best 4.1217 @ ep 9.0, 0/10000 since best)
+
+
                                               
 75%|███████▌  | 649/864 [11:51<11:05,  3.09s/it]
 75%|███████▌  | 650/864 [11:52<08:41,  2.44s/it]
                                                 
{'loss': '3.833', 'grad_norm': '0.1714', 'learning_rate': '0.0012', 'epoch': '9.028'}
+
 75%|███████▌  | 650/864 [11:52<08:41,  2.44s/it]
 75%|███████▌  | 651/864 [11:53<07:00,  1.97s/it]
 75%|███████▌  | 652/864 [11:54<05:50,  1.65s/it]
 76%|███████▌  | 653/864 [11:55<05:00,  1.43s/it]
 76%|███████▌  | 654/864 [11:56<04:26,  1.27s/it]
 76%|███████▌  | 655/864 [11:57<04:01,  1.16s/it]
 76%|███████▌  | 656/864 [11:58<03:44,  1.08s/it]
 76%|███████▌  | 657/864 [11:59<03:32,  1.03s/it]
 76%|███████▌  | 658/864 [12:00<03:23,  1.01it/s]
 76%|███████▋  | 659/864 [12:00<03:16,  1.04it/s]
 76%|███████▋  | 660/864 [12:01<03:12,  1.06it/s]
                                                 
{'loss': '3.731', 'grad_norm': '0.4772', 'learning_rate': '0.0012', 'epoch': '9.168'}
+
 76%|███████▋  | 660/864 [12:01<03:12,  1.06it/s]
 77%|███████▋  | 661/864 [12:02<03:08,  1.08it/s]
 77%|███████▋  | 662/864 [12:03<03:05,  1.09it/s]
 77%|███████▋  | 663/864 [12:04<03:03,  1.09it/s]
 77%|███████▋  | 664/864 [12:05<03:01,  1.10it/s]
 77%|███████▋  | 665/864 [12:06<03:00,  1.10it/s]
 77%|███████▋  | 666/864 [12:07<02:59,  1.11it/s]
 77%|███████▋  | 667/864 [12:08<02:57,  1.11it/s]
 77%|███████▋  | 668/864 [12:09<02:56,  1.11it/s]
 77%|███████▋  | 669/864 [12:09<02:55,  1.11it/s]
 78%|███████▊  | 670/864 [12:10<02:54,  1.11it/s]
                                                 
{'loss': '3.79', 'grad_norm': '0.3208', 'learning_rate': '0.0012', 'epoch': '9.307'}
+
 78%|███████▊  | 670/864 [12:10<02:54,  1.11it/s]
 78%|███████▊  | 671/864 [12:11<02:53,  1.11it/s]
 78%|███████▊  | 672/864 [12:12<02:52,  1.11it/s]
 78%|███████▊  | 673/864 [12:13<02:51,  1.11it/s]
 78%|███████▊  | 674/864 [12:14<02:50,  1.11it/s]
 78%|███████▊  | 675/864 [12:15<02:49,  1.11it/s]
 78%|███████▊  | 676/864 [12:16<02:49,  1.11it/s]
 78%|███████▊  | 677/864 [12:17<02:48,  1.11it/s]
 78%|███████▊  | 678/864 [12:17<02:47,  1.11it/s]
 79%|███████▊  | 679/864 [12:18<02:46,  1.11it/s]
 79%|███████▊  | 680/864 [12:19<02:45,  1.11it/s]
                                                 
{'loss': '3.795', 'grad_norm': '0.2693', 'learning_rate': '0.0012', 'epoch': '9.447'}
+
 79%|███████▊  | 680/864 [12:19<02:45,  1.11it/s]
 79%|███████▉  | 681/864 [12:20<02:44,  1.11it/s]
 79%|███████▉  | 682/864 [12:21<02:43,  1.11it/s]
 79%|███████▉  | 683/864 [12:22<02:42,  1.11it/s]
 79%|███████▉  | 684/864 [12:23<02:41,  1.11it/s]
 79%|███████▉  | 685/864 [12:24<02:40,  1.11it/s]
 79%|███████▉  | 686/864 [12:25<02:39,  1.11it/s]
 80%|███████▉  | 687/864 [12:26<02:39,  1.11it/s]
 80%|███████▉  | 688/864 [12:26<02:38,  1.11it/s]
 80%|███████▉  | 689/864 [12:27<02:37,  1.11it/s]
 80%|███████▉  | 690/864 [12:28<02:36,  1.11it/s]
                                                 
{'loss': '3.752', 'grad_norm': '0.1616', 'learning_rate': '0.0012', 'epoch': '9.586'}
+
 80%|███████▉  | 690/864 [12:28<02:36,  1.11it/s]
 80%|███████▉  | 691/864 [12:29<02:35,  1.11it/s]
 80%|████████  | 692/864 [12:30<02:34,  1.11it/s]
 80%|████████  | 693/864 [12:31<02:33,  1.11it/s]
 80%|████████  | 694/864 [12:32<02:32,  1.11it/s]
 80%|████████  | 695/864 [12:33<02:31,  1.11it/s]
 81%|████████  | 696/864 [12:34<02:31,  1.11it/s]
 81%|████████  | 697/864 [12:35<02:30,  1.11it/s]
 81%|████████  | 698/864 [12:35<02:29,  1.11it/s]
 81%|████████  | 699/864 [12:36<02:28,  1.11it/s]
 81%|████████  | 700/864 [12:37<02:27,  1.11it/s]
                                                 
{'loss': '3.742', 'grad_norm': '0.1375', 'learning_rate': '0.0012', 'epoch': '9.726'}
+
 81%|████████  | 700/864 [12:37<02:27,  1.11it/s]
 81%|████████  | 701/864 [12:38<02:26,  1.11it/s]
 81%|████████▏ | 702/864 [12:39<02:25,  1.11it/s]
 81%|████████▏ | 703/864 [12:40<02:24,  1.11it/s]
 81%|████████▏ | 704/864 [12:41<02:23,  1.11it/s]
 82%|████████▏ | 705/864 [12:42<02:22,  1.11it/s]
 82%|████████▏ | 706/864 [12:43<02:22,  1.11it/s]
 82%|████████▏ | 707/864 [12:44<02:21,  1.11it/s]
 82%|████████▏ | 708/864 [12:44<02:20,  1.11it/s]
 82%|████████▏ | 709/864 [12:45<02:19,  1.11it/s]
 82%|████████▏ | 710/864 [12:46<02:18,  1.11it/s]
                                                 
{'loss': '3.721', 'grad_norm': '0.1259', 'learning_rate': '0.0012', 'epoch': '9.866'}
+
 82%|████████▏ | 710/864 [12:46<02:18,  1.11it/s]
 82%|████████▏ | 711/864 [12:47<02:17,  1.11it/s]
 82%|████████▏ | 712/864 [12:48<02:16,  1.11it/s]
 83%|████████▎ | 713/864 [12:49<02:15,  1.11it/s]
 83%|████████▎ | 714/864 [12:50<02:14,  1.11it/s]
 83%|████████▎ | 715/864 [12:51<02:14,  1.11it/s]
 83%|████████▎ | 716/864 [12:52<02:13,  1.11it/s]
 83%|████████▎ | 717/864 [12:53<02:12,  1.11it/s]
 83%|████████▎ | 718/864 [12:53<02:11,  1.11it/s]
 83%|████████▎ | 719/864 [12:54<02:10,  1.11it/s]
 83%|████████▎ | 720/864 [12:55<01:55,  1.25it/s]
                                                 
{'loss': '3.708', 'grad_norm': '0.1329', 'learning_rate': '0.0012', 'epoch': '10'}
+
 83%|████████▎ | 720/864 [12:55<01:55,  1.25it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:02, 14.74it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 11.70it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 10.75it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 10.31it/s]
+
 28%|██▊       | 11/39 [00:01<00:02, 10.06it/s]
+
 33%|███▎      | 13/39 [00:01<00:02,  9.92it/s]
+
 36%|███▌      | 14/39 [00:01<00:02,  9.86it/s]
+
 38%|███▊      | 15/39 [00:01<00:02,  9.81it/s]
+
 41%|████      | 16/39 [00:01<00:02,  9.65it/s]
+
 46%|████▌     | 18/39 [00:01<00:02,  9.74it/s]
+
 49%|████▊     | 19/39 [00:01<00:02,  9.72it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01,  9.59it/s]
+
 56%|█████▋    | 22/39 [00:02<00:01,  9.70it/s]
+
 59%|█████▉    | 23/39 [00:02<00:01,  9.69it/s]
+
 62%|██████▏   | 24/39 [00:02<00:01,  9.68it/s]
+
 64%|██████▍   | 25/39 [00:02<00:01,  9.66it/s]
+
 67%|██████▋   | 26/39 [00:02<00:01,  9.67it/s]
+
 69%|██████▉   | 27/39 [00:02<00:01,  9.67it/s]
+
 72%|███████▏  | 28/39 [00:02<00:01,  9.66it/s]
+
 74%|███████▍  | 29/39 [00:02<00:01,  9.66it/s]
+
 77%|███████▋  | 30/39 [00:03<00:00,  9.67it/s]
+
 79%|███████▉  | 31/39 [00:03<00:00,  9.66it/s]
+
 82%|████████▏ | 32/39 [00:03<00:00,  9.66it/s]
+
 85%|████████▍ | 33/39 [00:03<00:00,  9.66it/s]
+
 87%|████████▋ | 34/39 [00:03<00:00,  9.66it/s]
+
 90%|████████▉ | 35/39 [00:03<00:00,  9.66it/s]
+
 92%|█████████▏| 36/39 [00:03<00:00,  9.66it/s]
+
 95%|█████████▍| 37/39 [00:03<00:00,  9.66it/s]
+
 97%|█████████▋| 38/39 [00:03<00:00,  9.66it/s]
+
100%|██████████| 39/39 [00:03<00:00,  9.66it/s]
                                                 
+
                                               
{'eval_loss': '4.057', 'eval_runtime': '4.091', 'eval_samples_per_second': '596.7', 'eval_steps_per_second': '9.534', 'epoch': '10'}
+
 83%|████████▎ | 720/864 [12:59<01:55,  1.25it/s]
+
100%|██████████| 39/39 [00:04<00:00,  9.66it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-720
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-720/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-720/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.13it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.13it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-720/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-720/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-720/tokenizer_config.json
+[eval] epoch 10.0 step 720 loss 4.0569 (best 4.0569 @ ep 10.0, 0/10000 since best)
+
+
                                               
 83%|████████▎ | 721/864 [13:03<07:21,  3.09s/it]
 84%|████████▎ | 722/864 [13:04<05:45,  2.43s/it]
 84%|████████▎ | 723/864 [13:05<04:38,  1.97s/it]
 84%|████████▍ | 724/864 [13:06<03:50,  1.65s/it]
 84%|████████▍ | 725/864 [13:07<03:17,  1.42s/it]
 84%|████████▍ | 726/864 [13:08<02:54,  1.27s/it]
 84%|████████▍ | 727/864 [13:09<02:38,  1.16s/it]
 84%|████████▍ | 728/864 [13:10<02:26,  1.08s/it]
 84%|████████▍ | 729/864 [13:11<02:18,  1.02s/it]
 84%|████████▍ | 730/864 [13:11<02:12,  1.01it/s]
                                                 
{'loss': '3.589', 'grad_norm': '0.5313', 'learning_rate': '0.0012', 'epoch': '10.14'}
+
 84%|████████▍ | 730/864 [13:11<02:12,  1.01it/s]
 85%|████████▍ | 731/864 [13:12<02:07,  1.04it/s]
 85%|████████▍ | 732/864 [13:13<02:04,  1.06it/s]
 85%|████████▍ | 733/864 [13:14<02:01,  1.08it/s]
 85%|████████▍ | 734/864 [13:15<01:59,  1.09it/s]
 85%|████████▌ | 735/864 [13:16<01:57,  1.09it/s]
 85%|████████▌ | 736/864 [13:17<01:56,  1.10it/s]
 85%|████████▌ | 737/864 [13:18<01:55,  1.10it/s]
 85%|████████▌ | 738/864 [13:19<01:53,  1.11it/s]
 86%|████████▌ | 739/864 [13:20<01:52,  1.11it/s]
 86%|████████▌ | 740/864 [13:20<01:51,  1.11it/s]
                                                 
{'loss': '3.693', 'grad_norm': '0.3751', 'learning_rate': '0.0012', 'epoch': '10.28'}
+
 86%|████████▌ | 740/864 [13:20<01:51,  1.11it/s]
 86%|████████▌ | 741/864 [13:21<01:50,  1.11it/s]
 86%|████████▌ | 742/864 [13:22<01:49,  1.11it/s]
 86%|████████▌ | 743/864 [13:23<01:48,  1.11it/s]
 86%|████████▌ | 744/864 [13:24<01:47,  1.11it/s]
 86%|████████▌ | 745/864 [13:25<01:47,  1.11it/s]
 86%|████████▋ | 746/864 [13:26<01:46,  1.11it/s]
 86%|████████▋ | 747/864 [13:27<01:45,  1.11it/s]
 87%|████████▋ | 748/864 [13:28<01:44,  1.11it/s]
 87%|████████▋ | 749/864 [13:29<01:43,  1.11it/s]
 87%|████████▋ | 750/864 [13:29<01:42,  1.11it/s]
                                                 
{'loss': '3.657', 'grad_norm': '0.1845', 'learning_rate': '0.0012', 'epoch': '10.42'}
+
 87%|████████▋ | 750/864 [13:29<01:42,  1.11it/s]
 87%|████████▋ | 751/864 [13:30<01:41,  1.11it/s]
 87%|████████▋ | 752/864 [13:31<01:40,  1.11it/s]
 87%|████████▋ | 753/864 [13:32<01:39,  1.11it/s]
 87%|████████▋ | 754/864 [13:33<01:38,  1.11it/s]
 87%|████████▋ | 755/864 [13:34<01:37,  1.11it/s]
 88%|████████▊ | 756/864 [13:35<01:37,  1.11it/s]
 88%|████████▊ | 757/864 [13:36<01:36,  1.11it/s]
 88%|████████▊ | 758/864 [13:37<01:35,  1.11it/s]
 88%|████████▊ | 759/864 [13:38<01:34,  1.11it/s]
 88%|████████▊ | 760/864 [13:38<01:33,  1.11it/s]
                                                 
{'loss': '3.639', 'grad_norm': '0.132', 'learning_rate': '0.0012', 'epoch': '10.56'}
+
 88%|████████▊ | 760/864 [13:38<01:33,  1.11it/s]
 88%|████████▊ | 761/864 [13:39<01:32,  1.11it/s]
 88%|████████▊ | 762/864 [13:40<01:31,  1.11it/s]
 88%|████████▊ | 763/864 [13:41<01:30,  1.11it/s]
 88%|████████▊ | 764/864 [13:42<01:29,  1.11it/s]
 89%|████████▊ | 765/864 [13:43<01:29,  1.11it/s]
 89%|████████▊ | 766/864 [13:44<01:28,  1.11it/s]
 89%|████████▉ | 767/864 [13:45<01:27,  1.11it/s]
 89%|████████▉ | 768/864 [13:46<01:26,  1.11it/s]
 89%|████████▉ | 769/864 [13:47<01:25,  1.11it/s]
 89%|████████▉ | 770/864 [13:47<01:24,  1.11it/s]
                                                 
{'loss': '3.619', 'grad_norm': '0.1276', 'learning_rate': '0.0012', 'epoch': '10.7'}
+
 89%|████████▉ | 770/864 [13:47<01:24,  1.11it/s]
 89%|████████▉ | 771/864 [13:48<01:23,  1.11it/s]
 89%|████████▉ | 772/864 [13:49<01:22,  1.11it/s]
 89%|████████▉ | 773/864 [13:50<01:21,  1.11it/s]
 90%|████████▉ | 774/864 [13:51<01:20,  1.11it/s]
 90%|████████▉ | 775/864 [13:52<01:19,  1.11it/s]
 90%|████████▉ | 776/864 [13:53<01:19,  1.11it/s]
 90%|████████▉ | 777/864 [13:54<01:28,  1.02s/it]
 90%|█████████ | 778/864 [13:55<01:24,  1.02it/s]
 90%|█████████ | 779/864 [13:56<01:21,  1.05it/s]
 90%|█████████ | 780/864 [13:57<01:18,  1.06it/s]
                                                 
{'loss': '3.618', 'grad_norm': '0.1232', 'learning_rate': '0.0012', 'epoch': '10.84'}
+
 90%|█████████ | 780/864 [13:57<01:18,  1.06it/s]
 90%|█████████ | 781/864 [13:58<01:17,  1.07it/s]
 91%|█████████ | 782/864 [13:59<01:15,  1.08it/s]
 91%|█████████ | 783/864 [14:00<01:14,  1.09it/s]
 91%|█████████ | 784/864 [14:00<01:12,  1.10it/s]
 91%|█████████ | 785/864 [14:01<01:11,  1.10it/s]
 91%|█████████ | 786/864 [14:02<01:10,  1.10it/s]
 91%|█████████ | 787/864 [14:03<01:09,  1.11it/s]
 91%|█████████ | 788/864 [14:04<01:08,  1.11it/s]
 91%|█████████▏| 789/864 [14:05<01:07,  1.11it/s]
 91%|█████████▏| 790/864 [14:06<01:06,  1.11it/s]
                                                 
{'loss': '3.608', 'grad_norm': '0.1296', 'learning_rate': '0.0012', 'epoch': '10.98'}
+
 91%|█████████▏| 790/864 [14:06<01:06,  1.11it/s]
 92%|█████████▏| 791/864 [14:07<01:05,  1.11it/s]
 92%|█████████▏| 792/864 [14:07<00:57,  1.25it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:01, 19.95it/s]
+
 10%|█         | 4/39 [00:00<00:02, 12.26it/s]
+
 15%|█▌        | 6/39 [00:00<00:03, 10.90it/s]
+
 21%|██        | 8/39 [00:00<00:02, 10.38it/s]
+
 26%|██▌       | 10/39 [00:00<00:02, 10.09it/s]
+
 31%|███       | 12/39 [00:01<00:02,  9.92it/s]
+
 36%|███▌      | 14/39 [00:01<00:02,  9.84it/s]
+
 41%|████      | 16/39 [00:01<00:02,  9.77it/s]
+
 44%|████▎     | 17/39 [00:01<00:02,  9.67it/s]
+
 49%|████▊     | 19/39 [00:01<00:02,  9.74it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01,  9.72it/s]
+
 54%|█████▍    | 21/39 [00:02<00:01,  9.60it/s]
+
 59%|█████▉    | 23/39 [00:02<00:01,  9.71it/s]
+
 62%|██████▏   | 24/39 [00:02<00:01,  9.70it/s]
+
 64%|██████▍   | 25/39 [00:02<00:01,  9.69it/s]
+
 67%|██████▋   | 26/39 [00:02<00:01,  9.68it/s]
+
 69%|██████▉   | 27/39 [00:02<00:01,  9.67it/s]
+
 72%|███████▏  | 28/39 [00:02<00:01,  9.66it/s]
+
 74%|███████▍  | 29/39 [00:02<00:01,  9.66it/s]
+
 77%|███████▋  | 30/39 [00:03<00:00,  9.65it/s]
+
 79%|███████▉  | 31/39 [00:03<00:00,  9.66it/s]
+
 82%|████████▏ | 32/39 [00:03<00:00,  9.66it/s]
+
 85%|████████▍ | 33/39 [00:03<00:00,  9.66it/s]
+
 87%|████████▋ | 34/39 [00:03<00:00,  9.66it/s]
+
 90%|████████▉ | 35/39 [00:03<00:00,  9.66it/s]
+
 92%|█████████▏| 36/39 [00:03<00:00,  9.66it/s]
+
 95%|█████████▍| 37/39 [00:03<00:00,  9.66it/s]
+
 97%|█████████▋| 38/39 [00:03<00:00,  9.65it/s]
+
100%|██████████| 39/39 [00:03<00:00,  9.65it/s]
                                                 
+
                                               
{'eval_loss': '4.024', 'eval_runtime': '4.095', 'eval_samples_per_second': '596.1', 'eval_steps_per_second': '9.523', 'epoch': '11'}
+
 92%|█████████▏| 792/864 [14:11<00:57,  1.25it/s]
+
100%|██████████| 39/39 [00:04<00:00,  9.65it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-792
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-792/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-792/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:01<00:00,  1.02s/it]
Writing model shards: 100%|██████████| 1/1 [00:01<00:00,  1.02s/it]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-792/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-792/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-792/tokenizer_config.json
+[eval] epoch 11.0 step 792 loss 4.0241 (best 4.0241 @ ep 11.0, 0/10000 since best)
+
+
                                               
 92%|█████████▏| 793/864 [14:16<03:42,  3.13s/it]
 92%|█████████▏| 794/864 [14:17<02:52,  2.46s/it]
 92%|█████████▏| 795/864 [14:18<02:17,  1.99s/it]
 92%|█████████▏| 796/864 [14:19<01:53,  1.66s/it]
 92%|█████████▏| 797/864 [14:19<01:36,  1.44s/it]
 92%|█████████▏| 798/864 [14:20<01:24,  1.27s/it]
 92%|█████████▏| 799/864 [14:21<01:15,  1.16s/it]
 93%|█████████▎| 800/864 [14:22<01:09,  1.08s/it]
                                                 
{'loss': '3.484', 'grad_norm': '0.2861', 'learning_rate': '0.0012', 'epoch': '11.11'}
+
 93%|█████████▎| 800/864 [14:22<01:09,  1.08s/it]
 93%|█████████▎| 801/864 [14:23<01:04,  1.03s/it]
 93%|█████████▎| 802/864 [14:24<01:01,  1.01it/s]
 93%|█████████▎| 803/864 [14:25<00:58,  1.04it/s]
 93%|█████████▎| 804/864 [14:26<00:56,  1.06it/s]
 93%|█████████▎| 805/864 [14:27<00:54,  1.07it/s]
 93%|█████████▎| 806/864 [14:28<00:53,  1.08it/s]
 93%|█████████▎| 807/864 [14:28<00:52,  1.09it/s]
 94%|█████████▎| 808/864 [14:29<00:51,  1.10it/s]
 94%|█████████▎| 809/864 [14:30<00:49,  1.10it/s]
 94%|█████████▍| 810/864 [14:31<00:48,  1.10it/s]
                                                 
{'loss': '3.458', 'grad_norm': '0.3106', 'learning_rate': '0.0012', 'epoch': '11.25'}
+
 94%|█████████▍| 810/864 [14:31<00:48,  1.10it/s]
 94%|█████████▍| 811/864 [14:32<00:47,  1.11it/s]
 94%|█████████▍| 812/864 [14:33<00:46,  1.11it/s]
 94%|█████████▍| 813/864 [14:34<00:46,  1.11it/s]
 94%|█████████▍| 814/864 [14:35<00:45,  1.11it/s]
 94%|█████████▍| 815/864 [14:36<00:44,  1.11it/s]
 94%|█████████▍| 816/864 [14:37<00:43,  1.11it/s]
 95%|█████████▍| 817/864 [14:37<00:42,  1.11it/s]
 95%|█████████▍| 818/864 [14:38<00:41,  1.11it/s]
 95%|█████████▍| 819/864 [14:39<00:40,  1.11it/s]
 95%|█████████▍| 820/864 [14:40<00:39,  1.11it/s]
                                                 
{'loss': '3.488', 'grad_norm': '0.2634', 'learning_rate': '0.0012', 'epoch': '11.39'}
+
 95%|█████████▍| 820/864 [14:40<00:39,  1.11it/s]
 95%|█████████▌| 821/864 [14:41<00:38,  1.11it/s]
 95%|█████████▌| 822/864 [14:42<00:37,  1.11it/s]
 95%|█████████▌| 823/864 [14:43<00:36,  1.11it/s]
 95%|█████████▌| 824/864 [14:44<00:36,  1.11it/s]
 95%|█████████▌| 825/864 [14:45<00:35,  1.11it/s]
 96%|█████████▌| 826/864 [14:46<00:34,  1.11it/s]
 96%|█████████▌| 827/864 [14:46<00:33,  1.11it/s]
 96%|█████████▌| 828/864 [14:47<00:32,  1.11it/s]
 96%|█████████▌| 829/864 [14:48<00:31,  1.11it/s]
 96%|█████████▌| 830/864 [14:49<00:30,  1.11it/s]
                                                 
{'loss': '3.499', 'grad_norm': '0.1477', 'learning_rate': '0.0012', 'epoch': '11.53'}
+
 96%|█████████▌| 830/864 [14:49<00:30,  1.11it/s]
 96%|█████████▌| 831/864 [14:50<00:29,  1.11it/s]
 96%|█████████▋| 832/864 [14:51<00:28,  1.11it/s]
 96%|█████████▋| 833/864 [14:52<00:27,  1.11it/s]
 97%|█████████▋| 834/864 [14:53<00:27,  1.11it/s]
 97%|█████████▋| 835/864 [14:54<00:26,  1.11it/s]
 97%|█████████▋| 836/864 [14:55<00:25,  1.11it/s]
 97%|█████████▋| 837/864 [14:55<00:24,  1.11it/s]
 97%|█████████▋| 838/864 [14:56<00:23,  1.11it/s]
 97%|█████████▋| 839/864 [14:57<00:22,  1.11it/s]
 97%|█████████▋| 840/864 [14:58<00:21,  1.11it/s]
                                                 
{'loss': '3.497', 'grad_norm': '0.1387', 'learning_rate': '0.0012', 'epoch': '11.67'}
+
 97%|█████████▋| 840/864 [14:58<00:21,  1.11it/s]
 97%|█████████▋| 841/864 [14:59<00:20,  1.11it/s]
 97%|█████████▋| 842/864 [15:00<00:19,  1.11it/s]
 98%|█████████▊| 843/864 [15:01<00:18,  1.11it/s]
 98%|█████████▊| 844/864 [15:02<00:18,  1.11it/s]
 98%|█████████▊| 845/864 [15:03<00:17,  1.11it/s]
 98%|█████████▊| 846/864 [15:04<00:16,  1.11it/s]
 98%|█████████▊| 847/864 [15:04<00:15,  1.11it/s]
 98%|█████████▊| 848/864 [15:05<00:14,  1.11it/s]
 98%|█████████▊| 849/864 [15:06<00:13,  1.11it/s]
 98%|█████████▊| 850/864 [15:07<00:12,  1.11it/s]
                                                 
{'loss': '3.503', 'grad_norm': '0.1274', 'learning_rate': '0.0012', 'epoch': '11.81'}
+
 98%|█████████▊| 850/864 [15:07<00:12,  1.11it/s]
 98%|█████████▊| 851/864 [15:08<00:11,  1.11it/s]
 99%|█████████▊| 852/864 [15:09<00:10,  1.11it/s]
 99%|█████████▊| 853/864 [15:10<00:09,  1.11it/s]
 99%|█████████▉| 854/864 [15:11<00:08,  1.11it/s]
 99%|█████████▉| 855/864 [15:12<00:08,  1.11it/s]
 99%|█████████▉| 856/864 [15:13<00:07,  1.11it/s]
 99%|█████████▉| 857/864 [15:13<00:06,  1.11it/s]
 99%|█████████▉| 858/864 [15:14<00:05,  1.11it/s]
 99%|█████████▉| 859/864 [15:15<00:04,  1.11it/s]
100%|█████████▉| 860/864 [15:16<00:03,  1.11it/s]
                                                 
{'loss': '3.514', 'grad_norm': '0.1485', 'learning_rate': '0.0012', 'epoch': '11.95'}
+
100%|█████████▉| 860/864 [15:16<00:03,  1.11it/s]
100%|█████████▉| 861/864 [15:17<00:02,  1.11it/s]
100%|█████████▉| 862/864 [15:18<00:01,  1.11it/s]
100%|█████████▉| 863/864 [15:19<00:00,  1.11it/s]
100%|██████████| 864/864 [15:19<00:00,  1.25it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:02, 14.88it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 11.77it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 10.80it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 10.34it/s]
+
 28%|██▊       | 11/39 [00:01<00:02, 10.09it/s]
+
 33%|███▎      | 13/39 [00:01<00:02,  9.94it/s]
+
 36%|███▌      | 14/39 [00:01<00:02,  9.88it/s]
+
 38%|███▊      | 15/39 [00:01<00:02,  9.83it/s]
+
 41%|████      | 16/39 [00:01<00:02,  9.67it/s]
+
 46%|████▌     | 18/39 [00:01<00:02,  9.76it/s]
+
 49%|████▊     | 19/39 [00:01<00:02,  9.74it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01,  9.72it/s]
+
 54%|█████▍    | 21/39 [00:02<00:01,  9.70it/s]
+
 56%|█████▋    | 22/39 [00:02<00:01,  9.69it/s]
+
 59%|█████▉    | 23/39 [00:02<00:01,  9.69it/s]
+
 62%|██████▏   | 24/39 [00:02<00:01,  9.69it/s]
+
 64%|██████▍   | 25/39 [00:02<00:01,  9.68it/s]
+
 67%|██████▋   | 26/39 [00:02<00:01,  9.67it/s]
+
 69%|██████▉   | 27/39 [00:02<00:01,  9.67it/s]
+
 72%|███████▏  | 28/39 [00:02<00:01,  9.67it/s]
+
 74%|███████▍  | 29/39 [00:02<00:01,  9.67it/s]
+
 77%|███████▋  | 30/39 [00:02<00:00,  9.66it/s]
+
 79%|███████▉  | 31/39 [00:03<00:00,  9.66it/s]
+
 82%|████████▏ | 32/39 [00:03<00:00,  9.66it/s]
+
 85%|████████▍ | 33/39 [00:03<00:00,  9.66it/s]
+
 87%|████████▋ | 34/39 [00:03<00:00,  9.66it/s]
+
 90%|████████▉ | 35/39 [00:03<00:00,  9.66it/s]
+
 92%|█████████▏| 36/39 [00:03<00:00,  9.66it/s]
+
 95%|█████████▍| 37/39 [00:03<00:00,  9.66it/s]
+
 97%|█████████▋| 38/39 [00:03<00:00,  9.66it/s]
+
100%|██████████| 39/39 [00:03<00:00,  9.66it/s]
                                                 
+
                                               
{'eval_loss': '4.01', 'eval_runtime': '4.111', 'eval_samples_per_second': '593.8', 'eval_steps_per_second': '9.488', 'epoch': '12'}
+
100%|██████████| 864/864 [15:24<00:00,  1.25it/s]
+
100%|██████████| 39/39 [00:04<00:00,  9.66it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-864
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-864/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-864/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.14it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.14it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-864/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-864/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S3_lr0.0012/checkpoint-864/tokenizer_config.json
+[eval] epoch 12.0 step 864 loss 4.0103 (best 4.0103 @ ep 12.0, 0/10000 since best)
+
+
                                               [transformers] 
+
+Training completed. Do not forget to share your model on huggingface.co/models =)
+
+
+
                                                 
{'train_runtime': '928.6', 'train_samples_per_second': '473.1', 'train_steps_per_second': '0.93', 'train_loss': '4.909', 'epoch': '12'}
+
100%|██████████| 864/864 [15:27<00:00,  1.25it/s]
100%|██████████| 864/864 [15:27<00:00,  1.07s/it]
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+wandb: 
+wandb: 🚀 View run t0_S3_lr0.0012 at: https://wandb.ai/slowrun-wankers/data-aug-pretraining/runs/hver985o
+wandb: Find logs at: wandb/run-20260817_200638-hver985o/logs
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+[rank0]:[W817 20:22:53.967428894 ProcessGroupNCCL.cpp:1575] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator())
diff --git a/runs/t0/t0_S4_lr0.0006/checkpoint-864/chat_template.jinja b/runs/t0/t0_S4_lr0.0006/checkpoint-864/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..699ff8df401fe4788525e9c1f9b86a99eadd6230
--- /dev/null
+++ b/runs/t0/t0_S4_lr0.0006/checkpoint-864/chat_template.jinja
@@ -0,0 +1,85 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0].role == 'system' %}
+        {{- messages[0].content + '\n\n' }}
+    {%- endif %}
+    {{- "# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0].content + '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
+{%- for message in messages[::-1] %}
+    {%- set index = (messages|length - 1) - loop.index0 %}
+    {%- if ns.multi_step_tool and message.role == "user" and not(message.content.startswith('<tool_response>') and message.content.endswith('</tool_response>')) %}
+        {%- set ns.multi_step_tool = false %}
+        {%- set ns.last_query_index = index %}
+    {%- endif %}
+{%- endfor %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {%- set content = message.content %}
+        {%- set reasoning_content = '' %}
+        {%- if message.reasoning_content is defined and message.reasoning_content is not none %}
+            {%- set reasoning_content = message.reasoning_content %}
+        {%- else %}
+            {%- if '</think>' in message.content %}
+                {%- set content = message.content.split('</think>')[-1].lstrip('\n') %}
+                {%- set reasoning_content = message.content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
+            {%- endif %}
+        {%- endif %}
+        {%- if loop.index0 > ns.last_query_index %}
+            {%- if loop.last or (not loop.last and reasoning_content) %}
+                {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content.strip('\n') + '\n</think>\n\n' + content.lstrip('\n') }}
+            {%- else %}
+                {{- '<|im_start|>' + message.role + '\n' + content }}
+            {%- endif %}
+        {%- else %}
+            {{- '<|im_start|>' + message.role + '\n' + content }}
+        {%- endif %}
+        {%- if message.tool_calls %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if (loop.first and content) or (not loop.first) %}
+                    {{- '\n' }}
+                {%- endif %}
+                {%- if tool_call.function %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {{- '<tool_call>\n{"name": "' }}
+                {{- tool_call.name }}
+                {{- '", "arguments": ' }}
+                {%- if tool_call.arguments is string %}
+                    {{- tool_call.arguments }}
+                {%- else %}
+                    {{- tool_call.arguments | tojson }}
+                {%- endif %}
+                {{- '}\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.first or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+    {%- if enable_thinking is defined and enable_thinking is false %}
+        {{- '<think>\n\n</think>\n\n' }}
+    {%- endif %}
+{%- endif %}
\ No newline at end of file
diff --git a/runs/t0/t0_S4_lr0.0006/checkpoint-864/config.json b/runs/t0/t0_S4_lr0.0006/checkpoint-864/config.json
new file mode 100644
index 0000000000000000000000000000000000000000..ef8760b121e086715460fb47a7acf8f81e61b283
--- /dev/null
+++ b/runs/t0/t0_S4_lr0.0006/checkpoint-864/config.json
@@ -0,0 +1,32 @@
+{
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": null,
+  "dtype": "float32",
+  "eos_token_id": 151645,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 2816,
+  "max_position_embeddings": 2048,
+  "mlp_bias": false,
+  "model_type": "llama",
+  "num_attention_heads": 8,
+  "num_hidden_layers": 24,
+  "num_key_value_heads": 8,
+  "pad_token_id": 151645,
+  "pretraining_tp": 1,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 10000.0,
+    "rope_type": "default"
+  },
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": false,
+  "vocab_size": 151671
+}
diff --git a/runs/t0/t0_S4_lr0.0006/checkpoint-864/generation_config.json b/runs/t0/t0_S4_lr0.0006/checkpoint-864/generation_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..8c4242f32d40df256857f87264f646130f684e33
--- /dev/null
+++ b/runs/t0/t0_S4_lr0.0006/checkpoint-864/generation_config.json
@@ -0,0 +1,11 @@
+{
+  "_from_model_config": true,
+  "eos_token_id": [
+    151645
+  ],
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "transformers_version": "5.15.0",
+  "use_cache": true
+}
diff --git a/runs/t0/t0_S4_lr0.0006/checkpoint-864/tokenizer_config.json b/runs/t0/t0_S4_lr0.0006/checkpoint-864/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..839a3743177a3ed77e1ffbae67520e8500ca0532
--- /dev/null
+++ b/runs/t0/t0_S4_lr0.0006/checkpoint-864/tokenizer_config.json
@@ -0,0 +1,19 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|l2r_pred|>",
+    "<|r2l_pred|>"
+  ],
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 131072,
+  "pad_token": "<|im_end|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}
diff --git a/runs/t0/t0_S4_lr0.0006/checkpoint-864/trainer_state.json b/runs/t0/t0_S4_lr0.0006/checkpoint-864/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..6f57ec813909c25a0a2e883ed765785db826d40b
--- /dev/null
+++ b/runs/t0/t0_S4_lr0.0006/checkpoint-864/trainer_state.json
@@ -0,0 +1,739 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 12.0,
+  "eval_steps": 72,
+  "global_step": 864,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.013961605584642234,
+      "grad_norm": 2.9206132888793945,
+      "learning_rate": 0.0,
+      "loss": 12.153533935546875,
+      "step": 1
+    },
+    {
+      "epoch": 0.13961605584642234,
+      "grad_norm": 2.1237220764160156,
+      "learning_rate": 5.399999999999999e-05,
+      "loss": 11.390463087293837,
+      "step": 10
+    },
+    {
+      "epoch": 0.2792321116928447,
+      "grad_norm": 1.830725073814392,
+      "learning_rate": 0.00011399999999999999,
+      "loss": 10.077462768554687,
+      "step": 20
+    },
+    {
+      "epoch": 0.418848167539267,
+      "grad_norm": 1.0628323554992676,
+      "learning_rate": 0.00017399999999999997,
+      "loss": 8.793680572509766,
+      "step": 30
+    },
+    {
+      "epoch": 0.5584642233856894,
+      "grad_norm": 0.45776990056037903,
+      "learning_rate": 0.000234,
+      "loss": 7.745941162109375,
+      "step": 40
+    },
+    {
+      "epoch": 0.6980802792321117,
+      "grad_norm": 1.1509424448013306,
+      "learning_rate": 0.000294,
+      "loss": 7.502726745605469,
+      "step": 50
+    },
+    {
+      "epoch": 0.837696335078534,
+      "grad_norm": 0.8295908570289612,
+      "learning_rate": 0.00035399999999999993,
+      "loss": 7.317385864257813,
+      "step": 60
+    },
+    {
+      "epoch": 0.9773123909249564,
+      "grad_norm": 0.8668380975723267,
+      "learning_rate": 0.0004139999999999999,
+      "loss": 7.089698028564453,
+      "step": 70
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 6.985106468200684,
+      "eval_runtime": 29.6678,
+      "eval_samples_per_second": 82.278,
+      "eval_steps_per_second": 1.315,
+      "step": 72
+    },
+    {
+      "epoch": 1.111692844677138,
+      "grad_norm": 0.6774813532829285,
+      "learning_rate": 0.000474,
+      "loss": 6.9334571838378904,
+      "step": 80
+    },
+    {
+      "epoch": 1.2513089005235603,
+      "grad_norm": 0.6552468538284302,
+      "learning_rate": 0.000534,
+      "loss": 6.771307373046875,
+      "step": 90
+    },
+    {
+      "epoch": 1.3909249563699826,
+      "grad_norm": 1.0316975116729736,
+      "learning_rate": 0.0005939999999999999,
+      "loss": 6.63712158203125,
+      "step": 100
+    },
+    {
+      "epoch": 1.530541012216405,
+      "grad_norm": 0.45376425981521606,
+      "learning_rate": 0.0006,
+      "loss": 6.486009979248047,
+      "step": 110
+    },
+    {
+      "epoch": 1.6701570680628273,
+      "grad_norm": 0.7860228419303894,
+      "learning_rate": 0.0006,
+      "loss": 6.39349594116211,
+      "step": 120
+    },
+    {
+      "epoch": 1.8097731239092496,
+      "grad_norm": 0.4239625632762909,
+      "learning_rate": 0.0006,
+      "loss": 6.256306457519531,
+      "step": 130
+    },
+    {
+      "epoch": 1.949389179755672,
+      "grad_norm": 0.6127851605415344,
+      "learning_rate": 0.0006,
+      "loss": 6.163679885864258,
+      "step": 140
+    },
+    {
+      "epoch": 2.0,
+      "eval_loss": 6.093886852264404,
+      "eval_runtime": 5.4342,
+      "eval_samples_per_second": 449.188,
+      "eval_steps_per_second": 7.177,
+      "step": 144
+    },
+    {
+      "epoch": 2.0837696335078535,
+      "grad_norm": 0.47936490178108215,
+      "learning_rate": 0.0006,
+      "loss": 6.04110107421875,
+      "step": 150
+    },
+    {
+      "epoch": 2.223385689354276,
+      "grad_norm": 0.5541312098503113,
+      "learning_rate": 0.0006,
+      "loss": 5.978498077392578,
+      "step": 160
+    },
+    {
+      "epoch": 2.363001745200698,
+      "grad_norm": 0.3599136471748352,
+      "learning_rate": 0.0006,
+      "loss": 5.869022369384766,
+      "step": 170
+    },
+    {
+      "epoch": 2.5026178010471205,
+      "grad_norm": 0.42818182706832886,
+      "learning_rate": 0.0006,
+      "loss": 5.82703742980957,
+      "step": 180
+    },
+    {
+      "epoch": 2.642233856893543,
+      "grad_norm": 0.6170725226402283,
+      "learning_rate": 0.0006,
+      "loss": 5.707303237915039,
+      "step": 190
+    },
+    {
+      "epoch": 2.781849912739965,
+      "grad_norm": 0.5079529881477356,
+      "learning_rate": 0.0006,
+      "loss": 5.721559524536133,
+      "step": 200
+    },
+    {
+      "epoch": 2.9214659685863875,
+      "grad_norm": 0.22502170503139496,
+      "learning_rate": 0.0006,
+      "loss": 5.599060821533203,
+      "step": 210
+    },
+    {
+      "epoch": 3.0,
+      "eval_loss": 5.532996654510498,
+      "eval_runtime": 5.4393,
+      "eval_samples_per_second": 448.773,
+      "eval_steps_per_second": 7.17,
+      "step": 216
+    },
+    {
+      "epoch": 3.055846422338569,
+      "grad_norm": 0.4988436996936798,
+      "learning_rate": 0.0006,
+      "loss": 5.476123046875,
+      "step": 220
+    },
+    {
+      "epoch": 3.195462478184991,
+      "grad_norm": 0.47404739260673523,
+      "learning_rate": 0.0006,
+      "loss": 5.449727630615234,
+      "step": 230
+    },
+    {
+      "epoch": 3.3350785340314135,
+      "grad_norm": 0.29689666628837585,
+      "learning_rate": 0.0006,
+      "loss": 5.350494003295898,
+      "step": 240
+    },
+    {
+      "epoch": 3.474694589877836,
+      "grad_norm": 0.5025255680084229,
+      "learning_rate": 0.0006,
+      "loss": 5.306208801269531,
+      "step": 250
+    },
+    {
+      "epoch": 3.614310645724258,
+      "grad_norm": 0.3087793290615082,
+      "learning_rate": 0.0006,
+      "loss": 5.252265167236328,
+      "step": 260
+    },
+    {
+      "epoch": 3.7539267015706805,
+      "grad_norm": 0.24143578112125397,
+      "learning_rate": 0.0006,
+      "loss": 5.147513580322266,
+      "step": 270
+    },
+    {
+      "epoch": 3.893542757417103,
+      "grad_norm": 0.6470257639884949,
+      "learning_rate": 0.0006,
+      "loss": 5.087617492675781,
+      "step": 280
+    },
+    {
+      "epoch": 4.0,
+      "eval_loss": 5.172598361968994,
+      "eval_runtime": 5.4512,
+      "eval_samples_per_second": 447.792,
+      "eval_steps_per_second": 7.154,
+      "step": 288
+    },
+    {
+      "epoch": 4.027923211169284,
+      "grad_norm": 0.6637756824493408,
+      "learning_rate": 0.0006,
+      "loss": 5.070102691650391,
+      "step": 290
+    },
+    {
+      "epoch": 4.167539267015707,
+      "grad_norm": 0.25461071729660034,
+      "learning_rate": 0.0006,
+      "loss": 4.9610851287841795,
+      "step": 300
+    },
+    {
+      "epoch": 4.307155322862129,
+      "grad_norm": 0.5295916795730591,
+      "learning_rate": 0.0006,
+      "loss": 4.859738159179687,
+      "step": 310
+    },
+    {
+      "epoch": 4.446771378708552,
+      "grad_norm": 0.5386121869087219,
+      "learning_rate": 0.0006,
+      "loss": 4.826572418212891,
+      "step": 320
+    },
+    {
+      "epoch": 4.5863874345549736,
+      "grad_norm": 0.42201587557792664,
+      "learning_rate": 0.0006,
+      "loss": 4.807993316650391,
+      "step": 330
+    },
+    {
+      "epoch": 4.726003490401396,
+      "grad_norm": 0.42141446471214294,
+      "learning_rate": 0.0006,
+      "loss": 4.7262310028076175,
+      "step": 340
+    },
+    {
+      "epoch": 4.865619546247818,
+      "grad_norm": 0.25011977553367615,
+      "learning_rate": 0.0006,
+      "loss": 4.645610046386719,
+      "step": 350
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 0.40314558148384094,
+      "learning_rate": 0.0006,
+      "loss": 4.607339477539062,
+      "step": 360
+    },
+    {
+      "epoch": 5.0,
+      "eval_loss": 4.691513538360596,
+      "eval_runtime": 5.4559,
+      "eval_samples_per_second": 447.402,
+      "eval_steps_per_second": 7.148,
+      "step": 360
+    },
+    {
+      "epoch": 5.139616055846423,
+      "grad_norm": 0.35727429389953613,
+      "learning_rate": 0.0006,
+      "loss": 4.493846130371094,
+      "step": 370
+    },
+    {
+      "epoch": 5.279232111692845,
+      "grad_norm": 0.579684317111969,
+      "learning_rate": 0.0006,
+      "loss": 4.477697372436523,
+      "step": 380
+    },
+    {
+      "epoch": 5.418848167539267,
+      "grad_norm": 0.4042253792285919,
+      "learning_rate": 0.0006,
+      "loss": 4.435718154907226,
+      "step": 390
+    },
+    {
+      "epoch": 5.558464223385689,
+      "grad_norm": 0.2034163773059845,
+      "learning_rate": 0.0006,
+      "loss": 4.361290740966797,
+      "step": 400
+    },
+    {
+      "epoch": 5.698080279232112,
+      "grad_norm": 0.21729686856269836,
+      "learning_rate": 0.0006,
+      "loss": 4.318072509765625,
+      "step": 410
+    },
+    {
+      "epoch": 5.837696335078534,
+      "grad_norm": 0.45357993245124817,
+      "learning_rate": 0.0006,
+      "loss": 4.285897827148437,
+      "step": 420
+    },
+    {
+      "epoch": 5.977312390924956,
+      "grad_norm": 0.4423562288284302,
+      "learning_rate": 0.0006,
+      "loss": 4.292565155029297,
+      "step": 430
+    },
+    {
+      "epoch": 6.0,
+      "eval_loss": 4.448753356933594,
+      "eval_runtime": 5.4351,
+      "eval_samples_per_second": 449.117,
+      "eval_steps_per_second": 7.176,
+      "step": 432
+    },
+    {
+      "epoch": 6.111692844677138,
+      "grad_norm": 0.6937609910964966,
+      "learning_rate": 0.0006,
+      "loss": 4.254433441162109,
+      "step": 440
+    },
+    {
+      "epoch": 6.2513089005235605,
+      "grad_norm": 0.501941442489624,
+      "learning_rate": 0.0006,
+      "loss": 4.290074157714844,
+      "step": 450
+    },
+    {
+      "epoch": 6.390924956369982,
+      "grad_norm": 0.2516627907752991,
+      "learning_rate": 0.0006,
+      "loss": 4.194451522827149,
+      "step": 460
+    },
+    {
+      "epoch": 6.530541012216405,
+      "grad_norm": 0.14753806591033936,
+      "learning_rate": 0.0006,
+      "loss": 4.124375915527343,
+      "step": 470
+    },
+    {
+      "epoch": 6.670157068062827,
+      "grad_norm": 0.14808453619480133,
+      "learning_rate": 0.0006,
+      "loss": 4.078779602050782,
+      "step": 480
+    },
+    {
+      "epoch": 6.80977312390925,
+      "grad_norm": 0.20621934533119202,
+      "learning_rate": 0.0006,
+      "loss": 4.063456344604492,
+      "step": 490
+    },
+    {
+      "epoch": 6.949389179755672,
+      "grad_norm": 0.19101426005363464,
+      "learning_rate": 0.0006,
+      "loss": 4.029528045654297,
+      "step": 500
+    },
+    {
+      "epoch": 7.0,
+      "eval_loss": 4.217341899871826,
+      "eval_runtime": 5.4543,
+      "eval_samples_per_second": 447.538,
+      "eval_steps_per_second": 7.15,
+      "step": 504
+    },
+    {
+      "epoch": 7.0837696335078535,
+      "grad_norm": 0.34916263818740845,
+      "learning_rate": 0.0006,
+      "loss": 3.9608695983886717,
+      "step": 510
+    },
+    {
+      "epoch": 7.223385689354275,
+      "grad_norm": 0.4681544005870819,
+      "learning_rate": 0.0006,
+      "loss": 3.9381080627441407,
+      "step": 520
+    },
+    {
+      "epoch": 7.363001745200698,
+      "grad_norm": 0.2090049833059311,
+      "learning_rate": 0.0006,
+      "loss": 3.907509994506836,
+      "step": 530
+    },
+    {
+      "epoch": 7.50261780104712,
+      "grad_norm": 0.17446690797805786,
+      "learning_rate": 0.0006,
+      "loss": 3.882748031616211,
+      "step": 540
+    },
+    {
+      "epoch": 7.642233856893543,
+      "grad_norm": 0.20346012711524963,
+      "learning_rate": 0.0006,
+      "loss": 3.867753601074219,
+      "step": 550
+    },
+    {
+      "epoch": 7.781849912739965,
+      "grad_norm": 0.24103151261806488,
+      "learning_rate": 0.0006,
+      "loss": 3.861027145385742,
+      "step": 560
+    },
+    {
+      "epoch": 7.9214659685863875,
+      "grad_norm": 0.22708670794963837,
+      "learning_rate": 0.0006,
+      "loss": 3.848551940917969,
+      "step": 570
+    },
+    {
+      "epoch": 8.0,
+      "eval_loss": 4.099960803985596,
+      "eval_runtime": 5.4497,
+      "eval_samples_per_second": 447.913,
+      "eval_steps_per_second": 7.156,
+      "step": 576
+    },
+    {
+      "epoch": 8.055846422338568,
+      "grad_norm": 0.3998148739337921,
+      "learning_rate": 0.0006,
+      "loss": 3.7770992279052735,
+      "step": 580
+    },
+    {
+      "epoch": 8.195462478184991,
+      "grad_norm": 0.7326574921607971,
+      "learning_rate": 0.0006,
+      "loss": 3.76361083984375,
+      "step": 590
+    },
+    {
+      "epoch": 8.335078534031414,
+      "grad_norm": 1.4837658405303955,
+      "learning_rate": 0.0006,
+      "loss": 4.08636245727539,
+      "step": 600
+    },
+    {
+      "epoch": 8.474694589877837,
+      "grad_norm": 1.0409595966339111,
+      "learning_rate": 0.0006,
+      "loss": 4.354316711425781,
+      "step": 610
+    },
+    {
+      "epoch": 8.614310645724258,
+      "grad_norm": 0.8818002343177795,
+      "learning_rate": 0.0006,
+      "loss": 4.53700065612793,
+      "step": 620
+    },
+    {
+      "epoch": 8.75392670157068,
+      "grad_norm": 0.197915181517601,
+      "learning_rate": 0.0006,
+      "loss": 4.277849578857422,
+      "step": 630
+    },
+    {
+      "epoch": 8.893542757417103,
+      "grad_norm": 0.13980717957019806,
+      "learning_rate": 0.0006,
+      "loss": 4.076632690429688,
+      "step": 640
+    },
+    {
+      "epoch": 9.0,
+      "eval_loss": 4.183110237121582,
+      "eval_runtime": 5.4533,
+      "eval_samples_per_second": 447.615,
+      "eval_steps_per_second": 7.152,
+      "step": 648
+    },
+    {
+      "epoch": 9.027923211169284,
+      "grad_norm": 0.12084736675024033,
+      "learning_rate": 0.0006,
+      "loss": 3.934686279296875,
+      "step": 650
+    },
+    {
+      "epoch": 9.167539267015707,
+      "grad_norm": 0.14144660532474518,
+      "learning_rate": 0.0006,
+      "loss": 3.8070610046386717,
+      "step": 660
+    },
+    {
+      "epoch": 9.30715532286213,
+      "grad_norm": 0.13604827225208282,
+      "learning_rate": 0.0006,
+      "loss": 3.7739948272705077,
+      "step": 670
+    },
+    {
+      "epoch": 9.44677137870855,
+      "grad_norm": 0.13959047198295593,
+      "learning_rate": 0.0006,
+      "loss": 3.7471954345703127,
+      "step": 680
+    },
+    {
+      "epoch": 9.586387434554974,
+      "grad_norm": 0.12865500152111053,
+      "learning_rate": 0.0006,
+      "loss": 3.7092533111572266,
+      "step": 690
+    },
+    {
+      "epoch": 9.726003490401396,
+      "grad_norm": 0.17941144108772278,
+      "learning_rate": 0.0006,
+      "loss": 3.7057247161865234,
+      "step": 700
+    },
+    {
+      "epoch": 9.86561954624782,
+      "grad_norm": 0.1411111056804657,
+      "learning_rate": 0.0006,
+      "loss": 3.679571533203125,
+      "step": 710
+    },
+    {
+      "epoch": 10.0,
+      "grad_norm": 0.16089625656604767,
+      "learning_rate": 0.0006,
+      "loss": 3.6586647033691406,
+      "step": 720
+    },
+    {
+      "epoch": 10.0,
+      "eval_loss": 4.012387752532959,
+      "eval_runtime": 5.4629,
+      "eval_samples_per_second": 446.835,
+      "eval_steps_per_second": 7.139,
+      "step": 720
+    },
+    {
+      "epoch": 10.139616055846423,
+      "grad_norm": 0.3115384876728058,
+      "learning_rate": 0.0006,
+      "loss": 3.541545867919922,
+      "step": 730
+    },
+    {
+      "epoch": 10.279232111692846,
+      "grad_norm": 0.27592477202415466,
+      "learning_rate": 0.0006,
+      "loss": 3.5572509765625,
+      "step": 740
+    },
+    {
+      "epoch": 10.418848167539267,
+      "grad_norm": 0.17261362075805664,
+      "learning_rate": 0.0006,
+      "loss": 3.5457481384277343,
+      "step": 750
+    },
+    {
+      "epoch": 10.55846422338569,
+      "grad_norm": 0.1407252699136734,
+      "learning_rate": 0.0006,
+      "loss": 3.5450862884521483,
+      "step": 760
+    },
+    {
+      "epoch": 10.698080279232112,
+      "grad_norm": 0.12504896521568298,
+      "learning_rate": 0.0006,
+      "loss": 3.5322113037109375,
+      "step": 770
+    },
+    {
+      "epoch": 10.837696335078533,
+      "grad_norm": 0.134646475315094,
+      "learning_rate": 0.0006,
+      "loss": 3.5337203979492187,
+      "step": 780
+    },
+    {
+      "epoch": 10.977312390924956,
+      "grad_norm": 0.14491793513298035,
+      "learning_rate": 0.0006,
+      "loss": 3.5220596313476564,
+      "step": 790
+    },
+    {
+      "epoch": 11.0,
+      "eval_loss": 3.970649480819702,
+      "eval_runtime": 5.4496,
+      "eval_samples_per_second": 447.925,
+      "eval_steps_per_second": 7.157,
+      "step": 792
+    },
+    {
+      "epoch": 11.111692844677139,
+      "grad_norm": 0.3893614709377289,
+      "learning_rate": 0.0006,
+      "loss": 3.3977737426757812,
+      "step": 800
+    },
+    {
+      "epoch": 11.25130890052356,
+      "grad_norm": 0.24212592840194702,
+      "learning_rate": 0.0006,
+      "loss": 3.3481903076171875,
+      "step": 810
+    },
+    {
+      "epoch": 11.390924956369982,
+      "grad_norm": 0.2399962693452835,
+      "learning_rate": 0.0006,
+      "loss": 3.3614341735839846,
+      "step": 820
+    },
+    {
+      "epoch": 11.530541012216405,
+      "grad_norm": 0.18185333907604218,
+      "learning_rate": 0.0006,
+      "loss": 3.377635192871094,
+      "step": 830
+    },
+    {
+      "epoch": 11.670157068062828,
+      "grad_norm": 0.1585046797990799,
+      "learning_rate": 0.0006,
+      "loss": 3.3790657043457033,
+      "step": 840
+    },
+    {
+      "epoch": 11.809773123909249,
+      "grad_norm": 0.17206858098506927,
+      "learning_rate": 0.0006,
+      "loss": 3.387134552001953,
+      "step": 850
+    },
+    {
+      "epoch": 11.949389179755672,
+      "grad_norm": 0.16680264472961426,
+      "learning_rate": 0.0006,
+      "loss": 3.397046661376953,
+      "step": 860
+    },
+    {
+      "epoch": 12.0,
+      "eval_loss": 3.9641687870025635,
+      "eval_runtime": 5.4473,
+      "eval_samples_per_second": 448.114,
+      "eval_steps_per_second": 7.16,
+      "step": 864
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 864,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 12,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": false,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.6673046800042557e+18,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/runs/t0/t0_S4_lr0.0006/train.log b/runs/t0/t0_S4_lr0.0006/train.log
new file mode 100644
index 0000000000000000000000000000000000000000..d80ac7484f15c8b68bcfcc12e80a0e471c94977d
--- /dev/null
+++ b/runs/t0/t0_S4_lr0.0006/train.log
@@ -0,0 +1,1655 @@
+W0817 20:46:49.863000 2279311 torch/distributed/run.py:851] 
+W0817 20:46:49.863000 2279311 torch/distributed/run.py:851] *****************************************
+W0817 20:46:49.863000 2279311 torch/distributed/run.py:851] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+W0817 20:46:49.863000 2279311 torch/distributed/run.py:851] *****************************************
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+EOS token: <|im_end|>
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+total params: 463642624
+[dataset] loaded packed cache: 36614 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+total params: 463642624
+total params:[dataset] loaded packed cache: 36614 train / 2441 eval blocks
+ 463642624
+[transformers] PyTorch: setting up devices
+[dataset] loaded packed cache: 36614 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+total params: 463642624
+[dataset] loaded packed cache: 36614 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+total params: 463642624
+[dataset] loaded packed cache: 36614 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+total params: 463642624
+[dataset] loaded packed cache: 36614 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+total params: 463642624
+[dataset] loaded packed cache: 36614 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+total params: 463642624
+[dataset] loaded packed cache: 36614 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+[transformers] ***** Running training *****
+[transformers]   Num examples = 36,614
+[transformers]   Num Epochs = 12
+[transformers]   Num update steps per epoch = 72
+[transformers]   Instantaneous batch size per device = 8
+[transformers]   Total train batch size (w. parallel, distributed & accumulation) = 512
+[transformers]   Gradient Accumulation steps = 8
+[transformers]   Total optimization steps = 864
+[transformers]   Number of trainable parameters = 463,642,624
+wandb: WARNING Changes to your `wandb` environment variables will be ignored because your `wandb` session has already started. For more information on how to modify your settings with `wandb.init()` arguments, please refer to https://wandb.me/wandb-init.
+wandb: setting up run xd89q9zj
+wandb: Tracking run with wandb version 0.28.2
+wandb: Run data is saved locally in /home/ubuntu/workspace/data-augmentations-for-pretraining/wandb/run-20260817_204713-xd89q9zj
+wandb: Run `wandb offline` to turn off syncing.
+wandb: Syncing run t0_S4_lr0.0006
+wandb: ⭐️ View project at https://wandb.ai/slowrun-wankers/data-aug-pretraining
+wandb: 🚀 View run at https://wandb.ai/slowrun-wankers/data-aug-pretraining/runs/xd89q9zj
+
  0%|          | 0/864 [00:00<?, ?it/s]wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+[rank4]:W0817 20:47:36.475000 2279486 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+[rank2]:W0817 20:47:36.630000 2279484 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+[rank7]:W0817 20:47:36.789000 2279489 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+[rank6]:W0817 20:47:36.890000 2279488 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+[rank5]:W0817 20:47:37.083000 2279487 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+[rank1]:W0817 20:47:37.200000 2279483 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+[rank3]:W0817 20:47:37.322000 2279485 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+[rank0]:W0817 20:47:38.405000 2279482 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+
  0%|          | 1/864 [00:44<10:35:26, 44.18s/it]
                                                  
{'loss': '12.15', 'grad_norm': '2.921', 'learning_rate': '0', 'epoch': '0.01396'}
+
  0%|          | 1/864 [00:44<10:35:26, 44.18s/it]
  0%|          | 2/864 [00:45<4:32:31, 18.97s/it] 
  0%|          | 3/864 [00:46<2:36:25, 10.90s/it]
  0%|          | 4/864 [00:48<1:41:54,  7.11s/it]
  1%|          | 5/864 [00:49<1:11:48,  5.02s/it]
  1%|          | 6/864 [00:50<53:39,  3.75s/it]  
  1%|          | 7/864 [00:52<42:08,  2.95s/it]
  1%|          | 8/864 [00:53<34:36,  2.43s/it]
  1%|          | 9/864 [00:54<29:33,  2.07s/it]
  1%|          | 10/864 [00:55<26:06,  1.83s/it]
                                                
{'loss': '11.39', 'grad_norm': '2.124', 'learning_rate': '5.4e-05', 'epoch': '0.1396'}
+
  1%|          | 10/864 [00:55<26:06,  1.83s/it]
  1%|▏         | 11/864 [00:57<23:44,  1.67s/it]
  1%|▏         | 12/864 [00:58<22:06,  1.56s/it]
  2%|▏         | 13/864 [00:59<20:58,  1.48s/it]
  2%|▏         | 14/864 [01:01<21:55,  1.55s/it]
  2%|▏         | 15/864 [01:03<22:53,  1.62s/it]
  2%|▏         | 16/864 [01:04<21:29,  1.52s/it]
  2%|▏         | 17/864 [01:05<20:31,  1.45s/it]
  2%|▏         | 18/864 [01:07<19:50,  1.41s/it]
  2%|▏         | 19/864 [01:08<19:21,  1.37s/it]
  2%|▏         | 20/864 [01:09<19:00,  1.35s/it]
                                                
{'loss': '10.08', 'grad_norm': '1.831', 'learning_rate': '0.000114', 'epoch': '0.2792'}
+
  2%|▏         | 20/864 [01:09<19:00,  1.35s/it]
  2%|▏         | 21/864 [01:11<18:45,  1.34s/it]
  3%|▎         | 22/864 [01:12<18:34,  1.32s/it]
  3%|▎         | 23/864 [01:13<18:26,  1.32s/it]
  3%|▎         | 24/864 [01:14<18:20,  1.31s/it]
  3%|▎         | 25/864 [01:16<18:16,  1.31s/it]
  3%|▎         | 26/864 [01:17<18:13,  1.30s/it]
  3%|▎         | 27/864 [01:18<18:10,  1.30s/it]
  3%|▎         | 28/864 [01:20<18:09,  1.30s/it]
  3%|▎         | 29/864 [01:21<18:08,  1.30s/it]
  3%|▎         | 30/864 [01:22<18:08,  1.31s/it]
                                                
{'loss': '8.794', 'grad_norm': '1.063', 'learning_rate': '0.000174', 'epoch': '0.4188'}
+
  3%|▎         | 30/864 [01:22<18:08,  1.31s/it]
  4%|▎         | 31/864 [01:24<18:07,  1.31s/it]
  4%|▎         | 32/864 [01:25<18:05,  1.30s/it]
  4%|▍         | 33/864 [01:26<18:03,  1.30s/it]
  4%|▍         | 34/864 [01:27<18:00,  1.30s/it]
  4%|▍         | 35/864 [01:29<17:58,  1.30s/it]
  4%|▍         | 36/864 [01:30<17:57,  1.30s/it]
  4%|▍         | 37/864 [01:31<17:56,  1.30s/it]
  4%|▍         | 38/864 [01:33<17:56,  1.30s/it]
  5%|▍         | 39/864 [01:34<17:55,  1.30s/it]
  5%|▍         | 40/864 [01:35<17:54,  1.30s/it]
                                                
{'loss': '7.746', 'grad_norm': '0.4578', 'learning_rate': '0.000234', 'epoch': '0.5585'}
+
  5%|▍         | 40/864 [01:35<17:54,  1.30s/it]
  5%|▍         | 41/864 [01:37<17:53,  1.30s/it]
  5%|▍         | 42/864 [01:38<17:52,  1.30s/it]
  5%|▍         | 43/864 [01:39<17:51,  1.31s/it]
  5%|▌         | 44/864 [01:41<17:49,  1.30s/it]
  5%|▌         | 45/864 [01:42<17:48,  1.31s/it]
  5%|▌         | 46/864 [01:43<17:48,  1.31s/it]
  5%|▌         | 47/864 [01:44<17:47,  1.31s/it]
  6%|▌         | 48/864 [01:46<17:45,  1.31s/it]
  6%|▌         | 49/864 [01:47<17:45,  1.31s/it]
  6%|▌         | 50/864 [01:48<17:44,  1.31s/it]
                                                
{'loss': '7.503', 'grad_norm': '1.151', 'learning_rate': '0.000294', 'epoch': '0.6981'}
+
  6%|▌         | 50/864 [01:48<17:44,  1.31s/it]
  6%|▌         | 51/864 [01:50<17:43,  1.31s/it]
  6%|▌         | 52/864 [01:51<17:42,  1.31s/it]
  6%|▌         | 53/864 [01:52<17:41,  1.31s/it]
  6%|▋         | 54/864 [01:54<17:40,  1.31s/it]
  6%|▋         | 55/864 [01:55<17:40,  1.31s/it]
  6%|▋         | 56/864 [01:56<17:38,  1.31s/it]
  7%|▋         | 57/864 [01:58<17:37,  1.31s/it]
  7%|▋         | 58/864 [01:59<17:36,  1.31s/it]
  7%|▋         | 59/864 [02:00<17:35,  1.31s/it]
  7%|▋         | 60/864 [02:01<17:33,  1.31s/it]
                                                
{'loss': '7.317', 'grad_norm': '0.8296', 'learning_rate': '0.000354', 'epoch': '0.8377'}
+
  7%|▋         | 60/864 [02:01<17:33,  1.31s/it]
  7%|▋         | 61/864 [02:03<17:32,  1.31s/it]
  7%|▋         | 62/864 [02:04<17:29,  1.31s/it]
  7%|▋         | 63/864 [02:05<17:30,  1.31s/it]
  7%|▋         | 64/864 [02:07<17:28,  1.31s/it]
  8%|▊         | 65/864 [02:08<17:27,  1.31s/it]
  8%|▊         | 66/864 [02:09<17:24,  1.31s/it]
  8%|▊         | 67/864 [02:11<17:24,  1.31s/it]
  8%|▊         | 68/864 [02:12<17:23,  1.31s/it]
  8%|▊         | 69/864 [02:13<17:22,  1.31s/it]
  8%|▊         | 70/864 [02:15<17:21,  1.31s/it]
                                                
{'loss': '7.09', 'grad_norm': '0.8668', 'learning_rate': '0.000414', 'epoch': '0.9773'}
+
  8%|▊         | 70/864 [02:15<17:21,  1.31s/it]
  8%|▊         | 71/864 [02:16<17:20,  1.31s/it]
  8%|▊         | 72/864 [02:17<15:24,  1.17s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.33it/s]
+
 13%|█▎        | 5/39 [00:00<00:03, 10.49it/s]
+
 18%|█▊        | 7/39 [00:00<00:03,  8.86it/s]
+
 21%|██        | 8/39 [00:00<00:03,  8.43it/s]
+
 23%|██▎       | 9/39 [00:00<00:03,  8.11it/s]
+
 26%|██▌       | 10/39 [00:01<00:03,  7.87it/s]
+
 28%|██▊       | 11/39 [00:01<00:03,  7.70it/s]
+
 31%|███       | 12/39 [00:01<00:03,  7.57it/s]
+
 33%|███▎      | 13/39 [00:01<00:03,  7.47it/s]
+
 36%|███▌      | 14/39 [00:01<00:03,  7.41it/s]
+
 38%|███▊      | 15/39 [00:01<00:03,  7.36it/s]
+
 41%|████      | 16/39 [00:01<00:03,  7.33it/s]
+
 44%|████▎     | 17/39 [00:02<00:03,  7.31it/s]
+
 46%|████▌     | 18/39 [00:02<00:02,  7.29it/s]
+
 49%|████▊     | 19/39 [00:02<00:02,  7.28it/s]
+
 51%|█████▏    | 20/39 [00:02<00:02,  7.28it/s]
+
 54%|█████▍    | 21/39 [00:02<00:02,  7.27it/s]
+
 56%|█████▋    | 22/39 [00:02<00:02,  7.27it/s]
+
 59%|█████▉    | 23/39 [00:02<00:02,  7.27it/s]
+
 62%|██████▏   | 24/39 [00:03<00:02,  7.26it/s]
+
 64%|██████▍   | 25/39 [00:03<00:01,  7.26it/s]
+
 67%|██████▋   | 26/39 [00:03<00:01,  7.26it/s]
+
 69%|██████▉   | 27/39 [00:03<00:01,  7.26it/s]
+
 72%|███████▏  | 28/39 [00:03<00:01,  7.26it/s]
+
 74%|███████▍  | 29/39 [00:03<00:01,  7.26it/s]
+
 77%|███████▋  | 30/39 [00:03<00:01,  7.26it/s]
+
 79%|███████▉  | 31/39 [00:04<00:01,  7.26it/s]
+
 82%|████████▏ | 32/39 [00:04<00:00,  7.26it/s]
+
 85%|████████▍ | 33/39 [00:04<00:00,  7.25it/s]
+
 87%|████████▋ | 34/39 [00:04<00:00,  7.25it/s]
+
 90%|████████▉ | 35/39 [00:04<00:00,  7.25it/s]
+
 92%|█████████▏| 36/39 [00:04<00:00,  7.26it/s]
+
 95%|█████████▍| 37/39 [00:04<00:00,  7.25it/s]
+
 97%|█████████▋| 38/39 [00:04<00:00,  7.26it/s]
+
100%|██████████| 39/39 [00:05<00:00,  7.26it/s]
                                                
+
                                               
{'eval_loss': '6.985', 'eval_runtime': '29.67', 'eval_samples_per_second': '82.28', 'eval_steps_per_second': '1.315', 'epoch': '1'}
+
  8%|▊         | 72/864 [02:46<15:24,  1.17s/it]
+
100%|██████████| 39/39 [00:05<00:00,  7.26it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0006/checkpoint-72
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0006/checkpoint-72/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0006/checkpoint-72/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:01<00:00,  1.82s/it]
Writing model shards: 100%|██████████| 1/1 [00:01<00:00,  1.82s/it]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0006/checkpoint-72/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0006/checkpoint-72/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0006/checkpoint-72/tokenizer_config.json
+[eval] epoch 1.0 step 72 loss 6.9851 (best 6.9851 @ ep 1.0, 0/10000 since best)
+
+
                                               
  8%|▊         | 73/864 [02:54<2:38:54, 12.05s/it]
  9%|▊         | 74/864 [02:55<1:56:14,  8.83s/it]
  9%|▊         | 75/864 [02:57<1:26:24,  6.57s/it]
  9%|▉         | 76/864 [02:58<1:05:33,  4.99s/it]
  9%|▉         | 77/864 [02:59<50:59,  3.89s/it]  
  9%|▉         | 78/864 [03:01<40:47,  3.11s/it]
  9%|▉         | 79/864 [03:02<33:39,  2.57s/it]
  9%|▉         | 80/864 [03:03<28:40,  2.19s/it]
                                                
{'loss': '6.933', 'grad_norm': '0.6775', 'learning_rate': '0.000474', 'epoch': '1.112'}
+
  9%|▉         | 80/864 [03:03<28:40,  2.19s/it]
  9%|▉         | 81/864 [03:05<25:10,  1.93s/it]
  9%|▉         | 82/864 [03:06<22:43,  1.74s/it]
 10%|▉         | 83/864 [03:07<21:00,  1.61s/it]
 10%|▉         | 84/864 [03:09<19:48,  1.52s/it]
 10%|▉         | 85/864 [03:10<18:57,  1.46s/it]
 10%|▉         | 86/864 [03:11<18:21,  1.42s/it]
 10%|█         | 87/864 [03:13<17:56,  1.38s/it]
 10%|█         | 88/864 [03:14<17:37,  1.36s/it]
 10%|█         | 89/864 [03:15<17:24,  1.35s/it]
 10%|█         | 90/864 [03:16<17:14,  1.34s/it]
                                                
{'loss': '6.771', 'grad_norm': '0.6552', 'learning_rate': '0.000534', 'epoch': '1.251'}
+
 10%|█         | 90/864 [03:16<17:14,  1.34s/it]
 11%|█         | 91/864 [03:18<17:07,  1.33s/it]
 11%|█         | 92/864 [03:19<17:02,  1.32s/it]
 11%|█         | 93/864 [03:20<16:58,  1.32s/it]
 11%|█         | 94/864 [03:22<16:54,  1.32s/it]
 11%|█         | 95/864 [03:23<16:51,  1.32s/it]
 11%|█         | 96/864 [03:24<16:49,  1.31s/it]
 11%|█         | 97/864 [03:26<16:48,  1.31s/it]
 11%|█▏        | 98/864 [03:27<16:46,  1.31s/it]
 11%|█▏        | 99/864 [03:28<16:45,  1.31s/it]
 12%|█▏        | 100/864 [03:30<16:44,  1.31s/it]
                                                 
{'loss': '6.637', 'grad_norm': '1.032', 'learning_rate': '0.000594', 'epoch': '1.391'}
+
 12%|█▏        | 100/864 [03:30<16:44,  1.31s/it]
 12%|█▏        | 101/864 [03:31<16:41,  1.31s/it]
 12%|█▏        | 102/864 [03:32<16:40,  1.31s/it]
 12%|█▏        | 103/864 [03:34<16:39,  1.31s/it]
 12%|█▏        | 104/864 [03:35<16:38,  1.31s/it]
 12%|█▏        | 105/864 [03:36<16:36,  1.31s/it]
 12%|█▏        | 106/864 [03:37<16:35,  1.31s/it]
 12%|█▏        | 107/864 [03:39<16:33,  1.31s/it]
 12%|█▎        | 108/864 [03:40<16:32,  1.31s/it]
 13%|█▎        | 109/864 [03:41<16:31,  1.31s/it]
 13%|█▎        | 110/864 [03:43<16:30,  1.31s/it]
                                                 
{'loss': '6.486', 'grad_norm': '0.4538', 'learning_rate': '0.0006', 'epoch': '1.531'}
+
 13%|█▎        | 110/864 [03:43<16:30,  1.31s/it]
 13%|█▎        | 111/864 [03:44<16:29,  1.31s/it]
 13%|█▎        | 112/864 [03:45<16:27,  1.31s/it]
 13%|█▎        | 113/864 [03:47<16:26,  1.31s/it]
 13%|█▎        | 114/864 [03:48<16:24,  1.31s/it]
 13%|█▎        | 115/864 [03:49<16:23,  1.31s/it]
 13%|█▎        | 116/864 [03:51<16:22,  1.31s/it]
 14%|█▎        | 117/864 [03:52<16:21,  1.31s/it]
 14%|█▎        | 118/864 [03:53<16:20,  1.31s/it]
 14%|█▍        | 119/864 [03:55<16:19,  1.31s/it]
 14%|█▍        | 120/864 [03:56<16:17,  1.31s/it]
                                                 
{'loss': '6.393', 'grad_norm': '0.786', 'learning_rate': '0.0006', 'epoch': '1.67'}
+
 14%|█▍        | 120/864 [03:56<16:17,  1.31s/it]
 14%|█▍        | 121/864 [03:57<16:16,  1.31s/it]
 14%|█▍        | 122/864 [03:58<16:14,  1.31s/it]
 14%|█▍        | 123/864 [04:00<16:14,  1.31s/it]
 14%|█▍        | 124/864 [04:01<16:12,  1.31s/it]
 14%|█▍        | 125/864 [04:02<16:11,  1.32s/it]
 15%|█▍        | 126/864 [04:04<16:10,  1.32s/it]
 15%|█▍        | 127/864 [04:05<16:08,  1.31s/it]
 15%|█▍        | 128/864 [04:06<16:07,  1.31s/it]
 15%|█▍        | 129/864 [04:08<16:06,  1.31s/it]
 15%|█▌        | 130/864 [04:09<16:03,  1.31s/it]
                                                 
{'loss': '6.256', 'grad_norm': '0.424', 'learning_rate': '0.0006', 'epoch': '1.81'}
+
 15%|█▌        | 130/864 [04:09<16:03,  1.31s/it]
 15%|█▌        | 131/864 [04:10<16:03,  1.31s/it]
 15%|█▌        | 132/864 [04:12<16:02,  1.31s/it]
 15%|█▌        | 133/864 [04:13<16:00,  1.31s/it]
 16%|█▌        | 134/864 [04:14<15:59,  1.31s/it]
 16%|█▌        | 135/864 [04:16<15:57,  1.31s/it]
 16%|█▌        | 136/864 [04:17<15:56,  1.31s/it]
 16%|█▌        | 137/864 [04:18<15:55,  1.31s/it]
 16%|█▌        | 138/864 [04:20<15:54,  1.31s/it]
 16%|█▌        | 139/864 [04:21<15:52,  1.31s/it]
 16%|█▌        | 140/864 [04:22<15:51,  1.31s/it]
                                                 
{'loss': '6.164', 'grad_norm': '0.6128', 'learning_rate': '0.0006', 'epoch': '1.949'}
+
 16%|█▌        | 140/864 [04:22<15:51,  1.31s/it]
 16%|█▋        | 141/864 [04:23<15:50,  1.31s/it]
 16%|█▋        | 142/864 [04:25<15:49,  1.32s/it]
 17%|█▋        | 143/864 [04:26<15:47,  1.31s/it]
 17%|█▋        | 144/864 [04:27<14:01,  1.17s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:02, 14.60it/s]
+
 10%|█         | 4/39 [00:00<00:03,  9.14it/s]
+
 15%|█▌        | 6/39 [00:00<00:04,  8.17it/s]
+
 18%|█▊        | 7/39 [00:00<00:04,  7.93it/s]
+
 21%|██        | 8/39 [00:00<00:04,  7.74it/s]
+
 23%|██▎       | 9/39 [00:01<00:03,  7.61it/s]
+
 26%|██▌       | 10/39 [00:01<00:03,  7.50it/s]
+
 28%|██▊       | 11/39 [00:01<00:03,  7.43it/s]
+
 31%|███       | 12/39 [00:01<00:03,  7.37it/s]
+
 33%|███▎      | 13/39 [00:01<00:03,  7.34it/s]
+
 36%|███▌      | 14/39 [00:01<00:03,  7.31it/s]
+
 38%|███▊      | 15/39 [00:01<00:03,  7.30it/s]
+
 41%|████      | 16/39 [00:02<00:03,  7.28it/s]
+
 44%|████▎     | 17/39 [00:02<00:03,  7.26it/s]
+
 46%|████▌     | 18/39 [00:02<00:02,  7.26it/s]
+
 49%|████▊     | 19/39 [00:02<00:02,  7.26it/s]
+
 51%|█████▏    | 20/39 [00:02<00:02,  7.25it/s]
+
 54%|█████▍    | 21/39 [00:02<00:02,  7.26it/s]
+
 56%|█████▋    | 22/39 [00:02<00:02,  7.26it/s]
+
 59%|█████▉    | 23/39 [00:03<00:02,  7.25it/s]
+
 62%|██████▏   | 24/39 [00:03<00:02,  7.26it/s]
+
 64%|██████▍   | 25/39 [00:03<00:01,  7.26it/s]
+
 67%|██████▋   | 26/39 [00:03<00:01,  7.25it/s]
+
 69%|██████▉   | 27/39 [00:03<00:01,  7.25it/s]
+
 72%|███████▏  | 28/39 [00:03<00:01,  7.25it/s]
+
 74%|███████▍  | 29/39 [00:03<00:01,  7.26it/s]
+
 77%|███████▋  | 30/39 [00:03<00:01,  7.26it/s]
+
 79%|███████▉  | 31/39 [00:04<00:01,  7.26it/s]
+
 82%|████████▏ | 32/39 [00:04<00:00,  7.26it/s]
+
 85%|████████▍ | 33/39 [00:04<00:00,  7.25it/s]
+
 87%|████████▋ | 34/39 [00:04<00:00,  7.23it/s]
+
 90%|████████▉ | 35/39 [00:04<00:00,  7.24it/s]
+
 92%|█████████▏| 36/39 [00:04<00:00,  7.24it/s]
+
 95%|█████████▍| 37/39 [00:04<00:00,  7.25it/s]
+
 97%|█████████▋| 38/39 [00:05<00:00,  7.25it/s]
+
100%|██████████| 39/39 [00:05<00:00,  7.25it/s]
                                                 
+
                                               
{'eval_loss': '6.094', 'eval_runtime': '5.434', 'eval_samples_per_second': '449.2', 'eval_steps_per_second': '7.177', 'epoch': '2'}
+
 17%|█▋        | 144/864 [04:32<14:01,  1.17s/it]
+
100%|██████████| 39/39 [00:05<00:00,  7.25it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0006/checkpoint-144
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0006/checkpoint-144/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0006/checkpoint-144/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:01<00:00,  1.84s/it]
Writing model shards: 100%|██████████| 1/1 [00:01<00:00,  1.84s/it]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0006/checkpoint-144/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0006/checkpoint-144/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0006/checkpoint-144/tokenizer_config.json
+[eval] epoch 2.0 step 144 loss 6.0939 (best 6.0939 @ ep 2.0, 0/10000 since best)
+
+
                                               
 17%|█▋        | 145/864 [04:40<55:57,  4.67s/it]
 17%|█▋        | 146/864 [04:41<43:48,  3.66s/it]
 17%|█▋        | 147/864 [04:42<35:19,  2.96s/it]
 17%|█▋        | 148/864 [04:44<29:23,  2.46s/it]
 17%|█▋        | 149/864 [04:45<25:13,  2.12s/it]
 17%|█▋        | 150/864 [04:46<22:19,  1.88s/it]
                                                 
{'loss': '6.041', 'grad_norm': '0.4794', 'learning_rate': '0.0006', 'epoch': '2.084'}
+
 17%|█▋        | 150/864 [04:46<22:19,  1.88s/it]
 17%|█▋        | 151/864 [04:48<20:17,  1.71s/it]
 18%|█▊        | 152/864 [04:49<18:51,  1.59s/it]
 18%|█▊        | 153/864 [04:50<17:51,  1.51s/it]
 18%|█▊        | 154/864 [04:52<17:08,  1.45s/it]
 18%|█▊        | 155/864 [04:53<16:38,  1.41s/it]
 18%|█▊        | 156/864 [04:54<16:17,  1.38s/it]
 18%|█▊        | 157/864 [04:55<16:01,  1.36s/it]
 18%|█▊        | 158/864 [04:57<15:49,  1.35s/it]
 18%|█▊        | 159/864 [04:58<15:41,  1.34s/it]
 19%|█▊        | 160/864 [04:59<15:35,  1.33s/it]
                                                 
{'loss': '5.978', 'grad_norm': '0.5541', 'learning_rate': '0.0006', 'epoch': '2.223'}
+
 19%|█▊        | 160/864 [04:59<15:35,  1.33s/it]
 19%|█▊        | 161/864 [05:01<15:30,  1.32s/it]
 19%|█▉        | 162/864 [05:02<15:27,  1.32s/it]
 19%|█▉        | 163/864 [05:03<15:24,  1.32s/it]
 19%|█▉        | 164/864 [05:05<15:22,  1.32s/it]
 19%|█▉        | 165/864 [05:06<15:20,  1.32s/it]
 19%|█▉        | 166/864 [05:07<15:18,  1.32s/it]
 19%|█▉        | 167/864 [05:09<15:16,  1.31s/it]
 19%|█▉        | 168/864 [05:10<15:14,  1.31s/it]
 20%|█▉        | 169/864 [05:11<15:12,  1.31s/it]
 20%|█▉        | 170/864 [05:13<15:11,  1.31s/it]
                                                 
{'loss': '5.869', 'grad_norm': '0.3599', 'learning_rate': '0.0006', 'epoch': '2.363'}
+
 20%|█▉        | 170/864 [05:13<15:11,  1.31s/it]
 20%|█▉        | 171/864 [05:14<15:10,  1.31s/it]
 20%|█▉        | 172/864 [05:15<15:09,  1.31s/it]
 20%|██        | 173/864 [05:17<15:07,  1.31s/it]
 20%|██        | 174/864 [05:18<15:06,  1.31s/it]
 20%|██        | 175/864 [05:19<15:05,  1.31s/it]
 20%|██        | 176/864 [05:20<15:03,  1.31s/it]
 20%|██        | 177/864 [05:22<15:02,  1.31s/it]
 21%|██        | 178/864 [05:23<15:00,  1.31s/it]
 21%|██        | 179/864 [05:24<14:59,  1.31s/it]
 21%|██        | 180/864 [05:26<14:58,  1.31s/it]
                                                 
{'loss': '5.827', 'grad_norm': '0.4282', 'learning_rate': '0.0006', 'epoch': '2.503'}
+
 21%|██        | 180/864 [05:26<14:58,  1.31s/it]
 21%|██        | 181/864 [05:27<14:56,  1.31s/it]
 21%|██        | 182/864 [05:28<14:55,  1.31s/it]
 21%|██        | 183/864 [05:30<14:54,  1.31s/it]
 21%|██▏       | 184/864 [05:31<14:53,  1.31s/it]
 21%|██▏       | 185/864 [05:32<14:52,  1.31s/it]
 22%|██▏       | 186/864 [05:34<14:51,  1.31s/it]
 22%|██▏       | 187/864 [05:35<14:49,  1.31s/it]
 22%|██▏       | 188/864 [05:36<14:47,  1.31s/it]
 22%|██▏       | 189/864 [05:38<14:46,  1.31s/it]
 22%|██▏       | 190/864 [05:39<14:44,  1.31s/it]
                                                 
{'loss': '5.707', 'grad_norm': '0.6171', 'learning_rate': '0.0006', 'epoch': '2.642'}
+
 22%|██▏       | 190/864 [05:39<14:44,  1.31s/it]
 22%|██▏       | 191/864 [05:40<14:43,  1.31s/it]
 22%|██▏       | 192/864 [05:41<14:42,  1.31s/it]
 22%|██▏       | 193/864 [05:43<14:41,  1.31s/it]
 22%|██▏       | 194/864 [05:44<14:39,  1.31s/it]
 23%|██▎       | 195/864 [05:45<14:39,  1.31s/it]
 23%|██▎       | 196/864 [05:47<14:37,  1.31s/it]
 23%|██▎       | 197/864 [05:48<14:35,  1.31s/it]
 23%|██▎       | 198/864 [05:49<14:34,  1.31s/it]
 23%|██▎       | 199/864 [05:51<14:34,  1.31s/it]
 23%|██▎       | 200/864 [05:52<14:32,  1.31s/it]
                                                 
{'loss': '5.722', 'grad_norm': '0.508', 'learning_rate': '0.0006', 'epoch': '2.782'}
+
 23%|██▎       | 200/864 [05:52<14:32,  1.31s/it]
 23%|██▎       | 201/864 [05:53<14:30,  1.31s/it]
 23%|██▎       | 202/864 [05:55<14:29,  1.31s/it]
 23%|██▎       | 203/864 [05:56<14:28,  1.31s/it]
 24%|██▎       | 204/864 [05:57<14:26,  1.31s/it]
 24%|██▎       | 205/864 [05:59<14:25,  1.31s/it]
 24%|██▍       | 206/864 [06:00<14:23,  1.31s/it]
 24%|██▍       | 207/864 [06:01<14:22,  1.31s/it]
 24%|██▍       | 208/864 [06:02<14:21,  1.31s/it]
 24%|██▍       | 209/864 [06:04<14:20,  1.31s/it]
 24%|██▍       | 210/864 [06:05<14:19,  1.31s/it]
                                                 
{'loss': '5.599', 'grad_norm': '0.225', 'learning_rate': '0.0006', 'epoch': '2.921'}
+
 24%|██▍       | 210/864 [06:05<14:19,  1.31s/it]
 24%|██▍       | 211/864 [06:06<14:18,  1.31s/it]
 25%|██▍       | 212/864 [06:08<14:16,  1.31s/it]
 25%|██▍       | 213/864 [06:09<14:14,  1.31s/it]
 25%|██▍       | 214/864 [06:10<14:14,  1.31s/it]
 25%|██▍       | 215/864 [06:12<14:12,  1.31s/it]
 25%|██▌       | 216/864 [06:13<12:36,  1.17s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:02, 14.44it/s]
+
 10%|█         | 4/39 [00:00<00:03,  9.09it/s]
+
 15%|█▌        | 6/39 [00:00<00:04,  8.14it/s]
+
 18%|█▊        | 7/39 [00:00<00:04,  7.90it/s]
+
 21%|██        | 8/39 [00:00<00:04,  7.72it/s]
+
 23%|██▎       | 9/39 [00:01<00:03,  7.58it/s]
+
 26%|██▌       | 10/39 [00:01<00:03,  7.47it/s]
+
 28%|██▊       | 11/39 [00:01<00:03,  7.40it/s]
+
 31%|███       | 12/39 [00:01<00:03,  7.35it/s]
+
 33%|███▎      | 13/39 [00:01<00:03,  7.32it/s]
+
 36%|███▌      | 14/39 [00:01<00:03,  7.30it/s]
+
 38%|███▊      | 15/39 [00:01<00:03,  7.28it/s]
+
 41%|████      | 16/39 [00:02<00:03,  7.27it/s]
+
 44%|████▎     | 17/39 [00:02<00:03,  7.26it/s]
+
 46%|████▌     | 18/39 [00:02<00:02,  7.25it/s]
+
 49%|████▊     | 19/39 [00:02<00:02,  7.25it/s]
+
 51%|█████▏    | 20/39 [00:02<00:02,  7.25it/s]
+
 54%|█████▍    | 21/39 [00:02<00:02,  7.25it/s]
+
 56%|█████▋    | 22/39 [00:02<00:02,  7.25it/s]
+
 59%|█████▉    | 23/39 [00:03<00:02,  7.24it/s]
+
 62%|██████▏   | 24/39 [00:03<00:02,  7.24it/s]
+
 64%|██████▍   | 25/39 [00:03<00:01,  7.24it/s]
+
 67%|██████▋   | 26/39 [00:03<00:01,  7.24it/s]
+
 69%|██████▉   | 27/39 [00:03<00:01,  7.24it/s]
+
 72%|███████▏  | 28/39 [00:03<00:01,  7.24it/s]
+
 74%|███████▍  | 29/39 [00:03<00:01,  7.25it/s]
+
 77%|███████▋  | 30/39 [00:04<00:01,  7.25it/s]
+
 79%|███████▉  | 31/39 [00:04<00:01,  7.25it/s]
+
 82%|████████▏ | 32/39 [00:04<00:00,  7.25it/s]
+
 85%|████████▍ | 33/39 [00:04<00:00,  7.25it/s]
+
 87%|████████▋ | 34/39 [00:04<00:00,  7.25it/s]
+
 90%|████████▉ | 35/39 [00:04<00:00,  7.26it/s]
+
 92%|█████████▏| 36/39 [00:04<00:00,  7.26it/s]
+
 95%|█████████▍| 37/39 [00:04<00:00,  7.26it/s]
+
 97%|█████████▋| 38/39 [00:05<00:00,  7.26it/s]
+
100%|██████████| 39/39 [00:05<00:00,  7.26it/s]
                                                 
+
                                               
{'eval_loss': '5.533', 'eval_runtime': '5.439', 'eval_samples_per_second': '448.8', 'eval_steps_per_second': '7.17', 'epoch': '3'}
+
 25%|██▌       | 216/864 [06:18<12:36,  1.17s/it]
+
100%|██████████| 39/39 [00:05<00:00,  7.26it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0006/checkpoint-216
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0006/checkpoint-216/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0006/checkpoint-216/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:01<00:00,  1.73s/it]
Writing model shards: 100%|██████████| 1/1 [00:01<00:00,  1.73s/it]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0006/checkpoint-216/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0006/checkpoint-216/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0006/checkpoint-216/tokenizer_config.json
+[eval] epoch 3.0 step 216 loss 5.5330 (best 5.5330 @ ep 3.0, 0/10000 since best)
+
+
                                               
 25%|██▌       | 217/864 [06:25<48:41,  4.51s/it]
 25%|██▌       | 218/864 [06:26<38:15,  3.55s/it]
 25%|██▌       | 219/864 [06:27<30:58,  2.88s/it]
 25%|██▌       | 220/864 [06:29<25:52,  2.41s/it]
                                                 
{'loss': '5.476', 'grad_norm': '0.4988', 'learning_rate': '0.0006', 'epoch': '3.056'}
+
 25%|██▌       | 220/864 [06:29<25:52,  2.41s/it]
 26%|██▌       | 221/864 [06:30<22:19,  2.08s/it]
 26%|██▌       | 222/864 [06:31<19:48,  1.85s/it]
 26%|██▌       | 223/864 [06:33<18:03,  1.69s/it]
 26%|██▌       | 224/864 [06:34<16:49,  1.58s/it]
 26%|██▌       | 225/864 [06:35<15:58,  1.50s/it]
 26%|██▌       | 226/864 [06:37<15:20,  1.44s/it]
 26%|██▋       | 227/864 [06:38<14:55,  1.41s/it]
 26%|██▋       | 228/864 [06:39<14:36,  1.38s/it]
 27%|██▋       | 229/864 [06:41<14:22,  1.36s/it]
 27%|██▋       | 230/864 [06:42<14:13,  1.35s/it]
                                                 
{'loss': '5.45', 'grad_norm': '0.474', 'learning_rate': '0.0006', 'epoch': '3.195'}
+
 27%|██▋       | 230/864 [06:42<14:13,  1.35s/it]
 27%|██▋       | 231/864 [06:43<14:06,  1.34s/it]
 27%|██▋       | 232/864 [06:45<14:01,  1.33s/it]
 27%|██▋       | 233/864 [06:46<13:56,  1.33s/it]
 27%|██▋       | 234/864 [06:47<13:52,  1.32s/it]
 27%|██▋       | 235/864 [06:48<13:50,  1.32s/it]
 27%|██▋       | 236/864 [06:50<13:47,  1.32s/it]
 27%|██▋       | 237/864 [06:51<13:44,  1.32s/it]
 28%|██▊       | 238/864 [06:52<13:42,  1.31s/it]
 28%|██▊       | 239/864 [06:54<13:40,  1.31s/it]
 28%|██▊       | 240/864 [06:55<13:39,  1.31s/it]
                                                 
{'loss': '5.35', 'grad_norm': '0.2969', 'learning_rate': '0.0006', 'epoch': '3.335'}
+
 28%|██▊       | 240/864 [06:55<13:39,  1.31s/it]
 28%|██▊       | 241/864 [06:56<13:38,  1.31s/it]
 28%|██▊       | 242/864 [06:58<13:37,  1.31s/it]
 28%|██▊       | 243/864 [06:59<13:36,  1.31s/it]
 28%|██▊       | 244/864 [07:00<13:34,  1.31s/it]
 28%|██▊       | 245/864 [07:02<13:33,  1.31s/it]
 28%|██▊       | 246/864 [07:03<13:31,  1.31s/it]
 29%|██▊       | 247/864 [07:04<13:30,  1.31s/it]
 29%|██▊       | 248/864 [07:06<13:28,  1.31s/it]
 29%|██▉       | 249/864 [07:07<13:27,  1.31s/it]
 29%|██▉       | 250/864 [07:08<13:26,  1.31s/it]
                                                 
{'loss': '5.306', 'grad_norm': '0.5025', 'learning_rate': '0.0006', 'epoch': '3.475'}
+
 29%|██▉       | 250/864 [07:08<13:26,  1.31s/it]
 29%|██▉       | 251/864 [07:09<13:24,  1.31s/it]
 29%|██▉       | 252/864 [07:11<13:23,  1.31s/it]
 29%|██▉       | 253/864 [07:12<13:22,  1.31s/it]
 29%|██▉       | 254/864 [07:13<13:20,  1.31s/it]
 30%|██▉       | 255/864 [07:15<13:19,  1.31s/it]
 30%|██▉       | 256/864 [07:16<13:17,  1.31s/it]
 30%|██▉       | 257/864 [07:17<13:16,  1.31s/it]
 30%|██▉       | 258/864 [07:19<13:15,  1.31s/it]
 30%|██▉       | 259/864 [07:20<13:13,  1.31s/it]
 30%|███       | 260/864 [07:21<13:12,  1.31s/it]
                                                 
{'loss': '5.252', 'grad_norm': '0.3088', 'learning_rate': '0.0006', 'epoch': '3.614'}
+
 30%|███       | 260/864 [07:21<13:12,  1.31s/it]
 30%|███       | 261/864 [07:23<13:10,  1.31s/it]
 30%|███       | 262/864 [07:24<13:09,  1.31s/it]
 30%|███       | 263/864 [07:25<13:09,  1.31s/it]
 31%|███       | 264/864 [07:27<13:07,  1.31s/it]
 31%|███       | 265/864 [07:28<13:06,  1.31s/it]
 31%|███       | 266/864 [07:29<13:04,  1.31s/it]
 31%|███       | 267/864 [07:30<13:03,  1.31s/it]
 31%|███       | 268/864 [07:32<13:02,  1.31s/it]
 31%|███       | 269/864 [07:33<13:00,  1.31s/it]
 31%|███▏      | 270/864 [07:34<12:59,  1.31s/it]
                                                 
{'loss': '5.148', 'grad_norm': '0.2414', 'learning_rate': '0.0006', 'epoch': '3.754'}
+
 31%|███▏      | 270/864 [07:34<12:59,  1.31s/it]
 31%|███▏      | 271/864 [07:36<12:59,  1.31s/it]
 31%|███▏      | 272/864 [07:37<12:57,  1.31s/it]
 32%|███▏      | 273/864 [07:38<12:56,  1.31s/it]
 32%|███▏      | 274/864 [07:40<12:55,  1.31s/it]
 32%|███▏      | 275/864 [07:41<12:53,  1.31s/it]
 32%|███▏      | 276/864 [07:42<12:52,  1.31s/it]
 32%|███▏      | 277/864 [07:44<12:51,  1.31s/it]
 32%|███▏      | 278/864 [07:45<12:50,  1.31s/it]
 32%|███▏      | 279/864 [07:46<12:48,  1.31s/it]
 32%|███▏      | 280/864 [07:48<12:46,  1.31s/it]
                                                 
{'loss': '5.088', 'grad_norm': '0.647', 'learning_rate': '0.0006', 'epoch': '3.894'}
+
 32%|███▏      | 280/864 [07:48<12:46,  1.31s/it]
 33%|███▎      | 281/864 [07:49<12:46,  1.31s/it]
 33%|███▎      | 282/864 [07:50<12:44,  1.31s/it]
 33%|███▎      | 283/864 [07:52<12:43,  1.31s/it]
 33%|███▎      | 284/864 [07:53<12:42,  1.32s/it]
 33%|███▎      | 285/864 [07:54<12:41,  1.32s/it]
 33%|███▎      | 286/864 [07:55<12:39,  1.31s/it]
 33%|███▎      | 287/864 [07:57<12:37,  1.31s/it]
 33%|███▎      | 288/864 [07:58<11:12,  1.17s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:02, 14.62it/s]
+
 10%|█         | 4/39 [00:00<00:03,  9.13it/s]
+
 15%|█▌        | 6/39 [00:00<00:04,  8.17it/s]
+
 18%|█▊        | 7/39 [00:00<00:04,  7.91it/s]
+
 21%|██        | 8/39 [00:00<00:04,  7.73it/s]
+
 23%|██▎       | 9/39 [00:01<00:03,  7.60it/s]
+
 26%|██▌       | 10/39 [00:01<00:03,  7.50it/s]
+
 28%|██▊       | 11/39 [00:01<00:03,  7.43it/s]
+
 31%|███       | 12/39 [00:01<00:03,  7.37it/s]
+
 33%|███▎      | 13/39 [00:01<00:03,  7.34it/s]
+
 36%|███▌      | 14/39 [00:01<00:03,  7.31it/s]
+
 38%|███▊      | 15/39 [00:01<00:03,  7.30it/s]
+
 41%|████      | 16/39 [00:02<00:03,  7.27it/s]
+
 44%|████▎     | 17/39 [00:02<00:03,  7.26it/s]
+
 46%|████▌     | 18/39 [00:02<00:02,  7.25it/s]
+
 49%|████▊     | 19/39 [00:02<00:02,  7.25it/s]
+
 51%|█████▏    | 20/39 [00:02<00:02,  7.25it/s]
+
 54%|█████▍    | 21/39 [00:02<00:02,  7.25it/s]
+
 56%|█████▋    | 22/39 [00:02<00:02,  7.25it/s]
+
 59%|█████▉    | 23/39 [00:03<00:02,  7.25it/s]
+
 62%|██████▏   | 24/39 [00:03<00:02,  7.25it/s]
+
 64%|██████▍   | 25/39 [00:03<00:01,  7.26it/s]
+
 67%|██████▋   | 26/39 [00:03<00:01,  7.26it/s]
+
 69%|██████▉   | 27/39 [00:03<00:01,  7.26it/s]
+
 72%|███████▏  | 28/39 [00:03<00:01,  7.26it/s]
+
 74%|███████▍  | 29/39 [00:03<00:01,  7.26it/s]
+
 77%|███████▋  | 30/39 [00:03<00:01,  7.26it/s]
+
 79%|███████▉  | 31/39 [00:04<00:01,  7.26it/s]
+
 82%|████████▏ | 32/39 [00:04<00:00,  7.26it/s]
+
 85%|████████▍ | 33/39 [00:04<00:00,  7.26it/s]
+
 87%|████████▋ | 34/39 [00:04<00:00,  7.26it/s]
+
 90%|████████▉ | 35/39 [00:04<00:00,  7.26it/s]
+
 92%|█████████▏| 36/39 [00:04<00:00,  7.26it/s]
+
 95%|█████████▍| 37/39 [00:04<00:00,  7.25it/s]
+
 97%|█████████▋| 38/39 [00:05<00:00,  7.25it/s]
+
100%|██████████| 39/39 [00:05<00:00,  7.25it/s]
                                                 
+
                                               
{'eval_loss': '5.173', 'eval_runtime': '5.451', 'eval_samples_per_second': '447.8', 'eval_steps_per_second': '7.154', 'epoch': '4'}
+
 33%|███▎      | 288/864 [08:03<11:12,  1.17s/it]
+
100%|██████████| 39/39 [00:05<00:00,  7.25it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0006/checkpoint-288
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0006/checkpoint-288/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0006/checkpoint-288/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:01<00:00,  1.58s/it]
Writing model shards: 100%|██████████| 1/1 [00:01<00:00,  1.58s/it]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0006/checkpoint-288/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0006/checkpoint-288/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0006/checkpoint-288/tokenizer_config.json
+[eval] epoch 4.0 step 288 loss 5.1726 (best 5.1726 @ ep 4.0, 0/10000 since best)
+
+
                                               
 33%|███▎      | 289/864 [08:11<46:27,  4.85s/it]
 34%|███▎      | 290/864 [08:12<36:12,  3.79s/it]
                                                 
{'loss': '5.07', 'grad_norm': '0.6638', 'learning_rate': '0.0006', 'epoch': '4.028'}
+
 34%|███▎      | 290/864 [08:12<36:12,  3.79s/it]
 34%|███▎      | 291/864 [08:14<29:03,  3.04s/it]
 34%|███▍      | 292/864 [08:15<24:03,  2.52s/it]
 34%|███▍      | 293/864 [08:16<20:33,  2.16s/it]
 34%|███▍      | 294/864 [08:18<18:06,  1.91s/it]
 34%|███▍      | 295/864 [08:19<16:22,  1.73s/it]
 34%|███▍      | 296/864 [08:20<15:10,  1.60s/it]
 34%|███▍      | 297/864 [08:22<14:19,  1.52s/it]
 34%|███▍      | 298/864 [08:23<13:43,  1.45s/it]
 35%|███▍      | 299/864 [08:24<13:17,  1.41s/it]
 35%|███▍      | 300/864 [08:25<12:59,  1.38s/it]
                                                 
{'loss': '4.961', 'grad_norm': '0.2546', 'learning_rate': '0.0006', 'epoch': '4.168'}
+
 35%|███▍      | 300/864 [08:25<12:59,  1.38s/it]
 35%|███▍      | 301/864 [08:27<12:45,  1.36s/it]
 35%|███▍      | 302/864 [08:28<12:36,  1.35s/it]
 35%|███▌      | 303/864 [08:29<12:29,  1.34s/it]
 35%|███▌      | 304/864 [08:31<12:23,  1.33s/it]
 35%|███▌      | 305/864 [08:32<12:19,  1.32s/it]
 35%|███▌      | 306/864 [08:33<12:16,  1.32s/it]
 36%|███▌      | 307/864 [08:35<12:13,  1.32s/it]
 36%|███▌      | 308/864 [08:36<12:11,  1.32s/it]
 36%|███▌      | 309/864 [08:37<12:09,  1.31s/it]
 36%|███▌      | 310/864 [08:39<12:07,  1.31s/it]
                                                 
{'loss': '4.86', 'grad_norm': '0.5296', 'learning_rate': '0.0006', 'epoch': '4.307'}
+
 36%|███▌      | 310/864 [08:39<12:07,  1.31s/it]
 36%|███▌      | 311/864 [08:40<12:05,  1.31s/it]
 36%|███▌      | 312/864 [08:41<12:04,  1.31s/it]
 36%|███▌      | 313/864 [08:42<12:02,  1.31s/it]
 36%|███▋      | 314/864 [08:44<12:01,  1.31s/it]
 36%|███▋      | 315/864 [08:45<11:59,  1.31s/it]
 37%|███▋      | 316/864 [08:46<11:58,  1.31s/it]
 37%|███▋      | 317/864 [08:48<11:57,  1.31s/it]
 37%|███▋      | 318/864 [08:49<11:56,  1.31s/it]
 37%|███▋      | 319/864 [08:50<11:55,  1.31s/it]
 37%|███▋      | 320/864 [08:52<12:48,  1.41s/it]
                                                 
{'loss': '4.827', 'grad_norm': '0.5386', 'learning_rate': '0.0006', 'epoch': '4.447'}
+
 37%|███▋      | 320/864 [08:52<12:48,  1.41s/it]
 37%|███▋      | 321/864 [08:53<12:30,  1.38s/it]
 37%|███▋      | 322/864 [08:55<12:17,  1.36s/it]
 37%|███▋      | 323/864 [08:56<12:08,  1.35s/it]
 38%|███▊      | 324/864 [08:57<12:01,  1.34s/it]
 38%|███▊      | 325/864 [08:59<11:56,  1.33s/it]
 38%|███▊      | 326/864 [09:00<11:53,  1.33s/it]
 38%|███▊      | 327/864 [09:01<11:50,  1.32s/it]
 38%|███▊      | 328/864 [09:03<11:47,  1.32s/it]
 38%|███▊      | 329/864 [09:04<11:44,  1.32s/it]
 38%|███▊      | 330/864 [09:05<11:43,  1.32s/it]
                                                 
{'loss': '4.808', 'grad_norm': '0.422', 'learning_rate': '0.0006', 'epoch': '4.586'}
+
 38%|███▊      | 330/864 [09:05<11:43,  1.32s/it]
 38%|███▊      | 331/864 [09:06<11:42,  1.32s/it]
 38%|███▊      | 332/864 [09:08<11:40,  1.32s/it]
 39%|███▊      | 333/864 [09:09<11:38,  1.32s/it]
 39%|███▊      | 334/864 [09:10<11:36,  1.31s/it]
 39%|███▉      | 335/864 [09:12<11:35,  1.31s/it]
 39%|███▉      | 336/864 [09:13<11:33,  1.31s/it]
 39%|███▉      | 337/864 [09:14<11:32,  1.31s/it]
 39%|███▉      | 338/864 [09:16<12:27,  1.42s/it]
 39%|███▉      | 339/864 [09:18<13:06,  1.50s/it]
 39%|███▉      | 340/864 [09:19<12:36,  1.44s/it]
                                                 
{'loss': '4.726', 'grad_norm': '0.4214', 'learning_rate': '0.0006', 'epoch': '4.726'}
+
 39%|███▉      | 340/864 [09:19<12:36,  1.44s/it]
 39%|███▉      | 341/864 [09:20<12:14,  1.40s/it]
 40%|███▉      | 342/864 [09:22<11:58,  1.38s/it]
 40%|███▉      | 343/864 [09:23<11:47,  1.36s/it]
 40%|███▉      | 344/864 [09:24<11:38,  1.34s/it]
 40%|███▉      | 345/864 [09:26<11:32,  1.33s/it]
 40%|████      | 346/864 [09:27<11:27,  1.33s/it]
 40%|████      | 347/864 [09:28<11:24,  1.32s/it]
 40%|████      | 348/864 [09:30<11:21,  1.32s/it]
 40%|████      | 349/864 [09:31<11:18,  1.32s/it]
 41%|████      | 350/864 [09:32<11:17,  1.32s/it]
                                                 
{'loss': '4.646', 'grad_norm': '0.2501', 'learning_rate': '0.0006', 'epoch': '4.866'}
+
 41%|████      | 350/864 [09:32<11:17,  1.32s/it]
 41%|████      | 351/864 [09:33<11:14,  1.32s/it]
 41%|████      | 352/864 [09:35<11:13,  1.31s/it]
 41%|████      | 353/864 [09:36<11:11,  1.31s/it]
 41%|████      | 354/864 [09:37<11:09,  1.31s/it]
 41%|████      | 355/864 [09:39<11:08,  1.31s/it]
 41%|████      | 356/864 [09:40<11:06,  1.31s/it]
 41%|████▏     | 357/864 [09:41<11:05,  1.31s/it]
 41%|████▏     | 358/864 [09:43<11:04,  1.31s/it]
 42%|████▏     | 359/864 [09:44<11:03,  1.31s/it]
 42%|████▏     | 360/864 [09:45<09:48,  1.17s/it]
                                                 
{'loss': '4.607', 'grad_norm': '0.4031', 'learning_rate': '0.0006', 'epoch': '5'}
+
 42%|████▏     | 360/864 [09:45<09:48,  1.17s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:02, 14.88it/s]
+
 10%|█         | 4/39 [00:00<00:03,  9.18it/s]
+
 15%|█▌        | 6/39 [00:00<00:04,  8.14it/s]
+
 18%|█▊        | 7/39 [00:00<00:04,  7.95it/s]
+
 21%|██        | 8/39 [00:00<00:03,  7.75it/s]
+
 23%|██▎       | 9/39 [00:01<00:03,  7.61it/s]
+
 26%|██▌       | 10/39 [00:01<00:03,  7.51it/s]
+
 28%|██▊       | 11/39 [00:01<00:03,  7.44it/s]
+
 31%|███       | 12/39 [00:01<00:03,  7.38it/s]
+
 33%|███▎      | 13/39 [00:01<00:03,  7.35it/s]
+
 36%|███▌      | 14/39 [00:01<00:03,  7.31it/s]
+
 38%|███▊      | 15/39 [00:01<00:03,  7.30it/s]
+
 41%|████      | 16/39 [00:02<00:03,  7.29it/s]
+
 44%|████▎     | 17/39 [00:02<00:03,  7.26it/s]
+
 46%|████▌     | 18/39 [00:02<00:02,  7.25it/s]
+
 49%|████▊     | 19/39 [00:02<00:02,  7.25it/s]
+
 51%|█████▏    | 20/39 [00:02<00:02,  7.24it/s]
+
 54%|█████▍    | 21/39 [00:02<00:02,  7.24it/s]
+
 56%|█████▋    | 22/39 [00:02<00:02,  7.24it/s]
+
 59%|█████▉    | 23/39 [00:03<00:02,  7.24it/s]
+
 62%|██████▏   | 24/39 [00:03<00:02,  7.23it/s]
+
 64%|██████▍   | 25/39 [00:03<00:01,  7.23it/s]
+
 67%|██████▋   | 26/39 [00:03<00:01,  7.23it/s]
+
 69%|██████▉   | 27/39 [00:03<00:01,  7.23it/s]
+
 72%|███████▏  | 28/39 [00:03<00:01,  7.22it/s]
+
 74%|███████▍  | 29/39 [00:03<00:01,  7.23it/s]
+
 77%|███████▋  | 30/39 [00:04<00:01,  7.22it/s]
+
 79%|███████▉  | 31/39 [00:04<00:01,  7.23it/s]
+
 82%|████████▏ | 32/39 [00:04<00:00,  7.23it/s]
+
 85%|████████▍ | 33/39 [00:04<00:00,  7.23it/s]
+
 87%|████████▋ | 34/39 [00:04<00:00,  7.24it/s]
+
 90%|████████▉ | 35/39 [00:04<00:00,  7.24it/s]
+
 92%|█████████▏| 36/39 [00:04<00:00,  7.24it/s]
+
 95%|█████████▍| 37/39 [00:04<00:00,  7.24it/s]
+
 97%|█████████▋| 38/39 [00:05<00:00,  7.25it/s]
+
100%|██████████| 39/39 [00:05<00:00,  7.25it/s]
                                                 
+
                                               
{'eval_loss': '4.692', 'eval_runtime': '5.456', 'eval_samples_per_second': '447.4', 'eval_steps_per_second': '7.148', 'epoch': '5'}
+
 42%|████▏     | 360/864 [09:50<09:48,  1.17s/it]
+
100%|██████████| 39/39 [00:05<00:00,  7.25it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0006/checkpoint-360
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0006/checkpoint-360/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0006/checkpoint-360/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:01<00:00,  1.83s/it]
Writing model shards: 100%|██████████| 1/1 [00:01<00:00,  1.83s/it]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0006/checkpoint-360/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0006/checkpoint-360/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0006/checkpoint-360/tokenizer_config.json
+[eval] epoch 5.0 step 360 loss 4.6915 (best 4.6915 @ ep 5.0, 0/10000 since best)
+
+
                                               
 42%|████▏     | 361/864 [09:58<39:07,  4.67s/it]
 42%|████▏     | 362/864 [09:59<30:37,  3.66s/it]
 42%|████▏     | 363/864 [10:00<24:41,  2.96s/it]
 42%|████▏     | 364/864 [10:02<20:31,  2.46s/it]
 42%|████▏     | 365/864 [10:03<17:36,  2.12s/it]
 42%|████▏     | 366/864 [10:04<15:35,  1.88s/it]
 42%|████▏     | 367/864 [10:06<14:09,  1.71s/it]
 43%|████▎     | 368/864 [10:07<13:08,  1.59s/it]
 43%|████▎     | 369/864 [10:08<12:26,  1.51s/it]
 43%|████▎     | 370/864 [10:09<11:56,  1.45s/it]
                                                 
{'loss': '4.494', 'grad_norm': '0.3573', 'learning_rate': '0.0006', 'epoch': '5.14'}
+
 43%|████▎     | 370/864 [10:09<11:56,  1.45s/it]
 43%|████▎     | 371/864 [10:11<11:34,  1.41s/it]
 43%|████▎     | 372/864 [10:12<11:18,  1.38s/it]
 43%|████▎     | 373/864 [10:13<11:07,  1.36s/it]
 43%|████▎     | 374/864 [10:15<10:59,  1.35s/it]
 43%|████▎     | 375/864 [10:16<10:53,  1.34s/it]
 44%|████▎     | 376/864 [10:17<10:48,  1.33s/it]
 44%|████▎     | 377/864 [10:19<10:44,  1.32s/it]
 44%|████▍     | 378/864 [10:20<10:42,  1.32s/it]
 44%|████▍     | 379/864 [10:21<10:39,  1.32s/it]
 44%|████▍     | 380/864 [10:23<10:37,  1.32s/it]
                                                 
{'loss': '4.478', 'grad_norm': '0.5797', 'learning_rate': '0.0006', 'epoch': '5.279'}
+
 44%|████▍     | 380/864 [10:23<10:37,  1.32s/it]
 44%|████▍     | 381/864 [10:24<10:35,  1.32s/it]
 44%|████▍     | 382/864 [10:25<10:34,  1.32s/it]
 44%|████▍     | 383/864 [10:27<10:32,  1.31s/it]
 44%|████▍     | 384/864 [10:28<10:31,  1.31s/it]
 45%|████▍     | 385/864 [10:29<10:29,  1.31s/it]
 45%|████▍     | 386/864 [10:30<10:27,  1.31s/it]
 45%|████▍     | 387/864 [10:32<10:26,  1.31s/it]
 45%|████▍     | 388/864 [10:33<10:25,  1.31s/it]
 45%|████▌     | 389/864 [10:34<10:24,  1.31s/it]
 45%|████▌     | 390/864 [10:36<10:22,  1.31s/it]
                                                 
{'loss': '4.436', 'grad_norm': '0.4042', 'learning_rate': '0.0006', 'epoch': '5.419'}
+
 45%|████▌     | 390/864 [10:36<10:22,  1.31s/it]
 45%|████▌     | 391/864 [10:37<10:21,  1.31s/it]
 45%|████▌     | 392/864 [10:38<10:19,  1.31s/it]
 45%|████▌     | 393/864 [10:40<10:18,  1.31s/it]
 46%|████▌     | 394/864 [10:41<10:17,  1.31s/it]
 46%|████▌     | 395/864 [10:42<10:15,  1.31s/it]
 46%|████▌     | 396/864 [10:44<10:14,  1.31s/it]
 46%|████▌     | 397/864 [10:45<10:13,  1.31s/it]
 46%|████▌     | 398/864 [10:46<10:11,  1.31s/it]
 46%|████▌     | 399/864 [10:48<10:10,  1.31s/it]
 46%|████▋     | 400/864 [10:49<10:08,  1.31s/it]
                                                 
{'loss': '4.361', 'grad_norm': '0.2034', 'learning_rate': '0.0006', 'epoch': '5.558'}
+
 46%|████▋     | 400/864 [10:49<10:08,  1.31s/it]
 46%|████▋     | 401/864 [10:50<10:07,  1.31s/it]
 47%|████▋     | 402/864 [10:51<10:06,  1.31s/it]
 47%|████▋     | 403/864 [10:53<10:05,  1.31s/it]
 47%|████▋     | 404/864 [10:54<10:04,  1.31s/it]
 47%|████▋     | 405/864 [10:55<10:03,  1.31s/it]
 47%|████▋     | 406/864 [10:57<10:01,  1.31s/it]
 47%|████▋     | 407/864 [10:58<10:00,  1.31s/it]
 47%|████▋     | 408/864 [10:59<09:58,  1.31s/it]
 47%|████▋     | 409/864 [11:01<09:57,  1.31s/it]
 47%|████▋     | 410/864 [11:02<09:56,  1.31s/it]
                                                 
{'loss': '4.318', 'grad_norm': '0.2173', 'learning_rate': '0.0006', 'epoch': '5.698'}
+
 47%|████▋     | 410/864 [11:02<09:56,  1.31s/it]
 48%|████▊     | 411/864 [11:03<09:55,  1.31s/it]
 48%|████▊     | 412/864 [11:05<09:53,  1.31s/it]
 48%|████▊     | 413/864 [11:06<09:52,  1.31s/it]
 48%|████▊     | 414/864 [11:07<09:50,  1.31s/it]
 48%|████▊     | 415/864 [11:09<09:49,  1.31s/it]
 48%|████▊     | 416/864 [11:10<09:48,  1.31s/it]
 48%|████▊     | 417/864 [11:11<09:47,  1.31s/it]
 48%|████▊     | 418/864 [11:12<09:46,  1.31s/it]
 48%|████▊     | 419/864 [11:14<09:44,  1.31s/it]
 49%|████▊     | 420/864 [11:15<09:43,  1.31s/it]
                                                 
{'loss': '4.286', 'grad_norm': '0.4536', 'learning_rate': '0.0006', 'epoch': '5.838'}
+
 49%|████▊     | 420/864 [11:15<09:43,  1.31s/it]
 49%|████▊     | 421/864 [11:16<09:42,  1.31s/it]
 49%|████▉     | 422/864 [11:18<09:40,  1.31s/it]
 49%|████▉     | 423/864 [11:19<09:39,  1.31s/it]
 49%|████▉     | 424/864 [11:20<09:37,  1.31s/it]
 49%|████▉     | 425/864 [11:22<09:36,  1.31s/it]
 49%|████▉     | 426/864 [11:23<09:35,  1.31s/it]
 49%|████▉     | 427/864 [11:24<09:33,  1.31s/it]
 50%|████▉     | 428/864 [11:26<09:32,  1.31s/it]
 50%|████▉     | 429/864 [11:27<09:31,  1.31s/it]
 50%|████▉     | 430/864 [11:28<09:29,  1.31s/it]
                                                 
{'loss': '4.293', 'grad_norm': '0.4424', 'learning_rate': '0.0006', 'epoch': '5.977'}
+
 50%|████▉     | 430/864 [11:28<09:29,  1.31s/it]
 50%|████▉     | 431/864 [11:30<09:28,  1.31s/it]
 50%|█████     | 432/864 [11:30<08:24,  1.17s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:02, 14.71it/s]
+
 10%|█         | 4/39 [00:00<00:03,  9.14it/s]
+
 15%|█▌        | 6/39 [00:00<00:04,  8.17it/s]
+
 18%|█▊        | 7/39 [00:00<00:04,  7.92it/s]
+
 21%|██        | 8/39 [00:00<00:04,  7.74it/s]
+
 23%|██▎       | 9/39 [00:01<00:03,  7.59it/s]
+
 26%|██▌       | 10/39 [00:01<00:03,  7.49it/s]
+
 28%|██▊       | 11/39 [00:01<00:03,  7.42it/s]
+
 31%|███       | 12/39 [00:01<00:03,  7.36it/s]
+
 33%|███▎      | 13/39 [00:01<00:03,  7.32it/s]
+
 36%|███▌      | 14/39 [00:01<00:03,  7.30it/s]
+
 38%|███▊      | 15/39 [00:01<00:03,  7.29it/s]
+
 41%|████      | 16/39 [00:02<00:03,  7.28it/s]
+
 44%|████▎     | 17/39 [00:02<00:03,  7.27it/s]
+
 46%|████▌     | 18/39 [00:02<00:02,  7.26it/s]
+
 49%|████▊     | 19/39 [00:02<00:02,  7.26it/s]
+
 51%|█████▏    | 20/39 [00:02<00:02,  7.26it/s]
+
 54%|█████▍    | 21/39 [00:02<00:02,  7.25it/s]
+
 56%|█████▋    | 22/39 [00:02<00:02,  7.25it/s]
+
 59%|█████▉    | 23/39 [00:03<00:02,  7.25it/s]
+
 62%|██████▏   | 24/39 [00:03<00:02,  7.25it/s]
+
 64%|██████▍   | 25/39 [00:03<00:01,  7.26it/s]
+
 67%|██████▋   | 26/39 [00:03<00:01,  7.25it/s]
+
 69%|██████▉   | 27/39 [00:03<00:01,  7.25it/s]
+
 72%|███████▏  | 28/39 [00:03<00:01,  7.25it/s]
+
 74%|███████▍  | 29/39 [00:03<00:01,  7.24it/s]
+
 77%|███████▋  | 30/39 [00:04<00:01,  7.24it/s]
+
 79%|███████▉  | 31/39 [00:04<00:01,  7.24it/s]
+
 82%|████████▏ | 32/39 [00:04<00:00,  7.25it/s]
+
 85%|████████▍ | 33/39 [00:04<00:00,  7.25it/s]
+
 87%|████████▋ | 34/39 [00:04<00:00,  7.25it/s]
+
 90%|████████▉ | 35/39 [00:04<00:00,  7.25it/s]
+
 92%|█████████▏| 36/39 [00:04<00:00,  7.25it/s]
+
 95%|█████████▍| 37/39 [00:04<00:00,  7.25it/s]
+
 97%|█████████▋| 38/39 [00:05<00:00,  7.26it/s]
+
100%|██████████| 39/39 [00:05<00:00,  7.26it/s]
                                                 
+
                                               
{'eval_loss': '4.449', 'eval_runtime': '5.435', 'eval_samples_per_second': '449.1', 'eval_steps_per_second': '7.176', 'epoch': '6'}
+
 50%|█████     | 432/864 [11:36<08:24,  1.17s/it]
+
100%|██████████| 39/39 [00:05<00:00,  7.26it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0006/checkpoint-432
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0006/checkpoint-432/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0006/checkpoint-432/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:01<00:00,  1.75s/it]
Writing model shards: 100%|██████████| 1/1 [00:01<00:00,  1.75s/it]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0006/checkpoint-432/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0006/checkpoint-432/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0006/checkpoint-432/tokenizer_config.json
+[eval] epoch 6.0 step 432 loss 4.4488 (best 4.4488 @ ep 6.0, 0/10000 since best)
+
+
                                               
 50%|█████     | 433/864 [11:43<33:22,  4.65s/it]
 50%|█████     | 434/864 [11:44<26:07,  3.64s/it]
 50%|█████     | 435/864 [11:46<21:03,  2.94s/it]
 50%|█████     | 436/864 [11:47<17:30,  2.45s/it]
 51%|█████     | 437/864 [11:48<15:01,  2.11s/it]
 51%|█████     | 438/864 [11:50<13:17,  1.87s/it]
 51%|█████     | 439/864 [11:51<12:04,  1.70s/it]
 51%|█████     | 440/864 [11:52<11:12,  1.59s/it]
                                                 
{'loss': '4.254', 'grad_norm': '0.6938', 'learning_rate': '0.0006', 'epoch': '6.112'}
+
 51%|█████     | 440/864 [11:52<11:12,  1.59s/it]
 51%|█████     | 441/864 [11:54<10:36,  1.50s/it]
 51%|█████     | 442/864 [11:55<10:10,  1.45s/it]
 51%|█████▏    | 443/864 [11:56<09:52,  1.41s/it]
 51%|█████▏    | 444/864 [11:58<09:38,  1.38s/it]
 52%|█████▏    | 445/864 [11:59<09:28,  1.36s/it]
 52%|█████▏    | 446/864 [12:00<09:21,  1.34s/it]
 52%|█████▏    | 447/864 [12:02<09:16,  1.33s/it]
 52%|█████▏    | 448/864 [12:03<09:12,  1.33s/it]
 52%|█████▏    | 449/864 [12:04<09:09,  1.32s/it]
 52%|█████▏    | 450/864 [12:05<09:06,  1.32s/it]
                                                 
{'loss': '4.29', 'grad_norm': '0.5019', 'learning_rate': '0.0006', 'epoch': '6.251'}
+
 52%|█████▏    | 450/864 [12:05<09:06,  1.32s/it]
 52%|█████▏    | 451/864 [12:07<09:04,  1.32s/it]
 52%|█████▏    | 452/864 [12:08<09:02,  1.32s/it]
 52%|█████▏    | 453/864 [12:09<09:00,  1.32s/it]
 53%|█████▎    | 454/864 [12:11<08:58,  1.31s/it]
 53%|█████▎    | 455/864 [12:12<08:57,  1.31s/it]
 53%|█████▎    | 456/864 [12:13<08:55,  1.31s/it]
 53%|█████▎    | 457/864 [12:15<08:54,  1.31s/it]
 53%|█████▎    | 458/864 [12:16<08:53,  1.31s/it]
 53%|█████▎    | 459/864 [12:17<08:51,  1.31s/it]
 53%|█████▎    | 460/864 [12:19<08:50,  1.31s/it]
                                                 
{'loss': '4.194', 'grad_norm': '0.2517', 'learning_rate': '0.0006', 'epoch': '6.391'}
+
 53%|█████▎    | 460/864 [12:19<08:50,  1.31s/it]
 53%|█████▎    | 461/864 [12:20<08:48,  1.31s/it]
 53%|█████▎    | 462/864 [12:21<08:47,  1.31s/it]
 54%|█████▎    | 463/864 [12:23<08:46,  1.31s/it]
 54%|█████▎    | 464/864 [12:24<08:44,  1.31s/it]
 54%|█████▍    | 465/864 [12:25<08:43,  1.31s/it]
 54%|█████▍    | 466/864 [12:26<08:42,  1.31s/it]
 54%|█████▍    | 467/864 [12:28<08:41,  1.31s/it]
 54%|█████▍    | 468/864 [12:29<08:39,  1.31s/it]
 54%|█████▍    | 469/864 [12:30<08:39,  1.31s/it]
 54%|█████▍    | 470/864 [12:32<08:37,  1.31s/it]
                                                 
{'loss': '4.124', 'grad_norm': '0.1475', 'learning_rate': '0.0006', 'epoch': '6.531'}
+
 54%|█████▍    | 470/864 [12:32<08:37,  1.31s/it]
 55%|█████▍    | 471/864 [12:33<08:36,  1.31s/it]
 55%|█████▍    | 472/864 [12:34<08:34,  1.31s/it]
 55%|█████▍    | 473/864 [12:36<08:33,  1.31s/it]
 55%|█████▍    | 474/864 [12:37<08:32,  1.31s/it]
 55%|█████▍    | 475/864 [12:38<08:30,  1.31s/it]
 55%|█████▌    | 476/864 [12:40<08:29,  1.31s/it]
 55%|█████▌    | 477/864 [12:41<08:28,  1.31s/it]
 55%|█████▌    | 478/864 [12:42<08:26,  1.31s/it]
 55%|█████▌    | 479/864 [12:44<08:25,  1.31s/it]
 56%|█████▌    | 480/864 [12:45<08:23,  1.31s/it]
                                                 
{'loss': '4.079', 'grad_norm': '0.1481', 'learning_rate': '0.0006', 'epoch': '6.67'}
+
 56%|█████▌    | 480/864 [12:45<08:23,  1.31s/it]
 56%|█████▌    | 481/864 [12:46<08:22,  1.31s/it]
 56%|█████▌    | 482/864 [12:47<08:21,  1.31s/it]
 56%|█████▌    | 483/864 [12:49<08:19,  1.31s/it]
 56%|█████▌    | 484/864 [12:50<08:18,  1.31s/it]
 56%|█████▌    | 485/864 [12:51<08:17,  1.31s/it]
 56%|█████▋    | 486/864 [12:53<08:16,  1.31s/it]
 56%|█████▋    | 487/864 [12:54<08:15,  1.31s/it]
 56%|█████▋    | 488/864 [12:55<08:13,  1.31s/it]
 57%|█████▋    | 489/864 [12:57<08:12,  1.31s/it]
 57%|█████▋    | 490/864 [12:58<08:10,  1.31s/it]
                                                 
{'loss': '4.063', 'grad_norm': '0.2062', 'learning_rate': '0.0006', 'epoch': '6.81'}
+
 57%|█████▋    | 490/864 [12:58<08:10,  1.31s/it]
 57%|█████▋    | 491/864 [12:59<08:09,  1.31s/it]
 57%|█████▋    | 492/864 [13:01<08:07,  1.31s/it]
 57%|█████▋    | 493/864 [13:02<08:06,  1.31s/it]
 57%|█████▋    | 494/864 [13:03<08:05,  1.31s/it]
 57%|█████▋    | 495/864 [13:05<08:04,  1.31s/it]
 57%|█████▋    | 496/864 [13:06<08:02,  1.31s/it]
 58%|█████▊    | 497/864 [13:07<08:01,  1.31s/it]
 58%|█████▊    | 498/864 [13:08<08:00,  1.31s/it]
 58%|█████▊    | 499/864 [13:10<07:59,  1.31s/it]
 58%|█████▊    | 500/864 [13:11<07:58,  1.31s/it]
                                                 
{'loss': '4.03', 'grad_norm': '0.191', 'learning_rate': '0.0006', 'epoch': '6.949'}
+
 58%|█████▊    | 500/864 [13:11<07:58,  1.31s/it]
 58%|█████▊    | 501/864 [13:12<07:56,  1.31s/it]
 58%|█████▊    | 502/864 [13:14<07:55,  1.31s/it]
 58%|█████▊    | 503/864 [13:15<07:53,  1.31s/it]
 58%|█████▊    | 504/864 [13:16<07:00,  1.17s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:02, 14.50it/s]
+
 10%|█         | 4/39 [00:00<00:03,  9.09it/s]
+
 15%|█▌        | 6/39 [00:00<00:04,  8.14it/s]
+
 18%|█▊        | 7/39 [00:00<00:04,  7.90it/s]
+
 21%|██        | 8/39 [00:00<00:04,  7.72it/s]
+
 23%|██▎       | 9/39 [00:01<00:03,  7.59it/s]
+
 26%|██▌       | 10/39 [00:01<00:03,  7.49it/s]
+
 28%|██▊       | 11/39 [00:01<00:03,  7.42it/s]
+
 31%|███       | 12/39 [00:01<00:03,  7.37it/s]
+
 33%|███▎      | 13/39 [00:01<00:03,  7.34it/s]
+
 36%|███▌      | 14/39 [00:01<00:03,  7.30it/s]
+
 38%|███▊      | 15/39 [00:01<00:03,  7.28it/s]
+
 41%|████      | 16/39 [00:02<00:03,  7.26it/s]
+
 44%|████▎     | 17/39 [00:02<00:03,  7.24it/s]
+
 46%|████▌     | 18/39 [00:02<00:02,  7.24it/s]
+
 49%|████▊     | 19/39 [00:02<00:02,  7.24it/s]
+
 51%|█████▏    | 20/39 [00:02<00:02,  7.25it/s]
+
 54%|█████▍    | 21/39 [00:02<00:02,  7.25it/s]
+
 56%|█████▋    | 22/39 [00:02<00:02,  7.25it/s]
+
 59%|█████▉    | 23/39 [00:03<00:02,  7.25it/s]
+
 62%|██████▏   | 24/39 [00:03<00:02,  7.25it/s]
+
 64%|██████▍   | 25/39 [00:03<00:01,  7.26it/s]
+
 67%|██████▋   | 26/39 [00:03<00:01,  7.26it/s]
+
 69%|██████▉   | 27/39 [00:03<00:01,  7.25it/s]
+
 72%|███████▏  | 28/39 [00:03<00:01,  7.25it/s]
+
 74%|███████▍  | 29/39 [00:03<00:01,  7.25it/s]
+
 77%|███████▋  | 30/39 [00:04<00:01,  7.24it/s]
+
 79%|███████▉  | 31/39 [00:04<00:01,  7.23it/s]
+
 82%|████████▏ | 32/39 [00:04<00:00,  7.24it/s]
+
 85%|████████▍ | 33/39 [00:04<00:00,  7.24it/s]
+
 87%|████████▋ | 34/39 [00:04<00:00,  7.24it/s]
+
 90%|████████▉ | 35/39 [00:04<00:00,  7.24it/s]
+
 92%|█████████▏| 36/39 [00:04<00:00,  7.25it/s]
+
 95%|█████████▍| 37/39 [00:04<00:00,  7.25it/s]
+
 97%|█████████▋| 38/39 [00:05<00:00,  7.25it/s]
+
100%|██████████| 39/39 [00:05<00:00,  7.26it/s]
                                                 
+
                                               
{'eval_loss': '4.217', 'eval_runtime': '5.454', 'eval_samples_per_second': '447.5', 'eval_steps_per_second': '7.15', 'epoch': '7'}
+
 58%|█████▊    | 504/864 [13:21<07:00,  1.17s/it]
+
100%|██████████| 39/39 [00:05<00:00,  7.26it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0006/checkpoint-504
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0006/checkpoint-504/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0006/checkpoint-504/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:01<00:00,  1.74s/it]
Writing model shards: 100%|██████████| 1/1 [00:01<00:00,  1.74s/it]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0006/checkpoint-504/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0006/checkpoint-504/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0006/checkpoint-504/tokenizer_config.json
+[eval] epoch 7.0 step 504 loss 4.2173 (best 4.2173 @ ep 7.0, 0/10000 since best)
+
+
                                               
 58%|█████▊    | 505/864 [13:29<27:51,  4.66s/it]
 59%|█████▊    | 506/864 [13:30<21:47,  3.65s/it]
 59%|█████▊    | 507/864 [13:31<17:33,  2.95s/it]
 59%|█████▉    | 508/864 [13:33<14:35,  2.46s/it]
 59%|█████▉    | 509/864 [13:34<12:31,  2.12s/it]
 59%|█████▉    | 510/864 [13:35<11:03,  1.88s/it]
                                                 
{'loss': '3.961', 'grad_norm': '0.3492', 'learning_rate': '0.0006', 'epoch': '7.084'}
+
 59%|█████▉    | 510/864 [13:35<11:03,  1.88s/it]
 59%|█████▉    | 511/864 [13:37<10:02,  1.71s/it]
 59%|█████▉    | 512/864 [13:38<09:19,  1.59s/it]
 59%|█████▉    | 513/864 [13:39<08:48,  1.51s/it]
 59%|█████▉    | 514/864 [13:40<08:26,  1.45s/it]
 60%|█████▉    | 515/864 [13:42<08:11,  1.41s/it]
 60%|█████▉    | 516/864 [13:43<07:59,  1.38s/it]
 60%|█████▉    | 517/864 [13:44<07:51,  1.36s/it]
 60%|█████▉    | 518/864 [13:46<07:45,  1.35s/it]
 60%|██████    | 519/864 [13:47<07:40,  1.34s/it]
 60%|██████    | 520/864 [13:48<07:36,  1.33s/it]
                                                 
{'loss': '3.938', 'grad_norm': '0.4682', 'learning_rate': '0.0006', 'epoch': '7.223'}
+
 60%|██████    | 520/864 [13:48<07:36,  1.33s/it]
 60%|██████    | 521/864 [13:50<07:34,  1.32s/it]
 60%|██████    | 522/864 [13:51<07:31,  1.32s/it]
 61%|██████    | 523/864 [13:52<07:29,  1.32s/it]
 61%|██████    | 524/864 [13:54<07:27,  1.32s/it]
 61%|██████    | 525/864 [13:55<07:25,  1.32s/it]
 61%|██████    | 526/864 [13:56<07:24,  1.32s/it]
 61%|██████    | 527/864 [13:58<07:23,  1.31s/it]
 61%|██████    | 528/864 [13:59<07:21,  1.31s/it]
 61%|██████    | 529/864 [14:00<07:20,  1.31s/it]
 61%|██████▏   | 530/864 [14:01<07:19,  1.31s/it]
                                                 
{'loss': '3.908', 'grad_norm': '0.209', 'learning_rate': '0.0006', 'epoch': '7.363'}
+
 61%|██████▏   | 530/864 [14:01<07:19,  1.31s/it]
 61%|██████▏   | 531/864 [14:03<07:17,  1.31s/it]
 62%|██████▏   | 532/864 [14:04<07:16,  1.31s/it]
 62%|██████▏   | 533/864 [14:05<07:14,  1.31s/it]
 62%|██████▏   | 534/864 [14:07<07:13,  1.31s/it]
 62%|██████▏   | 535/864 [14:08<07:12,  1.31s/it]
 62%|██████▏   | 536/864 [14:09<07:10,  1.31s/it]
 62%|██████▏   | 537/864 [14:11<07:09,  1.31s/it]
 62%|██████▏   | 538/864 [14:12<07:07,  1.31s/it]
 62%|██████▏   | 539/864 [14:13<07:06,  1.31s/it]
 62%|██████▎   | 540/864 [14:15<07:05,  1.31s/it]
                                                 
{'loss': '3.883', 'grad_norm': '0.1745', 'learning_rate': '0.0006', 'epoch': '7.503'}
+
 62%|██████▎   | 540/864 [14:15<07:05,  1.31s/it]
 63%|██████▎   | 541/864 [14:16<07:04,  1.31s/it]
 63%|██████▎   | 542/864 [14:17<07:02,  1.31s/it]
 63%|██████▎   | 543/864 [14:19<07:01,  1.31s/it]
 63%|██████▎   | 544/864 [14:20<06:59,  1.31s/it]
 63%|██████▎   | 545/864 [14:21<06:59,  1.31s/it]
 63%|██████▎   | 546/864 [14:22<06:57,  1.31s/it]
 63%|██████▎   | 547/864 [14:24<06:55,  1.31s/it]
 63%|██████▎   | 548/864 [14:25<06:54,  1.31s/it]
 64%|██████▎   | 549/864 [14:26<06:53,  1.31s/it]
 64%|██████▎   | 550/864 [14:28<06:52,  1.31s/it]
                                                 
{'loss': '3.868', 'grad_norm': '0.2035', 'learning_rate': '0.0006', 'epoch': '7.642'}
+
 64%|██████▎   | 550/864 [14:28<06:52,  1.31s/it]
 64%|██████▍   | 551/864 [14:29<06:51,  1.31s/it]
 64%|██████▍   | 552/864 [14:30<06:49,  1.31s/it]
 64%|██████▍   | 553/864 [14:32<06:48,  1.31s/it]
 64%|██████▍   | 554/864 [14:33<06:46,  1.31s/it]
 64%|██████▍   | 555/864 [14:34<06:45,  1.31s/it]
 64%|██████▍   | 556/864 [14:36<06:44,  1.31s/it]
 64%|██████▍   | 557/864 [14:37<06:42,  1.31s/it]
 65%|██████▍   | 558/864 [14:38<06:41,  1.31s/it]
 65%|██████▍   | 559/864 [14:40<06:40,  1.31s/it]
 65%|██████▍   | 560/864 [14:41<06:39,  1.31s/it]
                                                 
{'loss': '3.861', 'grad_norm': '0.241', 'learning_rate': '0.0006', 'epoch': '7.782'}
+
 65%|██████▍   | 560/864 [14:41<06:39,  1.31s/it]
 65%|██████▍   | 561/864 [14:42<06:37,  1.31s/it]
 65%|██████▌   | 562/864 [14:43<06:36,  1.31s/it]
 65%|██████▌   | 563/864 [14:45<06:35,  1.31s/it]
 65%|██████▌   | 564/864 [14:46<06:33,  1.31s/it]
 65%|██████▌   | 565/864 [14:47<06:32,  1.31s/it]
 66%|██████▌   | 566/864 [14:49<06:31,  1.31s/it]
 66%|██████▌   | 567/864 [14:50<06:30,  1.31s/it]
 66%|██████▌   | 568/864 [14:51<06:29,  1.31s/it]
 66%|██████▌   | 569/864 [14:53<06:27,  1.32s/it]
 66%|██████▌   | 570/864 [14:54<06:26,  1.31s/it]
                                                 
{'loss': '3.849', 'grad_norm': '0.2271', 'learning_rate': '0.0006', 'epoch': '7.921'}
+
 66%|██████▌   | 570/864 [14:54<06:26,  1.31s/it]
 66%|██████▌   | 571/864 [14:55<06:24,  1.31s/it]
 66%|██████▌   | 572/864 [14:57<06:23,  1.31s/it]
 66%|██████▋   | 573/864 [14:58<06:22,  1.31s/it]
 66%|██████▋   | 574/864 [14:59<06:20,  1.31s/it]
 67%|██████▋   | 575/864 [15:01<06:19,  1.31s/it]
 67%|██████▋   | 576/864 [15:01<05:36,  1.17s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:02, 14.65it/s]
+
 10%|█         | 4/39 [00:00<00:03,  9.06it/s]
+
 15%|█▌        | 6/39 [00:00<00:04,  8.17it/s]
+
 18%|█▊        | 7/39 [00:00<00:04,  7.91it/s]
+
 21%|██        | 8/39 [00:00<00:04,  7.67it/s]
+
 23%|██▎       | 9/39 [00:01<00:03,  7.60it/s]
+
 26%|██▌       | 10/39 [00:01<00:03,  7.49it/s]
+
 28%|██▊       | 11/39 [00:01<00:03,  7.42it/s]
+
 31%|███       | 12/39 [00:01<00:03,  7.32it/s]
+
 33%|███▎      | 13/39 [00:01<00:03,  7.35it/s]
+
 36%|███▌      | 14/39 [00:01<00:03,  7.32it/s]
+
 38%|███▊      | 15/39 [00:01<00:03,  7.30it/s]
+
 41%|████      | 16/39 [00:02<00:03,  7.24it/s]
+
 44%|████▎     | 17/39 [00:02<00:03,  7.29it/s]
+
 46%|████▌     | 18/39 [00:02<00:02,  7.28it/s]
+
 49%|████▊     | 19/39 [00:02<00:02,  7.26it/s]
+
 51%|█████▏    | 20/39 [00:02<00:02,  7.20it/s]
+
 54%|█████▍    | 21/39 [00:02<00:02,  7.27it/s]
+
 56%|█████▋    | 22/39 [00:02<00:02,  7.27it/s]
+
 59%|█████▉    | 23/39 [00:03<00:02,  7.26it/s]
+
 62%|██████▏   | 24/39 [00:03<00:02,  7.26it/s]
+
 64%|██████▍   | 25/39 [00:03<00:01,  7.25it/s]
+
 67%|██████▋   | 26/39 [00:03<00:01,  7.25it/s]
+
 69%|██████▉   | 27/39 [00:03<00:01,  7.25it/s]
+
 72%|███████▏  | 28/39 [00:03<00:01,  7.26it/s]
+
 74%|███████▍  | 29/39 [00:03<00:01,  7.26it/s]
+
 77%|███████▋  | 30/39 [00:04<00:01,  7.26it/s]
+
 79%|███████▉  | 31/39 [00:04<00:01,  7.26it/s]
+
 82%|████████▏ | 32/39 [00:04<00:00,  7.26it/s]
+
 85%|████████▍ | 33/39 [00:04<00:00,  7.26it/s]
+
 87%|████████▋ | 34/39 [00:04<00:00,  7.26it/s]
+
 90%|████████▉ | 35/39 [00:04<00:00,  7.26it/s]
+
 92%|█████████▏| 36/39 [00:04<00:00,  7.26it/s]
+
 95%|█████████▍| 37/39 [00:04<00:00,  7.26it/s]
+
 97%|█████████▋| 38/39 [00:05<00:00,  7.26it/s]
+
100%|██████████| 39/39 [00:05<00:00,  7.26it/s]
                                                 
+
                                               
{'eval_loss': '4.1', 'eval_runtime': '5.45', 'eval_samples_per_second': '447.9', 'eval_steps_per_second': '7.156', 'epoch': '8'}
+
 67%|██████▋   | 576/864 [15:07<05:36,  1.17s/it]
+
100%|██████████| 39/39 [00:05<00:00,  7.26it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0006/checkpoint-576
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0006/checkpoint-576/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0006/checkpoint-576/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:01<00:00,  1.68s/it]
Writing model shards: 100%|██████████| 1/1 [00:01<00:00,  1.68s/it]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0006/checkpoint-576/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0006/checkpoint-576/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0006/checkpoint-576/tokenizer_config.json
+[eval] epoch 8.0 step 576 loss 4.1000 (best 4.1000 @ ep 8.0, 0/10000 since best)
+
+
                                               
 67%|██████▋   | 577/864 [15:14<21:45,  4.55s/it]
 67%|██████▋   | 578/864 [15:15<17:02,  3.58s/it]
 67%|██████▋   | 579/864 [15:16<13:45,  2.90s/it]
 67%|██████▋   | 580/864 [15:18<11:27,  2.42s/it]
                                                 
{'loss': '3.777', 'grad_norm': '0.3998', 'learning_rate': '0.0006', 'epoch': '8.056'}
+
 67%|██████▋   | 580/864 [15:18<11:27,  2.42s/it]
 67%|██████▋   | 581/864 [15:19<09:50,  2.09s/it]
 67%|██████▋   | 582/864 [15:20<08:43,  1.85s/it]
 67%|██████▋   | 583/864 [15:22<07:55,  1.69s/it]
 68%|██████▊   | 584/864 [15:23<07:21,  1.58s/it]
 68%|██████▊   | 585/864 [15:24<06:57,  1.50s/it]
 68%|██████▊   | 586/864 [15:26<06:40,  1.44s/it]
 68%|██████▊   | 587/864 [15:27<06:28,  1.40s/it]
 68%|██████▊   | 588/864 [15:28<06:19,  1.38s/it]
 68%|██████▊   | 589/864 [15:30<06:13,  1.36s/it]
 68%|██████▊   | 590/864 [15:31<06:07,  1.34s/it]
                                                 
{'loss': '3.764', 'grad_norm': '0.7327', 'learning_rate': '0.0006', 'epoch': '8.195'}
+
 68%|██████▊   | 590/864 [15:31<06:07,  1.34s/it]
 68%|██████▊   | 591/864 [15:32<06:04,  1.33s/it]
 69%|██████▊   | 592/864 [15:33<06:01,  1.33s/it]
 69%|██████▊   | 593/864 [15:35<05:58,  1.32s/it]
 69%|██████▉   | 594/864 [15:36<05:56,  1.32s/it]
 69%|██████▉   | 595/864 [15:37<05:54,  1.32s/it]
 69%|██████▉   | 596/864 [15:39<05:52,  1.32s/it]
 69%|██████▉   | 597/864 [15:40<05:51,  1.32s/it]
 69%|██████▉   | 598/864 [15:41<05:49,  1.31s/it]
 69%|██████▉   | 599/864 [15:43<05:48,  1.31s/it]
 69%|██████▉   | 600/864 [15:44<05:46,  1.31s/it]
                                                 
{'loss': '4.086', 'grad_norm': '1.484', 'learning_rate': '0.0006', 'epoch': '8.335'}
+
 69%|██████▉   | 600/864 [15:44<05:46,  1.31s/it]
 70%|██████▉   | 601/864 [15:45<05:45,  1.31s/it]
 70%|██████▉   | 602/864 [15:47<05:44,  1.31s/it]
 70%|██████▉   | 603/864 [15:48<05:42,  1.31s/it]
 70%|██████▉   | 604/864 [15:49<05:40,  1.31s/it]
 70%|███████   | 605/864 [15:51<05:39,  1.31s/it]
 70%|███████   | 606/864 [15:52<05:38,  1.31s/it]
 70%|███████   | 607/864 [15:53<05:37,  1.31s/it]
 70%|███████   | 608/864 [15:54<05:35,  1.31s/it]
 70%|███████   | 609/864 [15:56<05:34,  1.31s/it]
 71%|███████   | 610/864 [15:57<05:33,  1.31s/it]
                                                 
{'loss': '4.354', 'grad_norm': '1.041', 'learning_rate': '0.0006', 'epoch': '8.475'}
+
 71%|███████   | 610/864 [15:57<05:33,  1.31s/it]
 71%|███████   | 611/864 [15:58<05:31,  1.31s/it]
 71%|███████   | 612/864 [16:00<05:30,  1.31s/it]
 71%|███████   | 613/864 [16:01<05:29,  1.31s/it]
 71%|███████   | 614/864 [16:02<05:28,  1.31s/it]
 71%|███████   | 615/864 [16:04<05:27,  1.31s/it]
 71%|███████▏  | 616/864 [16:05<05:25,  1.31s/it]
 71%|███████▏  | 617/864 [16:06<05:24,  1.31s/it]
 72%|███████▏  | 618/864 [16:08<05:22,  1.31s/it]
 72%|███████▏  | 619/864 [16:09<05:21,  1.31s/it]
 72%|███████▏  | 620/864 [16:10<05:20,  1.31s/it]
                                                 
{'loss': '4.537', 'grad_norm': '0.8818', 'learning_rate': '0.0006', 'epoch': '8.614'}
+
 72%|███████▏  | 620/864 [16:10<05:20,  1.31s/it]
 72%|███████▏  | 621/864 [16:12<05:19,  1.31s/it]
 72%|███████▏  | 622/864 [16:13<05:17,  1.31s/it]
 72%|███████▏  | 623/864 [16:14<05:16,  1.31s/it]
 72%|███████▏  | 624/864 [16:15<05:15,  1.32s/it]
 72%|███████▏  | 625/864 [16:17<05:14,  1.32s/it]
 72%|███████▏  | 626/864 [16:18<05:12,  1.31s/it]
 73%|███████▎  | 627/864 [16:19<05:11,  1.31s/it]
 73%|███████▎  | 628/864 [16:21<05:10,  1.31s/it]
 73%|███████▎  | 629/864 [16:22<05:08,  1.31s/it]
 73%|███████▎  | 630/864 [16:23<05:07,  1.31s/it]
                                                 
{'loss': '4.278', 'grad_norm': '0.1979', 'learning_rate': '0.0006', 'epoch': '8.754'}
+
 73%|███████▎  | 630/864 [16:23<05:07,  1.31s/it]
 73%|███████▎  | 631/864 [16:25<05:06,  1.32s/it]
 73%|███████▎  | 632/864 [16:26<05:05,  1.32s/it]
 73%|███████▎  | 633/864 [16:27<05:03,  1.31s/it]
 73%|███████▎  | 634/864 [16:29<05:02,  1.31s/it]
 73%|███████▎  | 635/864 [16:30<05:01,  1.31s/it]
 74%|███████▎  | 636/864 [16:31<04:59,  1.31s/it]
 74%|███████▎  | 637/864 [16:33<04:58,  1.31s/it]
 74%|███████▍  | 638/864 [16:34<04:56,  1.31s/it]
 74%|███████▍  | 639/864 [16:35<04:55,  1.31s/it]
 74%|███████▍  | 640/864 [16:37<05:16,  1.41s/it]
                                                 
{'loss': '4.077', 'grad_norm': '0.1398', 'learning_rate': '0.0006', 'epoch': '8.894'}
+
 74%|███████▍  | 640/864 [16:37<05:16,  1.41s/it]
 74%|███████▍  | 641/864 [16:38<05:08,  1.38s/it]
 74%|███████▍  | 642/864 [16:39<05:02,  1.36s/it]
 74%|███████▍  | 643/864 [16:41<04:58,  1.35s/it]
 75%|███████▍  | 644/864 [16:42<04:54,  1.34s/it]
 75%|███████▍  | 645/864 [16:43<04:51,  1.33s/it]
 75%|███████▍  | 646/864 [16:45<04:49,  1.33s/it]
 75%|███████▍  | 647/864 [16:46<04:46,  1.32s/it]
 75%|███████▌  | 648/864 [16:47<04:13,  1.17s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:02, 14.62it/s]
+
 10%|█         | 4/39 [00:00<00:03,  9.14it/s]
+
 15%|█▌        | 6/39 [00:00<00:04,  8.16it/s]
+
 18%|█▊        | 7/39 [00:00<00:04,  7.92it/s]
+
 21%|██        | 8/39 [00:00<00:04,  7.72it/s]
+
 23%|██▎       | 9/39 [00:01<00:03,  7.59it/s]
+
 26%|██▌       | 10/39 [00:01<00:03,  7.49it/s]
+
 28%|██▊       | 11/39 [00:01<00:03,  7.42it/s]
+
 31%|███       | 12/39 [00:01<00:03,  7.37it/s]
+
 33%|███▎      | 13/39 [00:01<00:03,  7.33it/s]
+
 36%|███▌      | 14/39 [00:01<00:03,  7.30it/s]
+
 38%|███▊      | 15/39 [00:01<00:03,  7.29it/s]
+
 41%|████      | 16/39 [00:02<00:03,  7.28it/s]
+
 44%|████▎     | 17/39 [00:02<00:03,  7.27it/s]
+
 46%|████▌     | 18/39 [00:02<00:02,  7.26it/s]
+
 49%|████▊     | 19/39 [00:02<00:02,  7.26it/s]
+
 51%|█████▏    | 20/39 [00:02<00:02,  7.25it/s]
+
 54%|█████▍    | 21/39 [00:02<00:02,  7.25it/s]
+
 56%|█████▋    | 22/39 [00:02<00:02,  7.25it/s]
+
 59%|█████▉    | 23/39 [00:03<00:02,  7.25it/s]
+
 62%|██████▏   | 24/39 [00:03<00:02,  7.25it/s]
+
 64%|██████▍   | 25/39 [00:03<00:01,  7.25it/s]
+
 67%|██████▋   | 26/39 [00:03<00:01,  7.25it/s]
+
 69%|██████▉   | 27/39 [00:03<00:01,  7.26it/s]
+
 72%|███████▏  | 28/39 [00:03<00:01,  7.25it/s]
+
 74%|███████▍  | 29/39 [00:03<00:01,  7.25it/s]
+
 77%|███████▋  | 30/39 [00:04<00:01,  7.24it/s]
+
 79%|███████▉  | 31/39 [00:04<00:01,  7.24it/s]
+
 82%|████████▏ | 32/39 [00:04<00:00,  7.24it/s]
+
 85%|████████▍ | 33/39 [00:04<00:00,  7.24it/s]
+
 87%|████████▋ | 34/39 [00:04<00:00,  7.25it/s]
+
 90%|████████▉ | 35/39 [00:04<00:00,  7.25it/s]
+
 92%|█████████▏| 36/39 [00:04<00:00,  7.25it/s]
+
 95%|█████████▍| 37/39 [00:04<00:00,  7.26it/s]
+
 97%|█████████▋| 38/39 [00:05<00:00,  7.26it/s]
+
100%|██████████| 39/39 [00:05<00:00,  7.25it/s]
                                                 
+
                                               
{'eval_loss': '4.183', 'eval_runtime': '5.453', 'eval_samples_per_second': '447.6', 'eval_steps_per_second': '7.152', 'epoch': '9'}
+
 75%|███████▌  | 648/864 [16:52<04:13,  1.17s/it]
+
100%|██████████| 39/39 [00:05<00:00,  7.25it/s][eval] epoch 9.0 step 648 loss 4.1831 (best 4.1000 @ ep 8.0, 1/10000 since best)
+
+
                                               
 75%|███████▌  | 649/864 [16:54<10:19,  2.88s/it]
 75%|███████▌  | 650/864 [16:55<08:35,  2.41s/it]
                                                 
{'loss': '3.935', 'grad_norm': '0.1208', 'learning_rate': '0.0006', 'epoch': '9.028'}
+
 75%|███████▌  | 650/864 [16:55<08:35,  2.41s/it]
 75%|███████▌  | 651/864 [16:56<07:23,  2.08s/it]
 75%|███████▌  | 652/864 [16:58<06:32,  1.85s/it]
 76%|███████▌  | 653/864 [16:59<05:56,  1.69s/it]
 76%|███████▌  | 654/864 [17:00<05:31,  1.58s/it]
 76%|███████▌  | 655/864 [17:02<05:13,  1.50s/it]
 76%|███████▌  | 656/864 [17:03<05:00,  1.44s/it]
 76%|███████▌  | 657/864 [17:04<04:50,  1.40s/it]
 76%|███████▌  | 658/864 [17:06<04:43,  1.38s/it]
 76%|███████▋  | 659/864 [17:07<04:38,  1.36s/it]
 76%|███████▋  | 660/864 [17:08<04:34,  1.34s/it]
                                                 
{'loss': '3.807', 'grad_norm': '0.1414', 'learning_rate': '0.0006', 'epoch': '9.168'}
+
 76%|███████▋  | 660/864 [17:08<04:34,  1.34s/it]
 77%|███████▋  | 661/864 [17:09<04:31,  1.34s/it]
 77%|███████▋  | 662/864 [17:11<04:49,  1.43s/it]
 77%|███████▋  | 663/864 [17:13<05:04,  1.51s/it]
 77%|███████▋  | 664/864 [17:14<04:50,  1.45s/it]
 77%|███████▋  | 665/864 [17:15<04:40,  1.41s/it]
 77%|███████▋  | 666/864 [17:17<04:33,  1.38s/it]
 77%|███████▋  | 667/864 [17:18<04:28,  1.36s/it]
 77%|███████▋  | 668/864 [17:19<04:24,  1.35s/it]
 77%|███████▋  | 669/864 [17:21<04:21,  1.34s/it]
 78%|███████▊  | 670/864 [17:22<04:18,  1.33s/it]
                                                 
{'loss': '3.774', 'grad_norm': '0.136', 'learning_rate': '0.0006', 'epoch': '9.307'}
+
 78%|███████▊  | 670/864 [17:22<04:18,  1.33s/it]
 78%|███████▊  | 671/864 [17:23<04:16,  1.33s/it]
 78%|███████▊  | 672/864 [17:25<04:14,  1.32s/it]
 78%|███████▊  | 673/864 [17:26<04:12,  1.32s/it]
 78%|███████▊  | 674/864 [17:27<04:10,  1.32s/it]
 78%|███████▊  | 675/864 [17:29<04:09,  1.32s/it]
 78%|███████▊  | 676/864 [17:30<04:07,  1.32s/it]
 78%|███████▊  | 677/864 [17:31<04:06,  1.32s/it]
 78%|███████▊  | 678/864 [17:33<04:04,  1.32s/it]
 79%|███████▊  | 679/864 [17:34<04:03,  1.32s/it]
 79%|███████▊  | 680/864 [17:35<04:01,  1.31s/it]
                                                 
{'loss': '3.747', 'grad_norm': '0.1396', 'learning_rate': '0.0006', 'epoch': '9.447'}
+
 79%|███████▊  | 680/864 [17:35<04:01,  1.31s/it]
 79%|███████▉  | 681/864 [17:37<04:00,  1.31s/it]
 79%|███████▉  | 682/864 [17:38<03:59,  1.32s/it]
 79%|███████▉  | 683/864 [17:39<03:58,  1.32s/it]
 79%|███████▉  | 684/864 [17:40<03:56,  1.32s/it]
 79%|███████▉  | 685/864 [17:42<03:55,  1.32s/it]
 79%|███████▉  | 686/864 [17:43<03:54,  1.31s/it]
 80%|███████▉  | 687/864 [17:44<03:52,  1.32s/it]
 80%|███████▉  | 688/864 [17:46<03:51,  1.32s/it]
 80%|███████▉  | 689/864 [17:47<03:50,  1.31s/it]
 80%|███████▉  | 690/864 [17:48<03:48,  1.32s/it]
                                                 
{'loss': '3.709', 'grad_norm': '0.1287', 'learning_rate': '0.0006', 'epoch': '9.586'}
+
 80%|███████▉  | 690/864 [17:48<03:48,  1.32s/it]
 80%|███████▉  | 691/864 [17:50<03:47,  1.32s/it]
 80%|████████  | 692/864 [17:51<03:46,  1.32s/it]
 80%|████████  | 693/864 [17:52<03:44,  1.31s/it]
 80%|████████  | 694/864 [17:54<03:43,  1.31s/it]
 80%|████████  | 695/864 [17:55<03:41,  1.31s/it]
 81%|████████  | 696/864 [17:56<03:40,  1.31s/it]
 81%|████████  | 697/864 [17:58<03:39,  1.31s/it]
 81%|████████  | 698/864 [17:59<03:37,  1.31s/it]
 81%|████████  | 699/864 [18:00<03:36,  1.31s/it]
 81%|████████  | 700/864 [18:02<03:35,  1.31s/it]
                                                 
{'loss': '3.706', 'grad_norm': '0.1794', 'learning_rate': '0.0006', 'epoch': '9.726'}
+
 81%|████████  | 700/864 [18:02<03:35,  1.31s/it]
 81%|████████  | 701/864 [18:03<03:34,  1.31s/it]
 81%|████████▏ | 702/864 [18:04<03:32,  1.31s/it]
 81%|████████▏ | 703/864 [18:05<03:31,  1.31s/it]
 81%|████████▏ | 704/864 [18:07<03:30,  1.31s/it]
 82%|████████▏ | 705/864 [18:08<03:28,  1.31s/it]
 82%|████████▏ | 706/864 [18:09<03:27,  1.31s/it]
 82%|████████▏ | 707/864 [18:11<03:26,  1.32s/it]
 82%|████████▏ | 708/864 [18:12<03:25,  1.32s/it]
 82%|████████▏ | 709/864 [18:13<03:23,  1.31s/it]
 82%|████████▏ | 710/864 [18:15<03:22,  1.31s/it]
                                                 
{'loss': '3.68', 'grad_norm': '0.1411', 'learning_rate': '0.0006', 'epoch': '9.866'}
+
 82%|████████▏ | 710/864 [18:15<03:22,  1.31s/it]
 82%|████████▏ | 711/864 [18:16<03:21,  1.31s/it]
 82%|████████▏ | 712/864 [18:17<03:19,  1.31s/it]
 83%|████████▎ | 713/864 [18:19<03:18,  1.31s/it]
 83%|████████▎ | 714/864 [18:20<03:16,  1.31s/it]
 83%|████████▎ | 715/864 [18:21<03:15,  1.31s/it]
 83%|████████▎ | 716/864 [18:23<03:14,  1.31s/it]
 83%|████████▎ | 717/864 [18:24<03:12,  1.31s/it]
 83%|████████▎ | 718/864 [18:25<03:11,  1.31s/it]
 83%|████████▎ | 719/864 [18:26<03:10,  1.31s/it]
 83%|████████▎ | 720/864 [18:27<02:48,  1.17s/it]
                                                 
{'loss': '3.659', 'grad_norm': '0.1609', 'learning_rate': '0.0006', 'epoch': '10'}
+
 83%|████████▎ | 720/864 [18:27<02:48,  1.17s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:02, 14.47it/s]
+
 10%|█         | 4/39 [00:00<00:03,  9.09it/s]
+
 15%|█▌        | 6/39 [00:00<00:04,  8.14it/s]
+
 18%|█▊        | 7/39 [00:00<00:04,  7.90it/s]
+
 21%|██        | 8/39 [00:00<00:04,  7.72it/s]
+
 23%|██▎       | 9/39 [00:01<00:03,  7.58it/s]
+
 26%|██▌       | 10/39 [00:01<00:03,  7.49it/s]
+
 28%|██▊       | 11/39 [00:01<00:03,  7.42it/s]
+
 31%|███       | 12/39 [00:01<00:03,  7.37it/s]
+
 33%|███▎      | 13/39 [00:01<00:03,  7.33it/s]
+
 36%|███▌      | 14/39 [00:01<00:03,  7.29it/s]
+
 38%|███▊      | 15/39 [00:01<00:03,  7.27it/s]
+
 41%|████      | 16/39 [00:02<00:03,  7.26it/s]
+
 44%|████▎     | 17/39 [00:02<00:03,  7.26it/s]
+
 46%|████▌     | 18/39 [00:02<00:02,  7.25it/s]
+
 49%|████▊     | 19/39 [00:02<00:02,  7.24it/s]
+
 51%|█████▏    | 20/39 [00:02<00:02,  7.23it/s]
+
 54%|█████▍    | 21/39 [00:02<00:02,  7.23it/s]
+
 56%|█████▋    | 22/39 [00:02<00:02,  7.24it/s]
+
 59%|█████▉    | 23/39 [00:03<00:02,  7.23it/s]
+
 62%|██████▏   | 24/39 [00:03<00:02,  7.23it/s]
+
 64%|██████▍   | 25/39 [00:03<00:01,  7.23it/s]
+
 67%|██████▋   | 26/39 [00:03<00:01,  7.23it/s]
+
 69%|██████▉   | 27/39 [00:03<00:01,  7.23it/s]
+
 72%|███████▏  | 28/39 [00:03<00:01,  7.23it/s]
+
 74%|███████▍  | 29/39 [00:03<00:01,  7.23it/s]
+
 77%|███████▋  | 30/39 [00:04<00:01,  7.22it/s]
+
 79%|███████▉  | 31/39 [00:04<00:01,  7.23it/s]
+
 82%|████████▏ | 32/39 [00:04<00:00,  7.23it/s]
+
 85%|████████▍ | 33/39 [00:04<00:00,  7.24it/s]
+
 87%|████████▋ | 34/39 [00:04<00:00,  7.24it/s]
+
 90%|████████▉ | 35/39 [00:04<00:00,  7.23it/s]
+
 92%|█████████▏| 36/39 [00:04<00:00,  7.23it/s]
+
 95%|█████████▍| 37/39 [00:04<00:00,  7.23it/s]
+
 97%|█████████▋| 38/39 [00:05<00:00,  7.23it/s]
+
100%|██████████| 39/39 [00:05<00:00,  7.22it/s]
                                                 
+
                                               
{'eval_loss': '4.012', 'eval_runtime': '5.463', 'eval_samples_per_second': '446.8', 'eval_steps_per_second': '7.139', 'epoch': '10'}
+
 83%|████████▎ | 720/864 [18:33<02:48,  1.17s/it]
+
100%|██████████| 39/39 [00:05<00:00,  7.22it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0006/checkpoint-720
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0006/checkpoint-720/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0006/checkpoint-720/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:01<00:00,  1.81s/it]
Writing model shards: 100%|██████████| 1/1 [00:01<00:00,  1.81s/it]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0006/checkpoint-720/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0006/checkpoint-720/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0006/checkpoint-720/tokenizer_config.json
+[eval] epoch 10.0 step 720 loss 4.0124 (best 4.0124 @ ep 10.0, 0/10000 since best)
+
+
                                               
 83%|████████▎ | 721/864 [18:40<11:09,  4.68s/it]
 84%|████████▎ | 722/864 [18:41<08:41,  3.67s/it]
 84%|████████▎ | 723/864 [18:43<06:57,  2.96s/it]
 84%|████████▍ | 724/864 [18:44<05:45,  2.47s/it]
 84%|████████▍ | 725/864 [18:45<04:54,  2.12s/it]
 84%|████████▍ | 726/864 [18:47<04:19,  1.88s/it]
 84%|████████▍ | 727/864 [18:48<03:54,  1.71s/it]
 84%|████████▍ | 728/864 [18:49<03:36,  1.59s/it]
 84%|████████▍ | 729/864 [18:51<03:23,  1.51s/it]
 84%|████████▍ | 730/864 [18:52<03:14,  1.45s/it]
                                                 
{'loss': '3.542', 'grad_norm': '0.3115', 'learning_rate': '0.0006', 'epoch': '10.14'}
+
 84%|████████▍ | 730/864 [18:52<03:14,  1.45s/it]
 85%|████████▍ | 731/864 [18:53<03:07,  1.41s/it]
 85%|████████▍ | 732/864 [18:55<03:02,  1.38s/it]
 85%|████████▍ | 733/864 [18:56<02:58,  1.36s/it]
 85%|████████▍ | 734/864 [18:57<02:55,  1.35s/it]
 85%|████████▌ | 735/864 [18:59<02:52,  1.34s/it]
 85%|████████▌ | 736/864 [19:00<02:50,  1.33s/it]
 85%|████████▌ | 737/864 [19:01<02:48,  1.32s/it]
 85%|████████▌ | 738/864 [19:02<02:46,  1.32s/it]
 86%|████████▌ | 739/864 [19:04<02:44,  1.32s/it]
 86%|████████▌ | 740/864 [19:05<02:43,  1.32s/it]
                                                 
{'loss': '3.557', 'grad_norm': '0.2759', 'learning_rate': '0.0006', 'epoch': '10.28'}
+
 86%|████████▌ | 740/864 [19:05<02:43,  1.32s/it]
 86%|████████▌ | 741/864 [19:06<02:41,  1.32s/it]
 86%|████████▌ | 742/864 [19:08<02:40,  1.32s/it]
 86%|████████▌ | 743/864 [19:09<02:39,  1.32s/it]
 86%|████████▌ | 744/864 [19:10<02:37,  1.31s/it]
 86%|████████▌ | 745/864 [19:12<02:36,  1.31s/it]
 86%|████████▋ | 746/864 [19:13<02:35,  1.31s/it]
 86%|████████▋ | 747/864 [19:14<02:33,  1.31s/it]
 87%|████████▋ | 748/864 [19:16<02:32,  1.31s/it]
 87%|████████▋ | 749/864 [19:17<02:31,  1.31s/it]
 87%|████████▋ | 750/864 [19:18<02:29,  1.31s/it]
                                                 
{'loss': '3.546', 'grad_norm': '0.1726', 'learning_rate': '0.0006', 'epoch': '10.42'}
+
 87%|████████▋ | 750/864 [19:18<02:29,  1.31s/it]
 87%|████████▋ | 751/864 [19:20<02:28,  1.32s/it]
 87%|████████▋ | 752/864 [19:21<02:27,  1.31s/it]
 87%|████████▋ | 753/864 [19:22<02:25,  1.32s/it]
 87%|████████▋ | 754/864 [19:24<02:24,  1.31s/it]
 87%|████████▋ | 755/864 [19:25<02:23,  1.31s/it]
 88%|████████▊ | 756/864 [19:26<02:22,  1.32s/it]
 88%|████████▊ | 757/864 [19:27<02:20,  1.31s/it]
 88%|████████▊ | 758/864 [19:29<02:19,  1.31s/it]
 88%|████████▊ | 759/864 [19:30<02:18,  1.31s/it]
 88%|████████▊ | 760/864 [19:31<02:16,  1.31s/it]
                                                 
{'loss': '3.545', 'grad_norm': '0.1407', 'learning_rate': '0.0006', 'epoch': '10.56'}
+
 88%|████████▊ | 760/864 [19:31<02:16,  1.31s/it]
 88%|████████▊ | 761/864 [19:33<02:15,  1.31s/it]
 88%|████████▊ | 762/864 [19:34<02:14,  1.32s/it]
 88%|████████▊ | 763/864 [19:35<02:12,  1.32s/it]
 88%|████████▊ | 764/864 [19:37<02:11,  1.32s/it]
 89%|████████▊ | 765/864 [19:38<02:10,  1.32s/it]
 89%|████████▊ | 766/864 [19:39<02:08,  1.31s/it]
 89%|████████▉ | 767/864 [19:41<02:07,  1.31s/it]
 89%|████████▉ | 768/864 [19:42<02:06,  1.32s/it]
 89%|████████▉ | 769/864 [19:43<02:04,  1.32s/it]
 89%|████████▉ | 770/864 [19:45<02:03,  1.31s/it]
                                                 
{'loss': '3.532', 'grad_norm': '0.125', 'learning_rate': '0.0006', 'epoch': '10.7'}
+
 89%|████████▉ | 770/864 [19:45<02:03,  1.31s/it]
 89%|████████▉ | 771/864 [19:46<02:02,  1.31s/it]
 89%|████████▉ | 772/864 [19:47<02:00,  1.32s/it]
 89%|████████▉ | 773/864 [19:49<01:59,  1.32s/it]
 90%|████████▉ | 774/864 [19:50<01:58,  1.31s/it]
 90%|████████▉ | 775/864 [19:51<01:57,  1.32s/it]
 90%|████████▉ | 776/864 [19:52<01:55,  1.32s/it]
 90%|████████▉ | 777/864 [19:54<01:54,  1.32s/it]
 90%|█████████ | 778/864 [19:55<01:53,  1.31s/it]
 90%|█████████ | 779/864 [19:56<01:51,  1.31s/it]
 90%|█████████ | 780/864 [19:58<01:50,  1.31s/it]
                                                 
{'loss': '3.534', 'grad_norm': '0.1346', 'learning_rate': '0.0006', 'epoch': '10.84'}
+
 90%|█████████ | 780/864 [19:58<01:50,  1.31s/it]
 90%|█████████ | 781/864 [19:59<01:49,  1.31s/it]
 91%|█████████ | 782/864 [20:00<01:47,  1.32s/it]
 91%|█████████ | 783/864 [20:02<01:46,  1.31s/it]
 91%|█████████ | 784/864 [20:03<01:45,  1.31s/it]
 91%|█████████ | 785/864 [20:04<01:43,  1.31s/it]
 91%|█████████ | 786/864 [20:06<01:42,  1.31s/it]
 91%|█████████ | 787/864 [20:07<01:41,  1.31s/it]
 91%|█████████ | 788/864 [20:08<01:39,  1.32s/it]
 91%|█████████▏| 789/864 [20:10<01:38,  1.31s/it]
 91%|█████████▏| 790/864 [20:11<01:37,  1.31s/it]
                                                 
{'loss': '3.522', 'grad_norm': '0.1449', 'learning_rate': '0.0006', 'epoch': '10.98'}
+
 91%|█████████▏| 790/864 [20:11<01:37,  1.31s/it]
 92%|█████████▏| 791/864 [20:12<01:36,  1.32s/it]
 92%|█████████▏| 792/864 [20:13<01:24,  1.17s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:02, 15.09it/s]
+
 10%|█         | 4/39 [00:00<00:03,  9.21it/s]
+
 15%|█▌        | 6/39 [00:00<00:04,  8.20it/s]
+
 18%|█▊        | 7/39 [00:00<00:04,  7.95it/s]
+
 21%|██        | 8/39 [00:00<00:04,  7.75it/s]
+
 23%|██▎       | 9/39 [00:01<00:03,  7.60it/s]
+
 26%|██▌       | 10/39 [00:01<00:03,  7.50it/s]
+
 28%|██▊       | 11/39 [00:01<00:03,  7.43it/s]
+
 31%|███       | 12/39 [00:01<00:03,  7.38it/s]
+
 33%|███▎      | 13/39 [00:01<00:03,  7.34it/s]
+
 36%|███▌      | 14/39 [00:01<00:03,  7.30it/s]
+
 38%|███▊      | 15/39 [00:01<00:03,  7.28it/s]
+
 41%|████      | 16/39 [00:02<00:03,  7.27it/s]
+
 44%|████▎     | 17/39 [00:02<00:03,  7.27it/s]
+
 46%|████▌     | 18/39 [00:02<00:02,  7.26it/s]
+
 49%|████▊     | 19/39 [00:02<00:02,  7.26it/s]
+
 51%|█████▏    | 20/39 [00:02<00:02,  7.25it/s]
+
 54%|█████▍    | 21/39 [00:02<00:02,  7.26it/s]
+
 56%|█████▋    | 22/39 [00:02<00:02,  7.26it/s]
+
 59%|█████▉    | 23/39 [00:03<00:02,  7.26it/s]
+
 62%|██████▏   | 24/39 [00:03<00:02,  7.26it/s]
+
 64%|██████▍   | 25/39 [00:03<00:01,  7.26it/s]
+
 67%|██████▋   | 26/39 [00:03<00:01,  7.26it/s]
+
 69%|██████▉   | 27/39 [00:03<00:01,  7.26it/s]
+
 72%|███████▏  | 28/39 [00:03<00:01,  7.26it/s]
+
 74%|███████▍  | 29/39 [00:03<00:01,  7.26it/s]
+
 77%|███████▋  | 30/39 [00:03<00:01,  7.26it/s]
+
 79%|███████▉  | 31/39 [00:04<00:01,  7.25it/s]
+
 82%|████████▏ | 32/39 [00:04<00:00,  7.25it/s]
+
 85%|████████▍ | 33/39 [00:04<00:00,  7.25it/s]
+
 87%|████████▋ | 34/39 [00:04<00:00,  7.25it/s]
+
 90%|████████▉ | 35/39 [00:04<00:00,  7.25it/s]
+
 92%|█████████▏| 36/39 [00:04<00:00,  7.25it/s]
+
 95%|█████████▍| 37/39 [00:04<00:00,  7.26it/s]
+
 97%|█████████▋| 38/39 [00:05<00:00,  7.26it/s]
+
100%|██████████| 39/39 [00:05<00:00,  7.25it/s]
                                                 
+
                                               
{'eval_loss': '3.971', 'eval_runtime': '5.45', 'eval_samples_per_second': '447.9', 'eval_steps_per_second': '7.157', 'epoch': '11'}
+
 92%|█████████▏| 792/864 [20:18<01:24,  1.17s/it]
+
100%|██████████| 39/39 [00:05<00:00,  7.25it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0006/checkpoint-792
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0006/checkpoint-792/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0006/checkpoint-792/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:01<00:00,  1.85s/it]
Writing model shards: 100%|██████████| 1/1 [00:01<00:00,  1.86s/it]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0006/checkpoint-792/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0006/checkpoint-792/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0006/checkpoint-792/tokenizer_config.json
+[eval] epoch 11.0 step 792 loss 3.9706 (best 3.9706 @ ep 11.0, 0/10000 since best)
+
+
                                               
 92%|█████████▏| 793/864 [20:26<05:32,  4.68s/it]
 92%|█████████▏| 794/864 [20:27<04:17,  3.67s/it]
 92%|█████████▏| 795/864 [20:29<03:24,  2.96s/it]
 92%|█████████▏| 796/864 [20:30<02:47,  2.47s/it]
 92%|█████████▏| 797/864 [20:31<02:22,  2.12s/it]
 92%|█████████▏| 798/864 [20:32<02:04,  1.88s/it]
 92%|█████████▏| 799/864 [20:34<01:51,  1.71s/it]
 93%|█████████▎| 800/864 [20:35<01:41,  1.59s/it]
                                                 
{'loss': '3.398', 'grad_norm': '0.3894', 'learning_rate': '0.0006', 'epoch': '11.11'}
+
 93%|█████████▎| 800/864 [20:35<01:41,  1.59s/it]
 93%|█████████▎| 801/864 [20:36<01:35,  1.51s/it]
 93%|█████████▎| 802/864 [20:38<01:29,  1.45s/it]
 93%|█████████▎| 803/864 [20:39<01:25,  1.41s/it]
 93%|█████████▎| 804/864 [20:40<01:22,  1.38s/it]
 93%|█████████▎| 805/864 [20:42<01:20,  1.36s/it]
 93%|█████████▎| 806/864 [20:43<01:18,  1.35s/it]
 93%|█████████▎| 807/864 [20:44<01:16,  1.34s/it]
 94%|█████████▎| 808/864 [20:46<01:14,  1.33s/it]
 94%|█████████▎| 809/864 [20:47<01:12,  1.32s/it]
 94%|█████████▍| 810/864 [20:48<01:11,  1.32s/it]
                                                 
{'loss': '3.348', 'grad_norm': '0.2421', 'learning_rate': '0.0006', 'epoch': '11.25'}
+
 94%|█████████▍| 810/864 [20:48<01:11,  1.32s/it]
 94%|█████████▍| 811/864 [20:50<01:09,  1.32s/it]
 94%|█████████▍| 812/864 [20:51<01:08,  1.32s/it]
 94%|█████████▍| 813/864 [20:52<01:07,  1.32s/it]
 94%|█████████▍| 814/864 [20:53<01:05,  1.31s/it]
 94%|█████████▍| 815/864 [20:55<01:04,  1.31s/it]
 94%|█████████▍| 816/864 [20:56<01:03,  1.31s/it]
 95%|█████████▍| 817/864 [20:57<01:01,  1.31s/it]
 95%|█████████▍| 818/864 [20:59<01:00,  1.31s/it]
 95%|█████████▍| 819/864 [21:00<00:59,  1.32s/it]
 95%|█████████▍| 820/864 [21:01<00:57,  1.32s/it]
                                                 
{'loss': '3.361', 'grad_norm': '0.24', 'learning_rate': '0.0006', 'epoch': '11.39'}
+
 95%|█████████▍| 820/864 [21:01<00:57,  1.32s/it]
 95%|█████████▌| 821/864 [21:03<00:56,  1.31s/it]
 95%|█████████▌| 822/864 [21:04<00:55,  1.31s/it]
 95%|█████████▌| 823/864 [21:05<00:53,  1.31s/it]
 95%|█████████▌| 824/864 [21:07<00:52,  1.31s/it]
 95%|█████████▌| 825/864 [21:08<00:51,  1.31s/it]
 96%|█████████▌| 826/864 [21:09<00:49,  1.31s/it]
 96%|█████████▌| 827/864 [21:11<00:48,  1.31s/it]
 96%|█████████▌| 828/864 [21:12<00:47,  1.31s/it]
 96%|█████████▌| 829/864 [21:13<00:45,  1.31s/it]
 96%|█████████▌| 830/864 [21:14<00:44,  1.31s/it]
                                                 
{'loss': '3.378', 'grad_norm': '0.1819', 'learning_rate': '0.0006', 'epoch': '11.53'}
+
 96%|█████████▌| 830/864 [21:15<00:44,  1.31s/it]
 96%|█████████▌| 831/864 [21:16<00:43,  1.31s/it]
 96%|█████████▋| 832/864 [21:17<00:42,  1.31s/it]
 96%|█████████▋| 833/864 [21:18<00:40,  1.31s/it]
 97%|█████████▋| 834/864 [21:20<00:39,  1.31s/it]
 97%|█████████▋| 835/864 [21:21<00:38,  1.31s/it]
 97%|█████████▋| 836/864 [21:22<00:36,  1.31s/it]
 97%|█████████▋| 837/864 [21:24<00:35,  1.31s/it]
 97%|█████████▋| 838/864 [21:25<00:34,  1.31s/it]
 97%|█████████▋| 839/864 [21:26<00:32,  1.31s/it]
 97%|█████████▋| 840/864 [21:28<00:31,  1.31s/it]
                                                 
{'loss': '3.379', 'grad_norm': '0.1585', 'learning_rate': '0.0006', 'epoch': '11.67'}
+
 97%|█████████▋| 840/864 [21:28<00:31,  1.31s/it]
 97%|█████████▋| 841/864 [21:29<00:30,  1.31s/it]
 97%|█████████▋| 842/864 [21:30<00:28,  1.31s/it]
 98%|█████████▊| 843/864 [21:32<00:27,  1.31s/it]
 98%|█████████▊| 844/864 [21:33<00:26,  1.31s/it]
 98%|█████████▊| 845/864 [21:34<00:24,  1.31s/it]
 98%|█████████▊| 846/864 [21:36<00:23,  1.31s/it]
 98%|█████████▊| 847/864 [21:37<00:22,  1.31s/it]
 98%|█████████▊| 848/864 [21:38<00:21,  1.31s/it]
 98%|█████████▊| 849/864 [21:39<00:19,  1.31s/it]
 98%|█████████▊| 850/864 [21:41<00:18,  1.31s/it]
                                                 
{'loss': '3.387', 'grad_norm': '0.1721', 'learning_rate': '0.0006', 'epoch': '11.81'}
+
 98%|█████████▊| 850/864 [21:41<00:18,  1.31s/it]
 98%|█████████▊| 851/864 [21:42<00:17,  1.31s/it]
 99%|█████████▊| 852/864 [21:43<00:15,  1.31s/it]
 99%|█████████▊| 853/864 [21:45<00:14,  1.31s/it]
 99%|█████████▉| 854/864 [21:46<00:13,  1.31s/it]
 99%|█████████▉| 855/864 [21:47<00:11,  1.31s/it]
 99%|█████████▉| 856/864 [21:49<00:10,  1.31s/it]
 99%|█████████▉| 857/864 [21:50<00:09,  1.31s/it]
 99%|█████████▉| 858/864 [21:51<00:07,  1.31s/it]
 99%|█████████▉| 859/864 [21:53<00:06,  1.31s/it]
100%|█████████▉| 860/864 [21:54<00:05,  1.31s/it]
                                                 
{'loss': '3.397', 'grad_norm': '0.1668', 'learning_rate': '0.0006', 'epoch': '11.95'}
+
100%|█████████▉| 860/864 [21:54<00:05,  1.31s/it]
100%|█████████▉| 861/864 [21:55<00:03,  1.31s/it]
100%|█████████▉| 862/864 [21:57<00:02,  1.32s/it]
100%|█████████▉| 863/864 [21:58<00:01,  1.31s/it]
100%|██████████| 864/864 [21:59<00:00,  1.17s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:02, 14.50it/s]
+
 10%|█         | 4/39 [00:00<00:03,  9.12it/s]
+
 15%|█▌        | 6/39 [00:00<00:04,  8.15it/s]
+
 18%|█▊        | 7/39 [00:00<00:04,  7.90it/s]
+
 21%|██        | 8/39 [00:00<00:04,  7.72it/s]
+
 23%|██▎       | 9/39 [00:01<00:03,  7.59it/s]
+
 26%|██▌       | 10/39 [00:01<00:03,  7.49it/s]
+
 28%|██▊       | 11/39 [00:01<00:03,  7.41it/s]
+
 31%|███       | 12/39 [00:01<00:03,  7.35it/s]
+
 33%|███▎      | 13/39 [00:01<00:03,  7.31it/s]
+
 36%|███▌      | 14/39 [00:01<00:03,  7.29it/s]
+
 38%|███▊      | 15/39 [00:01<00:03,  7.27it/s]
+
 41%|████      | 16/39 [00:02<00:03,  7.25it/s]
+
 44%|████▎     | 17/39 [00:02<00:03,  7.24it/s]
+
 46%|████▌     | 18/39 [00:02<00:02,  7.24it/s]
+
 49%|████▊     | 19/39 [00:02<00:02,  7.24it/s]
+
 51%|█████▏    | 20/39 [00:02<00:02,  7.24it/s]
+
 54%|█████▍    | 21/39 [00:02<00:02,  7.24it/s]
+
 56%|█████▋    | 22/39 [00:02<00:02,  7.24it/s]
+
 59%|█████▉    | 23/39 [00:03<00:02,  7.25it/s]
+
 62%|██████▏   | 24/39 [00:03<00:02,  7.25it/s]
+
 64%|██████▍   | 25/39 [00:03<00:01,  7.25it/s]
+
 67%|██████▋   | 26/39 [00:03<00:01,  7.25it/s]
+
 69%|██████▉   | 27/39 [00:03<00:01,  7.25it/s]
+
 72%|███████▏  | 28/39 [00:03<00:01,  7.26it/s]
+
 74%|███████▍  | 29/39 [00:03<00:01,  7.26it/s]
+
 77%|███████▋  | 30/39 [00:04<00:01,  7.26it/s]
+
 79%|███████▉  | 31/39 [00:04<00:01,  7.25it/s]
+
 82%|████████▏ | 32/39 [00:04<00:00,  7.26it/s]
+
 85%|████████▍ | 33/39 [00:04<00:00,  7.26it/s]
+
 87%|████████▋ | 34/39 [00:04<00:00,  7.24it/s]
+
 90%|████████▉ | 35/39 [00:04<00:00,  7.25it/s]
+
 92%|█████████▏| 36/39 [00:04<00:00,  7.25it/s]
+
 95%|█████████▍| 37/39 [00:04<00:00,  7.25it/s]
+
 97%|█████████▋| 38/39 [00:05<00:00,  7.24it/s]
+
100%|██████████| 39/39 [00:05<00:00,  7.24it/s]
                                                 
+
                                               
{'eval_loss': '3.964', 'eval_runtime': '5.447', 'eval_samples_per_second': '448.1', 'eval_steps_per_second': '7.16', 'epoch': '12'}
+
100%|██████████| 864/864 [22:04<00:00,  1.17s/it]
+
100%|██████████| 39/39 [00:05<00:00,  7.24it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0006/checkpoint-864
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0006/checkpoint-864/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0006/checkpoint-864/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:01<00:00,  1.80s/it]
Writing model shards: 100%|██████████| 1/1 [00:01<00:00,  1.80s/it]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0006/checkpoint-864/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0006/checkpoint-864/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0006/checkpoint-864/tokenizer_config.json
+[eval] epoch 12.0 step 864 loss 3.9642 (best 3.9642 @ ep 12.0, 0/10000 since best)
+
+
                                               [transformers] 
+
+Training completed. Do not forget to share your model on huggingface.co/models =)
+
+
+
                                                 
{'train_runtime': '1333', 'train_samples_per_second': '329.7', 'train_steps_per_second': '0.648', 'train_loss': '4.833', 'epoch': '12'}
+
100%|██████████| 864/864 [22:11<00:00,  1.17s/it]
100%|██████████| 864/864 [22:11<00:00,  1.54s/it]
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+wandb: 
+wandb: 🚀 View run t0_S4_lr0.0006 at: https://wandb.ai/slowrun-wankers/data-aug-pretraining/runs/xd89q9zj
+wandb: Find logs at: wandb/run-20260817_204713-xd89q9zj/logs
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+[rank0]:[W817 21:10:09.187974783 ProcessGroupNCCL.cpp:1575] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator())
diff --git a/runs/t0/t0_S4_lr0.0012/checkpoint-864/chat_template.jinja b/runs/t0/t0_S4_lr0.0012/checkpoint-864/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..699ff8df401fe4788525e9c1f9b86a99eadd6230
--- /dev/null
+++ b/runs/t0/t0_S4_lr0.0012/checkpoint-864/chat_template.jinja
@@ -0,0 +1,85 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0].role == 'system' %}
+        {{- messages[0].content + '\n\n' }}
+    {%- endif %}
+    {{- "# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0].content + '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
+{%- for message in messages[::-1] %}
+    {%- set index = (messages|length - 1) - loop.index0 %}
+    {%- if ns.multi_step_tool and message.role == "user" and not(message.content.startswith('<tool_response>') and message.content.endswith('</tool_response>')) %}
+        {%- set ns.multi_step_tool = false %}
+        {%- set ns.last_query_index = index %}
+    {%- endif %}
+{%- endfor %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {%- set content = message.content %}
+        {%- set reasoning_content = '' %}
+        {%- if message.reasoning_content is defined and message.reasoning_content is not none %}
+            {%- set reasoning_content = message.reasoning_content %}
+        {%- else %}
+            {%- if '</think>' in message.content %}
+                {%- set content = message.content.split('</think>')[-1].lstrip('\n') %}
+                {%- set reasoning_content = message.content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
+            {%- endif %}
+        {%- endif %}
+        {%- if loop.index0 > ns.last_query_index %}
+            {%- if loop.last or (not loop.last and reasoning_content) %}
+                {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content.strip('\n') + '\n</think>\n\n' + content.lstrip('\n') }}
+            {%- else %}
+                {{- '<|im_start|>' + message.role + '\n' + content }}
+            {%- endif %}
+        {%- else %}
+            {{- '<|im_start|>' + message.role + '\n' + content }}
+        {%- endif %}
+        {%- if message.tool_calls %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if (loop.first and content) or (not loop.first) %}
+                    {{- '\n' }}
+                {%- endif %}
+                {%- if tool_call.function %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {{- '<tool_call>\n{"name": "' }}
+                {{- tool_call.name }}
+                {{- '", "arguments": ' }}
+                {%- if tool_call.arguments is string %}
+                    {{- tool_call.arguments }}
+                {%- else %}
+                    {{- tool_call.arguments | tojson }}
+                {%- endif %}
+                {{- '}\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.first or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+    {%- if enable_thinking is defined and enable_thinking is false %}
+        {{- '<think>\n\n</think>\n\n' }}
+    {%- endif %}
+{%- endif %}
\ No newline at end of file
diff --git a/runs/t0/t0_S4_lr0.0012/checkpoint-864/config.json b/runs/t0/t0_S4_lr0.0012/checkpoint-864/config.json
new file mode 100644
index 0000000000000000000000000000000000000000..ef8760b121e086715460fb47a7acf8f81e61b283
--- /dev/null
+++ b/runs/t0/t0_S4_lr0.0012/checkpoint-864/config.json
@@ -0,0 +1,32 @@
+{
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": null,
+  "dtype": "float32",
+  "eos_token_id": 151645,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 2816,
+  "max_position_embeddings": 2048,
+  "mlp_bias": false,
+  "model_type": "llama",
+  "num_attention_heads": 8,
+  "num_hidden_layers": 24,
+  "num_key_value_heads": 8,
+  "pad_token_id": 151645,
+  "pretraining_tp": 1,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 10000.0,
+    "rope_type": "default"
+  },
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": false,
+  "vocab_size": 151671
+}
diff --git a/runs/t0/t0_S4_lr0.0012/checkpoint-864/generation_config.json b/runs/t0/t0_S4_lr0.0012/checkpoint-864/generation_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..8c4242f32d40df256857f87264f646130f684e33
--- /dev/null
+++ b/runs/t0/t0_S4_lr0.0012/checkpoint-864/generation_config.json
@@ -0,0 +1,11 @@
+{
+  "_from_model_config": true,
+  "eos_token_id": [
+    151645
+  ],
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "transformers_version": "5.15.0",
+  "use_cache": true
+}
diff --git a/runs/t0/t0_S4_lr0.0012/checkpoint-864/tokenizer_config.json b/runs/t0/t0_S4_lr0.0012/checkpoint-864/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..839a3743177a3ed77e1ffbae67520e8500ca0532
--- /dev/null
+++ b/runs/t0/t0_S4_lr0.0012/checkpoint-864/tokenizer_config.json
@@ -0,0 +1,19 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|l2r_pred|>",
+    "<|r2l_pred|>"
+  ],
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 131072,
+  "pad_token": "<|im_end|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}
diff --git a/runs/t0/t0_S4_lr0.0012/checkpoint-864/trainer_state.json b/runs/t0/t0_S4_lr0.0012/checkpoint-864/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..77a02dd7a019d0030acdfcd12541420f503ee9a4
--- /dev/null
+++ b/runs/t0/t0_S4_lr0.0012/checkpoint-864/trainer_state.json
@@ -0,0 +1,739 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 12.0,
+  "eval_steps": 72,
+  "global_step": 864,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.013961605584642234,
+      "grad_norm": 2.7830731868743896,
+      "learning_rate": 0.0,
+      "loss": 12.132288932800293,
+      "step": 1
+    },
+    {
+      "epoch": 0.13961605584642234,
+      "grad_norm": 2.0695760250091553,
+      "learning_rate": 0.00010799999999999998,
+      "loss": 11.117290920681423,
+      "step": 10
+    },
+    {
+      "epoch": 0.2792321116928447,
+      "grad_norm": 1.3685357570648193,
+      "learning_rate": 0.00022799999999999999,
+      "loss": 9.446217346191407,
+      "step": 20
+    },
+    {
+      "epoch": 0.418848167539267,
+      "grad_norm": 0.3884938955307007,
+      "learning_rate": 0.00034799999999999995,
+      "loss": 7.895368957519532,
+      "step": 30
+    },
+    {
+      "epoch": 0.5584642233856894,
+      "grad_norm": 0.487929105758667,
+      "learning_rate": 0.000468,
+      "loss": 7.554646301269531,
+      "step": 40
+    },
+    {
+      "epoch": 0.6980802792321117,
+      "grad_norm": 0.8012627363204956,
+      "learning_rate": 0.000588,
+      "loss": 7.401921081542969,
+      "step": 50
+    },
+    {
+      "epoch": 0.837696335078534,
+      "grad_norm": 0.6215198636054993,
+      "learning_rate": 0.0007079999999999999,
+      "loss": 7.233555603027344,
+      "step": 60
+    },
+    {
+      "epoch": 0.9773123909249564,
+      "grad_norm": 0.9021868705749512,
+      "learning_rate": 0.0008279999999999999,
+      "loss": 7.0121917724609375,
+      "step": 70
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 6.884247779846191,
+      "eval_runtime": 29.6155,
+      "eval_samples_per_second": 82.423,
+      "eval_steps_per_second": 1.317,
+      "step": 72
+    },
+    {
+      "epoch": 1.111692844677138,
+      "grad_norm": 0.7726740837097168,
+      "learning_rate": 0.000948,
+      "loss": 6.848681640625,
+      "step": 80
+    },
+    {
+      "epoch": 1.2513089005235603,
+      "grad_norm": 0.39272984862327576,
+      "learning_rate": 0.001068,
+      "loss": 6.674762725830078,
+      "step": 90
+    },
+    {
+      "epoch": 1.3909249563699826,
+      "grad_norm": 0.6953481435775757,
+      "learning_rate": 0.0011879999999999998,
+      "loss": 6.53614501953125,
+      "step": 100
+    },
+    {
+      "epoch": 1.530541012216405,
+      "grad_norm": 0.566683292388916,
+      "learning_rate": 0.0012,
+      "loss": 6.513034057617188,
+      "step": 110
+    },
+    {
+      "epoch": 1.6701570680628273,
+      "grad_norm": 0.29915207624435425,
+      "learning_rate": 0.0012,
+      "loss": 6.323231887817383,
+      "step": 120
+    },
+    {
+      "epoch": 1.8097731239092496,
+      "grad_norm": 0.4436410665512085,
+      "learning_rate": 0.0012,
+      "loss": 6.200662994384766,
+      "step": 130
+    },
+    {
+      "epoch": 1.949389179755672,
+      "grad_norm": 0.3148385286331177,
+      "learning_rate": 0.0012,
+      "loss": 6.173930740356445,
+      "step": 140
+    },
+    {
+      "epoch": 2.0,
+      "eval_loss": 6.08213472366333,
+      "eval_runtime": 5.4403,
+      "eval_samples_per_second": 448.691,
+      "eval_steps_per_second": 7.169,
+      "step": 144
+    },
+    {
+      "epoch": 2.0837696335078535,
+      "grad_norm": 0.5374071002006531,
+      "learning_rate": 0.0012,
+      "loss": 6.055215835571289,
+      "step": 150
+    },
+    {
+      "epoch": 2.223385689354276,
+      "grad_norm": 0.2520872950553894,
+      "learning_rate": 0.0012,
+      "loss": 5.946698760986328,
+      "step": 160
+    },
+    {
+      "epoch": 2.363001745200698,
+      "grad_norm": 0.44598719477653503,
+      "learning_rate": 0.0012,
+      "loss": 5.896925735473633,
+      "step": 170
+    },
+    {
+      "epoch": 2.5026178010471205,
+      "grad_norm": 0.22086943686008453,
+      "learning_rate": 0.0012,
+      "loss": 5.833065795898437,
+      "step": 180
+    },
+    {
+      "epoch": 2.642233856893543,
+      "grad_norm": 0.599107563495636,
+      "learning_rate": 0.0012,
+      "loss": 5.761848068237304,
+      "step": 190
+    },
+    {
+      "epoch": 2.781849912739965,
+      "grad_norm": 0.2769874930381775,
+      "learning_rate": 0.0012,
+      "loss": 5.68456802368164,
+      "step": 200
+    },
+    {
+      "epoch": 2.9214659685863875,
+      "grad_norm": 0.7282533049583435,
+      "learning_rate": 0.0012,
+      "loss": 5.649662780761719,
+      "step": 210
+    },
+    {
+      "epoch": 3.0,
+      "eval_loss": 5.649172782897949,
+      "eval_runtime": 5.4657,
+      "eval_samples_per_second": 446.6,
+      "eval_steps_per_second": 7.135,
+      "step": 216
+    },
+    {
+      "epoch": 3.055846422338569,
+      "grad_norm": 0.28119179606437683,
+      "learning_rate": 0.0012,
+      "loss": 5.5973163604736325,
+      "step": 220
+    },
+    {
+      "epoch": 3.195462478184991,
+      "grad_norm": 0.1549271196126938,
+      "learning_rate": 0.0012,
+      "loss": 5.466296005249023,
+      "step": 230
+    },
+    {
+      "epoch": 3.3350785340314135,
+      "grad_norm": 0.6101787090301514,
+      "learning_rate": 0.0012,
+      "loss": 5.418326568603516,
+      "step": 240
+    },
+    {
+      "epoch": 3.474694589877836,
+      "grad_norm": 0.27038276195526123,
+      "learning_rate": 0.0012,
+      "loss": 5.406241989135742,
+      "step": 250
+    },
+    {
+      "epoch": 3.614310645724258,
+      "grad_norm": 0.20545881986618042,
+      "learning_rate": 0.0012,
+      "loss": 5.286199951171875,
+      "step": 260
+    },
+    {
+      "epoch": 3.7539267015706805,
+      "grad_norm": 0.5039545297622681,
+      "learning_rate": 0.0012,
+      "loss": 5.312792587280273,
+      "step": 270
+    },
+    {
+      "epoch": 3.893542757417103,
+      "grad_norm": 0.21515797078609467,
+      "learning_rate": 0.0012,
+      "loss": 5.265902709960938,
+      "step": 280
+    },
+    {
+      "epoch": 4.0,
+      "eval_loss": 5.197559356689453,
+      "eval_runtime": 5.439,
+      "eval_samples_per_second": 448.799,
+      "eval_steps_per_second": 7.17,
+      "step": 288
+    },
+    {
+      "epoch": 4.027923211169284,
+      "grad_norm": 0.31526851654052734,
+      "learning_rate": 0.0012,
+      "loss": 5.1200202941894535,
+      "step": 290
+    },
+    {
+      "epoch": 4.167539267015707,
+      "grad_norm": 0.40778130292892456,
+      "learning_rate": 0.0012,
+      "loss": 5.063777923583984,
+      "step": 300
+    },
+    {
+      "epoch": 4.307155322862129,
+      "grad_norm": 0.16718265414237976,
+      "learning_rate": 0.0012,
+      "loss": 4.9782966613769535,
+      "step": 310
+    },
+    {
+      "epoch": 4.446771378708552,
+      "grad_norm": 0.257915735244751,
+      "learning_rate": 0.0012,
+      "loss": 4.8924823760986325,
+      "step": 320
+    },
+    {
+      "epoch": 4.5863874345549736,
+      "grad_norm": 0.5843560099601746,
+      "learning_rate": 0.0012,
+      "loss": 4.866177368164062,
+      "step": 330
+    },
+    {
+      "epoch": 4.726003490401396,
+      "grad_norm": 0.27052536606788635,
+      "learning_rate": 0.0012,
+      "loss": 4.885505676269531,
+      "step": 340
+    },
+    {
+      "epoch": 4.865619546247818,
+      "grad_norm": 0.128607839345932,
+      "learning_rate": 0.0012,
+      "loss": 4.7557823181152346,
+      "step": 350
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 0.3044080436229706,
+      "learning_rate": 0.0012,
+      "loss": 4.6753181457519535,
+      "step": 360
+    },
+    {
+      "epoch": 5.0,
+      "eval_loss": 4.7488484382629395,
+      "eval_runtime": 5.4378,
+      "eval_samples_per_second": 448.891,
+      "eval_steps_per_second": 7.172,
+      "step": 360
+    },
+    {
+      "epoch": 5.139616055846423,
+      "grad_norm": 0.49882882833480835,
+      "learning_rate": 0.0012,
+      "loss": 4.591573333740234,
+      "step": 370
+    },
+    {
+      "epoch": 5.279232111692845,
+      "grad_norm": 0.22627392411231995,
+      "learning_rate": 0.0012,
+      "loss": 4.567809677124023,
+      "step": 380
+    },
+    {
+      "epoch": 5.418848167539267,
+      "grad_norm": 0.17165814340114594,
+      "learning_rate": 0.0012,
+      "loss": 4.4708507537841795,
+      "step": 390
+    },
+    {
+      "epoch": 5.558464223385689,
+      "grad_norm": 0.21155276894569397,
+      "learning_rate": 0.0012,
+      "loss": 4.3975975036621096,
+      "step": 400
+    },
+    {
+      "epoch": 5.698080279232112,
+      "grad_norm": 0.29032468795776367,
+      "learning_rate": 0.0012,
+      "loss": 4.414227676391602,
+      "step": 410
+    },
+    {
+      "epoch": 5.837696335078534,
+      "grad_norm": 0.1890392005443573,
+      "learning_rate": 0.0012,
+      "loss": 4.374155807495117,
+      "step": 420
+    },
+    {
+      "epoch": 5.977312390924956,
+      "grad_norm": 0.17251808941364288,
+      "learning_rate": 0.0012,
+      "loss": 4.308379745483398,
+      "step": 430
+    },
+    {
+      "epoch": 6.0,
+      "eval_loss": 4.4369707107543945,
+      "eval_runtime": 5.4537,
+      "eval_samples_per_second": 447.585,
+      "eval_steps_per_second": 7.151,
+      "step": 432
+    },
+    {
+      "epoch": 6.111692844677138,
+      "grad_norm": 0.2730535566806793,
+      "learning_rate": 0.0012,
+      "loss": 4.220690536499023,
+      "step": 440
+    },
+    {
+      "epoch": 6.2513089005235605,
+      "grad_norm": 0.35956087708473206,
+      "learning_rate": 0.0012,
+      "loss": 4.216027450561524,
+      "step": 450
+    },
+    {
+      "epoch": 6.390924956369982,
+      "grad_norm": 0.17444954812526703,
+      "learning_rate": 0.0012,
+      "loss": 4.1690925598144535,
+      "step": 460
+    },
+    {
+      "epoch": 6.530541012216405,
+      "grad_norm": 0.14641860127449036,
+      "learning_rate": 0.0012,
+      "loss": 4.113532257080078,
+      "step": 470
+    },
+    {
+      "epoch": 6.670157068062827,
+      "grad_norm": 0.1628427356481552,
+      "learning_rate": 0.0012,
+      "loss": 4.076828384399414,
+      "step": 480
+    },
+    {
+      "epoch": 6.80977312390925,
+      "grad_norm": 0.18194392323493958,
+      "learning_rate": 0.0012,
+      "loss": 4.0723823547363285,
+      "step": 490
+    },
+    {
+      "epoch": 6.949389179755672,
+      "grad_norm": 0.18049317598342896,
+      "learning_rate": 0.0012,
+      "loss": 4.0408069610595705,
+      "step": 500
+    },
+    {
+      "epoch": 7.0,
+      "eval_loss": 4.229814052581787,
+      "eval_runtime": 5.4496,
+      "eval_samples_per_second": 447.926,
+      "eval_steps_per_second": 7.157,
+      "step": 504
+    },
+    {
+      "epoch": 7.0837696335078535,
+      "grad_norm": 0.33020293712615967,
+      "learning_rate": 0.0012,
+      "loss": 3.9618515014648437,
+      "step": 510
+    },
+    {
+      "epoch": 7.223385689354275,
+      "grad_norm": 0.25742554664611816,
+      "learning_rate": 0.0012,
+      "loss": 3.991061782836914,
+      "step": 520
+    },
+    {
+      "epoch": 7.363001745200698,
+      "grad_norm": 0.15890459716320038,
+      "learning_rate": 0.0012,
+      "loss": 3.946040725708008,
+      "step": 530
+    },
+    {
+      "epoch": 7.50261780104712,
+      "grad_norm": 0.11660423129796982,
+      "learning_rate": 0.0012,
+      "loss": 3.8959976196289063,
+      "step": 540
+    },
+    {
+      "epoch": 7.642233856893543,
+      "grad_norm": 0.11851505190134048,
+      "learning_rate": 0.0012,
+      "loss": 3.8700828552246094,
+      "step": 550
+    },
+    {
+      "epoch": 7.781849912739965,
+      "grad_norm": 0.24351678788661957,
+      "learning_rate": 0.0012,
+      "loss": 3.8552093505859375,
+      "step": 560
+    },
+    {
+      "epoch": 7.9214659685863875,
+      "grad_norm": 0.15510977804660797,
+      "learning_rate": 0.0012,
+      "loss": 3.8503486633300783,
+      "step": 570
+    },
+    {
+      "epoch": 8.0,
+      "eval_loss": 4.088932514190674,
+      "eval_runtime": 5.4365,
+      "eval_samples_per_second": 449.004,
+      "eval_steps_per_second": 7.174,
+      "step": 576
+    },
+    {
+      "epoch": 8.055846422338568,
+      "grad_norm": 0.27433669567108154,
+      "learning_rate": 0.0012,
+      "loss": 3.7618026733398438,
+      "step": 580
+    },
+    {
+      "epoch": 8.195462478184991,
+      "grad_norm": 0.3087034821510315,
+      "learning_rate": 0.0012,
+      "loss": 3.7158279418945312,
+      "step": 590
+    },
+    {
+      "epoch": 8.335078534031414,
+      "grad_norm": 0.19240885972976685,
+      "learning_rate": 0.0012,
+      "loss": 3.7282508850097655,
+      "step": 600
+    },
+    {
+      "epoch": 8.474694589877837,
+      "grad_norm": 0.12315202504396439,
+      "learning_rate": 0.0012,
+      "loss": 3.6883113861083983,
+      "step": 610
+    },
+    {
+      "epoch": 8.614310645724258,
+      "grad_norm": 0.134666308760643,
+      "learning_rate": 0.0012,
+      "loss": 3.6693538665771483,
+      "step": 620
+    },
+    {
+      "epoch": 8.75392670157068,
+      "grad_norm": 0.11143336445093155,
+      "learning_rate": 0.0012,
+      "loss": 3.6612457275390624,
+      "step": 630
+    },
+    {
+      "epoch": 8.893542757417103,
+      "grad_norm": 0.18448519706726074,
+      "learning_rate": 0.0012,
+      "loss": 3.672064208984375,
+      "step": 640
+    },
+    {
+      "epoch": 9.0,
+      "eval_loss": 4.005725860595703,
+      "eval_runtime": 5.4368,
+      "eval_samples_per_second": 448.974,
+      "eval_steps_per_second": 7.173,
+      "step": 648
+    },
+    {
+      "epoch": 9.027923211169284,
+      "grad_norm": 0.18339823186397552,
+      "learning_rate": 0.0012,
+      "loss": 3.6225852966308594,
+      "step": 650
+    },
+    {
+      "epoch": 9.167539267015707,
+      "grad_norm": 0.49299097061157227,
+      "learning_rate": 0.0012,
+      "loss": 3.556019592285156,
+      "step": 660
+    },
+    {
+      "epoch": 9.30715532286213,
+      "grad_norm": 1.2851951122283936,
+      "learning_rate": 0.0012,
+      "loss": 4.133036804199219,
+      "step": 670
+    },
+    {
+      "epoch": 9.44677137870855,
+      "grad_norm": 1.5363609790802002,
+      "learning_rate": 0.0012,
+      "loss": 5.060147857666015,
+      "step": 680
+    },
+    {
+      "epoch": 9.586387434554974,
+      "grad_norm": 0.5289552211761475,
+      "learning_rate": 0.0012,
+      "loss": 5.2387336730957035,
+      "step": 690
+    },
+    {
+      "epoch": 9.726003490401396,
+      "grad_norm": 0.1746184378862381,
+      "learning_rate": 0.0012,
+      "loss": 4.586937332153321,
+      "step": 700
+    },
+    {
+      "epoch": 9.86561954624782,
+      "grad_norm": 0.27418044209480286,
+      "learning_rate": 0.0012,
+      "loss": 4.650239181518555,
+      "step": 710
+    },
+    {
+      "epoch": 10.0,
+      "grad_norm": 0.15791937708854675,
+      "learning_rate": 0.0012,
+      "loss": 4.335153198242187,
+      "step": 720
+    },
+    {
+      "epoch": 10.0,
+      "eval_loss": 4.460577487945557,
+      "eval_runtime": 5.4539,
+      "eval_samples_per_second": 447.567,
+      "eval_steps_per_second": 7.151,
+      "step": 720
+    },
+    {
+      "epoch": 10.139616055846423,
+      "grad_norm": 0.1130480244755745,
+      "learning_rate": 0.0012,
+      "loss": 4.100911712646484,
+      "step": 730
+    },
+    {
+      "epoch": 10.279232111692846,
+      "grad_norm": 0.13805827498435974,
+      "learning_rate": 0.0012,
+      "loss": 3.986064147949219,
+      "step": 740
+    },
+    {
+      "epoch": 10.418848167539267,
+      "grad_norm": 0.12214994430541992,
+      "learning_rate": 0.0012,
+      "loss": 3.9097930908203127,
+      "step": 750
+    },
+    {
+      "epoch": 10.55846422338569,
+      "grad_norm": 0.10231558978557587,
+      "learning_rate": 0.0012,
+      "loss": 3.8575531005859376,
+      "step": 760
+    },
+    {
+      "epoch": 10.698080279232112,
+      "grad_norm": 0.2263040393590927,
+      "learning_rate": 0.0012,
+      "loss": 3.8134490966796877,
+      "step": 770
+    },
+    {
+      "epoch": 10.837696335078533,
+      "grad_norm": 0.11031533032655716,
+      "learning_rate": 0.0012,
+      "loss": 3.78343505859375,
+      "step": 780
+    },
+    {
+      "epoch": 10.977312390924956,
+      "grad_norm": 0.10530456155538559,
+      "learning_rate": 0.0012,
+      "loss": 3.733928680419922,
+      "step": 790
+    },
+    {
+      "epoch": 11.0,
+      "eval_loss": 4.046730041503906,
+      "eval_runtime": 5.4443,
+      "eval_samples_per_second": 448.359,
+      "eval_steps_per_second": 7.163,
+      "step": 792
+    },
+    {
+      "epoch": 11.111692844677139,
+      "grad_norm": 0.2062959522008896,
+      "learning_rate": 0.0012,
+      "loss": 3.6298500061035157,
+      "step": 800
+    },
+    {
+      "epoch": 11.25130890052356,
+      "grad_norm": 0.11892785876989365,
+      "learning_rate": 0.0012,
+      "loss": 3.5670894622802733,
+      "step": 810
+    },
+    {
+      "epoch": 11.390924956369982,
+      "grad_norm": 0.09811964631080627,
+      "learning_rate": 0.0012,
+      "loss": 3.5630123138427736,
+      "step": 820
+    },
+    {
+      "epoch": 11.530541012216405,
+      "grad_norm": 0.09262749552726746,
+      "learning_rate": 0.0012,
+      "loss": 3.5660240173339846,
+      "step": 830
+    },
+    {
+      "epoch": 11.670157068062828,
+      "grad_norm": 0.10465727001428604,
+      "learning_rate": 0.0012,
+      "loss": 3.560059356689453,
+      "step": 840
+    },
+    {
+      "epoch": 11.809773123909249,
+      "grad_norm": 0.11480753123760223,
+      "learning_rate": 0.0012,
+      "loss": 3.555225372314453,
+      "step": 850
+    },
+    {
+      "epoch": 11.949389179755672,
+      "grad_norm": 0.09398435801267624,
+      "learning_rate": 0.0012,
+      "loss": 3.5535049438476562,
+      "step": 860
+    },
+    {
+      "epoch": 12.0,
+      "eval_loss": 3.960569381713867,
+      "eval_runtime": 5.433,
+      "eval_samples_per_second": 449.291,
+      "eval_steps_per_second": 7.178,
+      "step": 864
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 864,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 12,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": false,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.6673046802726912e+18,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/runs/t0/t0_S4_lr0.0012/train.log b/runs/t0/t0_S4_lr0.0012/train.log
new file mode 100644
index 0000000000000000000000000000000000000000..32efb56afab1fbbd2aafad10dcec3ed973f1be7f
--- /dev/null
+++ b/runs/t0/t0_S4_lr0.0012/train.log
@@ -0,0 +1,1644 @@
+W0817 21:10:13.757000 2292876 torch/distributed/run.py:851] 
+W0817 21:10:13.757000 2292876 torch/distributed/run.py:851] *****************************************
+W0817 21:10:13.757000 2292876 torch/distributed/run.py:851] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+W0817 21:10:13.757000 2292876 torch/distributed/run.py:851] *****************************************
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+EOS token: <|im_end|>
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+total params: 463642624
+[dataset] loaded packed cache: 36614 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+total params: 463642624
+[dataset] loaded packed cache: 36614 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+total params: 463642624
+[dataset] loaded packed cache: 36614 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+total params: 463642624
+[dataset] loaded packed cache: 36614 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+total params: 463642624
+[dataset] loaded packed cache: 36614 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+total params: 463642624
+[dataset] loaded packed cache: 36614 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+total params: 463642624
+[dataset] loaded packed cache: 36614 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+total params: 463642624
+[dataset] loaded packed cache: 36614 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+[transformers] ***** Running training *****
+[transformers]   Num examples = 36,614
+[transformers]   Num Epochs = 12
+[transformers]   Num update steps per epoch = 72
+[transformers]   Instantaneous batch size per device = 8
+[transformers]   Total train batch size (w. parallel, distributed & accumulation) = 512
+[transformers]   Gradient Accumulation steps = 8
+[transformers]   Total optimization steps = 864
+[transformers]   Number of trainable parameters = 463,642,624
+wandb: WARNING Changes to your `wandb` environment variables will be ignored because your `wandb` session has already started. For more information on how to modify your settings with `wandb.init()` arguments, please refer to https://wandb.me/wandb-init.
+wandb: setting up run sc3anwdm
+wandb: Tracking run with wandb version 0.28.2
+wandb: Run data is saved locally in /home/ubuntu/workspace/data-augmentations-for-pretraining/wandb/run-20260817_211038-sc3anwdm
+wandb: Run `wandb offline` to turn off syncing.
+wandb: Syncing run t0_S4_lr0.0012
+wandb: ⭐️ View project at https://wandb.ai/slowrun-wankers/data-aug-pretraining
+wandb: 🚀 View run at https://wandb.ai/slowrun-wankers/data-aug-pretraining/runs/sc3anwdm
+
  0%|          | 0/864 [00:00<?, ?it/s]wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+[rank6]:W0817 21:11:01.301000 2292986 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+[rank7]:W0817 21:11:01.361000 2292987 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+[rank1]:W0817 21:11:01.475000 2292981 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+[rank3]:W0817 21:11:01.632000 2292983 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+[rank2]:W0817 21:11:01.728000 2292982 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+[rank4]:W0817 21:11:01.757000 2292984 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+[rank5]:W0817 21:11:02.020000 2292985 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+[rank0]:W0817 21:11:02.255000 2292980 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+
  0%|          | 1/864 [00:44<10:36:22, 44.24s/it]
                                                  
{'loss': '12.13', 'grad_norm': '2.783', 'learning_rate': '0', 'epoch': '0.01396'}
+
  0%|          | 1/864 [00:44<10:36:22, 44.24s/it]
  0%|          | 2/864 [00:45<4:33:48, 19.06s/it] 
  0%|          | 3/864 [00:46<2:37:08, 10.95s/it]
  0%|          | 4/864 [00:48<1:42:21,  7.14s/it]
  1%|          | 5/864 [00:49<1:12:05,  5.04s/it]
  1%|          | 6/864 [00:50<53:50,  3.76s/it]  
  1%|          | 7/864 [00:52<42:15,  2.96s/it]
  1%|          | 8/864 [00:53<34:40,  2.43s/it]
  1%|          | 9/864 [00:54<29:35,  2.08s/it]
  1%|          | 10/864 [00:56<26:08,  1.84s/it]
                                                
{'loss': '11.12', 'grad_norm': '2.07', 'learning_rate': '0.000108', 'epoch': '0.1396'}
+
  1%|          | 10/864 [00:56<26:08,  1.84s/it]
  1%|▏         | 11/864 [00:57<23:46,  1.67s/it]
  1%|▏         | 12/864 [00:58<22:07,  1.56s/it]
  2%|▏         | 13/864 [00:59<21:00,  1.48s/it]
  2%|▏         | 14/864 [01:01<22:01,  1.55s/it]
  2%|▏         | 15/864 [01:03<23:02,  1.63s/it]
  2%|▏         | 16/864 [01:04<21:35,  1.53s/it]
  2%|▏         | 17/864 [01:06<20:34,  1.46s/it]
  2%|▏         | 18/864 [01:07<19:52,  1.41s/it]
  2%|▏         | 19/864 [01:08<19:23,  1.38s/it]
  2%|▏         | 20/864 [01:09<19:01,  1.35s/it]
                                                
{'loss': '9.446', 'grad_norm': '1.369', 'learning_rate': '0.000228', 'epoch': '0.2792'}
+
  2%|▏         | 20/864 [01:09<19:01,  1.35s/it]
  2%|▏         | 21/864 [01:11<18:45,  1.33s/it]
  3%|▎         | 22/864 [01:12<18:34,  1.32s/it]
  3%|▎         | 23/864 [01:13<18:26,  1.32s/it]
  3%|▎         | 24/864 [01:15<18:20,  1.31s/it]
  3%|▎         | 25/864 [01:16<18:16,  1.31s/it]
  3%|▎         | 26/864 [01:17<18:14,  1.31s/it]
  3%|▎         | 27/864 [01:19<18:11,  1.30s/it]
  3%|▎         | 28/864 [01:20<18:09,  1.30s/it]
  3%|▎         | 29/864 [01:21<18:07,  1.30s/it]
  3%|▎         | 30/864 [01:22<18:07,  1.30s/it]
                                                
{'loss': '7.895', 'grad_norm': '0.3885', 'learning_rate': '0.000348', 'epoch': '0.4188'}
+
  3%|▎         | 30/864 [01:22<18:07,  1.30s/it]
  4%|▎         | 31/864 [01:24<18:05,  1.30s/it]
  4%|▎         | 32/864 [01:25<18:03,  1.30s/it]
  4%|▍         | 33/864 [01:26<18:02,  1.30s/it]
  4%|▍         | 34/864 [01:28<18:00,  1.30s/it]
  4%|▍         | 35/864 [01:29<18:00,  1.30s/it]
  4%|▍         | 36/864 [01:30<17:59,  1.30s/it]
  4%|▍         | 37/864 [01:32<17:58,  1.30s/it]
  4%|▍         | 38/864 [01:33<17:56,  1.30s/it]
  5%|▍         | 39/864 [01:34<17:54,  1.30s/it]
  5%|▍         | 40/864 [01:36<17:54,  1.30s/it]
                                                
{'loss': '7.555', 'grad_norm': '0.4879', 'learning_rate': '0.000468', 'epoch': '0.5585'}
+
  5%|▍         | 40/864 [01:36<17:54,  1.30s/it]
  5%|▍         | 41/864 [01:37<17:54,  1.31s/it]
  5%|▍         | 42/864 [01:38<17:52,  1.31s/it]
  5%|▍         | 43/864 [01:39<17:52,  1.31s/it]
  5%|▌         | 44/864 [01:41<17:51,  1.31s/it]
  5%|▌         | 45/864 [01:42<17:49,  1.31s/it]
  5%|▌         | 46/864 [01:43<17:48,  1.31s/it]
  5%|▌         | 47/864 [01:45<17:48,  1.31s/it]
  6%|▌         | 48/864 [01:46<17:47,  1.31s/it]
  6%|▌         | 49/864 [01:47<17:46,  1.31s/it]
  6%|▌         | 50/864 [01:49<17:45,  1.31s/it]
                                                
{'loss': '7.402', 'grad_norm': '0.8013', 'learning_rate': '0.000588', 'epoch': '0.6981'}
+
  6%|▌         | 50/864 [01:49<17:45,  1.31s/it]
  6%|▌         | 51/864 [01:50<17:44,  1.31s/it]
  6%|▌         | 52/864 [01:51<17:43,  1.31s/it]
  6%|▌         | 53/864 [01:53<17:42,  1.31s/it]
  6%|▋         | 54/864 [01:54<17:41,  1.31s/it]
  6%|▋         | 55/864 [01:55<17:39,  1.31s/it]
  6%|▋         | 56/864 [01:56<17:38,  1.31s/it]
  7%|▋         | 57/864 [01:58<17:37,  1.31s/it]
  7%|▋         | 58/864 [01:59<17:38,  1.31s/it]
  7%|▋         | 59/864 [02:00<17:36,  1.31s/it]
  7%|▋         | 60/864 [02:02<17:34,  1.31s/it]
                                                
{'loss': '7.234', 'grad_norm': '0.6215', 'learning_rate': '0.000708', 'epoch': '0.8377'}
+
  7%|▋         | 60/864 [02:02<17:34,  1.31s/it]
  7%|▋         | 61/864 [02:03<17:32,  1.31s/it]
  7%|▋         | 62/864 [02:04<17:31,  1.31s/it]
  7%|▋         | 63/864 [02:06<17:31,  1.31s/it]
  7%|▋         | 64/864 [02:07<17:30,  1.31s/it]
  8%|▊         | 65/864 [02:08<17:28,  1.31s/it]
  8%|▊         | 66/864 [02:10<17:28,  1.31s/it]
  8%|▊         | 67/864 [02:11<17:27,  1.31s/it]
  8%|▊         | 68/864 [02:12<17:25,  1.31s/it]
  8%|▊         | 69/864 [02:14<17:24,  1.31s/it]
  8%|▊         | 70/864 [02:15<17:24,  1.32s/it]
                                                
{'loss': '7.012', 'grad_norm': '0.9022', 'learning_rate': '0.000828', 'epoch': '0.9773'}
+
  8%|▊         | 70/864 [02:15<17:24,  1.32s/it]
  8%|▊         | 71/864 [02:16<17:22,  1.31s/it]
  8%|▊         | 72/864 [02:17<15:25,  1.17s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:02, 16.36it/s]
+
 13%|█▎        | 5/39 [00:00<00:03, 10.12it/s]
+
 18%|█▊        | 7/39 [00:00<00:03,  8.70it/s]
+
 21%|██        | 8/39 [00:00<00:03,  8.30it/s]
+
 23%|██▎       | 9/39 [00:01<00:03,  8.03it/s]
+
 26%|██▌       | 10/39 [00:01<00:03,  7.81it/s]
+
 28%|██▊       | 11/39 [00:01<00:03,  7.65it/s]
+
 31%|███       | 12/39 [00:01<00:03,  7.53it/s]
+
 33%|███▎      | 13/39 [00:01<00:03,  7.44it/s]
+
 36%|███▌      | 14/39 [00:01<00:03,  7.38it/s]
+
 38%|███▊      | 15/39 [00:01<00:03,  7.34it/s]
+
 41%|████      | 16/39 [00:01<00:03,  7.30it/s]
+
 44%|████▎     | 17/39 [00:02<00:03,  7.27it/s]
+
 46%|████▌     | 18/39 [00:02<00:02,  7.27it/s]
+
 49%|████▊     | 19/39 [00:02<00:02,  7.27it/s]
+
 51%|█████▏    | 20/39 [00:02<00:02,  7.26it/s]
+
 54%|█████▍    | 21/39 [00:02<00:02,  7.26it/s]
+
 56%|█████▋    | 22/39 [00:02<00:02,  7.26it/s]
+
 59%|█████▉    | 23/39 [00:02<00:02,  7.26it/s]
+
 62%|██████▏   | 24/39 [00:03<00:02,  7.26it/s]
+
 64%|██████▍   | 25/39 [00:03<00:01,  7.25it/s]
+
 67%|██████▋   | 26/39 [00:03<00:01,  7.24it/s]
+
 69%|██████▉   | 27/39 [00:03<00:01,  7.24it/s]
+
 72%|███████▏  | 28/39 [00:03<00:01,  7.24it/s]
+
 74%|███████▍  | 29/39 [00:03<00:01,  7.23it/s]
+
 77%|███████▋  | 30/39 [00:03<00:01,  7.22it/s]
+
 79%|███████▉  | 31/39 [00:04<00:01,  7.23it/s]
+
 82%|████████▏ | 32/39 [00:04<00:00,  7.23it/s]
+
 85%|████████▍ | 33/39 [00:04<00:00,  7.24it/s]
+
 87%|████████▋ | 34/39 [00:04<00:00,  7.25it/s]
+
 90%|████████▉ | 35/39 [00:04<00:00,  7.25it/s]
+
 92%|█████████▏| 36/39 [00:04<00:00,  7.25it/s]
+
 95%|█████████▍| 37/39 [00:04<00:00,  7.25it/s]
+
 97%|█████████▋| 38/39 [00:05<00:00,  7.25it/s]
+
100%|██████████| 39/39 [00:05<00:00,  7.25it/s]
                                                
+
                                               
{'eval_loss': '6.884', 'eval_runtime': '29.62', 'eval_samples_per_second': '82.42', 'eval_steps_per_second': '1.317', 'epoch': '1'}
+
  8%|▊         | 72/864 [02:47<15:25,  1.17s/it]
+
100%|██████████| 39/39 [00:05<00:00,  7.25it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0012/checkpoint-72
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0012/checkpoint-72/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0012/checkpoint-72/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:01<00:00,  1.83s/it]
Writing model shards: 100%|██████████| 1/1 [00:01<00:00,  1.83s/it]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0012/checkpoint-72/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0012/checkpoint-72/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0012/checkpoint-72/tokenizer_config.json
+[eval] epoch 1.0 step 72 loss 6.8842 (best 6.8842 @ ep 1.0, 0/10000 since best)
+
+
                                               
  8%|▊         | 73/864 [02:53<2:35:01, 11.76s/it]
  9%|▊         | 74/864 [02:55<1:53:33,  8.62s/it]
  9%|▊         | 75/864 [02:56<1:24:33,  6.43s/it]
  9%|▉         | 76/864 [02:57<1:04:16,  4.89s/it]
  9%|▉         | 77/864 [02:59<50:05,  3.82s/it]  
  9%|▉         | 78/864 [03:00<40:11,  3.07s/it]
  9%|▉         | 79/864 [03:01<33:14,  2.54s/it]
  9%|▉         | 80/864 [03:03<28:22,  2.17s/it]
                                                
{'loss': '6.849', 'grad_norm': '0.7727', 'learning_rate': '0.000948', 'epoch': '1.112'}
+
  9%|▉         | 80/864 [03:03<28:22,  2.17s/it]
  9%|▉         | 81/864 [03:04<24:58,  1.91s/it]
  9%|▉         | 82/864 [03:05<22:35,  1.73s/it]
 10%|▉         | 83/864 [03:07<20:56,  1.61s/it]
 10%|▉         | 84/864 [03:08<19:46,  1.52s/it]
 10%|▉         | 85/864 [03:09<18:55,  1.46s/it]
 10%|▉         | 86/864 [03:11<18:20,  1.41s/it]
 10%|█         | 87/864 [03:12<17:55,  1.38s/it]
 10%|█         | 88/864 [03:13<17:37,  1.36s/it]
 10%|█         | 89/864 [03:14<17:24,  1.35s/it]
 10%|█         | 90/864 [03:16<17:15,  1.34s/it]
                                                
{'loss': '6.675', 'grad_norm': '0.3927', 'learning_rate': '0.001068', 'epoch': '1.251'}
+
 10%|█         | 90/864 [03:16<17:15,  1.34s/it]
 11%|█         | 91/864 [03:17<17:08,  1.33s/it]
 11%|█         | 92/864 [03:18<17:03,  1.33s/it]
 11%|█         | 93/864 [03:20<16:58,  1.32s/it]
 11%|█         | 94/864 [03:21<16:55,  1.32s/it]
 11%|█         | 95/864 [03:22<16:52,  1.32s/it]
 11%|█         | 96/864 [03:24<16:50,  1.32s/it]
 11%|█         | 97/864 [03:25<16:47,  1.31s/it]
 11%|█▏        | 98/864 [03:26<16:45,  1.31s/it]
 11%|█▏        | 99/864 [03:28<16:44,  1.31s/it]
 12%|█▏        | 100/864 [03:29<16:42,  1.31s/it]
                                                 
{'loss': '6.536', 'grad_norm': '0.6953', 'learning_rate': '0.001188', 'epoch': '1.391'}
+
 12%|█▏        | 100/864 [03:29<16:42,  1.31s/it]
 12%|█▏        | 101/864 [03:30<16:41,  1.31s/it]
 12%|█▏        | 102/864 [03:32<16:40,  1.31s/it]
 12%|█▏        | 103/864 [03:33<16:38,  1.31s/it]
 12%|█▏        | 104/864 [03:34<16:37,  1.31s/it]
 12%|█▏        | 105/864 [03:35<16:36,  1.31s/it]
 12%|█▏        | 106/864 [03:37<16:35,  1.31s/it]
 12%|█▏        | 107/864 [03:38<16:33,  1.31s/it]
 12%|█▎        | 108/864 [03:39<16:32,  1.31s/it]
 13%|█▎        | 109/864 [03:41<16:31,  1.31s/it]
 13%|█▎        | 110/864 [03:42<16:29,  1.31s/it]
                                                 
{'loss': '6.513', 'grad_norm': '0.5667', 'learning_rate': '0.0012', 'epoch': '1.531'}
+
 13%|█▎        | 110/864 [03:42<16:29,  1.31s/it]
 13%|█▎        | 111/864 [03:43<16:28,  1.31s/it]
 13%|█▎        | 112/864 [03:45<16:27,  1.31s/it]
 13%|█▎        | 113/864 [03:46<16:25,  1.31s/it]
 13%|█▎        | 114/864 [03:47<16:24,  1.31s/it]
 13%|█▎        | 115/864 [03:49<16:22,  1.31s/it]
 13%|█▎        | 116/864 [03:50<16:21,  1.31s/it]
 14%|█▎        | 117/864 [03:51<16:20,  1.31s/it]
 14%|█▎        | 118/864 [03:53<16:18,  1.31s/it]
 14%|█▍        | 119/864 [03:54<16:18,  1.31s/it]
 14%|█▍        | 120/864 [03:55<16:17,  1.31s/it]
                                                 
{'loss': '6.323', 'grad_norm': '0.2992', 'learning_rate': '0.0012', 'epoch': '1.67'}
+
 14%|█▍        | 120/864 [03:55<16:17,  1.31s/it]
 14%|█▍        | 121/864 [03:56<16:15,  1.31s/it]
 14%|█▍        | 122/864 [03:58<16:14,  1.31s/it]
 14%|█▍        | 123/864 [03:59<16:12,  1.31s/it]
 14%|█▍        | 124/864 [04:00<16:11,  1.31s/it]
 14%|█▍        | 125/864 [04:02<16:10,  1.31s/it]
 15%|█▍        | 126/864 [04:03<16:08,  1.31s/it]
 15%|█▍        | 127/864 [04:04<16:07,  1.31s/it]
 15%|█▍        | 128/864 [04:06<16:06,  1.31s/it]
 15%|█▍        | 129/864 [04:07<16:04,  1.31s/it]
 15%|█▌        | 130/864 [04:08<16:03,  1.31s/it]
                                                 
{'loss': '6.201', 'grad_norm': '0.4436', 'learning_rate': '0.0012', 'epoch': '1.81'}
+
 15%|█▌        | 130/864 [04:08<16:03,  1.31s/it]
 15%|█▌        | 131/864 [04:10<16:02,  1.31s/it]
 15%|█▌        | 132/864 [04:11<16:00,  1.31s/it]
 15%|█▌        | 133/864 [04:12<15:59,  1.31s/it]
 16%|█▌        | 134/864 [04:14<15:57,  1.31s/it]
 16%|█▌        | 135/864 [04:15<15:56,  1.31s/it]
 16%|█▌        | 136/864 [04:16<15:54,  1.31s/it]
 16%|█▌        | 137/864 [04:17<15:53,  1.31s/it]
 16%|█▌        | 138/864 [04:19<15:52,  1.31s/it]
 16%|█▌        | 139/864 [04:20<15:50,  1.31s/it]
 16%|█▌        | 140/864 [04:21<15:50,  1.31s/it]
                                                 
{'loss': '6.174', 'grad_norm': '0.3148', 'learning_rate': '0.0012', 'epoch': '1.949'}
+
 16%|█▌        | 140/864 [04:21<15:50,  1.31s/it]
 16%|█▋        | 141/864 [04:23<15:50,  1.31s/it]
 16%|█▋        | 142/864 [04:24<15:49,  1.32s/it]
 17%|█▋        | 143/864 [04:25<15:47,  1.31s/it]
 17%|█▋        | 144/864 [04:26<14:01,  1.17s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:02, 15.63it/s]
+
 10%|█         | 4/39 [00:00<00:03,  9.28it/s]
+
 15%|█▌        | 6/39 [00:00<00:04,  8.22it/s]
+
 18%|█▊        | 7/39 [00:00<00:04,  7.95it/s]
+
 21%|██        | 8/39 [00:00<00:04,  7.74it/s]
+
 23%|██▎       | 9/39 [00:01<00:03,  7.60it/s]
+
 26%|██▌       | 10/39 [00:01<00:03,  7.50it/s]
+
 28%|██▊       | 11/39 [00:01<00:03,  7.42it/s]
+
 31%|███       | 12/39 [00:01<00:03,  7.37it/s]
+
 33%|███▎      | 13/39 [00:01<00:03,  7.32it/s]
+
 36%|███▌      | 14/39 [00:01<00:03,  7.30it/s]
+
 38%|███▊      | 15/39 [00:01<00:03,  7.28it/s]
+
 41%|████      | 16/39 [00:02<00:03,  7.26it/s]
+
 44%|████▎     | 17/39 [00:02<00:03,  7.27it/s]
+
 46%|████▌     | 18/39 [00:02<00:02,  7.26it/s]
+
 49%|████▊     | 19/39 [00:02<00:02,  7.26it/s]
+
 51%|█████▏    | 20/39 [00:02<00:02,  7.25it/s]
+
 54%|█████▍    | 21/39 [00:02<00:02,  7.25it/s]
+
 56%|█████▋    | 22/39 [00:02<00:02,  7.25it/s]
+
 59%|█████▉    | 23/39 [00:03<00:02,  7.25it/s]
+
 62%|██████▏   | 24/39 [00:03<00:02,  7.24it/s]
+
 64%|██████▍   | 25/39 [00:03<00:01,  7.24it/s]
+
 67%|██████▋   | 26/39 [00:03<00:01,  7.24it/s]
+
 69%|██████▉   | 27/39 [00:03<00:01,  7.25it/s]
+
 72%|███████▏  | 28/39 [00:03<00:01,  7.25it/s]
+
 74%|███████▍  | 29/39 [00:03<00:01,  7.24it/s]
+
 77%|███████▋  | 30/39 [00:03<00:01,  7.25it/s]
+
 79%|███████▉  | 31/39 [00:04<00:01,  7.25it/s]
+
 82%|████████▏ | 32/39 [00:04<00:00,  7.24it/s]
+
 85%|████████▍ | 33/39 [00:04<00:00,  7.24it/s]
+
 87%|████████▋ | 34/39 [00:04<00:00,  7.25it/s]
+
 90%|████████▉ | 35/39 [00:04<00:00,  7.25it/s]
+
 92%|█████████▏| 36/39 [00:04<00:00,  7.24it/s]
+
 95%|█████████▍| 37/39 [00:04<00:00,  7.24it/s]
+
 97%|█████████▋| 38/39 [00:05<00:00,  7.24it/s]
+
100%|██████████| 39/39 [00:05<00:00,  7.23it/s]
                                                 
+
                                               
{'eval_loss': '6.082', 'eval_runtime': '5.44', 'eval_samples_per_second': '448.7', 'eval_steps_per_second': '7.169', 'epoch': '2'}
+
 17%|█▋        | 144/864 [04:32<14:01,  1.17s/it]
+
100%|██████████| 39/39 [00:05<00:00,  7.23it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0012/checkpoint-144
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0012/checkpoint-144/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0012/checkpoint-144/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:01<00:00,  1.52s/it]
Writing model shards: 100%|██████████| 1/1 [00:01<00:00,  1.52s/it]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0012/checkpoint-144/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0012/checkpoint-144/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0012/checkpoint-144/tokenizer_config.json
+[eval] epoch 2.0 step 144 loss 6.0821 (best 6.0821 @ ep 2.0, 0/10000 since best)
+
+
                                               
 17%|█▋        | 145/864 [04:38<54:05,  4.51s/it]
 17%|█▋        | 146/864 [04:40<42:30,  3.55s/it]
 17%|█▋        | 147/864 [04:41<34:25,  2.88s/it]
 17%|█▋        | 148/864 [04:42<28:45,  2.41s/it]
 17%|█▋        | 149/864 [04:44<24:48,  2.08s/it]
 17%|█▋        | 150/864 [04:45<22:01,  1.85s/it]
                                                 
{'loss': '6.055', 'grad_norm': '0.5374', 'learning_rate': '0.0012', 'epoch': '2.084'}
+
 17%|█▋        | 150/864 [04:45<22:01,  1.85s/it]
 17%|█▋        | 151/864 [04:46<20:04,  1.69s/it]
 18%|█▊        | 152/864 [04:48<18:43,  1.58s/it]
 18%|█▊        | 153/864 [04:49<17:45,  1.50s/it]
 18%|█▊        | 154/864 [04:50<17:05,  1.44s/it]
 18%|█▊        | 155/864 [04:52<16:35,  1.40s/it]
 18%|█▊        | 156/864 [04:53<16:15,  1.38s/it]
 18%|█▊        | 157/864 [04:54<16:00,  1.36s/it]
 18%|█▊        | 158/864 [04:56<15:50,  1.35s/it]
 18%|█▊        | 159/864 [04:57<15:43,  1.34s/it]
 19%|█▊        | 160/864 [04:58<15:36,  1.33s/it]
                                                 
{'loss': '5.947', 'grad_norm': '0.2521', 'learning_rate': '0.0012', 'epoch': '2.223'}
+
 19%|█▊        | 160/864 [04:58<15:36,  1.33s/it]
 19%|█▊        | 161/864 [05:00<15:32,  1.33s/it]
 19%|█▉        | 162/864 [05:01<15:28,  1.32s/it]
 19%|█▉        | 163/864 [05:02<15:25,  1.32s/it]
 19%|█▉        | 164/864 [05:03<15:22,  1.32s/it]
 19%|█▉        | 165/864 [05:05<15:21,  1.32s/it]
 19%|█▉        | 166/864 [05:06<15:19,  1.32s/it]
 19%|█▉        | 167/864 [05:07<15:18,  1.32s/it]
 19%|█▉        | 168/864 [05:09<15:16,  1.32s/it]
 20%|█▉        | 169/864 [05:10<15:14,  1.32s/it]
 20%|█▉        | 170/864 [05:11<15:13,  1.32s/it]
                                                 
{'loss': '5.897', 'grad_norm': '0.446', 'learning_rate': '0.0012', 'epoch': '2.363'}
+
 20%|█▉        | 170/864 [05:11<15:13,  1.32s/it]
 20%|█▉        | 171/864 [05:13<15:11,  1.32s/it]
 20%|█▉        | 172/864 [05:14<15:09,  1.31s/it]
 20%|██        | 173/864 [05:15<15:08,  1.32s/it]
 20%|██        | 174/864 [05:17<15:07,  1.32s/it]
 20%|██        | 175/864 [05:18<15:06,  1.32s/it]
 20%|██        | 176/864 [05:19<15:04,  1.31s/it]
 20%|██        | 177/864 [05:21<15:03,  1.32s/it]
 21%|██        | 178/864 [05:22<15:02,  1.32s/it]
 21%|██        | 179/864 [05:23<15:01,  1.32s/it]
 21%|██        | 180/864 [05:25<15:00,  1.32s/it]
                                                 
{'loss': '5.833', 'grad_norm': '0.2209', 'learning_rate': '0.0012', 'epoch': '2.503'}
+
 21%|██        | 180/864 [05:25<15:00,  1.32s/it]
 21%|██        | 181/864 [05:26<14:58,  1.32s/it]
 21%|██        | 182/864 [05:27<14:57,  1.32s/it]
 21%|██        | 183/864 [05:28<14:55,  1.32s/it]
 21%|██▏       | 184/864 [05:30<14:54,  1.31s/it]
 21%|██▏       | 185/864 [05:31<14:52,  1.31s/it]
 22%|██▏       | 186/864 [05:32<14:51,  1.31s/it]
 22%|██▏       | 187/864 [05:34<14:49,  1.31s/it]
 22%|██▏       | 188/864 [05:35<14:47,  1.31s/it]
 22%|██▏       | 189/864 [05:36<14:47,  1.31s/it]
 22%|██▏       | 190/864 [05:38<14:45,  1.31s/it]
                                                 
{'loss': '5.762', 'grad_norm': '0.5991', 'learning_rate': '0.0012', 'epoch': '2.642'}
+
 22%|██▏       | 190/864 [05:38<14:45,  1.31s/it]
 22%|██▏       | 191/864 [05:39<14:44,  1.31s/it]
 22%|██▏       | 192/864 [05:40<14:43,  1.31s/it]
 22%|██▏       | 193/864 [05:42<14:42,  1.31s/it]
 22%|██▏       | 194/864 [05:43<14:40,  1.31s/it]
 23%|██▎       | 195/864 [05:44<14:39,  1.31s/it]
 23%|██▎       | 196/864 [05:46<14:38,  1.31s/it]
 23%|██▎       | 197/864 [05:47<14:37,  1.32s/it]
 23%|██▎       | 198/864 [05:48<14:36,  1.32s/it]
 23%|██▎       | 199/864 [05:49<14:34,  1.32s/it]
 23%|██▎       | 200/864 [05:51<14:33,  1.32s/it]
                                                 
{'loss': '5.685', 'grad_norm': '0.277', 'learning_rate': '0.0012', 'epoch': '2.782'}
+
 23%|██▎       | 200/864 [05:51<14:33,  1.32s/it]
 23%|██▎       | 201/864 [05:52<14:32,  1.32s/it]
 23%|██▎       | 202/864 [05:53<14:31,  1.32s/it]
 23%|██▎       | 203/864 [05:55<14:30,  1.32s/it]
 24%|██▎       | 204/864 [05:56<14:28,  1.32s/it]
 24%|██▎       | 205/864 [05:57<14:26,  1.32s/it]
 24%|██▍       | 206/864 [05:59<14:26,  1.32s/it]
 24%|██▍       | 207/864 [06:00<14:24,  1.32s/it]
 24%|██▍       | 208/864 [06:01<14:23,  1.32s/it]
 24%|██▍       | 209/864 [06:03<14:22,  1.32s/it]
 24%|██▍       | 210/864 [06:04<14:20,  1.32s/it]
                                                 
{'loss': '5.65', 'grad_norm': '0.7283', 'learning_rate': '0.0012', 'epoch': '2.921'}
+
 24%|██▍       | 210/864 [06:04<14:20,  1.32s/it]
 24%|██▍       | 211/864 [06:05<14:19,  1.32s/it]
 25%|██▍       | 212/864 [06:07<14:18,  1.32s/it]
 25%|██▍       | 213/864 [06:08<14:17,  1.32s/it]
 25%|██▍       | 214/864 [06:09<14:16,  1.32s/it]
 25%|██▍       | 215/864 [06:11<14:14,  1.32s/it]
 25%|██▌       | 216/864 [06:11<12:38,  1.17s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:02, 14.71it/s]
+
 10%|█         | 4/39 [00:00<00:03,  9.15it/s]
+
 15%|█▌        | 6/39 [00:00<00:04,  8.17it/s]
+
 18%|█▊        | 7/39 [00:00<00:04,  7.92it/s]
+
 21%|██        | 8/39 [00:00<00:04,  7.72it/s]
+
 23%|██▎       | 9/39 [00:01<00:03,  7.59it/s]
+
 26%|██▌       | 10/39 [00:01<00:03,  7.49it/s]
+
 28%|██▊       | 11/39 [00:01<00:03,  7.42it/s]
+
 31%|███       | 12/39 [00:01<00:03,  7.37it/s]
+
 33%|███▎      | 13/39 [00:01<00:03,  7.32it/s]
+
 36%|███▌      | 14/39 [00:01<00:03,  7.28it/s]
+
 38%|███▊      | 15/39 [00:01<00:03,  7.26it/s]
+
 41%|████      | 16/39 [00:02<00:03,  7.24it/s]
+
 44%|████▎     | 17/39 [00:02<00:03,  7.25it/s]
+
 46%|████▌     | 18/39 [00:02<00:02,  7.23it/s]
+
 49%|████▊     | 19/39 [00:02<00:02,  7.23it/s]
+
 51%|█████▏    | 20/39 [00:02<00:02,  7.23it/s]
+
 54%|█████▍    | 21/39 [00:02<00:02,  7.23it/s]
+
 56%|█████▋    | 22/39 [00:02<00:02,  7.24it/s]
+
 59%|█████▉    | 23/39 [00:03<00:02,  7.23it/s]
+
 62%|██████▏   | 24/39 [00:03<00:02,  7.23it/s]
+
 64%|██████▍   | 25/39 [00:03<00:01,  7.22it/s]
+
 67%|██████▋   | 26/39 [00:03<00:01,  7.22it/s]
+
 69%|██████▉   | 27/39 [00:03<00:01,  7.21it/s]
+
 72%|███████▏  | 28/39 [00:03<00:01,  7.22it/s]
+
 74%|███████▍  | 29/39 [00:03<00:01,  7.22it/s]
+
 77%|███████▋  | 30/39 [00:04<00:01,  7.23it/s]
+
 79%|███████▉  | 31/39 [00:04<00:01,  7.23it/s]
+
 82%|████████▏ | 32/39 [00:04<00:00,  7.23it/s]
+
 85%|████████▍ | 33/39 [00:04<00:00,  7.24it/s]
+
 87%|████████▋ | 34/39 [00:04<00:00,  7.24it/s]
+
 90%|████████▉ | 35/39 [00:04<00:00,  7.24it/s]
+
 92%|█████████▏| 36/39 [00:04<00:00,  7.21it/s]
+
 95%|█████████▍| 37/39 [00:04<00:00,  7.22it/s]
+
 97%|█████████▋| 38/39 [00:05<00:00,  7.22it/s]
+
100%|██████████| 39/39 [00:05<00:00,  7.23it/s]
                                                 
+
                                               
{'eval_loss': '5.649', 'eval_runtime': '5.466', 'eval_samples_per_second': '446.6', 'eval_steps_per_second': '7.135', 'epoch': '3'}
+
 25%|██▌       | 216/864 [06:17<12:38,  1.17s/it]
+
100%|██████████| 39/39 [00:05<00:00,  7.23it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0012/checkpoint-216
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0012/checkpoint-216/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0012/checkpoint-216/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:01<00:00,  1.48s/it]
Writing model shards: 100%|██████████| 1/1 [00:01<00:00,  1.48s/it]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0012/checkpoint-216/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0012/checkpoint-216/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0012/checkpoint-216/tokenizer_config.json
+[eval] epoch 3.0 step 216 loss 5.6492 (best 5.6492 @ ep 3.0, 0/10000 since best)
+
+
                                               
 25%|██▌       | 217/864 [06:24<48:28,  4.50s/it]
 25%|██▌       | 218/864 [06:25<38:07,  3.54s/it]
 25%|██▌       | 219/864 [06:26<30:52,  2.87s/it]
 25%|██▌       | 220/864 [06:28<25:48,  2.41s/it]
                                                 
{'loss': '5.597', 'grad_norm': '0.2812', 'learning_rate': '0.0012', 'epoch': '3.056'}
+
 25%|██▌       | 220/864 [06:28<25:48,  2.41s/it]
 26%|██▌       | 221/864 [06:29<22:16,  2.08s/it]
 26%|██▌       | 222/864 [06:30<19:48,  1.85s/it]
 26%|██▌       | 223/864 [06:32<18:02,  1.69s/it]
 26%|██▌       | 224/864 [06:33<16:49,  1.58s/it]
 26%|██▌       | 225/864 [06:34<15:57,  1.50s/it]
 26%|██▌       | 226/864 [06:35<15:20,  1.44s/it]
 26%|██▋       | 227/864 [06:37<14:54,  1.40s/it]
 26%|██▋       | 228/864 [06:38<14:35,  1.38s/it]
 27%|██▋       | 229/864 [06:39<14:22,  1.36s/it]
 27%|██▋       | 230/864 [06:41<14:12,  1.34s/it]
                                                 
{'loss': '5.466', 'grad_norm': '0.1549', 'learning_rate': '0.0012', 'epoch': '3.195'}
+
 27%|██▋       | 230/864 [06:41<14:12,  1.34s/it]
 27%|██▋       | 231/864 [06:42<14:04,  1.33s/it]
 27%|██▋       | 232/864 [06:43<14:00,  1.33s/it]
 27%|██▋       | 233/864 [06:45<13:55,  1.32s/it]
 27%|██▋       | 234/864 [06:46<13:51,  1.32s/it]
 27%|██▋       | 235/864 [06:47<13:49,  1.32s/it]
 27%|██▋       | 236/864 [06:49<13:47,  1.32s/it]
 27%|██▋       | 237/864 [06:50<13:45,  1.32s/it]
 28%|██▊       | 238/864 [06:51<13:43,  1.32s/it]
 28%|██▊       | 239/864 [06:53<13:41,  1.31s/it]
 28%|██▊       | 240/864 [06:54<13:40,  1.31s/it]
                                                 
{'loss': '5.418', 'grad_norm': '0.6102', 'learning_rate': '0.0012', 'epoch': '3.335'}
+
 28%|██▊       | 240/864 [06:54<13:40,  1.31s/it]
 28%|██▊       | 241/864 [06:55<13:39,  1.32s/it]
 28%|██▊       | 242/864 [06:56<13:38,  1.32s/it]
 28%|██▊       | 243/864 [06:58<13:36,  1.32s/it]
 28%|██▊       | 244/864 [06:59<13:34,  1.31s/it]
 28%|██▊       | 245/864 [07:00<13:33,  1.31s/it]
 28%|██▊       | 246/864 [07:02<13:32,  1.32s/it]
 29%|██▊       | 247/864 [07:03<13:31,  1.31s/it]
 29%|██▊       | 248/864 [07:04<13:29,  1.31s/it]
 29%|██▉       | 249/864 [07:06<13:28,  1.32s/it]
 29%|██▉       | 250/864 [07:07<13:28,  1.32s/it]
                                                 
{'loss': '5.406', 'grad_norm': '0.2704', 'learning_rate': '0.0012', 'epoch': '3.475'}
+
 29%|██▉       | 250/864 [07:07<13:28,  1.32s/it]
 29%|██▉       | 251/864 [07:08<13:26,  1.32s/it]
 29%|██▉       | 252/864 [07:10<13:24,  1.32s/it]
 29%|██▉       | 253/864 [07:11<13:23,  1.32s/it]
 29%|██▉       | 254/864 [07:12<13:22,  1.32s/it]
 30%|██▉       | 255/864 [07:14<13:20,  1.32s/it]
 30%|██▉       | 256/864 [07:15<13:19,  1.32s/it]
 30%|██▉       | 257/864 [07:16<13:17,  1.31s/it]
 30%|██▉       | 258/864 [07:18<13:16,  1.31s/it]
 30%|██▉       | 259/864 [07:19<13:15,  1.32s/it]
 30%|███       | 260/864 [07:20<13:13,  1.31s/it]
                                                 
{'loss': '5.286', 'grad_norm': '0.2055', 'learning_rate': '0.0012', 'epoch': '3.614'}
+
 30%|███       | 260/864 [07:20<13:13,  1.31s/it]
 30%|███       | 261/864 [07:21<13:12,  1.32s/it]
 30%|███       | 262/864 [07:23<13:11,  1.32s/it]
 30%|███       | 263/864 [07:24<13:10,  1.31s/it]
 31%|███       | 264/864 [07:25<13:08,  1.31s/it]
 31%|███       | 265/864 [07:27<13:07,  1.31s/it]
 31%|███       | 266/864 [07:28<13:06,  1.32s/it]
 31%|███       | 267/864 [07:29<13:04,  1.31s/it]
 31%|███       | 268/864 [07:31<13:03,  1.31s/it]
 31%|███       | 269/864 [07:32<13:02,  1.31s/it]
 31%|███▏      | 270/864 [07:33<13:01,  1.32s/it]
                                                 
{'loss': '5.313', 'grad_norm': '0.504', 'learning_rate': '0.0012', 'epoch': '3.754'}
+
 31%|███▏      | 270/864 [07:33<13:01,  1.32s/it]
 31%|███▏      | 271/864 [07:35<13:00,  1.32s/it]
 31%|███▏      | 272/864 [07:36<12:58,  1.32s/it]
 32%|███▏      | 273/864 [07:37<12:56,  1.31s/it]
 32%|███▏      | 274/864 [07:39<12:56,  1.32s/it]
 32%|███▏      | 275/864 [07:40<12:55,  1.32s/it]
 32%|███▏      | 276/864 [07:41<12:53,  1.32s/it]
 32%|███▏      | 277/864 [07:43<12:51,  1.31s/it]
 32%|███▏      | 278/864 [07:44<12:50,  1.32s/it]
 32%|███▏      | 279/864 [07:45<12:49,  1.32s/it]
 32%|███▏      | 280/864 [07:46<12:48,  1.32s/it]
                                                 
{'loss': '5.266', 'grad_norm': '0.2152', 'learning_rate': '0.0012', 'epoch': '3.894'}
+
 32%|███▏      | 280/864 [07:46<12:48,  1.32s/it]
 33%|███▎      | 281/864 [07:48<12:45,  1.31s/it]
 33%|███▎      | 282/864 [07:49<12:45,  1.32s/it]
 33%|███▎      | 283/864 [07:50<12:44,  1.32s/it]
 33%|███▎      | 284/864 [07:52<12:42,  1.31s/it]
 33%|███▎      | 285/864 [07:53<12:41,  1.31s/it]
 33%|███▎      | 286/864 [07:54<12:40,  1.32s/it]
 33%|███▎      | 287/864 [07:56<12:38,  1.32s/it]
 33%|███▎      | 288/864 [07:57<11:13,  1.17s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:02, 15.07it/s]
+
 10%|█         | 4/39 [00:00<00:03,  9.19it/s]
+
 15%|█▌        | 6/39 [00:00<00:04,  8.18it/s]
+
 18%|█▊        | 7/39 [00:00<00:04,  7.92it/s]
+
 21%|██        | 8/39 [00:00<00:04,  7.74it/s]
+
 23%|██▎       | 9/39 [00:01<00:03,  7.54it/s]
+
 26%|██▌       | 10/39 [00:01<00:03,  7.51it/s]
+
 28%|██▊       | 11/39 [00:01<00:03,  7.44it/s]
+
 31%|███       | 12/39 [00:01<00:03,  7.38it/s]
+
 33%|███▎      | 13/39 [00:01<00:03,  7.29it/s]
+
 36%|███▌      | 14/39 [00:01<00:03,  7.34it/s]
+
 38%|███▊      | 15/39 [00:01<00:03,  7.31it/s]
+
 41%|████      | 16/39 [00:02<00:03,  7.29it/s]
+
 44%|████▎     | 17/39 [00:02<00:03,  7.21it/s]
+
 46%|████▌     | 18/39 [00:02<00:02,  7.28it/s]
+
 49%|████▊     | 19/39 [00:02<00:02,  7.27it/s]
+
 51%|█████▏    | 20/39 [00:02<00:02,  7.27it/s]
+
 54%|█████▍    | 21/39 [00:02<00:02,  7.20it/s]
+
 56%|█████▋    | 22/39 [00:02<00:02,  7.27it/s]
+
 59%|█████▉    | 23/39 [00:03<00:02,  7.26it/s]
+
 62%|██████▏   | 24/39 [00:03<00:02,  7.26it/s]
+
 64%|██████▍   | 25/39 [00:03<00:01,  7.26it/s]
+
 67%|██████▋   | 26/39 [00:03<00:01,  7.26it/s]
+
 69%|██████▉   | 27/39 [00:03<00:01,  7.26it/s]
+
 72%|███████▏  | 28/39 [00:03<00:01,  7.26it/s]
+
 74%|███████▍  | 29/39 [00:03<00:01,  7.25it/s]
+
 77%|███████▋  | 30/39 [00:03<00:01,  7.25it/s]
+
 79%|███████▉  | 31/39 [00:04<00:01,  7.25it/s]
+
 82%|████████▏ | 32/39 [00:04<00:00,  7.24it/s]
+
 85%|████████▍ | 33/39 [00:04<00:00,  7.24it/s]
+
 87%|████████▋ | 34/39 [00:04<00:00,  7.24it/s]
+
 90%|████████▉ | 35/39 [00:04<00:00,  7.23it/s]
+
 92%|█████████▏| 36/39 [00:04<00:00,  7.23it/s]
+
 95%|█████████▍| 37/39 [00:04<00:00,  7.23it/s]
+
 97%|█████████▋| 38/39 [00:05<00:00,  7.23it/s]
+
100%|██████████| 39/39 [00:05<00:00,  7.24it/s]
                                                 
+
                                               
{'eval_loss': '5.198', 'eval_runtime': '5.439', 'eval_samples_per_second': '448.8', 'eval_steps_per_second': '7.17', 'epoch': '4'}
+
 33%|███▎      | 288/864 [08:02<11:13,  1.17s/it]
+
100%|██████████| 39/39 [00:05<00:00,  7.24it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0012/checkpoint-288
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0012/checkpoint-288/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0012/checkpoint-288/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:01<00:00,  1.51s/it]
Writing model shards: 100%|██████████| 1/1 [00:01<00:00,  1.51s/it]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0012/checkpoint-288/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0012/checkpoint-288/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0012/checkpoint-288/tokenizer_config.json
+[eval] epoch 4.0 step 288 loss 5.1976 (best 5.1976 @ ep 4.0, 0/10000 since best)
+
+
                                               
 33%|███▎      | 289/864 [08:09<42:49,  4.47s/it]
 34%|███▎      | 290/864 [08:10<33:41,  3.52s/it]
                                                 
{'loss': '5.12', 'grad_norm': '0.3153', 'learning_rate': '0.0012', 'epoch': '4.028'}
+
 34%|███▎      | 290/864 [08:10<33:41,  3.52s/it]
 34%|███▎      | 291/864 [08:11<27:18,  2.86s/it]
 34%|███▍      | 292/864 [08:13<22:49,  2.39s/it]
 34%|███▍      | 293/864 [08:14<19:42,  2.07s/it]
 34%|███▍      | 294/864 [08:15<17:30,  1.84s/it]
 34%|███▍      | 295/864 [08:17<15:58,  1.68s/it]
 34%|███▍      | 296/864 [08:18<14:53,  1.57s/it]
 34%|███▍      | 297/864 [08:19<14:08,  1.50s/it]
 34%|███▍      | 298/864 [08:20<13:36,  1.44s/it]
 35%|███▍      | 299/864 [08:22<13:12,  1.40s/it]
 35%|███▍      | 300/864 [08:23<12:56,  1.38s/it]
                                                 
{'loss': '5.064', 'grad_norm': '0.4078', 'learning_rate': '0.0012', 'epoch': '4.168'}
+
 35%|███▍      | 300/864 [08:23<12:56,  1.38s/it]
 35%|███▍      | 301/864 [08:24<12:45,  1.36s/it]
 35%|███▍      | 302/864 [08:26<12:37,  1.35s/it]
 35%|███▌      | 303/864 [08:27<12:31,  1.34s/it]
 35%|███▌      | 304/864 [08:28<12:25,  1.33s/it]
 35%|███▌      | 305/864 [08:30<12:21,  1.33s/it]
 35%|███▌      | 306/864 [08:31<12:17,  1.32s/it]
 36%|███▌      | 307/864 [08:32<12:16,  1.32s/it]
 36%|███▌      | 308/864 [08:34<12:14,  1.32s/it]
 36%|███▌      | 309/864 [08:35<12:12,  1.32s/it]
 36%|███▌      | 310/864 [08:36<12:11,  1.32s/it]
                                                 
{'loss': '4.978', 'grad_norm': '0.1672', 'learning_rate': '0.0012', 'epoch': '4.307'}
+
 36%|███▌      | 310/864 [08:36<12:11,  1.32s/it]
 36%|███▌      | 311/864 [08:38<12:09,  1.32s/it]
 36%|███▌      | 312/864 [08:39<12:07,  1.32s/it]
 36%|███▌      | 313/864 [08:40<12:06,  1.32s/it]
 36%|███▋      | 314/864 [08:42<12:04,  1.32s/it]
 36%|███▋      | 315/864 [08:43<12:02,  1.32s/it]
 37%|███▋      | 316/864 [08:44<12:01,  1.32s/it]
 37%|███▋      | 317/864 [08:46<12:00,  1.32s/it]
 37%|███▋      | 318/864 [08:47<11:59,  1.32s/it]
 37%|███▋      | 319/864 [08:48<11:57,  1.32s/it]
 37%|███▋      | 320/864 [08:49<11:56,  1.32s/it]
                                                 
{'loss': '4.892', 'grad_norm': '0.2579', 'learning_rate': '0.0012', 'epoch': '4.447'}
+
 37%|███▋      | 320/864 [08:49<11:56,  1.32s/it]
 37%|███▋      | 321/864 [08:51<12:50,  1.42s/it]
 37%|███▋      | 322/864 [08:52<12:31,  1.39s/it]
 37%|███▋      | 323/864 [08:54<12:18,  1.37s/it]
 38%|███▊      | 324/864 [08:55<12:09,  1.35s/it]
 38%|███▊      | 325/864 [08:56<12:02,  1.34s/it]
 38%|███▊      | 326/864 [08:58<11:57,  1.33s/it]
 38%|███▊      | 327/864 [08:59<11:53,  1.33s/it]
 38%|███▊      | 328/864 [09:00<11:50,  1.33s/it]
 38%|███▊      | 329/864 [09:02<11:47,  1.32s/it]
 38%|███▊      | 330/864 [09:03<11:45,  1.32s/it]
                                                 
{'loss': '4.866', 'grad_norm': '0.5844', 'learning_rate': '0.0012', 'epoch': '4.586'}
+
 38%|███▊      | 330/864 [09:03<11:45,  1.32s/it]
 38%|███▊      | 331/864 [09:04<11:43,  1.32s/it]
 38%|███▊      | 332/864 [09:06<11:42,  1.32s/it]
 39%|███▊      | 333/864 [09:07<11:40,  1.32s/it]
 39%|███▊      | 334/864 [09:08<11:38,  1.32s/it]
 39%|███▉      | 335/864 [09:10<11:37,  1.32s/it]
 39%|███▉      | 336/864 [09:11<11:35,  1.32s/it]
 39%|███▉      | 337/864 [09:12<11:34,  1.32s/it]
 39%|███▉      | 338/864 [09:14<11:32,  1.32s/it]
 39%|███▉      | 339/864 [09:15<13:02,  1.49s/it]
 39%|███▉      | 340/864 [09:17<12:33,  1.44s/it]
                                                 
{'loss': '4.886', 'grad_norm': '0.2705', 'learning_rate': '0.0012', 'epoch': '4.726'}
+
 39%|███▉      | 340/864 [09:17<12:33,  1.44s/it]
 39%|███▉      | 341/864 [09:18<12:12,  1.40s/it]
 40%|███▉      | 342/864 [09:19<11:57,  1.38s/it]
 40%|███▉      | 343/864 [09:21<11:46,  1.36s/it]
 40%|███▉      | 344/864 [09:22<11:40,  1.35s/it]
 40%|███▉      | 345/864 [09:23<11:35,  1.34s/it]
 40%|████      | 346/864 [09:25<11:31,  1.34s/it]
 40%|████      | 347/864 [09:26<11:27,  1.33s/it]
 40%|████      | 348/864 [09:27<11:25,  1.33s/it]
 40%|████      | 349/864 [09:29<11:22,  1.33s/it]
 41%|████      | 350/864 [09:30<11:20,  1.32s/it]
                                                 
{'loss': '4.756', 'grad_norm': '0.1286', 'learning_rate': '0.0012', 'epoch': '4.866'}
+
 41%|████      | 350/864 [09:30<11:20,  1.32s/it]
 41%|████      | 351/864 [09:31<11:18,  1.32s/it]
 41%|████      | 352/864 [09:33<11:15,  1.32s/it]
 41%|████      | 353/864 [09:34<11:14,  1.32s/it]
 41%|████      | 354/864 [09:35<11:12,  1.32s/it]
 41%|████      | 355/864 [09:37<11:11,  1.32s/it]
 41%|████      | 356/864 [09:38<11:10,  1.32s/it]
 41%|████▏     | 357/864 [09:39<11:08,  1.32s/it]
 41%|████▏     | 358/864 [09:40<11:07,  1.32s/it]
 42%|████▏     | 359/864 [09:42<11:06,  1.32s/it]
 42%|████▏     | 360/864 [09:43<09:50,  1.17s/it]
                                                 
{'loss': '4.675', 'grad_norm': '0.3044', 'learning_rate': '0.0012', 'epoch': '5'}
+
 42%|████▏     | 360/864 [09:43<09:50,  1.17s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:02, 15.30it/s]
+
 10%|█         | 4/39 [00:00<00:03,  9.25it/s]
+
 15%|█▌        | 6/39 [00:00<00:04,  8.20it/s]
+
 18%|█▊        | 7/39 [00:00<00:04,  7.94it/s]
+
 21%|██        | 8/39 [00:00<00:04,  7.74it/s]
+
 23%|██▎       | 9/39 [00:01<00:03,  7.60it/s]
+
 26%|██▌       | 10/39 [00:01<00:03,  7.49it/s]
+
 28%|██▊       | 11/39 [00:01<00:03,  7.41it/s]
+
 31%|███       | 12/39 [00:01<00:03,  7.36it/s]
+
 33%|███▎      | 13/39 [00:01<00:03,  7.33it/s]
+
 36%|███▌      | 14/39 [00:01<00:03,  7.30it/s]
+
 38%|███▊      | 15/39 [00:01<00:03,  7.28it/s]
+
 41%|████      | 16/39 [00:02<00:03,  7.25it/s]
+
 44%|████▎     | 17/39 [00:02<00:03,  7.24it/s]
+
 46%|████▌     | 18/39 [00:02<00:02,  7.24it/s]
+
 49%|████▊     | 19/39 [00:02<00:02,  7.25it/s]
+
 51%|█████▏    | 20/39 [00:02<00:02,  7.24it/s]
+
 54%|█████▍    | 21/39 [00:02<00:02,  7.24it/s]
+
 56%|█████▋    | 22/39 [00:02<00:02,  7.25it/s]
+
 59%|█████▉    | 23/39 [00:03<00:02,  7.25it/s]
+
 62%|██████▏   | 24/39 [00:03<00:02,  7.25it/s]
+
 64%|██████▍   | 25/39 [00:03<00:01,  7.25it/s]
+
 67%|██████▋   | 26/39 [00:03<00:01,  7.25it/s]
+
 69%|██████▉   | 27/39 [00:03<00:01,  7.25it/s]
+
 72%|███████▏  | 28/39 [00:03<00:01,  7.24it/s]
+
 74%|███████▍  | 29/39 [00:03<00:01,  7.24it/s]
+
 77%|███████▋  | 30/39 [00:03<00:01,  7.24it/s]
+
 79%|███████▉  | 31/39 [00:04<00:01,  7.25it/s]
+
 82%|████████▏ | 32/39 [00:04<00:00,  7.25it/s]
+
 85%|████████▍ | 33/39 [00:04<00:00,  7.24it/s]
+
 87%|████████▋ | 34/39 [00:04<00:00,  7.24it/s]
+
 90%|████████▉ | 35/39 [00:04<00:00,  7.24it/s]
+
 92%|█████████▏| 36/39 [00:04<00:00,  7.24it/s]
+
 95%|█████████▍| 37/39 [00:04<00:00,  7.24it/s]
+
 97%|█████████▋| 38/39 [00:05<00:00,  7.24it/s]
+
100%|██████████| 39/39 [00:05<00:00,  7.25it/s]
                                                 
+
                                               
{'eval_loss': '4.749', 'eval_runtime': '5.438', 'eval_samples_per_second': '448.9', 'eval_steps_per_second': '7.172', 'epoch': '5'}
+
 42%|████▏     | 360/864 [09:48<09:50,  1.17s/it]
+
100%|██████████| 39/39 [00:05<00:00,  7.25it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0012/checkpoint-360
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0012/checkpoint-360/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0012/checkpoint-360/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:01<00:00,  1.82s/it]
Writing model shards: 100%|██████████| 1/1 [00:01<00:00,  1.82s/it]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0012/checkpoint-360/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0012/checkpoint-360/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0012/checkpoint-360/tokenizer_config.json
+[eval] epoch 5.0 step 360 loss 4.7488 (best 4.7488 @ ep 5.0, 0/10000 since best)
+
+
                                               
 42%|████▏     | 361/864 [09:55<37:32,  4.48s/it]
 42%|████▏     | 362/864 [09:56<29:30,  3.53s/it]
 42%|████▏     | 363/864 [09:57<23:54,  2.86s/it]
 42%|████▏     | 364/864 [09:59<19:59,  2.40s/it]
 42%|████▏     | 365/864 [10:00<17:15,  2.07s/it]
 42%|████▏     | 366/864 [10:01<15:19,  1.85s/it]
 42%|████▏     | 367/864 [10:03<13:58,  1.69s/it]
 43%|████▎     | 368/864 [10:04<13:01,  1.58s/it]
 43%|████▎     | 369/864 [10:05<12:21,  1.50s/it]
 43%|████▎     | 370/864 [10:07<11:53,  1.44s/it]
                                                 
{'loss': '4.592', 'grad_norm': '0.4988', 'learning_rate': '0.0012', 'epoch': '5.14'}
+
 43%|████▎     | 370/864 [10:07<11:53,  1.44s/it]
 43%|████▎     | 371/864 [10:08<11:33,  1.41s/it]
 43%|████▎     | 372/864 [10:09<11:18,  1.38s/it]
 43%|████▎     | 373/864 [10:11<11:07,  1.36s/it]
 43%|████▎     | 374/864 [10:12<10:59,  1.35s/it]
 43%|████▎     | 375/864 [10:13<10:53,  1.34s/it]
 44%|████▎     | 376/864 [10:15<10:49,  1.33s/it]
 44%|████▎     | 377/864 [10:16<10:46,  1.33s/it]
 44%|████▍     | 378/864 [10:17<10:42,  1.32s/it]
 44%|████▍     | 379/864 [10:18<10:40,  1.32s/it]
 44%|████▍     | 380/864 [10:20<10:38,  1.32s/it]
                                                 
{'loss': '4.568', 'grad_norm': '0.2263', 'learning_rate': '0.0012', 'epoch': '5.279'}
+
 44%|████▍     | 380/864 [10:20<10:38,  1.32s/it]
 44%|████▍     | 381/864 [10:21<10:36,  1.32s/it]
 44%|████▍     | 382/864 [10:22<10:35,  1.32s/it]
 44%|████▍     | 383/864 [10:24<10:33,  1.32s/it]
 44%|████▍     | 384/864 [10:25<10:32,  1.32s/it]
 45%|████▍     | 385/864 [10:26<10:30,  1.32s/it]
 45%|████▍     | 386/864 [10:28<10:29,  1.32s/it]
 45%|████▍     | 387/864 [10:29<10:27,  1.31s/it]
 45%|████▍     | 388/864 [10:30<10:26,  1.32s/it]
 45%|████▌     | 389/864 [10:32<10:24,  1.31s/it]
 45%|████▌     | 390/864 [10:33<10:22,  1.31s/it]
                                                 
{'loss': '4.471', 'grad_norm': '0.1717', 'learning_rate': '0.0012', 'epoch': '5.419'}
+
 45%|████▌     | 390/864 [10:33<10:22,  1.31s/it]
 45%|████▌     | 391/864 [10:34<10:21,  1.31s/it]
 45%|████▌     | 392/864 [10:36<10:20,  1.32s/it]
 45%|████▌     | 393/864 [10:37<10:18,  1.31s/it]
 46%|████▌     | 394/864 [10:38<10:17,  1.31s/it]
 46%|████▌     | 395/864 [10:40<10:17,  1.32s/it]
 46%|████▌     | 396/864 [10:41<10:15,  1.32s/it]
 46%|████▌     | 397/864 [10:42<10:13,  1.31s/it]
 46%|████▌     | 398/864 [10:43<10:12,  1.31s/it]
 46%|████▌     | 399/864 [10:45<10:11,  1.31s/it]
 46%|████▋     | 400/864 [10:46<10:10,  1.32s/it]
                                                 
{'loss': '4.398', 'grad_norm': '0.2116', 'learning_rate': '0.0012', 'epoch': '5.558'}
+
 46%|████▋     | 400/864 [10:46<10:10,  1.32s/it]
 46%|████▋     | 401/864 [10:47<10:09,  1.32s/it]
 47%|████▋     | 402/864 [10:49<10:07,  1.32s/it]
 47%|████▋     | 403/864 [10:50<10:06,  1.32s/it]
 47%|████▋     | 404/864 [10:51<10:05,  1.32s/it]
 47%|████▋     | 405/864 [10:53<10:03,  1.32s/it]
 47%|████▋     | 406/864 [10:54<10:02,  1.32s/it]
 47%|████▋     | 407/864 [10:55<10:00,  1.31s/it]
 47%|████▋     | 408/864 [10:57<09:59,  1.32s/it]
 47%|████▋     | 409/864 [10:58<09:58,  1.31s/it]
 47%|████▋     | 410/864 [10:59<09:57,  1.31s/it]
                                                 
{'loss': '4.414', 'grad_norm': '0.2903', 'learning_rate': '0.0012', 'epoch': '5.698'}
+
 47%|████▋     | 410/864 [10:59<09:57,  1.31s/it]
 48%|████▊     | 411/864 [11:01<09:55,  1.32s/it]
 48%|████▊     | 412/864 [11:02<09:54,  1.32s/it]
 48%|████▊     | 413/864 [11:03<09:53,  1.32s/it]
 48%|████▊     | 414/864 [11:05<09:51,  1.32s/it]
 48%|████▊     | 415/864 [11:06<09:50,  1.31s/it]
 48%|████▊     | 416/864 [11:07<09:49,  1.32s/it]
 48%|████▊     | 417/864 [11:08<09:47,  1.32s/it]
 48%|████▊     | 418/864 [11:10<09:46,  1.31s/it]
 48%|████▊     | 419/864 [11:11<09:45,  1.32s/it]
 49%|████▊     | 420/864 [11:12<09:44,  1.32s/it]
                                                 
{'loss': '4.374', 'grad_norm': '0.189', 'learning_rate': '0.0012', 'epoch': '5.838'}
+
 49%|████▊     | 420/864 [11:12<09:44,  1.32s/it]
 49%|████▊     | 421/864 [11:14<09:43,  1.32s/it]
 49%|████▉     | 422/864 [11:15<09:41,  1.32s/it]
 49%|████▉     | 423/864 [11:16<09:40,  1.32s/it]
 49%|████▉     | 424/864 [11:18<09:38,  1.32s/it]
 49%|████▉     | 425/864 [11:19<09:37,  1.32s/it]
 49%|████▉     | 426/864 [11:20<09:36,  1.32s/it]
 49%|████▉     | 427/864 [11:22<09:35,  1.32s/it]
 50%|████▉     | 428/864 [11:23<09:33,  1.32s/it]
 50%|████▉     | 429/864 [11:24<09:32,  1.32s/it]
 50%|████▉     | 430/864 [11:26<09:31,  1.32s/it]
                                                 
{'loss': '4.308', 'grad_norm': '0.1725', 'learning_rate': '0.0012', 'epoch': '5.977'}
+
 50%|████▉     | 430/864 [11:26<09:31,  1.32s/it]
 50%|████▉     | 431/864 [11:27<09:30,  1.32s/it]
 50%|█████     | 432/864 [11:28<08:25,  1.17s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:02, 15.45it/s]
+
 10%|█         | 4/39 [00:00<00:03,  9.27it/s]
+
 15%|█▌        | 6/39 [00:00<00:04,  8.22it/s]
+
 18%|█▊        | 7/39 [00:00<00:04,  7.95it/s]
+
 21%|██        | 8/39 [00:00<00:04,  7.74it/s]
+
 23%|██▎       | 9/39 [00:01<00:03,  7.60it/s]
+
 26%|██▌       | 10/39 [00:01<00:03,  7.50it/s]
+
 28%|██▊       | 11/39 [00:01<00:03,  7.43it/s]
+
 31%|███       | 12/39 [00:01<00:03,  7.37it/s]
+
 33%|███▎      | 13/39 [00:01<00:03,  7.34it/s]
+
 36%|███▌      | 14/39 [00:01<00:03,  7.31it/s]
+
 38%|███▊      | 15/39 [00:01<00:03,  7.30it/s]
+
 41%|████      | 16/39 [00:02<00:03,  7.28it/s]
+
 44%|████▎     | 17/39 [00:02<00:03,  7.27it/s]
+
 46%|████▌     | 18/39 [00:02<00:02,  7.27it/s]
+
 49%|████▊     | 19/39 [00:02<00:02,  7.26it/s]
+
 51%|█████▏    | 20/39 [00:02<00:02,  7.26it/s]
+
 54%|█████▍    | 21/39 [00:02<00:02,  7.26it/s]
+
 56%|█████▋    | 22/39 [00:02<00:02,  7.24it/s]
+
 59%|█████▉    | 23/39 [00:03<00:02,  7.24it/s]
+
 62%|██████▏   | 24/39 [00:03<00:02,  7.23it/s]
+
 64%|██████▍   | 25/39 [00:03<00:01,  7.24it/s]
+
 67%|██████▋   | 26/39 [00:03<00:01,  7.24it/s]
+
 69%|██████▉   | 27/39 [00:03<00:01,  7.24it/s]
+
 72%|███████▏  | 28/39 [00:03<00:01,  7.25it/s]
+
 74%|███████▍  | 29/39 [00:03<00:01,  7.25it/s]
+
 77%|███████▋  | 30/39 [00:03<00:01,  7.25it/s]
+
 79%|███████▉  | 31/39 [00:04<00:01,  7.26it/s]
+
 82%|████████▏ | 32/39 [00:04<00:00,  7.25it/s]
+
 85%|████████▍ | 33/39 [00:04<00:00,  7.25it/s]
+
 87%|████████▋ | 34/39 [00:04<00:00,  7.25it/s]
+
 90%|████████▉ | 35/39 [00:04<00:00,  7.25it/s]
+
 92%|█████████▏| 36/39 [00:04<00:00,  7.23it/s]
+
 95%|█████████▍| 37/39 [00:04<00:00,  7.23it/s]
+
 97%|█████████▋| 38/39 [00:05<00:00,  7.23it/s]
+
100%|██████████| 39/39 [00:05<00:00,  7.23it/s]
                                                 
+
                                               
{'eval_loss': '4.437', 'eval_runtime': '5.454', 'eval_samples_per_second': '447.6', 'eval_steps_per_second': '7.151', 'epoch': '6'}
+
 50%|█████     | 432/864 [11:33<08:25,  1.17s/it]
+
100%|██████████| 39/39 [00:05<00:00,  7.23it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0012/checkpoint-432
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0012/checkpoint-432/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0012/checkpoint-432/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:01<00:00,  1.54s/it]
Writing model shards: 100%|██████████| 1/1 [00:01<00:00,  1.54s/it]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0012/checkpoint-432/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0012/checkpoint-432/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0012/checkpoint-432/tokenizer_config.json
+[eval] epoch 6.0 step 432 loss 4.4370 (best 4.4370 @ ep 6.0, 0/10000 since best)
+
+
                                               
 50%|█████     | 433/864 [11:40<31:32,  4.39s/it]
 50%|█████     | 434/864 [11:41<24:50,  3.47s/it]
 50%|█████     | 435/864 [11:42<20:09,  2.82s/it]
 50%|█████     | 436/864 [11:44<16:53,  2.37s/it]
 51%|█████     | 437/864 [11:45<14:36,  2.05s/it]
 51%|█████     | 438/864 [11:46<12:59,  1.83s/it]
 51%|█████     | 439/864 [11:47<11:52,  1.68s/it]
 51%|█████     | 440/864 [11:49<11:04,  1.57s/it]
                                                 
{'loss': '4.221', 'grad_norm': '0.2731', 'learning_rate': '0.0012', 'epoch': '6.112'}
+
 51%|█████     | 440/864 [11:49<11:04,  1.57s/it]
 51%|█████     | 441/864 [11:50<10:30,  1.49s/it]
 51%|█████     | 442/864 [11:51<10:06,  1.44s/it]
 51%|█████▏    | 443/864 [11:53<09:49,  1.40s/it]
 51%|█████▏    | 444/864 [11:54<09:37,  1.38s/it]
 52%|█████▏    | 445/864 [11:55<09:28,  1.36s/it]
 52%|█████▏    | 446/864 [11:57<09:22,  1.35s/it]
 52%|█████▏    | 447/864 [11:58<09:17,  1.34s/it]
 52%|█████▏    | 448/864 [11:59<09:13,  1.33s/it]
 52%|█████▏    | 449/864 [12:01<09:09,  1.33s/it]
 52%|█████▏    | 450/864 [12:02<09:07,  1.32s/it]
                                                 
{'loss': '4.216', 'grad_norm': '0.3596', 'learning_rate': '0.0012', 'epoch': '6.251'}
+
 52%|█████▏    | 450/864 [12:02<09:07,  1.32s/it]
 52%|█████▏    | 451/864 [12:03<09:05,  1.32s/it]
 52%|█████▏    | 452/864 [12:05<09:03,  1.32s/it]
 52%|█████▏    | 453/864 [12:06<09:01,  1.32s/it]
 53%|█████▎    | 454/864 [12:07<09:00,  1.32s/it]
 53%|█████▎    | 455/864 [12:09<08:57,  1.31s/it]
 53%|█████▎    | 456/864 [12:10<08:56,  1.31s/it]
 53%|█████▎    | 457/864 [12:11<08:55,  1.32s/it]
 53%|█████▎    | 458/864 [12:12<08:53,  1.32s/it]
 53%|█████▎    | 459/864 [12:14<08:52,  1.31s/it]
 53%|█████▎    | 460/864 [12:15<08:51,  1.32s/it]
                                                 
{'loss': '4.169', 'grad_norm': '0.1744', 'learning_rate': '0.0012', 'epoch': '6.391'}
+
 53%|█████▎    | 460/864 [12:15<08:51,  1.32s/it]
 53%|█████▎    | 461/864 [12:16<08:50,  1.32s/it]
 53%|█████▎    | 462/864 [12:18<08:48,  1.31s/it]
 54%|█████▎    | 463/864 [12:19<08:47,  1.31s/it]
 54%|█████▎    | 464/864 [12:20<08:45,  1.31s/it]
 54%|█████▍    | 465/864 [12:22<08:44,  1.31s/it]
 54%|█████▍    | 466/864 [12:23<08:42,  1.31s/it]
 54%|█████▍    | 467/864 [12:24<08:41,  1.31s/it]
 54%|█████▍    | 468/864 [12:26<08:40,  1.31s/it]
 54%|█████▍    | 469/864 [12:27<08:38,  1.31s/it]
 54%|█████▍    | 470/864 [12:28<08:37,  1.31s/it]
                                                 
{'loss': '4.114', 'grad_norm': '0.1464', 'learning_rate': '0.0012', 'epoch': '6.531'}
+
 54%|█████▍    | 470/864 [12:28<08:37,  1.31s/it]
 55%|█████▍    | 471/864 [12:30<08:36,  1.31s/it]
 55%|█████▍    | 472/864 [12:31<08:35,  1.31s/it]
 55%|█████▍    | 473/864 [12:32<08:33,  1.31s/it]
 55%|█████▍    | 474/864 [12:33<08:33,  1.32s/it]
 55%|█████▍    | 475/864 [12:35<08:32,  1.32s/it]
 55%|█████▌    | 476/864 [12:36<08:30,  1.32s/it]
 55%|█████▌    | 477/864 [12:37<08:29,  1.32s/it]
 55%|█████▌    | 478/864 [12:39<08:27,  1.31s/it]
 55%|█████▌    | 479/864 [12:40<08:26,  1.32s/it]
 56%|█████▌    | 480/864 [12:41<08:25,  1.32s/it]
                                                 
{'loss': '4.077', 'grad_norm': '0.1628', 'learning_rate': '0.0012', 'epoch': '6.67'}
+
 56%|█████▌    | 480/864 [12:41<08:25,  1.32s/it]
 56%|█████▌    | 481/864 [12:43<08:24,  1.32s/it]
 56%|█████▌    | 482/864 [12:44<08:22,  1.31s/it]
 56%|█████▌    | 483/864 [12:45<08:21,  1.32s/it]
 56%|█████▌    | 484/864 [12:47<08:19,  1.32s/it]
 56%|█████▌    | 485/864 [12:48<08:18,  1.32s/it]
 56%|█████▋    | 486/864 [12:49<08:16,  1.31s/it]
 56%|█████▋    | 487/864 [12:51<08:15,  1.31s/it]
 56%|█████▋    | 488/864 [12:52<08:14,  1.31s/it]
 57%|█████▋    | 489/864 [12:53<08:12,  1.31s/it]
 57%|█████▋    | 490/864 [12:55<08:11,  1.31s/it]
                                                 
{'loss': '4.072', 'grad_norm': '0.1819', 'learning_rate': '0.0012', 'epoch': '6.81'}
+
 57%|█████▋    | 490/864 [12:55<08:11,  1.31s/it]
 57%|█████▋    | 491/864 [12:56<08:10,  1.31s/it]
 57%|█████▋    | 492/864 [12:57<08:08,  1.31s/it]
 57%|█████▋    | 493/864 [12:58<08:07,  1.31s/it]
 57%|█████▋    | 494/864 [13:00<08:06,  1.31s/it]
 57%|█████▋    | 495/864 [13:01<08:05,  1.32s/it]
 57%|█████▋    | 496/864 [13:02<08:03,  1.31s/it]
 58%|█████▊    | 497/864 [13:04<08:02,  1.31s/it]
 58%|█████▊    | 498/864 [13:05<08:01,  1.32s/it]
 58%|█████▊    | 499/864 [13:06<08:00,  1.32s/it]
 58%|█████▊    | 500/864 [13:08<07:59,  1.32s/it]
                                                 
{'loss': '4.041', 'grad_norm': '0.1805', 'learning_rate': '0.0012', 'epoch': '6.949'}
+
 58%|█████▊    | 500/864 [13:08<07:59,  1.32s/it]
 58%|█████▊    | 501/864 [13:09<07:57,  1.32s/it]
 58%|█████▊    | 502/864 [13:10<07:56,  1.32s/it]
 58%|█████▊    | 503/864 [13:12<07:54,  1.32s/it]
 58%|█████▊    | 504/864 [13:12<07:01,  1.17s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:02, 15.13it/s]
+
 10%|█         | 4/39 [00:00<00:03,  9.21it/s]
+
 15%|█▌        | 6/39 [00:00<00:04,  8.21it/s]
+
 18%|█▊        | 7/39 [00:00<00:04,  7.94it/s]
+
 21%|██        | 8/39 [00:00<00:03,  7.75it/s]
+
 23%|██▎       | 9/39 [00:01<00:03,  7.60it/s]
+
 26%|██▌       | 10/39 [00:01<00:03,  7.50it/s]
+
 28%|██▊       | 11/39 [00:01<00:03,  7.43it/s]
+
 31%|███       | 12/39 [00:01<00:03,  7.37it/s]
+
 33%|███▎      | 13/39 [00:01<00:03,  7.34it/s]
+
 36%|███▌      | 14/39 [00:01<00:03,  7.31it/s]
+
 38%|███▊      | 15/39 [00:01<00:03,  7.29it/s]
+
 41%|████      | 16/39 [00:02<00:03,  7.28it/s]
+
 44%|████▎     | 17/39 [00:02<00:03,  7.26it/s]
+
 46%|████▌     | 18/39 [00:02<00:02,  7.25it/s]
+
 49%|████▊     | 19/39 [00:02<00:02,  7.25it/s]
+
 51%|█████▏    | 20/39 [00:02<00:02,  7.23it/s]
+
 54%|█████▍    | 21/39 [00:02<00:02,  7.23it/s]
+
 56%|█████▋    | 22/39 [00:02<00:02,  7.23it/s]
+
 59%|█████▉    | 23/39 [00:03<00:02,  7.24it/s]
+
 62%|██████▏   | 24/39 [00:03<00:02,  7.24it/s]
+
 64%|██████▍   | 25/39 [00:03<00:01,  7.24it/s]
+
 67%|██████▋   | 26/39 [00:03<00:01,  7.25it/s]
+
 69%|██████▉   | 27/39 [00:03<00:01,  7.24it/s]
+
 72%|███████▏  | 28/39 [00:03<00:01,  7.25it/s]
+
 74%|███████▍  | 29/39 [00:03<00:01,  7.25it/s]
+
 77%|███████▋  | 30/39 [00:03<00:01,  7.25it/s]
+
 79%|███████▉  | 31/39 [00:04<00:01,  7.24it/s]
+
 82%|████████▏ | 32/39 [00:04<00:00,  7.24it/s]
+
 85%|████████▍ | 33/39 [00:04<00:00,  7.24it/s]
+
 87%|████████▋ | 34/39 [00:04<00:00,  7.24it/s]
+
 90%|████████▉ | 35/39 [00:04<00:00,  7.25it/s]
+
 92%|█████████▏| 36/39 [00:04<00:00,  7.25it/s]
+
 95%|█████████▍| 37/39 [00:04<00:00,  7.25it/s]
+
 97%|█████████▋| 38/39 [00:05<00:00,  7.25it/s]
+
100%|██████████| 39/39 [00:05<00:00,  7.25it/s]
                                                 
+
                                               
{'eval_loss': '4.23', 'eval_runtime': '5.45', 'eval_samples_per_second': '447.9', 'eval_steps_per_second': '7.157', 'epoch': '7'}
+
 58%|█████▊    | 504/864 [13:18<07:01,  1.17s/it]
+
100%|██████████| 39/39 [00:05<00:00,  7.25it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0012/checkpoint-504
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0012/checkpoint-504/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0012/checkpoint-504/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:01<00:00,  1.57s/it]
Writing model shards: 100%|██████████| 1/1 [00:01<00:00,  1.58s/it]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0012/checkpoint-504/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0012/checkpoint-504/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0012/checkpoint-504/tokenizer_config.json
+[eval] epoch 7.0 step 504 loss 4.2298 (best 4.2298 @ ep 7.0, 0/10000 since best)
+
+
                                               
 58%|█████▊    | 505/864 [13:25<26:30,  4.43s/it]
 59%|█████▊    | 506/864 [13:26<20:51,  3.50s/it]
 59%|█████▊    | 507/864 [13:27<16:54,  2.84s/it]
 59%|█████▉    | 508/864 [13:28<14:08,  2.38s/it]
 59%|█████▉    | 509/864 [13:30<12:12,  2.06s/it]
 59%|█████▉    | 510/864 [13:31<10:51,  1.84s/it]
                                                 
{'loss': '3.962', 'grad_norm': '0.3302', 'learning_rate': '0.0012', 'epoch': '7.084'}
+
 59%|█████▉    | 510/864 [13:31<10:51,  1.84s/it]
 59%|█████▉    | 511/864 [13:32<09:54,  1.68s/it]
 59%|█████▉    | 512/864 [13:34<09:13,  1.57s/it]
 59%|█████▉    | 513/864 [13:35<08:45,  1.50s/it]
 59%|█████▉    | 514/864 [13:36<08:24,  1.44s/it]
 60%|█████▉    | 515/864 [13:38<08:10,  1.40s/it]
 60%|█████▉    | 516/864 [13:39<07:59,  1.38s/it]
 60%|█████▉    | 517/864 [13:40<07:51,  1.36s/it]
 60%|█████▉    | 518/864 [13:42<07:46,  1.35s/it]
 60%|██████    | 519/864 [13:43<07:41,  1.34s/it]
 60%|██████    | 520/864 [13:44<07:37,  1.33s/it]
                                                 
{'loss': '3.991', 'grad_norm': '0.2574', 'learning_rate': '0.0012', 'epoch': '7.223'}
+
 60%|██████    | 520/864 [13:44<07:37,  1.33s/it]
 60%|██████    | 521/864 [13:46<07:35,  1.33s/it]
 60%|██████    | 522/864 [13:47<07:32,  1.32s/it]
 61%|██████    | 523/864 [13:48<07:30,  1.32s/it]
 61%|██████    | 524/864 [13:50<07:28,  1.32s/it]
 61%|██████    | 525/864 [13:51<07:27,  1.32s/it]
 61%|██████    | 526/864 [13:52<07:25,  1.32s/it]
 61%|██████    | 527/864 [13:53<07:24,  1.32s/it]
 61%|██████    | 528/864 [13:55<07:22,  1.32s/it]
 61%|██████    | 529/864 [13:56<07:21,  1.32s/it]
 61%|██████▏   | 530/864 [13:57<07:19,  1.32s/it]
                                                 
{'loss': '3.946', 'grad_norm': '0.1589', 'learning_rate': '0.0012', 'epoch': '7.363'}
+
 61%|██████▏   | 530/864 [13:57<07:19,  1.32s/it]
 61%|██████▏   | 531/864 [13:59<07:18,  1.32s/it]
 62%|██████▏   | 532/864 [14:00<07:17,  1.32s/it]
 62%|██████▏   | 533/864 [14:01<07:15,  1.32s/it]
 62%|██████▏   | 534/864 [14:03<07:14,  1.32s/it]
 62%|██████▏   | 535/864 [14:04<07:13,  1.32s/it]
 62%|██████▏   | 536/864 [14:05<07:12,  1.32s/it]
 62%|██████▏   | 537/864 [14:07<07:10,  1.32s/it]
 62%|██████▏   | 538/864 [14:08<07:09,  1.32s/it]
 62%|██████▏   | 539/864 [14:09<07:07,  1.32s/it]
 62%|██████▎   | 540/864 [14:11<07:06,  1.32s/it]
                                                 
{'loss': '3.896', 'grad_norm': '0.1166', 'learning_rate': '0.0012', 'epoch': '7.503'}
+
 62%|██████▎   | 540/864 [14:11<07:06,  1.32s/it]
 63%|██████▎   | 541/864 [14:12<07:05,  1.32s/it]
 63%|██████▎   | 542/864 [14:13<07:04,  1.32s/it]
 63%|██████▎   | 543/864 [14:15<07:02,  1.32s/it]
 63%|██████▎   | 544/864 [14:16<07:00,  1.32s/it]
 63%|██████▎   | 545/864 [14:17<07:00,  1.32s/it]
 63%|██████▎   | 546/864 [14:18<06:58,  1.32s/it]
 63%|██████▎   | 547/864 [14:20<06:57,  1.32s/it]
 63%|██████▎   | 548/864 [14:21<06:55,  1.32s/it]
 64%|██████▎   | 549/864 [14:22<06:54,  1.32s/it]
 64%|██████▎   | 550/864 [14:24<06:53,  1.32s/it]
                                                 
{'loss': '3.87', 'grad_norm': '0.1185', 'learning_rate': '0.0012', 'epoch': '7.642'}
+
 64%|██████▎   | 550/864 [14:24<06:53,  1.32s/it]
 64%|██████▍   | 551/864 [14:25<06:52,  1.32s/it]
 64%|██████▍   | 552/864 [14:26<06:50,  1.32s/it]
 64%|██████▍   | 553/864 [14:28<06:49,  1.32s/it]
 64%|██████▍   | 554/864 [14:29<06:48,  1.32s/it]
 64%|██████▍   | 555/864 [14:30<06:46,  1.32s/it]
 64%|██████▍   | 556/864 [14:32<06:45,  1.32s/it]
 64%|██████▍   | 557/864 [14:33<06:44,  1.32s/it]
 65%|██████▍   | 558/864 [14:34<06:42,  1.32s/it]
 65%|██████▍   | 559/864 [14:36<06:41,  1.32s/it]
 65%|██████▍   | 560/864 [14:37<06:40,  1.32s/it]
                                                 
{'loss': '3.855', 'grad_norm': '0.2435', 'learning_rate': '0.0012', 'epoch': '7.782'}
+
 65%|██████▍   | 560/864 [14:37<06:40,  1.32s/it]
 65%|██████▍   | 561/864 [14:38<06:39,  1.32s/it]
 65%|██████▌   | 562/864 [14:40<06:37,  1.32s/it]
 65%|██████▌   | 563/864 [14:41<06:36,  1.32s/it]
 65%|██████▌   | 564/864 [14:42<06:35,  1.32s/it]
 65%|██████▌   | 565/864 [14:44<06:34,  1.32s/it]
 66%|██████▌   | 566/864 [14:45<06:32,  1.32s/it]
 66%|██████▌   | 567/864 [14:46<06:31,  1.32s/it]
 66%|██████▌   | 568/864 [14:47<06:30,  1.32s/it]
 66%|██████▌   | 569/864 [14:49<06:28,  1.32s/it]
 66%|██████▌   | 570/864 [14:50<06:27,  1.32s/it]
                                                 
{'loss': '3.85', 'grad_norm': '0.1551', 'learning_rate': '0.0012', 'epoch': '7.921'}
+
 66%|██████▌   | 570/864 [14:50<06:27,  1.32s/it]
 66%|██████▌   | 571/864 [14:51<06:26,  1.32s/it]
 66%|██████▌   | 572/864 [14:53<06:24,  1.32s/it]
 66%|██████▋   | 573/864 [14:54<06:23,  1.32s/it]
 66%|██████▋   | 574/864 [14:55<06:22,  1.32s/it]
 67%|██████▋   | 575/864 [14:57<06:20,  1.32s/it]
 67%|██████▋   | 576/864 [14:58<05:37,  1.17s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:02, 14.92it/s]
+
 10%|█         | 4/39 [00:00<00:03,  9.16it/s]
+
 15%|█▌        | 6/39 [00:00<00:04,  8.16it/s]
+
 18%|█▊        | 7/39 [00:00<00:04,  7.91it/s]
+
 21%|██        | 8/39 [00:00<00:04,  7.71it/s]
+
 23%|██▎       | 9/39 [00:01<00:03,  7.58it/s]
+
 26%|██▌       | 10/39 [00:01<00:03,  7.48it/s]
+
 28%|██▊       | 11/39 [00:01<00:03,  7.41it/s]
+
 31%|███       | 12/39 [00:01<00:03,  7.35it/s]
+
 33%|███▎      | 13/39 [00:01<00:03,  7.31it/s]
+
 36%|███▌      | 14/39 [00:01<00:03,  7.29it/s]
+
 38%|███▊      | 15/39 [00:01<00:03,  7.28it/s]
+
 41%|████      | 16/39 [00:02<00:03,  7.27it/s]
+
 44%|████▎     | 17/39 [00:02<00:03,  7.27it/s]
+
 46%|████▌     | 18/39 [00:02<00:02,  7.27it/s]
+
 49%|████▊     | 19/39 [00:02<00:02,  7.26it/s]
+
 51%|█████▏    | 20/39 [00:02<00:02,  7.25it/s]
+
 54%|█████▍    | 21/39 [00:02<00:02,  7.24it/s]
+
 56%|█████▋    | 22/39 [00:02<00:02,  7.24it/s]
+
 59%|█████▉    | 23/39 [00:03<00:02,  7.24it/s]
+
 62%|██████▏   | 24/39 [00:03<00:02,  7.24it/s]
+
 64%|██████▍   | 25/39 [00:03<00:01,  7.25it/s]
+
 67%|██████▋   | 26/39 [00:03<00:01,  7.25it/s]
+
 69%|██████▉   | 27/39 [00:03<00:01,  7.25it/s]
+
 72%|███████▏  | 28/39 [00:03<00:01,  7.24it/s]
+
 74%|███████▍  | 29/39 [00:03<00:01,  7.24it/s]
+
 77%|███████▋  | 30/39 [00:04<00:01,  7.24it/s]
+
 79%|███████▉  | 31/39 [00:04<00:01,  7.25it/s]
+
 82%|████████▏ | 32/39 [00:04<00:00,  7.25it/s]
+
 85%|████████▍ | 33/39 [00:04<00:00,  7.25it/s]
+
 87%|████████▋ | 34/39 [00:04<00:00,  7.26it/s]
+
 90%|████████▉ | 35/39 [00:04<00:00,  7.25it/s]
+
 92%|█████████▏| 36/39 [00:04<00:00,  7.26it/s]
+
 95%|█████████▍| 37/39 [00:04<00:00,  7.26it/s]
+
 97%|█████████▋| 38/39 [00:05<00:00,  7.26it/s]
+
100%|██████████| 39/39 [00:05<00:00,  7.24it/s]
                                                 
+
                                               
{'eval_loss': '4.089', 'eval_runtime': '5.436', 'eval_samples_per_second': '449', 'eval_steps_per_second': '7.174', 'epoch': '8'}
+
 67%|██████▋   | 576/864 [15:03<05:37,  1.17s/it]
+
100%|██████████| 39/39 [00:05<00:00,  7.24it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0012/checkpoint-576
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0012/checkpoint-576/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0012/checkpoint-576/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:01<00:00,  1.58s/it]
Writing model shards: 100%|██████████| 1/1 [00:01<00:00,  1.58s/it]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0012/checkpoint-576/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0012/checkpoint-576/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0012/checkpoint-576/tokenizer_config.json
+[eval] epoch 8.0 step 576 loss 4.0889 (best 4.0889 @ ep 8.0, 0/10000 since best)
+
+
                                               
 67%|██████▋   | 577/864 [15:10<21:10,  4.43s/it]
 67%|██████▋   | 578/864 [15:11<16:38,  3.49s/it]
 67%|██████▋   | 579/864 [15:12<13:28,  2.84s/it]
 67%|██████▋   | 580/864 [15:13<11:16,  2.38s/it]
                                                 
{'loss': '3.762', 'grad_norm': '0.2743', 'learning_rate': '0.0012', 'epoch': '8.056'}
+
 67%|██████▋   | 580/864 [15:13<11:16,  2.38s/it]
 67%|██████▋   | 581/864 [15:15<09:43,  2.06s/it]
 67%|██████▋   | 582/864 [15:16<08:37,  1.84s/it]
 67%|██████▋   | 583/864 [15:17<07:52,  1.68s/it]
 68%|██████▊   | 584/864 [15:19<07:19,  1.57s/it]
 68%|██████▊   | 585/864 [15:20<06:56,  1.49s/it]
 68%|██████▊   | 586/864 [15:21<06:40,  1.44s/it]
 68%|██████▊   | 587/864 [15:23<06:28,  1.40s/it]
 68%|██████▊   | 588/864 [15:24<06:19,  1.38s/it]
 68%|██████▊   | 589/864 [15:25<06:13,  1.36s/it]
 68%|██████▊   | 590/864 [15:27<06:08,  1.34s/it]
                                                 
{'loss': '3.716', 'grad_norm': '0.3087', 'learning_rate': '0.0012', 'epoch': '8.195'}
+
 68%|██████▊   | 590/864 [15:27<06:08,  1.34s/it]
 68%|██████▊   | 591/864 [15:28<06:04,  1.33s/it]
 69%|██████▊   | 592/864 [15:29<06:01,  1.33s/it]
 69%|██████▊   | 593/864 [15:31<05:58,  1.32s/it]
 69%|██████▉   | 594/864 [15:32<05:56,  1.32s/it]
 69%|██████▉   | 595/864 [15:33<05:54,  1.32s/it]
 69%|██████▉   | 596/864 [15:34<05:53,  1.32s/it]
 69%|██████▉   | 597/864 [15:36<05:51,  1.32s/it]
 69%|██████▉   | 598/864 [15:37<05:49,  1.31s/it]
 69%|██████▉   | 599/864 [15:38<05:48,  1.31s/it]
 69%|██████▉   | 600/864 [15:40<05:47,  1.32s/it]
                                                 
{'loss': '3.728', 'grad_norm': '0.1924', 'learning_rate': '0.0012', 'epoch': '8.335'}
+
 69%|██████▉   | 600/864 [15:40<05:47,  1.32s/it]
 70%|██████▉   | 601/864 [15:41<05:45,  1.31s/it]
 70%|██████▉   | 602/864 [15:42<05:44,  1.31s/it]
 70%|██████▉   | 603/864 [15:44<05:43,  1.31s/it]
 70%|██████▉   | 604/864 [15:45<05:41,  1.31s/it]
 70%|███████   | 605/864 [15:46<05:39,  1.31s/it]
 70%|███████   | 606/864 [15:48<05:38,  1.31s/it]
 70%|███████   | 607/864 [15:49<05:37,  1.31s/it]
 70%|███████   | 608/864 [15:50<05:36,  1.31s/it]
 70%|███████   | 609/864 [15:52<05:35,  1.31s/it]
 71%|███████   | 610/864 [15:53<05:33,  1.31s/it]
                                                 
{'loss': '3.688', 'grad_norm': '0.1232', 'learning_rate': '0.0012', 'epoch': '8.475'}
+
 71%|███████   | 610/864 [15:53<05:33,  1.31s/it]
 71%|███████   | 611/864 [15:54<05:32,  1.31s/it]
 71%|███████   | 612/864 [15:56<05:31,  1.31s/it]
 71%|███████   | 613/864 [15:57<05:29,  1.31s/it]
 71%|███████   | 614/864 [15:58<05:28,  1.31s/it]
 71%|███████   | 615/864 [15:59<05:26,  1.31s/it]
 71%|███████▏  | 616/864 [16:01<05:25,  1.31s/it]
 71%|███████▏  | 617/864 [16:02<05:24,  1.31s/it]
 72%|███████▏  | 618/864 [16:03<05:22,  1.31s/it]
 72%|███████▏  | 619/864 [16:05<05:21,  1.31s/it]
 72%|███████▏  | 620/864 [16:06<05:20,  1.31s/it]
                                                 
{'loss': '3.669', 'grad_norm': '0.1347', 'learning_rate': '0.0012', 'epoch': '8.614'}
+
 72%|███████▏  | 620/864 [16:06<05:20,  1.31s/it]
 72%|███████▏  | 621/864 [16:07<05:19,  1.31s/it]
 72%|███████▏  | 622/864 [16:09<05:17,  1.31s/it]
 72%|███████▏  | 623/864 [16:10<05:16,  1.31s/it]
 72%|███████▏  | 624/864 [16:11<05:15,  1.31s/it]
 72%|███████▏  | 625/864 [16:13<05:13,  1.31s/it]
 72%|███████▏  | 626/864 [16:14<05:12,  1.31s/it]
 73%|███████▎  | 627/864 [16:15<05:10,  1.31s/it]
 73%|███████▎  | 628/864 [16:17<05:09,  1.31s/it]
 73%|███████▎  | 629/864 [16:18<05:08,  1.31s/it]
 73%|███████▎  | 630/864 [16:19<05:07,  1.31s/it]
                                                 
{'loss': '3.661', 'grad_norm': '0.1114', 'learning_rate': '0.0012', 'epoch': '8.754'}
+
 73%|███████▎  | 630/864 [16:19<05:07,  1.31s/it]
 73%|███████▎  | 631/864 [16:20<05:06,  1.31s/it]
 73%|███████▎  | 632/864 [16:22<05:04,  1.31s/it]
 73%|███████▎  | 633/864 [16:23<05:03,  1.31s/it]
 73%|███████▎  | 634/864 [16:24<05:02,  1.31s/it]
 73%|███████▎  | 635/864 [16:26<05:00,  1.31s/it]
 74%|███████▎  | 636/864 [16:27<04:59,  1.31s/it]
 74%|███████▎  | 637/864 [16:29<05:21,  1.41s/it]
 74%|███████▍  | 638/864 [16:30<05:12,  1.38s/it]
 74%|███████▍  | 639/864 [16:31<05:06,  1.36s/it]
 74%|███████▍  | 640/864 [16:33<05:02,  1.35s/it]
                                                 
{'loss': '3.672', 'grad_norm': '0.1845', 'learning_rate': '0.0012', 'epoch': '8.894'}
+
 74%|███████▍  | 640/864 [16:33<05:02,  1.35s/it]
 74%|███████▍  | 641/864 [16:34<04:58,  1.34s/it]
 74%|███████▍  | 642/864 [16:35<04:55,  1.33s/it]
 74%|███████▍  | 643/864 [16:37<04:53,  1.33s/it]
 75%|███████▍  | 644/864 [16:38<04:51,  1.32s/it]
 75%|███████▍  | 645/864 [16:39<04:49,  1.32s/it]
 75%|███████▍  | 646/864 [16:41<04:47,  1.32s/it]
 75%|███████▍  | 647/864 [16:42<04:46,  1.32s/it]
 75%|███████▌  | 648/864 [16:43<04:13,  1.17s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:02, 14.83it/s]
+
 10%|█         | 4/39 [00:00<00:03,  9.16it/s]
+
 15%|█▌        | 6/39 [00:00<00:04,  8.18it/s]
+
 18%|█▊        | 7/39 [00:00<00:04,  7.94it/s]
+
 21%|██        | 8/39 [00:00<00:04,  7.74it/s]
+
 23%|██▎       | 9/39 [00:01<00:03,  7.60it/s]
+
 26%|██▌       | 10/39 [00:01<00:03,  7.51it/s]
+
 28%|██▊       | 11/39 [00:01<00:03,  7.43it/s]
+
 31%|███       | 12/39 [00:01<00:03,  7.37it/s]
+
 33%|███▎      | 13/39 [00:01<00:03,  7.33it/s]
+
 36%|███▌      | 14/39 [00:01<00:03,  7.30it/s]
+
 38%|███▊      | 15/39 [00:01<00:03,  7.28it/s]
+
 41%|████      | 16/39 [00:02<00:03,  7.27it/s]
+
 44%|████▎     | 17/39 [00:02<00:03,  7.27it/s]
+
 46%|████▌     | 18/39 [00:02<00:02,  7.26it/s]
+
 49%|████▊     | 19/39 [00:02<00:02,  7.26it/s]
+
 51%|█████▏    | 20/39 [00:02<00:02,  7.26it/s]
+
 54%|█████▍    | 21/39 [00:02<00:02,  7.25it/s]
+
 56%|█████▋    | 22/39 [00:02<00:02,  7.25it/s]
+
 59%|█████▉    | 23/39 [00:03<00:02,  7.25it/s]
+
 62%|██████▏   | 24/39 [00:03<00:02,  7.24it/s]
+
 64%|██████▍   | 25/39 [00:03<00:01,  7.22it/s]
+
 67%|██████▋   | 26/39 [00:03<00:01,  7.22it/s]
+
 69%|██████▉   | 27/39 [00:03<00:01,  7.23it/s]
+
 72%|███████▏  | 28/39 [00:03<00:01,  7.23it/s]
+
 74%|███████▍  | 29/39 [00:03<00:01,  7.24it/s]
+
 77%|███████▋  | 30/39 [00:04<00:01,  7.23it/s]
+
 79%|███████▉  | 31/39 [00:04<00:01,  7.24it/s]
+
 82%|████████▏ | 32/39 [00:04<00:00,  7.24it/s]
+
 85%|████████▍ | 33/39 [00:04<00:00,  7.24it/s]
+
 87%|████████▋ | 34/39 [00:04<00:00,  7.25it/s]
+
 90%|████████▉ | 35/39 [00:04<00:00,  7.25it/s]
+
 92%|█████████▏| 36/39 [00:04<00:00,  7.26it/s]
+
 95%|█████████▍| 37/39 [00:04<00:00,  7.26it/s]
+
 97%|█████████▋| 38/39 [00:05<00:00,  7.26it/s]
+
100%|██████████| 39/39 [00:05<00:00,  7.26it/s]
                                                 
+
                                               
{'eval_loss': '4.006', 'eval_runtime': '5.437', 'eval_samples_per_second': '449', 'eval_steps_per_second': '7.173', 'epoch': '9'}
+
 75%|███████▌  | 648/864 [16:48<04:13,  1.17s/it]
+
100%|██████████| 39/39 [00:05<00:00,  7.26it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0012/checkpoint-648
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0012/checkpoint-648/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0012/checkpoint-648/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:01<00:00,  1.49s/it]
Writing model shards: 100%|██████████| 1/1 [00:01<00:00,  1.49s/it]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0012/checkpoint-648/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0012/checkpoint-648/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0012/checkpoint-648/tokenizer_config.json
+[eval] epoch 9.0 step 648 loss 4.0057 (best 4.0057 @ ep 9.0, 0/10000 since best)
+
+
                                               
 75%|███████▌  | 649/864 [16:55<16:04,  4.49s/it]
 75%|███████▌  | 650/864 [16:56<12:36,  3.53s/it]
                                                 
{'loss': '3.623', 'grad_norm': '0.1834', 'learning_rate': '0.0012', 'epoch': '9.028'}
+
 75%|███████▌  | 650/864 [16:56<12:36,  3.53s/it]
 75%|███████▌  | 651/864 [16:58<10:10,  2.87s/it]
 75%|███████▌  | 652/864 [16:59<08:29,  2.40s/it]
 76%|███████▌  | 653/864 [17:00<07:17,  2.08s/it]
 76%|███████▌  | 654/864 [17:01<06:27,  1.85s/it]
 76%|███████▌  | 655/864 [17:03<05:52,  1.69s/it]
 76%|███████▌  | 656/864 [17:04<05:27,  1.57s/it]
 76%|███████▌  | 657/864 [17:05<05:09,  1.50s/it]
 76%|███████▌  | 658/864 [17:07<04:57,  1.44s/it]
 76%|███████▋  | 659/864 [17:08<04:47,  1.40s/it]
 76%|███████▋  | 660/864 [17:09<04:40,  1.38s/it]
                                                 
{'loss': '3.556', 'grad_norm': '0.493', 'learning_rate': '0.0012', 'epoch': '9.168'}
+
 76%|███████▋  | 660/864 [17:09<04:40,  1.38s/it]
 77%|███████▋  | 661/864 [17:11<04:35,  1.36s/it]
 77%|███████▋  | 662/864 [17:12<04:31,  1.35s/it]
 77%|███████▋  | 663/864 [17:14<04:57,  1.48s/it]
 77%|███████▋  | 664/864 [17:15<04:45,  1.43s/it]
 77%|███████▋  | 665/864 [17:17<04:57,  1.50s/it]
 77%|███████▋  | 666/864 [17:18<04:45,  1.44s/it]
 77%|███████▋  | 667/864 [17:19<04:36,  1.40s/it]
 77%|███████▋  | 668/864 [17:21<04:30,  1.38s/it]
 77%|███████▋  | 669/864 [17:22<04:24,  1.36s/it]
 78%|███████▊  | 670/864 [17:23<04:21,  1.35s/it]
                                                 
{'loss': '4.133', 'grad_norm': '1.285', 'learning_rate': '0.0012', 'epoch': '9.307'}
+
 78%|███████▊  | 670/864 [17:23<04:21,  1.35s/it]
 78%|███████▊  | 671/864 [17:25<04:18,  1.34s/it]
 78%|███████▊  | 672/864 [17:26<04:15,  1.33s/it]
 78%|███████▊  | 673/864 [17:27<04:13,  1.33s/it]
 78%|███████▊  | 674/864 [17:29<04:11,  1.32s/it]
 78%|███████▊  | 675/864 [17:30<04:09,  1.32s/it]
 78%|███████▊  | 676/864 [17:31<04:08,  1.32s/it]
 78%|███████▊  | 677/864 [17:33<04:06,  1.32s/it]
 78%|███████▊  | 678/864 [17:34<04:05,  1.32s/it]
 79%|███████▊  | 679/864 [17:35<04:04,  1.32s/it]
 79%|███████▊  | 680/864 [17:36<04:02,  1.32s/it]
                                                 
{'loss': '5.06', 'grad_norm': '1.536', 'learning_rate': '0.0012', 'epoch': '9.447'}
+
 79%|███████▊  | 680/864 [17:36<04:02,  1.32s/it]
 79%|███████▉  | 681/864 [17:38<04:01,  1.32s/it]
 79%|███████▉  | 682/864 [17:39<03:59,  1.32s/it]
 79%|███████▉  | 683/864 [17:40<03:58,  1.32s/it]
 79%|███████▉  | 684/864 [17:42<03:57,  1.32s/it]
 79%|███████▉  | 685/864 [17:43<03:55,  1.32s/it]
 79%|███████▉  | 686/864 [17:44<03:54,  1.32s/it]
 80%|███████▉  | 687/864 [17:46<03:53,  1.32s/it]
 80%|███████▉  | 688/864 [17:47<03:51,  1.32s/it]
 80%|███████▉  | 689/864 [17:48<03:50,  1.32s/it]
 80%|███████▉  | 690/864 [17:50<03:49,  1.32s/it]
                                                 
{'loss': '5.239', 'grad_norm': '0.529', 'learning_rate': '0.0012', 'epoch': '9.586'}
+
 80%|███████▉  | 690/864 [17:50<03:49,  1.32s/it]
 80%|███████▉  | 691/864 [17:51<03:47,  1.32s/it]
 80%|████████  | 692/864 [17:52<03:46,  1.32s/it]
 80%|████████  | 693/864 [17:54<03:45,  1.32s/it]
 80%|████████  | 694/864 [17:55<03:43,  1.32s/it]
 80%|████████  | 695/864 [17:56<03:42,  1.32s/it]
 81%|████████  | 696/864 [17:58<03:41,  1.32s/it]
 81%|████████  | 697/864 [17:59<03:40,  1.32s/it]
 81%|████████  | 698/864 [18:00<03:38,  1.32s/it]
 81%|████████  | 699/864 [18:02<03:37,  1.32s/it]
 81%|████████  | 700/864 [18:03<03:36,  1.32s/it]
                                                 
{'loss': '4.587', 'grad_norm': '0.1746', 'learning_rate': '0.0012', 'epoch': '9.726'}
+
 81%|████████  | 700/864 [18:03<03:36,  1.32s/it]
 81%|████████  | 701/864 [18:04<03:34,  1.32s/it]
 81%|████████▏ | 702/864 [18:05<03:33,  1.32s/it]
 81%|████████▏ | 703/864 [18:07<03:32,  1.32s/it]
 81%|████████▏ | 704/864 [18:08<03:30,  1.32s/it]
 82%|████████▏ | 705/864 [18:09<03:29,  1.32s/it]
 82%|████████▏ | 706/864 [18:11<03:28,  1.32s/it]
 82%|████████▏ | 707/864 [18:12<03:27,  1.32s/it]
 82%|████████▏ | 708/864 [18:13<03:25,  1.32s/it]
 82%|████████▏ | 709/864 [18:15<03:24,  1.32s/it]
 82%|████████▏ | 710/864 [18:16<03:22,  1.32s/it]
                                                 
{'loss': '4.65', 'grad_norm': '0.2742', 'learning_rate': '0.0012', 'epoch': '9.866'}
+
 82%|████████▏ | 710/864 [18:16<03:22,  1.32s/it]
 82%|████████▏ | 711/864 [18:17<03:21,  1.32s/it]
 82%|████████▏ | 712/864 [18:19<03:20,  1.32s/it]
 83%|████████▎ | 713/864 [18:20<03:18,  1.32s/it]
 83%|████████▎ | 714/864 [18:21<03:17,  1.32s/it]
 83%|████████▎ | 715/864 [18:23<03:16,  1.32s/it]
 83%|████████▎ | 716/864 [18:24<03:15,  1.32s/it]
 83%|████████▎ | 717/864 [18:25<03:13,  1.32s/it]
 83%|████████▎ | 718/864 [18:27<03:12,  1.32s/it]
 83%|████████▎ | 719/864 [18:28<03:11,  1.32s/it]
 83%|████████▎ | 720/864 [18:29<02:48,  1.17s/it]
                                                 
{'loss': '4.335', 'grad_norm': '0.1579', 'learning_rate': '0.0012', 'epoch': '10'}
+
 83%|████████▎ | 720/864 [18:29<02:48,  1.17s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:02, 14.53it/s]
+
 10%|█         | 4/39 [00:00<00:03,  9.13it/s]
+
 15%|█▌        | 6/39 [00:00<00:04,  8.16it/s]
+
 18%|█▊        | 7/39 [00:00<00:04,  7.92it/s]
+
 21%|██        | 8/39 [00:00<00:04,  7.74it/s]
+
 23%|██▎       | 9/39 [00:01<00:03,  7.60it/s]
+
 26%|██▌       | 10/39 [00:01<00:03,  7.48it/s]
+
 28%|██▊       | 11/39 [00:01<00:03,  7.41it/s]
+
 31%|███       | 12/39 [00:01<00:03,  7.36it/s]
+
 33%|███▎      | 13/39 [00:01<00:03,  7.33it/s]
+
 36%|███▌      | 14/39 [00:01<00:03,  7.31it/s]
+
 38%|███▊      | 15/39 [00:01<00:03,  7.27it/s]
+
 41%|████      | 16/39 [00:02<00:03,  7.26it/s]
+
 44%|████▎     | 17/39 [00:02<00:03,  7.25it/s]
+
 46%|████▌     | 18/39 [00:02<00:02,  7.24it/s]
+
 49%|████▊     | 19/39 [00:02<00:02,  7.24it/s]
+
 51%|█████▏    | 20/39 [00:02<00:02,  7.24it/s]
+
 54%|█████▍    | 21/39 [00:02<00:02,  7.23it/s]
+
 56%|█████▋    | 22/39 [00:02<00:02,  7.23it/s]
+
 59%|█████▉    | 23/39 [00:03<00:02,  7.24it/s]
+
 62%|██████▏   | 24/39 [00:03<00:02,  7.24it/s]
+
 64%|██████▍   | 25/39 [00:03<00:01,  7.25it/s]
+
 67%|██████▋   | 26/39 [00:03<00:01,  7.25it/s]
+
 69%|██████▉   | 27/39 [00:03<00:01,  7.25it/s]
+
 72%|███████▏  | 28/39 [00:03<00:01,  7.25it/s]
+
 74%|███████▍  | 29/39 [00:03<00:01,  7.25it/s]
+
 77%|███████▋  | 30/39 [00:04<00:01,  7.24it/s]
+
 79%|███████▉  | 31/39 [00:04<00:01,  7.23it/s]
+
 82%|████████▏ | 32/39 [00:04<00:00,  7.23it/s]
+
 85%|████████▍ | 33/39 [00:04<00:00,  7.24it/s]
+
 87%|████████▋ | 34/39 [00:04<00:00,  7.24it/s]
+
 90%|████████▉ | 35/39 [00:04<00:00,  7.25it/s]
+
 92%|█████████▏| 36/39 [00:04<00:00,  7.25it/s]
+
 95%|█████████▍| 37/39 [00:04<00:00,  7.25it/s]
+
 97%|█████████▋| 38/39 [00:05<00:00,  7.26it/s]
+
100%|██████████| 39/39 [00:05<00:00,  7.25it/s]
                                                 
+
                                               
{'eval_loss': '4.461', 'eval_runtime': '5.454', 'eval_samples_per_second': '447.6', 'eval_steps_per_second': '7.151', 'epoch': '10'}
+
 83%|████████▎ | 720/864 [18:34<02:48,  1.17s/it]
+
100%|██████████| 39/39 [00:05<00:00,  7.25it/s][eval] epoch 10.0 step 720 loss 4.4606 (best 4.0057 @ ep 9.0, 1/10000 since best)
+
+
                                               
 83%|████████▎ | 721/864 [18:36<06:51,  2.88s/it]
 84%|████████▎ | 722/864 [18:37<05:41,  2.41s/it]
 84%|████████▎ | 723/864 [18:38<04:53,  2.08s/it]
 84%|████████▍ | 724/864 [18:40<04:19,  1.85s/it]
 84%|████████▍ | 725/864 [18:41<03:55,  1.69s/it]
 84%|████████▍ | 726/864 [18:42<03:38,  1.58s/it]
 84%|████████▍ | 727/864 [18:43<03:25,  1.50s/it]
 84%|████████▍ | 728/864 [18:45<03:16,  1.45s/it]
 84%|████████▍ | 729/864 [18:46<03:10,  1.41s/it]
 84%|████████▍ | 730/864 [18:47<03:05,  1.38s/it]
                                                 
{'loss': '4.101', 'grad_norm': '0.113', 'learning_rate': '0.0012', 'epoch': '10.14'}
+
 84%|████████▍ | 730/864 [18:47<03:05,  1.38s/it]
 85%|████████▍ | 731/864 [18:49<03:01,  1.36s/it]
 85%|████████▍ | 732/864 [18:50<02:58,  1.35s/it]
 85%|████████▍ | 733/864 [18:51<02:55,  1.34s/it]
 85%|████████▍ | 734/864 [18:53<02:53,  1.33s/it]
 85%|████████▌ | 735/864 [18:54<02:51,  1.33s/it]
 85%|████████▌ | 736/864 [18:55<02:49,  1.33s/it]
 85%|████████▌ | 737/864 [18:57<02:47,  1.32s/it]
 85%|████████▌ | 738/864 [18:58<02:46,  1.32s/it]
 86%|████████▌ | 739/864 [18:59<02:45,  1.32s/it]
 86%|████████▌ | 740/864 [19:01<02:43,  1.32s/it]
                                                 
{'loss': '3.986', 'grad_norm': '0.1381', 'learning_rate': '0.0012', 'epoch': '10.28'}
+
 86%|████████▌ | 740/864 [19:01<02:43,  1.32s/it]
 86%|████████▌ | 741/864 [19:02<02:42,  1.32s/it]
 86%|████████▌ | 742/864 [19:03<02:40,  1.32s/it]
 86%|████████▌ | 743/864 [19:05<02:39,  1.32s/it]
 86%|████████▌ | 744/864 [19:06<02:38,  1.32s/it]
 86%|████████▌ | 745/864 [19:07<02:36,  1.32s/it]
 86%|████████▋ | 746/864 [19:09<02:35,  1.32s/it]
 86%|████████▋ | 747/864 [19:10<02:34,  1.32s/it]
 87%|████████▋ | 748/864 [19:11<02:32,  1.32s/it]
 87%|████████▋ | 749/864 [19:12<02:31,  1.32s/it]
 87%|████████▋ | 750/864 [19:14<02:30,  1.32s/it]
                                                 
{'loss': '3.91', 'grad_norm': '0.1221', 'learning_rate': '0.0012', 'epoch': '10.42'}
+
 87%|████████▋ | 750/864 [19:14<02:30,  1.32s/it]
 87%|████████▋ | 751/864 [19:15<02:28,  1.32s/it]
 87%|████████▋ | 752/864 [19:16<02:27,  1.32s/it]
 87%|████████▋ | 753/864 [19:18<02:26,  1.32s/it]
 87%|████████▋ | 754/864 [19:19<02:24,  1.32s/it]
 87%|████████▋ | 755/864 [19:20<02:23,  1.32s/it]
 88%|████████▊ | 756/864 [19:22<02:22,  1.32s/it]
 88%|████████▊ | 757/864 [19:23<02:21,  1.32s/it]
 88%|████████▊ | 758/864 [19:24<02:19,  1.32s/it]
 88%|████████▊ | 759/864 [19:26<02:18,  1.32s/it]
 88%|████████▊ | 760/864 [19:27<02:16,  1.32s/it]
                                                 
{'loss': '3.858', 'grad_norm': '0.1023', 'learning_rate': '0.0012', 'epoch': '10.56'}
+
 88%|████████▊ | 760/864 [19:27<02:16,  1.32s/it]
 88%|████████▊ | 761/864 [19:28<02:15,  1.32s/it]
 88%|████████▊ | 762/864 [19:30<02:14,  1.32s/it]
 88%|████████▊ | 763/864 [19:31<02:13,  1.32s/it]
 88%|████████▊ | 764/864 [19:32<02:11,  1.32s/it]
 89%|████████▊ | 765/864 [19:34<02:10,  1.32s/it]
 89%|████████▊ | 766/864 [19:35<02:09,  1.32s/it]
 89%|████████▉ | 767/864 [19:36<02:07,  1.32s/it]
 89%|████████▉ | 768/864 [19:37<02:06,  1.32s/it]
 89%|████████▉ | 769/864 [19:39<02:05,  1.32s/it]
 89%|████████▉ | 770/864 [19:40<02:03,  1.32s/it]
                                                 
{'loss': '3.813', 'grad_norm': '0.2263', 'learning_rate': '0.0012', 'epoch': '10.7'}
+
 89%|████████▉ | 770/864 [19:40<02:03,  1.32s/it]
 89%|████████▉ | 771/864 [19:41<02:02,  1.32s/it]
 89%|████████▉ | 772/864 [19:43<02:01,  1.32s/it]
 89%|████████▉ | 773/864 [19:44<02:00,  1.32s/it]
 90%|████████▉ | 774/864 [19:45<01:58,  1.32s/it]
 90%|████████▉ | 775/864 [19:47<01:57,  1.32s/it]
 90%|████████▉ | 776/864 [19:48<01:55,  1.32s/it]
 90%|████████▉ | 777/864 [19:49<01:54,  1.32s/it]
 90%|█████████ | 778/864 [19:51<01:53,  1.32s/it]
 90%|█████████ | 779/864 [19:52<01:51,  1.32s/it]
 90%|█████████ | 780/864 [19:53<01:50,  1.32s/it]
                                                 
{'loss': '3.783', 'grad_norm': '0.1103', 'learning_rate': '0.0012', 'epoch': '10.84'}
+
 90%|█████████ | 780/864 [19:53<01:50,  1.32s/it]
 90%|█████████ | 781/864 [19:55<01:49,  1.32s/it]
 91%|█████████ | 782/864 [19:56<01:48,  1.32s/it]
 91%|█████████ | 783/864 [19:57<01:46,  1.32s/it]
 91%|█████████ | 784/864 [19:59<01:45,  1.32s/it]
 91%|█████████ | 785/864 [20:00<01:44,  1.32s/it]
 91%|█████████ | 786/864 [20:01<01:42,  1.32s/it]
 91%|█████████ | 787/864 [20:03<01:41,  1.32s/it]
 91%|█████████ | 788/864 [20:04<01:40,  1.32s/it]
 91%|█████████▏| 789/864 [20:05<01:38,  1.32s/it]
 91%|█████████▏| 790/864 [20:06<01:37,  1.32s/it]
                                                 
{'loss': '3.734', 'grad_norm': '0.1053', 'learning_rate': '0.0012', 'epoch': '10.98'}
+
 91%|█████████▏| 790/864 [20:06<01:37,  1.32s/it]
 92%|█████████▏| 791/864 [20:08<01:36,  1.32s/it]
 92%|█████████▏| 792/864 [20:09<01:24,  1.17s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:02, 14.99it/s]
+
 10%|█         | 4/39 [00:00<00:03,  9.19it/s]
+
 15%|█▌        | 6/39 [00:00<00:04,  8.20it/s]
+
 18%|█▊        | 7/39 [00:00<00:04,  7.94it/s]
+
 21%|██        | 8/39 [00:00<00:04,  7.75it/s]
+
 23%|██▎       | 9/39 [00:01<00:03,  7.59it/s]
+
 26%|██▌       | 10/39 [00:01<00:03,  7.47it/s]
+
 28%|██▊       | 11/39 [00:01<00:03,  7.40it/s]
+
 31%|███       | 12/39 [00:01<00:03,  7.35it/s]
+
 33%|███▎      | 13/39 [00:01<00:03,  7.31it/s]
+
 36%|███▌      | 14/39 [00:01<00:03,  7.28it/s]
+
 38%|███▊      | 15/39 [00:01<00:03,  7.27it/s]
+
 41%|████      | 16/39 [00:02<00:03,  7.26it/s]
+
 44%|████▎     | 17/39 [00:02<00:03,  7.24it/s]
+
 46%|████▌     | 18/39 [00:02<00:02,  7.24it/s]
+
 49%|████▊     | 19/39 [00:02<00:02,  7.23it/s]
+
 51%|█████▏    | 20/39 [00:02<00:02,  7.24it/s]
+
 54%|█████▍    | 21/39 [00:02<00:02,  7.23it/s]
+
 56%|█████▋    | 22/39 [00:02<00:02,  7.24it/s]
+
 59%|█████▉    | 23/39 [00:03<00:02,  7.24it/s]
+
 62%|██████▏   | 24/39 [00:03<00:02,  7.24it/s]
+
 64%|██████▍   | 25/39 [00:03<00:01,  7.25it/s]
+
 67%|██████▋   | 26/39 [00:03<00:01,  7.25it/s]
+
 69%|██████▉   | 27/39 [00:03<00:01,  7.24it/s]
+
 72%|███████▏  | 28/39 [00:03<00:01,  7.24it/s]
+
 74%|███████▍  | 29/39 [00:03<00:01,  7.24it/s]
+
 77%|███████▋  | 30/39 [00:04<00:01,  7.24it/s]
+
 79%|███████▉  | 31/39 [00:04<00:01,  7.24it/s]
+
 82%|████████▏ | 32/39 [00:04<00:00,  7.25it/s]
+
 85%|████████▍ | 33/39 [00:04<00:00,  7.25it/s]
+
 87%|████████▋ | 34/39 [00:04<00:00,  7.25it/s]
+
 90%|████████▉ | 35/39 [00:04<00:00,  7.24it/s]
+
 92%|█████████▏| 36/39 [00:04<00:00,  7.23it/s]
+
 95%|█████████▍| 37/39 [00:04<00:00,  7.23it/s]
+
 97%|█████████▋| 38/39 [00:05<00:00,  7.23it/s]
+
100%|██████████| 39/39 [00:05<00:00,  7.24it/s]
                                                 
+
                                               
{'eval_loss': '4.047', 'eval_runtime': '5.444', 'eval_samples_per_second': '448.4', 'eval_steps_per_second': '7.163', 'epoch': '11'}
+
 92%|█████████▏| 792/864 [20:14<01:24,  1.17s/it]
+
100%|██████████| 39/39 [00:05<00:00,  7.24it/s][eval] epoch 11.0 step 792 loss 4.0467 (best 4.0057 @ ep 9.0, 2/10000 since best)
+
+
                                               
 92%|█████████▏| 793/864 [20:15<03:23,  2.87s/it]
 92%|█████████▏| 794/864 [20:17<02:48,  2.40s/it]
 92%|█████████▏| 795/864 [20:18<02:23,  2.08s/it]
 92%|█████████▏| 796/864 [20:19<02:05,  1.85s/it]
 92%|█████████▏| 797/864 [20:21<01:53,  1.69s/it]
 92%|█████████▏| 798/864 [20:22<01:44,  1.58s/it]
 92%|█████████▏| 799/864 [20:23<01:37,  1.50s/it]
 93%|█████████▎| 800/864 [20:25<01:32,  1.44s/it]
                                                 
{'loss': '3.63', 'grad_norm': '0.2063', 'learning_rate': '0.0012', 'epoch': '11.11'}
+
 93%|█████████▎| 800/864 [20:25<01:32,  1.44s/it]
 93%|█████████▎| 801/864 [20:26<01:28,  1.41s/it]
 93%|█████████▎| 802/864 [20:27<01:25,  1.38s/it]
 93%|█████████▎| 803/864 [20:29<01:23,  1.36s/it]
 93%|█████████▎| 804/864 [20:30<01:20,  1.35s/it]
 93%|█████████▎| 805/864 [20:31<01:19,  1.34s/it]
 93%|█████████▎| 806/864 [20:33<01:17,  1.33s/it]
 93%|█████████▎| 807/864 [20:34<01:15,  1.33s/it]
 94%|█████████▎| 808/864 [20:35<01:14,  1.32s/it]
 94%|█████████▎| 809/864 [20:37<01:12,  1.32s/it]
 94%|█████████▍| 810/864 [20:38<01:11,  1.32s/it]
                                                 
{'loss': '3.567', 'grad_norm': '0.1189', 'learning_rate': '0.0012', 'epoch': '11.25'}
+
 94%|█████████▍| 810/864 [20:38<01:11,  1.32s/it]
 94%|█████████▍| 811/864 [20:39<01:09,  1.32s/it]
 94%|█████████▍| 812/864 [20:40<01:08,  1.32s/it]
 94%|█████████▍| 813/864 [20:42<01:07,  1.32s/it]
 94%|█████████▍| 814/864 [20:43<01:05,  1.32s/it]
 94%|█████████▍| 815/864 [20:44<01:04,  1.32s/it]
 94%|█████████▍| 816/864 [20:46<01:03,  1.32s/it]
 95%|█████████▍| 817/864 [20:47<01:01,  1.32s/it]
 95%|█████████▍| 818/864 [20:48<01:00,  1.32s/it]
 95%|█████████▍| 819/864 [20:50<00:59,  1.32s/it]
 95%|█████████▍| 820/864 [20:51<00:57,  1.32s/it]
                                                 
{'loss': '3.563', 'grad_norm': '0.09812', 'learning_rate': '0.0012', 'epoch': '11.39'}
+
 95%|█████████▍| 820/864 [20:51<00:57,  1.32s/it]
 95%|█████████▌| 821/864 [20:52<00:56,  1.32s/it]
 95%|█████████▌| 822/864 [20:54<00:55,  1.31s/it]
 95%|█████████▌| 823/864 [20:55<00:53,  1.32s/it]
 95%|█████████▌| 824/864 [20:56<00:52,  1.32s/it]
 95%|█████████▌| 825/864 [20:58<00:51,  1.32s/it]
 96%|█████████▌| 826/864 [20:59<00:49,  1.32s/it]
 96%|█████████▌| 827/864 [21:00<00:48,  1.32s/it]
 96%|█████████▌| 828/864 [21:02<00:47,  1.32s/it]
 96%|█████████▌| 829/864 [21:03<00:46,  1.32s/it]
 96%|█████████▌| 830/864 [21:04<00:44,  1.32s/it]
                                                 
{'loss': '3.566', 'grad_norm': '0.09263', 'learning_rate': '0.0012', 'epoch': '11.53'}
+
 96%|█████████▌| 830/864 [21:04<00:44,  1.32s/it]
 96%|█████████▌| 831/864 [21:05<00:43,  1.32s/it]
 96%|█████████▋| 832/864 [21:07<00:42,  1.32s/it]
 96%|█████████▋| 833/864 [21:08<00:40,  1.32s/it]
 97%|█████████▋| 834/864 [21:09<00:39,  1.31s/it]
 97%|█████████▋| 835/864 [21:11<00:38,  1.31s/it]
 97%|█████████▋| 836/864 [21:12<00:36,  1.32s/it]
 97%|█████████▋| 837/864 [21:13<00:35,  1.32s/it]
 97%|█████████▋| 838/864 [21:15<00:34,  1.31s/it]
 97%|█████████▋| 839/864 [21:16<00:32,  1.31s/it]
 97%|█████████▋| 840/864 [21:17<00:31,  1.32s/it]
                                                 
{'loss': '3.56', 'grad_norm': '0.1047', 'learning_rate': '0.0012', 'epoch': '11.67'}
+
 97%|█████████▋| 840/864 [21:17<00:31,  1.32s/it]
 97%|█████████▋| 841/864 [21:19<00:30,  1.32s/it]
 97%|█████████▋| 842/864 [21:20<00:28,  1.32s/it]
 98%|█████████▊| 843/864 [21:21<00:27,  1.32s/it]
 98%|█████████▊| 844/864 [21:23<00:26,  1.32s/it]
 98%|█████████▊| 845/864 [21:24<00:25,  1.32s/it]
 98%|█████████▊| 846/864 [21:25<00:23,  1.32s/it]
 98%|█████████▊| 847/864 [21:27<00:22,  1.32s/it]
 98%|█████████▊| 848/864 [21:28<00:21,  1.31s/it]
 98%|█████████▊| 849/864 [21:29<00:19,  1.31s/it]
 98%|█████████▊| 850/864 [21:30<00:18,  1.32s/it]
                                                 
{'loss': '3.555', 'grad_norm': '0.1148', 'learning_rate': '0.0012', 'epoch': '11.81'}
+
 98%|█████████▊| 850/864 [21:30<00:18,  1.32s/it]
 98%|█████████▊| 851/864 [21:32<00:17,  1.32s/it]
 99%|█████████▊| 852/864 [21:33<00:15,  1.32s/it]
 99%|█████████▊| 853/864 [21:34<00:14,  1.32s/it]
 99%|█████████▉| 854/864 [21:36<00:13,  1.32s/it]
 99%|█████████▉| 855/864 [21:37<00:11,  1.31s/it]
 99%|█████████▉| 856/864 [21:38<00:10,  1.31s/it]
 99%|█████████▉| 857/864 [21:40<00:09,  1.31s/it]
 99%|█████████▉| 858/864 [21:41<00:07,  1.32s/it]
 99%|█████████▉| 859/864 [21:42<00:06,  1.32s/it]
100%|█████████▉| 860/864 [21:44<00:05,  1.31s/it]
                                                 
{'loss': '3.554', 'grad_norm': '0.09398', 'learning_rate': '0.0012', 'epoch': '11.95'}
+
100%|█████████▉| 860/864 [21:44<00:05,  1.31s/it]
100%|█████████▉| 861/864 [21:45<00:03,  1.32s/it]
100%|█████████▉| 862/864 [21:46<00:02,  1.32s/it]
100%|█████████▉| 863/864 [21:48<00:01,  1.32s/it]
100%|██████████| 864/864 [21:48<00:00,  1.17s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:02, 15.04it/s]
+
 10%|█         | 4/39 [00:00<00:03,  9.20it/s]
+
 15%|█▌        | 6/39 [00:00<00:04,  8.18it/s]
+
 18%|█▊        | 7/39 [00:00<00:04,  7.93it/s]
+
 21%|██        | 8/39 [00:00<00:04,  7.73it/s]
+
 23%|██▎       | 9/39 [00:01<00:03,  7.59it/s]
+
 26%|██▌       | 10/39 [00:01<00:03,  7.48it/s]
+
 28%|██▊       | 11/39 [00:01<00:03,  7.42it/s]
+
 31%|███       | 12/39 [00:01<00:03,  7.36it/s]
+
 33%|███▎      | 13/39 [00:01<00:03,  7.32it/s]
+
 36%|███▌      | 14/39 [00:01<00:03,  7.30it/s]
+
 38%|███▊      | 15/39 [00:01<00:03,  7.28it/s]
+
 41%|████      | 16/39 [00:02<00:03,  7.27it/s]
+
 44%|████▎     | 17/39 [00:02<00:03,  7.27it/s]
+
 46%|████▌     | 18/39 [00:02<00:02,  7.26it/s]
+
 49%|████▊     | 19/39 [00:02<00:02,  7.26it/s]
+
 51%|█████▏    | 20/39 [00:02<00:02,  7.26it/s]
+
 54%|█████▍    | 21/39 [00:02<00:02,  7.26it/s]
+
 56%|█████▋    | 22/39 [00:02<00:02,  7.25it/s]
+
 59%|█████▉    | 23/39 [00:03<00:02,  7.25it/s]
+
 62%|██████▏   | 24/39 [00:03<00:02,  7.24it/s]
+
 64%|██████▍   | 25/39 [00:03<00:01,  7.23it/s]
+
 67%|██████▋   | 26/39 [00:03<00:01,  7.23it/s]
+
 69%|██████▉   | 27/39 [00:03<00:01,  7.24it/s]
+
 72%|███████▏  | 28/39 [00:03<00:01,  7.25it/s]
+
 74%|███████▍  | 29/39 [00:03<00:01,  7.25it/s]
+
 77%|███████▋  | 30/39 [00:04<00:01,  7.25it/s]
+
 79%|███████▉  | 31/39 [00:04<00:01,  7.25it/s]
+
 82%|████████▏ | 32/39 [00:04<00:00,  7.24it/s]
+
 85%|████████▍ | 33/39 [00:04<00:00,  7.24it/s]
+
 87%|████████▋ | 34/39 [00:04<00:00,  7.24it/s]
+
 90%|████████▉ | 35/39 [00:04<00:00,  7.25it/s]
+
 92%|█████████▏| 36/39 [00:04<00:00,  7.25it/s]
+
 95%|█████████▍| 37/39 [00:04<00:00,  7.25it/s]
+
 97%|█████████▋| 38/39 [00:05<00:00,  7.25it/s]
+
100%|██████████| 39/39 [00:05<00:00,  7.24it/s]
                                                 
+
                                               
{'eval_loss': '3.961', 'eval_runtime': '5.433', 'eval_samples_per_second': '449.3', 'eval_steps_per_second': '7.178', 'epoch': '12'}
+
100%|██████████| 864/864 [21:54<00:00,  1.17s/it]
+
100%|██████████| 39/39 [00:05<00:00,  7.24it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0012/checkpoint-864
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0012/checkpoint-864/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0012/checkpoint-864/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:01<00:00,  1.84s/it]
Writing model shards: 100%|██████████| 1/1 [00:01<00:00,  1.84s/it]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0012/checkpoint-864/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0012/checkpoint-864/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S4_lr0.0012/checkpoint-864/tokenizer_config.json
+[eval] epoch 12.0 step 864 loss 3.9606 (best 3.9606 @ ep 12.0, 0/10000 since best)
+
+
                                               [transformers] 
+
+Training completed. Do not forget to share your model on huggingface.co/models =)
+
+
+
                                                 
{'train_runtime': '1321', 'train_samples_per_second': '332.7', 'train_steps_per_second': '0.654', 'train_loss': '4.895', 'epoch': '12'}
+
100%|██████████| 864/864 [21:59<00:00,  1.17s/it]
100%|██████████| 864/864 [21:59<00:00,  1.53s/it]
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+wandb: 
+wandb: 🚀 View run t0_S4_lr0.0012 at: https://wandb.ai/slowrun-wankers/data-aug-pretraining/runs/sc3anwdm
+wandb: Find logs at: wandb/run-20260817_211038-sc3anwdm/logs
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+[rank0]:[W817 21:33:23.610792834 ProcessGroupNCCL.cpp:1575] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator())
diff --git a/runs/t0/t0_S5_lr0.0003/checkpoint-720/chat_template.jinja b/runs/t0/t0_S5_lr0.0003/checkpoint-720/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..699ff8df401fe4788525e9c1f9b86a99eadd6230
--- /dev/null
+++ b/runs/t0/t0_S5_lr0.0003/checkpoint-720/chat_template.jinja
@@ -0,0 +1,85 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0].role == 'system' %}
+        {{- messages[0].content + '\n\n' }}
+    {%- endif %}
+    {{- "# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0].content + '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
+{%- for message in messages[::-1] %}
+    {%- set index = (messages|length - 1) - loop.index0 %}
+    {%- if ns.multi_step_tool and message.role == "user" and not(message.content.startswith('<tool_response>') and message.content.endswith('</tool_response>')) %}
+        {%- set ns.multi_step_tool = false %}
+        {%- set ns.last_query_index = index %}
+    {%- endif %}
+{%- endfor %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {%- set content = message.content %}
+        {%- set reasoning_content = '' %}
+        {%- if message.reasoning_content is defined and message.reasoning_content is not none %}
+            {%- set reasoning_content = message.reasoning_content %}
+        {%- else %}
+            {%- if '</think>' in message.content %}
+                {%- set content = message.content.split('</think>')[-1].lstrip('\n') %}
+                {%- set reasoning_content = message.content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
+            {%- endif %}
+        {%- endif %}
+        {%- if loop.index0 > ns.last_query_index %}
+            {%- if loop.last or (not loop.last and reasoning_content) %}
+                {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content.strip('\n') + '\n</think>\n\n' + content.lstrip('\n') }}
+            {%- else %}
+                {{- '<|im_start|>' + message.role + '\n' + content }}
+            {%- endif %}
+        {%- else %}
+            {{- '<|im_start|>' + message.role + '\n' + content }}
+        {%- endif %}
+        {%- if message.tool_calls %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if (loop.first and content) or (not loop.first) %}
+                    {{- '\n' }}
+                {%- endif %}
+                {%- if tool_call.function %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {{- '<tool_call>\n{"name": "' }}
+                {{- tool_call.name }}
+                {{- '", "arguments": ' }}
+                {%- if tool_call.arguments is string %}
+                    {{- tool_call.arguments }}
+                {%- else %}
+                    {{- tool_call.arguments | tojson }}
+                {%- endif %}
+                {{- '}\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.first or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+    {%- if enable_thinking is defined and enable_thinking is false %}
+        {{- '<think>\n\n</think>\n\n' }}
+    {%- endif %}
+{%- endif %}
\ No newline at end of file
diff --git a/runs/t0/t0_S5_lr0.0003/checkpoint-720/config.json b/runs/t0/t0_S5_lr0.0003/checkpoint-720/config.json
new file mode 100644
index 0000000000000000000000000000000000000000..58ed141896af03a1496676b060879ee607fff4bb
--- /dev/null
+++ b/runs/t0/t0_S5_lr0.0003/checkpoint-720/config.json
@@ -0,0 +1,32 @@
+{
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": null,
+  "dtype": "float32",
+  "eos_token_id": 151645,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1536,
+  "initializer_range": 0.02,
+  "intermediate_size": 4096,
+  "max_position_embeddings": 2048,
+  "mlp_bias": false,
+  "model_type": "llama",
+  "num_attention_heads": 12,
+  "num_hidden_layers": 24,
+  "num_key_value_heads": 12,
+  "pad_token_id": 151645,
+  "pretraining_tp": 1,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 10000.0,
+    "rope_type": "default"
+  },
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": false,
+  "vocab_size": 151671
+}
diff --git a/runs/t0/t0_S5_lr0.0003/checkpoint-720/generation_config.json b/runs/t0/t0_S5_lr0.0003/checkpoint-720/generation_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..8c4242f32d40df256857f87264f646130f684e33
--- /dev/null
+++ b/runs/t0/t0_S5_lr0.0003/checkpoint-720/generation_config.json
@@ -0,0 +1,11 @@
+{
+  "_from_model_config": true,
+  "eos_token_id": [
+    151645
+  ],
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "transformers_version": "5.15.0",
+  "use_cache": true
+}
diff --git a/runs/t0/t0_S5_lr0.0003/checkpoint-720/tokenizer_config.json b/runs/t0/t0_S5_lr0.0003/checkpoint-720/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..839a3743177a3ed77e1ffbae67520e8500ca0532
--- /dev/null
+++ b/runs/t0/t0_S5_lr0.0003/checkpoint-720/tokenizer_config.json
@@ -0,0 +1,19 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|l2r_pred|>",
+    "<|r2l_pred|>"
+  ],
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 131072,
+  "pad_token": "<|im_end|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}
diff --git a/runs/t0/t0_S5_lr0.0003/checkpoint-720/trainer_state.json b/runs/t0/t0_S5_lr0.0003/checkpoint-720/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..55649e025eac9e0894f2f585e5556ca867593098
--- /dev/null
+++ b/runs/t0/t0_S5_lr0.0003/checkpoint-720/trainer_state.json
@@ -0,0 +1,625 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 10.0,
+  "eval_steps": 72,
+  "global_step": 720,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.013973799126637555,
+      "grad_norm": 3.2618095874786377,
+      "learning_rate": 0.0,
+      "loss": 12.214674949645996,
+      "step": 1
+    },
+    {
+      "epoch": 0.13973799126637554,
+      "grad_norm": 2.4582679271698,
+      "learning_rate": 2.6999999999999996e-05,
+      "loss": 11.440606011284721,
+      "step": 10
+    },
+    {
+      "epoch": 0.2794759825327511,
+      "grad_norm": 1.7859525680541992,
+      "learning_rate": 5.6999999999999996e-05,
+      "loss": 9.868122863769532,
+      "step": 20
+    },
+    {
+      "epoch": 0.4192139737991266,
+      "grad_norm": 1.0912766456604004,
+      "learning_rate": 8.699999999999999e-05,
+      "loss": 8.853494262695312,
+      "step": 30
+    },
+    {
+      "epoch": 0.5589519650655022,
+      "grad_norm": 0.8212459683418274,
+      "learning_rate": 0.000117,
+      "loss": 7.825148010253907,
+      "step": 40
+    },
+    {
+      "epoch": 0.6986899563318777,
+      "grad_norm": 1.0021402835845947,
+      "learning_rate": 0.000147,
+      "loss": 7.4507301330566404,
+      "step": 50
+    },
+    {
+      "epoch": 0.8384279475982532,
+      "grad_norm": 0.9833669066429138,
+      "learning_rate": 0.00017699999999999997,
+      "loss": 7.263099670410156,
+      "step": 60
+    },
+    {
+      "epoch": 0.9781659388646288,
+      "grad_norm": 1.023337483406067,
+      "learning_rate": 0.00020699999999999996,
+      "loss": 7.0315185546875,
+      "step": 70
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 6.922646999359131,
+      "eval_runtime": 32.1357,
+      "eval_samples_per_second": 75.959,
+      "eval_steps_per_second": 2.396,
+      "step": 72
+    },
+    {
+      "epoch": 1.1117903930131003,
+      "grad_norm": 1.203994631767273,
+      "learning_rate": 0.000237,
+      "loss": 6.820037841796875,
+      "step": 80
+    },
+    {
+      "epoch": 1.251528384279476,
+      "grad_norm": 1.0269978046417236,
+      "learning_rate": 0.000267,
+      "loss": 6.610933685302735,
+      "step": 90
+    },
+    {
+      "epoch": 1.3912663755458516,
+      "grad_norm": 0.7635436058044434,
+      "learning_rate": 0.00029699999999999996,
+      "loss": 6.47060546875,
+      "step": 100
+    },
+    {
+      "epoch": 1.531004366812227,
+      "grad_norm": 0.9571962952613831,
+      "learning_rate": 0.0003,
+      "loss": 6.356604766845703,
+      "step": 110
+    },
+    {
+      "epoch": 1.6707423580786025,
+      "grad_norm": 0.47234103083610535,
+      "learning_rate": 0.0003,
+      "loss": 6.199187469482422,
+      "step": 120
+    },
+    {
+      "epoch": 1.8104803493449781,
+      "grad_norm": 0.6224037408828735,
+      "learning_rate": 0.0003,
+      "loss": 6.105295562744141,
+      "step": 130
+    },
+    {
+      "epoch": 1.9502183406113538,
+      "grad_norm": 0.7163454294204712,
+      "learning_rate": 0.0003,
+      "loss": 6.005741119384766,
+      "step": 140
+    },
+    {
+      "epoch": 2.0,
+      "eval_loss": 5.9517598152160645,
+      "eval_runtime": 7.9148,
+      "eval_samples_per_second": 308.408,
+      "eval_steps_per_second": 9.729,
+      "step": 144
+    },
+    {
+      "epoch": 2.0838427947598253,
+      "grad_norm": 0.5856307744979858,
+      "learning_rate": 0.0003,
+      "loss": 5.904288482666016,
+      "step": 150
+    },
+    {
+      "epoch": 2.2235807860262007,
+      "grad_norm": 0.6896644234657288,
+      "learning_rate": 0.0003,
+      "loss": 5.783623123168946,
+      "step": 160
+    },
+    {
+      "epoch": 2.3633187772925766,
+      "grad_norm": 1.0071693658828735,
+      "learning_rate": 0.0003,
+      "loss": 5.70606575012207,
+      "step": 170
+    },
+    {
+      "epoch": 2.503056768558952,
+      "grad_norm": 0.8469743728637695,
+      "learning_rate": 0.0003,
+      "loss": 5.638371276855469,
+      "step": 180
+    },
+    {
+      "epoch": 2.6427947598253274,
+      "grad_norm": 0.4349379241466522,
+      "learning_rate": 0.0003,
+      "loss": 5.536750793457031,
+      "step": 190
+    },
+    {
+      "epoch": 2.7825327510917033,
+      "grad_norm": 0.838222324848175,
+      "learning_rate": 0.0003,
+      "loss": 5.431597518920898,
+      "step": 200
+    },
+    {
+      "epoch": 2.9222707423580787,
+      "grad_norm": 0.76561439037323,
+      "learning_rate": 0.0003,
+      "loss": 5.4217266082763675,
+      "step": 210
+    },
+    {
+      "epoch": 3.0,
+      "eval_loss": 5.362564563751221,
+      "eval_runtime": 7.9146,
+      "eval_samples_per_second": 308.419,
+      "eval_steps_per_second": 9.729,
+      "step": 216
+    },
+    {
+      "epoch": 3.05589519650655,
+      "grad_norm": 0.4211660921573639,
+      "learning_rate": 0.0003,
+      "loss": 5.302315521240234,
+      "step": 220
+    },
+    {
+      "epoch": 3.1956331877729256,
+      "grad_norm": 0.32755231857299805,
+      "learning_rate": 0.0003,
+      "loss": 5.182415008544922,
+      "step": 230
+    },
+    {
+      "epoch": 3.3353711790393015,
+      "grad_norm": 1.0064932107925415,
+      "learning_rate": 0.0003,
+      "loss": 5.131160736083984,
+      "step": 240
+    },
+    {
+      "epoch": 3.475109170305677,
+      "grad_norm": 0.39928022027015686,
+      "learning_rate": 0.0003,
+      "loss": 5.130668258666992,
+      "step": 250
+    },
+    {
+      "epoch": 3.6148471615720523,
+      "grad_norm": 0.24958354234695435,
+      "learning_rate": 0.0003,
+      "loss": 5.007334136962891,
+      "step": 260
+    },
+    {
+      "epoch": 3.7545851528384278,
+      "grad_norm": 0.624622106552124,
+      "learning_rate": 0.0003,
+      "loss": 4.949510192871093,
+      "step": 270
+    },
+    {
+      "epoch": 3.8943231441048036,
+      "grad_norm": 1.041094422340393,
+      "learning_rate": 0.0003,
+      "loss": 4.952127075195312,
+      "step": 280
+    },
+    {
+      "epoch": 4.0,
+      "eval_loss": 5.030168056488037,
+      "eval_runtime": 7.9196,
+      "eval_samples_per_second": 308.221,
+      "eval_steps_per_second": 9.723,
+      "step": 288
+    },
+    {
+      "epoch": 4.0279475982532755,
+      "grad_norm": 0.591354250907898,
+      "learning_rate": 0.0003,
+      "loss": 4.971962356567383,
+      "step": 290
+    },
+    {
+      "epoch": 4.1676855895196505,
+      "grad_norm": 0.23682217299938202,
+      "learning_rate": 0.0003,
+      "loss": 4.798876953125,
+      "step": 300
+    },
+    {
+      "epoch": 4.307423580786026,
+      "grad_norm": 0.2569108009338379,
+      "learning_rate": 0.0003,
+      "loss": 4.705326843261719,
+      "step": 310
+    },
+    {
+      "epoch": 4.447161572052401,
+      "grad_norm": 0.46430492401123047,
+      "learning_rate": 0.0003,
+      "loss": 4.6480354309082035,
+      "step": 320
+    },
+    {
+      "epoch": 4.586899563318777,
+      "grad_norm": 0.5473710894584656,
+      "learning_rate": 0.0003,
+      "loss": 4.645205688476563,
+      "step": 330
+    },
+    {
+      "epoch": 4.726637554585153,
+      "grad_norm": 0.36366525292396545,
+      "learning_rate": 0.0003,
+      "loss": 4.587823486328125,
+      "step": 340
+    },
+    {
+      "epoch": 4.866375545851528,
+      "grad_norm": 0.39248207211494446,
+      "learning_rate": 0.0003,
+      "loss": 4.511217498779297,
+      "step": 350
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 0.4145806133747101,
+      "learning_rate": 0.0003,
+      "loss": 4.47677001953125,
+      "step": 360
+    },
+    {
+      "epoch": 5.0,
+      "eval_loss": 4.589375972747803,
+      "eval_runtime": 7.9133,
+      "eval_samples_per_second": 308.469,
+      "eval_steps_per_second": 9.73,
+      "step": 360
+    },
+    {
+      "epoch": 5.139737991266376,
+      "grad_norm": 0.5570223927497864,
+      "learning_rate": 0.0003,
+      "loss": 4.355342864990234,
+      "step": 370
+    },
+    {
+      "epoch": 5.279475982532751,
+      "grad_norm": 0.6994290351867676,
+      "learning_rate": 0.0003,
+      "loss": 4.36234130859375,
+      "step": 380
+    },
+    {
+      "epoch": 5.419213973799127,
+      "grad_norm": 0.5207724571228027,
+      "learning_rate": 0.0003,
+      "loss": 4.337489318847656,
+      "step": 390
+    },
+    {
+      "epoch": 5.558951965065502,
+      "grad_norm": 0.23361016809940338,
+      "learning_rate": 0.0003,
+      "loss": 4.260697937011718,
+      "step": 400
+    },
+    {
+      "epoch": 5.698689956331878,
+      "grad_norm": 0.267966091632843,
+      "learning_rate": 0.0003,
+      "loss": 4.221217346191406,
+      "step": 410
+    },
+    {
+      "epoch": 5.8384279475982535,
+      "grad_norm": 0.3150366544723511,
+      "learning_rate": 0.0003,
+      "loss": 4.19616470336914,
+      "step": 420
+    },
+    {
+      "epoch": 5.978165938864628,
+      "grad_norm": 0.35929128527641296,
+      "learning_rate": 0.0003,
+      "loss": 4.170941925048828,
+      "step": 430
+    },
+    {
+      "epoch": 6.0,
+      "eval_loss": 4.342624187469482,
+      "eval_runtime": 7.9157,
+      "eval_samples_per_second": 308.376,
+      "eval_steps_per_second": 9.728,
+      "step": 432
+    },
+    {
+      "epoch": 6.1117903930131,
+      "grad_norm": 0.8794248700141907,
+      "learning_rate": 0.0003,
+      "loss": 4.087378692626953,
+      "step": 440
+    },
+    {
+      "epoch": 6.251528384279476,
+      "grad_norm": 0.7270575165748596,
+      "learning_rate": 0.0003,
+      "loss": 4.194872283935547,
+      "step": 450
+    },
+    {
+      "epoch": 6.391266375545851,
+      "grad_norm": 1.0183628797531128,
+      "learning_rate": 0.0003,
+      "loss": 4.237209320068359,
+      "step": 460
+    },
+    {
+      "epoch": 6.531004366812227,
+      "grad_norm": 0.31741416454315186,
+      "learning_rate": 0.0003,
+      "loss": 4.177013778686524,
+      "step": 470
+    },
+    {
+      "epoch": 6.670742358078603,
+      "grad_norm": 0.1903829276561737,
+      "learning_rate": 0.0003,
+      "loss": 4.066764831542969,
+      "step": 480
+    },
+    {
+      "epoch": 6.810480349344978,
+      "grad_norm": 0.17531758546829224,
+      "learning_rate": 0.0003,
+      "loss": 4.023290634155273,
+      "step": 490
+    },
+    {
+      "epoch": 6.950218340611354,
+      "grad_norm": 0.2047526091337204,
+      "learning_rate": 0.0003,
+      "loss": 3.9738792419433593,
+      "step": 500
+    },
+    {
+      "epoch": 7.0,
+      "eval_loss": 4.2179388999938965,
+      "eval_runtime": 7.9122,
+      "eval_samples_per_second": 308.509,
+      "eval_steps_per_second": 9.732,
+      "step": 504
+    },
+    {
+      "epoch": 7.083842794759826,
+      "grad_norm": 0.22795313596725464,
+      "learning_rate": 0.0003,
+      "loss": 3.88709716796875,
+      "step": 510
+    },
+    {
+      "epoch": 7.223580786026201,
+      "grad_norm": 0.43237408995628357,
+      "learning_rate": 0.0003,
+      "loss": 3.8286155700683593,
+      "step": 520
+    },
+    {
+      "epoch": 7.3633187772925766,
+      "grad_norm": 0.2796283960342407,
+      "learning_rate": 0.0003,
+      "loss": 3.820693588256836,
+      "step": 530
+    },
+    {
+      "epoch": 7.503056768558952,
+      "grad_norm": 0.2597295939922333,
+      "learning_rate": 0.0003,
+      "loss": 3.8072124481201173,
+      "step": 540
+    },
+    {
+      "epoch": 7.642794759825327,
+      "grad_norm": 0.2075442373752594,
+      "learning_rate": 0.0003,
+      "loss": 3.7986984252929688,
+      "step": 550
+    },
+    {
+      "epoch": 7.782532751091703,
+      "grad_norm": 0.2210642695426941,
+      "learning_rate": 0.0003,
+      "loss": 3.789369583129883,
+      "step": 560
+    },
+    {
+      "epoch": 7.922270742358078,
+      "grad_norm": 0.2791687548160553,
+      "learning_rate": 0.0003,
+      "loss": 3.78194580078125,
+      "step": 570
+    },
+    {
+      "epoch": 8.0,
+      "eval_loss": 4.121953010559082,
+      "eval_runtime": 7.9121,
+      "eval_samples_per_second": 308.516,
+      "eval_steps_per_second": 9.732,
+      "step": 576
+    },
+    {
+      "epoch": 8.055895196506551,
+      "grad_norm": 0.6974738836288452,
+      "learning_rate": 0.0003,
+      "loss": 3.7090476989746093,
+      "step": 580
+    },
+    {
+      "epoch": 8.195633187772925,
+      "grad_norm": 0.8548513650894165,
+      "learning_rate": 0.0003,
+      "loss": 3.745708465576172,
+      "step": 590
+    },
+    {
+      "epoch": 8.335371179039301,
+      "grad_norm": 0.8406635522842407,
+      "learning_rate": 0.0003,
+      "loss": 3.828490447998047,
+      "step": 600
+    },
+    {
+      "epoch": 8.475109170305677,
+      "grad_norm": 0.43541979789733887,
+      "learning_rate": 0.0003,
+      "loss": 3.771062469482422,
+      "step": 610
+    },
+    {
+      "epoch": 8.614847161572053,
+      "grad_norm": 0.22296181321144104,
+      "learning_rate": 0.0003,
+      "loss": 3.6843185424804688,
+      "step": 620
+    },
+    {
+      "epoch": 8.754585152838429,
+      "grad_norm": 0.2070133239030838,
+      "learning_rate": 0.0003,
+      "loss": 3.6420669555664062,
+      "step": 630
+    },
+    {
+      "epoch": 8.894323144104803,
+      "grad_norm": 0.20196694135665894,
+      "learning_rate": 0.0003,
+      "loss": 3.6326332092285156,
+      "step": 640
+    },
+    {
+      "epoch": 9.0,
+      "eval_loss": 4.078823566436768,
+      "eval_runtime": 7.9041,
+      "eval_samples_per_second": 308.828,
+      "eval_steps_per_second": 9.742,
+      "step": 648
+    },
+    {
+      "epoch": 9.027947598253276,
+      "grad_norm": 0.3818998336791992,
+      "learning_rate": 0.0003,
+      "loss": 3.5768112182617187,
+      "step": 650
+    },
+    {
+      "epoch": 9.167685589519651,
+      "grad_norm": 0.6515381932258606,
+      "learning_rate": 0.0003,
+      "loss": 3.441845703125,
+      "step": 660
+    },
+    {
+      "epoch": 9.307423580786025,
+      "grad_norm": 0.41331514716148376,
+      "learning_rate": 0.0003,
+      "loss": 3.4659652709960938,
+      "step": 670
+    },
+    {
+      "epoch": 9.447161572052401,
+      "grad_norm": 0.3088191747665405,
+      "learning_rate": 0.0003,
+      "loss": 3.4546302795410155,
+      "step": 680
+    },
+    {
+      "epoch": 9.586899563318777,
+      "grad_norm": 0.2400057464838028,
+      "learning_rate": 0.0003,
+      "loss": 3.431840515136719,
+      "step": 690
+    },
+    {
+      "epoch": 9.726637554585153,
+      "grad_norm": 0.26047268509864807,
+      "learning_rate": 0.0003,
+      "loss": 3.4436744689941405,
+      "step": 700
+    },
+    {
+      "epoch": 9.866375545851529,
+      "grad_norm": 0.24156703054904938,
+      "learning_rate": 0.0003,
+      "loss": 3.435311508178711,
+      "step": 710
+    },
+    {
+      "epoch": 10.0,
+      "grad_norm": 0.290781170129776,
+      "learning_rate": 0.0003,
+      "loss": 3.428261566162109,
+      "step": 720
+    },
+    {
+      "epoch": 10.0,
+      "eval_loss": 4.061178207397461,
+      "eval_runtime": 7.9111,
+      "eval_samples_per_second": 308.552,
+      "eval_steps_per_second": 9.733,
+      "step": 720
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 864,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 12,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": false,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 3.059564909802029e+18,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/runs/t0/t0_S5_lr0.0003/train.log b/runs/t0/t0_S5_lr0.0003/train.log
new file mode 100644
index 0000000000000000000000000000000000000000..ff6504b1f1bff906173db49144a33561b068ec22
--- /dev/null
+++ b/runs/t0/t0_S5_lr0.0003/train.log
@@ -0,0 +1,2033 @@
+W0817 21:33:27.374000 2300522 torch/distributed/run.py:851] 
+W0817 21:33:27.374000 2300522 torch/distributed/run.py:851] *****************************************
+W0817 21:33:27.374000 2300522 torch/distributed/run.py:851] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+W0817 21:33:27.374000 2300522 torch/distributed/run.py:851] *****************************************
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+total params: 912519168
+[dataset] loaded packed cache: 36614 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+total params: 912519168
+[dataset] loaded packed cache: 36614 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+total params: 912519168
+total params: 912519168
+[dataset] loaded packed cache: 36614 train / 2441 eval blocks
+[dataset] loaded packed cache: 36614 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[transformers] PyTorch: setting up devices
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+total params: 912519168
+[dataset] loaded packed cache: 36614 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+total params: 912519168
+[dataset] loaded packed cache: 36614 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+total params: 912519168
+[dataset] loaded packed cache: 36614 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+total params: 912519168
+[dataset] loaded packed cache: 36614 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+[transformers] ***** Running training *****
+[transformers]   Num examples = 36,614
+[transformers]   Num Epochs = 12
+[transformers]   Num update steps per epoch = 72
+[transformers]   Instantaneous batch size per device = 4
+[transformers]   Total train batch size (w. parallel, distributed & accumulation) = 512
+[transformers]   Gradient Accumulation steps = 16
+[transformers]   Total optimization steps = 864
+[transformers]   Number of trainable parameters = 912,519,168
+wandb: WARNING Changes to your `wandb` environment variables will be ignored because your `wandb` session has already started. For more information on how to modify your settings with `wandb.init()` arguments, please refer to https://wandb.me/wandb-init.
+wandb: setting up run 7qnpc6jf
+wandb: Tracking run with wandb version 0.28.2
+wandb: Run data is saved locally in /home/ubuntu/workspace/data-augmentations-for-pretraining/wandb/run-20260817_213355-7qnpc6jf
+wandb: Run `wandb offline` to turn off syncing.
+wandb: Syncing run t0_S5_lr0.0003
+wandb: ⭐️ View project at https://wandb.ai/slowrun-wankers/data-aug-pretraining
+wandb: 🚀 View run at https://wandb.ai/slowrun-wankers/data-aug-pretraining/runs/7qnpc6jf
+
  0%|          | 0/864 [00:00<?, ?it/s]wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+[rank1]:W0817 21:34:18.242000 2300628 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+[rank7]:W0817 21:34:18.579000 2300634 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+[rank4]:W0817 21:34:18.586000 2300631 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+[rank2]:W0817 21:34:18.843000 2300629 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+[rank5]:W0817 21:34:18.873000 2300632 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+[rank3]:W0817 21:34:19.462000 2300630 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+[rank6]:W0817 21:34:19.571000 2300633 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+[rank0]:W0817 21:34:19.784000 2300627 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+
  0%|          | 1/864 [00:53<12:51:36, 53.65s/it]
                                                  
{'loss': '12.21', 'grad_norm': '3.262', 'learning_rate': '0', 'epoch': '0.01397'}
+
  0%|          | 1/864 [00:53<12:51:36, 53.65s/it]
  0%|          | 2/864 [00:56<5:38:18, 23.55s/it] 
  0%|          | 3/864 [00:58<3:18:41, 13.85s/it]
  0%|          | 4/864 [01:00<2:13:08,  9.29s/it]
  1%|          | 5/864 [01:03<1:36:56,  6.77s/it]
  1%|          | 6/864 [01:05<1:15:08,  5.25s/it]
  1%|          | 7/864 [01:07<1:01:17,  4.29s/it]
  1%|          | 8/864 [01:09<52:12,  3.66s/it]  
  1%|          | 9/864 [01:12<46:06,  3.24s/it]
  1%|          | 10/864 [01:14<41:57,  2.95s/it]
                                                
{'loss': '11.44', 'grad_norm': '2.458', 'learning_rate': '2.7e-05', 'epoch': '0.1397'}
+
  1%|          | 10/864 [01:14<41:57,  2.95s/it]
  1%|▏         | 11/864 [01:16<39:06,  2.75s/it]
  1%|▏         | 12/864 [01:19<37:08,  2.62s/it]
  2%|▏         | 13/864 [01:21<35:44,  2.52s/it]
  2%|▏         | 14/864 [01:23<34:45,  2.45s/it]
  2%|▏         | 15/864 [01:26<34:03,  2.41s/it]
  2%|▏         | 16/864 [01:28<33:33,  2.37s/it]
  2%|▏         | 17/864 [01:30<33:12,  2.35s/it]
  2%|▏         | 18/864 [01:32<32:56,  2.34s/it]
  2%|▏         | 19/864 [01:35<32:45,  2.33s/it]
  2%|▏         | 20/864 [01:37<32:36,  2.32s/it]
                                                
{'loss': '9.868', 'grad_norm': '1.786', 'learning_rate': '5.7e-05', 'epoch': '0.2795'}
+
  2%|▏         | 20/864 [01:37<32:36,  2.32s/it]
  2%|▏         | 21/864 [01:39<32:29,  2.31s/it]
  3%|▎         | 22/864 [01:42<32:23,  2.31s/it]
  3%|▎         | 23/864 [01:44<32:19,  2.31s/it]
  3%|▎         | 24/864 [01:46<32:14,  2.30s/it]
  3%|▎         | 25/864 [01:49<32:11,  2.30s/it]
  3%|▎         | 26/864 [01:51<32:07,  2.30s/it]
  3%|▎         | 27/864 [01:53<32:05,  2.30s/it]
  3%|▎         | 28/864 [01:55<32:02,  2.30s/it]
  3%|▎         | 29/864 [01:58<31:59,  2.30s/it]
  3%|▎         | 30/864 [02:00<31:57,  2.30s/it]
                                                
{'loss': '8.853', 'grad_norm': '1.091', 'learning_rate': '8.7e-05', 'epoch': '0.4192'}
+
  3%|▎         | 30/864 [02:00<31:57,  2.30s/it]
  4%|▎         | 31/864 [02:02<31:54,  2.30s/it]
  4%|▎         | 32/864 [02:05<31:53,  2.30s/it]
  4%|▍         | 33/864 [02:07<31:51,  2.30s/it]
  4%|▍         | 34/864 [02:09<31:49,  2.30s/it]
  4%|▍         | 35/864 [02:12<31:47,  2.30s/it]
  4%|▍         | 36/864 [02:14<31:45,  2.30s/it]
  4%|▍         | 37/864 [02:16<31:43,  2.30s/it]
  4%|▍         | 38/864 [02:18<31:40,  2.30s/it]
  5%|▍         | 39/864 [02:21<31:39,  2.30s/it]
  5%|▍         | 40/864 [02:23<31:38,  2.30s/it]
                                                
{'loss': '7.825', 'grad_norm': '0.8212', 'learning_rate': '0.000117', 'epoch': '0.559'}
+
  5%|▍         | 40/864 [02:23<31:38,  2.30s/it]
  5%|▍         | 41/864 [02:25<31:36,  2.30s/it]
  5%|▍         | 42/864 [02:28<31:34,  2.30s/it]
  5%|▍         | 43/864 [02:30<31:32,  2.31s/it]
  5%|▌         | 44/864 [02:32<31:30,  2.31s/it]
  5%|▌         | 45/864 [02:35<31:29,  2.31s/it]
  5%|▌         | 46/864 [02:37<31:27,  2.31s/it]
  5%|▌         | 47/864 [02:39<31:24,  2.31s/it]
  6%|▌         | 48/864 [02:42<31:23,  2.31s/it]
  6%|▌         | 49/864 [02:44<31:21,  2.31s/it]
  6%|▌         | 50/864 [02:46<31:20,  2.31s/it]
                                                
{'loss': '7.451', 'grad_norm': '1.002', 'learning_rate': '0.000147', 'epoch': '0.6987'}
+
  6%|▌         | 50/864 [02:46<31:20,  2.31s/it]
  6%|▌         | 51/864 [02:48<31:18,  2.31s/it]
  6%|▌         | 52/864 [02:51<31:17,  2.31s/it]
  6%|▌         | 53/864 [02:53<31:15,  2.31s/it]
  6%|▋         | 54/864 [02:55<31:14,  2.31s/it]
  6%|▋         | 55/864 [02:58<31:12,  2.31s/it]
  6%|▋         | 56/864 [03:00<31:09,  2.31s/it]
  7%|▋         | 57/864 [03:02<31:08,  2.32s/it]
  7%|▋         | 58/864 [03:05<31:06,  2.32s/it]
  7%|▋         | 59/864 [03:07<31:05,  2.32s/it]
  7%|▋         | 60/864 [03:09<31:03,  2.32s/it]
                                                
{'loss': '7.263', 'grad_norm': '0.9834', 'learning_rate': '0.000177', 'epoch': '0.8384'}
+
  7%|▋         | 60/864 [03:09<31:03,  2.32s/it]
  7%|▋         | 61/864 [03:12<31:01,  2.32s/it]
  7%|▋         | 62/864 [03:14<30:59,  2.32s/it]
  7%|▋         | 63/864 [03:16<30:57,  2.32s/it]
  7%|▋         | 64/864 [03:19<30:54,  2.32s/it]
  8%|▊         | 65/864 [03:21<30:53,  2.32s/it]
  8%|▊         | 66/864 [03:23<30:51,  2.32s/it]
  8%|▊         | 67/864 [03:26<30:49,  2.32s/it]
  8%|▊         | 68/864 [03:28<30:46,  2.32s/it]
  8%|▊         | 69/864 [03:30<30:45,  2.32s/it]
  8%|▊         | 70/864 [03:33<30:43,  2.32s/it]
                                                
{'loss': '7.032', 'grad_norm': '1.023', 'learning_rate': '0.000207', 'epoch': '0.9782'}
+
  8%|▊         | 70/864 [03:33<30:43,  2.32s/it]
  8%|▊         | 71/864 [03:35<30:42,  2.32s/it]
  8%|▊         | 72/864 [03:36<26:42,  2.02s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 4
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+
+
  0%|          | 0/77 [00:00<?, ?it/s]
+
  4%|▍         | 3/77 [00:00<00:03, 22.08it/s]
+
  8%|▊         | 6/77 [00:00<00:05, 12.70it/s]
+
 10%|█         | 8/77 [00:00<00:06, 11.49it/s]
+
 13%|█▎        | 10/77 [00:00<00:06, 10.85it/s]
+
 16%|█▌        | 12/77 [00:01<00:06, 10.47it/s]
+
 18%|█▊        | 14/77 [00:01<00:06, 10.24it/s]
+
 21%|██        | 16/77 [00:01<00:06, 10.09it/s]
+
 23%|██▎       | 18/77 [00:01<00:05, 10.00it/s]
+
 26%|██▌       | 20/77 [00:01<00:05,  9.94it/s]
+
 29%|██▊       | 22/77 [00:02<00:05,  9.89it/s]
+
 30%|██▉       | 23/77 [00:02<00:05,  9.87it/s]
+
 31%|███       | 24/77 [00:02<00:05,  9.85it/s]
+
 32%|███▏      | 25/77 [00:02<00:05,  9.84it/s]
+
 34%|███▍      | 26/77 [00:02<00:05,  9.82it/s]
+
 35%|███▌      | 27/77 [00:02<00:05,  9.81it/s]
+
 36%|███▋      | 28/77 [00:02<00:04,  9.80it/s]
+
 38%|███▊      | 29/77 [00:02<00:04,  9.80it/s]
+
 39%|███▉      | 30/77 [00:02<00:04,  9.79it/s]
+
 40%|████      | 31/77 [00:02<00:04,  9.79it/s]
+
 42%|████▏     | 32/77 [00:03<00:04,  9.79it/s]
+
 43%|████▎     | 33/77 [00:03<00:04,  9.79it/s]
+
 44%|████▍     | 34/77 [00:03<00:04,  9.78it/s]
+
 45%|████▌     | 35/77 [00:03<00:04,  9.78it/s]
+
 47%|████▋     | 36/77 [00:03<00:04,  9.78it/s]
+
 48%|████▊     | 37/77 [00:03<00:04,  9.77it/s]
+
 49%|████▉     | 38/77 [00:03<00:03,  9.78it/s]
+
 51%|█████     | 39/77 [00:03<00:03,  9.78it/s]
+
 52%|█████▏    | 40/77 [00:03<00:03,  9.77it/s]
+
 53%|█████▎    | 41/77 [00:04<00:03,  9.77it/s]
+
 55%|█████▍    | 42/77 [00:04<00:03,  9.77it/s]
+
 56%|█████▌    | 43/77 [00:04<00:03,  9.79it/s]
+
 57%|█████▋    | 44/77 [00:04<00:03,  9.78it/s]
+
 58%|█████▊    | 45/77 [00:04<00:03,  9.78it/s]
+
 60%|█████▉    | 46/77 [00:04<00:03,  9.78it/s]
+
 61%|██████    | 47/77 [00:04<00:03,  9.78it/s]
+
 62%|██████▏   | 48/77 [00:04<00:02,  9.79it/s]
+
 64%|██████▎   | 49/77 [00:04<00:02,  9.79it/s]
+
 65%|██████▍   | 50/77 [00:04<00:02,  9.78it/s]
+
 66%|██████▌   | 51/77 [00:05<00:02,  9.79it/s]
+
 68%|██████▊   | 52/77 [00:05<00:02,  9.78it/s]
+
 69%|██████▉   | 53/77 [00:05<00:02,  9.79it/s]
+
 70%|███████   | 54/77 [00:05<00:02,  9.79it/s]
+
 71%|███████▏  | 55/77 [00:05<00:02,  9.78it/s]
+
 73%|███████▎  | 56/77 [00:05<00:02,  9.78it/s]
+
 74%|███████▍  | 57/77 [00:05<00:02,  9.79it/s]
+
 75%|███████▌  | 58/77 [00:05<00:01,  9.78it/s]
+
 77%|███████▋  | 59/77 [00:05<00:01,  9.79it/s]
+
 78%|███████▊  | 60/77 [00:05<00:01,  9.80it/s]
+
 79%|███████▉  | 61/77 [00:06<00:01,  9.81it/s]
+
 81%|████████  | 62/77 [00:06<00:01,  9.80it/s]
+
 82%|████████▏ | 63/77 [00:06<00:01,  9.80it/s]
+
 83%|████████▎ | 64/77 [00:06<00:01,  9.81it/s]
+
 84%|████████▍ | 65/77 [00:06<00:01,  9.80it/s]
+
 86%|████████▌ | 66/77 [00:06<00:01,  9.79it/s]
+
 87%|████████▋ | 67/77 [00:06<00:01,  9.80it/s]
+
 88%|████████▊ | 68/77 [00:06<00:00,  9.80it/s]
+
 90%|████████▉ | 69/77 [00:06<00:00,  9.80it/s]
+
 91%|█████████ | 70/77 [00:06<00:00,  9.80it/s]
+
 92%|█████████▏| 71/77 [00:07<00:00,  9.78it/s]
+
 94%|█████████▎| 72/77 [00:07<00:00,  9.80it/s]
+
 95%|█████████▍| 73/77 [00:07<00:00,  9.80it/s]
+
 96%|█████████▌| 74/77 [00:07<00:00,  9.80it/s]
+
 97%|█████████▋| 75/77 [00:07<00:00,  9.80it/s]
+
 99%|█████████▊| 76/77 [00:07<00:00,  9.80it/s]
+
100%|██████████| 77/77 [00:07<00:00,  9.80it/s]
                                                
+
                                               
{'eval_loss': '6.923', 'eval_runtime': '32.14', 'eval_samples_per_second': '75.96', 'eval_steps_per_second': '2.396', 'epoch': '1'}
+
  8%|▊         | 72/864 [04:08<26:42,  2.02s/it]
+
100%|██████████| 77/77 [00:07<00:00,  9.80it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0003/checkpoint-72
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0003/checkpoint-72/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0003/checkpoint-72/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:03<00:00,  3.55s/it]
Writing model shards: 100%|██████████| 1/1 [00:03<00:00,  3.56s/it]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0003/checkpoint-72/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0003/checkpoint-72/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0003/checkpoint-72/tokenizer_config.json
+[eval] epoch 1.0 step 72 loss 6.9226 (best 6.9226 @ ep 1.0, 0/10000 since best)
+
+
                                               
  8%|▊         | 73/864 [04:21<3:16:06, 14.88s/it]
  9%|▊         | 74/864 [04:23<2:26:13, 11.11s/it]
  9%|▊         | 75/864 [04:26<1:51:22,  8.47s/it]
  9%|▉         | 76/864 [04:28<1:26:59,  6.62s/it]
  9%|▉         | 77/864 [04:30<1:09:56,  5.33s/it]
  9%|▉         | 78/864 [04:33<58:01,  4.43s/it]  
  9%|▉         | 79/864 [04:35<49:41,  3.80s/it]
  9%|▉         | 80/864 [04:37<43:50,  3.36s/it]
                                                
{'loss': '6.82', 'grad_norm': '1.204', 'learning_rate': '0.000237', 'epoch': '1.112'}
+
  9%|▉         | 80/864 [04:37<43:50,  3.36s/it]
  9%|▉         | 81/864 [04:40<39:44,  3.05s/it]
  9%|▉         | 82/864 [04:42<36:51,  2.83s/it]
 10%|▉         | 83/864 [04:44<34:49,  2.68s/it]
 10%|▉         | 84/864 [04:47<33:24,  2.57s/it]
 10%|▉         | 85/864 [04:49<32:23,  2.49s/it]
 10%|▉         | 86/864 [04:51<31:40,  2.44s/it]
 10%|█         | 87/864 [04:54<31:09,  2.41s/it]
 10%|█         | 88/864 [04:56<30:47,  2.38s/it]
 10%|█         | 89/864 [04:58<30:31,  2.36s/it]
 10%|█         | 90/864 [05:00<30:19,  2.35s/it]
                                                
{'loss': '6.611', 'grad_norm': '1.027', 'learning_rate': '0.000267', 'epoch': '1.252'}
+
 10%|█         | 90/864 [05:01<30:19,  2.35s/it]
 11%|█         | 91/864 [05:03<30:11,  2.34s/it]
 11%|█         | 92/864 [05:05<30:03,  2.34s/it]
 11%|█         | 93/864 [05:07<29:58,  2.33s/it]
 11%|█         | 94/864 [05:10<29:54,  2.33s/it]
 11%|█         | 95/864 [05:12<29:50,  2.33s/it]
 11%|█         | 96/864 [05:14<29:46,  2.33s/it]
 11%|█         | 97/864 [05:17<29:43,  2.32s/it]
 11%|█▏        | 98/864 [05:19<29:40,  2.32s/it]
 11%|█▏        | 99/864 [05:21<29:38,  2.32s/it]
 12%|█▏        | 100/864 [05:24<29:35,  2.32s/it]
                                                 
{'loss': '6.471', 'grad_norm': '0.7635', 'learning_rate': '0.000297', 'epoch': '1.391'}
+
 12%|█▏        | 100/864 [05:24<29:35,  2.32s/it]
 12%|█▏        | 101/864 [05:26<29:33,  2.32s/it]
 12%|█▏        | 102/864 [05:28<29:31,  2.32s/it]
 12%|█▏        | 103/864 [05:31<29:28,  2.32s/it]
 12%|█▏        | 104/864 [05:33<29:26,  2.32s/it]
 12%|█▏        | 105/864 [05:35<29:24,  2.32s/it]
 12%|█▏        | 106/864 [05:38<29:21,  2.32s/it]
 12%|█▏        | 107/864 [05:40<29:18,  2.32s/it]
 12%|█▎        | 108/864 [05:42<29:16,  2.32s/it]
 13%|█▎        | 109/864 [05:45<29:14,  2.32s/it]
 13%|█▎        | 110/864 [05:47<30:48,  2.45s/it]
                                                 
{'loss': '6.357', 'grad_norm': '0.9572', 'learning_rate': '0.0003', 'epoch': '1.531'}
+
 13%|█▎        | 110/864 [05:47<30:48,  2.45s/it]
 13%|█▎        | 111/864 [05:50<30:16,  2.41s/it]
 13%|█▎        | 112/864 [05:52<29:53,  2.38s/it]
 13%|█▎        | 113/864 [05:54<29:35,  2.36s/it]
 13%|█▎        | 114/864 [05:57<29:23,  2.35s/it]
 13%|█▎        | 115/864 [05:59<29:14,  2.34s/it]
 13%|█▎        | 116/864 [06:01<29:07,  2.34s/it]
 14%|█▎        | 117/864 [06:04<30:30,  2.45s/it]
 14%|█▎        | 118/864 [06:06<29:58,  2.41s/it]
 14%|█▍        | 119/864 [06:09<31:11,  2.51s/it]
 14%|█▍        | 120/864 [06:12<31:52,  2.57s/it]
                                                 
{'loss': '6.199', 'grad_norm': '0.4723', 'learning_rate': '0.0003', 'epoch': '1.671'}
+
 14%|█▍        | 120/864 [06:12<31:52,  2.57s/it]
 14%|█▍        | 121/864 [06:14<30:53,  2.50s/it]
 14%|█▍        | 122/864 [06:17<31:43,  2.56s/it]
 14%|█▍        | 123/864 [06:20<32:14,  2.61s/it]
 14%|█▍        | 124/864 [06:22<32:36,  2.64s/it]
 14%|█▍        | 125/864 [06:25<31:21,  2.55s/it]
 15%|█▍        | 126/864 [06:27<30:28,  2.48s/it]
 15%|█▍        | 127/864 [06:29<29:51,  2.43s/it]
 15%|█▍        | 128/864 [06:32<29:26,  2.40s/it]
 15%|█▍        | 129/864 [06:34<29:07,  2.38s/it]
 15%|█▌        | 130/864 [06:36<28:53,  2.36s/it]
                                                 
{'loss': '6.105', 'grad_norm': '0.6224', 'learning_rate': '0.0003', 'epoch': '1.81'}
+
 15%|█▌        | 130/864 [06:36<28:53,  2.36s/it]
 15%|█▌        | 131/864 [06:39<28:42,  2.35s/it]
 15%|█▌        | 132/864 [06:41<28:34,  2.34s/it]
 15%|█▌        | 133/864 [06:43<28:28,  2.34s/it]
 16%|█▌        | 134/864 [06:46<28:23,  2.33s/it]
 16%|█▌        | 135/864 [06:48<28:18,  2.33s/it]
 16%|█▌        | 136/864 [06:50<28:14,  2.33s/it]
 16%|█▌        | 137/864 [06:52<28:10,  2.33s/it]
 16%|█▌        | 138/864 [06:55<28:07,  2.32s/it]
 16%|█▌        | 139/864 [06:57<28:05,  2.32s/it]
 16%|█▌        | 140/864 [06:59<28:03,  2.33s/it]
                                                 
{'loss': '6.006', 'grad_norm': '0.7163', 'learning_rate': '0.0003', 'epoch': '1.95'}
+
 16%|█▌        | 140/864 [06:59<28:03,  2.33s/it]
 16%|█▋        | 141/864 [07:02<28:01,  2.33s/it]
 16%|█▋        | 142/864 [07:04<27:58,  2.33s/it]
 17%|█▋        | 143/864 [07:06<27:57,  2.33s/it]
 17%|█▋        | 144/864 [07:08<24:18,  2.03s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 4
+
+
  0%|          | 0/77 [00:00<?, ?it/s]
+
  3%|▎         | 2/77 [00:00<00:03, 19.41it/s]
+
  5%|▌         | 4/77 [00:00<00:05, 12.25it/s]
+
  8%|▊         | 6/77 [00:00<00:06, 10.99it/s]
+
 10%|█         | 8/77 [00:00<00:06, 10.47it/s]
+
 13%|█▎        | 10/77 [00:00<00:06, 10.21it/s]
+
 16%|█▌        | 12/77 [00:01<00:06, 10.06it/s]
+
 18%|█▊        | 14/77 [00:01<00:06,  9.95it/s]
+
 21%|██        | 16/77 [00:01<00:06,  9.90it/s]
+
 23%|██▎       | 18/77 [00:01<00:05,  9.86it/s]
+
 25%|██▍       | 19/77 [00:01<00:05,  9.84it/s]
+
 26%|██▌       | 20/77 [00:01<00:05,  9.83it/s]
+
 27%|██▋       | 21/77 [00:02<00:05,  9.82it/s]
+
 29%|██▊       | 22/77 [00:02<00:05,  9.81it/s]
+
 30%|██▉       | 23/77 [00:02<00:05,  9.79it/s]
+
 31%|███       | 24/77 [00:02<00:05,  9.79it/s]
+
 32%|███▏      | 25/77 [00:02<00:05,  9.79it/s]
+
 34%|███▍      | 26/77 [00:02<00:05,  9.78it/s]
+
 35%|███▌      | 27/77 [00:02<00:05,  9.78it/s]
+
 36%|███▋      | 28/77 [00:02<00:05,  9.78it/s]
+
 38%|███▊      | 29/77 [00:02<00:04,  9.78it/s]
+
 39%|███▉      | 30/77 [00:02<00:04,  9.78it/s]
+
 40%|████      | 31/77 [00:03<00:04,  9.77it/s]
+
 42%|████▏     | 32/77 [00:03<00:04,  9.77it/s]
+
 43%|████▎     | 33/77 [00:03<00:04,  9.78it/s]
+
 44%|████▍     | 34/77 [00:03<00:04,  9.78it/s]
+
 45%|████▌     | 35/77 [00:03<00:04,  9.77it/s]
+
 47%|████▋     | 36/77 [00:03<00:04,  9.78it/s]
+
 48%|████▊     | 37/77 [00:03<00:04,  9.78it/s]
+
 49%|████▉     | 38/77 [00:03<00:03,  9.77it/s]
+
 51%|█████     | 39/77 [00:03<00:03,  9.78it/s]
+
 52%|█████▏    | 40/77 [00:03<00:03,  9.78it/s]
+
 53%|█████▎    | 41/77 [00:04<00:03,  9.78it/s]
+
 55%|█████▍    | 42/77 [00:04<00:03,  9.78it/s]
+
 56%|█████▌    | 43/77 [00:04<00:03,  9.78it/s]
+
 57%|█████▋    | 44/77 [00:04<00:03,  9.77it/s]
+
 58%|█████▊    | 45/77 [00:04<00:03,  9.78it/s]
+
 60%|█████▉    | 46/77 [00:04<00:03,  9.78it/s]
+
 61%|██████    | 47/77 [00:04<00:03,  9.78it/s]
+
 62%|██████▏   | 48/77 [00:04<00:02,  9.78it/s]
+
 64%|██████▎   | 49/77 [00:04<00:02,  9.78it/s]
+
 65%|██████▍   | 50/77 [00:05<00:02,  9.78it/s]
+
 66%|██████▌   | 51/77 [00:05<00:02,  9.77it/s]
+
 68%|██████▊   | 52/77 [00:05<00:02,  9.78it/s]
+
 69%|██████▉   | 53/77 [00:05<00:02,  9.78it/s]
+
 70%|███████   | 54/77 [00:05<00:02,  9.78it/s]
+
 71%|███████▏  | 55/77 [00:05<00:02,  9.76it/s]
+
 73%|███████▎  | 56/77 [00:05<00:02,  9.78it/s]
+
 74%|███████▍  | 57/77 [00:05<00:02,  9.78it/s]
+
 75%|███████▌  | 58/77 [00:05<00:01,  9.77it/s]
+
 77%|███████▋  | 59/77 [00:05<00:01,  9.78it/s]
+
 78%|███████▊  | 60/77 [00:06<00:01,  9.78it/s]
+
 79%|███████▉  | 61/77 [00:06<00:01,  9.79it/s]
+
 81%|████████  | 62/77 [00:06<00:01,  9.79it/s]
+
 82%|████████▏ | 63/77 [00:06<00:01,  9.79it/s]
+
 83%|████████▎ | 64/77 [00:06<00:01,  9.79it/s]
+
 84%|████████▍ | 65/77 [00:06<00:01,  9.79it/s]
+
 86%|████████▌ | 66/77 [00:06<00:01,  9.79it/s]
+
 87%|████████▋ | 67/77 [00:06<00:01,  9.80it/s]
+
 88%|████████▊ | 68/77 [00:06<00:00,  9.79it/s]
+
 90%|████████▉ | 69/77 [00:06<00:00,  9.79it/s]
+
 91%|█████████ | 70/77 [00:07<00:00,  9.80it/s]
+
 92%|█████████▏| 71/77 [00:07<00:00,  9.79it/s]
+
 94%|█████████▎| 72/77 [00:07<00:00,  9.79it/s]
+
 95%|█████████▍| 73/77 [00:07<00:00,  9.78it/s]
+
 96%|█████████▌| 74/77 [00:07<00:00,  9.78it/s]
+
 97%|█████████▋| 75/77 [00:07<00:00,  9.79it/s]
+
 99%|█████████▊| 76/77 [00:07<00:00,  9.79it/s]
+
100%|██████████| 77/77 [00:07<00:00,  9.79it/s]
                                                 
+
                                               
{'eval_loss': '5.952', 'eval_runtime': '7.915', 'eval_samples_per_second': '308.4', 'eval_steps_per_second': '9.729', 'epoch': '2'}
+
 17%|█▋        | 144/864 [07:16<24:18,  2.03s/it]
+
100%|██████████| 77/77 [00:07<00:00,  9.79it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0003/checkpoint-144
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0003/checkpoint-144/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0003/checkpoint-144/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:03<00:00,  3.59s/it]
Writing model shards: 100%|██████████| 1/1 [00:03<00:00,  3.59s/it]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0003/checkpoint-144/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0003/checkpoint-144/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0003/checkpoint-144/tokenizer_config.json
+[eval] epoch 2.0 step 144 loss 5.9518 (best 5.9518 @ ep 2.0, 0/10000 since best)
+
+
                                               
 17%|█▋        | 145/864 [07:29<1:35:02,  7.93s/it]
 17%|█▋        | 146/864 [07:32<1:14:45,  6.25s/it]
 17%|█▋        | 147/864 [07:34<1:00:35,  5.07s/it]
 17%|█▋        | 148/864 [07:36<50:40,  4.25s/it]  
 17%|█▋        | 149/864 [07:39<43:43,  3.67s/it]
 17%|█▋        | 150/864 [07:41<38:50,  3.26s/it]
                                                 
{'loss': '5.904', 'grad_norm': '0.5856', 'learning_rate': '0.0003', 'epoch': '2.084'}
+
 17%|█▋        | 150/864 [07:41<38:50,  3.26s/it]
 17%|█▋        | 151/864 [07:43<35:26,  2.98s/it]
 18%|█▊        | 152/864 [07:46<33:02,  2.78s/it]
 18%|█▊        | 153/864 [07:48<31:21,  2.65s/it]
 18%|█▊        | 154/864 [07:50<30:10,  2.55s/it]
 18%|█▊        | 155/864 [07:53<29:19,  2.48s/it]
 18%|█▊        | 156/864 [07:55<28:42,  2.43s/it]
 18%|█▊        | 157/864 [07:57<28:16,  2.40s/it]
 18%|█▊        | 158/864 [08:00<27:57,  2.38s/it]
 18%|█▊        | 159/864 [08:02<27:43,  2.36s/it]
 19%|█▊        | 160/864 [08:04<27:33,  2.35s/it]
                                                 
{'loss': '5.784', 'grad_norm': '0.6897', 'learning_rate': '0.0003', 'epoch': '2.224'}
+
 19%|█▊        | 160/864 [08:04<27:33,  2.35s/it]
 19%|█▊        | 161/864 [08:07<27:25,  2.34s/it]
 19%|█▉        | 162/864 [08:09<27:19,  2.34s/it]
 19%|█▉        | 163/864 [08:11<27:14,  2.33s/it]
 19%|█▉        | 164/864 [08:14<27:11,  2.33s/it]
 19%|█▉        | 165/864 [08:16<27:06,  2.33s/it]
 19%|█▉        | 166/864 [08:18<27:03,  2.33s/it]
 19%|█▉        | 167/864 [08:21<27:00,  2.33s/it]
 19%|█▉        | 168/864 [08:23<26:57,  2.32s/it]
 20%|█▉        | 169/864 [08:25<26:55,  2.32s/it]
 20%|█▉        | 170/864 [08:28<26:52,  2.32s/it]
                                                 
{'loss': '5.706', 'grad_norm': '1.007', 'learning_rate': '0.0003', 'epoch': '2.363'}
+
 20%|█▉        | 170/864 [08:28<26:52,  2.32s/it]
 20%|█▉        | 171/864 [08:30<26:50,  2.32s/it]
 20%|█▉        | 172/864 [08:32<26:47,  2.32s/it]
 20%|██        | 173/864 [08:35<26:44,  2.32s/it]
 20%|██        | 174/864 [08:37<26:42,  2.32s/it]
 20%|██        | 175/864 [08:39<26:40,  2.32s/it]
 20%|██        | 176/864 [08:41<26:38,  2.32s/it]
 20%|██        | 177/864 [08:44<26:35,  2.32s/it]
 21%|██        | 178/864 [08:46<26:34,  2.32s/it]
 21%|██        | 179/864 [08:48<26:31,  2.32s/it]
 21%|██        | 180/864 [08:51<26:29,  2.32s/it]
                                                 
{'loss': '5.638', 'grad_norm': '0.847', 'learning_rate': '0.0003', 'epoch': '2.503'}
+
 21%|██        | 180/864 [08:51<26:29,  2.32s/it]
 21%|██        | 181/864 [08:53<26:27,  2.32s/it]
 21%|██        | 182/864 [08:55<26:25,  2.32s/it]
 21%|██        | 183/864 [08:58<26:22,  2.32s/it]
 21%|██▏       | 184/864 [09:00<26:20,  2.32s/it]
 21%|██▏       | 185/864 [09:02<26:18,  2.32s/it]
 22%|██▏       | 186/864 [09:05<26:16,  2.32s/it]
 22%|██▏       | 187/864 [09:07<26:13,  2.32s/it]
 22%|██▏       | 188/864 [09:09<26:11,  2.33s/it]
 22%|██▏       | 189/864 [09:12<26:09,  2.33s/it]
 22%|██▏       | 190/864 [09:14<26:06,  2.32s/it]
                                                 
{'loss': '5.537', 'grad_norm': '0.4349', 'learning_rate': '0.0003', 'epoch': '2.643'}
+
 22%|██▏       | 190/864 [09:14<26:06,  2.32s/it]
 22%|██▏       | 191/864 [09:16<26:03,  2.32s/it]
 22%|██▏       | 192/864 [09:19<26:01,  2.32s/it]
 22%|██▏       | 193/864 [09:21<25:59,  2.32s/it]
 22%|██▏       | 194/864 [09:23<25:57,  2.32s/it]
 23%|██▎       | 195/864 [09:26<25:55,  2.33s/it]
 23%|██▎       | 196/864 [09:28<25:52,  2.32s/it]
 23%|██▎       | 197/864 [09:30<25:49,  2.32s/it]
 23%|██▎       | 198/864 [09:33<25:47,  2.32s/it]
 23%|██▎       | 199/864 [09:35<25:44,  2.32s/it]
 23%|██▎       | 200/864 [09:37<25:42,  2.32s/it]
                                                 
{'loss': '5.432', 'grad_norm': '0.8382', 'learning_rate': '0.0003', 'epoch': '2.783'}
+
 23%|██▎       | 200/864 [09:37<25:42,  2.32s/it]
 23%|██▎       | 201/864 [09:40<25:41,  2.32s/it]
 23%|██▎       | 202/864 [09:42<25:38,  2.32s/it]
 23%|██▎       | 203/864 [09:44<25:35,  2.32s/it]
 24%|██▎       | 204/864 [09:47<25:33,  2.32s/it]
 24%|██▎       | 205/864 [09:49<25:31,  2.32s/it]
 24%|██▍       | 206/864 [09:51<25:29,  2.32s/it]
 24%|██▍       | 207/864 [09:54<25:27,  2.32s/it]
 24%|██▍       | 208/864 [09:56<25:24,  2.32s/it]
 24%|██▍       | 209/864 [09:58<25:23,  2.33s/it]
 24%|██▍       | 210/864 [10:01<25:20,  2.32s/it]
                                                 
{'loss': '5.422', 'grad_norm': '0.7656', 'learning_rate': '0.0003', 'epoch': '2.922'}
+
 24%|██▍       | 210/864 [10:01<25:20,  2.32s/it]
 24%|██▍       | 211/864 [10:03<25:18,  2.32s/it]
 25%|██▍       | 212/864 [10:05<25:15,  2.32s/it]
 25%|██▍       | 213/864 [10:07<25:14,  2.33s/it]
 25%|██▍       | 214/864 [10:10<25:11,  2.33s/it]
 25%|██▍       | 215/864 [10:12<25:09,  2.33s/it]
 25%|██▌       | 216/864 [10:13<21:53,  2.03s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 4
+
+
  0%|          | 0/77 [00:00<?, ?it/s]
+
  4%|▍         | 3/77 [00:00<00:04, 14.94it/s]
+
  6%|▋         | 5/77 [00:00<00:06, 11.88it/s]
+
  9%|▉         | 7/77 [00:00<00:06, 10.91it/s]
+
 12%|█▏        | 9/77 [00:00<00:06, 10.46it/s]
+
 14%|█▍        | 11/77 [00:01<00:06, 10.22it/s]
+
 17%|█▋        | 13/77 [00:01<00:06, 10.06it/s]
+
 19%|█▉        | 15/77 [00:01<00:06,  9.97it/s]
+
 21%|██        | 16/77 [00:01<00:06,  9.93it/s]
+
 22%|██▏       | 17/77 [00:01<00:06,  9.89it/s]
+
 23%|██▎       | 18/77 [00:01<00:05,  9.87it/s]
+
 25%|██▍       | 19/77 [00:01<00:05,  9.85it/s]
+
 26%|██▌       | 20/77 [00:01<00:05,  9.83it/s]
+
 27%|██▋       | 21/77 [00:02<00:05,  9.81it/s]
+
 29%|██▊       | 22/77 [00:02<00:05,  9.80it/s]
+
 30%|██▉       | 23/77 [00:02<00:05,  9.80it/s]
+
 31%|███       | 24/77 [00:02<00:05,  9.79it/s]
+
 32%|███▏      | 25/77 [00:02<00:05,  9.78it/s]
+
 34%|███▍      | 26/77 [00:02<00:05,  9.78it/s]
+
 35%|███▌      | 27/77 [00:02<00:05,  9.79it/s]
+
 36%|███▋      | 28/77 [00:02<00:05,  9.78it/s]
+
 38%|███▊      | 29/77 [00:02<00:04,  9.78it/s]
+
 39%|███▉      | 30/77 [00:02<00:04,  9.78it/s]
+
 40%|████      | 31/77 [00:03<00:04,  9.78it/s]
+
 42%|████▏     | 32/77 [00:03<00:04,  9.77it/s]
+
 43%|████▎     | 33/77 [00:03<00:04,  9.77it/s]
+
 44%|████▍     | 34/77 [00:03<00:04,  9.77it/s]
+
 45%|████▌     | 35/77 [00:03<00:04,  9.78it/s]
+
 47%|████▋     | 36/77 [00:03<00:04,  9.77it/s]
+
 48%|████▊     | 37/77 [00:03<00:04,  9.77it/s]
+
 49%|████▉     | 38/77 [00:03<00:03,  9.78it/s]
+
 51%|█████     | 39/77 [00:03<00:03,  9.78it/s]
+
 52%|█████▏    | 40/77 [00:03<00:03,  9.78it/s]
+
 53%|█████▎    | 41/77 [00:04<00:03,  9.75it/s]
+
 55%|█████▍    | 42/77 [00:04<00:03,  9.77it/s]
+
 56%|█████▌    | 43/77 [00:04<00:03,  9.78it/s]
+
 57%|█████▋    | 44/77 [00:04<00:03,  9.77it/s]
+
 58%|█████▊    | 45/77 [00:04<00:03,  9.77it/s]
+
 60%|█████▉    | 46/77 [00:04<00:03,  9.77it/s]
+
 61%|██████    | 47/77 [00:04<00:03,  9.79it/s]
+
 62%|██████▏   | 48/77 [00:04<00:02,  9.78it/s]
+
 64%|██████▎   | 49/77 [00:04<00:02,  9.77it/s]
+
 65%|██████▍   | 50/77 [00:05<00:02,  9.78it/s]
+
 66%|██████▌   | 51/77 [00:05<00:02,  9.79it/s]
+
 68%|██████▊   | 52/77 [00:05<00:02,  9.78it/s]
+
 69%|██████▉   | 53/77 [00:05<00:02,  9.78it/s]
+
 70%|███████   | 54/77 [00:05<00:02,  9.79it/s]
+
 71%|███████▏  | 55/77 [00:05<00:02,  9.77it/s]
+
 73%|███████▎  | 56/77 [00:05<00:02,  9.78it/s]
+
 74%|███████▍  | 57/77 [00:05<00:02,  9.77it/s]
+
 75%|███████▌  | 58/77 [00:05<00:01,  9.77it/s]
+
 77%|███████▋  | 59/77 [00:05<00:01,  9.79it/s]
+
 78%|███████▊  | 60/77 [00:06<00:01,  9.78it/s]
+
 79%|███████▉  | 61/77 [00:06<00:01,  9.79it/s]
+
 81%|████████  | 62/77 [00:06<00:01,  9.79it/s]
+
 82%|████████▏ | 63/77 [00:06<00:01,  9.80it/s]
+
 83%|████████▎ | 64/77 [00:06<00:01,  9.79it/s]
+
 84%|████████▍ | 65/77 [00:06<00:01,  9.79it/s]
+
 86%|████████▌ | 66/77 [00:06<00:01,  9.79it/s]
+
 87%|████████▋ | 67/77 [00:06<00:01,  9.79it/s]
+
 88%|████████▊ | 68/77 [00:06<00:00,  9.79it/s]
+
 90%|████████▉ | 69/77 [00:06<00:00,  9.79it/s]
+
 91%|█████████ | 70/77 [00:07<00:00,  9.79it/s]
+
 92%|█████████▏| 71/77 [00:07<00:00,  9.79it/s]
+
 94%|█████████▎| 72/77 [00:07<00:00,  9.79it/s]
+
 95%|█████████▍| 73/77 [00:07<00:00,  9.78it/s]
+
 96%|█████████▌| 74/77 [00:07<00:00,  9.79it/s]
+
 97%|█████████▋| 75/77 [00:07<00:00,  9.79it/s]
+
 99%|█████████▊| 76/77 [00:07<00:00,  9.79it/s]
+
100%|██████████| 77/77 [00:07<00:00,  9.79it/s]
                                                 
+
                                               
{'eval_loss': '5.363', 'eval_runtime': '7.915', 'eval_samples_per_second': '308.4', 'eval_steps_per_second': '9.729', 'epoch': '3'}
+
 25%|██▌       | 216/864 [10:21<21:53,  2.03s/it]
+
100%|██████████| 77/77 [00:07<00:00,  9.79it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0003/checkpoint-216
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0003/checkpoint-216/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0003/checkpoint-216/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:03<00:00,  3.53s/it]
Writing model shards: 100%|██████████| 1/1 [00:03<00:00,  3.53s/it]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0003/checkpoint-216/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0003/checkpoint-216/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0003/checkpoint-216/tokenizer_config.json
+[eval] epoch 3.0 step 216 loss 5.3626 (best 5.3626 @ ep 3.0, 0/10000 since best)
+
+
                                               
 25%|██▌       | 217/864 [10:34<1:22:25,  7.64s/it]
 25%|██▌       | 218/864 [10:37<1:05:05,  6.05s/it]
 25%|██▌       | 219/864 [10:39<52:58,  4.93s/it]  
 25%|██▌       | 220/864 [10:41<44:30,  4.15s/it]
                                                 
{'loss': '5.302', 'grad_norm': '0.4212', 'learning_rate': '0.0003', 'epoch': '3.056'}
+
 25%|██▌       | 220/864 [10:41<44:30,  4.15s/it]
 26%|██▌       | 221/864 [10:43<38:34,  3.60s/it]
 26%|██▌       | 222/864 [10:46<34:25,  3.22s/it]
 26%|██▌       | 223/864 [10:48<31:30,  2.95s/it]
 26%|██▌       | 224/864 [10:50<29:27,  2.76s/it]
 26%|██▌       | 225/864 [10:53<28:01,  2.63s/it]
 26%|██▌       | 226/864 [10:55<27:00,  2.54s/it]
 26%|██▋       | 227/864 [10:57<26:16,  2.47s/it]
 26%|██▋       | 228/864 [11:00<25:44,  2.43s/it]
 27%|██▋       | 229/864 [11:02<25:22,  2.40s/it]
 27%|██▋       | 230/864 [11:04<25:05,  2.38s/it]
                                                 
{'loss': '5.182', 'grad_norm': '0.3276', 'learning_rate': '0.0003', 'epoch': '3.196'}
+
 27%|██▋       | 230/864 [11:04<25:05,  2.38s/it]
 27%|██▋       | 231/864 [11:07<24:54,  2.36s/it]
 27%|██▋       | 232/864 [11:09<24:45,  2.35s/it]
 27%|██▋       | 233/864 [11:11<24:37,  2.34s/it]
 27%|██▋       | 234/864 [11:14<24:31,  2.34s/it]
 27%|██▋       | 235/864 [11:16<24:26,  2.33s/it]
 27%|██▋       | 236/864 [11:18<24:23,  2.33s/it]
 27%|██▋       | 237/864 [11:21<24:19,  2.33s/it]
 28%|██▊       | 238/864 [11:23<24:17,  2.33s/it]
 28%|██▊       | 239/864 [11:25<24:13,  2.33s/it]
 28%|██▊       | 240/864 [11:28<24:10,  2.33s/it]
                                                 
{'loss': '5.131', 'grad_norm': '1.006', 'learning_rate': '0.0003', 'epoch': '3.335'}
+
 28%|██▊       | 240/864 [11:28<24:10,  2.33s/it]
 28%|██▊       | 241/864 [11:30<24:09,  2.33s/it]
 28%|██▊       | 242/864 [11:32<24:06,  2.33s/it]
 28%|██▊       | 243/864 [11:35<24:03,  2.32s/it]
 28%|██▊       | 244/864 [11:37<24:01,  2.32s/it]
 28%|██▊       | 245/864 [11:39<23:59,  2.33s/it]
 28%|██▊       | 246/864 [11:42<23:56,  2.32s/it]
 29%|██▊       | 247/864 [11:44<23:54,  2.33s/it]
 29%|██▊       | 248/864 [11:46<23:51,  2.32s/it]
 29%|██▉       | 249/864 [11:49<23:49,  2.32s/it]
 29%|██▉       | 250/864 [11:51<23:46,  2.32s/it]
                                                 
{'loss': '5.131', 'grad_norm': '0.3993', 'learning_rate': '0.0003', 'epoch': '3.475'}
+
 29%|██▉       | 250/864 [11:51<23:46,  2.32s/it]
 29%|██▉       | 251/864 [11:53<23:44,  2.32s/it]
 29%|██▉       | 252/864 [11:56<23:42,  2.32s/it]
 29%|██▉       | 253/864 [11:58<23:39,  2.32s/it]
 29%|██▉       | 254/864 [12:00<23:36,  2.32s/it]
 30%|██▉       | 255/864 [12:03<23:34,  2.32s/it]
 30%|██▉       | 256/864 [12:05<23:32,  2.32s/it]
 30%|██▉       | 257/864 [12:07<23:30,  2.32s/it]
 30%|██▉       | 258/864 [12:09<23:28,  2.32s/it]
 30%|██▉       | 259/864 [12:12<23:25,  2.32s/it]
 30%|███       | 260/864 [12:14<23:23,  2.32s/it]
                                                 
{'loss': '5.007', 'grad_norm': '0.2496', 'learning_rate': '0.0003', 'epoch': '3.615'}
+
 30%|███       | 260/864 [12:14<23:23,  2.32s/it]
 30%|███       | 261/864 [12:16<23:20,  2.32s/it]
 30%|███       | 262/864 [12:19<23:18,  2.32s/it]
 30%|███       | 263/864 [12:21<23:15,  2.32s/it]
 31%|███       | 264/864 [12:23<23:12,  2.32s/it]
 31%|███       | 265/864 [12:26<23:10,  2.32s/it]
 31%|███       | 266/864 [12:28<23:08,  2.32s/it]
 31%|███       | 267/864 [12:30<23:06,  2.32s/it]
 31%|███       | 268/864 [12:33<23:03,  2.32s/it]
 31%|███       | 269/864 [12:35<23:01,  2.32s/it]
 31%|███▏      | 270/864 [12:37<22:58,  2.32s/it]
                                                 
{'loss': '4.95', 'grad_norm': '0.6246', 'learning_rate': '0.0003', 'epoch': '3.755'}
+
 31%|███▏      | 270/864 [12:37<22:58,  2.32s/it]
 31%|███▏      | 271/864 [12:40<22:56,  2.32s/it]
 31%|███▏      | 272/864 [12:42<22:54,  2.32s/it]
 32%|███▏      | 273/864 [12:44<22:52,  2.32s/it]
 32%|███▏      | 274/864 [12:47<22:51,  2.32s/it]
 32%|███▏      | 275/864 [12:49<22:48,  2.32s/it]
 32%|███▏      | 276/864 [12:51<22:46,  2.32s/it]
 32%|███▏      | 277/864 [12:54<22:43,  2.32s/it]
 32%|███▏      | 278/864 [12:56<22:41,  2.32s/it]
 32%|███▏      | 279/864 [12:58<22:38,  2.32s/it]
 32%|███▏      | 280/864 [13:01<23:39,  2.43s/it]
                                                 
{'loss': '4.952', 'grad_norm': '1.041', 'learning_rate': '0.0003', 'epoch': '3.894'}
+
 32%|███▏      | 280/864 [13:01<23:39,  2.43s/it]
 33%|███▎      | 281/864 [13:03<23:17,  2.40s/it]
 33%|███▎      | 282/864 [13:06<23:02,  2.37s/it]
 33%|███▎      | 283/864 [13:08<22:51,  2.36s/it]
 33%|███▎      | 284/864 [13:10<22:43,  2.35s/it]
 33%|███▎      | 285/864 [13:13<22:37,  2.34s/it]
 33%|███▎      | 286/864 [13:15<22:31,  2.34s/it]
 33%|███▎      | 287/864 [13:17<22:26,  2.33s/it]
 33%|███▎      | 288/864 [13:19<19:30,  2.03s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 4
+
+
  0%|          | 0/77 [00:00<?, ?it/s]
+
  4%|▍         | 3/77 [00:00<00:04, 14.97it/s]
+
  6%|▋         | 5/77 [00:00<00:06, 11.88it/s]
+
  9%|▉         | 7/77 [00:00<00:06, 10.91it/s]
+
 12%|█▏        | 9/77 [00:00<00:06, 10.46it/s]
+
 14%|█▍        | 11/77 [00:01<00:06, 10.21it/s]
+
 17%|█▋        | 13/77 [00:01<00:06, 10.06it/s]
+
 19%|█▉        | 15/77 [00:01<00:06,  9.97it/s]
+
 21%|██        | 16/77 [00:01<00:06,  9.93it/s]
+
 22%|██▏       | 17/77 [00:01<00:06,  9.90it/s]
+
 23%|██▎       | 18/77 [00:01<00:05,  9.87it/s]
+
 25%|██▍       | 19/77 [00:01<00:05,  9.85it/s]
+
 26%|██▌       | 20/77 [00:01<00:05,  9.82it/s]
+
 27%|██▋       | 21/77 [00:02<00:05,  9.80it/s]
+
 29%|██▊       | 22/77 [00:02<00:05,  9.81it/s]
+
 30%|██▉       | 23/77 [00:02<00:05,  9.80it/s]
+
 31%|███       | 24/77 [00:02<00:05,  9.79it/s]
+
 32%|███▏      | 25/77 [00:02<00:05,  9.78it/s]
+
 34%|███▍      | 26/77 [00:02<00:05,  9.78it/s]
+
 35%|███▌      | 27/77 [00:02<00:05,  9.78it/s]
+
 36%|███▋      | 28/77 [00:02<00:05,  9.78it/s]
+
 38%|███▊      | 29/77 [00:02<00:04,  9.78it/s]
+
 39%|███▉      | 30/77 [00:02<00:04,  9.78it/s]
+
 40%|████      | 31/77 [00:03<00:04,  9.78it/s]
+
 42%|████▏     | 32/77 [00:03<00:04,  9.79it/s]
+
 43%|████▎     | 33/77 [00:03<00:04,  9.78it/s]
+
 44%|████▍     | 34/77 [00:03<00:04,  9.78it/s]
+
 45%|████▌     | 35/77 [00:03<00:04,  9.78it/s]
+
 47%|████▋     | 36/77 [00:03<00:04,  9.77it/s]
+
 48%|████▊     | 37/77 [00:03<00:04,  9.78it/s]
+
 49%|████▉     | 38/77 [00:03<00:03,  9.78it/s]
+
 51%|█████     | 39/77 [00:03<00:03,  9.77it/s]
+
 52%|█████▏    | 40/77 [00:03<00:03,  9.76it/s]
+
 53%|█████▎    | 41/77 [00:04<00:03,  9.76it/s]
+
 55%|█████▍    | 42/77 [00:04<00:03,  9.76it/s]
+
 56%|█████▌    | 43/77 [00:04<00:03,  9.76it/s]
+
 57%|█████▋    | 44/77 [00:04<00:03,  9.76it/s]
+
 58%|█████▊    | 45/77 [00:04<00:03,  9.76it/s]
+
 60%|█████▉    | 46/77 [00:04<00:03,  9.77it/s]
+
 61%|██████    | 47/77 [00:04<00:03,  9.77it/s]
+
 62%|██████▏   | 48/77 [00:04<00:02,  9.76it/s]
+
 64%|██████▎   | 49/77 [00:04<00:02,  9.77it/s]
+
 65%|██████▍   | 50/77 [00:05<00:02,  9.77it/s]
+
 66%|██████▌   | 51/77 [00:05<00:02,  9.76it/s]
+
 68%|██████▊   | 52/77 [00:05<00:02,  9.76it/s]
+
 69%|██████▉   | 53/77 [00:05<00:02,  9.76it/s]
+
 70%|███████   | 54/77 [00:05<00:02,  9.71it/s]
+
 71%|███████▏  | 55/77 [00:05<00:02,  9.77it/s]
+
 73%|███████▎  | 56/77 [00:05<00:02,  9.77it/s]
+
 74%|███████▍  | 57/77 [00:05<00:02,  9.77it/s]
+
 75%|███████▌  | 58/77 [00:05<00:01,  9.78it/s]
+
 77%|███████▋  | 59/77 [00:05<00:01,  9.77it/s]
+
 78%|███████▊  | 60/77 [00:06<00:01,  9.77it/s]
+
 79%|███████▉  | 61/77 [00:06<00:01,  9.79it/s]
+
 81%|████████  | 62/77 [00:06<00:01,  9.79it/s]
+
 82%|████████▏ | 63/77 [00:06<00:01,  9.79it/s]
+
 83%|████████▎ | 64/77 [00:06<00:01,  9.78it/s]
+
 84%|████████▍ | 65/77 [00:06<00:01,  9.78it/s]
+
 86%|████████▌ | 66/77 [00:06<00:01,  9.77it/s]
+
 87%|████████▋ | 67/77 [00:06<00:01,  9.77it/s]
+
 88%|████████▊ | 68/77 [00:06<00:00,  9.77it/s]
+
 90%|████████▉ | 69/77 [00:06<00:00,  9.78it/s]
+
 91%|█████████ | 70/77 [00:07<00:00,  9.78it/s]
+
 92%|█████████▏| 71/77 [00:07<00:00,  9.78it/s]
+
 94%|█████████▎| 72/77 [00:07<00:00,  9.77it/s]
+
 95%|█████████▍| 73/77 [00:07<00:00,  9.78it/s]
+
 96%|█████████▌| 74/77 [00:07<00:00,  9.79it/s]
+
 97%|█████████▋| 75/77 [00:07<00:00,  9.78it/s]
+
 99%|█████████▊| 76/77 [00:07<00:00,  9.79it/s]
+
100%|██████████| 77/77 [00:07<00:00,  9.79it/s]
                                                 
+
                                               
{'eval_loss': '5.03', 'eval_runtime': '7.92', 'eval_samples_per_second': '308.2', 'eval_steps_per_second': '9.723', 'epoch': '4'}
+
 33%|███▎      | 288/864 [13:26<19:30,  2.03s/it]
+
100%|██████████| 77/77 [00:07<00:00,  9.79it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0003/checkpoint-288
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0003/checkpoint-288/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0003/checkpoint-288/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:03<00:00,  3.50s/it]
Writing model shards: 100%|██████████| 1/1 [00:03<00:00,  3.50s/it]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0003/checkpoint-288/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0003/checkpoint-288/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0003/checkpoint-288/tokenizer_config.json
+[eval] epoch 4.0 step 288 loss 5.0302 (best 5.0302 @ ep 4.0, 0/10000 since best)
+
+
                                               
 33%|███▎      | 289/864 [13:39<1:13:10,  7.64s/it]
 34%|███▎      | 290/864 [13:42<58:54,  6.16s/it]  
                                                 
{'loss': '4.972', 'grad_norm': '0.5914', 'learning_rate': '0.0003', 'epoch': '4.028'}
+
 34%|███▎      | 290/864 [13:42<58:54,  6.16s/it]
 34%|███▎      | 291/864 [13:45<48:59,  5.13s/it]
 34%|███▍      | 292/864 [13:47<40:51,  4.29s/it]
 34%|███▍      | 293/864 [13:50<36:13,  3.81s/it]
 34%|███▍      | 294/864 [13:52<31:55,  3.36s/it]
 34%|███▍      | 295/864 [13:55<29:59,  3.16s/it]
 34%|███▍      | 296/864 [13:57<27:33,  2.91s/it]
 34%|███▍      | 297/864 [14:00<26:53,  2.85s/it]
 34%|███▍      | 298/864 [14:02<26:26,  2.80s/it]
 35%|███▍      | 299/864 [14:05<25:01,  2.66s/it]
 35%|███▍      | 300/864 [14:07<24:01,  2.56s/it]
                                                 
{'loss': '4.799', 'grad_norm': '0.2368', 'learning_rate': '0.0003', 'epoch': '4.168'}
+
 35%|███▍      | 300/864 [14:07<24:01,  2.56s/it]
 35%|███▍      | 301/864 [14:09<23:19,  2.49s/it]
 35%|███▍      | 302/864 [14:12<22:49,  2.44s/it]
 35%|███▌      | 303/864 [14:14<22:27,  2.40s/it]
 35%|███▌      | 304/864 [14:16<22:11,  2.38s/it]
 35%|███▌      | 305/864 [14:19<21:59,  2.36s/it]
 35%|███▌      | 306/864 [14:21<21:50,  2.35s/it]
 36%|███▌      | 307/864 [14:23<21:43,  2.34s/it]
 36%|███▌      | 308/864 [14:26<21:38,  2.34s/it]
 36%|███▌      | 309/864 [14:28<21:33,  2.33s/it]
 36%|███▌      | 310/864 [14:30<21:29,  2.33s/it]
                                                 
{'loss': '4.705', 'grad_norm': '0.2569', 'learning_rate': '0.0003', 'epoch': '4.307'}
+
 36%|███▌      | 310/864 [14:30<21:29,  2.33s/it]
 36%|███▌      | 311/864 [14:33<21:26,  2.33s/it]
 36%|███▌      | 312/864 [14:35<21:23,  2.32s/it]
 36%|███▌      | 313/864 [14:37<21:20,  2.32s/it]
 36%|███▋      | 314/864 [14:40<21:17,  2.32s/it]
 36%|███▋      | 315/864 [14:42<21:14,  2.32s/it]
 37%|███▋      | 316/864 [14:44<21:11,  2.32s/it]
 37%|███▋      | 317/864 [14:47<21:09,  2.32s/it]
 37%|███▋      | 318/864 [14:49<21:06,  2.32s/it]
 37%|███▋      | 319/864 [14:51<21:04,  2.32s/it]
 37%|███▋      | 320/864 [14:53<21:02,  2.32s/it]
                                                 
{'loss': '4.648', 'grad_norm': '0.4643', 'learning_rate': '0.0003', 'epoch': '4.447'}
+
 37%|███▋      | 320/864 [14:54<21:02,  2.32s/it]
 37%|███▋      | 321/864 [14:56<21:00,  2.32s/it]
 37%|███▋      | 322/864 [14:58<20:58,  2.32s/it]
 37%|███▋      | 323/864 [15:00<20:55,  2.32s/it]
 38%|███▊      | 324/864 [15:03<20:52,  2.32s/it]
 38%|███▊      | 325/864 [15:05<20:50,  2.32s/it]
 38%|███▊      | 326/864 [15:07<20:48,  2.32s/it]
 38%|███▊      | 327/864 [15:10<20:45,  2.32s/it]
 38%|███▊      | 328/864 [15:12<20:43,  2.32s/it]
 38%|███▊      | 329/864 [15:14<20:41,  2.32s/it]
 38%|███▊      | 330/864 [15:17<20:39,  2.32s/it]
                                                 
{'loss': '4.645', 'grad_norm': '0.5474', 'learning_rate': '0.0003', 'epoch': '4.587'}
+
 38%|███▊      | 330/864 [15:17<20:39,  2.32s/it]
 38%|███▊      | 331/864 [15:19<20:36,  2.32s/it]
 38%|███▊      | 332/864 [15:21<20:34,  2.32s/it]
 39%|███▊      | 333/864 [15:24<20:31,  2.32s/it]
 39%|███▊      | 334/864 [15:26<20:30,  2.32s/it]
 39%|███▉      | 335/864 [15:28<20:27,  2.32s/it]
 39%|███▉      | 336/864 [15:31<20:25,  2.32s/it]
 39%|███▉      | 337/864 [15:33<20:23,  2.32s/it]
 39%|███▉      | 338/864 [15:35<20:21,  2.32s/it]
 39%|███▉      | 339/864 [15:38<20:19,  2.32s/it]
 39%|███▉      | 340/864 [15:40<20:16,  2.32s/it]
                                                 
{'loss': '4.588', 'grad_norm': '0.3637', 'learning_rate': '0.0003', 'epoch': '4.727'}
+
 39%|███▉      | 340/864 [15:40<20:16,  2.32s/it]
 39%|███▉      | 341/864 [15:42<20:14,  2.32s/it]
 40%|███▉      | 342/864 [15:45<20:11,  2.32s/it]
 40%|███▉      | 343/864 [15:47<20:09,  2.32s/it]
 40%|███▉      | 344/864 [15:49<20:06,  2.32s/it]
 40%|███▉      | 345/864 [15:52<20:04,  2.32s/it]
 40%|████      | 346/864 [15:54<20:02,  2.32s/it]
 40%|████      | 347/864 [15:56<20:00,  2.32s/it]
 40%|████      | 348/864 [15:58<19:57,  2.32s/it]
 40%|████      | 349/864 [16:01<19:55,  2.32s/it]
 41%|████      | 350/864 [16:03<19:53,  2.32s/it]
                                                 
{'loss': '4.511', 'grad_norm': '0.3925', 'learning_rate': '0.0003', 'epoch': '4.866'}
+
 41%|████      | 350/864 [16:03<19:53,  2.32s/it]
 41%|████      | 351/864 [16:05<19:50,  2.32s/it]
 41%|████      | 352/864 [16:08<19:48,  2.32s/it]
 41%|████      | 353/864 [16:10<19:45,  2.32s/it]
 41%|████      | 354/864 [16:12<19:43,  2.32s/it]
 41%|████      | 355/864 [16:15<19:41,  2.32s/it]
 41%|████      | 356/864 [16:17<19:39,  2.32s/it]
 41%|████▏     | 357/864 [16:19<19:37,  2.32s/it]
 41%|████▏     | 358/864 [16:22<19:35,  2.32s/it]
 42%|████▏     | 359/864 [16:24<19:32,  2.32s/it]
 42%|████▏     | 360/864 [16:25<16:59,  2.02s/it]
                                                 
{'loss': '4.477', 'grad_norm': '0.4146', 'learning_rate': '0.0003', 'epoch': '5'}
+
 42%|████▏     | 360/864 [16:25<16:59,  2.02s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 4
+
+
  0%|          | 0/77 [00:00<?, ?it/s]
+
  3%|▎         | 2/77 [00:00<00:03, 19.46it/s]
+
  5%|▌         | 4/77 [00:00<00:05, 12.28it/s]
+
  8%|▊         | 6/77 [00:00<00:06, 10.99it/s]
+
 10%|█         | 8/77 [00:00<00:06, 10.47it/s]
+
 13%|█▎        | 10/77 [00:00<00:06, 10.22it/s]
+
 16%|█▌        | 12/77 [00:01<00:06, 10.07it/s]
+
 18%|█▊        | 14/77 [00:01<00:06,  9.97it/s]
+
 21%|██        | 16/77 [00:01<00:06,  9.91it/s]
+
 23%|██▎       | 18/77 [00:01<00:05,  9.87it/s]
+
 25%|██▍       | 19/77 [00:01<00:05,  9.85it/s]
+
 26%|██▌       | 20/77 [00:01<00:05,  9.84it/s]
+
 27%|██▋       | 21/77 [00:02<00:05,  9.83it/s]
+
 29%|██▊       | 22/77 [00:02<00:05,  9.81it/s]
+
 30%|██▉       | 23/77 [00:02<00:05,  9.80it/s]
+
 31%|███       | 24/77 [00:02<00:05,  9.80it/s]
+
 32%|███▏      | 25/77 [00:02<00:05,  9.80it/s]
+
 34%|███▍      | 26/77 [00:02<00:05,  9.79it/s]
+
 35%|███▌      | 27/77 [00:02<00:05,  9.79it/s]
+
 36%|███▋      | 28/77 [00:02<00:05,  9.79it/s]
+
 38%|███▊      | 29/77 [00:02<00:04,  9.78it/s]
+
 39%|███▉      | 30/77 [00:02<00:04,  9.77it/s]
+
 40%|████      | 31/77 [00:03<00:04,  9.77it/s]
+
 42%|████▏     | 32/77 [00:03<00:04,  9.74it/s]
+
 43%|████▎     | 33/77 [00:03<00:04,  9.78it/s]
+
 44%|████▍     | 34/77 [00:03<00:04,  9.79it/s]
+
 45%|████▌     | 35/77 [00:03<00:04,  9.78it/s]
+
 47%|████▋     | 36/77 [00:03<00:04,  9.78it/s]
+
 48%|████▊     | 37/77 [00:03<00:04,  9.78it/s]
+
 49%|████▉     | 38/77 [00:03<00:03,  9.77it/s]
+
 51%|█████     | 39/77 [00:03<00:03,  9.77it/s]
+
 52%|█████▏    | 40/77 [00:03<00:03,  9.78it/s]
+
 53%|█████▎    | 41/77 [00:04<00:03,  9.77it/s]
+
 55%|█████▍    | 42/77 [00:04<00:03,  9.77it/s]
+
 56%|█████▌    | 43/77 [00:04<00:03,  9.77it/s]
+
 57%|█████▋    | 44/77 [00:04<00:03,  9.78it/s]
+
 58%|█████▊    | 45/77 [00:04<00:03,  9.78it/s]
+
 60%|█████▉    | 46/77 [00:04<00:03,  9.77it/s]
+
 61%|██████    | 47/77 [00:04<00:03,  9.77it/s]
+
 62%|██████▏   | 48/77 [00:04<00:02,  9.78it/s]
+
 64%|██████▎   | 49/77 [00:04<00:02,  9.78it/s]
+
 65%|██████▍   | 50/77 [00:05<00:02,  9.78it/s]
+
 66%|██████▌   | 51/77 [00:05<00:02,  9.78it/s]
+
 68%|██████▊   | 52/77 [00:05<00:02,  9.78it/s]
+
 69%|██████▉   | 53/77 [00:05<00:02,  9.78it/s]
+
 70%|███████   | 54/77 [00:05<00:02,  9.77it/s]
+
 71%|███████▏  | 55/77 [00:05<00:02,  9.78it/s]
+
 73%|███████▎  | 56/77 [00:05<00:02,  9.78it/s]
+
 74%|███████▍  | 57/77 [00:05<00:02,  9.75it/s]
+
 75%|███████▌  | 58/77 [00:05<00:01,  9.77it/s]
+
 77%|███████▋  | 59/77 [00:05<00:01,  9.78it/s]
+
 78%|███████▊  | 60/77 [00:06<00:01,  9.78it/s]
+
 79%|███████▉  | 61/77 [00:06<00:01,  9.79it/s]
+
 81%|████████  | 62/77 [00:06<00:01,  9.80it/s]
+
 82%|████████▏ | 63/77 [00:06<00:01,  9.79it/s]
+
 83%|████████▎ | 64/77 [00:06<00:01,  9.79it/s]
+
 84%|████████▍ | 65/77 [00:06<00:01,  9.79it/s]
+
 86%|████████▌ | 66/77 [00:06<00:01,  9.79it/s]
+
 87%|████████▋ | 67/77 [00:06<00:01,  9.79it/s]
+
 88%|████████▊ | 68/77 [00:06<00:00,  9.79it/s]
+
 90%|████████▉ | 69/77 [00:06<00:00,  9.79it/s]
+
 91%|█████████ | 70/77 [00:07<00:00,  9.78it/s]
+
 92%|█████████▏| 71/77 [00:07<00:00,  9.78it/s]
+
 94%|█████████▎| 72/77 [00:07<00:00,  9.79it/s]
+
 95%|█████████▍| 73/77 [00:07<00:00,  9.79it/s]
+
 96%|█████████▌| 74/77 [00:07<00:00,  9.79it/s]
+
 97%|█████████▋| 75/77 [00:07<00:00,  9.78it/s]
+
 99%|█████████▊| 76/77 [00:07<00:00,  9.78it/s]
+
100%|██████████| 77/77 [00:07<00:00,  9.79it/s]
                                                 
+
                                               
{'eval_loss': '4.589', 'eval_runtime': '7.913', 'eval_samples_per_second': '308.5', 'eval_steps_per_second': '9.73', 'epoch': '5'}
+
 42%|████▏     | 360/864 [16:33<16:59,  2.02s/it]
+
100%|██████████| 77/77 [00:07<00:00,  9.79it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0003/checkpoint-360
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0003/checkpoint-360/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0003/checkpoint-360/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:02<00:00,  2.56s/it]
Writing model shards: 100%|██████████| 1/1 [00:02<00:00,  2.56s/it]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0003/checkpoint-360/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0003/checkpoint-360/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0003/checkpoint-360/tokenizer_config.json
+[eval] epoch 5.0 step 360 loss 4.5894 (best 4.5894 @ ep 5.0, 0/10000 since best)
+
+
                                               
 42%|████▏     | 361/864 [16:45<1:01:33,  7.34s/it]
 42%|████▏     | 362/864 [16:47<48:48,  5.83s/it]  
 42%|████▏     | 363/864 [16:50<39:55,  4.78s/it]
 42%|████▏     | 364/864 [16:52<33:41,  4.04s/it]
 42%|████▏     | 365/864 [16:54<29:20,  3.53s/it]
 42%|████▏     | 366/864 [16:57<26:16,  3.17s/it]
 42%|████▏     | 367/864 [16:59<24:08,  2.91s/it]
 43%|████▎     | 368/864 [17:01<22:37,  2.74s/it]
 43%|████▎     | 369/864 [17:04<21:33,  2.61s/it]
 43%|████▎     | 370/864 [17:06<20:47,  2.53s/it]
                                                 
{'loss': '4.355', 'grad_norm': '0.557', 'learning_rate': '0.0003', 'epoch': '5.14'}
+
 43%|████▎     | 370/864 [17:06<20:47,  2.53s/it]
 43%|████▎     | 371/864 [17:08<20:15,  2.47s/it]
 43%|████▎     | 372/864 [17:11<19:52,  2.42s/it]
 43%|████▎     | 373/864 [17:13<19:34,  2.39s/it]
 43%|████▎     | 374/864 [17:15<19:22,  2.37s/it]
 43%|████▎     | 375/864 [17:18<19:12,  2.36s/it]
 44%|████▎     | 376/864 [17:20<19:05,  2.35s/it]
 44%|████▎     | 377/864 [17:22<19:00,  2.34s/it]
 44%|████▍     | 378/864 [17:25<18:55,  2.34s/it]
 44%|████▍     | 379/864 [17:27<18:51,  2.33s/it]
 44%|████▍     | 380/864 [17:29<18:47,  2.33s/it]
                                                 
{'loss': '4.362', 'grad_norm': '0.6994', 'learning_rate': '0.0003', 'epoch': '5.279'}
+
 44%|████▍     | 380/864 [17:29<18:47,  2.33s/it]
 44%|████▍     | 381/864 [17:32<18:44,  2.33s/it]
 44%|████▍     | 382/864 [17:34<18:41,  2.33s/it]
 44%|████▍     | 383/864 [17:36<18:39,  2.33s/it]
 44%|████▍     | 384/864 [17:39<18:36,  2.33s/it]
 45%|████▍     | 385/864 [17:41<18:34,  2.33s/it]
 45%|████▍     | 386/864 [17:43<18:31,  2.33s/it]
 45%|████▍     | 387/864 [17:46<18:29,  2.33s/it]
 45%|████▍     | 388/864 [17:48<18:26,  2.33s/it]
 45%|████▌     | 389/864 [17:50<18:24,  2.33s/it]
 45%|████▌     | 390/864 [17:52<18:21,  2.32s/it]
                                                 
{'loss': '4.337', 'grad_norm': '0.5208', 'learning_rate': '0.0003', 'epoch': '5.419'}
+
 45%|████▌     | 390/864 [17:53<18:21,  2.32s/it]
 45%|████▌     | 391/864 [17:55<18:19,  2.32s/it]
 45%|████▌     | 392/864 [17:57<18:17,  2.32s/it]
 45%|████▌     | 393/864 [17:59<18:14,  2.32s/it]
 46%|████▌     | 394/864 [18:02<18:12,  2.32s/it]
 46%|████▌     | 395/864 [18:04<18:09,  2.32s/it]
 46%|████▌     | 396/864 [18:06<18:07,  2.32s/it]
 46%|████▌     | 397/864 [18:09<18:05,  2.32s/it]
 46%|████▌     | 398/864 [18:11<18:02,  2.32s/it]
 46%|████▌     | 399/864 [18:13<18:00,  2.32s/it]
 46%|████▋     | 400/864 [18:16<17:58,  2.32s/it]
                                                 
{'loss': '4.261', 'grad_norm': '0.2336', 'learning_rate': '0.0003', 'epoch': '5.559'}
+
 46%|████▋     | 400/864 [18:16<17:58,  2.32s/it]
 46%|████▋     | 401/864 [18:18<17:56,  2.32s/it]
 47%|████▋     | 402/864 [18:20<17:53,  2.32s/it]
 47%|████▋     | 403/864 [18:23<17:51,  2.32s/it]
 47%|████▋     | 404/864 [18:25<17:48,  2.32s/it]
 47%|████▋     | 405/864 [18:27<17:46,  2.32s/it]
 47%|████▋     | 406/864 [18:30<17:44,  2.32s/it]
 47%|████▋     | 407/864 [18:32<17:42,  2.32s/it]
 47%|████▋     | 408/864 [18:34<17:40,  2.33s/it]
 47%|████▋     | 409/864 [18:37<17:37,  2.32s/it]
 47%|████▋     | 410/864 [18:39<17:34,  2.32s/it]
                                                 
{'loss': '4.221', 'grad_norm': '0.268', 'learning_rate': '0.0003', 'epoch': '5.699'}
+
 47%|████▋     | 410/864 [18:39<17:34,  2.32s/it]
 48%|████▊     | 411/864 [18:41<17:32,  2.32s/it]
 48%|████▊     | 412/864 [18:44<17:30,  2.32s/it]
 48%|████▊     | 413/864 [18:46<17:28,  2.32s/it]
 48%|████▊     | 414/864 [18:48<17:26,  2.33s/it]
 48%|████▊     | 415/864 [18:51<17:23,  2.32s/it]
 48%|████▊     | 416/864 [18:53<17:21,  2.32s/it]
 48%|████▊     | 417/864 [18:55<17:19,  2.32s/it]
 48%|████▊     | 418/864 [18:58<17:17,  2.33s/it]
 48%|████▊     | 419/864 [19:00<17:14,  2.33s/it]
 49%|████▊     | 420/864 [19:02<17:12,  2.33s/it]
                                                 
{'loss': '4.196', 'grad_norm': '0.315', 'learning_rate': '0.0003', 'epoch': '5.838'}
+
 49%|████▊     | 420/864 [19:02<17:12,  2.33s/it]
 49%|████▊     | 421/864 [19:05<17:10,  2.33s/it]
 49%|████▉     | 422/864 [19:07<17:08,  2.33s/it]
 49%|████▉     | 423/864 [19:09<17:05,  2.33s/it]
 49%|████▉     | 424/864 [19:12<17:03,  2.33s/it]
 49%|████▉     | 425/864 [19:14<17:00,  2.33s/it]
 49%|████▉     | 426/864 [19:16<16:58,  2.33s/it]
 49%|████▉     | 427/864 [19:18<16:56,  2.33s/it]
 50%|████▉     | 428/864 [19:21<16:53,  2.33s/it]
 50%|████▉     | 429/864 [19:23<16:51,  2.33s/it]
 50%|████▉     | 430/864 [19:25<16:48,  2.32s/it]
                                                 
{'loss': '4.171', 'grad_norm': '0.3593', 'learning_rate': '0.0003', 'epoch': '5.978'}
+
 50%|████▉     | 430/864 [19:25<16:48,  2.32s/it]
 50%|████▉     | 431/864 [19:28<16:46,  2.33s/it]
 50%|█████     | 432/864 [19:29<14:34,  2.03s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 4
+
+
  0%|          | 0/77 [00:00<?, ?it/s]
+
  4%|▍         | 3/77 [00:00<00:04, 15.02it/s]
+
  6%|▋         | 5/77 [00:00<00:06, 11.89it/s]
+
  9%|▉         | 7/77 [00:00<00:06, 10.92it/s]
+
 12%|█▏        | 9/77 [00:00<00:06, 10.46it/s]
+
 14%|█▍        | 11/77 [00:01<00:06, 10.22it/s]
+
 17%|█▋        | 13/77 [00:01<00:06, 10.06it/s]
+
 19%|█▉        | 15/77 [00:01<00:06,  9.97it/s]
+
 21%|██        | 16/77 [00:01<00:06,  9.93it/s]
+
 22%|██▏       | 17/77 [00:01<00:06,  9.90it/s]
+
 23%|██▎       | 18/77 [00:01<00:05,  9.87it/s]
+
 25%|██▍       | 19/77 [00:01<00:05,  9.85it/s]
+
 26%|██▌       | 20/77 [00:01<00:05,  9.83it/s]
+
 27%|██▋       | 21/77 [00:02<00:05,  9.81it/s]
+
 29%|██▊       | 22/77 [00:02<00:05,  9.80it/s]
+
 30%|██▉       | 23/77 [00:02<00:05,  9.80it/s]
+
 31%|███       | 24/77 [00:02<00:05,  9.76it/s]
+
 32%|███▏      | 25/77 [00:02<00:05,  9.78it/s]
+
 34%|███▍      | 26/77 [00:02<00:05,  9.79it/s]
+
 35%|███▌      | 27/77 [00:02<00:05,  9.78it/s]
+
 36%|███▋      | 28/77 [00:02<00:05,  9.78it/s]
+
 38%|███▊      | 29/77 [00:02<00:04,  9.78it/s]
+
 39%|███▉      | 30/77 [00:02<00:04,  9.77it/s]
+
 40%|████      | 31/77 [00:03<00:04,  9.78it/s]
+
 42%|████▏     | 32/77 [00:03<00:04,  9.78it/s]
+
 43%|████▎     | 33/77 [00:03<00:04,  9.78it/s]
+
 44%|████▍     | 34/77 [00:03<00:04,  9.78it/s]
+
 45%|████▌     | 35/77 [00:03<00:04,  9.78it/s]
+
 47%|████▋     | 36/77 [00:03<00:04,  9.78it/s]
+
 48%|████▊     | 37/77 [00:03<00:04,  9.78it/s]
+
 49%|████▉     | 38/77 [00:03<00:03,  9.78it/s]
+
 51%|█████     | 39/77 [00:03<00:03,  9.77it/s]
+
 52%|█████▏    | 40/77 [00:03<00:03,  9.76it/s]
+
 53%|█████▎    | 41/77 [00:04<00:03,  9.78it/s]
+
 55%|█████▍    | 42/77 [00:04<00:03,  9.77it/s]
+
 56%|█████▌    | 43/77 [00:04<00:03,  9.78it/s]
+
 57%|█████▋    | 44/77 [00:04<00:03,  9.78it/s]
+
 58%|█████▊    | 45/77 [00:04<00:03,  9.77it/s]
+
 60%|█████▉    | 46/77 [00:04<00:03,  9.78it/s]
+
 61%|██████    | 47/77 [00:04<00:03,  9.78it/s]
+
 62%|██████▏   | 48/77 [00:04<00:02,  9.78it/s]
+
 64%|██████▎   | 49/77 [00:04<00:02,  9.77it/s]
+
 65%|██████▍   | 50/77 [00:05<00:02,  9.78it/s]
+
 66%|██████▌   | 51/77 [00:05<00:02,  9.77it/s]
+
 68%|██████▊   | 52/77 [00:05<00:02,  9.78it/s]
+
 69%|██████▉   | 53/77 [00:05<00:02,  9.77it/s]
+
 70%|███████   | 54/77 [00:05<00:02,  9.77it/s]
+
 71%|███████▏  | 55/77 [00:05<00:02,  9.77it/s]
+
 73%|███████▎  | 56/77 [00:05<00:02,  9.78it/s]
+
 74%|███████▍  | 57/77 [00:05<00:02,  9.78it/s]
+
 75%|███████▌  | 58/77 [00:05<00:01,  9.78it/s]
+
 77%|███████▋  | 59/77 [00:05<00:01,  9.78it/s]
+
 78%|███████▊  | 60/77 [00:06<00:01,  9.78it/s]
+
 79%|███████▉  | 61/77 [00:06<00:01,  9.80it/s]
+
 81%|████████  | 62/77 [00:06<00:01,  9.79it/s]
+
 82%|████████▏ | 63/77 [00:06<00:01,  9.79it/s]
+
 83%|████████▎ | 64/77 [00:06<00:01,  9.79it/s]
+
 84%|████████▍ | 65/77 [00:06<00:01,  9.78it/s]
+
 86%|████████▌ | 66/77 [00:06<00:01,  9.79it/s]
+
 87%|████████▋ | 67/77 [00:06<00:01,  9.79it/s]
+
 88%|████████▊ | 68/77 [00:06<00:00,  9.79it/s]
+
 90%|████████▉ | 69/77 [00:06<00:00,  9.78it/s]
+
 91%|█████████ | 70/77 [00:07<00:00,  9.79it/s]
+
 92%|█████████▏| 71/77 [00:07<00:00,  9.79it/s]
+
 94%|█████████▎| 72/77 [00:07<00:00,  9.79it/s]
+
 95%|█████████▍| 73/77 [00:07<00:00,  9.77it/s]
+
 96%|█████████▌| 74/77 [00:07<00:00,  9.78it/s]
+
 97%|█████████▋| 75/77 [00:07<00:00,  9.78it/s]
+
 99%|█████████▊| 76/77 [00:07<00:00,  9.78it/s]
+
100%|██████████| 77/77 [00:07<00:00,  9.79it/s]
                                                 
+
                                               
{'eval_loss': '4.343', 'eval_runtime': '7.916', 'eval_samples_per_second': '308.4', 'eval_steps_per_second': '9.728', 'epoch': '6'}
+
 50%|█████     | 432/864 [19:37<14:34,  2.03s/it]
+
100%|██████████| 77/77 [00:07<00:00,  9.79it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0003/checkpoint-432
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0003/checkpoint-432/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0003/checkpoint-432/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:02<00:00,  2.58s/it]
Writing model shards: 100%|██████████| 1/1 [00:02<00:00,  2.58s/it]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0003/checkpoint-432/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0003/checkpoint-432/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0003/checkpoint-432/tokenizer_config.json
+[eval] epoch 6.0 step 432 loss 4.3426 (best 4.3426 @ ep 6.0, 0/10000 since best)
+
+
                                               
 50%|█████     | 433/864 [19:49<52:45,  7.34s/it]
 50%|█████     | 434/864 [19:51<41:49,  5.84s/it]
 50%|█████     | 435/864 [19:54<34:11,  4.78s/it]
 50%|█████     | 436/864 [19:56<28:50,  4.04s/it]
 51%|█████     | 437/864 [19:58<25:06,  3.53s/it]
 51%|█████     | 438/864 [20:00<22:28,  3.17s/it]
 51%|█████     | 439/864 [20:03<20:37,  2.91s/it]
 51%|█████     | 440/864 [20:05<19:20,  2.74s/it]
                                                 
{'loss': '4.087', 'grad_norm': '0.8794', 'learning_rate': '0.0003', 'epoch': '6.112'}
+
 51%|█████     | 440/864 [20:05<19:20,  2.74s/it]
 51%|█████     | 441/864 [20:07<18:24,  2.61s/it]
 51%|█████     | 442/864 [20:10<17:46,  2.53s/it]
 51%|█████▏    | 443/864 [20:12<17:18,  2.47s/it]
 51%|█████▏    | 444/864 [20:14<16:57,  2.42s/it]
 52%|█████▏    | 445/864 [20:17<16:42,  2.39s/it]
 52%|█████▏    | 446/864 [20:19<16:31,  2.37s/it]
 52%|█████▏    | 447/864 [20:21<16:22,  2.36s/it]
 52%|█████▏    | 448/864 [20:24<16:16,  2.35s/it]
 52%|█████▏    | 449/864 [20:26<16:10,  2.34s/it]
 52%|█████▏    | 450/864 [20:28<16:06,  2.34s/it]
                                                 
{'loss': '4.195', 'grad_norm': '0.7271', 'learning_rate': '0.0003', 'epoch': '6.252'}
+
 52%|█████▏    | 450/864 [20:28<16:06,  2.34s/it]
 52%|█████▏    | 451/864 [20:31<16:46,  2.44s/it]
 52%|█████▏    | 452/864 [20:33<16:29,  2.40s/it]
 52%|█████▏    | 453/864 [20:36<16:16,  2.38s/it]
 53%|█████▎    | 454/864 [20:38<16:07,  2.36s/it]
 53%|█████▎    | 455/864 [20:40<15:59,  2.35s/it]
 53%|█████▎    | 456/864 [20:43<15:54,  2.34s/it]
 53%|█████▎    | 457/864 [20:45<15:49,  2.33s/it]
 53%|█████▎    | 458/864 [20:47<15:45,  2.33s/it]
 53%|█████▎    | 459/864 [20:50<15:42,  2.33s/it]
 53%|█████▎    | 460/864 [20:52<15:38,  2.32s/it]
                                                 
{'loss': '4.237', 'grad_norm': '1.018', 'learning_rate': '0.0003', 'epoch': '6.391'}
+
 53%|█████▎    | 460/864 [20:52<15:38,  2.32s/it]
 53%|█████▎    | 461/864 [20:54<15:36,  2.32s/it]
 53%|█████▎    | 462/864 [20:57<15:33,  2.32s/it]
 54%|█████▎    | 463/864 [20:59<16:17,  2.44s/it]
 54%|█████▎    | 464/864 [21:02<16:46,  2.52s/it]
 54%|█████▍    | 465/864 [21:04<16:20,  2.46s/it]
 54%|█████▍    | 466/864 [21:07<16:45,  2.53s/it]
 54%|█████▍    | 467/864 [21:09<16:18,  2.47s/it]
 54%|█████▍    | 468/864 [21:12<16:43,  2.54s/it]
 54%|█████▍    | 469/864 [21:14<16:15,  2.47s/it]
 54%|█████▍    | 470/864 [21:17<15:55,  2.43s/it]
                                                 
{'loss': '4.177', 'grad_norm': '0.3174', 'learning_rate': '0.0003', 'epoch': '6.531'}
+
 54%|█████▍    | 470/864 [21:17<15:55,  2.43s/it]
 55%|█████▍    | 471/864 [21:19<16:25,  2.51s/it]
 55%|█████▍    | 472/864 [21:22<16:00,  2.45s/it]
 55%|█████▍    | 473/864 [21:24<16:26,  2.52s/it]
 55%|█████▍    | 474/864 [21:27<15:59,  2.46s/it]
 55%|█████▍    | 475/864 [21:29<15:41,  2.42s/it]
 55%|█████▌    | 476/864 [21:31<15:27,  2.39s/it]
 55%|█████▌    | 477/864 [21:34<15:17,  2.37s/it]
 55%|█████▌    | 478/864 [21:36<15:09,  2.36s/it]
 55%|█████▌    | 479/864 [21:38<15:03,  2.35s/it]
 56%|█████▌    | 480/864 [21:41<14:58,  2.34s/it]
                                                 
{'loss': '4.067', 'grad_norm': '0.1904', 'learning_rate': '0.0003', 'epoch': '6.671'}
+
 56%|█████▌    | 480/864 [21:41<14:58,  2.34s/it]
 56%|█████▌    | 481/864 [21:43<14:53,  2.33s/it]
 56%|█████▌    | 482/864 [21:45<14:50,  2.33s/it]
 56%|█████▌    | 483/864 [21:48<14:47,  2.33s/it]
 56%|█████▌    | 484/864 [21:50<14:44,  2.33s/it]
 56%|█████▌    | 485/864 [21:52<14:41,  2.33s/it]
 56%|█████▋    | 486/864 [21:55<14:38,  2.33s/it]
 56%|█████▋    | 487/864 [21:57<14:35,  2.32s/it]
 56%|█████▋    | 488/864 [21:59<14:33,  2.32s/it]
 57%|█████▋    | 489/864 [22:01<14:30,  2.32s/it]
 57%|█████▋    | 490/864 [22:04<14:28,  2.32s/it]
                                                 
{'loss': '4.023', 'grad_norm': '0.1753', 'learning_rate': '0.0003', 'epoch': '6.81'}
+
 57%|█████▋    | 490/864 [22:04<14:28,  2.32s/it]
 57%|█████▋    | 491/864 [22:06<14:26,  2.32s/it]
 57%|█████▋    | 492/864 [22:08<14:23,  2.32s/it]
 57%|█████▋    | 493/864 [22:11<14:21,  2.32s/it]
 57%|█████▋    | 494/864 [22:13<14:19,  2.32s/it]
 57%|█████▋    | 495/864 [22:15<14:16,  2.32s/it]
 57%|█████▋    | 496/864 [22:18<14:14,  2.32s/it]
 58%|█████▊    | 497/864 [22:20<14:12,  2.32s/it]
 58%|█████▊    | 498/864 [22:22<14:09,  2.32s/it]
 58%|█████▊    | 499/864 [22:25<14:07,  2.32s/it]
 58%|█████▊    | 500/864 [22:27<14:05,  2.32s/it]
                                                 
{'loss': '3.974', 'grad_norm': '0.2048', 'learning_rate': '0.0003', 'epoch': '6.95'}
+
 58%|█████▊    | 500/864 [22:27<14:05,  2.32s/it]
 58%|█████▊    | 501/864 [22:29<14:03,  2.32s/it]
 58%|█████▊    | 502/864 [22:32<14:00,  2.32s/it]
 58%|█████▊    | 503/864 [22:34<14:09,  2.35s/it]
 58%|█████▊    | 504/864 [22:35<12:15,  2.04s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 4
+
+
  0%|          | 0/77 [00:00<?, ?it/s]
+
  4%|▍         | 3/77 [00:00<00:04, 14.93it/s]
+
  6%|▋         | 5/77 [00:00<00:06, 11.86it/s]
+
  9%|▉         | 7/77 [00:00<00:06, 10.91it/s]
+
 12%|█▏        | 9/77 [00:00<00:06, 10.46it/s]
+
 14%|█▍        | 11/77 [00:01<00:06, 10.21it/s]
+
 17%|█▋        | 13/77 [00:01<00:06, 10.06it/s]
+
 19%|█▉        | 15/77 [00:01<00:06,  9.97it/s]
+
 21%|██        | 16/77 [00:01<00:06,  9.92it/s]
+
 22%|██▏       | 17/77 [00:01<00:06,  9.90it/s]
+
 23%|██▎       | 18/77 [00:01<00:05,  9.87it/s]
+
 25%|██▍       | 19/77 [00:01<00:05,  9.85it/s]
+
 26%|██▌       | 20/77 [00:01<00:05,  9.83it/s]
+
 27%|██▋       | 21/77 [00:02<00:05,  9.81it/s]
+
 29%|██▊       | 22/77 [00:02<00:05,  9.80it/s]
+
 30%|██▉       | 23/77 [00:02<00:05,  9.79it/s]
+
 31%|███       | 24/77 [00:02<00:05,  9.78it/s]
+
 32%|███▏      | 25/77 [00:02<00:05,  9.78it/s]
+
 34%|███▍      | 26/77 [00:02<00:05,  9.78it/s]
+
 35%|███▌      | 27/77 [00:02<00:05,  9.78it/s]
+
 36%|███▋      | 28/77 [00:02<00:05,  9.77it/s]
+
 38%|███▊      | 29/77 [00:02<00:04,  9.77it/s]
+
 39%|███▉      | 30/77 [00:02<00:04,  9.77it/s]
+
 40%|████      | 31/77 [00:03<00:04,  9.78it/s]
+
 42%|████▏     | 32/77 [00:03<00:04,  9.77it/s]
+
 43%|████▎     | 33/77 [00:03<00:04,  9.77it/s]
+
 44%|████▍     | 34/77 [00:03<00:04,  9.77it/s]
+
 45%|████▌     | 35/77 [00:03<00:04,  9.78it/s]
+
 47%|████▋     | 36/77 [00:03<00:04,  9.77it/s]
+
 48%|████▊     | 37/77 [00:03<00:04,  9.78it/s]
+
 49%|████▉     | 38/77 [00:03<00:03,  9.77it/s]
+
 51%|█████     | 39/77 [00:03<00:03,  9.77it/s]
+
 52%|█████▏    | 40/77 [00:03<00:03,  9.75it/s]
+
 53%|█████▎    | 41/77 [00:04<00:03,  9.77it/s]
+
 55%|█████▍    | 42/77 [00:04<00:03,  9.77it/s]
+
 56%|█████▌    | 43/77 [00:04<00:03,  9.78it/s]
+
 57%|█████▋    | 44/77 [00:04<00:03,  9.78it/s]
+
 58%|█████▊    | 45/77 [00:04<00:03,  9.78it/s]
+
 60%|█████▉    | 46/77 [00:04<00:03,  9.77it/s]
+
 61%|██████    | 47/77 [00:04<00:03,  9.78it/s]
+
 62%|██████▏   | 48/77 [00:04<00:02,  9.78it/s]
+
 64%|██████▎   | 49/77 [00:04<00:02,  9.79it/s]
+
 65%|██████▍   | 50/77 [00:05<00:02,  9.78it/s]
+
 66%|██████▌   | 51/77 [00:05<00:02,  9.78it/s]
+
 68%|██████▊   | 52/77 [00:05<00:02,  9.79it/s]
+
 69%|██████▉   | 53/77 [00:05<00:02,  9.78it/s]
+
 70%|███████   | 54/77 [00:05<00:02,  9.78it/s]
+
 71%|███████▏  | 55/77 [00:05<00:02,  9.79it/s]
+
 73%|███████▎  | 56/77 [00:05<00:02,  9.78it/s]
+
 74%|███████▍  | 57/77 [00:05<00:02,  9.77it/s]
+
 75%|███████▌  | 58/77 [00:05<00:01,  9.77it/s]
+
 77%|███████▋  | 59/77 [00:05<00:01,  9.78it/s]
+
 78%|███████▊  | 60/77 [00:06<00:01,  9.77it/s]
+
 79%|███████▉  | 61/77 [00:06<00:01,  9.78it/s]
+
 81%|████████  | 62/77 [00:06<00:01,  9.78it/s]
+
 82%|████████▏ | 63/77 [00:06<00:01,  9.79it/s]
+
 83%|████████▎ | 64/77 [00:06<00:01,  9.79it/s]
+
 84%|████████▍ | 65/77 [00:06<00:01,  9.77it/s]
+
 86%|████████▌ | 66/77 [00:06<00:01,  9.79it/s]
+
 87%|████████▋ | 67/77 [00:06<00:01,  9.79it/s]
+
 88%|████████▊ | 68/77 [00:06<00:00,  9.79it/s]
+
 90%|████████▉ | 69/77 [00:06<00:00,  9.80it/s]
+
 91%|█████████ | 70/77 [00:07<00:00,  9.79it/s]
+
 92%|█████████▏| 71/77 [00:07<00:00,  9.80it/s]
+
 94%|█████████▎| 72/77 [00:07<00:00,  9.79it/s]
+
 95%|█████████▍| 73/77 [00:07<00:00,  9.79it/s]
+
 96%|█████████▌| 74/77 [00:07<00:00,  9.78it/s]
+
 97%|█████████▋| 75/77 [00:07<00:00,  9.80it/s]
+
 99%|█████████▊| 76/77 [00:07<00:00,  9.80it/s]
+
100%|██████████| 77/77 [00:07<00:00,  9.79it/s]
                                                 
+
                                               
{'eval_loss': '4.218', 'eval_runtime': '7.912', 'eval_samples_per_second': '308.5', 'eval_steps_per_second': '9.732', 'epoch': '7'}
+
 58%|█████▊    | 504/864 [22:43<12:15,  2.04s/it]
+
100%|██████████| 77/77 [00:07<00:00,  9.79it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0003/checkpoint-504
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0003/checkpoint-504/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0003/checkpoint-504/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:03<00:00,  3.56s/it]
Writing model shards: 100%|██████████| 1/1 [00:03<00:00,  3.56s/it]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0003/checkpoint-504/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0003/checkpoint-504/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0003/checkpoint-504/tokenizer_config.json
+[eval] epoch 7.0 step 504 loss 4.2179 (best 4.2179 @ ep 7.0, 0/10000 since best)
+
+
                                               
 58%|█████▊    | 505/864 [22:56<45:47,  7.65s/it]
 59%|█████▊    | 506/864 [22:58<36:07,  6.05s/it]
 59%|█████▊    | 507/864 [23:01<29:21,  4.94s/it]
 59%|█████▉    | 508/864 [23:03<24:38,  4.15s/it]
 59%|█████▉    | 509/864 [23:05<21:19,  3.60s/it]
 59%|█████▉    | 510/864 [23:08<18:59,  3.22s/it]
                                                 
{'loss': '3.887', 'grad_norm': '0.228', 'learning_rate': '0.0003', 'epoch': '7.084'}
+
 59%|█████▉    | 510/864 [23:08<18:59,  3.22s/it]
 59%|█████▉    | 511/864 [23:10<17:22,  2.95s/it]
 59%|█████▉    | 512/864 [23:12<16:12,  2.76s/it]
 59%|█████▉    | 513/864 [23:15<15:23,  2.63s/it]
 59%|█████▉    | 514/864 [23:17<14:48,  2.54s/it]
 60%|█████▉    | 515/864 [23:19<14:23,  2.47s/it]
 60%|█████▉    | 516/864 [23:22<14:04,  2.43s/it]
 60%|█████▉    | 517/864 [23:24<13:51,  2.40s/it]
 60%|█████▉    | 518/864 [23:26<13:41,  2.37s/it]
 60%|██████    | 519/864 [23:29<13:33,  2.36s/it]
 60%|██████    | 520/864 [23:31<13:27,  2.35s/it]
                                                 
{'loss': '3.829', 'grad_norm': '0.4324', 'learning_rate': '0.0003', 'epoch': '7.224'}
+
 60%|██████    | 520/864 [23:31<13:27,  2.35s/it]
 60%|██████    | 521/864 [23:33<13:23,  2.34s/it]
 60%|██████    | 522/864 [23:36<13:19,  2.34s/it]
 61%|██████    | 523/864 [23:38<13:15,  2.33s/it]
 61%|██████    | 524/864 [23:40<13:12,  2.33s/it]
 61%|██████    | 525/864 [23:43<13:09,  2.33s/it]
 61%|██████    | 526/864 [23:45<13:06,  2.33s/it]
 61%|██████    | 527/864 [23:47<13:03,  2.33s/it]
 61%|██████    | 528/864 [23:50<13:01,  2.33s/it]
 61%|██████    | 529/864 [23:52<12:59,  2.33s/it]
 61%|██████▏   | 530/864 [23:54<12:56,  2.33s/it]
                                                 
{'loss': '3.821', 'grad_norm': '0.2796', 'learning_rate': '0.0003', 'epoch': '7.363'}
+
 61%|██████▏   | 530/864 [23:54<12:56,  2.33s/it]
 61%|██████▏   | 531/864 [23:57<12:54,  2.32s/it]
 62%|██████▏   | 532/864 [23:59<12:51,  2.32s/it]
 62%|██████▏   | 533/864 [24:01<12:49,  2.32s/it]
 62%|██████▏   | 534/864 [24:04<12:46,  2.32s/it]
 62%|██████▏   | 535/864 [24:06<12:44,  2.32s/it]
 62%|██████▏   | 536/864 [24:08<12:41,  2.32s/it]
 62%|██████▏   | 537/864 [24:11<12:39,  2.32s/it]
 62%|██████▏   | 538/864 [24:13<12:37,  2.32s/it]
 62%|██████▏   | 539/864 [24:15<12:35,  2.32s/it]
 62%|██████▎   | 540/864 [24:18<12:32,  2.32s/it]
                                                 
{'loss': '3.807', 'grad_norm': '0.2597', 'learning_rate': '0.0003', 'epoch': '7.503'}
+
 62%|██████▎   | 540/864 [24:18<12:32,  2.32s/it]
 63%|██████▎   | 541/864 [24:20<12:31,  2.33s/it]
 63%|██████▎   | 542/864 [24:22<12:28,  2.32s/it]
 63%|██████▎   | 543/864 [24:24<12:26,  2.32s/it]
 63%|██████▎   | 544/864 [24:27<12:23,  2.32s/it]
 63%|██████▎   | 545/864 [24:29<12:21,  2.32s/it]
 63%|██████▎   | 546/864 [24:31<12:18,  2.32s/it]
 63%|██████▎   | 547/864 [24:34<12:16,  2.32s/it]
 63%|██████▎   | 548/864 [24:36<12:14,  2.32s/it]
 64%|██████▎   | 549/864 [24:38<12:12,  2.32s/it]
 64%|██████▎   | 550/864 [24:41<12:09,  2.32s/it]
                                                 
{'loss': '3.799', 'grad_norm': '0.2075', 'learning_rate': '0.0003', 'epoch': '7.643'}
+
 64%|██████▎   | 550/864 [24:41<12:09,  2.32s/it]
 64%|██████▍   | 551/864 [24:43<12:07,  2.32s/it]
 64%|██████▍   | 552/864 [24:45<12:07,  2.33s/it]
 64%|██████▍   | 553/864 [24:48<12:04,  2.33s/it]
 64%|██████▍   | 554/864 [24:50<12:01,  2.33s/it]
 64%|██████▍   | 555/864 [24:52<11:58,  2.33s/it]
 64%|██████▍   | 556/864 [24:55<11:56,  2.33s/it]
 64%|██████▍   | 557/864 [24:57<11:53,  2.32s/it]
 65%|██████▍   | 558/864 [24:59<11:51,  2.33s/it]
 65%|██████▍   | 559/864 [25:02<11:49,  2.33s/it]
 65%|██████▍   | 560/864 [25:04<11:47,  2.33s/it]
                                                 
{'loss': '3.789', 'grad_norm': '0.2211', 'learning_rate': '0.0003', 'epoch': '7.783'}
+
 65%|██████▍   | 560/864 [25:04<11:47,  2.33s/it]
 65%|██████▍   | 561/864 [25:06<11:44,  2.33s/it]
 65%|██████▌   | 562/864 [25:09<11:42,  2.33s/it]
 65%|██████▌   | 563/864 [25:11<11:39,  2.33s/it]
 65%|██████▌   | 564/864 [25:13<11:37,  2.33s/it]
 65%|██████▌   | 565/864 [25:16<11:35,  2.33s/it]
 66%|██████▌   | 566/864 [25:18<11:32,  2.32s/it]
 66%|██████▌   | 567/864 [25:20<11:30,  2.33s/it]
 66%|██████▌   | 568/864 [25:23<11:27,  2.32s/it]
 66%|██████▌   | 569/864 [25:25<11:25,  2.32s/it]
 66%|██████▌   | 570/864 [25:27<11:23,  2.32s/it]
                                                 
{'loss': '3.782', 'grad_norm': '0.2792', 'learning_rate': '0.0003', 'epoch': '7.922'}
+
 66%|██████▌   | 570/864 [25:27<11:23,  2.32s/it]
 66%|██████▌   | 571/864 [25:30<11:21,  2.32s/it]
 66%|██████▌   | 572/864 [25:32<11:18,  2.33s/it]
 66%|██████▋   | 573/864 [25:34<11:16,  2.32s/it]
 66%|██████▋   | 574/864 [25:37<11:14,  2.32s/it]
 67%|██████▋   | 575/864 [25:39<11:12,  2.33s/it]
 67%|██████▋   | 576/864 [25:40<09:43,  2.03s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 4
+
+
  0%|          | 0/77 [00:00<?, ?it/s]
+
  3%|▎         | 2/77 [00:00<00:03, 19.58it/s]
+
  5%|▌         | 4/77 [00:00<00:05, 12.29it/s]
+
  8%|▊         | 6/77 [00:00<00:06, 10.99it/s]
+
 10%|█         | 8/77 [00:00<00:06, 10.48it/s]
+
 13%|█▎        | 10/77 [00:00<00:06, 10.22it/s]
+
 16%|█▌        | 12/77 [00:01<00:06, 10.06it/s]
+
 18%|█▊        | 14/77 [00:01<00:06,  9.97it/s]
+
 21%|██        | 16/77 [00:01<00:06,  9.91it/s]
+
 23%|██▎       | 18/77 [00:01<00:05,  9.86it/s]
+
 25%|██▍       | 19/77 [00:01<00:05,  9.85it/s]
+
 26%|██▌       | 20/77 [00:01<00:05,  9.83it/s]
+
 27%|██▋       | 21/77 [00:02<00:05,  9.81it/s]
+
 29%|██▊       | 22/77 [00:02<00:05,  9.81it/s]
+
 30%|██▉       | 23/77 [00:02<00:05,  9.80it/s]
+
 31%|███       | 24/77 [00:02<00:05,  9.80it/s]
+
 32%|███▏      | 25/77 [00:02<00:05,  9.79it/s]
+
 34%|███▍      | 26/77 [00:02<00:05,  9.78it/s]
+
 35%|███▌      | 27/77 [00:02<00:05,  9.78it/s]
+
 36%|███▋      | 28/77 [00:02<00:05,  9.78it/s]
+
 38%|███▊      | 29/77 [00:02<00:04,  9.77it/s]
+
 39%|███▉      | 30/77 [00:02<00:04,  9.79it/s]
+
 40%|████      | 31/77 [00:03<00:04,  9.79it/s]
+
 42%|████▏     | 32/77 [00:03<00:04,  9.78it/s]
+
 43%|████▎     | 33/77 [00:03<00:04,  9.78it/s]
+
 44%|████▍     | 34/77 [00:03<00:04,  9.78it/s]
+
 45%|████▌     | 35/77 [00:03<00:04,  9.78it/s]
+
 47%|████▋     | 36/77 [00:03<00:04,  9.78it/s]
+
 48%|████▊     | 37/77 [00:03<00:04,  9.78it/s]
+
 49%|████▉     | 38/77 [00:03<00:03,  9.77it/s]
+
 51%|█████     | 39/77 [00:03<00:03,  9.78it/s]
+
 52%|█████▏    | 40/77 [00:03<00:03,  9.78it/s]
+
 53%|█████▎    | 41/77 [00:04<00:03,  9.78it/s]
+
 55%|█████▍    | 42/77 [00:04<00:03,  9.78it/s]
+
 56%|█████▌    | 43/77 [00:04<00:03,  9.78it/s]
+
 57%|█████▋    | 44/77 [00:04<00:03,  9.78it/s]
+
 58%|█████▊    | 45/77 [00:04<00:03,  9.77it/s]
+
 60%|█████▉    | 46/77 [00:04<00:03,  9.79it/s]
+
 61%|██████    | 47/77 [00:04<00:03,  9.78it/s]
+
 62%|██████▏   | 48/77 [00:04<00:02,  9.78it/s]
+
 64%|██████▎   | 49/77 [00:04<00:02,  9.78it/s]
+
 65%|██████▍   | 50/77 [00:05<00:02,  9.78it/s]
+
 66%|██████▌   | 51/77 [00:05<00:02,  9.78it/s]
+
 68%|██████▊   | 52/77 [00:05<00:02,  9.78it/s]
+
 69%|██████▉   | 53/77 [00:05<00:02,  9.77it/s]
+
 70%|███████   | 54/77 [00:05<00:02,  9.78it/s]
+
 71%|███████▏  | 55/77 [00:05<00:02,  9.77it/s]
+
 73%|███████▎  | 56/77 [00:05<00:02,  9.79it/s]
+
 74%|███████▍  | 57/77 [00:05<00:02,  9.78it/s]
+
 75%|███████▌  | 58/77 [00:05<00:01,  9.77it/s]
+
 77%|███████▋  | 59/77 [00:05<00:01,  9.78it/s]
+
 78%|███████▊  | 60/77 [00:06<00:01,  9.78it/s]
+
 79%|███████▉  | 61/77 [00:06<00:01,  9.79it/s]
+
 81%|████████  | 62/77 [00:06<00:01,  9.79it/s]
+
 82%|████████▏ | 63/77 [00:06<00:01,  9.79it/s]
+
 83%|████████▎ | 64/77 [00:06<00:01,  9.79it/s]
+
 84%|████████▍ | 65/77 [00:06<00:01,  9.79it/s]
+
 86%|████████▌ | 66/77 [00:06<00:01,  9.79it/s]
+
 87%|████████▋ | 67/77 [00:06<00:01,  9.79it/s]
+
 88%|████████▊ | 68/77 [00:06<00:00,  9.79it/s]
+
 90%|████████▉ | 69/77 [00:06<00:00,  9.78it/s]
+
 91%|█████████ | 70/77 [00:07<00:00,  9.78it/s]
+
 92%|█████████▏| 71/77 [00:07<00:00,  9.78it/s]
+
 94%|█████████▎| 72/77 [00:07<00:00,  9.77it/s]
+
 95%|█████████▍| 73/77 [00:07<00:00,  9.78it/s]
+
 96%|█████████▌| 74/77 [00:07<00:00,  9.78it/s]
+
 97%|█████████▋| 75/77 [00:07<00:00,  9.79it/s]
+
 99%|█████████▊| 76/77 [00:07<00:00,  9.78it/s]
+
100%|██████████| 77/77 [00:07<00:00,  9.77it/s]
                                                 
+
                                               
{'eval_loss': '4.122', 'eval_runtime': '7.912', 'eval_samples_per_second': '308.5', 'eval_steps_per_second': '9.732', 'epoch': '8'}
+
 67%|██████▋   | 576/864 [25:48<09:43,  2.03s/it]
+
100%|██████████| 77/77 [00:07<00:00,  9.77it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0003/checkpoint-576
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0003/checkpoint-576/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0003/checkpoint-576/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:03<00:00,  3.51s/it]
Writing model shards: 100%|██████████| 1/1 [00:03<00:00,  3.51s/it]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0003/checkpoint-576/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0003/checkpoint-576/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0003/checkpoint-576/tokenizer_config.json
+[eval] epoch 8.0 step 576 loss 4.1220 (best 4.1220 @ ep 8.0, 0/10000 since best)
+
+
                                               
 67%|██████▋   | 577/864 [26:01<36:35,  7.65s/it]
 67%|██████▋   | 578/864 [26:03<28:50,  6.05s/it]
 67%|██████▋   | 579/864 [26:06<23:26,  4.93s/it]
 67%|██████▋   | 580/864 [26:08<19:38,  4.15s/it]
                                                 
{'loss': '3.709', 'grad_norm': '0.6975', 'learning_rate': '0.0003', 'epoch': '8.056'}
+
 67%|██████▋   | 580/864 [26:08<19:38,  4.15s/it]
 67%|██████▋   | 581/864 [26:10<16:59,  3.60s/it]
 67%|██████▋   | 582/864 [26:13<15:07,  3.22s/it]
 67%|██████▋   | 583/864 [26:15<13:49,  2.95s/it]
 68%|██████▊   | 584/864 [26:17<12:53,  2.76s/it]
 68%|██████▊   | 585/864 [26:20<12:13,  2.63s/it]
 68%|██████▊   | 586/864 [26:22<11:45,  2.54s/it]
 68%|██████▊   | 587/864 [26:24<11:25,  2.47s/it]
 68%|██████▊   | 588/864 [26:27<11:10,  2.43s/it]
 68%|██████▊   | 589/864 [26:29<10:59,  2.40s/it]
 68%|██████▊   | 590/864 [26:31<10:51,  2.38s/it]
                                                 
{'loss': '3.746', 'grad_norm': '0.8549', 'learning_rate': '0.0003', 'epoch': '8.196'}
+
 68%|██████▊   | 590/864 [26:31<10:51,  2.38s/it]
 68%|██████▊   | 591/864 [26:34<10:44,  2.36s/it]
 69%|██████▊   | 592/864 [26:36<10:39,  2.35s/it]
 69%|██████▊   | 593/864 [26:38<10:34,  2.34s/it]
 69%|██████▉   | 594/864 [26:41<10:31,  2.34s/it]
 69%|██████▉   | 595/864 [26:43<10:27,  2.33s/it]
 69%|██████▉   | 596/864 [26:45<10:25,  2.33s/it]
 69%|██████▉   | 597/864 [26:47<10:21,  2.33s/it]
 69%|██████▉   | 598/864 [26:50<10:18,  2.33s/it]
 69%|██████▉   | 599/864 [26:52<10:16,  2.33s/it]
 69%|██████▉   | 600/864 [26:54<10:13,  2.32s/it]
                                                 
{'loss': '3.828', 'grad_norm': '0.8407', 'learning_rate': '0.0003', 'epoch': '8.335'}
+
 69%|██████▉   | 600/864 [26:54<10:13,  2.32s/it]
 70%|██████▉   | 601/864 [26:57<10:11,  2.33s/it]
 70%|██████▉   | 602/864 [26:59<10:09,  2.32s/it]
 70%|██████▉   | 603/864 [27:01<10:06,  2.32s/it]
 70%|██████▉   | 604/864 [27:04<10:04,  2.32s/it]
 70%|███████   | 605/864 [27:06<10:01,  2.32s/it]
 70%|███████   | 606/864 [27:08<09:59,  2.32s/it]
 70%|███████   | 607/864 [27:11<09:57,  2.32s/it]
 70%|███████   | 608/864 [27:13<09:55,  2.32s/it]
 70%|███████   | 609/864 [27:15<09:52,  2.33s/it]
 71%|███████   | 610/864 [27:18<09:50,  2.32s/it]
                                                 
{'loss': '3.771', 'grad_norm': '0.4354', 'learning_rate': '0.0003', 'epoch': '8.475'}
+
 71%|███████   | 610/864 [27:18<09:50,  2.32s/it]
 71%|███████   | 611/864 [27:20<09:48,  2.33s/it]
 71%|███████   | 612/864 [27:22<09:45,  2.32s/it]
 71%|███████   | 613/864 [27:25<09:43,  2.32s/it]
 71%|███████   | 614/864 [27:27<09:40,  2.32s/it]
 71%|███████   | 615/864 [27:29<09:38,  2.32s/it]
 71%|███████▏  | 616/864 [27:32<09:36,  2.32s/it]
 71%|███████▏  | 617/864 [27:34<09:34,  2.32s/it]
 72%|███████▏  | 618/864 [27:36<09:31,  2.32s/it]
 72%|███████▏  | 619/864 [27:39<09:29,  2.33s/it]
 72%|███████▏  | 620/864 [27:41<09:27,  2.33s/it]
                                                 
{'loss': '3.684', 'grad_norm': '0.223', 'learning_rate': '0.0003', 'epoch': '8.615'}
+
 72%|███████▏  | 620/864 [27:41<09:27,  2.33s/it]
 72%|███████▏  | 621/864 [27:43<09:25,  2.33s/it]
 72%|███████▏  | 622/864 [27:46<09:48,  2.43s/it]
 72%|███████▏  | 623/864 [27:48<09:38,  2.40s/it]
 72%|███████▏  | 624/864 [27:51<09:30,  2.38s/it]
 72%|███████▏  | 625/864 [27:53<09:24,  2.36s/it]
 72%|███████▏  | 626/864 [27:55<09:19,  2.35s/it]
 73%|███████▎  | 627/864 [27:58<09:15,  2.34s/it]
 73%|███████▎  | 628/864 [28:00<09:11,  2.34s/it]
 73%|███████▎  | 629/864 [28:02<09:08,  2.33s/it]
 73%|███████▎  | 630/864 [28:05<09:05,  2.33s/it]
                                                 
{'loss': '3.642', 'grad_norm': '0.207', 'learning_rate': '0.0003', 'epoch': '8.755'}
+
 73%|███████▎  | 630/864 [28:05<09:05,  2.33s/it]
 73%|███████▎  | 631/864 [28:07<09:02,  2.33s/it]
 73%|███████▎  | 632/864 [28:09<09:00,  2.33s/it]
 73%|███████▎  | 633/864 [28:12<08:57,  2.33s/it]
 73%|███████▎  | 634/864 [28:14<09:20,  2.44s/it]
 73%|███████▎  | 635/864 [28:17<09:35,  2.51s/it]
 74%|███████▎  | 636/864 [28:19<09:20,  2.46s/it]
 74%|███████▎  | 637/864 [28:22<09:08,  2.42s/it]
 74%|███████▍  | 638/864 [28:24<09:24,  2.50s/it]
 74%|███████▍  | 639/864 [28:27<09:10,  2.44s/it]
 74%|███████▍  | 640/864 [28:29<09:25,  2.52s/it]
                                                 
{'loss': '3.633', 'grad_norm': '0.202', 'learning_rate': '0.0003', 'epoch': '8.894'}
+
 74%|███████▍  | 640/864 [28:29<09:25,  2.52s/it]
 74%|███████▍  | 641/864 [28:32<09:08,  2.46s/it]
 74%|███████▍  | 642/864 [28:34<08:57,  2.42s/it]
 74%|███████▍  | 643/864 [28:37<09:13,  2.50s/it]
 75%|███████▍  | 644/864 [28:39<09:22,  2.56s/it]
 75%|███████▍  | 645/864 [28:42<09:04,  2.49s/it]
 75%|███████▍  | 646/864 [28:44<08:51,  2.44s/it]
 75%|███████▍  | 647/864 [28:46<08:41,  2.40s/it]
 75%|███████▌  | 648/864 [28:48<07:29,  2.08s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 4
+
+
  0%|          | 0/77 [00:00<?, ?it/s]
+
  4%|▍         | 3/77 [00:00<00:04, 14.98it/s]
+
  6%|▋         | 5/77 [00:00<00:06, 11.89it/s]
+
  9%|▉         | 7/77 [00:00<00:06, 10.92it/s]
+
 12%|█▏        | 9/77 [00:00<00:06, 10.47it/s]
+
 14%|█▍        | 11/77 [00:01<00:06, 10.21it/s]
+
 17%|█▋        | 13/77 [00:01<00:06, 10.07it/s]
+
 19%|█▉        | 15/77 [00:01<00:06,  9.98it/s]
+
 21%|██        | 16/77 [00:01<00:06,  9.94it/s]
+
 22%|██▏       | 17/77 [00:01<00:06,  9.91it/s]
+
 23%|██▎       | 18/77 [00:01<00:05,  9.88it/s]
+
 25%|██▍       | 19/77 [00:01<00:05,  9.85it/s]
+
 26%|██▌       | 20/77 [00:01<00:05,  9.83it/s]
+
 27%|██▋       | 21/77 [00:02<00:05,  9.82it/s]
+
 29%|██▊       | 22/77 [00:02<00:05,  9.81it/s]
+
 30%|██▉       | 23/77 [00:02<00:05,  9.80it/s]
+
 31%|███       | 24/77 [00:02<00:05,  9.80it/s]
+
 32%|███▏      | 25/77 [00:02<00:05,  9.79it/s]
+
 34%|███▍      | 26/77 [00:02<00:05,  9.78it/s]
+
 35%|███▌      | 27/77 [00:02<00:05,  9.78it/s]
+
 36%|███▋      | 28/77 [00:02<00:05,  9.79it/s]
+
 38%|███▊      | 29/77 [00:02<00:04,  9.79it/s]
+
 39%|███▉      | 30/77 [00:02<00:04,  9.78it/s]
+
 40%|████      | 31/77 [00:03<00:04,  9.79it/s]
+
 42%|████▏     | 32/77 [00:03<00:04,  9.78it/s]
+
 43%|████▎     | 33/77 [00:03<00:04,  9.78it/s]
+
 44%|████▍     | 34/77 [00:03<00:04,  9.78it/s]
+
 45%|████▌     | 35/77 [00:03<00:04,  9.75it/s]
+
 47%|████▋     | 36/77 [00:03<00:04,  9.78it/s]
+
 48%|████▊     | 37/77 [00:03<00:04,  9.77it/s]
+
 49%|████▉     | 38/77 [00:03<00:03,  9.77it/s]
+
 51%|█████     | 39/77 [00:03<00:03,  9.79it/s]
+
 52%|█████▏    | 40/77 [00:03<00:03,  9.79it/s]
+
 53%|█████▎    | 41/77 [00:04<00:03,  9.78it/s]
+
 55%|█████▍    | 42/77 [00:04<00:03,  9.78it/s]
+
 56%|█████▌    | 43/77 [00:04<00:03,  9.79it/s]
+
 57%|█████▋    | 44/77 [00:04<00:03,  9.78it/s]
+
 58%|█████▊    | 45/77 [00:04<00:03,  9.78it/s]
+
 60%|█████▉    | 46/77 [00:04<00:03,  9.78it/s]
+
 61%|██████    | 47/77 [00:04<00:03,  9.78it/s]
+
 62%|██████▏   | 48/77 [00:04<00:02,  9.79it/s]
+
 64%|██████▎   | 49/77 [00:04<00:02,  9.78it/s]
+
 65%|██████▍   | 50/77 [00:05<00:02,  9.78it/s]
+
 66%|██████▌   | 51/77 [00:05<00:02,  9.78it/s]
+
 68%|██████▊   | 52/77 [00:05<00:02,  9.77it/s]
+
 69%|██████▉   | 53/77 [00:05<00:02,  9.77it/s]
+
 70%|███████   | 54/77 [00:05<00:02,  9.77it/s]
+
 71%|███████▏  | 55/77 [00:05<00:02,  9.78it/s]
+
 73%|███████▎  | 56/77 [00:05<00:02,  9.78it/s]
+
 74%|███████▍  | 57/77 [00:05<00:02,  9.78it/s]
+
 75%|███████▌  | 58/77 [00:05<00:01,  9.78it/s]
+
 77%|███████▋  | 59/77 [00:05<00:01,  9.78it/s]
+
 78%|███████▊  | 60/77 [00:06<00:01,  9.76it/s]
+
 79%|███████▉  | 61/77 [00:06<00:01,  9.79it/s]
+
 81%|████████  | 62/77 [00:06<00:01,  9.80it/s]
+
 82%|████████▏ | 63/77 [00:06<00:01,  9.79it/s]
+
 83%|████████▎ | 64/77 [00:06<00:01,  9.78it/s]
+
 84%|████████▍ | 65/77 [00:06<00:01,  9.78it/s]
+
 86%|████████▌ | 66/77 [00:06<00:01,  9.77it/s]
+
 87%|████████▋ | 67/77 [00:06<00:01,  9.79it/s]
+
 88%|████████▊ | 68/77 [00:06<00:00,  9.79it/s]
+
 90%|████████▉ | 69/77 [00:06<00:00,  9.79it/s]
+
 91%|█████████ | 70/77 [00:07<00:00,  9.79it/s]
+
 92%|█████████▏| 71/77 [00:07<00:00,  9.79it/s]
+
 94%|█████████▎| 72/77 [00:07<00:00,  9.79it/s]
+
 95%|█████████▍| 73/77 [00:07<00:00,  9.79it/s]
+
 96%|█████████▌| 74/77 [00:07<00:00,  9.79it/s]
+
 97%|█████████▋| 75/77 [00:07<00:00,  9.80it/s]
+
 99%|█████████▊| 76/77 [00:07<00:00,  9.79it/s]
+
100%|██████████| 77/77 [00:07<00:00,  9.78it/s]
                                                 
+
                                               
{'eval_loss': '4.079', 'eval_runtime': '7.904', 'eval_samples_per_second': '308.8', 'eval_steps_per_second': '9.742', 'epoch': '9'}
+
 75%|███████▌  | 648/864 [28:55<07:29,  2.08s/it]
+
100%|██████████| 77/77 [00:07<00:00,  9.78it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0003/checkpoint-648
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0003/checkpoint-648/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0003/checkpoint-648/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:02<00:00,  2.57s/it]
Writing model shards: 100%|██████████| 1/1 [00:02<00:00,  2.57s/it]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0003/checkpoint-648/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0003/checkpoint-648/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0003/checkpoint-648/tokenizer_config.json
+[eval] epoch 9.0 step 648 loss 4.0788 (best 4.0788 @ ep 9.0, 0/10000 since best)
+
+
                                               
 75%|███████▌  | 649/864 [29:07<26:23,  7.36s/it]
 75%|███████▌  | 650/864 [29:10<20:51,  5.85s/it]
                                                 
{'loss': '3.577', 'grad_norm': '0.3819', 'learning_rate': '0.0003', 'epoch': '9.028'}
+
 75%|███████▌  | 650/864 [29:10<20:51,  5.85s/it]
 75%|███████▌  | 651/864 [29:12<17:00,  4.79s/it]
 75%|███████▌  | 652/864 [29:14<14:18,  4.05s/it]
 76%|███████▌  | 653/864 [29:17<12:24,  3.53s/it]
 76%|███████▌  | 654/864 [29:19<11:05,  3.17s/it]
 76%|███████▌  | 655/864 [29:21<10:08,  2.91s/it]
 76%|███████▌  | 656/864 [29:24<09:28,  2.73s/it]
 76%|███████▌  | 657/864 [29:26<09:00,  2.61s/it]
 76%|███████▌  | 658/864 [29:28<08:39,  2.52s/it]
 76%|███████▋  | 659/864 [29:30<08:24,  2.46s/it]
 76%|███████▋  | 660/864 [29:33<08:13,  2.42s/it]
                                                 
{'loss': '3.442', 'grad_norm': '0.6515', 'learning_rate': '0.0003', 'epoch': '9.168'}
+
 76%|███████▋  | 660/864 [29:33<08:13,  2.42s/it]
 77%|███████▋  | 661/864 [29:35<08:05,  2.39s/it]
 77%|███████▋  | 662/864 [29:37<07:58,  2.37s/it]
 77%|███████▋  | 663/864 [29:40<07:53,  2.35s/it]
 77%|███████▋  | 664/864 [29:42<07:48,  2.34s/it]
 77%|███████▋  | 665/864 [29:44<07:44,  2.34s/it]
 77%|███████▋  | 666/864 [29:47<07:41,  2.33s/it]
 77%|███████▋  | 667/864 [29:49<07:38,  2.33s/it]
 77%|███████▋  | 668/864 [29:51<07:35,  2.33s/it]
 77%|███████▋  | 669/864 [29:54<07:33,  2.32s/it]
 78%|███████▊  | 670/864 [29:56<07:30,  2.32s/it]
                                                 
{'loss': '3.466', 'grad_norm': '0.4133', 'learning_rate': '0.0003', 'epoch': '9.307'}
+
 78%|███████▊  | 670/864 [29:56<07:30,  2.32s/it]
 78%|███████▊  | 671/864 [29:58<07:28,  2.32s/it]
 78%|███████▊  | 672/864 [30:01<07:25,  2.32s/it]
 78%|███████▊  | 673/864 [30:03<07:23,  2.32s/it]
 78%|███████▊  | 674/864 [30:05<07:21,  2.32s/it]
 78%|███████▊  | 675/864 [30:08<07:18,  2.32s/it]
 78%|███████▊  | 676/864 [30:10<07:16,  2.32s/it]
 78%|███████▊  | 677/864 [30:12<07:13,  2.32s/it]
 78%|███████▊  | 678/864 [30:15<07:11,  2.32s/it]
 79%|███████▊  | 679/864 [30:17<07:09,  2.32s/it]
 79%|███████▊  | 680/864 [30:19<07:06,  2.32s/it]
                                                 
{'loss': '3.455', 'grad_norm': '0.3088', 'learning_rate': '0.0003', 'epoch': '9.447'}
+
 79%|███████▊  | 680/864 [30:19<07:06,  2.32s/it]
 79%|███████▉  | 681/864 [30:22<07:04,  2.32s/it]
 79%|███████▉  | 682/864 [30:24<07:02,  2.32s/it]
 79%|███████▉  | 683/864 [30:26<06:59,  2.32s/it]
 79%|███████▉  | 684/864 [30:28<06:57,  2.32s/it]
 79%|███████▉  | 685/864 [30:31<06:55,  2.32s/it]
 79%|███████▉  | 686/864 [30:33<06:53,  2.32s/it]
 80%|███████▉  | 687/864 [30:35<06:50,  2.32s/it]
 80%|███████▉  | 688/864 [30:38<06:48,  2.32s/it]
 80%|███████▉  | 689/864 [30:40<06:46,  2.32s/it]
 80%|███████▉  | 690/864 [30:42<06:43,  2.32s/it]
                                                 
{'loss': '3.432', 'grad_norm': '0.24', 'learning_rate': '0.0003', 'epoch': '9.587'}
+
 80%|███████▉  | 690/864 [30:42<06:43,  2.32s/it]
 80%|███████▉  | 691/864 [30:45<06:41,  2.32s/it]
 80%|████████  | 692/864 [30:47<06:39,  2.32s/it]
 80%|████████  | 693/864 [30:49<06:37,  2.32s/it]
 80%|████████  | 694/864 [30:52<06:34,  2.32s/it]
 80%|████████  | 695/864 [30:54<06:32,  2.32s/it]
 81%|████████  | 696/864 [30:56<06:30,  2.32s/it]
 81%|████████  | 697/864 [30:59<06:27,  2.32s/it]
 81%|████████  | 698/864 [31:01<06:25,  2.32s/it]
 81%|████████  | 699/864 [31:03<06:23,  2.32s/it]
 81%|████████  | 700/864 [31:06<06:20,  2.32s/it]
                                                 
{'loss': '3.444', 'grad_norm': '0.2605', 'learning_rate': '0.0003', 'epoch': '9.727'}
+
 81%|████████  | 700/864 [31:06<06:20,  2.32s/it]
 81%|████████  | 701/864 [31:08<06:18,  2.32s/it]
 81%|████████▏ | 702/864 [31:10<06:16,  2.32s/it]
 81%|████████▏ | 703/864 [31:13<06:13,  2.32s/it]
 81%|████████▏ | 704/864 [31:15<06:11,  2.32s/it]
 82%|████████▏ | 705/864 [31:17<06:09,  2.32s/it]
 82%|████████▏ | 706/864 [31:20<06:06,  2.32s/it]
 82%|████████▏ | 707/864 [31:22<06:04,  2.32s/it]
 82%|████████▏ | 708/864 [31:24<06:02,  2.32s/it]
 82%|████████▏ | 709/864 [31:27<05:59,  2.32s/it]
 82%|████████▏ | 710/864 [31:29<05:57,  2.32s/it]
                                                 
{'loss': '3.435', 'grad_norm': '0.2416', 'learning_rate': '0.0003', 'epoch': '9.866'}
+
 82%|████████▏ | 710/864 [31:29<05:57,  2.32s/it]
 82%|████████▏ | 711/864 [31:31<05:55,  2.32s/it]
 82%|████████▏ | 712/864 [31:33<05:53,  2.32s/it]
 83%|████████▎ | 713/864 [31:36<05:50,  2.32s/it]
 83%|████████▎ | 714/864 [31:38<05:48,  2.32s/it]
 83%|████████▎ | 715/864 [31:40<05:46,  2.32s/it]
 83%|████████▎ | 716/864 [31:43<05:43,  2.32s/it]
 83%|████████▎ | 717/864 [31:45<05:41,  2.32s/it]
 83%|████████▎ | 718/864 [31:47<05:39,  2.32s/it]
 83%|████████▎ | 719/864 [31:50<05:36,  2.32s/it]
 83%|████████▎ | 720/864 [31:51<04:51,  2.02s/it]
                                                 
{'loss': '3.428', 'grad_norm': '0.2908', 'learning_rate': '0.0003', 'epoch': '10'}
+
 83%|████████▎ | 720/864 [31:51<04:51,  2.02s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 4
+
+
  0%|          | 0/77 [00:00<?, ?it/s]
+
  3%|▎         | 2/77 [00:00<00:03, 19.67it/s]
+
  5%|▌         | 4/77 [00:00<00:05, 12.26it/s]
+
  8%|▊         | 6/77 [00:00<00:06, 11.02it/s]
+
 10%|█         | 8/77 [00:00<00:06, 10.49it/s]
+
 13%|█▎        | 10/77 [00:00<00:06, 10.23it/s]
+
 16%|█▌        | 12/77 [00:01<00:06, 10.07it/s]
+
 18%|█▊        | 14/77 [00:01<00:06,  9.97it/s]
+
 21%|██        | 16/77 [00:01<00:06,  9.92it/s]
+
 23%|██▎       | 18/77 [00:01<00:05,  9.88it/s]
+
 25%|██▍       | 19/77 [00:01<00:05,  9.86it/s]
+
 26%|██▌       | 20/77 [00:01<00:05,  9.84it/s]
+
 27%|██▋       | 21/77 [00:02<00:05,  9.82it/s]
+
 29%|██▊       | 22/77 [00:02<00:05,  9.82it/s]
+
 30%|██▉       | 23/77 [00:02<00:05,  9.81it/s]
+
 31%|███       | 24/77 [00:02<00:05,  9.79it/s]
+
 32%|███▏      | 25/77 [00:02<00:05,  9.80it/s]
+
 34%|███▍      | 26/77 [00:02<00:05,  9.79it/s]
+
 35%|███▌      | 27/77 [00:02<00:05,  9.78it/s]
+
 36%|███▋      | 28/77 [00:02<00:05,  9.79it/s]
+
 38%|███▊      | 29/77 [00:02<00:04,  9.77it/s]
+
 39%|███▉      | 30/77 [00:02<00:04,  9.78it/s]
+
 40%|████      | 31/77 [00:03<00:04,  9.78it/s]
+
 42%|████▏     | 32/77 [00:03<00:04,  9.78it/s]
+
 43%|████▎     | 33/77 [00:03<00:04,  9.77it/s]
+
 44%|████▍     | 34/77 [00:03<00:04,  9.77it/s]
+
 45%|████▌     | 35/77 [00:03<00:04,  9.77it/s]
+
 47%|████▋     | 36/77 [00:03<00:04,  9.77it/s]
+
 48%|████▊     | 37/77 [00:03<00:04,  9.77it/s]
+
 49%|████▉     | 38/77 [00:03<00:03,  9.77it/s]
+
 51%|█████     | 39/77 [00:03<00:03,  9.78it/s]
+
 52%|█████▏    | 40/77 [00:03<00:03,  9.78it/s]
+
 53%|█████▎    | 41/77 [00:04<00:03,  9.76it/s]
+
 55%|█████▍    | 42/77 [00:04<00:03,  9.77it/s]
+
 56%|█████▌    | 43/77 [00:04<00:03,  9.78it/s]
+
 57%|█████▋    | 44/77 [00:04<00:03,  9.78it/s]
+
 58%|█████▊    | 45/77 [00:04<00:03,  9.78it/s]
+
 60%|█████▉    | 46/77 [00:04<00:03,  9.78it/s]
+
 61%|██████    | 47/77 [00:04<00:03,  9.78it/s]
+
 62%|██████▏   | 48/77 [00:04<00:02,  9.78it/s]
+
 64%|██████▎   | 49/77 [00:04<00:02,  9.78it/s]
+
 65%|██████▍   | 50/77 [00:05<00:02,  9.77it/s]
+
 66%|██████▌   | 51/77 [00:05<00:02,  9.78it/s]
+
 68%|██████▊   | 52/77 [00:05<00:02,  9.77it/s]
+
 69%|██████▉   | 53/77 [00:05<00:02,  9.75it/s]
+
 70%|███████   | 54/77 [00:05<00:02,  9.77it/s]
+
 71%|███████▏  | 55/77 [00:05<00:02,  9.77it/s]
+
 73%|███████▎  | 56/77 [00:05<00:02,  9.77it/s]
+
 74%|███████▍  | 57/77 [00:05<00:02,  9.78it/s]
+
 75%|███████▌  | 58/77 [00:05<00:01,  9.77it/s]
+
 77%|███████▋  | 59/77 [00:05<00:01,  9.78it/s]
+
 78%|███████▊  | 60/77 [00:06<00:01,  9.77it/s]
+
 79%|███████▉  | 61/77 [00:06<00:01,  9.80it/s]
+
 81%|████████  | 62/77 [00:06<00:01,  9.79it/s]
+
 82%|████████▏ | 63/77 [00:06<00:01,  9.79it/s]
+
 83%|████████▎ | 64/77 [00:06<00:01,  9.79it/s]
+
 84%|████████▍ | 65/77 [00:06<00:01,  9.79it/s]
+
 86%|████████▌ | 66/77 [00:06<00:01,  9.78it/s]
+
 87%|████████▋ | 67/77 [00:06<00:01,  9.79it/s]
+
 88%|████████▊ | 68/77 [00:06<00:00,  9.78it/s]
+
 90%|████████▉ | 69/77 [00:06<00:00,  9.78it/s]
+
 91%|█████████ | 70/77 [00:07<00:00,  9.78it/s]
+
 92%|█████████▏| 71/77 [00:07<00:00,  9.79it/s]
+
 94%|█████████▎| 72/77 [00:07<00:00,  9.78it/s]
+
 95%|█████████▍| 73/77 [00:07<00:00,  9.79it/s]
+
 96%|█████████▌| 74/77 [00:07<00:00,  9.79it/s]
+
 97%|█████████▋| 75/77 [00:07<00:00,  9.79it/s]
+
 99%|█████████▊| 76/77 [00:07<00:00,  9.79it/s]
+
100%|██████████| 77/77 [00:07<00:00,  9.79it/s]
                                                 
+
                                               
{'eval_loss': '4.061', 'eval_runtime': '7.911', 'eval_samples_per_second': '308.6', 'eval_steps_per_second': '9.733', 'epoch': '10'}
+
 83%|████████▎ | 720/864 [31:59<04:51,  2.02s/it]
+
100%|██████████| 77/77 [00:07<00:00,  9.79it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0003/checkpoint-720
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0003/checkpoint-720/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0003/checkpoint-720/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:02<00:00,  2.60s/it]
Writing model shards: 100%|██████████| 1/1 [00:02<00:00,  2.60s/it]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0003/checkpoint-720/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0003/checkpoint-720/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0003/checkpoint-720/tokenizer_config.json
+[eval] epoch 10.0 step 720 loss 4.0612 (best 4.0612 @ ep 10.0, 0/10000 since best)
+
+
                                               
 83%|████████▎ | 721/864 [32:11<17:34,  7.37s/it]
 84%|████████▎ | 722/864 [32:13<13:51,  5.86s/it]
 84%|████████▎ | 723/864 [32:16<11:16,  4.80s/it]
 84%|████████▍ | 724/864 [32:18<09:27,  4.05s/it]
 84%|████████▍ | 725/864 [32:20<08:11,  3.53s/it]
 84%|████████▍ | 726/864 [32:23<07:17,  3.17s/it]
 84%|████████▍ | 727/864 [32:25<06:40,  2.92s/it]
 84%|████████▍ | 728/864 [32:27<06:12,  2.74s/it]
 84%|████████▍ | 729/864 [32:30<05:53,  2.62s/it]
 84%|████████▍ | 730/864 [32:32<05:38,  2.53s/it]
                                                 
{'loss': '3.237', 'grad_norm': '1.045', 'learning_rate': '0.0003', 'epoch': '10.14'}
+
 84%|████████▍ | 730/864 [32:32<05:38,  2.53s/it]
 85%|████████▍ | 731/864 [32:34<05:27,  2.46s/it]
 85%|████████▍ | 732/864 [32:36<05:19,  2.42s/it]
 85%|████████▍ | 733/864 [32:39<05:13,  2.39s/it]
 85%|████████▍ | 734/864 [32:41<05:08,  2.37s/it]
 85%|████████▌ | 735/864 [32:43<05:03,  2.36s/it]
 85%|████████▌ | 736/864 [32:46<05:00,  2.35s/it]
 85%|████████▌ | 737/864 [32:48<04:57,  2.34s/it]
 85%|████████▌ | 738/864 [32:50<04:54,  2.34s/it]
 86%|████████▌ | 739/864 [32:53<04:51,  2.33s/it]
 86%|████████▌ | 740/864 [32:55<04:48,  2.33s/it]
                                                 
{'loss': '3.366', 'grad_norm': '0.954', 'learning_rate': '0.0003', 'epoch': '10.28'}
+
 86%|████████▌ | 740/864 [32:55<04:48,  2.33s/it]
 86%|████████▌ | 741/864 [32:57<04:46,  2.33s/it]
 86%|████████▌ | 742/864 [33:00<04:43,  2.33s/it]
 86%|████████▌ | 743/864 [33:02<04:41,  2.32s/it]
 86%|████████▌ | 744/864 [33:04<04:38,  2.32s/it]
 86%|████████▌ | 745/864 [33:07<04:36,  2.32s/it]
 86%|████████▋ | 746/864 [33:09<04:34,  2.32s/it]
 86%|████████▋ | 747/864 [33:11<04:31,  2.32s/it]
 87%|████████▋ | 748/864 [33:14<04:29,  2.32s/it]
 87%|████████▋ | 749/864 [33:16<04:27,  2.32s/it]
 87%|████████▋ | 750/864 [33:18<04:24,  2.32s/it]
                                                 
{'loss': '3.438', 'grad_norm': '0.8257', 'learning_rate': '0.0003', 'epoch': '10.42'}
+
 87%|████████▋ | 750/864 [33:18<04:24,  2.32s/it]
 87%|████████▋ | 751/864 [33:21<04:22,  2.32s/it]
 87%|████████▋ | 752/864 [33:23<04:19,  2.32s/it]
 87%|████████▋ | 753/864 [33:25<04:17,  2.32s/it]
 87%|████████▋ | 754/864 [33:28<04:15,  2.32s/it]
 87%|████████▋ | 755/864 [33:30<04:13,  2.32s/it]
 88%|████████▊ | 756/864 [33:32<04:10,  2.32s/it]
 88%|████████▊ | 757/864 [33:35<04:08,  2.32s/it]
 88%|████████▊ | 758/864 [33:37<04:06,  2.32s/it]
 88%|████████▊ | 759/864 [33:39<04:03,  2.32s/it]
 88%|████████▊ | 760/864 [33:42<04:01,  2.32s/it]
                                                 
{'loss': '3.374', 'grad_norm': '0.3496', 'learning_rate': '0.0003', 'epoch': '10.56'}
+
 88%|████████▊ | 760/864 [33:42<04:01,  2.32s/it]
 88%|████████▊ | 761/864 [33:44<03:59,  2.32s/it]
 88%|████████▊ | 762/864 [33:46<03:56,  2.32s/it]
 88%|████████▊ | 763/864 [33:48<03:54,  2.32s/it]
 88%|████████▊ | 764/864 [33:51<03:52,  2.32s/it]
 89%|████████▊ | 765/864 [33:53<03:49,  2.32s/it]
 89%|████████▊ | 766/864 [33:55<03:47,  2.32s/it]
 89%|████████▉ | 767/864 [33:58<03:45,  2.32s/it]
 89%|████████▉ | 768/864 [34:00<03:42,  2.32s/it]
 89%|████████▉ | 769/864 [34:02<03:40,  2.32s/it]
 89%|████████▉ | 770/864 [34:05<03:38,  2.32s/it]
                                                 
{'loss': '3.302', 'grad_norm': '0.2669', 'learning_rate': '0.0003', 'epoch': '10.7'}
+
 89%|████████▉ | 770/864 [34:05<03:38,  2.32s/it]
 89%|████████▉ | 771/864 [34:07<03:36,  2.32s/it]
 89%|████████▉ | 772/864 [34:09<03:33,  2.32s/it]
 89%|████████▉ | 773/864 [34:12<03:31,  2.32s/it]
 90%|████████▉ | 774/864 [34:14<03:29,  2.32s/it]
 90%|████████▉ | 775/864 [34:16<03:26,  2.32s/it]
 90%|████████▉ | 776/864 [34:19<03:24,  2.32s/it]
 90%|████████▉ | 777/864 [34:21<03:22,  2.32s/it]
 90%|█████████ | 778/864 [34:23<03:19,  2.32s/it]
 90%|█████████ | 779/864 [34:26<03:17,  2.32s/it]
 90%|█████████ | 780/864 [34:28<03:15,  2.32s/it]
                                                 
{'loss': '3.283', 'grad_norm': '0.2714', 'learning_rate': '0.0003', 'epoch': '10.84'}
+
 90%|█████████ | 780/864 [34:28<03:15,  2.32s/it]
 90%|█████████ | 781/864 [34:30<03:12,  2.33s/it]
 91%|█████████ | 782/864 [34:33<03:10,  2.33s/it]
 91%|█████████ | 783/864 [34:35<03:08,  2.33s/it]
 91%|█████████ | 784/864 [34:37<03:05,  2.32s/it]
 91%|█████████ | 785/864 [34:40<03:03,  2.32s/it]
 91%|█████████ | 786/864 [34:42<03:01,  2.32s/it]
 91%|█████████ | 787/864 [34:44<02:58,  2.32s/it]
 91%|█████████ | 788/864 [34:47<02:56,  2.33s/it]
 91%|█████████▏| 789/864 [34:49<02:54,  2.32s/it]
 91%|█████████▏| 790/864 [34:51<02:52,  2.32s/it]
                                                 
{'loss': '3.264', 'grad_norm': '0.2435', 'learning_rate': '0.0003', 'epoch': '10.98'}
+
 91%|█████████▏| 790/864 [34:51<02:52,  2.32s/it]
 92%|█████████▏| 791/864 [34:54<02:49,  2.33s/it]
 92%|█████████▏| 792/864 [34:55<02:25,  2.03s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 4
+
+
  0%|          | 0/77 [00:00<?, ?it/s]
+
  4%|▍         | 3/77 [00:00<00:04, 15.05it/s]
+
  6%|▋         | 5/77 [00:00<00:06, 11.89it/s]
+
  9%|▉         | 7/77 [00:00<00:06, 10.94it/s]
+
 12%|█▏        | 9/77 [00:00<00:06, 10.48it/s]
+
 14%|█▍        | 11/77 [00:01<00:06, 10.23it/s]
+
 17%|█▋        | 13/77 [00:01<00:06, 10.07it/s]
+
 19%|█▉        | 15/77 [00:01<00:06,  9.98it/s]
+
 21%|██        | 16/77 [00:01<00:06,  9.94it/s]
+
 22%|██▏       | 17/77 [00:01<00:06,  9.89it/s]
+
 23%|██▎       | 18/77 [00:01<00:05,  9.87it/s]
+
 25%|██▍       | 19/77 [00:01<00:05,  9.82it/s]
+
 26%|██▌       | 20/77 [00:01<00:05,  9.81it/s]
+
 27%|██▋       | 21/77 [00:02<00:05,  9.81it/s]
+
 29%|██▊       | 22/77 [00:02<00:05,  9.80it/s]
+
 30%|██▉       | 23/77 [00:02<00:05,  9.80it/s]
+
 31%|███       | 24/77 [00:02<00:05,  9.79it/s]
+
 32%|███▏      | 25/77 [00:02<00:05,  9.78it/s]
+
 34%|███▍      | 26/77 [00:02<00:05,  9.79it/s]
+
 35%|███▌      | 27/77 [00:02<00:05,  9.79it/s]
+
 36%|███▋      | 28/77 [00:02<00:05,  9.78it/s]
+
 38%|███▊      | 29/77 [00:02<00:04,  9.78it/s]
+
 39%|███▉      | 30/77 [00:02<00:04,  9.78it/s]
+
 40%|████      | 31/77 [00:03<00:04,  9.78it/s]
+
 42%|████▏     | 32/77 [00:03<00:04,  9.78it/s]
+
 43%|████▎     | 33/77 [00:03<00:04,  9.76it/s]
+
 44%|████▍     | 34/77 [00:03<00:04,  9.78it/s]
+
 45%|████▌     | 35/77 [00:03<00:04,  9.78it/s]
+
 47%|████▋     | 36/77 [00:03<00:04,  9.77it/s]
+
 48%|████▊     | 37/77 [00:03<00:04,  9.77it/s]
+
 49%|████▉     | 38/77 [00:03<00:03,  9.79it/s]
+
 51%|█████     | 39/77 [00:03<00:03,  9.78it/s]
+
 52%|█████▏    | 40/77 [00:03<00:03,  9.78it/s]
+
 53%|█████▎    | 41/77 [00:04<00:03,  9.77it/s]
+
 55%|█████▍    | 42/77 [00:04<00:03,  9.79it/s]
+
 56%|█████▌    | 43/77 [00:04<00:03,  9.79it/s]
+
 57%|█████▋    | 44/77 [00:04<00:03,  9.77it/s]
+
 58%|█████▊    | 45/77 [00:04<00:03,  9.77it/s]
+
 60%|█████▉    | 46/77 [00:04<00:03,  9.79it/s]
+
 61%|██████    | 47/77 [00:04<00:03,  9.79it/s]
+
 62%|██████▏   | 48/77 [00:04<00:02,  9.79it/s]
+
 64%|██████▎   | 49/77 [00:04<00:02,  9.78it/s]
+
 65%|██████▍   | 50/77 [00:05<00:02,  9.79it/s]
+
 66%|██████▌   | 51/77 [00:05<00:02,  9.78it/s]
+
 68%|██████▊   | 52/77 [00:05<00:02,  9.78it/s]
+
 69%|██████▉   | 53/77 [00:05<00:02,  9.77it/s]
+
 70%|███████   | 54/77 [00:05<00:02,  9.79it/s]
+
 71%|███████▏  | 55/77 [00:05<00:02,  9.78it/s]
+
 73%|███████▎  | 56/77 [00:05<00:02,  9.78it/s]
+
 74%|███████▍  | 57/77 [00:05<00:02,  9.78it/s]
+
 75%|███████▌  | 58/77 [00:05<00:01,  9.79it/s]
+
 77%|███████▋  | 59/77 [00:05<00:01,  9.78it/s]
+
 78%|███████▊  | 60/77 [00:06<00:01,  9.79it/s]
+
 79%|███████▉  | 61/77 [00:06<00:01,  9.81it/s]
+
 81%|████████  | 62/77 [00:06<00:01,  9.80it/s]
+
 82%|████████▏ | 63/77 [00:06<00:01,  9.78it/s]
+
 83%|████████▎ | 64/77 [00:06<00:01,  9.78it/s]
+
 84%|████████▍ | 65/77 [00:06<00:01,  9.78it/s]
+
 86%|████████▌ | 66/77 [00:06<00:01,  9.78it/s]
+
 87%|████████▋ | 67/77 [00:06<00:01,  9.78it/s]
+
 88%|████████▊ | 68/77 [00:06<00:00,  9.75it/s]
+
 90%|████████▉ | 69/77 [00:06<00:00,  9.79it/s]
+
 91%|█████████ | 70/77 [00:07<00:00,  9.80it/s]
+
 92%|█████████▏| 71/77 [00:07<00:00,  9.79it/s]
+
 94%|█████████▎| 72/77 [00:07<00:00,  9.76it/s]
+
 95%|█████████▍| 73/77 [00:07<00:00,  9.77it/s]
+
 96%|█████████▌| 74/77 [00:07<00:00,  9.79it/s]
+
 97%|█████████▋| 75/77 [00:07<00:00,  9.79it/s]
+
 99%|█████████▊| 76/77 [00:07<00:00,  9.79it/s]
+
100%|██████████| 77/77 [00:07<00:00,  9.80it/s]
                                                 
+
                                               
{'eval_loss': '4.088', 'eval_runtime': '7.915', 'eval_samples_per_second': '308.4', 'eval_steps_per_second': '9.729', 'epoch': '11'}
+
 92%|█████████▏| 792/864 [35:03<02:25,  2.03s/it]
+
100%|██████████| 77/77 [00:07<00:00,  9.80it/s][eval] epoch 11.0 step 792 loss 4.0877 (best 4.0612 @ ep 10.0, 1/10000 since best)
+
+
                                               
 92%|█████████▏| 793/864 [35:05<05:19,  4.50s/it]
 92%|█████████▏| 794/864 [35:08<04:37,  3.96s/it]
 92%|█████████▏| 795/864 [35:10<03:59,  3.47s/it]
 92%|█████████▏| 796/864 [35:12<03:32,  3.12s/it]
 92%|█████████▏| 797/864 [35:15<03:13,  2.88s/it]
 92%|█████████▏| 798/864 [35:17<02:59,  2.71s/it]
 92%|█████████▏| 799/864 [35:19<02:48,  2.60s/it]
 93%|█████████▎| 800/864 [35:22<02:40,  2.52s/it]
                                                 
{'loss': '3.068', 'grad_norm': '0.9555', 'learning_rate': '0.0003', 'epoch': '11.11'}
+
 93%|█████████▎| 800/864 [35:22<02:40,  2.52s/it]
 93%|█████████▎| 801/864 [35:24<02:34,  2.46s/it]
 93%|█████████▎| 802/864 [35:26<02:29,  2.42s/it]
 93%|█████████▎| 803/864 [35:29<02:25,  2.39s/it]
 93%|█████████▎| 804/864 [35:31<02:22,  2.37s/it]
 93%|█████████▎| 805/864 [35:33<02:18,  2.36s/it]
 93%|█████████▎| 806/864 [35:36<02:22,  2.46s/it]
 93%|█████████▎| 807/864 [35:38<02:17,  2.42s/it]
 94%|█████████▎| 808/864 [35:41<02:20,  2.50s/it]
 94%|█████████▎| 809/864 [35:44<02:20,  2.56s/it]
 94%|█████████▍| 810/864 [35:46<02:14,  2.49s/it]
                                                 
{'loss': '3.041', 'grad_norm': '0.8559', 'learning_rate': '0.0003', 'epoch': '11.25'}
+
 94%|█████████▍| 810/864 [35:46<02:14,  2.49s/it]
 94%|█████████▍| 811/864 [35:49<02:16,  2.58s/it]
 94%|█████████▍| 812/864 [35:52<02:15,  2.61s/it]
 94%|█████████▍| 813/864 [35:54<02:08,  2.53s/it]
 94%|█████████▍| 814/864 [35:56<02:03,  2.47s/it]
 94%|█████████▍| 815/864 [35:59<01:58,  2.42s/it]
 94%|█████████▍| 816/864 [36:01<02:00,  2.51s/it]
 95%|█████████▍| 817/864 [36:04<01:55,  2.45s/it]
 95%|█████████▍| 818/864 [36:06<01:50,  2.41s/it]
 95%|█████████▍| 819/864 [36:09<01:52,  2.50s/it]
 95%|█████████▍| 820/864 [36:11<01:47,  2.44s/it]
                                                 
{'loss': '3.079', 'grad_norm': '0.8699', 'learning_rate': '0.0003', 'epoch': '11.39'}
+
 95%|█████████▍| 820/864 [36:11<01:47,  2.44s/it]
 95%|█████████▌| 821/864 [36:13<01:43,  2.41s/it]
 95%|█████████▌| 822/864 [36:16<01:40,  2.38s/it]
 95%|█████████▌| 823/864 [36:18<01:36,  2.36s/it]
 95%|█████████▌| 824/864 [36:20<01:34,  2.35s/it]
 95%|█████████▌| 825/864 [36:23<01:31,  2.34s/it]
 96%|█████████▌| 826/864 [36:25<01:28,  2.34s/it]
 96%|█████████▌| 827/864 [36:27<01:26,  2.33s/it]
 96%|█████████▌| 828/864 [36:30<01:23,  2.33s/it]
 96%|█████████▌| 829/864 [36:32<01:21,  2.33s/it]
 96%|█████████▌| 830/864 [36:34<01:19,  2.33s/it]
                                                 
{'loss': '3.087', 'grad_norm': '0.4405', 'learning_rate': '0.0003', 'epoch': '11.53'}
+
 96%|█████████▌| 830/864 [36:34<01:19,  2.33s/it]
 96%|█████████▌| 831/864 [36:37<01:16,  2.33s/it]
 96%|█████████▋| 832/864 [36:39<01:14,  2.33s/it]
 96%|█████████▋| 833/864 [36:41<01:12,  2.32s/it]
 97%|█████████▋| 834/864 [36:43<01:09,  2.32s/it]
 97%|█████████▋| 835/864 [36:46<01:07,  2.32s/it]
 97%|█████████▋| 836/864 [36:48<01:05,  2.32s/it]
 97%|█████████▋| 837/864 [36:50<01:02,  2.32s/it]
 97%|█████████▋| 838/864 [36:53<01:00,  2.33s/it]
 97%|█████████▋| 839/864 [36:55<00:58,  2.32s/it]
 97%|█████████▋| 840/864 [36:57<00:55,  2.32s/it]
                                                 
{'loss': '3.071', 'grad_norm': '0.311', 'learning_rate': '0.0003', 'epoch': '11.67'}
+
 97%|█████████▋| 840/864 [36:57<00:55,  2.32s/it]
 97%|█████████▋| 841/864 [37:00<00:53,  2.33s/it]
 97%|█████████▋| 842/864 [37:02<00:51,  2.33s/it]
 98%|█████████▊| 843/864 [37:04<00:48,  2.32s/it]
 98%|█████████▊| 844/864 [37:07<00:46,  2.32s/it]
 98%|█████████▊| 845/864 [37:09<00:44,  2.32s/it]
 98%|█████████▊| 846/864 [37:11<00:41,  2.32s/it]
 98%|█████████▊| 847/864 [37:14<00:39,  2.32s/it]
 98%|█████████▊| 848/864 [37:16<00:37,  2.32s/it]
 98%|█████████▊| 849/864 [37:18<00:34,  2.32s/it]
 98%|█████████▊| 850/864 [37:21<00:32,  2.32s/it]
                                                 
{'loss': '3.075', 'grad_norm': '0.2971', 'learning_rate': '0.0003', 'epoch': '11.81'}
+
 98%|█████████▊| 850/864 [37:21<00:32,  2.32s/it]
 98%|█████████▊| 851/864 [37:23<00:30,  2.32s/it]
 99%|█████████▊| 852/864 [37:25<00:27,  2.32s/it]
 99%|█████████▊| 853/864 [37:28<00:25,  2.32s/it]
 99%|█████████▉| 854/864 [37:30<00:23,  2.32s/it]
 99%|█████████▉| 855/864 [37:32<00:20,  2.32s/it]
 99%|█████████▉| 856/864 [37:35<00:18,  2.32s/it]
 99%|█████████▉| 857/864 [37:37<00:16,  2.32s/it]
 99%|█████████▉| 858/864 [37:39<00:13,  2.32s/it]
 99%|█████████▉| 859/864 [37:42<00:11,  2.32s/it]
100%|█████████▉| 860/864 [37:44<00:09,  2.32s/it]
                                                 
{'loss': '3.087', 'grad_norm': '0.3015', 'learning_rate': '0.0003', 'epoch': '11.95'}
+
100%|█████████▉| 860/864 [37:44<00:09,  2.32s/it]
100%|█████████▉| 861/864 [37:46<00:06,  2.32s/it]
100%|█████████▉| 862/864 [37:49<00:04,  2.32s/it]
100%|█████████▉| 863/864 [37:51<00:02,  2.32s/it]
100%|██████████| 864/864 [37:52<00:00,  2.02s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 4
+
+
  0%|          | 0/77 [00:00<?, ?it/s]
+
  4%|▍         | 3/77 [00:00<00:04, 14.92it/s]
+
  6%|▋         | 5/77 [00:00<00:06, 11.85it/s]
+
  9%|▉         | 7/77 [00:00<00:06, 10.92it/s]
+
 12%|█▏        | 9/77 [00:00<00:06, 10.46it/s]
+
 14%|█▍        | 11/77 [00:01<00:06, 10.22it/s]
+
 17%|█▋        | 13/77 [00:01<00:06, 10.06it/s]
+
 19%|█▉        | 15/77 [00:01<00:06,  9.98it/s]
+
 21%|██        | 16/77 [00:01<00:06,  9.94it/s]
+
 22%|██▏       | 17/77 [00:01<00:06,  9.90it/s]
+
 23%|██▎       | 18/77 [00:01<00:05,  9.88it/s]
+
 25%|██▍       | 19/77 [00:01<00:05,  9.83it/s]
+
 26%|██▌       | 20/77 [00:01<00:05,  9.83it/s]
+
 27%|██▋       | 21/77 [00:02<00:05,  9.82it/s]
+
 29%|██▊       | 22/77 [00:02<00:05,  9.81it/s]
+
 30%|██▉       | 23/77 [00:02<00:05,  9.80it/s]
+
 31%|███       | 24/77 [00:02<00:05,  9.79it/s]
+
 32%|███▏      | 25/77 [00:02<00:05,  9.79it/s]
+
 34%|███▍      | 26/77 [00:02<00:05,  9.79it/s]
+
 35%|███▌      | 27/77 [00:02<00:05,  9.78it/s]
+
 36%|███▋      | 28/77 [00:02<00:05,  9.78it/s]
+
 38%|███▊      | 29/77 [00:02<00:04,  9.78it/s]
+
 39%|███▉      | 30/77 [00:02<00:04,  9.79it/s]
+
 40%|████      | 31/77 [00:03<00:04,  9.78it/s]
+
 42%|████▏     | 32/77 [00:03<00:04,  9.79it/s]
+
 43%|████▎     | 33/77 [00:03<00:04,  9.78it/s]
+
 44%|████▍     | 34/77 [00:03<00:04,  9.78it/s]
+
 45%|████▌     | 35/77 [00:03<00:04,  9.78it/s]
+
 47%|████▋     | 36/77 [00:03<00:04,  9.78it/s]
+
 48%|████▊     | 37/77 [00:03<00:04,  9.78it/s]
+
 49%|████▉     | 38/77 [00:03<00:03,  9.78it/s]
+
 51%|█████     | 39/77 [00:03<00:03,  9.77it/s]
+
 52%|█████▏    | 40/77 [00:03<00:03,  9.77it/s]
+
 53%|█████▎    | 41/77 [00:04<00:03,  9.78it/s]
+
 55%|█████▍    | 42/77 [00:04<00:03,  9.78it/s]
+
 56%|█████▌    | 43/77 [00:04<00:03,  9.77it/s]
+
 57%|█████▋    | 44/77 [00:04<00:03,  9.78it/s]
+
 58%|█████▊    | 45/77 [00:04<00:03,  9.78it/s]
+
 60%|█████▉    | 46/77 [00:04<00:03,  9.78it/s]
+
 61%|██████    | 47/77 [00:04<00:03,  9.78it/s]
+
 62%|██████▏   | 48/77 [00:04<00:02,  9.78it/s]
+
 64%|██████▎   | 49/77 [00:04<00:02,  9.78it/s]
+
 65%|██████▍   | 50/77 [00:05<00:02,  9.78it/s]
+
 66%|██████▌   | 51/77 [00:05<00:02,  9.76it/s]
+
 68%|██████▊   | 52/77 [00:05<00:02,  9.78it/s]
+
 69%|██████▉   | 53/77 [00:05<00:02,  9.78it/s]
+
 70%|███████   | 54/77 [00:05<00:02,  9.78it/s]
+
 71%|███████▏  | 55/77 [00:05<00:02,  9.76it/s]
+
 73%|███████▎  | 56/77 [00:05<00:02,  9.77it/s]
+
 74%|███████▍  | 57/77 [00:05<00:02,  9.78it/s]
+
 75%|███████▌  | 58/77 [00:05<00:01,  9.78it/s]
+
 77%|███████▋  | 59/77 [00:05<00:01,  9.77it/s]
+
 78%|███████▊  | 60/77 [00:06<00:01,  9.78it/s]
+
 79%|███████▉  | 61/77 [00:06<00:01,  9.79it/s]
+
 81%|████████  | 62/77 [00:06<00:01,  9.80it/s]
+
 82%|████████▏ | 63/77 [00:06<00:01,  9.79it/s]
+
 83%|████████▎ | 64/77 [00:06<00:01,  9.79it/s]
+
 84%|████████▍ | 65/77 [00:06<00:01,  9.79it/s]
+
 86%|████████▌ | 66/77 [00:06<00:01,  9.79it/s]
+
 87%|████████▋ | 67/77 [00:06<00:01,  9.78it/s]
+
 88%|████████▊ | 68/77 [00:06<00:00,  9.78it/s]
+
 90%|████████▉ | 69/77 [00:06<00:00,  9.79it/s]
+
 91%|█████████ | 70/77 [00:07<00:00,  9.79it/s]
+
 92%|█████████▏| 71/77 [00:07<00:00,  9.78it/s]
+
 94%|█████████▎| 72/77 [00:07<00:00,  9.79it/s]
+
 95%|█████████▍| 73/77 [00:07<00:00,  9.79it/s]
+
 96%|█████████▌| 74/77 [00:07<00:00,  9.79it/s]
+
 97%|█████████▋| 75/77 [00:07<00:00,  9.79it/s]
+
 99%|█████████▊| 76/77 [00:07<00:00,  9.79it/s]
+
100%|██████████| 77/77 [00:07<00:00,  9.78it/s]
                                                 
+
                                               
{'eval_loss': '4.137', 'eval_runtime': '7.904', 'eval_samples_per_second': '308.8', 'eval_steps_per_second': '9.742', 'epoch': '12'}
+
100%|██████████| 864/864 [38:00<00:00,  2.02s/it]
+
100%|██████████| 77/77 [00:07<00:00,  9.78it/s][eval] epoch 12.0 step 864 loss 4.1373 (best 4.0612 @ ep 10.0, 2/10000 since best)
+
+
                                               [transformers] 
+
+Training completed. Do not forget to share your model on huggingface.co/models =)
+
+
+
                                                 
{'train_runtime': '2282', 'train_samples_per_second': '192.6', 'train_steps_per_second': '0.379', 'train_loss': '4.651', 'epoch': '12'}
+
100%|██████████| 864/864 [38:00<00:00,  2.02s/it]
100%|██████████| 864/864 [38:00<00:00,  2.64s/it]
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+wandb: 
+wandb: 🚀 View run t0_S5_lr0.0003 at: https://wandb.ai/slowrun-wankers/data-aug-pretraining/runs/7qnpc6jf
+wandb: Find logs at: wandb/run-20260817_213355-7qnpc6jf/logs
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+[rank0]:[W817 22:12:43.825420359 ProcessGroupNCCL.cpp:1575] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator())
diff --git a/runs/t0/t0_S5_lr0.0012/checkpoint-720/chat_template.jinja b/runs/t0/t0_S5_lr0.0012/checkpoint-720/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..699ff8df401fe4788525e9c1f9b86a99eadd6230
--- /dev/null
+++ b/runs/t0/t0_S5_lr0.0012/checkpoint-720/chat_template.jinja
@@ -0,0 +1,85 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0].role == 'system' %}
+        {{- messages[0].content + '\n\n' }}
+    {%- endif %}
+    {{- "# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0].content + '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
+{%- for message in messages[::-1] %}
+    {%- set index = (messages|length - 1) - loop.index0 %}
+    {%- if ns.multi_step_tool and message.role == "user" and not(message.content.startswith('<tool_response>') and message.content.endswith('</tool_response>')) %}
+        {%- set ns.multi_step_tool = false %}
+        {%- set ns.last_query_index = index %}
+    {%- endif %}
+{%- endfor %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {%- set content = message.content %}
+        {%- set reasoning_content = '' %}
+        {%- if message.reasoning_content is defined and message.reasoning_content is not none %}
+            {%- set reasoning_content = message.reasoning_content %}
+        {%- else %}
+            {%- if '</think>' in message.content %}
+                {%- set content = message.content.split('</think>')[-1].lstrip('\n') %}
+                {%- set reasoning_content = message.content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
+            {%- endif %}
+        {%- endif %}
+        {%- if loop.index0 > ns.last_query_index %}
+            {%- if loop.last or (not loop.last and reasoning_content) %}
+                {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content.strip('\n') + '\n</think>\n\n' + content.lstrip('\n') }}
+            {%- else %}
+                {{- '<|im_start|>' + message.role + '\n' + content }}
+            {%- endif %}
+        {%- else %}
+            {{- '<|im_start|>' + message.role + '\n' + content }}
+        {%- endif %}
+        {%- if message.tool_calls %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if (loop.first and content) or (not loop.first) %}
+                    {{- '\n' }}
+                {%- endif %}
+                {%- if tool_call.function %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {{- '<tool_call>\n{"name": "' }}
+                {{- tool_call.name }}
+                {{- '", "arguments": ' }}
+                {%- if tool_call.arguments is string %}
+                    {{- tool_call.arguments }}
+                {%- else %}
+                    {{- tool_call.arguments | tojson }}
+                {%- endif %}
+                {{- '}\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.first or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+    {%- if enable_thinking is defined and enable_thinking is false %}
+        {{- '<think>\n\n</think>\n\n' }}
+    {%- endif %}
+{%- endif %}
\ No newline at end of file
diff --git a/runs/t0/t0_S5_lr0.0012/checkpoint-720/config.json b/runs/t0/t0_S5_lr0.0012/checkpoint-720/config.json
new file mode 100644
index 0000000000000000000000000000000000000000..58ed141896af03a1496676b060879ee607fff4bb
--- /dev/null
+++ b/runs/t0/t0_S5_lr0.0012/checkpoint-720/config.json
@@ -0,0 +1,32 @@
+{
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": null,
+  "dtype": "float32",
+  "eos_token_id": 151645,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1536,
+  "initializer_range": 0.02,
+  "intermediate_size": 4096,
+  "max_position_embeddings": 2048,
+  "mlp_bias": false,
+  "model_type": "llama",
+  "num_attention_heads": 12,
+  "num_hidden_layers": 24,
+  "num_key_value_heads": 12,
+  "pad_token_id": 151645,
+  "pretraining_tp": 1,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 10000.0,
+    "rope_type": "default"
+  },
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": false,
+  "vocab_size": 151671
+}
diff --git a/runs/t0/t0_S5_lr0.0012/checkpoint-720/generation_config.json b/runs/t0/t0_S5_lr0.0012/checkpoint-720/generation_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..8c4242f32d40df256857f87264f646130f684e33
--- /dev/null
+++ b/runs/t0/t0_S5_lr0.0012/checkpoint-720/generation_config.json
@@ -0,0 +1,11 @@
+{
+  "_from_model_config": true,
+  "eos_token_id": [
+    151645
+  ],
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "transformers_version": "5.15.0",
+  "use_cache": true
+}
diff --git a/runs/t0/t0_S5_lr0.0012/checkpoint-720/tokenizer_config.json b/runs/t0/t0_S5_lr0.0012/checkpoint-720/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..839a3743177a3ed77e1ffbae67520e8500ca0532
--- /dev/null
+++ b/runs/t0/t0_S5_lr0.0012/checkpoint-720/tokenizer_config.json
@@ -0,0 +1,19 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|l2r_pred|>",
+    "<|r2l_pred|>"
+  ],
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 131072,
+  "pad_token": "<|im_end|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}
diff --git a/runs/t0/t0_S5_lr0.0012/checkpoint-720/trainer_state.json b/runs/t0/t0_S5_lr0.0012/checkpoint-720/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..fa2466c91c410aa331b16f825f827fd65893ed63
--- /dev/null
+++ b/runs/t0/t0_S5_lr0.0012/checkpoint-720/trainer_state.json
@@ -0,0 +1,625 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 10.0,
+  "eval_steps": 72,
+  "global_step": 720,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.013973799126637555,
+      "grad_norm": 3.2854130268096924,
+      "learning_rate": 0.0,
+      "loss": 12.239278793334961,
+      "step": 1
+    },
+    {
+      "epoch": 0.13973799126637554,
+      "grad_norm": 2.8914339542388916,
+      "learning_rate": 0.00010799999999999998,
+      "loss": 10.772604200575087,
+      "step": 10
+    },
+    {
+      "epoch": 0.2794759825327511,
+      "grad_norm": 0.8951712846755981,
+      "learning_rate": 0.00022799999999999999,
+      "loss": 8.708420562744141,
+      "step": 20
+    },
+    {
+      "epoch": 0.4192139737991266,
+      "grad_norm": 0.8144592642784119,
+      "learning_rate": 0.00034799999999999995,
+      "loss": 7.609435272216797,
+      "step": 30
+    },
+    {
+      "epoch": 0.5589519650655022,
+      "grad_norm": 0.7392408847808838,
+      "learning_rate": 0.000468,
+      "loss": 7.431675720214844,
+      "step": 40
+    },
+    {
+      "epoch": 0.6986899563318777,
+      "grad_norm": 1.823486089706421,
+      "learning_rate": 0.000588,
+      "loss": 7.193846130371094,
+      "step": 50
+    },
+    {
+      "epoch": 0.8384279475982532,
+      "grad_norm": 1.0612823963165283,
+      "learning_rate": 0.0007079999999999999,
+      "loss": 6.98529052734375,
+      "step": 60
+    },
+    {
+      "epoch": 0.9781659388646288,
+      "grad_norm": 0.8120266795158386,
+      "learning_rate": 0.0008279999999999999,
+      "loss": 6.868883514404297,
+      "step": 70
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 6.711178302764893,
+      "eval_runtime": 32.1685,
+      "eval_samples_per_second": 75.882,
+      "eval_steps_per_second": 2.394,
+      "step": 72
+    },
+    {
+      "epoch": 1.1117903930131003,
+      "grad_norm": 0.6914841532707214,
+      "learning_rate": 0.000948,
+      "loss": 6.636103820800781,
+      "step": 80
+    },
+    {
+      "epoch": 1.251528384279476,
+      "grad_norm": 0.780213475227356,
+      "learning_rate": 0.001068,
+      "loss": 6.488282775878906,
+      "step": 90
+    },
+    {
+      "epoch": 1.3912663755458516,
+      "grad_norm": 0.9557600021362305,
+      "learning_rate": 0.0011879999999999998,
+      "loss": 6.354541778564453,
+      "step": 100
+    },
+    {
+      "epoch": 1.531004366812227,
+      "grad_norm": 0.7227393984794617,
+      "learning_rate": 0.0012,
+      "loss": 6.275666046142578,
+      "step": 110
+    },
+    {
+      "epoch": 1.6707423580786025,
+      "grad_norm": 2.0571846961975098,
+      "learning_rate": 0.0012,
+      "loss": 6.149044036865234,
+      "step": 120
+    },
+    {
+      "epoch": 1.8104803493449781,
+      "grad_norm": 0.3692125380039215,
+      "learning_rate": 0.0012,
+      "loss": 6.129460144042969,
+      "step": 130
+    },
+    {
+      "epoch": 1.9502183406113538,
+      "grad_norm": 0.35232868790626526,
+      "learning_rate": 0.0012,
+      "loss": 5.982979583740234,
+      "step": 140
+    },
+    {
+      "epoch": 2.0,
+      "eval_loss": 6.093167304992676,
+      "eval_runtime": 7.9136,
+      "eval_samples_per_second": 308.458,
+      "eval_steps_per_second": 9.73,
+      "step": 144
+    },
+    {
+      "epoch": 2.0838427947598253,
+      "grad_norm": 0.40879613161087036,
+      "learning_rate": 0.0012,
+      "loss": 5.902141571044922,
+      "step": 150
+    },
+    {
+      "epoch": 2.2235807860262007,
+      "grad_norm": 0.44311410188674927,
+      "learning_rate": 0.0012,
+      "loss": 5.77630844116211,
+      "step": 160
+    },
+    {
+      "epoch": 2.3633187772925766,
+      "grad_norm": 0.5734341144561768,
+      "learning_rate": 0.0012,
+      "loss": 5.730416107177734,
+      "step": 170
+    },
+    {
+      "epoch": 2.503056768558952,
+      "grad_norm": 0.3401224613189697,
+      "learning_rate": 0.0012,
+      "loss": 5.651076126098633,
+      "step": 180
+    },
+    {
+      "epoch": 2.6427947598253274,
+      "grad_norm": 0.7364341020584106,
+      "learning_rate": 0.0012,
+      "loss": 5.60278549194336,
+      "step": 190
+    },
+    {
+      "epoch": 2.7825327510917033,
+      "grad_norm": 0.27776798605918884,
+      "learning_rate": 0.0012,
+      "loss": 5.503947830200195,
+      "step": 200
+    },
+    {
+      "epoch": 2.9222707423580787,
+      "grad_norm": 0.5733747482299805,
+      "learning_rate": 0.0012,
+      "loss": 5.419222259521485,
+      "step": 210
+    },
+    {
+      "epoch": 3.0,
+      "eval_loss": 5.410129070281982,
+      "eval_runtime": 7.9145,
+      "eval_samples_per_second": 308.42,
+      "eval_steps_per_second": 9.729,
+      "step": 216
+    },
+    {
+      "epoch": 3.05589519650655,
+      "grad_norm": 0.22739177942276,
+      "learning_rate": 0.0012,
+      "loss": 5.340640640258789,
+      "step": 220
+    },
+    {
+      "epoch": 3.1956331877729256,
+      "grad_norm": 0.3654351830482483,
+      "learning_rate": 0.0012,
+      "loss": 5.246836090087891,
+      "step": 230
+    },
+    {
+      "epoch": 3.3353711790393015,
+      "grad_norm": 0.30522432923316956,
+      "learning_rate": 0.0012,
+      "loss": 5.183464813232422,
+      "step": 240
+    },
+    {
+      "epoch": 3.475109170305677,
+      "grad_norm": 0.7293229103088379,
+      "learning_rate": 0.0012,
+      "loss": 5.146945571899414,
+      "step": 250
+    },
+    {
+      "epoch": 3.6148471615720523,
+      "grad_norm": 0.27008751034736633,
+      "learning_rate": 0.0012,
+      "loss": 5.120269012451172,
+      "step": 260
+    },
+    {
+      "epoch": 3.7545851528384278,
+      "grad_norm": 0.25352606177330017,
+      "learning_rate": 0.0012,
+      "loss": 4.999597549438477,
+      "step": 270
+    },
+    {
+      "epoch": 3.8943231441048036,
+      "grad_norm": 0.622519314289093,
+      "learning_rate": 0.0012,
+      "loss": 4.967436218261719,
+      "step": 280
+    },
+    {
+      "epoch": 4.0,
+      "eval_loss": 5.000211238861084,
+      "eval_runtime": 7.9158,
+      "eval_samples_per_second": 308.369,
+      "eval_steps_per_second": 9.727,
+      "step": 288
+    },
+    {
+      "epoch": 4.0279475982532755,
+      "grad_norm": 0.33289751410484314,
+      "learning_rate": 0.0012,
+      "loss": 4.923779678344727,
+      "step": 290
+    },
+    {
+      "epoch": 4.1676855895196505,
+      "grad_norm": 0.2035096287727356,
+      "learning_rate": 0.0012,
+      "loss": 4.7559326171875,
+      "step": 300
+    },
+    {
+      "epoch": 4.307423580786026,
+      "grad_norm": 0.4597044587135315,
+      "learning_rate": 0.0012,
+      "loss": 4.681830596923828,
+      "step": 310
+    },
+    {
+      "epoch": 4.447161572052401,
+      "grad_norm": 0.39535853266716003,
+      "learning_rate": 0.0012,
+      "loss": 4.705263900756836,
+      "step": 320
+    },
+    {
+      "epoch": 4.586899563318777,
+      "grad_norm": 0.18191099166870117,
+      "learning_rate": 0.0012,
+      "loss": 4.623991394042969,
+      "step": 330
+    },
+    {
+      "epoch": 4.726637554585153,
+      "grad_norm": 0.3703513741493225,
+      "learning_rate": 0.0012,
+      "loss": 4.537522506713867,
+      "step": 340
+    },
+    {
+      "epoch": 4.866375545851528,
+      "grad_norm": 0.4233785569667816,
+      "learning_rate": 0.0012,
+      "loss": 4.576454162597656,
+      "step": 350
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 0.17962193489074707,
+      "learning_rate": 0.0012,
+      "loss": 4.482110214233399,
+      "step": 360
+    },
+    {
+      "epoch": 5.0,
+      "eval_loss": 4.558745384216309,
+      "eval_runtime": 7.9108,
+      "eval_samples_per_second": 308.566,
+      "eval_steps_per_second": 9.734,
+      "step": 360
+    },
+    {
+      "epoch": 5.139737991266376,
+      "grad_norm": 0.33794084191322327,
+      "learning_rate": 0.0012,
+      "loss": 4.3163707733154295,
+      "step": 370
+    },
+    {
+      "epoch": 5.279475982532751,
+      "grad_norm": 0.21855460107326508,
+      "learning_rate": 0.0012,
+      "loss": 4.3009895324707035,
+      "step": 380
+    },
+    {
+      "epoch": 5.419213973799127,
+      "grad_norm": 0.15166562795639038,
+      "learning_rate": 0.0012,
+      "loss": 4.227708435058593,
+      "step": 390
+    },
+    {
+      "epoch": 5.558951965065502,
+      "grad_norm": 0.29830625653266907,
+      "learning_rate": 0.0012,
+      "loss": 4.207986068725586,
+      "step": 400
+    },
+    {
+      "epoch": 5.698689956331878,
+      "grad_norm": 0.24197085201740265,
+      "learning_rate": 0.0012,
+      "loss": 4.186326599121093,
+      "step": 410
+    },
+    {
+      "epoch": 5.8384279475982535,
+      "grad_norm": 0.17020633816719055,
+      "learning_rate": 0.0012,
+      "loss": 4.153177642822266,
+      "step": 420
+    },
+    {
+      "epoch": 5.978165938864628,
+      "grad_norm": 0.23463435471057892,
+      "learning_rate": 0.0012,
+      "loss": 4.124895477294922,
+      "step": 430
+    },
+    {
+      "epoch": 6.0,
+      "eval_loss": 4.291956424713135,
+      "eval_runtime": 7.9141,
+      "eval_samples_per_second": 308.435,
+      "eval_steps_per_second": 9.729,
+      "step": 432
+    },
+    {
+      "epoch": 6.1117903930131,
+      "grad_norm": 0.33610767126083374,
+      "learning_rate": 0.0012,
+      "loss": 4.008880615234375,
+      "step": 440
+    },
+    {
+      "epoch": 6.251528384279476,
+      "grad_norm": 0.2952611744403839,
+      "learning_rate": 0.0012,
+      "loss": 4.0130470275878904,
+      "step": 450
+    },
+    {
+      "epoch": 6.391266375545851,
+      "grad_norm": 0.1474909633398056,
+      "learning_rate": 0.0012,
+      "loss": 3.9605411529541015,
+      "step": 460
+    },
+    {
+      "epoch": 6.531004366812227,
+      "grad_norm": 0.17840559780597687,
+      "learning_rate": 0.0012,
+      "loss": 3.9215927124023438,
+      "step": 470
+    },
+    {
+      "epoch": 6.670742358078603,
+      "grad_norm": 0.22309419512748718,
+      "learning_rate": 0.0012,
+      "loss": 3.9179962158203123,
+      "step": 480
+    },
+    {
+      "epoch": 6.810480349344978,
+      "grad_norm": 0.17490209639072418,
+      "learning_rate": 0.0012,
+      "loss": 3.9208335876464844,
+      "step": 490
+    },
+    {
+      "epoch": 6.950218340611354,
+      "grad_norm": 0.16962862014770508,
+      "learning_rate": 0.0012,
+      "loss": 3.8781494140625,
+      "step": 500
+    },
+    {
+      "epoch": 7.0,
+      "eval_loss": 4.123973369598389,
+      "eval_runtime": 7.9124,
+      "eval_samples_per_second": 308.502,
+      "eval_steps_per_second": 9.732,
+      "step": 504
+    },
+    {
+      "epoch": 7.083842794759826,
+      "grad_norm": 0.2539425194263458,
+      "learning_rate": 0.0012,
+      "loss": 3.77655029296875,
+      "step": 510
+    },
+    {
+      "epoch": 7.223580786026201,
+      "grad_norm": 0.26208364963531494,
+      "learning_rate": 0.0012,
+      "loss": 3.745014953613281,
+      "step": 520
+    },
+    {
+      "epoch": 7.3633187772925766,
+      "grad_norm": 0.1574983447790146,
+      "learning_rate": 0.0012,
+      "loss": 3.7169628143310547,
+      "step": 530
+    },
+    {
+      "epoch": 7.503056768558952,
+      "grad_norm": 0.18709805607795715,
+      "learning_rate": 0.0012,
+      "loss": 3.7050544738769533,
+      "step": 540
+    },
+    {
+      "epoch": 7.642794759825327,
+      "grad_norm": 0.14072749018669128,
+      "learning_rate": 0.0012,
+      "loss": 3.6973304748535156,
+      "step": 550
+    },
+    {
+      "epoch": 7.782532751091703,
+      "grad_norm": 0.13279055058956146,
+      "learning_rate": 0.0012,
+      "loss": 3.686246871948242,
+      "step": 560
+    },
+    {
+      "epoch": 7.922270742358078,
+      "grad_norm": 0.2193097621202469,
+      "learning_rate": 0.0012,
+      "loss": 3.6956031799316404,
+      "step": 570
+    },
+    {
+      "epoch": 8.0,
+      "eval_loss": 4.036094665527344,
+      "eval_runtime": 7.9031,
+      "eval_samples_per_second": 308.867,
+      "eval_steps_per_second": 9.743,
+      "step": 576
+    },
+    {
+      "epoch": 8.055895196506551,
+      "grad_norm": 0.31165802478790283,
+      "learning_rate": 0.0012,
+      "loss": 3.6125083923339845,
+      "step": 580
+    },
+    {
+      "epoch": 8.195633187772925,
+      "grad_norm": 0.4581586420536041,
+      "learning_rate": 0.0012,
+      "loss": 3.610798645019531,
+      "step": 590
+    },
+    {
+      "epoch": 8.335371179039301,
+      "grad_norm": 0.28279274702072144,
+      "learning_rate": 0.0012,
+      "loss": 3.702981948852539,
+      "step": 600
+    },
+    {
+      "epoch": 8.475109170305677,
+      "grad_norm": 0.15088902413845062,
+      "learning_rate": 0.0012,
+      "loss": 3.6176616668701174,
+      "step": 610
+    },
+    {
+      "epoch": 8.614847161572053,
+      "grad_norm": 0.10666224360466003,
+      "learning_rate": 0.0012,
+      "loss": 3.558070755004883,
+      "step": 620
+    },
+    {
+      "epoch": 8.754585152838429,
+      "grad_norm": 0.11449340730905533,
+      "learning_rate": 0.0012,
+      "loss": 3.536176300048828,
+      "step": 630
+    },
+    {
+      "epoch": 8.894323144104803,
+      "grad_norm": 0.11072063446044922,
+      "learning_rate": 0.0012,
+      "loss": 3.5396492004394533,
+      "step": 640
+    },
+    {
+      "epoch": 9.0,
+      "eval_loss": 3.974152088165283,
+      "eval_runtime": 7.9166,
+      "eval_samples_per_second": 308.338,
+      "eval_steps_per_second": 9.726,
+      "step": 648
+    },
+    {
+      "epoch": 9.027947598253276,
+      "grad_norm": 0.20924493670463562,
+      "learning_rate": 0.0012,
+      "loss": 3.4809524536132814,
+      "step": 650
+    },
+    {
+      "epoch": 9.167685589519651,
+      "grad_norm": 0.27390941977500916,
+      "learning_rate": 0.0012,
+      "loss": 3.3156246185302733,
+      "step": 660
+    },
+    {
+      "epoch": 9.307423580786025,
+      "grad_norm": 0.20567722618579865,
+      "learning_rate": 0.0012,
+      "loss": 3.3422481536865236,
+      "step": 670
+    },
+    {
+      "epoch": 9.447161572052401,
+      "grad_norm": 0.1480301320552826,
+      "learning_rate": 0.0012,
+      "loss": 3.34552001953125,
+      "step": 680
+    },
+    {
+      "epoch": 9.586899563318777,
+      "grad_norm": 0.11175384372472763,
+      "learning_rate": 0.0012,
+      "loss": 3.3335189819335938,
+      "step": 690
+    },
+    {
+      "epoch": 9.726637554585153,
+      "grad_norm": 0.11760509014129639,
+      "learning_rate": 0.0012,
+      "loss": 3.355607604980469,
+      "step": 700
+    },
+    {
+      "epoch": 9.866375545851529,
+      "grad_norm": 0.11177218705415726,
+      "learning_rate": 0.0012,
+      "loss": 3.355641174316406,
+      "step": 710
+    },
+    {
+      "epoch": 10.0,
+      "grad_norm": 0.13623695075511932,
+      "learning_rate": 0.0012,
+      "loss": 3.355918121337891,
+      "step": 720
+    },
+    {
+      "epoch": 10.0,
+      "eval_loss": 3.9539742469787598,
+      "eval_runtime": 7.9182,
+      "eval_samples_per_second": 308.278,
+      "eval_steps_per_second": 9.724,
+      "step": 720
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 864,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 12,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": false,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 3.059564909802029e+18,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/runs/t0/t0_S5_lr0.0012/train.log b/runs/t0/t0_S5_lr0.0012/train.log
new file mode 100644
index 0000000000000000000000000000000000000000..fa7c122a2d7eace660056e948469880abfc3cd65
--- /dev/null
+++ b/runs/t0/t0_S5_lr0.0012/train.log
@@ -0,0 +1,2034 @@
+W0817 22:51:54.353000 2348811 torch/distributed/run.py:851] 
+W0817 22:51:54.353000 2348811 torch/distributed/run.py:851] *****************************************
+W0817 22:51:54.353000 2348811 torch/distributed/run.py:851] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+W0817 22:51:54.353000 2348811 torch/distributed/run.py:851] *****************************************
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+total params: 912519168
+[dataset] loaded packed cache: 36614 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+total params: 912519168
+[dataset] loaded packed cache: 36614 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+total params: 912519168
+[dataset] loaded packed cache: 36614 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+total params: 912519168
+[dataset] loaded packed cache: 36614 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+total params: 912519168
+[dataset] loaded packed cache: 36614 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+total params: 912519168
+[dataset] loaded packed cache: 36614 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+total params: 912519168
+[dataset] loaded packed cache: 36614 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+total params: 912519168
+[dataset] loaded packed cache: 36614 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+[transformers] ***** Running training *****
+[transformers]   Num examples = 36,614
+[transformers]   Num Epochs = 12
+[transformers]   Num update steps per epoch = 72
+[transformers]   Instantaneous batch size per device = 4
+[transformers]   Total train batch size (w. parallel, distributed & accumulation) = 512
+[transformers]   Gradient Accumulation steps = 16
+[transformers]   Total optimization steps = 864
+[transformers]   Number of trainable parameters = 912,519,168
+wandb: WARNING Changes to your `wandb` environment variables will be ignored because your `wandb` session has already started. For more information on how to modify your settings with `wandb.init()` arguments, please refer to https://wandb.me/wandb-init.
+wandb: setting up run qkbq14t6
+wandb: Tracking run with wandb version 0.28.2
+wandb: Run data is saved locally in /home/ubuntu/workspace/data-augmentations-for-pretraining/wandb/run-20260817_225224-qkbq14t6
+wandb: Run `wandb offline` to turn off syncing.
+wandb: Syncing run t0_S5_lr0.0012
+wandb: ⭐️ View project at https://wandb.ai/slowrun-wankers/data-aug-pretraining
+wandb: 🚀 View run at https://wandb.ai/slowrun-wankers/data-aug-pretraining/runs/qkbq14t6
+
  0%|          | 0/864 [00:00<?, ?it/s]wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+[rank1]:W0817 22:52:48.033000 2348935 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+[rank3]:W0817 22:52:48.128000 2348937 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+[rank4]:W0817 22:52:48.291000 2348939 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+[rank7]:W0817 22:52:48.547000 2348942 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+[rank6]:W0817 22:52:48.592000 2348941 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+[rank5]:W0817 22:52:48.923000 2348940 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+[rank2]:W0817 22:52:48.995000 2348936 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+[rank0]:W0817 22:52:49.199000 2348934 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+
  0%|          | 1/864 [00:46<11:07:32, 46.41s/it]
                                                  
{'loss': '12.24', 'grad_norm': '3.285', 'learning_rate': '0', 'epoch': '0.01397'}
+
  0%|          | 1/864 [00:46<11:07:32, 46.41s/it]
  0%|          | 2/864 [00:48<4:55:37, 20.58s/it] 
  0%|          | 3/864 [00:51<2:55:32, 12.23s/it]
  0%|          | 4/864 [00:53<1:59:10,  8.31s/it]
  1%|          | 5/864 [00:55<1:28:00,  6.15s/it]
  1%|          | 6/864 [00:58<1:09:12,  4.84s/it]
  1%|          | 7/864 [01:00<57:15,  4.01s/it]  
  1%|          | 8/864 [01:02<49:25,  3.46s/it]
  1%|          | 9/864 [01:05<44:09,  3.10s/it]
  1%|          | 10/864 [01:07<40:34,  2.85s/it]
                                                
{'loss': '10.77', 'grad_norm': '2.891', 'learning_rate': '0.000108', 'epoch': '0.1397'}
+
  1%|          | 10/864 [01:07<40:34,  2.85s/it]
  1%|▏         | 11/864 [01:09<38:07,  2.68s/it]
  1%|▏         | 12/864 [01:11<36:26,  2.57s/it]
  2%|▏         | 13/864 [01:14<35:14,  2.48s/it]
  2%|▏         | 14/864 [01:16<34:23,  2.43s/it]
  2%|▏         | 15/864 [01:18<33:47,  2.39s/it]
  2%|▏         | 16/864 [01:21<33:21,  2.36s/it]
  2%|▏         | 17/864 [01:23<33:02,  2.34s/it]
  2%|▏         | 18/864 [01:25<32:50,  2.33s/it]
  2%|▏         | 19/864 [01:27<32:41,  2.32s/it]
  2%|▏         | 20/864 [01:30<32:35,  2.32s/it]
                                                
{'loss': '8.708', 'grad_norm': '0.8952', 'learning_rate': '0.000228', 'epoch': '0.2795'}
+
  2%|▏         | 20/864 [01:30<32:35,  2.32s/it]
  2%|▏         | 21/864 [01:32<32:30,  2.31s/it]
  3%|▎         | 22/864 [01:34<32:24,  2.31s/it]
  3%|▎         | 23/864 [01:37<32:19,  2.31s/it]
  3%|▎         | 24/864 [01:39<32:15,  2.30s/it]
  3%|▎         | 25/864 [01:41<32:12,  2.30s/it]
  3%|▎         | 26/864 [01:44<32:09,  2.30s/it]
  3%|▎         | 27/864 [01:46<32:07,  2.30s/it]
  3%|▎         | 28/864 [01:48<32:04,  2.30s/it]
  3%|▎         | 29/864 [01:51<32:02,  2.30s/it]
  3%|▎         | 30/864 [01:53<33:46,  2.43s/it]
                                                
{'loss': '7.609', 'grad_norm': '0.8145', 'learning_rate': '0.000348', 'epoch': '0.4192'}
+
  3%|▎         | 30/864 [01:53<33:46,  2.43s/it]
  4%|▎         | 31/864 [01:56<33:12,  2.39s/it]
  4%|▎         | 32/864 [01:58<32:48,  2.37s/it]
  4%|▍         | 33/864 [02:00<32:32,  2.35s/it]
  4%|▍         | 34/864 [02:02<32:19,  2.34s/it]
  4%|▍         | 35/864 [02:05<32:11,  2.33s/it]
  4%|▍         | 36/864 [02:07<32:04,  2.32s/it]
  4%|▍         | 37/864 [02:09<31:58,  2.32s/it]
  4%|▍         | 38/864 [02:12<31:55,  2.32s/it]
  5%|▍         | 39/864 [02:14<31:52,  2.32s/it]
  5%|▍         | 40/864 [02:16<31:48,  2.32s/it]
                                                
{'loss': '7.432', 'grad_norm': '0.7392', 'learning_rate': '0.000468', 'epoch': '0.559'}
+
  5%|▍         | 40/864 [02:16<31:48,  2.32s/it]
  5%|▍         | 41/864 [02:19<31:46,  2.32s/it]
  5%|▍         | 42/864 [02:22<34:42,  2.53s/it]
  5%|▍         | 43/864 [02:24<33:46,  2.47s/it]
  5%|▌         | 44/864 [02:26<33:07,  2.42s/it]
  5%|▌         | 45/864 [02:29<32:39,  2.39s/it]
  5%|▌         | 46/864 [02:31<32:19,  2.37s/it]
  5%|▌         | 47/864 [02:33<32:04,  2.36s/it]
  6%|▌         | 48/864 [02:36<31:53,  2.34s/it]
  6%|▌         | 49/864 [02:38<31:44,  2.34s/it]
  6%|▌         | 50/864 [02:40<31:37,  2.33s/it]
                                                
{'loss': '7.194', 'grad_norm': '1.823', 'learning_rate': '0.000588', 'epoch': '0.6987'}
+
  6%|▌         | 50/864 [02:40<31:37,  2.33s/it]
  6%|▌         | 51/864 [02:43<31:32,  2.33s/it]
  6%|▌         | 52/864 [02:45<31:30,  2.33s/it]
  6%|▌         | 53/864 [02:47<31:24,  2.32s/it]
  6%|▋         | 54/864 [02:50<31:22,  2.32s/it]
  6%|▋         | 55/864 [02:52<31:19,  2.32s/it]
  6%|▋         | 56/864 [02:54<31:16,  2.32s/it]
  7%|▋         | 57/864 [02:57<31:14,  2.32s/it]
  7%|▋         | 58/864 [02:59<31:12,  2.32s/it]
  7%|▋         | 59/864 [03:01<31:08,  2.32s/it]
  7%|▋         | 60/864 [03:03<31:05,  2.32s/it]
                                                
{'loss': '6.985', 'grad_norm': '1.061', 'learning_rate': '0.000708', 'epoch': '0.8384'}
+
  7%|▋         | 60/864 [03:03<31:05,  2.32s/it]
  7%|▋         | 61/864 [03:06<31:03,  2.32s/it]
  7%|▋         | 62/864 [03:08<31:00,  2.32s/it]
  7%|▋         | 63/864 [03:10<30:58,  2.32s/it]
  7%|▋         | 64/864 [03:13<30:56,  2.32s/it]
  8%|▊         | 65/864 [03:15<30:54,  2.32s/it]
  8%|▊         | 66/864 [03:17<30:52,  2.32s/it]
  8%|▊         | 67/864 [03:20<30:50,  2.32s/it]
  8%|▊         | 68/864 [03:22<30:47,  2.32s/it]
  8%|▊         | 69/864 [03:24<30:45,  2.32s/it]
  8%|▊         | 70/864 [03:27<30:43,  2.32s/it]
                                                
{'loss': '6.869', 'grad_norm': '0.812', 'learning_rate': '0.000828', 'epoch': '0.9782'}
+
  8%|▊         | 70/864 [03:27<30:43,  2.32s/it]
  8%|▊         | 71/864 [03:29<30:41,  2.32s/it]
  8%|▊         | 72/864 [03:30<26:41,  2.02s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 4
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+
+
  0%|          | 0/77 [00:00<?, ?it/s]
+
  4%|▍         | 3/77 [00:00<00:04, 17.21it/s]
+
  6%|▋         | 5/77 [00:00<00:05, 12.55it/s]
+
  9%|▉         | 7/77 [00:00<00:06, 11.25it/s]
+
 12%|█▏        | 9/77 [00:00<00:06, 10.65it/s]
+
 14%|█▍        | 11/77 [00:00<00:06, 10.33it/s]
+
 17%|█▋        | 13/77 [00:01<00:06, 10.14it/s]
+
 19%|█▉        | 15/77 [00:01<00:06, 10.03it/s]
+
 22%|██▏       | 17/77 [00:01<00:06,  9.94it/s]
+
 23%|██▎       | 18/77 [00:01<00:05,  9.91it/s]
+
 25%|██▍       | 19/77 [00:01<00:05,  9.87it/s]
+
 26%|██▌       | 20/77 [00:01<00:05,  9.86it/s]
+
 27%|██▋       | 21/77 [00:02<00:05,  9.84it/s]
+
 29%|██▊       | 22/77 [00:02<00:05,  9.83it/s]
+
 30%|██▉       | 23/77 [00:02<00:05,  9.82it/s]
+
 31%|███       | 24/77 [00:02<00:05,  9.81it/s]
+
 32%|███▏      | 25/77 [00:02<00:05,  9.80it/s]
+
 34%|███▍      | 26/77 [00:02<00:05,  9.78it/s]
+
 35%|███▌      | 27/77 [00:02<00:05,  9.79it/s]
+
 36%|███▋      | 28/77 [00:02<00:05,  9.78it/s]
+
 38%|███▊      | 29/77 [00:02<00:04,  9.79it/s]
+
 39%|███▉      | 30/77 [00:02<00:04,  9.79it/s]
+
 40%|████      | 31/77 [00:03<00:04,  9.79it/s]
+
 42%|████▏     | 32/77 [00:03<00:04,  9.79it/s]
+
 43%|████▎     | 33/77 [00:03<00:04,  9.79it/s]
+
 44%|████▍     | 34/77 [00:03<00:04,  9.78it/s]
+
 45%|████▌     | 35/77 [00:03<00:04,  9.78it/s]
+
 47%|████▋     | 36/77 [00:03<00:04,  9.78it/s]
+
 48%|████▊     | 37/77 [00:03<00:04,  9.78it/s]
+
 49%|████▉     | 38/77 [00:03<00:03,  9.77it/s]
+
 51%|█████     | 39/77 [00:03<00:03,  9.77it/s]
+
 52%|█████▏    | 40/77 [00:03<00:03,  9.76it/s]
+
 53%|█████▎    | 41/77 [00:04<00:03,  9.76it/s]
+
 55%|█████▍    | 42/77 [00:04<00:03,  9.77it/s]
+
 56%|█████▌    | 43/77 [00:04<00:03,  9.77it/s]
+
 57%|█████▋    | 44/77 [00:04<00:03,  9.78it/s]
+
 58%|█████▊    | 45/77 [00:04<00:03,  9.78it/s]
+
 60%|█████▉    | 46/77 [00:04<00:03,  9.78it/s]
+
 61%|██████    | 47/77 [00:04<00:03,  9.78it/s]
+
 62%|██████▏   | 48/77 [00:04<00:02,  9.78it/s]
+
 64%|██████▎   | 49/77 [00:04<00:02,  9.78it/s]
+
 65%|██████▍   | 50/77 [00:04<00:02,  9.77it/s]
+
 66%|██████▌   | 51/77 [00:05<00:02,  9.77it/s]
+
 68%|██████▊   | 52/77 [00:05<00:02,  9.78it/s]
+
 69%|██████▉   | 53/77 [00:05<00:02,  9.78it/s]
+
 70%|███████   | 54/77 [00:05<00:02,  9.77it/s]
+
 71%|███████▏  | 55/77 [00:05<00:02,  9.79it/s]
+
 73%|███████▎  | 56/77 [00:05<00:02,  9.79it/s]
+
 74%|███████▍  | 57/77 [00:05<00:02,  9.78it/s]
+
 75%|███████▌  | 58/77 [00:05<00:01,  9.79it/s]
+
 77%|███████▋  | 59/77 [00:05<00:01,  9.79it/s]
+
 78%|███████▊  | 60/77 [00:06<00:01,  9.76it/s]
+
 79%|███████▉  | 61/77 [00:06<00:01,  9.78it/s]
+
 81%|████████  | 62/77 [00:06<00:01,  9.79it/s]
+
 82%|████████▏ | 63/77 [00:06<00:01,  9.79it/s]
+
 83%|████████▎ | 64/77 [00:06<00:01,  9.79it/s]
+
 84%|████████▍ | 65/77 [00:06<00:01,  9.79it/s]
+
 86%|████████▌ | 66/77 [00:06<00:01,  9.79it/s]
+
 87%|████████▋ | 67/77 [00:06<00:01,  9.78it/s]
+
 88%|████████▊ | 68/77 [00:06<00:00,  9.79it/s]
+
 90%|████████▉ | 69/77 [00:06<00:00,  9.79it/s]
+
 91%|█████████ | 70/77 [00:07<00:00,  9.79it/s]
+
 92%|█████████▏| 71/77 [00:07<00:00,  9.80it/s]
+
 94%|█████████▎| 72/77 [00:07<00:00,  9.79it/s]
+
 95%|█████████▍| 73/77 [00:07<00:00,  9.80it/s]
+
 96%|█████████▌| 74/77 [00:07<00:00,  9.79it/s]
+
 97%|█████████▋| 75/77 [00:07<00:00,  9.79it/s]
+
 99%|█████████▊| 76/77 [00:07<00:00,  9.79it/s]
+
100%|██████████| 77/77 [00:07<00:00,  9.79it/s]
                                                
+
                                               
{'eval_loss': '6.711', 'eval_runtime': '32.17', 'eval_samples_per_second': '75.88', 'eval_steps_per_second': '2.394', 'epoch': '1'}
+
  8%|▊         | 72/864 [04:03<26:41,  2.02s/it]
+
100%|██████████| 77/77 [00:07<00:00,  9.79it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0012/checkpoint-72
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0012/checkpoint-72/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0012/checkpoint-72/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:03<00:00,  3.64s/it]
Writing model shards: 100%|██████████| 1/1 [00:03<00:00,  3.64s/it]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0012/checkpoint-72/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0012/checkpoint-72/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0012/checkpoint-72/tokenizer_config.json
+[eval] epoch 1.0 step 72 loss 6.7112 (best 6.7112 @ ep 1.0, 0/10000 since best)
+
+
                                               
  8%|▊         | 73/864 [04:15<3:16:54, 14.94s/it]
  9%|▊         | 74/864 [04:18<2:26:49, 11.15s/it]
  9%|▊         | 75/864 [04:20<1:51:49,  8.50s/it]
  9%|▉         | 76/864 [04:22<1:27:20,  6.65s/it]
  9%|▉         | 77/864 [04:25<1:10:12,  5.35s/it]
  9%|▉         | 78/864 [04:27<58:13,  4.44s/it]  
  9%|▉         | 79/864 [04:29<49:50,  3.81s/it]
  9%|▉         | 80/864 [04:32<43:57,  3.36s/it]
                                                
{'loss': '6.636', 'grad_norm': '0.6915', 'learning_rate': '0.000948', 'epoch': '1.112'}
+
  9%|▉         | 80/864 [04:32<43:57,  3.36s/it]
  9%|▉         | 81/864 [04:34<39:50,  3.05s/it]
  9%|▉         | 82/864 [04:36<36:58,  2.84s/it]
 10%|▉         | 83/864 [04:39<34:55,  2.68s/it]
 10%|▉         | 84/864 [04:41<33:28,  2.58s/it]
 10%|▉         | 85/864 [04:43<32:28,  2.50s/it]
 10%|▉         | 86/864 [04:46<31:45,  2.45s/it]
 10%|█         | 87/864 [04:48<31:13,  2.41s/it]
 10%|█         | 88/864 [04:50<30:51,  2.39s/it]
 10%|█         | 89/864 [04:53<30:35,  2.37s/it]
 10%|█         | 90/864 [04:55<30:23,  2.36s/it]
                                                
{'loss': '6.488', 'grad_norm': '0.7802', 'learning_rate': '0.001068', 'epoch': '1.252'}
+
 10%|█         | 90/864 [04:55<30:23,  2.36s/it]
 11%|█         | 91/864 [04:57<30:13,  2.35s/it]
 11%|█         | 92/864 [05:00<30:06,  2.34s/it]
 11%|█         | 93/864 [05:02<30:01,  2.34s/it]
 11%|█         | 94/864 [05:04<29:56,  2.33s/it]
 11%|█         | 95/864 [05:07<29:53,  2.33s/it]
 11%|█         | 96/864 [05:09<29:49,  2.33s/it]
 11%|█         | 97/864 [05:11<29:45,  2.33s/it]
 11%|█▏        | 98/864 [05:14<29:42,  2.33s/it]
 11%|█▏        | 99/864 [05:16<29:39,  2.33s/it]
 12%|█▏        | 100/864 [05:18<29:37,  2.33s/it]
                                                 
{'loss': '6.355', 'grad_norm': '0.9558', 'learning_rate': '0.001188', 'epoch': '1.391'}
+
 12%|█▏        | 100/864 [05:18<29:37,  2.33s/it]
 12%|█▏        | 101/864 [05:21<29:34,  2.33s/it]
 12%|█▏        | 102/864 [05:23<29:32,  2.33s/it]
 12%|█▏        | 103/864 [05:25<29:30,  2.33s/it]
 12%|█▏        | 104/864 [05:27<29:27,  2.33s/it]
 12%|█▏        | 105/864 [05:30<29:25,  2.33s/it]
 12%|█▏        | 106/864 [05:32<29:23,  2.33s/it]
 12%|█▏        | 107/864 [05:34<29:21,  2.33s/it]
 12%|█▎        | 108/864 [05:37<29:18,  2.33s/it]
 13%|█▎        | 109/864 [05:39<29:15,  2.33s/it]
 13%|█▎        | 110/864 [05:41<29:13,  2.33s/it]
                                                 
{'loss': '6.276', 'grad_norm': '0.7227', 'learning_rate': '0.0012', 'epoch': '1.531'}
+
 13%|█▎        | 110/864 [05:41<29:13,  2.33s/it]
 13%|█▎        | 111/864 [05:44<29:10,  2.32s/it]
 13%|█▎        | 112/864 [05:46<29:08,  2.33s/it]
 13%|█▎        | 113/864 [05:48<29:07,  2.33s/it]
 13%|█▎        | 114/864 [05:51<29:04,  2.33s/it]
 13%|█▎        | 115/864 [05:53<29:02,  2.33s/it]
 13%|█▎        | 116/864 [05:55<29:00,  2.33s/it]
 14%|█▎        | 117/864 [05:58<28:57,  2.33s/it]
 14%|█▎        | 118/864 [06:00<28:54,  2.33s/it]
 14%|█▍        | 119/864 [06:02<28:52,  2.33s/it]
 14%|█▍        | 120/864 [06:05<28:49,  2.32s/it]
                                                 
{'loss': '6.149', 'grad_norm': '2.057', 'learning_rate': '0.0012', 'epoch': '1.671'}
+
 14%|█▍        | 120/864 [06:05<28:49,  2.32s/it]
 14%|█▍        | 121/864 [06:07<28:46,  2.32s/it]
 14%|█▍        | 122/864 [06:09<28:45,  2.33s/it]
 14%|█▍        | 123/864 [06:12<28:43,  2.33s/it]
 14%|█▍        | 124/864 [06:14<28:41,  2.33s/it]
 14%|█▍        | 125/864 [06:16<28:38,  2.33s/it]
 15%|█▍        | 126/864 [06:19<28:36,  2.33s/it]
 15%|█▍        | 127/864 [06:21<28:34,  2.33s/it]
 15%|█▍        | 128/864 [06:23<28:32,  2.33s/it]
 15%|█▍        | 129/864 [06:26<28:29,  2.33s/it]
 15%|█▌        | 130/864 [06:28<28:27,  2.33s/it]
                                                 
{'loss': '6.129', 'grad_norm': '0.3692', 'learning_rate': '0.0012', 'epoch': '1.81'}
+
 15%|█▌        | 130/864 [06:28<28:27,  2.33s/it]
 15%|█▌        | 131/864 [06:30<28:25,  2.33s/it]
 15%|█▌        | 132/864 [06:33<28:22,  2.33s/it]
 15%|█▌        | 133/864 [06:35<28:21,  2.33s/it]
 16%|█▌        | 134/864 [06:37<28:18,  2.33s/it]
 16%|█▌        | 135/864 [06:40<28:15,  2.33s/it]
 16%|█▌        | 136/864 [06:42<28:13,  2.33s/it]
 16%|█▌        | 137/864 [06:44<28:11,  2.33s/it]
 16%|█▌        | 138/864 [06:47<28:09,  2.33s/it]
 16%|█▌        | 139/864 [06:49<28:07,  2.33s/it]
 16%|█▌        | 140/864 [06:51<28:04,  2.33s/it]
                                                 
{'loss': '5.983', 'grad_norm': '0.3523', 'learning_rate': '0.0012', 'epoch': '1.95'}
+
 16%|█▌        | 140/864 [06:51<28:04,  2.33s/it]
 16%|█▋        | 141/864 [06:54<28:02,  2.33s/it]
 16%|█▋        | 142/864 [06:56<28:00,  2.33s/it]
 17%|█▋        | 143/864 [06:58<27:58,  2.33s/it]
 17%|█▋        | 144/864 [07:00<24:20,  2.03s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 4
+
+
  0%|          | 0/77 [00:00<?, ?it/s]
+
  3%|▎         | 2/77 [00:00<00:03, 19.81it/s]
+
  5%|▌         | 4/77 [00:00<00:05, 12.39it/s]
+
  8%|▊         | 6/77 [00:00<00:06, 11.00it/s]
+
 10%|█         | 8/77 [00:00<00:06, 10.50it/s]
+
 13%|█▎        | 10/77 [00:00<00:06, 10.22it/s]
+
 16%|█▌        | 12/77 [00:01<00:06, 10.08it/s]
+
 18%|█▊        | 14/77 [00:01<00:06,  9.97it/s]
+
 21%|██        | 16/77 [00:01<00:06,  9.92it/s]
+
 23%|██▎       | 18/77 [00:01<00:05,  9.87it/s]
+
 25%|██▍       | 19/77 [00:01<00:05,  9.86it/s]
+
 26%|██▌       | 20/77 [00:01<00:05,  9.84it/s]
+
 27%|██▋       | 21/77 [00:02<00:05,  9.82it/s]
+
 29%|██▊       | 22/77 [00:02<00:05,  9.80it/s]
+
 30%|██▉       | 23/77 [00:02<00:05,  9.81it/s]
+
 31%|███       | 24/77 [00:02<00:05,  9.76it/s]
+
 32%|███▏      | 25/77 [00:02<00:05,  9.79it/s]
+
 34%|███▍      | 26/77 [00:02<00:05,  9.78it/s]
+
 35%|███▌      | 27/77 [00:02<00:05,  9.80it/s]
+
 36%|███▋      | 28/77 [00:02<00:05,  9.79it/s]
+
 38%|███▊      | 29/77 [00:02<00:04,  9.79it/s]
+
 39%|███▉      | 30/77 [00:02<00:04,  9.77it/s]
+
 40%|████      | 31/77 [00:03<00:04,  9.78it/s]
+
 42%|████▏     | 32/77 [00:03<00:04,  9.78it/s]
+
 43%|████▎     | 33/77 [00:03<00:04,  9.78it/s]
+
 44%|████▍     | 34/77 [00:03<00:04,  9.77it/s]
+
 45%|████▌     | 35/77 [00:03<00:04,  9.79it/s]
+
 47%|████▋     | 36/77 [00:03<00:04,  9.79it/s]
+
 48%|████▊     | 37/77 [00:03<00:04,  9.78it/s]
+
 49%|████▉     | 38/77 [00:03<00:03,  9.77it/s]
+
 51%|█████     | 39/77 [00:03<00:03,  9.78it/s]
+
 52%|█████▏    | 40/77 [00:03<00:03,  9.78it/s]
+
 53%|█████▎    | 41/77 [00:04<00:03,  9.78it/s]
+
 55%|█████▍    | 42/77 [00:04<00:03,  9.77it/s]
+
 56%|█████▌    | 43/77 [00:04<00:03,  9.79it/s]
+
 57%|█████▋    | 44/77 [00:04<00:03,  9.79it/s]
+
 58%|█████▊    | 45/77 [00:04<00:03,  9.78it/s]
+
 60%|█████▉    | 46/77 [00:04<00:03,  9.76it/s]
+
 61%|██████    | 47/77 [00:04<00:03,  9.79it/s]
+
 62%|██████▏   | 48/77 [00:04<00:02,  9.79it/s]
+
 64%|██████▎   | 49/77 [00:04<00:02,  9.78it/s]
+
 65%|██████▍   | 50/77 [00:05<00:02,  9.77it/s]
+
 66%|██████▌   | 51/77 [00:05<00:02,  9.79it/s]
+
 68%|██████▊   | 52/77 [00:05<00:02,  9.78it/s]
+
 69%|██████▉   | 53/77 [00:05<00:02,  9.78it/s]
+
 70%|███████   | 54/77 [00:05<00:02,  9.77it/s]
+
 71%|███████▏  | 55/77 [00:05<00:02,  9.79it/s]
+
 73%|███████▎  | 56/77 [00:05<00:02,  9.79it/s]
+
 74%|███████▍  | 57/77 [00:05<00:02,  9.78it/s]
+
 75%|███████▌  | 58/77 [00:05<00:01,  9.77it/s]
+
 77%|███████▋  | 59/77 [00:05<00:01,  9.78it/s]
+
 78%|███████▊  | 60/77 [00:06<00:01,  9.78it/s]
+
 79%|███████▉  | 61/77 [00:06<00:01,  9.79it/s]
+
 81%|████████  | 62/77 [00:06<00:01,  9.79it/s]
+
 82%|████████▏ | 63/77 [00:06<00:01,  9.79it/s]
+
 83%|████████▎ | 64/77 [00:06<00:01,  9.78it/s]
+
 84%|████████▍ | 65/77 [00:06<00:01,  9.79it/s]
+
 86%|████████▌ | 66/77 [00:06<00:01,  9.79it/s]
+
 87%|████████▋ | 67/77 [00:06<00:01,  9.79it/s]
+
 88%|████████▊ | 68/77 [00:06<00:00,  9.79it/s]
+
 90%|████████▉ | 69/77 [00:06<00:00,  9.79it/s]
+
 91%|█████████ | 70/77 [00:07<00:00,  9.78it/s]
+
 92%|█████████▏| 71/77 [00:07<00:00,  9.78it/s]
+
 94%|█████████▎| 72/77 [00:07<00:00,  9.79it/s]
+
 95%|█████████▍| 73/77 [00:07<00:00,  9.79it/s]
+
 96%|█████████▌| 74/77 [00:07<00:00,  9.78it/s]
+
 97%|█████████▋| 75/77 [00:07<00:00,  9.79it/s]
+
 99%|█████████▊| 76/77 [00:07<00:00,  9.78it/s]
+
100%|██████████| 77/77 [00:07<00:00,  9.78it/s]
                                                 
+
                                               
{'eval_loss': '6.093', 'eval_runtime': '7.914', 'eval_samples_per_second': '308.5', 'eval_steps_per_second': '9.73', 'epoch': '2'}
+
 17%|█▋        | 144/864 [07:07<24:20,  2.03s/it]
+
100%|██████████| 77/77 [00:07<00:00,  9.78it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0012/checkpoint-144
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0012/checkpoint-144/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0012/checkpoint-144/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:03<00:00,  3.61s/it]
Writing model shards: 100%|██████████| 1/1 [00:03<00:00,  3.61s/it]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0012/checkpoint-144/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0012/checkpoint-144/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0012/checkpoint-144/tokenizer_config.json
+[eval] epoch 2.0 step 144 loss 6.0932 (best 6.0932 @ ep 2.0, 0/10000 since best)
+
+
                                               
 17%|█▋        | 145/864 [07:21<1:35:25,  7.96s/it]
 17%|█▋        | 146/864 [07:24<1:15:01,  6.27s/it]
 17%|█▋        | 147/864 [07:26<1:00:45,  5.08s/it]
 17%|█▋        | 148/864 [07:28<50:47,  4.26s/it]  
 17%|█▋        | 149/864 [07:31<43:47,  3.68s/it]
 17%|█▋        | 150/864 [07:33<38:54,  3.27s/it]
                                                 
{'loss': '5.902', 'grad_norm': '0.4088', 'learning_rate': '0.0012', 'epoch': '2.084'}
+
 17%|█▋        | 150/864 [07:33<38:54,  3.27s/it]
 17%|█▋        | 151/864 [07:35<35:28,  2.99s/it]
 18%|█▊        | 152/864 [07:38<33:03,  2.79s/it]
 18%|█▊        | 153/864 [07:40<31:21,  2.65s/it]
 18%|█▊        | 154/864 [07:42<30:09,  2.55s/it]
 18%|█▊        | 155/864 [07:45<29:18,  2.48s/it]
 18%|█▊        | 156/864 [07:47<28:42,  2.43s/it]
 18%|█▊        | 157/864 [07:49<28:16,  2.40s/it]
 18%|█▊        | 158/864 [07:52<27:57,  2.38s/it]
 18%|█▊        | 159/864 [07:54<27:43,  2.36s/it]
 19%|█▊        | 160/864 [07:56<27:32,  2.35s/it]
                                                 
{'loss': '5.776', 'grad_norm': '0.4431', 'learning_rate': '0.0012', 'epoch': '2.224'}
+
 19%|█▊        | 160/864 [07:56<27:32,  2.35s/it]
 19%|█▊        | 161/864 [07:58<27:24,  2.34s/it]
 19%|█▉        | 162/864 [08:01<27:18,  2.33s/it]
 19%|█▉        | 163/864 [08:03<27:13,  2.33s/it]
 19%|█▉        | 164/864 [08:05<27:08,  2.33s/it]
 19%|█▉        | 165/864 [08:08<27:05,  2.33s/it]
 19%|█▉        | 166/864 [08:10<27:01,  2.32s/it]
 19%|█▉        | 167/864 [08:12<26:58,  2.32s/it]
 19%|█▉        | 168/864 [08:15<26:56,  2.32s/it]
 20%|█▉        | 169/864 [08:17<26:53,  2.32s/it]
 20%|█▉        | 170/864 [08:19<26:50,  2.32s/it]
                                                 
{'loss': '5.73', 'grad_norm': '0.5734', 'learning_rate': '0.0012', 'epoch': '2.363'}
+
 20%|█▉        | 170/864 [08:19<26:50,  2.32s/it]
 20%|█▉        | 171/864 [08:22<26:48,  2.32s/it]
 20%|█▉        | 172/864 [08:24<26:45,  2.32s/it]
 20%|██        | 173/864 [08:26<26:43,  2.32s/it]
 20%|██        | 174/864 [08:29<26:40,  2.32s/it]
 20%|██        | 175/864 [08:31<26:37,  2.32s/it]
 20%|██        | 176/864 [08:33<26:35,  2.32s/it]
 20%|██        | 177/864 [08:36<26:33,  2.32s/it]
 21%|██        | 178/864 [08:38<26:31,  2.32s/it]
 21%|██        | 179/864 [08:40<26:29,  2.32s/it]
 21%|██        | 180/864 [08:43<26:26,  2.32s/it]
                                                 
{'loss': '5.651', 'grad_norm': '0.3401', 'learning_rate': '0.0012', 'epoch': '2.503'}
+
 21%|██        | 180/864 [08:43<26:26,  2.32s/it]
 21%|██        | 181/864 [08:45<26:24,  2.32s/it]
 21%|██        | 182/864 [08:47<26:21,  2.32s/it]
 21%|██        | 183/864 [08:50<26:19,  2.32s/it]
 21%|██▏       | 184/864 [08:52<26:17,  2.32s/it]
 21%|██▏       | 185/864 [08:54<26:15,  2.32s/it]
 22%|██▏       | 186/864 [08:56<26:12,  2.32s/it]
 22%|██▏       | 187/864 [08:59<26:10,  2.32s/it]
 22%|██▏       | 188/864 [09:01<26:07,  2.32s/it]
 22%|██▏       | 189/864 [09:03<26:05,  2.32s/it]
 22%|██▏       | 190/864 [09:06<27:14,  2.42s/it]
                                                 
{'loss': '5.603', 'grad_norm': '0.7364', 'learning_rate': '0.0012', 'epoch': '2.643'}
+
 22%|██▏       | 190/864 [09:06<27:14,  2.42s/it]
 22%|██▏       | 191/864 [09:08<26:49,  2.39s/it]
 22%|██▏       | 192/864 [09:11<26:32,  2.37s/it]
 22%|██▏       | 193/864 [09:13<26:19,  2.35s/it]
 22%|██▏       | 194/864 [09:15<26:10,  2.34s/it]
 23%|██▎       | 195/864 [09:18<26:03,  2.34s/it]
 23%|██▎       | 196/864 [09:20<25:57,  2.33s/it]
 23%|██▎       | 197/864 [09:22<25:52,  2.33s/it]
 23%|██▎       | 198/864 [09:25<25:49,  2.33s/it]
 23%|██▎       | 199/864 [09:27<25:45,  2.32s/it]
 23%|██▎       | 200/864 [09:29<25:42,  2.32s/it]
                                                 
{'loss': '5.504', 'grad_norm': '0.2778', 'learning_rate': '0.0012', 'epoch': '2.783'}
+
 23%|██▎       | 200/864 [09:29<25:42,  2.32s/it]
 23%|██▎       | 201/864 [09:32<25:39,  2.32s/it]
 23%|██▎       | 202/864 [09:34<25:36,  2.32s/it]
 23%|██▎       | 203/864 [09:36<25:33,  2.32s/it]
 24%|██▎       | 204/864 [09:39<25:31,  2.32s/it]
 24%|██▎       | 205/864 [09:41<25:29,  2.32s/it]
 24%|██▍       | 206/864 [09:44<26:43,  2.44s/it]
 24%|██▍       | 207/864 [09:46<27:27,  2.51s/it]
 24%|██▍       | 208/864 [09:49<26:47,  2.45s/it]
 24%|██▍       | 209/864 [09:51<26:19,  2.41s/it]
 24%|██▍       | 210/864 [09:53<25:59,  2.38s/it]
                                                 
{'loss': '5.419', 'grad_norm': '0.5734', 'learning_rate': '0.0012', 'epoch': '2.922'}
+
 24%|██▍       | 210/864 [09:53<25:59,  2.38s/it]
 24%|██▍       | 211/864 [09:56<25:45,  2.37s/it]
 25%|██▍       | 212/864 [09:58<25:33,  2.35s/it]
 25%|██▍       | 213/864 [10:00<25:25,  2.34s/it]
 25%|██▍       | 214/864 [10:03<25:18,  2.34s/it]
 25%|██▍       | 215/864 [10:05<25:13,  2.33s/it]
 25%|██▌       | 216/864 [10:06<21:55,  2.03s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 4
+
+
  0%|          | 0/77 [00:00<?, ?it/s]
+
  3%|▎         | 2/77 [00:00<00:03, 19.99it/s]
+
  5%|▌         | 4/77 [00:00<00:05, 12.38it/s]
+
  8%|▊         | 6/77 [00:00<00:06, 11.03it/s]
+
 10%|█         | 8/77 [00:00<00:06, 10.51it/s]
+
 13%|█▎        | 10/77 [00:00<00:06, 10.22it/s]
+
 16%|█▌        | 12/77 [00:01<00:06, 10.08it/s]
+
 18%|█▊        | 14/77 [00:01<00:06,  9.97it/s]
+
 21%|██        | 16/77 [00:01<00:06,  9.91it/s]
+
 23%|██▎       | 18/77 [00:01<00:05,  9.87it/s]
+
 25%|██▍       | 19/77 [00:01<00:05,  9.85it/s]
+
 26%|██▌       | 20/77 [00:01<00:05,  9.84it/s]
+
 27%|██▋       | 21/77 [00:02<00:05,  9.83it/s]
+
 29%|██▊       | 22/77 [00:02<00:05,  9.82it/s]
+
 30%|██▉       | 23/77 [00:02<00:05,  9.81it/s]
+
 31%|███       | 24/77 [00:02<00:05,  9.80it/s]
+
 32%|███▏      | 25/77 [00:02<00:05,  9.79it/s]
+
 34%|███▍      | 26/77 [00:02<00:05,  9.79it/s]
+
 35%|███▌      | 27/77 [00:02<00:05,  9.78it/s]
+
 36%|███▋      | 28/77 [00:02<00:05,  9.78it/s]
+
 38%|███▊      | 29/77 [00:02<00:04,  9.77it/s]
+
 39%|███▉      | 30/77 [00:02<00:04,  9.78it/s]
+
 40%|████      | 31/77 [00:03<00:04,  9.79it/s]
+
 42%|████▏     | 32/77 [00:03<00:04,  9.79it/s]
+
 43%|████▎     | 33/77 [00:03<00:04,  9.78it/s]
+
 44%|████▍     | 34/77 [00:03<00:04,  9.78it/s]
+
 45%|████▌     | 35/77 [00:03<00:04,  9.78it/s]
+
 47%|████▋     | 36/77 [00:03<00:04,  9.78it/s]
+
 48%|████▊     | 37/77 [00:03<00:04,  9.78it/s]
+
 49%|████▉     | 38/77 [00:03<00:03,  9.78it/s]
+
 51%|█████     | 39/77 [00:03<00:03,  9.78it/s]
+
 52%|█████▏    | 40/77 [00:03<00:03,  9.78it/s]
+
 53%|█████▎    | 41/77 [00:04<00:03,  9.78it/s]
+
 55%|█████▍    | 42/77 [00:04<00:03,  9.78it/s]
+
 56%|█████▌    | 43/77 [00:04<00:03,  9.78it/s]
+
 57%|█████▋    | 44/77 [00:04<00:03,  9.77it/s]
+
 58%|█████▊    | 45/77 [00:04<00:03,  9.78it/s]
+
 60%|█████▉    | 46/77 [00:04<00:03,  9.78it/s]
+
 61%|██████    | 47/77 [00:04<00:03,  9.78it/s]
+
 62%|██████▏   | 48/77 [00:04<00:02,  9.78it/s]
+
 64%|██████▎   | 49/77 [00:04<00:02,  9.77it/s]
+
 65%|██████▍   | 50/77 [00:05<00:02,  9.78it/s]
+
 66%|██████▌   | 51/77 [00:05<00:02,  9.78it/s]
+
 68%|██████▊   | 52/77 [00:05<00:02,  9.78it/s]
+
 69%|██████▉   | 53/77 [00:05<00:02,  9.79it/s]
+
 70%|███████   | 54/77 [00:05<00:02,  9.77it/s]
+
 71%|███████▏  | 55/77 [00:05<00:02,  9.78it/s]
+
 73%|███████▎  | 56/77 [00:05<00:02,  9.76it/s]
+
 74%|███████▍  | 57/77 [00:05<00:02,  9.76it/s]
+
 75%|███████▌  | 58/77 [00:05<00:01,  9.76it/s]
+
 77%|███████▋  | 59/77 [00:05<00:01,  9.76it/s]
+
 78%|███████▊  | 60/77 [00:06<00:01,  9.76it/s]
+
 79%|███████▉  | 61/77 [00:06<00:01,  9.79it/s]
+
 81%|████████  | 62/77 [00:06<00:01,  9.79it/s]
+
 82%|████████▏ | 63/77 [00:06<00:01,  9.79it/s]
+
 83%|████████▎ | 64/77 [00:06<00:01,  9.79it/s]
+
 84%|████████▍ | 65/77 [00:06<00:01,  9.79it/s]
+
 86%|████████▌ | 66/77 [00:06<00:01,  9.80it/s]
+
 87%|████████▋ | 67/77 [00:06<00:01,  9.80it/s]
+
 88%|████████▊ | 68/77 [00:06<00:00,  9.79it/s]
+
 90%|████████▉ | 69/77 [00:06<00:00,  9.79it/s]
+
 91%|█████████ | 70/77 [00:07<00:00,  9.79it/s]
+
 92%|█████████▏| 71/77 [00:07<00:00,  9.79it/s]
+
 94%|█████████▎| 72/77 [00:07<00:00,  9.79it/s]
+
 95%|█████████▍| 73/77 [00:07<00:00,  9.79it/s]
+
 96%|█████████▌| 74/77 [00:07<00:00,  9.79it/s]
+
 97%|█████████▋| 75/77 [00:07<00:00,  9.79it/s]
+
 99%|█████████▊| 76/77 [00:07<00:00,  9.79it/s]
+
100%|██████████| 77/77 [00:07<00:00,  9.79it/s]
                                                 
+
                                               
{'eval_loss': '5.41', 'eval_runtime': '7.915', 'eval_samples_per_second': '308.4', 'eval_steps_per_second': '9.729', 'epoch': '3'}
+
 25%|██▌       | 216/864 [10:14<21:55,  2.03s/it]
+
100%|██████████| 77/77 [00:07<00:00,  9.79it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0012/checkpoint-216
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0012/checkpoint-216/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0012/checkpoint-216/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:03<00:00,  3.59s/it]
Writing model shards: 100%|██████████| 1/1 [00:03<00:00,  3.59s/it]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0012/checkpoint-216/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0012/checkpoint-216/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0012/checkpoint-216/tokenizer_config.json
+[eval] epoch 3.0 step 216 loss 5.4101 (best 5.4101 @ ep 3.0, 0/10000 since best)
+
+
                                               
 25%|██▌       | 217/864 [10:27<1:24:14,  7.81s/it]
 25%|██▌       | 218/864 [10:30<1:06:21,  6.16s/it]
 25%|██▌       | 219/864 [10:32<53:52,  5.01s/it]  
 25%|██▌       | 220/864 [10:34<45:08,  4.21s/it]
                                                 
{'loss': '5.341', 'grad_norm': '0.2274', 'learning_rate': '0.0012', 'epoch': '3.056'}
+
 25%|██▌       | 220/864 [10:34<45:08,  4.21s/it]
 26%|██▌       | 221/864 [10:37<39:01,  3.64s/it]
 26%|██▌       | 222/864 [10:39<34:43,  3.25s/it]
 26%|██▌       | 223/864 [10:41<31:43,  2.97s/it]
 26%|██▌       | 224/864 [10:44<29:36,  2.78s/it]
 26%|██▌       | 225/864 [10:46<28:07,  2.64s/it]
 26%|██▌       | 226/864 [10:48<27:03,  2.55s/it]
 26%|██▋       | 227/864 [10:51<26:20,  2.48s/it]
 26%|██▋       | 228/864 [10:53<25:47,  2.43s/it]
 27%|██▋       | 229/864 [10:55<25:23,  2.40s/it]
 27%|██▋       | 230/864 [10:58<25:06,  2.38s/it]
                                                 
{'loss': '5.247', 'grad_norm': '0.3654', 'learning_rate': '0.0012', 'epoch': '3.196'}
+
 27%|██▋       | 230/864 [10:58<25:06,  2.38s/it]
 27%|██▋       | 231/864 [11:00<24:54,  2.36s/it]
 27%|██▋       | 232/864 [11:02<24:45,  2.35s/it]
 27%|██▋       | 233/864 [11:05<24:38,  2.34s/it]
 27%|██▋       | 234/864 [11:07<24:31,  2.34s/it]
 27%|██▋       | 235/864 [11:09<24:27,  2.33s/it]
 27%|██▋       | 236/864 [11:12<24:23,  2.33s/it]
 27%|██▋       | 237/864 [11:14<24:20,  2.33s/it]
 28%|██▊       | 238/864 [11:16<24:16,  2.33s/it]
 28%|██▊       | 239/864 [11:19<24:14,  2.33s/it]
 28%|██▊       | 240/864 [11:21<24:11,  2.33s/it]
                                                 
{'loss': '5.183', 'grad_norm': '0.3052', 'learning_rate': '0.0012', 'epoch': '3.335'}
+
 28%|██▊       | 240/864 [11:21<24:11,  2.33s/it]
 28%|██▊       | 241/864 [11:23<24:09,  2.33s/it]
 28%|██▊       | 242/864 [11:26<24:06,  2.33s/it]
 28%|██▊       | 243/864 [11:28<24:03,  2.32s/it]
 28%|██▊       | 244/864 [11:30<24:00,  2.32s/it]
 28%|██▊       | 245/864 [11:33<23:58,  2.32s/it]
 28%|██▊       | 246/864 [11:35<23:56,  2.32s/it]
 29%|██▊       | 247/864 [11:37<23:54,  2.32s/it]
 29%|██▊       | 248/864 [11:40<23:52,  2.32s/it]
 29%|██▉       | 249/864 [11:42<23:49,  2.32s/it]
 29%|██▉       | 250/864 [11:44<23:47,  2.32s/it]
                                                 
{'loss': '5.147', 'grad_norm': '0.7293', 'learning_rate': '0.0012', 'epoch': '3.475'}
+
 29%|██▉       | 250/864 [11:44<23:47,  2.32s/it]
 29%|██▉       | 251/864 [11:47<23:45,  2.33s/it]
 29%|██▉       | 252/864 [11:49<23:43,  2.33s/it]
 29%|██▉       | 253/864 [11:51<23:40,  2.32s/it]
 29%|██▉       | 254/864 [11:53<23:38,  2.33s/it]
 30%|██▉       | 255/864 [11:56<23:35,  2.32s/it]
 30%|██▉       | 256/864 [11:58<23:33,  2.32s/it]
 30%|██▉       | 257/864 [12:00<23:31,  2.33s/it]
 30%|██▉       | 258/864 [12:03<23:29,  2.33s/it]
 30%|██▉       | 259/864 [12:05<23:26,  2.32s/it]
 30%|███       | 260/864 [12:07<23:24,  2.33s/it]
                                                 
{'loss': '5.12', 'grad_norm': '0.2701', 'learning_rate': '0.0012', 'epoch': '3.615'}
+
 30%|███       | 260/864 [12:07<23:24,  2.33s/it]
 30%|███       | 261/864 [12:10<23:21,  2.32s/it]
 30%|███       | 262/864 [12:12<23:20,  2.33s/it]
 30%|███       | 263/864 [12:14<23:17,  2.32s/it]
 31%|███       | 264/864 [12:17<23:14,  2.32s/it]
 31%|███       | 265/864 [12:19<23:12,  2.32s/it]
 31%|███       | 266/864 [12:21<23:10,  2.33s/it]
 31%|███       | 267/864 [12:24<23:07,  2.32s/it]
 31%|███       | 268/864 [12:26<23:05,  2.32s/it]
 31%|███       | 269/864 [12:28<23:02,  2.32s/it]
 31%|███▏      | 270/864 [12:31<23:00,  2.32s/it]
                                                 
{'loss': '5', 'grad_norm': '0.2535', 'learning_rate': '0.0012', 'epoch': '3.755'}
+
 31%|███▏      | 270/864 [12:31<23:00,  2.32s/it]
 31%|███▏      | 271/864 [12:33<22:58,  2.32s/it]
 31%|███▏      | 272/864 [12:35<22:56,  2.32s/it]
 32%|███▏      | 273/864 [12:38<22:53,  2.32s/it]
 32%|███▏      | 274/864 [12:40<22:51,  2.32s/it]
 32%|███▏      | 275/864 [12:42<22:49,  2.32s/it]
 32%|███▏      | 276/864 [12:45<22:46,  2.32s/it]
 32%|███▏      | 277/864 [12:47<22:44,  2.33s/it]
 32%|███▏      | 278/864 [12:49<22:42,  2.33s/it]
 32%|███▏      | 279/864 [12:52<22:40,  2.32s/it]
 32%|███▏      | 280/864 [12:54<22:37,  2.32s/it]
                                                 
{'loss': '4.967', 'grad_norm': '0.6225', 'learning_rate': '0.0012', 'epoch': '3.894'}
+
 32%|███▏      | 280/864 [12:54<22:37,  2.32s/it]
 33%|███▎      | 281/864 [12:56<22:35,  2.33s/it]
 33%|███▎      | 282/864 [12:59<22:33,  2.33s/it]
 33%|███▎      | 283/864 [13:01<22:31,  2.33s/it]
 33%|███▎      | 284/864 [13:03<22:28,  2.33s/it]
 33%|███▎      | 285/864 [13:06<22:26,  2.33s/it]
 33%|███▎      | 286/864 [13:08<22:24,  2.33s/it]
 33%|███▎      | 287/864 [13:10<22:22,  2.33s/it]
 33%|███▎      | 288/864 [13:12<19:27,  2.03s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 4
+
+
  0%|          | 0/77 [00:00<?, ?it/s]
+
  3%|▎         | 2/77 [00:00<00:03, 19.88it/s]
+
  5%|▌         | 4/77 [00:00<00:05, 12.34it/s]
+
  8%|▊         | 6/77 [00:00<00:06, 11.02it/s]
+
 10%|█         | 8/77 [00:00<00:06, 10.50it/s]
+
 13%|█▎        | 10/77 [00:00<00:06, 10.23it/s]
+
 16%|█▌        | 12/77 [00:01<00:06, 10.06it/s]
+
 18%|█▊        | 14/77 [00:01<00:06,  9.96it/s]
+
 21%|██        | 16/77 [00:01<00:06,  9.91it/s]
+
 23%|██▎       | 18/77 [00:01<00:05,  9.86it/s]
+
 25%|██▍       | 19/77 [00:01<00:05,  9.85it/s]
+
 26%|██▌       | 20/77 [00:01<00:05,  9.83it/s]
+
 27%|██▋       | 21/77 [00:02<00:05,  9.82it/s]
+
 29%|██▊       | 22/77 [00:02<00:05,  9.81it/s]
+
 30%|██▉       | 23/77 [00:02<00:05,  9.79it/s]
+
 31%|███       | 24/77 [00:02<00:05,  9.80it/s]
+
 32%|███▏      | 25/77 [00:02<00:05,  9.79it/s]
+
 34%|███▍      | 26/77 [00:02<00:05,  9.78it/s]
+
 35%|███▌      | 27/77 [00:02<00:05,  9.79it/s]
+
 36%|███▋      | 28/77 [00:02<00:05,  9.79it/s]
+
 38%|███▊      | 29/77 [00:02<00:04,  9.78it/s]
+
 39%|███▉      | 30/77 [00:02<00:04,  9.77it/s]
+
 40%|████      | 31/77 [00:03<00:04,  9.78it/s]
+
 42%|████▏     | 32/77 [00:03<00:04,  9.78it/s]
+
 43%|████▎     | 33/77 [00:03<00:04,  9.78it/s]
+
 44%|████▍     | 34/77 [00:03<00:04,  9.78it/s]
+
 45%|████▌     | 35/77 [00:03<00:04,  9.78it/s]
+
 47%|████▋     | 36/77 [00:03<00:04,  9.77it/s]
+
 48%|████▊     | 37/77 [00:03<00:04,  9.78it/s]
+
 49%|████▉     | 38/77 [00:03<00:03,  9.77it/s]
+
 51%|█████     | 39/77 [00:03<00:03,  9.78it/s]
+
 52%|█████▏    | 40/77 [00:03<00:03,  9.78it/s]
+
 53%|█████▎    | 41/77 [00:04<00:03,  9.78it/s]
+
 55%|█████▍    | 42/77 [00:04<00:03,  9.77it/s]
+
 56%|█████▌    | 43/77 [00:04<00:03,  9.78it/s]
+
 57%|█████▋    | 44/77 [00:04<00:03,  9.78it/s]
+
 58%|█████▊    | 45/77 [00:04<00:03,  9.77it/s]
+
 60%|█████▉    | 46/77 [00:04<00:03,  9.78it/s]
+
 61%|██████    | 47/77 [00:04<00:03,  9.78it/s]
+
 62%|██████▏   | 48/77 [00:04<00:02,  9.78it/s]
+
 64%|██████▎   | 49/77 [00:04<00:02,  9.78it/s]
+
 65%|██████▍   | 50/77 [00:05<00:02,  9.77it/s]
+
 66%|██████▌   | 51/77 [00:05<00:02,  9.77it/s]
+
 68%|██████▊   | 52/77 [00:05<00:02,  9.78it/s]
+
 69%|██████▉   | 53/77 [00:05<00:02,  9.77it/s]
+
 70%|███████   | 54/77 [00:05<00:02,  9.77it/s]
+
 71%|███████▏  | 55/77 [00:05<00:02,  9.78it/s]
+
 73%|███████▎  | 56/77 [00:05<00:02,  9.78it/s]
+
 74%|███████▍  | 57/77 [00:05<00:02,  9.78it/s]
+
 75%|███████▌  | 58/77 [00:05<00:01,  9.77it/s]
+
 77%|███████▋  | 59/77 [00:05<00:01,  9.78it/s]
+
 78%|███████▊  | 60/77 [00:06<00:01,  9.78it/s]
+
 79%|███████▉  | 61/77 [00:06<00:01,  9.76it/s]
+
 81%|████████  | 62/77 [00:06<00:01,  9.76it/s]
+
 82%|████████▏ | 63/77 [00:06<00:01,  9.79it/s]
+
 83%|████████▎ | 64/77 [00:06<00:01,  9.79it/s]
+
 84%|████████▍ | 65/77 [00:06<00:01,  9.79it/s]
+
 86%|████████▌ | 66/77 [00:06<00:01,  9.78it/s]
+
 87%|████████▋ | 67/77 [00:06<00:01,  9.79it/s]
+
 88%|████████▊ | 68/77 [00:06<00:00,  9.78it/s]
+
 90%|████████▉ | 69/77 [00:06<00:00,  9.78it/s]
+
 91%|█████████ | 70/77 [00:07<00:00,  9.78it/s]
+
 92%|█████████▏| 71/77 [00:07<00:00,  9.78it/s]
+
 94%|█████████▎| 72/77 [00:07<00:00,  9.78it/s]
+
 95%|█████████▍| 73/77 [00:07<00:00,  9.78it/s]
+
 96%|█████████▌| 74/77 [00:07<00:00,  9.78it/s]
+
 97%|█████████▋| 75/77 [00:07<00:00,  9.78it/s]
+
 99%|█████████▊| 76/77 [00:07<00:00,  9.78it/s]
+
100%|██████████| 77/77 [00:07<00:00,  9.78it/s]
                                                 
+
                                               
{'eval_loss': '5', 'eval_runtime': '7.916', 'eval_samples_per_second': '308.4', 'eval_steps_per_second': '9.727', 'epoch': '4'}
+
 33%|███▎      | 288/864 [13:19<19:27,  2.03s/it]
+
100%|██████████| 77/77 [00:07<00:00,  9.78it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0012/checkpoint-288
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0012/checkpoint-288/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0012/checkpoint-288/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:03<00:00,  3.60s/it]
Writing model shards: 100%|██████████| 1/1 [00:03<00:00,  3.60s/it]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0012/checkpoint-288/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0012/checkpoint-288/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0012/checkpoint-288/tokenizer_config.json
+[eval] epoch 4.0 step 288 loss 5.0002 (best 5.0002 @ ep 4.0, 0/10000 since best)
+
+
                                               
 33%|███▎      | 289/864 [13:33<1:16:14,  7.96s/it]
 34%|███▎      | 290/864 [13:36<59:56,  6.26s/it]  
                                                 
{'loss': '4.924', 'grad_norm': '0.3329', 'learning_rate': '0.0012', 'epoch': '4.028'}
+
 34%|███▎      | 290/864 [13:36<59:56,  6.26s/it]
 34%|███▎      | 291/864 [13:38<48:32,  5.08s/it]
 34%|███▍      | 292/864 [13:40<40:33,  4.25s/it]
 34%|███▍      | 293/864 [13:43<34:58,  3.67s/it]
 34%|███▍      | 294/864 [13:45<31:04,  3.27s/it]
 34%|███▍      | 295/864 [13:47<28:18,  2.99s/it]
 34%|███▍      | 296/864 [13:50<26:22,  2.79s/it]
 34%|███▍      | 297/864 [13:52<25:00,  2.65s/it]
 34%|███▍      | 298/864 [13:54<24:02,  2.55s/it]
 35%|███▍      | 299/864 [13:57<23:22,  2.48s/it]
 35%|███▍      | 300/864 [13:59<22:51,  2.43s/it]
                                                 
{'loss': '4.756', 'grad_norm': '0.2035', 'learning_rate': '0.0012', 'epoch': '4.168'}
+
 35%|███▍      | 300/864 [13:59<22:51,  2.43s/it]
 35%|███▍      | 301/864 [14:01<22:30,  2.40s/it]
 35%|███▍      | 302/864 [14:04<22:14,  2.38s/it]
 35%|███▌      | 303/864 [14:06<22:03,  2.36s/it]
 35%|███▌      | 304/864 [14:08<21:54,  2.35s/it]
 35%|███▌      | 305/864 [14:10<21:48,  2.34s/it]
 35%|███▌      | 306/864 [14:13<21:42,  2.33s/it]
 36%|███▌      | 307/864 [14:15<21:38,  2.33s/it]
 36%|███▌      | 308/864 [14:17<21:35,  2.33s/it]
 36%|███▌      | 309/864 [14:20<21:31,  2.33s/it]
 36%|███▌      | 310/864 [14:22<21:28,  2.33s/it]
                                                 
{'loss': '4.682', 'grad_norm': '0.4597', 'learning_rate': '0.0012', 'epoch': '4.307'}
+
 36%|███▌      | 310/864 [14:22<21:28,  2.33s/it]
 36%|███▌      | 311/864 [14:24<21:25,  2.32s/it]
 36%|███▌      | 312/864 [14:27<21:22,  2.32s/it]
 36%|███▌      | 313/864 [14:29<21:19,  2.32s/it]
 36%|███▋      | 314/864 [14:31<21:17,  2.32s/it]
 36%|███▋      | 315/864 [14:34<21:15,  2.32s/it]
 37%|███▋      | 316/864 [14:36<21:13,  2.32s/it]
 37%|███▋      | 317/864 [14:38<21:11,  2.32s/it]
 37%|███▋      | 318/864 [14:41<21:08,  2.32s/it]
 37%|███▋      | 319/864 [14:43<21:05,  2.32s/it]
 37%|███▋      | 320/864 [14:45<21:03,  2.32s/it]
                                                 
{'loss': '4.705', 'grad_norm': '0.3954', 'learning_rate': '0.0012', 'epoch': '4.447'}
+
 37%|███▋      | 320/864 [14:45<21:03,  2.32s/it]
 37%|███▋      | 321/864 [14:48<21:01,  2.32s/it]
 37%|███▋      | 322/864 [14:50<20:59,  2.32s/it]
 37%|███▋      | 323/864 [14:52<20:57,  2.32s/it]
 38%|███▊      | 324/864 [14:55<20:54,  2.32s/it]
 38%|███▊      | 325/864 [14:57<20:51,  2.32s/it]
 38%|███▊      | 326/864 [14:59<20:49,  2.32s/it]
 38%|███▊      | 327/864 [15:02<20:47,  2.32s/it]
 38%|███▊      | 328/864 [15:04<20:44,  2.32s/it]
 38%|███▊      | 329/864 [15:06<20:42,  2.32s/it]
 38%|███▊      | 330/864 [15:09<20:40,  2.32s/it]
                                                 
{'loss': '4.624', 'grad_norm': '0.1819', 'learning_rate': '0.0012', 'epoch': '4.587'}
+
 38%|███▊      | 330/864 [15:09<20:40,  2.32s/it]
 38%|███▊      | 331/864 [15:11<20:37,  2.32s/it]
 38%|███▊      | 332/864 [15:13<20:35,  2.32s/it]
 39%|███▊      | 333/864 [15:16<20:33,  2.32s/it]
 39%|███▊      | 334/864 [15:18<20:30,  2.32s/it]
 39%|███▉      | 335/864 [15:20<20:28,  2.32s/it]
 39%|███▉      | 336/864 [15:22<20:25,  2.32s/it]
 39%|███▉      | 337/864 [15:25<20:23,  2.32s/it]
 39%|███▉      | 338/864 [15:27<20:21,  2.32s/it]
 39%|███▉      | 339/864 [15:29<20:19,  2.32s/it]
 39%|███▉      | 340/864 [15:32<20:16,  2.32s/it]
                                                 
{'loss': '4.538', 'grad_norm': '0.3704', 'learning_rate': '0.0012', 'epoch': '4.727'}
+
 39%|███▉      | 340/864 [15:32<20:16,  2.32s/it]
 39%|███▉      | 341/864 [15:34<20:14,  2.32s/it]
 40%|███▉      | 342/864 [15:36<20:11,  2.32s/it]
 40%|███▉      | 343/864 [15:39<20:09,  2.32s/it]
 40%|███▉      | 344/864 [15:41<20:07,  2.32s/it]
 40%|███▉      | 345/864 [15:43<20:04,  2.32s/it]
 40%|████      | 346/864 [15:46<20:02,  2.32s/it]
 40%|████      | 347/864 [15:48<20:00,  2.32s/it]
 40%|████      | 348/864 [15:50<19:57,  2.32s/it]
 40%|████      | 349/864 [15:53<19:55,  2.32s/it]
 41%|████      | 350/864 [15:55<19:53,  2.32s/it]
                                                 
{'loss': '4.576', 'grad_norm': '0.4234', 'learning_rate': '0.0012', 'epoch': '4.866'}
+
 41%|████      | 350/864 [15:55<19:53,  2.32s/it]
 41%|████      | 351/864 [15:57<19:51,  2.32s/it]
 41%|████      | 352/864 [16:00<19:49,  2.32s/it]
 41%|████      | 353/864 [16:02<19:46,  2.32s/it]
 41%|████      | 354/864 [16:05<20:37,  2.43s/it]
 41%|████      | 355/864 [16:07<20:19,  2.40s/it]
 41%|████      | 356/864 [16:09<20:06,  2.37s/it]
 41%|████▏     | 357/864 [16:12<19:57,  2.36s/it]
 41%|████▏     | 358/864 [16:14<19:49,  2.35s/it]
 42%|████▏     | 359/864 [16:16<19:43,  2.34s/it]
 42%|████▏     | 360/864 [16:18<17:07,  2.04s/it]
                                                 
{'loss': '4.482', 'grad_norm': '0.1796', 'learning_rate': '0.0012', 'epoch': '5'}
+
 42%|████▏     | 360/864 [16:18<17:07,  2.04s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 4
+
+
  0%|          | 0/77 [00:00<?, ?it/s]
+
  3%|▎         | 2/77 [00:00<00:03, 19.88it/s]
+
  5%|▌         | 4/77 [00:00<00:05, 12.31it/s]
+
  8%|▊         | 6/77 [00:00<00:06, 11.01it/s]
+
 10%|█         | 8/77 [00:00<00:06, 10.49it/s]
+
 13%|█▎        | 10/77 [00:00<00:06, 10.22it/s]
+
 16%|█▌        | 12/77 [00:01<00:06, 10.07it/s]
+
 18%|█▊        | 14/77 [00:01<00:06,  9.97it/s]
+
 21%|██        | 16/77 [00:01<00:06,  9.91it/s]
+
 23%|██▎       | 18/77 [00:01<00:05,  9.87it/s]
+
 25%|██▍       | 19/77 [00:01<00:05,  9.85it/s]
+
 26%|██▌       | 20/77 [00:01<00:05,  9.83it/s]
+
 27%|██▋       | 21/77 [00:02<00:05,  9.82it/s]
+
 29%|██▊       | 22/77 [00:02<00:05,  9.81it/s]
+
 30%|██▉       | 23/77 [00:02<00:05,  9.80it/s]
+
 31%|███       | 24/77 [00:02<00:05,  9.81it/s]
+
 32%|███▏      | 25/77 [00:02<00:05,  9.79it/s]
+
 34%|███▍      | 26/77 [00:02<00:05,  9.79it/s]
+
 35%|███▌      | 27/77 [00:02<00:05,  9.79it/s]
+
 36%|███▋      | 28/77 [00:02<00:05,  9.78it/s]
+
 38%|███▊      | 29/77 [00:02<00:04,  9.78it/s]
+
 39%|███▉      | 30/77 [00:02<00:04,  9.78it/s]
+
 40%|████      | 31/77 [00:03<00:04,  9.78it/s]
+
 42%|████▏     | 32/77 [00:03<00:04,  9.78it/s]
+
 43%|████▎     | 33/77 [00:03<00:04,  9.77it/s]
+
 44%|████▍     | 34/77 [00:03<00:04,  9.78it/s]
+
 45%|████▌     | 35/77 [00:03<00:04,  9.77it/s]
+
 47%|████▋     | 36/77 [00:03<00:04,  9.78it/s]
+
 48%|████▊     | 37/77 [00:03<00:04,  9.78it/s]
+
 49%|████▉     | 38/77 [00:03<00:03,  9.78it/s]
+
 51%|█████     | 39/77 [00:03<00:03,  9.78it/s]
+
 52%|█████▏    | 40/77 [00:03<00:03,  9.78it/s]
+
 53%|█████▎    | 41/77 [00:04<00:03,  9.77it/s]
+
 55%|█████▍    | 42/77 [00:04<00:03,  9.77it/s]
+
 56%|█████▌    | 43/77 [00:04<00:03,  9.77it/s]
+
 57%|█████▋    | 44/77 [00:04<00:03,  9.77it/s]
+
 58%|█████▊    | 45/77 [00:04<00:03,  9.77it/s]
+
 60%|█████▉    | 46/77 [00:04<00:03,  9.77it/s]
+
 61%|██████    | 47/77 [00:04<00:03,  9.77it/s]
+
 62%|██████▏   | 48/77 [00:04<00:02,  9.77it/s]
+
 64%|██████▎   | 49/77 [00:04<00:02,  9.77it/s]
+
 65%|██████▍   | 50/77 [00:05<00:02,  9.77it/s]
+
 66%|██████▌   | 51/77 [00:05<00:02,  9.77it/s]
+
 68%|██████▊   | 52/77 [00:05<00:02,  9.78it/s]
+
 69%|██████▉   | 53/77 [00:05<00:02,  9.77it/s]
+
 70%|███████   | 54/77 [00:05<00:02,  9.78it/s]
+
 71%|███████▏  | 55/77 [00:05<00:02,  9.78it/s]
+
 73%|███████▎  | 56/77 [00:05<00:02,  9.73it/s]
+
 74%|███████▍  | 57/77 [00:05<00:02,  9.77it/s]
+
 75%|███████▌  | 58/77 [00:05<00:01,  9.79it/s]
+
 77%|███████▋  | 59/77 [00:05<00:01,  9.79it/s]
+
 78%|███████▊  | 60/77 [00:06<00:01,  9.79it/s]
+
 79%|███████▉  | 61/77 [00:06<00:01,  9.80it/s]
+
 81%|████████  | 62/77 [00:06<00:01,  9.80it/s]
+
 82%|████████▏ | 63/77 [00:06<00:01,  9.79it/s]
+
 83%|████████▎ | 64/77 [00:06<00:01,  9.79it/s]
+
 84%|████████▍ | 65/77 [00:06<00:01,  9.79it/s]
+
 86%|████████▌ | 66/77 [00:06<00:01,  9.79it/s]
+
 87%|████████▋ | 67/77 [00:06<00:01,  9.79it/s]
+
 88%|████████▊ | 68/77 [00:06<00:00,  9.79it/s]
+
 90%|████████▉ | 69/77 [00:06<00:00,  9.79it/s]
+
 91%|█████████ | 70/77 [00:07<00:00,  9.79it/s]
+
 92%|█████████▏| 71/77 [00:07<00:00,  9.79it/s]
+
 94%|█████████▎| 72/77 [00:07<00:00,  9.79it/s]
+
 95%|█████████▍| 73/77 [00:07<00:00,  9.79it/s]
+
 96%|█████████▌| 74/77 [00:07<00:00,  9.79it/s]
+
 97%|█████████▋| 75/77 [00:07<00:00,  9.79it/s]
+
 99%|█████████▊| 76/77 [00:07<00:00,  9.79it/s]
+
100%|██████████| 77/77 [00:07<00:00,  9.79it/s]
                                                 
+
                                               
{'eval_loss': '4.559', 'eval_runtime': '7.911', 'eval_samples_per_second': '308.6', 'eval_steps_per_second': '9.734', 'epoch': '5'}
+
 42%|████▏     | 360/864 [16:26<17:07,  2.04s/it]
+
100%|██████████| 77/77 [00:07<00:00,  9.79it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0012/checkpoint-360
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0012/checkpoint-360/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0012/checkpoint-360/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:03<00:00,  3.62s/it]
Writing model shards: 100%|██████████| 1/1 [00:03<00:00,  3.62s/it]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0012/checkpoint-360/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0012/checkpoint-360/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0012/checkpoint-360/tokenizer_config.json
+[eval] epoch 5.0 step 360 loss 4.5587 (best 4.5587 @ ep 5.0, 0/10000 since best)
+
+
                                               
 42%|████▏     | 361/864 [16:39<1:06:54,  7.98s/it]
 42%|████▏     | 362/864 [16:42<52:33,  6.28s/it]  
 42%|████▏     | 363/864 [16:44<42:32,  5.09s/it]
 42%|████▏     | 364/864 [16:46<35:31,  4.26s/it]
 42%|████▏     | 365/864 [16:49<30:36,  3.68s/it]
 42%|████▏     | 366/864 [16:51<27:09,  3.27s/it]
 42%|████▏     | 367/864 [16:53<24:45,  2.99s/it]
 43%|████▎     | 368/864 [16:56<23:03,  2.79s/it]
 43%|████▎     | 369/864 [16:58<21:51,  2.65s/it]
 43%|████▎     | 370/864 [17:00<21:00,  2.55s/it]
                                                 
{'loss': '4.316', 'grad_norm': '0.3379', 'learning_rate': '0.0012', 'epoch': '5.14'}
+
 43%|████▎     | 370/864 [17:00<21:00,  2.55s/it]
 43%|████▎     | 371/864 [17:03<21:21,  2.60s/it]
 43%|████▎     | 372/864 [17:06<21:30,  2.62s/it]
 43%|████▎     | 373/864 [17:08<20:42,  2.53s/it]
 43%|████▎     | 374/864 [17:10<20:09,  2.47s/it]
 43%|████▎     | 375/864 [17:13<19:45,  2.42s/it]
 44%|████▎     | 376/864 [17:15<19:28,  2.39s/it]
 44%|████▎     | 377/864 [17:17<19:15,  2.37s/it]
 44%|████▍     | 378/864 [17:20<19:05,  2.36s/it]
 44%|████▍     | 379/864 [17:22<18:57,  2.35s/it]
 44%|████▍     | 380/864 [17:24<18:51,  2.34s/it]
                                                 
{'loss': '4.301', 'grad_norm': '0.2186', 'learning_rate': '0.0012', 'epoch': '5.279'}
+
 44%|████▍     | 380/864 [17:24<18:51,  2.34s/it]
 44%|████▍     | 381/864 [17:27<18:47,  2.33s/it]
 44%|████▍     | 382/864 [17:29<18:43,  2.33s/it]
 44%|████▍     | 383/864 [17:31<18:39,  2.33s/it]
 44%|████▍     | 384/864 [17:34<18:36,  2.33s/it]
 45%|████▍     | 385/864 [17:36<18:33,  2.32s/it]
 45%|████▍     | 386/864 [17:38<18:30,  2.32s/it]
 45%|████▍     | 387/864 [17:41<18:27,  2.32s/it]
 45%|████▍     | 388/864 [17:43<18:24,  2.32s/it]
 45%|████▌     | 389/864 [17:45<18:22,  2.32s/it]
 45%|████▌     | 390/864 [17:48<18:20,  2.32s/it]
                                                 
{'loss': '4.228', 'grad_norm': '0.1517', 'learning_rate': '0.0012', 'epoch': '5.419'}
+
 45%|████▌     | 390/864 [17:48<18:20,  2.32s/it]
 45%|████▌     | 391/864 [17:50<18:17,  2.32s/it]
 45%|████▌     | 392/864 [17:52<18:15,  2.32s/it]
 45%|████▌     | 393/864 [17:54<18:12,  2.32s/it]
 46%|████▌     | 394/864 [17:57<18:10,  2.32s/it]
 46%|████▌     | 395/864 [17:59<18:08,  2.32s/it]
 46%|████▌     | 396/864 [18:01<18:05,  2.32s/it]
 46%|████▌     | 397/864 [18:04<18:04,  2.32s/it]
 46%|████▌     | 398/864 [18:06<18:02,  2.32s/it]
 46%|████▌     | 399/864 [18:08<17:59,  2.32s/it]
 46%|████▋     | 400/864 [18:11<17:57,  2.32s/it]
                                                 
{'loss': '4.208', 'grad_norm': '0.2983', 'learning_rate': '0.0012', 'epoch': '5.559'}
+
 46%|████▋     | 400/864 [18:11<17:57,  2.32s/it]
 46%|████▋     | 401/864 [18:13<17:55,  2.32s/it]
 47%|████▋     | 402/864 [18:15<17:53,  2.32s/it]
 47%|████▋     | 403/864 [18:18<17:50,  2.32s/it]
 47%|████▋     | 404/864 [18:20<17:47,  2.32s/it]
 47%|████▋     | 405/864 [18:22<17:45,  2.32s/it]
 47%|████▋     | 406/864 [18:25<17:42,  2.32s/it]
 47%|████▋     | 407/864 [18:27<17:40,  2.32s/it]
 47%|████▋     | 408/864 [18:29<17:38,  2.32s/it]
 47%|████▋     | 409/864 [18:32<17:36,  2.32s/it]
 47%|████▋     | 410/864 [18:34<17:33,  2.32s/it]
                                                 
{'loss': '4.186', 'grad_norm': '0.242', 'learning_rate': '0.0012', 'epoch': '5.699'}
+
 47%|████▋     | 410/864 [18:34<17:33,  2.32s/it]
 48%|████▊     | 411/864 [18:36<17:31,  2.32s/it]
 48%|████▊     | 412/864 [18:39<17:28,  2.32s/it]
 48%|████▊     | 413/864 [18:41<17:27,  2.32s/it]
 48%|████▊     | 414/864 [18:43<17:25,  2.32s/it]
 48%|████▊     | 415/864 [18:46<17:22,  2.32s/it]
 48%|████▊     | 416/864 [18:48<17:19,  2.32s/it]
 48%|████▊     | 417/864 [18:50<17:17,  2.32s/it]
 48%|████▊     | 418/864 [18:53<17:15,  2.32s/it]
 48%|████▊     | 419/864 [18:55<17:13,  2.32s/it]
 49%|████▊     | 420/864 [18:57<17:10,  2.32s/it]
                                                 
{'loss': '4.153', 'grad_norm': '0.1702', 'learning_rate': '0.0012', 'epoch': '5.838'}
+
 49%|████▊     | 420/864 [18:57<17:10,  2.32s/it]
 49%|████▊     | 421/864 [18:59<17:08,  2.32s/it]
 49%|████▉     | 422/864 [19:02<17:06,  2.32s/it]
 49%|████▉     | 423/864 [19:04<17:03,  2.32s/it]
 49%|████▉     | 424/864 [19:06<17:01,  2.32s/it]
 49%|████▉     | 425/864 [19:09<16:59,  2.32s/it]
 49%|████▉     | 426/864 [19:11<16:56,  2.32s/it]
 49%|████▉     | 427/864 [19:13<16:54,  2.32s/it]
 50%|████▉     | 428/864 [19:16<16:52,  2.32s/it]
 50%|████▉     | 429/864 [19:18<16:50,  2.32s/it]
 50%|████▉     | 430/864 [19:20<16:48,  2.32s/it]
                                                 
{'loss': '4.125', 'grad_norm': '0.2346', 'learning_rate': '0.0012', 'epoch': '5.978'}
+
 50%|████▉     | 430/864 [19:20<16:48,  2.32s/it]
 50%|████▉     | 431/864 [19:23<16:46,  2.33s/it]
 50%|█████     | 432/864 [19:24<14:34,  2.02s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 4
+
+
  0%|          | 0/77 [00:00<?, ?it/s]
+
  4%|▍         | 3/77 [00:00<00:04, 14.87it/s]
+
  6%|▋         | 5/77 [00:00<00:06, 11.86it/s]
+
  9%|▉         | 7/77 [00:00<00:06, 10.91it/s]
+
 12%|█▏        | 9/77 [00:00<00:06, 10.47it/s]
+
 14%|█▍        | 11/77 [00:01<00:06, 10.21it/s]
+
 17%|█▋        | 13/77 [00:01<00:06, 10.07it/s]
+
 19%|█▉        | 15/77 [00:01<00:06,  9.97it/s]
+
 21%|██        | 16/77 [00:01<00:06,  9.94it/s]
+
 22%|██▏       | 17/77 [00:01<00:06,  9.90it/s]
+
 23%|██▎       | 18/77 [00:01<00:05,  9.88it/s]
+
 25%|██▍       | 19/77 [00:01<00:05,  9.85it/s]
+
 26%|██▌       | 20/77 [00:01<00:05,  9.84it/s]
+
 27%|██▋       | 21/77 [00:02<00:05,  9.82it/s]
+
 29%|██▊       | 22/77 [00:02<00:05,  9.81it/s]
+
 30%|██▉       | 23/77 [00:02<00:05,  9.79it/s]
+
 31%|███       | 24/77 [00:02<00:05,  9.80it/s]
+
 32%|███▏      | 25/77 [00:02<00:05,  9.80it/s]
+
 34%|███▍      | 26/77 [00:02<00:05,  9.79it/s]
+
 35%|███▌      | 27/77 [00:02<00:05,  9.78it/s]
+
 36%|███▋      | 28/77 [00:02<00:05,  9.78it/s]
+
 38%|███▊      | 29/77 [00:02<00:04,  9.78it/s]
+
 39%|███▉      | 30/77 [00:02<00:04,  9.79it/s]
+
 40%|████      | 31/77 [00:03<00:04,  9.78it/s]
+
 42%|████▏     | 32/77 [00:03<00:04,  9.78it/s]
+
 43%|████▎     | 33/77 [00:03<00:04,  9.78it/s]
+
 44%|████▍     | 34/77 [00:03<00:04,  9.78it/s]
+
 45%|████▌     | 35/77 [00:03<00:04,  9.78it/s]
+
 47%|████▋     | 36/77 [00:03<00:04,  9.78it/s]
+
 48%|████▊     | 37/77 [00:03<00:04,  9.77it/s]
+
 49%|████▉     | 38/77 [00:03<00:03,  9.78it/s]
+
 51%|█████     | 39/77 [00:03<00:03,  9.78it/s]
+
 52%|█████▏    | 40/77 [00:03<00:03,  9.78it/s]
+
 53%|█████▎    | 41/77 [00:04<00:03,  9.78it/s]
+
 55%|█████▍    | 42/77 [00:04<00:03,  9.78it/s]
+
 56%|█████▌    | 43/77 [00:04<00:03,  9.76it/s]
+
 57%|█████▋    | 44/77 [00:04<00:03,  9.77it/s]
+
 58%|█████▊    | 45/77 [00:04<00:03,  9.78it/s]
+
 60%|█████▉    | 46/77 [00:04<00:03,  9.78it/s]
+
 61%|██████    | 47/77 [00:04<00:03,  9.77it/s]
+
 62%|██████▏   | 48/77 [00:04<00:02,  9.78it/s]
+
 64%|██████▎   | 49/77 [00:04<00:02,  9.78it/s]
+
 65%|██████▍   | 50/77 [00:05<00:02,  9.78it/s]
+
 66%|██████▌   | 51/77 [00:05<00:02,  9.77it/s]
+
 68%|██████▊   | 52/77 [00:05<00:02,  9.78it/s]
+
 69%|██████▉   | 53/77 [00:05<00:02,  9.78it/s]
+
 70%|███████   | 54/77 [00:05<00:02,  9.77it/s]
+
 71%|███████▏  | 55/77 [00:05<00:02,  9.77it/s]
+
 73%|███████▎  | 56/77 [00:05<00:02,  9.78it/s]
+
 74%|███████▍  | 57/77 [00:05<00:02,  9.78it/s]
+
 75%|███████▌  | 58/77 [00:05<00:01,  9.79it/s]
+
 77%|███████▋  | 59/77 [00:05<00:01,  9.77it/s]
+
 78%|███████▊  | 60/77 [00:06<00:01,  9.78it/s]
+
 79%|███████▉  | 61/77 [00:06<00:01,  9.80it/s]
+
 81%|████████  | 62/77 [00:06<00:01,  9.79it/s]
+
 82%|████████▏ | 63/77 [00:06<00:01,  9.78it/s]
+
 83%|████████▎ | 64/77 [00:06<00:01,  9.79it/s]
+
 84%|████████▍ | 65/77 [00:06<00:01,  9.78it/s]
+
 86%|████████▌ | 66/77 [00:06<00:01,  9.77it/s]
+
 87%|████████▋ | 67/77 [00:06<00:01,  9.77it/s]
+
 88%|████████▊ | 68/77 [00:06<00:00,  9.78it/s]
+
 90%|████████▉ | 69/77 [00:06<00:00,  9.79it/s]
+
 91%|█████████ | 70/77 [00:07<00:00,  9.78it/s]
+
 92%|█████████▏| 71/77 [00:07<00:00,  9.78it/s]
+
 94%|█████████▎| 72/77 [00:07<00:00,  9.79it/s]
+
 95%|█████████▍| 73/77 [00:07<00:00,  9.79it/s]
+
 96%|█████████▌| 74/77 [00:07<00:00,  9.79it/s]
+
 97%|█████████▋| 75/77 [00:07<00:00,  9.78it/s]
+
 99%|█████████▊| 76/77 [00:07<00:00,  9.79it/s]
+
100%|██████████| 77/77 [00:07<00:00,  9.79it/s]
                                                 
+
                                               
{'eval_loss': '4.292', 'eval_runtime': '7.914', 'eval_samples_per_second': '308.4', 'eval_steps_per_second': '9.729', 'epoch': '6'}
+
 50%|█████     | 432/864 [19:32<14:34,  2.02s/it]
+
100%|██████████| 77/77 [00:07<00:00,  9.79it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0012/checkpoint-432
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0012/checkpoint-432/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0012/checkpoint-432/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:03<00:00,  3.60s/it]
Writing model shards: 100%|██████████| 1/1 [00:03<00:00,  3.60s/it]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0012/checkpoint-432/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0012/checkpoint-432/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0012/checkpoint-432/tokenizer_config.json
+[eval] epoch 6.0 step 432 loss 4.2920 (best 4.2920 @ ep 6.0, 0/10000 since best)
+
+
                                               
 50%|█████     | 433/864 [19:46<57:13,  7.97s/it]
 50%|█████     | 434/864 [19:48<44:56,  6.27s/it]
 50%|█████     | 435/864 [19:50<36:22,  5.09s/it]
 50%|█████     | 436/864 [19:53<30:22,  4.26s/it]
 51%|█████     | 437/864 [19:55<26:09,  3.68s/it]
 51%|█████     | 438/864 [19:57<23:13,  3.27s/it]
 51%|█████     | 439/864 [20:00<21:08,  2.99s/it]
 51%|█████     | 440/864 [20:02<19:41,  2.79s/it]
                                                 
{'loss': '4.009', 'grad_norm': '0.3361', 'learning_rate': '0.0012', 'epoch': '6.112'}
+
 51%|█████     | 440/864 [20:02<19:41,  2.79s/it]
 51%|█████     | 441/864 [20:04<18:39,  2.65s/it]
 51%|█████     | 442/864 [20:07<17:55,  2.55s/it]
 51%|█████▏    | 443/864 [20:09<17:24,  2.48s/it]
 51%|█████▏    | 444/864 [20:11<17:01,  2.43s/it]
 52%|█████▏    | 445/864 [20:14<16:45,  2.40s/it]
 52%|█████▏    | 446/864 [20:16<16:33,  2.38s/it]
 52%|█████▏    | 447/864 [20:18<16:23,  2.36s/it]
 52%|█████▏    | 448/864 [20:21<16:16,  2.35s/it]
 52%|█████▏    | 449/864 [20:23<16:10,  2.34s/it]
 52%|█████▏    | 450/864 [20:25<16:05,  2.33s/it]
                                                 
{'loss': '4.013', 'grad_norm': '0.2953', 'learning_rate': '0.0012', 'epoch': '6.252'}
+
 52%|█████▏    | 450/864 [20:25<16:05,  2.33s/it]
 52%|█████▏    | 451/864 [20:28<16:01,  2.33s/it]
 52%|█████▏    | 452/864 [20:30<15:58,  2.33s/it]
 52%|█████▏    | 453/864 [20:32<15:55,  2.32s/it]
 53%|█████▎    | 454/864 [20:35<15:52,  2.32s/it]
 53%|█████▎    | 455/864 [20:37<15:49,  2.32s/it]
 53%|█████▎    | 456/864 [20:39<15:47,  2.32s/it]
 53%|█████▎    | 457/864 [20:42<15:44,  2.32s/it]
 53%|█████▎    | 458/864 [20:44<15:42,  2.32s/it]
 53%|█████▎    | 459/864 [20:46<15:39,  2.32s/it]
 53%|█████▎    | 460/864 [20:49<15:37,  2.32s/it]
                                                 
{'loss': '3.961', 'grad_norm': '0.1475', 'learning_rate': '0.0012', 'epoch': '6.391'}
+
 53%|█████▎    | 460/864 [20:49<15:37,  2.32s/it]
 53%|█████▎    | 461/864 [20:51<15:35,  2.32s/it]
 53%|█████▎    | 462/864 [20:53<15:32,  2.32s/it]
 54%|█████▎    | 463/864 [20:55<15:30,  2.32s/it]
 54%|█████▎    | 464/864 [20:58<15:28,  2.32s/it]
 54%|█████▍    | 465/864 [21:00<15:25,  2.32s/it]
 54%|█████▍    | 466/864 [21:02<15:23,  2.32s/it]
 54%|█████▍    | 467/864 [21:05<15:20,  2.32s/it]
 54%|█████▍    | 468/864 [21:07<15:18,  2.32s/it]
 54%|█████▍    | 469/864 [21:09<15:16,  2.32s/it]
 54%|█████▍    | 470/864 [21:12<15:13,  2.32s/it]
                                                 
{'loss': '3.922', 'grad_norm': '0.1784', 'learning_rate': '0.0012', 'epoch': '6.531'}
+
 54%|█████▍    | 470/864 [21:12<15:13,  2.32s/it]
 55%|█████▍    | 471/864 [21:14<15:11,  2.32s/it]
 55%|█████▍    | 472/864 [21:16<15:09,  2.32s/it]
 55%|█████▍    | 473/864 [21:19<15:07,  2.32s/it]
 55%|█████▍    | 474/864 [21:21<15:05,  2.32s/it]
 55%|█████▍    | 475/864 [21:23<15:02,  2.32s/it]
 55%|█████▌    | 476/864 [21:26<15:00,  2.32s/it]
 55%|█████▌    | 477/864 [21:28<14:58,  2.32s/it]
 55%|█████▌    | 478/864 [21:30<14:55,  2.32s/it]
 55%|█████▌    | 479/864 [21:33<14:53,  2.32s/it]
 56%|█████▌    | 480/864 [21:35<14:51,  2.32s/it]
                                                 
{'loss': '3.918', 'grad_norm': '0.2231', 'learning_rate': '0.0012', 'epoch': '6.671'}
+
 56%|█████▌    | 480/864 [21:35<14:51,  2.32s/it]
 56%|█████▌    | 481/864 [21:37<14:49,  2.32s/it]
 56%|█████▌    | 482/864 [21:40<14:47,  2.32s/it]
 56%|█████▌    | 483/864 [21:42<14:44,  2.32s/it]
 56%|█████▌    | 484/864 [21:44<14:42,  2.32s/it]
 56%|█████▌    | 485/864 [21:47<14:39,  2.32s/it]
 56%|█████▋    | 486/864 [21:49<14:37,  2.32s/it]
 56%|█████▋    | 487/864 [21:51<14:34,  2.32s/it]
 56%|█████▋    | 488/864 [21:53<14:32,  2.32s/it]
 57%|█████▋    | 489/864 [21:56<14:30,  2.32s/it]
 57%|█████▋    | 490/864 [21:58<14:27,  2.32s/it]
                                                 
{'loss': '3.921', 'grad_norm': '0.1749', 'learning_rate': '0.0012', 'epoch': '6.81'}
+
 57%|█████▋    | 490/864 [21:58<14:27,  2.32s/it]
 57%|█████▋    | 491/864 [22:00<14:25,  2.32s/it]
 57%|█████▋    | 492/864 [22:03<14:23,  2.32s/it]
 57%|█████▋    | 493/864 [22:05<14:21,  2.32s/it]
 57%|█████▋    | 494/864 [22:07<14:19,  2.32s/it]
 57%|█████▋    | 495/864 [22:10<14:16,  2.32s/it]
 57%|█████▋    | 496/864 [22:12<14:14,  2.32s/it]
 58%|█████▊    | 497/864 [22:14<14:12,  2.32s/it]
 58%|█████▊    | 498/864 [22:17<14:09,  2.32s/it]
 58%|█████▊    | 499/864 [22:19<14:07,  2.32s/it]
 58%|█████▊    | 500/864 [22:21<14:04,  2.32s/it]
                                                 
{'loss': '3.878', 'grad_norm': '0.1696', 'learning_rate': '0.0012', 'epoch': '6.95'}
+
 58%|█████▊    | 500/864 [22:21<14:04,  2.32s/it]
 58%|█████▊    | 501/864 [22:24<14:02,  2.32s/it]
 58%|█████▊    | 502/864 [22:26<14:00,  2.32s/it]
 58%|█████▊    | 503/864 [22:28<13:59,  2.32s/it]
 58%|█████▊    | 504/864 [22:30<12:08,  2.02s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 4
+
+
  0%|          | 0/77 [00:00<?, ?it/s]
+
  3%|▎         | 2/77 [00:00<00:03, 19.51it/s]
+
  5%|▌         | 4/77 [00:00<00:05, 12.26it/s]
+
  8%|▊         | 6/77 [00:00<00:06, 11.00it/s]
+
 10%|█         | 8/77 [00:00<00:06, 10.47it/s]
+
 13%|█▎        | 10/77 [00:00<00:06, 10.22it/s]
+
 16%|█▌        | 12/77 [00:01<00:06, 10.06it/s]
+
 18%|█▊        | 14/77 [00:01<00:06,  9.97it/s]
+
 21%|██        | 16/77 [00:01<00:06,  9.90it/s]
+
 23%|██▎       | 18/77 [00:01<00:05,  9.87it/s]
+
 25%|██▍       | 19/77 [00:01<00:05,  9.85it/s]
+
 26%|██▌       | 20/77 [00:01<00:05,  9.83it/s]
+
 27%|██▋       | 21/77 [00:02<00:05,  9.82it/s]
+
 29%|██▊       | 22/77 [00:02<00:05,  9.81it/s]
+
 30%|██▉       | 23/77 [00:02<00:05,  9.80it/s]
+
 31%|███       | 24/77 [00:02<00:05,  9.79it/s]
+
 32%|███▏      | 25/77 [00:02<00:05,  9.79it/s]
+
 34%|███▍      | 26/77 [00:02<00:05,  9.79it/s]
+
 35%|███▌      | 27/77 [00:02<00:05,  9.79it/s]
+
 36%|███▋      | 28/77 [00:02<00:05,  9.78it/s]
+
 38%|███▊      | 29/77 [00:02<00:04,  9.79it/s]
+
 39%|███▉      | 30/77 [00:02<00:04,  9.78it/s]
+
 40%|████      | 31/77 [00:03<00:04,  9.79it/s]
+
 42%|████▏     | 32/77 [00:03<00:04,  9.77it/s]
+
 43%|████▎     | 33/77 [00:03<00:04,  9.77it/s]
+
 44%|████▍     | 34/77 [00:03<00:04,  9.78it/s]
+
 45%|████▌     | 35/77 [00:03<00:04,  9.79it/s]
+
 47%|████▋     | 36/77 [00:03<00:04,  9.78it/s]
+
 48%|████▊     | 37/77 [00:03<00:04,  9.76it/s]
+
 49%|████▉     | 38/77 [00:03<00:03,  9.77it/s]
+
 51%|█████     | 39/77 [00:03<00:03,  9.78it/s]
+
 52%|█████▏    | 40/77 [00:03<00:03,  9.78it/s]
+
 53%|█████▎    | 41/77 [00:04<00:03,  9.78it/s]
+
 55%|█████▍    | 42/77 [00:04<00:03,  9.78it/s]
+
 56%|█████▌    | 43/77 [00:04<00:03,  9.78it/s]
+
 57%|█████▋    | 44/77 [00:04<00:03,  9.78it/s]
+
 58%|█████▊    | 45/77 [00:04<00:03,  9.78it/s]
+
 60%|█████▉    | 46/77 [00:04<00:03,  9.78it/s]
+
 61%|██████    | 47/77 [00:04<00:03,  9.78it/s]
+
 62%|██████▏   | 48/77 [00:04<00:02,  9.78it/s]
+
 64%|██████▎   | 49/77 [00:04<00:02,  9.78it/s]
+
 65%|██████▍   | 50/77 [00:05<00:02,  9.78it/s]
+
 66%|██████▌   | 51/77 [00:05<00:02,  9.78it/s]
+
 68%|██████▊   | 52/77 [00:05<00:02,  9.79it/s]
+
 69%|██████▉   | 53/77 [00:05<00:02,  9.78it/s]
+
 70%|███████   | 54/77 [00:05<00:02,  9.79it/s]
+
 71%|███████▏  | 55/77 [00:05<00:02,  9.78it/s]
+
 73%|███████▎  | 56/77 [00:05<00:02,  9.78it/s]
+
 74%|███████▍  | 57/77 [00:05<00:02,  9.78it/s]
+
 75%|███████▌  | 58/77 [00:05<00:01,  9.75it/s]
+
 77%|███████▋  | 59/77 [00:05<00:01,  9.77it/s]
+
 78%|███████▊  | 60/77 [00:06<00:01,  9.77it/s]
+
 79%|███████▉  | 61/77 [00:06<00:01,  9.80it/s]
+
 81%|████████  | 62/77 [00:06<00:01,  9.80it/s]
+
 82%|████████▏ | 63/77 [00:06<00:01,  9.80it/s]
+
 83%|████████▎ | 64/77 [00:06<00:01,  9.80it/s]
+
 84%|████████▍ | 65/77 [00:06<00:01,  9.79it/s]
+
 86%|████████▌ | 66/77 [00:06<00:01,  9.80it/s]
+
 87%|████████▋ | 67/77 [00:06<00:01,  9.79it/s]
+
 88%|████████▊ | 68/77 [00:06<00:00,  9.79it/s]
+
 90%|████████▉ | 69/77 [00:06<00:00,  9.79it/s]
+
 91%|█████████ | 70/77 [00:07<00:00,  9.79it/s]
+
 92%|█████████▏| 71/77 [00:07<00:00,  9.79it/s]
+
 94%|█████████▎| 72/77 [00:07<00:00,  9.79it/s]
+
 95%|█████████▍| 73/77 [00:07<00:00,  9.79it/s]
+
 96%|█████████▌| 74/77 [00:07<00:00,  9.79it/s]
+
 97%|█████████▋| 75/77 [00:07<00:00,  9.78it/s]
+
 99%|█████████▊| 76/77 [00:07<00:00,  9.78it/s]
+
100%|██████████| 77/77 [00:07<00:00,  9.78it/s]
                                                 
+
                                               
{'eval_loss': '4.124', 'eval_runtime': '7.912', 'eval_samples_per_second': '308.5', 'eval_steps_per_second': '9.732', 'epoch': '7'}
+
 58%|█████▊    | 504/864 [22:38<12:08,  2.02s/it]
+
100%|██████████| 77/77 [00:07<00:00,  9.78it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0012/checkpoint-504
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0012/checkpoint-504/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0012/checkpoint-504/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:03<00:00,  3.65s/it]
Writing model shards: 100%|██████████| 1/1 [00:03<00:00,  3.65s/it]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0012/checkpoint-504/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0012/checkpoint-504/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0012/checkpoint-504/tokenizer_config.json
+[eval] epoch 7.0 step 504 loss 4.1240 (best 4.1240 @ ep 7.0, 0/10000 since best)
+
+
                                               
 58%|█████▊    | 505/864 [22:52<47:42,  7.97s/it]
 59%|█████▊    | 506/864 [22:54<37:27,  6.28s/it]
 59%|█████▊    | 507/864 [22:56<30:16,  5.09s/it]
 59%|█████▉    | 508/864 [22:58<25:15,  4.26s/it]
 59%|█████▉    | 509/864 [23:01<21:45,  3.68s/it]
 59%|█████▉    | 510/864 [23:03<19:17,  3.27s/it]
                                                 
{'loss': '3.777', 'grad_norm': '0.2539', 'learning_rate': '0.0012', 'epoch': '7.084'}
+
 59%|█████▉    | 510/864 [23:03<19:17,  3.27s/it]
 59%|█████▉    | 511/864 [23:05<17:34,  2.99s/it]
 59%|█████▉    | 512/864 [23:08<16:21,  2.79s/it]
 59%|█████▉    | 513/864 [23:10<15:29,  2.65s/it]
 59%|█████▉    | 514/864 [23:12<14:52,  2.55s/it]
 60%|█████▉    | 515/864 [23:15<14:25,  2.48s/it]
 60%|█████▉    | 516/864 [23:17<14:06,  2.43s/it]
 60%|█████▉    | 517/864 [23:19<13:52,  2.40s/it]
 60%|█████▉    | 518/864 [23:22<14:19,  2.48s/it]
 60%|██████    | 519/864 [23:24<14:00,  2.44s/it]
 60%|██████    | 520/864 [23:27<13:46,  2.40s/it]
                                                 
{'loss': '3.745', 'grad_norm': '0.2621', 'learning_rate': '0.0012', 'epoch': '7.224'}
+
 60%|██████    | 520/864 [23:27<13:46,  2.40s/it]
 60%|██████    | 521/864 [23:29<13:36,  2.38s/it]
 60%|██████    | 522/864 [23:31<13:28,  2.36s/it]
 61%|██████    | 523/864 [23:34<13:21,  2.35s/it]
 61%|██████    | 524/864 [23:36<13:16,  2.34s/it]
 61%|██████    | 525/864 [23:38<13:12,  2.34s/it]
 61%|██████    | 526/864 [23:41<13:08,  2.33s/it]
 61%|██████    | 527/864 [23:43<13:05,  2.33s/it]
 61%|██████    | 528/864 [23:45<13:01,  2.33s/it]
 61%|██████    | 529/864 [23:48<12:59,  2.33s/it]
 61%|██████▏   | 530/864 [23:50<12:56,  2.32s/it]
                                                 
{'loss': '3.717', 'grad_norm': '0.1575', 'learning_rate': '0.0012', 'epoch': '7.363'}
+
 61%|██████▏   | 530/864 [23:50<12:56,  2.32s/it]
 61%|██████▏   | 531/864 [23:52<12:54,  2.33s/it]
 62%|██████▏   | 532/864 [23:55<12:51,  2.32s/it]
 62%|██████▏   | 533/864 [23:57<12:49,  2.32s/it]
 62%|██████▏   | 534/864 [23:59<12:46,  2.32s/it]
 62%|██████▏   | 535/864 [24:02<13:23,  2.44s/it]
 62%|██████▏   | 536/864 [24:05<13:44,  2.51s/it]
 62%|██████▏   | 537/864 [24:07<13:22,  2.46s/it]
 62%|██████▏   | 538/864 [24:09<13:07,  2.42s/it]
 62%|██████▏   | 539/864 [24:12<12:55,  2.39s/it]
 62%|██████▎   | 540/864 [24:14<12:47,  2.37s/it]
                                                 
{'loss': '3.705', 'grad_norm': '0.1871', 'learning_rate': '0.0012', 'epoch': '7.503'}
+
 62%|██████▎   | 540/864 [24:14<12:47,  2.37s/it]
 63%|██████▎   | 541/864 [24:16<12:40,  2.35s/it]
 63%|██████▎   | 542/864 [24:19<12:35,  2.35s/it]
 63%|██████▎   | 543/864 [24:21<12:30,  2.34s/it]
 63%|██████▎   | 544/864 [24:23<12:26,  2.33s/it]
 63%|██████▎   | 545/864 [24:26<12:23,  2.33s/it]
 63%|██████▎   | 546/864 [24:28<12:20,  2.33s/it]
 63%|██████▎   | 547/864 [24:30<12:17,  2.33s/it]
 63%|██████▎   | 548/864 [24:32<12:14,  2.33s/it]
 64%|██████▎   | 549/864 [24:35<12:12,  2.32s/it]
 64%|██████▎   | 550/864 [24:37<12:09,  2.32s/it]
                                                 
{'loss': '3.697', 'grad_norm': '0.1407', 'learning_rate': '0.0012', 'epoch': '7.643'}
+
 64%|██████▎   | 550/864 [24:37<12:09,  2.32s/it]
 64%|██████▍   | 551/864 [24:39<12:07,  2.32s/it]
 64%|██████▍   | 552/864 [24:42<12:04,  2.32s/it]
 64%|██████▍   | 553/864 [24:44<12:02,  2.32s/it]
 64%|██████▍   | 554/864 [24:46<11:59,  2.32s/it]
 64%|██████▍   | 555/864 [24:49<11:57,  2.32s/it]
 64%|██████▍   | 556/864 [24:51<11:55,  2.32s/it]
 64%|██████▍   | 557/864 [24:53<11:53,  2.32s/it]
 65%|██████▍   | 558/864 [24:56<11:51,  2.32s/it]
 65%|██████▍   | 559/864 [24:58<11:48,  2.32s/it]
 65%|██████▍   | 560/864 [25:00<11:46,  2.32s/it]
                                                 
{'loss': '3.686', 'grad_norm': '0.1328', 'learning_rate': '0.0012', 'epoch': '7.783'}
+
 65%|██████▍   | 560/864 [25:00<11:46,  2.32s/it]
 65%|██████▍   | 561/864 [25:03<11:44,  2.32s/it]
 65%|██████▌   | 562/864 [25:05<11:41,  2.32s/it]
 65%|██████▌   | 563/864 [25:07<11:39,  2.32s/it]
 65%|██████▌   | 564/864 [25:10<11:36,  2.32s/it]
 65%|██████▌   | 565/864 [25:12<11:34,  2.32s/it]
 66%|██████▌   | 566/864 [25:14<11:31,  2.32s/it]
 66%|██████▌   | 567/864 [25:17<11:29,  2.32s/it]
 66%|██████▌   | 568/864 [25:19<11:27,  2.32s/it]
 66%|██████▌   | 569/864 [25:21<11:25,  2.32s/it]
 66%|██████▌   | 570/864 [25:24<11:23,  2.32s/it]
                                                 
{'loss': '3.696', 'grad_norm': '0.2193', 'learning_rate': '0.0012', 'epoch': '7.922'}
+
 66%|██████▌   | 570/864 [25:24<11:23,  2.32s/it]
 66%|██████▌   | 571/864 [25:26<11:20,  2.32s/it]
 66%|██████▌   | 572/864 [25:28<11:18,  2.32s/it]
 66%|██████▋   | 573/864 [25:31<11:16,  2.32s/it]
 66%|██████▋   | 574/864 [25:33<11:14,  2.32s/it]
 67%|██████▋   | 575/864 [25:35<11:11,  2.32s/it]
 67%|██████▋   | 576/864 [25:37<09:43,  2.03s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 4
+
+
  0%|          | 0/77 [00:00<?, ?it/s]
+
  4%|▍         | 3/77 [00:00<00:04, 15.05it/s]
+
  6%|▋         | 5/77 [00:00<00:06, 11.91it/s]
+
  9%|▉         | 7/77 [00:00<00:06, 10.93it/s]
+
 12%|█▏        | 9/77 [00:00<00:06, 10.48it/s]
+
 14%|█▍        | 11/77 [00:01<00:06, 10.23it/s]
+
 17%|█▋        | 13/77 [00:01<00:06, 10.08it/s]
+
 19%|█▉        | 15/77 [00:01<00:06,  9.98it/s]
+
 21%|██        | 16/77 [00:01<00:06,  9.94it/s]
+
 22%|██▏       | 17/77 [00:01<00:06,  9.91it/s]
+
 23%|██▎       | 18/77 [00:01<00:05,  9.85it/s]
+
 25%|██▍       | 19/77 [00:01<00:05,  9.85it/s]
+
 26%|██▌       | 20/77 [00:01<00:05,  9.84it/s]
+
 27%|██▋       | 21/77 [00:02<00:05,  9.82it/s]
+
 29%|██▊       | 22/77 [00:02<00:05,  9.80it/s]
+
 30%|██▉       | 23/77 [00:02<00:05,  9.80it/s]
+
 31%|███       | 24/77 [00:02<00:05,  9.80it/s]
+
 32%|███▏      | 25/77 [00:02<00:05,  9.79it/s]
+
 34%|███▍      | 26/77 [00:02<00:05,  9.79it/s]
+
 35%|███▌      | 27/77 [00:02<00:05,  9.77it/s]
+
 36%|███▋      | 28/77 [00:02<00:05,  9.78it/s]
+
 38%|███▊      | 29/77 [00:02<00:04,  9.77it/s]
+
 39%|███▉      | 30/77 [00:02<00:04,  9.78it/s]
+
 40%|████      | 31/77 [00:03<00:04,  9.77it/s]
+
 42%|████▏     | 32/77 [00:03<00:04,  9.78it/s]
+
 43%|████▎     | 33/77 [00:03<00:04,  9.78it/s]
+
 44%|████▍     | 34/77 [00:03<00:04,  9.78it/s]
+
 45%|████▌     | 35/77 [00:03<00:04,  9.78it/s]
+
 47%|████▋     | 36/77 [00:03<00:04,  9.78it/s]
+
 48%|████▊     | 37/77 [00:03<00:04,  9.77it/s]
+
 49%|████▉     | 38/77 [00:03<00:03,  9.77it/s]
+
 51%|█████     | 39/77 [00:03<00:03,  9.78it/s]
+
 52%|█████▏    | 40/77 [00:03<00:03,  9.78it/s]
+
 53%|█████▎    | 41/77 [00:04<00:03,  9.78it/s]
+
 55%|█████▍    | 42/77 [00:04<00:03,  9.78it/s]
+
 56%|█████▌    | 43/77 [00:04<00:03,  9.77it/s]
+
 57%|█████▋    | 44/77 [00:04<00:03,  9.78it/s]
+
 58%|█████▊    | 45/77 [00:04<00:03,  9.78it/s]
+
 60%|█████▉    | 46/77 [00:04<00:03,  9.78it/s]
+
 61%|██████    | 47/77 [00:04<00:03,  9.78it/s]
+
 62%|██████▏   | 48/77 [00:04<00:02,  9.78it/s]
+
 64%|██████▎   | 49/77 [00:04<00:02,  9.78it/s]
+
 65%|██████▍   | 50/77 [00:05<00:02,  9.78it/s]
+
 66%|██████▌   | 51/77 [00:05<00:02,  9.78it/s]
+
 68%|██████▊   | 52/77 [00:05<00:02,  9.77it/s]
+
 69%|██████▉   | 53/77 [00:05<00:02,  9.77it/s]
+
 70%|███████   | 54/77 [00:05<00:02,  9.77it/s]
+
 71%|███████▏  | 55/77 [00:05<00:02,  9.77it/s]
+
 73%|███████▎  | 56/77 [00:05<00:02,  9.78it/s]
+
 74%|███████▍  | 57/77 [00:05<00:02,  9.78it/s]
+
 75%|███████▌  | 58/77 [00:05<00:01,  9.77it/s]
+
 77%|███████▋  | 59/77 [00:05<00:01,  9.78it/s]
+
 78%|███████▊  | 60/77 [00:06<00:01,  9.78it/s]
+
 79%|███████▉  | 61/77 [00:06<00:01,  9.79it/s]
+
 81%|████████  | 62/77 [00:06<00:01,  9.79it/s]
+
 82%|████████▏ | 63/77 [00:06<00:01,  9.79it/s]
+
 83%|████████▎ | 64/77 [00:06<00:01,  9.79it/s]
+
 84%|████████▍ | 65/77 [00:06<00:01,  9.79it/s]
+
 86%|████████▌ | 66/77 [00:06<00:01,  9.79it/s]
+
 87%|████████▋ | 67/77 [00:06<00:01,  9.79it/s]
+
 88%|████████▊ | 68/77 [00:06<00:00,  9.79it/s]
+
 90%|████████▉ | 69/77 [00:06<00:00,  9.79it/s]
+
 91%|█████████ | 70/77 [00:07<00:00,  9.78it/s]
+
 92%|█████████▏| 71/77 [00:07<00:00,  9.79it/s]
+
 94%|█████████▎| 72/77 [00:07<00:00,  9.78it/s]
+
 95%|█████████▍| 73/77 [00:07<00:00,  9.79it/s]
+
 96%|█████████▌| 74/77 [00:07<00:00,  9.79it/s]
+
 97%|█████████▋| 75/77 [00:07<00:00,  9.78it/s]
+
 99%|█████████▊| 76/77 [00:07<00:00,  9.79it/s]
+
100%|██████████| 77/77 [00:07<00:00,  9.79it/s]
                                                 
+
                                               
{'eval_loss': '4.036', 'eval_runtime': '7.903', 'eval_samples_per_second': '308.9', 'eval_steps_per_second': '9.743', 'epoch': '8'}
+
 67%|██████▋   | 576/864 [25:44<09:43,  2.03s/it]
+
100%|██████████| 77/77 [00:07<00:00,  9.79it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0012/checkpoint-576
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0012/checkpoint-576/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0012/checkpoint-576/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:03<00:00,  3.63s/it]
Writing model shards: 100%|██████████| 1/1 [00:03<00:00,  3.64s/it]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0012/checkpoint-576/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0012/checkpoint-576/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0012/checkpoint-576/tokenizer_config.json
+[eval] epoch 8.0 step 576 loss 4.0361 (best 4.0361 @ ep 8.0, 0/10000 since best)
+
+
                                               
 67%|██████▋   | 577/864 [25:58<38:06,  7.97s/it]
 67%|██████▋   | 578/864 [26:01<29:54,  6.27s/it]
 67%|██████▋   | 579/864 [26:03<24:10,  5.09s/it]
 67%|██████▋   | 580/864 [26:05<20:09,  4.26s/it]
                                                 
{'loss': '3.613', 'grad_norm': '0.3117', 'learning_rate': '0.0012', 'epoch': '8.056'}
+
 67%|██████▋   | 580/864 [26:05<20:09,  4.26s/it]
 67%|██████▋   | 581/864 [26:08<17:21,  3.68s/it]
 67%|██████▋   | 582/864 [26:10<15:23,  3.27s/it]
 67%|██████▋   | 583/864 [26:12<14:00,  2.99s/it]
 68%|██████▊   | 584/864 [26:15<13:01,  2.79s/it]
 68%|██████▊   | 585/864 [26:17<12:19,  2.65s/it]
 68%|██████▊   | 586/864 [26:19<11:50,  2.55s/it]
 68%|██████▊   | 587/864 [26:22<11:28,  2.49s/it]
 68%|██████▊   | 588/864 [26:24<11:12,  2.44s/it]
 68%|██████▊   | 589/864 [26:26<11:01,  2.41s/it]
 68%|██████▊   | 590/864 [26:29<10:52,  2.38s/it]
                                                 
{'loss': '3.611', 'grad_norm': '0.4582', 'learning_rate': '0.0012', 'epoch': '8.196'}
+
 68%|██████▊   | 590/864 [26:29<10:52,  2.38s/it]
 68%|██████▊   | 591/864 [26:31<10:46,  2.37s/it]
 69%|██████▊   | 592/864 [26:33<10:40,  2.35s/it]
 69%|██████▊   | 593/864 [26:36<10:35,  2.35s/it]
 69%|██████▉   | 594/864 [26:38<10:32,  2.34s/it]
 69%|██████▉   | 595/864 [26:40<10:28,  2.34s/it]
 69%|██████▉   | 596/864 [26:43<10:25,  2.33s/it]
 69%|██████▉   | 597/864 [26:45<10:22,  2.33s/it]
 69%|██████▉   | 598/864 [26:47<10:19,  2.33s/it]
 69%|██████▉   | 599/864 [26:50<10:17,  2.33s/it]
 69%|██████▉   | 600/864 [26:52<10:15,  2.33s/it]
                                                 
{'loss': '3.703', 'grad_norm': '0.2828', 'learning_rate': '0.0012', 'epoch': '8.335'}
+
 69%|██████▉   | 600/864 [26:52<10:15,  2.33s/it]
 70%|██████▉   | 601/864 [26:54<10:12,  2.33s/it]
 70%|██████▉   | 602/864 [26:57<10:10,  2.33s/it]
 70%|██████▉   | 603/864 [26:59<10:07,  2.33s/it]
 70%|██████▉   | 604/864 [27:01<10:05,  2.33s/it]
 70%|███████   | 605/864 [27:04<10:02,  2.33s/it]
 70%|███████   | 606/864 [27:06<10:00,  2.33s/it]
 70%|███████   | 607/864 [27:08<09:57,  2.33s/it]
 70%|███████   | 608/864 [27:11<09:55,  2.33s/it]
 70%|███████   | 609/864 [27:13<09:53,  2.33s/it]
 71%|███████   | 610/864 [27:15<09:50,  2.33s/it]
                                                 
{'loss': '3.618', 'grad_norm': '0.1509', 'learning_rate': '0.0012', 'epoch': '8.475'}
+
 71%|███████   | 610/864 [27:15<09:50,  2.33s/it]
 71%|███████   | 611/864 [27:17<09:49,  2.33s/it]
 71%|███████   | 612/864 [27:20<09:46,  2.33s/it]
 71%|███████   | 613/864 [27:22<09:44,  2.33s/it]
 71%|███████   | 614/864 [27:24<09:41,  2.33s/it]
 71%|███████   | 615/864 [27:27<09:39,  2.33s/it]
 71%|███████▏  | 616/864 [27:29<09:37,  2.33s/it]
 71%|███████▏  | 617/864 [27:31<09:34,  2.33s/it]
 72%|███████▏  | 618/864 [27:34<09:32,  2.33s/it]
 72%|███████▏  | 619/864 [27:36<09:30,  2.33s/it]
 72%|███████▏  | 620/864 [27:38<09:27,  2.33s/it]
                                                 
{'loss': '3.558', 'grad_norm': '0.1067', 'learning_rate': '0.0012', 'epoch': '8.615'}
+
 72%|███████▏  | 620/864 [27:38<09:27,  2.33s/it]
 72%|███████▏  | 621/864 [27:41<09:25,  2.33s/it]
 72%|███████▏  | 622/864 [27:43<09:23,  2.33s/it]
 72%|███████▏  | 623/864 [27:45<09:20,  2.33s/it]
 72%|███████▏  | 624/864 [27:48<09:18,  2.33s/it]
 72%|███████▏  | 625/864 [27:50<09:16,  2.33s/it]
 72%|███████▏  | 626/864 [27:52<09:14,  2.33s/it]
 73%|███████▎  | 627/864 [27:55<09:11,  2.33s/it]
 73%|███████▎  | 628/864 [27:57<09:09,  2.33s/it]
 73%|███████▎  | 629/864 [27:59<09:06,  2.33s/it]
 73%|███████▎  | 630/864 [28:02<09:04,  2.33s/it]
                                                 
{'loss': '3.536', 'grad_norm': '0.1145', 'learning_rate': '0.0012', 'epoch': '8.755'}
+
 73%|███████▎  | 630/864 [28:02<09:04,  2.33s/it]
 73%|███████▎  | 631/864 [28:04<09:02,  2.33s/it]
 73%|███████▎  | 632/864 [28:06<08:59,  2.33s/it]
 73%|███████▎  | 633/864 [28:09<08:57,  2.33s/it]
 73%|███████▎  | 634/864 [28:11<08:55,  2.33s/it]
 73%|███████▎  | 635/864 [28:13<08:52,  2.33s/it]
 74%|███████▎  | 636/864 [28:16<08:50,  2.33s/it]
 74%|███████▎  | 637/864 [28:18<08:47,  2.33s/it]
 74%|███████▍  | 638/864 [28:20<08:45,  2.33s/it]
 74%|███████▍  | 639/864 [28:23<08:43,  2.33s/it]
 74%|███████▍  | 640/864 [28:25<08:40,  2.33s/it]
                                                 
{'loss': '3.54', 'grad_norm': '0.1107', 'learning_rate': '0.0012', 'epoch': '8.894'}
+
 74%|███████▍  | 640/864 [28:25<08:40,  2.33s/it]
 74%|███████▍  | 641/864 [28:27<08:38,  2.33s/it]
 74%|███████▍  | 642/864 [28:30<08:36,  2.33s/it]
 74%|███████▍  | 643/864 [28:32<08:33,  2.33s/it]
 75%|███████▍  | 644/864 [28:34<08:31,  2.33s/it]
 75%|███████▍  | 645/864 [28:37<08:29,  2.33s/it]
 75%|███████▍  | 646/864 [28:39<08:27,  2.33s/it]
 75%|███████▍  | 647/864 [28:41<08:25,  2.33s/it]
 75%|███████▌  | 648/864 [28:43<07:17,  2.03s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 4
+
+
  0%|          | 0/77 [00:00<?, ?it/s]
+
  3%|▎         | 2/77 [00:00<00:03, 19.47it/s]
+
  5%|▌         | 4/77 [00:00<00:05, 12.27it/s]
+
  8%|▊         | 6/77 [00:00<00:06, 10.99it/s]
+
 10%|█         | 8/77 [00:00<00:06, 10.47it/s]
+
 13%|█▎        | 10/77 [00:00<00:06, 10.21it/s]
+
 16%|█▌        | 12/77 [00:01<00:06, 10.06it/s]
+
 18%|█▊        | 14/77 [00:01<00:06,  9.96it/s]
+
 21%|██        | 16/77 [00:01<00:06,  9.90it/s]
+
 23%|██▎       | 18/77 [00:01<00:05,  9.86it/s]
+
 25%|██▍       | 19/77 [00:01<00:05,  9.84it/s]
+
 26%|██▌       | 20/77 [00:01<00:05,  9.82it/s]
+
 27%|██▋       | 21/77 [00:02<00:05,  9.81it/s]
+
 29%|██▊       | 22/77 [00:02<00:05,  9.81it/s]
+
 30%|██▉       | 23/77 [00:02<00:05,  9.79it/s]
+
 31%|███       | 24/77 [00:02<00:05,  9.78it/s]
+
 32%|███▏      | 25/77 [00:02<00:05,  9.78it/s]
+
 34%|███▍      | 26/77 [00:02<00:05,  9.79it/s]
+
 35%|███▌      | 27/77 [00:02<00:05,  9.78it/s]
+
 36%|███▋      | 28/77 [00:02<00:05,  9.77it/s]
+
 38%|███▊      | 29/77 [00:02<00:04,  9.77it/s]
+
 39%|███▉      | 30/77 [00:02<00:04,  9.78it/s]
+
 40%|████      | 31/77 [00:03<00:04,  9.78it/s]
+
 42%|████▏     | 32/77 [00:03<00:04,  9.77it/s]
+
 43%|████▎     | 33/77 [00:03<00:04,  9.77it/s]
+
 44%|████▍     | 34/77 [00:03<00:04,  9.78it/s]
+
 45%|████▌     | 35/77 [00:03<00:04,  9.77it/s]
+
 47%|████▋     | 36/77 [00:03<00:04,  9.76it/s]
+
 48%|████▊     | 37/77 [00:03<00:04,  9.77it/s]
+
 49%|████▉     | 38/77 [00:03<00:03,  9.77it/s]
+
 51%|█████     | 39/77 [00:03<00:03,  9.78it/s]
+
 52%|█████▏    | 40/77 [00:03<00:03,  9.77it/s]
+
 53%|█████▎    | 41/77 [00:04<00:03,  9.78it/s]
+
 55%|█████▍    | 42/77 [00:04<00:03,  9.78it/s]
+
 56%|█████▌    | 43/77 [00:04<00:03,  9.78it/s]
+
 57%|█████▋    | 44/77 [00:04<00:03,  9.74it/s]
+
 58%|█████▊    | 45/77 [00:04<00:03,  9.77it/s]
+
 60%|█████▉    | 46/77 [00:04<00:03,  9.78it/s]
+
 61%|██████    | 47/77 [00:04<00:03,  9.78it/s]
+
 62%|██████▏   | 48/77 [00:04<00:02,  9.77it/s]
+
 64%|██████▎   | 49/77 [00:04<00:02,  9.78it/s]
+
 65%|██████▍   | 50/77 [00:05<00:02,  9.78it/s]
+
 66%|██████▌   | 51/77 [00:05<00:02,  9.78it/s]
+
 68%|██████▊   | 52/77 [00:05<00:02,  9.77it/s]
+
 69%|██████▉   | 53/77 [00:05<00:02,  9.78it/s]
+
 70%|███████   | 54/77 [00:05<00:02,  9.78it/s]
+
 71%|███████▏  | 55/77 [00:05<00:02,  9.78it/s]
+
 73%|███████▎  | 56/77 [00:05<00:02,  9.77it/s]
+
 74%|███████▍  | 57/77 [00:05<00:02,  9.78it/s]
+
 75%|███████▌  | 58/77 [00:05<00:01,  9.78it/s]
+
 77%|███████▋  | 59/77 [00:05<00:01,  9.78it/s]
+
 78%|███████▊  | 60/77 [00:06<00:01,  9.77it/s]
+
 79%|███████▉  | 61/77 [00:06<00:01,  9.79it/s]
+
 81%|████████  | 62/77 [00:06<00:01,  9.80it/s]
+
 82%|████████▏ | 63/77 [00:06<00:01,  9.79it/s]
+
 83%|████████▎ | 64/77 [00:06<00:01,  9.79it/s]
+
 84%|████████▍ | 65/77 [00:06<00:01,  9.78it/s]
+
 86%|████████▌ | 66/77 [00:06<00:01,  9.78it/s]
+
 87%|████████▋ | 67/77 [00:06<00:01,  9.79it/s]
+
 88%|████████▊ | 68/77 [00:06<00:00,  9.78it/s]
+
 90%|████████▉ | 69/77 [00:06<00:00,  9.78it/s]
+
 91%|█████████ | 70/77 [00:07<00:00,  9.78it/s]
+
 92%|█████████▏| 71/77 [00:07<00:00,  9.78it/s]
+
 94%|█████████▎| 72/77 [00:07<00:00,  9.78it/s]
+
 95%|█████████▍| 73/77 [00:07<00:00,  9.78it/s]
+
 96%|█████████▌| 74/77 [00:07<00:00,  9.79it/s]
+
 97%|█████████▋| 75/77 [00:07<00:00,  9.79it/s]
+
 99%|█████████▊| 76/77 [00:07<00:00,  9.79it/s]
+
100%|██████████| 77/77 [00:07<00:00,  9.78it/s]
                                                 
+
                                               
{'eval_loss': '3.974', 'eval_runtime': '7.917', 'eval_samples_per_second': '308.3', 'eval_steps_per_second': '9.726', 'epoch': '9'}
+
 75%|███████▌  | 648/864 [28:51<07:17,  2.03s/it]
+
100%|██████████| 77/77 [00:07<00:00,  9.78it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0012/checkpoint-648
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0012/checkpoint-648/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0012/checkpoint-648/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:03<00:00,  3.65s/it]
Writing model shards: 100%|██████████| 1/1 [00:03<00:00,  3.65s/it]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0012/checkpoint-648/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0012/checkpoint-648/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0012/checkpoint-648/tokenizer_config.json
+[eval] epoch 9.0 step 648 loss 3.9742 (best 3.9742 @ ep 9.0, 0/10000 since best)
+
+
                                               
 75%|███████▌  | 649/864 [29:04<28:31,  7.96s/it]
 75%|███████▌  | 650/864 [29:07<22:21,  6.27s/it]
                                                 
{'loss': '3.481', 'grad_norm': '0.2092', 'learning_rate': '0.0012', 'epoch': '9.028'}
+
 75%|███████▌  | 650/864 [29:07<22:21,  6.27s/it]
 75%|███████▌  | 651/864 [29:09<18:02,  5.08s/it]
 75%|███████▌  | 652/864 [29:11<15:02,  4.26s/it]
 76%|███████▌  | 653/864 [29:14<12:56,  3.68s/it]
 76%|███████▌  | 654/864 [29:16<11:27,  3.27s/it]
 76%|███████▌  | 655/864 [29:18<10:24,  2.99s/it]
 76%|███████▌  | 656/864 [29:21<09:40,  2.79s/it]
 76%|███████▌  | 657/864 [29:23<09:08,  2.65s/it]
 76%|███████▌  | 658/864 [29:25<08:45,  2.55s/it]
 76%|███████▋  | 659/864 [29:28<08:29,  2.48s/it]
 76%|███████▋  | 660/864 [29:30<08:17,  2.44s/it]
                                                 
{'loss': '3.316', 'grad_norm': '0.2739', 'learning_rate': '0.0012', 'epoch': '9.168'}
+
 76%|███████▋  | 660/864 [29:30<08:17,  2.44s/it]
 77%|███████▋  | 661/864 [29:32<08:08,  2.40s/it]
 77%|███████▋  | 662/864 [29:35<08:00,  2.38s/it]
 77%|███████▋  | 663/864 [29:37<07:55,  2.36s/it]
 77%|███████▋  | 664/864 [29:39<07:50,  2.35s/it]
 77%|███████▋  | 665/864 [29:42<07:46,  2.34s/it]
 77%|███████▋  | 666/864 [29:44<07:43,  2.34s/it]
 77%|███████▋  | 667/864 [29:46<07:39,  2.33s/it]
 77%|███████▋  | 668/864 [29:49<07:37,  2.33s/it]
 77%|███████▋  | 669/864 [29:51<07:34,  2.33s/it]
 78%|███████▊  | 670/864 [29:53<07:31,  2.33s/it]
                                                 
{'loss': '3.342', 'grad_norm': '0.2057', 'learning_rate': '0.0012', 'epoch': '9.307'}
+
 78%|███████▊  | 670/864 [29:53<07:31,  2.33s/it]
 78%|███████▊  | 671/864 [29:56<07:29,  2.33s/it]
 78%|███████▊  | 672/864 [29:58<07:26,  2.33s/it]
 78%|███████▊  | 673/864 [30:00<07:24,  2.33s/it]
 78%|███████▊  | 674/864 [30:03<07:21,  2.33s/it]
 78%|███████▊  | 675/864 [30:05<07:19,  2.33s/it]
 78%|███████▊  | 676/864 [30:07<07:17,  2.33s/it]
 78%|███████▊  | 677/864 [30:09<07:14,  2.33s/it]
 78%|███████▊  | 678/864 [30:12<07:12,  2.33s/it]
 79%|███████▊  | 679/864 [30:14<07:10,  2.33s/it]
 79%|███████▊  | 680/864 [30:16<07:07,  2.33s/it]
                                                 
{'loss': '3.346', 'grad_norm': '0.148', 'learning_rate': '0.0012', 'epoch': '9.447'}
+
 79%|███████▊  | 680/864 [30:17<07:07,  2.33s/it]
 79%|███████▉  | 681/864 [30:19<07:05,  2.33s/it]
 79%|███████▉  | 682/864 [30:21<07:22,  2.43s/it]
 79%|███████▉  | 683/864 [30:24<07:13,  2.40s/it]
 79%|███████▉  | 684/864 [30:26<07:07,  2.38s/it]
 79%|███████▉  | 685/864 [30:28<07:02,  2.36s/it]
 79%|███████▉  | 686/864 [30:31<06:58,  2.35s/it]
 80%|███████▉  | 687/864 [30:33<06:54,  2.34s/it]
 80%|███████▉  | 688/864 [30:35<06:50,  2.33s/it]
 80%|███████▉  | 689/864 [30:38<06:48,  2.33s/it]
 80%|███████▉  | 690/864 [30:40<06:45,  2.33s/it]
                                                 
{'loss': '3.334', 'grad_norm': '0.1118', 'learning_rate': '0.0012', 'epoch': '9.587'}
+
 80%|███████▉  | 690/864 [30:40<06:45,  2.33s/it]
 80%|███████▉  | 691/864 [30:42<06:42,  2.33s/it]
 80%|████████  | 692/864 [30:45<06:40,  2.33s/it]
 80%|████████  | 693/864 [30:47<06:37,  2.33s/it]
 80%|████████  | 694/864 [30:49<06:35,  2.32s/it]
 80%|████████  | 695/864 [30:52<06:32,  2.32s/it]
 81%|████████  | 696/864 [30:54<06:30,  2.32s/it]
 81%|████████  | 697/864 [30:56<06:27,  2.32s/it]
 81%|████████  | 698/864 [30:59<06:25,  2.32s/it]
 81%|████████  | 699/864 [31:01<06:23,  2.32s/it]
 81%|████████  | 700/864 [31:04<06:40,  2.44s/it]
                                                 
{'loss': '3.356', 'grad_norm': '0.1176', 'learning_rate': '0.0012', 'epoch': '9.727'}
+
 81%|████████  | 700/864 [31:04<06:40,  2.44s/it]
 81%|████████  | 701/864 [31:06<06:32,  2.41s/it]
 81%|████████▏ | 702/864 [31:08<06:25,  2.38s/it]
 81%|████████▏ | 703/864 [31:11<06:20,  2.37s/it]
 81%|████████▏ | 704/864 [31:13<06:16,  2.35s/it]
 82%|████████▏ | 705/864 [31:15<06:12,  2.34s/it]
 82%|████████▏ | 706/864 [31:18<06:09,  2.34s/it]
 82%|████████▏ | 707/864 [31:20<06:06,  2.33s/it]
 82%|████████▏ | 708/864 [31:22<06:03,  2.33s/it]
 82%|████████▏ | 709/864 [31:25<06:01,  2.33s/it]
 82%|████████▏ | 710/864 [31:27<05:58,  2.33s/it]
                                                 
{'loss': '3.356', 'grad_norm': '0.1118', 'learning_rate': '0.0012', 'epoch': '9.866'}
+
 82%|████████▏ | 710/864 [31:27<05:58,  2.33s/it]
 82%|████████▏ | 711/864 [31:29<05:55,  2.33s/it]
 82%|████████▏ | 712/864 [31:32<05:53,  2.32s/it]
 83%|████████▎ | 713/864 [31:34<05:51,  2.32s/it]
 83%|████████▎ | 714/864 [31:36<05:48,  2.33s/it]
 83%|████████▎ | 715/864 [31:39<05:46,  2.32s/it]
 83%|████████▎ | 716/864 [31:41<05:43,  2.32s/it]
 83%|████████▎ | 717/864 [31:43<05:41,  2.32s/it]
 83%|████████▎ | 718/864 [31:46<05:39,  2.32s/it]
 83%|████████▎ | 719/864 [31:48<05:37,  2.32s/it]
 83%|████████▎ | 720/864 [31:49<04:51,  2.03s/it]
                                                 
{'loss': '3.356', 'grad_norm': '0.1362', 'learning_rate': '0.0012', 'epoch': '10'}
+
 83%|████████▎ | 720/864 [31:49<04:51,  2.03s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 4
+
+
  0%|          | 0/77 [00:00<?, ?it/s]
+
  4%|▍         | 3/77 [00:00<00:04, 14.98it/s]
+
  6%|▋         | 5/77 [00:00<00:06, 11.89it/s]
+
  9%|▉         | 7/77 [00:00<00:06, 10.91it/s]
+
 12%|█▏        | 9/77 [00:00<00:06, 10.47it/s]
+
 14%|█▍        | 11/77 [00:01<00:06, 10.21it/s]
+
 17%|█▋        | 13/77 [00:01<00:06, 10.06it/s]
+
 19%|█▉        | 15/77 [00:01<00:06,  9.96it/s]
+
 21%|██        | 16/77 [00:01<00:06,  9.93it/s]
+
 22%|██▏       | 17/77 [00:01<00:06,  9.90it/s]
+
 23%|██▎       | 18/77 [00:01<00:05,  9.84it/s]
+
 25%|██▍       | 19/77 [00:01<00:05,  9.82it/s]
+
 26%|██▌       | 20/77 [00:01<00:05,  9.81it/s]
+
 27%|██▋       | 21/77 [00:02<00:05,  9.79it/s]
+
 29%|██▊       | 22/77 [00:02<00:05,  9.79it/s]
+
 30%|██▉       | 23/77 [00:02<00:05,  9.77it/s]
+
 31%|███       | 24/77 [00:02<00:05,  9.79it/s]
+
 32%|███▏      | 25/77 [00:02<00:05,  9.79it/s]
+
 34%|███▍      | 26/77 [00:02<00:05,  9.78it/s]
+
 35%|███▌      | 27/77 [00:02<00:05,  9.77it/s]
+
 36%|███▋      | 28/77 [00:02<00:05,  9.78it/s]
+
 38%|███▊      | 29/77 [00:02<00:04,  9.78it/s]
+
 39%|███▉      | 30/77 [00:02<00:04,  9.78it/s]
+
 40%|████      | 31/77 [00:03<00:04,  9.77it/s]
+
 42%|████▏     | 32/77 [00:03<00:04,  9.78it/s]
+
 43%|████▎     | 33/77 [00:03<00:04,  9.78it/s]
+
 44%|████▍     | 34/77 [00:03<00:04,  9.78it/s]
+
 45%|████▌     | 35/77 [00:03<00:04,  9.77it/s]
+
 47%|████▋     | 36/77 [00:03<00:04,  9.78it/s]
+
 48%|████▊     | 37/77 [00:03<00:04,  9.79it/s]
+
 49%|████▉     | 38/77 [00:03<00:03,  9.78it/s]
+
 51%|█████     | 39/77 [00:03<00:03,  9.76it/s]
+
 52%|█████▏    | 40/77 [00:03<00:03,  9.77it/s]
+
 53%|█████▎    | 41/77 [00:04<00:03,  9.78it/s]
+
 55%|█████▍    | 42/77 [00:04<00:03,  9.78it/s]
+
 56%|█████▌    | 43/77 [00:04<00:03,  9.77it/s]
+
 57%|█████▋    | 44/77 [00:04<00:03,  9.77it/s]
+
 58%|█████▊    | 45/77 [00:04<00:03,  9.78it/s]
+
 60%|█████▉    | 46/77 [00:04<00:03,  9.78it/s]
+
 61%|██████    | 47/77 [00:04<00:03,  9.77it/s]
+
 62%|██████▏   | 48/77 [00:04<00:02,  9.77it/s]
+
 64%|██████▎   | 49/77 [00:04<00:02,  9.78it/s]
+
 65%|██████▍   | 50/77 [00:05<00:02,  9.78it/s]
+
 66%|██████▌   | 51/77 [00:05<00:02,  9.77it/s]
+
 68%|██████▊   | 52/77 [00:05<00:02,  9.77it/s]
+
 69%|██████▉   | 53/77 [00:05<00:02,  9.78it/s]
+
 70%|███████   | 54/77 [00:05<00:02,  9.78it/s]
+
 71%|███████▏  | 55/77 [00:05<00:02,  9.78it/s]
+
 73%|███████▎  | 56/77 [00:05<00:02,  9.78it/s]
+
 74%|███████▍  | 57/77 [00:05<00:02,  9.77it/s]
+
 75%|███████▌  | 58/77 [00:05<00:01,  9.78it/s]
+
 77%|███████▋  | 59/77 [00:05<00:01,  9.78it/s]
+
 78%|███████▊  | 60/77 [00:06<00:01,  9.78it/s]
+
 79%|███████▉  | 61/77 [00:06<00:01,  9.80it/s]
+
 81%|████████  | 62/77 [00:06<00:01,  9.79it/s]
+
 82%|████████▏ | 63/77 [00:06<00:01,  9.79it/s]
+
 83%|████████▎ | 64/77 [00:06<00:01,  9.78it/s]
+
 84%|████████▍ | 65/77 [00:06<00:01,  9.79it/s]
+
 86%|████████▌ | 66/77 [00:06<00:01,  9.79it/s]
+
 87%|████████▋ | 67/77 [00:06<00:01,  9.78it/s]
+
 88%|████████▊ | 68/77 [00:06<00:00,  9.78it/s]
+
 90%|████████▉ | 69/77 [00:06<00:00,  9.79it/s]
+
 91%|█████████ | 70/77 [00:07<00:00,  9.79it/s]
+
 92%|█████████▏| 71/77 [00:07<00:00,  9.78it/s]
+
 94%|█████████▎| 72/77 [00:07<00:00,  9.78it/s]
+
 95%|█████████▍| 73/77 [00:07<00:00,  9.78it/s]
+
 96%|█████████▌| 74/77 [00:07<00:00,  9.78it/s]
+
 97%|█████████▋| 75/77 [00:07<00:00,  9.77it/s]
+
 99%|█████████▊| 76/77 [00:07<00:00,  9.78it/s]
+
100%|██████████| 77/77 [00:07<00:00,  9.78it/s]
                                                 
+
                                               
{'eval_loss': '3.954', 'eval_runtime': '7.918', 'eval_samples_per_second': '308.3', 'eval_steps_per_second': '9.724', 'epoch': '10'}
+
 83%|████████▎ | 720/864 [31:57<04:51,  2.03s/it]
+
100%|██████████| 77/77 [00:07<00:00,  9.78it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0012/checkpoint-720
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0012/checkpoint-720/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0012/checkpoint-720/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:03<00:00,  3.55s/it]
Writing model shards: 100%|██████████| 1/1 [00:03<00:00,  3.55s/it]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0012/checkpoint-720/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0012/checkpoint-720/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t0/t0_S5_lr0.0012/checkpoint-720/tokenizer_config.json
+[eval] epoch 10.0 step 720 loss 3.9540 (best 3.9540 @ ep 10.0, 0/10000 since best)
+
+
                                               
 83%|████████▎ | 721/864 [32:10<18:22,  7.71s/it]
 84%|████████▎ | 722/864 [32:12<14:25,  6.09s/it]
 84%|████████▎ | 723/864 [32:15<11:39,  4.96s/it]
 84%|████████▍ | 724/864 [32:17<09:43,  4.17s/it]
 84%|████████▍ | 725/864 [32:19<08:22,  3.62s/it]
 84%|████████▍ | 726/864 [32:22<07:25,  3.23s/it]
 84%|████████▍ | 727/864 [32:24<06:45,  2.96s/it]
 84%|████████▍ | 728/864 [32:26<06:16,  2.77s/it]
 84%|████████▍ | 729/864 [32:29<05:55,  2.63s/it]
 84%|████████▍ | 730/864 [32:31<05:40,  2.54s/it]
                                                 
{'loss': '3.218', 'grad_norm': '0.5767', 'learning_rate': '0.0012', 'epoch': '10.14'}
+
 84%|████████▍ | 730/864 [32:31<05:40,  2.54s/it]
 85%|████████▍ | 731/864 [32:33<05:29,  2.48s/it]
 85%|████████▍ | 732/864 [32:36<05:20,  2.43s/it]
 85%|████████▍ | 733/864 [32:38<05:14,  2.40s/it]
 85%|████████▍ | 734/864 [32:40<05:08,  2.38s/it]
 85%|████████▌ | 735/864 [32:43<05:04,  2.36s/it]
 85%|████████▌ | 736/864 [32:45<05:00,  2.35s/it]
 85%|████████▌ | 737/864 [32:47<04:57,  2.34s/it]
 85%|████████▌ | 738/864 [32:50<04:54,  2.33s/it]
 86%|████████▌ | 739/864 [32:52<04:51,  2.33s/it]
 86%|████████▌ | 740/864 [32:54<04:48,  2.33s/it]
                                                 
{'loss': '3.951', 'grad_norm': '1.034', 'learning_rate': '0.0012', 'epoch': '10.28'}
+
 86%|████████▌ | 740/864 [32:54<04:48,  2.33s/it]
 86%|████████▌ | 741/864 [32:57<04:46,  2.33s/it]
 86%|████████▌ | 742/864 [32:59<04:43,  2.33s/it]
 86%|████████▌ | 743/864 [33:01<04:41,  2.32s/it]
 86%|████████▌ | 744/864 [33:04<04:38,  2.32s/it]
 86%|████████▌ | 745/864 [33:06<04:36,  2.32s/it]
 86%|████████▋ | 746/864 [33:08<04:34,  2.32s/it]
 86%|████████▋ | 747/864 [33:11<04:32,  2.33s/it]
 87%|████████▋ | 748/864 [33:13<04:29,  2.32s/it]
 87%|████████▋ | 749/864 [33:15<04:27,  2.32s/it]
 87%|████████▋ | 750/864 [33:18<04:25,  2.33s/it]
                                                 
{'loss': '8.434', 'grad_norm': '2.058', 'learning_rate': '0.0012', 'epoch': '10.42'}
+
 87%|████████▋ | 750/864 [33:18<04:25,  2.33s/it]
 87%|████████▋ | 751/864 [33:20<04:22,  2.33s/it]
 87%|████████▋ | 752/864 [33:22<04:20,  2.32s/it]
 87%|████████▋ | 753/864 [33:24<04:17,  2.32s/it]
 87%|████████▋ | 754/864 [33:27<04:15,  2.32s/it]
 87%|████████▋ | 755/864 [33:29<04:13,  2.32s/it]
 88%|████████▊ | 756/864 [33:31<04:10,  2.32s/it]
 88%|████████▊ | 757/864 [33:34<04:08,  2.32s/it]
 88%|████████▊ | 758/864 [33:36<04:06,  2.32s/it]
 88%|████████▊ | 759/864 [33:38<04:03,  2.32s/it]
 88%|████████▊ | 760/864 [33:41<04:01,  2.32s/it]
                                                 
{'loss': '6.228', 'grad_norm': '1.03', 'learning_rate': '0.0012', 'epoch': '10.56'}
+
 88%|████████▊ | 760/864 [33:41<04:01,  2.32s/it]
 88%|████████▊ | 761/864 [33:43<03:59,  2.32s/it]
 88%|████████▊ | 762/864 [33:45<03:56,  2.32s/it]
 88%|████████▊ | 763/864 [33:48<03:54,  2.32s/it]
 88%|████████▊ | 764/864 [33:50<03:52,  2.32s/it]
 89%|████████▊ | 765/864 [33:52<03:50,  2.32s/it]
 89%|████████▊ | 766/864 [33:55<03:47,  2.32s/it]
 89%|████████▉ | 767/864 [33:57<03:45,  2.32s/it]
 89%|████████▉ | 768/864 [33:59<03:42,  2.32s/it]
 89%|████████▉ | 769/864 [34:02<03:40,  2.32s/it]
 89%|████████▉ | 770/864 [34:04<03:38,  2.32s/it]
                                                 
{'loss': '5.193', 'grad_norm': '0.6304', 'learning_rate': '0.0012', 'epoch': '10.7'}
+
 89%|████████▉ | 770/864 [34:04<03:38,  2.32s/it]
 89%|████████▉ | 771/864 [34:06<03:36,  2.32s/it]
 89%|████████▉ | 772/864 [34:09<03:33,  2.32s/it]
 89%|████████▉ | 773/864 [34:11<03:31,  2.32s/it]
 90%|████████▉ | 774/864 [34:13<03:29,  2.32s/it]
 90%|████████▉ | 775/864 [34:16<03:26,  2.32s/it]
 90%|████████▉ | 776/864 [34:18<03:24,  2.32s/it]
 90%|████████▉ | 777/864 [34:20<03:22,  2.32s/it]
 90%|█████████ | 778/864 [34:23<03:19,  2.32s/it]
 90%|█████████ | 779/864 [34:25<03:17,  2.32s/it]
 90%|█████████ | 780/864 [34:27<03:15,  2.32s/it]
                                                 
{'loss': '4.598', 'grad_norm': '0.5583', 'learning_rate': '0.0012', 'epoch': '10.84'}
+
 90%|█████████ | 780/864 [34:27<03:15,  2.32s/it]
 90%|█████████ | 781/864 [34:30<03:12,  2.32s/it]
 91%|█████████ | 782/864 [34:32<03:10,  2.32s/it]
 91%|█████████ | 783/864 [34:34<03:08,  2.33s/it]
 91%|█████████ | 784/864 [34:37<03:06,  2.33s/it]
 91%|█████████ | 785/864 [34:39<03:03,  2.32s/it]
 91%|█████████ | 786/864 [34:41<03:01,  2.32s/it]
 91%|█████████ | 787/864 [34:43<02:58,  2.32s/it]
 91%|█████████ | 788/864 [34:46<02:56,  2.32s/it]
 91%|█████████▏| 789/864 [34:48<02:54,  2.32s/it]
 91%|█████████▏| 790/864 [34:50<02:52,  2.32s/it]
                                                 
{'loss': '4.328', 'grad_norm': '0.2157', 'learning_rate': '0.0012', 'epoch': '10.98'}
+
 91%|█████████▏| 790/864 [34:50<02:52,  2.32s/it]
 92%|█████████▏| 791/864 [34:53<02:49,  2.33s/it]
 92%|█████████▏| 792/864 [34:54<02:25,  2.03s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 4
+
+
  0%|          | 0/77 [00:00<?, ?it/s]
+
  4%|▍         | 3/77 [00:00<00:04, 15.07it/s]
+
  6%|▋         | 5/77 [00:00<00:06, 11.90it/s]
+
  9%|▉         | 7/77 [00:00<00:06, 10.93it/s]
+
 12%|█▏        | 9/77 [00:00<00:06, 10.46it/s]
+
 14%|█▍        | 11/77 [00:01<00:06, 10.22it/s]
+
 17%|█▋        | 13/77 [00:01<00:06, 10.06it/s]
+
 19%|█▉        | 15/77 [00:01<00:06,  9.97it/s]
+
 21%|██        | 16/77 [00:01<00:06,  9.94it/s]
+
 22%|██▏       | 17/77 [00:01<00:06,  9.88it/s]
+
 23%|██▎       | 18/77 [00:01<00:05,  9.87it/s]
+
 25%|██▍       | 19/77 [00:01<00:05,  9.84it/s]
+
 26%|██▌       | 20/77 [00:01<00:05,  9.82it/s]
+
 27%|██▋       | 21/77 [00:02<00:05,  9.80it/s]
+
 29%|██▊       | 22/77 [00:02<00:05,  9.79it/s]
+
 30%|██▉       | 23/77 [00:02<00:05,  9.78it/s]
+
 31%|███       | 24/77 [00:02<00:05,  9.78it/s]
+
 32%|███▏      | 25/77 [00:02<00:05,  9.77it/s]
+
 34%|███▍      | 26/77 [00:02<00:05,  9.78it/s]
+
 35%|███▌      | 27/77 [00:02<00:05,  9.77it/s]
+
 36%|███▋      | 28/77 [00:02<00:05,  9.78it/s]
+
 38%|███▊      | 29/77 [00:02<00:04,  9.77it/s]
+
 39%|███▉      | 30/77 [00:02<00:04,  9.77it/s]
+
 40%|████      | 31/77 [00:03<00:04,  9.77it/s]
+
 42%|████▏     | 32/77 [00:03<00:04,  9.77it/s]
+
 43%|████▎     | 33/77 [00:03<00:04,  9.77it/s]
+
 44%|████▍     | 34/77 [00:03<00:04,  9.76it/s]
+
 45%|████▌     | 35/77 [00:03<00:04,  9.77it/s]
+
 47%|████▋     | 36/77 [00:03<00:04,  9.77it/s]
+
 48%|████▊     | 37/77 [00:03<00:04,  9.78it/s]
+
 49%|████▉     | 38/77 [00:03<00:03,  9.77it/s]
+
 51%|█████     | 39/77 [00:03<00:03,  9.77it/s]
+
 52%|█████▏    | 40/77 [00:03<00:03,  9.77it/s]
+
 53%|█████▎    | 41/77 [00:04<00:03,  9.77it/s]
+
 55%|█████▍    | 42/77 [00:04<00:03,  9.76it/s]
+
 56%|█████▌    | 43/77 [00:04<00:03,  9.77it/s]
+
 57%|█████▋    | 44/77 [00:04<00:03,  9.77it/s]
+
 58%|█████▊    | 45/77 [00:04<00:03,  9.77it/s]
+
 60%|█████▉    | 46/77 [00:04<00:03,  9.77it/s]
+
 61%|██████    | 47/77 [00:04<00:03,  9.77it/s]
+
 62%|██████▏   | 48/77 [00:04<00:02,  9.77it/s]
+
 64%|██████▎   | 49/77 [00:04<00:02,  9.77it/s]
+
 65%|██████▍   | 50/77 [00:05<00:02,  9.76it/s]
+
 66%|██████▌   | 51/77 [00:05<00:02,  9.77it/s]
+
 68%|██████▊   | 52/77 [00:05<00:02,  9.77it/s]
+
 69%|██████▉   | 53/77 [00:05<00:02,  9.77it/s]
+
 70%|███████   | 54/77 [00:05<00:02,  9.77it/s]
+
 71%|███████▏  | 55/77 [00:05<00:02,  9.77it/s]
+
 73%|███████▎  | 56/77 [00:05<00:02,  9.78it/s]
+
 74%|███████▍  | 57/77 [00:05<00:02,  9.77it/s]
+
 75%|███████▌  | 58/77 [00:05<00:01,  9.77it/s]
+
 77%|███████▋  | 59/77 [00:05<00:01,  9.77it/s]
+
 78%|███████▊  | 60/77 [00:06<00:01,  9.77it/s]
+
 79%|███████▉  | 61/77 [00:06<00:01,  9.79it/s]
+
 81%|████████  | 62/77 [00:06<00:01,  9.79it/s]
+
 82%|████████▏ | 63/77 [00:06<00:01,  9.78it/s]
+
 83%|████████▎ | 64/77 [00:06<00:01,  9.78it/s]
+
 84%|████████▍ | 65/77 [00:06<00:01,  9.77it/s]
+
 86%|████████▌ | 66/77 [00:06<00:01,  9.78it/s]
+
 87%|████████▋ | 67/77 [00:06<00:01,  9.78it/s]
+
 88%|████████▊ | 68/77 [00:06<00:00,  9.78it/s]
+
 90%|████████▉ | 69/77 [00:06<00:00,  9.78it/s]
+
 91%|█████████ | 70/77 [00:07<00:00,  9.78it/s]
+
 92%|█████████▏| 71/77 [00:07<00:00,  9.77it/s]
+
 94%|█████████▎| 72/77 [00:07<00:00,  9.77it/s]
+
 95%|█████████▍| 73/77 [00:07<00:00,  9.77it/s]
+
 96%|█████████▌| 74/77 [00:07<00:00,  9.77it/s]
+
 97%|█████████▋| 75/77 [00:07<00:00,  9.77it/s]
+
 99%|█████████▊| 76/77 [00:07<00:00,  9.78it/s]
+
100%|██████████| 77/77 [00:07<00:00,  9.78it/s]
                                                 
+
                                               
{'eval_loss': '4.448', 'eval_runtime': '7.919', 'eval_samples_per_second': '308.2', 'eval_steps_per_second': '9.723', 'epoch': '11'}
+
 92%|█████████▏| 792/864 [35:02<02:25,  2.03s/it]
+
100%|██████████| 77/77 [00:07<00:00,  9.78it/s][eval] epoch 11.0 step 792 loss 4.4479 (best 3.9540 @ ep 10.0, 1/10000 since best)
+
+
                                               
 92%|█████████▏| 793/864 [35:04<05:19,  4.50s/it]
 92%|█████████▏| 794/864 [35:07<04:29,  3.85s/it]
 92%|█████████▏| 795/864 [35:09<03:53,  3.39s/it]
 92%|█████████▏| 796/864 [35:11<03:28,  3.07s/it]
 92%|█████████▏| 797/864 [35:14<03:10,  2.85s/it]
 92%|█████████▏| 798/864 [35:16<02:57,  2.69s/it]
 92%|█████████▏| 799/864 [35:18<02:47,  2.58s/it]
 93%|█████████▎| 800/864 [35:21<02:40,  2.50s/it]
                                                 
{'loss': '4.035', 'grad_norm': '0.2781', 'learning_rate': '0.0012', 'epoch': '11.11'}
+
 93%|█████████▎| 800/864 [35:21<02:40,  2.50s/it]
 93%|█████████▎| 801/864 [35:23<02:34,  2.45s/it]
 93%|█████████▎| 802/864 [35:25<02:29,  2.41s/it]
 93%|█████████▎| 803/864 [35:28<02:25,  2.39s/it]
 93%|█████████▎| 804/864 [35:30<02:22,  2.37s/it]
 93%|█████████▎| 805/864 [35:32<02:18,  2.35s/it]
 93%|█████████▎| 806/864 [35:35<02:16,  2.35s/it]
 93%|█████████▎| 807/864 [35:37<02:13,  2.34s/it]
 94%|█████████▎| 808/864 [35:39<02:10,  2.33s/it]
 94%|█████████▎| 809/864 [35:42<02:08,  2.33s/it]
 94%|█████████▍| 810/864 [35:44<02:05,  2.33s/it]
                                                 
{'loss': '3.844', 'grad_norm': '0.1566', 'learning_rate': '0.0012', 'epoch': '11.25'}
+
 94%|█████████▍| 810/864 [35:44<02:05,  2.33s/it]
 94%|█████████▍| 811/864 [35:46<02:03,  2.33s/it]
 94%|█████████▍| 812/864 [35:49<02:00,  2.33s/it]
 94%|█████████▍| 813/864 [35:51<01:58,  2.33s/it]
 94%|█████████▍| 814/864 [35:53<01:56,  2.32s/it]
 94%|█████████▍| 815/864 [35:56<01:53,  2.32s/it]
 94%|█████████▍| 816/864 [35:58<01:51,  2.33s/it]
 95%|█████████▍| 817/864 [36:00<01:49,  2.32s/it]
 95%|█████████▍| 818/864 [36:02<01:46,  2.33s/it]
 95%|█████████▍| 819/864 [36:05<01:44,  2.32s/it]
 95%|█████████▍| 820/864 [36:07<01:42,  2.32s/it]
                                                 
{'loss': '3.74', 'grad_norm': '0.1661', 'learning_rate': '0.0012', 'epoch': '11.39'}
+
 95%|█████████▍| 820/864 [36:07<01:42,  2.32s/it]
 95%|█████████▌| 821/864 [36:09<01:39,  2.32s/it]
 95%|█████████▌| 822/864 [36:12<01:37,  2.32s/it]
 95%|█████████▌| 823/864 [36:14<01:35,  2.32s/it]
 95%|█████████▌| 824/864 [36:16<01:32,  2.32s/it]
 95%|█████████▌| 825/864 [36:19<01:30,  2.32s/it]
 96%|█████████▌| 826/864 [36:21<01:28,  2.32s/it]
 96%|█████████▌| 827/864 [36:23<01:25,  2.32s/it]
 96%|█████████▌| 828/864 [36:26<01:23,  2.32s/it]
 96%|█████████▌| 829/864 [36:28<01:21,  2.32s/it]
 96%|█████████▌| 830/864 [36:30<01:19,  2.32s/it]
                                                 
{'loss': '3.667', 'grad_norm': '0.1663', 'learning_rate': '0.0012', 'epoch': '11.53'}
+
 96%|█████████▌| 830/864 [36:30<01:19,  2.32s/it]
 96%|█████████▌| 831/864 [36:33<01:16,  2.32s/it]
 96%|█████████▋| 832/864 [36:35<01:14,  2.32s/it]
 96%|█████████▋| 833/864 [36:37<01:12,  2.32s/it]
 97%|█████████▋| 834/864 [36:40<01:09,  2.32s/it]
 97%|█████████▋| 835/864 [36:42<01:07,  2.32s/it]
 97%|█████████▋| 836/864 [36:44<01:05,  2.32s/it]
 97%|█████████▋| 837/864 [36:47<01:02,  2.32s/it]
 97%|█████████▋| 838/864 [36:49<01:00,  2.32s/it]
 97%|█████████▋| 839/864 [36:51<00:58,  2.32s/it]
 97%|█████████▋| 840/864 [36:54<00:55,  2.32s/it]
                                                 
{'loss': '3.605', 'grad_norm': '0.1635', 'learning_rate': '0.0012', 'epoch': '11.67'}
+
 97%|█████████▋| 840/864 [36:54<00:55,  2.32s/it]
 97%|█████████▋| 841/864 [36:56<00:53,  2.32s/it]
 97%|█████████▋| 842/864 [36:58<00:51,  2.32s/it]
 98%|█████████▊| 843/864 [37:01<00:48,  2.32s/it]
 98%|█████████▊| 844/864 [37:03<00:46,  2.32s/it]
 98%|█████████▊| 845/864 [37:05<00:44,  2.32s/it]
 98%|█████████▊| 846/864 [37:08<00:41,  2.32s/it]
 98%|█████████▊| 847/864 [37:10<00:41,  2.43s/it]
 98%|█████████▊| 848/864 [37:13<00:38,  2.40s/it]
 98%|█████████▊| 849/864 [37:15<00:35,  2.37s/it]
 98%|█████████▊| 850/864 [37:17<00:33,  2.36s/it]
                                                 
{'loss': '3.561', 'grad_norm': '0.1445', 'learning_rate': '0.0012', 'epoch': '11.81'}
+
 98%|█████████▊| 850/864 [37:17<00:33,  2.36s/it]
 98%|█████████▊| 851/864 [37:20<00:30,  2.35s/it]
 99%|█████████▊| 852/864 [37:22<00:28,  2.34s/it]
 99%|█████████▊| 853/864 [37:24<00:25,  2.34s/it]
 99%|█████████▉| 854/864 [37:26<00:23,  2.33s/it]
 99%|█████████▉| 855/864 [37:29<00:20,  2.33s/it]
 99%|█████████▉| 856/864 [37:31<00:18,  2.33s/it]
 99%|█████████▉| 857/864 [37:33<00:16,  2.33s/it]
 99%|█████████▉| 858/864 [37:36<00:13,  2.33s/it]
 99%|█████████▉| 859/864 [37:38<00:11,  2.33s/it]
100%|█████████▉| 860/864 [37:40<00:09,  2.33s/it]
                                                 
{'loss': '3.546', 'grad_norm': '0.1584', 'learning_rate': '0.0012', 'epoch': '11.95'}
+
100%|█████████▉| 860/864 [37:40<00:09,  2.33s/it]
100%|█████████▉| 861/864 [37:43<00:06,  2.33s/it]
100%|█████████▉| 862/864 [37:45<00:04,  2.33s/it]
100%|█████████▉| 863/864 [37:47<00:02,  2.33s/it]
100%|██████████| 864/864 [37:49<00:00,  2.15s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 4
+
+
  0%|          | 0/77 [00:00<?, ?it/s]
+
  4%|▍         | 3/77 [00:00<00:04, 15.01it/s]
+
  6%|▋         | 5/77 [00:00<00:06, 11.86it/s]
+
  9%|▉         | 7/77 [00:00<00:06, 10.90it/s]
+
 12%|█▏        | 9/77 [00:00<00:06, 10.46it/s]
+
 14%|█▍        | 11/77 [00:01<00:06, 10.20it/s]
+
 17%|█▋        | 13/77 [00:01<00:06, 10.06it/s]
+
 19%|█▉        | 15/77 [00:01<00:06,  9.96it/s]
+
 21%|██        | 16/77 [00:01<00:06,  9.93it/s]
+
 22%|██▏       | 17/77 [00:01<00:06,  9.90it/s]
+
 23%|██▎       | 18/77 [00:01<00:05,  9.86it/s]
+
 25%|██▍       | 19/77 [00:01<00:05,  9.84it/s]
+
 26%|██▌       | 20/77 [00:01<00:05,  9.83it/s]
+
 27%|██▋       | 21/77 [00:02<00:05,  9.81it/s]
+
 29%|██▊       | 22/77 [00:02<00:05,  9.80it/s]
+
 30%|██▉       | 23/77 [00:02<00:05,  9.78it/s]
+
 31%|███       | 24/77 [00:02<00:05,  9.78it/s]
+
 32%|███▏      | 25/77 [00:02<00:05,  9.79it/s]
+
 34%|███▍      | 26/77 [00:02<00:05,  9.77it/s]
+
 35%|███▌      | 27/77 [00:02<00:05,  9.76it/s]
+
 36%|███▋      | 28/77 [00:02<00:05,  9.78it/s]
+
 38%|███▊      | 29/77 [00:02<00:04,  9.78it/s]
+
 39%|███▉      | 30/77 [00:02<00:04,  9.78it/s]
+
 40%|████      | 31/77 [00:03<00:04,  9.76it/s]
+
 42%|████▏     | 32/77 [00:03<00:04,  9.78it/s]
+
 43%|████▎     | 33/77 [00:03<00:04,  9.78it/s]
+
 44%|████▍     | 34/77 [00:03<00:04,  9.77it/s]
+
 45%|████▌     | 35/77 [00:03<00:04,  9.76it/s]
+
 47%|████▋     | 36/77 [00:03<00:04,  9.75it/s]
+
 48%|████▊     | 37/77 [00:03<00:04,  9.78it/s]
+
 49%|████▉     | 38/77 [00:03<00:03,  9.78it/s]
+
 51%|█████     | 39/77 [00:03<00:03,  9.76it/s]
+
 52%|█████▏    | 40/77 [00:03<00:03,  9.77it/s]
+
 53%|█████▎    | 41/77 [00:04<00:03,  9.77it/s]
+
 55%|█████▍    | 42/77 [00:04<00:03,  9.76it/s]
+
 56%|█████▌    | 43/77 [00:04<00:03,  9.75it/s]
+
 57%|█████▋    | 44/77 [00:04<00:03,  9.77it/s]
+
 58%|█████▊    | 45/77 [00:04<00:03,  9.77it/s]
+
 60%|█████▉    | 46/77 [00:04<00:03,  9.77it/s]
+
 61%|██████    | 47/77 [00:04<00:03,  9.76it/s]
+
 62%|██████▏   | 48/77 [00:04<00:02,  9.77it/s]
+
 64%|██████▎   | 49/77 [00:04<00:02,  9.78it/s]
+
 65%|██████▍   | 50/77 [00:05<00:02,  9.77it/s]
+
 66%|██████▌   | 51/77 [00:05<00:02,  9.77it/s]
+
 68%|██████▊   | 52/77 [00:05<00:02,  9.77it/s]
+
 69%|██████▉   | 53/77 [00:05<00:02,  9.77it/s]
+
 70%|███████   | 54/77 [00:05<00:02,  9.77it/s]
+
 71%|███████▏  | 55/77 [00:05<00:02,  9.76it/s]
+
 73%|███████▎  | 56/77 [00:05<00:02,  9.77it/s]
+
 74%|███████▍  | 57/77 [00:05<00:02,  9.77it/s]
+
 75%|███████▌  | 58/77 [00:05<00:01,  9.77it/s]
+
 77%|███████▋  | 59/77 [00:05<00:01,  9.76it/s]
+
 78%|███████▊  | 60/77 [00:06<00:01,  9.78it/s]
+
 79%|███████▉  | 61/77 [00:06<00:01,  9.79it/s]
+
 81%|████████  | 62/77 [00:06<00:01,  9.79it/s]
+
 82%|████████▏ | 63/77 [00:06<00:01,  9.78it/s]
+
 83%|████████▎ | 64/77 [00:06<00:01,  9.79it/s]
+
 84%|████████▍ | 65/77 [00:06<00:01,  9.79it/s]
+
 86%|████████▌ | 66/77 [00:06<00:01,  9.79it/s]
+
 87%|████████▋ | 67/77 [00:06<00:01,  9.78it/s]
+
 88%|████████▊ | 68/77 [00:06<00:00,  9.78it/s]
+
 90%|████████▉ | 69/77 [00:06<00:00,  9.79it/s]
+
 91%|█████████ | 70/77 [00:07<00:00,  9.78it/s]
+
 92%|█████████▏| 71/77 [00:07<00:00,  9.78it/s]
+
 94%|█████████▎| 72/77 [00:07<00:00,  9.79it/s]
+
 95%|█████████▍| 73/77 [00:07<00:00,  9.78it/s]
+
 96%|█████████▌| 74/77 [00:07<00:00,  9.79it/s]
+
 97%|█████████▋| 75/77 [00:07<00:00,  9.78it/s]
+
 99%|█████████▊| 76/77 [00:07<00:00,  9.79it/s]
+
100%|██████████| 77/77 [00:07<00:00,  9.79it/s]
                                                 
+
                                               
{'eval_loss': '4.018', 'eval_runtime': '7.92', 'eval_samples_per_second': '308.2', 'eval_steps_per_second': '9.722', 'epoch': '12'}
+
100%|██████████| 864/864 [37:57<00:00,  2.15s/it]
+
100%|██████████| 77/77 [00:07<00:00,  9.79it/s][eval] epoch 12.0 step 864 loss 4.0176 (best 3.9540 @ ep 10.0, 2/10000 since best)
+
+
                                               [transformers] 
+
+Training completed. Do not forget to share your model on huggingface.co/models =)
+
+
+
                                                 
{'train_runtime': '2279', 'train_samples_per_second': '192.8', 'train_steps_per_second': '0.379', 'train_loss': '4.759', 'epoch': '12'}
+
100%|██████████| 864/864 [37:57<00:00,  2.15s/it]
100%|██████████| 864/864 [37:57<00:00,  2.64s/it]
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+wandb: 
+wandb: 🚀 View run t0_S5_lr0.0012 at: https://wandb.ai/slowrun-wankers/data-aug-pretraining/runs/qkbq14t6
+wandb: Find logs at: wandb/run-20260817_225224-qkbq14t6/logs
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+[rank0]:[W817 23:31:06.001347933 ProcessGroupNCCL.cpp:1575] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator())
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
diff --git a/runs/t2a/t2a_150M_C0_baseline/checkpoint-2232/chat_template.jinja b/runs/t2a/t2a_150M_C0_baseline/checkpoint-2232/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..699ff8df401fe4788525e9c1f9b86a99eadd6230
--- /dev/null
+++ b/runs/t2a/t2a_150M_C0_baseline/checkpoint-2232/chat_template.jinja
@@ -0,0 +1,85 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0].role == 'system' %}
+        {{- messages[0].content + '\n\n' }}
+    {%- endif %}
+    {{- "# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0].content + '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
+{%- for message in messages[::-1] %}
+    {%- set index = (messages|length - 1) - loop.index0 %}
+    {%- if ns.multi_step_tool and message.role == "user" and not(message.content.startswith('<tool_response>') and message.content.endswith('</tool_response>')) %}
+        {%- set ns.multi_step_tool = false %}
+        {%- set ns.last_query_index = index %}
+    {%- endif %}
+{%- endfor %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {%- set content = message.content %}
+        {%- set reasoning_content = '' %}
+        {%- if message.reasoning_content is defined and message.reasoning_content is not none %}
+            {%- set reasoning_content = message.reasoning_content %}
+        {%- else %}
+            {%- if '</think>' in message.content %}
+                {%- set content = message.content.split('</think>')[-1].lstrip('\n') %}
+                {%- set reasoning_content = message.content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
+            {%- endif %}
+        {%- endif %}
+        {%- if loop.index0 > ns.last_query_index %}
+            {%- if loop.last or (not loop.last and reasoning_content) %}
+                {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content.strip('\n') + '\n</think>\n\n' + content.lstrip('\n') }}
+            {%- else %}
+                {{- '<|im_start|>' + message.role + '\n' + content }}
+            {%- endif %}
+        {%- else %}
+            {{- '<|im_start|>' + message.role + '\n' + content }}
+        {%- endif %}
+        {%- if message.tool_calls %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if (loop.first and content) or (not loop.first) %}
+                    {{- '\n' }}
+                {%- endif %}
+                {%- if tool_call.function %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {{- '<tool_call>\n{"name": "' }}
+                {{- tool_call.name }}
+                {{- '", "arguments": ' }}
+                {%- if tool_call.arguments is string %}
+                    {{- tool_call.arguments }}
+                {%- else %}
+                    {{- tool_call.arguments | tojson }}
+                {%- endif %}
+                {{- '}\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.first or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+    {%- if enable_thinking is defined and enable_thinking is false %}
+        {{- '<think>\n\n</think>\n\n' }}
+    {%- endif %}
+{%- endif %}
\ No newline at end of file
diff --git a/runs/t2a/t2a_150M_C0_baseline/checkpoint-2232/config.json b/runs/t2a/t2a_150M_C0_baseline/checkpoint-2232/config.json
new file mode 100644
index 0000000000000000000000000000000000000000..f8ca584a0751672f280da7c54ff3916cdf24d6f4
--- /dev/null
+++ b/runs/t2a/t2a_150M_C0_baseline/checkpoint-2232/config.json
@@ -0,0 +1,32 @@
+{
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": null,
+  "dtype": "float32",
+  "eos_token_id": 151645,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 512,
+  "initializer_range": 0.02,
+  "intermediate_size": 1536,
+  "max_position_embeddings": 2048,
+  "mlp_bias": false,
+  "model_type": "llama",
+  "num_attention_heads": 4,
+  "num_hidden_layers": 20,
+  "num_key_value_heads": 4,
+  "pad_token_id": 151645,
+  "pretraining_tp": 1,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 10000.0,
+    "rope_type": "default"
+  },
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": false,
+  "vocab_size": 151671
+}
diff --git a/runs/t2a/t2a_150M_C0_baseline/checkpoint-2232/generation_config.json b/runs/t2a/t2a_150M_C0_baseline/checkpoint-2232/generation_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..8c4242f32d40df256857f87264f646130f684e33
--- /dev/null
+++ b/runs/t2a/t2a_150M_C0_baseline/checkpoint-2232/generation_config.json
@@ -0,0 +1,11 @@
+{
+  "_from_model_config": true,
+  "eos_token_id": [
+    151645
+  ],
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "transformers_version": "5.15.0",
+  "use_cache": true
+}
diff --git a/runs/t2a/t2a_150M_C0_baseline/checkpoint-2232/tokenizer_config.json b/runs/t2a/t2a_150M_C0_baseline/checkpoint-2232/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..839a3743177a3ed77e1ffbae67520e8500ca0532
--- /dev/null
+++ b/runs/t2a/t2a_150M_C0_baseline/checkpoint-2232/tokenizer_config.json
@@ -0,0 +1,19 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|l2r_pred|>",
+    "<|r2l_pred|>"
+  ],
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 131072,
+  "pad_token": "<|im_end|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}
diff --git a/runs/t2a/t2a_150M_C0_baseline/checkpoint-2232/trainer_state.json b/runs/t2a/t2a_150M_C0_baseline/checkpoint-2232/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..e37579147751e7b1a585d440248f9386ac314f05
--- /dev/null
+++ b/runs/t2a/t2a_150M_C0_baseline/checkpoint-2232/trainer_state.json
@@ -0,0 +1,1850 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 15.609457092819614,
+  "eval_steps": 72,
+  "global_step": 2232,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0070052539404553416,
+      "grad_norm": 1.7456934452056885,
+      "learning_rate": 0.0,
+      "loss": 12.037310600280762,
+      "step": 1
+    },
+    {
+      "epoch": 0.07005253940455342,
+      "grad_norm": 1.7848070859909058,
+      "learning_rate": 5.399999999999999e-05,
+      "loss": 11.75309583875868,
+      "step": 10
+    },
+    {
+      "epoch": 0.14010507880910683,
+      "grad_norm": 1.6792213916778564,
+      "learning_rate": 0.00011399999999999999,
+      "loss": 11.042081451416015,
+      "step": 20
+    },
+    {
+      "epoch": 0.21015761821366025,
+      "grad_norm": 1.6428947448730469,
+      "learning_rate": 0.00017399999999999997,
+      "loss": 10.273934173583985,
+      "step": 30
+    },
+    {
+      "epoch": 0.28021015761821366,
+      "grad_norm": 1.5020601749420166,
+      "learning_rate": 0.000234,
+      "loss": 9.222610473632812,
+      "step": 40
+    },
+    {
+      "epoch": 0.3502626970227671,
+      "grad_norm": 0.9085643887519836,
+      "learning_rate": 0.000294,
+      "loss": 8.245561218261718,
+      "step": 50
+    },
+    {
+      "epoch": 0.4203152364273205,
+      "grad_norm": 0.4325217604637146,
+      "learning_rate": 0.00035399999999999993,
+      "loss": 7.676107788085938,
+      "step": 60
+    },
+    {
+      "epoch": 0.4903677758318739,
+      "grad_norm": 0.4937305450439453,
+      "learning_rate": 0.0004139999999999999,
+      "loss": 7.524372100830078,
+      "step": 70
+    },
+    {
+      "epoch": 0.5043782837127846,
+      "eval_loss": 7.435024261474609,
+      "eval_runtime": 27.6967,
+      "eval_samples_per_second": 88.133,
+      "eval_steps_per_second": 1.408,
+      "step": 72
+    },
+    {
+      "epoch": 0.5604203152364273,
+      "grad_norm": 0.42533794045448303,
+      "learning_rate": 0.000474,
+      "loss": 7.372554016113281,
+      "step": 80
+    },
+    {
+      "epoch": 0.6304728546409807,
+      "grad_norm": 0.8951543569564819,
+      "learning_rate": 0.000534,
+      "loss": 7.21534423828125,
+      "step": 90
+    },
+    {
+      "epoch": 0.7005253940455342,
+      "grad_norm": 0.5405721664428711,
+      "learning_rate": 0.0005939999999999999,
+      "loss": 7.055476379394531,
+      "step": 100
+    },
+    {
+      "epoch": 0.7705779334500875,
+      "grad_norm": 0.4531368315219879,
+      "learning_rate": 0.0006,
+      "loss": 6.896892547607422,
+      "step": 110
+    },
+    {
+      "epoch": 0.840630472854641,
+      "grad_norm": 0.3535282611846924,
+      "learning_rate": 0.0006,
+      "loss": 6.763951110839844,
+      "step": 120
+    },
+    {
+      "epoch": 0.9106830122591943,
+      "grad_norm": 0.772462785243988,
+      "learning_rate": 0.0006,
+      "loss": 6.64959487915039,
+      "step": 130
+    },
+    {
+      "epoch": 0.9807355516637478,
+      "grad_norm": 0.6359318494796753,
+      "learning_rate": 0.0006,
+      "loss": 6.573255157470703,
+      "step": 140
+    },
+    {
+      "epoch": 1.0070052539404553,
+      "eval_loss": 6.504467487335205,
+      "eval_runtime": 3.2635,
+      "eval_samples_per_second": 747.973,
+      "eval_steps_per_second": 11.95,
+      "step": 144
+    },
+    {
+      "epoch": 1.0490367775831875,
+      "grad_norm": 0.6233965754508972,
+      "learning_rate": 0.0006,
+      "loss": 6.474125671386719,
+      "step": 150
+    },
+    {
+      "epoch": 1.1190893169877407,
+      "grad_norm": 1.1296820640563965,
+      "learning_rate": 0.0006,
+      "loss": 6.41492919921875,
+      "step": 160
+    },
+    {
+      "epoch": 1.1891418563922942,
+      "grad_norm": 0.3820130527019501,
+      "learning_rate": 0.0006,
+      "loss": 6.331824493408203,
+      "step": 170
+    },
+    {
+      "epoch": 1.2591943957968477,
+      "grad_norm": 1.0479296445846558,
+      "learning_rate": 0.0006,
+      "loss": 6.272692489624023,
+      "step": 180
+    },
+    {
+      "epoch": 1.3292469352014011,
+      "grad_norm": 0.6026280522346497,
+      "learning_rate": 0.0006,
+      "loss": 6.26782455444336,
+      "step": 190
+    },
+    {
+      "epoch": 1.3992994746059544,
+      "grad_norm": 1.3044335842132568,
+      "learning_rate": 0.0006,
+      "loss": 6.164206314086914,
+      "step": 200
+    },
+    {
+      "epoch": 1.4693520140105079,
+      "grad_norm": 0.414527952671051,
+      "learning_rate": 0.0006,
+      "loss": 6.096501159667969,
+      "step": 210
+    },
+    {
+      "epoch": 1.51138353765324,
+      "eval_loss": 6.077937126159668,
+      "eval_runtime": 3.2373,
+      "eval_samples_per_second": 754.014,
+      "eval_steps_per_second": 12.047,
+      "step": 216
+    },
+    {
+      "epoch": 1.5394045534150613,
+      "grad_norm": 1.2787466049194336,
+      "learning_rate": 0.0006,
+      "loss": 6.050628662109375,
+      "step": 220
+    },
+    {
+      "epoch": 1.6094570928196146,
+      "grad_norm": 0.6056044101715088,
+      "learning_rate": 0.0006,
+      "loss": 5.993916702270508,
+      "step": 230
+    },
+    {
+      "epoch": 1.679509632224168,
+      "grad_norm": 0.5148183107376099,
+      "learning_rate": 0.0006,
+      "loss": 5.94757080078125,
+      "step": 240
+    },
+    {
+      "epoch": 1.7495621716287215,
+      "grad_norm": 0.31190764904022217,
+      "learning_rate": 0.0006,
+      "loss": 5.883531188964843,
+      "step": 250
+    },
+    {
+      "epoch": 1.819614711033275,
+      "grad_norm": 0.36509600281715393,
+      "learning_rate": 0.0006,
+      "loss": 5.829424667358398,
+      "step": 260
+    },
+    {
+      "epoch": 1.8896672504378285,
+      "grad_norm": 1.0162628889083862,
+      "learning_rate": 0.0006,
+      "loss": 5.784934997558594,
+      "step": 270
+    },
+    {
+      "epoch": 1.959719789842382,
+      "grad_norm": 0.39998742938041687,
+      "learning_rate": 0.0006,
+      "loss": 5.737149810791015,
+      "step": 280
+    },
+    {
+      "epoch": 2.0140105078809105,
+      "eval_loss": 5.6931023597717285,
+      "eval_runtime": 3.7798,
+      "eval_samples_per_second": 645.803,
+      "eval_steps_per_second": 10.318,
+      "step": 288
+    },
+    {
+      "epoch": 2.0280210157618215,
+      "grad_norm": 0.4628411531448364,
+      "learning_rate": 0.0006,
+      "loss": 5.674631500244141,
+      "step": 290
+    },
+    {
+      "epoch": 2.098073555166375,
+      "grad_norm": 0.48857519030570984,
+      "learning_rate": 0.0006,
+      "loss": 5.590349197387695,
+      "step": 300
+    },
+    {
+      "epoch": 2.168126094570928,
+      "grad_norm": 0.55524742603302,
+      "learning_rate": 0.0006,
+      "loss": 5.591045761108399,
+      "step": 310
+    },
+    {
+      "epoch": 2.2381786339754814,
+      "grad_norm": 0.3205523192882538,
+      "learning_rate": 0.0006,
+      "loss": 5.515919494628906,
+      "step": 320
+    },
+    {
+      "epoch": 2.308231173380035,
+      "grad_norm": 0.6817094087600708,
+      "learning_rate": 0.0006,
+      "loss": 5.462646865844727,
+      "step": 330
+    },
+    {
+      "epoch": 2.3782837127845884,
+      "grad_norm": 0.782426655292511,
+      "learning_rate": 0.0006,
+      "loss": 5.457918167114258,
+      "step": 340
+    },
+    {
+      "epoch": 2.448336252189142,
+      "grad_norm": 0.30008888244628906,
+      "learning_rate": 0.0006,
+      "loss": 5.380028533935547,
+      "step": 350
+    },
+    {
+      "epoch": 2.5183887915936953,
+      "grad_norm": 0.49766749143600464,
+      "learning_rate": 0.0006,
+      "loss": 5.30742301940918,
+      "step": 360
+    },
+    {
+      "epoch": 2.5183887915936953,
+      "eval_loss": 5.35524320602417,
+      "eval_runtime": 3.985,
+      "eval_samples_per_second": 612.553,
+      "eval_steps_per_second": 9.787,
+      "step": 360
+    },
+    {
+      "epoch": 2.588441330998249,
+      "grad_norm": 1.1236608028411865,
+      "learning_rate": 0.0006,
+      "loss": 5.256686401367188,
+      "step": 370
+    },
+    {
+      "epoch": 2.6584938704028023,
+      "grad_norm": 0.9686716794967651,
+      "learning_rate": 0.0006,
+      "loss": 5.27575569152832,
+      "step": 380
+    },
+    {
+      "epoch": 2.7285464098073557,
+      "grad_norm": 0.29338839650154114,
+      "learning_rate": 0.0006,
+      "loss": 5.202019119262696,
+      "step": 390
+    },
+    {
+      "epoch": 2.7985989492119088,
+      "grad_norm": 0.34476038813591003,
+      "learning_rate": 0.0006,
+      "loss": 5.1141807556152346,
+      "step": 400
+    },
+    {
+      "epoch": 2.8686514886164622,
+      "grad_norm": 0.8026102185249329,
+      "learning_rate": 0.0006,
+      "loss": 5.106670379638672,
+      "step": 410
+    },
+    {
+      "epoch": 2.9387040280210157,
+      "grad_norm": 0.41981688141822815,
+      "learning_rate": 0.0006,
+      "loss": 5.072551727294922,
+      "step": 420
+    },
+    {
+      "epoch": 3.0070052539404553,
+      "grad_norm": 0.2799515426158905,
+      "learning_rate": 0.0006,
+      "loss": 4.986613464355469,
+      "step": 430
+    },
+    {
+      "epoch": 3.021015761821366,
+      "eval_loss": 5.031318664550781,
+      "eval_runtime": 3.2607,
+      "eval_samples_per_second": 748.624,
+      "eval_steps_per_second": 11.961,
+      "step": 432
+    },
+    {
+      "epoch": 3.0770577933450087,
+      "grad_norm": 0.7299258708953857,
+      "learning_rate": 0.0006,
+      "loss": 4.936912536621094,
+      "step": 440
+    },
+    {
+      "epoch": 3.147110332749562,
+      "grad_norm": 0.7612623572349548,
+      "learning_rate": 0.0006,
+      "loss": 4.938148117065429,
+      "step": 450
+    },
+    {
+      "epoch": 3.2171628721541157,
+      "grad_norm": 0.30043086409568787,
+      "learning_rate": 0.0006,
+      "loss": 4.860009002685547,
+      "step": 460
+    },
+    {
+      "epoch": 3.287215411558669,
+      "grad_norm": 0.34863752126693726,
+      "learning_rate": 0.0006,
+      "loss": 4.803009796142578,
+      "step": 470
+    },
+    {
+      "epoch": 3.357267950963222,
+      "grad_norm": 0.6028016805648804,
+      "learning_rate": 0.0006,
+      "loss": 4.771705245971679,
+      "step": 480
+    },
+    {
+      "epoch": 3.4273204903677756,
+      "grad_norm": 0.38058093190193176,
+      "learning_rate": 0.0006,
+      "loss": 4.7452739715576175,
+      "step": 490
+    },
+    {
+      "epoch": 3.497373029772329,
+      "grad_norm": 0.6069619059562683,
+      "learning_rate": 0.0006,
+      "loss": 4.701860809326172,
+      "step": 500
+    },
+    {
+      "epoch": 3.5253940455341506,
+      "eval_loss": 4.769557476043701,
+      "eval_runtime": 4.0934,
+      "eval_samples_per_second": 596.332,
+      "eval_steps_per_second": 9.528,
+      "step": 504
+    },
+    {
+      "epoch": 3.5674255691768826,
+      "grad_norm": 0.3816678524017334,
+      "learning_rate": 0.0006,
+      "loss": 4.671141815185547,
+      "step": 510
+    },
+    {
+      "epoch": 3.637478108581436,
+      "grad_norm": 0.34364593029022217,
+      "learning_rate": 0.0006,
+      "loss": 4.621066665649414,
+      "step": 520
+    },
+    {
+      "epoch": 3.7075306479859895,
+      "grad_norm": 0.7322856187820435,
+      "learning_rate": 0.0006,
+      "loss": 4.612473678588867,
+      "step": 530
+    },
+    {
+      "epoch": 3.777583187390543,
+      "grad_norm": 0.42280662059783936,
+      "learning_rate": 0.0006,
+      "loss": 4.61462173461914,
+      "step": 540
+    },
+    {
+      "epoch": 3.8476357267950965,
+      "grad_norm": 0.2599335312843323,
+      "learning_rate": 0.0006,
+      "loss": 4.5415092468261715,
+      "step": 550
+    },
+    {
+      "epoch": 3.91768826619965,
+      "grad_norm": 0.403529554605484,
+      "learning_rate": 0.0006,
+      "loss": 4.4888755798339846,
+      "step": 560
+    },
+    {
+      "epoch": 3.9877408056042034,
+      "grad_norm": 0.2667478621006012,
+      "learning_rate": 0.0006,
+      "loss": 4.468472290039062,
+      "step": 570
+    },
+    {
+      "epoch": 4.028021015761821,
+      "eval_loss": 4.5557074546813965,
+      "eval_runtime": 3.4966,
+      "eval_samples_per_second": 698.116,
+      "eval_steps_per_second": 11.154,
+      "step": 576
+    },
+    {
+      "epoch": 4.056042031523643,
+      "grad_norm": 0.4902321994304657,
+      "learning_rate": 0.0006,
+      "loss": 4.418440628051758,
+      "step": 580
+    },
+    {
+      "epoch": 4.126094570928196,
+      "grad_norm": 0.6080674529075623,
+      "learning_rate": 0.0006,
+      "loss": 4.395100402832031,
+      "step": 590
+    },
+    {
+      "epoch": 4.19614711033275,
+      "grad_norm": 0.3149462938308716,
+      "learning_rate": 0.0006,
+      "loss": 4.390229034423828,
+      "step": 600
+    },
+    {
+      "epoch": 4.266199649737303,
+      "grad_norm": 0.3042010962963104,
+      "learning_rate": 0.0006,
+      "loss": 4.345519638061523,
+      "step": 610
+    },
+    {
+      "epoch": 4.336252189141856,
+      "grad_norm": 0.24239438772201538,
+      "learning_rate": 0.0006,
+      "loss": 4.321981811523438,
+      "step": 620
+    },
+    {
+      "epoch": 4.406304728546409,
+      "grad_norm": 0.38201919198036194,
+      "learning_rate": 0.0006,
+      "loss": 4.2978263854980465,
+      "step": 630
+    },
+    {
+      "epoch": 4.476357267950963,
+      "grad_norm": 0.3829366862773895,
+      "learning_rate": 0.0006,
+      "loss": 4.280644226074219,
+      "step": 640
+    },
+    {
+      "epoch": 4.532399299474606,
+      "eval_loss": 4.381662845611572,
+      "eval_runtime": 3.691,
+      "eval_samples_per_second": 661.341,
+      "eval_steps_per_second": 10.566,
+      "step": 648
+    },
+    {
+      "epoch": 4.546409807355516,
+      "grad_norm": 0.32211166620254517,
+      "learning_rate": 0.0006,
+      "loss": 4.259928131103516,
+      "step": 650
+    },
+    {
+      "epoch": 4.61646234676007,
+      "grad_norm": 0.4348321259021759,
+      "learning_rate": 0.0006,
+      "loss": 4.251454925537109,
+      "step": 660
+    },
+    {
+      "epoch": 4.686514886164623,
+      "grad_norm": 0.3812483549118042,
+      "learning_rate": 0.0006,
+      "loss": 4.236824035644531,
+      "step": 670
+    },
+    {
+      "epoch": 4.756567425569177,
+      "grad_norm": 0.29503557085990906,
+      "learning_rate": 0.0006,
+      "loss": 4.216230010986328,
+      "step": 680
+    },
+    {
+      "epoch": 4.82661996497373,
+      "grad_norm": 0.2546517848968506,
+      "learning_rate": 0.0006,
+      "loss": 4.186763000488281,
+      "step": 690
+    },
+    {
+      "epoch": 4.896672504378284,
+      "grad_norm": 0.3478173613548279,
+      "learning_rate": 0.0006,
+      "loss": 4.161574172973633,
+      "step": 700
+    },
+    {
+      "epoch": 4.966725043782837,
+      "grad_norm": 0.21839696168899536,
+      "learning_rate": 0.0006,
+      "loss": 4.142048645019531,
+      "step": 710
+    },
+    {
+      "epoch": 5.035026269702277,
+      "grad_norm": 0.5530551671981812,
+      "learning_rate": 0.0006,
+      "loss": 4.121431732177735,
+      "step": 720
+    },
+    {
+      "epoch": 5.035026269702277,
+      "eval_loss": 4.278331756591797,
+      "eval_runtime": 3.908,
+      "eval_samples_per_second": 624.624,
+      "eval_steps_per_second": 9.98,
+      "step": 720
+    },
+    {
+      "epoch": 5.10507880910683,
+      "grad_norm": 0.4000762701034546,
+      "learning_rate": 0.0006,
+      "loss": 4.094532012939453,
+      "step": 730
+    },
+    {
+      "epoch": 5.175131348511384,
+      "grad_norm": 0.30992069840431213,
+      "learning_rate": 0.0006,
+      "loss": 4.089750671386719,
+      "step": 740
+    },
+    {
+      "epoch": 5.245183887915937,
+      "grad_norm": 0.23498931527137756,
+      "learning_rate": 0.0006,
+      "loss": 4.051603698730469,
+      "step": 750
+    },
+    {
+      "epoch": 5.315236427320491,
+      "grad_norm": 0.24305176734924316,
+      "learning_rate": 0.0006,
+      "loss": 4.043153381347656,
+      "step": 760
+    },
+    {
+      "epoch": 5.385288966725044,
+      "grad_norm": 0.24310357868671417,
+      "learning_rate": 0.0006,
+      "loss": 4.023551940917969,
+      "step": 770
+    },
+    {
+      "epoch": 5.455341506129598,
+      "grad_norm": 0.41956567764282227,
+      "learning_rate": 0.0006,
+      "loss": 4.035160064697266,
+      "step": 780
+    },
+    {
+      "epoch": 5.525394045534151,
+      "grad_norm": 0.32964956760406494,
+      "learning_rate": 0.0006,
+      "loss": 4.013400268554688,
+      "step": 790
+    },
+    {
+      "epoch": 5.5394045534150615,
+      "eval_loss": 4.1707587242126465,
+      "eval_runtime": 3.4448,
+      "eval_samples_per_second": 708.602,
+      "eval_steps_per_second": 11.321,
+      "step": 792
+    },
+    {
+      "epoch": 5.5954465849387045,
+      "grad_norm": 0.23922166228294373,
+      "learning_rate": 0.0006,
+      "loss": 4.003241729736328,
+      "step": 800
+    },
+    {
+      "epoch": 5.665499124343257,
+      "grad_norm": 0.22956278920173645,
+      "learning_rate": 0.0006,
+      "loss": 3.984221649169922,
+      "step": 810
+    },
+    {
+      "epoch": 5.735551663747811,
+      "grad_norm": 0.2851473391056061,
+      "learning_rate": 0.0006,
+      "loss": 3.9737266540527343,
+      "step": 820
+    },
+    {
+      "epoch": 5.805604203152364,
+      "grad_norm": 0.24042366445064545,
+      "learning_rate": 0.0006,
+      "loss": 3.972766876220703,
+      "step": 830
+    },
+    {
+      "epoch": 5.8756567425569175,
+      "grad_norm": 0.28479474782943726,
+      "learning_rate": 0.0006,
+      "loss": 3.972808837890625,
+      "step": 840
+    },
+    {
+      "epoch": 5.945709281961471,
+      "grad_norm": 0.38838624954223633,
+      "learning_rate": 0.0006,
+      "loss": 3.948868179321289,
+      "step": 850
+    },
+    {
+      "epoch": 6.0140105078809105,
+      "grad_norm": 0.3256464898586273,
+      "learning_rate": 0.0006,
+      "loss": 3.9456851959228514,
+      "step": 860
+    },
+    {
+      "epoch": 6.042031523642732,
+      "eval_loss": 4.092311859130859,
+      "eval_runtime": 3.3951,
+      "eval_samples_per_second": 718.982,
+      "eval_steps_per_second": 11.487,
+      "step": 864
+    },
+    {
+      "epoch": 6.084063047285464,
+      "grad_norm": 0.40261223912239075,
+      "learning_rate": 0.0006,
+      "loss": 3.878314208984375,
+      "step": 870
+    },
+    {
+      "epoch": 6.1541155866900175,
+      "grad_norm": 0.24834436178207397,
+      "learning_rate": 0.0006,
+      "loss": 3.8830192565917967,
+      "step": 880
+    },
+    {
+      "epoch": 6.224168126094571,
+      "grad_norm": 0.29639580845832825,
+      "learning_rate": 0.0006,
+      "loss": 3.8620361328125,
+      "step": 890
+    },
+    {
+      "epoch": 6.294220665499124,
+      "grad_norm": 0.24953483045101166,
+      "learning_rate": 0.0006,
+      "loss": 3.857699966430664,
+      "step": 900
+    },
+    {
+      "epoch": 6.364273204903678,
+      "grad_norm": 0.32599544525146484,
+      "learning_rate": 0.0006,
+      "loss": 3.8431934356689452,
+      "step": 910
+    },
+    {
+      "epoch": 6.434325744308231,
+      "grad_norm": 0.2872227132320404,
+      "learning_rate": 0.0006,
+      "loss": 3.8442756652832033,
+      "step": 920
+    },
+    {
+      "epoch": 6.504378283712785,
+      "grad_norm": 0.23899482190608978,
+      "learning_rate": 0.0006,
+      "loss": 3.8262901306152344,
+      "step": 930
+    },
+    {
+      "epoch": 6.546409807355516,
+      "eval_loss": 4.0181989669799805,
+      "eval_runtime": 3.4934,
+      "eval_samples_per_second": 698.739,
+      "eval_steps_per_second": 11.164,
+      "step": 936
+    },
+    {
+      "epoch": 6.574430823117338,
+      "grad_norm": 0.27845045924186707,
+      "learning_rate": 0.0006,
+      "loss": 3.8415252685546877,
+      "step": 940
+    },
+    {
+      "epoch": 6.644483362521892,
+      "grad_norm": 0.27635708451271057,
+      "learning_rate": 0.0006,
+      "loss": 3.828055572509766,
+      "step": 950
+    },
+    {
+      "epoch": 6.714535901926444,
+      "grad_norm": 0.27892908453941345,
+      "learning_rate": 0.0006,
+      "loss": 3.8211326599121094,
+      "step": 960
+    },
+    {
+      "epoch": 6.784588441330998,
+      "grad_norm": 0.26431673765182495,
+      "learning_rate": 0.0006,
+      "loss": 3.809326171875,
+      "step": 970
+    },
+    {
+      "epoch": 6.854640980735551,
+      "grad_norm": 0.2798311412334442,
+      "learning_rate": 0.0006,
+      "loss": 3.8122833251953123,
+      "step": 980
+    },
+    {
+      "epoch": 6.924693520140105,
+      "grad_norm": 0.33427661657333374,
+      "learning_rate": 0.0006,
+      "loss": 3.8152488708496093,
+      "step": 990
+    },
+    {
+      "epoch": 6.994746059544658,
+      "grad_norm": 0.2408551573753357,
+      "learning_rate": 0.0006,
+      "loss": 3.7974853515625,
+      "step": 1000
+    },
+    {
+      "epoch": 7.049036777583187,
+      "eval_loss": 3.9776611328125,
+      "eval_runtime": 3.5281,
+      "eval_samples_per_second": 691.883,
+      "eval_steps_per_second": 11.054,
+      "step": 1008
+    },
+    {
+      "epoch": 7.063047285464098,
+      "grad_norm": 0.4962931275367737,
+      "learning_rate": 0.0006,
+      "loss": 3.739080810546875,
+      "step": 1010
+    },
+    {
+      "epoch": 7.133099824868651,
+      "grad_norm": 0.3556423783302307,
+      "learning_rate": 0.0006,
+      "loss": 3.7520721435546873,
+      "step": 1020
+    },
+    {
+      "epoch": 7.203152364273205,
+      "grad_norm": 0.29888731241226196,
+      "learning_rate": 0.0006,
+      "loss": 3.739801788330078,
+      "step": 1030
+    },
+    {
+      "epoch": 7.273204903677758,
+      "grad_norm": 0.22400957345962524,
+      "learning_rate": 0.0006,
+      "loss": 3.7231216430664062,
+      "step": 1040
+    },
+    {
+      "epoch": 7.343257443082312,
+      "grad_norm": 0.22138261795043945,
+      "learning_rate": 0.0006,
+      "loss": 3.723231887817383,
+      "step": 1050
+    },
+    {
+      "epoch": 7.413309982486865,
+      "grad_norm": 0.24107135832309723,
+      "learning_rate": 0.0006,
+      "loss": 3.713814544677734,
+      "step": 1060
+    },
+    {
+      "epoch": 7.483362521891419,
+      "grad_norm": 0.2450643926858902,
+      "learning_rate": 0.0006,
+      "loss": 3.7051460266113283,
+      "step": 1070
+    },
+    {
+      "epoch": 7.553415061295972,
+      "grad_norm": 0.2829040586948395,
+      "learning_rate": 0.0006,
+      "loss": 3.710027313232422,
+      "step": 1080
+    },
+    {
+      "epoch": 7.553415061295972,
+      "eval_loss": 3.9317781925201416,
+      "eval_runtime": 3.9557,
+      "eval_samples_per_second": 617.084,
+      "eval_steps_per_second": 9.859,
+      "step": 1080
+    },
+    {
+      "epoch": 7.6234676007005255,
+      "grad_norm": 0.34311461448669434,
+      "learning_rate": 0.0006,
+      "loss": 3.7191070556640624,
+      "step": 1090
+    },
+    {
+      "epoch": 7.693520140105079,
+      "grad_norm": 0.2540404796600342,
+      "learning_rate": 0.0006,
+      "loss": 3.709515380859375,
+      "step": 1100
+    },
+    {
+      "epoch": 7.7635726795096325,
+      "grad_norm": 0.2629285454750061,
+      "learning_rate": 0.0006,
+      "loss": 3.6871673583984377,
+      "step": 1110
+    },
+    {
+      "epoch": 7.833625218914186,
+      "grad_norm": 0.23696254193782806,
+      "learning_rate": 0.0006,
+      "loss": 3.6875625610351563,
+      "step": 1120
+    },
+    {
+      "epoch": 7.903677758318739,
+      "grad_norm": 0.3052332401275635,
+      "learning_rate": 0.0006,
+      "loss": 3.695769500732422,
+      "step": 1130
+    },
+    {
+      "epoch": 7.973730297723293,
+      "grad_norm": 0.21110214293003082,
+      "learning_rate": 0.0006,
+      "loss": 3.6882614135742187,
+      "step": 1140
+    },
+    {
+      "epoch": 8.042031523642732,
+      "grad_norm": 0.422978937625885,
+      "learning_rate": 0.0006,
+      "loss": 3.6368148803710936,
+      "step": 1150
+    },
+    {
+      "epoch": 8.056042031523642,
+      "eval_loss": 3.90358304977417,
+      "eval_runtime": 3.3844,
+      "eval_samples_per_second": 721.261,
+      "eval_steps_per_second": 11.524,
+      "step": 1152
+    },
+    {
+      "epoch": 8.112084063047286,
+      "grad_norm": 0.34560948610305786,
+      "learning_rate": 0.0006,
+      "loss": 3.6172374725341796,
+      "step": 1160
+    },
+    {
+      "epoch": 8.18213660245184,
+      "grad_norm": 0.33484628796577454,
+      "learning_rate": 0.0006,
+      "loss": 3.613117218017578,
+      "step": 1170
+    },
+    {
+      "epoch": 8.252189141856393,
+      "grad_norm": 0.3299332559108734,
+      "learning_rate": 0.0006,
+      "loss": 3.6176902770996096,
+      "step": 1180
+    },
+    {
+      "epoch": 8.322241681260946,
+      "grad_norm": 0.25558748841285706,
+      "learning_rate": 0.0006,
+      "loss": 3.6145286560058594,
+      "step": 1190
+    },
+    {
+      "epoch": 8.3922942206655,
+      "grad_norm": 0.23964306712150574,
+      "learning_rate": 0.0006,
+      "loss": 3.6121849060058593,
+      "step": 1200
+    },
+    {
+      "epoch": 8.462346760070053,
+      "grad_norm": 0.23925594985485077,
+      "learning_rate": 0.0006,
+      "loss": 3.615177536010742,
+      "step": 1210
+    },
+    {
+      "epoch": 8.532399299474607,
+      "grad_norm": 0.23387187719345093,
+      "learning_rate": 0.0006,
+      "loss": 3.6193164825439452,
+      "step": 1220
+    },
+    {
+      "epoch": 8.560420315236428,
+      "eval_loss": 3.8646886348724365,
+      "eval_runtime": 3.4045,
+      "eval_samples_per_second": 716.998,
+      "eval_steps_per_second": 11.456,
+      "step": 1224
+    },
+    {
+      "epoch": 8.602451838879158,
+      "grad_norm": 0.24263112246990204,
+      "learning_rate": 0.0006,
+      "loss": 3.6070907592773436,
+      "step": 1230
+    },
+    {
+      "epoch": 8.672504378283712,
+      "grad_norm": 0.24607565999031067,
+      "learning_rate": 0.0006,
+      "loss": 3.6143115997314452,
+      "step": 1240
+    },
+    {
+      "epoch": 8.742556917688265,
+      "grad_norm": 0.273237943649292,
+      "learning_rate": 0.0006,
+      "loss": 3.607615661621094,
+      "step": 1250
+    },
+    {
+      "epoch": 8.812609457092819,
+      "grad_norm": 0.22331637144088745,
+      "learning_rate": 0.0006,
+      "loss": 3.6233566284179686,
+      "step": 1260
+    },
+    {
+      "epoch": 8.882661996497372,
+      "grad_norm": 0.3021933436393738,
+      "learning_rate": 0.0006,
+      "loss": 3.601115417480469,
+      "step": 1270
+    },
+    {
+      "epoch": 8.952714535901926,
+      "grad_norm": 0.249200239777565,
+      "learning_rate": 0.0006,
+      "loss": 3.597786712646484,
+      "step": 1280
+    },
+    {
+      "epoch": 9.021015761821365,
+      "grad_norm": 0.25300464034080505,
+      "learning_rate": 0.0006,
+      "loss": 3.591754913330078,
+      "step": 1290
+    },
+    {
+      "epoch": 9.063047285464098,
+      "eval_loss": 3.844236373901367,
+      "eval_runtime": 3.4005,
+      "eval_samples_per_second": 717.835,
+      "eval_steps_per_second": 11.469,
+      "step": 1296
+    },
+    {
+      "epoch": 9.091068301225919,
+      "grad_norm": 0.3442411720752716,
+      "learning_rate": 0.0006,
+      "loss": 3.5079620361328123,
+      "step": 1300
+    },
+    {
+      "epoch": 9.161120840630472,
+      "grad_norm": 0.35806402564048767,
+      "learning_rate": 0.0006,
+      "loss": 3.5409027099609376,
+      "step": 1310
+    },
+    {
+      "epoch": 9.231173380035026,
+      "grad_norm": 0.35811489820480347,
+      "learning_rate": 0.0006,
+      "loss": 3.54935302734375,
+      "step": 1320
+    },
+    {
+      "epoch": 9.30122591943958,
+      "grad_norm": 0.2234598696231842,
+      "learning_rate": 0.0006,
+      "loss": 3.5450748443603515,
+      "step": 1330
+    },
+    {
+      "epoch": 9.371278458844133,
+      "grad_norm": 0.21218106150627136,
+      "learning_rate": 0.0006,
+      "loss": 3.524667739868164,
+      "step": 1340
+    },
+    {
+      "epoch": 9.441330998248686,
+      "grad_norm": 0.2407359778881073,
+      "learning_rate": 0.0006,
+      "loss": 3.54589729309082,
+      "step": 1350
+    },
+    {
+      "epoch": 9.51138353765324,
+      "grad_norm": 0.29132887721061707,
+      "learning_rate": 0.0006,
+      "loss": 3.532334899902344,
+      "step": 1360
+    },
+    {
+      "epoch": 9.567425569176883,
+      "eval_loss": 3.823045015335083,
+      "eval_runtime": 3.5809,
+      "eval_samples_per_second": 681.681,
+      "eval_steps_per_second": 10.891,
+      "step": 1368
+    },
+    {
+      "epoch": 9.581436077057793,
+      "grad_norm": 0.27332165837287903,
+      "learning_rate": 0.0006,
+      "loss": 3.5407684326171873,
+      "step": 1370
+    },
+    {
+      "epoch": 9.651488616462347,
+      "grad_norm": 0.22323289513587952,
+      "learning_rate": 0.0006,
+      "loss": 3.5315406799316404,
+      "step": 1380
+    },
+    {
+      "epoch": 9.7215411558669,
+      "grad_norm": 0.23029769957065582,
+      "learning_rate": 0.0006,
+      "loss": 3.5354686737060548,
+      "step": 1390
+    },
+    {
+      "epoch": 9.791593695271454,
+      "grad_norm": 0.239872008562088,
+      "learning_rate": 0.0006,
+      "loss": 3.5226299285888674,
+      "step": 1400
+    },
+    {
+      "epoch": 9.861646234676007,
+      "grad_norm": 0.2355087697505951,
+      "learning_rate": 0.0006,
+      "loss": 3.5222801208496093,
+      "step": 1410
+    },
+    {
+      "epoch": 9.93169877408056,
+      "grad_norm": 0.22256873548030853,
+      "learning_rate": 0.0006,
+      "loss": 3.5412212371826173,
+      "step": 1420
+    },
+    {
+      "epoch": 10.0,
+      "grad_norm": 0.25026267766952515,
+      "learning_rate": 0.0006,
+      "loss": 3.519300079345703,
+      "step": 1430
+    },
+    {
+      "epoch": 10.070052539404553,
+      "grad_norm": 0.418517142534256,
+      "learning_rate": 0.0006,
+      "loss": 3.4538883209228515,
+      "step": 1440
+    },
+    {
+      "epoch": 10.070052539404553,
+      "eval_loss": 3.8263518810272217,
+      "eval_runtime": 3.5579,
+      "eval_samples_per_second": 686.082,
+      "eval_steps_per_second": 10.962,
+      "step": 1440
+    },
+    {
+      "epoch": 10.140105078809107,
+      "grad_norm": 0.4168876111507416,
+      "learning_rate": 0.0006,
+      "loss": 3.474140930175781,
+      "step": 1450
+    },
+    {
+      "epoch": 10.21015761821366,
+      "grad_norm": 0.3406018316745758,
+      "learning_rate": 0.0006,
+      "loss": 3.473613739013672,
+      "step": 1460
+    },
+    {
+      "epoch": 10.280210157618214,
+      "grad_norm": 0.21272197365760803,
+      "learning_rate": 0.0006,
+      "loss": 3.461037445068359,
+      "step": 1470
+    },
+    {
+      "epoch": 10.350262697022767,
+      "grad_norm": 0.2185119390487671,
+      "learning_rate": 0.0006,
+      "loss": 3.467035675048828,
+      "step": 1480
+    },
+    {
+      "epoch": 10.42031523642732,
+      "grad_norm": 0.2328590750694275,
+      "learning_rate": 0.0006,
+      "loss": 3.461333465576172,
+      "step": 1490
+    },
+    {
+      "epoch": 10.490367775831874,
+      "grad_norm": 0.2336580902338028,
+      "learning_rate": 0.0006,
+      "loss": 3.4777130126953124,
+      "step": 1500
+    },
+    {
+      "epoch": 10.560420315236428,
+      "grad_norm": 0.28968092799186707,
+      "learning_rate": 0.0006,
+      "loss": 3.466571807861328,
+      "step": 1510
+    },
+    {
+      "epoch": 10.574430823117337,
+      "eval_loss": 3.788224697113037,
+      "eval_runtime": 3.7388,
+      "eval_samples_per_second": 652.875,
+      "eval_steps_per_second": 10.431,
+      "step": 1512
+    },
+    {
+      "epoch": 10.630472854640981,
+      "grad_norm": 0.22140789031982422,
+      "learning_rate": 0.0006,
+      "loss": 3.46275634765625,
+      "step": 1520
+    },
+    {
+      "epoch": 10.700525394045535,
+      "grad_norm": 0.21839596331119537,
+      "learning_rate": 0.0006,
+      "loss": 3.4691925048828125,
+      "step": 1530
+    },
+    {
+      "epoch": 10.770577933450088,
+      "grad_norm": 0.26760339736938477,
+      "learning_rate": 0.0006,
+      "loss": 3.4725528717041017,
+      "step": 1540
+    },
+    {
+      "epoch": 10.840630472854642,
+      "grad_norm": 0.2242419421672821,
+      "learning_rate": 0.0006,
+      "loss": 3.475904846191406,
+      "step": 1550
+    },
+    {
+      "epoch": 10.910683012259195,
+      "grad_norm": 0.22520668804645538,
+      "learning_rate": 0.0006,
+      "loss": 3.4760223388671876,
+      "step": 1560
+    },
+    {
+      "epoch": 10.980735551663749,
+      "grad_norm": 0.23904766142368317,
+      "learning_rate": 0.0006,
+      "loss": 3.4671859741210938,
+      "step": 1570
+    },
+    {
+      "epoch": 11.049036777583188,
+      "grad_norm": 0.3444315791130066,
+      "learning_rate": 0.0006,
+      "loss": 3.4159122467041017,
+      "step": 1580
+    },
+    {
+      "epoch": 11.07705779334501,
+      "eval_loss": 3.791930913925171,
+      "eval_runtime": 3.4685,
+      "eval_samples_per_second": 703.773,
+      "eval_steps_per_second": 11.244,
+      "step": 1584
+    },
+    {
+      "epoch": 11.119089316987742,
+      "grad_norm": 0.30959776043891907,
+      "learning_rate": 0.0006,
+      "loss": 3.38648681640625,
+      "step": 1590
+    },
+    {
+      "epoch": 11.189141856392295,
+      "grad_norm": 0.2670822739601135,
+      "learning_rate": 0.0006,
+      "loss": 3.387676239013672,
+      "step": 1600
+    },
+    {
+      "epoch": 11.259194395796847,
+      "grad_norm": 0.2430114895105362,
+      "learning_rate": 0.0006,
+      "loss": 3.3922351837158202,
+      "step": 1610
+    },
+    {
+      "epoch": 11.3292469352014,
+      "grad_norm": 0.259003609418869,
+      "learning_rate": 0.0006,
+      "loss": 3.401170349121094,
+      "step": 1620
+    },
+    {
+      "epoch": 11.399299474605954,
+      "grad_norm": 0.28309041261672974,
+      "learning_rate": 0.0006,
+      "loss": 3.405831146240234,
+      "step": 1630
+    },
+    {
+      "epoch": 11.469352014010507,
+      "grad_norm": 0.2640569508075714,
+      "learning_rate": 0.0006,
+      "loss": 3.4174739837646486,
+      "step": 1640
+    },
+    {
+      "epoch": 11.53940455341506,
+      "grad_norm": 0.24435488879680634,
+      "learning_rate": 0.0006,
+      "loss": 3.40150146484375,
+      "step": 1650
+    },
+    {
+      "epoch": 11.581436077057793,
+      "eval_loss": 3.7666003704071045,
+      "eval_runtime": 3.5185,
+      "eval_samples_per_second": 693.764,
+      "eval_steps_per_second": 11.084,
+      "step": 1656
+    },
+    {
+      "epoch": 11.609457092819614,
+      "grad_norm": 0.23578906059265137,
+      "learning_rate": 0.0006,
+      "loss": 3.4165611267089844,
+      "step": 1660
+    },
+    {
+      "epoch": 11.679509632224168,
+      "grad_norm": 0.23872224986553192,
+      "learning_rate": 0.0006,
+      "loss": 3.4270652770996093,
+      "step": 1670
+    },
+    {
+      "epoch": 11.749562171628721,
+      "grad_norm": 0.2594001293182373,
+      "learning_rate": 0.0006,
+      "loss": 3.4211357116699217,
+      "step": 1680
+    },
+    {
+      "epoch": 11.819614711033275,
+      "grad_norm": 0.24929523468017578,
+      "learning_rate": 0.0006,
+      "loss": 3.423053741455078,
+      "step": 1690
+    },
+    {
+      "epoch": 11.889667250437828,
+      "grad_norm": 0.2450554370880127,
+      "learning_rate": 0.0006,
+      "loss": 3.4230796813964846,
+      "step": 1700
+    },
+    {
+      "epoch": 11.959719789842381,
+      "grad_norm": 0.2280256599187851,
+      "learning_rate": 0.0006,
+      "loss": 3.423610305786133,
+      "step": 1710
+    },
+    {
+      "epoch": 12.028021015761821,
+      "grad_norm": 0.40455758571624756,
+      "learning_rate": 0.0006,
+      "loss": 3.389678192138672,
+      "step": 1720
+    },
+    {
+      "epoch": 12.084063047285465,
+      "eval_loss": 3.7763803005218506,
+      "eval_runtime": 3.6455,
+      "eval_samples_per_second": 669.594,
+      "eval_steps_per_second": 10.698,
+      "step": 1728
+    },
+    {
+      "epoch": 12.098073555166375,
+      "grad_norm": 0.3537187874317169,
+      "learning_rate": 0.0006,
+      "loss": 3.333454132080078,
+      "step": 1730
+    },
+    {
+      "epoch": 12.168126094570928,
+      "grad_norm": 0.29340294003486633,
+      "learning_rate": 0.0006,
+      "loss": 3.344945526123047,
+      "step": 1740
+    },
+    {
+      "epoch": 12.238178633975481,
+      "grad_norm": 0.30197280645370483,
+      "learning_rate": 0.0006,
+      "loss": 3.360484313964844,
+      "step": 1750
+    },
+    {
+      "epoch": 12.308231173380035,
+      "grad_norm": 0.22601750493049622,
+      "learning_rate": 0.0006,
+      "loss": 3.350424957275391,
+      "step": 1760
+    },
+    {
+      "epoch": 12.378283712784588,
+      "grad_norm": 0.23398347198963165,
+      "learning_rate": 0.0006,
+      "loss": 3.3569656372070313,
+      "step": 1770
+    },
+    {
+      "epoch": 12.448336252189142,
+      "grad_norm": 0.2278864085674286,
+      "learning_rate": 0.0006,
+      "loss": 3.3536346435546873,
+      "step": 1780
+    },
+    {
+      "epoch": 12.518388791593695,
+      "grad_norm": 0.23192206025123596,
+      "learning_rate": 0.0006,
+      "loss": 3.361003875732422,
+      "step": 1790
+    },
+    {
+      "epoch": 12.588441330998249,
+      "grad_norm": 0.2884629964828491,
+      "learning_rate": 0.0006,
+      "loss": 3.3694381713867188,
+      "step": 1800
+    },
+    {
+      "epoch": 12.588441330998249,
+      "eval_loss": 3.753028154373169,
+      "eval_runtime": 3.5308,
+      "eval_samples_per_second": 691.353,
+      "eval_steps_per_second": 11.046,
+      "step": 1800
+    },
+    {
+      "epoch": 12.658493870402802,
+      "grad_norm": 0.22837434709072113,
+      "learning_rate": 0.0006,
+      "loss": 3.3743656158447264,
+      "step": 1810
+    },
+    {
+      "epoch": 12.728546409807356,
+      "grad_norm": 0.23902975022792816,
+      "learning_rate": 0.0006,
+      "loss": 3.3757350921630858,
+      "step": 1820
+    },
+    {
+      "epoch": 12.79859894921191,
+      "grad_norm": 0.24309372901916504,
+      "learning_rate": 0.0006,
+      "loss": 3.373929977416992,
+      "step": 1830
+    },
+    {
+      "epoch": 12.868651488616463,
+      "grad_norm": 0.265935480594635,
+      "learning_rate": 0.0006,
+      "loss": 3.3781661987304688,
+      "step": 1840
+    },
+    {
+      "epoch": 12.938704028021016,
+      "grad_norm": 0.2791769802570343,
+      "learning_rate": 0.0006,
+      "loss": 3.3764923095703123,
+      "step": 1850
+    },
+    {
+      "epoch": 13.007005253940456,
+      "grad_norm": 0.259432852268219,
+      "learning_rate": 0.0006,
+      "loss": 3.360120391845703,
+      "step": 1860
+    },
+    {
+      "epoch": 13.07705779334501,
+      "grad_norm": 0.29833364486694336,
+      "learning_rate": 0.0006,
+      "loss": 3.2793312072753906,
+      "step": 1870
+    },
+    {
+      "epoch": 13.091068301225919,
+      "eval_loss": 3.760260820388794,
+      "eval_runtime": 3.4544,
+      "eval_samples_per_second": 706.638,
+      "eval_steps_per_second": 11.29,
+      "step": 1872
+    },
+    {
+      "epoch": 13.147110332749563,
+      "grad_norm": 0.34551334381103516,
+      "learning_rate": 0.0006,
+      "loss": 3.28541259765625,
+      "step": 1880
+    },
+    {
+      "epoch": 13.217162872154116,
+      "grad_norm": 0.30898651480674744,
+      "learning_rate": 0.0006,
+      "loss": 3.2950508117675783,
+      "step": 1890
+    },
+    {
+      "epoch": 13.28721541155867,
+      "grad_norm": 0.2684871256351471,
+      "learning_rate": 0.0006,
+      "loss": 3.3074264526367188,
+      "step": 1900
+    },
+    {
+      "epoch": 13.357267950963223,
+      "grad_norm": 0.22169773280620575,
+      "learning_rate": 0.0006,
+      "loss": 3.3030105590820313,
+      "step": 1910
+    },
+    {
+      "epoch": 13.427320490367777,
+      "grad_norm": 0.27856582403182983,
+      "learning_rate": 0.0006,
+      "loss": 3.3179840087890624,
+      "step": 1920
+    },
+    {
+      "epoch": 13.49737302977233,
+      "grad_norm": 0.24769793450832367,
+      "learning_rate": 0.0006,
+      "loss": 3.3265071868896485,
+      "step": 1930
+    },
+    {
+      "epoch": 13.567425569176883,
+      "grad_norm": 0.224507674574852,
+      "learning_rate": 0.0006,
+      "loss": 3.3263214111328123,
+      "step": 1940
+    },
+    {
+      "epoch": 13.595446584938705,
+      "eval_loss": 3.7428011894226074,
+      "eval_runtime": 3.3783,
+      "eval_samples_per_second": 722.558,
+      "eval_steps_per_second": 11.544,
+      "step": 1944
+    },
+    {
+      "epoch": 13.637478108581437,
+      "grad_norm": 0.25914138555526733,
+      "learning_rate": 0.0006,
+      "loss": 3.332944869995117,
+      "step": 1950
+    },
+    {
+      "epoch": 13.70753064798599,
+      "grad_norm": 0.2343108355998993,
+      "learning_rate": 0.0006,
+      "loss": 3.3334304809570314,
+      "step": 1960
+    },
+    {
+      "epoch": 13.777583187390542,
+      "grad_norm": 0.23315395414829254,
+      "learning_rate": 0.0006,
+      "loss": 3.3350761413574217,
+      "step": 1970
+    },
+    {
+      "epoch": 13.847635726795096,
+      "grad_norm": 0.2602457106113434,
+      "learning_rate": 0.0006,
+      "loss": 3.3352535247802733,
+      "step": 1980
+    },
+    {
+      "epoch": 13.917688266199649,
+      "grad_norm": 0.24317029118537903,
+      "learning_rate": 0.0006,
+      "loss": 3.331440734863281,
+      "step": 1990
+    },
+    {
+      "epoch": 13.987740805604203,
+      "grad_norm": 0.23694735765457153,
+      "learning_rate": 0.0006,
+      "loss": 3.3404006958007812,
+      "step": 2000
+    },
+    {
+      "epoch": 14.056042031523642,
+      "grad_norm": 0.32518285512924194,
+      "learning_rate": 0.0006,
+      "loss": 3.254779052734375,
+      "step": 2010
+    },
+    {
+      "epoch": 14.098073555166375,
+      "eval_loss": 3.7570059299468994,
+      "eval_runtime": 3.4079,
+      "eval_samples_per_second": 716.283,
+      "eval_steps_per_second": 11.444,
+      "step": 2016
+    },
+    {
+      "epoch": 14.126094570928196,
+      "grad_norm": 0.3998780846595764,
+      "learning_rate": 0.0006,
+      "loss": 3.2511558532714844,
+      "step": 2020
+    },
+    {
+      "epoch": 14.196147110332749,
+      "grad_norm": 0.2794221043586731,
+      "learning_rate": 0.0006,
+      "loss": 3.2535163879394533,
+      "step": 2030
+    },
+    {
+      "epoch": 14.266199649737302,
+      "grad_norm": 0.2672958970069885,
+      "learning_rate": 0.0006,
+      "loss": 3.2557903289794923,
+      "step": 2040
+    },
+    {
+      "epoch": 14.336252189141856,
+      "grad_norm": 0.3607710599899292,
+      "learning_rate": 0.0006,
+      "loss": 3.2687118530273436,
+      "step": 2050
+    },
+    {
+      "epoch": 14.40630472854641,
+      "grad_norm": 0.26717787981033325,
+      "learning_rate": 0.0006,
+      "loss": 3.283890151977539,
+      "step": 2060
+    },
+    {
+      "epoch": 14.476357267950963,
+      "grad_norm": 0.22727353870868683,
+      "learning_rate": 0.0006,
+      "loss": 3.2842716217041015,
+      "step": 2070
+    },
+    {
+      "epoch": 14.546409807355516,
+      "grad_norm": 0.22838088870048523,
+      "learning_rate": 0.0006,
+      "loss": 3.278900909423828,
+      "step": 2080
+    },
+    {
+      "epoch": 14.602451838879158,
+      "eval_loss": 3.73828125,
+      "eval_runtime": 3.4709,
+      "eval_samples_per_second": 703.278,
+      "eval_steps_per_second": 11.236,
+      "step": 2088
+    },
+    {
+      "epoch": 14.61646234676007,
+      "grad_norm": 0.22517885267734528,
+      "learning_rate": 0.0006,
+      "loss": 3.290675735473633,
+      "step": 2090
+    },
+    {
+      "epoch": 14.686514886164623,
+      "grad_norm": 0.24148719012737274,
+      "learning_rate": 0.0006,
+      "loss": 3.2893081665039063,
+      "step": 2100
+    },
+    {
+      "epoch": 14.756567425569177,
+      "grad_norm": 0.22470487654209137,
+      "learning_rate": 0.0006,
+      "loss": 3.29949836730957,
+      "step": 2110
+    },
+    {
+      "epoch": 14.82661996497373,
+      "grad_norm": 0.2813766300678253,
+      "learning_rate": 0.0006,
+      "loss": 3.3061370849609375,
+      "step": 2120
+    },
+    {
+      "epoch": 14.896672504378284,
+      "grad_norm": 0.2263512760400772,
+      "learning_rate": 0.0006,
+      "loss": 3.3054107666015624,
+      "step": 2130
+    },
+    {
+      "epoch": 14.966725043782837,
+      "grad_norm": 0.24720951914787292,
+      "learning_rate": 0.0006,
+      "loss": 3.290102005004883,
+      "step": 2140
+    },
+    {
+      "epoch": 15.035026269702277,
+      "grad_norm": 0.3907608687877655,
+      "learning_rate": 0.0006,
+      "loss": 3.239105224609375,
+      "step": 2150
+    },
+    {
+      "epoch": 15.10507880910683,
+      "grad_norm": 0.4218903183937073,
+      "learning_rate": 0.0006,
+      "loss": 3.2074462890625,
+      "step": 2160
+    },
+    {
+      "epoch": 15.10507880910683,
+      "eval_loss": 3.769059896469116,
+      "eval_runtime": 4.3518,
+      "eval_samples_per_second": 560.917,
+      "eval_steps_per_second": 8.962,
+      "step": 2160
+    },
+    {
+      "epoch": 15.175131348511384,
+      "grad_norm": 0.3093777000904083,
+      "learning_rate": 0.0006,
+      "loss": 3.230516815185547,
+      "step": 2170
+    },
+    {
+      "epoch": 15.245183887915937,
+      "grad_norm": 0.24631434679031372,
+      "learning_rate": 0.0006,
+      "loss": 3.235198211669922,
+      "step": 2180
+    },
+    {
+      "epoch": 15.31523642732049,
+      "grad_norm": 0.22635217010974884,
+      "learning_rate": 0.0006,
+      "loss": 3.2339107513427736,
+      "step": 2190
+    },
+    {
+      "epoch": 15.385288966725044,
+      "grad_norm": 0.24024364352226257,
+      "learning_rate": 0.0006,
+      "loss": 3.2306838989257813,
+      "step": 2200
+    },
+    {
+      "epoch": 15.455341506129598,
+      "grad_norm": 0.25333160161972046,
+      "learning_rate": 0.0006,
+      "loss": 3.2406005859375,
+      "step": 2210
+    },
+    {
+      "epoch": 15.525394045534151,
+      "grad_norm": 0.2500869035720825,
+      "learning_rate": 0.0006,
+      "loss": 3.242127227783203,
+      "step": 2220
+    },
+    {
+      "epoch": 15.595446584938705,
+      "grad_norm": 0.2822413444519043,
+      "learning_rate": 0.0006,
+      "loss": 3.2631065368652346,
+      "step": 2230
+    },
+    {
+      "epoch": 15.609457092819614,
+      "eval_loss": 3.7376556396484375,
+      "eval_runtime": 3.2511,
+      "eval_samples_per_second": 750.83,
+      "eval_steps_per_second": 11.996,
+      "step": 2232
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 7150,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 50,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": false,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 9.557891315128074e+17,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/runs/t2a/t2a_150M_C0_baseline/train.log b/runs/t2a/t2a_150M_C0_baseline/train.log
new file mode 100644
index 0000000000000000000000000000000000000000..5a4c6ec6e16125a1c415d4369d2554454356eccd
--- /dev/null
+++ b/runs/t2a/t2a_150M_C0_baseline/train.log
@@ -0,0 +1,3042 @@
+W0818 21:30:27.653000 3060219 torch/distributed/run.py:851] 
+W0818 21:30:27.653000 3060219 torch/distributed/run.py:851] *****************************************
+W0818 21:30:27.653000 3060219 torch/distributed/run.py:851] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+W0818 21:30:27.653000 3060219 torch/distributed/run.py:851] *****************************************
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+total params: 145833984
+[dataset] loaded packed cache: 73061 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+total params: 145833984
+[dataset] loaded packed cache: 73061 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+total params: 145833984
+[dataset] loaded packed cache: 73061 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+total params: 145833984
+[dataset] loaded packed cache: 73061 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+total params: 145833984
+[dataset] loaded packed cache: 73061 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+total params: 145833984
+[dataset] loaded packed cache: 73061 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+total params: 145833984
+[dataset] loaded packed cache: 73061 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+total params: 145833984
+[dataset] loaded packed cache: 73061 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+[transformers] ***** Running training *****
+[transformers]   Num examples = 73,061
+[transformers]   Num Epochs = 50
+[transformers]   Num update steps per epoch = 143
+[transformers]   Instantaneous batch size per device = 8
+[transformers]   Total train batch size (w. parallel, distributed & accumulation) = 512
+[transformers]   Gradient Accumulation steps = 8
+[transformers]   Total optimization steps = 7,150
+[transformers]   Number of trainable parameters = 145,833,984
+wandb: WARNING Changes to your `wandb` environment variables will be ignored because your `wandb` session has already started. For more information on how to modify your settings with `wandb.init()` arguments, please refer to https://wandb.me/wandb-init.
+wandb: setting up run gc8jxphi
+wandb: Tracking run with wandb version 0.28.2
+wandb: Run data is saved locally in /home/ubuntu/workspace/data-augmentations-for-pretraining/wandb/run-20260818_213047-gc8jxphi
+wandb: Run `wandb offline` to turn off syncing.
+wandb: Syncing run t2a_150M_C0_baseline
+wandb: ⭐️ View project at https://wandb.ai/slowrun-wankers/data-aug-pretraining
+wandb: 🚀 View run at https://wandb.ai/slowrun-wankers/data-aug-pretraining/runs/gc8jxphi
+
  0%|          | 0/7150 [00:00<?, ?it/s]wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+[rank4]:W0818 21:31:10.418000 3060343 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+[rank7]:W0818 21:31:10.548000 3060346 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+[rank6]:W0818 21:31:10.877000 3060345 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+[rank3]:W0818 21:31:10.976000 3060342 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+[rank5]:W0818 21:31:11.283000 3060344 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+[rank1]:W0818 21:31:11.372000 3060340 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+[rank2]:W0818 21:31:11.430000 3060341 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+[rank0]:W0818 21:31:12.156000 3060339 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+
  0%|          | 1/7150 [00:41<81:38:25, 41.11s/it]
                                                   
{'loss': '12.04', 'grad_norm': '1.746', 'learning_rate': '0', 'epoch': '0.007005'}
+
  0%|          | 1/7150 [00:41<81:38:25, 41.11s/it]
  0%|          | 2/7150 [00:41<34:23:16, 17.32s/it]
  0%|          | 3/7150 [00:42<19:16:21,  9.71s/it]
  0%|          | 4/7150 [00:43<12:10:17,  6.13s/it]
  0%|          | 5/7150 [00:43<8:14:53,  4.16s/it] 
  0%|          | 6/7150 [00:44<5:52:55,  2.96s/it]
  0%|          | 7/7150 [00:45<4:22:57,  2.21s/it]
  0%|          | 8/7150 [00:45<3:23:53,  1.71s/it]
  0%|          | 9/7150 [00:46<2:44:21,  1.38s/it]
  0%|          | 10/7150 [00:46<2:17:31,  1.16s/it]
                                                   
{'loss': '11.75', 'grad_norm': '1.785', 'learning_rate': '5.4e-05', 'epoch': '0.07005'}
+
  0%|          | 10/7150 [00:46<2:17:31,  1.16s/it]
  0%|          | 11/7150 [00:47<1:59:09,  1.00s/it]
  0%|          | 12/7150 [00:48<1:46:26,  1.12it/s]
  0%|          | 13/7150 [00:48<1:37:38,  1.22it/s]
  0%|          | 14/7150 [00:49<1:31:33,  1.30it/s]
  0%|          | 15/7150 [00:50<1:27:17,  1.36it/s]
  0%|          | 16/7150 [00:50<1:24:21,  1.41it/s]
  0%|          | 17/7150 [00:51<1:22:12,  1.45it/s]
  0%|          | 18/7150 [00:52<1:20:45,  1.47it/s]
  0%|          | 19/7150 [00:52<1:19:39,  1.49it/s]
  0%|          | 20/7150 [00:53<1:18:57,  1.50it/s]
                                                   
{'loss': '11.04', 'grad_norm': '1.679', 'learning_rate': '0.000114', 'epoch': '0.1401'}
+
  0%|          | 20/7150 [00:53<1:18:57,  1.50it/s]
  0%|          | 21/7150 [00:54<1:18:25,  1.52it/s]
  0%|          | 22/7150 [00:54<1:18:01,  1.52it/s]
  0%|          | 23/7150 [00:55<1:17:43,  1.53it/s]
  0%|          | 24/7150 [00:56<1:17:35,  1.53it/s]
  0%|          | 25/7150 [00:56<1:17:22,  1.53it/s]
  0%|          | 26/7150 [00:57<1:17:19,  1.54it/s]
  0%|          | 27/7150 [00:58<1:17:13,  1.54it/s]
  0%|          | 28/7150 [00:58<1:17:10,  1.54it/s]
  0%|          | 29/7150 [00:59<1:17:08,  1.54it/s]
  0%|          | 30/7150 [00:59<1:17:06,  1.54it/s]
                                                   
{'loss': '10.27', 'grad_norm': '1.643', 'learning_rate': '0.000174', 'epoch': '0.2102'}
+
  0%|          | 30/7150 [00:59<1:17:06,  1.54it/s]
  0%|          | 31/7150 [01:00<1:17:06,  1.54it/s]
  0%|          | 32/7150 [01:01<1:17:02,  1.54it/s]
  0%|          | 33/7150 [01:01<1:16:59,  1.54it/s]
  0%|          | 34/7150 [01:02<1:16:55,  1.54it/s]
  0%|          | 35/7150 [01:03<1:16:57,  1.54it/s]
  1%|          | 36/7150 [01:03<1:16:55,  1.54it/s]
  1%|          | 37/7150 [01:04<1:16:56,  1.54it/s]
  1%|          | 38/7150 [01:05<1:16:53,  1.54it/s]
  1%|          | 39/7150 [01:05<1:16:51,  1.54it/s]
  1%|          | 40/7150 [01:06<1:16:47,  1.54it/s]
                                                   
{'loss': '9.223', 'grad_norm': '1.502', 'learning_rate': '0.000234', 'epoch': '0.2802'}
+
  1%|          | 40/7150 [01:06<1:16:47,  1.54it/s]
  1%|          | 41/7150 [01:07<1:16:50,  1.54it/s]
  1%|          | 42/7150 [01:07<1:16:50,  1.54it/s]
  1%|          | 43/7150 [01:08<1:16:49,  1.54it/s]
  1%|          | 44/7150 [01:09<1:16:51,  1.54it/s]
  1%|          | 45/7150 [01:09<1:16:48,  1.54it/s]
  1%|          | 46/7150 [01:10<1:16:51,  1.54it/s]
  1%|          | 47/7150 [01:11<1:16:49,  1.54it/s]
  1%|          | 48/7150 [01:11<1:16:49,  1.54it/s]
  1%|          | 49/7150 [01:12<1:16:46,  1.54it/s]
  1%|          | 50/7150 [01:12<1:16:45,  1.54it/s]
                                                   
{'loss': '8.246', 'grad_norm': '0.9086', 'learning_rate': '0.000294', 'epoch': '0.3503'}
+
  1%|          | 50/7150 [01:12<1:16:45,  1.54it/s]
  1%|          | 51/7150 [01:13<1:16:49,  1.54it/s]
  1%|          | 52/7150 [01:14<1:16:46,  1.54it/s]
  1%|          | 53/7150 [01:14<1:16:49,  1.54it/s]
  1%|          | 54/7150 [01:15<1:16:48,  1.54it/s]
  1%|          | 55/7150 [01:16<1:16:52,  1.54it/s]
  1%|          | 56/7150 [01:16<1:16:52,  1.54it/s]
  1%|          | 57/7150 [01:17<1:16:51,  1.54it/s]
  1%|          | 58/7150 [01:18<1:16:47,  1.54it/s]
  1%|          | 59/7150 [01:18<1:16:50,  1.54it/s]
  1%|          | 60/7150 [01:19<1:16:50,  1.54it/s]
                                                   
{'loss': '7.676', 'grad_norm': '0.4325', 'learning_rate': '0.000354', 'epoch': '0.4203'}
+
  1%|          | 60/7150 [01:19<1:16:50,  1.54it/s]
  1%|          | 61/7150 [01:20<1:16:54,  1.54it/s]
  1%|          | 62/7150 [01:20<1:16:51,  1.54it/s]
  1%|          | 63/7150 [01:21<1:16:48,  1.54it/s]
  1%|          | 64/7150 [01:22<1:16:50,  1.54it/s]
  1%|          | 65/7150 [01:22<1:16:51,  1.54it/s]
  1%|          | 66/7150 [01:23<1:16:53,  1.54it/s]
  1%|          | 67/7150 [01:24<1:16:50,  1.54it/s]
  1%|          | 68/7150 [01:24<1:16:53,  1.53it/s]
  1%|          | 69/7150 [01:25<1:16:52,  1.54it/s]
  1%|          | 70/7150 [01:25<1:16:50,  1.54it/s]
                                                   
{'loss': '7.524', 'grad_norm': '0.4937', 'learning_rate': '0.000414', 'epoch': '0.4904'}
+
  1%|          | 70/7150 [01:25<1:16:50,  1.54it/s]
  1%|          | 71/7150 [01:26<1:16:53,  1.53it/s]
  1%|          | 72/7150 [01:27<1:16:53,  1.53it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 28.53it/s]
+
 15%|█▌        | 6/39 [00:00<00:02, 15.95it/s]
+
 21%|██        | 8/39 [00:00<00:02, 14.36it/s]
+
 26%|██▌       | 10/39 [00:00<00:02, 13.56it/s]
+
 31%|███       | 12/39 [00:00<00:02, 13.09it/s]
+
 36%|███▌      | 14/39 [00:01<00:01, 12.79it/s]
+
 41%|████      | 16/39 [00:01<00:01, 12.61it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 12.51it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 12.42it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 12.34it/s]
+
 62%|██████▏   | 24/39 [00:01<00:01, 12.31it/s]
+
 67%|██████▋   | 26/39 [00:01<00:01, 12.28it/s]
+
 72%|███████▏  | 28/39 [00:02<00:00, 12.25it/s]
+
 77%|███████▋  | 30/39 [00:02<00:00, 12.24it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 12.24it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 12.21it/s]
+
 92%|█████████▏| 36/39 [00:02<00:00, 12.24it/s]
+
 97%|█████████▋| 38/39 [00:02<00:00, 12.24it/s]
                                                   
+
                                               
{'eval_loss': '7.435', 'eval_runtime': '27.7', 'eval_samples_per_second': '88.13', 'eval_steps_per_second': '1.408', 'epoch': '0.5044'}
+
  1%|          | 72/7150 [01:54<1:16:53,  1.53it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-72
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-72/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-72/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.71it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.71it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-72/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-72/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-72/tokenizer_config.json
+[eval] epoch 0.504 step 72 loss 7.4350 (best 7.4350 @ ep 0.504, 0/20 since best)
+
+
                                               
  1%|          | 73/7150 [01:57<18:46:08,  9.55s/it]
  1%|          | 74/7150 [01:58<13:31:22,  6.88s/it]
  1%|          | 75/7150 [01:58<9:51:01,  5.01s/it] 
  1%|          | 76/7150 [01:59<7:16:45,  3.70s/it]
  1%|          | 77/7150 [02:00<5:28:45,  2.79s/it]
  1%|          | 78/7150 [02:00<4:13:13,  2.15s/it]
  1%|          | 79/7150 [02:01<3:20:18,  1.70s/it]
  1%|          | 80/7150 [02:02<2:43:15,  1.39s/it]
                                                   
{'loss': '7.373', 'grad_norm': '0.4253', 'learning_rate': '0.000474', 'epoch': '0.5604'}
+
  1%|          | 80/7150 [02:02<2:43:15,  1.39s/it]
  1%|          | 81/7150 [02:02<2:17:24,  1.17s/it]
  1%|          | 82/7150 [02:03<1:59:15,  1.01s/it]
  1%|          | 83/7150 [02:04<1:46:34,  1.11it/s]
  1%|          | 84/7150 [02:04<1:37:40,  1.21it/s]
  1%|          | 85/7150 [02:05<1:31:26,  1.29it/s]
  1%|          | 86/7150 [02:06<1:27:06,  1.35it/s]
  1%|          | 87/7150 [02:06<1:24:01,  1.40it/s]
  1%|          | 88/7150 [02:07<1:21:54,  1.44it/s]
  1%|          | 89/7150 [02:08<1:20:24,  1.46it/s]
  1%|▏         | 90/7150 [02:08<1:19:19,  1.48it/s]
                                                   
{'loss': '7.215', 'grad_norm': '0.8952', 'learning_rate': '0.000534', 'epoch': '0.6305'}
+
  1%|▏         | 90/7150 [02:08<1:19:19,  1.48it/s]
  1%|▏         | 91/7150 [02:09<1:18:39,  1.50it/s]
  1%|▏         | 92/7150 [02:10<1:18:08,  1.51it/s]
  1%|▏         | 93/7150 [02:10<1:17:47,  1.51it/s]
  1%|▏         | 94/7150 [02:11<1:17:31,  1.52it/s]
  1%|▏         | 95/7150 [02:11<1:17:21,  1.52it/s]
  1%|▏         | 96/7150 [02:12<1:17:10,  1.52it/s]
  1%|▏         | 97/7150 [02:13<1:17:04,  1.53it/s]
  1%|▏         | 98/7150 [02:13<1:17:05,  1.52it/s]
  1%|▏         | 99/7150 [02:14<1:16:58,  1.53it/s]
  1%|▏         | 100/7150 [02:15<1:16:57,  1.53it/s]
                                                    
{'loss': '7.055', 'grad_norm': '0.5406', 'learning_rate': '0.000594', 'epoch': '0.7005'}
+
  1%|▏         | 100/7150 [02:15<1:16:57,  1.53it/s]
  1%|▏         | 101/7150 [02:15<1:16:56,  1.53it/s]
  1%|▏         | 102/7150 [02:16<1:16:53,  1.53it/s]
  1%|▏         | 103/7150 [02:17<1:16:51,  1.53it/s]
  1%|▏         | 104/7150 [02:17<1:16:46,  1.53it/s]
  1%|▏         | 105/7150 [02:18<1:16:47,  1.53it/s]
  1%|▏         | 106/7150 [02:19<1:16:45,  1.53it/s]
  1%|▏         | 107/7150 [02:19<1:16:41,  1.53it/s]
  2%|▏         | 108/7150 [02:20<1:16:41,  1.53it/s]
  2%|▏         | 109/7150 [02:21<1:16:39,  1.53it/s]
  2%|▏         | 110/7150 [02:21<1:16:38,  1.53it/s]
                                                    
{'loss': '6.897', 'grad_norm': '0.4531', 'learning_rate': '0.0006', 'epoch': '0.7706'}
+
  2%|▏         | 110/7150 [02:21<1:16:38,  1.53it/s]
  2%|▏         | 111/7150 [02:22<1:16:40,  1.53it/s]
  2%|▏         | 112/7150 [02:23<1:16:39,  1.53it/s]
  2%|▏         | 113/7150 [02:23<1:16:38,  1.53it/s]
  2%|▏         | 114/7150 [02:24<1:16:39,  1.53it/s]
  2%|▏         | 115/7150 [02:25<1:16:39,  1.53it/s]
  2%|▏         | 116/7150 [02:25<1:16:35,  1.53it/s]
  2%|▏         | 117/7150 [02:26<1:16:36,  1.53it/s]
  2%|▏         | 118/7150 [02:27<1:16:36,  1.53it/s]
  2%|▏         | 119/7150 [02:27<1:16:35,  1.53it/s]
  2%|▏         | 120/7150 [02:28<1:16:39,  1.53it/s]
                                                    
{'loss': '6.764', 'grad_norm': '0.3535', 'learning_rate': '0.0006', 'epoch': '0.8406'}
+
  2%|▏         | 120/7150 [02:28<1:16:39,  1.53it/s]
  2%|▏         | 121/7150 [02:28<1:16:37,  1.53it/s]
  2%|▏         | 122/7150 [02:29<1:16:35,  1.53it/s]
  2%|▏         | 123/7150 [02:30<1:16:34,  1.53it/s]
  2%|▏         | 124/7150 [02:30<1:16:33,  1.53it/s]
  2%|▏         | 125/7150 [02:31<1:16:32,  1.53it/s]
  2%|▏         | 126/7150 [02:32<1:16:29,  1.53it/s]
  2%|▏         | 127/7150 [02:32<1:16:29,  1.53it/s]
  2%|▏         | 128/7150 [02:33<1:16:27,  1.53it/s]
  2%|▏         | 129/7150 [02:34<1:16:28,  1.53it/s]
  2%|▏         | 130/7150 [02:34<1:16:31,  1.53it/s]
                                                    
{'loss': '6.65', 'grad_norm': '0.7725', 'learning_rate': '0.0006', 'epoch': '0.9107'}
+
  2%|▏         | 130/7150 [02:34<1:16:31,  1.53it/s]
  2%|▏         | 131/7150 [02:35<1:16:31,  1.53it/s]
  2%|▏         | 132/7150 [02:36<1:16:29,  1.53it/s]
  2%|▏         | 133/7150 [02:36<1:16:31,  1.53it/s]
  2%|▏         | 134/7150 [02:37<1:16:30,  1.53it/s]
  2%|▏         | 135/7150 [02:38<1:16:28,  1.53it/s]
  2%|▏         | 136/7150 [02:38<1:16:26,  1.53it/s]
  2%|▏         | 137/7150 [02:39<1:16:27,  1.53it/s]
  2%|▏         | 138/7150 [02:40<1:16:27,  1.53it/s]
  2%|▏         | 139/7150 [02:40<1:16:25,  1.53it/s]
  2%|▏         | 140/7150 [02:41<1:16:27,  1.53it/s]
                                                    
{'loss': '6.573', 'grad_norm': '0.6359', 'learning_rate': '0.0006', 'epoch': '0.9807'}
+
  2%|▏         | 140/7150 [02:41<1:16:27,  1.53it/s]
  2%|▏         | 141/7150 [02:42<1:16:40,  1.52it/s]
  2%|▏         | 142/7150 [02:42<1:16:28,  1.53it/s]
  2%|▏         | 143/7150 [02:43<1:10:45,  1.65it/s]
  2%|▏         | 144/7150 [02:43<1:14:49,  1.56it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.23it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.69it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.58it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.03it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.74it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.55it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.46it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.36it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.33it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.27it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.29it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.27it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.25it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.25it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.24it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.23it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.23it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.23it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.22it/s]
                                                    
+
                                               
{'eval_loss': '6.504', 'eval_runtime': '3.264', 'eval_samples_per_second': '748', 'eval_steps_per_second': '11.95', 'epoch': '1.007'}
+
  2%|▏         | 144/7150 [02:47<1:14:49,  1.56it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.22it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-144
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-144/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-144/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.90it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.90it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-144/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-144/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-144/tokenizer_config.json
+[eval] epoch 1.007 step 144 loss 6.5045 (best 6.5045 @ ep 1.007, 0/20 since best)
+
+
                                               
  2%|▏         | 145/7150 [02:49<4:20:36,  2.23s/it]
  2%|▏         | 146/7150 [02:50<3:25:14,  1.76s/it]
  2%|▏         | 147/7150 [02:51<2:46:32,  1.43s/it]
  2%|▏         | 148/7150 [02:51<2:19:26,  1.19s/it]
  2%|▏         | 149/7150 [02:52<2:00:26,  1.03s/it]
  2%|▏         | 150/7150 [02:53<1:47:10,  1.09it/s]
                                                    
{'loss': '6.474', 'grad_norm': '0.6234', 'learning_rate': '0.0006', 'epoch': '1.049'}
+
  2%|▏         | 150/7150 [02:53<1:47:10,  1.09it/s]
  2%|▏         | 151/7150 [02:53<1:37:57,  1.19it/s]
  2%|▏         | 152/7150 [02:54<1:31:25,  1.28it/s]
  2%|▏         | 153/7150 [02:55<1:26:54,  1.34it/s]
  2%|▏         | 154/7150 [02:55<1:23:42,  1.39it/s]
  2%|▏         | 155/7150 [02:56<1:21:26,  1.43it/s]
  2%|▏         | 156/7150 [02:57<1:19:50,  1.46it/s]
  2%|▏         | 157/7150 [02:57<1:18:46,  1.48it/s]
  2%|▏         | 158/7150 [02:58<1:17:59,  1.49it/s]
  2%|▏         | 159/7150 [02:59<1:17:26,  1.50it/s]
  2%|▏         | 160/7150 [02:59<1:17:04,  1.51it/s]
                                                    
{'loss': '6.415', 'grad_norm': '1.13', 'learning_rate': '0.0006', 'epoch': '1.119'}
+
  2%|▏         | 160/7150 [02:59<1:17:04,  1.51it/s]
  2%|▏         | 161/7150 [03:00<1:16:47,  1.52it/s]
  2%|▏         | 162/7150 [03:00<1:16:35,  1.52it/s]
  2%|▏         | 163/7150 [03:01<1:16:27,  1.52it/s]
  2%|▏         | 164/7150 [03:02<1:16:18,  1.53it/s]
  2%|▏         | 165/7150 [03:02<1:16:14,  1.53it/s]
  2%|▏         | 166/7150 [03:03<1:16:11,  1.53it/s]
  2%|▏         | 167/7150 [03:04<1:16:09,  1.53it/s]
  2%|▏         | 168/7150 [03:04<1:16:07,  1.53it/s]
  2%|▏         | 169/7150 [03:05<1:16:04,  1.53it/s]
  2%|▏         | 170/7150 [03:06<1:16:02,  1.53it/s]
                                                    
{'loss': '6.332', 'grad_norm': '0.382', 'learning_rate': '0.0006', 'epoch': '1.189'}
+
  2%|▏         | 170/7150 [03:06<1:16:02,  1.53it/s]
  2%|▏         | 171/7150 [03:06<1:16:02,  1.53it/s]
  2%|▏         | 172/7150 [03:07<1:16:00,  1.53it/s]
  2%|▏         | 173/7150 [03:08<1:15:57,  1.53it/s]
  2%|▏         | 174/7150 [03:08<1:15:57,  1.53it/s]
  2%|▏         | 175/7150 [03:09<1:15:57,  1.53it/s]
  2%|▏         | 176/7150 [03:10<1:15:56,  1.53it/s]
  2%|▏         | 177/7150 [03:10<1:15:56,  1.53it/s]
  2%|▏         | 178/7150 [03:11<1:15:55,  1.53it/s]
  3%|▎         | 179/7150 [03:12<1:15:54,  1.53it/s]
  3%|▎         | 180/7150 [03:12<1:15:54,  1.53it/s]
                                                    
{'loss': '6.273', 'grad_norm': '1.048', 'learning_rate': '0.0006', 'epoch': '1.259'}
+
  3%|▎         | 180/7150 [03:12<1:15:54,  1.53it/s]
  3%|▎         | 181/7150 [03:13<1:15:54,  1.53it/s]
  3%|▎         | 182/7150 [03:14<1:15:55,  1.53it/s]
  3%|▎         | 183/7150 [03:14<1:15:56,  1.53it/s]
  3%|▎         | 184/7150 [03:15<1:15:56,  1.53it/s]
  3%|▎         | 185/7150 [03:16<1:15:53,  1.53it/s]
  3%|▎         | 186/7150 [03:16<1:15:50,  1.53it/s]
  3%|▎         | 187/7150 [03:17<1:15:50,  1.53it/s]
  3%|▎         | 188/7150 [03:17<1:15:52,  1.53it/s]
  3%|▎         | 189/7150 [03:18<1:15:51,  1.53it/s]
  3%|▎         | 190/7150 [03:19<1:15:47,  1.53it/s]
                                                    
{'loss': '6.268', 'grad_norm': '0.6026', 'learning_rate': '0.0006', 'epoch': '1.329'}
+
  3%|▎         | 190/7150 [03:19<1:15:47,  1.53it/s]
  3%|▎         | 191/7150 [03:19<1:15:50,  1.53it/s]
  3%|▎         | 192/7150 [03:20<1:15:47,  1.53it/s]
  3%|▎         | 193/7150 [03:21<1:15:46,  1.53it/s]
  3%|▎         | 194/7150 [03:21<1:15:47,  1.53it/s]
  3%|▎         | 195/7150 [03:22<1:15:45,  1.53it/s]
  3%|▎         | 196/7150 [03:23<1:15:45,  1.53it/s]
  3%|▎         | 197/7150 [03:23<1:15:45,  1.53it/s]
  3%|▎         | 198/7150 [03:24<1:15:44,  1.53it/s]
  3%|▎         | 199/7150 [03:25<1:15:42,  1.53it/s]
  3%|▎         | 200/7150 [03:25<1:15:43,  1.53it/s]
                                                    
{'loss': '6.164', 'grad_norm': '1.304', 'learning_rate': '0.0006', 'epoch': '1.399'}
+
  3%|▎         | 200/7150 [03:25<1:15:43,  1.53it/s]
  3%|▎         | 201/7150 [03:26<1:15:44,  1.53it/s]
  3%|▎         | 202/7150 [03:27<1:15:43,  1.53it/s]
  3%|▎         | 203/7150 [03:27<1:15:46,  1.53it/s]
  3%|▎         | 204/7150 [03:28<1:15:45,  1.53it/s]
  3%|▎         | 205/7150 [03:29<1:15:42,  1.53it/s]
  3%|▎         | 206/7150 [03:29<1:15:42,  1.53it/s]
  3%|▎         | 207/7150 [03:30<1:15:42,  1.53it/s]
  3%|▎         | 208/7150 [03:31<1:15:41,  1.53it/s]
  3%|▎         | 209/7150 [03:31<1:15:37,  1.53it/s]
  3%|▎         | 210/7150 [03:32<1:15:39,  1.53it/s]
                                                    
{'loss': '6.097', 'grad_norm': '0.4145', 'learning_rate': '0.0006', 'epoch': '1.469'}
+
  3%|▎         | 210/7150 [03:32<1:15:39,  1.53it/s]
  3%|▎         | 211/7150 [03:33<1:15:41,  1.53it/s]
  3%|▎         | 212/7150 [03:33<1:15:38,  1.53it/s]
  3%|▎         | 213/7150 [03:34<1:15:38,  1.53it/s]
  3%|▎         | 214/7150 [03:34<1:15:35,  1.53it/s]
  3%|▎         | 215/7150 [03:35<1:15:36,  1.53it/s]
  3%|▎         | 216/7150 [03:36<1:15:32,  1.53it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.98it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.90it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.68it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.08it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.77it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.59it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.44it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.37it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.32it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.30it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.29it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.27it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.26it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.25it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.25it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.25it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.24it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.22it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s]
                                                    
+
                                               
{'eval_loss': '6.078', 'eval_runtime': '3.237', 'eval_samples_per_second': '754', 'eval_steps_per_second': '12.05', 'epoch': '1.511'}
+
  3%|▎         | 216/7150 [03:39<1:15:32,  1.53it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-216
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-216/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-216/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.94it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.94it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-216/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-216/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-216/tokenizer_config.json
+[eval] epoch 1.511 step 216 loss 6.0779 (best 6.0779 @ ep 1.511, 0/20 since best)
+
+
                                               
  3%|▎         | 217/7150 [03:42<4:17:16,  2.23s/it]
  3%|▎         | 218/7150 [03:42<3:22:47,  1.76s/it]
  3%|▎         | 219/7150 [03:43<2:44:32,  1.42s/it]
  3%|▎         | 220/7150 [03:44<2:17:47,  1.19s/it]
                                                    
{'loss': '6.051', 'grad_norm': '1.279', 'learning_rate': '0.0006', 'epoch': '1.539'}
+
  3%|▎         | 220/7150 [03:44<2:17:47,  1.19s/it]
  3%|▎         | 221/7150 [03:44<1:59:07,  1.03s/it]
  3%|▎         | 222/7150 [03:45<1:46:00,  1.09it/s]
  3%|▎         | 223/7150 [03:46<1:36:51,  1.19it/s]
  3%|▎         | 224/7150 [03:46<1:30:24,  1.28it/s]
  3%|▎         | 225/7150 [03:47<1:25:51,  1.34it/s]
  3%|▎         | 226/7150 [03:48<1:22:43,  1.39it/s]
  3%|▎         | 227/7150 [03:48<1:20:31,  1.43it/s]
  3%|▎         | 228/7150 [03:49<1:18:57,  1.46it/s]
  3%|▎         | 229/7150 [03:50<1:17:52,  1.48it/s]
  3%|▎         | 230/7150 [03:50<1:17:07,  1.50it/s]
                                                    
{'loss': '5.994', 'grad_norm': '0.6056', 'learning_rate': '0.0006', 'epoch': '1.609'}
+
  3%|▎         | 230/7150 [03:50<1:17:07,  1.50it/s]
  3%|▎         | 231/7150 [03:51<1:16:35,  1.51it/s]
  3%|▎         | 232/7150 [03:51<1:16:13,  1.51it/s]
  3%|▎         | 233/7150 [03:52<1:15:56,  1.52it/s]
  3%|▎         | 234/7150 [03:53<1:15:51,  1.52it/s]
  3%|▎         | 235/7150 [03:53<1:15:47,  1.52it/s]
  3%|▎         | 236/7150 [03:54<1:15:43,  1.52it/s]
  3%|▎         | 237/7150 [03:55<1:15:38,  1.52it/s]
  3%|▎         | 238/7150 [03:55<1:15:33,  1.52it/s]
  3%|▎         | 239/7150 [03:56<1:15:23,  1.53it/s]
  3%|▎         | 240/7150 [03:57<1:15:24,  1.53it/s]
                                                    
{'loss': '5.948', 'grad_norm': '0.5148', 'learning_rate': '0.0006', 'epoch': '1.68'}
+
  3%|▎         | 240/7150 [03:57<1:15:24,  1.53it/s]
  3%|▎         | 241/7150 [03:57<1:15:23,  1.53it/s]
  3%|▎         | 242/7150 [03:58<1:15:22,  1.53it/s]
  3%|▎         | 243/7150 [03:59<1:15:17,  1.53it/s]
  3%|▎         | 244/7150 [03:59<1:15:17,  1.53it/s]
  3%|▎         | 245/7150 [04:00<1:15:23,  1.53it/s]
  3%|▎         | 246/7150 [04:01<1:15:24,  1.53it/s]
  3%|▎         | 247/7150 [04:01<1:15:23,  1.53it/s]
  3%|▎         | 248/7150 [04:02<1:15:21,  1.53it/s]
  3%|▎         | 249/7150 [04:03<1:15:23,  1.53it/s]
  3%|▎         | 250/7150 [04:03<1:16:10,  1.51it/s]
                                                    
{'loss': '5.884', 'grad_norm': '0.3119', 'learning_rate': '0.0006', 'epoch': '1.75'}
+
  3%|▎         | 250/7150 [04:03<1:16:10,  1.51it/s]
  4%|▎         | 251/7150 [04:04<1:16:16,  1.51it/s]
  4%|▎         | 252/7150 [04:05<1:16:02,  1.51it/s]
  4%|▎         | 253/7150 [04:05<1:15:51,  1.52it/s]
  4%|▎         | 254/7150 [04:06<1:16:07,  1.51it/s]
  4%|▎         | 255/7150 [04:07<1:19:46,  1.44it/s]
  4%|▎         | 256/7150 [04:07<1:19:26,  1.45it/s]
  4%|▎         | 257/7150 [04:08<1:21:10,  1.42it/s]
  4%|▎         | 258/7150 [04:09<1:19:24,  1.45it/s]
  4%|▎         | 259/7150 [04:10<1:22:02,  1.40it/s]
  4%|▎         | 260/7150 [04:10<1:19:59,  1.44it/s]
                                                    
{'loss': '5.829', 'grad_norm': '0.3651', 'learning_rate': '0.0006', 'epoch': '1.82'}
+
  4%|▎         | 260/7150 [04:10<1:19:59,  1.44it/s]
  4%|▎         | 261/7150 [04:11<1:22:46,  1.39it/s]
  4%|▎         | 262/7150 [04:12<1:21:08,  1.41it/s]
  4%|▎         | 263/7150 [04:13<1:34:45,  1.21it/s]
  4%|▎         | 264/7150 [04:14<1:43:33,  1.11it/s]
  4%|▎         | 265/7150 [04:15<1:34:58,  1.21it/s]
  4%|▎         | 266/7150 [04:15<1:33:10,  1.23it/s]
  4%|▎         | 267/7150 [04:16<1:27:43,  1.31it/s]
  4%|▎         | 268/7150 [04:17<1:27:34,  1.31it/s]
  4%|▍         | 269/7150 [04:17<1:27:08,  1.32it/s]
  4%|▍         | 270/7150 [04:18<1:23:54,  1.37it/s]
                                                    
{'loss': '5.785', 'grad_norm': '1.016', 'learning_rate': '0.0006', 'epoch': '1.89'}
+
  4%|▍         | 270/7150 [04:18<1:23:54,  1.37it/s]
  4%|▍         | 271/7150 [04:19<1:26:20,  1.33it/s]
  4%|▍         | 272/7150 [04:20<1:23:40,  1.37it/s]
  4%|▍         | 273/7150 [04:20<1:23:42,  1.37it/s]
  4%|▍         | 274/7150 [04:21<1:25:03,  1.35it/s]
  4%|▍         | 275/7150 [04:22<1:22:01,  1.40it/s]
  4%|▍         | 276/7150 [04:23<1:26:36,  1.32it/s]
  4%|▍         | 277/7150 [04:23<1:23:22,  1.37it/s]
  4%|▍         | 278/7150 [04:24<1:30:18,  1.27it/s]
  4%|▍         | 279/7150 [04:25<1:34:42,  1.21it/s]
  4%|▍         | 280/7150 [04:26<1:30:15,  1.27it/s]
                                                    
{'loss': '5.737', 'grad_norm': '0.4', 'learning_rate': '0.0006', 'epoch': '1.96'}
+
  4%|▍         | 280/7150 [04:26<1:30:15,  1.27it/s]
  4%|▍         | 281/7150 [04:27<1:40:50,  1.14it/s]
  4%|▍         | 282/7150 [04:28<1:39:31,  1.15it/s]
  4%|▍         | 283/7150 [04:28<1:35:05,  1.20it/s]
  4%|▍         | 284/7150 [04:29<1:34:45,  1.21it/s]
  4%|▍         | 285/7150 [04:30<1:29:16,  1.28it/s]
  4%|▍         | 286/7150 [04:31<1:27:09,  1.31it/s]
  4%|▍         | 287/7150 [04:31<1:26:47,  1.32it/s]
  4%|▍         | 288/7150 [04:32<1:34:51,  1.21it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 19.79it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 15.20it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.78it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 10.42it/s]
+
 28%|██▊       | 11/39 [00:01<00:03,  9.31it/s]
+
 33%|███▎      | 13/39 [00:01<00:02,  9.81it/s]
+
 38%|███▊      | 15/39 [00:01<00:02, 10.16it/s]
+
 44%|████▎     | 17/39 [00:01<00:02, 10.74it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 11.13it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 11.46it/s]
+
 59%|█████▉    | 23/39 [00:02<00:01, 11.69it/s]
+
 64%|██████▍   | 25/39 [00:02<00:01, 11.02it/s]
+
 69%|██████▉   | 27/39 [00:02<00:01,  8.49it/s]
+
 74%|███████▍  | 29/39 [00:02<00:01,  8.99it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00,  9.51it/s]
+
 85%|████████▍ | 33/39 [00:03<00:00,  9.91it/s]
+
 90%|████████▉ | 35/39 [00:03<00:00, 10.51it/s]
+
 95%|█████████▍| 37/39 [00:03<00:00, 10.97it/s]
+
100%|██████████| 39/39 [00:03<00:00, 11.32it/s]
                                                    
+
                                               
{'eval_loss': '5.693', 'eval_runtime': '3.78', 'eval_samples_per_second': '645.8', 'eval_steps_per_second': '10.32', 'epoch': '2.014'}
+
  4%|▍         | 288/7150 [04:36<1:34:51,  1.21it/s]
+
100%|██████████| 39/39 [00:03<00:00, 11.32it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-288
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-288/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-288/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.96it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.96it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-288/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-288/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-288/tokenizer_config.json
+[eval] epoch 2.014 step 288 loss 5.6931 (best 5.6931 @ ep 2.014, 0/20 since best)
+
+
                                               
  4%|▍         | 289/7150 [04:39<4:56:29,  2.59s/it]
  4%|▍         | 290/7150 [04:40<4:00:05,  2.10s/it]
                                                    
{'loss': '5.675', 'grad_norm': '0.4628', 'learning_rate': '0.0006', 'epoch': '2.028'}
+
  4%|▍         | 290/7150 [04:40<4:00:05,  2.10s/it]
  4%|▍         | 291/7150 [04:41<3:12:14,  1.68s/it]
  4%|▍         | 292/7150 [04:42<2:52:32,  1.51s/it]
  4%|▍         | 293/7150 [04:43<2:23:44,  1.26s/it]
  4%|▍         | 294/7150 [04:43<2:11:03,  1.15s/it]
  4%|▍         | 295/7150 [04:44<2:05:10,  1.10s/it]
  4%|▍         | 296/7150 [04:45<1:49:58,  1.04it/s]
  4%|▍         | 297/7150 [04:46<1:47:54,  1.06it/s]
  4%|▍         | 298/7150 [04:47<1:38:00,  1.17it/s]
  4%|▍         | 299/7150 [04:48<1:38:54,  1.15it/s]
  4%|▍         | 300/7150 [04:48<1:33:02,  1.23it/s]
                                                    
{'loss': '5.59', 'grad_norm': '0.4886', 'learning_rate': '0.0006', 'epoch': '2.098'}
+
  4%|▍         | 300/7150 [04:48<1:33:02,  1.23it/s]
  4%|▍         | 301/7150 [04:49<1:35:49,  1.19it/s]
  4%|▍         | 302/7150 [04:50<1:29:49,  1.27it/s]
  4%|▍         | 303/7150 [04:51<1:42:45,  1.11it/s]
  4%|▍         | 304/7150 [04:52<1:45:11,  1.08it/s]
  4%|▍         | 305/7150 [04:53<1:36:05,  1.19it/s]
  4%|▍         | 306/7150 [04:53<1:37:15,  1.17it/s]
  4%|▍         | 307/7150 [04:54<1:30:26,  1.26it/s]
  4%|▍         | 308/7150 [04:55<1:34:25,  1.21it/s]
  4%|▍         | 309/7150 [04:56<1:36:56,  1.18it/s]
  4%|▍         | 310/7150 [04:57<1:31:38,  1.24it/s]
                                                    
{'loss': '5.591', 'grad_norm': '0.5552', 'learning_rate': '0.0006', 'epoch': '2.168'}
+
  4%|▍         | 310/7150 [04:57<1:31:38,  1.24it/s]
  4%|▍         | 311/7150 [04:58<1:34:08,  1.21it/s]
  4%|▍         | 312/7150 [04:59<1:52:06,  1.02it/s]
  4%|▍         | 313/7150 [05:00<1:55:35,  1.01s/it]
  4%|▍         | 314/7150 [05:01<1:52:14,  1.02it/s]
  4%|▍         | 315/7150 [05:02<1:41:30,  1.12it/s]
  4%|▍         | 316/7150 [05:02<1:41:50,  1.12it/s]
  4%|▍         | 317/7150 [05:03<1:34:09,  1.21it/s]
  4%|▍         | 318/7150 [05:04<1:36:21,  1.18it/s]
  4%|▍         | 319/7150 [05:05<1:30:17,  1.26it/s]
  4%|▍         | 320/7150 [05:06<1:35:02,  1.20it/s]
                                                    
{'loss': '5.516', 'grad_norm': '0.3206', 'learning_rate': '0.0006', 'epoch': '2.238'}
+
  4%|▍         | 320/7150 [05:06<1:35:02,  1.20it/s]
  4%|▍         | 321/7150 [05:06<1:29:27,  1.27it/s]
  5%|▍         | 322/7150 [05:07<1:34:35,  1.20it/s]
  5%|▍         | 323/7150 [05:08<1:29:10,  1.28it/s]
  5%|▍         | 324/7150 [05:09<1:34:27,  1.20it/s]
  5%|▍         | 325/7150 [05:09<1:28:29,  1.29it/s]
  5%|▍         | 326/7150 [05:10<1:32:11,  1.23it/s]
  5%|▍         | 327/7150 [05:11<1:32:02,  1.24it/s]
  5%|▍         | 328/7150 [05:12<1:34:01,  1.21it/s]
  5%|▍         | 329/7150 [05:13<1:28:48,  1.28it/s]
  5%|▍         | 330/7150 [05:14<1:32:09,  1.23it/s]
                                                    
{'loss': '5.463', 'grad_norm': '0.6817', 'learning_rate': '0.0006', 'epoch': '2.308'}
+
  5%|▍         | 330/7150 [05:14<1:32:09,  1.23it/s]
  5%|▍         | 331/7150 [05:15<1:37:02,  1.17it/s]
  5%|▍         | 332/7150 [05:15<1:39:29,  1.14it/s]
  5%|▍         | 333/7150 [05:16<1:31:56,  1.24it/s]
  5%|▍         | 334/7150 [05:17<1:36:28,  1.18it/s]
  5%|▍         | 335/7150 [05:18<1:29:51,  1.26it/s]
  5%|▍         | 336/7150 [05:19<1:37:37,  1.16it/s]
  5%|▍         | 337/7150 [05:19<1:31:41,  1.24it/s]
  5%|▍         | 338/7150 [05:20<1:34:13,  1.20it/s]
  5%|▍         | 339/7150 [05:21<1:38:53,  1.15it/s]
  5%|▍         | 340/7150 [05:22<1:31:57,  1.23it/s]
                                                    
{'loss': '5.458', 'grad_norm': '0.7824', 'learning_rate': '0.0006', 'epoch': '2.378'}
+
  5%|▍         | 340/7150 [05:22<1:31:57,  1.23it/s]
  5%|▍         | 341/7150 [05:23<1:36:32,  1.18it/s]
  5%|▍         | 342/7150 [05:24<1:33:48,  1.21it/s]
  5%|▍         | 343/7150 [05:25<1:34:45,  1.20it/s]
  5%|▍         | 344/7150 [05:25<1:38:50,  1.15it/s]
  5%|▍         | 345/7150 [05:26<1:31:59,  1.23it/s]
  5%|▍         | 346/7150 [05:27<1:36:59,  1.17it/s]
  5%|▍         | 347/7150 [05:28<1:30:08,  1.26it/s]
  5%|▍         | 348/7150 [05:29<1:35:37,  1.19it/s]
  5%|▍         | 349/7150 [05:29<1:30:25,  1.25it/s]
  5%|▍         | 350/7150 [05:30<1:36:53,  1.17it/s]
                                                    
{'loss': '5.38', 'grad_norm': '0.3001', 'learning_rate': '0.0006', 'epoch': '2.448'}
+
  5%|▍         | 350/7150 [05:30<1:36:53,  1.17it/s]
  5%|▍         | 351/7150 [05:31<1:38:56,  1.15it/s]
  5%|▍         | 352/7150 [05:32<1:34:01,  1.20it/s]
  5%|▍         | 353/7150 [05:33<1:33:38,  1.21it/s]
  5%|▍         | 354/7150 [05:34<1:33:42,  1.21it/s]
  5%|▍         | 355/7150 [05:34<1:29:33,  1.26it/s]
  5%|▍         | 356/7150 [05:35<1:34:24,  1.20it/s]
  5%|▍         | 357/7150 [05:36<1:28:15,  1.28it/s]
  5%|▌         | 358/7150 [05:37<1:36:04,  1.18it/s]
  5%|▌         | 359/7150 [05:38<1:35:47,  1.18it/s]
  5%|▌         | 360/7150 [05:39<1:34:01,  1.20it/s]
                                                    
{'loss': '5.307', 'grad_norm': '0.4977', 'learning_rate': '0.0006', 'epoch': '2.518'}
+
  5%|▌         | 360/7150 [05:39<1:34:01,  1.20it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.45it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.78it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 12.66it/s]
+
 23%|██▎       | 9/39 [00:00<00:03,  8.50it/s]
+
 26%|██▌       | 10/39 [00:01<00:03,  7.66it/s]
+
 31%|███       | 12/39 [00:01<00:03,  8.56it/s]
+
 36%|███▌      | 14/39 [00:01<00:02,  9.31it/s]
+
 38%|███▊      | 15/39 [00:01<00:02,  9.41it/s]
+
 44%|████▎     | 17/39 [00:01<00:02, 10.27it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 10.86it/s]
+
 54%|█████▍    | 21/39 [00:02<00:01, 11.27it/s]
+
 59%|█████▉    | 23/39 [00:02<00:01, 10.37it/s]
+
 64%|██████▍   | 25/39 [00:02<00:01,  8.08it/s]
+
 67%|██████▋   | 26/39 [00:02<00:01,  8.31it/s]
+
 72%|███████▏  | 28/39 [00:02<00:01,  9.12it/s]
+
 77%|███████▋  | 30/39 [00:03<00:00,  9.50it/s]
+
 82%|████████▏ | 32/39 [00:03<00:00, 10.15it/s]
+
 87%|████████▋ | 34/39 [00:03<00:00, 10.73it/s]
+
 92%|█████████▏| 36/39 [00:03<00:00, 11.16it/s]
+
 97%|█████████▋| 38/39 [00:03<00:00, 11.47it/s]
                                                    
+
                                               
{'eval_loss': '5.355', 'eval_runtime': '3.985', 'eval_samples_per_second': '612.6', 'eval_steps_per_second': '9.787', 'epoch': '2.518'}
+
  5%|▌         | 360/7150 [05:43<1:34:01,  1.20it/s]
+
100%|██████████| 39/39 [00:03<00:00, 11.47it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-360
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-360/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-360/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.94it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.94it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-360/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-360/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-360/tokenizer_config.json
+[eval] epoch 2.518 step 360 loss 5.3552 (best 5.3552 @ ep 2.518, 0/20 since best)
+
+
                                               
  5%|▌         | 361/7150 [05:46<5:05:47,  2.70s/it]
  5%|▌         | 362/7150 [05:46<3:56:25,  2.09s/it]
  5%|▌         | 363/7150 [05:47<3:18:58,  1.76s/it]
  5%|▌         | 364/7150 [05:48<2:42:07,  1.43s/it]
  5%|▌         | 365/7150 [05:49<2:23:22,  1.27s/it]
  5%|▌         | 366/7150 [05:50<2:05:22,  1.11s/it]
  5%|▌         | 367/7150 [05:51<1:59:00,  1.05s/it]
  5%|▌         | 368/7150 [05:51<1:54:40,  1.01s/it]
  5%|▌         | 369/7150 [05:52<1:43:01,  1.10it/s]
  5%|▌         | 370/7150 [05:53<1:43:51,  1.09it/s]
                                                    
{'loss': '5.257', 'grad_norm': '1.124', 'learning_rate': '0.0006', 'epoch': '2.588'}
+
  5%|▌         | 370/7150 [05:53<1:43:51,  1.09it/s]
  5%|▌         | 371/7150 [05:54<1:35:25,  1.18it/s]
  5%|▌         | 372/7150 [05:55<1:37:21,  1.16it/s]
  5%|▌         | 373/7150 [05:55<1:35:46,  1.18it/s]
  5%|▌         | 374/7150 [05:56<1:33:28,  1.21it/s]
  5%|▌         | 375/7150 [05:57<1:31:25,  1.24it/s]
  5%|▌         | 376/7150 [05:58<1:31:41,  1.23it/s]
  5%|▌         | 377/7150 [05:59<1:37:43,  1.16it/s]
  5%|▌         | 378/7150 [06:00<1:31:40,  1.23it/s]
  5%|▌         | 379/7150 [06:00<1:30:41,  1.24it/s]
  5%|▌         | 380/7150 [06:01<1:42:31,  1.10it/s]
                                                    
{'loss': '5.276', 'grad_norm': '0.9687', 'learning_rate': '0.0006', 'epoch': '2.658'}
+
  5%|▌         | 380/7150 [06:01<1:42:31,  1.10it/s]
  5%|▌         | 381/7150 [06:02<1:34:56,  1.19it/s]
  5%|▌         | 382/7150 [06:03<1:47:32,  1.05it/s]
  5%|▌         | 383/7150 [06:04<1:42:04,  1.10it/s]
  5%|▌         | 384/7150 [06:05<1:33:35,  1.20it/s]
  5%|▌         | 385/7150 [06:06<1:50:22,  1.02it/s]
  5%|▌         | 386/7150 [06:07<1:40:39,  1.12it/s]
  5%|▌         | 387/7150 [06:08<1:44:58,  1.07it/s]
  5%|▌         | 388/7150 [06:09<1:45:48,  1.07it/s]
  5%|▌         | 389/7150 [06:10<1:39:35,  1.13it/s]
  5%|▌         | 390/7150 [06:11<1:49:40,  1.03it/s]
                                                    
{'loss': '5.202', 'grad_norm': '0.2934', 'learning_rate': '0.0006', 'epoch': '2.729'}
+
  5%|▌         | 390/7150 [06:11<1:49:40,  1.03it/s]
  5%|▌         | 391/7150 [06:11<1:40:02,  1.13it/s]
  5%|▌         | 392/7150 [06:13<1:51:59,  1.01it/s]
  5%|▌         | 393/7150 [06:13<1:43:14,  1.09it/s]
  6%|▌         | 394/7150 [06:14<1:35:20,  1.18it/s]
  6%|▌         | 395/7150 [06:15<1:50:41,  1.02it/s]
  6%|▌         | 396/7150 [06:16<1:40:30,  1.12it/s]
  6%|▌         | 397/7150 [06:17<1:39:52,  1.13it/s]
  6%|▌         | 398/7150 [06:18<1:46:44,  1.05it/s]
  6%|▌         | 399/7150 [06:19<1:37:31,  1.15it/s]
  6%|▌         | 400/7150 [06:20<1:51:23,  1.01it/s]
                                                    
{'loss': '5.114', 'grad_norm': '0.3448', 'learning_rate': '0.0006', 'epoch': '2.799'}
+
  6%|▌         | 400/7150 [06:20<1:51:23,  1.01it/s]
  6%|▌         | 401/7150 [06:21<1:41:07,  1.11it/s]
  6%|▌         | 402/7150 [06:22<1:48:24,  1.04it/s]
  6%|▌         | 403/7150 [06:23<1:45:17,  1.07it/s]
  6%|▌         | 404/7150 [06:23<1:41:37,  1.11it/s]
  6%|▌         | 405/7150 [06:25<1:51:53,  1.00it/s]
  6%|▌         | 406/7150 [06:25<1:40:33,  1.12it/s]
  6%|▌         | 407/7150 [06:26<1:32:31,  1.21it/s]
  6%|▌         | 408/7150 [06:27<1:26:53,  1.29it/s]
  6%|▌         | 409/7150 [06:27<1:22:53,  1.36it/s]
  6%|▌         | 410/7150 [06:28<1:20:05,  1.40it/s]
                                                    
{'loss': '5.107', 'grad_norm': '0.8026', 'learning_rate': '0.0006', 'epoch': '2.869'}
+
  6%|▌         | 410/7150 [06:28<1:20:05,  1.40it/s]
  6%|▌         | 411/7150 [06:29<1:18:09,  1.44it/s]
  6%|▌         | 412/7150 [06:29<1:16:47,  1.46it/s]
  6%|▌         | 413/7150 [06:30<1:15:53,  1.48it/s]
  6%|▌         | 414/7150 [06:31<1:15:18,  1.49it/s]
  6%|▌         | 415/7150 [06:31<1:14:49,  1.50it/s]
  6%|▌         | 416/7150 [06:32<1:14:30,  1.51it/s]
  6%|▌         | 417/7150 [06:33<1:14:12,  1.51it/s]
  6%|▌         | 418/7150 [06:33<1:14:00,  1.52it/s]
  6%|▌         | 419/7150 [06:34<1:13:56,  1.52it/s]
  6%|▌         | 420/7150 [06:35<1:13:51,  1.52it/s]
                                                    
{'loss': '5.073', 'grad_norm': '0.4198', 'learning_rate': '0.0006', 'epoch': '2.939'}
+
  6%|▌         | 420/7150 [06:35<1:13:51,  1.52it/s]
  6%|▌         | 421/7150 [06:35<1:13:53,  1.52it/s]
  6%|▌         | 422/7150 [06:36<1:13:59,  1.52it/s]
  6%|▌         | 423/7150 [06:37<1:14:01,  1.51it/s]
  6%|▌         | 424/7150 [06:37<1:13:56,  1.52it/s]
  6%|▌         | 425/7150 [06:38<1:13:53,  1.52it/s]
  6%|▌         | 426/7150 [06:39<1:13:45,  1.52it/s]
  6%|▌         | 427/7150 [06:39<1:13:48,  1.52it/s]
  6%|▌         | 428/7150 [06:40<1:13:39,  1.52it/s]
  6%|▌         | 429/7150 [06:40<1:08:06,  1.64it/s]
  6%|▌         | 430/7150 [06:41<1:11:56,  1.56it/s]
                                                    
{'loss': '4.987', 'grad_norm': '0.28', 'learning_rate': '0.0006', 'epoch': '3.007'}
+
  6%|▌         | 430/7150 [06:41<1:11:56,  1.56it/s]
  6%|▌         | 431/7150 [06:42<1:12:24,  1.55it/s]
  6%|▌         | 432/7150 [06:42<1:12:49,  1.54it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.34it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.68it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.59it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.02it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.73it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.56it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.47it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.39it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.34it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.29it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.29it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.27it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.25it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.26it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.25it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.25it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.25it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.24it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.25it/s]
                                                    
+
                                               
{'eval_loss': '5.031', 'eval_runtime': '3.261', 'eval_samples_per_second': '748.6', 'eval_steps_per_second': '11.96', 'epoch': '3.021'}
+
  6%|▌         | 432/7150 [06:46<1:12:49,  1.54it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.25it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-432
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-432/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-432/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.65it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.65it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-432/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-432/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-432/tokenizer_config.json
+[eval] epoch 3.021 step 432 loss 5.0313 (best 5.0313 @ ep 3.021, 0/20 since best)
+
+
                                               
  6%|▌         | 433/7150 [06:48<4:14:21,  2.27s/it]
  6%|▌         | 434/7150 [06:49<3:20:02,  1.79s/it]
  6%|▌         | 435/7150 [06:50<2:41:58,  1.45s/it]
  6%|▌         | 436/7150 [06:50<2:15:23,  1.21s/it]
  6%|▌         | 437/7150 [06:51<1:56:48,  1.04s/it]
  6%|▌         | 438/7150 [06:52<1:43:46,  1.08it/s]
  6%|▌         | 439/7150 [06:52<1:34:39,  1.18it/s]
  6%|▌         | 440/7150 [06:53<1:28:12,  1.27it/s]
                                                    
{'loss': '4.937', 'grad_norm': '0.7299', 'learning_rate': '0.0006', 'epoch': '3.077'}
+
  6%|▌         | 440/7150 [06:53<1:28:12,  1.27it/s]
  6%|▌         | 441/7150 [06:54<1:23:45,  1.34it/s]
  6%|▌         | 442/7150 [06:54<1:20:40,  1.39it/s]
  6%|▌         | 443/7150 [06:55<1:18:30,  1.42it/s]
  6%|▌         | 444/7150 [06:56<1:16:54,  1.45it/s]
  6%|▌         | 445/7150 [06:56<1:15:47,  1.47it/s]
  6%|▌         | 446/7150 [06:57<1:14:58,  1.49it/s]
  6%|▋         | 447/7150 [06:58<1:14:29,  1.50it/s]
  6%|▋         | 448/7150 [06:58<1:14:07,  1.51it/s]
  6%|▋         | 449/7150 [06:59<1:13:48,  1.51it/s]
  6%|▋         | 450/7150 [07:00<1:13:34,  1.52it/s]
                                                    
{'loss': '4.938', 'grad_norm': '0.7613', 'learning_rate': '0.0006', 'epoch': '3.147'}
+
  6%|▋         | 450/7150 [07:00<1:13:34,  1.52it/s]
  6%|▋         | 451/7150 [07:00<1:13:34,  1.52it/s]
  6%|▋         | 452/7150 [07:01<1:13:31,  1.52it/s]
  6%|▋         | 453/7150 [07:02<1:13:27,  1.52it/s]
  6%|▋         | 454/7150 [07:02<1:13:22,  1.52it/s]
  6%|▋         | 455/7150 [07:03<1:13:22,  1.52it/s]
  6%|▋         | 456/7150 [07:04<1:13:20,  1.52it/s]
  6%|▋         | 457/7150 [07:04<1:13:18,  1.52it/s]
  6%|▋         | 458/7150 [07:05<1:13:16,  1.52it/s]
  6%|▋         | 459/7150 [07:05<1:13:20,  1.52it/s]
  6%|▋         | 460/7150 [07:06<1:13:27,  1.52it/s]
                                                    
{'loss': '4.86', 'grad_norm': '0.3004', 'learning_rate': '0.0006', 'epoch': '3.217'}
+
  6%|▋         | 460/7150 [07:06<1:13:27,  1.52it/s]
  6%|▋         | 461/7150 [07:07<1:13:30,  1.52it/s]
  6%|▋         | 462/7150 [07:07<1:13:24,  1.52it/s]
  6%|▋         | 463/7150 [07:08<1:13:23,  1.52it/s]
  6%|▋         | 464/7150 [07:09<1:13:24,  1.52it/s]
  7%|▋         | 465/7150 [07:09<1:13:22,  1.52it/s]
  7%|▋         | 466/7150 [07:10<1:13:18,  1.52it/s]
  7%|▋         | 467/7150 [07:11<1:13:15,  1.52it/s]
  7%|▋         | 468/7150 [07:11<1:13:11,  1.52it/s]
  7%|▋         | 469/7150 [07:12<1:13:08,  1.52it/s]
  7%|▋         | 470/7150 [07:13<1:13:10,  1.52it/s]
                                                    
{'loss': '4.803', 'grad_norm': '0.3486', 'learning_rate': '0.0006', 'epoch': '3.287'}
+
  7%|▋         | 470/7150 [07:13<1:13:10,  1.52it/s]
  7%|▋         | 471/7150 [07:13<1:13:12,  1.52it/s]
  7%|▋         | 472/7150 [07:14<1:13:10,  1.52it/s]
  7%|▋         | 473/7150 [07:15<1:13:09,  1.52it/s]
  7%|▋         | 474/7150 [07:15<1:14:54,  1.49it/s]
  7%|▋         | 475/7150 [07:16<1:14:35,  1.49it/s]
  7%|▋         | 476/7150 [07:17<1:14:11,  1.50it/s]
  7%|▋         | 477/7150 [07:17<1:13:47,  1.51it/s]
  7%|▋         | 478/7150 [07:18<1:13:30,  1.51it/s]
  7%|▋         | 479/7150 [07:19<1:13:57,  1.50it/s]
  7%|▋         | 480/7150 [07:19<1:14:01,  1.50it/s]
                                                    
{'loss': '4.772', 'grad_norm': '0.6028', 'learning_rate': '0.0006', 'epoch': '3.357'}
+
  7%|▋         | 480/7150 [07:19<1:14:01,  1.50it/s]
  7%|▋         | 481/7150 [07:20<1:17:54,  1.43it/s]
  7%|▋         | 482/7150 [07:21<1:20:48,  1.38it/s]
  7%|▋         | 483/7150 [07:22<1:18:22,  1.42it/s]
  7%|▋         | 484/7150 [07:22<1:20:48,  1.37it/s]
  7%|▋         | 485/7150 [07:23<1:24:12,  1.32it/s]
  7%|▋         | 486/7150 [07:24<1:20:43,  1.38it/s]
  7%|▋         | 487/7150 [07:25<1:23:39,  1.33it/s]
  7%|▋         | 488/7150 [07:25<1:25:43,  1.30it/s]
  7%|▋         | 489/7150 [07:26<1:26:23,  1.29it/s]
  7%|▋         | 490/7150 [07:27<1:23:31,  1.33it/s]
                                                    
{'loss': '4.745', 'grad_norm': '0.3806', 'learning_rate': '0.0006', 'epoch': '3.427'}
+
  7%|▋         | 490/7150 [07:27<1:23:31,  1.33it/s]
  7%|▋         | 491/7150 [07:28<1:24:38,  1.31it/s]
  7%|▋         | 492/7150 [07:29<1:25:36,  1.30it/s]
  7%|▋         | 493/7150 [07:29<1:25:57,  1.29it/s]
  7%|▋         | 494/7150 [07:30<1:26:41,  1.28it/s]
  7%|▋         | 495/7150 [07:31<1:22:38,  1.34it/s]
  7%|▋         | 496/7150 [07:32<1:24:24,  1.31it/s]
  7%|▋         | 497/7150 [07:32<1:25:20,  1.30it/s]
  7%|▋         | 498/7150 [07:33<1:21:41,  1.36it/s]
  7%|▋         | 499/7150 [07:34<1:23:31,  1.33it/s]
  7%|▋         | 500/7150 [07:35<1:24:23,  1.31it/s]
                                                    
{'loss': '4.702', 'grad_norm': '0.607', 'learning_rate': '0.0006', 'epoch': '3.497'}
+
  7%|▋         | 500/7150 [07:35<1:24:23,  1.31it/s]
  7%|▋         | 501/7150 [07:35<1:20:58,  1.37it/s]
  7%|▋         | 502/7150 [07:36<1:22:52,  1.34it/s]
  7%|▋         | 503/7150 [07:37<1:24:00,  1.32it/s]
  7%|▋         | 504/7150 [07:38<1:20:31,  1.38it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:02, 16.40it/s]
+
 13%|█▎        | 5/39 [00:00<00:03, 10.22it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 10.69it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 11.21it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 11.53it/s]
+
 33%|███▎      | 13/39 [00:01<00:02, 11.76it/s]
+
 38%|███▊      | 15/39 [00:01<00:02, 11.88it/s]
+
 44%|████▎     | 17/39 [00:01<00:02, 10.57it/s]
+
 49%|████▊     | 19/39 [00:01<00:02,  9.55it/s]
+
 51%|█████▏    | 20/39 [00:01<00:02,  8.37it/s]
+
 54%|█████▍    | 21/39 [00:02<00:02,  8.48it/s]
+
 56%|█████▋    | 22/39 [00:02<00:01,  8.52it/s]
+
 59%|█████▉    | 23/39 [00:02<00:02,  7.53it/s]
+
 64%|██████▍   | 25/39 [00:02<00:01,  8.58it/s]
+
 69%|██████▉   | 27/39 [00:02<00:01,  9.65it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 10.40it/s]
+
 79%|███████▉  | 31/39 [00:03<00:00, 10.95it/s]
+
 85%|████████▍ | 33/39 [00:03<00:00, 10.31it/s]
+
 90%|████████▉ | 35/39 [00:03<00:00,  8.44it/s]
+
 95%|█████████▍| 37/39 [00:03<00:00,  9.17it/s]
+
100%|██████████| 39/39 [00:03<00:00,  9.93it/s]
                                                    
+
                                               
{'eval_loss': '4.77', 'eval_runtime': '4.093', 'eval_samples_per_second': '596.3', 'eval_steps_per_second': '9.528', 'epoch': '3.525'}
+
  7%|▋         | 504/7150 [07:42<1:20:31,  1.38it/s]
+
100%|██████████| 39/39 [00:04<00:00,  9.93it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-504
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-504/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-504/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.79it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.78it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-504/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-504/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-504/tokenizer_config.json
+[eval] epoch 3.525 step 504 loss 4.7696 (best 4.7696 @ ep 3.525, 0/20 since best)
+
+
                                               
  7%|▋         | 505/7150 [07:44<4:43:45,  2.56s/it]
  7%|▋         | 506/7150 [07:45<3:49:07,  2.07s/it]
  7%|▋         | 507/7150 [07:46<3:10:28,  1.72s/it]
  7%|▋         | 508/7150 [07:47<2:39:32,  1.44s/it]
  7%|▋         | 509/7150 [07:48<2:16:17,  1.23s/it]
  7%|▋         | 510/7150 [07:49<2:05:21,  1.13s/it]
                                                    
{'loss': '4.671', 'grad_norm': '0.3817', 'learning_rate': '0.0006', 'epoch': '3.567'}
+
  7%|▋         | 510/7150 [07:49<2:05:21,  1.13s/it]
  7%|▋         | 511/7150 [07:49<1:49:27,  1.01it/s]
  7%|▋         | 512/7150 [07:50<1:46:22,  1.04it/s]
  7%|▋         | 513/7150 [07:51<1:43:09,  1.07it/s]
  7%|▋         | 514/7150 [07:52<1:34:02,  1.18it/s]
  7%|▋         | 515/7150 [07:53<1:35:15,  1.16it/s]
  7%|▋         | 516/7150 [07:53<1:34:19,  1.17it/s]
  7%|▋         | 517/7150 [07:54<1:29:36,  1.23it/s]
  7%|▋         | 518/7150 [07:55<1:32:07,  1.20it/s]
  7%|▋         | 519/7150 [07:56<1:26:21,  1.28it/s]
  7%|▋         | 520/7150 [07:57<1:28:31,  1.25it/s]
                                                    
{'loss': '4.621', 'grad_norm': '0.3436', 'learning_rate': '0.0006', 'epoch': '3.637'}
+
  7%|▋         | 520/7150 [07:57<1:28:31,  1.25it/s]
  7%|▋         | 521/7150 [07:57<1:30:36,  1.22it/s]
  7%|▋         | 522/7150 [07:58<1:25:09,  1.30it/s]
  7%|▋         | 523/7150 [07:59<1:29:16,  1.24it/s]
  7%|▋         | 524/7150 [08:00<1:24:17,  1.31it/s]
  7%|▋         | 525/7150 [08:00<1:28:18,  1.25it/s]
  7%|▋         | 526/7150 [08:01<1:23:25,  1.32it/s]
  7%|▋         | 527/7150 [08:02<1:27:11,  1.27it/s]
  7%|▋         | 528/7150 [08:03<1:25:53,  1.28it/s]
  7%|▋         | 529/7150 [08:04<1:26:18,  1.28it/s]
  7%|▋         | 530/7150 [08:04<1:28:35,  1.25it/s]
                                                    
{'loss': '4.612', 'grad_norm': '0.7323', 'learning_rate': '0.0006', 'epoch': '3.708'}
+
  7%|▋         | 530/7150 [08:04<1:28:35,  1.25it/s]
  7%|▋         | 531/7150 [08:05<1:24:53,  1.30it/s]
  7%|▋         | 532/7150 [08:06<1:28:57,  1.24it/s]
  7%|▋         | 533/7150 [08:07<1:24:02,  1.31it/s]
  7%|▋         | 534/7150 [08:08<1:28:07,  1.25it/s]
  7%|▋         | 535/7150 [08:08<1:32:08,  1.20it/s]
  7%|▋         | 536/7150 [08:09<1:27:34,  1.26it/s]
  8%|▊         | 537/7150 [08:10<1:28:30,  1.25it/s]
  8%|▊         | 538/7150 [08:11<1:30:55,  1.21it/s]
  8%|▊         | 539/7150 [08:12<1:26:25,  1.27it/s]
  8%|▊         | 540/7150 [08:12<1:30:47,  1.21it/s]
                                                    
{'loss': '4.615', 'grad_norm': '0.4228', 'learning_rate': '0.0006', 'epoch': '3.778'}
+
  8%|▊         | 540/7150 [08:12<1:30:47,  1.21it/s]
  8%|▊         | 541/7150 [08:13<1:25:11,  1.29it/s]
  8%|▊         | 542/7150 [08:14<1:29:30,  1.23it/s]
  8%|▊         | 543/7150 [08:15<1:24:44,  1.30it/s]
  8%|▊         | 544/7150 [08:16<1:36:32,  1.14it/s]
  8%|▊         | 545/7150 [08:16<1:30:10,  1.22it/s]
  8%|▊         | 546/7150 [08:17<1:33:14,  1.18it/s]
  8%|▊         | 547/7150 [08:18<1:30:42,  1.21it/s]
  8%|▊         | 548/7150 [08:19<1:29:15,  1.23it/s]
  8%|▊         | 549/7150 [08:20<1:32:48,  1.19it/s]
  8%|▊         | 550/7150 [08:21<1:31:27,  1.20it/s]
                                                    
{'loss': '4.542', 'grad_norm': '0.2599', 'learning_rate': '0.0006', 'epoch': '3.848'}
+
  8%|▊         | 550/7150 [08:21<1:31:27,  1.20it/s]
  8%|▊         | 551/7150 [08:21<1:30:11,  1.22it/s]
  8%|▊         | 552/7150 [08:23<1:39:38,  1.10it/s]
  8%|▊         | 553/7150 [08:23<1:33:59,  1.17it/s]
  8%|▊         | 554/7150 [08:24<1:36:01,  1.14it/s]
  8%|▊         | 555/7150 [08:25<1:29:54,  1.22it/s]
  8%|▊         | 556/7150 [08:26<1:29:25,  1.23it/s]
  8%|▊         | 557/7150 [08:26<1:26:16,  1.27it/s]
  8%|▊         | 558/7150 [08:27<1:23:11,  1.32it/s]
  8%|▊         | 559/7150 [08:28<1:19:46,  1.38it/s]
  8%|▊         | 560/7150 [08:28<1:18:55,  1.39it/s]
                                                    
{'loss': '4.489', 'grad_norm': '0.4035', 'learning_rate': '0.0006', 'epoch': '3.918'}
+
  8%|▊         | 560/7150 [08:28<1:18:55,  1.39it/s]
  8%|▊         | 561/7150 [08:29<1:18:59,  1.39it/s]
  8%|▊         | 562/7150 [08:30<1:18:18,  1.40it/s]
  8%|▊         | 563/7150 [08:31<1:16:57,  1.43it/s]
  8%|▊         | 564/7150 [08:31<1:17:35,  1.41it/s]
  8%|▊         | 565/7150 [08:32<1:16:55,  1.43it/s]
  8%|▊         | 566/7150 [08:33<1:18:03,  1.41it/s]
  8%|▊         | 567/7150 [08:33<1:16:31,  1.43it/s]
  8%|▊         | 568/7150 [08:34<1:16:34,  1.43it/s]
  8%|▊         | 569/7150 [08:35<1:17:48,  1.41it/s]
  8%|▊         | 570/7150 [08:36<1:17:48,  1.41it/s]
                                                    
{'loss': '4.468', 'grad_norm': '0.2667', 'learning_rate': '0.0006', 'epoch': '3.988'}
+
  8%|▊         | 570/7150 [08:36<1:17:48,  1.41it/s]
  8%|▊         | 571/7150 [08:36<1:18:03,  1.40it/s]
  8%|▊         | 572/7150 [08:37<1:11:03,  1.54it/s]
  8%|▊         | 573/7150 [08:38<1:15:01,  1.46it/s]
  8%|▊         | 574/7150 [08:38<1:16:00,  1.44it/s]
  8%|▊         | 575/7150 [08:39<1:14:41,  1.47it/s]
  8%|▊         | 576/7150 [08:40<1:15:24,  1.45it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:02, 17.34it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 12.78it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 12.40it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 12.46it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.26it/s]
+
 33%|███▎      | 13/39 [00:01<00:02, 12.36it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.20it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 11.32it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 11.49it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 11.79it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 11.94it/s]
+
 64%|██████▍   | 25/39 [00:02<00:01, 12.02it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.09it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 10.86it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 11.13it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 11.44it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 11.68it/s]
+
 95%|█████████▍| 37/39 [00:03<00:00, 11.47it/s]
+
100%|██████████| 39/39 [00:03<00:00, 10.40it/s]
                                                    
+
                                               
{'eval_loss': '4.556', 'eval_runtime': '3.497', 'eval_samples_per_second': '698.1', 'eval_steps_per_second': '11.15', 'epoch': '4.028'}
+
  8%|▊         | 576/7150 [08:43<1:15:24,  1.45it/s]
+
100%|██████████| 39/39 [00:03<00:00, 10.40it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-576
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-576/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-576/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.86it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.85it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-576/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-576/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-576/tokenizer_config.json
+[eval] epoch 4.028 step 576 loss 4.5557 (best 4.5557 @ ep 4.028, 0/20 since best)
+
+
                                               
  8%|▊         | 577/7150 [08:46<4:18:12,  2.36s/it]
  8%|▊         | 578/7150 [08:47<3:25:45,  1.88s/it]
  8%|▊         | 579/7150 [08:47<2:45:31,  1.51s/it]
  8%|▊         | 580/7150 [08:48<2:19:43,  1.28s/it]
                                                    
{'loss': '4.418', 'grad_norm': '0.4902', 'learning_rate': '0.0006', 'epoch': '4.056'}
+
  8%|▊         | 580/7150 [08:48<2:19:43,  1.28s/it]
  8%|▊         | 581/7150 [08:49<2:01:27,  1.11s/it]
  8%|▊         | 582/7150 [08:49<1:47:32,  1.02it/s]
  8%|▊         | 583/7150 [08:50<1:37:24,  1.12it/s]
  8%|▊         | 584/7150 [08:51<1:29:55,  1.22it/s]
  8%|▊         | 585/7150 [08:51<1:26:18,  1.27it/s]
  8%|▊         | 586/7150 [08:52<1:23:19,  1.31it/s]
  8%|▊         | 587/7150 [08:53<1:22:12,  1.33it/s]
  8%|▊         | 588/7150 [08:54<1:18:58,  1.38it/s]
  8%|▊         | 589/7150 [08:54<1:18:18,  1.40it/s]
  8%|▊         | 590/7150 [08:55<1:17:45,  1.41it/s]
                                                    
{'loss': '4.395', 'grad_norm': '0.6081', 'learning_rate': '0.0006', 'epoch': '4.126'}
+
  8%|▊         | 590/7150 [08:55<1:17:45,  1.41it/s]
  8%|▊         | 591/7150 [08:56<1:16:59,  1.42it/s]
  8%|▊         | 592/7150 [08:56<1:16:28,  1.43it/s]
  8%|▊         | 593/7150 [08:57<1:16:05,  1.44it/s]
  8%|▊         | 594/7150 [08:58<1:15:31,  1.45it/s]
  8%|▊         | 595/7150 [08:58<1:14:22,  1.47it/s]
  8%|▊         | 596/7150 [08:59<1:14:35,  1.46it/s]
  8%|▊         | 597/7150 [09:00<1:14:33,  1.47it/s]
  8%|▊         | 598/7150 [09:00<1:13:39,  1.48it/s]
  8%|▊         | 599/7150 [09:01<1:13:57,  1.48it/s]
  8%|▊         | 600/7150 [09:02<1:14:15,  1.47it/s]
                                                    
{'loss': '4.39', 'grad_norm': '0.3149', 'learning_rate': '0.0006', 'epoch': '4.196'}
+
  8%|▊         | 600/7150 [09:02<1:14:15,  1.47it/s]
  8%|▊         | 601/7150 [09:02<1:14:37,  1.46it/s]
  8%|▊         | 602/7150 [09:03<1:15:26,  1.45it/s]
  8%|▊         | 603/7150 [09:04<1:14:12,  1.47it/s]
  8%|▊         | 604/7150 [09:04<1:14:25,  1.47it/s]
  8%|▊         | 605/7150 [09:05<1:14:34,  1.46it/s]
  8%|▊         | 606/7150 [09:06<1:13:34,  1.48it/s]
  8%|▊         | 607/7150 [09:06<1:13:50,  1.48it/s]
  9%|▊         | 608/7150 [09:07<1:14:21,  1.47it/s]
  9%|▊         | 609/7150 [09:08<1:14:25,  1.46it/s]
  9%|▊         | 610/7150 [09:09<1:13:28,  1.48it/s]
                                                    
{'loss': '4.346', 'grad_norm': '0.3042', 'learning_rate': '0.0006', 'epoch': '4.266'}
+
  9%|▊         | 610/7150 [09:09<1:13:28,  1.48it/s]
  9%|▊         | 611/7150 [09:09<1:13:45,  1.48it/s]
  9%|▊         | 612/7150 [09:10<1:13:53,  1.47it/s]
  9%|▊         | 613/7150 [09:11<1:13:05,  1.49it/s]
  9%|▊         | 614/7150 [09:11<1:14:33,  1.46it/s]
  9%|▊         | 615/7150 [09:12<1:14:32,  1.46it/s]
  9%|▊         | 616/7150 [09:13<1:14:13,  1.47it/s]
  9%|▊         | 617/7150 [09:13<1:19:27,  1.37it/s]
  9%|▊         | 618/7150 [09:14<1:17:00,  1.41it/s]
  9%|▊         | 619/7150 [09:15<1:16:09,  1.43it/s]
  9%|▊         | 620/7150 [09:15<1:15:46,  1.44it/s]
                                                    
{'loss': '4.322', 'grad_norm': '0.2424', 'learning_rate': '0.0006', 'epoch': '4.336'}
+
  9%|▊         | 620/7150 [09:15<1:15:46,  1.44it/s]
  9%|▊         | 621/7150 [09:16<1:15:27,  1.44it/s]
  9%|▊         | 622/7150 [09:17<1:14:12,  1.47it/s]
  9%|▊         | 623/7150 [09:18<1:18:35,  1.38it/s]
  9%|▊         | 624/7150 [09:18<1:16:23,  1.42it/s]
  9%|▊         | 625/7150 [09:19<1:16:53,  1.41it/s]
  9%|▉         | 626/7150 [09:20<1:16:06,  1.43it/s]
  9%|▉         | 627/7150 [09:20<1:14:33,  1.46it/s]
  9%|▉         | 628/7150 [09:21<1:15:36,  1.44it/s]
  9%|▉         | 629/7150 [09:22<1:14:48,  1.45it/s]
  9%|▉         | 630/7150 [09:22<1:13:57,  1.47it/s]
                                                    
{'loss': '4.298', 'grad_norm': '0.382', 'learning_rate': '0.0006', 'epoch': '4.406'}
+
  9%|▉         | 630/7150 [09:22<1:13:57,  1.47it/s]
  9%|▉         | 631/7150 [09:23<1:14:13,  1.46it/s]
  9%|▉         | 632/7150 [09:24<1:15:22,  1.44it/s]
  9%|▉         | 633/7150 [09:25<1:15:59,  1.43it/s]
  9%|▉         | 634/7150 [09:25<1:15:00,  1.45it/s]
  9%|▉         | 635/7150 [09:26<1:21:57,  1.32it/s]
  9%|▉         | 636/7150 [09:27<1:19:30,  1.37it/s]
  9%|▉         | 637/7150 [09:27<1:17:57,  1.39it/s]
  9%|▉         | 638/7150 [09:28<1:16:38,  1.42it/s]
  9%|▉         | 639/7150 [09:29<1:15:52,  1.43it/s]
  9%|▉         | 640/7150 [09:29<1:14:26,  1.46it/s]
                                                    
{'loss': '4.281', 'grad_norm': '0.3829', 'learning_rate': '0.0006', 'epoch': '4.476'}
+
  9%|▉         | 640/7150 [09:29<1:14:26,  1.46it/s]
  9%|▉         | 641/7150 [09:30<1:14:22,  1.46it/s]
  9%|▉         | 642/7150 [09:31<1:16:20,  1.42it/s]
  9%|▉         | 643/7150 [09:32<1:15:22,  1.44it/s]
  9%|▉         | 644/7150 [09:32<1:14:00,  1.47it/s]
  9%|▉         | 645/7150 [09:33<1:13:45,  1.47it/s]
  9%|▉         | 646/7150 [09:34<1:13:46,  1.47it/s]
  9%|▉         | 647/7150 [09:34<1:12:52,  1.49it/s]
  9%|▉         | 648/7150 [09:35<1:15:11,  1.44it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.42it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.78it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.64it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 12.38it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 11.30it/s]
+
 33%|███▎      | 13/39 [00:01<00:02, 10.80it/s]
+
 38%|███▊      | 15/39 [00:01<00:02, 11.23it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 11.53it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 11.73it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 11.31it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 10.56it/s]
+
 64%|██████▍   | 25/39 [00:02<00:01, 10.21it/s]
+
 69%|██████▉   | 27/39 [00:02<00:01, 10.06it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 10.65it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 11.09it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 11.43it/s]
+
 90%|████████▉ | 35/39 [00:03<00:00, 11.17it/s]
+
 95%|█████████▍| 37/39 [00:03<00:00, 10.69it/s]
+
100%|██████████| 39/39 [00:03<00:00, 10.08it/s]
                                                    
+
                                               
{'eval_loss': '4.382', 'eval_runtime': '3.691', 'eval_samples_per_second': '661.3', 'eval_steps_per_second': '10.57', 'epoch': '4.532'}
+
  9%|▉         | 648/7150 [09:39<1:15:11,  1.44it/s]
+
100%|██████████| 39/39 [00:03<00:00, 10.08it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-648
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-648/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-648/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.92it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.92it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-648/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-648/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-648/tokenizer_config.json
+[eval] epoch 4.532 step 648 loss 4.3817 (best 4.3817 @ ep 4.532, 0/20 since best)
+
+
                                               
  9%|▉         | 649/7150 [09:41<4:22:58,  2.43s/it]
  9%|▉         | 650/7150 [09:42<3:26:04,  1.90s/it]
                                                    
{'loss': '4.26', 'grad_norm': '0.3221', 'learning_rate': '0.0006', 'epoch': '4.546'}
+
  9%|▉         | 650/7150 [09:42<3:26:04,  1.90s/it]
  9%|▉         | 651/7150 [09:43<2:46:39,  1.54s/it]
  9%|▉         | 652/7150 [09:43<2:18:54,  1.28s/it]
  9%|▉         | 653/7150 [09:44<1:59:38,  1.10s/it]
  9%|▉         | 654/7150 [09:45<1:46:19,  1.02it/s]
  9%|▉         | 655/7150 [09:46<1:36:29,  1.12it/s]
  9%|▉         | 656/7150 [09:46<1:29:43,  1.21it/s]
  9%|▉         | 657/7150 [09:47<1:25:06,  1.27it/s]
  9%|▉         | 658/7150 [09:48<1:21:46,  1.32it/s]
  9%|▉         | 659/7150 [09:48<1:19:41,  1.36it/s]
  9%|▉         | 660/7150 [09:49<1:18:04,  1.39it/s]
                                                    
{'loss': '4.251', 'grad_norm': '0.4348', 'learning_rate': '0.0006', 'epoch': '4.616'}
+
  9%|▉         | 660/7150 [09:49<1:18:04,  1.39it/s]
  9%|▉         | 661/7150 [09:50<1:17:02,  1.40it/s]
  9%|▉         | 662/7150 [09:50<1:17:30,  1.40it/s]
  9%|▉         | 663/7150 [09:51<1:17:46,  1.39it/s]
  9%|▉         | 664/7150 [09:52<1:16:45,  1.41it/s]
  9%|▉         | 665/7150 [09:53<1:18:34,  1.38it/s]
  9%|▉         | 666/7150 [09:53<1:16:12,  1.42it/s]
  9%|▉         | 667/7150 [09:54<1:15:34,  1.43it/s]
  9%|▉         | 668/7150 [09:55<1:19:41,  1.36it/s]
  9%|▉         | 669/7150 [09:56<1:26:20,  1.25it/s]
  9%|▉         | 670/7150 [09:56<1:25:48,  1.26it/s]
                                                    
{'loss': '4.237', 'grad_norm': '0.3812', 'learning_rate': '0.0006', 'epoch': '4.687'}
+
  9%|▉         | 670/7150 [09:56<1:25:48,  1.26it/s]
  9%|▉         | 671/7150 [09:57<1:22:06,  1.32it/s]
  9%|▉         | 672/7150 [09:58<1:20:37,  1.34it/s]
  9%|▉         | 673/7150 [09:59<1:20:01,  1.35it/s]
  9%|▉         | 674/7150 [09:59<1:18:01,  1.38it/s]
  9%|▉         | 675/7150 [10:00<1:16:42,  1.41it/s]
  9%|▉         | 676/7150 [10:01<1:24:04,  1.28it/s]
  9%|▉         | 677/7150 [10:02<1:27:16,  1.24it/s]
  9%|▉         | 678/7150 [10:03<1:24:36,  1.27it/s]
  9%|▉         | 679/7150 [10:03<1:22:54,  1.30it/s]
 10%|▉         | 680/7150 [10:04<1:23:06,  1.30it/s]
                                                    
{'loss': '4.216', 'grad_norm': '0.295', 'learning_rate': '0.0006', 'epoch': '4.757'}
+
 10%|▉         | 680/7150 [10:04<1:23:06,  1.30it/s]
 10%|▉         | 681/7150 [10:05<1:22:47,  1.30it/s]
 10%|▉         | 682/7150 [10:06<1:27:56,  1.23it/s]
 10%|▉         | 683/7150 [10:07<1:28:10,  1.22it/s]
 10%|▉         | 684/7150 [10:07<1:27:26,  1.23it/s]
 10%|▉         | 685/7150 [10:08<1:31:44,  1.17it/s]
 10%|▉         | 686/7150 [10:09<1:25:18,  1.26it/s]
 10%|▉         | 687/7150 [10:10<1:26:47,  1.24it/s]
 10%|▉         | 688/7150 [10:11<1:29:41,  1.20it/s]
 10%|▉         | 689/7150 [10:11<1:23:50,  1.28it/s]
 10%|▉         | 690/7150 [10:12<1:26:36,  1.24it/s]
                                                    
{'loss': '4.187', 'grad_norm': '0.2547', 'learning_rate': '0.0006', 'epoch': '4.827'}
+
 10%|▉         | 690/7150 [10:12<1:26:36,  1.24it/s]
 10%|▉         | 691/7150 [10:13<1:22:37,  1.30it/s]
 10%|▉         | 692/7150 [10:14<1:27:54,  1.22it/s]
 10%|▉         | 693/7150 [10:15<1:27:36,  1.23it/s]
 10%|▉         | 694/7150 [10:15<1:23:16,  1.29it/s]
 10%|▉         | 695/7150 [10:16<1:19:21,  1.36it/s]
 10%|▉         | 696/7150 [10:17<1:17:13,  1.39it/s]
 10%|▉         | 697/7150 [10:17<1:15:47,  1.42it/s]
 10%|▉         | 698/7150 [10:18<1:15:25,  1.43it/s]
 10%|▉         | 699/7150 [10:19<1:14:28,  1.44it/s]
 10%|▉         | 700/7150 [10:19<1:13:12,  1.47it/s]
                                                    
{'loss': '4.162', 'grad_norm': '0.3478', 'learning_rate': '0.0006', 'epoch': '4.897'}
+
 10%|▉         | 700/7150 [10:19<1:13:12,  1.47it/s]
 10%|▉         | 701/7150 [10:20<1:17:42,  1.38it/s]
 10%|▉         | 702/7150 [10:21<1:15:27,  1.42it/s]
 10%|▉         | 703/7150 [10:22<1:16:12,  1.41it/s]
 10%|▉         | 704/7150 [10:22<1:14:24,  1.44it/s]
 10%|▉         | 705/7150 [10:23<1:16:15,  1.41it/s]
 10%|▉         | 706/7150 [10:24<1:14:44,  1.44it/s]
 10%|▉         | 707/7150 [10:24<1:15:44,  1.42it/s]
 10%|▉         | 708/7150 [10:25<1:16:30,  1.40it/s]
 10%|▉         | 709/7150 [10:26<1:15:53,  1.41it/s]
 10%|▉         | 710/7150 [10:26<1:14:13,  1.45it/s]
                                                    
{'loss': '4.142', 'grad_norm': '0.2184', 'learning_rate': '0.0006', 'epoch': '4.967'}
+
 10%|▉         | 710/7150 [10:26<1:14:13,  1.45it/s]
 10%|▉         | 711/7150 [10:27<1:14:26,  1.44it/s]
 10%|▉         | 712/7150 [10:28<1:15:42,  1.42it/s]
 10%|▉         | 713/7150 [10:29<1:16:33,  1.40it/s]
 10%|▉         | 714/7150 [10:29<1:14:36,  1.44it/s]
 10%|█         | 715/7150 [10:30<1:10:35,  1.52it/s]
 10%|█         | 716/7150 [10:31<1:18:04,  1.37it/s]
 10%|█         | 717/7150 [10:31<1:17:07,  1.39it/s]
 10%|█         | 718/7150 [10:32<1:16:23,  1.40it/s]
 10%|█         | 719/7150 [10:33<1:19:39,  1.35it/s]
 10%|█         | 720/7150 [10:34<1:16:47,  1.40it/s]
                                                    
{'loss': '4.121', 'grad_norm': '0.5531', 'learning_rate': '0.0006', 'epoch': '5.035'}
+
 10%|█         | 720/7150 [10:34<1:16:47,  1.40it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:02, 17.69it/s]
+
 10%|█         | 4/39 [00:00<00:03, 10.88it/s]
+
 15%|█▌        | 6/39 [00:00<00:03, 10.79it/s]
+
 21%|██        | 8/39 [00:00<00:02, 11.31it/s]
+
 26%|██▌       | 10/39 [00:00<00:02, 11.63it/s]
+
 31%|███       | 12/39 [00:01<00:02, 11.84it/s]
+
 36%|███▌      | 14/39 [00:01<00:02, 11.96it/s]
+
 41%|████      | 16/39 [00:01<00:01, 11.67it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 10.60it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01,  9.70it/s]
+
 56%|█████▋    | 22/39 [00:02<00:01,  9.59it/s]
+
 59%|█████▉    | 23/39 [00:02<00:01,  9.31it/s]
+
 62%|██████▏   | 24/39 [00:02<00:01,  9.08it/s]
+
 64%|██████▍   | 25/39 [00:02<00:01,  8.90it/s]
+
 69%|██████▉   | 27/39 [00:02<00:01,  9.03it/s]
+
 72%|███████▏  | 28/39 [00:02<00:01,  8.81it/s]
+
 74%|███████▍  | 29/39 [00:02<00:01,  8.65it/s]
+
 77%|███████▋  | 30/39 [00:02<00:01,  8.78it/s]
+
 79%|███████▉  | 31/39 [00:03<00:00,  8.93it/s]
+
 85%|████████▍ | 33/39 [00:03<00:00, 10.04it/s]
+
 90%|████████▉ | 35/39 [00:03<00:00, 10.78it/s]
+
 95%|█████████▍| 37/39 [00:03<00:00, 11.25it/s]
+
100%|██████████| 39/39 [00:03<00:00, 11.57it/s]
                                                    
+
                                               
{'eval_loss': '4.278', 'eval_runtime': '3.908', 'eval_samples_per_second': '624.6', 'eval_steps_per_second': '9.98', 'epoch': '5.035'}
+
 10%|█         | 720/7150 [10:37<1:16:47,  1.40it/s]
+
100%|██████████| 39/39 [00:03<00:00, 11.57it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-720
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-720/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-720/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.88it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.88it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-720/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-720/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-720/tokenizer_config.json
+[eval] epoch 5.035 step 720 loss 4.2783 (best 4.2783 @ ep 5.035, 0/20 since best)
+
+
                                               
 10%|█         | 721/7150 [10:40<4:35:37,  2.57s/it]
 10%|█         | 722/7150 [10:41<3:41:17,  2.07s/it]
 10%|█         | 723/7150 [10:42<2:55:51,  1.64s/it]
 10%|█         | 724/7150 [10:43<2:25:10,  1.36s/it]
 10%|█         | 725/7150 [10:43<2:08:55,  1.20s/it]
 10%|█         | 726/7150 [10:44<1:51:13,  1.04s/it]
 10%|█         | 727/7150 [10:45<1:40:26,  1.07it/s]
 10%|█         | 728/7150 [10:46<1:32:51,  1.15it/s]
 10%|█         | 729/7150 [10:46<1:27:26,  1.22it/s]
 10%|█         | 730/7150 [10:47<1:22:12,  1.30it/s]
                                                    
{'loss': '4.095', 'grad_norm': '0.4001', 'learning_rate': '0.0006', 'epoch': '5.105'}
+
 10%|█         | 730/7150 [10:47<1:22:12,  1.30it/s]
 10%|█         | 731/7150 [10:48<1:21:45,  1.31it/s]
 10%|█         | 732/7150 [10:48<1:19:44,  1.34it/s]
 10%|█         | 733/7150 [10:49<1:20:00,  1.34it/s]
 10%|█         | 734/7150 [10:50<1:16:59,  1.39it/s]
 10%|█         | 735/7150 [10:50<1:16:23,  1.40it/s]
 10%|█         | 736/7150 [10:51<1:15:47,  1.41it/s]
 10%|█         | 737/7150 [10:52<1:15:46,  1.41it/s]
 10%|█         | 738/7150 [10:53<1:15:31,  1.42it/s]
 10%|█         | 739/7150 [10:53<1:15:23,  1.42it/s]
 10%|█         | 740/7150 [10:54<1:13:44,  1.45it/s]
                                                    
{'loss': '4.09', 'grad_norm': '0.3099', 'learning_rate': '0.0006', 'epoch': '5.175'}
+
 10%|█         | 740/7150 [10:54<1:13:44,  1.45it/s]
 10%|█         | 741/7150 [10:55<1:14:08,  1.44it/s]
 10%|█         | 742/7150 [10:55<1:14:29,  1.43it/s]
 10%|█         | 743/7150 [10:56<1:14:29,  1.43it/s]
 10%|█         | 744/7150 [10:57<1:22:04,  1.30it/s]
 10%|█         | 745/7150 [10:58<1:19:36,  1.34it/s]
 10%|█         | 746/7150 [10:58<1:18:25,  1.36it/s]
 10%|█         | 747/7150 [10:59<1:17:38,  1.37it/s]
 10%|█         | 748/7150 [11:00<1:17:49,  1.37it/s]
 10%|█         | 749/7150 [11:01<1:17:54,  1.37it/s]
 10%|█         | 750/7150 [11:01<1:18:51,  1.35it/s]
                                                    
{'loss': '4.052', 'grad_norm': '0.235', 'learning_rate': '0.0006', 'epoch': '5.245'}
+
 10%|█         | 750/7150 [11:01<1:18:51,  1.35it/s]
 11%|█         | 751/7150 [11:02<1:16:39,  1.39it/s]
 11%|█         | 752/7150 [11:03<1:15:14,  1.42it/s]
 11%|█         | 753/7150 [11:03<1:18:40,  1.36it/s]
 11%|█         | 754/7150 [11:04<1:19:16,  1.34it/s]
 11%|█         | 755/7150 [11:05<1:19:16,  1.34it/s]
 11%|█         | 756/7150 [11:06<1:21:51,  1.30it/s]
 11%|█         | 757/7150 [11:06<1:18:26,  1.36it/s]
 11%|█         | 758/7150 [11:07<1:22:20,  1.29it/s]
 11%|█         | 759/7150 [11:08<1:20:14,  1.33it/s]
 11%|█         | 760/7150 [11:09<1:17:01,  1.38it/s]
                                                    
{'loss': '4.043', 'grad_norm': '0.2431', 'learning_rate': '0.0006', 'epoch': '5.315'}
+
 11%|█         | 760/7150 [11:09<1:17:01,  1.38it/s]
 11%|█         | 761/7150 [11:10<1:23:16,  1.28it/s]
 11%|█         | 762/7150 [11:10<1:22:35,  1.29it/s]
 11%|█         | 763/7150 [11:11<1:21:53,  1.30it/s]
 11%|█         | 764/7150 [11:12<1:27:09,  1.22it/s]
 11%|█         | 765/7150 [11:13<1:28:46,  1.20it/s]
 11%|█         | 766/7150 [11:14<1:26:18,  1.23it/s]
 11%|█         | 767/7150 [11:15<1:31:15,  1.17it/s]
 11%|█         | 768/7150 [11:15<1:26:37,  1.23it/s]
 11%|█         | 769/7150 [11:16<1:29:54,  1.18it/s]
 11%|█         | 770/7150 [11:17<1:33:45,  1.13it/s]
                                                    
{'loss': '4.024', 'grad_norm': '0.2431', 'learning_rate': '0.0006', 'epoch': '5.385'}
+
 11%|█         | 770/7150 [11:17<1:33:45,  1.13it/s]
 11%|█         | 771/7150 [11:18<1:26:50,  1.22it/s]
 11%|█         | 772/7150 [11:19<1:23:23,  1.27it/s]
 11%|█         | 773/7150 [11:19<1:20:13,  1.32it/s]
 11%|█         | 774/7150 [11:20<1:18:22,  1.36it/s]
 11%|█         | 775/7150 [11:21<1:17:47,  1.37it/s]
 11%|█         | 776/7150 [11:21<1:16:37,  1.39it/s]
 11%|█         | 777/7150 [11:22<1:18:43,  1.35it/s]
 11%|█         | 778/7150 [11:23<1:17:15,  1.37it/s]
 11%|█         | 779/7150 [11:24<1:16:03,  1.40it/s]
 11%|█         | 780/7150 [11:24<1:19:40,  1.33it/s]
                                                    
{'loss': '4.035', 'grad_norm': '0.4196', 'learning_rate': '0.0006', 'epoch': '5.455'}
+
 11%|█         | 780/7150 [11:24<1:19:40,  1.33it/s]
 11%|█         | 781/7150 [11:25<1:17:28,  1.37it/s]
 11%|█         | 782/7150 [11:26<1:16:05,  1.39it/s]
 11%|█         | 783/7150 [11:27<1:17:29,  1.37it/s]
 11%|█         | 784/7150 [11:27<1:16:00,  1.40it/s]
 11%|█         | 785/7150 [11:28<1:18:19,  1.35it/s]
 11%|█         | 786/7150 [11:29<1:20:15,  1.32it/s]
 11%|█         | 787/7150 [11:30<1:18:04,  1.36it/s]
 11%|█         | 788/7150 [11:30<1:16:03,  1.39it/s]
 11%|█         | 789/7150 [11:31<1:13:59,  1.43it/s]
 11%|█         | 790/7150 [11:32<1:13:21,  1.44it/s]
                                                    
{'loss': '4.013', 'grad_norm': '0.3296', 'learning_rate': '0.0006', 'epoch': '5.525'}
+
 11%|█         | 790/7150 [11:32<1:13:21,  1.44it/s]
 11%|█         | 791/7150 [11:32<1:13:53,  1.43it/s]
 11%|█         | 792/7150 [11:33<1:13:28,  1.44it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.23it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.69it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.55it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.04it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.13it/s]
+
 33%|███▎      | 13/39 [00:01<00:02, 11.96it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.05it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.10it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.13it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.17it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 11.69it/s]
+
 64%|██████▍   | 25/39 [00:02<00:01, 11.03it/s]
+
 69%|██████▉   | 27/39 [00:02<00:01, 10.82it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 11.22it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 11.51it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 11.73it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 11.16it/s]
+
 95%|█████████▍| 37/39 [00:03<00:00, 10.99it/s]
+
100%|██████████| 39/39 [00:03<00:00, 11.35it/s]
                                                    
+
                                               
{'eval_loss': '4.171', 'eval_runtime': '3.445', 'eval_samples_per_second': '708.6', 'eval_steps_per_second': '11.32', 'epoch': '5.539'}
+
 11%|█         | 792/7150 [11:36<1:13:28,  1.44it/s]
+
100%|██████████| 39/39 [00:03<00:00, 11.35it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-792
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-792/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-792/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.92it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.92it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-792/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-792/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-792/tokenizer_config.json
+[eval] epoch 5.539 step 792 loss 4.1708 (best 4.1708 @ ep 5.539, 0/20 since best)
+
+
                                               
 11%|█         | 793/7150 [11:39<4:07:11,  2.33s/it]
 11%|█         | 794/7150 [11:40<3:14:05,  1.83s/it]
 11%|█         | 795/7150 [11:40<2:36:45,  1.48s/it]
 11%|█         | 796/7150 [11:41<2:11:12,  1.24s/it]
 11%|█         | 797/7150 [11:42<1:53:19,  1.07s/it]
 11%|█         | 798/7150 [11:42<1:42:36,  1.03it/s]
 11%|█         | 799/7150 [11:43<1:33:00,  1.14it/s]
 11%|█         | 800/7150 [11:44<1:26:31,  1.22it/s]
                                                    
{'loss': '4.003', 'grad_norm': '0.2392', 'learning_rate': '0.0006', 'epoch': '5.595'}
+
 11%|█         | 800/7150 [11:44<1:26:31,  1.22it/s]
 11%|█         | 801/7150 [11:45<1:22:06,  1.29it/s]
 11%|█         | 802/7150 [11:45<1:20:01,  1.32it/s]
 11%|█         | 803/7150 [11:46<1:18:33,  1.35it/s]
 11%|█         | 804/7150 [11:47<1:15:42,  1.40it/s]
 11%|█▏        | 805/7150 [11:47<1:14:16,  1.42it/s]
 11%|█▏        | 806/7150 [11:48<1:13:30,  1.44it/s]
 11%|█▏        | 807/7150 [11:49<1:12:10,  1.46it/s]
 11%|█▏        | 808/7150 [11:49<1:12:04,  1.47it/s]
 11%|█▏        | 809/7150 [11:50<1:12:01,  1.47it/s]
 11%|█▏        | 810/7150 [11:51<1:13:26,  1.44it/s]
                                                    
{'loss': '3.984', 'grad_norm': '0.2296', 'learning_rate': '0.0006', 'epoch': '5.665'}
+
 11%|█▏        | 810/7150 [11:51<1:13:26,  1.44it/s]
 11%|█▏        | 811/7150 [11:51<1:12:49,  1.45it/s]
 11%|█▏        | 812/7150 [11:52<1:13:17,  1.44it/s]
 11%|█▏        | 813/7150 [11:53<1:14:19,  1.42it/s]
 11%|█▏        | 814/7150 [11:53<1:14:10,  1.42it/s]
 11%|█▏        | 815/7150 [11:54<1:14:06,  1.42it/s]
 11%|█▏        | 816/7150 [11:55<1:14:00,  1.43it/s]
 11%|█▏        | 817/7150 [11:56<1:13:51,  1.43it/s]
 11%|█▏        | 818/7150 [11:56<1:13:44,  1.43it/s]
 11%|█▏        | 819/7150 [11:57<1:12:23,  1.46it/s]
 11%|█▏        | 820/7150 [11:58<1:12:49,  1.45it/s]
                                                    
{'loss': '3.974', 'grad_norm': '0.2851', 'learning_rate': '0.0006', 'epoch': '5.736'}
+
 11%|█▏        | 820/7150 [11:58<1:12:49,  1.45it/s]
 11%|█▏        | 821/7150 [11:58<1:15:57,  1.39it/s]
 11%|█▏        | 822/7150 [11:59<1:13:51,  1.43it/s]
 12%|█▏        | 823/7150 [12:00<1:13:46,  1.43it/s]
 12%|█▏        | 824/7150 [12:00<1:13:34,  1.43it/s]
 12%|█▏        | 825/7150 [12:01<1:12:14,  1.46it/s]
 12%|█▏        | 826/7150 [12:02<1:12:28,  1.45it/s]
 12%|█▏        | 827/7150 [12:03<1:12:46,  1.45it/s]
 12%|█▏        | 828/7150 [12:03<1:11:35,  1.47it/s]
 12%|█▏        | 829/7150 [12:04<1:13:28,  1.43it/s]
 12%|█▏        | 830/7150 [12:05<1:14:26,  1.41it/s]
                                                    
{'loss': '3.973', 'grad_norm': '0.2404', 'learning_rate': '0.0006', 'epoch': '5.806'}
+
 12%|█▏        | 830/7150 [12:05<1:14:26,  1.41it/s]
 12%|█▏        | 831/7150 [12:05<1:13:39,  1.43it/s]
 12%|█▏        | 832/7150 [12:06<1:13:46,  1.43it/s]
 12%|█▏        | 833/7150 [12:07<1:13:51,  1.43it/s]
 12%|█▏        | 834/7150 [12:07<1:15:09,  1.40it/s]
 12%|█▏        | 835/7150 [12:08<1:14:20,  1.42it/s]
 12%|█▏        | 836/7150 [12:09<1:17:57,  1.35it/s]
 12%|█▏        | 837/7150 [12:10<1:16:22,  1.38it/s]
 12%|█▏        | 838/7150 [12:10<1:14:59,  1.40it/s]
 12%|█▏        | 839/7150 [12:11<1:17:16,  1.36it/s]
 12%|█▏        | 840/7150 [12:12<1:15:53,  1.39it/s]
                                                    
{'loss': '3.973', 'grad_norm': '0.2848', 'learning_rate': '0.0006', 'epoch': '5.876'}
+
 12%|█▏        | 840/7150 [12:12<1:15:53,  1.39it/s]
 12%|█▏        | 841/7150 [12:13<1:19:14,  1.33it/s]
 12%|█▏        | 842/7150 [12:13<1:16:43,  1.37it/s]
 12%|█▏        | 843/7150 [12:14<1:14:20,  1.41it/s]
 12%|█▏        | 844/7150 [12:15<1:13:29,  1.43it/s]
 12%|█▏        | 845/7150 [12:15<1:12:55,  1.44it/s]
 12%|█▏        | 846/7150 [12:16<1:11:40,  1.47it/s]
 12%|█▏        | 847/7150 [12:17<1:11:48,  1.46it/s]
 12%|█▏        | 848/7150 [12:17<1:11:18,  1.47it/s]
 12%|█▏        | 849/7150 [12:18<1:11:31,  1.47it/s]
 12%|█▏        | 850/7150 [12:19<1:10:40,  1.49it/s]
                                                    
{'loss': '3.949', 'grad_norm': '0.3884', 'learning_rate': '0.0006', 'epoch': '5.946'}
+
 12%|█▏        | 850/7150 [12:19<1:10:40,  1.49it/s]
 12%|█▏        | 851/7150 [12:19<1:10:36,  1.49it/s]
 12%|█▏        | 852/7150 [12:20<1:12:03,  1.46it/s]
 12%|█▏        | 853/7150 [12:21<1:11:29,  1.47it/s]
 12%|█▏        | 854/7150 [12:21<1:11:55,  1.46it/s]
 12%|█▏        | 855/7150 [12:22<1:11:49,  1.46it/s]
 12%|█▏        | 856/7150 [12:23<1:12:07,  1.45it/s]
 12%|█▏        | 857/7150 [12:24<1:11:54,  1.46it/s]
 12%|█▏        | 858/7150 [12:24<1:05:53,  1.59it/s]
 12%|█▏        | 859/7150 [12:25<1:10:08,  1.49it/s]
 12%|█▏        | 860/7150 [12:25<1:10:17,  1.49it/s]
                                                    
{'loss': '3.946', 'grad_norm': '0.3256', 'learning_rate': '0.0006', 'epoch': '6.014'}
+
 12%|█▏        | 860/7150 [12:25<1:10:17,  1.49it/s]
 12%|█▏        | 861/7150 [12:26<1:09:51,  1.50it/s]
 12%|█▏        | 862/7150 [12:27<1:10:29,  1.49it/s]
 12%|█▏        | 863/7150 [12:27<1:10:56,  1.48it/s]
 12%|█▏        | 864/7150 [12:28<1:11:19,  1.47it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.79it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.87it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 12.99it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 12.70it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.52it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.41it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.34it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.29it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 11.38it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 11.65it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 11.83it/s]
+
 64%|██████▍   | 25/39 [00:02<00:01, 11.97it/s]
+
 69%|██████▉   | 27/39 [00:02<00:01, 11.46it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 11.50it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 11.70it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 11.86it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 11.97it/s]
+
 95%|█████████▍| 37/39 [00:03<00:00, 10.98it/s]
+
100%|██████████| 39/39 [00:03<00:00, 11.33it/s]
                                                    
+
                                               
{'eval_loss': '4.092', 'eval_runtime': '3.395', 'eval_samples_per_second': '719', 'eval_steps_per_second': '11.49', 'epoch': '6.042'}
+
 12%|█▏        | 864/7150 [12:32<1:11:19,  1.47it/s]
+
100%|██████████| 39/39 [00:03<00:00, 11.33it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-864
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-864/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-864/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.66it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.66it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-864/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-864/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-864/tokenizer_config.json
+[eval] epoch 6.042 step 864 loss 4.0923 (best 4.0923 @ ep 6.042, 0/20 since best)
+
+
                                               
 12%|█▏        | 865/7150 [12:34<4:05:20,  2.34s/it]
 12%|█▏        | 866/7150 [12:35<3:12:57,  1.84s/it]
 12%|█▏        | 867/7150 [12:36<2:35:41,  1.49s/it]
 12%|█▏        | 868/7150 [12:36<2:10:11,  1.24s/it]
 12%|█▏        | 869/7150 [12:37<1:56:11,  1.11s/it]
 12%|█▏        | 870/7150 [12:38<1:43:02,  1.02it/s]
                                                    
{'loss': '3.878', 'grad_norm': '0.4026', 'learning_rate': '0.0006', 'epoch': '6.084'}
+
 12%|█▏        | 870/7150 [12:38<1:43:02,  1.02it/s]
 12%|█▏        | 871/7150 [12:39<1:32:42,  1.13it/s]
 12%|█▏        | 872/7150 [12:39<1:25:38,  1.22it/s]
 12%|█▏        | 873/7150 [12:40<1:21:42,  1.28it/s]
 12%|█▏        | 874/7150 [12:41<1:20:00,  1.31it/s]
 12%|█▏        | 875/7150 [12:41<1:16:32,  1.37it/s]
 12%|█▏        | 876/7150 [12:42<1:15:17,  1.39it/s]
 12%|█▏        | 877/7150 [12:43<1:14:17,  1.41it/s]
 12%|█▏        | 878/7150 [12:43<1:13:43,  1.42it/s]
 12%|█▏        | 879/7150 [12:44<1:13:02,  1.43it/s]
 12%|█▏        | 880/7150 [12:45<1:11:36,  1.46it/s]
                                                    
{'loss': '3.883', 'grad_norm': '0.2483', 'learning_rate': '0.0006', 'epoch': '6.154'}
+
 12%|█▏        | 880/7150 [12:45<1:11:36,  1.46it/s]
 12%|█▏        | 881/7150 [12:45<1:10:50,  1.47it/s]
 12%|█▏        | 882/7150 [12:46<1:10:57,  1.47it/s]
 12%|█▏        | 883/7150 [12:47<1:12:05,  1.45it/s]
 12%|█▏        | 884/7150 [12:47<1:11:16,  1.47it/s]
 12%|█▏        | 885/7150 [12:48<1:10:47,  1.47it/s]
 12%|█▏        | 886/7150 [12:49<1:11:04,  1.47it/s]
 12%|█▏        | 887/7150 [12:49<1:11:19,  1.46it/s]
 12%|█▏        | 888/7150 [12:50<1:10:56,  1.47it/s]
 12%|█▏        | 889/7150 [12:51<1:10:06,  1.49it/s]
 12%|█▏        | 890/7150 [12:51<1:11:44,  1.45it/s]
                                                    
{'loss': '3.862', 'grad_norm': '0.2964', 'learning_rate': '0.0006', 'epoch': '6.224'}
+
 12%|█▏        | 890/7150 [12:51<1:11:44,  1.45it/s]
 12%|█▏        | 891/7150 [12:52<1:13:11,  1.43it/s]
 12%|█▏        | 892/7150 [12:53<1:11:39,  1.46it/s]
 12%|█▏        | 893/7150 [12:54<1:15:15,  1.39it/s]
 13%|█▎        | 894/7150 [12:54<1:13:59,  1.41it/s]
 13%|█▎        | 895/7150 [12:55<1:13:18,  1.42it/s]
 13%|█▎        | 896/7150 [12:56<1:11:44,  1.45it/s]
 13%|█▎        | 897/7150 [12:56<1:12:45,  1.43it/s]
 13%|█▎        | 898/7150 [12:57<1:13:25,  1.42it/s]
 13%|█▎        | 899/7150 [12:58<1:13:54,  1.41it/s]
 13%|█▎        | 900/7150 [12:59<1:14:20,  1.40it/s]
                                                    
{'loss': '3.858', 'grad_norm': '0.2495', 'learning_rate': '0.0006', 'epoch': '6.294'}
+
 13%|█▎        | 900/7150 [12:59<1:14:20,  1.40it/s]
 13%|█▎        | 901/7150 [12:59<1:12:29,  1.44it/s]
 13%|█▎        | 902/7150 [13:00<1:13:41,  1.41it/s]
 13%|█▎        | 903/7150 [13:01<1:13:00,  1.43it/s]
 13%|█▎        | 904/7150 [13:01<1:12:15,  1.44it/s]
 13%|█▎        | 905/7150 [13:02<1:11:49,  1.45it/s]
 13%|█▎        | 906/7150 [13:03<1:12:01,  1.44it/s]
 13%|█▎        | 907/7150 [13:03<1:12:03,  1.44it/s]
 13%|█▎        | 908/7150 [13:04<1:10:51,  1.47it/s]
 13%|█▎        | 909/7150 [13:05<1:13:17,  1.42it/s]
 13%|█▎        | 910/7150 [13:06<1:12:29,  1.43it/s]
                                                    
{'loss': '3.843', 'grad_norm': '0.326', 'learning_rate': '0.0006', 'epoch': '6.364'}
+
 13%|█▎        | 910/7150 [13:06<1:12:29,  1.43it/s]
 13%|█▎        | 911/7150 [13:06<1:12:09,  1.44it/s]
 13%|█▎        | 912/7150 [13:07<1:12:54,  1.43it/s]
 13%|█▎        | 913/7150 [13:08<1:12:00,  1.44it/s]
 13%|█▎        | 914/7150 [13:08<1:10:47,  1.47it/s]
 13%|█▎        | 915/7150 [13:09<1:10:45,  1.47it/s]
 13%|█▎        | 916/7150 [13:10<1:12:36,  1.43it/s]
 13%|█▎        | 917/7150 [13:10<1:12:31,  1.43it/s]
 13%|█▎        | 918/7150 [13:11<1:13:09,  1.42it/s]
 13%|█▎        | 919/7150 [13:12<1:12:29,  1.43it/s]
 13%|█▎        | 920/7150 [13:12<1:13:08,  1.42it/s]
                                                    
{'loss': '3.844', 'grad_norm': '0.2872', 'learning_rate': '0.0006', 'epoch': '6.434'}
+
 13%|█▎        | 920/7150 [13:12<1:13:08,  1.42it/s]
 13%|█▎        | 921/7150 [13:13<1:11:36,  1.45it/s]
 13%|█▎        | 922/7150 [13:14<1:11:20,  1.46it/s]
 13%|█▎        | 923/7150 [13:15<1:13:25,  1.41it/s]
 13%|█▎        | 924/7150 [13:15<1:13:25,  1.41it/s]
 13%|█▎        | 925/7150 [13:16<1:12:42,  1.43it/s]
 13%|█▎        | 926/7150 [13:17<1:11:12,  1.46it/s]
 13%|█▎        | 927/7150 [13:17<1:11:01,  1.46it/s]
 13%|█▎        | 928/7150 [13:18<1:12:01,  1.44it/s]
 13%|█▎        | 929/7150 [13:19<1:11:06,  1.46it/s]
 13%|█▎        | 930/7150 [13:19<1:10:46,  1.46it/s]
                                                    
{'loss': '3.826', 'grad_norm': '0.239', 'learning_rate': '0.0006', 'epoch': '6.504'}
+
 13%|█▎        | 930/7150 [13:19<1:10:46,  1.46it/s]
 13%|█▎        | 931/7150 [13:20<1:13:34,  1.41it/s]
 13%|█▎        | 932/7150 [13:21<1:13:53,  1.40it/s]
 13%|█▎        | 933/7150 [13:22<1:13:42,  1.41it/s]
 13%|█▎        | 934/7150 [13:22<1:11:53,  1.44it/s]
 13%|█▎        | 935/7150 [13:23<1:11:34,  1.45it/s]
 13%|█▎        | 936/7150 [13:24<1:11:37,  1.45it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.13it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.92it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 12.56it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 12.07it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.13it/s]
+
 33%|███▎      | 13/39 [00:01<00:02, 12.17it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.18it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 11.20it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 10.78it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 11.20it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 11.53it/s]
+
 64%|██████▍   | 25/39 [00:02<00:01, 11.74it/s]
+
 69%|██████▉   | 27/39 [00:02<00:01, 11.03it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 10.69it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 11.11it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 11.44it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 11.66it/s]
+
 95%|█████████▍| 37/39 [00:03<00:00, 11.82it/s]
+
100%|██████████| 39/39 [00:03<00:00, 11.74it/s]
                                                    
+
                                               
{'eval_loss': '4.018', 'eval_runtime': '3.493', 'eval_samples_per_second': '698.7', 'eval_steps_per_second': '11.16', 'epoch': '6.546'}
+
 13%|█▎        | 936/7150 [13:27<1:11:37,  1.45it/s]
+
100%|██████████| 39/39 [00:03<00:00, 11.74it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-936
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-936/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-936/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.92it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.92it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-936/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-936/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-936/tokenizer_config.json
+[eval] epoch 6.546 step 936 loss 4.0182 (best 4.0182 @ ep 6.546, 0/20 since best)
+
+
                                               
 13%|█▎        | 937/7150 [13:30<4:02:06,  2.34s/it]
 13%|█▎        | 938/7150 [13:30<3:11:56,  1.85s/it]
 13%|█▎        | 939/7150 [13:31<2:37:43,  1.52s/it]
 13%|█▎        | 940/7150 [13:32<2:10:38,  1.26s/it]
                                                    
{'loss': '3.842', 'grad_norm': '0.2785', 'learning_rate': '0.0006', 'epoch': '6.574'}
+
 13%|█▎        | 940/7150 [13:32<2:10:38,  1.26s/it]
 13%|█▎        | 941/7150 [13:33<1:56:22,  1.12s/it]
 13%|█▎        | 942/7150 [13:33<1:41:50,  1.02it/s]
 13%|█▎        | 943/7150 [13:34<1:32:08,  1.12it/s]
 13%|█▎        | 944/7150 [13:35<1:25:44,  1.21it/s]
 13%|█▎        | 945/7150 [13:35<1:21:09,  1.27it/s]
 13%|█▎        | 946/7150 [13:36<1:19:15,  1.30it/s]
 13%|█▎        | 947/7150 [13:37<1:15:46,  1.36it/s]
 13%|█▎        | 948/7150 [13:38<1:20:25,  1.29it/s]
 13%|█▎        | 949/7150 [13:38<1:21:53,  1.26it/s]
 13%|█▎        | 950/7150 [13:39<1:18:36,  1.31it/s]
                                                    
{'loss': '3.828', 'grad_norm': '0.2764', 'learning_rate': '0.0006', 'epoch': '6.644'}
+
 13%|█▎        | 950/7150 [13:39<1:18:36,  1.31it/s]
 13%|█▎        | 951/7150 [13:40<1:15:19,  1.37it/s]
 13%|█▎        | 952/7150 [13:41<1:17:35,  1.33it/s]
 13%|█▎        | 953/7150 [13:41<1:17:01,  1.34it/s]
 13%|█▎        | 954/7150 [13:42<1:14:10,  1.39it/s]
 13%|█▎        | 955/7150 [13:43<1:14:34,  1.38it/s]
 13%|█▎        | 956/7150 [13:43<1:14:13,  1.39it/s]
 13%|█▎        | 957/7150 [13:44<1:15:40,  1.36it/s]
 13%|█▎        | 958/7150 [13:45<1:14:58,  1.38it/s]
 13%|█▎        | 959/7150 [13:46<1:13:58,  1.39it/s]
 13%|█▎        | 960/7150 [13:46<1:15:04,  1.37it/s]
                                                    
{'loss': '3.821', 'grad_norm': '0.2789', 'learning_rate': '0.0006', 'epoch': '6.715'}
+
 13%|█▎        | 960/7150 [13:46<1:15:04,  1.37it/s]
 13%|█▎        | 961/7150 [13:47<1:12:50,  1.42it/s]
 13%|█▎        | 962/7150 [13:48<1:11:39,  1.44it/s]
 13%|█▎        | 963/7150 [13:48<1:11:55,  1.43it/s]
 13%|█▎        | 964/7150 [13:49<1:12:43,  1.42it/s]
 13%|█▎        | 965/7150 [13:50<1:13:01,  1.41it/s]
 14%|█▎        | 966/7150 [13:51<1:11:21,  1.44it/s]
 14%|█▎        | 967/7150 [13:51<1:10:53,  1.45it/s]
 14%|█▎        | 968/7150 [13:52<1:10:45,  1.46it/s]
 14%|█▎        | 969/7150 [13:53<1:09:45,  1.48it/s]
 14%|█▎        | 970/7150 [13:53<1:10:34,  1.46it/s]
                                                    
{'loss': '3.809', 'grad_norm': '0.2643', 'learning_rate': '0.0006', 'epoch': '6.785'}
+
 14%|█▎        | 970/7150 [13:53<1:10:34,  1.46it/s]
 14%|█▎        | 971/7150 [13:54<1:12:59,  1.41it/s]
 14%|█▎        | 972/7150 [13:55<1:11:16,  1.44it/s]
 14%|█▎        | 973/7150 [13:55<1:13:04,  1.41it/s]
 14%|█▎        | 974/7150 [13:56<1:13:22,  1.40it/s]
 14%|█▎        | 975/7150 [13:57<1:14:23,  1.38it/s]
 14%|█▎        | 976/7150 [13:58<1:14:20,  1.38it/s]
 14%|█▎        | 977/7150 [13:58<1:12:14,  1.42it/s]
 14%|█▎        | 978/7150 [13:59<1:12:50,  1.41it/s]
 14%|█▎        | 979/7150 [14:00<1:11:30,  1.44it/s]
 14%|█▎        | 980/7150 [14:00<1:10:12,  1.46it/s]
                                                    
{'loss': '3.812', 'grad_norm': '0.2798', 'learning_rate': '0.0006', 'epoch': '6.855'}
+
 14%|█▎        | 980/7150 [14:00<1:10:12,  1.46it/s]
 14%|█▎        | 981/7150 [14:01<1:13:19,  1.40it/s]
 14%|█▎        | 982/7150 [14:02<1:12:06,  1.43it/s]
 14%|█▎        | 983/7150 [14:02<1:11:15,  1.44it/s]
 14%|█▍        | 984/7150 [14:03<1:10:05,  1.47it/s]
 14%|█▍        | 985/7150 [14:04<1:12:11,  1.42it/s]
 14%|█▍        | 986/7150 [14:05<1:11:53,  1.43it/s]
 14%|█▍        | 987/7150 [14:05<1:10:30,  1.46it/s]
 14%|█▍        | 988/7150 [14:06<1:13:46,  1.39it/s]
 14%|█▍        | 989/7150 [14:07<1:14:59,  1.37it/s]
 14%|█▍        | 990/7150 [14:07<1:13:23,  1.40it/s]
                                                    
{'loss': '3.815', 'grad_norm': '0.3343', 'learning_rate': '0.0006', 'epoch': '6.925'}
+
 14%|█▍        | 990/7150 [14:07<1:13:23,  1.40it/s]
 14%|█▍        | 991/7150 [14:08<1:11:56,  1.43it/s]
 14%|█▍        | 992/7150 [14:09<1:11:59,  1.43it/s]
 14%|█▍        | 993/7150 [14:10<1:13:47,  1.39it/s]
 14%|█▍        | 994/7150 [14:10<1:11:44,  1.43it/s]
 14%|█▍        | 995/7150 [14:11<1:17:00,  1.33it/s]
 14%|█▍        | 996/7150 [14:12<1:15:43,  1.35it/s]
 14%|█▍        | 997/7150 [14:12<1:15:14,  1.36it/s]
 14%|█▍        | 998/7150 [14:13<1:12:46,  1.41it/s]
 14%|█▍        | 999/7150 [14:14<1:13:18,  1.40it/s]
 14%|█▍        | 1000/7150 [14:15<1:13:44,  1.39it/s]
                                                     
{'loss': '3.797', 'grad_norm': '0.2409', 'learning_rate': '0.0006', 'epoch': '6.995'}
+
 14%|█▍        | 1000/7150 [14:15<1:13:44,  1.39it/s]
 14%|█▍        | 1001/7150 [14:15<1:06:46,  1.53it/s]
 14%|█▍        | 1002/7150 [14:16<1:12:16,  1.42it/s]
 14%|█▍        | 1003/7150 [14:17<1:12:52,  1.41it/s]
 14%|█▍        | 1004/7150 [14:17<1:11:07,  1.44it/s]
 14%|█▍        | 1005/7150 [14:18<1:12:23,  1.41it/s]
 14%|█▍        | 1006/7150 [14:19<1:11:24,  1.43it/s]
 14%|█▍        | 1007/7150 [14:19<1:11:25,  1.43it/s]
 14%|█▍        | 1008/7150 [14:20<1:15:20,  1.36it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.57it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.81it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.62it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 11.74it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 10.93it/s]
+
 33%|███▎      | 13/39 [00:01<00:02, 11.33it/s]
+
 38%|███▊      | 15/39 [00:01<00:02, 11.60it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 11.79it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 11.93it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.00it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 10.41it/s]
+
 64%|██████▍   | 25/39 [00:02<00:01, 10.72it/s]
+
 69%|██████▉   | 27/39 [00:02<00:01, 11.13it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 11.44it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 11.67it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 11.84it/s]
+
 90%|████████▉ | 35/39 [00:03<00:00, 11.02it/s]
+
 95%|█████████▍| 37/39 [00:03<00:00, 10.78it/s]
+
100%|██████████| 39/39 [00:03<00:00, 11.17it/s]
                                                     
+
                                               
{'eval_loss': '3.978', 'eval_runtime': '3.528', 'eval_samples_per_second': '691.9', 'eval_steps_per_second': '11.05', 'epoch': '7.049'}
+
 14%|█▍        | 1008/7150 [14:24<1:15:20,  1.36it/s]
+
100%|██████████| 39/39 [00:03<00:00, 11.17it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-1008
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-1008/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-1008/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.89it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.89it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-1008/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-1008/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-1008/tokenizer_config.json
+[eval] epoch 7.049 step 1008 loss 3.9777 (best 3.9777 @ ep 7.049, 0/20 since best)
+
+
                                               
 14%|█▍        | 1009/7150 [14:27<4:05:30,  2.40s/it]
 14%|█▍        | 1010/7150 [14:27<3:12:12,  1.88s/it]
                                                     
{'loss': '3.739', 'grad_norm': '0.4963', 'learning_rate': '0.0006', 'epoch': '7.063'}
+
 14%|█▍        | 1010/7150 [14:27<3:12:12,  1.88s/it]
 14%|█▍        | 1011/7150 [14:28<2:37:02,  1.53s/it]
 14%|█▍        | 1012/7150 [14:29<2:12:11,  1.29s/it]
 14%|█▍        | 1013/7150 [14:29<1:54:35,  1.12s/it]
 14%|█▍        | 1014/7150 [14:30<1:41:04,  1.01it/s]
 14%|█▍        | 1015/7150 [14:31<1:31:59,  1.11it/s]
 14%|█▍        | 1016/7150 [14:31<1:26:55,  1.18it/s]
 14%|█▍        | 1017/7150 [14:32<1:22:53,  1.23it/s]
 14%|█▍        | 1018/7150 [14:33<1:20:17,  1.27it/s]
 14%|█▍        | 1019/7150 [14:34<1:18:22,  1.30it/s]
 14%|█▍        | 1020/7150 [14:34<1:20:04,  1.28it/s]
                                                     
{'loss': '3.752', 'grad_norm': '0.3556', 'learning_rate': '0.0006', 'epoch': '7.133'}
+
 14%|█▍        | 1020/7150 [14:34<1:20:04,  1.28it/s]
 14%|█▍        | 1021/7150 [14:35<1:17:58,  1.31it/s]
 14%|█▍        | 1022/7150 [14:36<1:19:16,  1.29it/s]
 14%|█▍        | 1023/7150 [14:37<1:18:02,  1.31it/s]
 14%|█▍        | 1024/7150 [14:37<1:16:54,  1.33it/s]
 14%|█▍        | 1025/7150 [14:38<1:13:52,  1.38it/s]
 14%|█▍        | 1026/7150 [14:39<1:13:59,  1.38it/s]
 14%|█▍        | 1027/7150 [14:40<1:14:03,  1.38it/s]
 14%|█▍        | 1028/7150 [14:40<1:14:28,  1.37it/s]
 14%|█▍        | 1029/7150 [14:41<1:14:44,  1.36it/s]
 14%|█▍        | 1030/7150 [14:42<1:12:20,  1.41it/s]
                                                     
{'loss': '3.74', 'grad_norm': '0.2989', 'learning_rate': '0.0006', 'epoch': '7.203'}
+
 14%|█▍        | 1030/7150 [14:42<1:12:20,  1.41it/s]
 14%|█▍        | 1031/7150 [14:42<1:13:10,  1.39it/s]
 14%|█▍        | 1032/7150 [14:43<1:13:56,  1.38it/s]
 14%|█▍        | 1033/7150 [14:44<1:13:50,  1.38it/s]
 14%|█▍        | 1034/7150 [14:45<1:12:53,  1.40it/s]
 14%|█▍        | 1035/7150 [14:45<1:11:35,  1.42it/s]
 14%|█▍        | 1036/7150 [14:46<1:12:24,  1.41it/s]
 15%|█▍        | 1037/7150 [14:47<1:10:42,  1.44it/s]
 15%|█▍        | 1038/7150 [14:47<1:12:19,  1.41it/s]
 15%|█▍        | 1039/7150 [14:48<1:13:01,  1.39it/s]
 15%|█▍        | 1040/7150 [14:49<1:11:03,  1.43it/s]
                                                     
{'loss': '3.723', 'grad_norm': '0.224', 'learning_rate': '0.0006', 'epoch': '7.273'}
+
 15%|█▍        | 1040/7150 [14:49<1:11:03,  1.43it/s]
 15%|█▍        | 1041/7150 [14:49<1:11:56,  1.42it/s]
 15%|█▍        | 1042/7150 [14:50<1:15:07,  1.36it/s]
 15%|█▍        | 1043/7150 [14:51<1:12:32,  1.40it/s]
 15%|█▍        | 1044/7150 [14:52<1:13:23,  1.39it/s]
 15%|█▍        | 1045/7150 [14:53<1:15:59,  1.34it/s]
 15%|█▍        | 1046/7150 [14:53<1:13:09,  1.39it/s]
 15%|█▍        | 1047/7150 [14:54<1:13:16,  1.39it/s]
 15%|█▍        | 1048/7150 [14:55<1:13:13,  1.39it/s]
 15%|█▍        | 1049/7150 [14:55<1:16:00,  1.34it/s]
 15%|█▍        | 1050/7150 [14:56<1:15:05,  1.35it/s]
                                                     
{'loss': '3.723', 'grad_norm': '0.2214', 'learning_rate': '0.0006', 'epoch': '7.343'}
+
 15%|█▍        | 1050/7150 [14:56<1:15:05,  1.35it/s]
 15%|█▍        | 1051/7150 [14:57<1:12:34,  1.40it/s]
 15%|█▍        | 1052/7150 [14:58<1:13:09,  1.39it/s]
 15%|█▍        | 1053/7150 [14:58<1:11:55,  1.41it/s]
 15%|█▍        | 1054/7150 [14:59<1:11:24,  1.42it/s]
 15%|█▍        | 1055/7150 [15:00<1:12:08,  1.41it/s]
 15%|█▍        | 1056/7150 [15:00<1:12:59,  1.39it/s]
 15%|█▍        | 1057/7150 [15:01<1:12:05,  1.41it/s]
 15%|█▍        | 1058/7150 [15:02<1:11:31,  1.42it/s]
 15%|█▍        | 1059/7150 [15:03<1:19:25,  1.28it/s]
 15%|█▍        | 1060/7150 [15:04<1:23:58,  1.21it/s]
                                                     
{'loss': '3.714', 'grad_norm': '0.2411', 'learning_rate': '0.0006', 'epoch': '7.413'}
+
 15%|█▍        | 1060/7150 [15:04<1:23:58,  1.21it/s]
 15%|█▍        | 1061/7150 [15:04<1:19:24,  1.28it/s]
 15%|█▍        | 1062/7150 [15:05<1:22:14,  1.23it/s]
 15%|█▍        | 1063/7150 [15:06<1:18:03,  1.30it/s]
 15%|█▍        | 1064/7150 [15:07<1:17:37,  1.31it/s]
 15%|█▍        | 1065/7150 [15:07<1:14:12,  1.37it/s]
 15%|█▍        | 1066/7150 [15:08<1:13:21,  1.38it/s]
 15%|█▍        | 1067/7150 [15:09<1:12:42,  1.39it/s]
 15%|█▍        | 1068/7150 [15:09<1:12:06,  1.41it/s]
 15%|█▍        | 1069/7150 [15:10<1:11:52,  1.41it/s]
 15%|█▍        | 1070/7150 [15:11<1:11:34,  1.42it/s]
                                                     
{'loss': '3.705', 'grad_norm': '0.2451', 'learning_rate': '0.0006', 'epoch': '7.483'}
+
 15%|█▍        | 1070/7150 [15:11<1:11:34,  1.42it/s]
 15%|█▍        | 1071/7150 [15:11<1:11:17,  1.42it/s]
 15%|█▍        | 1072/7150 [15:12<1:11:55,  1.41it/s]
 15%|█▌        | 1073/7150 [15:13<1:12:40,  1.39it/s]
 15%|█▌        | 1074/7150 [15:14<1:12:11,  1.40it/s]
 15%|█▌        | 1075/7150 [15:14<1:10:22,  1.44it/s]
 15%|█▌        | 1076/7150 [15:15<1:11:56,  1.41it/s]
 15%|█▌        | 1077/7150 [15:16<1:11:35,  1.41it/s]
 15%|█▌        | 1078/7150 [15:16<1:09:59,  1.45it/s]
 15%|█▌        | 1079/7150 [15:17<1:14:36,  1.36it/s]
 15%|█▌        | 1080/7150 [15:18<1:12:35,  1.39it/s]
                                                     
{'loss': '3.71', 'grad_norm': '0.2829', 'learning_rate': '0.0006', 'epoch': '7.553'}
+
 15%|█▌        | 1080/7150 [15:18<1:12:35,  1.39it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.65it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 12.77it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 10.88it/s]
+
 23%|██▎       | 9/39 [00:00<00:03,  9.87it/s]
+
 28%|██▊       | 11/39 [00:01<00:03,  9.19it/s]
+
 33%|███▎      | 13/39 [00:01<00:02,  9.09it/s]
+
 36%|███▌      | 14/39 [00:01<00:02,  9.02it/s]
+
 38%|███▊      | 15/39 [00:01<00:02,  9.08it/s]
+
 44%|████▎     | 17/39 [00:01<00:02,  9.98it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 10.66it/s]
+
 54%|█████▍    | 21/39 [00:02<00:01, 11.15it/s]
+
 59%|█████▉    | 23/39 [00:02<00:01, 11.18it/s]
+
 64%|██████▍   | 25/39 [00:02<00:01, 10.66it/s]
+
 69%|██████▉   | 27/39 [00:02<00:01, 11.10it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 11.44it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 11.67it/s]
+
 85%|████████▍ | 33/39 [00:03<00:00, 10.88it/s]
+
 90%|████████▉ | 35/39 [00:03<00:00, 10.23it/s]
+
 95%|█████████▍| 37/39 [00:03<00:00,  9.76it/s]
+
 97%|█████████▋| 38/39 [00:03<00:00,  9.43it/s]
+
100%|██████████| 39/39 [00:03<00:00,  9.14it/s]
                                                     
+
                                               
{'eval_loss': '3.932', 'eval_runtime': '3.956', 'eval_samples_per_second': '617.1', 'eval_steps_per_second': '9.859', 'epoch': '7.553'}
+
 15%|█▌        | 1080/7150 [15:22<1:12:35,  1.39it/s]
+
100%|██████████| 39/39 [00:03<00:00,  9.14it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-1080
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-1080/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-1080/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.94it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.94it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-1080/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-1080/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-1080/tokenizer_config.json
+[eval] epoch 7.553 step 1080 loss 3.9318 (best 3.9318 @ ep 7.553, 0/20 since best)
+
+
                                               
 15%|█▌        | 1081/7150 [15:25<4:20:00,  2.57s/it]
 15%|█▌        | 1082/7150 [15:26<3:23:17,  2.01s/it]
 15%|█▌        | 1083/7150 [15:26<2:49:40,  1.68s/it]
 15%|█▌        | 1084/7150 [15:27<2:23:16,  1.42s/it]
 15%|█▌        | 1085/7150 [15:28<2:05:59,  1.25s/it]
 15%|█▌        | 1086/7150 [15:29<1:52:56,  1.12s/it]
 15%|█▌        | 1087/7150 [15:30<1:42:49,  1.02s/it]
 15%|█▌        | 1088/7150 [15:31<1:39:58,  1.01it/s]
 15%|█▌        | 1089/7150 [15:31<1:30:59,  1.11it/s]
 15%|█▌        | 1090/7150 [15:32<1:32:58,  1.09it/s]
                                                     
{'loss': '3.719', 'grad_norm': '0.3431', 'learning_rate': '0.0006', 'epoch': '7.623'}
+
 15%|█▌        | 1090/7150 [15:32<1:32:58,  1.09it/s]
 15%|█▌        | 1091/7150 [15:33<1:34:25,  1.07it/s]
 15%|█▌        | 1092/7150 [15:34<1:35:16,  1.06it/s]
 15%|█▌        | 1093/7150 [15:35<1:36:01,  1.05it/s]
 15%|█▌        | 1094/7150 [15:36<1:27:29,  1.15it/s]
 15%|█▌        | 1095/7150 [15:37<1:30:07,  1.12it/s]
 15%|█▌        | 1096/7150 [15:37<1:24:30,  1.19it/s]
 15%|█▌        | 1097/7150 [15:38<1:26:48,  1.16it/s]
 15%|█▌        | 1098/7150 [15:39<1:29:53,  1.12it/s]
 15%|█▌        | 1099/7150 [15:40<1:23:14,  1.21it/s]
 15%|█▌        | 1100/7150 [15:41<1:21:25,  1.24it/s]
                                                     
{'loss': '3.71', 'grad_norm': '0.254', 'learning_rate': '0.0006', 'epoch': '7.694'}
+
 15%|█▌        | 1100/7150 [15:41<1:21:25,  1.24it/s]
 15%|█▌        | 1101/7150 [15:42<1:22:41,  1.22it/s]
 15%|█▌        | 1102/7150 [15:43<1:26:12,  1.17it/s]
 15%|█▌        | 1103/7150 [15:44<1:28:30,  1.14it/s]
 15%|█▌        | 1104/7150 [15:44<1:23:07,  1.21it/s]
 15%|█▌        | 1105/7150 [15:45<1:26:28,  1.17it/s]
 15%|█▌        | 1106/7150 [15:46<1:29:50,  1.12it/s]
 15%|█▌        | 1107/7150 [15:47<1:22:40,  1.22it/s]
 15%|█▌        | 1108/7150 [15:48<1:27:00,  1.16it/s]
 16%|█▌        | 1109/7150 [15:49<1:29:20,  1.13it/s]
 16%|█▌        | 1110/7150 [15:50<1:31:39,  1.10it/s]
                                                     
{'loss': '3.687', 'grad_norm': '0.2629', 'learning_rate': '0.0006', 'epoch': '7.764'}
+
 16%|█▌        | 1110/7150 [15:50<1:31:39,  1.10it/s]
 16%|█▌        | 1111/7150 [15:51<1:32:38,  1.09it/s]
 16%|█▌        | 1112/7150 [15:51<1:29:38,  1.12it/s]
 16%|█▌        | 1113/7150 [15:52<1:27:02,  1.16it/s]
 16%|█▌        | 1114/7150 [15:53<1:30:04,  1.12it/s]
 16%|█▌        | 1115/7150 [15:54<1:28:08,  1.14it/s]
 16%|█▌        | 1116/7150 [15:55<1:22:55,  1.21it/s]
 16%|█▌        | 1117/7150 [15:55<1:18:51,  1.27it/s]
 16%|█▌        | 1118/7150 [15:56<1:16:58,  1.31it/s]
 16%|█▌        | 1119/7150 [15:57<1:20:58,  1.24it/s]
 16%|█▌        | 1120/7150 [15:58<1:17:23,  1.30it/s]
                                                     
{'loss': '3.688', 'grad_norm': '0.237', 'learning_rate': '0.0006', 'epoch': '7.834'}
+
 16%|█▌        | 1120/7150 [15:58<1:17:23,  1.30it/s]
 16%|█▌        | 1121/7150 [15:58<1:16:22,  1.32it/s]
 16%|█▌        | 1122/7150 [15:59<1:15:19,  1.33it/s]
 16%|█▌        | 1123/7150 [16:00<1:13:22,  1.37it/s]
 16%|█▌        | 1124/7150 [16:01<1:18:14,  1.28it/s]
 16%|█▌        | 1125/7150 [16:02<1:22:29,  1.22it/s]
 16%|█▌        | 1126/7150 [16:02<1:19:55,  1.26it/s]
 16%|█▌        | 1127/7150 [16:03<1:16:23,  1.31it/s]
 16%|█▌        | 1128/7150 [16:04<1:13:45,  1.36it/s]
 16%|█▌        | 1129/7150 [16:04<1:11:40,  1.40it/s]
 16%|█▌        | 1130/7150 [16:05<1:09:51,  1.44it/s]
                                                     
{'loss': '3.696', 'grad_norm': '0.3052', 'learning_rate': '0.0006', 'epoch': '7.904'}
+
 16%|█▌        | 1130/7150 [16:05<1:09:51,  1.44it/s]
 16%|█▌        | 1131/7150 [16:06<1:08:54,  1.46it/s]
 16%|█▌        | 1132/7150 [16:06<1:08:28,  1.46it/s]
 16%|█▌        | 1133/7150 [16:07<1:07:56,  1.48it/s]
 16%|█▌        | 1134/7150 [16:08<1:07:16,  1.49it/s]
 16%|█▌        | 1135/7150 [16:08<1:07:14,  1.49it/s]
 16%|█▌        | 1136/7150 [16:09<1:07:23,  1.49it/s]
 16%|█▌        | 1137/7150 [16:10<1:06:49,  1.50it/s]
 16%|█▌        | 1138/7150 [16:10<1:06:50,  1.50it/s]
 16%|█▌        | 1139/7150 [16:11<1:07:26,  1.49it/s]
 16%|█▌        | 1140/7150 [16:12<1:07:50,  1.48it/s]
                                                     
{'loss': '3.688', 'grad_norm': '0.2111', 'learning_rate': '0.0006', 'epoch': '7.974'}
+
 16%|█▌        | 1140/7150 [16:12<1:07:50,  1.48it/s]
 16%|█▌        | 1141/7150 [16:12<1:07:11,  1.49it/s]
 16%|█▌        | 1142/7150 [16:13<1:07:43,  1.48it/s]
 16%|█▌        | 1143/7150 [16:14<1:08:31,  1.46it/s]
 16%|█▌        | 1144/7150 [16:14<1:03:03,  1.59it/s]
 16%|█▌        | 1145/7150 [16:15<1:06:49,  1.50it/s]
 16%|█▌        | 1146/7150 [16:16<1:06:23,  1.51it/s]
 16%|█▌        | 1147/7150 [16:16<1:06:35,  1.50it/s]
 16%|█▌        | 1148/7150 [16:17<1:06:45,  1.50it/s]
 16%|█▌        | 1149/7150 [16:18<1:06:52,  1.50it/s]
 16%|█▌        | 1150/7150 [16:18<1:06:26,  1.51it/s]
                                                     
{'loss': '3.637', 'grad_norm': '0.423', 'learning_rate': '0.0006', 'epoch': '8.042'}
+
 16%|█▌        | 1150/7150 [16:18<1:06:26,  1.51it/s]
 16%|█▌        | 1151/7150 [16:19<1:07:47,  1.47it/s]
 16%|█▌        | 1152/7150 [16:20<1:07:43,  1.48it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.47it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.81it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.60it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 12.42it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.35it/s]
+
 33%|███▎      | 13/39 [00:01<00:02, 12.31it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.28it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.27it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.01it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 11.82it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 11.97it/s]
+
 64%|██████▍   | 25/39 [00:02<00:01, 12.05it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.10it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 11.74it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 11.89it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 11.99it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.06it/s]
+
 95%|█████████▍| 37/39 [00:03<00:00, 11.29it/s]
+
100%|██████████| 39/39 [00:03<00:00, 11.11it/s]
                                                     
+
                                               
{'eval_loss': '3.904', 'eval_runtime': '3.384', 'eval_samples_per_second': '721.3', 'eval_steps_per_second': '11.52', 'epoch': '8.056'}
+
 16%|█▌        | 1152/7150 [16:23<1:07:43,  1.48it/s]
+
100%|██████████| 39/39 [00:03<00:00, 11.11it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-1152
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-1152/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-1152/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.90it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.90it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-1152/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-1152/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-1152/tokenizer_config.json
+[eval] epoch 8.056 step 1152 loss 3.9036 (best 3.9036 @ ep 8.056, 0/20 since best)
+
+
                                               
 16%|█▌        | 1153/7150 [16:26<3:50:04,  2.30s/it]
 16%|█▌        | 1154/7150 [16:27<3:01:01,  1.81s/it]
 16%|█▌        | 1155/7150 [16:27<2:28:06,  1.48s/it]
 16%|█▌        | 1156/7150 [16:28<2:04:13,  1.24s/it]
 16%|█▌        | 1157/7150 [16:29<1:47:00,  1.07s/it]
 16%|█▌        | 1158/7150 [16:29<1:34:29,  1.06it/s]
 16%|█▌        | 1159/7150 [16:30<1:26:12,  1.16it/s]
 16%|█▌        | 1160/7150 [16:31<1:20:52,  1.23it/s]
                                                     
{'loss': '3.617', 'grad_norm': '0.3456', 'learning_rate': '0.0006', 'epoch': '8.112'}
+
 16%|█▌        | 1160/7150 [16:31<1:20:52,  1.23it/s]
 16%|█▌        | 1161/7150 [16:31<1:16:59,  1.30it/s]
 16%|█▋        | 1162/7150 [16:32<1:13:56,  1.35it/s]
 16%|█▋        | 1163/7150 [16:33<1:12:06,  1.38it/s]
 16%|█▋        | 1164/7150 [16:33<1:10:01,  1.42it/s]
 16%|█▋        | 1165/7150 [16:34<1:12:29,  1.38it/s]
 16%|█▋        | 1166/7150 [16:35<1:10:20,  1.42it/s]
 16%|█▋        | 1167/7150 [16:35<1:10:24,  1.42it/s]
 16%|█▋        | 1168/7150 [16:36<1:09:56,  1.43it/s]
 16%|█▋        | 1169/7150 [16:37<1:08:29,  1.46it/s]
 16%|█▋        | 1170/7150 [16:38<1:08:40,  1.45it/s]
                                                     
{'loss': '3.613', 'grad_norm': '0.3348', 'learning_rate': '0.0006', 'epoch': '8.182'}
+
 16%|█▋        | 1170/7150 [16:38<1:08:40,  1.45it/s]
 16%|█▋        | 1171/7150 [16:38<1:08:10,  1.46it/s]
 16%|█▋        | 1172/7150 [16:39<1:08:05,  1.46it/s]
 16%|█▋        | 1173/7150 [16:40<1:11:00,  1.40it/s]
 16%|█▋        | 1174/7150 [16:40<1:09:14,  1.44it/s]
 16%|█▋        | 1175/7150 [16:41<1:08:25,  1.46it/s]
 16%|█▋        | 1176/7150 [16:42<1:08:29,  1.45it/s]
 16%|█▋        | 1177/7150 [16:42<1:09:38,  1.43it/s]
 16%|█▋        | 1178/7150 [16:43<1:08:38,  1.45it/s]
 16%|█▋        | 1179/7150 [16:44<1:07:33,  1.47it/s]
 17%|█▋        | 1180/7150 [16:44<1:07:28,  1.47it/s]
                                                     
{'loss': '3.618', 'grad_norm': '0.3299', 'learning_rate': '0.0006', 'epoch': '8.252'}
+
 17%|█▋        | 1180/7150 [16:44<1:07:28,  1.47it/s]
 17%|█▋        | 1181/7150 [16:45<1:07:38,  1.47it/s]
 17%|█▋        | 1182/7150 [16:46<1:07:27,  1.47it/s]
 17%|█▋        | 1183/7150 [16:46<1:07:43,  1.47it/s]
 17%|█▋        | 1184/7150 [16:47<1:07:46,  1.47it/s]
 17%|█▋        | 1185/7150 [16:48<1:06:56,  1.49it/s]
 17%|█▋        | 1186/7150 [16:48<1:07:28,  1.47it/s]
 17%|█▋        | 1187/7150 [16:49<1:07:48,  1.47it/s]
 17%|█▋        | 1188/7150 [16:50<1:06:59,  1.48it/s]
 17%|█▋        | 1189/7150 [16:51<1:07:19,  1.48it/s]
 17%|█▋        | 1190/7150 [16:51<1:07:31,  1.47it/s]
                                                     
{'loss': '3.615', 'grad_norm': '0.2556', 'learning_rate': '0.0006', 'epoch': '8.322'}
+
 17%|█▋        | 1190/7150 [16:51<1:07:31,  1.47it/s]
 17%|█▋        | 1191/7150 [16:52<1:07:51,  1.46it/s]
 17%|█▋        | 1192/7150 [16:53<1:06:58,  1.48it/s]
 17%|█▋        | 1193/7150 [16:53<1:07:06,  1.48it/s]
 17%|█▋        | 1194/7150 [16:54<1:07:21,  1.47it/s]
 17%|█▋        | 1195/7150 [16:55<1:07:34,  1.47it/s]
 17%|█▋        | 1196/7150 [16:55<1:06:47,  1.49it/s]
 17%|█▋        | 1197/7150 [16:56<1:07:55,  1.46it/s]
 17%|█▋        | 1198/7150 [16:57<1:07:42,  1.46it/s]
 17%|█▋        | 1199/7150 [16:57<1:06:49,  1.48it/s]
 17%|█▋        | 1200/7150 [16:58<1:07:03,  1.48it/s]
                                                     
{'loss': '3.612', 'grad_norm': '0.2396', 'learning_rate': '0.0006', 'epoch': '8.392'}
+
 17%|█▋        | 1200/7150 [16:58<1:07:03,  1.48it/s]
 17%|█▋        | 1201/7150 [16:59<1:07:13,  1.47it/s]
 17%|█▋        | 1202/7150 [16:59<1:07:14,  1.47it/s]
 17%|█▋        | 1203/7150 [17:00<1:07:23,  1.47it/s]
 17%|█▋        | 1204/7150 [17:01<1:07:19,  1.47it/s]
 17%|█▋        | 1205/7150 [17:01<1:07:22,  1.47it/s]
 17%|█▋        | 1206/7150 [17:02<1:06:34,  1.49it/s]
 17%|█▋        | 1207/7150 [17:03<1:08:00,  1.46it/s]
 17%|█▋        | 1208/7150 [17:03<1:07:45,  1.46it/s]
 17%|█▋        | 1209/7150 [17:04<1:07:31,  1.47it/s]
 17%|█▋        | 1210/7150 [17:05<1:07:14,  1.47it/s]
                                                     
{'loss': '3.615', 'grad_norm': '0.2393', 'learning_rate': '0.0006', 'epoch': '8.462'}
+
 17%|█▋        | 1210/7150 [17:05<1:07:14,  1.47it/s]
 17%|█▋        | 1211/7150 [17:05<1:07:05,  1.48it/s]
 17%|█▋        | 1212/7150 [17:06<1:06:24,  1.49it/s]
 17%|█▋        | 1213/7150 [17:07<1:08:27,  1.45it/s]
 17%|█▋        | 1214/7150 [17:07<1:07:19,  1.47it/s]
 17%|█▋        | 1215/7150 [17:08<1:07:18,  1.47it/s]
 17%|█▋        | 1216/7150 [17:09<1:08:11,  1.45it/s]
 17%|█▋        | 1217/7150 [17:10<1:08:44,  1.44it/s]
 17%|█▋        | 1218/7150 [17:10<1:07:29,  1.46it/s]
 17%|█▋        | 1219/7150 [17:11<1:07:33,  1.46it/s]
 17%|█▋        | 1220/7150 [17:12<1:08:39,  1.44it/s]
                                                     
{'loss': '3.619', 'grad_norm': '0.2339', 'learning_rate': '0.0006', 'epoch': '8.532'}
+
 17%|█▋        | 1220/7150 [17:12<1:08:39,  1.44it/s]
 17%|█▋        | 1221/7150 [17:12<1:08:26,  1.44it/s]
 17%|█▋        | 1222/7150 [17:13<1:07:43,  1.46it/s]
 17%|█▋        | 1223/7150 [17:14<1:08:01,  1.45it/s]
 17%|█▋        | 1224/7150 [17:14<1:07:44,  1.46it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.71it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.83it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 12.55it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 12.42it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.36it/s]
+
 33%|███▎      | 13/39 [00:01<00:02, 12.31it/s]
+
 38%|███▊      | 15/39 [00:01<00:02, 11.54it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 11.75it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 11.90it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.00it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.09it/s]
+
 64%|██████▍   | 25/39 [00:02<00:01, 12.14it/s]
+
 69%|██████▉   | 27/39 [00:02<00:01, 11.45it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 11.68it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 11.85it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 11.96it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.04it/s]
+
 95%|█████████▍| 37/39 [00:03<00:00, 11.89it/s]
+
100%|██████████| 39/39 [00:03<00:00, 10.95it/s]
                                                     
+
                                               
{'eval_loss': '3.865', 'eval_runtime': '3.405', 'eval_samples_per_second': '717', 'eval_steps_per_second': '11.46', 'epoch': '8.56'}
+
 17%|█▋        | 1224/7150 [17:18<1:07:44,  1.46it/s]
+
100%|██████████| 39/39 [00:03<00:00, 10.95it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-1224
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-1224/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-1224/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.94it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.94it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-1224/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-1224/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-1224/tokenizer_config.json
+[eval] epoch 8.56 step 1224 loss 3.8647 (best 3.8647 @ ep 8.56, 0/20 since best)
+
+
                                               
 17%|█▋        | 1225/7150 [17:20<3:47:10,  2.30s/it]
 17%|█▋        | 1226/7150 [17:21<2:58:58,  1.81s/it]
 17%|█▋        | 1227/7150 [17:22<2:25:17,  1.47s/it]
 17%|█▋        | 1228/7150 [17:23<2:03:55,  1.26s/it]
 17%|█▋        | 1229/7150 [17:23<1:46:06,  1.08s/it]
 17%|█▋        | 1230/7150 [17:24<1:34:37,  1.04it/s]
                                                     
{'loss': '3.607', 'grad_norm': '0.2426', 'learning_rate': '0.0006', 'epoch': '8.602'}
+
 17%|█▋        | 1230/7150 [17:24<1:34:37,  1.04it/s]
 17%|█▋        | 1231/7150 [17:25<1:26:21,  1.14it/s]
 17%|█▋        | 1232/7150 [17:25<1:20:32,  1.22it/s]
 17%|█▋        | 1233/7150 [17:26<1:15:45,  1.30it/s]
 17%|█▋        | 1234/7150 [17:27<1:13:06,  1.35it/s]
 17%|█▋        | 1235/7150 [17:27<1:12:20,  1.36it/s]
 17%|█▋        | 1236/7150 [17:28<1:10:50,  1.39it/s]
 17%|█▋        | 1237/7150 [17:29<1:08:54,  1.43it/s]
 17%|█▋        | 1238/7150 [17:29<1:08:40,  1.43it/s]
 17%|█▋        | 1239/7150 [17:30<1:08:29,  1.44it/s]
 17%|█▋        | 1240/7150 [17:31<1:08:05,  1.45it/s]
                                                     
{'loss': '3.614', 'grad_norm': '0.2461', 'learning_rate': '0.0006', 'epoch': '8.673'}
+
 17%|█▋        | 1240/7150 [17:31<1:08:05,  1.45it/s]
 17%|█▋        | 1241/7150 [17:31<1:07:55,  1.45it/s]
 17%|█▋        | 1242/7150 [17:32<1:07:44,  1.45it/s]
 17%|█▋        | 1243/7150 [17:33<1:07:42,  1.45it/s]
 17%|█▋        | 1244/7150 [17:33<1:07:18,  1.46it/s]
 17%|█▋        | 1245/7150 [17:34<1:07:14,  1.46it/s]
 17%|█▋        | 1246/7150 [17:35<1:09:08,  1.42it/s]
 17%|█▋        | 1247/7150 [17:36<1:08:23,  1.44it/s]
 17%|█▋        | 1248/7150 [17:36<1:09:05,  1.42it/s]
 17%|█▋        | 1249/7150 [17:37<1:08:26,  1.44it/s]
 17%|█▋        | 1250/7150 [17:38<1:07:12,  1.46it/s]
                                                     
{'loss': '3.608', 'grad_norm': '0.2732', 'learning_rate': '0.0006', 'epoch': '8.743'}
+
 17%|█▋        | 1250/7150 [17:38<1:07:12,  1.46it/s]
 17%|█▋        | 1251/7150 [17:38<1:07:12,  1.46it/s]
 18%|█▊        | 1252/7150 [17:39<1:09:24,  1.42it/s]
 18%|█▊        | 1253/7150 [17:40<1:08:42,  1.43it/s]
 18%|█▊        | 1254/7150 [17:40<1:09:17,  1.42it/s]
 18%|█▊        | 1255/7150 [17:41<1:08:40,  1.43it/s]
 18%|█▊        | 1256/7150 [17:42<1:07:21,  1.46it/s]
 18%|█▊        | 1257/7150 [17:42<1:07:09,  1.46it/s]
 18%|█▊        | 1258/7150 [17:43<1:09:05,  1.42it/s]
 18%|█▊        | 1259/7150 [17:44<1:08:45,  1.43it/s]
 18%|█▊        | 1260/7150 [17:45<1:14:23,  1.32it/s]
                                                     
{'loss': '3.623', 'grad_norm': '0.2233', 'learning_rate': '0.0006', 'epoch': '8.813'}
+
 18%|█▊        | 1260/7150 [17:45<1:14:23,  1.32it/s]
 18%|█▊        | 1261/7150 [17:46<1:15:52,  1.29it/s]
 18%|█▊        | 1262/7150 [17:46<1:14:06,  1.32it/s]
 18%|█▊        | 1263/7150 [17:47<1:11:07,  1.38it/s]
 18%|█▊        | 1264/7150 [17:48<1:12:15,  1.36it/s]
 18%|█▊        | 1265/7150 [17:49<1:13:43,  1.33it/s]
 18%|█▊        | 1266/7150 [17:49<1:11:30,  1.37it/s]
 18%|█▊        | 1267/7150 [17:50<1:15:15,  1.30it/s]
 18%|█▊        | 1268/7150 [17:51<1:13:51,  1.33it/s]
 18%|█▊        | 1269/7150 [17:51<1:11:47,  1.37it/s]
 18%|█▊        | 1270/7150 [17:52<1:12:31,  1.35it/s]
                                                     
{'loss': '3.601', 'grad_norm': '0.3022', 'learning_rate': '0.0006', 'epoch': '8.883'}
+
 18%|█▊        | 1270/7150 [17:52<1:12:31,  1.35it/s]
 18%|█▊        | 1271/7150 [17:53<1:10:50,  1.38it/s]
 18%|█▊        | 1272/7150 [17:54<1:08:50,  1.42it/s]
 18%|█▊        | 1273/7150 [17:54<1:08:15,  1.44it/s]
 18%|█▊        | 1274/7150 [17:55<1:09:37,  1.41it/s]
 18%|█▊        | 1275/7150 [17:56<1:11:37,  1.37it/s]
 18%|█▊        | 1276/7150 [17:57<1:14:06,  1.32it/s]
 18%|█▊        | 1277/7150 [17:58<1:19:07,  1.24it/s]
 18%|█▊        | 1278/7150 [17:58<1:18:12,  1.25it/s]
 18%|█▊        | 1279/7150 [17:59<1:15:00,  1.30it/s]
 18%|█▊        | 1280/7150 [18:00<1:14:10,  1.32it/s]
                                                     
{'loss': '3.598', 'grad_norm': '0.2492', 'learning_rate': '0.0006', 'epoch': '8.953'}
+
 18%|█▊        | 1280/7150 [18:00<1:14:10,  1.32it/s]
 18%|█▊        | 1281/7150 [18:00<1:11:59,  1.36it/s]
 18%|█▊        | 1282/7150 [18:01<1:09:35,  1.41it/s]
 18%|█▊        | 1283/7150 [18:02<1:09:44,  1.40it/s]
 18%|█▊        | 1284/7150 [18:02<1:08:50,  1.42it/s]
 18%|█▊        | 1285/7150 [18:03<1:12:25,  1.35it/s]
 18%|█▊        | 1286/7150 [18:04<1:10:24,  1.39it/s]
 18%|█▊        | 1287/7150 [18:04<1:03:43,  1.53it/s]
 18%|█▊        | 1288/7150 [18:05<1:07:04,  1.46it/s]
 18%|█▊        | 1289/7150 [18:06<1:07:00,  1.46it/s]
 18%|█▊        | 1290/7150 [18:07<1:06:04,  1.48it/s]
                                                     
{'loss': '3.592', 'grad_norm': '0.253', 'learning_rate': '0.0006', 'epoch': '9.021'}
+
 18%|█▊        | 1290/7150 [18:07<1:06:04,  1.48it/s]
 18%|█▊        | 1291/7150 [18:07<1:06:12,  1.47it/s]
 18%|█▊        | 1292/7150 [18:08<1:06:12,  1.47it/s]
 18%|█▊        | 1293/7150 [18:09<1:06:19,  1.47it/s]
 18%|█▊        | 1294/7150 [18:09<1:06:22,  1.47it/s]
 18%|█▊        | 1295/7150 [18:10<1:06:26,  1.47it/s]
 18%|█▊        | 1296/7150 [18:11<1:05:40,  1.49it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:02, 17.63it/s]
+
 10%|█         | 4/39 [00:00<00:02, 13.73it/s]
+
 15%|█▌        | 6/39 [00:00<00:02, 12.98it/s]
+
 21%|██        | 8/39 [00:00<00:02, 12.68it/s]
+
 26%|██▌       | 10/39 [00:00<00:02, 12.23it/s]
+
 31%|███       | 12/39 [00:00<00:02, 11.60it/s]
+
 36%|███▌      | 14/39 [00:01<00:02, 11.78it/s]
+
 41%|████      | 16/39 [00:01<00:01, 11.93it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 12.01it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 12.07it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 11.67it/s]
+
 62%|██████▏   | 24/39 [00:01<00:01, 11.62it/s]
+
 67%|██████▋   | 26/39 [00:02<00:01, 11.79it/s]
+
 72%|███████▏  | 28/39 [00:02<00:00, 11.93it/s]
+
 77%|███████▋  | 30/39 [00:02<00:00, 12.01it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 11.37it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 11.62it/s]
+
 92%|█████████▏| 36/39 [00:03<00:00, 11.80it/s]
+
 97%|█████████▋| 38/39 [00:03<00:00, 11.92it/s]
                                                     
+
                                               
{'eval_loss': '3.844', 'eval_runtime': '3.401', 'eval_samples_per_second': '717.8', 'eval_steps_per_second': '11.47', 'epoch': '9.063'}
+
 18%|█▊        | 1296/7150 [18:14<1:05:40,  1.49it/s]
+
100%|██████████| 39/39 [00:03<00:00, 11.92it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-1296
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-1296/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-1296/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.89it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.88it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-1296/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-1296/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-1296/tokenizer_config.json
+[eval] epoch 9.063 step 1296 loss 3.8442 (best 3.8442 @ ep 9.063, 0/20 since best)
+
+
                                               
 18%|█▊        | 1297/7150 [18:17<3:46:20,  2.32s/it]
 18%|█▊        | 1298/7150 [18:17<2:58:17,  1.83s/it]
 18%|█▊        | 1299/7150 [18:18<2:24:45,  1.48s/it]
 18%|█▊        | 1300/7150 [18:19<2:02:14,  1.25s/it]
                                                     
{'loss': '3.508', 'grad_norm': '0.3442', 'learning_rate': '0.0006', 'epoch': '9.091'}
+
 18%|█▊        | 1300/7150 [18:19<2:02:14,  1.25s/it]
 18%|█▊        | 1301/7150 [18:20<1:46:15,  1.09s/it]
 18%|█▊        | 1302/7150 [18:20<1:35:30,  1.02it/s]
 18%|█▊        | 1303/7150 [18:21<1:25:56,  1.13it/s]
 18%|█▊        | 1304/7150 [18:22<1:19:57,  1.22it/s]
 18%|█▊        | 1305/7150 [18:22<1:17:51,  1.25it/s]
 18%|█▊        | 1306/7150 [18:23<1:14:23,  1.31it/s]
 18%|█▊        | 1307/7150 [18:24<1:11:12,  1.37it/s]
 18%|█▊        | 1308/7150 [18:24<1:09:53,  1.39it/s]
 18%|█▊        | 1309/7150 [18:25<1:08:54,  1.41it/s]
 18%|█▊        | 1310/7150 [18:26<1:08:16,  1.43it/s]
                                                     
{'loss': '3.541', 'grad_norm': '0.3581', 'learning_rate': '0.0006', 'epoch': '9.161'}
+
 18%|█▊        | 1310/7150 [18:26<1:08:16,  1.43it/s]
 18%|█▊        | 1311/7150 [18:26<1:06:57,  1.45it/s]
 18%|█▊        | 1312/7150 [18:27<1:08:40,  1.42it/s]
 18%|█▊        | 1313/7150 [18:28<1:07:59,  1.43it/s]
 18%|█▊        | 1314/7150 [18:29<1:07:19,  1.44it/s]
 18%|█▊        | 1315/7150 [18:29<1:06:59,  1.45it/s]
 18%|█▊        | 1316/7150 [18:30<1:06:47,  1.46it/s]
 18%|█▊        | 1317/7150 [18:31<1:07:21,  1.44it/s]
 18%|█▊        | 1318/7150 [18:31<1:06:14,  1.47it/s]
 18%|█▊        | 1319/7150 [18:32<1:06:07,  1.47it/s]
 18%|█▊        | 1320/7150 [18:33<1:06:14,  1.47it/s]
                                                     
{'loss': '3.549', 'grad_norm': '0.3581', 'learning_rate': '0.0006', 'epoch': '9.231'}
+
 18%|█▊        | 1320/7150 [18:33<1:06:14,  1.47it/s]
 18%|█▊        | 1321/7150 [18:33<1:06:12,  1.47it/s]
 18%|█▊        | 1322/7150 [18:34<1:06:25,  1.46it/s]
 19%|█▊        | 1323/7150 [18:35<1:06:26,  1.46it/s]
 19%|█▊        | 1324/7150 [18:35<1:08:23,  1.42it/s]
 19%|█▊        | 1325/7150 [18:36<1:08:40,  1.41it/s]
 19%|█▊        | 1326/7150 [18:37<1:07:57,  1.43it/s]
 19%|█▊        | 1327/7150 [18:37<1:07:13,  1.44it/s]
 19%|█▊        | 1328/7150 [18:38<1:06:07,  1.47it/s]
 19%|█▊        | 1329/7150 [18:39<1:15:34,  1.28it/s]
 19%|█▊        | 1330/7150 [18:40<1:12:48,  1.33it/s]
                                                     
{'loss': '3.545', 'grad_norm': '0.2235', 'learning_rate': '0.0006', 'epoch': '9.301'}
+
 19%|█▊        | 1330/7150 [18:40<1:12:48,  1.33it/s]
 19%|█▊        | 1331/7150 [18:41<1:11:04,  1.36it/s]
 19%|█▊        | 1332/7150 [18:41<1:09:23,  1.40it/s]
 19%|█▊        | 1333/7150 [18:42<1:08:14,  1.42it/s]
 19%|█▊        | 1334/7150 [18:43<1:06:50,  1.45it/s]
 19%|█▊        | 1335/7150 [18:43<1:06:42,  1.45it/s]
 19%|█▊        | 1336/7150 [18:44<1:07:38,  1.43it/s]
 19%|█▊        | 1337/7150 [18:45<1:07:22,  1.44it/s]
 19%|█▊        | 1338/7150 [18:45<1:06:11,  1.46it/s]
 19%|█▊        | 1339/7150 [18:46<1:07:07,  1.44it/s]
 19%|█▊        | 1340/7150 [18:47<1:09:49,  1.39it/s]
                                                     
{'loss': '3.525', 'grad_norm': '0.2122', 'learning_rate': '0.0006', 'epoch': '9.371'}
+
 19%|█▊        | 1340/7150 [18:47<1:09:49,  1.39it/s]
 19%|█▉        | 1341/7150 [18:48<1:10:00,  1.38it/s]
 19%|█▉        | 1342/7150 [18:48<1:08:42,  1.41it/s]
 19%|█▉        | 1343/7150 [18:49<1:10:50,  1.37it/s]
 19%|█▉        | 1344/7150 [18:50<1:09:13,  1.40it/s]
 19%|█▉        | 1345/7150 [18:50<1:07:26,  1.43it/s]
 19%|█▉        | 1346/7150 [18:51<1:07:02,  1.44it/s]
 19%|█▉        | 1347/7150 [18:52<1:07:55,  1.42it/s]
 19%|█▉        | 1348/7150 [18:52<1:09:04,  1.40it/s]
 19%|█▉        | 1349/7150 [18:53<1:07:22,  1.44it/s]
 19%|█▉        | 1350/7150 [18:54<1:06:54,  1.44it/s]
                                                     
{'loss': '3.546', 'grad_norm': '0.2407', 'learning_rate': '0.0006', 'epoch': '9.441'}
+
 19%|█▉        | 1350/7150 [18:54<1:06:54,  1.44it/s]
 19%|█▉        | 1351/7150 [18:54<1:06:34,  1.45it/s]
 19%|█▉        | 1352/7150 [18:55<1:06:23,  1.46it/s]
 19%|█▉        | 1353/7150 [18:56<1:05:59,  1.46it/s]
 19%|█▉        | 1354/7150 [18:57<1:05:30,  1.47it/s]
 19%|█▉        | 1355/7150 [18:57<1:05:32,  1.47it/s]
 19%|█▉        | 1356/7150 [18:58<1:07:29,  1.43it/s]
 19%|█▉        | 1357/7150 [18:59<1:07:39,  1.43it/s]
 19%|█▉        | 1358/7150 [18:59<1:06:41,  1.45it/s]
 19%|█▉        | 1359/7150 [19:00<1:06:06,  1.46it/s]
 19%|█▉        | 1360/7150 [19:01<1:05:16,  1.48it/s]
                                                     
{'loss': '3.532', 'grad_norm': '0.2913', 'learning_rate': '0.0006', 'epoch': '9.511'}
+
 19%|█▉        | 1360/7150 [19:01<1:05:16,  1.48it/s]
 19%|█▉        | 1361/7150 [19:01<1:05:20,  1.48it/s]
 19%|█▉        | 1362/7150 [19:02<1:05:14,  1.48it/s]
 19%|█▉        | 1363/7150 [19:03<1:07:04,  1.44it/s]
 19%|█▉        | 1364/7150 [19:03<1:06:35,  1.45it/s]
 19%|█▉        | 1365/7150 [19:04<1:09:46,  1.38it/s]
 19%|█▉        | 1366/7150 [19:05<1:07:47,  1.42it/s]
 19%|█▉        | 1367/7150 [19:06<1:07:04,  1.44it/s]
 19%|█▉        | 1368/7150 [19:07<1:15:50,  1.27it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:01, 19.59it/s]
+
 10%|█         | 4/39 [00:00<00:02, 14.21it/s]
+
 15%|█▌        | 6/39 [00:00<00:02, 13.20it/s]
+
 21%|██        | 8/39 [00:00<00:02, 12.82it/s]
+
 26%|██▌       | 10/39 [00:00<00:02, 12.59it/s]
+
 31%|███       | 12/39 [00:00<00:02, 11.03it/s]
+
 36%|███▌      | 14/39 [00:01<00:02, 11.39it/s]
+
 41%|████      | 16/39 [00:01<00:01, 11.65it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 11.82it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 11.36it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 10.46it/s]
+
 62%|██████▏   | 24/39 [00:02<00:01, 10.86it/s]
+
 67%|██████▋   | 26/39 [00:02<00:01, 11.25it/s]
+
 72%|███████▏  | 28/39 [00:02<00:00, 11.55it/s]
+
 77%|███████▋  | 30/39 [00:02<00:00, 11.75it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 11.90it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 11.47it/s]
+
 92%|█████████▏| 36/39 [00:03<00:00, 10.52it/s]
+
 97%|█████████▋| 38/39 [00:03<00:00, 10.00it/s]
                                                     
+
                                               
{'eval_loss': '3.823', 'eval_runtime': '3.581', 'eval_samples_per_second': '681.7', 'eval_steps_per_second': '10.89', 'epoch': '9.567'}
+
 19%|█▉        | 1368/7150 [19:10<1:15:50,  1.27it/s]
+
100%|██████████| 39/39 [00:03<00:00, 10.00it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-1368
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-1368/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-1368/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.68it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.68it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-1368/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-1368/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-1368/tokenizer_config.json
+[eval] epoch 9.567 step 1368 loss 3.8230 (best 3.8230 @ ep 9.567, 0/20 since best)
+
+
                                               
 19%|█▉        | 1369/7150 [19:13<3:56:31,  2.45s/it]
 19%|█▉        | 1370/7150 [19:14<3:05:54,  1.93s/it]
                                                     
{'loss': '3.541', 'grad_norm': '0.2733', 'learning_rate': '0.0006', 'epoch': '9.581'}
+
 19%|█▉        | 1370/7150 [19:14<3:05:54,  1.93s/it]
 19%|█▉        | 1371/7150 [19:14<2:30:38,  1.56s/it]
 19%|█▉        | 1372/7150 [19:15<2:07:01,  1.32s/it]
 19%|█▉        | 1373/7150 [19:16<1:47:45,  1.12s/it]
 19%|█▉        | 1374/7150 [19:16<1:38:28,  1.02s/it]
 19%|█▉        | 1375/7150 [19:17<1:34:45,  1.02it/s]
 19%|█▉        | 1376/7150 [19:18<1:25:46,  1.12it/s]
 19%|█▉        | 1377/7150 [19:19<1:20:34,  1.19it/s]
 19%|█▉        | 1378/7150 [19:19<1:16:10,  1.26it/s]
 19%|█▉        | 1379/7150 [19:20<1:12:13,  1.33it/s]
 19%|█▉        | 1380/7150 [19:21<1:19:43,  1.21it/s]
                                                     
{'loss': '3.532', 'grad_norm': '0.2232', 'learning_rate': '0.0006', 'epoch': '9.651'}
+
 19%|█▉        | 1380/7150 [19:21<1:19:43,  1.21it/s]
 19%|█▉        | 1381/7150 [19:22<1:15:28,  1.27it/s]
 19%|█▉        | 1382/7150 [19:22<1:12:14,  1.33it/s]
 19%|█▉        | 1383/7150 [19:23<1:09:24,  1.38it/s]
 19%|█▉        | 1384/7150 [19:24<1:09:52,  1.38it/s]
 19%|█▉        | 1385/7150 [19:25<1:08:37,  1.40it/s]
 19%|█▉        | 1386/7150 [19:25<1:08:00,  1.41it/s]
 19%|█▉        | 1387/7150 [19:26<1:08:01,  1.41it/s]
 19%|█▉        | 1388/7150 [19:27<1:08:21,  1.40it/s]
 19%|█▉        | 1389/7150 [19:27<1:07:14,  1.43it/s]
 19%|█▉        | 1390/7150 [19:28<1:05:56,  1.46it/s]
                                                     
{'loss': '3.535', 'grad_norm': '0.2303', 'learning_rate': '0.0006', 'epoch': '9.722'}
+
 19%|█▉        | 1390/7150 [19:28<1:05:56,  1.46it/s]
 19%|█▉        | 1391/7150 [19:29<1:10:21,  1.36it/s]
 19%|█▉        | 1392/7150 [19:30<1:17:54,  1.23it/s]
 19%|█▉        | 1393/7150 [19:31<1:20:10,  1.20it/s]
 19%|█▉        | 1394/7150 [19:31<1:15:38,  1.27it/s]
 20%|█▉        | 1395/7150 [19:32<1:14:52,  1.28it/s]
 20%|█▉        | 1396/7150 [19:33<1:13:19,  1.31it/s]
 20%|█▉        | 1397/7150 [19:34<1:11:46,  1.34it/s]
 20%|█▉        | 1398/7150 [19:34<1:09:04,  1.39it/s]
 20%|█▉        | 1399/7150 [19:35<1:12:03,  1.33it/s]
 20%|█▉        | 1400/7150 [19:36<1:11:37,  1.34it/s]
                                                     
{'loss': '3.523', 'grad_norm': '0.2399', 'learning_rate': '0.0006', 'epoch': '9.792'}
+
 20%|█▉        | 1400/7150 [19:36<1:11:37,  1.34it/s]
 20%|█▉        | 1401/7150 [19:37<1:13:45,  1.30it/s]
 20%|█▉        | 1402/7150 [19:37<1:11:53,  1.33it/s]
 20%|█▉        | 1403/7150 [19:38<1:10:38,  1.36it/s]
 20%|█▉        | 1404/7150 [19:39<1:09:47,  1.37it/s]
 20%|█▉        | 1405/7150 [19:40<1:14:18,  1.29it/s]
 20%|█▉        | 1406/7150 [19:41<1:17:27,  1.24it/s]
 20%|█▉        | 1407/7150 [19:41<1:13:30,  1.30it/s]
 20%|█▉        | 1408/7150 [19:42<1:12:39,  1.32it/s]
 20%|█▉        | 1409/7150 [19:43<1:11:01,  1.35it/s]
 20%|█▉        | 1410/7150 [19:44<1:15:06,  1.27it/s]
                                                     
{'loss': '3.522', 'grad_norm': '0.2355', 'learning_rate': '0.0006', 'epoch': '9.862'}
+
 20%|█▉        | 1410/7150 [19:44<1:15:06,  1.27it/s]
 20%|█▉        | 1411/7150 [19:44<1:12:44,  1.31it/s]
 20%|█▉        | 1412/7150 [19:45<1:12:02,  1.33it/s]
 20%|█▉        | 1413/7150 [19:46<1:11:23,  1.34it/s]
 20%|█▉        | 1414/7150 [19:46<1:09:54,  1.37it/s]
 20%|█▉        | 1415/7150 [19:47<1:08:43,  1.39it/s]
 20%|█▉        | 1416/7150 [19:48<1:09:21,  1.38it/s]
 20%|█▉        | 1417/7150 [19:49<1:07:15,  1.42it/s]
 20%|█▉        | 1418/7150 [19:49<1:08:23,  1.40it/s]
 20%|█▉        | 1419/7150 [19:50<1:09:15,  1.38it/s]
 20%|█▉        | 1420/7150 [19:51<1:09:16,  1.38it/s]
                                                     
{'loss': '3.541', 'grad_norm': '0.2226', 'learning_rate': '0.0006', 'epoch': '9.932'}
+
 20%|█▉        | 1420/7150 [19:51<1:09:16,  1.38it/s]
 20%|█▉        | 1421/7150 [19:51<1:07:27,  1.42it/s]
 20%|█▉        | 1422/7150 [19:52<1:10:51,  1.35it/s]
 20%|█▉        | 1423/7150 [19:53<1:08:18,  1.40it/s]
 20%|█▉        | 1424/7150 [19:54<1:21:41,  1.17it/s]
 20%|█▉        | 1425/7150 [19:55<1:18:37,  1.21it/s]
 20%|█▉        | 1426/7150 [19:56<1:16:13,  1.25it/s]
 20%|█▉        | 1427/7150 [19:56<1:13:12,  1.30it/s]
 20%|█▉        | 1428/7150 [19:57<1:10:53,  1.35it/s]
 20%|█▉        | 1429/7150 [19:58<1:10:46,  1.35it/s]
 20%|██        | 1430/7150 [19:58<1:05:17,  1.46it/s]
                                                     
{'loss': '3.519', 'grad_norm': '0.2503', 'learning_rate': '0.0006', 'epoch': '10'}
+
 20%|██        | 1430/7150 [19:58<1:05:17,  1.46it/s]
 20%|██        | 1431/7150 [19:59<1:07:46,  1.41it/s]
 20%|██        | 1432/7150 [20:00<1:08:39,  1.39it/s]
 20%|██        | 1433/7150 [20:00<1:08:54,  1.38it/s]
 20%|██        | 1434/7150 [20:01<1:06:54,  1.42it/s]
 20%|██        | 1435/7150 [20:02<1:08:00,  1.40it/s]
 20%|██        | 1436/7150 [20:03<1:06:16,  1.44it/s]
 20%|██        | 1437/7150 [20:03<1:07:31,  1.41it/s]
 20%|██        | 1438/7150 [20:04<1:08:16,  1.39it/s]
 20%|██        | 1439/7150 [20:05<1:06:28,  1.43it/s]
 20%|██        | 1440/7150 [20:06<1:14:51,  1.27it/s]
                                                     
{'loss': '3.454', 'grad_norm': '0.4185', 'learning_rate': '0.0006', 'epoch': '10.07'}
+
 20%|██        | 1440/7150 [20:06<1:14:51,  1.27it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:03, 11.28it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 11.45it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 11.76it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 11.92it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.04it/s]
+
 33%|███▎      | 13/39 [00:01<00:02, 12.09it/s]
+
 38%|███▊      | 15/39 [00:01<00:02, 11.10it/s]
+
 44%|████▎     | 17/39 [00:01<00:02, 10.36it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 10.87it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 11.25it/s]
+
 59%|█████▉    | 23/39 [00:02<00:01, 11.55it/s]
+
 64%|██████▍   | 25/39 [00:02<00:01, 11.75it/s]
+
 69%|██████▉   | 27/39 [00:02<00:01, 11.88it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 10.28it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 10.73it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 11.14it/s]
+
 90%|████████▉ | 35/39 [00:03<00:00, 11.44it/s]
+
 95%|█████████▍| 37/39 [00:03<00:00, 11.67it/s]
+
100%|██████████| 39/39 [00:03<00:00, 11.83it/s]
                                                     
+
                                               
{'eval_loss': '3.826', 'eval_runtime': '3.558', 'eval_samples_per_second': '686.1', 'eval_steps_per_second': '10.96', 'epoch': '10.07'}
+
 20%|██        | 1440/7150 [20:09<1:14:51,  1.27it/s]
+
100%|██████████| 39/39 [00:03<00:00, 11.83it/s][eval] epoch 10.07 step 1440 loss 3.8264 (best 3.8230 @ ep 9.567, 1/20 since best)
+
+
                                               
 20%|██        | 1441/7150 [20:10<3:03:02,  1.92s/it]
 20%|██        | 1442/7150 [20:11<2:39:19,  1.67s/it]
 20%|██        | 1443/7150 [20:12<2:12:34,  1.39s/it]
 20%|██        | 1444/7150 [20:13<1:51:25,  1.17s/it]
 20%|██        | 1445/7150 [20:14<1:44:48,  1.10s/it]
 20%|██        | 1446/7150 [20:14<1:31:59,  1.03it/s]
 20%|██        | 1447/7150 [20:15<1:25:39,  1.11it/s]
 20%|██        | 1448/7150 [20:16<1:21:15,  1.17it/s]
 20%|██        | 1449/7150 [20:17<1:18:05,  1.22it/s]
 20%|██        | 1450/7150 [20:17<1:18:33,  1.21it/s]
                                                     
{'loss': '3.474', 'grad_norm': '0.4169', 'learning_rate': '0.0006', 'epoch': '10.14'}
+
 20%|██        | 1450/7150 [20:17<1:18:33,  1.21it/s]
 20%|██        | 1451/7150 [20:18<1:13:43,  1.29it/s]
 20%|██        | 1452/7150 [20:19<1:15:23,  1.26it/s]
 20%|██        | 1453/7150 [20:20<1:18:57,  1.20it/s]
 20%|██        | 1454/7150 [20:20<1:13:55,  1.28it/s]
 20%|██        | 1455/7150 [20:21<1:13:08,  1.30it/s]
 20%|██        | 1456/7150 [20:22<1:09:49,  1.36it/s]
 20%|██        | 1457/7150 [20:23<1:10:20,  1.35it/s]
 20%|██        | 1458/7150 [20:23<1:10:20,  1.35it/s]
 20%|██        | 1459/7150 [20:24<1:10:13,  1.35it/s]
 20%|██        | 1460/7150 [20:25<1:10:12,  1.35it/s]
                                                     
{'loss': '3.474', 'grad_norm': '0.3406', 'learning_rate': '0.0006', 'epoch': '10.21'}
+
 20%|██        | 1460/7150 [20:25<1:10:12,  1.35it/s]
 20%|██        | 1461/7150 [20:25<1:07:46,  1.40it/s]
 20%|██        | 1462/7150 [20:26<1:08:29,  1.38it/s]
 20%|██        | 1463/7150 [20:27<1:09:09,  1.37it/s]
 20%|██        | 1464/7150 [20:28<1:06:59,  1.41it/s]
 20%|██        | 1465/7150 [20:28<1:08:20,  1.39it/s]
 21%|██        | 1466/7150 [20:29<1:07:39,  1.40it/s]
 21%|██        | 1467/7150 [20:30<1:07:12,  1.41it/s]
 21%|██        | 1468/7150 [20:31<1:15:56,  1.25it/s]
 21%|██        | 1469/7150 [20:32<1:19:50,  1.19it/s]
 21%|██        | 1470/7150 [20:32<1:16:51,  1.23it/s]
                                                     
{'loss': '3.461', 'grad_norm': '0.2127', 'learning_rate': '0.0006', 'epoch': '10.28'}
+
 21%|██        | 1470/7150 [20:32<1:16:51,  1.23it/s]
 21%|██        | 1471/7150 [20:33<1:13:40,  1.28it/s]
 21%|██        | 1472/7150 [20:34<1:11:01,  1.33it/s]
 21%|██        | 1473/7150 [20:35<1:13:21,  1.29it/s]
 21%|██        | 1474/7150 [20:35<1:11:14,  1.33it/s]
 21%|██        | 1475/7150 [20:36<1:12:05,  1.31it/s]
 21%|██        | 1476/7150 [20:37<1:11:14,  1.33it/s]
 21%|██        | 1477/7150 [20:38<1:10:42,  1.34it/s]
 21%|██        | 1478/7150 [20:38<1:10:59,  1.33it/s]
 21%|██        | 1479/7150 [20:39<1:08:15,  1.38it/s]
 21%|██        | 1480/7150 [20:40<1:09:09,  1.37it/s]
                                                     
{'loss': '3.467', 'grad_norm': '0.2185', 'learning_rate': '0.0006', 'epoch': '10.35'}
+
 21%|██        | 1480/7150 [20:40<1:09:09,  1.37it/s]
 21%|██        | 1481/7150 [20:41<1:09:40,  1.36it/s]
 21%|██        | 1482/7150 [20:41<1:07:17,  1.40it/s]
 21%|██        | 1483/7150 [20:42<1:08:28,  1.38it/s]
 21%|██        | 1484/7150 [20:43<1:09:14,  1.36it/s]
 21%|██        | 1485/7150 [20:43<1:09:42,  1.35it/s]
 21%|██        | 1486/7150 [20:44<1:09:34,  1.36it/s]
 21%|██        | 1487/7150 [20:45<1:10:02,  1.35it/s]
 21%|██        | 1488/7150 [20:46<1:11:16,  1.32it/s]
 21%|██        | 1489/7150 [20:46<1:09:57,  1.35it/s]
 21%|██        | 1490/7150 [20:47<1:10:18,  1.34it/s]
                                                     
{'loss': '3.461', 'grad_norm': '0.2329', 'learning_rate': '0.0006', 'epoch': '10.42'}
+
 21%|██        | 1490/7150 [20:47<1:10:18,  1.34it/s]
 21%|██        | 1491/7150 [20:48<1:10:25,  1.34it/s]
 21%|██        | 1492/7150 [20:49<1:13:21,  1.29it/s]
 21%|██        | 1493/7150 [20:50<1:12:11,  1.31it/s]
 21%|██        | 1494/7150 [20:50<1:11:45,  1.31it/s]
 21%|██        | 1495/7150 [20:51<1:11:02,  1.33it/s]
 21%|██        | 1496/7150 [20:52<1:10:23,  1.34it/s]
 21%|██        | 1497/7150 [20:52<1:07:48,  1.39it/s]
 21%|██        | 1498/7150 [20:53<1:08:36,  1.37it/s]
 21%|██        | 1499/7150 [20:54<1:07:21,  1.40it/s]
 21%|██        | 1500/7150 [20:55<1:07:19,  1.40it/s]
                                                     
{'loss': '3.478', 'grad_norm': '0.2337', 'learning_rate': '0.0006', 'epoch': '10.49'}
+
 21%|██        | 1500/7150 [20:55<1:07:19,  1.40it/s]
 21%|██        | 1501/7150 [20:55<1:10:29,  1.34it/s]
 21%|██        | 1502/7150 [20:56<1:07:48,  1.39it/s]
 21%|██        | 1503/7150 [20:57<1:08:31,  1.37it/s]
 21%|██        | 1504/7150 [20:57<1:07:27,  1.39it/s]
 21%|██        | 1505/7150 [20:58<1:08:31,  1.37it/s]
 21%|██        | 1506/7150 [20:59<1:11:07,  1.32it/s]
 21%|██        | 1507/7150 [21:00<1:09:38,  1.35it/s]
 21%|██        | 1508/7150 [21:00<1:09:53,  1.35it/s]
 21%|██        | 1509/7150 [21:01<1:11:26,  1.32it/s]
 21%|██        | 1510/7150 [21:02<1:12:17,  1.30it/s]
                                                     
{'loss': '3.467', 'grad_norm': '0.2897', 'learning_rate': '0.0006', 'epoch': '10.56'}
+
 21%|██        | 1510/7150 [21:02<1:12:17,  1.30it/s]
 21%|██        | 1511/7150 [21:03<1:13:40,  1.28it/s]
 21%|██        | 1512/7150 [21:04<1:10:11,  1.34it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.87it/s]
+
 13%|█▎        | 5/39 [00:00<00:03, 10.36it/s]
+
 18%|█▊        | 7/39 [00:00<00:03,  9.69it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 10.49it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 11.03it/s]
+
 33%|███▎      | 13/39 [00:01<00:02, 11.39it/s]
+
 38%|███▊      | 15/39 [00:01<00:02, 11.65it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 11.82it/s]
+
 49%|████▊     | 19/39 [00:01<00:02,  9.42it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01,  9.94it/s]
+
 59%|█████▉    | 23/39 [00:02<00:01, 10.56it/s]
+
 64%|██████▍   | 25/39 [00:02<00:01, 11.01it/s]
+
 69%|██████▉   | 27/39 [00:02<00:01, 11.36it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 11.61it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 11.79it/s]
+
 85%|████████▍ | 33/39 [00:03<00:00,  9.45it/s]
+
 90%|████████▉ | 35/39 [00:03<00:00, 10.00it/s]
+
 95%|█████████▍| 37/39 [00:03<00:00, 10.58it/s]
+
100%|██████████| 39/39 [00:03<00:00, 11.02it/s]
                                                     
+
                                               
{'eval_loss': '3.788', 'eval_runtime': '3.739', 'eval_samples_per_second': '652.9', 'eval_steps_per_second': '10.43', 'epoch': '10.57'}
+
 21%|██        | 1512/7150 [21:07<1:10:11,  1.34it/s]
+
100%|██████████| 39/39 [00:03<00:00, 11.02it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-1512
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-1512/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-1512/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.83it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.83it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-1512/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-1512/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-1512/tokenizer_config.json
+[eval] epoch 10.574 step 1512 loss 3.7882 (best 3.7882 @ ep 10.574, 0/20 since best)
+
+
                                               
 21%|██        | 1513/7150 [21:10<3:55:47,  2.51s/it]
 21%|██        | 1514/7150 [21:11<3:07:16,  1.99s/it]
 21%|██        | 1515/7150 [21:12<2:29:47,  1.59s/it]
 21%|██        | 1516/7150 [21:12<2:06:58,  1.35s/it]
 21%|██        | 1517/7150 [21:13<1:51:20,  1.19s/it]
 21%|██        | 1518/7150 [21:14<1:37:24,  1.04s/it]
 21%|██        | 1519/7150 [21:15<1:29:54,  1.04it/s]
 21%|██▏       | 1520/7150 [21:15<1:25:47,  1.09it/s]
                                                     
{'loss': '3.463', 'grad_norm': '0.2214', 'learning_rate': '0.0006', 'epoch': '10.63'}
+
 21%|██▏       | 1520/7150 [21:16<1:25:47,  1.09it/s]
 21%|██▏       | 1521/7150 [21:16<1:18:42,  1.19it/s]
 21%|██▏       | 1522/7150 [21:17<1:17:25,  1.21it/s]
 21%|██▏       | 1523/7150 [21:18<1:16:24,  1.23it/s]
 21%|██▏       | 1524/7150 [21:18<1:13:19,  1.28it/s]
 21%|██▏       | 1525/7150 [21:19<1:11:44,  1.31it/s]
 21%|██▏       | 1526/7150 [21:20<1:12:40,  1.29it/s]
 21%|██▏       | 1527/7150 [21:21<1:11:40,  1.31it/s]
 21%|██▏       | 1528/7150 [21:21<1:09:59,  1.34it/s]
 21%|██▏       | 1529/7150 [21:22<1:11:35,  1.31it/s]
 21%|██▏       | 1530/7150 [21:23<1:12:02,  1.30it/s]
                                                     
{'loss': '3.469', 'grad_norm': '0.2184', 'learning_rate': '0.0006', 'epoch': '10.7'}
+
 21%|██▏       | 1530/7150 [21:23<1:12:02,  1.30it/s]
 21%|██▏       | 1531/7150 [21:24<1:08:54,  1.36it/s]
 21%|██▏       | 1532/7150 [21:24<1:10:27,  1.33it/s]
 21%|██▏       | 1533/7150 [21:25<1:11:21,  1.31it/s]
 21%|██▏       | 1534/7150 [21:26<1:09:48,  1.34it/s]
 21%|██▏       | 1535/7150 [21:27<1:14:03,  1.26it/s]
 21%|██▏       | 1536/7150 [21:28<1:14:19,  1.26it/s]
 21%|██▏       | 1537/7150 [21:28<1:14:19,  1.26it/s]
 22%|██▏       | 1538/7150 [21:29<1:15:00,  1.25it/s]
 22%|██▏       | 1539/7150 [21:30<1:14:16,  1.26it/s]
 22%|██▏       | 1540/7150 [21:31<1:10:23,  1.33it/s]
                                                     
{'loss': '3.473', 'grad_norm': '0.2676', 'learning_rate': '0.0006', 'epoch': '10.77'}
+
 22%|██▏       | 1540/7150 [21:31<1:10:23,  1.33it/s]
 22%|██▏       | 1541/7150 [21:32<1:12:24,  1.29it/s]
 22%|██▏       | 1542/7150 [21:32<1:12:52,  1.28it/s]
 22%|██▏       | 1543/7150 [21:33<1:09:24,  1.35it/s]
 22%|██▏       | 1544/7150 [21:34<1:10:53,  1.32it/s]
 22%|██▏       | 1545/7150 [21:35<1:12:28,  1.29it/s]
 22%|██▏       | 1546/7150 [21:35<1:09:09,  1.35it/s]
 22%|██▏       | 1547/7150 [21:36<1:11:34,  1.30it/s]
 22%|██▏       | 1548/7150 [21:37<1:12:44,  1.28it/s]
 22%|██▏       | 1549/7150 [21:38<1:09:23,  1.35it/s]
 22%|██▏       | 1550/7150 [21:38<1:11:40,  1.30it/s]
                                                     
{'loss': '3.476', 'grad_norm': '0.2242', 'learning_rate': '0.0006', 'epoch': '10.84'}
+
 22%|██▏       | 1550/7150 [21:38<1:11:40,  1.30it/s]
 22%|██▏       | 1551/7150 [21:39<1:12:30,  1.29it/s]
 22%|██▏       | 1552/7150 [21:40<1:09:09,  1.35it/s]
 22%|██▏       | 1553/7150 [21:41<1:11:08,  1.31it/s]
 22%|██▏       | 1554/7150 [21:41<1:08:04,  1.37it/s]
 22%|██▏       | 1555/7150 [21:42<1:09:50,  1.34it/s]
 22%|██▏       | 1556/7150 [21:43<1:12:35,  1.28it/s]
 22%|██▏       | 1557/7150 [21:44<1:12:28,  1.29it/s]
 22%|██▏       | 1558/7150 [21:44<1:09:01,  1.35it/s]
 22%|██▏       | 1559/7150 [21:45<1:15:59,  1.23it/s]
 22%|██▏       | 1560/7150 [21:46<1:11:34,  1.30it/s]
                                                     
{'loss': '3.476', 'grad_norm': '0.2252', 'learning_rate': '0.0006', 'epoch': '10.91'}
+
 22%|██▏       | 1560/7150 [21:46<1:11:34,  1.30it/s]
 22%|██▏       | 1561/7150 [21:47<1:13:09,  1.27it/s]
 22%|██▏       | 1562/7150 [21:48<1:14:19,  1.25it/s]
 22%|██▏       | 1563/7150 [21:48<1:10:28,  1.32it/s]
 22%|██▏       | 1564/7150 [21:49<1:13:04,  1.27it/s]
 22%|██▏       | 1565/7150 [21:50<1:09:27,  1.34it/s]
 22%|██▏       | 1566/7150 [21:51<1:11:38,  1.30it/s]
 22%|██▏       | 1567/7150 [21:51<1:09:22,  1.34it/s]
 22%|██▏       | 1568/7150 [21:52<1:10:36,  1.32it/s]
 22%|██▏       | 1569/7150 [21:53<1:12:24,  1.28it/s]
 22%|██▏       | 1570/7150 [21:54<1:12:51,  1.28it/s]
                                                     
{'loss': '3.467', 'grad_norm': '0.239', 'learning_rate': '0.0006', 'epoch': '10.98'}
+
 22%|██▏       | 1570/7150 [21:54<1:12:51,  1.28it/s]
 22%|██▏       | 1571/7150 [21:54<1:09:21,  1.34it/s]
 22%|██▏       | 1572/7150 [21:55<1:07:23,  1.38it/s]
 22%|██▏       | 1573/7150 [21:56<1:01:45,  1.51it/s]
 22%|██▏       | 1574/7150 [21:56<1:04:46,  1.43it/s]
 22%|██▏       | 1575/7150 [21:57<1:04:48,  1.43it/s]
 22%|██▏       | 1576/7150 [21:58<1:04:35,  1.44it/s]
 22%|██▏       | 1577/7150 [21:58<1:04:20,  1.44it/s]
 22%|██▏       | 1578/7150 [21:59<1:05:33,  1.42it/s]
 22%|██▏       | 1579/7150 [22:00<1:04:36,  1.44it/s]
 22%|██▏       | 1580/7150 [22:01<1:03:53,  1.45it/s]
                                                     
{'loss': '3.416', 'grad_norm': '0.3444', 'learning_rate': '0.0006', 'epoch': '11.05'}
+
 22%|██▏       | 1580/7150 [22:01<1:03:53,  1.45it/s]
 22%|██▏       | 1581/7150 [22:01<1:02:54,  1.48it/s]
 22%|██▏       | 1582/7150 [22:02<1:02:46,  1.48it/s]
 22%|██▏       | 1583/7150 [22:03<1:06:59,  1.39it/s]
 22%|██▏       | 1584/7150 [22:03<1:05:22,  1.42it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.69it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.83it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.62it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 12.06it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.11it/s]
+
 33%|███▎      | 13/39 [00:01<00:02, 12.16it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.17it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.20it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.19it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 11.68it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 11.87it/s]
+
 64%|██████▍   | 25/39 [00:02<00:01, 11.98it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.05it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 11.10it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 10.29it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 10.70it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 11.13it/s]
+
 95%|█████████▍| 37/39 [00:03<00:00, 11.45it/s]
+
100%|██████████| 39/39 [00:03<00:00, 11.24it/s]
                                                     
+
                                               
{'eval_loss': '3.792', 'eval_runtime': '3.469', 'eval_samples_per_second': '703.8', 'eval_steps_per_second': '11.24', 'epoch': '11.08'}
+
 22%|██▏       | 1584/7150 [22:07<1:05:22,  1.42it/s]
+
100%|██████████| 39/39 [00:03<00:00, 11.24it/s][eval] epoch 11.077 step 1584 loss 3.7919 (best 3.7882 @ ep 10.574, 1/20 since best)
+
+
                                               
 22%|██▏       | 1585/7150 [22:07<2:40:27,  1.73s/it]
 22%|██▏       | 1586/7150 [22:08<2:11:07,  1.41s/it]
 22%|██▏       | 1587/7150 [22:09<1:50:25,  1.19s/it]
 22%|██▏       | 1588/7150 [22:09<1:36:00,  1.04s/it]
 22%|██▏       | 1589/7150 [22:10<1:25:21,  1.09it/s]
 22%|██▏       | 1590/7150 [22:11<1:19:01,  1.17it/s]
                                                     
{'loss': '3.386', 'grad_norm': '0.3096', 'learning_rate': '0.0006', 'epoch': '11.12'}
+
 22%|██▏       | 1590/7150 [22:11<1:19:01,  1.17it/s]
 22%|██▏       | 1591/7150 [22:12<1:14:27,  1.24it/s]
 22%|██▏       | 1592/7150 [22:12<1:10:40,  1.31it/s]
 22%|██▏       | 1593/7150 [22:13<1:09:13,  1.34it/s]
 22%|██▏       | 1594/7150 [22:14<1:10:32,  1.31it/s]
 22%|██▏       | 1595/7150 [22:14<1:08:48,  1.35it/s]
 22%|██▏       | 1596/7150 [22:15<1:06:53,  1.38it/s]
 22%|██▏       | 1597/7150 [22:16<1:07:05,  1.38it/s]
 22%|██▏       | 1598/7150 [22:17<1:06:22,  1.39it/s]
 22%|██▏       | 1599/7150 [22:17<1:06:41,  1.39it/s]
 22%|██▏       | 1600/7150 [22:18<1:04:51,  1.43it/s]
                                                     
{'loss': '3.388', 'grad_norm': '0.2671', 'learning_rate': '0.0006', 'epoch': '11.19'}
+
 22%|██▏       | 1600/7150 [22:18<1:04:51,  1.43it/s]
 22%|██▏       | 1601/7150 [22:19<1:04:09,  1.44it/s]
 22%|██▏       | 1602/7150 [22:19<1:03:30,  1.46it/s]
 22%|██▏       | 1603/7150 [22:20<1:03:05,  1.47it/s]
 22%|██▏       | 1604/7150 [22:21<1:03:46,  1.45it/s]
 22%|██▏       | 1605/7150 [22:21<1:03:16,  1.46it/s]
 22%|██▏       | 1606/7150 [22:22<1:03:01,  1.47it/s]
 22%|██▏       | 1607/7150 [22:23<1:02:24,  1.48it/s]
 22%|██▏       | 1608/7150 [22:23<1:01:56,  1.49it/s]
 23%|██▎       | 1609/7150 [22:24<1:01:53,  1.49it/s]
 23%|██▎       | 1610/7150 [22:25<1:01:43,  1.50it/s]
                                                     
{'loss': '3.392', 'grad_norm': '0.243', 'learning_rate': '0.0006', 'epoch': '11.26'}
+
 23%|██▎       | 1610/7150 [22:25<1:01:43,  1.50it/s]
 23%|██▎       | 1611/7150 [22:25<1:02:29,  1.48it/s]
 23%|██▎       | 1612/7150 [22:26<1:01:50,  1.49it/s]
 23%|██▎       | 1613/7150 [22:27<1:01:37,  1.50it/s]
 23%|██▎       | 1614/7150 [22:27<1:01:33,  1.50it/s]
 23%|██▎       | 1615/7150 [22:28<1:01:22,  1.50it/s]
 23%|██▎       | 1616/7150 [22:29<1:01:05,  1.51it/s]
 23%|██▎       | 1617/7150 [22:29<1:01:35,  1.50it/s]
 23%|██▎       | 1618/7150 [22:30<1:01:32,  1.50it/s]
 23%|██▎       | 1619/7150 [22:31<1:01:29,  1.50it/s]
 23%|██▎       | 1620/7150 [22:31<1:01:09,  1.51it/s]
                                                     
{'loss': '3.401', 'grad_norm': '0.259', 'learning_rate': '0.0006', 'epoch': '11.33'}
+
 23%|██▎       | 1620/7150 [22:31<1:01:09,  1.51it/s]
 23%|██▎       | 1621/7150 [22:32<1:01:11,  1.51it/s]
 23%|██▎       | 1622/7150 [22:33<1:01:11,  1.51it/s]
 23%|██▎       | 1623/7150 [22:33<1:00:54,  1.51it/s]
 23%|██▎       | 1624/7150 [22:34<1:00:58,  1.51it/s]
 23%|██▎       | 1625/7150 [22:35<1:01:54,  1.49it/s]
 23%|██▎       | 1626/7150 [22:35<1:01:40,  1.49it/s]
 23%|██▎       | 1627/7150 [22:36<1:01:12,  1.50it/s]
 23%|██▎       | 1628/7150 [22:37<1:01:47,  1.49it/s]
 23%|██▎       | 1629/7150 [22:37<1:01:34,  1.49it/s]
 23%|██▎       | 1630/7150 [22:38<1:01:24,  1.50it/s]
                                                     
{'loss': '3.406', 'grad_norm': '0.2831', 'learning_rate': '0.0006', 'epoch': '11.4'}
+
 23%|██▎       | 1630/7150 [22:38<1:01:24,  1.50it/s]
 23%|██▎       | 1631/7150 [22:39<1:01:37,  1.49it/s]
 23%|██▎       | 1632/7150 [22:39<1:01:10,  1.50it/s]
 23%|██▎       | 1633/7150 [22:40<1:01:32,  1.49it/s]
 23%|██▎       | 1634/7150 [22:41<1:01:51,  1.49it/s]
 23%|██▎       | 1635/7150 [22:41<1:01:27,  1.50it/s]
 23%|██▎       | 1636/7150 [22:42<1:01:14,  1.50it/s]
 23%|██▎       | 1637/7150 [22:43<1:01:36,  1.49it/s]
 23%|██▎       | 1638/7150 [22:43<1:01:25,  1.50it/s]
 23%|██▎       | 1639/7150 [22:44<1:01:20,  1.50it/s]
 23%|██▎       | 1640/7150 [22:45<1:01:17,  1.50it/s]
                                                     
{'loss': '3.417', 'grad_norm': '0.2641', 'learning_rate': '0.0006', 'epoch': '11.47'}
+
 23%|██▎       | 1640/7150 [22:45<1:01:17,  1.50it/s]
 23%|██▎       | 1641/7150 [22:45<1:01:06,  1.50it/s]
 23%|██▎       | 1642/7150 [22:46<1:01:10,  1.50it/s]
 23%|██▎       | 1643/7150 [22:47<1:01:05,  1.50it/s]
 23%|██▎       | 1644/7150 [22:47<1:02:06,  1.48it/s]
 23%|██▎       | 1645/7150 [22:48<1:01:40,  1.49it/s]
 23%|██▎       | 1646/7150 [22:49<1:05:28,  1.40it/s]
 23%|██▎       | 1647/7150 [22:49<1:03:50,  1.44it/s]
 23%|██▎       | 1648/7150 [22:50<1:03:51,  1.44it/s]
 23%|██▎       | 1649/7150 [22:51<1:02:58,  1.46it/s]
 23%|██▎       | 1650/7150 [22:52<1:02:47,  1.46it/s]
                                                     
{'loss': '3.402', 'grad_norm': '0.2444', 'learning_rate': '0.0006', 'epoch': '11.54'}
+
 23%|██▎       | 1650/7150 [22:52<1:02:47,  1.46it/s]
 23%|██▎       | 1651/7150 [22:52<1:01:59,  1.48it/s]
 23%|██▎       | 1652/7150 [22:53<1:01:33,  1.49it/s]
 23%|██▎       | 1653/7150 [22:54<1:04:44,  1.41it/s]
 23%|██▎       | 1654/7150 [22:54<1:03:19,  1.45it/s]
 23%|██▎       | 1655/7150 [22:55<1:02:43,  1.46it/s]
 23%|██▎       | 1656/7150 [22:56<1:02:11,  1.47it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.40it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.75it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.59it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 12.21it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 11.10it/s]
+
 33%|███▎      | 13/39 [00:01<00:02, 11.13it/s]
+
 38%|███▊      | 15/39 [00:01<00:02, 11.46it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 11.70it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 11.86it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 11.31it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 11.60it/s]
+
 64%|██████▍   | 25/39 [00:02<00:01, 11.79it/s]
+
 69%|██████▉   | 27/39 [00:02<00:01, 11.92it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.01it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.08it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 11.27it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 10.68it/s]
+
 95%|█████████▍| 37/39 [00:03<00:00, 10.32it/s]
+
100%|██████████| 39/39 [00:03<00:00, 10.76it/s]
                                                     
+
                                               
{'eval_loss': '3.767', 'eval_runtime': '3.518', 'eval_samples_per_second': '693.8', 'eval_steps_per_second': '11.08', 'epoch': '11.58'}
+
 23%|██▎       | 1656/7150 [22:59<1:02:11,  1.47it/s]
+
100%|██████████| 39/39 [00:03<00:00, 10.76it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-1656
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-1656/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-1656/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.90it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.90it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-1656/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-1656/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-1656/tokenizer_config.json
+[eval] epoch 11.581 step 1656 loss 3.7666 (best 3.7666 @ ep 11.581, 0/20 since best)
+
+
                                               
 23%|██▎       | 1657/7150 [23:02<3:35:09,  2.35s/it]
 23%|██▎       | 1658/7150 [23:03<2:51:50,  1.88s/it]
 23%|██▎       | 1659/7150 [23:03<2:18:11,  1.51s/it]
 23%|██▎       | 1660/7150 [23:04<1:55:58,  1.27s/it]
                                                     
{'loss': '3.417', 'grad_norm': '0.2358', 'learning_rate': '0.0006', 'epoch': '11.61'}
+
 23%|██▎       | 1660/7150 [23:04<1:55:58,  1.27s/it]
 23%|██▎       | 1661/7150 [23:05<1:40:51,  1.10s/it]
 23%|██▎       | 1662/7150 [23:05<1:28:39,  1.03it/s]
 23%|██▎       | 1663/7150 [23:06<1:22:11,  1.11it/s]
 23%|██▎       | 1664/7150 [23:07<1:17:57,  1.17it/s]
 23%|██▎       | 1665/7150 [23:07<1:12:34,  1.26it/s]
 23%|██▎       | 1666/7150 [23:08<1:10:15,  1.30it/s]
 23%|██▎       | 1667/7150 [23:09<1:07:54,  1.35it/s]
 23%|██▎       | 1668/7150 [23:10<1:11:02,  1.29it/s]
 23%|██▎       | 1669/7150 [23:10<1:08:25,  1.33it/s]
 23%|██▎       | 1670/7150 [23:11<1:08:17,  1.34it/s]
                                                     
{'loss': '3.427', 'grad_norm': '0.2387', 'learning_rate': '0.0006', 'epoch': '11.68'}
+
 23%|██▎       | 1670/7150 [23:11<1:08:17,  1.34it/s]
 23%|██▎       | 1671/7150 [23:12<1:11:40,  1.27it/s]
 23%|██▎       | 1672/7150 [23:13<1:10:45,  1.29it/s]
 23%|██▎       | 1673/7150 [23:13<1:07:45,  1.35it/s]
 23%|██▎       | 1674/7150 [23:14<1:05:24,  1.40it/s]
 23%|██▎       | 1675/7150 [23:15<1:03:57,  1.43it/s]
 23%|██▎       | 1676/7150 [23:15<1:02:59,  1.45it/s]
 23%|██▎       | 1677/7150 [23:16<1:02:18,  1.46it/s]
 23%|██▎       | 1678/7150 [23:17<1:01:46,  1.48it/s]
 23%|██▎       | 1679/7150 [23:17<1:01:11,  1.49it/s]
 23%|██▎       | 1680/7150 [23:18<1:01:04,  1.49it/s]
                                                     
{'loss': '3.421', 'grad_norm': '0.2594', 'learning_rate': '0.0006', 'epoch': '11.75'}
+
 23%|██▎       | 1680/7150 [23:18<1:01:04,  1.49it/s]
 24%|██▎       | 1681/7150 [23:19<1:00:57,  1.50it/s]
 24%|██▎       | 1682/7150 [23:19<1:00:33,  1.51it/s]
 24%|██▎       | 1683/7150 [23:20<1:00:58,  1.49it/s]
 24%|██▎       | 1684/7150 [23:21<1:01:35,  1.48it/s]
 24%|██▎       | 1685/7150 [23:21<1:00:59,  1.49it/s]
 24%|██▎       | 1686/7150 [23:22<1:01:22,  1.48it/s]
 24%|██▎       | 1687/7150 [23:23<1:01:11,  1.49it/s]
 24%|██▎       | 1688/7150 [23:23<1:01:25,  1.48it/s]
 24%|██▎       | 1689/7150 [23:24<1:01:44,  1.47it/s]
 24%|██▎       | 1690/7150 [23:25<1:01:05,  1.49it/s]
                                                     
{'loss': '3.423', 'grad_norm': '0.2493', 'learning_rate': '0.0006', 'epoch': '11.82'}
+
 24%|██▎       | 1690/7150 [23:25<1:01:05,  1.49it/s]
 24%|██▎       | 1691/7150 [23:25<1:01:02,  1.49it/s]
 24%|██▎       | 1692/7150 [23:26<1:00:53,  1.49it/s]
 24%|██▎       | 1693/7150 [23:27<1:00:31,  1.50it/s]
 24%|██▎       | 1694/7150 [23:28<1:05:45,  1.38it/s]
 24%|██▎       | 1695/7150 [23:28<1:04:12,  1.42it/s]
 24%|██▎       | 1696/7150 [23:29<1:08:35,  1.33it/s]
 24%|██▎       | 1697/7150 [23:30<1:11:51,  1.26it/s]
 24%|██▎       | 1698/7150 [23:31<1:13:42,  1.23it/s]
 24%|██▍       | 1699/7150 [23:32<1:15:19,  1.21it/s]
 24%|██▍       | 1700/7150 [23:33<1:16:19,  1.19it/s]
                                                     
{'loss': '3.423', 'grad_norm': '0.2451', 'learning_rate': '0.0006', 'epoch': '11.89'}
+
 24%|██▍       | 1700/7150 [23:33<1:16:19,  1.19it/s]
 24%|██▍       | 1701/7150 [23:33<1:14:26,  1.22it/s]
 24%|██▍       | 1702/7150 [23:34<1:10:40,  1.28it/s]
 24%|██▍       | 1703/7150 [23:35<1:12:35,  1.25it/s]
 24%|██▍       | 1704/7150 [23:36<1:08:39,  1.32it/s]
 24%|██▍       | 1705/7150 [23:36<1:10:06,  1.29it/s]
 24%|██▍       | 1706/7150 [23:37<1:12:44,  1.25it/s]
 24%|██▍       | 1707/7150 [23:38<1:11:17,  1.27it/s]
 24%|██▍       | 1708/7150 [23:39<1:11:04,  1.28it/s]
 24%|██▍       | 1709/7150 [23:40<1:13:36,  1.23it/s]
 24%|██▍       | 1710/7150 [23:40<1:09:30,  1.30it/s]
                                                     
{'loss': '3.424', 'grad_norm': '0.228', 'learning_rate': '0.0006', 'epoch': '11.96'}
+
 24%|██▍       | 1710/7150 [23:40<1:09:30,  1.30it/s]
 24%|██▍       | 1711/7150 [23:41<1:12:38,  1.25it/s]
 24%|██▍       | 1712/7150 [23:42<1:14:56,  1.21it/s]
 24%|██▍       | 1713/7150 [23:43<1:16:01,  1.19it/s]
 24%|██▍       | 1714/7150 [23:44<1:17:07,  1.17it/s]
 24%|██▍       | 1715/7150 [23:45<1:11:57,  1.26it/s]
 24%|██▍       | 1716/7150 [23:45<1:04:09,  1.41it/s]
 24%|██▍       | 1717/7150 [23:46<1:05:21,  1.39it/s]
 24%|██▍       | 1718/7150 [23:46<1:03:52,  1.42it/s]
 24%|██▍       | 1719/7150 [23:47<1:02:29,  1.45it/s]
 24%|██▍       | 1720/7150 [23:48<1:02:12,  1.45it/s]
                                                     
{'loss': '3.39', 'grad_norm': '0.4046', 'learning_rate': '0.0006', 'epoch': '12.03'}
+
 24%|██▍       | 1720/7150 [23:48<1:02:12,  1.45it/s]
 24%|██▍       | 1721/7150 [23:49<1:02:00,  1.46it/s]
 24%|██▍       | 1722/7150 [23:49<1:01:46,  1.46it/s]
 24%|██▍       | 1723/7150 [23:50<1:01:41,  1.47it/s]
 24%|██▍       | 1724/7150 [23:51<1:02:00,  1.46it/s]
 24%|██▍       | 1725/7150 [23:51<1:03:01,  1.43it/s]
 24%|██▍       | 1726/7150 [23:52<1:05:55,  1.37it/s]
 24%|██▍       | 1727/7150 [23:53<1:04:13,  1.41it/s]
 24%|██▍       | 1728/7150 [23:53<1:02:55,  1.44it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.77it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.85it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.56it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 11.61it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 11.82it/s]
+
 33%|███▎      | 13/39 [00:01<00:02, 11.94it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.04it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 11.58it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 11.73it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 11.87it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 11.99it/s]
+
 64%|██████▍   | 25/39 [00:02<00:01, 12.06it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.11it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 11.26it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 10.53it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 10.09it/s]
+
 90%|████████▉ | 35/39 [00:03<00:00,  9.85it/s]
+
 92%|█████████▏| 36/39 [00:03<00:00,  9.63it/s]
+
 95%|█████████▍| 37/39 [00:03<00:00,  9.47it/s]
+
 97%|█████████▋| 38/39 [00:03<00:00,  9.35it/s]
                                                     
+
                                               
{'eval_loss': '3.776', 'eval_runtime': '3.646', 'eval_samples_per_second': '669.6', 'eval_steps_per_second': '10.7', 'epoch': '12.08'}
+
 24%|██▍       | 1728/7150 [23:57<1:02:55,  1.44it/s]
+
100%|██████████| 39/39 [00:03<00:00,  9.35it/s][eval] epoch 12.084 step 1728 loss 3.7764 (best 3.7666 @ ep 11.581, 1/20 since best)
+
+
                                               
 24%|██▍       | 1729/7150 [23:58<2:41:51,  1.79s/it]
 24%|██▍       | 1730/7150 [23:59<2:14:23,  1.49s/it]
                                                     
{'loss': '3.333', 'grad_norm': '0.3537', 'learning_rate': '0.0006', 'epoch': '12.1'}
+
 24%|██▍       | 1730/7150 [23:59<2:14:23,  1.49s/it]
 24%|██▍       | 1731/7150 [23:59<1:52:14,  1.24s/it]
 24%|██▍       | 1732/7150 [24:00<1:42:12,  1.13s/it]
 24%|██▍       | 1733/7150 [24:01<1:35:13,  1.05s/it]
 24%|██▍       | 1734/7150 [24:02<1:30:15,  1.00it/s]
 24%|██▍       | 1735/7150 [24:02<1:20:53,  1.12it/s]
 24%|██▍       | 1736/7150 [24:03<1:17:38,  1.16it/s]
 24%|██▍       | 1737/7150 [24:04<1:13:11,  1.23it/s]
 24%|██▍       | 1738/7150 [24:05<1:10:50,  1.27it/s]
 24%|██▍       | 1739/7150 [24:05<1:09:11,  1.30it/s]
 24%|██▍       | 1740/7150 [24:06<1:12:07,  1.25it/s]
                                                     
{'loss': '3.345', 'grad_norm': '0.2934', 'learning_rate': '0.0006', 'epoch': '12.17'}
+
 24%|██▍       | 1740/7150 [24:06<1:12:07,  1.25it/s]
 24%|██▍       | 1741/7150 [24:07<1:14:20,  1.21it/s]
 24%|██▍       | 1742/7150 [24:08<1:09:46,  1.29it/s]
 24%|██▍       | 1743/7150 [24:09<1:07:45,  1.33it/s]
 24%|██▍       | 1744/7150 [24:09<1:05:36,  1.37it/s]
 24%|██▍       | 1745/7150 [24:10<1:04:55,  1.39it/s]
 24%|██▍       | 1746/7150 [24:11<1:03:35,  1.42it/s]
 24%|██▍       | 1747/7150 [24:11<1:02:49,  1.43it/s]
 24%|██▍       | 1748/7150 [24:12<1:01:40,  1.46it/s]
 24%|██▍       | 1749/7150 [24:13<1:02:22,  1.44it/s]
 24%|██▍       | 1750/7150 [24:13<1:01:51,  1.45it/s]
                                                     
{'loss': '3.36', 'grad_norm': '0.302', 'learning_rate': '0.0006', 'epoch': '12.24'}
+
 24%|██▍       | 1750/7150 [24:13<1:01:51,  1.45it/s]
 24%|██▍       | 1751/7150 [24:14<1:01:36,  1.46it/s]
 25%|██▍       | 1752/7150 [24:15<1:01:54,  1.45it/s]
 25%|██▍       | 1753/7150 [24:15<1:01:31,  1.46it/s]
 25%|██▍       | 1754/7150 [24:16<1:01:25,  1.46it/s]
 25%|██▍       | 1755/7150 [24:17<1:01:13,  1.47it/s]
 25%|██▍       | 1756/7150 [24:17<1:01:51,  1.45it/s]
 25%|██▍       | 1757/7150 [24:18<1:00:59,  1.47it/s]
 25%|██▍       | 1758/7150 [24:19<1:00:50,  1.48it/s]
 25%|██▍       | 1759/7150 [24:19<1:01:38,  1.46it/s]
 25%|██▍       | 1760/7150 [24:20<1:01:24,  1.46it/s]
                                                     
{'loss': '3.35', 'grad_norm': '0.226', 'learning_rate': '0.0006', 'epoch': '12.31'}
+
 25%|██▍       | 1760/7150 [24:20<1:01:24,  1.46it/s]
 25%|██▍       | 1761/7150 [24:21<1:01:14,  1.47it/s]
 25%|██▍       | 1762/7150 [24:21<1:01:16,  1.47it/s]
 25%|██▍       | 1763/7150 [24:22<1:02:15,  1.44it/s]
 25%|██▍       | 1764/7150 [24:23<1:01:43,  1.45it/s]
 25%|██▍       | 1765/7150 [24:24<1:03:02,  1.42it/s]
 25%|██▍       | 1766/7150 [24:24<1:03:32,  1.41it/s]
 25%|██▍       | 1767/7150 [24:25<1:04:29,  1.39it/s]
 25%|██▍       | 1768/7150 [24:26<1:04:30,  1.39it/s]
 25%|██▍       | 1769/7150 [24:27<1:04:47,  1.38it/s]
 25%|██▍       | 1770/7150 [24:27<1:03:37,  1.41it/s]
                                                     
{'loss': '3.357', 'grad_norm': '0.234', 'learning_rate': '0.0006', 'epoch': '12.38'}
+
 25%|██▍       | 1770/7150 [24:27<1:03:37,  1.41it/s]
 25%|██▍       | 1771/7150 [24:28<1:02:36,  1.43it/s]
 25%|██▍       | 1772/7150 [24:29<1:02:05,  1.44it/s]
 25%|██▍       | 1773/7150 [24:29<1:01:42,  1.45it/s]
 25%|██▍       | 1774/7150 [24:30<1:03:38,  1.41it/s]
 25%|██▍       | 1775/7150 [24:31<1:02:24,  1.44it/s]
 25%|██▍       | 1776/7150 [24:31<1:02:33,  1.43it/s]
 25%|██▍       | 1777/7150 [24:32<1:04:14,  1.39it/s]
 25%|██▍       | 1778/7150 [24:33<1:07:17,  1.33it/s]
 25%|██▍       | 1779/7150 [24:34<1:06:43,  1.34it/s]
 25%|██▍       | 1780/7150 [24:34<1:05:34,  1.36it/s]
                                                     
{'loss': '3.354', 'grad_norm': '0.2279', 'learning_rate': '0.0006', 'epoch': '12.45'}
+
 25%|██▍       | 1780/7150 [24:34<1:05:34,  1.36it/s]
 25%|██▍       | 1781/7150 [24:35<1:09:08,  1.29it/s]
 25%|██▍       | 1782/7150 [24:36<1:12:33,  1.23it/s]
 25%|██▍       | 1783/7150 [24:37<1:11:36,  1.25it/s]
 25%|██▍       | 1784/7150 [24:38<1:08:13,  1.31it/s]
 25%|██▍       | 1785/7150 [24:38<1:05:57,  1.36it/s]
 25%|██▍       | 1786/7150 [24:39<1:06:24,  1.35it/s]
 25%|██▍       | 1787/7150 [24:40<1:08:17,  1.31it/s]
 25%|██▌       | 1788/7150 [24:41<1:06:36,  1.34it/s]
 25%|██▌       | 1789/7150 [24:41<1:07:00,  1.33it/s]
 25%|██▌       | 1790/7150 [24:42<1:11:01,  1.26it/s]
                                                     
{'loss': '3.361', 'grad_norm': '0.2319', 'learning_rate': '0.0006', 'epoch': '12.52'}
+
 25%|██▌       | 1790/7150 [24:42<1:11:01,  1.26it/s]
 25%|██▌       | 1791/7150 [24:43<1:13:35,  1.21it/s]
 25%|██▌       | 1792/7150 [24:44<1:09:01,  1.29it/s]
 25%|██▌       | 1793/7150 [24:44<1:06:21,  1.35it/s]
 25%|██▌       | 1794/7150 [24:45<1:05:11,  1.37it/s]
 25%|██▌       | 1795/7150 [24:46<1:04:24,  1.39it/s]
 25%|██▌       | 1796/7150 [24:47<1:05:30,  1.36it/s]
 25%|██▌       | 1797/7150 [24:47<1:05:07,  1.37it/s]
 25%|██▌       | 1798/7150 [24:48<1:03:30,  1.40it/s]
 25%|██▌       | 1799/7150 [24:49<1:07:33,  1.32it/s]
 25%|██▌       | 1800/7150 [24:50<1:10:35,  1.26it/s]
                                                     
{'loss': '3.369', 'grad_norm': '0.2885', 'learning_rate': '0.0006', 'epoch': '12.59'}
+
 25%|██▌       | 1800/7150 [24:50<1:10:35,  1.26it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:02, 17.70it/s]
+
 10%|█         | 4/39 [00:00<00:03, 11.61it/s]
+
 15%|█▌        | 6/39 [00:00<00:03, 10.35it/s]
+
 21%|██        | 8/39 [00:00<00:03,  9.88it/s]
+
 26%|██▌       | 10/39 [00:00<00:02, 10.13it/s]
+
 31%|███       | 12/39 [00:01<00:02, 10.77it/s]
+
 36%|███▌      | 14/39 [00:01<00:02, 11.21it/s]
+
 41%|████      | 16/39 [00:01<00:01, 11.52it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 11.36it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 11.61it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 11.78it/s]
+
 62%|██████▏   | 24/39 [00:02<00:01, 11.93it/s]
+
 67%|██████▋   | 26/39 [00:02<00:01, 12.02it/s]
+
 72%|███████▏  | 28/39 [00:02<00:00, 12.09it/s]
+
 77%|███████▋  | 30/39 [00:02<00:00, 11.34it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 11.61it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 11.80it/s]
+
 92%|█████████▏| 36/39 [00:03<00:00, 11.93it/s]
+
 97%|█████████▋| 38/39 [00:03<00:00, 11.69it/s]
                                                     
+
                                               
{'eval_loss': '3.753', 'eval_runtime': '3.531', 'eval_samples_per_second': '691.4', 'eval_steps_per_second': '11.05', 'epoch': '12.59'}
+
 25%|██▌       | 1800/7150 [24:53<1:10:35,  1.26it/s]
+
100%|██████████| 39/39 [00:03<00:00, 11.69it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-1800
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-1800/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-1800/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.64it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.64it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-1800/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-1800/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-1800/tokenizer_config.json
+[eval] epoch 12.588 step 1800 loss 3.7530 (best 3.7530 @ ep 12.588, 0/20 since best)
+
+
                                               
 25%|██▌       | 1801/7150 [24:56<3:39:05,  2.46s/it]
 25%|██▌       | 1802/7150 [24:57<2:50:50,  1.92s/it]
 25%|██▌       | 1803/7150 [24:57<2:17:47,  1.55s/it]
 25%|██▌       | 1804/7150 [24:58<1:54:20,  1.28s/it]
 25%|██▌       | 1805/7150 [24:59<1:37:52,  1.10s/it]
 25%|██▌       | 1806/7150 [24:59<1:26:22,  1.03it/s]
 25%|██▌       | 1807/7150 [25:00<1:18:16,  1.14it/s]
 25%|██▌       | 1808/7150 [25:01<1:12:18,  1.23it/s]
 25%|██▌       | 1809/7150 [25:01<1:08:49,  1.29it/s]
 25%|██▌       | 1810/7150 [25:02<1:05:57,  1.35it/s]
                                                     
{'loss': '3.374', 'grad_norm': '0.2284', 'learning_rate': '0.0006', 'epoch': '12.66'}
+
 25%|██▌       | 1810/7150 [25:02<1:05:57,  1.35it/s]
 25%|██▌       | 1811/7150 [25:03<1:03:56,  1.39it/s]
 25%|██▌       | 1812/7150 [25:03<1:02:27,  1.42it/s]
 25%|██▌       | 1813/7150 [25:04<1:01:11,  1.45it/s]
 25%|██▌       | 1814/7150 [25:05<1:01:55,  1.44it/s]
 25%|██▌       | 1815/7150 [25:05<1:01:06,  1.46it/s]
 25%|██▌       | 1816/7150 [25:06<1:00:12,  1.48it/s]
 25%|██▌       | 1817/7150 [25:07<59:52,  1.48it/s]  
 25%|██▌       | 1818/7150 [25:07<59:37,  1.49it/s]
 25%|██▌       | 1819/7150 [25:08<1:01:12,  1.45it/s]
 25%|██▌       | 1820/7150 [25:09<1:00:56,  1.46it/s]
                                                     
{'loss': '3.376', 'grad_norm': '0.239', 'learning_rate': '0.0006', 'epoch': '12.73'}
+
 25%|██▌       | 1820/7150 [25:09<1:00:56,  1.46it/s]
 25%|██▌       | 1821/7150 [25:10<1:02:14,  1.43it/s]
 25%|██▌       | 1822/7150 [25:10<1:01:14,  1.45it/s]
 25%|██▌       | 1823/7150 [25:11<1:00:36,  1.47it/s]
 26%|██▌       | 1824/7150 [25:12<1:00:21,  1.47it/s]
 26%|██▌       | 1825/7150 [25:12<59:59,  1.48it/s]  
 26%|██▌       | 1826/7150 [25:13<1:00:29,  1.47it/s]
 26%|██▌       | 1827/7150 [25:14<1:01:08,  1.45it/s]
 26%|██▌       | 1828/7150 [25:14<1:00:52,  1.46it/s]
 26%|██▌       | 1829/7150 [25:15<1:02:10,  1.43it/s]
 26%|██▌       | 1830/7150 [25:16<1:04:37,  1.37it/s]
                                                     
{'loss': '3.374', 'grad_norm': '0.2431', 'learning_rate': '0.0006', 'epoch': '12.8'}
+
 26%|██▌       | 1830/7150 [25:16<1:04:37,  1.37it/s]
 26%|██▌       | 1831/7150 [25:17<1:02:39,  1.41it/s]
 26%|██▌       | 1832/7150 [25:17<1:05:45,  1.35it/s]
 26%|██▌       | 1833/7150 [25:18<1:04:13,  1.38it/s]
 26%|██▌       | 1834/7150 [25:19<1:02:17,  1.42it/s]
 26%|██▌       | 1835/7150 [25:19<1:05:18,  1.36it/s]
 26%|██▌       | 1836/7150 [25:20<1:05:33,  1.35it/s]
 26%|██▌       | 1837/7150 [25:21<1:09:50,  1.27it/s]
 26%|██▌       | 1838/7150 [25:22<1:08:17,  1.30it/s]
 26%|██▌       | 1839/7150 [25:23<1:10:22,  1.26it/s]
 26%|██▌       | 1840/7150 [25:24<1:12:52,  1.21it/s]
                                                     
{'loss': '3.378', 'grad_norm': '0.2659', 'learning_rate': '0.0006', 'epoch': '12.87'}
+
 26%|██▌       | 1840/7150 [25:24<1:12:52,  1.21it/s]
 26%|██▌       | 1841/7150 [25:25<1:15:08,  1.18it/s]
 26%|██▌       | 1842/7150 [25:25<1:11:12,  1.24it/s]
 26%|██▌       | 1843/7150 [25:26<1:07:31,  1.31it/s]
 26%|██▌       | 1844/7150 [25:27<1:05:16,  1.35it/s]
 26%|██▌       | 1845/7150 [25:27<1:03:34,  1.39it/s]
 26%|██▌       | 1846/7150 [25:28<1:02:23,  1.42it/s]
 26%|██▌       | 1847/7150 [25:29<1:01:21,  1.44it/s]
 26%|██▌       | 1848/7150 [25:29<1:00:18,  1.47it/s]
 26%|██▌       | 1849/7150 [25:30<1:00:08,  1.47it/s]
 26%|██▌       | 1850/7150 [25:31<59:42,  1.48it/s]  
                                                   
{'loss': '3.376', 'grad_norm': '0.2792', 'learning_rate': '0.0006', 'epoch': '12.94'}
+
 26%|██▌       | 1850/7150 [25:31<59:42,  1.48it/s]
 26%|██▌       | 1851/7150 [25:31<59:06,  1.49it/s]
 26%|██▌       | 1852/7150 [25:32<59:15,  1.49it/s]
 26%|██▌       | 1853/7150 [25:33<59:09,  1.49it/s]
 26%|██▌       | 1854/7150 [25:33<58:58,  1.50it/s]
 26%|██▌       | 1855/7150 [25:34<59:57,  1.47it/s]
 26%|██▌       | 1856/7150 [25:35<59:15,  1.49it/s]
 26%|██▌       | 1857/7150 [25:35<59:30,  1.48it/s]
 26%|██▌       | 1858/7150 [25:36<59:30,  1.48it/s]
 26%|██▌       | 1859/7150 [25:36<55:00,  1.60it/s]
 26%|██▌       | 1860/7150 [25:37<57:58,  1.52it/s]
                                                   
{'loss': '3.36', 'grad_norm': '0.2594', 'learning_rate': '0.0006', 'epoch': '13.01'}
+
 26%|██▌       | 1860/7150 [25:37<57:58,  1.52it/s]
 26%|██▌       | 1861/7150 [25:38<58:35,  1.50it/s]
 26%|██▌       | 1862/7150 [25:39<58:51,  1.50it/s]
 26%|██▌       | 1863/7150 [25:39<58:29,  1.51it/s]
 26%|██▌       | 1864/7150 [25:40<58:46,  1.50it/s]
 26%|██▌       | 1865/7150 [25:41<59:02,  1.49it/s]
 26%|██▌       | 1866/7150 [25:41<59:10,  1.49it/s]
 26%|██▌       | 1867/7150 [25:42<58:43,  1.50it/s]
 26%|██▌       | 1868/7150 [25:43<58:54,  1.49it/s]
 26%|██▌       | 1869/7150 [25:43<58:54,  1.49it/s]
 26%|██▌       | 1870/7150 [25:44<59:12,  1.49it/s]
                                                   
{'loss': '3.279', 'grad_norm': '0.2983', 'learning_rate': '0.0006', 'epoch': '13.08'}
+
 26%|██▌       | 1870/7150 [25:44<59:12,  1.49it/s]
 26%|██▌       | 1871/7150 [25:45<59:04,  1.49it/s]
 26%|██▌       | 1872/7150 [25:45<58:55,  1.49it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.79it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.88it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.65it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.09it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 11.93it/s]
+
 33%|███▎      | 13/39 [00:01<00:02, 12.03it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.09it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.13it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 11.53it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 11.75it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 11.90it/s]
+
 64%|██████▍   | 25/39 [00:02<00:01, 12.00it/s]
+
 69%|██████▉   | 27/39 [00:02<00:01, 11.76it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 10.82it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 10.78it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 11.19it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 11.50it/s]
+
 95%|█████████▍| 37/39 [00:03<00:00, 11.73it/s]
+
100%|██████████| 39/39 [00:03<00:00, 11.05it/s]
                                                   
+
                                               
{'eval_loss': '3.76', 'eval_runtime': '3.454', 'eval_samples_per_second': '706.6', 'eval_steps_per_second': '11.29', 'epoch': '13.09'}
+
 26%|██▌       | 1872/7150 [25:49<58:55,  1.49it/s]
+
100%|██████████| 39/39 [00:03<00:00, 11.05it/s][eval] epoch 13.091 step 1872 loss 3.7603 (best 3.7530 @ ep 12.588, 1/20 since best)
+
+
                                               
 26%|██▌       | 1873/7150 [25:49<2:30:10,  1.71s/it]
 26%|██▌       | 1874/7150 [25:50<2:02:19,  1.39s/it]
 26%|██▌       | 1875/7150 [25:51<1:43:24,  1.18s/it]
 26%|██▌       | 1876/7150 [25:51<1:30:59,  1.04s/it]
 26%|██▋       | 1877/7150 [25:52<1:22:07,  1.07it/s]
 26%|██▋       | 1878/7150 [25:53<1:14:56,  1.17it/s]
 26%|██▋       | 1879/7150 [25:53<1:09:42,  1.26it/s]
 26%|██▋       | 1880/7150 [25:54<1:06:34,  1.32it/s]
                                                     
{'loss': '3.285', 'grad_norm': '0.3455', 'learning_rate': '0.0006', 'epoch': '13.15'}
+
 26%|██▋       | 1880/7150 [25:54<1:06:34,  1.32it/s]
 26%|██▋       | 1881/7150 [25:55<1:04:08,  1.37it/s]
 26%|██▋       | 1882/7150 [25:55<1:02:41,  1.40it/s]
 26%|██▋       | 1883/7150 [25:56<1:01:54,  1.42it/s]
 26%|██▋       | 1884/7150 [25:57<1:01:22,  1.43it/s]
 26%|██▋       | 1885/7150 [25:58<1:00:51,  1.44it/s]
 26%|██▋       | 1886/7150 [25:58<1:00:27,  1.45it/s]
 26%|██▋       | 1887/7150 [25:59<59:32,  1.47it/s]  
 26%|██▋       | 1888/7150 [26:00<59:21,  1.48it/s]
 26%|██▋       | 1889/7150 [26:00<59:00,  1.49it/s]
 26%|██▋       | 1890/7150 [26:01<58:29,  1.50it/s]
                                                   
{'loss': '3.295', 'grad_norm': '0.309', 'learning_rate': '0.0006', 'epoch': '13.22'}
+
 26%|██▋       | 1890/7150 [26:01<58:29,  1.50it/s]
 26%|██▋       | 1891/7150 [26:02<58:42,  1.49it/s]
 26%|██▋       | 1892/7150 [26:02<58:33,  1.50it/s]
 26%|██▋       | 1893/7150 [26:03<58:10,  1.51it/s]
 26%|██▋       | 1894/7150 [26:03<58:26,  1.50it/s]
 27%|██▋       | 1895/7150 [26:04<59:25,  1.47it/s]
 27%|██▋       | 1896/7150 [26:05<59:26,  1.47it/s]
 27%|██▋       | 1897/7150 [26:06<58:47,  1.49it/s]
 27%|██▋       | 1898/7150 [26:06<58:54,  1.49it/s]
 27%|██▋       | 1899/7150 [26:07<59:03,  1.48it/s]
 27%|██▋       | 1900/7150 [26:08<58:31,  1.50it/s]
                                                   
{'loss': '3.307', 'grad_norm': '0.2685', 'learning_rate': '0.0006', 'epoch': '13.29'}
+
 27%|██▋       | 1900/7150 [26:08<58:31,  1.50it/s]
 27%|██▋       | 1901/7150 [26:08<58:41,  1.49it/s]
 27%|██▋       | 1902/7150 [26:09<59:36,  1.47it/s]
 27%|██▋       | 1903/7150 [26:10<59:30,  1.47it/s]
 27%|██▋       | 1904/7150 [26:10<59:01,  1.48it/s]
 27%|██▋       | 1905/7150 [26:11<59:05,  1.48it/s]
 27%|██▋       | 1906/7150 [26:12<59:02,  1.48it/s]
 27%|██▋       | 1907/7150 [26:12<59:23,  1.47it/s]
 27%|██▋       | 1908/7150 [26:13<58:43,  1.49it/s]
 27%|██▋       | 1909/7150 [26:14<59:04,  1.48it/s]
 27%|██▋       | 1910/7150 [26:14<58:51,  1.48it/s]
                                                   
{'loss': '3.303', 'grad_norm': '0.2217', 'learning_rate': '0.0006', 'epoch': '13.36'}
+
 27%|██▋       | 1910/7150 [26:14<58:51,  1.48it/s]
 27%|██▋       | 1911/7150 [26:15<58:22,  1.50it/s]
 27%|██▋       | 1912/7150 [26:16<58:28,  1.49it/s]
 27%|██▋       | 1913/7150 [26:16<58:39,  1.49it/s]
 27%|██▋       | 1914/7150 [26:17<58:10,  1.50it/s]
 27%|██▋       | 1915/7150 [26:18<58:21,  1.50it/s]
 27%|██▋       | 1916/7150 [26:18<59:55,  1.46it/s]
 27%|██▋       | 1917/7150 [26:19<59:38,  1.46it/s]
 27%|██▋       | 1918/7150 [26:20<1:00:06,  1.45it/s]
 27%|██▋       | 1919/7150 [26:20<1:00:04,  1.45it/s]
 27%|██▋       | 1920/7150 [26:21<1:00:32,  1.44it/s]
                                                     
{'loss': '3.318', 'grad_norm': '0.2786', 'learning_rate': '0.0006', 'epoch': '13.43'}
+
 27%|██▋       | 1920/7150 [26:21<1:00:32,  1.44it/s]
 27%|██▋       | 1921/7150 [26:22<59:48,  1.46it/s]  
 27%|██▋       | 1922/7150 [26:22<59:06,  1.47it/s]
 27%|██▋       | 1923/7150 [26:23<59:02,  1.48it/s]
 27%|██▋       | 1924/7150 [26:24<59:05,  1.47it/s]
 27%|██▋       | 1925/7150 [26:25<58:49,  1.48it/s]
 27%|██▋       | 1926/7150 [26:25<58:30,  1.49it/s]
 27%|██▋       | 1927/7150 [26:26<58:58,  1.48it/s]
 27%|██▋       | 1928/7150 [26:27<58:56,  1.48it/s]
 27%|██▋       | 1929/7150 [26:27<58:27,  1.49it/s]
 27%|██▋       | 1930/7150 [26:28<1:00:06,  1.45it/s]
                                                     
{'loss': '3.327', 'grad_norm': '0.2477', 'learning_rate': '0.0006', 'epoch': '13.5'}
+
 27%|██▋       | 1930/7150 [26:28<1:00:06,  1.45it/s]
 27%|██▋       | 1931/7150 [26:29<59:31,  1.46it/s]  
 27%|██▋       | 1932/7150 [26:29<59:07,  1.47it/s]
 27%|██▋       | 1933/7150 [26:30<59:50,  1.45it/s]
 27%|██▋       | 1934/7150 [26:31<59:15,  1.47it/s]
 27%|██▋       | 1935/7150 [26:31<59:08,  1.47it/s]
 27%|██▋       | 1936/7150 [26:32<58:27,  1.49it/s]
 27%|██▋       | 1937/7150 [26:33<59:46,  1.45it/s]
 27%|██▋       | 1938/7150 [26:33<1:01:31,  1.41it/s]
 27%|██▋       | 1939/7150 [26:34<1:01:24,  1.41it/s]
 27%|██▋       | 1940/7150 [26:35<1:00:35,  1.43it/s]
                                                     
{'loss': '3.326', 'grad_norm': '0.2245', 'learning_rate': '0.0006', 'epoch': '13.57'}
+
 27%|██▋       | 1940/7150 [26:35<1:00:35,  1.43it/s]
 27%|██▋       | 1941/7150 [26:36<1:00:09,  1.44it/s]
 27%|██▋       | 1942/7150 [26:36<1:01:14,  1.42it/s]
 27%|██▋       | 1943/7150 [26:37<1:00:14,  1.44it/s]
 27%|██▋       | 1944/7150 [26:38<59:43,  1.45it/s]  [transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.79it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 13.45it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 12.91it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 12.66it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.50it/s]
+
 33%|███▎      | 13/39 [00:01<00:02, 11.96it/s]
+
 38%|███▊      | 15/39 [00:01<00:02, 11.89it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.00it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.07it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.13it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.17it/s]
+
 64%|██████▍   | 25/39 [00:02<00:01, 11.66it/s]
+
 69%|██████▉   | 27/39 [00:02<00:01, 11.23it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 11.54it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 11.75it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 11.90it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 11.99it/s]
+
 95%|█████████▍| 37/39 [00:03<00:00, 11.91it/s]
+
100%|██████████| 39/39 [00:03<00:00, 11.86it/s]
                                                   
+
                                               
{'eval_loss': '3.743', 'eval_runtime': '3.378', 'eval_samples_per_second': '722.6', 'eval_steps_per_second': '11.54', 'epoch': '13.6'}
+
 27%|██▋       | 1944/7150 [26:41<59:43,  1.45it/s]
+
100%|██████████| 39/39 [00:03<00:00, 11.86it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-1944
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-1944/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-1944/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.89it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.88it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-1944/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-1944/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-1944/tokenizer_config.json
+[eval] epoch 13.595 step 1944 loss 3.7428 (best 3.7428 @ ep 13.595, 0/20 since best)
+
+
                                               
 27%|██▋       | 1945/7150 [26:44<3:21:29,  2.32s/it]
 27%|██▋       | 1946/7150 [26:44<2:38:54,  1.83s/it]
 27%|██▋       | 1947/7150 [26:45<2:09:15,  1.49s/it]
 27%|██▋       | 1948/7150 [26:46<1:48:16,  1.25s/it]
 27%|██▋       | 1949/7150 [26:46<1:33:39,  1.08s/it]
 27%|██▋       | 1950/7150 [26:47<1:23:36,  1.04it/s]
                                                     
{'loss': '3.333', 'grad_norm': '0.2591', 'learning_rate': '0.0006', 'epoch': '13.64'}
+
 27%|██▋       | 1950/7150 [26:47<1:23:36,  1.04it/s]
 27%|██▋       | 1951/7150 [26:48<1:17:00,  1.13it/s]
 27%|██▋       | 1952/7150 [26:49<1:11:11,  1.22it/s]
 27%|██▋       | 1953/7150 [26:49<1:06:50,  1.30it/s]
 27%|██▋       | 1954/7150 [26:50<1:05:41,  1.32it/s]
 27%|██▋       | 1955/7150 [26:51<1:05:12,  1.33it/s]
 27%|██▋       | 1956/7150 [26:52<1:07:58,  1.27it/s]
 27%|██▋       | 1957/7150 [26:52<1:04:34,  1.34it/s]
 27%|██▋       | 1958/7150 [26:53<1:03:00,  1.37it/s]
 27%|██▋       | 1959/7150 [26:54<1:01:19,  1.41it/s]
 27%|██▋       | 1960/7150 [26:54<59:51,  1.44it/s]  
                                                   
{'loss': '3.333', 'grad_norm': '0.2343', 'learning_rate': '0.0006', 'epoch': '13.71'}
+
 27%|██▋       | 1960/7150 [26:54<59:51,  1.44it/s]
 27%|██▋       | 1961/7150 [26:55<1:00:11,  1.44it/s]
 27%|██▋       | 1962/7150 [26:56<59:29,  1.45it/s]  
 27%|██▋       | 1963/7150 [26:56<1:04:12,  1.35it/s]
 27%|██▋       | 1964/7150 [26:57<1:03:56,  1.35it/s]
 27%|██▋       | 1965/7150 [26:58<1:02:21,  1.39it/s]
 27%|██▋       | 1966/7150 [26:59<1:01:42,  1.40it/s]
 28%|██▊       | 1967/7150 [26:59<1:01:22,  1.41it/s]
 28%|██▊       | 1968/7150 [27:00<1:01:19,  1.41it/s]
 28%|██▊       | 1969/7150 [27:01<1:00:18,  1.43it/s]
 28%|██▊       | 1970/7150 [27:01<59:57,  1.44it/s]  
                                                   
{'loss': '3.335', 'grad_norm': '0.2332', 'learning_rate': '0.0006', 'epoch': '13.78'}
+
 28%|██▊       | 1970/7150 [27:01<59:57,  1.44it/s]
 28%|██▊       | 1971/7150 [27:02<59:13,  1.46it/s]
 28%|██▊       | 1972/7150 [27:03<1:00:17,  1.43it/s]
 28%|██▊       | 1973/7150 [27:03<59:52,  1.44it/s]  
 28%|██▊       | 1974/7150 [27:04<58:48,  1.47it/s]
 28%|██▊       | 1975/7150 [27:05<1:02:25,  1.38it/s]
 28%|██▊       | 1976/7150 [27:06<1:00:36,  1.42it/s]
 28%|██▊       | 1977/7150 [27:06<1:00:31,  1.42it/s]
 28%|██▊       | 1978/7150 [27:07<59:35,  1.45it/s]  
 28%|██▊       | 1979/7150 [27:08<58:36,  1.47it/s]
 28%|██▊       | 1980/7150 [27:08<59:49,  1.44it/s]
                                                   
{'loss': '3.335', 'grad_norm': '0.2602', 'learning_rate': '0.0006', 'epoch': '13.85'}
+
 28%|██▊       | 1980/7150 [27:08<59:49,  1.44it/s]
 28%|██▊       | 1981/7150 [27:09<58:48,  1.46it/s]
 28%|██▊       | 1982/7150 [27:10<58:54,  1.46it/s]
 28%|██▊       | 1983/7150 [27:10<59:11,  1.45it/s]
 28%|██▊       | 1984/7150 [27:11<58:20,  1.48it/s]
 28%|██▊       | 1985/7150 [27:12<1:02:43,  1.37it/s]
 28%|██▊       | 1986/7150 [27:12<1:01:15,  1.40it/s]
 28%|██▊       | 1987/7150 [27:13<1:02:47,  1.37it/s]
 28%|██▊       | 1988/7150 [27:14<1:00:54,  1.41it/s]
 28%|██▊       | 1989/7150 [27:15<1:00:45,  1.42it/s]
 28%|██▊       | 1990/7150 [27:15<1:00:44,  1.42it/s]
                                                     
{'loss': '3.331', 'grad_norm': '0.2432', 'learning_rate': '0.0006', 'epoch': '13.92'}
+
 28%|██▊       | 1990/7150 [27:15<1:00:44,  1.42it/s]
 28%|██▊       | 1991/7150 [27:16<59:45,  1.44it/s]  
 28%|██▊       | 1992/7150 [27:17<58:55,  1.46it/s]
 28%|██▊       | 1993/7150 [27:17<58:13,  1.48it/s]
 28%|██▊       | 1994/7150 [27:18<1:00:28,  1.42it/s]
 28%|██▊       | 1995/7150 [27:19<59:18,  1.45it/s]  
 28%|██▊       | 1996/7150 [27:19<58:40,  1.46it/s]
 28%|██▊       | 1997/7150 [27:20<1:00:22,  1.42it/s]
 28%|██▊       | 1998/7150 [27:21<1:02:14,  1.38it/s]
 28%|██▊       | 1999/7150 [27:22<1:00:25,  1.42it/s]
 28%|██▊       | 2000/7150 [27:22<1:01:13,  1.40it/s]
                                                     
{'loss': '3.34', 'grad_norm': '0.2369', 'learning_rate': '0.0006', 'epoch': '13.99'}
+
 28%|██▊       | 2000/7150 [27:22<1:01:13,  1.40it/s]
 28%|██▊       | 2001/7150 [27:23<1:00:24,  1.42it/s]
 28%|██▊       | 2002/7150 [27:23<54:58,  1.56it/s]  
 28%|██▊       | 2003/7150 [27:24<1:01:12,  1.40it/s]
 28%|██▊       | 2004/7150 [27:25<1:04:30,  1.33it/s]
 28%|██▊       | 2005/7150 [27:26<1:03:25,  1.35it/s]
 28%|██▊       | 2006/7150 [27:27<1:02:01,  1.38it/s]
 28%|██▊       | 2007/7150 [27:27<1:00:53,  1.41it/s]
 28%|██▊       | 2008/7150 [27:28<1:00:07,  1.43it/s]
 28%|██▊       | 2009/7150 [27:29<59:37,  1.44it/s]  
 28%|██▊       | 2010/7150 [27:29<59:02,  1.45it/s]
                                                   
{'loss': '3.255', 'grad_norm': '0.3252', 'learning_rate': '0.0006', 'epoch': '14.06'}
+
 28%|██▊       | 2010/7150 [27:29<59:02,  1.45it/s]
 28%|██▊       | 2011/7150 [27:30<58:10,  1.47it/s]
 28%|██▊       | 2012/7150 [27:31<59:10,  1.45it/s]
 28%|██▊       | 2013/7150 [27:31<59:07,  1.45it/s]
 28%|██▊       | 2014/7150 [27:32<58:53,  1.45it/s]
 28%|██▊       | 2015/7150 [27:33<58:44,  1.46it/s]
 28%|██▊       | 2016/7150 [27:33<58:42,  1.46it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.09it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.86it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.50it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.13it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 11.81it/s]
+
 33%|███▎      | 13/39 [00:01<00:02, 12.06it/s]
+
 38%|███▊      | 15/39 [00:01<00:02, 12.00it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.18it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 11.40it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 11.65it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 11.84it/s]
+
 64%|██████▍   | 25/39 [00:02<00:01, 11.96it/s]
+
 69%|██████▉   | 27/39 [00:02<00:01, 11.58it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 11.51it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 11.73it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 11.87it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 11.97it/s]
+
 95%|█████████▍| 37/39 [00:03<00:00, 12.05it/s]
+
100%|██████████| 39/39 [00:03<00:00, 11.30it/s]
                                                   
+
                                               
{'eval_loss': '3.757', 'eval_runtime': '3.408', 'eval_samples_per_second': '716.3', 'eval_steps_per_second': '11.44', 'epoch': '14.1'}
+
 28%|██▊       | 2016/7150 [27:37<58:42,  1.46it/s]
+
100%|██████████| 39/39 [00:03<00:00, 11.30it/s][eval] epoch 14.098 step 2016 loss 3.7570 (best 3.7428 @ ep 13.595, 1/20 since best)
+
+
                                               
 28%|██▊       | 2017/7150 [27:38<2:25:21,  1.70s/it]
 28%|██▊       | 2018/7150 [27:38<1:59:12,  1.39s/it]
 28%|██▊       | 2019/7150 [27:39<1:41:01,  1.18s/it]
 28%|██▊       | 2020/7150 [27:40<1:29:07,  1.04s/it]
                                                     
{'loss': '3.251', 'grad_norm': '0.3999', 'learning_rate': '0.0006', 'epoch': '14.13'}
+
 28%|██▊       | 2020/7150 [27:40<1:29:07,  1.04s/it]
 28%|██▊       | 2021/7150 [27:40<1:19:24,  1.08it/s]
 28%|██▊       | 2022/7150 [27:41<1:12:49,  1.17it/s]
 28%|██▊       | 2023/7150 [27:42<1:08:41,  1.24it/s]
 28%|██▊       | 2024/7150 [27:42<1:05:40,  1.30it/s]
 28%|██▊       | 2025/7150 [27:43<1:03:28,  1.35it/s]
 28%|██▊       | 2026/7150 [27:44<1:01:51,  1.38it/s]
 28%|██▊       | 2027/7150 [27:44<1:00:48,  1.40it/s]
 28%|██▊       | 2028/7150 [27:45<59:20,  1.44it/s]  
 28%|██▊       | 2029/7150 [27:46<59:02,  1.45it/s]
 28%|██▊       | 2030/7150 [27:46<58:49,  1.45it/s]
                                                   
{'loss': '3.254', 'grad_norm': '0.2794', 'learning_rate': '0.0006', 'epoch': '14.2'}
+
 28%|██▊       | 2030/7150 [27:46<58:49,  1.45it/s]
 28%|██▊       | 2031/7150 [27:47<57:56,  1.47it/s]
 28%|██▊       | 2032/7150 [27:48<58:33,  1.46it/s]
 28%|██▊       | 2033/7150 [27:48<58:35,  1.46it/s]
 28%|██▊       | 2034/7150 [27:49<58:34,  1.46it/s]
 28%|██▊       | 2035/7150 [27:50<58:26,  1.46it/s]
 28%|██▊       | 2036/7150 [27:50<57:40,  1.48it/s]
 28%|██▊       | 2037/7150 [27:51<57:55,  1.47it/s]
 29%|██▊       | 2038/7150 [27:52<58:03,  1.47it/s]
 29%|██▊       | 2039/7150 [27:52<57:19,  1.49it/s]
 29%|██▊       | 2040/7150 [27:53<57:39,  1.48it/s]
                                                   
{'loss': '3.256', 'grad_norm': '0.2673', 'learning_rate': '0.0006', 'epoch': '14.27'}
+
 29%|██▊       | 2040/7150 [27:53<57:39,  1.48it/s]
 29%|██▊       | 2041/7150 [27:54<57:50,  1.47it/s]
 29%|██▊       | 2042/7150 [27:55<58:47,  1.45it/s]
 29%|██▊       | 2043/7150 [27:55<58:24,  1.46it/s]
 29%|██▊       | 2044/7150 [27:56<57:34,  1.48it/s]
 29%|██▊       | 2045/7150 [27:57<58:39,  1.45it/s]
 29%|██▊       | 2046/7150 [27:57<59:14,  1.44it/s]
 29%|██▊       | 2047/7150 [27:58<58:09,  1.46it/s]
 29%|██▊       | 2048/7150 [27:59<58:58,  1.44it/s]
 29%|██▊       | 2049/7150 [27:59<58:44,  1.45it/s]
 29%|██▊       | 2050/7150 [28:00<59:35,  1.43it/s]
                                                   
{'loss': '3.269', 'grad_norm': '0.3608', 'learning_rate': '0.0006', 'epoch': '14.34'}
+
 29%|██▊       | 2050/7150 [28:00<59:35,  1.43it/s]
 29%|██▊       | 2051/7150 [28:01<1:00:05,  1.41it/s]
 29%|██▊       | 2052/7150 [28:02<1:02:03,  1.37it/s]
 29%|██▊       | 2053/7150 [28:02<1:00:07,  1.41it/s]
 29%|██▊       | 2054/7150 [28:03<1:01:19,  1.39it/s]
 29%|██▊       | 2055/7150 [28:04<59:33,  1.43it/s]  
 29%|██▉       | 2056/7150 [28:04<59:06,  1.44it/s]
 29%|██▉       | 2057/7150 [28:05<58:46,  1.44it/s]
 29%|██▉       | 2058/7150 [28:06<1:00:25,  1.40it/s]
 29%|██▉       | 2059/7150 [28:06<59:32,  1.43it/s]  
 29%|██▉       | 2060/7150 [28:07<59:05,  1.44it/s]
                                                   
{'loss': '3.284', 'grad_norm': '0.2672', 'learning_rate': '0.0006', 'epoch': '14.41'}
+
 29%|██▉       | 2060/7150 [28:07<59:05,  1.44it/s]
 29%|██▉       | 2061/7150 [28:08<58:31,  1.45it/s]
 29%|██▉       | 2062/7150 [28:09<59:40,  1.42it/s]
 29%|██▉       | 2063/7150 [28:09<1:00:10,  1.41it/s]
 29%|██▉       | 2064/7150 [28:10<59:29,  1.42it/s]  
 29%|██▉       | 2065/7150 [28:11<59:01,  1.44it/s]
 29%|██▉       | 2066/7150 [28:11<58:53,  1.44it/s]
 29%|██▉       | 2067/7150 [28:12<58:49,  1.44it/s]
 29%|██▉       | 2068/7150 [28:13<57:51,  1.46it/s]
 29%|██▉       | 2069/7150 [28:13<58:31,  1.45it/s]
 29%|██▉       | 2070/7150 [28:14<58:15,  1.45it/s]
                                                   
{'loss': '3.284', 'grad_norm': '0.2273', 'learning_rate': '0.0006', 'epoch': '14.48'}
+
 29%|██▉       | 2070/7150 [28:14<58:15,  1.45it/s]
 29%|██▉       | 2071/7150 [28:15<58:09,  1.46it/s]
 29%|██▉       | 2072/7150 [28:16<58:59,  1.43it/s]
 29%|██▉       | 2073/7150 [28:16<57:51,  1.46it/s]
 29%|██▉       | 2074/7150 [28:17<57:46,  1.46it/s]
 29%|██▉       | 2075/7150 [28:18<58:06,  1.46it/s]
 29%|██▉       | 2076/7150 [28:18<58:05,  1.46it/s]
 29%|██▉       | 2077/7150 [28:19<58:52,  1.44it/s]
 29%|██▉       | 2078/7150 [28:20<58:22,  1.45it/s]
 29%|██▉       | 2079/7150 [28:20<57:29,  1.47it/s]
 29%|██▉       | 2080/7150 [28:21<1:00:19,  1.40it/s]
                                                     
{'loss': '3.279', 'grad_norm': '0.2284', 'learning_rate': '0.0006', 'epoch': '14.55'}
+
 29%|██▉       | 2080/7150 [28:21<1:00:19,  1.40it/s]
 29%|██▉       | 2081/7150 [28:22<58:47,  1.44it/s]  
 29%|██▉       | 2082/7150 [28:22<58:23,  1.45it/s]
 29%|██▉       | 2083/7150 [28:23<59:09,  1.43it/s]
 29%|██▉       | 2084/7150 [28:24<58:55,  1.43it/s]
 29%|██▉       | 2085/7150 [28:24<58:19,  1.45it/s]
 29%|██▉       | 2086/7150 [28:25<57:24,  1.47it/s]
 29%|██▉       | 2087/7150 [28:26<57:24,  1.47it/s]
 29%|██▉       | 2088/7150 [28:27<58:23,  1.45it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:02, 16.55it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.10it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.22it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 12.84it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.61it/s]
+
 33%|███▎      | 13/39 [00:01<00:02, 11.77it/s]
+
 38%|███▊      | 15/39 [00:01<00:02, 11.79it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 11.94it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.03it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.08it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.14it/s]
+
 64%|██████▍   | 25/39 [00:02<00:01, 11.46it/s]
+
 69%|██████▉   | 27/39 [00:02<00:01, 10.36it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 10.87it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 11.25it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 11.54it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 11.74it/s]
+
 95%|█████████▍| 37/39 [00:03<00:00, 11.88it/s]
+
100%|██████████| 39/39 [00:03<00:00, 10.81it/s]
                                                   
+
                                               
{'eval_loss': '3.738', 'eval_runtime': '3.471', 'eval_samples_per_second': '703.3', 'eval_steps_per_second': '11.24', 'epoch': '14.6'}
+
 29%|██▉       | 2088/7150 [28:30<58:23,  1.45it/s]
+
100%|██████████| 39/39 [00:03<00:00, 10.81it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-2088
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-2088/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-2088/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.85it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.85it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-2088/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-2088/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-2088/tokenizer_config.json
+[eval] epoch 14.602 step 2088 loss 3.7383 (best 3.7383 @ ep 14.602, 0/20 since best)
+
+
                                               
 29%|██▉       | 2089/7150 [28:33<3:19:07,  2.36s/it]
 29%|██▉       | 2090/7150 [28:33<2:35:55,  1.85s/it]
                                                     
{'loss': '3.291', 'grad_norm': '0.2252', 'learning_rate': '0.0006', 'epoch': '14.62'}
+
 29%|██▉       | 2090/7150 [28:33<2:35:55,  1.85s/it]
 29%|██▉       | 2091/7150 [28:34<2:07:26,  1.51s/it]
 29%|██▉       | 2092/7150 [28:35<1:46:30,  1.26s/it]
 29%|██▉       | 2093/7150 [28:36<1:31:41,  1.09s/it]
 29%|██▉       | 2094/7150 [28:36<1:21:21,  1.04it/s]
 29%|██▉       | 2095/7150 [28:37<1:15:08,  1.12it/s]
 29%|██▉       | 2096/7150 [28:38<1:09:07,  1.22it/s]
 29%|██▉       | 2097/7150 [28:38<1:11:06,  1.18it/s]
 29%|██▉       | 2098/7150 [28:39<1:06:18,  1.27it/s]
 29%|██▉       | 2099/7150 [28:40<1:07:12,  1.25it/s]
 29%|██▉       | 2100/7150 [28:41<1:08:02,  1.24it/s]
                                                     
{'loss': '3.289', 'grad_norm': '0.2415', 'learning_rate': '0.0006', 'epoch': '14.69'}
+
 29%|██▉       | 2100/7150 [28:41<1:08:02,  1.24it/s]
 29%|██▉       | 2101/7150 [28:42<1:10:05,  1.20it/s]
 29%|██▉       | 2102/7150 [28:43<1:11:39,  1.17it/s]
 29%|██▉       | 2103/7150 [28:43<1:12:48,  1.16it/s]
 29%|██▉       | 2104/7150 [28:44<1:13:31,  1.14it/s]
 29%|██▉       | 2105/7150 [28:45<1:11:00,  1.18it/s]
 29%|██▉       | 2106/7150 [28:46<1:06:50,  1.26it/s]
 29%|██▉       | 2107/7150 [28:47<1:09:24,  1.21it/s]
 29%|██▉       | 2108/7150 [28:48<1:11:15,  1.18it/s]
 29%|██▉       | 2109/7150 [28:49<1:12:31,  1.16it/s]
 30%|██▉       | 2110/7150 [28:49<1:09:26,  1.21it/s]
                                                     
{'loss': '3.299', 'grad_norm': '0.2247', 'learning_rate': '0.0006', 'epoch': '14.76'}
+
 30%|██▉       | 2110/7150 [28:49<1:09:26,  1.21it/s]
 30%|██▉       | 2111/7150 [28:50<1:09:29,  1.21it/s]
 30%|██▉       | 2112/7150 [28:51<1:11:00,  1.18it/s]
 30%|██▉       | 2113/7150 [28:52<1:11:54,  1.17it/s]
 30%|██▉       | 2114/7150 [28:53<1:12:32,  1.16it/s]
 30%|██▉       | 2115/7150 [28:54<1:12:57,  1.15it/s]
 30%|██▉       | 2116/7150 [28:55<1:13:16,  1.15it/s]
 30%|██▉       | 2117/7150 [28:55<1:13:32,  1.14it/s]
 30%|██▉       | 2118/7150 [28:56<1:13:37,  1.14it/s]
 30%|██▉       | 2119/7150 [28:57<1:13:40,  1.14it/s]
 30%|██▉       | 2120/7150 [28:58<1:13:42,  1.14it/s]
                                                     
{'loss': '3.306', 'grad_norm': '0.2814', 'learning_rate': '0.0006', 'epoch': '14.83'}
+
 30%|██▉       | 2120/7150 [28:58<1:13:42,  1.14it/s]
 30%|██▉       | 2121/7150 [28:59<1:12:18,  1.16it/s]
 30%|██▉       | 2122/7150 [29:00<1:07:04,  1.25it/s]
 30%|██▉       | 2123/7150 [29:00<1:04:55,  1.29it/s]
 30%|██▉       | 2124/7150 [29:01<1:02:37,  1.34it/s]
 30%|██▉       | 2125/7150 [29:02<1:01:02,  1.37it/s]
 30%|██▉       | 2126/7150 [29:02<59:56,  1.40it/s]  
 30%|██▉       | 2127/7150 [29:03<58:24,  1.43it/s]
 30%|██▉       | 2128/7150 [29:04<57:57,  1.44it/s]
 30%|██▉       | 2129/7150 [29:04<57:42,  1.45it/s]
 30%|██▉       | 2130/7150 [29:05<57:27,  1.46it/s]
                                                   
{'loss': '3.305', 'grad_norm': '0.2264', 'learning_rate': '0.0006', 'epoch': '14.9'}
+
 30%|██▉       | 2130/7150 [29:05<57:27,  1.46it/s]
 30%|██▉       | 2131/7150 [29:06<57:21,  1.46it/s]
 30%|██▉       | 2132/7150 [29:06<56:35,  1.48it/s]
 30%|██▉       | 2133/7150 [29:07<58:19,  1.43it/s]
 30%|██▉       | 2134/7150 [29:08<57:30,  1.45it/s]
 30%|██▉       | 2135/7150 [29:08<57:44,  1.45it/s]
 30%|██▉       | 2136/7150 [29:09<1:00:04,  1.39it/s]
 30%|██▉       | 2137/7150 [29:10<1:01:38,  1.36it/s]
 30%|██▉       | 2138/7150 [29:11<1:00:15,  1.39it/s]
 30%|██▉       | 2139/7150 [29:11<59:05,  1.41it/s]  
 30%|██▉       | 2140/7150 [29:12<58:30,  1.43it/s]
                                                   
{'loss': '3.29', 'grad_norm': '0.2472', 'learning_rate': '0.0006', 'epoch': '14.97'}
+
 30%|██▉       | 2140/7150 [29:12<58:30,  1.43it/s]
 30%|██▉       | 2141/7150 [29:13<58:56,  1.42it/s]
 30%|██▉       | 2142/7150 [29:14<1:00:00,  1.39it/s]
 30%|██▉       | 2143/7150 [29:14<1:02:19,  1.34it/s]
 30%|██▉       | 2144/7150 [29:15<1:03:06,  1.32it/s]
 30%|███       | 2145/7150 [29:16<1:01:14,  1.36it/s]
 30%|███       | 2146/7150 [29:17<1:04:02,  1.30it/s]
 30%|███       | 2147/7150 [29:17<1:02:46,  1.33it/s]
 30%|███       | 2148/7150 [29:18<1:01:43,  1.35it/s]
 30%|███       | 2149/7150 [29:19<1:02:01,  1.34it/s]
 30%|███       | 2150/7150 [29:20<1:06:09,  1.26it/s]
                                                     
{'loss': '3.239', 'grad_norm': '0.3908', 'learning_rate': '0.0006', 'epoch': '15.04'}
+
 30%|███       | 2150/7150 [29:20<1:06:09,  1.26it/s]
 30%|███       | 2151/7150 [29:21<1:09:07,  1.21it/s]
 30%|███       | 2152/7150 [29:21<1:04:42,  1.29it/s]
 30%|███       | 2153/7150 [29:22<1:05:37,  1.27it/s]
 30%|███       | 2154/7150 [29:23<1:08:37,  1.21it/s]
 30%|███       | 2155/7150 [29:24<1:10:44,  1.18it/s]
 30%|███       | 2156/7150 [29:25<1:12:18,  1.15it/s]
 30%|███       | 2157/7150 [29:26<1:13:19,  1.13it/s]
 30%|███       | 2158/7150 [29:27<1:11:33,  1.16it/s]
 30%|███       | 2159/7150 [29:27<1:06:30,  1.25it/s]
 30%|███       | 2160/7150 [29:28<1:09:12,  1.20it/s]
                                                     
{'loss': '3.207', 'grad_norm': '0.4219', 'learning_rate': '0.0006', 'epoch': '15.11'}
+
 30%|███       | 2160/7150 [29:28<1:09:12,  1.20it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:01, 18.74it/s]
+
 10%|█         | 4/39 [00:00<00:02, 11.69it/s]
+
 15%|█▌        | 6/39 [00:00<00:03, 10.40it/s]
+
 21%|██        | 8/39 [00:00<00:03,  9.77it/s]
+
 26%|██▌       | 10/39 [00:00<00:03,  9.43it/s]
+
 28%|██▊       | 11/39 [00:01<00:02,  9.37it/s]
+
 31%|███       | 12/39 [00:01<00:02,  9.32it/s]
+
 33%|███▎      | 13/39 [00:01<00:02,  9.28it/s]
+
 36%|███▌      | 14/39 [00:01<00:02,  9.25it/s]
+
 38%|███▊      | 15/39 [00:01<00:02,  9.22it/s]
+
 41%|████      | 16/39 [00:01<00:02,  9.24it/s]
+
 44%|████▎     | 17/39 [00:01<00:02,  9.21it/s]
+
 46%|████▌     | 18/39 [00:01<00:02,  9.20it/s]
+
 49%|████▊     | 19/39 [00:01<00:02,  8.99it/s]
+
 51%|█████▏    | 20/39 [00:02<00:02,  8.95it/s]
+
 54%|█████▍    | 21/39 [00:02<00:01,  9.02it/s]
+
 56%|█████▋    | 22/39 [00:02<00:01,  9.07it/s]
+
 59%|█████▉    | 23/39 [00:02<00:01,  9.10it/s]
+
 62%|██████▏   | 24/39 [00:02<00:01,  9.13it/s]
+
 64%|██████▍   | 25/39 [00:02<00:01,  9.14it/s]
+
 67%|██████▋   | 26/39 [00:02<00:01,  9.18it/s]
+
 69%|██████▉   | 27/39 [00:02<00:01,  9.19it/s]
+
 72%|███████▏  | 28/39 [00:02<00:01,  9.18it/s]
+
 74%|███████▍  | 29/39 [00:03<00:01,  9.15it/s]
+
 77%|███████▋  | 30/39 [00:03<00:00,  9.07it/s]
+
 79%|███████▉  | 31/39 [00:03<00:00,  9.03it/s]
+
 82%|████████▏ | 32/39 [00:03<00:00,  9.01it/s]
+
 85%|████████▍ | 33/39 [00:03<00:00,  9.06it/s]
+
 87%|████████▋ | 34/39 [00:03<00:00,  9.09it/s]
+
 90%|████████▉ | 35/39 [00:03<00:00,  9.12it/s]
+
 92%|█████████▏| 36/39 [00:03<00:00,  9.14it/s]
+
 95%|█████████▍| 37/39 [00:03<00:00,  9.16it/s]
+
 97%|█████████▋| 38/39 [00:04<00:00,  9.17it/s]
+
100%|██████████| 39/39 [00:04<00:00,  9.19it/s]
                                                     
+
                                               
{'eval_loss': '3.769', 'eval_runtime': '4.352', 'eval_samples_per_second': '560.9', 'eval_steps_per_second': '8.962', 'epoch': '15.11'}
+
 30%|███       | 2160/7150 [29:33<1:09:12,  1.20it/s]
+
100%|██████████| 39/39 [00:04<00:00,  9.19it/s][eval] epoch 15.105 step 2160 loss 3.7691 (best 3.7383 @ ep 14.602, 1/20 since best)
+
+
                                               
 30%|███       | 2161/7150 [29:33<2:57:50,  2.14s/it]
 30%|███       | 2162/7150 [29:34<2:20:48,  1.69s/it]
 30%|███       | 2163/7150 [29:35<1:54:53,  1.38s/it]
 30%|███       | 2164/7150 [29:35<1:36:50,  1.17s/it]
 30%|███       | 2165/7150 [29:36<1:24:07,  1.01s/it]
 30%|███       | 2166/7150 [29:37<1:15:09,  1.11it/s]
 30%|███       | 2167/7150 [29:37<1:08:55,  1.20it/s]
 30%|███       | 2168/7150 [29:38<1:04:34,  1.29it/s]
 30%|███       | 2169/7150 [29:39<1:01:30,  1.35it/s]
 30%|███       | 2170/7150 [29:39<59:20,  1.40it/s]  
                                                   
{'loss': '3.231', 'grad_norm': '0.3094', 'learning_rate': '0.0006', 'epoch': '15.18'}
+
 30%|███       | 2170/7150 [29:39<59:20,  1.40it/s]
 30%|███       | 2171/7150 [29:40<57:52,  1.43it/s]
 30%|███       | 2172/7150 [29:41<56:48,  1.46it/s]
 30%|███       | 2173/7150 [29:41<56:04,  1.48it/s]
 30%|███       | 2174/7150 [29:42<55:30,  1.49it/s]
 30%|███       | 2175/7150 [29:43<55:08,  1.50it/s]
 30%|███       | 2176/7150 [29:43<54:52,  1.51it/s]
 30%|███       | 2177/7150 [29:44<54:42,  1.51it/s]
 30%|███       | 2178/7150 [29:44<54:32,  1.52it/s]
 30%|███       | 2179/7150 [29:45<54:28,  1.52it/s]
 30%|███       | 2180/7150 [29:46<54:27,  1.52it/s]
                                                   
{'loss': '3.235', 'grad_norm': '0.2463', 'learning_rate': '0.0006', 'epoch': '15.25'}
+
 30%|███       | 2180/7150 [29:46<54:27,  1.52it/s]
 31%|███       | 2181/7150 [29:46<54:29,  1.52it/s]
 31%|███       | 2182/7150 [29:47<54:27,  1.52it/s]
 31%|███       | 2183/7150 [29:48<54:23,  1.52it/s]
 31%|███       | 2184/7150 [29:48<54:18,  1.52it/s]
 31%|███       | 2185/7150 [29:49<54:14,  1.53it/s]
 31%|███       | 2186/7150 [29:50<54:12,  1.53it/s]
 31%|███       | 2187/7150 [29:50<54:11,  1.53it/s]
 31%|███       | 2188/7150 [29:51<54:10,  1.53it/s]
 31%|███       | 2189/7150 [29:52<54:08,  1.53it/s]
 31%|███       | 2190/7150 [29:52<54:08,  1.53it/s]
                                                   
{'loss': '3.234', 'grad_norm': '0.2264', 'learning_rate': '0.0006', 'epoch': '15.32'}
+
 31%|███       | 2190/7150 [29:52<54:08,  1.53it/s]
 31%|███       | 2191/7150 [29:53<54:09,  1.53it/s]
 31%|███       | 2192/7150 [29:54<54:09,  1.53it/s]
 31%|███       | 2193/7150 [29:54<54:12,  1.52it/s]
 31%|███       | 2194/7150 [29:55<54:09,  1.53it/s]
 31%|███       | 2195/7150 [29:56<54:08,  1.53it/s]
 31%|███       | 2196/7150 [29:56<54:07,  1.53it/s]
 31%|███       | 2197/7150 [29:57<54:07,  1.53it/s]
 31%|███       | 2198/7150 [29:58<54:04,  1.53it/s]
 31%|███       | 2199/7150 [29:58<54:03,  1.53it/s]
 31%|███       | 2200/7150 [29:59<54:01,  1.53it/s]
                                                   
{'loss': '3.231', 'grad_norm': '0.2402', 'learning_rate': '0.0006', 'epoch': '15.39'}
+
 31%|███       | 2200/7150 [29:59<54:01,  1.53it/s]
 31%|███       | 2201/7150 [30:00<54:03,  1.53it/s]
 31%|███       | 2202/7150 [30:00<54:01,  1.53it/s]
 31%|███       | 2203/7150 [30:01<54:00,  1.53it/s]
 31%|███       | 2204/7150 [30:02<53:59,  1.53it/s]
 31%|███       | 2205/7150 [30:02<53:59,  1.53it/s]
 31%|███       | 2206/7150 [30:03<53:58,  1.53it/s]
 31%|███       | 2207/7150 [30:03<53:56,  1.53it/s]
 31%|███       | 2208/7150 [30:04<53:55,  1.53it/s]
 31%|███       | 2209/7150 [30:05<53:54,  1.53it/s]
 31%|███       | 2210/7150 [30:05<53:55,  1.53it/s]
                                                   
{'loss': '3.241', 'grad_norm': '0.2533', 'learning_rate': '0.0006', 'epoch': '15.46'}
+
 31%|███       | 2210/7150 [30:05<53:55,  1.53it/s]
 31%|███       | 2211/7150 [30:06<53:55,  1.53it/s]
 31%|███       | 2212/7150 [30:07<53:52,  1.53it/s]
 31%|███       | 2213/7150 [30:07<53:52,  1.53it/s]
 31%|███       | 2214/7150 [30:08<53:53,  1.53it/s]
 31%|███       | 2215/7150 [30:09<53:53,  1.53it/s]
 31%|███       | 2216/7150 [30:09<53:50,  1.53it/s]
 31%|███       | 2217/7150 [30:10<53:50,  1.53it/s]
 31%|███       | 2218/7150 [30:11<53:51,  1.53it/s]
 31%|███       | 2219/7150 [30:11<53:50,  1.53it/s]
 31%|███       | 2220/7150 [30:12<53:47,  1.53it/s]
                                                   
{'loss': '3.242', 'grad_norm': '0.2501', 'learning_rate': '0.0006', 'epoch': '15.53'}
+
 31%|███       | 2220/7150 [30:12<53:47,  1.53it/s]
 31%|███       | 2221/7150 [30:13<53:47,  1.53it/s]
 31%|███       | 2222/7150 [30:13<53:48,  1.53it/s]
 31%|███       | 2223/7150 [30:14<53:48,  1.53it/s]
 31%|███       | 2224/7150 [30:15<53:45,  1.53it/s]
 31%|███       | 2225/7150 [30:15<53:44,  1.53it/s]
 31%|███       | 2226/7150 [30:16<53:43,  1.53it/s]
 31%|███       | 2227/7150 [30:17<53:45,  1.53it/s]
 31%|███       | 2228/7150 [30:17<53:42,  1.53it/s]
 31%|███       | 2229/7150 [30:18<53:42,  1.53it/s]
 31%|███       | 2230/7150 [30:19<53:41,  1.53it/s]
                                                   
{'loss': '3.263', 'grad_norm': '0.2822', 'learning_rate': '0.0006', 'epoch': '15.6'}
+
 31%|███       | 2230/7150 [30:19<53:41,  1.53it/s]
 31%|███       | 2231/7150 [30:19<53:43,  1.53it/s]
 31%|███       | 2232/7150 [30:20<53:39,  1.53it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.83it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.86it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.65it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.09it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.78it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.58it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.47it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.37it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.33it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.31it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.30it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.28it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.27it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.25it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.25it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.24it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.24it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.23it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s]
                                                   
+
                                               
{'eval_loss': '3.738', 'eval_runtime': '3.251', 'eval_samples_per_second': '750.8', 'eval_steps_per_second': '12', 'epoch': '15.61'}
+
 31%|███       | 2232/7150 [30:23<53:39,  1.53it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-2232
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-2232/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-2232/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.66it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.66it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-2232/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-2232/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_150M_C0_baseline/checkpoint-2232/tokenizer_config.json
+[eval] epoch 15.609 step 2232 loss 3.7377 (best 3.7377 @ ep 15.609, 0/20 since best)
+
+
                                               
 31%|███       | 2233/7150 [30:26<3:06:32,  2.28s/it]
 31%|███       | 2234/7150 [30:27<2:26:37,  1.79s/it]
 31%|███▏      | 2235/7150 [30:27<1:58:40,  1.45s/it]
 31%|███▏      | 2236/7150 [30:28<1:39:06,  1.21s/it]
 31%|███▏      | 2237/7150 [30:29<1:25:25,  1.04s/it]
 31%|███▏      | 2238/7150 [30:29<1:15:49,  1.08it/s]
 31%|███▏      | 2239/7150 [30:30<1:09:06,  1.18it/s]
 31%|███▏      | 2240/7150 [30:30<1:04:25,  1.27it/s]
                                                     
{'loss': '3.245', 'grad_norm': '0.2262', 'learning_rate': '0.0006', 'epoch': '15.67'}
+
 31%|███▏      | 2240/7150 [30:30<1:04:25,  1.27it/s]
 31%|███▏      | 2241/7150 [30:31<1:01:10,  1.34it/s]
 31%|███▏      | 2242/7150 [30:32<58:51,  1.39it/s]  
 31%|███▏      | 2243/7150 [30:32<57:16,  1.43it/s]
 31%|███▏      | 2244/7150 [30:33<56:07,  1.46it/s]
 31%|███▏      | 2245/7150 [30:34<55:19,  1.48it/s]
 31%|███▏      | 2246/7150 [30:34<54:44,  1.49it/s]
 31%|███▏      | 2247/7150 [30:35<54:22,  1.50it/s]
 31%|███▏      | 2248/7150 [30:36<54:05,  1.51it/s]
 31%|███▏      | 2249/7150 [30:36<53:52,  1.52it/s]
 31%|███▏      | 2250/7150 [30:37<53:45,  1.52it/s]
                                                   
{'loss': '3.261', 'grad_norm': '0.222', 'learning_rate': '0.0006', 'epoch': '15.74'}
+
 31%|███▏      | 2250/7150 [30:37<53:45,  1.52it/s]
 31%|███▏      | 2251/7150 [30:38<53:41,  1.52it/s]
 31%|███▏      | 2252/7150 [30:38<53:34,  1.52it/s]
 32%|███▏      | 2253/7150 [30:39<53:28,  1.53it/s]
 32%|███▏      | 2254/7150 [30:40<53:28,  1.53it/s]
 32%|███▏      | 2255/7150 [30:40<53:28,  1.53it/s]
 32%|███▏      | 2256/7150 [30:41<53:26,  1.53it/s]
 32%|███▏      | 2257/7150 [30:42<53:23,  1.53it/s]
 32%|███▏      | 2258/7150 [30:42<53:22,  1.53it/s]
 32%|███▏      | 2259/7150 [30:43<53:20,  1.53it/s]
 32%|███▏      | 2260/7150 [30:44<53:17,  1.53it/s]
                                                   
{'loss': '3.265', 'grad_norm': '0.2339', 'learning_rate': '0.0006', 'epoch': '15.81'}
+
 32%|███▏      | 2260/7150 [30:44<53:17,  1.53it/s]
 32%|███▏      | 2261/7150 [30:44<53:21,  1.53it/s]
 32%|███▏      | 2262/7150 [30:45<53:21,  1.53it/s]
 32%|███▏      | 2263/7150 [30:46<53:19,  1.53it/s]
 32%|███▏      | 2264/7150 [30:46<53:15,  1.53it/s]
 32%|███▏      | 2265/7150 [30:47<53:14,  1.53it/s]
 32%|███▏      | 2266/7150 [30:47<53:14,  1.53it/s]
 32%|███▏      | 2267/7150 [30:48<53:12,  1.53it/s]
 32%|███▏      | 2268/7150 [30:49<53:14,  1.53it/s]
 32%|███▏      | 2269/7150 [30:49<53:13,  1.53it/s]
 32%|███▏      | 2270/7150 [30:50<53:13,  1.53it/s]
                                                   
{'loss': '3.273', 'grad_norm': '0.2426', 'learning_rate': '0.0006', 'epoch': '15.88'}
+
 32%|███▏      | 2270/7150 [30:50<53:13,  1.53it/s]
 32%|███▏      | 2271/7150 [30:51<53:12,  1.53it/s]
 32%|███▏      | 2272/7150 [30:51<53:12,  1.53it/s]
 32%|███▏      | 2273/7150 [30:52<53:10,  1.53it/s]
 32%|███▏      | 2274/7150 [30:53<53:08,  1.53it/s]
 32%|███▏      | 2275/7150 [30:53<53:09,  1.53it/s]
 32%|███▏      | 2276/7150 [30:54<53:08,  1.53it/s]
 32%|███▏      | 2277/7150 [30:55<53:09,  1.53it/s]
 32%|███▏      | 2278/7150 [30:55<53:08,  1.53it/s]
 32%|███▏      | 2279/7150 [30:56<53:07,  1.53it/s]
 32%|███▏      | 2280/7150 [30:57<53:05,  1.53it/s]
                                                   
{'loss': '3.262', 'grad_norm': '0.2468', 'learning_rate': '0.0006', 'epoch': '15.95'}
+
 32%|███▏      | 2280/7150 [30:57<53:05,  1.53it/s]
 32%|███▏      | 2281/7150 [30:57<53:05,  1.53it/s]
 32%|███▏      | 2282/7150 [30:58<53:06,  1.53it/s]
 32%|███▏      | 2283/7150 [30:59<53:05,  1.53it/s]
 32%|███▏      | 2284/7150 [30:59<53:05,  1.53it/s]
 32%|███▏      | 2285/7150 [31:00<53:02,  1.53it/s]
 32%|███▏      | 2286/7150 [31:01<53:07,  1.53it/s]
 32%|███▏      | 2287/7150 [31:01<53:04,  1.53it/s]
 32%|███▏      | 2288/7150 [31:02<49:08,  1.65it/s]
 32%|███▏      | 2289/7150 [31:02<51:55,  1.56it/s]
 32%|███▏      | 2290/7150 [31:03<52:12,  1.55it/s]
                                                   
{'loss': '3.247', 'grad_norm': '0.3411', 'learning_rate': '0.0006', 'epoch': '16.01'}
+
 32%|███▏      | 2290/7150 [31:03<52:12,  1.55it/s]
 32%|███▏      | 2291/7150 [31:04<52:27,  1.54it/s]
 32%|███▏      | 2292/7150 [31:04<52:36,  1.54it/s]
 32%|███▏      | 2293/7150 [31:05<52:42,  1.54it/s]
 32%|███▏      | 2294/7150 [31:06<52:47,  1.53it/s]
 32%|███▏      | 2295/7150 [31:06<52:50,  1.53it/s]
 32%|███▏      | 2296/7150 [31:07<52:49,  1.53it/s]
 32%|███▏      | 2297/7150 [31:08<52:49,  1.53it/s]
 32%|███▏      | 2298/7150 [31:08<52:51,  1.53it/s]
 32%|███▏      | 2299/7150 [31:09<52:50,  1.53it/s]
 32%|███▏      | 2300/7150 [31:10<52:49,  1.53it/s]
                                                   
{'loss': '3.158', 'grad_norm': '0.4073', 'learning_rate': '0.0006', 'epoch': '16.08'}
+
 32%|███▏      | 2300/7150 [31:10<52:49,  1.53it/s]
 32%|███▏      | 2301/7150 [31:10<52:54,  1.53it/s]
 32%|███▏      | 2302/7150 [31:11<52:53,  1.53it/s]
 32%|███▏      | 2303/7150 [31:12<52:54,  1.53it/s]
 32%|███▏      | 2304/7150 [31:12<52:52,  1.53it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.74it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.85it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.61it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.06it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.74it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.57it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.44it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.38it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.32it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.31it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.30it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.28it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.26it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.25it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.25it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.25it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.24it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.25it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s]
                                                   
+
                                               
{'eval_loss': '3.756', 'eval_runtime': '3.247', 'eval_samples_per_second': '751.8', 'eval_steps_per_second': '12.01', 'epoch': '16.11'}
+
 32%|███▏      | 2304/7150 [31:16<52:52,  1.53it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s][eval] epoch 16.112 step 2304 loss 3.7562 (best 3.7377 @ ep 15.609, 1/20 since best)
+
+
                                               
 32%|███▏      | 2305/7150 [31:16<2:11:32,  1.63s/it]
 32%|███▏      | 2306/7150 [31:17<1:47:53,  1.34s/it]
 32%|███▏      | 2307/7150 [31:17<1:31:22,  1.13s/it]
 32%|███▏      | 2308/7150 [31:18<1:19:47,  1.01it/s]
 32%|███▏      | 2309/7150 [31:19<1:11:39,  1.13it/s]
 32%|███▏      | 2310/7150 [31:19<1:06:00,  1.22it/s]
                                                     
{'loss': '3.166', 'grad_norm': '0.291', 'learning_rate': '0.0006', 'epoch': '16.15'}
+
 32%|███▏      | 2310/7150 [31:19<1:06:00,  1.22it/s]
 32%|███▏      | 2311/7150 [31:20<1:02:02,  1.30it/s]
 32%|███▏      | 2312/7150 [31:21<59:15,  1.36it/s]  
 32%|███▏      | 2313/7150 [31:21<57:15,  1.41it/s]
 32%|███▏      | 2314/7150 [31:22<55:55,  1.44it/s]
 32%|███▏      | 2315/7150 [31:23<54:57,  1.47it/s]
 32%|███▏      | 2316/7150 [31:23<54:16,  1.48it/s]
 32%|███▏      | 2317/7150 [31:24<53:48,  1.50it/s]
 32%|███▏      | 2318/7150 [31:25<53:28,  1.51it/s]
 32%|███▏      | 2319/7150 [31:25<53:14,  1.51it/s]
 32%|███▏      | 2320/7150 [31:26<53:01,  1.52it/s]
                                                   
{'loss': '3.188', 'grad_norm': '0.2713', 'learning_rate': '0.0006', 'epoch': '16.22'}
+
 32%|███▏      | 2320/7150 [31:26<53:01,  1.52it/s]
 32%|███▏      | 2321/7150 [31:27<52:58,  1.52it/s]
 32%|███▏      | 2322/7150 [31:27<52:52,  1.52it/s]
 32%|███▏      | 2323/7150 [31:28<52:49,  1.52it/s]
 33%|███▎      | 2324/7150 [31:29<52:43,  1.53it/s]
 33%|███▎      | 2325/7150 [31:29<52:43,  1.53it/s]
 33%|███▎      | 2326/7150 [31:30<52:41,  1.53it/s]
 33%|███▎      | 2327/7150 [31:31<52:40,  1.53it/s]
 33%|███▎      | 2328/7150 [31:31<52:37,  1.53it/s]
 33%|███▎      | 2329/7150 [31:32<52:37,  1.53it/s]
 33%|███▎      | 2330/7150 [31:33<52:39,  1.53it/s]
                                                   
{'loss': '3.196', 'grad_norm': '0.2392', 'learning_rate': '0.0006', 'epoch': '16.29'}
+
 33%|███▎      | 2330/7150 [31:33<52:39,  1.53it/s]
 33%|███▎      | 2331/7150 [31:33<52:38,  1.53it/s]
 33%|███▎      | 2332/7150 [31:34<52:34,  1.53it/s]
 33%|███▎      | 2333/7150 [31:35<52:35,  1.53it/s]
 33%|███▎      | 2334/7150 [31:35<52:35,  1.53it/s]
 33%|███▎      | 2335/7150 [31:36<52:34,  1.53it/s]
 33%|███▎      | 2336/7150 [31:36<52:31,  1.53it/s]
 33%|███▎      | 2337/7150 [31:37<52:31,  1.53it/s]
 33%|███▎      | 2338/7150 [31:38<52:31,  1.53it/s]
 33%|███▎      | 2339/7150 [31:38<52:31,  1.53it/s]
 33%|███▎      | 2340/7150 [31:39<52:28,  1.53it/s]
                                                   
{'loss': '3.214', 'grad_norm': '0.24', 'learning_rate': '0.0006', 'epoch': '16.36'}
+
 33%|███▎      | 2340/7150 [31:39<52:28,  1.53it/s]
 33%|███▎      | 2341/7150 [31:40<52:31,  1.53it/s]
 33%|███▎      | 2342/7150 [31:40<52:30,  1.53it/s]
 33%|███▎      | 2343/7150 [31:41<52:29,  1.53it/s]
 33%|███▎      | 2344/7150 [31:42<52:25,  1.53it/s]
 33%|███▎      | 2345/7150 [31:42<52:26,  1.53it/s]
 33%|███▎      | 2346/7150 [31:43<52:27,  1.53it/s]
 33%|███▎      | 2347/7150 [31:44<52:26,  1.53it/s]
 33%|███▎      | 2348/7150 [31:44<52:22,  1.53it/s]
 33%|███▎      | 2349/7150 [31:45<52:23,  1.53it/s]
 33%|███▎      | 2350/7150 [31:46<52:23,  1.53it/s]
                                                   
{'loss': '3.21', 'grad_norm': '0.2215', 'learning_rate': '0.0006', 'epoch': '16.43'}
+
 33%|███▎      | 2350/7150 [31:46<52:23,  1.53it/s]
 33%|███▎      | 2351/7150 [31:46<52:23,  1.53it/s]
 33%|███▎      | 2352/7150 [31:47<52:22,  1.53it/s]
 33%|███▎      | 2353/7150 [31:48<52:22,  1.53it/s]
 33%|███▎      | 2354/7150 [31:48<52:21,  1.53it/s]
 33%|███▎      | 2355/7150 [31:49<52:21,  1.53it/s]
 33%|███▎      | 2356/7150 [31:50<52:18,  1.53it/s]
 33%|███▎      | 2357/7150 [31:50<52:18,  1.53it/s]
 33%|███▎      | 2358/7150 [31:51<52:17,  1.53it/s]
 33%|███▎      | 2359/7150 [31:52<52:18,  1.53it/s]
 33%|███▎      | 2360/7150 [31:52<52:16,  1.53it/s]
                                                   
{'loss': '3.228', 'grad_norm': '0.269', 'learning_rate': '0.0006', 'epoch': '16.5'}
+
 33%|███▎      | 2360/7150 [31:52<52:16,  1.53it/s]
 33%|███▎      | 2361/7150 [31:53<52:18,  1.53it/s]
 33%|███▎      | 2362/7150 [31:53<52:17,  1.53it/s]
 33%|███▎      | 2363/7150 [31:54<52:17,  1.53it/s]
 33%|███▎      | 2364/7150 [31:55<52:12,  1.53it/s]
 33%|███▎      | 2365/7150 [31:55<52:11,  1.53it/s]
 33%|███▎      | 2366/7150 [31:56<52:09,  1.53it/s]
 33%|███▎      | 2367/7150 [31:57<52:07,  1.53it/s]
 33%|███▎      | 2368/7150 [31:57<52:08,  1.53it/s]
 33%|███▎      | 2369/7150 [31:58<52:10,  1.53it/s]
 33%|███▎      | 2370/7150 [31:59<52:09,  1.53it/s]
                                                   
{'loss': '3.232', 'grad_norm': '0.252', 'learning_rate': '0.0006', 'epoch': '16.57'}
+
 33%|███▎      | 2370/7150 [31:59<52:09,  1.53it/s]
 33%|███▎      | 2371/7150 [31:59<52:07,  1.53it/s]
 33%|███▎      | 2372/7150 [32:00<52:08,  1.53it/s]
 33%|███▎      | 2373/7150 [32:01<52:06,  1.53it/s]
 33%|███▎      | 2374/7150 [32:01<52:04,  1.53it/s]
 33%|███▎      | 2375/7150 [32:02<52:06,  1.53it/s]
 33%|███▎      | 2376/7150 [32:03<52:05,  1.53it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.49it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.81it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.59it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.06it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.77it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.57it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.47it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.38it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.34it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.30it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.28it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.27it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.25it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.24it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.24it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.23it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.24it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.23it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.22it/s]
                                                   
+
                                               
{'eval_loss': '3.739', 'eval_runtime': '3.251', 'eval_samples_per_second': '750.8', 'eval_steps_per_second': '11.99', 'epoch': '16.62'}
+
 33%|███▎      | 2376/7150 [32:06<52:05,  1.53it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.22it/s][eval] epoch 16.616 step 2376 loss 3.7390 (best 3.7377 @ ep 15.609, 2/20 since best)
+
+
                                               
 33%|███▎      | 2377/7150 [32:07<2:09:40,  1.63s/it]
 33%|███▎      | 2378/7150 [32:07<1:46:23,  1.34s/it]
 33%|███▎      | 2379/7150 [32:08<1:30:05,  1.13s/it]
 33%|███▎      | 2380/7150 [32:09<1:18:40,  1.01it/s]
                                                     
{'loss': '3.22', 'grad_norm': '0.2497', 'learning_rate': '0.0006', 'epoch': '16.64'}
+
 33%|███▎      | 2380/7150 [32:09<1:18:40,  1.01it/s]
 33%|███▎      | 2381/7150 [32:09<1:10:39,  1.12it/s]
 33%|███▎      | 2382/7150 [32:10<1:05:04,  1.22it/s]
 33%|███▎      | 2383/7150 [32:10<1:01:08,  1.30it/s]
 33%|███▎      | 2384/7150 [32:11<58:24,  1.36it/s]  
 33%|███▎      | 2385/7150 [32:12<56:28,  1.41it/s]
 33%|███▎      | 2386/7150 [32:12<55:07,  1.44it/s]
 33%|███▎      | 2387/7150 [32:13<54:09,  1.47it/s]
 33%|███▎      | 2388/7150 [32:14<1:01:41,  1.29it/s]
 33%|███▎      | 2389/7150 [32:15<58:43,  1.35it/s]  
 33%|███▎      | 2390/7150 [32:15<56:42,  1.40it/s]
                                                   
{'loss': '3.229', 'grad_norm': '0.2292', 'learning_rate': '0.0006', 'epoch': '16.71'}
+
 33%|███▎      | 2390/7150 [32:15<56:42,  1.40it/s]
 33%|███▎      | 2391/7150 [32:16<55:16,  1.43it/s]
 33%|███▎      | 2392/7150 [32:17<54:16,  1.46it/s]
 33%|███▎      | 2393/7150 [32:17<53:32,  1.48it/s]
 33%|███▎      | 2394/7150 [32:18<53:04,  1.49it/s]
 33%|███▎      | 2395/7150 [32:19<52:43,  1.50it/s]
 34%|███▎      | 2396/7150 [32:19<52:27,  1.51it/s]
 34%|███▎      | 2397/7150 [32:20<52:14,  1.52it/s]
 34%|███▎      | 2398/7150 [32:21<52:06,  1.52it/s]
 34%|███▎      | 2399/7150 [32:21<52:00,  1.52it/s]
 34%|███▎      | 2400/7150 [32:22<51:56,  1.52it/s]
                                                   
{'loss': '3.241', 'grad_norm': '0.2712', 'learning_rate': '0.0006', 'epoch': '16.78'}
+
 34%|███▎      | 2400/7150 [32:22<51:56,  1.52it/s]
 34%|███▎      | 2401/7150 [32:23<51:54,  1.52it/s]
 34%|███▎      | 2402/7150 [32:23<51:52,  1.53it/s]
 34%|███▎      | 2403/7150 [32:24<51:49,  1.53it/s]
 34%|███▎      | 2404/7150 [32:25<51:47,  1.53it/s]
 34%|███▎      | 2405/7150 [32:25<51:47,  1.53it/s]
 34%|███▎      | 2406/7150 [32:26<51:46,  1.53it/s]
 34%|███▎      | 2407/7150 [32:27<51:46,  1.53it/s]
 34%|███▎      | 2408/7150 [32:27<51:43,  1.53it/s]
 34%|███▎      | 2409/7150 [32:28<51:44,  1.53it/s]
 34%|███▎      | 2410/7150 [32:29<51:42,  1.53it/s]
                                                   
{'loss': '3.242', 'grad_norm': '0.2475', 'learning_rate': '0.0006', 'epoch': '16.85'}
+
 34%|███▎      | 2410/7150 [32:29<51:42,  1.53it/s]
 34%|███▎      | 2411/7150 [32:29<51:43,  1.53it/s]
 34%|███▎      | 2412/7150 [32:30<51:42,  1.53it/s]
 34%|███▎      | 2413/7150 [32:30<51:41,  1.53it/s]
 34%|███▍      | 2414/7150 [32:31<51:39,  1.53it/s]
 34%|███▍      | 2415/7150 [32:32<51:36,  1.53it/s]
 34%|███▍      | 2416/7150 [32:32<51:37,  1.53it/s]
 34%|███▍      | 2417/7150 [32:33<51:37,  1.53it/s]
 34%|███▍      | 2418/7150 [32:34<51:37,  1.53it/s]
 34%|███▍      | 2419/7150 [32:34<51:36,  1.53it/s]
 34%|███▍      | 2420/7150 [32:35<51:37,  1.53it/s]
                                                   
{'loss': '3.227', 'grad_norm': '0.2306', 'learning_rate': '0.0006', 'epoch': '16.92'}
+
 34%|███▍      | 2420/7150 [32:35<51:37,  1.53it/s]
 34%|███▍      | 2421/7150 [32:36<51:39,  1.53it/s]
 34%|███▍      | 2422/7150 [32:36<51:37,  1.53it/s]
 34%|███▍      | 2423/7150 [32:37<51:34,  1.53it/s]
 34%|███▍      | 2424/7150 [32:38<51:34,  1.53it/s]
 34%|███▍      | 2425/7150 [32:38<51:32,  1.53it/s]
 34%|███▍      | 2426/7150 [32:39<51:29,  1.53it/s]
 34%|███▍      | 2427/7150 [32:40<51:30,  1.53it/s]
 34%|███▍      | 2428/7150 [32:40<51:30,  1.53it/s]
 34%|███▍      | 2429/7150 [32:41<51:33,  1.53it/s]
 34%|███▍      | 2430/7150 [32:42<51:28,  1.53it/s]
                                                   
{'loss': '3.241', 'grad_norm': '0.2406', 'learning_rate': '0.0006', 'epoch': '16.99'}
+
 34%|███▍      | 2430/7150 [32:42<51:28,  1.53it/s]
 34%|███▍      | 2431/7150 [32:42<47:40,  1.65it/s]
 34%|███▍      | 2432/7150 [32:43<50:25,  1.56it/s]
 34%|███▍      | 2433/7150 [32:43<50:44,  1.55it/s]
 34%|███▍      | 2434/7150 [32:44<50:55,  1.54it/s]
 34%|███▍      | 2435/7150 [32:45<51:03,  1.54it/s]
 34%|███▍      | 2436/7150 [32:45<51:09,  1.54it/s]
 34%|███▍      | 2437/7150 [32:46<51:12,  1.53it/s]
 34%|███▍      | 2438/7150 [32:47<51:15,  1.53it/s]
 34%|███▍      | 2439/7150 [32:47<51:18,  1.53it/s]
 34%|███▍      | 2440/7150 [32:48<51:21,  1.53it/s]
                                                   
{'loss': '3.138', 'grad_norm': '0.477', 'learning_rate': '0.0006', 'epoch': '17.06'}
+
 34%|███▍      | 2440/7150 [32:48<51:21,  1.53it/s]
 34%|███▍      | 2441/7150 [32:49<51:22,  1.53it/s]
 34%|███▍      | 2442/7150 [32:49<51:20,  1.53it/s]
 34%|███▍      | 2443/7150 [32:50<51:21,  1.53it/s]
 34%|███▍      | 2444/7150 [32:51<51:19,  1.53it/s]
 34%|███▍      | 2445/7150 [32:51<51:18,  1.53it/s]
 34%|███▍      | 2446/7150 [32:52<51:16,  1.53it/s]
 34%|███▍      | 2447/7150 [32:53<51:17,  1.53it/s]
 34%|███▍      | 2448/7150 [32:53<51:15,  1.53it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.64it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.80it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.63it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.05it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.76it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.56it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.46it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.38it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.33it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.29it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.29it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.26it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.25it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.24it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.23it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.22it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.23it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.22it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.20it/s]
                                                   
+
                                               
{'eval_loss': '3.758', 'eval_runtime': '3.246', 'eval_samples_per_second': '752', 'eval_steps_per_second': '12.02', 'epoch': '17.12'}
+
 34%|███▍      | 2448/7150 [32:57<51:15,  1.53it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.20it/s][eval] epoch 17.119 step 2448 loss 3.7583 (best 3.7377 @ ep 15.609, 3/20 since best)
+
+
                                               
 34%|███▍      | 2449/7150 [32:57<2:07:33,  1.63s/it]
 34%|███▍      | 2450/7150 [32:58<1:44:39,  1.34s/it]
                                                     
{'loss': '3.137', 'grad_norm': '0.348', 'learning_rate': '0.0006', 'epoch': '17.13'}
+
 34%|███▍      | 2450/7150 [32:58<1:44:39,  1.34s/it]
 34%|███▍      | 2451/7150 [32:59<1:28:37,  1.13s/it]
 34%|███▍      | 2452/7150 [32:59<1:17:25,  1.01it/s]
 34%|███▍      | 2453/7150 [33:00<1:09:33,  1.13it/s]
 34%|███▍      | 2454/7150 [33:00<1:04:05,  1.22it/s]
 34%|███▍      | 2455/7150 [33:01<1:00:12,  1.30it/s]
 34%|███▍      | 2456/7150 [33:02<57:30,  1.36it/s]  
 34%|███▍      | 2457/7150 [33:02<55:37,  1.41it/s]
 34%|███▍      | 2458/7150 [33:03<54:18,  1.44it/s]
 34%|███▍      | 2459/7150 [33:04<53:20,  1.47it/s]
 34%|███▍      | 2460/7150 [33:04<52:42,  1.48it/s]
                                                   
{'loss': '3.142', 'grad_norm': '0.2793', 'learning_rate': '0.0006', 'epoch': '17.2'}
+
 34%|███▍      | 2460/7150 [33:04<52:42,  1.48it/s]
 34%|███▍      | 2461/7150 [33:05<52:16,  1.49it/s]
 34%|███▍      | 2462/7150 [33:06<51:57,  1.50it/s]
 34%|███▍      | 2463/7150 [33:06<51:41,  1.51it/s]
 34%|███▍      | 2464/7150 [33:07<51:33,  1.51it/s]
 34%|███▍      | 2465/7150 [33:08<51:27,  1.52it/s]
 34%|███▍      | 2466/7150 [33:08<51:20,  1.52it/s]
 35%|███▍      | 2467/7150 [33:09<51:16,  1.52it/s]
 35%|███▍      | 2468/7150 [33:10<51:13,  1.52it/s]
 35%|███▍      | 2469/7150 [33:10<51:11,  1.52it/s]
 35%|███▍      | 2470/7150 [33:11<51:08,  1.53it/s]
                                                   
{'loss': '3.147', 'grad_norm': '0.2579', 'learning_rate': '0.0006', 'epoch': '17.27'}
+
 35%|███▍      | 2470/7150 [33:11<51:08,  1.53it/s]
 35%|███▍      | 2471/7150 [33:12<51:08,  1.52it/s]
 35%|███▍      | 2472/7150 [33:12<51:05,  1.53it/s]
 35%|███▍      | 2473/7150 [33:13<51:04,  1.53it/s]
 35%|███▍      | 2474/7150 [33:14<51:02,  1.53it/s]
 35%|███▍      | 2475/7150 [33:15<59:07,  1.32it/s]
 35%|███▍      | 2476/7150 [33:15<56:40,  1.37it/s]
 35%|███▍      | 2477/7150 [33:16<54:57,  1.42it/s]
 35%|███▍      | 2478/7150 [33:17<53:44,  1.45it/s]
 35%|███▍      | 2479/7150 [33:17<52:53,  1.47it/s]
 35%|███▍      | 2480/7150 [33:18<52:16,  1.49it/s]
                                                   
{'loss': '3.164', 'grad_norm': '0.2558', 'learning_rate': '0.0006', 'epoch': '17.34'}
+
 35%|███▍      | 2480/7150 [33:18<52:16,  1.49it/s]
 35%|███▍      | 2481/7150 [33:19<51:55,  1.50it/s]
 35%|███▍      | 2482/7150 [33:19<51:37,  1.51it/s]
 35%|███▍      | 2483/7150 [33:20<51:23,  1.51it/s]
 35%|███▍      | 2484/7150 [33:20<51:12,  1.52it/s]
 35%|███▍      | 2485/7150 [33:21<51:07,  1.52it/s]
 35%|███▍      | 2486/7150 [33:22<51:01,  1.52it/s]
 35%|███▍      | 2487/7150 [33:23<59:04,  1.32it/s]
 35%|███▍      | 2488/7150 [33:23<56:33,  1.37it/s]
 35%|███▍      | 2489/7150 [33:24<54:50,  1.42it/s]
 35%|███▍      | 2490/7150 [33:25<53:37,  1.45it/s]
                                                   
{'loss': '3.182', 'grad_norm': '0.2476', 'learning_rate': '0.0006', 'epoch': '17.41'}
+
 35%|███▍      | 2490/7150 [33:25<53:37,  1.45it/s]
 35%|███▍      | 2491/7150 [33:25<52:47,  1.47it/s]
 35%|███▍      | 2492/7150 [33:26<52:09,  1.49it/s]
 35%|███▍      | 2493/7150 [33:27<51:46,  1.50it/s]
 35%|███▍      | 2494/7150 [33:27<51:28,  1.51it/s]
 35%|███▍      | 2495/7150 [33:28<51:15,  1.51it/s]
 35%|███▍      | 2496/7150 [33:29<51:04,  1.52it/s]
 35%|███▍      | 2497/7150 [33:29<50:58,  1.52it/s]
 35%|███▍      | 2498/7150 [33:30<50:55,  1.52it/s]
 35%|███▍      | 2499/7150 [33:31<50:48,  1.53it/s]
 35%|███▍      | 2500/7150 [33:31<50:47,  1.53it/s]
                                                   
{'loss': '3.174', 'grad_norm': '0.2287', 'learning_rate': '0.0006', 'epoch': '17.48'}
+
 35%|███▍      | 2500/7150 [33:31<50:47,  1.53it/s]
 35%|███▍      | 2501/7150 [33:32<50:47,  1.53it/s]
 35%|███▍      | 2502/7150 [33:33<50:44,  1.53it/s]
 35%|███▌      | 2503/7150 [33:33<50:42,  1.53it/s]
 35%|███▌      | 2504/7150 [33:34<50:43,  1.53it/s]
 35%|███▌      | 2505/7150 [33:35<50:43,  1.53it/s]
 35%|███▌      | 2506/7150 [33:35<50:41,  1.53it/s]
 35%|███▌      | 2507/7150 [33:36<50:38,  1.53it/s]
 35%|███▌      | 2508/7150 [33:37<50:39,  1.53it/s]
 35%|███▌      | 2509/7150 [33:37<50:38,  1.53it/s]
 35%|███▌      | 2510/7150 [33:38<50:37,  1.53it/s]
                                                   
{'loss': '3.202', 'grad_norm': '0.2443', 'learning_rate': '0.0006', 'epoch': '17.55'}
+
 35%|███▌      | 2510/7150 [33:38<50:37,  1.53it/s]
 35%|███▌      | 2511/7150 [33:38<50:36,  1.53it/s]
 35%|███▌      | 2512/7150 [33:39<50:36,  1.53it/s]
 35%|███▌      | 2513/7150 [33:40<50:34,  1.53it/s]
 35%|███▌      | 2514/7150 [33:40<50:33,  1.53it/s]
 35%|███▌      | 2515/7150 [33:41<58:22,  1.32it/s]
 35%|███▌      | 2516/7150 [33:42<55:59,  1.38it/s]
 35%|███▌      | 2517/7150 [33:43<54:18,  1.42it/s]
 35%|███▌      | 2518/7150 [33:43<53:09,  1.45it/s]
 35%|███▌      | 2519/7150 [33:44<52:20,  1.47it/s]
 35%|███▌      | 2520/7150 [33:45<51:44,  1.49it/s]
                                                   
{'loss': '3.198', 'grad_norm': '0.2456', 'learning_rate': '0.0006', 'epoch': '17.62'}
+
 35%|███▌      | 2520/7150 [33:45<51:44,  1.49it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.19it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.64it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.57it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.03it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.75it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.57it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.45it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.37it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.32it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.30it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.30it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.28it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.26it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.27it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.25it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.25it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.24it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.24it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s]
                                                   
+
                                               
{'eval_loss': '3.74', 'eval_runtime': '3.248', 'eval_samples_per_second': '751.6', 'eval_steps_per_second': '12.01', 'epoch': '17.62'}
+
 35%|███▌      | 2520/7150 [33:48<51:44,  1.49it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s][eval] epoch 17.623 step 2520 loss 3.7395 (best 3.7377 @ ep 15.609, 4/20 since best)
+
+
                                               
 35%|███▌      | 2521/7150 [33:49<2:06:35,  1.64s/it]
 35%|███▌      | 2522/7150 [33:49<1:43:41,  1.34s/it]
 35%|███▌      | 2523/7150 [33:50<1:27:44,  1.14s/it]
 35%|███▌      | 2524/7150 [33:51<1:16:33,  1.01it/s]
 35%|███▌      | 2525/7150 [33:51<1:08:43,  1.12it/s]
 35%|███▌      | 2526/7150 [33:52<1:03:13,  1.22it/s]
 35%|███▌      | 2527/7150 [33:53<59:22,  1.30it/s]  
 35%|███▌      | 2528/7150 [33:53<56:39,  1.36it/s]
 35%|███▌      | 2529/7150 [33:54<54:44,  1.41it/s]
 35%|███▌      | 2530/7150 [33:55<53:26,  1.44it/s]
                                                   
{'loss': '3.214', 'grad_norm': '0.2632', 'learning_rate': '0.0006', 'epoch': '17.69'}
+
 35%|███▌      | 2530/7150 [33:55<53:26,  1.44it/s]
 35%|███▌      | 2531/7150 [33:55<52:32,  1.47it/s]
 35%|███▌      | 2532/7150 [33:56<51:53,  1.48it/s]
 35%|███▌      | 2533/7150 [33:56<51:24,  1.50it/s]
 35%|███▌      | 2534/7150 [33:57<51:05,  1.51it/s]
 35%|███▌      | 2535/7150 [33:58<50:52,  1.51it/s]
 35%|███▌      | 2536/7150 [33:58<50:41,  1.52it/s]
 35%|███▌      | 2537/7150 [33:59<50:34,  1.52it/s]
 35%|███▌      | 2538/7150 [34:00<50:29,  1.52it/s]
 36%|███▌      | 2539/7150 [34:00<50:25,  1.52it/s]
 36%|███▌      | 2540/7150 [34:01<50:19,  1.53it/s]
                                                   
{'loss': '3.205', 'grad_norm': '0.2264', 'learning_rate': '0.0006', 'epoch': '17.76'}
+
 36%|███▌      | 2540/7150 [34:01<50:19,  1.53it/s]
 36%|███▌      | 2541/7150 [34:02<50:21,  1.53it/s]
 36%|███▌      | 2542/7150 [34:02<50:19,  1.53it/s]
 36%|███▌      | 2543/7150 [34:03<50:18,  1.53it/s]
 36%|███▌      | 2544/7150 [34:04<50:15,  1.53it/s]
 36%|███▌      | 2545/7150 [34:05<58:02,  1.32it/s]
 36%|███▌      | 2546/7150 [34:05<55:39,  1.38it/s]
 36%|███▌      | 2547/7150 [34:06<53:59,  1.42it/s]
 36%|███▌      | 2548/7150 [34:07<52:53,  1.45it/s]
 36%|███▌      | 2549/7150 [34:07<52:04,  1.47it/s]
 36%|███▌      | 2550/7150 [34:08<51:28,  1.49it/s]
                                                   
{'loss': '3.211', 'grad_norm': '0.2394', 'learning_rate': '0.0006', 'epoch': '17.83'}
+
 36%|███▌      | 2550/7150 [34:08<51:28,  1.49it/s]
 36%|███▌      | 2551/7150 [34:09<58:58,  1.30it/s]
 36%|███▌      | 2552/7150 [34:10<56:18,  1.36it/s]
 36%|███▌      | 2553/7150 [34:10<54:27,  1.41it/s]
 36%|███▌      | 2554/7150 [34:11<1:00:57,  1.26it/s]
 36%|███▌      | 2555/7150 [34:12<57:41,  1.33it/s]  
 36%|███▌      | 2556/7150 [34:13<55:24,  1.38it/s]
 36%|███▌      | 2557/7150 [34:13<53:48,  1.42it/s]
 36%|███▌      | 2558/7150 [34:14<1:00:40,  1.26it/s]
 36%|███▌      | 2559/7150 [34:15<57:28,  1.33it/s]  
 36%|███▌      | 2560/7150 [34:16<55:14,  1.38it/s]
                                                   
{'loss': '3.225', 'grad_norm': '0.2551', 'learning_rate': '0.0006', 'epoch': '17.9'}
+
 36%|███▌      | 2560/7150 [34:16<55:14,  1.38it/s]
 36%|███▌      | 2561/7150 [34:16<53:42,  1.42it/s]
 36%|███▌      | 2562/7150 [34:17<52:34,  1.45it/s]
 36%|███▌      | 2563/7150 [34:17<51:47,  1.48it/s]
 36%|███▌      | 2564/7150 [34:18<51:17,  1.49it/s]
 36%|███▌      | 2565/7150 [34:19<50:54,  1.50it/s]
 36%|███▌      | 2566/7150 [34:19<50:38,  1.51it/s]
 36%|███▌      | 2567/7150 [34:20<50:24,  1.52it/s]
 36%|███▌      | 2568/7150 [34:21<50:17,  1.52it/s]
 36%|███▌      | 2569/7150 [34:21<50:11,  1.52it/s]
 36%|███▌      | 2570/7150 [34:22<50:06,  1.52it/s]
                                                   
{'loss': '3.213', 'grad_norm': '0.2507', 'learning_rate': '0.0006', 'epoch': '17.97'}
+
 36%|███▌      | 2570/7150 [34:22<50:06,  1.52it/s]
 36%|███▌      | 2571/7150 [34:23<50:03,  1.52it/s]
 36%|███▌      | 2572/7150 [34:23<50:06,  1.52it/s]
 36%|███▌      | 2573/7150 [34:24<50:02,  1.52it/s]
 36%|███▌      | 2574/7150 [34:25<46:19,  1.65it/s]
 36%|███▌      | 2575/7150 [34:25<49:19,  1.55it/s]
 36%|███▌      | 2576/7150 [34:26<49:28,  1.54it/s]
 36%|███▌      | 2577/7150 [34:27<49:35,  1.54it/s]
 36%|███▌      | 2578/7150 [34:27<49:39,  1.53it/s]
 36%|███▌      | 2579/7150 [34:28<49:42,  1.53it/s]
 36%|███▌      | 2580/7150 [34:29<49:46,  1.53it/s]
                                                   
{'loss': '3.152', 'grad_norm': '0.3791', 'learning_rate': '0.0006', 'epoch': '18.04'}
+
 36%|███▌      | 2580/7150 [34:29<49:46,  1.53it/s]
 36%|███▌      | 2581/7150 [34:29<49:48,  1.53it/s]
 36%|███▌      | 2582/7150 [34:30<49:48,  1.53it/s]
 36%|███▌      | 2583/7150 [34:30<49:46,  1.53it/s]
 36%|███▌      | 2584/7150 [34:31<49:48,  1.53it/s]
 36%|███▌      | 2585/7150 [34:32<49:47,  1.53it/s]
 36%|███▌      | 2586/7150 [34:32<49:47,  1.53it/s]
 36%|███▌      | 2587/7150 [34:33<49:44,  1.53it/s]
 36%|███▌      | 2588/7150 [34:34<49:45,  1.53it/s]
 36%|███▌      | 2589/7150 [34:34<49:44,  1.53it/s]
 36%|███▌      | 2590/7150 [34:35<49:42,  1.53it/s]
                                                   
{'loss': '3.1', 'grad_norm': '0.3227', 'learning_rate': '0.0006', 'epoch': '18.11'}
+
 36%|███▌      | 2590/7150 [34:35<49:42,  1.53it/s]
 36%|███▌      | 2591/7150 [34:36<49:44,  1.53it/s]
 36%|███▋      | 2592/7150 [34:36<49:44,  1.53it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.70it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.78it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.63it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.06it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.77it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.58it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.46it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.35it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.32it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.28it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.29it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.27it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.27it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.26it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.25it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.25it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.24it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.23it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s]
                                                   
+
                                               
{'eval_loss': '3.761', 'eval_runtime': '3.245', 'eval_samples_per_second': '752.2', 'eval_steps_per_second': '12.02', 'epoch': '18.13'}
+
 36%|███▋      | 2592/7150 [34:40<49:44,  1.53it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s][eval] epoch 18.126 step 2592 loss 3.7608 (best 3.7377 @ ep 15.609, 5/20 since best)
+
+
                                               
 36%|███▋      | 2593/7150 [34:40<2:03:42,  1.63s/it]
 36%|███▋      | 2594/7150 [34:41<1:41:28,  1.34s/it]
 36%|███▋      | 2595/7150 [34:42<1:25:54,  1.13s/it]
 36%|███▋      | 2596/7150 [34:42<1:15:03,  1.01it/s]
 36%|███▋      | 2597/7150 [34:43<1:07:28,  1.12it/s]
 36%|███▋      | 2598/7150 [34:44<1:02:07,  1.22it/s]
 36%|███▋      | 2599/7150 [34:44<58:23,  1.30it/s]  
 36%|███▋      | 2600/7150 [34:45<55:44,  1.36it/s]
                                                   
{'loss': '3.117', 'grad_norm': '0.3401', 'learning_rate': '0.0006', 'epoch': '18.18'}
+
 36%|███▋      | 2600/7150 [34:45<55:44,  1.36it/s]
 36%|███▋      | 2601/7150 [34:46<53:58,  1.40it/s]
 36%|███▋      | 2602/7150 [34:46<52:41,  1.44it/s]
 36%|███▋      | 2603/7150 [34:47<51:44,  1.46it/s]
 36%|███▋      | 2604/7150 [34:47<51:05,  1.48it/s]
 36%|███▋      | 2605/7150 [34:48<50:40,  1.49it/s]
 36%|███▋      | 2606/7150 [34:49<50:21,  1.50it/s]
 36%|███▋      | 2607/7150 [34:49<50:07,  1.51it/s]
 36%|███▋      | 2608/7150 [34:50<49:56,  1.52it/s]
 36%|███▋      | 2609/7150 [34:51<49:50,  1.52it/s]
 37%|███▋      | 2610/7150 [34:51<49:46,  1.52it/s]
                                                   
{'loss': '3.129', 'grad_norm': '0.3217', 'learning_rate': '0.0006', 'epoch': '18.25'}
+
 37%|███▋      | 2610/7150 [34:51<49:46,  1.52it/s]
 37%|███▋      | 2611/7150 [34:52<49:44,  1.52it/s]
 37%|███▋      | 2612/7150 [34:53<49:40,  1.52it/s]
 37%|███▋      | 2613/7150 [34:53<49:36,  1.52it/s]
 37%|███▋      | 2614/7150 [34:54<49:36,  1.52it/s]
 37%|███▋      | 2615/7150 [34:55<49:34,  1.52it/s]
 37%|███▋      | 2616/7150 [34:55<49:32,  1.53it/s]
 37%|███▋      | 2617/7150 [34:56<49:31,  1.53it/s]
 37%|███▋      | 2618/7150 [34:57<49:30,  1.53it/s]
 37%|███▋      | 2619/7150 [34:57<49:31,  1.53it/s]
 37%|███▋      | 2620/7150 [34:58<49:31,  1.52it/s]
                                                   
{'loss': '3.137', 'grad_norm': '0.2583', 'learning_rate': '0.0006', 'epoch': '18.32'}
+
 37%|███▋      | 2620/7150 [34:58<49:31,  1.52it/s]
 37%|███▋      | 2621/7150 [34:59<49:31,  1.52it/s]
 37%|███▋      | 2622/7150 [34:59<49:27,  1.53it/s]
 37%|███▋      | 2623/7150 [35:00<49:28,  1.53it/s]
 37%|███▋      | 2624/7150 [35:01<49:27,  1.53it/s]
 37%|███▋      | 2625/7150 [35:01<49:25,  1.53it/s]
 37%|███▋      | 2626/7150 [35:02<49:24,  1.53it/s]
 37%|███▋      | 2627/7150 [35:03<49:23,  1.53it/s]
 37%|███▋      | 2628/7150 [35:03<49:23,  1.53it/s]
 37%|███▋      | 2629/7150 [35:04<49:21,  1.53it/s]
 37%|███▋      | 2630/7150 [35:05<49:22,  1.53it/s]
                                                   
{'loss': '3.152', 'grad_norm': '0.2601', 'learning_rate': '0.0006', 'epoch': '18.39'}
+
 37%|███▋      | 2630/7150 [35:05<49:22,  1.53it/s]
 37%|███▋      | 2631/7150 [35:05<49:20,  1.53it/s]
 37%|███▋      | 2632/7150 [35:06<49:20,  1.53it/s]
 37%|███▋      | 2633/7150 [35:06<49:19,  1.53it/s]
 37%|███▋      | 2634/7150 [35:07<49:20,  1.53it/s]
 37%|███▋      | 2635/7150 [35:08<49:17,  1.53it/s]
 37%|███▋      | 2636/7150 [35:08<49:17,  1.53it/s]
 37%|███▋      | 2637/7150 [35:09<49:15,  1.53it/s]
 37%|███▋      | 2638/7150 [35:10<49:17,  1.53it/s]
 37%|███▋      | 2639/7150 [35:10<49:15,  1.53it/s]
 37%|███▋      | 2640/7150 [35:11<49:15,  1.53it/s]
                                                   
{'loss': '3.152', 'grad_norm': '0.2718', 'learning_rate': '0.0006', 'epoch': '18.46'}
+
 37%|███▋      | 2640/7150 [35:11<49:15,  1.53it/s]
 37%|███▋      | 2641/7150 [35:12<49:16,  1.53it/s]
 37%|███▋      | 2642/7150 [35:12<49:14,  1.53it/s]
 37%|███▋      | 2643/7150 [35:13<49:13,  1.53it/s]
 37%|███▋      | 2644/7150 [35:14<49:13,  1.53it/s]
 37%|███▋      | 2645/7150 [35:14<49:12,  1.53it/s]
 37%|███▋      | 2646/7150 [35:15<49:12,  1.53it/s]
 37%|███▋      | 2647/7150 [35:16<49:13,  1.52it/s]
 37%|███▋      | 2648/7150 [35:16<49:09,  1.53it/s]
 37%|███▋      | 2649/7150 [35:17<49:09,  1.53it/s]
 37%|███▋      | 2650/7150 [35:18<49:10,  1.53it/s]
                                                   
{'loss': '3.166', 'grad_norm': '0.282', 'learning_rate': '0.0006', 'epoch': '18.53'}
+
 37%|███▋      | 2650/7150 [35:18<49:10,  1.53it/s]
 37%|███▋      | 2651/7150 [35:18<49:10,  1.53it/s]
 37%|███▋      | 2652/7150 [35:19<49:08,  1.53it/s]
 37%|███▋      | 2653/7150 [35:20<49:07,  1.53it/s]
 37%|███▋      | 2654/7150 [35:20<49:06,  1.53it/s]
 37%|███▋      | 2655/7150 [35:21<49:04,  1.53it/s]
 37%|███▋      | 2656/7150 [35:22<49:04,  1.53it/s]
 37%|███▋      | 2657/7150 [35:22<49:04,  1.53it/s]
 37%|███▋      | 2658/7150 [35:23<49:03,  1.53it/s]
 37%|███▋      | 2659/7150 [35:24<49:02,  1.53it/s]
 37%|███▋      | 2660/7150 [35:24<49:01,  1.53it/s]
                                                   
{'loss': '3.174', 'grad_norm': '0.2459', 'learning_rate': '0.0006', 'epoch': '18.6'}
+
 37%|███▋      | 2660/7150 [35:24<49:01,  1.53it/s]
 37%|███▋      | 2661/7150 [35:25<49:01,  1.53it/s]
 37%|███▋      | 2662/7150 [35:26<49:01,  1.53it/s]
 37%|███▋      | 2663/7150 [35:26<48:59,  1.53it/s]
 37%|███▋      | 2664/7150 [35:27<49:00,  1.53it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.82it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.87it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.65it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.08it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.77it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.58it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.46it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.38it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.35it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.30it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.30it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.27it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.25it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.24it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.23it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.23it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.24it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.23it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.22it/s]
                                                   
+
                                               
{'eval_loss': '3.747', 'eval_runtime': '3.246', 'eval_samples_per_second': '752', 'eval_steps_per_second': '12.02', 'epoch': '18.63'}
+
 37%|███▋      | 2664/7150 [35:30<49:00,  1.53it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.22it/s][eval] epoch 18.63 step 2664 loss 3.7475 (best 3.7377 @ ep 15.609, 6/20 since best)
+
+
                                               
 37%|███▋      | 2665/7150 [35:31<2:01:45,  1.63s/it]
 37%|███▋      | 2666/7150 [35:31<1:39:51,  1.34s/it]
 37%|███▋      | 2667/7150 [35:32<1:24:33,  1.13s/it]
 37%|███▋      | 2668/7150 [35:33<1:13:50,  1.01it/s]
 37%|███▋      | 2669/7150 [35:33<1:06:17,  1.13it/s]
 37%|███▋      | 2670/7150 [35:34<1:01:03,  1.22it/s]
                                                     
{'loss': '3.177', 'grad_norm': '0.2397', 'learning_rate': '0.0006', 'epoch': '18.67'}
+
 37%|███▋      | 2670/7150 [35:34<1:01:03,  1.22it/s]
 37%|███▋      | 2671/7150 [35:35<57:25,  1.30it/s]  
 37%|███▋      | 2672/7150 [35:35<54:50,  1.36it/s]
 37%|███▋      | 2673/7150 [35:36<53:00,  1.41it/s]
 37%|███▋      | 2674/7150 [35:37<51:45,  1.44it/s]
 37%|███▋      | 2675/7150 [35:37<50:51,  1.47it/s]
 37%|███▋      | 2676/7150 [35:38<50:15,  1.48it/s]
 37%|███▋      | 2677/7150 [35:39<49:49,  1.50it/s]
 37%|███▋      | 2678/7150 [35:39<49:31,  1.51it/s]
 37%|███▋      | 2679/7150 [35:40<49:17,  1.51it/s]
 37%|███▋      | 2680/7150 [35:41<49:06,  1.52it/s]
                                                   
{'loss': '3.187', 'grad_norm': '0.2427', 'learning_rate': '0.0006', 'epoch': '18.74'}
+
 37%|███▋      | 2680/7150 [35:41<49:06,  1.52it/s]
 37%|███▋      | 2681/7150 [35:41<49:01,  1.52it/s]
 38%|███▊      | 2682/7150 [35:42<48:55,  1.52it/s]
 38%|███▊      | 2683/7150 [35:42<48:54,  1.52it/s]
 38%|███▊      | 2684/7150 [35:43<48:50,  1.52it/s]
 38%|███▊      | 2685/7150 [35:44<48:47,  1.52it/s]
 38%|███▊      | 2686/7150 [35:44<48:46,  1.53it/s]
 38%|███▊      | 2687/7150 [35:45<48:43,  1.53it/s]
 38%|███▊      | 2688/7150 [35:46<48:40,  1.53it/s]
 38%|███▊      | 2689/7150 [35:46<48:42,  1.53it/s]
 38%|███▊      | 2690/7150 [35:47<48:40,  1.53it/s]
                                                   
{'loss': '3.178', 'grad_norm': '0.2346', 'learning_rate': '0.0006', 'epoch': '18.81'}
+
 38%|███▊      | 2690/7150 [35:47<48:40,  1.53it/s]
 38%|███▊      | 2691/7150 [35:48<48:40,  1.53it/s]
 38%|███▊      | 2692/7150 [35:48<48:37,  1.53it/s]
 38%|███▊      | 2693/7150 [35:49<48:37,  1.53it/s]
 38%|███▊      | 2694/7150 [35:50<48:37,  1.53it/s]
 38%|███▊      | 2695/7150 [35:50<48:36,  1.53it/s]
 38%|███▊      | 2696/7150 [35:51<48:35,  1.53it/s]
 38%|███▊      | 2697/7150 [35:52<48:34,  1.53it/s]
 38%|███▊      | 2698/7150 [35:52<48:33,  1.53it/s]
 38%|███▊      | 2699/7150 [35:53<48:30,  1.53it/s]
 38%|███▊      | 2700/7150 [35:54<48:30,  1.53it/s]
                                                   
{'loss': '3.172', 'grad_norm': '0.2376', 'learning_rate': '0.0006', 'epoch': '18.88'}
+
 38%|███▊      | 2700/7150 [35:54<48:30,  1.53it/s]
 38%|███▊      | 2701/7150 [35:54<48:31,  1.53it/s]
 38%|███▊      | 2702/7150 [35:55<48:30,  1.53it/s]
 38%|███▊      | 2703/7150 [35:56<48:30,  1.53it/s]
 38%|███▊      | 2704/7150 [35:56<48:29,  1.53it/s]
 38%|███▊      | 2705/7150 [35:57<48:27,  1.53it/s]
 38%|███▊      | 2706/7150 [35:58<48:30,  1.53it/s]
 38%|███▊      | 2707/7150 [35:58<48:30,  1.53it/s]
 38%|███▊      | 2708/7150 [35:59<48:29,  1.53it/s]
 38%|███▊      | 2709/7150 [36:00<48:28,  1.53it/s]
 38%|███▊      | 2710/7150 [36:00<48:26,  1.53it/s]
                                                   
{'loss': '3.197', 'grad_norm': '0.2555', 'learning_rate': '0.0006', 'epoch': '18.95'}
+
 38%|███▊      | 2710/7150 [36:00<48:26,  1.53it/s]
 38%|███▊      | 2711/7150 [36:01<48:43,  1.52it/s]
 38%|███▊      | 2712/7150 [36:01<48:37,  1.52it/s]
 38%|███▊      | 2713/7150 [36:02<48:32,  1.52it/s]
 38%|███▊      | 2714/7150 [36:03<48:30,  1.52it/s]
 38%|███▊      | 2715/7150 [36:03<48:36,  1.52it/s]
 38%|███▊      | 2716/7150 [36:04<48:31,  1.52it/s]
 38%|███▊      | 2717/7150 [36:05<44:51,  1.65it/s]
 38%|███▊      | 2718/7150 [36:05<47:44,  1.55it/s]
 38%|███▊      | 2719/7150 [36:06<47:55,  1.54it/s]
 38%|███▊      | 2720/7150 [36:07<48:02,  1.54it/s]
                                                   
{'loss': '3.142', 'grad_norm': '0.2954', 'learning_rate': '0.0006', 'epoch': '19.02'}
+
 38%|███▊      | 2720/7150 [36:07<48:02,  1.54it/s]
 38%|███▊      | 2721/7150 [36:07<48:06,  1.53it/s]
 38%|███▊      | 2722/7150 [36:08<48:11,  1.53it/s]
 38%|███▊      | 2723/7150 [36:09<48:13,  1.53it/s]
 38%|███▊      | 2724/7150 [36:09<48:13,  1.53it/s]
 38%|███▊      | 2725/7150 [36:10<48:13,  1.53it/s]
 38%|███▊      | 2726/7150 [36:11<48:16,  1.53it/s]
 38%|███▊      | 2727/7150 [36:11<48:15,  1.53it/s]
 38%|███▊      | 2728/7150 [36:12<48:14,  1.53it/s]
 38%|███▊      | 2729/7150 [36:13<48:12,  1.53it/s]
 38%|███▊      | 2730/7150 [36:13<48:15,  1.53it/s]
                                                   
{'loss': '3.068', 'grad_norm': '0.4644', 'learning_rate': '0.0006', 'epoch': '19.09'}
+
 38%|███▊      | 2730/7150 [36:13<48:15,  1.53it/s]
 38%|███▊      | 2731/7150 [36:14<48:18,  1.52it/s]
 38%|███▊      | 2732/7150 [36:15<48:16,  1.53it/s]
 38%|███▊      | 2733/7150 [36:15<48:14,  1.53it/s]
 38%|███▊      | 2734/7150 [36:16<48:11,  1.53it/s]
 38%|███▊      | 2735/7150 [36:16<48:12,  1.53it/s]
 38%|███▊      | 2736/7150 [36:17<48:10,  1.53it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 19.31it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 15.02it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.73it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.13it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.79it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.59it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.47it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.38it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.34it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.29it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.31it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.29it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.27it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.25it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.25it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.24it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.23it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.24it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s]
                                                   
+
                                               
{'eval_loss': '3.772', 'eval_runtime': '3.241', 'eval_samples_per_second': '753.1', 'eval_steps_per_second': '12.03', 'epoch': '19.13'}
+
 38%|███▊      | 2736/7150 [36:20<48:10,  1.53it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s][eval] epoch 19.133 step 2736 loss 3.7719 (best 3.7377 @ ep 15.609, 7/20 since best)
+
+
                                               
 38%|███▊      | 2737/7150 [36:21<1:59:41,  1.63s/it]
 38%|███▊      | 2738/7150 [36:22<1:38:10,  1.34s/it]
 38%|███▊      | 2739/7150 [36:22<1:23:08,  1.13s/it]
 38%|███▊      | 2740/7150 [36:23<1:12:36,  1.01it/s]
                                                     
{'loss': '3.094', 'grad_norm': '0.3338', 'learning_rate': '0.0006', 'epoch': '19.16'}
+
 38%|███▊      | 2740/7150 [36:23<1:12:36,  1.01it/s]
 38%|███▊      | 2741/7150 [36:24<1:05:16,  1.13it/s]
 38%|███▊      | 2742/7150 [36:24<1:00:07,  1.22it/s]
 38%|███▊      | 2743/7150 [36:25<56:30,  1.30it/s]  
 38%|███▊      | 2744/7150 [36:26<53:59,  1.36it/s]
 38%|███▊      | 2745/7150 [36:26<52:10,  1.41it/s]
 38%|███▊      | 2746/7150 [36:27<50:56,  1.44it/s]
 38%|███▊      | 2747/7150 [36:28<50:05,  1.46it/s]
 38%|███▊      | 2748/7150 [36:28<49:29,  1.48it/s]
 38%|███▊      | 2749/7150 [36:29<49:02,  1.50it/s]
 38%|███▊      | 2750/7150 [36:30<48:44,  1.50it/s]
                                                   
{'loss': '3.115', 'grad_norm': '0.297', 'learning_rate': '0.0006', 'epoch': '19.23'}
+
 38%|███▊      | 2750/7150 [36:30<48:44,  1.50it/s]
 38%|███▊      | 2751/7150 [36:30<48:32,  1.51it/s]
 38%|███▊      | 2752/7150 [36:31<48:22,  1.52it/s]
 39%|███▊      | 2753/7150 [36:32<48:17,  1.52it/s]
 39%|███▊      | 2754/7150 [36:32<48:10,  1.52it/s]
 39%|███▊      | 2755/7150 [36:33<48:07,  1.52it/s]
 39%|███▊      | 2756/7150 [36:33<48:04,  1.52it/s]
 39%|███▊      | 2757/7150 [36:34<48:01,  1.52it/s]
 39%|███▊      | 2758/7150 [36:35<47:59,  1.53it/s]
 39%|███▊      | 2759/7150 [36:35<47:58,  1.53it/s]
 39%|███▊      | 2760/7150 [36:36<47:57,  1.53it/s]
                                                   
{'loss': '3.114', 'grad_norm': '0.2635', 'learning_rate': '0.0006', 'epoch': '19.3'}
+
 39%|███▊      | 2760/7150 [36:36<47:57,  1.53it/s]
 39%|███▊      | 2761/7150 [36:37<47:56,  1.53it/s]
 39%|███▊      | 2762/7150 [36:37<47:53,  1.53it/s]
 39%|███▊      | 2763/7150 [36:38<47:54,  1.53it/s]
 39%|███▊      | 2764/7150 [36:39<47:53,  1.53it/s]
 39%|███▊      | 2765/7150 [36:39<47:51,  1.53it/s]
 39%|███▊      | 2766/7150 [36:40<47:50,  1.53it/s]
 39%|███▊      | 2767/7150 [36:41<47:52,  1.53it/s]
 39%|███▊      | 2768/7150 [36:41<47:50,  1.53it/s]
 39%|███▊      | 2769/7150 [36:42<47:49,  1.53it/s]
 39%|███▊      | 2770/7150 [36:43<47:47,  1.53it/s]
                                                   
{'loss': '3.116', 'grad_norm': '0.2627', 'learning_rate': '0.0006', 'epoch': '19.37'}
+
 39%|███▊      | 2770/7150 [36:43<47:47,  1.53it/s]
 39%|███▉      | 2771/7150 [36:43<47:49,  1.53it/s]
 39%|███▉      | 2772/7150 [36:44<47:48,  1.53it/s]
 39%|███▉      | 2773/7150 [36:45<47:47,  1.53it/s]
 39%|███▉      | 2774/7150 [36:45<47:45,  1.53it/s]
 39%|███▉      | 2775/7150 [36:46<47:46,  1.53it/s]
 39%|███▉      | 2776/7150 [36:47<47:45,  1.53it/s]
 39%|███▉      | 2777/7150 [36:47<47:43,  1.53it/s]
 39%|███▉      | 2778/7150 [36:48<47:41,  1.53it/s]
 39%|███▉      | 2779/7150 [36:49<47:42,  1.53it/s]
 39%|███▉      | 2780/7150 [36:49<47:42,  1.53it/s]
                                                   
{'loss': '3.135', 'grad_norm': '0.2554', 'learning_rate': '0.0006', 'epoch': '19.44'}
+
 39%|███▉      | 2780/7150 [36:49<47:42,  1.53it/s]
 39%|███▉      | 2781/7150 [36:50<47:43,  1.53it/s]
 39%|███▉      | 2782/7150 [36:51<47:43,  1.53it/s]
 39%|███▉      | 2783/7150 [36:51<47:41,  1.53it/s]
 39%|███▉      | 2784/7150 [36:52<47:40,  1.53it/s]
 39%|███▉      | 2785/7150 [36:52<47:40,  1.53it/s]
 39%|███▉      | 2786/7150 [36:53<47:40,  1.53it/s]
 39%|███▉      | 2787/7150 [36:54<47:37,  1.53it/s]
 39%|███▉      | 2788/7150 [36:54<47:38,  1.53it/s]
 39%|███▉      | 2789/7150 [36:55<47:38,  1.53it/s]
 39%|███▉      | 2790/7150 [36:56<47:37,  1.53it/s]
                                                   
{'loss': '3.137', 'grad_norm': '0.2735', 'learning_rate': '0.0006', 'epoch': '19.51'}
+
 39%|███▉      | 2790/7150 [36:56<47:37,  1.53it/s]
 39%|███▉      | 2791/7150 [36:56<47:37,  1.53it/s]
 39%|███▉      | 2792/7150 [36:57<47:37,  1.53it/s]
 39%|███▉      | 2793/7150 [36:58<47:36,  1.53it/s]
 39%|███▉      | 2794/7150 [36:58<47:35,  1.53it/s]
 39%|███▉      | 2795/7150 [36:59<47:34,  1.53it/s]
 39%|███▉      | 2796/7150 [37:00<47:33,  1.53it/s]
 39%|███▉      | 2797/7150 [37:00<47:32,  1.53it/s]
 39%|███▉      | 2798/7150 [37:01<47:32,  1.53it/s]
 39%|███▉      | 2799/7150 [37:02<47:32,  1.53it/s]
 39%|███▉      | 2800/7150 [37:02<47:28,  1.53it/s]
                                                   
{'loss': '3.137', 'grad_norm': '0.2391', 'learning_rate': '0.0006', 'epoch': '19.58'}
+
 39%|███▉      | 2800/7150 [37:02<47:28,  1.53it/s]
 39%|███▉      | 2801/7150 [37:03<47:32,  1.52it/s]
 39%|███▉      | 2802/7150 [37:04<47:31,  1.52it/s]
 39%|███▉      | 2803/7150 [37:04<47:29,  1.53it/s]
 39%|███▉      | 2804/7150 [37:05<47:26,  1.53it/s]
 39%|███▉      | 2805/7150 [37:06<47:27,  1.53it/s]
 39%|███▉      | 2806/7150 [37:06<47:26,  1.53it/s]
 39%|███▉      | 2807/7150 [37:07<47:25,  1.53it/s]
 39%|███▉      | 2808/7150 [37:08<47:23,  1.53it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.38it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.70it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.57it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.01it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.74it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.53it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.44it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.35it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.31it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.28it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.27it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.26it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.25it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.24it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.23it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.23it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.22it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.23it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s]
                                                   
+
                                               
{'eval_loss': '3.756', 'eval_runtime': '3.265', 'eval_samples_per_second': '747.6', 'eval_steps_per_second': '11.94', 'epoch': '19.64'}
+
 39%|███▉      | 2808/7150 [37:11<47:23,  1.53it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s][eval] epoch 19.637 step 2808 loss 3.7555 (best 3.7377 @ ep 15.609, 8/20 since best)
+
+
                                               
 39%|███▉      | 2809/7150 [37:11<1:58:14,  1.63s/it]
 39%|███▉      | 2810/7150 [37:12<1:36:59,  1.34s/it]
                                                     
{'loss': '3.152', 'grad_norm': '0.2497', 'learning_rate': '0.0006', 'epoch': '19.65'}
+
 39%|███▉      | 2810/7150 [37:12<1:36:59,  1.34s/it]
 39%|███▉      | 2811/7150 [37:13<1:22:08,  1.14s/it]
 39%|███▉      | 2812/7150 [37:13<1:11:41,  1.01it/s]
 39%|███▉      | 2813/7150 [37:14<1:04:21,  1.12it/s]
 39%|███▉      | 2814/7150 [37:15<59:14,  1.22it/s]  
 39%|███▉      | 2815/7150 [37:15<55:42,  1.30it/s]
 39%|███▉      | 2816/7150 [37:16<53:10,  1.36it/s]
 39%|███▉      | 2817/7150 [37:17<51:24,  1.40it/s]
 39%|███▉      | 2818/7150 [37:17<50:07,  1.44it/s]
 39%|███▉      | 2819/7150 [37:18<49:16,  1.46it/s]
 39%|███▉      | 2820/7150 [37:19<48:40,  1.48it/s]
                                                   
{'loss': '3.133', 'grad_norm': '0.2605', 'learning_rate': '0.0006', 'epoch': '19.72'}
+
 39%|███▉      | 2820/7150 [37:19<48:40,  1.48it/s]
 39%|███▉      | 2821/7150 [37:19<48:16,  1.49it/s]
 39%|███▉      | 2822/7150 [37:20<47:55,  1.51it/s]
 39%|███▉      | 2823/7150 [37:21<47:44,  1.51it/s]
 39%|███▉      | 2824/7150 [37:21<47:35,  1.52it/s]
 40%|███▉      | 2825/7150 [37:22<47:27,  1.52it/s]
 40%|███▉      | 2826/7150 [37:23<47:21,  1.52it/s]
 40%|███▉      | 2827/7150 [37:23<47:19,  1.52it/s]
 40%|███▉      | 2828/7150 [37:24<47:17,  1.52it/s]
 40%|███▉      | 2829/7150 [37:25<47:13,  1.52it/s]
 40%|███▉      | 2830/7150 [37:25<47:11,  1.53it/s]
                                                   
{'loss': '3.158', 'grad_norm': '0.2529', 'learning_rate': '0.0006', 'epoch': '19.79'}
+
 40%|███▉      | 2830/7150 [37:25<47:11,  1.53it/s]
 40%|███▉      | 2831/7150 [37:26<47:12,  1.52it/s]
 40%|███▉      | 2832/7150 [37:27<47:11,  1.52it/s]
 40%|███▉      | 2833/7150 [37:27<47:09,  1.53it/s]
 40%|███▉      | 2834/7150 [37:28<47:07,  1.53it/s]
 40%|███▉      | 2835/7150 [37:29<47:07,  1.53it/s]
 40%|███▉      | 2836/7150 [37:29<47:07,  1.53it/s]
 40%|███▉      | 2837/7150 [37:30<47:05,  1.53it/s]
 40%|███▉      | 2838/7150 [37:30<47:03,  1.53it/s]
 40%|███▉      | 2839/7150 [37:31<47:02,  1.53it/s]
 40%|███▉      | 2840/7150 [37:32<47:03,  1.53it/s]
                                                   
{'loss': '3.173', 'grad_norm': '0.2335', 'learning_rate': '0.0006', 'epoch': '19.86'}
+
 40%|███▉      | 2840/7150 [37:32<47:03,  1.53it/s]
 40%|███▉      | 2841/7150 [37:32<47:03,  1.53it/s]
 40%|███▉      | 2842/7150 [37:33<47:02,  1.53it/s]
 40%|███▉      | 2843/7150 [37:34<47:01,  1.53it/s]
 40%|███▉      | 2844/7150 [37:34<47:03,  1.53it/s]
 40%|███▉      | 2845/7150 [37:35<47:02,  1.53it/s]
 40%|███▉      | 2846/7150 [37:36<47:01,  1.53it/s]
 40%|███▉      | 2847/7150 [37:36<46:59,  1.53it/s]
 40%|███▉      | 2848/7150 [37:37<47:00,  1.53it/s]
 40%|███▉      | 2849/7150 [37:38<46:58,  1.53it/s]
 40%|███▉      | 2850/7150 [37:38<46:56,  1.53it/s]
                                                   
{'loss': '3.166', 'grad_norm': '0.2838', 'learning_rate': '0.0006', 'epoch': '19.93'}
+
 40%|███▉      | 2850/7150 [37:38<46:56,  1.53it/s]
 40%|███▉      | 2851/7150 [37:39<46:55,  1.53it/s]
 40%|███▉      | 2852/7150 [37:40<46:55,  1.53it/s]
 40%|███▉      | 2853/7150 [37:40<46:55,  1.53it/s]
 40%|███▉      | 2854/7150 [37:41<46:53,  1.53it/s]
 40%|███▉      | 2855/7150 [37:42<46:50,  1.53it/s]
 40%|███▉      | 2856/7150 [37:42<46:51,  1.53it/s]
 40%|███▉      | 2857/7150 [37:43<46:50,  1.53it/s]
 40%|███▉      | 2858/7150 [37:44<46:52,  1.53it/s]
 40%|███▉      | 2859/7150 [37:44<46:47,  1.53it/s]
 40%|████      | 2860/7150 [37:45<43:20,  1.65it/s]
                                                   
{'loss': '3.168', 'grad_norm': '0.2678', 'learning_rate': '0.0006', 'epoch': '20'}
+
 40%|████      | 2860/7150 [37:45<43:20,  1.65it/s]
 40%|████      | 2861/7150 [37:45<46:25,  1.54it/s]
 40%|████      | 2862/7150 [37:46<46:31,  1.54it/s]
 40%|████      | 2863/7150 [37:47<46:36,  1.53it/s]
 40%|████      | 2864/7150 [37:47<46:37,  1.53it/s]
 40%|████      | 2865/7150 [37:48<46:37,  1.53it/s]
 40%|████      | 2866/7150 [37:49<46:41,  1.53it/s]
 40%|████      | 2867/7150 [37:49<46:42,  1.53it/s]
 40%|████      | 2868/7150 [37:50<46:42,  1.53it/s]
 40%|████      | 2869/7150 [37:51<46:40,  1.53it/s]
 40%|████      | 2870/7150 [37:51<46:42,  1.53it/s]
                                                   
{'loss': '3.048', 'grad_norm': '0.3443', 'learning_rate': '0.0006', 'epoch': '20.07'}
+
 40%|████      | 2870/7150 [37:51<46:42,  1.53it/s]
 40%|████      | 2871/7150 [37:52<46:43,  1.53it/s]
 40%|████      | 2872/7150 [37:53<46:41,  1.53it/s]
 40%|████      | 2873/7150 [37:53<46:38,  1.53it/s]
 40%|████      | 2874/7150 [37:54<46:39,  1.53it/s]
 40%|████      | 2875/7150 [37:55<46:38,  1.53it/s]
 40%|████      | 2876/7150 [37:55<46:38,  1.53it/s]
 40%|████      | 2877/7150 [37:56<46:36,  1.53it/s]
 40%|████      | 2878/7150 [37:57<46:35,  1.53it/s]
 40%|████      | 2879/7150 [37:57<46:35,  1.53it/s]
 40%|████      | 2880/7150 [37:58<46:33,  1.53it/s]
                                                   
{'loss': '3.059', 'grad_norm': '0.3208', 'learning_rate': '0.0006', 'epoch': '20.14'}
+
 40%|████      | 2880/7150 [37:58<46:33,  1.53it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.67it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.82it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.63it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.06it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.76it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.57it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.45it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.37it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.32it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.28it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.28it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.26it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.25it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.25it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.24it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.24it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.24it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.24it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s]
                                                   
+
                                               
{'eval_loss': '3.778', 'eval_runtime': '3.248', 'eval_samples_per_second': '751.6', 'eval_steps_per_second': '12.01', 'epoch': '20.14'}
+
 40%|████      | 2880/7150 [38:01<46:33,  1.53it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s][eval] epoch 20.14 step 2880 loss 3.7779 (best 3.7377 @ ep 15.609, 9/20 since best)
+
+
                                               
 40%|████      | 2881/7150 [38:02<1:55:55,  1.63s/it]
 40%|████      | 2882/7150 [38:02<1:35:03,  1.34s/it]
 40%|████      | 2883/7150 [38:03<1:20:29,  1.13s/it]
 40%|████      | 2884/7150 [38:04<1:10:16,  1.01it/s]
 40%|████      | 2885/7150 [38:04<1:03:06,  1.13it/s]
 40%|████      | 2886/7150 [38:05<58:07,  1.22it/s]  
 40%|████      | 2887/7150 [38:06<54:38,  1.30it/s]
 40%|████      | 2888/7150 [38:06<52:10,  1.36it/s]
 40%|████      | 2889/7150 [38:07<50:27,  1.41it/s]
 40%|████      | 2890/7150 [38:08<49:16,  1.44it/s]
                                                   
{'loss': '3.061', 'grad_norm': '0.2706', 'learning_rate': '0.0006', 'epoch': '20.21'}
+
 40%|████      | 2890/7150 [38:08<49:16,  1.44it/s]
 40%|████      | 2891/7150 [38:08<48:26,  1.47it/s]
 40%|████      | 2892/7150 [38:09<47:49,  1.48it/s]
 40%|████      | 2893/7150 [38:10<47:24,  1.50it/s]
 40%|████      | 2894/7150 [38:10<47:06,  1.51it/s]
 40%|████      | 2895/7150 [38:11<46:53,  1.51it/s]
 41%|████      | 2896/7150 [38:12<46:42,  1.52it/s]
 41%|████      | 2897/7150 [38:12<46:37,  1.52it/s]
 41%|████      | 2898/7150 [38:13<46:32,  1.52it/s]
 41%|████      | 2899/7150 [38:14<46:28,  1.52it/s]
 41%|████      | 2900/7150 [38:14<46:25,  1.53it/s]
                                                   
{'loss': '3.077', 'grad_norm': '0.2686', 'learning_rate': '0.0006', 'epoch': '20.28'}
+
 41%|████      | 2900/7150 [38:14<46:25,  1.53it/s]
 41%|████      | 2901/7150 [38:15<46:25,  1.53it/s]
 41%|████      | 2902/7150 [38:16<46:24,  1.53it/s]
 41%|████      | 2903/7150 [38:16<46:22,  1.53it/s]
 41%|████      | 2904/7150 [38:17<46:19,  1.53it/s]
 41%|████      | 2905/7150 [38:18<46:19,  1.53it/s]
 41%|████      | 2906/7150 [38:18<46:17,  1.53it/s]
 41%|████      | 2907/7150 [38:19<46:16,  1.53it/s]
 41%|████      | 2908/7150 [38:19<46:16,  1.53it/s]
 41%|████      | 2909/7150 [38:20<46:16,  1.53it/s]
 41%|████      | 2910/7150 [38:21<46:15,  1.53it/s]
                                                   
{'loss': '3.086', 'grad_norm': '0.2477', 'learning_rate': '0.0006', 'epoch': '20.35'}
+
 41%|████      | 2910/7150 [38:21<46:15,  1.53it/s]
 41%|████      | 2911/7150 [38:21<46:13,  1.53it/s]
 41%|████      | 2912/7150 [38:22<46:14,  1.53it/s]
 41%|████      | 2913/7150 [38:23<46:14,  1.53it/s]
 41%|████      | 2914/7150 [38:23<46:13,  1.53it/s]
 41%|████      | 2915/7150 [38:24<46:11,  1.53it/s]
 41%|████      | 2916/7150 [38:25<46:11,  1.53it/s]
 41%|████      | 2917/7150 [38:25<46:10,  1.53it/s]
 41%|████      | 2918/7150 [38:26<46:08,  1.53it/s]
 41%|████      | 2919/7150 [38:27<46:08,  1.53it/s]
 41%|████      | 2920/7150 [38:27<46:08,  1.53it/s]
                                                   
{'loss': '3.11', 'grad_norm': '0.2735', 'learning_rate': '0.0006', 'epoch': '20.42'}
+
 41%|████      | 2920/7150 [38:27<46:08,  1.53it/s]
 41%|████      | 2921/7150 [38:28<46:09,  1.53it/s]
 41%|████      | 2922/7150 [38:29<46:07,  1.53it/s]
 41%|████      | 2923/7150 [38:29<46:06,  1.53it/s]
 41%|████      | 2924/7150 [38:30<46:05,  1.53it/s]
 41%|████      | 2925/7150 [38:31<46:03,  1.53it/s]
 41%|████      | 2926/7150 [38:31<46:05,  1.53it/s]
 41%|████      | 2927/7150 [38:32<46:05,  1.53it/s]
 41%|████      | 2928/7150 [38:33<46:04,  1.53it/s]
 41%|████      | 2929/7150 [38:33<46:04,  1.53it/s]
 41%|████      | 2930/7150 [38:34<46:08,  1.52it/s]
                                                   
{'loss': '3.108', 'grad_norm': '0.2977', 'learning_rate': '0.0006', 'epoch': '20.49'}
+
 41%|████      | 2930/7150 [38:34<46:08,  1.52it/s]
 41%|████      | 2931/7150 [38:35<46:08,  1.52it/s]
 41%|████      | 2932/7150 [38:35<46:05,  1.53it/s]
 41%|████      | 2933/7150 [38:36<46:04,  1.53it/s]
 41%|████      | 2934/7150 [38:37<45:59,  1.53it/s]
 41%|████      | 2935/7150 [38:37<45:59,  1.53it/s]
 41%|████      | 2936/7150 [38:38<45:57,  1.53it/s]
 41%|████      | 2937/7150 [38:38<45:56,  1.53it/s]
 41%|████      | 2938/7150 [38:39<45:57,  1.53it/s]
 41%|████      | 2939/7150 [38:40<45:57,  1.53it/s]
 41%|████      | 2940/7150 [38:40<45:56,  1.53it/s]
                                                   
{'loss': '3.115', 'grad_norm': '0.238', 'learning_rate': '0.0006', 'epoch': '20.56'}
+
 41%|████      | 2940/7150 [38:40<45:56,  1.53it/s]
 41%|████      | 2941/7150 [38:41<45:55,  1.53it/s]
 41%|████      | 2942/7150 [38:42<45:55,  1.53it/s]
 41%|████      | 2943/7150 [38:42<45:54,  1.53it/s]
 41%|████      | 2944/7150 [38:43<45:54,  1.53it/s]
 41%|████      | 2945/7150 [38:44<45:51,  1.53it/s]
 41%|████      | 2946/7150 [38:44<45:50,  1.53it/s]
 41%|████      | 2947/7150 [38:45<45:49,  1.53it/s]
 41%|████      | 2948/7150 [38:46<45:47,  1.53it/s]
 41%|████      | 2949/7150 [38:46<45:48,  1.53it/s]
 41%|████▏     | 2950/7150 [38:47<45:48,  1.53it/s]
                                                   
{'loss': '3.12', 'grad_norm': '0.2507', 'learning_rate': '0.0006', 'epoch': '20.63'}
+
 41%|████▏     | 2950/7150 [38:47<45:48,  1.53it/s]
 41%|████▏     | 2951/7150 [38:48<45:49,  1.53it/s]
 41%|████▏     | 2952/7150 [38:48<45:48,  1.53it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 19.17it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.99it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.71it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.10it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.78it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.59it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.46it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.38it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.32it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.28it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.27it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.26it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.26it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.26it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.25it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.25it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.24it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.23it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s]
                                                   
+
                                               
{'eval_loss': '3.759', 'eval_runtime': '3.262', 'eval_samples_per_second': '748.4', 'eval_steps_per_second': '11.96', 'epoch': '20.64'}
+
 41%|████▏     | 2952/7150 [38:52<45:48,  1.53it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s][eval] epoch 20.644 step 2952 loss 3.7593 (best 3.7377 @ ep 15.609, 10/20 since best)
+
+
                                               
 41%|████▏     | 2953/7150 [38:52<1:54:15,  1.63s/it]
 41%|████▏     | 2954/7150 [38:53<1:33:42,  1.34s/it]
 41%|████▏     | 2955/7150 [38:54<1:19:19,  1.13s/it]
 41%|████▏     | 2956/7150 [38:54<1:09:12,  1.01it/s]
 41%|████▏     | 2957/7150 [38:55<1:02:09,  1.12it/s]
 41%|████▏     | 2958/7150 [38:55<57:14,  1.22it/s]  
 41%|████▏     | 2959/7150 [38:56<53:45,  1.30it/s]
 41%|████▏     | 2960/7150 [38:57<51:19,  1.36it/s]
                                                   
{'loss': '3.135', 'grad_norm': '0.2344', 'learning_rate': '0.0006', 'epoch': '20.7'}
+
 41%|████▏     | 2960/7150 [38:57<51:19,  1.36it/s]
 41%|████▏     | 2961/7150 [38:57<49:38,  1.41it/s]
 41%|████▏     | 2962/7150 [38:58<48:25,  1.44it/s]
 41%|████▏     | 2963/7150 [38:59<47:35,  1.47it/s]
 41%|████▏     | 2964/7150 [38:59<47:01,  1.48it/s]
 41%|████▏     | 2965/7150 [39:00<46:37,  1.50it/s]
 41%|████▏     | 2966/7150 [39:01<46:18,  1.51it/s]
 41%|████▏     | 2967/7150 [39:01<46:04,  1.51it/s]
 42%|████▏     | 2968/7150 [39:02<45:56,  1.52it/s]
 42%|████▏     | 2969/7150 [39:03<45:51,  1.52it/s]
 42%|████▏     | 2970/7150 [39:03<45:45,  1.52it/s]
                                                   
{'loss': '3.137', 'grad_norm': '0.254', 'learning_rate': '0.0006', 'epoch': '20.77'}
+
 42%|████▏     | 2970/7150 [39:03<45:45,  1.52it/s]
 42%|████▏     | 2971/7150 [39:04<45:41,  1.52it/s]
 42%|████▏     | 2972/7150 [39:05<45:40,  1.52it/s]
 42%|████▏     | 2973/7150 [39:05<45:38,  1.53it/s]
 42%|████▏     | 2974/7150 [39:06<45:37,  1.53it/s]
 42%|████▏     | 2975/7150 [39:07<45:33,  1.53it/s]
 42%|████▏     | 2976/7150 [39:07<45:33,  1.53it/s]
 42%|████▏     | 2977/7150 [39:08<45:31,  1.53it/s]
 42%|████▏     | 2978/7150 [39:09<45:29,  1.53it/s]
 42%|████▏     | 2979/7150 [39:09<45:31,  1.53it/s]
 42%|████▏     | 2980/7150 [39:10<45:31,  1.53it/s]
                                                   
{'loss': '3.143', 'grad_norm': '0.248', 'learning_rate': '0.0006', 'epoch': '20.84'}
+
 42%|████▏     | 2980/7150 [39:10<45:31,  1.53it/s]
 42%|████▏     | 2981/7150 [39:11<45:30,  1.53it/s]
 42%|████▏     | 2982/7150 [39:11<45:27,  1.53it/s]
 42%|████▏     | 2983/7150 [39:12<45:28,  1.53it/s]
 42%|████▏     | 2984/7150 [39:12<45:26,  1.53it/s]
 42%|████▏     | 2985/7150 [39:13<45:26,  1.53it/s]
 42%|████▏     | 2986/7150 [39:14<45:24,  1.53it/s]
 42%|████▏     | 2987/7150 [39:14<45:24,  1.53it/s]
 42%|████▏     | 2988/7150 [39:15<45:25,  1.53it/s]
 42%|████▏     | 2989/7150 [39:16<45:23,  1.53it/s]
 42%|████▏     | 2990/7150 [39:16<45:22,  1.53it/s]
                                                   
{'loss': '3.145', 'grad_norm': '0.2525', 'learning_rate': '0.0006', 'epoch': '20.91'}
+
 42%|████▏     | 2990/7150 [39:16<45:22,  1.53it/s]
 42%|████▏     | 2991/7150 [39:17<45:22,  1.53it/s]
 42%|████▏     | 2992/7150 [39:18<45:21,  1.53it/s]
 42%|████▏     | 2993/7150 [39:18<45:18,  1.53it/s]
 42%|████▏     | 2994/7150 [39:19<45:20,  1.53it/s]
 42%|████▏     | 2995/7150 [39:20<45:20,  1.53it/s]
 42%|████▏     | 2996/7150 [39:20<45:20,  1.53it/s]
 42%|████▏     | 2997/7150 [39:21<45:16,  1.53it/s]
 42%|████▏     | 2998/7150 [39:22<45:17,  1.53it/s]
 42%|████▏     | 2999/7150 [39:22<45:16,  1.53it/s]
 42%|████▏     | 3000/7150 [39:23<45:14,  1.53it/s]
                                                   
{'loss': '3.141', 'grad_norm': '0.2385', 'learning_rate': '0.0006', 'epoch': '20.98'}
+
 42%|████▏     | 3000/7150 [39:23<45:14,  1.53it/s]
 42%|████▏     | 3001/7150 [39:24<45:18,  1.53it/s]
 42%|████▏     | 3002/7150 [39:24<45:16,  1.53it/s]
 42%|████▏     | 3003/7150 [39:25<41:54,  1.65it/s]
 42%|████▏     | 3004/7150 [39:26<44:53,  1.54it/s]
 42%|████▏     | 3005/7150 [39:26<44:59,  1.54it/s]
 42%|████▏     | 3006/7150 [39:27<45:02,  1.53it/s]
 42%|████▏     | 3007/7150 [39:27<45:03,  1.53it/s]
 42%|████▏     | 3008/7150 [39:28<45:05,  1.53it/s]
 42%|████▏     | 3009/7150 [39:29<45:07,  1.53it/s]
 42%|████▏     | 3010/7150 [39:29<45:08,  1.53it/s]
                                                   
{'loss': '3.052', 'grad_norm': '0.4159', 'learning_rate': '0.0006', 'epoch': '21.05'}
+
 42%|████▏     | 3010/7150 [39:29<45:08,  1.53it/s]
 42%|████▏     | 3011/7150 [39:30<45:09,  1.53it/s]
 42%|████▏     | 3012/7150 [39:31<45:09,  1.53it/s]
 42%|████▏     | 3013/7150 [39:31<45:08,  1.53it/s]
 42%|████▏     | 3014/7150 [39:32<45:09,  1.53it/s]
 42%|████▏     | 3015/7150 [39:33<45:05,  1.53it/s]
 42%|████▏     | 3016/7150 [39:33<45:06,  1.53it/s]
 42%|████▏     | 3017/7150 [39:34<45:06,  1.53it/s]
 42%|████▏     | 3018/7150 [39:35<45:04,  1.53it/s]
 42%|████▏     | 3019/7150 [39:35<45:03,  1.53it/s]
 42%|████▏     | 3020/7150 [39:36<45:03,  1.53it/s]
                                                   
{'loss': '3.02', 'grad_norm': '0.4319', 'learning_rate': '0.0006', 'epoch': '21.12'}
+
 42%|████▏     | 3020/7150 [39:36<45:03,  1.53it/s]
 42%|████▏     | 3021/7150 [39:37<45:02,  1.53it/s]
 42%|████▏     | 3022/7150 [39:37<45:00,  1.53it/s]
 42%|████▏     | 3023/7150 [39:38<45:01,  1.53it/s]
 42%|████▏     | 3024/7150 [39:39<45:02,  1.53it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.47it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.78it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.60it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.06it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.74it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.57it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.45it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.37it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.32it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.29it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.29it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.28it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.26it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.25it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.25it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.25it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.25it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.25it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s]
                                                   
+
                                               
{'eval_loss': '3.796', 'eval_runtime': '3.26', 'eval_samples_per_second': '748.8', 'eval_steps_per_second': '11.96', 'epoch': '21.15'}
+
 42%|████▏     | 3024/7150 [39:42<45:02,  1.53it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s][eval] epoch 21.147 step 3024 loss 3.7956 (best 3.7377 @ ep 15.609, 11/20 since best)
+
+
                                               
 42%|████▏     | 3025/7150 [39:43<1:52:14,  1.63s/it]
 42%|████▏     | 3026/7150 [39:43<1:32:01,  1.34s/it]
 42%|████▏     | 3027/7150 [39:44<1:17:53,  1.13s/it]
 42%|████▏     | 3028/7150 [39:44<1:07:59,  1.01it/s]
 42%|████▏     | 3029/7150 [39:45<1:01:03,  1.13it/s]
 42%|████▏     | 3030/7150 [39:46<56:13,  1.22it/s]  
                                                   
{'loss': '3.054', 'grad_norm': '0.3718', 'learning_rate': '0.0006', 'epoch': '21.19'}
+
 42%|████▏     | 3030/7150 [39:46<56:13,  1.22it/s]
 42%|████▏     | 3031/7150 [39:46<52:50,  1.30it/s]
 42%|████▏     | 3032/7150 [39:47<50:28,  1.36it/s]
 42%|████▏     | 3033/7150 [39:48<48:47,  1.41it/s]
 42%|████▏     | 3034/7150 [39:48<47:38,  1.44it/s]
 42%|████▏     | 3035/7150 [39:49<46:49,  1.46it/s]
 42%|████▏     | 3036/7150 [39:50<46:13,  1.48it/s]
 42%|████▏     | 3037/7150 [39:50<45:49,  1.50it/s]
 42%|████▏     | 3038/7150 [39:51<45:32,  1.50it/s]
 43%|████▎     | 3039/7150 [39:52<45:20,  1.51it/s]
 43%|████▎     | 3040/7150 [39:52<45:11,  1.52it/s]
                                                   
{'loss': '3.057', 'grad_norm': '0.2839', 'learning_rate': '0.0006', 'epoch': '21.26'}
+
 43%|████▎     | 3040/7150 [39:52<45:11,  1.52it/s]
 43%|████▎     | 3041/7150 [39:53<45:04,  1.52it/s]
 43%|████▎     | 3042/7150 [39:54<45:00,  1.52it/s]
 43%|████▎     | 3043/7150 [39:54<44:57,  1.52it/s]
 43%|████▎     | 3044/7150 [39:55<44:53,  1.52it/s]
 43%|████▎     | 3045/7150 [39:56<44:49,  1.53it/s]
 43%|████▎     | 3046/7150 [39:56<44:48,  1.53it/s]
 43%|████▎     | 3047/7150 [39:57<44:46,  1.53it/s]
 43%|████▎     | 3048/7150 [39:58<44:46,  1.53it/s]
 43%|████▎     | 3049/7150 [39:58<44:44,  1.53it/s]
 43%|████▎     | 3050/7150 [39:59<44:45,  1.53it/s]
                                                   
{'loss': '3.07', 'grad_norm': '0.2609', 'learning_rate': '0.0006', 'epoch': '21.33'}
+
 43%|████▎     | 3050/7150 [39:59<44:45,  1.53it/s]
 43%|████▎     | 3051/7150 [40:00<44:47,  1.53it/s]
 43%|████▎     | 3052/7150 [40:00<44:46,  1.53it/s]
 43%|████▎     | 3053/7150 [40:01<44:43,  1.53it/s]
 43%|████▎     | 3054/7150 [40:02<44:43,  1.53it/s]
 43%|████▎     | 3055/7150 [40:02<44:41,  1.53it/s]
 43%|████▎     | 3056/7150 [40:03<44:41,  1.53it/s]
 43%|████▎     | 3057/7150 [40:03<44:40,  1.53it/s]
 43%|████▎     | 3058/7150 [40:04<44:39,  1.53it/s]
 43%|████▎     | 3059/7150 [40:05<44:38,  1.53it/s]
 43%|████▎     | 3060/7150 [40:05<44:38,  1.53it/s]
                                                   
{'loss': '3.074', 'grad_norm': '0.2601', 'learning_rate': '0.0006', 'epoch': '21.4'}
+
 43%|████▎     | 3060/7150 [40:05<44:38,  1.53it/s]
 43%|████▎     | 3061/7150 [40:06<44:37,  1.53it/s]
 43%|████▎     | 3062/7150 [40:07<44:36,  1.53it/s]
 43%|████▎     | 3063/7150 [40:07<44:35,  1.53it/s]
 43%|████▎     | 3064/7150 [40:08<44:34,  1.53it/s]
 43%|████▎     | 3065/7150 [40:09<44:34,  1.53it/s]
 43%|████▎     | 3066/7150 [40:09<44:34,  1.53it/s]
 43%|████▎     | 3067/7150 [40:10<44:33,  1.53it/s]
 43%|████▎     | 3068/7150 [40:11<44:31,  1.53it/s]
 43%|████▎     | 3069/7150 [40:11<44:31,  1.53it/s]
 43%|████▎     | 3070/7150 [40:12<44:30,  1.53it/s]
                                                   
{'loss': '3.084', 'grad_norm': '0.2554', 'learning_rate': '0.0006', 'epoch': '21.47'}
+
 43%|████▎     | 3070/7150 [40:12<44:30,  1.53it/s]
 43%|████▎     | 3071/7150 [40:13<44:31,  1.53it/s]
 43%|████▎     | 3072/7150 [40:13<44:28,  1.53it/s]
 43%|████▎     | 3073/7150 [40:14<44:29,  1.53it/s]
 43%|████▎     | 3074/7150 [40:15<44:28,  1.53it/s]
 43%|████▎     | 3075/7150 [40:15<44:26,  1.53it/s]
 43%|████▎     | 3076/7150 [40:16<44:26,  1.53it/s]
 43%|████▎     | 3077/7150 [40:17<44:26,  1.53it/s]
 43%|████▎     | 3078/7150 [40:17<44:25,  1.53it/s]
 43%|████▎     | 3079/7150 [40:18<44:24,  1.53it/s]
 43%|████▎     | 3080/7150 [40:19<44:24,  1.53it/s]
                                                   
{'loss': '3.097', 'grad_norm': '0.2647', 'learning_rate': '0.0006', 'epoch': '21.54'}
+
 43%|████▎     | 3080/7150 [40:19<44:24,  1.53it/s]
 43%|████▎     | 3081/7150 [40:19<44:25,  1.53it/s]
 43%|████▎     | 3082/7150 [40:20<44:24,  1.53it/s]
 43%|████▎     | 3083/7150 [40:21<44:22,  1.53it/s]
 43%|████▎     | 3084/7150 [40:21<44:22,  1.53it/s]
 43%|████▎     | 3085/7150 [40:22<44:21,  1.53it/s]
 43%|████▎     | 3086/7150 [40:22<44:22,  1.53it/s]
 43%|████▎     | 3087/7150 [40:23<44:18,  1.53it/s]
 43%|████▎     | 3088/7150 [40:24<44:18,  1.53it/s]
 43%|████▎     | 3089/7150 [40:24<44:17,  1.53it/s]
 43%|████▎     | 3090/7150 [40:25<44:15,  1.53it/s]
                                                   
{'loss': '3.108', 'grad_norm': '0.2529', 'learning_rate': '0.0006', 'epoch': '21.61'}
+
 43%|████▎     | 3090/7150 [40:25<44:15,  1.53it/s]
 43%|████▎     | 3091/7150 [40:26<44:18,  1.53it/s]
 43%|████▎     | 3092/7150 [40:26<44:17,  1.53it/s]
 43%|████▎     | 3093/7150 [40:27<44:16,  1.53it/s]
 43%|████▎     | 3094/7150 [40:28<44:14,  1.53it/s]
 43%|████▎     | 3095/7150 [40:28<44:14,  1.53it/s]
 43%|████▎     | 3096/7150 [40:29<44:12,  1.53it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.61it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.81it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.62it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.06it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.77it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.58it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.47it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.39it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.34it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.29it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.30it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.26it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.26it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.26it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.25it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.25it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.25it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.24it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s]
                                                   
+
                                               
{'eval_loss': '3.769', 'eval_runtime': '3.244', 'eval_samples_per_second': '752.3', 'eval_steps_per_second': '12.02', 'epoch': '21.65'}
+
 43%|████▎     | 3096/7150 [40:32<44:12,  1.53it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s][eval] epoch 21.651 step 3096 loss 3.7693 (best 3.7377 @ ep 15.609, 12/20 since best)
+
+
                                               
 43%|████▎     | 3097/7150 [40:33<1:49:57,  1.63s/it]
 43%|████▎     | 3098/7150 [40:34<1:30:11,  1.34s/it]
 43%|████▎     | 3099/7150 [40:34<1:16:22,  1.13s/it]
 43%|████▎     | 3100/7150 [40:35<1:06:42,  1.01it/s]
                                                     
{'loss': '3.106', 'grad_norm': '0.2555', 'learning_rate': '0.0006', 'epoch': '21.68'}
+
 43%|████▎     | 3100/7150 [40:35<1:06:42,  1.01it/s]
 43%|████▎     | 3101/7150 [40:36<59:57,  1.13it/s]  
 43%|████▎     | 3102/7150 [40:36<55:11,  1.22it/s]
 43%|████▎     | 3103/7150 [40:37<51:52,  1.30it/s]
 43%|████▎     | 3104/7150 [40:37<49:32,  1.36it/s]
 43%|████▎     | 3105/7150 [40:38<47:56,  1.41it/s]
 43%|████▎     | 3106/7150 [40:39<46:45,  1.44it/s]
 43%|████▎     | 3107/7150 [40:39<45:57,  1.47it/s]
 43%|████▎     | 3108/7150 [40:40<45:23,  1.48it/s]
 43%|████▎     | 3109/7150 [40:41<44:58,  1.50it/s]
 43%|████▎     | 3110/7150 [40:41<44:43,  1.51it/s]
                                                   
{'loss': '3.121', 'grad_norm': '0.2463', 'learning_rate': '0.0006', 'epoch': '21.75'}
+
 43%|████▎     | 3110/7150 [40:41<44:43,  1.51it/s]
 44%|████▎     | 3111/7150 [40:42<44:32,  1.51it/s]
 44%|████▎     | 3112/7150 [40:43<44:22,  1.52it/s]
 44%|████▎     | 3113/7150 [40:43<44:15,  1.52it/s]
 44%|████▎     | 3114/7150 [40:44<44:11,  1.52it/s]
 44%|████▎     | 3115/7150 [40:45<44:08,  1.52it/s]
 44%|████▎     | 3116/7150 [40:45<44:05,  1.52it/s]
 44%|████▎     | 3117/7150 [40:46<44:02,  1.53it/s]
 44%|████▎     | 3118/7150 [40:47<44:01,  1.53it/s]
 44%|████▎     | 3119/7150 [40:47<43:59,  1.53it/s]
 44%|████▎     | 3120/7150 [40:48<44:00,  1.53it/s]
                                                   
{'loss': '3.11', 'grad_norm': '0.2386', 'learning_rate': '0.0006', 'epoch': '21.82'}
+
 44%|████▎     | 3120/7150 [40:48<44:00,  1.53it/s]
 44%|████▎     | 3121/7150 [40:49<43:57,  1.53it/s]
 44%|████▎     | 3122/7150 [40:49<43:57,  1.53it/s]
 44%|████▎     | 3123/7150 [40:50<43:57,  1.53it/s]
 44%|████▎     | 3124/7150 [40:51<43:55,  1.53it/s]
 44%|████▎     | 3125/7150 [40:51<43:54,  1.53it/s]
 44%|████▎     | 3126/7150 [40:52<43:55,  1.53it/s]
 44%|████▎     | 3127/7150 [40:53<43:54,  1.53it/s]
 44%|████▎     | 3128/7150 [40:53<43:53,  1.53it/s]
 44%|████▍     | 3129/7150 [40:54<43:53,  1.53it/s]
 44%|████▍     | 3130/7150 [40:55<43:53,  1.53it/s]
                                                   
{'loss': '3.125', 'grad_norm': '0.2564', 'learning_rate': '0.0006', 'epoch': '21.89'}
+
 44%|████▍     | 3130/7150 [40:55<43:53,  1.53it/s]
 44%|████▍     | 3131/7150 [40:55<43:53,  1.53it/s]
 44%|████▍     | 3132/7150 [40:56<43:51,  1.53it/s]
 44%|████▍     | 3133/7150 [40:56<43:50,  1.53it/s]
 44%|████▍     | 3134/7150 [40:57<43:49,  1.53it/s]
 44%|████▍     | 3135/7150 [40:58<43:48,  1.53it/s]
 44%|████▍     | 3136/7150 [40:58<43:47,  1.53it/s]
 44%|████▍     | 3137/7150 [40:59<43:49,  1.53it/s]
 44%|████▍     | 3138/7150 [41:00<43:48,  1.53it/s]
 44%|████▍     | 3139/7150 [41:00<43:46,  1.53it/s]
 44%|████▍     | 3140/7150 [41:01<43:44,  1.53it/s]
                                                   
{'loss': '3.122', 'grad_norm': '0.2575', 'learning_rate': '0.0006', 'epoch': '21.96'}
+
 44%|████▍     | 3140/7150 [41:01<43:44,  1.53it/s]
 44%|████▍     | 3141/7150 [41:02<43:46,  1.53it/s]
 44%|████▍     | 3142/7150 [41:02<43:46,  1.53it/s]
 44%|████▍     | 3143/7150 [41:03<43:44,  1.53it/s]
 44%|████▍     | 3144/7150 [41:04<43:47,  1.52it/s]
 44%|████▍     | 3145/7150 [41:04<43:43,  1.53it/s]
 44%|████▍     | 3146/7150 [41:05<40:27,  1.65it/s]
 44%|████▍     | 3147/7150 [41:06<43:19,  1.54it/s]
 44%|████▍     | 3148/7150 [41:06<43:26,  1.54it/s]
 44%|████▍     | 3149/7150 [41:07<43:30,  1.53it/s]
 44%|████▍     | 3150/7150 [41:08<43:31,  1.53it/s]
                                                   
{'loss': '3.073', 'grad_norm': '0.4374', 'learning_rate': '0.0006', 'epoch': '22.03'}
+
 44%|████▍     | 3150/7150 [41:08<43:31,  1.53it/s]
 44%|████▍     | 3151/7150 [41:08<43:32,  1.53it/s]
 44%|████▍     | 3152/7150 [41:09<43:33,  1.53it/s]
 44%|████▍     | 3153/7150 [41:10<43:33,  1.53it/s]
 44%|████▍     | 3154/7150 [41:10<43:31,  1.53it/s]
 44%|████▍     | 3155/7150 [41:11<43:32,  1.53it/s]
 44%|████▍     | 3156/7150 [41:11<43:33,  1.53it/s]
 44%|████▍     | 3157/7150 [41:12<43:34,  1.53it/s]
 44%|████▍     | 3158/7150 [41:13<43:33,  1.53it/s]
 44%|████▍     | 3159/7150 [41:13<43:34,  1.53it/s]
 44%|████▍     | 3160/7150 [41:14<43:34,  1.53it/s]
                                                   
{'loss': '3.007', 'grad_norm': '0.435', 'learning_rate': '0.0006', 'epoch': '22.1'}
+
 44%|████▍     | 3160/7150 [41:14<43:34,  1.53it/s]
 44%|████▍     | 3161/7150 [41:15<43:37,  1.52it/s]
 44%|████▍     | 3162/7150 [41:15<43:35,  1.52it/s]
 44%|████▍     | 3163/7150 [41:16<43:32,  1.53it/s]
 44%|████▍     | 3164/7150 [41:17<43:33,  1.53it/s]
 44%|████▍     | 3165/7150 [41:17<43:34,  1.52it/s]
 44%|████▍     | 3166/7150 [41:18<43:32,  1.53it/s]
 44%|████▍     | 3167/7150 [41:19<43:30,  1.53it/s]
 44%|████▍     | 3168/7150 [41:19<43:28,  1.53it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.45it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.75it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.60it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.05it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.75it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.57it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.46it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.38it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.33it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.30it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.29it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.27it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.26it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.25it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.25it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.25it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.24it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.23it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s]
                                                   
+
                                               
{'eval_loss': '3.788', 'eval_runtime': '3.26', 'eval_samples_per_second': '748.8', 'eval_steps_per_second': '11.96', 'epoch': '22.15'}
+
 44%|████▍     | 3168/7150 [41:23<43:28,  1.53it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s][eval] epoch 22.154 step 3168 loss 3.7882 (best 3.7377 @ ep 15.609, 13/20 since best)
+
+
                                               
 44%|████▍     | 3169/7150 [41:23<1:48:21,  1.63s/it]
 44%|████▍     | 3170/7150 [41:24<1:28:50,  1.34s/it]
                                                     
{'loss': '3.015', 'grad_norm': '0.3402', 'learning_rate': '0.0006', 'epoch': '22.17'}
+
 44%|████▍     | 3170/7150 [41:24<1:28:50,  1.34s/it]
 44%|████▍     | 3171/7150 [41:25<1:15:12,  1.13s/it]
 44%|████▍     | 3172/7150 [41:25<1:05:40,  1.01it/s]
 44%|████▍     | 3173/7150 [41:26<58:58,  1.12it/s]  
 44%|████▍     | 3174/7150 [41:27<54:17,  1.22it/s]
 44%|████▍     | 3175/7150 [41:27<50:58,  1.30it/s]
 44%|████▍     | 3176/7150 [41:28<48:42,  1.36it/s]
 44%|████▍     | 3177/7150 [41:28<47:05,  1.41it/s]
 44%|████▍     | 3178/7150 [41:29<45:57,  1.44it/s]
 44%|████▍     | 3179/7150 [41:30<45:08,  1.47it/s]
 44%|████▍     | 3180/7150 [41:30<44:34,  1.48it/s]
                                                   
{'loss': '3.025', 'grad_norm': '0.2635', 'learning_rate': '0.0006', 'epoch': '22.24'}
+
 44%|████▍     | 3180/7150 [41:30<44:34,  1.48it/s]
 44%|████▍     | 3181/7150 [41:31<44:14,  1.50it/s]
 45%|████▍     | 3182/7150 [41:32<43:57,  1.50it/s]
 45%|████▍     | 3183/7150 [41:32<43:43,  1.51it/s]
 45%|████▍     | 3184/7150 [41:33<43:36,  1.52it/s]
 45%|████▍     | 3185/7150 [41:34<43:28,  1.52it/s]
 45%|████▍     | 3186/7150 [41:34<43:23,  1.52it/s]
 45%|████▍     | 3187/7150 [41:35<43:20,  1.52it/s]
 45%|████▍     | 3188/7150 [41:36<43:18,  1.52it/s]
 45%|████▍     | 3189/7150 [41:36<43:16,  1.53it/s]
 45%|████▍     | 3190/7150 [41:37<43:14,  1.53it/s]
                                                   
{'loss': '3.046', 'grad_norm': '0.26', 'learning_rate': '0.0006', 'epoch': '22.31'}
+
 45%|████▍     | 3190/7150 [41:37<43:14,  1.53it/s]
 45%|████▍     | 3191/7150 [41:38<43:14,  1.53it/s]
 45%|████▍     | 3192/7150 [41:38<43:13,  1.53it/s]
 45%|████▍     | 3193/7150 [41:39<43:12,  1.53it/s]
 45%|████▍     | 3194/7150 [41:40<43:09,  1.53it/s]
 45%|████▍     | 3195/7150 [41:40<43:10,  1.53it/s]
 45%|████▍     | 3196/7150 [41:41<43:08,  1.53it/s]
 45%|████▍     | 3197/7150 [41:42<43:08,  1.53it/s]
 45%|████▍     | 3198/7150 [41:42<43:07,  1.53it/s]
 45%|████▍     | 3199/7150 [41:43<43:08,  1.53it/s]
 45%|████▍     | 3200/7150 [41:44<43:06,  1.53it/s]
                                                   
{'loss': '3.052', 'grad_norm': '0.27', 'learning_rate': '0.0006', 'epoch': '22.38'}
+
 45%|████▍     | 3200/7150 [41:44<43:06,  1.53it/s]
 45%|████▍     | 3201/7150 [41:44<43:07,  1.53it/s]
 45%|████▍     | 3202/7150 [41:45<43:04,  1.53it/s]
 45%|████▍     | 3203/7150 [41:46<43:06,  1.53it/s]
 45%|████▍     | 3204/7150 [41:46<43:04,  1.53it/s]
 45%|████▍     | 3205/7150 [41:47<43:03,  1.53it/s]
 45%|████▍     | 3206/7150 [41:47<43:00,  1.53it/s]
 45%|████▍     | 3207/7150 [41:48<43:00,  1.53it/s]
 45%|████▍     | 3208/7150 [41:49<43:00,  1.53it/s]
 45%|████▍     | 3209/7150 [41:49<42:58,  1.53it/s]
 45%|████▍     | 3210/7150 [41:50<42:59,  1.53it/s]
                                                   
{'loss': '3.065', 'grad_norm': '0.2562', 'learning_rate': '0.0006', 'epoch': '22.45'}
+
 45%|████▍     | 3210/7150 [41:50<42:59,  1.53it/s]
 45%|████▍     | 3211/7150 [41:51<43:00,  1.53it/s]
 45%|████▍     | 3212/7150 [41:51<42:58,  1.53it/s]
 45%|████▍     | 3213/7150 [41:52<42:56,  1.53it/s]
 45%|████▍     | 3214/7150 [41:53<42:57,  1.53it/s]
 45%|████▍     | 3215/7150 [41:53<42:57,  1.53it/s]
 45%|████▍     | 3216/7150 [41:54<42:56,  1.53it/s]
 45%|████▍     | 3217/7150 [41:55<42:53,  1.53it/s]
 45%|████▌     | 3218/7150 [41:55<42:53,  1.53it/s]
 45%|████▌     | 3219/7150 [41:56<42:51,  1.53it/s]
 45%|████▌     | 3220/7150 [41:57<42:50,  1.53it/s]
                                                   
{'loss': '3.076', 'grad_norm': '0.2518', 'learning_rate': '0.0006', 'epoch': '22.52'}
+
 45%|████▌     | 3220/7150 [41:57<42:50,  1.53it/s]
 45%|████▌     | 3221/7150 [41:57<42:51,  1.53it/s]
 45%|████▌     | 3222/7150 [41:58<42:52,  1.53it/s]
 45%|████▌     | 3223/7150 [41:59<42:51,  1.53it/s]
 45%|████▌     | 3224/7150 [41:59<42:50,  1.53it/s]
 45%|████▌     | 3225/7150 [42:00<42:50,  1.53it/s]
 45%|████▌     | 3226/7150 [42:01<42:49,  1.53it/s]
 45%|████▌     | 3227/7150 [42:01<42:48,  1.53it/s]
 45%|████▌     | 3228/7150 [42:02<42:46,  1.53it/s]
 45%|████▌     | 3229/7150 [42:03<42:48,  1.53it/s]
 45%|████▌     | 3230/7150 [42:03<42:49,  1.53it/s]
                                                   
{'loss': '3.079', 'grad_norm': '0.2752', 'learning_rate': '0.0006', 'epoch': '22.59'}
+
 45%|████▌     | 3230/7150 [42:03<42:49,  1.53it/s]
 45%|████▌     | 3231/7150 [42:04<42:48,  1.53it/s]
 45%|████▌     | 3232/7150 [42:05<42:46,  1.53it/s]
 45%|████▌     | 3233/7150 [42:05<42:46,  1.53it/s]
 45%|████▌     | 3234/7150 [42:06<42:45,  1.53it/s]
 45%|████▌     | 3235/7150 [42:06<42:44,  1.53it/s]
 45%|████▌     | 3236/7150 [42:07<42:42,  1.53it/s]
 45%|████▌     | 3237/7150 [42:08<42:42,  1.53it/s]
 45%|████▌     | 3238/7150 [42:08<42:42,  1.53it/s]
 45%|████▌     | 3239/7150 [42:09<42:40,  1.53it/s]
 45%|████▌     | 3240/7150 [42:10<42:38,  1.53it/s]
                                                   
{'loss': '3.079', 'grad_norm': '0.2628', 'learning_rate': '0.0006', 'epoch': '22.66'}
+
 45%|████▌     | 3240/7150 [42:10<42:38,  1.53it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.81it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.87it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.65it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.10it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.78it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.59it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.45it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.38it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.31it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.30it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.28it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.28it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.26it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.25it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.25it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.25it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.24it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.24it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s]
                                                   
+
                                               
{'eval_loss': '3.772', 'eval_runtime': '3.245', 'eval_samples_per_second': '752.1', 'eval_steps_per_second': '12.02', 'epoch': '22.66'}
+
 45%|████▌     | 3240/7150 [42:13<42:38,  1.53it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s][eval] epoch 22.658 step 3240 loss 3.7721 (best 3.7377 @ ep 15.609, 14/20 since best)
+
+
                                               
 45%|████▌     | 3241/7150 [42:14<1:46:07,  1.63s/it]
 45%|████▌     | 3242/7150 [42:14<1:27:03,  1.34s/it]
 45%|████▌     | 3243/7150 [42:15<1:13:43,  1.13s/it]
 45%|████▌     | 3244/7150 [42:16<1:04:22,  1.01it/s]
 45%|████▌     | 3245/7150 [42:16<57:49,  1.13it/s]  
 45%|████▌     | 3246/7150 [42:17<53:15,  1.22it/s]
 45%|████▌     | 3247/7150 [42:18<50:03,  1.30it/s]
 45%|████▌     | 3248/7150 [42:18<47:48,  1.36it/s]
 45%|████▌     | 3249/7150 [42:19<46:13,  1.41it/s]
 45%|████▌     | 3250/7150 [42:20<45:07,  1.44it/s]
                                                   
{'loss': '3.092', 'grad_norm': '0.2482', 'learning_rate': '0.0006', 'epoch': '22.73'}
+
 45%|████▌     | 3250/7150 [42:20<45:07,  1.44it/s]
 45%|████▌     | 3251/7150 [42:20<44:22,  1.46it/s]
 45%|████▌     | 3252/7150 [42:21<43:48,  1.48it/s]
 45%|████▌     | 3253/7150 [42:22<43:24,  1.50it/s]
 46%|████▌     | 3254/7150 [42:22<43:08,  1.50it/s]
 46%|████▌     | 3255/7150 [42:23<43:01,  1.51it/s]
 46%|████▌     | 3256/7150 [42:23<42:51,  1.51it/s]
 46%|████▌     | 3257/7150 [42:24<42:44,  1.52it/s]
 46%|████▌     | 3258/7150 [42:25<42:36,  1.52it/s]
 46%|████▌     | 3259/7150 [42:25<42:34,  1.52it/s]
 46%|████▌     | 3260/7150 [42:26<42:32,  1.52it/s]
                                                   
{'loss': '3.09', 'grad_norm': '0.2479', 'learning_rate': '0.0006', 'epoch': '22.8'}
+
 46%|████▌     | 3260/7150 [42:26<42:32,  1.52it/s]
 46%|████▌     | 3261/7150 [42:27<42:31,  1.52it/s]
 46%|████▌     | 3262/7150 [42:27<42:27,  1.53it/s]
 46%|████▌     | 3263/7150 [42:28<42:27,  1.53it/s]
 46%|████▌     | 3264/7150 [42:29<42:26,  1.53it/s]
 46%|████▌     | 3265/7150 [42:29<42:24,  1.53it/s]
 46%|████▌     | 3266/7150 [42:30<42:22,  1.53it/s]
 46%|████▌     | 3267/7150 [42:31<42:22,  1.53it/s]
 46%|████▌     | 3268/7150 [42:31<42:20,  1.53it/s]
 46%|████▌     | 3269/7150 [42:32<42:19,  1.53it/s]
 46%|████▌     | 3270/7150 [42:33<42:20,  1.53it/s]
                                                   
{'loss': '3.09', 'grad_norm': '0.2549', 'learning_rate': '0.0006', 'epoch': '22.87'}
+
 46%|████▌     | 3270/7150 [42:33<42:20,  1.53it/s]
 46%|████▌     | 3271/7150 [42:33<42:20,  1.53it/s]
 46%|████▌     | 3272/7150 [42:34<42:20,  1.53it/s]
 46%|████▌     | 3273/7150 [42:35<42:18,  1.53it/s]
 46%|████▌     | 3274/7150 [42:35<42:19,  1.53it/s]
 46%|████▌     | 3275/7150 [42:36<42:18,  1.53it/s]
 46%|████▌     | 3276/7150 [42:37<42:18,  1.53it/s]
 46%|████▌     | 3277/7150 [42:37<42:15,  1.53it/s]
 46%|████▌     | 3278/7150 [42:38<42:16,  1.53it/s]
 46%|████▌     | 3279/7150 [42:39<42:16,  1.53it/s]
 46%|████▌     | 3280/7150 [42:39<42:16,  1.53it/s]
                                                   
{'loss': '3.119', 'grad_norm': '0.2571', 'learning_rate': '0.0006', 'epoch': '22.94'}
+
 46%|████▌     | 3280/7150 [42:39<42:16,  1.53it/s]
 46%|████▌     | 3281/7150 [42:40<42:16,  1.53it/s]
 46%|████▌     | 3282/7150 [42:41<42:13,  1.53it/s]
 46%|████▌     | 3283/7150 [42:41<42:12,  1.53it/s]
 46%|████▌     | 3284/7150 [42:42<42:11,  1.53it/s]
 46%|████▌     | 3285/7150 [42:42<42:09,  1.53it/s]
 46%|████▌     | 3286/7150 [42:43<42:09,  1.53it/s]
 46%|████▌     | 3287/7150 [42:44<42:13,  1.52it/s]
 46%|████▌     | 3288/7150 [42:44<42:10,  1.53it/s]
 46%|████▌     | 3289/7150 [42:45<39:00,  1.65it/s]
 46%|████▌     | 3290/7150 [42:46<41:33,  1.55it/s]
                                                   
{'loss': '3.088', 'grad_norm': '0.2844', 'learning_rate': '0.0006', 'epoch': '23.01'}
+
 46%|████▌     | 3290/7150 [42:46<41:33,  1.55it/s]
 46%|████▌     | 3291/7150 [42:46<41:44,  1.54it/s]
 46%|████▌     | 3292/7150 [42:47<41:49,  1.54it/s]
 46%|████▌     | 3293/7150 [42:48<41:54,  1.53it/s]
 46%|████▌     | 3294/7150 [42:48<41:56,  1.53it/s]
 46%|████▌     | 3295/7150 [42:49<41:59,  1.53it/s]
 46%|████▌     | 3296/7150 [42:50<42:00,  1.53it/s]
 46%|████▌     | 3297/7150 [42:50<42:00,  1.53it/s]
 46%|████▌     | 3298/7150 [42:51<42:00,  1.53it/s]
 46%|████▌     | 3299/7150 [42:52<42:00,  1.53it/s]
 46%|████▌     | 3300/7150 [42:52<41:59,  1.53it/s]
                                                   
{'loss': '3', 'grad_norm': '0.4554', 'learning_rate': '0.0006', 'epoch': '23.08'}
+
 46%|████▌     | 3300/7150 [42:52<41:59,  1.53it/s]
 46%|████▌     | 3301/7150 [42:53<41:58,  1.53it/s]
 46%|████▌     | 3302/7150 [42:54<41:59,  1.53it/s]
 46%|████▌     | 3303/7150 [42:54<41:59,  1.53it/s]
 46%|████▌     | 3304/7150 [42:55<41:58,  1.53it/s]
 46%|████▌     | 3305/7150 [42:55<41:56,  1.53it/s]
 46%|████▌     | 3306/7150 [42:56<41:57,  1.53it/s]
 46%|████▋     | 3307/7150 [42:57<41:57,  1.53it/s]
 46%|████▋     | 3308/7150 [42:57<41:57,  1.53it/s]
 46%|████▋     | 3309/7150 [42:58<41:56,  1.53it/s]
 46%|████▋     | 3310/7150 [42:59<41:56,  1.53it/s]
                                                   
{'loss': '3.011', 'grad_norm': '0.332', 'learning_rate': '0.0006', 'epoch': '23.15'}
+
 46%|████▋     | 3310/7150 [42:59<41:56,  1.53it/s]
 46%|████▋     | 3311/7150 [42:59<41:57,  1.53it/s]
 46%|████▋     | 3312/7150 [43:00<41:54,  1.53it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.60it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.81it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.61it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.06it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.76it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.57it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.45it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.37it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.32it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.30it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.28it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.26it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.26it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.24it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.24it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.24it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.25it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.24it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s]
                                                   
+
                                               
{'eval_loss': '3.807', 'eval_runtime': '3.247', 'eval_samples_per_second': '751.9', 'eval_steps_per_second': '12.01', 'epoch': '23.16'}
+
 46%|████▋     | 3312/7150 [43:03<41:54,  1.53it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s][eval] epoch 23.161 step 3312 loss 3.8074 (best 3.7377 @ ep 15.609, 15/20 since best)
+
+
                                               
 46%|████▋     | 3313/7150 [43:04<1:44:10,  1.63s/it]
 46%|████▋     | 3314/7150 [43:05<1:25:26,  1.34s/it]
 46%|████▋     | 3315/7150 [43:05<1:12:20,  1.13s/it]
 46%|████▋     | 3316/7150 [43:06<1:03:11,  1.01it/s]
 46%|████▋     | 3317/7150 [43:07<56:46,  1.13it/s]  
 46%|████▋     | 3318/7150 [43:07<52:15,  1.22it/s]
 46%|████▋     | 3319/7150 [43:08<49:07,  1.30it/s]
 46%|████▋     | 3320/7150 [43:09<46:55,  1.36it/s]
                                                   
{'loss': '3.005', 'grad_norm': '0.2811', 'learning_rate': '0.0006', 'epoch': '23.22'}
+
 46%|████▋     | 3320/7150 [43:09<46:55,  1.36it/s]
 46%|████▋     | 3321/7150 [43:09<45:25,  1.40it/s]
 46%|████▋     | 3322/7150 [43:10<44:19,  1.44it/s]
 46%|████▋     | 3323/7150 [43:11<43:31,  1.47it/s]
 46%|████▋     | 3324/7150 [43:11<42:59,  1.48it/s]
 47%|████▋     | 3325/7150 [43:12<42:36,  1.50it/s]
 47%|████▋     | 3326/7150 [43:12<42:18,  1.51it/s]
 47%|████▋     | 3327/7150 [43:13<42:08,  1.51it/s]
 47%|████▋     | 3328/7150 [43:14<42:00,  1.52it/s]
 47%|████▋     | 3329/7150 [43:14<41:54,  1.52it/s]
 47%|████▋     | 3330/7150 [43:15<41:49,  1.52it/s]
                                                   
{'loss': '3.02', 'grad_norm': '0.2545', 'learning_rate': '0.0006', 'epoch': '23.29'}
+
 47%|████▋     | 3330/7150 [43:15<41:49,  1.52it/s]
 47%|████▋     | 3331/7150 [43:16<41:47,  1.52it/s]
 47%|████▋     | 3332/7150 [43:16<41:46,  1.52it/s]
 47%|████▋     | 3333/7150 [43:17<41:43,  1.52it/s]
 47%|████▋     | 3334/7150 [43:18<41:41,  1.53it/s]
 47%|████▋     | 3335/7150 [43:18<41:40,  1.53it/s]
 47%|████▋     | 3336/7150 [43:19<41:39,  1.53it/s]
 47%|████▋     | 3337/7150 [43:20<41:38,  1.53it/s]
 47%|████▋     | 3338/7150 [43:20<41:37,  1.53it/s]
 47%|████▋     | 3339/7150 [43:21<41:35,  1.53it/s]
 47%|████▋     | 3340/7150 [43:22<41:35,  1.53it/s]
                                                   
{'loss': '3.029', 'grad_norm': '0.2547', 'learning_rate': '0.0006', 'epoch': '23.36'}
+
 47%|████▋     | 3340/7150 [43:22<41:35,  1.53it/s]
 47%|████▋     | 3341/7150 [43:22<41:36,  1.53it/s]
 47%|████▋     | 3342/7150 [43:23<41:36,  1.53it/s]
 47%|████▋     | 3343/7150 [43:24<41:33,  1.53it/s]
 47%|████▋     | 3344/7150 [43:24<41:33,  1.53it/s]
 47%|████▋     | 3345/7150 [43:25<41:33,  1.53it/s]
 47%|████▋     | 3346/7150 [43:26<41:33,  1.53it/s]
 47%|████▋     | 3347/7150 [43:26<41:30,  1.53it/s]
 47%|████▋     | 3348/7150 [43:27<41:32,  1.53it/s]
 47%|████▋     | 3349/7150 [43:28<41:31,  1.53it/s]
 47%|████▋     | 3350/7150 [43:28<41:34,  1.52it/s]
                                                   
{'loss': '3.032', 'grad_norm': '0.2677', 'learning_rate': '0.0006', 'epoch': '23.43'}
+
 47%|████▋     | 3350/7150 [43:28<41:34,  1.52it/s]
 47%|████▋     | 3351/7150 [43:29<41:35,  1.52it/s]
 47%|████▋     | 3352/7150 [43:30<41:32,  1.52it/s]
 47%|████▋     | 3353/7150 [43:30<41:30,  1.52it/s]
 47%|████▋     | 3354/7150 [43:31<41:29,  1.52it/s]
 47%|████▋     | 3355/7150 [43:31<41:28,  1.53it/s]
 47%|████▋     | 3356/7150 [43:32<41:27,  1.53it/s]
 47%|████▋     | 3357/7150 [43:33<41:24,  1.53it/s]
 47%|████▋     | 3358/7150 [43:33<41:24,  1.53it/s]
 47%|████▋     | 3359/7150 [43:34<41:24,  1.53it/s]
 47%|████▋     | 3360/7150 [43:35<41:24,  1.53it/s]
                                                   
{'loss': '3.041', 'grad_norm': '0.2574', 'learning_rate': '0.0006', 'epoch': '23.5'}
+
 47%|████▋     | 3360/7150 [43:35<41:24,  1.53it/s]
 47%|████▋     | 3361/7150 [43:35<41:24,  1.53it/s]
 47%|████▋     | 3362/7150 [43:36<41:23,  1.53it/s]
 47%|████▋     | 3363/7150 [43:37<41:24,  1.52it/s]
 47%|████▋     | 3364/7150 [43:37<41:23,  1.52it/s]
 47%|████▋     | 3365/7150 [43:38<41:22,  1.52it/s]
 47%|████▋     | 3366/7150 [43:39<41:19,  1.53it/s]
 47%|████▋     | 3367/7150 [43:39<41:20,  1.52it/s]
 47%|████▋     | 3368/7150 [43:40<41:21,  1.52it/s]
 47%|████▋     | 3369/7150 [43:41<41:20,  1.52it/s]
 47%|████▋     | 3370/7150 [43:41<41:18,  1.52it/s]
                                                   
{'loss': '3.057', 'grad_norm': '0.2789', 'learning_rate': '0.0006', 'epoch': '23.57'}
+
 47%|████▋     | 3370/7150 [43:41<41:18,  1.52it/s]
 47%|████▋     | 3371/7150 [43:42<41:17,  1.53it/s]
 47%|████▋     | 3372/7150 [43:43<41:18,  1.52it/s]
 47%|████▋     | 3373/7150 [43:43<41:16,  1.52it/s]
 47%|████▋     | 3374/7150 [43:44<41:15,  1.53it/s]
 47%|████▋     | 3375/7150 [43:45<41:13,  1.53it/s]
 47%|████▋     | 3376/7150 [43:45<41:13,  1.53it/s]
 47%|████▋     | 3377/7150 [43:46<41:15,  1.52it/s]
 47%|████▋     | 3378/7150 [43:47<41:14,  1.52it/s]
 47%|████▋     | 3379/7150 [43:47<41:13,  1.52it/s]
 47%|████▋     | 3380/7150 [43:48<41:10,  1.53it/s]
                                                   
{'loss': '3.053', 'grad_norm': '0.249', 'learning_rate': '0.0006', 'epoch': '23.64'}
+
 47%|████▋     | 3380/7150 [43:48<41:10,  1.53it/s]
 47%|████▋     | 3381/7150 [43:49<41:12,  1.52it/s]
 47%|████▋     | 3382/7150 [43:49<41:12,  1.52it/s]
 47%|████▋     | 3383/7150 [43:50<41:11,  1.52it/s]
 47%|████▋     | 3384/7150 [43:51<41:09,  1.52it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.65it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.80it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.61it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.06it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.76it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.58it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.47it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.39it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.33it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.29it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.28it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.28it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.26it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.25it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.24it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.23it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.25it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.24it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s]
                                                   
+
                                               
{'eval_loss': '3.785', 'eval_runtime': '3.252', 'eval_samples_per_second': '750.6', 'eval_steps_per_second': '11.99', 'epoch': '23.67'}
+
 47%|████▋     | 3384/7150 [43:54<41:09,  1.52it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s][eval] epoch 23.665 step 3384 loss 3.7846 (best 3.7377 @ ep 15.609, 16/20 since best)
+
+
                                               
 47%|████▋     | 3385/7150 [43:54<1:42:25,  1.63s/it]
 47%|████▋     | 3386/7150 [43:55<1:24:00,  1.34s/it]
 47%|████▋     | 3387/7150 [43:56<1:11:06,  1.13s/it]
 47%|████▋     | 3388/7150 [43:56<1:02:03,  1.01it/s]
 47%|████▋     | 3389/7150 [43:57<55:46,  1.12it/s]  
 47%|████▋     | 3390/7150 [43:58<51:21,  1.22it/s]
                                                   
{'loss': '3.062', 'grad_norm': '0.2477', 'learning_rate': '0.0006', 'epoch': '23.71'}
+
 47%|████▋     | 3390/7150 [43:58<51:21,  1.22it/s]
 47%|████▋     | 3391/7150 [43:58<48:17,  1.30it/s]
 47%|████▋     | 3392/7150 [43:59<46:07,  1.36it/s]
 47%|████▋     | 3393/7150 [44:00<44:34,  1.40it/s]
 47%|████▋     | 3394/7150 [44:00<43:31,  1.44it/s]
 47%|████▋     | 3395/7150 [44:01<42:46,  1.46it/s]
 47%|████▋     | 3396/7150 [44:02<42:13,  1.48it/s]
 48%|████▊     | 3397/7150 [44:02<41:50,  1.50it/s]
 48%|████▊     | 3398/7150 [44:03<41:34,  1.50it/s]
 48%|████▊     | 3399/7150 [44:04<41:24,  1.51it/s]
 48%|████▊     | 3400/7150 [44:04<41:16,  1.51it/s]
                                                   
{'loss': '3.074', 'grad_norm': '0.2403', 'learning_rate': '0.0006', 'epoch': '23.78'}
+
 48%|████▊     | 3400/7150 [44:04<41:16,  1.51it/s]
 48%|████▊     | 3401/7150 [44:05<41:11,  1.52it/s]
 48%|████▊     | 3402/7150 [44:06<41:04,  1.52it/s]
 48%|████▊     | 3403/7150 [44:06<41:03,  1.52it/s]
 48%|████▊     | 3404/7150 [44:07<40:59,  1.52it/s]
 48%|████▊     | 3405/7150 [44:08<40:57,  1.52it/s]
 48%|████▊     | 3406/7150 [44:08<40:53,  1.53it/s]
 48%|████▊     | 3407/7150 [44:09<40:54,  1.52it/s]
 48%|████▊     | 3408/7150 [44:10<40:53,  1.52it/s]
 48%|████▊     | 3409/7150 [44:10<40:51,  1.53it/s]
 48%|████▊     | 3410/7150 [44:11<40:48,  1.53it/s]
                                                   
{'loss': '3.081', 'grad_norm': '0.2475', 'learning_rate': '0.0006', 'epoch': '23.85'}
+
 48%|████▊     | 3410/7150 [44:11<40:48,  1.53it/s]
 48%|████▊     | 3411/7150 [44:11<40:50,  1.53it/s]
 48%|████▊     | 3412/7150 [44:12<40:49,  1.53it/s]
 48%|████▊     | 3413/7150 [44:13<40:47,  1.53it/s]
 48%|████▊     | 3414/7150 [44:13<40:46,  1.53it/s]
 48%|████▊     | 3415/7150 [44:14<40:46,  1.53it/s]
 48%|████▊     | 3416/7150 [44:15<40:47,  1.53it/s]
 48%|████▊     | 3417/7150 [44:15<40:45,  1.53it/s]
 48%|████▊     | 3418/7150 [44:16<40:43,  1.53it/s]
 48%|████▊     | 3419/7150 [44:17<40:44,  1.53it/s]
 48%|████▊     | 3420/7150 [44:17<40:44,  1.53it/s]
                                                   
{'loss': '3.095', 'grad_norm': '0.2763', 'learning_rate': '0.0006', 'epoch': '23.92'}
+
 48%|████▊     | 3420/7150 [44:17<40:44,  1.53it/s]
 48%|████▊     | 3421/7150 [44:18<40:45,  1.52it/s]
 48%|████▊     | 3422/7150 [44:19<40:45,  1.52it/s]
 48%|████▊     | 3423/7150 [44:19<40:42,  1.53it/s]
 48%|████▊     | 3424/7150 [44:20<40:43,  1.53it/s]
 48%|████▊     | 3425/7150 [44:21<40:43,  1.52it/s]
 48%|████▊     | 3426/7150 [44:21<40:41,  1.53it/s]
 48%|████▊     | 3427/7150 [44:22<40:40,  1.53it/s]
 48%|████▊     | 3428/7150 [44:23<40:37,  1.53it/s]
 48%|████▊     | 3429/7150 [44:23<40:38,  1.53it/s]
 48%|████▊     | 3430/7150 [44:24<40:45,  1.52it/s]
                                                   
{'loss': '3.088', 'grad_norm': '0.2539', 'learning_rate': '0.0006', 'epoch': '23.99'}
+
 48%|████▊     | 3430/7150 [44:24<40:45,  1.52it/s]
 48%|████▊     | 3431/7150 [44:25<40:42,  1.52it/s]
 48%|████▊     | 3432/7150 [44:25<37:38,  1.65it/s]
 48%|████▊     | 3433/7150 [44:26<40:03,  1.55it/s]
 48%|████▊     | 3434/7150 [44:26<40:12,  1.54it/s]
 48%|████▊     | 3435/7150 [44:27<40:20,  1.53it/s]
 48%|████▊     | 3436/7150 [44:28<40:26,  1.53it/s]
 48%|████▊     | 3437/7150 [44:28<40:28,  1.53it/s]
 48%|████▊     | 3438/7150 [44:29<40:28,  1.53it/s]
 48%|████▊     | 3439/7150 [44:30<40:29,  1.53it/s]
 48%|████▊     | 3440/7150 [44:30<40:30,  1.53it/s]
                                                   
{'loss': '2.991', 'grad_norm': '0.5128', 'learning_rate': '0.0006', 'epoch': '24.06'}
+
 48%|████▊     | 3440/7150 [44:30<40:30,  1.53it/s]
 48%|████▊     | 3441/7150 [44:31<40:34,  1.52it/s]
 48%|████▊     | 3442/7150 [44:32<40:32,  1.52it/s]
 48%|████▊     | 3443/7150 [44:32<40:31,  1.52it/s]
 48%|████▊     | 3444/7150 [44:33<40:28,  1.53it/s]
 48%|████▊     | 3445/7150 [44:34<40:28,  1.53it/s]
 48%|████▊     | 3446/7150 [44:34<40:27,  1.53it/s]
 48%|████▊     | 3447/7150 [44:35<40:26,  1.53it/s]
 48%|████▊     | 3448/7150 [44:36<40:25,  1.53it/s]
 48%|████▊     | 3449/7150 [44:36<40:24,  1.53it/s]
 48%|████▊     | 3450/7150 [44:37<40:25,  1.53it/s]
                                                   
{'loss': '2.96', 'grad_norm': '0.3154', 'learning_rate': '0.0006', 'epoch': '24.13'}
+
 48%|████▊     | 3450/7150 [44:37<40:25,  1.53it/s]
 48%|████▊     | 3451/7150 [44:38<40:25,  1.52it/s]
 48%|████▊     | 3452/7150 [44:38<40:23,  1.53it/s]
 48%|████▊     | 3453/7150 [44:39<46:52,  1.31it/s]
 48%|████▊     | 3454/7150 [44:40<44:53,  1.37it/s]
 48%|████▊     | 3455/7150 [44:41<43:31,  1.41it/s]
 48%|████▊     | 3456/7150 [44:41<42:36,  1.44it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.87it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.89it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.65it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.10it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.76it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.58it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.46it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.38it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.34it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.30it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.29it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.27it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.26it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.26it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.26it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.25it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.25it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.25it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.20it/s]
                                                   
+
                                               
{'eval_loss': '3.819', 'eval_runtime': '3.236', 'eval_samples_per_second': '754.3', 'eval_steps_per_second': '12.05', 'epoch': '24.17'}
+
 48%|████▊     | 3456/7150 [44:44<42:36,  1.44it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.20it/s][eval] epoch 24.168 step 3456 loss 3.8193 (best 3.7377 @ ep 15.609, 17/20 since best)
+
+
                                               
 48%|████▊     | 3457/7150 [44:45<1:41:41,  1.65s/it]
 48%|████▊     | 3458/7150 [44:46<1:23:14,  1.35s/it]
 48%|████▊     | 3459/7150 [44:46<1:10:21,  1.14s/it]
 48%|████▊     | 3460/7150 [44:47<1:01:20,  1.00it/s]
                                                     
{'loss': '2.968', 'grad_norm': '0.2892', 'learning_rate': '0.0006', 'epoch': '24.2'}
+
 48%|████▊     | 3460/7150 [44:47<1:01:20,  1.00it/s]
 48%|████▊     | 3461/7150 [44:48<55:03,  1.12it/s]  
 48%|████▊     | 3462/7150 [44:48<50:37,  1.21it/s]
 48%|████▊     | 3463/7150 [44:49<47:31,  1.29it/s]
 48%|████▊     | 3464/7150 [44:50<45:18,  1.36it/s]
 48%|████▊     | 3465/7150 [44:50<43:48,  1.40it/s]
 48%|████▊     | 3466/7150 [44:51<42:43,  1.44it/s]
 48%|████▊     | 3467/7150 [44:52<42:01,  1.46it/s]
 49%|████▊     | 3468/7150 [44:52<41:28,  1.48it/s]
 49%|████▊     | 3469/7150 [44:53<41:04,  1.49it/s]
 49%|████▊     | 3470/7150 [44:54<40:49,  1.50it/s]
                                                   
{'loss': '2.989', 'grad_norm': '0.3134', 'learning_rate': '0.0006', 'epoch': '24.27'}
+
 49%|████▊     | 3470/7150 [44:54<40:49,  1.50it/s]
 49%|████▊     | 3471/7150 [44:54<40:40,  1.51it/s]
 49%|████▊     | 3472/7150 [44:55<40:32,  1.51it/s]
 49%|████▊     | 3473/7150 [44:56<40:25,  1.52it/s]
 49%|████▊     | 3474/7150 [44:56<40:18,  1.52it/s]
 49%|████▊     | 3475/7150 [44:57<40:15,  1.52it/s]
 49%|████▊     | 3476/7150 [44:58<40:13,  1.52it/s]
 49%|████▊     | 3477/7150 [44:58<40:12,  1.52it/s]
 49%|████▊     | 3478/7150 [44:59<40:09,  1.52it/s]
 49%|████▊     | 3479/7150 [45:00<40:09,  1.52it/s]
 49%|████▊     | 3480/7150 [45:00<40:08,  1.52it/s]
                                                   
{'loss': '3.005', 'grad_norm': '0.2887', 'learning_rate': '0.0006', 'epoch': '24.34'}
+
 49%|████▊     | 3480/7150 [45:00<40:08,  1.52it/s]
 49%|████▊     | 3481/7150 [45:01<40:10,  1.52it/s]
 49%|████▊     | 3482/7150 [45:02<40:08,  1.52it/s]
 49%|████▊     | 3483/7150 [45:02<40:06,  1.52it/s]
 49%|████▊     | 3484/7150 [45:03<40:03,  1.52it/s]
 49%|████▊     | 3485/7150 [45:04<40:04,  1.52it/s]
 49%|████▉     | 3486/7150 [45:04<40:05,  1.52it/s]
 49%|████▉     | 3487/7150 [45:05<40:03,  1.52it/s]
 49%|████▉     | 3488/7150 [45:05<40:00,  1.53it/s]
 49%|████▉     | 3489/7150 [45:06<39:59,  1.53it/s]
 49%|████▉     | 3490/7150 [45:07<39:59,  1.53it/s]
                                                   
{'loss': '3.012', 'grad_norm': '0.2804', 'learning_rate': '0.0006', 'epoch': '24.41'}
+
 49%|████▉     | 3490/7150 [45:07<39:59,  1.53it/s]
 49%|████▉     | 3491/7150 [45:07<39:58,  1.53it/s]
 49%|████▉     | 3492/7150 [45:08<39:56,  1.53it/s]
 49%|████▉     | 3493/7150 [45:09<39:55,  1.53it/s]
 49%|████▉     | 3494/7150 [45:09<39:56,  1.53it/s]
 49%|████▉     | 3495/7150 [45:10<39:54,  1.53it/s]
 49%|████▉     | 3496/7150 [45:11<39:55,  1.53it/s]
 49%|████▉     | 3497/7150 [45:11<39:53,  1.53it/s]
 49%|████▉     | 3498/7150 [45:12<39:54,  1.53it/s]
 49%|████▉     | 3499/7150 [45:13<39:53,  1.53it/s]
 49%|████▉     | 3500/7150 [45:13<39:53,  1.53it/s]
                                                   
{'loss': '3.025', 'grad_norm': '0.2781', 'learning_rate': '0.0006', 'epoch': '24.48'}
+
 49%|████▉     | 3500/7150 [45:13<39:53,  1.53it/s]
 49%|████▉     | 3501/7150 [45:14<39:53,  1.52it/s]
 49%|████▉     | 3502/7150 [45:15<39:51,  1.53it/s]
 49%|████▉     | 3503/7150 [45:15<39:52,  1.52it/s]
 49%|████▉     | 3504/7150 [45:16<39:50,  1.52it/s]
 49%|████▉     | 3505/7150 [45:17<39:50,  1.52it/s]
 49%|████▉     | 3506/7150 [45:17<39:48,  1.53it/s]
 49%|████▉     | 3507/7150 [45:18<39:48,  1.53it/s]
 49%|████▉     | 3508/7150 [45:19<39:48,  1.52it/s]
 49%|████▉     | 3509/7150 [45:19<39:47,  1.53it/s]
 49%|████▉     | 3510/7150 [45:20<39:47,  1.52it/s]
                                                   
{'loss': '3.05', 'grad_norm': '0.2686', 'learning_rate': '0.0006', 'epoch': '24.55'}
+
 49%|████▉     | 3510/7150 [45:20<39:47,  1.52it/s]
 49%|████▉     | 3511/7150 [45:21<39:46,  1.53it/s]
 49%|████▉     | 3512/7150 [45:21<39:45,  1.52it/s]
 49%|████▉     | 3513/7150 [45:22<39:44,  1.53it/s]
 49%|████▉     | 3514/7150 [45:23<39:43,  1.53it/s]
 49%|████▉     | 3515/7150 [45:23<39:43,  1.52it/s]
 49%|████▉     | 3516/7150 [45:24<39:41,  1.53it/s]
 49%|████▉     | 3517/7150 [45:24<39:42,  1.52it/s]
 49%|████▉     | 3518/7150 [45:25<39:42,  1.52it/s]
 49%|████▉     | 3519/7150 [45:26<39:41,  1.52it/s]
 49%|████▉     | 3520/7150 [45:26<39:40,  1.53it/s]
                                                   
{'loss': '3.05', 'grad_norm': '0.2651', 'learning_rate': '0.0006', 'epoch': '24.62'}
+
 49%|████▉     | 3520/7150 [45:26<39:40,  1.53it/s]
 49%|████▉     | 3521/7150 [45:27<39:39,  1.53it/s]
 49%|████▉     | 3522/7150 [45:28<39:38,  1.53it/s]
 49%|████▉     | 3523/7150 [45:28<39:37,  1.53it/s]
 49%|████▉     | 3524/7150 [45:29<39:36,  1.53it/s]
 49%|████▉     | 3525/7150 [45:30<39:34,  1.53it/s]
 49%|████▉     | 3526/7150 [45:30<39:33,  1.53it/s]
 49%|████▉     | 3527/7150 [45:31<39:32,  1.53it/s]
 49%|████▉     | 3528/7150 [45:32<39:30,  1.53it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.34it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.72it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.57it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.04it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.74it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.56it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.46it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.39it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.33it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.30it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.28it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.26it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.26it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.26it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.25it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.23it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.24it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.24it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s]
                                                   
+
                                               
{'eval_loss': '3.792', 'eval_runtime': '3.261', 'eval_samples_per_second': '748.5', 'eval_steps_per_second': '11.96', 'epoch': '24.67'}
+
 49%|████▉     | 3528/7150 [45:35<39:30,  1.53it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s][eval] epoch 24.673 step 3528 loss 3.7917 (best 3.7377 @ ep 15.609, 18/20 since best)
+
+
                                               
 49%|████▉     | 3529/7150 [45:36<1:38:33,  1.63s/it]
 49%|████▉     | 3530/7150 [45:36<1:20:50,  1.34s/it]
                                                     
{'loss': '3.042', 'grad_norm': '0.2634', 'learning_rate': '0.0006', 'epoch': '24.69'}
+
 49%|████▉     | 3530/7150 [45:36<1:20:50,  1.34s/it]
 49%|████▉     | 3531/7150 [45:37<1:08:25,  1.13s/it]
 49%|████▉     | 3532/7150 [45:38<59:44,  1.01it/s]  
 49%|████▉     | 3533/7150 [45:38<53:37,  1.12it/s]
 49%|████▉     | 3534/7150 [45:39<49:24,  1.22it/s]
 49%|████▉     | 3535/7150 [45:40<46:24,  1.30it/s]
 49%|████▉     | 3536/7150 [45:40<44:17,  1.36it/s]
 49%|████▉     | 3537/7150 [45:41<42:49,  1.41it/s]
 49%|████▉     | 3538/7150 [45:42<41:48,  1.44it/s]
 49%|████▉     | 3539/7150 [45:42<41:04,  1.47it/s]
 50%|████▉     | 3540/7150 [45:43<40:32,  1.48it/s]
                                                   
{'loss': '3.057', 'grad_norm': '0.2559', 'learning_rate': '0.0006', 'epoch': '24.76'}
+
 50%|████▉     | 3540/7150 [45:43<40:32,  1.48it/s]
 50%|████▉     | 3541/7150 [45:43<40:12,  1.50it/s]
 50%|████▉     | 3542/7150 [45:44<39:57,  1.50it/s]
 50%|████▉     | 3543/7150 [45:45<39:48,  1.51it/s]
 50%|████▉     | 3544/7150 [45:45<39:38,  1.52it/s]
 50%|████▉     | 3545/7150 [45:46<39:32,  1.52it/s]
 50%|████▉     | 3546/7150 [45:47<39:29,  1.52it/s]
 50%|████▉     | 3547/7150 [45:47<39:25,  1.52it/s]
 50%|████▉     | 3548/7150 [45:48<39:22,  1.52it/s]
 50%|████▉     | 3549/7150 [45:49<39:21,  1.53it/s]
 50%|████▉     | 3550/7150 [45:49<39:22,  1.52it/s]
                                                   
{'loss': '3.066', 'grad_norm': '0.2655', 'learning_rate': '0.0006', 'epoch': '24.83'}
+
 50%|████▉     | 3550/7150 [45:49<39:22,  1.52it/s]
 50%|████▉     | 3551/7150 [45:50<39:21,  1.52it/s]
 50%|████▉     | 3552/7150 [45:51<39:20,  1.52it/s]
 50%|████▉     | 3553/7150 [45:51<39:19,  1.52it/s]
 50%|████▉     | 3554/7150 [45:52<39:18,  1.52it/s]
 50%|████▉     | 3555/7150 [45:53<39:16,  1.53it/s]
 50%|████▉     | 3556/7150 [45:53<39:16,  1.52it/s]
 50%|████▉     | 3557/7150 [45:54<39:14,  1.53it/s]
 50%|████▉     | 3558/7150 [45:55<39:13,  1.53it/s]
 50%|████▉     | 3559/7150 [45:55<39:13,  1.53it/s]
 50%|████▉     | 3560/7150 [45:56<39:12,  1.53it/s]
                                                   
{'loss': '3.067', 'grad_norm': '0.251', 'learning_rate': '0.0006', 'epoch': '24.9'}
+
 50%|████▉     | 3560/7150 [45:56<39:12,  1.53it/s]
 50%|████▉     | 3561/7150 [45:57<39:13,  1.53it/s]
 50%|████▉     | 3562/7150 [45:57<39:11,  1.53it/s]
 50%|████▉     | 3563/7150 [45:58<39:11,  1.53it/s]
 50%|████▉     | 3564/7150 [45:59<39:10,  1.53it/s]
 50%|████▉     | 3565/7150 [45:59<39:09,  1.53it/s]
 50%|████▉     | 3566/7150 [46:00<39:07,  1.53it/s]
 50%|████▉     | 3567/7150 [46:01<39:07,  1.53it/s]
 50%|████▉     | 3568/7150 [46:01<39:07,  1.53it/s]
 50%|████▉     | 3569/7150 [46:02<39:05,  1.53it/s]
 50%|████▉     | 3570/7150 [46:02<39:03,  1.53it/s]
                                                   
{'loss': '3.065', 'grad_norm': '0.2604', 'learning_rate': '0.0006', 'epoch': '24.97'}
+
 50%|████▉     | 3570/7150 [46:02<39:03,  1.53it/s]
 50%|████▉     | 3571/7150 [46:03<39:04,  1.53it/s]
 50%|████▉     | 3572/7150 [46:04<39:04,  1.53it/s]
 50%|████▉     | 3573/7150 [46:04<39:07,  1.52it/s]
 50%|████▉     | 3574/7150 [46:05<39:04,  1.53it/s]
 50%|█████     | 3575/7150 [46:06<36:08,  1.65it/s]
 50%|█████     | 3576/7150 [46:06<38:12,  1.56it/s]
 50%|█████     | 3577/7150 [46:07<38:27,  1.55it/s]
 50%|█████     | 3578/7150 [46:08<38:37,  1.54it/s]
 50%|█████     | 3579/7150 [46:08<38:42,  1.54it/s]
 50%|█████     | 3580/7150 [46:09<38:45,  1.54it/s]
                                                   
{'loss': '3', 'grad_norm': '0.5006', 'learning_rate': '0.0006', 'epoch': '25.04'}
+
 50%|█████     | 3580/7150 [46:09<38:45,  1.54it/s]
 50%|█████     | 3581/7150 [46:10<38:52,  1.53it/s]
 50%|█████     | 3582/7150 [46:10<38:52,  1.53it/s]
 50%|█████     | 3583/7150 [46:11<38:53,  1.53it/s]
 50%|█████     | 3584/7150 [46:12<38:51,  1.53it/s]
 50%|█████     | 3585/7150 [46:12<38:53,  1.53it/s]
 50%|█████     | 3586/7150 [46:13<38:52,  1.53it/s]
 50%|█████     | 3587/7150 [46:14<38:52,  1.53it/s]
 50%|█████     | 3588/7150 [46:14<38:52,  1.53it/s]
 50%|█████     | 3589/7150 [46:15<38:52,  1.53it/s]
 50%|█████     | 3590/7150 [46:15<38:53,  1.53it/s]
                                                   
{'loss': '2.937', 'grad_norm': '0.3781', 'learning_rate': '0.0006', 'epoch': '25.11'}
+
 50%|█████     | 3590/7150 [46:15<38:53,  1.53it/s]
 50%|█████     | 3591/7150 [46:16<38:53,  1.53it/s]
 50%|█████     | 3592/7150 [46:17<38:51,  1.53it/s]
 50%|█████     | 3593/7150 [46:17<38:49,  1.53it/s]
 50%|█████     | 3594/7150 [46:18<38:50,  1.53it/s]
 50%|█████     | 3595/7150 [46:19<38:50,  1.53it/s]
 50%|█████     | 3596/7150 [46:19<38:49,  1.53it/s]
 50%|█████     | 3597/7150 [46:20<45:03,  1.31it/s]
 50%|█████     | 3598/7150 [46:21<49:23,  1.20it/s]
 50%|█████     | 3599/7150 [46:22<46:11,  1.28it/s]
 50%|█████     | 3600/7150 [46:23<43:55,  1.35it/s]
                                                   
{'loss': '2.941', 'grad_norm': '0.321', 'learning_rate': '0.0006', 'epoch': '25.18'}
+
 50%|█████     | 3600/7150 [46:23<43:55,  1.35it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.15it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.64it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.53it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.01it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.72it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.55it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.45it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.36it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.31it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.27it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.27it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.26it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.25it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.25it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.25it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.24it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.23it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.23it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.22it/s]
                                                   
+
                                               
{'eval_loss': '3.82', 'eval_runtime': '3.247', 'eval_samples_per_second': '751.7', 'eval_steps_per_second': '12.01', 'epoch': '25.18'}
+
 50%|█████     | 3600/7150 [46:26<43:55,  1.35it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.22it/s][eval] epoch 25.175 step 3600 loss 3.8201 (best 3.7377 @ ep 15.609, 19/20 since best)
+
+
                                               
 50%|█████     | 3601/7150 [46:27<1:40:04,  1.69s/it]
 50%|█████     | 3602/7150 [46:27<1:21:40,  1.38s/it]
 50%|█████     | 3603/7150 [46:28<1:08:46,  1.16s/it]
 50%|█████     | 3604/7150 [46:29<59:43,  1.01s/it]  
 50%|█████     | 3605/7150 [46:29<53:26,  1.11it/s]
 50%|█████     | 3606/7150 [46:30<49:03,  1.20it/s]
 50%|█████     | 3607/7150 [46:31<45:56,  1.29it/s]
 50%|█████     | 3608/7150 [46:31<43:47,  1.35it/s]
 50%|█████     | 3609/7150 [46:32<42:13,  1.40it/s]
 50%|█████     | 3610/7150 [46:33<41:09,  1.43it/s]
                                                   
{'loss': '2.965', 'grad_norm': '0.3171', 'learning_rate': '0.0006', 'epoch': '25.25'}
+
 50%|█████     | 3610/7150 [46:33<41:09,  1.43it/s]
 51%|█████     | 3611/7150 [46:33<40:27,  1.46it/s]
 51%|█████     | 3612/7150 [46:34<39:56,  1.48it/s]
 51%|█████     | 3613/7150 [46:35<39:32,  1.49it/s]
 51%|█████     | 3614/7150 [46:35<39:16,  1.50it/s]
 51%|█████     | 3615/7150 [46:36<39:02,  1.51it/s]
 51%|█████     | 3616/7150 [46:36<38:55,  1.51it/s]
 51%|█████     | 3617/7150 [46:37<38:47,  1.52it/s]
 51%|█████     | 3618/7150 [46:38<38:44,  1.52it/s]
 51%|█████     | 3619/7150 [46:38<38:39,  1.52it/s]
 51%|█████     | 3620/7150 [46:39<38:37,  1.52it/s]
                                                   
{'loss': '2.987', 'grad_norm': '0.2892', 'learning_rate': '0.0006', 'epoch': '25.32'}
+
 51%|█████     | 3620/7150 [46:39<38:37,  1.52it/s]
 51%|█████     | 3621/7150 [46:40<38:37,  1.52it/s]
 51%|█████     | 3622/7150 [46:40<38:34,  1.52it/s]
 51%|█████     | 3623/7150 [46:41<38:34,  1.52it/s]
 51%|█████     | 3624/7150 [46:42<38:31,  1.53it/s]
 51%|█████     | 3625/7150 [46:42<38:29,  1.53it/s]
 51%|█████     | 3626/7150 [46:43<38:28,  1.53it/s]
 51%|█████     | 3627/7150 [46:44<38:27,  1.53it/s]
 51%|█████     | 3628/7150 [46:44<38:26,  1.53it/s]
 51%|█████     | 3629/7150 [46:45<38:26,  1.53it/s]
 51%|█████     | 3630/7150 [46:46<38:26,  1.53it/s]
                                                   
{'loss': '3.002', 'grad_norm': '0.2811', 'learning_rate': '0.0006', 'epoch': '25.39'}
+
 51%|█████     | 3630/7150 [46:46<38:26,  1.53it/s]
 51%|█████     | 3631/7150 [46:46<38:28,  1.52it/s]
 51%|█████     | 3632/7150 [46:47<38:25,  1.53it/s]
 51%|█████     | 3633/7150 [46:48<38:24,  1.53it/s]
 51%|█████     | 3634/7150 [46:48<38:25,  1.52it/s]
 51%|█████     | 3635/7150 [46:49<38:25,  1.52it/s]
 51%|█████     | 3636/7150 [46:50<44:18,  1.32it/s]
 51%|█████     | 3637/7150 [46:51<42:30,  1.38it/s]
 51%|█████     | 3638/7150 [46:51<41:14,  1.42it/s]
 51%|█████     | 3639/7150 [46:52<40:21,  1.45it/s]
 51%|█████     | 3640/7150 [46:53<39:44,  1.47it/s]
                                                   
{'loss': '3.004', 'grad_norm': '0.2884', 'learning_rate': '0.0006', 'epoch': '25.46'}
+
 51%|█████     | 3640/7150 [46:53<39:44,  1.47it/s]
 51%|█████     | 3641/7150 [46:53<39:21,  1.49it/s]
 51%|█████     | 3642/7150 [46:54<39:03,  1.50it/s]
 51%|█████     | 3643/7150 [46:55<38:51,  1.50it/s]
 51%|█████     | 3644/7150 [46:55<38:41,  1.51it/s]
 51%|█████     | 3645/7150 [46:56<38:35,  1.51it/s]
 51%|█████     | 3646/7150 [46:56<38:28,  1.52it/s]
 51%|█████     | 3647/7150 [46:57<38:23,  1.52it/s]
 51%|█████     | 3648/7150 [46:58<38:19,  1.52it/s]
 51%|█████     | 3649/7150 [46:58<38:16,  1.52it/s]
 51%|█████     | 3650/7150 [46:59<38:15,  1.52it/s]
                                                   
{'loss': '3.025', 'grad_norm': '0.2589', 'learning_rate': '0.0006', 'epoch': '25.53'}
+
 51%|█████     | 3650/7150 [46:59<38:15,  1.52it/s]
 51%|█████     | 3651/7150 [47:00<44:34,  1.31it/s]
 51%|█████     | 3652/7150 [47:01<42:37,  1.37it/s]
 51%|█████     | 3653/7150 [47:01<41:17,  1.41it/s]
 51%|█████     | 3654/7150 [47:02<40:19,  1.44it/s]
 51%|█████     | 3655/7150 [47:03<39:40,  1.47it/s]
 51%|█████     | 3656/7150 [47:03<39:12,  1.49it/s]
 51%|█████     | 3657/7150 [47:04<38:52,  1.50it/s]
 51%|█████     | 3658/7150 [47:05<38:37,  1.51it/s]
 51%|█████     | 3659/7150 [47:05<38:30,  1.51it/s]
 51%|█████     | 3660/7150 [47:06<38:23,  1.51it/s]
                                                   
{'loss': '3.023', 'grad_norm': '0.2532', 'learning_rate': '0.0006', 'epoch': '25.6'}
+
 51%|█████     | 3660/7150 [47:06<38:23,  1.51it/s]
 51%|█████     | 3661/7150 [47:07<44:14,  1.31it/s]
 51%|█████     | 3662/7150 [47:08<42:21,  1.37it/s]
 51%|█████     | 3663/7150 [47:08<41:04,  1.41it/s]
 51%|█████     | 3664/7150 [47:09<40:10,  1.45it/s]
 51%|█████▏    | 3665/7150 [47:10<39:32,  1.47it/s]
 51%|█████▏    | 3666/7150 [47:10<39:04,  1.49it/s]
 51%|█████▏    | 3667/7150 [47:11<38:44,  1.50it/s]
 51%|█████▏    | 3668/7150 [47:12<38:31,  1.51it/s]
 51%|█████▏    | 3669/7150 [47:12<38:21,  1.51it/s]
 51%|█████▏    | 3670/7150 [47:13<38:13,  1.52it/s]
                                                   
{'loss': '3.027', 'grad_norm': '0.2613', 'learning_rate': '0.0006', 'epoch': '25.67'}
+
 51%|█████▏    | 3670/7150 [47:13<38:13,  1.52it/s]
 51%|█████▏    | 3671/7150 [47:14<38:09,  1.52it/s]
 51%|█████▏    | 3672/7150 [47:14<38:06,  1.52it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.73it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.87it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.64it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.05it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.74it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.58it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.44it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.37it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.32it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.29it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.27it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.26it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.25it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.25it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.23it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.24it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.24it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.24it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s]
                                                   
+
                                               
{'eval_loss': '3.802', 'eval_runtime': '3.26', 'eval_samples_per_second': '748.7', 'eval_steps_per_second': '11.96', 'epoch': '25.68'}
+
 51%|█████▏    | 3672/7150 [47:17<38:06,  1.52it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s][eval] epoch 25.68 step 3672 loss 3.8017 (best 3.7377 @ ep 15.609, 20/20 since best)
+
+
                                               [transformers] 
+
+Training completed. Do not forget to share your model on huggingface.co/models =)
+
+
+
                                                   
{'train_runtime': '2839', 'train_samples_per_second': '1287', 'train_steps_per_second': '2.518', 'train_loss': '3.79', 'epoch': '25.68'}
+
 51%|█████▏    | 3672/7150 [47:17<38:06,  1.52it/s]
 51%|█████▏    | 3672/7150 [47:17<44:48,  1.29it/s]
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+wandb: 
+wandb: 🚀 View run t2a_150M_C0_baseline at: https://wandb.ai/slowrun-wankers/data-aug-pretraining/runs/gc8jxphi
+wandb: Find logs at: wandb/run-20260818_213047-gc8jxphi/logs
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+[rank0]:[W818 22:18:51.870054853 ProcessGroupNCCL.cpp:1575] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator())
diff --git a/runs/t2a/t2a_19M_C0_baseline/checkpoint-576/chat_template.jinja b/runs/t2a/t2a_19M_C0_baseline/checkpoint-576/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..699ff8df401fe4788525e9c1f9b86a99eadd6230
--- /dev/null
+++ b/runs/t2a/t2a_19M_C0_baseline/checkpoint-576/chat_template.jinja
@@ -0,0 +1,85 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0].role == 'system' %}
+        {{- messages[0].content + '\n\n' }}
+    {%- endif %}
+    {{- "# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0].content + '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
+{%- for message in messages[::-1] %}
+    {%- set index = (messages|length - 1) - loop.index0 %}
+    {%- if ns.multi_step_tool and message.role == "user" and not(message.content.startswith('<tool_response>') and message.content.endswith('</tool_response>')) %}
+        {%- set ns.multi_step_tool = false %}
+        {%- set ns.last_query_index = index %}
+    {%- endif %}
+{%- endfor %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {%- set content = message.content %}
+        {%- set reasoning_content = '' %}
+        {%- if message.reasoning_content is defined and message.reasoning_content is not none %}
+            {%- set reasoning_content = message.reasoning_content %}
+        {%- else %}
+            {%- if '</think>' in message.content %}
+                {%- set content = message.content.split('</think>')[-1].lstrip('\n') %}
+                {%- set reasoning_content = message.content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
+            {%- endif %}
+        {%- endif %}
+        {%- if loop.index0 > ns.last_query_index %}
+            {%- if loop.last or (not loop.last and reasoning_content) %}
+                {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content.strip('\n') + '\n</think>\n\n' + content.lstrip('\n') }}
+            {%- else %}
+                {{- '<|im_start|>' + message.role + '\n' + content }}
+            {%- endif %}
+        {%- else %}
+            {{- '<|im_start|>' + message.role + '\n' + content }}
+        {%- endif %}
+        {%- if message.tool_calls %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if (loop.first and content) or (not loop.first) %}
+                    {{- '\n' }}
+                {%- endif %}
+                {%- if tool_call.function %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {{- '<tool_call>\n{"name": "' }}
+                {{- tool_call.name }}
+                {{- '", "arguments": ' }}
+                {%- if tool_call.arguments is string %}
+                    {{- tool_call.arguments }}
+                {%- else %}
+                    {{- tool_call.arguments | tojson }}
+                {%- endif %}
+                {{- '}\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.first or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+    {%- if enable_thinking is defined and enable_thinking is false %}
+        {{- '<think>\n\n</think>\n\n' }}
+    {%- endif %}
+{%- endif %}
\ No newline at end of file
diff --git a/runs/t2a/t2a_19M_C0_baseline/checkpoint-576/config.json b/runs/t2a/t2a_19M_C0_baseline/checkpoint-576/config.json
new file mode 100644
index 0000000000000000000000000000000000000000..f8ca584a0751672f280da7c54ff3916cdf24d6f4
--- /dev/null
+++ b/runs/t2a/t2a_19M_C0_baseline/checkpoint-576/config.json
@@ -0,0 +1,32 @@
+{
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": null,
+  "dtype": "float32",
+  "eos_token_id": 151645,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 512,
+  "initializer_range": 0.02,
+  "intermediate_size": 1536,
+  "max_position_embeddings": 2048,
+  "mlp_bias": false,
+  "model_type": "llama",
+  "num_attention_heads": 4,
+  "num_hidden_layers": 20,
+  "num_key_value_heads": 4,
+  "pad_token_id": 151645,
+  "pretraining_tp": 1,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 10000.0,
+    "rope_type": "default"
+  },
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": false,
+  "vocab_size": 151671
+}
diff --git a/runs/t2a/t2a_19M_C0_baseline/checkpoint-576/generation_config.json b/runs/t2a/t2a_19M_C0_baseline/checkpoint-576/generation_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..8c4242f32d40df256857f87264f646130f684e33
--- /dev/null
+++ b/runs/t2a/t2a_19M_C0_baseline/checkpoint-576/generation_config.json
@@ -0,0 +1,11 @@
+{
+  "_from_model_config": true,
+  "eos_token_id": [
+    151645
+  ],
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "transformers_version": "5.15.0",
+  "use_cache": true
+}
diff --git a/runs/t2a/t2a_19M_C0_baseline/checkpoint-576/tokenizer_config.json b/runs/t2a/t2a_19M_C0_baseline/checkpoint-576/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..839a3743177a3ed77e1ffbae67520e8500ca0532
--- /dev/null
+++ b/runs/t2a/t2a_19M_C0_baseline/checkpoint-576/tokenizer_config.json
@@ -0,0 +1,19 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|l2r_pred|>",
+    "<|r2l_pred|>"
+  ],
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 131072,
+  "pad_token": "<|im_end|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}
diff --git a/runs/t2a/t2a_19M_C0_baseline/checkpoint-576/trainer_state.json b/runs/t2a/t2a_19M_C0_baseline/checkpoint-576/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..7fce671b1bdb8e6f3e7081e9d122a3516e0d8f61
--- /dev/null
+++ b/runs/t2a/t2a_19M_C0_baseline/checkpoint-576/trainer_state.json
@@ -0,0 +1,504 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 30.322147651006713,
+  "eval_steps": 72,
+  "global_step": 576,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.053691275167785234,
+      "grad_norm": 1.8142167329788208,
+      "learning_rate": 0.0,
+      "loss": 12.011314392089844,
+      "step": 1
+    },
+    {
+      "epoch": 0.5369127516778524,
+      "grad_norm": 1.7725874185562134,
+      "learning_rate": 5.399999999999999e-05,
+      "loss": 11.70957522922092,
+      "step": 10
+    },
+    {
+      "epoch": 1.0536912751677852,
+      "grad_norm": 1.685107707977295,
+      "learning_rate": 0.00011399999999999999,
+      "loss": 11.0172119140625,
+      "step": 20
+    },
+    {
+      "epoch": 1.5906040268456376,
+      "grad_norm": 1.6527771949768066,
+      "learning_rate": 0.00017399999999999997,
+      "loss": 10.258322143554688,
+      "step": 30
+    },
+    {
+      "epoch": 2.1073825503355703,
+      "grad_norm": 1.4742095470428467,
+      "learning_rate": 0.000234,
+      "loss": 9.209019470214844,
+      "step": 40
+    },
+    {
+      "epoch": 2.6442953020134228,
+      "grad_norm": 0.8990502953529358,
+      "learning_rate": 0.000294,
+      "loss": 8.233860778808594,
+      "step": 50
+    },
+    {
+      "epoch": 3.1610738255033555,
+      "grad_norm": 0.32120978832244873,
+      "learning_rate": 0.00035399999999999993,
+      "loss": 7.701448822021485,
+      "step": 60
+    },
+    {
+      "epoch": 3.697986577181208,
+      "grad_norm": 1.2037309408187866,
+      "learning_rate": 0.0004139999999999999,
+      "loss": 7.5403800964355465,
+      "step": 70
+    },
+    {
+      "epoch": 3.8053691275167782,
+      "eval_loss": 7.4664626121521,
+      "eval_runtime": 27.3161,
+      "eval_samples_per_second": 89.361,
+      "eval_steps_per_second": 1.428,
+      "step": 72
+    },
+    {
+      "epoch": 4.214765100671141,
+      "grad_norm": 0.43744996190071106,
+      "learning_rate": 0.000474,
+      "loss": 7.394523620605469,
+      "step": 80
+    },
+    {
+      "epoch": 4.751677852348993,
+      "grad_norm": 1.3217568397521973,
+      "learning_rate": 0.000534,
+      "loss": 7.174855804443359,
+      "step": 90
+    },
+    {
+      "epoch": 5.268456375838926,
+      "grad_norm": 0.5023097991943359,
+      "learning_rate": 0.0005939999999999999,
+      "loss": 7.081169891357422,
+      "step": 100
+    },
+    {
+      "epoch": 5.805369127516778,
+      "grad_norm": 0.30793970823287964,
+      "learning_rate": 0.0006,
+      "loss": 6.914817810058594,
+      "step": 110
+    },
+    {
+      "epoch": 6.322147651006711,
+      "grad_norm": 0.2508816123008728,
+      "learning_rate": 0.0006,
+      "loss": 6.747209167480468,
+      "step": 120
+    },
+    {
+      "epoch": 6.859060402684563,
+      "grad_norm": 0.6191902756690979,
+      "learning_rate": 0.0006,
+      "loss": 6.628437042236328,
+      "step": 130
+    },
+    {
+      "epoch": 7.375838926174497,
+      "grad_norm": 0.45684921741485596,
+      "learning_rate": 0.0006,
+      "loss": 6.524767303466797,
+      "step": 140
+    },
+    {
+      "epoch": 7.590604026845638,
+      "eval_loss": 6.43440580368042,
+      "eval_runtime": 3.2557,
+      "eval_samples_per_second": 749.764,
+      "eval_steps_per_second": 11.979,
+      "step": 144
+    },
+    {
+      "epoch": 7.912751677852349,
+      "grad_norm": 0.32593798637390137,
+      "learning_rate": 0.0006,
+      "loss": 6.398002624511719,
+      "step": 150
+    },
+    {
+      "epoch": 8.429530201342281,
+      "grad_norm": 0.49603816866874695,
+      "learning_rate": 0.0006,
+      "loss": 6.30065803527832,
+      "step": 160
+    },
+    {
+      "epoch": 8.966442953020135,
+      "grad_norm": 0.5765122771263123,
+      "learning_rate": 0.0006,
+      "loss": 6.225212097167969,
+      "step": 170
+    },
+    {
+      "epoch": 9.483221476510067,
+      "grad_norm": 0.35036981105804443,
+      "learning_rate": 0.0006,
+      "loss": 6.138700485229492,
+      "step": 180
+    },
+    {
+      "epoch": 10.0,
+      "grad_norm": 0.864263653755188,
+      "learning_rate": 0.0006,
+      "loss": 6.088552856445313,
+      "step": 190
+    },
+    {
+      "epoch": 10.536912751677852,
+      "grad_norm": 0.3657694160938263,
+      "learning_rate": 0.0006,
+      "loss": 6.047029876708985,
+      "step": 200
+    },
+    {
+      "epoch": 11.053691275167786,
+      "grad_norm": 0.32556647062301636,
+      "learning_rate": 0.0006,
+      "loss": 5.944820404052734,
+      "step": 210
+    },
+    {
+      "epoch": 11.375838926174497,
+      "eval_loss": 5.944965362548828,
+      "eval_runtime": 3.251,
+      "eval_samples_per_second": 750.841,
+      "eval_steps_per_second": 11.996,
+      "step": 216
+    },
+    {
+      "epoch": 11.590604026845638,
+      "grad_norm": 0.8797701001167297,
+      "learning_rate": 0.0006,
+      "loss": 5.853334426879883,
+      "step": 220
+    },
+    {
+      "epoch": 12.10738255033557,
+      "grad_norm": 0.7158633470535278,
+      "learning_rate": 0.0006,
+      "loss": 5.845497894287109,
+      "step": 230
+    },
+    {
+      "epoch": 12.644295302013422,
+      "grad_norm": 0.3370794355869293,
+      "learning_rate": 0.0006,
+      "loss": 5.740810775756836,
+      "step": 240
+    },
+    {
+      "epoch": 13.161073825503356,
+      "grad_norm": 1.0377658605575562,
+      "learning_rate": 0.0006,
+      "loss": 5.6702430725097654,
+      "step": 250
+    },
+    {
+      "epoch": 13.697986577181208,
+      "grad_norm": 0.6437246799468994,
+      "learning_rate": 0.0006,
+      "loss": 5.641211700439453,
+      "step": 260
+    },
+    {
+      "epoch": 14.21476510067114,
+      "grad_norm": 0.45554235577583313,
+      "learning_rate": 0.0006,
+      "loss": 5.558935928344726,
+      "step": 270
+    },
+    {
+      "epoch": 14.751677852348994,
+      "grad_norm": 0.7522556781768799,
+      "learning_rate": 0.0006,
+      "loss": 5.494405364990234,
+      "step": 280
+    },
+    {
+      "epoch": 15.161073825503356,
+      "eval_loss": 5.567083835601807,
+      "eval_runtime": 3.2645,
+      "eval_samples_per_second": 747.75,
+      "eval_steps_per_second": 11.947,
+      "step": 288
+    },
+    {
+      "epoch": 15.268456375838927,
+      "grad_norm": 0.8695115447044373,
+      "learning_rate": 0.0006,
+      "loss": 5.467164993286133,
+      "step": 290
+    },
+    {
+      "epoch": 15.805369127516778,
+      "grad_norm": 0.297993540763855,
+      "learning_rate": 0.0006,
+      "loss": 5.397716522216797,
+      "step": 300
+    },
+    {
+      "epoch": 16.322147651006713,
+      "grad_norm": 0.4352909326553345,
+      "learning_rate": 0.0006,
+      "loss": 5.3047119140625,
+      "step": 310
+    },
+    {
+      "epoch": 16.859060402684563,
+      "grad_norm": 0.6090894937515259,
+      "learning_rate": 0.0006,
+      "loss": 5.284146881103515,
+      "step": 320
+    },
+    {
+      "epoch": 17.375838926174495,
+      "grad_norm": 0.38252320885658264,
+      "learning_rate": 0.0006,
+      "loss": 5.229152679443359,
+      "step": 330
+    },
+    {
+      "epoch": 17.91275167785235,
+      "grad_norm": 0.5717241168022156,
+      "learning_rate": 0.0006,
+      "loss": 5.158631896972656,
+      "step": 340
+    },
+    {
+      "epoch": 18.42953020134228,
+      "grad_norm": 0.8796677589416504,
+      "learning_rate": 0.0006,
+      "loss": 5.094086837768555,
+      "step": 350
+    },
+    {
+      "epoch": 18.966442953020135,
+      "grad_norm": 0.5505955815315247,
+      "learning_rate": 0.0006,
+      "loss": 5.084417343139648,
+      "step": 360
+    },
+    {
+      "epoch": 18.966442953020135,
+      "eval_loss": 5.213481903076172,
+      "eval_runtime": 3.2663,
+      "eval_samples_per_second": 747.319,
+      "eval_steps_per_second": 11.94,
+      "step": 360
+    },
+    {
+      "epoch": 19.483221476510067,
+      "grad_norm": 0.3969738781452179,
+      "learning_rate": 0.0006,
+      "loss": 4.99476318359375,
+      "step": 370
+    },
+    {
+      "epoch": 20.0,
+      "grad_norm": 0.23706182837486267,
+      "learning_rate": 0.0006,
+      "loss": 4.9330909729003904,
+      "step": 380
+    },
+    {
+      "epoch": 20.536912751677853,
+      "grad_norm": 0.8291894197463989,
+      "learning_rate": 0.0006,
+      "loss": 4.859904861450195,
+      "step": 390
+    },
+    {
+      "epoch": 21.053691275167786,
+      "grad_norm": 0.483011931180954,
+      "learning_rate": 0.0006,
+      "loss": 4.925698089599609,
+      "step": 400
+    },
+    {
+      "epoch": 21.59060402684564,
+      "grad_norm": 0.4645936191082001,
+      "learning_rate": 0.0006,
+      "loss": 4.812581634521484,
+      "step": 410
+    },
+    {
+      "epoch": 22.107382550335572,
+      "grad_norm": 0.4653514623641968,
+      "learning_rate": 0.0006,
+      "loss": 4.752062225341797,
+      "step": 420
+    },
+    {
+      "epoch": 22.644295302013422,
+      "grad_norm": 0.45866936445236206,
+      "learning_rate": 0.0006,
+      "loss": 4.681341171264648,
+      "step": 430
+    },
+    {
+      "epoch": 22.751677852348994,
+      "eval_loss": 4.928530216217041,
+      "eval_runtime": 3.243,
+      "eval_samples_per_second": 752.706,
+      "eval_steps_per_second": 12.026,
+      "step": 432
+    },
+    {
+      "epoch": 23.161073825503355,
+      "grad_norm": 0.4152670204639435,
+      "learning_rate": 0.0006,
+      "loss": 4.620679473876953,
+      "step": 440
+    },
+    {
+      "epoch": 23.697986577181208,
+      "grad_norm": 0.8872674703598022,
+      "learning_rate": 0.0006,
+      "loss": 4.583133697509766,
+      "step": 450
+    },
+    {
+      "epoch": 24.21476510067114,
+      "grad_norm": 0.7488914132118225,
+      "learning_rate": 0.0006,
+      "loss": 4.62310905456543,
+      "step": 460
+    },
+    {
+      "epoch": 24.751677852348994,
+      "grad_norm": 0.404926061630249,
+      "learning_rate": 0.0006,
+      "loss": 4.546733856201172,
+      "step": 470
+    },
+    {
+      "epoch": 25.268456375838927,
+      "grad_norm": 0.31850552558898926,
+      "learning_rate": 0.0006,
+      "loss": 4.459835052490234,
+      "step": 480
+    },
+    {
+      "epoch": 25.80536912751678,
+      "grad_norm": 0.2806507647037506,
+      "learning_rate": 0.0006,
+      "loss": 4.37512435913086,
+      "step": 490
+    },
+    {
+      "epoch": 26.322147651006713,
+      "grad_norm": 0.5341393351554871,
+      "learning_rate": 0.0006,
+      "loss": 4.31119384765625,
+      "step": 500
+    },
+    {
+      "epoch": 26.536912751677853,
+      "eval_loss": 4.76123571395874,
+      "eval_runtime": 3.2635,
+      "eval_samples_per_second": 747.966,
+      "eval_steps_per_second": 11.95,
+      "step": 504
+    },
+    {
+      "epoch": 26.859060402684563,
+      "grad_norm": 1.043302297592163,
+      "learning_rate": 0.0006,
+      "loss": 4.311373138427735,
+      "step": 510
+    },
+    {
+      "epoch": 27.375838926174495,
+      "grad_norm": 0.9856887459754944,
+      "learning_rate": 0.0006,
+      "loss": 4.348341369628907,
+      "step": 520
+    },
+    {
+      "epoch": 27.91275167785235,
+      "grad_norm": 0.6892461180686951,
+      "learning_rate": 0.0006,
+      "loss": 4.312998962402344,
+      "step": 530
+    },
+    {
+      "epoch": 28.42953020134228,
+      "grad_norm": 0.3091641962528229,
+      "learning_rate": 0.0006,
+      "loss": 4.186044692993164,
+      "step": 540
+    },
+    {
+      "epoch": 28.966442953020135,
+      "grad_norm": 0.2504050135612488,
+      "learning_rate": 0.0006,
+      "loss": 4.133544540405273,
+      "step": 550
+    },
+    {
+      "epoch": 29.483221476510067,
+      "grad_norm": 0.5265411734580994,
+      "learning_rate": 0.0006,
+      "loss": 4.017610931396485,
+      "step": 560
+    },
+    {
+      "epoch": 30.0,
+      "grad_norm": 0.5696446299552917,
+      "learning_rate": 0.0006,
+      "loss": 4.023490905761719,
+      "step": 570
+    },
+    {
+      "epoch": 30.322147651006713,
+      "eval_loss": 4.713936805725098,
+      "eval_runtime": 3.2534,
+      "eval_samples_per_second": 750.303,
+      "eval_steps_per_second": 11.988,
+      "step": 576
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 7505,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 395,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": false,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2.4224477549376307e+17,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/runs/t2a/t2a_19M_C0_baseline/train.log b/runs/t2a/t2a_19M_C0_baseline/train.log
new file mode 100644
index 0000000000000000000000000000000000000000..d87c2fdafebfe9fd31f87f6eda06732b812a6ffb
--- /dev/null
+++ b/runs/t2a/t2a_19M_C0_baseline/train.log
@@ -0,0 +1,1990 @@
+W0819 17:24:01.513000 909438 torch/distributed/run.py:851] 
+W0819 17:24:01.513000 909438 torch/distributed/run.py:851] *****************************************
+W0819 17:24:01.513000 909438 torch/distributed/run.py:851] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+W0819 17:24:01.513000 909438 torch/distributed/run.py:851] *****************************************
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+total params: 145833984
+[dataset] loaded packed cache: 9536 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+total params: 145833984
+[dataset] loaded packed cache: 9536 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+total params: 145833984
+[dataset] loaded packed cache: 9536 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+total params: 145833984
+[dataset] loaded packed cache: 9536 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+total params: 145833984
+[dataset] loaded packed cache: 9536 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+total params: 145833984
+[dataset] loaded packed cache: 9536 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+total params: 145833984
+[dataset] loaded packed cache: 9536 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+total params: 145833984
+[dataset] loaded packed cache: 9536 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+[transformers] ***** Running training *****
+[transformers]   Num examples = 9,536
+[transformers]   Num Epochs = 395
+[transformers]   Num update steps per epoch = 19
+[transformers]   Instantaneous batch size per device = 8
+[transformers]   Total train batch size (w. parallel, distributed & accumulation) = 512
+[transformers]   Gradient Accumulation steps = 8
+[transformers]   Total optimization steps = 7,505
+[transformers]   Number of trainable parameters = 145,833,984
+wandb: WARNING Changes to your `wandb` environment variables will be ignored because your `wandb` session has already started. For more information on how to modify your settings with `wandb.init()` arguments, please refer to https://wandb.me/wandb-init.
+wandb: setting up run ws9il445
+wandb: Tracking run with wandb version 0.28.2
+wandb: Run data is saved locally in /home/ubuntu/workspace/data-augmentations-for-pretraining/wandb/run-20260819_172420-ws9il445
+wandb: Run `wandb offline` to turn off syncing.
+wandb: Syncing run t2a_19M_C0_baseline
+wandb: ⭐️ View project at https://wandb.ai/slowrun-wankers/data-aug-pretraining
+wandb: 🚀 View run at https://wandb.ai/slowrun-wankers/data-aug-pretraining/runs/ws9il445
+
  0%|          | 0/7505 [00:00<?, ?it/s]wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+[rank5]:W0819 17:24:42.304000 909567 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+[rank3]:W0819 17:24:42.505000 909565 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+[rank4]:W0819 17:24:42.669000 909566 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+[rank7]:W0819 17:24:42.715000 909569 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+[rank1]:W0819 17:24:42.734000 909563 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+[rank2]:W0819 17:24:42.821000 909564 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+[rank6]:W0819 17:24:42.969000 909568 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+[rank0]:W0819 17:24:46.019000 909562 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+
  0%|          | 1/7505 [00:41<86:53:28, 41.69s/it]
                                                   
{'loss': '12.01', 'grad_norm': '1.814', 'learning_rate': '0', 'epoch': '0.05369'}
+
  0%|          | 1/7505 [00:41<86:53:28, 41.69s/it]
  0%|          | 2/7505 [00:42<36:35:34, 17.56s/it]
  0%|          | 3/7505 [00:43<20:29:59,  9.84s/it]
  0%|          | 4/7505 [00:43<12:56:39,  6.21s/it]
  0%|          | 5/7505 [00:44<8:45:53,  4.21s/it] 
  0%|          | 6/7505 [00:44<6:14:43,  3.00s/it]
  0%|          | 7/7505 [00:45<4:38:47,  2.23s/it]
  0%|          | 8/7505 [00:46<3:35:54,  1.73s/it]
  0%|          | 9/7505 [00:46<2:53:46,  1.39s/it]
  0%|          | 10/7505 [00:47<2:25:12,  1.16s/it]
                                                   
{'loss': '11.71', 'grad_norm': '1.773', 'learning_rate': '5.4e-05', 'epoch': '0.5369'}
+
  0%|          | 10/7505 [00:47<2:25:12,  1.16s/it]
  0%|          | 11/7505 [00:48<2:05:38,  1.01s/it]
  0%|          | 12/7505 [00:48<1:52:11,  1.11it/s]
  0%|          | 13/7505 [00:49<1:42:48,  1.21it/s]
  0%|          | 14/7505 [00:50<1:36:19,  1.30it/s]
  0%|          | 15/7505 [00:50<1:31:47,  1.36it/s]
  0%|          | 16/7505 [00:51<1:28:35,  1.41it/s]
  0%|          | 17/7505 [00:52<1:26:27,  1.44it/s]
  0%|          | 18/7505 [00:52<1:24:50,  1.47it/s]
  0%|          | 19/7505 [00:53<1:14:45,  1.67it/s]
  0%|          | 20/7505 [00:53<1:19:56,  1.56it/s]
                                                   
{'loss': '11.02', 'grad_norm': '1.685', 'learning_rate': '0.000114', 'epoch': '1.054'}
+
  0%|          | 20/7505 [00:53<1:19:56,  1.56it/s]
  0%|          | 21/7505 [00:54<1:20:24,  1.55it/s]
  0%|          | 22/7505 [00:55<1:20:42,  1.55it/s]
  0%|          | 23/7505 [00:55<1:20:53,  1.54it/s]
  0%|          | 24/7505 [00:56<1:21:02,  1.54it/s]
  0%|          | 25/7505 [00:57<1:21:12,  1.54it/s]
  0%|          | 26/7505 [00:57<1:21:17,  1.53it/s]
  0%|          | 27/7505 [00:58<1:21:19,  1.53it/s]
  0%|          | 28/7505 [00:59<1:21:19,  1.53it/s]
  0%|          | 29/7505 [00:59<1:21:20,  1.53it/s]
  0%|          | 30/7505 [01:00<1:21:20,  1.53it/s]
                                                   
{'loss': '10.26', 'grad_norm': '1.653', 'learning_rate': '0.000174', 'epoch': '1.591'}
+
  0%|          | 30/7505 [01:00<1:21:20,  1.53it/s]
  0%|          | 31/7505 [01:01<1:21:23,  1.53it/s]
  0%|          | 32/7505 [01:01<1:21:19,  1.53it/s]
  0%|          | 33/7505 [01:02<1:21:16,  1.53it/s]
  0%|          | 34/7505 [01:03<1:21:19,  1.53it/s]
  0%|          | 35/7505 [01:03<1:21:26,  1.53it/s]
  0%|          | 36/7505 [01:04<1:21:30,  1.53it/s]
  0%|          | 37/7505 [01:05<1:21:17,  1.53it/s]
  1%|          | 38/7505 [01:05<1:12:08,  1.73it/s]
  1%|          | 39/7505 [01:06<1:17:36,  1.60it/s]
  1%|          | 40/7505 [01:06<1:18:36,  1.58it/s]
                                                   
{'loss': '9.209', 'grad_norm': '1.474', 'learning_rate': '0.000234', 'epoch': '2.107'}
+
  1%|          | 40/7505 [01:06<1:18:36,  1.58it/s]
  1%|          | 41/7505 [01:07<1:19:16,  1.57it/s]
  1%|          | 42/7505 [01:08<1:19:43,  1.56it/s]
  1%|          | 43/7505 [01:08<1:20:02,  1.55it/s]
  1%|          | 44/7505 [01:09<1:20:16,  1.55it/s]
  1%|          | 45/7505 [01:10<1:20:23,  1.55it/s]
  1%|          | 46/7505 [01:10<1:20:32,  1.54it/s]
  1%|          | 47/7505 [01:11<1:20:35,  1.54it/s]
  1%|          | 48/7505 [01:12<1:20:40,  1.54it/s]
  1%|          | 49/7505 [01:12<1:20:40,  1.54it/s]
  1%|          | 50/7505 [01:13<1:20:44,  1.54it/s]
                                                   
{'loss': '8.234', 'grad_norm': '0.8991', 'learning_rate': '0.000294', 'epoch': '2.644'}
+
  1%|          | 50/7505 [01:13<1:20:44,  1.54it/s]
  1%|          | 51/7505 [01:13<1:20:45,  1.54it/s]
  1%|          | 52/7505 [01:14<1:20:45,  1.54it/s]
  1%|          | 53/7505 [01:15<1:20:50,  1.54it/s]
  1%|          | 54/7505 [01:15<1:20:45,  1.54it/s]
  1%|          | 55/7505 [01:16<1:20:48,  1.54it/s]
  1%|          | 56/7505 [01:17<1:20:41,  1.54it/s]
  1%|          | 57/7505 [01:17<1:11:44,  1.73it/s]
  1%|          | 58/7505 [01:18<1:18:04,  1.59it/s]
  1%|          | 59/7505 [01:19<1:18:55,  1.57it/s]
  1%|          | 60/7505 [01:19<1:19:27,  1.56it/s]
                                                   
{'loss': '7.701', 'grad_norm': '0.3212', 'learning_rate': '0.000354', 'epoch': '3.161'}
+
  1%|          | 60/7505 [01:19<1:19:27,  1.56it/s]
  1%|          | 61/7505 [01:20<1:19:51,  1.55it/s]
  1%|          | 62/7505 [01:20<1:20:06,  1.55it/s]
  1%|          | 63/7505 [01:21<1:20:15,  1.55it/s]
  1%|          | 64/7505 [01:22<1:20:24,  1.54it/s]
  1%|          | 65/7505 [01:22<1:20:28,  1.54it/s]
  1%|          | 66/7505 [01:23<1:20:33,  1.54it/s]
  1%|          | 67/7505 [01:24<1:20:37,  1.54it/s]
  1%|          | 68/7505 [01:24<1:20:38,  1.54it/s]
  1%|          | 69/7505 [01:25<1:20:41,  1.54it/s]
  1%|          | 70/7505 [01:26<1:20:44,  1.53it/s]
                                                   
{'loss': '7.54', 'grad_norm': '1.204', 'learning_rate': '0.000414', 'epoch': '3.698'}
+
  1%|          | 70/7505 [01:26<1:20:44,  1.53it/s]
  1%|          | 71/7505 [01:26<1:20:51,  1.53it/s]
  1%|          | 72/7505 [01:27<1:20:49,  1.53it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 29.09it/s]
+
 15%|█▌        | 6/39 [00:00<00:02, 16.01it/s]
+
 21%|██        | 8/39 [00:00<00:02, 14.41it/s]
+
 26%|██▌       | 10/39 [00:00<00:02, 13.58it/s]
+
 31%|███       | 12/39 [00:00<00:02, 13.11it/s]
+
 36%|███▌      | 14/39 [00:01<00:01, 12.81it/s]
+
 41%|████      | 16/39 [00:01<00:01, 12.62it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 12.48it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 12.41it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 12.35it/s]
+
 62%|██████▏   | 24/39 [00:01<00:01, 12.32it/s]
+
 67%|██████▋   | 26/39 [00:01<00:01, 12.29it/s]
+
 72%|███████▏  | 28/39 [00:02<00:00, 12.27it/s]
+
 77%|███████▋  | 30/39 [00:02<00:00, 12.25it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 12.23it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 12.24it/s]
+
 92%|█████████▏| 36/39 [00:02<00:00, 12.24it/s]
+
 97%|█████████▋| 38/39 [00:02<00:00, 12.24it/s]
                                                   
+
                                               
{'eval_loss': '7.466', 'eval_runtime': '27.32', 'eval_samples_per_second': '89.36', 'eval_steps_per_second': '1.428', 'epoch': '3.805'}
+
  1%|          | 72/7505 [01:54<1:20:49,  1.53it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C0_baseline/checkpoint-72
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C0_baseline/checkpoint-72/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C0_baseline/checkpoint-72/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.66it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.66it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C0_baseline/checkpoint-72/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C0_baseline/checkpoint-72/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C0_baseline/checkpoint-72/tokenizer_config.json
+[eval] epoch 3.805 step 72 loss 7.4665 (best 7.4665 @ ep 3.805, 0/20 since best)
+
+
                                               
  1%|          | 73/7505 [01:57<19:29:58,  9.45s/it]
  1%|          | 74/7505 [01:58<14:03:15,  6.81s/it]
  1%|          | 75/7505 [01:58<10:14:21,  4.96s/it]
  1%|          | 76/7505 [01:59<7:25:15,  3.60s/it] 
  1%|          | 77/7505 [01:59<5:38:49,  2.74s/it]
  1%|          | 78/7505 [02:00<4:21:26,  2.11s/it]
  1%|          | 79/7505 [02:01<3:27:12,  1.67s/it]
  1%|          | 80/7505 [02:01<2:49:18,  1.37s/it]
                                                   
{'loss': '7.395', 'grad_norm': '0.4374', 'learning_rate': '0.000474', 'epoch': '4.215'}
+
  1%|          | 80/7505 [02:01<2:49:18,  1.37s/it]
  1%|          | 81/7505 [02:02<2:22:48,  1.15s/it]
  1%|          | 82/7505 [02:03<2:04:10,  1.00s/it]
  1%|          | 83/7505 [02:03<1:51:09,  1.11it/s]
  1%|          | 84/7505 [02:04<1:42:05,  1.21it/s]
  1%|          | 85/7505 [02:05<1:35:39,  1.29it/s]
  1%|          | 86/7505 [02:05<1:31:10,  1.36it/s]
  1%|          | 87/7505 [02:06<1:28:04,  1.40it/s]
  1%|          | 88/7505 [02:07<1:25:52,  1.44it/s]
  1%|          | 89/7505 [02:07<1:24:20,  1.47it/s]
  1%|          | 90/7505 [02:08<1:23:16,  1.48it/s]
                                                   
{'loss': '7.175', 'grad_norm': '1.322', 'learning_rate': '0.000534', 'epoch': '4.752'}
+
  1%|          | 90/7505 [02:08<1:23:16,  1.48it/s]
  1%|          | 91/7505 [02:09<1:22:30,  1.50it/s]
  1%|          | 92/7505 [02:09<1:21:59,  1.51it/s]
  1%|          | 93/7505 [02:10<1:21:38,  1.51it/s]
  1%|▏         | 94/7505 [02:11<1:21:24,  1.52it/s]
  1%|▏         | 95/7505 [02:11<1:12:12,  1.71it/s]
  1%|▏         | 96/7505 [02:12<1:17:47,  1.59it/s]
  1%|▏         | 97/7505 [02:12<1:18:44,  1.57it/s]
  1%|▏         | 98/7505 [02:13<1:19:27,  1.55it/s]
  1%|▏         | 99/7505 [02:14<1:19:49,  1.55it/s]
  1%|▏         | 100/7505 [02:14<1:20:02,  1.54it/s]
                                                    
{'loss': '7.081', 'grad_norm': '0.5023', 'learning_rate': '0.000594', 'epoch': '5.268'}
+
  1%|▏         | 100/7505 [02:14<1:20:02,  1.54it/s]
  1%|▏         | 101/7505 [02:15<1:20:19,  1.54it/s]
  1%|▏         | 102/7505 [02:16<1:20:22,  1.53it/s]
  1%|▏         | 103/7505 [02:16<1:20:28,  1.53it/s]
  1%|▏         | 104/7505 [02:17<1:20:29,  1.53it/s]
  1%|▏         | 105/7505 [02:18<1:20:33,  1.53it/s]
  1%|▏         | 106/7505 [02:18<1:20:35,  1.53it/s]
  1%|▏         | 107/7505 [02:19<1:20:35,  1.53it/s]
  1%|▏         | 108/7505 [02:20<1:20:41,  1.53it/s]
  1%|▏         | 109/7505 [02:20<1:20:44,  1.53it/s]
  1%|▏         | 110/7505 [02:21<1:20:44,  1.53it/s]
                                                    
{'loss': '6.915', 'grad_norm': '0.3079', 'learning_rate': '0.0006', 'epoch': '5.805'}
+
  1%|▏         | 110/7505 [02:21<1:20:44,  1.53it/s]
  1%|▏         | 111/7505 [02:21<1:20:39,  1.53it/s]
  1%|▏         | 112/7505 [02:22<1:20:41,  1.53it/s]
  2%|▏         | 113/7505 [02:23<1:20:36,  1.53it/s]
  2%|▏         | 114/7505 [02:23<1:11:36,  1.72it/s]
  2%|▏         | 115/7505 [02:24<1:17:15,  1.59it/s]
  2%|▏         | 116/7505 [02:25<1:18:14,  1.57it/s]
  2%|▏         | 117/7505 [02:25<1:18:59,  1.56it/s]
  2%|▏         | 118/7505 [02:26<1:19:29,  1.55it/s]
  2%|▏         | 119/7505 [02:27<1:19:51,  1.54it/s]
  2%|▏         | 120/7505 [02:27<1:20:04,  1.54it/s]
                                                    
{'loss': '6.747', 'grad_norm': '0.2509', 'learning_rate': '0.0006', 'epoch': '6.322'}
+
  2%|▏         | 120/7505 [02:27<1:20:04,  1.54it/s]
  2%|▏         | 121/7505 [02:28<1:20:14,  1.53it/s]
  2%|▏         | 122/7505 [02:29<1:20:18,  1.53it/s]
  2%|▏         | 123/7505 [02:29<1:20:25,  1.53it/s]
  2%|▏         | 124/7505 [02:30<1:20:33,  1.53it/s]
  2%|▏         | 125/7505 [02:31<1:20:34,  1.53it/s]
  2%|▏         | 126/7505 [02:31<1:20:32,  1.53it/s]
  2%|▏         | 127/7505 [02:32<1:20:33,  1.53it/s]
  2%|▏         | 128/7505 [02:32<1:20:35,  1.53it/s]
  2%|▏         | 129/7505 [02:33<1:20:34,  1.53it/s]
  2%|▏         | 130/7505 [02:34<1:20:34,  1.53it/s]
                                                    
{'loss': '6.628', 'grad_norm': '0.6192', 'learning_rate': '0.0006', 'epoch': '6.859'}
+
  2%|▏         | 130/7505 [02:34<1:20:34,  1.53it/s]
  2%|▏         | 131/7505 [02:34<1:20:38,  1.52it/s]
  2%|▏         | 132/7505 [02:35<1:20:32,  1.53it/s]
  2%|▏         | 133/7505 [02:36<1:11:34,  1.72it/s]
  2%|▏         | 134/7505 [02:36<1:17:16,  1.59it/s]
  2%|▏         | 135/7505 [02:37<1:18:12,  1.57it/s]
  2%|▏         | 136/7505 [02:38<1:18:54,  1.56it/s]
  2%|▏         | 137/7505 [02:38<1:19:21,  1.55it/s]
  2%|▏         | 138/7505 [02:39<1:19:38,  1.54it/s]
  2%|▏         | 139/7505 [02:40<1:19:51,  1.54it/s]
  2%|▏         | 140/7505 [02:40<1:20:00,  1.53it/s]
                                                    
{'loss': '6.525', 'grad_norm': '0.4568', 'learning_rate': '0.0006', 'epoch': '7.376'}
+
  2%|▏         | 140/7505 [02:40<1:20:00,  1.53it/s]
  2%|▏         | 141/7505 [02:41<1:20:11,  1.53it/s]
  2%|▏         | 142/7505 [02:41<1:20:11,  1.53it/s]
  2%|▏         | 143/7505 [02:42<1:20:12,  1.53it/s]
  2%|▏         | 144/7505 [02:43<1:20:13,  1.53it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.48it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.75it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.60it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.05it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.71it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.54it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.42it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.36it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.32it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.28it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.28it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.26it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.26it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.25it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.24it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.20it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.23it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.23it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.22it/s]
                                                    
+
                                               
{'eval_loss': '6.434', 'eval_runtime': '3.256', 'eval_samples_per_second': '749.8', 'eval_steps_per_second': '11.98', 'epoch': '7.591'}
+
  2%|▏         | 144/7505 [02:46<1:20:13,  1.53it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.22it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C0_baseline/checkpoint-144
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C0_baseline/checkpoint-144/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C0_baseline/checkpoint-144/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.93it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.93it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C0_baseline/checkpoint-144/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C0_baseline/checkpoint-144/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C0_baseline/checkpoint-144/tokenizer_config.json
+[eval] epoch 7.591 step 144 loss 6.4344 (best 6.4344 @ ep 7.591, 0/20 since best)
+
+
                                               
  2%|▏         | 145/7505 [02:49<4:35:37,  2.25s/it]
  2%|▏         | 146/7505 [02:49<3:36:58,  1.77s/it]
  2%|▏         | 147/7505 [02:50<2:55:57,  1.43s/it]
  2%|▏         | 148/7505 [02:51<2:27:12,  1.20s/it]
  2%|▏         | 149/7505 [02:51<2:07:06,  1.04s/it]
  2%|▏         | 150/7505 [02:52<1:53:05,  1.08it/s]
                                                    
{'loss': '6.398', 'grad_norm': '0.3259', 'learning_rate': '0.0006', 'epoch': '7.913'}
+
  2%|▏         | 150/7505 [02:52<1:53:05,  1.08it/s]
  2%|▏         | 151/7505 [02:53<1:43:11,  1.19it/s]
  2%|▏         | 152/7505 [02:53<1:27:20,  1.40it/s]
  2%|▏         | 153/7505 [02:54<1:27:35,  1.40it/s]
  2%|▏         | 154/7505 [02:54<1:25:24,  1.43it/s]
  2%|▏         | 155/7505 [02:55<1:23:47,  1.46it/s]
  2%|▏         | 156/7505 [02:56<1:22:41,  1.48it/s]
  2%|▏         | 157/7505 [02:56<1:21:54,  1.50it/s]
  2%|▏         | 158/7505 [02:57<1:21:31,  1.50it/s]
  2%|▏         | 159/7505 [02:58<1:21:04,  1.51it/s]
  2%|▏         | 160/7505 [02:58<1:20:47,  1.52it/s]
                                                    
{'loss': '6.301', 'grad_norm': '0.496', 'learning_rate': '0.0006', 'epoch': '8.43'}
+
  2%|▏         | 160/7505 [02:58<1:20:47,  1.52it/s]
  2%|▏         | 161/7505 [02:59<1:20:33,  1.52it/s]
  2%|▏         | 162/7505 [03:00<1:20:32,  1.52it/s]
  2%|▏         | 163/7505 [03:00<1:20:25,  1.52it/s]
  2%|▏         | 164/7505 [03:01<1:20:18,  1.52it/s]
  2%|▏         | 165/7505 [03:02<1:20:10,  1.53it/s]
  2%|▏         | 166/7505 [03:02<1:20:10,  1.53it/s]
  2%|▏         | 167/7505 [03:03<1:20:07,  1.53it/s]
  2%|▏         | 168/7505 [03:04<1:20:05,  1.53it/s]
  2%|▏         | 169/7505 [03:04<1:20:01,  1.53it/s]
  2%|▏         | 170/7505 [03:05<1:19:58,  1.53it/s]
                                                    
{'loss': '6.225', 'grad_norm': '0.5765', 'learning_rate': '0.0006', 'epoch': '8.966'}
+
  2%|▏         | 170/7505 [03:05<1:19:58,  1.53it/s]
  2%|▏         | 171/7505 [03:05<1:11:08,  1.72it/s]
  2%|▏         | 172/7505 [03:06<1:16:18,  1.60it/s]
  2%|▏         | 173/7505 [03:07<1:17:23,  1.58it/s]
  2%|▏         | 174/7505 [03:07<1:18:09,  1.56it/s]
  2%|▏         | 175/7505 [03:08<1:18:39,  1.55it/s]
  2%|▏         | 176/7505 [03:09<1:19:02,  1.55it/s]
  2%|▏         | 177/7505 [03:09<1:19:17,  1.54it/s]
  2%|▏         | 178/7505 [03:10<1:19:33,  1.54it/s]
  2%|▏         | 179/7505 [03:11<1:19:37,  1.53it/s]
  2%|▏         | 180/7505 [03:11<1:19:40,  1.53it/s]
                                                    
{'loss': '6.139', 'grad_norm': '0.3504', 'learning_rate': '0.0006', 'epoch': '9.483'}
+
  2%|▏         | 180/7505 [03:11<1:19:40,  1.53it/s]
  2%|▏         | 181/7505 [03:12<1:19:46,  1.53it/s]
  2%|▏         | 182/7505 [03:13<1:19:52,  1.53it/s]
  2%|▏         | 183/7505 [03:13<1:19:49,  1.53it/s]
  2%|▏         | 184/7505 [03:14<1:19:49,  1.53it/s]
  2%|▏         | 185/7505 [03:15<1:19:50,  1.53it/s]
  2%|▏         | 186/7505 [03:15<1:19:49,  1.53it/s]
  2%|▏         | 187/7505 [03:16<1:19:48,  1.53it/s]
  3%|▎         | 188/7505 [03:17<1:19:45,  1.53it/s]
  3%|▎         | 189/7505 [03:17<1:19:47,  1.53it/s]
  3%|▎         | 190/7505 [03:18<1:10:52,  1.72it/s]
                                                    
{'loss': '6.089', 'grad_norm': '0.8643', 'learning_rate': '0.0006', 'epoch': '10'}
+
  3%|▎         | 190/7505 [03:18<1:10:52,  1.72it/s]
  3%|▎         | 191/7505 [03:18<1:16:35,  1.59it/s]
  3%|▎         | 192/7505 [03:19<1:17:31,  1.57it/s]
  3%|▎         | 193/7505 [03:20<1:18:11,  1.56it/s]
  3%|▎         | 194/7505 [03:20<1:18:39,  1.55it/s]
  3%|▎         | 195/7505 [03:21<1:18:55,  1.54it/s]
  3%|▎         | 196/7505 [03:22<1:19:12,  1.54it/s]
  3%|▎         | 197/7505 [03:22<1:19:21,  1.53it/s]
  3%|▎         | 198/7505 [03:23<1:19:27,  1.53it/s]
  3%|▎         | 199/7505 [03:24<1:19:30,  1.53it/s]
  3%|▎         | 200/7505 [03:24<1:19:36,  1.53it/s]
                                                    
{'loss': '6.047', 'grad_norm': '0.3658', 'learning_rate': '0.0006', 'epoch': '10.54'}
+
  3%|▎         | 200/7505 [03:24<1:19:36,  1.53it/s]
  3%|▎         | 201/7505 [03:25<1:19:37,  1.53it/s]
  3%|▎         | 202/7505 [03:26<1:19:40,  1.53it/s]
  3%|▎         | 203/7505 [03:26<1:19:37,  1.53it/s]
  3%|▎         | 204/7505 [03:27<1:19:40,  1.53it/s]
  3%|▎         | 205/7505 [03:28<1:19:36,  1.53it/s]
  3%|▎         | 206/7505 [03:28<1:19:35,  1.53it/s]
  3%|▎         | 207/7505 [03:29<1:19:34,  1.53it/s]
  3%|▎         | 208/7505 [03:29<1:19:31,  1.53it/s]
  3%|▎         | 209/7505 [03:30<1:10:40,  1.72it/s]
  3%|▎         | 210/7505 [03:31<1:15:52,  1.60it/s]
                                                    
{'loss': '5.945', 'grad_norm': '0.3256', 'learning_rate': '0.0006', 'epoch': '11.05'}
+
  3%|▎         | 210/7505 [03:31<1:15:52,  1.60it/s]
  3%|▎         | 211/7505 [03:31<1:16:57,  1.58it/s]
  3%|▎         | 212/7505 [03:32<1:17:45,  1.56it/s]
  3%|▎         | 213/7505 [03:33<1:18:15,  1.55it/s]
  3%|▎         | 214/7505 [03:33<1:18:39,  1.54it/s]
  3%|▎         | 215/7505 [03:34<1:18:53,  1.54it/s]
  3%|▎         | 216/7505 [03:35<1:19:03,  1.54it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.32it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.74it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.54it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.03it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.72it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.56it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.43it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.38it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.31it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.30it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.28it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.26it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.25it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.24it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.25it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.24it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.24it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.24it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s]
                                                    
+
                                               
{'eval_loss': '5.945', 'eval_runtime': '3.251', 'eval_samples_per_second': '750.8', 'eval_steps_per_second': '12', 'epoch': '11.38'}
+
  3%|▎         | 216/7505 [03:38<1:19:03,  1.54it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C0_baseline/checkpoint-216
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C0_baseline/checkpoint-216/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C0_baseline/checkpoint-216/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.86it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.86it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C0_baseline/checkpoint-216/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C0_baseline/checkpoint-216/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C0_baseline/checkpoint-216/tokenizer_config.json
+[eval] epoch 11.376 step 216 loss 5.9450 (best 5.9450 @ ep 11.376, 0/20 since best)
+
+
                                               
  3%|▎         | 217/7505 [03:41<4:32:45,  2.25s/it]
  3%|▎         | 218/7505 [03:41<3:34:42,  1.77s/it]
  3%|▎         | 219/7505 [03:42<2:54:05,  1.43s/it]
  3%|▎         | 220/7505 [03:42<2:25:38,  1.20s/it]
                                                    
{'loss': '5.853', 'grad_norm': '0.8798', 'learning_rate': '0.0006', 'epoch': '11.59'}
+
  3%|▎         | 220/7505 [03:42<2:25:38,  1.20s/it]
  3%|▎         | 221/7505 [03:43<2:05:48,  1.04s/it]
  3%|▎         | 222/7505 [03:44<1:51:49,  1.09it/s]
  3%|▎         | 223/7505 [03:44<1:42:02,  1.19it/s]
  3%|▎         | 224/7505 [03:45<1:35:17,  1.27it/s]
  3%|▎         | 225/7505 [03:46<1:30:27,  1.34it/s]
  3%|▎         | 226/7505 [03:46<1:27:04,  1.39it/s]
  3%|▎         | 227/7505 [03:47<1:24:48,  1.43it/s]
  3%|▎         | 228/7505 [03:47<1:14:18,  1.63it/s]
  3%|▎         | 229/7505 [03:48<1:18:36,  1.54it/s]
  3%|▎         | 230/7505 [03:49<1:18:53,  1.54it/s]
                                                    
{'loss': '5.845', 'grad_norm': '0.7159', 'learning_rate': '0.0006', 'epoch': '12.11'}
+
  3%|▎         | 230/7505 [03:49<1:18:53,  1.54it/s]
  3%|▎         | 231/7505 [03:50<1:19:00,  1.53it/s]
  3%|▎         | 232/7505 [03:50<1:19:08,  1.53it/s]
  3%|▎         | 233/7505 [03:51<1:19:14,  1.53it/s]
  3%|▎         | 234/7505 [03:51<1:19:19,  1.53it/s]
  3%|▎         | 235/7505 [03:52<1:19:21,  1.53it/s]
  3%|▎         | 236/7505 [03:53<1:19:20,  1.53it/s]
  3%|▎         | 237/7505 [03:53<1:19:20,  1.53it/s]
  3%|▎         | 238/7505 [03:54<1:19:19,  1.53it/s]
  3%|▎         | 239/7505 [03:55<1:19:24,  1.53it/s]
  3%|▎         | 240/7505 [03:55<1:19:20,  1.53it/s]
                                                    
{'loss': '5.741', 'grad_norm': '0.3371', 'learning_rate': '0.0006', 'epoch': '12.64'}
+
  3%|▎         | 240/7505 [03:55<1:19:20,  1.53it/s]
  3%|▎         | 241/7505 [03:56<1:19:20,  1.53it/s]
  3%|▎         | 242/7505 [03:57<1:19:18,  1.53it/s]
  3%|▎         | 243/7505 [03:57<1:19:22,  1.52it/s]
  3%|▎         | 244/7505 [03:58<1:19:21,  1.53it/s]
  3%|▎         | 245/7505 [03:59<1:19:18,  1.53it/s]
  3%|▎         | 246/7505 [03:59<1:19:13,  1.53it/s]
  3%|▎         | 247/7505 [04:00<1:10:27,  1.72it/s]
  3%|▎         | 248/7505 [04:00<1:15:48,  1.60it/s]
  3%|▎         | 249/7505 [04:01<1:16:48,  1.57it/s]
  3%|▎         | 250/7505 [04:02<1:17:34,  1.56it/s]
                                                    
{'loss': '5.67', 'grad_norm': '1.038', 'learning_rate': '0.0006', 'epoch': '13.16'}
+
  3%|▎         | 250/7505 [04:02<1:17:34,  1.56it/s]
  3%|▎         | 251/7505 [04:02<1:18:03,  1.55it/s]
  3%|▎         | 252/7505 [04:03<1:18:23,  1.54it/s]
  3%|▎         | 253/7505 [04:04<1:18:36,  1.54it/s]
  3%|▎         | 254/7505 [04:04<1:18:47,  1.53it/s]
  3%|▎         | 255/7505 [04:05<1:18:54,  1.53it/s]
  3%|▎         | 256/7505 [04:06<1:18:59,  1.53it/s]
  3%|▎         | 257/7505 [04:06<1:18:59,  1.53it/s]
  3%|▎         | 258/7505 [04:07<1:19:03,  1.53it/s]
  3%|▎         | 259/7505 [04:08<1:19:06,  1.53it/s]
  3%|▎         | 260/7505 [04:08<1:19:04,  1.53it/s]
                                                    
{'loss': '5.641', 'grad_norm': '0.6437', 'learning_rate': '0.0006', 'epoch': '13.7'}
+
  3%|▎         | 260/7505 [04:08<1:19:04,  1.53it/s]
  3%|▎         | 261/7505 [04:09<1:19:09,  1.53it/s]
  3%|▎         | 262/7505 [04:10<1:19:04,  1.53it/s]
  4%|▎         | 263/7505 [04:10<1:19:07,  1.53it/s]
  4%|▎         | 264/7505 [04:11<1:19:06,  1.53it/s]
  4%|▎         | 265/7505 [04:12<1:19:02,  1.53it/s]
  4%|▎         | 266/7505 [04:12<1:10:13,  1.72it/s]
  4%|▎         | 267/7505 [04:13<1:15:17,  1.60it/s]
  4%|▎         | 268/7505 [04:13<1:16:27,  1.58it/s]
  4%|▎         | 269/7505 [04:14<1:17:11,  1.56it/s]
  4%|▎         | 270/7505 [04:15<1:17:43,  1.55it/s]
                                                    
{'loss': '5.559', 'grad_norm': '0.4555', 'learning_rate': '0.0006', 'epoch': '14.21'}
+
  4%|▎         | 270/7505 [04:15<1:17:43,  1.55it/s]
  4%|▎         | 271/7505 [04:15<1:18:06,  1.54it/s]
  4%|▎         | 272/7505 [04:16<1:18:21,  1.54it/s]
  4%|▎         | 273/7505 [04:17<1:18:29,  1.54it/s]
  4%|▎         | 274/7505 [04:18<1:35:03,  1.27it/s]
  4%|▎         | 275/7505 [04:18<1:30:12,  1.34it/s]
  4%|▎         | 276/7505 [04:20<1:42:27,  1.18it/s]
  4%|▎         | 277/7505 [04:20<1:35:28,  1.26it/s]
  4%|▎         | 278/7505 [04:21<1:30:29,  1.33it/s]
  4%|▎         | 279/7505 [04:22<1:26:59,  1.38it/s]
  4%|▎         | 280/7505 [04:22<1:24:31,  1.42it/s]
                                                    
{'loss': '5.494', 'grad_norm': '0.7523', 'learning_rate': '0.0006', 'epoch': '14.75'}
+
  4%|▎         | 280/7505 [04:22<1:24:31,  1.42it/s]
  4%|▎         | 281/7505 [04:23<1:22:49,  1.45it/s]
  4%|▍         | 282/7505 [04:23<1:21:39,  1.47it/s]
  4%|▍         | 283/7505 [04:24<1:20:45,  1.49it/s]
  4%|▍         | 284/7505 [04:25<1:20:05,  1.50it/s]
  4%|▍         | 285/7505 [04:25<1:10:58,  1.70it/s]
  4%|▍         | 286/7505 [04:26<1:15:36,  1.59it/s]
  4%|▍         | 287/7505 [04:27<1:16:32,  1.57it/s]
  4%|▍         | 288/7505 [04:27<1:17:14,  1.56it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.61it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.81it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.61it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.06it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.76it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.56it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.45it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.37it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.32it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.27it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.28it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.26it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.24it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.23it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.23it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.23it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.23it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.22it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.22it/s]
                                                    
+
                                               
{'eval_loss': '5.567', 'eval_runtime': '3.264', 'eval_samples_per_second': '747.8', 'eval_steps_per_second': '11.95', 'epoch': '15.16'}
+
  4%|▍         | 288/7505 [04:30<1:17:14,  1.56it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.22it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C0_baseline/checkpoint-288
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C0_baseline/checkpoint-288/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C0_baseline/checkpoint-288/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.93it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.93it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C0_baseline/checkpoint-288/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C0_baseline/checkpoint-288/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C0_baseline/checkpoint-288/tokenizer_config.json
+[eval] epoch 15.161 step 288 loss 5.5671 (best 5.5671 @ ep 15.161, 0/20 since best)
+
+
                                               
  4%|▍         | 289/7505 [04:33<4:28:53,  2.24s/it]
  4%|▍         | 290/7505 [04:34<3:31:48,  1.76s/it]
                                                    
{'loss': '5.467', 'grad_norm': '0.8695', 'learning_rate': '0.0006', 'epoch': '15.27'}
+
  4%|▍         | 290/7505 [04:34<3:31:48,  1.76s/it]
  4%|▍         | 291/7505 [04:34<2:51:50,  1.43s/it]
  4%|▍         | 292/7505 [04:35<2:23:54,  1.20s/it]
  4%|▍         | 293/7505 [04:36<2:04:21,  1.03s/it]
  4%|▍         | 294/7505 [04:36<1:50:39,  1.09it/s]
  4%|▍         | 295/7505 [04:37<1:41:01,  1.19it/s]
  4%|▍         | 296/7505 [04:38<1:34:23,  1.27it/s]
  4%|▍         | 297/7505 [04:38<1:29:42,  1.34it/s]
  4%|▍         | 298/7505 [04:39<1:26:25,  1.39it/s]
  4%|▍         | 299/7505 [04:40<1:24:04,  1.43it/s]
  4%|▍         | 300/7505 [04:40<1:22:26,  1.46it/s]
                                                    
{'loss': '5.398', 'grad_norm': '0.298', 'learning_rate': '0.0006', 'epoch': '15.81'}
+
  4%|▍         | 300/7505 [04:40<1:22:26,  1.46it/s]
  4%|▍         | 301/7505 [04:41<1:21:21,  1.48it/s]
  4%|▍         | 302/7505 [04:42<1:20:33,  1.49it/s]
  4%|▍         | 303/7505 [04:42<1:19:55,  1.50it/s]
  4%|▍         | 304/7505 [04:43<1:10:45,  1.70it/s]
  4%|▍         | 305/7505 [04:44<1:30:54,  1.32it/s]
  4%|▍         | 306/7505 [04:45<1:27:10,  1.38it/s]
  4%|▍         | 307/7505 [04:45<1:24:37,  1.42it/s]
  4%|▍         | 308/7505 [04:46<1:38:20,  1.22it/s]
  4%|▍         | 309/7505 [04:47<1:32:24,  1.30it/s]
  4%|▍         | 310/7505 [04:48<1:28:15,  1.36it/s]
                                                    
{'loss': '5.305', 'grad_norm': '0.4353', 'learning_rate': '0.0006', 'epoch': '16.32'}
+
  4%|▍         | 310/7505 [04:48<1:28:15,  1.36it/s]
  4%|▍         | 311/7505 [04:48<1:25:20,  1.40it/s]
  4%|▍         | 312/7505 [04:49<1:23:16,  1.44it/s]
  4%|▍         | 313/7505 [04:50<1:46:15,  1.13it/s]
  4%|▍         | 314/7505 [04:51<1:53:12,  1.06it/s]
  4%|▍         | 315/7505 [04:52<1:42:45,  1.17it/s]
  4%|▍         | 316/7505 [04:53<1:35:31,  1.25it/s]
  4%|▍         | 317/7505 [04:53<1:30:23,  1.33it/s]
  4%|▍         | 318/7505 [04:54<1:42:28,  1.17it/s]
  4%|▍         | 319/7505 [04:55<1:35:17,  1.26it/s]
  4%|▍         | 320/7505 [04:56<1:30:13,  1.33it/s]
                                                    
{'loss': '5.284', 'grad_norm': '0.6091', 'learning_rate': '0.0006', 'epoch': '16.86'}
+
  4%|▍         | 320/7505 [04:56<1:30:13,  1.33it/s]
  4%|▍         | 321/7505 [04:56<1:26:41,  1.38it/s]
  4%|▍         | 322/7505 [04:57<1:24:08,  1.42it/s]
  4%|▍         | 323/7505 [04:57<1:13:44,  1.62it/s]
  4%|▍         | 324/7505 [04:58<1:17:48,  1.54it/s]
  4%|▍         | 325/7505 [04:59<1:17:59,  1.53it/s]
  4%|▍         | 326/7505 [04:59<1:18:07,  1.53it/s]
  4%|▍         | 327/7505 [05:00<1:18:10,  1.53it/s]
  4%|▍         | 328/7505 [05:01<1:18:13,  1.53it/s]
  4%|▍         | 329/7505 [05:01<1:18:11,  1.53it/s]
  4%|▍         | 330/7505 [05:02<1:18:16,  1.53it/s]
                                                    
{'loss': '5.229', 'grad_norm': '0.3825', 'learning_rate': '0.0006', 'epoch': '17.38'}
+
  4%|▍         | 330/7505 [05:02<1:18:16,  1.53it/s]
  4%|▍         | 331/7505 [05:03<1:18:17,  1.53it/s]
  4%|▍         | 332/7505 [05:03<1:18:17,  1.53it/s]
  4%|▍         | 333/7505 [05:04<1:18:19,  1.53it/s]
  4%|▍         | 334/7505 [05:05<1:18:20,  1.53it/s]
  4%|▍         | 335/7505 [05:05<1:18:22,  1.52it/s]
  4%|▍         | 336/7505 [05:06<1:18:19,  1.53it/s]
  4%|▍         | 337/7505 [05:07<1:18:34,  1.52it/s]
  5%|▍         | 338/7505 [05:07<1:18:26,  1.52it/s]
  5%|▍         | 339/7505 [05:08<1:18:20,  1.52it/s]
  5%|▍         | 340/7505 [05:09<1:18:19,  1.52it/s]
                                                    
{'loss': '5.159', 'grad_norm': '0.5717', 'learning_rate': '0.0006', 'epoch': '17.91'}
+
  5%|▍         | 340/7505 [05:09<1:18:19,  1.52it/s]
  5%|▍         | 341/7505 [05:09<1:18:18,  1.52it/s]
  5%|▍         | 342/7505 [05:10<1:09:33,  1.72it/s]
  5%|▍         | 343/7505 [05:10<1:14:32,  1.60it/s]
  5%|▍         | 344/7505 [05:11<1:15:37,  1.58it/s]
  5%|▍         | 345/7505 [05:12<1:16:23,  1.56it/s]
  5%|▍         | 346/7505 [05:12<1:16:57,  1.55it/s]
  5%|▍         | 347/7505 [05:13<1:17:16,  1.54it/s]
  5%|▍         | 348/7505 [05:14<1:17:29,  1.54it/s]
  5%|▍         | 349/7505 [05:14<1:17:40,  1.54it/s]
  5%|▍         | 350/7505 [05:15<1:17:47,  1.53it/s]
                                                    
{'loss': '5.094', 'grad_norm': '0.8797', 'learning_rate': '0.0006', 'epoch': '18.43'}
+
  5%|▍         | 350/7505 [05:15<1:17:47,  1.53it/s]
  5%|▍         | 351/7505 [05:16<1:17:53,  1.53it/s]
  5%|▍         | 352/7505 [05:16<1:17:55,  1.53it/s]
  5%|▍         | 353/7505 [05:17<1:18:00,  1.53it/s]
  5%|▍         | 354/7505 [05:18<1:18:04,  1.53it/s]
  5%|▍         | 355/7505 [05:18<1:18:05,  1.53it/s]
  5%|▍         | 356/7505 [05:19<1:18:05,  1.53it/s]
  5%|▍         | 357/7505 [05:20<1:18:02,  1.53it/s]
  5%|▍         | 358/7505 [05:20<1:18:04,  1.53it/s]
  5%|▍         | 359/7505 [05:21<1:18:02,  1.53it/s]
  5%|▍         | 360/7505 [05:22<1:17:59,  1.53it/s]
                                                    
{'loss': '5.084', 'grad_norm': '0.5506', 'learning_rate': '0.0006', 'epoch': '18.97'}
+
  5%|▍         | 360/7505 [05:22<1:17:59,  1.53it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.49it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.78it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.58it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.06it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.72it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.52it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.41it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.36it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.30it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.29it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.28it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.27it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.25it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.25it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.24it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.24it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.24it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.23it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.22it/s]
                                                    
+
                                               
{'eval_loss': '5.213', 'eval_runtime': '3.266', 'eval_samples_per_second': '747.3', 'eval_steps_per_second': '11.94', 'epoch': '18.97'}
+
  5%|▍         | 360/7505 [05:25<1:17:59,  1.53it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.22it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C0_baseline/checkpoint-360
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C0_baseline/checkpoint-360/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C0_baseline/checkpoint-360/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.83it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.83it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C0_baseline/checkpoint-360/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C0_baseline/checkpoint-360/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C0_baseline/checkpoint-360/tokenizer_config.json
+[eval] epoch 18.966 step 360 loss 5.2135 (best 5.2135 @ ep 18.966, 0/20 since best)
+
+
                                               
  5%|▍         | 361/7505 [05:27<4:20:50,  2.19s/it]
  5%|▍         | 362/7505 [05:28<3:28:16,  1.75s/it]
  5%|▍         | 363/7505 [05:29<2:49:08,  1.42s/it]
  5%|▍         | 364/7505 [05:29<2:21:43,  1.19s/it]
  5%|▍         | 365/7505 [05:30<2:02:35,  1.03s/it]
  5%|▍         | 366/7505 [05:31<1:49:12,  1.09it/s]
  5%|▍         | 367/7505 [05:31<1:39:51,  1.19it/s]
  5%|▍         | 368/7505 [05:32<1:33:13,  1.28it/s]
  5%|▍         | 369/7505 [05:33<1:28:36,  1.34it/s]
  5%|▍         | 370/7505 [05:33<1:25:21,  1.39it/s]
                                                    
{'loss': '4.995', 'grad_norm': '0.397', 'learning_rate': '0.0006', 'epoch': '19.48'}
+
  5%|▍         | 370/7505 [05:33<1:25:21,  1.39it/s]
  5%|▍         | 371/7505 [05:34<1:23:10,  1.43it/s]
  5%|▍         | 372/7505 [05:35<1:21:33,  1.46it/s]
  5%|▍         | 373/7505 [05:35<1:20:26,  1.48it/s]
  5%|▍         | 374/7505 [05:36<1:19:38,  1.49it/s]
  5%|▍         | 375/7505 [05:37<1:19:09,  1.50it/s]
  5%|▌         | 376/7505 [05:37<1:18:43,  1.51it/s]
  5%|▌         | 377/7505 [05:38<1:18:29,  1.51it/s]
  5%|▌         | 378/7505 [05:39<1:18:15,  1.52it/s]
  5%|▌         | 379/7505 [05:39<1:18:07,  1.52it/s]
  5%|▌         | 380/7505 [05:40<1:09:19,  1.71it/s]
                                                    
{'loss': '4.933', 'grad_norm': '0.2371', 'learning_rate': '0.0006', 'epoch': '20'}
+
  5%|▌         | 380/7505 [05:40<1:09:19,  1.71it/s]
  5%|▌         | 381/7505 [05:40<1:14:11,  1.60it/s]
  5%|▌         | 382/7505 [05:41<1:15:14,  1.58it/s]
  5%|▌         | 383/7505 [05:42<1:16:00,  1.56it/s]
  5%|▌         | 384/7505 [05:42<1:16:38,  1.55it/s]
  5%|▌         | 385/7505 [05:43<1:17:00,  1.54it/s]
  5%|▌         | 386/7505 [05:44<1:17:12,  1.54it/s]
  5%|▌         | 387/7505 [05:44<1:17:20,  1.53it/s]
  5%|▌         | 388/7505 [05:45<1:17:30,  1.53it/s]
  5%|▌         | 389/7505 [05:46<1:17:35,  1.53it/s]
  5%|▌         | 390/7505 [05:46<1:17:35,  1.53it/s]
                                                    
{'loss': '4.86', 'grad_norm': '0.8292', 'learning_rate': '0.0006', 'epoch': '20.54'}
+
  5%|▌         | 390/7505 [05:46<1:17:35,  1.53it/s]
  5%|▌         | 391/7505 [05:47<1:17:40,  1.53it/s]
  5%|▌         | 392/7505 [05:48<1:17:38,  1.53it/s]
  5%|▌         | 393/7505 [05:48<1:17:38,  1.53it/s]
  5%|▌         | 394/7505 [05:49<1:17:38,  1.53it/s]
  5%|▌         | 395/7505 [05:50<1:17:36,  1.53it/s]
  5%|▌         | 396/7505 [05:50<1:17:34,  1.53it/s]
  5%|▌         | 397/7505 [05:51<1:17:38,  1.53it/s]
  5%|▌         | 398/7505 [05:51<1:17:34,  1.53it/s]
  5%|▌         | 399/7505 [05:52<1:08:57,  1.72it/s]
  5%|▌         | 400/7505 [05:53<1:13:57,  1.60it/s]
                                                    
{'loss': '4.926', 'grad_norm': '0.483', 'learning_rate': '0.0006', 'epoch': '21.05'}
+
  5%|▌         | 400/7505 [05:53<1:13:57,  1.60it/s]
  5%|▌         | 401/7505 [05:53<1:15:03,  1.58it/s]
  5%|▌         | 402/7505 [05:54<1:15:47,  1.56it/s]
  5%|▌         | 403/7505 [05:55<1:16:18,  1.55it/s]
  5%|▌         | 404/7505 [05:55<1:16:38,  1.54it/s]
  5%|▌         | 405/7505 [05:56<1:16:54,  1.54it/s]
  5%|▌         | 406/7505 [05:57<1:17:03,  1.54it/s]
  5%|▌         | 407/7505 [05:57<1:17:10,  1.53it/s]
  5%|▌         | 408/7505 [05:58<1:17:14,  1.53it/s]
  5%|▌         | 409/7505 [05:59<1:17:17,  1.53it/s]
  5%|▌         | 410/7505 [05:59<1:17:19,  1.53it/s]
                                                    
{'loss': '4.813', 'grad_norm': '0.4646', 'learning_rate': '0.0006', 'epoch': '21.59'}
+
  5%|▌         | 410/7505 [05:59<1:17:19,  1.53it/s]
  5%|▌         | 411/7505 [06:00<1:17:22,  1.53it/s]
  5%|▌         | 412/7505 [06:00<1:17:21,  1.53it/s]
  6%|▌         | 413/7505 [06:01<1:17:21,  1.53it/s]
  6%|▌         | 414/7505 [06:02<1:17:23,  1.53it/s]
  6%|▌         | 415/7505 [06:02<1:17:21,  1.53it/s]
  6%|▌         | 416/7505 [06:03<1:17:21,  1.53it/s]
  6%|▌         | 417/7505 [06:04<1:17:17,  1.53it/s]
  6%|▌         | 418/7505 [06:04<1:08:41,  1.72it/s]
  6%|▌         | 419/7505 [06:05<1:14:10,  1.59it/s]
  6%|▌         | 420/7505 [06:06<1:15:07,  1.57it/s]
                                                    
{'loss': '4.752', 'grad_norm': '0.4654', 'learning_rate': '0.0006', 'epoch': '22.11'}
+
  6%|▌         | 420/7505 [06:06<1:15:07,  1.57it/s]
  6%|▌         | 421/7505 [06:06<1:15:49,  1.56it/s]
  6%|▌         | 422/7505 [06:07<1:16:20,  1.55it/s]
  6%|▌         | 423/7505 [06:08<1:16:45,  1.54it/s]
  6%|▌         | 424/7505 [06:08<1:17:01,  1.53it/s]
  6%|▌         | 425/7505 [06:09<1:17:08,  1.53it/s]
  6%|▌         | 426/7505 [06:10<1:17:15,  1.53it/s]
  6%|▌         | 427/7505 [06:10<1:17:16,  1.53it/s]
  6%|▌         | 428/7505 [06:11<1:17:15,  1.53it/s]
  6%|▌         | 429/7505 [06:11<1:17:21,  1.52it/s]
  6%|▌         | 430/7505 [06:12<1:17:20,  1.52it/s]
                                                    
{'loss': '4.681', 'grad_norm': '0.4587', 'learning_rate': '0.0006', 'epoch': '22.64'}
+
  6%|▌         | 430/7505 [06:12<1:17:20,  1.52it/s]
  6%|▌         | 431/7505 [06:13<1:17:23,  1.52it/s]
  6%|▌         | 432/7505 [06:13<1:17:22,  1.52it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.94it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.88it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.65it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.06it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.75it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.55it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.45it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.37it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.32it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.28it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.29it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.26it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.25it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.23it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.22it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.23it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.22it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.22it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.22it/s]
                                                    
+
                                               
{'eval_loss': '4.929', 'eval_runtime': '3.243', 'eval_samples_per_second': '752.7', 'eval_steps_per_second': '12.03', 'epoch': '22.75'}
+
  6%|▌         | 432/7505 [06:17<1:17:22,  1.52it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.22it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C0_baseline/checkpoint-432
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C0_baseline/checkpoint-432/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C0_baseline/checkpoint-432/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.71it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.70it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C0_baseline/checkpoint-432/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C0_baseline/checkpoint-432/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C0_baseline/checkpoint-432/tokenizer_config.json
+[eval] epoch 22.752 step 432 loss 4.9285 (best 4.9285 @ ep 22.752, 0/20 since best)
+
+
                                               
  6%|▌         | 433/7505 [06:19<4:27:29,  2.27s/it]
  6%|▌         | 434/7505 [06:20<3:30:25,  1.79s/it]
  6%|▌         | 435/7505 [06:21<2:50:30,  1.45s/it]
  6%|▌         | 436/7505 [06:21<2:22:27,  1.21s/it]
  6%|▌         | 437/7505 [06:22<1:54:15,  1.03it/s]
  6%|▌         | 438/7505 [06:23<1:45:31,  1.12it/s]
  6%|▌         | 439/7505 [06:23<1:37:00,  1.21it/s]
  6%|▌         | 440/7505 [06:24<1:31:01,  1.29it/s]
                                                    
{'loss': '4.621', 'grad_norm': '0.4153', 'learning_rate': '0.0006', 'epoch': '23.16'}
+
  6%|▌         | 440/7505 [06:24<1:31:01,  1.29it/s]
  6%|▌         | 441/7505 [06:25<1:26:56,  1.35it/s]
  6%|▌         | 442/7505 [06:25<1:24:00,  1.40it/s]
  6%|▌         | 443/7505 [06:26<1:21:58,  1.44it/s]
  6%|▌         | 444/7505 [06:27<1:20:33,  1.46it/s]
  6%|▌         | 445/7505 [06:27<1:19:36,  1.48it/s]
  6%|▌         | 446/7505 [06:28<1:18:50,  1.49it/s]
  6%|▌         | 447/7505 [06:28<1:18:20,  1.50it/s]
  6%|▌         | 448/7505 [06:29<1:17:56,  1.51it/s]
  6%|▌         | 449/7505 [06:30<1:17:41,  1.51it/s]
  6%|▌         | 450/7505 [06:30<1:17:32,  1.52it/s]
                                                    
{'loss': '4.583', 'grad_norm': '0.8873', 'learning_rate': '0.0006', 'epoch': '23.7'}
+
  6%|▌         | 450/7505 [06:30<1:17:32,  1.52it/s]
  6%|▌         | 451/7505 [06:31<1:17:26,  1.52it/s]
  6%|▌         | 452/7505 [06:32<1:17:21,  1.52it/s]
  6%|▌         | 453/7505 [06:32<1:17:14,  1.52it/s]
  6%|▌         | 454/7505 [06:33<1:17:10,  1.52it/s]
  6%|▌         | 455/7505 [06:34<1:17:07,  1.52it/s]
  6%|▌         | 456/7505 [06:34<1:08:29,  1.72it/s]
  6%|▌         | 457/7505 [06:35<1:13:20,  1.60it/s]
  6%|▌         | 458/7505 [06:36<1:14:22,  1.58it/s]
  6%|▌         | 459/7505 [06:36<1:15:11,  1.56it/s]
  6%|▌         | 460/7505 [06:37<1:15:46,  1.55it/s]
                                                    
{'loss': '4.623', 'grad_norm': '0.7489', 'learning_rate': '0.0006', 'epoch': '24.21'}
+
  6%|▌         | 460/7505 [06:37<1:15:46,  1.55it/s]
  6%|▌         | 461/7505 [06:37<1:16:13,  1.54it/s]
  6%|▌         | 462/7505 [06:38<1:16:29,  1.53it/s]
  6%|▌         | 463/7505 [06:39<1:16:36,  1.53it/s]
  6%|▌         | 464/7505 [06:39<1:16:41,  1.53it/s]
  6%|▌         | 465/7505 [06:40<1:16:46,  1.53it/s]
  6%|▌         | 466/7505 [06:41<1:16:53,  1.53it/s]
  6%|▌         | 467/7505 [06:41<1:16:49,  1.53it/s]
  6%|▌         | 468/7505 [06:42<1:16:50,  1.53it/s]
  6%|▌         | 469/7505 [06:43<1:16:48,  1.53it/s]
  6%|▋         | 470/7505 [06:43<1:16:51,  1.53it/s]
                                                    
{'loss': '4.547', 'grad_norm': '0.4049', 'learning_rate': '0.0006', 'epoch': '24.75'}
+
  6%|▋         | 470/7505 [06:43<1:16:51,  1.53it/s]
  6%|▋         | 471/7505 [06:44<1:16:53,  1.52it/s]
  6%|▋         | 472/7505 [06:45<1:16:48,  1.53it/s]
  6%|▋         | 473/7505 [06:45<1:16:47,  1.53it/s]
  6%|▋         | 474/7505 [06:46<1:16:47,  1.53it/s]
  6%|▋         | 475/7505 [06:46<1:08:13,  1.72it/s]
  6%|▋         | 476/7505 [06:47<1:13:18,  1.60it/s]
  6%|▋         | 477/7505 [06:48<1:14:22,  1.57it/s]
  6%|▋         | 478/7505 [06:48<1:15:07,  1.56it/s]
  6%|▋         | 479/7505 [06:49<1:15:36,  1.55it/s]
  6%|▋         | 480/7505 [06:50<1:15:55,  1.54it/s]
                                                    
{'loss': '4.46', 'grad_norm': '0.3185', 'learning_rate': '0.0006', 'epoch': '25.27'}
+
  6%|▋         | 480/7505 [06:50<1:15:55,  1.54it/s]
  6%|▋         | 481/7505 [06:50<1:16:16,  1.53it/s]
  6%|▋         | 482/7505 [06:51<1:16:30,  1.53it/s]
  6%|▋         | 483/7505 [06:52<1:16:38,  1.53it/s]
  6%|▋         | 484/7505 [06:52<1:16:43,  1.53it/s]
  6%|▋         | 485/7505 [06:53<1:16:41,  1.53it/s]
  6%|▋         | 486/7505 [06:54<1:16:38,  1.53it/s]
  6%|▋         | 487/7505 [06:54<1:16:41,  1.53it/s]
  7%|▋         | 488/7505 [06:55<1:16:45,  1.52it/s]
  7%|▋         | 489/7505 [06:56<1:16:48,  1.52it/s]
  7%|▋         | 490/7505 [06:56<1:16:45,  1.52it/s]
                                                    
{'loss': '4.375', 'grad_norm': '0.2807', 'learning_rate': '0.0006', 'epoch': '25.81'}
+
  7%|▋         | 490/7505 [06:56<1:16:45,  1.52it/s]
  7%|▋         | 491/7505 [06:57<1:16:43,  1.52it/s]
  7%|▋         | 492/7505 [06:58<1:16:41,  1.52it/s]
  7%|▋         | 493/7505 [06:58<1:16:39,  1.52it/s]
  7%|▋         | 494/7505 [06:59<1:08:06,  1.72it/s]
  7%|▋         | 495/7505 [06:59<1:13:00,  1.60it/s]
  7%|▋         | 496/7505 [07:00<1:14:03,  1.58it/s]
  7%|▋         | 497/7505 [07:01<1:14:49,  1.56it/s]
  7%|▋         | 498/7505 [07:01<1:15:17,  1.55it/s]
  7%|▋         | 499/7505 [07:02<1:15:44,  1.54it/s]
  7%|▋         | 500/7505 [07:03<1:15:57,  1.54it/s]
                                                    
{'loss': '4.311', 'grad_norm': '0.5341', 'learning_rate': '0.0006', 'epoch': '26.32'}
+
  7%|▋         | 500/7505 [07:03<1:15:57,  1.54it/s]
  7%|▋         | 501/7505 [07:03<1:16:08,  1.53it/s]
  7%|▋         | 502/7505 [07:04<1:16:13,  1.53it/s]
  7%|▋         | 503/7505 [07:05<1:16:19,  1.53it/s]
  7%|▋         | 504/7505 [07:05<1:16:25,  1.53it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.34it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.72it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.58it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.02it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.74it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.55it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.43it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.36it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.32it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.27it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.26it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.26it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.23it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.23it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.23it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.23it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.24it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.21it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s]
                                                    
+
                                               
{'eval_loss': '4.761', 'eval_runtime': '3.264', 'eval_samples_per_second': '748', 'eval_steps_per_second': '11.95', 'epoch': '26.54'}
+
  7%|▋         | 504/7505 [07:09<1:16:25,  1.53it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C0_baseline/checkpoint-504
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C0_baseline/checkpoint-504/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C0_baseline/checkpoint-504/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.81it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.81it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C0_baseline/checkpoint-504/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C0_baseline/checkpoint-504/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C0_baseline/checkpoint-504/tokenizer_config.json
+[eval] epoch 26.537 step 504 loss 4.7612 (best 4.7612 @ ep 26.537, 0/20 since best)
+
+
                                               
  7%|▋         | 505/7505 [07:11<4:24:25,  2.27s/it]
  7%|▋         | 506/7505 [07:12<3:27:58,  1.78s/it]
  7%|▋         | 507/7505 [07:13<2:48:25,  1.44s/it]
  7%|▋         | 508/7505 [07:13<2:20:47,  1.21s/it]
  7%|▋         | 509/7505 [07:14<2:01:25,  1.04s/it]
  7%|▋         | 510/7505 [07:15<1:47:57,  1.08it/s]
                                                    
{'loss': '4.311', 'grad_norm': '1.043', 'learning_rate': '0.0006', 'epoch': '26.86'}
+
  7%|▋         | 510/7505 [07:15<1:47:57,  1.08it/s]
  7%|▋         | 511/7505 [07:15<1:38:31,  1.18it/s]
  7%|▋         | 512/7505 [07:16<1:31:48,  1.27it/s]
  7%|▋         | 513/7505 [07:16<1:18:39,  1.48it/s]
  7%|▋         | 514/7505 [07:17<1:20:16,  1.45it/s]
  7%|▋         | 515/7505 [07:18<1:19:06,  1.47it/s]
  7%|▋         | 516/7505 [07:18<1:18:21,  1.49it/s]
  7%|▋         | 517/7505 [07:19<1:17:45,  1.50it/s]
  7%|▋         | 518/7505 [07:20<1:17:19,  1.51it/s]
  7%|▋         | 519/7505 [07:20<1:17:02,  1.51it/s]
  7%|▋         | 520/7505 [07:21<1:16:48,  1.52it/s]
                                                    
{'loss': '4.348', 'grad_norm': '0.9857', 'learning_rate': '0.0006', 'epoch': '27.38'}
+
  7%|▋         | 520/7505 [07:21<1:16:48,  1.52it/s]
  7%|▋         | 521/7505 [07:22<1:16:42,  1.52it/s]
  7%|▋         | 522/7505 [07:22<1:16:33,  1.52it/s]
  7%|▋         | 523/7505 [07:23<1:16:28,  1.52it/s]
  7%|▋         | 524/7505 [07:24<1:16:25,  1.52it/s]
  7%|▋         | 525/7505 [07:24<1:16:22,  1.52it/s]
  7%|▋         | 526/7505 [07:25<1:16:19,  1.52it/s]
  7%|▋         | 527/7505 [07:26<1:16:14,  1.53it/s]
  7%|▋         | 528/7505 [07:26<1:16:12,  1.53it/s]
  7%|▋         | 529/7505 [07:27<1:16:14,  1.53it/s]
  7%|▋         | 530/7505 [07:28<1:16:16,  1.52it/s]
                                                    
{'loss': '4.313', 'grad_norm': '0.6892', 'learning_rate': '0.0006', 'epoch': '27.91'}
+
  7%|▋         | 530/7505 [07:28<1:16:16,  1.52it/s]
  7%|▋         | 531/7505 [07:28<1:16:13,  1.52it/s]
  7%|▋         | 532/7505 [07:29<1:07:42,  1.72it/s]
  7%|▋         | 533/7505 [07:29<1:13:38,  1.58it/s]
  7%|▋         | 534/7505 [07:30<1:14:24,  1.56it/s]
  7%|▋         | 535/7505 [07:31<1:14:58,  1.55it/s]
  7%|▋         | 536/7505 [07:31<1:15:16,  1.54it/s]
  7%|▋         | 537/7505 [07:32<1:15:33,  1.54it/s]
  7%|▋         | 538/7505 [07:33<1:15:46,  1.53it/s]
  7%|▋         | 539/7505 [07:33<1:15:56,  1.53it/s]
  7%|▋         | 540/7505 [07:34<1:15:58,  1.53it/s]
                                                    
{'loss': '4.186', 'grad_norm': '0.3092', 'learning_rate': '0.0006', 'epoch': '28.43'}
+
  7%|▋         | 540/7505 [07:34<1:15:58,  1.53it/s]
  7%|▋         | 541/7505 [07:35<1:16:02,  1.53it/s]
  7%|▋         | 542/7505 [07:35<1:15:59,  1.53it/s]
  7%|▋         | 543/7505 [07:36<1:16:03,  1.53it/s]
  7%|▋         | 544/7505 [07:37<1:16:07,  1.52it/s]
  7%|▋         | 545/7505 [07:37<1:16:06,  1.52it/s]
  7%|▋         | 546/7505 [07:38<1:16:03,  1.52it/s]
  7%|▋         | 547/7505 [07:39<1:15:59,  1.53it/s]
  7%|▋         | 548/7505 [07:39<1:16:01,  1.53it/s]
  7%|▋         | 549/7505 [07:40<1:16:07,  1.52it/s]
  7%|▋         | 550/7505 [07:41<1:16:01,  1.52it/s]
                                                    
{'loss': '4.134', 'grad_norm': '0.2504', 'learning_rate': '0.0006', 'epoch': '28.97'}
+
  7%|▋         | 550/7505 [07:41<1:16:01,  1.52it/s]
  7%|▋         | 551/7505 [07:41<1:07:35,  1.71it/s]
  7%|▋         | 552/7505 [07:42<1:12:41,  1.59it/s]
  7%|▋         | 553/7505 [07:42<1:13:40,  1.57it/s]
  7%|▋         | 554/7505 [07:43<1:14:18,  1.56it/s]
  7%|▋         | 555/7505 [07:44<1:14:47,  1.55it/s]
  7%|▋         | 556/7505 [07:44<1:15:08,  1.54it/s]
  7%|▋         | 557/7505 [07:45<1:15:21,  1.54it/s]
  7%|▋         | 558/7505 [07:46<1:15:29,  1.53it/s]
  7%|▋         | 559/7505 [07:46<1:15:33,  1.53it/s]
  7%|▋         | 560/7505 [07:47<1:15:40,  1.53it/s]
                                                    
{'loss': '4.018', 'grad_norm': '0.5265', 'learning_rate': '0.0006', 'epoch': '29.48'}
+
  7%|▋         | 560/7505 [07:47<1:15:40,  1.53it/s]
  7%|▋         | 561/7505 [07:48<1:15:44,  1.53it/s]
  7%|▋         | 562/7505 [07:48<1:15:46,  1.53it/s]
  8%|▊         | 563/7505 [07:49<1:15:47,  1.53it/s]
  8%|▊         | 564/7505 [07:50<1:15:46,  1.53it/s]
  8%|▊         | 565/7505 [07:50<1:15:47,  1.53it/s]
  8%|▊         | 566/7505 [07:51<1:15:44,  1.53it/s]
  8%|▊         | 567/7505 [07:52<1:15:46,  1.53it/s]
  8%|▊         | 568/7505 [07:52<1:15:45,  1.53it/s]
  8%|▊         | 569/7505 [07:53<1:15:45,  1.53it/s]
  8%|▊         | 570/7505 [07:53<1:07:18,  1.72it/s]
                                                    
{'loss': '4.023', 'grad_norm': '0.5696', 'learning_rate': '0.0006', 'epoch': '30'}
+
  8%|▊         | 570/7505 [07:53<1:07:18,  1.72it/s]
  8%|▊         | 571/7505 [07:54<1:12:14,  1.60it/s]
  8%|▊         | 572/7505 [07:55<1:13:14,  1.58it/s]
  8%|▊         | 573/7505 [07:55<1:14:00,  1.56it/s]
  8%|▊         | 574/7505 [07:56<1:14:33,  1.55it/s]
  8%|▊         | 575/7505 [07:57<1:14:59,  1.54it/s]
  8%|▊         | 576/7505 [07:57<1:15:15,  1.53it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.32it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.67it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.53it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.00it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.72it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.55it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.42it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.35it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.30it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.24it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.26it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.24it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.23it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.22it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.23it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.21it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.21it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.22it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.22it/s]
                                                    
+
                                               
{'eval_loss': '4.714', 'eval_runtime': '3.253', 'eval_samples_per_second': '750.3', 'eval_steps_per_second': '11.99', 'epoch': '30.32'}
+
  8%|▊         | 576/7505 [08:01<1:15:15,  1.53it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.22it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C0_baseline/checkpoint-576
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C0_baseline/checkpoint-576/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C0_baseline/checkpoint-576/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.92it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.92it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C0_baseline/checkpoint-576/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C0_baseline/checkpoint-576/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C0_baseline/checkpoint-576/tokenizer_config.json
+[eval] epoch 30.322 step 576 loss 4.7139 (best 4.7139 @ ep 30.322, 0/20 since best)
+
+
                                               
  8%|▊         | 577/7505 [08:03<4:19:09,  2.24s/it]
  8%|▊         | 578/7505 [08:04<3:24:03,  1.77s/it]
  8%|▊         | 579/7505 [08:05<2:45:31,  1.43s/it]
  8%|▊         | 580/7505 [08:05<2:18:29,  1.20s/it]
                                                    
{'loss': '3.977', 'grad_norm': '1.359', 'learning_rate': '0.0006', 'epoch': '30.54'}
+
  8%|▊         | 580/7505 [08:05<2:18:29,  1.20s/it]
  8%|▊         | 581/7505 [08:06<1:59:39,  1.04s/it]
  8%|▊         | 582/7505 [08:07<1:46:24,  1.08it/s]
  8%|▊         | 583/7505 [08:07<1:37:09,  1.19it/s]
  8%|▊         | 584/7505 [08:08<1:30:39,  1.27it/s]
  8%|▊         | 585/7505 [08:08<1:26:08,  1.34it/s]
  8%|▊         | 586/7505 [08:09<1:22:56,  1.39it/s]
  8%|▊         | 587/7505 [08:10<1:20:41,  1.43it/s]
  8%|▊         | 588/7505 [08:10<1:19:04,  1.46it/s]
  8%|▊         | 589/7505 [08:11<1:09:38,  1.65it/s]
  8%|▊         | 590/7505 [08:12<1:13:49,  1.56it/s]
                                                    
{'loss': '4.025', 'grad_norm': '0.9149', 'learning_rate': '0.0006', 'epoch': '31.05'}
+
  8%|▊         | 590/7505 [08:12<1:13:49,  1.56it/s]
  8%|▊         | 591/7505 [08:12<1:14:26,  1.55it/s]
  8%|▊         | 592/7505 [08:13<1:14:46,  1.54it/s]
  8%|▊         | 593/7505 [08:14<1:14:59,  1.54it/s]
  8%|▊         | 594/7505 [08:14<1:15:06,  1.53it/s]
  8%|▊         | 595/7505 [08:15<1:15:11,  1.53it/s]
  8%|▊         | 596/7505 [08:16<1:15:18,  1.53it/s]
  8%|▊         | 597/7505 [08:16<1:15:18,  1.53it/s]
  8%|▊         | 598/7505 [08:17<1:15:20,  1.53it/s]
  8%|▊         | 599/7505 [08:17<1:15:21,  1.53it/s]
  8%|▊         | 600/7505 [08:18<1:15:24,  1.53it/s]
                                                    
{'loss': '3.945', 'grad_norm': '0.9934', 'learning_rate': '0.0006', 'epoch': '31.59'}
+
  8%|▊         | 600/7505 [08:18<1:15:24,  1.53it/s]
  8%|▊         | 601/7505 [08:19<1:15:26,  1.53it/s]
  8%|▊         | 602/7505 [08:19<1:15:23,  1.53it/s]
  8%|▊         | 603/7505 [08:20<1:15:19,  1.53it/s]
  8%|▊         | 604/7505 [08:21<1:15:21,  1.53it/s]
  8%|▊         | 605/7505 [08:21<1:15:20,  1.53it/s]
  8%|▊         | 606/7505 [08:22<1:15:18,  1.53it/s]
  8%|▊         | 607/7505 [08:23<1:15:12,  1.53it/s]
  8%|▊         | 608/7505 [08:23<1:06:55,  1.72it/s]
  8%|▊         | 609/7505 [08:24<1:11:44,  1.60it/s]
  8%|▊         | 610/7505 [08:25<1:12:46,  1.58it/s]
                                                    
{'loss': '3.911', 'grad_norm': '0.4876', 'learning_rate': '0.0006', 'epoch': '32.11'}
+
  8%|▊         | 610/7505 [08:25<1:12:46,  1.58it/s]
  8%|▊         | 611/7505 [08:25<1:13:33,  1.56it/s]
  8%|▊         | 612/7505 [08:26<1:14:01,  1.55it/s]
  8%|▊         | 613/7505 [08:26<1:14:25,  1.54it/s]
  8%|▊         | 614/7505 [08:27<1:14:42,  1.54it/s]
  8%|▊         | 615/7505 [08:28<1:14:50,  1.53it/s]
  8%|▊         | 616/7505 [08:28<1:14:59,  1.53it/s]
  8%|▊         | 617/7505 [08:29<1:15:05,  1.53it/s]
  8%|▊         | 618/7505 [08:30<1:15:10,  1.53it/s]
  8%|▊         | 619/7505 [08:30<1:15:11,  1.53it/s]
  8%|▊         | 620/7505 [08:31<1:15:09,  1.53it/s]
                                                    
{'loss': '3.778', 'grad_norm': '0.4249', 'learning_rate': '0.0006', 'epoch': '32.64'}
+
  8%|▊         | 620/7505 [08:31<1:15:09,  1.53it/s]
  8%|▊         | 621/7505 [08:32<1:15:11,  1.53it/s]
  8%|▊         | 622/7505 [08:32<1:15:10,  1.53it/s]
  8%|▊         | 623/7505 [08:33<1:15:11,  1.53it/s]
  8%|▊         | 624/7505 [08:34<1:15:10,  1.53it/s]
  8%|▊         | 625/7505 [08:34<1:15:10,  1.53it/s]
  8%|▊         | 626/7505 [08:35<1:15:06,  1.53it/s]
  8%|▊         | 627/7505 [08:35<1:06:48,  1.72it/s]
  8%|▊         | 628/7505 [08:36<1:11:37,  1.60it/s]
  8%|▊         | 629/7505 [08:37<1:12:37,  1.58it/s]
  8%|▊         | 630/7505 [08:37<1:13:24,  1.56it/s]
                                                    
{'loss': '3.738', 'grad_norm': '0.6541', 'learning_rate': '0.0006', 'epoch': '33.16'}
+
  8%|▊         | 630/7505 [08:37<1:13:24,  1.56it/s]
  8%|▊         | 631/7505 [08:38<1:13:57,  1.55it/s]
  8%|▊         | 632/7505 [08:39<1:14:16,  1.54it/s]
  8%|▊         | 633/7505 [08:39<1:14:32,  1.54it/s]
  8%|▊         | 634/7505 [08:40<1:14:44,  1.53it/s]
  8%|▊         | 635/7505 [08:41<1:14:47,  1.53it/s]
  8%|▊         | 636/7505 [08:41<1:14:53,  1.53it/s]
  8%|▊         | 637/7505 [08:42<1:14:54,  1.53it/s]
  9%|▊         | 638/7505 [08:43<1:14:57,  1.53it/s]
  9%|▊         | 639/7505 [08:43<1:14:55,  1.53it/s]
  9%|▊         | 640/7505 [08:44<1:14:57,  1.53it/s]
                                                    
{'loss': '3.668', 'grad_norm': '0.9462', 'learning_rate': '0.0006', 'epoch': '33.7'}
+
  9%|▊         | 640/7505 [08:44<1:14:57,  1.53it/s]
  9%|▊         | 641/7505 [08:45<1:14:59,  1.53it/s]
  9%|▊         | 642/7505 [08:45<1:14:58,  1.53it/s]
  9%|▊         | 643/7505 [08:46<1:14:59,  1.52it/s]
  9%|▊         | 644/7505 [08:47<1:14:56,  1.53it/s]
  9%|▊         | 645/7505 [08:47<1:14:56,  1.53it/s]
  9%|▊         | 646/7505 [08:48<1:06:37,  1.72it/s]
  9%|▊         | 647/7505 [08:48<1:12:01,  1.59it/s]
  9%|▊         | 648/7505 [08:49<1:12:54,  1.57it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.42it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.72it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.59it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.02it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.75it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.55it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.44it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.35it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.33it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.28it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.27it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.24it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.23it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.24it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.23it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.23it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.22it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.22it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s]
                                                    
+
                                               
{'eval_loss': '4.762', 'eval_runtime': '3.239', 'eval_samples_per_second': '753.6', 'eval_steps_per_second': '12.04', 'epoch': '34.11'}
+
  9%|▊         | 648/7505 [08:52<1:12:54,  1.57it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s][eval] epoch 34.107 step 648 loss 4.7615 (best 4.7139 @ ep 30.322, 1/20 since best)
+
+
                                               
  9%|▊         | 649/7505 [08:53<3:04:28,  1.61s/it]
  9%|▊         | 650/7505 [08:54<2:31:33,  1.33s/it]
                                                    
{'loss': '3.654', 'grad_norm': '1.135', 'learning_rate': '0.0006', 'epoch': '34.21'}
+
  9%|▊         | 650/7505 [08:54<2:31:33,  1.33s/it]
  9%|▊         | 651/7505 [08:54<2:08:33,  1.13s/it]
  9%|▊         | 652/7505 [08:55<1:52:27,  1.02it/s]
  9%|▊         | 653/7505 [08:56<1:41:11,  1.13it/s]
  9%|▊         | 654/7505 [08:56<1:33:15,  1.22it/s]
  9%|▊         | 655/7505 [08:57<1:27:41,  1.30it/s]
  9%|▊         | 656/7505 [08:58<1:23:50,  1.36it/s]
  9%|▉         | 657/7505 [08:58<1:21:06,  1.41it/s]
  9%|▉         | 658/7505 [08:59<1:19:10,  1.44it/s]
  9%|▉         | 659/7505 [09:00<1:17:48,  1.47it/s]
  9%|▉         | 660/7505 [09:00<1:16:55,  1.48it/s]
                                                    
{'loss': '3.599', 'grad_norm': '1.07', 'learning_rate': '0.0006', 'epoch': '34.75'}
+
  9%|▉         | 660/7505 [09:00<1:16:55,  1.48it/s]
  9%|▉         | 661/7505 [09:01<1:16:18,  1.49it/s]
  9%|▉         | 662/7505 [09:02<1:15:47,  1.50it/s]
  9%|▉         | 663/7505 [09:02<1:15:28,  1.51it/s]
  9%|▉         | 664/7505 [09:03<1:15:13,  1.52it/s]
  9%|▉         | 665/7505 [09:03<1:06:46,  1.71it/s]
  9%|▉         | 666/7505 [09:04<1:11:55,  1.58it/s]
  9%|▉         | 667/7505 [09:05<1:12:43,  1.57it/s]
  9%|▉         | 668/7505 [09:05<1:13:18,  1.55it/s]
  9%|▉         | 669/7505 [09:06<1:13:43,  1.55it/s]
  9%|▉         | 670/7505 [09:07<1:13:59,  1.54it/s]
                                                    
{'loss': '3.56', 'grad_norm': '1.174', 'learning_rate': '0.0006', 'epoch': '35.27'}
+
  9%|▉         | 670/7505 [09:07<1:13:59,  1.54it/s]
  9%|▉         | 671/7505 [09:07<1:14:14,  1.53it/s]
  9%|▉         | 672/7505 [09:08<1:14:23,  1.53it/s]
  9%|▉         | 673/7505 [09:09<1:14:27,  1.53it/s]
  9%|▉         | 674/7505 [09:09<1:14:33,  1.53it/s]
  9%|▉         | 675/7505 [09:10<1:14:32,  1.53it/s]
  9%|▉         | 676/7505 [09:11<1:14:33,  1.53it/s]
  9%|▉         | 677/7505 [09:11<1:14:31,  1.53it/s]
  9%|▉         | 678/7505 [09:12<1:14:34,  1.53it/s]
  9%|▉         | 679/7505 [09:12<1:14:33,  1.53it/s]
  9%|▉         | 680/7505 [09:13<1:14:31,  1.53it/s]
                                                    
{'loss': '3.523', 'grad_norm': '1.044', 'learning_rate': '0.0006', 'epoch': '35.81'}
+
  9%|▉         | 680/7505 [09:13<1:14:31,  1.53it/s]
  9%|▉         | 681/7505 [09:14<1:14:33,  1.53it/s]
  9%|▉         | 682/7505 [09:14<1:14:33,  1.53it/s]
  9%|▉         | 683/7505 [09:15<1:14:48,  1.52it/s]
  9%|▉         | 684/7505 [09:16<1:06:26,  1.71it/s]
  9%|▉         | 685/7505 [09:16<1:11:09,  1.60it/s]
  9%|▉         | 686/7505 [09:17<1:12:18,  1.57it/s]
  9%|▉         | 687/7505 [09:18<1:13:02,  1.56it/s]
  9%|▉         | 688/7505 [09:18<1:13:26,  1.55it/s]
  9%|▉         | 689/7505 [09:19<1:14:07,  1.53it/s]
  9%|▉         | 690/7505 [09:20<1:27:36,  1.30it/s]
                                                    
{'loss': '3.451', 'grad_norm': '0.9442', 'learning_rate': '0.0006', 'epoch': '36.32'}
+
  9%|▉         | 690/7505 [09:20<1:27:36,  1.30it/s]
  9%|▉         | 691/7505 [09:21<1:45:37,  1.08it/s]
  9%|▉         | 692/7505 [09:23<1:58:12,  1.04s/it]
  9%|▉         | 693/7505 [09:24<2:07:01,  1.12s/it]
  9%|▉         | 694/7505 [09:25<2:13:12,  1.17s/it]
  9%|▉         | 695/7505 [09:26<2:17:31,  1.21s/it]
  9%|▉         | 696/7505 [09:28<2:20:29,  1.24s/it]
  9%|▉         | 697/7505 [09:29<2:21:09,  1.24s/it]
  9%|▉         | 698/7505 [09:30<2:22:20,  1.25s/it]
  9%|▉         | 699/7505 [09:32<2:22:36,  1.26s/it]
  9%|▉         | 700/7505 [09:33<2:22:51,  1.26s/it]
                                                    
{'loss': '3.411', 'grad_norm': '1.061', 'learning_rate': '0.0006', 'epoch': '36.86'}
+
  9%|▉         | 700/7505 [09:33<2:22:51,  1.26s/it]
  9%|▉         | 701/7505 [09:34<2:10:20,  1.15s/it]
  9%|▉         | 702/7505 [09:35<2:13:39,  1.18s/it]
  9%|▉         | 703/7505 [09:36<2:01:48,  1.07s/it]
  9%|▉         | 704/7505 [09:37<2:12:03,  1.17s/it]
  9%|▉         | 705/7505 [09:38<2:17:34,  1.21s/it]
  9%|▉         | 706/7505 [09:40<2:21:25,  1.25s/it]
  9%|▉         | 707/7505 [09:41<2:24:07,  1.27s/it]
  9%|▉         | 708/7505 [09:42<2:25:59,  1.29s/it]
  9%|▉         | 709/7505 [09:44<2:25:40,  1.29s/it]
  9%|▉         | 710/7505 [09:45<2:25:31,  1.28s/it]
                                                    
{'loss': '3.318', 'grad_norm': '1.345', 'learning_rate': '0.0006', 'epoch': '37.38'}
+
  9%|▉         | 710/7505 [09:45<2:25:31,  1.28s/it]
  9%|▉         | 711/7505 [09:46<2:25:28,  1.28s/it]
  9%|▉         | 712/7505 [09:48<2:25:19,  1.28s/it]
 10%|▉         | 713/7505 [09:49<2:13:37,  1.18s/it]
 10%|▉         | 714/7505 [09:50<2:13:43,  1.18s/it]
 10%|▉         | 715/7505 [09:51<2:17:22,  1.21s/it]
 10%|▉         | 716/7505 [09:52<2:19:42,  1.23s/it]
 10%|▉         | 717/7505 [09:54<2:21:28,  1.25s/it]
 10%|▉         | 718/7505 [09:55<2:22:44,  1.26s/it]
 10%|▉         | 719/7505 [09:56<2:23:31,  1.27s/it]
 10%|▉         | 720/7505 [09:57<2:24:00,  1.27s/it]
                                                    
{'loss': '3.296', 'grad_norm': '1.154', 'learning_rate': '0.0006', 'epoch': '37.91'}
+
 10%|▉         | 720/7505 [09:57<2:24:00,  1.27s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:02, 12.80it/s]
+
 10%|█         | 4/39 [00:00<00:04,  7.85it/s]
+
 13%|█▎        | 5/39 [00:00<00:04,  7.38it/s]
+
 15%|█▌        | 6/39 [00:00<00:04,  7.08it/s]
+
 18%|█▊        | 7/39 [00:00<00:04,  6.83it/s]
+
 21%|██        | 8/39 [00:01<00:04,  6.60it/s]
+
 23%|██▎       | 9/39 [00:01<00:04,  6.47it/s]
+
 26%|██▌       | 10/39 [00:01<00:04,  6.47it/s]
+
 28%|██▊       | 11/39 [00:01<00:04,  6.48it/s]
+
 31%|███       | 12/39 [00:01<00:04,  6.41it/s]
+
 33%|███▎      | 13/39 [00:01<00:04,  6.33it/s]
+
 36%|███▌      | 14/39 [00:02<00:03,  6.33it/s]
+
 38%|███▊      | 15/39 [00:02<00:03,  6.38it/s]
+
 41%|████      | 16/39 [00:02<00:03,  6.39it/s]
+
 44%|████▎     | 17/39 [00:02<00:03,  6.33it/s]
+
 46%|████▌     | 18/39 [00:02<00:03,  6.28it/s]
+
 49%|████▊     | 19/39 [00:02<00:03,  6.34it/s]
+
 51%|█████▏    | 20/39 [00:03<00:02,  6.39it/s]
+
 54%|█████▍    | 21/39 [00:03<00:02,  6.36it/s]
+
 56%|█████▋    | 22/39 [00:03<00:02,  6.29it/s]
+
 59%|█████▉    | 23/39 [00:03<00:02,  6.31it/s]
+
 62%|██████▏   | 24/39 [00:03<00:02,  6.37it/s]
+
 64%|██████▍   | 25/39 [00:03<00:02,  6.38it/s]
+
 67%|██████▋   | 26/39 [00:03<00:02,  6.33it/s]
+
 69%|██████▉   | 27/39 [00:04<00:01,  6.27it/s]
+
 72%|███████▏  | 28/39 [00:04<00:01,  6.34it/s]
+
 74%|███████▍  | 29/39 [00:04<00:01,  6.39it/s]
+
 77%|███████▋  | 30/39 [00:04<00:01,  6.36it/s]
+
 79%|███████▉  | 31/39 [00:04<00:01,  6.29it/s]
+
 82%|████████▏ | 32/39 [00:04<00:01,  6.30it/s]
+
 85%|████████▍ | 33/39 [00:05<00:00,  6.46it/s]
+
 90%|████████▉ | 35/39 [00:05<00:00,  8.26it/s]
+
 95%|█████████▍| 37/39 [00:05<00:00,  9.48it/s]
+
100%|██████████| 39/39 [00:05<00:00,  9.40it/s]
                                                    
+
                                               
{'eval_loss': '4.91', 'eval_runtime': '5.822', 'eval_samples_per_second': '419.3', 'eval_steps_per_second': '6.699', 'epoch': '37.91'}
+
 10%|▉         | 720/7505 [10:03<2:24:00,  1.27s/it]
+
100%|██████████| 39/39 [00:05<00:00,  9.40it/s][eval] epoch 37.913 step 720 loss 4.9105 (best 4.7139 @ ep 30.322, 2/20 since best)
+
+
                                               
 10%|▉         | 721/7505 [10:05<5:42:33,  3.03s/it]
 10%|▉         | 722/7505 [10:05<4:27:04,  2.36s/it]
 10%|▉         | 723/7505 [10:07<3:52:53,  2.06s/it]
 10%|▉         | 724/7505 [10:08<3:26:34,  1.83s/it]
 10%|▉         | 725/7505 [10:09<3:08:01,  1.66s/it]
 10%|▉         | 726/7505 [10:11<2:55:00,  1.55s/it]
 10%|▉         | 727/7505 [10:12<2:45:57,  1.47s/it]
 10%|▉         | 728/7505 [10:13<2:39:42,  1.41s/it]
 10%|▉         | 729/7505 [10:14<2:34:26,  1.37s/it]
 10%|▉         | 730/7505 [10:16<2:31:03,  1.34s/it]
                                                    
{'loss': '3.179', 'grad_norm': '1.147', 'learning_rate': '0.0006', 'epoch': '38.43'}
+
 10%|▉         | 730/7505 [10:16<2:31:03,  1.34s/it]
 10%|▉         | 731/7505 [10:17<2:26:00,  1.29s/it]
 10%|▉         | 732/7505 [10:18<2:04:25,  1.10s/it]
 10%|▉         | 733/7505 [10:18<1:49:15,  1.03it/s]
 10%|▉         | 734/7505 [10:19<1:38:36,  1.14it/s]
 10%|▉         | 735/7505 [10:19<1:31:11,  1.24it/s]
 10%|▉         | 736/7505 [10:20<1:26:00,  1.31it/s]
 10%|▉         | 737/7505 [10:21<1:22:24,  1.37it/s]
 10%|▉         | 738/7505 [10:21<1:19:50,  1.41it/s]
 10%|▉         | 739/7505 [10:22<1:18:04,  1.44it/s]
 10%|▉         | 740/7505 [10:23<1:16:46,  1.47it/s]
                                                    
{'loss': '3.185', 'grad_norm': '1.253', 'learning_rate': '0.0006', 'epoch': '38.97'}
+
 10%|▉         | 740/7505 [10:23<1:16:46,  1.47it/s]
 10%|▉         | 741/7505 [10:23<1:07:46,  1.66it/s]
 10%|▉         | 742/7505 [10:24<1:12:20,  1.56it/s]
 10%|▉         | 743/7505 [10:25<1:12:46,  1.55it/s]
 10%|▉         | 744/7505 [10:25<1:13:03,  1.54it/s]
 10%|▉         | 745/7505 [10:26<1:13:17,  1.54it/s]
 10%|▉         | 746/7505 [10:27<1:13:24,  1.53it/s]
 10%|▉         | 747/7505 [10:27<1:13:31,  1.53it/s]
 10%|▉         | 748/7505 [10:28<1:13:35,  1.53it/s]
 10%|▉         | 749/7505 [10:28<1:13:38,  1.53it/s]
 10%|▉         | 750/7505 [10:29<1:13:39,  1.53it/s]
                                                    
{'loss': '3.036', 'grad_norm': '1.268', 'learning_rate': '0.0006', 'epoch': '39.48'}
+
 10%|▉         | 750/7505 [10:29<1:13:39,  1.53it/s]
 10%|█         | 751/7505 [10:30<1:13:41,  1.53it/s]
 10%|█         | 752/7505 [10:30<1:13:43,  1.53it/s]
 10%|█         | 753/7505 [10:31<1:13:43,  1.53it/s]
 10%|█         | 754/7505 [10:32<1:13:45,  1.53it/s]
 10%|█         | 755/7505 [10:32<1:13:43,  1.53it/s]
 10%|█         | 756/7505 [10:33<1:13:44,  1.53it/s]
 10%|█         | 757/7505 [10:34<1:13:42,  1.53it/s]
 10%|█         | 758/7505 [10:34<1:13:44,  1.52it/s]
 10%|█         | 759/7505 [10:35<1:13:42,  1.53it/s]
 10%|█         | 760/7505 [10:35<1:05:31,  1.72it/s]
                                                    
{'loss': '3.056', 'grad_norm': '1.276', 'learning_rate': '0.0006', 'epoch': '40'}
+
 10%|█         | 760/7505 [10:35<1:05:31,  1.72it/s]
 10%|█         | 761/7505 [10:36<1:10:18,  1.60it/s]
 10%|█         | 762/7505 [10:37<1:11:18,  1.58it/s]
 10%|█         | 763/7505 [10:38<1:11:59,  1.56it/s]
 10%|█         | 764/7505 [10:38<1:12:29,  1.55it/s]
 10%|█         | 765/7505 [10:39<1:12:51,  1.54it/s]
 10%|█         | 766/7505 [10:39<1:13:05,  1.54it/s]
 10%|█         | 767/7505 [10:40<1:13:15,  1.53it/s]
 10%|█         | 768/7505 [10:41<1:13:20,  1.53it/s]
 10%|█         | 769/7505 [10:41<1:13:24,  1.53it/s]
 10%|█         | 770/7505 [10:42<1:13:27,  1.53it/s]
                                                    
{'loss': '2.882', 'grad_norm': '1.6', 'learning_rate': '0.0006', 'epoch': '40.54'}
+
 10%|█         | 770/7505 [10:42<1:13:27,  1.53it/s]
 10%|█         | 771/7505 [10:43<1:13:29,  1.53it/s]
 10%|█         | 772/7505 [10:43<1:13:31,  1.53it/s]
 10%|█         | 773/7505 [10:44<1:13:29,  1.53it/s]
 10%|█         | 774/7505 [10:45<1:13:32,  1.53it/s]
 10%|█         | 775/7505 [10:45<1:13:36,  1.52it/s]
 10%|█         | 776/7505 [10:46<1:13:32,  1.52it/s]
 10%|█         | 777/7505 [10:47<1:13:30,  1.53it/s]
 10%|█         | 778/7505 [10:47<1:13:28,  1.53it/s]
 10%|█         | 779/7505 [10:48<1:05:17,  1.72it/s]
 10%|█         | 780/7505 [10:48<1:10:00,  1.60it/s]
                                                    
{'loss': '2.93', 'grad_norm': '1.352', 'learning_rate': '0.0006', 'epoch': '41.05'}
+
 10%|█         | 780/7505 [10:48<1:10:00,  1.60it/s]
 10%|█         | 781/7505 [10:49<1:11:02,  1.58it/s]
 10%|█         | 782/7505 [10:50<1:11:43,  1.56it/s]
 10%|█         | 783/7505 [10:50<1:12:13,  1.55it/s]
 10%|█         | 784/7505 [10:51<1:12:37,  1.54it/s]
 10%|█         | 785/7505 [10:52<1:12:49,  1.54it/s]
 10%|█         | 786/7505 [10:52<1:13:00,  1.53it/s]
 10%|█         | 787/7505 [10:53<1:13:05,  1.53it/s]
 10%|█         | 788/7505 [10:54<1:13:11,  1.53it/s]
 11%|█         | 789/7505 [10:54<1:13:16,  1.53it/s]
 11%|█         | 790/7505 [10:55<1:13:14,  1.53it/s]
                                                    
{'loss': '2.771', 'grad_norm': '1.495', 'learning_rate': '0.0006', 'epoch': '41.59'}
+
 11%|█         | 790/7505 [10:55<1:13:14,  1.53it/s]
 11%|█         | 791/7505 [10:56<1:13:19,  1.53it/s]
 11%|█         | 792/7505 [10:56<1:13:18,  1.53it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.54it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.77it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.61it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.04it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.73it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.55it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.44it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.36it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.32it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.27it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.26it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.25it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.23it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.23it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.22it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.22it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.21it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.21it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.22it/s]
                                                    
+
                                               
{'eval_loss': '5.316', 'eval_runtime': '3.267', 'eval_samples_per_second': '747.2', 'eval_steps_per_second': '11.94', 'epoch': '41.7'}
+
 11%|█         | 792/7505 [11:00<1:13:18,  1.53it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.22it/s][eval] epoch 41.698 step 792 loss 5.3165 (best 4.7139 @ ep 30.322, 3/20 since best)
+
+
                                               
 11%|█         | 793/7505 [11:00<3:02:54,  1.64s/it]
 11%|█         | 794/7505 [11:01<2:30:01,  1.34s/it]
 11%|█         | 795/7505 [11:02<2:06:58,  1.14s/it]
 11%|█         | 796/7505 [11:02<1:50:49,  1.01it/s]
 11%|█         | 797/7505 [11:03<1:39:30,  1.12it/s]
 11%|█         | 798/7505 [11:03<1:23:29,  1.34it/s]
 11%|█         | 799/7505 [11:04<1:22:38,  1.35it/s]
 11%|█         | 800/7505 [11:05<1:19:46,  1.40it/s]
                                                    
{'loss': '2.787', 'grad_norm': '1.898', 'learning_rate': '0.0006', 'epoch': '42.11'}
+
 11%|█         | 800/7505 [11:05<1:19:46,  1.40it/s]
 11%|█         | 801/7505 [11:05<1:17:49,  1.44it/s]
 11%|█         | 802/7505 [11:06<1:16:25,  1.46it/s]
 11%|█         | 803/7505 [11:07<1:15:27,  1.48it/s]
 11%|█         | 804/7505 [11:07<1:14:48,  1.49it/s]
 11%|█         | 805/7505 [11:08<1:14:17,  1.50it/s]
 11%|█         | 806/7505 [11:09<1:13:58,  1.51it/s]
 11%|█         | 807/7505 [11:09<1:13:42,  1.51it/s]
 11%|█         | 808/7505 [11:10<1:13:34,  1.52it/s]
 11%|█         | 809/7505 [11:11<1:13:31,  1.52it/s]
 11%|█         | 810/7505 [11:11<1:13:21,  1.52it/s]
                                                    
{'loss': '2.677', 'grad_norm': '1.619', 'learning_rate': '0.0006', 'epoch': '42.64'}
+
 11%|█         | 810/7505 [11:11<1:13:21,  1.52it/s]
 11%|█         | 811/7505 [11:12<1:13:19,  1.52it/s]
 11%|█         | 812/7505 [11:13<1:13:16,  1.52it/s]
 11%|█         | 813/7505 [11:13<1:13:12,  1.52it/s]
 11%|█         | 814/7505 [11:14<1:13:12,  1.52it/s]
 11%|█         | 815/7505 [11:15<1:13:11,  1.52it/s]
 11%|█         | 816/7505 [11:15<1:13:09,  1.52it/s]
 11%|█         | 817/7505 [11:16<1:05:00,  1.71it/s]
 11%|█         | 818/7505 [11:16<1:09:37,  1.60it/s]
 11%|█         | 819/7505 [11:17<1:10:43,  1.58it/s]
 11%|█         | 820/7505 [11:18<1:11:25,  1.56it/s]
                                                    
{'loss': '2.674', 'grad_norm': '1.608', 'learning_rate': '0.0006', 'epoch': '43.16'}
+
 11%|█         | 820/7505 [11:18<1:11:25,  1.56it/s]
 11%|█         | 821/7505 [11:18<1:11:53,  1.55it/s]
 11%|█         | 822/7505 [11:19<1:12:15,  1.54it/s]
 11%|█         | 823/7505 [11:20<1:12:25,  1.54it/s]
 11%|█         | 824/7505 [11:20<1:12:36,  1.53it/s]
 11%|█         | 825/7505 [11:21<1:12:43,  1.53it/s]
 11%|█         | 826/7505 [11:22<1:12:46,  1.53it/s]
 11%|█         | 827/7505 [11:22<1:12:49,  1.53it/s]
 11%|█         | 828/7505 [11:23<1:12:51,  1.53it/s]
 11%|█         | 829/7505 [11:24<1:12:54,  1.53it/s]
 11%|█         | 830/7505 [11:24<1:12:55,  1.53it/s]
                                                    
{'loss': '2.565', 'grad_norm': '1.828', 'learning_rate': '0.0006', 'epoch': '43.7'}
+
 11%|█         | 830/7505 [11:24<1:12:55,  1.53it/s]
 11%|█         | 831/7505 [11:25<1:12:57,  1.52it/s]
 11%|█         | 832/7505 [11:25<1:12:57,  1.52it/s]
 11%|█         | 833/7505 [11:26<1:12:55,  1.52it/s]
 11%|█         | 834/7505 [11:27<1:12:57,  1.52it/s]
 11%|█         | 835/7505 [11:27<1:12:55,  1.52it/s]
 11%|█         | 836/7505 [11:28<1:04:48,  1.71it/s]
 11%|█         | 837/7505 [11:29<1:10:17,  1.58it/s]
 11%|█         | 838/7505 [11:29<1:11:06,  1.56it/s]
 11%|█         | 839/7505 [11:30<1:11:40,  1.55it/s]
 11%|█         | 840/7505 [11:31<1:12:04,  1.54it/s]
                                                    
{'loss': '2.563', 'grad_norm': '1.818', 'learning_rate': '0.0006', 'epoch': '44.21'}
+
 11%|█         | 840/7505 [11:31<1:12:04,  1.54it/s]
 11%|█         | 841/7505 [11:31<1:12:18,  1.54it/s]
 11%|█         | 842/7505 [11:32<1:12:32,  1.53it/s]
 11%|█         | 843/7505 [11:33<1:12:39,  1.53it/s]
 11%|█         | 844/7505 [11:33<1:12:42,  1.53it/s]
 11%|█▏        | 845/7505 [11:34<1:12:45,  1.53it/s]
 11%|█▏        | 846/7505 [11:35<1:12:45,  1.53it/s]
 11%|█▏        | 847/7505 [11:35<1:12:44,  1.53it/s]
 11%|█▏        | 848/7505 [11:36<1:12:48,  1.52it/s]
 11%|█▏        | 849/7505 [11:36<1:12:50,  1.52it/s]
 11%|█▏        | 850/7505 [11:37<1:12:48,  1.52it/s]
                                                    
{'loss': '2.474', 'grad_norm': '1.648', 'learning_rate': '0.0006', 'epoch': '44.75'}
+
 11%|█▏        | 850/7505 [11:37<1:12:48,  1.52it/s]
 11%|█▏        | 851/7505 [11:38<1:12:48,  1.52it/s]
 11%|█▏        | 852/7505 [11:38<1:12:47,  1.52it/s]
 11%|█▏        | 853/7505 [11:39<1:12:44,  1.52it/s]
 11%|█▏        | 854/7505 [11:40<1:12:46,  1.52it/s]
 11%|█▏        | 855/7505 [11:40<1:04:40,  1.71it/s]
 11%|█▏        | 856/7505 [11:41<1:09:22,  1.60it/s]
 11%|█▏        | 857/7505 [11:42<1:10:19,  1.58it/s]
 11%|█▏        | 858/7505 [11:42<1:11:01,  1.56it/s]
 11%|█▏        | 859/7505 [11:43<1:11:30,  1.55it/s]
 11%|█▏        | 860/7505 [11:44<1:11:50,  1.54it/s]
                                                    
{'loss': '2.432', 'grad_norm': '1.773', 'learning_rate': '0.0006', 'epoch': '45.27'}
+
 11%|█▏        | 860/7505 [11:44<1:11:50,  1.54it/s]
 11%|█▏        | 861/7505 [11:44<1:12:08,  1.53it/s]
 11%|█▏        | 862/7505 [11:45<1:12:20,  1.53it/s]
 11%|█▏        | 863/7505 [11:46<1:12:22,  1.53it/s]
 12%|█▏        | 864/7505 [11:46<1:12:28,  1.53it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 19.51it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 15.07it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.75it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.13it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.79it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.60it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.46it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.39it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.32it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.26it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.30it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.28it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.27it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.26it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.25it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.23it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.23it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.23it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.22it/s]
                                                    
+
                                               
{'eval_loss': '5.789', 'eval_runtime': '3.266', 'eval_samples_per_second': '747.5', 'eval_steps_per_second': '11.94', 'epoch': '45.48'}
+
 12%|█▏        | 864/7505 [11:49<1:12:28,  1.53it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.22it/s][eval] epoch 45.483 step 864 loss 5.7892 (best 4.7139 @ ep 30.322, 4/20 since best)
+
+
                                               
 12%|█▏        | 865/7505 [11:50<3:01:00,  1.64s/it]
 12%|█▏        | 866/7505 [11:51<2:28:52,  1.35s/it]
 12%|█▏        | 867/7505 [11:51<2:06:02,  1.14s/it]
 12%|█▏        | 868/7505 [11:52<1:50:04,  1.00it/s]
 12%|█▏        | 869/7505 [11:53<1:38:50,  1.12it/s]
 12%|█▏        | 870/7505 [11:53<1:30:58,  1.22it/s]
                                                    
{'loss': '2.381', 'grad_norm': '1.754', 'learning_rate': '0.0006', 'epoch': '45.81'}
+
 12%|█▏        | 870/7505 [11:53<1:30:58,  1.22it/s]
 12%|█▏        | 871/7505 [11:54<1:25:47,  1.29it/s]
 12%|█▏        | 872/7505 [11:55<1:21:51,  1.35it/s]
 12%|█▏        | 873/7505 [11:55<1:19:01,  1.40it/s]
 12%|█▏        | 874/7505 [11:56<1:08:58,  1.60it/s]
 12%|█▏        | 875/7505 [11:57<1:12:38,  1.52it/s]
 12%|█▏        | 876/7505 [11:57<1:12:35,  1.52it/s]
 12%|█▏        | 877/7505 [11:58<1:12:33,  1.52it/s]
 12%|█▏        | 878/7505 [11:58<1:12:31,  1.52it/s]
 12%|█▏        | 879/7505 [11:59<1:12:29,  1.52it/s]
 12%|█▏        | 880/7505 [12:00<1:12:29,  1.52it/s]
                                                    
{'loss': '2.324', 'grad_norm': '2.118', 'learning_rate': '0.0006', 'epoch': '46.32'}
+
 12%|█▏        | 880/7505 [12:00<1:12:29,  1.52it/s]
 12%|█▏        | 881/7505 [12:00<1:12:31,  1.52it/s]
 12%|█▏        | 882/7505 [12:01<1:12:34,  1.52it/s]
 12%|█▏        | 883/7505 [12:02<1:12:31,  1.52it/s]
 12%|█▏        | 884/7505 [12:02<1:12:29,  1.52it/s]
 12%|█▏        | 885/7505 [12:03<1:12:27,  1.52it/s]
 12%|█▏        | 886/7505 [12:04<1:12:25,  1.52it/s]
 12%|█▏        | 887/7505 [12:04<1:12:27,  1.52it/s]
 12%|█▏        | 888/7505 [12:05<1:12:29,  1.52it/s]
 12%|█▏        | 889/7505 [12:06<1:12:25,  1.52it/s]
 12%|█▏        | 890/7505 [12:06<1:12:24,  1.52it/s]
                                                    
{'loss': '2.3', 'grad_norm': '1.823', 'learning_rate': '0.0006', 'epoch': '46.86'}
+
 12%|█▏        | 890/7505 [12:06<1:12:24,  1.52it/s]
 12%|█▏        | 891/7505 [12:07<1:12:24,  1.52it/s]
 12%|█▏        | 892/7505 [12:08<1:12:18,  1.52it/s]
 12%|█▏        | 893/7505 [12:08<1:04:20,  1.71it/s]
 12%|█▏        | 894/7505 [12:09<1:09:21,  1.59it/s]
 12%|█▏        | 895/7505 [12:09<1:10:13,  1.57it/s]
 12%|█▏        | 896/7505 [12:10<1:10:49,  1.56it/s]
 12%|█▏        | 897/7505 [12:11<1:11:14,  1.55it/s]
 12%|█▏        | 898/7505 [12:11<1:11:38,  1.54it/s]
 12%|█▏        | 899/7505 [12:12<1:11:50,  1.53it/s]
 12%|█▏        | 900/7505 [12:13<1:11:54,  1.53it/s]
                                                    
{'loss': '2.233', 'grad_norm': '2.004', 'learning_rate': '0.0006', 'epoch': '47.38'}
+
 12%|█▏        | 900/7505 [12:13<1:11:54,  1.53it/s]
 12%|█▏        | 901/7505 [12:13<1:12:02,  1.53it/s]
 12%|█▏        | 902/7505 [12:14<1:12:04,  1.53it/s]
 12%|█▏        | 903/7505 [12:15<1:12:04,  1.53it/s]
 12%|█▏        | 904/7505 [12:15<1:12:09,  1.52it/s]
 12%|█▏        | 905/7505 [12:16<1:12:08,  1.52it/s]
 12%|█▏        | 906/7505 [12:17<1:12:07,  1.52it/s]
 12%|█▏        | 907/7505 [12:17<1:12:05,  1.53it/s]
 12%|█▏        | 908/7505 [12:18<1:12:06,  1.52it/s]
 12%|█▏        | 909/7505 [12:19<1:12:09,  1.52it/s]
 12%|█▏        | 910/7505 [12:19<1:12:09,  1.52it/s]
                                                    
{'loss': '2.232', 'grad_norm': '2.119', 'learning_rate': '0.0006', 'epoch': '47.91'}
+
 12%|█▏        | 910/7505 [12:19<1:12:09,  1.52it/s]
 12%|█▏        | 911/7505 [12:20<1:12:07,  1.52it/s]
 12%|█▏        | 912/7505 [12:20<1:04:07,  1.71it/s]
 12%|█▏        | 913/7505 [12:21<1:08:48,  1.60it/s]
 12%|█▏        | 914/7505 [12:22<1:09:46,  1.57it/s]
 12%|█▏        | 915/7505 [12:22<1:10:25,  1.56it/s]
 12%|█▏        | 916/7505 [12:23<1:10:57,  1.55it/s]
 12%|█▏        | 917/7505 [12:24<1:11:19,  1.54it/s]
 12%|█▏        | 918/7505 [12:24<1:11:29,  1.54it/s]
 12%|█▏        | 919/7505 [12:25<1:11:43,  1.53it/s]
 12%|█▏        | 920/7505 [12:26<1:11:47,  1.53it/s]
                                                    
{'loss': '2.123', 'grad_norm': '2.222', 'learning_rate': '0.0006', 'epoch': '48.43'}
+
 12%|█▏        | 920/7505 [12:26<1:11:47,  1.53it/s]
 12%|█▏        | 921/7505 [12:26<1:11:49,  1.53it/s]
 12%|█▏        | 922/7505 [12:27<1:11:54,  1.53it/s]
 12%|█▏        | 923/7505 [12:28<1:11:58,  1.52it/s]
 12%|█▏        | 924/7505 [12:28<1:11:57,  1.52it/s]
 12%|█▏        | 925/7505 [12:29<1:11:57,  1.52it/s]
 12%|█▏        | 926/7505 [12:30<1:11:55,  1.52it/s]
 12%|█▏        | 927/7505 [12:30<1:11:54,  1.52it/s]
 12%|█▏        | 928/7505 [12:31<1:11:55,  1.52it/s]
 12%|█▏        | 929/7505 [12:32<1:11:55,  1.52it/s]
 12%|█▏        | 930/7505 [12:32<1:11:52,  1.52it/s]
                                                    
{'loss': '2.157', 'grad_norm': '2.362', 'learning_rate': '0.0006', 'epoch': '48.97'}
+
 12%|█▏        | 930/7505 [12:32<1:11:52,  1.52it/s]
 12%|█▏        | 931/7505 [12:33<1:03:53,  1.71it/s]
 12%|█▏        | 932/7505 [12:33<1:08:33,  1.60it/s]
 12%|█▏        | 933/7505 [12:34<1:09:30,  1.58it/s]
 12%|█▏        | 934/7505 [12:35<1:10:16,  1.56it/s]
 12%|█▏        | 935/7505 [12:35<1:10:45,  1.55it/s]
 12%|█▏        | 936/7505 [12:36<1:11:04,  1.54it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.42it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.77it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.57it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.04it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.73it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.55it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.41it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.38it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.31it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.28it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.27it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.25it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.25it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.25it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.25it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.24it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.22it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.22it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.22it/s]
                                                    
+
                                               
{'eval_loss': '6.25', 'eval_runtime': '3.249', 'eval_samples_per_second': '751.2', 'eval_steps_per_second': '12', 'epoch': '49.27'}
+
 12%|█▏        | 936/7505 [12:39<1:11:04,  1.54it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.22it/s][eval] epoch 49.268 step 936 loss 6.2504 (best 4.7139 @ ep 30.322, 5/20 since best)
+
+
                                               
 12%|█▏        | 937/7505 [12:40<2:58:02,  1.63s/it]
 12%|█▏        | 938/7505 [12:41<2:26:12,  1.34s/it]
 13%|█▎        | 939/7505 [12:41<2:03:56,  1.13s/it]
 13%|█▎        | 940/7505 [12:42<1:48:19,  1.01it/s]
                                                    
{'loss': '2.018', 'grad_norm': '2.247', 'learning_rate': '0.0006', 'epoch': '49.48'}
+
 13%|█▎        | 940/7505 [12:42<1:48:19,  1.01it/s]
 13%|█▎        | 941/7505 [12:43<1:37:23,  1.12it/s]
 13%|█▎        | 942/7505 [12:43<1:29:48,  1.22it/s]
 13%|█▎        | 943/7505 [12:44<1:24:23,  1.30it/s]
 13%|█▎        | 944/7505 [12:45<1:20:34,  1.36it/s]
 13%|█▎        | 945/7505 [12:45<1:17:57,  1.40it/s]
 13%|█▎        | 946/7505 [12:46<1:16:06,  1.44it/s]
 13%|█▎        | 947/7505 [12:47<1:14:48,  1.46it/s]
 13%|█▎        | 948/7505 [12:47<1:13:53,  1.48it/s]
 13%|█▎        | 949/7505 [12:48<1:13:10,  1.49it/s]
 13%|█▎        | 950/7505 [12:48<1:04:47,  1.69it/s]
                                                    
{'loss': '2.082', 'grad_norm': '2.27', 'learning_rate': '0.0006', 'epoch': '50'}
+
 13%|█▎        | 950/7505 [12:48<1:04:47,  1.69it/s]
 13%|█▎        | 951/7505 [12:49<1:09:43,  1.57it/s]
 13%|█▎        | 952/7505 [12:50<1:10:18,  1.55it/s]
 13%|█▎        | 953/7505 [12:50<1:10:44,  1.54it/s]
 13%|█▎        | 954/7505 [12:51<1:10:58,  1.54it/s]
 13%|█▎        | 955/7505 [12:52<1:11:09,  1.53it/s]
 13%|█▎        | 956/7505 [12:52<1:11:14,  1.53it/s]
 13%|█▎        | 957/7505 [12:53<1:11:21,  1.53it/s]
 13%|█▎        | 958/7505 [12:54<1:11:28,  1.53it/s]
 13%|█▎        | 959/7505 [12:54<1:11:31,  1.53it/s]
 13%|█▎        | 960/7505 [12:55<1:11:30,  1.53it/s]
                                                    
{'loss': '1.93', 'grad_norm': '3.016', 'learning_rate': '0.0006', 'epoch': '50.54'}
+
 13%|█▎        | 960/7505 [12:55<1:11:30,  1.53it/s]
 13%|█▎        | 961/7505 [12:56<1:11:34,  1.52it/s]
 13%|█▎        | 962/7505 [12:56<1:11:32,  1.52it/s]
 13%|█▎        | 963/7505 [12:57<1:11:31,  1.52it/s]
 13%|█▎        | 964/7505 [12:58<1:11:33,  1.52it/s]
 13%|█▎        | 965/7505 [12:58<1:11:30,  1.52it/s]
 13%|█▎        | 966/7505 [12:59<1:11:31,  1.52it/s]
 13%|█▎        | 967/7505 [13:00<1:11:28,  1.52it/s]
 13%|█▎        | 968/7505 [13:00<1:11:27,  1.52it/s]
 13%|█▎        | 969/7505 [13:01<1:03:30,  1.72it/s]
 13%|█▎        | 970/7505 [13:01<1:08:45,  1.58it/s]
                                                    
{'loss': '2.022', 'grad_norm': '2.111', 'learning_rate': '0.0006', 'epoch': '51.05'}
+
 13%|█▎        | 970/7505 [13:01<1:08:45,  1.58it/s]
 13%|█▎        | 971/7505 [13:02<1:09:38,  1.56it/s]
 13%|█▎        | 972/7505 [13:03<1:10:10,  1.55it/s]
 13%|█▎        | 973/7505 [13:03<1:10:33,  1.54it/s]
 13%|█▎        | 974/7505 [13:04<1:10:46,  1.54it/s]
 13%|█▎        | 975/7505 [13:05<1:10:58,  1.53it/s]
 13%|█▎        | 976/7505 [13:05<1:11:07,  1.53it/s]
 13%|█▎        | 977/7505 [13:06<1:11:11,  1.53it/s]
 13%|█▎        | 978/7505 [13:07<1:11:14,  1.53it/s]
 13%|█▎        | 979/7505 [13:07<1:11:14,  1.53it/s]
 13%|█▎        | 980/7505 [13:08<1:11:15,  1.53it/s]
                                                    
{'loss': '1.894', 'grad_norm': '2.562', 'learning_rate': '0.0006', 'epoch': '51.59'}
+
 13%|█▎        | 980/7505 [13:08<1:11:15,  1.53it/s]
 13%|█▎        | 981/7505 [13:09<1:11:18,  1.52it/s]
 13%|█▎        | 982/7505 [13:09<1:11:18,  1.52it/s]
 13%|█▎        | 983/7505 [13:10<1:11:16,  1.52it/s]
 13%|█▎        | 984/7505 [13:11<1:11:17,  1.52it/s]
 13%|█▎        | 985/7505 [13:11<1:11:15,  1.52it/s]
 13%|█▎        | 986/7505 [13:12<1:11:19,  1.52it/s]
 13%|█▎        | 987/7505 [13:12<1:11:15,  1.52it/s]
 13%|█▎        | 988/7505 [13:13<1:03:21,  1.71it/s]
 13%|█▎        | 989/7505 [13:14<1:08:21,  1.59it/s]
 13%|█▎        | 990/7505 [13:14<1:09:12,  1.57it/s]
                                                    
{'loss': '1.943', 'grad_norm': '2.727', 'learning_rate': '0.0006', 'epoch': '52.11'}
+
 13%|█▎        | 990/7505 [13:14<1:09:12,  1.57it/s]
 13%|█▎        | 991/7505 [13:15<1:09:51,  1.55it/s]
 13%|█▎        | 992/7505 [13:16<1:10:14,  1.55it/s]
 13%|█▎        | 993/7505 [13:16<1:10:31,  1.54it/s]
 13%|█▎        | 994/7505 [13:17<1:10:44,  1.53it/s]
 13%|█▎        | 995/7505 [13:18<1:10:49,  1.53it/s]
 13%|█▎        | 996/7505 [13:18<1:10:57,  1.53it/s]
 13%|█▎        | 997/7505 [13:19<1:10:59,  1.53it/s]
 13%|█▎        | 998/7505 [13:20<1:11:01,  1.53it/s]
 13%|█▎        | 999/7505 [13:20<1:11:02,  1.53it/s]
 13%|█▎        | 1000/7505 [13:21<1:11:01,  1.53it/s]
                                                     
{'loss': '1.822', 'grad_norm': '2.63', 'learning_rate': '0.0006', 'epoch': '52.64'}
+
 13%|█▎        | 1000/7505 [13:21<1:11:01,  1.53it/s]
 13%|█▎        | 1001/7505 [13:21<1:11:06,  1.52it/s]
 13%|█▎        | 1002/7505 [13:22<1:11:07,  1.52it/s]
 13%|█▎        | 1003/7505 [13:23<1:11:04,  1.52it/s]
 13%|█▎        | 1004/7505 [13:23<1:11:06,  1.52it/s]
 13%|█▎        | 1005/7505 [13:24<1:11:03,  1.52it/s]
 13%|█▎        | 1006/7505 [13:25<1:11:03,  1.52it/s]
 13%|█▎        | 1007/7505 [13:25<1:03:08,  1.71it/s]
 13%|█▎        | 1008/7505 [13:26<1:08:11,  1.59it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.41it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.78it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.58it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.06it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.73it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.55it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.44it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.37it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.31it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.30it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.28it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.26it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.24it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.23it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.23it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.23it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.23it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.23it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s]
                                                     
+
                                               
{'eval_loss': '6.67', 'eval_runtime': '3.263', 'eval_samples_per_second': '748.1', 'eval_steps_per_second': '11.95', 'epoch': '53.05'}
+
 13%|█▎        | 1008/7505 [13:29<1:08:11,  1.59it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s][eval] epoch 53.054 step 1008 loss 6.6703 (best 4.7139 @ ep 30.322, 6/20 since best)
+
+
                                               
 13%|█▎        | 1009/7505 [13:30<2:54:59,  1.62s/it]
 13%|█▎        | 1010/7505 [13:31<2:23:45,  1.33s/it]
                                                     
{'loss': '1.848', 'grad_norm': '2.517', 'learning_rate': '0.0006', 'epoch': '53.16'}
+
 13%|█▎        | 1010/7505 [13:31<2:23:45,  1.33s/it]
 13%|█▎        | 1011/7505 [13:31<2:01:55,  1.13s/it]
 13%|█▎        | 1012/7505 [13:32<1:46:40,  1.01it/s]
 13%|█▎        | 1013/7505 [13:32<1:35:55,  1.13it/s]
 14%|█▎        | 1014/7505 [13:33<1:28:26,  1.22it/s]
 14%|█▎        | 1015/7505 [13:34<1:23:09,  1.30it/s]
 14%|█▎        | 1016/7505 [13:34<1:19:30,  1.36it/s]
 14%|█▎        | 1017/7505 [13:35<1:16:58,  1.40it/s]
 14%|█▎        | 1018/7505 [13:36<1:15:09,  1.44it/s]
 14%|█▎        | 1019/7505 [13:36<1:13:54,  1.46it/s]
 14%|█▎        | 1020/7505 [13:37<1:13:01,  1.48it/s]
                                                     
{'loss': '1.766', 'grad_norm': '2.478', 'learning_rate': '0.0006', 'epoch': '53.7'}
+
 14%|█▎        | 1020/7505 [13:37<1:13:01,  1.48it/s]
 14%|█▎        | 1021/7505 [13:38<1:12:23,  1.49it/s]
 14%|█▎        | 1022/7505 [13:38<1:11:55,  1.50it/s]
 14%|█▎        | 1023/7505 [13:39<1:11:39,  1.51it/s]
 14%|█▎        | 1024/7505 [13:40<1:11:25,  1.51it/s]
 14%|█▎        | 1025/7505 [13:40<1:11:10,  1.52it/s]
 14%|█▎        | 1026/7505 [13:41<1:03:10,  1.71it/s]
 14%|█▎        | 1027/7505 [13:41<1:08:09,  1.58it/s]
 14%|█▎        | 1028/7505 [13:42<1:08:56,  1.57it/s]
 14%|█▎        | 1029/7505 [13:43<1:09:27,  1.55it/s]
 14%|█▎        | 1030/7505 [13:43<1:09:51,  1.54it/s]
                                                     
{'loss': '1.764', 'grad_norm': '3.154', 'learning_rate': '0.0006', 'epoch': '54.21'}
+
 14%|█▎        | 1030/7505 [13:43<1:09:51,  1.54it/s]
 14%|█▎        | 1031/7505 [13:44<1:10:07,  1.54it/s]
 14%|█▍        | 1032/7505 [13:45<1:10:20,  1.53it/s]
 14%|█▍        | 1033/7505 [13:45<1:10:29,  1.53it/s]
 14%|█▍        | 1034/7505 [13:46<1:10:31,  1.53it/s]
 14%|█▍        | 1035/7505 [13:47<1:10:35,  1.53it/s]
 14%|█▍        | 1036/7505 [13:47<1:10:37,  1.53it/s]
 14%|█▍        | 1037/7505 [13:48<1:10:38,  1.53it/s]
 14%|█▍        | 1038/7505 [13:49<1:10:40,  1.53it/s]
 14%|█▍        | 1039/7505 [13:49<1:10:40,  1.52it/s]
 14%|█▍        | 1040/7505 [13:50<1:10:40,  1.52it/s]
                                                     
{'loss': '1.708', 'grad_norm': '2.574', 'learning_rate': '0.0006', 'epoch': '54.75'}
+
 14%|█▍        | 1040/7505 [13:50<1:10:40,  1.52it/s]
 14%|█▍        | 1041/7505 [13:51<1:10:40,  1.52it/s]
 14%|█▍        | 1042/7505 [13:51<1:10:39,  1.52it/s]
 14%|█▍        | 1043/7505 [13:52<1:10:40,  1.52it/s]
 14%|█▍        | 1044/7505 [13:53<1:10:37,  1.52it/s]
 14%|█▍        | 1045/7505 [13:53<1:02:45,  1.72it/s]
 14%|█▍        | 1046/7505 [13:54<1:07:40,  1.59it/s]
 14%|█▍        | 1047/7505 [13:54<1:08:36,  1.57it/s]
 14%|█▍        | 1048/7505 [13:55<1:09:10,  1.56it/s]
 14%|█▍        | 1049/7505 [13:56<1:09:34,  1.55it/s]
 14%|█▍        | 1050/7505 [13:56<1:09:53,  1.54it/s]
                                                     
{'loss': '1.687', 'grad_norm': '2.594', 'learning_rate': '0.0006', 'epoch': '55.27'}
+
 14%|█▍        | 1050/7505 [13:56<1:09:53,  1.54it/s]
 14%|█▍        | 1051/7505 [13:57<1:10:05,  1.53it/s]
 14%|█▍        | 1052/7505 [13:58<1:10:14,  1.53it/s]
 14%|█▍        | 1053/7505 [13:58<1:10:24,  1.53it/s]
 14%|█▍        | 1054/7505 [13:59<1:10:25,  1.53it/s]
 14%|█▍        | 1055/7505 [14:00<1:10:27,  1.53it/s]
 14%|█▍        | 1056/7505 [14:00<1:10:27,  1.53it/s]
 14%|█▍        | 1057/7505 [14:01<1:10:27,  1.53it/s]
 14%|█▍        | 1058/7505 [14:02<1:10:31,  1.52it/s]
 14%|█▍        | 1059/7505 [14:02<1:10:30,  1.52it/s]
 14%|█▍        | 1060/7505 [14:03<1:10:27,  1.52it/s]
                                                     
{'loss': '1.664', 'grad_norm': '2.685', 'learning_rate': '0.0006', 'epoch': '55.81'}
+
 14%|█▍        | 1060/7505 [14:03<1:10:27,  1.52it/s]
 14%|█▍        | 1061/7505 [14:04<1:10:30,  1.52it/s]
 14%|█▍        | 1062/7505 [14:04<1:10:26,  1.52it/s]
 14%|█▍        | 1063/7505 [14:05<1:10:26,  1.52it/s]
 14%|█▍        | 1064/7505 [14:05<1:02:35,  1.71it/s]
 14%|█▍        | 1065/7505 [14:06<1:07:37,  1.59it/s]
 14%|█▍        | 1066/7505 [14:07<1:08:27,  1.57it/s]
 14%|█▍        | 1067/7505 [14:07<1:09:00,  1.55it/s]
 14%|█▍        | 1068/7505 [14:08<1:09:25,  1.55it/s]
 14%|█▍        | 1069/7505 [14:09<1:09:43,  1.54it/s]
 14%|█▍        | 1070/7505 [14:09<1:09:55,  1.53it/s]
                                                     
{'loss': '1.62', 'grad_norm': '2.805', 'learning_rate': '0.0006', 'epoch': '56.32'}
+
 14%|█▍        | 1070/7505 [14:09<1:09:55,  1.53it/s]
 14%|█▍        | 1071/7505 [14:10<1:10:03,  1.53it/s]
 14%|█▍        | 1072/7505 [14:11<1:10:07,  1.53it/s]
 14%|█▍        | 1073/7505 [14:11<1:10:09,  1.53it/s]
 14%|█▍        | 1074/7505 [14:12<1:10:14,  1.53it/s]
 14%|█▍        | 1075/7505 [14:13<1:10:17,  1.52it/s]
 14%|█▍        | 1076/7505 [14:13<1:10:15,  1.53it/s]
 14%|█▍        | 1077/7505 [14:14<1:10:17,  1.52it/s]
 14%|█▍        | 1078/7505 [14:15<1:10:15,  1.52it/s]
 14%|█▍        | 1079/7505 [14:15<1:10:17,  1.52it/s]
 14%|█▍        | 1080/7505 [14:16<1:10:21,  1.52it/s]
                                                     
{'loss': '1.618', 'grad_norm': '2.879', 'learning_rate': '0.0006', 'epoch': '56.86'}
+
 14%|█▍        | 1080/7505 [14:16<1:10:21,  1.52it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.62it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.78it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.59it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.02it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.71it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.54it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.40it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.36it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.30it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.26it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.24it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.22it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.22it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.21it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.21it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.19it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.20it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.21it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.21it/s]
                                                     
+
                                               
{'eval_loss': '7.062', 'eval_runtime': '3.256', 'eval_samples_per_second': '749.7', 'eval_steps_per_second': '11.98', 'epoch': '56.86'}
+
 14%|█▍        | 1080/7505 [14:19<1:10:21,  1.52it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.21it/s][eval] epoch 56.859 step 1080 loss 7.0620 (best 4.7139 @ ep 30.322, 7/20 since best)
+
+
                                               
 14%|█▍        | 1081/7505 [14:20<2:55:03,  1.64s/it]
 14%|█▍        | 1082/7505 [14:21<2:23:33,  1.34s/it]
 14%|█▍        | 1083/7505 [14:21<1:53:41,  1.06s/it]
 14%|█▍        | 1084/7505 [14:22<1:43:16,  1.04it/s]
 14%|█▍        | 1085/7505 [14:22<1:33:21,  1.15it/s]
 14%|█▍        | 1086/7505 [14:23<1:26:26,  1.24it/s]
 14%|█▍        | 1087/7505 [14:24<1:21:33,  1.31it/s]
 14%|█▍        | 1088/7505 [14:24<1:18:09,  1.37it/s]
 15%|█▍        | 1089/7505 [14:25<1:15:42,  1.41it/s]
 15%|█▍        | 1090/7505 [14:26<1:14:03,  1.44it/s]
                                                     
{'loss': '1.551', 'grad_norm': '2.764', 'learning_rate': '0.0006', 'epoch': '57.38'}
+
 15%|█▍        | 1090/7505 [14:26<1:14:03,  1.44it/s]
 15%|█▍        | 1091/7505 [14:26<1:12:57,  1.47it/s]
 15%|█▍        | 1092/7505 [14:27<1:12:06,  1.48it/s]
 15%|█▍        | 1093/7505 [14:28<1:11:28,  1.50it/s]
 15%|█▍        | 1094/7505 [14:28<1:11:03,  1.50it/s]
 15%|█▍        | 1095/7505 [14:29<1:10:45,  1.51it/s]
 15%|█▍        | 1096/7505 [14:30<1:10:35,  1.51it/s]
 15%|█▍        | 1097/7505 [14:30<1:10:27,  1.52it/s]
 15%|█▍        | 1098/7505 [14:31<1:10:20,  1.52it/s]
 15%|█▍        | 1099/7505 [14:32<1:10:14,  1.52it/s]
 15%|█▍        | 1100/7505 [14:32<1:10:11,  1.52it/s]
                                                     
{'loss': '1.6', 'grad_norm': '2.94', 'learning_rate': '0.0006', 'epoch': '57.91'}
+
 15%|█▍        | 1100/7505 [14:32<1:10:11,  1.52it/s]
 15%|█▍        | 1101/7505 [14:33<1:10:06,  1.52it/s]
 15%|█▍        | 1102/7505 [14:33<1:02:19,  1.71it/s]
 15%|█▍        | 1103/7505 [14:34<1:07:17,  1.59it/s]
 15%|█▍        | 1104/7505 [14:35<1:08:08,  1.57it/s]
 15%|█▍        | 1105/7505 [14:35<1:08:42,  1.55it/s]
 15%|█▍        | 1106/7505 [14:36<1:09:02,  1.54it/s]
 15%|█▍        | 1107/7505 [14:37<1:09:19,  1.54it/s]
 15%|█▍        | 1108/7505 [14:37<1:09:33,  1.53it/s]
 15%|█▍        | 1109/7505 [14:38<1:09:41,  1.53it/s]
 15%|█▍        | 1110/7505 [14:39<1:09:43,  1.53it/s]
                                                     
{'loss': '1.534', 'grad_norm': '3.839', 'learning_rate': '0.0006', 'epoch': '58.43'}
+
 15%|█▍        | 1110/7505 [14:39<1:09:43,  1.53it/s]
 15%|█▍        | 1111/7505 [14:39<1:09:51,  1.53it/s]
 15%|█▍        | 1112/7505 [14:40<1:09:50,  1.53it/s]
 15%|█▍        | 1113/7505 [14:41<1:09:51,  1.53it/s]
 15%|█▍        | 1114/7505 [14:41<1:09:53,  1.52it/s]
 15%|█▍        | 1115/7505 [14:42<1:09:58,  1.52it/s]
 15%|█▍        | 1116/7505 [14:43<1:09:54,  1.52it/s]
 15%|█▍        | 1117/7505 [14:43<1:09:53,  1.52it/s]
 15%|█▍        | 1118/7505 [14:44<1:09:50,  1.52it/s]
 15%|█▍        | 1119/7505 [14:44<1:09:50,  1.52it/s]
 15%|█▍        | 1120/7505 [14:45<1:09:50,  1.52it/s]
                                                     
{'loss': '1.598', 'grad_norm': '3.035', 'learning_rate': '0.0006', 'epoch': '58.97'}
+
 15%|█▍        | 1120/7505 [14:45<1:09:50,  1.52it/s]
 15%|█▍        | 1121/7505 [14:46<1:02:04,  1.71it/s]
 15%|█▍        | 1122/7505 [14:46<1:07:28,  1.58it/s]
 15%|█▍        | 1123/7505 [14:47<1:08:11,  1.56it/s]
 15%|█▍        | 1124/7505 [14:48<1:08:41,  1.55it/s]
 15%|█▍        | 1125/7505 [14:48<1:08:59,  1.54it/s]
 15%|█▌        | 1126/7505 [14:49<1:09:12,  1.54it/s]
 15%|█▌        | 1127/7505 [14:50<1:09:20,  1.53it/s]
 15%|█▌        | 1128/7505 [14:50<1:09:26,  1.53it/s]
 15%|█▌        | 1129/7505 [14:51<1:09:33,  1.53it/s]
 15%|█▌        | 1130/7505 [14:52<1:09:35,  1.53it/s]
                                                     
{'loss': '1.476', 'grad_norm': '2.687', 'learning_rate': '0.0006', 'epoch': '59.48'}
+
 15%|█▌        | 1130/7505 [14:52<1:09:35,  1.53it/s]
 15%|█▌        | 1131/7505 [14:52<1:09:36,  1.53it/s]
 15%|█▌        | 1132/7505 [14:53<1:09:36,  1.53it/s]
 15%|█▌        | 1133/7505 [14:54<1:09:36,  1.53it/s]
 15%|█▌        | 1134/7505 [14:54<1:09:40,  1.52it/s]
 15%|█▌        | 1135/7505 [14:55<1:09:38,  1.52it/s]
 15%|█▌        | 1136/7505 [14:56<1:09:39,  1.52it/s]
 15%|█▌        | 1137/7505 [14:56<1:09:40,  1.52it/s]
 15%|█▌        | 1138/7505 [14:57<1:09:37,  1.52it/s]
 15%|█▌        | 1139/7505 [14:57<1:09:36,  1.52it/s]
 15%|█▌        | 1140/7505 [14:58<1:01:52,  1.71it/s]
                                                     
{'loss': '1.538', 'grad_norm': '3.269', 'learning_rate': '0.0006', 'epoch': '60'}
+
 15%|█▌        | 1140/7505 [14:58<1:01:52,  1.71it/s]
 15%|█▌        | 1141/7505 [14:59<1:06:52,  1.59it/s]
 15%|█▌        | 1142/7505 [14:59<1:07:41,  1.57it/s]
 15%|█▌        | 1143/7505 [15:00<1:08:14,  1.55it/s]
 15%|█▌        | 1144/7505 [15:01<1:08:37,  1.54it/s]
 15%|█▌        | 1145/7505 [15:01<1:08:56,  1.54it/s]
 15%|█▌        | 1146/7505 [15:02<1:09:10,  1.53it/s]
 15%|█▌        | 1147/7505 [15:03<1:09:16,  1.53it/s]
 15%|█▌        | 1148/7505 [15:03<1:09:19,  1.53it/s]
 15%|█▌        | 1149/7505 [15:04<1:09:24,  1.53it/s]
 15%|█▌        | 1150/7505 [15:05<1:09:24,  1.53it/s]
                                                     
{'loss': '1.406', 'grad_norm': '3.674', 'learning_rate': '0.0006', 'epoch': '60.54'}
+
 15%|█▌        | 1150/7505 [15:05<1:09:24,  1.53it/s]
 15%|█▌        | 1151/7505 [15:05<1:09:28,  1.52it/s]
 15%|█▌        | 1152/7505 [15:06<1:09:33,  1.52it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.02it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.56it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.49it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 12.98it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.69it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.50it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.39it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.33it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.28it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.26it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.26it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.25it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.23it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.22it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.22it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.22it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.22it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.22it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.22it/s]
                                                     
+
                                               
{'eval_loss': '7.343', 'eval_runtime': '3.254', 'eval_samples_per_second': '750.1', 'eval_steps_per_second': '11.98', 'epoch': '60.64'}
+
 15%|█▌        | 1152/7505 [15:09<1:09:33,  1.52it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.22it/s][eval] epoch 60.644 step 1152 loss 7.3426 (best 4.7139 @ ep 30.322, 8/20 since best)
+
+
                                               
 15%|█▌        | 1153/7505 [15:10<2:52:53,  1.63s/it]
 15%|█▌        | 1154/7505 [15:10<2:21:50,  1.34s/it]
 15%|█▌        | 1155/7505 [15:11<2:00:04,  1.13s/it]
 15%|█▌        | 1156/7505 [15:12<1:44:55,  1.01it/s]
 15%|█▌        | 1157/7505 [15:12<1:34:16,  1.12it/s]
 15%|█▌        | 1158/7505 [15:13<1:26:45,  1.22it/s]
 15%|█▌        | 1159/7505 [15:13<1:13:50,  1.43it/s]
 15%|█▌        | 1160/7505 [15:14<1:15:36,  1.40it/s]
                                                     
{'loss': '1.5', 'grad_norm': '3.122', 'learning_rate': '0.0006', 'epoch': '61.05'}
+
 15%|█▌        | 1160/7505 [15:14<1:15:36,  1.40it/s]
 15%|█▌        | 1161/7505 [15:15<1:13:45,  1.43it/s]
 15%|█▌        | 1162/7505 [15:16<1:12:26,  1.46it/s]
 15%|█▌        | 1163/7505 [15:16<1:11:31,  1.48it/s]
 16%|█▌        | 1164/7505 [15:17<1:10:50,  1.49it/s]
 16%|█▌        | 1165/7505 [15:17<1:10:24,  1.50it/s]
 16%|█▌        | 1166/7505 [15:18<1:10:06,  1.51it/s]
 16%|█▌        | 1167/7505 [15:19<1:09:56,  1.51it/s]
 16%|█▌        | 1168/7505 [15:19<1:09:43,  1.51it/s]
 16%|█▌        | 1169/7505 [15:20<1:09:35,  1.52it/s]
 16%|█▌        | 1170/7505 [15:21<1:09:28,  1.52it/s]
                                                     
{'loss': '1.374', 'grad_norm': '3.293', 'learning_rate': '0.0006', 'epoch': '61.59'}
+
 16%|█▌        | 1170/7505 [15:21<1:09:28,  1.52it/s]
 16%|█▌        | 1171/7505 [15:21<1:09:25,  1.52it/s]
 16%|█▌        | 1172/7505 [15:22<1:09:25,  1.52it/s]
 16%|█▌        | 1173/7505 [15:23<1:09:23,  1.52it/s]
 16%|█▌        | 1174/7505 [15:23<1:09:22,  1.52it/s]
 16%|█▌        | 1175/7505 [15:24<1:09:19,  1.52it/s]
 16%|█▌        | 1176/7505 [15:25<1:09:15,  1.52it/s]
 16%|█▌        | 1177/7505 [15:25<1:09:14,  1.52it/s]
 16%|█▌        | 1178/7505 [15:26<1:01:31,  1.71it/s]
 16%|█▌        | 1179/7505 [15:27<1:06:54,  1.58it/s]
 16%|█▌        | 1180/7505 [15:27<1:07:32,  1.56it/s]
                                                     
{'loss': '1.431', 'grad_norm': '3.137', 'learning_rate': '0.0006', 'epoch': '62.11'}
+
 16%|█▌        | 1180/7505 [15:27<1:07:32,  1.56it/s]
 16%|█▌        | 1181/7505 [15:28<1:08:02,  1.55it/s]
 16%|█▌        | 1182/7505 [15:29<1:08:25,  1.54it/s]
 16%|█▌        | 1183/7505 [15:29<1:08:40,  1.53it/s]
 16%|█▌        | 1184/7505 [15:30<1:08:50,  1.53it/s]
 16%|█▌        | 1185/7505 [15:30<1:08:54,  1.53it/s]
 16%|█▌        | 1186/7505 [15:31<1:08:58,  1.53it/s]
 16%|█▌        | 1187/7505 [15:32<1:08:58,  1.53it/s]
 16%|█▌        | 1188/7505 [15:32<1:09:01,  1.53it/s]
 16%|█▌        | 1189/7505 [15:33<1:09:03,  1.52it/s]
 16%|█▌        | 1190/7505 [15:34<1:09:01,  1.52it/s]
                                                     
{'loss': '1.33', 'grad_norm': '2.731', 'learning_rate': '0.0006', 'epoch': '62.64'}
+
 16%|█▌        | 1190/7505 [15:34<1:09:01,  1.52it/s]
 16%|█▌        | 1191/7505 [15:34<1:09:03,  1.52it/s]
 16%|█▌        | 1192/7505 [15:35<1:09:00,  1.52it/s]
 16%|█▌        | 1193/7505 [15:36<1:09:01,  1.52it/s]
 16%|█▌        | 1194/7505 [15:36<1:09:01,  1.52it/s]
 16%|█▌        | 1195/7505 [15:37<1:09:03,  1.52it/s]
 16%|█▌        | 1196/7505 [15:38<1:09:01,  1.52it/s]
 16%|█▌        | 1197/7505 [15:38<1:01:19,  1.71it/s]
 16%|█▌        | 1198/7505 [15:39<1:06:13,  1.59it/s]
 16%|█▌        | 1199/7505 [15:40<1:07:05,  1.57it/s]
 16%|█▌        | 1200/7505 [15:40<1:07:37,  1.55it/s]
                                                     
{'loss': '1.375', 'grad_norm': '2.83', 'learning_rate': '0.0006', 'epoch': '63.16'}
+
 16%|█▌        | 1200/7505 [15:40<1:07:37,  1.55it/s]
 16%|█▌        | 1201/7505 [15:41<1:08:00,  1.54it/s]
 16%|█▌        | 1202/7505 [15:41<1:08:16,  1.54it/s]
 16%|█▌        | 1203/7505 [15:42<1:08:25,  1.54it/s]
 16%|█▌        | 1204/7505 [15:43<1:08:36,  1.53it/s]
 16%|█▌        | 1205/7505 [15:43<1:08:40,  1.53it/s]
 16%|█▌        | 1206/7505 [15:44<1:08:44,  1.53it/s]
 16%|█▌        | 1207/7505 [15:45<1:08:49,  1.53it/s]
 16%|█▌        | 1208/7505 [15:45<1:08:48,  1.53it/s]
 16%|█▌        | 1209/7505 [15:46<1:08:48,  1.52it/s]
 16%|█▌        | 1210/7505 [15:47<1:08:51,  1.52it/s]
                                                     
{'loss': '1.318', 'grad_norm': '3.489', 'learning_rate': '0.0006', 'epoch': '63.7'}
+
 16%|█▌        | 1210/7505 [15:47<1:08:51,  1.52it/s]
 16%|█▌        | 1211/7505 [15:47<1:08:52,  1.52it/s]
 16%|█▌        | 1212/7505 [15:48<1:08:50,  1.52it/s]
 16%|█▌        | 1213/7505 [15:49<1:08:51,  1.52it/s]
 16%|█▌        | 1214/7505 [15:49<1:08:48,  1.52it/s]
 16%|█▌        | 1215/7505 [15:50<1:08:47,  1.52it/s]
 16%|█▌        | 1216/7505 [15:50<1:01:10,  1.71it/s]
 16%|█▌        | 1217/7505 [15:51<1:06:00,  1.59it/s]
 16%|█▌        | 1218/7505 [15:52<1:06:49,  1.57it/s]
 16%|█▌        | 1219/7505 [15:52<1:07:22,  1.56it/s]
 16%|█▋        | 1220/7505 [15:53<1:07:48,  1.54it/s]
                                                     
{'loss': '1.326', 'grad_norm': '3.167', 'learning_rate': '0.0006', 'epoch': '64.21'}
+
 16%|█▋        | 1220/7505 [15:53<1:07:48,  1.54it/s]
 16%|█▋        | 1221/7505 [15:54<1:08:08,  1.54it/s]
 16%|█▋        | 1222/7505 [15:54<1:08:22,  1.53it/s]
 16%|█▋        | 1223/7505 [15:55<1:08:29,  1.53it/s]
 16%|█▋        | 1224/7505 [15:56<1:08:31,  1.53it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.15it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.66it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.54it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.01it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.71it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.53it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.42it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.36it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.32it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.29it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.27it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.26it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.23it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.22it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.21it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.21it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.18it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.22it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.22it/s]
                                                     
+
                                               
{'eval_loss': '7.71', 'eval_runtime': '3.267', 'eval_samples_per_second': '747.2', 'eval_steps_per_second': '11.94', 'epoch': '64.43'}
+
 16%|█▋        | 1224/7505 [15:59<1:08:31,  1.53it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.22it/s][eval] epoch 64.43 step 1224 loss 7.7099 (best 4.7139 @ ep 30.322, 9/20 since best)
+
+
                                               
 16%|█▋        | 1225/7505 [16:00<2:51:09,  1.64s/it]
 16%|█▋        | 1226/7505 [16:00<2:20:22,  1.34s/it]
 16%|█▋        | 1227/7505 [16:01<1:58:50,  1.14s/it]
 16%|█▋        | 1228/7505 [16:02<1:43:45,  1.01it/s]
 16%|█▋        | 1229/7505 [16:02<1:33:14,  1.12it/s]
 16%|█▋        | 1230/7505 [16:03<1:25:50,  1.22it/s]
                                                     
{'loss': '1.287', 'grad_norm': '3.267', 'learning_rate': '0.0006', 'epoch': '64.75'}
+
 16%|█▋        | 1230/7505 [16:03<1:25:50,  1.22it/s]
 16%|█▋        | 1231/7505 [16:04<1:20:41,  1.30it/s]
 16%|█▋        | 1232/7505 [16:04<1:17:04,  1.36it/s]
 16%|█▋        | 1233/7505 [16:05<1:14:29,  1.40it/s]
 16%|█▋        | 1234/7505 [16:06<1:12:43,  1.44it/s]
 16%|█▋        | 1235/7505 [16:06<1:03:50,  1.64it/s]
 16%|█▋        | 1236/7505 [16:07<1:07:46,  1.54it/s]
 16%|█▋        | 1237/7505 [16:07<1:08:00,  1.54it/s]
 16%|█▋        | 1238/7505 [16:08<1:08:09,  1.53it/s]
 17%|█▋        | 1239/7505 [16:09<1:08:17,  1.53it/s]
 17%|█▋        | 1240/7505 [16:09<1:08:25,  1.53it/s]
                                                     
{'loss': '1.271', 'grad_norm': '2.599', 'learning_rate': '0.0006', 'epoch': '65.27'}
+
 17%|█▋        | 1240/7505 [16:09<1:08:25,  1.53it/s]
 17%|█▋        | 1241/7505 [16:10<1:08:33,  1.52it/s]
 17%|█▋        | 1242/7505 [16:11<1:08:33,  1.52it/s]
 17%|█▋        | 1243/7505 [16:11<1:08:30,  1.52it/s]
 17%|█▋        | 1244/7505 [16:12<1:08:31,  1.52it/s]
 17%|█▋        | 1245/7505 [16:13<1:08:30,  1.52it/s]
 17%|█▋        | 1246/7505 [16:13<1:08:28,  1.52it/s]
 17%|█▋        | 1247/7505 [16:14<1:08:31,  1.52it/s]
 17%|█▋        | 1248/7505 [16:15<1:08:31,  1.52it/s]
 17%|█▋        | 1249/7505 [16:15<1:08:27,  1.52it/s]
 17%|█▋        | 1250/7505 [16:16<1:08:28,  1.52it/s]
                                                     
{'loss': '1.266', 'grad_norm': '3.269', 'learning_rate': '0.0006', 'epoch': '65.81'}
+
 17%|█▋        | 1250/7505 [16:16<1:08:28,  1.52it/s]
 17%|█▋        | 1251/7505 [16:17<1:08:29,  1.52it/s]
 17%|█▋        | 1252/7505 [16:17<1:08:26,  1.52it/s]
 17%|█▋        | 1253/7505 [16:18<1:08:27,  1.52it/s]
 17%|█▋        | 1254/7505 [16:18<1:00:49,  1.71it/s]
 17%|█▋        | 1255/7505 [16:19<1:05:44,  1.58it/s]
 17%|█▋        | 1256/7505 [16:20<1:06:30,  1.57it/s]
 17%|█▋        | 1257/7505 [16:20<1:07:03,  1.55it/s]
 17%|█▋        | 1258/7505 [16:21<1:07:25,  1.54it/s]
 17%|█▋        | 1259/7505 [16:22<1:07:45,  1.54it/s]
 17%|█▋        | 1260/7505 [16:22<1:07:58,  1.53it/s]
                                                     
{'loss': '1.229', 'grad_norm': '3.714', 'learning_rate': '0.0006', 'epoch': '66.32'}
+
 17%|█▋        | 1260/7505 [16:22<1:07:58,  1.53it/s]
 17%|█▋        | 1261/7505 [16:23<1:08:09,  1.53it/s]
 17%|█▋        | 1262/7505 [16:24<1:08:12,  1.53it/s]
 17%|█▋        | 1263/7505 [16:24<1:08:15,  1.52it/s]
 17%|█▋        | 1264/7505 [16:25<1:08:15,  1.52it/s]
 17%|█▋        | 1265/7505 [16:26<1:08:17,  1.52it/s]
 17%|█▋        | 1266/7505 [16:26<1:08:15,  1.52it/s]
 17%|█▋        | 1267/7505 [16:27<1:08:15,  1.52it/s]
 17%|█▋        | 1268/7505 [16:28<1:08:21,  1.52it/s]
 17%|█▋        | 1269/7505 [16:28<1:08:20,  1.52it/s]
 17%|█▋        | 1270/7505 [16:29<1:08:18,  1.52it/s]
                                                     
{'loss': '1.229', 'grad_norm': '3.813', 'learning_rate': '0.0006', 'epoch': '66.86'}
+
 17%|█▋        | 1270/7505 [16:29<1:08:18,  1.52it/s]
 17%|█▋        | 1271/7505 [16:30<1:08:20,  1.52it/s]
 17%|█▋        | 1272/7505 [16:30<1:08:16,  1.52it/s]
 17%|█▋        | 1273/7505 [16:31<1:00:38,  1.71it/s]
 17%|█▋        | 1274/7505 [16:31<1:05:29,  1.59it/s]
 17%|█▋        | 1275/7505 [16:32<1:06:20,  1.57it/s]
 17%|█▋        | 1276/7505 [16:33<1:06:54,  1.55it/s]
 17%|█▋        | 1277/7505 [16:33<1:07:16,  1.54it/s]
 17%|█▋        | 1278/7505 [16:34<1:07:30,  1.54it/s]
 17%|█▋        | 1279/7505 [16:35<1:07:41,  1.53it/s]
 17%|█▋        | 1280/7505 [16:35<1:07:47,  1.53it/s]
                                                     
{'loss': '1.176', 'grad_norm': '3.25', 'learning_rate': '0.0006', 'epoch': '67.38'}
+
 17%|█▋        | 1280/7505 [16:35<1:07:47,  1.53it/s]
 17%|█▋        | 1281/7505 [16:36<1:07:55,  1.53it/s]
 17%|█▋        | 1282/7505 [16:37<1:08:00,  1.53it/s]
 17%|█▋        | 1283/7505 [16:37<1:08:00,  1.52it/s]
 17%|█▋        | 1284/7505 [16:38<1:07:59,  1.52it/s]
 17%|█▋        | 1285/7505 [16:39<1:08:03,  1.52it/s]
 17%|█▋        | 1286/7505 [16:39<1:08:02,  1.52it/s]
 17%|█▋        | 1287/7505 [16:40<1:08:00,  1.52it/s]
 17%|█▋        | 1288/7505 [16:41<1:08:02,  1.52it/s]
 17%|█▋        | 1289/7505 [16:41<1:08:02,  1.52it/s]
 17%|█▋        | 1290/7505 [16:42<1:08:03,  1.52it/s]
                                                     
{'loss': '1.231', 'grad_norm': '3.954', 'learning_rate': '0.0006', 'epoch': '67.91'}
+
 17%|█▋        | 1290/7505 [16:42<1:08:03,  1.52it/s]
 17%|█▋        | 1291/7505 [16:43<1:08:02,  1.52it/s]
 17%|█▋        | 1292/7505 [16:43<1:00:25,  1.71it/s]
 17%|█▋        | 1293/7505 [16:44<1:05:15,  1.59it/s]
 17%|█▋        | 1294/7505 [16:44<1:06:05,  1.57it/s]
 17%|█▋        | 1295/7505 [16:45<1:06:38,  1.55it/s]
 17%|█▋        | 1296/7505 [16:46<1:06:59,  1.54it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.82it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.86it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.65it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.06it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.75it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.55it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.43it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.36it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.31it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.28it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.27it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.26it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.25it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.22it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.24it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.23it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.23it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.22it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s]
                                                     
+
                                               
{'eval_loss': '7.98', 'eval_runtime': '3.249', 'eval_samples_per_second': '751.3', 'eval_steps_per_second': '12', 'epoch': '68.21'}
+
 17%|█▋        | 1296/7505 [16:49<1:06:59,  1.54it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s][eval] epoch 68.215 step 1296 loss 7.9798 (best 4.7139 @ ep 30.322, 10/20 since best)
+
+
                                               
 17%|█▋        | 1297/7505 [16:50<2:48:09,  1.63s/it]
 17%|█▋        | 1298/7505 [16:50<2:18:06,  1.34s/it]
 17%|█▋        | 1299/7505 [16:51<1:57:01,  1.13s/it]
 17%|█▋        | 1300/7505 [16:52<1:42:17,  1.01it/s]
                                                     
{'loss': '1.14', 'grad_norm': '3.02', 'learning_rate': '0.0006', 'epoch': '68.43'}
+
 17%|█▋        | 1300/7505 [16:52<1:42:17,  1.01it/s]
 17%|█▋        | 1301/7505 [16:52<1:32:01,  1.12it/s]
 17%|█▋        | 1302/7505 [16:53<1:24:44,  1.22it/s]
 17%|█▋        | 1303/7505 [16:54<1:19:40,  1.30it/s]
 17%|█▋        | 1304/7505 [16:54<1:16:09,  1.36it/s]
 17%|█▋        | 1305/7505 [16:55<1:13:39,  1.40it/s]
 17%|█▋        | 1306/7505 [16:56<1:11:54,  1.44it/s]
 17%|█▋        | 1307/7505 [16:56<1:10:41,  1.46it/s]
 17%|█▋        | 1308/7505 [16:57<1:09:47,  1.48it/s]
 17%|█▋        | 1309/7505 [16:57<1:09:13,  1.49it/s]
 17%|█▋        | 1310/7505 [16:58<1:08:45,  1.50it/s]
                                                     
{'loss': '1.179', 'grad_norm': '3.438', 'learning_rate': '0.0006', 'epoch': '68.97'}
+
 17%|█▋        | 1310/7505 [16:58<1:08:45,  1.50it/s]
 17%|█▋        | 1311/7505 [16:59<1:00:55,  1.69it/s]
 17%|█▋        | 1312/7505 [16:59<1:05:25,  1.58it/s]
 17%|█▋        | 1313/7505 [17:00<1:06:05,  1.56it/s]
 18%|█▊        | 1314/7505 [17:01<1:06:36,  1.55it/s]
 18%|█▊        | 1315/7505 [17:01<1:07:01,  1.54it/s]
 18%|█▊        | 1316/7505 [17:02<1:07:16,  1.53it/s]
 18%|█▊        | 1317/7505 [17:03<1:07:26,  1.53it/s]
 18%|█▊        | 1318/7505 [17:03<1:07:31,  1.53it/s]
 18%|█▊        | 1319/7505 [17:04<1:07:35,  1.53it/s]
 18%|█▊        | 1320/7505 [17:05<1:07:37,  1.52it/s]
                                                     
{'loss': '1.079', 'grad_norm': '2.598', 'learning_rate': '0.0006', 'epoch': '69.48'}
+
 18%|█▊        | 1320/7505 [17:05<1:07:37,  1.52it/s]
 18%|█▊        | 1321/7505 [17:05<1:07:38,  1.52it/s]
 18%|█▊        | 1322/7505 [17:06<1:07:39,  1.52it/s]
 18%|█▊        | 1323/7505 [17:07<1:07:43,  1.52it/s]
 18%|█▊        | 1324/7505 [17:07<1:07:45,  1.52it/s]
 18%|█▊        | 1325/7505 [17:08<1:07:46,  1.52it/s]
 18%|█▊        | 1326/7505 [17:08<1:07:42,  1.52it/s]
 18%|█▊        | 1327/7505 [17:09<1:07:40,  1.52it/s]
 18%|█▊        | 1328/7505 [17:10<1:07:41,  1.52it/s]
 18%|█▊        | 1329/7505 [17:10<1:07:35,  1.52it/s]
 18%|█▊        | 1330/7505 [17:11<1:00:07,  1.71it/s]
                                                     
{'loss': '1.219', 'grad_norm': '3.563', 'learning_rate': '0.0006', 'epoch': '70'}
+
 18%|█▊        | 1330/7505 [17:11<1:00:07,  1.71it/s]
 18%|█▊        | 1331/7505 [17:12<1:04:59,  1.58it/s]
 18%|█▊        | 1332/7505 [17:12<1:05:47,  1.56it/s]
 18%|█▊        | 1333/7505 [17:13<1:06:19,  1.55it/s]
 18%|█▊        | 1334/7505 [17:14<1:06:39,  1.54it/s]
 18%|█▊        | 1335/7505 [17:14<1:06:55,  1.54it/s]
 18%|█▊        | 1336/7505 [17:15<1:07:05,  1.53it/s]
 18%|█▊        | 1337/7505 [17:16<1:07:12,  1.53it/s]
 18%|█▊        | 1338/7505 [17:16<1:07:18,  1.53it/s]
 18%|█▊        | 1339/7505 [17:17<1:07:19,  1.53it/s]
 18%|█▊        | 1340/7505 [17:18<1:07:21,  1.53it/s]
                                                     
{'loss': '1.078', 'grad_norm': '3.129', 'learning_rate': '0.0006', 'epoch': '70.54'}
+
 18%|█▊        | 1340/7505 [17:18<1:07:21,  1.53it/s]
 18%|█▊        | 1341/7505 [17:18<1:07:24,  1.52it/s]
 18%|█▊        | 1342/7505 [17:19<1:07:24,  1.52it/s]
 18%|█▊        | 1343/7505 [17:20<1:07:23,  1.52it/s]
 18%|█▊        | 1344/7505 [17:20<1:07:26,  1.52it/s]
 18%|█▊        | 1345/7505 [17:21<1:07:25,  1.52it/s]
 18%|█▊        | 1346/7505 [17:21<1:07:23,  1.52it/s]
 18%|█▊        | 1347/7505 [17:22<1:07:24,  1.52it/s]
 18%|█▊        | 1348/7505 [17:23<1:08:27,  1.50it/s]
 18%|█▊        | 1349/7505 [17:23<1:01:05,  1.68it/s]
 18%|█▊        | 1350/7505 [17:24<1:05:33,  1.56it/s]
                                                     
{'loss': '1.167', 'grad_norm': '3.07', 'learning_rate': '0.0006', 'epoch': '71.05'}
+
 18%|█▊        | 1350/7505 [17:24<1:05:33,  1.56it/s]
 18%|█▊        | 1351/7505 [17:25<1:06:20,  1.55it/s]
 18%|█▊        | 1352/7505 [17:25<1:06:43,  1.54it/s]
 18%|█▊        | 1353/7505 [17:26<1:06:55,  1.53it/s]
 18%|█▊        | 1354/7505 [17:27<1:07:20,  1.52it/s]
 18%|█▊        | 1355/7505 [17:27<1:07:24,  1.52it/s]
 18%|█▊        | 1356/7505 [17:28<1:07:24,  1.52it/s]
 18%|█▊        | 1357/7505 [17:29<1:07:22,  1.52it/s]
 18%|█▊        | 1358/7505 [17:29<1:07:35,  1.52it/s]
 18%|█▊        | 1359/7505 [17:30<1:07:50,  1.51it/s]
 18%|█▊        | 1360/7505 [17:31<1:07:45,  1.51it/s]
                                                     
{'loss': '1.038', 'grad_norm': '3.142', 'learning_rate': '0.0006', 'epoch': '71.59'}
+
 18%|█▊        | 1360/7505 [17:31<1:07:45,  1.51it/s]
 18%|█▊        | 1361/7505 [17:31<1:07:45,  1.51it/s]
 18%|█▊        | 1362/7505 [17:32<1:07:50,  1.51it/s]
 18%|█▊        | 1363/7505 [17:33<1:08:02,  1.50it/s]
 18%|█▊        | 1364/7505 [17:33<1:07:59,  1.51it/s]
 18%|█▊        | 1365/7505 [17:34<1:08:22,  1.50it/s]
 18%|█▊        | 1366/7505 [17:35<1:08:27,  1.49it/s]
 18%|█▊        | 1367/7505 [17:35<1:08:37,  1.49it/s]
 18%|█▊        | 1368/7505 [17:36<1:01:12,  1.67it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:02, 17.99it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.33it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.25it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 12.76it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.46it/s]
+
 33%|███▎      | 13/39 [00:01<00:02, 12.28it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.26it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.15it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 11.81it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 11.60it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 11.67it/s]
+
 64%|██████▍   | 25/39 [00:02<00:01, 11.72it/s]
+
 69%|██████▉   | 27/39 [00:02<00:01, 10.58it/s]
+
 74%|███████▍  | 29/39 [00:02<00:01,  8.59it/s]
+
 77%|███████▋  | 30/39 [00:02<00:01,  7.92it/s]
+
 79%|███████▉  | 31/39 [00:02<00:01,  7.34it/s]
+
 82%|████████▏ | 32/39 [00:03<00:01,  6.90it/s]
+
 85%|████████▍ | 33/39 [00:03<00:00,  6.64it/s]
+
 87%|████████▋ | 34/39 [00:03<00:00,  6.41it/s]
+
 90%|████████▉ | 35/39 [00:03<00:00,  6.29it/s]
+
 92%|█████████▏| 36/39 [00:03<00:00,  6.16it/s]
+
 95%|█████████▍| 37/39 [00:03<00:00,  6.01it/s]
+
 97%|█████████▋| 38/39 [00:04<00:00,  5.96it/s]
+
100%|██████████| 39/39 [00:04<00:00,  5.91it/s]
                                                     
+
                                               
{'eval_loss': '8.196', 'eval_runtime': '4.546', 'eval_samples_per_second': '536.9', 'eval_steps_per_second': '8.578', 'epoch': '72'}
+
 18%|█▊        | 1368/7505 [17:40<1:01:12,  1.67it/s]
+
100%|██████████| 39/39 [00:04<00:00,  5.91it/s][eval] epoch 72.0 step 1368 loss 8.1963 (best 4.7139 @ ep 30.322, 11/20 since best)
+
+
                                               
 18%|█▊        | 1369/7505 [17:42<3:45:58,  2.21s/it]
 18%|█▊        | 1370/7505 [17:43<3:19:55,  1.96s/it]
                                                     
{'loss': '1.098', 'grad_norm': '3.363', 'learning_rate': '0.0006', 'epoch': '72.11'}
+
 18%|█▊        | 1370/7505 [17:43<3:19:55,  1.96s/it]
 18%|█▊        | 1371/7505 [17:44<3:01:38,  1.78s/it]
 18%|█▊        | 1372/7505 [17:46<2:47:58,  1.64s/it]
 18%|█▊        | 1373/7505 [17:47<2:38:33,  1.55s/it]
 18%|█▊        | 1374/7505 [17:48<2:31:02,  1.48s/it]
 18%|█▊        | 1375/7505 [17:50<2:25:40,  1.43s/it]
 18%|█▊        | 1376/7505 [17:51<2:22:19,  1.39s/it]
 18%|█▊        | 1377/7505 [17:52<2:19:06,  1.36s/it]
 18%|█▊        | 1378/7505 [17:54<2:18:29,  1.36s/it]
 18%|█▊        | 1379/7505 [17:55<2:17:13,  1.34s/it]
 18%|█▊        | 1380/7505 [17:56<2:16:48,  1.34s/it]
                                                     
{'loss': '1.013', 'grad_norm': '4.023', 'learning_rate': '0.0006', 'epoch': '72.64'}
+
 18%|█▊        | 1380/7505 [17:56<2:16:48,  1.34s/it]
 18%|█▊        | 1381/7505 [17:58<2:17:11,  1.34s/it]
 18%|█▊        | 1382/7505 [17:59<2:17:23,  1.35s/it]
 18%|█▊        | 1383/7505 [18:00<2:17:23,  1.35s/it]
 18%|█▊        | 1384/7505 [18:02<2:16:55,  1.34s/it]
 18%|█▊        | 1385/7505 [18:03<2:16:17,  1.34s/it]
 18%|█▊        | 1386/7505 [18:04<2:14:56,  1.32s/it]
 18%|█▊        | 1387/7505 [18:05<2:00:08,  1.18s/it]
 18%|█▊        | 1388/7505 [18:07<2:06:52,  1.24s/it]
 19%|█▊        | 1389/7505 [18:08<2:09:17,  1.27s/it]
 19%|█▊        | 1390/7505 [18:09<2:11:53,  1.29s/it]
                                                     
{'loss': '1.055', 'grad_norm': '3.447', 'learning_rate': '0.0006', 'epoch': '73.16'}
+
 19%|█▊        | 1390/7505 [18:09<2:11:53,  1.29s/it]
 19%|█▊        | 1391/7505 [18:11<2:13:55,  1.31s/it]
 19%|█▊        | 1392/7505 [18:12<2:14:31,  1.32s/it]
 19%|█▊        | 1393/7505 [18:14<2:24:34,  1.42s/it]
 19%|█▊        | 1394/7505 [18:15<2:22:27,  1.40s/it]
 19%|█▊        | 1395/7505 [18:16<2:20:42,  1.38s/it]
 19%|█▊        | 1396/7505 [18:18<2:18:36,  1.36s/it]
 19%|█▊        | 1397/7505 [18:19<2:16:57,  1.35s/it]
 19%|█▊        | 1398/7505 [18:20<2:15:57,  1.34s/it]
 19%|█▊        | 1399/7505 [18:21<2:15:13,  1.33s/it]
 19%|█▊        | 1400/7505 [18:23<2:15:29,  1.33s/it]
                                                     
{'loss': '1.002', 'grad_norm': '3.446', 'learning_rate': '0.0006', 'epoch': '73.7'}
+
 19%|█▊        | 1400/7505 [18:23<2:15:29,  1.33s/it]
 19%|█▊        | 1401/7505 [18:24<2:15:55,  1.34s/it]
 19%|█▊        | 1402/7505 [18:26<2:15:52,  1.34s/it]
 19%|█▊        | 1403/7505 [18:27<2:15:42,  1.33s/it]
 19%|█▊        | 1404/7505 [18:28<2:15:20,  1.33s/it]
 19%|█▊        | 1405/7505 [18:29<2:14:46,  1.33s/it]
 19%|█▊        | 1406/7505 [18:30<1:59:58,  1.18s/it]
 19%|█▊        | 1407/7505 [18:32<2:06:01,  1.24s/it]
 19%|█▉        | 1408/7505 [18:33<2:08:39,  1.27s/it]
 19%|█▉        | 1409/7505 [18:34<2:14:15,  1.32s/it]
 19%|█▉        | 1410/7505 [18:36<2:14:11,  1.32s/it]
                                                     
{'loss': '1.018', 'grad_norm': '3.771', 'learning_rate': '0.0006', 'epoch': '74.21'}
+
 19%|█▉        | 1410/7505 [18:36<2:14:11,  1.32s/it]
 19%|█▉        | 1411/7505 [18:37<2:15:13,  1.33s/it]
 19%|█▉        | 1412/7505 [18:38<2:15:04,  1.33s/it]
 19%|█▉        | 1413/7505 [18:40<2:15:41,  1.34s/it]
 19%|█▉        | 1414/7505 [18:41<2:15:11,  1.33s/it]
 19%|█▉        | 1415/7505 [18:42<2:15:25,  1.33s/it]
 19%|█▉        | 1416/7505 [18:44<2:15:24,  1.33s/it]
 19%|█▉        | 1417/7505 [18:45<2:15:13,  1.33s/it]
 19%|█▉        | 1418/7505 [18:46<2:14:44,  1.33s/it]
 19%|█▉        | 1419/7505 [18:48<2:14:07,  1.32s/it]
 19%|█▉        | 1420/7505 [18:49<2:14:44,  1.33s/it]
                                                     
{'loss': '0.9894', 'grad_norm': '3.023', 'learning_rate': '0.0006', 'epoch': '74.75'}
+
 19%|█▉        | 1420/7505 [18:49<2:14:44,  1.33s/it]
 19%|█▉        | 1421/7505 [18:50<2:15:45,  1.34s/it]
 19%|█▉        | 1422/7505 [18:52<2:16:15,  1.34s/it]
 19%|█▉        | 1423/7505 [18:53<2:16:23,  1.35s/it]
 19%|█▉        | 1424/7505 [18:55<2:16:10,  1.34s/it]
 19%|█▉        | 1425/7505 [18:55<2:01:42,  1.20s/it]
 19%|█▉        | 1426/7505 [18:57<2:08:44,  1.27s/it]
 19%|█▉        | 1427/7505 [18:58<2:10:54,  1.29s/it]
 19%|█▉        | 1428/7505 [19:00<2:12:20,  1.31s/it]
 19%|█▉        | 1429/7505 [19:01<2:12:37,  1.31s/it]
 19%|█▉        | 1430/7505 [19:02<2:12:31,  1.31s/it]
                                                     
{'loss': '0.9858', 'grad_norm': '3.238', 'learning_rate': '0.0006', 'epoch': '75.27'}
+
 19%|█▉        | 1430/7505 [19:02<2:12:31,  1.31s/it]
 19%|█▉        | 1431/7505 [19:03<2:13:03,  1.31s/it]
 19%|█▉        | 1432/7505 [19:05<2:12:53,  1.31s/it]
 19%|█▉        | 1433/7505 [19:06<2:14:11,  1.33s/it]
 19%|█▉        | 1434/7505 [19:07<2:14:34,  1.33s/it]
 19%|█▉        | 1435/7505 [19:09<2:14:43,  1.33s/it]
 19%|█▉        | 1436/7505 [19:10<2:14:54,  1.33s/it]
 19%|█▉        | 1437/7505 [19:11<2:14:56,  1.33s/it]
 19%|█▉        | 1438/7505 [19:13<2:15:39,  1.34s/it]
 19%|█▉        | 1439/7505 [19:14<2:15:23,  1.34s/it]
 19%|█▉        | 1440/7505 [19:15<2:14:27,  1.33s/it]
                                                     
{'loss': '0.9792', 'grad_norm': '3.269', 'learning_rate': '0.0006', 'epoch': '75.81'}
+
 19%|█▉        | 1440/7505 [19:15<2:14:27,  1.33s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:02, 12.38it/s]
+
 10%|█         | 4/39 [00:00<00:04,  7.50it/s]
+
 13%|█▎        | 5/39 [00:00<00:04,  6.91it/s]
+
 15%|█▌        | 6/39 [00:00<00:05,  6.50it/s]
+
 18%|█▊        | 7/39 [00:01<00:05,  6.36it/s]
+
 21%|██        | 8/39 [00:01<00:05,  6.17it/s]
+
 23%|██▎       | 9/39 [00:01<00:04,  6.04it/s]
+
 26%|██▌       | 10/39 [00:01<00:04,  5.92it/s]
+
 28%|██▊       | 11/39 [00:01<00:04,  5.92it/s]
+
 31%|███       | 12/39 [00:01<00:04,  5.83it/s]
+
 33%|███▎      | 13/39 [00:02<00:04,  5.84it/s]
+
 36%|███▌      | 14/39 [00:02<00:04,  5.79it/s]
+
 38%|███▊      | 15/39 [00:02<00:04,  5.93it/s]
+
 41%|████      | 16/39 [00:02<00:03,  5.92it/s]
+
 44%|████▎     | 17/39 [00:02<00:03,  5.91it/s]
+
 46%|████▌     | 18/39 [00:02<00:03,  5.87it/s]
+
 49%|████▊     | 19/39 [00:03<00:03,  5.88it/s]
+
 51%|█████▏    | 20/39 [00:03<00:03,  5.93it/s]
+
 54%|█████▍    | 21/39 [00:03<00:03,  5.80it/s]
+
 56%|█████▋    | 22/39 [00:03<00:02,  5.81it/s]
+
 59%|█████▉    | 23/39 [00:03<00:02,  5.84it/s]
+
 62%|██████▏   | 24/39 [00:03<00:02,  5.86it/s]
+
 64%|██████▍   | 25/39 [00:04<00:02,  5.85it/s]
+
 67%|██████▋   | 26/39 [00:04<00:02,  5.82it/s]
+
 69%|██████▉   | 27/39 [00:04<00:02,  5.76it/s]
+
 72%|███████▏  | 28/39 [00:04<00:01,  5.84it/s]
+
 74%|███████▍  | 29/39 [00:04<00:01,  5.85it/s]
+
 77%|███████▋  | 30/39 [00:04<00:01,  5.79it/s]
+
 79%|███████▉  | 31/39 [00:05<00:01,  5.80it/s]
+
 82%|████████▏ | 32/39 [00:05<00:01,  5.82it/s]
+
 85%|████████▍ | 33/39 [00:05<00:01,  5.80it/s]
+
 87%|████████▋ | 34/39 [00:05<00:00,  5.83it/s]
+
 90%|████████▉ | 35/39 [00:05<00:00,  5.80it/s]
+
 92%|█████████▏| 36/39 [00:05<00:00,  5.80it/s]
+
 95%|█████████▍| 37/39 [00:06<00:00,  5.80it/s]
+
 97%|█████████▋| 38/39 [00:06<00:00,  5.85it/s]
+
100%|██████████| 39/39 [00:06<00:00,  5.78it/s]
                                                     
+
                                               
{'eval_loss': '8.492', 'eval_runtime': '6.749', 'eval_samples_per_second': '361.7', 'eval_steps_per_second': '5.778', 'epoch': '75.81'}
+
 19%|█▉        | 1440/7505 [19:22<2:14:27,  1.33s/it]
+
100%|██████████| 39/39 [00:06<00:00,  5.78it/s][eval] epoch 75.805 step 1440 loss 8.4916 (best 4.7139 @ ep 30.322, 12/20 since best)
+
+
                                               
 19%|█▉        | 1441/7505 [19:24<5:39:36,  3.36s/it]
 19%|█▉        | 1442/7505 [19:25<4:38:20,  2.75s/it]
 19%|█▉        | 1443/7505 [19:26<3:55:46,  2.33s/it]
 19%|█▉        | 1444/7505 [19:27<3:10:23,  1.88s/it]
 19%|█▉        | 1445/7505 [19:29<2:55:37,  1.74s/it]
 19%|█▉        | 1446/7505 [19:30<2:43:12,  1.62s/it]
 19%|█▉        | 1447/7505 [19:31<2:33:44,  1.52s/it]
 19%|█▉        | 1448/7505 [19:32<2:28:04,  1.47s/it]
 19%|█▉        | 1449/7505 [19:34<2:23:39,  1.42s/it]
 19%|█▉        | 1450/7505 [19:35<2:20:52,  1.40s/it]
                                                     
{'loss': '0.9553', 'grad_norm': '3.558', 'learning_rate': '0.0006', 'epoch': '76.32'}
+
 19%|█▉        | 1450/7505 [19:35<2:20:52,  1.40s/it]
 19%|█▉        | 1451/7505 [19:36<2:19:39,  1.38s/it]
 19%|█▉        | 1452/7505 [19:38<2:18:19,  1.37s/it]
 19%|█▉        | 1453/7505 [19:39<2:17:13,  1.36s/it]
 19%|█▉        | 1454/7505 [19:40<2:16:15,  1.35s/it]
 19%|█▉        | 1455/7505 [19:42<2:15:11,  1.34s/it]
 19%|█▉        | 1456/7505 [19:43<2:14:34,  1.33s/it]
 19%|█▉        | 1457/7505 [19:44<2:14:10,  1.33s/it]
 19%|█▉        | 1458/7505 [19:46<2:13:51,  1.33s/it]
 19%|█▉        | 1459/7505 [19:47<2:13:55,  1.33s/it]
 19%|█▉        | 1460/7505 [19:48<2:14:27,  1.33s/it]
                                                     
{'loss': '0.9644', 'grad_norm': '3.279', 'learning_rate': '0.0006', 'epoch': '76.86'}
+
 19%|█▉        | 1460/7505 [19:48<2:14:27,  1.33s/it]
 19%|█▉        | 1461/7505 [19:50<2:15:39,  1.35s/it]
 19%|█▉        | 1462/7505 [19:51<2:15:52,  1.35s/it]
 19%|█▉        | 1463/7505 [19:52<2:01:04,  1.20s/it]
 20%|█▉        | 1464/7505 [19:53<2:08:09,  1.27s/it]
 20%|█▉        | 1465/7505 [19:55<2:10:34,  1.30s/it]
 20%|█▉        | 1466/7505 [19:56<2:20:00,  1.39s/it]
 20%|█▉        | 1467/7505 [19:58<2:18:30,  1.38s/it]
 20%|█▉        | 1468/7505 [19:59<2:17:12,  1.36s/it]
 20%|█▉        | 1469/7505 [20:00<2:16:03,  1.35s/it]
 20%|█▉        | 1470/7505 [20:02<2:15:38,  1.35s/it]
                                                     
{'loss': '0.9223', 'grad_norm': '3.124', 'learning_rate': '0.0006', 'epoch': '77.38'}
+
 20%|█▉        | 1470/7505 [20:02<2:15:38,  1.35s/it]
 20%|█▉        | 1471/7505 [20:03<2:16:09,  1.35s/it]
 20%|█▉        | 1472/7505 [20:05<2:25:10,  1.44s/it]
 20%|█▉        | 1473/7505 [20:06<2:22:18,  1.42s/it]
 20%|█▉        | 1474/7505 [20:08<2:30:08,  1.49s/it]
 20%|█▉        | 1475/7505 [20:09<2:25:57,  1.45s/it]
 20%|█▉        | 1476/7505 [20:11<2:22:53,  1.42s/it]
 20%|█▉        | 1477/7505 [20:12<2:19:58,  1.39s/it]
 20%|█▉        | 1478/7505 [20:13<2:18:14,  1.38s/it]
 20%|█▉        | 1479/7505 [20:15<2:16:20,  1.36s/it]
 20%|█▉        | 1480/7505 [20:16<2:24:33,  1.44s/it]
                                                     
{'loss': '0.9697', 'grad_norm': '3.347', 'learning_rate': '0.0006', 'epoch': '77.91'}
+
 20%|█▉        | 1480/7505 [20:16<2:24:33,  1.44s/it]
 20%|█▉        | 1481/7505 [20:18<2:22:13,  1.42s/it]
 20%|█▉        | 1482/7505 [20:18<2:05:06,  1.25s/it]
 20%|█▉        | 1483/7505 [20:20<2:09:36,  1.29s/it]
 20%|█▉        | 1484/7505 [20:21<2:11:34,  1.31s/it]
 20%|█▉        | 1485/7505 [20:22<2:12:20,  1.32s/it]
 20%|█▉        | 1486/7505 [20:24<2:13:08,  1.33s/it]
 20%|█▉        | 1487/7505 [20:25<2:13:37,  1.33s/it]
 20%|█▉        | 1488/7505 [20:26<2:13:45,  1.33s/it]
 20%|█▉        | 1489/7505 [20:28<2:13:49,  1.33s/it]
 20%|█▉        | 1490/7505 [20:29<2:13:58,  1.34s/it]
                                                     
{'loss': '0.8894', 'grad_norm': '3.461', 'learning_rate': '0.0006', 'epoch': '78.43'}
+
 20%|█▉        | 1490/7505 [20:29<2:13:58,  1.34s/it]
 20%|█▉        | 1491/7505 [20:30<2:14:04,  1.34s/it]
 20%|█▉        | 1492/7505 [20:32<2:14:14,  1.34s/it]
 20%|█▉        | 1493/7505 [20:33<2:14:14,  1.34s/it]
 20%|█▉        | 1494/7505 [20:35<2:14:17,  1.34s/it]
 20%|█▉        | 1495/7505 [20:36<2:14:13,  1.34s/it]
 20%|█▉        | 1496/7505 [20:37<2:14:23,  1.34s/it]
 20%|█▉        | 1497/7505 [20:39<2:14:22,  1.34s/it]
 20%|█▉        | 1498/7505 [20:40<2:14:24,  1.34s/it]
 20%|█▉        | 1499/7505 [20:41<2:14:19,  1.34s/it]
 20%|█▉        | 1500/7505 [20:43<2:13:57,  1.34s/it]
                                                     
{'loss': '0.9435', 'grad_norm': '3.536', 'learning_rate': '0.0006', 'epoch': '78.97'}
+
 20%|█▉        | 1500/7505 [20:43<2:13:57,  1.34s/it]
 20%|██        | 1501/7505 [20:43<1:59:05,  1.19s/it]
 20%|██        | 1502/7505 [20:45<2:04:25,  1.24s/it]
 20%|██        | 1503/7505 [20:46<2:07:29,  1.27s/it]
 20%|██        | 1504/7505 [20:47<2:09:29,  1.29s/it]
 20%|██        | 1505/7505 [20:49<2:11:23,  1.31s/it]
 20%|██        | 1506/7505 [20:50<2:12:13,  1.32s/it]
 20%|██        | 1507/7505 [20:52<2:13:11,  1.33s/it]
 20%|██        | 1508/7505 [20:53<2:13:27,  1.34s/it]
 20%|██        | 1509/7505 [20:54<2:14:08,  1.34s/it]
 20%|██        | 1510/7505 [20:56<2:14:34,  1.35s/it]
                                                     
{'loss': '0.8422', 'grad_norm': '3.331', 'learning_rate': '0.0006', 'epoch': '79.48'}
+
 20%|██        | 1510/7505 [20:56<2:14:34,  1.35s/it]
 20%|██        | 1511/7505 [20:57<2:14:46,  1.35s/it]
 20%|██        | 1512/7505 [20:58<2:13:59,  1.34s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:03, 11.96it/s]
+
 10%|█         | 4/39 [00:00<00:04,  7.41it/s]
+
 13%|█▎        | 5/39 [00:00<00:04,  6.88it/s]
+
 15%|█▌        | 6/39 [00:00<00:05,  6.54it/s]
+
 18%|█▊        | 7/39 [00:01<00:05,  6.35it/s]
+
 21%|██        | 8/39 [00:01<00:04,  6.21it/s]
+
 23%|██▎       | 9/39 [00:01<00:04,  6.09it/s]
+
 26%|██▌       | 10/39 [00:01<00:04,  5.99it/s]
+
 28%|██▊       | 11/39 [00:01<00:04,  5.90it/s]
+
 31%|███       | 12/39 [00:01<00:04,  5.92it/s]
+
 33%|███▎      | 13/39 [00:02<00:04,  5.82it/s]
+
 36%|███▌      | 14/39 [00:02<00:04,  5.84it/s]
+
 38%|███▊      | 15/39 [00:02<00:04,  5.80it/s]
+
 41%|████      | 16/39 [00:02<00:03,  5.82it/s]
+
 44%|████▎     | 17/39 [00:02<00:03,  5.86it/s]
+
 46%|████▌     | 18/39 [00:02<00:03,  5.82it/s]
+
 49%|████▊     | 19/39 [00:03<00:03,  5.86it/s]
+
 51%|█████▏    | 20/39 [00:03<00:03,  5.82it/s]
+
 54%|█████▍    | 21/39 [00:03<00:03,  5.83it/s]
+
 56%|█████▋    | 22/39 [00:03<00:02,  5.87it/s]
+
 59%|█████▉    | 23/39 [00:03<00:02,  5.88it/s]
+
 62%|██████▏   | 24/39 [00:03<00:02,  5.88it/s]
+
 64%|██████▍   | 25/39 [00:04<00:02,  5.83it/s]
+
 67%|██████▋   | 26/39 [00:04<00:02,  5.83it/s]
+
 69%|██████▉   | 27/39 [00:04<00:02,  5.87it/s]
+
 72%|███████▏  | 28/39 [00:04<00:01,  5.84it/s]
+
 74%|███████▍  | 29/39 [00:04<00:01,  5.89it/s]
+
 77%|███████▋  | 30/39 [00:04<00:01,  5.88it/s]
+
 79%|███████▉  | 31/39 [00:05<00:01,  5.81it/s]
+
 82%|████████▏ | 32/39 [00:05<00:01,  5.77it/s]
+
 85%|████████▍ | 33/39 [00:05<00:01,  5.80it/s]
+
 87%|████████▋ | 34/39 [00:05<00:00,  5.76it/s]
+
 90%|████████▉ | 35/39 [00:05<00:00,  5.73it/s]
+
 92%|█████████▏| 36/39 [00:06<00:00,  5.78it/s]
+
 95%|█████████▍| 37/39 [00:06<00:00,  5.74it/s]
+
 97%|█████████▋| 38/39 [00:06<00:00,  5.81it/s]
+
100%|██████████| 39/39 [00:06<00:00,  5.81it/s]
                                                     
+
                                               
{'eval_loss': '8.764', 'eval_runtime': '6.764', 'eval_samples_per_second': '360.9', 'eval_steps_per_second': '5.766', 'epoch': '79.59'}
+
 20%|██        | 1512/7505 [21:05<2:13:59,  1.34s/it]
+
100%|██████████| 39/39 [00:06<00:00,  5.81it/s][eval] epoch 79.591 step 1512 loss 8.7638 (best 4.7139 @ ep 30.322, 13/20 since best)
+
+
                                               
 20%|██        | 1513/7505 [21:06<5:36:33,  3.37s/it]
 20%|██        | 1514/7505 [21:08<4:36:01,  2.76s/it]
 20%|██        | 1515/7505 [21:09<3:53:40,  2.34s/it]
 20%|██        | 1516/7505 [21:10<3:23:35,  2.04s/it]
 20%|██        | 1517/7505 [21:12<3:02:53,  1.83s/it]
 20%|██        | 1518/7505 [21:13<2:47:56,  1.68s/it]
 20%|██        | 1519/7505 [21:14<2:37:34,  1.58s/it]
 20%|██        | 1520/7505 [21:15<2:15:51,  1.36s/it]
                                                     
{'loss': '0.9268', 'grad_norm': '3.069', 'learning_rate': '0.0006', 'epoch': '80'}
+
 20%|██        | 1520/7505 [21:15<2:15:51,  1.36s/it]
 20%|██        | 1521/7505 [21:17<2:17:52,  1.38s/it]
 20%|██        | 1522/7505 [21:18<2:16:26,  1.37s/it]
 20%|██        | 1523/7505 [21:19<2:16:10,  1.37s/it]
 20%|██        | 1524/7505 [21:21<2:15:48,  1.36s/it]
 20%|██        | 1525/7505 [21:22<2:15:17,  1.36s/it]
 20%|██        | 1526/7505 [21:23<2:15:04,  1.36s/it]
 20%|██        | 1527/7505 [21:25<2:14:46,  1.35s/it]
 20%|██        | 1528/7505 [21:26<2:14:30,  1.35s/it]
 20%|██        | 1529/7505 [21:27<2:13:31,  1.34s/it]
 20%|██        | 1530/7505 [21:29<2:13:30,  1.34s/it]
                                                     
{'loss': '0.8234', 'grad_norm': '3.739', 'learning_rate': '0.0006', 'epoch': '80.54'}
+
 20%|██        | 1530/7505 [21:29<2:13:30,  1.34s/it]
 20%|██        | 1531/7505 [21:30<2:12:53,  1.33s/it]
 20%|██        | 1532/7505 [21:31<2:12:50,  1.33s/it]
 20%|██        | 1533/7505 [21:33<2:13:25,  1.34s/it]
 20%|██        | 1534/7505 [21:34<2:13:44,  1.34s/it]
 20%|██        | 1535/7505 [21:36<2:14:01,  1.35s/it]
 20%|██        | 1536/7505 [21:37<2:14:02,  1.35s/it]
 20%|██        | 1537/7505 [21:38<2:14:04,  1.35s/it]
 20%|██        | 1538/7505 [21:40<2:13:33,  1.34s/it]
 21%|██        | 1539/7505 [21:40<1:59:05,  1.20s/it]
 21%|██        | 1540/7505 [21:42<2:04:40,  1.25s/it]
                                                     
{'loss': '0.907', 'grad_norm': '3.244', 'learning_rate': '0.0006', 'epoch': '81.05'}
+
 21%|██        | 1540/7505 [21:42<2:04:40,  1.25s/it]
 21%|██        | 1541/7505 [21:43<2:07:03,  1.28s/it]
 21%|██        | 1542/7505 [21:44<2:08:08,  1.29s/it]
 21%|██        | 1543/7505 [21:46<2:09:20,  1.30s/it]
 21%|██        | 1544/7505 [21:47<2:10:08,  1.31s/it]
 21%|██        | 1545/7505 [21:48<2:11:17,  1.32s/it]
 21%|██        | 1546/7505 [21:50<2:11:31,  1.32s/it]
 21%|██        | 1547/7505 [21:51<2:11:34,  1.33s/it]
 21%|██        | 1548/7505 [21:52<2:11:45,  1.33s/it]
 21%|██        | 1549/7505 [21:54<2:12:17,  1.33s/it]
 21%|██        | 1550/7505 [21:55<2:12:23,  1.33s/it]
                                                     
{'loss': '0.811', 'grad_norm': '3.097', 'learning_rate': '0.0006', 'epoch': '81.59'}
+
 21%|██        | 1550/7505 [21:55<2:12:23,  1.33s/it]
 21%|██        | 1551/7505 [21:56<2:12:31,  1.34s/it]
 21%|██        | 1552/7505 [21:58<2:11:57,  1.33s/it]
 21%|██        | 1553/7505 [21:59<2:11:37,  1.33s/it]
 21%|██        | 1554/7505 [22:00<2:11:47,  1.33s/it]
 21%|██        | 1555/7505 [22:02<2:12:11,  1.33s/it]
 21%|██        | 1556/7505 [22:03<2:12:37,  1.34s/it]
 21%|██        | 1557/7505 [22:04<2:13:00,  1.34s/it]
 21%|██        | 1558/7505 [22:05<1:58:40,  1.20s/it]
 21%|██        | 1559/7505 [22:07<2:05:07,  1.26s/it]
 21%|██        | 1560/7505 [22:08<2:07:58,  1.29s/it]
                                                     
{'loss': '0.8783', 'grad_norm': '3.108', 'learning_rate': '0.0006', 'epoch': '82.11'}
+
 21%|██        | 1560/7505 [22:08<2:07:58,  1.29s/it]
 21%|██        | 1561/7505 [22:09<2:10:23,  1.32s/it]
 21%|██        | 1562/7505 [22:11<2:11:28,  1.33s/it]
 21%|██        | 1563/7505 [22:12<2:12:19,  1.34s/it]
 21%|██        | 1564/7505 [22:14<2:12:33,  1.34s/it]
 21%|██        | 1565/7505 [22:15<2:11:47,  1.33s/it]
 21%|██        | 1566/7505 [22:16<2:11:57,  1.33s/it]
 21%|██        | 1567/7505 [22:18<2:11:58,  1.33s/it]
 21%|██        | 1568/7505 [22:19<2:12:17,  1.34s/it]
 21%|██        | 1569/7505 [22:20<2:12:19,  1.34s/it]
 21%|██        | 1570/7505 [22:22<2:12:40,  1.34s/it]
                                                     
{'loss': '0.8073', 'grad_norm': '3.697', 'learning_rate': '0.0006', 'epoch': '82.64'}
+
 21%|██        | 1570/7505 [22:22<2:12:40,  1.34s/it]
 21%|██        | 1571/7505 [22:23<2:13:02,  1.35s/it]
 21%|██        | 1572/7505 [22:24<2:12:45,  1.34s/it]
 21%|██        | 1573/7505 [22:26<2:12:38,  1.34s/it]
 21%|██        | 1574/7505 [22:27<2:12:48,  1.34s/it]
 21%|██        | 1575/7505 [22:28<2:11:48,  1.33s/it]
 21%|██        | 1576/7505 [22:30<2:12:25,  1.34s/it]
 21%|██        | 1577/7505 [22:30<1:58:06,  1.20s/it]
 21%|██        | 1578/7505 [22:32<2:04:40,  1.26s/it]
 21%|██        | 1579/7505 [22:33<2:07:02,  1.29s/it]
 21%|██        | 1580/7505 [22:35<2:09:10,  1.31s/it]
                                                     
{'loss': '0.8397', 'grad_norm': '3.239', 'learning_rate': '0.0006', 'epoch': '83.16'}
+
 21%|██        | 1580/7505 [22:35<2:09:10,  1.31s/it]
 21%|██        | 1581/7505 [22:36<2:11:08,  1.33s/it]
 21%|██        | 1582/7505 [22:37<2:11:54,  1.34s/it]
 21%|██        | 1583/7505 [22:39<2:12:35,  1.34s/it]
 21%|██        | 1584/7505 [22:40<2:13:07,  1.35s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:03, 11.42it/s]
+
 10%|█         | 4/39 [00:00<00:04,  7.07it/s]
+
 13%|█▎        | 5/39 [00:00<00:05,  6.69it/s]
+
 15%|█▌        | 6/39 [00:00<00:05,  6.35it/s]
+
 18%|█▊        | 7/39 [00:01<00:05,  6.20it/s]
+
 21%|██        | 8/39 [00:01<00:05,  6.11it/s]
+
 23%|██▎       | 9/39 [00:01<00:05,  6.00it/s]
+
 26%|██▌       | 10/39 [00:01<00:04,  5.87it/s]
+
 28%|██▊       | 11/39 [00:01<00:04,  5.85it/s]
+
 31%|███       | 12/39 [00:01<00:04,  5.82it/s]
+
 33%|███▎      | 13/39 [00:02<00:04,  5.80it/s]
+
 36%|███▌      | 14/39 [00:02<00:04,  5.86it/s]
+
 38%|███▊      | 15/39 [00:02<00:04,  5.80it/s]
+
 41%|████      | 16/39 [00:02<00:03,  5.80it/s]
+
 44%|████▎     | 17/39 [00:02<00:03,  5.85it/s]
+
 46%|████▌     | 18/39 [00:02<00:03,  5.92it/s]
+
 49%|████▊     | 19/39 [00:03<00:03,  5.93it/s]
+
 51%|█████▏    | 20/39 [00:03<00:03,  5.89it/s]
+
 54%|█████▍    | 21/39 [00:03<00:03,  5.87it/s]
+
 56%|█████▋    | 22/39 [00:03<00:02,  5.81it/s]
+
 59%|█████▉    | 23/39 [00:03<00:02,  5.79it/s]
+
 62%|██████▏   | 24/39 [00:03<00:02,  5.84it/s]
+
 64%|██████▍   | 25/39 [00:04<00:02,  5.87it/s]
+
 67%|██████▋   | 26/39 [00:04<00:02,  5.85it/s]
+
 69%|██████▉   | 27/39 [00:04<00:02,  5.83it/s]
+
 72%|███████▏  | 28/39 [00:04<00:01,  5.85it/s]
+
 74%|███████▍  | 29/39 [00:04<00:01,  5.84it/s]
+
 77%|███████▋  | 30/39 [00:04<00:01,  5.83it/s]
+
 79%|███████▉  | 31/39 [00:05<00:01,  5.85it/s]
+
 82%|████████▏ | 32/39 [00:05<00:01,  5.82it/s]
+
 85%|████████▍ | 33/39 [00:05<00:01,  5.82it/s]
+
 87%|████████▋ | 34/39 [00:05<00:00,  5.84it/s]
+
 90%|████████▉ | 35/39 [00:05<00:00,  5.82it/s]
+
 92%|█████████▏| 36/39 [00:06<00:00,  5.84it/s]
+
 95%|█████████▍| 37/39 [00:06<00:00,  5.82it/s]
+
 97%|█████████▋| 38/39 [00:06<00:00,  5.80it/s]
+
100%|██████████| 39/39 [00:06<00:00,  5.78it/s]
                                                     
+
                                               
{'eval_loss': '8.977', 'eval_runtime': '6.776', 'eval_samples_per_second': '360.2', 'eval_steps_per_second': '5.756', 'epoch': '83.38'}
+
 21%|██        | 1584/7505 [22:47<2:13:07,  1.35s/it]
+
100%|██████████| 39/39 [00:06<00:00,  5.78it/s][eval] epoch 83.376 step 1584 loss 8.9773 (best 4.7139 @ ep 30.322, 14/20 since best)
+
+
                                               
 21%|██        | 1585/7505 [22:48<5:33:38,  3.38s/it]
 21%|██        | 1586/7505 [22:50<4:33:34,  2.77s/it]
 21%|██        | 1587/7505 [22:51<3:51:56,  2.35s/it]
 21%|██        | 1588/7505 [22:52<3:22:33,  2.05s/it]
 21%|██        | 1589/7505 [22:54<3:02:00,  1.85s/it]
 21%|██        | 1590/7505 [22:55<2:47:32,  1.70s/it]
                                                     
{'loss': '0.7853', 'grad_norm': '2.98', 'learning_rate': '0.0006', 'epoch': '83.7'}
+
 21%|██        | 1590/7505 [22:55<2:47:32,  1.70s/it]
 21%|██        | 1591/7505 [22:56<2:37:19,  1.60s/it]
 21%|██        | 1592/7505 [22:58<2:30:12,  1.52s/it]
 21%|██        | 1593/7505 [22:59<2:25:04,  1.47s/it]
 21%|██        | 1594/7505 [23:00<2:21:19,  1.43s/it]
 21%|██▏       | 1595/7505 [23:02<2:18:22,  1.40s/it]
 21%|██▏       | 1596/7505 [23:03<2:01:43,  1.24s/it]
 21%|██▏       | 1597/7505 [23:04<2:06:37,  1.29s/it]
 21%|██▏       | 1598/7505 [23:05<2:07:42,  1.30s/it]
 21%|██▏       | 1599/7505 [23:07<2:08:51,  1.31s/it]
 21%|██▏       | 1600/7505 [23:08<2:09:37,  1.32s/it]
                                                     
{'loss': '0.8145', 'grad_norm': '3.543', 'learning_rate': '0.0006', 'epoch': '84.21'}
+
 21%|██▏       | 1600/7505 [23:08<2:09:37,  1.32s/it]
 21%|██▏       | 1601/7505 [23:09<2:10:38,  1.33s/it]
 21%|██▏       | 1602/7505 [23:11<2:10:41,  1.33s/it]
 21%|██▏       | 1603/7505 [23:12<2:10:49,  1.33s/it]
 21%|██▏       | 1604/7505 [23:13<2:11:16,  1.33s/it]
 21%|██▏       | 1605/7505 [23:15<2:11:27,  1.34s/it]
 21%|██▏       | 1606/7505 [23:16<2:11:47,  1.34s/it]
 21%|██▏       | 1607/7505 [23:17<2:10:56,  1.33s/it]
 21%|██▏       | 1608/7505 [23:19<2:10:58,  1.33s/it]
 21%|██▏       | 1609/7505 [23:20<2:10:59,  1.33s/it]
 21%|██▏       | 1610/7505 [23:21<2:10:48,  1.33s/it]
                                                     
{'loss': '0.7979', 'grad_norm': '3.774', 'learning_rate': '0.0006', 'epoch': '84.75'}
+
 21%|██▏       | 1610/7505 [23:21<2:10:48,  1.33s/it]
 21%|██▏       | 1611/7505 [23:23<2:11:22,  1.34s/it]
 21%|██▏       | 1612/7505 [23:24<2:11:08,  1.34s/it]
 21%|██▏       | 1613/7505 [23:25<2:10:38,  1.33s/it]
 22%|██▏       | 1614/7505 [23:27<2:09:54,  1.32s/it]
 22%|██▏       | 1615/7505 [23:27<1:55:48,  1.18s/it]
 22%|██▏       | 1616/7505 [23:29<2:01:33,  1.24s/it]
 22%|██▏       | 1617/7505 [23:30<2:03:21,  1.26s/it]
 22%|██▏       | 1618/7505 [23:31<2:05:14,  1.28s/it]
 22%|██▏       | 1619/7505 [23:33<2:05:30,  1.28s/it]
 22%|██▏       | 1620/7505 [23:34<2:06:31,  1.29s/it]
                                                     
{'loss': '0.8016', 'grad_norm': '3.215', 'learning_rate': '0.0006', 'epoch': '85.27'}
+
 22%|██▏       | 1620/7505 [23:34<2:06:31,  1.29s/it]
 22%|██▏       | 1621/7505 [23:35<2:05:19,  1.28s/it]
 22%|██▏       | 1622/7505 [23:37<2:05:16,  1.28s/it]
 22%|██▏       | 1623/7505 [23:38<2:05:30,  1.28s/it]
 22%|██▏       | 1624/7505 [23:39<2:05:40,  1.28s/it]
 22%|██▏       | 1625/7505 [23:40<2:05:46,  1.28s/it]
 22%|██▏       | 1626/7505 [23:42<2:06:04,  1.29s/it]
 22%|██▏       | 1627/7505 [23:43<2:04:24,  1.27s/it]
 22%|██▏       | 1628/7505 [23:44<2:03:31,  1.26s/it]
 22%|██▏       | 1629/7505 [23:45<2:03:40,  1.26s/it]
 22%|██▏       | 1630/7505 [23:47<2:03:42,  1.26s/it]
                                                     
{'loss': '0.7993', 'grad_norm': '4.033', 'learning_rate': '0.0006', 'epoch': '85.81'}
+
 22%|██▏       | 1630/7505 [23:47<2:03:42,  1.26s/it]
 22%|██▏       | 1631/7505 [23:48<2:04:02,  1.27s/it]
 22%|██▏       | 1632/7505 [23:49<2:03:46,  1.26s/it]
 22%|██▏       | 1633/7505 [23:51<2:03:22,  1.26s/it]
 22%|██▏       | 1634/7505 [23:51<1:49:57,  1.12s/it]
 22%|██▏       | 1635/7505 [23:53<1:55:20,  1.18s/it]
 22%|██▏       | 1636/7505 [23:54<1:57:31,  1.20s/it]
 22%|██▏       | 1637/7505 [23:55<1:58:50,  1.22s/it]
 22%|██▏       | 1638/7505 [23:56<1:59:35,  1.22s/it]
 22%|██▏       | 1639/7505 [23:58<2:00:30,  1.23s/it]
 22%|██▏       | 1640/7505 [23:59<2:01:16,  1.24s/it]
                                                     
{'loss': '0.7665', 'grad_norm': '3.771', 'learning_rate': '0.0006', 'epoch': '86.32'}
+
 22%|██▏       | 1640/7505 [23:59<2:01:16,  1.24s/it]
 22%|██▏       | 1641/7505 [24:00<2:01:42,  1.25s/it]
 22%|██▏       | 1642/7505 [24:01<2:01:52,  1.25s/it]
 22%|██▏       | 1643/7505 [24:03<2:01:41,  1.25s/it]
 22%|██▏       | 1644/7505 [24:04<2:01:12,  1.24s/it]
 22%|██▏       | 1645/7505 [24:05<2:01:25,  1.24s/it]
 22%|██▏       | 1646/7505 [24:06<2:01:39,  1.25s/it]
 22%|██▏       | 1647/7505 [24:08<2:02:02,  1.25s/it]
 22%|██▏       | 1648/7505 [24:09<2:02:20,  1.25s/it]
 22%|██▏       | 1649/7505 [24:10<2:01:55,  1.25s/it]
 22%|██▏       | 1650/7505 [24:11<2:02:00,  1.25s/it]
                                                     
{'loss': '0.7837', 'grad_norm': '3.496', 'learning_rate': '0.0006', 'epoch': '86.86'}
+
 22%|██▏       | 1650/7505 [24:11<2:02:00,  1.25s/it]
 22%|██▏       | 1651/7505 [24:13<2:02:18,  1.25s/it]
 22%|██▏       | 1652/7505 [24:14<2:02:00,  1.25s/it]
 22%|██▏       | 1653/7505 [24:15<1:48:43,  1.11s/it]
 22%|██▏       | 1654/7505 [24:16<1:54:46,  1.18s/it]
 22%|██▏       | 1655/7505 [24:17<1:57:11,  1.20s/it]
 22%|██▏       | 1656/7505 [24:19<1:58:59,  1.22s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:02, 12.94it/s]
+
 10%|█         | 4/39 [00:00<00:04,  7.90it/s]
+
 13%|█▎        | 5/39 [00:00<00:04,  7.29it/s]
+
 15%|█▌        | 6/39 [00:00<00:04,  6.91it/s]
+
 18%|█▊        | 7/39 [00:00<00:04,  6.73it/s]
+
 21%|██        | 8/39 [00:01<00:04,  6.59it/s]
+
 23%|██▎       | 9/39 [00:01<00:04,  6.46it/s]
+
 26%|██▌       | 10/39 [00:01<00:04,  6.41it/s]
+
 28%|██▊       | 11/39 [00:01<00:04,  6.31it/s]
+
 31%|███       | 12/39 [00:01<00:04,  6.25it/s]
+
 33%|███▎      | 13/39 [00:01<00:04,  6.24it/s]
+
 36%|███▌      | 14/39 [00:02<00:04,  6.21it/s]
+
 38%|███▊      | 15/39 [00:02<00:03,  6.21it/s]
+
 41%|████      | 16/39 [00:02<00:03,  6.17it/s]
+
 44%|████▎     | 17/39 [00:02<00:03,  6.25it/s]
+
 46%|████▌     | 18/39 [00:02<00:03,  6.27it/s]
+
 49%|████▊     | 19/39 [00:02<00:03,  6.29it/s]
+
 51%|█████▏    | 20/39 [00:03<00:03,  6.24it/s]
+
 54%|█████▍    | 21/39 [00:03<00:02,  6.26it/s]
+
 56%|█████▋    | 22/39 [00:03<00:02,  6.25it/s]
+
 59%|█████▉    | 23/39 [00:03<00:02,  6.32it/s]
+
 62%|██████▏   | 24/39 [00:03<00:02,  6.29it/s]
+
 64%|██████▍   | 25/39 [00:03<00:02,  6.37it/s]
+
 67%|██████▋   | 26/39 [00:04<00:02,  6.27it/s]
+
 69%|██████▉   | 27/39 [00:04<00:01,  6.34it/s]
+
 72%|███████▏  | 28/39 [00:04<00:01,  6.28it/s]
+
 74%|███████▍  | 29/39 [00:04<00:01,  6.27it/s]
+
 77%|███████▋  | 30/39 [00:04<00:01,  6.28it/s]
+
 79%|███████▉  | 31/39 [00:04<00:01,  6.34it/s]
+
 82%|████████▏ | 32/39 [00:04<00:01,  6.32it/s]
+
 85%|████████▍ | 33/39 [00:05<00:00,  6.29it/s]
+
 87%|████████▋ | 34/39 [00:05<00:00,  6.23it/s]
+
 90%|████████▉ | 35/39 [00:05<00:00,  6.26it/s]
+
 92%|█████████▏| 36/39 [00:05<00:00,  6.20it/s]
+
 95%|█████████▍| 37/39 [00:05<00:00,  6.20it/s]
+
 97%|█████████▋| 38/39 [00:05<00:00,  6.20it/s]
+
100%|██████████| 39/39 [00:06<00:00,  6.20it/s]
                                                     
+
                                               
{'eval_loss': '9.162', 'eval_runtime': '6.291', 'eval_samples_per_second': '388', 'eval_steps_per_second': '6.199', 'epoch': '87.16'}
+
 22%|██▏       | 1656/7505 [24:25<1:58:59,  1.22s/it]
+
100%|██████████| 39/39 [00:06<00:00,  6.20it/s][eval] epoch 87.161 step 1656 loss 9.1623 (best 4.7139 @ ep 30.322, 15/20 since best)
+
+
                                               
 22%|██▏       | 1657/7505 [24:26<5:03:40,  3.12s/it]
 22%|██▏       | 1658/7505 [24:27<4:08:41,  2.55s/it]
 22%|██▏       | 1659/7505 [24:29<3:30:48,  2.16s/it]
 22%|██▏       | 1660/7505 [24:30<3:04:09,  1.89s/it]
                                                     
{'loss': '0.7242', 'grad_norm': '3.091', 'learning_rate': '0.0006', 'epoch': '87.38'}
+
 22%|██▏       | 1660/7505 [24:30<3:04:09,  1.89s/it]
 22%|██▏       | 1661/7505 [24:31<2:45:44,  1.70s/it]
 22%|██▏       | 1662/7505 [24:32<2:32:08,  1.56s/it]
 22%|██▏       | 1663/7505 [24:34<2:22:59,  1.47s/it]
 22%|██▏       | 1664/7505 [24:35<2:16:54,  1.41s/it]
 22%|██▏       | 1665/7505 [24:36<2:12:13,  1.36s/it]
 22%|██▏       | 1666/7505 [24:37<2:08:52,  1.32s/it]
 22%|██▏       | 1667/7505 [24:39<2:06:39,  1.30s/it]
 22%|██▏       | 1668/7505 [24:40<2:05:09,  1.29s/it]
 22%|██▏       | 1669/7505 [24:41<2:04:26,  1.28s/it]
 22%|██▏       | 1670/7505 [24:42<2:03:38,  1.27s/it]
                                                     
{'loss': '0.7659', 'grad_norm': '3.487', 'learning_rate': '0.0006', 'epoch': '87.91'}
+
 22%|██▏       | 1670/7505 [24:42<2:03:38,  1.27s/it]
 22%|██▏       | 1671/7505 [24:44<2:02:56,  1.26s/it]
 22%|██▏       | 1672/7505 [24:44<1:49:28,  1.13s/it]
 22%|██▏       | 1673/7505 [24:46<1:55:08,  1.18s/it]
 22%|██▏       | 1674/7505 [24:47<1:57:00,  1.20s/it]
 22%|██▏       | 1675/7505 [24:48<1:58:19,  1.22s/it]
 22%|██▏       | 1676/7505 [24:49<1:59:09,  1.23s/it]
 22%|██▏       | 1677/7505 [24:51<1:59:21,  1.23s/it]
 22%|██▏       | 1678/7505 [24:52<1:59:43,  1.23s/it]
 22%|██▏       | 1679/7505 [24:53<2:00:37,  1.24s/it]
 22%|██▏       | 1680/7505 [24:54<2:01:02,  1.25s/it]
                                                     
{'loss': '0.7065', 'grad_norm': '4.136', 'learning_rate': '0.0006', 'epoch': '88.43'}
+
 22%|██▏       | 1680/7505 [24:54<2:01:02,  1.25s/it]
 22%|██▏       | 1681/7505 [24:56<2:02:01,  1.26s/it]
 22%|██▏       | 1682/7505 [24:57<2:02:02,  1.26s/it]
 22%|██▏       | 1683/7505 [24:58<2:01:40,  1.25s/it]
 22%|██▏       | 1684/7505 [24:59<2:01:38,  1.25s/it]
 22%|██▏       | 1685/7505 [25:01<2:01:01,  1.25s/it]
 22%|██▏       | 1686/7505 [25:02<2:00:33,  1.24s/it]
 22%|██▏       | 1687/7505 [25:03<2:01:10,  1.25s/it]
 22%|██▏       | 1688/7505 [25:04<2:01:26,  1.25s/it]
 23%|██▎       | 1689/7505 [25:06<2:01:43,  1.26s/it]
 23%|██▎       | 1690/7505 [25:07<2:01:34,  1.25s/it]
                                                     
{'loss': '0.7709', 'grad_norm': '3.339', 'learning_rate': '0.0006', 'epoch': '88.97'}
+
 23%|██▎       | 1690/7505 [25:07<2:01:34,  1.25s/it]
 23%|██▎       | 1691/7505 [25:08<1:48:32,  1.12s/it]
 23%|██▎       | 1692/7505 [25:09<1:53:34,  1.17s/it]
 23%|██▎       | 1693/7505 [25:10<1:55:39,  1.19s/it]
 23%|██▎       | 1694/7505 [25:12<1:57:16,  1.21s/it]
 23%|██▎       | 1695/7505 [25:13<1:58:01,  1.22s/it]
 23%|██▎       | 1696/7505 [25:14<1:58:47,  1.23s/it]
 23%|██▎       | 1697/7505 [25:15<1:59:38,  1.24s/it]
 23%|██▎       | 1698/7505 [25:17<1:59:55,  1.24s/it]
 23%|██▎       | 1699/7505 [25:18<1:59:52,  1.24s/it]
 23%|██▎       | 1700/7505 [25:19<2:00:10,  1.24s/it]
                                                     
{'loss': '0.6895', 'grad_norm': '3.167', 'learning_rate': '0.0006', 'epoch': '89.48'}
+
 23%|██▎       | 1700/7505 [25:19<2:00:10,  1.24s/it]
 23%|██▎       | 1701/7505 [25:20<2:00:25,  1.24s/it]
 23%|██▎       | 1702/7505 [25:22<2:00:49,  1.25s/it]
 23%|██▎       | 1703/7505 [25:23<2:01:11,  1.25s/it]
 23%|██▎       | 1704/7505 [25:24<2:01:28,  1.26s/it]
 23%|██▎       | 1705/7505 [25:25<2:01:29,  1.26s/it]
 23%|██▎       | 1706/7505 [25:27<2:01:22,  1.26s/it]
 23%|██▎       | 1707/7505 [25:28<2:00:39,  1.25s/it]
 23%|██▎       | 1708/7505 [25:29<2:00:38,  1.25s/it]
 23%|██▎       | 1709/7505 [25:30<2:00:05,  1.24s/it]
 23%|██▎       | 1710/7505 [25:31<1:47:12,  1.11s/it]
                                                     
{'loss': '0.7652', 'grad_norm': '3.589', 'learning_rate': '0.0006', 'epoch': '90'}
+
 23%|██▎       | 1710/7505 [25:31<1:47:12,  1.11s/it]
 23%|██▎       | 1711/7505 [25:32<1:53:51,  1.18s/it]
 23%|██▎       | 1712/7505 [25:34<1:55:57,  1.20s/it]
 23%|██▎       | 1713/7505 [25:35<1:57:42,  1.22s/it]
 23%|██▎       | 1714/7505 [25:36<1:59:13,  1.24s/it]
 23%|██▎       | 1715/7505 [25:38<2:00:05,  1.24s/it]
 23%|██▎       | 1716/7505 [25:39<2:00:37,  1.25s/it]
 23%|██▎       | 1717/7505 [25:40<2:00:23,  1.25s/it]
 23%|██▎       | 1718/7505 [25:41<2:00:46,  1.25s/it]
 23%|██▎       | 1719/7505 [25:43<2:00:34,  1.25s/it]
 23%|██▎       | 1720/7505 [25:44<2:00:48,  1.25s/it]
                                                     
{'loss': '0.657', 'grad_norm': '3.122', 'learning_rate': '0.0006', 'epoch': '90.54'}
+
 23%|██▎       | 1720/7505 [25:44<2:00:48,  1.25s/it]
 23%|██▎       | 1721/7505 [25:45<2:00:50,  1.25s/it]
 23%|██▎       | 1722/7505 [25:46<2:00:25,  1.25s/it]
 23%|██▎       | 1723/7505 [25:48<2:00:05,  1.25s/it]
 23%|██▎       | 1724/7505 [25:49<2:00:19,  1.25s/it]
 23%|██▎       | 1725/7505 [25:50<2:00:14,  1.25s/it]
 23%|██▎       | 1726/7505 [25:51<2:00:01,  1.25s/it]
 23%|██▎       | 1727/7505 [25:53<2:00:32,  1.25s/it]
 23%|██▎       | 1728/7505 [25:54<2:00:12,  1.25s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:02, 12.58it/s]
+
 10%|█         | 4/39 [00:00<00:04,  7.86it/s]
+
 13%|█▎        | 5/39 [00:00<00:04,  7.22it/s]
+
 15%|█▌        | 6/39 [00:00<00:04,  6.88it/s]
+
 18%|█▊        | 7/39 [00:00<00:04,  6.65it/s]
+
 21%|██        | 8/39 [00:01<00:04,  6.55it/s]
+
 23%|██▎       | 9/39 [00:01<00:04,  6.39it/s]
+
 26%|██▌       | 10/39 [00:01<00:04,  6.35it/s]
+
 28%|██▊       | 11/39 [00:01<00:04,  6.25it/s]
+
 31%|███       | 12/39 [00:01<00:04,  6.29it/s]
+
 33%|███▎      | 13/39 [00:01<00:04,  6.23it/s]
+
 36%|███▌      | 14/39 [00:02<00:04,  6.24it/s]
+
 38%|███▊      | 15/39 [00:02<00:03,  6.28it/s]
+
 41%|████      | 16/39 [00:02<00:03,  6.20it/s]
+
 44%|████▎     | 17/39 [00:02<00:03,  6.21it/s]
+
 46%|████▌     | 18/39 [00:02<00:03,  6.21it/s]
+
 49%|████▊     | 19/39 [00:02<00:03,  6.18it/s]
+
 51%|█████▏    | 20/39 [00:03<00:03,  6.14it/s]
+
 54%|█████▍    | 21/39 [00:03<00:02,  6.16it/s]
+
 56%|█████▋    | 22/39 [00:03<00:02,  6.14it/s]
+
 59%|█████▉    | 23/39 [00:03<00:02,  6.19it/s]
+
 62%|██████▏   | 24/39 [00:03<00:02,  6.18it/s]
+
 64%|██████▍   | 25/39 [00:03<00:02,  6.20it/s]
+
 67%|██████▋   | 26/39 [00:04<00:02,  6.23it/s]
+
 69%|██████▉   | 27/39 [00:04<00:01,  6.12it/s]
+
 72%|███████▏  | 28/39 [00:04<00:01,  6.11it/s]
+
 74%|███████▍  | 29/39 [00:04<00:01,  6.08it/s]
+
 77%|███████▋  | 30/39 [00:04<00:01,  6.15it/s]
+
 79%|███████▉  | 31/39 [00:04<00:01,  6.12it/s]
+
 82%|████████▏ | 32/39 [00:05<00:01,  6.14it/s]
+
 85%|████████▍ | 33/39 [00:05<00:00,  6.14it/s]
+
 87%|████████▋ | 34/39 [00:05<00:00,  6.09it/s]
+
 90%|████████▉ | 35/39 [00:05<00:00,  6.16it/s]
+
 92%|█████████▏| 36/39 [00:05<00:00,  6.12it/s]
+
 95%|█████████▍| 37/39 [00:05<00:00,  6.15it/s]
+
 97%|█████████▋| 38/39 [00:06<00:00,  6.13it/s]
+
100%|██████████| 39/39 [00:06<00:00,  6.18it/s]
                                                     
+
                                               
{'eval_loss': '9.328', 'eval_runtime': '6.375', 'eval_samples_per_second': '382.9', 'eval_steps_per_second': '6.118', 'epoch': '90.97'}
+
 23%|██▎       | 1728/7505 [26:00<2:00:12,  1.25s/it]
+
100%|██████████| 39/39 [00:06<00:00,  6.18it/s][eval] epoch 90.966 step 1728 loss 9.3278 (best 4.7139 @ ep 30.322, 16/20 since best)
+
+
                                               
 23%|██▎       | 1729/7505 [26:01<4:51:32,  3.03s/it]
 23%|██▎       | 1730/7505 [26:02<4:03:01,  2.52s/it]
                                                     
{'loss': '0.7405', 'grad_norm': '2.652', 'learning_rate': '0.0006', 'epoch': '91.05'}
+
 23%|██▎       | 1730/7505 [26:02<4:03:01,  2.52s/it]
 23%|██▎       | 1731/7505 [26:04<3:27:01,  2.15s/it]
 23%|██▎       | 1732/7505 [26:05<3:01:23,  1.89s/it]
 23%|██▎       | 1733/7505 [26:06<2:43:37,  1.70s/it]
 23%|██▎       | 1734/7505 [26:07<2:31:25,  1.57s/it]
 23%|██▎       | 1735/7505 [26:09<2:22:18,  1.48s/it]
 23%|██▎       | 1736/7505 [26:10<2:16:01,  1.41s/it]
 23%|██▎       | 1737/7505 [26:11<2:11:02,  1.36s/it]
 23%|██▎       | 1738/7505 [26:12<2:07:40,  1.33s/it]
 23%|██▎       | 1739/7505 [26:14<2:05:15,  1.30s/it]
 23%|██▎       | 1740/7505 [26:15<2:03:54,  1.29s/it]
                                                     
{'loss': '0.6755', 'grad_norm': '3.691', 'learning_rate': '0.0006', 'epoch': '91.59'}
+
 23%|██▎       | 1740/7505 [26:15<2:03:54,  1.29s/it]
 23%|██▎       | 1741/7505 [26:16<2:02:52,  1.28s/it]
 23%|██▎       | 1742/7505 [26:17<2:01:47,  1.27s/it]
 23%|██▎       | 1743/7505 [26:19<2:01:13,  1.26s/it]
 23%|██▎       | 1744/7505 [26:20<2:00:52,  1.26s/it]
 23%|██▎       | 1745/7505 [26:21<2:00:33,  1.26s/it]
 23%|██▎       | 1746/7505 [26:22<2:00:26,  1.25s/it]
 23%|██▎       | 1747/7505 [26:24<2:00:00,  1.25s/it]
 23%|██▎       | 1748/7505 [26:24<1:46:33,  1.11s/it]
 23%|██▎       | 1749/7505 [26:26<1:52:08,  1.17s/it]
 23%|██▎       | 1750/7505 [26:27<1:54:26,  1.19s/it]
                                                     
{'loss': '0.7313', 'grad_norm': '3.175', 'learning_rate': '0.0006', 'epoch': '92.11'}
+
 23%|██▎       | 1750/7505 [26:27<1:54:26,  1.19s/it]
 23%|██▎       | 1751/7505 [26:28<1:56:36,  1.22s/it]
 23%|██▎       | 1752/7505 [26:30<1:57:38,  1.23s/it]
 23%|██▎       | 1753/7505 [26:31<1:58:12,  1.23s/it]
 23%|██▎       | 1754/7505 [26:32<1:58:56,  1.24s/it]
 23%|██▎       | 1755/7505 [26:33<1:59:30,  1.25s/it]
 23%|██▎       | 1756/7505 [26:35<1:59:48,  1.25s/it]
 23%|██▎       | 1757/7505 [26:36<1:59:24,  1.25s/it]
 23%|██▎       | 1758/7505 [26:37<1:59:39,  1.25s/it]
 23%|██▎       | 1759/7505 [26:38<1:59:51,  1.25s/it]
 23%|██▎       | 1760/7505 [26:40<1:59:53,  1.25s/it]
                                                     
{'loss': '0.6613', 'grad_norm': '3.266', 'learning_rate': '0.0006', 'epoch': '92.64'}
+
 23%|██▎       | 1760/7505 [26:40<1:59:53,  1.25s/it]
 23%|██▎       | 1761/7505 [26:41<2:00:01,  1.25s/it]
 23%|██▎       | 1762/7505 [26:42<2:00:13,  1.26s/it]
 23%|██▎       | 1763/7505 [26:43<2:00:26,  1.26s/it]
 24%|██▎       | 1764/7505 [26:45<2:00:18,  1.26s/it]
 24%|██▎       | 1765/7505 [26:46<2:00:19,  1.26s/it]
 24%|██▎       | 1766/7505 [26:47<1:59:59,  1.25s/it]
 24%|██▎       | 1767/7505 [26:48<1:46:55,  1.12s/it]
 24%|██▎       | 1768/7505 [26:49<1:53:17,  1.18s/it]
 24%|██▎       | 1769/7505 [26:50<1:54:50,  1.20s/it]
 24%|██▎       | 1770/7505 [26:52<1:56:33,  1.22s/it]
                                                     
{'loss': '0.6981', 'grad_norm': '2.923', 'learning_rate': '0.0006', 'epoch': '93.16'}
+
 24%|██▎       | 1770/7505 [26:52<1:56:33,  1.22s/it]
 24%|██▎       | 1771/7505 [26:53<1:57:29,  1.23s/it]
 24%|██▎       | 1772/7505 [26:54<1:58:03,  1.24s/it]
 24%|██▎       | 1773/7505 [26:55<1:58:40,  1.24s/it]
 24%|██▎       | 1774/7505 [26:57<1:59:09,  1.25s/it]
 24%|██▎       | 1775/7505 [26:58<1:59:16,  1.25s/it]
 24%|██▎       | 1776/7505 [26:59<1:58:57,  1.25s/it]
 24%|██▎       | 1777/7505 [27:00<1:59:08,  1.25s/it]
 24%|██▎       | 1778/7505 [27:02<1:59:14,  1.25s/it]
 24%|██▎       | 1779/7505 [27:03<1:59:33,  1.25s/it]
 24%|██▎       | 1780/7505 [27:04<1:59:41,  1.25s/it]
                                                     
{'loss': '0.6609', 'grad_norm': '3.328', 'learning_rate': '0.0006', 'epoch': '93.7'}
+
 24%|██▎       | 1780/7505 [27:04<1:59:41,  1.25s/it]
 24%|██▎       | 1781/7505 [27:06<1:59:48,  1.26s/it]
 24%|██▎       | 1782/7505 [27:07<1:59:51,  1.26s/it]
 24%|██▍       | 1783/7505 [27:08<1:59:15,  1.25s/it]
 24%|██▍       | 1784/7505 [27:09<1:58:27,  1.24s/it]
 24%|██▍       | 1785/7505 [27:10<1:58:39,  1.24s/it]
 24%|██▍       | 1786/7505 [27:11<1:45:31,  1.11s/it]
 24%|██▍       | 1787/7505 [27:13<1:52:05,  1.18s/it]
 24%|██▍       | 1788/7505 [27:14<1:54:43,  1.20s/it]
 24%|██▍       | 1789/7505 [27:15<1:56:15,  1.22s/it]
 24%|██▍       | 1790/7505 [27:16<1:57:15,  1.23s/it]
                                                     
{'loss': '0.6831', 'grad_norm': '3.316', 'learning_rate': '0.0006', 'epoch': '94.21'}
+
 24%|██▍       | 1790/7505 [27:16<1:57:15,  1.23s/it]
 24%|██▍       | 1791/7505 [27:18<1:58:09,  1.24s/it]
 24%|██▍       | 1792/7505 [27:19<1:59:25,  1.25s/it]
 24%|██▍       | 1793/7505 [27:20<2:00:28,  1.27s/it]
 24%|██▍       | 1794/7505 [27:22<2:01:26,  1.28s/it]
 24%|██▍       | 1795/7505 [27:23<2:01:57,  1.28s/it]
 24%|██▍       | 1796/7505 [27:24<2:01:24,  1.28s/it]
 24%|██▍       | 1797/7505 [27:25<2:01:03,  1.27s/it]
 24%|██▍       | 1798/7505 [27:27<2:00:18,  1.26s/it]
 24%|██▍       | 1799/7505 [27:28<2:00:01,  1.26s/it]
 24%|██▍       | 1800/7505 [27:29<1:59:45,  1.26s/it]
                                                     
{'loss': '0.6582', 'grad_norm': '3.173', 'learning_rate': '0.0006', 'epoch': '94.75'}
+
 24%|██▍       | 1800/7505 [27:29<1:59:45,  1.26s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:02, 12.38it/s]
+
 10%|█         | 4/39 [00:00<00:04,  7.76it/s]
+
 13%|█▎        | 5/39 [00:00<00:04,  7.23it/s]
+
 15%|█▌        | 6/39 [00:00<00:04,  6.82it/s]
+
 18%|█▊        | 7/39 [00:00<00:04,  6.72it/s]
+
 21%|██        | 8/39 [00:01<00:04,  6.56it/s]
+
 23%|██▎       | 9/39 [00:01<00:04,  6.49it/s]
+
 26%|██▌       | 10/39 [00:01<00:04,  6.46it/s]
+
 28%|██▊       | 11/39 [00:01<00:04,  6.41it/s]
+
 31%|███       | 12/39 [00:01<00:04,  6.41it/s]
+
 33%|███▎      | 13/39 [00:01<00:04,  6.41it/s]
+
 36%|███▌      | 14/39 [00:02<00:03,  6.38it/s]
+
 38%|███▊      | 15/39 [00:02<00:03,  6.29it/s]
+
 41%|████      | 16/39 [00:02<00:03,  6.28it/s]
+
 44%|████▎     | 17/39 [00:02<00:03,  6.28it/s]
+
 46%|████▌     | 18/39 [00:02<00:03,  6.30it/s]
+
 49%|████▊     | 19/39 [00:02<00:03,  6.22it/s]
+
 51%|█████▏    | 20/39 [00:03<00:03,  6.23it/s]
+
 54%|█████▍    | 21/39 [00:03<00:02,  6.24it/s]
+
 56%|█████▋    | 22/39 [00:03<00:02,  6.20it/s]
+
 59%|█████▉    | 23/39 [00:03<00:02,  6.25it/s]
+
 62%|██████▏   | 24/39 [00:03<00:02,  6.22it/s]
+
 64%|██████▍   | 25/39 [00:03<00:02,  6.19it/s]
+
 67%|██████▋   | 26/39 [00:04<00:02,  6.18it/s]
+
 69%|██████▉   | 27/39 [00:04<00:01,  6.20it/s]
+
 72%|███████▏  | 28/39 [00:04<00:01,  6.17it/s]
+
 74%|███████▍  | 29/39 [00:04<00:01,  6.21it/s]
+
 77%|███████▋  | 30/39 [00:04<00:01,  6.26it/s]
+
 79%|███████▉  | 31/39 [00:04<00:01,  6.24it/s]
+
 82%|████████▏ | 32/39 [00:04<00:01,  6.19it/s]
+
 85%|████████▍ | 33/39 [00:05<00:00,  6.19it/s]
+
 87%|████████▋ | 34/39 [00:05<00:00,  6.24it/s]
+
 90%|████████▉ | 35/39 [00:05<00:00,  6.26it/s]
+
 92%|█████████▏| 36/39 [00:05<00:00,  6.29it/s]
+
 95%|█████████▍| 37/39 [00:05<00:00,  6.27it/s]
+
 97%|█████████▋| 38/39 [00:05<00:00,  6.28it/s]
+
100%|██████████| 39/39 [00:06<00:00,  6.24it/s]
                                                     
+
                                               
{'eval_loss': '9.502', 'eval_runtime': '6.308', 'eval_samples_per_second': '386.9', 'eval_steps_per_second': '6.182', 'epoch': '94.75'}
+
 24%|██▍       | 1800/7505 [27:35<1:59:45,  1.26s/it]
+
100%|██████████| 39/39 [00:06<00:00,  6.24it/s][eval] epoch 94.752 step 1800 loss 9.5018 (best 4.7139 @ ep 30.322, 17/20 since best)
+
+
                                               
 24%|██▍       | 1801/7505 [27:37<4:59:56,  3.16s/it]
 24%|██▍       | 1802/7505 [27:38<4:05:29,  2.58s/it]
 24%|██▍       | 1803/7505 [27:39<3:27:09,  2.18s/it]
 24%|██▍       | 1804/7505 [27:40<3:00:07,  1.90s/it]
 24%|██▍       | 1805/7505 [27:41<2:28:45,  1.57s/it]
 24%|██▍       | 1806/7505 [27:43<2:22:37,  1.50s/it]
 24%|██▍       | 1807/7505 [27:44<2:15:53,  1.43s/it]
 24%|██▍       | 1808/7505 [27:45<2:10:42,  1.38s/it]
 24%|██▍       | 1809/7505 [27:46<2:07:14,  1.34s/it]
 24%|██▍       | 1810/7505 [27:48<2:04:57,  1.32s/it]
                                                     
{'loss': '0.6497', 'grad_norm': '2.967', 'learning_rate': '0.0006', 'epoch': '95.27'}
+
 24%|██▍       | 1810/7505 [27:48<2:04:57,  1.32s/it]
 24%|██▍       | 1811/7505 [27:49<2:03:43,  1.30s/it]
 24%|██▍       | 1812/7505 [27:50<2:02:20,  1.29s/it]
 24%|██▍       | 1813/7505 [27:51<2:01:11,  1.28s/it]
 24%|██▍       | 1814/7505 [27:53<2:00:39,  1.27s/it]
 24%|██▍       | 1815/7505 [27:54<2:00:16,  1.27s/it]
 24%|██▍       | 1816/7505 [27:55<1:59:53,  1.26s/it]
 24%|██▍       | 1817/7505 [27:56<1:59:35,  1.26s/it]
 24%|██▍       | 1818/7505 [27:58<1:59:25,  1.26s/it]
 24%|██▍       | 1819/7505 [27:59<1:59:05,  1.26s/it]
 24%|██▍       | 1820/7505 [28:00<1:58:51,  1.25s/it]
                                                     
{'loss': '0.6569', 'grad_norm': '3.094', 'learning_rate': '0.0006', 'epoch': '95.81'}
+
 24%|██▍       | 1820/7505 [28:00<1:58:51,  1.25s/it]
 24%|██▍       | 1821/7505 [28:01<1:59:25,  1.26s/it]
 24%|██▍       | 1822/7505 [28:03<1:59:13,  1.26s/it]
 24%|██▍       | 1823/7505 [28:04<1:59:04,  1.26s/it]
 24%|██▍       | 1824/7505 [28:05<1:46:31,  1.13s/it]
 24%|██▍       | 1825/7505 [28:06<1:52:16,  1.19s/it]
 24%|██▍       | 1826/7505 [28:07<1:53:52,  1.20s/it]
 24%|██▍       | 1827/7505 [28:09<1:54:34,  1.21s/it]
 24%|██▍       | 1828/7505 [28:10<1:55:58,  1.23s/it]
 24%|██▍       | 1829/7505 [28:11<1:56:50,  1.24s/it]
 24%|██▍       | 1830/7505 [28:12<1:57:32,  1.24s/it]
                                                     
{'loss': '0.6338', 'grad_norm': '3.163', 'learning_rate': '0.0006', 'epoch': '96.32'}
+
 24%|██▍       | 1830/7505 [28:12<1:57:32,  1.24s/it]
 24%|██▍       | 1831/7505 [28:14<1:58:02,  1.25s/it]
 24%|██▍       | 1832/7505 [28:15<1:58:06,  1.25s/it]
 24%|██▍       | 1833/7505 [28:16<1:58:08,  1.25s/it]
 24%|██▍       | 1834/7505 [28:17<1:58:18,  1.25s/it]
 24%|██▍       | 1835/7505 [28:19<1:57:47,  1.25s/it]
 24%|██▍       | 1836/7505 [28:20<1:57:57,  1.25s/it]
 24%|██▍       | 1837/7505 [28:21<1:58:10,  1.25s/it]
 24%|██▍       | 1838/7505 [28:22<1:57:58,  1.25s/it]
 25%|██▍       | 1839/7505 [28:24<1:57:39,  1.25s/it]
 25%|██▍       | 1840/7505 [28:25<1:57:18,  1.24s/it]
                                                     
{'loss': '0.6427', 'grad_norm': '3.172', 'learning_rate': '0.0006', 'epoch': '96.86'}
+
 25%|██▍       | 1840/7505 [28:25<1:57:18,  1.24s/it]
 25%|██▍       | 1841/7505 [28:26<1:57:40,  1.25s/it]
 25%|██▍       | 1842/7505 [28:27<1:57:53,  1.25s/it]
 25%|██▍       | 1843/7505 [28:28<1:45:20,  1.12s/it]
 25%|██▍       | 1844/7505 [28:29<1:51:55,  1.19s/it]
 25%|██▍       | 1845/7505 [28:31<1:53:45,  1.21s/it]
 25%|██▍       | 1846/7505 [28:32<1:55:02,  1.22s/it]
 25%|██▍       | 1847/7505 [28:33<1:55:57,  1.23s/it]
 25%|██▍       | 1848/7505 [28:34<1:56:20,  1.23s/it]
 25%|██▍       | 1849/7505 [28:36<1:57:00,  1.24s/it]
 25%|██▍       | 1850/7505 [28:37<1:57:36,  1.25s/it]
                                                     
{'loss': '0.6009', 'grad_norm': '2.896', 'learning_rate': '0.0006', 'epoch': '97.38'}
+
 25%|██▍       | 1850/7505 [28:37<1:57:36,  1.25s/it]
 25%|██▍       | 1851/7505 [28:38<1:57:50,  1.25s/it]
 25%|██▍       | 1852/7505 [28:40<1:57:44,  1.25s/it]
 25%|██▍       | 1853/7505 [28:41<1:58:16,  1.26s/it]
 25%|██▍       | 1854/7505 [28:42<1:58:25,  1.26s/it]
 25%|██▍       | 1855/7505 [28:43<1:59:05,  1.26s/it]
 25%|██▍       | 1856/7505 [28:45<1:59:06,  1.27s/it]
 25%|██▍       | 1857/7505 [28:46<1:59:31,  1.27s/it]
 25%|██▍       | 1858/7505 [28:47<1:59:44,  1.27s/it]
 25%|██▍       | 1859/7505 [28:48<1:59:21,  1.27s/it]
 25%|██▍       | 1860/7505 [28:50<1:59:04,  1.27s/it]
                                                     
{'loss': '0.649', 'grad_norm': '3.572', 'learning_rate': '0.0006', 'epoch': '97.91'}
+
 25%|██▍       | 1860/7505 [28:50<1:59:04,  1.27s/it]
 25%|██▍       | 1861/7505 [28:51<1:59:04,  1.27s/it]
 25%|██▍       | 1862/7505 [28:52<1:45:33,  1.12s/it]
 25%|██▍       | 1863/7505 [28:53<1:52:08,  1.19s/it]
 25%|██▍       | 1864/7505 [28:54<1:53:56,  1.21s/it]
 25%|██▍       | 1865/7505 [28:56<1:55:22,  1.23s/it]
 25%|██▍       | 1866/7505 [28:57<1:56:20,  1.24s/it]
 25%|██▍       | 1867/7505 [28:58<1:56:57,  1.24s/it]
 25%|██▍       | 1868/7505 [28:59<1:56:56,  1.24s/it]
 25%|██▍       | 1869/7505 [29:01<1:57:29,  1.25s/it]
 25%|██▍       | 1870/7505 [29:02<1:57:09,  1.25s/it]
                                                     
{'loss': '0.5914', 'grad_norm': '3.396', 'learning_rate': '0.0006', 'epoch': '98.43'}
+
 25%|██▍       | 1870/7505 [29:02<1:57:09,  1.25s/it]
 25%|██▍       | 1871/7505 [29:03<1:57:04,  1.25s/it]
 25%|██▍       | 1872/7505 [29:04<1:58:11,  1.26s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:02, 12.53it/s]
+
 10%|█         | 4/39 [00:00<00:04,  7.79it/s]
+
 13%|█▎        | 5/39 [00:00<00:04,  7.26it/s]
+
 15%|█▌        | 6/39 [00:00<00:04,  6.90it/s]
+
 18%|█▊        | 7/39 [00:00<00:04,  6.66it/s]
+
 21%|██        | 8/39 [00:01<00:04,  6.55it/s]
+
 23%|██▎       | 9/39 [00:01<00:04,  6.44it/s]
+
 26%|██▌       | 10/39 [00:01<00:04,  6.35it/s]
+
 28%|██▊       | 11/39 [00:01<00:04,  6.26it/s]
+
 31%|███       | 12/39 [00:01<00:04,  6.20it/s]
+
 33%|███▎      | 13/39 [00:01<00:04,  6.21it/s]
+
 36%|███▌      | 14/39 [00:02<00:04,  6.16it/s]
+
 38%|███▊      | 15/39 [00:02<00:03,  6.18it/s]
+
 41%|████      | 16/39 [00:02<00:03,  6.21it/s]
+
 44%|████▎     | 17/39 [00:02<00:03,  6.17it/s]
+
 46%|████▌     | 18/39 [00:02<00:03,  6.12it/s]
+
 49%|████▊     | 19/39 [00:02<00:03,  6.11it/s]
+
 51%|█████▏    | 20/39 [00:03<00:03,  6.09it/s]
+
 54%|█████▍    | 21/39 [00:03<00:02,  6.14it/s]
+
 56%|█████▋    | 22/39 [00:03<00:02,  6.13it/s]
+
 59%|█████▉    | 23/39 [00:03<00:02,  6.14it/s]
+
 62%|██████▏   | 24/39 [00:03<00:02,  6.18it/s]
+
 64%|██████▍   | 25/39 [00:03<00:02,  6.14it/s]
+
 67%|██████▋   | 26/39 [00:04<00:02,  6.16it/s]
+
 69%|██████▉   | 27/39 [00:04<00:01,  6.18it/s]
+
 72%|███████▏  | 28/39 [00:04<00:01,  6.16it/s]
+
 74%|███████▍  | 29/39 [00:04<00:01,  6.24it/s]
+
 77%|███████▋  | 30/39 [00:04<00:01,  6.26it/s]
+
 79%|███████▉  | 31/39 [00:04<00:01,  6.24it/s]
+
 82%|████████▏ | 32/39 [00:05<00:01,  6.25it/s]
+
 85%|████████▍ | 33/39 [00:05<00:00,  6.18it/s]
+
 87%|████████▋ | 34/39 [00:05<00:00,  6.20it/s]
+
 90%|████████▉ | 35/39 [00:05<00:00,  6.24it/s]
+
 92%|█████████▏| 36/39 [00:05<00:00,  6.21it/s]
+
 95%|█████████▍| 37/39 [00:05<00:00,  6.17it/s]
+
 97%|█████████▋| 38/39 [00:05<00:00,  6.26it/s]
+
100%|██████████| 39/39 [00:06<00:00,  6.24it/s]
                                                     
+
                                               
{'eval_loss': '9.709', 'eval_runtime': '6.362', 'eval_samples_per_second': '383.7', 'eval_steps_per_second': '6.13', 'epoch': '98.54'}
+
 25%|██▍       | 1872/7505 [29:11<1:58:11,  1.26s/it]
+
100%|██████████| 39/39 [00:06<00:00,  6.24it/s][eval] epoch 98.537 step 1872 loss 9.7093 (best 4.7139 @ ep 30.322, 18/20 since best)
+
+
                                               
 25%|██▍       | 1873/7505 [29:12<4:57:11,  3.17s/it]
 25%|██▍       | 1874/7505 [29:13<4:02:52,  2.59s/it]
 25%|██▍       | 1875/7505 [29:15<3:25:16,  2.19s/it]
 25%|██▍       | 1876/7505 [29:16<2:59:12,  1.91s/it]
 25%|██▌       | 1877/7505 [29:17<2:40:43,  1.71s/it]
 25%|██▌       | 1878/7505 [29:18<2:27:53,  1.58s/it]
 25%|██▌       | 1879/7505 [29:20<2:18:50,  1.48s/it]
 25%|██▌       | 1880/7505 [29:21<2:12:39,  1.41s/it]
                                                     
{'loss': '0.6502', 'grad_norm': '3.213', 'learning_rate': '0.0006', 'epoch': '98.97'}
+
 25%|██▌       | 1880/7505 [29:21<2:12:39,  1.41s/it]
 25%|██▌       | 1881/7505 [29:22<1:55:43,  1.23s/it]
 25%|██▌       | 1882/7505 [29:23<1:58:55,  1.27s/it]
 25%|██▌       | 1883/7505 [29:24<1:58:08,  1.26s/it]
 25%|██▌       | 1884/7505 [29:25<1:57:51,  1.26s/it]
 25%|██▌       | 1885/7505 [29:27<1:57:14,  1.25s/it]
 25%|██▌       | 1886/7505 [29:28<1:56:58,  1.25s/it]
 25%|██▌       | 1887/7505 [29:29<1:57:13,  1.25s/it]
 25%|██▌       | 1888/7505 [29:30<1:57:21,  1.25s/it]
 25%|██▌       | 1889/7505 [29:32<1:57:14,  1.25s/it]
 25%|██▌       | 1890/7505 [29:33<1:57:02,  1.25s/it]
                                                     
{'loss': '0.563', 'grad_norm': '2.876', 'learning_rate': '0.0006', 'epoch': '99.48'}
+
 25%|██▌       | 1890/7505 [29:33<1:57:02,  1.25s/it]
 25%|██▌       | 1891/7505 [29:34<1:57:37,  1.26s/it]
 25%|██▌       | 1892/7505 [29:35<1:57:34,  1.26s/it]
 25%|██▌       | 1893/7505 [29:37<1:57:16,  1.25s/it]
 25%|██▌       | 1894/7505 [29:38<1:57:11,  1.25s/it]
 25%|██▌       | 1895/7505 [29:39<1:56:59,  1.25s/it]
 25%|██▌       | 1896/7505 [29:40<1:56:54,  1.25s/it]
 25%|██▌       | 1897/7505 [29:42<1:57:01,  1.25s/it]
 25%|██▌       | 1898/7505 [29:43<1:57:11,  1.25s/it]
 25%|██▌       | 1899/7505 [29:44<1:57:12,  1.25s/it]
 25%|██▌       | 1900/7505 [29:45<1:44:26,  1.12s/it]
                                                     
{'loss': '0.6461', 'grad_norm': '3.499', 'learning_rate': '0.0006', 'epoch': '100'}
+
 25%|██▌       | 1900/7505 [29:45<1:44:26,  1.12s/it]
 25%|██▌       | 1901/7505 [29:46<1:51:25,  1.19s/it]
 25%|██▌       | 1902/7505 [29:48<1:53:19,  1.21s/it]
 25%|██▌       | 1903/7505 [29:49<1:54:28,  1.23s/it]
 25%|██▌       | 1904/7505 [29:50<1:55:35,  1.24s/it]
 25%|██▌       | 1905/7505 [29:51<1:55:57,  1.24s/it]
 25%|██▌       | 1906/7505 [29:53<1:56:06,  1.24s/it]
 25%|██▌       | 1907/7505 [29:54<1:56:36,  1.25s/it]
 25%|██▌       | 1908/7505 [29:55<1:55:57,  1.24s/it]
 25%|██▌       | 1909/7505 [29:56<1:56:04,  1.24s/it]
 25%|██▌       | 1910/7505 [29:58<1:56:04,  1.24s/it]
                                                     
{'loss': '0.5434', 'grad_norm': '3.483', 'learning_rate': '0.0006', 'epoch': '100.5'}
+
 25%|██▌       | 1910/7505 [29:58<1:56:04,  1.24s/it]
 25%|██▌       | 1911/7505 [29:59<1:56:30,  1.25s/it]
 25%|██▌       | 1912/7505 [30:00<1:56:32,  1.25s/it]
 25%|██▌       | 1913/7505 [30:01<1:56:00,  1.24s/it]
 26%|██▌       | 1914/7505 [30:03<1:56:13,  1.25s/it]
 26%|██▌       | 1915/7505 [30:04<1:56:13,  1.25s/it]
 26%|██▌       | 1916/7505 [30:05<1:56:19,  1.25s/it]
 26%|██▌       | 1917/7505 [30:06<1:56:52,  1.25s/it]
 26%|██▌       | 1918/7505 [30:08<1:56:57,  1.26s/it]
 26%|██▌       | 1919/7505 [30:08<1:43:17,  1.11s/it]
 26%|██▌       | 1920/7505 [30:10<1:50:03,  1.18s/it]
                                                     
{'loss': '0.6299', 'grad_norm': '3.056', 'learning_rate': '0.0006', 'epoch': '101.1'}
+
 26%|██▌       | 1920/7505 [30:10<1:50:03,  1.18s/it]
 26%|██▌       | 1921/7505 [30:11<1:52:09,  1.21s/it]
 26%|██▌       | 1922/7505 [30:12<1:53:45,  1.22s/it]
 26%|██▌       | 1923/7505 [30:14<1:54:39,  1.23s/it]
 26%|██▌       | 1924/7505 [30:15<1:55:12,  1.24s/it]
 26%|██▌       | 1925/7505 [30:16<1:55:10,  1.24s/it]
 26%|██▌       | 1926/7505 [30:17<1:54:37,  1.23s/it]
 26%|██▌       | 1927/7505 [30:19<1:55:02,  1.24s/it]
 26%|██▌       | 1928/7505 [30:20<1:55:28,  1.24s/it]
 26%|██▌       | 1929/7505 [30:21<1:55:58,  1.25s/it]
 26%|██▌       | 1930/7505 [30:22<1:56:17,  1.25s/it]
                                                     
{'loss': '0.5443', 'grad_norm': '3.06', 'learning_rate': '0.0006', 'epoch': '101.6'}
+
 26%|██▌       | 1930/7505 [30:22<1:56:17,  1.25s/it]
 26%|██▌       | 1931/7505 [30:24<1:56:38,  1.26s/it]
 26%|██▌       | 1932/7505 [30:25<1:56:09,  1.25s/it]
 26%|██▌       | 1933/7505 [30:26<1:56:19,  1.25s/it]
 26%|██▌       | 1934/7505 [30:27<1:56:25,  1.25s/it]
 26%|██▌       | 1935/7505 [30:29<1:56:31,  1.26s/it]
 26%|██▌       | 1936/7505 [30:30<1:56:19,  1.25s/it]
 26%|██▌       | 1937/7505 [30:31<1:56:21,  1.25s/it]
 26%|██▌       | 1938/7505 [30:32<1:43:39,  1.12s/it]
 26%|██▌       | 1939/7505 [30:33<1:48:44,  1.17s/it]
 26%|██▌       | 1940/7505 [30:34<1:50:45,  1.19s/it]
                                                     
{'loss': '0.6006', 'grad_norm': '2.743', 'learning_rate': '0.0006', 'epoch': '102.1'}
+
 26%|██▌       | 1940/7505 [30:34<1:50:45,  1.19s/it]
 26%|██▌       | 1941/7505 [30:36<1:52:52,  1.22s/it]
 26%|██▌       | 1942/7505 [30:37<1:53:23,  1.22s/it]
 26%|██▌       | 1943/7505 [30:38<1:54:03,  1.23s/it]
 26%|██▌       | 1944/7505 [30:39<1:54:37,  1.24s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:02, 12.70it/s]
+
 10%|█         | 4/39 [00:00<00:04,  7.80it/s]
+
 13%|█▎        | 5/39 [00:00<00:04,  7.18it/s]
+
 15%|█▌        | 6/39 [00:00<00:04,  6.86it/s]
+
 18%|█▊        | 7/39 [00:00<00:04,  6.70it/s]
+
 21%|██        | 8/39 [00:01<00:04,  6.54it/s]
+
 23%|██▎       | 9/39 [00:01<00:04,  6.43it/s]
+
 26%|██▌       | 10/39 [00:01<00:04,  6.38it/s]
+
 28%|██▊       | 11/39 [00:01<00:04,  6.35it/s]
+
 31%|███       | 12/39 [00:01<00:04,  6.31it/s]
+
 33%|███▎      | 13/39 [00:01<00:04,  6.28it/s]
+
 36%|███▌      | 14/39 [00:02<00:03,  6.31it/s]
+
 38%|███▊      | 15/39 [00:02<00:03,  6.22it/s]
+
 41%|████      | 16/39 [00:02<00:03,  6.22it/s]
+
 44%|████▎     | 17/39 [00:02<00:03,  6.25it/s]
+
 46%|████▌     | 18/39 [00:02<00:03,  6.25it/s]
+
 49%|████▊     | 19/39 [00:02<00:03,  6.18it/s]
+
 51%|█████▏    | 20/39 [00:03<00:03,  6.18it/s]
+
 54%|█████▍    | 21/39 [00:03<00:02,  6.20it/s]
+
 56%|█████▋    | 22/39 [00:03<00:02,  6.15it/s]
+
 59%|█████▉    | 23/39 [00:03<00:02,  6.21it/s]
+
 62%|██████▏   | 24/39 [00:03<00:02,  6.23it/s]
+
 64%|██████▍   | 25/39 [00:03<00:02,  6.21it/s]
+
 67%|██████▋   | 26/39 [00:04<00:02,  6.19it/s]
+
 69%|██████▉   | 27/39 [00:04<00:01,  6.13it/s]
+
 72%|███████▏  | 28/39 [00:04<00:01,  6.12it/s]
+
 74%|███████▍  | 29/39 [00:04<00:01,  6.15it/s]
+
 77%|███████▋  | 30/39 [00:04<00:01,  6.12it/s]
+
 79%|███████▉  | 31/39 [00:04<00:01,  6.11it/s]
+
 82%|████████▏ | 32/39 [00:05<00:01,  6.06it/s]
+
 85%|████████▍ | 33/39 [00:05<00:00,  6.06it/s]
+
 87%|████████▋ | 34/39 [00:05<00:00,  6.04it/s]
+
 90%|████████▉ | 35/39 [00:05<00:00,  6.09it/s]
+
 92%|█████████▏| 36/39 [00:05<00:00,  6.12it/s]
+
 95%|█████████▍| 37/39 [00:05<00:00,  6.12it/s]
+
 97%|█████████▋| 38/39 [00:05<00:00,  6.12it/s]
+
100%|██████████| 39/39 [00:06<00:00,  6.18it/s]
                                                     
+
                                               
{'eval_loss': '9.89', 'eval_runtime': '6.369', 'eval_samples_per_second': '383.2', 'eval_steps_per_second': '6.123', 'epoch': '102.3'}
+
 26%|██▌       | 1944/7505 [30:46<1:54:37,  1.24s/it]
+
100%|██████████| 39/39 [00:06<00:00,  6.18it/s][eval] epoch 102.322 step 1944 loss 9.8902 (best 4.7139 @ ep 30.322, 19/20 since best)
+
+
                                               
 26%|██▌       | 1945/7505 [30:47<4:52:05,  3.15s/it]
 26%|██▌       | 1946/7505 [30:48<3:59:39,  2.59s/it]
 26%|██▌       | 1947/7505 [30:50<3:22:51,  2.19s/it]
 26%|██▌       | 1948/7505 [30:51<2:57:04,  1.91s/it]
 26%|██▌       | 1949/7505 [30:52<2:38:30,  1.71s/it]
 26%|██▌       | 1950/7505 [30:53<2:24:54,  1.57s/it]
                                                     
{'loss': '0.5442', 'grad_norm': '3.135', 'learning_rate': '0.0006', 'epoch': '102.6'}
+
 26%|██▌       | 1950/7505 [30:53<2:24:54,  1.57s/it]
 26%|██▌       | 1951/7505 [30:55<2:16:35,  1.48s/it]
 26%|██▌       | 1952/7505 [30:56<2:10:45,  1.41s/it]
 26%|██▌       | 1953/7505 [30:57<2:06:26,  1.37s/it]
 26%|██▌       | 1954/7505 [30:58<2:03:31,  1.34s/it]
 26%|██▌       | 1955/7505 [31:00<2:01:19,  1.31s/it]
 26%|██▌       | 1956/7505 [31:01<2:00:04,  1.30s/it]
 26%|██▌       | 1957/7505 [31:02<1:45:36,  1.14s/it]
 26%|██▌       | 1958/7505 [31:03<1:50:23,  1.19s/it]
 26%|██▌       | 1959/7505 [31:04<1:52:13,  1.21s/it]
 26%|██▌       | 1960/7505 [31:06<1:53:06,  1.22s/it]
                                                     
{'loss': '0.5816', 'grad_norm': '2.984', 'learning_rate': '0.0006', 'epoch': '103.2'}
+
 26%|██▌       | 1960/7505 [31:06<1:53:06,  1.22s/it]
 26%|██▌       | 1961/7505 [31:07<1:53:46,  1.23s/it]
 26%|██▌       | 1962/7505 [31:08<1:53:50,  1.23s/it]
 26%|██▌       | 1963/7505 [31:09<1:53:47,  1.23s/it]
 26%|██▌       | 1964/7505 [31:10<1:53:56,  1.23s/it]
 26%|██▌       | 1965/7505 [31:12<1:53:25,  1.23s/it]
 26%|██▌       | 1966/7505 [31:13<1:53:20,  1.23s/it]
 26%|██▌       | 1967/7505 [31:14<1:53:40,  1.23s/it]
 26%|██▌       | 1968/7505 [31:15<1:54:22,  1.24s/it]
 26%|██▌       | 1969/7505 [31:17<1:55:00,  1.25s/it]
 26%|██▌       | 1970/7505 [31:18<1:54:51,  1.25s/it]
                                                     
{'loss': '0.5482', 'grad_norm': '3.139', 'learning_rate': '0.0006', 'epoch': '103.7'}
+
 26%|██▌       | 1970/7505 [31:18<1:54:51,  1.25s/it]
 26%|██▋       | 1971/7505 [31:19<1:55:31,  1.25s/it]
 26%|██▋       | 1972/7505 [31:20<1:55:40,  1.25s/it]
 26%|██▋       | 1973/7505 [31:22<1:56:06,  1.26s/it]
 26%|██▋       | 1974/7505 [31:23<1:56:21,  1.26s/it]
 26%|██▋       | 1975/7505 [31:24<1:57:01,  1.27s/it]
 26%|██▋       | 1976/7505 [31:25<1:44:23,  1.13s/it]
 26%|██▋       | 1977/7505 [31:26<1:49:40,  1.19s/it]
 26%|██▋       | 1978/7505 [31:28<1:51:59,  1.22s/it]
 26%|██▋       | 1979/7505 [31:29<1:53:31,  1.23s/it]
 26%|██▋       | 1980/7505 [31:30<1:54:03,  1.24s/it]
                                                     
{'loss': '0.5633', 'grad_norm': '2.747', 'learning_rate': '0.0006', 'epoch': '104.2'}
+
 26%|██▋       | 1980/7505 [31:30<1:54:03,  1.24s/it]
 26%|██▋       | 1981/7505 [31:31<1:54:10,  1.24s/it]
 26%|██▋       | 1982/7505 [31:33<1:54:21,  1.24s/it]
 26%|██▋       | 1983/7505 [31:34<1:54:38,  1.25s/it]
 26%|██▋       | 1984/7505 [31:35<1:55:08,  1.25s/it]
 26%|██▋       | 1985/7505 [31:36<1:55:12,  1.25s/it]
 26%|██▋       | 1986/7505 [31:38<1:55:26,  1.26s/it]
 26%|██▋       | 1987/7505 [31:39<1:55:16,  1.25s/it]
 26%|██▋       | 1988/7505 [31:40<1:55:08,  1.25s/it]
 27%|██▋       | 1989/7505 [31:42<1:55:37,  1.26s/it]
 27%|██▋       | 1990/7505 [31:43<1:55:40,  1.26s/it]
                                                     
{'loss': '0.55', 'grad_norm': '3.327', 'learning_rate': '0.0006', 'epoch': '104.8'}
+
 27%|██▋       | 1990/7505 [31:43<1:55:40,  1.26s/it]
 27%|██▋       | 1991/7505 [31:44<1:55:55,  1.26s/it]
 27%|██▋       | 1992/7505 [31:45<1:55:35,  1.26s/it]
 27%|██▋       | 1993/7505 [31:47<1:55:19,  1.26s/it]
 27%|██▋       | 1994/7505 [31:48<1:55:19,  1.26s/it]
 27%|██▋       | 1995/7505 [31:49<1:42:24,  1.12s/it]
 27%|██▋       | 1996/7505 [31:50<1:48:52,  1.19s/it]
 27%|██▋       | 1997/7505 [31:51<1:50:43,  1.21s/it]
 27%|██▋       | 1998/7505 [31:52<1:52:10,  1.22s/it]
 27%|██▋       | 1999/7505 [31:54<1:52:53,  1.23s/it]
 27%|██▋       | 2000/7505 [31:55<1:53:32,  1.24s/it]
                                                     
{'loss': '0.5467', 'grad_norm': '2.815', 'learning_rate': '0.0006', 'epoch': '105.3'}
+
 27%|██▋       | 2000/7505 [31:55<1:53:32,  1.24s/it]
 27%|██▋       | 2001/7505 [31:56<1:54:05,  1.24s/it]
 27%|██▋       | 2002/7505 [31:57<1:53:28,  1.24s/it]
 27%|██▋       | 2003/7505 [31:59<1:52:53,  1.23s/it]
 27%|██▋       | 2004/7505 [32:00<1:53:12,  1.23s/it]
 27%|██▋       | 2005/7505 [32:01<1:53:44,  1.24s/it]
 27%|██▋       | 2006/7505 [32:02<1:54:02,  1.24s/it]
 27%|██▋       | 2007/7505 [32:04<1:54:01,  1.24s/it]
 27%|██▋       | 2008/7505 [32:05<1:53:59,  1.24s/it]
 27%|██▋       | 2009/7505 [32:06<1:54:08,  1.25s/it]
 27%|██▋       | 2010/7505 [32:07<1:54:40,  1.25s/it]
                                                     
{'loss': '0.5482', 'grad_norm': '3.208', 'learning_rate': '0.0006', 'epoch': '105.8'}
+
 27%|██▋       | 2010/7505 [32:07<1:54:40,  1.25s/it]
 27%|██▋       | 2011/7505 [32:09<1:54:45,  1.25s/it]
 27%|██▋       | 2012/7505 [32:10<1:54:22,  1.25s/it]
 27%|██▋       | 2013/7505 [32:11<1:54:30,  1.25s/it]
 27%|██▋       | 2014/7505 [32:12<1:42:05,  1.12s/it]
 27%|██▋       | 2015/7505 [32:13<1:47:41,  1.18s/it]
 27%|██▋       | 2016/7505 [32:15<1:49:30,  1.20s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:02, 12.92it/s]
+
 10%|█         | 4/39 [00:00<00:04,  7.79it/s]
+
 13%|█▎        | 5/39 [00:00<00:04,  7.25it/s]
+
 15%|█▌        | 6/39 [00:00<00:04,  6.93it/s]
+
 18%|█▊        | 7/39 [00:00<00:04,  6.69it/s]
+
 21%|██        | 8/39 [00:01<00:04,  6.46it/s]
+
 23%|██▎       | 9/39 [00:01<00:04,  6.38it/s]
+
 26%|██▌       | 10/39 [00:01<00:04,  6.35it/s]
+
 28%|██▊       | 11/39 [00:01<00:04,  6.34it/s]
+
 31%|███       | 12/39 [00:01<00:04,  6.31it/s]
+
 33%|███▎      | 13/39 [00:01<00:04,  6.29it/s]
+
 36%|███▌      | 14/39 [00:02<00:03,  6.26it/s]
+
 38%|███▊      | 15/39 [00:02<00:03,  6.19it/s]
+
 41%|████      | 16/39 [00:02<00:03,  6.25it/s]
+
 44%|████▎     | 17/39 [00:02<00:03,  6.13it/s]
+
 46%|████▌     | 18/39 [00:02<00:03,  6.17it/s]
+
 49%|████▊     | 19/39 [00:02<00:03,  6.19it/s]
+
 51%|█████▏    | 20/39 [00:03<00:03,  6.19it/s]
+
 54%|█████▍    | 21/39 [00:03<00:02,  6.18it/s]
+
 56%|█████▋    | 22/39 [00:03<00:02,  6.14it/s]
+
 59%|█████▉    | 23/39 [00:03<00:02,  6.16it/s]
+
 62%|██████▏   | 24/39 [00:03<00:02,  6.11it/s]
+
 64%|██████▍   | 25/39 [00:03<00:02,  6.14it/s]
+
 67%|██████▋   | 26/39 [00:04<00:02,  6.16it/s]
+
 69%|██████▉   | 27/39 [00:04<00:01,  6.22it/s]
+
 72%|███████▏  | 28/39 [00:04<00:01,  6.24it/s]
+
 74%|███████▍  | 29/39 [00:04<00:01,  6.23it/s]
+
 77%|███████▋  | 30/39 [00:04<00:01,  6.18it/s]
+
 79%|███████▉  | 31/39 [00:04<00:01,  6.21it/s]
+
 82%|████████▏ | 32/39 [00:05<00:01,  6.22it/s]
+
 85%|████████▍ | 33/39 [00:05<00:00,  6.23it/s]
+
 87%|████████▋ | 34/39 [00:05<00:00,  6.21it/s]
+
 90%|████████▉ | 35/39 [00:05<00:00,  6.18it/s]
+
 92%|█████████▏| 36/39 [00:05<00:00,  6.19it/s]
+
 95%|█████████▍| 37/39 [00:05<00:00,  6.15it/s]
+
 97%|█████████▋| 38/39 [00:05<00:00,  6.20it/s]
+
100%|██████████| 39/39 [00:06<00:00,  6.18it/s]
                                                     
+
                                               
{'eval_loss': '10.03', 'eval_runtime': '6.362', 'eval_samples_per_second': '383.7', 'eval_steps_per_second': '6.13', 'epoch': '106.1'}
+
 27%|██▋       | 2016/7505 [32:21<1:49:30,  1.20s/it]
+
100%|██████████| 39/39 [00:06<00:00,  6.18it/s][eval] epoch 106.107 step 2016 loss 10.0338 (best 4.7139 @ ep 30.322, 20/20 since best)
+
+
                                               [transformers] 
+
+Training completed. Do not forget to share your model on huggingface.co/models =)
+
+
+
                                                     
{'train_runtime': '1943', 'train_samples_per_second': '1939', 'train_steps_per_second': '3.863', 'train_loss': '2.766', 'epoch': '106.1'}
+
 27%|██▋       | 2016/7505 [32:21<1:49:30,  1.20s/it]
 27%|██▋       | 2016/7505 [32:21<1:28:05,  1.04it/s]
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+wandb: 
+wandb: 🚀 View run t2a_19M_C0_baseline at: https://wandb.ai/slowrun-wankers/data-aug-pretraining/runs/ws9il445
+wandb: Find logs at: wandb/run-20260819_172420-ws9il445/logs
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+[rank0]:[W819 17:57:27.783824120 ProcessGroupNCCL.cpp:1575] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator())
diff --git a/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-1152/chat_template.jinja b/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-1152/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..699ff8df401fe4788525e9c1f9b86a99eadd6230
--- /dev/null
+++ b/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-1152/chat_template.jinja
@@ -0,0 +1,85 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0].role == 'system' %}
+        {{- messages[0].content + '\n\n' }}
+    {%- endif %}
+    {{- "# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0].content + '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
+{%- for message in messages[::-1] %}
+    {%- set index = (messages|length - 1) - loop.index0 %}
+    {%- if ns.multi_step_tool and message.role == "user" and not(message.content.startswith('<tool_response>') and message.content.endswith('</tool_response>')) %}
+        {%- set ns.multi_step_tool = false %}
+        {%- set ns.last_query_index = index %}
+    {%- endif %}
+{%- endfor %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {%- set content = message.content %}
+        {%- set reasoning_content = '' %}
+        {%- if message.reasoning_content is defined and message.reasoning_content is not none %}
+            {%- set reasoning_content = message.reasoning_content %}
+        {%- else %}
+            {%- if '</think>' in message.content %}
+                {%- set content = message.content.split('</think>')[-1].lstrip('\n') %}
+                {%- set reasoning_content = message.content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
+            {%- endif %}
+        {%- endif %}
+        {%- if loop.index0 > ns.last_query_index %}
+            {%- if loop.last or (not loop.last and reasoning_content) %}
+                {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content.strip('\n') + '\n</think>\n\n' + content.lstrip('\n') }}
+            {%- else %}
+                {{- '<|im_start|>' + message.role + '\n' + content }}
+            {%- endif %}
+        {%- else %}
+            {{- '<|im_start|>' + message.role + '\n' + content }}
+        {%- endif %}
+        {%- if message.tool_calls %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if (loop.first and content) or (not loop.first) %}
+                    {{- '\n' }}
+                {%- endif %}
+                {%- if tool_call.function %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {{- '<tool_call>\n{"name": "' }}
+                {{- tool_call.name }}
+                {{- '", "arguments": ' }}
+                {%- if tool_call.arguments is string %}
+                    {{- tool_call.arguments }}
+                {%- else %}
+                    {{- tool_call.arguments | tojson }}
+                {%- endif %}
+                {{- '}\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.first or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+    {%- if enable_thinking is defined and enable_thinking is false %}
+        {{- '<think>\n\n</think>\n\n' }}
+    {%- endif %}
+{%- endif %}
\ No newline at end of file
diff --git a/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-1152/config.json b/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-1152/config.json
new file mode 100644
index 0000000000000000000000000000000000000000..a3474619f10931027097cc05d837296fa4f33a6c
--- /dev/null
+++ b/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-1152/config.json
@@ -0,0 +1,32 @@
+{
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": null,
+  "dtype": "float32",
+  "eos_token_id": 151645,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 512,
+  "initializer_range": 0.02,
+  "intermediate_size": 1536,
+  "max_position_embeddings": 2048,
+  "mlp_bias": false,
+  "model_type": "llama",
+  "num_attention_heads": 4,
+  "num_hidden_layers": 20,
+  "num_key_value_heads": 4,
+  "pad_token_id": 151645,
+  "pretraining_tp": 1,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 10000.0,
+    "rope_type": "default"
+  },
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": false,
+  "vocab_size": 151676
+}
diff --git a/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-1152/generation_config.json b/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-1152/generation_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..8c4242f32d40df256857f87264f646130f684e33
--- /dev/null
+++ b/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-1152/generation_config.json
@@ -0,0 +1,11 @@
+{
+  "_from_model_config": true,
+  "eos_token_id": [
+    151645
+  ],
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "transformers_version": "5.15.0",
+  "use_cache": true
+}
diff --git a/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-1152/tokenizer_config.json b/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-1152/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..3d43c97a447f4b25a8c36be5873f7b622935fa90
--- /dev/null
+++ b/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-1152/tokenizer_config.json
@@ -0,0 +1,24 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|l2r_pred|>",
+    "<|r2l_pred|>",
+    "<|next_1_pred|>",
+    "<|next_2_pred|>",
+    "<|next_3_pred|>",
+    "<|next_4_pred|>",
+    "<|next_5_pred|>"
+  ],
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 131072,
+  "pad_token": "<|im_end|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}
diff --git a/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-1152/trainer_state.json b/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-1152/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..e0ef3dea42eabb009044de22848d22059a172723
--- /dev/null
+++ b/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-1152/trainer_state.json
@@ -0,0 +1,974 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 60.644295302013425,
+  "eval_steps": 72,
+  "global_step": 1152,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.053691275167785234,
+      "grad_norm": 2.5292439460754395,
+      "learning_rate": 0.0,
+      "loss": 12.026820182800293,
+      "step": 1
+    },
+    {
+      "epoch": 0.5369127516778524,
+      "grad_norm": 1.6768239736557007,
+      "learning_rate": 5.399999999999999e-05,
+      "loss": 11.693177117241753,
+      "step": 10
+    },
+    {
+      "epoch": 1.0536912751677852,
+      "grad_norm": 1.5936330556869507,
+      "learning_rate": 0.00011399999999999999,
+      "loss": 11.058978271484374,
+      "step": 20
+    },
+    {
+      "epoch": 1.5906040268456376,
+      "grad_norm": 1.5629429817199707,
+      "learning_rate": 0.00017399999999999997,
+      "loss": 10.346524047851563,
+      "step": 30
+    },
+    {
+      "epoch": 2.1073825503355703,
+      "grad_norm": 1.3857123851776123,
+      "learning_rate": 0.000234,
+      "loss": 9.379820251464844,
+      "step": 40
+    },
+    {
+      "epoch": 2.6442953020134228,
+      "grad_norm": 0.852206826210022,
+      "learning_rate": 0.000294,
+      "loss": 8.4895751953125,
+      "step": 50
+    },
+    {
+      "epoch": 3.1610738255033555,
+      "grad_norm": 0.22521893680095673,
+      "learning_rate": 0.00035399999999999993,
+      "loss": 8.030628967285157,
+      "step": 60
+    },
+    {
+      "epoch": 3.697986577181208,
+      "grad_norm": 0.13907139003276825,
+      "learning_rate": 0.0004139999999999999,
+      "loss": 7.934052276611328,
+      "step": 70
+    },
+    {
+      "epoch": 3.8053691275167782,
+      "eval_loss": 7.538887977600098,
+      "eval_runtime": 29.1903,
+      "eval_samples_per_second": 83.624,
+      "eval_steps_per_second": 1.336,
+      "step": 72
+    },
+    {
+      "epoch": 4.214765100671141,
+      "grad_norm": 0.5181903839111328,
+      "learning_rate": 0.000474,
+      "loss": 7.865621948242188,
+      "step": 80
+    },
+    {
+      "epoch": 4.751677852348993,
+      "grad_norm": 0.7047372460365295,
+      "learning_rate": 0.000534,
+      "loss": 7.825517272949218,
+      "step": 90
+    },
+    {
+      "epoch": 5.268456375838926,
+      "grad_norm": 0.2291700839996338,
+      "learning_rate": 0.0005939999999999999,
+      "loss": 7.799723052978516,
+      "step": 100
+    },
+    {
+      "epoch": 5.805369127516778,
+      "grad_norm": 0.21737037599086761,
+      "learning_rate": 0.0006,
+      "loss": 7.775679016113282,
+      "step": 110
+    },
+    {
+      "epoch": 6.322147651006711,
+      "grad_norm": 0.2105957418680191,
+      "learning_rate": 0.0006,
+      "loss": 7.740351867675781,
+      "step": 120
+    },
+    {
+      "epoch": 6.859060402684563,
+      "grad_norm": 0.17692553997039795,
+      "learning_rate": 0.0006,
+      "loss": 7.707257080078125,
+      "step": 130
+    },
+    {
+      "epoch": 7.375838926174497,
+      "grad_norm": 0.34962159395217896,
+      "learning_rate": 0.0006,
+      "loss": 7.687887573242188,
+      "step": 140
+    },
+    {
+      "epoch": 7.590604026845638,
+      "eval_loss": 7.154483318328857,
+      "eval_runtime": 4.2429,
+      "eval_samples_per_second": 575.316,
+      "eval_steps_per_second": 9.192,
+      "step": 144
+    },
+    {
+      "epoch": 7.912751677852349,
+      "grad_norm": 0.3214983642101288,
+      "learning_rate": 0.0006,
+      "loss": 7.646513366699219,
+      "step": 150
+    },
+    {
+      "epoch": 8.429530201342281,
+      "grad_norm": 0.40536266565322876,
+      "learning_rate": 0.0006,
+      "loss": 7.597477722167969,
+      "step": 160
+    },
+    {
+      "epoch": 8.966442953020135,
+      "grad_norm": 0.24251580238342285,
+      "learning_rate": 0.0006,
+      "loss": 7.5954032897949215,
+      "step": 170
+    },
+    {
+      "epoch": 9.483221476510067,
+      "grad_norm": 0.6030511260032654,
+      "learning_rate": 0.0006,
+      "loss": 7.521504211425781,
+      "step": 180
+    },
+    {
+      "epoch": 10.0,
+      "grad_norm": 0.6346228122711182,
+      "learning_rate": 0.0006,
+      "loss": 7.483785247802734,
+      "step": 190
+    },
+    {
+      "epoch": 10.536912751677852,
+      "grad_norm": 0.31154632568359375,
+      "learning_rate": 0.0006,
+      "loss": 7.4339599609375,
+      "step": 200
+    },
+    {
+      "epoch": 11.053691275167786,
+      "grad_norm": 0.18996302783489227,
+      "learning_rate": 0.0006,
+      "loss": 7.282025146484375,
+      "step": 210
+    },
+    {
+      "epoch": 11.375838926174497,
+      "eval_loss": 6.644327163696289,
+      "eval_runtime": 4.0917,
+      "eval_samples_per_second": 596.572,
+      "eval_steps_per_second": 9.531,
+      "step": 216
+    },
+    {
+      "epoch": 11.590604026845638,
+      "grad_norm": 0.33199435472488403,
+      "learning_rate": 0.0006,
+      "loss": 7.253722381591797,
+      "step": 220
+    },
+    {
+      "epoch": 12.10738255033557,
+      "grad_norm": 0.5623189806938171,
+      "learning_rate": 0.0006,
+      "loss": 7.202415466308594,
+      "step": 230
+    },
+    {
+      "epoch": 12.644295302013422,
+      "grad_norm": 0.3505018353462219,
+      "learning_rate": 0.0006,
+      "loss": 7.122056579589843,
+      "step": 240
+    },
+    {
+      "epoch": 13.161073825503356,
+      "grad_norm": 0.39239782094955444,
+      "learning_rate": 0.0006,
+      "loss": 7.019824981689453,
+      "step": 250
+    },
+    {
+      "epoch": 13.697986577181208,
+      "grad_norm": 0.44103360176086426,
+      "learning_rate": 0.0006,
+      "loss": 7.00672607421875,
+      "step": 260
+    },
+    {
+      "epoch": 14.21476510067114,
+      "grad_norm": 0.43972906470298767,
+      "learning_rate": 0.0006,
+      "loss": 6.951579284667969,
+      "step": 270
+    },
+    {
+      "epoch": 14.751677852348994,
+      "grad_norm": 0.4134257137775421,
+      "learning_rate": 0.0006,
+      "loss": 6.868359375,
+      "step": 280
+    },
+    {
+      "epoch": 15.161073825503356,
+      "eval_loss": 6.103402614593506,
+      "eval_runtime": 3.8889,
+      "eval_samples_per_second": 627.689,
+      "eval_steps_per_second": 10.029,
+      "step": 288
+    },
+    {
+      "epoch": 15.268456375838927,
+      "grad_norm": 0.3197152018547058,
+      "learning_rate": 0.0006,
+      "loss": 6.876713562011719,
+      "step": 290
+    },
+    {
+      "epoch": 15.805369127516778,
+      "grad_norm": 0.1717618852853775,
+      "learning_rate": 0.0006,
+      "loss": 6.782273864746093,
+      "step": 300
+    },
+    {
+      "epoch": 16.322147651006713,
+      "grad_norm": 0.640656590461731,
+      "learning_rate": 0.0006,
+      "loss": 6.782866668701172,
+      "step": 310
+    },
+    {
+      "epoch": 16.859060402684563,
+      "grad_norm": 0.4703802764415741,
+      "learning_rate": 0.0006,
+      "loss": 6.722954559326172,
+      "step": 320
+    },
+    {
+      "epoch": 17.375838926174495,
+      "grad_norm": 0.2887628376483917,
+      "learning_rate": 0.0006,
+      "loss": 6.722260284423828,
+      "step": 330
+    },
+    {
+      "epoch": 17.91275167785235,
+      "grad_norm": 0.2506828010082245,
+      "learning_rate": 0.0006,
+      "loss": 6.618051147460937,
+      "step": 340
+    },
+    {
+      "epoch": 18.42953020134228,
+      "grad_norm": 0.4476231634616852,
+      "learning_rate": 0.0006,
+      "loss": 6.565214538574219,
+      "step": 350
+    },
+    {
+      "epoch": 18.966442953020135,
+      "grad_norm": 0.37019670009613037,
+      "learning_rate": 0.0006,
+      "loss": 6.546226501464844,
+      "step": 360
+    },
+    {
+      "epoch": 18.966442953020135,
+      "eval_loss": 5.720033645629883,
+      "eval_runtime": 4.5354,
+      "eval_samples_per_second": 538.212,
+      "eval_steps_per_second": 8.599,
+      "step": 360
+    },
+    {
+      "epoch": 19.483221476510067,
+      "grad_norm": 0.25988173484802246,
+      "learning_rate": 0.0006,
+      "loss": 6.515713500976562,
+      "step": 370
+    },
+    {
+      "epoch": 20.0,
+      "grad_norm": 0.3893851339817047,
+      "learning_rate": 0.0006,
+      "loss": 6.486627197265625,
+      "step": 380
+    },
+    {
+      "epoch": 20.536912751677853,
+      "grad_norm": 0.41074663400650024,
+      "learning_rate": 0.0006,
+      "loss": 6.41888427734375,
+      "step": 390
+    },
+    {
+      "epoch": 21.053691275167786,
+      "grad_norm": 0.2627493739128113,
+      "learning_rate": 0.0006,
+      "loss": 6.416921997070313,
+      "step": 400
+    },
+    {
+      "epoch": 21.59060402684564,
+      "grad_norm": 0.25566911697387695,
+      "learning_rate": 0.0006,
+      "loss": 6.365946960449219,
+      "step": 410
+    },
+    {
+      "epoch": 22.107382550335572,
+      "grad_norm": 0.46552973985671997,
+      "learning_rate": 0.0006,
+      "loss": 6.333673477172852,
+      "step": 420
+    },
+    {
+      "epoch": 22.644295302013422,
+      "grad_norm": 0.49429285526275635,
+      "learning_rate": 0.0006,
+      "loss": 6.27935791015625,
+      "step": 430
+    },
+    {
+      "epoch": 22.751677852348994,
+      "eval_loss": 5.414487361907959,
+      "eval_runtime": 2.6425,
+      "eval_samples_per_second": 923.757,
+      "eval_steps_per_second": 14.759,
+      "step": 432
+    },
+    {
+      "epoch": 23.161073825503355,
+      "grad_norm": 0.3953982889652252,
+      "learning_rate": 0.0006,
+      "loss": 6.2893318176269535,
+      "step": 440
+    },
+    {
+      "epoch": 23.697986577181208,
+      "grad_norm": 0.2119188904762268,
+      "learning_rate": 0.0006,
+      "loss": 6.219153594970703,
+      "step": 450
+    },
+    {
+      "epoch": 24.21476510067114,
+      "grad_norm": 0.2168317586183548,
+      "learning_rate": 0.0006,
+      "loss": 6.2345947265625,
+      "step": 460
+    },
+    {
+      "epoch": 24.751677852348994,
+      "grad_norm": 0.41650861501693726,
+      "learning_rate": 0.0006,
+      "loss": 6.129767990112304,
+      "step": 470
+    },
+    {
+      "epoch": 25.268456375838927,
+      "grad_norm": 0.5049124360084534,
+      "learning_rate": 0.0006,
+      "loss": 6.200005722045899,
+      "step": 480
+    },
+    {
+      "epoch": 25.80536912751678,
+      "grad_norm": 0.46824267506599426,
+      "learning_rate": 0.0006,
+      "loss": 6.174327850341797,
+      "step": 490
+    },
+    {
+      "epoch": 26.322147651006713,
+      "grad_norm": 0.1985160857439041,
+      "learning_rate": 0.0006,
+      "loss": 6.110338592529297,
+      "step": 500
+    },
+    {
+      "epoch": 26.536912751677853,
+      "eval_loss": 5.138503074645996,
+      "eval_runtime": 2.6434,
+      "eval_samples_per_second": 923.446,
+      "eval_steps_per_second": 14.754,
+      "step": 504
+    },
+    {
+      "epoch": 26.859060402684563,
+      "grad_norm": 0.19553250074386597,
+      "learning_rate": 0.0006,
+      "loss": 6.095131683349609,
+      "step": 510
+    },
+    {
+      "epoch": 27.375838926174495,
+      "grad_norm": 0.19958047568798065,
+      "learning_rate": 0.0006,
+      "loss": 5.995913696289063,
+      "step": 520
+    },
+    {
+      "epoch": 27.91275167785235,
+      "grad_norm": 0.24569685757160187,
+      "learning_rate": 0.0006,
+      "loss": 6.003170013427734,
+      "step": 530
+    },
+    {
+      "epoch": 28.42953020134228,
+      "grad_norm": 0.5316331386566162,
+      "learning_rate": 0.0006,
+      "loss": 5.934808349609375,
+      "step": 540
+    },
+    {
+      "epoch": 28.966442953020135,
+      "grad_norm": 0.3737245202064514,
+      "learning_rate": 0.0006,
+      "loss": 5.961774444580078,
+      "step": 550
+    },
+    {
+      "epoch": 29.483221476510067,
+      "grad_norm": 0.31760111451148987,
+      "learning_rate": 0.0006,
+      "loss": 5.928802108764648,
+      "step": 560
+    },
+    {
+      "epoch": 30.0,
+      "grad_norm": 0.2098255753517151,
+      "learning_rate": 0.0006,
+      "loss": 5.846025848388672,
+      "step": 570
+    },
+    {
+      "epoch": 30.322147651006713,
+      "eval_loss": 4.891351222991943,
+      "eval_runtime": 2.6423,
+      "eval_samples_per_second": 923.81,
+      "eval_steps_per_second": 14.76,
+      "step": 576
+    },
+    {
+      "epoch": 30.536912751677853,
+      "grad_norm": 0.3841542601585388,
+      "learning_rate": 0.0006,
+      "loss": 5.809355163574219,
+      "step": 580
+    },
+    {
+      "epoch": 31.053691275167786,
+      "grad_norm": 0.2733842432498932,
+      "learning_rate": 0.0006,
+      "loss": 5.817451095581054,
+      "step": 590
+    },
+    {
+      "epoch": 31.59060402684564,
+      "grad_norm": 0.48634791374206543,
+      "learning_rate": 0.0006,
+      "loss": 5.810215759277344,
+      "step": 600
+    },
+    {
+      "epoch": 32.10738255033557,
+      "grad_norm": 0.3789725601673126,
+      "learning_rate": 0.0006,
+      "loss": 5.801491928100586,
+      "step": 610
+    },
+    {
+      "epoch": 32.644295302013425,
+      "grad_norm": 0.2852318286895752,
+      "learning_rate": 0.0006,
+      "loss": 5.730844116210937,
+      "step": 620
+    },
+    {
+      "epoch": 33.16107382550336,
+      "grad_norm": 0.21317142248153687,
+      "learning_rate": 0.0006,
+      "loss": 5.756787109375,
+      "step": 630
+    },
+    {
+      "epoch": 33.69798657718121,
+      "grad_norm": 0.2483900785446167,
+      "learning_rate": 0.0006,
+      "loss": 5.73314208984375,
+      "step": 640
+    },
+    {
+      "epoch": 34.10738255033557,
+      "eval_loss": 4.715497970581055,
+      "eval_runtime": 2.6435,
+      "eval_samples_per_second": 923.387,
+      "eval_steps_per_second": 14.753,
+      "step": 648
+    },
+    {
+      "epoch": 34.214765100671144,
+      "grad_norm": 0.2699941098690033,
+      "learning_rate": 0.0006,
+      "loss": 5.675794219970703,
+      "step": 650
+    },
+    {
+      "epoch": 34.75167785234899,
+      "grad_norm": 0.3351564109325409,
+      "learning_rate": 0.0006,
+      "loss": 5.672748184204101,
+      "step": 660
+    },
+    {
+      "epoch": 35.26845637583892,
+      "grad_norm": 0.4180876612663269,
+      "learning_rate": 0.0006,
+      "loss": 5.5967670440673825,
+      "step": 670
+    },
+    {
+      "epoch": 35.80536912751678,
+      "grad_norm": 0.2991695702075958,
+      "learning_rate": 0.0006,
+      "loss": 5.660784912109375,
+      "step": 680
+    },
+    {
+      "epoch": 36.32214765100671,
+      "grad_norm": 0.29506081342697144,
+      "learning_rate": 0.0006,
+      "loss": 5.574668884277344,
+      "step": 690
+    },
+    {
+      "epoch": 36.85906040268456,
+      "grad_norm": 0.3907073736190796,
+      "learning_rate": 0.0006,
+      "loss": 5.627516555786133,
+      "step": 700
+    },
+    {
+      "epoch": 37.375838926174495,
+      "grad_norm": 0.3281225860118866,
+      "learning_rate": 0.0006,
+      "loss": 5.576396179199219,
+      "step": 710
+    },
+    {
+      "epoch": 37.91275167785235,
+      "grad_norm": 0.2437349110841751,
+      "learning_rate": 0.0006,
+      "loss": 5.558996963500976,
+      "step": 720
+    },
+    {
+      "epoch": 37.91275167785235,
+      "eval_loss": 4.60038948059082,
+      "eval_runtime": 2.6292,
+      "eval_samples_per_second": 928.426,
+      "eval_steps_per_second": 14.834,
+      "step": 720
+    },
+    {
+      "epoch": 38.42953020134228,
+      "grad_norm": 0.19309094548225403,
+      "learning_rate": 0.0006,
+      "loss": 5.529492950439453,
+      "step": 730
+    },
+    {
+      "epoch": 38.966442953020135,
+      "grad_norm": 0.22636272013187408,
+      "learning_rate": 0.0006,
+      "loss": 5.508900451660156,
+      "step": 740
+    },
+    {
+      "epoch": 39.48322147651007,
+      "grad_norm": 0.23093660175800323,
+      "learning_rate": 0.0006,
+      "loss": 5.457378005981445,
+      "step": 750
+    },
+    {
+      "epoch": 40.0,
+      "grad_norm": 0.29821616411209106,
+      "learning_rate": 0.0006,
+      "loss": 5.5060382843017575,
+      "step": 760
+    },
+    {
+      "epoch": 40.53691275167785,
+      "grad_norm": 0.4166114032268524,
+      "learning_rate": 0.0006,
+      "loss": 5.421920394897461,
+      "step": 770
+    },
+    {
+      "epoch": 41.053691275167786,
+      "grad_norm": 0.5481050610542297,
+      "learning_rate": 0.0006,
+      "loss": 5.5021240234375,
+      "step": 780
+    },
+    {
+      "epoch": 41.59060402684564,
+      "grad_norm": 0.4403257966041565,
+      "learning_rate": 0.0006,
+      "loss": 5.524919891357422,
+      "step": 790
+    },
+    {
+      "epoch": 41.69798657718121,
+      "eval_loss": 4.557999134063721,
+      "eval_runtime": 2.6465,
+      "eval_samples_per_second": 922.363,
+      "eval_steps_per_second": 14.737,
+      "step": 792
+    },
+    {
+      "epoch": 42.10738255033557,
+      "grad_norm": 0.2105201780796051,
+      "learning_rate": 0.0006,
+      "loss": 5.469275283813476,
+      "step": 800
+    },
+    {
+      "epoch": 42.644295302013425,
+      "grad_norm": 0.20716595649719238,
+      "learning_rate": 0.0006,
+      "loss": 5.373548126220703,
+      "step": 810
+    },
+    {
+      "epoch": 43.16107382550336,
+      "grad_norm": 0.21223177015781403,
+      "learning_rate": 0.0006,
+      "loss": 5.378014373779297,
+      "step": 820
+    },
+    {
+      "epoch": 43.69798657718121,
+      "grad_norm": 0.2113466113805771,
+      "learning_rate": 0.0006,
+      "loss": 5.346444320678711,
+      "step": 830
+    },
+    {
+      "epoch": 44.214765100671144,
+      "grad_norm": 0.2186392992734909,
+      "learning_rate": 0.0006,
+      "loss": 5.360507202148438,
+      "step": 840
+    },
+    {
+      "epoch": 44.75167785234899,
+      "grad_norm": 0.27498659491539,
+      "learning_rate": 0.0006,
+      "loss": 5.337562179565429,
+      "step": 850
+    },
+    {
+      "epoch": 45.26845637583892,
+      "grad_norm": 0.2543693482875824,
+      "learning_rate": 0.0006,
+      "loss": 5.3494621276855465,
+      "step": 860
+    },
+    {
+      "epoch": 45.48322147651007,
+      "eval_loss": 4.456889629364014,
+      "eval_runtime": 2.6315,
+      "eval_samples_per_second": 927.597,
+      "eval_steps_per_second": 14.82,
+      "step": 864
+    },
+    {
+      "epoch": 45.80536912751678,
+      "grad_norm": 0.3070426285266876,
+      "learning_rate": 0.0006,
+      "loss": 5.307219696044922,
+      "step": 870
+    },
+    {
+      "epoch": 46.32214765100671,
+      "grad_norm": 0.32272425293922424,
+      "learning_rate": 0.0006,
+      "loss": 5.3116401672363285,
+      "step": 880
+    },
+    {
+      "epoch": 46.85906040268456,
+      "grad_norm": 0.31429770588874817,
+      "learning_rate": 0.0006,
+      "loss": 5.34985580444336,
+      "step": 890
+    },
+    {
+      "epoch": 47.375838926174495,
+      "grad_norm": 0.3221947252750397,
+      "learning_rate": 0.0006,
+      "loss": 5.342152404785156,
+      "step": 900
+    },
+    {
+      "epoch": 47.91275167785235,
+      "grad_norm": 0.24101674556732178,
+      "learning_rate": 0.0006,
+      "loss": 5.268250274658203,
+      "step": 910
+    },
+    {
+      "epoch": 48.42953020134228,
+      "grad_norm": 0.2317378669977188,
+      "learning_rate": 0.0006,
+      "loss": 5.260585021972656,
+      "step": 920
+    },
+    {
+      "epoch": 48.966442953020135,
+      "grad_norm": 0.22440481185913086,
+      "learning_rate": 0.0006,
+      "loss": 5.240782165527344,
+      "step": 930
+    },
+    {
+      "epoch": 49.26845637583892,
+      "eval_loss": 4.423674583435059,
+      "eval_runtime": 2.6325,
+      "eval_samples_per_second": 927.249,
+      "eval_steps_per_second": 14.815,
+      "step": 936
+    },
+    {
+      "epoch": 49.48322147651007,
+      "grad_norm": 0.30753475427627563,
+      "learning_rate": 0.0006,
+      "loss": 5.258055114746094,
+      "step": 940
+    },
+    {
+      "epoch": 50.0,
+      "grad_norm": 0.2867712378501892,
+      "learning_rate": 0.0006,
+      "loss": 5.2097820281982425,
+      "step": 950
+    },
+    {
+      "epoch": 50.53691275167785,
+      "grad_norm": 0.28869014978408813,
+      "learning_rate": 0.0006,
+      "loss": 5.202045440673828,
+      "step": 960
+    },
+    {
+      "epoch": 51.053691275167786,
+      "grad_norm": 0.33684781193733215,
+      "learning_rate": 0.0006,
+      "loss": 5.205773162841797,
+      "step": 970
+    },
+    {
+      "epoch": 51.59060402684564,
+      "grad_norm": 0.33779972791671753,
+      "learning_rate": 0.0006,
+      "loss": 5.186534881591797,
+      "step": 980
+    },
+    {
+      "epoch": 52.10738255033557,
+      "grad_norm": 0.3655160665512085,
+      "learning_rate": 0.0006,
+      "loss": 5.227393341064453,
+      "step": 990
+    },
+    {
+      "epoch": 52.644295302013425,
+      "grad_norm": 0.3054697513580322,
+      "learning_rate": 0.0006,
+      "loss": 5.165704727172852,
+      "step": 1000
+    },
+    {
+      "epoch": 53.053691275167786,
+      "eval_loss": 4.391167640686035,
+      "eval_runtime": 2.6164,
+      "eval_samples_per_second": 932.964,
+      "eval_steps_per_second": 14.906,
+      "step": 1008
+    },
+    {
+      "epoch": 53.16107382550336,
+      "grad_norm": 0.24624931812286377,
+      "learning_rate": 0.0006,
+      "loss": 5.203597640991211,
+      "step": 1010
+    },
+    {
+      "epoch": 53.69798657718121,
+      "grad_norm": 0.23308025300502777,
+      "learning_rate": 0.0006,
+      "loss": 5.1529594421386715,
+      "step": 1020
+    },
+    {
+      "epoch": 54.214765100671144,
+      "grad_norm": 0.3395926356315613,
+      "learning_rate": 0.0006,
+      "loss": 5.155233764648438,
+      "step": 1030
+    },
+    {
+      "epoch": 54.75167785234899,
+      "grad_norm": 0.38559356331825256,
+      "learning_rate": 0.0006,
+      "loss": 5.216461181640625,
+      "step": 1040
+    },
+    {
+      "epoch": 55.26845637583892,
+      "grad_norm": 0.3221471309661865,
+      "learning_rate": 0.0006,
+      "loss": 5.181618499755859,
+      "step": 1050
+    },
+    {
+      "epoch": 55.80536912751678,
+      "grad_norm": 0.24010500311851501,
+      "learning_rate": 0.0006,
+      "loss": 5.106242370605469,
+      "step": 1060
+    },
+    {
+      "epoch": 56.32214765100671,
+      "grad_norm": 0.2482878714799881,
+      "learning_rate": 0.0006,
+      "loss": 5.09954605102539,
+      "step": 1070
+    },
+    {
+      "epoch": 56.85906040268456,
+      "grad_norm": 0.2414126992225647,
+      "learning_rate": 0.0006,
+      "loss": 5.142864227294922,
+      "step": 1080
+    },
+    {
+      "epoch": 56.85906040268456,
+      "eval_loss": 4.368844032287598,
+      "eval_runtime": 2.6419,
+      "eval_samples_per_second": 923.942,
+      "eval_steps_per_second": 14.762,
+      "step": 1080
+    },
+    {
+      "epoch": 57.375838926174495,
+      "grad_norm": 0.274541437625885,
+      "learning_rate": 0.0006,
+      "loss": 5.046080780029297,
+      "step": 1090
+    },
+    {
+      "epoch": 57.91275167785235,
+      "grad_norm": 0.26291564106941223,
+      "learning_rate": 0.0006,
+      "loss": 5.15833854675293,
+      "step": 1100
+    },
+    {
+      "epoch": 58.42953020134228,
+      "grad_norm": 0.37387144565582275,
+      "learning_rate": 0.0006,
+      "loss": 5.073323822021484,
+      "step": 1110
+    },
+    {
+      "epoch": 58.966442953020135,
+      "grad_norm": 0.2771231532096863,
+      "learning_rate": 0.0006,
+      "loss": 5.1004692077636715,
+      "step": 1120
+    },
+    {
+      "epoch": 59.48322147651007,
+      "grad_norm": 0.3077099919319153,
+      "learning_rate": 0.0006,
+      "loss": 5.069233703613281,
+      "step": 1130
+    },
+    {
+      "epoch": 60.0,
+      "grad_norm": 0.2758420705795288,
+      "learning_rate": 0.0006,
+      "loss": 5.102863311767578,
+      "step": 1140
+    },
+    {
+      "epoch": 60.53691275167785,
+      "grad_norm": 0.28643184900283813,
+      "learning_rate": 0.0006,
+      "loss": 5.020527648925781,
+      "step": 1150
+    },
+    {
+      "epoch": 60.644295302013425,
+      "eval_loss": 4.362539291381836,
+      "eval_runtime": 2.6422,
+      "eval_samples_per_second": 923.861,
+      "eval_steps_per_second": 14.761,
+      "step": 1152
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 7505,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 395,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": false,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 4.8448955098752614e+17,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/train.log b/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/train.log
new file mode 100644
index 0000000000000000000000000000000000000000..eb24265098b78faf0e6cf1fe02cf16da23058ab7
--- /dev/null
+++ b/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/train.log
@@ -0,0 +1,2467 @@
+W0818 18:21:30.512000 2926210 torch/distributed/run.py:851] 
+W0818 18:21:30.512000 2926210 torch/distributed/run.py:851] *****************************************
+W0818 18:21:30.512000 2926210 torch/distributed/run.py:851] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+W0818 18:21:30.512000 2926210 torch/distributed/run.py:851] *****************************************
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+EOS token: <|im_end|>
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+total params: 145836544
+[dataset] loaded packed cache: 9536 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+total params: 145836544
+total params: 145836544
+[dataset] loaded packed cache: 9536 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+total params: 145836544
+total params: 145836544
+[dataset] loaded packed cache: 9536 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[dataset] loaded packed cache: 9536 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[dataset] loaded packed cache: 9536 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+total params: 145836544
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+total params: 145836544
+[dataset] loaded packed cache: 9536 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[dataset] loaded packed cache: 9536 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+total params: 145836544
+[dataset] loaded packed cache: 9536 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+[transformers] ***** Running training *****
+[transformers]   Num examples = 9,536
+[transformers]   Num Epochs = 395
+[transformers]   Num update steps per epoch = 19
+[transformers]   Instantaneous batch size per device = 8
+[transformers]   Total train batch size (w. parallel, distributed & accumulation) = 512
+[transformers]   Gradient Accumulation steps = 8
+[transformers]   Total optimization steps = 7,505
+[transformers]   Number of trainable parameters = 145,836,544
+wandb: WARNING Changes to your `wandb` environment variables will be ignored because your `wandb` session has already started. For more information on how to modify your settings with `wandb.init()` arguments, please refer to https://wandb.me/wandb-init.
+wandb: setting up run obs138g0
+wandb: Tracking run with wandb version 0.28.2
+wandb: Run data is saved locally in /home/ubuntu/workspace/data-augmentations-for-pretraining/wandb/run-20260818_182149-obs138g0
+wandb: Run `wandb offline` to turn off syncing.
+wandb: Syncing run t2a_19M_C3_rand5_r2l50_i5exp
+wandb: ⭐️ View project at https://wandb.ai/slowrun-wankers/data-aug-pretraining
+wandb: 🚀 View run at https://wandb.ai/slowrun-wankers/data-aug-pretraining/runs/obs138g0
+
  0%|          | 0/7505 [00:00<?, ?it/s]wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+[rank2]:W0818 18:22:12.842000 2926782 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+[rank7]:W0818 18:22:12.948000 2926787 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+[rank3]:W0818 18:22:13.024000 2926783 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+[rank5]:W0818 18:22:13.043000 2926785 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+[rank6]:W0818 18:22:13.197000 2926786 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+[rank1]:W0818 18:22:13.259000 2926781 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+[rank4]:W0818 18:22:13.731000 2926784 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+[rank0]:W0818 18:22:14.846000 2926780 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+
  0%|          | 1/7505 [00:41<86:00:16, 41.26s/it]
                                                   
{'loss': '12.03', 'grad_norm': '2.529', 'learning_rate': '0', 'epoch': '0.05369'}
+
  0%|          | 1/7505 [00:41<86:00:16, 41.26s/it]
  0%|          | 2/7505 [00:41<36:13:04, 17.38s/it]
  0%|          | 3/7505 [00:42<20:18:56,  9.75s/it]
  0%|          | 4/7505 [00:43<12:49:51,  6.16s/it]
  0%|          | 5/7505 [00:43<8:41:20,  4.17s/it] 
  0%|          | 6/7505 [00:44<6:15:01,  3.00s/it]
  0%|          | 7/7505 [00:45<4:50:53,  2.33s/it]
  0%|          | 8/7505 [00:46<3:43:52,  1.79s/it]
  0%|          | 9/7505 [00:46<3:01:32,  1.45s/it]
  0%|          | 10/7505 [00:47<2:31:15,  1.21s/it]
                                                   
{'loss': '11.69', 'grad_norm': '1.677', 'learning_rate': '5.4e-05', 'epoch': '0.5369'}
+
  0%|          | 10/7505 [00:47<2:31:15,  1.21s/it]
  0%|          | 11/7505 [00:48<2:09:43,  1.04s/it]
  0%|          | 12/7505 [00:48<1:55:06,  1.08it/s]
  0%|          | 13/7505 [00:49<1:44:39,  1.19it/s]
  0%|          | 14/7505 [00:50<1:38:14,  1.27it/s]
  0%|          | 15/7505 [00:50<1:33:18,  1.34it/s]
  0%|          | 16/7505 [00:51<1:29:26,  1.40it/s]
  0%|          | 17/7505 [00:52<1:26:42,  1.44it/s]
  0%|          | 18/7505 [00:52<1:25:31,  1.46it/s]
  0%|          | 19/7505 [00:53<1:15:08,  1.66it/s]
  0%|          | 20/7505 [00:53<1:20:45,  1.54it/s]
                                                   
{'loss': '11.06', 'grad_norm': '1.594', 'learning_rate': '0.000114', 'epoch': '1.054'}
+
  0%|          | 20/7505 [00:53<1:20:45,  1.54it/s]
  0%|          | 21/7505 [00:54<1:20:37,  1.55it/s]
  0%|          | 22/7505 [00:55<1:20:33,  1.55it/s]
  0%|          | 23/7505 [00:56<1:24:16,  1.48it/s]
  0%|          | 24/7505 [00:56<1:29:06,  1.40it/s]
  0%|          | 25/7505 [00:57<1:27:43,  1.42it/s]
  0%|          | 26/7505 [00:58<1:26:24,  1.44it/s]
  0%|          | 27/7505 [00:58<1:24:29,  1.48it/s]
  0%|          | 28/7505 [00:59<1:23:32,  1.49it/s]
  0%|          | 29/7505 [01:00<1:23:20,  1.49it/s]
  0%|          | 30/7505 [01:00<1:22:24,  1.51it/s]
                                                   
{'loss': '10.35', 'grad_norm': '1.563', 'learning_rate': '0.000174', 'epoch': '1.591'}
+
  0%|          | 30/7505 [01:00<1:22:24,  1.51it/s]
  0%|          | 31/7505 [01:01<1:21:45,  1.52it/s]
  0%|          | 32/7505 [01:02<1:21:45,  1.52it/s]
  0%|          | 33/7505 [01:02<1:22:07,  1.52it/s]
  0%|          | 34/7505 [01:03<1:36:23,  1.29it/s]
  0%|          | 35/7505 [01:04<1:46:57,  1.16it/s]
  0%|          | 36/7505 [01:05<1:38:57,  1.26it/s]
  0%|          | 37/7505 [01:06<1:46:30,  1.17it/s]
  1%|          | 38/7505 [01:06<1:30:23,  1.38it/s]
  1%|          | 39/7505 [01:07<1:32:05,  1.35it/s]
  1%|          | 40/7505 [01:08<1:41:35,  1.22it/s]
                                                   
{'loss': '9.38', 'grad_norm': '1.386', 'learning_rate': '0.000234', 'epoch': '2.107'}
+
  1%|          | 40/7505 [01:08<1:41:35,  1.22it/s]
  1%|          | 41/7505 [01:09<1:36:50,  1.28it/s]
  1%|          | 42/7505 [01:10<1:35:33,  1.30it/s]
  1%|          | 43/7505 [01:11<1:52:31,  1.11it/s]
  1%|          | 44/7505 [01:11<1:42:40,  1.21it/s]
  1%|          | 45/7505 [01:12<1:48:38,  1.14it/s]
  1%|          | 46/7505 [01:13<1:43:51,  1.20it/s]
  1%|          | 47/7505 [01:14<1:37:30,  1.27it/s]
  1%|          | 48/7505 [01:15<1:32:16,  1.35it/s]
  1%|          | 49/7505 [01:15<1:28:37,  1.40it/s]
  1%|          | 50/7505 [01:16<1:27:01,  1.43it/s]
                                                   
{'loss': '8.49', 'grad_norm': '0.8522', 'learning_rate': '0.000294', 'epoch': '2.644'}
+
  1%|          | 50/7505 [01:16<1:27:01,  1.43it/s]
  1%|          | 51/7505 [01:16<1:25:01,  1.46it/s]
  1%|          | 52/7505 [01:17<1:23:30,  1.49it/s]
  1%|          | 53/7505 [01:18<1:24:55,  1.46it/s]
  1%|          | 54/7505 [01:18<1:23:43,  1.48it/s]
  1%|          | 55/7505 [01:20<1:45:55,  1.17it/s]
  1%|          | 56/7505 [01:20<1:40:03,  1.24it/s]
  1%|          | 57/7505 [01:21<1:31:27,  1.36it/s]
  1%|          | 58/7505 [01:22<1:47:10,  1.16it/s]
  1%|          | 59/7505 [01:23<1:38:55,  1.25it/s]
  1%|          | 60/7505 [01:24<1:35:35,  1.30it/s]
                                                   
{'loss': '8.031', 'grad_norm': '0.2252', 'learning_rate': '0.000354', 'epoch': '3.161'}
+
  1%|          | 60/7505 [01:24<1:35:35,  1.30it/s]
  1%|          | 61/7505 [01:24<1:30:55,  1.36it/s]
  1%|          | 62/7505 [01:25<1:28:26,  1.40it/s]
  1%|          | 63/7505 [01:26<1:32:51,  1.34it/s]
  1%|          | 64/7505 [01:26<1:29:01,  1.39it/s]
  1%|          | 65/7505 [01:27<1:26:44,  1.43it/s]
  1%|          | 66/7505 [01:28<1:24:43,  1.46it/s]
  1%|          | 67/7505 [01:28<1:23:40,  1.48it/s]
  1%|          | 68/7505 [01:29<1:38:08,  1.26it/s]
  1%|          | 69/7505 [01:30<1:32:34,  1.34it/s]
  1%|          | 70/7505 [01:31<1:43:33,  1.20it/s]
                                                   
{'loss': '7.934', 'grad_norm': '0.1391', 'learning_rate': '0.000414', 'epoch': '3.698'}
+
  1%|          | 70/7505 [01:31<1:43:33,  1.20it/s]
  1%|          | 71/7505 [01:32<1:46:12,  1.17it/s]
  1%|          | 72/7505 [01:33<1:46:54,  1.16it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
 10%|█         | 4/39 [00:00<00:01, 25.34it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 10.94it/s]
+
 23%|██▎       | 9/39 [00:00<00:03,  9.21it/s]
+
 28%|██▊       | 11/39 [00:01<00:03,  8.42it/s]
+
 31%|███       | 12/39 [00:01<00:03,  8.15it/s]
+
 33%|███▎      | 13/39 [00:01<00:03,  7.90it/s]
+
 36%|███▌      | 14/39 [00:01<00:03,  7.72it/s]
+
 38%|███▊      | 15/39 [00:01<00:03,  7.58it/s]
+
 41%|████      | 16/39 [00:01<00:02,  7.98it/s]
+
 44%|████▎     | 17/39 [00:01<00:02,  7.72it/s]
+
 46%|████▌     | 18/39 [00:02<00:02,  7.48it/s]
+
 49%|████▊     | 19/39 [00:02<00:02,  7.35it/s]
+
 51%|█████▏    | 20/39 [00:02<00:02,  7.25it/s]
+
 54%|█████▍    | 21/39 [00:02<00:02,  7.15it/s]
+
 56%|█████▋    | 22/39 [00:02<00:02,  7.58it/s]
+
 59%|█████▉    | 23/39 [00:02<00:02,  7.42it/s]
+
 62%|██████▏   | 24/39 [00:02<00:02,  7.29it/s]
+
 64%|██████▍   | 25/39 [00:03<00:01,  7.21it/s]
+
 67%|██████▋   | 26/39 [00:03<00:01,  7.16it/s]
+
 69%|██████▉   | 27/39 [00:03<00:01,  7.11it/s]
+
 72%|███████▏  | 28/39 [00:03<00:01,  7.08it/s]
+
 74%|███████▍  | 29/39 [00:03<00:01,  7.00it/s]
+
 77%|███████▋  | 30/39 [00:03<00:01,  6.98it/s]
+
 79%|███████▉  | 31/39 [00:03<00:01,  7.03it/s]
+
 82%|████████▏ | 32/39 [00:04<00:00,  7.08it/s]
+
 87%|████████▋ | 34/39 [00:04<00:00,  8.86it/s]
+
 92%|█████████▏| 36/39 [00:04<00:00, 10.62it/s]
+
 97%|█████████▋| 38/39 [00:04<00:00, 11.92it/s]
                                                   
+
                                               
{'eval_loss': '7.539', 'eval_runtime': '29.19', 'eval_samples_per_second': '83.62', 'eval_steps_per_second': '1.336', 'epoch': '3.805'}
+
  1%|          | 72/7505 [02:02<1:46:54,  1.16it/s]
+
100%|██████████| 39/39 [00:04<00:00, 11.92it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-72
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-72/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-72/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.74it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.73it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-72/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-72/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-72/tokenizer_config.json
+[eval] epoch 3.805 step 72 loss 7.5389 (best 7.5389 @ ep 3.805, 0/20 since best)
+
+
                                               
  1%|          | 73/7505 [02:05<20:55:41, 10.14s/it]
  1%|          | 74/7505 [02:05<15:05:43,  7.31s/it]
  1%|          | 75/7505 [02:06<10:58:54,  5.32s/it]
  1%|          | 76/7505 [02:06<7:56:22,  3.85s/it] 
  1%|          | 77/7505 [02:07<6:01:28,  2.92s/it]
  1%|          | 78/7505 [02:08<4:37:06,  2.24s/it]
  1%|          | 79/7505 [02:08<3:38:01,  1.76s/it]
  1%|          | 80/7505 [02:09<3:04:25,  1.49s/it]
                                                   
{'loss': '7.866', 'grad_norm': '0.5182', 'learning_rate': '0.000474', 'epoch': '4.215'}
+
  1%|          | 80/7505 [02:09<3:04:25,  1.49s/it]
  1%|          | 81/7505 [02:10<2:52:18,  1.39s/it]
  1%|          | 82/7505 [02:12<2:42:41,  1.32s/it]
  1%|          | 83/7505 [02:12<2:18:04,  1.12s/it]
  1%|          | 84/7505 [02:13<2:15:14,  1.09s/it]
  1%|          | 85/7505 [02:14<1:59:40,  1.03it/s]
  1%|          | 86/7505 [02:15<2:01:24,  1.02it/s]
  1%|          | 87/7505 [02:16<1:54:08,  1.08it/s]
  1%|          | 88/7505 [02:17<1:53:51,  1.09it/s]
  1%|          | 89/7505 [02:17<1:43:51,  1.19it/s]
  1%|          | 90/7505 [02:18<1:52:40,  1.10it/s]
                                                   
{'loss': '7.826', 'grad_norm': '0.7047', 'learning_rate': '0.000534', 'epoch': '4.752'}
+
  1%|          | 90/7505 [02:18<1:52:40,  1.10it/s]
  1%|          | 91/7505 [02:19<1:44:17,  1.18it/s]
  1%|          | 92/7505 [02:20<1:50:51,  1.11it/s]
  1%|          | 93/7505 [02:21<1:45:39,  1.17it/s]
  1%|▏         | 94/7505 [02:22<1:48:30,  1.14it/s]
  1%|▏         | 95/7505 [02:22<1:36:29,  1.28it/s]
  1%|▏         | 96/7505 [02:24<1:51:53,  1.10it/s]
  1%|▏         | 97/7505 [02:24<1:46:06,  1.16it/s]
  1%|▏         | 98/7505 [02:26<2:00:27,  1.02it/s]
  1%|▏         | 99/7505 [02:27<2:02:55,  1.00it/s]
  1%|▏         | 100/7505 [02:28<2:09:31,  1.05s/it]
                                                    
{'loss': '7.8', 'grad_norm': '0.2292', 'learning_rate': '0.000594', 'epoch': '5.268'}
+
  1%|▏         | 100/7505 [02:28<2:09:31,  1.05s/it]
  1%|▏         | 101/7505 [02:29<2:21:20,  1.15s/it]
  1%|▏         | 102/7505 [02:30<2:13:28,  1.08s/it]
  1%|▏         | 103/7505 [02:31<2:16:16,  1.10s/it]
  1%|▏         | 104/7505 [02:32<2:18:36,  1.12s/it]
  1%|▏         | 105/7505 [02:33<2:14:53,  1.09s/it]
  1%|▏         | 106/7505 [02:34<2:09:40,  1.05s/it]
  1%|▏         | 107/7505 [02:36<2:12:17,  1.07s/it]
  1%|▏         | 108/7505 [02:37<2:15:31,  1.10s/it]
  1%|▏         | 109/7505 [02:37<2:01:59,  1.01it/s]
  1%|▏         | 110/7505 [02:38<1:54:03,  1.08it/s]
                                                    
{'loss': '7.776', 'grad_norm': '0.2174', 'learning_rate': '0.0006', 'epoch': '5.805'}
+
  1%|▏         | 110/7505 [02:38<1:54:03,  1.08it/s]
  1%|▏         | 111/7505 [02:39<2:03:18,  1.00s/it]
  1%|▏         | 112/7505 [02:40<1:54:28,  1.08it/s]
  2%|▏         | 113/7505 [02:41<1:47:50,  1.14it/s]
  2%|▏         | 114/7505 [02:42<1:40:33,  1.23it/s]
  2%|▏         | 115/7505 [02:43<1:48:13,  1.14it/s]
  2%|▏         | 116/7505 [02:43<1:39:47,  1.23it/s]
  2%|▏         | 117/7505 [02:44<1:48:52,  1.13it/s]
  2%|▏         | 118/7505 [02:45<1:41:45,  1.21it/s]
  2%|▏         | 119/7505 [02:46<1:49:36,  1.12it/s]
  2%|▏         | 120/7505 [02:47<1:40:42,  1.22it/s]
                                                    
{'loss': '7.74', 'grad_norm': '0.2106', 'learning_rate': '0.0006', 'epoch': '6.322'}
+
  2%|▏         | 120/7505 [02:47<1:40:42,  1.22it/s]
  2%|▏         | 121/7505 [02:48<1:41:54,  1.21it/s]
  2%|▏         | 122/7505 [02:49<1:53:45,  1.08it/s]
  2%|▏         | 123/7505 [02:49<1:47:27,  1.14it/s]
  2%|▏         | 124/7505 [02:50<1:42:08,  1.20it/s]
  2%|▏         | 125/7505 [02:51<1:46:04,  1.16it/s]
  2%|▏         | 126/7505 [02:52<1:46:11,  1.16it/s]
  2%|▏         | 127/7505 [02:53<1:45:03,  1.17it/s]
  2%|▏         | 128/7505 [02:53<1:38:19,  1.25it/s]
  2%|▏         | 129/7505 [02:55<1:55:27,  1.06it/s]
  2%|▏         | 130/7505 [02:56<2:03:19,  1.00s/it]
                                                    
{'loss': '7.707', 'grad_norm': '0.1769', 'learning_rate': '0.0006', 'epoch': '6.859'}
+
  2%|▏         | 130/7505 [02:56<2:03:19,  1.00s/it]
  2%|▏         | 131/7505 [02:57<2:08:25,  1.05s/it]
  2%|▏         | 132/7505 [02:58<1:55:06,  1.07it/s]
  2%|▏         | 133/7505 [02:58<1:35:15,  1.29it/s]
  2%|▏         | 134/7505 [03:00<1:58:53,  1.03it/s]
  2%|▏         | 135/7505 [03:00<1:46:58,  1.15it/s]
  2%|▏         | 136/7505 [03:01<2:02:10,  1.01it/s]
  2%|▏         | 137/7505 [03:02<1:51:44,  1.10it/s]
  2%|▏         | 138/7505 [03:03<1:47:18,  1.14it/s]
  2%|▏         | 139/7505 [03:04<1:59:17,  1.03it/s]
  2%|▏         | 140/7505 [03:05<2:11:50,  1.07s/it]
                                                    
{'loss': '7.688', 'grad_norm': '0.3496', 'learning_rate': '0.0006', 'epoch': '7.376'}
+
  2%|▏         | 140/7505 [03:05<2:11:50,  1.07s/it]
  2%|▏         | 141/7505 [03:06<1:57:21,  1.05it/s]
  2%|▏         | 142/7505 [03:07<2:03:25,  1.01s/it]
  2%|▏         | 143/7505 [03:08<2:06:30,  1.03s/it]
  2%|▏         | 144/7505 [03:09<1:52:04,  1.09it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 22.67it/s]
+
 15%|█▌        | 6/39 [00:00<00:02, 13.15it/s]
+
 21%|██        | 8/39 [00:00<00:03, 10.12it/s]
+
 26%|██▌       | 10/39 [00:00<00:03,  8.97it/s]
+
 28%|██▊       | 11/39 [00:01<00:03,  8.80it/s]
+
 33%|███▎      | 13/39 [00:01<00:02, 10.36it/s]
+
 38%|███▊      | 15/39 [00:01<00:02,  9.25it/s]
+
 44%|████▎     | 17/39 [00:01<00:02,  8.35it/s]
+
 46%|████▌     | 18/39 [00:01<00:02,  8.07it/s]
+
 49%|████▊     | 19/39 [00:02<00:02,  7.82it/s]
+
 51%|█████▏    | 20/39 [00:02<00:02,  7.61it/s]
+
 54%|█████▍    | 21/39 [00:02<00:02,  7.46it/s]
+
 56%|█████▋    | 22/39 [00:02<00:02,  7.31it/s]
+
 59%|█████▉    | 23/39 [00:02<00:02,  7.20it/s]
+
 62%|██████▏   | 24/39 [00:02<00:02,  7.11it/s]
+
 64%|██████▍   | 25/39 [00:02<00:01,  7.04it/s]
+
 67%|██████▋   | 26/39 [00:03<00:01,  7.44it/s]
+
 72%|███████▏  | 28/39 [00:03<00:01,  9.69it/s]
+
 77%|███████▋  | 30/39 [00:03<00:00, 11.23it/s]
+
 82%|████████▏ | 32/39 [00:03<00:00, 12.40it/s]
+
 87%|████████▋ | 34/39 [00:03<00:00, 13.22it/s]
+
 92%|█████████▏| 36/39 [00:03<00:00, 13.81it/s]
+
 97%|█████████▋| 38/39 [00:03<00:00, 12.52it/s]
                                                    
+
                                               
{'eval_loss': '7.154', 'eval_runtime': '4.243', 'eval_samples_per_second': '575.3', 'eval_steps_per_second': '9.192', 'epoch': '7.591'}
+
  2%|▏         | 144/7505 [03:13<1:52:04,  1.09it/s]
+
100%|██████████| 39/39 [00:04<00:00, 12.52it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-144
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-144/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-144/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.99it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.98it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-144/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-144/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-144/tokenizer_config.json
+[eval] epoch 7.591 step 144 loss 7.1545 (best 7.1545 @ ep 7.591, 0/20 since best)
+
+
                                               
  2%|▏         | 145/7505 [03:16<5:40:19,  2.77s/it]
  2%|▏         | 146/7505 [03:17<4:32:50,  2.22s/it]
  2%|▏         | 147/7505 [03:18<3:37:41,  1.78s/it]
  2%|▏         | 148/7505 [03:19<3:01:30,  1.48s/it]
  2%|▏         | 149/7505 [03:19<2:40:25,  1.31s/it]
  2%|▏         | 150/7505 [03:20<2:24:16,  1.18s/it]
                                                    
{'loss': '7.647', 'grad_norm': '0.3215', 'learning_rate': '0.0006', 'epoch': '7.913'}
+
  2%|▏         | 150/7505 [03:20<2:24:16,  1.18s/it]
  2%|▏         | 151/7505 [03:21<2:23:41,  1.17s/it]
  2%|▏         | 152/7505 [03:22<2:00:31,  1.02it/s]
  2%|▏         | 153/7505 [03:23<2:09:52,  1.06s/it]
  2%|▏         | 154/7505 [03:24<2:14:05,  1.09s/it]
  2%|▏         | 155/7505 [03:26<2:14:52,  1.10s/it]
  2%|▏         | 156/7505 [03:26<1:58:19,  1.04it/s]
  2%|▏         | 157/7505 [03:27<1:56:15,  1.05it/s]
  2%|▏         | 158/7505 [03:28<2:04:06,  1.01s/it]
  2%|▏         | 159/7505 [03:29<1:51:15,  1.10it/s]
  2%|▏         | 160/7505 [03:30<1:48:20,  1.13it/s]
                                                    
{'loss': '7.597', 'grad_norm': '0.4054', 'learning_rate': '0.0006', 'epoch': '8.43'}
+
  2%|▏         | 160/7505 [03:30<1:48:20,  1.13it/s]
  2%|▏         | 161/7505 [03:31<1:58:41,  1.03it/s]
  2%|▏         | 162/7505 [03:32<2:06:07,  1.03s/it]
  2%|▏         | 163/7505 [03:33<2:00:22,  1.02it/s]
  2%|▏         | 164/7505 [03:34<2:00:22,  1.02it/s]
  2%|▏         | 165/7505 [03:35<1:55:46,  1.06it/s]
  2%|▏         | 166/7505 [03:36<2:03:58,  1.01s/it]
  2%|▏         | 167/7505 [03:37<2:07:39,  1.04s/it]
  2%|▏         | 168/7505 [03:38<1:54:02,  1.07it/s]
  2%|▏         | 169/7505 [03:39<1:56:49,  1.05it/s]
  2%|▏         | 170/7505 [03:40<2:01:10,  1.01it/s]
                                                    
{'loss': '7.595', 'grad_norm': '0.2425', 'learning_rate': '0.0006', 'epoch': '8.966'}
+
  2%|▏         | 170/7505 [03:40<2:01:10,  1.01it/s]
  2%|▏         | 171/7505 [03:40<1:46:42,  1.15it/s]
  2%|▏         | 172/7505 [03:42<1:57:52,  1.04it/s]
  2%|▏         | 173/7505 [03:43<2:02:33,  1.00s/it]
  2%|▏         | 174/7505 [03:44<2:08:16,  1.05s/it]
  2%|▏         | 175/7505 [03:45<2:05:15,  1.03s/it]
  2%|▏         | 176/7505 [03:46<1:57:32,  1.04it/s]
  2%|▏         | 177/7505 [03:47<2:04:21,  1.02s/it]
  2%|▏         | 178/7505 [03:48<2:11:53,  1.08s/it]
  2%|▏         | 179/7505 [03:49<1:55:50,  1.05it/s]
  2%|▏         | 180/7505 [03:50<2:05:00,  1.02s/it]
                                                    
{'loss': '7.522', 'grad_norm': '0.6031', 'learning_rate': '0.0006', 'epoch': '9.483'}
+
  2%|▏         | 180/7505 [03:50<2:05:00,  1.02s/it]
  2%|▏         | 181/7505 [03:51<1:57:22,  1.04it/s]
  2%|▏         | 182/7505 [03:52<1:58:54,  1.03it/s]
  2%|▏         | 183/7505 [03:53<2:01:38,  1.00it/s]
  2%|▏         | 184/7505 [03:54<1:58:11,  1.03it/s]
  2%|▏         | 185/7505 [03:55<2:05:53,  1.03s/it]
  2%|▏         | 186/7505 [03:56<2:14:22,  1.10s/it]
  2%|▏         | 187/7505 [03:57<2:20:07,  1.15s/it]
  3%|▎         | 188/7505 [03:59<2:20:51,  1.16s/it]
  3%|▎         | 189/7505 [04:00<2:16:38,  1.12s/it]
  3%|▎         | 190/7505 [04:00<1:50:22,  1.10it/s]
                                                    
{'loss': '7.484', 'grad_norm': '0.6346', 'learning_rate': '0.0006', 'epoch': '10'}
+
  3%|▎         | 190/7505 [04:00<1:50:22,  1.10it/s]
  3%|▎         | 191/7505 [04:01<2:04:27,  1.02s/it]
  3%|▎         | 192/7505 [04:02<2:09:21,  1.06s/it]
  3%|▎         | 193/7505 [04:04<2:13:10,  1.09s/it]
  3%|▎         | 194/7505 [04:05<2:16:11,  1.12s/it]
  3%|▎         | 195/7505 [04:06<2:17:34,  1.13s/it]
  3%|▎         | 196/7505 [04:07<2:25:13,  1.19s/it]
  3%|▎         | 197/7505 [04:08<2:23:40,  1.18s/it]
  3%|▎         | 198/7505 [04:10<2:23:01,  1.17s/it]
  3%|▎         | 199/7505 [04:11<2:19:21,  1.14s/it]
  3%|▎         | 200/7505 [04:11<2:01:19,  1.00it/s]
                                                    
{'loss': '7.434', 'grad_norm': '0.3115', 'learning_rate': '0.0006', 'epoch': '10.54'}
+
  3%|▎         | 200/7505 [04:11<2:01:19,  1.00it/s]
  3%|▎         | 201/7505 [04:12<2:04:39,  1.02s/it]
  3%|▎         | 202/7505 [04:13<1:52:25,  1.08it/s]
  3%|▎         | 203/7505 [04:14<2:01:02,  1.01it/s]
  3%|▎         | 204/7505 [04:15<1:57:29,  1.04it/s]
  3%|▎         | 205/7505 [04:16<1:46:02,  1.15it/s]
  3%|▎         | 206/7505 [04:17<1:53:33,  1.07it/s]
  3%|▎         | 207/7505 [04:18<1:45:17,  1.16it/s]
  3%|▎         | 208/7505 [04:19<1:50:15,  1.10it/s]
  3%|▎         | 209/7505 [04:19<1:32:02,  1.32it/s]
  3%|▎         | 210/7505 [04:20<1:49:02,  1.11it/s]
                                                    
{'loss': '7.282', 'grad_norm': '0.19', 'learning_rate': '0.0006', 'epoch': '11.05'}
+
  3%|▎         | 210/7505 [04:20<1:49:02,  1.11it/s]
  3%|▎         | 211/7505 [04:21<1:52:39,  1.08it/s]
  3%|▎         | 212/7505 [04:22<1:42:39,  1.18it/s]
  3%|▎         | 213/7505 [04:23<1:51:08,  1.09it/s]
  3%|▎         | 214/7505 [04:24<1:41:29,  1.20it/s]
  3%|▎         | 215/7505 [04:25<1:51:03,  1.09it/s]
  3%|▎         | 216/7505 [04:25<1:42:24,  1.19it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:02, 13.41it/s]
+
 13%|█▎        | 5/39 [00:00<00:03,  9.23it/s]
+
 15%|█▌        | 6/39 [00:00<00:03,  8.49it/s]
+
 18%|█▊        | 7/39 [00:00<00:03,  8.07it/s]
+
 23%|██▎       | 9/39 [00:00<00:03,  9.35it/s]
+
 28%|██▊       | 11/39 [00:01<00:02, 10.92it/s]
+
 33%|███▎      | 13/39 [00:01<00:02, 12.14it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 13.01it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.91it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 13.56it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 14.02it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.54it/s]
+
 64%|██████▍   | 25/39 [00:02<00:01, 10.27it/s]
+
 69%|██████▉   | 27/39 [00:02<00:01,  9.10it/s]
+
 74%|███████▍  | 29/39 [00:02<00:01,  8.96it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00,  9.97it/s]
+
 85%|████████▍ | 33/39 [00:03<00:00,  8.82it/s]
+
 87%|████████▋ | 34/39 [00:03<00:00,  8.43it/s]
+
 90%|████████▉ | 35/39 [00:03<00:00,  8.10it/s]
+
 92%|█████████▏| 36/39 [00:03<00:00,  7.83it/s]
+
 97%|█████████▋| 38/39 [00:03<00:00,  9.45it/s]
+
100%|██████████| 39/39 [00:03<00:00,  9.43it/s]
                                                    
+
                                               
{'eval_loss': '6.644', 'eval_runtime': '4.092', 'eval_samples_per_second': '596.6', 'eval_steps_per_second': '9.531', 'epoch': '11.38'}
+
  3%|▎         | 216/7505 [04:29<1:42:24,  1.19it/s]
+
100%|██████████| 39/39 [00:04<00:00,  9.43it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-216
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-216/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-216/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.96it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.96it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-216/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-216/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-216/tokenizer_config.json
+[eval] epoch 11.376 step 216 loss 6.6443 (best 6.6443 @ ep 11.376, 0/20 since best)
+
+
                                               
  3%|▎         | 217/7505 [04:32<5:21:22,  2.65s/it]
  3%|▎         | 218/7505 [04:33<4:25:13,  2.18s/it]
  3%|▎         | 219/7505 [04:34<3:45:40,  1.86s/it]
  3%|▎         | 220/7505 [04:35<3:01:36,  1.50s/it]
                                                    
{'loss': '7.254', 'grad_norm': '0.332', 'learning_rate': '0.0006', 'epoch': '11.59'}
+
  3%|▎         | 220/7505 [04:35<3:01:36,  1.50s/it]
  3%|▎         | 221/7505 [04:36<2:50:05,  1.40s/it]
  3%|▎         | 222/7505 [04:37<2:42:03,  1.34s/it]
  3%|▎         | 223/7505 [04:38<2:30:58,  1.24s/it]
  3%|▎         | 224/7505 [04:39<2:09:45,  1.07s/it]
  3%|▎         | 225/7505 [04:40<2:00:03,  1.01it/s]
  3%|▎         | 226/7505 [04:41<1:48:40,  1.12it/s]
  3%|▎         | 227/7505 [04:41<1:44:49,  1.16it/s]
  3%|▎         | 228/7505 [04:42<1:44:02,  1.17it/s]
  3%|▎         | 229/7505 [04:43<1:39:43,  1.22it/s]
  3%|▎         | 230/7505 [04:44<1:36:44,  1.25it/s]
                                                    
{'loss': '7.202', 'grad_norm': '0.5623', 'learning_rate': '0.0006', 'epoch': '12.11'}
+
  3%|▎         | 230/7505 [04:44<1:36:44,  1.25it/s]
  3%|▎         | 231/7505 [04:44<1:31:33,  1.32it/s]
  3%|▎         | 232/7505 [04:45<1:28:55,  1.36it/s]
  3%|▎         | 233/7505 [04:46<1:26:18,  1.40it/s]
  3%|▎         | 234/7505 [04:46<1:24:07,  1.44it/s]
  3%|▎         | 235/7505 [04:47<1:22:39,  1.47it/s]
  3%|▎         | 236/7505 [04:48<1:22:35,  1.47it/s]
  3%|▎         | 237/7505 [04:48<1:21:30,  1.49it/s]
  3%|▎         | 238/7505 [04:49<1:35:06,  1.27it/s]
  3%|▎         | 239/7505 [04:51<1:49:36,  1.10it/s]
  3%|▎         | 240/7505 [04:52<1:58:42,  1.02it/s]
                                                    
{'loss': '7.122', 'grad_norm': '0.3505', 'learning_rate': '0.0006', 'epoch': '12.64'}
+
  3%|▎         | 240/7505 [04:52<1:58:42,  1.02it/s]
  3%|▎         | 241/7505 [04:53<2:04:21,  1.03s/it]
  3%|▎         | 242/7505 [04:54<1:53:30,  1.07it/s]
  3%|▎         | 243/7505 [04:55<2:00:25,  1.01it/s]
  3%|▎         | 244/7505 [04:55<1:48:05,  1.12it/s]
  3%|▎         | 245/7505 [04:57<1:55:11,  1.05it/s]
  3%|▎         | 246/7505 [04:58<1:59:01,  1.02it/s]
  3%|▎         | 247/7505 [04:58<1:52:45,  1.07it/s]
  3%|▎         | 248/7505 [04:59<1:46:00,  1.14it/s]
  3%|▎         | 249/7505 [05:00<1:51:13,  1.09it/s]
  3%|▎         | 250/7505 [05:01<1:57:01,  1.03it/s]
                                                    
{'loss': '7.02', 'grad_norm': '0.3924', 'learning_rate': '0.0006', 'epoch': '13.16'}
+
  3%|▎         | 250/7505 [05:01<1:57:01,  1.03it/s]
  3%|▎         | 251/7505 [05:02<2:00:15,  1.01it/s]
  3%|▎         | 252/7505 [05:04<2:09:32,  1.07s/it]
  3%|▎         | 253/7505 [05:05<2:11:32,  1.09s/it]
  3%|▎         | 254/7505 [05:06<2:10:07,  1.08s/it]
  3%|▎         | 255/7505 [05:07<2:08:50,  1.07s/it]
  3%|▎         | 256/7505 [05:07<1:53:50,  1.06it/s]
  3%|▎         | 257/7505 [05:08<1:58:47,  1.02it/s]
  3%|▎         | 258/7505 [05:10<2:05:45,  1.04s/it]
  3%|▎         | 259/7505 [05:11<2:10:08,  1.08s/it]
  3%|▎         | 260/7505 [05:12<1:59:17,  1.01it/s]
                                                    
{'loss': '7.007', 'grad_norm': '0.441', 'learning_rate': '0.0006', 'epoch': '13.7'}
+
  3%|▎         | 260/7505 [05:12<1:59:17,  1.01it/s]
  3%|▎         | 261/7505 [05:12<1:49:27,  1.10it/s]
  3%|▎         | 262/7505 [05:13<1:52:17,  1.08it/s]
  4%|▎         | 263/7505 [05:14<1:42:13,  1.18it/s]
  4%|▎         | 264/7505 [05:15<1:50:15,  1.09it/s]
  4%|▎         | 265/7505 [05:16<1:40:48,  1.20it/s]
  4%|▎         | 266/7505 [05:16<1:30:56,  1.33it/s]
  4%|▎         | 267/7505 [05:18<1:49:47,  1.10it/s]
  4%|▎         | 268/7505 [05:19<1:55:07,  1.05it/s]
  4%|▎         | 269/7505 [05:20<1:59:24,  1.01it/s]
  4%|▎         | 270/7505 [05:21<2:05:35,  1.04s/it]
                                                    
{'loss': '6.952', 'grad_norm': '0.4397', 'learning_rate': '0.0006', 'epoch': '14.21'}
+
  4%|▎         | 270/7505 [05:21<2:05:35,  1.04s/it]
  4%|▎         | 271/7505 [05:22<2:10:03,  1.08s/it]
  4%|▎         | 272/7505 [05:23<2:19:00,  1.15s/it]
  4%|▎         | 273/7505 [05:24<2:18:20,  1.15s/it]
  4%|▎         | 274/7505 [05:25<2:12:02,  1.10s/it]
  4%|▎         | 275/7505 [05:27<2:14:44,  1.12s/it]
  4%|▎         | 276/7505 [05:27<2:02:39,  1.02s/it]
  4%|▎         | 277/7505 [05:28<2:03:17,  1.02s/it]
  4%|▎         | 278/7505 [05:29<1:51:53,  1.08it/s]
  4%|▎         | 279/7505 [05:30<1:47:54,  1.12it/s]
  4%|▎         | 280/7505 [05:31<1:56:30,  1.03it/s]
                                                    
{'loss': '6.868', 'grad_norm': '0.4134', 'learning_rate': '0.0006', 'epoch': '14.75'}
+
  4%|▎         | 280/7505 [05:31<1:56:30,  1.03it/s]
  4%|▎         | 281/7505 [05:32<1:50:34,  1.09it/s]
  4%|▍         | 282/7505 [05:33<1:43:26,  1.16it/s]
  4%|▍         | 283/7505 [05:34<1:48:49,  1.11it/s]
  4%|▍         | 284/7505 [05:34<1:40:11,  1.20it/s]
  4%|▍         | 285/7505 [05:35<1:25:18,  1.41it/s]
  4%|▍         | 286/7505 [05:35<1:26:21,  1.39it/s]
  4%|▍         | 287/7505 [05:36<1:36:45,  1.24it/s]
  4%|▍         | 288/7505 [05:37<1:31:16,  1.32it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.94it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 16.88it/s]
+
 18%|█▊        | 7/39 [00:00<00:01, 16.12it/s]
+
 23%|██▎       | 9/39 [00:00<00:01, 15.19it/s]
+
 28%|██▊       | 11/39 [00:00<00:01, 14.27it/s]
+
 33%|███▎      | 13/39 [00:00<00:01, 14.56it/s]
+
 38%|███▊      | 15/39 [00:00<00:01, 14.47it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 14.59it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 14.01it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 14.34it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 13.44it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 10.57it/s]
+
 69%|██████▉   | 27/39 [00:02<00:01, 10.27it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 10.79it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00,  9.30it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00,  8.43it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00,  8.12it/s]
+
 90%|████████▉ | 35/39 [00:03<00:00,  7.88it/s]
+
 92%|█████████▏| 36/39 [00:03<00:00,  7.67it/s]
+
 95%|█████████▍| 37/39 [00:03<00:00,  7.48it/s]
+
 97%|█████████▋| 38/39 [00:03<00:00,  7.33it/s]
+
100%|██████████| 39/39 [00:03<00:00,  7.22it/s]
                                                    
+
                                               
{'eval_loss': '6.103', 'eval_runtime': '3.889', 'eval_samples_per_second': '627.7', 'eval_steps_per_second': '10.03', 'epoch': '15.16'}
+
  4%|▍         | 288/7505 [05:41<1:31:16,  1.32it/s]
+
100%|██████████| 39/39 [00:03<00:00,  7.22it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-288
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-288/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-288/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.98it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.98it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-288/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-288/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-288/tokenizer_config.json
+[eval] epoch 15.161 step 288 loss 6.1034 (best 6.1034 @ ep 15.161, 0/20 since best)
+
+
                                               
  4%|▍         | 289/7505 [05:44<5:00:20,  2.50s/it]
  4%|▍         | 290/7505 [05:45<4:07:40,  2.06s/it]
                                                    
{'loss': '6.877', 'grad_norm': '0.3197', 'learning_rate': '0.0006', 'epoch': '15.27'}
+
  4%|▍         | 290/7505 [05:45<4:07:40,  2.06s/it]
  4%|▍         | 291/7505 [05:45<3:17:59,  1.65s/it]
  4%|▍         | 292/7505 [05:46<2:42:13,  1.35s/it]
  4%|▍         | 293/7505 [05:47<2:18:12,  1.15s/it]
  4%|▍         | 294/7505 [05:47<2:00:23,  1.00s/it]
  4%|▍         | 295/7505 [05:48<1:47:56,  1.11it/s]
  4%|▍         | 296/7505 [05:49<1:39:08,  1.21it/s]
  4%|▍         | 297/7505 [05:49<1:33:34,  1.28it/s]
  4%|▍         | 298/7505 [05:50<1:29:20,  1.34it/s]
  4%|▍         | 299/7505 [05:51<1:26:11,  1.39it/s]
  4%|▍         | 300/7505 [05:51<1:23:59,  1.43it/s]
                                                    
{'loss': '6.782', 'grad_norm': '0.1718', 'learning_rate': '0.0006', 'epoch': '15.81'}
+
  4%|▍         | 300/7505 [05:51<1:23:59,  1.43it/s]
  4%|▍         | 301/7505 [05:52<1:23:30,  1.44it/s]
  4%|▍         | 302/7505 [05:53<1:22:06,  1.46it/s]
  4%|▍         | 303/7505 [05:54<1:34:23,  1.27it/s]
  4%|▍         | 304/7505 [05:54<1:30:03,  1.33it/s]
  4%|▍         | 305/7505 [05:56<1:47:44,  1.11it/s]
  4%|▍         | 306/7505 [05:56<1:39:54,  1.20it/s]
  4%|▍         | 307/7505 [05:57<1:49:52,  1.09it/s]
  4%|▍         | 308/7505 [05:58<1:40:31,  1.19it/s]
  4%|▍         | 309/7505 [05:59<1:47:42,  1.11it/s]
  4%|▍         | 310/7505 [06:00<1:58:30,  1.01it/s]
                                                    
{'loss': '6.783', 'grad_norm': '0.6407', 'learning_rate': '0.0006', 'epoch': '16.32'}
+
  4%|▍         | 310/7505 [06:00<1:58:30,  1.01it/s]
  4%|▍         | 311/7505 [06:01<2:05:27,  1.05s/it]
  4%|▍         | 312/7505 [06:03<2:10:07,  1.09s/it]
  4%|▍         | 313/7505 [06:04<2:13:30,  1.11s/it]
  4%|▍         | 314/7505 [06:05<2:12:04,  1.10s/it]
  4%|▍         | 315/7505 [06:06<2:11:09,  1.09s/it]
  4%|▍         | 316/7505 [06:07<2:11:45,  1.10s/it]
  4%|▍         | 317/7505 [06:08<1:56:43,  1.03it/s]
  4%|▍         | 318/7505 [06:09<1:59:02,  1.01it/s]
  4%|▍         | 319/7505 [06:09<1:46:49,  1.12it/s]
  4%|▍         | 320/7505 [06:11<1:55:43,  1.03it/s]
                                                    
{'loss': '6.723', 'grad_norm': '0.4704', 'learning_rate': '0.0006', 'epoch': '16.86'}
+
  4%|▍         | 320/7505 [06:11<1:55:43,  1.03it/s]
  4%|▍         | 321/7505 [06:12<1:56:52,  1.02it/s]
  4%|▍         | 322/7505 [06:13<1:58:57,  1.01it/s]
  4%|▍         | 323/7505 [06:13<1:52:01,  1.07it/s]
  4%|▍         | 324/7505 [06:14<1:51:09,  1.08it/s]
  4%|▍         | 325/7505 [06:15<1:59:06,  1.00it/s]
  4%|▍         | 326/7505 [06:17<2:05:06,  1.05s/it]
  4%|▍         | 327/7505 [06:18<2:05:24,  1.05s/it]
  4%|▍         | 328/7505 [06:19<2:06:53,  1.06s/it]
  4%|▍         | 329/7505 [06:19<1:51:43,  1.07it/s]
  4%|▍         | 330/7505 [06:20<1:54:24,  1.05it/s]
                                                    
{'loss': '6.722', 'grad_norm': '0.2888', 'learning_rate': '0.0006', 'epoch': '17.38'}
+
  4%|▍         | 330/7505 [06:20<1:54:24,  1.05it/s]
  4%|▍         | 331/7505 [06:21<1:46:16,  1.13it/s]
  4%|▍         | 332/7505 [06:22<1:49:39,  1.09it/s]
  4%|▍         | 333/7505 [06:23<1:56:55,  1.02it/s]
  4%|▍         | 334/7505 [06:24<1:59:44,  1.00s/it]
  4%|▍         | 335/7505 [06:25<1:46:40,  1.12it/s]
  4%|▍         | 336/7505 [06:26<1:53:48,  1.05it/s]
  4%|▍         | 337/7505 [06:27<1:42:34,  1.16it/s]
  5%|▍         | 338/7505 [06:28<1:53:38,  1.05it/s]
  5%|▍         | 339/7505 [06:29<1:52:14,  1.06it/s]
  5%|▍         | 340/7505 [06:30<1:50:41,  1.08it/s]
                                                    
{'loss': '6.618', 'grad_norm': '0.2507', 'learning_rate': '0.0006', 'epoch': '17.91'}
+
  5%|▍         | 340/7505 [06:30<1:50:41,  1.08it/s]
  5%|▍         | 341/7505 [06:30<1:40:30,  1.19it/s]
  5%|▍         | 342/7505 [06:31<1:25:03,  1.40it/s]
  5%|▍         | 343/7505 [06:31<1:25:58,  1.39it/s]
  5%|▍         | 344/7505 [06:32<1:23:15,  1.43it/s]
  5%|▍         | 345/7505 [06:33<1:21:57,  1.46it/s]
  5%|▍         | 346/7505 [06:33<1:20:26,  1.48it/s]
  5%|▍         | 347/7505 [06:34<1:19:22,  1.50it/s]
  5%|▍         | 348/7505 [06:35<1:18:37,  1.52it/s]
  5%|▍         | 349/7505 [06:35<1:18:26,  1.52it/s]
  5%|▍         | 350/7505 [06:36<1:17:57,  1.53it/s]
                                                    
{'loss': '6.565', 'grad_norm': '0.4476', 'learning_rate': '0.0006', 'epoch': '18.43'}
+
  5%|▍         | 350/7505 [06:36<1:17:57,  1.53it/s]
  5%|▍         | 351/7505 [06:37<1:28:47,  1.34it/s]
  5%|▍         | 352/7505 [06:38<1:43:22,  1.15it/s]
  5%|▍         | 353/7505 [06:39<1:36:43,  1.23it/s]
  5%|▍         | 354/7505 [06:40<1:47:58,  1.10it/s]
  5%|▍         | 355/7505 [06:41<1:53:18,  1.05it/s]
  5%|▍         | 356/7505 [06:42<1:42:03,  1.17it/s]
  5%|▍         | 357/7505 [06:43<1:47:12,  1.11it/s]
  5%|▍         | 358/7505 [06:43<1:40:38,  1.18it/s]
  5%|▍         | 359/7505 [06:44<1:38:49,  1.21it/s]
  5%|▍         | 360/7505 [06:45<1:42:06,  1.17it/s]
                                                    
{'loss': '6.546', 'grad_norm': '0.3702', 'learning_rate': '0.0006', 'epoch': '18.97'}
+
  5%|▍         | 360/7505 [06:45<1:42:06,  1.17it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 21.19it/s]
+
 15%|█▌        | 6/39 [00:00<00:02, 16.00it/s]
+
 21%|██        | 8/39 [00:00<00:02, 11.01it/s]
+
 26%|██▌       | 10/39 [00:00<00:03,  9.24it/s]
+
 31%|███       | 12/39 [00:01<00:03,  8.42it/s]
+
 36%|███▌      | 14/39 [00:01<00:02,  9.80it/s]
+
 41%|████      | 16/39 [00:01<00:02,  8.94it/s]
+
 44%|████▎     | 17/39 [00:01<00:02,  8.48it/s]
+
 46%|████▌     | 18/39 [00:01<00:02,  8.10it/s]
+
 49%|████▊     | 19/39 [00:02<00:02,  7.91it/s]
+
 51%|█████▏    | 20/39 [00:02<00:02,  7.92it/s]
+
 56%|█████▋    | 22/39 [00:02<00:01,  9.87it/s]
+
 62%|██████▏   | 24/39 [00:02<00:01,  8.79it/s]
+
 64%|██████▍   | 25/39 [00:02<00:01,  8.32it/s]
+
 67%|██████▋   | 26/39 [00:02<00:01,  7.94it/s]
+
 69%|██████▉   | 27/39 [00:02<00:01,  7.64it/s]
+
 72%|███████▏  | 28/39 [00:03<00:01,  8.07it/s]
+
 77%|███████▋  | 30/39 [00:03<00:00, 10.00it/s]
+
 82%|████████▏ | 32/39 [00:03<00:00,  8.56it/s]
+
 85%|████████▍ | 33/39 [00:03<00:00,  8.20it/s]
+
 87%|████████▋ | 34/39 [00:03<00:00,  7.95it/s]
+
 90%|████████▉ | 35/39 [00:03<00:00,  7.81it/s]
+
 92%|█████████▏| 36/39 [00:04<00:00,  8.02it/s]
+
 97%|█████████▋| 38/39 [00:04<00:00, 10.06it/s]
                                                    
+
                                               
{'eval_loss': '5.72', 'eval_runtime': '4.535', 'eval_samples_per_second': '538.2', 'eval_steps_per_second': '8.599', 'epoch': '18.97'}
+
  5%|▍         | 360/7505 [06:50<1:42:06,  1.17it/s]
+
100%|██████████| 39/39 [00:04<00:00, 10.06it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-360
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-360/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-360/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.86it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.86it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-360/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-360/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-360/tokenizer_config.json
+[eval] epoch 18.966 step 360 loss 5.7200 (best 5.7200 @ ep 18.966, 0/20 since best)
+
+
                                               
  5%|▍         | 361/7505 [06:52<5:29:42,  2.77s/it]
  5%|▍         | 362/7505 [06:53<4:27:40,  2.25s/it]
  5%|▍         | 363/7505 [06:54<3:30:04,  1.76s/it]
  5%|▍         | 364/7505 [06:55<3:06:26,  1.57s/it]
  5%|▍         | 365/7505 [06:56<2:33:08,  1.29s/it]
  5%|▍         | 366/7505 [06:56<2:10:18,  1.10s/it]
  5%|▍         | 367/7505 [06:57<1:54:17,  1.04it/s]
  5%|▍         | 368/7505 [06:58<1:42:44,  1.16it/s]
  5%|▍         | 369/7505 [06:58<1:34:39,  1.26it/s]
  5%|▍         | 370/7505 [06:59<1:29:21,  1.33it/s]
                                                    
{'loss': '6.516', 'grad_norm': '0.2599', 'learning_rate': '0.0006', 'epoch': '19.48'}
+
  5%|▍         | 370/7505 [06:59<1:29:21,  1.33it/s]
  5%|▍         | 371/7505 [07:00<1:25:18,  1.39it/s]
  5%|▍         | 372/7505 [07:00<1:22:26,  1.44it/s]
  5%|▍         | 373/7505 [07:01<1:20:46,  1.47it/s]
  5%|▍         | 374/7505 [07:01<1:19:16,  1.50it/s]
  5%|▍         | 375/7505 [07:02<1:18:12,  1.52it/s]
  5%|▌         | 376/7505 [07:03<1:17:28,  1.53it/s]
  5%|▌         | 377/7505 [07:03<1:17:09,  1.54it/s]
  5%|▌         | 378/7505 [07:04<1:16:43,  1.55it/s]
  5%|▌         | 379/7505 [07:05<1:16:22,  1.56it/s]
  5%|▌         | 380/7505 [07:05<1:07:51,  1.75it/s]
                                                    
{'loss': '6.487', 'grad_norm': '0.3894', 'learning_rate': '0.0006', 'epoch': '20'}
+
  5%|▌         | 380/7505 [07:05<1:07:51,  1.75it/s]
  5%|▌         | 381/7505 [07:06<1:13:48,  1.61it/s]
  5%|▌         | 382/7505 [07:06<1:14:22,  1.60it/s]
  5%|▌         | 383/7505 [07:07<1:14:57,  1.58it/s]
  5%|▌         | 384/7505 [07:08<1:15:12,  1.58it/s]
  5%|▌         | 385/7505 [07:08<1:15:20,  1.57it/s]
  5%|▌         | 386/7505 [07:09<1:15:26,  1.57it/s]
  5%|▌         | 387/7505 [07:10<1:15:49,  1.56it/s]
  5%|▌         | 388/7505 [07:10<1:15:44,  1.57it/s]
  5%|▌         | 389/7505 [07:11<1:15:42,  1.57it/s]
  5%|▌         | 390/7505 [07:12<1:15:42,  1.57it/s]
                                                    
{'loss': '6.419', 'grad_norm': '0.4107', 'learning_rate': '0.0006', 'epoch': '20.54'}
+
  5%|▌         | 390/7505 [07:12<1:15:42,  1.57it/s]
  5%|▌         | 391/7505 [07:12<1:15:46,  1.56it/s]
  5%|▌         | 392/7505 [07:13<1:15:49,  1.56it/s]
  5%|▌         | 393/7505 [07:13<1:15:47,  1.56it/s]
  5%|▌         | 394/7505 [07:14<1:15:44,  1.56it/s]
  5%|▌         | 395/7505 [07:15<1:15:53,  1.56it/s]
  5%|▌         | 396/7505 [07:15<1:15:47,  1.56it/s]
  5%|▌         | 397/7505 [07:16<1:15:46,  1.56it/s]
  5%|▌         | 398/7505 [07:17<1:15:35,  1.57it/s]
  5%|▌         | 399/7505 [07:17<1:07:05,  1.77it/s]
  5%|▌         | 400/7505 [07:18<1:13:01,  1.62it/s]
                                                    
{'loss': '6.417', 'grad_norm': '0.2627', 'learning_rate': '0.0006', 'epoch': '21.05'}
+
  5%|▌         | 400/7505 [07:18<1:13:01,  1.62it/s]
  5%|▌         | 401/7505 [07:18<1:13:55,  1.60it/s]
  5%|▌         | 402/7505 [07:19<1:14:27,  1.59it/s]
  5%|▌         | 403/7505 [07:20<1:14:44,  1.58it/s]
  5%|▌         | 404/7505 [07:20<1:14:58,  1.58it/s]
  5%|▌         | 405/7505 [07:21<1:15:08,  1.57it/s]
  5%|▌         | 406/7505 [07:22<1:15:13,  1.57it/s]
  5%|▌         | 407/7505 [07:22<1:15:24,  1.57it/s]
  5%|▌         | 408/7505 [07:23<1:15:21,  1.57it/s]
  5%|▌         | 409/7505 [07:24<1:15:20,  1.57it/s]
  5%|▌         | 410/7505 [07:24<1:15:24,  1.57it/s]
                                                    
{'loss': '6.366', 'grad_norm': '0.2557', 'learning_rate': '0.0006', 'epoch': '21.59'}
+
  5%|▌         | 410/7505 [07:24<1:15:24,  1.57it/s]
  5%|▌         | 411/7505 [07:25<1:15:24,  1.57it/s]
  5%|▌         | 412/7505 [07:25<1:15:30,  1.57it/s]
  6%|▌         | 413/7505 [07:26<1:15:33,  1.56it/s]
  6%|▌         | 414/7505 [07:27<1:15:28,  1.57it/s]
  6%|▌         | 415/7505 [07:27<1:15:24,  1.57it/s]
  6%|▌         | 416/7505 [07:28<1:15:26,  1.57it/s]
  6%|▌         | 417/7505 [07:29<1:15:23,  1.57it/s]
  6%|▌         | 418/7505 [07:29<1:06:56,  1.76it/s]
  6%|▌         | 419/7505 [07:30<1:12:48,  1.62it/s]
  6%|▌         | 420/7505 [07:30<1:13:33,  1.61it/s]
                                                    
{'loss': '6.334', 'grad_norm': '0.4655', 'learning_rate': '0.0006', 'epoch': '22.11'}
+
  6%|▌         | 420/7505 [07:30<1:13:33,  1.61it/s]
  6%|▌         | 421/7505 [07:31<1:14:03,  1.59it/s]
  6%|▌         | 422/7505 [07:32<1:14:22,  1.59it/s]
  6%|▌         | 423/7505 [07:32<1:14:35,  1.58it/s]
  6%|▌         | 424/7505 [07:33<1:14:44,  1.58it/s]
  6%|▌         | 425/7505 [07:34<1:14:51,  1.58it/s]
  6%|▌         | 426/7505 [07:34<1:14:52,  1.58it/s]
  6%|▌         | 427/7505 [07:35<1:14:56,  1.57it/s]
  6%|▌         | 428/7505 [07:36<1:14:59,  1.57it/s]
  6%|▌         | 429/7505 [07:36<1:15:05,  1.57it/s]
  6%|▌         | 430/7505 [07:37<1:15:05,  1.57it/s]
                                                    
{'loss': '6.279', 'grad_norm': '0.4943', 'learning_rate': '0.0006', 'epoch': '22.64'}
+
  6%|▌         | 430/7505 [07:37<1:15:05,  1.57it/s]
  6%|▌         | 431/7505 [07:37<1:15:06,  1.57it/s]
  6%|▌         | 432/7505 [07:38<1:15:04,  1.57it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 22.50it/s]
+
 15%|█▌        | 6/39 [00:00<00:01, 17.54it/s]
+
 21%|██        | 8/39 [00:00<00:01, 16.58it/s]
+
 26%|██▌       | 10/39 [00:00<00:01, 16.09it/s]
+
 31%|███       | 12/39 [00:00<00:01, 15.75it/s]
+
 36%|███▌      | 14/39 [00:00<00:01, 15.56it/s]
+
 41%|████      | 16/39 [00:00<00:01, 15.42it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 15.35it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 15.28it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 15.26it/s]
+
 62%|██████▏   | 24/39 [00:01<00:00, 15.22it/s]
+
 67%|██████▋   | 26/39 [00:01<00:00, 15.22it/s]
+
 72%|███████▏  | 28/39 [00:01<00:00, 15.21it/s]
+
 77%|███████▋  | 30/39 [00:01<00:00, 15.20it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 15.18it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 15.18it/s]
+
 92%|█████████▏| 36/39 [00:02<00:00, 15.19it/s]
+
 97%|█████████▋| 38/39 [00:02<00:00, 15.19it/s]
                                                    
+
                                               
{'eval_loss': '5.414', 'eval_runtime': '2.643', 'eval_samples_per_second': '923.8', 'eval_steps_per_second': '14.76', 'epoch': '22.75'}
+
  6%|▌         | 432/7505 [07:41<1:15:04,  1.57it/s]
+
100%|██████████| 39/39 [00:02<00:00, 15.19it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-432
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-432/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-432/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.73it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.72it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-432/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-432/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-432/tokenizer_config.json
+[eval] epoch 22.752 step 432 loss 5.4145 (best 5.4145 @ ep 22.752, 0/20 since best)
+
+
                                               
  6%|▌         | 433/7505 [07:43<4:01:55,  2.05s/it]
  6%|▌         | 434/7505 [07:44<3:11:44,  1.63s/it]
  6%|▌         | 435/7505 [07:45<2:36:37,  1.33s/it]
  6%|▌         | 436/7505 [07:45<2:11:59,  1.12s/it]
  6%|▌         | 437/7505 [07:46<1:46:29,  1.11it/s]
  6%|▌         | 438/7505 [07:46<1:40:23,  1.17it/s]
  6%|▌         | 439/7505 [07:47<1:32:44,  1.27it/s]
  6%|▌         | 440/7505 [07:48<1:27:23,  1.35it/s]
                                                    
{'loss': '6.289', 'grad_norm': '0.3954', 'learning_rate': '0.0006', 'epoch': '23.16'}
+
  6%|▌         | 440/7505 [07:48<1:27:23,  1.35it/s]
  6%|▌         | 441/7505 [07:48<1:23:38,  1.41it/s]
  6%|▌         | 442/7505 [07:49<1:20:58,  1.45it/s]
  6%|▌         | 443/7505 [07:50<1:19:06,  1.49it/s]
  6%|▌         | 444/7505 [07:50<1:17:50,  1.51it/s]
  6%|▌         | 445/7505 [07:51<1:16:55,  1.53it/s]
  6%|▌         | 446/7505 [07:52<1:16:16,  1.54it/s]
  6%|▌         | 447/7505 [07:52<1:15:49,  1.55it/s]
  6%|▌         | 448/7505 [07:53<1:15:30,  1.56it/s]
  6%|▌         | 449/7505 [07:53<1:15:17,  1.56it/s]
  6%|▌         | 450/7505 [07:54<1:15:07,  1.57it/s]
                                                    
{'loss': '6.219', 'grad_norm': '0.2119', 'learning_rate': '0.0006', 'epoch': '23.7'}
+
  6%|▌         | 450/7505 [07:54<1:15:07,  1.57it/s]
  6%|▌         | 451/7505 [07:55<1:15:00,  1.57it/s]
  6%|▌         | 452/7505 [07:55<1:14:54,  1.57it/s]
  6%|▌         | 453/7505 [07:56<1:14:51,  1.57it/s]
  6%|▌         | 454/7505 [07:57<1:14:48,  1.57it/s]
  6%|▌         | 455/7505 [07:57<1:14:41,  1.57it/s]
  6%|▌         | 456/7505 [07:58<1:06:21,  1.77it/s]
  6%|▌         | 457/7505 [07:58<1:12:12,  1.63it/s]
  6%|▌         | 458/7505 [07:59<1:12:53,  1.61it/s]
  6%|▌         | 459/7505 [08:00<1:13:21,  1.60it/s]
  6%|▌         | 460/7505 [08:00<1:13:41,  1.59it/s]
                                                    
{'loss': '6.235', 'grad_norm': '0.2168', 'learning_rate': '0.0006', 'epoch': '24.21'}
+
  6%|▌         | 460/7505 [08:00<1:13:41,  1.59it/s]
  6%|▌         | 461/7505 [08:01<1:14:01,  1.59it/s]
  6%|▌         | 462/7505 [08:02<1:14:08,  1.58it/s]
  6%|▌         | 463/7505 [08:02<1:14:13,  1.58it/s]
  6%|▌         | 464/7505 [08:03<1:14:15,  1.58it/s]
  6%|▌         | 465/7505 [08:03<1:14:18,  1.58it/s]
  6%|▌         | 466/7505 [08:04<1:14:21,  1.58it/s]
  6%|▌         | 467/7505 [08:05<1:14:22,  1.58it/s]
  6%|▌         | 468/7505 [08:05<1:14:23,  1.58it/s]
  6%|▌         | 469/7505 [08:06<1:14:23,  1.58it/s]
  6%|▋         | 470/7505 [08:07<1:14:26,  1.58it/s]
                                                    
{'loss': '6.13', 'grad_norm': '0.4165', 'learning_rate': '0.0006', 'epoch': '24.75'}
+
  6%|▋         | 470/7505 [08:07<1:14:26,  1.58it/s]
  6%|▋         | 471/7505 [08:07<1:14:29,  1.57it/s]
  6%|▋         | 472/7505 [08:08<1:14:32,  1.57it/s]
  6%|▋         | 473/7505 [08:09<1:14:32,  1.57it/s]
  6%|▋         | 474/7505 [08:09<1:14:26,  1.57it/s]
  6%|▋         | 475/7505 [08:10<1:06:08,  1.77it/s]
  6%|▋         | 476/7505 [08:10<1:12:02,  1.63it/s]
  6%|▋         | 477/7505 [08:11<1:12:46,  1.61it/s]
  6%|▋         | 478/7505 [08:12<1:13:16,  1.60it/s]
  6%|▋         | 479/7505 [08:12<1:13:32,  1.59it/s]
  6%|▋         | 480/7505 [08:13<1:13:45,  1.59it/s]
                                                    
{'loss': '6.2', 'grad_norm': '0.5049', 'learning_rate': '0.0006', 'epoch': '25.27'}
+
  6%|▋         | 480/7505 [08:13<1:13:45,  1.59it/s]
  6%|▋         | 481/7505 [08:14<1:13:55,  1.58it/s]
  6%|▋         | 482/7505 [08:14<1:14:00,  1.58it/s]
  6%|▋         | 483/7505 [08:15<1:14:05,  1.58it/s]
  6%|▋         | 484/7505 [08:15<1:14:05,  1.58it/s]
  6%|▋         | 485/7505 [08:16<1:14:08,  1.58it/s]
  6%|▋         | 486/7505 [08:17<1:14:08,  1.58it/s]
  6%|▋         | 487/7505 [08:17<1:14:08,  1.58it/s]
  7%|▋         | 488/7505 [08:18<1:14:07,  1.58it/s]
  7%|▋         | 489/7505 [08:19<1:14:08,  1.58it/s]
  7%|▋         | 490/7505 [08:19<1:14:07,  1.58it/s]
                                                    
{'loss': '6.174', 'grad_norm': '0.4682', 'learning_rate': '0.0006', 'epoch': '25.81'}
+
  7%|▋         | 490/7505 [08:19<1:14:07,  1.58it/s]
  7%|▋         | 491/7505 [08:20<1:14:09,  1.58it/s]
  7%|▋         | 492/7505 [08:20<1:14:08,  1.58it/s]
  7%|▋         | 493/7505 [08:21<1:14:08,  1.58it/s]
  7%|▋         | 494/7505 [08:22<1:05:52,  1.77it/s]
  7%|▋         | 495/7505 [08:22<1:11:48,  1.63it/s]
  7%|▋         | 496/7505 [08:23<1:12:29,  1.61it/s]
  7%|▋         | 497/7505 [08:24<1:12:59,  1.60it/s]
  7%|▋         | 498/7505 [08:24<1:13:18,  1.59it/s]
  7%|▋         | 499/7505 [08:25<1:13:31,  1.59it/s]
  7%|▋         | 500/7505 [08:25<1:13:42,  1.58it/s]
                                                    
{'loss': '6.11', 'grad_norm': '0.1985', 'learning_rate': '0.0006', 'epoch': '26.32'}
+
  7%|▋         | 500/7505 [08:25<1:13:42,  1.58it/s]
  7%|▋         | 501/7505 [08:26<1:13:49,  1.58it/s]
  7%|▋         | 502/7505 [08:27<1:13:51,  1.58it/s]
  7%|▋         | 503/7505 [08:27<1:13:55,  1.58it/s]
  7%|▋         | 504/7505 [08:28<1:13:55,  1.58it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 22.93it/s]
+
 15%|█▌        | 6/39 [00:00<00:01, 17.60it/s]
+
 21%|██        | 8/39 [00:00<00:01, 16.65it/s]
+
 26%|██▌       | 10/39 [00:00<00:01, 16.11it/s]
+
 31%|███       | 12/39 [00:00<00:01, 15.79it/s]
+
 36%|███▌      | 14/39 [00:00<00:01, 15.56it/s]
+
 41%|████      | 16/39 [00:00<00:01, 15.45it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 15.34it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 15.30it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 15.22it/s]
+
 62%|██████▏   | 24/39 [00:01<00:00, 15.23it/s]
+
 67%|██████▋   | 26/39 [00:01<00:00, 15.20it/s]
+
 72%|███████▏  | 28/39 [00:01<00:00, 15.20it/s]
+
 77%|███████▋  | 30/39 [00:01<00:00, 15.19it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 15.19it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 15.18it/s]
+
 92%|█████████▏| 36/39 [00:02<00:00, 15.19it/s]
+
 97%|█████████▋| 38/39 [00:02<00:00, 15.18it/s]
                                                    
+
                                               
{'eval_loss': '5.139', 'eval_runtime': '2.643', 'eval_samples_per_second': '923.4', 'eval_steps_per_second': '14.75', 'epoch': '26.54'}
+
  7%|▋         | 504/7505 [08:31<1:13:55,  1.58it/s]
+
100%|██████████| 39/39 [00:02<00:00, 15.18it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-504
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-504/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-504/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.86it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.86it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-504/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-504/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-504/tokenizer_config.json
+[eval] epoch 26.537 step 504 loss 5.1385 (best 5.1385 @ ep 26.537, 0/20 since best)
+
+
                                               
  7%|▋         | 505/7505 [08:33<3:57:37,  2.04s/it]
  7%|▋         | 506/7505 [08:34<3:08:29,  1.62s/it]
  7%|▋         | 507/7505 [08:35<2:34:09,  1.32s/it]
  7%|▋         | 508/7505 [08:35<2:10:04,  1.12s/it]
  7%|▋         | 509/7505 [08:36<1:53:12,  1.03it/s]
  7%|▋         | 510/7505 [08:36<1:41:24,  1.15it/s]
                                                    
{'loss': '6.095', 'grad_norm': '0.1955', 'learning_rate': '0.0006', 'epoch': '26.86'}
+
  7%|▋         | 510/7505 [08:36<1:41:24,  1.15it/s]
  7%|▋         | 511/7505 [08:37<1:33:11,  1.25it/s]
  7%|▋         | 512/7505 [08:38<1:27:20,  1.33it/s]
  7%|▋         | 513/7505 [08:38<1:15:04,  1.55it/s]
  7%|▋         | 514/7505 [08:39<1:18:05,  1.49it/s]
  7%|▋         | 515/7505 [08:39<1:16:49,  1.52it/s]
  7%|▋         | 516/7505 [08:40<1:15:59,  1.53it/s]
  7%|▋         | 517/7505 [08:41<1:15:21,  1.55it/s]
  7%|▋         | 518/7505 [08:41<1:14:55,  1.55it/s]
  7%|▋         | 519/7505 [08:42<1:14:39,  1.56it/s]
  7%|▋         | 520/7505 [08:43<1:14:26,  1.56it/s]
                                                    
{'loss': '5.996', 'grad_norm': '0.1996', 'learning_rate': '0.0006', 'epoch': '27.38'}
+
  7%|▋         | 520/7505 [08:43<1:14:26,  1.56it/s]
  7%|▋         | 521/7505 [08:43<1:14:17,  1.57it/s]
  7%|▋         | 522/7505 [08:44<1:14:10,  1.57it/s]
  7%|▋         | 523/7505 [08:45<1:14:06,  1.57it/s]
  7%|▋         | 524/7505 [08:45<1:14:02,  1.57it/s]
  7%|▋         | 525/7505 [08:46<1:14:01,  1.57it/s]
  7%|▋         | 526/7505 [08:46<1:13:58,  1.57it/s]
  7%|▋         | 527/7505 [08:47<1:13:56,  1.57it/s]
  7%|▋         | 528/7505 [08:48<1:13:54,  1.57it/s]
  7%|▋         | 529/7505 [08:48<1:13:54,  1.57it/s]
  7%|▋         | 530/7505 [08:49<1:13:54,  1.57it/s]
                                                    
{'loss': '6.003', 'grad_norm': '0.2457', 'learning_rate': '0.0006', 'epoch': '27.91'}
+
  7%|▋         | 530/7505 [08:49<1:13:54,  1.57it/s]
  7%|▋         | 531/7505 [08:50<1:13:51,  1.57it/s]
  7%|▋         | 532/7505 [08:50<1:05:37,  1.77it/s]
  7%|▋         | 533/7505 [08:51<1:11:35,  1.62it/s]
  7%|▋         | 534/7505 [08:51<1:12:15,  1.61it/s]
  7%|▋         | 535/7505 [08:52<1:12:44,  1.60it/s]
  7%|▋         | 536/7505 [08:53<1:13:04,  1.59it/s]
  7%|▋         | 537/7505 [08:53<1:13:19,  1.58it/s]
  7%|▋         | 538/7505 [08:54<1:13:26,  1.58it/s]
  7%|▋         | 539/7505 [08:55<1:13:32,  1.58it/s]
  7%|▋         | 540/7505 [08:55<1:13:35,  1.58it/s]
                                                    
{'loss': '5.935', 'grad_norm': '0.5316', 'learning_rate': '0.0006', 'epoch': '28.43'}
+
  7%|▋         | 540/7505 [08:55<1:13:35,  1.58it/s]
  7%|▋         | 541/7505 [08:56<1:13:40,  1.58it/s]
  7%|▋         | 542/7505 [08:57<1:13:41,  1.57it/s]
  7%|▋         | 543/7505 [08:57<1:13:41,  1.57it/s]
  7%|▋         | 544/7505 [08:58<1:13:41,  1.57it/s]
  7%|▋         | 545/7505 [08:58<1:13:42,  1.57it/s]
  7%|▋         | 546/7505 [08:59<1:13:41,  1.57it/s]
  7%|▋         | 547/7505 [09:00<1:13:41,  1.57it/s]
  7%|▋         | 548/7505 [09:00<1:13:40,  1.57it/s]
  7%|▋         | 549/7505 [09:01<1:13:41,  1.57it/s]
  7%|▋         | 550/7505 [09:02<1:13:37,  1.57it/s]
                                                    
{'loss': '5.962', 'grad_norm': '0.3737', 'learning_rate': '0.0006', 'epoch': '28.97'}
+
  7%|▋         | 550/7505 [09:02<1:13:37,  1.57it/s]
  7%|▋         | 551/7505 [09:02<1:05:26,  1.77it/s]
  7%|▋         | 552/7505 [09:03<1:11:14,  1.63it/s]
  7%|▋         | 553/7505 [09:03<1:11:57,  1.61it/s]
  7%|▋         | 554/7505 [09:04<1:12:26,  1.60it/s]
  7%|▋         | 555/7505 [09:05<1:12:46,  1.59it/s]
  7%|▋         | 556/7505 [09:05<1:13:03,  1.59it/s]
  7%|▋         | 557/7505 [09:06<1:13:11,  1.58it/s]
  7%|▋         | 558/7505 [09:07<1:13:18,  1.58it/s]
  7%|▋         | 559/7505 [09:07<1:13:21,  1.58it/s]
  7%|▋         | 560/7505 [09:08<1:13:24,  1.58it/s]
                                                    
{'loss': '5.929', 'grad_norm': '0.3176', 'learning_rate': '0.0006', 'epoch': '29.48'}
+
  7%|▋         | 560/7505 [09:08<1:13:24,  1.58it/s]
  7%|▋         | 561/7505 [09:08<1:13:26,  1.58it/s]
  7%|▋         | 562/7505 [09:09<1:13:26,  1.58it/s]
  8%|▊         | 563/7505 [09:10<1:13:25,  1.58it/s]
  8%|▊         | 564/7505 [09:10<1:13:26,  1.58it/s]
  8%|▊         | 565/7505 [09:11<1:13:25,  1.58it/s]
  8%|▊         | 566/7505 [09:12<1:13:24,  1.58it/s]
  8%|▊         | 567/7505 [09:12<1:13:22,  1.58it/s]
  8%|▊         | 568/7505 [09:13<1:13:26,  1.57it/s]
  8%|▊         | 569/7505 [09:14<1:13:22,  1.58it/s]
  8%|▊         | 570/7505 [09:14<1:05:10,  1.77it/s]
                                                    
{'loss': '5.846', 'grad_norm': '0.2098', 'learning_rate': '0.0006', 'epoch': '30'}
+
  8%|▊         | 570/7505 [09:14<1:05:10,  1.77it/s]
  8%|▊         | 571/7505 [09:15<1:11:01,  1.63it/s]
  8%|▊         | 572/7505 [09:15<1:11:45,  1.61it/s]
  8%|▊         | 573/7505 [09:16<1:12:13,  1.60it/s]
  8%|▊         | 574/7505 [09:17<1:12:31,  1.59it/s]
  8%|▊         | 575/7505 [09:17<1:12:44,  1.59it/s]
  8%|▊         | 576/7505 [09:18<1:12:54,  1.58it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 23.39it/s]
+
 15%|█▌        | 6/39 [00:00<00:01, 17.75it/s]
+
 21%|██        | 8/39 [00:00<00:01, 16.70it/s]
+
 26%|██▌       | 10/39 [00:00<00:01, 16.15it/s]
+
 31%|███       | 12/39 [00:00<00:01, 15.76it/s]
+
 36%|███▌      | 14/39 [00:00<00:01, 15.61it/s]
+
 41%|████      | 16/39 [00:00<00:01, 15.45it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 15.37it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 15.29it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 15.27it/s]
+
 62%|██████▏   | 24/39 [00:01<00:00, 15.24it/s]
+
 67%|██████▋   | 26/39 [00:01<00:00, 15.22it/s]
+
 72%|███████▏  | 28/39 [00:01<00:00, 15.21it/s]
+
 77%|███████▋  | 30/39 [00:01<00:00, 15.21it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 15.20it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 15.19it/s]
+
 92%|█████████▏| 36/39 [00:02<00:00, 15.19it/s]
+
 97%|█████████▋| 38/39 [00:02<00:00, 15.19it/s]
                                                    
+
                                               
{'eval_loss': '4.891', 'eval_runtime': '2.642', 'eval_samples_per_second': '923.8', 'eval_steps_per_second': '14.76', 'epoch': '30.32'}
+
  8%|▊         | 576/7505 [09:20<1:12:54,  1.58it/s]
+
100%|██████████| 39/39 [00:02<00:00, 15.19it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-576
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-576/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-576/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.97it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.97it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-576/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-576/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-576/tokenizer_config.json
+[eval] epoch 30.322 step 576 loss 4.8914 (best 4.8914 @ ep 30.322, 0/20 since best)
+
+
                                               
  8%|▊         | 577/7505 [09:23<3:54:04,  2.03s/it]
  8%|▊         | 578/7505 [09:24<3:05:46,  1.61s/it]
  8%|▊         | 579/7505 [09:24<2:31:58,  1.32s/it]
  8%|▊         | 580/7505 [09:25<2:08:21,  1.11s/it]
                                                    
{'loss': '5.809', 'grad_norm': '0.3842', 'learning_rate': '0.0006', 'epoch': '30.54'}
+
  8%|▊         | 580/7505 [09:25<2:08:21,  1.11s/it]
  8%|▊         | 581/7505 [09:26<1:51:49,  1.03it/s]
  8%|▊         | 582/7505 [09:26<1:40:15,  1.15it/s]
  8%|▊         | 583/7505 [09:27<1:32:07,  1.25it/s]
  8%|▊         | 584/7505 [09:28<1:26:25,  1.33it/s]
  8%|▊         | 585/7505 [09:28<1:22:26,  1.40it/s]
  8%|▊         | 586/7505 [09:29<1:19:39,  1.45it/s]
  8%|▊         | 587/7505 [09:29<1:17:44,  1.48it/s]
  8%|▊         | 588/7505 [09:30<1:16:19,  1.51it/s]
  8%|▊         | 589/7505 [09:30<1:07:14,  1.71it/s]
  8%|▊         | 590/7505 [09:31<1:11:43,  1.61it/s]
                                                    
{'loss': '5.817', 'grad_norm': '0.2734', 'learning_rate': '0.0006', 'epoch': '31.05'}
+
  8%|▊         | 590/7505 [09:31<1:11:43,  1.61it/s]
  8%|▊         | 591/7505 [09:32<1:12:09,  1.60it/s]
  8%|▊         | 592/7505 [09:32<1:12:25,  1.59it/s]
  8%|▊         | 593/7505 [09:33<1:12:37,  1.59it/s]
  8%|▊         | 594/7505 [09:34<1:12:45,  1.58it/s]
  8%|▊         | 595/7505 [09:34<1:12:49,  1.58it/s]
  8%|▊         | 596/7505 [09:35<1:12:52,  1.58it/s]
  8%|▊         | 597/7505 [09:36<1:12:55,  1.58it/s]
  8%|▊         | 598/7505 [09:36<1:12:57,  1.58it/s]
  8%|▊         | 599/7505 [09:37<1:12:58,  1.58it/s]
  8%|▊         | 600/7505 [09:38<1:13:02,  1.58it/s]
                                                    
{'loss': '5.81', 'grad_norm': '0.4863', 'learning_rate': '0.0006', 'epoch': '31.59'}
+
  8%|▊         | 600/7505 [09:38<1:13:02,  1.58it/s]
  8%|▊         | 601/7505 [09:38<1:13:02,  1.58it/s]
  8%|▊         | 602/7505 [09:39<1:13:03,  1.57it/s]
  8%|▊         | 603/7505 [09:39<1:13:01,  1.58it/s]
  8%|▊         | 604/7505 [09:40<1:13:01,  1.58it/s]
  8%|▊         | 605/7505 [09:41<1:13:01,  1.57it/s]
  8%|▊         | 606/7505 [09:41<1:12:59,  1.58it/s]
  8%|▊         | 607/7505 [09:42<1:12:55,  1.58it/s]
  8%|▊         | 608/7505 [09:42<1:04:49,  1.77it/s]
  8%|▊         | 609/7505 [09:43<1:10:37,  1.63it/s]
  8%|▊         | 610/7505 [09:44<1:11:19,  1.61it/s]
                                                    
{'loss': '5.801', 'grad_norm': '0.379', 'learning_rate': '0.0006', 'epoch': '32.11'}
+
  8%|▊         | 610/7505 [09:44<1:11:19,  1.61it/s]
  8%|▊         | 611/7505 [09:44<1:11:49,  1.60it/s]
  8%|▊         | 612/7505 [09:45<1:12:10,  1.59it/s]
  8%|▊         | 613/7505 [09:46<1:12:24,  1.59it/s]
  8%|▊         | 614/7505 [09:46<1:12:34,  1.58it/s]
  8%|▊         | 615/7505 [09:47<1:12:39,  1.58it/s]
  8%|▊         | 616/7505 [09:48<1:12:42,  1.58it/s]
  8%|▊         | 617/7505 [09:48<1:12:46,  1.58it/s]
  8%|▊         | 618/7505 [09:49<1:12:47,  1.58it/s]
  8%|▊         | 619/7505 [09:49<1:12:49,  1.58it/s]
  8%|▊         | 620/7505 [09:50<1:12:51,  1.58it/s]
                                                    
{'loss': '5.731', 'grad_norm': '0.2852', 'learning_rate': '0.0006', 'epoch': '32.64'}
+
  8%|▊         | 620/7505 [09:50<1:12:51,  1.58it/s]
  8%|▊         | 621/7505 [09:51<1:12:51,  1.57it/s]
  8%|▊         | 622/7505 [09:51<1:12:50,  1.57it/s]
  8%|▊         | 623/7505 [09:52<1:12:50,  1.57it/s]
  8%|▊         | 624/7505 [09:53<1:12:48,  1.58it/s]
  8%|▊         | 625/7505 [09:53<1:12:49,  1.57it/s]
  8%|▊         | 626/7505 [09:54<1:12:46,  1.58it/s]
  8%|▊         | 627/7505 [09:54<1:04:39,  1.77it/s]
  8%|▊         | 628/7505 [09:55<1:10:27,  1.63it/s]
  8%|▊         | 629/7505 [09:56<1:11:09,  1.61it/s]
  8%|▊         | 630/7505 [09:56<1:11:37,  1.60it/s]
                                                    
{'loss': '5.757', 'grad_norm': '0.2132', 'learning_rate': '0.0006', 'epoch': '33.16'}
+
  8%|▊         | 630/7505 [09:56<1:11:37,  1.60it/s]
  8%|▊         | 631/7505 [09:57<1:11:57,  1.59it/s]
  8%|▊         | 632/7505 [09:58<1:12:11,  1.59it/s]
  8%|▊         | 633/7505 [09:58<1:12:20,  1.58it/s]
  8%|▊         | 634/7505 [09:59<1:12:27,  1.58it/s]
  8%|▊         | 635/7505 [09:59<1:12:33,  1.58it/s]
  8%|▊         | 636/7505 [10:00<1:12:34,  1.58it/s]
  8%|▊         | 637/7505 [10:01<1:12:37,  1.58it/s]
  9%|▊         | 638/7505 [10:01<1:12:42,  1.57it/s]
  9%|▊         | 639/7505 [10:02<1:12:40,  1.57it/s]
  9%|▊         | 640/7505 [10:03<1:12:38,  1.57it/s]
                                                    
{'loss': '5.733', 'grad_norm': '0.2484', 'learning_rate': '0.0006', 'epoch': '33.7'}
+
  9%|▊         | 640/7505 [10:03<1:12:38,  1.57it/s]
  9%|▊         | 641/7505 [10:03<1:12:38,  1.58it/s]
  9%|▊         | 642/7505 [10:04<1:12:38,  1.57it/s]
  9%|▊         | 643/7505 [10:05<1:12:36,  1.58it/s]
  9%|▊         | 644/7505 [10:05<1:12:36,  1.57it/s]
  9%|▊         | 645/7505 [10:06<1:12:35,  1.58it/s]
  9%|▊         | 646/7505 [10:06<1:04:30,  1.77it/s]
  9%|▊         | 647/7505 [10:07<1:10:16,  1.63it/s]
  9%|▊         | 648/7505 [10:08<1:10:56,  1.61it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 22.80it/s]
+
 15%|█▌        | 6/39 [00:00<00:01, 17.58it/s]
+
 21%|██        | 8/39 [00:00<00:01, 16.62it/s]
+
 26%|██▌       | 10/39 [00:00<00:01, 16.10it/s]
+
 31%|███       | 12/39 [00:00<00:01, 15.77it/s]
+
 36%|███▌      | 14/39 [00:00<00:01, 15.55it/s]
+
 41%|████      | 16/39 [00:00<00:01, 15.42it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 15.34it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 15.28it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 15.25it/s]
+
 62%|██████▏   | 24/39 [00:01<00:00, 15.25it/s]
+
 67%|██████▋   | 26/39 [00:01<00:00, 15.23it/s]
+
 72%|███████▏  | 28/39 [00:01<00:00, 15.21it/s]
+
 77%|███████▋  | 30/39 [00:01<00:00, 15.20it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 15.20it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 15.19it/s]
+
 92%|█████████▏| 36/39 [00:02<00:00, 15.18it/s]
+
 97%|█████████▋| 38/39 [00:02<00:00, 15.19it/s]
                                                    
+
                                               
{'eval_loss': '4.715', 'eval_runtime': '2.643', 'eval_samples_per_second': '923.4', 'eval_steps_per_second': '14.75', 'epoch': '34.11'}
+
  9%|▊         | 648/7505 [10:10<1:10:56,  1.61it/s]
+
100%|██████████| 39/39 [00:02<00:00, 15.19it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-648
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-648/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-648/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.98it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.97it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-648/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-648/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-648/tokenizer_config.json
+[eval] epoch 34.107 step 648 loss 4.7155 (best 4.7155 @ ep 34.107, 0/20 since best)
+
+
                                               
  9%|▊         | 649/7505 [10:13<3:50:16,  2.02s/it]
  9%|▊         | 650/7505 [10:14<3:02:54,  1.60s/it]
                                                    
{'loss': '5.676', 'grad_norm': '0.27', 'learning_rate': '0.0006', 'epoch': '34.21'}
+
  9%|▊         | 650/7505 [10:14<3:02:54,  1.60s/it]
  9%|▊         | 651/7505 [10:14<2:29:47,  1.31s/it]
  9%|▊         | 652/7505 [10:15<2:06:33,  1.11s/it]
  9%|▊         | 653/7505 [10:15<1:50:20,  1.03it/s]
  9%|▊         | 654/7505 [10:16<1:38:57,  1.15it/s]
  9%|▊         | 655/7505 [10:17<1:31:00,  1.25it/s]
  9%|▊         | 656/7505 [10:17<1:25:25,  1.34it/s]
  9%|▉         | 657/7505 [10:18<1:21:41,  1.40it/s]
  9%|▉         | 658/7505 [10:19<1:18:57,  1.45it/s]
  9%|▉         | 659/7505 [10:19<1:16:58,  1.48it/s]
  9%|▉         | 660/7505 [10:20<1:15:37,  1.51it/s]
                                                    
{'loss': '5.673', 'grad_norm': '0.3352', 'learning_rate': '0.0006', 'epoch': '34.75'}
+
  9%|▉         | 660/7505 [10:20<1:15:37,  1.51it/s]
  9%|▉         | 661/7505 [10:21<1:14:41,  1.53it/s]
  9%|▉         | 662/7505 [10:21<1:14:00,  1.54it/s]
  9%|▉         | 663/7505 [10:22<1:13:34,  1.55it/s]
  9%|▉         | 664/7505 [10:22<1:13:10,  1.56it/s]
  9%|▉         | 665/7505 [10:23<1:04:52,  1.76it/s]
  9%|▉         | 666/7505 [10:24<1:10:20,  1.62it/s]
  9%|▉         | 667/7505 [10:24<1:10:54,  1.61it/s]
  9%|▉         | 668/7505 [10:25<1:11:19,  1.60it/s]
  9%|▉         | 669/7505 [10:25<1:11:36,  1.59it/s]
  9%|▉         | 670/7505 [10:26<1:11:52,  1.59it/s]
                                                    
{'loss': '5.597', 'grad_norm': '0.4181', 'learning_rate': '0.0006', 'epoch': '35.27'}
+
  9%|▉         | 670/7505 [10:26<1:11:52,  1.59it/s]
  9%|▉         | 671/7505 [10:27<1:12:00,  1.58it/s]
  9%|▉         | 672/7505 [10:27<1:12:05,  1.58it/s]
  9%|▉         | 673/7505 [10:28<1:12:09,  1.58it/s]
  9%|▉         | 674/7505 [10:29<1:12:13,  1.58it/s]
  9%|▉         | 675/7505 [10:29<1:12:14,  1.58it/s]
  9%|▉         | 676/7505 [10:30<1:12:13,  1.58it/s]
  9%|▉         | 677/7505 [10:31<1:12:12,  1.58it/s]
  9%|▉         | 678/7505 [10:31<1:12:12,  1.58it/s]
  9%|▉         | 679/7505 [10:32<1:12:10,  1.58it/s]
  9%|▉         | 680/7505 [10:32<1:12:09,  1.58it/s]
                                                    
{'loss': '5.661', 'grad_norm': '0.2992', 'learning_rate': '0.0006', 'epoch': '35.81'}
+
  9%|▉         | 680/7505 [10:32<1:12:09,  1.58it/s]
  9%|▉         | 681/7505 [10:33<1:12:10,  1.58it/s]
  9%|▉         | 682/7505 [10:34<1:12:09,  1.58it/s]
  9%|▉         | 683/7505 [10:34<1:12:06,  1.58it/s]
  9%|▉         | 684/7505 [10:35<1:04:05,  1.77it/s]
  9%|▉         | 685/7505 [10:35<1:09:21,  1.64it/s]
  9%|▉         | 686/7505 [10:36<1:10:15,  1.62it/s]
  9%|▉         | 687/7505 [10:37<1:10:52,  1.60it/s]
  9%|▉         | 688/7505 [10:37<1:11:13,  1.60it/s]
  9%|▉         | 689/7505 [10:38<1:11:28,  1.59it/s]
  9%|▉         | 690/7505 [10:39<1:11:41,  1.58it/s]
                                                    
{'loss': '5.575', 'grad_norm': '0.2951', 'learning_rate': '0.0006', 'epoch': '36.32'}
+
  9%|▉         | 690/7505 [10:39<1:11:41,  1.58it/s]
  9%|▉         | 691/7505 [10:39<1:11:51,  1.58it/s]
  9%|▉         | 692/7505 [10:40<1:11:56,  1.58it/s]
  9%|▉         | 693/7505 [10:41<1:11:57,  1.58it/s]
  9%|▉         | 694/7505 [10:41<1:12:01,  1.58it/s]
  9%|▉         | 695/7505 [10:42<1:12:02,  1.58it/s]
  9%|▉         | 696/7505 [10:42<1:12:04,  1.57it/s]
  9%|▉         | 697/7505 [10:43<1:12:04,  1.57it/s]
  9%|▉         | 698/7505 [10:44<1:12:05,  1.57it/s]
  9%|▉         | 699/7505 [10:44<1:12:06,  1.57it/s]
  9%|▉         | 700/7505 [10:45<1:12:05,  1.57it/s]
                                                    
{'loss': '5.628', 'grad_norm': '0.3907', 'learning_rate': '0.0006', 'epoch': '36.86'}
+
  9%|▉         | 700/7505 [10:45<1:12:05,  1.57it/s]
  9%|▉         | 701/7505 [10:46<1:12:04,  1.57it/s]
  9%|▉         | 702/7505 [10:46<1:12:04,  1.57it/s]
  9%|▉         | 703/7505 [10:47<1:04:01,  1.77it/s]
  9%|▉         | 704/7505 [10:47<1:09:16,  1.64it/s]
  9%|▉         | 705/7505 [10:48<1:10:05,  1.62it/s]
  9%|▉         | 706/7505 [10:49<1:10:38,  1.60it/s]
  9%|▉         | 707/7505 [10:49<1:11:01,  1.60it/s]
  9%|▉         | 708/7505 [10:50<1:11:17,  1.59it/s]
  9%|▉         | 709/7505 [10:51<1:11:27,  1.59it/s]
  9%|▉         | 710/7505 [10:51<1:11:34,  1.58it/s]
                                                    
{'loss': '5.576', 'grad_norm': '0.3281', 'learning_rate': '0.0006', 'epoch': '37.38'}
+
  9%|▉         | 710/7505 [10:51<1:11:34,  1.58it/s]
  9%|▉         | 711/7505 [10:52<1:11:44,  1.58it/s]
  9%|▉         | 712/7505 [10:52<1:11:46,  1.58it/s]
 10%|▉         | 713/7505 [10:53<1:11:46,  1.58it/s]
 10%|▉         | 714/7505 [10:54<1:11:49,  1.58it/s]
 10%|▉         | 715/7505 [10:54<1:11:51,  1.57it/s]
 10%|▉         | 716/7505 [10:55<1:11:49,  1.58it/s]
 10%|▉         | 717/7505 [10:56<1:11:49,  1.58it/s]
 10%|▉         | 718/7505 [10:56<1:11:49,  1.57it/s]
 10%|▉         | 719/7505 [10:57<1:11:47,  1.58it/s]
 10%|▉         | 720/7505 [10:58<1:11:49,  1.57it/s]
                                                    
{'loss': '5.559', 'grad_norm': '0.2437', 'learning_rate': '0.0006', 'epoch': '37.91'}
+
 10%|▉         | 720/7505 [10:58<1:11:49,  1.57it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 23.21it/s]
+
 15%|█▌        | 6/39 [00:00<00:01, 17.41it/s]
+
 21%|██        | 8/39 [00:00<00:01, 16.73it/s]
+
 26%|██▌       | 10/39 [00:00<00:01, 16.16it/s]
+
 31%|███       | 12/39 [00:00<00:01, 15.80it/s]
+
 36%|███▌      | 14/39 [00:00<00:01, 15.55it/s]
+
 41%|████      | 16/39 [00:00<00:01, 15.44it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 15.35it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 15.29it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 15.24it/s]
+
 62%|██████▏   | 24/39 [00:01<00:00, 15.24it/s]
+
 67%|██████▋   | 26/39 [00:01<00:00, 15.22it/s]
+
 72%|███████▏  | 28/39 [00:01<00:00, 15.20it/s]
+
 77%|███████▋  | 30/39 [00:01<00:00, 15.20it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 15.20it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 15.19it/s]
+
 92%|█████████▏| 36/39 [00:02<00:00, 15.19it/s]
+
 97%|█████████▋| 38/39 [00:02<00:00, 15.18it/s]
                                                    
+
                                               
{'eval_loss': '4.6', 'eval_runtime': '2.629', 'eval_samples_per_second': '928.4', 'eval_steps_per_second': '14.83', 'epoch': '37.91'}
+
 10%|▉         | 720/7505 [11:00<1:11:49,  1.57it/s]
+
100%|██████████| 39/39 [00:02<00:00, 15.18it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-720
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-720/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-720/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.97it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.96it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-720/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-720/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-720/tokenizer_config.json
+[eval] epoch 37.913 step 720 loss 4.6004 (best 4.6004 @ ep 37.913, 0/20 since best)
+
+
                                               
 10%|▉         | 721/7505 [11:03<3:49:02,  2.03s/it]
 10%|▉         | 722/7505 [11:03<2:53:51,  1.54s/it]
 10%|▉         | 723/7505 [11:04<2:26:29,  1.30s/it]
 10%|▉         | 724/7505 [11:05<2:04:05,  1.10s/it]
 10%|▉         | 725/7505 [11:05<1:48:22,  1.04it/s]
 10%|▉         | 726/7505 [11:06<1:37:24,  1.16it/s]
 10%|▉         | 727/7505 [11:06<1:29:41,  1.26it/s]
 10%|▉         | 728/7505 [11:07<1:24:16,  1.34it/s]
 10%|▉         | 729/7505 [11:08<1:20:31,  1.40it/s]
 10%|▉         | 730/7505 [11:08<1:17:53,  1.45it/s]
                                                    
{'loss': '5.529', 'grad_norm': '0.1931', 'learning_rate': '0.0006', 'epoch': '38.43'}
+
 10%|▉         | 730/7505 [11:08<1:17:53,  1.45it/s]
 10%|▉         | 731/7505 [11:09<1:16:07,  1.48it/s]
 10%|▉         | 732/7505 [11:10<1:14:49,  1.51it/s]
 10%|▉         | 733/7505 [11:10<1:13:51,  1.53it/s]
 10%|▉         | 734/7505 [11:11<1:13:14,  1.54it/s]
 10%|▉         | 735/7505 [11:12<1:12:47,  1.55it/s]
 10%|▉         | 736/7505 [11:12<1:12:28,  1.56it/s]
 10%|▉         | 737/7505 [11:13<1:12:16,  1.56it/s]
 10%|▉         | 738/7505 [11:13<1:12:03,  1.57it/s]
 10%|▉         | 739/7505 [11:14<1:11:54,  1.57it/s]
 10%|▉         | 740/7505 [11:15<1:11:49,  1.57it/s]
                                                    
{'loss': '5.509', 'grad_norm': '0.2264', 'learning_rate': '0.0006', 'epoch': '38.97'}
+
 10%|▉         | 740/7505 [11:15<1:11:49,  1.57it/s]
 10%|▉         | 741/7505 [11:15<1:03:48,  1.77it/s]
 10%|▉         | 742/7505 [11:16<1:09:28,  1.62it/s]
 10%|▉         | 743/7505 [11:17<1:10:07,  1.61it/s]
 10%|▉         | 744/7505 [11:17<1:10:34,  1.60it/s]
 10%|▉         | 745/7505 [11:18<1:10:51,  1.59it/s]
 10%|▉         | 746/7505 [11:18<1:11:03,  1.59it/s]
 10%|▉         | 747/7505 [11:19<1:11:10,  1.58it/s]
 10%|▉         | 748/7505 [11:20<1:11:14,  1.58it/s]
 10%|▉         | 749/7505 [11:20<1:11:19,  1.58it/s]
 10%|▉         | 750/7505 [11:21<1:11:24,  1.58it/s]
                                                    
{'loss': '5.457', 'grad_norm': '0.2309', 'learning_rate': '0.0006', 'epoch': '39.48'}
+
 10%|▉         | 750/7505 [11:21<1:11:24,  1.58it/s]
 10%|█         | 751/7505 [11:22<1:11:28,  1.58it/s]
 10%|█         | 752/7505 [11:22<1:11:27,  1.57it/s]
 10%|█         | 753/7505 [11:23<1:11:27,  1.57it/s]
 10%|█         | 754/7505 [11:24<1:11:26,  1.57it/s]
 10%|█         | 755/7505 [11:24<1:11:26,  1.57it/s]
 10%|█         | 756/7505 [11:25<1:11:28,  1.57it/s]
 10%|█         | 757/7505 [11:25<1:11:26,  1.57it/s]
 10%|█         | 758/7505 [11:26<1:11:28,  1.57it/s]
 10%|█         | 759/7505 [11:27<1:11:25,  1.57it/s]
 10%|█         | 760/7505 [11:27<1:03:29,  1.77it/s]
                                                    
{'loss': '5.506', 'grad_norm': '0.2982', 'learning_rate': '0.0006', 'epoch': '40'}
+
 10%|█         | 760/7505 [11:27<1:03:29,  1.77it/s]
 10%|█         | 761/7505 [11:28<1:08:28,  1.64it/s]
 10%|█         | 762/7505 [11:28<1:09:19,  1.62it/s]
 10%|█         | 763/7505 [11:29<1:09:55,  1.61it/s]
 10%|█         | 764/7505 [11:30<1:10:22,  1.60it/s]
 10%|█         | 765/7505 [11:30<1:10:39,  1.59it/s]
 10%|█         | 766/7505 [11:31<1:10:50,  1.59it/s]
 10%|█         | 767/7505 [11:32<1:11:00,  1.58it/s]
 10%|█         | 768/7505 [11:32<1:11:05,  1.58it/s]
 10%|█         | 769/7505 [11:33<1:11:08,  1.58it/s]
 10%|█         | 770/7505 [11:34<1:11:10,  1.58it/s]
                                                    
{'loss': '5.422', 'grad_norm': '0.4166', 'learning_rate': '0.0006', 'epoch': '40.54'}
+
 10%|█         | 770/7505 [11:34<1:11:10,  1.58it/s]
 10%|█         | 771/7505 [11:34<1:11:12,  1.58it/s]
 10%|█         | 772/7505 [11:35<1:11:12,  1.58it/s]
 10%|█         | 773/7505 [11:35<1:11:11,  1.58it/s]
 10%|█         | 774/7505 [11:36<1:11:12,  1.58it/s]
 10%|█         | 775/7505 [11:37<1:11:13,  1.57it/s]
 10%|█         | 776/7505 [11:37<1:11:13,  1.57it/s]
 10%|█         | 777/7505 [11:38<1:11:16,  1.57it/s]
 10%|█         | 778/7505 [11:39<1:11:13,  1.57it/s]
 10%|█         | 779/7505 [11:39<1:03:18,  1.77it/s]
 10%|█         | 780/7505 [11:40<1:08:29,  1.64it/s]
                                                    
{'loss': '5.502', 'grad_norm': '0.5481', 'learning_rate': '0.0006', 'epoch': '41.05'}
+
 10%|█         | 780/7505 [11:40<1:08:29,  1.64it/s]
 10%|█         | 781/7505 [11:40<1:09:18,  1.62it/s]
 10%|█         | 782/7505 [11:41<1:09:50,  1.60it/s]
 10%|█         | 783/7505 [11:42<1:10:19,  1.59it/s]
 10%|█         | 784/7505 [11:42<1:10:33,  1.59it/s]
 10%|█         | 785/7505 [11:43<1:10:48,  1.58it/s]
 10%|█         | 786/7505 [11:44<1:10:56,  1.58it/s]
 10%|█         | 787/7505 [11:44<1:11:03,  1.58it/s]
 10%|█         | 788/7505 [11:45<1:11:09,  1.57it/s]
 11%|█         | 789/7505 [11:45<1:11:09,  1.57it/s]
 11%|█         | 790/7505 [11:46<1:11:10,  1.57it/s]
                                                    
{'loss': '5.525', 'grad_norm': '0.4403', 'learning_rate': '0.0006', 'epoch': '41.59'}
+
 11%|█         | 790/7505 [11:46<1:11:10,  1.57it/s]
 11%|█         | 791/7505 [11:47<1:11:13,  1.57it/s]
 11%|█         | 792/7505 [11:47<1:11:12,  1.57it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 22.56it/s]
+
 15%|█▌        | 6/39 [00:00<00:01, 17.50it/s]
+
 21%|██        | 8/39 [00:00<00:01, 16.55it/s]
+
 26%|██▌       | 10/39 [00:00<00:01, 16.04it/s]
+
 31%|███       | 12/39 [00:00<00:01, 15.70it/s]
+
 36%|███▌      | 14/39 [00:00<00:01, 15.51it/s]
+
 41%|████      | 16/39 [00:00<00:01, 15.39it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 15.32it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 15.26it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 15.23it/s]
+
 62%|██████▏   | 24/39 [00:01<00:00, 15.22it/s]
+
 67%|██████▋   | 26/39 [00:01<00:00, 15.20it/s]
+
 72%|███████▏  | 28/39 [00:01<00:00, 15.19it/s]
+
 77%|███████▋  | 30/39 [00:01<00:00, 15.19it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 15.18it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 15.18it/s]
+
 92%|█████████▏| 36/39 [00:02<00:00, 15.17it/s]
+
 97%|█████████▋| 38/39 [00:02<00:00, 15.17it/s]
                                                    
+
                                               
{'eval_loss': '4.558', 'eval_runtime': '2.647', 'eval_samples_per_second': '922.4', 'eval_steps_per_second': '14.74', 'epoch': '41.7'}
+
 11%|█         | 792/7505 [11:50<1:11:12,  1.57it/s]
+
100%|██████████| 39/39 [00:02<00:00, 15.17it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-792
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-792/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-792/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.97it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.97it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-792/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-792/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-792/tokenizer_config.json
+[eval] epoch 41.698 step 792 loss 4.5580 (best 4.5580 @ ep 41.698, 0/20 since best)
+
+
                                               
 11%|█         | 793/7505 [11:53<3:47:31,  2.03s/it]
 11%|█         | 794/7505 [11:53<3:00:32,  1.61s/it]
 11%|█         | 795/7505 [11:54<2:27:39,  1.32s/it]
 11%|█         | 796/7505 [11:55<2:04:38,  1.11s/it]
 11%|█         | 797/7505 [11:55<1:48:30,  1.03it/s]
 11%|█         | 798/7505 [11:56<1:29:21,  1.25it/s]
 11%|█         | 799/7505 [11:56<1:26:41,  1.29it/s]
 11%|█         | 800/7505 [11:57<1:22:18,  1.36it/s]
                                                    
{'loss': '5.469', 'grad_norm': '0.2105', 'learning_rate': '0.0006', 'epoch': '42.11'}
+
 11%|█         | 800/7505 [11:57<1:22:18,  1.36it/s]
 11%|█         | 801/7505 [11:58<1:19:00,  1.41it/s]
 11%|█         | 802/7505 [11:58<1:16:36,  1.46it/s]
 11%|█         | 803/7505 [11:59<1:14:56,  1.49it/s]
 11%|█         | 804/7505 [11:59<1:13:42,  1.52it/s]
 11%|█         | 805/7505 [12:00<1:12:51,  1.53it/s]
 11%|█         | 806/7505 [12:01<1:12:15,  1.55it/s]
 11%|█         | 807/7505 [12:01<1:11:52,  1.55it/s]
 11%|█         | 808/7505 [12:02<1:11:34,  1.56it/s]
 11%|█         | 809/7505 [12:03<1:11:21,  1.56it/s]
 11%|█         | 810/7505 [12:03<1:11:11,  1.57it/s]
                                                    
{'loss': '5.374', 'grad_norm': '0.2072', 'learning_rate': '0.0006', 'epoch': '42.64'}
+
 11%|█         | 810/7505 [12:03<1:11:11,  1.57it/s]
 11%|█         | 811/7505 [12:04<1:11:09,  1.57it/s]
 11%|█         | 812/7505 [12:05<1:11:06,  1.57it/s]
 11%|█         | 813/7505 [12:05<1:11:01,  1.57it/s]
 11%|█         | 814/7505 [12:06<1:10:56,  1.57it/s]
 11%|█         | 815/7505 [12:06<1:10:54,  1.57it/s]
 11%|█         | 816/7505 [12:07<1:10:50,  1.57it/s]
 11%|█         | 817/7505 [12:08<1:02:58,  1.77it/s]
 11%|█         | 818/7505 [12:08<1:08:00,  1.64it/s]
 11%|█         | 819/7505 [12:09<1:08:49,  1.62it/s]
 11%|█         | 820/7505 [12:10<1:09:25,  1.61it/s]
                                                    
{'loss': '5.378', 'grad_norm': '0.2122', 'learning_rate': '0.0006', 'epoch': '43.16'}
+
 11%|█         | 820/7505 [12:10<1:09:25,  1.61it/s]
 11%|█         | 821/7505 [12:10<1:09:53,  1.59it/s]
 11%|█         | 822/7505 [12:11<1:10:09,  1.59it/s]
 11%|█         | 823/7505 [12:11<1:10:21,  1.58it/s]
 11%|█         | 824/7505 [12:12<1:10:26,  1.58it/s]
 11%|█         | 825/7505 [12:13<1:10:31,  1.58it/s]
 11%|█         | 826/7505 [12:13<1:10:35,  1.58it/s]
 11%|█         | 827/7505 [12:14<1:10:35,  1.58it/s]
 11%|█         | 828/7505 [12:15<1:10:38,  1.58it/s]
 11%|█         | 829/7505 [12:15<1:10:39,  1.57it/s]
 11%|█         | 830/7505 [12:16<1:10:40,  1.57it/s]
                                                    
{'loss': '5.346', 'grad_norm': '0.2113', 'learning_rate': '0.0006', 'epoch': '43.7'}
+
 11%|█         | 830/7505 [12:16<1:10:40,  1.57it/s]
 11%|█         | 831/7505 [12:16<1:10:42,  1.57it/s]
 11%|█         | 832/7505 [12:17<1:10:40,  1.57it/s]
 11%|█         | 833/7505 [12:18<1:10:37,  1.57it/s]
 11%|█         | 834/7505 [12:18<1:10:38,  1.57it/s]
 11%|█         | 835/7505 [12:19<1:10:36,  1.57it/s]
 11%|█         | 836/7505 [12:19<1:02:45,  1.77it/s]
 11%|█         | 837/7505 [12:20<1:07:49,  1.64it/s]
 11%|█         | 838/7505 [12:21<1:08:36,  1.62it/s]
 11%|█         | 839/7505 [12:21<1:09:10,  1.61it/s]
 11%|█         | 840/7505 [12:22<1:09:36,  1.60it/s]
                                                    
{'loss': '5.361', 'grad_norm': '0.2186', 'learning_rate': '0.0006', 'epoch': '44.21'}
+
 11%|█         | 840/7505 [12:22<1:09:36,  1.60it/s]
 11%|█         | 841/7505 [12:23<1:09:56,  1.59it/s]
 11%|█         | 842/7505 [12:23<1:10:05,  1.58it/s]
 11%|█         | 843/7505 [12:24<1:10:13,  1.58it/s]
 11%|█         | 844/7505 [12:25<1:10:18,  1.58it/s]
 11%|█▏        | 845/7505 [12:25<1:10:22,  1.58it/s]
 11%|█▏        | 846/7505 [12:26<1:10:25,  1.58it/s]
 11%|█▏        | 847/7505 [12:27<1:10:26,  1.58it/s]
 11%|█▏        | 848/7505 [12:27<1:10:25,  1.58it/s]
 11%|█▏        | 849/7505 [12:28<1:10:25,  1.58it/s]
 11%|█▏        | 850/7505 [12:28<1:10:25,  1.58it/s]
                                                    
{'loss': '5.338', 'grad_norm': '0.275', 'learning_rate': '0.0006', 'epoch': '44.75'}
+
 11%|█▏        | 850/7505 [12:28<1:10:25,  1.58it/s]
 11%|█▏        | 851/7505 [12:29<1:10:25,  1.57it/s]
 11%|█▏        | 852/7505 [12:30<1:10:23,  1.58it/s]
 11%|█▏        | 853/7505 [12:30<1:10:25,  1.57it/s]
 11%|█▏        | 854/7505 [12:31<1:10:22,  1.58it/s]
 11%|█▏        | 855/7505 [12:31<1:02:32,  1.77it/s]
 11%|█▏        | 856/7505 [12:32<1:07:37,  1.64it/s]
 11%|█▏        | 857/7505 [12:33<1:08:26,  1.62it/s]
 11%|█▏        | 858/7505 [12:33<1:09:01,  1.60it/s]
 11%|█▏        | 859/7505 [12:34<1:09:23,  1.60it/s]
 11%|█▏        | 860/7505 [12:35<1:09:40,  1.59it/s]
                                                    
{'loss': '5.349', 'grad_norm': '0.2544', 'learning_rate': '0.0006', 'epoch': '45.27'}
+
 11%|█▏        | 860/7505 [12:35<1:09:40,  1.59it/s]
 11%|█▏        | 861/7505 [12:35<1:09:51,  1.58it/s]
 11%|█▏        | 862/7505 [12:36<1:09:59,  1.58it/s]
 11%|█▏        | 863/7505 [12:37<1:10:04,  1.58it/s]
 12%|█▏        | 864/7505 [12:37<1:10:08,  1.58it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 22.77it/s]
+
 15%|█▌        | 6/39 [00:00<00:01, 17.62it/s]
+
 21%|██        | 8/39 [00:00<00:01, 16.64it/s]
+
 26%|██▌       | 10/39 [00:00<00:01, 16.13it/s]
+
 31%|███       | 12/39 [00:00<00:01, 15.79it/s]
+
 36%|███▌      | 14/39 [00:00<00:01, 15.59it/s]
+
 41%|████      | 16/39 [00:00<00:01, 15.43it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 15.35it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 15.30it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 15.26it/s]
+
 62%|██████▏   | 24/39 [00:01<00:00, 15.24it/s]
+
 67%|██████▋   | 26/39 [00:01<00:00, 15.24it/s]
+
 72%|███████▏  | 28/39 [00:01<00:00, 15.17it/s]
+
 77%|███████▋  | 30/39 [00:01<00:00, 15.22it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 15.19it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 15.17it/s]
+
 92%|█████████▏| 36/39 [00:02<00:00, 15.16it/s]
+
 97%|█████████▋| 38/39 [00:02<00:00, 15.18it/s]
                                                    
+
                                               
{'eval_loss': '4.457', 'eval_runtime': '2.631', 'eval_samples_per_second': '927.6', 'eval_steps_per_second': '14.82', 'epoch': '45.48'}
+
 12%|█▏        | 864/7505 [12:40<1:10:08,  1.58it/s]
+
100%|██████████| 39/39 [00:02<00:00, 15.18it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-864
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-864/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-864/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.73it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.73it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-864/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-864/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-864/tokenizer_config.json
+[eval] epoch 45.483 step 864 loss 4.4569 (best 4.4569 @ ep 45.483, 0/20 since best)
+
+
                                               
 12%|█▏        | 865/7505 [12:42<3:46:18,  2.05s/it]
 12%|█▏        | 866/7505 [12:43<2:59:27,  1.62s/it]
 12%|█▏        | 867/7505 [12:44<2:26:41,  1.33s/it]
 12%|█▏        | 868/7505 [12:44<2:03:44,  1.12s/it]
 12%|█▏        | 869/7505 [12:45<1:47:42,  1.03it/s]
 12%|█▏        | 870/7505 [12:46<1:36:25,  1.15it/s]
                                                    
{'loss': '5.307', 'grad_norm': '0.307', 'learning_rate': '0.0006', 'epoch': '45.81'}
+
 12%|█▏        | 870/7505 [12:46<1:36:25,  1.15it/s]
 12%|█▏        | 871/7505 [12:46<1:28:36,  1.25it/s]
 12%|█▏        | 872/7505 [12:47<1:23:05,  1.33it/s]
 12%|█▏        | 873/7505 [12:48<1:19:12,  1.40it/s]
 12%|█▏        | 874/7505 [12:48<1:08:43,  1.61it/s]
 12%|█▏        | 875/7505 [12:49<1:11:52,  1.54it/s]
 12%|█▏        | 876/7505 [12:49<1:11:26,  1.55it/s]
 12%|█▏        | 877/7505 [12:50<1:11:07,  1.55it/s]
 12%|█▏        | 878/7505 [12:51<1:10:55,  1.56it/s]
 12%|█▏        | 879/7505 [12:51<1:10:44,  1.56it/s]
 12%|█▏        | 880/7505 [12:52<1:10:33,  1.56it/s]
                                                    
{'loss': '5.312', 'grad_norm': '0.3227', 'learning_rate': '0.0006', 'epoch': '46.32'}
+
 12%|█▏        | 880/7505 [12:52<1:10:33,  1.56it/s]
 12%|█▏        | 881/7505 [12:53<1:10:29,  1.57it/s]
 12%|█▏        | 882/7505 [12:53<1:10:21,  1.57it/s]
 12%|█▏        | 883/7505 [12:54<1:10:16,  1.57it/s]
 12%|█▏        | 884/7505 [12:54<1:10:15,  1.57it/s]
 12%|█▏        | 885/7505 [12:55<1:10:12,  1.57it/s]
 12%|█▏        | 886/7505 [12:56<1:10:11,  1.57it/s]
 12%|█▏        | 887/7505 [12:56<1:10:12,  1.57it/s]
 12%|█▏        | 888/7505 [12:57<1:10:12,  1.57it/s]
 12%|█▏        | 889/7505 [12:58<1:10:08,  1.57it/s]
 12%|█▏        | 890/7505 [12:58<1:10:06,  1.57it/s]
                                                    
{'loss': '5.35', 'grad_norm': '0.3143', 'learning_rate': '0.0006', 'epoch': '46.86'}
+
 12%|█▏        | 890/7505 [12:58<1:10:06,  1.57it/s]
 12%|█▏        | 891/7505 [12:59<1:10:06,  1.57it/s]
 12%|█▏        | 892/7505 [13:00<1:10:04,  1.57it/s]
 12%|█▏        | 893/7505 [13:00<1:02:18,  1.77it/s]
 12%|█▏        | 894/7505 [13:01<1:07:18,  1.64it/s]
 12%|█▏        | 895/7505 [13:01<1:08:07,  1.62it/s]
 12%|█▏        | 896/7505 [13:02<1:08:39,  1.60it/s]
 12%|█▏        | 897/7505 [13:03<1:09:03,  1.59it/s]
 12%|█▏        | 898/7505 [13:03<1:09:21,  1.59it/s]
 12%|█▏        | 899/7505 [13:04<1:09:32,  1.58it/s]
 12%|█▏        | 900/7505 [13:04<1:09:43,  1.58it/s]
                                                    
{'loss': '5.342', 'grad_norm': '0.3222', 'learning_rate': '0.0006', 'epoch': '47.38'}
+
 12%|█▏        | 900/7505 [13:04<1:09:43,  1.58it/s]
 12%|█▏        | 901/7505 [13:05<1:09:53,  1.57it/s]
 12%|█▏        | 902/7505 [13:06<1:09:58,  1.57it/s]
 12%|█▏        | 903/7505 [13:06<1:09:58,  1.57it/s]
 12%|█▏        | 904/7505 [13:07<1:09:58,  1.57it/s]
 12%|█▏        | 905/7505 [13:08<1:09:59,  1.57it/s]
 12%|█▏        | 906/7505 [13:08<1:10:01,  1.57it/s]
 12%|█▏        | 907/7505 [13:09<1:09:59,  1.57it/s]
 12%|█▏        | 908/7505 [13:10<1:09:54,  1.57it/s]
 12%|█▏        | 909/7505 [13:10<1:09:53,  1.57it/s]
 12%|█▏        | 910/7505 [13:11<1:09:53,  1.57it/s]
                                                    
{'loss': '5.268', 'grad_norm': '0.241', 'learning_rate': '0.0006', 'epoch': '47.91'}
+
 12%|█▏        | 910/7505 [13:11<1:09:53,  1.57it/s]
 12%|█▏        | 911/7505 [13:11<1:09:49,  1.57it/s]
 12%|█▏        | 912/7505 [13:12<1:02:04,  1.77it/s]
 12%|█▏        | 913/7505 [13:13<1:06:38,  1.65it/s]
 12%|█▏        | 914/7505 [13:13<1:07:35,  1.63it/s]
 12%|█▏        | 915/7505 [13:14<1:08:16,  1.61it/s]
 12%|█▏        | 916/7505 [13:14<1:08:42,  1.60it/s]
 12%|█▏        | 917/7505 [13:15<1:09:00,  1.59it/s]
 12%|█▏        | 918/7505 [13:16<1:09:15,  1.59it/s]
 12%|█▏        | 919/7505 [13:16<1:09:25,  1.58it/s]
 12%|█▏        | 920/7505 [13:17<1:09:31,  1.58it/s]
                                                    
{'loss': '5.261', 'grad_norm': '0.2317', 'learning_rate': '0.0006', 'epoch': '48.43'}
+
 12%|█▏        | 920/7505 [13:17<1:09:31,  1.58it/s]
 12%|█▏        | 921/7505 [13:18<1:09:33,  1.58it/s]
 12%|█▏        | 922/7505 [13:18<1:09:35,  1.58it/s]
 12%|█▏        | 923/7505 [13:19<1:09:35,  1.58it/s]
 12%|█▏        | 924/7505 [13:20<1:09:36,  1.58it/s]
 12%|█▏        | 925/7505 [13:20<1:09:36,  1.58it/s]
 12%|█▏        | 926/7505 [13:21<1:09:38,  1.57it/s]
 12%|█▏        | 927/7505 [13:21<1:09:37,  1.57it/s]
 12%|█▏        | 928/7505 [13:22<1:09:36,  1.57it/s]
 12%|█▏        | 929/7505 [13:23<1:09:35,  1.57it/s]
 12%|█▏        | 930/7505 [13:23<1:09:33,  1.58it/s]
                                                    
{'loss': '5.241', 'grad_norm': '0.2244', 'learning_rate': '0.0006', 'epoch': '48.97'}
+
 12%|█▏        | 930/7505 [13:23<1:09:33,  1.58it/s]
 12%|█▏        | 931/7505 [13:24<1:01:52,  1.77it/s]
 12%|█▏        | 932/7505 [13:24<1:06:23,  1.65it/s]
 12%|█▏        | 933/7505 [13:25<1:07:19,  1.63it/s]
 12%|█▏        | 934/7505 [13:26<1:08:00,  1.61it/s]
 12%|█▏        | 935/7505 [13:26<1:08:27,  1.60it/s]
 12%|█▏        | 936/7505 [13:27<1:08:48,  1.59it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 23.18it/s]
+
 15%|█▌        | 6/39 [00:00<00:01, 17.63it/s]
+
 21%|██        | 8/39 [00:00<00:01, 16.64it/s]
+
 26%|██▌       | 10/39 [00:00<00:01, 16.09it/s]
+
 31%|███       | 12/39 [00:00<00:01, 15.77it/s]
+
 36%|███▌      | 14/39 [00:00<00:01, 15.54it/s]
+
 41%|████      | 16/39 [00:00<00:01, 15.41it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 15.30it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 15.25it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 15.20it/s]
+
 62%|██████▏   | 24/39 [00:01<00:00, 15.23it/s]
+
 67%|██████▋   | 26/39 [00:01<00:00, 15.21it/s]
+
 72%|███████▏  | 28/39 [00:01<00:00, 15.21it/s]
+
 77%|███████▋  | 30/39 [00:01<00:00, 15.20it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 15.20it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 15.19it/s]
+
 92%|█████████▏| 36/39 [00:02<00:00, 15.20it/s]
+
 97%|█████████▋| 38/39 [00:02<00:00, 15.19it/s]
                                                    
+
                                               
{'eval_loss': '4.424', 'eval_runtime': '2.632', 'eval_samples_per_second': '927.2', 'eval_steps_per_second': '14.81', 'epoch': '49.27'}
+
 12%|█▏        | 936/7505 [13:30<1:08:48,  1.59it/s]
+
100%|██████████| 39/39 [00:02<00:00, 15.19it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-936
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-936/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-936/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.97it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.97it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-936/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-936/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-936/tokenizer_config.json
+[eval] epoch 49.268 step 936 loss 4.4237 (best 4.4237 @ ep 49.268, 0/20 since best)
+
+
                                               
 12%|█▏        | 937/7505 [13:32<3:41:05,  2.02s/it]
 12%|█▏        | 938/7505 [13:33<2:55:32,  1.60s/it]
 13%|█▎        | 939/7505 [13:34<2:23:41,  1.31s/it]
 13%|█▎        | 940/7505 [13:34<2:01:25,  1.11s/it]
                                                    
{'loss': '5.258', 'grad_norm': '0.3075', 'learning_rate': '0.0006', 'epoch': '49.48'}
+
 13%|█▎        | 940/7505 [13:34<2:01:25,  1.11s/it]
 13%|█▎        | 941/7505 [13:35<1:45:48,  1.03it/s]
 13%|█▎        | 942/7505 [13:35<1:34:53,  1.15it/s]
 13%|█▎        | 943/7505 [13:36<1:27:18,  1.25it/s]
 13%|█▎        | 944/7505 [13:37<1:21:56,  1.33it/s]
 13%|█▎        | 945/7505 [13:37<1:18:10,  1.40it/s]
 13%|█▎        | 946/7505 [13:38<1:15:32,  1.45it/s]
 13%|█▎        | 947/7505 [13:39<1:13:41,  1.48it/s]
 13%|█▎        | 948/7505 [13:39<1:12:25,  1.51it/s]
 13%|█▎        | 949/7505 [13:40<1:11:28,  1.53it/s]
 13%|█▎        | 950/7505 [13:40<1:03:07,  1.73it/s]
                                                    
{'loss': '5.21', 'grad_norm': '0.2868', 'learning_rate': '0.0006', 'epoch': '50'}
+
 13%|█▎        | 950/7505 [13:40<1:03:07,  1.73it/s]
 13%|█▎        | 951/7505 [13:41<1:08:21,  1.60it/s]
 13%|█▎        | 952/7505 [13:42<1:08:38,  1.59it/s]
 13%|█▎        | 953/7505 [13:42<1:08:49,  1.59it/s]
 13%|█▎        | 954/7505 [13:43<1:08:59,  1.58it/s]
 13%|█▎        | 955/7505 [13:44<1:09:05,  1.58it/s]
 13%|█▎        | 956/7505 [13:44<1:09:09,  1.58it/s]
 13%|█▎        | 957/7505 [13:45<1:09:13,  1.58it/s]
 13%|█▎        | 958/7505 [13:45<1:09:14,  1.58it/s]
 13%|█▎        | 959/7505 [13:46<1:09:14,  1.58it/s]
 13%|█▎        | 960/7505 [13:47<1:09:14,  1.58it/s]
                                                    
{'loss': '5.202', 'grad_norm': '0.2887', 'learning_rate': '0.0006', 'epoch': '50.54'}
+
 13%|█▎        | 960/7505 [13:47<1:09:14,  1.58it/s]
 13%|█▎        | 961/7505 [13:47<1:09:15,  1.57it/s]
 13%|█▎        | 962/7505 [13:48<1:09:16,  1.57it/s]
 13%|█▎        | 963/7505 [13:49<1:09:16,  1.57it/s]
 13%|█▎        | 964/7505 [13:49<1:09:16,  1.57it/s]
 13%|█▎        | 965/7505 [13:50<1:09:15,  1.57it/s]
 13%|█▎        | 966/7505 [13:51<1:09:14,  1.57it/s]
 13%|█▎        | 967/7505 [13:51<1:09:15,  1.57it/s]
 13%|█▎        | 968/7505 [13:52<1:09:11,  1.57it/s]
 13%|█▎        | 969/7505 [13:52<1:01:29,  1.77it/s]
 13%|█▎        | 970/7505 [13:53<1:06:51,  1.63it/s]
                                                    
{'loss': '5.206', 'grad_norm': '0.3368', 'learning_rate': '0.0006', 'epoch': '51.05'}
+
 13%|█▎        | 970/7505 [13:53<1:06:51,  1.63it/s]
 13%|█▎        | 971/7505 [13:54<1:07:33,  1.61it/s]
 13%|█▎        | 972/7505 [13:54<1:08:00,  1.60it/s]
 13%|█▎        | 973/7505 [13:55<1:08:19,  1.59it/s]
 13%|█▎        | 974/7505 [13:55<1:08:36,  1.59it/s]
 13%|█▎        | 975/7505 [13:56<1:08:45,  1.58it/s]
 13%|█▎        | 976/7505 [13:57<1:08:51,  1.58it/s]
 13%|█▎        | 977/7505 [13:57<1:08:54,  1.58it/s]
 13%|█▎        | 978/7505 [13:58<1:08:56,  1.58it/s]
 13%|█▎        | 979/7505 [13:59<1:08:59,  1.58it/s]
 13%|█▎        | 980/7505 [13:59<1:09:00,  1.58it/s]
                                                    
{'loss': '5.187', 'grad_norm': '0.3378', 'learning_rate': '0.0006', 'epoch': '51.59'}
+
 13%|█▎        | 980/7505 [13:59<1:09:00,  1.58it/s]
 13%|█▎        | 981/7505 [14:00<1:09:02,  1.57it/s]
 13%|█▎        | 982/7505 [14:01<1:09:02,  1.57it/s]
 13%|█▎        | 983/7505 [14:01<1:09:03,  1.57it/s]
 13%|█▎        | 984/7505 [14:02<1:09:02,  1.57it/s]
 13%|█▎        | 985/7505 [14:02<1:08:59,  1.58it/s]
 13%|█▎        | 986/7505 [14:03<1:09:00,  1.57it/s]
 13%|█▎        | 987/7505 [14:04<1:08:58,  1.57it/s]
 13%|█▎        | 988/7505 [14:04<1:01:18,  1.77it/s]
 13%|█▎        | 989/7505 [14:05<1:05:58,  1.65it/s]
 13%|█▎        | 990/7505 [14:05<1:06:52,  1.62it/s]
                                                    
{'loss': '5.227', 'grad_norm': '0.3655', 'learning_rate': '0.0006', 'epoch': '52.11'}
+
 13%|█▎        | 990/7505 [14:05<1:06:52,  1.62it/s]
 13%|█▎        | 991/7505 [14:06<1:07:30,  1.61it/s]
 13%|█▎        | 992/7505 [14:07<1:07:55,  1.60it/s]
 13%|█▎        | 993/7505 [14:07<1:08:13,  1.59it/s]
 13%|█▎        | 994/7505 [14:08<1:08:25,  1.59it/s]
 13%|█▎        | 995/7505 [14:09<1:08:33,  1.58it/s]
 13%|█▎        | 996/7505 [14:09<1:08:38,  1.58it/s]
 13%|█▎        | 997/7505 [14:10<1:08:41,  1.58it/s]
 13%|█▎        | 998/7505 [14:11<1:08:43,  1.58it/s]
 13%|█▎        | 999/7505 [14:11<1:08:45,  1.58it/s]
 13%|█▎        | 1000/7505 [14:12<1:08:46,  1.58it/s]
                                                     
{'loss': '5.166', 'grad_norm': '0.3055', 'learning_rate': '0.0006', 'epoch': '52.64'}
+
 13%|█▎        | 1000/7505 [14:12<1:08:46,  1.58it/s]
 13%|█▎        | 1001/7505 [14:12<1:08:48,  1.58it/s]
 13%|█▎        | 1002/7505 [14:13<1:08:49,  1.57it/s]
 13%|█▎        | 1003/7505 [14:14<1:08:49,  1.57it/s]
 13%|█▎        | 1004/7505 [14:14<1:08:48,  1.57it/s]
 13%|█▎        | 1005/7505 [14:15<1:08:47,  1.57it/s]
 13%|█▎        | 1006/7505 [14:16<1:08:46,  1.57it/s]
 13%|█▎        | 1007/7505 [14:16<1:01:07,  1.77it/s]
 13%|█▎        | 1008/7505 [14:17<1:06:11,  1.64it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 23.56it/s]
+
 15%|█▌        | 6/39 [00:00<00:01, 17.77it/s]
+
 21%|██        | 8/39 [00:00<00:01, 16.74it/s]
+
 26%|██▌       | 10/39 [00:00<00:01, 16.16it/s]
+
 31%|███       | 12/39 [00:00<00:01, 15.82it/s]
+
 36%|███▌      | 14/39 [00:00<00:01, 15.59it/s]
+
 41%|████      | 16/39 [00:00<00:01, 15.46it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 15.35it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 15.31it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 15.25it/s]
+
 62%|██████▏   | 24/39 [00:01<00:00, 15.24it/s]
+
 67%|██████▋   | 26/39 [00:01<00:00, 15.23it/s]
+
 72%|███████▏  | 28/39 [00:01<00:00, 15.21it/s]
+
 77%|███████▋  | 30/39 [00:01<00:00, 15.20it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 15.20it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 15.19it/s]
+
 92%|█████████▏| 36/39 [00:02<00:00, 15.19it/s]
+
 97%|█████████▋| 38/39 [00:02<00:00, 15.19it/s]
                                                     
+
                                               
{'eval_loss': '4.391', 'eval_runtime': '2.616', 'eval_samples_per_second': '933', 'eval_steps_per_second': '14.91', 'epoch': '53.05'}
+
 13%|█▎        | 1008/7505 [14:19<1:06:11,  1.64it/s]
+
100%|██████████| 39/39 [00:02<00:00, 15.19it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-1008
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-1008/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-1008/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  2.18it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  2.18it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-1008/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-1008/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-1008/tokenizer_config.json
+[eval] epoch 53.054 step 1008 loss 4.3912 (best 4.3912 @ ep 53.054, 0/20 since best)
+
+
                                               
 13%|█▎        | 1009/7505 [14:22<3:35:25,  1.99s/it]
 13%|█▎        | 1010/7505 [14:23<2:51:22,  1.58s/it]
                                                     
{'loss': '5.204', 'grad_norm': '0.2462', 'learning_rate': '0.0006', 'epoch': '53.16'}
+
 13%|█▎        | 1010/7505 [14:23<2:51:22,  1.58s/it]
 13%|█▎        | 1011/7505 [14:23<2:20:33,  1.30s/it]
 13%|█▎        | 1012/7505 [14:24<1:58:57,  1.10s/it]
 13%|█▎        | 1013/7505 [14:25<1:43:51,  1.04it/s]
 14%|█▎        | 1014/7505 [14:25<1:33:17,  1.16it/s]
 14%|█▎        | 1015/7505 [14:26<1:25:54,  1.26it/s]
 14%|█▎        | 1016/7505 [14:26<1:20:43,  1.34it/s]
 14%|█▎        | 1017/7505 [14:27<1:17:05,  1.40it/s]
 14%|█▎        | 1018/7505 [14:28<1:14:33,  1.45it/s]
 14%|█▎        | 1019/7505 [14:28<1:12:45,  1.49it/s]
 14%|█▎        | 1020/7505 [14:29<1:11:30,  1.51it/s]
                                                     
{'loss': '5.153', 'grad_norm': '0.2331', 'learning_rate': '0.0006', 'epoch': '53.7'}
+
 14%|█▎        | 1020/7505 [14:29<1:11:30,  1.51it/s]
 14%|█▎        | 1021/7505 [14:30<1:10:37,  1.53it/s]
 14%|█▎        | 1022/7505 [14:30<1:10:01,  1.54it/s]
 14%|█▎        | 1023/7505 [14:31<1:09:36,  1.55it/s]
 14%|█▎        | 1024/7505 [14:31<1:09:16,  1.56it/s]
 14%|█▎        | 1025/7505 [14:32<1:09:01,  1.56it/s]
 14%|█▎        | 1026/7505 [14:33<1:01:15,  1.76it/s]
 14%|█▎        | 1027/7505 [14:33<1:06:00,  1.64it/s]
 14%|█▎        | 1028/7505 [14:34<1:06:45,  1.62it/s]
 14%|█▎        | 1029/7505 [14:35<1:07:15,  1.60it/s]
 14%|█▎        | 1030/7505 [14:35<1:07:36,  1.60it/s]
                                                     
{'loss': '5.155', 'grad_norm': '0.3396', 'learning_rate': '0.0006', 'epoch': '54.21'}
+
 14%|█▎        | 1030/7505 [14:35<1:07:36,  1.60it/s]
 14%|█▎        | 1031/7505 [14:36<1:07:56,  1.59it/s]
 14%|█▍        | 1032/7505 [14:36<1:08:06,  1.58it/s]
 14%|█▍        | 1033/7505 [14:37<1:08:11,  1.58it/s]
 14%|█▍        | 1034/7505 [14:38<1:08:17,  1.58it/s]
 14%|█▍        | 1035/7505 [14:38<1:08:20,  1.58it/s]
 14%|█▍        | 1036/7505 [14:39<1:08:21,  1.58it/s]
 14%|█▍        | 1037/7505 [14:40<1:08:22,  1.58it/s]
 14%|█▍        | 1038/7505 [14:40<1:08:23,  1.58it/s]
 14%|█▍        | 1039/7505 [14:41<1:08:25,  1.57it/s]
 14%|█▍        | 1040/7505 [14:42<1:08:45,  1.57it/s]
                                                     
{'loss': '5.216', 'grad_norm': '0.3856', 'learning_rate': '0.0006', 'epoch': '54.75'}
+
 14%|█▍        | 1040/7505 [14:42<1:08:45,  1.57it/s]
 14%|█▍        | 1041/7505 [14:42<1:08:48,  1.57it/s]
 14%|█▍        | 1042/7505 [14:43<1:08:40,  1.57it/s]
 14%|█▍        | 1043/7505 [14:43<1:08:34,  1.57it/s]
 14%|█▍        | 1044/7505 [14:44<1:08:28,  1.57it/s]
 14%|█▍        | 1045/7505 [14:44<1:00:50,  1.77it/s]
 14%|█▍        | 1046/7505 [14:45<1:05:43,  1.64it/s]
 14%|█▍        | 1047/7505 [14:46<1:06:28,  1.62it/s]
 14%|█▍        | 1048/7505 [14:46<1:07:01,  1.61it/s]
 14%|█▍        | 1049/7505 [14:47<1:07:25,  1.60it/s]
 14%|█▍        | 1050/7505 [14:48<1:07:39,  1.59it/s]
                                                     
{'loss': '5.182', 'grad_norm': '0.3221', 'learning_rate': '0.0006', 'epoch': '55.27'}
+
 14%|█▍        | 1050/7505 [14:48<1:07:39,  1.59it/s]
 14%|█▍        | 1051/7505 [14:48<1:07:51,  1.59it/s]
 14%|█▍        | 1052/7505 [14:49<1:08:00,  1.58it/s]
 14%|█▍        | 1053/7505 [14:50<1:08:04,  1.58it/s]
 14%|█▍        | 1054/7505 [14:50<1:08:07,  1.58it/s]
 14%|█▍        | 1055/7505 [14:51<1:08:10,  1.58it/s]
 14%|█▍        | 1056/7505 [14:52<1:08:12,  1.58it/s]
 14%|█▍        | 1057/7505 [14:52<1:08:12,  1.58it/s]
 14%|█▍        | 1058/7505 [14:53<1:08:13,  1.57it/s]
 14%|█▍        | 1059/7505 [14:53<1:08:13,  1.57it/s]
 14%|█▍        | 1060/7505 [14:54<1:08:14,  1.57it/s]
                                                     
{'loss': '5.106', 'grad_norm': '0.2401', 'learning_rate': '0.0006', 'epoch': '55.81'}
+
 14%|█▍        | 1060/7505 [14:54<1:08:14,  1.57it/s]
 14%|█▍        | 1061/7505 [14:55<1:08:15,  1.57it/s]
 14%|█▍        | 1062/7505 [14:55<1:08:14,  1.57it/s]
 14%|█▍        | 1063/7505 [14:56<1:08:11,  1.57it/s]
 14%|█▍        | 1064/7505 [14:56<1:00:35,  1.77it/s]
 14%|█▍        | 1065/7505 [14:57<1:05:04,  1.65it/s]
 14%|█▍        | 1066/7505 [14:58<1:06:00,  1.63it/s]
 14%|█▍        | 1067/7505 [14:58<1:06:40,  1.61it/s]
 14%|█▍        | 1068/7505 [14:59<1:07:07,  1.60it/s]
 14%|█▍        | 1069/7505 [15:00<1:07:27,  1.59it/s]
 14%|█▍        | 1070/7505 [15:00<1:07:38,  1.59it/s]
                                                     
{'loss': '5.1', 'grad_norm': '0.2483', 'learning_rate': '0.0006', 'epoch': '56.32'}
+
 14%|█▍        | 1070/7505 [15:00<1:07:38,  1.59it/s]
 14%|█▍        | 1071/7505 [15:01<1:07:48,  1.58it/s]
 14%|█▍        | 1072/7505 [15:02<1:07:53,  1.58it/s]
 14%|█▍        | 1073/7505 [15:02<1:07:55,  1.58it/s]
 14%|█▍        | 1074/7505 [15:03<1:07:57,  1.58it/s]
 14%|█▍        | 1075/7505 [15:03<1:08:00,  1.58it/s]
 14%|█▍        | 1076/7505 [15:04<1:08:00,  1.58it/s]
 14%|█▍        | 1077/7505 [15:05<1:08:01,  1.58it/s]
 14%|█▍        | 1078/7505 [15:05<1:08:02,  1.57it/s]
 14%|█▍        | 1079/7505 [15:06<1:08:00,  1.57it/s]
 14%|█▍        | 1080/7505 [15:07<1:08:00,  1.57it/s]
                                                     
{'loss': '5.143', 'grad_norm': '0.2414', 'learning_rate': '0.0006', 'epoch': '56.86'}
+
 14%|█▍        | 1080/7505 [15:07<1:08:00,  1.57it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 22.16it/s]
+
 15%|█▌        | 6/39 [00:00<00:01, 17.44it/s]
+
 21%|██        | 8/39 [00:00<00:01, 16.54it/s]
+
 26%|██▌       | 10/39 [00:00<00:01, 16.05it/s]
+
 31%|███       | 12/39 [00:00<00:01, 15.73it/s]
+
 36%|███▌      | 14/39 [00:00<00:01, 15.56it/s]
+
 41%|████      | 16/39 [00:00<00:01, 15.43it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 15.36it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 15.29it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 15.26it/s]
+
 62%|██████▏   | 24/39 [00:01<00:00, 15.24it/s]
+
 67%|██████▋   | 26/39 [00:01<00:00, 15.23it/s]
+
 72%|███████▏  | 28/39 [00:01<00:00, 15.21it/s]
+
 77%|███████▋  | 30/39 [00:01<00:00, 15.22it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 15.20it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 15.20it/s]
+
 92%|█████████▏| 36/39 [00:02<00:00, 15.18it/s]
+
 97%|█████████▋| 38/39 [00:02<00:00, 15.20it/s]
                                                     
+
                                               
{'eval_loss': '4.369', 'eval_runtime': '2.642', 'eval_samples_per_second': '923.9', 'eval_steps_per_second': '14.76', 'epoch': '56.86'}
+
 14%|█▍        | 1080/7505 [15:09<1:08:00,  1.57it/s]
+
100%|██████████| 39/39 [00:02<00:00, 15.20it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-1080
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-1080/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-1080/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.98it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.98it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-1080/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-1080/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-1080/tokenizer_config.json
+[eval] epoch 56.859 step 1080 loss 4.3688 (best 4.3688 @ ep 56.859, 0/20 since best)
+
+
                                               
 14%|█▍        | 1081/7505 [15:12<3:36:55,  2.03s/it]
 14%|█▍        | 1082/7505 [15:13<2:52:10,  1.61s/it]
 14%|█▍        | 1083/7505 [15:13<2:13:21,  1.25s/it]
 14%|█▍        | 1084/7505 [15:14<1:56:19,  1.09s/it]
 14%|█▍        | 1085/7505 [15:14<1:41:49,  1.05it/s]
 14%|█▍        | 1086/7505 [15:15<1:31:42,  1.17it/s]
 14%|█▍        | 1087/7505 [15:16<1:24:35,  1.26it/s]
 14%|█▍        | 1088/7505 [15:16<1:19:36,  1.34it/s]
 15%|█▍        | 1089/7505 [15:17<1:16:06,  1.41it/s]
 15%|█▍        | 1090/7505 [15:17<1:13:40,  1.45it/s]
                                                     
{'loss': '5.046', 'grad_norm': '0.2745', 'learning_rate': '0.0006', 'epoch': '57.38'}
+
 15%|█▍        | 1090/7505 [15:17<1:13:40,  1.45it/s]
 15%|█▍        | 1091/7505 [15:18<1:12:07,  1.48it/s]
 15%|█▍        | 1092/7505 [15:19<1:10:58,  1.51it/s]
 15%|█▍        | 1093/7505 [15:19<1:10:00,  1.53it/s]
 15%|█▍        | 1094/7505 [15:20<1:09:21,  1.54it/s]
 15%|█▍        | 1095/7505 [15:21<1:08:54,  1.55it/s]
 15%|█▍        | 1096/7505 [15:21<1:08:35,  1.56it/s]
 15%|█▍        | 1097/7505 [15:22<1:08:21,  1.56it/s]
 15%|█▍        | 1098/7505 [15:23<1:08:11,  1.57it/s]
 15%|█▍        | 1099/7505 [15:23<1:08:04,  1.57it/s]
 15%|█▍        | 1100/7505 [15:24<1:07:58,  1.57it/s]
                                                     
{'loss': '5.158', 'grad_norm': '0.2629', 'learning_rate': '0.0006', 'epoch': '57.91'}
+
 15%|█▍        | 1100/7505 [15:24<1:07:58,  1.57it/s]
 15%|█▍        | 1101/7505 [15:24<1:07:53,  1.57it/s]
 15%|█▍        | 1102/7505 [15:25<1:00:17,  1.77it/s]
 15%|█▍        | 1103/7505 [15:26<1:05:15,  1.63it/s]
 15%|█▍        | 1104/7505 [15:26<1:05:59,  1.62it/s]
 15%|█▍        | 1105/7505 [15:27<1:06:29,  1.60it/s]
 15%|█▍        | 1106/7505 [15:27<1:06:51,  1.60it/s]
 15%|█▍        | 1107/7505 [15:28<1:07:06,  1.59it/s]
 15%|█▍        | 1108/7505 [15:29<1:07:15,  1.59it/s]
 15%|█▍        | 1109/7505 [15:29<1:07:24,  1.58it/s]
 15%|█▍        | 1110/7505 [15:30<1:07:30,  1.58it/s]
                                                     
{'loss': '5.073', 'grad_norm': '0.3739', 'learning_rate': '0.0006', 'epoch': '58.43'}
+
 15%|█▍        | 1110/7505 [15:30<1:07:30,  1.58it/s]
 15%|█▍        | 1111/7505 [15:31<1:07:32,  1.58it/s]
 15%|█▍        | 1112/7505 [15:31<1:07:34,  1.58it/s]
 15%|█▍        | 1113/7505 [15:32<1:07:35,  1.58it/s]
 15%|█▍        | 1114/7505 [15:33<1:07:35,  1.58it/s]
 15%|█▍        | 1115/7505 [15:33<1:07:36,  1.58it/s]
 15%|█▍        | 1116/7505 [15:34<1:07:36,  1.57it/s]
 15%|█▍        | 1117/7505 [15:34<1:07:34,  1.58it/s]
 15%|█▍        | 1118/7505 [15:35<1:07:34,  1.58it/s]
 15%|█▍        | 1119/7505 [15:36<1:07:33,  1.58it/s]
 15%|█▍        | 1120/7505 [15:36<1:07:29,  1.58it/s]
                                                     
{'loss': '5.1', 'grad_norm': '0.2771', 'learning_rate': '0.0006', 'epoch': '58.97'}
+
 15%|█▍        | 1120/7505 [15:36<1:07:29,  1.58it/s]
 15%|█▍        | 1121/7505 [15:37<1:00:00,  1.77it/s]
 15%|█▍        | 1122/7505 [15:37<1:05:23,  1.63it/s]
 15%|█▍        | 1123/7505 [15:38<1:06:01,  1.61it/s]
 15%|█▍        | 1124/7505 [15:39<1:06:28,  1.60it/s]
 15%|█▍        | 1125/7505 [15:39<1:06:47,  1.59it/s]
 15%|█▌        | 1126/7505 [15:40<1:06:59,  1.59it/s]
 15%|█▌        | 1127/7505 [15:41<1:07:08,  1.58it/s]
 15%|█▌        | 1128/7505 [15:41<1:07:13,  1.58it/s]
 15%|█▌        | 1129/7505 [15:42<1:07:17,  1.58it/s]
 15%|█▌        | 1130/7505 [15:43<1:07:21,  1.58it/s]
                                                     
{'loss': '5.069', 'grad_norm': '0.3077', 'learning_rate': '0.0006', 'epoch': '59.48'}
+
 15%|█▌        | 1130/7505 [15:43<1:07:21,  1.58it/s]
 15%|█▌        | 1131/7505 [15:43<1:07:25,  1.58it/s]
 15%|█▌        | 1132/7505 [15:44<1:07:26,  1.58it/s]
 15%|█▌        | 1133/7505 [15:44<1:07:25,  1.58it/s]
 15%|█▌        | 1134/7505 [15:45<1:07:24,  1.58it/s]
 15%|█▌        | 1135/7505 [15:46<1:07:23,  1.58it/s]
 15%|█▌        | 1136/7505 [15:46<1:07:23,  1.58it/s]
 15%|█▌        | 1137/7505 [15:47<1:07:23,  1.57it/s]
 15%|█▌        | 1138/7505 [15:48<1:07:24,  1.57it/s]
 15%|█▌        | 1139/7505 [15:48<1:07:22,  1.57it/s]
 15%|█▌        | 1140/7505 [15:49<59:52,  1.77it/s]  
                                                   
{'loss': '5.103', 'grad_norm': '0.2758', 'learning_rate': '0.0006', 'epoch': '60'}
+
 15%|█▌        | 1140/7505 [15:49<59:52,  1.77it/s]
 15%|█▌        | 1141/7505 [15:49<1:05:11,  1.63it/s]
 15%|█▌        | 1142/7505 [15:50<1:05:49,  1.61it/s]
 15%|█▌        | 1143/7505 [15:51<1:06:18,  1.60it/s]
 15%|█▌        | 1144/7505 [15:51<1:06:35,  1.59it/s]
 15%|█▌        | 1145/7505 [15:52<1:06:48,  1.59it/s]
 15%|█▌        | 1146/7505 [15:53<1:06:57,  1.58it/s]
 15%|█▌        | 1147/7505 [15:53<1:07:04,  1.58it/s]
 15%|█▌        | 1148/7505 [15:54<1:07:07,  1.58it/s]
 15%|█▌        | 1149/7505 [15:54<1:07:09,  1.58it/s]
 15%|█▌        | 1150/7505 [15:55<1:07:11,  1.58it/s]
                                                     
{'loss': '5.021', 'grad_norm': '0.2864', 'learning_rate': '0.0006', 'epoch': '60.54'}
+
 15%|█▌        | 1150/7505 [15:55<1:07:11,  1.58it/s]
 15%|█▌        | 1151/7505 [15:56<1:07:15,  1.57it/s]
 15%|█▌        | 1152/7505 [15:56<1:07:15,  1.57it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 23.07it/s]
+
 15%|█▌        | 6/39 [00:00<00:01, 17.64it/s]
+
 21%|██        | 8/39 [00:00<00:01, 16.66it/s]
+
 26%|██▌       | 10/39 [00:00<00:01, 16.13it/s]
+
 31%|███       | 12/39 [00:00<00:01, 15.78it/s]
+
 36%|███▌      | 14/39 [00:00<00:01, 15.58it/s]
+
 41%|████      | 16/39 [00:00<00:01, 15.42it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 15.33it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 15.28it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 15.25it/s]
+
 62%|██████▏   | 24/39 [00:01<00:00, 15.24it/s]
+
 67%|██████▋   | 26/39 [00:01<00:00, 15.22it/s]
+
 72%|███████▏  | 28/39 [00:01<00:00, 15.21it/s]
+
 77%|███████▋  | 30/39 [00:01<00:00, 15.19it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 15.19it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 15.18it/s]
+
 92%|█████████▏| 36/39 [00:02<00:00, 15.18it/s]
+
 97%|█████████▋| 38/39 [00:02<00:00, 15.18it/s]
                                                     
+
                                               
{'eval_loss': '4.363', 'eval_runtime': '2.642', 'eval_samples_per_second': '923.9', 'eval_steps_per_second': '14.76', 'epoch': '60.64'}
+
 15%|█▌        | 1152/7505 [15:59<1:07:15,  1.57it/s]
+
100%|██████████| 39/39 [00:02<00:00, 15.18it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-1152
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-1152/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-1152/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  2.00it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  2.00it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-1152/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-1152/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_19M_C3_rand5_r2l50_i5exp/checkpoint-1152/tokenizer_config.json
+[eval] epoch 60.644 step 1152 loss 4.3625 (best 4.3625 @ ep 60.644, 0/20 since best)
+
+
                                               
 15%|█▌        | 1153/7505 [16:02<3:34:28,  2.03s/it]
 15%|█▌        | 1154/7505 [16:02<2:50:16,  1.61s/it]
 15%|█▌        | 1155/7505 [16:03<2:19:19,  1.32s/it]
 15%|█▌        | 1156/7505 [16:04<1:57:39,  1.11s/it]
 15%|█▌        | 1157/7505 [16:04<1:42:29,  1.03it/s]
 15%|█▌        | 1158/7505 [16:05<1:31:52,  1.15it/s]
 15%|█▌        | 1159/7505 [16:05<1:16:59,  1.37it/s]
 15%|█▌        | 1160/7505 [16:06<1:16:36,  1.38it/s]
                                                     
{'loss': '5.006', 'grad_norm': '0.2502', 'learning_rate': '0.0006', 'epoch': '61.05'}
+
 15%|█▌        | 1160/7505 [16:06<1:16:36,  1.38it/s]
 15%|█▌        | 1161/7505 [16:07<1:13:48,  1.43it/s]
 15%|█▌        | 1162/7505 [16:07<1:11:49,  1.47it/s]
 15%|█▌        | 1163/7505 [16:08<1:10:25,  1.50it/s]
 16%|█▌        | 1164/7505 [16:09<1:09:26,  1.52it/s]
 16%|█▌        | 1165/7505 [16:09<1:08:45,  1.54it/s]
 16%|█▌        | 1166/7505 [16:10<1:08:15,  1.55it/s]
 16%|█▌        | 1167/7505 [16:10<1:07:55,  1.56it/s]
 16%|█▌        | 1168/7505 [16:11<1:07:39,  1.56it/s]
 16%|█▌        | 1169/7505 [16:12<1:07:28,  1.56it/s]
 16%|█▌        | 1170/7505 [16:12<1:07:20,  1.57it/s]
                                                     
{'loss': '4.972', 'grad_norm': '0.2448', 'learning_rate': '0.0006', 'epoch': '61.59'}
+
 16%|█▌        | 1170/7505 [16:12<1:07:20,  1.57it/s]
 16%|█▌        | 1171/7505 [16:13<1:07:18,  1.57it/s]
 16%|█▌        | 1172/7505 [16:14<1:07:13,  1.57it/s]
 16%|█▌        | 1173/7505 [16:14<1:07:10,  1.57it/s]
 16%|█▌        | 1174/7505 [16:15<1:07:07,  1.57it/s]
 16%|█▌        | 1175/7505 [16:15<1:07:07,  1.57it/s]
 16%|█▌        | 1176/7505 [16:16<1:07:05,  1.57it/s]
 16%|█▌        | 1177/7505 [16:17<1:07:00,  1.57it/s]
 16%|█▌        | 1178/7505 [16:17<59:33,  1.77it/s]  
 16%|█▌        | 1179/7505 [16:18<1:04:20,  1.64it/s]
 16%|█▌        | 1180/7505 [16:19<1:05:07,  1.62it/s]
                                                     
{'loss': '5.042', 'grad_norm': '0.2589', 'learning_rate': '0.0006', 'epoch': '62.11'}
+
 16%|█▌        | 1180/7505 [16:19<1:05:07,  1.62it/s]
 16%|█▌        | 1181/7505 [16:19<1:05:43,  1.60it/s]
 16%|█▌        | 1182/7505 [16:20<1:06:05,  1.59it/s]
 16%|█▌        | 1183/7505 [16:20<1:06:20,  1.59it/s]
 16%|█▌        | 1184/7505 [16:21<1:06:30,  1.58it/s]
 16%|█▌        | 1185/7505 [16:22<1:06:37,  1.58it/s]
 16%|█▌        | 1186/7505 [16:22<1:06:42,  1.58it/s]
 16%|█▌        | 1187/7505 [16:23<1:06:44,  1.58it/s]
 16%|█▌        | 1188/7505 [16:24<1:06:46,  1.58it/s]
 16%|█▌        | 1189/7505 [16:24<1:06:47,  1.58it/s]
 16%|█▌        | 1190/7505 [16:25<1:06:50,  1.57it/s]
                                                     
{'loss': '5.007', 'grad_norm': '0.3443', 'learning_rate': '0.0006', 'epoch': '62.64'}
+
 16%|█▌        | 1190/7505 [16:25<1:06:50,  1.57it/s]
 16%|█▌        | 1191/7505 [16:26<1:06:49,  1.57it/s]
 16%|█▌        | 1192/7505 [16:26<1:06:51,  1.57it/s]
 16%|█▌        | 1193/7505 [16:27<1:06:52,  1.57it/s]
 16%|█▌        | 1194/7505 [16:27<1:06:49,  1.57it/s]
 16%|█▌        | 1195/7505 [16:28<1:06:49,  1.57it/s]
 16%|█▌        | 1196/7505 [16:29<1:06:46,  1.57it/s]
 16%|█▌        | 1197/7505 [16:29<59:21,  1.77it/s]  
 16%|█▌        | 1198/7505 [16:30<1:03:36,  1.65it/s]
 16%|█▌        | 1199/7505 [16:30<1:04:34,  1.63it/s]
 16%|█▌        | 1200/7505 [16:31<1:05:11,  1.61it/s]
                                                     
{'loss': '5.058', 'grad_norm': '0.3078', 'learning_rate': '0.0006', 'epoch': '63.16'}
+
 16%|█▌        | 1200/7505 [16:31<1:05:11,  1.61it/s]
 16%|█▌        | 1201/7505 [16:32<1:05:40,  1.60it/s]
 16%|█▌        | 1202/7505 [16:32<1:05:58,  1.59it/s]
 16%|█▌        | 1203/7505 [16:33<1:06:10,  1.59it/s]
 16%|█▌        | 1204/7505 [16:34<1:06:17,  1.58it/s]
 16%|█▌        | 1205/7505 [16:34<1:06:25,  1.58it/s]
 16%|█▌        | 1206/7505 [16:35<1:06:29,  1.58it/s]
 16%|█▌        | 1207/7505 [16:36<1:06:32,  1.58it/s]
 16%|█▌        | 1208/7505 [16:36<1:06:35,  1.58it/s]
 16%|█▌        | 1209/7505 [16:37<1:06:35,  1.58it/s]
 16%|█▌        | 1210/7505 [16:37<1:06:37,  1.57it/s]
                                                     
{'loss': '4.992', 'grad_norm': '0.3481', 'learning_rate': '0.0006', 'epoch': '63.7'}
+
 16%|█▌        | 1210/7505 [16:37<1:06:37,  1.57it/s]
 16%|█▌        | 1211/7505 [16:38<1:06:38,  1.57it/s]
 16%|█▌        | 1212/7505 [16:39<1:06:35,  1.57it/s]
 16%|█▌        | 1213/7505 [16:39<1:06:34,  1.58it/s]
 16%|█▌        | 1214/7505 [16:40<1:06:34,  1.57it/s]
 16%|█▌        | 1215/7505 [16:41<1:06:32,  1.58it/s]
 16%|█▌        | 1216/7505 [16:41<59:08,  1.77it/s]  
 16%|█▌        | 1217/7505 [16:42<1:03:54,  1.64it/s]
 16%|█▌        | 1218/7505 [16:42<1:04:41,  1.62it/s]
 16%|█▌        | 1219/7505 [16:43<1:05:14,  1.61it/s]
 16%|█▋        | 1220/7505 [16:44<1:05:38,  1.60it/s]
                                                     
{'loss': '4.987', 'grad_norm': '0.3037', 'learning_rate': '0.0006', 'epoch': '64.21'}
+
 16%|█▋        | 1220/7505 [16:44<1:05:38,  1.60it/s]
 16%|█▋        | 1221/7505 [16:44<1:05:54,  1.59it/s]
 16%|█▋        | 1222/7505 [16:45<1:06:07,  1.58it/s]
 16%|█▋        | 1223/7505 [16:46<1:06:14,  1.58it/s]
 16%|█▋        | 1224/7505 [16:46<1:06:18,  1.58it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 22.87it/s]
+
 15%|█▌        | 6/39 [00:00<00:01, 17.59it/s]
+
 21%|██        | 8/39 [00:00<00:01, 16.60it/s]
+
 26%|██▌       | 10/39 [00:00<00:01, 16.09it/s]
+
 31%|███       | 12/39 [00:00<00:01, 15.76it/s]
+
 36%|███▌      | 14/39 [00:00<00:01, 15.55it/s]
+
 41%|████      | 16/39 [00:00<00:01, 15.42it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 15.32it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 15.27it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 15.22it/s]
+
 62%|██████▏   | 24/39 [00:01<00:00, 15.21it/s]
+
 67%|██████▋   | 26/39 [00:01<00:00, 15.20it/s]
+
 72%|███████▏  | 28/39 [00:01<00:00, 15.20it/s]
+
 77%|███████▋  | 30/39 [00:01<00:00, 15.19it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 15.19it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 15.19it/s]
+
 92%|█████████▏| 36/39 [00:02<00:00, 15.18it/s]
+
 97%|█████████▋| 38/39 [00:02<00:00, 15.19it/s]
                                                     
+
                                               
{'eval_loss': '4.383', 'eval_runtime': '2.629', 'eval_samples_per_second': '928.4', 'eval_steps_per_second': '14.83', 'epoch': '64.43'}
+
 16%|█▋        | 1224/7505 [16:49<1:06:18,  1.58it/s]
+
100%|██████████| 39/39 [00:02<00:00, 15.19it/s][eval] epoch 64.43 step 1224 loss 4.3829 (best 4.3625 @ ep 60.644, 1/20 since best)
+
+
                                               
 16%|█▋        | 1225/7505 [16:49<2:28:53,  1.42s/it]
 16%|█▋        | 1226/7505 [16:50<2:04:11,  1.19s/it]
 16%|█▋        | 1227/7505 [16:51<1:46:53,  1.02s/it]
 16%|█▋        | 1228/7505 [16:51<1:34:45,  1.10it/s]
 16%|█▋        | 1229/7505 [16:52<1:26:14,  1.21it/s]
 16%|█▋        | 1230/7505 [16:53<1:20:16,  1.30it/s]
                                                     
{'loss': '4.982', 'grad_norm': '0.3093', 'learning_rate': '0.0006', 'epoch': '64.75'}
+
 16%|█▋        | 1230/7505 [16:53<1:20:16,  1.30it/s]
 16%|█▋        | 1231/7505 [16:53<1:16:07,  1.37it/s]
 16%|█▋        | 1232/7505 [16:54<1:13:13,  1.43it/s]
 16%|█▋        | 1233/7505 [16:54<1:11:09,  1.47it/s]
 16%|█▋        | 1234/7505 [16:55<1:09:41,  1.50it/s]
 16%|█▋        | 1235/7505 [16:56<1:01:18,  1.70it/s]
 16%|█▋        | 1236/7505 [16:56<1:05:24,  1.60it/s]
 16%|█▋        | 1237/7505 [16:57<1:05:41,  1.59it/s]
 16%|█▋        | 1238/7505 [16:58<1:05:52,  1.59it/s]
 17%|█▋        | 1239/7505 [16:58<1:05:59,  1.58it/s]
 17%|█▋        | 1240/7505 [16:59<1:06:05,  1.58it/s]
                                                     
{'loss': '4.92', 'grad_norm': '0.2871', 'learning_rate': '0.0006', 'epoch': '65.27'}
+
 17%|█▋        | 1240/7505 [16:59<1:06:05,  1.58it/s]
 17%|█▋        | 1241/7505 [16:59<1:06:09,  1.58it/s]
 17%|█▋        | 1242/7505 [17:00<1:06:21,  1.57it/s]
 17%|█▋        | 1243/7505 [17:01<1:06:31,  1.57it/s]
 17%|█▋        | 1244/7505 [17:01<1:06:29,  1.57it/s]
 17%|█▋        | 1245/7505 [17:02<1:06:24,  1.57it/s]
 17%|█▋        | 1246/7505 [17:03<1:06:20,  1.57it/s]
 17%|█▋        | 1247/7505 [17:03<1:06:18,  1.57it/s]
 17%|█▋        | 1248/7505 [17:04<1:17:01,  1.35it/s]
 17%|█▋        | 1249/7505 [17:05<1:13:44,  1.41it/s]
 17%|█▋        | 1250/7505 [17:05<1:11:29,  1.46it/s]
                                                     
{'loss': '4.93', 'grad_norm': '0.291', 'learning_rate': '0.0006', 'epoch': '65.81'}
+
 17%|█▋        | 1250/7505 [17:05<1:11:29,  1.46it/s]
 17%|█▋        | 1251/7505 [17:06<1:09:54,  1.49it/s]
 17%|█▋        | 1252/7505 [17:07<1:08:48,  1.51it/s]
 17%|█▋        | 1253/7505 [17:07<1:08:00,  1.53it/s]
 17%|█▋        | 1254/7505 [17:08<1:00:06,  1.73it/s]
 17%|█▋        | 1255/7505 [17:09<1:04:25,  1.62it/s]
 17%|█▋        | 1256/7505 [17:09<1:04:57,  1.60it/s]
 17%|█▋        | 1257/7505 [17:10<1:05:18,  1.59it/s]
 17%|█▋        | 1258/7505 [17:10<1:05:35,  1.59it/s]
 17%|█▋        | 1259/7505 [17:11<1:05:46,  1.58it/s]
 17%|█▋        | 1260/7505 [17:12<1:05:52,  1.58it/s]
                                                     
{'loss': '4.934', 'grad_norm': '0.3711', 'learning_rate': '0.0006', 'epoch': '66.32'}
+
 17%|█▋        | 1260/7505 [17:12<1:05:52,  1.58it/s]
 17%|█▋        | 1261/7505 [17:12<1:05:56,  1.58it/s]
 17%|█▋        | 1262/7505 [17:13<1:05:58,  1.58it/s]
 17%|█▋        | 1263/7505 [17:14<1:06:00,  1.58it/s]
 17%|█▋        | 1264/7505 [17:14<1:06:01,  1.58it/s]
 17%|█▋        | 1265/7505 [17:15<1:06:01,  1.58it/s]
 17%|█▋        | 1266/7505 [17:16<1:06:00,  1.58it/s]
 17%|█▋        | 1267/7505 [17:16<1:06:02,  1.57it/s]
 17%|█▋        | 1268/7505 [17:17<1:06:02,  1.57it/s]
 17%|█▋        | 1269/7505 [17:17<1:06:02,  1.57it/s]
 17%|█▋        | 1270/7505 [17:18<1:06:02,  1.57it/s]
                                                     
{'loss': '4.875', 'grad_norm': '0.3208', 'learning_rate': '0.0006', 'epoch': '66.86'}
+
 17%|█▋        | 1270/7505 [17:18<1:06:02,  1.57it/s]
 17%|█▋        | 1271/7505 [17:19<1:06:01,  1.57it/s]
 17%|█▋        | 1272/7505 [17:19<1:05:59,  1.57it/s]
 17%|█▋        | 1273/7505 [17:20<58:38,  1.77it/s]  
 17%|█▋        | 1274/7505 [17:20<1:03:21,  1.64it/s]
 17%|█▋        | 1275/7505 [17:21<1:04:08,  1.62it/s]
 17%|█▋        | 1276/7505 [17:22<1:04:39,  1.61it/s]
 17%|█▋        | 1277/7505 [17:22<1:05:03,  1.60it/s]
 17%|█▋        | 1278/7505 [17:23<1:05:18,  1.59it/s]
 17%|█▋        | 1279/7505 [17:24<1:05:28,  1.58it/s]
 17%|█▋        | 1280/7505 [17:24<1:05:36,  1.58it/s]
                                                     
{'loss': '4.905', 'grad_norm': '0.3594', 'learning_rate': '0.0006', 'epoch': '67.38'}
+
 17%|█▋        | 1280/7505 [17:24<1:05:36,  1.58it/s]
 17%|█▋        | 1281/7505 [17:25<1:05:42,  1.58it/s]
 17%|█▋        | 1282/7505 [17:26<1:05:45,  1.58it/s]
 17%|█▋        | 1283/7505 [17:26<1:05:46,  1.58it/s]
 17%|█▋        | 1284/7505 [17:27<1:05:47,  1.58it/s]
 17%|█▋        | 1285/7505 [17:27<1:05:47,  1.58it/s]
 17%|█▋        | 1286/7505 [17:28<1:05:47,  1.58it/s]
 17%|█▋        | 1287/7505 [17:29<1:05:47,  1.58it/s]
 17%|█▋        | 1288/7505 [17:29<1:05:48,  1.57it/s]
 17%|█▋        | 1289/7505 [17:30<1:05:46,  1.58it/s]
 17%|█▋        | 1290/7505 [17:31<1:05:46,  1.57it/s]
                                                     
{'loss': '4.976', 'grad_norm': '0.266', 'learning_rate': '0.0006', 'epoch': '67.91'}
+
 17%|█▋        | 1290/7505 [17:31<1:05:46,  1.57it/s]
 17%|█▋        | 1291/7505 [17:31<1:05:45,  1.57it/s]
 17%|█▋        | 1292/7505 [17:32<58:26,  1.77it/s]  
 17%|█▋        | 1293/7505 [17:32<1:03:09,  1.64it/s]
 17%|█▋        | 1294/7505 [17:33<1:14:36,  1.39it/s]
 17%|█▋        | 1295/7505 [17:34<1:11:55,  1.44it/s]
 17%|█▋        | 1296/7505 [17:35<1:10:03,  1.48it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 22.82it/s]
+
 15%|█▌        | 6/39 [00:00<00:01, 17.57it/s]
+
 21%|██        | 8/39 [00:00<00:01, 16.60it/s]
+
 26%|██▌       | 10/39 [00:00<00:01, 16.09it/s]
+
 31%|███       | 12/39 [00:00<00:01, 15.78it/s]
+
 36%|███▌      | 14/39 [00:00<00:01, 15.56it/s]
+
 41%|████      | 16/39 [00:00<00:01, 15.45it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 15.36it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 15.27it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 15.24it/s]
+
 62%|██████▏   | 24/39 [00:01<00:00, 15.24it/s]
+
 67%|██████▋   | 26/39 [00:01<00:00, 15.22it/s]
+
 72%|███████▏  | 28/39 [00:01<00:00, 15.21it/s]
+
 77%|███████▋  | 30/39 [00:01<00:00, 15.21it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 15.20it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 15.21it/s]
+
 92%|█████████▏| 36/39 [00:02<00:00, 15.19it/s]
+
 97%|█████████▋| 38/39 [00:02<00:00, 15.19it/s]
                                                     
+
                                               
{'eval_loss': '4.372', 'eval_runtime': '2.627', 'eval_samples_per_second': '929.1', 'eval_steps_per_second': '14.85', 'epoch': '68.21'}
+
 17%|█▋        | 1296/7505 [17:37<1:10:03,  1.48it/s]
+
100%|██████████| 39/39 [00:02<00:00, 15.19it/s][eval] epoch 68.215 step 1296 loss 4.3717 (best 4.3625 @ ep 60.644, 2/20 since best)
+
+
                                               
 17%|█▋        | 1297/7505 [17:38<2:30:17,  1.45s/it]
 17%|█▋        | 1298/7505 [17:38<2:04:53,  1.21s/it]
 17%|█▋        | 1299/7505 [17:39<1:47:07,  1.04s/it]
 17%|█▋        | 1300/7505 [17:40<1:34:41,  1.09it/s]
                                                     
{'loss': '4.918', 'grad_norm': '0.2867', 'learning_rate': '0.0006', 'epoch': '68.43'}
+
 17%|█▋        | 1300/7505 [17:40<1:34:41,  1.09it/s]
 17%|█▋        | 1301/7505 [17:40<1:25:59,  1.20it/s]
 17%|█▋        | 1302/7505 [17:41<1:19:52,  1.29it/s]
 17%|█▋        | 1303/7505 [17:42<1:26:28,  1.20it/s]
 17%|█▋        | 1304/7505 [17:43<1:30:50,  1.14it/s]
 17%|█▋        | 1305/7505 [17:44<1:23:14,  1.24it/s]
 17%|█▋        | 1306/7505 [17:45<1:37:46,  1.06it/s]
 17%|█▋        | 1307/7505 [17:46<1:28:04,  1.17it/s]
 17%|█▋        | 1308/7505 [17:46<1:21:18,  1.27it/s]
 17%|█▋        | 1309/7505 [17:47<1:16:33,  1.35it/s]
 17%|█▋        | 1310/7505 [17:47<1:13:12,  1.41it/s]
                                                     
{'loss': '4.952', 'grad_norm': '0.318', 'learning_rate': '0.0006', 'epoch': '68.97'}
+
 17%|█▋        | 1310/7505 [17:47<1:13:12,  1.41it/s]
 17%|█▋        | 1311/7505 [17:48<1:03:38,  1.62it/s]
 17%|█▋        | 1312/7505 [17:49<1:06:48,  1.54it/s]
 17%|█▋        | 1313/7505 [17:49<1:06:25,  1.55it/s]
 18%|█▊        | 1314/7505 [17:50<1:06:07,  1.56it/s]
 18%|█▊        | 1315/7505 [17:50<1:06:03,  1.56it/s]
 18%|█▊        | 1316/7505 [17:51<1:06:02,  1.56it/s]
 18%|█▊        | 1317/7505 [17:52<1:05:52,  1.57it/s]
 18%|█▊        | 1318/7505 [17:52<1:05:59,  1.56it/s]
 18%|█▊        | 1319/7505 [17:53<1:05:53,  1.56it/s]
 18%|█▊        | 1320/7505 [17:54<1:05:51,  1.57it/s]
                                                     
{'loss': '4.925', 'grad_norm': '0.3315', 'learning_rate': '0.0006', 'epoch': '69.48'}
+
 18%|█▊        | 1320/7505 [17:54<1:05:51,  1.57it/s]
 18%|█▊        | 1321/7505 [17:54<1:05:44,  1.57it/s]
 18%|█▊        | 1322/7505 [17:55<1:05:39,  1.57it/s]
 18%|█▊        | 1323/7505 [17:56<1:05:33,  1.57it/s]
 18%|█▊        | 1324/7505 [17:56<1:05:30,  1.57it/s]
 18%|█▊        | 1325/7505 [17:57<1:05:28,  1.57it/s]
 18%|█▊        | 1326/7505 [17:57<1:05:27,  1.57it/s]
 18%|█▊        | 1327/7505 [17:58<1:05:23,  1.57it/s]
 18%|█▊        | 1328/7505 [17:59<1:05:23,  1.57it/s]
 18%|█▊        | 1329/7505 [17:59<1:05:20,  1.58it/s]
 18%|█▊        | 1330/7505 [18:00<58:06,  1.77it/s]  
                                                   
{'loss': '4.933', 'grad_norm': '0.4027', 'learning_rate': '0.0006', 'epoch': '70'}
+
 18%|█▊        | 1330/7505 [18:00<58:06,  1.77it/s]
 18%|█▊        | 1331/7505 [18:01<1:02:55,  1.64it/s]
 18%|█▊        | 1332/7505 [18:01<1:03:41,  1.62it/s]
 18%|█▊        | 1333/7505 [18:02<1:04:13,  1.60it/s]
 18%|█▊        | 1334/7505 [18:02<1:06:28,  1.55it/s]
 18%|█▊        | 1335/7505 [18:03<1:06:13,  1.55it/s]
 18%|█▊        | 1336/7505 [18:04<1:06:11,  1.55it/s]
 18%|█▊        | 1337/7505 [18:04<1:06:28,  1.55it/s]
 18%|█▊        | 1338/7505 [18:05<1:06:30,  1.55it/s]
 18%|█▊        | 1339/7505 [18:06<1:08:37,  1.50it/s]
 18%|█▊        | 1340/7505 [18:06<1:08:07,  1.51it/s]
                                                     
{'loss': '4.882', 'grad_norm': '0.384', 'learning_rate': '0.0006', 'epoch': '70.54'}
+
 18%|█▊        | 1340/7505 [18:06<1:08:07,  1.51it/s]
 18%|█▊        | 1341/7505 [18:07<1:08:28,  1.50it/s]
 18%|█▊        | 1342/7505 [18:08<1:08:18,  1.50it/s]
 18%|█▊        | 1343/7505 [18:08<1:08:32,  1.50it/s]
 18%|█▊        | 1344/7505 [18:09<1:08:00,  1.51it/s]
 18%|█▊        | 1345/7505 [18:10<1:07:48,  1.51it/s]
 18%|█▊        | 1346/7505 [18:11<1:17:58,  1.32it/s]
 18%|█▊        | 1347/7505 [18:11<1:14:46,  1.37it/s]
 18%|█▊        | 1348/7505 [18:12<1:13:14,  1.40it/s]
 18%|█▊        | 1349/7505 [18:12<1:03:50,  1.61it/s]
 18%|█▊        | 1350/7505 [18:13<1:07:43,  1.51it/s]
                                                     
{'loss': '4.875', 'grad_norm': '0.2963', 'learning_rate': '0.0006', 'epoch': '71.05'}
+
 18%|█▊        | 1350/7505 [18:13<1:07:43,  1.51it/s]
 18%|█▊        | 1351/7505 [18:14<1:07:36,  1.52it/s]
 18%|█▊        | 1352/7505 [18:15<1:07:32,  1.52it/s]
 18%|█▊        | 1353/7505 [18:15<1:08:04,  1.51it/s]
 18%|█▊        | 1354/7505 [18:16<1:07:46,  1.51it/s]
 18%|█▊        | 1355/7505 [18:17<1:07:27,  1.52it/s]
 18%|█▊        | 1356/7505 [18:17<1:07:13,  1.52it/s]
 18%|█▊        | 1357/7505 [18:18<1:07:43,  1.51it/s]
 18%|█▊        | 1358/7505 [18:19<1:18:19,  1.31it/s]
 18%|█▊        | 1359/7505 [18:20<1:14:48,  1.37it/s]
 18%|█▊        | 1360/7505 [18:20<1:12:21,  1.42it/s]
                                                     
{'loss': '4.825', 'grad_norm': '0.2879', 'learning_rate': '0.0006', 'epoch': '71.59'}
+
 18%|█▊        | 1360/7505 [18:20<1:12:21,  1.42it/s]
 18%|█▊        | 1361/7505 [18:21<1:10:43,  1.45it/s]
 18%|█▊        | 1362/7505 [18:21<1:10:09,  1.46it/s]
 18%|█▊        | 1363/7505 [18:22<1:09:06,  1.48it/s]
 18%|█▊        | 1364/7505 [18:23<1:08:21,  1.50it/s]
 18%|█▊        | 1365/7505 [18:23<1:07:54,  1.51it/s]
 18%|█▊        | 1366/7505 [18:24<1:08:03,  1.50it/s]
 18%|█▊        | 1367/7505 [18:25<1:07:32,  1.51it/s]
 18%|█▊        | 1368/7505 [18:25<1:00:05,  1.70it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 20.66it/s]
+
 15%|█▌        | 6/39 [00:00<00:02, 15.99it/s]
+
 21%|██        | 8/39 [00:00<00:01, 15.65it/s]
+
 26%|██▌       | 10/39 [00:00<00:01, 14.95it/s]
+
 31%|███       | 12/39 [00:00<00:01, 15.00it/s]
+
 36%|███▌      | 14/39 [00:00<00:01, 14.33it/s]
+
 41%|████      | 16/39 [00:01<00:01, 14.57it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 14.39it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 13.93it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 14.29it/s]
+
 62%|██████▏   | 24/39 [00:01<00:01, 14.15it/s]
+
 67%|██████▋   | 26/39 [00:01<00:00, 14.44it/s]
+
 72%|███████▏  | 28/39 [00:01<00:00, 14.01it/s]
+
 77%|███████▋  | 30/39 [00:02<00:00, 14.34it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 14.18it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 14.45it/s]
+
 92%|█████████▏| 36/39 [00:02<00:00, 14.00it/s]
+
 97%|█████████▋| 38/39 [00:02<00:00, 13.93it/s]
                                                     
+
                                               
{'eval_loss': '4.372', 'eval_runtime': '2.825', 'eval_samples_per_second': '864', 'eval_steps_per_second': '13.8', 'epoch': '72'}
+
 18%|█▊        | 1368/7505 [18:28<1:00:05,  1.70it/s]
+
100%|██████████| 39/39 [00:02<00:00, 13.93it/s][eval] epoch 72.0 step 1368 loss 4.3716 (best 4.3625 @ ep 60.644, 3/20 since best)
+
+
                                               
 18%|█▊        | 1369/7505 [18:29<2:31:59,  1.49s/it]
 18%|█▊        | 1370/7505 [18:29<2:06:35,  1.24s/it]
                                                     
{'loss': '4.883', 'grad_norm': '0.2724', 'learning_rate': '0.0006', 'epoch': '72.11'}
+
 18%|█▊        | 1370/7505 [18:29<2:06:35,  1.24s/it]
 18%|█▊        | 1371/7505 [18:30<1:48:37,  1.06s/it]
 18%|█▊        | 1372/7505 [18:31<1:36:10,  1.06it/s]
 18%|█▊        | 1373/7505 [18:31<1:27:33,  1.17it/s]
 18%|█▊        | 1374/7505 [18:32<1:32:01,  1.11it/s]
 18%|█▊        | 1375/7505 [18:33<1:24:30,  1.21it/s]
 18%|█▊        | 1376/7505 [18:34<1:19:06,  1.29it/s]
 18%|█▊        | 1377/7505 [18:34<1:15:19,  1.36it/s]
 18%|█▊        | 1378/7505 [18:35<1:13:08,  1.40it/s]
 18%|█▊        | 1379/7505 [18:36<1:11:04,  1.44it/s]
 18%|█▊        | 1380/7505 [18:36<1:09:41,  1.46it/s]
                                                     
{'loss': '4.892', 'grad_norm': '0.2668', 'learning_rate': '0.0006', 'epoch': '72.64'}
+
 18%|█▊        | 1380/7505 [18:36<1:09:41,  1.46it/s]
 18%|█▊        | 1381/7505 [18:37<1:09:30,  1.47it/s]
 18%|█▊        | 1382/7505 [18:38<1:08:36,  1.49it/s]
 18%|█▊        | 1383/7505 [18:38<1:07:58,  1.50it/s]
 18%|█▊        | 1384/7505 [18:39<1:07:44,  1.51it/s]
 18%|█▊        | 1385/7505 [18:40<1:08:02,  1.50it/s]
 18%|█▊        | 1386/7505 [18:40<1:07:32,  1.51it/s]
 18%|█▊        | 1387/7505 [18:41<1:00:00,  1.70it/s]
 18%|█▊        | 1388/7505 [18:41<1:04:22,  1.58it/s]
 19%|█▊        | 1389/7505 [18:42<1:05:10,  1.56it/s]
 19%|█▊        | 1390/7505 [18:43<1:06:13,  1.54it/s]
                                                     
{'loss': '4.837', 'grad_norm': '0.3377', 'learning_rate': '0.0006', 'epoch': '73.16'}
+
 19%|█▊        | 1390/7505 [18:43<1:06:13,  1.54it/s]
 19%|█▊        | 1391/7505 [18:43<1:06:17,  1.54it/s]
 19%|█▊        | 1392/7505 [18:44<1:06:30,  1.53it/s]
 19%|█▊        | 1393/7505 [18:45<1:06:23,  1.53it/s]
 19%|█▊        | 1394/7505 [18:45<1:07:05,  1.52it/s]
 19%|█▊        | 1395/7505 [18:46<1:07:00,  1.52it/s]
 19%|█▊        | 1396/7505 [18:47<1:06:47,  1.52it/s]
 19%|█▊        | 1397/7505 [18:47<1:06:54,  1.52it/s]
 19%|█▊        | 1398/7505 [18:48<1:07:19,  1.51it/s]
 19%|█▊        | 1399/7505 [18:49<1:07:02,  1.52it/s]
 19%|█▊        | 1400/7505 [18:49<1:06:48,  1.52it/s]
                                                     
{'loss': '4.851', 'grad_norm': '0.289', 'learning_rate': '0.0006', 'epoch': '73.7'}
+
 19%|█▊        | 1400/7505 [18:49<1:06:48,  1.52it/s]
 19%|█▊        | 1401/7505 [18:50<1:07:01,  1.52it/s]
 19%|█▊        | 1402/7505 [18:51<1:06:56,  1.52it/s]
 19%|█▊        | 1403/7505 [18:51<1:06:44,  1.52it/s]
 19%|█▊        | 1404/7505 [18:52<1:06:38,  1.53it/s]
 19%|█▊        | 1405/7505 [18:53<1:07:11,  1.51it/s]
 19%|█▊        | 1406/7505 [18:53<59:33,  1.71it/s]  
 19%|█▊        | 1407/7505 [18:54<1:03:49,  1.59it/s]
 19%|█▉        | 1408/7505 [18:54<1:04:34,  1.57it/s]
 19%|█▉        | 1409/7505 [18:55<1:05:04,  1.56it/s]
 19%|█▉        | 1410/7505 [18:56<1:16:38,  1.33it/s]
                                                     
{'loss': '4.776', 'grad_norm': '0.3709', 'learning_rate': '0.0006', 'epoch': '74.21'}
+
 19%|█▉        | 1410/7505 [18:56<1:16:38,  1.33it/s]
 19%|█▉        | 1411/7505 [18:57<1:34:07,  1.08it/s]
 19%|█▉        | 1412/7505 [18:58<1:27:31,  1.16it/s]
 19%|█▉        | 1413/7505 [18:59<1:21:02,  1.25it/s]
 19%|█▉        | 1414/7505 [18:59<1:16:31,  1.33it/s]
 19%|█▉        | 1415/7505 [19:00<1:17:35,  1.31it/s]
 19%|█▉        | 1416/7505 [19:02<1:35:24,  1.06it/s]
 19%|█▉        | 1417/7505 [19:03<1:47:59,  1.06s/it]
 19%|█▉        | 1418/7505 [19:04<1:43:10,  1.02s/it]
 19%|█▉        | 1419/7505 [19:05<1:34:17,  1.08it/s]
 19%|█▉        | 1420/7505 [19:05<1:25:53,  1.18it/s]
                                                     
{'loss': '4.915', 'grad_norm': '0.4502', 'learning_rate': '0.0006', 'epoch': '74.75'}
+
 19%|█▉        | 1420/7505 [19:05<1:25:53,  1.18it/s]
 19%|█▉        | 1421/7505 [19:06<1:20:06,  1.27it/s]
 19%|█▉        | 1422/7505 [19:07<1:16:38,  1.32it/s]
 19%|█▉        | 1423/7505 [19:07<1:13:38,  1.38it/s]
 19%|█▉        | 1424/7505 [19:08<1:11:39,  1.41it/s]
 19%|█▉        | 1425/7505 [19:08<1:03:15,  1.60it/s]
 19%|█▉        | 1426/7505 [19:09<1:06:49,  1.52it/s]
 19%|█▉        | 1427/7505 [19:10<1:07:25,  1.50it/s]
 19%|█▉        | 1428/7505 [19:10<1:07:13,  1.51it/s]
 19%|█▉        | 1429/7505 [19:11<1:06:54,  1.51it/s]
 19%|█▉        | 1430/7505 [19:12<1:07:28,  1.50it/s]
                                                     
{'loss': '4.904', 'grad_norm': '0.3192', 'learning_rate': '0.0006', 'epoch': '75.27'}
+
 19%|█▉        | 1430/7505 [19:12<1:07:28,  1.50it/s]
 19%|█▉        | 1431/7505 [19:12<1:07:06,  1.51it/s]
 19%|█▉        | 1432/7505 [19:13<1:06:57,  1.51it/s]
 19%|█▉        | 1433/7505 [19:14<1:06:42,  1.52it/s]
 19%|█▉        | 1434/7505 [19:14<1:07:11,  1.51it/s]
 19%|█▉        | 1435/7505 [19:15<1:07:00,  1.51it/s]
 19%|█▉        | 1436/7505 [19:16<1:06:44,  1.52it/s]
 19%|█▉        | 1437/7505 [19:16<1:06:32,  1.52it/s]
 19%|█▉        | 1438/7505 [19:17<1:07:07,  1.51it/s]
 19%|█▉        | 1439/7505 [19:18<1:06:53,  1.51it/s]
 19%|█▉        | 1440/7505 [19:18<1:06:38,  1.52it/s]
                                                     
{'loss': '4.838', 'grad_norm': '0.2737', 'learning_rate': '0.0006', 'epoch': '75.81'}
+
 19%|█▉        | 1440/7505 [19:18<1:06:38,  1.52it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 19.12it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 16.95it/s]
+
 18%|█▊        | 7/39 [00:00<00:01, 16.18it/s]
+
 23%|██▎       | 9/39 [00:00<00:01, 15.77it/s]
+
 28%|██▊       | 11/39 [00:00<00:01, 14.61it/s]
+
 33%|███▎      | 13/39 [00:00<00:01, 14.82it/s]
+
 38%|███▊      | 15/39 [00:00<00:01, 14.94it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 14.57it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 13.97it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 14.30it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 14.58it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 14.00it/s]
+
 69%|██████▉   | 27/39 [00:01<00:00, 14.34it/s]
+
 74%|███████▍  | 29/39 [00:01<00:00, 14.58it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 14.76it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 13.71it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 14.13it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 14.43it/s]
+
100%|██████████| 39/39 [00:02<00:00, 14.23it/s]
                                                     
+
                                               
{'eval_loss': '4.388', 'eval_runtime': '2.826', 'eval_samples_per_second': '863.8', 'eval_steps_per_second': '13.8', 'epoch': '75.81'}
+
 19%|█▉        | 1440/7505 [19:21<1:06:38,  1.52it/s]
+
100%|██████████| 39/39 [00:02<00:00, 14.23it/s][eval] epoch 75.805 step 1440 loss 4.3881 (best 4.3625 @ ep 60.644, 4/20 since best)
+
+
                                               
 19%|█▉        | 1441/7505 [19:22<2:32:15,  1.51s/it]
 19%|█▉        | 1442/7505 [19:22<2:06:23,  1.25s/it]
 19%|█▉        | 1443/7505 [19:23<1:48:23,  1.07s/it]
 19%|█▉        | 1444/7505 [19:24<1:28:36,  1.14it/s]
 19%|█▉        | 1445/7505 [19:24<1:25:32,  1.18it/s]
 19%|█▉        | 1446/7505 [19:25<1:19:36,  1.27it/s]
 19%|█▉        | 1447/7505 [19:26<1:15:41,  1.33it/s]
 19%|█▉        | 1448/7505 [19:26<1:12:48,  1.39it/s]
 19%|█▉        | 1449/7505 [19:27<1:11:34,  1.41it/s]
 19%|█▉        | 1450/7505 [19:28<1:10:05,  1.44it/s]
                                                     
{'loss': '4.882', 'grad_norm': '0.3237', 'learning_rate': '0.0006', 'epoch': '76.32'}
+
 19%|█▉        | 1450/7505 [19:28<1:10:05,  1.44it/s]
 19%|█▉        | 1451/7505 [19:28<1:08:52,  1.47it/s]
 19%|█▉        | 1452/7505 [19:29<1:08:27,  1.47it/s]
 19%|█▉        | 1453/7505 [19:30<1:08:31,  1.47it/s]
 19%|█▉        | 1454/7505 [19:30<1:07:44,  1.49it/s]
 19%|█▉        | 1455/7505 [19:31<1:07:09,  1.50it/s]
 19%|█▉        | 1456/7505 [19:32<1:10:19,  1.43it/s]
 19%|█▉        | 1457/7505 [19:33<1:29:47,  1.12it/s]
 19%|█▉        | 1458/7505 [19:34<1:29:41,  1.12it/s]
 19%|█▉        | 1459/7505 [19:35<1:22:35,  1.22it/s]
 19%|█▉        | 1460/7505 [19:36<1:38:18,  1.02it/s]
                                                     
{'loss': '4.783', 'grad_norm': '0.3192', 'learning_rate': '0.0006', 'epoch': '76.86'}
+
 19%|█▉        | 1460/7505 [19:36<1:38:18,  1.02it/s]
 19%|█▉        | 1461/7505 [19:37<1:42:53,  1.02s/it]
 19%|█▉        | 1462/7505 [19:38<1:33:18,  1.08it/s]
 19%|█▉        | 1463/7505 [19:39<1:30:13,  1.12it/s]
 20%|█▉        | 1464/7505 [19:40<1:45:04,  1.04s/it]
 20%|█▉        | 1465/7505 [19:41<1:33:33,  1.08it/s]
 20%|█▉        | 1466/7505 [19:42<1:43:07,  1.02s/it]
 20%|█▉        | 1467/7505 [19:43<1:50:09,  1.09s/it]
 20%|█▉        | 1468/7505 [19:44<1:36:53,  1.04it/s]
 20%|█▉        | 1469/7505 [19:45<1:45:22,  1.05s/it]
 20%|█▉        | 1470/7505 [19:46<1:51:51,  1.11s/it]
                                                     
{'loss': '4.863', 'grad_norm': '0.2739', 'learning_rate': '0.0006', 'epoch': '77.38'}
+
 20%|█▉        | 1470/7505 [19:46<1:51:51,  1.11s/it]
 20%|█▉        | 1471/7505 [19:47<1:38:03,  1.03it/s]
 20%|█▉        | 1472/7505 [19:48<1:28:23,  1.14it/s]
 20%|█▉        | 1473/7505 [19:48<1:21:32,  1.23it/s]
 20%|█▉        | 1474/7505 [19:49<1:26:54,  1.16it/s]
 20%|█▉        | 1475/7505 [19:51<1:40:29,  1.00it/s]
 20%|█▉        | 1476/7505 [19:51<1:34:47,  1.06it/s]
 20%|█▉        | 1477/7505 [19:52<1:26:07,  1.17it/s]
 20%|█▉        | 1478/7505 [19:53<1:20:47,  1.24it/s]
 20%|█▉        | 1479/7505 [19:53<1:16:16,  1.32it/s]
 20%|█▉        | 1480/7505 [19:54<1:13:12,  1.37it/s]
                                                     
{'loss': '4.774', 'grad_norm': '0.2929', 'learning_rate': '0.0006', 'epoch': '77.91'}
+
 20%|█▉        | 1480/7505 [19:54<1:13:12,  1.37it/s]
 20%|█▉        | 1481/7505 [19:55<1:11:00,  1.41it/s]
 20%|█▉        | 1482/7505 [19:55<1:07:05,  1.50it/s]
 20%|█▉        | 1483/7505 [19:57<1:30:40,  1.11it/s]
 20%|█▉        | 1484/7505 [19:58<1:31:35,  1.10it/s]
 20%|█▉        | 1485/7505 [19:59<1:30:23,  1.11it/s]
 20%|█▉        | 1486/7505 [20:00<1:42:24,  1.02s/it]
 20%|█▉        | 1487/7505 [20:00<1:31:58,  1.09it/s]
 20%|█▉        | 1488/7505 [20:01<1:23:56,  1.19it/s]
 20%|█▉        | 1489/7505 [20:02<1:18:18,  1.28it/s]
 20%|█▉        | 1490/7505 [20:02<1:15:07,  1.33it/s]
                                                     
{'loss': '4.765', 'grad_norm': '0.2776', 'learning_rate': '0.0006', 'epoch': '78.43'}
+
 20%|█▉        | 1490/7505 [20:02<1:15:07,  1.33it/s]
 20%|█▉        | 1491/7505 [20:03<1:12:13,  1.39it/s]
 20%|█▉        | 1492/7505 [20:04<1:10:10,  1.43it/s]
 20%|█▉        | 1493/7505 [20:04<1:08:57,  1.45it/s]
 20%|█▉        | 1494/7505 [20:05<1:08:37,  1.46it/s]
 20%|█▉        | 1495/7505 [20:06<1:07:40,  1.48it/s]
 20%|█▉        | 1496/7505 [20:06<1:07:12,  1.49it/s]
 20%|█▉        | 1497/7505 [20:07<1:06:39,  1.50it/s]
 20%|█▉        | 1498/7505 [20:08<1:07:00,  1.49it/s]
 20%|█▉        | 1499/7505 [20:08<1:06:45,  1.50it/s]
 20%|█▉        | 1500/7505 [20:09<1:06:13,  1.51it/s]
                                                     
{'loss': '4.762', 'grad_norm': '0.2955', 'learning_rate': '0.0006', 'epoch': '78.97'}
+
 20%|█▉        | 1500/7505 [20:09<1:06:13,  1.51it/s]
 20%|██        | 1501/7505 [20:09<58:57,  1.70it/s]  
 20%|██        | 1502/7505 [20:10<1:03:15,  1.58it/s]
 20%|██        | 1503/7505 [20:11<1:04:36,  1.55it/s]
 20%|██        | 1504/7505 [20:12<1:05:03,  1.54it/s]
 20%|██        | 1505/7505 [20:12<1:05:03,  1.54it/s]
 20%|██        | 1506/7505 [20:13<1:05:50,  1.52it/s]
 20%|██        | 1507/7505 [20:14<1:05:44,  1.52it/s]
 20%|██        | 1508/7505 [20:14<1:05:35,  1.52it/s]
 20%|██        | 1509/7505 [20:15<1:05:27,  1.53it/s]
 20%|██        | 1510/7505 [20:16<1:06:06,  1.51it/s]
                                                     
{'loss': '4.763', 'grad_norm': '0.3928', 'learning_rate': '0.0006', 'epoch': '79.48'}
+
 20%|██        | 1510/7505 [20:16<1:06:06,  1.51it/s]
 20%|██        | 1511/7505 [20:17<1:17:50,  1.28it/s]
 20%|██        | 1512/7505 [20:18<1:33:28,  1.07it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:02, 13.76it/s]
+
 10%|█         | 4/39 [00:00<00:02, 13.67it/s]
+
 15%|█▌        | 6/39 [00:00<00:03, 10.64it/s]
+
 21%|██        | 8/39 [00:00<00:03,  8.98it/s]
+
 23%|██▎       | 9/39 [00:00<00:03,  8.55it/s]
+
 26%|██▌       | 10/39 [00:01<00:03,  8.22it/s]
+
 28%|██▊       | 11/39 [00:01<00:03,  7.95it/s]
+
 31%|███       | 12/39 [00:01<00:03,  7.77it/s]
+
 33%|███▎      | 13/39 [00:01<00:03,  7.62it/s]
+
 36%|███▌      | 14/39 [00:01<00:03,  7.53it/s]
+
 38%|███▊      | 15/39 [00:01<00:03,  7.48it/s]
+
 41%|████      | 16/39 [00:01<00:03,  7.41it/s]
+
 44%|████▎     | 17/39 [00:02<00:02,  7.34it/s]
+
 46%|████▌     | 18/39 [00:02<00:02,  7.31it/s]
+
 49%|████▊     | 19/39 [00:02<00:02,  7.24it/s]
+
 51%|█████▏    | 20/39 [00:02<00:02,  7.26it/s]
+
 54%|█████▍    | 21/39 [00:02<00:02,  7.22it/s]
+
 56%|█████▋    | 22/39 [00:02<00:02,  7.20it/s]
+
 62%|██████▏   | 24/39 [00:02<00:01,  8.97it/s]
+
 67%|██████▋   | 26/39 [00:03<00:01, 10.67it/s]
+
 72%|███████▏  | 28/39 [00:03<00:00, 11.97it/s]
+
 77%|███████▋  | 30/39 [00:03<00:00, 12.31it/s]
+
 82%|████████▏ | 32/39 [00:03<00:00, 13.13it/s]
+
 87%|████████▋ | 34/39 [00:03<00:00, 13.70it/s]
+
 92%|█████████▏| 36/39 [00:03<00:00, 14.14it/s]
+
 97%|█████████▋| 38/39 [00:03<00:00, 13.68it/s]
                                                     
+
                                               
{'eval_loss': '4.423', 'eval_runtime': '4.218', 'eval_samples_per_second': '578.8', 'eval_steps_per_second': '9.247', 'epoch': '79.59'}
+
 20%|██        | 1512/7505 [20:22<1:33:28,  1.07it/s]
+
100%|██████████| 39/39 [00:04<00:00, 13.68it/s][eval] epoch 79.591 step 1512 loss 4.4226 (best 4.3625 @ ep 60.644, 5/20 since best)
+
+
                                               
 20%|██        | 1513/7505 [20:23<3:51:31,  2.32s/it]
 20%|██        | 1514/7505 [20:24<3:11:29,  1.92s/it]
 20%|██        | 1515/7505 [20:25<2:39:30,  1.60s/it]
 20%|██        | 1516/7505 [20:27<2:30:25,  1.51s/it]
 20%|██        | 1517/7505 [20:28<2:25:43,  1.46s/it]
 20%|██        | 1518/7505 [20:29<2:22:37,  1.43s/it]
 20%|██        | 1519/7505 [20:30<2:06:32,  1.27s/it]
 20%|██        | 1520/7505 [20:31<1:43:07,  1.03s/it]
                                                     
{'loss': '4.816', 'grad_norm': '0.35', 'learning_rate': '0.0006', 'epoch': '80'}
+
 20%|██        | 1520/7505 [20:31<1:43:07,  1.03s/it]
 20%|██        | 1521/7505 [20:32<1:55:13,  1.16s/it]
 20%|██        | 1522/7505 [20:33<2:01:17,  1.22s/it]
 20%|██        | 1523/7505 [20:35<1:57:16,  1.18s/it]
 20%|██        | 1524/7505 [20:35<1:41:21,  1.02s/it]
 20%|██        | 1525/7505 [20:36<1:32:19,  1.08it/s]
 20%|██        | 1526/7505 [20:37<1:24:43,  1.18it/s]
 20%|██        | 1527/7505 [20:37<1:18:44,  1.27it/s]
 20%|██        | 1528/7505 [20:38<1:14:28,  1.34it/s]
 20%|██        | 1529/7505 [20:39<1:11:52,  1.39it/s]
 20%|██        | 1530/7505 [20:39<1:10:25,  1.41it/s]
                                                     
{'loss': '4.751', 'grad_norm': '0.3203', 'learning_rate': '0.0006', 'epoch': '80.54'}
+
 20%|██        | 1530/7505 [20:39<1:10:25,  1.41it/s]
 20%|██        | 1531/7505 [20:40<1:08:41,  1.45it/s]
 20%|██        | 1532/7505 [20:41<1:07:49,  1.47it/s]
 20%|██        | 1533/7505 [20:41<1:07:35,  1.47it/s]
 20%|██        | 1534/7505 [20:42<1:06:42,  1.49it/s]
 20%|██        | 1535/7505 [20:42<1:06:02,  1.51it/s]
 20%|██        | 1536/7505 [20:44<1:19:34,  1.25it/s]
 20%|██        | 1537/7505 [20:45<1:36:07,  1.03it/s]
 20%|██        | 1538/7505 [20:46<1:40:42,  1.01s/it]
 21%|██        | 1539/7505 [20:46<1:22:37,  1.20it/s]
 21%|██        | 1540/7505 [20:47<1:21:50,  1.21it/s]
                                                     
{'loss': '4.784', 'grad_norm': '0.3225', 'learning_rate': '0.0006', 'epoch': '81.05'}
+
 21%|██        | 1540/7505 [20:47<1:21:50,  1.21it/s]
 21%|██        | 1541/7505 [20:48<1:16:44,  1.30it/s]
 21%|██        | 1542/7505 [20:49<1:13:40,  1.35it/s]
 21%|██        | 1543/7505 [20:49<1:11:00,  1.40it/s]
 21%|██        | 1544/7505 [20:50<1:09:07,  1.44it/s]
 21%|██        | 1545/7505 [20:51<1:07:50,  1.46it/s]
 21%|██        | 1546/7505 [20:51<1:07:21,  1.47it/s]
 21%|██        | 1547/7505 [20:52<1:06:33,  1.49it/s]
 21%|██        | 1548/7505 [20:53<1:06:01,  1.50it/s]
 21%|██        | 1549/7505 [20:53<1:06:05,  1.50it/s]
 21%|██        | 1550/7505 [20:54<1:05:49,  1.51it/s]
                                                     
{'loss': '4.686', 'grad_norm': '0.3572', 'learning_rate': '0.0006', 'epoch': '81.59'}
+
 21%|██        | 1550/7505 [20:54<1:05:49,  1.51it/s]
 21%|██        | 1551/7505 [20:55<1:05:30,  1.51it/s]
 21%|██        | 1552/7505 [20:55<1:05:14,  1.52it/s]
 21%|██        | 1553/7505 [20:56<1:05:34,  1.51it/s]
 21%|██        | 1554/7505 [20:56<1:05:18,  1.52it/s]
 21%|██        | 1555/7505 [20:57<1:05:06,  1.52it/s]
 21%|██        | 1556/7505 [20:58<1:05:21,  1.52it/s]
 21%|██        | 1557/7505 [20:58<1:05:15,  1.52it/s]
 21%|██        | 1558/7505 [20:59<58:04,  1.71it/s]  
 21%|██        | 1559/7505 [21:00<1:02:27,  1.59it/s]
 21%|██        | 1560/7505 [21:00<1:03:02,  1.57it/s]
                                                     
{'loss': '4.748', 'grad_norm': '0.2917', 'learning_rate': '0.0006', 'epoch': '82.11'}
+
 21%|██        | 1560/7505 [21:00<1:03:02,  1.57it/s]
 21%|██        | 1561/7505 [21:01<1:20:40,  1.23it/s]
 21%|██        | 1562/7505 [21:03<1:36:19,  1.03it/s]
 21%|██        | 1563/7505 [21:04<1:30:35,  1.09it/s]
 21%|██        | 1564/7505 [21:05<1:43:33,  1.05s/it]
 21%|██        | 1565/7505 [21:06<1:52:21,  1.13s/it]
 21%|██        | 1566/7505 [21:08<1:58:59,  1.20s/it]
 21%|██        | 1567/7505 [21:09<2:02:29,  1.24s/it]
 21%|██        | 1568/7505 [21:10<2:06:00,  1.27s/it]
 21%|██        | 1569/7505 [21:12<2:08:58,  1.30s/it]
 21%|██        | 1570/7505 [21:13<2:02:12,  1.24s/it]
                                                     
{'loss': '4.705', 'grad_norm': '0.3103', 'learning_rate': '0.0006', 'epoch': '82.64'}
+
 21%|██        | 1570/7505 [21:13<2:02:12,  1.24s/it]
 21%|██        | 1571/7505 [21:14<1:57:40,  1.19s/it]
 21%|██        | 1572/7505 [21:15<2:02:14,  1.24s/it]
 21%|██        | 1573/7505 [21:17<2:05:52,  1.27s/it]
 21%|██        | 1574/7505 [21:18<2:04:45,  1.26s/it]
 21%|██        | 1575/7505 [21:19<2:07:21,  1.29s/it]
 21%|██        | 1576/7505 [21:21<2:09:28,  1.31s/it]
 21%|██        | 1577/7505 [21:21<1:56:04,  1.17s/it]
 21%|██        | 1578/7505 [21:22<1:45:18,  1.07s/it]
 21%|██        | 1579/7505 [21:23<1:33:37,  1.05it/s]
 21%|██        | 1580/7505 [21:24<1:24:49,  1.16it/s]
                                                     
{'loss': '4.767', 'grad_norm': '0.3036', 'learning_rate': '0.0006', 'epoch': '83.16'}
+
 21%|██        | 1580/7505 [21:24<1:24:49,  1.16it/s]
 21%|██        | 1581/7505 [21:24<1:18:58,  1.25it/s]
 21%|██        | 1582/7505 [21:25<1:14:32,  1.32it/s]
 21%|██        | 1583/7505 [21:26<1:12:04,  1.37it/s]
 21%|██        | 1584/7505 [21:26<1:09:42,  1.42it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 23.15it/s]
+
 15%|█▌        | 6/39 [00:00<00:01, 16.59it/s]
+
 21%|██        | 8/39 [00:00<00:02, 15.45it/s]
+
 26%|██▌       | 10/39 [00:00<00:01, 15.33it/s]
+
 31%|███       | 12/39 [00:00<00:01, 14.57it/s]
+
 36%|███▌      | 14/39 [00:00<00:01, 14.76it/s]
+
 41%|████      | 16/39 [00:01<00:01, 14.32it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 14.58it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 14.10it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 14.42it/s]
+
 62%|██████▏   | 24/39 [00:01<00:01, 14.14it/s]
+
 67%|██████▋   | 26/39 [00:01<00:00, 14.45it/s]
+
 72%|███████▏  | 28/39 [00:01<00:00, 14.02it/s]
+
 77%|███████▋  | 30/39 [00:02<00:00, 13.95it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 14.30it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 14.56it/s]
+
 92%|█████████▏| 36/39 [00:02<00:00, 14.10it/s]
+
 97%|█████████▋| 38/39 [00:02<00:00, 13.92it/s]
                                                     
+
                                               
{'eval_loss': '4.438', 'eval_runtime': '2.802', 'eval_samples_per_second': '871.3', 'eval_steps_per_second': '13.92', 'epoch': '83.38'}
+
 21%|██        | 1584/7505 [21:29<1:09:42,  1.42it/s]
+
100%|██████████| 39/39 [00:02<00:00, 13.92it/s][eval] epoch 83.376 step 1584 loss 4.4381 (best 4.3625 @ ep 60.644, 6/20 since best)
+
+
                                               
 21%|██        | 1585/7505 [21:30<2:31:35,  1.54s/it]
 21%|██        | 1586/7505 [21:30<2:05:42,  1.27s/it]
 21%|██        | 1587/7505 [21:31<1:57:42,  1.19s/it]
 21%|██        | 1588/7505 [21:33<2:01:52,  1.24s/it]
 21%|██        | 1589/7505 [21:34<1:59:21,  1.21s/it]
 21%|██        | 1590/7505 [21:35<1:45:21,  1.07s/it]
                                                     
{'loss': '4.801', 'grad_norm': '0.3766', 'learning_rate': '0.0006', 'epoch': '83.7'}
+
 21%|██        | 1590/7505 [21:35<1:45:21,  1.07s/it]
 21%|██        | 1591/7505 [21:36<1:53:53,  1.16s/it]
 21%|██        | 1592/7505 [21:37<1:59:18,  1.21s/it]
 21%|██        | 1593/7505 [21:38<1:55:51,  1.18s/it]
 21%|██        | 1594/7505 [21:40<2:00:39,  1.22s/it]
 21%|██▏       | 1595/7505 [21:41<2:04:30,  1.26s/it]
 21%|██▏       | 1596/7505 [21:42<1:52:29,  1.14s/it]
 21%|██▏       | 1597/7505 [21:43<1:57:04,  1.19s/it]
 21%|██▏       | 1598/7505 [21:44<2:01:25,  1.23s/it]
 21%|██▏       | 1599/7505 [21:46<2:05:06,  1.27s/it]
 21%|██▏       | 1600/7505 [21:47<2:01:34,  1.24s/it]
                                                     
{'loss': '4.735', 'grad_norm': '0.3869', 'learning_rate': '0.0006', 'epoch': '84.21'}
+
 21%|██▏       | 1600/7505 [21:47<2:01:34,  1.24s/it]
 21%|██▏       | 1601/7505 [21:48<1:55:31,  1.17s/it]
 21%|██▏       | 1602/7505 [21:49<2:00:23,  1.22s/it]
 21%|██▏       | 1603/7505 [21:51<2:04:22,  1.26s/it]
 21%|██▏       | 1604/7505 [21:52<2:03:25,  1.25s/it]
 21%|██▏       | 1605/7505 [21:53<2:05:11,  1.27s/it]
 21%|██▏       | 1606/7505 [21:55<2:07:15,  1.29s/it]
 21%|██▏       | 1607/7505 [21:56<2:08:58,  1.31s/it]
 21%|██▏       | 1608/7505 [21:57<1:59:55,  1.22s/it]
 21%|██▏       | 1609/7505 [21:58<1:53:22,  1.15s/it]
 21%|██▏       | 1610/7505 [21:59<1:59:23,  1.22s/it]
                                                     
{'loss': '4.778', 'grad_norm': '0.3143', 'learning_rate': '0.0006', 'epoch': '84.75'}
+
 21%|██▏       | 1610/7505 [21:59<1:59:23,  1.22s/it]
 21%|██▏       | 1611/7505 [22:01<2:04:03,  1.26s/it]
 21%|██▏       | 1612/7505 [22:02<2:02:34,  1.25s/it]
 21%|██▏       | 1613/7505 [22:03<2:05:45,  1.28s/it]
 22%|██▏       | 1614/7505 [22:05<2:07:59,  1.30s/it]
 22%|██▏       | 1615/7505 [22:05<1:48:30,  1.11s/it]
 22%|██▏       | 1616/7505 [22:06<1:36:56,  1.01it/s]
 22%|██▏       | 1617/7505 [22:07<1:29:29,  1.10it/s]
 22%|██▏       | 1618/7505 [22:07<1:21:48,  1.20it/s]
 22%|██▏       | 1619/7505 [22:08<1:16:23,  1.28it/s]
 22%|██▏       | 1620/7505 [22:09<1:12:36,  1.35it/s]
                                                     
{'loss': '4.706', 'grad_norm': '0.3238', 'learning_rate': '0.0006', 'epoch': '85.27'}
+
 22%|██▏       | 1620/7505 [22:09<1:12:36,  1.35it/s]
 22%|██▏       | 1621/7505 [22:09<1:10:33,  1.39it/s]
 22%|██▏       | 1622/7505 [22:10<1:08:34,  1.43it/s]
 22%|██▏       | 1623/7505 [22:11<1:07:10,  1.46it/s]
 22%|██▏       | 1624/7505 [22:11<1:06:09,  1.48it/s]
 22%|██▏       | 1625/7505 [22:12<1:06:02,  1.48it/s]
 22%|██▏       | 1626/7505 [22:13<1:17:51,  1.26it/s]
 22%|██▏       | 1627/7505 [22:14<1:34:41,  1.03it/s]
 22%|██▏       | 1628/7505 [22:16<1:45:28,  1.08s/it]
 22%|██▏       | 1629/7505 [22:17<1:52:27,  1.15s/it]
 22%|██▏       | 1630/7505 [22:18<1:58:44,  1.21s/it]
                                                     
{'loss': '4.7', 'grad_norm': '0.3188', 'learning_rate': '0.0006', 'epoch': '85.81'}
+
 22%|██▏       | 1630/7505 [22:18<1:58:44,  1.21s/it]
 22%|██▏       | 1631/7505 [22:20<2:02:34,  1.25s/it]
 22%|██▏       | 1632/7505 [22:21<1:58:51,  1.21s/it]
 22%|██▏       | 1633/7505 [22:22<2:02:19,  1.25s/it]
 22%|██▏       | 1634/7505 [22:23<1:50:10,  1.13s/it]
 22%|██▏       | 1635/7505 [22:24<1:58:15,  1.21s/it]
 22%|██▏       | 1636/7505 [22:26<2:01:39,  1.24s/it]
 22%|██▏       | 1637/7505 [22:27<2:01:19,  1.24s/it]
 22%|██▏       | 1638/7505 [22:28<2:04:48,  1.28s/it]
 22%|██▏       | 1639/7505 [22:30<2:07:42,  1.31s/it]
 22%|██▏       | 1640/7505 [22:31<2:08:07,  1.31s/it]
                                                     
{'loss': '4.725', 'grad_norm': '0.3407', 'learning_rate': '0.0006', 'epoch': '86.32'}
+
 22%|██▏       | 1640/7505 [22:31<2:08:07,  1.31s/it]
 22%|██▏       | 1641/7505 [22:32<2:09:11,  1.32s/it]
 22%|██▏       | 1642/7505 [22:34<2:09:57,  1.33s/it]
 22%|██▏       | 1643/7505 [22:35<2:09:25,  1.32s/it]
 22%|██▏       | 1644/7505 [22:36<1:53:52,  1.17s/it]
 22%|██▏       | 1645/7505 [22:37<1:59:25,  1.22s/it]
 22%|██▏       | 1646/7505 [22:39<2:03:39,  1.27s/it]
 22%|██▏       | 1647/7505 [22:40<1:58:46,  1.22s/it]
 22%|██▏       | 1648/7505 [22:41<2:01:42,  1.25s/it]
 22%|██▏       | 1649/7505 [22:42<2:04:46,  1.28s/it]
 22%|██▏       | 1650/7505 [22:44<2:08:02,  1.31s/it]
                                                     
{'loss': '4.679', 'grad_norm': '0.2698', 'learning_rate': '0.0006', 'epoch': '86.86'}
+
 22%|██▏       | 1650/7505 [22:44<2:08:02,  1.31s/it]
 22%|██▏       | 1651/7505 [22:45<2:08:06,  1.31s/it]
 22%|██▏       | 1652/7505 [22:46<2:00:17,  1.23s/it]
 22%|██▏       | 1653/7505 [22:47<1:49:06,  1.12s/it]
 22%|██▏       | 1654/7505 [22:48<1:57:53,  1.21s/it]
 22%|██▏       | 1655/7505 [22:49<1:51:24,  1.14s/it]
 22%|██▏       | 1656/7505 [22:50<1:39:12,  1.02s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:02, 14.98it/s]
+
 10%|█         | 4/39 [00:00<00:03,  8.93it/s]
+
 15%|█▌        | 6/39 [00:00<00:04,  7.93it/s]
+
 18%|█▊        | 7/39 [00:00<00:04,  7.64it/s]
+
 21%|██        | 8/39 [00:00<00:04,  7.47it/s]
+
 23%|██▎       | 9/39 [00:01<00:04,  7.29it/s]
+
 26%|██▌       | 10/39 [00:01<00:04,  7.16it/s]
+
 28%|██▊       | 11/39 [00:01<00:03,  7.11it/s]
+
 31%|███       | 12/39 [00:01<00:03,  7.06it/s]
+
 33%|███▎      | 13/39 [00:01<00:03,  7.02it/s]
+
 36%|███▌      | 14/39 [00:01<00:03,  7.01it/s]
+
 38%|███▊      | 15/39 [00:02<00:03,  6.99it/s]
+
 41%|████      | 16/39 [00:02<00:03,  6.96it/s]
+
 44%|████▎     | 17/39 [00:02<00:03,  6.97it/s]
+
 46%|████▌     | 18/39 [00:02<00:03,  6.96it/s]
+
 49%|████▊     | 19/39 [00:02<00:02,  6.93it/s]
+
 51%|█████▏    | 20/39 [00:02<00:02,  6.93it/s]
+
 54%|█████▍    | 21/39 [00:02<00:02,  6.92it/s]
+
 56%|█████▋    | 22/39 [00:03<00:02,  6.91it/s]
+
 59%|█████▉    | 23/39 [00:03<00:02,  6.92it/s]
+
 62%|██████▏   | 24/39 [00:03<00:02,  6.92it/s]
+
 64%|██████▍   | 25/39 [00:03<00:02,  6.90it/s]
+
 67%|██████▋   | 26/39 [00:03<00:01,  6.85it/s]
+
 69%|██████▉   | 27/39 [00:03<00:01,  6.85it/s]
+
 72%|███████▏  | 28/39 [00:03<00:01,  6.97it/s]
+
 74%|███████▍  | 29/39 [00:04<00:01,  7.03it/s]
+
 77%|███████▋  | 30/39 [00:04<00:01,  7.08it/s]
+
 79%|███████▉  | 31/39 [00:04<00:01,  7.09it/s]
+
 82%|████████▏ | 32/39 [00:04<00:00,  7.10it/s]
+
 85%|████████▍ | 33/39 [00:04<00:00,  7.09it/s]
+
 87%|████████▋ | 34/39 [00:04<00:00,  7.04it/s]
+
 90%|████████▉ | 35/39 [00:04<00:00,  6.98it/s]
+
 92%|█████████▏| 36/39 [00:05<00:00,  6.93it/s]
+
 95%|█████████▍| 37/39 [00:05<00:00,  6.90it/s]
+
 97%|█████████▋| 38/39 [00:05<00:00,  6.88it/s]
+
100%|██████████| 39/39 [00:05<00:00,  6.85it/s]
                                                     
+
                                               
{'eval_loss': '4.467', 'eval_runtime': '5.663', 'eval_samples_per_second': '431', 'eval_steps_per_second': '6.887', 'epoch': '87.16'}
+
 22%|██▏       | 1656/7505 [22:56<1:39:12,  1.02s/it]
+
100%|██████████| 39/39 [00:05<00:00,  6.85it/s][eval] epoch 87.161 step 1656 loss 4.4671 (best 4.3625 @ ep 60.644, 7/20 since best)
+
+
                                               
 22%|██▏       | 1657/7505 [22:57<4:34:18,  2.81s/it]
 22%|██▏       | 1658/7505 [22:58<3:50:49,  2.37s/it]
 22%|██▏       | 1659/7505 [23:00<3:17:48,  2.03s/it]
 22%|██▏       | 1660/7505 [23:01<2:56:43,  1.81s/it]
                                                     
{'loss': '4.673', 'grad_norm': '0.3355', 'learning_rate': '0.0006', 'epoch': '87.38'}
+
 22%|██▏       | 1660/7505 [23:01<2:56:43,  1.81s/it]
 22%|██▏       | 1661/7505 [23:02<2:37:37,  1.62s/it]
 22%|██▏       | 1662/7505 [23:03<2:09:38,  1.33s/it]
 22%|██▏       | 1663/7505 [23:04<2:09:27,  1.33s/it]
 22%|██▏       | 1664/7505 [23:05<2:06:19,  1.30s/it]
 22%|██▏       | 1665/7505 [23:07<2:05:32,  1.29s/it]
 22%|██▏       | 1666/7505 [23:08<2:04:25,  1.28s/it]
 22%|██▏       | 1667/7505 [23:09<1:46:04,  1.09s/it]
 22%|██▏       | 1668/7505 [23:10<1:51:31,  1.15s/it]
 22%|██▏       | 1669/7505 [23:11<1:54:30,  1.18s/it]
 22%|██▏       | 1670/7505 [23:12<1:56:26,  1.20s/it]
                                                     
{'loss': '4.731', 'grad_norm': '0.3388', 'learning_rate': '0.0006', 'epoch': '87.91'}
+
 22%|██▏       | 1670/7505 [23:12<1:56:26,  1.20s/it]
 22%|██▏       | 1671/7505 [23:14<1:58:23,  1.22s/it]
 22%|██▏       | 1672/7505 [23:14<1:45:17,  1.08s/it]
 22%|██▏       | 1673/7505 [23:16<1:53:22,  1.17s/it]
 22%|██▏       | 1674/7505 [23:17<1:54:35,  1.18s/it]
 22%|██▏       | 1675/7505 [23:18<1:58:10,  1.22s/it]
 22%|██▏       | 1676/7505 [23:19<1:43:59,  1.07s/it]
 22%|██▏       | 1677/7505 [23:20<1:44:42,  1.08s/it]
 22%|██▏       | 1678/7505 [23:21<1:51:22,  1.15s/it]
 22%|██▏       | 1679/7505 [23:22<1:40:10,  1.03s/it]
 22%|██▏       | 1680/7505 [23:23<1:32:25,  1.05it/s]
                                                     
{'loss': '4.747', 'grad_norm': '0.3387', 'learning_rate': '0.0006', 'epoch': '88.43'}
+
 22%|██▏       | 1680/7505 [23:23<1:32:25,  1.05it/s]
 22%|██▏       | 1681/7505 [23:24<1:44:24,  1.08s/it]
 22%|██▏       | 1682/7505 [23:25<1:48:36,  1.12s/it]
 22%|██▏       | 1683/7505 [23:26<1:34:58,  1.02it/s]
 22%|██▏       | 1684/7505 [23:27<1:40:33,  1.04s/it]
 22%|██▏       | 1685/7505 [23:29<1:48:45,  1.12s/it]
 22%|██▏       | 1686/7505 [23:30<1:50:37,  1.14s/it]
 22%|██▏       | 1687/7505 [23:31<1:56:54,  1.21s/it]
 22%|██▏       | 1688/7505 [23:33<2:01:44,  1.26s/it]
 23%|██▎       | 1689/7505 [23:34<2:03:23,  1.27s/it]
 23%|██▎       | 1690/7505 [23:35<1:57:27,  1.21s/it]
                                                     
{'loss': '4.652', 'grad_norm': '0.3867', 'learning_rate': '0.0006', 'epoch': '88.97'}
+
 23%|██▎       | 1690/7505 [23:35<1:57:27,  1.21s/it]
 23%|██▎       | 1691/7505 [23:36<1:47:03,  1.10s/it]
 23%|██▎       | 1692/7505 [23:37<1:56:07,  1.20s/it]
 23%|██▎       | 1693/7505 [23:38<1:54:12,  1.18s/it]
 23%|██▎       | 1694/7505 [23:39<1:39:35,  1.03s/it]
 23%|██▎       | 1695/7505 [23:40<1:49:01,  1.13s/it]
 23%|██▎       | 1696/7505 [23:42<1:55:56,  1.20s/it]
 23%|██▎       | 1697/7505 [23:43<1:56:52,  1.21s/it]
 23%|██▎       | 1698/7505 [23:44<2:01:09,  1.25s/it]
 23%|██▎       | 1699/7505 [23:46<2:04:22,  1.29s/it]
 23%|██▎       | 1700/7505 [23:47<2:05:15,  1.29s/it]
                                                     
{'loss': '4.677', 'grad_norm': '0.3662', 'learning_rate': '0.0006', 'epoch': '89.48'}
+
 23%|██▎       | 1700/7505 [23:47<2:05:15,  1.29s/it]
 23%|██▎       | 1701/7505 [23:48<2:02:07,  1.26s/it]
 23%|██▎       | 1702/7505 [23:50<2:04:56,  1.29s/it]
 23%|██▎       | 1703/7505 [23:51<2:07:02,  1.31s/it]
 23%|██▎       | 1704/7505 [23:52<1:57:30,  1.22s/it]
 23%|██▎       | 1705/7505 [23:53<2:01:19,  1.26s/it]
 23%|██▎       | 1706/7505 [23:55<2:04:13,  1.29s/it]
 23%|██▎       | 1707/7505 [23:56<2:06:43,  1.31s/it]
 23%|██▎       | 1708/7505 [23:57<2:06:32,  1.31s/it]
 23%|██▎       | 1709/7505 [23:59<2:07:51,  1.32s/it]
 23%|██▎       | 1710/7505 [24:00<1:54:27,  1.19s/it]
                                                     
{'loss': '4.671', 'grad_norm': '0.3343', 'learning_rate': '0.0006', 'epoch': '90'}
+
 23%|██▎       | 1710/7505 [24:00<1:54:27,  1.19s/it]
 23%|██▎       | 1711/7505 [24:01<2:02:03,  1.26s/it]
 23%|██▎       | 1712/7505 [24:02<1:44:19,  1.08s/it]
 23%|██▎       | 1713/7505 [24:03<1:50:43,  1.15s/it]
 23%|██▎       | 1714/7505 [24:04<1:57:29,  1.22s/it]
 23%|██▎       | 1715/7505 [24:05<1:41:56,  1.06s/it]
 23%|██▎       | 1716/7505 [24:06<1:43:56,  1.08s/it]
 23%|██▎       | 1717/7505 [24:07<1:51:15,  1.15s/it]
 23%|██▎       | 1718/7505 [24:08<1:47:35,  1.12s/it]
 23%|██▎       | 1719/7505 [24:09<1:33:41,  1.03it/s]
 23%|██▎       | 1720/7505 [24:10<1:41:16,  1.05s/it]
                                                     
{'loss': '4.574', 'grad_norm': '0.3438', 'learning_rate': '0.0006', 'epoch': '90.54'}
+
 23%|██▎       | 1720/7505 [24:10<1:41:16,  1.05s/it]
 23%|██▎       | 1721/7505 [24:12<1:51:15,  1.15s/it]
 23%|██▎       | 1722/7505 [24:13<1:56:56,  1.21s/it]
 23%|██▎       | 1723/7505 [24:14<1:42:41,  1.07s/it]
 23%|██▎       | 1724/7505 [24:15<1:35:43,  1.01it/s]
 23%|██▎       | 1725/7505 [24:16<1:46:28,  1.11s/it]
 23%|██▎       | 1726/7505 [24:17<1:54:23,  1.19s/it]
 23%|██▎       | 1727/7505 [24:19<1:57:45,  1.22s/it]
 23%|██▎       | 1728/7505 [24:20<1:53:24,  1.18s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:02, 14.51it/s]
+
 10%|█         | 4/39 [00:00<00:03,  8.85it/s]
+
 15%|█▌        | 6/39 [00:00<00:04,  7.77it/s]
+
 18%|█▊        | 7/39 [00:00<00:04,  7.53it/s]
+
 21%|██        | 8/39 [00:01<00:04,  7.37it/s]
+
 23%|██▎       | 9/39 [00:01<00:04,  7.23it/s]
+
 26%|██▌       | 10/39 [00:01<00:04,  7.07it/s]
+
 28%|██▊       | 11/39 [00:01<00:03,  7.00it/s]
+
 31%|███       | 12/39 [00:01<00:03,  6.98it/s]
+
 33%|███▎      | 13/39 [00:01<00:03,  6.92it/s]
+
 36%|███▌      | 14/39 [00:01<00:03,  6.91it/s]
+
 38%|███▊      | 15/39 [00:02<00:03,  6.90it/s]
+
 41%|████      | 16/39 [00:02<00:03,  6.89it/s]
+
 44%|████▎     | 17/39 [00:02<00:03,  6.88it/s]
+
 46%|████▌     | 18/39 [00:02<00:03,  6.87it/s]
+
 49%|████▊     | 19/39 [00:02<00:02,  6.87it/s]
+
 51%|█████▏    | 20/39 [00:02<00:02,  6.83it/s]
+
 54%|█████▍    | 21/39 [00:02<00:02,  6.81it/s]
+
 56%|█████▋    | 22/39 [00:03<00:02,  6.84it/s]
+
 59%|█████▉    | 23/39 [00:03<00:02,  6.89it/s]
+
 62%|██████▏   | 24/39 [00:03<00:02,  6.95it/s]
+
 64%|██████▍   | 25/39 [00:03<00:01,  7.03it/s]
+
 67%|██████▋   | 26/39 [00:03<00:01,  7.11it/s]
+
 69%|██████▉   | 27/39 [00:03<00:01,  7.13it/s]
+
 72%|███████▏  | 28/39 [00:03<00:01,  7.09it/s]
+
 74%|███████▍  | 29/39 [00:04<00:01,  7.06it/s]
+
 77%|███████▋  | 30/39 [00:04<00:01,  7.00it/s]
+
 79%|███████▉  | 31/39 [00:04<00:01,  6.96it/s]
+
 82%|████████▏ | 32/39 [00:04<00:01,  6.98it/s]
+
 85%|████████▍ | 33/39 [00:04<00:00,  6.97it/s]
+
 87%|████████▋ | 34/39 [00:04<00:00,  6.97it/s]
+
 90%|████████▉ | 35/39 [00:04<00:00,  6.98it/s]
+
 92%|█████████▏| 36/39 [00:05<00:00,  6.96it/s]
+
 95%|█████████▍| 37/39 [00:05<00:00,  6.97it/s]
+
 97%|█████████▋| 38/39 [00:05<00:00,  6.97it/s]
+
100%|██████████| 39/39 [00:05<00:00,  6.94it/s]
                                                     
+
                                               
{'eval_loss': '4.478', 'eval_runtime': '5.684', 'eval_samples_per_second': '429.5', 'eval_steps_per_second': '6.862', 'epoch': '90.97'}
+
 23%|██▎       | 1728/7505 [24:25<1:53:24,  1.18s/it]
+
100%|██████████| 39/39 [00:05<00:00,  6.94it/s][eval] epoch 90.966 step 1728 loss 4.4778 (best 4.3625 @ ep 60.644, 8/20 since best)
+
+
                                               
 23%|██▎       | 1729/7505 [24:26<4:28:17,  2.79s/it]
 23%|██▎       | 1730/7505 [24:28<3:45:02,  2.34s/it]
                                                     
{'loss': '4.629', 'grad_norm': '0.3395', 'learning_rate': '0.0006', 'epoch': '91.05'}
+
 23%|██▎       | 1730/7505 [24:28<3:45:02,  2.34s/it]
 23%|██▎       | 1731/7505 [24:29<3:17:07,  2.05s/it]
 23%|██▎       | 1732/7505 [24:30<2:57:26,  1.84s/it]
 23%|██▎       | 1733/7505 [24:32<2:39:27,  1.66s/it]
 23%|██▎       | 1734/7505 [24:33<2:29:47,  1.56s/it]
 23%|██▎       | 1735/7505 [24:34<2:24:51,  1.51s/it]
 23%|██▎       | 1736/7505 [24:36<2:20:37,  1.46s/it]
 23%|██▎       | 1737/7505 [24:37<2:10:52,  1.36s/it]
 23%|██▎       | 1738/7505 [24:37<1:52:32,  1.17s/it]
 23%|██▎       | 1739/7505 [24:39<1:57:37,  1.22s/it]
 23%|██▎       | 1740/7505 [24:40<2:01:47,  1.27s/it]
                                                     
{'loss': '4.6', 'grad_norm': '0.3734', 'learning_rate': '0.0006', 'epoch': '91.59'}
+
 23%|██▎       | 1740/7505 [24:40<2:01:47,  1.27s/it]
 23%|██▎       | 1741/7505 [24:42<2:05:10,  1.30s/it]
 23%|██▎       | 1742/7505 [24:43<2:03:15,  1.28s/it]
 23%|██▎       | 1743/7505 [24:44<2:05:51,  1.31s/it]
 23%|██▎       | 1744/7505 [24:46<2:07:53,  1.33s/it]
 23%|██▎       | 1745/7505 [24:47<1:59:21,  1.24s/it]
 23%|██▎       | 1746/7505 [24:48<2:01:26,  1.27s/it]
 23%|██▎       | 1747/7505 [24:49<2:04:22,  1.30s/it]
 23%|██▎       | 1748/7505 [24:50<1:52:25,  1.17s/it]
 23%|██▎       | 1749/7505 [24:51<1:51:59,  1.17s/it]
 23%|██▎       | 1750/7505 [24:52<1:38:39,  1.03s/it]
                                                     
{'loss': '4.664', 'grad_norm': '0.4725', 'learning_rate': '0.0006', 'epoch': '92.11'}
+
 23%|██▎       | 1750/7505 [24:52<1:38:39,  1.03s/it]
 23%|██▎       | 1751/7505 [24:53<1:49:35,  1.14s/it]
 23%|██▎       | 1752/7505 [24:55<1:55:10,  1.20s/it]
 23%|██▎       | 1753/7505 [24:56<1:59:25,  1.25s/it]
 23%|██▎       | 1754/7505 [24:57<2:01:03,  1.26s/it]
 23%|██▎       | 1755/7505 [24:59<2:03:26,  1.29s/it]
 23%|██▎       | 1756/7505 [25:00<2:04:58,  1.30s/it]
 23%|██▎       | 1757/7505 [25:01<2:05:30,  1.31s/it]
 23%|██▎       | 1758/7505 [25:03<2:05:35,  1.31s/it]
 23%|██▎       | 1759/7505 [25:04<2:07:35,  1.33s/it]
 23%|██▎       | 1760/7505 [25:05<2:08:13,  1.34s/it]
                                                     
{'loss': '4.693', 'grad_norm': '0.3343', 'learning_rate': '0.0006', 'epoch': '92.64'}
+
 23%|██▎       | 1760/7505 [25:05<2:08:13,  1.34s/it]
 23%|██▎       | 1761/7505 [25:07<2:03:52,  1.29s/it]
 23%|██▎       | 1762/7505 [25:08<2:04:48,  1.30s/it]
 23%|██▎       | 1763/7505 [25:09<2:06:26,  1.32s/it]
 24%|██▎       | 1764/7505 [25:11<2:08:35,  1.34s/it]
 24%|██▎       | 1765/7505 [25:12<2:00:26,  1.26s/it]
 24%|██▎       | 1766/7505 [25:13<1:51:27,  1.17s/it]
 24%|██▎       | 1767/7505 [25:14<1:42:56,  1.08s/it]
 24%|██▎       | 1768/7505 [25:15<1:53:42,  1.19s/it]
 24%|██▎       | 1769/7505 [25:16<1:57:38,  1.23s/it]
 24%|██▎       | 1770/7505 [25:17<1:50:06,  1.15s/it]
                                                     
{'loss': '4.641', 'grad_norm': '0.3393', 'learning_rate': '0.0006', 'epoch': '93.16'}
+
 24%|██▎       | 1770/7505 [25:17<1:50:06,  1.15s/it]
 24%|██▎       | 1771/7505 [25:19<1:55:33,  1.21s/it]
 24%|██▎       | 1772/7505 [25:20<1:59:19,  1.25s/it]
 24%|██▎       | 1773/7505 [25:21<1:57:03,  1.23s/it]
 24%|██▎       | 1774/7505 [25:23<1:59:24,  1.25s/it]
 24%|██▎       | 1775/7505 [25:24<2:03:07,  1.29s/it]
 24%|██▎       | 1776/7505 [25:25<2:05:47,  1.32s/it]
 24%|██▎       | 1777/7505 [25:26<2:02:04,  1.28s/it]
 24%|██▎       | 1778/7505 [25:28<2:03:05,  1.29s/it]
 24%|██▎       | 1779/7505 [25:29<2:04:27,  1.30s/it]
 24%|██▎       | 1780/7505 [25:31<2:06:27,  1.33s/it]
                                                     
{'loss': '4.668', 'grad_norm': '0.3518', 'learning_rate': '0.0006', 'epoch': '93.7'}
+
 24%|██▎       | 1780/7505 [25:31<2:06:27,  1.33s/it]
 24%|██▎       | 1781/7505 [25:32<2:06:30,  1.33s/it]
 24%|██▎       | 1782/7505 [25:33<2:07:46,  1.34s/it]
 24%|██▍       | 1783/7505 [25:35<2:08:48,  1.35s/it]
 24%|██▍       | 1784/7505 [25:36<2:04:46,  1.31s/it]
 24%|██▍       | 1785/7505 [25:37<2:06:01,  1.32s/it]
 24%|██▍       | 1786/7505 [25:38<1:52:48,  1.18s/it]
 24%|██▍       | 1787/7505 [25:39<1:51:51,  1.17s/it]
 24%|██▍       | 1788/7505 [25:41<1:57:06,  1.23s/it]
 24%|██▍       | 1789/7505 [25:42<2:01:37,  1.28s/it]
 24%|██▍       | 1790/7505 [25:43<2:03:45,  1.30s/it]
                                                     
{'loss': '4.665', 'grad_norm': '0.3569', 'learning_rate': '0.0006', 'epoch': '94.21'}
+
 24%|██▍       | 1790/7505 [25:43<2:03:45,  1.30s/it]
 24%|██▍       | 1791/7505 [25:44<2:01:36,  1.28s/it]
 24%|██▍       | 1792/7505 [25:46<2:03:58,  1.30s/it]
 24%|██▍       | 1793/7505 [25:47<2:06:02,  1.32s/it]
 24%|██▍       | 1794/7505 [25:49<2:05:39,  1.32s/it]
 24%|██▍       | 1795/7505 [25:50<2:02:14,  1.28s/it]
 24%|██▍       | 1796/7505 [25:51<2:04:48,  1.31s/it]
 24%|██▍       | 1797/7505 [25:52<2:06:34,  1.33s/it]
 24%|██▍       | 1798/7505 [25:54<2:06:51,  1.33s/it]
 24%|██▍       | 1799/7505 [25:55<2:06:41,  1.33s/it]
 24%|██▍       | 1800/7505 [25:57<2:07:40,  1.34s/it]
                                                     
{'loss': '4.693', 'grad_norm': '0.3349', 'learning_rate': '0.0006', 'epoch': '94.75'}
+
 24%|██▍       | 1800/7505 [25:57<2:07:40,  1.34s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:02, 14.08it/s]
+
 10%|█         | 4/39 [00:00<00:03,  8.86it/s]
+
 13%|█▎        | 5/39 [00:00<00:04,  8.26it/s]
+
 15%|█▌        | 6/39 [00:00<00:04,  7.82it/s]
+
 18%|█▊        | 7/39 [00:00<00:04,  7.51it/s]
+
 21%|██        | 8/39 [00:00<00:04,  7.36it/s]
+
 23%|██▎       | 9/39 [00:01<00:04,  7.27it/s]
+
 26%|██▌       | 10/39 [00:01<00:04,  7.19it/s]
+
 28%|██▊       | 11/39 [00:01<00:03,  7.26it/s]
+
 31%|███       | 12/39 [00:01<00:03,  7.29it/s]
+
 33%|███▎      | 13/39 [00:01<00:03,  7.30it/s]
+
 36%|███▌      | 14/39 [00:01<00:03,  7.30it/s]
+
 38%|███▊      | 15/39 [00:01<00:03,  7.28it/s]
+
 41%|████      | 16/39 [00:02<00:03,  7.28it/s]
+
 44%|████▎     | 17/39 [00:02<00:03,  7.17it/s]
+
 46%|████▌     | 18/39 [00:02<00:02,  7.11it/s]
+
 49%|████▊     | 19/39 [00:02<00:02,  7.06it/s]
+
 51%|█████▏    | 20/39 [00:02<00:02,  7.02it/s]
+
 54%|█████▍    | 21/39 [00:02<00:02,  7.00it/s]
+
 56%|█████▋    | 22/39 [00:02<00:02,  6.99it/s]
+
 59%|█████▉    | 23/39 [00:03<00:02,  6.96it/s]
+
 62%|██████▏   | 24/39 [00:03<00:02,  6.95it/s]
+
 64%|██████▍   | 25/39 [00:03<00:02,  6.94it/s]
+
 67%|██████▋   | 26/39 [00:03<00:01,  6.92it/s]
+
 69%|██████▉   | 27/39 [00:03<00:01,  6.91it/s]
+
 72%|███████▏  | 28/39 [00:03<00:01,  6.91it/s]
+
 74%|███████▍  | 29/39 [00:03<00:01,  6.90it/s]
+
 77%|███████▋  | 30/39 [00:04<00:01,  6.90it/s]
+
 79%|███████▉  | 31/39 [00:04<00:01,  6.90it/s]
+
 82%|████████▏ | 32/39 [00:04<00:01,  6.89it/s]
+
 85%|████████▍ | 33/39 [00:04<00:00,  6.89it/s]
+
 87%|████████▋ | 34/39 [00:04<00:00,  6.85it/s]
+
 90%|████████▉ | 35/39 [00:04<00:00,  6.82it/s]
+
 92%|█████████▏| 36/39 [00:05<00:00,  6.80it/s]
+
 95%|█████████▍| 37/39 [00:05<00:00,  6.81it/s]
+
 97%|█████████▋| 38/39 [00:05<00:00,  6.82it/s]
+
100%|██████████| 39/39 [00:05<00:00,  6.81it/s]
                                                     
+
                                               
{'eval_loss': '4.504', 'eval_runtime': '5.664', 'eval_samples_per_second': '431', 'eval_steps_per_second': '6.886', 'epoch': '94.75'}
+
 24%|██▍       | 1800/7505 [26:02<2:07:40,  1.34s/it]
+
100%|██████████| 39/39 [00:05<00:00,  6.81it/s][eval] epoch 94.752 step 1800 loss 4.5043 (best 4.3625 @ ep 60.644, 9/20 since best)
+
+
                                               
 24%|██▍       | 1801/7505 [26:04<4:48:46,  3.04s/it]
 24%|██▍       | 1802/7505 [26:05<4:01:13,  2.54s/it]
 24%|██▍       | 1803/7505 [26:06<3:28:07,  2.19s/it]
 24%|██▍       | 1804/7505 [26:07<2:59:54,  1.89s/it]
 24%|██▍       | 1805/7505 [26:08<2:29:31,  1.57s/it]
 24%|██▍       | 1806/7505 [26:10<2:24:40,  1.52s/it]
 24%|██▍       | 1807/7505 [26:11<2:19:47,  1.47s/it]
 24%|██▍       | 1808/7505 [26:12<2:16:30,  1.44s/it]
 24%|██▍       | 1809/7505 [26:14<2:12:35,  1.40s/it]
 24%|██▍       | 1810/7505 [26:15<2:11:51,  1.39s/it]
                                                     
{'loss': '4.643', 'grad_norm': '0.3156', 'learning_rate': '0.0006', 'epoch': '95.27'}
+
 24%|██▍       | 1810/7505 [26:15<2:11:51,  1.39s/it]
 24%|██▍       | 1811/7505 [26:16<2:09:52,  1.37s/it]
 24%|██▍       | 1812/7505 [26:18<2:04:42,  1.31s/it]
 24%|██▍       | 1813/7505 [26:19<2:04:25,  1.31s/it]
 24%|██▍       | 1814/7505 [26:20<1:59:46,  1.26s/it]
 24%|██▍       | 1815/7505 [26:21<2:01:04,  1.28s/it]
 24%|██▍       | 1816/7505 [26:23<1:58:05,  1.25s/it]
 24%|██▍       | 1817/7505 [26:24<1:59:55,  1.27s/it]
 24%|██▍       | 1818/7505 [26:25<1:58:16,  1.25s/it]
 24%|██▍       | 1819/7505 [26:26<1:59:30,  1.26s/it]
 24%|██▍       | 1820/7505 [26:28<1:58:11,  1.25s/it]
                                                     
{'loss': '4.622', 'grad_norm': '0.3488', 'learning_rate': '0.0006', 'epoch': '95.81'}
+
 24%|██▍       | 1820/7505 [26:28<1:58:11,  1.25s/it]
 24%|██▍       | 1821/7505 [26:29<1:59:25,  1.26s/it]
 24%|██▍       | 1822/7505 [26:30<2:00:52,  1.28s/it]
 24%|██▍       | 1823/7505 [26:31<1:43:07,  1.09s/it]
 24%|██▍       | 1824/7505 [26:31<1:25:21,  1.11it/s]
 24%|██▍       | 1825/7505 [26:33<1:38:27,  1.04s/it]
 24%|██▍       | 1826/7505 [26:34<1:42:05,  1.08s/it]
 24%|██▍       | 1827/7505 [26:35<1:48:32,  1.15s/it]
 24%|██▍       | 1828/7505 [26:36<1:54:42,  1.21s/it]
 24%|██▍       | 1829/7505 [26:38<1:58:46,  1.26s/it]
 24%|██▍       | 1830/7505 [26:39<2:01:16,  1.28s/it]
                                                     
{'loss': '4.617', 'grad_norm': '0.337', 'learning_rate': '0.0006', 'epoch': '96.32'}
+
 24%|██▍       | 1830/7505 [26:39<2:01:16,  1.28s/it]
 24%|██▍       | 1831/7505 [26:40<1:58:31,  1.25s/it]
 24%|██▍       | 1832/7505 [26:42<2:02:09,  1.29s/it]
 24%|██▍       | 1833/7505 [26:43<2:03:33,  1.31s/it]
 24%|██▍       | 1834/7505 [26:44<2:00:29,  1.27s/it]
 24%|██▍       | 1835/7505 [26:46<2:02:01,  1.29s/it]
 24%|██▍       | 1836/7505 [26:47<2:04:10,  1.31s/it]
 24%|██▍       | 1837/7505 [26:48<2:05:44,  1.33s/it]
 24%|██▍       | 1838/7505 [26:49<1:53:20,  1.20s/it]
 25%|██▍       | 1839/7505 [26:51<1:56:15,  1.23s/it]
 25%|██▍       | 1840/7505 [26:52<2:00:26,  1.28s/it]
                                                     
{'loss': '4.619', 'grad_norm': '0.3332', 'learning_rate': '0.0006', 'epoch': '96.86'}
+
 25%|██▍       | 1840/7505 [26:52<2:00:26,  1.28s/it]
 25%|██▍       | 1841/7505 [26:53<2:02:30,  1.30s/it]
 25%|██▍       | 1842/7505 [26:55<2:04:11,  1.32s/it]
 25%|██▍       | 1843/7505 [26:55<1:50:43,  1.17s/it]
 25%|██▍       | 1844/7505 [26:57<1:58:44,  1.26s/it]
 25%|██▍       | 1845/7505 [26:58<2:01:51,  1.29s/it]
 25%|██▍       | 1846/7505 [26:59<1:54:09,  1.21s/it]
 25%|██▍       | 1847/7505 [27:01<1:55:08,  1.22s/it]
 25%|██▍       | 1848/7505 [27:02<1:58:26,  1.26s/it]
 25%|██▍       | 1849/7505 [27:03<1:55:55,  1.23s/it]
 25%|██▍       | 1850/7505 [27:04<1:58:07,  1.25s/it]
                                                     
{'loss': '4.537', 'grad_norm': '0.3514', 'learning_rate': '0.0006', 'epoch': '97.38'}
+
 25%|██▍       | 1850/7505 [27:04<1:58:07,  1.25s/it]
 25%|██▍       | 1851/7505 [27:06<1:55:01,  1.22s/it]
 25%|██▍       | 1852/7505 [27:07<1:57:46,  1.25s/it]
 25%|██▍       | 1853/7505 [27:08<1:56:20,  1.23s/it]
 25%|██▍       | 1854/7505 [27:09<1:58:42,  1.26s/it]
 25%|██▍       | 1855/7505 [27:10<1:44:50,  1.11s/it]
 25%|██▍       | 1856/7505 [27:11<1:34:14,  1.00s/it]
 25%|██▍       | 1857/7505 [27:12<1:42:52,  1.09s/it]
 25%|██▍       | 1858/7505 [27:13<1:31:00,  1.03it/s]
 25%|██▍       | 1859/7505 [27:14<1:22:14,  1.14it/s]
 25%|██▍       | 1860/7505 [27:14<1:16:06,  1.24it/s]
                                                     
{'loss': '4.543', 'grad_norm': '0.3616', 'learning_rate': '0.0006', 'epoch': '97.91'}
+
 25%|██▍       | 1860/7505 [27:14<1:16:06,  1.24it/s]
 25%|██▍       | 1861/7505 [27:15<1:12:35,  1.30it/s]
 25%|██▍       | 1862/7505 [27:15<1:02:08,  1.51it/s]
 25%|██▍       | 1863/7505 [27:16<1:04:36,  1.46it/s]
 25%|██▍       | 1864/7505 [27:17<1:03:44,  1.47it/s]
 25%|██▍       | 1865/7505 [27:17<1:03:22,  1.48it/s]
 25%|██▍       | 1866/7505 [27:18<1:03:34,  1.48it/s]
 25%|██▍       | 1867/7505 [27:19<1:02:58,  1.49it/s]
 25%|██▍       | 1868/7505 [27:19<1:02:32,  1.50it/s]
 25%|██▍       | 1869/7505 [27:20<1:03:04,  1.49it/s]
 25%|██▍       | 1870/7505 [27:21<1:02:40,  1.50it/s]
                                                     
{'loss': '4.49', 'grad_norm': '0.336', 'learning_rate': '0.0006', 'epoch': '98.43'}
+
 25%|██▍       | 1870/7505 [27:21<1:02:40,  1.50it/s]
 25%|██▍       | 1871/7505 [27:21<1:02:25,  1.50it/s]
 25%|██▍       | 1872/7505 [27:22<1:02:23,  1.50it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 19.20it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 16.99it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 15.33it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 14.88it/s]
+
 28%|██▊       | 11/39 [00:00<00:01, 14.42it/s]
+
 33%|███▎      | 13/39 [00:00<00:01, 14.51it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 14.36it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 13.82it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 14.22it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 13.97it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 14.34it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 13.77it/s]
+
 69%|██████▉   | 27/39 [00:01<00:00, 14.17it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 13.96it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 13.51it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 13.96it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 14.18it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 14.15it/s]
+
100%|██████████| 39/39 [00:02<00:00, 13.62it/s]
                                                     
+
                                               
{'eval_loss': '4.546', 'eval_runtime': '2.864', 'eval_samples_per_second': '852.4', 'eval_steps_per_second': '13.62', 'epoch': '98.54'}
+
 25%|██▍       | 1872/7505 [27:25<1:02:23,  1.50it/s]
+
100%|██████████| 39/39 [00:02<00:00, 13.62it/s][eval] epoch 98.537 step 1872 loss 4.5458 (best 4.3625 @ ep 60.644, 10/20 since best)
+
+
                                               
 25%|██▍       | 1873/7505 [27:26<2:22:54,  1.52s/it]
 25%|██▍       | 1874/7505 [27:26<1:58:42,  1.26s/it]
 25%|██▍       | 1875/7505 [27:27<1:41:36,  1.08s/it]
 25%|██▍       | 1876/7505 [27:28<1:30:17,  1.04it/s]
 25%|██▌       | 1877/7505 [27:28<1:21:52,  1.15it/s]
 25%|██▌       | 1878/7505 [27:29<1:15:47,  1.24it/s]
 25%|██▌       | 1879/7505 [27:30<1:12:23,  1.30it/s]
 25%|██▌       | 1880/7505 [27:30<1:09:16,  1.35it/s]
                                                     
{'loss': '4.555', 'grad_norm': '0.3243', 'learning_rate': '0.0006', 'epoch': '98.97'}
+
 25%|██▌       | 1880/7505 [27:30<1:09:16,  1.35it/s]
 25%|██▌       | 1881/7505 [27:31<1:00:29,  1.55it/s]
 25%|██▌       | 1882/7505 [27:31<1:03:31,  1.48it/s]
 25%|██▌       | 1883/7505 [27:32<1:02:52,  1.49it/s]
 25%|██▌       | 1884/7505 [27:33<1:03:15,  1.48it/s]
 25%|██▌       | 1885/7505 [27:33<1:02:45,  1.49it/s]
 25%|██▌       | 1886/7505 [27:34<1:02:23,  1.50it/s]
 25%|██▌       | 1887/7505 [27:35<1:02:03,  1.51it/s]
 25%|██▌       | 1888/7505 [27:35<1:02:38,  1.49it/s]
 25%|██▌       | 1889/7505 [27:36<1:02:16,  1.50it/s]
 25%|██▌       | 1890/7505 [27:37<1:13:54,  1.27it/s]
                                                     
{'loss': '4.551', 'grad_norm': '0.3689', 'learning_rate': '0.0006', 'epoch': '99.48'}
+
 25%|██▌       | 1890/7505 [27:37<1:13:54,  1.27it/s]
 25%|██▌       | 1891/7505 [27:38<1:28:37,  1.06it/s]
 25%|██▌       | 1892/7505 [27:40<1:35:41,  1.02s/it]
 25%|██▌       | 1893/7505 [27:41<1:43:33,  1.11s/it]
 25%|██▌       | 1894/7505 [27:42<1:45:47,  1.13s/it]
 25%|██▌       | 1895/7505 [27:43<1:50:33,  1.18s/it]
 25%|██▌       | 1896/7505 [27:45<1:50:03,  1.18s/it]
 25%|██▌       | 1897/7505 [27:46<1:53:49,  1.22s/it]
 25%|██▌       | 1898/7505 [27:47<1:51:57,  1.20s/it]
 25%|██▌       | 1899/7505 [27:48<1:55:12,  1.23s/it]
 25%|██▌       | 1900/7505 [27:49<1:39:55,  1.07s/it]
                                                     
{'loss': '4.596', 'grad_norm': '0.3767', 'learning_rate': '0.0006', 'epoch': '100'}
+
 25%|██▌       | 1900/7505 [27:49<1:39:55,  1.07s/it]
 25%|██▌       | 1901/7505 [27:50<1:30:46,  1.03it/s]
 25%|██▌       | 1902/7505 [27:51<1:36:10,  1.03s/it]
 25%|██▌       | 1903/7505 [27:52<1:43:59,  1.11s/it]
 25%|██▌       | 1904/7505 [27:54<1:50:26,  1.18s/it]
 25%|██▌       | 1905/7505 [27:55<1:54:59,  1.23s/it]
 25%|██▌       | 1906/7505 [27:56<1:53:48,  1.22s/it]
 25%|██▌       | 1907/7505 [27:57<1:56:18,  1.25s/it]
 25%|██▌       | 1908/7505 [27:59<1:59:51,  1.28s/it]
 25%|██▌       | 1909/7505 [28:00<1:56:22,  1.25s/it]
 25%|██▌       | 1910/7505 [28:01<1:58:03,  1.27s/it]
                                                     
{'loss': '4.549', 'grad_norm': '0.3827', 'learning_rate': '0.0006', 'epoch': '100.5'}
+
 25%|██▌       | 1910/7505 [28:01<1:58:03,  1.27s/it]
 25%|██▌       | 1911/7505 [28:02<1:43:07,  1.11s/it]
 25%|██▌       | 1912/7505 [28:03<1:41:04,  1.08s/it]
 25%|██▌       | 1913/7505 [28:04<1:47:13,  1.15s/it]
 26%|██▌       | 1914/7505 [28:06<1:48:06,  1.16s/it]
 26%|██▌       | 1915/7505 [28:07<1:52:05,  1.20s/it]
 26%|██▌       | 1916/7505 [28:08<1:50:50,  1.19s/it]
 26%|██▌       | 1917/7505 [28:09<1:54:02,  1.22s/it]
 26%|██▌       | 1918/7505 [28:10<1:44:35,  1.12s/it]
 26%|██▌       | 1919/7505 [28:11<1:24:58,  1.10it/s]
 26%|██▌       | 1920/7505 [28:12<1:34:27,  1.01s/it]
                                                     
{'loss': '4.601', 'grad_norm': '0.3776', 'learning_rate': '0.0006', 'epoch': '101.1'}
+
 26%|██▌       | 1920/7505 [28:12<1:34:27,  1.01s/it]
 26%|██▌       | 1921/7505 [28:13<1:43:13,  1.11s/it]
 26%|██▌       | 1922/7505 [28:14<1:31:12,  1.02it/s]
 26%|██▌       | 1923/7505 [28:15<1:27:26,  1.06it/s]
 26%|██▌       | 1924/7505 [28:16<1:37:36,  1.05s/it]
 26%|██▌       | 1925/7505 [28:17<1:40:45,  1.08s/it]
 26%|██▌       | 1926/7505 [28:19<1:48:51,  1.17s/it]
 26%|██▌       | 1927/7505 [28:20<1:54:41,  1.23s/it]
 26%|██▌       | 1928/7505 [28:21<1:54:17,  1.23s/it]
 26%|██▌       | 1929/7505 [28:22<1:55:48,  1.25s/it]
 26%|██▌       | 1930/7505 [28:24<1:59:41,  1.29s/it]
                                                     
{'loss': '4.489', 'grad_norm': '0.3201', 'learning_rate': '0.0006', 'epoch': '101.6'}
+
 26%|██▌       | 1930/7505 [28:24<1:59:41,  1.29s/it]
 26%|██▌       | 1931/7505 [28:25<2:02:20,  1.32s/it]
 26%|██▌       | 1932/7505 [28:27<2:02:10,  1.32s/it]
 26%|██▌       | 1933/7505 [28:28<1:57:53,  1.27s/it]
 26%|██▌       | 1934/7505 [28:29<1:58:58,  1.28s/it]
 26%|██▌       | 1935/7505 [28:30<1:56:08,  1.25s/it]
 26%|██▌       | 1936/7505 [28:31<1:56:51,  1.26s/it]
 26%|██▌       | 1937/7505 [28:33<1:57:09,  1.26s/it]
 26%|██▌       | 1938/7505 [28:33<1:33:42,  1.01s/it]
 26%|██▌       | 1939/7505 [28:34<1:37:21,  1.05s/it]
 26%|██▌       | 1940/7505 [28:36<1:44:55,  1.13s/it]
                                                     
{'loss': '4.586', 'grad_norm': '0.3781', 'learning_rate': '0.0006', 'epoch': '102.1'}
+
 26%|██▌       | 1940/7505 [28:36<1:44:55,  1.13s/it]
 26%|██▌       | 1941/7505 [28:37<1:51:12,  1.20s/it]
 26%|██▌       | 1942/7505 [28:38<1:55:43,  1.25s/it]
 26%|██▌       | 1943/7505 [28:40<1:59:04,  1.28s/it]
 26%|██▌       | 1944/7505 [28:41<1:52:08,  1.21s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:02, 14.06it/s]
+
 10%|█         | 4/39 [00:00<00:04,  8.72it/s]
+
 13%|█▎        | 5/39 [00:00<00:04,  8.07it/s]
+
 15%|█▌        | 6/39 [00:00<00:04,  7.68it/s]
+
 18%|█▊        | 7/39 [00:00<00:04,  7.42it/s]
+
 21%|██        | 8/39 [00:01<00:04,  7.22it/s]
+
 23%|██▎       | 9/39 [00:01<00:04,  7.11it/s]
+
 26%|██▌       | 10/39 [00:01<00:04,  7.04it/s]
+
 28%|██▊       | 11/39 [00:01<00:04,  6.98it/s]
+
 31%|███       | 12/39 [00:01<00:03,  6.94it/s]
+
 33%|███▎      | 13/39 [00:01<00:03,  7.52it/s]
+
 36%|███▌      | 14/39 [00:01<00:03,  7.41it/s]
+
 38%|███▊      | 15/39 [00:01<00:03,  7.26it/s]
+
 41%|████      | 16/39 [00:02<00:03,  7.15it/s]
+
 44%|████▎     | 17/39 [00:02<00:03,  7.07it/s]
+
 46%|████▌     | 18/39 [00:02<00:02,  7.04it/s]
+
 49%|████▊     | 19/39 [00:02<00:02,  7.11it/s]
+
 51%|█████▏    | 20/39 [00:02<00:02,  7.09it/s]
+
 54%|█████▍    | 21/39 [00:02<00:02,  7.19it/s]
+
 56%|█████▋    | 22/39 [00:02<00:02,  7.24it/s]
+
 59%|█████▉    | 23/39 [00:03<00:02,  7.29it/s]
+
 62%|██████▏   | 24/39 [00:03<00:02,  7.24it/s]
+
 64%|██████▍   | 25/39 [00:03<00:01,  7.24it/s]
+
 67%|██████▋   | 26/39 [00:03<00:01,  7.25it/s]
+
 69%|██████▉   | 27/39 [00:03<00:01,  7.23it/s]
+
 72%|███████▏  | 28/39 [00:03<00:01,  7.24it/s]
+
 74%|███████▍  | 29/39 [00:03<00:01,  7.22it/s]
+
 79%|███████▉  | 31/39 [00:04<00:00,  9.13it/s]
+
 82%|████████▏ | 32/39 [00:04<00:00,  8.88it/s]
+
 85%|████████▍ | 33/39 [00:04<00:00,  8.24it/s]
+
 87%|████████▋ | 34/39 [00:04<00:00,  7.88it/s]
+
 90%|████████▉ | 35/39 [00:04<00:00,  7.69it/s]
+
 92%|█████████▏| 36/39 [00:04<00:00,  7.61it/s]
+
 95%|█████████▍| 37/39 [00:04<00:00,  7.53it/s]
+
 97%|█████████▋| 38/39 [00:05<00:00,  7.48it/s]
+
100%|██████████| 39/39 [00:05<00:00,  7.41it/s]
                                                     
+
                                               
{'eval_loss': '4.566', 'eval_runtime': '5.372', 'eval_samples_per_second': '454.4', 'eval_steps_per_second': '7.26', 'epoch': '102.3'}
+
 26%|██▌       | 1944/7505 [28:46<1:52:08,  1.21s/it]
+
100%|██████████| 39/39 [00:05<00:00,  7.41it/s][eval] epoch 102.322 step 1944 loss 4.5658 (best 4.3625 @ ep 60.644, 11/20 since best)
+
+
                                               
 26%|██▌       | 1945/7505 [28:47<4:22:59,  2.84s/it]
 26%|██▌       | 1946/7505 [28:49<3:37:38,  2.35s/it]
 26%|██▌       | 1947/7505 [28:50<3:09:04,  2.04s/it]
 26%|██▌       | 1948/7505 [28:51<2:45:21,  1.79s/it]
 26%|██▌       | 1949/7505 [28:52<2:31:55,  1.64s/it]
 26%|██▌       | 1950/7505 [28:54<2:19:07,  1.50s/it]
                                                     
{'loss': '4.545', 'grad_norm': '0.3377', 'learning_rate': '0.0006', 'epoch': '102.6'}
+
 26%|██▌       | 1950/7505 [28:54<2:19:07,  1.50s/it]
 26%|██▌       | 1951/7505 [28:55<2:13:55,  1.45s/it]
 26%|██▌       | 1952/7505 [28:56<2:05:00,  1.35s/it]
 26%|██▌       | 1953/7505 [28:57<2:04:00,  1.34s/it]
 26%|██▌       | 1954/7505 [28:58<1:49:32,  1.18s/it]
 26%|██▌       | 1955/7505 [28:59<1:34:59,  1.03s/it]
 26%|██▌       | 1956/7505 [29:00<1:42:56,  1.11s/it]
 26%|██▌       | 1957/7505 [29:01<1:35:26,  1.03s/it]
 26%|██▌       | 1958/7505 [29:02<1:40:59,  1.09s/it]
 26%|██▌       | 1959/7505 [29:04<1:47:24,  1.16s/it]
 26%|██▌       | 1960/7505 [29:04<1:38:00,  1.06s/it]
                                                     
{'loss': '4.558', 'grad_norm': '0.3449', 'learning_rate': '0.0006', 'epoch': '103.2'}
+
 26%|██▌       | 1960/7505 [29:04<1:38:00,  1.06s/it]
 26%|██▌       | 1961/7505 [29:05<1:29:00,  1.04it/s]
 26%|██▌       | 1962/7505 [29:06<1:39:29,  1.08s/it]
 26%|██▌       | 1963/7505 [29:08<1:42:49,  1.11s/it]
 26%|██▌       | 1964/7505 [29:09<1:48:39,  1.18s/it]
 26%|██▌       | 1965/7505 [29:10<1:44:09,  1.13s/it]
 26%|██▌       | 1966/7505 [29:11<1:31:01,  1.01it/s]
 26%|██▌       | 1967/7505 [29:12<1:39:11,  1.07s/it]
 26%|██▌       | 1968/7505 [29:13<1:46:05,  1.15s/it]
 26%|██▌       | 1969/7505 [29:14<1:34:14,  1.02s/it]
 26%|██▌       | 1970/7505 [29:15<1:36:16,  1.04s/it]
                                                     
{'loss': '4.508', 'grad_norm': '0.3257', 'learning_rate': '0.0006', 'epoch': '103.7'}
+
 26%|██▌       | 1970/7505 [29:15<1:36:16,  1.04s/it]
 26%|██▋       | 1971/7505 [29:16<1:43:52,  1.13s/it]
 26%|██▋       | 1972/7505 [29:18<1:44:40,  1.14s/it]
 26%|██▋       | 1973/7505 [29:19<1:49:59,  1.19s/it]
 26%|██▋       | 1974/7505 [29:20<1:40:17,  1.09s/it]
 26%|██▋       | 1975/7505 [29:21<1:47:17,  1.16s/it]
 26%|██▋       | 1976/7505 [29:22<1:39:03,  1.08s/it]
 26%|██▋       | 1977/7505 [29:23<1:49:11,  1.19s/it]
 26%|██▋       | 1978/7505 [29:24<1:39:48,  1.08s/it]
 26%|██▋       | 1979/7505 [29:25<1:28:33,  1.04it/s]
 26%|██▋       | 1980/7505 [29:26<1:39:34,  1.08s/it]
                                                     
{'loss': '4.489', 'grad_norm': '0.4044', 'learning_rate': '0.0006', 'epoch': '104.2'}
+
 26%|██▋       | 1980/7505 [29:26<1:39:34,  1.08s/it]
 26%|██▋       | 1981/7505 [29:28<1:47:37,  1.17s/it]
 26%|██▋       | 1982/7505 [29:29<1:53:36,  1.23s/it]
 26%|██▋       | 1983/7505 [29:30<1:53:37,  1.23s/it]
 26%|██▋       | 1984/7505 [29:32<1:57:53,  1.28s/it]
 26%|██▋       | 1985/7505 [29:33<2:00:30,  1.31s/it]
 26%|██▋       | 1986/7505 [29:34<2:00:52,  1.31s/it]
 26%|██▋       | 1987/7505 [29:35<1:55:07,  1.25s/it]
 26%|██▋       | 1988/7505 [29:36<1:39:51,  1.09s/it]
 27%|██▋       | 1989/7505 [29:37<1:45:53,  1.15s/it]
 27%|██▋       | 1990/7505 [29:39<1:46:09,  1.15s/it]
                                                     
{'loss': '4.62', 'grad_norm': '0.3768', 'learning_rate': '0.0006', 'epoch': '104.8'}
+
 27%|██▋       | 1990/7505 [29:39<1:46:09,  1.15s/it]
 27%|██▋       | 1991/7505 [29:39<1:32:27,  1.01s/it]
 27%|██▋       | 1992/7505 [29:41<1:41:36,  1.11s/it]
 27%|██▋       | 1993/7505 [29:42<1:46:16,  1.16s/it]
 27%|██▋       | 1994/7505 [29:43<1:51:16,  1.21s/it]
 27%|██▋       | 1995/7505 [29:44<1:41:15,  1.10s/it]
 27%|██▋       | 1996/7505 [29:45<1:49:48,  1.20s/it]
 27%|██▋       | 1997/7505 [29:46<1:34:19,  1.03s/it]
 27%|██▋       | 1998/7505 [29:47<1:23:38,  1.10it/s]
 27%|██▋       | 1999/7505 [29:47<1:16:04,  1.21it/s]
 27%|██▋       | 2000/7505 [29:48<1:10:42,  1.30it/s]
                                                     
{'loss': '4.58', 'grad_norm': '0.3718', 'learning_rate': '0.0006', 'epoch': '105.3'}
+
 27%|██▋       | 2000/7505 [29:48<1:10:42,  1.30it/s]
 27%|██▋       | 2001/7505 [29:49<1:06:57,  1.37it/s]
 27%|██▋       | 2002/7505 [29:49<1:04:19,  1.43it/s]
 27%|██▋       | 2003/7505 [29:50<1:02:29,  1.47it/s]
 27%|██▋       | 2004/7505 [29:51<1:18:47,  1.16it/s]
 27%|██▋       | 2005/7505 [29:52<1:30:11,  1.02it/s]
 27%|██▋       | 2006/7505 [29:54<1:35:42,  1.04s/it]
 27%|██▋       | 2007/7505 [29:55<1:43:02,  1.12s/it]
 27%|██▋       | 2008/7505 [29:56<1:43:57,  1.13s/it]
 27%|██▋       | 2009/7505 [29:57<1:48:26,  1.18s/it]
 27%|██▋       | 2010/7505 [29:58<1:35:37,  1.04s/it]
                                                     
{'loss': '4.582', 'grad_norm': '0.3677', 'learning_rate': '0.0006', 'epoch': '105.8'}
+
 27%|██▋       | 2010/7505 [29:58<1:35:37,  1.04s/it]
 27%|██▋       | 2011/7505 [29:59<1:27:55,  1.04it/s]
 27%|██▋       | 2012/7505 [30:00<1:38:52,  1.08s/it]
 27%|██▋       | 2013/7505 [30:01<1:41:35,  1.11s/it]
 27%|██▋       | 2014/7505 [30:02<1:22:04,  1.12it/s]
 27%|██▋       | 2015/7505 [30:03<1:31:56,  1.00s/it]
 27%|██▋       | 2016/7505 [30:04<1:41:24,  1.11s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:02, 17.37it/s]
+
 10%|█         | 4/39 [00:00<00:02, 15.94it/s]
+
 15%|█▌        | 6/39 [00:00<00:02, 15.58it/s]
+
 21%|██        | 8/39 [00:00<00:02, 15.43it/s]
+
 26%|██▌       | 10/39 [00:00<00:01, 15.33it/s]
+
 31%|███       | 12/39 [00:00<00:01, 15.28it/s]
+
 36%|███▌      | 14/39 [00:00<00:01, 13.58it/s]
+
 41%|████      | 16/39 [00:01<00:02, 10.40it/s]
+
 46%|████▌     | 18/39 [00:01<00:02,  9.00it/s]
+
 51%|█████▏    | 20/39 [00:01<00:02,  8.24it/s]
+
 54%|█████▍    | 21/39 [00:01<00:02,  7.95it/s]
+
 56%|█████▋    | 22/39 [00:02<00:02,  7.74it/s]
+
 59%|█████▉    | 23/39 [00:02<00:02,  7.55it/s]
+
 62%|██████▏   | 24/39 [00:02<00:02,  7.37it/s]
+
 64%|██████▍   | 25/39 [00:02<00:01,  7.33it/s]
+
 67%|██████▋   | 26/39 [00:02<00:01,  7.25it/s]
+
 69%|██████▉   | 27/39 [00:02<00:01,  7.25it/s]
+
 72%|███████▏  | 28/39 [00:02<00:01,  7.27it/s]
+
 74%|███████▍  | 29/39 [00:03<00:01,  7.29it/s]
+
 77%|███████▋  | 30/39 [00:03<00:01,  7.30it/s]
+
 79%|███████▉  | 31/39 [00:03<00:01,  7.29it/s]
+
 82%|████████▏ | 32/39 [00:03<00:00,  7.28it/s]
+
 87%|████████▋ | 34/39 [00:03<00:00,  8.95it/s]
+
 92%|█████████▏| 36/39 [00:03<00:00,  8.91it/s]
+
 95%|█████████▍| 37/39 [00:04<00:00,  8.37it/s]
+
 97%|█████████▋| 38/39 [00:04<00:00,  7.97it/s]
+
100%|██████████| 39/39 [00:04<00:00,  7.66it/s]
                                                     
+
                                               
{'eval_loss': '4.596', 'eval_runtime': '4.516', 'eval_samples_per_second': '540.5', 'eval_steps_per_second': '8.635', 'epoch': '106.1'}
+
 27%|██▋       | 2016/7505 [30:09<1:41:24,  1.11s/it]
+
100%|██████████| 39/39 [00:04<00:00,  7.66it/s][eval] epoch 106.107 step 2016 loss 4.5960 (best 4.3625 @ ep 60.644, 12/20 since best)
+
+
                                               
 27%|██▋       | 2017/7505 [30:10<3:50:34,  2.52s/it]
 27%|██▋       | 2018/7505 [30:11<3:13:23,  2.11s/it]
 27%|██▋       | 2019/7505 [30:13<2:51:49,  1.88s/it]
 27%|██▋       | 2020/7505 [30:14<2:21:13,  1.54s/it]
                                                     
{'loss': '4.545', 'grad_norm': '0.3526', 'learning_rate': '0.0006', 'epoch': '106.3'}
+
 27%|██▋       | 2020/7505 [30:14<2:21:13,  1.54s/it]
 27%|██▋       | 2021/7505 [30:15<2:08:16,  1.40s/it]
 27%|██▋       | 2022/7505 [30:16<2:05:36,  1.37s/it]
 27%|██▋       | 2023/7505 [30:17<1:59:39,  1.31s/it]
 27%|██▋       | 2024/7505 [30:18<1:59:19,  1.31s/it]
 27%|██▋       | 2025/7505 [30:20<1:54:34,  1.25s/it]
 27%|██▋       | 2026/7505 [30:21<1:56:17,  1.27s/it]
 27%|██▋       | 2027/7505 [30:22<1:53:16,  1.24s/it]
 27%|██▋       | 2028/7505 [30:23<1:56:14,  1.27s/it]
 27%|██▋       | 2029/7505 [30:24<1:52:46,  1.24s/it]
 27%|██▋       | 2030/7505 [30:26<1:55:30,  1.27s/it]
                                                     
{'loss': '4.511', 'grad_norm': '0.3407', 'learning_rate': '0.0006', 'epoch': '106.9'}
+
 27%|██▋       | 2030/7505 [30:26<1:55:30,  1.27s/it]
 27%|██▋       | 2031/7505 [30:27<1:50:31,  1.21s/it]
 27%|██▋       | 2032/7505 [30:28<1:34:41,  1.04s/it]
 27%|██▋       | 2033/7505 [30:28<1:23:00,  1.10it/s]
 27%|██▋       | 2034/7505 [30:30<1:36:25,  1.06s/it]
 27%|██▋       | 2035/7505 [30:31<1:38:55,  1.09s/it]
 27%|██▋       | 2036/7505 [30:32<1:45:01,  1.15s/it]
 27%|██▋       | 2037/7505 [30:33<1:42:37,  1.13s/it]
 27%|██▋       | 2038/7505 [30:34<1:31:42,  1.01s/it]
 27%|██▋       | 2039/7505 [30:35<1:40:43,  1.11s/it]
 27%|██▋       | 2040/7505 [30:36<1:42:56,  1.13s/it]
                                                     
{'loss': '4.469', 'grad_norm': '0.4419', 'learning_rate': '0.0006', 'epoch': '107.4'}
+
 27%|██▋       | 2040/7505 [30:36<1:42:56,  1.13s/it]
 27%|██▋       | 2041/7505 [30:37<1:29:23,  1.02it/s]
 27%|██▋       | 2042/7505 [30:38<1:31:29,  1.00s/it]
 27%|██▋       | 2043/7505 [30:39<1:40:02,  1.10s/it]
 27%|██▋       | 2044/7505 [30:40<1:40:52,  1.11s/it]
 27%|██▋       | 2045/7505 [30:42<1:46:23,  1.17s/it]
 27%|██▋       | 2046/7505 [30:43<1:46:43,  1.17s/it]
 27%|██▋       | 2047/7505 [30:44<1:50:14,  1.21s/it]
 27%|██▋       | 2048/7505 [30:45<1:48:48,  1.20s/it]
 27%|██▋       | 2049/7505 [30:47<1:51:39,  1.23s/it]
 27%|██▋       | 2050/7505 [30:48<1:49:29,  1.20s/it]
                                                     
{'loss': '4.495', 'grad_norm': '0.382', 'learning_rate': '0.0006', 'epoch': '107.9'}
+
 27%|██▋       | 2050/7505 [30:48<1:49:29,  1.20s/it]
 27%|██▋       | 2051/7505 [30:49<1:53:13,  1.25s/it]
 27%|██▋       | 2052/7505 [30:50<1:41:05,  1.11s/it]
 27%|██▋       | 2053/7505 [30:51<1:30:19,  1.01it/s]
 27%|██▋       | 2054/7505 [30:52<1:39:48,  1.10s/it]
 27%|██▋       | 2055/7505 [30:53<1:45:54,  1.17s/it]
 27%|██▋       | 2056/7505 [30:55<1:46:32,  1.17s/it]
 27%|██▋       | 2057/7505 [30:56<1:50:10,  1.21s/it]
 27%|██▋       | 2058/7505 [30:57<1:34:22,  1.04s/it]
 27%|██▋       | 2059/7505 [30:58<1:38:17,  1.08s/it]
 27%|██▋       | 2060/7505 [30:59<1:44:55,  1.16s/it]
                                                     
{'loss': '4.503', 'grad_norm': '0.3155', 'learning_rate': '0.0006', 'epoch': '108.4'}
+
 27%|██▋       | 2060/7505 [30:59<1:44:55,  1.16s/it]
 27%|██▋       | 2061/7505 [31:00<1:44:54,  1.16s/it]
 27%|██▋       | 2062/7505 [31:02<1:49:00,  1.20s/it]
 27%|██▋       | 2063/7505 [31:02<1:36:59,  1.07s/it]
 28%|██▊       | 2064/7505 [31:03<1:28:26,  1.03it/s]
 28%|██▊       | 2065/7505 [31:04<1:38:02,  1.08s/it]
 28%|██▊       | 2066/7505 [31:06<1:40:44,  1.11s/it]
 28%|██▊       | 2067/7505 [31:07<1:45:58,  1.17s/it]
 28%|██▊       | 2068/7505 [31:08<1:46:51,  1.18s/it]
 28%|██▊       | 2069/7505 [31:09<1:49:38,  1.21s/it]
 28%|██▊       | 2070/7505 [31:11<1:49:43,  1.21s/it]
                                                     
{'loss': '4.515', 'grad_norm': '0.3011', 'learning_rate': '0.0006', 'epoch': '109'}
+
 28%|██▊       | 2070/7505 [31:11<1:49:43,  1.21s/it]
 28%|██▊       | 2071/7505 [31:11<1:27:40,  1.03it/s]
 28%|██▊       | 2072/7505 [31:12<1:24:25,  1.07it/s]
 28%|██▊       | 2073/7505 [31:13<1:35:50,  1.06s/it]
 28%|██▊       | 2074/7505 [31:14<1:39:55,  1.10s/it]
 28%|██▊       | 2075/7505 [31:16<1:43:36,  1.14s/it]
 28%|██▊       | 2076/7505 [31:17<1:47:59,  1.19s/it]
 28%|██▊       | 2077/7505 [31:18<1:47:59,  1.19s/it]
 28%|██▊       | 2078/7505 [31:19<1:50:43,  1.22s/it]
 28%|██▊       | 2079/7505 [31:20<1:34:41,  1.05s/it]
 28%|██▊       | 2080/7505 [31:21<1:30:07,  1.00it/s]
                                                     
{'loss': '4.489', 'grad_norm': '0.4102', 'learning_rate': '0.0006', 'epoch': '109.5'}
+
 28%|██▊       | 2080/7505 [31:21<1:30:07,  1.00it/s]
 28%|██▊       | 2081/7505 [31:22<1:39:14,  1.10s/it]
 28%|██▊       | 2082/7505 [31:23<1:40:43,  1.11s/it]
 28%|██▊       | 2083/7505 [31:25<1:45:46,  1.17s/it]
 28%|██▊       | 2084/7505 [31:26<1:47:21,  1.19s/it]
 28%|██▊       | 2085/7505 [31:27<1:32:19,  1.02s/it]
 28%|██▊       | 2086/7505 [31:28<1:31:05,  1.01s/it]
 28%|██▊       | 2087/7505 [31:29<1:39:57,  1.11s/it]
 28%|██▊       | 2088/7505 [31:30<1:34:53,  1.05s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 22.97it/s]
+
 15%|█▌        | 6/39 [00:00<00:01, 16.67it/s]
+
 21%|██        | 8/39 [00:00<00:02, 11.32it/s]
+
 26%|██▌       | 10/39 [00:00<00:03,  9.47it/s]
+
 31%|███       | 12/39 [00:01<00:03,  8.56it/s]
+
 33%|███▎      | 13/39 [00:01<00:03,  8.24it/s]
+
 36%|███▌      | 14/39 [00:01<00:03,  7.99it/s]
+
 38%|███▊      | 15/39 [00:01<00:03,  7.75it/s]
+
 41%|████      | 16/39 [00:01<00:03,  7.58it/s]
+
 44%|████▎     | 17/39 [00:01<00:02,  7.43it/s]
+
 46%|████▌     | 18/39 [00:02<00:02,  7.34it/s]
+
 49%|████▊     | 19/39 [00:02<00:02,  7.24it/s]
+
 51%|█████▏    | 20/39 [00:02<00:02,  7.18it/s]
+
 54%|█████▍    | 21/39 [00:02<00:02,  7.13it/s]
+
 56%|█████▋    | 22/39 [00:02<00:02,  7.10it/s]
+
 59%|█████▉    | 23/39 [00:02<00:02,  7.06it/s]
+
 62%|██████▏   | 24/39 [00:02<00:02,  7.45it/s]
+
 67%|██████▋   | 26/39 [00:02<00:01,  9.70it/s]
+
 72%|███████▏  | 28/39 [00:03<00:00, 11.32it/s]
+
 77%|███████▋  | 30/39 [00:03<00:00, 12.48it/s]
+
 82%|████████▏ | 32/39 [00:03<00:00, 13.28it/s]
+
 87%|████████▋ | 34/39 [00:03<00:00, 13.85it/s]
+
 92%|█████████▏| 36/39 [00:03<00:00, 14.25it/s]
+
 97%|█████████▋| 38/39 [00:03<00:00, 14.54it/s]
                                                     
+
                                               
{'eval_loss': '4.612', 'eval_runtime': '4.072', 'eval_samples_per_second': '599.4', 'eval_steps_per_second': '9.577', 'epoch': '109.9'}
+
 28%|██▊       | 2088/7505 [31:34<1:34:53,  1.05s/it]
+
100%|██████████| 39/39 [00:03<00:00, 14.54it/s][eval] epoch 109.913 step 2088 loss 4.6123 (best 4.3625 @ ep 60.644, 13/20 since best)
+
+
                                               
 28%|██▊       | 2089/7505 [31:35<3:33:12,  2.36s/it]
 28%|██▊       | 2090/7505 [31:36<2:52:13,  1.91s/it]
                                                     
{'loss': '4.546', 'grad_norm': '0.3707', 'learning_rate': '0.0006', 'epoch': '110'}
+
 28%|██▊       | 2090/7505 [31:36<2:52:13,  1.91s/it]
 28%|██▊       | 2091/7505 [31:37<2:34:19,  1.71s/it]
 28%|██▊       | 2092/7505 [31:39<2:23:21,  1.59s/it]
 28%|██▊       | 2093/7505 [31:40<2:12:03,  1.46s/it]
 28%|██▊       | 2094/7505 [31:41<2:07:40,  1.42s/it]
 28%|██▊       | 2095/7505 [31:42<2:00:07,  1.33s/it]
 28%|██▊       | 2096/7505 [31:44<1:59:20,  1.32s/it]
 28%|██▊       | 2097/7505 [31:45<1:55:18,  1.28s/it]
 28%|██▊       | 2098/7505 [31:46<1:56:32,  1.29s/it]
 28%|██▊       | 2099/7505 [31:47<1:52:16,  1.25s/it]
 28%|██▊       | 2100/7505 [31:48<1:53:44,  1.26s/it]
                                                     
{'loss': '4.487', 'grad_norm': '0.3927', 'learning_rate': '0.0006', 'epoch': '110.5'}
+
 28%|██▊       | 2100/7505 [31:48<1:53:44,  1.26s/it]
 28%|██▊       | 2101/7505 [31:50<1:51:00,  1.23s/it]
 28%|██▊       | 2102/7505 [31:51<1:52:52,  1.25s/it]
 28%|██▊       | 2103/7505 [31:52<1:50:10,  1.22s/it]
 28%|██▊       | 2104/7505 [31:53<1:52:08,  1.25s/it]
 28%|██▊       | 2105/7505 [31:55<1:49:29,  1.22s/it]
 28%|██▊       | 2106/7505 [31:55<1:33:51,  1.04s/it]
 28%|██▊       | 2107/7505 [31:56<1:34:20,  1.05s/it]
 28%|██▊       | 2108/7505 [31:58<1:42:34,  1.14s/it]
 28%|██▊       | 2109/7505 [31:58<1:28:30,  1.02it/s]
 28%|██▊       | 2110/7505 [31:59<1:22:03,  1.10it/s]
                                                     
{'loss': '4.523', 'grad_norm': '0.3864', 'learning_rate': '0.0006', 'epoch': '111.1'}
+
 28%|██▊       | 2110/7505 [31:59<1:22:03,  1.10it/s]
 28%|██▊       | 2111/7505 [32:00<1:14:39,  1.20it/s]
 28%|██▊       | 2112/7505 [32:00<1:09:26,  1.29it/s]
 28%|██▊       | 2113/7505 [32:01<1:06:04,  1.36it/s]
 28%|██▊       | 2114/7505 [32:02<1:03:23,  1.42it/s]
 28%|██▊       | 2115/7505 [32:02<1:01:34,  1.46it/s]
 28%|██▊       | 2116/7505 [32:03<1:00:31,  1.48it/s]
 28%|██▊       | 2117/7505 [32:03<59:32,  1.51it/s]  
 28%|██▊       | 2118/7505 [32:04<58:52,  1.53it/s]
 28%|██▊       | 2119/7505 [32:05<58:38,  1.53it/s]
 28%|██▊       | 2120/7505 [32:05<58:14,  1.54it/s]
                                                   
{'loss': '4.418', 'grad_norm': '0.3977', 'learning_rate': '0.0006', 'epoch': '111.6'}
+
 28%|██▊       | 2120/7505 [32:05<58:14,  1.54it/s]
 28%|██▊       | 2121/7505 [32:06<57:57,  1.55it/s]
 28%|██▊       | 2122/7505 [32:07<57:59,  1.55it/s]
 28%|██▊       | 2123/7505 [32:07<57:43,  1.55it/s]
 28%|██▊       | 2124/7505 [32:08<57:33,  1.56it/s]
 28%|██▊       | 2125/7505 [32:09<57:40,  1.55it/s]
 28%|██▊       | 2126/7505 [32:09<57:30,  1.56it/s]
 28%|██▊       | 2127/7505 [32:10<57:20,  1.56it/s]
 28%|██▊       | 2128/7505 [32:10<51:00,  1.76it/s]
 28%|██▊       | 2129/7505 [32:11<55:05,  1.63it/s]
 28%|██▊       | 2130/7505 [32:12<55:59,  1.60it/s]
                                                   
{'loss': '4.466', 'grad_norm': '0.4051', 'learning_rate': '0.0006', 'epoch': '112.1'}
+
 28%|██▊       | 2130/7505 [32:12<55:59,  1.60it/s]
 28%|██▊       | 2131/7505 [32:12<56:21,  1.59it/s]
 28%|██▊       | 2132/7505 [32:13<56:35,  1.58it/s]
 28%|██▊       | 2133/7505 [32:14<57:02,  1.57it/s]
 28%|██▊       | 2134/7505 [32:14<57:03,  1.57it/s]
 28%|██▊       | 2135/7505 [32:15<57:03,  1.57it/s]
 28%|██▊       | 2136/7505 [32:15<57:17,  1.56it/s]
 28%|██▊       | 2137/7505 [32:16<57:14,  1.56it/s]
 28%|██▊       | 2138/7505 [32:17<57:09,  1.56it/s]
 29%|██▊       | 2139/7505 [32:17<57:22,  1.56it/s]
 29%|██▊       | 2140/7505 [32:18<57:15,  1.56it/s]
                                                   
{'loss': '4.473', 'grad_norm': '0.3274', 'learning_rate': '0.0006', 'epoch': '112.6'}
+
 29%|██▊       | 2140/7505 [32:18<57:15,  1.56it/s]
 29%|██▊       | 2141/7505 [32:19<57:10,  1.56it/s]
 29%|██▊       | 2142/7505 [32:19<57:23,  1.56it/s]
 29%|██▊       | 2143/7505 [32:20<57:16,  1.56it/s]
 29%|██▊       | 2144/7505 [32:21<57:12,  1.56it/s]
 29%|██▊       | 2145/7505 [32:21<57:25,  1.56it/s]
 29%|██▊       | 2146/7505 [32:22<57:16,  1.56it/s]
 29%|██▊       | 2147/7505 [32:22<50:56,  1.75it/s]
 29%|██▊       | 2148/7505 [32:23<55:24,  1.61it/s]
 29%|██▊       | 2149/7505 [32:24<55:56,  1.60it/s]
 29%|██▊       | 2150/7505 [32:24<56:21,  1.58it/s]
                                                   
{'loss': '4.437', 'grad_norm': '0.3965', 'learning_rate': '0.0006', 'epoch': '113.2'}
+
 29%|██▊       | 2150/7505 [32:24<56:21,  1.58it/s]
 29%|██▊       | 2151/7505 [32:25<56:32,  1.58it/s]
 29%|██▊       | 2152/7505 [32:26<56:39,  1.57it/s]
 29%|██▊       | 2153/7505 [32:26<56:57,  1.57it/s]
 29%|██▊       | 2154/7505 [32:27<56:57,  1.57it/s]
 29%|██▊       | 2155/7505 [32:28<56:54,  1.57it/s]
 29%|██▊       | 2156/7505 [32:28<57:09,  1.56it/s]
 29%|██▊       | 2157/7505 [32:29<57:04,  1.56it/s]
 29%|██▉       | 2158/7505 [32:29<57:02,  1.56it/s]
 29%|██▉       | 2159/7505 [32:30<57:16,  1.56it/s]
 29%|██▉       | 2160/7505 [32:31<57:09,  1.56it/s]
                                                   
{'loss': '4.449', 'grad_norm': '0.3895', 'learning_rate': '0.0006', 'epoch': '113.7'}
+
 29%|██▉       | 2160/7505 [32:31<57:09,  1.56it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 23.73it/s]
+
 15%|█▌        | 6/39 [00:00<00:01, 17.26it/s]
+
 21%|██        | 8/39 [00:00<00:01, 16.44it/s]
+
 26%|██▌       | 10/39 [00:00<00:01, 15.98it/s]
+
 31%|███       | 12/39 [00:00<00:01, 15.34it/s]
+
 36%|███▌      | 14/39 [00:00<00:01, 15.30it/s]
+
 41%|████      | 16/39 [00:01<00:01, 15.25it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 14.92it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 14.98it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 15.04it/s]
+
 62%|██████▏   | 24/39 [00:01<00:00, 15.10it/s]
+
 67%|██████▋   | 26/39 [00:01<00:00, 14.81it/s]
+
 72%|███████▏  | 28/39 [00:01<00:00, 14.93it/s]
+
 77%|███████▋  | 30/39 [00:01<00:00, 15.00it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 15.07it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 14.78it/s]
+
 92%|█████████▏| 36/39 [00:02<00:00, 14.91it/s]
+
 97%|█████████▋| 38/39 [00:02<00:00, 14.99it/s]
                                                   
+
                                               
{'eval_loss': '4.635', 'eval_runtime': '2.66', 'eval_samples_per_second': '917.7', 'eval_steps_per_second': '14.66', 'epoch': '113.7'}
+
 29%|██▉       | 2160/7505 [32:33<57:09,  1.56it/s]
+
100%|██████████| 39/39 [00:02<00:00, 14.99it/s][eval] epoch 113.698 step 2160 loss 4.6346 (best 4.3625 @ ep 60.644, 14/20 since best)
+
+
                                               
 29%|██▉       | 2161/7505 [32:34<2:08:22,  1.44s/it]
 29%|██▉       | 2162/7505 [32:35<1:46:55,  1.20s/it]
 29%|██▉       | 2163/7505 [32:35<1:31:49,  1.03s/it]
 29%|██▉       | 2164/7505 [32:36<1:21:34,  1.09it/s]
 29%|██▉       | 2165/7505 [32:37<1:14:07,  1.20it/s]
 29%|██▉       | 2166/7505 [32:37<1:02:41,  1.42it/s]
 29%|██▉       | 2167/7505 [32:38<1:03:10,  1.41it/s]
 29%|██▉       | 2168/7505 [32:38<1:01:27,  1.45it/s]
 29%|██▉       | 2169/7505 [32:39<1:00:03,  1.48it/s]
 29%|██▉       | 2170/7505 [32:40<59:01,  1.51it/s]  
                                                   
{'loss': '4.501', 'grad_norm': '0.44', 'learning_rate': '0.0006', 'epoch': '114.2'}
+
 29%|██▉       | 2170/7505 [32:40<59:01,  1.51it/s]
 29%|██▉       | 2171/7505 [32:40<58:37,  1.52it/s]
 29%|██▉       | 2172/7505 [32:41<58:03,  1.53it/s]
 29%|██▉       | 2173/7505 [32:42<57:38,  1.54it/s]
 29%|██▉       | 2174/7505 [32:42<57:38,  1.54it/s]
 29%|██▉       | 2175/7505 [32:43<57:20,  1.55it/s]
 29%|██▉       | 2176/7505 [32:43<57:08,  1.55it/s]
 29%|██▉       | 2177/7505 [32:44<57:16,  1.55it/s]
 29%|██▉       | 2178/7505 [32:45<57:07,  1.55it/s]
 29%|██▉       | 2179/7505 [32:45<57:02,  1.56it/s]
 29%|██▉       | 2180/7505 [32:46<56:56,  1.56it/s]
                                                   
{'loss': '4.509', 'grad_norm': '0.422', 'learning_rate': '0.0006', 'epoch': '114.8'}
+
 29%|██▉       | 2180/7505 [32:46<56:56,  1.56it/s]
 29%|██▉       | 2181/7505 [32:47<56:52,  1.56it/s]
 29%|██▉       | 2182/7505 [32:47<56:44,  1.56it/s]
 29%|██▉       | 2183/7505 [32:48<56:36,  1.57it/s]
 29%|██▉       | 2184/7505 [32:49<56:29,  1.57it/s]
 29%|██▉       | 2185/7505 [32:49<50:09,  1.77it/s]
 29%|██▉       | 2186/7505 [32:50<54:39,  1.62it/s]
 29%|██▉       | 2187/7505 [32:50<55:09,  1.61it/s]
 29%|██▉       | 2188/7505 [32:51<55:29,  1.60it/s]
 29%|██▉       | 2189/7505 [32:52<55:43,  1.59it/s]
 29%|██▉       | 2190/7505 [32:52<55:51,  1.59it/s]
                                                   
{'loss': '4.393', 'grad_norm': '0.3924', 'learning_rate': '0.0006', 'epoch': '115.3'}
+
 29%|██▉       | 2190/7505 [32:52<55:51,  1.59it/s]
 29%|██▉       | 2191/7505 [32:53<55:58,  1.58it/s]
 29%|██▉       | 2192/7505 [32:54<56:04,  1.58it/s]
 29%|██▉       | 2193/7505 [32:54<56:07,  1.58it/s]
 29%|██▉       | 2194/7505 [32:55<56:07,  1.58it/s]
 29%|██▉       | 2195/7505 [32:55<56:09,  1.58it/s]
 29%|██▉       | 2196/7505 [32:56<56:09,  1.58it/s]
 29%|██▉       | 2197/7505 [32:57<56:10,  1.57it/s]
 29%|██▉       | 2198/7505 [32:57<56:09,  1.58it/s]
 29%|██▉       | 2199/7505 [32:58<56:09,  1.57it/s]
 29%|██▉       | 2200/7505 [32:59<56:10,  1.57it/s]
                                                   
{'loss': '4.443', 'grad_norm': '0.3703', 'learning_rate': '0.0006', 'epoch': '115.8'}
+
 29%|██▉       | 2200/7505 [32:59<56:10,  1.57it/s]
 29%|██▉       | 2201/7505 [32:59<56:10,  1.57it/s]
 29%|██▉       | 2202/7505 [33:00<56:10,  1.57it/s]
 29%|██▉       | 2203/7505 [33:01<56:07,  1.57it/s]
 29%|██▉       | 2204/7505 [33:01<49:53,  1.77it/s]
 29%|██▉       | 2205/7505 [33:02<53:58,  1.64it/s]
 29%|██▉       | 2206/7505 [33:02<54:36,  1.62it/s]
 29%|██▉       | 2207/7505 [33:03<55:02,  1.60it/s]
 29%|██▉       | 2208/7505 [33:04<55:19,  1.60it/s]
 29%|██▉       | 2209/7505 [33:04<55:31,  1.59it/s]
 29%|██▉       | 2210/7505 [33:05<55:40,  1.58it/s]
                                                   
{'loss': '4.434', 'grad_norm': '0.4873', 'learning_rate': '0.0006', 'epoch': '116.3'}
+
 29%|██▉       | 2210/7505 [33:05<55:40,  1.58it/s]
 29%|██▉       | 2211/7505 [33:05<55:48,  1.58it/s]
 29%|██▉       | 2212/7505 [33:06<55:51,  1.58it/s]
 29%|██▉       | 2213/7505 [33:07<55:54,  1.58it/s]
 30%|██▉       | 2214/7505 [33:07<55:55,  1.58it/s]
 30%|██▉       | 2215/7505 [33:08<55:55,  1.58it/s]
 30%|██▉       | 2216/7505 [33:09<55:57,  1.58it/s]
 30%|██▉       | 2217/7505 [33:09<55:57,  1.57it/s]
 30%|██▉       | 2218/7505 [33:10<55:57,  1.57it/s]
 30%|██▉       | 2219/7505 [33:11<55:56,  1.57it/s]
 30%|██▉       | 2220/7505 [33:11<55:57,  1.57it/s]
                                                   
{'loss': '4.507', 'grad_norm': '0.3283', 'learning_rate': '0.0006', 'epoch': '116.9'}
+
 30%|██▉       | 2220/7505 [33:11<55:57,  1.57it/s]
 30%|██▉       | 2221/7505 [33:12<55:57,  1.57it/s]
 30%|██▉       | 2222/7505 [33:12<55:54,  1.57it/s]
 30%|██▉       | 2223/7505 [33:13<49:40,  1.77it/s]
 30%|██▉       | 2224/7505 [33:14<53:42,  1.64it/s]
 30%|██▉       | 2225/7505 [33:14<54:22,  1.62it/s]
 30%|██▉       | 2226/7505 [33:15<54:49,  1.60it/s]
 30%|██▉       | 2227/7505 [33:15<55:07,  1.60it/s]
 30%|██▉       | 2228/7505 [33:16<55:19,  1.59it/s]
 30%|██▉       | 2229/7505 [33:17<55:29,  1.58it/s]
 30%|██▉       | 2230/7505 [33:17<55:34,  1.58it/s]
                                                   
{'loss': '4.39', 'grad_norm': '0.3573', 'learning_rate': '0.0006', 'epoch': '117.4'}
+
 30%|██▉       | 2230/7505 [33:17<55:34,  1.58it/s]
 30%|██▉       | 2231/7505 [33:18<55:39,  1.58it/s]
 30%|██▉       | 2232/7505 [33:19<55:41,  1.58it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 22.50it/s]
+
 15%|█▌        | 6/39 [00:00<00:01, 17.45it/s]
+
 21%|██        | 8/39 [00:00<00:01, 16.53it/s]
+
 26%|██▌       | 10/39 [00:00<00:01, 16.03it/s]
+
 31%|███       | 12/39 [00:00<00:01, 15.73it/s]
+
 36%|███▌      | 14/39 [00:00<00:01, 15.52it/s]
+
 41%|████      | 16/39 [00:00<00:01, 15.39it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 15.32it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 15.26it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 15.23it/s]
+
 62%|██████▏   | 24/39 [00:01<00:00, 15.25it/s]
+
 67%|██████▋   | 26/39 [00:01<00:00, 15.24it/s]
+
 72%|███████▏  | 28/39 [00:01<00:00, 15.22it/s]
+
 77%|███████▋  | 30/39 [00:01<00:00, 15.22it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 15.21it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 15.21it/s]
+
 92%|█████████▏| 36/39 [00:02<00:00, 15.20it/s]
+
 97%|█████████▋| 38/39 [00:02<00:00, 15.20it/s]
                                                   
+
                                               
{'eval_loss': '4.67', 'eval_runtime': '2.632', 'eval_samples_per_second': '927.4', 'eval_steps_per_second': '14.82', 'epoch': '117.5'}
+
 30%|██▉       | 2232/7505 [33:21<55:41,  1.58it/s]
+
100%|██████████| 39/39 [00:02<00:00, 15.20it/s][eval] epoch 117.483 step 2232 loss 4.6705 (best 4.3625 @ ep 60.644, 15/20 since best)
+
+
                                               
 30%|██▉       | 2233/7505 [33:22<2:05:06,  1.42s/it]
 30%|██▉       | 2234/7505 [33:23<1:44:19,  1.19s/it]
 30%|██▉       | 2235/7505 [33:23<1:29:46,  1.02s/it]
 30%|██▉       | 2236/7505 [33:24<1:19:38,  1.10it/s]
 30%|██▉       | 2237/7505 [33:24<1:12:30,  1.21it/s]
 30%|██▉       | 2238/7505 [33:25<1:07:30,  1.30it/s]
 30%|██▉       | 2239/7505 [33:26<1:03:59,  1.37it/s]
 30%|██▉       | 2240/7505 [33:26<1:01:33,  1.43it/s]
                                                     
{'loss': '4.396', 'grad_norm': '0.3665', 'learning_rate': '0.0006', 'epoch': '117.9'}
+
 30%|██▉       | 2240/7505 [33:26<1:01:33,  1.43it/s]
 30%|██▉       | 2241/7505 [33:27<59:46,  1.47it/s]  
 30%|██▉       | 2242/7505 [33:27<52:22,  1.67it/s]
 30%|██▉       | 2243/7505 [33:28<55:43,  1.57it/s]
 30%|██▉       | 2244/7505 [33:29<55:54,  1.57it/s]
 30%|██▉       | 2245/7505 [33:29<55:57,  1.57it/s]
 30%|██▉       | 2246/7505 [33:30<55:56,  1.57it/s]
 30%|██▉       | 2247/7505 [33:31<55:54,  1.57it/s]
 30%|██▉       | 2248/7505 [33:31<55:54,  1.57it/s]
 30%|██▉       | 2249/7505 [33:32<55:53,  1.57it/s]
 30%|██▉       | 2250/7505 [33:33<55:49,  1.57it/s]
                                                   
{'loss': '4.269', 'grad_norm': '0.366', 'learning_rate': '0.0006', 'epoch': '118.4'}
+
 30%|██▉       | 2250/7505 [33:33<55:49,  1.57it/s]
 30%|██▉       | 2251/7505 [33:33<55:47,  1.57it/s]
 30%|███       | 2252/7505 [33:34<55:44,  1.57it/s]
 30%|███       | 2253/7505 [33:35<55:44,  1.57it/s]
 30%|███       | 2254/7505 [33:35<55:42,  1.57it/s]
 30%|███       | 2255/7505 [33:36<55:40,  1.57it/s]
 30%|███       | 2256/7505 [33:36<55:40,  1.57it/s]
 30%|███       | 2257/7505 [33:37<55:40,  1.57it/s]
 30%|███       | 2258/7505 [33:38<55:38,  1.57it/s]
 30%|███       | 2259/7505 [33:38<55:37,  1.57it/s]
 30%|███       | 2260/7505 [33:39<55:37,  1.57it/s]
                                                   
{'loss': '4.504', 'grad_norm': '0.3594', 'learning_rate': '0.0006', 'epoch': '119'}
+
 30%|███       | 2260/7505 [33:39<55:37,  1.57it/s]
 30%|███       | 2261/7505 [33:39<49:27,  1.77it/s]
 30%|███       | 2262/7505 [33:40<53:27,  1.63it/s]
 30%|███       | 2263/7505 [33:41<54:04,  1.62it/s]
 30%|███       | 2264/7505 [33:41<54:28,  1.60it/s]
 30%|███       | 2265/7505 [33:42<54:45,  1.59it/s]
 30%|███       | 2266/7505 [33:43<54:57,  1.59it/s]
 30%|███       | 2267/7505 [33:43<55:05,  1.58it/s]
 30%|███       | 2268/7505 [33:44<55:11,  1.58it/s]
 30%|███       | 2269/7505 [33:45<55:14,  1.58it/s]
 30%|███       | 2270/7505 [33:45<55:16,  1.58it/s]
                                                   
{'loss': '4.413', 'grad_norm': '0.4164', 'learning_rate': '0.0006', 'epoch': '119.5'}
+
 30%|███       | 2270/7505 [33:45<55:16,  1.58it/s]
 30%|███       | 2271/7505 [33:46<55:20,  1.58it/s]
 30%|███       | 2272/7505 [33:46<55:21,  1.58it/s]
 30%|███       | 2273/7505 [33:47<55:21,  1.58it/s]
 30%|███       | 2274/7505 [33:48<55:22,  1.57it/s]
 30%|███       | 2275/7505 [33:48<55:23,  1.57it/s]
 30%|███       | 2276/7505 [33:49<55:23,  1.57it/s]
 30%|███       | 2277/7505 [33:50<55:21,  1.57it/s]
 30%|███       | 2278/7505 [33:50<55:21,  1.57it/s]
 30%|███       | 2279/7505 [33:51<55:19,  1.57it/s]
 30%|███       | 2280/7505 [33:51<49:14,  1.77it/s]
                                                   
{'loss': '4.411', 'grad_norm': '0.3968', 'learning_rate': '0.0006', 'epoch': '120'}
+
 30%|███       | 2280/7505 [33:51<49:14,  1.77it/s]
 30%|███       | 2281/7505 [33:52<53:42,  1.62it/s]
 30%|███       | 2282/7505 [33:53<54:10,  1.61it/s]
 30%|███       | 2283/7505 [33:53<54:29,  1.60it/s]
 30%|███       | 2284/7505 [33:54<54:42,  1.59it/s]
 30%|███       | 2285/7505 [33:55<54:53,  1.58it/s]
 30%|███       | 2286/7505 [33:55<55:00,  1.58it/s]
 30%|███       | 2287/7505 [33:56<55:03,  1.58it/s]
 30%|███       | 2288/7505 [33:56<55:09,  1.58it/s]
 30%|███       | 2289/7505 [33:57<55:15,  1.57it/s]
 31%|███       | 2290/7505 [33:58<55:15,  1.57it/s]
                                                   
{'loss': '4.334', 'grad_norm': '0.3769', 'learning_rate': '0.0006', 'epoch': '120.5'}
+
 31%|███       | 2290/7505 [33:58<55:15,  1.57it/s]
 31%|███       | 2291/7505 [33:58<55:15,  1.57it/s]
 31%|███       | 2292/7505 [33:59<55:16,  1.57it/s]
 31%|███       | 2293/7505 [34:00<55:15,  1.57it/s]
 31%|███       | 2294/7505 [34:00<55:13,  1.57it/s]
 31%|███       | 2295/7505 [34:01<55:12,  1.57it/s]
 31%|███       | 2296/7505 [34:02<55:12,  1.57it/s]
 31%|███       | 2297/7505 [34:02<55:12,  1.57it/s]
 31%|███       | 2298/7505 [34:03<55:09,  1.57it/s]
 31%|███       | 2299/7505 [34:03<48:59,  1.77it/s]
 31%|███       | 2300/7505 [34:04<52:26,  1.65it/s]
                                                   
{'loss': '4.342', 'grad_norm': '0.3851', 'learning_rate': '0.0006', 'epoch': '121.1'}
+
 31%|███       | 2300/7505 [34:04<52:26,  1.65it/s]
 31%|███       | 2301/7505 [34:05<53:15,  1.63it/s]
 31%|███       | 2302/7505 [34:05<53:48,  1.61it/s]
 31%|███       | 2303/7505 [34:06<54:11,  1.60it/s]
 31%|███       | 2304/7505 [34:06<54:27,  1.59it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 22.72it/s]
+
 15%|█▌        | 6/39 [00:00<00:01, 17.55it/s]
+
 21%|██        | 8/39 [00:00<00:01, 16.63it/s]
+
 26%|██▌       | 10/39 [00:00<00:01, 16.09it/s]
+
 31%|███       | 12/39 [00:00<00:01, 15.77it/s]
+
 36%|███▌      | 14/39 [00:00<00:01, 15.58it/s]
+
 41%|████      | 16/39 [00:00<00:01, 15.46it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 15.37it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 15.32it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 15.27it/s]
+
 62%|██████▏   | 24/39 [00:01<00:00, 15.28it/s]
+
 67%|██████▋   | 26/39 [00:01<00:00, 15.24it/s]
+
 72%|███████▏  | 28/39 [00:01<00:00, 15.23it/s]
+
 77%|███████▋  | 30/39 [00:01<00:00, 15.21it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 15.21it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 15.19it/s]
+
 92%|█████████▏| 36/39 [00:02<00:00, 15.20it/s]
+
 97%|█████████▋| 38/39 [00:02<00:00, 15.19it/s]
                                                   
+
                                               
{'eval_loss': '4.707', 'eval_runtime': '2.638', 'eval_samples_per_second': '925.2', 'eval_steps_per_second': '14.78', 'epoch': '121.3'}
+
 31%|███       | 2304/7505 [34:09<54:27,  1.59it/s]
+
100%|██████████| 39/39 [00:02<00:00, 15.19it/s][eval] epoch 121.268 step 2304 loss 4.7068 (best 4.3625 @ ep 60.644, 16/20 since best)
+
+
                                               
 31%|███       | 2305/7505 [34:10<2:03:13,  1.42s/it]
 31%|███       | 2306/7505 [34:10<1:42:46,  1.19s/it]
 31%|███       | 2307/7505 [34:11<1:28:26,  1.02s/it]
 31%|███       | 2308/7505 [34:12<1:18:23,  1.10it/s]
 31%|███       | 2309/7505 [34:12<1:11:21,  1.21it/s]
 31%|███       | 2310/7505 [34:13<1:06:27,  1.30it/s]
                                                     
{'loss': '4.315', 'grad_norm': '0.3904', 'learning_rate': '0.0006', 'epoch': '121.6'}
+
 31%|███       | 2310/7505 [34:13<1:06:27,  1.30it/s]
 31%|███       | 2311/7505 [34:14<1:03:02,  1.37it/s]
 31%|███       | 2312/7505 [34:14<1:00:37,  1.43it/s]
 31%|███       | 2313/7505 [34:15<58:55,  1.47it/s]  
 31%|███       | 2314/7505 [34:15<57:43,  1.50it/s]
 31%|███       | 2315/7505 [34:16<56:54,  1.52it/s]
 31%|███       | 2316/7505 [34:17<56:20,  1.54it/s]
 31%|███       | 2317/7505 [34:17<55:53,  1.55it/s]
 31%|███       | 2318/7505 [34:18<49:30,  1.75it/s]
 31%|███       | 2319/7505 [34:18<52:45,  1.64it/s]
 31%|███       | 2320/7505 [34:19<53:22,  1.62it/s]
                                                   
{'loss': '4.458', 'grad_norm': '0.3851', 'learning_rate': '0.0006', 'epoch': '122.1'}
+
 31%|███       | 2320/7505 [34:19<53:22,  1.62it/s]
 31%|███       | 2321/7505 [34:20<53:50,  1.60it/s]
 31%|███       | 2322/7505 [34:20<54:09,  1.59it/s]
 31%|███       | 2323/7505 [34:21<54:21,  1.59it/s]
 31%|███       | 2324/7505 [34:22<54:30,  1.58it/s]
 31%|███       | 2325/7505 [34:22<54:36,  1.58it/s]
 31%|███       | 2326/7505 [34:23<54:41,  1.58it/s]
 31%|███       | 2327/7505 [34:24<54:42,  1.58it/s]
 31%|███       | 2328/7505 [34:24<54:44,  1.58it/s]
 31%|███       | 2329/7505 [34:25<54:45,  1.58it/s]
 31%|███       | 2330/7505 [34:25<54:45,  1.58it/s]
                                                   
{'loss': '4.373', 'grad_norm': '0.3785', 'learning_rate': '0.0006', 'epoch': '122.6'}
+
 31%|███       | 2330/7505 [34:25<54:45,  1.58it/s]
 31%|███       | 2331/7505 [34:26<54:47,  1.57it/s]
 31%|███       | 2332/7505 [34:27<54:47,  1.57it/s]
 31%|███       | 2333/7505 [34:27<54:47,  1.57it/s]
 31%|███       | 2334/7505 [34:28<54:47,  1.57it/s]
 31%|███       | 2335/7505 [34:29<54:48,  1.57it/s]
 31%|███       | 2336/7505 [34:29<54:44,  1.57it/s]
 31%|███       | 2337/7505 [34:30<48:39,  1.77it/s]
 31%|███       | 2338/7505 [34:30<52:34,  1.64it/s]
 31%|███       | 2339/7505 [34:31<1:01:57,  1.39it/s]
 31%|███       | 2340/7505 [34:32<59:45,  1.44it/s]  
                                                   
{'loss': '4.437', 'grad_norm': '0.3792', 'learning_rate': '0.0006', 'epoch': '123.2'}
+
 31%|███       | 2340/7505 [34:32<59:45,  1.44it/s]
 31%|███       | 2341/7505 [34:33<58:13,  1.48it/s]
 31%|███       | 2342/7505 [34:33<57:08,  1.51it/s]
 31%|███       | 2343/7505 [34:34<56:22,  1.53it/s]
 31%|███       | 2344/7505 [34:35<55:50,  1.54it/s]
 31%|███       | 2345/7505 [34:35<55:29,  1.55it/s]
 31%|███▏      | 2346/7505 [34:36<55:14,  1.56it/s]
 31%|███▏      | 2347/7505 [34:36<55:02,  1.56it/s]
 31%|███▏      | 2348/7505 [34:37<54:54,  1.57it/s]
 31%|███▏      | 2349/7505 [34:38<54:47,  1.57it/s]
 31%|███▏      | 2350/7505 [34:38<54:45,  1.57it/s]
                                                   
{'loss': '4.337', 'grad_norm': '0.4237', 'learning_rate': '0.0006', 'epoch': '123.7'}
+
 31%|███▏      | 2350/7505 [34:38<54:45,  1.57it/s]
 31%|███▏      | 2351/7505 [34:39<54:42,  1.57it/s]
 31%|███▏      | 2352/7505 [34:40<54:40,  1.57it/s]
 31%|███▏      | 2353/7505 [34:40<54:39,  1.57it/s]
 31%|███▏      | 2354/7505 [34:41<54:39,  1.57it/s]
 31%|███▏      | 2355/7505 [34:42<54:35,  1.57it/s]
 31%|███▏      | 2356/7505 [34:42<48:29,  1.77it/s]
 31%|███▏      | 2357/7505 [34:43<52:23,  1.64it/s]
 31%|███▏      | 2358/7505 [34:43<53:01,  1.62it/s]
 31%|███▏      | 2359/7505 [34:44<53:26,  1.61it/s]
 31%|███▏      | 2360/7505 [34:45<53:45,  1.60it/s]
                                                   
{'loss': '4.422', 'grad_norm': '0.3829', 'learning_rate': '0.0006', 'epoch': '124.2'}
+
 31%|███▏      | 2360/7505 [34:45<53:45,  1.60it/s]
 31%|███▏      | 2361/7505 [34:45<53:58,  1.59it/s]
 31%|███▏      | 2362/7505 [34:46<54:06,  1.58it/s]
 31%|███▏      | 2363/7505 [34:46<54:12,  1.58it/s]
 31%|███▏      | 2364/7505 [34:47<1:03:07,  1.36it/s]
 32%|███▏      | 2365/7505 [34:48<1:00:28,  1.42it/s]
 32%|███▏      | 2366/7505 [34:49<58:39,  1.46it/s]  
 32%|███▏      | 2367/7505 [34:49<57:21,  1.49it/s]
 32%|███▏      | 2368/7505 [34:50<56:26,  1.52it/s]
 32%|███▏      | 2369/7505 [34:51<55:49,  1.53it/s]
 32%|███▏      | 2370/7505 [34:51<55:22,  1.55it/s]
                                                   
{'loss': '4.359', 'grad_norm': '0.388', 'learning_rate': '0.0006', 'epoch': '124.8'}
+
 32%|███▏      | 2370/7505 [34:51<55:22,  1.55it/s]
 32%|███▏      | 2371/7505 [34:52<55:04,  1.55it/s]
 32%|███▏      | 2372/7505 [34:53<54:51,  1.56it/s]
 32%|███▏      | 2373/7505 [34:53<54:42,  1.56it/s]
 32%|███▏      | 2374/7505 [34:54<54:35,  1.57it/s]
 32%|███▏      | 2375/7505 [34:54<48:27,  1.76it/s]
 32%|███▏      | 2376/7505 [34:55<52:13,  1.64it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 23.44it/s]
+
 15%|█▌        | 6/39 [00:00<00:01, 17.74it/s]
+
 21%|██        | 8/39 [00:00<00:01, 16.72it/s]
+
 26%|██▌       | 10/39 [00:00<00:01, 16.06it/s]
+
 31%|███       | 12/39 [00:00<00:01, 15.74it/s]
+
 36%|███▌      | 14/39 [00:00<00:01, 15.56it/s]
+
 41%|████      | 16/39 [00:00<00:01, 15.43it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 15.36it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 15.29it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 15.27it/s]
+
 62%|██████▏   | 24/39 [00:01<00:00, 15.26it/s]
+
 67%|██████▋   | 26/39 [00:01<00:00, 15.25it/s]
+
 72%|███████▏  | 28/39 [00:01<00:00, 15.22it/s]
+
 77%|███████▋  | 30/39 [00:01<00:00, 15.23it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 15.21it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 15.22it/s]
+
 92%|█████████▏| 36/39 [00:02<00:00, 15.20it/s]
+
 97%|█████████▋| 38/39 [00:02<00:00, 15.21it/s]
                                                   
+
                                               
{'eval_loss': '4.736', 'eval_runtime': '2.617', 'eval_samples_per_second': '932.7', 'eval_steps_per_second': '14.9', 'epoch': '125.1'}
+
 32%|███▏      | 2376/7505 [34:58<52:13,  1.64it/s]
+
100%|██████████| 39/39 [00:02<00:00, 15.21it/s][eval] epoch 125.054 step 2376 loss 4.7356 (best 4.3625 @ ep 60.644, 17/20 since best)
+
+
                                               
 32%|███▏      | 2377/7505 [34:58<1:59:57,  1.40s/it]
 32%|███▏      | 2378/7505 [34:59<1:40:15,  1.17s/it]
 32%|███▏      | 2379/7505 [34:59<1:26:27,  1.01s/it]
 32%|███▏      | 2380/7505 [35:00<1:16:47,  1.11it/s]
                                                     
{'loss': '4.373', 'grad_norm': '0.3824', 'learning_rate': '0.0006', 'epoch': '125.3'}
+
 32%|███▏      | 2380/7505 [35:00<1:16:47,  1.11it/s]
 32%|███▏      | 2381/7505 [35:01<1:10:01,  1.22it/s]
 32%|███▏      | 2382/7505 [35:01<1:05:16,  1.31it/s]
 32%|███▏      | 2383/7505 [35:02<1:01:57,  1.38it/s]
 32%|███▏      | 2384/7505 [35:03<59:37,  1.43it/s]  
 32%|███▏      | 2385/7505 [35:04<1:06:49,  1.28it/s]
 32%|███▏      | 2386/7505 [35:04<1:03:00,  1.35it/s]
 32%|███▏      | 2387/7505 [35:05<1:00:23,  1.41it/s]
 32%|███▏      | 2388/7505 [35:06<58:30,  1.46it/s]  
 32%|███▏      | 2389/7505 [35:06<57:11,  1.49it/s]
 32%|███▏      | 2390/7505 [35:07<56:16,  1.51it/s]
                                                   
{'loss': '4.365', 'grad_norm': '0.3951', 'learning_rate': '0.0006', 'epoch': '125.8'}
+
 32%|███▏      | 2390/7505 [35:07<56:16,  1.51it/s]
 32%|███▏      | 2391/7505 [35:07<55:38,  1.53it/s]
 32%|███▏      | 2392/7505 [35:08<55:11,  1.54it/s]
 32%|███▏      | 2393/7505 [35:09<54:51,  1.55it/s]
 32%|███▏      | 2394/7505 [35:09<48:36,  1.75it/s]
 32%|███▏      | 2395/7505 [35:10<52:17,  1.63it/s]
 32%|███▏      | 2396/7505 [35:10<52:50,  1.61it/s]
 32%|███▏      | 2397/7505 [35:11<53:12,  1.60it/s]
 32%|███▏      | 2398/7505 [35:12<53:29,  1.59it/s]
 32%|███▏      | 2399/7505 [35:12<53:39,  1.59it/s]
 32%|███▏      | 2400/7505 [35:13<53:45,  1.58it/s]
                                                   
{'loss': '4.404', 'grad_norm': '0.3724', 'learning_rate': '0.0006', 'epoch': '126.3'}
+
 32%|███▏      | 2400/7505 [35:13<53:45,  1.58it/s]
 32%|███▏      | 2401/7505 [35:14<53:52,  1.58it/s]
 32%|███▏      | 2402/7505 [35:15<1:02:42,  1.36it/s]
 32%|███▏      | 2403/7505 [35:15<1:00:04,  1.42it/s]
 32%|███▏      | 2404/7505 [35:16<58:15,  1.46it/s]  
 32%|███▏      | 2405/7505 [35:17<56:57,  1.49it/s]
 32%|███▏      | 2406/7505 [35:17<56:03,  1.52it/s]
 32%|███▏      | 2407/7505 [35:18<55:25,  1.53it/s]
 32%|███▏      | 2408/7505 [35:18<54:58,  1.55it/s]
 32%|███▏      | 2409/7505 [35:19<54:38,  1.55it/s]
 32%|███▏      | 2410/7505 [35:20<1:03:16,  1.34it/s]
                                                     
{'loss': '4.443', 'grad_norm': '0.3521', 'learning_rate': '0.0006', 'epoch': '126.9'}
+
 32%|███▏      | 2410/7505 [35:20<1:03:16,  1.34it/s]
 32%|███▏      | 2411/7505 [35:21<1:00:28,  1.40it/s]
 32%|███▏      | 2412/7505 [35:21<58:27,  1.45it/s]  
 32%|███▏      | 2413/7505 [35:22<51:06,  1.66it/s]
 32%|███▏      | 2414/7505 [35:22<53:45,  1.58it/s]
 32%|███▏      | 2415/7505 [35:23<53:48,  1.58it/s]
 32%|███▏      | 2416/7505 [35:24<53:50,  1.58it/s]
 32%|███▏      | 2417/7505 [35:24<53:52,  1.57it/s]
 32%|███▏      | 2418/7505 [35:25<53:52,  1.57it/s]
 32%|███▏      | 2419/7505 [35:26<53:52,  1.57it/s]
 32%|███▏      | 2420/7505 [35:26<53:51,  1.57it/s]
                                                   
{'loss': '4.392', 'grad_norm': '0.3815', 'learning_rate': '0.0006', 'epoch': '127.4'}
+
 32%|███▏      | 2420/7505 [35:26<53:51,  1.57it/s]
 32%|███▏      | 2421/7505 [35:27<53:52,  1.57it/s]
 32%|███▏      | 2422/7505 [35:27<53:53,  1.57it/s]
 32%|███▏      | 2423/7505 [35:28<53:53,  1.57it/s]
 32%|███▏      | 2424/7505 [35:29<53:52,  1.57it/s]
 32%|███▏      | 2425/7505 [35:29<53:53,  1.57it/s]
 32%|███▏      | 2426/7505 [35:30<53:51,  1.57it/s]
 32%|███▏      | 2427/7505 [35:31<53:49,  1.57it/s]
 32%|███▏      | 2428/7505 [35:31<53:47,  1.57it/s]
 32%|███▏      | 2429/7505 [35:32<53:45,  1.57it/s]
 32%|███▏      | 2430/7505 [35:33<53:44,  1.57it/s]
                                                   
{'loss': '4.379', 'grad_norm': '0.3796', 'learning_rate': '0.0006', 'epoch': '127.9'}
+
 32%|███▏      | 2430/7505 [35:33<53:44,  1.57it/s]
 32%|███▏      | 2431/7505 [35:33<53:42,  1.57it/s]
 32%|███▏      | 2432/7505 [35:34<47:45,  1.77it/s]
 32%|███▏      | 2433/7505 [35:34<51:33,  1.64it/s]
 32%|███▏      | 2434/7505 [35:35<52:12,  1.62it/s]
 32%|███▏      | 2435/7505 [35:36<52:40,  1.60it/s]
 32%|███▏      | 2436/7505 [35:36<53:00,  1.59it/s]
 32%|███▏      | 2437/7505 [35:37<53:12,  1.59it/s]
 32%|███▏      | 2438/7505 [35:38<53:21,  1.58it/s]
 32%|███▏      | 2439/7505 [35:38<53:26,  1.58it/s]
 33%|███▎      | 2440/7505 [35:39<1:02:19,  1.35it/s]
                                                     
{'loss': '4.336', 'grad_norm': '0.3725', 'learning_rate': '0.0006', 'epoch': '128.4'}
+
 33%|███▎      | 2440/7505 [35:39<1:02:19,  1.35it/s]
 33%|███▎      | 2441/7505 [35:40<59:44,  1.41it/s]  
 33%|███▎      | 2442/7505 [35:40<57:54,  1.46it/s]
 33%|███▎      | 2443/7505 [35:41<56:37,  1.49it/s]
 33%|███▎      | 2444/7505 [35:42<55:42,  1.51it/s]
 33%|███▎      | 2445/7505 [35:43<1:03:50,  1.32it/s]
 33%|███▎      | 2446/7505 [35:43<1:00:43,  1.39it/s]
 33%|███▎      | 2447/7505 [35:44<58:32,  1.44it/s]  
 33%|███▎      | 2448/7505 [35:45<57:01,  1.48it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 22.87it/s]
+
 15%|█▌        | 6/39 [00:00<00:01, 17.58it/s]
+
 21%|██        | 8/39 [00:00<00:01, 16.64it/s]
+
 26%|██▌       | 10/39 [00:00<00:01, 16.12it/s]
+
 31%|███       | 12/39 [00:00<00:01, 15.77it/s]
+
 36%|███▌      | 14/39 [00:00<00:01, 15.60it/s]
+
 41%|████      | 16/39 [00:00<00:01, 15.46it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 15.35it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 15.32it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 15.26it/s]
+
 62%|██████▏   | 24/39 [00:01<00:00, 15.27it/s]
+
 67%|██████▋   | 26/39 [00:01<00:00, 15.25it/s]
+
 72%|███████▏  | 28/39 [00:01<00:00, 15.23it/s]
+
 77%|███████▋  | 30/39 [00:01<00:00, 15.21it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 15.21it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 15.19it/s]
+
 92%|█████████▏| 36/39 [00:02<00:00, 15.21it/s]
+
 97%|█████████▋| 38/39 [00:02<00:00, 15.19it/s]
                                                   
+
                                               
{'eval_loss': '4.749', 'eval_runtime': '2.627', 'eval_samples_per_second': '929.3', 'eval_steps_per_second': '14.85', 'epoch': '128.9'}
+
 33%|███▎      | 2448/7505 [35:47<57:01,  1.48it/s]
+
100%|██████████| 39/39 [00:02<00:00, 15.19it/s][eval] epoch 128.859 step 2448 loss 4.7488 (best 4.3625 @ ep 60.644, 18/20 since best)
+
+
                                               
 33%|███▎      | 2449/7505 [35:48<2:02:20,  1.45s/it]
 33%|███▎      | 2450/7505 [35:48<1:41:41,  1.21s/it]
                                                     
{'loss': '4.419', 'grad_norm': '0.4035', 'learning_rate': '0.0006', 'epoch': '129'}
+
 33%|███▎      | 2450/7505 [35:48<1:41:41,  1.21s/it]
 33%|███▎      | 2451/7505 [35:49<1:21:17,  1.04it/s]
 33%|███▎      | 2452/7505 [35:50<1:15:02,  1.12it/s]
 33%|███▎      | 2453/7505 [35:50<1:08:34,  1.23it/s]
 33%|███▎      | 2454/7505 [35:51<1:04:02,  1.31it/s]
 33%|███▎      | 2455/7505 [35:51<1:00:50,  1.38it/s]
 33%|███▎      | 2456/7505 [35:52<58:37,  1.44it/s]  
 33%|███▎      | 2457/7505 [35:53<57:04,  1.47it/s]
 33%|███▎      | 2458/7505 [35:53<55:58,  1.50it/s]
 33%|███▎      | 2459/7505 [35:54<55:11,  1.52it/s]
 33%|███▎      | 2460/7505 [35:55<54:39,  1.54it/s]
                                                   
{'loss': '4.352', 'grad_norm': '0.4184', 'learning_rate': '0.0006', 'epoch': '129.5'}
+
 33%|███▎      | 2460/7505 [35:55<54:39,  1.54it/s]
 33%|███▎      | 2461/7505 [35:55<54:17,  1.55it/s]
 33%|███▎      | 2462/7505 [35:56<1:02:55,  1.34it/s]
 33%|███▎      | 2463/7505 [35:57<1:00:02,  1.40it/s]
 33%|███▎      | 2464/7505 [35:58<58:02,  1.45it/s]  
 33%|███▎      | 2465/7505 [35:58<56:37,  1.48it/s]
 33%|███▎      | 2466/7505 [35:59<55:42,  1.51it/s]
 33%|███▎      | 2467/7505 [35:59<54:59,  1.53it/s]
 33%|███▎      | 2468/7505 [36:00<54:31,  1.54it/s]
 33%|███▎      | 2469/7505 [36:01<54:07,  1.55it/s]
 33%|███▎      | 2470/7505 [36:01<47:56,  1.75it/s]
                                                   
{'loss': '4.321', 'grad_norm': '0.4469', 'learning_rate': '0.0006', 'epoch': '130'}
+
 33%|███▎      | 2470/7505 [36:01<47:56,  1.75it/s]
 33%|███▎      | 2471/7505 [36:02<51:14,  1.64it/s]
 33%|███▎      | 2472/7505 [36:02<51:53,  1.62it/s]
 33%|███▎      | 2473/7505 [36:03<52:17,  1.60it/s]
 33%|███▎      | 2474/7505 [36:04<52:34,  1.59it/s]
 33%|███▎      | 2475/7505 [36:04<52:46,  1.59it/s]
 33%|███▎      | 2476/7505 [36:05<52:56,  1.58it/s]
 33%|███▎      | 2477/7505 [36:06<53:01,  1.58it/s]
 33%|███▎      | 2478/7505 [36:06<53:04,  1.58it/s]
 33%|███▎      | 2479/7505 [36:07<53:05,  1.58it/s]
 33%|███▎      | 2480/7505 [36:08<53:07,  1.58it/s]
                                                   
{'loss': '4.337', 'grad_norm': '0.3775', 'learning_rate': '0.0006', 'epoch': '130.5'}
+
 33%|███▎      | 2480/7505 [36:08<53:07,  1.58it/s]
 33%|███▎      | 2481/7505 [36:08<53:08,  1.58it/s]
 33%|███▎      | 2482/7505 [36:09<53:08,  1.58it/s]
 33%|███▎      | 2483/7505 [36:09<53:07,  1.58it/s]
 33%|███▎      | 2484/7505 [36:10<53:09,  1.57it/s]
 33%|███▎      | 2485/7505 [36:11<53:09,  1.57it/s]
 33%|███▎      | 2486/7505 [36:11<53:07,  1.57it/s]
 33%|███▎      | 2487/7505 [36:12<53:07,  1.57it/s]
 33%|███▎      | 2488/7505 [36:13<53:05,  1.57it/s]
 33%|███▎      | 2489/7505 [36:13<47:12,  1.77it/s]
 33%|███▎      | 2490/7505 [36:14<51:02,  1.64it/s]
                                                   
{'loss': '4.294', 'grad_norm': '0.3545', 'learning_rate': '0.0006', 'epoch': '131.1'}
+
 33%|███▎      | 2490/7505 [36:14<51:02,  1.64it/s]
 33%|███▎      | 2491/7505 [36:14<51:40,  1.62it/s]
 33%|███▎      | 2492/7505 [36:15<52:05,  1.60it/s]
 33%|███▎      | 2493/7505 [36:16<52:22,  1.59it/s]
 33%|███▎      | 2494/7505 [36:16<52:34,  1.59it/s]
 33%|███▎      | 2495/7505 [36:17<52:42,  1.58it/s]
 33%|███▎      | 2496/7505 [36:18<52:47,  1.58it/s]
 33%|███▎      | 2497/7505 [36:18<52:51,  1.58it/s]
 33%|███▎      | 2498/7505 [36:19<52:52,  1.58it/s]
 33%|███▎      | 2499/7505 [36:19<52:54,  1.58it/s]
 33%|███▎      | 2500/7505 [36:20<52:55,  1.58it/s]
                                                   
{'loss': '4.303', 'grad_norm': '0.3677', 'learning_rate': '0.0006', 'epoch': '131.6'}
+
 33%|███▎      | 2500/7505 [36:20<52:55,  1.58it/s]
 33%|███▎      | 2501/7505 [36:21<52:57,  1.57it/s]
 33%|███▎      | 2502/7505 [36:21<52:58,  1.57it/s]
 33%|███▎      | 2503/7505 [36:22<52:57,  1.57it/s]
 33%|███▎      | 2504/7505 [36:23<52:57,  1.57it/s]
 33%|███▎      | 2505/7505 [36:23<52:55,  1.57it/s]
 33%|███▎      | 2506/7505 [36:24<52:54,  1.57it/s]
 33%|███▎      | 2507/7505 [36:25<52:52,  1.58it/s]
 33%|███▎      | 2508/7505 [36:25<47:00,  1.77it/s]
 33%|███▎      | 2509/7505 [36:26<50:46,  1.64it/s]
 33%|███▎      | 2510/7505 [36:26<51:23,  1.62it/s]
                                                   
{'loss': '4.337', 'grad_norm': '0.3705', 'learning_rate': '0.0006', 'epoch': '132.1'}
+
 33%|███▎      | 2510/7505 [36:26<51:23,  1.62it/s]
 33%|███▎      | 2511/7505 [36:27<51:50,  1.61it/s]
 33%|███▎      | 2512/7505 [36:28<52:08,  1.60it/s]
 33%|███▎      | 2513/7505 [36:28<52:20,  1.59it/s]
 33%|███▎      | 2514/7505 [36:29<52:29,  1.58it/s]
 34%|███▎      | 2515/7505 [36:29<52:34,  1.58it/s]
 34%|███▎      | 2516/7505 [36:30<52:38,  1.58it/s]
 34%|███▎      | 2517/7505 [36:31<52:41,  1.58it/s]
 34%|███▎      | 2518/7505 [36:31<52:43,  1.58it/s]
 34%|███▎      | 2519/7505 [36:32<52:43,  1.58it/s]
 34%|███▎      | 2520/7505 [36:33<52:43,  1.58it/s]
                                                   
{'loss': '4.272', 'grad_norm': '0.4199', 'learning_rate': '0.0006', 'epoch': '132.6'}
+
 34%|███▎      | 2520/7505 [36:33<52:43,  1.58it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 22.67it/s]
+
 15%|█▌        | 6/39 [00:00<00:01, 17.54it/s]
+
 21%|██        | 8/39 [00:00<00:01, 16.61it/s]
+
 26%|██▌       | 10/39 [00:00<00:01, 16.06it/s]
+
 31%|███       | 12/39 [00:00<00:01, 15.76it/s]
+
 36%|███▌      | 14/39 [00:00<00:01, 15.57it/s]
+
 41%|████      | 16/39 [00:00<00:01, 15.43it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 15.36it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 15.29it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 15.26it/s]
+
 62%|██████▏   | 24/39 [00:01<00:00, 15.26it/s]
+
 67%|██████▋   | 26/39 [00:01<00:00, 15.23it/s]
+
 72%|███████▏  | 28/39 [00:01<00:00, 15.23it/s]
+
 77%|███████▋  | 30/39 [00:01<00:00, 15.21it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 15.22it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 15.18it/s]
+
 92%|█████████▏| 36/39 [00:02<00:00, 15.19it/s]
+
 97%|█████████▋| 38/39 [00:02<00:00, 15.19it/s]
                                                   
+
                                               
{'eval_loss': '4.78', 'eval_runtime': '2.627', 'eval_samples_per_second': '929', 'eval_steps_per_second': '14.84', 'epoch': '132.6'}
+
 34%|███▎      | 2520/7505 [36:35<52:43,  1.58it/s]
+
100%|██████████| 39/39 [00:02<00:00, 15.19it/s][eval] epoch 132.644 step 2520 loss 4.7796 (best 4.3625 @ ep 60.644, 19/20 since best)
+
+
                                               
 34%|███▎      | 2521/7505 [36:36<1:58:18,  1.42s/it]
 34%|███▎      | 2522/7505 [36:37<1:38:35,  1.19s/it]
 34%|███▎      | 2523/7505 [36:37<1:24:49,  1.02s/it]
 34%|███▎      | 2524/7505 [36:38<1:15:11,  1.10it/s]
 34%|███▎      | 2525/7505 [36:38<1:08:25,  1.21it/s]
 34%|███▎      | 2526/7505 [36:39<1:03:40,  1.30it/s]
 34%|███▎      | 2527/7505 [36:39<54:32,  1.52it/s]  
 34%|███▎      | 2528/7505 [36:40<56:24,  1.47it/s]
 34%|███▎      | 2529/7505 [36:41<55:16,  1.50it/s]
 34%|███▎      | 2530/7505 [36:41<54:28,  1.52it/s]
                                                   
{'loss': '4.314', 'grad_norm': '0.3948', 'learning_rate': '0.0006', 'epoch': '133.2'}
+
 34%|███▎      | 2530/7505 [36:42<54:28,  1.52it/s]
 34%|███▎      | 2531/7505 [36:42<53:56,  1.54it/s]
 34%|███▎      | 2532/7505 [36:43<53:32,  1.55it/s]
 34%|███▍      | 2533/7505 [36:43<53:14,  1.56it/s]
 34%|███▍      | 2534/7505 [36:44<53:02,  1.56it/s]
 34%|███▍      | 2535/7505 [36:45<52:54,  1.57it/s]
 34%|███▍      | 2536/7505 [36:45<52:48,  1.57it/s]
 34%|███▍      | 2537/7505 [36:46<52:43,  1.57it/s]
 34%|███▍      | 2538/7505 [36:47<52:39,  1.57it/s]
 34%|███▍      | 2539/7505 [36:47<52:37,  1.57it/s]
 34%|███▍      | 2540/7505 [36:48<52:36,  1.57it/s]
                                                   
{'loss': '4.36', 'grad_norm': '0.3865', 'learning_rate': '0.0006', 'epoch': '133.7'}
+
 34%|███▍      | 2540/7505 [36:48<52:36,  1.57it/s]
 34%|███▍      | 2541/7505 [36:48<52:35,  1.57it/s]
 34%|███▍      | 2542/7505 [36:49<52:33,  1.57it/s]
 34%|███▍      | 2543/7505 [36:50<52:33,  1.57it/s]
 34%|███▍      | 2544/7505 [36:50<52:31,  1.57it/s]
 34%|███▍      | 2545/7505 [36:51<52:28,  1.58it/s]
 34%|███▍      | 2546/7505 [36:51<46:37,  1.77it/s]
 34%|███▍      | 2547/7505 [36:52<50:23,  1.64it/s]
 34%|███▍      | 2548/7505 [36:53<51:02,  1.62it/s]
 34%|███▍      | 2549/7505 [36:53<51:29,  1.60it/s]
 34%|███▍      | 2550/7505 [36:54<51:49,  1.59it/s]
                                                   
{'loss': '4.346', 'grad_norm': '0.4091', 'learning_rate': '0.0006', 'epoch': '134.2'}
+
 34%|███▍      | 2550/7505 [36:54<51:49,  1.59it/s]
 34%|███▍      | 2551/7505 [36:55<52:01,  1.59it/s]
 34%|███▍      | 2552/7505 [36:55<52:10,  1.58it/s]
 34%|███▍      | 2553/7505 [36:56<52:14,  1.58it/s]
 34%|███▍      | 2554/7505 [36:57<52:18,  1.58it/s]
 34%|███▍      | 2555/7505 [36:57<52:22,  1.58it/s]
 34%|███▍      | 2556/7505 [36:58<52:23,  1.57it/s]
 34%|███▍      | 2557/7505 [36:59<52:23,  1.57it/s]
 34%|███▍      | 2558/7505 [36:59<52:23,  1.57it/s]
 34%|███▍      | 2559/7505 [37:00<52:23,  1.57it/s]
 34%|███▍      | 2560/7505 [37:00<52:22,  1.57it/s]
                                                   
{'loss': '4.274', 'grad_norm': '0.4347', 'learning_rate': '0.0006', 'epoch': '134.8'}
+
 34%|███▍      | 2560/7505 [37:00<52:22,  1.57it/s]
 34%|███▍      | 2561/7505 [37:01<52:22,  1.57it/s]
 34%|███▍      | 2562/7505 [37:02<52:23,  1.57it/s]
 34%|███▍      | 2563/7505 [37:02<52:23,  1.57it/s]
 34%|███▍      | 2564/7505 [37:03<52:22,  1.57it/s]
 34%|███▍      | 2565/7505 [37:03<46:31,  1.77it/s]
 34%|███▍      | 2566/7505 [37:04<50:39,  1.63it/s]
 34%|███▍      | 2567/7505 [37:05<51:07,  1.61it/s]
 34%|███▍      | 2568/7505 [37:05<51:28,  1.60it/s]
 34%|███▍      | 2569/7505 [37:06<51:41,  1.59it/s]
 34%|███▍      | 2570/7505 [37:07<51:54,  1.58it/s]
                                                   
{'loss': '4.315', 'grad_norm': '0.3872', 'learning_rate': '0.0006', 'epoch': '135.3'}
+
 34%|███▍      | 2570/7505 [37:07<51:54,  1.58it/s]
 34%|███▍      | 2571/7505 [37:07<52:02,  1.58it/s]
 34%|███▍      | 2572/7505 [37:08<52:04,  1.58it/s]
 34%|███▍      | 2573/7505 [37:09<52:05,  1.58it/s]
 34%|███▍      | 2574/7505 [37:09<52:07,  1.58it/s]
 34%|███▍      | 2575/7505 [37:10<52:08,  1.58it/s]
 34%|███▍      | 2576/7505 [37:10<52:07,  1.58it/s]
 34%|███▍      | 2577/7505 [37:11<52:08,  1.58it/s]
 34%|███▍      | 2578/7505 [37:12<52:08,  1.57it/s]
 34%|███▍      | 2579/7505 [37:12<52:07,  1.58it/s]
 34%|███▍      | 2580/7505 [37:13<52:08,  1.57it/s]
                                                   
{'loss': '4.355', 'grad_norm': '0.3705', 'learning_rate': '0.0006', 'epoch': '135.8'}
+
 34%|███▍      | 2580/7505 [37:13<52:08,  1.57it/s]
 34%|███▍      | 2581/7505 [37:14<52:07,  1.57it/s]
 34%|███▍      | 2582/7505 [37:14<52:07,  1.57it/s]
 34%|███▍      | 2583/7505 [37:15<52:05,  1.57it/s]
 34%|███▍      | 2584/7505 [37:15<46:17,  1.77it/s]
 34%|███▍      | 2585/7505 [37:16<50:02,  1.64it/s]
 34%|███▍      | 2586/7505 [37:17<50:39,  1.62it/s]
 34%|███▍      | 2587/7505 [37:17<51:03,  1.61it/s]
 34%|███▍      | 2588/7505 [37:18<51:22,  1.60it/s]
 34%|███▍      | 2589/7505 [37:19<51:33,  1.59it/s]
 35%|███▍      | 2590/7505 [37:19<51:45,  1.58it/s]
                                                   
{'loss': '4.256', 'grad_norm': '0.4266', 'learning_rate': '0.0006', 'epoch': '136.3'}
+
 35%|███▍      | 2590/7505 [37:19<51:45,  1.58it/s]
 35%|███▍      | 2591/7505 [37:20<51:51,  1.58it/s]
 35%|███▍      | 2592/7505 [37:20<51:54,  1.58it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 23.56it/s]
+
 15%|█▌        | 6/39 [00:00<00:01, 17.74it/s]
+
 21%|██        | 8/39 [00:00<00:01, 16.72it/s]
+
 26%|██▌       | 10/39 [00:00<00:01, 16.16it/s]
+
 31%|███       | 12/39 [00:00<00:01, 15.81it/s]
+
 36%|███▌      | 14/39 [00:00<00:01, 15.59it/s]
+
 41%|████      | 16/39 [00:00<00:01, 15.45it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 15.37it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 15.30it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 15.26it/s]
+
 62%|██████▏   | 24/39 [00:01<00:00, 15.26it/s]
+
 67%|██████▋   | 26/39 [00:01<00:00, 15.25it/s]
+
 72%|███████▏  | 28/39 [00:01<00:00, 15.22it/s]
+
 77%|███████▋  | 30/39 [00:01<00:00, 15.21it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 15.20it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 15.20it/s]
+
 92%|█████████▏| 36/39 [00:02<00:00, 15.18it/s]
+
 97%|█████████▋| 38/39 [00:02<00:00, 15.19it/s]
                                                   
+
                                               
{'eval_loss': '4.829', 'eval_runtime': '2.62', 'eval_samples_per_second': '931.6', 'eval_steps_per_second': '14.88', 'epoch': '136.4'}
+
 35%|███▍      | 2592/7505 [37:23<51:54,  1.58it/s]
+
100%|██████████| 39/39 [00:02<00:00, 15.19it/s][eval] epoch 136.43 step 2592 loss 4.8295 (best 4.3625 @ ep 60.644, 20/20 since best)
+
+
                                               [transformers] 
+
+Training completed. Do not forget to share your model on huggingface.co/models =)
+
+
+
                                                   
{'train_runtime': '2245', 'train_samples_per_second': '1678', 'train_steps_per_second': '3.343', 'train_loss': '5.338', 'epoch': '136.4'}
+
 35%|███▍      | 2592/7505 [37:23<51:54,  1.58it/s]
 35%|███▍      | 2592/7505 [37:23<1:10:52,  1.16it/s]
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+wandb: 
+wandb: 🚀 View run t2a_19M_C3_rand5_r2l50_i5exp at: https://wandb.ai/slowrun-wankers/data-aug-pretraining/runs/obs138g0
+wandb: Find logs at: wandb/run-20260818_182149-obs138g0/logs
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+[rank0]:[W818 18:59:59.117821772 ProcessGroupNCCL.cpp:1575] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator())
diff --git a/runs/t2a/t2a_300M_C0_baseline/checkpoint-5688/chat_template.jinja b/runs/t2a/t2a_300M_C0_baseline/checkpoint-5688/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..699ff8df401fe4788525e9c1f9b86a99eadd6230
--- /dev/null
+++ b/runs/t2a/t2a_300M_C0_baseline/checkpoint-5688/chat_template.jinja
@@ -0,0 +1,85 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0].role == 'system' %}
+        {{- messages[0].content + '\n\n' }}
+    {%- endif %}
+    {{- "# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0].content + '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
+{%- for message in messages[::-1] %}
+    {%- set index = (messages|length - 1) - loop.index0 %}
+    {%- if ns.multi_step_tool and message.role == "user" and not(message.content.startswith('<tool_response>') and message.content.endswith('</tool_response>')) %}
+        {%- set ns.multi_step_tool = false %}
+        {%- set ns.last_query_index = index %}
+    {%- endif %}
+{%- endfor %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {%- set content = message.content %}
+        {%- set reasoning_content = '' %}
+        {%- if message.reasoning_content is defined and message.reasoning_content is not none %}
+            {%- set reasoning_content = message.reasoning_content %}
+        {%- else %}
+            {%- if '</think>' in message.content %}
+                {%- set content = message.content.split('</think>')[-1].lstrip('\n') %}
+                {%- set reasoning_content = message.content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
+            {%- endif %}
+        {%- endif %}
+        {%- if loop.index0 > ns.last_query_index %}
+            {%- if loop.last or (not loop.last and reasoning_content) %}
+                {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content.strip('\n') + '\n</think>\n\n' + content.lstrip('\n') }}
+            {%- else %}
+                {{- '<|im_start|>' + message.role + '\n' + content }}
+            {%- endif %}
+        {%- else %}
+            {{- '<|im_start|>' + message.role + '\n' + content }}
+        {%- endif %}
+        {%- if message.tool_calls %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if (loop.first and content) or (not loop.first) %}
+                    {{- '\n' }}
+                {%- endif %}
+                {%- if tool_call.function %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {{- '<tool_call>\n{"name": "' }}
+                {{- tool_call.name }}
+                {{- '", "arguments": ' }}
+                {%- if tool_call.arguments is string %}
+                    {{- tool_call.arguments }}
+                {%- else %}
+                    {{- tool_call.arguments | tojson }}
+                {%- endif %}
+                {{- '}\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.first or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+    {%- if enable_thinking is defined and enable_thinking is false %}
+        {{- '<think>\n\n</think>\n\n' }}
+    {%- endif %}
+{%- endif %}
\ No newline at end of file
diff --git a/runs/t2a/t2a_300M_C0_baseline/checkpoint-5688/config.json b/runs/t2a/t2a_300M_C0_baseline/checkpoint-5688/config.json
new file mode 100644
index 0000000000000000000000000000000000000000..f8ca584a0751672f280da7c54ff3916cdf24d6f4
--- /dev/null
+++ b/runs/t2a/t2a_300M_C0_baseline/checkpoint-5688/config.json
@@ -0,0 +1,32 @@
+{
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": null,
+  "dtype": "float32",
+  "eos_token_id": 151645,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 512,
+  "initializer_range": 0.02,
+  "intermediate_size": 1536,
+  "max_position_embeddings": 2048,
+  "mlp_bias": false,
+  "model_type": "llama",
+  "num_attention_heads": 4,
+  "num_hidden_layers": 20,
+  "num_key_value_heads": 4,
+  "pad_token_id": 151645,
+  "pretraining_tp": 1,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 10000.0,
+    "rope_type": "default"
+  },
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": false,
+  "vocab_size": 151671
+}
diff --git a/runs/t2a/t2a_300M_C0_baseline/checkpoint-5688/generation_config.json b/runs/t2a/t2a_300M_C0_baseline/checkpoint-5688/generation_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..8c4242f32d40df256857f87264f646130f684e33
--- /dev/null
+++ b/runs/t2a/t2a_300M_C0_baseline/checkpoint-5688/generation_config.json
@@ -0,0 +1,11 @@
+{
+  "_from_model_config": true,
+  "eos_token_id": [
+    151645
+  ],
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "transformers_version": "5.15.0",
+  "use_cache": true
+}
diff --git a/runs/t2a/t2a_300M_C0_baseline/checkpoint-5688/tokenizer_config.json b/runs/t2a/t2a_300M_C0_baseline/checkpoint-5688/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..839a3743177a3ed77e1ffbae67520e8500ca0532
--- /dev/null
+++ b/runs/t2a/t2a_300M_C0_baseline/checkpoint-5688/tokenizer_config.json
@@ -0,0 +1,19 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|l2r_pred|>",
+    "<|r2l_pred|>"
+  ],
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 131072,
+  "pad_token": "<|im_end|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}
diff --git a/runs/t2a/t2a_300M_C0_baseline/checkpoint-5688/trainer_state.json b/runs/t2a/t2a_300M_C0_baseline/checkpoint-5688/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..15c4b183ed9b346cc53ce0c41822eaeebd105d1f
--- /dev/null
+++ b/runs/t2a/t2a_300M_C0_baseline/checkpoint-5688/trainer_state.json
@@ -0,0 +1,4649 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 19.821319353429445,
+  "eval_steps": 72,
+  "global_step": 5688,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0034949759720401923,
+      "grad_norm": 1.8103264570236206,
+      "learning_rate": 0.0,
+      "loss": 12.027359962463379,
+      "step": 1
+    },
+    {
+      "epoch": 0.03494975972040192,
+      "grad_norm": 1.7561959028244019,
+      "learning_rate": 5.399999999999999e-05,
+      "loss": 11.73574490017361,
+      "step": 10
+    },
+    {
+      "epoch": 0.06989951944080385,
+      "grad_norm": 1.68845534324646,
+      "learning_rate": 0.00011399999999999999,
+      "loss": 11.051107788085938,
+      "step": 20
+    },
+    {
+      "epoch": 0.10484927916120576,
+      "grad_norm": 1.67290198802948,
+      "learning_rate": 0.00017399999999999997,
+      "loss": 10.294818115234374,
+      "step": 30
+    },
+    {
+      "epoch": 0.1397990388816077,
+      "grad_norm": 1.539352297782898,
+      "learning_rate": 0.000234,
+      "loss": 9.248680114746094,
+      "step": 40
+    },
+    {
+      "epoch": 0.17474879860200962,
+      "grad_norm": 0.9665805697441101,
+      "learning_rate": 0.000294,
+      "loss": 8.243627166748047,
+      "step": 50
+    },
+    {
+      "epoch": 0.20969855832241152,
+      "grad_norm": 0.2849996089935303,
+      "learning_rate": 0.00035399999999999993,
+      "loss": 7.690145111083984,
+      "step": 60
+    },
+    {
+      "epoch": 0.24464831804281345,
+      "grad_norm": 1.6108201742172241,
+      "learning_rate": 0.0004139999999999999,
+      "loss": 7.504912567138672,
+      "step": 70
+    },
+    {
+      "epoch": 0.2516382699868938,
+      "eval_loss": 7.463537216186523,
+      "eval_runtime": 27.9811,
+      "eval_samples_per_second": 87.237,
+      "eval_steps_per_second": 1.394,
+      "step": 72
+    },
+    {
+      "epoch": 0.2795980777632154,
+      "grad_norm": 0.4956258535385132,
+      "learning_rate": 0.000474,
+      "loss": 7.347062683105468,
+      "step": 80
+    },
+    {
+      "epoch": 0.3145478374836173,
+      "grad_norm": 0.2999236285686493,
+      "learning_rate": 0.000534,
+      "loss": 7.1412605285644535,
+      "step": 90
+    },
+    {
+      "epoch": 0.34949759720401924,
+      "grad_norm": 0.8744467496871948,
+      "learning_rate": 0.0005939999999999999,
+      "loss": 6.99285659790039,
+      "step": 100
+    },
+    {
+      "epoch": 0.38444735692442117,
+      "grad_norm": 0.5292134881019592,
+      "learning_rate": 0.0006,
+      "loss": 6.847041320800781,
+      "step": 110
+    },
+    {
+      "epoch": 0.41939711664482304,
+      "grad_norm": 1.0331865549087524,
+      "learning_rate": 0.0006,
+      "loss": 6.714336395263672,
+      "step": 120
+    },
+    {
+      "epoch": 0.45434687636522497,
+      "grad_norm": 1.0205185413360596,
+      "learning_rate": 0.0006,
+      "loss": 6.611071014404297,
+      "step": 130
+    },
+    {
+      "epoch": 0.4892966360856269,
+      "grad_norm": 0.7076956629753113,
+      "learning_rate": 0.0006,
+      "loss": 6.489197540283203,
+      "step": 140
+    },
+    {
+      "epoch": 0.5032765399737876,
+      "eval_loss": 6.43985652923584,
+      "eval_runtime": 3.2682,
+      "eval_samples_per_second": 746.895,
+      "eval_steps_per_second": 11.933,
+      "step": 144
+    },
+    {
+      "epoch": 0.5242463958060288,
+      "grad_norm": 0.9724865555763245,
+      "learning_rate": 0.0006,
+      "loss": 6.3917724609375,
+      "step": 150
+    },
+    {
+      "epoch": 0.5591961555264308,
+      "grad_norm": 0.5225860476493835,
+      "learning_rate": 0.0006,
+      "loss": 6.300282669067383,
+      "step": 160
+    },
+    {
+      "epoch": 0.5941459152468327,
+      "grad_norm": 0.5105811953544617,
+      "learning_rate": 0.0006,
+      "loss": 6.2310630798339846,
+      "step": 170
+    },
+    {
+      "epoch": 0.6290956749672346,
+      "grad_norm": 0.3318553566932678,
+      "learning_rate": 0.0006,
+      "loss": 6.132600021362305,
+      "step": 180
+    },
+    {
+      "epoch": 0.6640454346876365,
+      "grad_norm": 0.6217828989028931,
+      "learning_rate": 0.0006,
+      "loss": 6.086975479125977,
+      "step": 190
+    },
+    {
+      "epoch": 0.6989951944080385,
+      "grad_norm": 0.47685033082962036,
+      "learning_rate": 0.0006,
+      "loss": 6.034068298339844,
+      "step": 200
+    },
+    {
+      "epoch": 0.7339449541284404,
+      "grad_norm": 0.7540782690048218,
+      "learning_rate": 0.0006,
+      "loss": 5.938161468505859,
+      "step": 210
+    },
+    {
+      "epoch": 0.7549148099606815,
+      "eval_loss": 5.9513702392578125,
+      "eval_runtime": 3.2628,
+      "eval_samples_per_second": 748.133,
+      "eval_steps_per_second": 11.953,
+      "step": 216
+    },
+    {
+      "epoch": 0.7688947138488423,
+      "grad_norm": 1.2809175252914429,
+      "learning_rate": 0.0006,
+      "loss": 5.887511444091797,
+      "step": 220
+    },
+    {
+      "epoch": 0.8038444735692443,
+      "grad_norm": 0.4060598909854889,
+      "learning_rate": 0.0006,
+      "loss": 5.8352806091308596,
+      "step": 230
+    },
+    {
+      "epoch": 0.8387942332896461,
+      "grad_norm": 1.100441336631775,
+      "learning_rate": 0.0006,
+      "loss": 5.752750396728516,
+      "step": 240
+    },
+    {
+      "epoch": 0.873743993010048,
+      "grad_norm": 0.5426212549209595,
+      "learning_rate": 0.0006,
+      "loss": 5.726037216186524,
+      "step": 250
+    },
+    {
+      "epoch": 0.9086937527304499,
+      "grad_norm": 0.6426259875297546,
+      "learning_rate": 0.0006,
+      "loss": 5.642111587524414,
+      "step": 260
+    },
+    {
+      "epoch": 0.9436435124508519,
+      "grad_norm": 0.5986626148223877,
+      "learning_rate": 0.0006,
+      "loss": 5.604520797729492,
+      "step": 270
+    },
+    {
+      "epoch": 0.9785932721712538,
+      "grad_norm": 0.5398047566413879,
+      "learning_rate": 0.0006,
+      "loss": 5.5412956237792965,
+      "step": 280
+    },
+    {
+      "epoch": 1.0034949759720402,
+      "eval_loss": 5.596405506134033,
+      "eval_runtime": 3.2459,
+      "eval_samples_per_second": 752.023,
+      "eval_steps_per_second": 12.015,
+      "step": 288
+    },
+    {
+      "epoch": 1.0104849279161205,
+      "grad_norm": 0.8156829476356506,
+      "learning_rate": 0.0006,
+      "loss": 5.484656524658203,
+      "step": 290
+    },
+    {
+      "epoch": 1.0454346876365226,
+      "grad_norm": 0.5032617449760437,
+      "learning_rate": 0.0006,
+      "loss": 5.479055786132813,
+      "step": 300
+    },
+    {
+      "epoch": 1.0803844473569244,
+      "grad_norm": 0.25261226296424866,
+      "learning_rate": 0.0006,
+      "loss": 5.3620353698730465,
+      "step": 310
+    },
+    {
+      "epoch": 1.1153342070773264,
+      "grad_norm": 0.5416950583457947,
+      "learning_rate": 0.0006,
+      "loss": 5.3263084411621096,
+      "step": 320
+    },
+    {
+      "epoch": 1.1502839667977283,
+      "grad_norm": 0.5336521863937378,
+      "learning_rate": 0.0006,
+      "loss": 5.2896873474121096,
+      "step": 330
+    },
+    {
+      "epoch": 1.1852337265181303,
+      "grad_norm": 0.5503013134002686,
+      "learning_rate": 0.0006,
+      "loss": 5.233787918090821,
+      "step": 340
+    },
+    {
+      "epoch": 1.2201834862385321,
+      "grad_norm": 0.46227598190307617,
+      "learning_rate": 0.0006,
+      "loss": 5.189003753662109,
+      "step": 350
+    },
+    {
+      "epoch": 1.255133245958934,
+      "grad_norm": 0.34390494227409363,
+      "learning_rate": 0.0006,
+      "loss": 5.134808731079102,
+      "step": 360
+    },
+    {
+      "epoch": 1.255133245958934,
+      "eval_loss": 5.222589015960693,
+      "eval_runtime": 3.2655,
+      "eval_samples_per_second": 747.514,
+      "eval_steps_per_second": 11.943,
+      "step": 360
+    },
+    {
+      "epoch": 1.290083005679336,
+      "grad_norm": 0.7792335152626038,
+      "learning_rate": 0.0006,
+      "loss": 5.0798194885253904,
+      "step": 370
+    },
+    {
+      "epoch": 1.3250327653997378,
+      "grad_norm": 0.9799200892448425,
+      "learning_rate": 0.0006,
+      "loss": 5.088294982910156,
+      "step": 380
+    },
+    {
+      "epoch": 1.3599825251201398,
+      "grad_norm": 0.3464367687702179,
+      "learning_rate": 0.0006,
+      "loss": 5.027822494506836,
+      "step": 390
+    },
+    {
+      "epoch": 1.3949322848405417,
+      "grad_norm": 0.2372007817029953,
+      "learning_rate": 0.0006,
+      "loss": 4.9623668670654295,
+      "step": 400
+    },
+    {
+      "epoch": 1.4298820445609437,
+      "grad_norm": 0.9554475545883179,
+      "learning_rate": 0.0006,
+      "loss": 4.910638427734375,
+      "step": 410
+    },
+    {
+      "epoch": 1.4648318042813455,
+      "grad_norm": 0.7165559530258179,
+      "learning_rate": 0.0006,
+      "loss": 4.910945129394531,
+      "step": 420
+    },
+    {
+      "epoch": 1.4997815640017476,
+      "grad_norm": 0.34058812260627747,
+      "learning_rate": 0.0006,
+      "loss": 4.862317657470703,
+      "step": 430
+    },
+    {
+      "epoch": 1.506771515945828,
+      "eval_loss": 4.936007976531982,
+      "eval_runtime": 3.2616,
+      "eval_samples_per_second": 748.417,
+      "eval_steps_per_second": 11.958,
+      "step": 432
+    },
+    {
+      "epoch": 1.5347313237221494,
+      "grad_norm": 0.5273450613021851,
+      "learning_rate": 0.0006,
+      "loss": 4.783837127685547,
+      "step": 440
+    },
+    {
+      "epoch": 1.5696810834425512,
+      "grad_norm": 0.43669116497039795,
+      "learning_rate": 0.0006,
+      "loss": 4.758808517456055,
+      "step": 450
+    },
+    {
+      "epoch": 1.6046308431629532,
+      "grad_norm": 0.4376815855503082,
+      "learning_rate": 0.0006,
+      "loss": 4.690071105957031,
+      "step": 460
+    },
+    {
+      "epoch": 1.6395806028833553,
+      "grad_norm": 0.43130210041999817,
+      "learning_rate": 0.0006,
+      "loss": 4.665610504150391,
+      "step": 470
+    },
+    {
+      "epoch": 1.674530362603757,
+      "grad_norm": 0.5736966729164124,
+      "learning_rate": 0.0006,
+      "loss": 4.625357055664063,
+      "step": 480
+    },
+    {
+      "epoch": 1.709480122324159,
+      "grad_norm": 0.5190726518630981,
+      "learning_rate": 0.0006,
+      "loss": 4.6028491973876955,
+      "step": 490
+    },
+    {
+      "epoch": 1.744429882044561,
+      "grad_norm": 0.28841209411621094,
+      "learning_rate": 0.0006,
+      "loss": 4.553189086914062,
+      "step": 500
+    },
+    {
+      "epoch": 1.7584097859327217,
+      "eval_loss": 4.659235000610352,
+      "eval_runtime": 3.2621,
+      "eval_samples_per_second": 748.282,
+      "eval_steps_per_second": 11.955,
+      "step": 504
+    },
+    {
+      "epoch": 1.779379641764963,
+      "grad_norm": 0.3278319537639618,
+      "learning_rate": 0.0006,
+      "loss": 4.5001678466796875,
+      "step": 510
+    },
+    {
+      "epoch": 1.8143294014853648,
+      "grad_norm": 0.3519146740436554,
+      "learning_rate": 0.0006,
+      "loss": 4.465358734130859,
+      "step": 520
+    },
+    {
+      "epoch": 1.8492791612057666,
+      "grad_norm": 0.6919175982475281,
+      "learning_rate": 0.0006,
+      "loss": 4.433049011230469,
+      "step": 530
+    },
+    {
+      "epoch": 1.8842289209261687,
+      "grad_norm": 0.3892306983470917,
+      "learning_rate": 0.0006,
+      "loss": 4.427141571044922,
+      "step": 540
+    },
+    {
+      "epoch": 1.9191786806465707,
+      "grad_norm": 0.39456647634506226,
+      "learning_rate": 0.0006,
+      "loss": 4.387229156494141,
+      "step": 550
+    },
+    {
+      "epoch": 1.9541284403669725,
+      "grad_norm": 0.3719537854194641,
+      "learning_rate": 0.0006,
+      "loss": 4.344993972778321,
+      "step": 560
+    },
+    {
+      "epoch": 1.9890782000873743,
+      "grad_norm": 0.3394317030906677,
+      "learning_rate": 0.0006,
+      "loss": 4.321841430664063,
+      "step": 570
+    },
+    {
+      "epoch": 2.0069899519440804,
+      "eval_loss": 4.468684196472168,
+      "eval_runtime": 3.2528,
+      "eval_samples_per_second": 750.441,
+      "eval_steps_per_second": 11.99,
+      "step": 576
+    },
+    {
+      "epoch": 2.020969855832241,
+      "grad_norm": 0.49324706196784973,
+      "learning_rate": 0.0006,
+      "loss": 4.296000671386719,
+      "step": 580
+    },
+    {
+      "epoch": 2.055919615552643,
+      "grad_norm": 0.40804269909858704,
+      "learning_rate": 0.0006,
+      "loss": 4.260264205932617,
+      "step": 590
+    },
+    {
+      "epoch": 2.090869375273045,
+      "grad_norm": 0.25117993354797363,
+      "learning_rate": 0.0006,
+      "loss": 4.231197357177734,
+      "step": 600
+    },
+    {
+      "epoch": 2.125819134993447,
+      "grad_norm": 0.4513896107673645,
+      "learning_rate": 0.0006,
+      "loss": 4.214814376831055,
+      "step": 610
+    },
+    {
+      "epoch": 2.160768894713849,
+      "grad_norm": 0.42641469836235046,
+      "learning_rate": 0.0006,
+      "loss": 4.201937866210938,
+      "step": 620
+    },
+    {
+      "epoch": 2.1957186544342506,
+      "grad_norm": 0.3763907849788666,
+      "learning_rate": 0.0006,
+      "loss": 4.169511032104492,
+      "step": 630
+    },
+    {
+      "epoch": 2.230668414154653,
+      "grad_norm": 0.2426532804965973,
+      "learning_rate": 0.0006,
+      "loss": 4.150532150268555,
+      "step": 640
+    },
+    {
+      "epoch": 2.2586282219309743,
+      "eval_loss": 4.305553436279297,
+      "eval_runtime": 3.2473,
+      "eval_samples_per_second": 751.699,
+      "eval_steps_per_second": 12.01,
+      "step": 648
+    },
+    {
+      "epoch": 2.2656181738750547,
+      "grad_norm": 0.2809857130050659,
+      "learning_rate": 0.0006,
+      "loss": 4.1320953369140625,
+      "step": 650
+    },
+    {
+      "epoch": 2.3005679335954565,
+      "grad_norm": 0.3201979994773865,
+      "learning_rate": 0.0006,
+      "loss": 4.094313049316407,
+      "step": 660
+    },
+    {
+      "epoch": 2.3355176933158583,
+      "grad_norm": 0.4085295498371124,
+      "learning_rate": 0.0006,
+      "loss": 4.0859321594238285,
+      "step": 670
+    },
+    {
+      "epoch": 2.3704674530362606,
+      "grad_norm": 0.37784844636917114,
+      "learning_rate": 0.0006,
+      "loss": 4.077365112304688,
+      "step": 680
+    },
+    {
+      "epoch": 2.4054172127566624,
+      "grad_norm": 0.3630412220954895,
+      "learning_rate": 0.0006,
+      "loss": 4.085600280761719,
+      "step": 690
+    },
+    {
+      "epoch": 2.4403669724770642,
+      "grad_norm": 0.2659319341182709,
+      "learning_rate": 0.0006,
+      "loss": 4.0376838684082035,
+      "step": 700
+    },
+    {
+      "epoch": 2.475316732197466,
+      "grad_norm": 0.3372329771518707,
+      "learning_rate": 0.0006,
+      "loss": 4.028413391113281,
+      "step": 710
+    },
+    {
+      "epoch": 2.510266491917868,
+      "grad_norm": 0.3702462315559387,
+      "learning_rate": 0.0006,
+      "loss": 4.04058609008789,
+      "step": 720
+    },
+    {
+      "epoch": 2.510266491917868,
+      "eval_loss": 4.205882549285889,
+      "eval_runtime": 3.2492,
+      "eval_samples_per_second": 751.272,
+      "eval_steps_per_second": 12.003,
+      "step": 720
+    },
+    {
+      "epoch": 2.54521625163827,
+      "grad_norm": 0.353391170501709,
+      "learning_rate": 0.0006,
+      "loss": 4.026080322265625,
+      "step": 730
+    },
+    {
+      "epoch": 2.580166011358672,
+      "grad_norm": 0.31685870885849,
+      "learning_rate": 0.0006,
+      "loss": 3.9829071044921873,
+      "step": 740
+    },
+    {
+      "epoch": 2.615115771079074,
+      "grad_norm": 0.3026818335056305,
+      "learning_rate": 0.0006,
+      "loss": 3.974071502685547,
+      "step": 750
+    },
+    {
+      "epoch": 2.6500655307994756,
+      "grad_norm": 0.38419532775878906,
+      "learning_rate": 0.0006,
+      "loss": 3.9538246154785157,
+      "step": 760
+    },
+    {
+      "epoch": 2.6850152905198774,
+      "grad_norm": 0.3027503192424774,
+      "learning_rate": 0.0006,
+      "loss": 3.9572322845458983,
+      "step": 770
+    },
+    {
+      "epoch": 2.7199650502402797,
+      "grad_norm": 0.35563647747039795,
+      "learning_rate": 0.0006,
+      "loss": 3.9424922943115233,
+      "step": 780
+    },
+    {
+      "epoch": 2.7549148099606815,
+      "grad_norm": 0.30176934599876404,
+      "learning_rate": 0.0006,
+      "loss": 3.9279945373535154,
+      "step": 790
+    },
+    {
+      "epoch": 2.761904761904762,
+      "eval_loss": 4.107356548309326,
+      "eval_runtime": 3.2551,
+      "eval_samples_per_second": 749.899,
+      "eval_steps_per_second": 11.981,
+      "step": 792
+    },
+    {
+      "epoch": 2.7898645696810833,
+      "grad_norm": 0.26767697930336,
+      "learning_rate": 0.0006,
+      "loss": 3.9147575378417967,
+      "step": 800
+    },
+    {
+      "epoch": 2.8248143294014856,
+      "grad_norm": 0.36032038927078247,
+      "learning_rate": 0.0006,
+      "loss": 3.8961360931396483,
+      "step": 810
+    },
+    {
+      "epoch": 2.8597640891218874,
+      "grad_norm": 0.3800591826438904,
+      "learning_rate": 0.0006,
+      "loss": 3.9047950744628905,
+      "step": 820
+    },
+    {
+      "epoch": 2.894713848842289,
+      "grad_norm": 0.2939601242542267,
+      "learning_rate": 0.0006,
+      "loss": 3.8852943420410155,
+      "step": 830
+    },
+    {
+      "epoch": 2.929663608562691,
+      "grad_norm": 0.3686358630657196,
+      "learning_rate": 0.0006,
+      "loss": 3.8656280517578123,
+      "step": 840
+    },
+    {
+      "epoch": 2.964613368283093,
+      "grad_norm": 0.25963079929351807,
+      "learning_rate": 0.0006,
+      "loss": 3.8637855529785154,
+      "step": 850
+    },
+    {
+      "epoch": 2.999563128003495,
+      "grad_norm": 0.282455712556839,
+      "learning_rate": 0.0006,
+      "loss": 3.862845611572266,
+      "step": 860
+    },
+    {
+      "epoch": 3.0104849279161208,
+      "eval_loss": 4.062587261199951,
+      "eval_runtime": 3.2512,
+      "eval_samples_per_second": 750.79,
+      "eval_steps_per_second": 11.995,
+      "step": 864
+    },
+    {
+      "epoch": 3.031454783748362,
+      "grad_norm": 0.40367621183395386,
+      "learning_rate": 0.0006,
+      "loss": 3.8319114685058593,
+      "step": 870
+    },
+    {
+      "epoch": 3.0664045434687637,
+      "grad_norm": 0.35334888100624084,
+      "learning_rate": 0.0006,
+      "loss": 3.8218406677246093,
+      "step": 880
+    },
+    {
+      "epoch": 3.1013543031891655,
+      "grad_norm": 0.3136698007583618,
+      "learning_rate": 0.0006,
+      "loss": 3.797529602050781,
+      "step": 890
+    },
+    {
+      "epoch": 3.1363040629095673,
+      "grad_norm": 0.2592521011829376,
+      "learning_rate": 0.0006,
+      "loss": 3.778206634521484,
+      "step": 900
+    },
+    {
+      "epoch": 3.1712538226299696,
+      "grad_norm": 0.3828992545604706,
+      "learning_rate": 0.0006,
+      "loss": 3.7767139434814454,
+      "step": 910
+    },
+    {
+      "epoch": 3.2062035823503714,
+      "grad_norm": 0.3110186457633972,
+      "learning_rate": 0.0006,
+      "loss": 3.7636386871337892,
+      "step": 920
+    },
+    {
+      "epoch": 3.241153342070773,
+      "grad_norm": 0.23242363333702087,
+      "learning_rate": 0.0006,
+      "loss": 3.7548202514648437,
+      "step": 930
+    },
+    {
+      "epoch": 3.2621231979030143,
+      "eval_loss": 3.974292516708374,
+      "eval_runtime": 3.2458,
+      "eval_samples_per_second": 752.055,
+      "eval_steps_per_second": 12.016,
+      "step": 936
+    },
+    {
+      "epoch": 3.276103101791175,
+      "grad_norm": 0.26155588030815125,
+      "learning_rate": 0.0006,
+      "loss": 3.750012969970703,
+      "step": 940
+    },
+    {
+      "epoch": 3.3110528615115773,
+      "grad_norm": 0.23901717364788055,
+      "learning_rate": 0.0006,
+      "loss": 3.737981414794922,
+      "step": 950
+    },
+    {
+      "epoch": 3.346002621231979,
+      "grad_norm": 0.366420179605484,
+      "learning_rate": 0.0006,
+      "loss": 3.7385795593261717,
+      "step": 960
+    },
+    {
+      "epoch": 3.380952380952381,
+      "grad_norm": 0.25742533802986145,
+      "learning_rate": 0.0006,
+      "loss": 3.749012756347656,
+      "step": 970
+    },
+    {
+      "epoch": 3.4159021406727827,
+      "grad_norm": 0.29228854179382324,
+      "learning_rate": 0.0006,
+      "loss": 3.727323532104492,
+      "step": 980
+    },
+    {
+      "epoch": 3.450851900393185,
+      "grad_norm": 0.2535940408706665,
+      "learning_rate": 0.0006,
+      "loss": 3.7173316955566404,
+      "step": 990
+    },
+    {
+      "epoch": 3.485801660113587,
+      "grad_norm": 0.376825213432312,
+      "learning_rate": 0.0006,
+      "loss": 3.7095382690429686,
+      "step": 1000
+    },
+    {
+      "epoch": 3.5137614678899083,
+      "eval_loss": 3.941437005996704,
+      "eval_runtime": 3.261,
+      "eval_samples_per_second": 748.538,
+      "eval_steps_per_second": 11.959,
+      "step": 1008
+    },
+    {
+      "epoch": 3.5207514198339886,
+      "grad_norm": 0.43542033433914185,
+      "learning_rate": 0.0006,
+      "loss": 3.7163040161132814,
+      "step": 1010
+    },
+    {
+      "epoch": 3.5557011795543905,
+      "grad_norm": 0.27645036578178406,
+      "learning_rate": 0.0006,
+      "loss": 3.722352981567383,
+      "step": 1020
+    },
+    {
+      "epoch": 3.5906509392747923,
+      "grad_norm": 0.28030532598495483,
+      "learning_rate": 0.0006,
+      "loss": 3.7020999908447267,
+      "step": 1030
+    },
+    {
+      "epoch": 3.6256006989951945,
+      "grad_norm": 0.24973303079605103,
+      "learning_rate": 0.0006,
+      "loss": 3.7000885009765625,
+      "step": 1040
+    },
+    {
+      "epoch": 3.6605504587155964,
+      "grad_norm": 0.22721338272094727,
+      "learning_rate": 0.0006,
+      "loss": 3.6920379638671874,
+      "step": 1050
+    },
+    {
+      "epoch": 3.695500218435998,
+      "grad_norm": 0.28152787685394287,
+      "learning_rate": 0.0006,
+      "loss": 3.6847507476806642,
+      "step": 1060
+    },
+    {
+      "epoch": 3.7304499781564004,
+      "grad_norm": 0.30779653787612915,
+      "learning_rate": 0.0006,
+      "loss": 3.6724178314208986,
+      "step": 1070
+    },
+    {
+      "epoch": 3.765399737876802,
+      "grad_norm": 0.26307567954063416,
+      "learning_rate": 0.0006,
+      "loss": 3.6704265594482424,
+      "step": 1080
+    },
+    {
+      "epoch": 3.765399737876802,
+      "eval_loss": 3.88352108001709,
+      "eval_runtime": 3.2493,
+      "eval_samples_per_second": 751.245,
+      "eval_steps_per_second": 12.003,
+      "step": 1080
+    },
+    {
+      "epoch": 3.800349497597204,
+      "grad_norm": 0.3203473389148712,
+      "learning_rate": 0.0006,
+      "loss": 3.6614990234375,
+      "step": 1090
+    },
+    {
+      "epoch": 3.835299257317606,
+      "grad_norm": 0.2821556031703949,
+      "learning_rate": 0.0006,
+      "loss": 3.6521583557128907,
+      "step": 1100
+    },
+    {
+      "epoch": 3.8702490170380077,
+      "grad_norm": 0.25545603036880493,
+      "learning_rate": 0.0006,
+      "loss": 3.6598255157470705,
+      "step": 1110
+    },
+    {
+      "epoch": 3.90519877675841,
+      "grad_norm": 0.2474914938211441,
+      "learning_rate": 0.0006,
+      "loss": 3.6521820068359374,
+      "step": 1120
+    },
+    {
+      "epoch": 3.940148536478812,
+      "grad_norm": 0.35863569378852844,
+      "learning_rate": 0.0006,
+      "loss": 3.641748046875,
+      "step": 1130
+    },
+    {
+      "epoch": 3.9750982961992136,
+      "grad_norm": 0.27923640608787537,
+      "learning_rate": 0.0006,
+      "loss": 3.6296718597412108,
+      "step": 1140
+    },
+    {
+      "epoch": 4.006989951944081,
+      "grad_norm": 0.40943944454193115,
+      "learning_rate": 0.0006,
+      "loss": 3.6224037170410157,
+      "step": 1150
+    },
+    {
+      "epoch": 4.013979903888161,
+      "eval_loss": 3.8851940631866455,
+      "eval_runtime": 3.2447,
+      "eval_samples_per_second": 752.295,
+      "eval_steps_per_second": 12.019,
+      "step": 1152
+    },
+    {
+      "epoch": 4.041939711664482,
+      "grad_norm": 0.3625980317592621,
+      "learning_rate": 0.0006,
+      "loss": 3.6101612091064452,
+      "step": 1160
+    },
+    {
+      "epoch": 4.0768894713848844,
+      "grad_norm": 0.2343684732913971,
+      "learning_rate": 0.0006,
+      "loss": 3.5837188720703126,
+      "step": 1170
+    },
+    {
+      "epoch": 4.111839231105286,
+      "grad_norm": 0.24874775111675262,
+      "learning_rate": 0.0006,
+      "loss": 3.5648468017578123,
+      "step": 1180
+    },
+    {
+      "epoch": 4.146788990825688,
+      "grad_norm": 0.22182610630989075,
+      "learning_rate": 0.0006,
+      "loss": 3.5701034545898436,
+      "step": 1190
+    },
+    {
+      "epoch": 4.18173875054609,
+      "grad_norm": 0.27770084142684937,
+      "learning_rate": 0.0006,
+      "loss": 3.569261169433594,
+      "step": 1200
+    },
+    {
+      "epoch": 4.216688510266492,
+      "grad_norm": 0.2494732290506363,
+      "learning_rate": 0.0006,
+      "loss": 3.5702476501464844,
+      "step": 1210
+    },
+    {
+      "epoch": 4.251638269986894,
+      "grad_norm": 0.25190049409866333,
+      "learning_rate": 0.0006,
+      "loss": 3.5553802490234374,
+      "step": 1220
+    },
+    {
+      "epoch": 4.265618173875055,
+      "eval_loss": 3.818894624710083,
+      "eval_runtime": 3.2612,
+      "eval_samples_per_second": 748.501,
+      "eval_steps_per_second": 11.959,
+      "step": 1224
+    },
+    {
+      "epoch": 4.286588029707295,
+      "grad_norm": 0.33044248819351196,
+      "learning_rate": 0.0006,
+      "loss": 3.5529201507568358,
+      "step": 1230
+    },
+    {
+      "epoch": 4.321537789427698,
+      "grad_norm": 0.2686653733253479,
+      "learning_rate": 0.0006,
+      "loss": 3.56262092590332,
+      "step": 1240
+    },
+    {
+      "epoch": 4.3564875491481,
+      "grad_norm": 0.28370386362075806,
+      "learning_rate": 0.0006,
+      "loss": 3.554620361328125,
+      "step": 1250
+    },
+    {
+      "epoch": 4.391437308868501,
+      "grad_norm": 0.3933729827404022,
+      "learning_rate": 0.0006,
+      "loss": 3.575749969482422,
+      "step": 1260
+    },
+    {
+      "epoch": 4.4263870685889035,
+      "grad_norm": 0.2854246199131012,
+      "learning_rate": 0.0006,
+      "loss": 3.5564411163330076,
+      "step": 1270
+    },
+    {
+      "epoch": 4.461336828309306,
+      "grad_norm": 0.2771267592906952,
+      "learning_rate": 0.0006,
+      "loss": 3.5506031036376955,
+      "step": 1280
+    },
+    {
+      "epoch": 4.496286588029707,
+      "grad_norm": 0.22956126928329468,
+      "learning_rate": 0.0006,
+      "loss": 3.5430240631103516,
+      "step": 1290
+    },
+    {
+      "epoch": 4.517256443861949,
+      "eval_loss": 3.788468837738037,
+      "eval_runtime": 3.2599,
+      "eval_samples_per_second": 748.788,
+      "eval_steps_per_second": 11.963,
+      "step": 1296
+    },
+    {
+      "epoch": 4.531236347750109,
+      "grad_norm": 0.2599964737892151,
+      "learning_rate": 0.0006,
+      "loss": 3.551839065551758,
+      "step": 1300
+    },
+    {
+      "epoch": 4.566186107470511,
+      "grad_norm": 0.2577304244041443,
+      "learning_rate": 0.0006,
+      "loss": 3.5488304138183593,
+      "step": 1310
+    },
+    {
+      "epoch": 4.601135867190913,
+      "grad_norm": 0.27179989218711853,
+      "learning_rate": 0.0006,
+      "loss": 3.550336456298828,
+      "step": 1320
+    },
+    {
+      "epoch": 4.636085626911315,
+      "grad_norm": 0.24012631177902222,
+      "learning_rate": 0.0006,
+      "loss": 3.530913162231445,
+      "step": 1330
+    },
+    {
+      "epoch": 4.671035386631717,
+      "grad_norm": 0.251501202583313,
+      "learning_rate": 0.0006,
+      "loss": 3.5298126220703123,
+      "step": 1340
+    },
+    {
+      "epoch": 4.705985146352119,
+      "grad_norm": 0.2944290041923523,
+      "learning_rate": 0.0006,
+      "loss": 3.513424301147461,
+      "step": 1350
+    },
+    {
+      "epoch": 4.740934906072521,
+      "grad_norm": 0.21871249377727509,
+      "learning_rate": 0.0006,
+      "loss": 3.5299285888671874,
+      "step": 1360
+    },
+    {
+      "epoch": 4.768894713848843,
+      "eval_loss": 3.7646024227142334,
+      "eval_runtime": 3.2633,
+      "eval_samples_per_second": 748.027,
+      "eval_steps_per_second": 11.951,
+      "step": 1368
+    },
+    {
+      "epoch": 4.775884665792923,
+      "grad_norm": 0.2700722813606262,
+      "learning_rate": 0.0006,
+      "loss": 3.5179073333740236,
+      "step": 1370
+    },
+    {
+      "epoch": 4.810834425513325,
+      "grad_norm": 0.28921589255332947,
+      "learning_rate": 0.0006,
+      "loss": 3.5324832916259767,
+      "step": 1380
+    },
+    {
+      "epoch": 4.845784185233726,
+      "grad_norm": 0.29661330580711365,
+      "learning_rate": 0.0006,
+      "loss": 3.5182342529296875,
+      "step": 1390
+    },
+    {
+      "epoch": 4.8807339449541285,
+      "grad_norm": 0.22164201736450195,
+      "learning_rate": 0.0006,
+      "loss": 3.5170005798339843,
+      "step": 1400
+    },
+    {
+      "epoch": 4.91568370467453,
+      "grad_norm": 0.263325572013855,
+      "learning_rate": 0.0006,
+      "loss": 3.5143806457519533,
+      "step": 1410
+    },
+    {
+      "epoch": 4.950633464394932,
+      "grad_norm": 0.2468360960483551,
+      "learning_rate": 0.0006,
+      "loss": 3.521328353881836,
+      "step": 1420
+    },
+    {
+      "epoch": 4.985583224115334,
+      "grad_norm": 0.29565975069999695,
+      "learning_rate": 0.0006,
+      "loss": 3.5046390533447265,
+      "step": 1430
+    },
+    {
+      "epoch": 5.017474879860201,
+      "grad_norm": 0.3554621934890747,
+      "learning_rate": 0.0006,
+      "loss": 3.4706165313720705,
+      "step": 1440
+    },
+    {
+      "epoch": 5.017474879860201,
+      "eval_loss": 3.7661139965057373,
+      "eval_runtime": 3.2615,
+      "eval_samples_per_second": 748.426,
+      "eval_steps_per_second": 11.958,
+      "step": 1440
+    },
+    {
+      "epoch": 5.052424639580603,
+      "grad_norm": 0.31832361221313477,
+      "learning_rate": 0.0006,
+      "loss": 3.449114227294922,
+      "step": 1450
+    },
+    {
+      "epoch": 5.087374399301005,
+      "grad_norm": 0.25010085105895996,
+      "learning_rate": 0.0006,
+      "loss": 3.4600296020507812,
+      "step": 1460
+    },
+    {
+      "epoch": 5.122324159021407,
+      "grad_norm": 0.20949716866016388,
+      "learning_rate": 0.0006,
+      "loss": 3.434639739990234,
+      "step": 1470
+    },
+    {
+      "epoch": 5.157273918741809,
+      "grad_norm": 0.21873407065868378,
+      "learning_rate": 0.0006,
+      "loss": 3.437639617919922,
+      "step": 1480
+    },
+    {
+      "epoch": 5.19222367846221,
+      "grad_norm": 0.23436321318149567,
+      "learning_rate": 0.0006,
+      "loss": 3.450662612915039,
+      "step": 1490
+    },
+    {
+      "epoch": 5.2271734381826125,
+      "grad_norm": 0.3336713910102844,
+      "learning_rate": 0.0006,
+      "loss": 3.454402542114258,
+      "step": 1500
+    },
+    {
+      "epoch": 5.262123197903015,
+      "grad_norm": 0.283755362033844,
+      "learning_rate": 0.0006,
+      "loss": 3.4522064208984373,
+      "step": 1510
+    },
+    {
+      "epoch": 5.269113149847095,
+      "eval_loss": 3.7274162769317627,
+      "eval_runtime": 3.263,
+      "eval_samples_per_second": 748.079,
+      "eval_steps_per_second": 11.952,
+      "step": 1512
+    },
+    {
+      "epoch": 5.297072957623416,
+      "grad_norm": 0.25525152683258057,
+      "learning_rate": 0.0006,
+      "loss": 3.4496952056884767,
+      "step": 1520
+    },
+    {
+      "epoch": 5.332022717343818,
+      "grad_norm": 0.2811771333217621,
+      "learning_rate": 0.0006,
+      "loss": 3.451160430908203,
+      "step": 1530
+    },
+    {
+      "epoch": 5.36697247706422,
+      "grad_norm": 0.24571454524993896,
+      "learning_rate": 0.0006,
+      "loss": 3.439900207519531,
+      "step": 1540
+    },
+    {
+      "epoch": 5.401922236784622,
+      "grad_norm": 0.20756185054779053,
+      "learning_rate": 0.0006,
+      "loss": 3.435158538818359,
+      "step": 1550
+    },
+    {
+      "epoch": 5.436871996505024,
+      "grad_norm": 0.3258594572544098,
+      "learning_rate": 0.0006,
+      "loss": 3.4386497497558595,
+      "step": 1560
+    },
+    {
+      "epoch": 5.471821756225426,
+      "grad_norm": 0.2695145606994629,
+      "learning_rate": 0.0006,
+      "loss": 3.442052459716797,
+      "step": 1570
+    },
+    {
+      "epoch": 5.506771515945828,
+      "grad_norm": 0.24291402101516724,
+      "learning_rate": 0.0006,
+      "loss": 3.4464370727539064,
+      "step": 1580
+    },
+    {
+      "epoch": 5.520751419833989,
+      "eval_loss": 3.7075228691101074,
+      "eval_runtime": 3.2626,
+      "eval_samples_per_second": 748.182,
+      "eval_steps_per_second": 11.954,
+      "step": 1584
+    },
+    {
+      "epoch": 5.54172127566623,
+      "grad_norm": 0.2379208207130432,
+      "learning_rate": 0.0006,
+      "loss": 3.4302520751953125,
+      "step": 1590
+    },
+    {
+      "epoch": 5.576671035386632,
+      "grad_norm": 0.24483603239059448,
+      "learning_rate": 0.0006,
+      "loss": 3.4376548767089843,
+      "step": 1600
+    },
+    {
+      "epoch": 5.611620795107034,
+      "grad_norm": 0.24607770144939423,
+      "learning_rate": 0.0006,
+      "loss": 3.435297393798828,
+      "step": 1610
+    },
+    {
+      "epoch": 5.646570554827435,
+      "grad_norm": 0.2616729140281677,
+      "learning_rate": 0.0006,
+      "loss": 3.409037399291992,
+      "step": 1620
+    },
+    {
+      "epoch": 5.6815203145478375,
+      "grad_norm": 0.24463753402233124,
+      "learning_rate": 0.0006,
+      "loss": 3.4296310424804686,
+      "step": 1630
+    },
+    {
+      "epoch": 5.71647007426824,
+      "grad_norm": 0.26358485221862793,
+      "learning_rate": 0.0006,
+      "loss": 3.439252471923828,
+      "step": 1640
+    },
+    {
+      "epoch": 5.751419833988641,
+      "grad_norm": 0.2910812199115753,
+      "learning_rate": 0.0006,
+      "loss": 3.4348064422607423,
+      "step": 1650
+    },
+    {
+      "epoch": 5.772389689820883,
+      "eval_loss": 3.6882729530334473,
+      "eval_runtime": 3.2478,
+      "eval_samples_per_second": 751.597,
+      "eval_steps_per_second": 12.008,
+      "step": 1656
+    },
+    {
+      "epoch": 5.786369593709043,
+      "grad_norm": 0.22973617911338806,
+      "learning_rate": 0.0006,
+      "loss": 3.426691436767578,
+      "step": 1660
+    },
+    {
+      "epoch": 5.821319353429445,
+      "grad_norm": 0.24030563235282898,
+      "learning_rate": 0.0006,
+      "loss": 3.4158489227294924,
+      "step": 1670
+    },
+    {
+      "epoch": 5.856269113149847,
+      "grad_norm": 0.25955885648727417,
+      "learning_rate": 0.0006,
+      "loss": 3.428885650634766,
+      "step": 1680
+    },
+    {
+      "epoch": 5.891218872870249,
+      "grad_norm": 0.25044476985931396,
+      "learning_rate": 0.0006,
+      "loss": 3.424819564819336,
+      "step": 1690
+    },
+    {
+      "epoch": 5.926168632590651,
+      "grad_norm": 0.23120065033435822,
+      "learning_rate": 0.0006,
+      "loss": 3.4295394897460936,
+      "step": 1700
+    },
+    {
+      "epoch": 5.961118392311053,
+      "grad_norm": 0.21678978204727173,
+      "learning_rate": 0.0006,
+      "loss": 3.4210220336914063,
+      "step": 1710
+    },
+    {
+      "epoch": 5.996068152031455,
+      "grad_norm": 0.2809757888317108,
+      "learning_rate": 0.0006,
+      "loss": 3.4109947204589846,
+      "step": 1720
+    },
+    {
+      "epoch": 6.0209698558322415,
+      "eval_loss": 3.6969563961029053,
+      "eval_runtime": 3.2539,
+      "eval_samples_per_second": 750.176,
+      "eval_steps_per_second": 11.986,
+      "step": 1728
+    },
+    {
+      "epoch": 6.0279598077763215,
+      "grad_norm": 0.31643757224082947,
+      "learning_rate": 0.0006,
+      "loss": 3.3572032928466795,
+      "step": 1730
+    },
+    {
+      "epoch": 6.062909567496724,
+      "grad_norm": 0.29343917965888977,
+      "learning_rate": 0.0006,
+      "loss": 3.3697841644287108,
+      "step": 1740
+    },
+    {
+      "epoch": 6.097859327217125,
+      "grad_norm": 0.2565082013607025,
+      "learning_rate": 0.0006,
+      "loss": 3.369162750244141,
+      "step": 1750
+    },
+    {
+      "epoch": 6.132809086937527,
+      "grad_norm": 0.20671115815639496,
+      "learning_rate": 0.0006,
+      "loss": 3.3570404052734375,
+      "step": 1760
+    },
+    {
+      "epoch": 6.16775884665793,
+      "grad_norm": 0.23868943750858307,
+      "learning_rate": 0.0006,
+      "loss": 3.3632354736328125,
+      "step": 1770
+    },
+    {
+      "epoch": 6.202708606378331,
+      "grad_norm": 0.2508518099784851,
+      "learning_rate": 0.0006,
+      "loss": 3.348232650756836,
+      "step": 1780
+    },
+    {
+      "epoch": 6.237658366098733,
+      "grad_norm": 0.22632752358913422,
+      "learning_rate": 0.0006,
+      "loss": 3.360157775878906,
+      "step": 1790
+    },
+    {
+      "epoch": 6.272608125819135,
+      "grad_norm": 0.26951679587364197,
+      "learning_rate": 0.0006,
+      "loss": 3.3667224884033202,
+      "step": 1800
+    },
+    {
+      "epoch": 6.272608125819135,
+      "eval_loss": 3.6704163551330566,
+      "eval_runtime": 3.2601,
+      "eval_samples_per_second": 748.741,
+      "eval_steps_per_second": 11.963,
+      "step": 1800
+    },
+    {
+      "epoch": 6.307557885539537,
+      "grad_norm": 0.26842039823532104,
+      "learning_rate": 0.0006,
+      "loss": 3.357311248779297,
+      "step": 1810
+    },
+    {
+      "epoch": 6.342507645259939,
+      "grad_norm": 0.2667703926563263,
+      "learning_rate": 0.0006,
+      "loss": 3.3607868194580077,
+      "step": 1820
+    },
+    {
+      "epoch": 6.3774574049803405,
+      "grad_norm": 0.23163355886936188,
+      "learning_rate": 0.0006,
+      "loss": 3.347637939453125,
+      "step": 1830
+    },
+    {
+      "epoch": 6.412407164700743,
+      "grad_norm": 0.22703006863594055,
+      "learning_rate": 0.0006,
+      "loss": 3.3697723388671874,
+      "step": 1840
+    },
+    {
+      "epoch": 6.447356924421145,
+      "grad_norm": 0.2799169421195984,
+      "learning_rate": 0.0006,
+      "loss": 3.369439697265625,
+      "step": 1850
+    },
+    {
+      "epoch": 6.482306684141546,
+      "grad_norm": 0.2718331217765808,
+      "learning_rate": 0.0006,
+      "loss": 3.3738807678222655,
+      "step": 1860
+    },
+    {
+      "epoch": 6.517256443861949,
+      "grad_norm": 0.2377900928258896,
+      "learning_rate": 0.0006,
+      "loss": 3.363282012939453,
+      "step": 1870
+    },
+    {
+      "epoch": 6.524246395806029,
+      "eval_loss": 3.653632879257202,
+      "eval_runtime": 3.2399,
+      "eval_samples_per_second": 753.407,
+      "eval_steps_per_second": 12.037,
+      "step": 1872
+    },
+    {
+      "epoch": 6.55220620358235,
+      "grad_norm": 0.2151111215353012,
+      "learning_rate": 0.0006,
+      "loss": 3.3598587036132814,
+      "step": 1880
+    },
+    {
+      "epoch": 6.587155963302752,
+      "grad_norm": 0.2511304020881653,
+      "learning_rate": 0.0006,
+      "loss": 3.358686828613281,
+      "step": 1890
+    },
+    {
+      "epoch": 6.622105723023155,
+      "grad_norm": 0.2504633665084839,
+      "learning_rate": 0.0006,
+      "loss": 3.357570266723633,
+      "step": 1900
+    },
+    {
+      "epoch": 6.657055482743556,
+      "grad_norm": 0.3591446280479431,
+      "learning_rate": 0.0006,
+      "loss": 3.3733283996582033,
+      "step": 1910
+    },
+    {
+      "epoch": 6.692005242463958,
+      "grad_norm": 0.2559049427509308,
+      "learning_rate": 0.0006,
+      "loss": 3.3553306579589846,
+      "step": 1920
+    },
+    {
+      "epoch": 6.72695500218436,
+      "grad_norm": 0.21506226062774658,
+      "learning_rate": 0.0006,
+      "loss": 3.3603141784667967,
+      "step": 1930
+    },
+    {
+      "epoch": 6.761904761904762,
+      "grad_norm": 0.2192126214504242,
+      "learning_rate": 0.0006,
+      "loss": 3.348493957519531,
+      "step": 1940
+    },
+    {
+      "epoch": 6.775884665792923,
+      "eval_loss": 3.6372475624084473,
+      "eval_runtime": 3.245,
+      "eval_samples_per_second": 752.235,
+      "eval_steps_per_second": 12.019,
+      "step": 1944
+    },
+    {
+      "epoch": 6.796854521625164,
+      "grad_norm": 0.2293984293937683,
+      "learning_rate": 0.0006,
+      "loss": 3.3508243560791016,
+      "step": 1950
+    },
+    {
+      "epoch": 6.8318042813455655,
+      "grad_norm": 0.2845034897327423,
+      "learning_rate": 0.0006,
+      "loss": 3.3553226470947264,
+      "step": 1960
+    },
+    {
+      "epoch": 6.866754041065968,
+      "grad_norm": 0.20659707486629486,
+      "learning_rate": 0.0006,
+      "loss": 3.3627620697021485,
+      "step": 1970
+    },
+    {
+      "epoch": 6.90170380078637,
+      "grad_norm": 0.20515389740467072,
+      "learning_rate": 0.0006,
+      "loss": 3.3460277557373046,
+      "step": 1980
+    },
+    {
+      "epoch": 6.936653560506771,
+      "grad_norm": 0.28660303354263306,
+      "learning_rate": 0.0006,
+      "loss": 3.3561187744140626,
+      "step": 1990
+    },
+    {
+      "epoch": 6.971603320227174,
+      "grad_norm": 0.2599312365055084,
+      "learning_rate": 0.0006,
+      "loss": 3.3571990966796874,
+      "step": 2000
+    },
+    {
+      "epoch": 7.00349497597204,
+      "grad_norm": 0.2783736288547516,
+      "learning_rate": 0.0006,
+      "loss": 3.319508361816406,
+      "step": 2010
+    },
+    {
+      "epoch": 7.0244648318042815,
+      "eval_loss": 3.6518635749816895,
+      "eval_runtime": 3.247,
+      "eval_samples_per_second": 751.772,
+      "eval_steps_per_second": 12.011,
+      "step": 2016
+    },
+    {
+      "epoch": 7.038444735692442,
+      "grad_norm": 0.3239979147911072,
+      "learning_rate": 0.0006,
+      "loss": 3.3044384002685545,
+      "step": 2020
+    },
+    {
+      "epoch": 7.073394495412844,
+      "grad_norm": 0.2413933426141739,
+      "learning_rate": 0.0006,
+      "loss": 3.298838806152344,
+      "step": 2030
+    },
+    {
+      "epoch": 7.108344255133246,
+      "grad_norm": 0.24074682593345642,
+      "learning_rate": 0.0006,
+      "loss": 3.3026168823242186,
+      "step": 2040
+    },
+    {
+      "epoch": 7.143294014853648,
+      "grad_norm": 0.24079196155071259,
+      "learning_rate": 0.0006,
+      "loss": 3.2917388916015624,
+      "step": 2050
+    },
+    {
+      "epoch": 7.1782437745740495,
+      "grad_norm": 0.1919202357530594,
+      "learning_rate": 0.0006,
+      "loss": 3.275748062133789,
+      "step": 2060
+    },
+    {
+      "epoch": 7.213193534294452,
+      "grad_norm": 0.26559439301490784,
+      "learning_rate": 0.0006,
+      "loss": 3.295817565917969,
+      "step": 2070
+    },
+    {
+      "epoch": 7.248143294014854,
+      "grad_norm": 0.22799427807331085,
+      "learning_rate": 0.0006,
+      "loss": 3.296709442138672,
+      "step": 2080
+    },
+    {
+      "epoch": 7.2761031017911755,
+      "eval_loss": 3.624324321746826,
+      "eval_runtime": 3.2433,
+      "eval_samples_per_second": 752.631,
+      "eval_steps_per_second": 12.025,
+      "step": 2088
+    },
+    {
+      "epoch": 7.283093053735255,
+      "grad_norm": 0.2224215418100357,
+      "learning_rate": 0.0006,
+      "loss": 3.2963180541992188,
+      "step": 2090
+    },
+    {
+      "epoch": 7.318042813455658,
+      "grad_norm": 0.2444535791873932,
+      "learning_rate": 0.0006,
+      "loss": 3.2914241790771483,
+      "step": 2100
+    },
+    {
+      "epoch": 7.352992573176059,
+      "grad_norm": 0.21932783722877502,
+      "learning_rate": 0.0006,
+      "loss": 3.303139495849609,
+      "step": 2110
+    },
+    {
+      "epoch": 7.387942332896461,
+      "grad_norm": 0.23804153501987457,
+      "learning_rate": 0.0006,
+      "loss": 3.3035308837890627,
+      "step": 2120
+    },
+    {
+      "epoch": 7.422892092616864,
+      "grad_norm": 0.2395394891500473,
+      "learning_rate": 0.0006,
+      "loss": 3.2892124176025392,
+      "step": 2130
+    },
+    {
+      "epoch": 7.457841852337265,
+      "grad_norm": 0.25124213099479675,
+      "learning_rate": 0.0006,
+      "loss": 3.2951118469238283,
+      "step": 2140
+    },
+    {
+      "epoch": 7.492791612057667,
+      "grad_norm": 0.23319879174232483,
+      "learning_rate": 0.0006,
+      "loss": 3.2998283386230467,
+      "step": 2150
+    },
+    {
+      "epoch": 7.527741371778069,
+      "grad_norm": 0.24116648733615875,
+      "learning_rate": 0.0006,
+      "loss": 3.308794403076172,
+      "step": 2160
+    },
+    {
+      "epoch": 7.527741371778069,
+      "eval_loss": 3.6152615547180176,
+      "eval_runtime": 3.2532,
+      "eval_samples_per_second": 750.331,
+      "eval_steps_per_second": 11.988,
+      "step": 2160
+    },
+    {
+      "epoch": 7.562691131498471,
+      "grad_norm": 0.2400725781917572,
+      "learning_rate": 0.0006,
+      "loss": 3.3068809509277344,
+      "step": 2170
+    },
+    {
+      "epoch": 7.597640891218873,
+      "grad_norm": 0.2281012386083603,
+      "learning_rate": 0.0006,
+      "loss": 3.301954650878906,
+      "step": 2180
+    },
+    {
+      "epoch": 7.6325906509392745,
+      "grad_norm": 0.2728268802165985,
+      "learning_rate": 0.0006,
+      "loss": 3.302402877807617,
+      "step": 2190
+    },
+    {
+      "epoch": 7.667540410659677,
+      "grad_norm": 0.23581810295581818,
+      "learning_rate": 0.0006,
+      "loss": 3.307527542114258,
+      "step": 2200
+    },
+    {
+      "epoch": 7.702490170380079,
+      "grad_norm": 0.2632705867290497,
+      "learning_rate": 0.0006,
+      "loss": 3.317687225341797,
+      "step": 2210
+    },
+    {
+      "epoch": 7.73743993010048,
+      "grad_norm": 0.25068581104278564,
+      "learning_rate": 0.0006,
+      "loss": 3.3130027770996096,
+      "step": 2220
+    },
+    {
+      "epoch": 7.772389689820883,
+      "grad_norm": 0.21420809626579285,
+      "learning_rate": 0.0006,
+      "loss": 3.298267364501953,
+      "step": 2230
+    },
+    {
+      "epoch": 7.7793796417649625,
+      "eval_loss": 3.6043670177459717,
+      "eval_runtime": 3.2394,
+      "eval_samples_per_second": 753.527,
+      "eval_steps_per_second": 12.039,
+      "step": 2232
+    },
+    {
+      "epoch": 7.807339449541285,
+      "grad_norm": 0.21798232197761536,
+      "learning_rate": 0.0006,
+      "loss": 3.3103466033935547,
+      "step": 2240
+    },
+    {
+      "epoch": 7.842289209261686,
+      "grad_norm": 0.20122861862182617,
+      "learning_rate": 0.0006,
+      "loss": 3.3017601013183593,
+      "step": 2250
+    },
+    {
+      "epoch": 7.8772389689820885,
+      "grad_norm": 0.2941531240940094,
+      "learning_rate": 0.0006,
+      "loss": 3.2926555633544923,
+      "step": 2260
+    },
+    {
+      "epoch": 7.91218872870249,
+      "grad_norm": 0.2505897879600525,
+      "learning_rate": 0.0006,
+      "loss": 3.3025276184082033,
+      "step": 2270
+    },
+    {
+      "epoch": 7.947138488422892,
+      "grad_norm": 0.22626236081123352,
+      "learning_rate": 0.0006,
+      "loss": 3.295301818847656,
+      "step": 2280
+    },
+    {
+      "epoch": 7.9820882481432935,
+      "grad_norm": 0.23794551193714142,
+      "learning_rate": 0.0006,
+      "loss": 3.296539306640625,
+      "step": 2290
+    },
+    {
+      "epoch": 8.013979903888162,
+      "grad_norm": 0.31783896684646606,
+      "learning_rate": 0.0006,
+      "loss": 3.271350860595703,
+      "step": 2300
+    },
+    {
+      "epoch": 8.027959807776321,
+      "eval_loss": 3.6203079223632812,
+      "eval_runtime": 3.2516,
+      "eval_samples_per_second": 750.714,
+      "eval_steps_per_second": 11.994,
+      "step": 2304
+    },
+    {
+      "epoch": 8.048929663608563,
+      "grad_norm": 0.2704427242279053,
+      "learning_rate": 0.0006,
+      "loss": 3.2576080322265626,
+      "step": 2310
+    },
+    {
+      "epoch": 8.083879423328964,
+      "grad_norm": 0.26199305057525635,
+      "learning_rate": 0.0006,
+      "loss": 3.233934020996094,
+      "step": 2320
+    },
+    {
+      "epoch": 8.118829183049366,
+      "grad_norm": 0.22090311348438263,
+      "learning_rate": 0.0006,
+      "loss": 3.235779571533203,
+      "step": 2330
+    },
+    {
+      "epoch": 8.153778942769769,
+      "grad_norm": 0.24668550491333008,
+      "learning_rate": 0.0006,
+      "loss": 3.237989044189453,
+      "step": 2340
+    },
+    {
+      "epoch": 8.18872870249017,
+      "grad_norm": 0.22720353305339813,
+      "learning_rate": 0.0006,
+      "loss": 3.2465499877929687,
+      "step": 2350
+    },
+    {
+      "epoch": 8.223678462210572,
+      "grad_norm": 0.2963132858276367,
+      "learning_rate": 0.0006,
+      "loss": 3.250588226318359,
+      "step": 2360
+    },
+    {
+      "epoch": 8.258628221930975,
+      "grad_norm": 0.2263713777065277,
+      "learning_rate": 0.0006,
+      "loss": 3.244337463378906,
+      "step": 2370
+    },
+    {
+      "epoch": 8.279598077763215,
+      "eval_loss": 3.597494125366211,
+      "eval_runtime": 3.2475,
+      "eval_samples_per_second": 751.644,
+      "eval_steps_per_second": 12.009,
+      "step": 2376
+    },
+    {
+      "epoch": 8.293577981651376,
+      "grad_norm": 0.24112579226493835,
+      "learning_rate": 0.0006,
+      "loss": 3.2619468688964846,
+      "step": 2380
+    },
+    {
+      "epoch": 8.328527741371778,
+      "grad_norm": 0.2483445405960083,
+      "learning_rate": 0.0006,
+      "loss": 3.2471988677978514,
+      "step": 2390
+    },
+    {
+      "epoch": 8.36347750109218,
+      "grad_norm": 0.23584790527820587,
+      "learning_rate": 0.0006,
+      "loss": 3.2458396911621095,
+      "step": 2400
+    },
+    {
+      "epoch": 8.398427260812582,
+      "grad_norm": 0.24326704442501068,
+      "learning_rate": 0.0006,
+      "loss": 3.258475112915039,
+      "step": 2410
+    },
+    {
+      "epoch": 8.433377020532983,
+      "grad_norm": 0.25203439593315125,
+      "learning_rate": 0.0006,
+      "loss": 3.2588306427001954,
+      "step": 2420
+    },
+    {
+      "epoch": 8.468326780253387,
+      "grad_norm": 0.21486154198646545,
+      "learning_rate": 0.0006,
+      "loss": 3.2324851989746093,
+      "step": 2430
+    },
+    {
+      "epoch": 8.503276539973788,
+      "grad_norm": 0.23271426558494568,
+      "learning_rate": 0.0006,
+      "loss": 3.264314651489258,
+      "step": 2440
+    },
+    {
+      "epoch": 8.53123634775011,
+      "eval_loss": 3.5861639976501465,
+      "eval_runtime": 3.2549,
+      "eval_samples_per_second": 749.949,
+      "eval_steps_per_second": 11.982,
+      "step": 2448
+    },
+    {
+      "epoch": 8.53822629969419,
+      "grad_norm": 0.21104109287261963,
+      "learning_rate": 0.0006,
+      "loss": 3.2432979583740233,
+      "step": 2450
+    },
+    {
+      "epoch": 8.57317605941459,
+      "grad_norm": 0.2308482676744461,
+      "learning_rate": 0.0006,
+      "loss": 3.2519115447998046,
+      "step": 2460
+    },
+    {
+      "epoch": 8.608125819134994,
+      "grad_norm": 0.26735422015190125,
+      "learning_rate": 0.0006,
+      "loss": 3.2551074981689454,
+      "step": 2470
+    },
+    {
+      "epoch": 8.643075578855395,
+      "grad_norm": 0.23472614586353302,
+      "learning_rate": 0.0006,
+      "loss": 3.254596710205078,
+      "step": 2480
+    },
+    {
+      "epoch": 8.678025338575797,
+      "grad_norm": 0.20107829570770264,
+      "learning_rate": 0.0006,
+      "loss": 3.2619773864746096,
+      "step": 2490
+    },
+    {
+      "epoch": 8.7129750982962,
+      "grad_norm": 0.2232264131307602,
+      "learning_rate": 0.0006,
+      "loss": 3.2473453521728515,
+      "step": 2500
+    },
+    {
+      "epoch": 8.747924858016601,
+      "grad_norm": 0.2425001561641693,
+      "learning_rate": 0.0006,
+      "loss": 3.245287322998047,
+      "step": 2510
+    },
+    {
+      "epoch": 8.782874617737003,
+      "grad_norm": 0.24481889605522156,
+      "learning_rate": 0.0006,
+      "loss": 3.2641754150390625,
+      "step": 2520
+    },
+    {
+      "epoch": 8.782874617737003,
+      "eval_loss": 3.5801212787628174,
+      "eval_runtime": 3.2446,
+      "eval_samples_per_second": 752.333,
+      "eval_steps_per_second": 12.02,
+      "step": 2520
+    },
+    {
+      "epoch": 8.817824377457406,
+      "grad_norm": 0.20684371888637543,
+      "learning_rate": 0.0006,
+      "loss": 3.2686134338378907,
+      "step": 2530
+    },
+    {
+      "epoch": 8.852774137177807,
+      "grad_norm": 0.22612686455249786,
+      "learning_rate": 0.0006,
+      "loss": 3.2567123413085937,
+      "step": 2540
+    },
+    {
+      "epoch": 8.887723896898208,
+      "grad_norm": 0.2703574299812317,
+      "learning_rate": 0.0006,
+      "loss": 3.2661643981933595,
+      "step": 2550
+    },
+    {
+      "epoch": 8.922673656618612,
+      "grad_norm": 0.24275517463684082,
+      "learning_rate": 0.0006,
+      "loss": 3.272625732421875,
+      "step": 2560
+    },
+    {
+      "epoch": 8.957623416339013,
+      "grad_norm": 0.25982919335365295,
+      "learning_rate": 0.0006,
+      "loss": 3.256539154052734,
+      "step": 2570
+    },
+    {
+      "epoch": 8.992573176059414,
+      "grad_norm": 0.2267482578754425,
+      "learning_rate": 0.0006,
+      "loss": 3.2591373443603517,
+      "step": 2580
+    },
+    {
+      "epoch": 9.02446483180428,
+      "grad_norm": 0.38567808270454407,
+      "learning_rate": 0.0006,
+      "loss": 3.2312469482421875,
+      "step": 2590
+    },
+    {
+      "epoch": 9.031454783748362,
+      "eval_loss": 3.605508327484131,
+      "eval_runtime": 3.2604,
+      "eval_samples_per_second": 748.676,
+      "eval_steps_per_second": 11.962,
+      "step": 2592
+    },
+    {
+      "epoch": 9.059414591524684,
+      "grad_norm": 0.2434706836938858,
+      "learning_rate": 0.0006,
+      "loss": 3.208142852783203,
+      "step": 2600
+    },
+    {
+      "epoch": 9.094364351245085,
+      "grad_norm": 0.2534368336200714,
+      "learning_rate": 0.0006,
+      "loss": 3.202614593505859,
+      "step": 2610
+    },
+    {
+      "epoch": 9.129314110965487,
+      "grad_norm": 0.2170858532190323,
+      "learning_rate": 0.0006,
+      "loss": 3.2168445587158203,
+      "step": 2620
+    },
+    {
+      "epoch": 9.16426387068589,
+      "grad_norm": 0.21068714559078217,
+      "learning_rate": 0.0006,
+      "loss": 3.1942779541015627,
+      "step": 2630
+    },
+    {
+      "epoch": 9.199213630406291,
+      "grad_norm": 0.2258630245923996,
+      "learning_rate": 0.0006,
+      "loss": 3.2012100219726562,
+      "step": 2640
+    },
+    {
+      "epoch": 9.234163390126692,
+      "grad_norm": 0.24640101194381714,
+      "learning_rate": 0.0006,
+      "loss": 3.1907680511474608,
+      "step": 2650
+    },
+    {
+      "epoch": 9.269113149847096,
+      "grad_norm": 0.21594661474227905,
+      "learning_rate": 0.0006,
+      "loss": 3.21116943359375,
+      "step": 2660
+    },
+    {
+      "epoch": 9.283093053735255,
+      "eval_loss": 3.5735225677490234,
+      "eval_runtime": 3.2479,
+      "eval_samples_per_second": 751.56,
+      "eval_steps_per_second": 12.008,
+      "step": 2664
+    },
+    {
+      "epoch": 9.304062909567497,
+      "grad_norm": 0.22658808529376984,
+      "learning_rate": 0.0006,
+      "loss": 3.187892532348633,
+      "step": 2670
+    },
+    {
+      "epoch": 9.339012669287898,
+      "grad_norm": 0.24525444209575653,
+      "learning_rate": 0.0006,
+      "loss": 3.1912796020507814,
+      "step": 2680
+    },
+    {
+      "epoch": 9.373962429008301,
+      "grad_norm": 0.2433639019727707,
+      "learning_rate": 0.0006,
+      "loss": 3.2131378173828127,
+      "step": 2690
+    },
+    {
+      "epoch": 9.408912188728703,
+      "grad_norm": 0.2769683599472046,
+      "learning_rate": 0.0006,
+      "loss": 3.219586181640625,
+      "step": 2700
+    },
+    {
+      "epoch": 9.443861948449104,
+      "grad_norm": 0.220779687166214,
+      "learning_rate": 0.0006,
+      "loss": 3.224137878417969,
+      "step": 2710
+    },
+    {
+      "epoch": 9.478811708169506,
+      "grad_norm": 0.2180791050195694,
+      "learning_rate": 0.0006,
+      "loss": 3.2175228118896486,
+      "step": 2720
+    },
+    {
+      "epoch": 9.513761467889909,
+      "grad_norm": 0.22729653120040894,
+      "learning_rate": 0.0006,
+      "loss": 3.203997802734375,
+      "step": 2730
+    },
+    {
+      "epoch": 9.534731323722148,
+      "eval_loss": 3.5659878253936768,
+      "eval_runtime": 3.2486,
+      "eval_samples_per_second": 751.4,
+      "eval_steps_per_second": 12.005,
+      "step": 2736
+    },
+    {
+      "epoch": 9.54871122761031,
+      "grad_norm": 0.208607479929924,
+      "learning_rate": 0.0006,
+      "loss": 3.213786315917969,
+      "step": 2740
+    },
+    {
+      "epoch": 9.583660987330711,
+      "grad_norm": 0.24215640127658844,
+      "learning_rate": 0.0006,
+      "loss": 3.2111549377441406,
+      "step": 2750
+    },
+    {
+      "epoch": 9.618610747051115,
+      "grad_norm": 0.21286575496196747,
+      "learning_rate": 0.0006,
+      "loss": 3.237984466552734,
+      "step": 2760
+    },
+    {
+      "epoch": 9.653560506771516,
+      "grad_norm": 0.23427526652812958,
+      "learning_rate": 0.0006,
+      "loss": 3.215163803100586,
+      "step": 2770
+    },
+    {
+      "epoch": 9.688510266491917,
+      "grad_norm": 0.21158790588378906,
+      "learning_rate": 0.0006,
+      "loss": 3.2104469299316407,
+      "step": 2780
+    },
+    {
+      "epoch": 9.72346002621232,
+      "grad_norm": 0.22658434510231018,
+      "learning_rate": 0.0006,
+      "loss": 3.230307388305664,
+      "step": 2790
+    },
+    {
+      "epoch": 9.758409785932722,
+      "grad_norm": 0.2290172427892685,
+      "learning_rate": 0.0006,
+      "loss": 3.217431640625,
+      "step": 2800
+    },
+    {
+      "epoch": 9.786369593709043,
+      "eval_loss": 3.5584716796875,
+      "eval_runtime": 3.2494,
+      "eval_samples_per_second": 751.214,
+      "eval_steps_per_second": 12.002,
+      "step": 2808
+    },
+    {
+      "epoch": 9.793359545653123,
+      "grad_norm": 0.2156689316034317,
+      "learning_rate": 0.0006,
+      "loss": 3.2219924926757812,
+      "step": 2810
+    },
+    {
+      "epoch": 9.828309305373526,
+      "grad_norm": 0.19849006831645966,
+      "learning_rate": 0.0006,
+      "loss": 3.2274627685546875,
+      "step": 2820
+    },
+    {
+      "epoch": 9.863259065093928,
+      "grad_norm": 0.21485635638237,
+      "learning_rate": 0.0006,
+      "loss": 3.232157897949219,
+      "step": 2830
+    },
+    {
+      "epoch": 9.89820882481433,
+      "grad_norm": 0.22270803153514862,
+      "learning_rate": 0.0006,
+      "loss": 3.230732727050781,
+      "step": 2840
+    },
+    {
+      "epoch": 9.93315858453473,
+      "grad_norm": 0.24504123628139496,
+      "learning_rate": 0.0006,
+      "loss": 3.2274757385253907,
+      "step": 2850
+    },
+    {
+      "epoch": 9.968108344255134,
+      "grad_norm": 0.2725991904735565,
+      "learning_rate": 0.0006,
+      "loss": 3.219687652587891,
+      "step": 2860
+    },
+    {
+      "epoch": 10.0,
+      "grad_norm": 0.40203312039375305,
+      "learning_rate": 0.0006,
+      "loss": 3.220264434814453,
+      "step": 2870
+    },
+    {
+      "epoch": 10.034949759720401,
+      "grad_norm": 0.34610697627067566,
+      "learning_rate": 0.0006,
+      "loss": 3.168370819091797,
+      "step": 2880
+    },
+    {
+      "epoch": 10.034949759720401,
+      "eval_loss": 3.5853285789489746,
+      "eval_runtime": 3.2636,
+      "eval_samples_per_second": 747.95,
+      "eval_steps_per_second": 11.95,
+      "step": 2880
+    },
+    {
+      "epoch": 10.069899519440805,
+      "grad_norm": 0.27846553921699524,
+      "learning_rate": 0.0006,
+      "loss": 3.1650123596191406,
+      "step": 2890
+    },
+    {
+      "epoch": 10.104849279161206,
+      "grad_norm": 0.21352824568748474,
+      "learning_rate": 0.0006,
+      "loss": 3.165480041503906,
+      "step": 2900
+    },
+    {
+      "epoch": 10.139799038881607,
+      "grad_norm": 0.20953257381916046,
+      "learning_rate": 0.0006,
+      "loss": 3.1576595306396484,
+      "step": 2910
+    },
+    {
+      "epoch": 10.17474879860201,
+      "grad_norm": 0.2514585256576538,
+      "learning_rate": 0.0006,
+      "loss": 3.167644500732422,
+      "step": 2920
+    },
+    {
+      "epoch": 10.209698558322412,
+      "grad_norm": 0.20475761592388153,
+      "learning_rate": 0.0006,
+      "loss": 3.176905632019043,
+      "step": 2930
+    },
+    {
+      "epoch": 10.244648318042813,
+      "grad_norm": 0.22510138154029846,
+      "learning_rate": 0.0006,
+      "loss": 3.1697763442993163,
+      "step": 2940
+    },
+    {
+      "epoch": 10.279598077763215,
+      "grad_norm": 0.22730804979801178,
+      "learning_rate": 0.0006,
+      "loss": 3.1725616455078125,
+      "step": 2950
+    },
+    {
+      "epoch": 10.286588029707296,
+      "eval_loss": 3.55889630317688,
+      "eval_runtime": 3.2616,
+      "eval_samples_per_second": 748.403,
+      "eval_steps_per_second": 11.957,
+      "step": 2952
+    },
+    {
+      "epoch": 10.314547837483618,
+      "grad_norm": 0.19970236718654633,
+      "learning_rate": 0.0006,
+      "loss": 3.1804214477539063,
+      "step": 2960
+    },
+    {
+      "epoch": 10.349497597204019,
+      "grad_norm": 0.22887524962425232,
+      "learning_rate": 0.0006,
+      "loss": 3.1779220581054686,
+      "step": 2970
+    },
+    {
+      "epoch": 10.38444735692442,
+      "grad_norm": 0.22860968112945557,
+      "learning_rate": 0.0006,
+      "loss": 3.169704818725586,
+      "step": 2980
+    },
+    {
+      "epoch": 10.419397116644824,
+      "grad_norm": 0.24533846974372864,
+      "learning_rate": 0.0006,
+      "loss": 3.177603530883789,
+      "step": 2990
+    },
+    {
+      "epoch": 10.454346876365225,
+      "grad_norm": 0.23936660587787628,
+      "learning_rate": 0.0006,
+      "loss": 3.1877105712890623,
+      "step": 3000
+    },
+    {
+      "epoch": 10.489296636085626,
+      "grad_norm": 0.21400554478168488,
+      "learning_rate": 0.0006,
+      "loss": 3.1808149337768556,
+      "step": 3010
+    },
+    {
+      "epoch": 10.52424639580603,
+      "grad_norm": 0.2217673361301422,
+      "learning_rate": 0.0006,
+      "loss": 3.1751018524169923,
+      "step": 3020
+    },
+    {
+      "epoch": 10.53822629969419,
+      "eval_loss": 3.55299711227417,
+      "eval_runtime": 3.2592,
+      "eval_samples_per_second": 748.957,
+      "eval_steps_per_second": 11.966,
+      "step": 3024
+    },
+    {
+      "epoch": 10.55919615552643,
+      "grad_norm": 0.20337224006652832,
+      "learning_rate": 0.0006,
+      "loss": 3.182303810119629,
+      "step": 3030
+    },
+    {
+      "epoch": 10.594145915246832,
+      "grad_norm": 0.22320085763931274,
+      "learning_rate": 0.0006,
+      "loss": 3.19226188659668,
+      "step": 3040
+    },
+    {
+      "epoch": 10.629095674967235,
+      "grad_norm": 0.2297360897064209,
+      "learning_rate": 0.0006,
+      "loss": 3.1898571014404298,
+      "step": 3050
+    },
+    {
+      "epoch": 10.664045434687637,
+      "grad_norm": 0.22680862247943878,
+      "learning_rate": 0.0006,
+      "loss": 3.2022167205810548,
+      "step": 3060
+    },
+    {
+      "epoch": 10.698995194408038,
+      "grad_norm": 0.2387961447238922,
+      "learning_rate": 0.0006,
+      "loss": 3.1674932479858398,
+      "step": 3070
+    },
+    {
+      "epoch": 10.73394495412844,
+      "grad_norm": 0.23046232759952545,
+      "learning_rate": 0.0006,
+      "loss": 3.1918613433837892,
+      "step": 3080
+    },
+    {
+      "epoch": 10.768894713848843,
+      "grad_norm": 0.22830675542354584,
+      "learning_rate": 0.0006,
+      "loss": 3.183994102478027,
+      "step": 3090
+    },
+    {
+      "epoch": 10.789864569681084,
+      "eval_loss": 3.5444297790527344,
+      "eval_runtime": 3.2615,
+      "eval_samples_per_second": 748.43,
+      "eval_steps_per_second": 11.958,
+      "step": 3096
+    },
+    {
+      "epoch": 10.803844473569244,
+      "grad_norm": 0.22133506834506989,
+      "learning_rate": 0.0006,
+      "loss": 3.191306304931641,
+      "step": 3100
+    },
+    {
+      "epoch": 10.838794233289645,
+      "grad_norm": 0.2325688898563385,
+      "learning_rate": 0.0006,
+      "loss": 3.1874507904052733,
+      "step": 3110
+    },
+    {
+      "epoch": 10.873743993010049,
+      "grad_norm": 0.23838211596012115,
+      "learning_rate": 0.0006,
+      "loss": 3.1907676696777343,
+      "step": 3120
+    },
+    {
+      "epoch": 10.90869375273045,
+      "grad_norm": 0.2471194714307785,
+      "learning_rate": 0.0006,
+      "loss": 3.2024204254150392,
+      "step": 3130
+    },
+    {
+      "epoch": 10.943643512450851,
+      "grad_norm": 0.21122467517852783,
+      "learning_rate": 0.0006,
+      "loss": 3.200571823120117,
+      "step": 3140
+    },
+    {
+      "epoch": 10.978593272171254,
+      "grad_norm": 0.21754302084445953,
+      "learning_rate": 0.0006,
+      "loss": 3.1950096130371093,
+      "step": 3150
+    },
+    {
+      "epoch": 11.01048492791612,
+      "grad_norm": 0.3732515275478363,
+      "learning_rate": 0.0006,
+      "loss": 3.1760208129882814,
+      "step": 3160
+    },
+    {
+      "epoch": 11.038444735692442,
+      "eval_loss": 3.5765671730041504,
+      "eval_runtime": 3.2509,
+      "eval_samples_per_second": 750.859,
+      "eval_steps_per_second": 11.997,
+      "step": 3168
+    },
+    {
+      "epoch": 11.045434687636522,
+      "grad_norm": 0.3300546407699585,
+      "learning_rate": 0.0006,
+      "loss": 3.1458770751953127,
+      "step": 3170
+    },
+    {
+      "epoch": 11.080384447356925,
+      "grad_norm": 0.23498955368995667,
+      "learning_rate": 0.0006,
+      "loss": 3.142800521850586,
+      "step": 3180
+    },
+    {
+      "epoch": 11.115334207077327,
+      "grad_norm": 0.23064017295837402,
+      "learning_rate": 0.0006,
+      "loss": 3.1359394073486326,
+      "step": 3190
+    },
+    {
+      "epoch": 11.150283966797728,
+      "grad_norm": 0.21482722461223602,
+      "learning_rate": 0.0006,
+      "loss": 3.129222869873047,
+      "step": 3200
+    },
+    {
+      "epoch": 11.18523372651813,
+      "grad_norm": 0.2106955498456955,
+      "learning_rate": 0.0006,
+      "loss": 3.1208629608154297,
+      "step": 3210
+    },
+    {
+      "epoch": 11.220183486238533,
+      "grad_norm": 0.23341742157936096,
+      "learning_rate": 0.0006,
+      "loss": 3.1269767761230467,
+      "step": 3220
+    },
+    {
+      "epoch": 11.255133245958934,
+      "grad_norm": 0.23724929988384247,
+      "learning_rate": 0.0006,
+      "loss": 3.1442237854003907,
+      "step": 3230
+    },
+    {
+      "epoch": 11.290083005679335,
+      "grad_norm": 0.21010470390319824,
+      "learning_rate": 0.0006,
+      "loss": 3.1327152252197266,
+      "step": 3240
+    },
+    {
+      "epoch": 11.290083005679335,
+      "eval_loss": 3.547455072402954,
+      "eval_runtime": 3.2619,
+      "eval_samples_per_second": 748.343,
+      "eval_steps_per_second": 11.956,
+      "step": 3240
+    },
+    {
+      "epoch": 11.325032765399738,
+      "grad_norm": 0.21863767504692078,
+      "learning_rate": 0.0006,
+      "loss": 3.1528892517089844,
+      "step": 3250
+    },
+    {
+      "epoch": 11.35998252512014,
+      "grad_norm": 0.22152899205684662,
+      "learning_rate": 0.0006,
+      "loss": 3.1434192657470703,
+      "step": 3260
+    },
+    {
+      "epoch": 11.394932284840541,
+      "grad_norm": 0.23242661356925964,
+      "learning_rate": 0.0006,
+      "loss": 3.148727035522461,
+      "step": 3270
+    },
+    {
+      "epoch": 11.429882044560944,
+      "grad_norm": 0.24980394542217255,
+      "learning_rate": 0.0006,
+      "loss": 3.1498046875,
+      "step": 3280
+    },
+    {
+      "epoch": 11.464831804281346,
+      "grad_norm": 0.23047639429569244,
+      "learning_rate": 0.0006,
+      "loss": 3.1563995361328123,
+      "step": 3290
+    },
+    {
+      "epoch": 11.499781564001747,
+      "grad_norm": 0.22669772803783417,
+      "learning_rate": 0.0006,
+      "loss": 3.165831184387207,
+      "step": 3300
+    },
+    {
+      "epoch": 11.534731323722148,
+      "grad_norm": 0.24344085156917572,
+      "learning_rate": 0.0006,
+      "loss": 3.1575668334960936,
+      "step": 3310
+    },
+    {
+      "epoch": 11.54172127566623,
+      "eval_loss": 3.540040969848633,
+      "eval_runtime": 3.254,
+      "eval_samples_per_second": 750.154,
+      "eval_steps_per_second": 11.985,
+      "step": 3312
+    },
+    {
+      "epoch": 11.569681083442552,
+      "grad_norm": 0.2195751667022705,
+      "learning_rate": 0.0006,
+      "loss": 3.159097099304199,
+      "step": 3320
+    },
+    {
+      "epoch": 11.604630843162953,
+      "grad_norm": 0.20429736375808716,
+      "learning_rate": 0.0006,
+      "loss": 3.1552000045776367,
+      "step": 3330
+    },
+    {
+      "epoch": 11.639580602883354,
+      "grad_norm": 0.21613851189613342,
+      "learning_rate": 0.0006,
+      "loss": 3.1508174896240235,
+      "step": 3340
+    },
+    {
+      "epoch": 11.674530362603758,
+      "grad_norm": 0.21004842221736908,
+      "learning_rate": 0.0006,
+      "loss": 3.1665166854858398,
+      "step": 3350
+    },
+    {
+      "epoch": 11.709480122324159,
+      "grad_norm": 0.23241327702999115,
+      "learning_rate": 0.0006,
+      "loss": 3.164842414855957,
+      "step": 3360
+    },
+    {
+      "epoch": 11.74442988204456,
+      "grad_norm": 0.2207593023777008,
+      "learning_rate": 0.0006,
+      "loss": 3.1644824981689452,
+      "step": 3370
+    },
+    {
+      "epoch": 11.779379641764963,
+      "grad_norm": 0.2405390739440918,
+      "learning_rate": 0.0006,
+      "loss": 3.1764152526855467,
+      "step": 3380
+    },
+    {
+      "epoch": 11.793359545653123,
+      "eval_loss": 3.5332000255584717,
+      "eval_runtime": 3.2472,
+      "eval_samples_per_second": 751.735,
+      "eval_steps_per_second": 12.011,
+      "step": 3384
+    },
+    {
+      "epoch": 11.814329401485365,
+      "grad_norm": 0.24024921655654907,
+      "learning_rate": 0.0006,
+      "loss": 3.1589113235473634,
+      "step": 3390
+    },
+    {
+      "epoch": 11.849279161205766,
+      "grad_norm": 0.23888683319091797,
+      "learning_rate": 0.0006,
+      "loss": 3.169368934631348,
+      "step": 3400
+    },
+    {
+      "epoch": 11.88422892092617,
+      "grad_norm": 0.20942534506320953,
+      "learning_rate": 0.0006,
+      "loss": 3.1747886657714846,
+      "step": 3410
+    },
+    {
+      "epoch": 11.91917868064657,
+      "grad_norm": 0.2112586498260498,
+      "learning_rate": 0.0006,
+      "loss": 3.166824150085449,
+      "step": 3420
+    },
+    {
+      "epoch": 11.954128440366972,
+      "grad_norm": 0.2044767439365387,
+      "learning_rate": 0.0006,
+      "loss": 3.1647735595703126,
+      "step": 3430
+    },
+    {
+      "epoch": 11.989078200087373,
+      "grad_norm": 0.22425387799739838,
+      "learning_rate": 0.0006,
+      "loss": 3.1585723876953127,
+      "step": 3440
+    },
+    {
+      "epoch": 12.020969855832242,
+      "grad_norm": 0.30950942635536194,
+      "learning_rate": 0.0006,
+      "loss": 3.129406547546387,
+      "step": 3450
+    },
+    {
+      "epoch": 12.041939711664483,
+      "eval_loss": 3.5562198162078857,
+      "eval_runtime": 3.2604,
+      "eval_samples_per_second": 748.671,
+      "eval_steps_per_second": 11.962,
+      "step": 3456
+    },
+    {
+      "epoch": 12.055919615552643,
+      "grad_norm": 0.25703173875808716,
+      "learning_rate": 0.0006,
+      "loss": 3.115077018737793,
+      "step": 3460
+    },
+    {
+      "epoch": 12.090869375273044,
+      "grad_norm": 0.257079541683197,
+      "learning_rate": 0.0006,
+      "loss": 3.0938657760620116,
+      "step": 3470
+    },
+    {
+      "epoch": 12.125819134993447,
+      "grad_norm": 0.27722257375717163,
+      "learning_rate": 0.0006,
+      "loss": 3.105236053466797,
+      "step": 3480
+    },
+    {
+      "epoch": 12.160768894713849,
+      "grad_norm": 0.21611487865447998,
+      "learning_rate": 0.0006,
+      "loss": 3.114472198486328,
+      "step": 3490
+    },
+    {
+      "epoch": 12.19571865443425,
+      "grad_norm": 0.19990737736225128,
+      "learning_rate": 0.0006,
+      "loss": 3.0992176055908205,
+      "step": 3500
+    },
+    {
+      "epoch": 12.230668414154653,
+      "grad_norm": 0.22954604029655457,
+      "learning_rate": 0.0006,
+      "loss": 3.108814811706543,
+      "step": 3510
+    },
+    {
+      "epoch": 12.265618173875055,
+      "grad_norm": 0.21943169832229614,
+      "learning_rate": 0.0006,
+      "loss": 3.117928314208984,
+      "step": 3520
+    },
+    {
+      "epoch": 12.293577981651376,
+      "eval_loss": 3.541825294494629,
+      "eval_runtime": 3.2618,
+      "eval_samples_per_second": 748.363,
+      "eval_steps_per_second": 11.957,
+      "step": 3528
+    },
+    {
+      "epoch": 12.300567933595456,
+      "grad_norm": 0.231853649020195,
+      "learning_rate": 0.0006,
+      "loss": 3.1182844161987306,
+      "step": 3530
+    },
+    {
+      "epoch": 12.33551769331586,
+      "grad_norm": 0.2511996626853943,
+      "learning_rate": 0.0006,
+      "loss": 3.113280487060547,
+      "step": 3540
+    },
+    {
+      "epoch": 12.37046745303626,
+      "grad_norm": 0.20694443583488464,
+      "learning_rate": 0.0006,
+      "loss": 3.1264551162719725,
+      "step": 3550
+    },
+    {
+      "epoch": 12.405417212756662,
+      "grad_norm": 0.21675296127796173,
+      "learning_rate": 0.0006,
+      "loss": 3.1372774124145506,
+      "step": 3560
+    },
+    {
+      "epoch": 12.440366972477065,
+      "grad_norm": 0.2077033817768097,
+      "learning_rate": 0.0006,
+      "loss": 3.13992919921875,
+      "step": 3570
+    },
+    {
+      "epoch": 12.475316732197467,
+      "grad_norm": 0.22091925144195557,
+      "learning_rate": 0.0006,
+      "loss": 3.124655914306641,
+      "step": 3580
+    },
+    {
+      "epoch": 12.510266491917868,
+      "grad_norm": 0.23417656123638153,
+      "learning_rate": 0.0006,
+      "loss": 3.12007999420166,
+      "step": 3590
+    },
+    {
+      "epoch": 12.54521625163827,
+      "grad_norm": 0.23963961005210876,
+      "learning_rate": 0.0006,
+      "loss": 3.1335426330566407,
+      "step": 3600
+    },
+    {
+      "epoch": 12.54521625163827,
+      "eval_loss": 3.536545991897583,
+      "eval_runtime": 3.2476,
+      "eval_samples_per_second": 751.631,
+      "eval_steps_per_second": 12.009,
+      "step": 3600
+    },
+    {
+      "epoch": 12.580166011358672,
+      "grad_norm": 0.2441488802433014,
+      "learning_rate": 0.0006,
+      "loss": 3.144057846069336,
+      "step": 3610
+    },
+    {
+      "epoch": 12.615115771079074,
+      "grad_norm": 0.23208114504814148,
+      "learning_rate": 0.0006,
+      "loss": 3.125559425354004,
+      "step": 3620
+    },
+    {
+      "epoch": 12.650065530799475,
+      "grad_norm": 0.20495259761810303,
+      "learning_rate": 0.0006,
+      "loss": 3.138299751281738,
+      "step": 3630
+    },
+    {
+      "epoch": 12.685015290519878,
+      "grad_norm": 0.20879854261875153,
+      "learning_rate": 0.0006,
+      "loss": 3.1344417572021483,
+      "step": 3640
+    },
+    {
+      "epoch": 12.71996505024028,
+      "grad_norm": 0.19176813960075378,
+      "learning_rate": 0.0006,
+      "loss": 3.1462285995483397,
+      "step": 3650
+    },
+    {
+      "epoch": 12.754914809960681,
+      "grad_norm": 0.24343250691890717,
+      "learning_rate": 0.0006,
+      "loss": 3.1446929931640626,
+      "step": 3660
+    },
+    {
+      "epoch": 12.789864569681084,
+      "grad_norm": 0.20799709856510162,
+      "learning_rate": 0.0006,
+      "loss": 3.142406463623047,
+      "step": 3670
+    },
+    {
+      "epoch": 12.796854521625164,
+      "eval_loss": 3.5257339477539062,
+      "eval_runtime": 3.2552,
+      "eval_samples_per_second": 749.871,
+      "eval_steps_per_second": 11.981,
+      "step": 3672
+    },
+    {
+      "epoch": 12.824814329401486,
+      "grad_norm": 0.24613051116466522,
+      "learning_rate": 0.0006,
+      "loss": 3.148510551452637,
+      "step": 3680
+    },
+    {
+      "epoch": 12.859764089121887,
+      "grad_norm": 0.22636881470680237,
+      "learning_rate": 0.0006,
+      "loss": 3.1353824615478514,
+      "step": 3690
+    },
+    {
+      "epoch": 12.89471384884229,
+      "grad_norm": 0.236853688955307,
+      "learning_rate": 0.0006,
+      "loss": 3.146282196044922,
+      "step": 3700
+    },
+    {
+      "epoch": 12.929663608562691,
+      "grad_norm": 0.21839898824691772,
+      "learning_rate": 0.0006,
+      "loss": 3.136213684082031,
+      "step": 3710
+    },
+    {
+      "epoch": 12.964613368283093,
+      "grad_norm": 0.21705345809459686,
+      "learning_rate": 0.0006,
+      "loss": 3.1376064300537108,
+      "step": 3720
+    },
+    {
+      "epoch": 12.999563128003494,
+      "grad_norm": 0.21565082669258118,
+      "learning_rate": 0.0006,
+      "loss": 3.13662109375,
+      "step": 3730
+    },
+    {
+      "epoch": 13.031454783748362,
+      "grad_norm": 0.3086640536785126,
+      "learning_rate": 0.0006,
+      "loss": 3.0977493286132813,
+      "step": 3740
+    },
+    {
+      "epoch": 13.045434687636522,
+      "eval_loss": 3.5486865043640137,
+      "eval_runtime": 3.2594,
+      "eval_samples_per_second": 748.906,
+      "eval_steps_per_second": 11.965,
+      "step": 3744
+    },
+    {
+      "epoch": 13.066404543468764,
+      "grad_norm": 0.2622837722301483,
+      "learning_rate": 0.0006,
+      "loss": 3.0872968673706054,
+      "step": 3750
+    },
+    {
+      "epoch": 13.101354303189165,
+      "grad_norm": 0.2383953481912613,
+      "learning_rate": 0.0006,
+      "loss": 3.0848209381103517,
+      "step": 3760
+    },
+    {
+      "epoch": 13.136304062909568,
+      "grad_norm": 0.2120991200208664,
+      "learning_rate": 0.0006,
+      "loss": 3.0751800537109375,
+      "step": 3770
+    },
+    {
+      "epoch": 13.17125382262997,
+      "grad_norm": 0.22427545487880707,
+      "learning_rate": 0.0006,
+      "loss": 3.070135498046875,
+      "step": 3780
+    },
+    {
+      "epoch": 13.206203582350371,
+      "grad_norm": 0.23202325403690338,
+      "learning_rate": 0.0006,
+      "loss": 3.0774150848388673,
+      "step": 3790
+    },
+    {
+      "epoch": 13.241153342070774,
+      "grad_norm": 0.23149551451206207,
+      "learning_rate": 0.0006,
+      "loss": 3.107881546020508,
+      "step": 3800
+    },
+    {
+      "epoch": 13.276103101791175,
+      "grad_norm": 0.23152674734592438,
+      "learning_rate": 0.0006,
+      "loss": 3.0857669830322267,
+      "step": 3810
+    },
+    {
+      "epoch": 13.297072957623417,
+      "eval_loss": 3.5316123962402344,
+      "eval_runtime": 3.2599,
+      "eval_samples_per_second": 748.794,
+      "eval_steps_per_second": 11.964,
+      "step": 3816
+    },
+    {
+      "epoch": 13.311052861511577,
+      "grad_norm": 0.2318379133939743,
+      "learning_rate": 0.0006,
+      "loss": 3.089146041870117,
+      "step": 3820
+    },
+    {
+      "epoch": 13.346002621231978,
+      "grad_norm": 0.24000167846679688,
+      "learning_rate": 0.0006,
+      "loss": 3.108494758605957,
+      "step": 3830
+    },
+    {
+      "epoch": 13.380952380952381,
+      "grad_norm": 0.2473367303609848,
+      "learning_rate": 0.0006,
+      "loss": 3.100922393798828,
+      "step": 3840
+    },
+    {
+      "epoch": 13.415902140672783,
+      "grad_norm": 0.21381697058677673,
+      "learning_rate": 0.0006,
+      "loss": 3.109054946899414,
+      "step": 3850
+    },
+    {
+      "epoch": 13.450851900393184,
+      "grad_norm": 0.22325816750526428,
+      "learning_rate": 0.0006,
+      "loss": 3.1112930297851564,
+      "step": 3860
+    },
+    {
+      "epoch": 13.485801660113587,
+      "grad_norm": 0.22084319591522217,
+      "learning_rate": 0.0006,
+      "loss": 3.1059669494628905,
+      "step": 3870
+    },
+    {
+      "epoch": 13.520751419833989,
+      "grad_norm": 0.22149917483329773,
+      "learning_rate": 0.0006,
+      "loss": 3.1137500762939454,
+      "step": 3880
+    },
+    {
+      "epoch": 13.54871122761031,
+      "eval_loss": 3.5243887901306152,
+      "eval_runtime": 3.2447,
+      "eval_samples_per_second": 752.297,
+      "eval_steps_per_second": 12.02,
+      "step": 3888
+    },
+    {
+      "epoch": 13.55570117955439,
+      "grad_norm": 0.20428219437599182,
+      "learning_rate": 0.0006,
+      "loss": 3.1112882614135744,
+      "step": 3890
+    },
+    {
+      "epoch": 13.590650939274793,
+      "grad_norm": 0.22512096166610718,
+      "learning_rate": 0.0006,
+      "loss": 3.097181701660156,
+      "step": 3900
+    },
+    {
+      "epoch": 13.625600698995195,
+      "grad_norm": 0.20783133804798126,
+      "learning_rate": 0.0006,
+      "loss": 3.115766906738281,
+      "step": 3910
+    },
+    {
+      "epoch": 13.660550458715596,
+      "grad_norm": 0.227244034409523,
+      "learning_rate": 0.0006,
+      "loss": 3.1228054046630858,
+      "step": 3920
+    },
+    {
+      "epoch": 13.695500218435999,
+      "grad_norm": 0.23311980068683624,
+      "learning_rate": 0.0006,
+      "loss": 3.116320037841797,
+      "step": 3930
+    },
+    {
+      "epoch": 13.7304499781564,
+      "grad_norm": 0.22328364849090576,
+      "learning_rate": 0.0006,
+      "loss": 3.133649444580078,
+      "step": 3940
+    },
+    {
+      "epoch": 13.765399737876802,
+      "grad_norm": 0.23404401540756226,
+      "learning_rate": 0.0006,
+      "loss": 3.1179473876953123,
+      "step": 3950
+    },
+    {
+      "epoch": 13.800349497597203,
+      "grad_norm": 0.21677330136299133,
+      "learning_rate": 0.0006,
+      "loss": 3.124909210205078,
+      "step": 3960
+    },
+    {
+      "epoch": 13.800349497597203,
+      "eval_loss": 3.5214247703552246,
+      "eval_runtime": 3.251,
+      "eval_samples_per_second": 750.856,
+      "eval_steps_per_second": 11.996,
+      "step": 3960
+    },
+    {
+      "epoch": 13.835299257317606,
+      "grad_norm": 0.2086809277534485,
+      "learning_rate": 0.0006,
+      "loss": 3.1245502471923827,
+      "step": 3970
+    },
+    {
+      "epoch": 13.870249017038008,
+      "grad_norm": 0.2378126084804535,
+      "learning_rate": 0.0006,
+      "loss": 3.111708068847656,
+      "step": 3980
+    },
+    {
+      "epoch": 13.905198776758409,
+      "grad_norm": 0.22283878922462463,
+      "learning_rate": 0.0006,
+      "loss": 3.1183197021484377,
+      "step": 3990
+    },
+    {
+      "epoch": 13.940148536478812,
+      "grad_norm": 0.21770469844341278,
+      "learning_rate": 0.0006,
+      "loss": 3.1238304138183595,
+      "step": 4000
+    },
+    {
+      "epoch": 13.975098296199214,
+      "grad_norm": 0.22705180943012238,
+      "learning_rate": 0.0006,
+      "loss": 3.123646354675293,
+      "step": 4010
+    },
+    {
+      "epoch": 14.00698995194408,
+      "grad_norm": 0.2921150326728821,
+      "learning_rate": 0.0006,
+      "loss": 3.110612678527832,
+      "step": 4020
+    },
+    {
+      "epoch": 14.041939711664483,
+      "grad_norm": 0.3078233301639557,
+      "learning_rate": 0.0006,
+      "loss": 3.0525794982910157,
+      "step": 4030
+    },
+    {
+      "epoch": 14.048929663608563,
+      "eval_loss": 3.5491466522216797,
+      "eval_runtime": 3.2446,
+      "eval_samples_per_second": 752.337,
+      "eval_steps_per_second": 12.02,
+      "step": 4032
+    },
+    {
+      "epoch": 14.076889471384884,
+      "grad_norm": 0.24426162242889404,
+      "learning_rate": 0.0006,
+      "loss": 3.058322715759277,
+      "step": 4040
+    },
+    {
+      "epoch": 14.111839231105286,
+      "grad_norm": 0.24375401437282562,
+      "learning_rate": 0.0006,
+      "loss": 3.0558265686035155,
+      "step": 4050
+    },
+    {
+      "epoch": 14.146788990825687,
+      "grad_norm": 0.22461462020874023,
+      "learning_rate": 0.0006,
+      "loss": 3.0573266983032226,
+      "step": 4060
+    },
+    {
+      "epoch": 14.18173875054609,
+      "grad_norm": 0.2431996613740921,
+      "learning_rate": 0.0006,
+      "loss": 3.0655328750610353,
+      "step": 4070
+    },
+    {
+      "epoch": 14.216688510266492,
+      "grad_norm": 0.21942028403282166,
+      "learning_rate": 0.0006,
+      "loss": 3.076056480407715,
+      "step": 4080
+    },
+    {
+      "epoch": 14.251638269986893,
+      "grad_norm": 0.2372651845216751,
+      "learning_rate": 0.0006,
+      "loss": 3.0641759872436523,
+      "step": 4090
+    },
+    {
+      "epoch": 14.286588029707296,
+      "grad_norm": 0.2462606132030487,
+      "learning_rate": 0.0006,
+      "loss": 3.083588790893555,
+      "step": 4100
+    },
+    {
+      "epoch": 14.300567933595456,
+      "eval_loss": 3.528660774230957,
+      "eval_runtime": 3.2711,
+      "eval_samples_per_second": 746.228,
+      "eval_steps_per_second": 11.923,
+      "step": 4104
+    },
+    {
+      "epoch": 14.321537789427698,
+      "grad_norm": 0.21411260962486267,
+      "learning_rate": 0.0006,
+      "loss": 3.083404541015625,
+      "step": 4110
+    },
+    {
+      "epoch": 14.356487549148099,
+      "grad_norm": 0.2222680300474167,
+      "learning_rate": 0.0006,
+      "loss": 3.089621734619141,
+      "step": 4120
+    },
+    {
+      "epoch": 14.391437308868502,
+      "grad_norm": 0.21534475684165955,
+      "learning_rate": 0.0006,
+      "loss": 3.0944677352905274,
+      "step": 4130
+    },
+    {
+      "epoch": 14.426387068588904,
+      "grad_norm": 0.2287389487028122,
+      "learning_rate": 0.0006,
+      "loss": 3.0854726791381837,
+      "step": 4140
+    },
+    {
+      "epoch": 14.461336828309305,
+      "grad_norm": 0.21588699519634247,
+      "learning_rate": 0.0006,
+      "loss": 3.0785106658935546,
+      "step": 4150
+    },
+    {
+      "epoch": 14.496286588029708,
+      "grad_norm": 0.2173336297273636,
+      "learning_rate": 0.0006,
+      "loss": 3.0937904357910155,
+      "step": 4160
+    },
+    {
+      "epoch": 14.53123634775011,
+      "grad_norm": 0.21985146403312683,
+      "learning_rate": 0.0006,
+      "loss": 3.0891979217529295,
+      "step": 4170
+    },
+    {
+      "epoch": 14.552206203582351,
+      "eval_loss": 3.519540786743164,
+      "eval_runtime": 3.251,
+      "eval_samples_per_second": 750.839,
+      "eval_steps_per_second": 11.996,
+      "step": 4176
+    },
+    {
+      "epoch": 14.56618610747051,
+      "grad_norm": 0.2155863493680954,
+      "learning_rate": 0.0006,
+      "loss": 3.078486442565918,
+      "step": 4180
+    },
+    {
+      "epoch": 14.601135867190912,
+      "grad_norm": 0.2322133183479309,
+      "learning_rate": 0.0006,
+      "loss": 3.0994333267211913,
+      "step": 4190
+    },
+    {
+      "epoch": 14.636085626911315,
+      "grad_norm": 0.20761388540267944,
+      "learning_rate": 0.0006,
+      "loss": 3.0935203552246096,
+      "step": 4200
+    },
+    {
+      "epoch": 14.671035386631717,
+      "grad_norm": 0.22560258209705353,
+      "learning_rate": 0.0006,
+      "loss": 3.080997657775879,
+      "step": 4210
+    },
+    {
+      "epoch": 14.705985146352118,
+      "grad_norm": 0.23228737711906433,
+      "learning_rate": 0.0006,
+      "loss": 3.103913497924805,
+      "step": 4220
+    },
+    {
+      "epoch": 14.740934906072521,
+      "grad_norm": 0.21750810742378235,
+      "learning_rate": 0.0006,
+      "loss": 3.112936019897461,
+      "step": 4230
+    },
+    {
+      "epoch": 14.775884665792923,
+      "grad_norm": 0.19724830985069275,
+      "learning_rate": 0.0006,
+      "loss": 3.0930578231811525,
+      "step": 4240
+    },
+    {
+      "epoch": 14.803844473569244,
+      "eval_loss": 3.513853073120117,
+      "eval_runtime": 3.2626,
+      "eval_samples_per_second": 748.169,
+      "eval_steps_per_second": 11.954,
+      "step": 4248
+    },
+    {
+      "epoch": 14.810834425513324,
+      "grad_norm": 0.2039424479007721,
+      "learning_rate": 0.0006,
+      "loss": 3.1030202865600587,
+      "step": 4250
+    },
+    {
+      "epoch": 14.845784185233727,
+      "grad_norm": 0.21879424154758453,
+      "learning_rate": 0.0006,
+      "loss": 3.105409622192383,
+      "step": 4260
+    },
+    {
+      "epoch": 14.880733944954128,
+      "grad_norm": 0.2264726758003235,
+      "learning_rate": 0.0006,
+      "loss": 3.099333953857422,
+      "step": 4270
+    },
+    {
+      "epoch": 14.91568370467453,
+      "grad_norm": 0.22432805597782135,
+      "learning_rate": 0.0006,
+      "loss": 3.1190414428710938,
+      "step": 4280
+    },
+    {
+      "epoch": 14.950633464394933,
+      "grad_norm": 0.22344641387462616,
+      "learning_rate": 0.0006,
+      "loss": 3.1041292190551757,
+      "step": 4290
+    },
+    {
+      "epoch": 14.985583224115334,
+      "grad_norm": 0.24332766234874725,
+      "learning_rate": 0.0006,
+      "loss": 3.092624473571777,
+      "step": 4300
+    },
+    {
+      "epoch": 15.0174748798602,
+      "grad_norm": 0.39760899543762207,
+      "learning_rate": 0.0006,
+      "loss": 3.083616256713867,
+      "step": 4310
+    },
+    {
+      "epoch": 15.052424639580602,
+      "grad_norm": 0.2631627023220062,
+      "learning_rate": 0.0006,
+      "loss": 3.0473527908325195,
+      "step": 4320
+    },
+    {
+      "epoch": 15.052424639580602,
+      "eval_loss": 3.538908004760742,
+      "eval_runtime": 3.2667,
+      "eval_samples_per_second": 747.246,
+      "eval_steps_per_second": 11.939,
+      "step": 4320
+    },
+    {
+      "epoch": 15.087374399301005,
+      "grad_norm": 0.20825771987438202,
+      "learning_rate": 0.0006,
+      "loss": 3.0433197021484375,
+      "step": 4330
+    },
+    {
+      "epoch": 15.122324159021407,
+      "grad_norm": 0.23742148280143738,
+      "learning_rate": 0.0006,
+      "loss": 3.040219306945801,
+      "step": 4340
+    },
+    {
+      "epoch": 15.157273918741808,
+      "grad_norm": 0.25612542033195496,
+      "learning_rate": 0.0006,
+      "loss": 3.050073432922363,
+      "step": 4350
+    },
+    {
+      "epoch": 15.192223678462211,
+      "grad_norm": 0.21400567889213562,
+      "learning_rate": 0.0006,
+      "loss": 3.0390625,
+      "step": 4360
+    },
+    {
+      "epoch": 15.227173438182612,
+      "grad_norm": 0.22660741209983826,
+      "learning_rate": 0.0006,
+      "loss": 3.045100784301758,
+      "step": 4370
+    },
+    {
+      "epoch": 15.262123197903014,
+      "grad_norm": 0.2552482485771179,
+      "learning_rate": 0.0006,
+      "loss": 3.059720993041992,
+      "step": 4380
+    },
+    {
+      "epoch": 15.297072957623417,
+      "grad_norm": 0.23194336891174316,
+      "learning_rate": 0.0006,
+      "loss": 3.0541946411132814,
+      "step": 4390
+    },
+    {
+      "epoch": 15.304062909567497,
+      "eval_loss": 3.524683713912964,
+      "eval_runtime": 3.251,
+      "eval_samples_per_second": 750.848,
+      "eval_steps_per_second": 11.996,
+      "step": 4392
+    },
+    {
+      "epoch": 15.332022717343818,
+      "grad_norm": 0.2294771820306778,
+      "learning_rate": 0.0006,
+      "loss": 3.070947265625,
+      "step": 4400
+    },
+    {
+      "epoch": 15.36697247706422,
+      "grad_norm": 0.2262076437473297,
+      "learning_rate": 0.0006,
+      "loss": 3.061265563964844,
+      "step": 4410
+    },
+    {
+      "epoch": 15.401922236784623,
+      "grad_norm": 0.2301194667816162,
+      "learning_rate": 0.0006,
+      "loss": 3.0521621704101562,
+      "step": 4420
+    },
+    {
+      "epoch": 15.436871996505024,
+      "grad_norm": 0.2328759878873825,
+      "learning_rate": 0.0006,
+      "loss": 3.0701065063476562,
+      "step": 4430
+    },
+    {
+      "epoch": 15.471821756225426,
+      "grad_norm": 0.2372756451368332,
+      "learning_rate": 0.0006,
+      "loss": 3.059339904785156,
+      "step": 4440
+    },
+    {
+      "epoch": 15.506771515945829,
+      "grad_norm": 0.2372213751077652,
+      "learning_rate": 0.0006,
+      "loss": 3.0750497817993163,
+      "step": 4450
+    },
+    {
+      "epoch": 15.54172127566623,
+      "grad_norm": 0.222769096493721,
+      "learning_rate": 0.0006,
+      "loss": 3.0793590545654297,
+      "step": 4460
+    },
+    {
+      "epoch": 15.55570117955439,
+      "eval_loss": 3.5129148960113525,
+      "eval_runtime": 3.2642,
+      "eval_samples_per_second": 747.806,
+      "eval_steps_per_second": 11.948,
+      "step": 4464
+    },
+    {
+      "epoch": 15.576671035386632,
+      "grad_norm": 0.2298610508441925,
+      "learning_rate": 0.0006,
+      "loss": 3.080717849731445,
+      "step": 4470
+    },
+    {
+      "epoch": 15.611620795107033,
+      "grad_norm": 0.22263580560684204,
+      "learning_rate": 0.0006,
+      "loss": 3.0787052154541015,
+      "step": 4480
+    },
+    {
+      "epoch": 15.646570554827436,
+      "grad_norm": 0.2506346106529236,
+      "learning_rate": 0.0006,
+      "loss": 3.062752914428711,
+      "step": 4490
+    },
+    {
+      "epoch": 15.681520314547837,
+      "grad_norm": 0.22322842478752136,
+      "learning_rate": 0.0006,
+      "loss": 3.0809616088867187,
+      "step": 4500
+    },
+    {
+      "epoch": 15.716470074268239,
+      "grad_norm": 0.2202921211719513,
+      "learning_rate": 0.0006,
+      "loss": 3.0854751586914064,
+      "step": 4510
+    },
+    {
+      "epoch": 15.751419833988642,
+      "grad_norm": 0.21363793313503265,
+      "learning_rate": 0.0006,
+      "loss": 3.085614776611328,
+      "step": 4520
+    },
+    {
+      "epoch": 15.786369593709043,
+      "grad_norm": 0.21110957860946655,
+      "learning_rate": 0.0006,
+      "loss": 3.077484893798828,
+      "step": 4530
+    },
+    {
+      "epoch": 15.807339449541285,
+      "eval_loss": 3.5101675987243652,
+      "eval_runtime": 3.2591,
+      "eval_samples_per_second": 748.989,
+      "eval_steps_per_second": 11.967,
+      "step": 4536
+    },
+    {
+      "epoch": 15.821319353429445,
+      "grad_norm": 0.25022223591804504,
+      "learning_rate": 0.0006,
+      "loss": 3.088617706298828,
+      "step": 4540
+    },
+    {
+      "epoch": 15.856269113149848,
+      "grad_norm": 0.2117125242948532,
+      "learning_rate": 0.0006,
+      "loss": 3.0797492980957033,
+      "step": 4550
+    },
+    {
+      "epoch": 15.89121887287025,
+      "grad_norm": 0.21591436862945557,
+      "learning_rate": 0.0006,
+      "loss": 3.090804672241211,
+      "step": 4560
+    },
+    {
+      "epoch": 15.92616863259065,
+      "grad_norm": 0.2173730432987213,
+      "learning_rate": 0.0006,
+      "loss": 3.0834587097167967,
+      "step": 4570
+    },
+    {
+      "epoch": 15.961118392311054,
+      "grad_norm": 0.2097509503364563,
+      "learning_rate": 0.0006,
+      "loss": 3.0933273315429686,
+      "step": 4580
+    },
+    {
+      "epoch": 15.996068152031455,
+      "grad_norm": 0.22636964917182922,
+      "learning_rate": 0.0006,
+      "loss": 3.0893348693847655,
+      "step": 4590
+    },
+    {
+      "epoch": 16.027959807776323,
+      "grad_norm": 0.2694890797138214,
+      "learning_rate": 0.0006,
+      "loss": 3.0205278396606445,
+      "step": 4600
+    },
+    {
+      "epoch": 16.055919615552643,
+      "eval_loss": 3.5364325046539307,
+      "eval_runtime": 3.261,
+      "eval_samples_per_second": 748.545,
+      "eval_steps_per_second": 11.96,
+      "step": 4608
+    },
+    {
+      "epoch": 16.062909567496725,
+      "grad_norm": 0.25205299258232117,
+      "learning_rate": 0.0006,
+      "loss": 3.016530227661133,
+      "step": 4610
+    },
+    {
+      "epoch": 16.097859327217126,
+      "grad_norm": 0.24426576495170593,
+      "learning_rate": 0.0006,
+      "loss": 3.0180795669555662,
+      "step": 4620
+    },
+    {
+      "epoch": 16.132809086937527,
+      "grad_norm": 0.2393605262041092,
+      "learning_rate": 0.0006,
+      "loss": 3.0269186019897463,
+      "step": 4630
+    },
+    {
+      "epoch": 16.16775884665793,
+      "grad_norm": 0.23526199162006378,
+      "learning_rate": 0.0006,
+      "loss": 3.0400951385498045,
+      "step": 4640
+    },
+    {
+      "epoch": 16.20270860637833,
+      "grad_norm": 0.21889753639698029,
+      "learning_rate": 0.0006,
+      "loss": 3.023430824279785,
+      "step": 4650
+    },
+    {
+      "epoch": 16.23765836609873,
+      "grad_norm": 0.2326001077890396,
+      "learning_rate": 0.0006,
+      "loss": 3.042998504638672,
+      "step": 4660
+    },
+    {
+      "epoch": 16.272608125819136,
+      "grad_norm": 0.21788732707500458,
+      "learning_rate": 0.0006,
+      "loss": 3.0386539459228517,
+      "step": 4670
+    },
+    {
+      "epoch": 16.307557885539538,
+      "grad_norm": 0.22534075379371643,
+      "learning_rate": 0.0006,
+      "loss": 3.0418207168579103,
+      "step": 4680
+    },
+    {
+      "epoch": 16.307557885539538,
+      "eval_loss": 3.5240864753723145,
+      "eval_runtime": 3.3067,
+      "eval_samples_per_second": 738.198,
+      "eval_steps_per_second": 11.794,
+      "step": 4680
+    },
+    {
+      "epoch": 16.34250764525994,
+      "grad_norm": 0.26475340127944946,
+      "learning_rate": 0.0006,
+      "loss": 3.0590129852294923,
+      "step": 4690
+    },
+    {
+      "epoch": 16.37745740498034,
+      "grad_norm": 0.24520041048526764,
+      "learning_rate": 0.0006,
+      "loss": 3.038155364990234,
+      "step": 4700
+    },
+    {
+      "epoch": 16.412407164700742,
+      "grad_norm": 0.22435322403907776,
+      "learning_rate": 0.0006,
+      "loss": 3.0530975341796873,
+      "step": 4710
+    },
+    {
+      "epoch": 16.447356924421143,
+      "grad_norm": 0.20739896595478058,
+      "learning_rate": 0.0006,
+      "loss": 3.060129928588867,
+      "step": 4720
+    },
+    {
+      "epoch": 16.482306684141548,
+      "grad_norm": 0.27902841567993164,
+      "learning_rate": 0.0006,
+      "loss": 3.0550994873046875,
+      "step": 4730
+    },
+    {
+      "epoch": 16.51725644386195,
+      "grad_norm": 0.21780024468898773,
+      "learning_rate": 0.0006,
+      "loss": 3.0552465438842775,
+      "step": 4740
+    },
+    {
+      "epoch": 16.55220620358235,
+      "grad_norm": 0.20766344666481018,
+      "learning_rate": 0.0006,
+      "loss": 3.052708053588867,
+      "step": 4750
+    },
+    {
+      "epoch": 16.55919615552643,
+      "eval_loss": 3.513065814971924,
+      "eval_runtime": 3.3655,
+      "eval_samples_per_second": 725.296,
+      "eval_steps_per_second": 11.588,
+      "step": 4752
+    },
+    {
+      "epoch": 16.587155963302752,
+      "grad_norm": 0.23465462028980255,
+      "learning_rate": 0.0006,
+      "loss": 3.062264060974121,
+      "step": 4760
+    },
+    {
+      "epoch": 16.622105723023154,
+      "grad_norm": 0.20552541315555573,
+      "learning_rate": 0.0006,
+      "loss": 3.0664562225341796,
+      "step": 4770
+    },
+    {
+      "epoch": 16.657055482743555,
+      "grad_norm": 0.21770046651363373,
+      "learning_rate": 0.0006,
+      "loss": 3.0766639709472656,
+      "step": 4780
+    },
+    {
+      "epoch": 16.692005242463956,
+      "grad_norm": 0.22245261073112488,
+      "learning_rate": 0.0006,
+      "loss": 3.055426025390625,
+      "step": 4790
+    },
+    {
+      "epoch": 16.72695500218436,
+      "grad_norm": 0.21591559052467346,
+      "learning_rate": 0.0006,
+      "loss": 3.064352607727051,
+      "step": 4800
+    },
+    {
+      "epoch": 16.761904761904763,
+      "grad_norm": 0.23161092400550842,
+      "learning_rate": 0.0006,
+      "loss": 3.062070846557617,
+      "step": 4810
+    },
+    {
+      "epoch": 16.796854521625164,
+      "grad_norm": 0.2139396071434021,
+      "learning_rate": 0.0006,
+      "loss": 3.078191566467285,
+      "step": 4820
+    },
+    {
+      "epoch": 16.810834425513324,
+      "eval_loss": 3.5091755390167236,
+      "eval_runtime": 7.5303,
+      "eval_samples_per_second": 324.159,
+      "eval_steps_per_second": 5.179,
+      "step": 4824
+    },
+    {
+      "epoch": 16.831804281345565,
+      "grad_norm": 0.2298831343650818,
+      "learning_rate": 0.0006,
+      "loss": 3.0621850967407225,
+      "step": 4830
+    },
+    {
+      "epoch": 16.866754041065967,
+      "grad_norm": 0.1998245120048523,
+      "learning_rate": 0.0006,
+      "loss": 3.064693260192871,
+      "step": 4840
+    },
+    {
+      "epoch": 16.90170380078637,
+      "grad_norm": 0.21080723404884338,
+      "learning_rate": 0.0006,
+      "loss": 3.0811368942260744,
+      "step": 4850
+    },
+    {
+      "epoch": 16.936653560506773,
+      "grad_norm": 0.20483531057834625,
+      "learning_rate": 0.0006,
+      "loss": 3.056428337097168,
+      "step": 4860
+    },
+    {
+      "epoch": 16.971603320227175,
+      "grad_norm": 0.221528559923172,
+      "learning_rate": 0.0006,
+      "loss": 3.069818305969238,
+      "step": 4870
+    },
+    {
+      "epoch": 17.00349497597204,
+      "grad_norm": 0.3038545846939087,
+      "learning_rate": 0.0006,
+      "loss": 3.0453723907470702,
+      "step": 4880
+    },
+    {
+      "epoch": 17.03844473569244,
+      "grad_norm": 0.2575819492340088,
+      "learning_rate": 0.0006,
+      "loss": 2.9968231201171873,
+      "step": 4890
+    },
+    {
+      "epoch": 17.059414591524682,
+      "eval_loss": 3.5293376445770264,
+      "eval_runtime": 7.3793,
+      "eval_samples_per_second": 330.788,
+      "eval_steps_per_second": 5.285,
+      "step": 4896
+    },
+    {
+      "epoch": 17.073394495412845,
+      "grad_norm": 0.24228203296661377,
+      "learning_rate": 0.0006,
+      "loss": 3.012477493286133,
+      "step": 4900
+    },
+    {
+      "epoch": 17.108344255133247,
+      "grad_norm": 0.24087928235530853,
+      "learning_rate": 0.0006,
+      "loss": 3.000431442260742,
+      "step": 4910
+    },
+    {
+      "epoch": 17.143294014853648,
+      "grad_norm": 0.2340719848871231,
+      "learning_rate": 0.0006,
+      "loss": 3.0005977630615233,
+      "step": 4920
+    },
+    {
+      "epoch": 17.17824377457405,
+      "grad_norm": 0.23718640208244324,
+      "learning_rate": 0.0006,
+      "loss": 3.0187118530273436,
+      "step": 4930
+    },
+    {
+      "epoch": 17.21319353429445,
+      "grad_norm": 0.23856772482395172,
+      "learning_rate": 0.0006,
+      "loss": 3.019180679321289,
+      "step": 4940
+    },
+    {
+      "epoch": 17.248143294014852,
+      "grad_norm": 0.23096665740013123,
+      "learning_rate": 0.0006,
+      "loss": 3.0042135238647463,
+      "step": 4950
+    },
+    {
+      "epoch": 17.283093053735257,
+      "grad_norm": 0.22185878455638885,
+      "learning_rate": 0.0006,
+      "loss": 3.024405097961426,
+      "step": 4960
+    },
+    {
+      "epoch": 17.311052861511577,
+      "eval_loss": 3.5166492462158203,
+      "eval_runtime": 7.4626,
+      "eval_samples_per_second": 327.099,
+      "eval_steps_per_second": 5.226,
+      "step": 4968
+    },
+    {
+      "epoch": 17.31804281345566,
+      "grad_norm": 0.23213611543178558,
+      "learning_rate": 0.0006,
+      "loss": 3.029388999938965,
+      "step": 4970
+    },
+    {
+      "epoch": 17.35299257317606,
+      "grad_norm": 0.247343048453331,
+      "learning_rate": 0.0006,
+      "loss": 3.012360382080078,
+      "step": 4980
+    },
+    {
+      "epoch": 17.38794233289646,
+      "grad_norm": 0.24156036972999573,
+      "learning_rate": 0.0006,
+      "loss": 3.026628875732422,
+      "step": 4990
+    },
+    {
+      "epoch": 17.422892092616863,
+      "grad_norm": 0.23647816479206085,
+      "learning_rate": 0.0006,
+      "loss": 3.042314910888672,
+      "step": 5000
+    },
+    {
+      "epoch": 17.457841852337264,
+      "grad_norm": 0.2525269091129303,
+      "learning_rate": 0.0006,
+      "loss": 3.0462404251098634,
+      "step": 5010
+    },
+    {
+      "epoch": 17.492791612057665,
+      "grad_norm": 0.22319558262825012,
+      "learning_rate": 0.0006,
+      "loss": 3.048808479309082,
+      "step": 5020
+    },
+    {
+      "epoch": 17.52774137177807,
+      "grad_norm": 0.22997041046619415,
+      "learning_rate": 0.0006,
+      "loss": 3.0349891662597654,
+      "step": 5030
+    },
+    {
+      "epoch": 17.56269113149847,
+      "grad_norm": 0.21920591592788696,
+      "learning_rate": 0.0006,
+      "loss": 3.0409507751464844,
+      "step": 5040
+    },
+    {
+      "epoch": 17.56269113149847,
+      "eval_loss": 3.5158889293670654,
+      "eval_runtime": 7.4233,
+      "eval_samples_per_second": 328.828,
+      "eval_steps_per_second": 5.254,
+      "step": 5040
+    },
+    {
+      "epoch": 17.597640891218873,
+      "grad_norm": 0.23242522776126862,
+      "learning_rate": 0.0006,
+      "loss": 3.0345521926879884,
+      "step": 5050
+    },
+    {
+      "epoch": 17.632590650939274,
+      "grad_norm": 0.23279322683811188,
+      "learning_rate": 0.0006,
+      "loss": 3.0375698089599608,
+      "step": 5060
+    },
+    {
+      "epoch": 17.667540410659676,
+      "grad_norm": 0.22396472096443176,
+      "learning_rate": 0.0006,
+      "loss": 3.056574249267578,
+      "step": 5070
+    },
+    {
+      "epoch": 17.702490170380077,
+      "grad_norm": 0.19807346165180206,
+      "learning_rate": 0.0006,
+      "loss": 3.0603185653686524,
+      "step": 5080
+    },
+    {
+      "epoch": 17.737439930100482,
+      "grad_norm": 0.21861477196216583,
+      "learning_rate": 0.0006,
+      "loss": 3.0521718978881838,
+      "step": 5090
+    },
+    {
+      "epoch": 17.772389689820884,
+      "grad_norm": 0.2207152396440506,
+      "learning_rate": 0.0006,
+      "loss": 3.0594032287597654,
+      "step": 5100
+    },
+    {
+      "epoch": 17.807339449541285,
+      "grad_norm": 0.21890032291412354,
+      "learning_rate": 0.0006,
+      "loss": 3.063242721557617,
+      "step": 5110
+    },
+    {
+      "epoch": 17.814329401485367,
+      "eval_loss": 3.5121800899505615,
+      "eval_runtime": 7.4008,
+      "eval_samples_per_second": 329.829,
+      "eval_steps_per_second": 5.27,
+      "step": 5112
+    },
+    {
+      "epoch": 17.842289209261686,
+      "grad_norm": 0.21062296628952026,
+      "learning_rate": 0.0006,
+      "loss": 3.059429168701172,
+      "step": 5120
+    },
+    {
+      "epoch": 17.877238968982088,
+      "grad_norm": 0.20210039615631104,
+      "learning_rate": 0.0006,
+      "loss": 3.0671924591064452,
+      "step": 5130
+    },
+    {
+      "epoch": 17.91218872870249,
+      "grad_norm": 0.24482186138629913,
+      "learning_rate": 0.0006,
+      "loss": 3.0585289001464844,
+      "step": 5140
+    },
+    {
+      "epoch": 17.94713848842289,
+      "grad_norm": 0.2553030252456665,
+      "learning_rate": 0.0006,
+      "loss": 3.058705520629883,
+      "step": 5150
+    },
+    {
+      "epoch": 17.982088248143295,
+      "grad_norm": 0.21517649292945862,
+      "learning_rate": 0.0006,
+      "loss": 3.071492385864258,
+      "step": 5160
+    },
+    {
+      "epoch": 18.01397990388816,
+      "grad_norm": 0.40742236375808716,
+      "learning_rate": 0.0006,
+      "loss": 3.031229019165039,
+      "step": 5170
+    },
+    {
+      "epoch": 18.04892966360856,
+      "grad_norm": 0.3224371671676636,
+      "learning_rate": 0.0006,
+      "loss": 3.0113935470581055,
+      "step": 5180
+    },
+    {
+      "epoch": 18.062909567496725,
+      "eval_loss": 3.534085988998413,
+      "eval_runtime": 7.4409,
+      "eval_samples_per_second": 328.053,
+      "eval_steps_per_second": 5.241,
+      "step": 5184
+    },
+    {
+      "epoch": 18.083879423328966,
+      "grad_norm": 0.2747146487236023,
+      "learning_rate": 0.0006,
+      "loss": 2.9980716705322266,
+      "step": 5190
+    },
+    {
+      "epoch": 18.118829183049368,
+      "grad_norm": 0.2297176569700241,
+      "learning_rate": 0.0006,
+      "loss": 2.9935012817382813,
+      "step": 5200
+    },
+    {
+      "epoch": 18.15377894276977,
+      "grad_norm": 0.21988452970981598,
+      "learning_rate": 0.0006,
+      "loss": 2.99840087890625,
+      "step": 5210
+    },
+    {
+      "epoch": 18.18872870249017,
+      "grad_norm": 0.22812554240226746,
+      "learning_rate": 0.0006,
+      "loss": 3.0035776138305663,
+      "step": 5220
+    },
+    {
+      "epoch": 18.22367846221057,
+      "grad_norm": 0.20533959567546844,
+      "learning_rate": 0.0006,
+      "loss": 2.9987850189208984,
+      "step": 5230
+    },
+    {
+      "epoch": 18.258628221930973,
+      "grad_norm": 0.21126243472099304,
+      "learning_rate": 0.0006,
+      "loss": 3.0022491455078124,
+      "step": 5240
+    },
+    {
+      "epoch": 18.293577981651374,
+      "grad_norm": 0.22381673753261566,
+      "learning_rate": 0.0006,
+      "loss": 3.0073720932006838,
+      "step": 5250
+    },
+    {
+      "epoch": 18.314547837483616,
+      "eval_loss": 3.5182130336761475,
+      "eval_runtime": 7.437,
+      "eval_samples_per_second": 328.222,
+      "eval_steps_per_second": 5.244,
+      "step": 5256
+    },
+    {
+      "epoch": 18.32852774137178,
+      "grad_norm": 0.2268986701965332,
+      "learning_rate": 0.0006,
+      "loss": 3.01855525970459,
+      "step": 5260
+    },
+    {
+      "epoch": 18.36347750109218,
+      "grad_norm": 0.23119987547397614,
+      "learning_rate": 0.0006,
+      "loss": 3.0249351501464843,
+      "step": 5270
+    },
+    {
+      "epoch": 18.398427260812582,
+      "grad_norm": 0.2085934430360794,
+      "learning_rate": 0.0006,
+      "loss": 3.0055150985717773,
+      "step": 5280
+    },
+    {
+      "epoch": 18.433377020532983,
+      "grad_norm": 0.2297414243221283,
+      "learning_rate": 0.0006,
+      "loss": 3.024668502807617,
+      "step": 5290
+    },
+    {
+      "epoch": 18.468326780253385,
+      "grad_norm": 0.22136586904525757,
+      "learning_rate": 0.0006,
+      "loss": 3.016472816467285,
+      "step": 5300
+    },
+    {
+      "epoch": 18.503276539973786,
+      "grad_norm": 0.2561214864253998,
+      "learning_rate": 0.0006,
+      "loss": 3.0173313140869142,
+      "step": 5310
+    },
+    {
+      "epoch": 18.53822629969419,
+      "grad_norm": 0.21839040517807007,
+      "learning_rate": 0.0006,
+      "loss": 3.017107391357422,
+      "step": 5320
+    },
+    {
+      "epoch": 18.56618610747051,
+      "eval_loss": 3.509559154510498,
+      "eval_runtime": 7.4899,
+      "eval_samples_per_second": 325.907,
+      "eval_steps_per_second": 5.207,
+      "step": 5328
+    },
+    {
+      "epoch": 18.573176059414592,
+      "grad_norm": 0.24576397240161896,
+      "learning_rate": 0.0006,
+      "loss": 3.0373584747314455,
+      "step": 5330
+    },
+    {
+      "epoch": 18.608125819134994,
+      "grad_norm": 0.21707676351070404,
+      "learning_rate": 0.0006,
+      "loss": 3.0267887115478516,
+      "step": 5340
+    },
+    {
+      "epoch": 18.643075578855395,
+      "grad_norm": 0.22613626718521118,
+      "learning_rate": 0.0006,
+      "loss": 3.0409290313720705,
+      "step": 5350
+    },
+    {
+      "epoch": 18.678025338575797,
+      "grad_norm": 0.1976986527442932,
+      "learning_rate": 0.0006,
+      "loss": 3.0386127471923827,
+      "step": 5360
+    },
+    {
+      "epoch": 18.712975098296198,
+      "grad_norm": 0.2118467539548874,
+      "learning_rate": 0.0006,
+      "loss": 3.040230941772461,
+      "step": 5370
+    },
+    {
+      "epoch": 18.747924858016603,
+      "grad_norm": 0.20487086474895477,
+      "learning_rate": 0.0006,
+      "loss": 3.0428600311279297,
+      "step": 5380
+    },
+    {
+      "epoch": 18.782874617737004,
+      "grad_norm": 0.22586505115032196,
+      "learning_rate": 0.0006,
+      "loss": 3.047272491455078,
+      "step": 5390
+    },
+    {
+      "epoch": 18.817824377457406,
+      "grad_norm": 0.2086966633796692,
+      "learning_rate": 0.0006,
+      "loss": 3.024753761291504,
+      "step": 5400
+    },
+    {
+      "epoch": 18.817824377457406,
+      "eval_loss": 3.503643751144409,
+      "eval_runtime": 7.4719,
+      "eval_samples_per_second": 326.689,
+      "eval_steps_per_second": 5.22,
+      "step": 5400
+    },
+    {
+      "epoch": 18.852774137177807,
+      "grad_norm": 0.2347467690706253,
+      "learning_rate": 0.0006,
+      "loss": 3.038570785522461,
+      "step": 5410
+    },
+    {
+      "epoch": 18.88772389689821,
+      "grad_norm": 0.20837339758872986,
+      "learning_rate": 0.0006,
+      "loss": 3.033599281311035,
+      "step": 5420
+    },
+    {
+      "epoch": 18.92267365661861,
+      "grad_norm": 0.21020616590976715,
+      "learning_rate": 0.0006,
+      "loss": 3.0428997039794923,
+      "step": 5430
+    },
+    {
+      "epoch": 18.95762341633901,
+      "grad_norm": 0.19686657190322876,
+      "learning_rate": 0.0006,
+      "loss": 3.0497291564941404,
+      "step": 5440
+    },
+    {
+      "epoch": 18.992573176059416,
+      "grad_norm": 0.2169448435306549,
+      "learning_rate": 0.0006,
+      "loss": 3.0556671142578127,
+      "step": 5450
+    },
+    {
+      "epoch": 19.02446483180428,
+      "grad_norm": 0.32634595036506653,
+      "learning_rate": 0.0006,
+      "loss": 3.014866256713867,
+      "step": 5460
+    },
+    {
+      "epoch": 19.059414591524682,
+      "grad_norm": 0.24913258850574493,
+      "learning_rate": 0.0006,
+      "loss": 2.989482879638672,
+      "step": 5470
+    },
+    {
+      "epoch": 19.066404543468764,
+      "eval_loss": 3.5264627933502197,
+      "eval_runtime": 7.4619,
+      "eval_samples_per_second": 327.13,
+      "eval_steps_per_second": 5.227,
+      "step": 5472
+    },
+    {
+      "epoch": 19.094364351245087,
+      "grad_norm": 0.2381090670824051,
+      "learning_rate": 0.0006,
+      "loss": 2.976517105102539,
+      "step": 5480
+    },
+    {
+      "epoch": 19.12931411096549,
+      "grad_norm": 0.22422131896018982,
+      "learning_rate": 0.0006,
+      "loss": 2.969605255126953,
+      "step": 5490
+    },
+    {
+      "epoch": 19.16426387068589,
+      "grad_norm": 0.21318356692790985,
+      "learning_rate": 0.0006,
+      "loss": 2.98162956237793,
+      "step": 5500
+    },
+    {
+      "epoch": 19.19921363040629,
+      "grad_norm": 0.2570667266845703,
+      "learning_rate": 0.0006,
+      "loss": 2.9917987823486327,
+      "step": 5510
+    },
+    {
+      "epoch": 19.234163390126692,
+      "grad_norm": 0.23020485043525696,
+      "learning_rate": 0.0006,
+      "loss": 3.0056358337402345,
+      "step": 5520
+    },
+    {
+      "epoch": 19.269113149847094,
+      "grad_norm": 0.2144797295331955,
+      "learning_rate": 0.0006,
+      "loss": 3.0110912322998047,
+      "step": 5530
+    },
+    {
+      "epoch": 19.304062909567495,
+      "grad_norm": 0.23170755803585052,
+      "learning_rate": 0.0006,
+      "loss": 2.9843938827514647,
+      "step": 5540
+    },
+    {
+      "epoch": 19.31804281345566,
+      "eval_loss": 3.5164332389831543,
+      "eval_runtime": 7.4585,
+      "eval_samples_per_second": 327.28,
+      "eval_steps_per_second": 5.229,
+      "step": 5544
+    },
+    {
+      "epoch": 19.3390126692879,
+      "grad_norm": 0.20028246939182281,
+      "learning_rate": 0.0006,
+      "loss": 3.016252899169922,
+      "step": 5550
+    },
+    {
+      "epoch": 19.3739624290083,
+      "grad_norm": 0.23646661639213562,
+      "learning_rate": 0.0006,
+      "loss": 2.9998050689697267,
+      "step": 5560
+    },
+    {
+      "epoch": 19.408912188728703,
+      "grad_norm": 0.23107370734214783,
+      "learning_rate": 0.0006,
+      "loss": 3.0041255950927734,
+      "step": 5570
+    },
+    {
+      "epoch": 19.443861948449104,
+      "grad_norm": 0.30640512704849243,
+      "learning_rate": 0.0006,
+      "loss": 3.003022003173828,
+      "step": 5580
+    },
+    {
+      "epoch": 19.478811708169506,
+      "grad_norm": 0.23508583009243011,
+      "learning_rate": 0.0006,
+      "loss": 3.004806900024414,
+      "step": 5590
+    },
+    {
+      "epoch": 19.513761467889907,
+      "grad_norm": 0.22322413325309753,
+      "learning_rate": 0.0006,
+      "loss": 3.0096115112304687,
+      "step": 5600
+    },
+    {
+      "epoch": 19.548711227610312,
+      "grad_norm": 0.21756328642368317,
+      "learning_rate": 0.0006,
+      "loss": 3.0236843109130858,
+      "step": 5610
+    },
+    {
+      "epoch": 19.56968108344255,
+      "eval_loss": 3.509248733520508,
+      "eval_runtime": 7.4624,
+      "eval_samples_per_second": 327.105,
+      "eval_steps_per_second": 5.226,
+      "step": 5616
+    },
+    {
+      "epoch": 19.583660987330713,
+      "grad_norm": 0.21684585511684418,
+      "learning_rate": 0.0006,
+      "loss": 3.023478126525879,
+      "step": 5620
+    },
+    {
+      "epoch": 19.618610747051115,
+      "grad_norm": 0.21128204464912415,
+      "learning_rate": 0.0006,
+      "loss": 3.0205398559570313,
+      "step": 5630
+    },
+    {
+      "epoch": 19.653560506771516,
+      "grad_norm": 0.22280649840831757,
+      "learning_rate": 0.0006,
+      "loss": 3.010708045959473,
+      "step": 5640
+    },
+    {
+      "epoch": 19.688510266491917,
+      "grad_norm": 0.2070789784193039,
+      "learning_rate": 0.0006,
+      "loss": 3.0157245635986327,
+      "step": 5650
+    },
+    {
+      "epoch": 19.72346002621232,
+      "grad_norm": 0.202907532453537,
+      "learning_rate": 0.0006,
+      "loss": 3.014574432373047,
+      "step": 5660
+    },
+    {
+      "epoch": 19.75840978593272,
+      "grad_norm": 0.22183173894882202,
+      "learning_rate": 0.0006,
+      "loss": 3.0195291519165037,
+      "step": 5670
+    },
+    {
+      "epoch": 19.793359545653125,
+      "grad_norm": 0.2025201916694641,
+      "learning_rate": 0.0006,
+      "loss": 3.032331848144531,
+      "step": 5680
+    },
+    {
+      "epoch": 19.821319353429445,
+      "eval_loss": 3.503469705581665,
+      "eval_runtime": 7.5059,
+      "eval_samples_per_second": 325.212,
+      "eval_steps_per_second": 5.196,
+      "step": 5688
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 7175,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 25,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": false,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2.432688697775227e+18,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/runs/t2a/t2a_300M_C0_baseline/train.log b/runs/t2a/t2a_300M_C0_baseline/train.log
new file mode 100644
index 0000000000000000000000000000000000000000..4ecd397306af0ce3abceb5bf9813ae2a51710b81
--- /dev/null
+++ b/runs/t2a/t2a_300M_C0_baseline/train.log
@@ -0,0 +1,5420 @@
+W0819 03:25:45.295000 448278 torch/distributed/run.py:851] 
+W0819 03:25:45.295000 448278 torch/distributed/run.py:851] *****************************************
+W0819 03:25:45.295000 448278 torch/distributed/run.py:851] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+W0819 03:25:45.295000 448278 torch/distributed/run.py:851] *****************************************
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+total params: 145833984
+[dataset] loaded packed cache: 146474 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+total params: 145833984
+[dataset] loaded packed cache: 146474 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+total params: 145833984
+[dataset] loaded packed cache: 146474 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+total params: 145833984
+[dataset] loaded packed cache: 146474 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+total params: 145833984
+total params: 145833984
+total params: 145833984
+[dataset] loaded packed cache: 146474 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[dataset] loaded packed cache: 146474 train / 2441 eval blocks
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+[transformers] PyTorch: setting up devices
+[dataset] loaded packed cache: 146474 train / 2441 eval blocks
+total params: 145833984
+[transformers] PyTorch: setting up devices
+[dataset] loaded packed cache: 146474 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+[transformers] ***** Running training *****
+[transformers]   Num examples = 146,474
+[transformers]   Num Epochs = 25
+[transformers]   Num update steps per epoch = 287
+[transformers]   Instantaneous batch size per device = 8
+[transformers]   Total train batch size (w. parallel, distributed & accumulation) = 512
+[transformers]   Gradient Accumulation steps = 8
+[transformers]   Total optimization steps = 7,175
+[transformers]   Number of trainable parameters = 145,833,984
+wandb: WARNING Changes to your `wandb` environment variables will be ignored because your `wandb` session has already started. For more information on how to modify your settings with `wandb.init()` arguments, please refer to https://wandb.me/wandb-init.
+wandb: setting up run xarz21cw
+wandb: Tracking run with wandb version 0.28.2
+wandb: Run data is saved locally in /home/ubuntu/workspace/data-augmentations-for-pretraining/wandb/run-20260819_032604-xarz21cw
+wandb: Run `wandb offline` to turn off syncing.
+wandb: Syncing run t2a_300M_C0_baseline
+wandb: ⭐️ View project at https://wandb.ai/slowrun-wankers/data-aug-pretraining
+wandb: 🚀 View run at https://wandb.ai/slowrun-wankers/data-aug-pretraining/runs/xarz21cw
+
  0%|          | 0/7175 [00:00<?, ?it/s]wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+[rank1]:W0819 03:26:25.627000 448395 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+[rank5]:W0819 03:26:25.882000 448399 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+[rank3]:W0819 03:26:25.896000 448397 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+[rank7]:W0819 03:26:25.948000 448401 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+[rank6]:W0819 03:26:26.217000 448400 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+[rank4]:W0819 03:26:26.220000 448398 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+[rank2]:W0819 03:26:26.235000 448396 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+[rank0]:W0819 03:26:29.023000 448394 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+
  0%|          | 1/7175 [00:40<81:08:58, 40.72s/it]
                                                   
{'loss': '12.03', 'grad_norm': '1.81', 'learning_rate': '0', 'epoch': '0.003495'}
+
  0%|          | 1/7175 [00:40<81:08:58, 40.72s/it]
  0%|          | 2/7175 [00:41<34:11:45, 17.16s/it]
  0%|          | 3/7175 [00:42<19:10:16,  9.62s/it]
  0%|          | 4/7175 [00:42<12:06:46,  6.08s/it]
  0%|          | 5/7175 [00:43<8:12:42,  4.12s/it] 
  0%|          | 6/7175 [00:43<5:51:39,  2.94s/it]
  0%|          | 7/7175 [00:44<4:22:09,  2.19s/it]
  0%|          | 8/7175 [00:45<3:23:29,  1.70s/it]
  0%|          | 9/7175 [00:45<2:44:08,  1.37s/it]
  0%|          | 10/7175 [00:46<2:17:29,  1.15s/it]
                                                   
{'loss': '11.74', 'grad_norm': '1.756', 'learning_rate': '5.4e-05', 'epoch': '0.03495'}
+
  0%|          | 10/7175 [00:46<2:17:29,  1.15s/it]
  0%|          | 11/7175 [00:47<1:59:15,  1.00it/s]
  0%|          | 12/7175 [00:47<1:46:37,  1.12it/s]
  0%|          | 13/7175 [00:48<1:37:52,  1.22it/s]
  0%|          | 14/7175 [00:49<1:31:44,  1.30it/s]
  0%|          | 15/7175 [00:49<1:27:29,  1.36it/s]
  0%|          | 16/7175 [00:50<1:24:29,  1.41it/s]
  0%|          | 17/7175 [00:51<1:22:24,  1.45it/s]
  0%|          | 18/7175 [00:51<1:20:57,  1.47it/s]
  0%|          | 19/7175 [00:52<1:19:55,  1.49it/s]
  0%|          | 20/7175 [00:53<1:19:15,  1.50it/s]
                                                   
{'loss': '11.05', 'grad_norm': '1.688', 'learning_rate': '0.000114', 'epoch': '0.0699'}
+
  0%|          | 20/7175 [00:53<1:19:15,  1.50it/s]
  0%|          | 21/7175 [00:53<1:18:50,  1.51it/s]
  0%|          | 22/7175 [00:54<1:18:29,  1.52it/s]
  0%|          | 23/7175 [00:55<1:18:10,  1.52it/s]
  0%|          | 24/7175 [00:55<1:17:59,  1.53it/s]
  0%|          | 25/7175 [00:56<1:17:48,  1.53it/s]
  0%|          | 26/7175 [00:57<1:17:47,  1.53it/s]
  0%|          | 27/7175 [00:57<1:17:34,  1.54it/s]
  0%|          | 28/7175 [00:58<1:17:32,  1.54it/s]
  0%|          | 29/7175 [00:58<1:17:29,  1.54it/s]
  0%|          | 30/7175 [00:59<1:17:29,  1.54it/s]
                                                   
{'loss': '10.29', 'grad_norm': '1.673', 'learning_rate': '0.000174', 'epoch': '0.1048'}
+
  0%|          | 30/7175 [00:59<1:17:29,  1.54it/s]
  0%|          | 31/7175 [01:00<1:17:30,  1.54it/s]
  0%|          | 32/7175 [01:00<1:17:26,  1.54it/s]
  0%|          | 33/7175 [01:01<1:17:25,  1.54it/s]
  0%|          | 34/7175 [01:02<1:17:24,  1.54it/s]
  0%|          | 35/7175 [01:02<1:17:22,  1.54it/s]
  1%|          | 36/7175 [01:03<1:17:23,  1.54it/s]
  1%|          | 37/7175 [01:04<1:17:20,  1.54it/s]
  1%|          | 38/7175 [01:04<1:17:17,  1.54it/s]
  1%|          | 39/7175 [01:05<1:17:16,  1.54it/s]
  1%|          | 40/7175 [01:06<1:17:16,  1.54it/s]
                                                   
{'loss': '9.249', 'grad_norm': '1.539', 'learning_rate': '0.000234', 'epoch': '0.1398'}
+
  1%|          | 40/7175 [01:06<1:17:16,  1.54it/s]
  1%|          | 41/7175 [01:06<1:17:17,  1.54it/s]
  1%|          | 42/7175 [01:07<1:17:15,  1.54it/s]
  1%|          | 43/7175 [01:08<1:17:15,  1.54it/s]
  1%|          | 44/7175 [01:08<1:17:14,  1.54it/s]
  1%|          | 45/7175 [01:09<1:17:15,  1.54it/s]
  1%|          | 46/7175 [01:10<1:17:13,  1.54it/s]
  1%|          | 47/7175 [01:10<1:17:17,  1.54it/s]
  1%|          | 48/7175 [01:11<1:17:17,  1.54it/s]
  1%|          | 49/7175 [01:11<1:17:17,  1.54it/s]
  1%|          | 50/7175 [01:12<1:17:18,  1.54it/s]
                                                   
{'loss': '8.244', 'grad_norm': '0.9666', 'learning_rate': '0.000294', 'epoch': '0.1747'}
+
  1%|          | 50/7175 [01:12<1:17:18,  1.54it/s]
  1%|          | 51/7175 [01:13<1:17:23,  1.53it/s]
  1%|          | 52/7175 [01:13<1:17:23,  1.53it/s]
  1%|          | 53/7175 [01:14<1:17:25,  1.53it/s]
  1%|          | 54/7175 [01:15<1:17:29,  1.53it/s]
  1%|          | 55/7175 [01:15<1:17:33,  1.53it/s]
  1%|          | 56/7175 [01:16<1:17:31,  1.53it/s]
  1%|          | 57/7175 [01:17<1:17:30,  1.53it/s]
  1%|          | 58/7175 [01:17<1:17:27,  1.53it/s]
  1%|          | 59/7175 [01:18<1:17:22,  1.53it/s]
  1%|          | 60/7175 [01:19<1:17:27,  1.53it/s]
                                                   
{'loss': '7.69', 'grad_norm': '0.285', 'learning_rate': '0.000354', 'epoch': '0.2097'}
+
  1%|          | 60/7175 [01:19<1:17:27,  1.53it/s]
  1%|          | 61/7175 [01:19<1:17:35,  1.53it/s]
  1%|          | 62/7175 [01:20<1:17:34,  1.53it/s]
  1%|          | 63/7175 [01:21<1:17:28,  1.53it/s]
  1%|          | 64/7175 [01:21<1:17:27,  1.53it/s]
  1%|          | 65/7175 [01:22<1:17:25,  1.53it/s]
  1%|          | 66/7175 [01:23<1:17:24,  1.53it/s]
  1%|          | 67/7175 [01:23<1:17:31,  1.53it/s]
  1%|          | 68/7175 [01:24<1:17:33,  1.53it/s]
  1%|          | 69/7175 [01:25<1:17:35,  1.53it/s]
  1%|          | 70/7175 [01:25<1:17:29,  1.53it/s]
                                                   
{'loss': '7.505', 'grad_norm': '1.611', 'learning_rate': '0.000414', 'epoch': '0.2446'}
+
  1%|          | 70/7175 [01:25<1:17:29,  1.53it/s]
  1%|          | 71/7175 [01:26<1:17:33,  1.53it/s]
  1%|          | 72/7175 [01:27<1:17:31,  1.53it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 28.67it/s]
+
 15%|█▌        | 6/39 [00:00<00:02, 16.01it/s]
+
 21%|██        | 8/39 [00:00<00:02, 14.39it/s]
+
 26%|██▌       | 10/39 [00:00<00:02, 13.57it/s]
+
 31%|███       | 12/39 [00:00<00:02, 13.02it/s]
+
 36%|███▌      | 14/39 [00:01<00:01, 12.82it/s]
+
 41%|████      | 16/39 [00:01<00:01, 12.61it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 12.49it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 12.40it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 12.35it/s]
+
 62%|██████▏   | 24/39 [00:01<00:01, 12.32it/s]
+
 67%|██████▋   | 26/39 [00:01<00:01, 12.29it/s]
+
 72%|███████▏  | 28/39 [00:02<00:00, 12.27it/s]
+
 77%|███████▋  | 30/39 [00:02<00:00, 12.26it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 12.25it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 12.23it/s]
+
 92%|█████████▏| 36/39 [00:02<00:00, 12.24it/s]
+
 97%|█████████▋| 38/39 [00:02<00:00, 12.22it/s]
                                                   
+
                                               
{'eval_loss': '7.464', 'eval_runtime': '27.98', 'eval_samples_per_second': '87.24', 'eval_steps_per_second': '1.394', 'epoch': '0.2516'}
+
  1%|          | 72/7175 [01:55<1:17:31,  1.53it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.22it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-72
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-72/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-72/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.72it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.72it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-72/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-72/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-72/tokenizer_config.json
+[eval] epoch 0.252 step 72 loss 7.4635 (best 7.4635 @ ep 0.252, 0/20 since best)
+
+
                                               
  1%|          | 73/7175 [01:57<19:00:18,  9.63s/it]
  1%|          | 74/7175 [01:58<13:41:20,  6.94s/it]
  1%|          | 75/7175 [01:58<9:58:05,  5.05s/it] 
  1%|          | 76/7175 [01:59<7:21:47,  3.73s/it]
  1%|          | 77/7175 [02:00<5:32:33,  2.81s/it]
  1%|          | 78/7175 [02:00<4:16:03,  2.16s/it]
  1%|          | 79/7175 [02:01<3:22:32,  1.71s/it]
  1%|          | 80/7175 [02:02<2:44:56,  1.39s/it]
                                                   
{'loss': '7.347', 'grad_norm': '0.4956', 'learning_rate': '0.000474', 'epoch': '0.2796'}
+
  1%|          | 80/7175 [02:02<2:44:56,  1.39s/it]
  1%|          | 81/7175 [02:02<2:18:41,  1.17s/it]
  1%|          | 82/7175 [02:03<2:00:16,  1.02s/it]
  1%|          | 83/7175 [02:04<1:47:21,  1.10it/s]
  1%|          | 84/7175 [02:04<1:38:17,  1.20it/s]
  1%|          | 85/7175 [02:05<1:32:00,  1.28it/s]
  1%|          | 86/7175 [02:06<1:27:36,  1.35it/s]
  1%|          | 87/7175 [02:06<1:24:30,  1.40it/s]
  1%|          | 88/7175 [02:07<1:22:23,  1.43it/s]
  1%|          | 89/7175 [02:08<1:20:55,  1.46it/s]
  1%|▏         | 90/7175 [02:08<1:19:58,  1.48it/s]
                                                   
{'loss': '7.141', 'grad_norm': '0.2999', 'learning_rate': '0.000534', 'epoch': '0.3145'}
+
  1%|▏         | 90/7175 [02:08<1:19:58,  1.48it/s]
  1%|▏         | 91/7175 [02:09<1:19:14,  1.49it/s]
  1%|▏         | 92/7175 [02:10<1:18:43,  1.50it/s]
  1%|▏         | 93/7175 [02:10<1:18:17,  1.51it/s]
  1%|▏         | 94/7175 [02:11<1:18:00,  1.51it/s]
  1%|▏         | 95/7175 [02:12<1:17:48,  1.52it/s]
  1%|▏         | 96/7175 [02:12<1:17:36,  1.52it/s]
  1%|▏         | 97/7175 [02:13<1:17:26,  1.52it/s]
  1%|▏         | 98/7175 [02:13<1:17:32,  1.52it/s]
  1%|▏         | 99/7175 [02:14<1:17:37,  1.52it/s]
  1%|▏         | 100/7175 [02:15<1:17:39,  1.52it/s]
                                                    
{'loss': '6.993', 'grad_norm': '0.8744', 'learning_rate': '0.000594', 'epoch': '0.3495'}
+
  1%|▏         | 100/7175 [02:15<1:17:39,  1.52it/s]
  1%|▏         | 101/7175 [02:15<1:17:37,  1.52it/s]
  1%|▏         | 102/7175 [02:16<1:17:33,  1.52it/s]
  1%|▏         | 103/7175 [02:17<1:17:36,  1.52it/s]
  1%|▏         | 104/7175 [02:17<1:17:37,  1.52it/s]
  1%|▏         | 105/7175 [02:18<1:17:29,  1.52it/s]
  1%|▏         | 106/7175 [02:19<1:17:18,  1.52it/s]
  1%|▏         | 107/7175 [02:19<1:17:21,  1.52it/s]
  2%|▏         | 108/7175 [02:20<1:17:23,  1.52it/s]
  2%|▏         | 109/7175 [02:21<1:17:18,  1.52it/s]
  2%|▏         | 110/7175 [02:21<1:17:11,  1.53it/s]
                                                    
{'loss': '6.847', 'grad_norm': '0.5292', 'learning_rate': '0.0006', 'epoch': '0.3844'}
+
  2%|▏         | 110/7175 [02:21<1:17:11,  1.53it/s]
  2%|▏         | 111/7175 [02:22<1:17:10,  1.53it/s]
  2%|▏         | 112/7175 [02:23<1:17:09,  1.53it/s]
  2%|▏         | 113/7175 [02:23<1:17:06,  1.53it/s]
  2%|▏         | 114/7175 [02:24<1:17:04,  1.53it/s]
  2%|▏         | 115/7175 [02:25<1:17:01,  1.53it/s]
  2%|▏         | 116/7175 [02:25<1:17:04,  1.53it/s]
  2%|▏         | 117/7175 [02:26<1:16:59,  1.53it/s]
  2%|▏         | 118/7175 [02:27<1:16:58,  1.53it/s]
  2%|▏         | 119/7175 [02:27<1:16:57,  1.53it/s]
  2%|▏         | 120/7175 [02:28<1:16:57,  1.53it/s]
                                                    
{'loss': '6.714', 'grad_norm': '1.033', 'learning_rate': '0.0006', 'epoch': '0.4194'}
+
  2%|▏         | 120/7175 [02:28<1:16:57,  1.53it/s]
  2%|▏         | 121/7175 [02:29<1:17:00,  1.53it/s]
  2%|▏         | 122/7175 [02:29<1:16:59,  1.53it/s]
  2%|▏         | 123/7175 [02:30<1:16:57,  1.53it/s]
  2%|▏         | 124/7175 [02:31<1:16:53,  1.53it/s]
  2%|▏         | 125/7175 [02:31<1:16:54,  1.53it/s]
  2%|▏         | 126/7175 [02:32<1:16:53,  1.53it/s]
  2%|▏         | 127/7175 [02:33<1:16:54,  1.53it/s]
  2%|▏         | 128/7175 [02:33<1:16:51,  1.53it/s]
  2%|▏         | 129/7175 [02:34<1:16:50,  1.53it/s]
  2%|▏         | 130/7175 [02:34<1:16:49,  1.53it/s]
                                                    
{'loss': '6.611', 'grad_norm': '1.021', 'learning_rate': '0.0006', 'epoch': '0.4543'}
+
  2%|▏         | 130/7175 [02:34<1:16:49,  1.53it/s]
  2%|▏         | 131/7175 [02:35<1:16:50,  1.53it/s]
  2%|▏         | 132/7175 [02:36<1:16:49,  1.53it/s]
  2%|▏         | 133/7175 [02:36<1:16:49,  1.53it/s]
  2%|▏         | 134/7175 [02:37<1:16:52,  1.53it/s]
  2%|▏         | 135/7175 [02:38<1:16:52,  1.53it/s]
  2%|▏         | 136/7175 [02:38<1:16:49,  1.53it/s]
  2%|▏         | 137/7175 [02:39<1:16:47,  1.53it/s]
  2%|▏         | 138/7175 [02:40<1:16:49,  1.53it/s]
  2%|▏         | 139/7175 [02:40<1:16:48,  1.53it/s]
  2%|▏         | 140/7175 [02:41<1:16:47,  1.53it/s]
                                                    
{'loss': '6.489', 'grad_norm': '0.7077', 'learning_rate': '0.0006', 'epoch': '0.4893'}
+
  2%|▏         | 140/7175 [02:41<1:16:47,  1.53it/s]
  2%|▏         | 141/7175 [02:42<1:16:49,  1.53it/s]
  2%|▏         | 142/7175 [02:42<1:16:46,  1.53it/s]
  2%|▏         | 143/7175 [02:43<1:16:46,  1.53it/s]
  2%|▏         | 144/7175 [02:44<1:16:46,  1.53it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:02, 17.87it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.81it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.42it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.10it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.62it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.64it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.36it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.42it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.23it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.33it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.30it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.29it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.23it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.26it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.26it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.25it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.24it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.24it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.21it/s]
                                                    
+
                                               
{'eval_loss': '6.44', 'eval_runtime': '3.268', 'eval_samples_per_second': '746.9', 'eval_steps_per_second': '11.93', 'epoch': '0.5033'}
+
  2%|▏         | 144/7175 [02:47<1:16:46,  1.53it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.21it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-144
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-144/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-144/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.91it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.91it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-144/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-144/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-144/tokenizer_config.json
+[eval] epoch 0.503 step 144 loss 6.4399 (best 6.4399 @ ep 0.503, 0/20 since best)
+
+
                                               
  2%|▏         | 145/7175 [02:50<4:23:43,  2.25s/it]
  2%|▏         | 146/7175 [02:50<3:27:34,  1.77s/it]
  2%|▏         | 147/7175 [02:51<2:48:12,  1.44s/it]
  2%|▏         | 148/7175 [02:52<2:20:40,  1.20s/it]
  2%|▏         | 149/7175 [02:52<2:01:27,  1.04s/it]
  2%|▏         | 150/7175 [02:53<1:47:56,  1.08it/s]
                                                    
{'loss': '6.392', 'grad_norm': '0.9725', 'learning_rate': '0.0006', 'epoch': '0.5242'}
+
  2%|▏         | 150/7175 [02:53<1:47:56,  1.08it/s]
  2%|▏         | 151/7175 [02:54<1:38:31,  1.19it/s]
  2%|▏         | 152/7175 [02:54<1:31:55,  1.27it/s]
  2%|▏         | 153/7175 [02:55<1:27:17,  1.34it/s]
  2%|▏         | 154/7175 [02:55<1:24:04,  1.39it/s]
  2%|▏         | 155/7175 [02:56<1:21:48,  1.43it/s]
  2%|▏         | 156/7175 [02:57<1:20:17,  1.46it/s]
  2%|▏         | 157/7175 [02:57<1:19:08,  1.48it/s]
  2%|▏         | 158/7175 [02:58<1:18:19,  1.49it/s]
  2%|▏         | 159/7175 [02:59<1:17:46,  1.50it/s]
  2%|▏         | 160/7175 [02:59<1:17:25,  1.51it/s]
                                                    
{'loss': '6.3', 'grad_norm': '0.5226', 'learning_rate': '0.0006', 'epoch': '0.5592'}
+
  2%|▏         | 160/7175 [02:59<1:17:25,  1.51it/s]
  2%|▏         | 161/7175 [03:00<1:17:10,  1.51it/s]
  2%|▏         | 162/7175 [03:01<1:16:56,  1.52it/s]
  2%|▏         | 163/7175 [03:01<1:16:48,  1.52it/s]
  2%|▏         | 164/7175 [03:02<1:16:43,  1.52it/s]
  2%|▏         | 165/7175 [03:03<1:16:36,  1.52it/s]
  2%|▏         | 166/7175 [03:03<1:16:35,  1.53it/s]
  2%|▏         | 167/7175 [03:04<1:16:33,  1.53it/s]
  2%|▏         | 168/7175 [03:05<1:16:31,  1.53it/s]
  2%|▏         | 169/7175 [03:05<1:16:29,  1.53it/s]
  2%|▏         | 170/7175 [03:06<1:16:27,  1.53it/s]
                                                    
{'loss': '6.231', 'grad_norm': '0.5106', 'learning_rate': '0.0006', 'epoch': '0.5941'}
+
  2%|▏         | 170/7175 [03:06<1:16:27,  1.53it/s]
  2%|▏         | 171/7175 [03:07<1:16:31,  1.53it/s]
  2%|▏         | 172/7175 [03:07<1:16:28,  1.53it/s]
  2%|▏         | 173/7175 [03:08<1:16:24,  1.53it/s]
  2%|▏         | 174/7175 [03:09<1:16:20,  1.53it/s]
  2%|▏         | 175/7175 [03:09<1:16:23,  1.53it/s]
  2%|▏         | 176/7175 [03:10<1:16:18,  1.53it/s]
  2%|▏         | 177/7175 [03:11<1:16:17,  1.53it/s]
  2%|▏         | 178/7175 [03:11<1:16:17,  1.53it/s]
  2%|▏         | 179/7175 [03:12<1:16:15,  1.53it/s]
  3%|▎         | 180/7175 [03:13<1:16:13,  1.53it/s]
                                                    
{'loss': '6.133', 'grad_norm': '0.3319', 'learning_rate': '0.0006', 'epoch': '0.6291'}
+
  3%|▎         | 180/7175 [03:13<1:16:13,  1.53it/s]
  3%|▎         | 181/7175 [03:13<1:16:15,  1.53it/s]
  3%|▎         | 182/7175 [03:14<1:16:16,  1.53it/s]
  3%|▎         | 183/7175 [03:14<1:16:15,  1.53it/s]
  3%|▎         | 184/7175 [03:15<1:16:18,  1.53it/s]
  3%|▎         | 185/7175 [03:16<1:16:18,  1.53it/s]
  3%|▎         | 186/7175 [03:16<1:16:14,  1.53it/s]
  3%|▎         | 187/7175 [03:17<1:16:12,  1.53it/s]
  3%|▎         | 188/7175 [03:18<1:16:12,  1.53it/s]
  3%|▎         | 189/7175 [03:18<1:16:12,  1.53it/s]
  3%|▎         | 190/7175 [03:19<1:16:09,  1.53it/s]
                                                    
{'loss': '6.087', 'grad_norm': '0.6218', 'learning_rate': '0.0006', 'epoch': '0.664'}
+
  3%|▎         | 190/7175 [03:19<1:16:09,  1.53it/s]
  3%|▎         | 191/7175 [03:20<1:16:12,  1.53it/s]
  3%|▎         | 192/7175 [03:20<1:16:12,  1.53it/s]
  3%|▎         | 193/7175 [03:21<1:16:15,  1.53it/s]
  3%|▎         | 194/7175 [03:22<1:16:20,  1.52it/s]
  3%|▎         | 195/7175 [03:22<1:16:13,  1.53it/s]
  3%|▎         | 196/7175 [03:23<1:16:11,  1.53it/s]
  3%|▎         | 197/7175 [03:24<1:16:09,  1.53it/s]
  3%|▎         | 198/7175 [03:24<1:16:06,  1.53it/s]
  3%|▎         | 199/7175 [03:25<1:16:03,  1.53it/s]
  3%|▎         | 200/7175 [03:26<1:16:00,  1.53it/s]
                                                    
{'loss': '6.034', 'grad_norm': '0.4769', 'learning_rate': '0.0006', 'epoch': '0.699'}
+
  3%|▎         | 200/7175 [03:26<1:16:00,  1.53it/s]
  3%|▎         | 201/7175 [03:26<1:16:00,  1.53it/s]
  3%|▎         | 202/7175 [03:27<1:15:58,  1.53it/s]
  3%|▎         | 203/7175 [03:28<1:16:09,  1.53it/s]
  3%|▎         | 204/7175 [03:28<1:16:05,  1.53it/s]
  3%|▎         | 205/7175 [03:29<1:16:01,  1.53it/s]
  3%|▎         | 206/7175 [03:30<1:15:58,  1.53it/s]
  3%|▎         | 207/7175 [03:30<1:15:56,  1.53it/s]
  3%|▎         | 208/7175 [03:31<1:15:53,  1.53it/s]
  3%|▎         | 209/7175 [03:31<1:15:52,  1.53it/s]
  3%|▎         | 210/7175 [03:32<1:15:50,  1.53it/s]
                                                    
{'loss': '5.938', 'grad_norm': '0.7541', 'learning_rate': '0.0006', 'epoch': '0.7339'}
+
  3%|▎         | 210/7175 [03:32<1:15:50,  1.53it/s]
  3%|▎         | 211/7175 [03:33<1:15:52,  1.53it/s]
  3%|▎         | 212/7175 [03:33<1:15:51,  1.53it/s]
  3%|▎         | 213/7175 [03:34<1:15:51,  1.53it/s]
  3%|▎         | 214/7175 [03:35<1:15:53,  1.53it/s]
  3%|▎         | 215/7175 [03:35<1:15:51,  1.53it/s]
  3%|▎         | 216/7175 [03:36<1:15:48,  1.53it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.37it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.73it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.58it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.04it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.74it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.56it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.46it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.39it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.33it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.29it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.27it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.25it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.25it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.23it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.24it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.23it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.24it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.24it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s]
                                                    
+
                                               
{'eval_loss': '5.951', 'eval_runtime': '3.263', 'eval_samples_per_second': '748.1', 'eval_steps_per_second': '11.95', 'epoch': '0.7549'}
+
  3%|▎         | 216/7175 [03:39<1:15:48,  1.53it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-216
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-216/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-216/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.94it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.94it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-216/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-216/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-216/tokenizer_config.json
+[eval] epoch 0.755 step 216 loss 5.9514 (best 5.9514 @ ep 0.755, 0/20 since best)
+
+
                                               
  3%|▎         | 217/7175 [03:42<4:19:51,  2.24s/it]
  3%|▎         | 218/7175 [03:43<3:24:38,  1.76s/it]
  3%|▎         | 219/7175 [03:43<2:45:55,  1.43s/it]
  3%|▎         | 220/7175 [03:44<2:18:51,  1.20s/it]
                                                    
{'loss': '5.888', 'grad_norm': '1.281', 'learning_rate': '0.0006', 'epoch': '0.7689'}
+
  3%|▎         | 220/7175 [03:44<2:18:51,  1.20s/it]
  3%|▎         | 221/7175 [03:45<1:59:57,  1.04s/it]
  3%|▎         | 222/7175 [03:45<1:46:44,  1.09it/s]
  3%|▎         | 223/7175 [03:46<1:37:24,  1.19it/s]
  3%|▎         | 224/7175 [03:47<1:30:52,  1.27it/s]
  3%|▎         | 225/7175 [03:47<1:26:21,  1.34it/s]
  3%|▎         | 226/7175 [03:48<1:23:13,  1.39it/s]
  3%|▎         | 227/7175 [03:49<1:20:58,  1.43it/s]
  3%|▎         | 228/7175 [03:49<1:19:21,  1.46it/s]
  3%|▎         | 229/7175 [03:50<1:18:20,  1.48it/s]
  3%|▎         | 230/7175 [03:51<1:17:31,  1.49it/s]
                                                    
{'loss': '5.835', 'grad_norm': '0.4061', 'learning_rate': '0.0006', 'epoch': '0.8038'}
+
  3%|▎         | 230/7175 [03:51<1:17:31,  1.49it/s]
  3%|▎         | 231/7175 [03:51<1:16:58,  1.50it/s]
  3%|▎         | 232/7175 [03:52<1:16:33,  1.51it/s]
  3%|▎         | 233/7175 [03:52<1:16:20,  1.52it/s]
  3%|▎         | 234/7175 [03:53<1:16:06,  1.52it/s]
  3%|▎         | 235/7175 [03:54<1:15:55,  1.52it/s]
  3%|▎         | 236/7175 [03:54<1:15:51,  1.52it/s]
  3%|▎         | 237/7175 [03:55<1:15:46,  1.53it/s]
  3%|▎         | 238/7175 [03:56<1:15:42,  1.53it/s]
  3%|▎         | 239/7175 [03:56<1:15:43,  1.53it/s]
  3%|▎         | 240/7175 [03:57<1:15:39,  1.53it/s]
                                                    
{'loss': '5.753', 'grad_norm': '1.1', 'learning_rate': '0.0006', 'epoch': '0.8388'}
+
  3%|▎         | 240/7175 [03:57<1:15:39,  1.53it/s]
  3%|▎         | 241/7175 [03:58<1:15:37,  1.53it/s]
  3%|▎         | 242/7175 [03:58<1:15:33,  1.53it/s]
  3%|▎         | 243/7175 [03:59<1:15:34,  1.53it/s]
  3%|▎         | 244/7175 [04:00<1:15:32,  1.53it/s]
  3%|▎         | 245/7175 [04:00<1:15:30,  1.53it/s]
  3%|▎         | 246/7175 [04:01<1:15:31,  1.53it/s]
  3%|▎         | 247/7175 [04:02<1:15:29,  1.53it/s]
  3%|▎         | 248/7175 [04:02<1:15:27,  1.53it/s]
  3%|▎         | 249/7175 [04:03<1:15:30,  1.53it/s]
  3%|▎         | 250/7175 [04:04<1:15:27,  1.53it/s]
                                                    
{'loss': '5.726', 'grad_norm': '0.5426', 'learning_rate': '0.0006', 'epoch': '0.8737'}
+
  3%|▎         | 250/7175 [04:04<1:15:27,  1.53it/s]
  3%|▎         | 251/7175 [04:04<1:15:28,  1.53it/s]
  4%|▎         | 252/7175 [04:05<1:15:28,  1.53it/s]
  4%|▎         | 253/7175 [04:06<1:15:26,  1.53it/s]
  4%|▎         | 254/7175 [04:06<1:15:24,  1.53it/s]
  4%|▎         | 255/7175 [04:07<1:15:22,  1.53it/s]
  4%|▎         | 256/7175 [04:08<1:15:24,  1.53it/s]
  4%|▎         | 257/7175 [04:08<1:15:21,  1.53it/s]
  4%|▎         | 258/7175 [04:09<1:30:32,  1.27it/s]
  4%|▎         | 259/7175 [04:10<1:25:55,  1.34it/s]
  4%|▎         | 260/7175 [04:11<1:22:43,  1.39it/s]
                                                    
{'loss': '5.642', 'grad_norm': '0.6426', 'learning_rate': '0.0006', 'epoch': '0.9087'}
+
  4%|▎         | 260/7175 [04:11<1:22:43,  1.39it/s]
  4%|▎         | 261/7175 [04:11<1:20:32,  1.43it/s]
  4%|▎         | 262/7175 [04:12<1:18:56,  1.46it/s]
  4%|▎         | 263/7175 [04:13<1:17:51,  1.48it/s]
  4%|▎         | 264/7175 [04:13<1:17:10,  1.49it/s]
  4%|▎         | 265/7175 [04:14<1:16:37,  1.50it/s]
  4%|▎         | 266/7175 [04:14<1:16:11,  1.51it/s]
  4%|▎         | 267/7175 [04:15<1:15:57,  1.52it/s]
  4%|▎         | 268/7175 [04:16<1:15:44,  1.52it/s]
  4%|▎         | 269/7175 [04:16<1:15:36,  1.52it/s]
  4%|▍         | 270/7175 [04:17<1:15:29,  1.52it/s]
                                                    
{'loss': '5.605', 'grad_norm': '0.5987', 'learning_rate': '0.0006', 'epoch': '0.9436'}
+
  4%|▍         | 270/7175 [04:17<1:15:29,  1.52it/s]
  4%|▍         | 271/7175 [04:18<1:15:27,  1.52it/s]
  4%|▍         | 272/7175 [04:18<1:15:25,  1.53it/s]
  4%|▍         | 273/7175 [04:19<1:15:22,  1.53it/s]
  4%|▍         | 274/7175 [04:20<1:15:17,  1.53it/s]
  4%|▍         | 275/7175 [04:20<1:15:16,  1.53it/s]
  4%|▍         | 276/7175 [04:21<1:15:16,  1.53it/s]
  4%|▍         | 277/7175 [04:22<1:15:12,  1.53it/s]
  4%|▍         | 278/7175 [04:22<1:15:19,  1.53it/s]
  4%|▍         | 279/7175 [04:23<1:15:20,  1.53it/s]
  4%|▍         | 280/7175 [04:24<1:15:20,  1.53it/s]
                                                    
{'loss': '5.541', 'grad_norm': '0.5398', 'learning_rate': '0.0006', 'epoch': '0.9786'}
+
  4%|▍         | 280/7175 [04:24<1:15:20,  1.53it/s]
  4%|▍         | 281/7175 [04:24<1:15:20,  1.53it/s]
  4%|▍         | 282/7175 [04:25<1:15:16,  1.53it/s]
  4%|▍         | 283/7175 [04:26<1:15:18,  1.53it/s]
  4%|▍         | 284/7175 [04:26<1:15:35,  1.52it/s]
  4%|▍         | 285/7175 [04:27<1:15:26,  1.52it/s]
  4%|▍         | 286/7175 [04:28<1:15:20,  1.52it/s]
  4%|▍         | 288/7175 [04:28<1:02:27,  1.84it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.45it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.75it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.58it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.04it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.75it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.57it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.45it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.38it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.33it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.30it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.29it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.26it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.25it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.24it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.24it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.23it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.24it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.23it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s]
                                                    
+
                                               
{'eval_loss': '5.596', 'eval_runtime': '3.246', 'eval_samples_per_second': '752', 'eval_steps_per_second': '12.02', 'epoch': '1.003'}
+
  4%|▍         | 288/7175 [04:32<1:02:27,  1.84it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-288
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-288/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-288/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.91it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.91it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-288/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-288/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-288/tokenizer_config.json
+[eval] epoch 1.003 step 288 loss 5.5964 (best 5.5964 @ ep 1.003, 0/20 since best)
+
+
                                               
  4%|▍         | 289/7175 [04:34<3:37:10,  1.89s/it]
  4%|▍         | 290/7175 [04:35<2:59:56,  1.57s/it]
                                                    
{'loss': '5.485', 'grad_norm': '0.8157', 'learning_rate': '0.0006', 'epoch': '1.01'}
+
  4%|▍         | 290/7175 [04:35<2:59:56,  1.57s/it]
  4%|▍         | 291/7175 [04:36<2:53:23,  1.51s/it]
  4%|▍         | 292/7175 [04:37<2:25:47,  1.27s/it]
  4%|▍         | 293/7175 [04:38<2:05:32,  1.09s/it]
  4%|▍         | 294/7175 [04:38<1:50:55,  1.03it/s]
  4%|▍         | 295/7175 [04:39<1:40:24,  1.14it/s]
  4%|▍         | 296/7175 [04:40<1:32:56,  1.23it/s]
  4%|▍         | 297/7175 [04:41<1:43:01,  1.11it/s]
  4%|▍         | 298/7175 [04:42<1:50:04,  1.04it/s]
  4%|▍         | 299/7175 [04:43<1:54:22,  1.00it/s]
  4%|▍         | 300/7175 [04:44<1:57:12,  1.02s/it]
                                                    
{'loss': '5.479', 'grad_norm': '0.5033', 'learning_rate': '0.0006', 'epoch': '1.045'}
+
  4%|▍         | 300/7175 [04:44<1:57:12,  1.02s/it]
  4%|▍         | 301/7175 [04:45<1:44:33,  1.10it/s]
  4%|▍         | 302/7175 [04:46<1:54:03,  1.00it/s]
  4%|▍         | 303/7175 [04:47<1:42:19,  1.12it/s]
  4%|▍         | 304/7175 [04:47<1:34:05,  1.22it/s]
  4%|▍         | 305/7175 [04:48<1:28:22,  1.30it/s]
  4%|▍         | 306/7175 [04:49<1:24:25,  1.36it/s]
  4%|▍         | 307/7175 [04:49<1:21:32,  1.40it/s]
  4%|▍         | 308/7175 [04:50<1:19:31,  1.44it/s]
  4%|▍         | 309/7175 [04:51<1:32:42,  1.23it/s]
  4%|▍         | 310/7175 [04:52<1:27:21,  1.31it/s]
                                                    
{'loss': '5.362', 'grad_norm': '0.2526', 'learning_rate': '0.0006', 'epoch': '1.08'}
+
  4%|▍         | 310/7175 [04:52<1:27:21,  1.31it/s]
  4%|▍         | 311/7175 [04:52<1:23:37,  1.37it/s]
  4%|▍         | 312/7175 [04:53<1:20:57,  1.41it/s]
  4%|▍         | 313/7175 [04:54<1:19:06,  1.45it/s]
  4%|▍         | 314/7175 [04:54<1:17:52,  1.47it/s]
  4%|▍         | 315/7175 [04:55<1:16:56,  1.49it/s]
  4%|▍         | 316/7175 [04:56<1:16:17,  1.50it/s]
  4%|▍         | 317/7175 [04:56<1:15:51,  1.51it/s]
  4%|▍         | 318/7175 [04:57<1:15:35,  1.51it/s]
  4%|▍         | 319/7175 [04:57<1:15:20,  1.52it/s]
  4%|▍         | 320/7175 [04:58<1:15:08,  1.52it/s]
                                                    
{'loss': '5.326', 'grad_norm': '0.5417', 'learning_rate': '0.0006', 'epoch': '1.115'}
+
  4%|▍         | 320/7175 [04:58<1:15:08,  1.52it/s]
  4%|▍         | 321/7175 [04:59<1:15:03,  1.52it/s]
  4%|▍         | 322/7175 [04:59<1:14:57,  1.52it/s]
  5%|▍         | 323/7175 [05:00<1:14:53,  1.53it/s]
  5%|▍         | 324/7175 [05:01<1:14:50,  1.53it/s]
  5%|▍         | 325/7175 [05:01<1:14:50,  1.53it/s]
  5%|▍         | 326/7175 [05:02<1:14:47,  1.53it/s]
  5%|▍         | 327/7175 [05:03<1:14:46,  1.53it/s]
  5%|▍         | 328/7175 [05:03<1:14:45,  1.53it/s]
  5%|▍         | 329/7175 [05:04<1:14:45,  1.53it/s]
  5%|▍         | 330/7175 [05:05<1:14:42,  1.53it/s]
                                                    
{'loss': '5.29', 'grad_norm': '0.5337', 'learning_rate': '0.0006', 'epoch': '1.15'}
+
  5%|▍         | 330/7175 [05:05<1:14:42,  1.53it/s]
  5%|▍         | 331/7175 [05:05<1:14:43,  1.53it/s]
  5%|▍         | 332/7175 [05:06<1:14:42,  1.53it/s]
  5%|▍         | 333/7175 [05:07<1:14:42,  1.53it/s]
  5%|▍         | 334/7175 [05:07<1:14:39,  1.53it/s]
  5%|▍         | 335/7175 [05:08<1:14:40,  1.53it/s]
  5%|▍         | 336/7175 [05:09<1:14:37,  1.53it/s]
  5%|▍         | 337/7175 [05:09<1:14:39,  1.53it/s]
  5%|▍         | 338/7175 [05:10<1:14:35,  1.53it/s]
  5%|▍         | 339/7175 [05:11<1:14:35,  1.53it/s]
  5%|▍         | 340/7175 [05:11<1:14:33,  1.53it/s]
                                                    
{'loss': '5.234', 'grad_norm': '0.5503', 'learning_rate': '0.0006', 'epoch': '1.185'}
+
  5%|▍         | 340/7175 [05:11<1:14:33,  1.53it/s]
  5%|▍         | 341/7175 [05:12<1:14:38,  1.53it/s]
  5%|▍         | 342/7175 [05:13<1:14:35,  1.53it/s]
  5%|▍         | 343/7175 [05:13<1:14:33,  1.53it/s]
  5%|▍         | 344/7175 [05:14<1:14:33,  1.53it/s]
  5%|▍         | 345/7175 [05:15<1:14:36,  1.53it/s]
  5%|▍         | 346/7175 [05:15<1:14:34,  1.53it/s]
  5%|▍         | 347/7175 [05:16<1:14:33,  1.53it/s]
  5%|▍         | 348/7175 [05:16<1:14:31,  1.53it/s]
  5%|▍         | 349/7175 [05:17<1:14:33,  1.53it/s]
  5%|▍         | 350/7175 [05:18<1:14:30,  1.53it/s]
                                                    
{'loss': '5.189', 'grad_norm': '0.4623', 'learning_rate': '0.0006', 'epoch': '1.22'}
+
  5%|▍         | 350/7175 [05:18<1:14:30,  1.53it/s]
  5%|▍         | 351/7175 [05:18<1:14:35,  1.52it/s]
  5%|▍         | 352/7175 [05:19<1:14:32,  1.53it/s]
  5%|▍         | 353/7175 [05:20<1:14:30,  1.53it/s]
  5%|▍         | 354/7175 [05:20<1:14:31,  1.53it/s]
  5%|▍         | 355/7175 [05:21<1:14:28,  1.53it/s]
  5%|▍         | 356/7175 [05:22<1:14:27,  1.53it/s]
  5%|▍         | 357/7175 [05:22<1:14:26,  1.53it/s]
  5%|▍         | 358/7175 [05:23<1:14:29,  1.53it/s]
  5%|▌         | 359/7175 [05:24<1:14:26,  1.53it/s]
  5%|▌         | 360/7175 [05:24<1:14:24,  1.53it/s]
                                                    
{'loss': '5.135', 'grad_norm': '0.3439', 'learning_rate': '0.0006', 'epoch': '1.255'}
+
  5%|▌         | 360/7175 [05:24<1:14:24,  1.53it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.88it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.87it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.64it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.09it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.76it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.57it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.43it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.37it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.32it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.28it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.27it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.25it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.23it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.23it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.23it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.23it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.23it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.23it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.22it/s]
                                                    
+
                                               
{'eval_loss': '5.223', 'eval_runtime': '3.265', 'eval_samples_per_second': '747.5', 'eval_steps_per_second': '11.94', 'epoch': '1.255'}
+
  5%|▌         | 360/7175 [05:28<1:14:24,  1.53it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.22it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-360
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-360/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-360/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.84it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.84it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-360/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-360/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-360/tokenizer_config.json
+[eval] epoch 1.255 step 360 loss 5.2226 (best 5.2226 @ ep 1.255, 0/20 since best)
+
+
                                               
  5%|▌         | 361/7175 [05:30<4:17:38,  2.27s/it]
  5%|▌         | 362/7175 [05:31<3:22:34,  1.78s/it]
  5%|▌         | 363/7175 [05:32<2:44:03,  1.45s/it]
  5%|▌         | 364/7175 [05:32<2:17:05,  1.21s/it]
  5%|▌         | 365/7175 [05:33<1:58:15,  1.04s/it]
  5%|▌         | 366/7175 [05:34<1:44:59,  1.08it/s]
  5%|▌         | 367/7175 [05:34<1:35:45,  1.19it/s]
  5%|▌         | 368/7175 [05:35<1:29:18,  1.27it/s]
  5%|▌         | 369/7175 [05:36<1:24:46,  1.34it/s]
  5%|▌         | 370/7175 [05:36<1:21:34,  1.39it/s]
                                                    
{'loss': '5.08', 'grad_norm': '0.7792', 'learning_rate': '0.0006', 'epoch': '1.29'}
+
  5%|▌         | 370/7175 [05:36<1:21:34,  1.39it/s]
  5%|▌         | 371/7175 [05:37<1:19:23,  1.43it/s]
  5%|▌         | 372/7175 [05:38<1:17:51,  1.46it/s]
  5%|▌         | 373/7175 [05:38<1:16:50,  1.48it/s]
  5%|▌         | 374/7175 [05:39<1:16:02,  1.49it/s]
  5%|▌         | 375/7175 [05:40<1:15:26,  1.50it/s]
  5%|▌         | 376/7175 [05:40<1:15:04,  1.51it/s]
  5%|▌         | 377/7175 [05:41<1:14:45,  1.52it/s]
  5%|▌         | 378/7175 [05:41<1:14:32,  1.52it/s]
  5%|▌         | 379/7175 [05:42<1:14:22,  1.52it/s]
  5%|▌         | 380/7175 [05:43<1:14:17,  1.52it/s]
                                                    
{'loss': '5.088', 'grad_norm': '0.9799', 'learning_rate': '0.0006', 'epoch': '1.325'}
+
  5%|▌         | 380/7175 [05:43<1:14:17,  1.52it/s]
  5%|▌         | 381/7175 [05:43<1:14:15,  1.52it/s]
  5%|▌         | 382/7175 [05:44<1:14:11,  1.53it/s]
  5%|▌         | 383/7175 [05:45<1:14:11,  1.53it/s]
  5%|▌         | 384/7175 [05:45<1:14:06,  1.53it/s]
  5%|▌         | 385/7175 [05:46<1:14:05,  1.53it/s]
  5%|▌         | 386/7175 [05:47<1:14:06,  1.53it/s]
  5%|▌         | 387/7175 [05:47<1:14:08,  1.53it/s]
  5%|▌         | 388/7175 [05:48<1:14:06,  1.53it/s]
  5%|▌         | 389/7175 [05:49<1:14:06,  1.53it/s]
  5%|▌         | 390/7175 [05:49<1:14:05,  1.53it/s]
                                                    
{'loss': '5.028', 'grad_norm': '0.3464', 'learning_rate': '0.0006', 'epoch': '1.36'}
+
  5%|▌         | 390/7175 [05:49<1:14:05,  1.53it/s]
  5%|▌         | 391/7175 [05:50<1:14:06,  1.53it/s]
  5%|▌         | 392/7175 [05:51<1:14:06,  1.53it/s]
  5%|▌         | 393/7175 [05:51<1:14:03,  1.53it/s]
  5%|▌         | 394/7175 [05:52<1:14:03,  1.53it/s]
  6%|▌         | 395/7175 [05:53<1:14:01,  1.53it/s]
  6%|▌         | 396/7175 [05:53<1:14:02,  1.53it/s]
  6%|▌         | 397/7175 [05:54<1:13:59,  1.53it/s]
  6%|▌         | 398/7175 [05:55<1:13:59,  1.53it/s]
  6%|▌         | 399/7175 [05:55<1:13:57,  1.53it/s]
  6%|▌         | 400/7175 [05:56<1:13:57,  1.53it/s]
                                                    
{'loss': '4.962', 'grad_norm': '0.2372', 'learning_rate': '0.0006', 'epoch': '1.395'}
+
  6%|▌         | 400/7175 [05:56<1:13:57,  1.53it/s]
  6%|▌         | 401/7175 [05:57<1:13:57,  1.53it/s]
  6%|▌         | 402/7175 [05:57<1:13:56,  1.53it/s]
  6%|▌         | 403/7175 [05:58<1:13:55,  1.53it/s]
  6%|▌         | 404/7175 [05:59<1:13:57,  1.53it/s]
  6%|▌         | 405/7175 [05:59<1:13:55,  1.53it/s]
  6%|▌         | 406/7175 [06:00<1:13:54,  1.53it/s]
  6%|▌         | 407/7175 [06:00<1:14:03,  1.52it/s]
  6%|▌         | 408/7175 [06:01<1:13:58,  1.52it/s]
  6%|▌         | 409/7175 [06:02<1:13:58,  1.52it/s]
  6%|▌         | 410/7175 [06:02<1:13:53,  1.53it/s]
                                                    
{'loss': '4.911', 'grad_norm': '0.9554', 'learning_rate': '0.0006', 'epoch': '1.43'}
+
  6%|▌         | 410/7175 [06:02<1:13:53,  1.53it/s]
  6%|▌         | 411/7175 [06:03<1:13:56,  1.52it/s]
  6%|▌         | 412/7175 [06:04<1:13:54,  1.53it/s]
  6%|▌         | 413/7175 [06:04<1:13:54,  1.52it/s]
  6%|▌         | 414/7175 [06:05<1:13:49,  1.53it/s]
  6%|▌         | 415/7175 [06:06<1:13:47,  1.53it/s]
  6%|▌         | 416/7175 [06:06<1:13:43,  1.53it/s]
  6%|▌         | 417/7175 [06:07<1:13:44,  1.53it/s]
  6%|▌         | 418/7175 [06:08<1:13:42,  1.53it/s]
  6%|▌         | 419/7175 [06:08<1:13:41,  1.53it/s]
  6%|▌         | 420/7175 [06:09<1:13:43,  1.53it/s]
                                                    
{'loss': '4.911', 'grad_norm': '0.7166', 'learning_rate': '0.0006', 'epoch': '1.465'}
+
  6%|▌         | 420/7175 [06:09<1:13:43,  1.53it/s]
  6%|▌         | 421/7175 [06:10<1:13:44,  1.53it/s]
  6%|▌         | 422/7175 [06:10<1:13:42,  1.53it/s]
  6%|▌         | 423/7175 [06:11<1:13:42,  1.53it/s]
  6%|▌         | 424/7175 [06:12<1:13:41,  1.53it/s]
  6%|▌         | 425/7175 [06:12<1:13:39,  1.53it/s]
  6%|▌         | 426/7175 [06:13<1:13:39,  1.53it/s]
  6%|▌         | 427/7175 [06:14<1:13:38,  1.53it/s]
  6%|▌         | 428/7175 [06:14<1:13:40,  1.53it/s]
  6%|▌         | 429/7175 [06:15<1:13:38,  1.53it/s]
  6%|▌         | 430/7175 [06:16<1:13:37,  1.53it/s]
                                                    
{'loss': '4.862', 'grad_norm': '0.3406', 'learning_rate': '0.0006', 'epoch': '1.5'}
+
  6%|▌         | 430/7175 [06:16<1:13:37,  1.53it/s]
  6%|▌         | 431/7175 [06:16<1:13:41,  1.53it/s]
  6%|▌         | 432/7175 [06:17<1:13:39,  1.53it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.47it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.75it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.60it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.04it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.76it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.57it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.47it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.37it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.33it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.30it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.28it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.27it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.26it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.25it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.24it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.24it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.23it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.24it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s]
                                                    
+
                                               
{'eval_loss': '4.936', 'eval_runtime': '3.262', 'eval_samples_per_second': '748.4', 'eval_steps_per_second': '11.96', 'epoch': '1.507'}
+
  6%|▌         | 432/7175 [06:20<1:13:39,  1.53it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-432
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-432/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-432/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.69it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.69it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-432/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-432/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-432/tokenizer_config.json
+[eval] epoch 1.507 step 432 loss 4.9360 (best 4.9360 @ ep 1.507, 0/20 since best)
+
+
                                               
  6%|▌         | 433/7175 [06:23<4:15:06,  2.27s/it]
  6%|▌         | 434/7175 [06:24<3:20:35,  1.79s/it]
  6%|▌         | 435/7175 [06:24<2:42:26,  1.45s/it]
  6%|▌         | 436/7175 [06:25<2:15:43,  1.21s/it]
  6%|▌         | 437/7175 [06:26<1:57:03,  1.04s/it]
  6%|▌         | 438/7175 [06:26<1:43:55,  1.08it/s]
  6%|▌         | 439/7175 [06:27<1:34:45,  1.18it/s]
  6%|▌         | 440/7175 [06:27<1:28:24,  1.27it/s]
                                                    
{'loss': '4.784', 'grad_norm': '0.5273', 'learning_rate': '0.0006', 'epoch': '1.535'}
+
  6%|▌         | 440/7175 [06:27<1:28:24,  1.27it/s]
  6%|▌         | 441/7175 [06:28<1:23:56,  1.34it/s]
  6%|▌         | 442/7175 [06:29<1:20:46,  1.39it/s]
  6%|▌         | 443/7175 [06:29<1:18:34,  1.43it/s]
  6%|▌         | 444/7175 [06:30<1:17:03,  1.46it/s]
  6%|▌         | 445/7175 [06:31<1:16:00,  1.48it/s]
  6%|▌         | 446/7175 [06:31<1:15:14,  1.49it/s]
  6%|▌         | 447/7175 [06:32<1:14:41,  1.50it/s]
  6%|▌         | 448/7175 [06:33<1:14:17,  1.51it/s]
  6%|▋         | 449/7175 [06:33<1:14:00,  1.51it/s]
  6%|▋         | 450/7175 [06:34<1:13:48,  1.52it/s]
                                                    
{'loss': '4.759', 'grad_norm': '0.4367', 'learning_rate': '0.0006', 'epoch': '1.57'}
+
  6%|▋         | 450/7175 [06:34<1:13:48,  1.52it/s]
  6%|▋         | 451/7175 [06:35<1:13:41,  1.52it/s]
  6%|▋         | 452/7175 [06:35<1:13:35,  1.52it/s]
  6%|▋         | 453/7175 [06:36<1:13:31,  1.52it/s]
  6%|▋         | 454/7175 [06:37<1:13:26,  1.53it/s]
  6%|▋         | 455/7175 [06:37<1:13:25,  1.53it/s]
  6%|▋         | 456/7175 [06:38<1:13:23,  1.53it/s]
  6%|▋         | 457/7175 [06:39<1:13:21,  1.53it/s]
  6%|▋         | 458/7175 [06:39<1:13:20,  1.53it/s]
  6%|▋         | 459/7175 [06:40<1:13:18,  1.53it/s]
  6%|▋         | 460/7175 [06:41<1:13:18,  1.53it/s]
                                                    
{'loss': '4.69', 'grad_norm': '0.4377', 'learning_rate': '0.0006', 'epoch': '1.605'}
+
  6%|▋         | 460/7175 [06:41<1:13:18,  1.53it/s]
  6%|▋         | 461/7175 [06:41<1:13:19,  1.53it/s]
  6%|▋         | 462/7175 [06:42<1:13:17,  1.53it/s]
  6%|▋         | 463/7175 [06:43<1:13:15,  1.53it/s]
  6%|▋         | 464/7175 [06:43<1:13:16,  1.53it/s]
  6%|▋         | 465/7175 [06:44<1:13:16,  1.53it/s]
  6%|▋         | 466/7175 [06:45<1:13:14,  1.53it/s]
  7%|▋         | 467/7175 [06:45<1:13:13,  1.53it/s]
  7%|▋         | 468/7175 [06:46<1:13:13,  1.53it/s]
  7%|▋         | 469/7175 [06:46<1:13:13,  1.53it/s]
  7%|▋         | 470/7175 [06:47<1:13:10,  1.53it/s]
                                                    
{'loss': '4.666', 'grad_norm': '0.4313', 'learning_rate': '0.0006', 'epoch': '1.64'}
+
  7%|▋         | 470/7175 [06:47<1:13:10,  1.53it/s]
  7%|▋         | 471/7175 [06:48<1:13:11,  1.53it/s]
  7%|▋         | 472/7175 [06:48<1:13:11,  1.53it/s]
  7%|▋         | 473/7175 [06:49<1:13:13,  1.53it/s]
  7%|▋         | 474/7175 [06:50<1:13:09,  1.53it/s]
  7%|▋         | 475/7175 [06:50<1:13:09,  1.53it/s]
  7%|▋         | 476/7175 [06:51<1:13:06,  1.53it/s]
  7%|▋         | 477/7175 [06:52<1:13:05,  1.53it/s]
  7%|▋         | 478/7175 [06:52<1:13:04,  1.53it/s]
  7%|▋         | 479/7175 [06:53<1:13:03,  1.53it/s]
  7%|▋         | 480/7175 [06:54<1:13:04,  1.53it/s]
                                                    
{'loss': '4.625', 'grad_norm': '0.5737', 'learning_rate': '0.0006', 'epoch': '1.675'}
+
  7%|▋         | 480/7175 [06:54<1:13:04,  1.53it/s]
  7%|▋         | 481/7175 [06:54<1:13:05,  1.53it/s]
  7%|▋         | 482/7175 [06:55<1:13:03,  1.53it/s]
  7%|▋         | 483/7175 [06:56<1:13:01,  1.53it/s]
  7%|▋         | 484/7175 [06:56<1:13:03,  1.53it/s]
  7%|▋         | 485/7175 [06:57<1:13:00,  1.53it/s]
  7%|▋         | 486/7175 [06:58<1:13:00,  1.53it/s]
  7%|▋         | 487/7175 [06:58<1:12:58,  1.53it/s]
  7%|▋         | 488/7175 [06:59<1:13:02,  1.53it/s]
  7%|▋         | 489/7175 [07:00<1:12:59,  1.53it/s]
  7%|▋         | 490/7175 [07:00<1:12:59,  1.53it/s]
                                                    
{'loss': '4.603', 'grad_norm': '0.5191', 'learning_rate': '0.0006', 'epoch': '1.709'}
+
  7%|▋         | 490/7175 [07:00<1:12:59,  1.53it/s]
  7%|▋         | 491/7175 [07:01<1:12:59,  1.53it/s]
  7%|▋         | 492/7175 [07:02<1:12:57,  1.53it/s]
  7%|▋         | 493/7175 [07:02<1:12:56,  1.53it/s]
  7%|▋         | 494/7175 [07:03<1:12:56,  1.53it/s]
  7%|▋         | 495/7175 [07:04<1:12:54,  1.53it/s]
  7%|▋         | 496/7175 [07:04<1:12:54,  1.53it/s]
  7%|▋         | 497/7175 [07:05<1:12:52,  1.53it/s]
  7%|▋         | 498/7175 [07:05<1:12:52,  1.53it/s]
  7%|▋         | 499/7175 [07:06<1:12:49,  1.53it/s]
  7%|▋         | 500/7175 [07:07<1:12:54,  1.53it/s]
                                                    
{'loss': '4.553', 'grad_norm': '0.2884', 'learning_rate': '0.0006', 'epoch': '1.744'}
+
  7%|▋         | 500/7175 [07:07<1:12:54,  1.53it/s]
  7%|▋         | 501/7175 [07:07<1:12:52,  1.53it/s]
  7%|▋         | 502/7175 [07:08<1:12:51,  1.53it/s]
  7%|▋         | 503/7175 [07:09<1:12:48,  1.53it/s]
  7%|▋         | 504/7175 [07:09<1:12:50,  1.53it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.45it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.73it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.61it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.05it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.76it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.57it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.47it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.38it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.34it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.29it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.28it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.25it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.25it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.24it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.23it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.23it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.23it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.23it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s]
                                                    
+
                                               
{'eval_loss': '4.659', 'eval_runtime': '3.262', 'eval_samples_per_second': '748.3', 'eval_steps_per_second': '11.96', 'epoch': '1.758'}
+
  7%|▋         | 504/7175 [07:13<1:12:50,  1.53it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-504
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-504/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-504/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.83it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.83it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-504/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-504/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-504/tokenizer_config.json
+[eval] epoch 1.758 step 504 loss 4.6592 (best 4.6592 @ ep 1.758, 0/20 since best)
+
+
                                               
  7%|▋         | 505/7175 [07:15<4:11:57,  2.27s/it]
  7%|▋         | 506/7175 [07:16<3:18:08,  1.78s/it]
  7%|▋         | 507/7175 [07:17<2:40:25,  1.44s/it]
  7%|▋         | 508/7175 [07:17<2:14:04,  1.21s/it]
  7%|▋         | 509/7175 [07:18<1:55:39,  1.04s/it]
  7%|▋         | 510/7175 [07:19<1:42:46,  1.08it/s]
                                                    
{'loss': '4.5', 'grad_norm': '0.3278', 'learning_rate': '0.0006', 'epoch': '1.779'}
+
  7%|▋         | 510/7175 [07:19<1:42:46,  1.08it/s]
  7%|▋         | 511/7175 [07:19<1:33:45,  1.18it/s]
  7%|▋         | 512/7175 [07:20<1:27:24,  1.27it/s]
  7%|▋         | 513/7175 [07:21<1:23:00,  1.34it/s]
  7%|▋         | 514/7175 [07:21<1:19:53,  1.39it/s]
  7%|▋         | 515/7175 [07:22<1:17:43,  1.43it/s]
  7%|▋         | 516/7175 [07:23<1:16:11,  1.46it/s]
  7%|▋         | 517/7175 [07:23<1:15:11,  1.48it/s]
  7%|▋         | 518/7175 [07:24<1:14:24,  1.49it/s]
  7%|▋         | 519/7175 [07:25<1:13:53,  1.50it/s]
  7%|▋         | 520/7175 [07:25<1:13:30,  1.51it/s]
                                                    
{'loss': '4.465', 'grad_norm': '0.3519', 'learning_rate': '0.0006', 'epoch': '1.814'}
+
  7%|▋         | 520/7175 [07:25<1:13:30,  1.51it/s]
  7%|▋         | 521/7175 [07:26<1:13:16,  1.51it/s]
  7%|▋         | 522/7175 [07:27<1:13:01,  1.52it/s]
  7%|▋         | 523/7175 [07:27<1:12:53,  1.52it/s]
  7%|▋         | 524/7175 [07:28<1:12:47,  1.52it/s]
  7%|▋         | 525/7175 [07:29<1:12:44,  1.52it/s]
  7%|▋         | 526/7175 [07:29<1:12:40,  1.52it/s]
  7%|▋         | 527/7175 [07:30<1:12:39,  1.52it/s]
  7%|▋         | 528/7175 [07:30<1:12:36,  1.53it/s]
  7%|▋         | 529/7175 [07:31<1:12:36,  1.53it/s]
  7%|▋         | 530/7175 [07:32<1:12:35,  1.53it/s]
                                                    
{'loss': '4.433', 'grad_norm': '0.6919', 'learning_rate': '0.0006', 'epoch': '1.849'}
+
  7%|▋         | 530/7175 [07:32<1:12:35,  1.53it/s]
  7%|▋         | 531/7175 [07:32<1:12:35,  1.53it/s]
  7%|▋         | 532/7175 [07:33<1:12:33,  1.53it/s]
  7%|▋         | 533/7175 [07:34<1:12:32,  1.53it/s]
  7%|▋         | 534/7175 [07:34<1:12:34,  1.52it/s]
  7%|▋         | 535/7175 [07:35<1:12:30,  1.53it/s]
  7%|▋         | 536/7175 [07:36<1:12:27,  1.53it/s]
  7%|▋         | 537/7175 [07:36<1:12:26,  1.53it/s]
  7%|▋         | 538/7175 [07:37<1:12:27,  1.53it/s]
  8%|▊         | 539/7175 [07:38<1:12:24,  1.53it/s]
  8%|▊         | 540/7175 [07:38<1:12:24,  1.53it/s]
                                                    
{'loss': '4.427', 'grad_norm': '0.3892', 'learning_rate': '0.0006', 'epoch': '1.884'}
+
  8%|▊         | 540/7175 [07:38<1:12:24,  1.53it/s]
  8%|▊         | 541/7175 [07:39<1:12:25,  1.53it/s]
  8%|▊         | 542/7175 [07:40<1:12:24,  1.53it/s]
  8%|▊         | 543/7175 [07:40<1:12:22,  1.53it/s]
  8%|▊         | 544/7175 [07:41<1:12:21,  1.53it/s]
  8%|▊         | 545/7175 [07:42<1:12:22,  1.53it/s]
  8%|▊         | 546/7175 [07:42<1:12:20,  1.53it/s]
  8%|▊         | 547/7175 [07:43<1:12:20,  1.53it/s]
  8%|▊         | 548/7175 [07:44<1:12:20,  1.53it/s]
  8%|▊         | 549/7175 [07:44<1:12:20,  1.53it/s]
  8%|▊         | 550/7175 [07:45<1:12:20,  1.53it/s]
                                                    
{'loss': '4.387', 'grad_norm': '0.3946', 'learning_rate': '0.0006', 'epoch': '1.919'}
+
  8%|▊         | 550/7175 [07:45<1:12:20,  1.53it/s]
  8%|▊         | 551/7175 [07:46<1:12:19,  1.53it/s]
  8%|▊         | 552/7175 [07:46<1:12:17,  1.53it/s]
  8%|▊         | 553/7175 [07:47<1:12:17,  1.53it/s]
  8%|▊         | 554/7175 [07:48<1:12:15,  1.53it/s]
  8%|▊         | 555/7175 [07:48<1:12:14,  1.53it/s]
  8%|▊         | 556/7175 [07:49<1:12:13,  1.53it/s]
  8%|▊         | 557/7175 [07:49<1:12:15,  1.53it/s]
  8%|▊         | 558/7175 [07:50<1:12:13,  1.53it/s]
  8%|▊         | 559/7175 [07:51<1:12:26,  1.52it/s]
  8%|▊         | 560/7175 [07:51<1:12:22,  1.52it/s]
                                                    
{'loss': '4.345', 'grad_norm': '0.372', 'learning_rate': '0.0006', 'epoch': '1.954'}
+
  8%|▊         | 560/7175 [07:51<1:12:22,  1.52it/s]
  8%|▊         | 561/7175 [07:52<1:12:21,  1.52it/s]
  8%|▊         | 562/7175 [07:53<1:12:20,  1.52it/s]
  8%|▊         | 563/7175 [07:53<1:12:18,  1.52it/s]
  8%|▊         | 564/7175 [07:54<1:12:14,  1.53it/s]
  8%|▊         | 565/7175 [07:55<1:12:16,  1.52it/s]
  8%|▊         | 566/7175 [07:55<1:12:14,  1.52it/s]
  8%|▊         | 567/7175 [07:56<1:12:11,  1.53it/s]
  8%|▊         | 568/7175 [07:57<1:12:14,  1.52it/s]
  8%|▊         | 569/7175 [07:57<1:12:11,  1.53it/s]
  8%|▊         | 570/7175 [07:58<1:12:11,  1.52it/s]
                                                    
{'loss': '4.322', 'grad_norm': '0.3394', 'learning_rate': '0.0006', 'epoch': '1.989'}
+
  8%|▊         | 570/7175 [07:58<1:12:11,  1.52it/s]
  8%|▊         | 571/7175 [07:59<1:12:34,  1.52it/s]
  8%|▊         | 572/7175 [07:59<1:12:23,  1.52it/s]
  8%|▊         | 573/7175 [08:00<1:12:14,  1.52it/s]
  8%|▊         | 575/7175 [08:01<59:57,  1.83it/s]  
  8%|▊         | 576/7175 [08:01<1:02:57,  1.75it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.32it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.69it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.52it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.00it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.71it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.54it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.44it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.36it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.31it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.29it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.26it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.24it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.23it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.23it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.23it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.22it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.21it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.21it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.21it/s]
                                                    
+
                                               
{'eval_loss': '4.469', 'eval_runtime': '3.253', 'eval_samples_per_second': '750.4', 'eval_steps_per_second': '11.99', 'epoch': '2.007'}
+
  8%|▊         | 576/7175 [08:05<1:02:57,  1.75it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.21it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-576
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-576/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-576/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.93it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.93it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-576/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-576/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-576/tokenizer_config.json
+[eval] epoch 2.007 step 576 loss 4.4687 (best 4.4687 @ ep 2.007, 0/20 since best)
+
+
                                               
  8%|▊         | 577/7175 [08:07<3:37:21,  1.98s/it]
  8%|▊         | 578/7175 [08:08<2:57:45,  1.62s/it]
  8%|▊         | 579/7175 [08:09<2:28:08,  1.35s/it]
  8%|▊         | 580/7175 [08:09<2:06:20,  1.15s/it]
                                                    
{'loss': '4.296', 'grad_norm': '0.4932', 'learning_rate': '0.0006', 'epoch': '2.021'}
+
  8%|▊         | 580/7175 [08:09<2:06:20,  1.15s/it]
  8%|▊         | 581/7175 [08:10<1:50:38,  1.01s/it]
  8%|▊         | 582/7175 [08:11<1:39:18,  1.11it/s]
  8%|▊         | 583/7175 [08:11<1:31:14,  1.20it/s]
  8%|▊         | 584/7175 [08:12<1:25:30,  1.28it/s]
  8%|▊         | 585/7175 [08:13<1:21:27,  1.35it/s]
  8%|▊         | 586/7175 [08:13<1:18:37,  1.40it/s]
  8%|▊         | 587/7175 [08:14<1:16:35,  1.43it/s]
  8%|▊         | 588/7175 [08:15<1:15:10,  1.46it/s]
  8%|▊         | 589/7175 [08:15<1:14:09,  1.48it/s]
  8%|▊         | 590/7175 [08:16<1:13:26,  1.49it/s]
                                                    
{'loss': '4.26', 'grad_norm': '0.408', 'learning_rate': '0.0006', 'epoch': '2.056'}
+
  8%|▊         | 590/7175 [08:16<1:13:26,  1.49it/s]
  8%|▊         | 591/7175 [08:17<1:13:01,  1.50it/s]
  8%|▊         | 592/7175 [08:17<1:12:38,  1.51it/s]
  8%|▊         | 593/7175 [08:18<1:12:22,  1.52it/s]
  8%|▊         | 594/7175 [08:19<1:12:11,  1.52it/s]
  8%|▊         | 595/7175 [08:19<1:12:06,  1.52it/s]
  8%|▊         | 596/7175 [08:20<1:12:00,  1.52it/s]
  8%|▊         | 597/7175 [08:21<1:11:54,  1.52it/s]
  8%|▊         | 598/7175 [08:21<1:11:49,  1.53it/s]
  8%|▊         | 599/7175 [08:22<1:11:51,  1.53it/s]
  8%|▊         | 600/7175 [08:22<1:11:46,  1.53it/s]
                                                    
{'loss': '4.231', 'grad_norm': '0.2512', 'learning_rate': '0.0006', 'epoch': '2.091'}
+
  8%|▊         | 600/7175 [08:22<1:11:46,  1.53it/s]
  8%|▊         | 601/7175 [08:23<1:11:45,  1.53it/s]
  8%|▊         | 602/7175 [08:24<1:11:43,  1.53it/s]
  8%|▊         | 603/7175 [08:24<1:11:43,  1.53it/s]
  8%|▊         | 604/7175 [08:25<1:11:42,  1.53it/s]
  8%|▊         | 605/7175 [08:26<1:11:40,  1.53it/s]
  8%|▊         | 606/7175 [08:26<1:11:42,  1.53it/s]
  8%|▊         | 607/7175 [08:27<1:11:45,  1.53it/s]
  8%|▊         | 608/7175 [08:28<1:11:41,  1.53it/s]
  8%|▊         | 609/7175 [08:28<1:11:39,  1.53it/s]
  9%|▊         | 610/7175 [08:29<1:11:41,  1.53it/s]
                                                    
{'loss': '4.215', 'grad_norm': '0.4514', 'learning_rate': '0.0006', 'epoch': '2.126'}
+
  9%|▊         | 610/7175 [08:29<1:11:41,  1.53it/s]
  9%|▊         | 611/7175 [08:30<1:11:43,  1.53it/s]
  9%|▊         | 612/7175 [08:30<1:11:39,  1.53it/s]
  9%|▊         | 613/7175 [08:31<1:11:36,  1.53it/s]
  9%|▊         | 614/7175 [08:32<1:11:35,  1.53it/s]
  9%|▊         | 615/7175 [08:32<1:11:36,  1.53it/s]
  9%|▊         | 616/7175 [08:33<1:11:35,  1.53it/s]
  9%|▊         | 617/7175 [08:34<1:11:34,  1.53it/s]
  9%|▊         | 618/7175 [08:34<1:11:34,  1.53it/s]
  9%|▊         | 619/7175 [08:35<1:11:33,  1.53it/s]
  9%|▊         | 620/7175 [08:36<1:11:31,  1.53it/s]
                                                    
{'loss': '4.202', 'grad_norm': '0.4264', 'learning_rate': '0.0006', 'epoch': '2.161'}
+
  9%|▊         | 620/7175 [08:36<1:11:31,  1.53it/s]
  9%|▊         | 621/7175 [08:36<1:11:31,  1.53it/s]
  9%|▊         | 622/7175 [08:37<1:11:31,  1.53it/s]
  9%|▊         | 623/7175 [08:38<1:11:30,  1.53it/s]
  9%|▊         | 624/7175 [08:38<1:11:28,  1.53it/s]
  9%|▊         | 625/7175 [08:39<1:11:27,  1.53it/s]
  9%|▊         | 626/7175 [08:40<1:11:31,  1.53it/s]
  9%|▊         | 627/7175 [08:40<1:11:28,  1.53it/s]
  9%|▉         | 628/7175 [08:41<1:11:25,  1.53it/s]
  9%|▉         | 629/7175 [08:41<1:11:23,  1.53it/s]
  9%|▉         | 630/7175 [08:42<1:11:26,  1.53it/s]
                                                    
{'loss': '4.17', 'grad_norm': '0.3764', 'learning_rate': '0.0006', 'epoch': '2.196'}
+
  9%|▉         | 630/7175 [08:42<1:11:26,  1.53it/s]
  9%|▉         | 631/7175 [08:43<1:11:25,  1.53it/s]
  9%|▉         | 632/7175 [08:43<1:11:23,  1.53it/s]
  9%|▉         | 633/7175 [08:44<1:11:24,  1.53it/s]
  9%|▉         | 634/7175 [08:45<1:11:26,  1.53it/s]
  9%|▉         | 635/7175 [08:45<1:11:22,  1.53it/s]
  9%|▉         | 636/7175 [08:46<1:11:19,  1.53it/s]
  9%|▉         | 637/7175 [08:47<1:11:18,  1.53it/s]
  9%|▉         | 638/7175 [08:47<1:11:18,  1.53it/s]
  9%|▉         | 639/7175 [08:48<1:11:16,  1.53it/s]
  9%|▉         | 640/7175 [08:49<1:11:16,  1.53it/s]
                                                    
{'loss': '4.151', 'grad_norm': '0.2427', 'learning_rate': '0.0006', 'epoch': '2.231'}
+
  9%|▉         | 640/7175 [08:49<1:11:16,  1.53it/s]
  9%|▉         | 641/7175 [08:49<1:11:19,  1.53it/s]
  9%|▉         | 642/7175 [08:50<1:11:18,  1.53it/s]
  9%|▉         | 643/7175 [08:51<1:11:23,  1.53it/s]
  9%|▉         | 644/7175 [08:51<1:11:17,  1.53it/s]
  9%|▉         | 645/7175 [08:52<1:11:16,  1.53it/s]
  9%|▉         | 646/7175 [08:53<1:11:16,  1.53it/s]
  9%|▉         | 647/7175 [08:53<1:11:13,  1.53it/s]
  9%|▉         | 648/7175 [08:54<1:11:12,  1.53it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.50it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.78it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.60it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.05it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.73it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.56it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.42it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.38it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.30it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.29it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.28it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.26it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.25it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.24it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.24it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.24it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.23it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.22it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.21it/s]
                                                    
+
                                               
{'eval_loss': '4.306', 'eval_runtime': '3.247', 'eval_samples_per_second': '751.7', 'eval_steps_per_second': '12.01', 'epoch': '2.259'}
+
  9%|▉         | 648/7175 [08:57<1:11:12,  1.53it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.21it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-648
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-648/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-648/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.84it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.83it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-648/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-648/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-648/tokenizer_config.json
+[eval] epoch 2.259 step 648 loss 4.3056 (best 4.3056 @ ep 2.259, 0/20 since best)
+
+
                                               
  9%|▉         | 649/7175 [09:00<4:05:08,  2.25s/it]
  9%|▉         | 650/7175 [09:01<3:12:55,  1.77s/it]
                                                    
{'loss': '4.132', 'grad_norm': '0.281', 'learning_rate': '0.0006', 'epoch': '2.266'}
+
  9%|▉         | 650/7175 [09:01<3:12:55,  1.77s/it]
  9%|▉         | 651/7175 [09:01<2:36:24,  1.44s/it]
  9%|▉         | 652/7175 [09:02<2:10:47,  1.20s/it]
  9%|▉         | 653/7175 [09:03<1:52:54,  1.04s/it]
  9%|▉         | 654/7175 [09:03<1:40:22,  1.08it/s]
  9%|▉         | 655/7175 [09:04<1:31:35,  1.19it/s]
  9%|▉         | 656/7175 [09:04<1:25:25,  1.27it/s]
  9%|▉         | 657/7175 [09:05<1:21:05,  1.34it/s]
  9%|▉         | 658/7175 [09:06<1:18:05,  1.39it/s]
  9%|▉         | 659/7175 [09:06<1:15:57,  1.43it/s]
  9%|▉         | 660/7175 [09:07<1:14:28,  1.46it/s]
                                                    
{'loss': '4.094', 'grad_norm': '0.3202', 'learning_rate': '0.0006', 'epoch': '2.301'}
+
  9%|▉         | 660/7175 [09:07<1:14:28,  1.46it/s]
  9%|▉         | 661/7175 [09:08<1:13:27,  1.48it/s]
  9%|▉         | 662/7175 [09:08<1:12:46,  1.49it/s]
  9%|▉         | 663/7175 [09:09<1:12:13,  1.50it/s]
  9%|▉         | 664/7175 [09:10<1:11:50,  1.51it/s]
  9%|▉         | 665/7175 [09:10<1:11:33,  1.52it/s]
  9%|▉         | 666/7175 [09:11<1:11:23,  1.52it/s]
  9%|▉         | 667/7175 [09:12<1:11:13,  1.52it/s]
  9%|▉         | 668/7175 [09:12<1:11:10,  1.52it/s]
  9%|▉         | 669/7175 [09:13<1:11:08,  1.52it/s]
  9%|▉         | 670/7175 [09:14<1:11:02,  1.53it/s]
                                                    
{'loss': '4.086', 'grad_norm': '0.4085', 'learning_rate': '0.0006', 'epoch': '2.336'}
+
  9%|▉         | 670/7175 [09:14<1:11:02,  1.53it/s]
  9%|▉         | 671/7175 [09:14<1:11:02,  1.53it/s]
  9%|▉         | 672/7175 [09:15<1:10:59,  1.53it/s]
  9%|▉         | 673/7175 [09:16<1:11:00,  1.53it/s]
  9%|▉         | 674/7175 [09:16<1:10:57,  1.53it/s]
  9%|▉         | 675/7175 [09:17<1:10:56,  1.53it/s]
  9%|▉         | 676/7175 [09:18<1:10:52,  1.53it/s]
  9%|▉         | 677/7175 [09:18<1:10:53,  1.53it/s]
  9%|▉         | 678/7175 [09:19<1:10:52,  1.53it/s]
  9%|▉         | 679/7175 [09:20<1:10:51,  1.53it/s]
  9%|▉         | 680/7175 [09:20<1:10:52,  1.53it/s]
                                                    
{'loss': '4.077', 'grad_norm': '0.3778', 'learning_rate': '0.0006', 'epoch': '2.37'}
+
  9%|▉         | 680/7175 [09:20<1:10:52,  1.53it/s]
  9%|▉         | 681/7175 [09:21<1:10:51,  1.53it/s]
 10%|▉         | 682/7175 [09:21<1:10:51,  1.53it/s]
 10%|▉         | 683/7175 [09:22<1:10:49,  1.53it/s]
 10%|▉         | 684/7175 [09:23<1:10:51,  1.53it/s]
 10%|▉         | 685/7175 [09:23<1:10:49,  1.53it/s]
 10%|▉         | 686/7175 [09:24<1:10:47,  1.53it/s]
 10%|▉         | 687/7175 [09:25<1:10:46,  1.53it/s]
 10%|▉         | 688/7175 [09:25<1:10:51,  1.53it/s]
 10%|▉         | 689/7175 [09:26<1:10:48,  1.53it/s]
 10%|▉         | 690/7175 [09:27<1:10:45,  1.53it/s]
                                                    
{'loss': '4.086', 'grad_norm': '0.363', 'learning_rate': '0.0006', 'epoch': '2.405'}
+
 10%|▉         | 690/7175 [09:27<1:10:45,  1.53it/s]
 10%|▉         | 691/7175 [09:27<1:10:45,  1.53it/s]
 10%|▉         | 692/7175 [09:28<1:10:48,  1.53it/s]
 10%|▉         | 693/7175 [09:29<1:10:45,  1.53it/s]
 10%|▉         | 694/7175 [09:29<1:10:42,  1.53it/s]
 10%|▉         | 695/7175 [09:30<1:10:43,  1.53it/s]
 10%|▉         | 696/7175 [09:31<1:10:45,  1.53it/s]
 10%|▉         | 697/7175 [09:31<1:10:42,  1.53it/s]
 10%|▉         | 698/7175 [09:32<1:10:42,  1.53it/s]
 10%|▉         | 699/7175 [09:33<1:10:41,  1.53it/s]
 10%|▉         | 700/7175 [09:33<1:10:43,  1.53it/s]
                                                    
{'loss': '4.038', 'grad_norm': '0.2659', 'learning_rate': '0.0006', 'epoch': '2.44'}
+
 10%|▉         | 700/7175 [09:33<1:10:43,  1.53it/s]
 10%|▉         | 701/7175 [09:34<1:10:44,  1.53it/s]
 10%|▉         | 702/7175 [09:35<1:10:41,  1.53it/s]
 10%|▉         | 703/7175 [09:35<1:10:39,  1.53it/s]
 10%|▉         | 704/7175 [09:36<1:10:39,  1.53it/s]
 10%|▉         | 705/7175 [09:37<1:10:37,  1.53it/s]
 10%|▉         | 706/7175 [09:37<1:10:36,  1.53it/s]
 10%|▉         | 707/7175 [09:38<1:10:35,  1.53it/s]
 10%|▉         | 708/7175 [09:39<1:10:37,  1.53it/s]
 10%|▉         | 709/7175 [09:39<1:10:34,  1.53it/s]
 10%|▉         | 710/7175 [09:40<1:10:34,  1.53it/s]
                                                    
{'loss': '4.028', 'grad_norm': '0.3372', 'learning_rate': '0.0006', 'epoch': '2.475'}
+
 10%|▉         | 710/7175 [09:40<1:10:34,  1.53it/s]
 10%|▉         | 711/7175 [09:40<1:10:35,  1.53it/s]
 10%|▉         | 712/7175 [09:41<1:10:34,  1.53it/s]
 10%|▉         | 713/7175 [09:42<1:10:34,  1.53it/s]
 10%|▉         | 714/7175 [09:42<1:10:31,  1.53it/s]
 10%|▉         | 715/7175 [09:43<1:10:30,  1.53it/s]
 10%|▉         | 716/7175 [09:44<1:10:30,  1.53it/s]
 10%|▉         | 717/7175 [09:44<1:10:29,  1.53it/s]
 10%|█         | 718/7175 [09:45<1:10:28,  1.53it/s]
 10%|█         | 719/7175 [09:46<1:10:27,  1.53it/s]
 10%|█         | 720/7175 [09:46<1:10:28,  1.53it/s]
                                                    
{'loss': '4.041', 'grad_norm': '0.3702', 'learning_rate': '0.0006', 'epoch': '2.51'}
+
 10%|█         | 720/7175 [09:46<1:10:28,  1.53it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.63it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.82it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.60it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.06it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.76it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.57it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.44it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.36it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.32it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.29it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.28it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.26it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.26it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.24it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.24it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.25it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.24it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.24it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s]
                                                    
+
                                               
{'eval_loss': '4.206', 'eval_runtime': '3.249', 'eval_samples_per_second': '751.3', 'eval_steps_per_second': '12', 'epoch': '2.51'}
+
 10%|█         | 720/7175 [09:50<1:10:28,  1.53it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-720
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-720/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-720/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  2.14it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  2.14it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-720/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-720/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-720/tokenizer_config.json
+[eval] epoch 2.51 step 720 loss 4.2059 (best 4.2059 @ ep 2.51, 0/20 since best)
+
+
                                               
 10%|█         | 721/7175 [09:52<3:59:31,  2.23s/it]
 10%|█         | 722/7175 [09:53<3:08:44,  1.75s/it]
 10%|█         | 723/7175 [09:54<2:33:10,  1.42s/it]
 10%|█         | 724/7175 [09:54<2:08:17,  1.19s/it]
 10%|█         | 725/7175 [09:55<1:50:54,  1.03s/it]
 10%|█         | 726/7175 [09:56<1:38:43,  1.09it/s]
 10%|█         | 727/7175 [09:56<1:30:13,  1.19it/s]
 10%|█         | 728/7175 [09:57<1:24:14,  1.28it/s]
 10%|█         | 729/7175 [09:58<1:20:04,  1.34it/s]
 10%|█         | 730/7175 [09:58<1:17:12,  1.39it/s]
                                                    
{'loss': '4.026', 'grad_norm': '0.3534', 'learning_rate': '0.0006', 'epoch': '2.545'}
+
 10%|█         | 730/7175 [09:58<1:17:12,  1.39it/s]
 10%|█         | 731/7175 [09:59<1:15:10,  1.43it/s]
 10%|█         | 732/7175 [09:59<1:13:42,  1.46it/s]
 10%|█         | 733/7175 [10:00<1:12:39,  1.48it/s]
 10%|█         | 734/7175 [10:01<1:11:55,  1.49it/s]
 10%|█         | 735/7175 [10:01<1:11:27,  1.50it/s]
 10%|█         | 736/7175 [10:02<1:11:06,  1.51it/s]
 10%|█         | 737/7175 [10:03<1:10:50,  1.51it/s]
 10%|█         | 738/7175 [10:03<1:10:38,  1.52it/s]
 10%|█         | 739/7175 [10:04<1:10:31,  1.52it/s]
 10%|█         | 740/7175 [10:05<1:10:24,  1.52it/s]
                                                    
{'loss': '3.983', 'grad_norm': '0.3169', 'learning_rate': '0.0006', 'epoch': '2.58'}
+
 10%|█         | 740/7175 [10:05<1:10:24,  1.52it/s]
 10%|█         | 741/7175 [10:05<1:10:22,  1.52it/s]
 10%|█         | 742/7175 [10:06<1:10:17,  1.53it/s]
 10%|█         | 743/7175 [10:07<1:10:17,  1.52it/s]
 10%|█         | 744/7175 [10:07<1:10:14,  1.53it/s]
 10%|█         | 745/7175 [10:08<1:10:13,  1.53it/s]
 10%|█         | 746/7175 [10:09<1:10:11,  1.53it/s]
 10%|█         | 747/7175 [10:09<1:10:12,  1.53it/s]
 10%|█         | 748/7175 [10:10<1:10:09,  1.53it/s]
 10%|█         | 749/7175 [10:11<1:10:07,  1.53it/s]
 10%|█         | 750/7175 [10:11<1:10:06,  1.53it/s]
                                                    
{'loss': '3.974', 'grad_norm': '0.3027', 'learning_rate': '0.0006', 'epoch': '2.615'}
+
 10%|█         | 750/7175 [10:11<1:10:06,  1.53it/s]
 10%|█         | 751/7175 [10:12<1:10:09,  1.53it/s]
 10%|█         | 752/7175 [10:13<1:10:06,  1.53it/s]
 10%|█         | 753/7175 [10:13<1:10:03,  1.53it/s]
 11%|█         | 754/7175 [10:14<1:10:03,  1.53it/s]
 11%|█         | 755/7175 [10:15<1:10:06,  1.53it/s]
 11%|█         | 756/7175 [10:15<1:10:04,  1.53it/s]
 11%|█         | 757/7175 [10:16<1:10:02,  1.53it/s]
 11%|█         | 758/7175 [10:17<1:10:01,  1.53it/s]
 11%|█         | 759/7175 [10:17<1:10:03,  1.53it/s]
 11%|█         | 760/7175 [10:18<1:09:59,  1.53it/s]
                                                    
{'loss': '3.954', 'grad_norm': '0.3842', 'learning_rate': '0.0006', 'epoch': '2.65'}
+
 11%|█         | 760/7175 [10:18<1:09:59,  1.53it/s]
 11%|█         | 761/7175 [10:18<1:10:00,  1.53it/s]
 11%|█         | 762/7175 [10:19<1:09:57,  1.53it/s]
 11%|█         | 763/7175 [10:20<1:09:56,  1.53it/s]
 11%|█         | 764/7175 [10:20<1:09:56,  1.53it/s]
 11%|█         | 765/7175 [10:21<1:09:55,  1.53it/s]
 11%|█         | 766/7175 [10:22<1:09:56,  1.53it/s]
 11%|█         | 767/7175 [10:22<1:09:54,  1.53it/s]
 11%|█         | 768/7175 [10:23<1:09:55,  1.53it/s]
 11%|█         | 769/7175 [10:24<1:09:54,  1.53it/s]
 11%|█         | 770/7175 [10:24<1:09:56,  1.53it/s]
                                                    
{'loss': '3.957', 'grad_norm': '0.3028', 'learning_rate': '0.0006', 'epoch': '2.685'}
+
 11%|█         | 770/7175 [10:24<1:09:56,  1.53it/s]
 11%|█         | 771/7175 [10:25<1:09:55,  1.53it/s]
 11%|█         | 772/7175 [10:26<1:09:54,  1.53it/s]
 11%|█         | 773/7175 [10:26<1:09:52,  1.53it/s]
 11%|█         | 774/7175 [10:27<1:09:54,  1.53it/s]
 11%|█         | 775/7175 [10:28<1:09:53,  1.53it/s]
 11%|█         | 776/7175 [10:28<1:09:51,  1.53it/s]
 11%|█         | 777/7175 [10:29<1:09:50,  1.53it/s]
 11%|█         | 778/7175 [10:30<1:09:50,  1.53it/s]
 11%|█         | 779/7175 [10:30<1:09:47,  1.53it/s]
 11%|█         | 780/7175 [10:31<1:09:46,  1.53it/s]
                                                    
{'loss': '3.942', 'grad_norm': '0.3556', 'learning_rate': '0.0006', 'epoch': '2.72'}
+
 11%|█         | 780/7175 [10:31<1:09:46,  1.53it/s]
 11%|█         | 781/7175 [10:32<1:09:47,  1.53it/s]
 11%|█         | 782/7175 [10:32<1:09:49,  1.53it/s]
 11%|█         | 783/7175 [10:33<1:09:45,  1.53it/s]
 11%|█         | 784/7175 [10:34<1:09:45,  1.53it/s]
 11%|█         | 785/7175 [10:34<1:09:42,  1.53it/s]
 11%|█         | 786/7175 [10:35<1:09:46,  1.53it/s]
 11%|█         | 787/7175 [10:36<1:09:43,  1.53it/s]
 11%|█         | 788/7175 [10:36<1:09:44,  1.53it/s]
 11%|█         | 789/7175 [10:37<1:09:41,  1.53it/s]
 11%|█         | 790/7175 [10:37<1:09:43,  1.53it/s]
                                                    
{'loss': '3.928', 'grad_norm': '0.3018', 'learning_rate': '0.0006', 'epoch': '2.755'}
+
 11%|█         | 790/7175 [10:37<1:09:43,  1.53it/s]
 11%|█         | 791/7175 [10:38<1:09:42,  1.53it/s]
 11%|█         | 792/7175 [10:39<1:09:41,  1.53it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.16it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.69it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.53it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.04it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.72it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.57it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.44it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.38it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.31it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.30it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.28it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.26it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.25it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.24it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.23it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.22it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.22it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.23it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s]
                                                    
+
                                               
{'eval_loss': '4.107', 'eval_runtime': '3.255', 'eval_samples_per_second': '749.9', 'eval_steps_per_second': '11.98', 'epoch': '2.762'}
+
 11%|█         | 792/7175 [10:42<1:09:41,  1.53it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-792
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-792/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-792/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.85it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.85it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-792/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-792/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-792/tokenizer_config.json
+[eval] epoch 2.762 step 792 loss 4.1074 (best 4.1074 @ ep 2.762, 0/20 since best)
+
+
                                               
 11%|█         | 793/7175 [10:45<3:59:46,  2.25s/it]
 11%|█         | 794/7175 [10:45<3:08:41,  1.77s/it]
 11%|█         | 795/7175 [10:46<2:32:56,  1.44s/it]
 11%|█         | 796/7175 [10:47<2:07:56,  1.20s/it]
 11%|█         | 797/7175 [10:47<1:50:24,  1.04s/it]
 11%|█         | 798/7175 [10:48<1:38:10,  1.08it/s]
 11%|█         | 799/7175 [10:49<1:29:34,  1.19it/s]
 11%|█         | 800/7175 [10:49<1:23:37,  1.27it/s]
                                                    
{'loss': '3.915', 'grad_norm': '0.2677', 'learning_rate': '0.0006', 'epoch': '2.79'}
+
 11%|█         | 800/7175 [10:49<1:23:37,  1.27it/s]
 11%|█         | 801/7175 [10:50<1:19:25,  1.34it/s]
 11%|█         | 802/7175 [10:51<1:16:26,  1.39it/s]
 11%|█         | 803/7175 [10:51<1:14:21,  1.43it/s]
 11%|█         | 804/7175 [10:52<1:12:54,  1.46it/s]
 11%|█         | 805/7175 [10:53<1:11:53,  1.48it/s]
 11%|█         | 806/7175 [10:53<1:11:10,  1.49it/s]
 11%|█         | 807/7175 [10:54<1:10:38,  1.50it/s]
 11%|█▏        | 808/7175 [10:55<1:10:18,  1.51it/s]
 11%|█▏        | 809/7175 [10:55<1:10:03,  1.51it/s]
 11%|█▏        | 810/7175 [10:56<1:09:52,  1.52it/s]
                                                    
{'loss': '3.896', 'grad_norm': '0.3603', 'learning_rate': '0.0006', 'epoch': '2.825'}
+
 11%|█▏        | 810/7175 [10:56<1:09:52,  1.52it/s]
 11%|█▏        | 811/7175 [10:57<1:09:44,  1.52it/s]
 11%|█▏        | 812/7175 [10:57<1:09:40,  1.52it/s]
 11%|█▏        | 813/7175 [10:58<1:09:34,  1.52it/s]
 11%|█▏        | 814/7175 [10:59<1:09:32,  1.52it/s]
 11%|█▏        | 815/7175 [10:59<1:09:29,  1.53it/s]
 11%|█▏        | 816/7175 [11:00<1:09:28,  1.53it/s]
 11%|█▏        | 817/7175 [11:00<1:09:27,  1.53it/s]
 11%|█▏        | 818/7175 [11:01<1:09:25,  1.53it/s]
 11%|█▏        | 819/7175 [11:02<1:09:24,  1.53it/s]
 11%|█▏        | 820/7175 [11:02<1:09:24,  1.53it/s]
                                                    
{'loss': '3.905', 'grad_norm': '0.3801', 'learning_rate': '0.0006', 'epoch': '2.86'}
+
 11%|█▏        | 820/7175 [11:02<1:09:24,  1.53it/s]
 11%|█▏        | 821/7175 [11:03<1:09:26,  1.53it/s]
 11%|█▏        | 822/7175 [11:04<1:09:22,  1.53it/s]
 11%|█▏        | 823/7175 [11:04<1:09:20,  1.53it/s]
 11%|█▏        | 824/7175 [11:05<1:09:18,  1.53it/s]
 11%|█▏        | 825/7175 [11:06<1:09:18,  1.53it/s]
 12%|█▏        | 826/7175 [11:06<1:09:18,  1.53it/s]
 12%|█▏        | 827/7175 [11:07<1:09:16,  1.53it/s]
 12%|█▏        | 828/7175 [11:08<1:09:17,  1.53it/s]
 12%|█▏        | 829/7175 [11:08<1:09:15,  1.53it/s]
 12%|█▏        | 830/7175 [11:09<1:09:15,  1.53it/s]
                                                    
{'loss': '3.885', 'grad_norm': '0.294', 'learning_rate': '0.0006', 'epoch': '2.895'}
+
 12%|█▏        | 830/7175 [11:09<1:09:15,  1.53it/s]
 12%|█▏        | 831/7175 [11:10<1:09:14,  1.53it/s]
 12%|█▏        | 832/7175 [11:10<1:09:15,  1.53it/s]
 12%|█▏        | 833/7175 [11:11<1:09:17,  1.53it/s]
 12%|█▏        | 834/7175 [11:12<1:09:14,  1.53it/s]
 12%|█▏        | 835/7175 [11:12<1:09:13,  1.53it/s]
 12%|█▏        | 836/7175 [11:13<1:09:13,  1.53it/s]
 12%|█▏        | 837/7175 [11:14<1:09:15,  1.53it/s]
 12%|█▏        | 838/7175 [11:14<1:09:12,  1.53it/s]
 12%|█▏        | 839/7175 [11:15<1:09:11,  1.53it/s]
 12%|█▏        | 840/7175 [11:16<1:09:11,  1.53it/s]
                                                    
{'loss': '3.866', 'grad_norm': '0.3686', 'learning_rate': '0.0006', 'epoch': '2.93'}
+
 12%|█▏        | 840/7175 [11:16<1:09:11,  1.53it/s]
 12%|█▏        | 841/7175 [11:16<1:09:14,  1.52it/s]
 12%|█▏        | 842/7175 [11:17<1:09:11,  1.53it/s]
 12%|█▏        | 843/7175 [11:18<1:09:10,  1.53it/s]
 12%|█▏        | 844/7175 [11:18<1:09:07,  1.53it/s]
 12%|█▏        | 845/7175 [11:19<1:09:07,  1.53it/s]
 12%|█▏        | 846/7175 [11:19<1:09:05,  1.53it/s]
 12%|█▏        | 847/7175 [11:20<1:09:05,  1.53it/s]
 12%|█▏        | 848/7175 [11:21<1:09:03,  1.53it/s]
 12%|█▏        | 849/7175 [11:21<1:09:04,  1.53it/s]
 12%|█▏        | 850/7175 [11:22<1:09:03,  1.53it/s]
                                                    
{'loss': '3.864', 'grad_norm': '0.2596', 'learning_rate': '0.0006', 'epoch': '2.965'}
+
 12%|█▏        | 850/7175 [11:22<1:09:03,  1.53it/s]
 12%|█▏        | 851/7175 [11:23<1:09:04,  1.53it/s]
 12%|█▏        | 852/7175 [11:23<1:09:03,  1.53it/s]
 12%|█▏        | 853/7175 [11:24<1:09:02,  1.53it/s]
 12%|█▏        | 854/7175 [11:25<1:09:04,  1.53it/s]
 12%|█▏        | 855/7175 [11:25<1:09:02,  1.53it/s]
 12%|█▏        | 856/7175 [11:26<1:09:01,  1.53it/s]
 12%|█▏        | 857/7175 [11:27<1:08:59,  1.53it/s]
 12%|█▏        | 858/7175 [11:27<1:09:19,  1.52it/s]
 12%|█▏        | 859/7175 [11:28<1:09:08,  1.52it/s]
 12%|█▏        | 860/7175 [11:29<1:09:04,  1.52it/s]
                                                    
{'loss': '3.863', 'grad_norm': '0.2825', 'learning_rate': '0.0006', 'epoch': '3'}
+
 12%|█▏        | 860/7175 [11:29<1:09:04,  1.52it/s]
 12%|█▏        | 862/7175 [11:29<57:19,  1.84it/s]  
 12%|█▏        | 863/7175 [11:30<1:00:12,  1.75it/s]
 12%|█▏        | 864/7175 [11:31<1:02:30,  1.68it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:02, 17.96it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.64it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.51it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.00it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.71it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.56it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.43it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.39it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.32it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.30it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.27it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.25it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.25it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.24it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.24it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.24it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.22it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.22it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s]
                                                    
+
                                               
{'eval_loss': '4.063', 'eval_runtime': '3.251', 'eval_samples_per_second': '750.8', 'eval_steps_per_second': '11.99', 'epoch': '3.01'}
+
 12%|█▏        | 864/7175 [11:34<1:02:30,  1.68it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-864
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-864/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-864/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.93it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.93it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-864/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-864/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-864/tokenizer_config.json
+[eval] epoch 3.01 step 864 loss 4.0626 (best 4.0626 @ ep 3.01, 0/20 since best)
+
+
                                               
 12%|█▏        | 865/7175 [11:37<3:37:06,  2.06s/it]
 12%|█▏        | 866/7175 [11:37<2:55:32,  1.67s/it]
 12%|█▏        | 867/7175 [11:38<2:25:01,  1.38s/it]
 12%|█▏        | 868/7175 [11:39<2:02:55,  1.17s/it]
 12%|█▏        | 869/7175 [11:39<1:47:05,  1.02s/it]
 12%|█▏        | 870/7175 [11:40<1:35:47,  1.10it/s]
                                                    
{'loss': '3.832', 'grad_norm': '0.4037', 'learning_rate': '0.0006', 'epoch': '3.031'}
+
 12%|█▏        | 870/7175 [11:40<1:35:47,  1.10it/s]
 12%|█▏        | 871/7175 [11:41<1:27:46,  1.20it/s]
 12%|█▏        | 872/7175 [11:41<1:22:09,  1.28it/s]
 12%|█▏        | 873/7175 [11:42<1:18:07,  1.34it/s]
 12%|█▏        | 874/7175 [11:43<1:15:17,  1.39it/s]
 12%|█▏        | 875/7175 [11:43<1:13:18,  1.43it/s]
 12%|█▏        | 876/7175 [11:44<1:12:00,  1.46it/s]
 12%|█▏        | 877/7175 [11:45<1:11:00,  1.48it/s]
 12%|█▏        | 878/7175 [11:45<1:10:16,  1.49it/s]
 12%|█▏        | 879/7175 [11:46<1:09:47,  1.50it/s]
 12%|█▏        | 880/7175 [11:47<1:09:30,  1.51it/s]
                                                    
{'loss': '3.822', 'grad_norm': '0.3533', 'learning_rate': '0.0006', 'epoch': '3.066'}
+
 12%|█▏        | 880/7175 [11:47<1:09:30,  1.51it/s]
 12%|█▏        | 881/7175 [11:47<1:09:16,  1.51it/s]
 12%|█▏        | 882/7175 [11:48<1:09:03,  1.52it/s]
 12%|█▏        | 883/7175 [11:49<1:08:55,  1.52it/s]
 12%|█▏        | 884/7175 [11:49<1:08:52,  1.52it/s]
 12%|█▏        | 885/7175 [11:50<1:08:46,  1.52it/s]
 12%|█▏        | 886/7175 [11:51<1:08:41,  1.53it/s]
 12%|█▏        | 887/7175 [11:51<1:08:38,  1.53it/s]
 12%|█▏        | 888/7175 [11:52<1:08:38,  1.53it/s]
 12%|█▏        | 889/7175 [11:53<1:08:37,  1.53it/s]
 12%|█▏        | 890/7175 [11:53<1:08:35,  1.53it/s]
                                                    
{'loss': '3.798', 'grad_norm': '0.3137', 'learning_rate': '0.0006', 'epoch': '3.101'}
+
 12%|█▏        | 890/7175 [11:53<1:08:35,  1.53it/s]
 12%|█▏        | 891/7175 [11:54<1:08:39,  1.53it/s]
 12%|█▏        | 892/7175 [11:54<1:08:40,  1.52it/s]
 12%|█▏        | 893/7175 [11:55<1:08:39,  1.52it/s]
 12%|█▏        | 894/7175 [11:56<1:08:36,  1.53it/s]
 12%|█▏        | 895/7175 [11:56<1:08:34,  1.53it/s]
 12%|█▏        | 896/7175 [11:57<1:08:35,  1.53it/s]
 13%|█▎        | 897/7175 [11:58<1:08:32,  1.53it/s]
 13%|█▎        | 898/7175 [11:58<1:08:29,  1.53it/s]
 13%|█▎        | 899/7175 [11:59<1:08:28,  1.53it/s]
 13%|█▎        | 900/7175 [12:00<1:08:29,  1.53it/s]
                                                    
{'loss': '3.778', 'grad_norm': '0.2593', 'learning_rate': '0.0006', 'epoch': '3.136'}
+
 13%|█▎        | 900/7175 [12:00<1:08:29,  1.53it/s]
 13%|█▎        | 901/7175 [12:00<1:08:28,  1.53it/s]
 13%|█▎        | 902/7175 [12:01<1:08:24,  1.53it/s]
 13%|█▎        | 903/7175 [12:02<1:08:27,  1.53it/s]
 13%|█▎        | 904/7175 [12:02<1:08:27,  1.53it/s]
 13%|█▎        | 905/7175 [12:03<1:08:26,  1.53it/s]
 13%|█▎        | 906/7175 [12:04<1:08:23,  1.53it/s]
 13%|█▎        | 907/7175 [12:04<1:08:24,  1.53it/s]
 13%|█▎        | 908/7175 [12:05<1:08:23,  1.53it/s]
 13%|█▎        | 909/7175 [12:06<1:08:22,  1.53it/s]
 13%|█▎        | 910/7175 [12:06<1:08:21,  1.53it/s]
                                                    
{'loss': '3.777', 'grad_norm': '0.3829', 'learning_rate': '0.0006', 'epoch': '3.171'}
+
 13%|█▎        | 910/7175 [12:06<1:08:21,  1.53it/s]
 13%|█▎        | 911/7175 [12:07<1:08:22,  1.53it/s]
 13%|█▎        | 912/7175 [12:08<1:08:21,  1.53it/s]
 13%|█▎        | 913/7175 [12:08<1:08:19,  1.53it/s]
 13%|█▎        | 914/7175 [12:09<1:08:20,  1.53it/s]
 13%|█▎        | 915/7175 [12:10<1:08:21,  1.53it/s]
 13%|█▎        | 916/7175 [12:10<1:08:20,  1.53it/s]
 13%|█▎        | 917/7175 [12:11<1:08:19,  1.53it/s]
 13%|█▎        | 918/7175 [12:12<1:08:17,  1.53it/s]
 13%|█▎        | 919/7175 [12:12<1:08:17,  1.53it/s]
 13%|█▎        | 920/7175 [12:13<1:08:16,  1.53it/s]
                                                    
{'loss': '3.764', 'grad_norm': '0.311', 'learning_rate': '0.0006', 'epoch': '3.206'}
+
 13%|█▎        | 920/7175 [12:13<1:08:16,  1.53it/s]
 13%|█▎        | 921/7175 [12:13<1:08:16,  1.53it/s]
 13%|█▎        | 922/7175 [12:14<1:08:14,  1.53it/s]
 13%|█▎        | 923/7175 [12:15<1:08:15,  1.53it/s]
 13%|█▎        | 924/7175 [12:15<1:08:15,  1.53it/s]
 13%|█▎        | 925/7175 [12:16<1:08:13,  1.53it/s]
 13%|█▎        | 926/7175 [12:17<1:08:11,  1.53it/s]
 13%|█▎        | 927/7175 [12:17<1:08:13,  1.53it/s]
 13%|█▎        | 928/7175 [12:18<1:08:12,  1.53it/s]
 13%|█▎        | 929/7175 [12:19<1:08:11,  1.53it/s]
 13%|█▎        | 930/7175 [12:19<1:08:09,  1.53it/s]
                                                    
{'loss': '3.755', 'grad_norm': '0.2324', 'learning_rate': '0.0006', 'epoch': '3.241'}
+
 13%|█▎        | 930/7175 [12:19<1:08:09,  1.53it/s]
 13%|█▎        | 931/7175 [12:20<1:08:10,  1.53it/s]
 13%|█▎        | 932/7175 [12:21<1:08:10,  1.53it/s]
 13%|█▎        | 933/7175 [12:21<1:08:08,  1.53it/s]
 13%|█▎        | 934/7175 [12:22<1:08:06,  1.53it/s]
 13%|█▎        | 935/7175 [12:23<1:08:06,  1.53it/s]
 13%|█▎        | 936/7175 [12:23<1:08:06,  1.53it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.52it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.79it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.61it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.07it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.76it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.58it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.45it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.38it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.33it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.30it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.29it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.28it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.27it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.25it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.24it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.24it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.24it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.23it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s]
                                                    
+
                                               
{'eval_loss': '3.974', 'eval_runtime': '3.246', 'eval_samples_per_second': '752.1', 'eval_steps_per_second': '12.02', 'epoch': '3.262'}
+
 13%|█▎        | 936/7175 [12:27<1:08:06,  1.53it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-936
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-936/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-936/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.74it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.74it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-936/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-936/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-936/tokenizer_config.json
+[eval] epoch 3.262 step 936 loss 3.9743 (best 3.9743 @ ep 3.262, 0/20 since best)
+
+
                                               
 13%|█▎        | 937/7175 [12:29<3:54:56,  2.26s/it]
 13%|█▎        | 938/7175 [12:30<3:04:47,  1.78s/it]
 13%|█▎        | 939/7175 [12:31<2:29:44,  1.44s/it]
 13%|█▎        | 940/7175 [12:31<2:05:14,  1.21s/it]
                                                    
{'loss': '3.75', 'grad_norm': '0.2616', 'learning_rate': '0.0006', 'epoch': '3.276'}
+
 13%|█▎        | 940/7175 [12:31<2:05:14,  1.21s/it]
 13%|█▎        | 941/7175 [12:32<1:48:06,  1.04s/it]
 13%|█▎        | 942/7175 [12:33<1:36:01,  1.08it/s]
 13%|█▎        | 943/7175 [12:33<1:27:39,  1.18it/s]
 13%|█▎        | 944/7175 [12:34<1:21:46,  1.27it/s]
 13%|█▎        | 945/7175 [12:35<1:17:38,  1.34it/s]
 13%|█▎        | 946/7175 [12:35<1:14:41,  1.39it/s]
 13%|█▎        | 947/7175 [12:36<1:12:38,  1.43it/s]
 13%|█▎        | 948/7175 [12:37<1:11:12,  1.46it/s]
 13%|█▎        | 949/7175 [12:37<1:10:14,  1.48it/s]
 13%|█▎        | 950/7175 [12:38<1:09:30,  1.49it/s]
                                                    
{'loss': '3.738', 'grad_norm': '0.239', 'learning_rate': '0.0006', 'epoch': '3.311'}
+
 13%|█▎        | 950/7175 [12:38<1:09:30,  1.49it/s]
 13%|█▎        | 951/7175 [12:38<1:09:01,  1.50it/s]
 13%|█▎        | 952/7175 [12:39<1:08:41,  1.51it/s]
 13%|█▎        | 953/7175 [12:40<1:08:27,  1.51it/s]
 13%|█▎        | 954/7175 [12:40<1:08:17,  1.52it/s]
 13%|█▎        | 955/7175 [12:41<1:08:14,  1.52it/s]
 13%|█▎        | 956/7175 [12:42<1:08:06,  1.52it/s]
 13%|█▎        | 957/7175 [12:42<1:08:04,  1.52it/s]
 13%|█▎        | 958/7175 [12:43<1:07:59,  1.52it/s]
 13%|█▎        | 959/7175 [12:44<1:07:56,  1.52it/s]
 13%|█▎        | 960/7175 [12:44<1:07:54,  1.53it/s]
                                                    
{'loss': '3.739', 'grad_norm': '0.3664', 'learning_rate': '0.0006', 'epoch': '3.346'}
+
 13%|█▎        | 960/7175 [12:44<1:07:54,  1.53it/s]
 13%|█▎        | 961/7175 [12:45<1:07:55,  1.52it/s]
 13%|█▎        | 962/7175 [12:46<1:07:52,  1.53it/s]
 13%|█▎        | 963/7175 [12:46<1:07:51,  1.53it/s]
 13%|█▎        | 964/7175 [12:47<1:07:48,  1.53it/s]
 13%|█▎        | 965/7175 [12:48<1:07:49,  1.53it/s]
 13%|█▎        | 966/7175 [12:48<1:07:46,  1.53it/s]
 13%|█▎        | 967/7175 [12:49<1:07:49,  1.53it/s]
 13%|█▎        | 968/7175 [12:50<1:07:46,  1.53it/s]
 14%|█▎        | 969/7175 [12:50<1:07:46,  1.53it/s]
 14%|█▎        | 970/7175 [12:51<1:07:44,  1.53it/s]
                                                    
{'loss': '3.749', 'grad_norm': '0.2574', 'learning_rate': '0.0006', 'epoch': '3.381'}
+
 14%|█▎        | 970/7175 [12:51<1:07:44,  1.53it/s]
 14%|█▎        | 971/7175 [12:52<1:07:45,  1.53it/s]
 14%|█▎        | 972/7175 [12:52<1:07:42,  1.53it/s]
 14%|█▎        | 973/7175 [12:53<1:07:42,  1.53it/s]
 14%|█▎        | 974/7175 [12:54<1:07:42,  1.53it/s]
 14%|█▎        | 975/7175 [12:54<1:07:41,  1.53it/s]
 14%|█▎        | 976/7175 [12:55<1:07:39,  1.53it/s]
 14%|█▎        | 977/7175 [12:56<1:07:41,  1.53it/s]
 14%|█▎        | 978/7175 [12:56<1:07:42,  1.53it/s]
 14%|█▎        | 979/7175 [12:57<1:07:39,  1.53it/s]
 14%|█▎        | 980/7175 [12:57<1:07:38,  1.53it/s]
                                                    
{'loss': '3.727', 'grad_norm': '0.2923', 'learning_rate': '0.0006', 'epoch': '3.416'}
+
 14%|█▎        | 980/7175 [12:57<1:07:38,  1.53it/s]
 14%|█▎        | 981/7175 [12:58<1:07:39,  1.53it/s]
 14%|█▎        | 982/7175 [12:59<1:07:40,  1.53it/s]
 14%|█▎        | 983/7175 [12:59<1:07:38,  1.53it/s]
 14%|█▎        | 984/7175 [13:00<1:07:38,  1.53it/s]
 14%|█▎        | 985/7175 [13:01<1:07:36,  1.53it/s]
 14%|█▎        | 986/7175 [13:01<1:07:37,  1.53it/s]
 14%|█▍        | 987/7175 [13:02<1:07:37,  1.52it/s]
 14%|█▍        | 988/7175 [13:03<1:07:34,  1.53it/s]
 14%|█▍        | 989/7175 [13:03<1:07:32,  1.53it/s]
 14%|█▍        | 990/7175 [13:04<1:07:31,  1.53it/s]
                                                    
{'loss': '3.717', 'grad_norm': '0.2536', 'learning_rate': '0.0006', 'epoch': '3.451'}
+
 14%|█▍        | 990/7175 [13:04<1:07:31,  1.53it/s]
 14%|█▍        | 991/7175 [13:05<1:07:33,  1.53it/s]
 14%|█▍        | 992/7175 [13:05<1:07:30,  1.53it/s]
 14%|█▍        | 993/7175 [13:06<1:07:28,  1.53it/s]
 14%|█▍        | 994/7175 [13:07<1:07:29,  1.53it/s]
 14%|█▍        | 995/7175 [13:07<1:07:28,  1.53it/s]
 14%|█▍        | 996/7175 [13:08<1:07:26,  1.53it/s]
 14%|█▍        | 997/7175 [13:09<1:07:24,  1.53it/s]
 14%|█▍        | 998/7175 [13:09<1:07:24,  1.53it/s]
 14%|█▍        | 999/7175 [13:10<1:07:22,  1.53it/s]
 14%|█▍        | 1000/7175 [13:11<1:07:22,  1.53it/s]
                                                     
{'loss': '3.71', 'grad_norm': '0.3768', 'learning_rate': '0.0006', 'epoch': '3.486'}
+
 14%|█▍        | 1000/7175 [13:11<1:07:22,  1.53it/s]
 14%|█▍        | 1001/7175 [13:11<1:07:23,  1.53it/s]
 14%|█▍        | 1002/7175 [13:12<1:07:24,  1.53it/s]
 14%|█▍        | 1003/7175 [13:13<1:07:23,  1.53it/s]
 14%|█▍        | 1004/7175 [13:13<1:07:23,  1.53it/s]
 14%|█▍        | 1005/7175 [13:14<1:07:20,  1.53it/s]
 14%|█▍        | 1006/7175 [13:15<1:07:21,  1.53it/s]
 14%|█▍        | 1007/7175 [13:15<1:07:25,  1.52it/s]
 14%|█▍        | 1008/7175 [13:16<1:07:21,  1.53it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.67it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.83it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.64it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.09it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.77it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.59it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.45it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.38it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.31it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.29it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.27it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.26it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.25it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.24it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.23it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.22it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.23it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.22it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s]
                                                     
+
                                               
{'eval_loss': '3.941', 'eval_runtime': '3.261', 'eval_samples_per_second': '748.5', 'eval_steps_per_second': '11.96', 'epoch': '3.514'}
+
 14%|█▍        | 1008/7175 [13:19<1:07:21,  1.53it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-1008
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-1008/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-1008/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  2.00it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  2.00it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-1008/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-1008/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-1008/tokenizer_config.json
+[eval] epoch 3.514 step 1008 loss 3.9414 (best 3.9414 @ ep 3.514, 0/20 since best)
+
+
                                               
 14%|█▍        | 1009/7175 [13:22<3:49:39,  2.23s/it]
 14%|█▍        | 1010/7175 [13:22<3:00:52,  1.76s/it]
                                                     
{'loss': '3.716', 'grad_norm': '0.4354', 'learning_rate': '0.0006', 'epoch': '3.521'}
+
 14%|█▍        | 1010/7175 [13:22<3:00:52,  1.76s/it]
 14%|█▍        | 1011/7175 [13:23<2:26:47,  1.43s/it]
 14%|█▍        | 1012/7175 [13:24<2:02:52,  1.20s/it]
 14%|█▍        | 1013/7175 [13:24<1:46:09,  1.03s/it]
 14%|█▍        | 1014/7175 [13:25<1:34:27,  1.09it/s]
 14%|█▍        | 1015/7175 [13:26<1:26:18,  1.19it/s]
 14%|█▍        | 1016/7175 [13:26<1:20:33,  1.27it/s]
 14%|█▍        | 1017/7175 [13:27<1:16:31,  1.34it/s]
 14%|█▍        | 1018/7175 [13:28<1:13:42,  1.39it/s]
 14%|█▍        | 1019/7175 [13:28<1:11:46,  1.43it/s]
 14%|█▍        | 1020/7175 [13:29<1:10:23,  1.46it/s]
                                                     
{'loss': '3.722', 'grad_norm': '0.2765', 'learning_rate': '0.0006', 'epoch': '3.556'}
+
 14%|█▍        | 1020/7175 [13:29<1:10:23,  1.46it/s]
 14%|█▍        | 1021/7175 [13:30<1:09:28,  1.48it/s]
 14%|█▍        | 1022/7175 [13:30<1:08:45,  1.49it/s]
 14%|█▍        | 1023/7175 [13:31<1:08:18,  1.50it/s]
 14%|█▍        | 1024/7175 [13:32<1:07:55,  1.51it/s]
 14%|█▍        | 1025/7175 [13:32<1:07:41,  1.51it/s]
 14%|█▍        | 1026/7175 [13:33<1:07:29,  1.52it/s]
 14%|█▍        | 1027/7175 [13:34<1:07:23,  1.52it/s]
 14%|█▍        | 1028/7175 [13:34<1:07:17,  1.52it/s]
 14%|█▍        | 1029/7175 [13:35<1:07:14,  1.52it/s]
 14%|█▍        | 1030/7175 [13:35<1:07:12,  1.52it/s]
                                                     
{'loss': '3.702', 'grad_norm': '0.2803', 'learning_rate': '0.0006', 'epoch': '3.591'}
+
 14%|█▍        | 1030/7175 [13:35<1:07:12,  1.52it/s]
 14%|█▍        | 1031/7175 [13:36<1:07:11,  1.52it/s]
 14%|█▍        | 1032/7175 [13:37<1:07:12,  1.52it/s]
 14%|█▍        | 1033/7175 [13:37<1:07:08,  1.52it/s]
 14%|█▍        | 1034/7175 [13:38<1:07:05,  1.53it/s]
 14%|█▍        | 1035/7175 [13:39<1:07:04,  1.53it/s]
 14%|█▍        | 1036/7175 [13:39<1:07:05,  1.53it/s]
 14%|█▍        | 1037/7175 [13:40<1:07:04,  1.53it/s]
 14%|█▍        | 1038/7175 [13:41<1:07:03,  1.53it/s]
 14%|█▍        | 1039/7175 [13:41<1:07:02,  1.53it/s]
 14%|█▍        | 1040/7175 [13:42<1:07:00,  1.53it/s]
                                                     
{'loss': '3.7', 'grad_norm': '0.2497', 'learning_rate': '0.0006', 'epoch': '3.626'}
+
 14%|█▍        | 1040/7175 [13:42<1:07:00,  1.53it/s]
 15%|█▍        | 1041/7175 [13:43<1:07:03,  1.52it/s]
 15%|█▍        | 1042/7175 [13:43<1:06:59,  1.53it/s]
 15%|█▍        | 1043/7175 [13:44<1:06:57,  1.53it/s]
 15%|█▍        | 1044/7175 [13:45<1:06:55,  1.53it/s]
 15%|█▍        | 1045/7175 [13:45<1:06:56,  1.53it/s]
 15%|█▍        | 1046/7175 [13:46<1:06:55,  1.53it/s]
 15%|█▍        | 1047/7175 [13:47<1:06:53,  1.53it/s]
 15%|█▍        | 1048/7175 [13:47<1:06:53,  1.53it/s]
 15%|█▍        | 1049/7175 [13:48<1:06:54,  1.53it/s]
 15%|█▍        | 1050/7175 [13:49<1:06:53,  1.53it/s]
                                                     
{'loss': '3.692', 'grad_norm': '0.2272', 'learning_rate': '0.0006', 'epoch': '3.661'}
+
 15%|█▍        | 1050/7175 [13:49<1:06:53,  1.53it/s]
 15%|█▍        | 1051/7175 [13:49<1:06:52,  1.53it/s]
 15%|█▍        | 1052/7175 [13:50<1:06:50,  1.53it/s]
 15%|█▍        | 1053/7175 [13:51<1:06:53,  1.53it/s]
 15%|█▍        | 1054/7175 [13:51<1:06:50,  1.53it/s]
 15%|█▍        | 1055/7175 [13:52<1:06:49,  1.53it/s]
 15%|█▍        | 1056/7175 [13:53<1:06:47,  1.53it/s]
 15%|█▍        | 1057/7175 [13:53<1:06:49,  1.53it/s]
 15%|█▍        | 1058/7175 [13:54<1:06:48,  1.53it/s]
 15%|█▍        | 1059/7175 [13:55<1:06:47,  1.53it/s]
 15%|█▍        | 1060/7175 [13:55<1:06:44,  1.53it/s]
                                                     
{'loss': '3.685', 'grad_norm': '0.2815', 'learning_rate': '0.0006', 'epoch': '3.696'}
+
 15%|█▍        | 1060/7175 [13:55<1:06:44,  1.53it/s]
 15%|█▍        | 1061/7175 [13:56<1:06:46,  1.53it/s]
 15%|█▍        | 1062/7175 [13:56<1:06:45,  1.53it/s]
 15%|█▍        | 1063/7175 [13:57<1:06:44,  1.53it/s]
 15%|█▍        | 1064/7175 [13:58<1:06:40,  1.53it/s]
 15%|█▍        | 1065/7175 [13:58<1:06:41,  1.53it/s]
 15%|█▍        | 1066/7175 [13:59<1:06:40,  1.53it/s]
 15%|█▍        | 1067/7175 [14:00<1:06:41,  1.53it/s]
 15%|█▍        | 1068/7175 [14:00<1:06:38,  1.53it/s]
 15%|█▍        | 1069/7175 [14:01<1:06:40,  1.53it/s]
 15%|█▍        | 1070/7175 [14:02<1:06:37,  1.53it/s]
                                                     
{'loss': '3.672', 'grad_norm': '0.3078', 'learning_rate': '0.0006', 'epoch': '3.73'}
+
 15%|█▍        | 1070/7175 [14:02<1:06:37,  1.53it/s]
 15%|█▍        | 1071/7175 [14:02<1:06:39,  1.53it/s]
 15%|█▍        | 1072/7175 [14:03<1:06:36,  1.53it/s]
 15%|█▍        | 1073/7175 [14:04<1:06:36,  1.53it/s]
 15%|█▍        | 1074/7175 [14:04<1:06:34,  1.53it/s]
 15%|█▍        | 1075/7175 [14:05<1:06:36,  1.53it/s]
 15%|█▍        | 1076/7175 [14:06<1:06:33,  1.53it/s]
 15%|█▌        | 1077/7175 [14:06<1:06:36,  1.53it/s]
 15%|█▌        | 1078/7175 [14:07<1:06:35,  1.53it/s]
 15%|█▌        | 1079/7175 [14:08<1:06:34,  1.53it/s]
 15%|█▌        | 1080/7175 [14:08<1:06:34,  1.53it/s]
                                                     
{'loss': '3.67', 'grad_norm': '0.2631', 'learning_rate': '0.0006', 'epoch': '3.765'}
+
 15%|█▌        | 1080/7175 [14:08<1:06:34,  1.53it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.51it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.78it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.60it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.04it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.73it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.58it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.44it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.37it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.33it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.29it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.27it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.27it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.25it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.25it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.25it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.23it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.24it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.24it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s]
                                                     
+
                                               
{'eval_loss': '3.884', 'eval_runtime': '3.249', 'eval_samples_per_second': '751.2', 'eval_steps_per_second': '12', 'epoch': '3.765'}
+
 15%|█▌        | 1080/7175 [14:12<1:06:34,  1.53it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-1080
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-1080/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-1080/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.82it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.82it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-1080/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-1080/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-1080/tokenizer_config.json
+[eval] epoch 3.765 step 1080 loss 3.8835 (best 3.8835 @ ep 3.765, 0/20 since best)
+
+
                                               
 15%|█▌        | 1081/7175 [14:14<3:49:22,  2.26s/it]
 15%|█▌        | 1082/7175 [14:15<3:00:28,  1.78s/it]
 15%|█▌        | 1083/7175 [14:16<2:26:13,  1.44s/it]
 15%|█▌        | 1084/7175 [14:16<2:02:15,  1.20s/it]
 15%|█▌        | 1085/7175 [14:17<1:45:29,  1.04s/it]
 15%|█▌        | 1086/7175 [14:18<1:33:45,  1.08it/s]
 15%|█▌        | 1087/7175 [14:18<1:25:34,  1.19it/s]
 15%|█▌        | 1088/7175 [14:19<1:19:48,  1.27it/s]
 15%|█▌        | 1089/7175 [14:19<1:15:46,  1.34it/s]
 15%|█▌        | 1090/7175 [14:20<1:12:58,  1.39it/s]
                                                     
{'loss': '3.661', 'grad_norm': '0.3203', 'learning_rate': '0.0006', 'epoch': '3.8'}
+
 15%|█▌        | 1090/7175 [14:20<1:12:58,  1.39it/s]
 15%|█▌        | 1091/7175 [14:21<1:11:00,  1.43it/s]
 15%|█▌        | 1092/7175 [14:21<1:09:36,  1.46it/s]
 15%|█▌        | 1093/7175 [14:22<1:08:38,  1.48it/s]
 15%|█▌        | 1094/7175 [14:23<1:07:57,  1.49it/s]
 15%|█▌        | 1095/7175 [14:23<1:07:28,  1.50it/s]
 15%|█▌        | 1096/7175 [14:24<1:07:05,  1.51it/s]
 15%|█▌        | 1097/7175 [14:25<1:06:53,  1.51it/s]
 15%|█▌        | 1098/7175 [14:25<1:06:42,  1.52it/s]
 15%|█▌        | 1099/7175 [14:26<1:06:35,  1.52it/s]
 15%|█▌        | 1100/7175 [14:27<1:06:28,  1.52it/s]
                                                     
{'loss': '3.652', 'grad_norm': '0.2822', 'learning_rate': '0.0006', 'epoch': '3.835'}
+
 15%|█▌        | 1100/7175 [14:27<1:06:28,  1.52it/s]
 15%|█▌        | 1101/7175 [14:27<1:06:29,  1.52it/s]
 15%|█▌        | 1102/7175 [14:28<1:06:27,  1.52it/s]
 15%|█▌        | 1103/7175 [14:29<1:06:22,  1.52it/s]
 15%|█▌        | 1104/7175 [14:29<1:06:18,  1.53it/s]
 15%|█▌        | 1105/7175 [14:30<1:06:16,  1.53it/s]
 15%|█▌        | 1106/7175 [14:31<1:06:17,  1.53it/s]
 15%|█▌        | 1107/7175 [14:31<1:06:16,  1.53it/s]
 15%|█▌        | 1108/7175 [14:32<1:06:13,  1.53it/s]
 15%|█▌        | 1109/7175 [14:33<1:06:11,  1.53it/s]
 15%|█▌        | 1110/7175 [14:33<1:06:11,  1.53it/s]
                                                     
{'loss': '3.66', 'grad_norm': '0.2555', 'learning_rate': '0.0006', 'epoch': '3.87'}
+
 15%|█▌        | 1110/7175 [14:33<1:06:11,  1.53it/s]
 15%|█▌        | 1111/7175 [14:34<1:06:12,  1.53it/s]
 15%|█▌        | 1112/7175 [14:35<1:06:10,  1.53it/s]
 16%|█▌        | 1113/7175 [14:35<1:06:10,  1.53it/s]
 16%|█▌        | 1114/7175 [14:36<1:06:10,  1.53it/s]
 16%|█▌        | 1115/7175 [14:37<1:06:09,  1.53it/s]
 16%|█▌        | 1116/7175 [14:37<1:06:06,  1.53it/s]
 16%|█▌        | 1117/7175 [14:38<1:06:07,  1.53it/s]
 16%|█▌        | 1118/7175 [14:38<1:06:06,  1.53it/s]
 16%|█▌        | 1119/7175 [14:39<1:06:07,  1.53it/s]
 16%|█▌        | 1120/7175 [14:40<1:06:04,  1.53it/s]
                                                     
{'loss': '3.652', 'grad_norm': '0.2475', 'learning_rate': '0.0006', 'epoch': '3.905'}
+
 16%|█▌        | 1120/7175 [14:40<1:06:04,  1.53it/s]
 16%|█▌        | 1121/7175 [14:40<1:06:06,  1.53it/s]
 16%|█▌        | 1122/7175 [14:41<1:06:06,  1.53it/s]
 16%|█▌        | 1123/7175 [14:42<1:06:04,  1.53it/s]
 16%|█▌        | 1124/7175 [14:42<1:06:03,  1.53it/s]
 16%|█▌        | 1125/7175 [14:43<1:06:03,  1.53it/s]
 16%|█▌        | 1126/7175 [14:44<1:06:04,  1.53it/s]
 16%|█▌        | 1127/7175 [14:44<1:06:02,  1.53it/s]
 16%|█▌        | 1128/7175 [14:45<1:06:02,  1.53it/s]
 16%|█▌        | 1129/7175 [14:46<1:06:00,  1.53it/s]
 16%|█▌        | 1130/7175 [14:46<1:06:03,  1.53it/s]
                                                     
{'loss': '3.642', 'grad_norm': '0.3586', 'learning_rate': '0.0006', 'epoch': '3.94'}
+
 16%|█▌        | 1130/7175 [14:46<1:06:03,  1.53it/s]
 16%|█▌        | 1131/7175 [14:47<1:06:02,  1.53it/s]
 16%|█▌        | 1132/7175 [14:48<1:06:00,  1.53it/s]
 16%|█▌        | 1133/7175 [14:48<1:05:59,  1.53it/s]
 16%|█▌        | 1134/7175 [14:49<1:06:00,  1.53it/s]
 16%|█▌        | 1135/7175 [14:50<1:05:58,  1.53it/s]
 16%|█▌        | 1136/7175 [14:50<1:05:58,  1.53it/s]
 16%|█▌        | 1137/7175 [14:51<1:05:55,  1.53it/s]
 16%|█▌        | 1138/7175 [14:52<1:05:55,  1.53it/s]
 16%|█▌        | 1139/7175 [14:52<1:05:56,  1.53it/s]
 16%|█▌        | 1140/7175 [14:53<1:05:53,  1.53it/s]
                                                     
{'loss': '3.63', 'grad_norm': '0.2792', 'learning_rate': '0.0006', 'epoch': '3.975'}
+
 16%|█▌        | 1140/7175 [14:53<1:05:53,  1.53it/s]
 16%|█▌        | 1141/7175 [14:54<1:05:53,  1.53it/s]
 16%|█▌        | 1142/7175 [14:54<1:05:52,  1.53it/s]
 16%|█▌        | 1143/7175 [14:55<1:05:52,  1.53it/s]
 16%|█▌        | 1144/7175 [14:56<1:05:50,  1.53it/s]
 16%|█▌        | 1145/7175 [14:56<1:06:10,  1.52it/s]
 16%|█▌        | 1146/7175 [14:57<1:06:01,  1.52it/s]
 16%|█▌        | 1147/7175 [14:57<1:05:54,  1.52it/s]
 16%|█▌        | 1149/7175 [14:58<55:06,  1.82it/s]  
 16%|█▌        | 1150/7175 [14:59<57:43,  1.74it/s]
                                                   
{'loss': '3.622', 'grad_norm': '0.4094', 'learning_rate': '0.0006', 'epoch': '4.007'}
+
 16%|█▌        | 1150/7175 [14:59<57:43,  1.74it/s]
 16%|█▌        | 1151/7175 [15:00<59:50,  1.68it/s]
 16%|█▌        | 1152/7175 [15:00<1:01:28,  1.63it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.72it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.86it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.64it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.08it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.76it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.58it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.46it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.39it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.33it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.30it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.29it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.27it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.26it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.25it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.24it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.25it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.25it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.25it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s]
                                                     
+
                                               
{'eval_loss': '3.885', 'eval_runtime': '3.245', 'eval_samples_per_second': '752.3', 'eval_steps_per_second': '12.02', 'epoch': '4.014'}
+
 16%|█▌        | 1152/7175 [15:04<1:01:28,  1.63it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s][eval] epoch 4.014 step 1152 loss 3.8852 (best 3.8835 @ ep 3.765, 1/20 since best)
+
+
                                               
 16%|█▌        | 1153/7175 [15:04<2:33:48,  1.53s/it]
 16%|█▌        | 1154/7175 [15:05<2:08:37,  1.28s/it]
 16%|█▌        | 1155/7175 [15:06<1:50:21,  1.10s/it]
 16%|█▌        | 1156/7175 [15:06<1:37:15,  1.03it/s]
 16%|█▌        | 1157/7175 [15:07<1:27:58,  1.14it/s]
 16%|█▌        | 1158/7175 [15:07<1:21:19,  1.23it/s]
 16%|█▌        | 1159/7175 [15:08<1:16:38,  1.31it/s]
 16%|█▌        | 1160/7175 [15:09<1:13:19,  1.37it/s]
                                                     
{'loss': '3.61', 'grad_norm': '0.3626', 'learning_rate': '0.0006', 'epoch': '4.042'}
+
 16%|█▌        | 1160/7175 [15:09<1:13:19,  1.37it/s]
 16%|█▌        | 1161/7175 [15:09<1:11:04,  1.41it/s]
 16%|█▌        | 1162/7175 [15:10<1:09:24,  1.44it/s]
 16%|█▌        | 1163/7175 [15:11<1:08:14,  1.47it/s]
 16%|█▌        | 1164/7175 [15:11<1:07:27,  1.49it/s]
 16%|█▌        | 1165/7175 [15:12<1:06:55,  1.50it/s]
 16%|█▋        | 1166/7175 [15:13<1:06:31,  1.51it/s]
 16%|█▋        | 1167/7175 [15:13<1:06:11,  1.51it/s]
 16%|█▋        | 1168/7175 [15:14<1:06:00,  1.52it/s]
 16%|█▋        | 1169/7175 [15:15<1:05:52,  1.52it/s]
 16%|█▋        | 1170/7175 [15:15<1:05:56,  1.52it/s]
                                                     
{'loss': '3.584', 'grad_norm': '0.2344', 'learning_rate': '0.0006', 'epoch': '4.077'}
+
 16%|█▋        | 1170/7175 [15:15<1:05:56,  1.52it/s]
 16%|█▋        | 1171/7175 [15:16<1:05:51,  1.52it/s]
 16%|█▋        | 1172/7175 [15:17<1:05:43,  1.52it/s]
 16%|█▋        | 1173/7175 [15:17<1:05:39,  1.52it/s]
 16%|█▋        | 1174/7175 [15:18<1:05:42,  1.52it/s]
 16%|█▋        | 1175/7175 [15:19<1:05:38,  1.52it/s]
 16%|█▋        | 1176/7175 [15:19<1:05:35,  1.52it/s]
 16%|█▋        | 1177/7175 [15:20<1:05:31,  1.53it/s]
 16%|█▋        | 1178/7175 [15:21<1:05:31,  1.53it/s]
 16%|█▋        | 1179/7175 [15:21<1:05:30,  1.53it/s]
 16%|█▋        | 1180/7175 [15:22<1:05:28,  1.53it/s]
                                                     
{'loss': '3.565', 'grad_norm': '0.2487', 'learning_rate': '0.0006', 'epoch': '4.112'}
+
 16%|█▋        | 1180/7175 [15:22<1:05:28,  1.53it/s]
 16%|█▋        | 1181/7175 [15:23<1:05:27,  1.53it/s]
 16%|█▋        | 1182/7175 [15:23<1:05:26,  1.53it/s]
 16%|█▋        | 1183/7175 [15:24<1:05:27,  1.53it/s]
 17%|█▋        | 1184/7175 [15:25<1:05:25,  1.53it/s]
 17%|█▋        | 1185/7175 [15:25<1:05:21,  1.53it/s]
 17%|█▋        | 1186/7175 [15:26<1:05:23,  1.53it/s]
 17%|█▋        | 1187/7175 [15:26<1:05:26,  1.53it/s]
 17%|█▋        | 1188/7175 [15:27<1:05:22,  1.53it/s]
 17%|█▋        | 1189/7175 [15:28<1:05:19,  1.53it/s]
 17%|█▋        | 1190/7175 [15:28<1:05:19,  1.53it/s]
                                                     
{'loss': '3.57', 'grad_norm': '0.2218', 'learning_rate': '0.0006', 'epoch': '4.147'}
+
 17%|█▋        | 1190/7175 [15:28<1:05:19,  1.53it/s]
 17%|█▋        | 1191/7175 [15:29<1:05:24,  1.52it/s]
 17%|█▋        | 1192/7175 [15:30<1:05:21,  1.53it/s]
 17%|█▋        | 1193/7175 [15:30<1:05:19,  1.53it/s]
 17%|█▋        | 1194/7175 [15:31<1:05:18,  1.53it/s]
 17%|█▋        | 1195/7175 [15:32<1:05:21,  1.52it/s]
 17%|█▋        | 1196/7175 [15:32<1:05:24,  1.52it/s]
 17%|█▋        | 1197/7175 [15:33<1:05:21,  1.52it/s]
 17%|█▋        | 1198/7175 [15:34<1:05:18,  1.53it/s]
 17%|█▋        | 1199/7175 [15:34<1:05:17,  1.53it/s]
 17%|█▋        | 1200/7175 [15:35<1:05:18,  1.52it/s]
                                                     
{'loss': '3.569', 'grad_norm': '0.2777', 'learning_rate': '0.0006', 'epoch': '4.182'}
+
 17%|█▋        | 1200/7175 [15:35<1:05:18,  1.52it/s]
 17%|█▋        | 1201/7175 [15:36<1:05:21,  1.52it/s]
 17%|█▋        | 1202/7175 [15:36<1:05:17,  1.52it/s]
 17%|█▋        | 1203/7175 [15:37<1:05:12,  1.53it/s]
 17%|█▋        | 1204/7175 [15:38<1:05:11,  1.53it/s]
 17%|█▋        | 1205/7175 [15:38<1:05:14,  1.53it/s]
 17%|█▋        | 1206/7175 [15:39<1:05:11,  1.53it/s]
 17%|█▋        | 1207/7175 [15:40<1:05:06,  1.53it/s]
 17%|█▋        | 1208/7175 [15:40<1:05:08,  1.53it/s]
 17%|█▋        | 1209/7175 [15:41<1:05:09,  1.53it/s]
 17%|█▋        | 1210/7175 [15:42<1:05:07,  1.53it/s]
                                                     
{'loss': '3.57', 'grad_norm': '0.2495', 'learning_rate': '0.0006', 'epoch': '4.217'}
+
 17%|█▋        | 1210/7175 [15:42<1:05:07,  1.53it/s]
 17%|█▋        | 1211/7175 [15:42<1:05:07,  1.53it/s]
 17%|█▋        | 1212/7175 [15:43<1:05:05,  1.53it/s]
 17%|█▋        | 1213/7175 [15:44<1:05:06,  1.53it/s]
 17%|█▋        | 1214/7175 [15:44<1:05:04,  1.53it/s]
 17%|█▋        | 1215/7175 [15:45<1:05:02,  1.53it/s]
 17%|█▋        | 1216/7175 [15:45<1:05:03,  1.53it/s]
 17%|█▋        | 1217/7175 [15:46<1:05:04,  1.53it/s]
 17%|█▋        | 1218/7175 [15:47<1:05:04,  1.53it/s]
 17%|█▋        | 1219/7175 [15:47<1:05:01,  1.53it/s]
 17%|█▋        | 1220/7175 [15:48<1:04:59,  1.53it/s]
                                                     
{'loss': '3.555', 'grad_norm': '0.2519', 'learning_rate': '0.0006', 'epoch': '4.252'}
+
 17%|█▋        | 1220/7175 [15:48<1:04:59,  1.53it/s]
 17%|█▋        | 1221/7175 [15:49<1:05:04,  1.53it/s]
 17%|█▋        | 1222/7175 [15:49<1:05:01,  1.53it/s]
 17%|█▋        | 1223/7175 [15:50<1:04:59,  1.53it/s]
 17%|█▋        | 1224/7175 [15:51<1:04:57,  1.53it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.36it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.69it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.57it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.03it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.73it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.56it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.44it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.36it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.32it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.28it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.27it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.27it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.24it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.25it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.24it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.24it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.24it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.23it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s]
                                                     
+
                                               
{'eval_loss': '3.819', 'eval_runtime': '3.261', 'eval_samples_per_second': '748.5', 'eval_steps_per_second': '11.96', 'epoch': '4.266'}
+
 17%|█▋        | 1224/7175 [15:54<1:04:57,  1.53it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-1224
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-1224/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-1224/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.90it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.90it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-1224/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-1224/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-1224/tokenizer_config.json
+[eval] epoch 4.266 step 1224 loss 3.8189 (best 3.8189 @ ep 4.266, 0/20 since best)
+
+
                                               
 17%|█▋        | 1225/7175 [15:57<3:42:45,  2.25s/it]
 17%|█▋        | 1226/7175 [15:57<2:55:20,  1.77s/it]
 17%|█▋        | 1227/7175 [15:58<2:22:10,  1.43s/it]
 17%|█▋        | 1228/7175 [15:59<1:58:56,  1.20s/it]
 17%|█▋        | 1229/7175 [15:59<1:42:43,  1.04s/it]
 17%|█▋        | 1230/7175 [16:00<1:31:20,  1.08it/s]
                                                     
{'loss': '3.553', 'grad_norm': '0.3304', 'learning_rate': '0.0006', 'epoch': '4.287'}
+
 17%|█▋        | 1230/7175 [16:00<1:31:20,  1.08it/s]
 17%|█▋        | 1231/7175 [16:01<1:23:25,  1.19it/s]
 17%|█▋        | 1232/7175 [16:01<1:17:49,  1.27it/s]
 17%|█▋        | 1233/7175 [16:02<1:13:58,  1.34it/s]
 17%|█▋        | 1234/7175 [16:03<1:11:11,  1.39it/s]
 17%|█▋        | 1235/7175 [16:03<1:09:16,  1.43it/s]
 17%|█▋        | 1236/7175 [16:04<1:07:54,  1.46it/s]
 17%|█▋        | 1237/7175 [16:05<1:07:00,  1.48it/s]
 17%|█▋        | 1238/7175 [16:05<1:06:19,  1.49it/s]
 17%|█▋        | 1239/7175 [16:06<1:05:49,  1.50it/s]
 17%|█▋        | 1240/7175 [16:07<1:05:29,  1.51it/s]
                                                     
{'loss': '3.563', 'grad_norm': '0.2687', 'learning_rate': '0.0006', 'epoch': '4.322'}
+
 17%|█▋        | 1240/7175 [16:07<1:05:29,  1.51it/s]
 17%|█▋        | 1241/7175 [16:07<1:05:18,  1.51it/s]
 17%|█▋        | 1242/7175 [16:08<1:05:08,  1.52it/s]
 17%|█▋        | 1243/7175 [16:08<1:04:58,  1.52it/s]
 17%|█▋        | 1244/7175 [16:09<1:04:53,  1.52it/s]
 17%|█▋        | 1245/7175 [16:10<1:04:54,  1.52it/s]
 17%|█▋        | 1246/7175 [16:10<1:04:49,  1.52it/s]
 17%|█▋        | 1247/7175 [16:11<1:04:45,  1.53it/s]
 17%|█▋        | 1248/7175 [16:12<1:04:43,  1.53it/s]
 17%|█▋        | 1249/7175 [16:12<1:04:44,  1.53it/s]
 17%|█▋        | 1250/7175 [16:13<1:04:42,  1.53it/s]
                                                     
{'loss': '3.555', 'grad_norm': '0.2837', 'learning_rate': '0.0006', 'epoch': '4.356'}
+
 17%|█▋        | 1250/7175 [16:13<1:04:42,  1.53it/s]
 17%|█▋        | 1251/7175 [16:14<1:04:40,  1.53it/s]
 17%|█▋        | 1252/7175 [16:14<1:04:38,  1.53it/s]
 17%|█▋        | 1253/7175 [16:15<1:04:40,  1.53it/s]
 17%|█▋        | 1254/7175 [16:16<1:04:57,  1.52it/s]
 17%|█▋        | 1255/7175 [16:16<1:04:57,  1.52it/s]
 18%|█▊        | 1256/7175 [16:17<1:04:55,  1.52it/s]
 18%|█▊        | 1257/7175 [16:18<1:04:48,  1.52it/s]
 18%|█▊        | 1258/7175 [16:18<1:04:42,  1.52it/s]
 18%|█▊        | 1259/7175 [16:19<1:04:39,  1.52it/s]
 18%|█▊        | 1260/7175 [16:20<1:04:40,  1.52it/s]
                                                     
{'loss': '3.576', 'grad_norm': '0.3934', 'learning_rate': '0.0006', 'epoch': '4.391'}
+
 18%|█▊        | 1260/7175 [16:20<1:04:40,  1.52it/s]
 18%|█▊        | 1261/7175 [16:20<1:04:39,  1.52it/s]
 18%|█▊        | 1262/7175 [16:21<1:04:37,  1.52it/s]
 18%|█▊        | 1263/7175 [16:22<1:04:33,  1.53it/s]
 18%|█▊        | 1264/7175 [16:22<1:04:34,  1.53it/s]
 18%|█▊        | 1265/7175 [16:23<1:04:33,  1.53it/s]
 18%|█▊        | 1266/7175 [16:24<1:04:32,  1.53it/s]
 18%|█▊        | 1267/7175 [16:24<1:04:28,  1.53it/s]
 18%|█▊        | 1268/7175 [16:25<1:04:29,  1.53it/s]
 18%|█▊        | 1269/7175 [16:26<1:04:28,  1.53it/s]
 18%|█▊        | 1270/7175 [16:26<1:04:28,  1.53it/s]
                                                     
{'loss': '3.556', 'grad_norm': '0.2854', 'learning_rate': '0.0006', 'epoch': '4.426'}
+
 18%|█▊        | 1270/7175 [16:26<1:04:28,  1.53it/s]
 18%|█▊        | 1271/7175 [16:27<1:04:27,  1.53it/s]
 18%|█▊        | 1272/7175 [16:27<1:04:27,  1.53it/s]
 18%|█▊        | 1273/7175 [16:28<1:04:29,  1.53it/s]
 18%|█▊        | 1274/7175 [16:29<1:04:27,  1.53it/s]
 18%|█▊        | 1275/7175 [16:29<1:04:24,  1.53it/s]
 18%|█▊        | 1276/7175 [16:30<1:04:23,  1.53it/s]
 18%|█▊        | 1277/7175 [16:31<1:04:23,  1.53it/s]
 18%|█▊        | 1278/7175 [16:31<1:04:22,  1.53it/s]
 18%|█▊        | 1279/7175 [16:32<1:04:19,  1.53it/s]
 18%|█▊        | 1280/7175 [16:33<1:04:20,  1.53it/s]
                                                     
{'loss': '3.551', 'grad_norm': '0.2771', 'learning_rate': '0.0006', 'epoch': '4.461'}
+
 18%|█▊        | 1280/7175 [16:33<1:04:20,  1.53it/s]
 18%|█▊        | 1281/7175 [16:33<1:04:21,  1.53it/s]
 18%|█▊        | 1282/7175 [16:34<1:04:23,  1.53it/s]
 18%|█▊        | 1283/7175 [16:35<1:04:20,  1.53it/s]
 18%|█▊        | 1284/7175 [16:35<1:04:18,  1.53it/s]
 18%|█▊        | 1285/7175 [16:36<1:04:21,  1.53it/s]
 18%|█▊        | 1286/7175 [16:37<1:04:18,  1.53it/s]
 18%|█▊        | 1287/7175 [16:37<1:04:15,  1.53it/s]
 18%|█▊        | 1288/7175 [16:38<1:04:14,  1.53it/s]
 18%|█▊        | 1289/7175 [16:39<1:04:18,  1.53it/s]
 18%|█▊        | 1290/7175 [16:39<1:04:16,  1.53it/s]
                                                     
{'loss': '3.543', 'grad_norm': '0.2296', 'learning_rate': '0.0006', 'epoch': '4.496'}
+
 18%|█▊        | 1290/7175 [16:39<1:04:16,  1.53it/s]
 18%|█▊        | 1291/7175 [16:40<1:14:36,  1.31it/s]
 18%|█▊        | 1292/7175 [16:41<1:11:26,  1.37it/s]
 18%|█▊        | 1293/7175 [16:42<1:09:14,  1.42it/s]
 18%|█▊        | 1294/7175 [16:42<1:07:42,  1.45it/s]
 18%|█▊        | 1295/7175 [16:43<1:06:41,  1.47it/s]
 18%|█▊        | 1296/7175 [16:44<1:05:52,  1.49it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.34it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.71it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.57it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.05it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.75it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.56it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.46it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.37it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.32it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.30it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.29it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.27it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.26it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.25it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.25it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.24it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.24it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.24it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s]
                                                     
+
                                               
{'eval_loss': '3.788', 'eval_runtime': '3.26', 'eval_samples_per_second': '748.8', 'eval_steps_per_second': '11.96', 'epoch': '4.517'}
+
 18%|█▊        | 1296/7175 [16:47<1:05:52,  1.49it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-1296
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-1296/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-1296/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.96it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.96it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-1296/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-1296/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-1296/tokenizer_config.json
+[eval] epoch 4.517 step 1296 loss 3.7885 (best 3.7885 @ ep 4.517, 0/20 since best)
+
+
                                               
 18%|█▊        | 1297/7175 [16:50<3:41:08,  2.26s/it]
 18%|█▊        | 1298/7175 [16:50<2:54:00,  1.78s/it]
 18%|█▊        | 1299/7175 [16:51<2:21:01,  1.44s/it]
 18%|█▊        | 1300/7175 [16:51<1:57:54,  1.20s/it]
                                                     
{'loss': '3.552', 'grad_norm': '0.26', 'learning_rate': '0.0006', 'epoch': '4.531'}
+
 18%|█▊        | 1300/7175 [16:51<1:57:54,  1.20s/it]
 18%|█▊        | 1301/7175 [16:52<1:41:45,  1.04s/it]
 18%|█▊        | 1302/7175 [16:53<1:30:26,  1.08it/s]
 18%|█▊        | 1303/7175 [16:53<1:22:33,  1.19it/s]
 18%|█▊        | 1304/7175 [16:54<1:17:00,  1.27it/s]
 18%|█▊        | 1305/7175 [16:55<1:13:06,  1.34it/s]
 18%|█▊        | 1306/7175 [16:55<1:10:21,  1.39it/s]
 18%|█▊        | 1307/7175 [16:56<1:08:29,  1.43it/s]
 18%|█▊        | 1308/7175 [16:57<1:07:07,  1.46it/s]
 18%|█▊        | 1309/7175 [16:57<1:06:09,  1.48it/s]
 18%|█▊        | 1310/7175 [16:58<1:05:31,  1.49it/s]
                                                     
{'loss': '3.549', 'grad_norm': '0.2577', 'learning_rate': '0.0006', 'epoch': '4.566'}
+
 18%|█▊        | 1310/7175 [16:58<1:05:31,  1.49it/s]
 18%|█▊        | 1311/7175 [16:59<1:05:08,  1.50it/s]
 18%|█▊        | 1312/7175 [16:59<1:04:46,  1.51it/s]
 18%|█▊        | 1313/7175 [17:00<1:04:30,  1.51it/s]
 18%|█▊        | 1314/7175 [17:01<1:04:20,  1.52it/s]
 18%|█▊        | 1315/7175 [17:01<1:04:17,  1.52it/s]
 18%|█▊        | 1316/7175 [17:02<1:04:09,  1.52it/s]
 18%|█▊        | 1317/7175 [17:03<1:04:02,  1.52it/s]
 18%|█▊        | 1318/7175 [17:03<1:04:00,  1.53it/s]
 18%|█▊        | 1319/7175 [17:04<1:04:02,  1.52it/s]
 18%|█▊        | 1320/7175 [17:05<1:03:59,  1.53it/s]
                                                     
{'loss': '3.55', 'grad_norm': '0.2718', 'learning_rate': '0.0006', 'epoch': '4.601'}
+
 18%|█▊        | 1320/7175 [17:05<1:03:59,  1.53it/s]
 18%|█▊        | 1321/7175 [17:05<1:03:58,  1.52it/s]
 18%|█▊        | 1322/7175 [17:06<1:03:54,  1.53it/s]
 18%|█▊        | 1323/7175 [17:07<1:03:57,  1.53it/s]
 18%|█▊        | 1324/7175 [17:07<1:03:59,  1.52it/s]
 18%|█▊        | 1325/7175 [17:08<1:03:55,  1.53it/s]
 18%|█▊        | 1326/7175 [17:09<1:03:50,  1.53it/s]
 18%|█▊        | 1327/7175 [17:09<1:03:50,  1.53it/s]
 19%|█▊        | 1328/7175 [17:10<1:03:50,  1.53it/s]
 19%|█▊        | 1329/7175 [17:10<1:03:49,  1.53it/s]
 19%|█▊        | 1330/7175 [17:11<1:03:45,  1.53it/s]
                                                     
{'loss': '3.531', 'grad_norm': '0.2401', 'learning_rate': '0.0006', 'epoch': '4.636'}
+
 19%|█▊        | 1330/7175 [17:11<1:03:45,  1.53it/s]
 19%|█▊        | 1331/7175 [17:12<1:04:10,  1.52it/s]
 19%|█▊        | 1332/7175 [17:12<1:04:00,  1.52it/s]
 19%|█▊        | 1333/7175 [17:13<1:03:55,  1.52it/s]
 19%|█▊        | 1334/7175 [17:14<1:03:55,  1.52it/s]
 19%|█▊        | 1335/7175 [17:14<1:03:51,  1.52it/s]
 19%|█▊        | 1336/7175 [17:15<1:03:46,  1.53it/s]
 19%|█▊        | 1337/7175 [17:16<1:03:44,  1.53it/s]
 19%|█▊        | 1338/7175 [17:16<1:03:44,  1.53it/s]
 19%|█▊        | 1339/7175 [17:17<1:03:42,  1.53it/s]
 19%|█▊        | 1340/7175 [17:18<1:03:39,  1.53it/s]
                                                     
{'loss': '3.53', 'grad_norm': '0.2515', 'learning_rate': '0.0006', 'epoch': '4.671'}
+
 19%|█▊        | 1340/7175 [17:18<1:03:39,  1.53it/s]
 19%|█▊        | 1341/7175 [17:18<1:03:41,  1.53it/s]
 19%|█▊        | 1342/7175 [17:19<1:03:43,  1.53it/s]
 19%|█▊        | 1343/7175 [17:20<1:03:40,  1.53it/s]
 19%|█▊        | 1344/7175 [17:20<1:03:38,  1.53it/s]
 19%|█▊        | 1345/7175 [17:21<1:03:38,  1.53it/s]
 19%|█▉        | 1346/7175 [17:22<1:03:41,  1.53it/s]
 19%|█▉        | 1347/7175 [17:22<1:03:38,  1.53it/s]
 19%|█▉        | 1348/7175 [17:23<1:03:36,  1.53it/s]
 19%|█▉        | 1349/7175 [17:24<1:03:36,  1.53it/s]
 19%|█▉        | 1350/7175 [17:24<1:03:38,  1.53it/s]
                                                     
{'loss': '3.513', 'grad_norm': '0.2944', 'learning_rate': '0.0006', 'epoch': '4.706'}
+
 19%|█▉        | 1350/7175 [17:24<1:03:38,  1.53it/s]
 19%|█▉        | 1351/7175 [17:25<1:03:37,  1.53it/s]
 19%|█▉        | 1352/7175 [17:26<1:03:35,  1.53it/s]
 19%|█▉        | 1353/7175 [17:26<1:03:32,  1.53it/s]
 19%|█▉        | 1354/7175 [17:27<1:03:35,  1.53it/s]
 19%|█▉        | 1355/7175 [17:28<1:03:33,  1.53it/s]
 19%|█▉        | 1356/7175 [17:28<1:03:31,  1.53it/s]
 19%|█▉        | 1357/7175 [17:29<1:03:29,  1.53it/s]
 19%|█▉        | 1358/7175 [17:29<1:03:31,  1.53it/s]
 19%|█▉        | 1359/7175 [17:30<1:03:30,  1.53it/s]
 19%|█▉        | 1360/7175 [17:31<1:03:28,  1.53it/s]
                                                     
{'loss': '3.53', 'grad_norm': '0.2187', 'learning_rate': '0.0006', 'epoch': '4.741'}
+
 19%|█▉        | 1360/7175 [17:31<1:03:28,  1.53it/s]
 19%|█▉        | 1361/7175 [17:31<1:03:26,  1.53it/s]
 19%|█▉        | 1362/7175 [17:32<1:03:29,  1.53it/s]
 19%|█▉        | 1363/7175 [17:33<1:03:28,  1.53it/s]
 19%|█▉        | 1364/7175 [17:33<1:03:26,  1.53it/s]
 19%|█▉        | 1365/7175 [17:34<1:03:23,  1.53it/s]
 19%|█▉        | 1366/7175 [17:35<1:03:26,  1.53it/s]
 19%|█▉        | 1367/7175 [17:35<1:03:25,  1.53it/s]
 19%|█▉        | 1368/7175 [17:36<1:03:22,  1.53it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.37it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.73it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.57it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.03it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.74it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.56it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.44it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.37it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.33it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.29it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.28it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.26it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.24it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.23it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.23it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.23it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.23it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.23it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.22it/s]
                                                     
+
                                               
{'eval_loss': '3.765', 'eval_runtime': '3.263', 'eval_samples_per_second': '748', 'eval_steps_per_second': '11.95', 'epoch': '4.769'}
+
 19%|█▉        | 1368/7175 [17:39<1:03:22,  1.53it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.22it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-1368
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-1368/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-1368/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.66it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.66it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-1368/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-1368/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-1368/tokenizer_config.json
+[eval] epoch 4.769 step 1368 loss 3.7646 (best 3.7646 @ ep 4.769, 0/20 since best)
+
+
                                               
 19%|█▉        | 1369/7175 [17:42<3:39:58,  2.27s/it]
 19%|█▉        | 1370/7175 [17:43<2:52:55,  1.79s/it]
                                                     
{'loss': '3.518', 'grad_norm': '0.2701', 'learning_rate': '0.0006', 'epoch': '4.776'}
+
 19%|█▉        | 1370/7175 [17:43<2:52:55,  1.79s/it]
 19%|█▉        | 1371/7175 [17:43<2:20:03,  1.45s/it]
 19%|█▉        | 1372/7175 [17:44<1:57:00,  1.21s/it]
 19%|█▉        | 1373/7175 [17:45<1:40:57,  1.04s/it]
 19%|█▉        | 1374/7175 [17:45<1:29:40,  1.08it/s]
 19%|█▉        | 1375/7175 [17:46<1:21:46,  1.18it/s]
 19%|█▉        | 1376/7175 [17:47<1:16:11,  1.27it/s]
 19%|█▉        | 1377/7175 [17:47<1:12:20,  1.34it/s]
 19%|█▉        | 1378/7175 [17:48<1:09:40,  1.39it/s]
 19%|█▉        | 1379/7175 [17:49<1:07:45,  1.43it/s]
 19%|█▉        | 1380/7175 [17:49<1:06:23,  1.45it/s]
                                                     
{'loss': '3.532', 'grad_norm': '0.2892', 'learning_rate': '0.0006', 'epoch': '4.811'}
+
 19%|█▉        | 1380/7175 [17:49<1:06:23,  1.45it/s]
 19%|█▉        | 1381/7175 [17:50<1:05:26,  1.48it/s]
 19%|█▉        | 1382/7175 [17:51<1:04:50,  1.49it/s]
 19%|█▉        | 1383/7175 [17:51<1:04:19,  1.50it/s]
 19%|█▉        | 1384/7175 [17:52<1:04:00,  1.51it/s]
 19%|█▉        | 1385/7175 [17:53<1:03:42,  1.51it/s]
 19%|█▉        | 1386/7175 [17:53<1:03:36,  1.52it/s]
 19%|█▉        | 1387/7175 [17:54<1:03:27,  1.52it/s]
 19%|█▉        | 1388/7175 [17:55<1:03:22,  1.52it/s]
 19%|█▉        | 1389/7175 [17:55<1:03:15,  1.52it/s]
 19%|█▉        | 1390/7175 [17:56<1:03:15,  1.52it/s]
                                                     
{'loss': '3.518', 'grad_norm': '0.2966', 'learning_rate': '0.0006', 'epoch': '4.846'}
+
 19%|█▉        | 1390/7175 [17:56<1:03:15,  1.52it/s]
 19%|█▉        | 1391/7175 [17:57<1:03:17,  1.52it/s]
 19%|█▉        | 1392/7175 [17:57<1:03:12,  1.52it/s]
 19%|█▉        | 1393/7175 [17:58<1:03:09,  1.53it/s]
 19%|█▉        | 1394/7175 [17:58<1:03:07,  1.53it/s]
 19%|█▉        | 1395/7175 [17:59<1:03:09,  1.53it/s]
 19%|█▉        | 1396/7175 [18:00<1:03:06,  1.53it/s]
 19%|█▉        | 1397/7175 [18:00<1:03:05,  1.53it/s]
 19%|█▉        | 1398/7175 [18:01<1:03:04,  1.53it/s]
 19%|█▉        | 1399/7175 [18:02<1:03:06,  1.53it/s]
 20%|█▉        | 1400/7175 [18:02<1:03:04,  1.53it/s]
                                                     
{'loss': '3.517', 'grad_norm': '0.2216', 'learning_rate': '0.0006', 'epoch': '4.881'}
+
 20%|█▉        | 1400/7175 [18:02<1:03:04,  1.53it/s]
 20%|█▉        | 1401/7175 [18:03<1:13:12,  1.31it/s]
 20%|█▉        | 1402/7175 [18:04<1:10:04,  1.37it/s]
 20%|█▉        | 1403/7175 [18:05<1:07:53,  1.42it/s]
 20%|█▉        | 1404/7175 [18:05<1:06:24,  1.45it/s]
 20%|█▉        | 1405/7175 [18:06<1:05:21,  1.47it/s]
 20%|█▉        | 1406/7175 [18:07<1:04:37,  1.49it/s]
 20%|█▉        | 1407/7175 [18:07<1:04:05,  1.50it/s]
 20%|█▉        | 1408/7175 [18:08<1:03:46,  1.51it/s]
 20%|█▉        | 1409/7175 [18:09<1:13:22,  1.31it/s]
 20%|█▉        | 1410/7175 [18:10<1:23:12,  1.15it/s]
                                                     
{'loss': '3.514', 'grad_norm': '0.2633', 'learning_rate': '0.0006', 'epoch': '4.916'}
+
 20%|█▉        | 1410/7175 [18:10<1:23:12,  1.15it/s]
 20%|█▉        | 1411/7175 [18:11<1:17:08,  1.25it/s]
 20%|█▉        | 1412/7175 [18:11<1:12:50,  1.32it/s]
 20%|█▉        | 1413/7175 [18:12<1:09:49,  1.38it/s]
 20%|█▉        | 1414/7175 [18:13<1:07:44,  1.42it/s]
 20%|█▉        | 1415/7175 [18:13<1:06:15,  1.45it/s]
 20%|█▉        | 1416/7175 [18:14<1:15:18,  1.27it/s]
 20%|█▉        | 1417/7175 [18:15<1:21:25,  1.18it/s]
 20%|█▉        | 1418/7175 [18:16<1:15:47,  1.27it/s]
 20%|█▉        | 1419/7175 [18:17<1:11:51,  1.33it/s]
 20%|█▉        | 1420/7175 [18:17<1:09:09,  1.39it/s]
                                                     
{'loss': '3.521', 'grad_norm': '0.2468', 'learning_rate': '0.0006', 'epoch': '4.951'}
+
 20%|█▉        | 1420/7175 [18:17<1:09:09,  1.39it/s]
 20%|█▉        | 1421/7175 [18:18<1:17:05,  1.24it/s]
 20%|█▉        | 1422/7175 [18:19<1:12:46,  1.32it/s]
 20%|█▉        | 1423/7175 [18:20<1:09:42,  1.38it/s]
 20%|█▉        | 1424/7175 [18:20<1:07:39,  1.42it/s]
 20%|█▉        | 1425/7175 [18:21<1:06:11,  1.45it/s]
 20%|█▉        | 1426/7175 [18:22<1:05:08,  1.47it/s]
 20%|█▉        | 1427/7175 [18:22<1:04:22,  1.49it/s]
 20%|█▉        | 1428/7175 [18:23<1:03:54,  1.50it/s]
 20%|█▉        | 1429/7175 [18:24<1:03:33,  1.51it/s]
 20%|█▉        | 1430/7175 [18:24<1:03:17,  1.51it/s]
                                                     
{'loss': '3.505', 'grad_norm': '0.2957', 'learning_rate': '0.0006', 'epoch': '4.986'}
+
 20%|█▉        | 1430/7175 [18:24<1:03:17,  1.51it/s]
 20%|█▉        | 1431/7175 [18:25<1:03:05,  1.52it/s]
 20%|█▉        | 1432/7175 [18:26<1:03:07,  1.52it/s]
 20%|█▉        | 1433/7175 [18:26<1:03:00,  1.52it/s]
 20%|█▉        | 1434/7175 [18:27<1:02:54,  1.52it/s]
 20%|██        | 1436/7175 [18:28<52:08,  1.83it/s]  
 20%|██        | 1437/7175 [18:28<54:46,  1.75it/s]
 20%|██        | 1438/7175 [18:29<56:49,  1.68it/s]
 20%|██        | 1439/7175 [18:30<58:24,  1.64it/s]
 20%|██        | 1440/7175 [18:30<59:33,  1.61it/s]
                                                   
{'loss': '3.471', 'grad_norm': '0.3555', 'learning_rate': '0.0006', 'epoch': '5.017'}
+
 20%|██        | 1440/7175 [18:30<59:33,  1.61it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.52it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.79it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.61it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.09it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.79it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.58it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.46it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.36it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.30it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.29it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.28it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.26it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.25it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.25it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.24it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.23it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.23it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.23it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s]
                                                   
+
                                               
{'eval_loss': '3.766', 'eval_runtime': '3.261', 'eval_samples_per_second': '748.4', 'eval_steps_per_second': '11.96', 'epoch': '5.017'}
+
 20%|██        | 1440/7175 [18:34<59:33,  1.61it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s][eval] epoch 5.017 step 1440 loss 3.7661 (best 3.7646 @ ep 4.769, 1/20 since best)
+
+
                                               
 20%|██        | 1441/7175 [18:34<2:29:34,  1.57s/it]
 20%|██        | 1442/7175 [18:35<2:04:19,  1.30s/it]
 20%|██        | 1443/7175 [18:36<1:46:13,  1.11s/it]
 20%|██        | 1444/7175 [18:36<1:33:20,  1.02it/s]
 20%|██        | 1445/7175 [18:37<1:24:09,  1.13it/s]
 20%|██        | 1446/7175 [18:37<1:17:44,  1.23it/s]
 20%|██        | 1447/7175 [18:38<1:13:14,  1.30it/s]
 20%|██        | 1448/7175 [18:39<1:10:00,  1.36it/s]
 20%|██        | 1449/7175 [18:39<1:07:43,  1.41it/s]
 20%|██        | 1450/7175 [18:40<1:06:10,  1.44it/s]
                                                     
{'loss': '3.449', 'grad_norm': '0.3183', 'learning_rate': '0.0006', 'epoch': '5.052'}
+
 20%|██        | 1450/7175 [18:40<1:06:10,  1.44it/s]
 20%|██        | 1451/7175 [18:41<1:05:07,  1.46it/s]
 20%|██        | 1452/7175 [18:41<1:04:18,  1.48it/s]
 20%|██        | 1453/7175 [18:42<1:03:45,  1.50it/s]
 20%|██        | 1454/7175 [18:43<1:03:19,  1.51it/s]
 20%|██        | 1455/7175 [18:43<1:03:07,  1.51it/s]
 20%|██        | 1456/7175 [18:44<1:02:54,  1.52it/s]
 20%|██        | 1457/7175 [18:45<1:02:46,  1.52it/s]
 20%|██        | 1458/7175 [18:45<1:02:36,  1.52it/s]
 20%|██        | 1459/7175 [18:46<1:02:35,  1.52it/s]
 20%|██        | 1460/7175 [18:47<1:02:39,  1.52it/s]
                                                     
{'loss': '3.46', 'grad_norm': '0.2501', 'learning_rate': '0.0006', 'epoch': '5.087'}
+
 20%|██        | 1460/7175 [18:47<1:02:39,  1.52it/s]
 20%|██        | 1461/7175 [18:47<1:02:35,  1.52it/s]
 20%|██        | 1462/7175 [18:48<1:02:30,  1.52it/s]
 20%|██        | 1463/7175 [18:49<1:02:29,  1.52it/s]
 20%|██        | 1464/7175 [18:49<1:02:25,  1.52it/s]
 20%|██        | 1465/7175 [18:50<1:02:26,  1.52it/s]
 20%|██        | 1466/7175 [18:51<1:02:24,  1.52it/s]
 20%|██        | 1467/7175 [18:51<1:02:19,  1.53it/s]
 20%|██        | 1468/7175 [18:52<1:02:18,  1.53it/s]
 20%|██        | 1469/7175 [18:53<1:02:21,  1.52it/s]
 20%|██        | 1470/7175 [18:53<1:02:18,  1.53it/s]
                                                     
{'loss': '3.435', 'grad_norm': '0.2095', 'learning_rate': '0.0006', 'epoch': '5.122'}
+
 20%|██        | 1470/7175 [18:53<1:02:18,  1.53it/s]
 21%|██        | 1471/7175 [18:54<1:02:20,  1.52it/s]
 21%|██        | 1472/7175 [18:55<1:02:17,  1.53it/s]
 21%|██        | 1473/7175 [18:55<1:02:17,  1.53it/s]
 21%|██        | 1474/7175 [18:56<1:02:18,  1.52it/s]
 21%|██        | 1475/7175 [18:57<1:02:16,  1.53it/s]
 21%|██        | 1476/7175 [18:57<1:02:13,  1.53it/s]
 21%|██        | 1477/7175 [18:58<1:02:12,  1.53it/s]
 21%|██        | 1478/7175 [18:58<1:02:15,  1.53it/s]
 21%|██        | 1479/7175 [18:59<1:02:13,  1.53it/s]
 21%|██        | 1480/7175 [19:00<1:02:11,  1.53it/s]
                                                     
{'loss': '3.438', 'grad_norm': '0.2187', 'learning_rate': '0.0006', 'epoch': '5.157'}
+
 21%|██        | 1480/7175 [19:00<1:02:11,  1.53it/s]
 21%|██        | 1481/7175 [19:00<1:02:10,  1.53it/s]
 21%|██        | 1482/7175 [19:01<1:02:12,  1.53it/s]
 21%|██        | 1483/7175 [19:02<1:02:10,  1.53it/s]
 21%|██        | 1484/7175 [19:02<1:02:09,  1.53it/s]
 21%|██        | 1485/7175 [19:03<1:02:07,  1.53it/s]
 21%|██        | 1486/7175 [19:04<1:02:09,  1.53it/s]
 21%|██        | 1487/7175 [19:04<1:02:08,  1.53it/s]
 21%|██        | 1488/7175 [19:05<1:02:07,  1.53it/s]
 21%|██        | 1489/7175 [19:06<1:02:04,  1.53it/s]
 21%|██        | 1490/7175 [19:06<1:02:05,  1.53it/s]
                                                     
{'loss': '3.451', 'grad_norm': '0.2344', 'learning_rate': '0.0006', 'epoch': '5.192'}
+
 21%|██        | 1490/7175 [19:06<1:02:05,  1.53it/s]
 21%|██        | 1491/7175 [19:07<1:02:08,  1.52it/s]
 21%|██        | 1492/7175 [19:08<1:02:03,  1.53it/s]
 21%|██        | 1493/7175 [19:08<1:02:00,  1.53it/s]
 21%|██        | 1494/7175 [19:09<1:01:59,  1.53it/s]
 21%|██        | 1495/7175 [19:10<1:02:01,  1.53it/s]
 21%|██        | 1496/7175 [19:10<1:02:00,  1.53it/s]
 21%|██        | 1497/7175 [19:11<1:01:58,  1.53it/s]
 21%|██        | 1498/7175 [19:12<1:01:58,  1.53it/s]
 21%|██        | 1499/7175 [19:12<1:02:02,  1.52it/s]
 21%|██        | 1500/7175 [19:13<1:02:00,  1.53it/s]
                                                     
{'loss': '3.454', 'grad_norm': '0.3337', 'learning_rate': '0.0006', 'epoch': '5.227'}
+
 21%|██        | 1500/7175 [19:13<1:02:00,  1.53it/s]
 21%|██        | 1501/7175 [19:14<1:02:02,  1.52it/s]
 21%|██        | 1502/7175 [19:14<1:01:59,  1.53it/s]
 21%|██        | 1503/7175 [19:15<1:02:00,  1.52it/s]
 21%|██        | 1504/7175 [19:16<1:02:01,  1.52it/s]
 21%|██        | 1505/7175 [19:16<1:01:57,  1.53it/s]
 21%|██        | 1506/7175 [19:17<1:01:56,  1.53it/s]
 21%|██        | 1507/7175 [19:17<1:01:54,  1.53it/s]
 21%|██        | 1508/7175 [19:18<1:01:55,  1.53it/s]
 21%|██        | 1509/7175 [19:19<1:01:53,  1.53it/s]
 21%|██        | 1510/7175 [19:19<1:01:54,  1.53it/s]
                                                     
{'loss': '3.452', 'grad_norm': '0.2838', 'learning_rate': '0.0006', 'epoch': '5.262'}
+
 21%|██        | 1510/7175 [19:19<1:01:54,  1.53it/s]
 21%|██        | 1511/7175 [19:20<1:01:52,  1.53it/s]
 21%|██        | 1512/7175 [19:21<1:01:52,  1.53it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.65it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.83it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.63it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.05it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.75it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.54it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.45it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.36it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.33it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.28it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.28it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.27it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.25it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.24it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.25it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.24it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.24it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.23it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s]
                                                     
+
                                               
{'eval_loss': '3.727', 'eval_runtime': '3.263', 'eval_samples_per_second': '748.1', 'eval_steps_per_second': '11.95', 'epoch': '5.269'}
+
 21%|██        | 1512/7175 [19:24<1:01:52,  1.53it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-1512
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-1512/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-1512/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.82it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.82it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-1512/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-1512/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-1512/tokenizer_config.json
+[eval] epoch 5.269 step 1512 loss 3.7274 (best 3.7274 @ ep 5.269, 0/20 since best)
+
+
                                               
 21%|██        | 1513/7175 [19:27<3:33:10,  2.26s/it]
 21%|██        | 1514/7175 [19:27<2:47:45,  1.78s/it]
 21%|██        | 1515/7175 [19:28<2:15:53,  1.44s/it]
 21%|██        | 1516/7175 [19:29<1:53:37,  1.20s/it]
 21%|██        | 1517/7175 [19:29<1:38:03,  1.04s/it]
 21%|██        | 1518/7175 [19:30<1:27:10,  1.08it/s]
 21%|██        | 1519/7175 [19:31<1:19:31,  1.19it/s]
 21%|██        | 1520/7175 [19:31<1:14:09,  1.27it/s]
                                                     
{'loss': '3.45', 'grad_norm': '0.2553', 'learning_rate': '0.0006', 'epoch': '5.297'}
+
 21%|██        | 1520/7175 [19:31<1:14:09,  1.27it/s]
 21%|██        | 1521/7175 [19:32<1:10:26,  1.34it/s]
 21%|██        | 1522/7175 [19:33<1:07:50,  1.39it/s]
 21%|██        | 1523/7175 [19:33<1:05:58,  1.43it/s]
 21%|██        | 1524/7175 [19:34<1:04:39,  1.46it/s]
 21%|██▏       | 1525/7175 [19:35<1:03:44,  1.48it/s]
 21%|██▏       | 1526/7175 [19:35<1:03:06,  1.49it/s]
 21%|██▏       | 1527/7175 [19:36<1:02:39,  1.50it/s]
 21%|██▏       | 1528/7175 [19:37<1:02:20,  1.51it/s]
 21%|██▏       | 1529/7175 [19:37<1:02:08,  1.51it/s]
 21%|██▏       | 1530/7175 [19:38<1:01:58,  1.52it/s]
                                                     
{'loss': '3.451', 'grad_norm': '0.2812', 'learning_rate': '0.0006', 'epoch': '5.332'}
+
 21%|██▏       | 1530/7175 [19:38<1:01:58,  1.52it/s]
 21%|██▏       | 1531/7175 [19:39<1:01:52,  1.52it/s]
 21%|██▏       | 1532/7175 [19:39<1:01:45,  1.52it/s]
 21%|██▏       | 1533/7175 [19:40<1:01:43,  1.52it/s]
 21%|██▏       | 1534/7175 [19:41<1:01:41,  1.52it/s]
 21%|██▏       | 1535/7175 [19:41<1:01:37,  1.53it/s]
 21%|██▏       | 1536/7175 [19:42<1:01:35,  1.53it/s]
 21%|██▏       | 1537/7175 [19:42<1:01:34,  1.53it/s]
 21%|██▏       | 1538/7175 [19:43<1:01:34,  1.53it/s]
 21%|██▏       | 1539/7175 [19:44<1:01:32,  1.53it/s]
 21%|██▏       | 1540/7175 [19:44<1:01:30,  1.53it/s]
                                                     
{'loss': '3.44', 'grad_norm': '0.2457', 'learning_rate': '0.0006', 'epoch': '5.367'}
+
 21%|██▏       | 1540/7175 [19:44<1:01:30,  1.53it/s]
 21%|██▏       | 1541/7175 [19:45<1:01:30,  1.53it/s]
 21%|██▏       | 1542/7175 [19:46<1:01:31,  1.53it/s]
 22%|██▏       | 1543/7175 [19:46<1:01:30,  1.53it/s]
 22%|██▏       | 1544/7175 [19:47<1:01:27,  1.53it/s]
 22%|██▏       | 1545/7175 [19:48<1:01:27,  1.53it/s]
 22%|██▏       | 1546/7175 [19:48<1:01:25,  1.53it/s]
 22%|██▏       | 1547/7175 [19:49<1:01:24,  1.53it/s]
 22%|██▏       | 1548/7175 [19:50<1:01:22,  1.53it/s]
 22%|██▏       | 1549/7175 [19:50<1:01:25,  1.53it/s]
 22%|██▏       | 1550/7175 [19:51<1:01:24,  1.53it/s]
                                                     
{'loss': '3.435', 'grad_norm': '0.2076', 'learning_rate': '0.0006', 'epoch': '5.402'}
+
 22%|██▏       | 1550/7175 [19:51<1:01:24,  1.53it/s]
 22%|██▏       | 1551/7175 [19:52<1:01:23,  1.53it/s]
 22%|██▏       | 1552/7175 [19:52<1:01:23,  1.53it/s]
 22%|██▏       | 1553/7175 [19:53<1:01:22,  1.53it/s]
 22%|██▏       | 1554/7175 [19:54<1:01:23,  1.53it/s]
 22%|██▏       | 1555/7175 [19:54<1:01:20,  1.53it/s]
 22%|██▏       | 1556/7175 [19:55<1:01:20,  1.53it/s]
 22%|██▏       | 1557/7175 [19:56<1:01:21,  1.53it/s]
 22%|██▏       | 1558/7175 [19:56<1:01:22,  1.53it/s]
 22%|██▏       | 1559/7175 [19:57<1:01:19,  1.53it/s]
 22%|██▏       | 1560/7175 [19:58<1:01:16,  1.53it/s]
                                                     
{'loss': '3.439', 'grad_norm': '0.3259', 'learning_rate': '0.0006', 'epoch': '5.437'}
+
 22%|██▏       | 1560/7175 [19:58<1:01:16,  1.53it/s]
 22%|██▏       | 1561/7175 [19:58<1:01:17,  1.53it/s]
 22%|██▏       | 1562/7175 [19:59<1:01:17,  1.53it/s]
 22%|██▏       | 1563/7175 [20:00<1:01:16,  1.53it/s]
 22%|██▏       | 1564/7175 [20:00<1:01:13,  1.53it/s]
 22%|██▏       | 1565/7175 [20:01<1:01:14,  1.53it/s]
 22%|██▏       | 1566/7175 [20:01<1:01:13,  1.53it/s]
 22%|██▏       | 1567/7175 [20:02<1:01:12,  1.53it/s]
 22%|██▏       | 1568/7175 [20:03<1:01:07,  1.53it/s]
 22%|██▏       | 1569/7175 [20:03<1:01:10,  1.53it/s]
 22%|██▏       | 1570/7175 [20:04<1:01:09,  1.53it/s]
                                                     
{'loss': '3.442', 'grad_norm': '0.2695', 'learning_rate': '0.0006', 'epoch': '5.472'}
+
 22%|██▏       | 1570/7175 [20:04<1:01:09,  1.53it/s]
 22%|██▏       | 1571/7175 [20:05<1:01:09,  1.53it/s]
 22%|██▏       | 1572/7175 [20:05<1:01:07,  1.53it/s]
 22%|██▏       | 1573/7175 [20:06<1:01:09,  1.53it/s]
 22%|██▏       | 1574/7175 [20:07<1:01:08,  1.53it/s]
 22%|██▏       | 1575/7175 [20:07<1:01:05,  1.53it/s]
 22%|██▏       | 1576/7175 [20:08<1:01:05,  1.53it/s]
 22%|██▏       | 1577/7175 [20:09<1:01:08,  1.53it/s]
 22%|██▏       | 1578/7175 [20:09<1:01:06,  1.53it/s]
 22%|██▏       | 1579/7175 [20:10<1:01:03,  1.53it/s]
 22%|██▏       | 1580/7175 [20:11<1:01:02,  1.53it/s]
                                                     
{'loss': '3.446', 'grad_norm': '0.2429', 'learning_rate': '0.0006', 'epoch': '5.507'}
+
 22%|██▏       | 1580/7175 [20:11<1:01:02,  1.53it/s]
 22%|██▏       | 1581/7175 [20:11<1:01:07,  1.53it/s]
 22%|██▏       | 1582/7175 [20:12<1:01:04,  1.53it/s]
 22%|██▏       | 1583/7175 [20:13<1:01:01,  1.53it/s]
 22%|██▏       | 1584/7175 [20:13<1:01:00,  1.53it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.38it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.72it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.57it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.04it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.74it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.56it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.41it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.37it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.32it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.29it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.26it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.27it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.25it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.25it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.25it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.24it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.22it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.23it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.21it/s]
                                                     
+
                                               
{'eval_loss': '3.708', 'eval_runtime': '3.263', 'eval_samples_per_second': '748.2', 'eval_steps_per_second': '11.95', 'epoch': '5.521'}
+
 22%|██▏       | 1584/7175 [20:17<1:01:00,  1.53it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.21it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-1584
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-1584/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-1584/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.87it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.87it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-1584/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-1584/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-1584/tokenizer_config.json
+[eval] epoch 5.521 step 1584 loss 3.7075 (best 3.7075 @ ep 5.521, 0/20 since best)
+
+
                                               
 22%|██▏       | 1585/7175 [20:19<3:29:52,  2.25s/it]
 22%|██▏       | 1586/7175 [20:20<2:45:11,  1.77s/it]
 22%|██▏       | 1587/7175 [20:21<2:13:52,  1.44s/it]
 22%|██▏       | 1588/7175 [20:21<1:51:57,  1.20s/it]
 22%|██▏       | 1589/7175 [20:22<1:36:37,  1.04s/it]
 22%|██▏       | 1590/7175 [20:23<1:25:55,  1.08it/s]
                                                     
{'loss': '3.43', 'grad_norm': '0.2379', 'learning_rate': '0.0006', 'epoch': '5.542'}
+
 22%|██▏       | 1590/7175 [20:23<1:25:55,  1.08it/s]
 22%|██▏       | 1591/7175 [20:23<1:18:27,  1.19it/s]
 22%|██▏       | 1592/7175 [20:24<1:13:10,  1.27it/s]
 22%|██▏       | 1593/7175 [20:24<1:09:29,  1.34it/s]
 22%|██▏       | 1594/7175 [20:25<1:06:56,  1.39it/s]
 22%|██▏       | 1595/7175 [20:26<1:05:06,  1.43it/s]
 22%|██▏       | 1596/7175 [20:26<1:03:47,  1.46it/s]
 22%|██▏       | 1597/7175 [20:27<1:02:57,  1.48it/s]
 22%|██▏       | 1598/7175 [20:28<1:02:21,  1.49it/s]
 22%|██▏       | 1599/7175 [20:28<1:01:55,  1.50it/s]
 22%|██▏       | 1600/7175 [20:29<1:01:33,  1.51it/s]
                                                     
{'loss': '3.438', 'grad_norm': '0.2448', 'learning_rate': '0.0006', 'epoch': '5.577'}
+
 22%|██▏       | 1600/7175 [20:29<1:01:33,  1.51it/s]
 22%|██▏       | 1601/7175 [20:30<1:01:22,  1.51it/s]
 22%|██▏       | 1602/7175 [20:30<1:01:17,  1.52it/s]
 22%|██▏       | 1603/7175 [20:31<1:01:05,  1.52it/s]
 22%|██▏       | 1604/7175 [20:32<1:00:58,  1.52it/s]
 22%|██▏       | 1605/7175 [20:32<1:00:54,  1.52it/s]
 22%|██▏       | 1606/7175 [20:33<1:00:55,  1.52it/s]
 22%|██▏       | 1607/7175 [20:34<1:00:52,  1.52it/s]
 22%|██▏       | 1608/7175 [20:34<1:00:47,  1.53it/s]
 22%|██▏       | 1609/7175 [20:35<1:00:44,  1.53it/s]
 22%|██▏       | 1610/7175 [20:36<1:00:49,  1.52it/s]
                                                     
{'loss': '3.435', 'grad_norm': '0.2461', 'learning_rate': '0.0006', 'epoch': '5.612'}
+
 22%|██▏       | 1610/7175 [20:36<1:00:49,  1.52it/s]
 22%|██▏       | 1611/7175 [20:36<1:00:49,  1.52it/s]
 22%|██▏       | 1612/7175 [20:37<1:00:46,  1.53it/s]
 22%|██▏       | 1613/7175 [20:38<1:00:41,  1.53it/s]
 22%|██▏       | 1614/7175 [20:38<1:00:44,  1.53it/s]
 23%|██▎       | 1615/7175 [20:39<1:00:43,  1.53it/s]
 23%|██▎       | 1616/7175 [20:40<1:00:42,  1.53it/s]
 23%|██▎       | 1617/7175 [20:40<1:00:39,  1.53it/s]
 23%|██▎       | 1618/7175 [20:41<1:00:41,  1.53it/s]
 23%|██▎       | 1619/7175 [20:42<1:00:47,  1.52it/s]
 23%|██▎       | 1620/7175 [20:42<1:00:44,  1.52it/s]
                                                     
{'loss': '3.409', 'grad_norm': '0.2617', 'learning_rate': '0.0006', 'epoch': '5.647'}
+
 23%|██▎       | 1620/7175 [20:42<1:00:44,  1.52it/s]
 23%|██▎       | 1621/7175 [20:43<1:00:43,  1.52it/s]
 23%|██▎       | 1622/7175 [20:43<1:00:37,  1.53it/s]
 23%|██▎       | 1623/7175 [20:44<1:00:39,  1.53it/s]
 23%|██▎       | 1624/7175 [20:45<1:00:39,  1.53it/s]
 23%|██▎       | 1625/7175 [20:45<1:00:37,  1.53it/s]
 23%|██▎       | 1626/7175 [20:46<1:00:33,  1.53it/s]
 23%|██▎       | 1627/7175 [20:47<1:00:37,  1.53it/s]
 23%|██▎       | 1628/7175 [20:47<1:00:38,  1.52it/s]
 23%|██▎       | 1629/7175 [20:48<1:00:35,  1.53it/s]
 23%|██▎       | 1630/7175 [20:49<1:00:30,  1.53it/s]
                                                     
{'loss': '3.43', 'grad_norm': '0.2446', 'learning_rate': '0.0006', 'epoch': '5.682'}
+
 23%|██▎       | 1630/7175 [20:49<1:00:30,  1.53it/s]
 23%|██▎       | 1631/7175 [20:49<1:00:31,  1.53it/s]
 23%|██▎       | 1632/7175 [20:50<1:00:32,  1.53it/s]
 23%|██▎       | 1633/7175 [20:51<1:00:31,  1.53it/s]
 23%|██▎       | 1634/7175 [20:51<1:00:26,  1.53it/s]
 23%|██▎       | 1635/7175 [20:52<1:00:26,  1.53it/s]
 23%|██▎       | 1636/7175 [20:53<1:00:29,  1.53it/s]
 23%|██▎       | 1637/7175 [20:53<1:00:28,  1.53it/s]
 23%|██▎       | 1638/7175 [20:54<1:00:24,  1.53it/s]
 23%|██▎       | 1639/7175 [20:55<1:00:25,  1.53it/s]
 23%|██▎       | 1640/7175 [20:55<1:00:25,  1.53it/s]
                                                     
{'loss': '3.439', 'grad_norm': '0.2636', 'learning_rate': '0.0006', 'epoch': '5.716'}
+
 23%|██▎       | 1640/7175 [20:55<1:00:25,  1.53it/s]
 23%|██▎       | 1641/7175 [20:56<1:00:25,  1.53it/s]
 23%|██▎       | 1642/7175 [20:57<1:00:22,  1.53it/s]
 23%|██▎       | 1643/7175 [20:57<1:00:23,  1.53it/s]
 23%|██▎       | 1644/7175 [20:58<1:00:25,  1.53it/s]
 23%|██▎       | 1645/7175 [20:59<1:00:23,  1.53it/s]
 23%|██▎       | 1646/7175 [20:59<1:00:19,  1.53it/s]
 23%|██▎       | 1647/7175 [21:00<1:00:19,  1.53it/s]
 23%|██▎       | 1648/7175 [21:01<1:00:22,  1.53it/s]
 23%|██▎       | 1649/7175 [21:01<1:00:22,  1.53it/s]
 23%|██▎       | 1650/7175 [21:02<1:00:18,  1.53it/s]
                                                     
{'loss': '3.435', 'grad_norm': '0.2911', 'learning_rate': '0.0006', 'epoch': '5.751'}
+
 23%|██▎       | 1650/7175 [21:02<1:00:18,  1.53it/s]
 23%|██▎       | 1651/7175 [21:02<1:00:18,  1.53it/s]
 23%|██▎       | 1652/7175 [21:03<1:00:21,  1.53it/s]
 23%|██▎       | 1653/7175 [21:04<1:00:19,  1.53it/s]
 23%|██▎       | 1654/7175 [21:04<1:00:17,  1.53it/s]
 23%|██▎       | 1655/7175 [21:05<1:00:14,  1.53it/s]
 23%|██▎       | 1656/7175 [21:06<1:00:17,  1.53it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.66it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.82it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.62it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.08it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.75it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.57it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.45it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.37it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.32it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.26it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.27it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.26it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.25it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.25it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.24it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.23it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.23it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.23it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s]
                                                     
+
                                               
{'eval_loss': '3.688', 'eval_runtime': '3.248', 'eval_samples_per_second': '751.6', 'eval_steps_per_second': '12.01', 'epoch': '5.772'}
+
 23%|██▎       | 1656/7175 [21:09<1:00:17,  1.53it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-1656
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-1656/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-1656/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.85it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.85it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-1656/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-1656/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-1656/tokenizer_config.json
+[eval] epoch 5.772 step 1656 loss 3.6883 (best 3.6883 @ ep 5.772, 0/20 since best)
+
+
                                               
 23%|██▎       | 1657/7175 [21:12<3:27:25,  2.26s/it]
 23%|██▎       | 1658/7175 [21:12<2:43:12,  1.77s/it]
 23%|██▎       | 1659/7175 [21:13<2:12:16,  1.44s/it]
 23%|██▎       | 1660/7175 [21:14<1:50:37,  1.20s/it]
                                                     
{'loss': '3.427', 'grad_norm': '0.2297', 'learning_rate': '0.0006', 'epoch': '5.786'}
+
 23%|██▎       | 1660/7175 [21:14<1:50:37,  1.20s/it]
 23%|██▎       | 1661/7175 [21:14<1:35:28,  1.04s/it]
 23%|██▎       | 1662/7175 [21:15<1:24:52,  1.08it/s]
 23%|██▎       | 1663/7175 [21:16<1:17:27,  1.19it/s]
 23%|██▎       | 1664/7175 [21:16<1:12:15,  1.27it/s]
 23%|██▎       | 1665/7175 [21:17<1:08:35,  1.34it/s]
 23%|██▎       | 1666/7175 [21:18<1:06:03,  1.39it/s]
 23%|██▎       | 1667/7175 [21:18<1:04:17,  1.43it/s]
 23%|██▎       | 1668/7175 [21:19<1:03:02,  1.46it/s]
 23%|██▎       | 1669/7175 [21:20<1:02:07,  1.48it/s]
 23%|██▎       | 1670/7175 [21:20<1:01:30,  1.49it/s]
                                                     
{'loss': '3.416', 'grad_norm': '0.2403', 'learning_rate': '0.0006', 'epoch': '5.821'}
+
 23%|██▎       | 1670/7175 [21:20<1:01:30,  1.49it/s]
 23%|██▎       | 1671/7175 [21:21<1:01:08,  1.50it/s]
 23%|██▎       | 1672/7175 [21:22<1:00:48,  1.51it/s]
 23%|██▎       | 1673/7175 [21:22<1:00:31,  1.51it/s]
 23%|██▎       | 1674/7175 [21:23<1:00:22,  1.52it/s]
 23%|██▎       | 1675/7175 [21:24<1:00:19,  1.52it/s]
 23%|██▎       | 1676/7175 [21:24<1:00:14,  1.52it/s]
 23%|██▎       | 1677/7175 [21:25<1:00:07,  1.52it/s]
 23%|██▎       | 1678/7175 [21:25<1:00:04,  1.53it/s]
 23%|██▎       | 1679/7175 [21:26<1:00:04,  1.52it/s]
 23%|██▎       | 1680/7175 [21:27<1:00:02,  1.53it/s]
                                                     
{'loss': '3.429', 'grad_norm': '0.2596', 'learning_rate': '0.0006', 'epoch': '5.856'}
+
 23%|██▎       | 1680/7175 [21:27<1:00:02,  1.53it/s]
 23%|██▎       | 1681/7175 [21:27<59:59,  1.53it/s]  
 23%|██▎       | 1682/7175 [21:28<59:57,  1.53it/s]
 23%|██▎       | 1683/7175 [21:29<59:59,  1.53it/s]
 23%|██▎       | 1684/7175 [21:29<59:58,  1.53it/s]
 23%|██▎       | 1685/7175 [21:30<59:54,  1.53it/s]
 23%|██▎       | 1686/7175 [21:31<59:53,  1.53it/s]
 24%|██▎       | 1687/7175 [21:31<59:55,  1.53it/s]
 24%|██▎       | 1688/7175 [21:32<59:54,  1.53it/s]
 24%|██▎       | 1689/7175 [21:33<59:50,  1.53it/s]
 24%|██▎       | 1690/7175 [21:33<59:51,  1.53it/s]
                                                   
{'loss': '3.425', 'grad_norm': '0.2504', 'learning_rate': '0.0006', 'epoch': '5.891'}
+
 24%|██▎       | 1690/7175 [21:33<59:51,  1.53it/s]
 24%|██▎       | 1691/7175 [21:34<59:54,  1.53it/s]
 24%|██▎       | 1692/7175 [21:35<59:52,  1.53it/s]
 24%|██▎       | 1693/7175 [21:35<59:48,  1.53it/s]
 24%|██▎       | 1694/7175 [21:36<59:48,  1.53it/s]
 24%|██▎       | 1695/7175 [21:37<59:50,  1.53it/s]
 24%|██▎       | 1696/7175 [21:37<59:49,  1.53it/s]
 24%|██▎       | 1697/7175 [21:38<59:45,  1.53it/s]
 24%|██▎       | 1698/7175 [21:39<59:46,  1.53it/s]
 24%|██▎       | 1699/7175 [21:39<59:46,  1.53it/s]
 24%|██▎       | 1700/7175 [21:40<59:46,  1.53it/s]
                                                   
{'loss': '3.43', 'grad_norm': '0.2312', 'learning_rate': '0.0006', 'epoch': '5.926'}
+
 24%|██▎       | 1700/7175 [21:40<59:46,  1.53it/s]
 24%|██▎       | 1701/7175 [21:41<59:44,  1.53it/s]
 24%|██▎       | 1702/7175 [21:41<59:44,  1.53it/s]
 24%|██▎       | 1703/7175 [21:42<59:43,  1.53it/s]
 24%|██▎       | 1704/7175 [21:43<59:42,  1.53it/s]
 24%|██▍       | 1705/7175 [21:43<59:40,  1.53it/s]
 24%|██▍       | 1706/7175 [21:44<59:42,  1.53it/s]
 24%|██▍       | 1707/7175 [21:44<59:42,  1.53it/s]
 24%|██▍       | 1708/7175 [21:45<59:41,  1.53it/s]
 24%|██▍       | 1709/7175 [21:46<59:39,  1.53it/s]
 24%|██▍       | 1710/7175 [21:46<59:39,  1.53it/s]
                                                   
{'loss': '3.421', 'grad_norm': '0.2168', 'learning_rate': '0.0006', 'epoch': '5.961'}
+
 24%|██▍       | 1710/7175 [21:46<59:39,  1.53it/s]
 24%|██▍       | 1711/7175 [21:47<59:41,  1.53it/s]
 24%|██▍       | 1712/7175 [21:48<59:39,  1.53it/s]
 24%|██▍       | 1713/7175 [21:48<59:36,  1.53it/s]
 24%|██▍       | 1714/7175 [21:49<59:37,  1.53it/s]
 24%|██▍       | 1715/7175 [21:50<59:37,  1.53it/s]
 24%|██▍       | 1716/7175 [21:50<59:35,  1.53it/s]
 24%|██▍       | 1717/7175 [21:51<59:33,  1.53it/s]
 24%|██▍       | 1718/7175 [21:52<59:33,  1.53it/s]
 24%|██▍       | 1719/7175 [21:52<59:43,  1.52it/s]
 24%|██▍       | 1720/7175 [21:53<59:41,  1.52it/s]
                                                   
{'loss': '3.411', 'grad_norm': '0.281', 'learning_rate': '0.0006', 'epoch': '5.996'}
+
 24%|██▍       | 1720/7175 [21:53<59:41,  1.52it/s]
 24%|██▍       | 1721/7175 [21:54<59:37,  1.52it/s]
 24%|██▍       | 1723/7175 [21:55<49:51,  1.82it/s]
 24%|██▍       | 1724/7175 [21:55<52:13,  1.74it/s]
 24%|██▍       | 1725/7175 [21:56<54:06,  1.68it/s]
 24%|██▍       | 1726/7175 [21:56<55:32,  1.64it/s]
 24%|██▍       | 1727/7175 [21:57<56:41,  1.60it/s]
 24%|██▍       | 1728/7175 [21:58<57:27,  1.58it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.00it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.84it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.45it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.13it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.66it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.65it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.39it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.44it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.26it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.36it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.33it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.30it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.28it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.24it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.24it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.22it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.22it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.22it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s]
                                                   
+
                                               
{'eval_loss': '3.697', 'eval_runtime': '3.254', 'eval_samples_per_second': '750.2', 'eval_steps_per_second': '11.99', 'epoch': '6.021'}
+
 24%|██▍       | 1728/7175 [22:01<57:27,  1.58it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s][eval] epoch 6.021 step 1728 loss 3.6970 (best 3.6883 @ ep 5.772, 1/20 since best)
+
+
                                               
 24%|██▍       | 1729/7175 [22:02<2:23:34,  1.58s/it]
 24%|██▍       | 1730/7175 [22:02<1:58:59,  1.31s/it]
                                                     
{'loss': '3.357', 'grad_norm': '0.3164', 'learning_rate': '0.0006', 'epoch': '6.028'}
+
 24%|██▍       | 1730/7175 [22:02<1:58:59,  1.31s/it]
 24%|██▍       | 1731/7175 [22:03<1:41:28,  1.12s/it]
 24%|██▍       | 1732/7175 [22:04<1:29:00,  1.02it/s]
 24%|██▍       | 1733/7175 [22:04<1:20:11,  1.13it/s]
 24%|██▍       | 1734/7175 [22:05<1:13:59,  1.23it/s]
 24%|██▍       | 1735/7175 [22:06<1:09:40,  1.30it/s]
 24%|██▍       | 1736/7175 [22:06<1:06:35,  1.36it/s]
 24%|██▍       | 1737/7175 [22:07<1:04:27,  1.41it/s]
 24%|██▍       | 1738/7175 [22:08<1:02:54,  1.44it/s]
 24%|██▍       | 1739/7175 [22:08<1:01:51,  1.46it/s]
 24%|██▍       | 1740/7175 [22:09<1:01:10,  1.48it/s]
                                                     
{'loss': '3.37', 'grad_norm': '0.2934', 'learning_rate': '0.0006', 'epoch': '6.063'}
+
 24%|██▍       | 1740/7175 [22:09<1:01:10,  1.48it/s]
 24%|██▍       | 1741/7175 [22:10<1:00:42,  1.49it/s]
 24%|██▍       | 1742/7175 [22:10<1:00:17,  1.50it/s]
 24%|██▍       | 1743/7175 [22:11<59:57,  1.51it/s]  
 24%|██▍       | 1744/7175 [22:12<59:44,  1.52it/s]
 24%|██▍       | 1745/7175 [22:12<59:38,  1.52it/s]
 24%|██▍       | 1746/7175 [22:13<59:30,  1.52it/s]
 24%|██▍       | 1747/7175 [22:14<59:24,  1.52it/s]
 24%|██▍       | 1748/7175 [22:14<59:20,  1.52it/s]
 24%|██▍       | 1749/7175 [22:15<59:23,  1.52it/s]
 24%|██▍       | 1750/7175 [22:15<59:20,  1.52it/s]
                                                   
{'loss': '3.369', 'grad_norm': '0.2565', 'learning_rate': '0.0006', 'epoch': '6.098'}
+
 24%|██▍       | 1750/7175 [22:15<59:20,  1.52it/s]
 24%|██▍       | 1751/7175 [22:16<59:19,  1.52it/s]
 24%|██▍       | 1752/7175 [22:17<59:14,  1.53it/s]
 24%|██▍       | 1753/7175 [22:17<59:13,  1.53it/s]
 24%|██▍       | 1754/7175 [22:18<59:13,  1.53it/s]
 24%|██▍       | 1755/7175 [22:19<59:12,  1.53it/s]
 24%|██▍       | 1756/7175 [22:19<59:10,  1.53it/s]
 24%|██▍       | 1757/7175 [22:20<59:09,  1.53it/s]
 25%|██▍       | 1758/7175 [22:21<59:10,  1.53it/s]
 25%|██▍       | 1759/7175 [22:21<59:08,  1.53it/s]
 25%|██▍       | 1760/7175 [22:22<59:06,  1.53it/s]
                                                   
{'loss': '3.357', 'grad_norm': '0.2067', 'learning_rate': '0.0006', 'epoch': '6.133'}
+
 25%|██▍       | 1760/7175 [22:22<59:06,  1.53it/s]
 25%|██▍       | 1761/7175 [22:23<59:06,  1.53it/s]
 25%|██▍       | 1762/7175 [22:23<59:09,  1.52it/s]
 25%|██▍       | 1763/7175 [22:24<59:08,  1.53it/s]
 25%|██▍       | 1764/7175 [22:25<59:06,  1.53it/s]
 25%|██▍       | 1765/7175 [22:25<59:04,  1.53it/s]
 25%|██▍       | 1766/7175 [22:26<59:06,  1.53it/s]
 25%|██▍       | 1767/7175 [22:27<59:04,  1.53it/s]
 25%|██▍       | 1768/7175 [22:27<59:03,  1.53it/s]
 25%|██▍       | 1769/7175 [22:28<59:00,  1.53it/s]
 25%|██▍       | 1770/7175 [22:29<59:00,  1.53it/s]
                                                   
{'loss': '3.363', 'grad_norm': '0.2387', 'learning_rate': '0.0006', 'epoch': '6.168'}
+
 25%|██▍       | 1770/7175 [22:29<59:00,  1.53it/s]
 25%|██▍       | 1771/7175 [22:29<59:02,  1.53it/s]
 25%|██▍       | 1772/7175 [22:30<59:00,  1.53it/s]
 25%|██▍       | 1773/7175 [22:31<58:59,  1.53it/s]
 25%|██▍       | 1774/7175 [22:31<58:57,  1.53it/s]
 25%|██▍       | 1775/7175 [22:32<58:59,  1.53it/s]
 25%|██▍       | 1776/7175 [22:33<58:58,  1.53it/s]
 25%|██▍       | 1777/7175 [22:33<58:55,  1.53it/s]
 25%|██▍       | 1778/7175 [22:34<58:54,  1.53it/s]
 25%|██▍       | 1779/7175 [22:34<58:56,  1.53it/s]
 25%|██▍       | 1780/7175 [22:35<58:55,  1.53it/s]
                                                   
{'loss': '3.348', 'grad_norm': '0.2509', 'learning_rate': '0.0006', 'epoch': '6.203'}
+
 25%|██▍       | 1780/7175 [22:35<58:55,  1.53it/s]
 25%|██▍       | 1781/7175 [22:36<58:56,  1.53it/s]
 25%|██▍       | 1782/7175 [22:36<58:53,  1.53it/s]
 25%|██▍       | 1783/7175 [22:37<58:54,  1.53it/s]
 25%|██▍       | 1784/7175 [22:38<58:53,  1.53it/s]
 25%|██▍       | 1785/7175 [22:38<58:53,  1.53it/s]
 25%|██▍       | 1786/7175 [22:39<58:49,  1.53it/s]
 25%|██▍       | 1787/7175 [22:40<58:50,  1.53it/s]
 25%|██▍       | 1788/7175 [22:40<58:49,  1.53it/s]
 25%|██▍       | 1789/7175 [22:41<58:48,  1.53it/s]
 25%|██▍       | 1790/7175 [22:42<58:46,  1.53it/s]
                                                   
{'loss': '3.36', 'grad_norm': '0.2263', 'learning_rate': '0.0006', 'epoch': '6.238'}
+
 25%|██▍       | 1790/7175 [22:42<58:46,  1.53it/s]
 25%|██▍       | 1791/7175 [22:42<58:47,  1.53it/s]
 25%|██▍       | 1792/7175 [22:43<58:47,  1.53it/s]
 25%|██▍       | 1793/7175 [22:44<58:46,  1.53it/s]
 25%|██▌       | 1794/7175 [22:44<58:44,  1.53it/s]
 25%|██▌       | 1795/7175 [22:45<58:44,  1.53it/s]
 25%|██▌       | 1796/7175 [22:46<58:45,  1.53it/s]
 25%|██▌       | 1797/7175 [22:46<58:44,  1.53it/s]
 25%|██▌       | 1798/7175 [22:47<58:42,  1.53it/s]
 25%|██▌       | 1799/7175 [22:48<58:40,  1.53it/s]
 25%|██▌       | 1800/7175 [22:48<58:42,  1.53it/s]
                                                   
{'loss': '3.367', 'grad_norm': '0.2695', 'learning_rate': '0.0006', 'epoch': '6.273'}
+
 25%|██▌       | 1800/7175 [22:48<58:42,  1.53it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.49it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.76it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.61it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.06it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.75it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.57it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.46it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.38it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.34it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.27it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.29it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.27it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.25it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.25it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.24it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.24it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.24it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.23it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s]
                                                   
+
                                               
{'eval_loss': '3.67', 'eval_runtime': '3.26', 'eval_samples_per_second': '748.7', 'eval_steps_per_second': '11.96', 'epoch': '6.273'}
+
 25%|██▌       | 1800/7175 [22:51<58:42,  1.53it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-1800
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-1800/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-1800/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.65it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.65it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-1800/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-1800/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-1800/tokenizer_config.json
+[eval] epoch 6.273 step 1800 loss 3.6704 (best 3.6704 @ ep 6.273, 0/20 since best)
+
+
                                               
 25%|██▌       | 1801/7175 [22:54<3:24:01,  2.28s/it]
 25%|██▌       | 1802/7175 [22:55<2:40:22,  1.79s/it]
 25%|██▌       | 1803/7175 [22:56<2:09:50,  1.45s/it]
 25%|██▌       | 1804/7175 [22:56<1:48:25,  1.21s/it]
 25%|██▌       | 1805/7175 [22:57<1:33:29,  1.04s/it]
 25%|██▌       | 1806/7175 [22:58<1:23:00,  1.08it/s]
 25%|██▌       | 1807/7175 [22:58<1:15:38,  1.18it/s]
 25%|██▌       | 1808/7175 [22:59<1:10:29,  1.27it/s]
 25%|██▌       | 1809/7175 [23:00<1:06:57,  1.34it/s]
 25%|██▌       | 1810/7175 [23:00<1:04:28,  1.39it/s]
                                                     
{'loss': '3.357', 'grad_norm': '0.2684', 'learning_rate': '0.0006', 'epoch': '6.308'}
+
 25%|██▌       | 1810/7175 [23:00<1:04:28,  1.39it/s]
 25%|██▌       | 1811/7175 [23:01<1:02:42,  1.43it/s]
 25%|██▌       | 1812/7175 [23:01<1:01:25,  1.46it/s]
 25%|██▌       | 1813/7175 [23:02<1:00:32,  1.48it/s]
 25%|██▌       | 1814/7175 [23:03<59:56,  1.49it/s]  
 25%|██▌       | 1815/7175 [23:03<59:31,  1.50it/s]
 25%|██▌       | 1816/7175 [23:04<59:12,  1.51it/s]
 25%|██▌       | 1817/7175 [23:05<58:58,  1.51it/s]
 25%|██▌       | 1818/7175 [23:05<58:51,  1.52it/s]
 25%|██▌       | 1819/7175 [23:06<58:44,  1.52it/s]
 25%|██▌       | 1820/7175 [23:07<58:41,  1.52it/s]
                                                   
{'loss': '3.361', 'grad_norm': '0.2668', 'learning_rate': '0.0006', 'epoch': '6.343'}
+
 25%|██▌       | 1820/7175 [23:07<58:41,  1.52it/s]
 25%|██▌       | 1821/7175 [23:07<58:36,  1.52it/s]
 25%|██▌       | 1822/7175 [23:08<58:33,  1.52it/s]
 25%|██▌       | 1823/7175 [23:09<58:29,  1.52it/s]
 25%|██▌       | 1824/7175 [23:09<58:28,  1.53it/s]
 25%|██▌       | 1825/7175 [23:10<58:24,  1.53it/s]
 25%|██▌       | 1826/7175 [23:11<58:25,  1.53it/s]
 25%|██▌       | 1827/7175 [23:11<58:22,  1.53it/s]
 25%|██▌       | 1828/7175 [23:12<58:21,  1.53it/s]
 25%|██▌       | 1829/7175 [23:13<58:20,  1.53it/s]
 26%|██▌       | 1830/7175 [23:13<58:24,  1.53it/s]
                                                   
{'loss': '3.348', 'grad_norm': '0.2316', 'learning_rate': '0.0006', 'epoch': '6.377'}
+
 26%|██▌       | 1830/7175 [23:13<58:24,  1.53it/s]
 26%|██▌       | 1831/7175 [23:14<58:25,  1.52it/s]
 26%|██▌       | 1832/7175 [23:15<58:22,  1.53it/s]
 26%|██▌       | 1833/7175 [23:15<58:19,  1.53it/s]
 26%|██▌       | 1834/7175 [23:16<58:19,  1.53it/s]
 26%|██▌       | 1835/7175 [23:17<58:21,  1.53it/s]
 26%|██▌       | 1836/7175 [23:17<58:18,  1.53it/s]
 26%|██▌       | 1837/7175 [23:18<58:17,  1.53it/s]
 26%|██▌       | 1838/7175 [23:19<58:14,  1.53it/s]
 26%|██▌       | 1839/7175 [23:19<58:18,  1.53it/s]
 26%|██▌       | 1840/7175 [23:20<58:16,  1.53it/s]
                                                   
{'loss': '3.37', 'grad_norm': '0.227', 'learning_rate': '0.0006', 'epoch': '6.412'}
+
 26%|██▌       | 1840/7175 [23:20<58:16,  1.53it/s]
 26%|██▌       | 1841/7175 [23:21<58:17,  1.53it/s]
 26%|██▌       | 1842/7175 [23:21<58:13,  1.53it/s]
 26%|██▌       | 1843/7175 [23:22<58:15,  1.53it/s]
 26%|██▌       | 1844/7175 [23:22<58:13,  1.53it/s]
 26%|██▌       | 1845/7175 [23:23<58:13,  1.53it/s]
 26%|██▌       | 1846/7175 [23:24<58:09,  1.53it/s]
 26%|██▌       | 1847/7175 [23:24<58:12,  1.53it/s]
 26%|██▌       | 1848/7175 [23:25<58:13,  1.52it/s]
 26%|██▌       | 1849/7175 [23:26<58:11,  1.53it/s]
 26%|██▌       | 1850/7175 [23:26<58:06,  1.53it/s]
                                                   
{'loss': '3.369', 'grad_norm': '0.2799', 'learning_rate': '0.0006', 'epoch': '6.447'}
+
 26%|██▌       | 1850/7175 [23:26<58:06,  1.53it/s]
 26%|██▌       | 1851/7175 [23:27<58:06,  1.53it/s]
 26%|██▌       | 1852/7175 [23:28<58:07,  1.53it/s]
 26%|██▌       | 1853/7175 [23:28<58:06,  1.53it/s]
 26%|██▌       | 1854/7175 [23:29<58:04,  1.53it/s]
 26%|██▌       | 1855/7175 [23:30<58:04,  1.53it/s]
 26%|██▌       | 1856/7175 [23:30<58:04,  1.53it/s]
 26%|██▌       | 1857/7175 [23:31<58:03,  1.53it/s]
 26%|██▌       | 1858/7175 [23:32<58:01,  1.53it/s]
 26%|██▌       | 1859/7175 [23:32<58:00,  1.53it/s]
 26%|██▌       | 1860/7175 [23:33<58:00,  1.53it/s]
                                                   
{'loss': '3.374', 'grad_norm': '0.2718', 'learning_rate': '0.0006', 'epoch': '6.482'}
+
 26%|██▌       | 1860/7175 [23:33<58:00,  1.53it/s]
 26%|██▌       | 1861/7175 [23:34<58:01,  1.53it/s]
 26%|██▌       | 1862/7175 [23:34<57:57,  1.53it/s]
 26%|██▌       | 1863/7175 [23:35<57:58,  1.53it/s]
 26%|██▌       | 1864/7175 [23:36<57:57,  1.53it/s]
 26%|██▌       | 1865/7175 [23:36<57:56,  1.53it/s]
 26%|██▌       | 1866/7175 [23:37<57:55,  1.53it/s]
 26%|██▌       | 1867/7175 [23:38<57:57,  1.53it/s]
 26%|██▌       | 1868/7175 [23:38<57:56,  1.53it/s]
 26%|██▌       | 1869/7175 [23:39<57:56,  1.53it/s]
 26%|██▌       | 1870/7175 [23:39<57:54,  1.53it/s]
                                                   
{'loss': '3.363', 'grad_norm': '0.2378', 'learning_rate': '0.0006', 'epoch': '6.517'}
+
 26%|██▌       | 1870/7175 [23:40<57:54,  1.53it/s]
 26%|██▌       | 1871/7175 [23:40<57:55,  1.53it/s]
 26%|██▌       | 1872/7175 [23:41<57:55,  1.53it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.92it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.63it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.71it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 12.95it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.83it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.48it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.52it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.32it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.39it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.23it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.35it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.31it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.29it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.27it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.26it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.26it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.24it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.25it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s]
                                                   
+
                                               
{'eval_loss': '3.654', 'eval_runtime': '3.24', 'eval_samples_per_second': '753.4', 'eval_steps_per_second': '12.04', 'epoch': '6.524'}
+
 26%|██▌       | 1872/7175 [23:44<57:55,  1.53it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-1872
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-1872/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-1872/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.84it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.84it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-1872/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-1872/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-1872/tokenizer_config.json
+[eval] epoch 6.524 step 1872 loss 3.6536 (best 3.6536 @ ep 6.524, 0/20 since best)
+
+
                                               
 26%|██▌       | 1873/7175 [23:47<3:19:11,  2.25s/it]
 26%|██▌       | 1874/7175 [23:47<2:36:45,  1.77s/it]
 26%|██▌       | 1875/7175 [23:48<2:07:03,  1.44s/it]
 26%|██▌       | 1876/7175 [23:49<1:46:14,  1.20s/it]
 26%|██▌       | 1877/7175 [23:49<1:31:43,  1.04s/it]
 26%|██▌       | 1878/7175 [23:50<1:21:31,  1.08it/s]
 26%|██▌       | 1879/7175 [23:51<1:14:23,  1.19it/s]
 26%|██▌       | 1880/7175 [23:51<1:09:23,  1.27it/s]
                                                     
{'loss': '3.36', 'grad_norm': '0.2151', 'learning_rate': '0.0006', 'epoch': '6.552'}
+
 26%|██▌       | 1880/7175 [23:51<1:09:23,  1.27it/s]
 26%|██▌       | 1881/7175 [23:52<1:05:58,  1.34it/s]
 26%|██▌       | 1882/7175 [23:53<1:03:31,  1.39it/s]
 26%|██▌       | 1883/7175 [23:53<1:01:47,  1.43it/s]
 26%|██▋       | 1884/7175 [23:54<1:00:31,  1.46it/s]
 26%|██▋       | 1885/7175 [23:55<59:42,  1.48it/s]  
 26%|██▋       | 1886/7175 [23:55<59:10,  1.49it/s]
 26%|██▋       | 1887/7175 [23:56<58:42,  1.50it/s]
 26%|██▋       | 1888/7175 [23:57<58:22,  1.51it/s]
 26%|██▋       | 1889/7175 [23:57<58:09,  1.51it/s]
 26%|██▋       | 1890/7175 [23:58<58:03,  1.52it/s]
                                                   
{'loss': '3.359', 'grad_norm': '0.2511', 'learning_rate': '0.0006', 'epoch': '6.587'}
+
 26%|██▋       | 1890/7175 [23:58<58:03,  1.52it/s]
 26%|██▋       | 1891/7175 [23:59<57:57,  1.52it/s]
 26%|██▋       | 1892/7175 [23:59<57:51,  1.52it/s]
 26%|██▋       | 1893/7175 [24:00<57:45,  1.52it/s]
 26%|██▋       | 1894/7175 [24:01<57:47,  1.52it/s]
 26%|██▋       | 1895/7175 [24:01<57:43,  1.52it/s]
 26%|██▋       | 1896/7175 [24:02<57:42,  1.52it/s]
 26%|██▋       | 1897/7175 [24:03<57:39,  1.53it/s]
 26%|██▋       | 1898/7175 [24:03<57:40,  1.53it/s]
 26%|██▋       | 1899/7175 [24:04<57:39,  1.53it/s]
 26%|██▋       | 1900/7175 [24:04<57:38,  1.53it/s]
                                                   
{'loss': '3.358', 'grad_norm': '0.2505', 'learning_rate': '0.0006', 'epoch': '6.622'}
+
 26%|██▋       | 1900/7175 [24:04<57:38,  1.53it/s]
 26%|██▋       | 1901/7175 [24:05<57:36,  1.53it/s]
 27%|██▋       | 1902/7175 [24:06<57:34,  1.53it/s]
 27%|██▋       | 1903/7175 [24:06<57:33,  1.53it/s]
 27%|██▋       | 1904/7175 [24:07<57:34,  1.53it/s]
 27%|██▋       | 1905/7175 [24:08<57:30,  1.53it/s]
 27%|██▋       | 1906/7175 [24:08<57:30,  1.53it/s]
 27%|██▋       | 1907/7175 [24:09<57:30,  1.53it/s]
 27%|██▋       | 1908/7175 [24:10<57:29,  1.53it/s]
 27%|██▋       | 1909/7175 [24:10<57:27,  1.53it/s]
 27%|██▋       | 1910/7175 [24:11<57:28,  1.53it/s]
                                                   
{'loss': '3.373', 'grad_norm': '0.3591', 'learning_rate': '0.0006', 'epoch': '6.657'}
+
 27%|██▋       | 1910/7175 [24:11<57:28,  1.53it/s]
 27%|██▋       | 1911/7175 [24:12<57:32,  1.52it/s]
 27%|██▋       | 1912/7175 [24:12<57:29,  1.53it/s]
 27%|██▋       | 1913/7175 [24:13<57:26,  1.53it/s]
 27%|██▋       | 1914/7175 [24:14<57:24,  1.53it/s]
 27%|██▋       | 1915/7175 [24:14<57:27,  1.53it/s]
 27%|██▋       | 1916/7175 [24:15<57:25,  1.53it/s]
 27%|██▋       | 1917/7175 [24:16<57:23,  1.53it/s]
 27%|██▋       | 1918/7175 [24:16<57:21,  1.53it/s]
 27%|██▋       | 1919/7175 [24:17<57:22,  1.53it/s]
 27%|██▋       | 1920/7175 [24:18<57:22,  1.53it/s]
                                                   
{'loss': '3.355', 'grad_norm': '0.2559', 'learning_rate': '0.0006', 'epoch': '6.692'}
+
 27%|██▋       | 1920/7175 [24:18<57:22,  1.53it/s]
 27%|██▋       | 1921/7175 [24:18<57:23,  1.53it/s]
 27%|██▋       | 1922/7175 [24:19<57:20,  1.53it/s]
 27%|██▋       | 1923/7175 [24:20<57:24,  1.52it/s]
 27%|██▋       | 1924/7175 [24:20<57:21,  1.53it/s]
 27%|██▋       | 1925/7175 [24:21<57:20,  1.53it/s]
 27%|██▋       | 1926/7175 [24:22<57:17,  1.53it/s]
 27%|██▋       | 1927/7175 [24:22<57:20,  1.53it/s]
 27%|██▋       | 1928/7175 [24:23<57:18,  1.53it/s]
 27%|██▋       | 1929/7175 [24:23<57:17,  1.53it/s]
 27%|██▋       | 1930/7175 [24:24<57:13,  1.53it/s]
                                                   
{'loss': '3.36', 'grad_norm': '0.2151', 'learning_rate': '0.0006', 'epoch': '6.727'}
+
 27%|██▋       | 1930/7175 [24:24<57:13,  1.53it/s]
 27%|██▋       | 1931/7175 [24:25<57:16,  1.53it/s]
 27%|██▋       | 1932/7175 [24:25<57:15,  1.53it/s]
 27%|██▋       | 1933/7175 [24:26<57:13,  1.53it/s]
 27%|██▋       | 1934/7175 [24:27<57:11,  1.53it/s]
 27%|██▋       | 1935/7175 [24:27<57:13,  1.53it/s]
 27%|██▋       | 1936/7175 [24:28<57:11,  1.53it/s]
 27%|██▋       | 1937/7175 [24:29<57:10,  1.53it/s]
 27%|██▋       | 1938/7175 [24:29<57:08,  1.53it/s]
 27%|██▋       | 1939/7175 [24:30<57:10,  1.53it/s]
 27%|██▋       | 1940/7175 [24:31<57:09,  1.53it/s]
                                                   
{'loss': '3.348', 'grad_norm': '0.2192', 'learning_rate': '0.0006', 'epoch': '6.762'}
+
 27%|██▋       | 1940/7175 [24:31<57:09,  1.53it/s]
 27%|██▋       | 1941/7175 [24:31<57:11,  1.53it/s]
 27%|██▋       | 1942/7175 [24:32<57:07,  1.53it/s]
 27%|██▋       | 1943/7175 [24:33<57:08,  1.53it/s]
 27%|██▋       | 1944/7175 [24:33<57:06,  1.53it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.91it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.90it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.67it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.09it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.77it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.58it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.45it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.39it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.32it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.30it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.28it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.27it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.26it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.24it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.23it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.24it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.24it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.23it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s]
                                                   
+
                                               
{'eval_loss': '3.637', 'eval_runtime': '3.245', 'eval_samples_per_second': '752.2', 'eval_steps_per_second': '12.02', 'epoch': '6.776'}
+
 27%|██▋       | 1944/7175 [24:37<57:06,  1.53it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-1944
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-1944/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-1944/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.88it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.88it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-1944/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-1944/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-1944/tokenizer_config.json
+[eval] epoch 6.776 step 1944 loss 3.6372 (best 3.6372 @ ep 6.776, 0/20 since best)
+
+
                                               
 27%|██▋       | 1945/7175 [24:39<3:16:02,  2.25s/it]
 27%|██▋       | 1946/7175 [24:40<2:34:19,  1.77s/it]
 27%|██▋       | 1947/7175 [24:41<2:05:05,  1.44s/it]
 27%|██▋       | 1948/7175 [24:41<1:44:37,  1.20s/it]
 27%|██▋       | 1949/7175 [24:42<1:30:19,  1.04s/it]
 27%|██▋       | 1950/7175 [24:43<1:20:21,  1.08it/s]
                                                     
{'loss': '3.351', 'grad_norm': '0.2294', 'learning_rate': '0.0006', 'epoch': '6.797'}
+
 27%|██▋       | 1950/7175 [24:43<1:20:21,  1.08it/s]
 27%|██▋       | 1951/7175 [24:43<1:13:22,  1.19it/s]
 27%|██▋       | 1952/7175 [24:44<1:08:25,  1.27it/s]
 27%|██▋       | 1953/7175 [24:45<1:04:59,  1.34it/s]
 27%|██▋       | 1954/7175 [24:45<1:02:36,  1.39it/s]
 27%|██▋       | 1955/7175 [24:46<1:00:54,  1.43it/s]
 27%|██▋       | 1956/7175 [24:46<59:39,  1.46it/s]  
 27%|██▋       | 1957/7175 [24:47<58:51,  1.48it/s]
 27%|██▋       | 1958/7175 [24:48<58:18,  1.49it/s]
 27%|██▋       | 1959/7175 [24:48<57:52,  1.50it/s]
 27%|██▋       | 1960/7175 [24:49<57:32,  1.51it/s]
                                                   
{'loss': '3.355', 'grad_norm': '0.2845', 'learning_rate': '0.0006', 'epoch': '6.832'}
+
 27%|██▋       | 1960/7175 [24:49<57:32,  1.51it/s]
 27%|██▋       | 1961/7175 [24:50<57:22,  1.51it/s]
 27%|██▋       | 1962/7175 [24:50<57:14,  1.52it/s]
 27%|██▋       | 1963/7175 [24:51<57:06,  1.52it/s]
 27%|██▋       | 1964/7175 [24:52<57:00,  1.52it/s]
 27%|██▋       | 1965/7175 [24:52<56:58,  1.52it/s]
 27%|██▋       | 1966/7175 [24:53<56:56,  1.52it/s]
 27%|██▋       | 1967/7175 [24:54<56:52,  1.53it/s]
 27%|██▋       | 1968/7175 [24:54<56:49,  1.53it/s]
 27%|██▋       | 1969/7175 [24:55<56:53,  1.53it/s]
 27%|██▋       | 1970/7175 [24:56<56:50,  1.53it/s]
                                                   
{'loss': '3.363', 'grad_norm': '0.2066', 'learning_rate': '0.0006', 'epoch': '6.867'}
+
 27%|██▋       | 1970/7175 [24:56<56:50,  1.53it/s]
 27%|██▋       | 1971/7175 [24:56<56:51,  1.53it/s]
 27%|██▋       | 1972/7175 [24:57<56:47,  1.53it/s]
 27%|██▋       | 1973/7175 [24:58<56:48,  1.53it/s]
 28%|██▊       | 1974/7175 [24:58<56:48,  1.53it/s]
 28%|██▊       | 1975/7175 [24:59<56:46,  1.53it/s]
 28%|██▊       | 1976/7175 [25:00<56:44,  1.53it/s]
 28%|██▊       | 1977/7175 [25:00<56:46,  1.53it/s]
 28%|██▊       | 1978/7175 [25:01<56:44,  1.53it/s]
 28%|██▊       | 1979/7175 [25:02<56:42,  1.53it/s]
 28%|██▊       | 1980/7175 [25:02<56:40,  1.53it/s]
                                                   
{'loss': '3.346', 'grad_norm': '0.2052', 'learning_rate': '0.0006', 'epoch': '6.902'}
+
 28%|██▊       | 1980/7175 [25:02<56:40,  1.53it/s]
 28%|██▊       | 1981/7175 [25:03<56:43,  1.53it/s]
 28%|██▊       | 1982/7175 [25:04<56:42,  1.53it/s]
 28%|██▊       | 1983/7175 [25:04<56:39,  1.53it/s]
 28%|██▊       | 1984/7175 [25:05<56:38,  1.53it/s]
 28%|██▊       | 1985/7175 [25:05<56:39,  1.53it/s]
 28%|██▊       | 1986/7175 [25:06<56:39,  1.53it/s]
 28%|██▊       | 1987/7175 [25:07<56:37,  1.53it/s]
 28%|██▊       | 1988/7175 [25:07<56:38,  1.53it/s]
 28%|██▊       | 1989/7175 [25:08<56:37,  1.53it/s]
 28%|██▊       | 1990/7175 [25:09<56:36,  1.53it/s]
                                                   
{'loss': '3.356', 'grad_norm': '0.2866', 'learning_rate': '0.0006', 'epoch': '6.937'}
+
 28%|██▊       | 1990/7175 [25:09<56:36,  1.53it/s]
 28%|██▊       | 1991/7175 [25:09<56:36,  1.53it/s]
 28%|██▊       | 1992/7175 [25:10<56:33,  1.53it/s]
 28%|██▊       | 1993/7175 [25:11<56:34,  1.53it/s]
 28%|██▊       | 1994/7175 [25:11<56:34,  1.53it/s]
 28%|██▊       | 1995/7175 [25:12<56:33,  1.53it/s]
 28%|██▊       | 1996/7175 [25:13<56:30,  1.53it/s]
 28%|██▊       | 1997/7175 [25:13<56:31,  1.53it/s]
 28%|██▊       | 1998/7175 [25:14<56:30,  1.53it/s]
 28%|██▊       | 1999/7175 [25:15<56:27,  1.53it/s]
 28%|██▊       | 2000/7175 [25:15<56:26,  1.53it/s]
                                                   
{'loss': '3.357', 'grad_norm': '0.2599', 'learning_rate': '0.0006', 'epoch': '6.972'}
+
 28%|██▊       | 2000/7175 [25:15<56:26,  1.53it/s]
 28%|██▊       | 2001/7175 [25:16<56:30,  1.53it/s]
 28%|██▊       | 2002/7175 [25:17<56:29,  1.53it/s]
 28%|██▊       | 2003/7175 [25:17<56:26,  1.53it/s]
 28%|██▊       | 2004/7175 [25:18<56:27,  1.53it/s]
 28%|██▊       | 2005/7175 [25:19<56:29,  1.53it/s]
 28%|██▊       | 2006/7175 [25:19<56:35,  1.52it/s]
 28%|██▊       | 2007/7175 [25:20<56:32,  1.52it/s]
 28%|██▊       | 2008/7175 [25:21<56:28,  1.52it/s]
 28%|██▊       | 2010/7175 [25:21<46:31,  1.85it/s]
                                                   
{'loss': '3.32', 'grad_norm': '0.2784', 'learning_rate': '0.0006', 'epoch': '7.003'}
+
 28%|██▊       | 2010/7175 [25:21<46:31,  1.85it/s]
 28%|██▊       | 2011/7175 [25:22<49:01,  1.76it/s]
 28%|██▊       | 2012/7175 [25:23<50:55,  1.69it/s]
 28%|██▊       | 2013/7175 [25:23<52:25,  1.64it/s]
 28%|██▊       | 2014/7175 [25:24<53:29,  1.61it/s]
 28%|██▊       | 2015/7175 [25:25<54:18,  1.58it/s]
 28%|██▊       | 2016/7175 [25:25<54:53,  1.57it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 19.05it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.97it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.68it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.13it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.78it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.59it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.47it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.39it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.32it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.30it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.28it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.27it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.26it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.24it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.24it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.24it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.23it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.23it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.22it/s]
                                                   
+
                                               
{'eval_loss': '3.652', 'eval_runtime': '3.247', 'eval_samples_per_second': '751.8', 'eval_steps_per_second': '12.01', 'epoch': '7.024'}
+
 28%|██▊       | 2016/7175 [25:29<54:53,  1.57it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.22it/s][eval] epoch 7.024 step 2016 loss 3.6519 (best 3.6372 @ ep 6.776, 1/20 since best)
+
+
                                               
 28%|██▊       | 2017/7175 [25:29<2:17:03,  1.59s/it]
 28%|██▊       | 2018/7175 [25:30<1:53:13,  1.32s/it]
 28%|██▊       | 2019/7175 [25:31<1:36:19,  1.12s/it]
 28%|██▊       | 2020/7175 [25:31<1:24:23,  1.02it/s]
                                                     
{'loss': '3.304', 'grad_norm': '0.324', 'learning_rate': '0.0006', 'epoch': '7.038'}
+
 28%|██▊       | 2020/7175 [25:31<1:24:23,  1.02it/s]
 28%|██▊       | 2021/7175 [25:32<1:16:01,  1.13it/s]
 28%|██▊       | 2022/7175 [25:32<1:10:08,  1.22it/s]
 28%|██▊       | 2023/7175 [25:33<1:05:58,  1.30it/s]
 28%|██▊       | 2024/7175 [25:34<1:03:02,  1.36it/s]
 28%|██▊       | 2025/7175 [25:34<1:00:59,  1.41it/s]
 28%|██▊       | 2026/7175 [25:35<59:34,  1.44it/s]  
 28%|██▊       | 2027/7175 [25:36<58:33,  1.47it/s]
 28%|██▊       | 2028/7175 [25:36<57:48,  1.48it/s]
 28%|██▊       | 2029/7175 [25:37<57:19,  1.50it/s]
 28%|██▊       | 2030/7175 [25:38<57:00,  1.50it/s]
                                                   
{'loss': '3.299', 'grad_norm': '0.2414', 'learning_rate': '0.0006', 'epoch': '7.073'}
+
 28%|██▊       | 2030/7175 [25:38<57:00,  1.50it/s]
 28%|██▊       | 2031/7175 [25:38<56:45,  1.51it/s]
 28%|██▊       | 2032/7175 [25:39<56:32,  1.52it/s]
 28%|██▊       | 2033/7175 [25:40<56:23,  1.52it/s]
 28%|██▊       | 2034/7175 [25:40<56:21,  1.52it/s]
 28%|██▊       | 2035/7175 [25:41<56:17,  1.52it/s]
 28%|██▊       | 2036/7175 [25:42<56:13,  1.52it/s]
 28%|██▊       | 2037/7175 [25:42<56:09,  1.52it/s]
 28%|██▊       | 2038/7175 [25:43<56:09,  1.52it/s]
 28%|██▊       | 2039/7175 [25:44<56:07,  1.53it/s]
 28%|██▊       | 2040/7175 [25:44<56:07,  1.53it/s]
                                                   
{'loss': '3.303', 'grad_norm': '0.2407', 'learning_rate': '0.0006', 'epoch': '7.108'}
+
 28%|██▊       | 2040/7175 [25:44<56:07,  1.53it/s]
 28%|██▊       | 2041/7175 [25:45<56:06,  1.53it/s]
 28%|██▊       | 2042/7175 [25:46<56:04,  1.53it/s]
 28%|██▊       | 2043/7175 [25:46<56:05,  1.53it/s]
 28%|██▊       | 2044/7175 [25:47<56:03,  1.53it/s]
 29%|██▊       | 2045/7175 [25:48<56:00,  1.53it/s]
 29%|██▊       | 2046/7175 [25:48<55:58,  1.53it/s]
 29%|██▊       | 2047/7175 [25:49<56:02,  1.53it/s]
 29%|██▊       | 2048/7175 [25:50<56:01,  1.53it/s]
 29%|██▊       | 2049/7175 [25:50<55:59,  1.53it/s]
 29%|██▊       | 2050/7175 [25:51<55:57,  1.53it/s]
                                                   
{'loss': '3.292', 'grad_norm': '0.2408', 'learning_rate': '0.0006', 'epoch': '7.143'}
+
 29%|██▊       | 2050/7175 [25:51<55:57,  1.53it/s]
 29%|██▊       | 2051/7175 [25:51<55:59,  1.53it/s]
 29%|██▊       | 2052/7175 [25:52<55:57,  1.53it/s]
 29%|██▊       | 2053/7175 [25:53<55:56,  1.53it/s]
 29%|██▊       | 2054/7175 [25:53<55:54,  1.53it/s]
 29%|██▊       | 2055/7175 [25:54<55:55,  1.53it/s]
 29%|██▊       | 2056/7175 [25:55<55:53,  1.53it/s]
 29%|██▊       | 2057/7175 [25:55<55:53,  1.53it/s]
 29%|██▊       | 2058/7175 [25:56<55:50,  1.53it/s]
 29%|██▊       | 2059/7175 [25:57<55:51,  1.53it/s]
 29%|██▊       | 2060/7175 [25:57<55:51,  1.53it/s]
                                                   
{'loss': '3.276', 'grad_norm': '0.1919', 'learning_rate': '0.0006', 'epoch': '7.178'}
+
 29%|██▊       | 2060/7175 [25:57<55:51,  1.53it/s]
 29%|██▊       | 2061/7175 [25:58<55:51,  1.53it/s]
 29%|██▊       | 2062/7175 [25:59<55:50,  1.53it/s]
 29%|██▉       | 2063/7175 [25:59<55:48,  1.53it/s]
 29%|██▉       | 2064/7175 [26:00<55:49,  1.53it/s]
 29%|██▉       | 2065/7175 [26:01<55:48,  1.53it/s]
 29%|██▉       | 2066/7175 [26:01<55:47,  1.53it/s]
 29%|██▉       | 2067/7175 [26:02<55:44,  1.53it/s]
 29%|██▉       | 2068/7175 [26:03<55:47,  1.53it/s]
 29%|██▉       | 2069/7175 [26:03<55:45,  1.53it/s]
 29%|██▉       | 2070/7175 [26:04<55:43,  1.53it/s]
                                                   
{'loss': '3.296', 'grad_norm': '0.2656', 'learning_rate': '0.0006', 'epoch': '7.213'}
+
 29%|██▉       | 2070/7175 [26:04<55:43,  1.53it/s]
 29%|██▉       | 2071/7175 [26:05<55:42,  1.53it/s]
 29%|██▉       | 2072/7175 [26:05<55:44,  1.53it/s]
 29%|██▉       | 2073/7175 [26:06<55:42,  1.53it/s]
 29%|██▉       | 2074/7175 [26:07<55:40,  1.53it/s]
 29%|██▉       | 2075/7175 [26:07<55:38,  1.53it/s]
 29%|██▉       | 2076/7175 [26:08<55:41,  1.53it/s]
 29%|██▉       | 2077/7175 [26:09<55:39,  1.53it/s]
 29%|██▉       | 2078/7175 [26:09<55:37,  1.53it/s]
 29%|██▉       | 2079/7175 [26:10<55:35,  1.53it/s]
 29%|██▉       | 2080/7175 [26:10<55:37,  1.53it/s]
                                                   
{'loss': '3.297', 'grad_norm': '0.228', 'learning_rate': '0.0006', 'epoch': '7.248'}
+
 29%|██▉       | 2080/7175 [26:10<55:37,  1.53it/s]
 29%|██▉       | 2081/7175 [26:11<55:37,  1.53it/s]
 29%|██▉       | 2082/7175 [26:12<55:35,  1.53it/s]
 29%|██▉       | 2083/7175 [26:12<55:34,  1.53it/s]
 29%|██▉       | 2084/7175 [26:13<55:37,  1.53it/s]
 29%|██▉       | 2085/7175 [26:14<55:34,  1.53it/s]
 29%|██▉       | 2086/7175 [26:14<55:32,  1.53it/s]
 29%|██▉       | 2087/7175 [26:15<55:30,  1.53it/s]
 29%|██▉       | 2088/7175 [26:16<55:33,  1.53it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 19.73it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 15.15it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.77it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.15it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.82it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.61it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.48it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.40it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.34it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.30it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.31it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.25it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.26it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.24it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.24it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.25it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.24it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.23it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s]
                                                   
+
                                               
{'eval_loss': '3.624', 'eval_runtime': '3.243', 'eval_samples_per_second': '752.6', 'eval_steps_per_second': '12.03', 'epoch': '7.276'}
+
 29%|██▉       | 2088/7175 [26:19<55:33,  1.53it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-2088
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-2088/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-2088/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.84it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.84it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-2088/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-2088/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-2088/tokenizer_config.json
+[eval] epoch 7.276 step 2088 loss 3.6243 (best 3.6243 @ ep 7.276, 0/20 since best)
+
+
                                               
 29%|██▉       | 2089/7175 [26:22<3:11:17,  2.26s/it]
 29%|██▉       | 2090/7175 [26:22<2:30:28,  1.78s/it]
                                                     
{'loss': '3.296', 'grad_norm': '0.2224', 'learning_rate': '0.0006', 'epoch': '7.283'}
+
 29%|██▉       | 2090/7175 [26:22<2:30:28,  1.78s/it]
 29%|██▉       | 2091/7175 [26:23<2:01:59,  1.44s/it]
 29%|██▉       | 2092/7175 [26:24<1:42:00,  1.20s/it]
 29%|██▉       | 2093/7175 [26:24<1:28:01,  1.04s/it]
 29%|██▉       | 2094/7175 [26:25<1:18:14,  1.08it/s]
 29%|██▉       | 2095/7175 [26:26<1:11:25,  1.19it/s]
 29%|██▉       | 2096/7175 [26:26<1:06:37,  1.27it/s]
 29%|██▉       | 2097/7175 [26:27<1:03:15,  1.34it/s]
 29%|██▉       | 2098/7175 [26:28<1:00:52,  1.39it/s]
 29%|██▉       | 2099/7175 [26:28<59:15,  1.43it/s]  
 29%|██▉       | 2100/7175 [26:29<58:06,  1.46it/s]
                                                   
{'loss': '3.291', 'grad_norm': '0.2445', 'learning_rate': '0.0006', 'epoch': '7.318'}
+
 29%|██▉       | 2100/7175 [26:29<58:06,  1.46it/s]
 29%|██▉       | 2101/7175 [26:30<57:18,  1.48it/s]
 29%|██▉       | 2102/7175 [26:30<56:41,  1.49it/s]
 29%|██▉       | 2103/7175 [26:31<56:17,  1.50it/s]
 29%|██▉       | 2104/7175 [26:32<56:02,  1.51it/s]
 29%|██▉       | 2105/7175 [26:32<55:50,  1.51it/s]
 29%|██▉       | 2106/7175 [26:33<55:39,  1.52it/s]
 29%|██▉       | 2107/7175 [26:33<55:32,  1.52it/s]
 29%|██▉       | 2108/7175 [26:34<55:31,  1.52it/s]
 29%|██▉       | 2109/7175 [26:35<55:27,  1.52it/s]
 29%|██▉       | 2110/7175 [26:35<55:22,  1.52it/s]
                                                   
{'loss': '3.303', 'grad_norm': '0.2193', 'learning_rate': '0.0006', 'epoch': '7.353'}
+
 29%|██▉       | 2110/7175 [26:35<55:22,  1.52it/s]
 29%|██▉       | 2111/7175 [26:36<55:20,  1.52it/s]
 29%|██▉       | 2112/7175 [26:37<55:21,  1.52it/s]
 29%|██▉       | 2113/7175 [26:37<55:19,  1.52it/s]
 29%|██▉       | 2114/7175 [26:38<55:17,  1.53it/s]
 29%|██▉       | 2115/7175 [26:39<55:14,  1.53it/s]
 29%|██▉       | 2116/7175 [26:39<55:15,  1.53it/s]
 30%|██▉       | 2117/7175 [26:40<55:15,  1.53it/s]
 30%|██▉       | 2118/7175 [26:41<55:14,  1.53it/s]
 30%|██▉       | 2119/7175 [26:41<55:12,  1.53it/s]
 30%|██▉       | 2120/7175 [26:42<55:12,  1.53it/s]
                                                   
{'loss': '3.304', 'grad_norm': '0.238', 'learning_rate': '0.0006', 'epoch': '7.388'}
+
 30%|██▉       | 2120/7175 [26:42<55:12,  1.53it/s]
 30%|██▉       | 2121/7175 [26:43<55:17,  1.52it/s]
 30%|██▉       | 2122/7175 [26:43<55:19,  1.52it/s]
 30%|██▉       | 2123/7175 [26:44<55:18,  1.52it/s]
 30%|██▉       | 2124/7175 [26:45<55:17,  1.52it/s]
 30%|██▉       | 2125/7175 [26:45<55:12,  1.52it/s]
 30%|██▉       | 2126/7175 [26:46<55:10,  1.52it/s]
 30%|██▉       | 2127/7175 [26:47<55:14,  1.52it/s]
 30%|██▉       | 2128/7175 [26:47<55:13,  1.52it/s]
 30%|██▉       | 2129/7175 [26:48<55:11,  1.52it/s]
 30%|██▉       | 2130/7175 [26:49<55:06,  1.53it/s]
                                                   
{'loss': '3.289', 'grad_norm': '0.2395', 'learning_rate': '0.0006', 'epoch': '7.423'}
+
 30%|██▉       | 2130/7175 [26:49<55:06,  1.53it/s]
 30%|██▉       | 2131/7175 [26:49<55:06,  1.53it/s]
 30%|██▉       | 2132/7175 [26:50<55:08,  1.52it/s]
 30%|██▉       | 2133/7175 [26:51<55:06,  1.52it/s]
 30%|██▉       | 2134/7175 [26:51<55:01,  1.53it/s]
 30%|██▉       | 2135/7175 [26:52<55:01,  1.53it/s]
 30%|██▉       | 2136/7175 [26:53<55:04,  1.52it/s]
 30%|██▉       | 2137/7175 [26:53<55:02,  1.53it/s]
 30%|██▉       | 2138/7175 [26:54<55:03,  1.52it/s]
 30%|██▉       | 2139/7175 [26:54<54:59,  1.53it/s]
 30%|██▉       | 2140/7175 [26:55<55:00,  1.53it/s]
                                                   
{'loss': '3.295', 'grad_norm': '0.2512', 'learning_rate': '0.0006', 'epoch': '7.458'}
+
 30%|██▉       | 2140/7175 [26:55<55:00,  1.53it/s]
 30%|██▉       | 2141/7175 [26:56<55:03,  1.52it/s]
 30%|██▉       | 2142/7175 [26:56<55:00,  1.52it/s]
 30%|██▉       | 2143/7175 [26:57<54:57,  1.53it/s]
 30%|██▉       | 2144/7175 [26:58<54:56,  1.53it/s]
 30%|██▉       | 2145/7175 [26:58<54:59,  1.52it/s]
 30%|██▉       | 2146/7175 [26:59<54:57,  1.53it/s]
 30%|██▉       | 2147/7175 [27:00<54:55,  1.53it/s]
 30%|██▉       | 2148/7175 [27:00<54:53,  1.53it/s]
 30%|██▉       | 2149/7175 [27:01<54:54,  1.53it/s]
 30%|██▉       | 2150/7175 [27:02<54:53,  1.53it/s]
                                                   
{'loss': '3.3', 'grad_norm': '0.2332', 'learning_rate': '0.0006', 'epoch': '7.493'}
+
 30%|██▉       | 2150/7175 [27:02<54:53,  1.53it/s]
 30%|██▉       | 2151/7175 [27:02<54:53,  1.53it/s]
 30%|██▉       | 2152/7175 [27:03<54:50,  1.53it/s]
 30%|███       | 2153/7175 [27:04<54:49,  1.53it/s]
 30%|███       | 2154/7175 [27:04<54:50,  1.53it/s]
 30%|███       | 2155/7175 [27:05<54:49,  1.53it/s]
 30%|███       | 2156/7175 [27:06<54:47,  1.53it/s]
 30%|███       | 2157/7175 [27:06<54:45,  1.53it/s]
 30%|███       | 2158/7175 [27:07<54:48,  1.53it/s]
 30%|███       | 2159/7175 [27:08<54:47,  1.53it/s]
 30%|███       | 2160/7175 [27:08<54:44,  1.53it/s]
                                                   
{'loss': '3.309', 'grad_norm': '0.2412', 'learning_rate': '0.0006', 'epoch': '7.528'}
+
 30%|███       | 2160/7175 [27:08<54:44,  1.53it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.34it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.46it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.61it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 12.92it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.80it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.48it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.51it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.30it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.39it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.22it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.34it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.30it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.27it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.25it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.26it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.25it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.24it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.23it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.22it/s]
                                                   
+
                                               
{'eval_loss': '3.615', 'eval_runtime': '3.253', 'eval_samples_per_second': '750.3', 'eval_steps_per_second': '11.99', 'epoch': '7.528'}
+
 30%|███       | 2160/7175 [27:11<54:44,  1.53it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.22it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-2160
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-2160/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-2160/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.87it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.87it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-2160/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-2160/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-2160/tokenizer_config.json
+[eval] epoch 7.528 step 2160 loss 3.6153 (best 3.6153 @ ep 7.528, 0/20 since best)
+
+
                                               
 30%|███       | 2161/7175 [27:14<3:08:12,  2.25s/it]
 30%|███       | 2162/7175 [27:15<2:28:06,  1.77s/it]
 30%|███       | 2163/7175 [27:16<2:00:01,  1.44s/it]
 30%|███       | 2164/7175 [27:16<1:40:22,  1.20s/it]
 30%|███       | 2165/7175 [27:17<1:26:36,  1.04s/it]
 30%|███       | 2166/7175 [27:17<1:17:01,  1.08it/s]
 30%|███       | 2167/7175 [27:18<1:10:18,  1.19it/s]
 30%|███       | 2168/7175 [27:19<1:05:33,  1.27it/s]
 30%|███       | 2169/7175 [27:19<1:02:18,  1.34it/s]
 30%|███       | 2170/7175 [27:20<59:58,  1.39it/s]  
                                                   
{'loss': '3.307', 'grad_norm': '0.2401', 'learning_rate': '0.0006', 'epoch': '7.563'}
+
 30%|███       | 2170/7175 [27:20<59:58,  1.39it/s]
 30%|███       | 2171/7175 [27:21<58:22,  1.43it/s]
 30%|███       | 2172/7175 [27:21<57:12,  1.46it/s]
 30%|███       | 2173/7175 [27:22<56:27,  1.48it/s]
 30%|███       | 2174/7175 [27:23<55:53,  1.49it/s]
 30%|███       | 2175/7175 [27:23<55:31,  1.50it/s]
 30%|███       | 2176/7175 [27:24<55:10,  1.51it/s]
 30%|███       | 2177/7175 [27:25<55:00,  1.51it/s]
 30%|███       | 2178/7175 [27:25<54:51,  1.52it/s]
 30%|███       | 2179/7175 [27:26<54:44,  1.52it/s]
 30%|███       | 2180/7175 [27:27<54:37,  1.52it/s]
                                                   
{'loss': '3.302', 'grad_norm': '0.2281', 'learning_rate': '0.0006', 'epoch': '7.598'}
+
 30%|███       | 2180/7175 [27:27<54:37,  1.52it/s]
 30%|███       | 2181/7175 [27:27<54:39,  1.52it/s]
 30%|███       | 2182/7175 [27:28<54:36,  1.52it/s]
 30%|███       | 2183/7175 [27:29<54:32,  1.53it/s]
 30%|███       | 2184/7175 [27:29<54:29,  1.53it/s]
 30%|███       | 2185/7175 [27:30<54:31,  1.53it/s]
 30%|███       | 2186/7175 [27:31<54:29,  1.53it/s]
 30%|███       | 2187/7175 [27:31<54:27,  1.53it/s]
 30%|███       | 2188/7175 [27:32<54:25,  1.53it/s]
 31%|███       | 2189/7175 [27:33<54:27,  1.53it/s]
 31%|███       | 2190/7175 [27:33<54:27,  1.53it/s]
                                                   
{'loss': '3.302', 'grad_norm': '0.2728', 'learning_rate': '0.0006', 'epoch': '7.633'}
+
 31%|███       | 2190/7175 [27:33<54:27,  1.53it/s]
 31%|███       | 2191/7175 [27:34<54:27,  1.53it/s]
 31%|███       | 2192/7175 [27:35<54:24,  1.53it/s]
 31%|███       | 2193/7175 [27:35<54:23,  1.53it/s]
 31%|███       | 2194/7175 [27:36<54:25,  1.53it/s]
 31%|███       | 2195/7175 [27:36<54:25,  1.53it/s]
 31%|███       | 2196/7175 [27:37<54:23,  1.53it/s]
 31%|███       | 2197/7175 [27:38<54:20,  1.53it/s]
 31%|███       | 2198/7175 [27:38<54:23,  1.52it/s]
 31%|███       | 2199/7175 [27:39<54:22,  1.53it/s]
 31%|███       | 2200/7175 [27:40<54:21,  1.53it/s]
                                                   
{'loss': '3.308', 'grad_norm': '0.2358', 'learning_rate': '0.0006', 'epoch': '7.668'}
+
 31%|███       | 2200/7175 [27:40<54:21,  1.53it/s]
 31%|███       | 2201/7175 [27:40<54:19,  1.53it/s]
 31%|███       | 2202/7175 [27:41<54:19,  1.53it/s]
 31%|███       | 2203/7175 [27:42<54:21,  1.52it/s]
 31%|███       | 2204/7175 [27:42<54:19,  1.53it/s]
 31%|███       | 2205/7175 [27:43<54:18,  1.53it/s]
 31%|███       | 2206/7175 [27:44<54:15,  1.53it/s]
 31%|███       | 2207/7175 [27:44<54:17,  1.53it/s]
 31%|███       | 2208/7175 [27:45<54:17,  1.52it/s]
 31%|███       | 2209/7175 [27:46<54:15,  1.53it/s]
 31%|███       | 2210/7175 [27:46<54:10,  1.53it/s]
                                                   
{'loss': '3.318', 'grad_norm': '0.2633', 'learning_rate': '0.0006', 'epoch': '7.702'}
+
 31%|███       | 2210/7175 [27:46<54:10,  1.53it/s]
 31%|███       | 2211/7175 [27:47<54:12,  1.53it/s]
 31%|███       | 2212/7175 [27:48<54:14,  1.52it/s]
 31%|███       | 2213/7175 [27:48<54:12,  1.53it/s]
 31%|███       | 2214/7175 [27:49<54:07,  1.53it/s]
 31%|███       | 2215/7175 [27:50<54:08,  1.53it/s]
 31%|███       | 2216/7175 [27:50<54:08,  1.53it/s]
 31%|███       | 2217/7175 [27:51<54:07,  1.53it/s]
 31%|███       | 2218/7175 [27:52<54:03,  1.53it/s]
 31%|███       | 2219/7175 [27:52<54:06,  1.53it/s]
 31%|███       | 2220/7175 [27:53<54:06,  1.53it/s]
                                                   
{'loss': '3.313', 'grad_norm': '0.2507', 'learning_rate': '0.0006', 'epoch': '7.737'}
+
 31%|███       | 2220/7175 [27:53<54:06,  1.53it/s]
 31%|███       | 2221/7175 [27:54<54:06,  1.53it/s]
 31%|███       | 2222/7175 [27:54<54:02,  1.53it/s]
 31%|███       | 2223/7175 [27:55<54:03,  1.53it/s]
 31%|███       | 2224/7175 [27:55<54:03,  1.53it/s]
 31%|███       | 2225/7175 [27:56<54:03,  1.53it/s]
 31%|███       | 2226/7175 [27:57<53:59,  1.53it/s]
 31%|███       | 2227/7175 [27:57<54:01,  1.53it/s]
 31%|███       | 2228/7175 [27:58<54:01,  1.53it/s]
 31%|███       | 2229/7175 [27:59<53:58,  1.53it/s]
 31%|███       | 2230/7175 [27:59<53:56,  1.53it/s]
                                                   
{'loss': '3.298', 'grad_norm': '0.2142', 'learning_rate': '0.0006', 'epoch': '7.772'}
+
 31%|███       | 2230/7175 [27:59<53:56,  1.53it/s]
 31%|███       | 2231/7175 [28:00<53:59,  1.53it/s]
 31%|███       | 2232/7175 [28:01<53:58,  1.53it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.54it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.78it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.59it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.04it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.74it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.56it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.43it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.36it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.31it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.28it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.28it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.27it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.25it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.24it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.23it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.23it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.23it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.22it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s]
                                                   
+
                                               
{'eval_loss': '3.604', 'eval_runtime': '3.239', 'eval_samples_per_second': '753.5', 'eval_steps_per_second': '12.04', 'epoch': '7.779'}
+
 31%|███       | 2232/7175 [28:04<53:58,  1.53it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-2232
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-2232/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-2232/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.86it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.86it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-2232/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-2232/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-2232/tokenizer_config.json
+[eval] epoch 7.779 step 2232 loss 3.6044 (best 3.6044 @ ep 7.779, 0/20 since best)
+
+
                                               
 31%|███       | 2233/7175 [28:07<3:05:48,  2.26s/it]
 31%|███       | 2234/7175 [28:07<2:26:12,  1.78s/it]
 31%|███       | 2235/7175 [28:08<1:58:32,  1.44s/it]
 31%|███       | 2236/7175 [28:09<1:39:08,  1.20s/it]
 31%|███       | 2237/7175 [28:09<1:25:31,  1.04s/it]
 31%|███       | 2238/7175 [28:10<1:16:00,  1.08it/s]
 31%|███       | 2239/7175 [28:11<1:09:23,  1.19it/s]
 31%|███       | 2240/7175 [28:11<1:04:43,  1.27it/s]
                                                     
{'loss': '3.31', 'grad_norm': '0.218', 'learning_rate': '0.0006', 'epoch': '7.807'}
+
 31%|███       | 2240/7175 [28:11<1:04:43,  1.27it/s]
 31%|███       | 2241/7175 [28:12<1:01:28,  1.34it/s]
 31%|███       | 2242/7175 [28:13<59:09,  1.39it/s]  
 31%|███▏      | 2243/7175 [28:13<57:37,  1.43it/s]
 31%|███▏      | 2244/7175 [28:14<56:29,  1.45it/s]
 31%|███▏      | 2245/7175 [28:15<55:41,  1.48it/s]
 31%|███▏      | 2246/7175 [28:15<55:05,  1.49it/s]
 31%|███▏      | 2247/7175 [28:16<54:43,  1.50it/s]
 31%|███▏      | 2248/7175 [28:17<54:29,  1.51it/s]
 31%|███▏      | 2249/7175 [28:17<54:17,  1.51it/s]
 31%|███▏      | 2250/7175 [28:18<54:08,  1.52it/s]
                                                   
{'loss': '3.302', 'grad_norm': '0.2012', 'learning_rate': '0.0006', 'epoch': '7.842'}
+
 31%|███▏      | 2250/7175 [28:18<54:08,  1.52it/s]
 31%|███▏      | 2251/7175 [28:19<54:00,  1.52it/s]
 31%|███▏      | 2252/7175 [28:19<53:56,  1.52it/s]
 31%|███▏      | 2253/7175 [28:20<53:55,  1.52it/s]
 31%|███▏      | 2254/7175 [28:20<53:51,  1.52it/s]
 31%|███▏      | 2255/7175 [28:21<53:48,  1.52it/s]
 31%|███▏      | 2256/7175 [28:22<53:43,  1.53it/s]
 31%|███▏      | 2257/7175 [28:22<53:46,  1.52it/s]
 31%|███▏      | 2258/7175 [28:23<53:45,  1.52it/s]
 31%|███▏      | 2259/7175 [28:24<53:43,  1.53it/s]
 31%|███▏      | 2260/7175 [28:24<53:40,  1.53it/s]
                                                   
{'loss': '3.293', 'grad_norm': '0.2942', 'learning_rate': '0.0006', 'epoch': '7.877'}
+
 31%|███▏      | 2260/7175 [28:24<53:40,  1.53it/s]
 32%|███▏      | 2261/7175 [28:25<53:41,  1.53it/s]
 32%|███▏      | 2262/7175 [28:26<53:39,  1.53it/s]
 32%|███▏      | 2263/7175 [28:26<53:39,  1.53it/s]
 32%|███▏      | 2264/7175 [28:27<53:37,  1.53it/s]
 32%|███▏      | 2265/7175 [28:28<53:37,  1.53it/s]
 32%|███▏      | 2266/7175 [28:28<53:36,  1.53it/s]
 32%|███▏      | 2267/7175 [28:29<53:36,  1.53it/s]
 32%|███▏      | 2268/7175 [28:30<53:36,  1.53it/s]
 32%|███▏      | 2269/7175 [28:30<53:34,  1.53it/s]
 32%|███▏      | 2270/7175 [28:31<53:36,  1.52it/s]
                                                   
{'loss': '3.303', 'grad_norm': '0.2506', 'learning_rate': '0.0006', 'epoch': '7.912'}
+
 32%|███▏      | 2270/7175 [28:31<53:36,  1.52it/s]
 32%|███▏      | 2271/7175 [28:32<53:34,  1.53it/s]
 32%|███▏      | 2272/7175 [28:32<53:32,  1.53it/s]
 32%|███▏      | 2273/7175 [28:33<53:29,  1.53it/s]
 32%|███▏      | 2274/7175 [28:34<53:32,  1.53it/s]
 32%|███▏      | 2275/7175 [28:34<53:30,  1.53it/s]
 32%|███▏      | 2276/7175 [28:35<53:28,  1.53it/s]
 32%|███▏      | 2277/7175 [28:36<53:26,  1.53it/s]
 32%|███▏      | 2278/7175 [28:36<53:28,  1.53it/s]
 32%|███▏      | 2279/7175 [28:37<53:26,  1.53it/s]
 32%|███▏      | 2280/7175 [28:38<53:25,  1.53it/s]
                                                   
{'loss': '3.295', 'grad_norm': '0.2263', 'learning_rate': '0.0006', 'epoch': '7.947'}
+
 32%|███▏      | 2280/7175 [28:38<53:25,  1.53it/s]
 32%|███▏      | 2281/7175 [28:38<53:24,  1.53it/s]
 32%|███▏      | 2282/7175 [28:39<53:27,  1.53it/s]
 32%|███▏      | 2283/7175 [28:39<53:25,  1.53it/s]
 32%|███▏      | 2284/7175 [28:40<53:24,  1.53it/s]
 32%|███▏      | 2285/7175 [28:41<53:22,  1.53it/s]
 32%|███▏      | 2286/7175 [28:41<53:25,  1.53it/s]
 32%|███▏      | 2287/7175 [28:42<53:25,  1.52it/s]
 32%|███▏      | 2288/7175 [28:43<53:23,  1.53it/s]
 32%|███▏      | 2289/7175 [28:43<53:20,  1.53it/s]
 32%|███▏      | 2290/7175 [28:44<53:19,  1.53it/s]
                                                   
{'loss': '3.297', 'grad_norm': '0.2379', 'learning_rate': '0.0006', 'epoch': '7.982'}
+
 32%|███▏      | 2290/7175 [28:44<53:19,  1.53it/s]
 32%|███▏      | 2291/7175 [28:45<53:21,  1.53it/s]
 32%|███▏      | 2292/7175 [28:45<53:19,  1.53it/s]
 32%|███▏      | 2293/7175 [28:46<53:27,  1.52it/s]
 32%|███▏      | 2294/7175 [28:47<53:22,  1.52it/s]
 32%|███▏      | 2295/7175 [28:47<53:17,  1.53it/s]
 32%|███▏      | 2297/7175 [28:48<44:31,  1.83it/s]
 32%|███▏      | 2298/7175 [28:49<46:42,  1.74it/s]
 32%|███▏      | 2299/7175 [28:50<48:24,  1.68it/s]
 32%|███▏      | 2300/7175 [28:50<49:42,  1.63it/s]
                                                   
{'loss': '3.271', 'grad_norm': '0.3178', 'learning_rate': '0.0006', 'epoch': '8.014'}
+
 32%|███▏      | 2300/7175 [28:50<49:42,  1.63it/s]
 32%|███▏      | 2301/7175 [28:51<50:40,  1.60it/s]
 32%|███▏      | 2302/7175 [28:51<51:22,  1.58it/s]
 32%|███▏      | 2303/7175 [28:52<51:56,  1.56it/s]
 32%|███▏      | 2304/7175 [28:53<52:17,  1.55it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.40it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.74it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.57it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.05it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.73it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.55it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.45it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.37it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.32it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.30it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.29it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.27it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.26it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.24it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.25it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.25it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.24it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.24it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s]
                                                   
+
                                               
{'eval_loss': '3.62', 'eval_runtime': '3.252', 'eval_samples_per_second': '750.7', 'eval_steps_per_second': '11.99', 'epoch': '8.028'}
+
 32%|███▏      | 2304/7175 [28:56<52:17,  1.55it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s][eval] epoch 8.028 step 2304 loss 3.6203 (best 3.6044 @ ep 7.779, 1/20 since best)
+
+
                                               
 32%|███▏      | 2305/7175 [28:57<2:10:23,  1.61s/it]
 32%|███▏      | 2306/7175 [28:57<1:47:31,  1.32s/it]
 32%|███▏      | 2307/7175 [28:58<1:31:19,  1.13s/it]
 32%|███▏      | 2308/7175 [28:59<1:19:53,  1.02it/s]
 32%|███▏      | 2309/7175 [28:59<1:11:53,  1.13it/s]
 32%|███▏      | 2310/7175 [29:00<1:06:17,  1.22it/s]
                                                     
{'loss': '3.258', 'grad_norm': '0.2704', 'learning_rate': '0.0006', 'epoch': '8.049'}
+
 32%|███▏      | 2310/7175 [29:00<1:06:17,  1.22it/s]
 32%|███▏      | 2311/7175 [29:01<1:02:20,  1.30it/s]
 32%|███▏      | 2312/7175 [29:01<59:32,  1.36it/s]  
 32%|███▏      | 2313/7175 [29:02<57:34,  1.41it/s]
 32%|███▏      | 2314/7175 [29:03<56:15,  1.44it/s]
 32%|███▏      | 2315/7175 [29:03<55:18,  1.46it/s]
 32%|███▏      | 2316/7175 [29:04<54:36,  1.48it/s]
 32%|███▏      | 2317/7175 [29:05<54:06,  1.50it/s]
 32%|███▏      | 2318/7175 [29:05<53:49,  1.50it/s]
 32%|███▏      | 2319/7175 [29:06<53:34,  1.51it/s]
 32%|███▏      | 2320/7175 [29:07<53:23,  1.52it/s]
                                                   
{'loss': '3.234', 'grad_norm': '0.262', 'learning_rate': '0.0006', 'epoch': '8.084'}
+
 32%|███▏      | 2320/7175 [29:07<53:23,  1.52it/s]
 32%|███▏      | 2321/7175 [29:07<53:14,  1.52it/s]
 32%|███▏      | 2322/7175 [29:08<53:11,  1.52it/s]
 32%|███▏      | 2323/7175 [29:08<53:06,  1.52it/s]
 32%|███▏      | 2324/7175 [29:09<53:03,  1.52it/s]
 32%|███▏      | 2325/7175 [29:10<52:58,  1.53it/s]
 32%|███▏      | 2326/7175 [29:10<52:59,  1.53it/s]
 32%|███▏      | 2327/7175 [29:11<52:57,  1.53it/s]
 32%|███▏      | 2328/7175 [29:12<52:54,  1.53it/s]
 32%|███▏      | 2329/7175 [29:12<52:53,  1.53it/s]
 32%|███▏      | 2330/7175 [29:13<52:55,  1.53it/s]
                                                   
{'loss': '3.236', 'grad_norm': '0.2209', 'learning_rate': '0.0006', 'epoch': '8.119'}
+
 32%|███▏      | 2330/7175 [29:13<52:55,  1.53it/s]
 32%|███▏      | 2331/7175 [29:14<52:55,  1.53it/s]
 33%|███▎      | 2332/7175 [29:14<52:53,  1.53it/s]
 33%|███▎      | 2333/7175 [29:15<52:51,  1.53it/s]
 33%|███▎      | 2334/7175 [29:16<52:51,  1.53it/s]
 33%|███▎      | 2335/7175 [29:16<52:49,  1.53it/s]
 33%|███▎      | 2336/7175 [29:17<52:47,  1.53it/s]
 33%|███▎      | 2337/7175 [29:18<52:45,  1.53it/s]
 33%|███▎      | 2338/7175 [29:18<52:49,  1.53it/s]
 33%|███▎      | 2339/7175 [29:19<52:49,  1.53it/s]
 33%|███▎      | 2340/7175 [29:20<52:49,  1.53it/s]
                                                   
{'loss': '3.238', 'grad_norm': '0.2467', 'learning_rate': '0.0006', 'epoch': '8.154'}
+
 33%|███▎      | 2340/7175 [29:20<52:49,  1.53it/s]
 33%|███▎      | 2341/7175 [29:20<52:47,  1.53it/s]
 33%|███▎      | 2342/7175 [29:21<52:46,  1.53it/s]
 33%|███▎      | 2343/7175 [29:22<52:46,  1.53it/s]
 33%|███▎      | 2344/7175 [29:22<52:46,  1.53it/s]
 33%|███▎      | 2345/7175 [29:23<52:43,  1.53it/s]
 33%|███▎      | 2346/7175 [29:24<52:42,  1.53it/s]
 33%|███▎      | 2347/7175 [29:24<52:43,  1.53it/s]
 33%|███▎      | 2348/7175 [29:25<52:42,  1.53it/s]
 33%|███▎      | 2349/7175 [29:26<52:40,  1.53it/s]
 33%|███▎      | 2350/7175 [29:26<52:40,  1.53it/s]
                                                   
{'loss': '3.247', 'grad_norm': '0.2272', 'learning_rate': '0.0006', 'epoch': '8.189'}
+
 33%|███▎      | 2350/7175 [29:26<52:40,  1.53it/s]
 33%|███▎      | 2351/7175 [29:27<52:44,  1.52it/s]
 33%|███▎      | 2352/7175 [29:27<52:41,  1.53it/s]
 33%|███▎      | 2353/7175 [29:28<52:38,  1.53it/s]
 33%|███▎      | 2354/7175 [29:29<52:36,  1.53it/s]
 33%|███▎      | 2355/7175 [29:29<52:39,  1.53it/s]
 33%|███▎      | 2356/7175 [29:30<52:38,  1.53it/s]
 33%|███▎      | 2357/7175 [29:31<52:36,  1.53it/s]
 33%|███▎      | 2358/7175 [29:31<52:34,  1.53it/s]
 33%|███▎      | 2359/7175 [29:32<52:35,  1.53it/s]
 33%|███▎      | 2360/7175 [29:33<52:34,  1.53it/s]
                                                   
{'loss': '3.251', 'grad_norm': '0.2963', 'learning_rate': '0.0006', 'epoch': '8.224'}
+
 33%|███▎      | 2360/7175 [29:33<52:34,  1.53it/s]
 33%|███▎      | 2361/7175 [29:33<52:35,  1.53it/s]
 33%|███▎      | 2362/7175 [29:34<52:32,  1.53it/s]
 33%|███▎      | 2363/7175 [29:35<52:33,  1.53it/s]
 33%|███▎      | 2364/7175 [29:35<52:33,  1.53it/s]
 33%|███▎      | 2365/7175 [29:36<52:32,  1.53it/s]
 33%|███▎      | 2366/7175 [29:37<52:28,  1.53it/s]
 33%|███▎      | 2367/7175 [29:37<52:30,  1.53it/s]
 33%|███▎      | 2368/7175 [29:38<52:31,  1.53it/s]
 33%|███▎      | 2369/7175 [29:39<52:29,  1.53it/s]
 33%|███▎      | 2370/7175 [29:39<52:27,  1.53it/s]
                                                   
{'loss': '3.244', 'grad_norm': '0.2264', 'learning_rate': '0.0006', 'epoch': '8.259'}
+
 33%|███▎      | 2370/7175 [29:39<52:27,  1.53it/s]
 33%|███▎      | 2371/7175 [29:40<52:26,  1.53it/s]
 33%|███▎      | 2372/7175 [29:41<1:00:45,  1.32it/s]
 33%|███▎      | 2373/7175 [29:42<58:16,  1.37it/s]  
 33%|███▎      | 2374/7175 [29:42<56:28,  1.42it/s]
 33%|███▎      | 2375/7175 [29:43<55:13,  1.45it/s]
 33%|███▎      | 2376/7175 [29:44<54:20,  1.47it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.66it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.84it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.63it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.08it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.74it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.55it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.44it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.37it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.33it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.30it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.27it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.26it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.25it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.25it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.24it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.24it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.23it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.23it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s]
                                                   
+
                                               
{'eval_loss': '3.597', 'eval_runtime': '3.248', 'eval_samples_per_second': '751.6', 'eval_steps_per_second': '12.01', 'epoch': '8.28'}
+
 33%|███▎      | 2376/7175 [29:47<54:20,  1.47it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-2376
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-2376/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-2376/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.70it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.70it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-2376/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-2376/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-2376/tokenizer_config.json
+[eval] epoch 8.28 step 2376 loss 3.5975 (best 3.5975 @ ep 8.28, 0/20 since best)
+
+
                                               
 33%|███▎      | 2377/7175 [29:50<3:02:26,  2.28s/it]
 33%|███▎      | 2378/7175 [29:50<2:23:21,  1.79s/it]
 33%|███▎      | 2379/7175 [29:51<1:56:01,  1.45s/it]
 33%|███▎      | 2380/7175 [29:52<1:36:52,  1.21s/it]
                                                     
{'loss': '3.262', 'grad_norm': '0.2411', 'learning_rate': '0.0006', 'epoch': '8.294'}
+
 33%|███▎      | 2380/7175 [29:52<1:36:52,  1.21s/it]
 33%|███▎      | 2381/7175 [29:52<1:23:33,  1.05s/it]
 33%|███▎      | 2382/7175 [29:53<1:14:09,  1.08it/s]
 33%|███▎      | 2383/7175 [29:54<1:07:35,  1.18it/s]
 33%|███▎      | 2384/7175 [29:54<1:03:00,  1.27it/s]
 33%|███▎      | 2385/7175 [29:55<59:49,  1.33it/s]  
 33%|███▎      | 2386/7175 [29:55<57:32,  1.39it/s]
 33%|███▎      | 2387/7175 [29:56<55:59,  1.43it/s]
 33%|███▎      | 2388/7175 [29:57<54:51,  1.45it/s]
 33%|███▎      | 2389/7175 [29:57<54:03,  1.48it/s]
 33%|███▎      | 2390/7175 [29:58<53:32,  1.49it/s]
                                                   
{'loss': '3.247', 'grad_norm': '0.2483', 'learning_rate': '0.0006', 'epoch': '8.329'}
+
 33%|███▎      | 2390/7175 [29:58<53:32,  1.49it/s]
 33%|███▎      | 2391/7175 [29:59<53:08,  1.50it/s]
 33%|███▎      | 2392/7175 [29:59<52:49,  1.51it/s]
 33%|███▎      | 2393/7175 [30:00<52:36,  1.52it/s]
 33%|███▎      | 2394/7175 [30:01<52:31,  1.52it/s]
 33%|███▎      | 2395/7175 [30:01<52:25,  1.52it/s]
 33%|███▎      | 2396/7175 [30:02<52:18,  1.52it/s]
 33%|███▎      | 2397/7175 [30:03<52:15,  1.52it/s]
 33%|███▎      | 2398/7175 [30:03<52:16,  1.52it/s]
 33%|███▎      | 2399/7175 [30:04<52:13,  1.52it/s]
 33%|███▎      | 2400/7175 [30:05<52:11,  1.52it/s]
                                                   
{'loss': '3.246', 'grad_norm': '0.2358', 'learning_rate': '0.0006', 'epoch': '8.363'}
+
 33%|███▎      | 2400/7175 [30:05<52:11,  1.52it/s]
 33%|███▎      | 2401/7175 [30:05<52:09,  1.53it/s]
 33%|███▎      | 2402/7175 [30:06<52:09,  1.53it/s]
 33%|███▎      | 2403/7175 [30:07<52:09,  1.52it/s]
 34%|███▎      | 2404/7175 [30:07<52:07,  1.53it/s]
 34%|███▎      | 2405/7175 [30:08<52:06,  1.53it/s]
 34%|███▎      | 2406/7175 [30:09<52:04,  1.53it/s]
 34%|███▎      | 2407/7175 [30:09<52:05,  1.53it/s]
 34%|███▎      | 2408/7175 [30:10<52:05,  1.53it/s]
 34%|███▎      | 2409/7175 [30:11<52:02,  1.53it/s]
 34%|███▎      | 2410/7175 [30:11<52:02,  1.53it/s]
                                                   
{'loss': '3.258', 'grad_norm': '0.2433', 'learning_rate': '0.0006', 'epoch': '8.398'}
+
 34%|███▎      | 2410/7175 [30:11<52:02,  1.53it/s]
 34%|███▎      | 2411/7175 [30:12<52:04,  1.52it/s]
 34%|███▎      | 2412/7175 [30:13<52:04,  1.52it/s]
 34%|███▎      | 2413/7175 [30:13<52:03,  1.52it/s]
 34%|███▎      | 2414/7175 [30:14<52:00,  1.53it/s]
 34%|███▎      | 2415/7175 [30:14<51:58,  1.53it/s]
 34%|███▎      | 2416/7175 [30:15<52:00,  1.53it/s]
 34%|███▎      | 2417/7175 [30:16<51:58,  1.53it/s]
 34%|███▎      | 2418/7175 [30:16<51:55,  1.53it/s]
 34%|███▎      | 2419/7175 [30:17<51:54,  1.53it/s]
 34%|███▎      | 2420/7175 [30:18<51:56,  1.53it/s]
                                                   
{'loss': '3.259', 'grad_norm': '0.252', 'learning_rate': '0.0006', 'epoch': '8.433'}
+
 34%|███▎      | 2420/7175 [30:18<51:56,  1.53it/s]
 34%|███▎      | 2421/7175 [30:18<51:57,  1.53it/s]
 34%|███▍      | 2422/7175 [30:19<51:55,  1.53it/s]
 34%|███▍      | 2423/7175 [30:20<51:52,  1.53it/s]
 34%|███▍      | 2424/7175 [30:20<51:53,  1.53it/s]
 34%|███▍      | 2425/7175 [30:21<51:52,  1.53it/s]
 34%|███▍      | 2426/7175 [30:22<51:51,  1.53it/s]
 34%|███▍      | 2427/7175 [30:22<51:47,  1.53it/s]
 34%|███▍      | 2428/7175 [30:23<51:51,  1.53it/s]
 34%|███▍      | 2429/7175 [30:24<51:49,  1.53it/s]
 34%|███▍      | 2430/7175 [30:24<51:47,  1.53it/s]
                                                   
{'loss': '3.232', 'grad_norm': '0.2149', 'learning_rate': '0.0006', 'epoch': '8.468'}
+
 34%|███▍      | 2430/7175 [30:24<51:47,  1.53it/s]
 34%|███▍      | 2431/7175 [30:25<51:46,  1.53it/s]
 34%|███▍      | 2432/7175 [30:26<51:48,  1.53it/s]
 34%|███▍      | 2433/7175 [30:26<51:47,  1.53it/s]
 34%|███▍      | 2434/7175 [30:27<51:45,  1.53it/s]
 34%|███▍      | 2435/7175 [30:28<51:43,  1.53it/s]
 34%|███▍      | 2436/7175 [30:28<51:44,  1.53it/s]
 34%|███▍      | 2437/7175 [30:29<51:43,  1.53it/s]
 34%|███▍      | 2438/7175 [30:30<51:42,  1.53it/s]
 34%|███▍      | 2439/7175 [30:30<51:39,  1.53it/s]
 34%|███▍      | 2440/7175 [30:31<51:42,  1.53it/s]
                                                   
{'loss': '3.264', 'grad_norm': '0.2327', 'learning_rate': '0.0006', 'epoch': '8.503'}
+
 34%|███▍      | 2440/7175 [30:31<51:42,  1.53it/s]
 34%|███▍      | 2441/7175 [30:32<51:42,  1.53it/s]
 34%|███▍      | 2442/7175 [30:32<51:42,  1.53it/s]
 34%|███▍      | 2443/7175 [30:33<51:39,  1.53it/s]
 34%|███▍      | 2444/7175 [30:33<51:39,  1.53it/s]
 34%|███▍      | 2445/7175 [30:34<51:41,  1.52it/s]
 34%|███▍      | 2446/7175 [30:35<51:40,  1.53it/s]
 34%|███▍      | 2447/7175 [30:35<51:39,  1.53it/s]
 34%|███▍      | 2448/7175 [30:36<51:38,  1.53it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.07it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.89it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.48it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.12it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.66it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.63it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.37it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.44it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.23it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.35it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.33it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.30it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.28it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.26it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.25it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.25it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.24it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.23it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s]
                                                   
+
                                               
{'eval_loss': '3.586', 'eval_runtime': '3.255', 'eval_samples_per_second': '749.9', 'eval_steps_per_second': '11.98', 'epoch': '8.531'}
+
 34%|███▍      | 2448/7175 [30:39<51:38,  1.53it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-2448
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-2448/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-2448/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.85it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.84it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-2448/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-2448/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-2448/tokenizer_config.json
+[eval] epoch 8.531 step 2448 loss 3.5862 (best 3.5862 @ ep 8.531, 0/20 since best)
+
+
                                               
 34%|███▍      | 2449/7175 [30:42<2:57:35,  2.25s/it]
 34%|███▍      | 2450/7175 [30:43<2:19:44,  1.77s/it]
                                                     
{'loss': '3.243', 'grad_norm': '0.211', 'learning_rate': '0.0006', 'epoch': '8.538'}
+
 34%|███▍      | 2450/7175 [30:43<2:19:44,  1.77s/it]
 34%|███▍      | 2451/7175 [30:43<1:53:17,  1.44s/it]
 34%|███▍      | 2452/7175 [30:44<1:34:44,  1.20s/it]
 34%|███▍      | 2453/7175 [30:45<1:21:48,  1.04s/it]
 34%|███▍      | 2454/7175 [30:45<1:12:43,  1.08it/s]
 34%|███▍      | 2455/7175 [30:46<1:06:22,  1.19it/s]
 34%|███▍      | 2456/7175 [30:47<1:01:56,  1.27it/s]
 34%|███▍      | 2457/7175 [30:47<58:47,  1.34it/s]  
 34%|███▍      | 2458/7175 [30:48<56:35,  1.39it/s]
 34%|███▍      | 2459/7175 [30:49<55:03,  1.43it/s]
 34%|███▍      | 2460/7175 [30:49<54:01,  1.45it/s]
                                                   
{'loss': '3.252', 'grad_norm': '0.2308', 'learning_rate': '0.0006', 'epoch': '8.573'}
+
 34%|███▍      | 2460/7175 [30:49<54:01,  1.45it/s]
 34%|███▍      | 2461/7175 [30:50<53:18,  1.47it/s]
 34%|███▍      | 2462/7175 [30:51<52:43,  1.49it/s]
 34%|███▍      | 2463/7175 [30:51<52:22,  1.50it/s]
 34%|███▍      | 2464/7175 [30:52<52:05,  1.51it/s]
 34%|███▍      | 2465/7175 [30:53<51:54,  1.51it/s]
 34%|███▍      | 2466/7175 [30:53<51:47,  1.52it/s]
 34%|███▍      | 2467/7175 [30:54<51:40,  1.52it/s]
 34%|███▍      | 2468/7175 [30:55<51:38,  1.52it/s]
 34%|███▍      | 2469/7175 [30:55<51:33,  1.52it/s]
 34%|███▍      | 2470/7175 [30:56<51:29,  1.52it/s]
                                                   
{'loss': '3.255', 'grad_norm': '0.2674', 'learning_rate': '0.0006', 'epoch': '8.608'}
+
 34%|███▍      | 2470/7175 [30:56<51:29,  1.52it/s]
 34%|███▍      | 2471/7175 [30:57<51:31,  1.52it/s]
 34%|███▍      | 2472/7175 [30:57<51:27,  1.52it/s]
 34%|███▍      | 2473/7175 [30:58<51:28,  1.52it/s]
 34%|███▍      | 2474/7175 [30:58<51:26,  1.52it/s]
 34%|███▍      | 2475/7175 [30:59<51:24,  1.52it/s]
 35%|███▍      | 2476/7175 [31:00<51:23,  1.52it/s]
 35%|███▍      | 2477/7175 [31:00<51:21,  1.52it/s]
 35%|███▍      | 2478/7175 [31:01<51:18,  1.53it/s]
 35%|███▍      | 2479/7175 [31:02<51:15,  1.53it/s]
 35%|███▍      | 2480/7175 [31:02<51:18,  1.53it/s]
                                                   
{'loss': '3.255', 'grad_norm': '0.2347', 'learning_rate': '0.0006', 'epoch': '8.643'}
+
 35%|███▍      | 2480/7175 [31:02<51:18,  1.53it/s]
 35%|███▍      | 2481/7175 [31:03<51:18,  1.52it/s]
 35%|███▍      | 2482/7175 [31:04<59:15,  1.32it/s]
 35%|███▍      | 2483/7175 [31:05<56:47,  1.38it/s]
 35%|███▍      | 2484/7175 [31:05<55:07,  1.42it/s]
 35%|███▍      | 2485/7175 [31:06<53:55,  1.45it/s]
 35%|███▍      | 2486/7175 [31:07<53:05,  1.47it/s]
 35%|███▍      | 2487/7175 [31:07<52:28,  1.49it/s]
 35%|███▍      | 2488/7175 [31:08<52:06,  1.50it/s]
 35%|███▍      | 2489/7175 [31:09<51:47,  1.51it/s]
 35%|███▍      | 2490/7175 [31:09<51:39,  1.51it/s]
                                                   
{'loss': '3.262', 'grad_norm': '0.2011', 'learning_rate': '0.0006', 'epoch': '8.678'}
+
 35%|███▍      | 2490/7175 [31:09<51:39,  1.51it/s]
 35%|███▍      | 2491/7175 [31:10<51:29,  1.52it/s]
 35%|███▍      | 2492/7175 [31:11<51:21,  1.52it/s]
 35%|███▍      | 2493/7175 [31:11<51:16,  1.52it/s]
 35%|███▍      | 2494/7175 [31:12<51:11,  1.52it/s]
 35%|███▍      | 2495/7175 [31:13<51:06,  1.53it/s]
 35%|███▍      | 2496/7175 [31:13<51:08,  1.52it/s]
 35%|███▍      | 2497/7175 [31:14<51:05,  1.53it/s]
 35%|███▍      | 2498/7175 [31:15<51:01,  1.53it/s]
 35%|███▍      | 2499/7175 [31:15<50:59,  1.53it/s]
 35%|███▍      | 2500/7175 [31:16<51:00,  1.53it/s]
                                                   
{'loss': '3.247', 'grad_norm': '0.2232', 'learning_rate': '0.0006', 'epoch': '8.713'}
+
 35%|███▍      | 2500/7175 [31:16<51:00,  1.53it/s]
 35%|███▍      | 2501/7175 [31:17<51:01,  1.53it/s]
 35%|███▍      | 2502/7175 [31:17<50:57,  1.53it/s]
 35%|███▍      | 2503/7175 [31:18<50:58,  1.53it/s]
 35%|███▍      | 2504/7175 [31:18<50:59,  1.53it/s]
 35%|███▍      | 2505/7175 [31:19<50:58,  1.53it/s]
 35%|███▍      | 2506/7175 [31:20<50:54,  1.53it/s]
 35%|███▍      | 2507/7175 [31:20<50:56,  1.53it/s]
 35%|███▍      | 2508/7175 [31:21<50:56,  1.53it/s]
 35%|███▍      | 2509/7175 [31:22<50:53,  1.53it/s]
 35%|███▍      | 2510/7175 [31:22<50:52,  1.53it/s]
                                                   
{'loss': '3.245', 'grad_norm': '0.2425', 'learning_rate': '0.0006', 'epoch': '8.748'}
+
 35%|███▍      | 2510/7175 [31:22<50:52,  1.53it/s]
 35%|███▍      | 2511/7175 [31:23<50:56,  1.53it/s]
 35%|███▌      | 2512/7175 [31:24<50:54,  1.53it/s]
 35%|███▌      | 2513/7175 [31:24<50:52,  1.53it/s]
 35%|███▌      | 2514/7175 [31:25<50:51,  1.53it/s]
 35%|███▌      | 2515/7175 [31:26<50:53,  1.53it/s]
 35%|███▌      | 2516/7175 [31:26<50:52,  1.53it/s]
 35%|███▌      | 2517/7175 [31:27<58:58,  1.32it/s]
 35%|███▌      | 2518/7175 [31:28<56:31,  1.37it/s]
 35%|███▌      | 2519/7175 [31:29<54:49,  1.42it/s]
 35%|███▌      | 2520/7175 [31:30<1:01:38,  1.26it/s]
                                                     
{'loss': '3.264', 'grad_norm': '0.2448', 'learning_rate': '0.0006', 'epoch': '8.783'}
+
 35%|███▌      | 2520/7175 [31:30<1:01:38,  1.26it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.53it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.82it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.60it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.08it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.73it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.59it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.44it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.38it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.31it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.29it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.27it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.26it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.24it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.23it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.23it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.23it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.23it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.23it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.22it/s]
                                                     
+
                                               
{'eval_loss': '3.58', 'eval_runtime': '3.245', 'eval_samples_per_second': '752.3', 'eval_steps_per_second': '12.02', 'epoch': '8.783'}
+
 35%|███▌      | 2520/7175 [31:33<1:01:38,  1.26it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.22it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-2520
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-2520/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-2520/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  2.10it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  2.09it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-2520/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-2520/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-2520/tokenizer_config.json
+[eval] epoch 8.783 step 2520 loss 3.5801 (best 3.5801 @ ep 8.783, 0/20 since best)
+
+
                                               
 35%|███▌      | 2521/7175 [31:36<3:00:44,  2.33s/it]
 35%|███▌      | 2522/7175 [31:36<2:21:44,  1.83s/it]
 35%|███▌      | 2523/7175 [31:37<1:54:27,  1.48s/it]
 35%|███▌      | 2524/7175 [31:38<1:35:21,  1.23s/it]
 35%|███▌      | 2525/7175 [31:38<1:21:57,  1.06s/it]
 35%|███▌      | 2526/7175 [31:39<1:12:36,  1.07it/s]
 35%|███▌      | 2527/7175 [31:39<1:06:01,  1.17it/s]
 35%|███▌      | 2528/7175 [31:40<1:01:23,  1.26it/s]
 35%|███▌      | 2529/7175 [31:41<58:10,  1.33it/s]  
 35%|███▌      | 2530/7175 [31:41<55:57,  1.38it/s]
                                                   
{'loss': '3.269', 'grad_norm': '0.2068', 'learning_rate': '0.0006', 'epoch': '8.818'}
+
 35%|███▌      | 2530/7175 [31:41<55:57,  1.38it/s]
 35%|███▌      | 2531/7175 [31:42<1:02:29,  1.24it/s]
 35%|███▌      | 2532/7175 [31:43<58:57,  1.31it/s]  
 35%|███▌      | 2533/7175 [31:44<56:28,  1.37it/s]
 35%|███▌      | 2534/7175 [31:44<54:41,  1.41it/s]
 35%|███▌      | 2535/7175 [31:45<53:28,  1.45it/s]
 35%|███▌      | 2536/7175 [31:46<52:38,  1.47it/s]
 35%|███▌      | 2537/7175 [31:47<59:47,  1.29it/s]
 35%|███▌      | 2538/7175 [31:47<56:58,  1.36it/s]
 35%|███▌      | 2539/7175 [31:48<55:05,  1.40it/s]
 35%|███▌      | 2540/7175 [31:49<53:44,  1.44it/s]
                                                   
{'loss': '3.257', 'grad_norm': '0.2261', 'learning_rate': '0.0006', 'epoch': '8.853'}
+
 35%|███▌      | 2540/7175 [31:49<53:44,  1.44it/s]
 35%|███▌      | 2541/7175 [31:49<52:49,  1.46it/s]
 35%|███▌      | 2542/7175 [31:50<1:00:00,  1.29it/s]
 35%|███▌      | 2543/7175 [31:51<57:11,  1.35it/s]  
 35%|███▌      | 2544/7175 [31:52<55:10,  1.40it/s]
 35%|███▌      | 2545/7175 [31:52<53:46,  1.44it/s]
 35%|███▌      | 2546/7175 [31:53<52:45,  1.46it/s]
 35%|███▌      | 2547/7175 [31:54<52:06,  1.48it/s]
 36%|███▌      | 2548/7175 [31:54<51:37,  1.49it/s]
 36%|███▌      | 2549/7175 [31:55<51:16,  1.50it/s]
 36%|███▌      | 2550/7175 [31:56<51:00,  1.51it/s]
                                                   
{'loss': '3.266', 'grad_norm': '0.2704', 'learning_rate': '0.0006', 'epoch': '8.888'}
+
 36%|███▌      | 2550/7175 [31:56<51:00,  1.51it/s]
 36%|███▌      | 2551/7175 [31:56<50:52,  1.51it/s]
 36%|███▌      | 2552/7175 [31:57<50:46,  1.52it/s]
 36%|███▌      | 2553/7175 [31:58<50:40,  1.52it/s]
 36%|███▌      | 2554/7175 [31:58<50:34,  1.52it/s]
 36%|███▌      | 2555/7175 [31:59<50:31,  1.52it/s]
 36%|███▌      | 2556/7175 [32:00<50:30,  1.52it/s]
 36%|███▌      | 2557/7175 [32:00<50:28,  1.52it/s]
 36%|███▌      | 2558/7175 [32:01<50:25,  1.53it/s]
 36%|███▌      | 2559/7175 [32:01<50:24,  1.53it/s]
 36%|███▌      | 2560/7175 [32:02<50:26,  1.52it/s]
                                                   
{'loss': '3.273', 'grad_norm': '0.2428', 'learning_rate': '0.0006', 'epoch': '8.923'}
+
 36%|███▌      | 2560/7175 [32:02<50:26,  1.52it/s]
 36%|███▌      | 2561/7175 [32:03<50:26,  1.52it/s]
 36%|███▌      | 2562/7175 [32:03<50:26,  1.52it/s]
 36%|███▌      | 2563/7175 [32:04<50:21,  1.53it/s]
 36%|███▌      | 2564/7175 [32:05<50:22,  1.53it/s]
 36%|███▌      | 2565/7175 [32:05<50:21,  1.53it/s]
 36%|███▌      | 2566/7175 [32:06<50:20,  1.53it/s]
 36%|███▌      | 2567/7175 [32:07<50:18,  1.53it/s]
 36%|███▌      | 2568/7175 [32:07<50:17,  1.53it/s]
 36%|███▌      | 2569/7175 [32:08<50:17,  1.53it/s]
 36%|███▌      | 2570/7175 [32:09<50:17,  1.53it/s]
                                                   
{'loss': '3.257', 'grad_norm': '0.2598', 'learning_rate': '0.0006', 'epoch': '8.958'}
+
 36%|███▌      | 2570/7175 [32:09<50:17,  1.53it/s]
 36%|███▌      | 2571/7175 [32:09<50:17,  1.53it/s]
 36%|███▌      | 2572/7175 [32:10<50:16,  1.53it/s]
 36%|███▌      | 2573/7175 [32:11<50:17,  1.53it/s]
 36%|███▌      | 2574/7175 [32:11<50:15,  1.53it/s]
 36%|███▌      | 2575/7175 [32:12<50:12,  1.53it/s]
 36%|███▌      | 2576/7175 [32:13<50:11,  1.53it/s]
 36%|███▌      | 2577/7175 [32:13<50:11,  1.53it/s]
 36%|███▌      | 2578/7175 [32:14<50:12,  1.53it/s]
 36%|███▌      | 2579/7175 [32:15<50:09,  1.53it/s]
 36%|███▌      | 2580/7175 [32:15<50:16,  1.52it/s]
                                                   
{'loss': '3.259', 'grad_norm': '0.2267', 'learning_rate': '0.0006', 'epoch': '8.993'}
+
 36%|███▌      | 2580/7175 [32:15<50:16,  1.52it/s]
 36%|███▌      | 2581/7175 [32:16<50:13,  1.52it/s]
 36%|███▌      | 2582/7175 [32:17<50:10,  1.53it/s]
 36%|███▌      | 2584/7175 [32:17<41:56,  1.82it/s]
 36%|███▌      | 2585/7175 [32:18<43:57,  1.74it/s]
 36%|███▌      | 2586/7175 [32:19<45:33,  1.68it/s]
 36%|███▌      | 2587/7175 [32:19<46:46,  1.63it/s]
 36%|███▌      | 2588/7175 [32:20<47:41,  1.60it/s]
 36%|███▌      | 2589/7175 [32:21<48:22,  1.58it/s]
 36%|███▌      | 2590/7175 [32:21<48:51,  1.56it/s]
                                                   
{'loss': '3.231', 'grad_norm': '0.3857', 'learning_rate': '0.0006', 'epoch': '9.024'}
+
 36%|███▌      | 2590/7175 [32:21<48:51,  1.56it/s]
 36%|███▌      | 2591/7175 [32:22<49:11,  1.55it/s]
 36%|███▌      | 2592/7175 [32:23<49:25,  1.55it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.22it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.69it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.57it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.03it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.73it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.58it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.45it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.37it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.32it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.28it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.28it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.27it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.26it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.25it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.25it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.24it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.24it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.24it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s]
                                                   
+
                                               
{'eval_loss': '3.606', 'eval_runtime': '3.26', 'eval_samples_per_second': '748.7', 'eval_steps_per_second': '11.96', 'epoch': '9.031'}
+
 36%|███▌      | 2592/7175 [32:26<49:25,  1.55it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s][eval] epoch 9.031 step 2592 loss 3.6055 (best 3.5801 @ ep 8.783, 1/20 since best)
+
+
                                               
 36%|███▌      | 2593/7175 [32:27<2:03:24,  1.62s/it]
 36%|███▌      | 2594/7175 [32:27<1:41:32,  1.33s/it]
 36%|███▌      | 2595/7175 [32:28<1:26:08,  1.13s/it]
 36%|███▌      | 2596/7175 [32:29<1:15:19,  1.01it/s]
 36%|███▌      | 2597/7175 [32:29<1:07:44,  1.13it/s]
 36%|███▌      | 2598/7175 [32:30<1:02:24,  1.22it/s]
 36%|███▌      | 2599/7175 [32:30<58:39,  1.30it/s]  
 36%|███▌      | 2600/7175 [32:31<56:03,  1.36it/s]
                                                   
{'loss': '3.208', 'grad_norm': '0.2435', 'learning_rate': '0.0006', 'epoch': '9.059'}
+
 36%|███▌      | 2600/7175 [32:31<56:03,  1.36it/s]
 36%|███▋      | 2601/7175 [32:32<54:17,  1.40it/s]
 36%|███▋      | 2602/7175 [32:32<53:00,  1.44it/s]
 36%|███▋      | 2603/7175 [32:33<52:05,  1.46it/s]
 36%|███▋      | 2604/7175 [32:34<51:23,  1.48it/s]
 36%|███▋      | 2605/7175 [32:34<50:55,  1.50it/s]
 36%|███▋      | 2606/7175 [32:35<50:35,  1.51it/s]
 36%|███▋      | 2607/7175 [32:36<50:22,  1.51it/s]
 36%|███▋      | 2608/7175 [32:36<50:10,  1.52it/s]
 36%|███▋      | 2609/7175 [32:37<50:06,  1.52it/s]
 36%|███▋      | 2610/7175 [32:38<50:01,  1.52it/s]
                                                   
{'loss': '3.203', 'grad_norm': '0.2534', 'learning_rate': '0.0006', 'epoch': '9.094'}
+
 36%|███▋      | 2610/7175 [32:38<50:01,  1.52it/s]
 36%|███▋      | 2611/7175 [32:38<50:00,  1.52it/s]
 36%|███▋      | 2612/7175 [32:39<49:55,  1.52it/s]
 36%|███▋      | 2613/7175 [32:40<49:51,  1.53it/s]
 36%|███▋      | 2614/7175 [32:40<49:52,  1.52it/s]
 36%|███▋      | 2615/7175 [32:41<49:51,  1.52it/s]
 36%|███▋      | 2616/7175 [32:42<49:49,  1.52it/s]
 36%|███▋      | 2617/7175 [32:42<49:47,  1.53it/s]
 36%|███▋      | 2618/7175 [32:43<49:47,  1.53it/s]
 37%|███▋      | 2619/7175 [32:44<49:47,  1.53it/s]
 37%|███▋      | 2620/7175 [32:44<49:45,  1.53it/s]
                                                   
{'loss': '3.217', 'grad_norm': '0.2171', 'learning_rate': '0.0006', 'epoch': '9.129'}
+
 37%|███▋      | 2620/7175 [32:44<49:45,  1.53it/s]
 37%|███▋      | 2621/7175 [32:45<49:46,  1.53it/s]
 37%|███▋      | 2622/7175 [32:46<49:44,  1.53it/s]
 37%|███▋      | 2623/7175 [32:46<49:46,  1.52it/s]
 37%|███▋      | 2624/7175 [32:47<49:44,  1.52it/s]
 37%|███▋      | 2625/7175 [32:48<49:44,  1.52it/s]
 37%|███▋      | 2626/7175 [32:48<49:43,  1.52it/s]
 37%|███▋      | 2627/7175 [32:49<49:41,  1.53it/s]
 37%|███▋      | 2628/7175 [32:49<49:43,  1.52it/s]
 37%|███▋      | 2629/7175 [32:50<49:43,  1.52it/s]
 37%|███▋      | 2630/7175 [32:51<49:42,  1.52it/s]
                                                   
{'loss': '3.194', 'grad_norm': '0.2107', 'learning_rate': '0.0006', 'epoch': '9.164'}
+
 37%|███▋      | 2630/7175 [32:51<49:42,  1.52it/s]
 37%|███▋      | 2631/7175 [32:51<49:42,  1.52it/s]
 37%|███▋      | 2632/7175 [32:52<49:42,  1.52it/s]
 37%|███▋      | 2633/7175 [32:53<49:40,  1.52it/s]
 37%|███▋      | 2634/7175 [32:53<49:41,  1.52it/s]
 37%|███▋      | 2635/7175 [32:54<49:38,  1.52it/s]
 37%|███▋      | 2636/7175 [32:55<49:35,  1.53it/s]
 37%|███▋      | 2637/7175 [32:55<49:33,  1.53it/s]
 37%|███▋      | 2638/7175 [32:56<49:35,  1.52it/s]
 37%|███▋      | 2639/7175 [32:57<49:34,  1.53it/s]
 37%|███▋      | 2640/7175 [32:57<49:33,  1.53it/s]
                                                   
{'loss': '3.201', 'grad_norm': '0.2259', 'learning_rate': '0.0006', 'epoch': '9.199'}
+
 37%|███▋      | 2640/7175 [32:57<49:33,  1.53it/s]
 37%|███▋      | 2641/7175 [32:58<49:31,  1.53it/s]
 37%|███▋      | 2642/7175 [32:59<49:30,  1.53it/s]
 37%|███▋      | 2643/7175 [32:59<49:31,  1.53it/s]
 37%|███▋      | 2644/7175 [33:00<49:30,  1.53it/s]
 37%|███▋      | 2645/7175 [33:01<49:28,  1.53it/s]
 37%|███▋      | 2646/7175 [33:01<49:27,  1.53it/s]
 37%|███▋      | 2647/7175 [33:02<49:27,  1.53it/s]
 37%|███▋      | 2648/7175 [33:03<49:27,  1.53it/s]
 37%|███▋      | 2649/7175 [33:03<49:24,  1.53it/s]
 37%|███▋      | 2650/7175 [33:04<49:25,  1.53it/s]
                                                   
{'loss': '3.191', 'grad_norm': '0.2464', 'learning_rate': '0.0006', 'epoch': '9.234'}
+
 37%|███▋      | 2650/7175 [33:04<49:25,  1.53it/s]
 37%|███▋      | 2651/7175 [33:05<49:28,  1.52it/s]
 37%|███▋      | 2652/7175 [33:05<49:26,  1.52it/s]
 37%|███▋      | 2653/7175 [33:06<49:25,  1.52it/s]
 37%|███▋      | 2654/7175 [33:07<49:22,  1.53it/s]
 37%|███▋      | 2655/7175 [33:07<49:21,  1.53it/s]
 37%|███▋      | 2656/7175 [33:08<49:22,  1.53it/s]
 37%|███▋      | 2657/7175 [33:09<49:21,  1.53it/s]
 37%|███▋      | 2658/7175 [33:09<49:20,  1.53it/s]
 37%|███▋      | 2659/7175 [33:10<49:17,  1.53it/s]
 37%|███▋      | 2660/7175 [33:10<49:19,  1.53it/s]
                                                   
{'loss': '3.211', 'grad_norm': '0.2159', 'learning_rate': '0.0006', 'epoch': '9.269'}
+
 37%|███▋      | 2660/7175 [33:10<49:19,  1.53it/s]
 37%|███▋      | 2661/7175 [33:11<49:20,  1.52it/s]
 37%|███▋      | 2662/7175 [33:12<49:18,  1.53it/s]
 37%|███▋      | 2663/7175 [33:12<49:15,  1.53it/s]
 37%|███▋      | 2664/7175 [33:13<49:16,  1.53it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.54it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.76it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.61it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.05it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.76it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.56it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.47it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.37it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.32it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.28it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.29it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.25it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.25it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.22it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.24it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.22it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.24it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.23it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s]
                                                   
+
                                               
{'eval_loss': '3.574', 'eval_runtime': '3.248', 'eval_samples_per_second': '751.6', 'eval_steps_per_second': '12.01', 'epoch': '9.283'}
+
 37%|███▋      | 2664/7175 [33:16<49:16,  1.53it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-2664
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-2664/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-2664/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.83it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.83it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-2664/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-2664/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-2664/tokenizer_config.json
+[eval] epoch 9.283 step 2664 loss 3.5735 (best 3.5735 @ ep 9.283, 0/20 since best)
+
+
                                               
 37%|███▋      | 2665/7175 [33:19<2:50:37,  2.27s/it]
 37%|███▋      | 2666/7175 [33:20<2:14:09,  1.79s/it]
 37%|███▋      | 2667/7175 [33:20<1:48:37,  1.45s/it]
 37%|███▋      | 2668/7175 [33:21<1:30:47,  1.21s/it]
 37%|███▋      | 2669/7175 [33:22<1:18:17,  1.04s/it]
 37%|███▋      | 2670/7175 [33:22<1:09:32,  1.08it/s]
                                                     
{'loss': '3.188', 'grad_norm': '0.2266', 'learning_rate': '0.0006', 'epoch': '9.304'}
+
 37%|███▋      | 2670/7175 [33:22<1:09:32,  1.08it/s]
 37%|███▋      | 2671/7175 [33:23<1:03:25,  1.18it/s]
 37%|███▋      | 2672/7175 [33:24<59:07,  1.27it/s]  
 37%|███▋      | 2673/7175 [33:24<56:07,  1.34it/s]
 37%|███▋      | 2674/7175 [33:25<54:01,  1.39it/s]
 37%|███▋      | 2675/7175 [33:26<52:30,  1.43it/s]
 37%|███▋      | 2676/7175 [33:26<51:30,  1.46it/s]
 37%|███▋      | 2677/7175 [33:27<50:48,  1.48it/s]
 37%|███▋      | 2678/7175 [33:28<50:17,  1.49it/s]
 37%|███▋      | 2679/7175 [33:28<49:52,  1.50it/s]
 37%|███▋      | 2680/7175 [33:29<49:38,  1.51it/s]
                                                   
{'loss': '3.191', 'grad_norm': '0.2453', 'learning_rate': '0.0006', 'epoch': '9.339'}
+
 37%|███▋      | 2680/7175 [33:29<49:38,  1.51it/s]
 37%|███▋      | 2681/7175 [33:30<49:30,  1.51it/s]
 37%|███▋      | 2682/7175 [33:30<49:21,  1.52it/s]
 37%|███▋      | 2683/7175 [33:31<49:13,  1.52it/s]
 37%|███▋      | 2684/7175 [33:32<49:09,  1.52it/s]
 37%|███▋      | 2685/7175 [33:32<49:06,  1.52it/s]
 37%|███▋      | 2686/7175 [33:33<49:05,  1.52it/s]
 37%|███▋      | 2687/7175 [33:34<49:02,  1.53it/s]
 37%|███▋      | 2688/7175 [33:34<49:00,  1.53it/s]
 37%|███▋      | 2689/7175 [33:35<49:01,  1.52it/s]
 37%|███▋      | 2690/7175 [33:36<48:59,  1.53it/s]
                                                   
{'loss': '3.213', 'grad_norm': '0.2434', 'learning_rate': '0.0006', 'epoch': '9.374'}
+
 37%|███▋      | 2690/7175 [33:36<48:59,  1.53it/s]
 38%|███▊      | 2691/7175 [33:36<48:58,  1.53it/s]
 38%|███▊      | 2692/7175 [33:37<48:56,  1.53it/s]
 38%|███▊      | 2693/7175 [33:37<48:58,  1.53it/s]
 38%|███▊      | 2694/7175 [33:38<48:56,  1.53it/s]
 38%|███▊      | 2695/7175 [33:39<48:53,  1.53it/s]
 38%|███▊      | 2696/7175 [33:39<48:52,  1.53it/s]
 38%|███▊      | 2697/7175 [33:40<48:53,  1.53it/s]
 38%|███▊      | 2698/7175 [33:41<48:52,  1.53it/s]
 38%|███▊      | 2699/7175 [33:41<48:50,  1.53it/s]
 38%|███▊      | 2700/7175 [33:42<48:49,  1.53it/s]
                                                   
{'loss': '3.22', 'grad_norm': '0.277', 'learning_rate': '0.0006', 'epoch': '9.409'}
+
 38%|███▊      | 2700/7175 [33:42<48:49,  1.53it/s]
 38%|███▊      | 2701/7175 [33:43<48:53,  1.53it/s]
 38%|███▊      | 2702/7175 [33:43<48:52,  1.53it/s]
 38%|███▊      | 2703/7175 [33:44<49:00,  1.52it/s]
 38%|███▊      | 2704/7175 [33:45<48:57,  1.52it/s]
 38%|███▊      | 2705/7175 [33:45<48:53,  1.52it/s]
 38%|███▊      | 2706/7175 [33:46<48:49,  1.53it/s]
 38%|███▊      | 2707/7175 [33:47<48:50,  1.52it/s]
 38%|███▊      | 2708/7175 [33:47<48:49,  1.53it/s]
 38%|███▊      | 2709/7175 [33:48<48:46,  1.53it/s]
 38%|███▊      | 2710/7175 [33:49<48:44,  1.53it/s]
                                                   
{'loss': '3.224', 'grad_norm': '0.2208', 'learning_rate': '0.0006', 'epoch': '9.444'}
+
 38%|███▊      | 2710/7175 [33:49<48:44,  1.53it/s]
 38%|███▊      | 2711/7175 [33:49<48:57,  1.52it/s]
 38%|███▊      | 2712/7175 [33:50<48:51,  1.52it/s]
 38%|███▊      | 2713/7175 [33:51<48:50,  1.52it/s]
 38%|███▊      | 2714/7175 [33:51<48:48,  1.52it/s]
 38%|███▊      | 2715/7175 [33:52<48:43,  1.53it/s]
 38%|███▊      | 2716/7175 [33:53<48:42,  1.53it/s]
 38%|███▊      | 2717/7175 [33:53<48:41,  1.53it/s]
 38%|███▊      | 2718/7175 [33:54<48:40,  1.53it/s]
 38%|███▊      | 2719/7175 [33:55<48:38,  1.53it/s]
 38%|███▊      | 2720/7175 [33:55<48:37,  1.53it/s]
                                                   
{'loss': '3.218', 'grad_norm': '0.2181', 'learning_rate': '0.0006', 'epoch': '9.479'}
+
 38%|███▊      | 2720/7175 [33:55<48:37,  1.53it/s]
 38%|███▊      | 2721/7175 [33:56<48:40,  1.53it/s]
 38%|███▊      | 2722/7175 [33:56<48:38,  1.53it/s]
 38%|███▊      | 2723/7175 [33:57<48:36,  1.53it/s]
 38%|███▊      | 2724/7175 [33:58<48:34,  1.53it/s]
 38%|███▊      | 2725/7175 [33:58<48:36,  1.53it/s]
 38%|███▊      | 2726/7175 [33:59<48:35,  1.53it/s]
 38%|███▊      | 2727/7175 [34:00<48:33,  1.53it/s]
 38%|███▊      | 2728/7175 [34:00<48:31,  1.53it/s]
 38%|███▊      | 2729/7175 [34:01<48:34,  1.53it/s]
 38%|███▊      | 2730/7175 [34:02<48:32,  1.53it/s]
                                                   
{'loss': '3.204', 'grad_norm': '0.2273', 'learning_rate': '0.0006', 'epoch': '9.514'}
+
 38%|███▊      | 2730/7175 [34:02<48:32,  1.53it/s]
 38%|███▊      | 2731/7175 [34:02<48:32,  1.53it/s]
 38%|███▊      | 2732/7175 [34:03<48:29,  1.53it/s]
 38%|███▊      | 2733/7175 [34:04<48:30,  1.53it/s]
 38%|███▊      | 2734/7175 [34:04<48:29,  1.53it/s]
 38%|███▊      | 2735/7175 [34:05<48:26,  1.53it/s]
 38%|███▊      | 2736/7175 [34:06<48:24,  1.53it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:02, 17.61it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.72it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.39it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.07it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.63it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.60it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.36it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.42it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.26it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.35it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.32it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.28it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.27it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.26it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.25it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.24it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.25it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.24it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s]
                                                   
+
                                               
{'eval_loss': '3.566', 'eval_runtime': '3.249', 'eval_samples_per_second': '751.4', 'eval_steps_per_second': '12.01', 'epoch': '9.535'}
+
 38%|███▊      | 2736/7175 [34:09<48:24,  1.53it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-2736
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-2736/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-2736/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.65it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.65it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-2736/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-2736/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-2736/tokenizer_config.json
+[eval] epoch 9.535 step 2736 loss 3.5660 (best 3.5660 @ ep 9.535, 0/20 since best)
+
+
                                               
 38%|███▊      | 2737/7175 [34:12<2:48:39,  2.28s/it]
 38%|███▊      | 2738/7175 [34:12<2:12:34,  1.79s/it]
 38%|███▊      | 2739/7175 [34:13<1:47:16,  1.45s/it]
 38%|███▊      | 2740/7175 [34:14<1:29:37,  1.21s/it]
                                                     
{'loss': '3.214', 'grad_norm': '0.2086', 'learning_rate': '0.0006', 'epoch': '9.549'}
+
 38%|███▊      | 2740/7175 [34:14<1:29:37,  1.21s/it]
 38%|███▊      | 2741/7175 [34:14<1:17:16,  1.05s/it]
 38%|███▊      | 2742/7175 [34:15<1:08:36,  1.08it/s]
 38%|███▊      | 2743/7175 [34:16<1:02:30,  1.18it/s]
 38%|███▊      | 2744/7175 [34:16<58:15,  1.27it/s]  
 38%|███▊      | 2745/7175 [34:17<55:18,  1.33it/s]
 38%|███▊      | 2746/7175 [34:18<53:13,  1.39it/s]
 38%|███▊      | 2747/7175 [34:18<51:44,  1.43it/s]
 38%|███▊      | 2748/7175 [34:19<50:41,  1.46it/s]
 38%|███▊      | 2749/7175 [34:20<50:00,  1.48it/s]
 38%|███▊      | 2750/7175 [34:20<49:29,  1.49it/s]
                                                   
{'loss': '3.211', 'grad_norm': '0.2422', 'learning_rate': '0.0006', 'epoch': '9.584'}
+
 38%|███▊      | 2750/7175 [34:20<49:29,  1.49it/s]
 38%|███▊      | 2751/7175 [34:21<49:08,  1.50it/s]
 38%|███▊      | 2752/7175 [34:22<48:50,  1.51it/s]
 38%|███▊      | 2753/7175 [34:22<48:43,  1.51it/s]
 38%|███▊      | 2754/7175 [34:23<48:36,  1.52it/s]
 38%|███▊      | 2755/7175 [34:24<48:29,  1.52it/s]
 38%|███▊      | 2756/7175 [34:24<48:23,  1.52it/s]
 38%|███▊      | 2757/7175 [34:25<48:19,  1.52it/s]
 38%|███▊      | 2758/7175 [34:25<48:17,  1.52it/s]
 38%|███▊      | 2759/7175 [34:26<48:14,  1.53it/s]
 38%|███▊      | 2760/7175 [34:27<48:10,  1.53it/s]
                                                   
{'loss': '3.238', 'grad_norm': '0.2129', 'learning_rate': '0.0006', 'epoch': '9.619'}
+
 38%|███▊      | 2760/7175 [34:27<48:10,  1.53it/s]
 38%|███▊      | 2761/7175 [34:27<48:13,  1.53it/s]
 38%|███▊      | 2762/7175 [34:28<48:11,  1.53it/s]
 39%|███▊      | 2763/7175 [34:29<48:09,  1.53it/s]
 39%|███▊      | 2764/7175 [34:29<48:07,  1.53it/s]
 39%|███▊      | 2765/7175 [34:30<48:08,  1.53it/s]
 39%|███▊      | 2766/7175 [34:31<48:08,  1.53it/s]
 39%|███▊      | 2767/7175 [34:31<48:06,  1.53it/s]
 39%|███▊      | 2768/7175 [34:32<48:03,  1.53it/s]
 39%|███▊      | 2769/7175 [34:33<48:06,  1.53it/s]
 39%|███▊      | 2770/7175 [34:33<48:04,  1.53it/s]
                                                   
{'loss': '3.215', 'grad_norm': '0.2343', 'learning_rate': '0.0006', 'epoch': '9.654'}
+
 39%|███▊      | 2770/7175 [34:33<48:04,  1.53it/s]
 39%|███▊      | 2771/7175 [34:34<48:04,  1.53it/s]
 39%|███▊      | 2772/7175 [34:35<48:01,  1.53it/s]
 39%|███▊      | 2773/7175 [34:35<48:03,  1.53it/s]
 39%|███▊      | 2774/7175 [34:36<48:02,  1.53it/s]
 39%|███▊      | 2775/7175 [34:37<47:58,  1.53it/s]
 39%|███▊      | 2776/7175 [34:37<48:00,  1.53it/s]
 39%|███▊      | 2777/7175 [34:38<47:59,  1.53it/s]
 39%|███▊      | 2778/7175 [34:39<47:57,  1.53it/s]
 39%|███▊      | 2779/7175 [34:39<47:58,  1.53it/s]
 39%|███▊      | 2780/7175 [34:40<47:59,  1.53it/s]
                                                   
{'loss': '3.21', 'grad_norm': '0.2116', 'learning_rate': '0.0006', 'epoch': '9.689'}
+
 39%|███▊      | 2780/7175 [34:40<47:59,  1.53it/s]
 39%|███▉      | 2781/7175 [34:41<48:00,  1.53it/s]
 39%|███▉      | 2782/7175 [34:41<47:58,  1.53it/s]
 39%|███▉      | 2783/7175 [34:42<47:55,  1.53it/s]
 39%|███▉      | 2784/7175 [34:43<47:54,  1.53it/s]
 39%|███▉      | 2785/7175 [34:43<47:56,  1.53it/s]
 39%|███▉      | 2786/7175 [34:44<47:54,  1.53it/s]
 39%|███▉      | 2787/7175 [34:44<47:52,  1.53it/s]
 39%|███▉      | 2788/7175 [34:45<47:53,  1.53it/s]
 39%|███▉      | 2789/7175 [34:46<47:53,  1.53it/s]
 39%|███▉      | 2790/7175 [34:46<47:52,  1.53it/s]
                                                   
{'loss': '3.23', 'grad_norm': '0.2266', 'learning_rate': '0.0006', 'epoch': '9.723'}
+
 39%|███▉      | 2790/7175 [34:46<47:52,  1.53it/s]
 39%|███▉      | 2791/7175 [34:47<47:51,  1.53it/s]
 39%|███▉      | 2792/7175 [34:48<47:51,  1.53it/s]
 39%|███▉      | 2793/7175 [34:48<47:50,  1.53it/s]
 39%|███▉      | 2794/7175 [34:49<47:49,  1.53it/s]
 39%|███▉      | 2795/7175 [34:50<47:45,  1.53it/s]
 39%|███▉      | 2796/7175 [34:50<47:47,  1.53it/s]
 39%|███▉      | 2797/7175 [34:51<47:46,  1.53it/s]
 39%|███▉      | 2798/7175 [34:52<47:44,  1.53it/s]
 39%|███▉      | 2799/7175 [34:52<47:42,  1.53it/s]
 39%|███▉      | 2800/7175 [34:53<47:45,  1.53it/s]
                                                   
{'loss': '3.217', 'grad_norm': '0.229', 'learning_rate': '0.0006', 'epoch': '9.758'}
+
 39%|███▉      | 2800/7175 [34:53<47:45,  1.53it/s]
 39%|███▉      | 2801/7175 [34:54<47:47,  1.53it/s]
 39%|███▉      | 2802/7175 [34:54<47:45,  1.53it/s]
 39%|███▉      | 2803/7175 [34:55<47:42,  1.53it/s]
 39%|███▉      | 2804/7175 [34:56<47:44,  1.53it/s]
 39%|███▉      | 2805/7175 [34:56<47:44,  1.53it/s]
 39%|███▉      | 2806/7175 [34:57<47:41,  1.53it/s]
 39%|███▉      | 2807/7175 [34:58<47:38,  1.53it/s]
 39%|███▉      | 2808/7175 [34:58<47:41,  1.53it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.29it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.69it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.57it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 12.99it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.73it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.55it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.45it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.37it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.33it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.27it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.29it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.26it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.25it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.25it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.24it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.22it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.23it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.22it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s]
                                                   
+
                                               
{'eval_loss': '3.558', 'eval_runtime': '3.249', 'eval_samples_per_second': '751.2', 'eval_steps_per_second': '12', 'epoch': '9.786'}
+
 39%|███▉      | 2808/7175 [35:01<47:41,  1.53it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-2808
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-2808/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-2808/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.85it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.85it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-2808/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-2808/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-2808/tokenizer_config.json
+[eval] epoch 9.786 step 2808 loss 3.5585 (best 3.5585 @ ep 9.786, 0/20 since best)
+
+
                                               
 39%|███▉      | 2809/7175 [35:04<2:44:26,  2.26s/it]
 39%|███▉      | 2810/7175 [35:05<2:09:22,  1.78s/it]
                                                     
{'loss': '3.222', 'grad_norm': '0.2157', 'learning_rate': '0.0006', 'epoch': '9.793'}
+
 39%|███▉      | 2810/7175 [35:05<2:09:22,  1.78s/it]
 39%|███▉      | 2811/7175 [35:06<1:44:51,  1.44s/it]
 39%|███▉      | 2812/7175 [35:06<1:27:39,  1.21s/it]
 39%|███▉      | 2813/7175 [35:07<1:15:35,  1.04s/it]
 39%|███▉      | 2814/7175 [35:08<1:07:13,  1.08it/s]
 39%|███▉      | 2815/7175 [35:08<1:01:21,  1.18it/s]
 39%|███▉      | 2816/7175 [35:09<57:12,  1.27it/s]  
 39%|███▉      | 2817/7175 [35:09<54:16,  1.34it/s]
 39%|███▉      | 2818/7175 [35:10<52:17,  1.39it/s]
 39%|███▉      | 2819/7175 [35:11<50:51,  1.43it/s]
 39%|███▉      | 2820/7175 [35:11<49:53,  1.45it/s]
                                                   
{'loss': '3.227', 'grad_norm': '0.1985', 'learning_rate': '0.0006', 'epoch': '9.828'}
+
 39%|███▉      | 2820/7175 [35:11<49:53,  1.45it/s]
 39%|███▉      | 2821/7175 [35:12<49:11,  1.48it/s]
 39%|███▉      | 2822/7175 [35:13<48:39,  1.49it/s]
 39%|███▉      | 2823/7175 [35:13<48:20,  1.50it/s]
 39%|███▉      | 2824/7175 [35:14<48:05,  1.51it/s]
 39%|███▉      | 2825/7175 [35:15<47:53,  1.51it/s]
 39%|███▉      | 2826/7175 [35:15<47:43,  1.52it/s]
 39%|███▉      | 2827/7175 [35:16<47:41,  1.52it/s]
 39%|███▉      | 2828/7175 [35:17<47:37,  1.52it/s]
 39%|███▉      | 2829/7175 [35:17<47:36,  1.52it/s]
 39%|███▉      | 2830/7175 [35:18<47:32,  1.52it/s]
                                                   
{'loss': '3.232', 'grad_norm': '0.2149', 'learning_rate': '0.0006', 'epoch': '9.863'}
+
 39%|███▉      | 2830/7175 [35:18<47:32,  1.52it/s]
 39%|███▉      | 2831/7175 [35:19<47:30,  1.52it/s]
 39%|███▉      | 2832/7175 [35:19<47:30,  1.52it/s]
 39%|███▉      | 2833/7175 [35:20<47:29,  1.52it/s]
 39%|███▉      | 2834/7175 [35:21<47:28,  1.52it/s]
 40%|███▉      | 2835/7175 [35:21<47:27,  1.52it/s]
 40%|███▉      | 2836/7175 [35:22<47:24,  1.53it/s]
 40%|███▉      | 2837/7175 [35:23<47:27,  1.52it/s]
 40%|███▉      | 2838/7175 [35:23<47:26,  1.52it/s]
 40%|███▉      | 2839/7175 [35:24<47:26,  1.52it/s]
 40%|███▉      | 2840/7175 [35:25<47:25,  1.52it/s]
                                                   
{'loss': '3.231', 'grad_norm': '0.2227', 'learning_rate': '0.0006', 'epoch': '9.898'}
+
 40%|███▉      | 2840/7175 [35:25<47:25,  1.52it/s]
 40%|███▉      | 2841/7175 [35:25<47:23,  1.52it/s]
 40%|███▉      | 2842/7175 [35:26<47:23,  1.52it/s]
 40%|███▉      | 2843/7175 [35:27<47:21,  1.52it/s]
 40%|███▉      | 2844/7175 [35:27<47:21,  1.52it/s]
 40%|███▉      | 2845/7175 [35:28<47:20,  1.52it/s]
 40%|███▉      | 2846/7175 [35:29<47:19,  1.52it/s]
 40%|███▉      | 2847/7175 [35:29<47:19,  1.52it/s]
 40%|███▉      | 2848/7175 [35:30<47:19,  1.52it/s]
 40%|███▉      | 2849/7175 [35:30<47:18,  1.52it/s]
 40%|███▉      | 2850/7175 [35:31<47:17,  1.52it/s]
                                                   
{'loss': '3.227', 'grad_norm': '0.245', 'learning_rate': '0.0006', 'epoch': '9.933'}
+
 40%|███▉      | 2850/7175 [35:31<47:17,  1.52it/s]
 40%|███▉      | 2851/7175 [35:32<47:16,  1.52it/s]
 40%|███▉      | 2852/7175 [35:32<47:16,  1.52it/s]
 40%|███▉      | 2853/7175 [35:33<47:14,  1.52it/s]
 40%|███▉      | 2854/7175 [35:34<47:15,  1.52it/s]
 40%|███▉      | 2855/7175 [35:34<47:16,  1.52it/s]
 40%|███▉      | 2856/7175 [35:35<47:11,  1.53it/s]
 40%|███▉      | 2857/7175 [35:36<47:11,  1.52it/s]
 40%|███▉      | 2858/7175 [35:36<47:11,  1.52it/s]
 40%|███▉      | 2859/7175 [35:37<47:08,  1.53it/s]
 40%|███▉      | 2860/7175 [35:38<47:06,  1.53it/s]
                                                   
{'loss': '3.22', 'grad_norm': '0.2726', 'learning_rate': '0.0006', 'epoch': '9.968'}
+
 40%|███▉      | 2860/7175 [35:38<47:06,  1.53it/s]
 40%|███▉      | 2861/7175 [35:38<47:07,  1.53it/s]
 40%|███▉      | 2862/7175 [35:39<47:08,  1.52it/s]
 40%|███▉      | 2863/7175 [35:40<47:06,  1.53it/s]
 40%|███▉      | 2864/7175 [35:40<47:04,  1.53it/s]
 40%|███▉      | 2865/7175 [35:41<47:02,  1.53it/s]
 40%|███▉      | 2866/7175 [35:42<47:02,  1.53it/s]
 40%|███▉      | 2867/7175 [35:42<47:11,  1.52it/s]
 40%|███▉      | 2868/7175 [35:43<47:07,  1.52it/s]
 40%|███▉      | 2869/7175 [35:44<47:04,  1.52it/s]
                                                   
{'loss': '3.22', 'grad_norm': '0.402', 'learning_rate': '0.0006', 'epoch': '10'}
+
 40%|████      | 2870/7175 [35:44<47:04,  1.52it/s]
 40%|████      | 2871/7175 [35:44<39:25,  1.82it/s]
 40%|████      | 2872/7175 [35:45<41:17,  1.74it/s]
 40%|████      | 2873/7175 [35:46<42:46,  1.68it/s]
 40%|████      | 2874/7175 [35:46<43:53,  1.63it/s]
 40%|████      | 2875/7175 [35:47<44:44,  1.60it/s]
 40%|████      | 2876/7175 [35:48<45:23,  1.58it/s]
 40%|████      | 2877/7175 [35:48<45:49,  1.56it/s]
 40%|████      | 2878/7175 [35:49<46:07,  1.55it/s]
 40%|████      | 2879/7175 [35:50<46:20,  1.54it/s]
 40%|████      | 2880/7175 [35:50<46:31,  1.54it/s]
                                                   
{'loss': '3.168', 'grad_norm': '0.3461', 'learning_rate': '0.0006', 'epoch': '10.03'}
+
 40%|████      | 2880/7175 [35:50<46:31,  1.54it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.42it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.47it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.67it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 12.95it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.82it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.50it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.52it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.31it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.39it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.20it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.35it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.31it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.28it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.27it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.25it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.25it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.23it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.23it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.20it/s]
                                                   
+
                                               
{'eval_loss': '3.585', 'eval_runtime': '3.264', 'eval_samples_per_second': '748', 'eval_steps_per_second': '11.95', 'epoch': '10.03'}
+
 40%|████      | 2880/7175 [35:54<46:31,  1.54it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.20it/s][eval] epoch 10.035 step 2880 loss 3.5853 (best 3.5585 @ ep 9.786, 1/20 since best)
+
+
                                               
 40%|████      | 2881/7175 [35:54<1:56:11,  1.62s/it]
 40%|████      | 2882/7175 [35:55<1:35:29,  1.33s/it]
 40%|████      | 2883/7175 [35:56<1:20:56,  1.13s/it]
 40%|████      | 2884/7175 [35:56<1:10:43,  1.01it/s]
 40%|████      | 2885/7175 [35:57<1:03:34,  1.12it/s]
 40%|████      | 2886/7175 [35:58<58:32,  1.22it/s]  
 40%|████      | 2887/7175 [35:58<55:00,  1.30it/s]
 40%|████      | 2888/7175 [35:59<52:31,  1.36it/s]
 40%|████      | 2889/7175 [35:59<50:49,  1.41it/s]
 40%|████      | 2890/7175 [36:00<49:37,  1.44it/s]
                                                   
{'loss': '3.165', 'grad_norm': '0.2785', 'learning_rate': '0.0006', 'epoch': '10.07'}
+
 40%|████      | 2890/7175 [36:00<49:37,  1.44it/s]
 40%|████      | 2891/7175 [36:01<48:47,  1.46it/s]
 40%|████      | 2892/7175 [36:01<48:09,  1.48it/s]
 40%|████      | 2893/7175 [36:02<47:43,  1.50it/s]
 40%|████      | 2894/7175 [36:03<47:27,  1.50it/s]
 40%|████      | 2895/7175 [36:03<47:15,  1.51it/s]
 40%|████      | 2896/7175 [36:04<47:06,  1.51it/s]
 40%|████      | 2897/7175 [36:05<46:58,  1.52it/s]
 40%|████      | 2898/7175 [36:05<46:53,  1.52it/s]
 40%|████      | 2899/7175 [36:06<46:50,  1.52it/s]
 40%|████      | 2900/7175 [36:07<46:47,  1.52it/s]
                                                   
{'loss': '3.165', 'grad_norm': '0.2135', 'learning_rate': '0.0006', 'epoch': '10.1'}
+
 40%|████      | 2900/7175 [36:07<46:47,  1.52it/s]
 40%|████      | 2901/7175 [36:07<46:45,  1.52it/s]
 40%|████      | 2902/7175 [36:08<46:41,  1.53it/s]
 40%|████      | 2903/7175 [36:09<46:42,  1.52it/s]
 40%|████      | 2904/7175 [36:09<46:41,  1.52it/s]
 40%|████      | 2905/7175 [36:10<46:39,  1.53it/s]
 41%|████      | 2906/7175 [36:11<46:35,  1.53it/s]
 41%|████      | 2907/7175 [36:11<46:37,  1.53it/s]
 41%|████      | 2908/7175 [36:12<46:37,  1.53it/s]
 41%|████      | 2909/7175 [36:13<46:36,  1.53it/s]
 41%|████      | 2910/7175 [36:13<46:33,  1.53it/s]
                                                   
{'loss': '3.158', 'grad_norm': '0.2095', 'learning_rate': '0.0006', 'epoch': '10.14'}
+
 41%|████      | 2910/7175 [36:13<46:33,  1.53it/s]
 41%|████      | 2911/7175 [36:14<46:35,  1.53it/s]
 41%|████      | 2912/7175 [36:15<46:35,  1.52it/s]
 41%|████      | 2913/7175 [36:15<46:35,  1.52it/s]
 41%|████      | 2914/7175 [36:16<46:36,  1.52it/s]
 41%|████      | 2915/7175 [36:17<46:31,  1.53it/s]
 41%|████      | 2916/7175 [36:17<46:31,  1.53it/s]
 41%|████      | 2917/7175 [36:18<46:30,  1.53it/s]
 41%|████      | 2918/7175 [36:19<46:29,  1.53it/s]
 41%|████      | 2919/7175 [36:19<46:25,  1.53it/s]
 41%|████      | 2920/7175 [36:20<46:27,  1.53it/s]
                                                   
{'loss': '3.168', 'grad_norm': '0.2515', 'learning_rate': '0.0006', 'epoch': '10.17'}
+
 41%|████      | 2920/7175 [36:20<46:27,  1.53it/s]
 41%|████      | 2921/7175 [36:20<46:29,  1.53it/s]
 41%|████      | 2922/7175 [36:21<46:27,  1.53it/s]
 41%|████      | 2923/7175 [36:22<46:25,  1.53it/s]
 41%|████      | 2924/7175 [36:22<46:25,  1.53it/s]
 41%|████      | 2925/7175 [36:23<46:25,  1.53it/s]
 41%|████      | 2926/7175 [36:24<46:25,  1.53it/s]
 41%|████      | 2927/7175 [36:24<46:22,  1.53it/s]
 41%|████      | 2928/7175 [36:25<46:21,  1.53it/s]
 41%|████      | 2929/7175 [36:26<46:22,  1.53it/s]
 41%|████      | 2930/7175 [36:26<46:21,  1.53it/s]
                                                   
{'loss': '3.177', 'grad_norm': '0.2048', 'learning_rate': '0.0006', 'epoch': '10.21'}
+
 41%|████      | 2930/7175 [36:26<46:21,  1.53it/s]
 41%|████      | 2931/7175 [36:27<46:24,  1.52it/s]
 41%|████      | 2932/7175 [36:28<46:19,  1.53it/s]
 41%|████      | 2933/7175 [36:28<46:21,  1.53it/s]
 41%|████      | 2934/7175 [36:29<46:20,  1.53it/s]
 41%|████      | 2935/7175 [36:30<46:18,  1.53it/s]
 41%|████      | 2936/7175 [36:30<46:15,  1.53it/s]
 41%|████      | 2937/7175 [36:31<46:16,  1.53it/s]
 41%|████      | 2938/7175 [36:32<46:15,  1.53it/s]
 41%|████      | 2939/7175 [36:32<46:13,  1.53it/s]
 41%|████      | 2940/7175 [36:33<46:10,  1.53it/s]
                                                   
{'loss': '3.17', 'grad_norm': '0.2251', 'learning_rate': '0.0006', 'epoch': '10.24'}
+
 41%|████      | 2940/7175 [36:33<46:10,  1.53it/s]
 41%|████      | 2941/7175 [36:34<46:14,  1.53it/s]
 41%|████      | 2942/7175 [36:34<46:12,  1.53it/s]
 41%|████      | 2943/7175 [36:35<46:10,  1.53it/s]
 41%|████      | 2944/7175 [36:36<46:08,  1.53it/s]
 41%|████      | 2945/7175 [36:36<46:10,  1.53it/s]
 41%|████      | 2946/7175 [36:37<46:09,  1.53it/s]
 41%|████      | 2947/7175 [36:38<46:07,  1.53it/s]
 41%|████      | 2948/7175 [36:38<46:07,  1.53it/s]
 41%|████      | 2949/7175 [36:39<46:07,  1.53it/s]
 41%|████      | 2950/7175 [36:39<46:07,  1.53it/s]
                                                   
{'loss': '3.173', 'grad_norm': '0.2273', 'learning_rate': '0.0006', 'epoch': '10.28'}
+
 41%|████      | 2950/7175 [36:39<46:07,  1.53it/s]
 41%|████      | 2951/7175 [36:40<46:06,  1.53it/s]
 41%|████      | 2952/7175 [36:41<46:05,  1.53it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.70it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.83it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.64it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.08it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.77it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.56it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.45it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.37it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.30it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.29it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.29it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.27it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.26it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.25it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.25it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.24it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.23it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.24it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.22it/s]
                                                   
+
                                               
{'eval_loss': '3.559', 'eval_runtime': '3.262', 'eval_samples_per_second': '748.4', 'eval_steps_per_second': '11.96', 'epoch': '10.29'}
+
 41%|████      | 2952/7175 [36:44<46:05,  1.53it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.22it/s][eval] epoch 10.287 step 2952 loss 3.5589 (best 3.5585 @ ep 9.786, 2/20 since best)
+
+
                                               
 41%|████      | 2953/7175 [36:45<1:54:58,  1.63s/it]
 41%|████      | 2954/7175 [36:45<1:34:16,  1.34s/it]
 41%|████      | 2955/7175 [36:46<1:19:47,  1.13s/it]
 41%|████      | 2956/7175 [36:47<1:09:37,  1.01it/s]
 41%|████      | 2957/7175 [36:47<1:02:34,  1.12it/s]
 41%|████      | 2958/7175 [36:48<57:36,  1.22it/s]  
 41%|████      | 2959/7175 [36:49<54:06,  1.30it/s]
 41%|████▏     | 2960/7175 [36:49<51:37,  1.36it/s]
                                                   
{'loss': '3.18', 'grad_norm': '0.1997', 'learning_rate': '0.0006', 'epoch': '10.31'}
+
 41%|████▏     | 2960/7175 [36:49<51:37,  1.36it/s]
 41%|████▏     | 2961/7175 [36:50<49:59,  1.41it/s]
 41%|████▏     | 2962/7175 [36:51<48:47,  1.44it/s]
 41%|████▏     | 2963/7175 [36:51<47:55,  1.46it/s]
 41%|████▏     | 2964/7175 [36:52<47:17,  1.48it/s]
 41%|████▏     | 2965/7175 [36:53<46:56,  1.50it/s]
 41%|████▏     | 2966/7175 [36:53<46:38,  1.50it/s]
 41%|████▏     | 2967/7175 [36:54<46:23,  1.51it/s]
 41%|████▏     | 2968/7175 [36:55<46:14,  1.52it/s]
 41%|████▏     | 2969/7175 [36:55<46:10,  1.52it/s]
 41%|████▏     | 2970/7175 [36:56<46:05,  1.52it/s]
                                                   
{'loss': '3.178', 'grad_norm': '0.2289', 'learning_rate': '0.0006', 'epoch': '10.35'}
+
 41%|████▏     | 2970/7175 [36:56<46:05,  1.52it/s]
 41%|████▏     | 2971/7175 [36:56<46:04,  1.52it/s]
 41%|████▏     | 2972/7175 [36:57<45:59,  1.52it/s]
 41%|████▏     | 2973/7175 [36:58<45:56,  1.52it/s]
 41%|████▏     | 2974/7175 [36:58<45:57,  1.52it/s]
 41%|████▏     | 2975/7175 [36:59<45:56,  1.52it/s]
 41%|████▏     | 2976/7175 [37:00<45:54,  1.52it/s]
 41%|████▏     | 2977/7175 [37:00<45:51,  1.53it/s]
 42%|████▏     | 2978/7175 [37:01<45:52,  1.52it/s]
 42%|████▏     | 2979/7175 [37:02<45:51,  1.52it/s]
 42%|████▏     | 2980/7175 [37:02<45:50,  1.52it/s]
                                                   
{'loss': '3.17', 'grad_norm': '0.2286', 'learning_rate': '0.0006', 'epoch': '10.38'}
+
 42%|████▏     | 2980/7175 [37:02<45:50,  1.52it/s]
 42%|████▏     | 2981/7175 [37:03<45:50,  1.52it/s]
 42%|████▏     | 2982/7175 [37:04<45:47,  1.53it/s]
 42%|████▏     | 2983/7175 [37:04<45:48,  1.52it/s]
 42%|████▏     | 2984/7175 [37:05<45:47,  1.53it/s]
 42%|████▏     | 2985/7175 [37:06<45:44,  1.53it/s]
 42%|████▏     | 2986/7175 [37:06<45:42,  1.53it/s]
 42%|████▏     | 2987/7175 [37:07<45:45,  1.53it/s]
 42%|████▏     | 2988/7175 [37:08<45:44,  1.53it/s]
 42%|████▏     | 2989/7175 [37:08<45:43,  1.53it/s]
 42%|████▏     | 2990/7175 [37:09<45:41,  1.53it/s]
                                                   
{'loss': '3.178', 'grad_norm': '0.2453', 'learning_rate': '0.0006', 'epoch': '10.42'}
+
 42%|████▏     | 2990/7175 [37:09<45:41,  1.53it/s]
 42%|████▏     | 2991/7175 [37:10<45:43,  1.53it/s]
 42%|████▏     | 2992/7175 [37:10<45:44,  1.52it/s]
 42%|████▏     | 2993/7175 [37:11<45:42,  1.52it/s]
 42%|████▏     | 2994/7175 [37:12<45:40,  1.53it/s]
 42%|████▏     | 2995/7175 [37:12<45:38,  1.53it/s]
 42%|████▏     | 2996/7175 [37:13<45:40,  1.53it/s]
 42%|████▏     | 2997/7175 [37:14<45:39,  1.52it/s]
 42%|████▏     | 2998/7175 [37:14<45:37,  1.53it/s]
 42%|████▏     | 2999/7175 [37:15<45:36,  1.53it/s]
 42%|████▏     | 3000/7175 [37:15<45:35,  1.53it/s]
                                                   
{'loss': '3.188', 'grad_norm': '0.2394', 'learning_rate': '0.0006', 'epoch': '10.45'}
+
 42%|████▏     | 3000/7175 [37:16<45:35,  1.53it/s]
 42%|████▏     | 3001/7175 [37:16<45:38,  1.52it/s]
 42%|████▏     | 3002/7175 [37:17<45:36,  1.52it/s]
 42%|████▏     | 3003/7175 [37:17<45:35,  1.52it/s]
 42%|████▏     | 3004/7175 [37:18<45:33,  1.53it/s]
 42%|████▏     | 3005/7175 [37:19<45:34,  1.52it/s]
 42%|████▏     | 3006/7175 [37:19<45:34,  1.52it/s]
 42%|████▏     | 3007/7175 [37:20<45:33,  1.52it/s]
 42%|████▏     | 3008/7175 [37:21<45:32,  1.52it/s]
 42%|████▏     | 3009/7175 [37:21<45:29,  1.53it/s]
 42%|████▏     | 3010/7175 [37:22<45:31,  1.52it/s]
                                                   
{'loss': '3.181', 'grad_norm': '0.214', 'learning_rate': '0.0006', 'epoch': '10.49'}
+
 42%|████▏     | 3010/7175 [37:22<45:31,  1.52it/s]
 42%|████▏     | 3011/7175 [37:23<45:34,  1.52it/s]
 42%|████▏     | 3012/7175 [37:23<45:34,  1.52it/s]
 42%|████▏     | 3013/7175 [37:24<45:33,  1.52it/s]
 42%|████▏     | 3014/7175 [37:25<45:30,  1.52it/s]
 42%|████▏     | 3015/7175 [37:25<45:28,  1.52it/s]
 42%|████▏     | 3016/7175 [37:26<45:30,  1.52it/s]
 42%|████▏     | 3017/7175 [37:27<45:29,  1.52it/s]
 42%|████▏     | 3018/7175 [37:27<45:28,  1.52it/s]
 42%|████▏     | 3019/7175 [37:28<45:26,  1.52it/s]
 42%|████▏     | 3020/7175 [37:29<45:23,  1.53it/s]
                                                   
{'loss': '3.175', 'grad_norm': '0.2218', 'learning_rate': '0.0006', 'epoch': '10.52'}
+
 42%|████▏     | 3020/7175 [37:29<45:23,  1.53it/s]
 42%|████▏     | 3021/7175 [37:29<45:27,  1.52it/s]
 42%|████▏     | 3022/7175 [37:30<45:26,  1.52it/s]
 42%|████▏     | 3023/7175 [37:31<45:25,  1.52it/s]
 42%|████▏     | 3024/7175 [37:31<45:23,  1.52it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.54it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.69it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.60it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.02it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.74it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.54it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.43it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.35it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.31it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.27it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.24it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.26it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.23it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.23it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.21it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.22it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.22it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.23it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.20it/s]
                                                   
+
                                               
{'eval_loss': '3.553', 'eval_runtime': '3.259', 'eval_samples_per_second': '749', 'eval_steps_per_second': '11.97', 'epoch': '10.54'}
+
 42%|████▏     | 3024/7175 [37:35<45:23,  1.52it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.20it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-3024
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-3024/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-3024/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.85it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.85it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-3024/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-3024/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-3024/tokenizer_config.json
+[eval] epoch 10.538 step 3024 loss 3.5530 (best 3.5530 @ ep 10.538, 0/20 since best)
+
+
                                               
 42%|████▏     | 3025/7175 [37:37<2:36:31,  2.26s/it]
 42%|████▏     | 3026/7175 [37:38<2:03:06,  1.78s/it]
 42%|████▏     | 3027/7175 [37:39<1:39:43,  1.44s/it]
 42%|████▏     | 3028/7175 [37:39<1:23:19,  1.21s/it]
 42%|████▏     | 3029/7175 [37:40<1:11:55,  1.04s/it]
 42%|████▏     | 3030/7175 [37:41<1:03:55,  1.08it/s]
                                                     
{'loss': '3.182', 'grad_norm': '0.2034', 'learning_rate': '0.0006', 'epoch': '10.56'}
+
 42%|████▏     | 3030/7175 [37:41<1:03:55,  1.08it/s]
 42%|████▏     | 3031/7175 [37:41<58:18,  1.18it/s]  
 42%|████▏     | 3032/7175 [37:42<54:20,  1.27it/s]
 42%|████▏     | 3033/7175 [37:42<51:38,  1.34it/s]
 42%|████▏     | 3034/7175 [37:43<49:43,  1.39it/s]
 42%|████▏     | 3035/7175 [37:44<48:22,  1.43it/s]
 42%|████▏     | 3036/7175 [37:44<47:24,  1.45it/s]
 42%|████▏     | 3037/7175 [37:45<46:44,  1.48it/s]
 42%|████▏     | 3038/7175 [37:46<46:18,  1.49it/s]
 42%|████▏     | 3039/7175 [37:46<45:58,  1.50it/s]
 42%|████▏     | 3040/7175 [37:47<45:43,  1.51it/s]
                                                   
{'loss': '3.192', 'grad_norm': '0.2232', 'learning_rate': '0.0006', 'epoch': '10.59'}
+
 42%|████▏     | 3040/7175 [37:47<45:43,  1.51it/s]
 42%|████▏     | 3041/7175 [37:48<45:33,  1.51it/s]
 42%|████▏     | 3042/7175 [37:48<45:25,  1.52it/s]
 42%|████▏     | 3043/7175 [37:49<45:21,  1.52it/s]
 42%|████▏     | 3044/7175 [37:50<45:18,  1.52it/s]
 42%|████▏     | 3045/7175 [37:50<45:14,  1.52it/s]
 42%|████▏     | 3046/7175 [37:51<45:10,  1.52it/s]
 42%|████▏     | 3047/7175 [37:52<45:08,  1.52it/s]
 42%|████▏     | 3048/7175 [37:52<45:07,  1.52it/s]
 42%|████▏     | 3049/7175 [37:53<45:07,  1.52it/s]
 43%|████▎     | 3050/7175 [37:54<45:05,  1.52it/s]
                                                   
{'loss': '3.19', 'grad_norm': '0.2297', 'learning_rate': '0.0006', 'epoch': '10.63'}
+
 43%|████▎     | 3050/7175 [37:54<45:05,  1.52it/s]
 43%|████▎     | 3051/7175 [37:54<45:03,  1.53it/s]
 43%|████▎     | 3052/7175 [37:55<45:04,  1.52it/s]
 43%|████▎     | 3053/7175 [37:56<45:04,  1.52it/s]
 43%|████▎     | 3054/7175 [37:56<45:03,  1.52it/s]
 43%|████▎     | 3055/7175 [37:57<45:01,  1.53it/s]
 43%|████▎     | 3056/7175 [37:58<44:59,  1.53it/s]
 43%|████▎     | 3057/7175 [37:58<45:00,  1.52it/s]
 43%|████▎     | 3058/7175 [37:59<45:00,  1.52it/s]
 43%|████▎     | 3059/7175 [38:00<44:59,  1.52it/s]
 43%|████▎     | 3060/7175 [38:00<44:57,  1.53it/s]
                                                   
{'loss': '3.202', 'grad_norm': '0.2268', 'learning_rate': '0.0006', 'epoch': '10.66'}
+
 43%|████▎     | 3060/7175 [38:00<44:57,  1.53it/s]
 43%|████▎     | 3061/7175 [38:01<44:56,  1.53it/s]
 43%|████▎     | 3062/7175 [38:02<44:57,  1.52it/s]
 43%|████▎     | 3063/7175 [38:02<44:57,  1.52it/s]
 43%|████▎     | 3064/7175 [38:03<44:55,  1.53it/s]
 43%|████▎     | 3065/7175 [38:03<44:51,  1.53it/s]
 43%|████▎     | 3066/7175 [38:04<44:53,  1.53it/s]
 43%|████▎     | 3067/7175 [38:05<44:53,  1.53it/s]
 43%|████▎     | 3068/7175 [38:05<44:53,  1.52it/s]
 43%|████▎     | 3069/7175 [38:06<44:50,  1.53it/s]
 43%|████▎     | 3070/7175 [38:07<44:50,  1.53it/s]
                                                   
{'loss': '3.167', 'grad_norm': '0.2388', 'learning_rate': '0.0006', 'epoch': '10.7'}
+
 43%|████▎     | 3070/7175 [38:07<44:50,  1.53it/s]
 43%|████▎     | 3071/7175 [38:07<44:52,  1.52it/s]
 43%|████▎     | 3072/7175 [38:08<44:52,  1.52it/s]
 43%|████▎     | 3073/7175 [38:09<44:50,  1.52it/s]
 43%|████▎     | 3074/7175 [38:09<44:47,  1.53it/s]
 43%|████▎     | 3075/7175 [38:10<44:47,  1.53it/s]
 43%|████▎     | 3076/7175 [38:11<44:47,  1.53it/s]
 43%|████▎     | 3077/7175 [38:11<44:46,  1.53it/s]
 43%|████▎     | 3078/7175 [38:12<44:44,  1.53it/s]
 43%|████▎     | 3079/7175 [38:13<44:43,  1.53it/s]
 43%|████▎     | 3080/7175 [38:13<44:43,  1.53it/s]
                                                   
{'loss': '3.192', 'grad_norm': '0.2305', 'learning_rate': '0.0006', 'epoch': '10.73'}
+
 43%|████▎     | 3080/7175 [38:13<44:43,  1.53it/s]
 43%|████▎     | 3081/7175 [38:14<44:45,  1.52it/s]
 43%|████▎     | 3082/7175 [38:15<44:42,  1.53it/s]
 43%|████▎     | 3083/7175 [38:15<44:38,  1.53it/s]
 43%|████▎     | 3084/7175 [38:16<44:40,  1.53it/s]
 43%|████▎     | 3085/7175 [38:17<44:40,  1.53it/s]
 43%|████▎     | 3086/7175 [38:17<44:40,  1.53it/s]
 43%|████▎     | 3087/7175 [38:18<44:37,  1.53it/s]
 43%|████▎     | 3088/7175 [38:19<44:39,  1.53it/s]
 43%|████▎     | 3089/7175 [38:19<44:38,  1.53it/s]
 43%|████▎     | 3090/7175 [38:20<44:35,  1.53it/s]
                                                   
{'loss': '3.184', 'grad_norm': '0.2283', 'learning_rate': '0.0006', 'epoch': '10.77'}
+
 43%|████▎     | 3090/7175 [38:20<44:35,  1.53it/s]
 43%|████▎     | 3091/7175 [38:21<44:34,  1.53it/s]
 43%|████▎     | 3092/7175 [38:21<44:35,  1.53it/s]
 43%|████▎     | 3093/7175 [38:22<44:35,  1.53it/s]
 43%|████▎     | 3094/7175 [38:22<44:34,  1.53it/s]
 43%|████▎     | 3095/7175 [38:23<44:32,  1.53it/s]
 43%|████▎     | 3096/7175 [38:24<44:32,  1.53it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.35it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.69it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.57it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.02it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.75it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.56it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.46it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.38it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.34it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.29it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.30it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.28it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.27it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.26it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.23it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.25it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.24it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.23it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s]
                                                   
+
                                               
{'eval_loss': '3.544', 'eval_runtime': '3.261', 'eval_samples_per_second': '748.4', 'eval_steps_per_second': '11.96', 'epoch': '10.79'}
+
 43%|████▎     | 3096/7175 [38:27<44:32,  1.53it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-3096
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-3096/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-3096/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.86it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.86it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-3096/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-3096/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-3096/tokenizer_config.json
+[eval] epoch 10.79 step 3096 loss 3.5444 (best 3.5444 @ ep 10.79, 0/20 since best)
+
+
                                               
 43%|████▎     | 3097/7175 [38:30<2:33:25,  2.26s/it]
 43%|████▎     | 3098/7175 [38:30<2:00:44,  1.78s/it]
 43%|████▎     | 3099/7175 [38:31<1:37:48,  1.44s/it]
 43%|████▎     | 3100/7175 [38:32<1:21:47,  1.20s/it]
                                                     
{'loss': '3.191', 'grad_norm': '0.2213', 'learning_rate': '0.0006', 'epoch': '10.8'}
+
 43%|████▎     | 3100/7175 [38:32<1:21:47,  1.20s/it]
 43%|████▎     | 3101/7175 [38:32<1:10:37,  1.04s/it]
 43%|████▎     | 3102/7175 [38:33<1:02:46,  1.08it/s]
 43%|████▎     | 3103/7175 [38:34<57:15,  1.19it/s]  
 43%|████▎     | 3104/7175 [38:34<53:23,  1.27it/s]
 43%|████▎     | 3105/7175 [38:35<50:44,  1.34it/s]
 43%|████▎     | 3106/7175 [38:36<48:51,  1.39it/s]
 43%|████▎     | 3107/7175 [38:36<47:29,  1.43it/s]
 43%|████▎     | 3108/7175 [38:37<46:31,  1.46it/s]
 43%|████▎     | 3109/7175 [38:38<45:55,  1.48it/s]
 43%|████▎     | 3110/7175 [38:38<45:27,  1.49it/s]
                                                   
{'loss': '3.187', 'grad_norm': '0.2326', 'learning_rate': '0.0006', 'epoch': '10.84'}
+
 43%|████▎     | 3110/7175 [38:38<45:27,  1.49it/s]
 43%|████▎     | 3111/7175 [38:39<45:09,  1.50it/s]
 43%|████▎     | 3112/7175 [38:40<44:54,  1.51it/s]
 43%|████▎     | 3113/7175 [38:40<44:44,  1.51it/s]
 43%|████▎     | 3114/7175 [38:41<44:37,  1.52it/s]
 43%|████▎     | 3115/7175 [38:42<44:33,  1.52it/s]
 43%|████▎     | 3116/7175 [38:42<44:29,  1.52it/s]
 43%|████▎     | 3117/7175 [38:43<44:25,  1.52it/s]
 43%|████▎     | 3118/7175 [38:44<44:24,  1.52it/s]
 43%|████▎     | 3119/7175 [38:44<44:23,  1.52it/s]
 43%|████▎     | 3120/7175 [38:45<44:23,  1.52it/s]
                                                   
{'loss': '3.191', 'grad_norm': '0.2384', 'learning_rate': '0.0006', 'epoch': '10.87'}
+
 43%|████▎     | 3120/7175 [38:45<44:23,  1.52it/s]
 43%|████▎     | 3121/7175 [38:46<44:24,  1.52it/s]
 44%|████▎     | 3122/7175 [38:46<44:20,  1.52it/s]
 44%|████▎     | 3123/7175 [38:47<44:16,  1.53it/s]
 44%|████▎     | 3124/7175 [38:47<44:18,  1.52it/s]
 44%|████▎     | 3125/7175 [38:48<44:17,  1.52it/s]
 44%|████▎     | 3126/7175 [38:49<44:16,  1.52it/s]
 44%|████▎     | 3127/7175 [38:49<44:15,  1.52it/s]
 44%|████▎     | 3128/7175 [38:50<44:13,  1.53it/s]
 44%|████▎     | 3129/7175 [38:51<44:13,  1.52it/s]
 44%|████▎     | 3130/7175 [38:51<44:13,  1.52it/s]
                                                   
{'loss': '3.202', 'grad_norm': '0.2471', 'learning_rate': '0.0006', 'epoch': '10.91'}
+
 44%|████▎     | 3130/7175 [38:51<44:13,  1.52it/s]
 44%|████▎     | 3131/7175 [38:52<44:12,  1.52it/s]
 44%|████▎     | 3132/7175 [38:53<44:11,  1.53it/s]
 44%|████▎     | 3133/7175 [38:53<44:09,  1.53it/s]
 44%|████▎     | 3134/7175 [38:54<44:10,  1.52it/s]
 44%|████▎     | 3135/7175 [38:55<44:10,  1.52it/s]
 44%|████▎     | 3136/7175 [38:55<44:08,  1.53it/s]
 44%|████▎     | 3137/7175 [38:56<44:05,  1.53it/s]
 44%|████▎     | 3138/7175 [38:57<44:06,  1.53it/s]
 44%|████▎     | 3139/7175 [38:57<44:06,  1.53it/s]
 44%|████▍     | 3140/7175 [38:58<44:06,  1.52it/s]
                                                   
{'loss': '3.201', 'grad_norm': '0.2112', 'learning_rate': '0.0006', 'epoch': '10.94'}
+
 44%|████▍     | 3140/7175 [38:58<44:06,  1.52it/s]
 44%|████▍     | 3141/7175 [38:59<44:05,  1.53it/s]
 44%|████▍     | 3142/7175 [38:59<44:03,  1.53it/s]
 44%|████▍     | 3143/7175 [39:00<44:03,  1.53it/s]
 44%|████▍     | 3144/7175 [39:01<44:04,  1.52it/s]
 44%|████▍     | 3145/7175 [39:01<44:01,  1.53it/s]
 44%|████▍     | 3146/7175 [39:02<43:59,  1.53it/s]
 44%|████▍     | 3147/7175 [39:03<44:01,  1.53it/s]
 44%|████▍     | 3148/7175 [39:03<44:00,  1.53it/s]
 44%|████▍     | 3149/7175 [39:04<43:58,  1.53it/s]
 44%|████▍     | 3150/7175 [39:05<43:55,  1.53it/s]
                                                   
{'loss': '3.195', 'grad_norm': '0.2175', 'learning_rate': '0.0006', 'epoch': '10.98'}
+
 44%|████▍     | 3150/7175 [39:05<43:55,  1.53it/s]
 44%|████▍     | 3151/7175 [39:05<43:57,  1.53it/s]
 44%|████▍     | 3152/7175 [39:06<43:56,  1.53it/s]
 44%|████▍     | 3153/7175 [39:07<43:55,  1.53it/s]
 44%|████▍     | 3154/7175 [39:07<44:03,  1.52it/s]
 44%|████▍     | 3155/7175 [39:08<43:58,  1.52it/s]
 44%|████▍     | 3156/7175 [39:08<43:54,  1.53it/s]
 44%|████▍     | 3158/7175 [39:09<36:44,  1.82it/s]
 44%|████▍     | 3159/7175 [39:10<38:27,  1.74it/s]
 44%|████▍     | 3160/7175 [39:11<39:53,  1.68it/s]
                                                   
{'loss': '3.176', 'grad_norm': '0.3733', 'learning_rate': '0.0006', 'epoch': '11.01'}
+
 44%|████▍     | 3160/7175 [39:11<39:53,  1.68it/s]
 44%|████▍     | 3161/7175 [39:11<40:58,  1.63it/s]
 44%|████▍     | 3162/7175 [39:12<41:45,  1.60it/s]
 44%|████▍     | 3163/7175 [39:13<42:18,  1.58it/s]
 44%|████▍     | 3164/7175 [39:13<42:45,  1.56it/s]
 44%|████▍     | 3165/7175 [39:14<43:03,  1.55it/s]
 44%|████▍     | 3166/7175 [39:15<43:15,  1.54it/s]
 44%|████▍     | 3167/7175 [39:15<43:22,  1.54it/s]
 44%|████▍     | 3168/7175 [39:16<43:29,  1.54it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.11it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.65it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.53it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.00it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.71it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.55it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.45it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.37it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.33it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.29it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.28it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.26it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.26it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.24it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.23it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.22it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.22it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.23it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.22it/s]
                                                   
+
                                               
{'eval_loss': '3.577', 'eval_runtime': '3.251', 'eval_samples_per_second': '750.9', 'eval_steps_per_second': '12', 'epoch': '11.04'}
+
 44%|████▍     | 3168/7175 [39:19<43:29,  1.54it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.22it/s][eval] epoch 11.038 step 3168 loss 3.5766 (best 3.5444 @ ep 10.79, 1/20 since best)
+
+
                                               
 44%|████▍     | 3169/7175 [39:20<1:48:16,  1.62s/it]
 44%|████▍     | 3170/7175 [39:20<1:28:58,  1.33s/it]
                                                     
{'loss': '3.146', 'grad_norm': '0.3301', 'learning_rate': '0.0006', 'epoch': '11.05'}
+
 44%|████▍     | 3170/7175 [39:20<1:28:58,  1.33s/it]
 44%|████▍     | 3171/7175 [39:21<1:15:27,  1.13s/it]
 44%|████▍     | 3172/7175 [39:22<1:05:56,  1.01it/s]
 44%|████▍     | 3173/7175 [39:22<59:14,  1.13it/s]  
 44%|████▍     | 3174/7175 [39:23<54:34,  1.22it/s]
 44%|████▍     | 3175/7175 [39:24<51:20,  1.30it/s]
 44%|████▍     | 3176/7175 [39:24<49:01,  1.36it/s]
 44%|████▍     | 3177/7175 [39:25<47:24,  1.41it/s]
 44%|████▍     | 3178/7175 [39:26<46:17,  1.44it/s]
 44%|████▍     | 3179/7175 [39:26<45:29,  1.46it/s]
 44%|████▍     | 3180/7175 [39:27<45:01,  1.48it/s]
                                                   
{'loss': '3.143', 'grad_norm': '0.235', 'learning_rate': '0.0006', 'epoch': '11.08'}
+
 44%|████▍     | 3180/7175 [39:27<45:01,  1.48it/s]
 44%|████▍     | 3181/7175 [39:28<44:38,  1.49it/s]
 44%|████▍     | 3182/7175 [39:28<44:20,  1.50it/s]
 44%|████▍     | 3183/7175 [39:29<44:09,  1.51it/s]
 44%|████▍     | 3184/7175 [39:30<43:58,  1.51it/s]
 44%|████▍     | 3185/7175 [39:30<43:48,  1.52it/s]
 44%|████▍     | 3186/7175 [39:31<43:44,  1.52it/s]
 44%|████▍     | 3187/7175 [39:32<43:41,  1.52it/s]
 44%|████▍     | 3188/7175 [39:32<43:38,  1.52it/s]
 44%|████▍     | 3189/7175 [39:33<43:35,  1.52it/s]
 44%|████▍     | 3190/7175 [39:34<43:35,  1.52it/s]
                                                   
{'loss': '3.136', 'grad_norm': '0.2306', 'learning_rate': '0.0006', 'epoch': '11.12'}
+
 44%|████▍     | 3190/7175 [39:34<43:35,  1.52it/s]
 44%|████▍     | 3191/7175 [39:34<43:35,  1.52it/s]
 44%|████▍     | 3192/7175 [39:35<43:34,  1.52it/s]
 45%|████▍     | 3193/7175 [39:36<43:32,  1.52it/s]
 45%|████▍     | 3194/7175 [39:36<43:31,  1.52it/s]
 45%|████▍     | 3195/7175 [39:37<43:31,  1.52it/s]
 45%|████▍     | 3196/7175 [39:37<43:30,  1.52it/s]
 45%|████▍     | 3197/7175 [39:38<43:29,  1.52it/s]
 45%|████▍     | 3198/7175 [39:39<43:27,  1.52it/s]
 45%|████▍     | 3199/7175 [39:39<43:24,  1.53it/s]
 45%|████▍     | 3200/7175 [39:40<43:26,  1.53it/s]
                                                   
{'loss': '3.129', 'grad_norm': '0.2148', 'learning_rate': '0.0006', 'epoch': '11.15'}
+
 45%|████▍     | 3200/7175 [39:40<43:26,  1.53it/s]
 45%|████▍     | 3201/7175 [39:41<43:27,  1.52it/s]
 45%|████▍     | 3202/7175 [39:41<43:25,  1.52it/s]
 45%|████▍     | 3203/7175 [39:42<43:23,  1.53it/s]
 45%|████▍     | 3204/7175 [39:43<43:22,  1.53it/s]
 45%|████▍     | 3205/7175 [39:43<43:22,  1.53it/s]
 45%|████▍     | 3206/7175 [39:44<43:22,  1.53it/s]
 45%|████▍     | 3207/7175 [39:45<43:20,  1.53it/s]
 45%|████▍     | 3208/7175 [39:45<43:24,  1.52it/s]
 45%|████▍     | 3209/7175 [39:46<43:21,  1.52it/s]
 45%|████▍     | 3210/7175 [39:47<43:19,  1.53it/s]
                                                   
{'loss': '3.121', 'grad_norm': '0.2107', 'learning_rate': '0.0006', 'epoch': '11.19'}
+
 45%|████▍     | 3210/7175 [39:47<43:19,  1.53it/s]
 45%|████▍     | 3211/7175 [39:47<43:20,  1.52it/s]
 45%|████▍     | 3212/7175 [39:48<43:18,  1.52it/s]
 45%|████▍     | 3213/7175 [39:49<43:17,  1.53it/s]
 45%|████▍     | 3214/7175 [39:49<43:17,  1.52it/s]
 45%|████▍     | 3215/7175 [39:50<43:16,  1.53it/s]
 45%|████▍     | 3216/7175 [39:51<43:14,  1.53it/s]
 45%|████▍     | 3217/7175 [39:51<43:12,  1.53it/s]
 45%|████▍     | 3218/7175 [39:52<43:13,  1.53it/s]
 45%|████▍     | 3219/7175 [39:53<43:13,  1.53it/s]
 45%|████▍     | 3220/7175 [39:53<43:13,  1.52it/s]
                                                   
{'loss': '3.127', 'grad_norm': '0.2334', 'learning_rate': '0.0006', 'epoch': '11.22'}
+
 45%|████▍     | 3220/7175 [39:53<43:13,  1.52it/s]
 45%|████▍     | 3221/7175 [39:54<43:12,  1.53it/s]
 45%|████▍     | 3222/7175 [39:55<43:11,  1.53it/s]
 45%|████▍     | 3223/7175 [39:55<43:11,  1.52it/s]
 45%|████▍     | 3224/7175 [39:56<43:10,  1.52it/s]
 45%|████▍     | 3225/7175 [39:57<43:09,  1.53it/s]
 45%|████▍     | 3226/7175 [39:57<43:07,  1.53it/s]
 45%|████▍     | 3227/7175 [39:58<43:08,  1.52it/s]
 45%|████▍     | 3228/7175 [39:58<43:09,  1.52it/s]
 45%|████▌     | 3229/7175 [39:59<43:08,  1.52it/s]
 45%|████▌     | 3230/7175 [40:00<43:06,  1.53it/s]
                                                   
{'loss': '3.144', 'grad_norm': '0.2372', 'learning_rate': '0.0006', 'epoch': '11.26'}
+
 45%|████▌     | 3230/7175 [40:00<43:06,  1.53it/s]
 45%|████▌     | 3231/7175 [40:00<43:04,  1.53it/s]
 45%|████▌     | 3232/7175 [40:01<43:06,  1.52it/s]
 45%|████▌     | 3233/7175 [40:02<43:04,  1.53it/s]
 45%|████▌     | 3234/7175 [40:02<43:03,  1.53it/s]
 45%|████▌     | 3235/7175 [40:03<43:03,  1.53it/s]
 45%|████▌     | 3236/7175 [40:04<43:01,  1.53it/s]
 45%|████▌     | 3237/7175 [40:04<43:03,  1.52it/s]
 45%|████▌     | 3238/7175 [40:05<43:01,  1.52it/s]
 45%|████▌     | 3239/7175 [40:06<43:01,  1.52it/s]
 45%|████▌     | 3240/7175 [40:06<43:00,  1.53it/s]
                                                   
{'loss': '3.133', 'grad_norm': '0.2101', 'learning_rate': '0.0006', 'epoch': '11.29'}
+
 45%|████▌     | 3240/7175 [40:06<43:00,  1.53it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.39it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.75it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.59it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.06it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.75it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.57it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.43it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.37it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.33it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.29it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.27it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.27it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.24it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.24it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.23it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.23it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.23it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.23it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.22it/s]
                                                   
+
                                               
{'eval_loss': '3.547', 'eval_runtime': '3.262', 'eval_samples_per_second': '748.3', 'eval_steps_per_second': '11.96', 'epoch': '11.29'}
+
 45%|████▌     | 3240/7175 [40:10<43:00,  1.53it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.22it/s][eval] epoch 11.29 step 3240 loss 3.5475 (best 3.5444 @ ep 10.79, 2/20 since best)
+
+
                                               
 45%|████▌     | 3241/7175 [40:10<1:47:14,  1.64s/it]
 45%|████▌     | 3242/7175 [40:11<1:27:54,  1.34s/it]
 45%|████▌     | 3243/7175 [40:12<1:14:24,  1.14s/it]
 45%|████▌     | 3244/7175 [40:12<1:04:57,  1.01it/s]
 45%|████▌     | 3245/7175 [40:13<58:19,  1.12it/s]  
 45%|████▌     | 3246/7175 [40:14<53:41,  1.22it/s]
 45%|████▌     | 3247/7175 [40:14<50:28,  1.30it/s]
 45%|████▌     | 3248/7175 [40:15<48:11,  1.36it/s]
 45%|████▌     | 3249/7175 [40:16<46:34,  1.40it/s]
 45%|████▌     | 3250/7175 [40:16<45:26,  1.44it/s]
                                                   
{'loss': '3.153', 'grad_norm': '0.2186', 'learning_rate': '0.0006', 'epoch': '11.33'}
+
 45%|████▌     | 3250/7175 [40:16<45:26,  1.44it/s]
 45%|████▌     | 3251/7175 [40:17<44:41,  1.46it/s]
 45%|████▌     | 3252/7175 [40:17<44:10,  1.48it/s]
 45%|████▌     | 3253/7175 [40:18<43:45,  1.49it/s]
 45%|████▌     | 3254/7175 [40:19<43:26,  1.50it/s]
 45%|████▌     | 3255/7175 [40:19<43:14,  1.51it/s]
 45%|████▌     | 3256/7175 [40:20<43:08,  1.51it/s]
 45%|████▌     | 3257/7175 [40:21<43:01,  1.52it/s]
 45%|████▌     | 3258/7175 [40:21<42:56,  1.52it/s]
 45%|████▌     | 3259/7175 [40:22<42:51,  1.52it/s]
 45%|████▌     | 3260/7175 [40:23<42:50,  1.52it/s]
                                                   
{'loss': '3.143', 'grad_norm': '0.2215', 'learning_rate': '0.0006', 'epoch': '11.36'}
+
 45%|████▌     | 3260/7175 [40:23<42:50,  1.52it/s]
 45%|████▌     | 3261/7175 [40:23<42:50,  1.52it/s]
 45%|████▌     | 3262/7175 [40:24<42:47,  1.52it/s]
 45%|████▌     | 3263/7175 [40:25<42:45,  1.52it/s]
 45%|████▌     | 3264/7175 [40:25<42:43,  1.53it/s]
 46%|████▌     | 3265/7175 [40:26<42:44,  1.52it/s]
 46%|████▌     | 3266/7175 [40:27<42:43,  1.53it/s]
 46%|████▌     | 3267/7175 [40:27<42:41,  1.53it/s]
 46%|████▌     | 3268/7175 [40:28<42:39,  1.53it/s]
 46%|████▌     | 3269/7175 [40:29<42:40,  1.53it/s]
 46%|████▌     | 3270/7175 [40:29<42:39,  1.53it/s]
                                                   
{'loss': '3.149', 'grad_norm': '0.2324', 'learning_rate': '0.0006', 'epoch': '11.39'}
+
 46%|████▌     | 3270/7175 [40:29<42:39,  1.53it/s]
 46%|████▌     | 3271/7175 [40:30<42:40,  1.52it/s]
 46%|████▌     | 3272/7175 [40:31<42:38,  1.53it/s]
 46%|████▌     | 3273/7175 [40:31<42:37,  1.53it/s]
 46%|████▌     | 3274/7175 [40:32<42:39,  1.52it/s]
 46%|████▌     | 3275/7175 [40:33<42:37,  1.52it/s]
 46%|████▌     | 3276/7175 [40:33<42:37,  1.52it/s]
 46%|████▌     | 3277/7175 [40:34<42:34,  1.53it/s]
 46%|████▌     | 3278/7175 [40:35<42:34,  1.53it/s]
 46%|████▌     | 3279/7175 [40:35<42:33,  1.53it/s]
 46%|████▌     | 3280/7175 [40:36<42:31,  1.53it/s]
                                                   
{'loss': '3.15', 'grad_norm': '0.2498', 'learning_rate': '0.0006', 'epoch': '11.43'}
+
 46%|████▌     | 3280/7175 [40:36<42:31,  1.53it/s]
 46%|████▌     | 3281/7175 [40:36<42:34,  1.52it/s]
 46%|████▌     | 3282/7175 [40:37<42:33,  1.52it/s]
 46%|████▌     | 3283/7175 [40:38<42:33,  1.52it/s]
 46%|████▌     | 3284/7175 [40:38<42:31,  1.53it/s]
 46%|████▌     | 3285/7175 [40:39<42:30,  1.53it/s]
 46%|████▌     | 3286/7175 [40:40<42:28,  1.53it/s]
 46%|████▌     | 3287/7175 [40:40<42:28,  1.53it/s]
 46%|████▌     | 3288/7175 [40:41<42:27,  1.53it/s]
 46%|████▌     | 3289/7175 [40:42<42:27,  1.53it/s]
 46%|████▌     | 3290/7175 [40:42<42:25,  1.53it/s]
                                                   
{'loss': '3.156', 'grad_norm': '0.2305', 'learning_rate': '0.0006', 'epoch': '11.46'}
+
 46%|████▌     | 3290/7175 [40:42<42:25,  1.53it/s]
 46%|████▌     | 3291/7175 [40:43<42:25,  1.53it/s]
 46%|████▌     | 3292/7175 [40:44<42:26,  1.52it/s]
 46%|████▌     | 3293/7175 [40:44<42:24,  1.53it/s]
 46%|████▌     | 3294/7175 [40:45<42:24,  1.53it/s]
 46%|████▌     | 3295/7175 [40:46<42:23,  1.53it/s]
 46%|████▌     | 3296/7175 [40:46<42:23,  1.53it/s]
 46%|████▌     | 3297/7175 [40:47<42:22,  1.53it/s]
 46%|████▌     | 3298/7175 [40:48<42:21,  1.53it/s]
 46%|████▌     | 3299/7175 [40:48<42:19,  1.53it/s]
 46%|████▌     | 3300/7175 [40:49<42:19,  1.53it/s]
                                                   
{'loss': '3.166', 'grad_norm': '0.2267', 'learning_rate': '0.0006', 'epoch': '11.5'}
+
 46%|████▌     | 3300/7175 [40:49<42:19,  1.53it/s]
 46%|████▌     | 3301/7175 [40:50<42:20,  1.52it/s]
 46%|████▌     | 3302/7175 [40:50<42:20,  1.52it/s]
 46%|████▌     | 3303/7175 [40:51<42:17,  1.53it/s]
 46%|████▌     | 3304/7175 [40:52<42:14,  1.53it/s]
 46%|████▌     | 3305/7175 [40:52<42:17,  1.53it/s]
 46%|████▌     | 3306/7175 [40:53<42:15,  1.53it/s]
 46%|████▌     | 3307/7175 [40:54<42:14,  1.53it/s]
 46%|████▌     | 3308/7175 [40:54<42:12,  1.53it/s]
 46%|████▌     | 3309/7175 [40:55<42:14,  1.53it/s]
 46%|████▌     | 3310/7175 [40:55<42:12,  1.53it/s]
                                                   
{'loss': '3.158', 'grad_norm': '0.2434', 'learning_rate': '0.0006', 'epoch': '11.53'}
+
 46%|████▌     | 3310/7175 [40:55<42:12,  1.53it/s]
 46%|████▌     | 3311/7175 [40:56<42:14,  1.52it/s]
 46%|████▌     | 3312/7175 [40:57<42:11,  1.53it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.45it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.74it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.58it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.04it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.75it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.56it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.45it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.37it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.32it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.30it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.29it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.27it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.25it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.24it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.23it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.23it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.24it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.23it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.22it/s]
                                                   
+
                                               
{'eval_loss': '3.54', 'eval_runtime': '3.254', 'eval_samples_per_second': '750.2', 'eval_steps_per_second': '11.98', 'epoch': '11.54'}
+
 46%|████▌     | 3312/7175 [41:00<42:11,  1.53it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.22it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-3312
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-3312/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-3312/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.66it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.66it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-3312/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-3312/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-3312/tokenizer_config.json
+[eval] epoch 11.542 step 3312 loss 3.5400 (best 3.5400 @ ep 11.542, 0/20 since best)
+
+
                                               
 46%|████▌     | 3313/7175 [41:03<2:26:25,  2.27s/it]
 46%|████▌     | 3314/7175 [41:04<1:55:06,  1.79s/it]
 46%|████▌     | 3315/7175 [41:04<1:33:10,  1.45s/it]
 46%|████▌     | 3316/7175 [41:05<1:17:48,  1.21s/it]
 46%|████▌     | 3317/7175 [41:05<1:07:06,  1.04s/it]
 46%|████▌     | 3318/7175 [41:06<59:36,  1.08it/s]  
 46%|████▋     | 3319/7175 [41:07<54:19,  1.18it/s]
 46%|████▋     | 3320/7175 [41:07<50:37,  1.27it/s]
                                                   
{'loss': '3.159', 'grad_norm': '0.2196', 'learning_rate': '0.0006', 'epoch': '11.57'}
+
 46%|████▋     | 3320/7175 [41:07<50:37,  1.27it/s]
 46%|████▋     | 3321/7175 [41:08<48:06,  1.34it/s]
 46%|████▋     | 3322/7175 [41:09<46:17,  1.39it/s]
 46%|████▋     | 3323/7175 [41:09<44:59,  1.43it/s]
 46%|████▋     | 3324/7175 [41:10<44:04,  1.46it/s]
 46%|████▋     | 3325/7175 [41:11<43:29,  1.48it/s]
 46%|████▋     | 3326/7175 [41:11<43:02,  1.49it/s]
 46%|████▋     | 3327/7175 [41:12<42:42,  1.50it/s]
 46%|████▋     | 3328/7175 [41:13<42:27,  1.51it/s]
 46%|████▋     | 3329/7175 [41:13<42:20,  1.51it/s]
 46%|████▋     | 3330/7175 [41:14<42:13,  1.52it/s]
                                                   
{'loss': '3.155', 'grad_norm': '0.2043', 'learning_rate': '0.0006', 'epoch': '11.6'}
+
 46%|████▋     | 3330/7175 [41:14<42:13,  1.52it/s]
 46%|████▋     | 3331/7175 [41:15<42:09,  1.52it/s]
 46%|████▋     | 3332/7175 [41:15<42:02,  1.52it/s]
 46%|████▋     | 3333/7175 [41:16<42:02,  1.52it/s]
 46%|████▋     | 3334/7175 [41:17<41:59,  1.52it/s]
 46%|████▋     | 3335/7175 [41:17<41:57,  1.53it/s]
 46%|████▋     | 3336/7175 [41:18<41:54,  1.53it/s]
 47%|████▋     | 3337/7175 [41:19<41:55,  1.53it/s]
 47%|████▋     | 3338/7175 [41:19<41:54,  1.53it/s]
 47%|████▋     | 3339/7175 [41:20<41:52,  1.53it/s]
 47%|████▋     | 3340/7175 [41:21<41:51,  1.53it/s]
                                                   
{'loss': '3.151', 'grad_norm': '0.2161', 'learning_rate': '0.0006', 'epoch': '11.64'}
+
 47%|████▋     | 3340/7175 [41:21<41:51,  1.53it/s]
 47%|████▋     | 3341/7175 [41:21<41:53,  1.53it/s]
 47%|████▋     | 3342/7175 [41:22<41:51,  1.53it/s]
 47%|████▋     | 3343/7175 [41:23<41:50,  1.53it/s]
 47%|████▋     | 3344/7175 [41:23<41:47,  1.53it/s]
 47%|████▋     | 3345/7175 [41:24<41:49,  1.53it/s]
 47%|████▋     | 3346/7175 [41:24<41:48,  1.53it/s]
 47%|████▋     | 3347/7175 [41:25<41:46,  1.53it/s]
 47%|████▋     | 3348/7175 [41:26<41:44,  1.53it/s]
 47%|████▋     | 3349/7175 [41:26<41:46,  1.53it/s]
 47%|████▋     | 3350/7175 [41:27<41:47,  1.53it/s]
                                                   
{'loss': '3.167', 'grad_norm': '0.21', 'learning_rate': '0.0006', 'epoch': '11.67'}
+
 47%|████▋     | 3350/7175 [41:27<41:47,  1.53it/s]
 47%|████▋     | 3351/7175 [41:28<41:46,  1.53it/s]
 47%|████▋     | 3352/7175 [41:28<41:43,  1.53it/s]
 47%|████▋     | 3353/7175 [41:29<41:44,  1.53it/s]
 47%|████▋     | 3354/7175 [41:30<41:43,  1.53it/s]
 47%|████▋     | 3355/7175 [41:30<41:47,  1.52it/s]
 47%|████▋     | 3356/7175 [41:31<41:43,  1.53it/s]
 47%|████▋     | 3357/7175 [41:32<41:41,  1.53it/s]
 47%|████▋     | 3358/7175 [41:32<41:42,  1.53it/s]
 47%|████▋     | 3359/7175 [41:33<41:40,  1.53it/s]
 47%|████▋     | 3360/7175 [41:34<41:38,  1.53it/s]
                                                   
{'loss': '3.165', 'grad_norm': '0.2324', 'learning_rate': '0.0006', 'epoch': '11.71'}
+
 47%|████▋     | 3360/7175 [41:34<41:38,  1.53it/s]
 47%|████▋     | 3361/7175 [41:34<41:37,  1.53it/s]
 47%|████▋     | 3362/7175 [41:35<41:39,  1.53it/s]
 47%|████▋     | 3363/7175 [41:36<41:37,  1.53it/s]
 47%|████▋     | 3364/7175 [41:36<41:35,  1.53it/s]
 47%|████▋     | 3365/7175 [41:37<41:33,  1.53it/s]
 47%|████▋     | 3366/7175 [41:38<41:35,  1.53it/s]
 47%|████▋     | 3367/7175 [41:38<41:34,  1.53it/s]
 47%|████▋     | 3368/7175 [41:39<41:31,  1.53it/s]
 47%|████▋     | 3369/7175 [41:40<41:30,  1.53it/s]
 47%|████▋     | 3370/7175 [41:40<41:31,  1.53it/s]
                                                   
{'loss': '3.164', 'grad_norm': '0.2208', 'learning_rate': '0.0006', 'epoch': '11.74'}
+
 47%|████▋     | 3370/7175 [41:40<41:31,  1.53it/s]
 47%|████▋     | 3371/7175 [41:41<41:33,  1.53it/s]
 47%|████▋     | 3372/7175 [41:42<41:30,  1.53it/s]
 47%|████▋     | 3373/7175 [41:42<41:29,  1.53it/s]
 47%|████▋     | 3374/7175 [41:43<41:30,  1.53it/s]
 47%|████▋     | 3375/7175 [41:43<41:28,  1.53it/s]
 47%|████▋     | 3376/7175 [41:44<41:26,  1.53it/s]
 47%|████▋     | 3377/7175 [41:45<41:27,  1.53it/s]
 47%|████▋     | 3378/7175 [41:45<41:25,  1.53it/s]
 47%|████▋     | 3379/7175 [41:46<41:25,  1.53it/s]
 47%|████▋     | 3380/7175 [41:47<41:23,  1.53it/s]
                                                   
{'loss': '3.176', 'grad_norm': '0.2405', 'learning_rate': '0.0006', 'epoch': '11.78'}
+
 47%|████▋     | 3380/7175 [41:47<41:23,  1.53it/s]
 47%|████▋     | 3381/7175 [41:47<41:25,  1.53it/s]
 47%|████▋     | 3382/7175 [41:48<41:25,  1.53it/s]
 47%|████▋     | 3383/7175 [41:49<41:24,  1.53it/s]
 47%|████▋     | 3384/7175 [41:49<41:22,  1.53it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.42it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.75it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.59it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.05it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.74it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.57it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.45it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.38it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.33it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.30it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.27it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.27it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.25it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.25it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.25it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.24it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.24it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.24it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s]
                                                   
+
                                               
{'eval_loss': '3.533', 'eval_runtime': '3.247', 'eval_samples_per_second': '751.7', 'eval_steps_per_second': '12.01', 'epoch': '11.79'}
+
 47%|████▋     | 3384/7175 [41:53<41:22,  1.53it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-3384
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-3384/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-3384/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.84it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.84it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-3384/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-3384/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-3384/tokenizer_config.json
+[eval] epoch 11.793 step 3384 loss 3.5332 (best 3.5332 @ ep 11.793, 0/20 since best)
+
+
                                               
 47%|████▋     | 3385/7175 [41:55<2:22:06,  2.25s/it]
 47%|████▋     | 3386/7175 [41:56<1:51:49,  1.77s/it]
 47%|████▋     | 3387/7175 [41:57<1:30:37,  1.44s/it]
 47%|████▋     | 3388/7175 [41:57<1:15:48,  1.20s/it]
 47%|████▋     | 3389/7175 [41:58<1:05:26,  1.04s/it]
 47%|████▋     | 3390/7175 [41:59<58:09,  1.08it/s]  
                                                   
{'loss': '3.159', 'grad_norm': '0.2402', 'learning_rate': '0.0006', 'epoch': '11.81'}
+
 47%|████▋     | 3390/7175 [41:59<58:09,  1.08it/s]
 47%|████▋     | 3391/7175 [41:59<53:07,  1.19it/s]
 47%|████▋     | 3392/7175 [42:00<49:35,  1.27it/s]
 47%|████▋     | 3393/7175 [42:01<47:06,  1.34it/s]
 47%|████▋     | 3394/7175 [42:01<45:19,  1.39it/s]
 47%|████▋     | 3395/7175 [42:02<44:06,  1.43it/s]
 47%|████▋     | 3396/7175 [42:03<43:16,  1.46it/s]
 47%|████▋     | 3397/7175 [42:03<42:39,  1.48it/s]
 47%|████▋     | 3398/7175 [42:04<42:11,  1.49it/s]
 47%|████▋     | 3399/7175 [42:04<41:52,  1.50it/s]
 47%|████▋     | 3400/7175 [42:05<41:41,  1.51it/s]
                                                   
{'loss': '3.169', 'grad_norm': '0.2389', 'learning_rate': '0.0006', 'epoch': '11.85'}
+
 47%|████▋     | 3400/7175 [42:05<41:41,  1.51it/s]
 47%|████▋     | 3401/7175 [42:06<41:33,  1.51it/s]
 47%|████▋     | 3402/7175 [42:06<41:25,  1.52it/s]
 47%|████▋     | 3403/7175 [42:07<41:19,  1.52it/s]
 47%|████▋     | 3404/7175 [42:08<41:18,  1.52it/s]
 47%|████▋     | 3405/7175 [42:08<41:15,  1.52it/s]
 47%|████▋     | 3406/7175 [42:09<41:11,  1.52it/s]
 47%|████▋     | 3407/7175 [42:10<41:09,  1.53it/s]
 47%|████▋     | 3408/7175 [42:10<41:10,  1.52it/s]
 48%|████▊     | 3409/7175 [42:11<41:09,  1.53it/s]
 48%|████▊     | 3410/7175 [42:12<41:06,  1.53it/s]
                                                   
{'loss': '3.175', 'grad_norm': '0.2094', 'learning_rate': '0.0006', 'epoch': '11.88'}
+
 48%|████▊     | 3410/7175 [42:12<41:06,  1.53it/s]
 48%|████▊     | 3411/7175 [42:12<41:05,  1.53it/s]
 48%|████▊     | 3412/7175 [42:13<41:07,  1.53it/s]
 48%|████▊     | 3413/7175 [42:14<41:05,  1.53it/s]
 48%|████▊     | 3414/7175 [42:14<41:03,  1.53it/s]
 48%|████▊     | 3415/7175 [42:15<41:01,  1.53it/s]
 48%|████▊     | 3416/7175 [42:16<41:03,  1.53it/s]
 48%|████▊     | 3417/7175 [42:16<41:01,  1.53it/s]
 48%|████▊     | 3418/7175 [42:17<40:59,  1.53it/s]
 48%|████▊     | 3419/7175 [42:18<40:57,  1.53it/s]
 48%|████▊     | 3420/7175 [42:18<41:00,  1.53it/s]
                                                   
{'loss': '3.167', 'grad_norm': '0.2113', 'learning_rate': '0.0006', 'epoch': '11.92'}
+
 48%|████▊     | 3420/7175 [42:18<41:00,  1.53it/s]
 48%|████▊     | 3421/7175 [42:19<41:03,  1.52it/s]
 48%|████▊     | 3422/7175 [42:20<41:02,  1.52it/s]
 48%|████▊     | 3423/7175 [42:20<40:59,  1.53it/s]
 48%|████▊     | 3424/7175 [42:21<40:57,  1.53it/s]
 48%|████▊     | 3425/7175 [42:22<40:59,  1.52it/s]
 48%|████▊     | 3426/7175 [42:22<40:57,  1.53it/s]
 48%|████▊     | 3427/7175 [42:23<40:56,  1.53it/s]
 48%|████▊     | 3428/7175 [42:24<40:54,  1.53it/s]
 48%|████▊     | 3429/7175 [42:24<40:56,  1.53it/s]
 48%|████▊     | 3430/7175 [42:25<40:55,  1.53it/s]
                                                   
{'loss': '3.165', 'grad_norm': '0.2045', 'learning_rate': '0.0006', 'epoch': '11.95'}
+
 48%|████▊     | 3430/7175 [42:25<40:55,  1.53it/s]
 48%|████▊     | 3431/7175 [42:25<40:56,  1.52it/s]
 48%|████▊     | 3432/7175 [42:26<40:55,  1.52it/s]
 48%|████▊     | 3433/7175 [42:27<40:52,  1.53it/s]
 48%|████▊     | 3434/7175 [42:27<40:53,  1.52it/s]
 48%|████▊     | 3435/7175 [42:28<40:52,  1.52it/s]
 48%|████▊     | 3436/7175 [42:29<40:52,  1.52it/s]
 48%|████▊     | 3437/7175 [42:29<40:49,  1.53it/s]
 48%|████▊     | 3438/7175 [42:30<40:49,  1.53it/s]
 48%|████▊     | 3439/7175 [42:31<40:51,  1.52it/s]
 48%|████▊     | 3440/7175 [42:31<40:49,  1.52it/s]
                                                   
{'loss': '3.159', 'grad_norm': '0.2243', 'learning_rate': '0.0006', 'epoch': '11.99'}
+
 48%|████▊     | 3440/7175 [42:31<40:49,  1.52it/s]
 48%|████▊     | 3441/7175 [42:32<41:00,  1.52it/s]
 48%|████▊     | 3442/7175 [42:33<40:56,  1.52it/s]
 48%|████▊     | 3443/7175 [42:33<40:52,  1.52it/s]
 48%|████▊     | 3445/7175 [42:34<33:49,  1.84it/s]
 48%|████▊     | 3446/7175 [42:35<35:32,  1.75it/s]
 48%|████▊     | 3447/7175 [42:35<36:53,  1.68it/s]
 48%|████▊     | 3448/7175 [42:36<43:46,  1.42it/s]
 48%|████▊     | 3449/7175 [42:37<42:51,  1.45it/s]
 48%|████▊     | 3450/7175 [42:38<42:15,  1.47it/s]
                                                   
{'loss': '3.129', 'grad_norm': '0.3095', 'learning_rate': '0.0006', 'epoch': '12.02'}
+
 48%|████▊     | 3450/7175 [42:38<42:15,  1.47it/s]
 48%|████▊     | 3451/7175 [42:38<41:49,  1.48it/s]
 48%|████▊     | 3452/7175 [42:39<41:29,  1.50it/s]
 48%|████▊     | 3453/7175 [42:40<41:13,  1.50it/s]
 48%|████▊     | 3454/7175 [42:40<41:02,  1.51it/s]
 48%|████▊     | 3455/7175 [42:41<40:56,  1.51it/s]
 48%|████▊     | 3456/7175 [42:42<40:50,  1.52it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.51it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.75it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.60it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.06it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.75it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.57it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.45it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.37it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.33it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.28it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.29it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.28it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.26it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.26it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.25it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.23it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.25it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.24it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s]
                                                   
+
                                               
{'eval_loss': '3.556', 'eval_runtime': '3.26', 'eval_samples_per_second': '748.7', 'eval_steps_per_second': '11.96', 'epoch': '12.04'}
+
 48%|████▊     | 3456/7175 [42:45<40:50,  1.52it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s][eval] epoch 12.042 step 3456 loss 3.5562 (best 3.5332 @ ep 11.793, 1/20 since best)
+
+
                                               
 48%|████▊     | 3457/7175 [42:46<1:41:10,  1.63s/it]
 48%|████▊     | 3458/7175 [42:46<1:23:00,  1.34s/it]
 48%|████▊     | 3459/7175 [42:47<1:10:16,  1.13s/it]
 48%|████▊     | 3460/7175 [42:48<1:01:23,  1.01it/s]
                                                     
{'loss': '3.115', 'grad_norm': '0.257', 'learning_rate': '0.0006', 'epoch': '12.06'}
+
 48%|████▊     | 3460/7175 [42:48<1:01:23,  1.01it/s]
 48%|████▊     | 3461/7175 [42:48<55:09,  1.12it/s]  
 48%|████▊     | 3462/7175 [42:49<50:45,  1.22it/s]
 48%|████▊     | 3463/7175 [42:50<47:40,  1.30it/s]
 48%|████▊     | 3464/7175 [42:50<45:32,  1.36it/s]
 48%|████▊     | 3465/7175 [42:51<44:03,  1.40it/s]
 48%|████▊     | 3466/7175 [42:52<42:59,  1.44it/s]
 48%|████▊     | 3467/7175 [42:52<42:13,  1.46it/s]
 48%|████▊     | 3468/7175 [42:53<41:40,  1.48it/s]
 48%|████▊     | 3469/7175 [42:54<41:20,  1.49it/s]
 48%|████▊     | 3470/7175 [42:54<41:03,  1.50it/s]
                                                   
{'loss': '3.094', 'grad_norm': '0.2571', 'learning_rate': '0.0006', 'epoch': '12.09'}
+
 48%|████▊     | 3470/7175 [42:54<41:03,  1.50it/s]
 48%|████▊     | 3471/7175 [42:55<40:52,  1.51it/s]
 48%|████▊     | 3472/7175 [42:55<40:41,  1.52it/s]
 48%|████▊     | 3473/7175 [42:56<40:38,  1.52it/s]
 48%|████▊     | 3474/7175 [42:57<40:33,  1.52it/s]
 48%|████▊     | 3475/7175 [42:57<40:29,  1.52it/s]
 48%|████▊     | 3476/7175 [42:58<40:25,  1.52it/s]
 48%|████▊     | 3477/7175 [42:59<40:26,  1.52it/s]
 48%|████▊     | 3478/7175 [42:59<40:24,  1.53it/s]
 48%|████▊     | 3479/7175 [43:00<40:22,  1.53it/s]
 49%|████▊     | 3480/7175 [43:01<40:19,  1.53it/s]
                                                   
{'loss': '3.105', 'grad_norm': '0.2772', 'learning_rate': '0.0006', 'epoch': '12.13'}
+
 49%|████▊     | 3480/7175 [43:01<40:19,  1.53it/s]
 49%|████▊     | 3481/7175 [43:01<40:21,  1.53it/s]
 49%|████▊     | 3482/7175 [43:02<40:19,  1.53it/s]
 49%|████▊     | 3483/7175 [43:03<40:18,  1.53it/s]
 49%|████▊     | 3484/7175 [43:03<40:15,  1.53it/s]
 49%|████▊     | 3485/7175 [43:04<40:18,  1.53it/s]
 49%|████▊     | 3486/7175 [43:05<40:17,  1.53it/s]
 49%|████▊     | 3487/7175 [43:05<40:15,  1.53it/s]
 49%|████▊     | 3488/7175 [43:06<40:15,  1.53it/s]
 49%|████▊     | 3489/7175 [43:07<40:15,  1.53it/s]
 49%|████▊     | 3490/7175 [43:07<40:14,  1.53it/s]
                                                   
{'loss': '3.114', 'grad_norm': '0.2161', 'learning_rate': '0.0006', 'epoch': '12.16'}
+
 49%|████▊     | 3490/7175 [43:07<40:14,  1.53it/s]
 49%|████▊     | 3491/7175 [43:08<40:14,  1.53it/s]
 49%|████▊     | 3492/7175 [43:09<40:12,  1.53it/s]
 49%|████▊     | 3493/7175 [43:09<40:12,  1.53it/s]
 49%|████▊     | 3494/7175 [43:10<40:13,  1.53it/s]
 49%|████▊     | 3495/7175 [43:11<40:12,  1.53it/s]
 49%|████▊     | 3496/7175 [43:11<40:12,  1.53it/s]
 49%|████▊     | 3497/7175 [43:12<40:08,  1.53it/s]
 49%|████▉     | 3498/7175 [43:13<40:11,  1.53it/s]
 49%|████▉     | 3499/7175 [43:13<40:09,  1.53it/s]
 49%|████▉     | 3500/7175 [43:14<40:07,  1.53it/s]
                                                   
{'loss': '3.099', 'grad_norm': '0.1999', 'learning_rate': '0.0006', 'epoch': '12.2'}
+
 49%|████▉     | 3500/7175 [43:14<40:07,  1.53it/s]
 49%|████▉     | 3501/7175 [43:14<40:06,  1.53it/s]
 49%|████▉     | 3502/7175 [43:15<40:07,  1.53it/s]
 49%|████▉     | 3503/7175 [43:16<40:05,  1.53it/s]
 49%|████▉     | 3504/7175 [43:16<40:04,  1.53it/s]
 49%|████▉     | 3505/7175 [43:17<40:03,  1.53it/s]
 49%|████▉     | 3506/7175 [43:18<40:04,  1.53it/s]
 49%|████▉     | 3507/7175 [43:18<40:02,  1.53it/s]
 49%|████▉     | 3508/7175 [43:19<40:01,  1.53it/s]
 49%|████▉     | 3509/7175 [43:20<39:59,  1.53it/s]
 49%|████▉     | 3510/7175 [43:20<40:01,  1.53it/s]
                                                   
{'loss': '3.109', 'grad_norm': '0.2295', 'learning_rate': '0.0006', 'epoch': '12.23'}
+
 49%|████▉     | 3510/7175 [43:20<40:01,  1.53it/s]
 49%|████▉     | 3511/7175 [43:21<40:00,  1.53it/s]
 49%|████▉     | 3512/7175 [43:22<39:59,  1.53it/s]
 49%|████▉     | 3513/7175 [43:22<39:56,  1.53it/s]
 49%|████▉     | 3514/7175 [43:23<39:59,  1.53it/s]
 49%|████▉     | 3515/7175 [43:24<39:58,  1.53it/s]
 49%|████▉     | 3516/7175 [43:24<39:57,  1.53it/s]
 49%|████▉     | 3517/7175 [43:25<39:54,  1.53it/s]
 49%|████▉     | 3518/7175 [43:26<39:57,  1.53it/s]
 49%|████▉     | 3519/7175 [43:26<39:56,  1.53it/s]
 49%|████▉     | 3520/7175 [43:27<39:55,  1.53it/s]
                                                   
{'loss': '3.118', 'grad_norm': '0.2194', 'learning_rate': '0.0006', 'epoch': '12.27'}
+
 49%|████▉     | 3520/7175 [43:27<39:55,  1.53it/s]
 49%|████▉     | 3521/7175 [43:28<39:54,  1.53it/s]
 49%|████▉     | 3522/7175 [43:28<39:54,  1.53it/s]
 49%|████▉     | 3523/7175 [43:29<39:54,  1.53it/s]
 49%|████▉     | 3524/7175 [43:30<39:53,  1.53it/s]
 49%|████▉     | 3525/7175 [43:30<39:51,  1.53it/s]
 49%|████▉     | 3526/7175 [43:31<39:49,  1.53it/s]
 49%|████▉     | 3527/7175 [43:32<39:52,  1.52it/s]
 49%|████▉     | 3528/7175 [43:32<39:50,  1.53it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.37it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.74it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.54it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.05it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.73it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.57it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.44it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.38it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.31it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.31it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.27it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.26it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.25it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.24it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.24it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.24it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.23it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.23it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s]
                                                   
+
                                               
{'eval_loss': '3.542', 'eval_runtime': '3.262', 'eval_samples_per_second': '748.4', 'eval_steps_per_second': '11.96', 'epoch': '12.29'}
+
 49%|████▉     | 3528/7175 [43:35<39:50,  1.53it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s][eval] epoch 12.294 step 3528 loss 3.5418 (best 3.5332 @ ep 11.793, 2/20 since best)
+
+
                                               
 49%|████▉     | 3529/7175 [43:36<1:39:18,  1.63s/it]
 49%|████▉     | 3530/7175 [43:37<1:21:24,  1.34s/it]
                                                     
{'loss': '3.118', 'grad_norm': '0.2319', 'learning_rate': '0.0006', 'epoch': '12.3'}
+
 49%|████▉     | 3530/7175 [43:37<1:21:24,  1.34s/it]
 49%|████▉     | 3531/7175 [43:37<1:08:54,  1.13s/it]
 49%|████▉     | 3532/7175 [43:38<1:00:10,  1.01it/s]
 49%|████▉     | 3533/7175 [43:39<54:02,  1.12it/s]  
 49%|████▉     | 3534/7175 [43:39<49:43,  1.22it/s]
 49%|████▉     | 3535/7175 [43:40<46:42,  1.30it/s]
 49%|████▉     | 3536/7175 [43:41<44:38,  1.36it/s]
 49%|████▉     | 3537/7175 [43:41<43:09,  1.40it/s]
 49%|████▉     | 3538/7175 [43:42<42:06,  1.44it/s]
 49%|████▉     | 3539/7175 [43:43<41:20,  1.47it/s]
 49%|████▉     | 3540/7175 [43:43<40:52,  1.48it/s]
                                                   
{'loss': '3.113', 'grad_norm': '0.2512', 'learning_rate': '0.0006', 'epoch': '12.34'}
+
 49%|████▉     | 3540/7175 [43:43<40:52,  1.48it/s]
 49%|████▉     | 3541/7175 [43:44<40:31,  1.49it/s]
 49%|████▉     | 3542/7175 [43:45<40:14,  1.50it/s]
 49%|████▉     | 3543/7175 [43:45<40:02,  1.51it/s]
 49%|████▉     | 3544/7175 [43:46<39:57,  1.51it/s]
 49%|████▉     | 3545/7175 [43:47<39:50,  1.52it/s]
 49%|████▉     | 3546/7175 [43:47<39:44,  1.52it/s]
 49%|████▉     | 3547/7175 [43:48<39:40,  1.52it/s]
 49%|████▉     | 3548/7175 [43:49<39:41,  1.52it/s]
 49%|████▉     | 3549/7175 [43:49<39:38,  1.52it/s]
 49%|████▉     | 3550/7175 [43:50<39:36,  1.53it/s]
                                                   
{'loss': '3.126', 'grad_norm': '0.2069', 'learning_rate': '0.0006', 'epoch': '12.37'}
+
 49%|████▉     | 3550/7175 [43:50<39:36,  1.53it/s]
 49%|████▉     | 3551/7175 [43:51<39:35,  1.53it/s]
 50%|████▉     | 3552/7175 [43:51<39:34,  1.53it/s]
 50%|████▉     | 3553/7175 [43:52<39:34,  1.53it/s]
 50%|████▉     | 3554/7175 [43:52<39:32,  1.53it/s]
 50%|████▉     | 3555/7175 [43:53<39:31,  1.53it/s]
 50%|████▉     | 3556/7175 [43:54<39:31,  1.53it/s]
 50%|████▉     | 3557/7175 [43:54<39:30,  1.53it/s]
 50%|████▉     | 3558/7175 [43:55<39:29,  1.53it/s]
 50%|████▉     | 3559/7175 [43:56<39:27,  1.53it/s]
 50%|████▉     | 3560/7175 [43:56<39:28,  1.53it/s]
                                                   
{'loss': '3.137', 'grad_norm': '0.2168', 'learning_rate': '0.0006', 'epoch': '12.41'}
+
 50%|████▉     | 3560/7175 [43:56<39:28,  1.53it/s]
 50%|████▉     | 3561/7175 [43:57<39:31,  1.52it/s]
 50%|████▉     | 3562/7175 [43:58<39:29,  1.52it/s]
 50%|████▉     | 3563/7175 [43:58<39:28,  1.53it/s]
 50%|████▉     | 3564/7175 [43:59<39:26,  1.53it/s]
 50%|████▉     | 3565/7175 [44:00<39:27,  1.52it/s]
 50%|████▉     | 3566/7175 [44:00<39:25,  1.53it/s]
 50%|████▉     | 3567/7175 [44:01<39:25,  1.53it/s]
 50%|████▉     | 3568/7175 [44:02<39:24,  1.53it/s]
 50%|████▉     | 3569/7175 [44:02<39:23,  1.53it/s]
 50%|████▉     | 3570/7175 [44:03<39:23,  1.53it/s]
                                                   
{'loss': '3.14', 'grad_norm': '0.2077', 'learning_rate': '0.0006', 'epoch': '12.44'}
+
 50%|████▉     | 3570/7175 [44:03<39:23,  1.53it/s]
 50%|████▉     | 3571/7175 [44:04<39:23,  1.52it/s]
 50%|████▉     | 3572/7175 [44:04<39:24,  1.52it/s]
 50%|████▉     | 3573/7175 [44:05<39:23,  1.52it/s]
 50%|████▉     | 3574/7175 [44:06<39:21,  1.53it/s]
 50%|████▉     | 3575/7175 [44:06<39:22,  1.52it/s]
 50%|████▉     | 3576/7175 [44:07<39:20,  1.52it/s]
 50%|████▉     | 3577/7175 [44:08<39:19,  1.53it/s]
 50%|████▉     | 3578/7175 [44:08<39:17,  1.53it/s]
 50%|████▉     | 3579/7175 [44:09<39:17,  1.53it/s]
 50%|████▉     | 3580/7175 [44:10<39:15,  1.53it/s]
                                                   
{'loss': '3.125', 'grad_norm': '0.2209', 'learning_rate': '0.0006', 'epoch': '12.48'}
+
 50%|████▉     | 3580/7175 [44:10<39:15,  1.53it/s]
 50%|████▉     | 3581/7175 [44:10<39:16,  1.53it/s]
 50%|████▉     | 3582/7175 [44:11<39:15,  1.53it/s]
 50%|████▉     | 3583/7175 [44:11<39:14,  1.53it/s]
 50%|████▉     | 3584/7175 [44:12<39:15,  1.52it/s]
 50%|████▉     | 3585/7175 [44:13<39:14,  1.52it/s]
 50%|████▉     | 3586/7175 [44:13<39:15,  1.52it/s]
 50%|████▉     | 3587/7175 [44:14<39:13,  1.52it/s]
 50%|█████     | 3588/7175 [44:15<39:11,  1.53it/s]
 50%|█████     | 3589/7175 [44:15<39:12,  1.52it/s]
 50%|█████     | 3590/7175 [44:16<39:11,  1.52it/s]
                                                   
{'loss': '3.12', 'grad_norm': '0.2342', 'learning_rate': '0.0006', 'epoch': '12.51'}
+
 50%|█████     | 3590/7175 [44:16<39:11,  1.52it/s]
 50%|█████     | 3591/7175 [44:17<39:12,  1.52it/s]
 50%|█████     | 3592/7175 [44:17<39:11,  1.52it/s]
 50%|█████     | 3593/7175 [44:18<39:08,  1.53it/s]
 50%|█████     | 3594/7175 [44:19<39:10,  1.52it/s]
 50%|█████     | 3595/7175 [44:19<39:08,  1.52it/s]
 50%|█████     | 3596/7175 [44:20<39:08,  1.52it/s]
 50%|█████     | 3597/7175 [44:21<39:07,  1.52it/s]
 50%|█████     | 3598/7175 [44:21<39:06,  1.52it/s]
 50%|█████     | 3599/7175 [44:22<39:06,  1.52it/s]
 50%|█████     | 3600/7175 [44:23<39:05,  1.52it/s]
                                                   
{'loss': '3.134', 'grad_norm': '0.2396', 'learning_rate': '0.0006', 'epoch': '12.55'}
+
 50%|█████     | 3600/7175 [44:23<39:05,  1.52it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.45it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.72it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.57it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.04it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.73it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.57it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.45it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.38it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.34it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.27it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.30it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.28it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.26it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.27it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.25it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.24it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.24it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.23it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s]
                                                   
+
                                               
{'eval_loss': '3.537', 'eval_runtime': '3.248', 'eval_samples_per_second': '751.6', 'eval_steps_per_second': '12.01', 'epoch': '12.55'}
+
 50%|█████     | 3600/7175 [44:26<39:05,  1.52it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s][eval] epoch 12.545 step 3600 loss 3.5365 (best 3.5332 @ ep 11.793, 3/20 since best)
+
+
                                               
 50%|█████     | 3601/7175 [44:27<1:37:07,  1.63s/it]
 50%|█████     | 3602/7175 [44:27<1:19:42,  1.34s/it]
 50%|█████     | 3603/7175 [44:28<1:07:30,  1.13s/it]
 50%|█████     | 3604/7175 [44:29<58:56,  1.01it/s]  
 50%|█████     | 3605/7175 [44:30<58:59,  1.01it/s]
 50%|█████     | 3606/7175 [44:30<52:57,  1.12it/s]
 50%|█████     | 3607/7175 [44:31<55:18,  1.08it/s]
 50%|█████     | 3608/7175 [44:32<50:21,  1.18it/s]
 50%|█████     | 3609/7175 [44:32<46:54,  1.27it/s]
 50%|█████     | 3610/7175 [44:33<44:28,  1.34it/s]
                                                   
{'loss': '3.144', 'grad_norm': '0.2441', 'learning_rate': '0.0006', 'epoch': '12.58'}
+
 50%|█████     | 3610/7175 [44:33<44:28,  1.34it/s]
 50%|█████     | 3611/7175 [44:34<42:50,  1.39it/s]
 50%|█████     | 3612/7175 [44:34<41:38,  1.43it/s]
 50%|█████     | 3613/7175 [44:35<40:47,  1.46it/s]
 50%|█████     | 3614/7175 [44:36<40:10,  1.48it/s]
 50%|█████     | 3615/7175 [44:36<39:48,  1.49it/s]
 50%|█████     | 3616/7175 [44:37<39:30,  1.50it/s]
 50%|█████     | 3617/7175 [44:38<39:16,  1.51it/s]
 50%|█████     | 3618/7175 [44:38<39:07,  1.52it/s]
 50%|█████     | 3619/7175 [44:39<45:05,  1.31it/s]
 50%|█████     | 3620/7175 [44:40<43:13,  1.37it/s]
                                                   
{'loss': '3.126', 'grad_norm': '0.2321', 'learning_rate': '0.0006', 'epoch': '12.62'}
+
 50%|█████     | 3620/7175 [44:40<43:13,  1.37it/s]
 50%|█████     | 3621/7175 [44:41<41:54,  1.41it/s]
 50%|█████     | 3622/7175 [44:41<40:55,  1.45it/s]
 50%|█████     | 3623/7175 [44:42<40:18,  1.47it/s]
 51%|█████     | 3624/7175 [44:43<39:50,  1.49it/s]
 51%|█████     | 3625/7175 [44:43<39:32,  1.50it/s]
 51%|█████     | 3626/7175 [44:44<39:17,  1.51it/s]
 51%|█████     | 3627/7175 [44:45<39:06,  1.51it/s]
 51%|█████     | 3628/7175 [44:45<39:00,  1.52it/s]
 51%|█████     | 3629/7175 [44:46<38:55,  1.52it/s]
 51%|█████     | 3630/7175 [44:47<38:51,  1.52it/s]
                                                   
{'loss': '3.138', 'grad_norm': '0.205', 'learning_rate': '0.0006', 'epoch': '12.65'}
+
 51%|█████     | 3630/7175 [44:47<38:51,  1.52it/s]
 51%|█████     | 3631/7175 [44:47<38:49,  1.52it/s]
 51%|█████     | 3632/7175 [44:48<38:47,  1.52it/s]
 51%|█████     | 3633/7175 [44:49<38:46,  1.52it/s]
 51%|█████     | 3634/7175 [44:49<38:44,  1.52it/s]
 51%|█████     | 3635/7175 [44:50<38:43,  1.52it/s]
 51%|█████     | 3636/7175 [44:51<44:46,  1.32it/s]
 51%|█████     | 3637/7175 [44:52<42:52,  1.38it/s]
 51%|█████     | 3638/7175 [44:52<41:36,  1.42it/s]
 51%|█████     | 3639/7175 [44:53<40:41,  1.45it/s]
 51%|█████     | 3640/7175 [44:53<40:03,  1.47it/s]
                                                   
{'loss': '3.134', 'grad_norm': '0.2088', 'learning_rate': '0.0006', 'epoch': '12.69'}
+
 51%|█████     | 3640/7175 [44:53<40:03,  1.47it/s]
 51%|█████     | 3641/7175 [44:54<39:52,  1.48it/s]
 51%|█████     | 3642/7175 [44:55<39:28,  1.49it/s]
 51%|█████     | 3643/7175 [44:55<39:10,  1.50it/s]
 51%|█████     | 3644/7175 [44:56<38:58,  1.51it/s]
 51%|█████     | 3645/7175 [44:57<38:52,  1.51it/s]
 51%|█████     | 3646/7175 [44:57<38:45,  1.52it/s]
 51%|█████     | 3647/7175 [44:58<38:39,  1.52it/s]
 51%|█████     | 3648/7175 [44:59<38:35,  1.52it/s]
 51%|█████     | 3649/7175 [44:59<38:35,  1.52it/s]
 51%|█████     | 3650/7175 [45:00<44:34,  1.32it/s]
                                                   
{'loss': '3.146', 'grad_norm': '0.1918', 'learning_rate': '0.0006', 'epoch': '12.72'}
+
 51%|█████     | 3650/7175 [45:00<44:34,  1.32it/s]
 51%|█████     | 3651/7175 [45:01<42:45,  1.37it/s]
 51%|█████     | 3652/7175 [45:02<41:26,  1.42it/s]
 51%|█████     | 3653/7175 [45:02<40:31,  1.45it/s]
 51%|█████     | 3654/7175 [45:03<39:54,  1.47it/s]
 51%|█████     | 3655/7175 [45:04<39:27,  1.49it/s]
 51%|█████     | 3656/7175 [45:04<39:07,  1.50it/s]
 51%|█████     | 3657/7175 [45:05<38:54,  1.51it/s]
 51%|█████     | 3658/7175 [45:06<38:47,  1.51it/s]
 51%|█████     | 3659/7175 [45:06<38:38,  1.52it/s]
 51%|█████     | 3660/7175 [45:07<38:32,  1.52it/s]
                                                   
{'loss': '3.145', 'grad_norm': '0.2434', 'learning_rate': '0.0006', 'epoch': '12.75'}
+
 51%|█████     | 3660/7175 [45:07<38:32,  1.52it/s]
 51%|█████     | 3661/7175 [45:08<38:28,  1.52it/s]
 51%|█████     | 3662/7175 [45:08<38:28,  1.52it/s]
 51%|█████     | 3663/7175 [45:09<38:25,  1.52it/s]
 51%|█████     | 3664/7175 [45:10<38:22,  1.52it/s]
 51%|█████     | 3665/7175 [45:10<38:19,  1.53it/s]
 51%|█████     | 3666/7175 [45:11<38:21,  1.52it/s]
 51%|█████     | 3667/7175 [45:12<38:19,  1.53it/s]
 51%|█████     | 3668/7175 [45:13<44:07,  1.32it/s]
 51%|█████     | 3669/7175 [45:13<42:20,  1.38it/s]
 51%|█████     | 3670/7175 [45:14<41:05,  1.42it/s]
                                                   
{'loss': '3.142', 'grad_norm': '0.208', 'learning_rate': '0.0006', 'epoch': '12.79'}
+
 51%|█████     | 3670/7175 [45:14<41:05,  1.42it/s]
 51%|█████     | 3671/7175 [45:14<40:12,  1.45it/s]
 51%|█████     | 3672/7175 [45:15<45:33,  1.28it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.34it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.76it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.57it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.05it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.72it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.55it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.42it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.37it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.31it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.29it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.26it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.26it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.25it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.25it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.23it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.24it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.24it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.25it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.25it/s]
                                                   
+
                                               
{'eval_loss': '3.526', 'eval_runtime': '3.255', 'eval_samples_per_second': '749.9', 'eval_steps_per_second': '11.98', 'epoch': '12.8'}
+
 51%|█████     | 3672/7175 [45:19<45:33,  1.28it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.25it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-3672
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-3672/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-3672/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.66it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.66it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-3672/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-3672/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-3672/tokenizer_config.json
+[eval] epoch 12.797 step 3672 loss 3.5257 (best 3.5257 @ ep 12.797, 0/20 since best)
+
+
                                               
 51%|█████     | 3673/7175 [45:22<2:17:53,  2.36s/it]
 51%|█████     | 3674/7175 [45:22<1:47:56,  1.85s/it]
 51%|█████     | 3675/7175 [45:23<1:27:00,  1.49s/it]
 51%|█████     | 3676/7175 [45:23<1:12:23,  1.24s/it]
 51%|█████     | 3677/7175 [45:24<1:02:07,  1.07s/it]
 51%|█████▏    | 3678/7175 [45:25<54:55,  1.06it/s]  
 51%|█████▏    | 3679/7175 [45:25<49:52,  1.17it/s]
 51%|█████▏    | 3680/7175 [45:26<46:22,  1.26it/s]
                                                   
{'loss': '3.149', 'grad_norm': '0.2461', 'learning_rate': '0.0006', 'epoch': '12.82'}
+
 51%|█████▏    | 3680/7175 [45:26<46:22,  1.26it/s]
 51%|█████▏    | 3681/7175 [45:27<43:57,  1.32it/s]
 51%|█████▏    | 3682/7175 [45:27<42:12,  1.38it/s]
 51%|█████▏    | 3683/7175 [45:28<40:58,  1.42it/s]
 51%|█████▏    | 3684/7175 [45:29<40:06,  1.45it/s]
 51%|█████▏    | 3685/7175 [45:29<39:31,  1.47it/s]
 51%|█████▏    | 3686/7175 [45:30<39:07,  1.49it/s]
 51%|█████▏    | 3687/7175 [45:31<38:50,  1.50it/s]
 51%|█████▏    | 3688/7175 [45:31<38:37,  1.50it/s]
 51%|█████▏    | 3689/7175 [45:32<38:26,  1.51it/s]
 51%|█████▏    | 3690/7175 [45:33<38:19,  1.52it/s]
                                                   
{'loss': '3.135', 'grad_norm': '0.2264', 'learning_rate': '0.0006', 'epoch': '12.86'}
+
 51%|█████▏    | 3690/7175 [45:33<38:19,  1.52it/s]
 51%|█████▏    | 3691/7175 [45:33<38:19,  1.52it/s]
 51%|█████▏    | 3692/7175 [45:34<38:16,  1.52it/s]
 51%|█████▏    | 3693/7175 [45:35<38:12,  1.52it/s]
 51%|█████▏    | 3694/7175 [45:35<38:10,  1.52it/s]
 51%|█████▏    | 3695/7175 [45:36<38:07,  1.52it/s]
 52%|█████▏    | 3696/7175 [45:37<38:04,  1.52it/s]
 52%|█████▏    | 3697/7175 [45:37<38:05,  1.52it/s]
 52%|█████▏    | 3698/7175 [45:38<38:04,  1.52it/s]
 52%|█████▏    | 3699/7175 [45:39<38:02,  1.52it/s]
 52%|█████▏    | 3700/7175 [45:39<38:02,  1.52it/s]
                                                   
{'loss': '3.146', 'grad_norm': '0.2369', 'learning_rate': '0.0006', 'epoch': '12.89'}
+
 52%|█████▏    | 3700/7175 [45:39<38:02,  1.52it/s]
 52%|█████▏    | 3701/7175 [45:40<38:02,  1.52it/s]
 52%|█████▏    | 3702/7175 [45:41<37:59,  1.52it/s]
 52%|█████▏    | 3703/7175 [45:41<37:59,  1.52it/s]
 52%|█████▏    | 3704/7175 [45:42<37:58,  1.52it/s]
 52%|█████▏    | 3705/7175 [45:43<37:55,  1.52it/s]
 52%|█████▏    | 3706/7175 [45:43<37:54,  1.53it/s]
 52%|█████▏    | 3707/7175 [45:44<37:52,  1.53it/s]
 52%|█████▏    | 3708/7175 [45:44<37:53,  1.52it/s]
 52%|█████▏    | 3709/7175 [45:45<37:52,  1.53it/s]
 52%|█████▏    | 3710/7175 [45:46<37:50,  1.53it/s]
                                                   
{'loss': '3.136', 'grad_norm': '0.2184', 'learning_rate': '0.0006', 'epoch': '12.93'}
+
 52%|█████▏    | 3710/7175 [45:46<37:50,  1.53it/s]
 52%|█████▏    | 3711/7175 [45:46<37:50,  1.53it/s]
 52%|█████▏    | 3712/7175 [45:47<37:50,  1.52it/s]
 52%|█████▏    | 3713/7175 [45:48<37:48,  1.53it/s]
 52%|█████▏    | 3714/7175 [45:48<37:47,  1.53it/s]
 52%|█████▏    | 3715/7175 [45:49<37:46,  1.53it/s]
 52%|█████▏    | 3716/7175 [45:50<37:47,  1.53it/s]
 52%|█████▏    | 3717/7175 [45:50<37:48,  1.52it/s]
 52%|█████▏    | 3718/7175 [45:51<37:47,  1.52it/s]
 52%|█████▏    | 3719/7175 [45:52<37:46,  1.52it/s]
 52%|█████▏    | 3720/7175 [45:52<37:44,  1.53it/s]
                                                   
{'loss': '3.138', 'grad_norm': '0.2171', 'learning_rate': '0.0006', 'epoch': '12.96'}
+
 52%|█████▏    | 3720/7175 [45:52<37:44,  1.53it/s]
 52%|█████▏    | 3721/7175 [45:53<37:45,  1.52it/s]
 52%|█████▏    | 3722/7175 [45:54<37:45,  1.52it/s]
 52%|█████▏    | 3723/7175 [45:54<37:44,  1.52it/s]
 52%|█████▏    | 3724/7175 [45:55<37:42,  1.53it/s]
 52%|█████▏    | 3725/7175 [45:56<37:40,  1.53it/s]
 52%|█████▏    | 3726/7175 [45:56<37:42,  1.52it/s]
 52%|█████▏    | 3727/7175 [45:57<37:41,  1.52it/s]
 52%|█████▏    | 3728/7175 [45:58<37:50,  1.52it/s]
 52%|█████▏    | 3729/7175 [45:58<37:46,  1.52it/s]
 52%|█████▏    | 3730/7175 [45:59<37:42,  1.52it/s]
                                                   
{'loss': '3.137', 'grad_norm': '0.2157', 'learning_rate': '0.0006', 'epoch': '13'}
+
 52%|█████▏    | 3730/7175 [45:59<37:42,  1.52it/s]
 52%|█████▏    | 3732/7175 [46:00<31:24,  1.83it/s]
 52%|█████▏    | 3733/7175 [46:00<32:55,  1.74it/s]
 52%|█████▏    | 3734/7175 [46:01<34:08,  1.68it/s]
 52%|█████▏    | 3735/7175 [46:02<35:04,  1.63it/s]
 52%|█████▏    | 3736/7175 [46:02<35:46,  1.60it/s]
 52%|█████▏    | 3737/7175 [46:03<36:17,  1.58it/s]
 52%|█████▏    | 3738/7175 [46:04<36:37,  1.56it/s]
 52%|█████▏    | 3739/7175 [46:04<36:54,  1.55it/s]
 52%|█████▏    | 3740/7175 [46:05<37:07,  1.54it/s]
                                                   
{'loss': '3.098', 'grad_norm': '0.3087', 'learning_rate': '0.0006', 'epoch': '13.03'}
+
 52%|█████▏    | 3740/7175 [46:05<37:07,  1.54it/s]
 52%|█████▏    | 3741/7175 [46:06<37:15,  1.54it/s]
 52%|█████▏    | 3742/7175 [46:06<37:20,  1.53it/s]
 52%|█████▏    | 3743/7175 [46:07<37:22,  1.53it/s]
 52%|█████▏    | 3744/7175 [46:08<37:24,  1.53it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.94it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.93it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.67it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.10it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.76it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.59it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.46it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.39it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.33it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.29it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.29it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.26it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.24it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.25it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.23it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.23it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.23it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.22it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.21it/s]
                                                   
+
                                               
{'eval_loss': '3.549', 'eval_runtime': '3.259', 'eval_samples_per_second': '748.9', 'eval_steps_per_second': '11.96', 'epoch': '13.05'}
+
 52%|█████▏    | 3744/7175 [46:11<37:24,  1.53it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.21it/s][eval] epoch 13.045 step 3744 loss 3.5487 (best 3.5257 @ ep 12.797, 1/20 since best)
+
+
                                               
 52%|█████▏    | 3745/7175 [46:12<1:33:11,  1.63s/it]
 52%|█████▏    | 3746/7175 [46:12<1:16:28,  1.34s/it]
 52%|█████▏    | 3747/7175 [46:13<1:04:45,  1.13s/it]
 52%|█████▏    | 3748/7175 [46:14<56:32,  1.01it/s]  
 52%|█████▏    | 3749/7175 [46:14<50:48,  1.12it/s]
 52%|█████▏    | 3750/7175 [46:15<46:46,  1.22it/s]
                                                   
{'loss': '3.087', 'grad_norm': '0.2623', 'learning_rate': '0.0006', 'epoch': '13.07'}
+
 52%|█████▏    | 3750/7175 [46:15<46:46,  1.22it/s]
 52%|█████▏    | 3751/7175 [46:15<43:59,  1.30it/s]
 52%|█████▏    | 3752/7175 [46:16<41:59,  1.36it/s]
 52%|█████▏    | 3753/7175 [46:17<40:34,  1.41it/s]
 52%|█████▏    | 3754/7175 [46:17<39:39,  1.44it/s]
 52%|█████▏    | 3755/7175 [46:18<38:57,  1.46it/s]
 52%|█████▏    | 3756/7175 [46:19<38:27,  1.48it/s]
 52%|█████▏    | 3757/7175 [46:19<38:05,  1.50it/s]
 52%|█████▏    | 3758/7175 [46:20<37:52,  1.50it/s]
 52%|█████▏    | 3759/7175 [46:21<37:41,  1.51it/s]
 52%|█████▏    | 3760/7175 [46:21<37:34,  1.51it/s]
                                                   
{'loss': '3.085', 'grad_norm': '0.2384', 'learning_rate': '0.0006', 'epoch': '13.1'}
+
 52%|█████▏    | 3760/7175 [46:21<37:34,  1.51it/s]
 52%|█████▏    | 3761/7175 [46:22<37:30,  1.52it/s]
 52%|█████▏    | 3762/7175 [46:23<37:26,  1.52it/s]
 52%|█████▏    | 3763/7175 [46:23<37:23,  1.52it/s]
 52%|█████▏    | 3764/7175 [46:24<37:20,  1.52it/s]
 52%|█████▏    | 3765/7175 [46:25<37:19,  1.52it/s]
 52%|█████▏    | 3766/7175 [46:25<37:16,  1.52it/s]
 53%|█████▎    | 3767/7175 [46:26<37:14,  1.52it/s]
 53%|█████▎    | 3768/7175 [46:27<37:14,  1.52it/s]
 53%|█████▎    | 3769/7175 [46:27<37:14,  1.52it/s]
 53%|█████▎    | 3770/7175 [46:28<37:13,  1.52it/s]
                                                   
{'loss': '3.075', 'grad_norm': '0.2121', 'learning_rate': '0.0006', 'epoch': '13.14'}
+
 53%|█████▎    | 3770/7175 [46:28<37:13,  1.52it/s]
 53%|█████▎    | 3771/7175 [46:29<37:13,  1.52it/s]
 53%|█████▎    | 3772/7175 [46:29<37:11,  1.52it/s]
 53%|█████▎    | 3773/7175 [46:30<37:12,  1.52it/s]
 53%|█████▎    | 3774/7175 [46:31<37:10,  1.52it/s]
 53%|█████▎    | 3775/7175 [46:31<37:10,  1.52it/s]
 53%|█████▎    | 3776/7175 [46:32<37:08,  1.53it/s]
 53%|█████▎    | 3777/7175 [46:33<37:07,  1.53it/s]
 53%|█████▎    | 3778/7175 [46:33<37:06,  1.53it/s]
 53%|█████▎    | 3779/7175 [46:34<37:07,  1.52it/s]
 53%|█████▎    | 3780/7175 [46:35<37:05,  1.53it/s]
                                                   
{'loss': '3.07', 'grad_norm': '0.2243', 'learning_rate': '0.0006', 'epoch': '13.17'}
+
 53%|█████▎    | 3780/7175 [46:35<37:05,  1.53it/s]
 53%|█████▎    | 3781/7175 [46:35<37:05,  1.53it/s]
 53%|█████▎    | 3782/7175 [46:36<37:05,  1.52it/s]
 53%|█████▎    | 3783/7175 [46:36<37:04,  1.53it/s]
 53%|█████▎    | 3784/7175 [46:37<37:03,  1.52it/s]
 53%|█████▎    | 3785/7175 [46:38<37:01,  1.53it/s]
 53%|█████▎    | 3786/7175 [46:38<37:02,  1.53it/s]
 53%|█████▎    | 3787/7175 [46:39<37:02,  1.52it/s]
 53%|█████▎    | 3788/7175 [46:40<37:01,  1.52it/s]
 53%|█████▎    | 3789/7175 [46:40<37:00,  1.53it/s]
 53%|█████▎    | 3790/7175 [46:41<36:58,  1.53it/s]
                                                   
{'loss': '3.077', 'grad_norm': '0.232', 'learning_rate': '0.0006', 'epoch': '13.21'}
+
 53%|█████▎    | 3790/7175 [46:41<36:58,  1.53it/s]
 53%|█████▎    | 3791/7175 [46:42<36:59,  1.52it/s]
 53%|█████▎    | 3792/7175 [46:42<36:58,  1.52it/s]
 53%|█████▎    | 3793/7175 [46:43<36:58,  1.52it/s]
 53%|█████▎    | 3794/7175 [46:44<36:57,  1.52it/s]
 53%|█████▎    | 3795/7175 [46:44<36:54,  1.53it/s]
 53%|█████▎    | 3796/7175 [46:45<36:56,  1.52it/s]
 53%|█████▎    | 3797/7175 [46:46<36:55,  1.52it/s]
 53%|█████▎    | 3798/7175 [46:46<36:54,  1.52it/s]
 53%|█████▎    | 3799/7175 [46:47<36:54,  1.52it/s]
 53%|█████▎    | 3800/7175 [46:48<36:52,  1.53it/s]
                                                   
{'loss': '3.108', 'grad_norm': '0.2315', 'learning_rate': '0.0006', 'epoch': '13.24'}
+
 53%|█████▎    | 3800/7175 [46:48<36:52,  1.53it/s]
 53%|█████▎    | 3801/7175 [46:48<36:54,  1.52it/s]
 53%|█████▎    | 3802/7175 [46:49<36:52,  1.52it/s]
 53%|█████▎    | 3803/7175 [46:50<36:52,  1.52it/s]
 53%|█████▎    | 3804/7175 [46:50<36:50,  1.53it/s]
 53%|█████▎    | 3805/7175 [46:51<36:49,  1.53it/s]
 53%|█████▎    | 3806/7175 [46:52<36:50,  1.52it/s]
 53%|█████▎    | 3807/7175 [46:52<36:49,  1.52it/s]
 53%|█████▎    | 3808/7175 [46:53<36:48,  1.52it/s]
 53%|█████▎    | 3809/7175 [46:54<36:45,  1.53it/s]
 53%|█████▎    | 3810/7175 [46:54<36:46,  1.53it/s]
                                                   
{'loss': '3.086', 'grad_norm': '0.2315', 'learning_rate': '0.0006', 'epoch': '13.28'}
+
 53%|█████▎    | 3810/7175 [46:54<36:46,  1.53it/s]
 53%|█████▎    | 3811/7175 [46:55<36:48,  1.52it/s]
 53%|█████▎    | 3812/7175 [46:55<36:46,  1.52it/s]
 53%|█████▎    | 3813/7175 [46:56<36:46,  1.52it/s]
 53%|█████▎    | 3814/7175 [46:57<36:44,  1.52it/s]
 53%|█████▎    | 3815/7175 [46:57<36:43,  1.52it/s]
 53%|█████▎    | 3816/7175 [46:58<36:44,  1.52it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.80it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.87it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.64it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.08it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.78it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.59it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.46it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.39it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.33it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.30it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.29it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.27it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.26it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.25it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.24it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.24it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.24it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.23it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s]
                                                   
+
                                               
{'eval_loss': '3.532', 'eval_runtime': '3.26', 'eval_samples_per_second': '748.8', 'eval_steps_per_second': '11.96', 'epoch': '13.3'}
+
 53%|█████▎    | 3816/7175 [47:01<36:44,  1.52it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s][eval] epoch 13.297 step 3816 loss 3.5316 (best 3.5257 @ ep 12.797, 2/20 since best)
+
+
                                               
 53%|█████▎    | 3817/7175 [47:02<1:31:27,  1.63s/it]
 53%|█████▎    | 3818/7175 [47:03<1:15:00,  1.34s/it]
 53%|█████▎    | 3819/7175 [47:03<1:03:28,  1.13s/it]
 53%|█████▎    | 3820/7175 [47:04<55:25,  1.01it/s]  
                                                   
{'loss': '3.089', 'grad_norm': '0.2318', 'learning_rate': '0.0006', 'epoch': '13.31'}
+
 53%|█████▎    | 3820/7175 [47:04<55:25,  1.01it/s]
 53%|█████▎    | 3821/7175 [47:05<49:48,  1.12it/s]
 53%|█████▎    | 3822/7175 [47:05<45:50,  1.22it/s]
 53%|█████▎    | 3823/7175 [47:06<43:04,  1.30it/s]
 53%|█████▎    | 3824/7175 [47:07<41:07,  1.36it/s]
 53%|█████▎    | 3825/7175 [47:07<39:45,  1.40it/s]
 53%|█████▎    | 3826/7175 [47:08<38:49,  1.44it/s]
 53%|█████▎    | 3827/7175 [47:09<38:09,  1.46it/s]
 53%|█████▎    | 3828/7175 [47:09<37:40,  1.48it/s]
 53%|█████▎    | 3829/7175 [47:10<37:18,  1.49it/s]
 53%|█████▎    | 3830/7175 [47:11<37:04,  1.50it/s]
                                                   
{'loss': '3.108', 'grad_norm': '0.24', 'learning_rate': '0.0006', 'epoch': '13.35'}
+
 53%|█████▎    | 3830/7175 [47:11<37:04,  1.50it/s]
 53%|█████▎    | 3831/7175 [47:11<36:57,  1.51it/s]
 53%|█████▎    | 3832/7175 [47:12<36:49,  1.51it/s]
 53%|█████▎    | 3833/7175 [47:13<36:44,  1.52it/s]
 53%|█████▎    | 3834/7175 [47:13<36:38,  1.52it/s]
 53%|█████▎    | 3835/7175 [47:14<36:34,  1.52it/s]
 53%|█████▎    | 3836/7175 [47:14<36:34,  1.52it/s]
 53%|█████▎    | 3837/7175 [47:15<36:32,  1.52it/s]
 53%|█████▎    | 3838/7175 [47:16<36:30,  1.52it/s]
 54%|█████▎    | 3839/7175 [47:16<36:29,  1.52it/s]
 54%|█████▎    | 3840/7175 [47:17<36:28,  1.52it/s]
                                                   
{'loss': '3.101', 'grad_norm': '0.2473', 'learning_rate': '0.0006', 'epoch': '13.38'}
+
 54%|█████▎    | 3840/7175 [47:17<36:28,  1.52it/s]
 54%|█████▎    | 3841/7175 [47:18<36:29,  1.52it/s]
 54%|█████▎    | 3842/7175 [47:18<36:27,  1.52it/s]
 54%|█████▎    | 3843/7175 [47:19<36:26,  1.52it/s]
 54%|█████▎    | 3844/7175 [47:20<36:24,  1.52it/s]
 54%|█████▎    | 3845/7175 [47:20<36:23,  1.53it/s]
 54%|█████▎    | 3846/7175 [47:21<36:24,  1.52it/s]
 54%|█████▎    | 3847/7175 [47:22<36:22,  1.52it/s]
 54%|█████▎    | 3848/7175 [47:22<36:22,  1.52it/s]
 54%|█████▎    | 3849/7175 [47:23<36:20,  1.53it/s]
 54%|█████▎    | 3850/7175 [47:24<36:20,  1.53it/s]
                                                   
{'loss': '3.109', 'grad_norm': '0.2138', 'learning_rate': '0.0006', 'epoch': '13.42'}
+
 54%|█████▎    | 3850/7175 [47:24<36:20,  1.53it/s]
 54%|█████▎    | 3851/7175 [47:24<36:22,  1.52it/s]
 54%|█████▎    | 3852/7175 [47:25<36:20,  1.52it/s]
 54%|█████▎    | 3853/7175 [47:26<36:19,  1.52it/s]
 54%|█████▎    | 3854/7175 [47:26<36:18,  1.52it/s]
 54%|█████▎    | 3855/7175 [47:27<36:16,  1.53it/s]
 54%|█████▎    | 3856/7175 [47:28<36:17,  1.52it/s]
 54%|█████▍    | 3857/7175 [47:28<36:16,  1.52it/s]
 54%|█████▍    | 3858/7175 [47:29<36:16,  1.52it/s]
 54%|█████▍    | 3859/7175 [47:30<36:14,  1.53it/s]
 54%|█████▍    | 3860/7175 [47:30<36:13,  1.53it/s]
                                                   
{'loss': '3.111', 'grad_norm': '0.2233', 'learning_rate': '0.0006', 'epoch': '13.45'}
+
 54%|█████▍    | 3860/7175 [47:30<36:13,  1.53it/s]
 54%|█████▍    | 3861/7175 [47:31<36:14,  1.52it/s]
 54%|█████▍    | 3862/7175 [47:32<36:13,  1.52it/s]
 54%|█████▍    | 3863/7175 [47:32<36:13,  1.52it/s]
 54%|█████▍    | 3864/7175 [47:33<36:11,  1.52it/s]
 54%|█████▍    | 3865/7175 [47:34<36:10,  1.52it/s]
 54%|█████▍    | 3866/7175 [47:34<36:10,  1.52it/s]
 54%|█████▍    | 3867/7175 [47:35<36:09,  1.52it/s]
 54%|█████▍    | 3868/7175 [47:35<36:08,  1.53it/s]
 54%|█████▍    | 3869/7175 [47:36<36:06,  1.53it/s]
 54%|█████▍    | 3870/7175 [47:37<36:07,  1.52it/s]
                                                   
{'loss': '3.106', 'grad_norm': '0.2208', 'learning_rate': '0.0006', 'epoch': '13.49'}
+
 54%|█████▍    | 3870/7175 [47:37<36:07,  1.52it/s]
 54%|█████▍    | 3871/7175 [47:37<36:08,  1.52it/s]
 54%|█████▍    | 3872/7175 [47:38<36:07,  1.52it/s]
 54%|█████▍    | 3873/7175 [47:39<36:06,  1.52it/s]
 54%|█████▍    | 3874/7175 [47:39<36:03,  1.53it/s]
 54%|█████▍    | 3875/7175 [47:40<36:04,  1.52it/s]
 54%|█████▍    | 3876/7175 [47:41<36:02,  1.53it/s]
 54%|█████▍    | 3877/7175 [47:41<36:02,  1.53it/s]
 54%|█████▍    | 3878/7175 [47:42<36:01,  1.53it/s]
 54%|█████▍    | 3879/7175 [47:43<36:00,  1.53it/s]
 54%|█████▍    | 3880/7175 [47:43<35:59,  1.53it/s]
                                                   
{'loss': '3.114', 'grad_norm': '0.2215', 'learning_rate': '0.0006', 'epoch': '13.52'}
+
 54%|█████▍    | 3880/7175 [47:43<35:59,  1.53it/s]
 54%|█████▍    | 3881/7175 [47:44<36:00,  1.52it/s]
 54%|█████▍    | 3882/7175 [47:45<35:59,  1.53it/s]
 54%|█████▍    | 3883/7175 [47:45<35:57,  1.53it/s]
 54%|█████▍    | 3884/7175 [47:46<35:58,  1.52it/s]
 54%|█████▍    | 3885/7175 [47:47<35:56,  1.53it/s]
 54%|█████▍    | 3886/7175 [47:47<35:57,  1.52it/s]
 54%|█████▍    | 3887/7175 [47:48<35:56,  1.52it/s]
 54%|█████▍    | 3888/7175 [47:49<35:54,  1.53it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.25it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.71it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.55it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.04it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.75it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.57it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.45it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.38it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.32it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.30it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.30it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.27it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.26it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.24it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.25it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.24it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.23it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.23it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s]
                                                   
+
                                               
{'eval_loss': '3.524', 'eval_runtime': '3.245', 'eval_samples_per_second': '752.3', 'eval_steps_per_second': '12.02', 'epoch': '13.55'}
+
 54%|█████▍    | 3888/7175 [47:52<35:54,  1.53it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-3888
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-3888/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-3888/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.84it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.84it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-3888/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-3888/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-3888/tokenizer_config.json
+[eval] epoch 13.549 step 3888 loss 3.5244 (best 3.5244 @ ep 13.549, 0/20 since best)
+
+
                                               
 54%|█████▍    | 3889/7175 [47:55<2:03:18,  2.25s/it]
 54%|█████▍    | 3890/7175 [47:55<1:37:02,  1.77s/it]
                                                     
{'loss': '3.111', 'grad_norm': '0.2043', 'learning_rate': '0.0006', 'epoch': '13.56'}
+
 54%|█████▍    | 3890/7175 [47:55<1:37:02,  1.77s/it]
 54%|█████▍    | 3891/7175 [47:56<1:18:39,  1.44s/it]
 54%|█████▍    | 3892/7175 [47:57<1:05:46,  1.20s/it]
 54%|█████▍    | 3893/7175 [47:57<56:45,  1.04s/it]  
 54%|█████▍    | 3894/7175 [47:58<50:27,  1.08it/s]
 54%|█████▍    | 3895/7175 [47:59<46:02,  1.19it/s]
 54%|█████▍    | 3896/7175 [47:59<42:59,  1.27it/s]
 54%|█████▍    | 3897/7175 [48:00<40:48,  1.34it/s]
 54%|█████▍    | 3898/7175 [48:00<39:18,  1.39it/s]
 54%|█████▍    | 3899/7175 [48:01<38:13,  1.43it/s]
 54%|█████▍    | 3900/7175 [48:02<37:29,  1.46it/s]
                                                   
{'loss': '3.097', 'grad_norm': '0.2251', 'learning_rate': '0.0006', 'epoch': '13.59'}
+
 54%|█████▍    | 3900/7175 [48:02<37:29,  1.46it/s]
 54%|█████▍    | 3901/7175 [48:02<37:01,  1.47it/s]
 54%|█████▍    | 3902/7175 [48:03<36:37,  1.49it/s]
 54%|█████▍    | 3903/7175 [48:04<36:21,  1.50it/s]
 54%|█████▍    | 3904/7175 [48:04<36:09,  1.51it/s]
 54%|█████▍    | 3905/7175 [48:05<36:01,  1.51it/s]
 54%|█████▍    | 3906/7175 [48:06<35:57,  1.52it/s]
 54%|█████▍    | 3907/7175 [48:06<35:51,  1.52it/s]
 54%|█████▍    | 3908/7175 [48:07<35:48,  1.52it/s]
 54%|█████▍    | 3909/7175 [48:08<35:44,  1.52it/s]
 54%|█████▍    | 3910/7175 [48:08<35:43,  1.52it/s]
                                                   
{'loss': '3.116', 'grad_norm': '0.2078', 'learning_rate': '0.0006', 'epoch': '13.63'}
+
 54%|█████▍    | 3910/7175 [48:08<35:43,  1.52it/s]
 55%|█████▍    | 3911/7175 [48:09<35:44,  1.52it/s]
 55%|█████▍    | 3912/7175 [48:10<35:41,  1.52it/s]
 55%|█████▍    | 3913/7175 [48:10<35:40,  1.52it/s]
 55%|█████▍    | 3914/7175 [48:11<35:38,  1.53it/s]
 55%|█████▍    | 3915/7175 [48:12<35:38,  1.52it/s]
 55%|█████▍    | 3916/7175 [48:12<35:36,  1.53it/s]
 55%|█████▍    | 3917/7175 [48:13<35:36,  1.53it/s]
 55%|█████▍    | 3918/7175 [48:14<35:34,  1.53it/s]
 55%|█████▍    | 3919/7175 [48:14<35:33,  1.53it/s]
 55%|█████▍    | 3920/7175 [48:15<35:33,  1.53it/s]
                                                   
{'loss': '3.123', 'grad_norm': '0.2272', 'learning_rate': '0.0006', 'epoch': '13.66'}
+
 55%|█████▍    | 3920/7175 [48:15<35:33,  1.53it/s]
 55%|█████▍    | 3921/7175 [48:16<35:32,  1.53it/s]
 55%|█████▍    | 3922/7175 [48:16<35:32,  1.53it/s]
 55%|█████▍    | 3923/7175 [48:17<35:30,  1.53it/s]
 55%|█████▍    | 3924/7175 [48:18<35:31,  1.53it/s]
 55%|█████▍    | 3925/7175 [48:18<35:29,  1.53it/s]
 55%|█████▍    | 3926/7175 [48:19<35:28,  1.53it/s]
 55%|█████▍    | 3927/7175 [48:19<35:27,  1.53it/s]
 55%|█████▍    | 3928/7175 [48:20<35:28,  1.53it/s]
 55%|█████▍    | 3929/7175 [48:21<35:26,  1.53it/s]
 55%|█████▍    | 3930/7175 [48:21<35:25,  1.53it/s]
                                                   
{'loss': '3.116', 'grad_norm': '0.2331', 'learning_rate': '0.0006', 'epoch': '13.7'}
+
 55%|█████▍    | 3930/7175 [48:21<35:25,  1.53it/s]
 55%|█████▍    | 3931/7175 [48:22<35:24,  1.53it/s]
 55%|█████▍    | 3932/7175 [48:23<35:25,  1.53it/s]
 55%|█████▍    | 3933/7175 [48:23<35:24,  1.53it/s]
 55%|█████▍    | 3934/7175 [48:24<35:23,  1.53it/s]
 55%|█████▍    | 3935/7175 [48:25<35:21,  1.53it/s]
 55%|█████▍    | 3936/7175 [48:25<35:22,  1.53it/s]
 55%|█████▍    | 3937/7175 [48:26<35:20,  1.53it/s]
 55%|█████▍    | 3938/7175 [48:27<35:19,  1.53it/s]
 55%|█████▍    | 3939/7175 [48:27<35:18,  1.53it/s]
 55%|█████▍    | 3940/7175 [48:28<35:19,  1.53it/s]
                                                   
{'loss': '3.134', 'grad_norm': '0.2233', 'learning_rate': '0.0006', 'epoch': '13.73'}
+
 55%|█████▍    | 3940/7175 [48:28<35:19,  1.53it/s]
 55%|█████▍    | 3941/7175 [48:29<35:19,  1.53it/s]
 55%|█████▍    | 3942/7175 [48:29<35:17,  1.53it/s]
 55%|█████▍    | 3943/7175 [48:30<35:16,  1.53it/s]
 55%|█████▍    | 3944/7175 [48:31<35:17,  1.53it/s]
 55%|█████▍    | 3945/7175 [48:31<35:15,  1.53it/s]
 55%|█████▍    | 3946/7175 [48:32<35:14,  1.53it/s]
 55%|█████▌    | 3947/7175 [48:33<35:13,  1.53it/s]
 55%|█████▌    | 3948/7175 [48:33<35:15,  1.53it/s]
 55%|█████▌    | 3949/7175 [48:34<35:13,  1.53it/s]
 55%|█████▌    | 3950/7175 [48:35<35:12,  1.53it/s]
                                                   
{'loss': '3.118', 'grad_norm': '0.234', 'learning_rate': '0.0006', 'epoch': '13.77'}
+
 55%|█████▌    | 3950/7175 [48:35<35:12,  1.53it/s]
 55%|█████▌    | 3951/7175 [48:35<35:10,  1.53it/s]
 55%|█████▌    | 3952/7175 [48:36<35:11,  1.53it/s]
 55%|█████▌    | 3953/7175 [48:37<35:09,  1.53it/s]
 55%|█████▌    | 3954/7175 [48:37<35:08,  1.53it/s]
 55%|█████▌    | 3955/7175 [48:38<35:06,  1.53it/s]
 55%|█████▌    | 3956/7175 [48:38<35:08,  1.53it/s]
 55%|█████▌    | 3957/7175 [48:39<35:07,  1.53it/s]
 55%|█████▌    | 3958/7175 [48:40<35:06,  1.53it/s]
 55%|█████▌    | 3959/7175 [48:40<35:04,  1.53it/s]
 55%|█████▌    | 3960/7175 [48:41<35:04,  1.53it/s]
                                                   
{'loss': '3.125', 'grad_norm': '0.2168', 'learning_rate': '0.0006', 'epoch': '13.8'}
+
 55%|█████▌    | 3960/7175 [48:41<35:04,  1.53it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:02, 17.97it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.60it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.51it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.01it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.69it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.53it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.42it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.35it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.29it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.27it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.24it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.23it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.24it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.23it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.23it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.23it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.22it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.24it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s]
                                                   
+
                                               
{'eval_loss': '3.521', 'eval_runtime': '3.251', 'eval_samples_per_second': '750.9', 'eval_steps_per_second': '12', 'epoch': '13.8'}
+
 55%|█████▌    | 3960/7175 [48:44<35:04,  1.53it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-3960
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-3960/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-3960/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.86it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.86it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-3960/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-3960/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-3960/tokenizer_config.json
+[eval] epoch 13.8 step 3960 loss 3.5214 (best 3.5214 @ ep 13.8, 0/20 since best)
+
+
                                               
 55%|█████▌    | 3961/7175 [48:47<2:00:55,  2.26s/it]
 55%|█████▌    | 3962/7175 [48:48<1:35:07,  1.78s/it]
 55%|█████▌    | 3963/7175 [48:48<1:17:03,  1.44s/it]
 55%|█████▌    | 3964/7175 [48:49<1:04:25,  1.20s/it]
 55%|█████▌    | 3965/7175 [48:50<55:34,  1.04s/it]  
 55%|█████▌    | 3966/7175 [48:50<49:22,  1.08it/s]
 55%|█████▌    | 3967/7175 [48:51<45:04,  1.19it/s]
 55%|█████▌    | 3968/7175 [48:52<42:02,  1.27it/s]
 55%|█████▌    | 3969/7175 [48:52<39:54,  1.34it/s]
 55%|█████▌    | 3970/7175 [48:53<38:25,  1.39it/s]
                                                   
{'loss': '3.125', 'grad_norm': '0.2087', 'learning_rate': '0.0006', 'epoch': '13.84'}
+
 55%|█████▌    | 3970/7175 [48:53<38:25,  1.39it/s]
 55%|█████▌    | 3971/7175 [48:54<37:24,  1.43it/s]
 55%|█████▌    | 3972/7175 [48:54<36:40,  1.46it/s]
 55%|█████▌    | 3973/7175 [48:55<36:08,  1.48it/s]
 55%|█████▌    | 3974/7175 [48:56<35:45,  1.49it/s]
 55%|█████▌    | 3975/7175 [48:56<35:32,  1.50it/s]
 55%|█████▌    | 3976/7175 [48:57<35:21,  1.51it/s]
 55%|█████▌    | 3977/7175 [48:58<35:12,  1.51it/s]
 55%|█████▌    | 3978/7175 [48:58<35:04,  1.52it/s]
 55%|█████▌    | 3979/7175 [48:59<35:02,  1.52it/s]
 55%|█████▌    | 3980/7175 [49:00<34:59,  1.52it/s]
                                                   
{'loss': '3.112', 'grad_norm': '0.2378', 'learning_rate': '0.0006', 'epoch': '13.87'}
+
 55%|█████▌    | 3980/7175 [49:00<34:59,  1.52it/s]
 55%|█████▌    | 3981/7175 [49:00<34:57,  1.52it/s]
 55%|█████▌    | 3982/7175 [49:01<34:54,  1.52it/s]
 56%|█████▌    | 3983/7175 [49:02<34:53,  1.52it/s]
 56%|█████▌    | 3984/7175 [49:02<34:51,  1.53it/s]
 56%|█████▌    | 3985/7175 [49:03<34:50,  1.53it/s]
 56%|█████▌    | 3986/7175 [49:03<34:48,  1.53it/s]
 56%|█████▌    | 3987/7175 [49:04<34:49,  1.53it/s]
 56%|█████▌    | 3988/7175 [49:05<34:48,  1.53it/s]
 56%|█████▌    | 3989/7175 [49:05<34:48,  1.53it/s]
 56%|█████▌    | 3990/7175 [49:06<34:46,  1.53it/s]
                                                   
{'loss': '3.118', 'grad_norm': '0.2228', 'learning_rate': '0.0006', 'epoch': '13.91'}
+
 56%|█████▌    | 3990/7175 [49:06<34:46,  1.53it/s]
 56%|█████▌    | 3991/7175 [49:07<34:45,  1.53it/s]
 56%|█████▌    | 3992/7175 [49:07<34:45,  1.53it/s]
 56%|█████▌    | 3993/7175 [49:08<34:45,  1.53it/s]
 56%|█████▌    | 3994/7175 [49:09<34:42,  1.53it/s]
 56%|█████▌    | 3995/7175 [49:09<34:42,  1.53it/s]
 56%|█████▌    | 3996/7175 [49:10<34:42,  1.53it/s]
 56%|█████▌    | 3997/7175 [49:11<34:41,  1.53it/s]
 56%|█████▌    | 3998/7175 [49:11<34:40,  1.53it/s]
 56%|█████▌    | 3999/7175 [49:12<34:40,  1.53it/s]
 56%|█████▌    | 4000/7175 [49:13<34:39,  1.53it/s]
                                                   
{'loss': '3.124', 'grad_norm': '0.2177', 'learning_rate': '0.0006', 'epoch': '13.94'}
+
 56%|█████▌    | 4000/7175 [49:13<34:39,  1.53it/s]
 56%|█████▌    | 4001/7175 [49:13<34:39,  1.53it/s]
 56%|█████▌    | 4002/7175 [49:14<34:37,  1.53it/s]
 56%|█████▌    | 4003/7175 [49:15<34:37,  1.53it/s]
 56%|█████▌    | 4004/7175 [49:15<34:41,  1.52it/s]
 56%|█████▌    | 4005/7175 [49:16<34:39,  1.52it/s]
 56%|█████▌    | 4006/7175 [49:17<34:37,  1.53it/s]
 56%|█████▌    | 4007/7175 [49:17<34:34,  1.53it/s]
 56%|█████▌    | 4008/7175 [49:18<34:35,  1.53it/s]
 56%|█████▌    | 4009/7175 [49:19<34:34,  1.53it/s]
 56%|█████▌    | 4010/7175 [49:19<34:33,  1.53it/s]
                                                   
{'loss': '3.124', 'grad_norm': '0.2271', 'learning_rate': '0.0006', 'epoch': '13.98'}
+
 56%|█████▌    | 4010/7175 [49:19<34:33,  1.53it/s]
 56%|█████▌    | 4011/7175 [49:20<34:32,  1.53it/s]
 56%|█████▌    | 4012/7175 [49:21<34:32,  1.53it/s]
 56%|█████▌    | 4013/7175 [49:21<34:31,  1.53it/s]
 56%|█████▌    | 4014/7175 [49:22<34:31,  1.53it/s]
 56%|█████▌    | 4015/7175 [49:22<34:37,  1.52it/s]
 56%|█████▌    | 4016/7175 [49:23<34:33,  1.52it/s]
 56%|█████▌    | 4017/7175 [49:24<34:30,  1.53it/s]
 56%|█████▌    | 4019/7175 [49:25<28:51,  1.82it/s]
 56%|█████▌    | 4020/7175 [49:25<30:12,  1.74it/s]
                                                   
{'loss': '3.111', 'grad_norm': '0.2921', 'learning_rate': '0.0006', 'epoch': '14.01'}
+
 56%|█████▌    | 4020/7175 [49:25<30:12,  1.74it/s]
 56%|█████▌    | 4021/7175 [49:26<31:20,  1.68it/s]
 56%|█████▌    | 4022/7175 [49:27<32:10,  1.63it/s]
 56%|█████▌    | 4023/7175 [49:27<32:47,  1.60it/s]
 56%|█████▌    | 4024/7175 [49:28<33:12,  1.58it/s]
 56%|█████▌    | 4025/7175 [49:29<33:34,  1.56it/s]
 56%|█████▌    | 4026/7175 [49:29<33:48,  1.55it/s]
 56%|█████▌    | 4027/7175 [49:30<33:57,  1.55it/s]
 56%|█████▌    | 4028/7175 [49:31<34:02,  1.54it/s]
 56%|█████▌    | 4029/7175 [49:31<34:09,  1.54it/s]
 56%|█████▌    | 4030/7175 [49:32<34:11,  1.53it/s]
                                                   
{'loss': '3.053', 'grad_norm': '0.3078', 'learning_rate': '0.0006', 'epoch': '14.04'}
+
 56%|█████▌    | 4030/7175 [49:32<34:11,  1.53it/s]
 56%|█████▌    | 4031/7175 [49:32<34:14,  1.53it/s]
 56%|█████▌    | 4032/7175 [49:33<34:14,  1.53it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.39it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.75it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.56it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.05it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.72it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.58it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.44it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.40it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.32it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.31it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.28it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.27it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.25it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.26it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.25it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.24it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.24it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.22it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s]
                                                   
+
                                               
{'eval_loss': '3.549', 'eval_runtime': '3.245', 'eval_samples_per_second': '752.3', 'eval_steps_per_second': '12.02', 'epoch': '14.05'}
+
 56%|█████▌    | 4032/7175 [49:36<34:14,  1.53it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s][eval] epoch 14.049 step 4032 loss 3.5491 (best 3.5214 @ ep 13.8, 1/20 since best)
+
+
                                               
 56%|█████▌    | 4033/7175 [49:37<1:25:05,  1.63s/it]
 56%|█████▌    | 4034/7175 [49:38<1:09:53,  1.34s/it]
 56%|█████▌    | 4035/7175 [49:38<59:12,  1.13s/it]  
 56%|█████▋    | 4036/7175 [49:39<51:42,  1.01it/s]
 56%|█████▋    | 4037/7175 [49:40<46:26,  1.13it/s]
 56%|█████▋    | 4038/7175 [49:40<42:48,  1.22it/s]
 56%|█████▋    | 4039/7175 [49:41<40:14,  1.30it/s]
 56%|█████▋    | 4040/7175 [49:42<38:25,  1.36it/s]
                                                   
{'loss': '3.058', 'grad_norm': '0.2443', 'learning_rate': '0.0006', 'epoch': '14.08'}
+
 56%|█████▋    | 4040/7175 [49:42<38:25,  1.36it/s]
 56%|█████▋    | 4041/7175 [49:42<37:09,  1.41it/s]
 56%|█████▋    | 4042/7175 [49:43<36:15,  1.44it/s]
 56%|█████▋    | 4043/7175 [49:44<35:40,  1.46it/s]
 56%|█████▋    | 4044/7175 [49:44<35:14,  1.48it/s]
 56%|█████▋    | 4045/7175 [49:45<34:55,  1.49it/s]
 56%|█████▋    | 4046/7175 [49:46<34:41,  1.50it/s]
 56%|█████▋    | 4047/7175 [49:46<34:31,  1.51it/s]
 56%|█████▋    | 4048/7175 [49:47<34:25,  1.51it/s]
 56%|█████▋    | 4049/7175 [49:48<34:20,  1.52it/s]
 56%|█████▋    | 4050/7175 [49:48<34:16,  1.52it/s]
                                                   
{'loss': '3.056', 'grad_norm': '0.2438', 'learning_rate': '0.0006', 'epoch': '14.11'}
+
 56%|█████▋    | 4050/7175 [49:48<34:16,  1.52it/s]
 56%|█████▋    | 4051/7175 [49:49<34:13,  1.52it/s]
 56%|█████▋    | 4052/7175 [49:50<34:11,  1.52it/s]
 56%|█████▋    | 4053/7175 [49:50<34:10,  1.52it/s]
 57%|█████▋    | 4054/7175 [49:51<34:08,  1.52it/s]
 57%|█████▋    | 4055/7175 [49:51<34:05,  1.52it/s]
 57%|█████▋    | 4056/7175 [49:52<34:04,  1.53it/s]
 57%|█████▋    | 4057/7175 [49:53<34:05,  1.52it/s]
 57%|█████▋    | 4058/7175 [49:53<34:05,  1.52it/s]
 57%|█████▋    | 4059/7175 [49:54<34:04,  1.52it/s]
 57%|█████▋    | 4060/7175 [49:55<34:02,  1.52it/s]
                                                   
{'loss': '3.057', 'grad_norm': '0.2246', 'learning_rate': '0.0006', 'epoch': '14.15'}
+
 57%|█████▋    | 4060/7175 [49:55<34:02,  1.52it/s]
 57%|█████▋    | 4061/7175 [49:55<34:01,  1.53it/s]
 57%|█████▋    | 4062/7175 [49:56<34:02,  1.52it/s]
 57%|█████▋    | 4063/7175 [49:57<34:00,  1.52it/s]
 57%|█████▋    | 4064/7175 [49:57<33:59,  1.53it/s]
 57%|█████▋    | 4065/7175 [49:58<33:57,  1.53it/s]
 57%|█████▋    | 4066/7175 [49:59<33:58,  1.53it/s]
 57%|█████▋    | 4067/7175 [49:59<33:57,  1.53it/s]
 57%|█████▋    | 4068/7175 [50:00<33:57,  1.52it/s]
 57%|█████▋    | 4069/7175 [50:01<33:56,  1.53it/s]
 57%|█████▋    | 4070/7175 [50:01<33:54,  1.53it/s]
                                                   
{'loss': '3.066', 'grad_norm': '0.2432', 'learning_rate': '0.0006', 'epoch': '14.18'}
+
 57%|█████▋    | 4070/7175 [50:01<33:54,  1.53it/s]
 57%|█████▋    | 4071/7175 [50:02<33:57,  1.52it/s]
 57%|█████▋    | 4072/7175 [50:03<33:55,  1.52it/s]
 57%|█████▋    | 4073/7175 [50:03<33:54,  1.52it/s]
 57%|█████▋    | 4074/7175 [50:04<33:51,  1.53it/s]
 57%|█████▋    | 4075/7175 [50:05<33:51,  1.53it/s]
 57%|█████▋    | 4076/7175 [50:05<33:50,  1.53it/s]
 57%|█████▋    | 4077/7175 [50:06<33:49,  1.53it/s]
 57%|█████▋    | 4078/7175 [50:07<33:48,  1.53it/s]
 57%|█████▋    | 4079/7175 [50:07<33:48,  1.53it/s]
 57%|█████▋    | 4080/7175 [50:08<33:48,  1.53it/s]
                                                   
{'loss': '3.076', 'grad_norm': '0.2194', 'learning_rate': '0.0006', 'epoch': '14.22'}
+
 57%|█████▋    | 4080/7175 [50:08<33:48,  1.53it/s]
 57%|█████▋    | 4081/7175 [50:09<33:47,  1.53it/s]
 57%|█████▋    | 4082/7175 [50:09<33:46,  1.53it/s]
 57%|█████▋    | 4083/7175 [50:10<33:45,  1.53it/s]
 57%|█████▋    | 4084/7175 [50:10<33:44,  1.53it/s]
 57%|█████▋    | 4085/7175 [50:11<33:43,  1.53it/s]
 57%|█████▋    | 4086/7175 [50:12<33:43,  1.53it/s]
 57%|█████▋    | 4087/7175 [50:12<33:43,  1.53it/s]
 57%|█████▋    | 4088/7175 [50:13<33:44,  1.53it/s]
 57%|█████▋    | 4089/7175 [50:14<33:43,  1.52it/s]
 57%|█████▋    | 4090/7175 [50:14<33:43,  1.52it/s]
                                                   
{'loss': '3.064', 'grad_norm': '0.2373', 'learning_rate': '0.0006', 'epoch': '14.25'}
+
 57%|█████▋    | 4090/7175 [50:14<33:43,  1.52it/s]
 57%|█████▋    | 4091/7175 [50:15<33:41,  1.53it/s]
 57%|█████▋    | 4092/7175 [50:16<33:41,  1.53it/s]
 57%|█████▋    | 4093/7175 [50:16<33:40,  1.53it/s]
 57%|█████▋    | 4094/7175 [50:17<33:38,  1.53it/s]
 57%|█████▋    | 4095/7175 [50:18<33:37,  1.53it/s]
 57%|█████▋    | 4096/7175 [50:18<33:36,  1.53it/s]
 57%|█████▋    | 4097/7175 [50:19<33:36,  1.53it/s]
 57%|█████▋    | 4098/7175 [50:20<33:36,  1.53it/s]
 57%|█████▋    | 4099/7175 [50:20<33:35,  1.53it/s]
 57%|█████▋    | 4100/7175 [50:21<33:34,  1.53it/s]
                                                   
{'loss': '3.084', 'grad_norm': '0.2463', 'learning_rate': '0.0006', 'epoch': '14.29'}
+
 57%|█████▋    | 4100/7175 [50:21<33:34,  1.53it/s]
 57%|█████▋    | 4101/7175 [50:22<33:36,  1.52it/s]
 57%|█████▋    | 4102/7175 [50:22<33:36,  1.52it/s]
 57%|█████▋    | 4103/7175 [50:23<33:36,  1.52it/s]
 57%|█████▋    | 4104/7175 [50:24<33:36,  1.52it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.69it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.76it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.57it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 12.99it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.72it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.51it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.42it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.33it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.29it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.28it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.27it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.26it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.25it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.23it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.22it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.23it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.22it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.23it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s]
                                                   
+
                                               
{'eval_loss': '3.529', 'eval_runtime': '3.271', 'eval_samples_per_second': '746.2', 'eval_steps_per_second': '11.92', 'epoch': '14.3'}
+
 57%|█████▋    | 4104/7175 [50:27<33:36,  1.52it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s][eval] epoch 14.301 step 4104 loss 3.5287 (best 3.5214 @ ep 13.8, 2/20 since best)
+
+
                                               
 57%|█████▋    | 4105/7175 [50:28<1:23:47,  1.64s/it]
 57%|█████▋    | 4106/7175 [50:28<1:08:43,  1.34s/it]
 57%|█████▋    | 4107/7175 [50:29<58:11,  1.14s/it]  
 57%|█████▋    | 4108/7175 [50:29<50:50,  1.01it/s]
 57%|█████▋    | 4109/7175 [50:30<45:37,  1.12it/s]
 57%|█████▋    | 4110/7175 [50:31<41:56,  1.22it/s]
                                                   
{'loss': '3.083', 'grad_norm': '0.2141', 'learning_rate': '0.0006', 'epoch': '14.32'}
+
 57%|█████▋    | 4110/7175 [50:31<41:56,  1.22it/s]
 57%|█████▋    | 4111/7175 [50:31<39:43,  1.29it/s]
 57%|█████▋    | 4112/7175 [50:32<39:37,  1.29it/s]
 57%|█████▋    | 4113/7175 [50:33<37:51,  1.35it/s]
 57%|█████▋    | 4114/7175 [50:34<36:29,  1.40it/s]
 57%|█████▋    | 4115/7175 [50:34<35:56,  1.42it/s]
 57%|█████▋    | 4116/7175 [50:35<35:16,  1.45it/s]
 57%|█████▋    | 4117/7175 [50:36<34:51,  1.46it/s]
 57%|█████▋    | 4118/7175 [50:36<34:35,  1.47it/s]
 57%|█████▋    | 4119/7175 [50:37<34:18,  1.48it/s]
 57%|█████▋    | 4120/7175 [50:38<34:05,  1.49it/s]
                                                   
{'loss': '3.09', 'grad_norm': '0.2223', 'learning_rate': '0.0006', 'epoch': '14.36'}
+
 57%|█████▋    | 4120/7175 [50:38<34:05,  1.49it/s]
 57%|█████▋    | 4121/7175 [50:38<33:54,  1.50it/s]
 57%|█████▋    | 4122/7175 [50:39<33:43,  1.51it/s]
 57%|█████▋    | 4123/7175 [50:40<33:36,  1.51it/s]
 57%|█████▋    | 4124/7175 [50:40<33:30,  1.52it/s]
 57%|█████▋    | 4125/7175 [50:41<33:26,  1.52it/s]
 58%|█████▊    | 4126/7175 [50:41<33:23,  1.52it/s]
 58%|█████▊    | 4127/7175 [50:42<33:21,  1.52it/s]
 58%|█████▊    | 4128/7175 [50:43<33:18,  1.52it/s]
 58%|█████▊    | 4129/7175 [50:43<33:17,  1.53it/s]
 58%|█████▊    | 4130/7175 [50:44<33:16,  1.53it/s]
                                                   
{'loss': '3.094', 'grad_norm': '0.2153', 'learning_rate': '0.0006', 'epoch': '14.39'}
+
 58%|█████▊    | 4130/7175 [50:44<33:16,  1.53it/s]
 58%|█████▊    | 4131/7175 [50:45<33:16,  1.52it/s]
 58%|█████▊    | 4132/7175 [50:45<33:15,  1.53it/s]
 58%|█████▊    | 4133/7175 [50:46<33:12,  1.53it/s]
 58%|█████▊    | 4134/7175 [50:47<33:14,  1.52it/s]
 58%|█████▊    | 4135/7175 [50:47<33:13,  1.53it/s]
 58%|█████▊    | 4136/7175 [50:48<33:11,  1.53it/s]
 58%|█████▊    | 4137/7175 [50:49<33:09,  1.53it/s]
 58%|█████▊    | 4138/7175 [50:49<33:10,  1.53it/s]
 58%|█████▊    | 4139/7175 [50:50<33:09,  1.53it/s]
 58%|█████▊    | 4140/7175 [50:51<33:08,  1.53it/s]
                                                   
{'loss': '3.085', 'grad_norm': '0.2287', 'learning_rate': '0.0006', 'epoch': '14.43'}
+
 58%|█████▊    | 4140/7175 [50:51<33:08,  1.53it/s]
 58%|█████▊    | 4141/7175 [50:51<33:07,  1.53it/s]
 58%|█████▊    | 4142/7175 [50:52<33:07,  1.53it/s]
 58%|█████▊    | 4143/7175 [50:53<33:06,  1.53it/s]
 58%|█████▊    | 4144/7175 [50:53<33:05,  1.53it/s]
 58%|█████▊    | 4145/7175 [50:54<33:04,  1.53it/s]
 58%|█████▊    | 4146/7175 [50:55<33:05,  1.53it/s]
 58%|█████▊    | 4147/7175 [50:55<33:04,  1.53it/s]
 58%|█████▊    | 4148/7175 [50:56<33:03,  1.53it/s]
 58%|█████▊    | 4149/7175 [50:57<33:01,  1.53it/s]
 58%|█████▊    | 4150/7175 [50:57<33:01,  1.53it/s]
                                                   
{'loss': '3.079', 'grad_norm': '0.2159', 'learning_rate': '0.0006', 'epoch': '14.46'}
+
 58%|█████▊    | 4150/7175 [50:57<33:01,  1.53it/s]
 58%|█████▊    | 4151/7175 [50:58<33:03,  1.52it/s]
 58%|█████▊    | 4152/7175 [50:59<33:01,  1.53it/s]
 58%|█████▊    | 4153/7175 [50:59<33:00,  1.53it/s]
 58%|█████▊    | 4154/7175 [51:00<32:58,  1.53it/s]
 58%|█████▊    | 4155/7175 [51:01<32:59,  1.53it/s]
 58%|█████▊    | 4156/7175 [51:01<32:58,  1.53it/s]
 58%|█████▊    | 4157/7175 [51:02<32:57,  1.53it/s]
 58%|█████▊    | 4158/7175 [51:02<32:54,  1.53it/s]
 58%|█████▊    | 4159/7175 [51:03<32:56,  1.53it/s]
 58%|█████▊    | 4160/7175 [51:04<32:55,  1.53it/s]
                                                   
{'loss': '3.094', 'grad_norm': '0.2173', 'learning_rate': '0.0006', 'epoch': '14.5'}
+
 58%|█████▊    | 4160/7175 [51:04<32:55,  1.53it/s]
 58%|█████▊    | 4161/7175 [51:04<32:55,  1.53it/s]
 58%|█████▊    | 4162/7175 [51:05<32:54,  1.53it/s]
 58%|█████▊    | 4163/7175 [51:06<32:53,  1.53it/s]
 58%|█████▊    | 4164/7175 [51:06<32:53,  1.53it/s]
 58%|█████▊    | 4165/7175 [51:07<32:53,  1.53it/s]
 58%|█████▊    | 4166/7175 [51:08<32:54,  1.52it/s]
 58%|█████▊    | 4167/7175 [51:08<32:52,  1.52it/s]
 58%|█████▊    | 4168/7175 [51:09<32:52,  1.52it/s]
 58%|█████▊    | 4169/7175 [51:10<32:52,  1.52it/s]
 58%|█████▊    | 4170/7175 [51:10<32:51,  1.52it/s]
                                                   
{'loss': '3.089', 'grad_norm': '0.2199', 'learning_rate': '0.0006', 'epoch': '14.53'}
+
 58%|█████▊    | 4170/7175 [51:10<32:51,  1.52it/s]
 58%|█████▊    | 4171/7175 [51:11<32:51,  1.52it/s]
 58%|█████▊    | 4172/7175 [51:12<32:49,  1.52it/s]
 58%|█████▊    | 4173/7175 [51:12<32:48,  1.52it/s]
 58%|█████▊    | 4174/7175 [51:13<32:49,  1.52it/s]
 58%|█████▊    | 4175/7175 [51:14<32:48,  1.52it/s]
 58%|█████▊    | 4176/7175 [51:14<32:47,  1.52it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.38it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.68it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.58it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.04it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.75it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.56it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.45it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.36it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.34it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.30it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.29it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.27it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.26it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.24it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.24it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.23it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.23it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.23it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s]
                                                   
+
                                               
{'eval_loss': '3.52', 'eval_runtime': '3.251', 'eval_samples_per_second': '750.8', 'eval_steps_per_second': '12', 'epoch': '14.55'}
+
 58%|█████▊    | 4176/7175 [51:18<32:47,  1.52it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-4176
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-4176/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-4176/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.61it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.61it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-4176/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-4176/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-4176/tokenizer_config.json
+[eval] epoch 14.552 step 4176 loss 3.5195 (best 3.5195 @ ep 14.552, 0/20 since best)
+
+
                                               
 58%|█████▊    | 4177/7175 [51:20<1:54:15,  2.29s/it]
 58%|█████▊    | 4178/7175 [51:21<1:29:45,  1.80s/it]
 58%|█████▊    | 4179/7175 [51:22<1:12:37,  1.45s/it]
 58%|█████▊    | 4180/7175 [51:22<1:00:36,  1.21s/it]
                                                     
{'loss': '3.078', 'grad_norm': '0.2156', 'learning_rate': '0.0006', 'epoch': '14.57'}
+
 58%|█████▊    | 4180/7175 [51:22<1:00:36,  1.21s/it]
 58%|█████▊    | 4181/7175 [51:23<52:13,  1.05s/it]  
 58%|█████▊    | 4182/7175 [51:24<46:20,  1.08it/s]
 58%|█████▊    | 4183/7175 [51:24<42:14,  1.18it/s]
 58%|█████▊    | 4184/7175 [51:25<39:20,  1.27it/s]
 58%|█████▊    | 4185/7175 [51:26<37:19,  1.33it/s]
 58%|█████▊    | 4186/7175 [51:26<35:54,  1.39it/s]
 58%|█████▊    | 4187/7175 [51:27<34:54,  1.43it/s]
 58%|█████▊    | 4188/7175 [51:28<34:11,  1.46it/s]
 58%|█████▊    | 4189/7175 [51:28<33:43,  1.48it/s]
 58%|█████▊    | 4190/7175 [51:29<33:22,  1.49it/s]
                                                   
{'loss': '3.099', 'grad_norm': '0.2322', 'learning_rate': '0.0006', 'epoch': '14.6'}
+
 58%|█████▊    | 4190/7175 [51:29<33:22,  1.49it/s]
 58%|█████▊    | 4191/7175 [51:30<33:08,  1.50it/s]
 58%|█████▊    | 4192/7175 [51:30<32:59,  1.51it/s]
 58%|█████▊    | 4193/7175 [51:31<32:50,  1.51it/s]
 58%|█████▊    | 4194/7175 [51:32<32:46,  1.52it/s]
 58%|█████▊    | 4195/7175 [51:32<32:41,  1.52it/s]
 58%|█████▊    | 4196/7175 [51:33<32:38,  1.52it/s]
 58%|█████▊    | 4197/7175 [51:33<32:34,  1.52it/s]
 59%|█████▊    | 4198/7175 [51:34<32:33,  1.52it/s]
 59%|█████▊    | 4199/7175 [51:35<32:31,  1.52it/s]
 59%|█████▊    | 4200/7175 [51:35<32:30,  1.53it/s]
                                                   
{'loss': '3.094', 'grad_norm': '0.2076', 'learning_rate': '0.0006', 'epoch': '14.64'}
+
 59%|█████▊    | 4200/7175 [51:35<32:30,  1.53it/s]
 59%|█████▊    | 4201/7175 [51:36<32:29,  1.53it/s]
 59%|█████▊    | 4202/7175 [51:37<32:28,  1.53it/s]
 59%|█████▊    | 4203/7175 [51:37<32:27,  1.53it/s]
 59%|█████▊    | 4204/7175 [51:38<32:28,  1.52it/s]
 59%|█████▊    | 4205/7175 [51:39<32:26,  1.53it/s]
 59%|█████▊    | 4206/7175 [51:39<32:25,  1.53it/s]
 59%|█████▊    | 4207/7175 [51:40<32:26,  1.52it/s]
 59%|█████▊    | 4208/7175 [51:41<32:25,  1.53it/s]
 59%|█████▊    | 4209/7175 [51:41<32:24,  1.53it/s]
 59%|█████▊    | 4210/7175 [51:42<32:21,  1.53it/s]
                                                   
{'loss': '3.081', 'grad_norm': '0.2256', 'learning_rate': '0.0006', 'epoch': '14.67'}
+
 59%|█████▊    | 4210/7175 [51:42<32:21,  1.53it/s]
 59%|█████▊    | 4211/7175 [51:43<32:23,  1.53it/s]
 59%|█████▊    | 4212/7175 [51:43<32:22,  1.53it/s]
 59%|█████▊    | 4213/7175 [51:44<32:21,  1.53it/s]
 59%|█████▊    | 4214/7175 [51:45<32:19,  1.53it/s]
 59%|█████▊    | 4215/7175 [51:45<32:19,  1.53it/s]
 59%|█████▉    | 4216/7175 [51:46<32:18,  1.53it/s]
 59%|█████▉    | 4217/7175 [51:47<32:18,  1.53it/s]
 59%|█████▉    | 4218/7175 [51:47<32:16,  1.53it/s]
 59%|█████▉    | 4219/7175 [51:48<32:16,  1.53it/s]
 59%|█████▉    | 4220/7175 [51:49<32:16,  1.53it/s]
                                                   
{'loss': '3.104', 'grad_norm': '0.2323', 'learning_rate': '0.0006', 'epoch': '14.71'}
+
 59%|█████▉    | 4220/7175 [51:49<32:16,  1.53it/s]
 59%|█████▉    | 4221/7175 [51:49<32:18,  1.52it/s]
 59%|█████▉    | 4222/7175 [51:50<32:17,  1.52it/s]
 59%|█████▉    | 4223/7175 [51:51<32:14,  1.53it/s]
 59%|█████▉    | 4224/7175 [51:51<32:14,  1.53it/s]
 59%|█████▉    | 4225/7175 [51:52<32:14,  1.53it/s]
 59%|█████▉    | 4226/7175 [51:52<32:12,  1.53it/s]
 59%|█████▉    | 4227/7175 [51:53<32:11,  1.53it/s]
 59%|█████▉    | 4228/7175 [51:54<32:11,  1.53it/s]
 59%|█████▉    | 4229/7175 [51:54<32:11,  1.53it/s]
 59%|█████▉    | 4230/7175 [51:55<32:10,  1.53it/s]
                                                   
{'loss': '3.113', 'grad_norm': '0.2175', 'learning_rate': '0.0006', 'epoch': '14.74'}
+
 59%|█████▉    | 4230/7175 [51:55<32:10,  1.53it/s]
 59%|█████▉    | 4231/7175 [51:56<32:10,  1.53it/s]
 59%|█████▉    | 4232/7175 [51:56<32:07,  1.53it/s]
 59%|█████▉    | 4233/7175 [51:57<32:07,  1.53it/s]
 59%|█████▉    | 4234/7175 [51:58<32:06,  1.53it/s]
 59%|█████▉    | 4235/7175 [51:58<32:04,  1.53it/s]
 59%|█████▉    | 4236/7175 [51:59<32:04,  1.53it/s]
 59%|█████▉    | 4237/7175 [52:00<32:04,  1.53it/s]
 59%|█████▉    | 4238/7175 [52:00<32:03,  1.53it/s]
 59%|█████▉    | 4239/7175 [52:01<32:02,  1.53it/s]
 59%|█████▉    | 4240/7175 [52:02<32:02,  1.53it/s]
                                                   
{'loss': '3.093', 'grad_norm': '0.1972', 'learning_rate': '0.0006', 'epoch': '14.78'}
+
 59%|█████▉    | 4240/7175 [52:02<32:02,  1.53it/s]
 59%|█████▉    | 4241/7175 [52:02<32:03,  1.52it/s]
 59%|█████▉    | 4242/7175 [52:03<32:02,  1.53it/s]
 59%|█████▉    | 4243/7175 [52:04<32:02,  1.53it/s]
 59%|█████▉    | 4244/7175 [52:04<31:59,  1.53it/s]
 59%|█████▉    | 4245/7175 [52:05<31:59,  1.53it/s]
 59%|█████▉    | 4246/7175 [52:06<31:59,  1.53it/s]
 59%|█████▉    | 4247/7175 [52:06<31:58,  1.53it/s]
 59%|█████▉    | 4248/7175 [52:07<31:55,  1.53it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.38it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.74it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.57it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.05it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.75it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.57it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.44it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.38it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.31it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.29it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.28it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.26it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.26it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.26it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.25it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.24it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.23it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.23it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s]
                                                   
+
                                               
{'eval_loss': '3.514', 'eval_runtime': '3.263', 'eval_samples_per_second': '748.2', 'eval_steps_per_second': '11.95', 'epoch': '14.8'}
+
 59%|█████▉    | 4248/7175 [52:10<31:55,  1.53it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-4248
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-4248/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-4248/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.84it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.84it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-4248/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-4248/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-4248/tokenizer_config.json
+[eval] epoch 14.804 step 4248 loss 3.5139 (best 3.5139 @ ep 14.804, 0/20 since best)
+
+
                                               
 59%|█████▉    | 4249/7175 [52:13<1:50:15,  2.26s/it]
 59%|█████▉    | 4250/7175 [52:14<1:26:42,  1.78s/it]
                                                     
{'loss': '3.103', 'grad_norm': '0.2039', 'learning_rate': '0.0006', 'epoch': '14.81'}
+
 59%|█████▉    | 4250/7175 [52:14<1:26:42,  1.78s/it]
 59%|█████▉    | 4251/7175 [52:14<1:10:15,  1.44s/it]
 59%|█████▉    | 4252/7175 [52:15<58:44,  1.21s/it]  
 59%|█████▉    | 4253/7175 [52:16<50:40,  1.04s/it]
 59%|█████▉    | 4254/7175 [52:16<45:02,  1.08it/s]
 59%|█████▉    | 4255/7175 [52:17<41:04,  1.18it/s]
 59%|█████▉    | 4256/7175 [52:17<38:18,  1.27it/s]
 59%|█████▉    | 4257/7175 [52:18<36:22,  1.34it/s]
 59%|█████▉    | 4258/7175 [52:19<35:02,  1.39it/s]
 59%|█████▉    | 4259/7175 [52:19<34:04,  1.43it/s]
 59%|█████▉    | 4260/7175 [52:20<33:24,  1.45it/s]
                                                   
{'loss': '3.105', 'grad_norm': '0.2188', 'learning_rate': '0.0006', 'epoch': '14.85'}
+
 59%|█████▉    | 4260/7175 [52:20<33:24,  1.45it/s]
 59%|█████▉    | 4261/7175 [52:21<32:56,  1.47it/s]
 59%|█████▉    | 4262/7175 [52:21<32:34,  1.49it/s]
 59%|█████▉    | 4263/7175 [52:22<32:19,  1.50it/s]
 59%|█████▉    | 4264/7175 [52:23<32:09,  1.51it/s]
 59%|█████▉    | 4265/7175 [52:23<32:02,  1.51it/s]
 59%|█████▉    | 4266/7175 [52:24<31:56,  1.52it/s]
 59%|█████▉    | 4267/7175 [52:25<31:52,  1.52it/s]
 59%|█████▉    | 4268/7175 [52:25<31:49,  1.52it/s]
 59%|█████▉    | 4269/7175 [52:26<31:47,  1.52it/s]
 60%|█████▉    | 4270/7175 [52:27<31:45,  1.52it/s]
                                                   
{'loss': '3.099', 'grad_norm': '0.2265', 'learning_rate': '0.0006', 'epoch': '14.88'}
+
 60%|█████▉    | 4270/7175 [52:27<31:45,  1.52it/s]
 60%|█████▉    | 4271/7175 [52:27<31:44,  1.53it/s]
 60%|█████▉    | 4272/7175 [52:28<31:43,  1.53it/s]
 60%|█████▉    | 4273/7175 [52:29<31:43,  1.52it/s]
 60%|█████▉    | 4274/7175 [52:29<31:42,  1.53it/s]
 60%|█████▉    | 4275/7175 [52:30<31:41,  1.53it/s]
 60%|█████▉    | 4276/7175 [52:31<31:39,  1.53it/s]
 60%|█████▉    | 4277/7175 [52:31<31:40,  1.52it/s]
 60%|█████▉    | 4278/7175 [52:32<31:39,  1.52it/s]
 60%|█████▉    | 4279/7175 [52:33<31:39,  1.52it/s]
 60%|█████▉    | 4280/7175 [52:33<31:38,  1.53it/s]
                                                   
{'loss': '3.119', 'grad_norm': '0.2243', 'learning_rate': '0.0006', 'epoch': '14.92'}
+
 60%|█████▉    | 4280/7175 [52:33<31:38,  1.53it/s]
 60%|█████▉    | 4281/7175 [52:34<31:36,  1.53it/s]
 60%|█████▉    | 4282/7175 [52:35<31:37,  1.52it/s]
 60%|█████▉    | 4283/7175 [52:35<31:35,  1.53it/s]
 60%|█████▉    | 4284/7175 [52:36<31:35,  1.53it/s]
 60%|█████▉    | 4285/7175 [52:36<31:33,  1.53it/s]
 60%|█████▉    | 4286/7175 [52:37<31:33,  1.53it/s]
 60%|█████▉    | 4287/7175 [52:38<31:33,  1.53it/s]
 60%|█████▉    | 4288/7175 [52:38<31:32,  1.53it/s]
 60%|█████▉    | 4289/7175 [52:39<31:31,  1.53it/s]
 60%|█████▉    | 4290/7175 [52:40<31:30,  1.53it/s]
                                                   
{'loss': '3.104', 'grad_norm': '0.2234', 'learning_rate': '0.0006', 'epoch': '14.95'}
+
 60%|█████▉    | 4290/7175 [52:40<31:30,  1.53it/s]
 60%|█████▉    | 4291/7175 [52:40<31:31,  1.52it/s]
 60%|█████▉    | 4292/7175 [52:41<31:30,  1.53it/s]
 60%|█████▉    | 4293/7175 [52:42<31:29,  1.53it/s]
 60%|█████▉    | 4294/7175 [52:42<31:27,  1.53it/s]
 60%|█████▉    | 4295/7175 [52:43<31:28,  1.52it/s]
 60%|█████▉    | 4296/7175 [52:44<31:28,  1.52it/s]
 60%|█████▉    | 4297/7175 [52:44<31:26,  1.53it/s]
 60%|█████▉    | 4298/7175 [52:45<31:26,  1.53it/s]
 60%|█████▉    | 4299/7175 [52:46<31:25,  1.53it/s]
 60%|█████▉    | 4300/7175 [52:46<31:26,  1.52it/s]
                                                   
{'loss': '3.093', 'grad_norm': '0.2433', 'learning_rate': '0.0006', 'epoch': '14.99'}
+
 60%|█████▉    | 4300/7175 [52:46<31:26,  1.52it/s]
 60%|█████▉    | 4301/7175 [52:47<31:25,  1.52it/s]
 60%|█████▉    | 4302/7175 [52:48<31:31,  1.52it/s]
 60%|█████▉    | 4303/7175 [52:48<31:27,  1.52it/s]
 60%|█████▉    | 4304/7175 [52:49<31:25,  1.52it/s]
 60%|██████    | 4306/7175 [52:50<26:06,  1.83it/s]
 60%|██████    | 4307/7175 [52:50<27:24,  1.74it/s]
 60%|██████    | 4308/7175 [52:51<28:25,  1.68it/s]
 60%|██████    | 4309/7175 [52:52<29:12,  1.64it/s]
 60%|██████    | 4310/7175 [52:52<29:46,  1.60it/s]
                                                   
{'loss': '3.084', 'grad_norm': '0.3976', 'learning_rate': '0.0006', 'epoch': '15.02'}
+
 60%|██████    | 4310/7175 [52:52<29:46,  1.60it/s]
 60%|██████    | 4311/7175 [52:53<30:12,  1.58it/s]
 60%|██████    | 4312/7175 [52:54<30:32,  1.56it/s]
 60%|██████    | 4313/7175 [52:54<30:44,  1.55it/s]
 60%|██████    | 4314/7175 [52:55<30:53,  1.54it/s]
 60%|██████    | 4315/7175 [52:56<30:58,  1.54it/s]
 60%|██████    | 4316/7175 [52:56<31:03,  1.53it/s]
 60%|██████    | 4317/7175 [52:57<31:06,  1.53it/s]
 60%|██████    | 4318/7175 [52:58<31:08,  1.53it/s]
 60%|██████    | 4319/7175 [52:58<31:09,  1.53it/s]
 60%|██████    | 4320/7175 [52:59<31:09,  1.53it/s]
                                                   
{'loss': '3.047', 'grad_norm': '0.2632', 'learning_rate': '0.0006', 'epoch': '15.05'}
+
 60%|██████    | 4320/7175 [52:59<31:09,  1.53it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.48it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.77it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.58it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.05it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.75it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.53it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.44it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.35it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.31it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.27it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.25it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.25it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.23it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.23it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.24it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.23it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.22it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.23it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.22it/s]
                                                   
+
                                               
{'eval_loss': '3.539', 'eval_runtime': '3.267', 'eval_samples_per_second': '747.2', 'eval_steps_per_second': '11.94', 'epoch': '15.05'}
+
 60%|██████    | 4320/7175 [53:02<31:09,  1.53it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.22it/s][eval] epoch 15.052 step 4320 loss 3.5389 (best 3.5139 @ ep 14.804, 1/20 since best)
+
+
                                               
 60%|██████    | 4321/7175 [53:03<1:17:45,  1.63s/it]
 60%|██████    | 4322/7175 [53:04<1:03:45,  1.34s/it]
 60%|██████    | 4323/7175 [53:04<53:58,  1.14s/it]  
 60%|██████    | 4324/7175 [53:05<47:07,  1.01it/s]
 60%|██████    | 4325/7175 [53:06<42:18,  1.12it/s]
 60%|██████    | 4326/7175 [53:06<38:55,  1.22it/s]
 60%|██████    | 4327/7175 [53:07<36:34,  1.30it/s]
 60%|██████    | 4328/7175 [53:07<34:54,  1.36it/s]
 60%|██████    | 4329/7175 [53:08<33:45,  1.41it/s]
 60%|██████    | 4330/7175 [53:09<32:56,  1.44it/s]
                                                   
{'loss': '3.043', 'grad_norm': '0.2083', 'learning_rate': '0.0006', 'epoch': '15.09'}
+
 60%|██████    | 4330/7175 [53:09<32:56,  1.44it/s]
 60%|██████    | 4331/7175 [53:09<32:22,  1.46it/s]
 60%|██████    | 4332/7175 [53:10<31:59,  1.48it/s]
 60%|██████    | 4333/7175 [53:11<31:42,  1.49it/s]
 60%|██████    | 4334/7175 [53:11<31:30,  1.50it/s]
 60%|██████    | 4335/7175 [53:12<31:20,  1.51it/s]
 60%|██████    | 4336/7175 [53:13<31:15,  1.51it/s]
 60%|██████    | 4337/7175 [53:13<31:11,  1.52it/s]
 60%|██████    | 4338/7175 [53:14<31:07,  1.52it/s]
 60%|██████    | 4339/7175 [53:15<31:04,  1.52it/s]
 60%|██████    | 4340/7175 [53:15<31:01,  1.52it/s]
                                                   
{'loss': '3.04', 'grad_norm': '0.2374', 'learning_rate': '0.0006', 'epoch': '15.12'}
+
 60%|██████    | 4340/7175 [53:15<31:01,  1.52it/s]
 61%|██████    | 4341/7175 [53:16<31:01,  1.52it/s]
 61%|██████    | 4342/7175 [53:17<31:00,  1.52it/s]
 61%|██████    | 4343/7175 [53:17<30:58,  1.52it/s]
 61%|██████    | 4344/7175 [53:18<30:57,  1.52it/s]
 61%|██████    | 4345/7175 [53:19<30:55,  1.53it/s]
 61%|██████    | 4346/7175 [53:19<30:55,  1.52it/s]
 61%|██████    | 4347/7175 [53:20<30:54,  1.52it/s]
 61%|██████    | 4348/7175 [53:21<30:53,  1.52it/s]
 61%|██████    | 4349/7175 [53:21<30:52,  1.53it/s]
 61%|██████    | 4350/7175 [53:22<30:52,  1.53it/s]
                                                   
{'loss': '3.05', 'grad_norm': '0.2561', 'learning_rate': '0.0006', 'epoch': '15.16'}
+
 61%|██████    | 4350/7175 [53:22<30:52,  1.53it/s]
 61%|██████    | 4351/7175 [53:23<30:53,  1.52it/s]
 61%|██████    | 4352/7175 [53:23<30:51,  1.52it/s]
 61%|██████    | 4353/7175 [53:24<30:52,  1.52it/s]
 61%|██████    | 4354/7175 [53:25<30:50,  1.52it/s]
 61%|██████    | 4355/7175 [53:25<30:49,  1.52it/s]
 61%|██████    | 4356/7175 [53:26<30:49,  1.52it/s]
 61%|██████    | 4357/7175 [53:26<30:48,  1.52it/s]
 61%|██████    | 4358/7175 [53:27<30:47,  1.52it/s]
 61%|██████    | 4359/7175 [53:28<30:45,  1.53it/s]
 61%|██████    | 4360/7175 [53:28<30:45,  1.53it/s]
                                                   
{'loss': '3.039', 'grad_norm': '0.214', 'learning_rate': '0.0006', 'epoch': '15.19'}
+
 61%|██████    | 4360/7175 [53:28<30:45,  1.53it/s]
 61%|██████    | 4361/7175 [53:29<30:46,  1.52it/s]
 61%|██████    | 4362/7175 [53:30<30:44,  1.52it/s]
 61%|██████    | 4363/7175 [53:30<30:43,  1.53it/s]
 61%|██████    | 4364/7175 [53:31<30:41,  1.53it/s]
 61%|██████    | 4365/7175 [53:32<30:42,  1.52it/s]
 61%|██████    | 4366/7175 [53:32<30:41,  1.53it/s]
 61%|██████    | 4367/7175 [53:33<30:41,  1.53it/s]
 61%|██████    | 4368/7175 [53:34<30:40,  1.53it/s]
 61%|██████    | 4369/7175 [53:34<30:39,  1.53it/s]
 61%|██████    | 4370/7175 [53:35<30:39,  1.53it/s]
                                                   
{'loss': '3.045', 'grad_norm': '0.2266', 'learning_rate': '0.0006', 'epoch': '15.23'}
+
 61%|██████    | 4370/7175 [53:35<30:39,  1.53it/s]
 61%|██████    | 4371/7175 [53:36<30:39,  1.52it/s]
 61%|██████    | 4372/7175 [53:36<30:38,  1.52it/s]
 61%|██████    | 4373/7175 [53:37<30:37,  1.53it/s]
 61%|██████    | 4374/7175 [53:38<30:37,  1.52it/s]
 61%|██████    | 4375/7175 [53:38<30:36,  1.52it/s]
 61%|██████    | 4376/7175 [53:39<30:36,  1.52it/s]
 61%|██████    | 4377/7175 [53:40<30:35,  1.52it/s]
 61%|██████    | 4378/7175 [53:40<30:33,  1.53it/s]
 61%|██████    | 4379/7175 [53:41<30:33,  1.52it/s]
 61%|██████    | 4380/7175 [53:42<30:32,  1.53it/s]
                                                   
{'loss': '3.06', 'grad_norm': '0.2552', 'learning_rate': '0.0006', 'epoch': '15.26'}
+
 61%|██████    | 4380/7175 [53:42<30:32,  1.53it/s]
 61%|██████    | 4381/7175 [53:42<30:32,  1.52it/s]
 61%|██████    | 4382/7175 [53:43<30:30,  1.53it/s]
 61%|██████    | 4383/7175 [53:44<30:29,  1.53it/s]
 61%|██████    | 4384/7175 [53:44<30:30,  1.52it/s]
 61%|██████    | 4385/7175 [53:45<30:29,  1.53it/s]
 61%|██████    | 4386/7175 [53:46<30:28,  1.53it/s]
 61%|██████    | 4387/7175 [53:46<30:26,  1.53it/s]
 61%|██████    | 4388/7175 [53:47<30:27,  1.52it/s]
 61%|██████    | 4389/7175 [53:47<30:26,  1.53it/s]
 61%|██████    | 4390/7175 [53:48<30:25,  1.53it/s]
                                                   
{'loss': '3.054', 'grad_norm': '0.2319', 'learning_rate': '0.0006', 'epoch': '15.3'}
+
 61%|██████    | 4390/7175 [53:48<30:25,  1.53it/s]
 61%|██████    | 4391/7175 [53:49<30:24,  1.53it/s]
 61%|██████    | 4392/7175 [53:49<30:24,  1.53it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.02it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.59it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.52it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.02it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.73it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.56it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.44it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.36it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.31it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.27it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.29it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.27it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.26it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.25it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.25it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.23it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.24it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.23it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s]
                                                   
+
                                               
{'eval_loss': '3.525', 'eval_runtime': '3.251', 'eval_samples_per_second': '750.8', 'eval_steps_per_second': '12', 'epoch': '15.3'}
+
 61%|██████    | 4392/7175 [53:53<30:24,  1.53it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s][eval] epoch 15.304 step 4392 loss 3.5247 (best 3.5139 @ ep 14.804, 2/20 since best)
+
+
                                               
 61%|██████    | 4393/7175 [53:53<1:15:36,  1.63s/it]
 61%|██████    | 4394/7175 [53:54<1:02:00,  1.34s/it]
 61%|██████▏   | 4395/7175 [53:55<52:32,  1.13s/it]  
 61%|██████▏   | 4396/7175 [53:55<45:51,  1.01it/s]
 61%|██████▏   | 4397/7175 [53:56<41:12,  1.12it/s]
 61%|██████▏   | 4398/7175 [53:57<37:54,  1.22it/s]
 61%|██████▏   | 4399/7175 [53:57<35:38,  1.30it/s]
 61%|██████▏   | 4400/7175 [53:58<34:01,  1.36it/s]
                                                   
{'loss': '3.071', 'grad_norm': '0.2295', 'learning_rate': '0.0006', 'epoch': '15.33'}
+
 61%|██████▏   | 4400/7175 [53:58<34:01,  1.36it/s]
 61%|██████▏   | 4401/7175 [53:59<32:54,  1.41it/s]
 61%|██████▏   | 4402/7175 [53:59<32:06,  1.44it/s]
 61%|██████▏   | 4403/7175 [54:00<31:32,  1.46it/s]
 61%|██████▏   | 4404/7175 [54:01<31:10,  1.48it/s]
 61%|██████▏   | 4405/7175 [54:01<30:53,  1.49it/s]
 61%|██████▏   | 4406/7175 [54:02<30:41,  1.50it/s]
 61%|██████▏   | 4407/7175 [54:03<30:32,  1.51it/s]
 61%|██████▏   | 4408/7175 [54:03<30:26,  1.51it/s]
 61%|██████▏   | 4409/7175 [54:04<30:22,  1.52it/s]
 61%|██████▏   | 4410/7175 [54:04<30:18,  1.52it/s]
                                                   
{'loss': '3.061', 'grad_norm': '0.2262', 'learning_rate': '0.0006', 'epoch': '15.37'}
+
 61%|██████▏   | 4410/7175 [54:04<30:18,  1.52it/s]
 61%|██████▏   | 4411/7175 [54:05<30:15,  1.52it/s]
 61%|██████▏   | 4412/7175 [54:06<30:13,  1.52it/s]
 62%|██████▏   | 4413/7175 [54:06<30:12,  1.52it/s]
 62%|██████▏   | 4414/7175 [54:07<30:10,  1.53it/s]
 62%|██████▏   | 4415/7175 [54:08<30:09,  1.53it/s]
 62%|██████▏   | 4416/7175 [54:08<30:08,  1.53it/s]
 62%|██████▏   | 4417/7175 [54:09<30:08,  1.52it/s]
 62%|██████▏   | 4418/7175 [54:10<30:08,  1.52it/s]
 62%|██████▏   | 4419/7175 [54:10<30:07,  1.52it/s]
 62%|██████▏   | 4420/7175 [54:11<30:05,  1.53it/s]
                                                   
{'loss': '3.052', 'grad_norm': '0.2301', 'learning_rate': '0.0006', 'epoch': '15.4'}
+
 62%|██████▏   | 4420/7175 [54:11<30:05,  1.53it/s]
 62%|██████▏   | 4421/7175 [54:12<30:05,  1.53it/s]
 62%|██████▏   | 4422/7175 [54:12<30:05,  1.52it/s]
 62%|██████▏   | 4423/7175 [54:13<30:03,  1.53it/s]
 62%|██████▏   | 4424/7175 [54:14<30:03,  1.53it/s]
 62%|██████▏   | 4425/7175 [54:14<30:01,  1.53it/s]
 62%|██████▏   | 4426/7175 [54:15<30:01,  1.53it/s]
 62%|██████▏   | 4427/7175 [54:16<30:00,  1.53it/s]
 62%|██████▏   | 4428/7175 [54:16<29:59,  1.53it/s]
 62%|██████▏   | 4429/7175 [54:17<29:58,  1.53it/s]
 62%|██████▏   | 4430/7175 [54:18<29:59,  1.53it/s]
                                                   
{'loss': '3.07', 'grad_norm': '0.2329', 'learning_rate': '0.0006', 'epoch': '15.44'}
+
 62%|██████▏   | 4430/7175 [54:18<29:59,  1.53it/s]
 62%|██████▏   | 4431/7175 [54:18<29:59,  1.52it/s]
 62%|██████▏   | 4432/7175 [54:19<29:58,  1.53it/s]
 62%|██████▏   | 4433/7175 [54:20<29:57,  1.53it/s]
 62%|██████▏   | 4434/7175 [54:20<29:56,  1.53it/s]
 62%|██████▏   | 4435/7175 [54:21<29:56,  1.52it/s]
 62%|██████▏   | 4436/7175 [54:22<29:55,  1.53it/s]
 62%|██████▏   | 4437/7175 [54:22<29:54,  1.53it/s]
 62%|██████▏   | 4438/7175 [54:23<29:53,  1.53it/s]
 62%|██████▏   | 4439/7175 [54:24<29:54,  1.52it/s]
 62%|██████▏   | 4440/7175 [54:24<29:53,  1.52it/s]
                                                   
{'loss': '3.059', 'grad_norm': '0.2373', 'learning_rate': '0.0006', 'epoch': '15.47'}
+
 62%|██████▏   | 4440/7175 [54:24<29:53,  1.52it/s]
 62%|██████▏   | 4441/7175 [54:25<29:52,  1.52it/s]
 62%|██████▏   | 4442/7175 [54:25<29:52,  1.52it/s]
 62%|██████▏   | 4443/7175 [54:26<29:51,  1.52it/s]
 62%|██████▏   | 4444/7175 [54:27<29:51,  1.52it/s]
 62%|██████▏   | 4445/7175 [54:27<29:50,  1.52it/s]
 62%|██████▏   | 4446/7175 [54:28<29:49,  1.52it/s]
 62%|██████▏   | 4447/7175 [54:29<29:49,  1.52it/s]
 62%|██████▏   | 4448/7175 [54:29<29:48,  1.52it/s]
 62%|██████▏   | 4449/7175 [54:30<29:48,  1.52it/s]
 62%|██████▏   | 4450/7175 [54:31<29:47,  1.52it/s]
                                                   
{'loss': '3.075', 'grad_norm': '0.2372', 'learning_rate': '0.0006', 'epoch': '15.51'}
+
 62%|██████▏   | 4450/7175 [54:31<29:47,  1.52it/s]
 62%|██████▏   | 4451/7175 [54:31<29:46,  1.52it/s]
 62%|██████▏   | 4452/7175 [54:32<29:46,  1.52it/s]
 62%|██████▏   | 4453/7175 [54:33<29:44,  1.53it/s]
 62%|██████▏   | 4454/7175 [54:33<29:45,  1.52it/s]
 62%|██████▏   | 4455/7175 [54:34<29:45,  1.52it/s]
 62%|██████▏   | 4456/7175 [54:35<29:43,  1.52it/s]
 62%|██████▏   | 4457/7175 [54:35<29:42,  1.52it/s]
 62%|██████▏   | 4458/7175 [54:36<29:41,  1.53it/s]
 62%|██████▏   | 4459/7175 [54:37<29:41,  1.52it/s]
 62%|██████▏   | 4460/7175 [54:37<29:40,  1.52it/s]
                                                   
{'loss': '3.079', 'grad_norm': '0.2228', 'learning_rate': '0.0006', 'epoch': '15.54'}
+
 62%|██████▏   | 4460/7175 [54:37<29:40,  1.52it/s]
 62%|██████▏   | 4461/7175 [54:38<29:40,  1.52it/s]
 62%|██████▏   | 4462/7175 [54:39<29:39,  1.52it/s]
 62%|██████▏   | 4463/7175 [54:39<29:38,  1.53it/s]
 62%|██████▏   | 4464/7175 [54:40<29:37,  1.52it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.47it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.79it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.61it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.06it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.76it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.59it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.45it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.38it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.31it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.29it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.28it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.27it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.25it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.24it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.23it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.23it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.22it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.23it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.22it/s]
                                                   
+
                                               
{'eval_loss': '3.513', 'eval_runtime': '3.264', 'eval_samples_per_second': '747.8', 'eval_steps_per_second': '11.95', 'epoch': '15.56'}
+
 62%|██████▏   | 4464/7175 [54:43<29:37,  1.52it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.22it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-4464
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-4464/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-4464/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.83it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.83it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-4464/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-4464/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-4464/tokenizer_config.json
+[eval] epoch 15.556 step 4464 loss 3.5129 (best 3.5129 @ ep 15.556, 0/20 since best)
+
+
                                               
 62%|██████▏   | 4465/7175 [54:46<1:42:26,  2.27s/it]
 62%|██████▏   | 4466/7175 [54:47<1:20:31,  1.78s/it]
 62%|██████▏   | 4467/7175 [54:47<1:05:12,  1.44s/it]
 62%|██████▏   | 4468/7175 [54:48<54:29,  1.21s/it]  
 62%|██████▏   | 4469/7175 [54:49<46:58,  1.04s/it]
 62%|██████▏   | 4470/7175 [54:50<46:29,  1.03s/it]
                                                   
{'loss': '3.081', 'grad_norm': '0.2299', 'learning_rate': '0.0006', 'epoch': '15.58'}
+
 62%|██████▏   | 4470/7175 [54:50<46:29,  1.03s/it]
 62%|██████▏   | 4471/7175 [54:50<41:23,  1.09it/s]
 62%|██████▏   | 4472/7175 [54:51<37:48,  1.19it/s]
 62%|██████▏   | 4473/7175 [54:52<35:18,  1.28it/s]
 62%|██████▏   | 4474/7175 [54:52<33:33,  1.34it/s]
 62%|██████▏   | 4475/7175 [54:53<32:19,  1.39it/s]
 62%|██████▏   | 4476/7175 [54:53<31:27,  1.43it/s]
 62%|██████▏   | 4477/7175 [54:54<30:52,  1.46it/s]
 62%|██████▏   | 4478/7175 [54:55<30:25,  1.48it/s]
 62%|██████▏   | 4479/7175 [54:55<30:07,  1.49it/s]
 62%|██████▏   | 4480/7175 [54:56<29:53,  1.50it/s]
                                                   
{'loss': '3.079', 'grad_norm': '0.2226', 'learning_rate': '0.0006', 'epoch': '15.61'}
+
 62%|██████▏   | 4480/7175 [54:56<29:53,  1.50it/s]
 62%|██████▏   | 4481/7175 [54:57<29:46,  1.51it/s]
 62%|██████▏   | 4482/7175 [54:57<29:40,  1.51it/s]
 62%|██████▏   | 4483/7175 [54:58<29:34,  1.52it/s]
 62%|██████▏   | 4484/7175 [54:59<29:29,  1.52it/s]
 63%|██████▎   | 4485/7175 [54:59<29:27,  1.52it/s]
 63%|██████▎   | 4486/7175 [55:00<29:26,  1.52it/s]
 63%|██████▎   | 4487/7175 [55:01<29:26,  1.52it/s]
 63%|██████▎   | 4488/7175 [55:01<29:24,  1.52it/s]
 63%|██████▎   | 4489/7175 [55:02<29:23,  1.52it/s]
 63%|██████▎   | 4490/7175 [55:03<29:21,  1.52it/s]
                                                   
{'loss': '3.063', 'grad_norm': '0.2506', 'learning_rate': '0.0006', 'epoch': '15.65'}
+
 63%|██████▎   | 4490/7175 [55:03<29:21,  1.52it/s]
 63%|██████▎   | 4491/7175 [55:03<29:22,  1.52it/s]
 63%|██████▎   | 4492/7175 [55:04<29:20,  1.52it/s]
 63%|██████▎   | 4493/7175 [55:05<29:19,  1.52it/s]
 63%|██████▎   | 4494/7175 [55:05<29:17,  1.53it/s]
 63%|██████▎   | 4495/7175 [55:06<29:17,  1.52it/s]
 63%|██████▎   | 4496/7175 [55:07<29:16,  1.52it/s]
 63%|██████▎   | 4497/7175 [55:07<29:14,  1.53it/s]
 63%|██████▎   | 4498/7175 [55:08<29:13,  1.53it/s]
 63%|██████▎   | 4499/7175 [55:09<29:13,  1.53it/s]
 63%|██████▎   | 4500/7175 [55:09<29:14,  1.53it/s]
                                                   
{'loss': '3.081', 'grad_norm': '0.2232', 'learning_rate': '0.0006', 'epoch': '15.68'}
+
 63%|██████▎   | 4500/7175 [55:09<29:14,  1.53it/s]
 63%|██████▎   | 4501/7175 [55:10<29:13,  1.53it/s]
 63%|██████▎   | 4502/7175 [55:11<29:11,  1.53it/s]
 63%|██████▎   | 4503/7175 [55:11<29:09,  1.53it/s]
 63%|██████▎   | 4504/7175 [55:12<29:10,  1.53it/s]
 63%|██████▎   | 4505/7175 [55:12<29:10,  1.53it/s]
 63%|██████▎   | 4506/7175 [55:13<29:09,  1.53it/s]
 63%|██████▎   | 4507/7175 [55:14<29:08,  1.53it/s]
 63%|██████▎   | 4508/7175 [55:14<29:08,  1.53it/s]
 63%|██████▎   | 4509/7175 [55:15<29:07,  1.53it/s]
 63%|██████▎   | 4510/7175 [55:16<29:06,  1.53it/s]
                                                   
{'loss': '3.085', 'grad_norm': '0.2203', 'learning_rate': '0.0006', 'epoch': '15.72'}
+
 63%|██████▎   | 4510/7175 [55:16<29:06,  1.53it/s]
 63%|██████▎   | 4511/7175 [55:16<29:06,  1.53it/s]
 63%|██████▎   | 4512/7175 [55:17<29:04,  1.53it/s]
 63%|██████▎   | 4513/7175 [55:18<29:04,  1.53it/s]
 63%|██████▎   | 4514/7175 [55:18<29:03,  1.53it/s]
 63%|██████▎   | 4515/7175 [55:19<29:02,  1.53it/s]
 63%|██████▎   | 4516/7175 [55:20<29:01,  1.53it/s]
 63%|██████▎   | 4517/7175 [55:20<29:02,  1.53it/s]
 63%|██████▎   | 4518/7175 [55:21<29:01,  1.53it/s]
 63%|██████▎   | 4519/7175 [55:22<29:00,  1.53it/s]
 63%|██████▎   | 4520/7175 [55:22<28:58,  1.53it/s]
                                                   
{'loss': '3.086', 'grad_norm': '0.2136', 'learning_rate': '0.0006', 'epoch': '15.75'}
+
 63%|██████▎   | 4520/7175 [55:22<28:58,  1.53it/s]
 63%|██████▎   | 4521/7175 [55:23<29:00,  1.53it/s]
 63%|██████▎   | 4522/7175 [55:24<28:59,  1.53it/s]
 63%|██████▎   | 4523/7175 [55:24<28:58,  1.53it/s]
 63%|██████▎   | 4524/7175 [55:25<28:56,  1.53it/s]
 63%|██████▎   | 4525/7175 [55:26<28:57,  1.53it/s]
 63%|██████▎   | 4526/7175 [55:26<28:56,  1.53it/s]
 63%|██████▎   | 4527/7175 [55:27<28:54,  1.53it/s]
 63%|██████▎   | 4528/7175 [55:28<28:53,  1.53it/s]
 63%|██████▎   | 4529/7175 [55:28<28:54,  1.53it/s]
 63%|██████▎   | 4530/7175 [55:29<28:53,  1.53it/s]
                                                   
{'loss': '3.077', 'grad_norm': '0.2111', 'learning_rate': '0.0006', 'epoch': '15.79'}
+
 63%|██████▎   | 4530/7175 [55:29<28:53,  1.53it/s]
 63%|██████▎   | 4531/7175 [55:30<28:52,  1.53it/s]
 63%|██████▎   | 4532/7175 [55:30<28:50,  1.53it/s]
 63%|██████▎   | 4533/7175 [55:31<28:50,  1.53it/s]
 63%|██████▎   | 4534/7175 [55:32<28:50,  1.53it/s]
 63%|██████▎   | 4535/7175 [55:32<28:49,  1.53it/s]
 63%|██████▎   | 4536/7175 [55:33<28:47,  1.53it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.93it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.96it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.67it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.09it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.78it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.58it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.47it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.39it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.33it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.30it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.30it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.27it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.26it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.24it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.24it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.17it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.19it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.19it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.21it/s]
                                                   
+
                                               
{'eval_loss': '3.51', 'eval_runtime': '3.259', 'eval_samples_per_second': '749', 'eval_steps_per_second': '11.97', 'epoch': '15.81'}
+
 63%|██████▎   | 4536/7175 [55:36<28:47,  1.53it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.21it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-4536
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-4536/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-4536/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.87it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.87it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-4536/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-4536/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-4536/tokenizer_config.json
+[eval] epoch 15.807 step 4536 loss 3.5102 (best 3.5102 @ ep 15.807, 0/20 since best)
+
+
                                               
 63%|██████▎   | 4537/7175 [55:39<1:39:13,  2.26s/it]
 63%|██████▎   | 4538/7175 [55:39<1:18:04,  1.78s/it]
 63%|██████▎   | 4539/7175 [55:40<1:03:16,  1.44s/it]
 63%|██████▎   | 4540/7175 [55:41<52:54,  1.20s/it]  
                                                   
{'loss': '3.089', 'grad_norm': '0.2502', 'learning_rate': '0.0006', 'epoch': '15.82'}
+
 63%|██████▎   | 4540/7175 [55:41<52:54,  1.20s/it]
 63%|██████▎   | 4541/7175 [55:41<45:38,  1.04s/it]
 63%|██████▎   | 4542/7175 [55:42<40:34,  1.08it/s]
 63%|██████▎   | 4543/7175 [55:43<37:00,  1.19it/s]
 63%|██████▎   | 4544/7175 [55:43<34:30,  1.27it/s]
 63%|██████▎   | 4545/7175 [55:44<32:46,  1.34it/s]
 63%|██████▎   | 4546/7175 [55:45<31:33,  1.39it/s]
 63%|██████▎   | 4547/7175 [55:45<30:41,  1.43it/s]
 63%|██████▎   | 4548/7175 [55:46<30:05,  1.46it/s]
 63%|██████▎   | 4549/7175 [55:47<29:39,  1.48it/s]
 63%|██████▎   | 4550/7175 [55:47<29:21,  1.49it/s]
                                                   
{'loss': '3.08', 'grad_norm': '0.2117', 'learning_rate': '0.0006', 'epoch': '15.86'}
+
 63%|██████▎   | 4550/7175 [55:47<29:21,  1.49it/s]
 63%|██████▎   | 4551/7175 [55:48<29:09,  1.50it/s]
 63%|██████▎   | 4552/7175 [55:49<28:59,  1.51it/s]
 63%|██████▎   | 4553/7175 [55:49<28:52,  1.51it/s]
 63%|██████▎   | 4554/7175 [55:50<28:46,  1.52it/s]
 63%|██████▎   | 4555/7175 [55:51<28:44,  1.52it/s]
 63%|██████▎   | 4556/7175 [55:51<28:41,  1.52it/s]
 64%|██████▎   | 4557/7175 [55:52<28:39,  1.52it/s]
 64%|██████▎   | 4558/7175 [55:53<28:36,  1.52it/s]
 64%|██████▎   | 4559/7175 [55:53<28:37,  1.52it/s]
 64%|██████▎   | 4560/7175 [55:54<28:35,  1.52it/s]
                                                   
{'loss': '3.091', 'grad_norm': '0.2159', 'learning_rate': '0.0006', 'epoch': '15.89'}
+
 64%|██████▎   | 4560/7175 [55:54<28:35,  1.52it/s]
 64%|██████▎   | 4561/7175 [55:55<28:34,  1.52it/s]
 64%|██████▎   | 4562/7175 [55:55<28:33,  1.52it/s]
 64%|██████▎   | 4563/7175 [55:56<28:32,  1.53it/s]
 64%|██████▎   | 4564/7175 [55:57<28:32,  1.52it/s]
 64%|██████▎   | 4565/7175 [55:57<28:31,  1.52it/s]
 64%|██████▎   | 4566/7175 [55:58<28:30,  1.53it/s]
 64%|██████▎   | 4567/7175 [55:58<28:28,  1.53it/s]
 64%|██████▎   | 4568/7175 [55:59<28:28,  1.53it/s]
 64%|██████▎   | 4569/7175 [56:00<28:37,  1.52it/s]
 64%|██████▎   | 4570/7175 [56:00<28:33,  1.52it/s]
                                                   
{'loss': '3.083', 'grad_norm': '0.2174', 'learning_rate': '0.0006', 'epoch': '15.93'}
+
 64%|██████▎   | 4570/7175 [56:00<28:33,  1.52it/s]
 64%|██████▎   | 4571/7175 [56:01<28:32,  1.52it/s]
 64%|██████▎   | 4572/7175 [56:02<28:31,  1.52it/s]
 64%|██████▎   | 4573/7175 [56:02<28:30,  1.52it/s]
 64%|██████▎   | 4574/7175 [56:03<28:28,  1.52it/s]
 64%|██████▍   | 4575/7175 [56:04<28:25,  1.52it/s]
 64%|██████▍   | 4576/7175 [56:04<28:25,  1.52it/s]
 64%|██████▍   | 4577/7175 [56:05<28:24,  1.52it/s]
 64%|██████▍   | 4578/7175 [56:06<28:22,  1.53it/s]
 64%|██████▍   | 4579/7175 [56:06<28:21,  1.53it/s]
 64%|██████▍   | 4580/7175 [56:07<28:21,  1.53it/s]
                                                   
{'loss': '3.093', 'grad_norm': '0.2098', 'learning_rate': '0.0006', 'epoch': '15.96'}
+
 64%|██████▍   | 4580/7175 [56:07<28:21,  1.53it/s]
 64%|██████▍   | 4581/7175 [56:08<28:20,  1.53it/s]
 64%|██████▍   | 4582/7175 [56:08<28:19,  1.53it/s]
 64%|██████▍   | 4583/7175 [56:09<28:17,  1.53it/s]
 64%|██████▍   | 4584/7175 [56:10<28:17,  1.53it/s]
 64%|██████▍   | 4585/7175 [56:10<28:16,  1.53it/s]
 64%|██████▍   | 4586/7175 [56:11<28:15,  1.53it/s]
 64%|██████▍   | 4587/7175 [56:12<28:14,  1.53it/s]
 64%|██████▍   | 4588/7175 [56:12<28:14,  1.53it/s]
 64%|██████▍   | 4589/7175 [56:13<28:21,  1.52it/s]
 64%|██████▍   | 4590/7175 [56:14<28:18,  1.52it/s]
                                                   
{'loss': '3.089', 'grad_norm': '0.2264', 'learning_rate': '0.0006', 'epoch': '16'}
+
 64%|██████▍   | 4590/7175 [56:14<28:18,  1.52it/s]
 64%|██████▍   | 4591/7175 [56:14<28:16,  1.52it/s]
 64%|██████▍   | 4593/7175 [56:15<23:27,  1.83it/s]
 64%|██████▍   | 4594/7175 [56:16<24:37,  1.75it/s]
 64%|██████▍   | 4595/7175 [56:16<25:33,  1.68it/s]
 64%|██████▍   | 4596/7175 [56:17<26:15,  1.64it/s]
 64%|██████▍   | 4597/7175 [56:18<26:47,  1.60it/s]
 64%|██████▍   | 4598/7175 [56:18<27:08,  1.58it/s]
 64%|██████▍   | 4599/7175 [56:19<27:26,  1.56it/s]
 64%|██████▍   | 4600/7175 [56:20<27:38,  1.55it/s]
                                                   
{'loss': '3.021', 'grad_norm': '0.2695', 'learning_rate': '0.0006', 'epoch': '16.03'}
+
 64%|██████▍   | 4600/7175 [56:20<27:38,  1.55it/s]
 64%|██████▍   | 4601/7175 [56:20<27:47,  1.54it/s]
 64%|██████▍   | 4602/7175 [56:21<27:52,  1.54it/s]
 64%|██████▍   | 4603/7175 [56:22<27:55,  1.53it/s]
 64%|██████▍   | 4604/7175 [56:22<27:58,  1.53it/s]
 64%|██████▍   | 4605/7175 [56:23<27:59,  1.53it/s]
 64%|██████▍   | 4606/7175 [56:24<28:00,  1.53it/s]
 64%|██████▍   | 4607/7175 [56:24<27:59,  1.53it/s]
 64%|██████▍   | 4608/7175 [56:25<28:01,  1.53it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.54it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.77it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.63it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.05it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.77it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.57it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.46it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.38it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.34it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.30it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.28it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.26it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.26it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.24it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.24it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.24it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.23it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.24it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s]
                                                   
+
                                               
{'eval_loss': '3.536', 'eval_runtime': '3.261', 'eval_samples_per_second': '748.5', 'eval_steps_per_second': '11.96', 'epoch': '16.06'}
+
 64%|██████▍   | 4608/7175 [56:28<28:01,  1.53it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s][eval] epoch 16.056 step 4608 loss 3.5364 (best 3.5102 @ ep 15.807, 1/20 since best)
+
+
                                               
 64%|██████▍   | 4609/7175 [56:29<1:09:48,  1.63s/it]
 64%|██████▍   | 4610/7175 [56:29<57:15,  1.34s/it]  
                                                   
{'loss': '3.017', 'grad_norm': '0.2521', 'learning_rate': '0.0006', 'epoch': '16.06'}
+
 64%|██████▍   | 4610/7175 [56:29<57:15,  1.34s/it]
 64%|██████▍   | 4611/7175 [56:30<48:27,  1.13s/it]
 64%|██████▍   | 4612/7175 [56:31<42:19,  1.01it/s]
 64%|██████▍   | 4613/7175 [56:31<38:00,  1.12it/s]
 64%|██████▍   | 4614/7175 [56:32<34:58,  1.22it/s]
 64%|██████▍   | 4615/7175 [56:33<32:51,  1.30it/s]
 64%|██████▍   | 4616/7175 [56:33<31:22,  1.36it/s]
 64%|██████▍   | 4617/7175 [56:34<30:21,  1.40it/s]
 64%|██████▍   | 4618/7175 [56:35<29:36,  1.44it/s]
 64%|██████▍   | 4619/7175 [56:35<29:06,  1.46it/s]
 64%|██████▍   | 4620/7175 [56:36<28:43,  1.48it/s]
                                                   
{'loss': '3.018', 'grad_norm': '0.2443', 'learning_rate': '0.0006', 'epoch': '16.1'}
+
 64%|██████▍   | 4620/7175 [56:36<28:43,  1.48it/s]
 64%|██████▍   | 4621/7175 [56:37<28:30,  1.49it/s]
 64%|██████▍   | 4622/7175 [56:37<28:20,  1.50it/s]
 64%|██████▍   | 4623/7175 [56:38<28:11,  1.51it/s]
 64%|██████▍   | 4624/7175 [56:39<28:05,  1.51it/s]
 64%|██████▍   | 4625/7175 [56:39<28:00,  1.52it/s]
 64%|██████▍   | 4626/7175 [56:40<27:57,  1.52it/s]
 64%|██████▍   | 4627/7175 [56:41<27:56,  1.52it/s]
 65%|██████▍   | 4628/7175 [56:41<27:53,  1.52it/s]
 65%|██████▍   | 4629/7175 [56:42<27:51,  1.52it/s]
 65%|██████▍   | 4630/7175 [56:43<27:50,  1.52it/s]
                                                   
{'loss': '3.027', 'grad_norm': '0.2394', 'learning_rate': '0.0006', 'epoch': '16.13'}
+
 65%|██████▍   | 4630/7175 [56:43<27:50,  1.52it/s]
 65%|██████▍   | 4631/7175 [56:43<27:49,  1.52it/s]
 65%|██████▍   | 4632/7175 [56:44<27:49,  1.52it/s]
 65%|██████▍   | 4633/7175 [56:45<27:47,  1.52it/s]
 65%|██████▍   | 4634/7175 [56:45<27:46,  1.53it/s]
 65%|██████▍   | 4635/7175 [56:46<27:44,  1.53it/s]
 65%|██████▍   | 4636/7175 [56:47<27:45,  1.52it/s]
 65%|██████▍   | 4637/7175 [56:47<27:44,  1.52it/s]
 65%|██████▍   | 4638/7175 [56:48<27:43,  1.52it/s]
 65%|██████▍   | 4639/7175 [56:48<27:42,  1.53it/s]
 65%|██████▍   | 4640/7175 [56:49<27:41,  1.53it/s]
                                                   
{'loss': '3.04', 'grad_norm': '0.2353', 'learning_rate': '0.0006', 'epoch': '16.17'}
+
 65%|██████▍   | 4640/7175 [56:49<27:41,  1.53it/s]
 65%|██████▍   | 4641/7175 [56:50<27:42,  1.52it/s]
 65%|██████▍   | 4642/7175 [56:50<27:41,  1.52it/s]
 65%|██████▍   | 4643/7175 [56:51<27:40,  1.53it/s]
 65%|██████▍   | 4644/7175 [56:52<27:38,  1.53it/s]
 65%|██████▍   | 4645/7175 [56:52<27:39,  1.52it/s]
 65%|██████▍   | 4646/7175 [56:53<27:39,  1.52it/s]
 65%|██████▍   | 4647/7175 [56:54<27:38,  1.52it/s]
 65%|██████▍   | 4648/7175 [56:54<27:37,  1.52it/s]
 65%|██████▍   | 4649/7175 [56:55<27:36,  1.53it/s]
 65%|██████▍   | 4650/7175 [56:56<27:36,  1.52it/s]
                                                   
{'loss': '3.023', 'grad_norm': '0.2189', 'learning_rate': '0.0006', 'epoch': '16.2'}
+
 65%|██████▍   | 4650/7175 [56:56<27:36,  1.52it/s]
 65%|██████▍   | 4651/7175 [56:56<27:36,  1.52it/s]
 65%|██████▍   | 4652/7175 [56:57<27:40,  1.52it/s]
 65%|██████▍   | 4653/7175 [56:58<27:39,  1.52it/s]
 65%|██████▍   | 4654/7175 [56:58<27:37,  1.52it/s]
 65%|██████▍   | 4655/7175 [56:59<27:35,  1.52it/s]
 65%|██████▍   | 4656/7175 [57:00<27:32,  1.52it/s]
 65%|██████▍   | 4657/7175 [57:00<27:32,  1.52it/s]
 65%|██████▍   | 4658/7175 [57:01<27:33,  1.52it/s]
 65%|██████▍   | 4659/7175 [57:02<27:35,  1.52it/s]
 65%|██████▍   | 4660/7175 [57:02<27:37,  1.52it/s]
                                                   
{'loss': '3.043', 'grad_norm': '0.2326', 'learning_rate': '0.0006', 'epoch': '16.24'}
+
 65%|██████▍   | 4660/7175 [57:02<27:37,  1.52it/s]
 65%|██████▍   | 4661/7175 [57:03<27:40,  1.51it/s]
 65%|██████▍   | 4662/7175 [57:04<27:36,  1.52it/s]
 65%|██████▍   | 4663/7175 [57:04<28:01,  1.49it/s]
 65%|██████▌   | 4664/7175 [57:05<29:20,  1.43it/s]
 65%|██████▌   | 4665/7175 [57:06<28:47,  1.45it/s]
 65%|██████▌   | 4666/7175 [57:06<28:22,  1.47it/s]
 65%|██████▌   | 4667/7175 [57:07<28:24,  1.47it/s]
 65%|██████▌   | 4668/7175 [57:08<28:11,  1.48it/s]
 65%|██████▌   | 4669/7175 [57:08<28:19,  1.47it/s]
 65%|██████▌   | 4670/7175 [57:09<28:03,  1.49it/s]
                                                   
{'loss': '3.039', 'grad_norm': '0.2179', 'learning_rate': '0.0006', 'epoch': '16.27'}
+
 65%|██████▌   | 4670/7175 [57:09<28:03,  1.49it/s]
 65%|██████▌   | 4671/7175 [57:10<27:53,  1.50it/s]
 65%|██████▌   | 4672/7175 [57:10<27:45,  1.50it/s]
 65%|██████▌   | 4673/7175 [57:11<27:41,  1.51it/s]
 65%|██████▌   | 4674/7175 [57:12<27:35,  1.51it/s]
 65%|██████▌   | 4675/7175 [57:12<27:31,  1.51it/s]
 65%|██████▌   | 4676/7175 [57:13<27:32,  1.51it/s]
 65%|██████▌   | 4677/7175 [57:14<27:32,  1.51it/s]
 65%|██████▌   | 4678/7175 [57:14<27:38,  1.51it/s]
 65%|██████▌   | 4679/7175 [57:15<27:34,  1.51it/s]
 65%|██████▌   | 4680/7175 [57:16<27:34,  1.51it/s]
                                                   
{'loss': '3.042', 'grad_norm': '0.2253', 'learning_rate': '0.0006', 'epoch': '16.31'}
+
 65%|██████▌   | 4680/7175 [57:16<27:34,  1.51it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:02, 16.84it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 13.68it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 12.93it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 12.66it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.50it/s]
+
 33%|███▎      | 13/39 [00:01<00:02, 12.38it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.19it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.15it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.14it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 11.71it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 11.84it/s]
+
 64%|██████▍   | 25/39 [00:02<00:01, 11.96it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.00it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.08it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.12it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.18it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.20it/s]
+
 95%|█████████▍| 37/39 [00:03<00:00, 12.22it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.21it/s]
                                                   
+
                                               
{'eval_loss': '3.524', 'eval_runtime': '3.307', 'eval_samples_per_second': '738.2', 'eval_steps_per_second': '11.79', 'epoch': '16.31'}
+
 65%|██████▌   | 4680/7175 [57:19<27:34,  1.51it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.21it/s][eval] epoch 16.308 step 4680 loss 3.5241 (best 3.5102 @ ep 15.807, 2/20 since best)
+
+
                                               
 65%|██████▌   | 4681/7175 [57:20<1:08:45,  1.65s/it]
 65%|██████▌   | 4682/7175 [57:20<56:18,  1.36s/it]  
 65%|██████▌   | 4683/7175 [57:21<47:54,  1.15s/it]
 65%|██████▌   | 4684/7175 [57:22<41:45,  1.01s/it]
 65%|██████▌   | 4685/7175 [57:22<37:24,  1.11it/s]
 65%|██████▌   | 4686/7175 [57:23<34:41,  1.20it/s]
 65%|██████▌   | 4687/7175 [57:24<32:28,  1.28it/s]
 65%|██████▌   | 4688/7175 [57:25<35:48,  1.16it/s]
 65%|██████▌   | 4689/7175 [57:25<33:18,  1.24it/s]
 65%|██████▌   | 4690/7175 [57:26<31:42,  1.31it/s]
                                                   
{'loss': '3.059', 'grad_norm': '0.2648', 'learning_rate': '0.0006', 'epoch': '16.34'}
+
 65%|██████▌   | 4690/7175 [57:26<31:42,  1.31it/s]
 65%|██████▌   | 4691/7175 [57:27<30:41,  1.35it/s]
 65%|██████▌   | 4692/7175 [57:27<30:01,  1.38it/s]
 65%|██████▌   | 4693/7175 [57:28<34:10,  1.21it/s]
 65%|██████▌   | 4694/7175 [57:30<41:37,  1.01s/it]
 65%|██████▌   | 4695/7175 [57:31<47:26,  1.15s/it]
 65%|██████▌   | 4696/7175 [57:33<51:21,  1.24s/it]
 65%|██████▌   | 4697/7175 [57:34<53:58,  1.31s/it]
 65%|██████▌   | 4698/7175 [57:36<55:57,  1.36s/it]
 65%|██████▌   | 4699/7175 [57:37<57:23,  1.39s/it]
 66%|██████▌   | 4700/7175 [57:39<58:22,  1.41s/it]
                                                   
{'loss': '3.038', 'grad_norm': '0.2452', 'learning_rate': '0.0006', 'epoch': '16.38'}
+
 66%|██████▌   | 4700/7175 [57:39<58:22,  1.41s/it]
 66%|██████▌   | 4701/7175 [57:40<55:15,  1.34s/it]
 66%|██████▌   | 4702/7175 [57:41<46:44,  1.13s/it]
 66%|██████▌   | 4703/7175 [57:41<40:48,  1.01it/s]
 66%|██████▌   | 4704/7175 [57:42<36:39,  1.12it/s]
 66%|██████▌   | 4705/7175 [57:42<33:45,  1.22it/s]
 66%|██████▌   | 4706/7175 [57:43<31:43,  1.30it/s]
 66%|██████▌   | 4707/7175 [57:44<34:38,  1.19it/s]
 66%|██████▌   | 4708/7175 [57:45<32:22,  1.27it/s]
 66%|██████▌   | 4709/7175 [57:45<30:44,  1.34it/s]
 66%|██████▌   | 4710/7175 [57:46<29:36,  1.39it/s]
                                                   
{'loss': '3.053', 'grad_norm': '0.2244', 'learning_rate': '0.0006', 'epoch': '16.41'}
+
 66%|██████▌   | 4710/7175 [57:46<29:36,  1.39it/s]
 66%|██████▌   | 4711/7175 [57:47<28:48,  1.43it/s]
 66%|██████▌   | 4712/7175 [57:47<28:19,  1.45it/s]
 66%|██████▌   | 4713/7175 [57:48<28:45,  1.43it/s]
 66%|██████▌   | 4714/7175 [57:49<28:18,  1.45it/s]
 66%|██████▌   | 4715/7175 [57:50<27:55,  1.47it/s]
 66%|██████▌   | 4716/7175 [57:50<27:40,  1.48it/s]
 66%|██████▌   | 4717/7175 [57:51<31:59,  1.28it/s]
 66%|██████▌   | 4718/7175 [57:52<30:25,  1.35it/s]
 66%|██████▌   | 4719/7175 [57:53<29:22,  1.39it/s]
 66%|██████▌   | 4720/7175 [57:53<28:38,  1.43it/s]
                                                   
{'loss': '3.06', 'grad_norm': '0.2074', 'learning_rate': '0.0006', 'epoch': '16.45'}
+
 66%|██████▌   | 4720/7175 [57:53<28:38,  1.43it/s]
 66%|██████▌   | 4721/7175 [57:54<28:06,  1.45it/s]
 66%|██████▌   | 4722/7175 [57:54<27:46,  1.47it/s]
 66%|██████▌   | 4723/7175 [57:55<31:53,  1.28it/s]
 66%|██████▌   | 4724/7175 [57:56<30:21,  1.35it/s]
 66%|██████▌   | 4725/7175 [57:57<29:17,  1.39it/s]
 66%|██████▌   | 4726/7175 [57:57<28:32,  1.43it/s]
 66%|██████▌   | 4727/7175 [57:58<28:08,  1.45it/s]
 66%|██████▌   | 4728/7175 [57:59<27:49,  1.47it/s]
 66%|██████▌   | 4729/7175 [57:59<27:34,  1.48it/s]
 66%|██████▌   | 4730/7175 [58:00<27:41,  1.47it/s]
                                                   
{'loss': '3.055', 'grad_norm': '0.279', 'learning_rate': '0.0006', 'epoch': '16.48'}
+
 66%|██████▌   | 4730/7175 [58:00<27:41,  1.47it/s]
 66%|██████▌   | 4731/7175 [58:01<27:44,  1.47it/s]
 66%|██████▌   | 4732/7175 [58:02<32:34,  1.25it/s]
 66%|██████▌   | 4733/7175 [58:03<35:44,  1.14it/s]
 66%|██████▌   | 4734/7175 [58:04<33:13,  1.22it/s]
 66%|██████▌   | 4735/7175 [58:04<31:15,  1.30it/s]
 66%|██████▌   | 4736/7175 [58:05<29:54,  1.36it/s]
 66%|██████▌   | 4737/7175 [58:06<28:59,  1.40it/s]
 66%|██████▌   | 4738/7175 [58:06<28:19,  1.43it/s]
 66%|██████▌   | 4739/7175 [58:07<27:50,  1.46it/s]
 66%|██████▌   | 4740/7175 [58:08<27:30,  1.48it/s]
                                                   
{'loss': '3.055', 'grad_norm': '0.2178', 'learning_rate': '0.0006', 'epoch': '16.52'}
+
 66%|██████▌   | 4740/7175 [58:08<27:30,  1.48it/s]
 66%|██████▌   | 4741/7175 [58:08<27:16,  1.49it/s]
 66%|██████▌   | 4742/7175 [58:09<27:04,  1.50it/s]
 66%|██████▌   | 4743/7175 [58:10<26:59,  1.50it/s]
 66%|██████▌   | 4744/7175 [58:10<26:54,  1.51it/s]
 66%|██████▌   | 4745/7175 [58:11<26:51,  1.51it/s]
 66%|██████▌   | 4746/7175 [58:12<26:49,  1.51it/s]
 66%|██████▌   | 4747/7175 [58:12<26:47,  1.51it/s]
 66%|██████▌   | 4748/7175 [58:13<26:45,  1.51it/s]
 66%|██████▌   | 4749/7175 [58:14<26:44,  1.51it/s]
 66%|██████▌   | 4750/7175 [58:14<26:45,  1.51it/s]
                                                   
{'loss': '3.053', 'grad_norm': '0.2077', 'learning_rate': '0.0006', 'epoch': '16.55'}
+
 66%|██████▌   | 4750/7175 [58:14<26:45,  1.51it/s]
 66%|██████▌   | 4751/7175 [58:15<26:47,  1.51it/s]
 66%|██████▌   | 4752/7175 [58:16<26:48,  1.51it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:02, 17.31it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.34it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.28it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 12.86it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.58it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.42it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.32it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.26it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.21it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.17it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.16it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.04it/s]
+
 69%|██████▉   | 27/39 [00:02<00:01, 11.86it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 11.73it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 11.53it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 11.38it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 11.53it/s]
+
 95%|█████████▍| 37/39 [00:03<00:00, 11.66it/s]
+
100%|██████████| 39/39 [00:03<00:00, 11.74it/s]
                                                   
+
                                               
{'eval_loss': '3.513', 'eval_runtime': '3.365', 'eval_samples_per_second': '725.3', 'eval_steps_per_second': '11.59', 'epoch': '16.56'}
+
 66%|██████▌   | 4752/7175 [58:19<26:48,  1.51it/s]
+
100%|██████████| 39/39 [00:03<00:00, 11.74it/s][eval] epoch 16.559 step 4752 loss 3.5131 (best 3.5102 @ ep 15.807, 3/20 since best)
+
+
                                               
 66%|██████▌   | 4753/7175 [58:20<1:07:33,  1.67s/it]
 66%|██████▋   | 4754/7175 [58:20<55:16,  1.37s/it]  
 66%|██████▋   | 4755/7175 [58:21<51:08,  1.27s/it]
 66%|██████▋   | 4756/7175 [58:22<43:48,  1.09s/it]
 66%|██████▋   | 4757/7175 [58:23<38:40,  1.04it/s]
 66%|██████▋   | 4758/7175 [58:23<35:02,  1.15it/s]
 66%|██████▋   | 4759/7175 [58:24<32:30,  1.24it/s]
 66%|██████▋   | 4760/7175 [58:25<30:43,  1.31it/s]
                                                   
{'loss': '3.062', 'grad_norm': '0.2347', 'learning_rate': '0.0006', 'epoch': '16.59'}
+
 66%|██████▋   | 4760/7175 [58:25<30:43,  1.31it/s]
 66%|██████▋   | 4761/7175 [58:26<35:42,  1.13it/s]
 66%|██████▋   | 4762/7175 [58:27<41:53,  1.04s/it]
 66%|██████▋   | 4763/7175 [58:29<46:22,  1.15s/it]
 66%|██████▋   | 4764/7175 [58:30<49:35,  1.23s/it]
 66%|██████▋   | 4765/7175 [58:31<51:48,  1.29s/it]
 66%|██████▋   | 4766/7175 [58:33<54:00,  1.35s/it]
 66%|██████▋   | 4767/7175 [58:34<50:11,  1.25s/it]
 66%|██████▋   | 4768/7175 [58:35<48:34,  1.21s/it]
 66%|██████▋   | 4769/7175 [58:36<41:52,  1.04s/it]
 66%|██████▋   | 4770/7175 [58:36<37:24,  1.07it/s]
                                                   
{'loss': '3.066', 'grad_norm': '0.2055', 'learning_rate': '0.0006', 'epoch': '16.62'}
+
 66%|██████▋   | 4770/7175 [58:36<37:24,  1.07it/s]
 66%|██████▋   | 4771/7175 [58:37<39:10,  1.02it/s]
 67%|██████▋   | 4772/7175 [58:38<36:43,  1.09it/s]
 67%|██████▋   | 4773/7175 [58:39<39:51,  1.00it/s]
 67%|██████▋   | 4774/7175 [58:41<41:29,  1.04s/it]
 67%|██████▋   | 4775/7175 [58:42<46:48,  1.17s/it]
 67%|██████▋   | 4776/7175 [58:44<54:17,  1.36s/it]
 67%|██████▋   | 4777/7175 [58:45<55:46,  1.40s/it]
 67%|██████▋   | 4778/7175 [58:47<56:47,  1.42s/it]
 67%|██████▋   | 4779/7175 [58:48<57:24,  1.44s/it]
 67%|██████▋   | 4780/7175 [58:50<57:50,  1.45s/it]
                                                   
{'loss': '3.077', 'grad_norm': '0.2177', 'learning_rate': '0.0006', 'epoch': '16.66'}
+
 67%|██████▋   | 4780/7175 [58:50<57:50,  1.45s/it]
 67%|██████▋   | 4781/7175 [58:51<58:28,  1.47s/it]
 67%|██████▋   | 4782/7175 [58:53<58:29,  1.47s/it]
 67%|██████▋   | 4783/7175 [58:54<58:38,  1.47s/it]
 67%|██████▋   | 4784/7175 [58:56<58:40,  1.47s/it]
 67%|██████▋   | 4785/7175 [58:57<58:43,  1.47s/it]
 67%|██████▋   | 4786/7175 [58:59<58:44,  1.48s/it]
 67%|██████▋   | 4787/7175 [59:00<58:44,  1.48s/it]
 67%|██████▋   | 4788/7175 [59:02<58:46,  1.48s/it]
 67%|██████▋   | 4789/7175 [59:03<1:02:34,  1.57s/it]
 67%|██████▋   | 4790/7175 [59:05<1:01:24,  1.54s/it]
                                                     
{'loss': '3.055', 'grad_norm': '0.2225', 'learning_rate': '0.0006', 'epoch': '16.69'}
+
 67%|██████▋   | 4790/7175 [59:05<1:01:24,  1.54s/it]
 67%|██████▋   | 4791/7175 [59:06<1:00:55,  1.53s/it]
 67%|██████▋   | 4792/7175 [59:08<1:00:18,  1.52s/it]
 67%|██████▋   | 4793/7175 [59:09<59:47,  1.51s/it]  
 67%|██████▋   | 4794/7175 [59:11<59:27,  1.50s/it]
 67%|██████▋   | 4795/7175 [59:12<59:07,  1.49s/it]
 67%|██████▋   | 4796/7175 [59:14<58:59,  1.49s/it]
 67%|██████▋   | 4797/7175 [59:15<58:55,  1.49s/it]
 67%|██████▋   | 4798/7175 [59:17<58:46,  1.48s/it]
 67%|██████▋   | 4799/7175 [59:18<58:30,  1.48s/it]
 67%|██████▋   | 4800/7175 [59:20<58:32,  1.48s/it]
                                                   
{'loss': '3.064', 'grad_norm': '0.2159', 'learning_rate': '0.0006', 'epoch': '16.73'}
+
 67%|██████▋   | 4800/7175 [59:20<58:32,  1.48s/it]
 67%|██████▋   | 4801/7175 [59:21<58:48,  1.49s/it]
 67%|██████▋   | 4802/7175 [59:23<58:44,  1.49s/it]
 67%|██████▋   | 4803/7175 [59:24<58:35,  1.48s/it]
 67%|██████▋   | 4804/7175 [59:26<58:31,  1.48s/it]
 67%|██████▋   | 4805/7175 [59:27<58:33,  1.48s/it]
 67%|██████▋   | 4806/7175 [59:29<58:33,  1.48s/it]
 67%|██████▋   | 4807/7175 [59:30<58:23,  1.48s/it]
 67%|██████▋   | 4808/7175 [59:32<58:22,  1.48s/it]
 67%|██████▋   | 4809/7175 [59:33<58:21,  1.48s/it]
 67%|██████▋   | 4810/7175 [59:34<58:34,  1.49s/it]
                                                   
{'loss': '3.062', 'grad_norm': '0.2316', 'learning_rate': '0.0006', 'epoch': '16.76'}
+
 67%|██████▋   | 4810/7175 [59:35<58:34,  1.49s/it]
 67%|██████▋   | 4811/7175 [59:36<58:47,  1.49s/it]
 67%|██████▋   | 4812/7175 [59:37<58:36,  1.49s/it]
 67%|██████▋   | 4813/7175 [59:39<58:28,  1.49s/it]
 67%|██████▋   | 4814/7175 [59:40<58:20,  1.48s/it]
 67%|██████▋   | 4815/7175 [59:42<58:19,  1.48s/it]
 67%|██████▋   | 4816/7175 [59:43<58:20,  1.48s/it]
 67%|██████▋   | 4817/7175 [59:45<58:11,  1.48s/it]
 67%|██████▋   | 4818/7175 [59:46<58:09,  1.48s/it]
 67%|██████▋   | 4819/7175 [59:48<58:11,  1.48s/it]
 67%|██████▋   | 4820/7175 [59:49<58:07,  1.48s/it]
                                                   
{'loss': '3.078', 'grad_norm': '0.2139', 'learning_rate': '0.0006', 'epoch': '16.8'}
+
 67%|██████▋   | 4820/7175 [59:49<58:07,  1.48s/it]
 67%|██████▋   | 4821/7175 [59:51<58:15,  1.48s/it]
 67%|██████▋   | 4822/7175 [59:52<58:10,  1.48s/it]
 67%|██████▋   | 4823/7175 [59:54<58:13,  1.49s/it]
 67%|██████▋   | 4824/7175 [59:55<58:03,  1.48s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:03, 10.13it/s]
+
 10%|█         | 4/39 [00:00<00:05,  6.32it/s]
+
 13%|█▎        | 5/39 [00:00<00:05,  6.01it/s]
+
 15%|█▌        | 6/39 [00:00<00:05,  5.82it/s]
+
 18%|█▊        | 7/39 [00:01<00:05,  5.69it/s]
+
 21%|██        | 8/39 [00:01<00:05,  5.52it/s]
+
 23%|██▎       | 9/39 [00:01<00:05,  5.43it/s]
+
 26%|██▌       | 10/39 [00:01<00:05,  5.39it/s]
+
 28%|██▊       | 11/39 [00:01<00:05,  5.34it/s]
+
 31%|███       | 12/39 [00:02<00:05,  5.36it/s]
+
 33%|███▎      | 13/39 [00:02<00:04,  5.41it/s]
+
 36%|███▌      | 14/39 [00:02<00:04,  5.35it/s]
+
 38%|███▊      | 15/39 [00:02<00:04,  5.35it/s]
+
 41%|████      | 16/39 [00:02<00:04,  5.27it/s]
+
 44%|████▎     | 17/39 [00:03<00:04,  5.28it/s]
+
 46%|████▌     | 18/39 [00:03<00:03,  5.30it/s]
+
 49%|████▊     | 19/39 [00:03<00:03,  5.31it/s]
+
 51%|█████▏    | 20/39 [00:03<00:03,  5.26it/s]
+
 54%|█████▍    | 21/39 [00:03<00:03,  5.26it/s]
+
 56%|█████▋    | 22/39 [00:03<00:03,  5.24it/s]
+
 59%|█████▉    | 23/39 [00:04<00:03,  5.27it/s]
+
 62%|██████▏   | 24/39 [00:04<00:02,  5.29it/s]
+
 64%|██████▍   | 25/39 [00:04<00:02,  5.26it/s]
+
 67%|██████▋   | 26/39 [00:04<00:02,  5.17it/s]
+
 69%|██████▉   | 27/39 [00:04<00:02,  5.18it/s]
+
 72%|███████▏  | 28/39 [00:05<00:02,  5.11it/s]
+
 74%|███████▍  | 29/39 [00:05<00:01,  5.07it/s]
+
 77%|███████▋  | 30/39 [00:05<00:01,  5.05it/s]
+
 79%|███████▉  | 31/39 [00:05<00:01,  5.14it/s]
+
 82%|████████▏ | 32/39 [00:05<00:01,  5.09it/s]
+
 85%|████████▍ | 33/39 [00:06<00:01,  5.08it/s]
+
 87%|████████▋ | 34/39 [00:06<00:00,  5.11it/s]
+
 90%|████████▉ | 35/39 [00:06<00:00,  5.17it/s]
+
 92%|█████████▏| 36/39 [00:06<00:00,  5.18it/s]
+
 95%|█████████▍| 37/39 [00:06<00:00,  5.24it/s]
+
 97%|█████████▋| 38/39 [00:07<00:00,  5.22it/s]
+
100%|██████████| 39/39 [00:07<00:00,  5.26it/s]
                                                   
+
                                               
{'eval_loss': '3.509', 'eval_runtime': '7.53', 'eval_samples_per_second': '324.2', 'eval_steps_per_second': '5.179', 'epoch': '16.81'}
+
 67%|██████▋   | 4824/7175 [1:00:03<58:03,  1.48s/it]
+
100%|██████████| 39/39 [00:07<00:00,  5.26it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-4824
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-4824/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-4824/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.67it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.67it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-4824/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-4824/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-4824/tokenizer_config.json
+[eval] epoch 16.811 step 4824 loss 3.5092 (best 3.5092 @ ep 16.811, 0/20 since best)
+
+
                                               
 67%|██████▋   | 4825/7175 [1:00:06<2:51:29,  4.38s/it]
 67%|██████▋   | 4826/7175 [1:00:08<2:17:22,  3.51s/it]
 67%|██████▋   | 4827/7175 [1:00:09<1:53:29,  2.90s/it]
 67%|██████▋   | 4828/7175 [1:00:11<1:36:36,  2.47s/it]
 67%|██████▋   | 4829/7175 [1:00:12<1:24:53,  2.17s/it]
 67%|██████▋   | 4830/7175 [1:00:14<1:16:43,  1.96s/it]
                                                       
{'loss': '3.062', 'grad_norm': '0.2299', 'learning_rate': '0.0006', 'epoch': '16.83'}
+
 67%|██████▋   | 4830/7175 [1:00:14<1:16:43,  1.96s/it]
 67%|██████▋   | 4831/7175 [1:00:15<1:10:45,  1.81s/it]
 67%|██████▋   | 4832/7175 [1:00:17<1:06:49,  1.71s/it]
 67%|██████▋   | 4833/7175 [1:00:18<1:04:05,  1.64s/it]
 67%|██████▋   | 4834/7175 [1:00:20<1:02:10,  1.59s/it]
 67%|██████▋   | 4835/7175 [1:00:21<1:00:55,  1.56s/it]
 67%|██████▋   | 4836/7175 [1:00:23<59:57,  1.54s/it]  
 67%|██████▋   | 4837/7175 [1:00:24<59:10,  1.52s/it]
 67%|██████▋   | 4838/7175 [1:00:26<58:34,  1.50s/it]
 67%|██████▋   | 4839/7175 [1:00:27<58:22,  1.50s/it]
 67%|██████▋   | 4840/7175 [1:00:29<58:12,  1.50s/it]
                                                     
{'loss': '3.065', 'grad_norm': '0.1998', 'learning_rate': '0.0006', 'epoch': '16.87'}
+
 67%|██████▋   | 4840/7175 [1:00:29<58:12,  1.50s/it]
 67%|██████▋   | 4841/7175 [1:00:30<58:12,  1.50s/it]
 67%|██████▋   | 4842/7175 [1:00:32<57:58,  1.49s/it]
 67%|██████▋   | 4843/7175 [1:00:33<57:33,  1.48s/it]
 68%|██████▊   | 4844/7175 [1:00:34<57:26,  1.48s/it]
 68%|██████▊   | 4845/7175 [1:00:36<57:26,  1.48s/it]
 68%|██████▊   | 4846/7175 [1:00:37<57:24,  1.48s/it]
 68%|██████▊   | 4847/7175 [1:00:39<57:18,  1.48s/it]
 68%|██████▊   | 4848/7175 [1:00:40<57:17,  1.48s/it]
 68%|██████▊   | 4849/7175 [1:00:42<57:17,  1.48s/it]
 68%|██████▊   | 4850/7175 [1:00:43<57:15,  1.48s/it]
                                                     
{'loss': '3.081', 'grad_norm': '0.2108', 'learning_rate': '0.0006', 'epoch': '16.9'}
+
 68%|██████▊   | 4850/7175 [1:00:43<57:15,  1.48s/it]
 68%|██████▊   | 4851/7175 [1:00:45<57:26,  1.48s/it]
 68%|██████▊   | 4852/7175 [1:00:46<57:20,  1.48s/it]
 68%|██████▊   | 4853/7175 [1:00:48<57:19,  1.48s/it]
 68%|██████▊   | 4854/7175 [1:00:49<57:14,  1.48s/it]
 68%|██████▊   | 4855/7175 [1:00:51<57:11,  1.48s/it]
 68%|██████▊   | 4856/7175 [1:00:52<57:18,  1.48s/it]
 68%|██████▊   | 4857/7175 [1:00:54<57:11,  1.48s/it]
 68%|██████▊   | 4858/7175 [1:00:55<57:18,  1.48s/it]
 68%|██████▊   | 4859/7175 [1:00:57<57:09,  1.48s/it]
 68%|██████▊   | 4860/7175 [1:00:58<57:05,  1.48s/it]
                                                     
{'loss': '3.056', 'grad_norm': '0.2048', 'learning_rate': '0.0006', 'epoch': '16.94'}
+
 68%|██████▊   | 4860/7175 [1:00:58<57:05,  1.48s/it]
 68%|██████▊   | 4861/7175 [1:01:00<57:17,  1.49s/it]
 68%|██████▊   | 4862/7175 [1:01:01<57:11,  1.48s/it]
 68%|██████▊   | 4863/7175 [1:01:03<57:03,  1.48s/it]
 68%|██████▊   | 4864/7175 [1:01:04<57:00,  1.48s/it]
 68%|██████▊   | 4865/7175 [1:01:06<56:57,  1.48s/it]
 68%|██████▊   | 4866/7175 [1:01:07<56:57,  1.48s/it]
 68%|██████▊   | 4867/7175 [1:01:09<56:52,  1.48s/it]
 68%|██████▊   | 4868/7175 [1:01:10<56:59,  1.48s/it]
 68%|██████▊   | 4869/7175 [1:01:11<56:55,  1.48s/it]
 68%|██████▊   | 4870/7175 [1:01:13<56:52,  1.48s/it]
                                                     
{'loss': '3.07', 'grad_norm': '0.2215', 'learning_rate': '0.0006', 'epoch': '16.97'}
+
 68%|██████▊   | 4870/7175 [1:01:13<56:52,  1.48s/it]
 68%|██████▊   | 4871/7175 [1:01:14<57:04,  1.49s/it]
 68%|██████▊   | 4872/7175 [1:01:16<56:59,  1.48s/it]
 68%|██████▊   | 4873/7175 [1:01:17<56:57,  1.48s/it]
 68%|██████▊   | 4874/7175 [1:01:19<56:57,  1.49s/it]
 68%|██████▊   | 4875/7175 [1:01:20<56:50,  1.48s/it]
 68%|██████▊   | 4876/7175 [1:01:22<56:50,  1.48s/it]
 68%|██████▊   | 4877/7175 [1:01:23<56:50,  1.48s/it]
 68%|██████▊   | 4878/7175 [1:01:25<56:44,  1.48s/it]
 68%|██████▊   | 4879/7175 [1:01:25<42:17,  1.11s/it]
 68%|██████▊   | 4880/7175 [1:01:27<47:39,  1.25s/it]
                                                     
{'loss': '3.045', 'grad_norm': '0.3039', 'learning_rate': '0.0006', 'epoch': '17'}
+
 68%|██████▊   | 4880/7175 [1:01:27<47:39,  1.25s/it]
 68%|██████▊   | 4881/7175 [1:01:28<50:28,  1.32s/it]
 68%|██████▊   | 4882/7175 [1:01:30<52:11,  1.37s/it]
 68%|██████▊   | 4883/7175 [1:01:31<53:30,  1.40s/it]
 68%|██████▊   | 4884/7175 [1:01:33<54:23,  1.42s/it]
 68%|██████▊   | 4885/7175 [1:01:34<55:01,  1.44s/it]
 68%|██████▊   | 4886/7175 [1:01:36<55:14,  1.45s/it]
 68%|██████▊   | 4887/7175 [1:01:37<55:32,  1.46s/it]
 68%|██████▊   | 4888/7175 [1:01:38<55:45,  1.46s/it]
 68%|██████▊   | 4889/7175 [1:01:40<55:59,  1.47s/it]
 68%|██████▊   | 4890/7175 [1:01:41<56:04,  1.47s/it]
                                                     
{'loss': '2.997', 'grad_norm': '0.2576', 'learning_rate': '0.0006', 'epoch': '17.04'}
+
 68%|██████▊   | 4890/7175 [1:01:41<56:04,  1.47s/it]
 68%|██████▊   | 4891/7175 [1:01:43<56:19,  1.48s/it]
 68%|██████▊   | 4892/7175 [1:01:44<56:20,  1.48s/it]
 68%|██████▊   | 4893/7175 [1:01:46<56:12,  1.48s/it]
 68%|██████▊   | 4894/7175 [1:01:47<56:12,  1.48s/it]
 68%|██████▊   | 4895/7175 [1:01:49<56:10,  1.48s/it]
 68%|██████▊   | 4896/7175 [1:01:50<56:07,  1.48s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:03, 10.63it/s]
+
 10%|█         | 4/39 [00:00<00:05,  6.56it/s]
+
 13%|█▎        | 5/39 [00:00<00:05,  6.16it/s]
+
 15%|█▌        | 6/39 [00:00<00:05,  5.85it/s]
+
 18%|█▊        | 7/39 [00:01<00:05,  5.70it/s]
+
 21%|██        | 8/39 [00:01<00:05,  5.54it/s]
+
 23%|██▎       | 9/39 [00:01<00:05,  5.47it/s]
+
 26%|██▌       | 10/39 [00:01<00:05,  5.42it/s]
+
 28%|██▊       | 11/39 [00:01<00:05,  5.35it/s]
+
 31%|███       | 12/39 [00:02<00:05,  5.29it/s]
+
 33%|███▎      | 13/39 [00:02<00:04,  5.36it/s]
+
 36%|███▌      | 14/39 [00:02<00:04,  5.37it/s]
+
 38%|███▊      | 15/39 [00:02<00:04,  5.36it/s]
+
 41%|████      | 16/39 [00:02<00:04,  5.34it/s]
+
 44%|████▎     | 17/39 [00:03<00:04,  5.39it/s]
+
 46%|████▌     | 18/39 [00:03<00:03,  5.45it/s]
+
 49%|████▊     | 19/39 [00:03<00:03,  5.44it/s]
+
 51%|█████▏    | 20/39 [00:03<00:03,  5.39it/s]
+
 54%|█████▍    | 21/39 [00:03<00:03,  5.37it/s]
+
 56%|█████▋    | 22/39 [00:03<00:03,  5.36it/s]
+
 59%|█████▉    | 23/39 [00:04<00:03,  5.31it/s]
+
 62%|██████▏   | 24/39 [00:04<00:02,  5.32it/s]
+
 64%|██████▍   | 25/39 [00:04<00:02,  5.28it/s]
+
 67%|██████▋   | 26/39 [00:04<00:02,  5.26it/s]
+
 69%|██████▉   | 27/39 [00:04<00:02,  5.27it/s]
+
 72%|███████▏  | 28/39 [00:05<00:02,  5.28it/s]
+
 74%|███████▍  | 29/39 [00:05<00:01,  5.36it/s]
+
 77%|███████▋  | 30/39 [00:05<00:01,  5.37it/s]
+
 79%|███████▉  | 31/39 [00:05<00:01,  5.38it/s]
+
 82%|████████▏ | 32/39 [00:05<00:01,  5.33it/s]
+
 85%|████████▍ | 33/39 [00:06<00:01,  5.32it/s]
+
 87%|████████▋ | 34/39 [00:06<00:00,  5.40it/s]
+
 90%|████████▉ | 35/39 [00:06<00:00,  5.41it/s]
+
 92%|█████████▏| 36/39 [00:06<00:00,  5.40it/s]
+
 95%|█████████▍| 37/39 [00:06<00:00,  5.35it/s]
+
 97%|█████████▋| 38/39 [00:06<00:00,  5.34it/s]
+
100%|██████████| 39/39 [00:07<00:00,  5.29it/s]
                                                     
+
                                               
{'eval_loss': '3.529', 'eval_runtime': '7.379', 'eval_samples_per_second': '330.8', 'eval_steps_per_second': '5.285', 'epoch': '17.06'}
+
 68%|██████▊   | 4896/7175 [1:01:58<56:07,  1.48s/it]
+
100%|██████████| 39/39 [00:07<00:00,  5.29it/s][eval] epoch 17.059 step 4896 loss 3.5293 (best 3.5092 @ ep 16.811, 1/20 since best)
+
+
                                               
 68%|██████▊   | 4897/7175 [1:01:59<2:20:00,  3.69s/it]
 68%|██████▊   | 4898/7175 [1:02:01<1:54:46,  3.02s/it]
 68%|██████▊   | 4899/7175 [1:02:02<1:37:09,  2.56s/it]
 68%|██████▊   | 4900/7175 [1:02:04<1:24:54,  2.24s/it]
                                                       
{'loss': '3.012', 'grad_norm': '0.2423', 'learning_rate': '0.0006', 'epoch': '17.07'}
+
 68%|██████▊   | 4900/7175 [1:02:04<1:24:54,  2.24s/it]
 68%|██████▊   | 4901/7175 [1:02:05<1:16:23,  2.02s/it]
 68%|██████▊   | 4902/7175 [1:02:07<1:10:13,  1.85s/it]
 68%|██████▊   | 4903/7175 [1:02:08<1:05:54,  1.74s/it]
 68%|██████▊   | 4904/7175 [1:02:10<1:03:03,  1.67s/it]
 68%|██████▊   | 4905/7175 [1:02:11<1:00:59,  1.61s/it]
 68%|██████▊   | 4906/7175 [1:02:13<59:25,  1.57s/it]  
 68%|██████▊   | 4907/7175 [1:02:14<58:18,  1.54s/it]
 68%|██████▊   | 4908/7175 [1:02:15<57:40,  1.53s/it]
 68%|██████▊   | 4909/7175 [1:02:17<57:12,  1.51s/it]
 68%|██████▊   | 4910/7175 [1:02:18<56:47,  1.50s/it]
                                                     
{'loss': '3', 'grad_norm': '0.2409', 'learning_rate': '0.0006', 'epoch': '17.11'}
+
 68%|██████▊   | 4910/7175 [1:02:18<56:47,  1.50s/it]
 68%|██████▊   | 4911/7175 [1:02:20<56:37,  1.50s/it]
 68%|██████▊   | 4912/7175 [1:02:21<56:24,  1.50s/it]
 68%|██████▊   | 4913/7175 [1:02:23<56:09,  1.49s/it]
 68%|██████▊   | 4914/7175 [1:02:24<50:08,  1.33s/it]
 69%|██████▊   | 4915/7175 [1:02:25<45:59,  1.22s/it]
 69%|██████▊   | 4916/7175 [1:02:26<46:26,  1.23s/it]
 69%|██████▊   | 4917/7175 [1:02:28<49:04,  1.30s/it]
 69%|██████▊   | 4918/7175 [1:02:29<50:51,  1.35s/it]
 69%|██████▊   | 4919/7175 [1:02:30<52:07,  1.39s/it]
 69%|██████▊   | 4920/7175 [1:02:32<53:19,  1.42s/it]
                                                     
{'loss': '3.001', 'grad_norm': '0.2341', 'learning_rate': '0.0006', 'epoch': '17.14'}
+
 69%|██████▊   | 4920/7175 [1:02:32<53:19,  1.42s/it]
 69%|██████▊   | 4921/7175 [1:02:33<54:19,  1.45s/it]
 69%|██████▊   | 4922/7175 [1:02:35<54:41,  1.46s/it]
 69%|██████▊   | 4923/7175 [1:02:36<54:53,  1.46s/it]
 69%|██████▊   | 4924/7175 [1:02:38<55:05,  1.47s/it]
 69%|██████▊   | 4925/7175 [1:02:39<55:12,  1.47s/it]
 69%|██████▊   | 4926/7175 [1:02:41<55:16,  1.47s/it]
 69%|██████▊   | 4927/7175 [1:02:42<55:20,  1.48s/it]
 69%|██████▊   | 4928/7175 [1:02:44<55:21,  1.48s/it]
 69%|██████▊   | 4929/7175 [1:02:45<55:24,  1.48s/it]
 69%|██████▊   | 4930/7175 [1:02:47<55:24,  1.48s/it]
                                                     
{'loss': '3.019', 'grad_norm': '0.2372', 'learning_rate': '0.0006', 'epoch': '17.18'}
+
 69%|██████▊   | 4930/7175 [1:02:47<55:24,  1.48s/it]
 69%|██████▊   | 4931/7175 [1:02:48<55:32,  1.48s/it]
 69%|██████▊   | 4932/7175 [1:02:50<55:26,  1.48s/it]
 69%|██████▉   | 4933/7175 [1:02:51<55:22,  1.48s/it]
 69%|██████▉   | 4934/7175 [1:02:53<55:16,  1.48s/it]
 69%|██████▉   | 4935/7175 [1:02:54<55:15,  1.48s/it]
 69%|██████▉   | 4936/7175 [1:02:56<55:13,  1.48s/it]
 69%|██████▉   | 4937/7175 [1:02:57<55:20,  1.48s/it]
 69%|██████▉   | 4938/7175 [1:02:59<55:22,  1.49s/it]
 69%|██████▉   | 4939/7175 [1:03:00<55:17,  1.48s/it]
 69%|██████▉   | 4940/7175 [1:03:02<55:15,  1.48s/it]
                                                     
{'loss': '3.019', 'grad_norm': '0.2386', 'learning_rate': '0.0006', 'epoch': '17.21'}
+
 69%|██████▉   | 4940/7175 [1:03:02<55:15,  1.48s/it]
 69%|██████▉   | 4941/7175 [1:03:03<55:17,  1.49s/it]
 69%|██████▉   | 4942/7175 [1:03:05<55:12,  1.48s/it]
 69%|██████▉   | 4943/7175 [1:03:06<55:08,  1.48s/it]
 69%|██████▉   | 4944/7175 [1:03:08<55:06,  1.48s/it]
 69%|██████▉   | 4945/7175 [1:03:09<55:01,  1.48s/it]
 69%|██████▉   | 4946/7175 [1:03:11<54:58,  1.48s/it]
 69%|██████▉   | 4947/7175 [1:03:12<55:01,  1.48s/it]
 69%|██████▉   | 4948/7175 [1:03:13<55:01,  1.48s/it]
 69%|██████▉   | 4949/7175 [1:03:15<54:56,  1.48s/it]
 69%|██████▉   | 4950/7175 [1:03:16<54:51,  1.48s/it]
                                                     
{'loss': '3.004', 'grad_norm': '0.231', 'learning_rate': '0.0006', 'epoch': '17.25'}
+
 69%|██████▉   | 4950/7175 [1:03:16<54:51,  1.48s/it]
 69%|██████▉   | 4951/7175 [1:03:18<55:00,  1.48s/it]
 69%|██████▉   | 4952/7175 [1:03:19<54:58,  1.48s/it]
 69%|██████▉   | 4953/7175 [1:03:21<54:50,  1.48s/it]
 69%|██████▉   | 4954/7175 [1:03:22<54:47,  1.48s/it]
 69%|██████▉   | 4955/7175 [1:03:24<54:46,  1.48s/it]
 69%|██████▉   | 4956/7175 [1:03:25<54:49,  1.48s/it]
 69%|██████▉   | 4957/7175 [1:03:27<54:45,  1.48s/it]
 69%|██████▉   | 4958/7175 [1:03:28<54:42,  1.48s/it]
 69%|██████▉   | 4959/7175 [1:03:30<54:42,  1.48s/it]
 69%|██████▉   | 4960/7175 [1:03:31<54:40,  1.48s/it]
                                                     
{'loss': '3.024', 'grad_norm': '0.2219', 'learning_rate': '0.0006', 'epoch': '17.28'}
+
 69%|██████▉   | 4960/7175 [1:03:31<54:40,  1.48s/it]
 69%|██████▉   | 4961/7175 [1:03:33<54:45,  1.48s/it]
 69%|██████▉   | 4962/7175 [1:03:34<54:43,  1.48s/it]
 69%|██████▉   | 4963/7175 [1:03:36<54:39,  1.48s/it]
 69%|██████▉   | 4964/7175 [1:03:37<54:35,  1.48s/it]
 69%|██████▉   | 4965/7175 [1:03:39<54:38,  1.48s/it]
 69%|██████▉   | 4966/7175 [1:03:40<54:35,  1.48s/it]
 69%|██████▉   | 4967/7175 [1:03:42<54:38,  1.48s/it]
 69%|██████▉   | 4968/7175 [1:03:43<54:39,  1.49s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:03, 10.68it/s]
+
 10%|█         | 4/39 [00:00<00:05,  6.64it/s]
+
 13%|█▎        | 5/39 [00:00<00:05,  6.21it/s]
+
 15%|█▌        | 6/39 [00:00<00:05,  5.83it/s]
+
 18%|█▊        | 7/39 [00:01<00:05,  5.62it/s]
+
 21%|██        | 8/39 [00:01<00:05,  5.51it/s]
+
 23%|██▎       | 9/39 [00:01<00:05,  5.56it/s]
+
 26%|██▌       | 10/39 [00:01<00:05,  5.51it/s]
+
 28%|██▊       | 11/39 [00:01<00:05,  5.48it/s]
+
 31%|███       | 12/39 [00:02<00:04,  5.40it/s]
+
 33%|███▎      | 13/39 [00:02<00:04,  5.37it/s]
+
 36%|███▌      | 14/39 [00:02<00:04,  5.32it/s]
+
 38%|███▊      | 15/39 [00:02<00:04,  5.33it/s]
+
 41%|████      | 16/39 [00:02<00:04,  5.30it/s]
+
 44%|████▎     | 17/39 [00:03<00:04,  5.29it/s]
+
 46%|████▌     | 18/39 [00:03<00:04,  5.23it/s]
+
 49%|████▊     | 19/39 [00:03<00:03,  5.22it/s]
+
 51%|█████▏    | 20/39 [00:03<00:03,  5.23it/s]
+
 54%|█████▍    | 21/39 [00:03<00:03,  5.29it/s]
+
 56%|█████▋    | 22/39 [00:03<00:03,  5.26it/s]
+
 59%|█████▉    | 23/39 [00:04<00:03,  5.25it/s]
+
 62%|██████▏   | 24/39 [00:04<00:02,  5.27it/s]
+
 64%|██████▍   | 25/39 [00:04<00:02,  5.25it/s]
+
 67%|██████▋   | 26/39 [00:04<00:02,  5.24it/s]
+
 69%|██████▉   | 27/39 [00:04<00:02,  5.28it/s]
+
 72%|███████▏  | 28/39 [00:05<00:02,  5.29it/s]
+
 74%|███████▍  | 29/39 [00:05<00:01,  5.26it/s]
+
 77%|███████▋  | 30/39 [00:05<00:01,  5.20it/s]
+
 79%|███████▉  | 31/39 [00:05<00:01,  5.20it/s]
+
 82%|████████▏ | 32/39 [00:05<00:01,  5.20it/s]
+
 85%|████████▍ | 33/39 [00:06<00:01,  5.25it/s]
+
 87%|████████▋ | 34/39 [00:06<00:00,  5.27it/s]
+
 90%|████████▉ | 35/39 [00:06<00:00,  5.24it/s]
+
 92%|█████████▏| 36/39 [00:06<00:00,  5.21it/s]
+
 95%|█████████▍| 37/39 [00:06<00:00,  5.24it/s]
+
 97%|█████████▋| 38/39 [00:07<00:00,  5.23it/s]
+
100%|██████████| 39/39 [00:07<00:00,  5.27it/s]
                                                     
+
                                               
{'eval_loss': '3.517', 'eval_runtime': '7.463', 'eval_samples_per_second': '327.1', 'eval_steps_per_second': '5.226', 'epoch': '17.31'}
+
 69%|██████▉   | 4968/7175 [1:03:51<54:39,  1.49s/it]
+
100%|██████████| 39/39 [00:07<00:00,  5.27it/s][eval] epoch 17.311 step 4968 loss 3.5166 (best 3.5092 @ ep 16.811, 2/20 since best)
+
+
                                               
 69%|██████▉   | 4969/7175 [1:03:52<2:16:49,  3.72s/it]
 69%|██████▉   | 4970/7175 [1:03:54<1:52:11,  3.05s/it]
                                                       
{'loss': '3.029', 'grad_norm': '0.2321', 'learning_rate': '0.0006', 'epoch': '17.32'}
+
 69%|██████▉   | 4970/7175 [1:03:54<1:52:11,  3.05s/it]
 69%|██████▉   | 4971/7175 [1:03:55<1:35:05,  2.59s/it]
 69%|██████▉   | 4972/7175 [1:03:57<1:22:50,  2.26s/it]
 69%|██████▉   | 4973/7175 [1:03:58<1:14:17,  2.02s/it]
 69%|██████▉   | 4974/7175 [1:04:00<1:08:14,  1.86s/it]
 69%|██████▉   | 4975/7175 [1:04:01<1:03:58,  1.74s/it]
 69%|██████▉   | 4976/7175 [1:04:02<1:00:57,  1.66s/it]
 69%|██████▉   | 4977/7175 [1:04:04<58:53,  1.61s/it]  
 69%|██████▉   | 4978/7175 [1:04:05<57:31,  1.57s/it]
 69%|██████▉   | 4979/7175 [1:04:07<56:29,  1.54s/it]
 69%|██████▉   | 4980/7175 [1:04:08<55:27,  1.52s/it]
                                                     
{'loss': '3.012', 'grad_norm': '0.2473', 'learning_rate': '0.0006', 'epoch': '17.35'}
+
 69%|██████▉   | 4980/7175 [1:04:08<55:27,  1.52s/it]
 69%|██████▉   | 4981/7175 [1:04:10<55:15,  1.51s/it]
 69%|██████▉   | 4982/7175 [1:04:11<54:38,  1.49s/it]
 69%|██████▉   | 4983/7175 [1:04:13<54:32,  1.49s/it]
 69%|██████▉   | 4984/7175 [1:04:14<54:25,  1.49s/it]
 69%|██████▉   | 4985/7175 [1:04:16<54:22,  1.49s/it]
 69%|██████▉   | 4986/7175 [1:04:17<54:14,  1.49s/it]
 70%|██████▉   | 4987/7175 [1:04:19<54:10,  1.49s/it]
 70%|██████▉   | 4988/7175 [1:04:20<54:03,  1.48s/it]
 70%|██████▉   | 4989/7175 [1:04:22<54:01,  1.48s/it]
 70%|██████▉   | 4990/7175 [1:04:23<53:58,  1.48s/it]
                                                     
{'loss': '3.027', 'grad_norm': '0.2416', 'learning_rate': '0.0006', 'epoch': '17.39'}
+
 70%|██████▉   | 4990/7175 [1:04:23<53:58,  1.48s/it]
 70%|██████▉   | 4991/7175 [1:04:25<54:13,  1.49s/it]
 70%|██████▉   | 4992/7175 [1:04:26<54:11,  1.49s/it]
 70%|██████▉   | 4993/7175 [1:04:28<54:04,  1.49s/it]
 70%|██████▉   | 4994/7175 [1:04:29<53:56,  1.48s/it]
 70%|██████▉   | 4995/7175 [1:04:31<53:51,  1.48s/it]
 70%|██████▉   | 4996/7175 [1:04:32<53:57,  1.49s/it]
 70%|██████▉   | 4997/7175 [1:04:34<53:53,  1.48s/it]
 70%|██████▉   | 4998/7175 [1:04:35<53:51,  1.48s/it]
 70%|██████▉   | 4999/7175 [1:04:37<53:45,  1.48s/it]
 70%|██████▉   | 5000/7175 [1:04:38<53:39,  1.48s/it]
                                                     
{'loss': '3.042', 'grad_norm': '0.2365', 'learning_rate': '0.0006', 'epoch': '17.42'}
+
 70%|██████▉   | 5000/7175 [1:04:38<53:39,  1.48s/it]
 70%|██████▉   | 5001/7175 [1:04:40<53:49,  1.49s/it]
 70%|██████▉   | 5002/7175 [1:04:41<53:43,  1.48s/it]
 70%|██████▉   | 5003/7175 [1:04:42<53:36,  1.48s/it]
 70%|██████▉   | 5004/7175 [1:04:44<53:33,  1.48s/it]
 70%|██████▉   | 5005/7175 [1:04:45<53:34,  1.48s/it]
 70%|██████▉   | 5006/7175 [1:04:47<53:31,  1.48s/it]
 70%|██████▉   | 5007/7175 [1:04:48<53:23,  1.48s/it]
 70%|██████▉   | 5008/7175 [1:04:50<53:36,  1.48s/it]
 70%|██████▉   | 5009/7175 [1:04:51<53:32,  1.48s/it]
 70%|██████▉   | 5010/7175 [1:04:53<53:26,  1.48s/it]
                                                     
{'loss': '3.046', 'grad_norm': '0.2525', 'learning_rate': '0.0006', 'epoch': '17.46'}
+
 70%|██████▉   | 5010/7175 [1:04:53<53:26,  1.48s/it]
 70%|██████▉   | 5011/7175 [1:04:54<53:33,  1.49s/it]
 70%|██████▉   | 5012/7175 [1:04:56<53:35,  1.49s/it]
 70%|██████▉   | 5013/7175 [1:04:57<53:28,  1.48s/it]
 70%|██████▉   | 5014/7175 [1:04:59<53:18,  1.48s/it]
 70%|██████▉   | 5015/7175 [1:05:00<53:16,  1.48s/it]
 70%|██████▉   | 5016/7175 [1:05:02<53:16,  1.48s/it]
 70%|██████▉   | 5017/7175 [1:05:03<53:10,  1.48s/it]
 70%|██████▉   | 5018/7175 [1:05:05<53:14,  1.48s/it]
 70%|██████▉   | 5019/7175 [1:05:06<53:10,  1.48s/it]
 70%|██████▉   | 5020/7175 [1:05:08<53:13,  1.48s/it]
                                                     
{'loss': '3.049', 'grad_norm': '0.2232', 'learning_rate': '0.0006', 'epoch': '17.49'}
+
 70%|██████▉   | 5020/7175 [1:05:08<53:13,  1.48s/it]
 70%|██████▉   | 5021/7175 [1:05:09<53:20,  1.49s/it]
 70%|██████▉   | 5022/7175 [1:05:11<53:16,  1.48s/it]
 70%|███████   | 5023/7175 [1:05:12<53:13,  1.48s/it]
 70%|███████   | 5024/7175 [1:05:14<53:11,  1.48s/it]
 70%|███████   | 5025/7175 [1:05:15<53:12,  1.49s/it]
 70%|███████   | 5026/7175 [1:05:17<53:04,  1.48s/it]
 70%|███████   | 5027/7175 [1:05:18<52:57,  1.48s/it]
 70%|███████   | 5028/7175 [1:05:20<52:56,  1.48s/it]
 70%|███████   | 5029/7175 [1:05:21<52:52,  1.48s/it]
 70%|███████   | 5030/7175 [1:05:22<52:52,  1.48s/it]
                                                     
{'loss': '3.035', 'grad_norm': '0.23', 'learning_rate': '0.0006', 'epoch': '17.53'}
+
 70%|███████   | 5030/7175 [1:05:23<52:52,  1.48s/it]
 70%|███████   | 5031/7175 [1:05:24<53:03,  1.48s/it]
 70%|███████   | 5032/7175 [1:05:25<52:57,  1.48s/it]
 70%|███████   | 5033/7175 [1:05:27<52:55,  1.48s/it]
 70%|███████   | 5034/7175 [1:05:28<52:50,  1.48s/it]
 70%|███████   | 5035/7175 [1:05:30<52:47,  1.48s/it]
 70%|███████   | 5036/7175 [1:05:31<52:46,  1.48s/it]
 70%|███████   | 5037/7175 [1:05:33<52:47,  1.48s/it]
 70%|███████   | 5038/7175 [1:05:34<52:39,  1.48s/it]
 70%|███████   | 5039/7175 [1:05:36<52:36,  1.48s/it]
 70%|███████   | 5040/7175 [1:05:37<52:42,  1.48s/it]
                                                     
{'loss': '3.041', 'grad_norm': '0.2192', 'learning_rate': '0.0006', 'epoch': '17.56'}
+
 70%|███████   | 5040/7175 [1:05:37<52:42,  1.48s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:03, 10.82it/s]
+
 10%|█         | 4/39 [00:00<00:05,  6.70it/s]
+
 13%|█▎        | 5/39 [00:00<00:05,  6.17it/s]
+
 15%|█▌        | 6/39 [00:00<00:05,  5.85it/s]
+
 18%|█▊        | 7/39 [00:01<00:05,  5.68it/s]
+
 21%|██        | 8/39 [00:01<00:05,  5.53it/s]
+
 23%|██▎       | 9/39 [00:01<00:05,  5.48it/s]
+
 26%|██▌       | 10/39 [00:01<00:05,  5.39it/s]
+
 28%|██▊       | 11/39 [00:01<00:05,  5.34it/s]
+
 31%|███       | 12/39 [00:02<00:05,  5.33it/s]
+
 33%|███▎      | 13/39 [00:02<00:04,  5.29it/s]
+
 36%|███▌      | 14/39 [00:02<00:04,  5.27it/s]
+
 38%|███▊      | 15/39 [00:02<00:04,  5.32it/s]
+
 41%|████      | 16/39 [00:02<00:04,  5.35it/s]
+
 44%|████▎     | 17/39 [00:03<00:04,  5.37it/s]
+
 46%|████▌     | 18/39 [00:03<00:03,  5.32it/s]
+
 49%|████▊     | 19/39 [00:03<00:03,  5.32it/s]
+
 51%|█████▏    | 20/39 [00:03<00:03,  5.28it/s]
+
 54%|█████▍    | 21/39 [00:03<00:03,  5.30it/s]
+
 56%|█████▋    | 22/39 [00:03<00:03,  5.27it/s]
+
 59%|█████▉    | 23/39 [00:04<00:03,  5.29it/s]
+
 62%|██████▏   | 24/39 [00:04<00:02,  5.30it/s]
+
 64%|██████▍   | 25/39 [00:04<00:02,  5.33it/s]
+
 67%|██████▋   | 26/39 [00:04<00:02,  5.40it/s]
+
 69%|██████▉   | 27/39 [00:04<00:02,  5.40it/s]
+
 72%|███████▏  | 28/39 [00:05<00:02,  5.40it/s]
+
 74%|███████▍  | 29/39 [00:05<00:01,  5.35it/s]
+
 77%|███████▋  | 30/39 [00:05<00:01,  5.30it/s]
+
 79%|███████▉  | 31/39 [00:05<00:01,  5.30it/s]
+
 82%|████████▏ | 32/39 [00:05<00:01,  5.33it/s]
+
 85%|████████▍ | 33/39 [00:06<00:01,  5.29it/s]
+
 87%|████████▋ | 34/39 [00:06<00:00,  5.26it/s]
+
 90%|████████▉ | 35/39 [00:06<00:00,  5.25it/s]
+
 92%|█████████▏| 36/39 [00:06<00:00,  5.26it/s]
+
 95%|█████████▍| 37/39 [00:06<00:00,  5.25it/s]
+
 97%|█████████▋| 38/39 [00:06<00:00,  5.28it/s]
+
100%|██████████| 39/39 [00:07<00:00,  5.25it/s]
                                                     
+
                                               
{'eval_loss': '3.516', 'eval_runtime': '7.423', 'eval_samples_per_second': '328.8', 'eval_steps_per_second': '5.254', 'epoch': '17.56'}
+
 70%|███████   | 5040/7175 [1:05:45<52:42,  1.48s/it]
+
100%|██████████| 39/39 [00:07<00:00,  5.25it/s][eval] epoch 17.563 step 5040 loss 3.5159 (best 3.5092 @ ep 16.811, 3/20 since best)
+
+
                                               
 70%|███████   | 5041/7175 [1:05:46<2:12:08,  3.72s/it]
 70%|███████   | 5042/7175 [1:05:48<1:48:13,  3.04s/it]
 70%|███████   | 5043/7175 [1:05:49<1:31:27,  2.57s/it]
 70%|███████   | 5044/7175 [1:05:51<1:19:43,  2.24s/it]
 70%|███████   | 5045/7175 [1:05:52<1:11:37,  2.02s/it]
 70%|███████   | 5046/7175 [1:05:54<1:05:47,  1.85s/it]
 70%|███████   | 5047/7175 [1:05:55<1:01:49,  1.74s/it]
 70%|███████   | 5048/7175 [1:05:57<59:05,  1.67s/it]  
 70%|███████   | 5049/7175 [1:05:58<57:00,  1.61s/it]
 70%|███████   | 5050/7175 [1:06:00<55:38,  1.57s/it]
                                                     
{'loss': '3.035', 'grad_norm': '0.2324', 'learning_rate': '0.0006', 'epoch': '17.6'}
+
 70%|███████   | 5050/7175 [1:06:00<55:38,  1.57s/it]
 70%|███████   | 5051/7175 [1:06:01<54:55,  1.55s/it]
 70%|███████   | 5052/7175 [1:06:03<54:13,  1.53s/it]
 70%|███████   | 5053/7175 [1:06:04<53:38,  1.52s/it]
 70%|███████   | 5054/7175 [1:06:06<53:14,  1.51s/it]
 70%|███████   | 5055/7175 [1:06:07<52:54,  1.50s/it]
 70%|███████   | 5056/7175 [1:06:08<52:42,  1.49s/it]
 70%|███████   | 5057/7175 [1:06:10<52:32,  1.49s/it]
 70%|███████   | 5058/7175 [1:06:11<52:27,  1.49s/it]
 71%|███████   | 5059/7175 [1:06:13<52:23,  1.49s/it]
 71%|███████   | 5060/7175 [1:06:14<52:06,  1.48s/it]
                                                     
{'loss': '3.038', 'grad_norm': '0.2328', 'learning_rate': '0.0006', 'epoch': '17.63'}
+
 71%|███████   | 5060/7175 [1:06:14<52:06,  1.48s/it]
 71%|███████   | 5061/7175 [1:06:16<51:59,  1.48s/it]
 71%|███████   | 5062/7175 [1:06:17<51:59,  1.48s/it]
 71%|███████   | 5063/7175 [1:06:19<51:18,  1.46s/it]
 71%|███████   | 5064/7175 [1:06:19<42:49,  1.22s/it]
 71%|███████   | 5065/7175 [1:06:20<41:07,  1.17s/it]
 71%|███████   | 5066/7175 [1:06:22<44:13,  1.26s/it]
 71%|███████   | 5067/7175 [1:06:23<46:21,  1.32s/it]
 71%|███████   | 5068/7175 [1:06:25<47:47,  1.36s/it]
 71%|███████   | 5069/7175 [1:06:26<48:52,  1.39s/it]
 71%|███████   | 5070/7175 [1:06:28<49:55,  1.42s/it]
                                                     
{'loss': '3.057', 'grad_norm': '0.224', 'learning_rate': '0.0006', 'epoch': '17.67'}
+
 71%|███████   | 5070/7175 [1:06:28<49:55,  1.42s/it]
 71%|███████   | 5071/7175 [1:06:29<50:29,  1.44s/it]
 71%|███████   | 5072/7175 [1:06:31<50:55,  1.45s/it]
 71%|███████   | 5073/7175 [1:06:32<51:13,  1.46s/it]
 71%|███████   | 5074/7175 [1:06:34<51:19,  1.47s/it]
 71%|███████   | 5075/7175 [1:06:35<51:26,  1.47s/it]
 71%|███████   | 5076/7175 [1:06:37<51:32,  1.47s/it]
 71%|███████   | 5077/7175 [1:06:38<51:37,  1.48s/it]
 71%|███████   | 5078/7175 [1:06:40<51:35,  1.48s/it]
 71%|███████   | 5079/7175 [1:06:41<51:38,  1.48s/it]
 71%|███████   | 5080/7175 [1:06:43<51:41,  1.48s/it]
                                                     
{'loss': '3.06', 'grad_norm': '0.1981', 'learning_rate': '0.0006', 'epoch': '17.7'}
+
 71%|███████   | 5080/7175 [1:06:43<51:41,  1.48s/it]
 71%|███████   | 5081/7175 [1:06:44<51:47,  1.48s/it]
 71%|███████   | 5082/7175 [1:06:46<51:43,  1.48s/it]
 71%|███████   | 5083/7175 [1:06:47<51:38,  1.48s/it]
 71%|███████   | 5084/7175 [1:06:49<51:35,  1.48s/it]
 71%|███████   | 5085/7175 [1:06:50<51:34,  1.48s/it]
 71%|███████   | 5086/7175 [1:06:51<51:32,  1.48s/it]
 71%|███████   | 5087/7175 [1:06:53<51:29,  1.48s/it]
 71%|███████   | 5088/7175 [1:06:54<51:28,  1.48s/it]
 71%|███████   | 5089/7175 [1:06:56<51:28,  1.48s/it]
 71%|███████   | 5090/7175 [1:06:57<51:21,  1.48s/it]
                                                     
{'loss': '3.052', 'grad_norm': '0.2186', 'learning_rate': '0.0006', 'epoch': '17.74'}
+
 71%|███████   | 5090/7175 [1:06:57<51:21,  1.48s/it]
 71%|███████   | 5091/7175 [1:06:59<51:32,  1.48s/it]
 71%|███████   | 5092/7175 [1:07:00<51:27,  1.48s/it]
 71%|███████   | 5093/7175 [1:07:02<51:22,  1.48s/it]
 71%|███████   | 5094/7175 [1:07:03<51:18,  1.48s/it]
 71%|███████   | 5095/7175 [1:07:05<51:21,  1.48s/it]
 71%|███████   | 5096/7175 [1:07:06<51:17,  1.48s/it]
 71%|███████   | 5097/7175 [1:07:08<51:14,  1.48s/it]
 71%|███████   | 5098/7175 [1:07:09<51:14,  1.48s/it]
 71%|███████   | 5099/7175 [1:07:11<51:09,  1.48s/it]
 71%|███████   | 5100/7175 [1:07:12<51:08,  1.48s/it]
                                                     
{'loss': '3.059', 'grad_norm': '0.2207', 'learning_rate': '0.0006', 'epoch': '17.77'}
+
 71%|███████   | 5100/7175 [1:07:12<51:08,  1.48s/it]
 71%|███████   | 5101/7175 [1:07:14<51:13,  1.48s/it]
 71%|███████   | 5102/7175 [1:07:15<51:12,  1.48s/it]
 71%|███████   | 5103/7175 [1:07:17<51:11,  1.48s/it]
 71%|███████   | 5104/7175 [1:07:18<51:13,  1.48s/it]
 71%|███████   | 5105/7175 [1:07:20<51:10,  1.48s/it]
 71%|███████   | 5106/7175 [1:07:21<51:07,  1.48s/it]
 71%|███████   | 5107/7175 [1:07:23<51:02,  1.48s/it]
 71%|███████   | 5108/7175 [1:07:24<50:57,  1.48s/it]
 71%|███████   | 5109/7175 [1:07:26<50:53,  1.48s/it]
 71%|███████   | 5110/7175 [1:07:27<50:55,  1.48s/it]
                                                     
{'loss': '3.063', 'grad_norm': '0.2189', 'learning_rate': '0.0006', 'epoch': '17.81'}
+
 71%|███████   | 5110/7175 [1:07:27<50:55,  1.48s/it]
 71%|███████   | 5111/7175 [1:07:29<51:05,  1.49s/it]
 71%|███████   | 5112/7175 [1:07:30<51:01,  1.48s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:03, 10.63it/s]
+
 10%|█         | 4/39 [00:00<00:05,  6.70it/s]
+
 13%|█▎        | 5/39 [00:00<00:05,  6.26it/s]
+
 15%|█▌        | 6/39 [00:00<00:05,  6.04it/s]
+
 18%|█▊        | 7/39 [00:01<00:05,  5.84it/s]
+
 21%|██        | 8/39 [00:01<00:05,  5.71it/s]
+
 23%|██▎       | 9/39 [00:01<00:05,  5.54it/s]
+
 26%|██▌       | 10/39 [00:01<00:05,  5.47it/s]
+
 28%|██▊       | 11/39 [00:01<00:05,  5.42it/s]
+
 31%|███       | 12/39 [00:02<00:05,  5.36it/s]
+
 33%|███▎      | 13/39 [00:02<00:04,  5.36it/s]
+
 36%|███▌      | 14/39 [00:02<00:04,  5.33it/s]
+
 38%|███▊      | 15/39 [00:02<00:04,  5.33it/s]
+
 41%|████      | 16/39 [00:02<00:04,  5.33it/s]
+
 44%|████▎     | 17/39 [00:02<00:04,  5.35it/s]
+
 46%|████▌     | 18/39 [00:03<00:03,  5.34it/s]
+
 49%|████▊     | 19/39 [00:03<00:03,  5.28it/s]
+
 51%|█████▏    | 20/39 [00:03<00:03,  5.26it/s]
+
 54%|█████▍    | 21/39 [00:03<00:03,  5.24it/s]
+
 56%|█████▋    | 22/39 [00:03<00:03,  5.26it/s]
+
 59%|█████▉    | 23/39 [00:04<00:03,  5.25it/s]
+
 62%|██████▏   | 24/39 [00:04<00:02,  5.29it/s]
+
 64%|██████▍   | 25/39 [00:04<00:02,  5.26it/s]
+
 67%|██████▋   | 26/39 [00:04<00:02,  5.28it/s]
+
 69%|██████▉   | 27/39 [00:04<00:02,  5.26it/s]
+
 72%|███████▏  | 28/39 [00:05<00:02,  5.24it/s]
+
 74%|███████▍  | 29/39 [00:05<00:01,  5.23it/s]
+
 77%|███████▋  | 30/39 [00:05<00:01,  5.30it/s]
+
 79%|███████▉  | 31/39 [00:05<00:01,  5.27it/s]
+
 82%|████████▏ | 32/39 [00:05<00:01,  5.25it/s]
+
 85%|████████▍ | 33/39 [00:06<00:01,  5.26it/s]
+
 87%|████████▋ | 34/39 [00:06<00:00,  5.30it/s]
+
 90%|████████▉ | 35/39 [00:06<00:00,  5.38it/s]
+
 92%|█████████▏| 36/39 [00:06<00:00,  5.39it/s]
+
 95%|█████████▍| 37/39 [00:06<00:00,  5.40it/s]
+
 97%|█████████▋| 38/39 [00:06<00:00,  5.38it/s]
+
100%|██████████| 39/39 [00:07<00:00,  5.32it/s]
                                                     
+
                                               
{'eval_loss': '3.512', 'eval_runtime': '7.401', 'eval_samples_per_second': '329.8', 'eval_steps_per_second': '5.27', 'epoch': '17.81'}
+
 71%|███████   | 5112/7175 [1:07:37<51:01,  1.48s/it]
+
100%|██████████| 39/39 [00:07<00:00,  5.32it/s][eval] epoch 17.814 step 5112 loss 3.5122 (best 3.5092 @ ep 16.811, 4/20 since best)
+
+
                                               
 71%|███████▏  | 5113/7175 [1:07:39<2:07:09,  3.70s/it]
 71%|███████▏  | 5114/7175 [1:07:40<1:44:11,  3.03s/it]
 71%|███████▏  | 5115/7175 [1:07:42<1:28:06,  2.57s/it]
 71%|███████▏  | 5116/7175 [1:07:43<1:16:52,  2.24s/it]
 71%|███████▏  | 5117/7175 [1:07:45<1:09:04,  2.01s/it]
 71%|███████▏  | 5118/7175 [1:07:46<1:03:32,  1.85s/it]
 71%|███████▏  | 5119/7175 [1:07:48<59:37,  1.74s/it]  
 71%|███████▏  | 5120/7175 [1:07:49<56:54,  1.66s/it]
                                                     
{'loss': '3.059', 'grad_norm': '0.2106', 'learning_rate': '0.0006', 'epoch': '17.84'}
+
 71%|███████▏  | 5120/7175 [1:07:49<56:54,  1.66s/it]
 71%|███████▏  | 5121/7175 [1:07:51<55:08,  1.61s/it]
 71%|███████▏  | 5122/7175 [1:07:52<53:42,  1.57s/it]
 71%|███████▏  | 5123/7175 [1:07:54<52:45,  1.54s/it]
 71%|███████▏  | 5124/7175 [1:07:55<52:05,  1.52s/it]
 71%|███████▏  | 5125/7175 [1:07:57<51:32,  1.51s/it]
 71%|███████▏  | 5126/7175 [1:07:58<51:14,  1.50s/it]
 71%|███████▏  | 5127/7175 [1:08:00<50:59,  1.49s/it]
 71%|███████▏  | 5128/7175 [1:08:01<50:50,  1.49s/it]
 71%|███████▏  | 5129/7175 [1:08:03<50:37,  1.48s/it]
 71%|███████▏  | 5130/7175 [1:08:04<50:34,  1.48s/it]
                                                     
{'loss': '3.067', 'grad_norm': '0.2021', 'learning_rate': '0.0006', 'epoch': '17.88'}
+
 71%|███████▏  | 5130/7175 [1:08:04<50:34,  1.48s/it]
 72%|███████▏  | 5131/7175 [1:08:06<50:44,  1.49s/it]
 72%|███████▏  | 5132/7175 [1:08:07<50:39,  1.49s/it]
 72%|███████▏  | 5133/7175 [1:08:08<50:33,  1.49s/it]
 72%|███████▏  | 5134/7175 [1:08:10<50:32,  1.49s/it]
 72%|███████▏  | 5135/7175 [1:08:11<50:28,  1.48s/it]
 72%|███████▏  | 5136/7175 [1:08:13<50:21,  1.48s/it]
 72%|███████▏  | 5137/7175 [1:08:14<50:27,  1.49s/it]
 72%|███████▏  | 5138/7175 [1:08:16<50:26,  1.49s/it]
 72%|███████▏  | 5139/7175 [1:08:17<50:16,  1.48s/it]
 72%|███████▏  | 5140/7175 [1:08:19<50:12,  1.48s/it]
                                                     
{'loss': '3.059', 'grad_norm': '0.2448', 'learning_rate': '0.0006', 'epoch': '17.91'}
+
 72%|███████▏  | 5140/7175 [1:08:19<50:12,  1.48s/it]
 72%|███████▏  | 5141/7175 [1:08:20<50:22,  1.49s/it]
 72%|███████▏  | 5142/7175 [1:08:22<50:17,  1.48s/it]
 72%|███████▏  | 5143/7175 [1:08:23<50:09,  1.48s/it]
 72%|███████▏  | 5144/7175 [1:08:25<49:58,  1.48s/it]
 72%|███████▏  | 5145/7175 [1:08:26<49:58,  1.48s/it]
 72%|███████▏  | 5146/7175 [1:08:28<50:01,  1.48s/it]
 72%|███████▏  | 5147/7175 [1:08:29<50:00,  1.48s/it]
 72%|███████▏  | 5148/7175 [1:08:31<49:58,  1.48s/it]
 72%|███████▏  | 5149/7175 [1:08:32<50:04,  1.48s/it]
 72%|███████▏  | 5150/7175 [1:08:34<49:59,  1.48s/it]
                                                     
{'loss': '3.059', 'grad_norm': '0.2553', 'learning_rate': '0.0006', 'epoch': '17.95'}
+
 72%|███████▏  | 5150/7175 [1:08:34<49:59,  1.48s/it]
 72%|███████▏  | 5151/7175 [1:08:35<50:04,  1.48s/it]
 72%|███████▏  | 5152/7175 [1:08:37<49:58,  1.48s/it]
 72%|███████▏  | 5153/7175 [1:08:38<50:01,  1.48s/it]
 72%|███████▏  | 5154/7175 [1:08:40<49:58,  1.48s/it]
 72%|███████▏  | 5155/7175 [1:08:41<49:56,  1.48s/it]
 72%|███████▏  | 5156/7175 [1:08:43<49:49,  1.48s/it]
 72%|███████▏  | 5157/7175 [1:08:44<49:48,  1.48s/it]
 72%|███████▏  | 5158/7175 [1:08:46<49:48,  1.48s/it]
 72%|███████▏  | 5159/7175 [1:08:47<49:50,  1.48s/it]
 72%|███████▏  | 5160/7175 [1:08:49<49:58,  1.49s/it]
                                                     
{'loss': '3.071', 'grad_norm': '0.2152', 'learning_rate': '0.0006', 'epoch': '17.98'}
+
 72%|███████▏  | 5160/7175 [1:08:49<49:58,  1.49s/it]
 72%|███████▏  | 5161/7175 [1:08:50<50:04,  1.49s/it]
 72%|███████▏  | 5162/7175 [1:08:51<49:56,  1.49s/it]
 72%|███████▏  | 5163/7175 [1:08:53<49:53,  1.49s/it]
 72%|███████▏  | 5164/7175 [1:08:54<49:48,  1.49s/it]
 72%|███████▏  | 5165/7175 [1:08:56<49:41,  1.48s/it]
 72%|███████▏  | 5166/7175 [1:08:56<37:05,  1.11s/it]
 72%|███████▏  | 5167/7175 [1:08:58<41:43,  1.25s/it]
 72%|███████▏  | 5168/7175 [1:08:59<44:02,  1.32s/it]
 72%|███████▏  | 5169/7175 [1:09:01<45:38,  1.37s/it]
 72%|███████▏  | 5170/7175 [1:09:02<46:45,  1.40s/it]
                                                     
{'loss': '3.031', 'grad_norm': '0.4074', 'learning_rate': '0.0006', 'epoch': '18.01'}
+
 72%|███████▏  | 5170/7175 [1:09:02<46:45,  1.40s/it]
 72%|███████▏  | 5171/7175 [1:09:04<47:42,  1.43s/it]
 72%|███████▏  | 5172/7175 [1:09:05<48:11,  1.44s/it]
 72%|███████▏  | 5173/7175 [1:09:07<48:31,  1.45s/it]
 72%|███████▏  | 5174/7175 [1:09:08<48:47,  1.46s/it]
 72%|███████▏  | 5175/7175 [1:09:10<48:53,  1.47s/it]
 72%|███████▏  | 5176/7175 [1:09:11<49:01,  1.47s/it]
 72%|███████▏  | 5177/7175 [1:09:13<49:06,  1.47s/it]
 72%|███████▏  | 5178/7175 [1:09:14<49:06,  1.48s/it]
 72%|███████▏  | 5179/7175 [1:09:16<49:13,  1.48s/it]
 72%|███████▏  | 5180/7175 [1:09:17<49:10,  1.48s/it]
                                                     
{'loss': '3.011', 'grad_norm': '0.3224', 'learning_rate': '0.0006', 'epoch': '18.05'}
+
 72%|███████▏  | 5180/7175 [1:09:17<49:10,  1.48s/it]
 72%|███████▏  | 5181/7175 [1:09:18<49:17,  1.48s/it]
 72%|███████▏  | 5182/7175 [1:09:20<49:16,  1.48s/it]
 72%|███████▏  | 5183/7175 [1:09:21<49:14,  1.48s/it]
 72%|███████▏  | 5184/7175 [1:09:23<49:13,  1.48s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:03, 10.07it/s]
+
 10%|█         | 4/39 [00:00<00:05,  6.44it/s]
+
 13%|█▎        | 5/39 [00:00<00:05,  6.05it/s]
+
 15%|█▌        | 6/39 [00:00<00:05,  5.85it/s]
+
 18%|█▊        | 7/39 [00:01<00:05,  5.75it/s]
+
 21%|██        | 8/39 [00:01<00:05,  5.65it/s]
+
 23%|██▎       | 9/39 [00:01<00:05,  5.54it/s]
+
 26%|██▌       | 10/39 [00:01<00:05,  5.50it/s]
+
 28%|██▊       | 11/39 [00:01<00:05,  5.47it/s]
+
 31%|███       | 12/39 [00:02<00:04,  5.43it/s]
+
 33%|███▎      | 13/39 [00:02<00:04,  5.35it/s]
+
 36%|███▌      | 14/39 [00:02<00:04,  5.35it/s]
+
 38%|███▊      | 15/39 [00:02<00:04,  5.30it/s]
+
 41%|████      | 16/39 [00:02<00:04,  5.27it/s]
+
 44%|████▎     | 17/39 [00:03<00:04,  5.25it/s]
+
 46%|████▌     | 18/39 [00:03<00:03,  5.32it/s]
+
 49%|████▊     | 19/39 [00:03<00:03,  5.28it/s]
+
 51%|█████▏    | 20/39 [00:03<00:03,  5.26it/s]
+
 54%|█████▍    | 21/39 [00:03<00:03,  5.24it/s]
+
 56%|█████▋    | 22/39 [00:03<00:03,  5.26it/s]
+
 59%|█████▉    | 23/39 [00:04<00:03,  5.31it/s]
+
 62%|██████▏   | 24/39 [00:04<00:02,  5.38it/s]
+
 64%|██████▍   | 25/39 [00:04<00:02,  5.35it/s]
+
 67%|██████▋   | 26/39 [00:04<00:02,  5.31it/s]
+
 69%|██████▉   | 27/39 [00:04<00:02,  5.28it/s]
+
 72%|███████▏  | 28/39 [00:05<00:02,  5.25it/s]
+
 74%|███████▍  | 29/39 [00:05<00:01,  5.27it/s]
+
 77%|███████▋  | 30/39 [00:05<00:01,  5.30it/s]
+
 79%|███████▉  | 31/39 [00:05<00:01,  5.27it/s]
+
 82%|████████▏ | 32/39 [00:05<00:01,  5.26it/s]
+
 85%|████████▍ | 33/39 [00:06<00:01,  5.24it/s]
+
 87%|████████▋ | 34/39 [00:06<00:00,  5.28it/s]
+
 90%|████████▉ | 35/39 [00:06<00:00,  5.29it/s]
+
 92%|█████████▏| 36/39 [00:06<00:00,  5.26it/s]
+
 95%|█████████▍| 37/39 [00:06<00:00,  5.29it/s]
+
 97%|█████████▋| 38/39 [00:07<00:00,  5.28it/s]
+
100%|██████████| 39/39 [00:07<00:00,  5.25it/s]
                                                     
+
                                               
{'eval_loss': '3.534', 'eval_runtime': '7.441', 'eval_samples_per_second': '328.1', 'eval_steps_per_second': '5.241', 'epoch': '18.06'}
+
 72%|███████▏  | 5184/7175 [1:09:30<49:13,  1.48s/it]
+
100%|██████████| 39/39 [00:07<00:00,  5.25it/s][eval] epoch 18.063 step 5184 loss 3.5341 (best 3.5092 @ ep 16.811, 5/20 since best)
+
+
                                               
 72%|███████▏  | 5185/7175 [1:09:32<2:03:05,  3.71s/it]
 72%|███████▏  | 5186/7175 [1:09:33<1:40:52,  3.04s/it]
 72%|███████▏  | 5187/7175 [1:09:35<1:25:18,  2.57s/it]
 72%|███████▏  | 5188/7175 [1:09:36<1:14:22,  2.25s/it]
 72%|███████▏  | 5189/7175 [1:09:38<1:06:46,  2.02s/it]
 72%|███████▏  | 5190/7175 [1:09:39<1:01:23,  1.86s/it]
                                                       
{'loss': '2.998', 'grad_norm': '0.2747', 'learning_rate': '0.0006', 'epoch': '18.08'}
+
 72%|███████▏  | 5190/7175 [1:09:39<1:01:23,  1.86s/it]
 72%|███████▏  | 5191/7175 [1:09:41<57:48,  1.75s/it]  
 72%|███████▏  | 5192/7175 [1:09:42<55:08,  1.67s/it]
 72%|███████▏  | 5193/7175 [1:09:44<53:13,  1.61s/it]
 72%|███████▏  | 5194/7175 [1:09:45<51:53,  1.57s/it]
 72%|███████▏  | 5195/7175 [1:09:47<50:58,  1.54s/it]
 72%|███████▏  | 5196/7175 [1:09:48<50:20,  1.53s/it]
 72%|███████▏  | 5197/7175 [1:09:50<49:48,  1.51s/it]
 72%|███████▏  | 5198/7175 [1:09:51<49:33,  1.50s/it]
 72%|███████▏  | 5199/7175 [1:09:53<49:17,  1.50s/it]
 72%|███████▏  | 5200/7175 [1:09:54<49:10,  1.49s/it]
                                                     
{'loss': '2.994', 'grad_norm': '0.2297', 'learning_rate': '0.0006', 'epoch': '18.12'}
+
 72%|███████▏  | 5200/7175 [1:09:54<49:10,  1.49s/it]
 72%|███████▏  | 5201/7175 [1:09:56<49:04,  1.49s/it]
 73%|███████▎  | 5202/7175 [1:09:57<48:54,  1.49s/it]
 73%|███████▎  | 5203/7175 [1:09:59<48:53,  1.49s/it]
 73%|███████▎  | 5204/7175 [1:10:00<48:54,  1.49s/it]
 73%|███████▎  | 5205/7175 [1:10:02<48:53,  1.49s/it]
 73%|███████▎  | 5206/7175 [1:10:03<48:45,  1.49s/it]
 73%|███████▎  | 5207/7175 [1:10:04<48:45,  1.49s/it]
 73%|███████▎  | 5208/7175 [1:10:06<48:40,  1.48s/it]
 73%|███████▎  | 5209/7175 [1:10:07<48:33,  1.48s/it]
 73%|███████▎  | 5210/7175 [1:10:09<48:30,  1.48s/it]
                                                     
{'loss': '2.998', 'grad_norm': '0.2199', 'learning_rate': '0.0006', 'epoch': '18.15'}
+
 73%|███████▎  | 5210/7175 [1:10:09<48:30,  1.48s/it]
 73%|███████▎  | 5211/7175 [1:10:10<48:43,  1.49s/it]
 73%|███████▎  | 5212/7175 [1:10:12<48:38,  1.49s/it]
 73%|███████▎  | 5213/7175 [1:10:13<43:19,  1.32s/it]
 73%|███████▎  | 5214/7175 [1:10:14<40:01,  1.22s/it]
 73%|███████▎  | 5215/7175 [1:10:15<40:17,  1.23s/it]
 73%|███████▎  | 5216/7175 [1:10:17<42:34,  1.30s/it]
 73%|███████▎  | 5217/7175 [1:10:18<44:06,  1.35s/it]
 73%|███████▎  | 5218/7175 [1:10:20<45:08,  1.38s/it]
 73%|███████▎  | 5219/7175 [1:10:21<46:03,  1.41s/it]
 73%|███████▎  | 5220/7175 [1:10:22<46:46,  1.44s/it]
                                                     
{'loss': '3.004', 'grad_norm': '0.2281', 'learning_rate': '0.0006', 'epoch': '18.19'}
+
 73%|███████▎  | 5220/7175 [1:10:22<46:46,  1.44s/it]
 73%|███████▎  | 5221/7175 [1:10:24<47:17,  1.45s/it]
 73%|███████▎  | 5222/7175 [1:10:25<47:26,  1.46s/it]
 73%|███████▎  | 5223/7175 [1:10:27<47:38,  1.46s/it]
 73%|███████▎  | 5224/7175 [1:10:28<47:43,  1.47s/it]
 73%|███████▎  | 5225/7175 [1:10:30<47:48,  1.47s/it]
 73%|███████▎  | 5226/7175 [1:10:31<47:52,  1.47s/it]
 73%|███████▎  | 5227/7175 [1:10:33<47:54,  1.48s/it]
 73%|███████▎  | 5228/7175 [1:10:34<47:56,  1.48s/it]
 73%|███████▎  | 5229/7175 [1:10:36<47:57,  1.48s/it]
 73%|███████▎  | 5230/7175 [1:10:37<48:01,  1.48s/it]
                                                     
{'loss': '2.999', 'grad_norm': '0.2053', 'learning_rate': '0.0006', 'epoch': '18.22'}
+
 73%|███████▎  | 5230/7175 [1:10:37<48:01,  1.48s/it]
 73%|███████▎  | 5231/7175 [1:10:39<48:13,  1.49s/it]
 73%|███████▎  | 5232/7175 [1:10:40<48:03,  1.48s/it]
 73%|███████▎  | 5233/7175 [1:10:42<48:02,  1.48s/it]
 73%|███████▎  | 5234/7175 [1:10:43<47:59,  1.48s/it]
 73%|███████▎  | 5235/7175 [1:10:45<47:52,  1.48s/it]
 73%|███████▎  | 5236/7175 [1:10:46<47:52,  1.48s/it]
 73%|███████▎  | 5237/7175 [1:10:48<47:58,  1.49s/it]
 73%|███████▎  | 5238/7175 [1:10:49<47:47,  1.48s/it]
 73%|███████▎  | 5239/7175 [1:10:51<47:46,  1.48s/it]
 73%|███████▎  | 5240/7175 [1:10:52<47:45,  1.48s/it]
                                                     
{'loss': '3.002', 'grad_norm': '0.2113', 'learning_rate': '0.0006', 'epoch': '18.26'}
+
 73%|███████▎  | 5240/7175 [1:10:52<47:45,  1.48s/it]
 73%|███████▎  | 5241/7175 [1:10:54<47:55,  1.49s/it]
 73%|███████▎  | 5242/7175 [1:10:55<47:50,  1.49s/it]
 73%|███████▎  | 5243/7175 [1:10:57<47:43,  1.48s/it]
 73%|███████▎  | 5244/7175 [1:10:58<47:40,  1.48s/it]
 73%|███████▎  | 5245/7175 [1:11:00<47:42,  1.48s/it]
 73%|███████▎  | 5246/7175 [1:11:01<47:42,  1.48s/it]
 73%|███████▎  | 5247/7175 [1:11:03<47:45,  1.49s/it]
 73%|███████▎  | 5248/7175 [1:11:04<47:41,  1.48s/it]
 73%|███████▎  | 5249/7175 [1:11:05<47:43,  1.49s/it]
 73%|███████▎  | 5250/7175 [1:11:07<47:39,  1.49s/it]
                                                     
{'loss': '3.007', 'grad_norm': '0.2238', 'learning_rate': '0.0006', 'epoch': '18.29'}
+
 73%|███████▎  | 5250/7175 [1:11:07<47:39,  1.49s/it]
 73%|███████▎  | 5251/7175 [1:11:08<47:44,  1.49s/it]
 73%|███████▎  | 5252/7175 [1:11:10<47:45,  1.49s/it]
 73%|███████▎  | 5253/7175 [1:11:11<47:45,  1.49s/it]
 73%|███████▎  | 5254/7175 [1:11:13<47:40,  1.49s/it]
 73%|███████▎  | 5255/7175 [1:11:14<47:45,  1.49s/it]
 73%|███████▎  | 5256/7175 [1:11:16<47:46,  1.49s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:03, 10.55it/s]
+
 10%|█         | 4/39 [00:00<00:05,  6.64it/s]
+
 13%|█▎        | 5/39 [00:00<00:05,  6.18it/s]
+
 15%|█▌        | 6/39 [00:00<00:05,  5.85it/s]
+
 18%|█▊        | 7/39 [00:01<00:05,  5.64it/s]
+
 21%|██        | 8/39 [00:01<00:05,  5.54it/s]
+
 23%|██▎       | 9/39 [00:01<00:05,  5.52it/s]
+
 26%|██▌       | 10/39 [00:01<00:05,  5.48it/s]
+
 28%|██▊       | 11/39 [00:01<00:05,  5.46it/s]
+
 31%|███       | 12/39 [00:02<00:05,  5.38it/s]
+
 33%|███▎      | 13/39 [00:02<00:04,  5.33it/s]
+
 36%|███▌      | 14/39 [00:02<00:04,  5.32it/s]
+
 38%|███▊      | 15/39 [00:02<00:04,  5.33it/s]
+
 41%|████      | 16/39 [00:02<00:04,  5.29it/s]
+
 44%|████▎     | 17/39 [00:03<00:04,  5.31it/s]
+
 46%|████▌     | 18/39 [00:03<00:03,  5.27it/s]
+
 49%|████▊     | 19/39 [00:03<00:03,  5.25it/s]
+
 51%|█████▏    | 20/39 [00:03<00:03,  5.23it/s]
+
 54%|█████▍    | 21/39 [00:03<00:03,  5.30it/s]
+
 56%|█████▋    | 22/39 [00:03<00:03,  5.27it/s]
+
 59%|█████▉    | 23/39 [00:04<00:03,  5.26it/s]
+
 62%|██████▏   | 24/39 [00:04<00:02,  5.23it/s]
+
 64%|██████▍   | 25/39 [00:04<00:02,  5.26it/s]
+
 67%|██████▋   | 26/39 [00:04<00:02,  5.28it/s]
+
 69%|██████▉   | 27/39 [00:04<00:02,  5.30it/s]
+
 72%|███████▏  | 28/39 [00:05<00:02,  5.27it/s]
+
 74%|███████▍  | 29/39 [00:05<00:01,  5.24it/s]
+
 77%|███████▋  | 30/39 [00:05<00:01,  5.15it/s]
+
 79%|███████▉  | 31/39 [00:05<00:01,  5.15it/s]
+
 82%|████████▏ | 32/39 [00:05<00:01,  5.17it/s]
+
 85%|████████▍ | 33/39 [00:06<00:01,  5.26it/s]
+
 87%|████████▋ | 34/39 [00:06<00:00,  5.24it/s]
+
 90%|████████▉ | 35/39 [00:06<00:00,  5.23it/s]
+
 92%|█████████▏| 36/39 [00:06<00:00,  5.26it/s]
+
 95%|█████████▍| 37/39 [00:06<00:00,  5.30it/s]
+
 97%|█████████▋| 38/39 [00:07<00:00,  5.39it/s]
+
100%|██████████| 39/39 [00:07<00:00,  5.40it/s]
                                                     
+
                                               
{'eval_loss': '3.518', 'eval_runtime': '7.437', 'eval_samples_per_second': '328.2', 'eval_steps_per_second': '5.244', 'epoch': '18.31'}
+
 73%|███████▎  | 5256/7175 [1:11:23<47:46,  1.49s/it]
+
100%|██████████| 39/39 [00:07<00:00,  5.40it/s][eval] epoch 18.315 step 5256 loss 3.5182 (best 3.5092 @ ep 16.811, 6/20 since best)
+
+
                                               
 73%|███████▎  | 5257/7175 [1:11:25<1:58:53,  3.72s/it]
 73%|███████▎  | 5258/7175 [1:11:26<1:37:19,  3.05s/it]
 73%|███████▎  | 5259/7175 [1:11:28<1:22:12,  2.57s/it]
 73%|███████▎  | 5260/7175 [1:11:29<1:11:46,  2.25s/it]
                                                       
{'loss': '3.019', 'grad_norm': '0.2269', 'learning_rate': '0.0006', 'epoch': '18.33'}
+
 73%|███████▎  | 5260/7175 [1:11:29<1:11:46,  2.25s/it]
 73%|███████▎  | 5261/7175 [1:11:31<1:04:37,  2.03s/it]
 73%|███████▎  | 5262/7175 [1:11:32<59:28,  1.87s/it]  
 73%|███████▎  | 5263/7175 [1:11:34<55:51,  1.75s/it]
 73%|███████▎  | 5264/7175 [1:11:35<53:15,  1.67s/it]
 73%|███████▎  | 5265/7175 [1:11:37<51:32,  1.62s/it]
 73%|███████▎  | 5266/7175 [1:11:38<50:09,  1.58s/it]
 73%|███████▎  | 5267/7175 [1:11:40<49:15,  1.55s/it]
 73%|███████▎  | 5268/7175 [1:11:41<48:35,  1.53s/it]
 73%|███████▎  | 5269/7175 [1:11:43<48:08,  1.52s/it]
 73%|███████▎  | 5270/7175 [1:11:44<47:48,  1.51s/it]
                                                     
{'loss': '3.025', 'grad_norm': '0.2312', 'learning_rate': '0.0006', 'epoch': '18.36'}
+
 73%|███████▎  | 5270/7175 [1:11:44<47:48,  1.51s/it]
 73%|███████▎  | 5271/7175 [1:11:46<47:45,  1.51s/it]
 73%|███████▎  | 5272/7175 [1:11:47<47:31,  1.50s/it]
 73%|███████▎  | 5273/7175 [1:11:49<47:20,  1.49s/it]
 74%|███████▎  | 5274/7175 [1:11:50<47:11,  1.49s/it]
 74%|███████▎  | 5275/7175 [1:11:52<47:06,  1.49s/it]
 74%|███████▎  | 5276/7175 [1:11:53<47:04,  1.49s/it]
 74%|███████▎  | 5277/7175 [1:11:55<46:57,  1.48s/it]
 74%|███████▎  | 5278/7175 [1:11:56<46:55,  1.48s/it]
 74%|███████▎  | 5279/7175 [1:11:58<46:46,  1.48s/it]
 74%|███████▎  | 5280/7175 [1:11:59<46:44,  1.48s/it]
                                                     
{'loss': '3.006', 'grad_norm': '0.2086', 'learning_rate': '0.0006', 'epoch': '18.4'}
+
 74%|███████▎  | 5280/7175 [1:11:59<46:44,  1.48s/it]
 74%|███████▎  | 5281/7175 [1:12:00<46:53,  1.49s/it]
 74%|███████▎  | 5282/7175 [1:12:02<46:48,  1.48s/it]
 74%|███████▎  | 5283/7175 [1:12:03<46:41,  1.48s/it]
 74%|███████▎  | 5284/7175 [1:12:05<46:40,  1.48s/it]
 74%|███████▎  | 5285/7175 [1:12:06<46:38,  1.48s/it]
 74%|███████▎  | 5286/7175 [1:12:08<46:36,  1.48s/it]
 74%|███████▎  | 5287/7175 [1:12:09<46:38,  1.48s/it]
 74%|███████▎  | 5288/7175 [1:12:11<46:40,  1.48s/it]
 74%|███████▎  | 5289/7175 [1:12:12<46:36,  1.48s/it]
 74%|███████▎  | 5290/7175 [1:12:14<46:31,  1.48s/it]
                                                     
{'loss': '3.025', 'grad_norm': '0.2297', 'learning_rate': '0.0006', 'epoch': '18.43'}
+
 74%|███████▎  | 5290/7175 [1:12:14<46:31,  1.48s/it]
 74%|███████▎  | 5291/7175 [1:12:15<46:35,  1.48s/it]
 74%|███████▍  | 5292/7175 [1:12:17<46:28,  1.48s/it]
 74%|███████▍  | 5293/7175 [1:12:18<46:25,  1.48s/it]
 74%|███████▍  | 5294/7175 [1:12:20<46:14,  1.47s/it]
 74%|███████▍  | 5295/7175 [1:12:21<46:16,  1.48s/it]
 74%|███████▍  | 5296/7175 [1:12:23<46:19,  1.48s/it]
 74%|███████▍  | 5297/7175 [1:12:24<46:20,  1.48s/it]
 74%|███████▍  | 5298/7175 [1:12:26<46:16,  1.48s/it]
 74%|███████▍  | 5299/7175 [1:12:27<46:00,  1.47s/it]
 74%|███████▍  | 5300/7175 [1:12:29<46:03,  1.47s/it]
                                                     
{'loss': '3.016', 'grad_norm': '0.2214', 'learning_rate': '0.0006', 'epoch': '18.47'}
+
 74%|███████▍  | 5300/7175 [1:12:29<46:03,  1.47s/it]
 74%|███████▍  | 5301/7175 [1:12:30<46:15,  1.48s/it]
 74%|███████▍  | 5302/7175 [1:12:32<46:11,  1.48s/it]
 74%|███████▍  | 5303/7175 [1:12:33<46:12,  1.48s/it]
 74%|███████▍  | 5304/7175 [1:12:35<46:07,  1.48s/it]
 74%|███████▍  | 5305/7175 [1:12:36<46:07,  1.48s/it]
 74%|███████▍  | 5306/7175 [1:12:37<46:06,  1.48s/it]
 74%|███████▍  | 5307/7175 [1:12:39<46:02,  1.48s/it]
 74%|███████▍  | 5308/7175 [1:12:40<46:01,  1.48s/it]
 74%|███████▍  | 5309/7175 [1:12:42<45:59,  1.48s/it]
 74%|███████▍  | 5310/7175 [1:12:43<45:56,  1.48s/it]
                                                     
{'loss': '3.017', 'grad_norm': '0.2561', 'learning_rate': '0.0006', 'epoch': '18.5'}
+
 74%|███████▍  | 5310/7175 [1:12:43<45:56,  1.48s/it]
 74%|███████▍  | 5311/7175 [1:12:45<45:57,  1.48s/it]
 74%|███████▍  | 5312/7175 [1:12:46<45:55,  1.48s/it]
 74%|███████▍  | 5313/7175 [1:12:48<45:54,  1.48s/it]
 74%|███████▍  | 5314/7175 [1:12:49<45:46,  1.48s/it]
 74%|███████▍  | 5315/7175 [1:12:51<45:35,  1.47s/it]
 74%|███████▍  | 5316/7175 [1:12:52<45:41,  1.47s/it]
 74%|███████▍  | 5317/7175 [1:12:54<45:41,  1.48s/it]
 74%|███████▍  | 5318/7175 [1:12:55<45:42,  1.48s/it]
 74%|███████▍  | 5319/7175 [1:12:57<45:42,  1.48s/it]
 74%|███████▍  | 5320/7175 [1:12:58<45:43,  1.48s/it]
                                                     
{'loss': '3.017', 'grad_norm': '0.2184', 'learning_rate': '0.0006', 'epoch': '18.54'}
+
 74%|███████▍  | 5320/7175 [1:12:58<45:43,  1.48s/it]
 74%|███████▍  | 5321/7175 [1:13:00<45:51,  1.48s/it]
 74%|███████▍  | 5322/7175 [1:13:01<45:51,  1.48s/it]
 74%|███████▍  | 5323/7175 [1:13:03<45:49,  1.48s/it]
 74%|███████▍  | 5324/7175 [1:13:04<45:41,  1.48s/it]
 74%|███████▍  | 5325/7175 [1:13:06<45:42,  1.48s/it]
 74%|███████▍  | 5326/7175 [1:13:07<45:39,  1.48s/it]
 74%|███████▍  | 5327/7175 [1:13:09<45:36,  1.48s/it]
 74%|███████▍  | 5328/7175 [1:13:10<45:33,  1.48s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:03, 10.37it/s]
+
 10%|█         | 4/39 [00:00<00:05,  6.44it/s]
+
 13%|█▎        | 5/39 [00:00<00:05,  6.01it/s]
+
 15%|█▌        | 6/39 [00:00<00:05,  5.66it/s]
+
 18%|█▊        | 7/39 [00:01<00:05,  5.51it/s]
+
 21%|██        | 8/39 [00:01<00:05,  5.46it/s]
+
 23%|██▎       | 9/39 [00:01<00:05,  5.41it/s]
+
 26%|██▌       | 10/39 [00:01<00:05,  5.37it/s]
+
 28%|██▊       | 11/39 [00:01<00:05,  5.31it/s]
+
 31%|███       | 12/39 [00:02<00:05,  5.28it/s]
+
 33%|███▎      | 13/39 [00:02<00:04,  5.26it/s]
+
 36%|███▌      | 14/39 [00:02<00:04,  5.32it/s]
+
 38%|███▊      | 15/39 [00:02<00:04,  5.28it/s]
+
 41%|████      | 16/39 [00:02<00:04,  5.26it/s]
+
 44%|████▎     | 17/39 [00:03<00:04,  5.25it/s]
+
 46%|████▌     | 18/39 [00:03<00:03,  5.26it/s]
+
 49%|████▊     | 19/39 [00:03<00:03,  5.24it/s]
+
 51%|█████▏    | 20/39 [00:03<00:03,  5.27it/s]
+
 54%|█████▍    | 21/39 [00:03<00:03,  5.25it/s]
+
 56%|█████▋    | 22/39 [00:04<00:03,  5.24it/s]
+
 59%|█████▉    | 23/39 [00:04<00:03,  5.23it/s]
+
 62%|██████▏   | 24/39 [00:04<00:02,  5.22it/s]
+
 64%|██████▍   | 25/39 [00:04<00:02,  5.22it/s]
+
 67%|██████▋   | 26/39 [00:04<00:02,  5.32it/s]
+
 69%|██████▉   | 27/39 [00:04<00:02,  5.34it/s]
+
 72%|███████▏  | 28/39 [00:05<00:02,  5.36it/s]
+
 74%|███████▍  | 29/39 [00:05<00:01,  5.32it/s]
+
 77%|███████▋  | 30/39 [00:05<00:01,  5.32it/s]
+
 79%|███████▉  | 31/39 [00:05<00:01,  5.29it/s]
+
 82%|████████▏ | 32/39 [00:05<00:01,  5.31it/s]
+
 85%|████████▍ | 33/39 [00:06<00:01,  5.27it/s]
+
 87%|████████▋ | 34/39 [00:06<00:00,  5.25it/s]
+
 90%|████████▉ | 35/39 [00:06<00:00,  5.23it/s]
+
 92%|█████████▏| 36/39 [00:06<00:00,  5.22it/s]
+
 95%|█████████▍| 37/39 [00:06<00:00,  5.22it/s]
+
 97%|█████████▋| 38/39 [00:07<00:00,  5.30it/s]
+
100%|██████████| 39/39 [00:07<00:00,  5.26it/s]
                                                     
+
                                               
{'eval_loss': '3.51', 'eval_runtime': '7.49', 'eval_samples_per_second': '325.9', 'eval_steps_per_second': '5.207', 'epoch': '18.57'}
+
 74%|███████▍  | 5328/7175 [1:13:18<45:33,  1.48s/it]
+
100%|██████████| 39/39 [00:07<00:00,  5.26it/s][eval] epoch 18.566 step 5328 loss 3.5096 (best 3.5092 @ ep 16.811, 7/20 since best)
+
+
                                               
 74%|███████▍  | 5329/7175 [1:13:19<1:54:50,  3.73s/it]
 74%|███████▍  | 5330/7175 [1:13:20<1:34:04,  3.06s/it]
                                                       
{'loss': '3.037', 'grad_norm': '0.2458', 'learning_rate': '0.0006', 'epoch': '18.57'}
+
 74%|███████▍  | 5330/7175 [1:13:21<1:34:04,  3.06s/it]
 74%|███████▍  | 5331/7175 [1:13:22<1:19:32,  2.59s/it]
 74%|███████▍  | 5332/7175 [1:13:23<1:09:18,  2.26s/it]
 74%|███████▍  | 5333/7175 [1:13:25<1:02:04,  2.02s/it]
 74%|███████▍  | 5334/7175 [1:13:26<57:06,  1.86s/it]  
 74%|███████▍  | 5335/7175 [1:13:28<53:32,  1.75s/it]
 74%|███████▍  | 5336/7175 [1:13:29<51:03,  1.67s/it]
 74%|███████▍  | 5337/7175 [1:13:31<49:21,  1.61s/it]
 74%|███████▍  | 5338/7175 [1:13:32<48:06,  1.57s/it]
 74%|███████▍  | 5339/7175 [1:13:34<47:12,  1.54s/it]
 74%|███████▍  | 5340/7175 [1:13:35<46:33,  1.52s/it]
                                                     
{'loss': '3.027', 'grad_norm': '0.2171', 'learning_rate': '0.0006', 'epoch': '18.61'}
+
 74%|███████▍  | 5340/7175 [1:13:35<46:33,  1.52s/it]
 74%|███████▍  | 5341/7175 [1:13:37<46:20,  1.52s/it]
 74%|███████▍  | 5342/7175 [1:13:38<46:00,  1.51s/it]
 74%|███████▍  | 5343/7175 [1:13:40<45:43,  1.50s/it]
 74%|███████▍  | 5344/7175 [1:13:41<45:21,  1.49s/it]
 74%|███████▍  | 5345/7175 [1:13:43<45:13,  1.48s/it]
 75%|███████▍  | 5346/7175 [1:13:44<45:11,  1.48s/it]
 75%|███████▍  | 5347/7175 [1:13:46<45:09,  1.48s/it]
 75%|███████▍  | 5348/7175 [1:13:47<45:10,  1.48s/it]
 75%|███████▍  | 5349/7175 [1:13:49<45:07,  1.48s/it]
 75%|███████▍  | 5350/7175 [1:13:50<45:05,  1.48s/it]
                                                     
{'loss': '3.041', 'grad_norm': '0.2261', 'learning_rate': '0.0006', 'epoch': '18.64'}
+
 75%|███████▍  | 5350/7175 [1:13:50<45:05,  1.48s/it]
 75%|███████▍  | 5351/7175 [1:13:52<45:12,  1.49s/it]
 75%|███████▍  | 5352/7175 [1:13:53<45:07,  1.49s/it]
 75%|███████▍  | 5353/7175 [1:13:55<45:01,  1.48s/it]
 75%|███████▍  | 5354/7175 [1:13:56<44:57,  1.48s/it]
 75%|███████▍  | 5355/7175 [1:13:58<44:55,  1.48s/it]
 75%|███████▍  | 5356/7175 [1:13:59<44:57,  1.48s/it]
 75%|███████▍  | 5357/7175 [1:14:00<44:50,  1.48s/it]
 75%|███████▍  | 5358/7175 [1:14:02<44:49,  1.48s/it]
 75%|███████▍  | 5359/7175 [1:14:03<44:48,  1.48s/it]
 75%|███████▍  | 5360/7175 [1:14:05<44:45,  1.48s/it]
                                                     
{'loss': '3.039', 'grad_norm': '0.1977', 'learning_rate': '0.0006', 'epoch': '18.68'}
+
 75%|███████▍  | 5360/7175 [1:14:05<44:45,  1.48s/it]
 75%|███████▍  | 5361/7175 [1:14:06<44:57,  1.49s/it]
 75%|███████▍  | 5362/7175 [1:14:07<40:57,  1.36s/it]
 75%|███████▍  | 5363/7175 [1:14:08<36:01,  1.19s/it]
 75%|███████▍  | 5364/7175 [1:14:10<36:51,  1.22s/it]
 75%|███████▍  | 5365/7175 [1:14:11<39:01,  1.29s/it]
 75%|███████▍  | 5366/7175 [1:14:12<40:33,  1.35s/it]
 75%|███████▍  | 5367/7175 [1:14:14<41:36,  1.38s/it]
 75%|███████▍  | 5368/7175 [1:14:15<42:27,  1.41s/it]
 75%|███████▍  | 5369/7175 [1:14:17<43:07,  1.43s/it]
 75%|███████▍  | 5370/7175 [1:14:18<43:28,  1.44s/it]
                                                     
{'loss': '3.04', 'grad_norm': '0.2118', 'learning_rate': '0.0006', 'epoch': '18.71'}
+
 75%|███████▍  | 5370/7175 [1:14:18<43:28,  1.44s/it]
 75%|███████▍  | 5371/7175 [1:14:20<43:55,  1.46s/it]
 75%|███████▍  | 5372/7175 [1:14:21<44:08,  1.47s/it]
 75%|███████▍  | 5373/7175 [1:14:23<44:10,  1.47s/it]
 75%|███████▍  | 5374/7175 [1:14:24<44:16,  1.47s/it]
 75%|███████▍  | 5375/7175 [1:14:26<44:16,  1.48s/it]
 75%|███████▍  | 5376/7175 [1:14:27<44:19,  1.48s/it]
 75%|███████▍  | 5377/7175 [1:14:29<44:19,  1.48s/it]
 75%|███████▍  | 5378/7175 [1:14:30<44:16,  1.48s/it]
 75%|███████▍  | 5379/7175 [1:14:32<44:14,  1.48s/it]
 75%|███████▍  | 5380/7175 [1:14:33<44:13,  1.48s/it]
                                                     
{'loss': '3.043', 'grad_norm': '0.2049', 'learning_rate': '0.0006', 'epoch': '18.75'}
+
 75%|███████▍  | 5380/7175 [1:14:33<44:13,  1.48s/it]
 75%|███████▍  | 5381/7175 [1:14:35<44:21,  1.48s/it]
 75%|███████▌  | 5382/7175 [1:14:36<44:18,  1.48s/it]
 75%|███████▌  | 5383/7175 [1:14:38<44:17,  1.48s/it]
 75%|███████▌  | 5384/7175 [1:14:39<44:15,  1.48s/it]
 75%|███████▌  | 5385/7175 [1:14:41<44:09,  1.48s/it]
 75%|███████▌  | 5386/7175 [1:14:42<44:09,  1.48s/it]
 75%|███████▌  | 5387/7175 [1:14:44<44:09,  1.48s/it]
 75%|███████▌  | 5388/7175 [1:14:45<44:07,  1.48s/it]
 75%|███████▌  | 5389/7175 [1:14:47<44:06,  1.48s/it]
 75%|███████▌  | 5390/7175 [1:14:48<44:01,  1.48s/it]
                                                     
{'loss': '3.047', 'grad_norm': '0.2259', 'learning_rate': '0.0006', 'epoch': '18.78'}
+
 75%|███████▌  | 5390/7175 [1:14:48<44:01,  1.48s/it]
 75%|███████▌  | 5391/7175 [1:14:50<44:07,  1.48s/it]
 75%|███████▌  | 5392/7175 [1:14:51<44:06,  1.48s/it]
 75%|███████▌  | 5393/7175 [1:14:52<44:01,  1.48s/it]
 75%|███████▌  | 5394/7175 [1:14:54<43:55,  1.48s/it]
 75%|███████▌  | 5395/7175 [1:14:55<43:57,  1.48s/it]
 75%|███████▌  | 5396/7175 [1:14:57<43:54,  1.48s/it]
 75%|███████▌  | 5397/7175 [1:14:58<43:54,  1.48s/it]
 75%|███████▌  | 5398/7175 [1:15:00<43:51,  1.48s/it]
 75%|███████▌  | 5399/7175 [1:15:01<43:50,  1.48s/it]
 75%|███████▌  | 5400/7175 [1:15:03<43:50,  1.48s/it]
                                                     
{'loss': '3.025', 'grad_norm': '0.2087', 'learning_rate': '0.0006', 'epoch': '18.82'}
+
 75%|███████▌  | 5400/7175 [1:15:03<43:50,  1.48s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:03, 10.81it/s]
+
 10%|█         | 4/39 [00:00<00:05,  6.78it/s]
+
 13%|█▎        | 5/39 [00:00<00:05,  6.32it/s]
+
 15%|█▌        | 6/39 [00:00<00:05,  5.94it/s]
+
 18%|█▊        | 7/39 [00:01<00:05,  5.70it/s]
+
 21%|██        | 8/39 [00:01<00:05,  5.57it/s]
+
 23%|██▎       | 9/39 [00:01<00:05,  5.47it/s]
+
 26%|██▌       | 10/39 [00:01<00:05,  5.43it/s]
+
 28%|██▊       | 11/39 [00:01<00:05,  5.36it/s]
+
 31%|███       | 12/39 [00:02<00:05,  5.31it/s]
+
 33%|███▎      | 13/39 [00:02<00:04,  5.28it/s]
+
 36%|███▌      | 14/39 [00:02<00:04,  5.25it/s]
+
 38%|███▊      | 15/39 [00:02<00:04,  5.30it/s]
+
 41%|████      | 16/39 [00:02<00:04,  5.34it/s]
+
 44%|████▎     | 17/39 [00:03<00:04,  5.30it/s]
+
 46%|████▌     | 18/39 [00:03<00:04,  5.24it/s]
+
 49%|████▊     | 19/39 [00:03<00:03,  5.26it/s]
+
 51%|█████▏    | 20/39 [00:03<00:03,  5.24it/s]
+
 54%|█████▍    | 21/39 [00:03<00:03,  5.24it/s]
+
 56%|█████▋    | 22/39 [00:03<00:03,  5.28it/s]
+
 59%|█████▉    | 23/39 [00:04<00:03,  5.29it/s]
+
 62%|██████▏   | 24/39 [00:04<00:02,  5.26it/s]
+
 64%|██████▍   | 25/39 [00:04<00:02,  5.25it/s]
+
 67%|██████▋   | 26/39 [00:04<00:02,  5.23it/s]
+
 69%|██████▉   | 27/39 [00:04<00:02,  5.26it/s]
+
 72%|███████▏  | 28/39 [00:05<00:02,  5.29it/s]
+
 74%|███████▍  | 29/39 [00:05<00:01,  5.26it/s]
+
 77%|███████▋  | 30/39 [00:05<00:01,  5.24it/s]
+
 79%|███████▉  | 31/39 [00:05<00:01,  5.23it/s]
+
 82%|████████▏ | 32/39 [00:05<00:01,  5.25it/s]
+
 85%|████████▍ | 33/39 [00:06<00:01,  5.24it/s]
+
 87%|████████▋ | 34/39 [00:06<00:00,  5.28it/s]
+
 90%|████████▉ | 35/39 [00:06<00:00,  5.25it/s]
+
 92%|█████████▏| 36/39 [00:06<00:00,  5.27it/s]
+
 95%|█████████▍| 37/39 [00:06<00:00,  5.25it/s]
+
 97%|█████████▋| 38/39 [00:07<00:00,  5.23it/s]
+
100%|██████████| 39/39 [00:07<00:00,  5.22it/s]
                                                     
+
                                               
{'eval_loss': '3.504', 'eval_runtime': '7.472', 'eval_samples_per_second': '326.7', 'eval_steps_per_second': '5.22', 'epoch': '18.82'}
+
 75%|███████▌  | 5400/7175 [1:15:10<43:50,  1.48s/it]
+
100%|██████████| 39/39 [00:07<00:00,  5.22it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-5400
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-5400/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-5400/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.68it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.68it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-5400/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-5400/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-5400/tokenizer_config.json
+[eval] epoch 18.818 step 5400 loss 3.5036 (best 3.5036 @ ep 18.818, 0/20 since best)
+
+
                                               
 75%|███████▌  | 5401/7175 [1:15:14<2:09:05,  4.37s/it]
 75%|███████▌  | 5402/7175 [1:15:15<1:43:25,  3.50s/it]
 75%|███████▌  | 5403/7175 [1:15:17<1:25:29,  2.89s/it]
 75%|███████▌  | 5404/7175 [1:15:18<1:12:59,  2.47s/it]
 75%|███████▌  | 5405/7175 [1:15:20<1:04:08,  2.17s/it]
 75%|███████▌  | 5406/7175 [1:15:21<57:58,  1.97s/it]  
 75%|███████▌  | 5407/7175 [1:15:23<53:38,  1.82s/it]
 75%|███████▌  | 5408/7175 [1:15:24<50:36,  1.72s/it]
 75%|███████▌  | 5409/7175 [1:15:26<48:19,  1.64s/it]
 75%|███████▌  | 5410/7175 [1:15:27<46:56,  1.60s/it]
                                                     
{'loss': '3.039', 'grad_norm': '0.2347', 'learning_rate': '0.0006', 'epoch': '18.85'}
+
 75%|███████▌  | 5410/7175 [1:15:27<46:56,  1.60s/it]
 75%|███████▌  | 5411/7175 [1:15:29<46:04,  1.57s/it]
 75%|███████▌  | 5412/7175 [1:15:30<45:17,  1.54s/it]
 75%|███████▌  | 5413/7175 [1:15:32<44:46,  1.52s/it]
 75%|███████▌  | 5414/7175 [1:15:33<44:06,  1.50s/it]
 75%|███████▌  | 5415/7175 [1:15:35<43:51,  1.50s/it]
 75%|███████▌  | 5416/7175 [1:15:36<43:42,  1.49s/it]
 75%|███████▌  | 5417/7175 [1:15:38<43:35,  1.49s/it]
 76%|███████▌  | 5418/7175 [1:15:39<43:33,  1.49s/it]
 76%|███████▌  | 5419/7175 [1:15:41<43:31,  1.49s/it]
 76%|███████▌  | 5420/7175 [1:15:42<43:25,  1.48s/it]
                                                     
{'loss': '3.034', 'grad_norm': '0.2084', 'learning_rate': '0.0006', 'epoch': '18.89'}
+
 76%|███████▌  | 5420/7175 [1:15:42<43:25,  1.48s/it]
 76%|███████▌  | 5421/7175 [1:15:44<43:32,  1.49s/it]
 76%|███████▌  | 5422/7175 [1:15:45<43:22,  1.48s/it]
 76%|███████▌  | 5423/7175 [1:15:47<43:16,  1.48s/it]
 76%|███████▌  | 5424/7175 [1:15:48<43:18,  1.48s/it]
 76%|███████▌  | 5425/7175 [1:15:50<43:16,  1.48s/it]
 76%|███████▌  | 5426/7175 [1:15:51<43:17,  1.49s/it]
 76%|███████▌  | 5427/7175 [1:15:52<43:12,  1.48s/it]
 76%|███████▌  | 5428/7175 [1:15:54<43:09,  1.48s/it]
 76%|███████▌  | 5429/7175 [1:15:55<43:09,  1.48s/it]
 76%|███████▌  | 5430/7175 [1:15:57<43:05,  1.48s/it]
                                                     
{'loss': '3.043', 'grad_norm': '0.2102', 'learning_rate': '0.0006', 'epoch': '18.92'}
+
 76%|███████▌  | 5430/7175 [1:15:57<43:05,  1.48s/it]
 76%|███████▌  | 5431/7175 [1:15:58<43:10,  1.49s/it]
 76%|███████▌  | 5432/7175 [1:16:00<43:06,  1.48s/it]
 76%|███████▌  | 5433/7175 [1:16:01<43:05,  1.48s/it]
 76%|███████▌  | 5434/7175 [1:16:03<43:00,  1.48s/it]
 76%|███████▌  | 5435/7175 [1:16:04<42:56,  1.48s/it]
 76%|███████▌  | 5436/7175 [1:16:06<42:58,  1.48s/it]
 76%|███████▌  | 5437/7175 [1:16:07<42:59,  1.48s/it]
 76%|███████▌  | 5438/7175 [1:16:09<42:40,  1.47s/it]
 76%|███████▌  | 5439/7175 [1:16:10<42:41,  1.48s/it]
 76%|███████▌  | 5440/7175 [1:16:12<42:44,  1.48s/it]
                                                     
{'loss': '3.05', 'grad_norm': '0.1969', 'learning_rate': '0.0006', 'epoch': '18.96'}
+
 76%|███████▌  | 5440/7175 [1:16:12<42:44,  1.48s/it]
 76%|███████▌  | 5441/7175 [1:16:13<42:52,  1.48s/it]
 76%|███████▌  | 5442/7175 [1:16:15<42:50,  1.48s/it]
 76%|███████▌  | 5443/7175 [1:16:16<42:45,  1.48s/it]
 76%|███████▌  | 5444/7175 [1:16:18<42:42,  1.48s/it]
 76%|███████▌  | 5445/7175 [1:16:19<42:42,  1.48s/it]
 76%|███████▌  | 5446/7175 [1:16:21<42:38,  1.48s/it]
 76%|███████▌  | 5447/7175 [1:16:22<42:34,  1.48s/it]
 76%|███████▌  | 5448/7175 [1:16:24<42:31,  1.48s/it]
 76%|███████▌  | 5449/7175 [1:16:25<42:32,  1.48s/it]
 76%|███████▌  | 5450/7175 [1:16:27<42:45,  1.49s/it]
                                                     
{'loss': '3.056', 'grad_norm': '0.2169', 'learning_rate': '0.0006', 'epoch': '18.99'}
+
 76%|███████▌  | 5450/7175 [1:16:27<42:45,  1.49s/it]
 76%|███████▌  | 5451/7175 [1:16:28<42:44,  1.49s/it]
 76%|███████▌  | 5452/7175 [1:16:30<42:37,  1.48s/it]
 76%|███████▌  | 5453/7175 [1:16:30<31:49,  1.11s/it]
 76%|███████▌  | 5454/7175 [1:16:31<35:55,  1.25s/it]
 76%|███████▌  | 5455/7175 [1:16:33<37:53,  1.32s/it]
 76%|███████▌  | 5456/7175 [1:16:34<39:11,  1.37s/it]
 76%|███████▌  | 5457/7175 [1:16:36<40:08,  1.40s/it]
 76%|███████▌  | 5458/7175 [1:16:37<40:47,  1.43s/it]
 76%|███████▌  | 5459/7175 [1:16:39<41:15,  1.44s/it]
 76%|███████▌  | 5460/7175 [1:16:40<41:31,  1.45s/it]
                                                     
{'loss': '3.015', 'grad_norm': '0.3263', 'learning_rate': '0.0006', 'epoch': '19.02'}
+
 76%|███████▌  | 5460/7175 [1:16:40<41:31,  1.45s/it]
 76%|███████▌  | 5461/7175 [1:16:42<41:51,  1.47s/it]
 76%|███████▌  | 5462/7175 [1:16:43<41:55,  1.47s/it]
 76%|███████▌  | 5463/7175 [1:16:45<41:58,  1.47s/it]
 76%|███████▌  | 5464/7175 [1:16:46<42:02,  1.47s/it]
 76%|███████▌  | 5465/7175 [1:16:48<42:04,  1.48s/it]
 76%|███████▌  | 5466/7175 [1:16:49<42:03,  1.48s/it]
 76%|███████▌  | 5467/7175 [1:16:51<42:04,  1.48s/it]
 76%|███████▌  | 5468/7175 [1:16:52<42:01,  1.48s/it]
 76%|███████▌  | 5469/7175 [1:16:54<42:02,  1.48s/it]
 76%|███████▌  | 5470/7175 [1:16:55<42:03,  1.48s/it]
                                                     
{'loss': '2.989', 'grad_norm': '0.2491', 'learning_rate': '0.0006', 'epoch': '19.06'}
+
 76%|███████▌  | 5470/7175 [1:16:55<42:03,  1.48s/it]
 76%|███████▋  | 5471/7175 [1:16:57<42:13,  1.49s/it]
 76%|███████▋  | 5472/7175 [1:16:58<42:07,  1.48s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:03, 10.36it/s]
+
 10%|█         | 4/39 [00:00<00:05,  6.54it/s]
+
 13%|█▎        | 5/39 [00:00<00:05,  6.14it/s]
+
 15%|█▌        | 6/39 [00:00<00:05,  5.87it/s]
+
 18%|█▊        | 7/39 [00:01<00:05,  5.65it/s]
+
 21%|██        | 8/39 [00:01<00:05,  5.52it/s]
+
 23%|██▎       | 9/39 [00:01<00:05,  5.45it/s]
+
 26%|██▌       | 10/39 [00:01<00:05,  5.37it/s]
+
 28%|██▊       | 11/39 [00:01<00:05,  5.33it/s]
+
 31%|███       | 12/39 [00:02<00:05,  5.38it/s]
+
 33%|███▎      | 13/39 [00:02<00:04,  5.39it/s]
+
 36%|███▌      | 14/39 [00:02<00:04,  5.34it/s]
+
 38%|███▊      | 15/39 [00:02<00:04,  5.33it/s]
+
 41%|████      | 16/39 [00:02<00:04,  5.35it/s]
+
 44%|████▎     | 17/39 [00:03<00:04,  5.41it/s]
+
 46%|████▌     | 18/39 [00:03<00:03,  5.41it/s]
+
 49%|████▊     | 19/39 [00:03<00:03,  5.41it/s]
+
 51%|█████▏    | 20/39 [00:03<00:03,  5.35it/s]
+
 54%|█████▍    | 21/39 [00:03<00:03,  5.30it/s]
+
 56%|█████▋    | 22/39 [00:03<00:03,  5.27it/s]
+
 59%|█████▉    | 23/39 [00:04<00:02,  5.34it/s]
+
 62%|██████▏   | 24/39 [00:04<00:02,  5.30it/s]
+
 64%|██████▍   | 25/39 [00:04<00:02,  5.27it/s]
+
 67%|██████▋   | 26/39 [00:04<00:02,  5.17it/s]
+
 69%|██████▉   | 27/39 [00:04<00:02,  5.20it/s]
+
 72%|███████▏  | 28/39 [00:05<00:02,  5.20it/s]
+
 74%|███████▍  | 29/39 [00:05<00:01,  5.25it/s]
+
 77%|███████▋  | 30/39 [00:05<00:01,  5.23it/s]
+
 79%|███████▉  | 31/39 [00:05<00:01,  5.26it/s]
+
 82%|████████▏ | 32/39 [00:05<00:01,  5.24it/s]
+
 85%|████████▍ | 33/39 [00:06<00:01,  5.23it/s]
+
 87%|████████▋ | 34/39 [00:06<00:00,  5.22it/s]
+
 90%|████████▉ | 35/39 [00:06<00:00,  5.30it/s]
+
 92%|█████████▏| 36/39 [00:06<00:00,  5.27it/s]
+
 95%|█████████▍| 37/39 [00:06<00:00,  5.18it/s]
+
 97%|█████████▋| 38/39 [00:07<00:00,  5.17it/s]
+
100%|██████████| 39/39 [00:07<00:00,  5.21it/s]
                                                     
+
                                               
{'eval_loss': '3.526', 'eval_runtime': '7.462', 'eval_samples_per_second': '327.1', 'eval_steps_per_second': '5.227', 'epoch': '19.07'}
+
 76%|███████▋  | 5472/7175 [1:17:05<42:07,  1.48s/it]
+
100%|██████████| 39/39 [00:07<00:00,  5.21it/s][eval] epoch 19.066 step 5472 loss 3.5265 (best 3.5036 @ ep 18.818, 1/20 since best)
+
+
                                               
 76%|███████▋  | 5473/7175 [1:17:07<1:45:20,  3.71s/it]
 76%|███████▋  | 5474/7175 [1:17:08<1:26:09,  3.04s/it]
 76%|███████▋  | 5475/7175 [1:17:10<1:12:57,  2.57s/it]
 76%|███████▋  | 5476/7175 [1:17:11<1:03:37,  2.25s/it]
 76%|███████▋  | 5477/7175 [1:17:13<57:07,  2.02s/it]  
 76%|███████▋  | 5478/7175 [1:17:14<52:36,  1.86s/it]
 76%|███████▋  | 5479/7175 [1:17:16<49:23,  1.75s/it]
 76%|███████▋  | 5480/7175 [1:17:17<46:56,  1.66s/it]
                                                     
{'loss': '2.977', 'grad_norm': '0.2381', 'learning_rate': '0.0006', 'epoch': '19.09'}
+
 76%|███████▋  | 5480/7175 [1:17:17<46:56,  1.66s/it]
 76%|███████▋  | 5481/7175 [1:17:19<45:26,  1.61s/it]
 76%|███████▋  | 5482/7175 [1:17:20<44:19,  1.57s/it]
 76%|███████▋  | 5483/7175 [1:17:22<43:34,  1.54s/it]
 76%|███████▋  | 5484/7175 [1:17:23<42:57,  1.52s/it]
 76%|███████▋  | 5485/7175 [1:17:25<42:36,  1.51s/it]
 76%|███████▋  | 5486/7175 [1:17:26<42:19,  1.50s/it]
 76%|███████▋  | 5487/7175 [1:17:28<42:03,  1.49s/it]
 76%|███████▋  | 5488/7175 [1:17:29<41:50,  1.49s/it]
 77%|███████▋  | 5489/7175 [1:17:31<41:44,  1.49s/it]
 77%|███████▋  | 5490/7175 [1:17:32<41:43,  1.49s/it]
                                                     
{'loss': '2.97', 'grad_norm': '0.2242', 'learning_rate': '0.0006', 'epoch': '19.13'}
+
 77%|███████▋  | 5490/7175 [1:17:32<41:43,  1.49s/it]
 77%|███████▋  | 5491/7175 [1:17:34<41:42,  1.49s/it]
 77%|███████▋  | 5492/7175 [1:17:35<41:41,  1.49s/it]
 77%|███████▋  | 5493/7175 [1:17:37<41:34,  1.48s/it]
 77%|███████▋  | 5494/7175 [1:17:38<41:37,  1.49s/it]
 77%|███████▋  | 5495/7175 [1:17:40<41:32,  1.48s/it]
 77%|███████▋  | 5496/7175 [1:17:41<41:32,  1.48s/it]
 77%|███████▋  | 5497/7175 [1:17:42<41:32,  1.49s/it]
 77%|███████▋  | 5498/7175 [1:17:44<41:31,  1.49s/it]
 77%|███████▋  | 5499/7175 [1:17:45<41:27,  1.48s/it]
 77%|███████▋  | 5500/7175 [1:17:47<41:23,  1.48s/it]
                                                     
{'loss': '2.982', 'grad_norm': '0.2132', 'learning_rate': '0.0006', 'epoch': '19.16'}
+
 77%|███████▋  | 5500/7175 [1:17:47<41:23,  1.48s/it]
 77%|███████▋  | 5501/7175 [1:17:48<41:29,  1.49s/it]
 77%|███████▋  | 5502/7175 [1:17:50<41:23,  1.48s/it]
 77%|███████▋  | 5503/7175 [1:17:51<41:20,  1.48s/it]
 77%|███████▋  | 5504/7175 [1:17:53<41:17,  1.48s/it]
 77%|███████▋  | 5505/7175 [1:17:54<41:18,  1.48s/it]
 77%|███████▋  | 5506/7175 [1:17:56<41:17,  1.48s/it]
 77%|███████▋  | 5507/7175 [1:17:57<41:13,  1.48s/it]
 77%|███████▋  | 5508/7175 [1:17:59<43:53,  1.58s/it]
 77%|███████▋  | 5509/7175 [1:18:01<43:04,  1.55s/it]
 77%|███████▋  | 5510/7175 [1:18:02<42:29,  1.53s/it]
                                                     
{'loss': '2.992', 'grad_norm': '0.2571', 'learning_rate': '0.0006', 'epoch': '19.2'}
+
 77%|███████▋  | 5510/7175 [1:18:02<42:29,  1.53s/it]
 77%|███████▋  | 5511/7175 [1:18:03<40:21,  1.46s/it]
 77%|███████▋  | 5512/7175 [1:18:04<33:41,  1.22s/it]
 77%|███████▋  | 5513/7175 [1:18:05<34:06,  1.23s/it]
 77%|███████▋  | 5514/7175 [1:18:07<35:59,  1.30s/it]
 77%|███████▋  | 5515/7175 [1:18:08<37:18,  1.35s/it]
 77%|███████▋  | 5516/7175 [1:18:10<38:17,  1.39s/it]
 77%|███████▋  | 5517/7175 [1:18:11<38:52,  1.41s/it]
 77%|███████▋  | 5518/7175 [1:18:13<39:26,  1.43s/it]
 77%|███████▋  | 5519/7175 [1:18:14<39:48,  1.44s/it]
 77%|███████▋  | 5520/7175 [1:18:16<40:03,  1.45s/it]
                                                     
{'loss': '3.006', 'grad_norm': '0.2302', 'learning_rate': '0.0006', 'epoch': '19.23'}
+
 77%|███████▋  | 5520/7175 [1:18:16<40:03,  1.45s/it]
 77%|███████▋  | 5521/7175 [1:18:17<40:27,  1.47s/it]
 77%|███████▋  | 5522/7175 [1:18:19<40:31,  1.47s/it]
 77%|███████▋  | 5523/7175 [1:18:20<40:34,  1.47s/it]
 77%|███████▋  | 5524/7175 [1:18:22<40:41,  1.48s/it]
 77%|███████▋  | 5525/7175 [1:18:23<40:41,  1.48s/it]
 77%|███████▋  | 5526/7175 [1:18:24<40:38,  1.48s/it]
 77%|███████▋  | 5527/7175 [1:18:26<40:36,  1.48s/it]
 77%|███████▋  | 5528/7175 [1:18:27<40:39,  1.48s/it]
 77%|███████▋  | 5529/7175 [1:18:29<40:35,  1.48s/it]
 77%|███████▋  | 5530/7175 [1:18:30<40:43,  1.49s/it]
                                                     
{'loss': '3.011', 'grad_norm': '0.2145', 'learning_rate': '0.0006', 'epoch': '19.27'}
+
 77%|███████▋  | 5530/7175 [1:18:30<40:43,  1.49s/it]
 77%|███████▋  | 5531/7175 [1:18:32<40:46,  1.49s/it]
 77%|███████▋  | 5532/7175 [1:18:33<40:38,  1.48s/it]
 77%|███████▋  | 5533/7175 [1:18:35<40:33,  1.48s/it]
 77%|███████▋  | 5534/7175 [1:18:36<40:29,  1.48s/it]
 77%|███████▋  | 5535/7175 [1:18:38<40:28,  1.48s/it]
 77%|███████▋  | 5536/7175 [1:18:39<40:25,  1.48s/it]
 77%|███████▋  | 5537/7175 [1:18:41<40:22,  1.48s/it]
 77%|███████▋  | 5538/7175 [1:18:42<40:20,  1.48s/it]
 77%|███████▋  | 5539/7175 [1:18:44<40:24,  1.48s/it]
 77%|███████▋  | 5540/7175 [1:18:45<40:21,  1.48s/it]
                                                     
{'loss': '2.984', 'grad_norm': '0.2317', 'learning_rate': '0.0006', 'epoch': '19.3'}
+
 77%|███████▋  | 5540/7175 [1:18:45<40:21,  1.48s/it]
 77%|███████▋  | 5541/7175 [1:18:47<40:27,  1.49s/it]
 77%|███████▋  | 5542/7175 [1:18:48<40:22,  1.48s/it]
 77%|███████▋  | 5543/7175 [1:18:50<40:18,  1.48s/it]
 77%|███████▋  | 5544/7175 [1:18:51<40:11,  1.48s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:03, 10.45it/s]
+
 10%|█         | 4/39 [00:00<00:05,  6.51it/s]
+
 13%|█▎        | 5/39 [00:00<00:05,  6.11it/s]
+
 15%|█▌        | 6/39 [00:00<00:05,  5.82it/s]
+
 18%|█▊        | 7/39 [00:01<00:05,  5.67it/s]
+
 21%|██        | 8/39 [00:01<00:05,  5.52it/s]
+
 23%|██▎       | 9/39 [00:01<00:05,  5.45it/s]
+
 26%|██▌       | 10/39 [00:01<00:05,  5.37it/s]
+
 28%|██▊       | 11/39 [00:01<00:05,  5.32it/s]
+
 31%|███       | 12/39 [00:02<00:05,  5.29it/s]
+
 33%|███▎      | 13/39 [00:02<00:04,  5.36it/s]
+
 36%|███▌      | 14/39 [00:02<00:04,  5.33it/s]
+
 38%|███▊      | 15/39 [00:02<00:04,  5.32it/s]
+
 41%|████      | 16/39 [00:02<00:04,  5.28it/s]
+
 44%|████▎     | 17/39 [00:03<00:04,  5.26it/s]
+
 46%|████▌     | 18/39 [00:03<00:04,  5.24it/s]
+
 49%|████▊     | 19/39 [00:03<00:03,  5.32it/s]
+
 51%|█████▏    | 20/39 [00:03<00:03,  5.28it/s]
+
 54%|█████▍    | 21/39 [00:03<00:03,  5.28it/s]
+
 56%|█████▋    | 22/39 [00:03<00:03,  5.25it/s]
+
 59%|█████▉    | 23/39 [00:04<00:03,  5.23it/s]
+
 62%|██████▏   | 24/39 [00:04<00:02,  5.26it/s]
+
 64%|██████▍   | 25/39 [00:04<00:02,  5.29it/s]
+
 67%|██████▋   | 26/39 [00:04<00:02,  5.26it/s]
+
 69%|██████▉   | 27/39 [00:04<00:02,  5.24it/s]
+
 72%|███████▏  | 28/39 [00:05<00:02,  5.22it/s]
+
 74%|███████▍  | 29/39 [00:05<00:01,  5.25it/s]
+
 77%|███████▋  | 30/39 [00:05<00:01,  5.24it/s]
+
 79%|███████▉  | 31/39 [00:05<00:01,  5.28it/s]
+
 82%|████████▏ | 32/39 [00:05<00:01,  5.25it/s]
+
 85%|████████▍ | 33/39 [00:06<00:01,  5.23it/s]
+
 87%|████████▋ | 34/39 [00:06<00:00,  5.26it/s]
+
 90%|████████▉ | 35/39 [00:06<00:00,  5.30it/s]
+
 92%|█████████▏| 36/39 [00:06<00:00,  5.38it/s]
+
 95%|█████████▍| 37/39 [00:06<00:00,  5.39it/s]
+
 97%|█████████▋| 38/39 [00:07<00:00,  5.39it/s]
+
100%|██████████| 39/39 [00:07<00:00,  5.37it/s]
                                                     
+
                                               
{'eval_loss': '3.516', 'eval_runtime': '7.458', 'eval_samples_per_second': '327.3', 'eval_steps_per_second': '5.229', 'epoch': '19.32'}
+
 77%|███████▋  | 5544/7175 [1:18:59<40:11,  1.48s/it]
+
100%|██████████| 39/39 [00:07<00:00,  5.37it/s][eval] epoch 19.318 step 5544 loss 3.5164 (best 3.5036 @ ep 18.818, 2/20 since best)
+
+
                                               
 77%|███████▋  | 5545/7175 [1:19:00<1:40:54,  3.71s/it]
 77%|███████▋  | 5546/7175 [1:19:02<1:22:37,  3.04s/it]
 77%|███████▋  | 5547/7175 [1:19:03<1:09:51,  2.57s/it]
 77%|███████▋  | 5548/7175 [1:19:05<1:00:56,  2.25s/it]
 77%|███████▋  | 5549/7175 [1:19:06<54:39,  2.02s/it]  
 77%|███████▋  | 5550/7175 [1:19:07<50:16,  1.86s/it]
                                                     
{'loss': '3.016', 'grad_norm': '0.2003', 'learning_rate': '0.0006', 'epoch': '19.34'}
+
 77%|███████▋  | 5550/7175 [1:19:08<50:16,  1.86s/it]
 77%|███████▋  | 5551/7175 [1:19:09<47:18,  1.75s/it]
 77%|███████▋  | 5552/7175 [1:19:10<45:11,  1.67s/it]
 77%|███████▋  | 5553/7175 [1:19:12<43:36,  1.61s/it]
 77%|███████▋  | 5554/7175 [1:19:13<42:29,  1.57s/it]
 77%|███████▋  | 5555/7175 [1:19:15<41:47,  1.55s/it]
 77%|███████▋  | 5556/7175 [1:19:16<41:14,  1.53s/it]
 77%|███████▋  | 5557/7175 [1:19:18<40:51,  1.52s/it]
 77%|███████▋  | 5558/7175 [1:19:19<40:33,  1.51s/it]
 77%|███████▋  | 5559/7175 [1:19:21<40:21,  1.50s/it]
 77%|███████▋  | 5560/7175 [1:19:22<40:02,  1.49s/it]
                                                     
{'loss': '3', 'grad_norm': '0.2365', 'learning_rate': '0.0006', 'epoch': '19.37'}
+
 77%|███████▋  | 5560/7175 [1:19:22<40:02,  1.49s/it]
 78%|███████▊  | 5561/7175 [1:19:24<40:08,  1.49s/it]
 78%|███████▊  | 5562/7175 [1:19:25<40:02,  1.49s/it]
 78%|███████▊  | 5563/7175 [1:19:27<39:52,  1.48s/it]
 78%|███████▊  | 5564/7175 [1:19:28<39:49,  1.48s/it]
 78%|███████▊  | 5565/7175 [1:19:30<39:45,  1.48s/it]
 78%|███████▊  | 5566/7175 [1:19:31<39:44,  1.48s/it]
 78%|███████▊  | 5567/7175 [1:19:33<39:40,  1.48s/it]
 78%|███████▊  | 5568/7175 [1:19:34<39:40,  1.48s/it]
 78%|███████▊  | 5569/7175 [1:19:36<39:38,  1.48s/it]
 78%|███████▊  | 5570/7175 [1:19:37<39:35,  1.48s/it]
                                                     
{'loss': '3.004', 'grad_norm': '0.2311', 'learning_rate': '0.0006', 'epoch': '19.41'}
+
 78%|███████▊  | 5570/7175 [1:19:37<39:35,  1.48s/it]
 78%|███████▊  | 5571/7175 [1:19:39<39:37,  1.48s/it]
 78%|███████▊  | 5572/7175 [1:19:40<39:34,  1.48s/it]
 78%|███████▊  | 5573/7175 [1:19:42<39:33,  1.48s/it]
 78%|███████▊  | 5574/7175 [1:19:43<39:30,  1.48s/it]
 78%|███████▊  | 5575/7175 [1:19:45<39:29,  1.48s/it]
 78%|███████▊  | 5576/7175 [1:19:46<39:30,  1.48s/it]
 78%|███████▊  | 5577/7175 [1:19:48<39:26,  1.48s/it]
 78%|███████▊  | 5578/7175 [1:19:49<39:28,  1.48s/it]
 78%|███████▊  | 5579/7175 [1:19:50<39:26,  1.48s/it]
 78%|███████▊  | 5580/7175 [1:19:52<39:26,  1.48s/it]
                                                     
{'loss': '3.003', 'grad_norm': '0.3064', 'learning_rate': '0.0006', 'epoch': '19.44'}
+
 78%|███████▊  | 5580/7175 [1:19:52<39:26,  1.48s/it]
 78%|███████▊  | 5581/7175 [1:19:53<39:31,  1.49s/it]
 78%|███████▊  | 5582/7175 [1:19:55<39:32,  1.49s/it]
 78%|███████▊  | 5583/7175 [1:19:56<39:29,  1.49s/it]
 78%|███████▊  | 5584/7175 [1:19:58<39:21,  1.48s/it]
 78%|███████▊  | 5585/7175 [1:19:59<39:19,  1.48s/it]
 78%|███████▊  | 5586/7175 [1:20:01<39:16,  1.48s/it]
 78%|███████▊  | 5587/7175 [1:20:02<39:12,  1.48s/it]
 78%|███████▊  | 5588/7175 [1:20:04<39:12,  1.48s/it]
 78%|███████▊  | 5589/7175 [1:20:05<39:13,  1.48s/it]
 78%|███████▊  | 5590/7175 [1:20:07<39:10,  1.48s/it]
                                                     
{'loss': '3.005', 'grad_norm': '0.2351', 'learning_rate': '0.0006', 'epoch': '19.48'}
+
 78%|███████▊  | 5590/7175 [1:20:07<39:10,  1.48s/it]
 78%|███████▊  | 5591/7175 [1:20:08<39:16,  1.49s/it]
 78%|███████▊  | 5592/7175 [1:20:10<39:12,  1.49s/it]
 78%|███████▊  | 5593/7175 [1:20:11<39:07,  1.48s/it]
 78%|███████▊  | 5594/7175 [1:20:13<39:05,  1.48s/it]
 78%|███████▊  | 5595/7175 [1:20:14<39:03,  1.48s/it]
 78%|███████▊  | 5596/7175 [1:20:16<39:02,  1.48s/it]
 78%|███████▊  | 5597/7175 [1:20:17<39:05,  1.49s/it]
 78%|███████▊  | 5598/7175 [1:20:19<38:57,  1.48s/it]
 78%|███████▊  | 5599/7175 [1:20:20<38:57,  1.48s/it]
 78%|███████▊  | 5600/7175 [1:20:22<38:53,  1.48s/it]
                                                     
{'loss': '3.01', 'grad_norm': '0.2232', 'learning_rate': '0.0006', 'epoch': '19.51'}
+
 78%|███████▊  | 5600/7175 [1:20:22<38:53,  1.48s/it]
 78%|███████▊  | 5601/7175 [1:20:23<39:00,  1.49s/it]
 78%|███████▊  | 5602/7175 [1:20:25<38:56,  1.49s/it]
 78%|███████▊  | 5603/7175 [1:20:26<38:51,  1.48s/it]
 78%|███████▊  | 5604/7175 [1:20:28<38:47,  1.48s/it]
 78%|███████▊  | 5605/7175 [1:20:29<38:49,  1.48s/it]
 78%|███████▊  | 5606/7175 [1:20:31<38:45,  1.48s/it]
 78%|███████▊  | 5607/7175 [1:20:32<38:37,  1.48s/it]
 78%|███████▊  | 5608/7175 [1:20:34<38:38,  1.48s/it]
 78%|███████▊  | 5609/7175 [1:20:35<38:38,  1.48s/it]
 78%|███████▊  | 5610/7175 [1:20:36<38:39,  1.48s/it]
                                                     
{'loss': '3.024', 'grad_norm': '0.2176', 'learning_rate': '0.0006', 'epoch': '19.55'}
+
 78%|███████▊  | 5610/7175 [1:20:37<38:39,  1.48s/it]
 78%|███████▊  | 5611/7175 [1:20:38<38:44,  1.49s/it]
 78%|███████▊  | 5612/7175 [1:20:39<38:40,  1.48s/it]
 78%|███████▊  | 5613/7175 [1:20:41<38:36,  1.48s/it]
 78%|███████▊  | 5614/7175 [1:20:42<38:34,  1.48s/it]
 78%|███████▊  | 5615/7175 [1:20:44<38:28,  1.48s/it]
 78%|███████▊  | 5616/7175 [1:20:45<38:25,  1.48s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:03, 10.92it/s]
+
 10%|█         | 4/39 [00:00<00:05,  6.66it/s]
+
 13%|█▎        | 5/39 [00:00<00:05,  6.20it/s]
+
 15%|█▌        | 6/39 [00:00<00:05,  5.90it/s]
+
 18%|█▊        | 7/39 [00:01<00:05,  5.67it/s]
+
 21%|██        | 8/39 [00:01<00:05,  5.53it/s]
+
 23%|██▎       | 9/39 [00:01<00:05,  5.47it/s]
+
 26%|██▌       | 10/39 [00:01<00:05,  5.51it/s]
+
 28%|██▊       | 11/39 [00:01<00:05,  5.48it/s]
+
 31%|███       | 12/39 [00:02<00:04,  5.46it/s]
+
 33%|███▎      | 13/39 [00:02<00:04,  5.42it/s]
+
 36%|███▌      | 14/39 [00:02<00:04,  5.41it/s]
+
 38%|███▊      | 15/39 [00:02<00:04,  5.41it/s]
+
 41%|████      | 16/39 [00:02<00:04,  5.42it/s]
+
 44%|████▎     | 17/39 [00:02<00:04,  5.34it/s]
+
 46%|████▌     | 18/39 [00:03<00:04,  5.23it/s]
+
 49%|████▊     | 19/39 [00:03<00:03,  5.14it/s]
+
 51%|█████▏    | 20/39 [00:03<00:03,  5.15it/s]
+
 54%|█████▍    | 21/39 [00:03<00:03,  5.10it/s]
+
 56%|█████▋    | 22/39 [00:03<00:03,  5.12it/s]
+
 59%|█████▉    | 23/39 [00:04<00:03,  5.20it/s]
+
 62%|██████▏   | 24/39 [00:04<00:02,  5.23it/s]
+
 64%|██████▍   | 25/39 [00:04<00:02,  5.22it/s]
+
 67%|██████▋   | 26/39 [00:04<00:02,  5.21it/s]
+
 69%|██████▉   | 27/39 [00:04<00:02,  5.21it/s]
+
 72%|███████▏  | 28/39 [00:05<00:02,  5.25it/s]
+
 74%|███████▍  | 29/39 [00:05<00:01,  5.28it/s]
+
 77%|███████▋  | 30/39 [00:05<00:01,  5.25it/s]
+
 79%|███████▉  | 31/39 [00:05<00:01,  5.25it/s]
+
 82%|████████▏ | 32/39 [00:05<00:01,  5.23it/s]
+
 85%|████████▍ | 33/39 [00:06<00:01,  5.22it/s]
+
 87%|████████▋ | 34/39 [00:06<00:00,  5.25it/s]
+
 90%|████████▉ | 35/39 [00:06<00:00,  5.28it/s]
+
 92%|█████████▏| 36/39 [00:06<00:00,  5.25it/s]
+
 95%|█████████▍| 37/39 [00:06<00:00,  5.23it/s]
+
 97%|█████████▋| 38/39 [00:07<00:00,  5.27it/s]
+
100%|██████████| 39/39 [00:07<00:00,  5.24it/s]
                                                     
+
                                               
{'eval_loss': '3.509', 'eval_runtime': '7.462', 'eval_samples_per_second': '327.1', 'eval_steps_per_second': '5.226', 'epoch': '19.57'}
+
 78%|███████▊  | 5616/7175 [1:20:53<38:25,  1.48s/it]
+
100%|██████████| 39/39 [00:07<00:00,  5.24it/s][eval] epoch 19.57 step 5616 loss 3.5092 (best 3.5036 @ ep 18.818, 3/20 since best)
+
+
                                               
 78%|███████▊  | 5617/7175 [1:20:54<1:36:29,  3.72s/it]
 78%|███████▊  | 5618/7175 [1:20:56<1:19:02,  3.05s/it]
 78%|███████▊  | 5619/7175 [1:20:57<1:06:47,  2.58s/it]
 78%|███████▊  | 5620/7175 [1:20:59<58:14,  2.25s/it]  
                                                     
{'loss': '3.023', 'grad_norm': '0.2168', 'learning_rate': '0.0006', 'epoch': '19.58'}
+
 78%|███████▊  | 5620/7175 [1:20:59<58:14,  2.25s/it]
 78%|███████▊  | 5621/7175 [1:21:00<52:20,  2.02s/it]
 78%|███████▊  | 5622/7175 [1:21:02<48:02,  1.86s/it]
 78%|███████▊  | 5623/7175 [1:21:03<45:04,  1.74s/it]
 78%|███████▊  | 5624/7175 [1:21:05<42:59,  1.66s/it]
 78%|███████▊  | 5625/7175 [1:21:06<41:33,  1.61s/it]
 78%|███████▊  | 5626/7175 [1:21:08<40:32,  1.57s/it]
 78%|███████▊  | 5627/7175 [1:21:09<39:50,  1.54s/it]
 78%|███████▊  | 5628/7175 [1:21:11<39:20,  1.53s/it]
 78%|███████▊  | 5629/7175 [1:21:12<38:58,  1.51s/it]
 78%|███████▊  | 5630/7175 [1:21:14<38:42,  1.50s/it]
                                                     
{'loss': '3.021', 'grad_norm': '0.2113', 'learning_rate': '0.0006', 'epoch': '19.62'}
+
 78%|███████▊  | 5630/7175 [1:21:14<38:42,  1.50s/it]
 78%|███████▊  | 5631/7175 [1:21:15<38:38,  1.50s/it]
 78%|███████▊  | 5632/7175 [1:21:17<38:27,  1.50s/it]
 79%|███████▊  | 5633/7175 [1:21:18<38:21,  1.49s/it]
 79%|███████▊  | 5634/7175 [1:21:20<38:15,  1.49s/it]
 79%|███████▊  | 5635/7175 [1:21:21<38:11,  1.49s/it]
 79%|███████▊  | 5636/7175 [1:21:22<38:08,  1.49s/it]
 79%|███████▊  | 5637/7175 [1:21:24<38:00,  1.48s/it]
 79%|███████▊  | 5638/7175 [1:21:25<37:56,  1.48s/it]
 79%|███████▊  | 5639/7175 [1:21:27<37:57,  1.48s/it]
 79%|███████▊  | 5640/7175 [1:21:28<37:55,  1.48s/it]
                                                     
{'loss': '3.011', 'grad_norm': '0.2228', 'learning_rate': '0.0006', 'epoch': '19.65'}
+
 79%|███████▊  | 5640/7175 [1:21:28<37:55,  1.48s/it]
 79%|███████▊  | 5641/7175 [1:21:30<38:00,  1.49s/it]
 79%|███████▊  | 5642/7175 [1:21:31<37:47,  1.48s/it]
 79%|███████▊  | 5643/7175 [1:21:33<37:49,  1.48s/it]
 79%|███████▊  | 5644/7175 [1:21:34<37:46,  1.48s/it]
 79%|███████▊  | 5645/7175 [1:21:36<37:45,  1.48s/it]
 79%|███████▊  | 5646/7175 [1:21:37<37:44,  1.48s/it]
 79%|███████▊  | 5647/7175 [1:21:39<37:40,  1.48s/it]
 79%|███████▊  | 5648/7175 [1:21:40<37:40,  1.48s/it]
 79%|███████▊  | 5649/7175 [1:21:42<37:43,  1.48s/it]
 79%|███████▊  | 5650/7175 [1:21:43<37:42,  1.48s/it]
                                                     
{'loss': '3.016', 'grad_norm': '0.2071', 'learning_rate': '0.0006', 'epoch': '19.69'}
+
 79%|███████▊  | 5650/7175 [1:21:43<37:42,  1.48s/it]
 79%|███████▉  | 5651/7175 [1:21:45<37:45,  1.49s/it]
 79%|███████▉  | 5652/7175 [1:21:46<37:44,  1.49s/it]
 79%|███████▉  | 5653/7175 [1:21:48<37:41,  1.49s/it]
 79%|███████▉  | 5654/7175 [1:21:49<37:35,  1.48s/it]
 79%|███████▉  | 5655/7175 [1:21:51<37:25,  1.48s/it]
 79%|███████▉  | 5656/7175 [1:21:52<37:25,  1.48s/it]
 79%|███████▉  | 5657/7175 [1:21:54<37:26,  1.48s/it]
 79%|███████▉  | 5658/7175 [1:21:55<37:24,  1.48s/it]
 79%|███████▉  | 5659/7175 [1:21:57<37:23,  1.48s/it]
 79%|███████▉  | 5660/7175 [1:21:58<37:20,  1.48s/it]
                                                     
{'loss': '3.015', 'grad_norm': '0.2029', 'learning_rate': '0.0006', 'epoch': '19.72'}
+
 79%|███████▉  | 5660/7175 [1:21:58<37:20,  1.48s/it]
 79%|███████▉  | 5661/7175 [1:22:00<37:28,  1.49s/it]
 79%|███████▉  | 5662/7175 [1:22:00<32:26,  1.29s/it]
 79%|███████▉  | 5663/7175 [1:22:01<29:15,  1.16s/it]
 79%|███████▉  | 5664/7175 [1:22:03<31:34,  1.25s/it]
 79%|███████▉  | 5665/7175 [1:22:04<33:07,  1.32s/it]
 79%|███████▉  | 5666/7175 [1:22:06<34:11,  1.36s/it]
 79%|███████▉  | 5667/7175 [1:22:07<34:55,  1.39s/it]
 79%|███████▉  | 5668/7175 [1:22:09<35:37,  1.42s/it]
 79%|███████▉  | 5669/7175 [1:22:10<36:05,  1.44s/it]
 79%|███████▉  | 5670/7175 [1:22:11<36:18,  1.45s/it]
                                                     
{'loss': '3.02', 'grad_norm': '0.2218', 'learning_rate': '0.0006', 'epoch': '19.76'}
+
 79%|███████▉  | 5670/7175 [1:22:12<36:18,  1.45s/it]
 79%|███████▉  | 5671/7175 [1:22:13<36:39,  1.46s/it]
 79%|███████▉  | 5672/7175 [1:22:14<36:45,  1.47s/it]
 79%|███████▉  | 5673/7175 [1:22:16<36:50,  1.47s/it]
 79%|███████▉  | 5674/7175 [1:22:17<36:49,  1.47s/it]
 79%|███████▉  | 5675/7175 [1:22:19<36:53,  1.48s/it]
 79%|███████▉  | 5676/7175 [1:22:20<36:53,  1.48s/it]
 79%|███████▉  | 5677/7175 [1:22:22<36:54,  1.48s/it]
 79%|███████▉  | 5678/7175 [1:22:23<36:51,  1.48s/it]
 79%|███████▉  | 5679/7175 [1:22:25<36:50,  1.48s/it]
 79%|███████▉  | 5680/7175 [1:22:26<36:50,  1.48s/it]
                                                     
{'loss': '3.032', 'grad_norm': '0.2025', 'learning_rate': '0.0006', 'epoch': '19.79'}
+
 79%|███████▉  | 5680/7175 [1:22:26<36:50,  1.48s/it]
 79%|███████▉  | 5681/7175 [1:22:28<36:57,  1.48s/it]
 79%|███████▉  | 5682/7175 [1:22:29<36:43,  1.48s/it]
 79%|███████▉  | 5683/7175 [1:22:31<36:43,  1.48s/it]
 79%|███████▉  | 5684/7175 [1:22:32<36:43,  1.48s/it]
 79%|███████▉  | 5685/7175 [1:22:34<36:47,  1.48s/it]
 79%|███████▉  | 5686/7175 [1:22:35<36:44,  1.48s/it]
 79%|███████▉  | 5687/7175 [1:22:37<36:40,  1.48s/it]
 79%|███████▉  | 5688/7175 [1:22:38<36:40,  1.48s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:03, 10.48it/s]
+
 10%|█         | 4/39 [00:00<00:05,  6.42it/s]
+
 13%|█▎        | 5/39 [00:00<00:05,  6.02it/s]
+
 15%|█▌        | 6/39 [00:00<00:05,  5.66it/s]
+
 18%|█▊        | 7/39 [00:01<00:05,  5.48it/s]
+
 21%|██        | 8/39 [00:01<00:05,  5.42it/s]
+
 23%|██▎       | 9/39 [00:01<00:05,  5.39it/s]
+
 26%|██▌       | 10/39 [00:01<00:05,  5.34it/s]
+
 28%|██▊       | 11/39 [00:01<00:05,  5.34it/s]
+
 31%|███       | 12/39 [00:02<00:05,  5.30it/s]
+
 33%|███▎      | 13/39 [00:02<00:04,  5.27it/s]
+
 36%|███▌      | 14/39 [00:02<00:04,  5.24it/s]
+
 38%|███▊      | 15/39 [00:02<00:04,  5.23it/s]
+
 41%|████      | 16/39 [00:02<00:04,  5.23it/s]
+
 44%|████▎     | 17/39 [00:03<00:04,  5.29it/s]
+
 46%|████▌     | 18/39 [00:03<00:03,  5.26it/s]
+
 49%|████▊     | 19/39 [00:03<00:03,  5.26it/s]
+
 51%|█████▏    | 20/39 [00:03<00:03,  5.24it/s]
+
 54%|█████▍    | 21/39 [00:03<00:03,  5.22it/s]
+
 56%|█████▋    | 22/39 [00:04<00:03,  5.22it/s]
+
 59%|█████▉    | 23/39 [00:04<00:03,  5.29it/s]
+
 62%|██████▏   | 24/39 [00:04<00:02,  5.26it/s]
+
 64%|██████▍   | 25/39 [00:04<00:02,  5.23it/s]
+
 67%|██████▋   | 26/39 [00:04<00:02,  5.26it/s]
+
 69%|██████▉   | 27/39 [00:04<00:02,  5.24it/s]
+
 72%|███████▏  | 28/39 [00:05<00:02,  5.23it/s]
+
 74%|███████▍  | 29/39 [00:05<00:01,  5.26it/s]
+
 77%|███████▋  | 30/39 [00:05<00:01,  5.30it/s]
+
 79%|███████▉  | 31/39 [00:05<00:01,  5.33it/s]
+
 82%|████████▏ | 32/39 [00:05<00:01,  5.30it/s]
+
 85%|████████▍ | 33/39 [00:06<00:01,  5.27it/s]
+
 87%|████████▋ | 34/39 [00:06<00:00,  5.27it/s]
+
 90%|████████▉ | 35/39 [00:06<00:00,  5.30it/s]
+
 92%|█████████▏| 36/39 [00:06<00:00,  5.27it/s]
+
 95%|█████████▍| 37/39 [00:06<00:00,  5.28it/s]
+
 97%|█████████▋| 38/39 [00:07<00:00,  5.26it/s]
+
100%|██████████| 39/39 [00:07<00:00,  5.23it/s]
                                                     
+
                                               
{'eval_loss': '3.503', 'eval_runtime': '7.506', 'eval_samples_per_second': '325.2', 'eval_steps_per_second': '5.196', 'epoch': '19.82'}
+
 79%|███████▉  | 5688/7175 [1:22:46<36:40,  1.48s/it]
+
100%|██████████| 39/39 [00:07<00:00,  5.23it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-5688
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-5688/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-5688/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.88it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.88it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-5688/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-5688/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_300M_C0_baseline/checkpoint-5688/tokenizer_config.json
+[eval] epoch 19.821 step 5688 loss 3.5035 (best 3.5035 @ ep 19.821, 0/20 since best)
+
+
                                               
 79%|███████▉  | 5689/7175 [1:22:49<1:47:57,  4.36s/it]
 79%|███████▉  | 5690/7175 [1:22:51<1:26:31,  3.50s/it]
                                                       
{'loss': '3.029', 'grad_norm': '0.2425', 'learning_rate': '0.0006', 'epoch': '19.83'}
+
 79%|███████▉  | 5690/7175 [1:22:51<1:26:31,  3.50s/it]
 79%|███████▉  | 5691/7175 [1:22:52<1:11:36,  2.90s/it]
 79%|███████▉  | 5692/7175 [1:22:54<1:01:03,  2.47s/it]
 79%|███████▉  | 5693/7175 [1:22:55<53:40,  2.17s/it]  
 79%|███████▉  | 5694/7175 [1:22:57<48:31,  1.97s/it]
 79%|███████▉  | 5695/7175 [1:22:58<44:52,  1.82s/it]
 79%|███████▉  | 5696/7175 [1:23:00<42:20,  1.72s/it]
 79%|███████▉  | 5697/7175 [1:23:01<40:31,  1.65s/it]
 79%|███████▉  | 5698/7175 [1:23:03<39:17,  1.60s/it]
 79%|███████▉  | 5699/7175 [1:23:04<38:26,  1.56s/it]
 79%|███████▉  | 5700/7175 [1:23:06<37:49,  1.54s/it]
                                                     
{'loss': '3.033', 'grad_norm': '0.201', 'learning_rate': '0.0006', 'epoch': '19.86'}
+
 79%|███████▉  | 5700/7175 [1:23:06<37:49,  1.54s/it]
 79%|███████▉  | 5701/7175 [1:23:07<37:41,  1.53s/it]
 79%|███████▉  | 5702/7175 [1:23:09<37:21,  1.52s/it]
 79%|███████▉  | 5703/7175 [1:23:10<37:01,  1.51s/it]
 79%|███████▉  | 5704/7175 [1:23:11<36:44,  1.50s/it]
 80%|███████▉  | 5705/7175 [1:23:13<36:35,  1.49s/it]
 80%|███████▉  | 5706/7175 [1:23:14<36:27,  1.49s/it]
 80%|███████▉  | 5707/7175 [1:23:16<36:23,  1.49s/it]
 80%|███████▉  | 5708/7175 [1:23:17<36:22,  1.49s/it]
 80%|███████▉  | 5709/7175 [1:23:19<36:19,  1.49s/it]
 80%|███████▉  | 5710/7175 [1:23:20<36:14,  1.48s/it]
                                                     
{'loss': '3.033', 'grad_norm': '0.2168', 'learning_rate': '0.0006', 'epoch': '19.9'}
+
 80%|███████▉  | 5710/7175 [1:23:20<36:14,  1.48s/it]
 80%|███████▉  | 5711/7175 [1:23:22<36:22,  1.49s/it]
 80%|███████▉  | 5712/7175 [1:23:23<36:16,  1.49s/it]
 80%|███████▉  | 5713/7175 [1:23:25<36:12,  1.49s/it]
 80%|███████▉  | 5714/7175 [1:23:26<36:09,  1.48s/it]
 80%|███████▉  | 5715/7175 [1:23:28<36:06,  1.48s/it]
 80%|███████▉  | 5716/7175 [1:23:29<36:02,  1.48s/it]
 80%|███████▉  | 5717/7175 [1:23:31<35:59,  1.48s/it]
 80%|███████▉  | 5718/7175 [1:23:32<35:56,  1.48s/it]
 80%|███████▉  | 5719/7175 [1:23:34<35:56,  1.48s/it]
 80%|███████▉  | 5720/7175 [1:23:35<35:54,  1.48s/it]
                                                     
{'loss': '3.044', 'grad_norm': '0.2075', 'learning_rate': '0.0006', 'epoch': '19.93'}
+
 80%|███████▉  | 5720/7175 [1:23:35<35:54,  1.48s/it]
 80%|███████▉  | 5721/7175 [1:23:37<35:57,  1.48s/it]
 80%|███████▉  | 5722/7175 [1:23:38<35:56,  1.48s/it]
 80%|███████▉  | 5723/7175 [1:23:40<35:55,  1.48s/it]
 80%|███████▉  | 5724/7175 [1:23:41<35:54,  1.48s/it]
 80%|███████▉  | 5725/7175 [1:23:43<35:48,  1.48s/it]
 80%|███████▉  | 5726/7175 [1:23:44<35:46,  1.48s/it]
 80%|███████▉  | 5727/7175 [1:23:46<35:45,  1.48s/it]
 80%|███████▉  | 5728/7175 [1:23:47<35:45,  1.48s/it]
 80%|███████▉  | 5729/7175 [1:23:49<35:43,  1.48s/it]
 80%|███████▉  | 5730/7175 [1:23:50<35:44,  1.48s/it]
                                                     
{'loss': '3.034', 'grad_norm': '0.2144', 'learning_rate': '0.0006', 'epoch': '19.97'}
+
 80%|███████▉  | 5730/7175 [1:23:50<35:44,  1.48s/it]
 80%|███████▉  | 5731/7175 [1:23:52<35:47,  1.49s/it]
 80%|███████▉  | 5732/7175 [1:23:53<35:42,  1.48s/it]
 80%|███████▉  | 5733/7175 [1:23:55<35:39,  1.48s/it]
 80%|███████▉  | 5734/7175 [1:23:56<35:31,  1.48s/it]
 80%|███████▉  | 5735/7175 [1:23:57<35:31,  1.48s/it]
 80%|███████▉  | 5736/7175 [1:23:59<35:23,  1.48s/it]
 80%|███████▉  | 5737/7175 [1:24:00<35:25,  1.48s/it]
 80%|███████▉  | 5738/7175 [1:24:02<35:27,  1.48s/it]
 80%|███████▉  | 5739/7175 [1:24:03<35:24,  1.48s/it]
 80%|████████  | 5740/7175 [1:24:04<26:27,  1.11s/it]
                                                     
{'loss': '3.02', 'grad_norm': '0.4237', 'learning_rate': '0.0006', 'epoch': '20'}
+
 80%|████████  | 5740/7175 [1:24:04<26:27,  1.11s/it]
 80%|████████  | 5741/7175 [1:24:05<29:51,  1.25s/it]
 80%|████████  | 5742/7175 [1:24:07<31:34,  1.32s/it]
 80%|████████  | 5743/7175 [1:24:08<32:42,  1.37s/it]
 80%|████████  | 5744/7175 [1:24:10<33:25,  1.40s/it]
 80%|████████  | 5745/7175 [1:24:11<33:57,  1.42s/it]
 80%|████████  | 5746/7175 [1:24:13<34:20,  1.44s/it]
 80%|████████  | 5747/7175 [1:24:14<34:36,  1.45s/it]
 80%|████████  | 5748/7175 [1:24:16<34:47,  1.46s/it]
 80%|████████  | 5749/7175 [1:24:17<36:53,  1.55s/it]
 80%|████████  | 5750/7175 [1:24:19<40:24,  1.70s/it]
                                                     
{'loss': '2.952', 'grad_norm': '0.2509', 'learning_rate': '0.0006', 'epoch': '20.03'}
+
 80%|████████  | 5750/7175 [1:24:19<40:24,  1.70s/it]
 80%|████████  | 5751/7175 [1:24:21<42:58,  1.81s/it]
 80%|████████  | 5752/7175 [1:24:23<44:38,  1.88s/it]
 80%|████████  | 5753/7175 [1:24:26<45:44,  1.93s/it]
 80%|████████  | 5754/7175 [1:24:28<46:26,  1.96s/it]
 80%|████████  | 5755/7175 [1:24:30<46:56,  1.98s/it]
 80%|████████  | 5756/7175 [1:24:32<47:09,  1.99s/it]
 80%|████████  | 5757/7175 [1:24:34<47:23,  2.01s/it]
 80%|████████  | 5758/7175 [1:24:36<47:32,  2.01s/it]
 80%|████████  | 5759/7175 [1:24:38<47:31,  2.01s/it]
 80%|████████  | 5760/7175 [1:24:40<47:27,  2.01s/it]
                                                     
{'loss': '2.97', 'grad_norm': '0.2533', 'learning_rate': '0.0006', 'epoch': '20.07'}
+
 80%|████████  | 5760/7175 [1:24:40<47:27,  2.01s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:04,  7.93it/s]
+
  8%|▊         | 3/39 [00:00<00:06,  5.46it/s]
+
 10%|█         | 4/39 [00:00<00:07,  4.69it/s]
+
 13%|█▎        | 5/39 [00:01<00:07,  4.39it/s]
+
 15%|█▌        | 6/39 [00:01<00:07,  4.35it/s]
+
 18%|█▊        | 7/39 [00:01<00:06,  4.60it/s]
+
 21%|██        | 8/39 [00:01<00:06,  4.77it/s]
+
 23%|██▎       | 9/39 [00:01<00:06,  4.90it/s]
+
 26%|██▌       | 10/39 [00:02<00:05,  4.95it/s]
+
 28%|██▊       | 11/39 [00:02<00:05,  5.03it/s]
+
 31%|███       | 12/39 [00:02<00:05,  5.08it/s]
+
 33%|███▎      | 13/39 [00:02<00:05,  5.20it/s]
+
 36%|███▌      | 14/39 [00:02<00:04,  5.19it/s]
+
 38%|███▊      | 15/39 [00:02<00:04,  5.19it/s]
+
 41%|████      | 16/39 [00:03<00:04,  5.19it/s]
+
 44%|████▎     | 17/39 [00:03<00:04,  5.19it/s]
+
 46%|████▌     | 18/39 [00:03<00:04,  5.24it/s]
+
 49%|████▊     | 19/39 [00:03<00:03,  5.27it/s]
+
 51%|█████▏    | 20/39 [00:03<00:03,  5.25it/s]
+
 54%|█████▍    | 21/39 [00:04<00:03,  5.23it/s]
+
 56%|█████▋    | 22/39 [00:04<00:03,  5.22it/s]
+
 59%|█████▉    | 23/39 [00:04<00:03,  5.26it/s]
+
 62%|██████▏   | 24/39 [00:04<00:02,  5.24it/s]
+
 64%|██████▍   | 25/39 [00:04<00:02,  5.26it/s]
+
 67%|██████▋   | 26/39 [00:05<00:02,  5.29it/s]
+
 69%|██████▉   | 27/39 [00:05<00:02,  5.32it/s]
+
 72%|███████▏  | 28/39 [00:05<00:02,  5.32it/s]
+
 74%|███████▍  | 29/39 [00:05<00:01,  5.35it/s]
+
 77%|███████▋  | 30/39 [00:05<00:01,  5.41it/s]
+
 79%|███████▉  | 31/39 [00:06<00:01,  5.41it/s]
+
 82%|████████▏ | 32/39 [00:06<00:01,  5.41it/s]
+
 85%|████████▍ | 33/39 [00:06<00:01,  5.38it/s]
+
 87%|████████▋ | 34/39 [00:06<00:00,  5.39it/s]
+
 90%|████████▉ | 35/39 [00:06<00:00,  5.39it/s]
+
 92%|█████████▏| 36/39 [00:06<00:00,  5.45it/s]
+
 95%|█████████▍| 37/39 [00:07<00:00,  5.44it/s]
+
 97%|█████████▋| 38/39 [00:07<00:00,  5.43it/s]
+
100%|██████████| 39/39 [00:07<00:00,  5.43it/s]
                                                     
+
                                               
{'eval_loss': '3.53', 'eval_runtime': '7.729', 'eval_samples_per_second': '315.8', 'eval_steps_per_second': '5.046', 'epoch': '20.07'}
+
 80%|████████  | 5760/7175 [1:24:47<47:27,  2.01s/it]
+
100%|██████████| 39/39 [00:07<00:00,  5.43it/s][eval] epoch 20.07 step 5760 loss 3.5296 (best 3.5035 @ ep 19.821, 1/20 since best)
+
+
                                               
 80%|████████  | 5761/7175 [1:24:49<1:38:25,  4.18s/it]
 80%|████████  | 5762/7175 [1:24:50<1:19:17,  3.37s/it]
 80%|████████  | 5763/7175 [1:24:52<1:05:56,  2.80s/it]
 80%|████████  | 5764/7175 [1:24:53<56:33,  2.40s/it]  
 80%|████████  | 5765/7175 [1:24:55<49:58,  2.13s/it]
 80%|████████  | 5766/7175 [1:24:56<45:25,  1.93s/it]
 80%|████████  | 5767/7175 [1:24:58<42:09,  1.80s/it]
 80%|████████  | 5768/7175 [1:24:59<39:53,  1.70s/it]
 80%|████████  | 5769/7175 [1:25:01<38:10,  1.63s/it]
 80%|████████  | 5770/7175 [1:25:02<37:04,  1.58s/it]
                                                     
{'loss': '2.974', 'grad_norm': '0.271', 'learning_rate': '0.0006', 'epoch': '20.1'}
+
 80%|████████  | 5770/7175 [1:25:02<37:04,  1.58s/it]
 80%|████████  | 5771/7175 [1:25:04<36:25,  1.56s/it]
 80%|████████  | 5772/7175 [1:25:05<35:53,  1.53s/it]
 80%|████████  | 5773/7175 [1:25:07<35:26,  1.52s/it]
 80%|████████  | 5774/7175 [1:25:08<35:13,  1.51s/it]
 80%|████████  | 5775/7175 [1:25:10<34:59,  1.50s/it]
 81%|████████  | 5776/7175 [1:25:11<34:49,  1.49s/it]
 81%|████████  | 5777/7175 [1:25:13<34:39,  1.49s/it]
 81%|████████  | 5778/7175 [1:25:14<34:33,  1.48s/it]
 81%|████████  | 5779/7175 [1:25:16<34:29,  1.48s/it]
 81%|████████  | 5780/7175 [1:25:17<34:26,  1.48s/it]
                                                     
{'loss': '2.966', 'grad_norm': '0.2175', 'learning_rate': '0.0006', 'epoch': '20.14'}
+
 81%|████████  | 5780/7175 [1:25:17<34:26,  1.48s/it]
 81%|████████  | 5781/7175 [1:25:19<34:28,  1.48s/it]
 81%|████████  | 5782/7175 [1:25:20<34:24,  1.48s/it]
 81%|████████  | 5783/7175 [1:25:21<34:22,  1.48s/it]
 81%|████████  | 5784/7175 [1:25:23<34:19,  1.48s/it]
 81%|████████  | 5785/7175 [1:25:24<34:20,  1.48s/it]
 81%|████████  | 5786/7175 [1:25:26<34:18,  1.48s/it]
 81%|████████  | 5787/7175 [1:25:27<34:16,  1.48s/it]
 81%|████████  | 5788/7175 [1:25:29<34:13,  1.48s/it]
 81%|████████  | 5789/7175 [1:25:30<34:10,  1.48s/it]
 81%|████████  | 5790/7175 [1:25:32<34:11,  1.48s/it]
                                                     
{'loss': '2.966', 'grad_norm': '0.2354', 'learning_rate': '0.0006', 'epoch': '20.17'}
+
 81%|████████  | 5790/7175 [1:25:32<34:11,  1.48s/it]
 81%|████████  | 5791/7175 [1:25:33<34:16,  1.49s/it]
 81%|████████  | 5792/7175 [1:25:35<34:11,  1.48s/it]
 81%|████████  | 5793/7175 [1:25:36<34:06,  1.48s/it]
 81%|████████  | 5794/7175 [1:25:38<34:03,  1.48s/it]
 81%|████████  | 5795/7175 [1:25:39<34:02,  1.48s/it]
 81%|████████  | 5796/7175 [1:25:41<34:01,  1.48s/it]
 81%|████████  | 5797/7175 [1:25:42<33:59,  1.48s/it]
 81%|████████  | 5798/7175 [1:25:44<33:57,  1.48s/it]
 81%|████████  | 5799/7175 [1:25:45<33:56,  1.48s/it]
 81%|████████  | 5800/7175 [1:25:47<33:45,  1.47s/it]
                                                     
{'loss': '2.974', 'grad_norm': '0.2169', 'learning_rate': '0.0006', 'epoch': '20.21'}
+
 81%|████████  | 5800/7175 [1:25:47<33:45,  1.47s/it]
 81%|████████  | 5801/7175 [1:25:48<33:51,  1.48s/it]
 81%|████████  | 5802/7175 [1:25:50<33:48,  1.48s/it]
 81%|████████  | 5803/7175 [1:25:51<33:51,  1.48s/it]
 81%|████████  | 5804/7175 [1:25:53<33:51,  1.48s/it]
 81%|████████  | 5805/7175 [1:25:54<33:48,  1.48s/it]
 81%|████████  | 5806/7175 [1:25:56<33:48,  1.48s/it]
 81%|████████  | 5807/7175 [1:25:57<33:49,  1.48s/it]
 81%|████████  | 5808/7175 [1:25:59<33:46,  1.48s/it]
 81%|████████  | 5809/7175 [1:26:00<33:44,  1.48s/it]
 81%|████████  | 5810/7175 [1:26:01<33:43,  1.48s/it]
                                                     
{'loss': '2.988', 'grad_norm': '0.2471', 'learning_rate': '0.0006', 'epoch': '20.24'}
+
 81%|████████  | 5810/7175 [1:26:01<33:43,  1.48s/it]
 81%|████████  | 5811/7175 [1:26:03<33:43,  1.48s/it]
 81%|████████  | 5812/7175 [1:26:04<29:42,  1.31s/it]
 81%|████████  | 5813/7175 [1:26:05<28:02,  1.24s/it]
 81%|████████  | 5814/7175 [1:26:06<28:15,  1.25s/it]
 81%|████████  | 5815/7175 [1:26:08<29:47,  1.31s/it]
 81%|████████  | 5816/7175 [1:26:09<31:02,  1.37s/it]
 81%|████████  | 5817/7175 [1:26:11<31:44,  1.40s/it]
 81%|████████  | 5818/7175 [1:26:12<32:19,  1.43s/it]
 81%|████████  | 5819/7175 [1:26:14<32:41,  1.45s/it]
 81%|████████  | 5820/7175 [1:26:15<32:55,  1.46s/it]
                                                     
{'loss': '2.994', 'grad_norm': '0.2115', 'learning_rate': '0.0006', 'epoch': '20.28'}
+
 81%|████████  | 5820/7175 [1:26:15<32:55,  1.46s/it]
 81%|████████  | 5821/7175 [1:26:17<33:10,  1.47s/it]
 81%|████████  | 5822/7175 [1:26:18<33:12,  1.47s/it]
 81%|████████  | 5823/7175 [1:26:20<33:14,  1.48s/it]
 81%|████████  | 5824/7175 [1:26:21<33:14,  1.48s/it]
 81%|████████  | 5825/7175 [1:26:23<33:15,  1.48s/it]
 81%|████████  | 5826/7175 [1:26:24<33:16,  1.48s/it]
 81%|████████  | 5827/7175 [1:26:25<33:15,  1.48s/it]
 81%|████████  | 5828/7175 [1:26:27<33:14,  1.48s/it]
 81%|████████  | 5829/7175 [1:26:28<33:13,  1.48s/it]
 81%|████████▏ | 5830/7175 [1:26:30<33:11,  1.48s/it]
                                                     
{'loss': '2.992', 'grad_norm': '0.229', 'learning_rate': '0.0006', 'epoch': '20.31'}
+
 81%|████████▏ | 5830/7175 [1:26:30<33:11,  1.48s/it]
 81%|████████▏ | 5831/7175 [1:26:31<33:11,  1.48s/it]
 81%|████████▏ | 5832/7175 [1:26:33<33:10,  1.48s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:03, 10.34it/s]
+
 10%|█         | 4/39 [00:00<00:05,  6.38it/s]
+
 13%|█▎        | 5/39 [00:00<00:05,  5.89it/s]
+
 15%|█▌        | 6/39 [00:00<00:05,  5.69it/s]
+
 18%|█▊        | 7/39 [00:01<00:05,  5.49it/s]
+
 21%|██        | 8/39 [00:01<00:05,  5.36it/s]
+
 23%|██▎       | 9/39 [00:01<00:05,  5.24it/s]
+
 26%|██▌       | 10/39 [00:01<00:05,  5.25it/s]
+
 28%|██▊       | 11/39 [00:01<00:05,  5.23it/s]
+
 31%|███       | 12/39 [00:02<00:05,  5.27it/s]
+
 33%|███▎      | 13/39 [00:02<00:04,  5.25it/s]
+
 36%|███▌      | 14/39 [00:02<00:04,  5.26it/s]
+
 38%|███▊      | 15/39 [00:02<00:04,  5.22it/s]
+
 41%|████      | 16/39 [00:02<00:04,  5.25it/s]
+
 44%|████▎     | 17/39 [00:03<00:04,  5.23it/s]
+
 46%|████▌     | 18/39 [00:03<00:03,  5.27it/s]
+
 49%|████▊     | 19/39 [00:03<00:03,  5.25it/s]
+
 51%|█████▏    | 20/39 [00:03<00:03,  5.24it/s]
+
 54%|█████▍    | 21/39 [00:03<00:03,  5.23it/s]
+
 56%|█████▋    | 22/39 [00:04<00:03,  5.22it/s]
+
 59%|█████▉    | 23/39 [00:04<00:03,  5.25it/s]
+
 62%|██████▏   | 24/39 [00:04<00:02,  5.28it/s]
+
 64%|██████▍   | 25/39 [00:04<00:02,  5.26it/s]
+
 67%|██████▋   | 26/39 [00:04<00:02,  5.29it/s]
+
 69%|██████▉   | 27/39 [00:04<00:02,  5.32it/s]
+
 72%|███████▏  | 28/39 [00:05<00:02,  5.35it/s]
+
 74%|███████▍  | 29/39 [00:05<00:01,  5.41it/s]
+
 77%|███████▋  | 30/39 [00:05<00:01,  5.41it/s]
+
 79%|███████▉  | 31/39 [00:05<00:01,  5.41it/s]
+
 82%|████████▏ | 32/39 [00:05<00:01,  5.38it/s]
+
 85%|████████▍ | 33/39 [00:06<00:01,  5.25it/s]
+
 87%|████████▋ | 34/39 [00:06<00:00,  5.19it/s]
+
 90%|████████▉ | 35/39 [00:06<00:00,  5.21it/s]
+
 92%|█████████▏| 36/39 [00:06<00:00,  5.25it/s]
+
 95%|█████████▍| 37/39 [00:06<00:00,  5.23it/s]
+
 97%|█████████▋| 38/39 [00:07<00:00,  5.24it/s]
+
100%|██████████| 39/39 [00:07<00:00,  5.22it/s]
                                                     
+
                                               
{'eval_loss': '3.519', 'eval_runtime': '7.515', 'eval_samples_per_second': '324.8', 'eval_steps_per_second': '5.19', 'epoch': '20.32'}
+
 81%|████████▏ | 5832/7175 [1:26:40<33:10,  1.48s/it]
+
100%|██████████| 39/39 [00:07<00:00,  5.22it/s][eval] epoch 20.322 step 5832 loss 3.5187 (best 3.5035 @ ep 19.821, 2/20 since best)
+
+
                                               
 81%|████████▏ | 5833/7175 [1:26:42<1:23:29,  3.73s/it]
 81%|████████▏ | 5834/7175 [1:26:43<1:08:20,  3.06s/it]
 81%|████████▏ | 5835/7175 [1:26:45<59:58,  2.69s/it]  
 81%|████████▏ | 5836/7175 [1:26:47<51:52,  2.32s/it]
 81%|████████▏ | 5837/7175 [1:26:48<46:09,  2.07s/it]
 81%|████████▏ | 5838/7175 [1:26:50<42:15,  1.90s/it]
 81%|████████▏ | 5839/7175 [1:26:51<41:38,  1.87s/it]
 81%|████████▏ | 5840/7175 [1:26:53<39:02,  1.75s/it]
                                                     
{'loss': '2.992', 'grad_norm': '0.2243', 'learning_rate': '0.0006', 'epoch': '20.35'}
+
 81%|████████▏ | 5840/7175 [1:26:53<39:02,  1.75s/it]
 81%|████████▏ | 5841/7175 [1:26:54<37:16,  1.68s/it]
 81%|████████▏ | 5842/7175 [1:26:56<35:57,  1.62s/it]
 81%|████████▏ | 5843/7175 [1:26:57<35:03,  1.58s/it]
 81%|████████▏ | 5844/7175 [1:26:59<36:27,  1.64s/it]
 81%|████████▏ | 5845/7175 [1:27:01<35:22,  1.60s/it]
 81%|████████▏ | 5846/7175 [1:27:02<34:35,  1.56s/it]
 81%|████████▏ | 5847/7175 [1:27:04<34:05,  1.54s/it]
 82%|████████▏ | 5848/7175 [1:27:05<33:43,  1.52s/it]
 82%|████████▏ | 5849/7175 [1:27:07<33:32,  1.52s/it]
 82%|████████▏ | 5850/7175 [1:27:08<33:16,  1.51s/it]
                                                     
{'loss': '2.983', 'grad_norm': '0.2085', 'learning_rate': '0.0006', 'epoch': '20.38'}
+
 82%|████████▏ | 5850/7175 [1:27:08<33:16,  1.51s/it]
 82%|████████▏ | 5851/7175 [1:27:10<33:14,  1.51s/it]
 82%|████████▏ | 5852/7175 [1:27:11<35:35,  1.61s/it]
 82%|████████▏ | 5853/7175 [1:27:13<34:42,  1.58s/it]
 82%|████████▏ | 5854/7175 [1:27:14<34:03,  1.55s/it]
 82%|████████▏ | 5855/7175 [1:27:16<33:43,  1.53s/it]
 82%|████████▏ | 5856/7175 [1:27:17<33:32,  1.53s/it]
 82%|████████▏ | 5857/7175 [1:27:19<33:21,  1.52s/it]
 82%|████████▏ | 5858/7175 [1:27:21<36:33,  1.67s/it]
 82%|████████▏ | 5859/7175 [1:27:23<39:07,  1.78s/it]
 82%|████████▏ | 5860/7175 [1:27:25<40:57,  1.87s/it]
                                                     
{'loss': '2.991', 'grad_norm': '0.2189', 'learning_rate': '0.0006', 'epoch': '20.42'}
+
 82%|████████▏ | 5860/7175 [1:27:25<40:57,  1.87s/it]
 82%|████████▏ | 5861/7175 [1:27:27<42:24,  1.94s/it]
 82%|████████▏ | 5862/7175 [1:27:29<43:06,  1.97s/it]
 82%|████████▏ | 5863/7175 [1:27:31<43:28,  1.99s/it]
 82%|████████▏ | 5864/7175 [1:27:33<43:45,  2.00s/it]
 82%|████████▏ | 5865/7175 [1:27:35<43:57,  2.01s/it]
 82%|████████▏ | 5866/7175 [1:27:37<44:05,  2.02s/it]
 82%|████████▏ | 5867/7175 [1:27:39<44:08,  2.03s/it]
 82%|████████▏ | 5868/7175 [1:27:41<44:05,  2.02s/it]
 82%|████████▏ | 5869/7175 [1:27:43<44:04,  2.03s/it]
 82%|████████▏ | 5870/7175 [1:27:45<43:49,  2.01s/it]
                                                     
{'loss': '2.998', 'grad_norm': '0.2179', 'learning_rate': '0.0006', 'epoch': '20.45'}
+
 82%|████████▏ | 5870/7175 [1:27:46<43:49,  2.01s/it]
 82%|████████▏ | 5871/7175 [1:27:48<44:15,  2.04s/it]
 82%|████████▏ | 5872/7175 [1:27:50<44:30,  2.05s/it]
 82%|████████▏ | 5873/7175 [1:27:52<44:35,  2.05s/it]
 82%|████████▏ | 5874/7175 [1:27:54<44:35,  2.06s/it]
 82%|████████▏ | 5875/7175 [1:27:56<44:25,  2.05s/it]
 82%|████████▏ | 5876/7175 [1:27:58<44:14,  2.04s/it]
 82%|████████▏ | 5877/7175 [1:28:00<44:08,  2.04s/it]
 82%|████████▏ | 5878/7175 [1:28:02<44:07,  2.04s/it]
 82%|████████▏ | 5879/7175 [1:28:04<44:03,  2.04s/it]
 82%|████████▏ | 5880/7175 [1:28:06<44:00,  2.04s/it]
                                                     
{'loss': '3.003', 'grad_norm': '0.2148', 'learning_rate': '0.0006', 'epoch': '20.49'}
+
 82%|████████▏ | 5880/7175 [1:28:06<44:00,  2.04s/it]
 82%|████████▏ | 5881/7175 [1:28:08<44:06,  2.04s/it]
 82%|████████▏ | 5882/7175 [1:28:10<43:41,  2.03s/it]
 82%|████████▏ | 5883/7175 [1:28:12<43:50,  2.04s/it]
 82%|████████▏ | 5884/7175 [1:28:14<44:04,  2.05s/it]
 82%|████████▏ | 5885/7175 [1:28:16<44:09,  2.05s/it]
 82%|████████▏ | 5886/7175 [1:28:19<45:45,  2.13s/it]
 82%|████████▏ | 5887/7175 [1:28:21<45:07,  2.10s/it]
 82%|████████▏ | 5888/7175 [1:28:23<44:37,  2.08s/it]
 82%|████████▏ | 5889/7175 [1:28:25<44:13,  2.06s/it]
 82%|████████▏ | 5890/7175 [1:28:27<43:58,  2.05s/it]
                                                     
{'loss': '2.996', 'grad_norm': '0.2131', 'learning_rate': '0.0006', 'epoch': '20.52'}
+
 82%|████████▏ | 5890/7175 [1:28:27<43:58,  2.05s/it]
 82%|████████▏ | 5891/7175 [1:28:29<43:59,  2.06s/it]
 82%|████████▏ | 5892/7175 [1:28:31<43:51,  2.05s/it]
 82%|████████▏ | 5893/7175 [1:28:33<43:44,  2.05s/it]
 82%|████████▏ | 5894/7175 [1:28:35<43:08,  2.02s/it]
 82%|████████▏ | 5895/7175 [1:28:37<43:26,  2.04s/it]
 82%|████████▏ | 5896/7175 [1:28:39<43:32,  2.04s/it]
 82%|████████▏ | 5897/7175 [1:28:41<43:39,  2.05s/it]
 82%|████████▏ | 5898/7175 [1:28:43<43:45,  2.06s/it]
 82%|████████▏ | 5899/7175 [1:28:45<43:40,  2.05s/it]
 82%|████████▏ | 5900/7175 [1:28:47<45:14,  2.13s/it]
                                                     
{'loss': '3.003', 'grad_norm': '0.2362', 'learning_rate': '0.0006', 'epoch': '20.56'}
+
 82%|████████▏ | 5900/7175 [1:28:47<45:14,  2.13s/it]
 82%|████████▏ | 5901/7175 [1:28:49<44:44,  2.11s/it]
 82%|████████▏ | 5902/7175 [1:28:51<44:12,  2.08s/it]
 82%|████████▏ | 5903/7175 [1:28:53<43:52,  2.07s/it]
 82%|████████▏ | 5904/7175 [1:28:56<43:34,  2.06s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:04,  8.19it/s]
+
  8%|▊         | 3/39 [00:00<00:06,  5.48it/s]
+
 10%|█         | 4/39 [00:00<00:07,  4.71it/s]
+
 13%|█▎        | 5/39 [00:01<00:07,  4.43it/s]
+
 15%|█▌        | 6/39 [00:01<00:07,  4.20it/s]
+
 18%|█▊        | 7/39 [00:01<00:07,  4.08it/s]
+
 21%|██        | 8/39 [00:01<00:07,  3.97it/s]
+
 23%|██▎       | 9/39 [00:02<00:07,  3.92it/s]
+
 26%|██▌       | 10/39 [00:02<00:07,  3.89it/s]
+
 28%|██▊       | 11/39 [00:02<00:07,  3.87it/s]
+
 31%|███       | 12/39 [00:02<00:07,  3.83it/s]
+
 33%|███▎      | 13/39 [00:03<00:06,  3.82it/s]
+
 36%|███▌      | 14/39 [00:03<00:06,  3.79it/s]
+
 38%|███▊      | 15/39 [00:03<00:06,  3.81it/s]
+
 41%|████      | 16/39 [00:03<00:05,  3.83it/s]
+
 44%|████▎     | 17/39 [00:04<00:05,  3.82it/s]
+
 46%|████▌     | 18/39 [00:04<00:05,  3.82it/s]
+
 49%|████▊     | 19/39 [00:04<00:05,  3.80it/s]
+
 51%|█████▏    | 20/39 [00:04<00:05,  3.79it/s]
+
 54%|█████▍    | 21/39 [00:05<00:04,  3.76it/s]
+
 56%|█████▋    | 22/39 [00:05<00:04,  3.76it/s]
+
 59%|█████▉    | 23/39 [00:05<00:04,  3.74it/s]
+
 62%|██████▏   | 24/39 [00:06<00:03,  3.79it/s]
+
 64%|██████▍   | 25/39 [00:06<00:03,  3.78it/s]
+
 67%|██████▋   | 26/39 [00:06<00:03,  3.77it/s]
+
 69%|██████▉   | 27/39 [00:06<00:03,  3.76it/s]
+
 72%|███████▏  | 28/39 [00:07<00:02,  3.76it/s]
+
 74%|███████▍  | 29/39 [00:07<00:02,  3.75it/s]
+
 77%|███████▋  | 30/39 [00:07<00:02,  3.75it/s]
+
 79%|███████▉  | 31/39 [00:07<00:02,  3.77it/s]
+
 82%|████████▏ | 32/39 [00:08<00:01,  3.79it/s]
+
 85%|████████▍ | 33/39 [00:08<00:01,  3.78it/s]
+
 87%|████████▋ | 34/39 [00:08<00:01,  3.72it/s]
+
 90%|████████▉ | 35/39 [00:08<00:01,  3.70it/s]
+
 92%|█████████▏| 36/39 [00:09<00:00,  3.69it/s]
+
 95%|█████████▍| 37/39 [00:09<00:00,  3.70it/s]
+
 97%|█████████▋| 38/39 [00:09<00:00,  3.65it/s]
+
100%|██████████| 39/39 [00:10<00:00,  3.68it/s]
                                                     
+
                                               
{'eval_loss': '3.514', 'eval_runtime': '10.4', 'eval_samples_per_second': '234.7', 'eval_steps_per_second': '3.749', 'epoch': '20.57'}
+
 82%|████████▏ | 5904/7175 [1:29:06<43:34,  2.06s/it]
+
100%|██████████| 39/39 [00:10<00:00,  3.68it/s][eval] epoch 20.573 step 5904 loss 3.5141 (best 3.5035 @ ep 19.821, 3/20 since best)
+
+
                                               
 82%|████████▏ | 5905/7175 [1:29:08<1:49:33,  5.18s/it]
 82%|████████▏ | 5906/7175 [1:29:10<1:29:37,  4.24s/it]
 82%|████████▏ | 5907/7175 [1:29:12<1:15:34,  3.58s/it]
 82%|████████▏ | 5908/7175 [1:29:14<1:05:45,  3.11s/it]
 82%|████████▏ | 5909/7175 [1:29:16<58:50,  2.79s/it]  
 82%|████████▏ | 5910/7175 [1:29:18<54:00,  2.56s/it]
                                                     
{'loss': '3.005', 'grad_norm': '0.2274', 'learning_rate': '0.0006', 'epoch': '20.59'}
+
 82%|████████▏ | 5910/7175 [1:29:18<54:00,  2.56s/it]
 82%|████████▏ | 5911/7175 [1:29:20<50:44,  2.41s/it]
 82%|████████▏ | 5912/7175 [1:29:22<48:21,  2.30s/it]
 82%|████████▏ | 5913/7175 [1:29:24<46:19,  2.20s/it]
 82%|████████▏ | 5914/7175 [1:29:26<44:43,  2.13s/it]
 82%|████████▏ | 5915/7175 [1:29:28<44:14,  2.11s/it]
 82%|████████▏ | 5916/7175 [1:29:30<44:00,  2.10s/it]
 82%|████████▏ | 5917/7175 [1:29:32<43:48,  2.09s/it]
 82%|████████▏ | 5918/7175 [1:29:34<43:33,  2.08s/it]
 82%|████████▏ | 5919/7175 [1:29:36<43:21,  2.07s/it]
 83%|████████▎ | 5920/7175 [1:29:39<43:07,  2.06s/it]
                                                     
{'loss': '3.021', 'grad_norm': '0.2035', 'learning_rate': '0.0006', 'epoch': '20.63'}
+
 83%|████████▎ | 5920/7175 [1:29:39<43:07,  2.06s/it]
 83%|████████▎ | 5921/7175 [1:29:41<43:11,  2.07s/it]
 83%|████████▎ | 5922/7175 [1:29:43<42:55,  2.06s/it]
 83%|████████▎ | 5923/7175 [1:29:45<42:45,  2.05s/it]
 83%|████████▎ | 5924/7175 [1:29:47<42:35,  2.04s/it]
 83%|████████▎ | 5925/7175 [1:29:49<42:34,  2.04s/it]
 83%|████████▎ | 5926/7175 [1:29:51<42:14,  2.03s/it]
 83%|████████▎ | 5927/7175 [1:29:53<42:28,  2.04s/it]
 83%|████████▎ | 5928/7175 [1:29:55<42:37,  2.05s/it]
 83%|████████▎ | 5929/7175 [1:29:57<42:38,  2.05s/it]
 83%|████████▎ | 5930/7175 [1:29:59<42:33,  2.05s/it]
                                                     
{'loss': '3.02', 'grad_norm': '0.2055', 'learning_rate': '0.0006', 'epoch': '20.66'}
+
 83%|████████▎ | 5930/7175 [1:29:59<42:33,  2.05s/it]
 83%|████████▎ | 5931/7175 [1:30:01<42:38,  2.06s/it]
 83%|████████▎ | 5932/7175 [1:30:03<42:29,  2.05s/it]
 83%|████████▎ | 5933/7175 [1:30:05<42:19,  2.04s/it]
 83%|████████▎ | 5934/7175 [1:30:07<42:14,  2.04s/it]
 83%|████████▎ | 5935/7175 [1:30:09<42:09,  2.04s/it]
 83%|████████▎ | 5936/7175 [1:30:11<42:05,  2.04s/it]
 83%|████████▎ | 5937/7175 [1:30:13<42:00,  2.04s/it]
 83%|████████▎ | 5938/7175 [1:30:15<42:02,  2.04s/it]
 83%|████████▎ | 5939/7175 [1:30:17<42:07,  2.04s/it]
 83%|████████▎ | 5940/7175 [1:30:19<42:15,  2.05s/it]
                                                     
{'loss': '3.015', 'grad_norm': '0.2467', 'learning_rate': '0.0006', 'epoch': '20.7'}
+
 83%|████████▎ | 5940/7175 [1:30:19<42:15,  2.05s/it]
 83%|████████▎ | 5941/7175 [1:30:22<42:28,  2.07s/it]
 83%|████████▎ | 5942/7175 [1:30:24<42:28,  2.07s/it]
 83%|████████▎ | 5943/7175 [1:30:26<42:16,  2.06s/it]
 83%|████████▎ | 5944/7175 [1:30:28<42:10,  2.06s/it]
 83%|████████▎ | 5945/7175 [1:30:30<42:01,  2.05s/it]
 83%|████████▎ | 5946/7175 [1:30:32<41:52,  2.04s/it]
 83%|████████▎ | 5947/7175 [1:30:34<41:49,  2.04s/it]
 83%|████████▎ | 5948/7175 [1:30:36<41:47,  2.04s/it]
 83%|████████▎ | 5949/7175 [1:30:38<41:36,  2.04s/it]
 83%|████████▎ | 5950/7175 [1:30:40<41:36,  2.04s/it]
                                                     
{'loss': '3.01', 'grad_norm': '0.2404', 'learning_rate': '0.0006', 'epoch': '20.73'}
+
 83%|████████▎ | 5950/7175 [1:30:40<41:36,  2.04s/it]
 83%|████████▎ | 5951/7175 [1:30:42<42:01,  2.06s/it]
 83%|████████▎ | 5952/7175 [1:30:44<42:01,  2.06s/it]
 83%|████████▎ | 5953/7175 [1:30:46<42:03,  2.07s/it]
 83%|████████▎ | 5954/7175 [1:30:48<42:00,  2.06s/it]
 83%|████████▎ | 5955/7175 [1:30:50<41:46,  2.05s/it]
 83%|████████▎ | 5956/7175 [1:30:52<41:33,  2.05s/it]
 83%|████████▎ | 5957/7175 [1:30:54<41:26,  2.04s/it]
 83%|████████▎ | 5958/7175 [1:30:56<41:27,  2.04s/it]
 83%|████████▎ | 5959/7175 [1:30:58<41:20,  2.04s/it]
 83%|████████▎ | 5960/7175 [1:31:00<41:14,  2.04s/it]
                                                     
{'loss': '3.012', 'grad_norm': '0.2234', 'learning_rate': '0.0006', 'epoch': '20.77'}
+
 83%|████████▎ | 5960/7175 [1:31:00<41:14,  2.04s/it]
 83%|████████▎ | 5961/7175 [1:31:02<41:23,  2.05s/it]
 83%|████████▎ | 5962/7175 [1:31:05<41:22,  2.05s/it]
 83%|████████▎ | 5963/7175 [1:31:07<41:29,  2.05s/it]
 83%|████████▎ | 5964/7175 [1:31:09<41:34,  2.06s/it]
 83%|████████▎ | 5965/7175 [1:31:11<41:38,  2.06s/it]
 83%|████████▎ | 5966/7175 [1:31:13<41:37,  2.07s/it]
 83%|████████▎ | 5967/7175 [1:31:15<40:55,  2.03s/it]
 83%|████████▎ | 5968/7175 [1:31:16<38:09,  1.90s/it]
 83%|████████▎ | 5969/7175 [1:31:18<38:47,  1.93s/it]
 83%|████████▎ | 5970/7175 [1:31:20<39:19,  1.96s/it]
                                                     
{'loss': '3.025', 'grad_norm': '0.2184', 'learning_rate': '0.0006', 'epoch': '20.8'}
+
 83%|████████▎ | 5970/7175 [1:31:20<39:19,  1.96s/it]
 83%|████████▎ | 5971/7175 [1:31:22<39:50,  1.99s/it]
 83%|████████▎ | 5972/7175 [1:31:24<39:46,  1.98s/it]
 83%|████████▎ | 5973/7175 [1:31:26<39:43,  1.98s/it]
 83%|████████▎ | 5974/7175 [1:31:28<39:36,  1.98s/it]
 83%|████████▎ | 5975/7175 [1:31:30<40:04,  2.00s/it]
 83%|████████▎ | 5976/7175 [1:31:32<40:30,  2.03s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:04,  7.87it/s]
+
  8%|▊         | 3/39 [00:00<00:06,  5.33it/s]
+
 10%|█         | 4/39 [00:00<00:07,  4.59it/s]
+
 13%|█▎        | 5/39 [00:01<00:07,  4.28it/s]
+
 15%|█▌        | 6/39 [00:01<00:08,  4.09it/s]
+
 18%|█▊        | 7/39 [00:01<00:07,  4.00it/s]
+
 21%|██        | 8/39 [00:01<00:07,  3.89it/s]
+
 23%|██▎       | 9/39 [00:02<00:07,  3.87it/s]
+
 26%|██▌       | 10/39 [00:02<00:07,  3.83it/s]
+
 28%|██▊       | 11/39 [00:02<00:07,  3.77it/s]
+
 31%|███       | 12/39 [00:02<00:07,  3.74it/s]
+
 33%|███▎      | 13/39 [00:03<00:06,  3.74it/s]
+
 36%|███▌      | 14/39 [00:03<00:06,  3.73it/s]
+
 38%|███▊      | 15/39 [00:03<00:06,  3.78it/s]
+
 41%|████      | 16/39 [00:03<00:06,  3.75it/s]
+
 44%|████▎     | 17/39 [00:04<00:05,  3.72it/s]
+
 46%|████▌     | 18/39 [00:04<00:05,  3.75it/s]
+
 49%|████▊     | 19/39 [00:04<00:05,  3.78it/s]
+
 51%|█████▏    | 20/39 [00:05<00:05,  3.77it/s]
+
 54%|█████▍    | 21/39 [00:05<00:04,  3.76it/s]
+
 56%|█████▋    | 22/39 [00:05<00:04,  3.78it/s]
+
 59%|█████▉    | 23/39 [00:05<00:04,  3.76it/s]
+
 62%|██████▏   | 24/39 [00:06<00:03,  3.75it/s]
+
 64%|██████▍   | 25/39 [00:06<00:03,  3.74it/s]
+
 67%|██████▋   | 26/39 [00:06<00:03,  3.74it/s]
+
 69%|██████▉   | 27/39 [00:06<00:03,  3.72it/s]
+
 72%|███████▏  | 28/39 [00:07<00:02,  3.70it/s]
+
 74%|███████▍  | 29/39 [00:07<00:02,  3.79it/s]
+
 77%|███████▋  | 30/39 [00:07<00:02,  3.79it/s]
+
 79%|███████▉  | 31/39 [00:07<00:02,  3.77it/s]
+
 82%|████████▏ | 32/39 [00:08<00:01,  3.82it/s]
+
 85%|████████▍ | 33/39 [00:08<00:01,  3.79it/s]
+
 87%|████████▋ | 34/39 [00:08<00:01,  3.79it/s]
+
 90%|████████▉ | 35/39 [00:09<00:01,  3.80it/s]
+
 92%|█████████▏| 36/39 [00:09<00:00,  3.80it/s]
+
 95%|█████████▍| 37/39 [00:09<00:00,  3.78it/s]
+
 97%|█████████▋| 38/39 [00:09<00:00,  3.81it/s]
+
100%|██████████| 39/39 [00:10<00:00,  3.79it/s]
                                                     
+
                                               
{'eval_loss': '3.507', 'eval_runtime': '10.41', 'eval_samples_per_second': '234.4', 'eval_steps_per_second': '3.746', 'epoch': '20.82'}
+
 83%|████████▎ | 5976/7175 [1:31:43<40:30,  2.03s/it]
+
100%|██████████| 39/39 [00:10<00:00,  3.79it/s][eval] epoch 20.825 step 5976 loss 3.5071 (best 3.5035 @ ep 19.821, 4/20 since best)
+
+
                                               
 83%|████████▎ | 5977/7175 [1:31:45<1:42:47,  5.15s/it]
 83%|████████▎ | 5978/7175 [1:31:47<1:24:01,  4.21s/it]
 83%|████████▎ | 5979/7175 [1:31:49<1:10:52,  3.56s/it]
 83%|████████▎ | 5980/7175 [1:31:51<1:01:45,  3.10s/it]
                                                       
{'loss': '3.007', 'grad_norm': '0.2263', 'learning_rate': '0.0006', 'epoch': '20.84'}
+
 83%|████████▎ | 5980/7175 [1:31:51<1:01:45,  3.10s/it]
 83%|████████▎ | 5981/7175 [1:31:53<55:16,  2.78s/it]  
 83%|████████▎ | 5982/7175 [1:31:55<51:12,  2.58s/it]
 83%|████████▎ | 5983/7175 [1:31:57<48:19,  2.43s/it]
 83%|████████▎ | 5984/7175 [1:31:59<46:10,  2.33s/it]
 83%|████████▎ | 5985/7175 [1:32:01<44:40,  2.25s/it]
 83%|████████▎ | 5986/7175 [1:32:03<43:23,  2.19s/it]
 83%|████████▎ | 5987/7175 [1:32:05<42:27,  2.14s/it]
 83%|████████▎ | 5988/7175 [1:32:08<41:53,  2.12s/it]
 83%|████████▎ | 5989/7175 [1:32:10<41:29,  2.10s/it]
 83%|████████▎ | 5990/7175 [1:32:12<41:13,  2.09s/it]
                                                     
{'loss': '3.029', 'grad_norm': '0.2185', 'learning_rate': '0.0006', 'epoch': '20.87'}
+
 83%|████████▎ | 5990/7175 [1:32:12<41:13,  2.09s/it]
 83%|████████▎ | 5991/7175 [1:32:14<41:06,  2.08s/it]
 84%|████████▎ | 5992/7175 [1:32:16<40:50,  2.07s/it]
 84%|████████▎ | 5993/7175 [1:32:18<40:41,  2.07s/it]
 84%|████████▎ | 5994/7175 [1:32:20<40:42,  2.07s/it]
 84%|████████▎ | 5995/7175 [1:32:22<40:42,  2.07s/it]
 84%|████████▎ | 5996/7175 [1:32:24<40:38,  2.07s/it]
 84%|████████▎ | 5997/7175 [1:32:26<40:35,  2.07s/it]
 84%|████████▎ | 5998/7175 [1:32:28<40:22,  2.06s/it]
 84%|████████▎ | 5999/7175 [1:32:30<40:11,  2.05s/it]
 84%|████████▎ | 6000/7175 [1:32:32<40:06,  2.05s/it]
                                                     
{'loss': '3.013', 'grad_norm': '0.2298', 'learning_rate': '0.0006', 'epoch': '20.91'}
+
 84%|████████▎ | 6000/7175 [1:32:32<40:06,  2.05s/it]
 84%|████████▎ | 6001/7175 [1:32:34<40:13,  2.06s/it]
 84%|████████▎ | 6002/7175 [1:32:36<40:02,  2.05s/it]
 84%|████████▎ | 6003/7175 [1:32:38<39:55,  2.04s/it]
 84%|████████▎ | 6004/7175 [1:32:40<39:48,  2.04s/it]
 84%|████████▎ | 6005/7175 [1:32:42<39:52,  2.04s/it]
 84%|████████▎ | 6006/7175 [1:32:45<40:01,  2.05s/it]
 84%|████████▎ | 6007/7175 [1:32:47<40:05,  2.06s/it]
 84%|████████▎ | 6008/7175 [1:32:49<40:09,  2.06s/it]
 84%|████████▎ | 6009/7175 [1:32:51<40:04,  2.06s/it]
 84%|████████▍ | 6010/7175 [1:32:53<39:58,  2.06s/it]
                                                     
{'loss': '3.018', 'grad_norm': '0.2082', 'learning_rate': '0.0006', 'epoch': '20.94'}
+
 84%|████████▍ | 6010/7175 [1:32:53<39:58,  2.06s/it]
 84%|████████▍ | 6011/7175 [1:32:55<39:59,  2.06s/it]
 84%|████████▍ | 6012/7175 [1:32:57<39:47,  2.05s/it]
 84%|████████▍ | 6013/7175 [1:32:59<39:43,  2.05s/it]
 84%|████████▍ | 6014/7175 [1:33:01<39:33,  2.04s/it]
 84%|████████▍ | 6015/7175 [1:33:03<39:26,  2.04s/it]
 84%|████████▍ | 6016/7175 [1:33:05<39:20,  2.04s/it]
 84%|████████▍ | 6017/7175 [1:33:07<39:21,  2.04s/it]
 84%|████████▍ | 6018/7175 [1:33:09<39:37,  2.05s/it]
 84%|████████▍ | 6019/7175 [1:33:11<39:39,  2.06s/it]
 84%|████████▍ | 6020/7175 [1:33:13<39:37,  2.06s/it]
                                                     
{'loss': '3.005', 'grad_norm': '0.2036', 'learning_rate': '0.0006', 'epoch': '20.98'}
+
 84%|████████▍ | 6020/7175 [1:33:13<39:37,  2.06s/it]
 84%|████████▍ | 6021/7175 [1:33:15<39:49,  2.07s/it]
 84%|████████▍ | 6022/7175 [1:33:17<39:37,  2.06s/it]
 84%|████████▍ | 6023/7175 [1:33:19<39:30,  2.06s/it]
 84%|████████▍ | 6024/7175 [1:33:21<39:23,  2.05s/it]
 84%|████████▍ | 6025/7175 [1:33:24<39:16,  2.05s/it]
 84%|████████▍ | 6026/7175 [1:33:26<39:09,  2.04s/it]
 84%|████████▍ | 6027/7175 [1:33:26<29:14,  1.53s/it]
 84%|████████▍ | 6028/7175 [1:33:28<32:34,  1.70s/it]
 84%|████████▍ | 6029/7175 [1:33:30<34:24,  1.80s/it]
 84%|████████▍ | 6030/7175 [1:33:32<35:49,  1.88s/it]
                                                     
{'loss': '2.987', 'grad_norm': '0.3117', 'learning_rate': '0.0006', 'epoch': '21.01'}
+
 84%|████████▍ | 6030/7175 [1:33:32<35:49,  1.88s/it]
 84%|████████▍ | 6031/7175 [1:33:34<37:04,  1.94s/it]
 84%|████████▍ | 6032/7175 [1:33:36<37:41,  1.98s/it]
 84%|████████▍ | 6033/7175 [1:33:38<38:12,  2.01s/it]
 84%|████████▍ | 6034/7175 [1:33:40<38:29,  2.02s/it]
 84%|████████▍ | 6035/7175 [1:33:42<38:37,  2.03s/it]
 84%|████████▍ | 6036/7175 [1:33:44<38:38,  2.04s/it]
 84%|████████▍ | 6037/7175 [1:33:47<38:36,  2.04s/it]
 84%|████████▍ | 6038/7175 [1:33:49<38:40,  2.04s/it]
 84%|████████▍ | 6039/7175 [1:33:51<38:37,  2.04s/it]
 84%|████████▍ | 6040/7175 [1:33:53<38:33,  2.04s/it]
                                                     
{'loss': '2.953', 'grad_norm': '0.2275', 'learning_rate': '0.0006', 'epoch': '21.05'}
+
 84%|████████▍ | 6040/7175 [1:33:53<38:33,  2.04s/it]
 84%|████████▍ | 6041/7175 [1:33:55<38:39,  2.05s/it]
 84%|████████▍ | 6042/7175 [1:33:57<38:42,  2.05s/it]
 84%|████████▍ | 6043/7175 [1:33:59<38:46,  2.06s/it]
 84%|████████▍ | 6044/7175 [1:34:01<38:50,  2.06s/it]
 84%|████████▍ | 6045/7175 [1:34:03<38:45,  2.06s/it]
 84%|████████▍ | 6046/7175 [1:34:05<38:42,  2.06s/it]
 84%|████████▍ | 6047/7175 [1:34:07<38:35,  2.05s/it]
 84%|████████▍ | 6048/7175 [1:34:09<38:26,  2.05s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:04,  7.73it/s]
+
  8%|▊         | 3/39 [00:00<00:06,  5.32it/s]
+
 10%|█         | 4/39 [00:00<00:07,  4.60it/s]
+
 13%|█▎        | 5/39 [00:01<00:07,  4.25it/s]
+
 15%|█▌        | 6/39 [00:01<00:08,  4.07it/s]
+
 18%|█▊        | 7/39 [00:01<00:08,  3.98it/s]
+
 21%|██        | 8/39 [00:01<00:07,  3.92it/s]
+
 23%|██▎       | 9/39 [00:02<00:07,  3.87it/s]
+
 26%|██▌       | 10/39 [00:02<00:07,  3.82it/s]
+
 28%|██▊       | 11/39 [00:02<00:07,  3.80it/s]
+
 31%|███       | 12/39 [00:02<00:07,  3.83it/s]
+
 33%|███▎      | 13/39 [00:03<00:06,  3.79it/s]
+
 36%|███▌      | 14/39 [00:03<00:06,  3.82it/s]
+
 38%|███▊      | 15/39 [00:03<00:06,  3.77it/s]
+
 41%|████      | 16/39 [00:03<00:06,  3.77it/s]
+
 44%|████▎     | 17/39 [00:04<00:05,  3.77it/s]
+
 46%|████▌     | 18/39 [00:04<00:05,  3.78it/s]
+
 49%|████▊     | 19/39 [00:04<00:05,  3.79it/s]
+
 51%|█████▏    | 20/39 [00:05<00:05,  3.79it/s]
+
 54%|█████▍    | 21/39 [00:05<00:04,  3.79it/s]
+
 56%|█████▋    | 22/39 [00:05<00:04,  3.77it/s]
+
 59%|█████▉    | 23/39 [00:05<00:04,  3.78it/s]
+
 62%|██████▏   | 24/39 [00:06<00:04,  3.73it/s]
+
 64%|██████▍   | 25/39 [00:06<00:03,  3.76it/s]
+
 67%|██████▋   | 26/39 [00:06<00:03,  3.75it/s]
+
 69%|██████▉   | 27/39 [00:06<00:03,  3.76it/s]
+
 72%|███████▏  | 28/39 [00:07<00:02,  3.76it/s]
+
 74%|███████▍  | 29/39 [00:07<00:02,  3.76it/s]
+
 77%|███████▋  | 30/39 [00:07<00:02,  3.76it/s]
+
 79%|███████▉  | 31/39 [00:07<00:02,  3.77it/s]
+
 82%|████████▏ | 32/39 [00:08<00:01,  3.78it/s]
+
 85%|████████▍ | 33/39 [00:08<00:01,  3.75it/s]
+
 87%|████████▋ | 34/39 [00:08<00:01,  3.74it/s]
+
 90%|████████▉ | 35/39 [00:09<00:01,  3.73it/s]
+
 92%|█████████▏| 36/39 [00:09<00:00,  3.76it/s]
+
 95%|█████████▍| 37/39 [00:09<00:00,  3.76it/s]
+
 97%|█████████▋| 38/39 [00:09<00:00,  3.76it/s]
+
100%|██████████| 39/39 [00:10<00:00,  3.78it/s]
                                                     
+
                                               
{'eval_loss': '3.526', 'eval_runtime': '10.43', 'eval_samples_per_second': '234.1', 'eval_steps_per_second': '3.74', 'epoch': '21.07'}
+
 84%|████████▍ | 6048/7175 [1:34:20<38:26,  2.05s/it]
+
100%|██████████| 39/39 [00:10<00:00,  3.78it/s][eval] epoch 21.073 step 6048 loss 3.5262 (best 3.5035 @ ep 19.821, 5/20 since best)
+
+
                                               
 84%|████████▍ | 6049/7175 [1:34:22<1:37:09,  5.18s/it]
 84%|████████▍ | 6050/7175 [1:34:24<1:19:36,  4.25s/it]
                                                       
{'loss': '2.946', 'grad_norm': '0.2249', 'learning_rate': '0.0006', 'epoch': '21.08'}
+
 84%|████████▍ | 6050/7175 [1:34:24<1:19:36,  4.25s/it]
 84%|████████▍ | 6051/7175 [1:34:26<1:07:30,  3.60s/it]
 84%|████████▍ | 6052/7175 [1:34:28<58:53,  3.15s/it]  
 84%|████████▍ | 6053/7175 [1:34:30<52:37,  2.81s/it]
 84%|████████▍ | 6054/7175 [1:34:32<48:15,  2.58s/it]
 84%|████████▍ | 6055/7175 [1:34:34<45:11,  2.42s/it]
 84%|████████▍ | 6056/7175 [1:34:36<43:04,  2.31s/it]
 84%|████████▍ | 6057/7175 [1:34:38<41:28,  2.23s/it]
 84%|████████▍ | 6058/7175 [1:34:40<40:18,  2.17s/it]
 84%|████████▍ | 6059/7175 [1:34:42<39:31,  2.12s/it]
 84%|████████▍ | 6060/7175 [1:34:44<38:57,  2.10s/it]
                                                     
{'loss': '2.951', 'grad_norm': '0.2743', 'learning_rate': '0.0006', 'epoch': '21.12'}
+
 84%|████████▍ | 6060/7175 [1:34:44<38:57,  2.10s/it]
 84%|████████▍ | 6061/7175 [1:34:46<38:54,  2.10s/it]
 84%|████████▍ | 6062/7175 [1:34:48<38:47,  2.09s/it]
 85%|████████▍ | 6063/7175 [1:34:50<38:34,  2.08s/it]
 85%|████████▍ | 6064/7175 [1:34:52<38:27,  2.08s/it]
 85%|████████▍ | 6065/7175 [1:34:54<38:16,  2.07s/it]
 85%|████████▍ | 6066/7175 [1:34:57<38:01,  2.06s/it]
 85%|████████▍ | 6067/7175 [1:34:59<37:52,  2.05s/it]
 85%|████████▍ | 6068/7175 [1:35:01<37:50,  2.05s/it]
 85%|████████▍ | 6069/7175 [1:35:03<37:44,  2.05s/it]
 85%|████████▍ | 6070/7175 [1:35:05<37:41,  2.05s/it]
                                                     
{'loss': '2.953', 'grad_norm': '0.2292', 'learning_rate': '0.0006', 'epoch': '21.15'}
+
 85%|████████▍ | 6070/7175 [1:35:05<37:41,  2.05s/it]
 85%|████████▍ | 6071/7175 [1:35:07<37:45,  2.05s/it]
 85%|████████▍ | 6072/7175 [1:35:09<37:35,  2.04s/it]
 85%|████████▍ | 6073/7175 [1:35:11<37:38,  2.05s/it]
 85%|████████▍ | 6074/7175 [1:35:13<37:44,  2.06s/it]
 85%|████████▍ | 6075/7175 [1:35:15<37:40,  2.05s/it]
 85%|████████▍ | 6076/7175 [1:35:17<37:44,  2.06s/it]
 85%|████████▍ | 6077/7175 [1:35:19<37:38,  2.06s/it]
 85%|████████▍ | 6078/7175 [1:35:21<37:35,  2.06s/it]
 85%|████████▍ | 6079/7175 [1:35:23<37:28,  2.05s/it]
 85%|████████▍ | 6080/7175 [1:35:25<37:21,  2.05s/it]
                                                     
{'loss': '2.954', 'grad_norm': '0.2282', 'learning_rate': '0.0006', 'epoch': '21.19'}
+
 85%|████████▍ | 6080/7175 [1:35:25<37:21,  2.05s/it]
 85%|████████▍ | 6081/7175 [1:35:27<37:21,  2.05s/it]
 85%|████████▍ | 6082/7175 [1:35:29<37:17,  2.05s/it]
 85%|████████▍ | 6083/7175 [1:35:31<37:11,  2.04s/it]
 85%|████████▍ | 6084/7175 [1:35:33<37:08,  2.04s/it]
 85%|████████▍ | 6085/7175 [1:35:35<36:52,  2.03s/it]
 85%|████████▍ | 6086/7175 [1:35:37<37:02,  2.04s/it]
 85%|████████▍ | 6087/7175 [1:35:39<37:05,  2.05s/it]
 85%|████████▍ | 6088/7175 [1:35:42<37:11,  2.05s/it]
 85%|████████▍ | 6089/7175 [1:35:44<37:08,  2.05s/it]
 85%|████████▍ | 6090/7175 [1:35:46<37:01,  2.05s/it]
                                                     
{'loss': '2.955', 'grad_norm': '0.2138', 'learning_rate': '0.0006', 'epoch': '21.22'}
+
 85%|████████▍ | 6090/7175 [1:35:46<37:01,  2.05s/it]
 85%|████████▍ | 6091/7175 [1:35:48<37:06,  2.05s/it]
 85%|████████▍ | 6092/7175 [1:35:50<37:01,  2.05s/it]
 85%|████████▍ | 6093/7175 [1:35:52<36:52,  2.04s/it]
 85%|████████▍ | 6094/7175 [1:35:54<36:48,  2.04s/it]
 85%|████████▍ | 6095/7175 [1:35:56<36:43,  2.04s/it]
 85%|████████▍ | 6096/7175 [1:35:58<36:39,  2.04s/it]
 85%|████████▍ | 6097/7175 [1:36:00<36:19,  2.02s/it]
 85%|████████▍ | 6098/7175 [1:36:02<36:34,  2.04s/it]
 85%|████████▌ | 6099/7175 [1:36:04<36:35,  2.04s/it]
 85%|████████▌ | 6100/7175 [1:36:06<36:46,  2.05s/it]
                                                     
{'loss': '2.962', 'grad_norm': '0.213', 'learning_rate': '0.0006', 'epoch': '21.26'}
+
 85%|████████▌ | 6100/7175 [1:36:06<36:46,  2.05s/it]
 85%|████████▌ | 6101/7175 [1:36:08<36:58,  2.07s/it]
 85%|████████▌ | 6102/7175 [1:36:10<36:45,  2.06s/it]
 85%|████████▌ | 6103/7175 [1:36:12<36:35,  2.05s/it]
 85%|████████▌ | 6104/7175 [1:36:14<36:28,  2.04s/it]
 85%|████████▌ | 6105/7175 [1:36:16<36:22,  2.04s/it]
 85%|████████▌ | 6106/7175 [1:36:18<36:15,  2.04s/it]
 85%|████████▌ | 6107/7175 [1:36:20<36:14,  2.04s/it]
 85%|████████▌ | 6108/7175 [1:36:22<36:15,  2.04s/it]
 85%|████████▌ | 6109/7175 [1:36:24<36:14,  2.04s/it]
 85%|████████▌ | 6110/7175 [1:36:27<36:15,  2.04s/it]
                                                     
{'loss': '2.981', 'grad_norm': '0.233', 'learning_rate': '0.0006', 'epoch': '21.29'}
+
 85%|████████▌ | 6110/7175 [1:36:27<36:15,  2.04s/it]
 85%|████████▌ | 6111/7175 [1:36:29<36:29,  2.06s/it]
 85%|████████▌ | 6112/7175 [1:36:31<36:33,  2.06s/it]
 85%|████████▌ | 6113/7175 [1:36:33<36:32,  2.06s/it]
 85%|████████▌ | 6114/7175 [1:36:35<36:22,  2.06s/it]
 85%|████████▌ | 6115/7175 [1:36:37<36:15,  2.05s/it]
 85%|████████▌ | 6116/7175 [1:36:39<36:09,  2.05s/it]
 85%|████████▌ | 6117/7175 [1:36:41<36:06,  2.05s/it]
 85%|████████▌ | 6118/7175 [1:36:43<35:56,  2.04s/it]
 85%|████████▌ | 6119/7175 [1:36:45<35:50,  2.04s/it]
 85%|████████▌ | 6120/7175 [1:36:47<35:48,  2.04s/it]
                                                     
{'loss': '2.985', 'grad_norm': '0.2457', 'learning_rate': '0.0006', 'epoch': '21.33'}
+
 85%|████████▌ | 6120/7175 [1:36:47<35:48,  2.04s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:04,  7.52it/s]
+
  8%|▊         | 3/39 [00:00<00:06,  5.30it/s]
+
 10%|█         | 4/39 [00:00<00:07,  4.59it/s]
+
 13%|█▎        | 5/39 [00:01<00:07,  4.31it/s]
+
 15%|█▌        | 6/39 [00:01<00:07,  4.13it/s]
+
 18%|█▊        | 7/39 [00:01<00:07,  4.01it/s]
+
 21%|██        | 8/39 [00:01<00:07,  3.90it/s]
+
 23%|██▎       | 9/39 [00:02<00:07,  3.84it/s]
+
 26%|██▌       | 10/39 [00:02<00:07,  3.81it/s]
+
 28%|██▊       | 11/39 [00:02<00:07,  3.78it/s]
+
 31%|███       | 12/39 [00:02<00:07,  3.76it/s]
+
 33%|███▎      | 13/39 [00:03<00:06,  3.75it/s]
+
 36%|███▌      | 14/39 [00:03<00:06,  3.72it/s]
+
 38%|███▊      | 15/39 [00:03<00:06,  3.69it/s]
+
 41%|████      | 16/39 [00:04<00:06,  3.70it/s]
+
 44%|████▎     | 17/39 [00:04<00:05,  3.77it/s]
+
 46%|████▌     | 18/39 [00:04<00:05,  3.78it/s]
+
 49%|████▊     | 19/39 [00:04<00:05,  3.77it/s]
+
 51%|█████▏    | 20/39 [00:05<00:05,  3.75it/s]
+
 54%|█████▍    | 21/39 [00:05<00:04,  3.73it/s]
+
 56%|█████▋    | 22/39 [00:05<00:04,  3.73it/s]
+
 59%|█████▉    | 23/39 [00:05<00:04,  3.75it/s]
+
 62%|██████▏   | 24/39 [00:06<00:03,  3.77it/s]
+
 64%|██████▍   | 25/39 [00:06<00:03,  3.78it/s]
+
 67%|██████▋   | 26/39 [00:06<00:03,  3.72it/s]
+
 69%|██████▉   | 27/39 [00:06<00:03,  3.69it/s]
+
 72%|███████▏  | 28/39 [00:07<00:02,  3.69it/s]
+
 74%|███████▍  | 29/39 [00:07<00:02,  3.71it/s]
+
 77%|███████▋  | 30/39 [00:07<00:02,  3.72it/s]
+
 79%|███████▉  | 31/39 [00:08<00:02,  3.77it/s]
+
 82%|████████▏ | 32/39 [00:08<00:01,  3.69it/s]
+
 85%|████████▍ | 33/39 [00:08<00:01,  3.72it/s]
+
 87%|████████▋ | 34/39 [00:08<00:01,  3.70it/s]
+
 90%|████████▉ | 35/39 [00:09<00:01,  3.70it/s]
+
 92%|█████████▏| 36/39 [00:09<00:00,  3.73it/s]
+
 95%|█████████▍| 37/39 [00:09<00:00,  3.70it/s]
+
 97%|█████████▋| 38/39 [00:09<00:00,  3.69it/s]
+
100%|██████████| 39/39 [00:10<00:00,  3.71it/s]
                                                     
+
                                               
{'eval_loss': '3.522', 'eval_runtime': '10.5', 'eval_samples_per_second': '232.5', 'eval_steps_per_second': '3.715', 'epoch': '21.33'}
+
 85%|████████▌ | 6120/7175 [1:36:58<35:48,  2.04s/it]
+
100%|██████████| 39/39 [00:10<00:00,  3.71it/s][eval] epoch 21.325 step 6120 loss 3.5215 (best 3.5035 @ ep 19.821, 6/20 since best)
+
+
                                               
 85%|████████▌ | 6121/7175 [1:37:00<1:31:21,  5.20s/it]
 85%|████████▌ | 6122/7175 [1:37:02<1:14:37,  4.25s/it]
 85%|████████▌ | 6123/7175 [1:37:03<1:00:50,  3.47s/it]
 85%|████████▌ | 6124/7175 [1:37:05<52:38,  3.01s/it]  
 85%|████████▌ | 6125/7175 [1:37:07<47:26,  2.71s/it]
 85%|████████▌ | 6126/7175 [1:37:09<43:46,  2.50s/it]
 85%|████████▌ | 6127/7175 [1:37:11<41:10,  2.36s/it]
 85%|████████▌ | 6128/7175 [1:37:13<39:11,  2.25s/it]
 85%|████████▌ | 6129/7175 [1:37:15<37:49,  2.17s/it]
 85%|████████▌ | 6130/7175 [1:37:17<37:17,  2.14s/it]
                                                     
{'loss': '2.993', 'grad_norm': '0.2294', 'learning_rate': '0.0006', 'epoch': '21.36'}
+
 85%|████████▌ | 6130/7175 [1:37:17<37:17,  2.14s/it]
 85%|████████▌ | 6131/7175 [1:37:19<37:08,  2.13s/it]
 85%|████████▌ | 6132/7175 [1:37:21<36:45,  2.11s/it]
 85%|████████▌ | 6133/7175 [1:37:24<36:15,  2.09s/it]
 85%|████████▌ | 6134/7175 [1:37:26<36:00,  2.08s/it]
 86%|████████▌ | 6135/7175 [1:37:28<35:47,  2.06s/it]
 86%|████████▌ | 6136/7175 [1:37:30<35:33,  2.05s/it]
 86%|████████▌ | 6137/7175 [1:37:32<35:26,  2.05s/it]
 86%|████████▌ | 6138/7175 [1:37:34<35:17,  2.04s/it]
 86%|████████▌ | 6139/7175 [1:37:36<35:11,  2.04s/it]
 86%|████████▌ | 6140/7175 [1:37:38<35:09,  2.04s/it]
                                                     
{'loss': '2.983', 'grad_norm': '0.2442', 'learning_rate': '0.0006', 'epoch': '21.39'}
+
 86%|████████▌ | 6140/7175 [1:37:38<35:09,  2.04s/it]
 86%|████████▌ | 6141/7175 [1:37:40<35:23,  2.05s/it]
 86%|████████▌ | 6142/7175 [1:37:42<35:29,  2.06s/it]
 86%|████████▌ | 6143/7175 [1:37:44<35:27,  2.06s/it]
 86%|████████▌ | 6144/7175 [1:37:46<35:24,  2.06s/it]
 86%|████████▌ | 6145/7175 [1:37:48<35:16,  2.05s/it]
 86%|████████▌ | 6146/7175 [1:37:50<35:10,  2.05s/it]
 86%|████████▌ | 6147/7175 [1:37:52<35:04,  2.05s/it]
 86%|████████▌ | 6148/7175 [1:37:54<34:59,  2.04s/it]
 86%|████████▌ | 6149/7175 [1:37:56<34:55,  2.04s/it]
 86%|████████▌ | 6150/7175 [1:37:58<34:50,  2.04s/it]
                                                     
{'loss': '2.995', 'grad_norm': '0.2142', 'learning_rate': '0.0006', 'epoch': '21.43'}
+
 86%|████████▌ | 6150/7175 [1:37:58<34:50,  2.04s/it]
 86%|████████▌ | 6151/7175 [1:38:00<34:56,  2.05s/it]
 86%|████████▌ | 6152/7175 [1:38:02<34:37,  2.03s/it]
 86%|████████▌ | 6153/7175 [1:38:04<34:51,  2.05s/it]
 86%|████████▌ | 6154/7175 [1:38:06<34:55,  2.05s/it]
 86%|████████▌ | 6155/7175 [1:38:09<34:58,  2.06s/it]
 86%|████████▌ | 6156/7175 [1:38:11<35:03,  2.06s/it]
 86%|████████▌ | 6157/7175 [1:38:13<34:52,  2.06s/it]
 86%|████████▌ | 6158/7175 [1:38:15<34:45,  2.05s/it]
 86%|████████▌ | 6159/7175 [1:38:17<34:43,  2.05s/it]
 86%|████████▌ | 6160/7175 [1:38:19<34:32,  2.04s/it]
                                                     
{'loss': '2.993', 'grad_norm': '0.2275', 'learning_rate': '0.0006', 'epoch': '21.46'}
+
 86%|████████▌ | 6160/7175 [1:38:19<34:32,  2.04s/it]
 86%|████████▌ | 6161/7175 [1:38:21<34:37,  2.05s/it]
 86%|████████▌ | 6162/7175 [1:38:23<34:31,  2.05s/it]
 86%|████████▌ | 6163/7175 [1:38:25<34:26,  2.04s/it]
 86%|████████▌ | 6164/7175 [1:38:27<34:26,  2.04s/it]
 86%|████████▌ | 6165/7175 [1:38:29<34:30,  2.05s/it]
 86%|████████▌ | 6166/7175 [1:38:31<34:35,  2.06s/it]
 86%|████████▌ | 6167/7175 [1:38:33<34:38,  2.06s/it]
 86%|████████▌ | 6168/7175 [1:38:35<34:38,  2.06s/it]
 86%|████████▌ | 6169/7175 [1:38:37<34:27,  2.06s/it]
 86%|████████▌ | 6170/7175 [1:38:39<34:22,  2.05s/it]
                                                     
{'loss': '2.991', 'grad_norm': '0.2376', 'learning_rate': '0.0006', 'epoch': '21.5'}
+
 86%|████████▌ | 6170/7175 [1:38:39<34:22,  2.05s/it]
 86%|████████▌ | 6171/7175 [1:38:41<34:20,  2.05s/it]
 86%|████████▌ | 6172/7175 [1:38:43<34:12,  2.05s/it]
 86%|████████▌ | 6173/7175 [1:38:45<34:05,  2.04s/it]
 86%|████████▌ | 6174/7175 [1:38:47<34:01,  2.04s/it]
 86%|████████▌ | 6175/7175 [1:38:50<33:56,  2.04s/it]
 86%|████████▌ | 6176/7175 [1:38:52<33:55,  2.04s/it]
 86%|████████▌ | 6177/7175 [1:38:54<34:01,  2.05s/it]
 86%|████████▌ | 6178/7175 [1:38:56<34:06,  2.05s/it]
 86%|████████▌ | 6179/7175 [1:38:58<34:07,  2.06s/it]
 86%|████████▌ | 6180/7175 [1:39:00<34:09,  2.06s/it]
                                                     
{'loss': '2.982', 'grad_norm': '0.2281', 'learning_rate': '0.0006', 'epoch': '21.53'}
+
 86%|████████▌ | 6180/7175 [1:39:00<34:09,  2.06s/it]
 86%|████████▌ | 6181/7175 [1:39:02<34:15,  2.07s/it]
 86%|████████▌ | 6182/7175 [1:39:04<34:11,  2.07s/it]
 86%|████████▌ | 6183/7175 [1:39:06<33:58,  2.06s/it]
 86%|████████▌ | 6184/7175 [1:39:08<33:52,  2.05s/it]
 86%|████████▌ | 6185/7175 [1:39:10<33:50,  2.05s/it]
 86%|████████▌ | 6186/7175 [1:39:12<33:41,  2.04s/it]
 86%|████████▌ | 6187/7175 [1:39:14<33:31,  2.04s/it]
 86%|████████▌ | 6188/7175 [1:39:16<33:31,  2.04s/it]
 86%|████████▋ | 6189/7175 [1:39:18<33:40,  2.05s/it]
 86%|████████▋ | 6190/7175 [1:39:20<33:42,  2.05s/it]
                                                     
{'loss': '2.997', 'grad_norm': '0.2326', 'learning_rate': '0.0006', 'epoch': '21.57'}
+
 86%|████████▋ | 6190/7175 [1:39:20<33:42,  2.05s/it]
 86%|████████▋ | 6191/7175 [1:39:22<33:51,  2.06s/it]
 86%|████████▋ | 6192/7175 [1:39:24<33:53,  2.07s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:04,  7.65it/s]
+
  8%|▊         | 3/39 [00:00<00:06,  5.24it/s]
+
 10%|█         | 4/39 [00:00<00:07,  4.56it/s]
+
 13%|█▎        | 5/39 [00:01<00:08,  4.23it/s]
+
 15%|█▌        | 6/39 [00:01<00:08,  4.09it/s]
+
 18%|█▊        | 7/39 [00:01<00:07,  4.00it/s]
+
 21%|██        | 8/39 [00:01<00:07,  3.91it/s]
+
 23%|██▎       | 9/39 [00:02<00:07,  3.88it/s]
+
 26%|██▌       | 10/39 [00:02<00:07,  3.84it/s]
+
 28%|██▊       | 11/39 [00:02<00:07,  3.85it/s]
+
 31%|███       | 12/39 [00:02<00:07,  3.82it/s]
+
 33%|███▎      | 13/39 [00:03<00:06,  3.79it/s]
+
 36%|███▌      | 14/39 [00:03<00:06,  3.78it/s]
+
 38%|███▊      | 15/39 [00:03<00:06,  3.79it/s]
+
 41%|████      | 16/39 [00:03<00:06,  3.75it/s]
+
 44%|████▎     | 17/39 [00:04<00:05,  3.72it/s]
+
 46%|████▌     | 18/39 [00:04<00:05,  3.73it/s]
+
 49%|████▊     | 19/39 [00:04<00:05,  3.76it/s]
+
 51%|█████▏    | 20/39 [00:05<00:05,  3.77it/s]
+
 54%|█████▍    | 21/39 [00:05<00:04,  3.78it/s]
+
 56%|█████▋    | 22/39 [00:05<00:04,  3.82it/s]
+
 59%|█████▉    | 23/39 [00:05<00:04,  3.74it/s]
+
 62%|██████▏   | 24/39 [00:06<00:03,  3.81it/s]
+
 64%|██████▍   | 25/39 [00:06<00:03,  3.79it/s]
+
 67%|██████▋   | 26/39 [00:06<00:03,  3.78it/s]
+
 69%|██████▉   | 27/39 [00:06<00:03,  3.76it/s]
+
 72%|███████▏  | 28/39 [00:07<00:02,  3.78it/s]
+
 74%|███████▍  | 29/39 [00:07<00:02,  3.82it/s]
+
 77%|███████▋  | 30/39 [00:07<00:02,  3.79it/s]
+
 79%|███████▉  | 31/39 [00:07<00:02,  3.79it/s]
+
 82%|████████▏ | 32/39 [00:08<00:01,  3.80it/s]
+
 85%|████████▍ | 33/39 [00:08<00:01,  3.80it/s]
+
 87%|████████▋ | 34/39 [00:08<00:01,  3.77it/s]
+
 90%|████████▉ | 35/39 [00:09<00:01,  3.82it/s]
+
 92%|█████████▏| 36/39 [00:09<00:00,  3.77it/s]
+
 95%|█████████▍| 37/39 [00:09<00:00,  3.77it/s]
+
 97%|█████████▋| 38/39 [00:09<00:00,  3.77it/s]
+
100%|██████████| 39/39 [00:10<00:00,  3.78it/s]
                                                     
+
                                               
{'eval_loss': '3.511', 'eval_runtime': '10.38', 'eval_samples_per_second': '235.3', 'eval_steps_per_second': '3.759', 'epoch': '21.58'}
+
 86%|████████▋ | 6192/7175 [1:39:35<33:53,  2.07s/it]
+
100%|██████████| 39/39 [00:10<00:00,  3.78it/s][eval] epoch 21.577 step 6192 loss 3.5110 (best 3.5035 @ ep 19.821, 7/20 since best)
+
+
                                               
 86%|████████▋ | 6193/7175 [1:39:37<1:24:33,  5.17s/it]
 86%|████████▋ | 6194/7175 [1:39:39<1:09:03,  4.22s/it]
 86%|████████▋ | 6195/7175 [1:39:41<58:19,  3.57s/it]  
 86%|████████▋ | 6196/7175 [1:39:43<50:57,  3.12s/it]
 86%|████████▋ | 6197/7175 [1:39:45<45:44,  2.81s/it]
 86%|████████▋ | 6198/7175 [1:39:47<42:03,  2.58s/it]
 86%|████████▋ | 6199/7175 [1:39:49<39:34,  2.43s/it]
 86%|████████▋ | 6200/7175 [1:39:51<37:38,  2.32s/it]
                                                     
{'loss': '3.002', 'grad_norm': '0.227', 'learning_rate': '0.0006', 'epoch': '21.6'}
+
 86%|████████▋ | 6200/7175 [1:39:51<37:38,  2.32s/it]
 86%|████████▋ | 6201/7175 [1:39:53<36:25,  2.24s/it]
 86%|████████▋ | 6202/7175 [1:39:55<35:25,  2.18s/it]
 86%|████████▋ | 6203/7175 [1:39:57<34:39,  2.14s/it]
 86%|████████▋ | 6204/7175 [1:39:59<34:08,  2.11s/it]
 86%|████████▋ | 6205/7175 [1:40:02<33:46,  2.09s/it]
 86%|████████▋ | 6206/7175 [1:40:04<33:28,  2.07s/it]
 87%|████████▋ | 6207/7175 [1:40:06<33:18,  2.06s/it]
 87%|████████▋ | 6208/7175 [1:40:08<33:17,  2.07s/it]
 87%|████████▋ | 6209/7175 [1:40:10<33:19,  2.07s/it]
 87%|████████▋ | 6210/7175 [1:40:12<33:19,  2.07s/it]
                                                     
{'loss': '2.996', 'grad_norm': '0.2191', 'learning_rate': '0.0006', 'epoch': '21.64'}
+
 87%|████████▋ | 6210/7175 [1:40:12<33:19,  2.07s/it]
 87%|████████▋ | 6211/7175 [1:40:14<33:26,  2.08s/it]
 87%|████████▋ | 6212/7175 [1:40:16<33:13,  2.07s/it]
 87%|████████▋ | 6213/7175 [1:40:18<33:04,  2.06s/it]
 87%|████████▋ | 6214/7175 [1:40:20<32:56,  2.06s/it]
 87%|████████▋ | 6215/7175 [1:40:22<32:50,  2.05s/it]
 87%|████████▋ | 6216/7175 [1:40:24<32:42,  2.05s/it]
 87%|████████▋ | 6217/7175 [1:40:26<32:39,  2.05s/it]
 87%|████████▋ | 6218/7175 [1:40:28<32:36,  2.04s/it]
 87%|████████▋ | 6219/7175 [1:40:30<32:38,  2.05s/it]
 87%|████████▋ | 6220/7175 [1:40:32<32:37,  2.05s/it]
                                                     
{'loss': '2.997', 'grad_norm': '0.2367', 'learning_rate': '0.0006', 'epoch': '21.67'}
+
 87%|████████▋ | 6220/7175 [1:40:32<32:37,  2.05s/it]
 87%|████████▋ | 6221/7175 [1:40:34<32:47,  2.06s/it]
 87%|████████▋ | 6222/7175 [1:40:36<32:48,  2.07s/it]
 87%|████████▋ | 6223/7175 [1:40:39<32:48,  2.07s/it]
 87%|████████▋ | 6224/7175 [1:40:41<32:39,  2.06s/it]
 87%|████████▋ | 6225/7175 [1:40:43<32:31,  2.05s/it]
 87%|████████▋ | 6226/7175 [1:40:45<32:25,  2.05s/it]
 87%|████████▋ | 6227/7175 [1:40:47<32:19,  2.05s/it]
 87%|████████▋ | 6228/7175 [1:40:49<32:11,  2.04s/it]
 87%|████████▋ | 6229/7175 [1:40:51<32:11,  2.04s/it]
 87%|████████▋ | 6230/7175 [1:40:53<32:10,  2.04s/it]
                                                     
{'loss': '3.012', 'grad_norm': '0.2174', 'learning_rate': '0.0006', 'epoch': '21.71'}
+
 87%|████████▋ | 6230/7175 [1:40:53<32:10,  2.04s/it]
 87%|████████▋ | 6231/7175 [1:40:55<32:07,  2.04s/it]
 87%|████████▋ | 6232/7175 [1:40:57<32:08,  2.05s/it]
 87%|████████▋ | 6233/7175 [1:40:59<32:14,  2.05s/it]
 87%|████████▋ | 6234/7175 [1:41:01<32:15,  2.06s/it]
 87%|████████▋ | 6235/7175 [1:41:03<32:14,  2.06s/it]
 87%|████████▋ | 6236/7175 [1:41:05<32:10,  2.06s/it]
 87%|████████▋ | 6237/7175 [1:41:07<32:05,  2.05s/it]
 87%|████████▋ | 6238/7175 [1:41:09<32:00,  2.05s/it]
 87%|████████▋ | 6239/7175 [1:41:11<31:57,  2.05s/it]
 87%|████████▋ | 6240/7175 [1:41:13<31:54,  2.05s/it]
                                                     
{'loss': '3.006', 'grad_norm': '0.2203', 'learning_rate': '0.0006', 'epoch': '21.74'}
+
 87%|████████▋ | 6240/7175 [1:41:13<31:54,  2.05s/it]
 87%|████████▋ | 6241/7175 [1:41:15<31:54,  2.05s/it]
 87%|████████▋ | 6242/7175 [1:41:17<31:48,  2.05s/it]
 87%|████████▋ | 6243/7175 [1:41:19<31:45,  2.04s/it]
 87%|████████▋ | 6244/7175 [1:41:22<31:50,  2.05s/it]
 87%|████████▋ | 6245/7175 [1:41:24<31:51,  2.05s/it]
 87%|████████▋ | 6246/7175 [1:41:26<31:58,  2.07s/it]
 87%|████████▋ | 6247/7175 [1:41:28<32:00,  2.07s/it]
 87%|████████▋ | 6248/7175 [1:41:30<31:54,  2.06s/it]
 87%|████████▋ | 6249/7175 [1:41:32<31:45,  2.06s/it]
 87%|████████▋ | 6250/7175 [1:41:34<31:38,  2.05s/it]
                                                     
{'loss': '3.017', 'grad_norm': '0.2229', 'learning_rate': '0.0006', 'epoch': '21.78'}
+
 87%|████████▋ | 6250/7175 [1:41:34<31:38,  2.05s/it]
 87%|████████▋ | 6251/7175 [1:41:36<31:42,  2.06s/it]
 87%|████████▋ | 6252/7175 [1:41:38<31:37,  2.06s/it]
 87%|████████▋ | 6253/7175 [1:41:40<31:33,  2.05s/it]
 87%|████████▋ | 6254/7175 [1:41:42<31:24,  2.05s/it]
 87%|████████▋ | 6255/7175 [1:41:44<31:21,  2.04s/it]
 87%|████████▋ | 6256/7175 [1:41:46<31:20,  2.05s/it]
 87%|████████▋ | 6257/7175 [1:41:48<31:23,  2.05s/it]
 87%|████████▋ | 6258/7175 [1:41:50<31:24,  2.05s/it]
 87%|████████▋ | 6259/7175 [1:41:52<31:23,  2.06s/it]
 87%|████████▋ | 6260/7175 [1:41:54<31:20,  2.06s/it]
                                                     
{'loss': '3.001', 'grad_norm': '0.2243', 'learning_rate': '0.0006', 'epoch': '21.81'}
+
 87%|████████▋ | 6260/7175 [1:41:55<31:20,  2.06s/it]
 87%|████████▋ | 6261/7175 [1:41:57<31:24,  2.06s/it]
 87%|████████▋ | 6262/7175 [1:41:59<31:14,  2.05s/it]
 87%|████████▋ | 6263/7175 [1:42:01<31:09,  2.05s/it]
 87%|████████▋ | 6264/7175 [1:42:03<31:08,  2.05s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:04,  7.90it/s]
+
  8%|▊         | 3/39 [00:00<00:06,  5.46it/s]
+
 10%|█         | 4/39 [00:00<00:07,  4.65it/s]
+
 13%|█▎        | 5/39 [00:01<00:07,  4.32it/s]
+
 15%|█▌        | 6/39 [00:01<00:07,  4.18it/s]
+
 18%|█▊        | 7/39 [00:01<00:07,  4.00it/s]
+
 21%|██        | 8/39 [00:01<00:07,  3.95it/s]
+
 23%|██▎       | 9/39 [00:02<00:07,  3.92it/s]
+
 26%|██▌       | 10/39 [00:02<00:07,  3.84it/s]
+
 28%|██▊       | 11/39 [00:02<00:07,  3.85it/s]
+
 31%|███       | 12/39 [00:02<00:07,  3.79it/s]
+
 33%|███▎      | 13/39 [00:03<00:06,  3.81it/s]
+
 36%|███▌      | 14/39 [00:03<00:06,  3.82it/s]
+
 38%|███▊      | 15/39 [00:03<00:06,  3.76it/s]
+
 41%|████      | 16/39 [00:03<00:06,  3.78it/s]
+
 44%|████▎     | 17/39 [00:04<00:05,  3.76it/s]
+
 46%|████▌     | 18/39 [00:04<00:05,  3.78it/s]
+
 49%|████▊     | 19/39 [00:04<00:05,  3.79it/s]
+
 51%|█████▏    | 20/39 [00:05<00:05,  3.77it/s]
+
 54%|█████▍    | 21/39 [00:05<00:04,  3.73it/s]
+
 56%|█████▋    | 22/39 [00:05<00:04,  3.76it/s]
+
 59%|█████▉    | 23/39 [00:05<00:04,  3.75it/s]
+
 62%|██████▏   | 24/39 [00:06<00:03,  3.83it/s]
+
 64%|██████▍   | 25/39 [00:06<00:03,  3.86it/s]
+
 67%|██████▋   | 26/39 [00:06<00:03,  3.84it/s]
+
 69%|██████▉   | 27/39 [00:06<00:03,  3.82it/s]
+
 72%|███████▏  | 28/39 [00:07<00:02,  3.77it/s]
+
 74%|███████▍  | 29/39 [00:07<00:02,  3.74it/s]
+
 77%|███████▋  | 30/39 [00:07<00:02,  3.75it/s]
+
 79%|███████▉  | 31/39 [00:07<00:02,  3.78it/s]
+
 82%|████████▏ | 32/39 [00:08<00:01,  3.77it/s]
+
 85%|████████▍ | 33/39 [00:08<00:01,  3.75it/s]
+
 87%|████████▋ | 34/39 [00:08<00:01,  3.76it/s]
+
 90%|████████▉ | 35/39 [00:08<00:01,  3.75it/s]
+
 92%|█████████▏| 36/39 [00:09<00:00,  3.70it/s]
+
 95%|█████████▍| 37/39 [00:09<00:00,  3.69it/s]
+
 97%|█████████▋| 38/39 [00:09<00:00,  3.73it/s]
+
100%|██████████| 39/39 [00:10<00:00,  3.73it/s]
                                                     
+
                                               
{'eval_loss': '3.504', 'eval_runtime': '10.39', 'eval_samples_per_second': '234.8', 'eval_steps_per_second': '3.752', 'epoch': '21.83'}
+
 87%|████████▋ | 6264/7175 [1:42:13<31:08,  2.05s/it]
+
100%|██████████| 39/39 [00:10<00:00,  3.73it/s][eval] epoch 21.828 step 6264 loss 3.5036 (best 3.5035 @ ep 19.821, 8/20 since best)
+
+
                                               
 87%|████████▋ | 6265/7175 [1:42:15<1:18:28,  5.17s/it]
 87%|████████▋ | 6266/7175 [1:42:17<1:04:19,  4.25s/it]
 87%|████████▋ | 6267/7175 [1:42:19<54:19,  3.59s/it]  
 87%|████████▋ | 6268/7175 [1:42:21<47:16,  3.13s/it]
 87%|████████▋ | 6269/7175 [1:42:23<42:14,  2.80s/it]
 87%|████████▋ | 6270/7175 [1:42:25<38:44,  2.57s/it]
                                                     
{'loss': '3.011', 'grad_norm': '0.2135', 'learning_rate': '0.0006', 'epoch': '21.85'}
+
 87%|████████▋ | 6270/7175 [1:42:25<38:44,  2.57s/it]
 87%|████████▋ | 6271/7175 [1:42:27<36:28,  2.42s/it]
 87%|████████▋ | 6272/7175 [1:42:30<34:44,  2.31s/it]
 87%|████████▋ | 6273/7175 [1:42:32<33:33,  2.23s/it]
 87%|████████▋ | 6274/7175 [1:42:34<32:36,  2.17s/it]
 87%|████████▋ | 6275/7175 [1:42:36<32:07,  2.14s/it]
 87%|████████▋ | 6276/7175 [1:42:38<31:43,  2.12s/it]
 87%|████████▋ | 6277/7175 [1:42:40<31:29,  2.10s/it]
 87%|████████▋ | 6278/7175 [1:42:42<31:16,  2.09s/it]
 88%|████████▊ | 6279/7175 [1:42:44<31:00,  2.08s/it]
 88%|████████▊ | 6280/7175 [1:42:46<30:52,  2.07s/it]
                                                     
{'loss': '3.011', 'grad_norm': '0.2052', 'learning_rate': '0.0006', 'epoch': '21.88'}
+
 88%|████████▊ | 6280/7175 [1:42:46<30:52,  2.07s/it]
 88%|████████▊ | 6281/7175 [1:42:48<30:50,  2.07s/it]
 88%|████████▊ | 6282/7175 [1:42:50<30:38,  2.06s/it]
 88%|████████▊ | 6283/7175 [1:42:52<30:28,  2.05s/it]
 88%|████████▊ | 6284/7175 [1:42:54<28:57,  1.95s/it]
 88%|████████▊ | 6285/7175 [1:42:56<28:10,  1.90s/it]
 88%|████████▊ | 6286/7175 [1:42:58<28:37,  1.93s/it]
 88%|████████▊ | 6287/7175 [1:43:00<29:08,  1.97s/it]
 88%|████████▊ | 6288/7175 [1:43:02<29:26,  1.99s/it]
 88%|████████▊ | 6289/7175 [1:43:04<29:31,  2.00s/it]
 88%|████████▊ | 6290/7175 [1:43:06<29:28,  2.00s/it]
                                                     
{'loss': '3.018', 'grad_norm': '0.2365', 'learning_rate': '0.0006', 'epoch': '21.92'}
+
 88%|████████▊ | 6290/7175 [1:43:06<29:28,  2.00s/it]
 88%|████████▊ | 6291/7175 [1:43:08<29:46,  2.02s/it]
 88%|████████▊ | 6292/7175 [1:43:10<29:52,  2.03s/it]
 88%|████████▊ | 6293/7175 [1:43:12<29:52,  2.03s/it]
 88%|████████▊ | 6294/7175 [1:43:14<29:54,  2.04s/it]
 88%|████████▊ | 6295/7175 [1:43:16<29:53,  2.04s/it]
 88%|████████▊ | 6296/7175 [1:43:18<29:55,  2.04s/it]
 88%|████████▊ | 6297/7175 [1:43:20<29:52,  2.04s/it]
 88%|████████▊ | 6298/7175 [1:43:22<29:40,  2.03s/it]
 88%|████████▊ | 6299/7175 [1:43:24<29:48,  2.04s/it]
 88%|████████▊ | 6300/7175 [1:43:26<29:52,  2.05s/it]
                                                     
{'loss': '3.006', 'grad_norm': '0.2326', 'learning_rate': '0.0006', 'epoch': '21.95'}
+
 88%|████████▊ | 6300/7175 [1:43:26<29:52,  2.05s/it]
 88%|████████▊ | 6301/7175 [1:43:28<30:04,  2.06s/it]
 88%|████████▊ | 6302/7175 [1:43:30<30:06,  2.07s/it]
 88%|████████▊ | 6303/7175 [1:43:32<30:00,  2.07s/it]
 88%|████████▊ | 6304/7175 [1:43:34<29:49,  2.05s/it]
 88%|████████▊ | 6305/7175 [1:43:36<29:39,  2.05s/it]
 88%|████████▊ | 6306/7175 [1:43:39<29:40,  2.05s/it]
 88%|████████▊ | 6307/7175 [1:43:41<29:35,  2.05s/it]
 88%|████████▊ | 6308/7175 [1:43:43<29:35,  2.05s/it]
 88%|████████▊ | 6309/7175 [1:43:45<29:30,  2.04s/it]
 88%|████████▊ | 6310/7175 [1:43:47<29:25,  2.04s/it]
                                                     
{'loss': '2.996', 'grad_norm': '0.2249', 'learning_rate': '0.0006', 'epoch': '21.99'}
+
 88%|████████▊ | 6310/7175 [1:43:47<29:25,  2.04s/it]
 88%|████████▊ | 6311/7175 [1:43:49<29:35,  2.06s/it]
 88%|████████▊ | 6312/7175 [1:43:51<29:39,  2.06s/it]
 88%|████████▊ | 6313/7175 [1:43:53<29:38,  2.06s/it]
 88%|████████▊ | 6314/7175 [1:43:53<22:06,  1.54s/it]
 88%|████████▊ | 6315/7175 [1:43:55<24:44,  1.73s/it]
 88%|████████▊ | 6316/7175 [1:43:57<26:04,  1.82s/it]
 88%|████████▊ | 6317/7175 [1:43:59<26:57,  1.89s/it]
 88%|████████▊ | 6318/7175 [1:44:02<27:37,  1.93s/it]
 88%|████████▊ | 6319/7175 [1:44:04<28:01,  1.96s/it]
 88%|████████▊ | 6320/7175 [1:44:06<28:17,  1.99s/it]
                                                     
{'loss': '2.975', 'grad_norm': '0.3149', 'learning_rate': '0.0006', 'epoch': '22.02'}
+
 88%|████████▊ | 6320/7175 [1:44:06<28:17,  1.99s/it]
 88%|████████▊ | 6321/7175 [1:44:08<28:37,  2.01s/it]
 88%|████████▊ | 6322/7175 [1:44:10<28:41,  2.02s/it]
 88%|████████▊ | 6323/7175 [1:44:12<28:45,  2.03s/it]
 88%|████████▊ | 6324/7175 [1:44:14<28:55,  2.04s/it]
 88%|████████▊ | 6325/7175 [1:44:16<29:05,  2.05s/it]
 88%|████████▊ | 6326/7175 [1:44:18<29:12,  2.06s/it]
 88%|████████▊ | 6327/7175 [1:44:20<29:17,  2.07s/it]
 88%|████████▊ | 6328/7175 [1:44:22<29:05,  2.06s/it]
 88%|████████▊ | 6329/7175 [1:44:24<28:57,  2.05s/it]
 88%|████████▊ | 6330/7175 [1:44:26<28:49,  2.05s/it]
                                                     
{'loss': '2.95', 'grad_norm': '0.2619', 'learning_rate': '0.0006', 'epoch': '22.06'}
+
 88%|████████▊ | 6330/7175 [1:44:26<28:49,  2.05s/it]
 88%|████████▊ | 6331/7175 [1:44:28<28:51,  2.05s/it]
 88%|████████▊ | 6332/7175 [1:44:30<28:47,  2.05s/it]
 88%|████████▊ | 6333/7175 [1:44:32<28:43,  2.05s/it]
 88%|████████▊ | 6334/7175 [1:44:34<28:40,  2.05s/it]
 88%|████████▊ | 6335/7175 [1:44:36<28:42,  2.05s/it]
 88%|████████▊ | 6336/7175 [1:44:39<28:48,  2.06s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:04,  7.67it/s]
+
  8%|▊         | 3/39 [00:00<00:06,  5.31it/s]
+
 10%|█         | 4/39 [00:00<00:07,  4.58it/s]
+
 13%|█▎        | 5/39 [00:01<00:08,  4.09it/s]
+
 15%|█▌        | 6/39 [00:01<00:08,  4.08it/s]
+
 18%|█▊        | 7/39 [00:01<00:08,  3.98it/s]
+
 21%|██        | 8/39 [00:01<00:08,  3.87it/s]
+
 23%|██▎       | 9/39 [00:02<00:07,  3.85it/s]
+
 26%|██▌       | 10/39 [00:02<00:07,  3.78it/s]
+
 28%|██▊       | 11/39 [00:02<00:07,  3.77it/s]
+
 31%|███       | 12/39 [00:02<00:07,  3.73it/s]
+
 33%|███▎      | 13/39 [00:03<00:06,  3.73it/s]
+
 36%|███▌      | 14/39 [00:03<00:06,  3.69it/s]
+
 38%|███▊      | 15/39 [00:03<00:06,  3.71it/s]
+
 41%|████      | 16/39 [00:04<00:06,  3.75it/s]
+
 44%|████▎     | 17/39 [00:04<00:05,  3.77it/s]
+
 46%|████▌     | 18/39 [00:04<00:05,  3.79it/s]
+
 49%|████▊     | 19/39 [00:04<00:05,  3.79it/s]
+
 51%|█████▏    | 20/39 [00:05<00:05,  3.77it/s]
+
 54%|█████▍    | 21/39 [00:05<00:04,  3.74it/s]
+
 56%|█████▋    | 22/39 [00:05<00:04,  3.75it/s]
+
 59%|█████▉    | 23/39 [00:05<00:04,  3.76it/s]
+
 62%|██████▏   | 24/39 [00:06<00:03,  3.77it/s]
+
 64%|██████▍   | 25/39 [00:06<00:03,  3.73it/s]
+
 67%|██████▋   | 26/39 [00:06<00:03,  3.79it/s]
+
 69%|██████▉   | 27/39 [00:06<00:03,  3.76it/s]
+
 72%|███████▏  | 28/39 [00:07<00:02,  3.73it/s]
+
 74%|███████▍  | 29/39 [00:07<00:02,  3.74it/s]
+
 77%|███████▋  | 30/39 [00:07<00:02,  3.66it/s]
+
 79%|███████▉  | 31/39 [00:08<00:02,  3.72it/s]
+
 82%|████████▏ | 32/39 [00:08<00:01,  3.75it/s]
+
 85%|████████▍ | 33/39 [00:08<00:01,  3.74it/s]
+
 87%|████████▋ | 34/39 [00:08<00:01,  3.79it/s]
+
 90%|████████▉ | 35/39 [00:09<00:01,  3.78it/s]
+
 92%|█████████▏| 36/39 [00:09<00:00,  3.80it/s]
+
 95%|█████████▍| 37/39 [00:09<00:00,  3.84it/s]
+
 97%|█████████▋| 38/39 [00:09<00:00,  3.78it/s]
+
100%|██████████| 39/39 [00:10<00:00,  3.77it/s]
                                                     
+
                                               
{'eval_loss': '3.534', 'eval_runtime': '10.46', 'eval_samples_per_second': '233.4', 'eval_steps_per_second': '3.729', 'epoch': '22.08'}
+
 88%|████████▊ | 6336/7175 [1:44:49<28:48,  2.06s/it]
+
100%|██████████| 39/39 [00:10<00:00,  3.77it/s][eval] epoch 22.077 step 6336 loss 3.5345 (best 3.5035 @ ep 19.821, 9/20 since best)
+
+
                                               
 88%|████████▊ | 6337/7175 [1:44:51<1:12:27,  5.19s/it]
 88%|████████▊ | 6338/7175 [1:44:53<59:11,  4.24s/it]  
 88%|████████▊ | 6339/7175 [1:44:55<49:52,  3.58s/it]
 88%|████████▊ | 6340/7175 [1:44:57<43:25,  3.12s/it]
                                                     
{'loss': '2.94', 'grad_norm': '0.227', 'learning_rate': '0.0006', 'epoch': '22.09'}
+
 88%|████████▊ | 6340/7175 [1:44:57<43:25,  3.12s/it]
 88%|████████▊ | 6341/7175 [1:44:59<38:58,  2.80s/it]
 88%|████████▊ | 6342/7175 [1:45:01<35:53,  2.58s/it]
 88%|████████▊ | 6343/7175 [1:45:03<33:39,  2.43s/it]
 88%|████████▊ | 6344/7175 [1:45:05<32:06,  2.32s/it]
 88%|████████▊ | 6345/7175 [1:45:07<31:04,  2.25s/it]
 88%|████████▊ | 6346/7175 [1:45:10<30:16,  2.19s/it]
 88%|████████▊ | 6347/7175 [1:45:12<29:39,  2.15s/it]
 88%|████████▊ | 6348/7175 [1:45:14<29:09,  2.12s/it]
 88%|████████▊ | 6349/7175 [1:45:16<28:48,  2.09s/it]
 89%|████████▊ | 6350/7175 [1:45:18<28:32,  2.08s/it]
                                                     
{'loss': '2.946', 'grad_norm': '0.2168', 'learning_rate': '0.0006', 'epoch': '22.13'}
+
 89%|████████▊ | 6350/7175 [1:45:18<28:32,  2.08s/it]
 89%|████████▊ | 6351/7175 [1:45:20<28:27,  2.07s/it]
 89%|████████▊ | 6352/7175 [1:45:22<28:17,  2.06s/it]
 89%|████████▊ | 6353/7175 [1:45:24<28:08,  2.05s/it]
 89%|████████▊ | 6354/7175 [1:45:26<27:56,  2.04s/it]
 89%|████████▊ | 6355/7175 [1:45:28<28:03,  2.05s/it]
 89%|████████▊ | 6356/7175 [1:45:30<28:08,  2.06s/it]
 89%|████████▊ | 6357/7175 [1:45:32<28:07,  2.06s/it]
 89%|████████▊ | 6358/7175 [1:45:34<28:04,  2.06s/it]
 89%|████████▊ | 6359/7175 [1:45:36<27:55,  2.05s/it]
 89%|████████▊ | 6360/7175 [1:45:38<27:52,  2.05s/it]
                                                     
{'loss': '2.951', 'grad_norm': '0.2299', 'learning_rate': '0.0006', 'epoch': '22.16'}
+
 89%|████████▊ | 6360/7175 [1:45:38<27:52,  2.05s/it]
 89%|████████▊ | 6361/7175 [1:45:40<27:54,  2.06s/it]
 89%|████████▊ | 6362/7175 [1:45:42<27:47,  2.05s/it]
 89%|████████▊ | 6363/7175 [1:45:44<27:46,  2.05s/it]
 89%|████████▊ | 6364/7175 [1:45:46<27:39,  2.05s/it]
 89%|████████▊ | 6365/7175 [1:45:48<27:35,  2.04s/it]
 89%|████████▊ | 6366/7175 [1:45:50<27:34,  2.05s/it]
 89%|████████▊ | 6367/7175 [1:45:53<27:38,  2.05s/it]
 89%|████████▉ | 6368/7175 [1:45:55<27:41,  2.06s/it]
 89%|████████▉ | 6369/7175 [1:45:57<27:42,  2.06s/it]
 89%|████████▉ | 6370/7175 [1:45:59<27:42,  2.07s/it]
                                                     
{'loss': '2.938', 'grad_norm': '0.2204', 'learning_rate': '0.0006', 'epoch': '22.2'}
+
 89%|████████▉ | 6370/7175 [1:45:59<27:42,  2.07s/it]
 89%|████████▉ | 6371/7175 [1:46:01<27:41,  2.07s/it]
 89%|████████▉ | 6372/7175 [1:46:03<27:38,  2.07s/it]
 89%|████████▉ | 6373/7175 [1:46:05<27:32,  2.06s/it]
 89%|████████▉ | 6374/7175 [1:46:07<27:28,  2.06s/it]
 89%|████████▉ | 6375/7175 [1:46:09<27:21,  2.05s/it]
 89%|████████▉ | 6376/7175 [1:46:11<27:17,  2.05s/it]
 89%|████████▉ | 6377/7175 [1:46:13<27:10,  2.04s/it]
 89%|████████▉ | 6378/7175 [1:46:15<27:10,  2.05s/it]
 89%|████████▉ | 6379/7175 [1:46:17<27:13,  2.05s/it]
 89%|████████▉ | 6380/7175 [1:46:19<27:16,  2.06s/it]
                                                     
{'loss': '2.948', 'grad_norm': '0.2274', 'learning_rate': '0.0006', 'epoch': '22.23'}
+
 89%|████████▉ | 6380/7175 [1:46:19<27:16,  2.06s/it]
 89%|████████▉ | 6381/7175 [1:46:21<27:26,  2.07s/it]
 89%|████████▉ | 6382/7175 [1:46:23<27:19,  2.07s/it]
 89%|████████▉ | 6383/7175 [1:46:26<27:10,  2.06s/it]
 89%|████████▉ | 6384/7175 [1:46:28<27:02,  2.05s/it]
 89%|████████▉ | 6385/7175 [1:46:30<26:58,  2.05s/it]
 89%|████████▉ | 6386/7175 [1:46:32<26:52,  2.04s/it]
 89%|████████▉ | 6387/7175 [1:46:34<26:47,  2.04s/it]
 89%|████████▉ | 6388/7175 [1:46:36<26:46,  2.04s/it]
 89%|████████▉ | 6389/7175 [1:46:38<26:42,  2.04s/it]
 89%|████████▉ | 6390/7175 [1:46:40<26:40,  2.04s/it]
                                                     
{'loss': '2.957', 'grad_norm': '0.2345', 'learning_rate': '0.0006', 'epoch': '22.27'}
+
 89%|████████▉ | 6390/7175 [1:46:40<26:40,  2.04s/it]
 89%|████████▉ | 6391/7175 [1:46:42<26:53,  2.06s/it]
 89%|████████▉ | 6392/7175 [1:46:44<26:51,  2.06s/it]
 89%|████████▉ | 6393/7175 [1:46:46<26:53,  2.06s/it]
 89%|████████▉ | 6394/7175 [1:46:48<26:51,  2.06s/it]
 89%|████████▉ | 6395/7175 [1:46:50<26:46,  2.06s/it]
 89%|████████▉ | 6396/7175 [1:46:52<26:41,  2.06s/it]
 89%|████████▉ | 6397/7175 [1:46:54<26:34,  2.05s/it]
 89%|████████▉ | 6398/7175 [1:46:56<26:31,  2.05s/it]
 89%|████████▉ | 6399/7175 [1:46:58<26:26,  2.04s/it]
 89%|████████▉ | 6400/7175 [1:47:00<26:24,  2.05s/it]
                                                     
{'loss': '2.955', 'grad_norm': '0.2436', 'learning_rate': '0.0006', 'epoch': '22.3'}
+
 89%|████████▉ | 6400/7175 [1:47:00<26:24,  2.05s/it]
 89%|████████▉ | 6401/7175 [1:47:02<26:28,  2.05s/it]
 89%|████████▉ | 6402/7175 [1:47:04<26:27,  2.05s/it]
 89%|████████▉ | 6403/7175 [1:47:07<26:27,  2.06s/it]
 89%|████████▉ | 6404/7175 [1:47:09<26:28,  2.06s/it]
 89%|████████▉ | 6405/7175 [1:47:11<26:28,  2.06s/it]
 89%|████████▉ | 6406/7175 [1:47:13<26:28,  2.07s/it]
 89%|████████▉ | 6407/7175 [1:47:15<26:20,  2.06s/it]
 89%|████████▉ | 6408/7175 [1:47:17<26:13,  2.05s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:04,  7.82it/s]
+
  8%|▊         | 3/39 [00:00<00:06,  5.37it/s]
+
 10%|█         | 4/39 [00:00<00:07,  4.67it/s]
+
 13%|█▎        | 5/39 [00:01<00:07,  4.32it/s]
+
 15%|█▌        | 6/39 [00:01<00:08,  4.11it/s]
+
 18%|█▊        | 7/39 [00:01<00:08,  3.97it/s]
+
 21%|██        | 8/39 [00:01<00:07,  3.88it/s]
+
 23%|██▎       | 9/39 [00:02<00:07,  3.87it/s]
+
 26%|██▌       | 10/39 [00:02<00:07,  3.82it/s]
+
 28%|██▊       | 11/39 [00:02<00:07,  3.83it/s]
+
 31%|███       | 12/39 [00:02<00:07,  3.79it/s]
+
 33%|███▎      | 13/39 [00:03<00:06,  3.76it/s]
+
 36%|███▌      | 14/39 [00:03<00:06,  3.81it/s]
+
 38%|███▊      | 15/39 [00:03<00:06,  3.80it/s]
+
 41%|████      | 16/39 [00:03<00:06,  3.80it/s]
+
 44%|████▎     | 17/39 [00:04<00:05,  3.81it/s]
+
 46%|████▌     | 18/39 [00:04<00:05,  3.80it/s]
+
 49%|████▊     | 19/39 [00:04<00:05,  3.80it/s]
+
 51%|█████▏    | 20/39 [00:05<00:05,  3.79it/s]
+
 54%|█████▍    | 21/39 [00:05<00:04,  3.79it/s]
+
 56%|█████▋    | 22/39 [00:05<00:04,  3.75it/s]
+
 59%|█████▉    | 23/39 [00:05<00:04,  3.72it/s]
+
 62%|██████▏   | 24/39 [00:06<00:03,  3.76it/s]
+
 64%|██████▍   | 25/39 [00:06<00:03,  3.76it/s]
+
 67%|██████▋   | 26/39 [00:06<00:03,  3.77it/s]
+
 69%|██████▉   | 27/39 [00:06<00:03,  3.79it/s]
+
 72%|███████▏  | 28/39 [00:07<00:02,  3.80it/s]
+
 74%|███████▍  | 29/39 [00:07<00:02,  3.78it/s]
+
 77%|███████▋  | 30/39 [00:07<00:02,  3.82it/s]
+
 79%|███████▉  | 31/39 [00:07<00:02,  3.77it/s]
+
 82%|████████▏ | 32/39 [00:08<00:01,  3.76it/s]
+
 85%|████████▍ | 33/39 [00:08<00:01,  3.76it/s]
+
 87%|████████▋ | 34/39 [00:08<00:01,  3.74it/s]
+
 90%|████████▉ | 35/39 [00:09<00:01,  3.72it/s]
+
 92%|█████████▏| 36/39 [00:09<00:00,  3.74it/s]
+
 95%|█████████▍| 37/39 [00:09<00:00,  3.74it/s]
+
 97%|█████████▋| 38/39 [00:09<00:00,  3.80it/s]
+
100%|██████████| 39/39 [00:10<00:00,  3.82it/s]
                                                     
+
                                               
{'eval_loss': '3.526', 'eval_runtime': '10.39', 'eval_samples_per_second': '234.9', 'eval_steps_per_second': '3.754', 'epoch': '22.33'}
+
 89%|████████▉ | 6408/7175 [1:47:27<26:13,  2.05s/it]
+
100%|██████████| 39/39 [00:10<00:00,  3.82it/s][eval] epoch 22.329 step 6408 loss 3.5256 (best 3.5035 @ ep 19.821, 10/20 since best)
+
+
                                               
 89%|████████▉ | 6409/7175 [1:47:29<1:05:58,  5.17s/it]
 89%|████████▉ | 6410/7175 [1:47:31<54:04,  4.24s/it]  
                                                     
{'loss': '2.96', 'grad_norm': '0.2422', 'learning_rate': '0.0006', 'epoch': '22.34'}
+
 89%|████████▉ | 6410/7175 [1:47:31<54:04,  4.24s/it]
 89%|████████▉ | 6411/7175 [1:47:33<45:47,  3.60s/it]
 89%|████████▉ | 6412/7175 [1:47:35<39:53,  3.14s/it]
 89%|████████▉ | 6413/7175 [1:47:38<35:42,  2.81s/it]
 89%|████████▉ | 6414/7175 [1:47:40<32:45,  2.58s/it]
 89%|████████▉ | 6415/7175 [1:47:42<30:39,  2.42s/it]
 89%|████████▉ | 6416/7175 [1:47:44<29:08,  2.30s/it]
 89%|████████▉ | 6417/7175 [1:47:46<28:05,  2.22s/it]
 89%|████████▉ | 6418/7175 [1:47:48<27:21,  2.17s/it]
 89%|████████▉ | 6419/7175 [1:47:50<26:51,  2.13s/it]
 89%|████████▉ | 6420/7175 [1:47:52<26:26,  2.10s/it]
                                                     
{'loss': '2.978', 'grad_norm': '0.23', 'learning_rate': '0.0006', 'epoch': '22.37'}
+
 89%|████████▉ | 6420/7175 [1:47:52<26:26,  2.10s/it]
 89%|████████▉ | 6421/7175 [1:47:54<26:21,  2.10s/it]
 90%|████████▉ | 6422/7175 [1:47:56<26:13,  2.09s/it]
 90%|████████▉ | 6423/7175 [1:47:58<26:09,  2.09s/it]
 90%|████████▉ | 6424/7175 [1:48:00<26:02,  2.08s/it]
 90%|████████▉ | 6425/7175 [1:48:02<25:54,  2.07s/it]
 90%|████████▉ | 6426/7175 [1:48:04<25:43,  2.06s/it]
 90%|████████▉ | 6427/7175 [1:48:06<25:36,  2.05s/it]
 90%|████████▉ | 6428/7175 [1:48:08<25:31,  2.05s/it]
 90%|████████▉ | 6429/7175 [1:48:10<25:26,  2.05s/it]
 90%|████████▉ | 6430/7175 [1:48:12<25:22,  2.04s/it]
                                                     
{'loss': '2.985', 'grad_norm': '0.2132', 'learning_rate': '0.0006', 'epoch': '22.41'}
+
 90%|████████▉ | 6430/7175 [1:48:12<25:22,  2.04s/it]
 90%|████████▉ | 6431/7175 [1:48:14<25:24,  2.05s/it]
 90%|████████▉ | 6432/7175 [1:48:16<25:17,  2.04s/it]
 90%|████████▉ | 6433/7175 [1:48:19<25:18,  2.05s/it]
 90%|████████▉ | 6434/7175 [1:48:21<25:23,  2.06s/it]
 90%|████████▉ | 6435/7175 [1:48:23<25:24,  2.06s/it]
 90%|████████▉ | 6436/7175 [1:48:25<25:23,  2.06s/it]
 90%|████████▉ | 6437/7175 [1:48:27<25:22,  2.06s/it]
 90%|████████▉ | 6438/7175 [1:48:29<25:14,  2.05s/it]
 90%|████████▉ | 6439/7175 [1:48:31<25:10,  2.05s/it]
 90%|████████▉ | 6440/7175 [1:48:33<25:05,  2.05s/it]
                                                     
{'loss': '2.961', 'grad_norm': '0.2448', 'learning_rate': '0.0006', 'epoch': '22.44'}
+
 90%|████████▉ | 6440/7175 [1:48:33<25:05,  2.05s/it]
 90%|████████▉ | 6441/7175 [1:48:35<25:10,  2.06s/it]
 90%|████████▉ | 6442/7175 [1:48:37<25:03,  2.05s/it]
 90%|████████▉ | 6443/7175 [1:48:39<24:59,  2.05s/it]
 90%|████████▉ | 6444/7175 [1:48:41<24:55,  2.05s/it]
 90%|████████▉ | 6445/7175 [1:48:43<24:45,  2.04s/it]
 90%|████████▉ | 6446/7175 [1:48:45<23:29,  1.93s/it]
 90%|████████▉ | 6447/7175 [1:48:47<23:30,  1.94s/it]
 90%|████████▉ | 6448/7175 [1:48:49<23:53,  1.97s/it]
 90%|████████▉ | 6449/7175 [1:48:51<24:06,  1.99s/it]
 90%|████████▉ | 6450/7175 [1:48:53<24:10,  2.00s/it]
                                                     
{'loss': '2.979', 'grad_norm': '0.2572', 'learning_rate': '0.0006', 'epoch': '22.48'}
+
 90%|████████▉ | 6450/7175 [1:48:53<24:10,  2.00s/it]
 90%|████████▉ | 6451/7175 [1:48:55<24:07,  2.00s/it]
 90%|████████▉ | 6452/7175 [1:48:57<24:00,  1.99s/it]
 90%|████████▉ | 6453/7175 [1:48:59<24:07,  2.01s/it]
 90%|████████▉ | 6454/7175 [1:49:01<24:10,  2.01s/it]
 90%|████████▉ | 6455/7175 [1:49:03<24:14,  2.02s/it]
 90%|████████▉ | 6456/7175 [1:49:05<24:17,  2.03s/it]
 90%|████████▉ | 6457/7175 [1:49:07<24:22,  2.04s/it]
 90%|█████████ | 6458/7175 [1:49:09<24:26,  2.05s/it]
 90%|█████████ | 6459/7175 [1:49:11<24:30,  2.05s/it]
 90%|█████████ | 6460/7175 [1:49:13<24:29,  2.06s/it]
                                                     
{'loss': '2.992', 'grad_norm': '0.2222', 'learning_rate': '0.0006', 'epoch': '22.51'}
+
 90%|█████████ | 6460/7175 [1:49:13<24:29,  2.06s/it]
 90%|█████████ | 6461/7175 [1:49:15<24:28,  2.06s/it]
 90%|█████████ | 6462/7175 [1:49:17<24:22,  2.05s/it]
 90%|█████████ | 6463/7175 [1:49:19<24:15,  2.04s/it]
 90%|█████████ | 6464/7175 [1:49:21<24:15,  2.05s/it]
 90%|█████████ | 6465/7175 [1:49:23<24:11,  2.04s/it]
 90%|█████████ | 6466/7175 [1:49:25<24:06,  2.04s/it]
 90%|█████████ | 6467/7175 [1:49:28<24:03,  2.04s/it]
 90%|█████████ | 6468/7175 [1:49:30<24:02,  2.04s/it]
 90%|█████████ | 6469/7175 [1:49:32<24:07,  2.05s/it]
 90%|█████████ | 6470/7175 [1:49:34<24:10,  2.06s/it]
                                                     
{'loss': '2.984', 'grad_norm': '0.2275', 'learning_rate': '0.0006', 'epoch': '22.55'}
+
 90%|█████████ | 6470/7175 [1:49:34<24:10,  2.06s/it]
 90%|█████████ | 6471/7175 [1:49:36<24:18,  2.07s/it]
 90%|█████████ | 6472/7175 [1:49:38<24:14,  2.07s/it]
 90%|█████████ | 6473/7175 [1:49:40<24:08,  2.06s/it]
 90%|█████████ | 6474/7175 [1:49:42<24:02,  2.06s/it]
 90%|█████████ | 6475/7175 [1:49:44<23:56,  2.05s/it]
 90%|█████████ | 6476/7175 [1:49:46<23:51,  2.05s/it]
 90%|█████████ | 6477/7175 [1:49:48<23:45,  2.04s/it]
 90%|█████████ | 6478/7175 [1:49:50<23:40,  2.04s/it]
 90%|█████████ | 6479/7175 [1:49:52<23:35,  2.03s/it]
 90%|█████████ | 6480/7175 [1:49:54<23:33,  2.03s/it]
                                                     
{'loss': '2.981', 'grad_norm': '0.2171', 'learning_rate': '0.0006', 'epoch': '22.58'}
+
 90%|█████████ | 6480/7175 [1:49:54<23:33,  2.03s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:04,  7.68it/s]
+
  8%|▊         | 3/39 [00:00<00:06,  5.25it/s]
+
 10%|█         | 4/39 [00:00<00:07,  4.51it/s]
+
 13%|█▎        | 5/39 [00:01<00:08,  4.17it/s]
+
 15%|█▌        | 6/39 [00:01<00:08,  4.05it/s]
+
 18%|█▊        | 7/39 [00:01<00:08,  3.98it/s]
+
 21%|██        | 8/39 [00:01<00:07,  3.89it/s]
+
 23%|██▎       | 9/39 [00:02<00:07,  3.84it/s]
+
 26%|██▌       | 10/39 [00:02<00:07,  3.79it/s]
+
 28%|██▊       | 11/39 [00:02<00:07,  3.75it/s]
+
 31%|███       | 12/39 [00:02<00:07,  3.71it/s]
+
 33%|███▎      | 13/39 [00:03<00:06,  3.73it/s]
+
 36%|███▌      | 14/39 [00:03<00:06,  3.71it/s]
+
 38%|███▊      | 15/39 [00:03<00:06,  3.74it/s]
+
 41%|████      | 16/39 [00:04<00:06,  3.74it/s]
+
 44%|████▎     | 17/39 [00:04<00:05,  3.74it/s]
+
 46%|████▌     | 18/39 [00:04<00:05,  3.77it/s]
+
 49%|████▊     | 19/39 [00:04<00:05,  3.77it/s]
+
 51%|█████▏    | 20/39 [00:05<00:05,  3.80it/s]
+
 54%|█████▍    | 21/39 [00:05<00:04,  3.77it/s]
+
 56%|█████▋    | 22/39 [00:05<00:04,  3.77it/s]
+
 59%|█████▉    | 23/39 [00:05<00:04,  3.75it/s]
+
 62%|██████▏   | 24/39 [00:06<00:03,  3.77it/s]
+
 64%|██████▍   | 25/39 [00:06<00:03,  3.73it/s]
+
 67%|██████▋   | 26/39 [00:06<00:03,  3.72it/s]
+
 69%|██████▉   | 27/39 [00:06<00:03,  3.72it/s]
+
 72%|███████▏  | 28/39 [00:07<00:02,  3.71it/s]
+
 74%|███████▍  | 29/39 [00:07<00:02,  3.70it/s]
+
 77%|███████▋  | 30/39 [00:07<00:02,  3.71it/s]
+
 79%|███████▉  | 31/39 [00:08<00:02,  3.72it/s]
+
 82%|████████▏ | 32/39 [00:08<00:01,  3.72it/s]
+
 85%|████████▍ | 33/39 [00:08<00:01,  3.73it/s]
+
 87%|████████▋ | 34/39 [00:08<00:01,  3.71it/s]
+
 90%|████████▉ | 35/39 [00:09<00:01,  3.72it/s]
+
 92%|█████████▏| 36/39 [00:09<00:00,  3.69it/s]
+
 95%|█████████▍| 37/39 [00:09<00:00,  3.74it/s]
+
 97%|█████████▋| 38/39 [00:09<00:00,  3.71it/s]
+
100%|██████████| 39/39 [00:10<00:00,  3.69it/s]
                                                     
+
                                               
{'eval_loss': '3.516', 'eval_runtime': '10.52', 'eval_samples_per_second': '232', 'eval_steps_per_second': '3.707', 'epoch': '22.58'}
+
 90%|█████████ | 6480/7175 [1:50:05<23:33,  2.03s/it]
+
100%|██████████| 39/39 [00:10<00:00,  3.69it/s][eval] epoch 22.58 step 6480 loss 3.5157 (best 3.5035 @ ep 19.821, 11/20 since best)
+
+
                                               
 90%|█████████ | 6481/7175 [1:50:07<1:00:08,  5.20s/it]
 90%|█████████ | 6482/7175 [1:50:09<49:04,  4.25s/it]  
 90%|█████████ | 6483/7175 [1:50:11<41:20,  3.58s/it]
 90%|█████████ | 6484/7175 [1:50:13<35:56,  3.12s/it]
 90%|█████████ | 6485/7175 [1:50:15<32:11,  2.80s/it]
 90%|█████████ | 6486/7175 [1:50:17<29:30,  2.57s/it]
 90%|█████████ | 6487/7175 [1:50:19<27:39,  2.41s/it]
 90%|█████████ | 6488/7175 [1:50:21<26:24,  2.31s/it]
 90%|█████████ | 6489/7175 [1:50:23<25:35,  2.24s/it]
 90%|█████████ | 6490/7175 [1:50:25<25:00,  2.19s/it]
                                                     
{'loss': '2.995', 'grad_norm': '0.213', 'learning_rate': '0.0006', 'epoch': '22.62'}
+
 90%|█████████ | 6490/7175 [1:50:25<25:00,  2.19s/it]
 90%|█████████ | 6491/7175 [1:50:27<24:38,  2.16s/it]
 90%|█████████ | 6492/7175 [1:50:29<24:10,  2.12s/it]
 90%|█████████ | 6493/7175 [1:50:31<23:51,  2.10s/it]
 91%|█████████ | 6494/7175 [1:50:33<23:37,  2.08s/it]
 91%|█████████ | 6495/7175 [1:50:35<23:25,  2.07s/it]
 91%|█████████ | 6496/7175 [1:50:37<23:18,  2.06s/it]
 91%|█████████ | 6497/7175 [1:50:40<23:14,  2.06s/it]
 91%|█████████ | 6498/7175 [1:50:42<23:07,  2.05s/it]
 91%|█████████ | 6499/7175 [1:50:44<23:03,  2.05s/it]
 91%|█████████ | 6500/7175 [1:50:46<23:04,  2.05s/it]
                                                     
{'loss': '2.987', 'grad_norm': '0.2337', 'learning_rate': '0.0006', 'epoch': '22.65'}
+
 91%|█████████ | 6500/7175 [1:50:46<23:04,  2.05s/it]
 91%|█████████ | 6501/7175 [1:50:48<23:13,  2.07s/it]
 91%|█████████ | 6502/7175 [1:50:50<23:09,  2.06s/it]
 91%|█████████ | 6503/7175 [1:50:52<23:06,  2.06s/it]
 91%|█████████ | 6504/7175 [1:50:54<23:00,  2.06s/it]
 91%|█████████ | 6505/7175 [1:50:56<22:54,  2.05s/it]
 91%|█████████ | 6506/7175 [1:50:58<22:47,  2.04s/it]
 91%|█████████ | 6507/7175 [1:51:00<22:45,  2.04s/it]
 91%|█████████ | 6508/7175 [1:51:02<22:43,  2.04s/it]
 91%|█████████ | 6509/7175 [1:51:04<22:39,  2.04s/it]
 91%|█████████ | 6510/7175 [1:51:06<22:39,  2.04s/it]
                                                     
{'loss': '2.999', 'grad_norm': '0.2521', 'learning_rate': '0.0006', 'epoch': '22.69'}
+
 91%|█████████ | 6510/7175 [1:51:06<22:39,  2.04s/it]
 91%|█████████ | 6511/7175 [1:51:08<22:43,  2.05s/it]
 91%|█████████ | 6512/7175 [1:51:10<22:46,  2.06s/it]
 91%|█████████ | 6513/7175 [1:51:12<22:46,  2.06s/it]
 91%|█████████ | 6514/7175 [1:51:14<22:43,  2.06s/it]
 91%|█████████ | 6515/7175 [1:51:17<22:42,  2.06s/it]
 91%|█████████ | 6516/7175 [1:51:19<22:35,  2.06s/it]
 91%|█████████ | 6517/7175 [1:51:21<22:28,  2.05s/it]
 91%|█████████ | 6518/7175 [1:51:23<22:23,  2.05s/it]
 91%|█████████ | 6519/7175 [1:51:25<22:20,  2.04s/it]
 91%|█████████ | 6520/7175 [1:51:27<22:18,  2.04s/it]
                                                     
{'loss': '2.995', 'grad_norm': '0.2359', 'learning_rate': '0.0006', 'epoch': '22.72'}
+
 91%|█████████ | 6520/7175 [1:51:27<22:18,  2.04s/it]
 91%|█████████ | 6521/7175 [1:51:29<22:19,  2.05s/it]
 91%|█████████ | 6522/7175 [1:51:31<22:16,  2.05s/it]
 91%|█████████ | 6523/7175 [1:51:33<22:13,  2.04s/it]
 91%|█████████ | 6524/7175 [1:51:35<22:14,  2.05s/it]
 91%|█████████ | 6525/7175 [1:51:37<22:15,  2.05s/it]
 91%|█████████ | 6526/7175 [1:51:39<22:15,  2.06s/it]
 91%|█████████ | 6527/7175 [1:51:41<22:16,  2.06s/it]
 91%|█████████ | 6528/7175 [1:51:43<22:08,  2.05s/it]
 91%|█████████ | 6529/7175 [1:51:45<22:00,  2.04s/it]
 91%|█████████ | 6530/7175 [1:51:47<21:57,  2.04s/it]
                                                     
{'loss': '2.989', 'grad_norm': '0.2662', 'learning_rate': '0.0006', 'epoch': '22.75'}
+
 91%|█████████ | 6530/7175 [1:51:47<21:57,  2.04s/it]
 91%|█████████ | 6531/7175 [1:51:49<22:01,  2.05s/it]
 91%|█████████ | 6532/7175 [1:51:51<21:55,  2.05s/it]
 91%|█████████ | 6533/7175 [1:51:53<21:52,  2.05s/it]
 91%|█████████ | 6534/7175 [1:51:55<21:49,  2.04s/it]
 91%|█████████ | 6535/7175 [1:51:57<21:48,  2.04s/it]
 91%|█████████ | 6536/7175 [1:52:00<21:51,  2.05s/it]
 91%|█████████ | 6537/7175 [1:52:02<21:54,  2.06s/it]
 91%|█████████ | 6538/7175 [1:52:04<21:53,  2.06s/it]
 91%|█████████ | 6539/7175 [1:52:06<21:53,  2.07s/it]
 91%|█████████ | 6540/7175 [1:52:08<21:46,  2.06s/it]
                                                     
{'loss': '2.994', 'grad_norm': '0.2218', 'learning_rate': '0.0006', 'epoch': '22.79'}
+
 91%|█████████ | 6540/7175 [1:52:08<21:46,  2.06s/it]
 91%|█████████ | 6541/7175 [1:52:10<21:47,  2.06s/it]
 91%|█████████ | 6542/7175 [1:52:12<21:42,  2.06s/it]
 91%|█████████ | 6543/7175 [1:52:14<21:37,  2.05s/it]
 91%|█████████ | 6544/7175 [1:52:16<21:34,  2.05s/it]
 91%|█████████ | 6545/7175 [1:52:18<21:27,  2.04s/it]
 91%|█████████ | 6546/7175 [1:52:20<21:24,  2.04s/it]
 91%|█████████ | 6547/7175 [1:52:22<21:22,  2.04s/it]
 91%|█████████▏| 6548/7175 [1:52:24<21:24,  2.05s/it]
 91%|█████████▏| 6549/7175 [1:52:26<21:25,  2.05s/it]
 91%|█████████▏| 6550/7175 [1:52:28<21:26,  2.06s/it]
                                                     
{'loss': '2.978', 'grad_norm': '0.214', 'learning_rate': '0.0006', 'epoch': '22.82'}
+
 91%|█████████▏| 6550/7175 [1:52:28<21:26,  2.06s/it]
 91%|█████████▏| 6551/7175 [1:52:30<21:31,  2.07s/it]
 91%|█████████▏| 6552/7175 [1:52:32<21:23,  2.06s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:04,  7.50it/s]
+
  8%|▊         | 3/39 [00:00<00:06,  5.37it/s]
+
 10%|█         | 4/39 [00:00<00:07,  4.65it/s]
+
 13%|█▎        | 5/39 [00:01<00:07,  4.26it/s]
+
 15%|█▌        | 6/39 [00:01<00:08,  4.05it/s]
+
 18%|█▊        | 7/39 [00:01<00:08,  3.94it/s]
+
 21%|██        | 8/39 [00:01<00:08,  3.87it/s]
+
 23%|██▎       | 9/39 [00:02<00:07,  3.81it/s]
+
 26%|██▌       | 10/39 [00:02<00:07,  3.82it/s]
+
 28%|██▊       | 11/39 [00:02<00:07,  3.80it/s]
+
 31%|███       | 12/39 [00:02<00:07,  3.76it/s]
+
 33%|███▎      | 13/39 [00:03<00:06,  3.73it/s]
+
 36%|███▌      | 14/39 [00:03<00:06,  3.75it/s]
+
 38%|███▊      | 15/39 [00:03<00:06,  3.78it/s]
+
 41%|████      | 16/39 [00:04<00:06,  3.77it/s]
+
 44%|████▎     | 17/39 [00:04<00:05,  3.72it/s]
+
 46%|████▌     | 18/39 [00:04<00:05,  3.70it/s]
+
 49%|████▊     | 19/39 [00:04<00:05,  3.76it/s]
+
 51%|█████▏    | 20/39 [00:05<00:05,  3.75it/s]
+
 54%|█████▍    | 21/39 [00:05<00:04,  3.73it/s]
+
 56%|█████▋    | 22/39 [00:05<00:04,  3.75it/s]
+
 59%|█████▉    | 23/39 [00:05<00:04,  3.75it/s]
+
 62%|██████▏   | 24/39 [00:06<00:03,  3.76it/s]
+
 64%|██████▍   | 25/39 [00:06<00:03,  3.74it/s]
+
 67%|██████▋   | 26/39 [00:06<00:03,  3.72it/s]
+
 69%|██████▉   | 27/39 [00:06<00:03,  3.69it/s]
+
 72%|███████▏  | 28/39 [00:07<00:02,  3.72it/s]
+
 74%|███████▍  | 29/39 [00:07<00:02,  3.67it/s]
+
 77%|███████▋  | 30/39 [00:07<00:02,  3.74it/s]
+
 79%|███████▉  | 31/39 [00:08<00:02,  3.73it/s]
+
 82%|████████▏ | 32/39 [00:08<00:01,  3.72it/s]
+
 85%|████████▍ | 33/39 [00:08<00:01,  3.74it/s]
+
 87%|████████▋ | 34/39 [00:08<00:01,  3.76it/s]
+
 90%|████████▉ | 35/39 [00:09<00:01,  3.75it/s]
+
 92%|█████████▏| 36/39 [00:09<00:00,  3.77it/s]
+
 95%|█████████▍| 37/39 [00:09<00:00,  3.82it/s]
+
 97%|█████████▋| 38/39 [00:09<00:00,  3.77it/s]
+
100%|██████████| 39/39 [00:10<00:00,  3.79it/s]
                                                     
+
                                               
{'eval_loss': '3.504', 'eval_runtime': '10.49', 'eval_samples_per_second': '232.7', 'eval_steps_per_second': '3.718', 'epoch': '22.83'}
+
 91%|█████████▏| 6552/7175 [1:52:43<21:23,  2.06s/it]
+
100%|██████████| 39/39 [00:10<00:00,  3.79it/s][eval] epoch 22.832 step 6552 loss 3.5042 (best 3.5035 @ ep 19.821, 12/20 since best)
+
+
                                               
 91%|█████████▏| 6553/7175 [1:52:45<53:52,  5.20s/it]
 91%|█████████▏| 6554/7175 [1:52:47<44:02,  4.25s/it]
 91%|█████████▏| 6555/7175 [1:52:49<37:08,  3.59s/it]
 91%|█████████▏| 6556/7175 [1:52:51<32:21,  3.14s/it]
 91%|█████████▏| 6557/7175 [1:52:53<29:01,  2.82s/it]
 91%|█████████▏| 6558/7175 [1:52:55<26:39,  2.59s/it]
 91%|█████████▏| 6559/7175 [1:52:57<24:54,  2.43s/it]
 91%|█████████▏| 6560/7175 [1:52:59<23:40,  2.31s/it]
                                                     
{'loss': '3.002', 'grad_norm': '0.2071', 'learning_rate': '0.0006', 'epoch': '22.86'}
+
 91%|█████████▏| 6560/7175 [1:52:59<23:40,  2.31s/it]
 91%|█████████▏| 6561/7175 [1:53:01<22:54,  2.24s/it]
 91%|█████████▏| 6562/7175 [1:53:03<22:16,  2.18s/it]
 91%|█████████▏| 6563/7175 [1:53:05<21:47,  2.14s/it]
 91%|█████████▏| 6564/7175 [1:53:08<21:25,  2.10s/it]
 91%|█████████▏| 6565/7175 [1:53:10<21:10,  2.08s/it]
 92%|█████████▏| 6566/7175 [1:53:12<21:02,  2.07s/it]
 92%|█████████▏| 6567/7175 [1:53:14<20:59,  2.07s/it]
 92%|█████████▏| 6568/7175 [1:53:16<20:57,  2.07s/it]
 92%|█████████▏| 6569/7175 [1:53:18<20:53,  2.07s/it]
 92%|█████████▏| 6570/7175 [1:53:20<20:51,  2.07s/it]
                                                     
{'loss': '3.011', 'grad_norm': '0.1994', 'learning_rate': '0.0006', 'epoch': '22.89'}
+
 92%|█████████▏| 6570/7175 [1:53:20<20:51,  2.07s/it]
 92%|█████████▏| 6571/7175 [1:53:22<20:52,  2.07s/it]
 92%|█████████▏| 6572/7175 [1:53:24<20:46,  2.07s/it]
 92%|█████████▏| 6573/7175 [1:53:26<20:37,  2.06s/it]
 92%|█████████▏| 6574/7175 [1:53:28<20:31,  2.05s/it]
 92%|█████████▏| 6575/7175 [1:53:30<20:24,  2.04s/it]
 92%|█████████▏| 6576/7175 [1:53:32<21:21,  2.14s/it]
 92%|█████████▏| 6577/7175 [1:53:34<21:00,  2.11s/it]
 92%|█████████▏| 6578/7175 [1:53:37<20:49,  2.09s/it]
 92%|█████████▏| 6579/7175 [1:53:39<20:43,  2.09s/it]
 92%|█████████▏| 6580/7175 [1:53:41<20:38,  2.08s/it]
                                                     
{'loss': '3.004', 'grad_norm': '0.215', 'learning_rate': '0.0006', 'epoch': '22.93'}
+
 92%|█████████▏| 6580/7175 [1:53:41<20:38,  2.08s/it]
 92%|█████████▏| 6581/7175 [1:53:43<20:38,  2.08s/it]
 92%|█████████▏| 6582/7175 [1:53:45<20:30,  2.08s/it]
 92%|█████████▏| 6583/7175 [1:53:47<20:23,  2.07s/it]
 92%|█████████▏| 6584/7175 [1:53:49<20:14,  2.05s/it]
 92%|█████████▏| 6585/7175 [1:53:51<20:08,  2.05s/it]
 92%|█████████▏| 6586/7175 [1:53:53<20:06,  2.05s/it]
 92%|█████████▏| 6587/7175 [1:53:55<20:02,  2.05s/it]
 92%|█████████▏| 6588/7175 [1:53:57<19:57,  2.04s/it]
 92%|█████████▏| 6589/7175 [1:53:59<19:53,  2.04s/it]
 92%|█████████▏| 6590/7175 [1:54:01<19:43,  2.02s/it]
                                                     
{'loss': '3.003', 'grad_norm': '0.2108', 'learning_rate': '0.0006', 'epoch': '22.96'}
+
 92%|█████████▏| 6590/7175 [1:54:01<19:43,  2.02s/it]
 92%|█████████▏| 6591/7175 [1:54:03<19:54,  2.04s/it]
 92%|█████████▏| 6592/7175 [1:54:05<19:56,  2.05s/it]
 92%|█████████▏| 6593/7175 [1:54:07<19:57,  2.06s/it]
 92%|█████████▏| 6594/7175 [1:54:09<19:55,  2.06s/it]
 92%|█████████▏| 6595/7175 [1:54:11<19:50,  2.05s/it]
 92%|█████████▏| 6596/7175 [1:54:13<19:47,  2.05s/it]
 92%|█████████▏| 6597/7175 [1:54:16<19:45,  2.05s/it]
 92%|█████████▏| 6598/7175 [1:54:18<19:43,  2.05s/it]
 92%|█████████▏| 6599/7175 [1:54:20<19:39,  2.05s/it]
 92%|█████████▏| 6600/7175 [1:54:22<19:34,  2.04s/it]
                                                     
{'loss': '3', 'grad_norm': '0.253', 'learning_rate': '0.0006', 'epoch': '23'}
+
 92%|█████████▏| 6600/7175 [1:54:22<19:34,  2.04s/it]
 92%|█████████▏| 6601/7175 [1:54:22<14:38,  1.53s/it]
 92%|█████████▏| 6602/7175 [1:54:24<16:16,  1.70s/it]
 92%|█████████▏| 6603/7175 [1:54:26<17:15,  1.81s/it]
 92%|█████████▏| 6604/7175 [1:54:28<17:57,  1.89s/it]
 92%|█████████▏| 6605/7175 [1:54:30<18:26,  1.94s/it]
 92%|█████████▏| 6606/7175 [1:54:32<18:45,  1.98s/it]
 92%|█████████▏| 6607/7175 [1:54:34<19:01,  2.01s/it]
 92%|█████████▏| 6608/7175 [1:54:36<17:55,  1.90s/it]
 92%|█████████▏| 6609/7175 [1:54:38<17:58,  1.91s/it]
 92%|█████████▏| 6610/7175 [1:54:40<18:14,  1.94s/it]
                                                     
{'loss': '2.956', 'grad_norm': '0.3441', 'learning_rate': '0.0006', 'epoch': '23.03'}
+
 92%|█████████▏| 6610/7175 [1:54:40<18:14,  1.94s/it]
 92%|█████████▏| 6611/7175 [1:54:42<18:29,  1.97s/it]
 92%|█████████▏| 6612/7175 [1:54:44<18:37,  1.98s/it]
 92%|█████████▏| 6613/7175 [1:54:46<18:36,  1.99s/it]
 92%|█████████▏| 6614/7175 [1:54:48<18:29,  1.98s/it]
 92%|█████████▏| 6615/7175 [1:54:50<18:44,  2.01s/it]
 92%|█████████▏| 6616/7175 [1:54:52<18:55,  2.03s/it]
 92%|█████████▏| 6617/7175 [1:54:54<19:00,  2.04s/it]
 92%|█████████▏| 6618/7175 [1:54:56<19:04,  2.05s/it]
 92%|█████████▏| 6619/7175 [1:54:58<18:59,  2.05s/it]
 92%|█████████▏| 6620/7175 [1:55:00<18:55,  2.05s/it]
                                                     
{'loss': '2.939', 'grad_norm': '0.2336', 'learning_rate': '0.0006', 'epoch': '23.07'}
+
 92%|█████████▏| 6620/7175 [1:55:00<18:55,  2.05s/it]
 92%|█████████▏| 6621/7175 [1:55:02<19:01,  2.06s/it]
 92%|█████████▏| 6622/7175 [1:55:05<18:56,  2.05s/it]
 92%|█████████▏| 6623/7175 [1:55:07<18:52,  2.05s/it]
 92%|█████████▏| 6624/7175 [1:55:09<18:48,  2.05s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:04,  8.23it/s]
+
  8%|▊         | 3/39 [00:00<00:06,  5.49it/s]
+
 10%|█         | 4/39 [00:00<00:07,  4.70it/s]
+
 13%|█▎        | 5/39 [00:01<00:07,  4.29it/s]
+
 15%|█▌        | 6/39 [00:01<00:07,  4.13it/s]
+
 18%|█▊        | 7/39 [00:01<00:08,  3.99it/s]
+
 21%|██        | 8/39 [00:01<00:07,  3.94it/s]
+
 23%|██▎       | 9/39 [00:02<00:07,  3.93it/s]
+
 26%|██▌       | 10/39 [00:02<00:07,  3.86it/s]
+
 28%|██▊       | 11/39 [00:02<00:07,  3.83it/s]
+
 31%|███       | 12/39 [00:02<00:07,  3.84it/s]
+
 33%|███▎      | 13/39 [00:03<00:06,  3.80it/s]
+
 36%|███▌      | 14/39 [00:03<00:06,  3.79it/s]
+
 38%|███▊      | 15/39 [00:03<00:06,  3.85it/s]
+
 41%|████      | 16/39 [00:03<00:06,  3.79it/s]
+
 44%|████▎     | 17/39 [00:04<00:05,  3.76it/s]
+
 46%|████▌     | 18/39 [00:04<00:05,  3.74it/s]
+
 49%|████▊     | 19/39 [00:04<00:05,  3.72it/s]
+
 51%|█████▏    | 20/39 [00:05<00:05,  3.76it/s]
+
 54%|█████▍    | 21/39 [00:05<00:04,  3.80it/s]
+
 56%|█████▋    | 22/39 [00:05<00:04,  3.76it/s]
+
 59%|█████▉    | 23/39 [00:05<00:04,  3.78it/s]
+
 62%|██████▏   | 24/39 [00:06<00:04,  3.74it/s]
+
 64%|██████▍   | 25/39 [00:06<00:03,  3.73it/s]
+
 67%|██████▋   | 26/39 [00:06<00:03,  3.75it/s]
+
 69%|██████▉   | 27/39 [00:06<00:03,  3.75it/s]
+
 72%|███████▏  | 28/39 [00:07<00:02,  3.77it/s]
+
 74%|███████▍  | 29/39 [00:07<00:02,  3.73it/s]
+
 77%|███████▋  | 30/39 [00:07<00:02,  3.76it/s]
+
 79%|███████▉  | 31/39 [00:07<00:02,  3.73it/s]
+
 82%|████████▏ | 32/39 [00:08<00:01,  3.71it/s]
+
 85%|████████▍ | 33/39 [00:08<00:01,  3.72it/s]
+
 87%|████████▋ | 34/39 [00:08<00:01,  3.77it/s]
+
 90%|████████▉ | 35/39 [00:09<00:01,  3.77it/s]
+
 92%|█████████▏| 36/39 [00:09<00:00,  3.75it/s]
+
 95%|█████████▍| 37/39 [00:09<00:00,  3.72it/s]
+
 97%|█████████▋| 38/39 [00:09<00:00,  3.73it/s]
+
100%|██████████| 39/39 [00:10<00:00,  3.74it/s]
                                                     
+
                                               
{'eval_loss': '3.533', 'eval_runtime': '10.43', 'eval_samples_per_second': '233.9', 'eval_steps_per_second': '3.738', 'epoch': '23.08'}
+
 92%|█████████▏| 6624/7175 [1:55:19<18:48,  2.05s/it]
+
100%|██████████| 39/39 [00:10<00:00,  3.74it/s][eval] epoch 23.08 step 6624 loss 3.5328 (best 3.5035 @ ep 19.821, 13/20 since best)
+
+
                                               
 92%|█████████▏| 6625/7175 [1:55:21<47:28,  5.18s/it]
 92%|█████████▏| 6626/7175 [1:55:23<38:47,  4.24s/it]
 92%|█████████▏| 6627/7175 [1:55:25<32:42,  3.58s/it]
 92%|█████████▏| 6628/7175 [1:55:27<28:26,  3.12s/it]
 92%|█████████▏| 6629/7175 [1:55:29<25:25,  2.79s/it]
 92%|█████████▏| 6630/7175 [1:55:31<23:18,  2.57s/it]
                                                     
{'loss': '2.933', 'grad_norm': '0.2245', 'learning_rate': '0.0006', 'epoch': '23.1'}
+
 92%|█████████▏| 6630/7175 [1:55:31<23:18,  2.57s/it]
 92%|█████████▏| 6631/7175 [1:55:33<21:51,  2.41s/it]
 92%|█████████▏| 6632/7175 [1:55:35<20:49,  2.30s/it]
 92%|█████████▏| 6633/7175 [1:55:37<19:54,  2.20s/it]
 92%|█████████▏| 6634/7175 [1:55:39<19:17,  2.14s/it]
 92%|█████████▏| 6635/7175 [1:55:41<19:06,  2.12s/it]
 92%|█████████▏| 6636/7175 [1:55:44<18:54,  2.10s/it]
 93%|█████████▎| 6637/7175 [1:55:46<18:46,  2.09s/it]
 93%|█████████▎| 6638/7175 [1:55:48<18:37,  2.08s/it]
 93%|█████████▎| 6639/7175 [1:55:50<18:28,  2.07s/it]
 93%|█████████▎| 6640/7175 [1:55:52<18:21,  2.06s/it]
                                                     
{'loss': '2.934', 'grad_norm': '0.223', 'learning_rate': '0.0006', 'epoch': '23.14'}
+
 93%|█████████▎| 6640/7175 [1:55:52<18:21,  2.06s/it]
 93%|█████████▎| 6641/7175 [1:55:54<18:22,  2.07s/it]
 93%|█████████▎| 6642/7175 [1:55:56<18:16,  2.06s/it]
 93%|█████████▎| 6643/7175 [1:55:58<18:12,  2.05s/it]
 93%|█████████▎| 6644/7175 [1:56:00<18:10,  2.05s/it]
 93%|█████████▎| 6645/7175 [1:56:02<18:04,  2.05s/it]
 93%|█████████▎| 6646/7175 [1:56:04<17:53,  2.03s/it]
 93%|█████████▎| 6647/7175 [1:56:06<17:57,  2.04s/it]
 93%|█████████▎| 6648/7175 [1:56:08<18:00,  2.05s/it]
 93%|█████████▎| 6649/7175 [1:56:10<18:01,  2.06s/it]
 93%|█████████▎| 6650/7175 [1:56:12<18:01,  2.06s/it]
                                                     
{'loss': '2.928', 'grad_norm': '0.2219', 'learning_rate': '0.0006', 'epoch': '23.17'}
+
 93%|█████████▎| 6650/7175 [1:56:12<18:01,  2.06s/it]
 93%|█████████▎| 6651/7175 [1:56:14<18:02,  2.07s/it]
 93%|█████████▎| 6652/7175 [1:56:16<17:57,  2.06s/it]
 93%|█████████▎| 6653/7175 [1:56:18<17:51,  2.05s/it]
 93%|█████████▎| 6654/7175 [1:56:20<17:48,  2.05s/it]
 93%|█████████▎| 6655/7175 [1:56:22<17:46,  2.05s/it]
 93%|█████████▎| 6656/7175 [1:56:25<17:42,  2.05s/it]
 93%|█████████▎| 6657/7175 [1:56:27<17:38,  2.04s/it]
 93%|█████████▎| 6658/7175 [1:56:29<17:26,  2.02s/it]
 93%|█████████▎| 6659/7175 [1:56:31<17:33,  2.04s/it]
 93%|█████████▎| 6660/7175 [1:56:33<17:37,  2.05s/it]
                                                     
{'loss': '2.941', 'grad_norm': '0.2382', 'learning_rate': '0.0006', 'epoch': '23.21'}
+
 93%|█████████▎| 6660/7175 [1:56:33<17:37,  2.05s/it]
 93%|█████████▎| 6661/7175 [1:56:35<17:44,  2.07s/it]
 93%|█████████▎| 6662/7175 [1:56:37<17:44,  2.08s/it]
 93%|█████████▎| 6663/7175 [1:56:39<17:36,  2.06s/it]
 93%|█████████▎| 6664/7175 [1:56:41<17:31,  2.06s/it]
 93%|█████████▎| 6665/7175 [1:56:43<17:28,  2.06s/it]
 93%|█████████▎| 6666/7175 [1:56:45<17:25,  2.05s/it]
 93%|█████████▎| 6667/7175 [1:56:47<17:23,  2.05s/it]
 93%|█████████▎| 6668/7175 [1:56:49<17:18,  2.05s/it]
 93%|█████████▎| 6669/7175 [1:56:51<17:13,  2.04s/it]
 93%|█████████▎| 6670/7175 [1:56:53<17:11,  2.04s/it]
                                                     
{'loss': '2.95', 'grad_norm': '0.2237', 'learning_rate': '0.0006', 'epoch': '23.24'}
+
 93%|█████████▎| 6670/7175 [1:56:53<17:11,  2.04s/it]
 93%|█████████▎| 6671/7175 [1:56:55<17:18,  2.06s/it]
 93%|█████████▎| 6672/7175 [1:56:57<17:17,  2.06s/it]
 93%|█████████▎| 6673/7175 [1:56:59<17:17,  2.07s/it]
 93%|█████████▎| 6674/7175 [1:57:02<17:18,  2.07s/it]
 93%|█████████▎| 6675/7175 [1:57:04<17:13,  2.07s/it]
 93%|█████████▎| 6676/7175 [1:57:06<17:08,  2.06s/it]
 93%|█████████▎| 6677/7175 [1:57:08<17:04,  2.06s/it]
 93%|█████████▎| 6678/7175 [1:57:10<17:00,  2.05s/it]
 93%|█████████▎| 6679/7175 [1:57:12<16:57,  2.05s/it]
 93%|█████████▎| 6680/7175 [1:57:14<16:54,  2.05s/it]
                                                     
{'loss': '2.947', 'grad_norm': '0.2273', 'learning_rate': '0.0006', 'epoch': '23.28'}
+
 93%|█████████▎| 6680/7175 [1:57:14<16:54,  2.05s/it]
 93%|█████████▎| 6681/7175 [1:57:16<16:57,  2.06s/it]
 93%|█████████▎| 6682/7175 [1:57:18<16:46,  2.04s/it]
 93%|█████████▎| 6683/7175 [1:57:20<16:50,  2.05s/it]
 93%|█████████▎| 6684/7175 [1:57:22<16:52,  2.06s/it]
 93%|█████████▎| 6685/7175 [1:57:24<16:54,  2.07s/it]
 93%|█████████▎| 6686/7175 [1:57:26<16:53,  2.07s/it]
 93%|█████████▎| 6687/7175 [1:57:28<16:47,  2.06s/it]
 93%|█████████▎| 6688/7175 [1:57:30<16:43,  2.06s/it]
 93%|█████████▎| 6689/7175 [1:57:32<16:39,  2.06s/it]
 93%|█████████▎| 6690/7175 [1:57:34<16:35,  2.05s/it]
                                                     
{'loss': '2.948', 'grad_norm': '0.2131', 'learning_rate': '0.0006', 'epoch': '23.31'}
+
 93%|█████████▎| 6690/7175 [1:57:34<16:35,  2.05s/it]
 93%|█████████▎| 6691/7175 [1:57:37<16:36,  2.06s/it]
 93%|█████████▎| 6692/7175 [1:57:39<16:32,  2.06s/it]
 93%|█████████▎| 6693/7175 [1:57:41<16:29,  2.05s/it]
 93%|█████████▎| 6694/7175 [1:57:43<16:18,  2.03s/it]
 93%|█████████▎| 6695/7175 [1:57:45<16:20,  2.04s/it]
 93%|█████████▎| 6696/7175 [1:57:47<16:21,  2.05s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:04,  7.76it/s]
+
  8%|▊         | 3/39 [00:00<00:06,  5.33it/s]
+
 10%|█         | 4/39 [00:00<00:07,  4.56it/s]
+
 13%|█▎        | 5/39 [00:01<00:08,  4.22it/s]
+
 15%|█▌        | 6/39 [00:01<00:08,  4.07it/s]
+
 18%|█▊        | 7/39 [00:01<00:08,  3.94it/s]
+
 21%|██        | 8/39 [00:01<00:08,  3.85it/s]
+
 23%|██▎       | 9/39 [00:02<00:07,  3.80it/s]
+
 26%|██▌       | 10/39 [00:02<00:07,  3.79it/s]
+
 28%|██▊       | 11/39 [00:02<00:07,  3.77it/s]
+
 31%|███       | 12/39 [00:02<00:07,  3.79it/s]
+
 33%|███▎      | 13/39 [00:03<00:06,  3.76it/s]
+
 36%|███▌      | 14/39 [00:03<00:06,  3.76it/s]
+
 38%|███▊      | 15/39 [00:03<00:06,  3.77it/s]
+
 41%|████      | 16/39 [00:04<00:06,  3.69it/s]
+
 44%|████▎     | 17/39 [00:04<00:05,  3.70it/s]
+
 46%|████▌     | 18/39 [00:04<00:05,  3.70it/s]
+
 49%|████▊     | 19/39 [00:04<00:05,  3.71it/s]
+
 51%|█████▏    | 20/39 [00:05<00:05,  3.77it/s]
+
 54%|█████▍    | 21/39 [00:05<00:04,  3.79it/s]
+
 56%|█████▋    | 22/39 [00:05<00:04,  3.78it/s]
+
 59%|█████▉    | 23/39 [00:05<00:04,  3.78it/s]
+
 62%|██████▏   | 24/39 [00:06<00:04,  3.74it/s]
+
 64%|██████▍   | 25/39 [00:06<00:03,  3.76it/s]
+
 67%|██████▋   | 26/39 [00:06<00:03,  3.80it/s]
+
 69%|██████▉   | 27/39 [00:06<00:03,  3.78it/s]
+
 72%|███████▏  | 28/39 [00:07<00:02,  3.77it/s]
+
 74%|███████▍  | 29/39 [00:07<00:02,  3.79it/s]
+
 77%|███████▋  | 30/39 [00:07<00:02,  3.80it/s]
+
 79%|███████▉  | 31/39 [00:07<00:02,  3.78it/s]
+
 82%|████████▏ | 32/39 [00:08<00:01,  3.75it/s]
+
 85%|████████▍ | 33/39 [00:08<00:01,  3.77it/s]
+
 87%|████████▋ | 34/39 [00:08<00:01,  3.74it/s]
+
 90%|████████▉ | 35/39 [00:09<00:01,  3.79it/s]
+
 92%|█████████▏| 36/39 [00:09<00:00,  3.79it/s]
+
 95%|█████████▍| 37/39 [00:09<00:00,  3.78it/s]
+
 97%|█████████▋| 38/39 [00:09<00:00,  3.80it/s]
+
100%|██████████| 39/39 [00:10<00:00,  3.78it/s]
                                                     
+
                                               
{'eval_loss': '3.528', 'eval_runtime': '10.44', 'eval_samples_per_second': '233.8', 'eval_steps_per_second': '3.735', 'epoch': '23.33'}
+
 93%|█████████▎| 6696/7175 [1:57:57<16:21,  2.05s/it]
+
100%|██████████| 39/39 [00:10<00:00,  3.78it/s][eval] epoch 23.332 step 6696 loss 3.5278 (best 3.5035 @ ep 19.821, 14/20 since best)
+
+
                                               
 93%|█████████▎| 6697/7175 [1:57:59<41:16,  5.18s/it]
 93%|█████████▎| 6698/7175 [1:58:01<33:45,  4.25s/it]
 93%|█████████▎| 6699/7175 [1:58:03<28:27,  3.59s/it]
 93%|█████████▎| 6700/7175 [1:58:05<24:43,  3.12s/it]
                                                     
{'loss': '2.951', 'grad_norm': '0.2197', 'learning_rate': '0.0006', 'epoch': '23.35'}
+
 93%|█████████▎| 6700/7175 [1:58:05<24:43,  3.12s/it]
 93%|█████████▎| 6701/7175 [1:58:07<22:04,  2.79s/it]
 93%|█████████▎| 6702/7175 [1:58:09<20:08,  2.56s/it]
 93%|█████████▎| 6703/7175 [1:58:12<19:01,  2.42s/it]
 93%|█████████▎| 6704/7175 [1:58:14<18:09,  2.31s/it]
 93%|█████████▎| 6705/7175 [1:58:16<17:34,  2.24s/it]
 93%|█████████▎| 6706/7175 [1:58:18<17:07,  2.19s/it]
 93%|█████████▎| 6707/7175 [1:58:20<16:47,  2.15s/it]
 93%|█████████▎| 6708/7175 [1:58:22<16:30,  2.12s/it]
 94%|█████████▎| 6709/7175 [1:58:24<16:19,  2.10s/it]
 94%|█████████▎| 6710/7175 [1:58:26<16:10,  2.09s/it]
                                                     
{'loss': '2.951', 'grad_norm': '0.2309', 'learning_rate': '0.0006', 'epoch': '23.38'}
+
 94%|█████████▎| 6710/7175 [1:58:26<16:10,  2.09s/it]
 94%|█████████▎| 6711/7175 [1:58:28<16:10,  2.09s/it]
 94%|█████████▎| 6712/7175 [1:58:30<16:03,  2.08s/it]
 94%|█████████▎| 6713/7175 [1:58:32<15:57,  2.07s/it]
 94%|█████████▎| 6714/7175 [1:58:34<15:46,  2.05s/it]
 94%|█████████▎| 6715/7175 [1:58:36<15:47,  2.06s/it]
 94%|█████████▎| 6716/7175 [1:58:38<15:47,  2.06s/it]
 94%|█████████▎| 6717/7175 [1:58:40<15:48,  2.07s/it]
 94%|█████████▎| 6718/7175 [1:58:42<15:46,  2.07s/it]
 94%|█████████▎| 6719/7175 [1:58:45<15:41,  2.06s/it]
 94%|█████████▎| 6720/7175 [1:58:47<15:37,  2.06s/it]
                                                     
{'loss': '2.963', 'grad_norm': '0.2244', 'learning_rate': '0.0006', 'epoch': '23.42'}
+
 94%|█████████▎| 6720/7175 [1:58:47<15:37,  2.06s/it]
 94%|█████████▎| 6721/7175 [1:58:49<15:34,  2.06s/it]
 94%|█████████▎| 6722/7175 [1:58:51<15:29,  2.05s/it]
 94%|█████████▎| 6723/7175 [1:58:53<15:26,  2.05s/it]
 94%|█████████▎| 6724/7175 [1:58:55<15:24,  2.05s/it]
 94%|█████████▎| 6725/7175 [1:58:57<15:20,  2.04s/it]
 94%|█████████▎| 6726/7175 [1:58:59<15:13,  2.03s/it]
 94%|█████████▍| 6727/7175 [1:59:01<15:16,  2.05s/it]
 94%|█████████▍| 6728/7175 [1:59:03<15:21,  2.06s/it]
 94%|█████████▍| 6729/7175 [1:59:05<15:21,  2.07s/it]
 94%|█████████▍| 6730/7175 [1:59:07<15:18,  2.06s/it]
                                                     
{'loss': '2.97', 'grad_norm': '0.2125', 'learning_rate': '0.0006', 'epoch': '23.45'}
+
 94%|█████████▍| 6730/7175 [1:59:07<15:18,  2.06s/it]
 94%|█████████▍| 6731/7175 [1:59:09<15:21,  2.07s/it]
 94%|█████████▍| 6732/7175 [1:59:11<15:17,  2.07s/it]
 94%|█████████▍| 6733/7175 [1:59:13<15:10,  2.06s/it]
 94%|█████████▍| 6734/7175 [1:59:15<15:08,  2.06s/it]
 94%|█████████▍| 6735/7175 [1:59:17<15:05,  2.06s/it]
 94%|█████████▍| 6736/7175 [1:59:19<15:01,  2.05s/it]
 94%|█████████▍| 6737/7175 [1:59:22<15:00,  2.05s/it]
 94%|█████████▍| 6738/7175 [1:59:24<14:58,  2.06s/it]
 94%|█████████▍| 6739/7175 [1:59:26<14:58,  2.06s/it]
 94%|█████████▍| 6740/7175 [1:59:28<15:00,  2.07s/it]
                                                     
{'loss': '2.963', 'grad_norm': '0.2312', 'learning_rate': '0.0006', 'epoch': '23.49'}
+
 94%|█████████▍| 6740/7175 [1:59:28<15:00,  2.07s/it]
 94%|█████████▍| 6741/7175 [1:59:30<15:02,  2.08s/it]
 94%|█████████▍| 6742/7175 [1:59:32<14:58,  2.08s/it]
 94%|█████████▍| 6743/7175 [1:59:34<14:52,  2.07s/it]
 94%|█████████▍| 6744/7175 [1:59:36<14:46,  2.06s/it]
 94%|█████████▍| 6745/7175 [1:59:38<14:43,  2.05s/it]
 94%|█████████▍| 6746/7175 [1:59:40<14:41,  2.06s/it]
 94%|█████████▍| 6747/7175 [1:59:42<14:37,  2.05s/it]
 94%|█████████▍| 6748/7175 [1:59:44<14:33,  2.05s/it]
 94%|█████████▍| 6749/7175 [1:59:46<14:32,  2.05s/it]
 94%|█████████▍| 6750/7175 [1:59:48<14:34,  2.06s/it]
                                                     
{'loss': '2.971', 'grad_norm': '0.2107', 'learning_rate': '0.0006', 'epoch': '23.52'}
+
 94%|█████████▍| 6750/7175 [1:59:48<14:34,  2.06s/it]
 94%|█████████▍| 6751/7175 [1:59:50<14:40,  2.08s/it]
 94%|█████████▍| 6752/7175 [1:59:53<14:40,  2.08s/it]
 94%|█████████▍| 6753/7175 [1:59:55<14:40,  2.09s/it]
 94%|█████████▍| 6754/7175 [1:59:57<14:37,  2.09s/it]
 94%|█████████▍| 6755/7175 [1:59:59<14:33,  2.08s/it]
 94%|█████████▍| 6756/7175 [2:00:01<14:31,  2.08s/it]
 94%|█████████▍| 6757/7175 [2:00:03<14:26,  2.07s/it]
 94%|█████████▍| 6758/7175 [2:00:05<14:24,  2.07s/it]
 94%|█████████▍| 6759/7175 [2:00:07<14:21,  2.07s/it]
 94%|█████████▍| 6760/7175 [2:00:09<14:18,  2.07s/it]
                                                     
{'loss': '2.978', 'grad_norm': '0.2356', 'learning_rate': '0.0006', 'epoch': '23.56'}
+
 94%|█████████▍| 6760/7175 [2:00:09<14:18,  2.07s/it]
 94%|█████████▍| 6761/7175 [2:00:11<14:18,  2.07s/it]
 94%|█████████▍| 6762/7175 [2:00:13<14:16,  2.07s/it]
 94%|█████████▍| 6763/7175 [2:00:15<14:16,  2.08s/it]
 94%|█████████▍| 6764/7175 [2:00:17<14:14,  2.08s/it]
 94%|█████████▍| 6765/7175 [2:00:20<14:13,  2.08s/it]
 94%|█████████▍| 6766/7175 [2:00:22<14:11,  2.08s/it]
 94%|█████████▍| 6767/7175 [2:00:24<14:03,  2.07s/it]
 94%|█████████▍| 6768/7175 [2:00:26<13:59,  2.06s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:04,  7.67it/s]
+
  8%|▊         | 3/39 [00:00<00:05,  6.48it/s]
+
 10%|█         | 4/39 [00:00<00:05,  6.34it/s]
+
 13%|█▎        | 5/39 [00:00<00:05,  6.37it/s]
+
 15%|█▌        | 6/39 [00:00<00:05,  6.14it/s]
+
 18%|█▊        | 7/39 [00:01<00:05,  6.01it/s]
+
 21%|██        | 8/39 [00:01<00:06,  5.03it/s]
+
 23%|██▎       | 9/39 [00:01<00:06,  4.54it/s]
+
 26%|██▌       | 10/39 [00:01<00:06,  4.23it/s]
+
 28%|██▊       | 11/39 [00:02<00:06,  4.20it/s]
+
 31%|███       | 12/39 [00:02<00:05,  4.66it/s]
+
 33%|███▎      | 13/39 [00:02<00:05,  4.40it/s]
+
 36%|███▌      | 14/39 [00:02<00:06,  4.14it/s]
+
 38%|███▊      | 15/39 [00:03<00:05,  4.03it/s]
+
 41%|████      | 16/39 [00:03<00:05,  3.99it/s]
+
 44%|████▎     | 17/39 [00:03<00:05,  3.95it/s]
+
 46%|████▌     | 18/39 [00:03<00:05,  3.89it/s]
+
 49%|████▊     | 19/39 [00:04<00:05,  3.85it/s]
+
 51%|█████▏    | 20/39 [00:04<00:04,  3.81it/s]
+
 54%|█████▍    | 21/39 [00:04<00:04,  3.83it/s]
+
 56%|█████▋    | 22/39 [00:04<00:04,  3.81it/s]
+
 59%|█████▉    | 23/39 [00:05<00:04,  3.82it/s]
+
 62%|██████▏   | 24/39 [00:05<00:03,  3.84it/s]
+
 64%|██████▍   | 25/39 [00:05<00:03,  3.84it/s]
+
 67%|██████▋   | 26/39 [00:06<00:03,  3.83it/s]
+
 69%|██████▉   | 27/39 [00:06<00:03,  3.84it/s]
+
 72%|███████▏  | 28/39 [00:06<00:02,  3.82it/s]
+
 74%|███████▍  | 29/39 [00:06<00:02,  3.83it/s]
+
 77%|███████▋  | 30/39 [00:07<00:02,  3.82it/s]
+
 79%|███████▉  | 31/39 [00:07<00:02,  3.84it/s]
+
 82%|████████▏ | 32/39 [00:07<00:01,  3.85it/s]
+
 85%|████████▍ | 33/39 [00:07<00:01,  3.85it/s]
+
 87%|████████▋ | 34/39 [00:08<00:01,  3.83it/s]
+
 90%|████████▉ | 35/39 [00:08<00:01,  3.83it/s]
+
 92%|█████████▏| 36/39 [00:08<00:00,  3.84it/s]
+
 95%|█████████▍| 37/39 [00:08<00:00,  3.83it/s]
+
 97%|█████████▋| 38/39 [00:09<00:00,  3.86it/s]
+
100%|██████████| 39/39 [00:09<00:00,  3.84it/s]
                                                     
+
                                               
{'eval_loss': '3.512', 'eval_runtime': '9.719', 'eval_samples_per_second': '251.2', 'eval_steps_per_second': '4.013', 'epoch': '23.58'}
+
 94%|█████████▍| 6768/7175 [2:00:35<13:59,  2.06s/it]
+
100%|██████████| 39/39 [00:09<00:00,  3.84it/s][eval] epoch 23.584 step 6768 loss 3.5116 (best 3.5035 @ ep 19.821, 15/20 since best)
+
+
                                               
 94%|█████████▍| 6769/7175 [2:00:37<33:31,  4.95s/it]
 94%|█████████▍| 6770/7175 [2:00:39<27:26,  4.06s/it]
                                                     
{'loss': '2.97', 'grad_norm': '0.2337', 'learning_rate': '0.0006', 'epoch': '23.59'}
+
 94%|█████████▍| 6770/7175 [2:00:39<27:26,  4.06s/it]
 94%|█████████▍| 6771/7175 [2:00:42<23:29,  3.49s/it]
 94%|█████████▍| 6772/7175 [2:00:44<20:33,  3.06s/it]
 94%|█████████▍| 6773/7175 [2:00:46<18:29,  2.76s/it]
 94%|█████████▍| 6774/7175 [2:00:48<17:00,  2.55s/it]
 94%|█████████▍| 6775/7175 [2:00:50<15:58,  2.40s/it]
 94%|█████████▍| 6776/7175 [2:00:52<15:14,  2.29s/it]
 94%|█████████▍| 6777/7175 [2:00:54<14:40,  2.21s/it]
 94%|█████████▍| 6778/7175 [2:00:56<14:19,  2.16s/it]
 94%|█████████▍| 6779/7175 [2:00:58<14:03,  2.13s/it]
 94%|█████████▍| 6780/7175 [2:01:00<13:45,  2.09s/it]
                                                     
{'loss': '2.976', 'grad_norm': '0.2101', 'learning_rate': '0.0006', 'epoch': '23.63'}
+
 94%|█████████▍| 6780/7175 [2:01:00<13:45,  2.09s/it]
 95%|█████████▍| 6781/7175 [2:01:02<13:47,  2.10s/it]
 95%|█████████▍| 6782/7175 [2:01:04<13:44,  2.10s/it]
 95%|█████████▍| 6783/7175 [2:01:06<13:41,  2.10s/it]
 95%|█████████▍| 6784/7175 [2:01:08<13:37,  2.09s/it]
 95%|█████████▍| 6785/7175 [2:01:10<13:31,  2.08s/it]
 95%|█████████▍| 6786/7175 [2:01:12<13:25,  2.07s/it]
 95%|█████████▍| 6787/7175 [2:01:14<13:24,  2.07s/it]
 95%|█████████▍| 6788/7175 [2:01:17<13:20,  2.07s/it]
 95%|█████████▍| 6789/7175 [2:01:19<13:16,  2.06s/it]
 95%|█████████▍| 6790/7175 [2:01:21<13:12,  2.06s/it]
                                                     
{'loss': '2.975', 'grad_norm': '0.2296', 'learning_rate': '0.0006', 'epoch': '23.66'}
+
 95%|█████████▍| 6790/7175 [2:01:21<13:12,  2.06s/it]
 95%|█████████▍| 6791/7175 [2:01:23<13:11,  2.06s/it]
 95%|█████████▍| 6792/7175 [2:01:25<13:07,  2.06s/it]
 95%|█████████▍| 6793/7175 [2:01:27<13:08,  2.06s/it]
 95%|█████████▍| 6794/7175 [2:01:29<13:09,  2.07s/it]
 95%|█████████▍| 6795/7175 [2:01:31<13:07,  2.07s/it]
 95%|█████████▍| 6796/7175 [2:01:33<13:06,  2.07s/it]
 95%|█████████▍| 6797/7175 [2:01:35<13:01,  2.07s/it]
 95%|█████████▍| 6798/7175 [2:01:37<12:56,  2.06s/it]
 95%|█████████▍| 6799/7175 [2:01:39<12:51,  2.05s/it]
 95%|█████████▍| 6800/7175 [2:01:41<12:48,  2.05s/it]
                                                     
{'loss': '2.996', 'grad_norm': '0.2525', 'learning_rate': '0.0006', 'epoch': '23.7'}
+
 95%|█████████▍| 6800/7175 [2:01:41<12:48,  2.05s/it]
 95%|█████████▍| 6801/7175 [2:01:43<12:51,  2.06s/it]
 95%|█████████▍| 6802/7175 [2:01:45<12:46,  2.06s/it]
 95%|█████████▍| 6803/7175 [2:01:47<12:43,  2.05s/it]
 95%|█████████▍| 6804/7175 [2:01:49<12:44,  2.06s/it]
 95%|█████████▍| 6805/7175 [2:01:52<12:45,  2.07s/it]
 95%|█████████▍| 6806/7175 [2:01:54<12:43,  2.07s/it]
 95%|█████████▍| 6807/7175 [2:01:56<12:42,  2.07s/it]
 95%|█████████▍| 6808/7175 [2:01:58<12:41,  2.07s/it]
 95%|█████████▍| 6809/7175 [2:02:00<12:36,  2.07s/it]
 95%|█████████▍| 6810/7175 [2:02:02<12:34,  2.07s/it]
                                                     
{'loss': '2.981', 'grad_norm': '0.2182', 'learning_rate': '0.0006', 'epoch': '23.73'}
+
 95%|█████████▍| 6810/7175 [2:02:02<12:34,  2.07s/it]
 95%|█████████▍| 6811/7175 [2:02:04<12:34,  2.07s/it]
 95%|█████████▍| 6812/7175 [2:02:06<12:30,  2.07s/it]
 95%|█████████▍| 6813/7175 [2:02:08<12:27,  2.06s/it]
 95%|█████████▍| 6814/7175 [2:02:10<12:23,  2.06s/it]
 95%|█████████▍| 6815/7175 [2:02:12<12:18,  2.05s/it]
 95%|█████████▍| 6816/7175 [2:02:14<12:18,  2.06s/it]
 95%|█████████▌| 6817/7175 [2:02:16<12:16,  2.06s/it]
 95%|█████████▌| 6818/7175 [2:02:18<12:15,  2.06s/it]
 95%|█████████▌| 6819/7175 [2:02:20<12:13,  2.06s/it]
 95%|█████████▌| 6820/7175 [2:02:23<12:11,  2.06s/it]
                                                     
{'loss': '2.996', 'grad_norm': '0.2222', 'learning_rate': '0.0006', 'epoch': '23.77'}
+
 95%|█████████▌| 6820/7175 [2:02:23<12:11,  2.06s/it]
 95%|█████████▌| 6821/7175 [2:02:25<12:12,  2.07s/it]
 95%|█████████▌| 6822/7175 [2:02:27<12:07,  2.06s/it]
 95%|█████████▌| 6823/7175 [2:02:29<12:04,  2.06s/it]
 95%|█████████▌| 6824/7175 [2:02:31<12:02,  2.06s/it]
 95%|█████████▌| 6825/7175 [2:02:33<11:59,  2.06s/it]
 95%|█████████▌| 6826/7175 [2:02:35<11:54,  2.05s/it]
 95%|█████████▌| 6827/7175 [2:02:37<11:51,  2.05s/it]
 95%|█████████▌| 6828/7175 [2:02:39<11:42,  2.03s/it]
 95%|█████████▌| 6829/7175 [2:02:41<11:45,  2.04s/it]
 95%|█████████▌| 6830/7175 [2:02:43<11:48,  2.05s/it]
                                                     
{'loss': '2.996', 'grad_norm': '0.2212', 'learning_rate': '0.0006', 'epoch': '23.8'}
+
 95%|█████████▌| 6830/7175 [2:02:43<11:48,  2.05s/it]
 95%|█████████▌| 6831/7175 [2:02:45<11:52,  2.07s/it]
 95%|█████████▌| 6832/7175 [2:02:47<11:51,  2.07s/it]
 95%|█████████▌| 6833/7175 [2:02:49<11:46,  2.07s/it]
 95%|█████████▌| 6834/7175 [2:02:51<11:43,  2.06s/it]
 95%|█████████▌| 6835/7175 [2:02:53<11:40,  2.06s/it]
 95%|█████████▌| 6836/7175 [2:02:55<11:38,  2.06s/it]
 95%|█████████▌| 6837/7175 [2:02:57<11:35,  2.06s/it]
 95%|█████████▌| 6838/7175 [2:03:00<11:33,  2.06s/it]
 95%|█████████▌| 6839/7175 [2:03:02<11:31,  2.06s/it]
 95%|█████████▌| 6840/7175 [2:03:04<11:22,  2.04s/it]
                                                     
{'loss': '2.974', 'grad_norm': '0.2174', 'learning_rate': '0.0006', 'epoch': '23.84'}
+
 95%|█████████▌| 6840/7175 [2:03:04<11:22,  2.04s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:04,  8.07it/s]
+
  8%|▊         | 3/39 [00:00<00:06,  5.43it/s]
+
 10%|█         | 4/39 [00:00<00:07,  4.68it/s]
+
 13%|█▎        | 5/39 [00:01<00:08,  4.23it/s]
+
 15%|█▌        | 6/39 [00:01<00:08,  4.06it/s]
+
 18%|█▊        | 7/39 [00:01<00:08,  3.92it/s]
+
 21%|██        | 8/39 [00:01<00:08,  3.87it/s]
+
 23%|██▎       | 9/39 [00:02<00:08,  3.74it/s]
+
 26%|██▌       | 10/39 [00:02<00:07,  3.70it/s]
+
 28%|██▊       | 11/39 [00:02<00:07,  3.67it/s]
+
 31%|███       | 12/39 [00:02<00:07,  3.67it/s]
+
 33%|███▎      | 13/39 [00:03<00:07,  3.68it/s]
+
 36%|███▌      | 14/39 [00:03<00:06,  3.69it/s]
+
 38%|███▊      | 15/39 [00:03<00:06,  3.73it/s]
+
 41%|████      | 16/39 [00:04<00:06,  3.72it/s]
+
 44%|████▎     | 17/39 [00:04<00:05,  3.69it/s]
+
 46%|████▌     | 18/39 [00:04<00:05,  3.69it/s]
+
 49%|████▊     | 19/39 [00:04<00:05,  3.64it/s]
+
 51%|█████▏    | 20/39 [00:05<00:05,  3.64it/s]
+
 54%|█████▍    | 21/39 [00:05<00:04,  3.67it/s]
+
 56%|█████▋    | 22/39 [00:05<00:04,  3.67it/s]
+
 59%|█████▉    | 23/39 [00:05<00:04,  3.62it/s]
+
 62%|██████▏   | 24/39 [00:06<00:04,  3.68it/s]
+
 64%|██████▍   | 25/39 [00:06<00:03,  3.69it/s]
+
 67%|██████▋   | 26/39 [00:06<00:03,  3.71it/s]
+
 69%|██████▉   | 27/39 [00:07<00:03,  3.72it/s]
+
 72%|███████▏  | 28/39 [00:07<00:02,  3.74it/s]
+
 74%|███████▍  | 29/39 [00:07<00:02,  3.71it/s]
+
 77%|███████▋  | 30/39 [00:07<00:02,  3.68it/s]
+
 79%|███████▉  | 31/39 [00:08<00:02,  3.67it/s]
+
 82%|████████▏ | 32/39 [00:08<00:01,  3.64it/s]
+
 85%|████████▍ | 33/39 [00:08<00:01,  3.68it/s]
+
 87%|████████▋ | 34/39 [00:08<00:01,  3.68it/s]
+
 90%|████████▉ | 35/39 [00:09<00:01,  3.73it/s]
+
 92%|█████████▏| 36/39 [00:09<00:00,  3.71it/s]
+
 95%|█████████▍| 37/39 [00:09<00:00,  3.72it/s]
+
 97%|█████████▋| 38/39 [00:10<00:00,  3.72it/s]
+
100%|██████████| 39/39 [00:10<00:00,  3.72it/s]
                                                     
+
                                               
{'eval_loss': '3.51', 'eval_runtime': '10.62', 'eval_samples_per_second': '229.9', 'eval_steps_per_second': '3.673', 'epoch': '23.84'}
+
 95%|█████████▌| 6840/7175 [2:03:14<11:22,  2.04s/it]
+
100%|██████████| 39/39 [00:10<00:00,  3.72it/s][eval] epoch 23.835 step 6840 loss 3.5099 (best 3.5035 @ ep 19.821, 16/20 since best)
+
+
                                               
 95%|█████████▌| 6841/7175 [2:03:16<29:09,  5.24s/it]
 95%|█████████▌| 6842/7175 [2:03:18<23:46,  4.28s/it]
 95%|█████████▌| 6843/7175 [2:03:20<20:00,  3.62s/it]
 95%|█████████▌| 6844/7175 [2:03:22<17:21,  3.15s/it]
 95%|█████████▌| 6845/7175 [2:03:25<15:30,  2.82s/it]
 95%|█████████▌| 6846/7175 [2:03:27<14:12,  2.59s/it]
 95%|█████████▌| 6847/7175 [2:03:29<13:11,  2.41s/it]
 95%|█████████▌| 6848/7175 [2:03:31<12:35,  2.31s/it]
 95%|█████████▌| 6849/7175 [2:03:33<12:10,  2.24s/it]
 95%|█████████▌| 6850/7175 [2:03:35<11:52,  2.19s/it]
                                                     
{'loss': '2.988', 'grad_norm': '0.1969', 'learning_rate': '0.0006', 'epoch': '23.87'}
+
 95%|█████████▌| 6850/7175 [2:03:35<11:52,  2.19s/it]
 95%|█████████▌| 6851/7175 [2:03:37<11:43,  2.17s/it]
 95%|█████████▌| 6852/7175 [2:03:39<11:30,  2.14s/it]
 96%|█████████▌| 6853/7175 [2:03:41<11:19,  2.11s/it]
 96%|█████████▌| 6854/7175 [2:03:43<11:11,  2.09s/it]
 96%|█████████▌| 6855/7175 [2:03:45<11:05,  2.08s/it]
 96%|█████████▌| 6856/7175 [2:03:47<11:01,  2.08s/it]
 96%|█████████▌| 6857/7175 [2:03:49<10:57,  2.07s/it]
 96%|█████████▌| 6858/7175 [2:03:51<10:54,  2.06s/it]
 96%|█████████▌| 6859/7175 [2:03:53<10:46,  2.05s/it]
 96%|█████████▌| 6860/7175 [2:03:55<10:42,  2.04s/it]
                                                     
{'loss': '2.998', 'grad_norm': '0.2159', 'learning_rate': '0.0006', 'epoch': '23.91'}
+
 96%|█████████▌| 6860/7175 [2:03:55<10:42,  2.04s/it]
 96%|█████████▌| 6861/7175 [2:03:57<10:46,  2.06s/it]
 96%|█████████▌| 6862/7175 [2:04:00<10:45,  2.06s/it]
 96%|█████████▌| 6863/7175 [2:04:02<10:46,  2.07s/it]
 96%|█████████▌| 6864/7175 [2:04:04<10:44,  2.07s/it]
 96%|█████████▌| 6865/7175 [2:04:06<10:41,  2.07s/it]
 96%|█████████▌| 6866/7175 [2:04:08<10:38,  2.07s/it]
 96%|█████████▌| 6867/7175 [2:04:10<10:35,  2.06s/it]
 96%|█████████▌| 6868/7175 [2:04:12<10:33,  2.06s/it]
 96%|█████████▌| 6869/7175 [2:04:14<10:29,  2.06s/it]
 96%|█████████▌| 6870/7175 [2:04:16<10:28,  2.06s/it]
                                                     
{'loss': '2.995', 'grad_norm': '0.2499', 'learning_rate': '0.0006', 'epoch': '23.94'}
+
 96%|█████████▌| 6870/7175 [2:04:16<10:28,  2.06s/it]
 96%|█████████▌| 6871/7175 [2:04:18<10:28,  2.07s/it]
 96%|█████████▌| 6872/7175 [2:04:20<10:26,  2.07s/it]
 96%|█████████▌| 6873/7175 [2:04:22<10:26,  2.07s/it]
 96%|█████████▌| 6874/7175 [2:04:24<10:24,  2.08s/it]
 96%|█████████▌| 6875/7175 [2:04:26<10:22,  2.07s/it]
 96%|█████████▌| 6876/7175 [2:04:28<10:18,  2.07s/it]
 96%|█████████▌| 6877/7175 [2:04:31<10:14,  2.06s/it]
 96%|█████████▌| 6878/7175 [2:04:33<10:12,  2.06s/it]
 96%|█████████▌| 6879/7175 [2:04:35<10:09,  2.06s/it]
 96%|█████████▌| 6880/7175 [2:04:37<10:06,  2.06s/it]
                                                     
{'loss': '2.99', 'grad_norm': '0.225', 'learning_rate': '0.0006', 'epoch': '23.98'}
+
 96%|█████████▌| 6880/7175 [2:04:37<10:06,  2.06s/it]
 96%|█████████▌| 6881/7175 [2:04:39<10:06,  2.06s/it]
 96%|█████████▌| 6882/7175 [2:04:41<10:02,  2.06s/it]
 96%|█████████▌| 6883/7175 [2:04:43<09:58,  2.05s/it]
 96%|█████████▌| 6884/7175 [2:04:45<10:00,  2.06s/it]
 96%|█████████▌| 6885/7175 [2:04:47<10:00,  2.07s/it]
 96%|█████████▌| 6886/7175 [2:04:49<09:59,  2.07s/it]
 96%|█████████▌| 6887/7175 [2:04:51<09:54,  2.07s/it]
 96%|█████████▌| 6888/7175 [2:04:51<07:22,  1.54s/it]
 96%|█████████▌| 6889/7175 [2:04:54<08:11,  1.72s/it]
 96%|█████████▌| 6890/7175 [2:04:56<08:37,  1.82s/it]
                                                     
{'loss': '2.975', 'grad_norm': '0.3746', 'learning_rate': '0.0006', 'epoch': '24.01'}
+
 96%|█████████▌| 6890/7175 [2:04:56<08:37,  1.82s/it]
 96%|█████████▌| 6891/7175 [2:04:58<08:59,  1.90s/it]
 96%|█████████▌| 6892/7175 [2:05:00<09:09,  1.94s/it]
 96%|█████████▌| 6893/7175 [2:05:02<09:16,  1.97s/it]
 96%|█████████▌| 6894/7175 [2:05:04<09:20,  2.00s/it]
 96%|█████████▌| 6895/7175 [2:05:06<09:23,  2.01s/it]
 96%|█████████▌| 6896/7175 [2:05:08<09:20,  2.01s/it]
 96%|█████████▌| 6897/7175 [2:05:10<09:25,  2.03s/it]
 96%|█████████▌| 6898/7175 [2:05:12<09:27,  2.05s/it]
 96%|█████████▌| 6899/7175 [2:05:14<09:28,  2.06s/it]
 96%|█████████▌| 6900/7175 [2:05:16<09:30,  2.07s/it]
                                                     
{'loss': '2.914', 'grad_norm': '0.2787', 'learning_rate': '0.0006', 'epoch': '24.04'}
+
 96%|█████████▌| 6900/7175 [2:05:16<09:30,  2.07s/it]
 96%|█████████▌| 6901/7175 [2:05:18<09:31,  2.09s/it]
 96%|█████████▌| 6902/7175 [2:05:20<09:27,  2.08s/it]
 96%|█████████▌| 6903/7175 [2:05:23<09:22,  2.07s/it]
 96%|█████████▌| 6904/7175 [2:05:25<09:19,  2.06s/it]
 96%|█████████▌| 6905/7175 [2:05:27<09:17,  2.06s/it]
 96%|█████████▋| 6906/7175 [2:05:29<09:14,  2.06s/it]
 96%|█████████▋| 6907/7175 [2:05:31<09:11,  2.06s/it]
 96%|█████████▋| 6908/7175 [2:05:33<09:04,  2.04s/it]
 96%|█████████▋| 6909/7175 [2:05:35<09:03,  2.04s/it]
 96%|█████████▋| 6910/7175 [2:05:37<09:06,  2.06s/it]
                                                     
{'loss': '2.932', 'grad_norm': '0.2244', 'learning_rate': '0.0006', 'epoch': '24.08'}
+
 96%|█████████▋| 6910/7175 [2:05:37<09:06,  2.06s/it]
 96%|█████████▋| 6911/7175 [2:05:39<09:09,  2.08s/it]
 96%|█████████▋| 6912/7175 [2:05:41<09:08,  2.08s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:05,  7.34it/s]
+
  8%|▊         | 3/39 [00:00<00:06,  5.30it/s]
+
 10%|█         | 4/39 [00:00<00:07,  4.60it/s]
+
 13%|█▎        | 5/39 [00:01<00:07,  4.31it/s]
+
 15%|█▌        | 6/39 [00:01<00:08,  4.09it/s]
+
 18%|█▊        | 7/39 [00:01<00:08,  3.98it/s]
+
 21%|██        | 8/39 [00:01<00:07,  3.89it/s]
+
 23%|██▎       | 9/39 [00:02<00:07,  3.78it/s]
+
 26%|██▌       | 10/39 [00:02<00:07,  3.76it/s]
+
 28%|██▊       | 11/39 [00:02<00:07,  3.74it/s]
+
 31%|███       | 12/39 [00:02<00:07,  3.73it/s]
+
 33%|███▎      | 13/39 [00:03<00:06,  3.71it/s]
+
 36%|███▌      | 14/39 [00:03<00:06,  3.70it/s]
+
 38%|███▊      | 15/39 [00:03<00:06,  3.72it/s]
+
 41%|████      | 16/39 [00:04<00:06,  3.68it/s]
+
 44%|████▎     | 17/39 [00:04<00:06,  3.67it/s]
+
 46%|████▌     | 18/39 [00:04<00:05,  3.74it/s]
+
 49%|████▊     | 19/39 [00:04<00:05,  3.76it/s]
+
 51%|█████▏    | 20/39 [00:05<00:05,  3.75it/s]
+
 54%|█████▍    | 21/39 [00:05<00:04,  3.74it/s]
+
 56%|█████▋    | 22/39 [00:05<00:04,  3.72it/s]
+
 59%|█████▉    | 23/39 [00:05<00:04,  3.74it/s]
+
 62%|██████▏   | 24/39 [00:06<00:03,  3.75it/s]
+
 64%|██████▍   | 25/39 [00:06<00:03,  3.73it/s]
+
 67%|██████▋   | 26/39 [00:06<00:03,  3.78it/s]
+
 69%|██████▉   | 27/39 [00:06<00:03,  3.77it/s]
+
 72%|███████▏  | 28/39 [00:07<00:02,  3.80it/s]
+
 74%|███████▍  | 29/39 [00:07<00:02,  3.81it/s]
+
 77%|███████▋  | 30/39 [00:07<00:02,  3.77it/s]
+
 79%|███████▉  | 31/39 [00:08<00:02,  3.78it/s]
+
 82%|████████▏ | 32/39 [00:08<00:01,  3.78it/s]
+
 85%|████████▍ | 33/39 [00:08<00:01,  3.74it/s]
+
 87%|████████▋ | 34/39 [00:08<00:01,  3.74it/s]
+
 90%|████████▉ | 35/39 [00:09<00:01,  3.74it/s]
+
 92%|█████████▏| 36/39 [00:09<00:00,  3.73it/s]
+
 95%|█████████▍| 37/39 [00:09<00:00,  3.72it/s]
+
 97%|█████████▋| 38/39 [00:09<00:00,  3.79it/s]
+
100%|██████████| 39/39 [00:10<00:00,  3.78it/s]
                                                     
+
                                               
{'eval_loss': '3.531', 'eval_runtime': '10.49', 'eval_samples_per_second': '232.8', 'eval_steps_per_second': '3.719', 'epoch': '24.08'}
+
 96%|█████████▋| 6912/7175 [2:05:52<09:08,  2.08s/it]
+
100%|██████████| 39/39 [00:10<00:00,  3.78it/s][eval] epoch 24.084 step 6912 loss 3.5307 (best 3.5035 @ ep 19.821, 17/20 since best)
+
+
                                               
 96%|█████████▋| 6913/7175 [2:05:54<22:47,  5.22s/it]
 96%|█████████▋| 6914/7175 [2:05:56<18:34,  4.27s/it]
 96%|█████████▋| 6915/7175 [2:05:58<15:38,  3.61s/it]
 96%|█████████▋| 6916/7175 [2:06:00<13:37,  3.16s/it]
 96%|█████████▋| 6917/7175 [2:06:02<12:11,  2.83s/it]
 96%|█████████▋| 6918/7175 [2:06:04<11:10,  2.61s/it]
 96%|█████████▋| 6919/7175 [2:06:06<10:28,  2.45s/it]
 96%|█████████▋| 6920/7175 [2:06:08<09:56,  2.34s/it]
                                                     
{'loss': '2.921', 'grad_norm': '0.2118', 'learning_rate': '0.0006', 'epoch': '24.11'}
+
 96%|█████████▋| 6920/7175 [2:06:08<09:56,  2.34s/it]
 96%|█████████▋| 6921/7175 [2:06:10<09:36,  2.27s/it]
 96%|█████████▋| 6922/7175 [2:06:12<09:19,  2.21s/it]
 96%|█████████▋| 6923/7175 [2:06:14<09:06,  2.17s/it]
 97%|█████████▋| 6924/7175 [2:06:16<08:42,  2.08s/it]
 97%|█████████▋| 6925/7175 [2:06:18<08:08,  1.95s/it]
 97%|█████████▋| 6926/7175 [2:06:20<08:12,  1.98s/it]
 97%|█████████▋| 6927/7175 [2:06:22<08:12,  1.99s/it]
 97%|█████████▋| 6928/7175 [2:06:24<08:09,  1.98s/it]
 97%|█████████▋| 6929/7175 [2:06:26<08:08,  1.98s/it]
 97%|█████████▋| 6930/7175 [2:06:28<08:08,  1.99s/it]
                                                     
{'loss': '2.933', 'grad_norm': '0.2201', 'learning_rate': '0.0006', 'epoch': '24.15'}
+
 97%|█████████▋| 6930/7175 [2:06:28<08:08,  1.99s/it]
 97%|█████████▋| 6931/7175 [2:06:30<08:15,  2.03s/it]
 97%|█████████▋| 6932/7175 [2:06:32<08:15,  2.04s/it]
 97%|█████████▋| 6933/7175 [2:06:34<08:16,  2.05s/it]
 97%|█████████▋| 6934/7175 [2:06:36<08:14,  2.05s/it]
 97%|█████████▋| 6935/7175 [2:06:38<08:12,  2.05s/it]
 97%|█████████▋| 6936/7175 [2:06:40<08:10,  2.05s/it]
 97%|█████████▋| 6937/7175 [2:06:42<08:08,  2.05s/it]
 97%|█████████▋| 6938/7175 [2:06:45<08:06,  2.05s/it]
 97%|█████████▋| 6939/7175 [2:06:47<08:41,  2.21s/it]
 97%|█████████▋| 6940/7175 [2:06:49<08:31,  2.18s/it]
                                                     
{'loss': '2.927', 'grad_norm': '0.2345', 'learning_rate': '0.0006', 'epoch': '24.18'}
+
 97%|█████████▋| 6940/7175 [2:06:49<08:31,  2.18s/it]
 97%|█████████▋| 6941/7175 [2:06:51<08:24,  2.16s/it]
 97%|█████████▋| 6942/7175 [2:06:53<08:17,  2.14s/it]
 97%|█████████▋| 6943/7175 [2:06:56<08:12,  2.12s/it]
 97%|█████████▋| 6944/7175 [2:06:58<08:06,  2.11s/it]
 97%|█████████▋| 6945/7175 [2:07:00<07:59,  2.09s/it]
 97%|█████████▋| 6946/7175 [2:07:02<07:54,  2.07s/it]
 97%|█████████▋| 6947/7175 [2:07:04<07:50,  2.06s/it]
 97%|█████████▋| 6948/7175 [2:07:06<07:48,  2.07s/it]
 97%|█████████▋| 6949/7175 [2:07:08<07:46,  2.06s/it]
 97%|█████████▋| 6950/7175 [2:07:10<07:42,  2.06s/it]
                                                     
{'loss': '2.924', 'grad_norm': '0.2158', 'learning_rate': '0.0006', 'epoch': '24.22'}
+
 97%|█████████▋| 6950/7175 [2:07:10<07:42,  2.06s/it]
 97%|█████████▋| 6951/7175 [2:07:12<07:42,  2.07s/it]
 97%|█████████▋| 6952/7175 [2:07:14<07:40,  2.07s/it]
 97%|█████████▋| 6953/7175 [2:07:16<07:39,  2.07s/it]
 97%|█████████▋| 6954/7175 [2:07:18<07:37,  2.07s/it]
 97%|█████████▋| 6955/7175 [2:07:20<07:36,  2.07s/it]
 97%|█████████▋| 6956/7175 [2:07:22<07:32,  2.06s/it]
 97%|█████████▋| 6957/7175 [2:07:24<07:30,  2.07s/it]
 97%|█████████▋| 6958/7175 [2:07:26<07:27,  2.06s/it]
 97%|█████████▋| 6959/7175 [2:07:28<07:27,  2.07s/it]
 97%|█████████▋| 6960/7175 [2:07:31<07:42,  2.15s/it]
                                                     
{'loss': '2.938', 'grad_norm': '0.2109', 'learning_rate': '0.0006', 'epoch': '24.25'}
+
 97%|█████████▋| 6960/7175 [2:07:31<07:42,  2.15s/it]
 97%|█████████▋| 6961/7175 [2:07:33<07:35,  2.13s/it]
 97%|█████████▋| 6962/7175 [2:07:35<07:28,  2.10s/it]
 97%|█████████▋| 6963/7175 [2:07:37<07:20,  2.08s/it]
 97%|█████████▋| 6964/7175 [2:07:39<07:18,  2.08s/it]
 97%|█████████▋| 6965/7175 [2:07:41<07:16,  2.08s/it]
 97%|█████████▋| 6966/7175 [2:07:43<07:13,  2.08s/it]
 97%|█████████▋| 6967/7175 [2:07:45<07:10,  2.07s/it]
 97%|█████████▋| 6968/7175 [2:07:47<07:07,  2.06s/it]
 97%|█████████▋| 6969/7175 [2:07:49<07:05,  2.07s/it]
 97%|█████████▋| 6970/7175 [2:07:51<07:01,  2.06s/it]
                                                     
{'loss': '2.948', 'grad_norm': '0.2119', 'learning_rate': '0.0006', 'epoch': '24.29'}
+
 97%|█████████▋| 6970/7175 [2:07:51<07:01,  2.06s/it]
 97%|█████████▋| 6971/7175 [2:07:54<07:16,  2.14s/it]
 97%|█████████▋| 6972/7175 [2:07:56<07:08,  2.11s/it]
 97%|█████████▋| 6973/7175 [2:07:58<07:02,  2.09s/it]
 97%|█████████▋| 6974/7175 [2:08:00<06:58,  2.08s/it]
 97%|█████████▋| 6975/7175 [2:08:02<06:51,  2.06s/it]
 97%|█████████▋| 6976/7175 [2:08:04<06:51,  2.07s/it]
 97%|█████████▋| 6977/7175 [2:08:06<06:50,  2.07s/it]
 97%|█████████▋| 6978/7175 [2:08:08<06:48,  2.07s/it]
 97%|█████████▋| 6979/7175 [2:08:10<06:47,  2.08s/it]
 97%|█████████▋| 6980/7175 [2:08:12<06:43,  2.07s/it]
                                                     
{'loss': '2.954', 'grad_norm': '0.2344', 'learning_rate': '0.0006', 'epoch': '24.32'}
+
 97%|█████████▋| 6980/7175 [2:08:12<06:43,  2.07s/it]
 97%|█████████▋| 6981/7175 [2:08:14<06:42,  2.07s/it]
 97%|█████████▋| 6982/7175 [2:08:16<06:39,  2.07s/it]
 97%|█████████▋| 6983/7175 [2:08:18<06:35,  2.06s/it]
 97%|█████████▋| 6984/7175 [2:08:21<06:32,  2.06s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:04,  7.90it/s]
+
  8%|▊         | 3/39 [00:00<00:06,  5.38it/s]
+
 10%|█         | 4/39 [00:00<00:07,  4.68it/s]
+
 13%|█▎        | 5/39 [00:01<00:07,  4.32it/s]
+
 15%|█▌        | 6/39 [00:01<00:07,  4.13it/s]
+
 18%|█▊        | 7/39 [00:01<00:07,  4.01it/s]
+
 21%|██        | 8/39 [00:01<00:07,  3.91it/s]
+
 23%|██▎       | 9/39 [00:02<00:07,  3.85it/s]
+
 26%|██▌       | 10/39 [00:02<00:07,  3.77it/s]
+
 28%|██▊       | 11/39 [00:02<00:07,  3.80it/s]
+
 31%|███       | 12/39 [00:02<00:07,  3.75it/s]
+
 33%|███▎      | 13/39 [00:03<00:06,  3.74it/s]
+
 36%|███▌      | 14/39 [00:03<00:06,  3.79it/s]
+
 38%|███▊      | 15/39 [00:03<00:06,  3.79it/s]
+
 41%|████      | 16/39 [00:03<00:06,  3.81it/s]
+
 44%|████▎     | 17/39 [00:04<00:05,  3.74it/s]
+
 46%|████▌     | 18/39 [00:04<00:05,  3.76it/s]
+
 49%|████▊     | 19/39 [00:04<00:05,  3.77it/s]
+
 51%|█████▏    | 20/39 [00:05<00:05,  3.78it/s]
+
 54%|█████▍    | 21/39 [00:05<00:04,  3.78it/s]
+
 56%|█████▋    | 22/39 [00:05<00:04,  3.72it/s]
+
 59%|█████▉    | 23/39 [00:05<00:04,  3.69it/s]
+
 62%|██████▏   | 24/39 [00:06<00:04,  3.68it/s]
+
 64%|██████▍   | 25/39 [00:06<00:03,  3.70it/s]
+
 67%|██████▋   | 26/39 [00:06<00:03,  3.71it/s]
+
 69%|██████▉   | 27/39 [00:06<00:03,  3.75it/s]
+
 72%|███████▏  | 28/39 [00:07<00:02,  3.77it/s]
+
 74%|███████▍  | 29/39 [00:07<00:02,  3.76it/s]
+
 77%|███████▋  | 30/39 [00:07<00:02,  3.69it/s]
+
 79%|███████▉  | 31/39 [00:08<00:02,  3.68it/s]
+
 82%|████████▏ | 32/39 [00:08<00:01,  3.68it/s]
+
 85%|████████▍ | 33/39 [00:08<00:01,  3.69it/s]
+
 87%|████████▋ | 34/39 [00:08<00:01,  3.73it/s]
+
 90%|████████▉ | 35/39 [00:09<00:01,  3.72it/s]
+
 92%|█████████▏| 36/39 [00:09<00:00,  3.72it/s]
+
 95%|█████████▍| 37/39 [00:09<00:00,  3.72it/s]
+
 97%|█████████▋| 38/39 [00:09<00:00,  3.70it/s]
+
100%|██████████| 39/39 [00:10<00:00,  3.71it/s]
                                                     
+
                                               
{'eval_loss': '3.527', 'eval_runtime': '10.51', 'eval_samples_per_second': '232.4', 'eval_steps_per_second': '3.712', 'epoch': '24.34'}
+
 97%|█████████▋| 6984/7175 [2:08:31<06:32,  2.06s/it]
+
100%|██████████| 39/39 [00:10<00:00,  3.71it/s][eval] epoch 24.336 step 6984 loss 3.5269 (best 3.5035 @ ep 19.821, 18/20 since best)
+
+
                                               
 97%|█████████▋| 6985/7175 [2:08:33<16:32,  5.23s/it]
 97%|█████████▋| 6986/7175 [2:08:35<13:29,  4.28s/it]
 97%|█████████▋| 6987/7175 [2:08:37<11:19,  3.61s/it]
 97%|█████████▋| 6988/7175 [2:08:39<09:48,  3.15s/it]
 97%|█████████▋| 6989/7175 [2:08:41<08:43,  2.82s/it]
 97%|█████████▋| 6990/7175 [2:08:43<07:58,  2.59s/it]
                                                     
{'loss': '2.952', 'grad_norm': '0.2255', 'learning_rate': '0.0006', 'epoch': '24.36'}
+
 97%|█████████▋| 6990/7175 [2:08:43<07:58,  2.59s/it]
 97%|█████████▋| 6991/7175 [2:08:46<07:27,  2.43s/it]
 97%|█████████▋| 6992/7175 [2:08:48<07:05,  2.32s/it]
 97%|█████████▋| 6993/7175 [2:08:50<06:48,  2.24s/it]
 97%|█████████▋| 6994/7175 [2:08:52<06:32,  2.17s/it]
 97%|█████████▋| 6995/7175 [2:08:54<06:25,  2.14s/it]
 98%|█████████▊| 6996/7175 [2:08:56<06:18,  2.12s/it]
 98%|█████████▊| 6997/7175 [2:08:58<06:14,  2.10s/it]
 98%|█████████▊| 6998/7175 [2:09:00<06:10,  2.09s/it]
 98%|█████████▊| 6999/7175 [2:09:02<06:06,  2.08s/it]
 98%|█████████▊| 7000/7175 [2:09:04<06:02,  2.07s/it]
                                                     
{'loss': '2.946', 'grad_norm': '0.2324', 'learning_rate': '0.0006', 'epoch': '24.39'}
+
 98%|█████████▊| 7000/7175 [2:09:04<06:02,  2.07s/it]
 98%|█████████▊| 7001/7175 [2:09:06<06:00,  2.07s/it]
 98%|█████████▊| 7002/7175 [2:09:08<05:58,  2.07s/it]
 98%|█████████▊| 7003/7175 [2:09:10<05:54,  2.06s/it]
 98%|█████████▊| 7004/7175 [2:09:12<05:51,  2.05s/it]
 98%|█████████▊| 7005/7175 [2:09:15<06:05,  2.15s/it]
 98%|█████████▊| 7006/7175 [2:09:17<05:58,  2.12s/it]
 98%|█████████▊| 7007/7175 [2:09:19<05:54,  2.11s/it]
 98%|█████████▊| 7008/7175 [2:09:21<05:50,  2.10s/it]
 98%|█████████▊| 7009/7175 [2:09:23<05:47,  2.09s/it]
 98%|█████████▊| 7010/7175 [2:09:25<05:44,  2.09s/it]
                                                     
{'loss': '2.955', 'grad_norm': '0.2349', 'learning_rate': '0.0006', 'epoch': '24.43'}
+
 98%|█████████▊| 7010/7175 [2:09:25<05:44,  2.09s/it]
 98%|█████████▊| 7011/7175 [2:09:27<05:42,  2.09s/it]
 98%|█████████▊| 7012/7175 [2:09:29<05:37,  2.07s/it]
 98%|█████████▊| 7013/7175 [2:09:31<05:35,  2.07s/it]
 98%|█████████▊| 7014/7175 [2:09:33<05:32,  2.06s/it]
 98%|█████████▊| 7015/7175 [2:09:35<05:29,  2.06s/it]
 98%|█████████▊| 7016/7175 [2:09:37<05:26,  2.05s/it]
 98%|█████████▊| 7017/7175 [2:09:39<05:24,  2.05s/it]
 98%|█████████▊| 7018/7175 [2:09:41<05:19,  2.04s/it]
 98%|█████████▊| 7019/7175 [2:09:43<05:19,  2.05s/it]
 98%|█████████▊| 7020/7175 [2:09:45<05:18,  2.05s/it]
                                                     
{'loss': '2.955', 'grad_norm': '0.2264', 'learning_rate': '0.0006', 'epoch': '24.46'}
+
 98%|█████████▊| 7020/7175 [2:09:46<05:18,  2.05s/it]
 98%|█████████▊| 7021/7175 [2:09:48<05:19,  2.07s/it]
 98%|█████████▊| 7022/7175 [2:09:50<05:17,  2.08s/it]
 98%|█████████▊| 7023/7175 [2:09:52<05:14,  2.07s/it]
 98%|█████████▊| 7024/7175 [2:09:54<05:12,  2.07s/it]
 98%|█████████▊| 7025/7175 [2:09:56<05:09,  2.06s/it]
 98%|█████████▊| 7026/7175 [2:09:58<05:07,  2.06s/it]
 98%|█████████▊| 7027/7175 [2:10:00<05:03,  2.05s/it]
 98%|█████████▊| 7028/7175 [2:10:02<05:01,  2.05s/it]
 98%|█████████▊| 7029/7175 [2:10:04<04:59,  2.05s/it]
 98%|█████████▊| 7030/7175 [2:10:06<04:57,  2.05s/it]
                                                     
{'loss': '2.952', 'grad_norm': '0.2162', 'learning_rate': '0.0006', 'epoch': '24.5'}
+
 98%|█████████▊| 7030/7175 [2:10:06<04:57,  2.05s/it]
 98%|█████████▊| 7031/7175 [2:10:08<04:57,  2.07s/it]
 98%|█████████▊| 7032/7175 [2:10:10<04:56,  2.07s/it]
 98%|█████████▊| 7033/7175 [2:10:12<04:53,  2.07s/it]
 98%|█████████▊| 7034/7175 [2:10:14<04:52,  2.07s/it]
 98%|█████████▊| 7035/7175 [2:10:16<04:49,  2.07s/it]
 98%|█████████▊| 7036/7175 [2:10:19<04:47,  2.07s/it]
 98%|█████████▊| 7037/7175 [2:10:21<04:44,  2.06s/it]
 98%|█████████▊| 7038/7175 [2:10:23<04:42,  2.06s/it]
 98%|█████████▊| 7039/7175 [2:10:25<04:40,  2.06s/it]
 98%|█████████▊| 7040/7175 [2:10:27<04:37,  2.06s/it]
                                                     
{'loss': '2.963', 'grad_norm': '0.2534', 'learning_rate': '0.0006', 'epoch': '24.53'}
+
 98%|█████████▊| 7040/7175 [2:10:27<04:37,  2.06s/it]
 98%|█████████▊| 7041/7175 [2:10:29<04:36,  2.06s/it]
 98%|█████████▊| 7042/7175 [2:10:31<04:34,  2.06s/it]
 98%|█████████▊| 7043/7175 [2:10:33<04:33,  2.07s/it]
 98%|█████████▊| 7044/7175 [2:10:35<04:30,  2.07s/it]
 98%|█████████▊| 7045/7175 [2:10:37<04:30,  2.08s/it]
 98%|█████████▊| 7046/7175 [2:10:39<04:27,  2.08s/it]
 98%|█████████▊| 7047/7175 [2:10:41<04:24,  2.07s/it]
 98%|█████████▊| 7048/7175 [2:10:43<04:21,  2.06s/it]
 98%|█████████▊| 7049/7175 [2:10:45<04:19,  2.06s/it]
 98%|█████████▊| 7050/7175 [2:10:47<04:16,  2.05s/it]
                                                     
{'loss': '2.966', 'grad_norm': '0.2439', 'learning_rate': '0.0006', 'epoch': '24.57'}
+
 98%|█████████▊| 7050/7175 [2:10:47<04:16,  2.05s/it]
 98%|█████████▊| 7051/7175 [2:10:49<04:15,  2.06s/it]
 98%|█████████▊| 7052/7175 [2:10:52<04:12,  2.05s/it]
 98%|█████████▊| 7053/7175 [2:10:54<04:10,  2.06s/it]
 98%|█████████▊| 7054/7175 [2:10:56<04:08,  2.05s/it]
 98%|█████████▊| 7055/7175 [2:10:58<04:07,  2.06s/it]
 98%|█████████▊| 7056/7175 [2:11:00<04:06,  2.07s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:04,  7.96it/s]
+
  8%|▊         | 3/39 [00:00<00:06,  5.39it/s]
+
 10%|█         | 4/39 [00:00<00:07,  4.61it/s]
+
 13%|█▎        | 5/39 [00:01<00:07,  4.27it/s]
+
 15%|█▌        | 6/39 [00:01<00:08,  4.05it/s]
+
 18%|█▊        | 7/39 [00:01<00:08,  3.98it/s]
+
 21%|██        | 8/39 [00:01<00:07,  3.90it/s]
+
 23%|██▎       | 9/39 [00:02<00:07,  3.80it/s]
+
 26%|██▌       | 10/39 [00:02<00:07,  3.76it/s]
+
 28%|██▊       | 11/39 [00:02<00:07,  3.76it/s]
+
 31%|███       | 12/39 [00:02<00:07,  3.73it/s]
+
 33%|███▎      | 13/39 [00:03<00:07,  3.71it/s]
+
 36%|███▌      | 14/39 [00:03<00:06,  3.74it/s]
+
 38%|███▊      | 15/39 [00:03<00:06,  3.77it/s]
+
 41%|████      | 16/39 [00:04<00:06,  3.76it/s]
+
 44%|████▎     | 17/39 [00:04<00:05,  3.74it/s]
+
 46%|████▌     | 18/39 [00:04<00:05,  3.71it/s]
+
 49%|████▊     | 19/39 [00:04<00:05,  3.70it/s]
+
 51%|█████▏    | 20/39 [00:05<00:05,  3.72it/s]
+
 54%|█████▍    | 21/39 [00:05<00:04,  3.70it/s]
+
 56%|█████▋    | 22/39 [00:05<00:04,  3.69it/s]
+
 59%|█████▉    | 23/39 [00:05<00:04,  3.75it/s]
+
 62%|██████▏   | 24/39 [00:06<00:03,  3.76it/s]
+
 64%|██████▍   | 25/39 [00:06<00:03,  3.74it/s]
+
 67%|██████▋   | 26/39 [00:06<00:03,  3.81it/s]
+
 69%|██████▉   | 27/39 [00:06<00:03,  3.79it/s]
+
 72%|███████▏  | 28/39 [00:07<00:02,  3.77it/s]
+
 74%|███████▍  | 29/39 [00:07<00:02,  3.77it/s]
+
 77%|███████▋  | 30/39 [00:07<00:02,  3.78it/s]
+
 79%|███████▉  | 31/39 [00:08<00:02,  3.80it/s]
+
 82%|████████▏ | 32/39 [00:08<00:01,  3.80it/s]
+
 85%|████████▍ | 33/39 [00:08<00:01,  3.81it/s]
+
 87%|████████▋ | 34/39 [00:08<00:01,  3.83it/s]
+
 90%|████████▉ | 35/39 [00:09<00:01,  3.82it/s]
+
 92%|█████████▏| 36/39 [00:09<00:00,  3.78it/s]
+
 95%|█████████▍| 37/39 [00:09<00:00,  3.79it/s]
+
 97%|█████████▋| 38/39 [00:09<00:00,  3.78it/s]
+
100%|██████████| 39/39 [00:10<00:00,  3.79it/s]
                                                     
+
                                               
{'eval_loss': '3.515', 'eval_runtime': '10.43', 'eval_samples_per_second': '234.1', 'eval_steps_per_second': '3.74', 'epoch': '24.59'}
+
 98%|█████████▊| 7056/7175 [2:11:10<04:06,  2.07s/it]
+
100%|██████████| 39/39 [00:10<00:00,  3.79it/s][eval] epoch 24.587 step 7056 loss 3.5152 (best 3.5035 @ ep 19.821, 19/20 since best)
+
+
                                               
 98%|█████████▊| 7057/7175 [2:11:12<10:12,  5.19s/it]
 98%|█████████▊| 7058/7175 [2:11:14<08:16,  4.25s/it]
 98%|█████████▊| 7059/7175 [2:11:16<06:56,  3.59s/it]
 98%|█████████▊| 7060/7175 [2:11:18<05:59,  3.13s/it]
                                                     
{'loss': '2.961', 'grad_norm': '0.2099', 'learning_rate': '0.0006', 'epoch': '24.6'}
+
 98%|█████████▊| 7060/7175 [2:11:18<05:59,  3.13s/it]
 98%|█████████▊| 7061/7175 [2:11:21<05:21,  2.82s/it]
 98%|█████████▊| 7062/7175 [2:11:23<04:53,  2.60s/it]
 98%|█████████▊| 7063/7175 [2:11:25<04:33,  2.44s/it]
 98%|█████████▊| 7064/7175 [2:11:27<04:19,  2.33s/it]
 98%|█████████▊| 7065/7175 [2:11:29<04:07,  2.25s/it]
 98%|█████████▊| 7066/7175 [2:11:31<03:59,  2.19s/it]
 98%|█████████▊| 7067/7175 [2:11:33<03:52,  2.15s/it]
 99%|█████████▊| 7068/7175 [2:11:35<03:46,  2.12s/it]
 99%|█████████▊| 7069/7175 [2:11:37<03:42,  2.10s/it]
 99%|█████████▊| 7070/7175 [2:11:39<03:38,  2.08s/it]
                                                     
{'loss': '2.974', 'grad_norm': '0.229', 'learning_rate': '0.0006', 'epoch': '24.64'}
+
 99%|█████████▊| 7070/7175 [2:11:39<03:38,  2.08s/it]
 99%|█████████▊| 7071/7175 [2:11:41<03:36,  2.09s/it]
 99%|█████████▊| 7072/7175 [2:11:43<03:33,  2.07s/it]
 99%|█████████▊| 7073/7175 [2:11:45<03:31,  2.07s/it]
 99%|█████████▊| 7074/7175 [2:11:47<03:30,  2.08s/it]
 99%|█████████▊| 7075/7175 [2:11:49<03:28,  2.09s/it]
 99%|█████████▊| 7076/7175 [2:11:52<03:26,  2.08s/it]
 99%|█████████▊| 7077/7175 [2:11:54<03:23,  2.08s/it]
 99%|█████████▊| 7078/7175 [2:11:56<03:21,  2.08s/it]
 99%|█████████▊| 7079/7175 [2:11:58<03:18,  2.07s/it]
 99%|█████████▊| 7080/7175 [2:12:00<03:16,  2.07s/it]
                                                     
{'loss': '2.97', 'grad_norm': '0.2238', 'learning_rate': '0.0006', 'epoch': '24.67'}
+
 99%|█████████▊| 7080/7175 [2:12:00<03:16,  2.07s/it]
 99%|█████████▊| 7081/7175 [2:12:02<03:14,  2.07s/it]
 99%|█████████▊| 7082/7175 [2:12:04<03:12,  2.07s/it]
 99%|█████████▊| 7083/7175 [2:12:06<03:09,  2.06s/it]
 99%|█████████▊| 7084/7175 [2:12:08<03:03,  2.01s/it]
 99%|█████████▊| 7085/7175 [2:12:09<02:49,  1.89s/it]
 99%|█████████▉| 7086/7175 [2:12:11<02:50,  1.92s/it]
 99%|█████████▉| 7087/7175 [2:12:14<02:52,  1.96s/it]
 99%|█████████▉| 7088/7175 [2:12:16<02:52,  1.99s/it]
 99%|█████████▉| 7089/7175 [2:12:18<02:52,  2.00s/it]
 99%|█████████▉| 7090/7175 [2:12:20<02:50,  2.00s/it]
                                                     
{'loss': '2.975', 'grad_norm': '0.2251', 'learning_rate': '0.0006', 'epoch': '24.71'}
+
 99%|█████████▉| 7090/7175 [2:12:20<02:50,  2.00s/it]
 99%|█████████▉| 7091/7175 [2:12:22<02:50,  2.03s/it]
 99%|█████████▉| 7092/7175 [2:12:24<02:49,  2.05s/it]
 99%|█████████▉| 7093/7175 [2:12:26<02:47,  2.05s/it]
 99%|█████████▉| 7094/7175 [2:12:28<02:45,  2.05s/it]
 99%|█████████▉| 7095/7175 [2:12:30<02:44,  2.05s/it]
 99%|█████████▉| 7096/7175 [2:12:32<02:41,  2.05s/it]
 99%|█████████▉| 7097/7175 [2:12:34<02:40,  2.05s/it]
 99%|█████████▉| 7098/7175 [2:12:36<02:38,  2.06s/it]
 99%|█████████▉| 7099/7175 [2:12:38<02:36,  2.06s/it]
 99%|█████████▉| 7100/7175 [2:12:40<02:34,  2.07s/it]
                                                     
{'loss': '2.968', 'grad_norm': '0.2116', 'learning_rate': '0.0006', 'epoch': '24.74'}
+
 99%|█████████▉| 7100/7175 [2:12:40<02:34,  2.07s/it]
 99%|█████████▉| 7101/7175 [2:12:42<02:33,  2.07s/it]
 99%|█████████▉| 7102/7175 [2:12:44<02:31,  2.07s/it]
 99%|█████████▉| 7103/7175 [2:12:46<02:28,  2.06s/it]
 99%|█████████▉| 7104/7175 [2:12:49<02:25,  2.05s/it]
 99%|█████████▉| 7105/7175 [2:12:51<02:23,  2.06s/it]
 99%|█████████▉| 7106/7175 [2:12:53<02:21,  2.05s/it]
 99%|█████████▉| 7107/7175 [2:12:55<02:19,  2.05s/it]
 99%|█████████▉| 7108/7175 [2:12:57<02:16,  2.04s/it]
 99%|█████████▉| 7109/7175 [2:12:59<02:13,  2.03s/it]
 99%|█████████▉| 7110/7175 [2:13:01<02:13,  2.05s/it]
                                                     
{'loss': '2.977', 'grad_norm': '0.2564', 'learning_rate': '0.0006', 'epoch': '24.78'}
+
 99%|█████████▉| 7110/7175 [2:13:01<02:13,  2.05s/it]
 99%|█████████▉| 7111/7175 [2:13:03<02:12,  2.06s/it]
 99%|█████████▉| 7112/7175 [2:13:05<02:10,  2.07s/it]
 99%|█████████▉| 7113/7175 [2:13:07<02:08,  2.07s/it]
 99%|█████████▉| 7114/7175 [2:13:09<02:05,  2.06s/it]
 99%|█████████▉| 7115/7175 [2:13:11<02:03,  2.06s/it]
 99%|█████████▉| 7116/7175 [2:13:13<02:01,  2.05s/it]
 99%|█████████▉| 7117/7175 [2:13:15<01:58,  2.05s/it]
 99%|█████████▉| 7118/7175 [2:13:17<01:56,  2.05s/it]
 99%|█████████▉| 7119/7175 [2:13:19<01:54,  2.05s/it]
 99%|█████████▉| 7120/7175 [2:13:21<01:52,  2.05s/it]
                                                     
{'loss': '2.979', 'grad_norm': '0.2108', 'learning_rate': '0.0006', 'epoch': '24.81'}
+
 99%|█████████▉| 7120/7175 [2:13:21<01:52,  2.05s/it]
 99%|█████████▉| 7121/7175 [2:13:23<01:51,  2.07s/it]
 99%|█████████▉| 7122/7175 [2:13:26<01:49,  2.07s/it]
 99%|█████████▉| 7123/7175 [2:13:28<01:47,  2.08s/it]
 99%|█████████▉| 7124/7175 [2:13:30<01:45,  2.08s/it]
 99%|█████████▉| 7125/7175 [2:13:32<01:43,  2.07s/it]
 99%|█████████▉| 7126/7175 [2:13:34<01:41,  2.07s/it]
 99%|█████████▉| 7127/7175 [2:13:36<01:38,  2.06s/it]
 99%|█████████▉| 7128/7175 [2:13:38<01:36,  2.06s/it][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  5%|▌         | 2/39 [00:00<00:04,  7.82it/s]
+
  8%|▊         | 3/39 [00:00<00:06,  5.50it/s]
+
 10%|█         | 4/39 [00:00<00:07,  4.63it/s]
+
 13%|█▎        | 5/39 [00:01<00:07,  4.26it/s]
+
 15%|█▌        | 6/39 [00:01<00:07,  4.14it/s]
+
 18%|█▊        | 7/39 [00:01<00:07,  4.04it/s]
+
 21%|██        | 8/39 [00:01<00:07,  3.94it/s]
+
 23%|██▎       | 9/39 [00:02<00:07,  3.85it/s]
+
 26%|██▌       | 10/39 [00:02<00:07,  3.83it/s]
+
 28%|██▊       | 11/39 [00:02<00:07,  3.79it/s]
+
 31%|███       | 12/39 [00:02<00:07,  3.77it/s]
+
 33%|███▎      | 13/39 [00:03<00:06,  3.78it/s]
+
 36%|███▌      | 14/39 [00:03<00:06,  3.80it/s]
+
 38%|███▊      | 15/39 [00:03<00:06,  3.74it/s]
+
 41%|████      | 16/39 [00:03<00:06,  3.72it/s]
+
 44%|████▎     | 17/39 [00:04<00:05,  3.75it/s]
+
 46%|████▌     | 18/39 [00:04<00:05,  3.70it/s]
+
 49%|████▊     | 19/39 [00:04<00:05,  3.69it/s]
+
 51%|█████▏    | 20/39 [00:05<00:05,  3.71it/s]
+
 54%|█████▍    | 21/39 [00:05<00:04,  3.71it/s]
+
 56%|█████▋    | 22/39 [00:05<00:04,  3.72it/s]
+
 59%|█████▉    | 23/39 [00:05<00:04,  3.71it/s]
+
 62%|██████▏   | 24/39 [00:06<00:04,  3.71it/s]
+
 64%|██████▍   | 25/39 [00:06<00:03,  3.77it/s]
+
 67%|██████▋   | 26/39 [00:06<00:03,  3.76it/s]
+
 69%|██████▉   | 27/39 [00:06<00:03,  3.77it/s]
+
 72%|███████▏  | 28/39 [00:07<00:02,  3.80it/s]
+
 74%|███████▍  | 29/39 [00:07<00:02,  3.81it/s]
+
 77%|███████▋  | 30/39 [00:07<00:02,  3.83it/s]
+
 79%|███████▉  | 31/39 [00:07<00:02,  3.83it/s]
+
 82%|████████▏ | 32/39 [00:08<00:01,  3.77it/s]
+
 85%|████████▍ | 33/39 [00:08<00:01,  3.80it/s]
+
 87%|████████▋ | 34/39 [00:08<00:01,  3.80it/s]
+
 90%|████████▉ | 35/39 [00:09<00:01,  3.86it/s]
+
 92%|█████████▏| 36/39 [00:09<00:00,  3.85it/s]
+
 95%|█████████▍| 37/39 [00:09<00:00,  3.81it/s]
+
 97%|█████████▋| 38/39 [00:09<00:00,  3.77it/s]
+
100%|██████████| 39/39 [00:10<00:00,  3.78it/s]
                                                     
+
                                               
{'eval_loss': '3.512', 'eval_runtime': '10.4', 'eval_samples_per_second': '234.7', 'eval_steps_per_second': '3.75', 'epoch': '24.84'}
+
 99%|█████████▉| 7128/7175 [2:13:48<01:36,  2.06s/it]
+
100%|██████████| 39/39 [00:10<00:00,  3.78it/s][eval] epoch 24.839 step 7128 loss 3.5122 (best 3.5035 @ ep 19.821, 20/20 since best)
+
+
                                               [transformers] 
+
+Training completed. Do not forget to share your model on huggingface.co/models =)
+
+
+
                                                     
{'train_runtime': '8030', 'train_samples_per_second': '456', 'train_steps_per_second': '0.893', 'train_loss': '3.42', 'epoch': '24.84'}
+
 99%|█████████▉| 7128/7175 [2:13:48<01:36,  2.06s/it]
 99%|█████████▉| 7128/7175 [2:13:48<00:52,  1.13s/it]
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+wandb: 
+wandb: 🚀 View run t2a_300M_C0_baseline at: https://wandb.ai/slowrun-wankers/data-aug-pretraining/runs/xarz21cw
+wandb: Find logs at: wandb/run-20260819_032604-xarz21cw/logs
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+[rank0]:[W819 05:40:41.983036568 ProcessGroupNCCL.cpp:1575] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator())
diff --git a/runs/t2a/t2a_37M_C0_baseline/checkpoint-792/config.json b/runs/t2a/t2a_37M_C0_baseline/checkpoint-792/config.json
new file mode 100644
index 0000000000000000000000000000000000000000..f8ca584a0751672f280da7c54ff3916cdf24d6f4
--- /dev/null
+++ b/runs/t2a/t2a_37M_C0_baseline/checkpoint-792/config.json
@@ -0,0 +1,32 @@
+{
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": null,
+  "dtype": "float32",
+  "eos_token_id": 151645,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 512,
+  "initializer_range": 0.02,
+  "intermediate_size": 1536,
+  "max_position_embeddings": 2048,
+  "mlp_bias": false,
+  "model_type": "llama",
+  "num_attention_heads": 4,
+  "num_hidden_layers": 20,
+  "num_key_value_heads": 4,
+  "pad_token_id": 151645,
+  "pretraining_tp": 1,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 10000.0,
+    "rope_type": "default"
+  },
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": false,
+  "vocab_size": 151671
+}
diff --git a/runs/t2a/t2a_37M_C1_rand15/checkpoint-1224/chat_template.jinja b/runs/t2a/t2a_37M_C1_rand15/checkpoint-1224/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..699ff8df401fe4788525e9c1f9b86a99eadd6230
--- /dev/null
+++ b/runs/t2a/t2a_37M_C1_rand15/checkpoint-1224/chat_template.jinja
@@ -0,0 +1,85 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0].role == 'system' %}
+        {{- messages[0].content + '\n\n' }}
+    {%- endif %}
+    {{- "# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0].content + '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
+{%- for message in messages[::-1] %}
+    {%- set index = (messages|length - 1) - loop.index0 %}
+    {%- if ns.multi_step_tool and message.role == "user" and not(message.content.startswith('<tool_response>') and message.content.endswith('</tool_response>')) %}
+        {%- set ns.multi_step_tool = false %}
+        {%- set ns.last_query_index = index %}
+    {%- endif %}
+{%- endfor %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {%- set content = message.content %}
+        {%- set reasoning_content = '' %}
+        {%- if message.reasoning_content is defined and message.reasoning_content is not none %}
+            {%- set reasoning_content = message.reasoning_content %}
+        {%- else %}
+            {%- if '</think>' in message.content %}
+                {%- set content = message.content.split('</think>')[-1].lstrip('\n') %}
+                {%- set reasoning_content = message.content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
+            {%- endif %}
+        {%- endif %}
+        {%- if loop.index0 > ns.last_query_index %}
+            {%- if loop.last or (not loop.last and reasoning_content) %}
+                {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content.strip('\n') + '\n</think>\n\n' + content.lstrip('\n') }}
+            {%- else %}
+                {{- '<|im_start|>' + message.role + '\n' + content }}
+            {%- endif %}
+        {%- else %}
+            {{- '<|im_start|>' + message.role + '\n' + content }}
+        {%- endif %}
+        {%- if message.tool_calls %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if (loop.first and content) or (not loop.first) %}
+                    {{- '\n' }}
+                {%- endif %}
+                {%- if tool_call.function %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {{- '<tool_call>\n{"name": "' }}
+                {{- tool_call.name }}
+                {{- '", "arguments": ' }}
+                {%- if tool_call.arguments is string %}
+                    {{- tool_call.arguments }}
+                {%- else %}
+                    {{- tool_call.arguments | tojson }}
+                {%- endif %}
+                {{- '}\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.first or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+    {%- if enable_thinking is defined and enable_thinking is false %}
+        {{- '<think>\n\n</think>\n\n' }}
+    {%- endif %}
+{%- endif %}
\ No newline at end of file
diff --git a/runs/t2a/t2a_37M_C1_rand15/checkpoint-1224/config.json b/runs/t2a/t2a_37M_C1_rand15/checkpoint-1224/config.json
new file mode 100644
index 0000000000000000000000000000000000000000..f8ca584a0751672f280da7c54ff3916cdf24d6f4
--- /dev/null
+++ b/runs/t2a/t2a_37M_C1_rand15/checkpoint-1224/config.json
@@ -0,0 +1,32 @@
+{
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": null,
+  "dtype": "float32",
+  "eos_token_id": 151645,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 512,
+  "initializer_range": 0.02,
+  "intermediate_size": 1536,
+  "max_position_embeddings": 2048,
+  "mlp_bias": false,
+  "model_type": "llama",
+  "num_attention_heads": 4,
+  "num_hidden_layers": 20,
+  "num_key_value_heads": 4,
+  "pad_token_id": 151645,
+  "pretraining_tp": 1,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 10000.0,
+    "rope_type": "default"
+  },
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": false,
+  "vocab_size": 151671
+}
diff --git a/runs/t2a/t2a_37M_C1_rand15/checkpoint-1224/generation_config.json b/runs/t2a/t2a_37M_C1_rand15/checkpoint-1224/generation_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..8c4242f32d40df256857f87264f646130f684e33
--- /dev/null
+++ b/runs/t2a/t2a_37M_C1_rand15/checkpoint-1224/generation_config.json
@@ -0,0 +1,11 @@
+{
+  "_from_model_config": true,
+  "eos_token_id": [
+    151645
+  ],
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "transformers_version": "5.15.0",
+  "use_cache": true
+}
diff --git a/runs/t2a/t2a_37M_C1_rand15/checkpoint-1224/tokenizer_config.json b/runs/t2a/t2a_37M_C1_rand15/checkpoint-1224/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..839a3743177a3ed77e1ffbae67520e8500ca0532
--- /dev/null
+++ b/runs/t2a/t2a_37M_C1_rand15/checkpoint-1224/tokenizer_config.json
@@ -0,0 +1,19 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|l2r_pred|>",
+    "<|r2l_pred|>"
+  ],
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 131072,
+  "pad_token": "<|im_end|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}
diff --git a/runs/t2a/t2a_37M_C1_rand15/checkpoint-1224/trainer_state.json b/runs/t2a/t2a_37M_C1_rand15/checkpoint-1224/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..37bef7833ab9c6a67b3254a8d3791cc4fa6e83c5
--- /dev/null
+++ b/runs/t2a/t2a_37M_C1_rand15/checkpoint-1224/trainer_state.json
@@ -0,0 +1,1031 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 34.0,
+  "eval_steps": 72,
+  "global_step": 1224,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.028268551236749116,
+      "grad_norm": 1.9435242414474487,
+      "learning_rate": 0.0,
+      "loss": 12.012422561645508,
+      "step": 1
+    },
+    {
+      "epoch": 0.2826855123674912,
+      "grad_norm": 1.7593657970428467,
+      "learning_rate": 5.399999999999999e-05,
+      "loss": 11.689207288953993,
+      "step": 10
+    },
+    {
+      "epoch": 0.5653710247349824,
+      "grad_norm": 1.6745258569717407,
+      "learning_rate": 0.00011399999999999999,
+      "loss": 10.962593841552735,
+      "step": 20
+    },
+    {
+      "epoch": 0.8480565371024735,
+      "grad_norm": 1.6189308166503906,
+      "learning_rate": 0.00017399999999999997,
+      "loss": 10.20816879272461,
+      "step": 30
+    },
+    {
+      "epoch": 1.1130742049469964,
+      "grad_norm": 1.4237518310546875,
+      "learning_rate": 0.000234,
+      "loss": 9.165228271484375,
+      "step": 40
+    },
+    {
+      "epoch": 1.3957597173144876,
+      "grad_norm": 0.7442731857299805,
+      "learning_rate": 0.000294,
+      "loss": 8.201170349121094,
+      "step": 50
+    },
+    {
+      "epoch": 1.6784452296819787,
+      "grad_norm": 0.2590465545654297,
+      "learning_rate": 0.00035399999999999993,
+      "loss": 7.70286865234375,
+      "step": 60
+    },
+    {
+      "epoch": 1.96113074204947,
+      "grad_norm": 0.37517449259757996,
+      "learning_rate": 0.0004139999999999999,
+      "loss": 7.561809539794922,
+      "step": 70
+    },
+    {
+      "epoch": 2.0,
+      "eval_loss": 7.520596981048584,
+      "eval_runtime": 27.7744,
+      "eval_samples_per_second": 87.887,
+      "eval_steps_per_second": 1.404,
+      "step": 72
+    },
+    {
+      "epoch": 2.2261484098939928,
+      "grad_norm": 0.8139720559120178,
+      "learning_rate": 0.000474,
+      "loss": 7.445631408691407,
+      "step": 80
+    },
+    {
+      "epoch": 2.508833922261484,
+      "grad_norm": 0.9102348685264587,
+      "learning_rate": 0.000534,
+      "loss": 7.280931091308593,
+      "step": 90
+    },
+    {
+      "epoch": 2.791519434628975,
+      "grad_norm": 0.7658299803733826,
+      "learning_rate": 0.0005939999999999999,
+      "loss": 7.13553466796875,
+      "step": 100
+    },
+    {
+      "epoch": 3.056537102473498,
+      "grad_norm": 0.4161047637462616,
+      "learning_rate": 0.0006,
+      "loss": 7.008511352539062,
+      "step": 110
+    },
+    {
+      "epoch": 3.3392226148409896,
+      "grad_norm": 0.5928809642791748,
+      "learning_rate": 0.0006,
+      "loss": 6.889783477783203,
+      "step": 120
+    },
+    {
+      "epoch": 3.6219081272084805,
+      "grad_norm": 1.145796298980713,
+      "learning_rate": 0.0006,
+      "loss": 6.782656860351563,
+      "step": 130
+    },
+    {
+      "epoch": 3.904593639575972,
+      "grad_norm": 0.528510332107544,
+      "learning_rate": 0.0006,
+      "loss": 6.717856597900391,
+      "step": 140
+    },
+    {
+      "epoch": 4.0,
+      "eval_loss": 6.781717300415039,
+      "eval_runtime": 3.2332,
+      "eval_samples_per_second": 754.985,
+      "eval_steps_per_second": 12.062,
+      "step": 144
+    },
+    {
+      "epoch": 4.169611307420495,
+      "grad_norm": 0.4031285047531128,
+      "learning_rate": 0.0006,
+      "loss": 6.618006896972656,
+      "step": 150
+    },
+    {
+      "epoch": 4.4522968197879855,
+      "grad_norm": 1.0299594402313232,
+      "learning_rate": 0.0006,
+      "loss": 6.570128631591797,
+      "step": 160
+    },
+    {
+      "epoch": 4.734982332155477,
+      "grad_norm": 0.28047454357147217,
+      "learning_rate": 0.0006,
+      "loss": 6.497016906738281,
+      "step": 170
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 0.9915837645530701,
+      "learning_rate": 0.0006,
+      "loss": 6.436015319824219,
+      "step": 180
+    },
+    {
+      "epoch": 5.282685512367491,
+      "grad_norm": 0.5041624307632446,
+      "learning_rate": 0.0006,
+      "loss": 6.406034851074219,
+      "step": 190
+    },
+    {
+      "epoch": 5.565371024734983,
+      "grad_norm": 0.2719726264476776,
+      "learning_rate": 0.0006,
+      "loss": 6.329710388183594,
+      "step": 200
+    },
+    {
+      "epoch": 5.848056537102473,
+      "grad_norm": 0.6014756560325623,
+      "learning_rate": 0.0006,
+      "loss": 6.2951713562011715,
+      "step": 210
+    },
+    {
+      "epoch": 6.0,
+      "eval_loss": 6.37517786026001,
+      "eval_runtime": 3.2468,
+      "eval_samples_per_second": 751.813,
+      "eval_steps_per_second": 12.012,
+      "step": 216
+    },
+    {
+      "epoch": 6.113074204946996,
+      "grad_norm": 0.6955145597457886,
+      "learning_rate": 0.0006,
+      "loss": 6.227598571777344,
+      "step": 220
+    },
+    {
+      "epoch": 6.395759717314488,
+      "grad_norm": 0.6271640658378601,
+      "learning_rate": 0.0006,
+      "loss": 6.220753479003906,
+      "step": 230
+    },
+    {
+      "epoch": 6.678445229681979,
+      "grad_norm": 0.2919083535671234,
+      "learning_rate": 0.0006,
+      "loss": 6.147015380859375,
+      "step": 240
+    },
+    {
+      "epoch": 6.96113074204947,
+      "grad_norm": 0.7793846130371094,
+      "learning_rate": 0.0006,
+      "loss": 6.0990653991699215,
+      "step": 250
+    },
+    {
+      "epoch": 7.226148409893993,
+      "grad_norm": 0.4031132757663727,
+      "learning_rate": 0.0006,
+      "loss": 6.054906845092773,
+      "step": 260
+    },
+    {
+      "epoch": 7.508833922261484,
+      "grad_norm": 0.8691778779029846,
+      "learning_rate": 0.0006,
+      "loss": 5.986634826660156,
+      "step": 270
+    },
+    {
+      "epoch": 7.791519434628976,
+      "grad_norm": 0.8808174133300781,
+      "learning_rate": 0.0006,
+      "loss": 5.985994720458985,
+      "step": 280
+    },
+    {
+      "epoch": 8.0,
+      "eval_loss": 6.045180320739746,
+      "eval_runtime": 3.251,
+      "eval_samples_per_second": 750.848,
+      "eval_steps_per_second": 11.996,
+      "step": 288
+    },
+    {
+      "epoch": 8.056537102473499,
+      "grad_norm": 0.34325718879699707,
+      "learning_rate": 0.0006,
+      "loss": 5.947795486450195,
+      "step": 290
+    },
+    {
+      "epoch": 8.33922261484099,
+      "grad_norm": 0.3509552776813507,
+      "learning_rate": 0.0006,
+      "loss": 5.856074523925781,
+      "step": 300
+    },
+    {
+      "epoch": 8.621908127208481,
+      "grad_norm": 0.9778263568878174,
+      "learning_rate": 0.0006,
+      "loss": 5.836148834228515,
+      "step": 310
+    },
+    {
+      "epoch": 8.904593639575971,
+      "grad_norm": 0.3471936881542206,
+      "learning_rate": 0.0006,
+      "loss": 5.850526428222656,
+      "step": 320
+    },
+    {
+      "epoch": 9.169611307420494,
+      "grad_norm": 0.6656075119972229,
+      "learning_rate": 0.0006,
+      "loss": 5.764811706542969,
+      "step": 330
+    },
+    {
+      "epoch": 9.452296819787986,
+      "grad_norm": 0.3785649836063385,
+      "learning_rate": 0.0006,
+      "loss": 5.722269058227539,
+      "step": 340
+    },
+    {
+      "epoch": 9.734982332155477,
+      "grad_norm": 0.640816867351532,
+      "learning_rate": 0.0006,
+      "loss": 5.698089599609375,
+      "step": 350
+    },
+    {
+      "epoch": 10.0,
+      "grad_norm": 0.5507412552833557,
+      "learning_rate": 0.0006,
+      "loss": 5.697988891601563,
+      "step": 360
+    },
+    {
+      "epoch": 10.0,
+      "eval_loss": 5.7081451416015625,
+      "eval_runtime": 3.2381,
+      "eval_samples_per_second": 753.847,
+      "eval_steps_per_second": 12.044,
+      "step": 360
+    },
+    {
+      "epoch": 10.282685512367491,
+      "grad_norm": 0.2801050841808319,
+      "learning_rate": 0.0006,
+      "loss": 5.614902496337891,
+      "step": 370
+    },
+    {
+      "epoch": 10.565371024734983,
+      "grad_norm": 0.36252981424331665,
+      "learning_rate": 0.0006,
+      "loss": 5.569301605224609,
+      "step": 380
+    },
+    {
+      "epoch": 10.848056537102474,
+      "grad_norm": 0.6907970309257507,
+      "learning_rate": 0.0006,
+      "loss": 5.55896224975586,
+      "step": 390
+    },
+    {
+      "epoch": 11.113074204946997,
+      "grad_norm": 0.596554160118103,
+      "learning_rate": 0.0006,
+      "loss": 5.538333129882813,
+      "step": 400
+    },
+    {
+      "epoch": 11.395759717314487,
+      "grad_norm": 0.5550813674926758,
+      "learning_rate": 0.0006,
+      "loss": 5.47388916015625,
+      "step": 410
+    },
+    {
+      "epoch": 11.678445229681978,
+      "grad_norm": 0.3134074807167053,
+      "learning_rate": 0.0006,
+      "loss": 5.451878356933594,
+      "step": 420
+    },
+    {
+      "epoch": 11.96113074204947,
+      "grad_norm": 0.6855193972587585,
+      "learning_rate": 0.0006,
+      "loss": 5.408074951171875,
+      "step": 430
+    },
+    {
+      "epoch": 12.0,
+      "eval_loss": 5.40466833114624,
+      "eval_runtime": 3.2602,
+      "eval_samples_per_second": 748.72,
+      "eval_steps_per_second": 11.962,
+      "step": 432
+    },
+    {
+      "epoch": 12.226148409893993,
+      "grad_norm": 0.4941120147705078,
+      "learning_rate": 0.0006,
+      "loss": 5.414258575439453,
+      "step": 440
+    },
+    {
+      "epoch": 12.508833922261484,
+      "grad_norm": 0.3882310688495636,
+      "learning_rate": 0.0006,
+      "loss": 5.360256195068359,
+      "step": 450
+    },
+    {
+      "epoch": 12.791519434628976,
+      "grad_norm": 0.2722965478897095,
+      "learning_rate": 0.0006,
+      "loss": 5.3089244842529295,
+      "step": 460
+    },
+    {
+      "epoch": 13.056537102473499,
+      "grad_norm": 0.2843002378940582,
+      "learning_rate": 0.0006,
+      "loss": 5.265041351318359,
+      "step": 470
+    },
+    {
+      "epoch": 13.33922261484099,
+      "grad_norm": 0.5850459933280945,
+      "learning_rate": 0.0006,
+      "loss": 5.233289337158203,
+      "step": 480
+    },
+    {
+      "epoch": 13.621908127208481,
+      "grad_norm": 0.8874235153198242,
+      "learning_rate": 0.0006,
+      "loss": 5.281147384643555,
+      "step": 490
+    },
+    {
+      "epoch": 13.904593639575971,
+      "grad_norm": 0.2751728296279907,
+      "learning_rate": 0.0006,
+      "loss": 5.273110961914062,
+      "step": 500
+    },
+    {
+      "epoch": 14.0,
+      "eval_loss": 5.117500305175781,
+      "eval_runtime": 3.2488,
+      "eval_samples_per_second": 751.356,
+      "eval_steps_per_second": 12.004,
+      "step": 504
+    },
+    {
+      "epoch": 14.169611307420494,
+      "grad_norm": 0.24036535620689392,
+      "learning_rate": 0.0006,
+      "loss": 5.167061614990234,
+      "step": 510
+    },
+    {
+      "epoch": 14.452296819787986,
+      "grad_norm": 0.6117097735404968,
+      "learning_rate": 0.0006,
+      "loss": 5.118548583984375,
+      "step": 520
+    },
+    {
+      "epoch": 14.734982332155477,
+      "grad_norm": 0.31319838762283325,
+      "learning_rate": 0.0006,
+      "loss": 5.109169006347656,
+      "step": 530
+    },
+    {
+      "epoch": 15.0,
+      "grad_norm": 0.3651270270347595,
+      "learning_rate": 0.0006,
+      "loss": 5.070779418945312,
+      "step": 540
+    },
+    {
+      "epoch": 15.282685512367491,
+      "grad_norm": 0.4581730365753174,
+      "learning_rate": 0.0006,
+      "loss": 5.021930694580078,
+      "step": 550
+    },
+    {
+      "epoch": 15.565371024734983,
+      "grad_norm": 0.4816060960292816,
+      "learning_rate": 0.0006,
+      "loss": 5.014722442626953,
+      "step": 560
+    },
+    {
+      "epoch": 15.848056537102474,
+      "grad_norm": 0.39352262020111084,
+      "learning_rate": 0.0006,
+      "loss": 4.996958160400391,
+      "step": 570
+    },
+    {
+      "epoch": 16.0,
+      "eval_loss": 4.801138877868652,
+      "eval_runtime": 3.2323,
+      "eval_samples_per_second": 755.184,
+      "eval_steps_per_second": 12.066,
+      "step": 576
+    },
+    {
+      "epoch": 16.113074204946997,
+      "grad_norm": 0.2454403191804886,
+      "learning_rate": 0.0006,
+      "loss": 4.9322021484375,
+      "step": 580
+    },
+    {
+      "epoch": 16.395759717314487,
+      "grad_norm": 0.36802002787590027,
+      "learning_rate": 0.0006,
+      "loss": 4.8944751739501955,
+      "step": 590
+    },
+    {
+      "epoch": 16.67844522968198,
+      "grad_norm": 0.5228356122970581,
+      "learning_rate": 0.0006,
+      "loss": 4.886128616333008,
+      "step": 600
+    },
+    {
+      "epoch": 16.96113074204947,
+      "grad_norm": 0.2989749610424042,
+      "learning_rate": 0.0006,
+      "loss": 4.879407119750977,
+      "step": 610
+    },
+    {
+      "epoch": 17.226148409893995,
+      "grad_norm": 0.4083768129348755,
+      "learning_rate": 0.0006,
+      "loss": 4.809102630615234,
+      "step": 620
+    },
+    {
+      "epoch": 17.508833922261484,
+      "grad_norm": 0.3537263870239258,
+      "learning_rate": 0.0006,
+      "loss": 4.807111358642578,
+      "step": 630
+    },
+    {
+      "epoch": 17.791519434628974,
+      "grad_norm": 0.3576992154121399,
+      "learning_rate": 0.0006,
+      "loss": 4.804788208007812,
+      "step": 640
+    },
+    {
+      "epoch": 18.0,
+      "eval_loss": 4.593893051147461,
+      "eval_runtime": 3.2611,
+      "eval_samples_per_second": 748.51,
+      "eval_steps_per_second": 11.959,
+      "step": 648
+    },
+    {
+      "epoch": 18.0565371024735,
+      "grad_norm": 0.39931267499923706,
+      "learning_rate": 0.0006,
+      "loss": 4.764279174804687,
+      "step": 650
+    },
+    {
+      "epoch": 18.33922261484099,
+      "grad_norm": 0.4623089134693146,
+      "learning_rate": 0.0006,
+      "loss": 4.723663330078125,
+      "step": 660
+    },
+    {
+      "epoch": 18.62190812720848,
+      "grad_norm": 0.38529741764068604,
+      "learning_rate": 0.0006,
+      "loss": 4.716490173339844,
+      "step": 670
+    },
+    {
+      "epoch": 18.90459363957597,
+      "grad_norm": 0.2664492428302765,
+      "learning_rate": 0.0006,
+      "loss": 4.695201873779297,
+      "step": 680
+    },
+    {
+      "epoch": 19.169611307420496,
+      "grad_norm": 0.2968577742576599,
+      "learning_rate": 0.0006,
+      "loss": 4.65149154663086,
+      "step": 690
+    },
+    {
+      "epoch": 19.452296819787986,
+      "grad_norm": 0.34656834602355957,
+      "learning_rate": 0.0006,
+      "loss": 4.619742584228516,
+      "step": 700
+    },
+    {
+      "epoch": 19.73498233215548,
+      "grad_norm": 0.3387688100337982,
+      "learning_rate": 0.0006,
+      "loss": 4.623148345947266,
+      "step": 710
+    },
+    {
+      "epoch": 20.0,
+      "grad_norm": 0.4601157307624817,
+      "learning_rate": 0.0006,
+      "loss": 4.637310791015625,
+      "step": 720
+    },
+    {
+      "epoch": 20.0,
+      "eval_loss": 4.474363803863525,
+      "eval_runtime": 3.2398,
+      "eval_samples_per_second": 753.439,
+      "eval_steps_per_second": 12.038,
+      "step": 720
+    },
+    {
+      "epoch": 20.28268551236749,
+      "grad_norm": 0.4556927978992462,
+      "learning_rate": 0.0006,
+      "loss": 4.572430801391602,
+      "step": 730
+    },
+    {
+      "epoch": 20.565371024734983,
+      "grad_norm": 0.3392789959907532,
+      "learning_rate": 0.0006,
+      "loss": 4.582429504394531,
+      "step": 740
+    },
+    {
+      "epoch": 20.848056537102472,
+      "grad_norm": 0.21901702880859375,
+      "learning_rate": 0.0006,
+      "loss": 4.5568183898925785,
+      "step": 750
+    },
+    {
+      "epoch": 21.113074204946997,
+      "grad_norm": 0.2501514256000519,
+      "learning_rate": 0.0006,
+      "loss": 4.528646087646484,
+      "step": 760
+    },
+    {
+      "epoch": 21.395759717314487,
+      "grad_norm": 0.280891478061676,
+      "learning_rate": 0.0006,
+      "loss": 4.492322158813477,
+      "step": 770
+    },
+    {
+      "epoch": 21.67844522968198,
+      "grad_norm": 0.3456341624259949,
+      "learning_rate": 0.0006,
+      "loss": 4.489433288574219,
+      "step": 780
+    },
+    {
+      "epoch": 21.96113074204947,
+      "grad_norm": 0.46443817019462585,
+      "learning_rate": 0.0006,
+      "loss": 4.494069671630859,
+      "step": 790
+    },
+    {
+      "epoch": 22.0,
+      "eval_loss": 4.342863082885742,
+      "eval_runtime": 3.2433,
+      "eval_samples_per_second": 752.633,
+      "eval_steps_per_second": 12.025,
+      "step": 792
+    },
+    {
+      "epoch": 22.226148409893995,
+      "grad_norm": 0.45519575476646423,
+      "learning_rate": 0.0006,
+      "loss": 4.467262268066406,
+      "step": 800
+    },
+    {
+      "epoch": 22.508833922261484,
+      "grad_norm": 0.3256106972694397,
+      "learning_rate": 0.0006,
+      "loss": 4.453234100341797,
+      "step": 810
+    },
+    {
+      "epoch": 22.791519434628974,
+      "grad_norm": 0.27822867035865784,
+      "learning_rate": 0.0006,
+      "loss": 4.4489891052246096,
+      "step": 820
+    },
+    {
+      "epoch": 23.0565371024735,
+      "grad_norm": 0.2985180914402008,
+      "learning_rate": 0.0006,
+      "loss": 4.424115371704102,
+      "step": 830
+    },
+    {
+      "epoch": 23.33922261484099,
+      "grad_norm": 0.38306093215942383,
+      "learning_rate": 0.0006,
+      "loss": 4.376193237304688,
+      "step": 840
+    },
+    {
+      "epoch": 23.62190812720848,
+      "grad_norm": 0.3414123058319092,
+      "learning_rate": 0.0006,
+      "loss": 4.3761547088623045,
+      "step": 850
+    },
+    {
+      "epoch": 23.90459363957597,
+      "grad_norm": 0.2492820769548416,
+      "learning_rate": 0.0006,
+      "loss": 4.388767242431641,
+      "step": 860
+    },
+    {
+      "epoch": 24.0,
+      "eval_loss": 4.2626824378967285,
+      "eval_runtime": 3.2466,
+      "eval_samples_per_second": 751.864,
+      "eval_steps_per_second": 12.013,
+      "step": 864
+    },
+    {
+      "epoch": 24.169611307420496,
+      "grad_norm": 0.31275811791419983,
+      "learning_rate": 0.0006,
+      "loss": 4.337926864624023,
+      "step": 870
+    },
+    {
+      "epoch": 24.452296819787986,
+      "grad_norm": 0.34109243750572205,
+      "learning_rate": 0.0006,
+      "loss": 4.322794342041016,
+      "step": 880
+    },
+    {
+      "epoch": 24.73498233215548,
+      "grad_norm": 0.4022921323776245,
+      "learning_rate": 0.0006,
+      "loss": 4.344395446777344,
+      "step": 890
+    },
+    {
+      "epoch": 25.0,
+      "grad_norm": 0.3676358461380005,
+      "learning_rate": 0.0006,
+      "loss": 4.3362274169921875,
+      "step": 900
+    },
+    {
+      "epoch": 25.28268551236749,
+      "grad_norm": 0.4200904071331024,
+      "learning_rate": 0.0006,
+      "loss": 4.272370529174805,
+      "step": 910
+    },
+    {
+      "epoch": 25.565371024734983,
+      "grad_norm": 0.3631523847579956,
+      "learning_rate": 0.0006,
+      "loss": 4.287709808349609,
+      "step": 920
+    },
+    {
+      "epoch": 25.848056537102472,
+      "grad_norm": 0.28186315298080444,
+      "learning_rate": 0.0006,
+      "loss": 4.280107879638672,
+      "step": 930
+    },
+    {
+      "epoch": 26.0,
+      "eval_loss": 4.21533203125,
+      "eval_runtime": 3.2558,
+      "eval_samples_per_second": 749.742,
+      "eval_steps_per_second": 11.979,
+      "step": 936
+    },
+    {
+      "epoch": 26.113074204946997,
+      "grad_norm": 0.34260958433151245,
+      "learning_rate": 0.0006,
+      "loss": 4.2594554901123045,
+      "step": 940
+    },
+    {
+      "epoch": 26.395759717314487,
+      "grad_norm": 0.2899376153945923,
+      "learning_rate": 0.0006,
+      "loss": 4.222721099853516,
+      "step": 950
+    },
+    {
+      "epoch": 26.67844522968198,
+      "grad_norm": 0.2744390666484833,
+      "learning_rate": 0.0006,
+      "loss": 4.229715728759766,
+      "step": 960
+    },
+    {
+      "epoch": 26.96113074204947,
+      "grad_norm": 0.30746665596961975,
+      "learning_rate": 0.0006,
+      "loss": 4.24095344543457,
+      "step": 970
+    },
+    {
+      "epoch": 27.226148409893995,
+      "grad_norm": 0.38828665018081665,
+      "learning_rate": 0.0006,
+      "loss": 4.19915542602539,
+      "step": 980
+    },
+    {
+      "epoch": 27.508833922261484,
+      "grad_norm": 0.34221720695495605,
+      "learning_rate": 0.0006,
+      "loss": 4.187008285522461,
+      "step": 990
+    },
+    {
+      "epoch": 27.791519434628974,
+      "grad_norm": 0.29330554604530334,
+      "learning_rate": 0.0006,
+      "loss": 4.196577072143555,
+      "step": 1000
+    },
+    {
+      "epoch": 28.0,
+      "eval_loss": 4.183407783508301,
+      "eval_runtime": 3.2429,
+      "eval_samples_per_second": 752.713,
+      "eval_steps_per_second": 12.026,
+      "step": 1008
+    },
+    {
+      "epoch": 28.0565371024735,
+      "grad_norm": 0.352918416261673,
+      "learning_rate": 0.0006,
+      "loss": 4.186087799072266,
+      "step": 1010
+    },
+    {
+      "epoch": 28.33922261484099,
+      "grad_norm": 0.33987584710121155,
+      "learning_rate": 0.0006,
+      "loss": 4.145789718627929,
+      "step": 1020
+    },
+    {
+      "epoch": 28.62190812720848,
+      "grad_norm": 0.3465867042541504,
+      "learning_rate": 0.0006,
+      "loss": 4.147980499267578,
+      "step": 1030
+    },
+    {
+      "epoch": 28.90459363957597,
+      "grad_norm": 0.2727724015712738,
+      "learning_rate": 0.0006,
+      "loss": 4.156186294555664,
+      "step": 1040
+    },
+    {
+      "epoch": 29.169611307420496,
+      "grad_norm": 0.36561259627342224,
+      "learning_rate": 0.0006,
+      "loss": 4.114311218261719,
+      "step": 1050
+    },
+    {
+      "epoch": 29.452296819787986,
+      "grad_norm": 0.4864152669906616,
+      "learning_rate": 0.0006,
+      "loss": 4.119881057739258,
+      "step": 1060
+    },
+    {
+      "epoch": 29.73498233215548,
+      "grad_norm": 0.29262229800224304,
+      "learning_rate": 0.0006,
+      "loss": 4.122491836547852,
+      "step": 1070
+    },
+    {
+      "epoch": 30.0,
+      "grad_norm": 0.3393446207046509,
+      "learning_rate": 0.0006,
+      "loss": 4.119888687133789,
+      "step": 1080
+    },
+    {
+      "epoch": 30.0,
+      "eval_loss": 4.1612467765808105,
+      "eval_runtime": 3.2402,
+      "eval_samples_per_second": 753.348,
+      "eval_steps_per_second": 12.036,
+      "step": 1080
+    },
+    {
+      "epoch": 30.28268551236749,
+      "grad_norm": 0.3970426321029663,
+      "learning_rate": 0.0006,
+      "loss": 4.058244323730468,
+      "step": 1090
+    },
+    {
+      "epoch": 30.565371024734983,
+      "grad_norm": 0.31837108731269836,
+      "learning_rate": 0.0006,
+      "loss": 4.070771789550781,
+      "step": 1100
+    },
+    {
+      "epoch": 30.848056537102472,
+      "grad_norm": 0.26694807410240173,
+      "learning_rate": 0.0006,
+      "loss": 4.077159118652344,
+      "step": 1110
+    },
+    {
+      "epoch": 31.113074204946997,
+      "grad_norm": 0.2974282205104828,
+      "learning_rate": 0.0006,
+      "loss": 4.0485107421875,
+      "step": 1120
+    },
+    {
+      "epoch": 31.395759717314487,
+      "grad_norm": 0.3916138708591461,
+      "learning_rate": 0.0006,
+      "loss": 4.019329452514649,
+      "step": 1130
+    },
+    {
+      "epoch": 31.67844522968198,
+      "grad_norm": 0.5596138834953308,
+      "learning_rate": 0.0006,
+      "loss": 4.059733581542969,
+      "step": 1140
+    },
+    {
+      "epoch": 31.96113074204947,
+      "grad_norm": 0.3571491539478302,
+      "learning_rate": 0.0006,
+      "loss": 4.074971771240234,
+      "step": 1150
+    },
+    {
+      "epoch": 32.0,
+      "eval_loss": 4.163784027099609,
+      "eval_runtime": 3.2447,
+      "eval_samples_per_second": 752.296,
+      "eval_steps_per_second": 12.019,
+      "step": 1152
+    },
+    {
+      "epoch": 32.226148409893995,
+      "grad_norm": 0.301995187997818,
+      "learning_rate": 0.0006,
+      "loss": 4.007966613769531,
+      "step": 1160
+    },
+    {
+      "epoch": 32.50883392226148,
+      "grad_norm": 0.25756651163101196,
+      "learning_rate": 0.0006,
+      "loss": 3.9878776550292967,
+      "step": 1170
+    },
+    {
+      "epoch": 32.791519434628974,
+      "grad_norm": 0.2359946072101593,
+      "learning_rate": 0.0006,
+      "loss": 4.008602905273437,
+      "step": 1180
+    },
+    {
+      "epoch": 33.0565371024735,
+      "grad_norm": 0.3307354152202606,
+      "learning_rate": 0.0006,
+      "loss": 3.9916606903076173,
+      "step": 1190
+    },
+    {
+      "epoch": 33.33922261484099,
+      "grad_norm": 0.34522730112075806,
+      "learning_rate": 0.0006,
+      "loss": 3.940735626220703,
+      "step": 1200
+    },
+    {
+      "epoch": 33.62190812720848,
+      "grad_norm": 0.34154391288757324,
+      "learning_rate": 0.0006,
+      "loss": 3.9729019165039063,
+      "step": 1210
+    },
+    {
+      "epoch": 33.90459363957597,
+      "grad_norm": 0.3281477987766266,
+      "learning_rate": 0.0006,
+      "loss": 3.9945247650146483,
+      "step": 1220
+    },
+    {
+      "epoch": 34.0,
+      "eval_loss": 4.151440620422363,
+      "eval_runtime": 3.2619,
+      "eval_samples_per_second": 748.347,
+      "eval_steps_per_second": 11.956,
+      "step": 1224
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 7308,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 203,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": false,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 5.159095118824735e+17,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/runs/t2a/t2a_37M_C1_rand15/train.log b/runs/t2a/t2a_37M_C1_rand15/train.log
new file mode 100644
index 0000000000000000000000000000000000000000..363770efe4f526c0fe83cd16e314452169a2c4a4
--- /dev/null
+++ b/runs/t2a/t2a_37M_C1_rand15/train.log
@@ -0,0 +1,2410 @@
+W0818 19:32:21.651000 2998827 torch/distributed/run.py:851] 
+W0818 19:32:21.651000 2998827 torch/distributed/run.py:851] *****************************************
+W0818 19:32:21.651000 2998827 torch/distributed/run.py:851] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+W0818 19:32:21.651000 2998827 torch/distributed/run.py:851] *****************************************
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+total params: 145833984
+[dataset] loaded packed cache: 18063 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+total params: 145833984
+total params: 145833984
+[dataset] loaded packed cache: 18063 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[dataset] loaded packed cache: 18063 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+total params: 145833984
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+total params: 145833984
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+[dataset] loaded packed cache: 18063 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[dataset] loaded packed cache: 18063 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+total params: 145833984
+[dataset] loaded packed cache: 18063 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+total params: 145833984
+[dataset] loaded packed cache: 18063 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+total params: 145833984
+[dataset] loaded packed cache: 18063 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+[transformers] ***** Running training *****
+[transformers]   Num examples = 18,063
+[transformers]   Num Epochs = 203
+[transformers]   Num update steps per epoch = 36
+[transformers]   Instantaneous batch size per device = 8
+[transformers]   Total train batch size (w. parallel, distributed & accumulation) = 512
+[transformers]   Gradient Accumulation steps = 8
+[transformers]   Total optimization steps = 7,308
+[transformers]   Number of trainable parameters = 145,833,984
+wandb: WARNING Changes to your `wandb` environment variables will be ignored because your `wandb` session has already started. For more information on how to modify your settings with `wandb.init()` arguments, please refer to https://wandb.me/wandb-init.
+wandb: Tracking run with wandb version 0.28.2
+wandb: Run data is saved locally in /home/ubuntu/workspace/data-augmentations-for-pretraining/wandb/run-20260818_193241-n0ad7csv
+wandb: Run `wandb offline` to turn off syncing.
+wandb: Syncing run t2a_37M_C1_rand15
+wandb: ⭐️ View project at https://wandb.ai/slowrun-wankers/data-aug-pretraining
+wandb: 🚀 View run at https://wandb.ai/slowrun-wankers/data-aug-pretraining/runs/n0ad7csv
+
  0%|          | 0/7308 [00:00<?, ?it/s]wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+[rank7]:W0818 19:33:05.113000 2998947 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+[rank2]:W0818 19:33:05.129000 2998942 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+[rank4]:W0818 19:33:05.468000 2998944 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+[rank3]:W0818 19:33:05.523000 2998943 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+[rank5]:W0818 19:33:05.605000 2998945 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+[rank6]:W0818 19:33:05.634000 2998946 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+[rank1]:W0818 19:33:05.971000 2998941 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+[rank0]:W0818 19:33:06.510000 2998940 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+
  0%|          | 1/7308 [00:41<84:23:18, 41.58s/it]
                                                   
{'loss': '12.01', 'grad_norm': '1.944', 'learning_rate': '0', 'epoch': '0.02827'}
+
  0%|          | 1/7308 [00:41<84:23:18, 41.58s/it]
  0%|          | 2/7308 [00:42<35:32:45, 17.52s/it]
  0%|          | 3/7308 [00:42<19:54:57,  9.81s/it]
  0%|          | 4/7308 [00:43<12:34:25,  6.20s/it]
  0%|          | 5/7308 [00:44<8:30:54,  4.20s/it] 
  0%|          | 6/7308 [00:44<6:04:14,  2.99s/it]
  0%|          | 7/7308 [00:45<4:31:06,  2.23s/it]
  0%|          | 8/7308 [00:46<3:30:08,  1.73s/it]
  0%|          | 9/7308 [00:46<2:49:17,  1.39s/it]
  0%|          | 10/7308 [00:47<2:21:38,  1.16s/it]
                                                   
{'loss': '11.69', 'grad_norm': '1.759', 'learning_rate': '5.4e-05', 'epoch': '0.2827'}
+
  0%|          | 10/7308 [00:47<2:21:38,  1.16s/it]
  0%|          | 11/7308 [00:48<2:02:40,  1.01s/it]
  0%|          | 12/7308 [00:48<1:49:31,  1.11it/s]
  0%|          | 13/7308 [00:49<1:40:28,  1.21it/s]
  0%|          | 14/7308 [00:50<1:34:05,  1.29it/s]
  0%|          | 15/7308 [00:50<1:29:37,  1.36it/s]
  0%|          | 16/7308 [00:51<1:26:37,  1.40it/s]
  0%|          | 17/7308 [00:52<1:24:28,  1.44it/s]
  0%|          | 18/7308 [00:52<1:23:01,  1.46it/s]
  0%|          | 19/7308 [00:53<1:21:51,  1.48it/s]
  0%|          | 20/7308 [00:54<1:21:16,  1.49it/s]
                                                   
{'loss': '10.96', 'grad_norm': '1.675', 'learning_rate': '0.000114', 'epoch': '0.5654'}
+
  0%|          | 20/7308 [00:54<1:21:16,  1.49it/s]
  0%|          | 21/7308 [00:54<1:20:48,  1.50it/s]
  0%|          | 22/7308 [00:55<1:20:14,  1.51it/s]
  0%|          | 23/7308 [00:55<1:20:01,  1.52it/s]
  0%|          | 24/7308 [00:56<1:19:48,  1.52it/s]
  0%|          | 25/7308 [00:57<1:19:43,  1.52it/s]
  0%|          | 26/7308 [00:57<1:19:26,  1.53it/s]
  0%|          | 27/7308 [00:58<1:19:13,  1.53it/s]
  0%|          | 28/7308 [00:59<1:19:03,  1.53it/s]
  0%|          | 29/7308 [00:59<1:18:57,  1.54it/s]
  0%|          | 30/7308 [01:00<1:18:56,  1.54it/s]
                                                   
{'loss': '10.21', 'grad_norm': '1.619', 'learning_rate': '0.000174', 'epoch': '0.8481'}
+
  0%|          | 30/7308 [01:00<1:18:56,  1.54it/s]
  0%|          | 31/7308 [01:01<1:18:58,  1.54it/s]
  0%|          | 32/7308 [01:01<1:18:54,  1.54it/s]
  0%|          | 33/7308 [01:02<1:18:57,  1.54it/s]
  0%|          | 34/7308 [01:03<1:18:52,  1.54it/s]
  0%|          | 35/7308 [01:03<1:18:47,  1.54it/s]
  0%|          | 36/7308 [01:04<1:04:12,  1.89it/s]
  1%|          | 37/7308 [01:04<1:11:07,  1.70it/s]
  1%|          | 38/7308 [01:05<1:13:21,  1.65it/s]
  1%|          | 39/7308 [01:06<1:14:58,  1.62it/s]
  1%|          | 40/7308 [01:06<1:16:07,  1.59it/s]
                                                   
{'loss': '9.165', 'grad_norm': '1.424', 'learning_rate': '0.000234', 'epoch': '1.113'}
+
  1%|          | 40/7308 [01:06<1:16:07,  1.59it/s]
  1%|          | 41/7308 [01:07<1:16:54,  1.57it/s]
  1%|          | 42/7308 [01:08<1:17:26,  1.56it/s]
  1%|          | 43/7308 [01:08<1:17:47,  1.56it/s]
  1%|          | 44/7308 [01:09<1:18:00,  1.55it/s]
  1%|          | 45/7308 [01:09<1:18:13,  1.55it/s]
  1%|          | 46/7308 [01:10<1:18:27,  1.54it/s]
  1%|          | 47/7308 [01:11<1:18:29,  1.54it/s]
  1%|          | 48/7308 [01:11<1:18:37,  1.54it/s]
  1%|          | 49/7308 [01:12<1:18:41,  1.54it/s]
  1%|          | 50/7308 [01:13<1:18:41,  1.54it/s]
                                                   
{'loss': '8.201', 'grad_norm': '0.7443', 'learning_rate': '0.000294', 'epoch': '1.396'}
+
  1%|          | 50/7308 [01:13<1:18:41,  1.54it/s]
  1%|          | 51/7308 [01:13<1:18:44,  1.54it/s]
  1%|          | 52/7308 [01:14<1:18:47,  1.53it/s]
  1%|          | 53/7308 [01:15<1:18:54,  1.53it/s]
  1%|          | 54/7308 [01:15<1:18:53,  1.53it/s]
  1%|          | 55/7308 [01:16<1:18:54,  1.53it/s]
  1%|          | 56/7308 [01:17<1:18:57,  1.53it/s]
  1%|          | 57/7308 [01:17<1:18:57,  1.53it/s]
  1%|          | 58/7308 [01:18<1:18:57,  1.53it/s]
  1%|          | 59/7308 [01:19<1:19:02,  1.53it/s]
  1%|          | 60/7308 [01:19<1:19:10,  1.53it/s]
                                                   
{'loss': '7.703', 'grad_norm': '0.259', 'learning_rate': '0.000354', 'epoch': '1.678'}
+
  1%|          | 60/7308 [01:19<1:19:10,  1.53it/s]
  1%|          | 61/7308 [01:20<1:19:07,  1.53it/s]
  1%|          | 62/7308 [01:21<1:19:03,  1.53it/s]
  1%|          | 63/7308 [01:21<1:19:00,  1.53it/s]
  1%|          | 64/7308 [01:22<1:18:56,  1.53it/s]
  1%|          | 65/7308 [01:23<1:18:53,  1.53it/s]
  1%|          | 66/7308 [01:23<1:18:54,  1.53it/s]
  1%|          | 67/7308 [01:24<1:18:51,  1.53it/s]
  1%|          | 68/7308 [01:24<1:18:46,  1.53it/s]
  1%|          | 69/7308 [01:25<1:18:42,  1.53it/s]
  1%|          | 70/7308 [01:26<1:18:40,  1.53it/s]
                                                   
{'loss': '7.562', 'grad_norm': '0.3752', 'learning_rate': '0.000414', 'epoch': '1.961'}
+
  1%|          | 70/7308 [01:26<1:18:40,  1.53it/s]
  1%|          | 71/7308 [01:26<1:18:43,  1.53it/s]
  1%|          | 72/7308 [01:27<1:04:06,  1.88it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 27.30it/s]
+
 15%|█▌        | 6/39 [00:00<00:02, 15.77it/s]
+
 21%|██        | 8/39 [00:00<00:02, 14.31it/s]
+
 26%|██▌       | 10/39 [00:00<00:02, 13.55it/s]
+
 31%|███       | 12/39 [00:00<00:02, 13.08it/s]
+
 36%|███▌      | 14/39 [00:01<00:01, 12.79it/s]
+
 41%|████      | 16/39 [00:01<00:01, 12.59it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 12.46it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 12.38it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 12.34it/s]
+
 62%|██████▏   | 24/39 [00:01<00:01, 12.33it/s]
+
 67%|██████▋   | 26/39 [00:01<00:01, 12.25it/s]
+
 72%|███████▏  | 28/39 [00:02<00:00, 12.24it/s]
+
 77%|███████▋  | 30/39 [00:02<00:00, 12.23it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 12.23it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 12.22it/s]
+
 92%|█████████▏| 36/39 [00:02<00:00, 12.22it/s]
+
 97%|█████████▋| 38/39 [00:02<00:00, 12.20it/s]
                                                   
+
                                               
{'eval_loss': '7.521', 'eval_runtime': '27.77', 'eval_samples_per_second': '87.89', 'eval_steps_per_second': '1.404', 'epoch': '2'}
+
  1%|          | 72/7308 [01:54<1:04:06,  1.88it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.20it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-72
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-72/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-72/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.70it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.69it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-72/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-72/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-72/tokenizer_config.json
+[eval] epoch 2.0 step 72 loss 7.5206 (best 7.5206 @ ep 2.0, 0/20 since best)
+
+
                                               
  1%|          | 73/7308 [01:57<19:07:27,  9.52s/it]
  1%|          | 74/7308 [01:58<13:46:41,  6.86s/it]
  1%|          | 75/7308 [01:58<10:02:15,  5.00s/it]
  1%|          | 76/7308 [01:59<7:25:06,  3.69s/it] 
  1%|          | 77/7308 [02:00<5:35:10,  2.78s/it]
  1%|          | 78/7308 [02:00<4:18:13,  2.14s/it]
  1%|          | 79/7308 [02:01<3:24:18,  1.70s/it]
  1%|          | 80/7308 [02:02<2:46:41,  1.38s/it]
                                                   
{'loss': '7.446', 'grad_norm': '0.814', 'learning_rate': '0.000474', 'epoch': '2.226'}
+
  1%|          | 80/7308 [02:02<2:46:41,  1.38s/it]
  1%|          | 81/7308 [02:02<2:20:24,  1.17s/it]
  1%|          | 82/7308 [02:03<2:01:53,  1.01s/it]
  1%|          | 83/7308 [02:04<1:48:55,  1.11it/s]
  1%|          | 84/7308 [02:04<1:39:51,  1.21it/s]
  1%|          | 85/7308 [02:05<1:33:28,  1.29it/s]
  1%|          | 86/7308 [02:06<1:29:00,  1.35it/s]
  1%|          | 87/7308 [02:06<1:25:55,  1.40it/s]
  1%|          | 88/7308 [02:07<1:23:45,  1.44it/s]
  1%|          | 89/7308 [02:08<1:22:09,  1.46it/s]
  1%|          | 90/7308 [02:08<1:21:03,  1.48it/s]
                                                   
{'loss': '7.281', 'grad_norm': '0.9102', 'learning_rate': '0.000534', 'epoch': '2.509'}
+
  1%|          | 90/7308 [02:08<1:21:03,  1.48it/s]
  1%|          | 91/7308 [02:09<1:20:20,  1.50it/s]
  1%|▏         | 92/7308 [02:10<1:19:47,  1.51it/s]
  1%|▏         | 93/7308 [02:10<1:19:24,  1.51it/s]
  1%|▏         | 94/7308 [02:11<1:19:11,  1.52it/s]
  1%|▏         | 95/7308 [02:12<1:18:58,  1.52it/s]
  1%|▏         | 96/7308 [02:12<1:18:47,  1.53it/s]
  1%|▏         | 97/7308 [02:13<1:18:42,  1.53it/s]
  1%|▏         | 98/7308 [02:14<1:18:37,  1.53it/s]
  1%|▏         | 99/7308 [02:14<1:18:35,  1.53it/s]
  1%|▏         | 100/7308 [02:15<1:18:32,  1.53it/s]
                                                    
{'loss': '7.136', 'grad_norm': '0.7658', 'learning_rate': '0.000594', 'epoch': '2.792'}
+
  1%|▏         | 100/7308 [02:15<1:18:32,  1.53it/s]
  1%|▏         | 101/7308 [02:15<1:18:32,  1.53it/s]
  1%|▏         | 102/7308 [02:16<1:18:31,  1.53it/s]
  1%|▏         | 103/7308 [02:17<1:18:27,  1.53it/s]
  1%|▏         | 104/7308 [02:17<1:18:28,  1.53it/s]
  1%|▏         | 105/7308 [02:18<1:18:27,  1.53it/s]
  1%|▏         | 106/7308 [02:19<1:18:27,  1.53it/s]
  1%|▏         | 107/7308 [02:19<1:18:23,  1.53it/s]
  1%|▏         | 108/7308 [02:20<1:03:55,  1.88it/s]
  1%|▏         | 109/7308 [02:20<1:10:45,  1.70it/s]
  2%|▏         | 110/7308 [02:21<1:13:09,  1.64it/s]
                                                    
{'loss': '7.009', 'grad_norm': '0.4161', 'learning_rate': '0.0006', 'epoch': '3.057'}
+
  2%|▏         | 110/7308 [02:21<1:13:09,  1.64it/s]
  2%|▏         | 111/7308 [02:22<1:14:48,  1.60it/s]
  2%|▏         | 112/7308 [02:22<1:15:51,  1.58it/s]
  2%|▏         | 113/7308 [02:23<1:16:36,  1.57it/s]
  2%|▏         | 114/7308 [02:24<1:17:10,  1.55it/s]
  2%|▏         | 115/7308 [02:24<1:17:28,  1.55it/s]
  2%|▏         | 116/7308 [02:25<1:17:42,  1.54it/s]
  2%|▏         | 117/7308 [02:26<1:17:54,  1.54it/s]
  2%|▏         | 118/7308 [02:26<1:18:00,  1.54it/s]
  2%|▏         | 119/7308 [02:27<1:18:04,  1.53it/s]
  2%|▏         | 120/7308 [02:28<1:18:13,  1.53it/s]
                                                    
{'loss': '6.89', 'grad_norm': '0.5929', 'learning_rate': '0.0006', 'epoch': '3.339'}
+
  2%|▏         | 120/7308 [02:28<1:18:13,  1.53it/s]
  2%|▏         | 121/7308 [02:28<1:18:20,  1.53it/s]
  2%|▏         | 122/7308 [02:29<1:18:16,  1.53it/s]
  2%|▏         | 123/7308 [02:30<1:18:17,  1.53it/s]
  2%|▏         | 124/7308 [02:30<1:18:20,  1.53it/s]
  2%|▏         | 125/7308 [02:31<1:18:18,  1.53it/s]
  2%|▏         | 126/7308 [02:31<1:18:23,  1.53it/s]
  2%|▏         | 127/7308 [02:32<1:18:21,  1.53it/s]
  2%|▏         | 128/7308 [02:33<1:18:20,  1.53it/s]
  2%|▏         | 129/7308 [02:33<1:18:17,  1.53it/s]
  2%|▏         | 130/7308 [02:34<1:18:12,  1.53it/s]
                                                    
{'loss': '6.783', 'grad_norm': '1.146', 'learning_rate': '0.0006', 'epoch': '3.622'}
+
  2%|▏         | 130/7308 [02:34<1:18:12,  1.53it/s]
  2%|▏         | 131/7308 [02:35<1:18:21,  1.53it/s]
  2%|▏         | 132/7308 [02:35<1:18:26,  1.52it/s]
  2%|▏         | 133/7308 [02:36<1:18:18,  1.53it/s]
  2%|▏         | 134/7308 [02:37<1:18:14,  1.53it/s]
  2%|▏         | 135/7308 [02:37<1:18:13,  1.53it/s]
  2%|▏         | 136/7308 [02:38<1:18:14,  1.53it/s]
  2%|▏         | 137/7308 [02:39<1:18:10,  1.53it/s]
  2%|▏         | 138/7308 [02:39<1:18:11,  1.53it/s]
  2%|▏         | 139/7308 [02:40<1:18:09,  1.53it/s]
  2%|▏         | 140/7308 [02:41<1:18:09,  1.53it/s]
                                                    
{'loss': '6.718', 'grad_norm': '0.5285', 'learning_rate': '0.0006', 'epoch': '3.905'}
+
  2%|▏         | 140/7308 [02:41<1:18:09,  1.53it/s]
  2%|▏         | 141/7308 [02:41<1:18:08,  1.53it/s]
  2%|▏         | 142/7308 [02:42<1:18:09,  1.53it/s]
  2%|▏         | 143/7308 [02:43<1:18:05,  1.53it/s]
  2%|▏         | 144/7308 [02:43<1:03:36,  1.88it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 19.16it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 15.01it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.71it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.13it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.81it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.61it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.49it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.41it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.35it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.32it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.31it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.28it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.26it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.26it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.26it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.25it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.26it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.23it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s]
                                                    
+
                                               
{'eval_loss': '6.782', 'eval_runtime': '3.233', 'eval_samples_per_second': '755', 'eval_steps_per_second': '12.06', 'epoch': '4'}
+
  2%|▏         | 144/7308 [02:46<1:03:36,  1.88it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-144
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-144/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-144/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.94it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.94it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-144/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-144/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-144/tokenizer_config.json
+[eval] epoch 4.0 step 144 loss 6.7817 (best 6.7817 @ ep 4.0, 0/20 since best)
+
+
                                               
  2%|▏         | 145/7308 [02:49<4:18:21,  2.16s/it]
  2%|▏         | 146/7308 [02:49<3:24:09,  1.71s/it]
  2%|▏         | 147/7308 [02:50<2:46:20,  1.39s/it]
  2%|▏         | 148/7308 [02:51<2:19:48,  1.17s/it]
  2%|▏         | 149/7308 [02:51<2:01:11,  1.02s/it]
  2%|▏         | 150/7308 [02:52<1:48:15,  1.10it/s]
                                                    
{'loss': '6.618', 'grad_norm': '0.4031', 'learning_rate': '0.0006', 'epoch': '4.17'}
+
  2%|▏         | 150/7308 [02:52<1:48:15,  1.10it/s]
  2%|▏         | 151/7308 [02:53<1:55:06,  1.04it/s]
  2%|▏         | 152/7308 [02:54<1:43:58,  1.15it/s]
  2%|▏         | 153/7308 [02:55<1:36:05,  1.24it/s]
  2%|▏         | 154/7308 [02:55<1:30:38,  1.32it/s]
  2%|▏         | 155/7308 [02:56<1:26:47,  1.37it/s]
  2%|▏         | 156/7308 [02:56<1:24:07,  1.42it/s]
  2%|▏         | 157/7308 [02:57<1:22:14,  1.45it/s]
  2%|▏         | 158/7308 [02:58<1:20:52,  1.47it/s]
  2%|▏         | 159/7308 [02:58<1:19:59,  1.49it/s]
  2%|▏         | 160/7308 [02:59<1:19:20,  1.50it/s]
                                                    
{'loss': '6.57', 'grad_norm': '1.03', 'learning_rate': '0.0006', 'epoch': '4.452'}
+
  2%|▏         | 160/7308 [02:59<1:19:20,  1.50it/s]
  2%|▏         | 161/7308 [03:00<1:34:27,  1.26it/s]
  2%|▏         | 162/7308 [03:01<1:29:28,  1.33it/s]
  2%|▏         | 163/7308 [03:02<1:42:00,  1.17it/s]
  2%|▏         | 164/7308 [03:03<1:34:50,  1.26it/s]
  2%|▏         | 165/7308 [03:03<1:29:43,  1.33it/s]
  2%|▏         | 166/7308 [03:04<1:41:59,  1.17it/s]
  2%|▏         | 167/7308 [03:05<1:34:45,  1.26it/s]
  2%|▏         | 168/7308 [03:06<1:46:14,  1.12it/s]
  2%|▏         | 169/7308 [03:07<1:37:46,  1.22it/s]
  2%|▏         | 170/7308 [03:08<1:54:41,  1.04it/s]
                                                    
{'loss': '6.497', 'grad_norm': '0.2805', 'learning_rate': '0.0006', 'epoch': '4.735'}
+
  2%|▏         | 170/7308 [03:08<1:54:41,  1.04it/s]
  2%|▏         | 171/7308 [03:09<1:43:37,  1.15it/s]
  2%|▏         | 172/7308 [03:09<1:35:56,  1.24it/s]
  2%|▏         | 173/7308 [03:11<1:47:48,  1.10it/s]
  2%|▏         | 174/7308 [03:11<1:38:50,  1.20it/s]
  2%|▏         | 175/7308 [03:12<1:32:34,  1.28it/s]
  2%|▏         | 176/7308 [03:13<1:52:43,  1.05it/s]
  2%|▏         | 177/7308 [03:14<1:42:14,  1.16it/s]
  2%|▏         | 178/7308 [03:14<1:34:51,  1.25it/s]
  2%|▏         | 179/7308 [03:15<1:29:38,  1.33it/s]
  2%|▏         | 180/7308 [03:15<1:11:40,  1.66it/s]
                                                    
{'loss': '6.436', 'grad_norm': '0.9916', 'learning_rate': '0.0006', 'epoch': '5'}
+
  2%|▏         | 180/7308 [03:15<1:11:40,  1.66it/s]
  2%|▏         | 181/7308 [03:16<1:16:02,  1.56it/s]
  2%|▏         | 182/7308 [03:17<1:16:34,  1.55it/s]
  3%|▎         | 183/7308 [03:17<1:16:52,  1.54it/s]
  3%|▎         | 184/7308 [03:18<1:17:06,  1.54it/s]
  3%|▎         | 185/7308 [03:19<1:17:18,  1.54it/s]
  3%|▎         | 186/7308 [03:19<1:17:24,  1.53it/s]
  3%|▎         | 187/7308 [03:20<1:17:28,  1.53it/s]
  3%|▎         | 188/7308 [03:21<1:17:33,  1.53it/s]
  3%|▎         | 189/7308 [03:21<1:17:36,  1.53it/s]
  3%|▎         | 190/7308 [03:22<1:17:35,  1.53it/s]
                                                    
{'loss': '6.406', 'grad_norm': '0.5042', 'learning_rate': '0.0006', 'epoch': '5.283'}
+
  3%|▎         | 190/7308 [03:22<1:17:35,  1.53it/s]
  3%|▎         | 191/7308 [03:23<1:17:41,  1.53it/s]
  3%|▎         | 192/7308 [03:23<1:17:39,  1.53it/s]
  3%|▎         | 193/7308 [03:24<1:17:41,  1.53it/s]
  3%|▎         | 194/7308 [03:25<1:17:41,  1.53it/s]
  3%|▎         | 195/7308 [03:25<1:17:38,  1.53it/s]
  3%|▎         | 196/7308 [03:26<1:17:35,  1.53it/s]
  3%|▎         | 197/7308 [03:27<1:17:35,  1.53it/s]
  3%|▎         | 198/7308 [03:27<1:17:35,  1.53it/s]
  3%|▎         | 199/7308 [03:28<1:17:33,  1.53it/s]
  3%|▎         | 200/7308 [03:29<1:17:35,  1.53it/s]
                                                    
{'loss': '6.33', 'grad_norm': '0.272', 'learning_rate': '0.0006', 'epoch': '5.565'}
+
  3%|▎         | 200/7308 [03:29<1:17:35,  1.53it/s]
  3%|▎         | 201/7308 [03:29<1:17:38,  1.53it/s]
  3%|▎         | 202/7308 [03:30<1:17:36,  1.53it/s]
  3%|▎         | 203/7308 [03:31<1:17:34,  1.53it/s]
  3%|▎         | 204/7308 [03:31<1:17:31,  1.53it/s]
  3%|▎         | 205/7308 [03:32<1:17:32,  1.53it/s]
  3%|▎         | 206/7308 [03:32<1:17:31,  1.53it/s]
  3%|▎         | 207/7308 [03:33<1:17:27,  1.53it/s]
  3%|▎         | 208/7308 [03:34<1:17:32,  1.53it/s]
  3%|▎         | 209/7308 [03:34<1:17:30,  1.53it/s]
  3%|▎         | 210/7308 [03:35<1:17:27,  1.53it/s]
                                                    
{'loss': '6.295', 'grad_norm': '0.6015', 'learning_rate': '0.0006', 'epoch': '5.848'}
+
  3%|▎         | 210/7308 [03:35<1:17:27,  1.53it/s]
  3%|▎         | 211/7308 [03:36<1:17:28,  1.53it/s]
  3%|▎         | 212/7308 [03:36<1:17:27,  1.53it/s]
  3%|▎         | 213/7308 [03:37<1:17:28,  1.53it/s]
  3%|▎         | 214/7308 [03:38<1:17:28,  1.53it/s]
  3%|▎         | 215/7308 [03:38<1:17:25,  1.53it/s]
  3%|▎         | 216/7308 [03:39<1:03:07,  1.87it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.63it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.81it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.55it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.02it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.74it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.57it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.46it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.38it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.33it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.29it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.29it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.27it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.26it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.26it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.25it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.25it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.26it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.24it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s]
                                                    
+
                                               
{'eval_loss': '6.375', 'eval_runtime': '3.247', 'eval_samples_per_second': '751.8', 'eval_steps_per_second': '12.01', 'epoch': '6'}
+
  3%|▎         | 216/7308 [03:42<1:03:07,  1.87it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-216
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-216/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-216/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.78it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.77it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-216/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-216/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-216/tokenizer_config.json
+[eval] epoch 6.0 step 216 loss 6.3752 (best 6.3752 @ ep 6.0, 0/20 since best)
+
+
                                               
  3%|▎         | 217/7308 [03:45<4:19:40,  2.20s/it]
  3%|▎         | 218/7308 [03:45<3:24:52,  1.73s/it]
  3%|▎         | 219/7308 [03:46<2:46:30,  1.41s/it]
  3%|▎         | 220/7308 [03:47<2:19:42,  1.18s/it]
                                                    
{'loss': '6.228', 'grad_norm': '0.6955', 'learning_rate': '0.0006', 'epoch': '6.113'}
+
  3%|▎         | 220/7308 [03:47<2:19:42,  1.18s/it]
  3%|▎         | 221/7308 [03:47<2:00:56,  1.02s/it]
  3%|▎         | 222/7308 [03:48<1:47:48,  1.10it/s]
  3%|▎         | 223/7308 [03:49<1:38:39,  1.20it/s]
  3%|▎         | 224/7308 [03:49<1:32:21,  1.28it/s]
  3%|▎         | 225/7308 [03:50<1:27:44,  1.35it/s]
  3%|▎         | 226/7308 [03:51<1:24:34,  1.40it/s]
  3%|▎         | 227/7308 [03:51<1:22:22,  1.43it/s]
  3%|▎         | 228/7308 [03:52<1:20:46,  1.46it/s]
  3%|▎         | 229/7308 [03:53<1:19:38,  1.48it/s]
  3%|▎         | 230/7308 [03:53<1:18:53,  1.50it/s]
                                                    
{'loss': '6.221', 'grad_norm': '0.6272', 'learning_rate': '0.0006', 'epoch': '6.396'}
+
  3%|▎         | 230/7308 [03:53<1:18:53,  1.50it/s]
  3%|▎         | 231/7308 [03:54<1:18:26,  1.50it/s]
  3%|▎         | 232/7308 [03:55<1:18:03,  1.51it/s]
  3%|▎         | 233/7308 [03:55<1:17:47,  1.52it/s]
  3%|▎         | 234/7308 [03:56<1:17:33,  1.52it/s]
  3%|▎         | 235/7308 [03:56<1:17:22,  1.52it/s]
  3%|▎         | 236/7308 [03:57<1:17:17,  1.52it/s]
  3%|▎         | 237/7308 [03:58<1:17:16,  1.53it/s]
  3%|▎         | 238/7308 [03:58<1:17:15,  1.53it/s]
  3%|▎         | 239/7308 [03:59<1:17:10,  1.53it/s]
  3%|▎         | 240/7308 [04:00<1:17:07,  1.53it/s]
                                                    
{'loss': '6.147', 'grad_norm': '0.2919', 'learning_rate': '0.0006', 'epoch': '6.678'}
+
  3%|▎         | 240/7308 [04:00<1:17:07,  1.53it/s]
  3%|▎         | 241/7308 [04:00<1:17:10,  1.53it/s]
  3%|▎         | 242/7308 [04:01<1:17:04,  1.53it/s]
  3%|▎         | 243/7308 [04:02<1:16:58,  1.53it/s]
  3%|▎         | 244/7308 [04:02<1:16:58,  1.53it/s]
  3%|▎         | 245/7308 [04:03<1:16:58,  1.53it/s]
  3%|▎         | 246/7308 [04:04<1:17:03,  1.53it/s]
  3%|▎         | 247/7308 [04:04<1:17:02,  1.53it/s]
  3%|▎         | 248/7308 [04:05<1:17:00,  1.53it/s]
  3%|▎         | 249/7308 [04:06<1:16:57,  1.53it/s]
  3%|▎         | 250/7308 [04:06<1:16:56,  1.53it/s]
                                                    
{'loss': '6.099', 'grad_norm': '0.7794', 'learning_rate': '0.0006', 'epoch': '6.961'}
+
  3%|▎         | 250/7308 [04:06<1:16:56,  1.53it/s]
  3%|▎         | 251/7308 [04:07<1:16:57,  1.53it/s]
  3%|▎         | 252/7308 [04:07<1:02:41,  1.88it/s]
  3%|▎         | 253/7308 [04:08<1:10:20,  1.67it/s]
  3%|▎         | 254/7308 [04:09<1:12:24,  1.62it/s]
  3%|▎         | 255/7308 [04:09<1:13:45,  1.59it/s]
  4%|▎         | 256/7308 [04:10<1:14:38,  1.57it/s]
  4%|▎         | 257/7308 [04:11<1:15:20,  1.56it/s]
  4%|▎         | 258/7308 [04:11<1:15:49,  1.55it/s]
  4%|▎         | 259/7308 [04:12<1:16:09,  1.54it/s]
  4%|▎         | 260/7308 [04:13<1:16:20,  1.54it/s]
                                                    
{'loss': '6.055', 'grad_norm': '0.4031', 'learning_rate': '0.0006', 'epoch': '7.226'}
+
  4%|▎         | 260/7308 [04:13<1:16:20,  1.54it/s]
  4%|▎         | 261/7308 [04:13<1:16:32,  1.53it/s]
  4%|▎         | 262/7308 [04:14<1:16:37,  1.53it/s]
  4%|▎         | 263/7308 [04:14<1:16:44,  1.53it/s]
  4%|▎         | 264/7308 [04:15<1:16:43,  1.53it/s]
  4%|▎         | 265/7308 [04:16<1:16:43,  1.53it/s]
  4%|▎         | 266/7308 [04:16<1:16:43,  1.53it/s]
  4%|▎         | 267/7308 [04:17<1:16:41,  1.53it/s]
  4%|▎         | 268/7308 [04:18<1:16:41,  1.53it/s]
  4%|▎         | 269/7308 [04:18<1:16:38,  1.53it/s]
  4%|▎         | 270/7308 [04:19<1:16:37,  1.53it/s]
                                                    
{'loss': '5.987', 'grad_norm': '0.8692', 'learning_rate': '0.0006', 'epoch': '7.509'}
+
  4%|▎         | 270/7308 [04:19<1:16:37,  1.53it/s]
  4%|▎         | 271/7308 [04:20<1:16:41,  1.53it/s]
  4%|▎         | 272/7308 [04:20<1:16:45,  1.53it/s]
  4%|▎         | 273/7308 [04:21<1:16:44,  1.53it/s]
  4%|▎         | 274/7308 [04:22<1:16:41,  1.53it/s]
  4%|▍         | 275/7308 [04:22<1:16:41,  1.53it/s]
  4%|▍         | 276/7308 [04:23<1:16:39,  1.53it/s]
  4%|▍         | 277/7308 [04:24<1:16:39,  1.53it/s]
  4%|▍         | 278/7308 [04:24<1:16:39,  1.53it/s]
  4%|▍         | 279/7308 [04:25<1:16:37,  1.53it/s]
  4%|▍         | 280/7308 [04:26<1:16:39,  1.53it/s]
                                                    
{'loss': '5.986', 'grad_norm': '0.8808', 'learning_rate': '0.0006', 'epoch': '7.792'}
+
  4%|▍         | 280/7308 [04:26<1:16:39,  1.53it/s]
  4%|▍         | 281/7308 [04:26<1:16:39,  1.53it/s]
  4%|▍         | 282/7308 [04:27<1:16:36,  1.53it/s]
  4%|▍         | 283/7308 [04:28<1:16:41,  1.53it/s]
  4%|▍         | 284/7308 [04:28<1:16:40,  1.53it/s]
  4%|▍         | 285/7308 [04:29<1:16:38,  1.53it/s]
  4%|▍         | 286/7308 [04:30<1:16:37,  1.53it/s]
  4%|▍         | 287/7308 [04:30<1:16:32,  1.53it/s]
  4%|▍         | 288/7308 [04:30<1:02:22,  1.88it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.70it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.80it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.63it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.09it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.78it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.59it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.46it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.38it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.30it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.27it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.26it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.26it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.25it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.25it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.21it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.22it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.21it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.21it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.22it/s]
                                                    
+
                                               
{'eval_loss': '6.045', 'eval_runtime': '3.251', 'eval_samples_per_second': '750.8', 'eval_steps_per_second': '12', 'epoch': '8'}
+
  4%|▍         | 288/7308 [04:34<1:02:22,  1.88it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.22it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-288
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-288/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-288/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.94it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.94it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-288/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-288/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-288/tokenizer_config.json
+[eval] epoch 8.0 step 288 loss 6.0452 (best 6.0452 @ ep 8.0, 0/20 since best)
+
+
                                               
  4%|▍         | 289/7308 [04:36<4:14:19,  2.17s/it]
  4%|▍         | 290/7308 [04:37<3:21:01,  1.72s/it]
                                                    
{'loss': '5.948', 'grad_norm': '0.3433', 'learning_rate': '0.0006', 'epoch': '8.057'}
+
  4%|▍         | 290/7308 [04:37<3:21:01,  1.72s/it]
  4%|▍         | 291/7308 [04:38<2:43:39,  1.40s/it]
  4%|▍         | 292/7308 [04:38<2:17:28,  1.18s/it]
  4%|▍         | 293/7308 [04:39<1:59:12,  1.02s/it]
  4%|▍         | 294/7308 [04:40<1:46:28,  1.10it/s]
  4%|▍         | 295/7308 [04:40<1:37:29,  1.20it/s]
  4%|▍         | 296/7308 [04:41<1:31:09,  1.28it/s]
  4%|▍         | 297/7308 [04:42<1:26:43,  1.35it/s]
  4%|▍         | 298/7308 [04:42<1:23:37,  1.40it/s]
  4%|▍         | 299/7308 [04:43<1:21:25,  1.43it/s]
  4%|▍         | 300/7308 [04:44<1:19:53,  1.46it/s]
                                                    
{'loss': '5.856', 'grad_norm': '0.351', 'learning_rate': '0.0006', 'epoch': '8.339'}
+
  4%|▍         | 300/7308 [04:44<1:19:53,  1.46it/s]
  4%|▍         | 301/7308 [04:44<1:18:52,  1.48it/s]
  4%|▍         | 302/7308 [04:45<1:18:05,  1.50it/s]
  4%|▍         | 303/7308 [04:46<1:17:32,  1.51it/s]
  4%|▍         | 304/7308 [04:46<1:17:12,  1.51it/s]
  4%|▍         | 305/7308 [04:47<1:16:55,  1.52it/s]
  4%|▍         | 306/7308 [04:48<1:16:45,  1.52it/s]
  4%|▍         | 307/7308 [04:48<1:16:37,  1.52it/s]
  4%|▍         | 308/7308 [04:49<1:16:33,  1.52it/s]
  4%|▍         | 309/7308 [04:50<1:16:29,  1.52it/s]
  4%|▍         | 310/7308 [04:50<1:16:27,  1.53it/s]
                                                    
{'loss': '5.836', 'grad_norm': '0.9778', 'learning_rate': '0.0006', 'epoch': '8.622'}
+
  4%|▍         | 310/7308 [04:50<1:16:27,  1.53it/s]
  4%|▍         | 311/7308 [04:51<1:16:25,  1.53it/s]
  4%|▍         | 312/7308 [04:51<1:16:22,  1.53it/s]
  4%|▍         | 313/7308 [04:52<1:16:19,  1.53it/s]
  4%|▍         | 314/7308 [04:53<1:16:18,  1.53it/s]
  4%|▍         | 315/7308 [04:53<1:16:20,  1.53it/s]
  4%|▍         | 316/7308 [04:54<1:16:17,  1.53it/s]
  4%|▍         | 317/7308 [04:55<1:16:15,  1.53it/s]
  4%|▍         | 318/7308 [04:55<1:16:14,  1.53it/s]
  4%|▍         | 319/7308 [04:56<1:16:14,  1.53it/s]
  4%|▍         | 320/7308 [04:57<1:16:12,  1.53it/s]
                                                    
{'loss': '5.851', 'grad_norm': '0.3472', 'learning_rate': '0.0006', 'epoch': '8.905'}
+
  4%|▍         | 320/7308 [04:57<1:16:12,  1.53it/s]
  4%|▍         | 321/7308 [04:57<1:16:14,  1.53it/s]
  4%|▍         | 322/7308 [04:58<1:16:14,  1.53it/s]
  4%|▍         | 323/7308 [04:59<1:16:11,  1.53it/s]
  4%|▍         | 324/7308 [04:59<1:02:05,  1.87it/s]
  4%|▍         | 325/7308 [05:00<1:09:06,  1.68it/s]
  4%|▍         | 326/7308 [05:00<1:11:17,  1.63it/s]
  4%|▍         | 327/7308 [05:01<1:12:44,  1.60it/s]
  4%|▍         | 328/7308 [05:02<1:13:44,  1.58it/s]
  5%|▍         | 329/7308 [05:02<1:14:25,  1.56it/s]
  5%|▍         | 330/7308 [05:03<1:14:56,  1.55it/s]
                                                    
{'loss': '5.765', 'grad_norm': '0.6656', 'learning_rate': '0.0006', 'epoch': '9.17'}
+
  5%|▍         | 330/7308 [05:03<1:14:56,  1.55it/s]
  5%|▍         | 331/7308 [05:04<1:15:20,  1.54it/s]
  5%|▍         | 332/7308 [05:04<1:15:30,  1.54it/s]
  5%|▍         | 333/7308 [05:05<1:15:43,  1.54it/s]
  5%|▍         | 334/7308 [05:06<1:15:50,  1.53it/s]
  5%|▍         | 335/7308 [05:06<1:15:52,  1.53it/s]
  5%|▍         | 336/7308 [05:07<1:15:55,  1.53it/s]
  5%|▍         | 337/7308 [05:08<1:15:59,  1.53it/s]
  5%|▍         | 338/7308 [05:08<1:16:00,  1.53it/s]
  5%|▍         | 339/7308 [05:09<1:16:00,  1.53it/s]
  5%|▍         | 340/7308 [05:10<1:15:58,  1.53it/s]
                                                    
{'loss': '5.722', 'grad_norm': '0.3786', 'learning_rate': '0.0006', 'epoch': '9.452'}
+
  5%|▍         | 340/7308 [05:10<1:15:58,  1.53it/s]
  5%|▍         | 341/7308 [05:10<1:16:04,  1.53it/s]
  5%|▍         | 342/7308 [05:11<1:16:03,  1.53it/s]
  5%|▍         | 343/7308 [05:11<1:15:59,  1.53it/s]
  5%|▍         | 344/7308 [05:12<1:15:57,  1.53it/s]
  5%|▍         | 345/7308 [05:13<1:15:58,  1.53it/s]
  5%|▍         | 346/7308 [05:13<1:16:01,  1.53it/s]
  5%|▍         | 347/7308 [05:14<1:15:58,  1.53it/s]
  5%|▍         | 348/7308 [05:15<1:15:55,  1.53it/s]
  5%|▍         | 349/7308 [05:15<1:16:00,  1.53it/s]
  5%|▍         | 350/7308 [05:16<1:15:59,  1.53it/s]
                                                    
{'loss': '5.698', 'grad_norm': '0.6408', 'learning_rate': '0.0006', 'epoch': '9.735'}
+
  5%|▍         | 350/7308 [05:16<1:15:59,  1.53it/s]
  5%|▍         | 351/7308 [05:17<1:15:57,  1.53it/s]
  5%|▍         | 352/7308 [05:17<1:15:52,  1.53it/s]
  5%|▍         | 353/7308 [05:18<1:15:52,  1.53it/s]
  5%|▍         | 354/7308 [05:19<1:15:50,  1.53it/s]
  5%|▍         | 355/7308 [05:19<1:15:49,  1.53it/s]
  5%|▍         | 356/7308 [05:20<1:15:54,  1.53it/s]
  5%|▍         | 357/7308 [05:21<1:15:55,  1.53it/s]
  5%|▍         | 358/7308 [05:21<1:15:53,  1.53it/s]
  5%|▍         | 359/7308 [05:22<1:15:51,  1.53it/s]
  5%|▍         | 360/7308 [05:22<1:01:48,  1.87it/s]
                                                    
{'loss': '5.698', 'grad_norm': '0.5507', 'learning_rate': '0.0006', 'epoch': '10'}
+
  5%|▍         | 360/7308 [05:22<1:01:48,  1.87it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 19.37it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 15.03it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.72it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.11it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.82it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.59it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.49it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.39it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.36it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.30it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.31it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.29it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.27it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.26it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.25it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.25it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.24it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.24it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.25it/s]
                                                    
+
                                               
{'eval_loss': '5.708', 'eval_runtime': '3.238', 'eval_samples_per_second': '753.8', 'eval_steps_per_second': '12.04', 'epoch': '10'}
+
  5%|▍         | 360/7308 [05:25<1:01:48,  1.87it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.25it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-360
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-360/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-360/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.78it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.78it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-360/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-360/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-360/tokenizer_config.json
+[eval] epoch 10.0 step 360 loss 5.7081 (best 5.7081 @ ep 10.0, 0/20 since best)
+
+
                                               
  5%|▍         | 361/7308 [05:28<4:13:52,  2.19s/it]
  5%|▍         | 362/7308 [05:29<3:20:30,  1.73s/it]
  5%|▍         | 363/7308 [05:30<2:43:01,  1.41s/it]
  5%|▍         | 364/7308 [05:30<2:16:49,  1.18s/it]
  5%|▍         | 365/7308 [05:31<1:58:30,  1.02s/it]
  5%|▌         | 366/7308 [05:32<1:45:42,  1.09it/s]
  5%|▌         | 367/7308 [05:32<1:36:41,  1.20it/s]
  5%|▌         | 368/7308 [05:33<1:30:24,  1.28it/s]
  5%|▌         | 369/7308 [05:34<1:26:03,  1.34it/s]
  5%|▌         | 370/7308 [05:34<1:22:57,  1.39it/s]
                                                    
{'loss': '5.615', 'grad_norm': '0.2801', 'learning_rate': '0.0006', 'epoch': '10.28'}
+
  5%|▌         | 370/7308 [05:34<1:22:57,  1.39it/s]
  5%|▌         | 371/7308 [05:35<1:20:49,  1.43it/s]
  5%|▌         | 372/7308 [05:35<1:19:12,  1.46it/s]
  5%|▌         | 373/7308 [05:36<1:18:07,  1.48it/s]
  5%|▌         | 374/7308 [05:37<1:17:26,  1.49it/s]
  5%|▌         | 375/7308 [05:37<1:16:57,  1.50it/s]
  5%|▌         | 376/7308 [05:38<1:16:30,  1.51it/s]
  5%|▌         | 377/7308 [05:39<1:16:13,  1.52it/s]
  5%|▌         | 378/7308 [05:39<1:16:06,  1.52it/s]
  5%|▌         | 379/7308 [05:40<1:16:00,  1.52it/s]
  5%|▌         | 380/7308 [05:41<1:15:51,  1.52it/s]
                                                    
{'loss': '5.569', 'grad_norm': '0.3625', 'learning_rate': '0.0006', 'epoch': '10.57'}
+
  5%|▌         | 380/7308 [05:41<1:15:51,  1.52it/s]
  5%|▌         | 381/7308 [05:41<1:15:48,  1.52it/s]
  5%|▌         | 382/7308 [05:42<1:15:46,  1.52it/s]
  5%|▌         | 383/7308 [05:43<1:15:50,  1.52it/s]
  5%|▌         | 384/7308 [05:43<1:15:45,  1.52it/s]
  5%|▌         | 385/7308 [05:44<1:15:39,  1.53it/s]
  5%|▌         | 386/7308 [05:45<1:15:34,  1.53it/s]
  5%|▌         | 387/7308 [05:45<1:15:34,  1.53it/s]
  5%|▌         | 388/7308 [05:46<1:15:32,  1.53it/s]
  5%|▌         | 389/7308 [05:47<1:15:31,  1.53it/s]
  5%|▌         | 390/7308 [05:47<1:15:29,  1.53it/s]
                                                    
{'loss': '5.559', 'grad_norm': '0.6908', 'learning_rate': '0.0006', 'epoch': '10.85'}
+
  5%|▌         | 390/7308 [05:47<1:15:29,  1.53it/s]
  5%|▌         | 391/7308 [05:48<1:15:36,  1.52it/s]
  5%|▌         | 392/7308 [05:49<1:15:33,  1.53it/s]
  5%|▌         | 393/7308 [05:49<1:15:30,  1.53it/s]
  5%|▌         | 394/7308 [05:50<1:15:33,  1.52it/s]
  5%|▌         | 395/7308 [05:51<1:15:30,  1.53it/s]
  5%|▌         | 396/7308 [05:51<1:01:34,  1.87it/s]
  5%|▌         | 397/7308 [05:52<1:09:07,  1.67it/s]
  5%|▌         | 398/7308 [05:52<1:11:00,  1.62it/s]
  5%|▌         | 399/7308 [05:53<1:12:20,  1.59it/s]
  5%|▌         | 400/7308 [05:54<1:13:16,  1.57it/s]
                                                    
{'loss': '5.538', 'grad_norm': '0.5966', 'learning_rate': '0.0006', 'epoch': '11.11'}
+
  5%|▌         | 400/7308 [05:54<1:13:16,  1.57it/s]
  5%|▌         | 401/7308 [05:54<1:13:56,  1.56it/s]
  6%|▌         | 402/7308 [05:55<1:14:21,  1.55it/s]
  6%|▌         | 403/7308 [05:55<1:14:44,  1.54it/s]
  6%|▌         | 404/7308 [05:56<1:14:58,  1.53it/s]
  6%|▌         | 405/7308 [05:57<1:15:03,  1.53it/s]
  6%|▌         | 406/7308 [05:57<1:15:09,  1.53it/s]
  6%|▌         | 407/7308 [05:58<1:15:10,  1.53it/s]
  6%|▌         | 408/7308 [05:59<1:15:12,  1.53it/s]
  6%|▌         | 409/7308 [05:59<1:15:13,  1.53it/s]
  6%|▌         | 410/7308 [06:00<1:15:15,  1.53it/s]
                                                    
{'loss': '5.474', 'grad_norm': '0.5551', 'learning_rate': '0.0006', 'epoch': '11.4'}
+
  6%|▌         | 410/7308 [06:00<1:15:15,  1.53it/s]
  6%|▌         | 411/7308 [06:01<1:15:16,  1.53it/s]
  6%|▌         | 412/7308 [06:01<1:15:18,  1.53it/s]
  6%|▌         | 413/7308 [06:02<1:15:18,  1.53it/s]
  6%|▌         | 414/7308 [06:03<1:15:15,  1.53it/s]
  6%|▌         | 415/7308 [06:03<1:15:14,  1.53it/s]
  6%|▌         | 416/7308 [06:04<1:15:17,  1.53it/s]
  6%|▌         | 417/7308 [06:05<1:15:24,  1.52it/s]
  6%|▌         | 418/7308 [06:05<1:15:21,  1.52it/s]
  6%|▌         | 419/7308 [06:06<1:15:15,  1.53it/s]
  6%|▌         | 420/7308 [06:07<1:15:11,  1.53it/s]
                                                    
{'loss': '5.452', 'grad_norm': '0.3134', 'learning_rate': '0.0006', 'epoch': '11.68'}
+
  6%|▌         | 420/7308 [06:07<1:15:11,  1.53it/s]
  6%|▌         | 421/7308 [06:07<1:15:15,  1.53it/s]
  6%|▌         | 422/7308 [06:08<1:15:12,  1.53it/s]
  6%|▌         | 423/7308 [06:09<1:15:10,  1.53it/s]
  6%|▌         | 424/7308 [06:09<1:15:07,  1.53it/s]
  6%|▌         | 425/7308 [06:10<1:15:09,  1.53it/s]
  6%|▌         | 426/7308 [06:11<1:15:08,  1.53it/s]
  6%|▌         | 427/7308 [06:11<1:15:06,  1.53it/s]
  6%|▌         | 428/7308 [06:12<1:15:05,  1.53it/s]
  6%|▌         | 429/7308 [06:13<1:15:06,  1.53it/s]
  6%|▌         | 430/7308 [06:13<1:15:07,  1.53it/s]
                                                    
{'loss': '5.408', 'grad_norm': '0.6855', 'learning_rate': '0.0006', 'epoch': '11.96'}
+
  6%|▌         | 430/7308 [06:13<1:15:07,  1.53it/s]
  6%|▌         | 431/7308 [06:14<1:15:03,  1.53it/s]
  6%|▌         | 432/7308 [06:14<1:01:12,  1.87it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.72it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.78it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.63it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.04it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.78it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.58it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.43it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.41it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.33it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.30it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.29it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.30it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.28it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.27it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.25it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.25it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.26it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.25it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.25it/s]
                                                    
+
                                               
{'eval_loss': '5.405', 'eval_runtime': '3.26', 'eval_samples_per_second': '748.7', 'eval_steps_per_second': '11.96', 'epoch': '12'}
+
  6%|▌         | 432/7308 [06:17<1:01:12,  1.87it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.25it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-432
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-432/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-432/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.65it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.65it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-432/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-432/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-432/tokenizer_config.json
+[eval] epoch 12.0 step 432 loss 5.4047 (best 5.4047 @ ep 12.0, 0/20 since best)
+
+
                                               
  6%|▌         | 433/7308 [06:20<4:14:33,  2.22s/it]
  6%|▌         | 434/7308 [06:21<3:20:37,  1.75s/it]
  6%|▌         | 435/7308 [06:22<2:42:53,  1.42s/it]
  6%|▌         | 436/7308 [06:22<2:16:29,  1.19s/it]
  6%|▌         | 437/7308 [06:23<1:58:02,  1.03s/it]
  6%|▌         | 438/7308 [06:24<1:45:07,  1.09it/s]
  6%|▌         | 439/7308 [06:24<1:36:03,  1.19it/s]
  6%|▌         | 440/7308 [06:25<1:29:43,  1.28it/s]
                                                    
{'loss': '5.414', 'grad_norm': '0.4941', 'learning_rate': '0.0006', 'epoch': '12.23'}
+
  6%|▌         | 440/7308 [06:25<1:29:43,  1.28it/s]
  6%|▌         | 441/7308 [06:25<1:25:22,  1.34it/s]
  6%|▌         | 442/7308 [06:26<1:22:14,  1.39it/s]
  6%|▌         | 443/7308 [06:27<1:20:00,  1.43it/s]
  6%|▌         | 444/7308 [06:27<1:18:26,  1.46it/s]
  6%|▌         | 445/7308 [06:28<1:17:23,  1.48it/s]
  6%|▌         | 446/7308 [06:29<1:16:40,  1.49it/s]
  6%|▌         | 447/7308 [06:29<1:16:07,  1.50it/s]
  6%|▌         | 448/7308 [06:30<1:15:42,  1.51it/s]
  6%|▌         | 449/7308 [06:31<1:15:27,  1.51it/s]
  6%|▌         | 450/7308 [06:31<1:15:15,  1.52it/s]
                                                    
{'loss': '5.36', 'grad_norm': '0.3882', 'learning_rate': '0.0006', 'epoch': '12.51'}
+
  6%|▌         | 450/7308 [06:31<1:15:15,  1.52it/s]
  6%|▌         | 451/7308 [06:32<1:15:09,  1.52it/s]
  6%|▌         | 452/7308 [06:33<1:15:02,  1.52it/s]
  6%|▌         | 453/7308 [06:33<1:15:00,  1.52it/s]
  6%|▌         | 454/7308 [06:34<1:14:54,  1.52it/s]
  6%|▌         | 455/7308 [06:35<1:14:53,  1.53it/s]
  6%|▌         | 456/7308 [06:35<1:14:48,  1.53it/s]
  6%|▋         | 457/7308 [06:36<1:14:46,  1.53it/s]
  6%|▋         | 458/7308 [06:37<1:14:42,  1.53it/s]
  6%|▋         | 459/7308 [06:37<1:14:43,  1.53it/s]
  6%|▋         | 460/7308 [06:38<1:14:44,  1.53it/s]
                                                    
{'loss': '5.309', 'grad_norm': '0.2723', 'learning_rate': '0.0006', 'epoch': '12.79'}
+
  6%|▋         | 460/7308 [06:38<1:14:44,  1.53it/s]
  6%|▋         | 461/7308 [06:39<1:14:44,  1.53it/s]
  6%|▋         | 462/7308 [06:39<1:14:42,  1.53it/s]
  6%|▋         | 463/7308 [06:40<1:14:39,  1.53it/s]
  6%|▋         | 464/7308 [06:41<1:14:39,  1.53it/s]
  6%|▋         | 465/7308 [06:41<1:14:35,  1.53it/s]
  6%|▋         | 466/7308 [06:42<1:14:39,  1.53it/s]
  6%|▋         | 467/7308 [06:43<1:14:40,  1.53it/s]
  6%|▋         | 468/7308 [06:43<1:00:49,  1.87it/s]
  6%|▋         | 469/7308 [06:44<1:08:25,  1.67it/s]
  6%|▋         | 470/7308 [06:44<1:10:14,  1.62it/s]
                                                    
{'loss': '5.265', 'grad_norm': '0.2843', 'learning_rate': '0.0006', 'epoch': '13.06'}
+
  6%|▋         | 470/7308 [06:44<1:10:14,  1.62it/s]
  6%|▋         | 471/7308 [06:45<1:11:37,  1.59it/s]
  6%|▋         | 472/7308 [06:45<1:12:31,  1.57it/s]
  6%|▋         | 473/7308 [06:46<1:13:09,  1.56it/s]
  6%|▋         | 474/7308 [06:47<1:13:32,  1.55it/s]
  6%|▋         | 475/7308 [06:47<1:13:53,  1.54it/s]
  7%|▋         | 476/7308 [06:48<1:14:03,  1.54it/s]
  7%|▋         | 477/7308 [06:49<1:14:08,  1.54it/s]
  7%|▋         | 478/7308 [06:49<1:14:15,  1.53it/s]
  7%|▋         | 479/7308 [06:50<1:14:22,  1.53it/s]
  7%|▋         | 480/7308 [06:51<1:14:24,  1.53it/s]
                                                    
{'loss': '5.233', 'grad_norm': '0.585', 'learning_rate': '0.0006', 'epoch': '13.34'}
+
  7%|▋         | 480/7308 [06:51<1:14:24,  1.53it/s]
  7%|▋         | 481/7308 [06:51<1:14:26,  1.53it/s]
  7%|▋         | 482/7308 [06:52<1:14:25,  1.53it/s]
  7%|▋         | 483/7308 [06:53<1:14:26,  1.53it/s]
  7%|▋         | 484/7308 [06:53<1:14:30,  1.53it/s]
  7%|▋         | 485/7308 [06:54<1:14:28,  1.53it/s]
  7%|▋         | 486/7308 [06:55<1:14:24,  1.53it/s]
  7%|▋         | 487/7308 [06:55<1:14:28,  1.53it/s]
  7%|▋         | 488/7308 [06:56<1:14:28,  1.53it/s]
  7%|▋         | 489/7308 [06:57<1:14:23,  1.53it/s]
  7%|▋         | 490/7308 [06:57<1:14:23,  1.53it/s]
                                                    
{'loss': '5.281', 'grad_norm': '0.8874', 'learning_rate': '0.0006', 'epoch': '13.62'}
+
  7%|▋         | 490/7308 [06:57<1:14:23,  1.53it/s]
  7%|▋         | 491/7308 [06:58<1:14:27,  1.53it/s]
  7%|▋         | 492/7308 [06:59<1:14:26,  1.53it/s]
  7%|▋         | 493/7308 [06:59<1:14:23,  1.53it/s]
  7%|▋         | 494/7308 [07:00<1:14:27,  1.53it/s]
  7%|▋         | 495/7308 [07:01<1:14:22,  1.53it/s]
  7%|▋         | 496/7308 [07:01<1:14:20,  1.53it/s]
  7%|▋         | 497/7308 [07:02<1:14:20,  1.53it/s]
  7%|▋         | 498/7308 [07:03<1:14:17,  1.53it/s]
  7%|▋         | 499/7308 [07:03<1:14:18,  1.53it/s]
  7%|▋         | 500/7308 [07:04<1:14:19,  1.53it/s]
                                                    
{'loss': '5.273', 'grad_norm': '0.2752', 'learning_rate': '0.0006', 'epoch': '13.9'}
+
  7%|▋         | 500/7308 [07:04<1:14:19,  1.53it/s]
  7%|▋         | 501/7308 [07:04<1:14:19,  1.53it/s]
  7%|▋         | 502/7308 [07:05<1:14:16,  1.53it/s]
  7%|▋         | 503/7308 [07:06<1:14:15,  1.53it/s]
  7%|▋         | 504/7308 [07:06<1:00:29,  1.87it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.29it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.62it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.53it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.00it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.72it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.56it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.44it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.37it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.33it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.29it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.27it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.26it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.25it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.25it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.23it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.23it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.21it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.19it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.22it/s]
                                                    
+
                                               
{'eval_loss': '5.118', 'eval_runtime': '3.249', 'eval_samples_per_second': '751.4', 'eval_steps_per_second': '12', 'epoch': '14'}
+
  7%|▋         | 504/7308 [07:09<1:00:29,  1.87it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.22it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-504
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-504/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-504/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.78it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.77it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-504/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-504/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-504/tokenizer_config.json
+[eval] epoch 14.0 step 504 loss 5.1175 (best 5.1175 @ ep 14.0, 0/20 since best)
+
+
                                               
  7%|▋         | 505/7308 [07:12<4:09:35,  2.20s/it]
  7%|▋         | 506/7308 [07:13<3:16:56,  1.74s/it]
  7%|▋         | 507/7308 [07:13<2:40:01,  1.41s/it]
  7%|▋         | 508/7308 [07:14<2:14:17,  1.18s/it]
  7%|▋         | 509/7308 [07:15<1:56:17,  1.03s/it]
  7%|▋         | 510/7308 [07:15<1:43:37,  1.09it/s]
                                                    
{'loss': '5.167', 'grad_norm': '0.2404', 'learning_rate': '0.0006', 'epoch': '14.17'}
+
  7%|▋         | 510/7308 [07:15<1:43:37,  1.09it/s]
  7%|▋         | 511/7308 [07:16<1:34:53,  1.19it/s]
  7%|▋         | 512/7308 [07:17<1:28:35,  1.28it/s]
  7%|▋         | 513/7308 [07:17<1:24:16,  1.34it/s]
  7%|▋         | 514/7308 [07:18<1:21:14,  1.39it/s]
  7%|▋         | 515/7308 [07:19<1:19:05,  1.43it/s]
  7%|▋         | 516/7308 [07:19<1:17:33,  1.46it/s]
  7%|▋         | 517/7308 [07:20<1:16:31,  1.48it/s]
  7%|▋         | 518/7308 [07:21<1:15:46,  1.49it/s]
  7%|▋         | 519/7308 [07:21<1:15:16,  1.50it/s]
  7%|▋         | 520/7308 [07:22<1:14:54,  1.51it/s]
                                                    
{'loss': '5.119', 'grad_norm': '0.6117', 'learning_rate': '0.0006', 'epoch': '14.45'}
+
  7%|▋         | 520/7308 [07:22<1:14:54,  1.51it/s]
  7%|▋         | 521/7308 [07:23<1:14:40,  1.51it/s]
  7%|▋         | 522/7308 [07:23<1:14:30,  1.52it/s]
  7%|▋         | 523/7308 [07:24<1:14:21,  1.52it/s]
  7%|▋         | 524/7308 [07:25<1:14:15,  1.52it/s]
  7%|▋         | 525/7308 [07:25<1:14:12,  1.52it/s]
  7%|▋         | 526/7308 [07:26<1:14:09,  1.52it/s]
  7%|▋         | 527/7308 [07:27<1:14:04,  1.53it/s]
  7%|▋         | 528/7308 [07:27<1:14:02,  1.53it/s]
  7%|▋         | 529/7308 [07:28<1:14:06,  1.52it/s]
  7%|▋         | 530/7308 [07:28<1:14:04,  1.53it/s]
                                                    
{'loss': '5.109', 'grad_norm': '0.3132', 'learning_rate': '0.0006', 'epoch': '14.73'}
+
  7%|▋         | 530/7308 [07:29<1:14:04,  1.53it/s]
  7%|▋         | 531/7308 [07:29<1:14:00,  1.53it/s]
  7%|▋         | 532/7308 [07:30<1:13:55,  1.53it/s]
  7%|▋         | 533/7308 [07:30<1:13:57,  1.53it/s]
  7%|▋         | 534/7308 [07:31<1:13:54,  1.53it/s]
  7%|▋         | 535/7308 [07:32<1:13:54,  1.53it/s]
  7%|▋         | 536/7308 [07:32<1:13:56,  1.53it/s]
  7%|▋         | 537/7308 [07:33<1:13:57,  1.53it/s]
  7%|▋         | 538/7308 [07:34<1:13:58,  1.53it/s]
  7%|▋         | 539/7308 [07:34<1:13:55,  1.53it/s]
  7%|▋         | 540/7308 [07:35<1:00:15,  1.87it/s]
                                                    
{'loss': '5.071', 'grad_norm': '0.3651', 'learning_rate': '0.0006', 'epoch': '15'}
+
  7%|▋         | 540/7308 [07:35<1:00:15,  1.87it/s]
  7%|▋         | 541/7308 [07:35<1:06:56,  1.68it/s]
  7%|▋         | 542/7308 [07:36<1:08:59,  1.63it/s]
  7%|▋         | 543/7308 [07:37<1:10:25,  1.60it/s]
  7%|▋         | 544/7308 [07:37<1:11:28,  1.58it/s]
  7%|▋         | 545/7308 [07:38<1:12:10,  1.56it/s]
  7%|▋         | 546/7308 [07:39<1:12:39,  1.55it/s]
  7%|▋         | 547/7308 [07:39<1:12:59,  1.54it/s]
  7%|▋         | 548/7308 [07:40<1:13:14,  1.54it/s]
  8%|▊         | 549/7308 [07:41<1:13:23,  1.54it/s]
  8%|▊         | 550/7308 [07:41<1:13:28,  1.53it/s]
                                                    
{'loss': '5.022', 'grad_norm': '0.4582', 'learning_rate': '0.0006', 'epoch': '15.28'}
+
  8%|▊         | 550/7308 [07:41<1:13:28,  1.53it/s]
  8%|▊         | 551/7308 [07:42<1:13:34,  1.53it/s]
  8%|▊         | 552/7308 [07:43<1:13:38,  1.53it/s]
  8%|▊         | 553/7308 [07:43<1:13:43,  1.53it/s]
  8%|▊         | 554/7308 [07:44<1:13:40,  1.53it/s]
  8%|▊         | 555/7308 [07:45<1:13:39,  1.53it/s]
  8%|▊         | 556/7308 [07:45<1:13:43,  1.53it/s]
  8%|▊         | 557/7308 [07:46<1:13:42,  1.53it/s]
  8%|▊         | 558/7308 [07:47<1:13:39,  1.53it/s]
  8%|▊         | 559/7308 [07:47<1:13:37,  1.53it/s]
  8%|▊         | 560/7308 [07:48<1:13:39,  1.53it/s]
                                                    
{'loss': '5.015', 'grad_norm': '0.4816', 'learning_rate': '0.0006', 'epoch': '15.57'}
+
  8%|▊         | 560/7308 [07:48<1:13:39,  1.53it/s]
  8%|▊         | 561/7308 [07:48<1:13:39,  1.53it/s]
  8%|▊         | 562/7308 [07:49<1:13:37,  1.53it/s]
  8%|▊         | 563/7308 [07:50<1:13:39,  1.53it/s]
  8%|▊         | 564/7308 [07:50<1:13:43,  1.52it/s]
  8%|▊         | 565/7308 [07:51<1:13:40,  1.53it/s]
  8%|▊         | 566/7308 [07:52<1:13:37,  1.53it/s]
  8%|▊         | 567/7308 [07:52<1:13:33,  1.53it/s]
  8%|▊         | 568/7308 [07:53<1:13:36,  1.53it/s]
  8%|▊         | 569/7308 [07:54<1:13:37,  1.53it/s]
  8%|▊         | 570/7308 [07:54<1:13:35,  1.53it/s]
                                                    
{'loss': '4.997', 'grad_norm': '0.3935', 'learning_rate': '0.0006', 'epoch': '15.85'}
+
  8%|▊         | 570/7308 [07:54<1:13:35,  1.53it/s]
  8%|▊         | 571/7308 [07:55<1:13:35,  1.53it/s]
  8%|▊         | 572/7308 [07:56<1:13:36,  1.53it/s]
  8%|▊         | 573/7308 [07:56<1:13:37,  1.52it/s]
  8%|▊         | 574/7308 [07:57<1:13:37,  1.52it/s]
  8%|▊         | 575/7308 [07:58<1:13:29,  1.53it/s]
  8%|▊         | 576/7308 [07:58<59:53,  1.87it/s]  [transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.94it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.94it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.72it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.10it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.80it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.59it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.49it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.40it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.36it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.29it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.31it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.28it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.27it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.26it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.25it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.25it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.24it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.25it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s]
                                                  
+
                                               
{'eval_loss': '4.801', 'eval_runtime': '3.232', 'eval_samples_per_second': '755.2', 'eval_steps_per_second': '12.07', 'epoch': '16'}
+
  8%|▊         | 576/7308 [08:01<59:53,  1.87it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-576
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-576/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-576/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.93it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.92it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-576/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-576/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-576/tokenizer_config.json
+[eval] epoch 16.0 step 576 loss 4.8011 (best 4.8011 @ ep 16.0, 0/20 since best)
+
+
                                               
  8%|▊         | 577/7308 [08:04<4:04:20,  2.18s/it]
  8%|▊         | 578/7308 [08:05<3:12:58,  1.72s/it]
  8%|▊         | 579/7308 [08:05<2:37:06,  1.40s/it]
  8%|▊         | 580/7308 [08:06<2:12:04,  1.18s/it]
                                                    
{'loss': '4.932', 'grad_norm': '0.2454', 'learning_rate': '0.0006', 'epoch': '16.11'}
+
  8%|▊         | 580/7308 [08:06<2:12:04,  1.18s/it]
  8%|▊         | 581/7308 [08:07<1:54:32,  1.02s/it]
  8%|▊         | 582/7308 [08:07<1:42:14,  1.10it/s]
  8%|▊         | 583/7308 [08:08<1:33:33,  1.20it/s]
  8%|▊         | 584/7308 [08:09<1:27:28,  1.28it/s]
  8%|▊         | 585/7308 [08:09<1:23:20,  1.34it/s]
  8%|▊         | 586/7308 [08:10<1:20:21,  1.39it/s]
  8%|▊         | 587/7308 [08:10<1:18:12,  1.43it/s]
  8%|▊         | 588/7308 [08:11<1:16:42,  1.46it/s]
  8%|▊         | 589/7308 [08:12<1:15:41,  1.48it/s]
  8%|▊         | 590/7308 [08:12<1:14:59,  1.49it/s]
                                                    
{'loss': '4.894', 'grad_norm': '0.368', 'learning_rate': '0.0006', 'epoch': '16.4'}
+
  8%|▊         | 590/7308 [08:12<1:14:59,  1.49it/s]
  8%|▊         | 591/7308 [08:13<1:14:32,  1.50it/s]
  8%|▊         | 592/7308 [08:14<1:14:07,  1.51it/s]
  8%|▊         | 593/7308 [08:14<1:13:55,  1.51it/s]
  8%|▊         | 594/7308 [08:15<1:13:44,  1.52it/s]
  8%|▊         | 595/7308 [08:16<1:13:37,  1.52it/s]
  8%|▊         | 596/7308 [08:16<1:13:34,  1.52it/s]
  8%|▊         | 597/7308 [08:17<1:13:28,  1.52it/s]
  8%|▊         | 598/7308 [08:18<1:13:30,  1.52it/s]
  8%|▊         | 599/7308 [08:18<1:13:27,  1.52it/s]
  8%|▊         | 600/7308 [08:19<1:13:23,  1.52it/s]
                                                    
{'loss': '4.886', 'grad_norm': '0.5228', 'learning_rate': '0.0006', 'epoch': '16.68'}
+
  8%|▊         | 600/7308 [08:19<1:13:23,  1.52it/s]
  8%|▊         | 601/7308 [08:20<1:13:21,  1.52it/s]
  8%|▊         | 602/7308 [08:20<1:13:20,  1.52it/s]
  8%|▊         | 603/7308 [08:21<1:13:21,  1.52it/s]
  8%|▊         | 604/7308 [08:22<1:13:25,  1.52it/s]
  8%|▊         | 605/7308 [08:22<1:13:22,  1.52it/s]
  8%|▊         | 606/7308 [08:23<1:13:20,  1.52it/s]
  8%|▊         | 607/7308 [08:24<1:13:19,  1.52it/s]
  8%|▊         | 608/7308 [08:24<1:13:13,  1.52it/s]
  8%|▊         | 609/7308 [08:25<1:13:12,  1.52it/s]
  8%|▊         | 610/7308 [08:26<1:13:11,  1.53it/s]
                                                    
{'loss': '4.879', 'grad_norm': '0.299', 'learning_rate': '0.0006', 'epoch': '16.96'}
+
  8%|▊         | 610/7308 [08:26<1:13:11,  1.53it/s]
  8%|▊         | 611/7308 [08:26<1:13:07,  1.53it/s]
  8%|▊         | 612/7308 [08:26<59:37,  1.87it/s]  
  8%|▊         | 613/7308 [08:27<1:05:59,  1.69it/s]
  8%|▊         | 614/7308 [08:28<1:08:09,  1.64it/s]
  8%|▊         | 615/7308 [08:29<1:09:39,  1.60it/s]
  8%|▊         | 616/7308 [08:29<1:10:44,  1.58it/s]
  8%|▊         | 617/7308 [08:30<1:11:21,  1.56it/s]
  8%|▊         | 618/7308 [08:30<1:11:50,  1.55it/s]
  8%|▊         | 619/7308 [08:31<1:12:17,  1.54it/s]
  8%|▊         | 620/7308 [08:32<1:12:36,  1.54it/s]
                                                    
{'loss': '4.809', 'grad_norm': '0.4084', 'learning_rate': '0.0006', 'epoch': '17.23'}
+
  8%|▊         | 620/7308 [08:32<1:12:36,  1.54it/s]
  8%|▊         | 621/7308 [08:32<1:12:46,  1.53it/s]
  9%|▊         | 622/7308 [08:33<1:12:55,  1.53it/s]
  9%|▊         | 623/7308 [08:34<1:12:59,  1.53it/s]
  9%|▊         | 624/7308 [08:34<1:12:55,  1.53it/s]
  9%|▊         | 625/7308 [08:35<1:12:58,  1.53it/s]
  9%|▊         | 626/7308 [08:36<1:13:09,  1.52it/s]
  9%|▊         | 627/7308 [08:36<1:13:13,  1.52it/s]
  9%|▊         | 628/7308 [08:37<1:13:08,  1.52it/s]
  9%|▊         | 629/7308 [08:38<1:13:00,  1.52it/s]
  9%|▊         | 630/7308 [08:38<1:12:56,  1.53it/s]
                                                    
{'loss': '4.807', 'grad_norm': '0.3537', 'learning_rate': '0.0006', 'epoch': '17.51'}
+
  9%|▊         | 630/7308 [08:38<1:12:56,  1.53it/s]
  9%|▊         | 631/7308 [08:39<1:12:59,  1.52it/s]
  9%|▊         | 632/7308 [08:40<1:13:07,  1.52it/s]
  9%|▊         | 633/7308 [08:40<1:13:08,  1.52it/s]
  9%|▊         | 634/7308 [08:41<1:13:02,  1.52it/s]
  9%|▊         | 635/7308 [08:42<1:12:56,  1.52it/s]
  9%|▊         | 636/7308 [08:42<1:12:54,  1.53it/s]
  9%|▊         | 637/7308 [08:43<1:12:55,  1.52it/s]
  9%|▊         | 638/7308 [08:44<1:12:54,  1.52it/s]
  9%|▊         | 639/7308 [08:44<1:12:52,  1.53it/s]
  9%|▉         | 640/7308 [08:45<1:12:47,  1.53it/s]
                                                    
{'loss': '4.805', 'grad_norm': '0.3577', 'learning_rate': '0.0006', 'epoch': '17.79'}
+
  9%|▉         | 640/7308 [08:45<1:12:47,  1.53it/s]
  9%|▉         | 641/7308 [08:46<1:12:51,  1.53it/s]
  9%|▉         | 642/7308 [08:46<1:12:51,  1.52it/s]
  9%|▉         | 643/7308 [08:47<1:12:54,  1.52it/s]
  9%|▉         | 644/7308 [08:48<1:12:48,  1.53it/s]
  9%|▉         | 645/7308 [08:48<1:12:47,  1.53it/s]
  9%|▉         | 646/7308 [08:49<1:12:50,  1.52it/s]
  9%|▉         | 647/7308 [08:49<1:12:47,  1.53it/s]
  9%|▉         | 648/7308 [08:50<59:19,  1.87it/s]  [transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.44it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.76it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.61it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.05it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.75it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.54it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.46it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.38it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.33it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.29it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.29it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.27it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.26it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.26it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.25it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.24it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.24it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.23it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.21it/s]
                                                  
+
                                               
{'eval_loss': '4.594', 'eval_runtime': '3.261', 'eval_samples_per_second': '748.5', 'eval_steps_per_second': '11.96', 'epoch': '18'}
+
  9%|▉         | 648/7308 [08:53<59:19,  1.87it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.21it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-648
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-648/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-648/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.79it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.79it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-648/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-648/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-648/tokenizer_config.json
+[eval] epoch 18.0 step 648 loss 4.5939 (best 4.5939 @ ep 18.0, 0/20 since best)
+
+
                                               
  9%|▉         | 649/7308 [08:56<4:04:11,  2.20s/it]
  9%|▉         | 650/7308 [08:56<3:12:42,  1.74s/it]
                                                    
{'loss': '4.764', 'grad_norm': '0.3993', 'learning_rate': '0.0006', 'epoch': '18.06'}
+
  9%|▉         | 650/7308 [08:56<3:12:42,  1.74s/it]
  9%|▉         | 651/7308 [08:57<2:36:38,  1.41s/it]
  9%|▉         | 652/7308 [08:58<2:11:26,  1.18s/it]
  9%|▉         | 653/7308 [08:58<1:53:47,  1.03s/it]
  9%|▉         | 654/7308 [08:59<1:41:25,  1.09it/s]
  9%|▉         | 655/7308 [09:00<1:32:45,  1.20it/s]
  9%|▉         | 656/7308 [09:00<1:26:43,  1.28it/s]
  9%|▉         | 657/7308 [09:01<1:22:27,  1.34it/s]
  9%|▉         | 658/7308 [09:02<1:19:30,  1.39it/s]
  9%|▉         | 659/7308 [09:02<1:17:24,  1.43it/s]
  9%|▉         | 660/7308 [09:03<1:15:59,  1.46it/s]
                                                    
{'loss': '4.724', 'grad_norm': '0.4623', 'learning_rate': '0.0006', 'epoch': '18.34'}
+
  9%|▉         | 660/7308 [09:03<1:15:59,  1.46it/s]
  9%|▉         | 661/7308 [09:04<1:15:01,  1.48it/s]
  9%|▉         | 662/7308 [09:04<1:14:17,  1.49it/s]
  9%|▉         | 663/7308 [09:05<1:13:45,  1.50it/s]
  9%|▉         | 664/7308 [09:06<1:13:24,  1.51it/s]
  9%|▉         | 665/7308 [09:06<1:13:09,  1.51it/s]
  9%|▉         | 666/7308 [09:07<1:12:55,  1.52it/s]
  9%|▉         | 667/7308 [09:08<1:12:49,  1.52it/s]
  9%|▉         | 668/7308 [09:08<1:12:42,  1.52it/s]
  9%|▉         | 669/7308 [09:09<1:12:39,  1.52it/s]
  9%|▉         | 670/7308 [09:10<1:12:34,  1.52it/s]
                                                    
{'loss': '4.716', 'grad_norm': '0.3853', 'learning_rate': '0.0006', 'epoch': '18.62'}
+
  9%|▉         | 670/7308 [09:10<1:12:34,  1.52it/s]
  9%|▉         | 671/7308 [09:10<1:12:33,  1.52it/s]
  9%|▉         | 672/7308 [09:11<1:12:30,  1.53it/s]
  9%|▉         | 673/7308 [09:12<1:12:29,  1.53it/s]
  9%|▉         | 674/7308 [09:12<1:12:27,  1.53it/s]
  9%|▉         | 675/7308 [09:13<1:12:25,  1.53it/s]
  9%|▉         | 676/7308 [09:14<1:12:30,  1.52it/s]
  9%|▉         | 677/7308 [09:14<1:12:29,  1.52it/s]
  9%|▉         | 678/7308 [09:15<1:12:30,  1.52it/s]
  9%|▉         | 679/7308 [09:15<1:12:31,  1.52it/s]
  9%|▉         | 680/7308 [09:16<1:12:26,  1.52it/s]
                                                    
{'loss': '4.695', 'grad_norm': '0.2664', 'learning_rate': '0.0006', 'epoch': '18.9'}
+
  9%|▉         | 680/7308 [09:16<1:12:26,  1.52it/s]
  9%|▉         | 681/7308 [09:17<1:12:25,  1.52it/s]
  9%|▉         | 682/7308 [09:17<1:12:26,  1.52it/s]
  9%|▉         | 683/7308 [09:18<1:12:22,  1.53it/s]
  9%|▉         | 684/7308 [09:18<58:57,  1.87it/s]  
  9%|▉         | 685/7308 [09:19<1:06:15,  1.67it/s]
  9%|▉         | 686/7308 [09:20<1:08:04,  1.62it/s]
  9%|▉         | 687/7308 [09:20<1:09:22,  1.59it/s]
  9%|▉         | 688/7308 [09:21<1:10:12,  1.57it/s]
  9%|▉         | 689/7308 [09:22<1:10:50,  1.56it/s]
  9%|▉         | 690/7308 [09:22<1:11:17,  1.55it/s]
                                                    
{'loss': '4.651', 'grad_norm': '0.2969', 'learning_rate': '0.0006', 'epoch': '19.17'}
+
  9%|▉         | 690/7308 [09:22<1:11:17,  1.55it/s]
  9%|▉         | 691/7308 [09:23<1:11:38,  1.54it/s]
  9%|▉         | 692/7308 [09:24<1:11:48,  1.54it/s]
  9%|▉         | 693/7308 [09:24<1:11:55,  1.53it/s]
  9%|▉         | 694/7308 [09:25<1:12:00,  1.53it/s]
 10%|▉         | 695/7308 [09:26<1:12:07,  1.53it/s]
 10%|▉         | 696/7308 [09:26<1:12:08,  1.53it/s]
 10%|▉         | 697/7308 [09:27<1:12:05,  1.53it/s]
 10%|▉         | 698/7308 [09:28<1:12:05,  1.53it/s]
 10%|▉         | 699/7308 [09:28<1:12:13,  1.53it/s]
 10%|▉         | 700/7308 [09:29<1:12:14,  1.52it/s]
                                                    
{'loss': '4.62', 'grad_norm': '0.3466', 'learning_rate': '0.0006', 'epoch': '19.45'}
+
 10%|▉         | 700/7308 [09:29<1:12:14,  1.52it/s]
 10%|▉         | 701/7308 [09:30<1:12:13,  1.52it/s]
 10%|▉         | 702/7308 [09:30<1:12:08,  1.53it/s]
 10%|▉         | 703/7308 [09:31<1:12:06,  1.53it/s]
 10%|▉         | 704/7308 [09:32<1:12:05,  1.53it/s]
 10%|▉         | 705/7308 [09:32<1:12:06,  1.53it/s]
 10%|▉         | 706/7308 [09:33<1:12:05,  1.53it/s]
 10%|▉         | 707/7308 [09:34<1:12:03,  1.53it/s]
 10%|▉         | 708/7308 [09:34<1:12:01,  1.53it/s]
 10%|▉         | 709/7308 [09:35<1:11:58,  1.53it/s]
 10%|▉         | 710/7308 [09:36<1:12:00,  1.53it/s]
                                                    
{'loss': '4.623', 'grad_norm': '0.3388', 'learning_rate': '0.0006', 'epoch': '19.73'}
+
 10%|▉         | 710/7308 [09:36<1:12:00,  1.53it/s]
 10%|▉         | 711/7308 [09:36<1:12:03,  1.53it/s]
 10%|▉         | 712/7308 [09:37<1:12:01,  1.53it/s]
 10%|▉         | 713/7308 [09:37<1:12:00,  1.53it/s]
 10%|▉         | 714/7308 [09:38<1:12:00,  1.53it/s]
 10%|▉         | 715/7308 [09:39<1:11:56,  1.53it/s]
 10%|▉         | 716/7308 [09:39<1:11:54,  1.53it/s]
 10%|▉         | 717/7308 [09:40<1:11:51,  1.53it/s]
 10%|▉         | 718/7308 [09:41<1:11:50,  1.53it/s]
 10%|▉         | 719/7308 [09:41<1:11:48,  1.53it/s]
 10%|▉         | 720/7308 [09:42<58:30,  1.88it/s]  
                                                  
{'loss': '4.637', 'grad_norm': '0.4601', 'learning_rate': '0.0006', 'epoch': '20'}
+
 10%|▉         | 720/7308 [09:42<58:30,  1.88it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 19.05it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.92it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.70it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.09it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.79it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.58it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.47it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.37it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.34it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.28it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.29it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.26it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.26it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.26it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.25it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.24it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.24it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.24it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s]
                                                  
+
                                               
{'eval_loss': '4.474', 'eval_runtime': '3.24', 'eval_samples_per_second': '753.4', 'eval_steps_per_second': '12.04', 'epoch': '20'}
+
 10%|▉         | 720/7308 [09:45<58:30,  1.88it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-720
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-720/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-720/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.93it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.93it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-720/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-720/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-720/tokenizer_config.json
+[eval] epoch 20.0 step 720 loss 4.4744 (best 4.4744 @ ep 20.0, 0/20 since best)
+
+
                                               
 10%|▉         | 721/7308 [09:48<3:59:00,  2.18s/it]
 10%|▉         | 722/7308 [09:48<3:08:53,  1.72s/it]
 10%|▉         | 723/7308 [09:49<2:33:46,  1.40s/it]
 10%|▉         | 724/7308 [09:50<2:09:10,  1.18s/it]
 10%|▉         | 725/7308 [09:50<1:51:59,  1.02s/it]
 10%|▉         | 726/7308 [09:51<1:39:57,  1.10it/s]
 10%|▉         | 727/7308 [09:52<1:31:30,  1.20it/s]
 10%|▉         | 728/7308 [09:52<1:25:36,  1.28it/s]
 10%|▉         | 729/7308 [09:53<1:21:27,  1.35it/s]
 10%|▉         | 730/7308 [09:54<1:18:38,  1.39it/s]
                                                    
{'loss': '4.572', 'grad_norm': '0.4557', 'learning_rate': '0.0006', 'epoch': '20.28'}
+
 10%|▉         | 730/7308 [09:54<1:18:38,  1.39it/s]
 10%|█         | 731/7308 [09:54<1:16:39,  1.43it/s]
 10%|█         | 732/7308 [09:55<1:15:09,  1.46it/s]
 10%|█         | 733/7308 [09:56<1:14:10,  1.48it/s]
 10%|█         | 734/7308 [09:56<1:13:24,  1.49it/s]
 10%|█         | 735/7308 [09:57<1:12:56,  1.50it/s]
 10%|█         | 736/7308 [09:57<1:12:34,  1.51it/s]
 10%|█         | 737/7308 [09:58<1:12:20,  1.51it/s]
 10%|█         | 738/7308 [09:59<1:12:09,  1.52it/s]
 10%|█         | 739/7308 [09:59<1:12:02,  1.52it/s]
 10%|█         | 740/7308 [10:00<1:11:55,  1.52it/s]
                                                    
{'loss': '4.582', 'grad_norm': '0.3393', 'learning_rate': '0.0006', 'epoch': '20.57'}
+
 10%|█         | 740/7308 [10:00<1:11:55,  1.52it/s]
 10%|█         | 741/7308 [10:01<1:11:52,  1.52it/s]
 10%|█         | 742/7308 [10:01<1:11:52,  1.52it/s]
 10%|█         | 743/7308 [10:02<1:11:48,  1.52it/s]
 10%|█         | 744/7308 [10:03<1:11:53,  1.52it/s]
 10%|█         | 745/7308 [10:03<1:11:49,  1.52it/s]
 10%|█         | 746/7308 [10:04<1:11:47,  1.52it/s]
 10%|█         | 747/7308 [10:05<1:11:45,  1.52it/s]
 10%|█         | 748/7308 [10:05<1:11:44,  1.52it/s]
 10%|█         | 749/7308 [10:06<1:11:44,  1.52it/s]
 10%|█         | 750/7308 [10:07<1:11:44,  1.52it/s]
                                                    
{'loss': '4.557', 'grad_norm': '0.219', 'learning_rate': '0.0006', 'epoch': '20.85'}
+
 10%|█         | 750/7308 [10:07<1:11:44,  1.52it/s]
 10%|█         | 751/7308 [10:07<1:11:46,  1.52it/s]
 10%|█         | 752/7308 [10:08<1:11:40,  1.52it/s]
 10%|█         | 753/7308 [10:09<1:11:41,  1.52it/s]
 10%|█         | 754/7308 [10:09<1:11:42,  1.52it/s]
 10%|█         | 755/7308 [10:10<1:11:35,  1.53it/s]
 10%|█         | 756/7308 [10:10<58:18,  1.87it/s]  
 10%|█         | 757/7308 [10:11<1:04:44,  1.69it/s]
 10%|█         | 758/7308 [10:12<1:06:47,  1.63it/s]
 10%|█         | 759/7308 [10:12<1:08:10,  1.60it/s]
 10%|█         | 760/7308 [10:13<1:09:09,  1.58it/s]
                                                    
{'loss': '4.529', 'grad_norm': '0.2502', 'learning_rate': '0.0006', 'epoch': '21.11'}
+
 10%|█         | 760/7308 [10:13<1:09:09,  1.58it/s]
 10%|█         | 761/7308 [10:14<1:09:56,  1.56it/s]
 10%|█         | 762/7308 [10:14<1:10:25,  1.55it/s]
 10%|█         | 763/7308 [10:15<1:10:43,  1.54it/s]
 10%|█         | 764/7308 [10:16<1:10:57,  1.54it/s]
 10%|█         | 765/7308 [10:16<1:11:04,  1.53it/s]
 10%|█         | 766/7308 [10:17<1:11:12,  1.53it/s]
 10%|█         | 767/7308 [10:17<1:11:18,  1.53it/s]
 11%|█         | 768/7308 [10:18<1:11:17,  1.53it/s]
 11%|█         | 769/7308 [10:19<1:11:17,  1.53it/s]
 11%|█         | 770/7308 [10:19<1:11:21,  1.53it/s]
                                                    
{'loss': '4.492', 'grad_norm': '0.2809', 'learning_rate': '0.0006', 'epoch': '21.4'}
+
 11%|█         | 770/7308 [10:19<1:11:21,  1.53it/s]
 11%|█         | 771/7308 [10:20<1:11:26,  1.52it/s]
 11%|█         | 772/7308 [10:21<1:11:24,  1.53it/s]
 11%|█         | 773/7308 [10:21<1:11:22,  1.53it/s]
 11%|█         | 774/7308 [10:22<1:11:26,  1.52it/s]
 11%|█         | 775/7308 [10:23<1:11:31,  1.52it/s]
 11%|█         | 776/7308 [10:23<1:11:33,  1.52it/s]
 11%|█         | 777/7308 [10:24<1:11:37,  1.52it/s]
 11%|█         | 778/7308 [10:25<1:11:39,  1.52it/s]
 11%|█         | 779/7308 [10:25<1:11:42,  1.52it/s]
 11%|█         | 780/7308 [10:26<1:11:42,  1.52it/s]
                                                    
{'loss': '4.489', 'grad_norm': '0.3456', 'learning_rate': '0.0006', 'epoch': '21.68'}
+
 11%|█         | 780/7308 [10:26<1:11:42,  1.52it/s]
 11%|█         | 781/7308 [10:27<1:11:47,  1.52it/s]
 11%|█         | 782/7308 [10:27<1:11:46,  1.52it/s]
 11%|█         | 783/7308 [10:28<1:11:44,  1.52it/s]
 11%|█         | 784/7308 [10:29<1:11:41,  1.52it/s]
 11%|█         | 785/7308 [10:29<1:11:42,  1.52it/s]
 11%|█         | 786/7308 [10:30<1:11:42,  1.52it/s]
 11%|█         | 787/7308 [10:31<1:11:40,  1.52it/s]
 11%|█         | 788/7308 [10:31<1:11:32,  1.52it/s]
 11%|█         | 789/7308 [10:32<1:11:28,  1.52it/s]
 11%|█         | 790/7308 [10:33<1:11:23,  1.52it/s]
                                                    
{'loss': '4.494', 'grad_norm': '0.4644', 'learning_rate': '0.0006', 'epoch': '21.96'}
+
 11%|█         | 790/7308 [10:33<1:11:23,  1.52it/s]
 11%|█         | 791/7308 [10:33<1:11:20,  1.52it/s]
 11%|█         | 792/7308 [10:34<58:10,  1.87it/s]  [transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.59it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.80it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.63it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.07it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.76it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.58it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.46it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.39it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.33it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.30it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.29it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.27it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.25it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.24it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.24it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.24it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.24it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.25it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s]
                                                  
+
                                               
{'eval_loss': '4.343', 'eval_runtime': '3.243', 'eval_samples_per_second': '752.6', 'eval_steps_per_second': '12.03', 'epoch': '22'}
+
 11%|█         | 792/7308 [10:37<58:10,  1.87it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-792
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-792/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-792/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.79it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.79it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-792/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-792/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-792/tokenizer_config.json
+[eval] epoch 22.0 step 792 loss 4.3429 (best 4.3429 @ ep 22.0, 0/20 since best)
+
+
                                               
 11%|█         | 793/7308 [10:40<3:58:07,  2.19s/it]
 11%|█         | 794/7308 [10:40<3:08:01,  1.73s/it]
 11%|█         | 795/7308 [10:41<2:32:53,  1.41s/it]
 11%|█         | 796/7308 [10:42<2:08:24,  1.18s/it]
 11%|█         | 797/7308 [10:42<1:51:17,  1.03s/it]
 11%|█         | 798/7308 [10:43<1:39:16,  1.09it/s]
 11%|█         | 799/7308 [10:44<1:30:48,  1.19it/s]
 11%|█         | 800/7308 [10:44<1:24:50,  1.28it/s]
                                                    
{'loss': '4.467', 'grad_norm': '0.4552', 'learning_rate': '0.0006', 'epoch': '22.23'}
+
 11%|█         | 800/7308 [10:44<1:24:50,  1.28it/s]
 11%|█         | 801/7308 [10:45<1:20:40,  1.34it/s]
 11%|█         | 802/7308 [10:46<1:17:49,  1.39it/s]
 11%|█         | 803/7308 [10:46<1:15:45,  1.43it/s]
 11%|█         | 804/7308 [10:47<1:14:19,  1.46it/s]
 11%|█         | 805/7308 [10:47<1:13:16,  1.48it/s]
 11%|█         | 806/7308 [10:48<1:12:36,  1.49it/s]
 11%|█         | 807/7308 [10:49<1:12:08,  1.50it/s]
 11%|█         | 808/7308 [10:49<1:11:49,  1.51it/s]
 11%|█         | 809/7308 [10:50<1:11:31,  1.51it/s]
 11%|█         | 810/7308 [10:51<1:11:21,  1.52it/s]
                                                    
{'loss': '4.453', 'grad_norm': '0.3256', 'learning_rate': '0.0006', 'epoch': '22.51'}
+
 11%|█         | 810/7308 [10:51<1:11:21,  1.52it/s]
 11%|█         | 811/7308 [10:51<1:11:13,  1.52it/s]
 11%|█         | 812/7308 [10:52<1:11:05,  1.52it/s]
 11%|█         | 813/7308 [10:53<1:10:57,  1.53it/s]
 11%|█         | 814/7308 [10:53<1:10:56,  1.53it/s]
 11%|█         | 815/7308 [10:54<1:10:52,  1.53it/s]
 11%|█         | 816/7308 [10:55<1:10:54,  1.53it/s]
 11%|█         | 817/7308 [10:55<1:10:52,  1.53it/s]
 11%|█         | 818/7308 [10:56<1:10:50,  1.53it/s]
 11%|█         | 819/7308 [10:57<1:10:47,  1.53it/s]
 11%|█         | 820/7308 [10:57<1:10:46,  1.53it/s]
                                                    
{'loss': '4.449', 'grad_norm': '0.2782', 'learning_rate': '0.0006', 'epoch': '22.79'}
+
 11%|█         | 820/7308 [10:57<1:10:46,  1.53it/s]
 11%|█         | 821/7308 [10:58<1:10:50,  1.53it/s]
 11%|█         | 822/7308 [10:59<1:10:47,  1.53it/s]
 11%|█▏        | 823/7308 [10:59<1:10:43,  1.53it/s]
 11%|█▏        | 824/7308 [11:00<1:10:43,  1.53it/s]
 11%|█▏        | 825/7308 [11:01<1:10:44,  1.53it/s]
 11%|█▏        | 826/7308 [11:01<1:10:43,  1.53it/s]
 11%|█▏        | 827/7308 [11:02<1:10:42,  1.53it/s]
 11%|█▏        | 828/7308 [11:02<57:38,  1.87it/s]  
 11%|█▏        | 829/7308 [11:03<1:04:20,  1.68it/s]
 11%|█▏        | 830/7308 [11:04<1:06:22,  1.63it/s]
                                                    
{'loss': '4.424', 'grad_norm': '0.2985', 'learning_rate': '0.0006', 'epoch': '23.06'}
+
 11%|█▏        | 830/7308 [11:04<1:06:22,  1.63it/s]
 11%|█▏        | 831/7308 [11:04<1:07:44,  1.59it/s]
 11%|█▏        | 832/7308 [11:05<1:08:39,  1.57it/s]
 11%|█▏        | 833/7308 [11:05<1:09:15,  1.56it/s]
 11%|█▏        | 834/7308 [11:06<1:09:40,  1.55it/s]
 11%|█▏        | 835/7308 [11:07<1:09:58,  1.54it/s]
 11%|█▏        | 836/7308 [11:07<1:10:10,  1.54it/s]
 11%|█▏        | 837/7308 [11:08<1:10:18,  1.53it/s]
 11%|█▏        | 838/7308 [11:09<1:10:23,  1.53it/s]
 11%|█▏        | 839/7308 [11:09<1:10:28,  1.53it/s]
 11%|█▏        | 840/7308 [11:10<1:10:32,  1.53it/s]
                                                    
{'loss': '4.376', 'grad_norm': '0.3831', 'learning_rate': '0.0006', 'epoch': '23.34'}
+
 11%|█▏        | 840/7308 [11:10<1:10:32,  1.53it/s]
 12%|█▏        | 841/7308 [11:11<1:10:32,  1.53it/s]
 12%|█▏        | 842/7308 [11:11<1:10:29,  1.53it/s]
 12%|█▏        | 843/7308 [11:12<1:10:34,  1.53it/s]
 12%|█▏        | 844/7308 [11:13<1:10:35,  1.53it/s]
 12%|█▏        | 845/7308 [11:13<1:10:31,  1.53it/s]
 12%|█▏        | 846/7308 [11:14<1:10:30,  1.53it/s]
 12%|█▏        | 847/7308 [11:15<1:10:32,  1.53it/s]
 12%|█▏        | 848/7308 [11:15<1:10:36,  1.53it/s]
 12%|█▏        | 849/7308 [11:16<1:10:36,  1.52it/s]
 12%|█▏        | 850/7308 [11:17<1:10:34,  1.53it/s]
                                                    
{'loss': '4.376', 'grad_norm': '0.3414', 'learning_rate': '0.0006', 'epoch': '23.62'}
+
 12%|█▏        | 850/7308 [11:17<1:10:34,  1.53it/s]
 12%|█▏        | 851/7308 [11:17<1:10:32,  1.53it/s]
 12%|█▏        | 852/7308 [11:18<1:10:33,  1.53it/s]
 12%|█▏        | 853/7308 [11:19<1:10:33,  1.52it/s]
 12%|█▏        | 854/7308 [11:19<1:10:29,  1.53it/s]
 12%|█▏        | 855/7308 [11:20<1:10:25,  1.53it/s]
 12%|█▏        | 856/7308 [11:21<1:10:28,  1.53it/s]
 12%|█▏        | 857/7308 [11:21<1:10:27,  1.53it/s]
 12%|█▏        | 858/7308 [11:22<1:10:24,  1.53it/s]
 12%|█▏        | 859/7308 [11:23<1:10:23,  1.53it/s]
 12%|█▏        | 860/7308 [11:23<1:10:25,  1.53it/s]
                                                    
{'loss': '4.389', 'grad_norm': '0.2493', 'learning_rate': '0.0006', 'epoch': '23.9'}
+
 12%|█▏        | 860/7308 [11:23<1:10:25,  1.53it/s]
 12%|█▏        | 861/7308 [11:24<1:10:28,  1.52it/s]
 12%|█▏        | 862/7308 [11:24<1:10:27,  1.52it/s]
 12%|█▏        | 863/7308 [11:25<1:10:23,  1.53it/s]
 12%|█▏        | 864/7308 [11:25<57:21,  1.87it/s]  [transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.70it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.80it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.63it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.06it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.78it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.57it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.47it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.39it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.36it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.28it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.29it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.26it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.26it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.25it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.24it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.25it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.25it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.24it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s]
                                                  
+
                                               
{'eval_loss': '4.263', 'eval_runtime': '3.247', 'eval_samples_per_second': '751.9', 'eval_steps_per_second': '12.01', 'epoch': '24'}
+
 12%|█▏        | 864/7308 [11:29<57:21,  1.87it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-864
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-864/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-864/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.93it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.93it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-864/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-864/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-864/tokenizer_config.json
+[eval] epoch 24.0 step 864 loss 4.2627 (best 4.2627 @ ep 24.0, 0/20 since best)
+
+
                                               
 12%|█▏        | 865/7308 [11:31<3:53:59,  2.18s/it]
 12%|█▏        | 866/7308 [11:32<3:04:47,  1.72s/it]
 12%|█▏        | 867/7308 [11:33<2:30:23,  1.40s/it]
 12%|█▏        | 868/7308 [11:33<2:06:24,  1.18s/it]
 12%|█▏        | 869/7308 [11:34<1:49:37,  1.02s/it]
 12%|█▏        | 870/7308 [11:35<1:37:50,  1.10it/s]
                                                    
{'loss': '4.338', 'grad_norm': '0.3128', 'learning_rate': '0.0006', 'epoch': '24.17'}
+
 12%|█▏        | 870/7308 [11:35<1:37:50,  1.10it/s]
 12%|█▏        | 871/7308 [11:35<1:29:33,  1.20it/s]
 12%|█▏        | 872/7308 [11:36<1:23:46,  1.28it/s]
 12%|█▏        | 873/7308 [11:37<1:19:43,  1.35it/s]
 12%|█▏        | 874/7308 [11:37<1:16:54,  1.39it/s]
 12%|█▏        | 875/7308 [11:38<1:14:52,  1.43it/s]
 12%|█▏        | 876/7308 [11:39<1:13:26,  1.46it/s]
 12%|█▏        | 877/7308 [11:39<1:12:30,  1.48it/s]
 12%|█▏        | 878/7308 [11:40<1:11:49,  1.49it/s]
 12%|█▏        | 879/7308 [11:41<1:11:17,  1.50it/s]
 12%|█▏        | 880/7308 [11:41<1:10:58,  1.51it/s]
                                                    
{'loss': '4.323', 'grad_norm': '0.3411', 'learning_rate': '0.0006', 'epoch': '24.45'}
+
 12%|█▏        | 880/7308 [11:41<1:10:58,  1.51it/s]
 12%|█▏        | 881/7308 [11:42<1:10:46,  1.51it/s]
 12%|█▏        | 882/7308 [11:43<1:10:36,  1.52it/s]
 12%|█▏        | 883/7308 [11:43<1:10:26,  1.52it/s]
 12%|█▏        | 884/7308 [11:44<1:10:22,  1.52it/s]
 12%|█▏        | 885/7308 [11:45<1:10:17,  1.52it/s]
 12%|█▏        | 886/7308 [11:45<1:10:19,  1.52it/s]
 12%|█▏        | 887/7308 [11:46<1:10:20,  1.52it/s]
 12%|█▏        | 888/7308 [11:46<1:10:18,  1.52it/s]
 12%|█▏        | 889/7308 [11:47<1:10:14,  1.52it/s]
 12%|█▏        | 890/7308 [11:48<1:10:11,  1.52it/s]
                                                    
{'loss': '4.344', 'grad_norm': '0.4023', 'learning_rate': '0.0006', 'epoch': '24.73'}
+
 12%|█▏        | 890/7308 [11:48<1:10:11,  1.52it/s]
 12%|█▏        | 891/7308 [11:48<1:10:11,  1.52it/s]
 12%|█▏        | 892/7308 [11:49<1:10:13,  1.52it/s]
 12%|█▏        | 893/7308 [11:50<1:10:13,  1.52it/s]
 12%|█▏        | 894/7308 [11:50<1:10:11,  1.52it/s]
 12%|█▏        | 895/7308 [11:51<1:10:10,  1.52it/s]
 12%|█▏        | 896/7308 [11:52<1:10:08,  1.52it/s]
 12%|█▏        | 897/7308 [11:52<1:10:07,  1.52it/s]
 12%|█▏        | 898/7308 [11:53<1:10:06,  1.52it/s]
 12%|█▏        | 899/7308 [11:54<1:10:01,  1.53it/s]
 12%|█▏        | 900/7308 [11:54<57:06,  1.87it/s]  
                                                  
{'loss': '4.336', 'grad_norm': '0.3676', 'learning_rate': '0.0006', 'epoch': '25'}
+
 12%|█▏        | 900/7308 [11:54<57:06,  1.87it/s]
 12%|█▏        | 901/7308 [11:55<1:03:16,  1.69it/s]
 12%|█▏        | 902/7308 [11:55<1:05:15,  1.64it/s]
 12%|█▏        | 903/7308 [11:56<1:06:41,  1.60it/s]
 12%|█▏        | 904/7308 [11:57<1:07:44,  1.58it/s]
 12%|█▏        | 905/7308 [11:57<1:08:25,  1.56it/s]
 12%|█▏        | 906/7308 [11:58<1:08:49,  1.55it/s]
 12%|█▏        | 907/7308 [11:59<1:09:08,  1.54it/s]
 12%|█▏        | 908/7308 [11:59<1:09:21,  1.54it/s]
 12%|█▏        | 909/7308 [12:00<1:09:31,  1.53it/s]
 12%|█▏        | 910/7308 [12:01<1:09:36,  1.53it/s]
                                                    
{'loss': '4.272', 'grad_norm': '0.4201', 'learning_rate': '0.0006', 'epoch': '25.28'}
+
 12%|█▏        | 910/7308 [12:01<1:09:36,  1.53it/s]
 12%|█▏        | 911/7308 [12:01<1:09:40,  1.53it/s]
 12%|█▏        | 912/7308 [12:02<1:09:42,  1.53it/s]
 12%|█▏        | 913/7308 [12:03<1:09:45,  1.53it/s]
 13%|█▎        | 914/7308 [12:03<1:09:46,  1.53it/s]
 13%|█▎        | 915/7308 [12:04<1:09:44,  1.53it/s]
 13%|█▎        | 916/7308 [12:05<1:09:44,  1.53it/s]
 13%|█▎        | 917/7308 [12:05<1:09:45,  1.53it/s]
 13%|█▎        | 918/7308 [12:06<1:09:43,  1.53it/s]
 13%|█▎        | 919/7308 [12:06<1:09:43,  1.53it/s]
 13%|█▎        | 920/7308 [12:07<1:09:42,  1.53it/s]
                                                    
{'loss': '4.288', 'grad_norm': '0.3632', 'learning_rate': '0.0006', 'epoch': '25.57'}
+
 13%|█▎        | 920/7308 [12:07<1:09:42,  1.53it/s]
 13%|█▎        | 921/7308 [12:08<1:09:45,  1.53it/s]
 13%|█▎        | 922/7308 [12:08<1:09:43,  1.53it/s]
 13%|█▎        | 923/7308 [12:09<1:09:41,  1.53it/s]
 13%|█▎        | 924/7308 [12:10<1:09:39,  1.53it/s]
 13%|█▎        | 925/7308 [12:10<1:09:40,  1.53it/s]
 13%|█▎        | 926/7308 [12:11<1:09:40,  1.53it/s]
 13%|█▎        | 927/7308 [12:12<1:09:39,  1.53it/s]
 13%|█▎        | 928/7308 [12:12<1:09:38,  1.53it/s]
 13%|█▎        | 929/7308 [12:13<1:09:37,  1.53it/s]
 13%|█▎        | 930/7308 [12:14<1:09:40,  1.53it/s]
                                                    
{'loss': '4.28', 'grad_norm': '0.2819', 'learning_rate': '0.0006', 'epoch': '25.85'}
+
 13%|█▎        | 930/7308 [12:14<1:09:40,  1.53it/s]
 13%|█▎        | 931/7308 [12:14<1:09:37,  1.53it/s]
 13%|█▎        | 932/7308 [12:15<1:09:36,  1.53it/s]
 13%|█▎        | 933/7308 [12:16<1:09:37,  1.53it/s]
 13%|█▎        | 934/7308 [12:16<1:09:36,  1.53it/s]
 13%|█▎        | 935/7308 [12:17<1:09:33,  1.53it/s]
 13%|█▎        | 936/7308 [12:17<56:42,  1.87it/s]  [transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.43it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.78it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.59it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.07it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.76it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.59it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.47it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.40it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.33it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.31it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.30it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.27it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.28it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.26it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.24it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.25it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.25it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.25it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.25it/s]
                                                  
+
                                               
{'eval_loss': '4.215', 'eval_runtime': '3.256', 'eval_samples_per_second': '749.7', 'eval_steps_per_second': '11.98', 'epoch': '26'}
+
 13%|█▎        | 936/7308 [12:20<56:42,  1.87it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.25it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-936
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-936/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-936/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.62it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.62it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-936/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-936/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-936/tokenizer_config.json
+[eval] epoch 26.0 step 936 loss 4.2153 (best 4.2153 @ ep 26.0, 0/20 since best)
+
+
                                               
 13%|█▎        | 937/7308 [12:23<3:55:03,  2.21s/it]
 13%|█▎        | 938/7308 [12:24<3:05:20,  1.75s/it]
 13%|█▎        | 939/7308 [12:25<2:30:35,  1.42s/it]
 13%|█▎        | 940/7308 [12:25<2:06:14,  1.19s/it]
                                                    
{'loss': '4.259', 'grad_norm': '0.3426', 'learning_rate': '0.0006', 'epoch': '26.11'}
+
 13%|█▎        | 940/7308 [12:25<2:06:14,  1.19s/it]
 13%|█▎        | 941/7308 [12:26<1:49:11,  1.03s/it]
 13%|█▎        | 942/7308 [12:27<1:37:15,  1.09it/s]
 13%|█▎        | 943/7308 [12:27<1:28:56,  1.19it/s]
 13%|█▎        | 944/7308 [12:28<1:23:05,  1.28it/s]
 13%|█▎        | 945/7308 [12:29<1:18:57,  1.34it/s]
 13%|█▎        | 946/7308 [12:29<1:16:04,  1.39it/s]
 13%|█▎        | 947/7308 [12:30<1:14:07,  1.43it/s]
 13%|█▎        | 948/7308 [12:31<1:12:42,  1.46it/s]
 13%|█▎        | 949/7308 [12:31<1:11:41,  1.48it/s]
 13%|█▎        | 950/7308 [12:32<1:10:59,  1.49it/s]
                                                    
{'loss': '4.223', 'grad_norm': '0.2899', 'learning_rate': '0.0006', 'epoch': '26.4'}
+
 13%|█▎        | 950/7308 [12:32<1:10:59,  1.49it/s]
 13%|█▎        | 951/7308 [12:33<1:10:37,  1.50it/s]
 13%|█▎        | 952/7308 [12:33<1:10:18,  1.51it/s]
 13%|█▎        | 953/7308 [12:34<1:10:11,  1.51it/s]
 13%|█▎        | 954/7308 [12:35<1:10:06,  1.51it/s]
 13%|█▎        | 955/7308 [12:35<1:09:57,  1.51it/s]
 13%|█▎        | 956/7308 [12:36<1:09:49,  1.52it/s]
 13%|█▎        | 957/7308 [12:36<1:09:45,  1.52it/s]
 13%|█▎        | 958/7308 [12:37<1:09:37,  1.52it/s]
 13%|█▎        | 959/7308 [12:38<1:09:34,  1.52it/s]
 13%|█▎        | 960/7308 [12:38<1:09:32,  1.52it/s]
                                                    
{'loss': '4.23', 'grad_norm': '0.2744', 'learning_rate': '0.0006', 'epoch': '26.68'}
+
 13%|█▎        | 960/7308 [12:38<1:09:32,  1.52it/s]
 13%|█▎        | 961/7308 [12:39<1:09:34,  1.52it/s]
 13%|█▎        | 962/7308 [12:40<1:09:32,  1.52it/s]
 13%|█▎        | 963/7308 [12:40<1:09:26,  1.52it/s]
 13%|█▎        | 964/7308 [12:41<1:09:22,  1.52it/s]
 13%|█▎        | 965/7308 [12:42<1:09:18,  1.53it/s]
 13%|█▎        | 966/7308 [12:42<1:09:21,  1.52it/s]
 13%|█▎        | 967/7308 [12:43<1:09:21,  1.52it/s]
 13%|█▎        | 968/7308 [12:44<1:09:17,  1.52it/s]
 13%|█▎        | 969/7308 [12:44<1:09:16,  1.52it/s]
 13%|█▎        | 970/7308 [12:45<1:09:15,  1.53it/s]
                                                    
{'loss': '4.241', 'grad_norm': '0.3075', 'learning_rate': '0.0006', 'epoch': '26.96'}
+
 13%|█▎        | 970/7308 [12:45<1:09:15,  1.53it/s]
 13%|█▎        | 971/7308 [12:46<1:09:15,  1.53it/s]
 13%|█▎        | 972/7308 [12:46<56:25,  1.87it/s]  
 13%|█▎        | 973/7308 [12:47<1:03:19,  1.67it/s]
 13%|█▎        | 974/7308 [12:47<1:05:06,  1.62it/s]
 13%|█▎        | 975/7308 [12:48<1:06:18,  1.59it/s]
 13%|█▎        | 976/7308 [12:49<1:07:07,  1.57it/s]
 13%|█▎        | 977/7308 [12:49<1:07:42,  1.56it/s]
 13%|█▎        | 978/7308 [12:50<1:08:08,  1.55it/s]
 13%|█▎        | 979/7308 [12:51<1:08:25,  1.54it/s]
 13%|█▎        | 980/7308 [12:51<1:08:36,  1.54it/s]
                                                    
{'loss': '4.199', 'grad_norm': '0.3883', 'learning_rate': '0.0006', 'epoch': '27.23'}
+
 13%|█▎        | 980/7308 [12:51<1:08:36,  1.54it/s]
 13%|█▎        | 981/7308 [12:52<1:08:43,  1.53it/s]
 13%|█▎        | 982/7308 [12:53<1:08:53,  1.53it/s]
 13%|█▎        | 983/7308 [12:53<1:09:05,  1.53it/s]
 13%|█▎        | 984/7308 [12:54<1:09:03,  1.53it/s]
 13%|█▎        | 985/7308 [12:55<1:09:00,  1.53it/s]
 13%|█▎        | 986/7308 [12:55<1:08:57,  1.53it/s]
 14%|█▎        | 987/7308 [12:56<1:08:56,  1.53it/s]
 14%|█▎        | 988/7308 [12:56<1:08:52,  1.53it/s]
 14%|█▎        | 989/7308 [12:57<1:08:50,  1.53it/s]
 14%|█▎        | 990/7308 [12:58<1:08:51,  1.53it/s]
                                                    
{'loss': '4.187', 'grad_norm': '0.3422', 'learning_rate': '0.0006', 'epoch': '27.51'}
+
 14%|█▎        | 990/7308 [12:58<1:08:51,  1.53it/s]
 14%|█▎        | 991/7308 [12:58<1:08:52,  1.53it/s]
 14%|█▎        | 992/7308 [12:59<1:08:54,  1.53it/s]
 14%|█▎        | 993/7308 [13:00<1:08:54,  1.53it/s]
 14%|█▎        | 994/7308 [13:00<1:08:53,  1.53it/s]
 14%|█▎        | 995/7308 [13:01<1:08:50,  1.53it/s]
 14%|█▎        | 996/7308 [13:02<1:08:51,  1.53it/s]
 14%|█▎        | 997/7308 [13:02<1:08:54,  1.53it/s]
 14%|█▎        | 998/7308 [13:03<1:08:57,  1.53it/s]
 14%|█▎        | 999/7308 [13:04<1:09:01,  1.52it/s]
 14%|█▎        | 1000/7308 [13:04<1:08:58,  1.52it/s]
                                                     
{'loss': '4.197', 'grad_norm': '0.2933', 'learning_rate': '0.0006', 'epoch': '27.79'}
+
 14%|█▎        | 1000/7308 [13:04<1:08:58,  1.52it/s]
 14%|█▎        | 1001/7308 [13:05<1:08:55,  1.53it/s]
 14%|█▎        | 1002/7308 [13:06<1:08:53,  1.53it/s]
 14%|█▎        | 1003/7308 [13:06<1:08:53,  1.53it/s]
 14%|█▎        | 1004/7308 [13:07<1:08:49,  1.53it/s]
 14%|█▍        | 1005/7308 [13:08<1:08:50,  1.53it/s]
 14%|█▍        | 1006/7308 [13:08<1:08:47,  1.53it/s]
 14%|█▍        | 1007/7308 [13:09<1:08:45,  1.53it/s]
 14%|█▍        | 1008/7308 [13:09<56:00,  1.87it/s]  [transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.67it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.84it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.64it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.09it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.78it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.60it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.49it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.40it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.35it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.31it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.27it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.27it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.27it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.26it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.25it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.26it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.26it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.25it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s]
                                                   
+
                                               
{'eval_loss': '4.183', 'eval_runtime': '3.243', 'eval_samples_per_second': '752.7', 'eval_steps_per_second': '12.03', 'epoch': '28'}
+
 14%|█▍        | 1008/7308 [13:12<56:00,  1.87it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-1008
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-1008/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-1008/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.89it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.89it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-1008/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-1008/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-1008/tokenizer_config.json
+[eval] epoch 28.0 step 1008 loss 4.1834 (best 4.1834 @ ep 28.0, 0/20 since best)
+
+
                                               
 14%|█▍        | 1009/7308 [13:15<3:48:26,  2.18s/it]
 14%|█▍        | 1010/7308 [13:16<3:00:27,  1.72s/it]
                                                     
{'loss': '4.186', 'grad_norm': '0.3529', 'learning_rate': '0.0006', 'epoch': '28.06'}
+
 14%|█▍        | 1010/7308 [13:16<3:00:27,  1.72s/it]
 14%|█▍        | 1011/7308 [13:17<2:26:54,  1.40s/it]
 14%|█▍        | 1012/7308 [13:17<2:03:33,  1.18s/it]
 14%|█▍        | 1013/7308 [13:18<1:47:08,  1.02s/it]
 14%|█▍        | 1014/7308 [13:18<1:35:35,  1.10it/s]
 14%|█▍        | 1015/7308 [13:19<1:27:30,  1.20it/s]
 14%|█▍        | 1016/7308 [13:20<1:21:50,  1.28it/s]
 14%|█▍        | 1017/7308 [13:20<1:17:55,  1.35it/s]
 14%|█▍        | 1018/7308 [13:21<1:15:10,  1.39it/s]
 14%|█▍        | 1019/7308 [13:22<1:13:09,  1.43it/s]
 14%|█▍        | 1020/7308 [13:22<1:11:49,  1.46it/s]
                                                     
{'loss': '4.146', 'grad_norm': '0.3399', 'learning_rate': '0.0006', 'epoch': '28.34'}
+
 14%|█▍        | 1020/7308 [13:22<1:11:49,  1.46it/s]
 14%|█▍        | 1021/7308 [13:23<1:10:55,  1.48it/s]
 14%|█▍        | 1022/7308 [13:24<1:10:19,  1.49it/s]
 14%|█▍        | 1023/7308 [13:24<1:09:54,  1.50it/s]
 14%|█▍        | 1024/7308 [13:25<1:09:31,  1.51it/s]
 14%|█▍        | 1025/7308 [13:26<1:09:13,  1.51it/s]
 14%|█▍        | 1026/7308 [13:26<1:09:03,  1.52it/s]
 14%|█▍        | 1027/7308 [13:27<1:08:56,  1.52it/s]
 14%|█▍        | 1028/7308 [13:28<1:08:50,  1.52it/s]
 14%|█▍        | 1029/7308 [13:28<1:08:44,  1.52it/s]
 14%|█▍        | 1030/7308 [13:29<1:08:41,  1.52it/s]
                                                     
{'loss': '4.148', 'grad_norm': '0.3466', 'learning_rate': '0.0006', 'epoch': '28.62'}
+
 14%|█▍        | 1030/7308 [13:29<1:08:41,  1.52it/s]
 14%|█▍        | 1031/7308 [13:30<1:08:39,  1.52it/s]
 14%|█▍        | 1032/7308 [13:30<1:08:36,  1.52it/s]
 14%|█▍        | 1033/7308 [13:31<1:08:31,  1.53it/s]
 14%|█▍        | 1034/7308 [13:32<1:08:32,  1.53it/s]
 14%|█▍        | 1035/7308 [13:32<1:08:31,  1.53it/s]
 14%|█▍        | 1036/7308 [13:33<1:08:30,  1.53it/s]
 14%|█▍        | 1037/7308 [13:34<1:08:31,  1.53it/s]
 14%|█▍        | 1038/7308 [13:34<1:08:32,  1.52it/s]
 14%|█▍        | 1039/7308 [13:35<1:08:30,  1.53it/s]
 14%|█▍        | 1040/7308 [13:36<1:08:30,  1.52it/s]
                                                     
{'loss': '4.156', 'grad_norm': '0.2728', 'learning_rate': '0.0006', 'epoch': '28.9'}
+
 14%|█▍        | 1040/7308 [13:36<1:08:30,  1.52it/s]
 14%|█▍        | 1041/7308 [13:36<1:08:34,  1.52it/s]
 14%|█▍        | 1042/7308 [13:37<1:08:33,  1.52it/s]
 14%|█▍        | 1043/7308 [13:37<1:08:26,  1.53it/s]
 14%|█▍        | 1044/7308 [13:38<55:46,  1.87it/s]  
 14%|█▍        | 1045/7308 [13:38<1:01:39,  1.69it/s]
 14%|█▍        | 1046/7308 [13:39<1:03:42,  1.64it/s]
 14%|█▍        | 1047/7308 [13:40<1:05:07,  1.60it/s]
 14%|█▍        | 1048/7308 [13:40<1:06:01,  1.58it/s]
 14%|█▍        | 1049/7308 [13:41<1:06:43,  1.56it/s]
 14%|█▍        | 1050/7308 [13:42<1:07:14,  1.55it/s]
                                                     
{'loss': '4.114', 'grad_norm': '0.3656', 'learning_rate': '0.0006', 'epoch': '29.17'}
+
 14%|█▍        | 1050/7308 [13:42<1:07:14,  1.55it/s]
 14%|█▍        | 1051/7308 [13:42<1:07:35,  1.54it/s]
 14%|█▍        | 1052/7308 [13:43<1:07:47,  1.54it/s]
 14%|█▍        | 1053/7308 [13:44<1:07:56,  1.53it/s]
 14%|█▍        | 1054/7308 [13:44<1:08:04,  1.53it/s]
 14%|█▍        | 1055/7308 [13:45<1:08:12,  1.53it/s]
 14%|█▍        | 1056/7308 [13:46<1:08:15,  1.53it/s]
 14%|█▍        | 1057/7308 [13:46<1:08:17,  1.53it/s]
 14%|█▍        | 1058/7308 [13:47<1:08:15,  1.53it/s]
 14%|█▍        | 1059/7308 [13:48<1:08:13,  1.53it/s]
 15%|█▍        | 1060/7308 [13:48<1:08:15,  1.53it/s]
                                                     
{'loss': '4.12', 'grad_norm': '0.4864', 'learning_rate': '0.0006', 'epoch': '29.45'}
+
 15%|█▍        | 1060/7308 [13:48<1:08:15,  1.53it/s]
 15%|█▍        | 1061/7308 [13:49<1:08:16,  1.53it/s]
 15%|█▍        | 1062/7308 [13:50<1:08:13,  1.53it/s]
 15%|█▍        | 1063/7308 [13:50<1:08:11,  1.53it/s]
 15%|█▍        | 1064/7308 [13:51<1:08:12,  1.53it/s]
 15%|█▍        | 1065/7308 [13:52<1:08:13,  1.53it/s]
 15%|█▍        | 1066/7308 [13:52<1:08:11,  1.53it/s]
 15%|█▍        | 1067/7308 [13:53<1:08:09,  1.53it/s]
 15%|█▍        | 1068/7308 [13:54<1:08:09,  1.53it/s]
 15%|█▍        | 1069/7308 [13:54<1:08:15,  1.52it/s]
 15%|█▍        | 1070/7308 [13:55<1:08:15,  1.52it/s]
                                                     
{'loss': '4.122', 'grad_norm': '0.2926', 'learning_rate': '0.0006', 'epoch': '29.73'}
+
 15%|█▍        | 1070/7308 [13:55<1:08:15,  1.52it/s]
 15%|█▍        | 1071/7308 [13:56<1:08:15,  1.52it/s]
 15%|█▍        | 1072/7308 [13:56<1:08:11,  1.52it/s]
 15%|█▍        | 1073/7308 [13:57<1:08:07,  1.53it/s]
 15%|█▍        | 1074/7308 [13:57<1:08:10,  1.52it/s]
 15%|█▍        | 1075/7308 [13:58<1:08:12,  1.52it/s]
 15%|█▍        | 1076/7308 [13:59<1:08:10,  1.52it/s]
 15%|█▍        | 1077/7308 [13:59<1:08:06,  1.52it/s]
 15%|█▍        | 1078/7308 [14:00<1:08:05,  1.52it/s]
 15%|█▍        | 1079/7308 [14:01<1:08:04,  1.53it/s]
 15%|█▍        | 1080/7308 [14:01<55:28,  1.87it/s]  
                                                   
{'loss': '4.12', 'grad_norm': '0.3393', 'learning_rate': '0.0006', 'epoch': '30'}
+
 15%|█▍        | 1080/7308 [14:01<55:28,  1.87it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.59it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.81it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.63it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.07it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.78it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.58it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.48it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.38it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.34it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.31it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.31it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.29it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.27it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.27it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.27it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.26it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.25it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.26it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s]
                                                   
+
                                               
{'eval_loss': '4.161', 'eval_runtime': '3.24', 'eval_samples_per_second': '753.3', 'eval_steps_per_second': '12.04', 'epoch': '30'}
+
 15%|█▍        | 1080/7308 [14:04<55:28,  1.87it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-1080
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-1080/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-1080/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.77it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.77it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-1080/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-1080/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-1080/tokenizer_config.json
+[eval] epoch 30.0 step 1080 loss 4.1612 (best 4.1612 @ ep 30.0, 0/20 since best)
+
+
                                               
 15%|█▍        | 1081/7308 [14:07<3:47:43,  2.19s/it]
 15%|█▍        | 1082/7308 [14:08<2:59:44,  1.73s/it]
 15%|█▍        | 1083/7308 [14:08<2:26:08,  1.41s/it]
 15%|█▍        | 1084/7308 [14:09<2:02:39,  1.18s/it]
 15%|█▍        | 1085/7308 [14:10<1:46:14,  1.02s/it]
 15%|█▍        | 1086/7308 [14:10<1:34:44,  1.09it/s]
 15%|█▍        | 1087/7308 [14:11<1:26:39,  1.20it/s]
 15%|█▍        | 1088/7308 [14:12<1:21:01,  1.28it/s]
 15%|█▍        | 1089/7308 [14:12<1:17:06,  1.34it/s]
 15%|█▍        | 1090/7308 [14:13<1:14:20,  1.39it/s]
                                                     
{'loss': '4.058', 'grad_norm': '0.397', 'learning_rate': '0.0006', 'epoch': '30.28'}
+
 15%|█▍        | 1090/7308 [14:13<1:14:20,  1.39it/s]
 15%|█▍        | 1091/7308 [14:14<1:12:25,  1.43it/s]
 15%|█▍        | 1092/7308 [14:14<1:11:04,  1.46it/s]
 15%|█▍        | 1093/7308 [14:15<1:10:09,  1.48it/s]
 15%|█▍        | 1094/7308 [14:16<1:09:37,  1.49it/s]
 15%|█▍        | 1095/7308 [14:16<1:09:10,  1.50it/s]
 15%|█▍        | 1096/7308 [14:17<1:08:52,  1.50it/s]
 15%|█▌        | 1097/7308 [14:18<1:08:38,  1.51it/s]
 15%|█▌        | 1098/7308 [14:18<1:08:24,  1.51it/s]
 15%|█▌        | 1099/7308 [14:19<1:08:15,  1.52it/s]
 15%|█▌        | 1100/7308 [14:20<1:08:05,  1.52it/s]
                                                     
{'loss': '4.071', 'grad_norm': '0.3184', 'learning_rate': '0.0006', 'epoch': '30.57'}
+
 15%|█▌        | 1100/7308 [14:20<1:08:05,  1.52it/s]
 15%|█▌        | 1101/7308 [14:20<1:08:03,  1.52it/s]
 15%|█▌        | 1102/7308 [14:21<1:07:59,  1.52it/s]
 15%|█▌        | 1103/7308 [14:22<1:07:58,  1.52it/s]
 15%|█▌        | 1104/7308 [14:22<1:07:53,  1.52it/s]
 15%|█▌        | 1105/7308 [14:23<1:07:50,  1.52it/s]
 15%|█▌        | 1106/7308 [14:23<1:07:50,  1.52it/s]
 15%|█▌        | 1107/7308 [14:24<1:07:52,  1.52it/s]
 15%|█▌        | 1108/7308 [14:25<1:07:52,  1.52it/s]
 15%|█▌        | 1109/7308 [14:25<1:07:53,  1.52it/s]
 15%|█▌        | 1110/7308 [14:26<1:07:50,  1.52it/s]
                                                     
{'loss': '4.077', 'grad_norm': '0.2669', 'learning_rate': '0.0006', 'epoch': '30.85'}
+
 15%|█▌        | 1110/7308 [14:26<1:07:50,  1.52it/s]
 15%|█▌        | 1111/7308 [14:27<1:07:48,  1.52it/s]
 15%|█▌        | 1112/7308 [14:27<1:07:45,  1.52it/s]
 15%|█▌        | 1113/7308 [14:28<1:07:46,  1.52it/s]
 15%|█▌        | 1114/7308 [14:29<1:07:48,  1.52it/s]
 15%|█▌        | 1115/7308 [14:29<1:07:46,  1.52it/s]
 15%|█▌        | 1116/7308 [14:30<55:13,  1.87it/s]  
 15%|█▌        | 1117/7308 [14:30<1:01:25,  1.68it/s]
 15%|█▌        | 1118/7308 [14:31<1:03:19,  1.63it/s]
 15%|█▌        | 1119/7308 [14:32<1:04:38,  1.60it/s]
 15%|█▌        | 1120/7308 [14:32<1:05:35,  1.57it/s]
                                                     
{'loss': '4.049', 'grad_norm': '0.2974', 'learning_rate': '0.0006', 'epoch': '31.11'}
+
 15%|█▌        | 1120/7308 [14:32<1:05:35,  1.57it/s]
 15%|█▌        | 1121/7308 [14:33<1:06:13,  1.56it/s]
 15%|█▌        | 1122/7308 [14:34<1:06:34,  1.55it/s]
 15%|█▌        | 1123/7308 [14:34<1:06:54,  1.54it/s]
 15%|█▌        | 1124/7308 [14:35<1:07:09,  1.53it/s]
 15%|█▌        | 1125/7308 [14:36<1:07:19,  1.53it/s]
 15%|█▌        | 1126/7308 [14:36<1:07:23,  1.53it/s]
 15%|█▌        | 1127/7308 [14:37<1:07:21,  1.53it/s]
 15%|█▌        | 1128/7308 [14:38<1:18:19,  1.31it/s]
 15%|█▌        | 1129/7308 [14:39<1:15:02,  1.37it/s]
 15%|█▌        | 1130/7308 [14:39<1:12:45,  1.42it/s]
                                                     
{'loss': '4.019', 'grad_norm': '0.3916', 'learning_rate': '0.0006', 'epoch': '31.4'}
+
 15%|█▌        | 1130/7308 [14:39<1:12:45,  1.42it/s]
 15%|█▌        | 1131/7308 [14:40<1:11:17,  1.44it/s]
 15%|█▌        | 1132/7308 [14:41<1:10:10,  1.47it/s]
 16%|█▌        | 1133/7308 [14:41<1:09:22,  1.48it/s]
 16%|█▌        | 1134/7308 [14:42<1:08:48,  1.50it/s]
 16%|█▌        | 1135/7308 [14:43<1:08:19,  1.51it/s]
 16%|█▌        | 1136/7308 [14:43<1:08:05,  1.51it/s]
 16%|█▌        | 1137/7308 [14:44<1:07:53,  1.52it/s]
 16%|█▌        | 1138/7308 [14:45<1:07:43,  1.52it/s]
 16%|█▌        | 1139/7308 [14:45<1:07:38,  1.52it/s]
 16%|█▌        | 1140/7308 [14:46<1:07:31,  1.52it/s]
                                                     
{'loss': '4.06', 'grad_norm': '0.5596', 'learning_rate': '0.0006', 'epoch': '31.68'}
+
 16%|█▌        | 1140/7308 [14:46<1:07:31,  1.52it/s]
 16%|█▌        | 1141/7308 [14:46<1:07:33,  1.52it/s]
 16%|█▌        | 1142/7308 [14:47<1:07:31,  1.52it/s]
 16%|█▌        | 1143/7308 [14:48<1:07:28,  1.52it/s]
 16%|█▌        | 1144/7308 [14:48<1:07:24,  1.52it/s]
 16%|█▌        | 1145/7308 [14:49<1:07:20,  1.53it/s]
 16%|█▌        | 1146/7308 [14:50<1:07:20,  1.52it/s]
 16%|█▌        | 1147/7308 [14:50<1:07:20,  1.52it/s]
 16%|█▌        | 1148/7308 [14:51<1:07:16,  1.53it/s]
 16%|█▌        | 1149/7308 [14:52<1:07:14,  1.53it/s]
 16%|█▌        | 1150/7308 [14:52<1:07:19,  1.52it/s]
                                                     
{'loss': '4.075', 'grad_norm': '0.3571', 'learning_rate': '0.0006', 'epoch': '31.96'}
+
 16%|█▌        | 1150/7308 [14:52<1:07:19,  1.52it/s]
 16%|█▌        | 1151/7308 [14:53<1:07:20,  1.52it/s]
 16%|█▌        | 1152/7308 [14:53<54:51,  1.87it/s]  [transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.57it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.83it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.62it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.07it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.77it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.58it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.47it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.39it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.34it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.30it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.29it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.28it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.26it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.26it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.25it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.25it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.24it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.24it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s]
                                                   
+
                                               
{'eval_loss': '4.164', 'eval_runtime': '3.245', 'eval_samples_per_second': '752.3', 'eval_steps_per_second': '12.02', 'epoch': '32'}
+
 16%|█▌        | 1152/7308 [14:57<54:51,  1.87it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s][eval] epoch 32.0 step 1152 loss 4.1638 (best 4.1612 @ ep 30.0, 1/20 since best)
+
+
                                               
 16%|█▌        | 1153/7308 [14:57<2:41:03,  1.57s/it]
 16%|█▌        | 1154/7308 [14:58<2:12:50,  1.30s/it]
 16%|█▌        | 1155/7308 [14:59<1:53:05,  1.10s/it]
 16%|█▌        | 1156/7308 [14:59<1:39:18,  1.03it/s]
 16%|█▌        | 1157/7308 [15:00<1:29:38,  1.14it/s]
 16%|█▌        | 1158/7308 [15:01<1:22:52,  1.24it/s]
 16%|█▌        | 1159/7308 [15:01<1:18:09,  1.31it/s]
 16%|█▌        | 1160/7308 [15:02<1:14:53,  1.37it/s]
                                                     
{'loss': '4.008', 'grad_norm': '0.302', 'learning_rate': '0.0006', 'epoch': '32.23'}
+
 16%|█▌        | 1160/7308 [15:02<1:14:53,  1.37it/s]
 16%|█▌        | 1161/7308 [15:03<1:12:40,  1.41it/s]
 16%|█▌        | 1162/7308 [15:03<1:10:59,  1.44it/s]
 16%|█▌        | 1163/7308 [15:04<1:09:47,  1.47it/s]
 16%|█▌        | 1164/7308 [15:04<1:08:58,  1.48it/s]
 16%|█▌        | 1165/7308 [15:05<1:08:29,  1.49it/s]
 16%|█▌        | 1166/7308 [15:06<1:08:09,  1.50it/s]
 16%|█▌        | 1167/7308 [15:06<1:07:50,  1.51it/s]
 16%|█▌        | 1168/7308 [15:07<1:07:38,  1.51it/s]
 16%|█▌        | 1169/7308 [15:08<1:07:24,  1.52it/s]
 16%|█▌        | 1170/7308 [15:08<1:07:17,  1.52it/s]
                                                     
{'loss': '3.988', 'grad_norm': '0.2576', 'learning_rate': '0.0006', 'epoch': '32.51'}
+
 16%|█▌        | 1170/7308 [15:08<1:07:17,  1.52it/s]
 16%|█▌        | 1171/7308 [15:09<1:07:16,  1.52it/s]
 16%|█▌        | 1172/7308 [15:10<1:07:14,  1.52it/s]
 16%|█▌        | 1173/7308 [15:10<1:07:07,  1.52it/s]
 16%|█▌        | 1174/7308 [15:11<1:07:04,  1.52it/s]
 16%|█▌        | 1175/7308 [15:12<1:07:04,  1.52it/s]
 16%|█▌        | 1176/7308 [15:12<1:07:07,  1.52it/s]
 16%|█▌        | 1177/7308 [15:13<1:07:04,  1.52it/s]
 16%|█▌        | 1178/7308 [15:14<1:07:01,  1.52it/s]
 16%|█▌        | 1179/7308 [15:14<1:06:59,  1.52it/s]
 16%|█▌        | 1180/7308 [15:15<1:06:58,  1.52it/s]
                                                     
{'loss': '4.009', 'grad_norm': '0.236', 'learning_rate': '0.0006', 'epoch': '32.79'}
+
 16%|█▌        | 1180/7308 [15:15<1:06:58,  1.52it/s]
 16%|█▌        | 1181/7308 [15:16<1:07:01,  1.52it/s]
 16%|█▌        | 1182/7308 [15:17<1:18:07,  1.31it/s]
 16%|█▌        | 1183/7308 [15:17<1:14:44,  1.37it/s]
 16%|█▌        | 1184/7308 [15:18<1:12:20,  1.41it/s]
 16%|█▌        | 1185/7308 [15:19<1:10:39,  1.44it/s]
 16%|█▌        | 1186/7308 [15:19<1:09:35,  1.47it/s]
 16%|█▌        | 1187/7308 [15:20<1:08:43,  1.48it/s]
 16%|█▋        | 1188/7308 [15:20<55:47,  1.83it/s]  
 16%|█▋        | 1189/7308 [15:21<1:01:03,  1.67it/s]
 16%|█▋        | 1190/7308 [15:22<1:02:47,  1.62it/s]
                                                     
{'loss': '3.992', 'grad_norm': '0.3307', 'learning_rate': '0.0006', 'epoch': '33.06'}
+
 16%|█▋        | 1190/7308 [15:22<1:02:47,  1.62it/s]
 16%|█▋        | 1191/7308 [15:22<1:04:02,  1.59it/s]
 16%|█▋        | 1192/7308 [15:23<1:04:51,  1.57it/s]
 16%|█▋        | 1193/7308 [15:24<1:05:24,  1.56it/s]
 16%|█▋        | 1194/7308 [15:24<1:05:48,  1.55it/s]
 16%|█▋        | 1195/7308 [15:25<1:06:06,  1.54it/s]
 16%|█▋        | 1196/7308 [15:26<1:06:18,  1.54it/s]
 16%|█▋        | 1197/7308 [15:26<1:06:22,  1.53it/s]
 16%|█▋        | 1198/7308 [15:27<1:06:28,  1.53it/s]
 16%|█▋        | 1199/7308 [15:27<1:06:35,  1.53it/s]
 16%|█▋        | 1200/7308 [15:28<1:06:42,  1.53it/s]
                                                     
{'loss': '3.941', 'grad_norm': '0.3452', 'learning_rate': '0.0006', 'epoch': '33.34'}
+
 16%|█▋        | 1200/7308 [15:28<1:06:42,  1.53it/s]
 16%|█▋        | 1201/7308 [15:29<1:06:45,  1.52it/s]
 16%|█▋        | 1202/7308 [15:29<1:06:44,  1.52it/s]
 16%|█▋        | 1203/7308 [15:30<1:06:43,  1.52it/s]
 16%|█▋        | 1204/7308 [15:31<1:06:41,  1.53it/s]
 16%|█▋        | 1205/7308 [15:31<1:06:41,  1.53it/s]
 17%|█▋        | 1206/7308 [15:32<1:06:40,  1.53it/s]
 17%|█▋        | 1207/7308 [15:33<1:06:37,  1.53it/s]
 17%|█▋        | 1208/7308 [15:33<1:06:35,  1.53it/s]
 17%|█▋        | 1209/7308 [15:34<1:06:37,  1.53it/s]
 17%|█▋        | 1210/7308 [15:35<1:06:36,  1.53it/s]
                                                     
{'loss': '3.973', 'grad_norm': '0.3415', 'learning_rate': '0.0006', 'epoch': '33.62'}
+
 17%|█▋        | 1210/7308 [15:35<1:06:36,  1.53it/s]
 17%|█▋        | 1211/7308 [15:35<1:06:36,  1.53it/s]
 17%|█▋        | 1212/7308 [15:36<1:06:32,  1.53it/s]
 17%|█▋        | 1213/7308 [15:37<1:06:37,  1.52it/s]
 17%|█▋        | 1214/7308 [15:37<1:06:36,  1.52it/s]
 17%|█▋        | 1215/7308 [15:38<1:06:35,  1.53it/s]
 17%|█▋        | 1216/7308 [15:39<1:17:37,  1.31it/s]
 17%|█▋        | 1217/7308 [15:40<1:14:12,  1.37it/s]
 17%|█▋        | 1218/7308 [15:40<1:11:55,  1.41it/s]
 17%|█▋        | 1219/7308 [15:41<1:10:18,  1.44it/s]
 17%|█▋        | 1220/7308 [15:42<1:09:07,  1.47it/s]
                                                     
{'loss': '3.995', 'grad_norm': '0.3281', 'learning_rate': '0.0006', 'epoch': '33.9'}
+
 17%|█▋        | 1220/7308 [15:42<1:09:07,  1.47it/s]
 17%|█▋        | 1221/7308 [15:42<1:08:20,  1.48it/s]
 17%|█▋        | 1222/7308 [15:43<1:07:47,  1.50it/s]
 17%|█▋        | 1223/7308 [15:44<1:07:22,  1.51it/s]
 17%|█▋        | 1224/7308 [15:44<54:47,  1.85it/s]  [transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 19.05it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.93it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.67it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.08it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.76it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.57it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.48it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.40it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.34it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.30it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.30it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.28it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.27it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.26it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.25it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.24it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.24it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.24it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s]
                                                   
+
                                               
{'eval_loss': '4.151', 'eval_runtime': '3.262', 'eval_samples_per_second': '748.3', 'eval_steps_per_second': '11.96', 'epoch': '34'}
+
 17%|█▋        | 1224/7308 [15:47<54:47,  1.85it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-1224
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-1224/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-1224/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.87it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.87it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-1224/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-1224/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C1_rand15/checkpoint-1224/tokenizer_config.json
+[eval] epoch 34.0 step 1224 loss 4.1514 (best 4.1514 @ ep 34.0, 0/20 since best)
+
+
                                               
 17%|█▋        | 1225/7308 [15:50<3:41:26,  2.18s/it]
 17%|█▋        | 1226/7308 [15:50<2:54:53,  1.73s/it]
 17%|█▋        | 1227/7308 [15:51<2:22:22,  1.40s/it]
 17%|█▋        | 1228/7308 [15:52<1:59:35,  1.18s/it]
 17%|█▋        | 1229/7308 [15:52<1:43:39,  1.02s/it]
 17%|█▋        | 1230/7308 [15:53<1:43:16,  1.02s/it]
                                                     
{'loss': '3.933', 'grad_norm': '0.385', 'learning_rate': '0.0006', 'epoch': '34.17'}
+
 17%|█▋        | 1230/7308 [15:53<1:43:16,  1.02s/it]
 17%|█▋        | 1231/7308 [15:54<1:32:14,  1.10it/s]
 17%|█▋        | 1232/7308 [15:55<1:24:35,  1.20it/s]
 17%|█▋        | 1233/7308 [15:55<1:19:12,  1.28it/s]
 17%|█▋        | 1234/7308 [15:56<1:15:20,  1.34it/s]
 17%|█▋        | 1235/7308 [15:57<1:12:35,  1.39it/s]
 17%|█▋        | 1236/7308 [15:57<1:10:40,  1.43it/s]
 17%|█▋        | 1237/7308 [15:58<1:09:23,  1.46it/s]
 17%|█▋        | 1238/7308 [15:59<1:08:27,  1.48it/s]
 17%|█▋        | 1239/7308 [15:59<1:07:46,  1.49it/s]
 17%|█▋        | 1240/7308 [16:00<1:07:18,  1.50it/s]
                                                     
{'loss': '3.931', 'grad_norm': '0.3232', 'learning_rate': '0.0006', 'epoch': '34.45'}
+
 17%|█▋        | 1240/7308 [16:00<1:07:18,  1.50it/s]
 17%|█▋        | 1241/7308 [16:01<1:07:03,  1.51it/s]
 17%|█▋        | 1242/7308 [16:01<1:06:51,  1.51it/s]
 17%|█▋        | 1243/7308 [16:02<1:06:41,  1.52it/s]
 17%|█▋        | 1244/7308 [16:03<1:06:30,  1.52it/s]
 17%|█▋        | 1245/7308 [16:03<1:06:23,  1.52it/s]
 17%|█▋        | 1246/7308 [16:04<1:06:24,  1.52it/s]
 17%|█▋        | 1247/7308 [16:05<1:06:21,  1.52it/s]
 17%|█▋        | 1248/7308 [16:05<1:06:21,  1.52it/s]
 17%|█▋        | 1249/7308 [16:06<1:06:15,  1.52it/s]
 17%|█▋        | 1250/7308 [16:07<1:06:12,  1.53it/s]
                                                     
{'loss': '3.943', 'grad_norm': '0.3357', 'learning_rate': '0.0006', 'epoch': '34.73'}
+
 17%|█▋        | 1250/7308 [16:07<1:06:12,  1.53it/s]
 17%|█▋        | 1251/7308 [16:08<1:16:39,  1.32it/s]
 17%|█▋        | 1252/7308 [16:08<1:13:31,  1.37it/s]
 17%|█▋        | 1253/7308 [16:09<1:11:19,  1.42it/s]
 17%|█▋        | 1254/7308 [16:10<1:09:43,  1.45it/s]
 17%|█▋        | 1255/7308 [16:10<1:08:36,  1.47it/s]
 17%|█▋        | 1256/7308 [16:11<1:07:52,  1.49it/s]
 17%|█▋        | 1257/7308 [16:12<1:07:20,  1.50it/s]
 17%|█▋        | 1258/7308 [16:12<1:06:56,  1.51it/s]
 17%|█▋        | 1259/7308 [16:13<1:06:40,  1.51it/s]
 17%|█▋        | 1260/7308 [16:13<54:16,  1.86it/s]  
                                                   
{'loss': '3.943', 'grad_norm': '0.3397', 'learning_rate': '0.0006', 'epoch': '35'}
+
 17%|█▋        | 1260/7308 [16:13<54:16,  1.86it/s]
 17%|█▋        | 1261/7308 [16:14<1:00:24,  1.67it/s]
 17%|█▋        | 1262/7308 [16:14<1:02:07,  1.62it/s]
 17%|█▋        | 1263/7308 [16:15<1:03:15,  1.59it/s]
 17%|█▋        | 1264/7308 [16:16<1:04:05,  1.57it/s]
 17%|█▋        | 1265/7308 [16:16<1:04:40,  1.56it/s]
 17%|█▋        | 1266/7308 [16:17<1:05:04,  1.55it/s]
 17%|█▋        | 1267/7308 [16:18<1:05:23,  1.54it/s]
 17%|█▋        | 1268/7308 [16:18<1:05:36,  1.53it/s]
 17%|█▋        | 1269/7308 [16:19<1:05:40,  1.53it/s]
 17%|█▋        | 1270/7308 [16:20<1:15:54,  1.33it/s]
                                                     
{'loss': '3.877', 'grad_norm': '0.3528', 'learning_rate': '0.0006', 'epoch': '35.28'}
+
 17%|█▋        | 1270/7308 [16:20<1:15:54,  1.33it/s]
 17%|█▋        | 1271/7308 [16:21<1:12:55,  1.38it/s]
 17%|█▋        | 1272/7308 [16:22<1:21:37,  1.23it/s]
 17%|█▋        | 1273/7308 [16:22<1:16:52,  1.31it/s]
 17%|█▋        | 1274/7308 [16:23<1:13:35,  1.37it/s]
 17%|█▋        | 1275/7308 [16:24<1:21:34,  1.23it/s]
 17%|█▋        | 1276/7308 [16:25<1:16:49,  1.31it/s]
 17%|█▋        | 1277/7308 [16:25<1:13:28,  1.37it/s]
 17%|█▋        | 1278/7308 [16:26<1:11:12,  1.41it/s]
 18%|█▊        | 1279/7308 [16:27<1:09:34,  1.44it/s]
 18%|█▊        | 1280/7308 [16:27<1:08:23,  1.47it/s]
                                                     
{'loss': '3.894', 'grad_norm': '0.3096', 'learning_rate': '0.0006', 'epoch': '35.57'}
+
 18%|█▊        | 1280/7308 [16:27<1:08:23,  1.47it/s]
 18%|█▊        | 1281/7308 [16:28<1:07:36,  1.49it/s]
 18%|█▊        | 1282/7308 [16:29<1:07:04,  1.50it/s]
 18%|█▊        | 1283/7308 [16:29<1:06:40,  1.51it/s]
 18%|█▊        | 1284/7308 [16:30<1:06:21,  1.51it/s]
 18%|█▊        | 1285/7308 [16:31<1:06:10,  1.52it/s]
 18%|█▊        | 1286/7308 [16:31<1:06:05,  1.52it/s]
 18%|█▊        | 1287/7308 [16:32<1:05:57,  1.52it/s]
 18%|█▊        | 1288/7308 [16:33<1:05:51,  1.52it/s]
 18%|█▊        | 1289/7308 [16:33<1:05:47,  1.52it/s]
 18%|█▊        | 1290/7308 [16:34<1:05:46,  1.52it/s]
                                                     
{'loss': '3.915', 'grad_norm': '0.3125', 'learning_rate': '0.0006', 'epoch': '35.85'}
+
 18%|█▊        | 1290/7308 [16:34<1:05:46,  1.52it/s]
 18%|█▊        | 1291/7308 [16:35<1:05:48,  1.52it/s]
 18%|█▊        | 1292/7308 [16:35<1:05:42,  1.53it/s]
 18%|█▊        | 1293/7308 [16:36<1:05:40,  1.53it/s]
 18%|█▊        | 1294/7308 [16:36<1:05:43,  1.53it/s]
 18%|█▊        | 1295/7308 [16:37<1:05:41,  1.53it/s]
 18%|█▊        | 1296/7308 [16:37<53:32,  1.87it/s]  [transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 19.12it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.95it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.71it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.11it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.79it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.61it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.49it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.41it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.31it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.32it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.31it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.27it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.27it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.26it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.26it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.25it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.26it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.25it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.25it/s]
                                                   
+
                                               
{'eval_loss': '4.155', 'eval_runtime': '3.232', 'eval_samples_per_second': '755.2', 'eval_steps_per_second': '12.06', 'epoch': '36'}
+
 18%|█▊        | 1296/7308 [16:41<53:32,  1.87it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.25it/s][eval] epoch 36.0 step 1296 loss 4.1550 (best 4.1514 @ ep 34.0, 1/20 since best)
+
+
                                               
 18%|█▊        | 1297/7308 [16:41<2:36:47,  1.57s/it]
 18%|█▊        | 1298/7308 [16:42<2:09:24,  1.29s/it]
 18%|█▊        | 1299/7308 [16:43<1:50:14,  1.10s/it]
 18%|█▊        | 1300/7308 [16:43<1:36:51,  1.03it/s]
                                                     
{'loss': '3.889', 'grad_norm': '0.3985', 'learning_rate': '0.0006', 'epoch': '36.11'}
+
 18%|█▊        | 1300/7308 [16:43<1:36:51,  1.03it/s]
 18%|█▊        | 1301/7308 [16:44<1:27:31,  1.14it/s]
 18%|█▊        | 1302/7308 [16:45<1:20:59,  1.24it/s]
 18%|█▊        | 1303/7308 [16:45<1:16:22,  1.31it/s]
 18%|█▊        | 1304/7308 [16:46<1:13:07,  1.37it/s]
 18%|█▊        | 1305/7308 [16:47<1:10:48,  1.41it/s]
 18%|█▊        | 1306/7308 [16:47<1:09:11,  1.45it/s]
 18%|█▊        | 1307/7308 [16:48<1:08:02,  1.47it/s]
 18%|█▊        | 1308/7308 [16:49<1:07:17,  1.49it/s]
 18%|█▊        | 1309/7308 [16:49<1:06:46,  1.50it/s]
 18%|█▊        | 1310/7308 [16:50<1:06:21,  1.51it/s]
                                                     
{'loss': '3.859', 'grad_norm': '0.4323', 'learning_rate': '0.0006', 'epoch': '36.4'}
+
 18%|█▊        | 1310/7308 [16:50<1:06:21,  1.51it/s]
 18%|█▊        | 1311/7308 [16:51<1:06:04,  1.51it/s]
 18%|█▊        | 1312/7308 [16:51<1:05:53,  1.52it/s]
 18%|█▊        | 1313/7308 [16:52<1:05:48,  1.52it/s]
 18%|█▊        | 1314/7308 [16:53<1:05:40,  1.52it/s]
 18%|█▊        | 1315/7308 [16:53<1:05:33,  1.52it/s]
 18%|█▊        | 1316/7308 [16:54<1:05:31,  1.52it/s]
 18%|█▊        | 1317/7308 [16:54<1:05:32,  1.52it/s]
 18%|█▊        | 1318/7308 [16:55<1:05:29,  1.52it/s]
 18%|█▊        | 1319/7308 [16:56<1:05:25,  1.53it/s]
 18%|█▊        | 1320/7308 [16:56<1:05:23,  1.53it/s]
                                                     
{'loss': '3.891', 'grad_norm': '0.3711', 'learning_rate': '0.0006', 'epoch': '36.68'}
+
 18%|█▊        | 1320/7308 [16:56<1:05:23,  1.53it/s]
 18%|█▊        | 1321/7308 [16:57<1:05:21,  1.53it/s]
 18%|█▊        | 1322/7308 [16:58<1:05:20,  1.53it/s]
 18%|█▊        | 1323/7308 [16:58<1:05:17,  1.53it/s]
 18%|█▊        | 1324/7308 [16:59<1:05:20,  1.53it/s]
 18%|█▊        | 1325/7308 [17:00<1:05:21,  1.53it/s]
 18%|█▊        | 1326/7308 [17:00<1:05:20,  1.53it/s]
 18%|█▊        | 1327/7308 [17:01<1:05:18,  1.53it/s]
 18%|█▊        | 1328/7308 [17:02<1:05:17,  1.53it/s]
 18%|█▊        | 1329/7308 [17:02<1:05:15,  1.53it/s]
 18%|█▊        | 1330/7308 [17:03<1:05:16,  1.53it/s]
                                                     
{'loss': '3.898', 'grad_norm': '0.3177', 'learning_rate': '0.0006', 'epoch': '36.96'}
+
 18%|█▊        | 1330/7308 [17:03<1:05:16,  1.53it/s]
 18%|█▊        | 1331/7308 [17:04<1:05:14,  1.53it/s]
 18%|█▊        | 1332/7308 [17:04<53:11,  1.87it/s]  
 18%|█▊        | 1333/7308 [17:05<59:14,  1.68it/s]
 18%|█▊        | 1334/7308 [17:05<1:01:02,  1.63it/s]
 18%|█▊        | 1335/7308 [17:06<1:02:17,  1.60it/s]
 18%|█▊        | 1336/7308 [17:07<1:03:10,  1.58it/s]
 18%|█▊        | 1337/7308 [17:07<1:03:46,  1.56it/s]
 18%|█▊        | 1338/7308 [17:08<1:04:13,  1.55it/s]
 18%|█▊        | 1339/7308 [17:09<1:04:31,  1.54it/s]
 18%|█▊        | 1340/7308 [17:09<1:04:40,  1.54it/s]
                                                     
{'loss': '3.833', 'grad_norm': '0.338', 'learning_rate': '0.0006', 'epoch': '37.23'}
+
 18%|█▊        | 1340/7308 [17:09<1:04:40,  1.54it/s]
 18%|█▊        | 1341/7308 [17:10<1:05:20,  1.52it/s]
 18%|█▊        | 1342/7308 [17:11<1:05:19,  1.52it/s]
 18%|█▊        | 1343/7308 [17:11<1:05:17,  1.52it/s]
 18%|█▊        | 1344/7308 [17:12<1:05:11,  1.52it/s]
 18%|█▊        | 1345/7308 [17:13<1:05:08,  1.53it/s]
 18%|█▊        | 1346/7308 [17:13<1:05:09,  1.52it/s]
 18%|█▊        | 1347/7308 [17:14<1:05:08,  1.53it/s]
 18%|█▊        | 1348/7308 [17:14<1:05:05,  1.53it/s]
 18%|█▊        | 1349/7308 [17:15<1:05:03,  1.53it/s]
 18%|█▊        | 1350/7308 [17:16<1:05:05,  1.53it/s]
                                                     
{'loss': '3.833', 'grad_norm': '0.3717', 'learning_rate': '0.0006', 'epoch': '37.51'}
+
 18%|█▊        | 1350/7308 [17:16<1:05:05,  1.53it/s]
 18%|█▊        | 1351/7308 [17:16<1:05:06,  1.52it/s]
 19%|█▊        | 1352/7308 [17:17<1:05:04,  1.53it/s]
 19%|█▊        | 1353/7308 [17:18<1:05:03,  1.53it/s]
 19%|█▊        | 1354/7308 [17:18<1:05:00,  1.53it/s]
 19%|█▊        | 1355/7308 [17:19<1:05:02,  1.53it/s]
 19%|█▊        | 1356/7308 [17:20<1:05:00,  1.53it/s]
 19%|█▊        | 1357/7308 [17:20<1:04:56,  1.53it/s]
 19%|█▊        | 1358/7308 [17:21<1:04:56,  1.53it/s]
 19%|█▊        | 1359/7308 [17:22<1:05:00,  1.53it/s]
 19%|█▊        | 1360/7308 [17:22<1:04:59,  1.53it/s]
                                                     
{'loss': '3.857', 'grad_norm': '0.3025', 'learning_rate': '0.0006', 'epoch': '37.79'}
+
 19%|█▊        | 1360/7308 [17:22<1:04:59,  1.53it/s]
 19%|█▊        | 1361/7308 [17:23<1:05:00,  1.52it/s]
 19%|█▊        | 1362/7308 [17:24<1:04:58,  1.53it/s]
 19%|█▊        | 1363/7308 [17:24<1:04:59,  1.52it/s]
 19%|█▊        | 1364/7308 [17:25<1:04:59,  1.52it/s]
 19%|█▊        | 1365/7308 [17:26<1:04:57,  1.52it/s]
 19%|█▊        | 1366/7308 [17:26<1:04:55,  1.53it/s]
 19%|█▊        | 1367/7308 [17:27<1:04:51,  1.53it/s]
 19%|█▊        | 1368/7308 [17:27<52:50,  1.87it/s]  [transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.99it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.92it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.68it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.10it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.77it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.58it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.46it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.40it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.34it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.31it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.30it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.28it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.27it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.26it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.25it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.24it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.24it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.24it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s]
                                                   
+
                                               
{'eval_loss': '4.172', 'eval_runtime': '3.261', 'eval_samples_per_second': '748.5', 'eval_steps_per_second': '11.96', 'epoch': '38'}
+
 19%|█▊        | 1368/7308 [17:30<52:50,  1.87it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s][eval] epoch 38.0 step 1368 loss 4.1715 (best 4.1514 @ ep 34.0, 2/20 since best)
+
+
                                               
 19%|█▊        | 1369/7308 [17:31<2:35:41,  1.57s/it]
 19%|█▊        | 1370/7308 [17:32<2:08:24,  1.30s/it]
                                                     
{'loss': '3.848', 'grad_norm': '0.4403', 'learning_rate': '0.0006', 'epoch': '38.06'}
+
 19%|█▊        | 1370/7308 [17:32<2:08:24,  1.30s/it]
 19%|█▉        | 1371/7308 [17:32<1:49:20,  1.11s/it]
 19%|█▉        | 1372/7308 [17:33<1:36:01,  1.03it/s]
 19%|█▉        | 1373/7308 [17:34<1:26:40,  1.14it/s]
 19%|█▉        | 1374/7308 [17:34<1:20:06,  1.23it/s]
 19%|█▉        | 1375/7308 [17:35<1:15:28,  1.31it/s]
 19%|█▉        | 1376/7308 [17:36<1:12:17,  1.37it/s]
 19%|█▉        | 1377/7308 [17:36<1:10:01,  1.41it/s]
 19%|█▉        | 1378/7308 [17:37<1:08:28,  1.44it/s]
 19%|█▉        | 1379/7308 [17:38<1:07:19,  1.47it/s]
 19%|█▉        | 1380/7308 [17:38<1:06:31,  1.49it/s]
                                                     
{'loss': '3.812', 'grad_norm': '0.4382', 'learning_rate': '0.0006', 'epoch': '38.34'}
+
 19%|█▉        | 1380/7308 [17:38<1:06:31,  1.49it/s]
 19%|█▉        | 1381/7308 [17:39<1:06:02,  1.50it/s]
 19%|█▉        | 1382/7308 [17:40<1:05:39,  1.50it/s]
 19%|█▉        | 1383/7308 [17:40<1:05:19,  1.51it/s]
 19%|█▉        | 1384/7308 [17:41<1:05:08,  1.52it/s]
 19%|█▉        | 1385/7308 [17:42<1:05:00,  1.52it/s]
 19%|█▉        | 1386/7308 [17:42<1:04:54,  1.52it/s]
 19%|█▉        | 1387/7308 [17:43<1:04:49,  1.52it/s]
 19%|█▉        | 1388/7308 [17:44<1:04:45,  1.52it/s]
 19%|█▉        | 1389/7308 [17:44<1:04:44,  1.52it/s]
 19%|█▉        | 1390/7308 [17:45<1:04:44,  1.52it/s]
                                                     
{'loss': '3.821', 'grad_norm': '0.377', 'learning_rate': '0.0006', 'epoch': '38.62'}
+
 19%|█▉        | 1390/7308 [17:45<1:04:44,  1.52it/s]
 19%|█▉        | 1391/7308 [17:46<1:04:44,  1.52it/s]
 19%|█▉        | 1392/7308 [17:46<1:04:39,  1.53it/s]
 19%|█▉        | 1393/7308 [17:47<1:04:35,  1.53it/s]
 19%|█▉        | 1394/7308 [17:48<1:04:36,  1.53it/s]
 19%|█▉        | 1395/7308 [17:48<1:04:38,  1.52it/s]
 19%|█▉        | 1396/7308 [17:49<1:04:36,  1.53it/s]
 19%|█▉        | 1397/7308 [17:50<1:04:33,  1.53it/s]
 19%|█▉        | 1398/7308 [17:50<1:04:33,  1.53it/s]
 19%|█▉        | 1399/7308 [17:51<1:04:35,  1.52it/s]
 19%|█▉        | 1400/7308 [17:52<1:04:33,  1.53it/s]
                                                     
{'loss': '3.832', 'grad_norm': '0.2985', 'learning_rate': '0.0006', 'epoch': '38.9'}
+
 19%|█▉        | 1400/7308 [17:52<1:04:33,  1.53it/s]
 19%|█▉        | 1401/7308 [17:52<1:04:31,  1.53it/s]
 19%|█▉        | 1402/7308 [17:53<1:04:28,  1.53it/s]
 19%|█▉        | 1403/7308 [17:53<1:04:28,  1.53it/s]
 19%|█▉        | 1404/7308 [17:54<52:31,  1.87it/s]  
 19%|█▉        | 1405/7308 [17:54<58:29,  1.68it/s]
 19%|█▉        | 1406/7308 [17:55<1:00:18,  1.63it/s]
 19%|█▉        | 1407/7308 [17:56<1:01:34,  1.60it/s]
 19%|█▉        | 1408/7308 [17:56<1:02:26,  1.57it/s]
 19%|█▉        | 1409/7308 [17:57<1:02:59,  1.56it/s]
 19%|█▉        | 1410/7308 [17:58<1:03:24,  1.55it/s]
                                                     
{'loss': '3.784', 'grad_norm': '0.3613', 'learning_rate': '0.0006', 'epoch': '39.17'}
+
 19%|█▉        | 1410/7308 [17:58<1:03:24,  1.55it/s]
 19%|█▉        | 1411/7308 [17:58<1:03:44,  1.54it/s]
 19%|█▉        | 1412/7308 [17:59<1:03:56,  1.54it/s]
 19%|█▉        | 1413/7308 [18:00<1:04:03,  1.53it/s]
 19%|█▉        | 1414/7308 [18:00<1:04:06,  1.53it/s]
 19%|█▉        | 1415/7308 [18:01<1:04:12,  1.53it/s]
 19%|█▉        | 1416/7308 [18:02<1:04:17,  1.53it/s]
 19%|█▉        | 1417/7308 [18:02<1:04:19,  1.53it/s]
 19%|█▉        | 1418/7308 [18:03<1:04:16,  1.53it/s]
 19%|█▉        | 1419/7308 [18:04<1:04:15,  1.53it/s]
 19%|█▉        | 1420/7308 [18:04<1:04:21,  1.52it/s]
                                                     
{'loss': '3.775', 'grad_norm': '0.4271', 'learning_rate': '0.0006', 'epoch': '39.45'}
+
 19%|█▉        | 1420/7308 [18:04<1:04:21,  1.52it/s]
 19%|█▉        | 1421/7308 [18:05<1:04:24,  1.52it/s]
 19%|█▉        | 1422/7308 [18:06<1:04:21,  1.52it/s]
 19%|█▉        | 1423/7308 [18:06<1:04:17,  1.53it/s]
 19%|█▉        | 1424/7308 [18:07<1:04:15,  1.53it/s]
 19%|█▉        | 1425/7308 [18:08<1:04:15,  1.53it/s]
 20%|█▉        | 1426/7308 [18:08<1:04:15,  1.53it/s]
 20%|█▉        | 1427/7308 [18:09<1:04:12,  1.53it/s]
 20%|█▉        | 1428/7308 [18:10<1:04:11,  1.53it/s]
 20%|█▉        | 1429/7308 [18:10<1:04:13,  1.53it/s]
 20%|█▉        | 1430/7308 [18:11<1:04:14,  1.53it/s]
                                                     
{'loss': '3.803', 'grad_norm': '0.389', 'learning_rate': '0.0006', 'epoch': '39.73'}
+
 20%|█▉        | 1430/7308 [18:11<1:04:14,  1.53it/s]
 20%|█▉        | 1431/7308 [18:11<1:04:14,  1.52it/s]
 20%|█▉        | 1432/7308 [18:12<1:04:11,  1.53it/s]
 20%|█▉        | 1433/7308 [18:13<1:04:12,  1.53it/s]
 20%|█▉        | 1434/7308 [18:13<1:04:12,  1.52it/s]
 20%|█▉        | 1435/7308 [18:14<1:04:11,  1.52it/s]
 20%|█▉        | 1436/7308 [18:15<1:04:08,  1.53it/s]
 20%|█▉        | 1437/7308 [18:15<1:04:05,  1.53it/s]
 20%|█▉        | 1438/7308 [18:16<1:04:09,  1.52it/s]
 20%|█▉        | 1439/7308 [18:17<1:04:06,  1.53it/s]
 20%|█▉        | 1440/7308 [18:17<52:13,  1.87it/s]  
                                                   
{'loss': '3.812', 'grad_norm': '0.3911', 'learning_rate': '0.0006', 'epoch': '40'}
+
 20%|█▉        | 1440/7308 [18:17<52:13,  1.87it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.40it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.74it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.58it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.05it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.74it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.58it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.46it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.35it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.33it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.29it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.28it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.26it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.26it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.24it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.25it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.24it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.24it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.24it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.25it/s]
                                                   
+
                                               
{'eval_loss': '4.177', 'eval_runtime': '3.251', 'eval_samples_per_second': '750.9', 'eval_steps_per_second': '12', 'epoch': '40'}
+
 20%|█▉        | 1440/7308 [18:20<52:13,  1.87it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.25it/s][eval] epoch 40.0 step 1440 loss 4.1767 (best 4.1514 @ ep 34.0, 3/20 since best)
+
+
                                               
 20%|█▉        | 1441/7308 [18:21<2:33:37,  1.57s/it]
 20%|█▉        | 1442/7308 [18:22<2:06:44,  1.30s/it]
 20%|█▉        | 1443/7308 [18:22<1:47:53,  1.10s/it]
 20%|█▉        | 1444/7308 [18:23<1:34:44,  1.03it/s]
 20%|█▉        | 1445/7308 [18:24<1:25:32,  1.14it/s]
 20%|█▉        | 1446/7308 [18:24<1:19:03,  1.24it/s]
 20%|█▉        | 1447/7308 [18:25<1:14:30,  1.31it/s]
 20%|█▉        | 1448/7308 [18:26<1:11:19,  1.37it/s]
 20%|█▉        | 1449/7308 [18:26<1:09:08,  1.41it/s]
 20%|█▉        | 1450/7308 [18:27<1:07:35,  1.44it/s]
                                                     
{'loss': '3.736', 'grad_norm': '0.467', 'learning_rate': '0.0006', 'epoch': '40.28'}
+
 20%|█▉        | 1450/7308 [18:27<1:07:35,  1.44it/s]
 20%|█▉        | 1451/7308 [18:28<1:06:29,  1.47it/s]
 20%|█▉        | 1452/7308 [18:28<1:05:43,  1.49it/s]
 20%|█▉        | 1453/7308 [18:29<1:05:14,  1.50it/s]
 20%|█▉        | 1454/7308 [18:30<1:04:52,  1.50it/s]
 20%|█▉        | 1455/7308 [18:30<1:04:35,  1.51it/s]
 20%|█▉        | 1456/7308 [18:31<1:04:20,  1.52it/s]
 20%|█▉        | 1457/7308 [18:31<1:04:11,  1.52it/s]
 20%|█▉        | 1458/7308 [18:32<1:04:09,  1.52it/s]
 20%|█▉        | 1459/7308 [18:33<1:04:05,  1.52it/s]
 20%|█▉        | 1460/7308 [18:33<1:04:01,  1.52it/s]
                                                     
{'loss': '3.763', 'grad_norm': '0.3896', 'learning_rate': '0.0006', 'epoch': '40.57'}
+
 20%|█▉        | 1460/7308 [18:33<1:04:01,  1.52it/s]
 20%|█▉        | 1461/7308 [18:34<1:03:58,  1.52it/s]
 20%|██        | 1462/7308 [18:35<1:03:55,  1.52it/s]
 20%|██        | 1463/7308 [18:35<1:03:55,  1.52it/s]
 20%|██        | 1464/7308 [18:36<1:03:57,  1.52it/s]
 20%|██        | 1465/7308 [18:37<1:03:55,  1.52it/s]
 20%|██        | 1466/7308 [18:37<1:03:50,  1.53it/s]
 20%|██        | 1467/7308 [18:38<1:03:49,  1.53it/s]
 20%|██        | 1468/7308 [18:39<1:03:50,  1.52it/s]
 20%|██        | 1469/7308 [18:39<1:03:51,  1.52it/s]
 20%|██        | 1470/7308 [18:40<1:03:48,  1.52it/s]
                                                     
{'loss': '3.779', 'grad_norm': '0.3593', 'learning_rate': '0.0006', 'epoch': '40.85'}
+
 20%|██        | 1470/7308 [18:40<1:03:48,  1.52it/s]
 20%|██        | 1471/7308 [18:41<1:03:49,  1.52it/s]
 20%|██        | 1472/7308 [18:41<1:03:47,  1.52it/s]
 20%|██        | 1473/7308 [18:42<1:03:49,  1.52it/s]
 20%|██        | 1474/7308 [18:43<1:03:48,  1.52it/s]
 20%|██        | 1475/7308 [18:43<1:03:44,  1.53it/s]
 20%|██        | 1476/7308 [18:44<51:56,  1.87it/s]  
 20%|██        | 1477/7308 [18:44<57:30,  1.69it/s]
 20%|██        | 1478/7308 [18:45<59:25,  1.64it/s]
 20%|██        | 1479/7308 [18:46<1:00:42,  1.60it/s]
 20%|██        | 1480/7308 [18:46<1:01:35,  1.58it/s]
                                                     
{'loss': '3.759', 'grad_norm': '0.3327', 'learning_rate': '0.0006', 'epoch': '41.11'}
+
 20%|██        | 1480/7308 [18:46<1:01:35,  1.58it/s]
 20%|██        | 1481/7308 [18:47<1:02:14,  1.56it/s]
 20%|██        | 1482/7308 [18:48<1:02:37,  1.55it/s]
 20%|██        | 1483/7308 [18:48<1:02:54,  1.54it/s]
 20%|██        | 1484/7308 [18:49<1:03:10,  1.54it/s]
 20%|██        | 1485/7308 [18:50<1:03:22,  1.53it/s]
 20%|██        | 1486/7308 [18:50<1:03:26,  1.53it/s]
 20%|██        | 1487/7308 [18:51<1:03:28,  1.53it/s]
 20%|██        | 1488/7308 [18:51<1:03:27,  1.53it/s]
 20%|██        | 1489/7308 [18:52<1:03:32,  1.53it/s]
 20%|██        | 1490/7308 [18:53<1:03:34,  1.53it/s]
                                                     
{'loss': '3.716', 'grad_norm': '0.4036', 'learning_rate': '0.0006', 'epoch': '41.4'}
+
 20%|██        | 1490/7308 [18:53<1:03:34,  1.53it/s]
 20%|██        | 1491/7308 [18:53<1:03:35,  1.52it/s]
 20%|██        | 1492/7308 [18:54<1:03:31,  1.53it/s]
 20%|██        | 1493/7308 [18:55<1:03:30,  1.53it/s]
 20%|██        | 1494/7308 [18:55<1:03:32,  1.53it/s]
 20%|██        | 1495/7308 [18:56<1:03:35,  1.52it/s]
 20%|██        | 1496/7308 [18:57<1:03:38,  1.52it/s]
 20%|██        | 1497/7308 [18:57<1:03:34,  1.52it/s]
 20%|██        | 1498/7308 [18:58<1:03:31,  1.52it/s]
 21%|██        | 1499/7308 [18:59<1:03:30,  1.52it/s]
 21%|██        | 1500/7308 [18:59<1:03:30,  1.52it/s]
                                                     
{'loss': '3.738', 'grad_norm': '0.3366', 'learning_rate': '0.0006', 'epoch': '41.68'}
+
 21%|██        | 1500/7308 [18:59<1:03:30,  1.52it/s]
 21%|██        | 1501/7308 [19:00<1:03:31,  1.52it/s]
 21%|██        | 1502/7308 [19:01<1:03:26,  1.53it/s]
 21%|██        | 1503/7308 [19:01<1:03:22,  1.53it/s]
 21%|██        | 1504/7308 [19:02<1:03:25,  1.53it/s]
 21%|██        | 1505/7308 [19:03<1:03:26,  1.52it/s]
 21%|██        | 1506/7308 [19:03<1:03:25,  1.52it/s]
 21%|██        | 1507/7308 [19:04<1:03:22,  1.53it/s]
 21%|██        | 1508/7308 [19:05<1:03:22,  1.53it/s]
 21%|██        | 1509/7308 [19:05<1:03:24,  1.52it/s]
 21%|██        | 1510/7308 [19:06<1:03:23,  1.52it/s]
                                                     
{'loss': '3.764', 'grad_norm': '0.3543', 'learning_rate': '0.0006', 'epoch': '41.96'}
+
 21%|██        | 1510/7308 [19:06<1:03:23,  1.52it/s]
 21%|██        | 1511/7308 [19:07<1:03:22,  1.52it/s]
 21%|██        | 1512/7308 [19:07<51:37,  1.87it/s]  [transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.31it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.71it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.56it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.02it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.74it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.54it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.45it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.38it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.33it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.30it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.28it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.26it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.25it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.25it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.23it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.25it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.20it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.22it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.22it/s]
                                                   
+
                                               
{'eval_loss': '4.195', 'eval_runtime': '3.25', 'eval_samples_per_second': '751.2', 'eval_steps_per_second': '12', 'epoch': '42'}
+
 21%|██        | 1512/7308 [19:10<51:37,  1.87it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.22it/s][eval] epoch 42.0 step 1512 loss 4.1945 (best 4.1514 @ ep 34.0, 4/20 since best)
+
+
                                               
 21%|██        | 1513/7308 [19:11<2:31:47,  1.57s/it]
 21%|██        | 1514/7308 [19:11<2:05:10,  1.30s/it]
 21%|██        | 1515/7308 [19:12<1:46:34,  1.10s/it]
 21%|██        | 1516/7308 [19:13<1:33:38,  1.03it/s]
 21%|██        | 1517/7308 [19:13<1:24:39,  1.14it/s]
 21%|██        | 1518/7308 [19:14<1:18:16,  1.23it/s]
 21%|██        | 1519/7308 [19:15<1:13:46,  1.31it/s]
 21%|██        | 1520/7308 [19:15<1:10:33,  1.37it/s]
                                                     
{'loss': '3.7', 'grad_norm': '0.3973', 'learning_rate': '0.0006', 'epoch': '42.23'}
+
 21%|██        | 1520/7308 [19:15<1:10:33,  1.37it/s]
 21%|██        | 1521/7308 [19:16<1:08:19,  1.41it/s]
 21%|██        | 1522/7308 [19:17<1:06:48,  1.44it/s]
 21%|██        | 1523/7308 [19:17<1:05:43,  1.47it/s]
 21%|██        | 1524/7308 [19:18<1:04:55,  1.48it/s]
 21%|██        | 1525/7308 [19:19<1:04:21,  1.50it/s]
 21%|██        | 1526/7308 [19:19<1:04:02,  1.50it/s]
 21%|██        | 1527/7308 [19:20<1:03:46,  1.51it/s]
 21%|██        | 1528/7308 [19:21<1:03:35,  1.51it/s]
 21%|██        | 1529/7308 [19:21<1:03:25,  1.52it/s]
 21%|██        | 1530/7308 [19:22<1:03:18,  1.52it/s]
                                                     
{'loss': '3.709', 'grad_norm': '0.3535', 'learning_rate': '0.0006', 'epoch': '42.51'}
+
 21%|██        | 1530/7308 [19:22<1:03:18,  1.52it/s]
 21%|██        | 1531/7308 [19:23<1:03:16,  1.52it/s]
 21%|██        | 1532/7308 [19:23<1:03:14,  1.52it/s]
 21%|██        | 1533/7308 [19:24<1:03:11,  1.52it/s]
 21%|██        | 1534/7308 [19:25<1:03:06,  1.52it/s]
 21%|██        | 1535/7308 [19:25<1:03:07,  1.52it/s]
 21%|██        | 1536/7308 [19:26<1:03:06,  1.52it/s]
 21%|██        | 1537/7308 [19:27<1:03:04,  1.52it/s]
 21%|██        | 1538/7308 [19:27<1:03:01,  1.53it/s]
 21%|██        | 1539/7308 [19:28<1:03:00,  1.53it/s]
 21%|██        | 1540/7308 [19:29<1:03:02,  1.52it/s]
                                                     
{'loss': '3.732', 'grad_norm': '0.3926', 'learning_rate': '0.0006', 'epoch': '42.79'}
+
 21%|██        | 1540/7308 [19:29<1:03:02,  1.52it/s]
 21%|██        | 1541/7308 [19:29<1:03:04,  1.52it/s]
 21%|██        | 1542/7308 [19:30<1:03:01,  1.52it/s]
 21%|██        | 1543/7308 [19:30<1:03:00,  1.52it/s]
 21%|██        | 1544/7308 [19:31<1:02:59,  1.53it/s]
 21%|██        | 1545/7308 [19:32<1:02:59,  1.52it/s]
 21%|██        | 1546/7308 [19:32<1:02:59,  1.52it/s]
 21%|██        | 1547/7308 [19:33<1:02:55,  1.53it/s]
 21%|██        | 1548/7308 [19:33<51:17,  1.87it/s]  
 21%|██        | 1549/7308 [19:34<57:39,  1.66it/s]
 21%|██        | 1550/7308 [19:35<59:13,  1.62it/s]
                                                   
{'loss': '3.727', 'grad_norm': '0.3447', 'learning_rate': '0.0006', 'epoch': '43.06'}
+
 21%|██        | 1550/7308 [19:35<59:13,  1.62it/s]
 21%|██        | 1551/7308 [19:35<1:00:18,  1.59it/s]
 21%|██        | 1552/7308 [19:36<1:01:06,  1.57it/s]
 21%|██▏       | 1553/7308 [19:37<1:01:38,  1.56it/s]
 21%|██▏       | 1554/7308 [19:37<1:01:59,  1.55it/s]
 21%|██▏       | 1555/7308 [19:38<1:02:14,  1.54it/s]
 21%|██▏       | 1556/7308 [19:39<1:02:22,  1.54it/s]
 21%|██▏       | 1557/7308 [19:39<1:02:32,  1.53it/s]
 21%|██▏       | 1558/7308 [19:40<1:02:36,  1.53it/s]
 21%|██▏       | 1559/7308 [19:41<1:02:39,  1.53it/s]
 21%|██▏       | 1560/7308 [19:41<1:02:40,  1.53it/s]
                                                     
{'loss': '3.652', 'grad_norm': '0.4371', 'learning_rate': '0.0006', 'epoch': '43.34'}
+
 21%|██▏       | 1560/7308 [19:41<1:02:40,  1.53it/s]
 21%|██▏       | 1561/7308 [19:42<1:02:43,  1.53it/s]
 21%|██▏       | 1562/7308 [19:43<1:02:45,  1.53it/s]
 21%|██▏       | 1563/7308 [19:43<1:02:45,  1.53it/s]
 21%|██▏       | 1564/7308 [19:44<1:02:42,  1.53it/s]
 21%|██▏       | 1565/7308 [19:45<1:02:41,  1.53it/s]
 21%|██▏       | 1566/7308 [19:45<1:02:45,  1.53it/s]
 21%|██▏       | 1567/7308 [19:46<1:02:44,  1.53it/s]
 21%|██▏       | 1568/7308 [19:47<1:02:43,  1.53it/s]
 21%|██▏       | 1569/7308 [19:47<1:02:41,  1.53it/s]
 21%|██▏       | 1570/7308 [19:48<1:02:40,  1.53it/s]
                                                     
{'loss': '3.705', 'grad_norm': '0.3961', 'learning_rate': '0.0006', 'epoch': '43.62'}
+
 21%|██▏       | 1570/7308 [19:48<1:02:40,  1.53it/s]
 21%|██▏       | 1571/7308 [19:49<1:02:42,  1.52it/s]
 22%|██▏       | 1572/7308 [19:49<1:02:40,  1.53it/s]
 22%|██▏       | 1573/7308 [19:50<1:02:37,  1.53it/s]
 22%|██▏       | 1574/7308 [19:50<1:02:35,  1.53it/s]
 22%|██▏       | 1575/7308 [19:51<1:02:38,  1.53it/s]
 22%|██▏       | 1576/7308 [19:52<1:02:39,  1.52it/s]
 22%|██▏       | 1577/7308 [19:52<1:02:37,  1.53it/s]
 22%|██▏       | 1578/7308 [19:53<1:02:35,  1.53it/s]
 22%|██▏       | 1579/7308 [19:54<1:02:35,  1.53it/s]
 22%|██▏       | 1580/7308 [19:54<1:02:45,  1.52it/s]
                                                     
{'loss': '3.724', 'grad_norm': '0.3278', 'learning_rate': '0.0006', 'epoch': '43.9'}
+
 22%|██▏       | 1580/7308 [19:54<1:02:45,  1.52it/s]
 22%|██▏       | 1581/7308 [19:55<1:02:44,  1.52it/s]
 22%|██▏       | 1582/7308 [19:56<1:02:43,  1.52it/s]
 22%|██▏       | 1583/7308 [19:56<1:02:37,  1.52it/s]
 22%|██▏       | 1584/7308 [19:57<51:03,  1.87it/s]  [transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.87it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.90it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.67it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.08it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.76it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.59it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.46it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.39it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.33it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.29it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.28it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.26it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.25it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.26it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.26it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.25it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.25it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.25it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.25it/s]
                                                   
+
                                               
{'eval_loss': '4.214', 'eval_runtime': '3.263', 'eval_samples_per_second': '748.2', 'eval_steps_per_second': '11.95', 'epoch': '44'}
+
 22%|██▏       | 1584/7308 [20:00<51:03,  1.87it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.25it/s][eval] epoch 44.0 step 1584 loss 4.2142 (best 4.1514 @ ep 34.0, 5/20 since best)
+
+
                                               
 22%|██▏       | 1585/7308 [20:01<2:30:39,  1.58s/it]
 22%|██▏       | 1586/7308 [20:01<2:04:10,  1.30s/it]
 22%|██▏       | 1587/7308 [20:02<1:45:39,  1.11s/it]
 22%|██▏       | 1588/7308 [20:03<1:32:41,  1.03it/s]
 22%|██▏       | 1589/7308 [20:03<1:23:36,  1.14it/s]
 22%|██▏       | 1590/7308 [20:04<1:17:17,  1.23it/s]
                                                     
{'loss': '3.648', 'grad_norm': '0.3793', 'learning_rate': '0.0006', 'epoch': '44.17'}
+
 22%|██▏       | 1590/7308 [20:04<1:17:17,  1.23it/s]
 22%|██▏       | 1591/7308 [20:05<1:12:54,  1.31it/s]
 22%|██▏       | 1592/7308 [20:05<1:09:47,  1.37it/s]
 22%|██▏       | 1593/7308 [20:06<1:07:34,  1.41it/s]
 22%|██▏       | 1594/7308 [20:07<1:05:59,  1.44it/s]
 22%|██▏       | 1595/7308 [20:07<1:04:56,  1.47it/s]
 22%|██▏       | 1596/7308 [20:08<1:04:10,  1.48it/s]
 22%|██▏       | 1597/7308 [20:09<1:03:36,  1.50it/s]
 22%|██▏       | 1598/7308 [20:09<1:03:13,  1.51it/s]
 22%|██▏       | 1599/7308 [20:10<1:02:58,  1.51it/s]
 22%|██▏       | 1600/7308 [20:11<1:02:48,  1.51it/s]
                                                     
{'loss': '3.656', 'grad_norm': '0.3855', 'learning_rate': '0.0006', 'epoch': '44.45'}
+
 22%|██▏       | 1600/7308 [20:11<1:02:48,  1.51it/s]
 22%|██▏       | 1601/7308 [20:11<1:02:41,  1.52it/s]
 22%|██▏       | 1602/7308 [20:12<1:02:33,  1.52it/s]
 22%|██▏       | 1603/7308 [20:12<1:02:26,  1.52it/s]
 22%|██▏       | 1604/7308 [20:13<1:02:25,  1.52it/s]
 22%|██▏       | 1605/7308 [20:14<1:02:23,  1.52it/s]
 22%|██▏       | 1606/7308 [20:14<1:02:22,  1.52it/s]
 22%|██▏       | 1607/7308 [20:15<1:02:18,  1.52it/s]
 22%|██▏       | 1608/7308 [20:16<1:02:18,  1.52it/s]
 22%|██▏       | 1609/7308 [20:16<1:02:18,  1.52it/s]
 22%|██▏       | 1610/7308 [20:17<1:02:16,  1.52it/s]
                                                     
{'loss': '3.67', 'grad_norm': '0.3506', 'learning_rate': '0.0006', 'epoch': '44.73'}
+
 22%|██▏       | 1610/7308 [20:17<1:02:16,  1.52it/s]
 22%|██▏       | 1611/7308 [20:18<1:02:15,  1.53it/s]
 22%|██▏       | 1612/7308 [20:18<1:02:13,  1.53it/s]
 22%|██▏       | 1613/7308 [20:19<1:02:15,  1.52it/s]
 22%|██▏       | 1614/7308 [20:20<1:02:14,  1.52it/s]
 22%|██▏       | 1615/7308 [20:20<1:02:13,  1.52it/s]
 22%|██▏       | 1616/7308 [20:21<1:02:11,  1.53it/s]
 22%|██▏       | 1617/7308 [20:22<1:02:09,  1.53it/s]
 22%|██▏       | 1618/7308 [20:22<1:02:13,  1.52it/s]
 22%|██▏       | 1619/7308 [20:23<1:02:09,  1.53it/s]
 22%|██▏       | 1620/7308 [20:23<50:38,  1.87it/s]  
                                                   
{'loss': '3.693', 'grad_norm': '0.4098', 'learning_rate': '0.0006', 'epoch': '45'}
+
 22%|██▏       | 1620/7308 [20:23<50:38,  1.87it/s]
 22%|██▏       | 1621/7308 [20:24<56:51,  1.67it/s]
 22%|██▏       | 1622/7308 [20:25<58:24,  1.62it/s]
 22%|██▏       | 1623/7308 [20:25<59:28,  1.59it/s]
 22%|██▏       | 1624/7308 [20:26<1:00:13,  1.57it/s]
 22%|██▏       | 1625/7308 [20:27<1:00:46,  1.56it/s]
 22%|██▏       | 1626/7308 [20:27<1:01:09,  1.55it/s]
 22%|██▏       | 1627/7308 [20:28<1:01:24,  1.54it/s]
 22%|██▏       | 1628/7308 [20:29<1:01:34,  1.54it/s]
 22%|██▏       | 1629/7308 [20:29<1:01:42,  1.53it/s]
 22%|██▏       | 1630/7308 [20:30<1:01:47,  1.53it/s]
                                                     
{'loss': '3.596', 'grad_norm': '0.4543', 'learning_rate': '0.0006', 'epoch': '45.28'}
+
 22%|██▏       | 1630/7308 [20:30<1:01:47,  1.53it/s]
 22%|██▏       | 1631/7308 [20:31<1:01:51,  1.53it/s]
 22%|██▏       | 1632/7308 [20:31<1:01:54,  1.53it/s]
 22%|██▏       | 1633/7308 [20:32<1:01:55,  1.53it/s]
 22%|██▏       | 1634/7308 [20:32<1:01:58,  1.53it/s]
 22%|██▏       | 1635/7308 [20:33<1:01:55,  1.53it/s]
 22%|██▏       | 1636/7308 [20:34<1:01:55,  1.53it/s]
 22%|██▏       | 1637/7308 [20:34<1:01:54,  1.53it/s]
 22%|██▏       | 1638/7308 [20:35<1:01:55,  1.53it/s]
 22%|██▏       | 1639/7308 [20:36<1:01:54,  1.53it/s]
 22%|██▏       | 1640/7308 [20:36<1:01:55,  1.53it/s]
                                                     
{'loss': '3.643', 'grad_norm': '0.3737', 'learning_rate': '0.0006', 'epoch': '45.57'}
+
 22%|██▏       | 1640/7308 [20:36<1:01:55,  1.53it/s]
 22%|██▏       | 1641/7308 [20:37<1:01:55,  1.53it/s]
 22%|██▏       | 1642/7308 [20:38<1:01:56,  1.52it/s]
 22%|██▏       | 1643/7308 [20:38<1:01:57,  1.52it/s]
 22%|██▏       | 1644/7308 [20:39<1:01:54,  1.52it/s]
 23%|██▎       | 1645/7308 [20:40<1:01:55,  1.52it/s]
 23%|██▎       | 1646/7308 [20:40<1:01:54,  1.52it/s]
 23%|██▎       | 1647/7308 [20:41<1:01:53,  1.52it/s]
 23%|██▎       | 1648/7308 [20:42<1:01:53,  1.52it/s]
 23%|██▎       | 1649/7308 [20:42<1:01:50,  1.53it/s]
 23%|██▎       | 1650/7308 [20:43<1:01:51,  1.52it/s]
                                                     
{'loss': '3.67', 'grad_norm': '0.3768', 'learning_rate': '0.0006', 'epoch': '45.85'}
+
 23%|██▎       | 1650/7308 [20:43<1:01:51,  1.52it/s]
 23%|██▎       | 1651/7308 [20:44<1:01:51,  1.52it/s]
 23%|██▎       | 1652/7308 [20:44<1:01:56,  1.52it/s]
 23%|██▎       | 1653/7308 [20:45<1:01:54,  1.52it/s]
 23%|██▎       | 1654/7308 [20:46<1:01:55,  1.52it/s]
 23%|██▎       | 1655/7308 [20:46<1:01:52,  1.52it/s]
 23%|██▎       | 1656/7308 [20:47<50:22,  1.87it/s]  [transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 19.13it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.86it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.67it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.09it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.76it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.57it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.43it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.34it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.30it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.26it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.26it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.26it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.22it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.25it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.26it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.26it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.25it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.23it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s]
                                                   
+
                                               
{'eval_loss': '4.249', 'eval_runtime': '3.239', 'eval_samples_per_second': '753.6', 'eval_steps_per_second': '12.04', 'epoch': '46'}
+
 23%|██▎       | 1656/7308 [20:50<50:22,  1.87it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s][eval] epoch 46.0 step 1656 loss 4.2492 (best 4.1514 @ ep 34.0, 6/20 since best)
+
+
                                               
 23%|██▎       | 1657/7308 [20:51<2:27:56,  1.57s/it]
 23%|██▎       | 1658/7308 [20:51<2:02:00,  1.30s/it]
 23%|██▎       | 1659/7308 [20:52<1:43:53,  1.10s/it]
 23%|██▎       | 1660/7308 [20:52<1:31:20,  1.03it/s]
                                                     
{'loss': '3.65', 'grad_norm': '0.4626', 'learning_rate': '0.0006', 'epoch': '46.11'}
+
 23%|██▎       | 1660/7308 [20:52<1:31:20,  1.03it/s]
 23%|██▎       | 1661/7308 [20:53<1:22:30,  1.14it/s]
 23%|██▎       | 1662/7308 [20:54<1:16:14,  1.23it/s]
 23%|██▎       | 1663/7308 [20:54<1:11:54,  1.31it/s]
 23%|██▎       | 1664/7308 [20:55<1:08:47,  1.37it/s]
 23%|██▎       | 1665/7308 [20:56<1:06:38,  1.41it/s]
 23%|██▎       | 1666/7308 [20:56<1:05:09,  1.44it/s]
 23%|██▎       | 1667/7308 [20:57<1:04:05,  1.47it/s]
 23%|██▎       | 1668/7308 [20:58<1:03:19,  1.48it/s]
 23%|██▎       | 1669/7308 [20:58<1:02:46,  1.50it/s]
 23%|██▎       | 1670/7308 [20:59<1:02:26,  1.50it/s]
                                                     
{'loss': '3.591', 'grad_norm': '0.4254', 'learning_rate': '0.0006', 'epoch': '46.4'}
+
 23%|██▎       | 1670/7308 [20:59<1:02:26,  1.50it/s]
 23%|██▎       | 1671/7308 [21:00<1:02:12,  1.51it/s]
 23%|██▎       | 1672/7308 [21:00<1:02:02,  1.51it/s]
 23%|██▎       | 1673/7308 [21:01<1:01:51,  1.52it/s]
 23%|██▎       | 1674/7308 [21:02<1:01:43,  1.52it/s]
 23%|██▎       | 1675/7308 [21:02<1:01:40,  1.52it/s]
 23%|██▎       | 1676/7308 [21:03<1:01:38,  1.52it/s]
 23%|██▎       | 1677/7308 [21:04<1:01:34,  1.52it/s]
 23%|██▎       | 1678/7308 [21:04<1:01:31,  1.52it/s]
 23%|██▎       | 1679/7308 [21:05<1:01:33,  1.52it/s]
 23%|██▎       | 1680/7308 [21:06<1:01:33,  1.52it/s]
                                                     
{'loss': '3.626', 'grad_norm': '0.3405', 'learning_rate': '0.0006', 'epoch': '46.68'}
+
 23%|██▎       | 1680/7308 [21:06<1:01:33,  1.52it/s]
 23%|██▎       | 1681/7308 [21:06<1:01:36,  1.52it/s]
 23%|██▎       | 1682/7308 [21:07<1:01:32,  1.52it/s]
 23%|██▎       | 1683/7308 [21:08<1:01:30,  1.52it/s]
 23%|██▎       | 1684/7308 [21:08<1:01:27,  1.52it/s]
 23%|██▎       | 1685/7308 [21:09<1:01:29,  1.52it/s]
 23%|██▎       | 1686/7308 [21:10<1:01:29,  1.52it/s]
 23%|██▎       | 1687/7308 [21:10<1:01:27,  1.52it/s]
 23%|██▎       | 1688/7308 [21:11<1:01:25,  1.52it/s]
 23%|██▎       | 1689/7308 [21:11<1:01:25,  1.52it/s]
 23%|██▎       | 1690/7308 [21:12<1:01:27,  1.52it/s]
                                                     
{'loss': '3.65', 'grad_norm': '0.3401', 'learning_rate': '0.0006', 'epoch': '46.96'}
+
 23%|██▎       | 1690/7308 [21:12<1:01:27,  1.52it/s]
 23%|██▎       | 1691/7308 [21:13<1:01:25,  1.52it/s]
 23%|██▎       | 1692/7308 [21:13<50:01,  1.87it/s]  
 23%|██▎       | 1693/7308 [21:14<56:11,  1.67it/s]
 23%|██▎       | 1694/7308 [21:14<57:45,  1.62it/s]
 23%|██▎       | 1695/7308 [21:15<58:47,  1.59it/s]
 23%|██▎       | 1696/7308 [21:16<59:33,  1.57it/s]
 23%|██▎       | 1697/7308 [21:16<1:00:04,  1.56it/s]
 23%|██▎       | 1698/7308 [21:17<1:00:27,  1.55it/s]
 23%|██▎       | 1699/7308 [21:18<1:00:41,  1.54it/s]
 23%|██▎       | 1700/7308 [21:18<1:00:51,  1.54it/s]
                                                     
{'loss': '3.574', 'grad_norm': '0.4223', 'learning_rate': '0.0006', 'epoch': '47.23'}
+
 23%|██▎       | 1700/7308 [21:18<1:00:51,  1.54it/s]
 23%|██▎       | 1701/7308 [21:19<1:00:58,  1.53it/s]
 23%|██▎       | 1702/7308 [21:20<1:01:03,  1.53it/s]
 23%|██▎       | 1703/7308 [21:20<1:01:07,  1.53it/s]
 23%|██▎       | 1704/7308 [21:21<1:01:08,  1.53it/s]
 23%|██▎       | 1705/7308 [21:22<1:01:07,  1.53it/s]
 23%|██▎       | 1706/7308 [21:22<1:01:07,  1.53it/s]
 23%|██▎       | 1707/7308 [21:23<1:01:09,  1.53it/s]
 23%|██▎       | 1708/7308 [21:24<1:01:09,  1.53it/s]
 23%|██▎       | 1709/7308 [21:24<1:01:09,  1.53it/s]
 23%|██▎       | 1710/7308 [21:25<1:01:09,  1.53it/s]
                                                     
{'loss': '3.604', 'grad_norm': '0.5106', 'learning_rate': '0.0006', 'epoch': '47.51'}
+
 23%|██▎       | 1710/7308 [21:25<1:01:09,  1.53it/s]
 23%|██▎       | 1711/7308 [21:26<1:01:09,  1.53it/s]
 23%|██▎       | 1712/7308 [21:26<1:01:09,  1.52it/s]
 23%|██▎       | 1713/7308 [21:27<1:01:08,  1.52it/s]
 23%|██▎       | 1714/7308 [21:28<1:01:08,  1.53it/s]
 23%|██▎       | 1715/7308 [21:28<1:01:05,  1.53it/s]
 23%|██▎       | 1716/7308 [21:29<1:01:07,  1.52it/s]
 23%|██▎       | 1717/7308 [21:30<1:01:06,  1.52it/s]
 24%|██▎       | 1718/7308 [21:30<1:01:04,  1.53it/s]
 24%|██▎       | 1719/7308 [21:31<1:01:02,  1.53it/s]
 24%|██▎       | 1720/7308 [21:32<1:01:02,  1.53it/s]
                                                     
{'loss': '3.631', 'grad_norm': '0.4309', 'learning_rate': '0.0006', 'epoch': '47.79'}
+
 24%|██▎       | 1720/7308 [21:32<1:01:02,  1.53it/s]
 24%|██▎       | 1721/7308 [21:32<1:01:03,  1.52it/s]
 24%|██▎       | 1722/7308 [21:33<1:01:05,  1.52it/s]
 24%|██▎       | 1723/7308 [21:33<1:01:04,  1.52it/s]
 24%|██▎       | 1724/7308 [21:34<1:01:01,  1.53it/s]
 24%|██▎       | 1725/7308 [21:35<1:01:01,  1.52it/s]
 24%|██▎       | 1726/7308 [21:35<1:01:03,  1.52it/s]
 24%|██▎       | 1727/7308 [21:36<1:00:59,  1.52it/s]
 24%|██▎       | 1728/7308 [21:36<49:42,  1.87it/s]  [transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 19.03it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.94it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.70it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.06it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.80it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.60it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.49it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.39it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.35it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.30it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.29it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.27it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.26it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.25it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.26it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.25it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.25it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.23it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s]
                                                   
+
                                               
{'eval_loss': '4.267', 'eval_runtime': '3.245', 'eval_samples_per_second': '752.3', 'eval_steps_per_second': '12.02', 'epoch': '48'}
+
 24%|██▎       | 1728/7308 [21:40<49:42,  1.87it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s][eval] epoch 48.0 step 1728 loss 4.2668 (best 4.1514 @ ep 34.0, 7/20 since best)
+
+
                                               
 24%|██▎       | 1729/7308 [21:40<2:25:51,  1.57s/it]
 24%|██▎       | 1730/7308 [21:41<2:00:20,  1.29s/it]
                                                     
{'loss': '3.608', 'grad_norm': '0.4071', 'learning_rate': '0.0006', 'epoch': '48.06'}
+
 24%|██▎       | 1730/7308 [21:41<2:00:20,  1.29s/it]
 24%|██▎       | 1731/7308 [21:42<1:42:33,  1.10s/it]
 24%|██▎       | 1732/7308 [21:42<1:30:04,  1.03it/s]
 24%|██▎       | 1733/7308 [21:43<1:21:18,  1.14it/s]
 24%|██▎       | 1734/7308 [21:44<1:15:08,  1.24it/s]
 24%|██▎       | 1735/7308 [21:44<1:10:55,  1.31it/s]
 24%|██▍       | 1736/7308 [21:45<1:07:56,  1.37it/s]
 24%|██▍       | 1737/7308 [21:46<1:05:48,  1.41it/s]
 24%|██▍       | 1738/7308 [21:46<1:04:18,  1.44it/s]
 24%|██▍       | 1739/7308 [21:47<1:03:15,  1.47it/s]
 24%|██▍       | 1740/7308 [21:48<1:02:30,  1.48it/s]
                                                     
{'loss': '3.554', 'grad_norm': '0.3597', 'learning_rate': '0.0006', 'epoch': '48.34'}
+
 24%|██▍       | 1740/7308 [21:48<1:02:30,  1.48it/s]
 24%|██▍       | 1741/7308 [21:48<1:01:59,  1.50it/s]
 24%|██▍       | 1742/7308 [21:49<1:01:35,  1.51it/s]
 24%|██▍       | 1743/7308 [21:50<1:01:19,  1.51it/s]
 24%|██▍       | 1744/7308 [21:50<1:01:08,  1.52it/s]
 24%|██▍       | 1745/7308 [21:51<1:01:00,  1.52it/s]
 24%|██▍       | 1746/7308 [21:51<1:00:54,  1.52it/s]
 24%|██▍       | 1747/7308 [21:52<1:00:51,  1.52it/s]
 24%|██▍       | 1748/7308 [21:53<1:00:47,  1.52it/s]
 24%|██▍       | 1749/7308 [21:53<1:00:47,  1.52it/s]
 24%|██▍       | 1750/7308 [21:54<1:00:45,  1.52it/s]
                                                     
{'loss': '3.566', 'grad_norm': '0.3497', 'learning_rate': '0.0006', 'epoch': '48.62'}
+
 24%|██▍       | 1750/7308 [21:54<1:00:45,  1.52it/s]
 24%|██▍       | 1751/7308 [21:55<1:00:44,  1.52it/s]
 24%|██▍       | 1752/7308 [21:55<1:00:43,  1.52it/s]
 24%|██▍       | 1753/7308 [21:56<1:00:40,  1.53it/s]
 24%|██▍       | 1754/7308 [21:57<1:00:38,  1.53it/s]
 24%|██▍       | 1755/7308 [21:57<1:00:36,  1.53it/s]
 24%|██▍       | 1756/7308 [21:58<1:00:35,  1.53it/s]
 24%|██▍       | 1757/7308 [21:59<1:00:33,  1.53it/s]
 24%|██▍       | 1758/7308 [21:59<1:00:33,  1.53it/s]
 24%|██▍       | 1759/7308 [22:00<1:00:33,  1.53it/s]
 24%|██▍       | 1760/7308 [22:01<1:00:33,  1.53it/s]
                                                     
{'loss': '3.601', 'grad_norm': '0.3339', 'learning_rate': '0.0006', 'epoch': '48.9'}
+
 24%|██▍       | 1760/7308 [22:01<1:00:33,  1.53it/s]
 24%|██▍       | 1761/7308 [22:01<1:00:34,  1.53it/s]
 24%|██▍       | 1762/7308 [22:02<1:00:32,  1.53it/s]
 24%|██▍       | 1763/7308 [22:03<1:00:31,  1.53it/s]
 24%|██▍       | 1764/7308 [22:03<49:18,  1.87it/s]  
 24%|██▍       | 1765/7308 [22:04<55:24,  1.67it/s]
 24%|██▍       | 1766/7308 [22:04<57:00,  1.62it/s]
 24%|██▍       | 1767/7308 [22:05<58:01,  1.59it/s]
 24%|██▍       | 1768/7308 [22:06<58:49,  1.57it/s]
 24%|██▍       | 1769/7308 [22:06<59:20,  1.56it/s]
 24%|██▍       | 1770/7308 [22:07<59:41,  1.55it/s]
                                                   
{'loss': '3.544', 'grad_norm': '0.3936', 'learning_rate': '0.0006', 'epoch': '49.17'}
+
 24%|██▍       | 1770/7308 [22:07<59:41,  1.55it/s]
 24%|██▍       | 1771/7308 [22:08<59:55,  1.54it/s]
 24%|██▍       | 1772/7308 [22:08<1:00:03,  1.54it/s]
 24%|██▍       | 1773/7308 [22:09<1:00:13,  1.53it/s]
 24%|██▍       | 1774/7308 [22:10<1:00:19,  1.53it/s]
 24%|██▍       | 1775/7308 [22:10<1:00:21,  1.53it/s]
 24%|██▍       | 1776/7308 [22:11<1:00:20,  1.53it/s]
 24%|██▍       | 1777/7308 [22:11<1:00:21,  1.53it/s]
 24%|██▍       | 1778/7308 [22:12<1:00:24,  1.53it/s]
 24%|██▍       | 1779/7308 [22:13<1:00:23,  1.53it/s]
 24%|██▍       | 1780/7308 [22:13<1:00:21,  1.53it/s]
                                                     
{'loss': '3.543', 'grad_norm': '0.3956', 'learning_rate': '0.0006', 'epoch': '49.45'}
+
 24%|██▍       | 1780/7308 [22:13<1:00:21,  1.53it/s]
 24%|██▍       | 1781/7308 [22:14<1:00:22,  1.53it/s]
 24%|██▍       | 1782/7308 [22:15<1:00:24,  1.52it/s]
 24%|██▍       | 1783/7308 [22:15<1:00:26,  1.52it/s]
 24%|██▍       | 1784/7308 [22:16<1:00:24,  1.52it/s]
 24%|██▍       | 1785/7308 [22:17<1:00:22,  1.52it/s]
 24%|██▍       | 1786/7308 [22:17<1:00:20,  1.53it/s]
 24%|██▍       | 1787/7308 [22:18<1:00:20,  1.52it/s]
 24%|██▍       | 1788/7308 [22:19<1:00:21,  1.52it/s]
 24%|██▍       | 1789/7308 [22:19<1:00:19,  1.52it/s]
 24%|██▍       | 1790/7308 [22:20<1:00:21,  1.52it/s]
                                                     
{'loss': '3.558', 'grad_norm': '0.3594', 'learning_rate': '0.0006', 'epoch': '49.73'}
+
 24%|██▍       | 1790/7308 [22:20<1:00:21,  1.52it/s]
 25%|██▍       | 1791/7308 [22:21<1:00:20,  1.52it/s]
 25%|██▍       | 1792/7308 [22:21<1:00:19,  1.52it/s]
 25%|██▍       | 1793/7308 [22:22<1:00:19,  1.52it/s]
 25%|██▍       | 1794/7308 [22:23<1:00:19,  1.52it/s]
 25%|██▍       | 1795/7308 [22:23<1:00:16,  1.52it/s]
 25%|██▍       | 1796/7308 [22:24<1:00:16,  1.52it/s]
 25%|██▍       | 1797/7308 [22:25<1:00:14,  1.52it/s]
 25%|██▍       | 1798/7308 [22:25<1:00:15,  1.52it/s]
 25%|██▍       | 1799/7308 [22:26<1:00:13,  1.52it/s]
 25%|██▍       | 1800/7308 [22:26<49:04,  1.87it/s]  
                                                   
{'loss': '3.592', 'grad_norm': '0.4403', 'learning_rate': '0.0006', 'epoch': '50'}
+
 25%|██▍       | 1800/7308 [22:26<49:04,  1.87it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.97it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.91it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.69it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.09it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.80it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.59it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.49it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.40it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.36it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.30it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.29it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.27it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.27it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.25it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.25it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.24it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.23it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.22it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.20it/s]
                                                   
+
                                               
{'eval_loss': '4.298', 'eval_runtime': '3.263', 'eval_samples_per_second': '748.1', 'eval_steps_per_second': '11.95', 'epoch': '50'}
+
 25%|██▍       | 1800/7308 [22:29<49:04,  1.87it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.20it/s][eval] epoch 50.0 step 1800 loss 4.2984 (best 4.1514 @ ep 34.0, 8/20 since best)
+
+
                                               
 25%|██▍       | 1801/7308 [22:30<2:24:59,  1.58s/it]
 25%|██▍       | 1802/7308 [22:31<1:59:32,  1.30s/it]
 25%|██▍       | 1803/7308 [22:32<1:41:40,  1.11s/it]
 25%|██▍       | 1804/7308 [22:32<1:29:12,  1.03it/s]
 25%|██▍       | 1805/7308 [22:33<1:20:28,  1.14it/s]
 25%|██▍       | 1806/7308 [22:33<1:14:21,  1.23it/s]
 25%|██▍       | 1807/7308 [22:34<1:10:06,  1.31it/s]
 25%|██▍       | 1808/7308 [22:35<1:07:07,  1.37it/s]
 25%|██▍       | 1809/7308 [22:35<1:05:00,  1.41it/s]
 25%|██▍       | 1810/7308 [22:36<1:03:33,  1.44it/s]
                                                     
{'loss': '3.495', 'grad_norm': '0.4257', 'learning_rate': '0.0006', 'epoch': '50.28'}
+
 25%|██▍       | 1810/7308 [22:36<1:03:33,  1.44it/s]
 25%|██▍       | 1811/7308 [22:37<1:02:33,  1.46it/s]
 25%|██▍       | 1812/7308 [22:37<1:01:46,  1.48it/s]
 25%|██▍       | 1813/7308 [22:38<1:01:17,  1.49it/s]
 25%|██▍       | 1814/7308 [22:39<1:00:55,  1.50it/s]
 25%|██▍       | 1815/7308 [22:39<1:00:39,  1.51it/s]
 25%|██▍       | 1816/7308 [22:40<1:00:29,  1.51it/s]
 25%|██▍       | 1817/7308 [22:41<1:00:18,  1.52it/s]
 25%|██▍       | 1818/7308 [22:41<1:00:15,  1.52it/s]
 25%|██▍       | 1819/7308 [22:42<1:00:10,  1.52it/s]
 25%|██▍       | 1820/7308 [22:43<1:00:07,  1.52it/s]
                                                     
{'loss': '3.535', 'grad_norm': '0.4663', 'learning_rate': '0.0006', 'epoch': '50.57'}
+
 25%|██▍       | 1820/7308 [22:43<1:00:07,  1.52it/s]
 25%|██▍       | 1821/7308 [22:43<1:00:04,  1.52it/s]
 25%|██▍       | 1822/7308 [22:44<59:59,  1.52it/s]  
 25%|██▍       | 1823/7308 [22:45<1:00:03,  1.52it/s]
 25%|██▍       | 1824/7308 [22:45<1:00:02,  1.52it/s]
 25%|██▍       | 1825/7308 [22:46<59:59,  1.52it/s]  
 25%|██▍       | 1826/7308 [22:47<59:59,  1.52it/s]
 25%|██▌       | 1827/7308 [22:47<59:56,  1.52it/s]
 25%|██▌       | 1828/7308 [22:48<59:55,  1.52it/s]
 25%|██▌       | 1829/7308 [22:49<59:55,  1.52it/s]
 25%|██▌       | 1830/7308 [22:49<59:54,  1.52it/s]
                                                   
{'loss': '3.561', 'grad_norm': '0.4071', 'learning_rate': '0.0006', 'epoch': '50.85'}
+
 25%|██▌       | 1830/7308 [22:49<59:54,  1.52it/s]
 25%|██▌       | 1831/7308 [22:50<59:54,  1.52it/s]
 25%|██▌       | 1832/7308 [22:51<59:53,  1.52it/s]
 25%|██▌       | 1833/7308 [22:51<59:51,  1.52it/s]
 25%|██▌       | 1834/7308 [22:52<59:54,  1.52it/s]
 25%|██▌       | 1835/7308 [22:53<59:50,  1.52it/s]
 25%|██▌       | 1836/7308 [22:53<48:44,  1.87it/s]
 25%|██▌       | 1837/7308 [22:54<54:37,  1.67it/s]
 25%|██▌       | 1838/7308 [22:54<56:10,  1.62it/s]
 25%|██▌       | 1839/7308 [22:55<57:12,  1.59it/s]
 25%|██▌       | 1840/7308 [22:55<57:58,  1.57it/s]
                                                   
{'loss': '3.533', 'grad_norm': '0.3547', 'learning_rate': '0.0006', 'epoch': '51.11'}
+
 25%|██▌       | 1840/7308 [22:55<57:58,  1.57it/s]
 25%|██▌       | 1841/7308 [22:56<58:33,  1.56it/s]
 25%|██▌       | 1842/7308 [22:57<58:53,  1.55it/s]
 25%|██▌       | 1843/7308 [22:57<59:08,  1.54it/s]
 25%|██▌       | 1844/7308 [22:58<59:17,  1.54it/s]
 25%|██▌       | 1845/7308 [22:59<59:25,  1.53it/s]
 25%|██▌       | 1846/7308 [22:59<59:32,  1.53it/s]
 25%|██▌       | 1847/7308 [23:00<59:34,  1.53it/s]
 25%|██▌       | 1848/7308 [23:01<59:35,  1.53it/s]
 25%|██▌       | 1849/7308 [23:01<59:36,  1.53it/s]
 25%|██▌       | 1850/7308 [23:02<59:37,  1.53it/s]
                                                   
{'loss': '3.493', 'grad_norm': '0.5284', 'learning_rate': '0.0006', 'epoch': '51.4'}
+
 25%|██▌       | 1850/7308 [23:02<59:37,  1.53it/s]
 25%|██▌       | 1851/7308 [23:03<59:41,  1.52it/s]
 25%|██▌       | 1852/7308 [23:03<59:42,  1.52it/s]
 25%|██▌       | 1853/7308 [23:04<59:39,  1.52it/s]
 25%|██▌       | 1854/7308 [23:05<59:37,  1.52it/s]
 25%|██▌       | 1855/7308 [23:05<59:34,  1.53it/s]
 25%|██▌       | 1856/7308 [23:06<59:36,  1.52it/s]
 25%|██▌       | 1857/7308 [23:07<59:36,  1.52it/s]
 25%|██▌       | 1858/7308 [23:07<59:34,  1.52it/s]
 25%|██▌       | 1859/7308 [23:08<59:31,  1.53it/s]
 25%|██▌       | 1860/7308 [23:09<59:29,  1.53it/s]
                                                   
{'loss': '3.528', 'grad_norm': '0.4248', 'learning_rate': '0.0006', 'epoch': '51.68'}
+
 25%|██▌       | 1860/7308 [23:09<59:29,  1.53it/s]
 25%|██▌       | 1861/7308 [23:09<59:35,  1.52it/s]
 25%|██▌       | 1862/7308 [23:10<59:33,  1.52it/s]
 25%|██▌       | 1863/7308 [23:11<59:31,  1.52it/s]
 26%|██▌       | 1864/7308 [23:11<59:28,  1.53it/s]
 26%|██▌       | 1865/7308 [23:12<59:27,  1.53it/s]
 26%|██▌       | 1866/7308 [23:13<59:30,  1.52it/s]
 26%|██▌       | 1867/7308 [23:13<59:29,  1.52it/s]
 26%|██▌       | 1868/7308 [23:14<59:26,  1.53it/s]
 26%|██▌       | 1869/7308 [23:14<59:26,  1.53it/s]
 26%|██▌       | 1870/7308 [23:15<59:27,  1.52it/s]
                                                   
{'loss': '3.549', 'grad_norm': '0.3683', 'learning_rate': '0.0006', 'epoch': '51.96'}
+
 26%|██▌       | 1870/7308 [23:15<59:27,  1.52it/s]
 26%|██▌       | 1871/7308 [23:16<59:26,  1.52it/s]
 26%|██▌       | 1872/7308 [23:16<48:26,  1.87it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.49it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.71it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.58it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.03it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.74it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.56it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.46it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.38it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.34it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.29it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.29it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.28it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.26it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.25it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.24it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.25it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.24it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.24it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s]
                                                   
+
                                               
{'eval_loss': '4.318', 'eval_runtime': '3.234', 'eval_samples_per_second': '754.8', 'eval_steps_per_second': '12.06', 'epoch': '52'}
+
 26%|██▌       | 1872/7308 [23:19<48:26,  1.87it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s][eval] epoch 52.0 step 1872 loss 4.3181 (best 4.1514 @ ep 34.0, 9/20 since best)
+
+
                                               
 26%|██▌       | 1873/7308 [23:20<2:22:09,  1.57s/it]
 26%|██▌       | 1874/7308 [23:21<1:57:16,  1.29s/it]
 26%|██▌       | 1875/7308 [23:21<1:39:50,  1.10s/it]
 26%|██▌       | 1876/7308 [23:22<1:27:42,  1.03it/s]
 26%|██▌       | 1877/7308 [23:23<1:19:10,  1.14it/s]
 26%|██▌       | 1878/7308 [23:23<1:13:11,  1.24it/s]
 26%|██▌       | 1879/7308 [23:24<1:08:59,  1.31it/s]
 26%|██▌       | 1880/7308 [23:25<1:06:07,  1.37it/s]
                                                     
{'loss': '3.484', 'grad_norm': '0.4871', 'learning_rate': '0.0006', 'epoch': '52.23'}
+
 26%|██▌       | 1880/7308 [23:25<1:06:07,  1.37it/s]
 26%|██▌       | 1881/7308 [23:25<1:04:07,  1.41it/s]
 26%|██▌       | 1882/7308 [23:26<1:02:39,  1.44it/s]
 26%|██▌       | 1883/7308 [23:27<1:01:36,  1.47it/s]
 26%|██▌       | 1884/7308 [23:27<1:00:53,  1.48it/s]
 26%|██▌       | 1885/7308 [23:28<1:00:23,  1.50it/s]
 26%|██▌       | 1886/7308 [23:29<1:00:02,  1.50it/s]
 26%|██▌       | 1887/7308 [23:29<59:46,  1.51it/s]  
 26%|██▌       | 1888/7308 [23:30<59:36,  1.52it/s]
 26%|██▌       | 1889/7308 [23:31<59:31,  1.52it/s]
 26%|██▌       | 1890/7308 [23:31<59:25,  1.52it/s]
                                                   
{'loss': '3.482', 'grad_norm': '0.4655', 'learning_rate': '0.0006', 'epoch': '52.51'}
+
 26%|██▌       | 1890/7308 [23:31<59:25,  1.52it/s]
 26%|██▌       | 1891/7308 [23:32<59:22,  1.52it/s]
 26%|██▌       | 1892/7308 [23:33<59:17,  1.52it/s]
 26%|██▌       | 1893/7308 [23:33<59:14,  1.52it/s]
 26%|██▌       | 1894/7308 [23:34<59:12,  1.52it/s]
 26%|██▌       | 1895/7308 [23:34<59:13,  1.52it/s]
 26%|██▌       | 1896/7308 [23:35<59:10,  1.52it/s]
 26%|██▌       | 1897/7308 [23:36<59:09,  1.52it/s]
 26%|██▌       | 1898/7308 [23:36<59:06,  1.53it/s]
 26%|██▌       | 1899/7308 [23:37<59:05,  1.53it/s]
 26%|██▌       | 1900/7308 [23:38<59:02,  1.53it/s]
                                                   
{'loss': '3.523', 'grad_norm': '0.3858', 'learning_rate': '0.0006', 'epoch': '52.79'}
+
 26%|██▌       | 1900/7308 [23:38<59:02,  1.53it/s]
 26%|██▌       | 1901/7308 [23:38<59:04,  1.53it/s]
 26%|██▌       | 1902/7308 [23:39<59:04,  1.53it/s]
 26%|██▌       | 1903/7308 [23:40<59:04,  1.52it/s]
 26%|██▌       | 1904/7308 [23:40<59:03,  1.52it/s]
 26%|██▌       | 1905/7308 [23:41<59:00,  1.53it/s]
 26%|██▌       | 1906/7308 [23:42<58:58,  1.53it/s]
 26%|██▌       | 1907/7308 [23:42<58:57,  1.53it/s]
 26%|██▌       | 1908/7308 [23:43<48:02,  1.87it/s]
 26%|██▌       | 1909/7308 [23:43<53:53,  1.67it/s]
 26%|██▌       | 1910/7308 [23:44<55:26,  1.62it/s]
                                                   
{'loss': '3.512', 'grad_norm': '0.5233', 'learning_rate': '0.0006', 'epoch': '53.06'}
+
 26%|██▌       | 1910/7308 [23:44<55:26,  1.62it/s]
 26%|██▌       | 1911/7308 [23:45<56:33,  1.59it/s]
 26%|██▌       | 1912/7308 [23:45<57:17,  1.57it/s]
 26%|██▌       | 1913/7308 [23:46<57:46,  1.56it/s]
 26%|██▌       | 1914/7308 [23:47<58:06,  1.55it/s]
 26%|██▌       | 1915/7308 [23:47<58:22,  1.54it/s]
 26%|██▌       | 1916/7308 [23:48<58:34,  1.53it/s]
 26%|██▌       | 1917/7308 [23:49<58:40,  1.53it/s]
 26%|██▌       | 1918/7308 [23:49<58:43,  1.53it/s]
 26%|██▋       | 1919/7308 [23:50<58:45,  1.53it/s]
 26%|██▋       | 1920/7308 [23:51<58:49,  1.53it/s]
                                                   
{'loss': '3.448', 'grad_norm': '0.4564', 'learning_rate': '0.0006', 'epoch': '53.34'}
+
 26%|██▋       | 1920/7308 [23:51<58:49,  1.53it/s]
 26%|██▋       | 1921/7308 [23:51<58:53,  1.52it/s]
 26%|██▋       | 1922/7308 [23:52<58:52,  1.52it/s]
 26%|██▋       | 1923/7308 [23:53<58:51,  1.53it/s]
 26%|██▋       | 1924/7308 [23:53<58:48,  1.53it/s]
 26%|██▋       | 1925/7308 [23:54<58:48,  1.53it/s]
 26%|██▋       | 1926/7308 [23:54<58:49,  1.52it/s]
 26%|██▋       | 1927/7308 [23:55<58:49,  1.52it/s]
 26%|██▋       | 1928/7308 [23:56<58:48,  1.52it/s]
 26%|██▋       | 1929/7308 [23:56<58:47,  1.53it/s]
 26%|██▋       | 1930/7308 [23:57<58:47,  1.52it/s]
                                                   
{'loss': '3.483', 'grad_norm': '0.3726', 'learning_rate': '0.0006', 'epoch': '53.62'}
+
 26%|██▋       | 1930/7308 [23:57<58:47,  1.52it/s]
 26%|██▋       | 1931/7308 [23:58<58:48,  1.52it/s]
 26%|██▋       | 1932/7308 [23:58<58:48,  1.52it/s]
 26%|██▋       | 1933/7308 [23:59<58:45,  1.52it/s]
 26%|██▋       | 1934/7308 [24:00<58:43,  1.53it/s]
 26%|██▋       | 1935/7308 [24:00<58:44,  1.52it/s]
 26%|██▋       | 1936/7308 [24:01<58:45,  1.52it/s]
 27%|██▋       | 1937/7308 [24:02<58:45,  1.52it/s]
 27%|██▋       | 1938/7308 [24:02<58:44,  1.52it/s]
 27%|██▋       | 1939/7308 [24:03<58:41,  1.52it/s]
 27%|██▋       | 1940/7308 [24:04<58:40,  1.52it/s]
                                                   
{'loss': '3.505', 'grad_norm': '0.3759', 'learning_rate': '0.0006', 'epoch': '53.9'}
+
 27%|██▋       | 1940/7308 [24:04<58:40,  1.52it/s]
 27%|██▋       | 1941/7308 [24:04<58:43,  1.52it/s]
 27%|██▋       | 1942/7308 [24:05<58:42,  1.52it/s]
 27%|██▋       | 1943/7308 [24:06<58:37,  1.53it/s]
 27%|██▋       | 1944/7308 [24:06<47:46,  1.87it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.46it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.77it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.61it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.05it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.76it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.58it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.46it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.40it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.34it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.31it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.31it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.29it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.28it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.26it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.27it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.26it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.25it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.25it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.25it/s]
                                                   
+
                                               
{'eval_loss': '4.354', 'eval_runtime': '3.26', 'eval_samples_per_second': '748.8', 'eval_steps_per_second': '11.96', 'epoch': '54'}
+
 27%|██▋       | 1944/7308 [24:09<47:46,  1.87it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.25it/s][eval] epoch 54.0 step 1944 loss 4.3544 (best 4.1514 @ ep 34.0, 10/20 since best)
+
+
                                               
 27%|██▋       | 1945/7308 [24:10<2:20:56,  1.58s/it]
 27%|██▋       | 1946/7308 [24:11<1:56:11,  1.30s/it]
 27%|██▋       | 1947/7308 [24:11<1:38:54,  1.11s/it]
 27%|██▋       | 1948/7308 [24:12<1:26:47,  1.03it/s]
 27%|██▋       | 1949/7308 [24:13<1:18:18,  1.14it/s]
 27%|██▋       | 1950/7308 [24:13<1:12:21,  1.23it/s]
                                                     
{'loss': '3.462', 'grad_norm': '0.5015', 'learning_rate': '0.0006', 'epoch': '54.17'}
+
 27%|██▋       | 1950/7308 [24:13<1:12:21,  1.23it/s]
 27%|██▋       | 1951/7308 [24:14<1:08:13,  1.31it/s]
 27%|██▋       | 1952/7308 [24:14<1:05:20,  1.37it/s]
 27%|██▋       | 1953/7308 [24:15<1:03:18,  1.41it/s]
 27%|██▋       | 1954/7308 [24:16<1:01:54,  1.44it/s]
 27%|██▋       | 1955/7308 [24:16<1:00:52,  1.47it/s]
 27%|██▋       | 1956/7308 [24:17<1:00:09,  1.48it/s]
 27%|██▋       | 1957/7308 [24:18<59:38,  1.50it/s]  
 27%|██▋       | 1958/7308 [24:18<59:18,  1.50it/s]
 27%|██▋       | 1959/7308 [24:19<59:03,  1.51it/s]
 27%|██▋       | 1960/7308 [24:20<58:51,  1.51it/s]
                                                   
{'loss': '3.447', 'grad_norm': '0.4381', 'learning_rate': '0.0006', 'epoch': '54.45'}
+
 27%|██▋       | 1960/7308 [24:20<58:51,  1.51it/s]
 27%|██▋       | 1961/7308 [24:20<58:44,  1.52it/s]
 27%|██▋       | 1962/7308 [24:21<58:40,  1.52it/s]
 27%|██▋       | 1963/7308 [24:22<58:35,  1.52it/s]
 27%|██▋       | 1964/7308 [24:22<58:31,  1.52it/s]
 27%|██▋       | 1965/7308 [24:23<58:30,  1.52it/s]
 27%|██▋       | 1966/7308 [24:24<58:25,  1.52it/s]
 27%|██▋       | 1967/7308 [24:24<58:24,  1.52it/s]
 27%|██▋       | 1968/7308 [24:25<58:28,  1.52it/s]
 27%|██▋       | 1969/7308 [24:26<58:28,  1.52it/s]
 27%|██▋       | 1970/7308 [24:26<58:26,  1.52it/s]
                                                   
{'loss': '3.472', 'grad_norm': '0.3937', 'learning_rate': '0.0006', 'epoch': '54.73'}
+
 27%|██▋       | 1970/7308 [24:26<58:26,  1.52it/s]
 27%|██▋       | 1971/7308 [24:27<58:24,  1.52it/s]
 27%|██▋       | 1972/7308 [24:28<58:22,  1.52it/s]
 27%|██▋       | 1973/7308 [24:28<58:21,  1.52it/s]
 27%|██▋       | 1974/7308 [24:29<58:20,  1.52it/s]
 27%|██▋       | 1975/7308 [24:30<58:19,  1.52it/s]
 27%|██▋       | 1976/7308 [24:30<58:16,  1.52it/s]
 27%|██▋       | 1977/7308 [24:31<58:15,  1.53it/s]
 27%|██▋       | 1978/7308 [24:32<58:17,  1.52it/s]
 27%|██▋       | 1979/7308 [24:32<58:16,  1.52it/s]
 27%|██▋       | 1980/7308 [24:32<47:28,  1.87it/s]
                                                   
{'loss': '3.503', 'grad_norm': '0.4473', 'learning_rate': '0.0006', 'epoch': '55'}
+
 27%|██▋       | 1980/7308 [24:32<47:28,  1.87it/s]
 27%|██▋       | 1981/7308 [24:33<52:41,  1.68it/s]
 27%|██▋       | 1982/7308 [24:34<54:20,  1.63it/s]
 27%|██▋       | 1983/7308 [24:35<55:29,  1.60it/s]
 27%|██▋       | 1984/7308 [24:35<56:21,  1.57it/s]
 27%|██▋       | 1985/7308 [24:36<56:52,  1.56it/s]
 27%|██▋       | 1986/7308 [24:36<57:13,  1.55it/s]
 27%|██▋       | 1987/7308 [24:37<57:30,  1.54it/s]
 27%|██▋       | 1988/7308 [24:38<57:40,  1.54it/s]
 27%|██▋       | 1989/7308 [24:38<57:46,  1.53it/s]
 27%|██▋       | 1990/7308 [24:39<57:51,  1.53it/s]
                                                   
{'loss': '3.403', 'grad_norm': '0.4893', 'learning_rate': '0.0006', 'epoch': '55.28'}
+
 27%|██▋       | 1990/7308 [24:39<57:51,  1.53it/s]
 27%|██▋       | 1991/7308 [24:40<58:00,  1.53it/s]
 27%|██▋       | 1992/7308 [24:40<58:02,  1.53it/s]
 27%|██▋       | 1993/7308 [24:41<58:01,  1.53it/s]
 27%|██▋       | 1994/7308 [24:42<58:02,  1.53it/s]
 27%|██▋       | 1995/7308 [24:42<58:00,  1.53it/s]
 27%|██▋       | 1996/7308 [24:43<58:02,  1.53it/s]
 27%|██▋       | 1997/7308 [24:44<58:01,  1.53it/s]
 27%|██▋       | 1998/7308 [24:44<58:00,  1.53it/s]
 27%|██▋       | 1999/7308 [24:45<57:59,  1.53it/s]
 27%|██▋       | 2000/7308 [24:46<58:01,  1.52it/s]
                                                   
{'loss': '3.443', 'grad_norm': '0.4805', 'learning_rate': '0.0006', 'epoch': '55.57'}
+
 27%|██▋       | 2000/7308 [24:46<58:01,  1.52it/s]
 27%|██▋       | 2001/7308 [24:46<58:03,  1.52it/s]
 27%|██▋       | 2002/7308 [24:47<58:02,  1.52it/s]
 27%|██▋       | 2003/7308 [24:48<57:59,  1.52it/s]
 27%|██▋       | 2004/7308 [24:48<57:57,  1.53it/s]
 27%|██▋       | 2005/7308 [24:49<57:58,  1.52it/s]
 27%|██▋       | 2006/7308 [24:50<57:57,  1.52it/s]
 27%|██▋       | 2007/7308 [24:50<57:57,  1.52it/s]
 27%|██▋       | 2008/7308 [24:51<57:55,  1.52it/s]
 27%|██▋       | 2009/7308 [24:52<57:53,  1.53it/s]
 28%|██▊       | 2010/7308 [24:52<57:55,  1.52it/s]
                                                   
{'loss': '3.467', 'grad_norm': '0.3953', 'learning_rate': '0.0006', 'epoch': '55.85'}
+
 28%|██▊       | 2010/7308 [24:52<57:55,  1.52it/s]
 28%|██▊       | 2011/7308 [24:53<57:56,  1.52it/s]
 28%|██▊       | 2012/7308 [24:54<57:55,  1.52it/s]
 28%|██▊       | 2013/7308 [24:54<57:53,  1.52it/s]
 28%|██▊       | 2014/7308 [24:55<57:52,  1.52it/s]
 28%|██▊       | 2015/7308 [24:55<57:51,  1.52it/s]
 28%|██▊       | 2016/7308 [24:56<47:08,  1.87it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 19.21it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 15.01it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.72it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.12it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.79it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.61it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.48it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.39it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.35it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.31it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.30it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.27it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.27it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.26it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.26it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.25it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.25it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.25it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.22it/s]
                                                   
+
                                               
{'eval_loss': '4.384', 'eval_runtime': '3.234', 'eval_samples_per_second': '754.8', 'eval_steps_per_second': '12.06', 'epoch': '56'}
+
 28%|██▊       | 2016/7308 [24:59<47:08,  1.87it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.22it/s][eval] epoch 56.0 step 2016 loss 4.3844 (best 4.1514 @ ep 34.0, 11/20 since best)
+
+
                                               
 28%|██▊       | 2017/7308 [25:00<2:17:44,  1.56s/it]
 28%|██▊       | 2018/7308 [25:00<1:53:42,  1.29s/it]
 28%|██▊       | 2019/7308 [25:01<1:36:54,  1.10s/it]
 28%|██▊       | 2020/7308 [25:02<1:25:10,  1.03it/s]
                                                     
{'loss': '3.448', 'grad_norm': '0.4527', 'learning_rate': '0.0006', 'epoch': '56.11'}
+
 28%|██▊       | 2020/7308 [25:02<1:25:10,  1.03it/s]
 28%|██▊       | 2021/7308 [25:02<1:16:58,  1.14it/s]
 28%|██▊       | 2022/7308 [25:03<1:11:11,  1.24it/s]
 28%|██▊       | 2023/7308 [25:04<1:07:10,  1.31it/s]
 28%|██▊       | 2024/7308 [25:04<1:04:20,  1.37it/s]
 28%|██▊       | 2025/7308 [25:05<1:02:23,  1.41it/s]
 28%|██▊       | 2026/7308 [25:06<1:00:59,  1.44it/s]
 28%|██▊       | 2027/7308 [25:06<59:59,  1.47it/s]  
 28%|██▊       | 2028/7308 [25:07<59:18,  1.48it/s]
 28%|██▊       | 2029/7308 [25:08<58:48,  1.50it/s]
 28%|██▊       | 2030/7308 [25:08<58:30,  1.50it/s]
                                                   
{'loss': '3.395', 'grad_norm': '0.4529', 'learning_rate': '0.0006', 'epoch': '56.4'}
+
 28%|██▊       | 2030/7308 [25:08<58:30,  1.50it/s]
 28%|██▊       | 2031/7308 [25:09<58:16,  1.51it/s]
 28%|██▊       | 2032/7308 [25:10<58:05,  1.51it/s]
 28%|██▊       | 2033/7308 [25:10<57:55,  1.52it/s]
 28%|██▊       | 2034/7308 [25:11<57:50,  1.52it/s]
 28%|██▊       | 2035/7308 [25:12<57:48,  1.52it/s]
 28%|██▊       | 2036/7308 [25:12<57:44,  1.52it/s]
 28%|██▊       | 2037/7308 [25:13<57:42,  1.52it/s]
 28%|██▊       | 2038/7308 [25:13<57:41,  1.52it/s]
 28%|██▊       | 2039/7308 [25:14<57:40,  1.52it/s]
 28%|██▊       | 2040/7308 [25:15<57:39,  1.52it/s]
                                                   
{'loss': '3.432', 'grad_norm': '0.4218', 'learning_rate': '0.0006', 'epoch': '56.68'}
+
 28%|██▊       | 2040/7308 [25:15<57:39,  1.52it/s]
 28%|██▊       | 2041/7308 [25:15<57:42,  1.52it/s]
 28%|██▊       | 2042/7308 [25:16<57:39,  1.52it/s]
 28%|██▊       | 2043/7308 [25:17<57:38,  1.52it/s]
 28%|██▊       | 2044/7308 [25:17<57:35,  1.52it/s]
 28%|██▊       | 2045/7308 [25:18<57:34,  1.52it/s]
 28%|██▊       | 2046/7308 [25:19<57:34,  1.52it/s]
 28%|██▊       | 2047/7308 [25:19<57:33,  1.52it/s]
 28%|██▊       | 2048/7308 [25:20<57:31,  1.52it/s]
 28%|██▊       | 2049/7308 [25:21<57:29,  1.52it/s]
 28%|██▊       | 2050/7308 [25:21<57:30,  1.52it/s]
                                                   
{'loss': '3.471', 'grad_norm': '0.365', 'learning_rate': '0.0006', 'epoch': '56.96'}
+
 28%|██▊       | 2050/7308 [25:21<57:30,  1.52it/s]
 28%|██▊       | 2051/7308 [25:22<57:28,  1.52it/s]
 28%|██▊       | 2052/7308 [25:22<46:50,  1.87it/s]
 28%|██▊       | 2053/7308 [25:23<51:54,  1.69it/s]
 28%|██▊       | 2054/7308 [25:24<53:35,  1.63it/s]
 28%|██▊       | 2055/7308 [25:24<54:45,  1.60it/s]
 28%|██▊       | 2056/7308 [25:25<55:38,  1.57it/s]
 28%|██▊       | 2057/7308 [25:26<56:07,  1.56it/s]
 28%|██▊       | 2058/7308 [25:26<56:30,  1.55it/s]
 28%|██▊       | 2059/7308 [25:27<56:45,  1.54it/s]
 28%|██▊       | 2060/7308 [25:28<56:56,  1.54it/s]
                                                   
{'loss': '3.377', 'grad_norm': '0.4726', 'learning_rate': '0.0006', 'epoch': '57.23'}
+
 28%|██▊       | 2060/7308 [25:28<56:56,  1.54it/s]
 28%|██▊       | 2061/7308 [25:28<57:04,  1.53it/s]
 28%|██▊       | 2062/7308 [25:29<57:08,  1.53it/s]
 28%|██▊       | 2063/7308 [25:30<57:11,  1.53it/s]
 28%|██▊       | 2064/7308 [25:30<57:13,  1.53it/s]
 28%|██▊       | 2065/7308 [25:31<57:15,  1.53it/s]
 28%|██▊       | 2066/7308 [25:32<57:16,  1.53it/s]
 28%|██▊       | 2067/7308 [25:32<57:15,  1.53it/s]
 28%|██▊       | 2068/7308 [25:33<57:15,  1.53it/s]
 28%|██▊       | 2069/7308 [25:33<57:15,  1.53it/s]
 28%|██▊       | 2070/7308 [25:34<57:15,  1.52it/s]
                                                   
{'loss': '3.407', 'grad_norm': '0.4546', 'learning_rate': '0.0006', 'epoch': '57.51'}
+
 28%|██▊       | 2070/7308 [25:34<57:15,  1.52it/s]
 28%|██▊       | 2071/7308 [25:35<57:15,  1.52it/s]
 28%|██▊       | 2072/7308 [25:35<57:15,  1.52it/s]
 28%|██▊       | 2073/7308 [25:36<57:12,  1.53it/s]
 28%|██▊       | 2074/7308 [25:37<57:12,  1.52it/s]
 28%|██▊       | 2075/7308 [25:37<57:12,  1.52it/s]
 28%|██▊       | 2076/7308 [25:38<57:11,  1.52it/s]
 28%|██▊       | 2077/7308 [25:39<57:10,  1.52it/s]
 28%|██▊       | 2078/7308 [25:39<57:08,  1.53it/s]
 28%|██▊       | 2079/7308 [25:40<57:09,  1.52it/s]
 28%|██▊       | 2080/7308 [25:41<57:08,  1.52it/s]
                                                   
{'loss': '3.428', 'grad_norm': '0.3961', 'learning_rate': '0.0006', 'epoch': '57.79'}
+
 28%|██▊       | 2080/7308 [25:41<57:08,  1.52it/s]
 28%|██▊       | 2081/7308 [25:41<57:08,  1.52it/s]
 28%|██▊       | 2082/7308 [25:42<57:07,  1.52it/s]
 29%|██▊       | 2083/7308 [25:43<57:04,  1.53it/s]
 29%|██▊       | 2084/7308 [25:43<57:06,  1.52it/s]
 29%|██▊       | 2085/7308 [25:44<57:05,  1.52it/s]
 29%|██▊       | 2086/7308 [25:45<57:05,  1.52it/s]
 29%|██▊       | 2087/7308 [25:45<57:02,  1.53it/s]
 29%|██▊       | 2088/7308 [25:46<46:30,  1.87it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 19.03it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.91it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.71it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.09it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.80it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.58it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.48it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.38it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.36it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.30it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.30it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.28it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.27it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.26it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.26it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.25it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.25it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.25it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.25it/s]
                                                   
+
                                               
{'eval_loss': '4.411', 'eval_runtime': '3.233', 'eval_samples_per_second': '755', 'eval_steps_per_second': '12.06', 'epoch': '58'}
+
 29%|██▊       | 2088/7308 [25:49<46:30,  1.87it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.25it/s][eval] epoch 58.0 step 2088 loss 4.4114 (best 4.1514 @ ep 34.0, 12/20 since best)
+
+
                                               
 29%|██▊       | 2089/7308 [25:50<2:15:47,  1.56s/it]
 29%|██▊       | 2090/7308 [25:50<1:52:07,  1.29s/it]
                                                     
{'loss': '3.431', 'grad_norm': '0.4161', 'learning_rate': '0.0006', 'epoch': '58.06'}
+
 29%|██▊       | 2090/7308 [25:50<1:52:07,  1.29s/it]
 29%|██▊       | 2091/7308 [25:51<1:35:34,  1.10s/it]
 29%|██▊       | 2092/7308 [25:51<1:23:59,  1.04it/s]
 29%|██▊       | 2093/7308 [25:52<1:15:52,  1.15it/s]
 29%|██▊       | 2094/7308 [25:53<1:10:12,  1.24it/s]
 29%|██▊       | 2095/7308 [25:53<1:06:14,  1.31it/s]
 29%|██▊       | 2096/7308 [25:54<1:03:26,  1.37it/s]
 29%|██▊       | 2097/7308 [25:55<1:01:30,  1.41it/s]
 29%|██▊       | 2098/7308 [25:55<1:00:09,  1.44it/s]
 29%|██▊       | 2099/7308 [25:56<59:09,  1.47it/s]  
 29%|██▊       | 2100/7308 [25:57<58:26,  1.49it/s]
                                                   
{'loss': '3.357', 'grad_norm': '0.5581', 'learning_rate': '0.0006', 'epoch': '58.34'}
+
 29%|██▊       | 2100/7308 [25:57<58:26,  1.49it/s]
 29%|██▊       | 2101/7308 [25:57<58:01,  1.50it/s]
 29%|██▉       | 2102/7308 [25:58<57:41,  1.50it/s]
 29%|██▉       | 2103/7308 [25:59<57:29,  1.51it/s]
 29%|██▉       | 2104/7308 [25:59<57:22,  1.51it/s]
 29%|██▉       | 2105/7308 [26:00<57:12,  1.52it/s]
 29%|██▉       | 2106/7308 [26:01<57:04,  1.52it/s]
 29%|██▉       | 2107/7308 [26:01<57:01,  1.52it/s]
 29%|██▉       | 2108/7308 [26:02<56:59,  1.52it/s]
 29%|██▉       | 2109/7308 [26:03<56:56,  1.52it/s]
 29%|██▉       | 2110/7308 [26:03<56:53,  1.52it/s]
                                                   
{'loss': '3.414', 'grad_norm': '0.4237', 'learning_rate': '0.0006', 'epoch': '58.62'}
+
 29%|██▉       | 2110/7308 [26:03<56:53,  1.52it/s]
 29%|██▉       | 2111/7308 [26:04<56:51,  1.52it/s]
 29%|██▉       | 2112/7308 [26:05<56:48,  1.52it/s]
 29%|██▉       | 2113/7308 [26:05<56:50,  1.52it/s]
 29%|██▉       | 2114/7308 [26:06<56:51,  1.52it/s]
 29%|██▉       | 2115/7308 [26:07<56:48,  1.52it/s]
 29%|██▉       | 2116/7308 [26:07<56:43,  1.53it/s]
 29%|██▉       | 2117/7308 [26:08<56:40,  1.53it/s]
 29%|██▉       | 2118/7308 [26:09<56:41,  1.53it/s]
 29%|██▉       | 2119/7308 [26:09<56:41,  1.53it/s]
 29%|██▉       | 2120/7308 [26:10<56:38,  1.53it/s]
                                                   
{'loss': '3.43', 'grad_norm': '0.3515', 'learning_rate': '0.0006', 'epoch': '58.9'}
+
 29%|██▉       | 2120/7308 [26:10<56:38,  1.53it/s]
 29%|██▉       | 2121/7308 [26:10<56:38,  1.53it/s]
 29%|██▉       | 2122/7308 [26:11<56:40,  1.53it/s]
 29%|██▉       | 2123/7308 [26:12<56:38,  1.53it/s]
 29%|██▉       | 2124/7308 [26:12<46:08,  1.87it/s]
 29%|██▉       | 2125/7308 [26:13<51:49,  1.67it/s]
 29%|██▉       | 2126/7308 [26:13<53:14,  1.62it/s]
 29%|██▉       | 2127/7308 [26:14<54:16,  1.59it/s]
 29%|██▉       | 2128/7308 [26:15<54:57,  1.57it/s]
 29%|██▉       | 2129/7308 [26:15<55:26,  1.56it/s]
 29%|██▉       | 2130/7308 [26:16<55:51,  1.55it/s]
                                                   
{'loss': '3.369', 'grad_norm': '0.4208', 'learning_rate': '0.0006', 'epoch': '59.17'}
+
 29%|██▉       | 2130/7308 [26:16<55:51,  1.55it/s]
 29%|██▉       | 2131/7308 [26:17<56:05,  1.54it/s]
 29%|██▉       | 2132/7308 [26:17<56:14,  1.53it/s]
 29%|██▉       | 2133/7308 [26:18<56:18,  1.53it/s]
 29%|██▉       | 2134/7308 [26:19<56:20,  1.53it/s]
 29%|██▉       | 2135/7308 [26:19<56:23,  1.53it/s]
 29%|██▉       | 2136/7308 [26:20<56:24,  1.53it/s]
 29%|██▉       | 2137/7308 [26:21<56:29,  1.53it/s]
 29%|██▉       | 2138/7308 [26:21<56:28,  1.53it/s]
 29%|██▉       | 2139/7308 [26:22<56:25,  1.53it/s]
 29%|██▉       | 2140/7308 [26:23<56:28,  1.53it/s]
                                                   
{'loss': '3.352', 'grad_norm': '0.4645', 'learning_rate': '0.0006', 'epoch': '59.45'}
+
 29%|██▉       | 2140/7308 [26:23<56:28,  1.53it/s]
 29%|██▉       | 2141/7308 [26:23<56:28,  1.52it/s]
 29%|██▉       | 2142/7308 [26:24<56:27,  1.53it/s]
 29%|██▉       | 2143/7308 [26:25<56:26,  1.53it/s]
 29%|██▉       | 2144/7308 [26:25<56:26,  1.53it/s]
 29%|██▉       | 2145/7308 [26:26<56:27,  1.52it/s]
 29%|██▉       | 2146/7308 [26:27<56:27,  1.52it/s]
 29%|██▉       | 2147/7308 [26:27<56:25,  1.52it/s]
 29%|██▉       | 2148/7308 [26:28<56:24,  1.52it/s]
 29%|██▉       | 2149/7308 [26:29<56:21,  1.53it/s]
 29%|██▉       | 2150/7308 [26:29<56:21,  1.53it/s]
                                                   
{'loss': '3.387', 'grad_norm': '0.394', 'learning_rate': '0.0006', 'epoch': '59.73'}
+
 29%|██▉       | 2150/7308 [26:29<56:21,  1.53it/s]
 29%|██▉       | 2151/7308 [26:30<56:23,  1.52it/s]
 29%|██▉       | 2152/7308 [26:31<56:21,  1.52it/s]
 29%|██▉       | 2153/7308 [26:31<56:20,  1.53it/s]
 29%|██▉       | 2154/7308 [26:32<56:20,  1.52it/s]
 29%|██▉       | 2155/7308 [26:32<56:18,  1.53it/s]
 30%|██▉       | 2156/7308 [26:33<56:18,  1.53it/s]
 30%|██▉       | 2157/7308 [26:34<56:18,  1.52it/s]
 30%|██▉       | 2158/7308 [26:35<1:05:28,  1.31it/s]
 30%|██▉       | 2159/7308 [26:35<1:02:40,  1.37it/s]
 30%|██▉       | 2160/7308 [26:36<50:19,  1.71it/s]  
                                                   
{'loss': '3.419', 'grad_norm': '0.4583', 'learning_rate': '0.0006', 'epoch': '60'}
+
 30%|██▉       | 2160/7308 [26:36<50:19,  1.71it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.55it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.77it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.60it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.04it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.75it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.57it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.45it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.37it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.33it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.29it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.29it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.27it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.26it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.25it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.25it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.26it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.25it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.25it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s]
                                                   
+
                                               
{'eval_loss': '4.45', 'eval_runtime': '3.238', 'eval_samples_per_second': '753.9', 'eval_steps_per_second': '12.05', 'epoch': '60'}
+
 30%|██▉       | 2160/7308 [26:39<50:19,  1.71it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s][eval] epoch 60.0 step 2160 loss 4.4498 (best 4.1514 @ ep 34.0, 13/20 since best)
+
+
                                               
 30%|██▉       | 2161/7308 [26:40<2:18:00,  1.61s/it]
 30%|██▉       | 2162/7308 [26:40<1:53:27,  1.32s/it]
 30%|██▉       | 2163/7308 [26:41<1:36:15,  1.12s/it]
 30%|██▉       | 2164/7308 [26:42<1:24:12,  1.02it/s]
 30%|██▉       | 2165/7308 [26:42<1:15:48,  1.13it/s]
 30%|██▉       | 2166/7308 [26:43<1:09:55,  1.23it/s]
 30%|██▉       | 2167/7308 [26:44<1:05:51,  1.30it/s]
 30%|██▉       | 2168/7308 [26:44<1:02:56,  1.36it/s]
 30%|██▉       | 2169/7308 [26:45<1:00:53,  1.41it/s]
 30%|██▉       | 2170/7308 [26:46<59:25,  1.44it/s]  
                                                   
{'loss': '3.321', 'grad_norm': '0.5267', 'learning_rate': '0.0006', 'epoch': '60.28'}
+
 30%|██▉       | 2170/7308 [26:46<59:25,  1.44it/s]
 30%|██▉       | 2171/7308 [26:46<58:29,  1.46it/s]
 30%|██▉       | 2172/7308 [26:47<57:47,  1.48it/s]
 30%|██▉       | 2173/7308 [26:48<57:17,  1.49it/s]
 30%|██▉       | 2174/7308 [26:48<56:53,  1.50it/s]
 30%|██▉       | 2175/7308 [26:49<56:38,  1.51it/s]
 30%|██▉       | 2176/7308 [26:50<56:30,  1.51it/s]
 30%|██▉       | 2177/7308 [26:50<56:22,  1.52it/s]
 30%|██▉       | 2178/7308 [26:51<56:15,  1.52it/s]
 30%|██▉       | 2179/7308 [26:52<56:10,  1.52it/s]
 30%|██▉       | 2180/7308 [26:52<56:07,  1.52it/s]
                                                   
{'loss': '3.362', 'grad_norm': '0.4435', 'learning_rate': '0.0006', 'epoch': '60.57'}
+
 30%|██▉       | 2180/7308 [26:52<56:07,  1.52it/s]
 30%|██▉       | 2181/7308 [26:53<56:08,  1.52it/s]
 30%|██▉       | 2182/7308 [26:53<56:05,  1.52it/s]
 30%|██▉       | 2183/7308 [26:54<56:03,  1.52it/s]
 30%|██▉       | 2184/7308 [26:55<56:01,  1.52it/s]
 30%|██▉       | 2185/7308 [26:55<56:02,  1.52it/s]
 30%|██▉       | 2186/7308 [26:56<56:03,  1.52it/s]
 30%|██▉       | 2187/7308 [26:57<56:01,  1.52it/s]
 30%|██▉       | 2188/7308 [26:57<56:01,  1.52it/s]
 30%|██▉       | 2189/7308 [26:58<55:58,  1.52it/s]
 30%|██▉       | 2190/7308 [26:59<55:55,  1.53it/s]
                                                   
{'loss': '3.396', 'grad_norm': '0.4028', 'learning_rate': '0.0006', 'epoch': '60.85'}
+
 30%|██▉       | 2190/7308 [26:59<55:55,  1.53it/s]
 30%|██▉       | 2191/7308 [26:59<55:57,  1.52it/s]
 30%|██▉       | 2192/7308 [27:00<55:57,  1.52it/s]
 30%|███       | 2193/7308 [27:01<55:56,  1.52it/s]
 30%|███       | 2194/7308 [27:01<55:54,  1.52it/s]
 30%|███       | 2195/7308 [27:02<55:51,  1.53it/s]
 30%|███       | 2196/7308 [27:02<45:31,  1.87it/s]
 30%|███       | 2197/7308 [27:03<50:27,  1.69it/s]
 30%|███       | 2198/7308 [27:04<52:05,  1.63it/s]
 30%|███       | 2199/7308 [27:04<53:13,  1.60it/s]
 30%|███       | 2200/7308 [27:05<54:00,  1.58it/s]
                                                   
{'loss': '3.364', 'grad_norm': '0.4524', 'learning_rate': '0.0006', 'epoch': '61.11'}
+
 30%|███       | 2200/7308 [27:05<54:00,  1.58it/s]
 30%|███       | 2201/7308 [27:06<54:34,  1.56it/s]
 30%|███       | 2202/7308 [27:06<54:54,  1.55it/s]
 30%|███       | 2203/7308 [27:07<55:13,  1.54it/s]
 30%|███       | 2204/7308 [27:08<55:28,  1.53it/s]
 30%|███       | 2205/7308 [27:08<55:32,  1.53it/s]
 30%|███       | 2206/7308 [27:09<55:34,  1.53it/s]
 30%|███       | 2207/7308 [27:10<55:36,  1.53it/s]
 30%|███       | 2208/7308 [27:10<55:39,  1.53it/s]
 30%|███       | 2209/7308 [27:11<55:40,  1.53it/s]
 30%|███       | 2210/7308 [27:12<55:40,  1.53it/s]
                                                   
{'loss': '3.32', 'grad_norm': '0.5108', 'learning_rate': '0.0006', 'epoch': '61.4'}
+
 30%|███       | 2210/7308 [27:12<55:40,  1.53it/s]
 30%|███       | 2211/7308 [27:12<55:40,  1.53it/s]
 30%|███       | 2212/7308 [27:13<55:39,  1.53it/s]
 30%|███       | 2213/7308 [27:13<55:40,  1.53it/s]
 30%|███       | 2214/7308 [27:14<55:39,  1.53it/s]
 30%|███       | 2215/7308 [27:15<55:41,  1.52it/s]
 30%|███       | 2216/7308 [27:15<55:42,  1.52it/s]
 30%|███       | 2217/7308 [27:16<55:40,  1.52it/s]
 30%|███       | 2218/7308 [27:17<55:39,  1.52it/s]
 30%|███       | 2219/7308 [27:17<55:39,  1.52it/s]
 30%|███       | 2220/7308 [27:18<55:38,  1.52it/s]
                                                   
{'loss': '3.349', 'grad_norm': '0.4091', 'learning_rate': '0.0006', 'epoch': '61.68'}
+
 30%|███       | 2220/7308 [27:18<55:38,  1.52it/s]
 30%|███       | 2221/7308 [27:19<55:39,  1.52it/s]
 30%|███       | 2222/7308 [27:19<55:37,  1.52it/s]
 30%|███       | 2223/7308 [27:20<55:34,  1.52it/s]
 30%|███       | 2224/7308 [27:21<55:35,  1.52it/s]
 30%|███       | 2225/7308 [27:21<55:34,  1.52it/s]
 30%|███       | 2226/7308 [27:22<55:32,  1.52it/s]
 30%|███       | 2227/7308 [27:23<55:31,  1.53it/s]
 30%|███       | 2228/7308 [27:23<55:32,  1.52it/s]
 31%|███       | 2229/7308 [27:24<55:33,  1.52it/s]
 31%|███       | 2230/7308 [27:25<55:31,  1.52it/s]
                                                   
{'loss': '3.39', 'grad_norm': '0.3782', 'learning_rate': '0.0006', 'epoch': '61.96'}
+
 31%|███       | 2230/7308 [27:25<55:31,  1.52it/s]
 31%|███       | 2231/7308 [27:25<55:30,  1.52it/s]
 31%|███       | 2232/7308 [27:26<45:13,  1.87it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.80it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.87it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.67it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.08it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.79it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.59it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.48it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.40it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.35it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.31it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.31it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.28it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.27it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.27it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.25it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.26it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.25it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.26it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.25it/s]
                                                   
+
                                               
{'eval_loss': '4.468', 'eval_runtime': '3.233', 'eval_samples_per_second': '754.9', 'eval_steps_per_second': '12.06', 'epoch': '62'}
+
 31%|███       | 2232/7308 [27:29<45:13,  1.87it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.25it/s][eval] epoch 62.0 step 2232 loss 4.4679 (best 4.1514 @ ep 34.0, 14/20 since best)
+
+
                                               
 31%|███       | 2233/7308 [27:30<2:12:07,  1.56s/it]
 31%|███       | 2234/7308 [27:30<1:49:06,  1.29s/it]
 31%|███       | 2235/7308 [27:31<1:33:00,  1.10s/it]
 31%|███       | 2236/7308 [27:31<1:21:43,  1.03it/s]
 31%|███       | 2237/7308 [27:32<1:13:47,  1.15it/s]
 31%|███       | 2238/7308 [27:33<1:08:18,  1.24it/s]
 31%|███       | 2239/7308 [27:33<1:04:26,  1.31it/s]
 31%|███       | 2240/7308 [27:34<1:01:43,  1.37it/s]
                                                     
{'loss': '3.31', 'grad_norm': '0.4825', 'learning_rate': '0.0006', 'epoch': '62.23'}
+
 31%|███       | 2240/7308 [27:34<1:01:43,  1.37it/s]
 31%|███       | 2241/7308 [27:35<59:49,  1.41it/s]  
 31%|███       | 2242/7308 [27:35<58:27,  1.44it/s]
 31%|███       | 2243/7308 [27:36<57:34,  1.47it/s]
 31%|███       | 2244/7308 [27:37<56:56,  1.48it/s]
 31%|███       | 2245/7308 [27:37<56:29,  1.49it/s]
 31%|███       | 2246/7308 [27:38<1:04:47,  1.30it/s]
 31%|███       | 2247/7308 [27:39<1:01:57,  1.36it/s]
 31%|███       | 2248/7308 [27:40<59:56,  1.41it/s]  
 31%|███       | 2249/7308 [27:40<58:31,  1.44it/s]
 31%|███       | 2250/7308 [27:41<57:34,  1.46it/s]
                                                   
{'loss': '3.315', 'grad_norm': '0.4115', 'learning_rate': '0.0006', 'epoch': '62.51'}
+
 31%|███       | 2250/7308 [27:41<57:34,  1.46it/s]
 31%|███       | 2251/7308 [27:42<56:53,  1.48it/s]
 31%|███       | 2252/7308 [27:42<56:24,  1.49it/s]
 31%|███       | 2253/7308 [27:43<56:01,  1.50it/s]
 31%|███       | 2254/7308 [27:44<55:46,  1.51it/s]
 31%|███       | 2255/7308 [27:44<55:39,  1.51it/s]
 31%|███       | 2256/7308 [27:45<55:32,  1.52it/s]
 31%|███       | 2257/7308 [27:46<55:27,  1.52it/s]
 31%|███       | 2258/7308 [27:46<55:21,  1.52it/s]
 31%|███       | 2259/7308 [27:47<55:16,  1.52it/s]
 31%|███       | 2260/7308 [27:48<55:16,  1.52it/s]
                                                   
{'loss': '3.349', 'grad_norm': '0.3789', 'learning_rate': '0.0006', 'epoch': '62.79'}
+
 31%|███       | 2260/7308 [27:48<55:16,  1.52it/s]
 31%|███       | 2261/7308 [27:48<55:16,  1.52it/s]
 31%|███       | 2262/7308 [27:49<55:15,  1.52it/s]
 31%|███       | 2263/7308 [27:50<55:10,  1.52it/s]
 31%|███       | 2264/7308 [27:50<55:08,  1.52it/s]
 31%|███       | 2265/7308 [27:51<55:09,  1.52it/s]
 31%|███       | 2266/7308 [27:52<55:09,  1.52it/s]
 31%|███       | 2267/7308 [27:52<55:07,  1.52it/s]
 31%|███       | 2268/7308 [27:52<44:54,  1.87it/s]
 31%|███       | 2269/7308 [27:53<49:36,  1.69it/s]
 31%|███       | 2270/7308 [27:54<51:14,  1.64it/s]
                                                   
{'loss': '3.358', 'grad_norm': '0.5119', 'learning_rate': '0.0006', 'epoch': '63.06'}
+
 31%|███       | 2270/7308 [27:54<51:14,  1.64it/s]
 31%|███       | 2271/7308 [27:54<52:28,  1.60it/s]
 31%|███       | 2272/7308 [27:55<53:18,  1.57it/s]
 31%|███       | 2273/7308 [27:56<53:52,  1.56it/s]
 31%|███       | 2274/7308 [27:56<54:17,  1.55it/s]
 31%|███       | 2275/7308 [27:57<54:32,  1.54it/s]
 31%|███       | 2276/7308 [27:58<54:42,  1.53it/s]
 31%|███       | 2277/7308 [27:58<54:46,  1.53it/s]
 31%|███       | 2278/7308 [27:59<54:49,  1.53it/s]
 31%|███       | 2279/7308 [28:00<54:55,  1.53it/s]
 31%|███       | 2280/7308 [28:00<55:00,  1.52it/s]
                                                   
{'loss': '3.279', 'grad_norm': '0.5376', 'learning_rate': '0.0006', 'epoch': '63.34'}
+
 31%|███       | 2280/7308 [28:00<55:00,  1.52it/s]
 31%|███       | 2281/7308 [28:01<55:05,  1.52it/s]
 31%|███       | 2282/7308 [28:02<55:06,  1.52it/s]
 31%|███       | 2283/7308 [28:02<55:05,  1.52it/s]
 31%|███▏      | 2284/7308 [28:03<55:03,  1.52it/s]
 31%|███▏      | 2285/7308 [28:04<55:01,  1.52it/s]
 31%|███▏      | 2286/7308 [28:04<54:56,  1.52it/s]
 31%|███▏      | 2287/7308 [28:05<54:56,  1.52it/s]
 31%|███▏      | 2288/7308 [28:06<54:59,  1.52it/s]
 31%|███▏      | 2289/7308 [28:06<55:01,  1.52it/s]
 31%|███▏      | 2290/7308 [28:07<55:01,  1.52it/s]
                                                   
{'loss': '3.323', 'grad_norm': '0.4593', 'learning_rate': '0.0006', 'epoch': '63.62'}
+
 31%|███▏      | 2290/7308 [28:07<55:01,  1.52it/s]
 31%|███▏      | 2291/7308 [28:08<55:00,  1.52it/s]
 31%|███▏      | 2292/7308 [28:08<54:58,  1.52it/s]
 31%|███▏      | 2293/7308 [28:09<54:56,  1.52it/s]
 31%|███▏      | 2294/7308 [28:10<54:53,  1.52it/s]
 31%|███▏      | 2295/7308 [28:10<54:52,  1.52it/s]
 31%|███▏      | 2296/7308 [28:11<54:55,  1.52it/s]
 31%|███▏      | 2297/7308 [28:12<54:52,  1.52it/s]
 31%|███▏      | 2298/7308 [28:12<54:49,  1.52it/s]
 31%|███▏      | 2299/7308 [28:13<54:45,  1.52it/s]
 31%|███▏      | 2300/7308 [28:14<54:43,  1.53it/s]
                                                   
{'loss': '3.358', 'grad_norm': '0.4369', 'learning_rate': '0.0006', 'epoch': '63.9'}
+
 31%|███▏      | 2300/7308 [28:14<54:43,  1.53it/s]
 31%|███▏      | 2301/7308 [28:14<54:45,  1.52it/s]
 31%|███▏      | 2302/7308 [28:15<54:45,  1.52it/s]
 32%|███▏      | 2303/7308 [28:15<54:43,  1.52it/s]
 32%|███▏      | 2304/7308 [28:16<44:36,  1.87it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 19.03it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.95it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.69it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.10it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.78it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.59it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.46it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.38it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.33it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.31it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.29it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.28it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.27it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.27it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.24it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.24it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.24it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.25it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.25it/s]
                                                   
+
                                               
{'eval_loss': '4.501', 'eval_runtime': '3.235', 'eval_samples_per_second': '754.5', 'eval_steps_per_second': '12.05', 'epoch': '64'}
+
 32%|███▏      | 2304/7308 [28:19<44:36,  1.87it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.25it/s][eval] epoch 64.0 step 2304 loss 4.5005 (best 4.1514 @ ep 34.0, 15/20 since best)
+
+
                                               
 32%|███▏      | 2305/7308 [28:20<2:10:14,  1.56s/it]
 32%|███▏      | 2306/7308 [28:20<1:47:41,  1.29s/it]
 32%|███▏      | 2307/7308 [28:21<1:31:46,  1.10s/it]
 32%|███▏      | 2308/7308 [28:22<1:20:36,  1.03it/s]
 32%|███▏      | 2309/7308 [28:22<1:12:50,  1.14it/s]
 32%|███▏      | 2310/7308 [28:23<1:07:20,  1.24it/s]
                                                     
{'loss': '3.301', 'grad_norm': '0.5515', 'learning_rate': '0.0006', 'epoch': '64.17'}
+
 32%|███▏      | 2310/7308 [28:23<1:07:20,  1.24it/s]
 32%|███▏      | 2311/7308 [28:24<1:03:32,  1.31it/s]
 32%|███▏      | 2312/7308 [28:24<1:00:52,  1.37it/s]
 32%|███▏      | 2313/7308 [28:25<59:00,  1.41it/s]  
 32%|███▏      | 2314/7308 [28:26<57:39,  1.44it/s]
 32%|███▏      | 2315/7308 [28:26<56:42,  1.47it/s]
 32%|███▏      | 2316/7308 [28:27<56:05,  1.48it/s]
 32%|███▏      | 2317/7308 [28:28<55:39,  1.49it/s]
 32%|███▏      | 2318/7308 [28:28<55:21,  1.50it/s]
 32%|███▏      | 2319/7308 [28:29<55:06,  1.51it/s]
 32%|███▏      | 2320/7308 [28:30<54:54,  1.51it/s]
                                                   
{'loss': '3.29', 'grad_norm': '0.4395', 'learning_rate': '0.0006', 'epoch': '64.45'}
+
 32%|███▏      | 2320/7308 [28:30<54:54,  1.51it/s]
 32%|███▏      | 2321/7308 [28:30<54:48,  1.52it/s]
 32%|███▏      | 2322/7308 [28:31<54:49,  1.52it/s]
 32%|███▏      | 2323/7308 [28:32<54:43,  1.52it/s]
 32%|███▏      | 2324/7308 [28:32<54:41,  1.52it/s]
 32%|███▏      | 2325/7308 [28:33<54:41,  1.52it/s]
 32%|███▏      | 2326/7308 [28:33<54:39,  1.52it/s]
 32%|███▏      | 2327/7308 [28:34<54:38,  1.52it/s]
 32%|███▏      | 2328/7308 [28:35<54:35,  1.52it/s]
 32%|███▏      | 2329/7308 [28:35<54:33,  1.52it/s]
 32%|███▏      | 2330/7308 [28:36<54:32,  1.52it/s]
                                                   
{'loss': '3.318', 'grad_norm': '0.4523', 'learning_rate': '0.0006', 'epoch': '64.73'}
+
 32%|███▏      | 2330/7308 [28:36<54:32,  1.52it/s]
 32%|███▏      | 2331/7308 [28:37<54:30,  1.52it/s]
 32%|███▏      | 2332/7308 [28:37<54:31,  1.52it/s]
 32%|███▏      | 2333/7308 [28:38<54:31,  1.52it/s]
 32%|███▏      | 2334/7308 [28:39<54:32,  1.52it/s]
 32%|███▏      | 2335/7308 [28:39<54:30,  1.52it/s]
 32%|███▏      | 2336/7308 [28:40<1:03:00,  1.32it/s]
 32%|███▏      | 2337/7308 [28:41<1:00:24,  1.37it/s]
 32%|███▏      | 2338/7308 [28:42<58:36,  1.41it/s]  
 32%|███▏      | 2339/7308 [28:42<57:14,  1.45it/s]
 32%|███▏      | 2340/7308 [28:43<46:18,  1.79it/s]
                                                   
{'loss': '3.346', 'grad_norm': '0.517', 'learning_rate': '0.0006', 'epoch': '65'}
+
 32%|███▏      | 2340/7308 [28:43<46:18,  1.79it/s]
 32%|███▏      | 2341/7308 [28:43<50:30,  1.64it/s]
 32%|███▏      | 2342/7308 [28:44<51:41,  1.60it/s]
 32%|███▏      | 2343/7308 [28:45<52:30,  1.58it/s]
 32%|███▏      | 2344/7308 [28:45<53:03,  1.56it/s]
 32%|███▏      | 2345/7308 [28:46<53:27,  1.55it/s]
 32%|███▏      | 2346/7308 [28:47<53:41,  1.54it/s]
 32%|███▏      | 2347/7308 [28:48<1:02:14,  1.33it/s]
 32%|███▏      | 2348/7308 [28:48<59:49,  1.38it/s]  
 32%|███▏      | 2349/7308 [28:49<58:07,  1.42it/s]
 32%|███▏      | 2350/7308 [28:50<1:05:34,  1.26it/s]
                                                     
{'loss': '3.244', 'grad_norm': '0.4738', 'learning_rate': '0.0006', 'epoch': '65.28'}
+
 32%|███▏      | 2350/7308 [28:50<1:05:34,  1.26it/s]
 32%|███▏      | 2351/7308 [28:51<1:10:38,  1.17it/s]
 32%|███▏      | 2352/7308 [28:52<1:05:38,  1.26it/s]
 32%|███▏      | 2353/7308 [28:52<1:02:11,  1.33it/s]
 32%|███▏      | 2354/7308 [28:53<59:44,  1.38it/s]  
 32%|███▏      | 2355/7308 [28:54<58:07,  1.42it/s]
 32%|███▏      | 2356/7308 [28:54<56:59,  1.45it/s]
 32%|███▏      | 2357/7308 [28:55<56:10,  1.47it/s]
 32%|███▏      | 2358/7308 [28:56<55:34,  1.48it/s]
 32%|███▏      | 2359/7308 [28:56<55:07,  1.50it/s]
 32%|███▏      | 2360/7308 [28:57<54:46,  1.51it/s]
                                                   
{'loss': '3.273', 'grad_norm': '0.4361', 'learning_rate': '0.0006', 'epoch': '65.57'}
+
 32%|███▏      | 2360/7308 [28:57<54:46,  1.51it/s]
 32%|███▏      | 2361/7308 [28:58<54:35,  1.51it/s]
 32%|███▏      | 2362/7308 [28:58<54:29,  1.51it/s]
 32%|███▏      | 2363/7308 [28:59<54:24,  1.52it/s]
 32%|███▏      | 2364/7308 [28:59<54:20,  1.52it/s]
 32%|███▏      | 2365/7308 [29:00<54:16,  1.52it/s]
 32%|███▏      | 2366/7308 [29:01<54:12,  1.52it/s]
 32%|███▏      | 2367/7308 [29:01<54:07,  1.52it/s]
 32%|███▏      | 2368/7308 [29:02<54:05,  1.52it/s]
 32%|███▏      | 2369/7308 [29:03<54:06,  1.52it/s]
 32%|███▏      | 2370/7308 [29:03<54:06,  1.52it/s]
                                                   
{'loss': '3.32', 'grad_norm': '0.3724', 'learning_rate': '0.0006', 'epoch': '65.85'}
+
 32%|███▏      | 2370/7308 [29:03<54:06,  1.52it/s]
 32%|███▏      | 2371/7308 [29:04<54:06,  1.52it/s]
 32%|███▏      | 2372/7308 [29:05<54:04,  1.52it/s]
 32%|███▏      | 2373/7308 [29:05<54:00,  1.52it/s]
 32%|███▏      | 2374/7308 [29:06<54:00,  1.52it/s]
 32%|███▏      | 2375/7308 [29:07<53:58,  1.52it/s]
 33%|███▎      | 2376/7308 [29:07<43:58,  1.87it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.85it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.85it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.66it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.07it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.78it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.59it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.48it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.40it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.36it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.31it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.31it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.28it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.28it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.27it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.27it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.26it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.26it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.25it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.25it/s]
                                                   
+
                                               
{'eval_loss': '4.534', 'eval_runtime': '3.242', 'eval_samples_per_second': '753', 'eval_steps_per_second': '12.03', 'epoch': '66'}
+
 33%|███▎      | 2376/7308 [29:10<43:58,  1.87it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.25it/s][eval] epoch 66.0 step 2376 loss 4.5342 (best 4.1514 @ ep 34.0, 16/20 since best)
+
+
                                               
 33%|███▎      | 2377/7308 [29:11<2:16:43,  1.66s/it]
 33%|███▎      | 2378/7308 [29:12<1:51:49,  1.36s/it]
 33%|███▎      | 2379/7308 [29:13<1:34:29,  1.15s/it]
 33%|███▎      | 2380/7308 [29:13<1:22:19,  1.00s/it]
                                                     
{'loss': '3.298', 'grad_norm': '0.5008', 'learning_rate': '0.0006', 'epoch': '66.11'}
+
 33%|███▎      | 2380/7308 [29:13<1:22:19,  1.00s/it]
 33%|███▎      | 2381/7308 [29:14<1:13:50,  1.11it/s]
 33%|███▎      | 2382/7308 [29:15<1:07:51,  1.21it/s]
 33%|███▎      | 2383/7308 [29:15<1:03:39,  1.29it/s]
 33%|███▎      | 2384/7308 [29:16<1:09:06,  1.19it/s]
 33%|███▎      | 2385/7308 [29:17<1:04:29,  1.27it/s]
 33%|███▎      | 2386/7308 [29:18<1:01:18,  1.34it/s]
 33%|███▎      | 2387/7308 [29:18<59:06,  1.39it/s]  
 33%|███▎      | 2388/7308 [29:19<57:33,  1.42it/s]
 33%|███▎      | 2389/7308 [29:19<56:27,  1.45it/s]
 33%|███▎      | 2390/7308 [29:20<55:40,  1.47it/s]
                                                   
{'loss': '3.24', 'grad_norm': '0.4738', 'learning_rate': '0.0006', 'epoch': '66.4'}
+
 33%|███▎      | 2390/7308 [29:20<55:40,  1.47it/s]
 33%|███▎      | 2391/7308 [29:21<55:04,  1.49it/s]
 33%|███▎      | 2392/7308 [29:21<54:40,  1.50it/s]
 33%|███▎      | 2393/7308 [29:22<54:25,  1.50it/s]
 33%|███▎      | 2394/7308 [29:23<54:16,  1.51it/s]
 33%|███▎      | 2395/7308 [29:23<54:04,  1.51it/s]
 33%|███▎      | 2396/7308 [29:24<53:56,  1.52it/s]
 33%|███▎      | 2397/7308 [29:25<53:51,  1.52it/s]
 33%|███▎      | 2398/7308 [29:25<53:49,  1.52it/s]
 33%|███▎      | 2399/7308 [29:26<53:47,  1.52it/s]
 33%|███▎      | 2400/7308 [29:27<53:49,  1.52it/s]
                                                   
{'loss': '3.289', 'grad_norm': '0.4751', 'learning_rate': '0.0006', 'epoch': '66.68'}
+
 33%|███▎      | 2400/7308 [29:27<53:49,  1.52it/s]
 33%|███▎      | 2401/7308 [29:27<53:49,  1.52it/s]
 33%|███▎      | 2402/7308 [29:28<53:44,  1.52it/s]
 33%|███▎      | 2403/7308 [29:29<53:44,  1.52it/s]
 33%|███▎      | 2404/7308 [29:29<53:40,  1.52it/s]
 33%|███▎      | 2405/7308 [29:30<53:40,  1.52it/s]
 33%|███▎      | 2406/7308 [29:31<53:39,  1.52it/s]
 33%|███▎      | 2407/7308 [29:31<53:37,  1.52it/s]
 33%|███▎      | 2408/7308 [29:32<53:35,  1.52it/s]
 33%|███▎      | 2409/7308 [29:33<53:33,  1.52it/s]
 33%|███▎      | 2410/7308 [29:33<53:34,  1.52it/s]
                                                   
{'loss': '3.322', 'grad_norm': '0.417', 'learning_rate': '0.0006', 'epoch': '66.96'}
+
 33%|███▎      | 2410/7308 [29:33<53:34,  1.52it/s]
 33%|███▎      | 2411/7308 [29:34<53:33,  1.52it/s]
 33%|███▎      | 2412/7308 [29:34<43:38,  1.87it/s]
 33%|███▎      | 2413/7308 [29:35<48:22,  1.69it/s]
 33%|███▎      | 2414/7308 [29:36<49:54,  1.63it/s]
 33%|███▎      | 2415/7308 [29:36<50:59,  1.60it/s]
 33%|███▎      | 2416/7308 [29:37<51:41,  1.58it/s]
 33%|███▎      | 2417/7308 [29:38<52:14,  1.56it/s]
 33%|███▎      | 2418/7308 [29:38<52:36,  1.55it/s]
 33%|███▎      | 2419/7308 [29:39<52:52,  1.54it/s]
 33%|███▎      | 2420/7308 [29:40<53:00,  1.54it/s]
                                                   
{'loss': '3.242', 'grad_norm': '0.5211', 'learning_rate': '0.0006', 'epoch': '67.23'}
+
 33%|███▎      | 2420/7308 [29:40<53:00,  1.54it/s]
 33%|███▎      | 2421/7308 [29:40<53:06,  1.53it/s]
 33%|███▎      | 2422/7308 [29:41<53:12,  1.53it/s]
 33%|███▎      | 2423/7308 [29:41<53:15,  1.53it/s]
 33%|███▎      | 2424/7308 [29:42<53:17,  1.53it/s]
 33%|███▎      | 2425/7308 [29:43<53:16,  1.53it/s]
 33%|███▎      | 2426/7308 [29:43<53:18,  1.53it/s]
 33%|███▎      | 2427/7308 [29:44<53:20,  1.53it/s]
 33%|███▎      | 2428/7308 [29:45<53:21,  1.52it/s]
 33%|███▎      | 2429/7308 [29:45<53:19,  1.52it/s]
 33%|███▎      | 2430/7308 [29:46<53:19,  1.52it/s]
                                                   
{'loss': '3.258', 'grad_norm': '0.5921', 'learning_rate': '0.0006', 'epoch': '67.51'}
+
 33%|███▎      | 2430/7308 [29:46<53:19,  1.52it/s]
 33%|███▎      | 2431/7308 [29:47<53:18,  1.52it/s]
 33%|███▎      | 2432/7308 [29:47<53:19,  1.52it/s]
 33%|███▎      | 2433/7308 [29:48<53:18,  1.52it/s]
 33%|███▎      | 2434/7308 [29:49<53:16,  1.52it/s]
 33%|███▎      | 2435/7308 [29:49<53:15,  1.52it/s]
 33%|███▎      | 2436/7308 [29:50<53:15,  1.52it/s]
 33%|███▎      | 2437/7308 [29:51<53:15,  1.52it/s]
 33%|███▎      | 2438/7308 [29:51<53:15,  1.52it/s]
 33%|███▎      | 2439/7308 [29:52<53:14,  1.52it/s]
 33%|███▎      | 2440/7308 [29:53<53:13,  1.52it/s]
                                                   
{'loss': '3.295', 'grad_norm': '0.3996', 'learning_rate': '0.0006', 'epoch': '67.79'}
+
 33%|███▎      | 2440/7308 [29:53<53:13,  1.52it/s]
 33%|███▎      | 2441/7308 [29:53<53:13,  1.52it/s]
 33%|███▎      | 2442/7308 [29:54<53:12,  1.52it/s]
 33%|███▎      | 2443/7308 [29:55<53:11,  1.52it/s]
 33%|███▎      | 2444/7308 [29:55<53:10,  1.52it/s]
 33%|███▎      | 2445/7308 [29:56<53:09,  1.52it/s]
 33%|███▎      | 2446/7308 [29:57<53:08,  1.53it/s]
 33%|███▎      | 2447/7308 [29:57<53:04,  1.53it/s]
 33%|███▎      | 2448/7308 [29:57<43:15,  1.87it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.99it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.91it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.71it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.13it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.80it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.59it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.48it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.39it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.34it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.30it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.31it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.28it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.27it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.26it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.25it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.26it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.25it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.25it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s]
                                                   
+
                                               
{'eval_loss': '4.564', 'eval_runtime': '3.24', 'eval_samples_per_second': '753.3', 'eval_steps_per_second': '12.04', 'epoch': '68'}
+
 33%|███▎      | 2448/7308 [30:01<43:15,  1.87it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.24it/s][eval] epoch 68.0 step 2448 loss 4.5642 (best 4.1514 @ ep 34.0, 17/20 since best)
+
+
                                               
 34%|███▎      | 2449/7308 [30:01<2:06:42,  1.56s/it]
 34%|███▎      | 2450/7308 [30:02<1:44:37,  1.29s/it]
                                                     
{'loss': '3.283', 'grad_norm': '0.4472', 'learning_rate': '0.0006', 'epoch': '68.06'}
+
 34%|███▎      | 2450/7308 [30:02<1:44:37,  1.29s/it]
 34%|███▎      | 2451/7308 [30:03<1:29:08,  1.10s/it]
 34%|███▎      | 2452/7308 [30:03<1:18:18,  1.03it/s]
 34%|███▎      | 2453/7308 [30:04<1:10:42,  1.14it/s]
 34%|███▎      | 2454/7308 [30:05<1:05:24,  1.24it/s]
 34%|███▎      | 2455/7308 [30:05<1:01:40,  1.31it/s]
 34%|███▎      | 2456/7308 [30:06<59:05,  1.37it/s]  
 34%|███▎      | 2457/7308 [30:07<57:15,  1.41it/s]
 34%|███▎      | 2458/7308 [30:07<55:59,  1.44it/s]
 34%|███▎      | 2459/7308 [30:08<55:05,  1.47it/s]
 34%|███▎      | 2460/7308 [30:09<54:27,  1.48it/s]
                                                   
{'loss': '3.197', 'grad_norm': '0.511', 'learning_rate': '0.0006', 'epoch': '68.34'}
+
 34%|███▎      | 2460/7308 [30:09<54:27,  1.48it/s]
 34%|███▎      | 2461/7308 [30:09<54:01,  1.50it/s]
 34%|███▎      | 2462/7308 [30:10<53:42,  1.50it/s]
 34%|███▎      | 2463/7308 [30:11<53:28,  1.51it/s]
 34%|███▎      | 2464/7308 [30:11<53:19,  1.51it/s]
 34%|███▎      | 2465/7308 [30:12<53:12,  1.52it/s]
 34%|███▎      | 2466/7308 [30:13<53:07,  1.52it/s]
 34%|███▍      | 2467/7308 [30:13<53:02,  1.52it/s]
 34%|███▍      | 2468/7308 [30:14<52:58,  1.52it/s]
 34%|███▍      | 2469/7308 [30:15<52:58,  1.52it/s]
 34%|███▍      | 2470/7308 [30:15<52:56,  1.52it/s]
                                                   
{'loss': '3.257', 'grad_norm': '0.5156', 'learning_rate': '0.0006', 'epoch': '68.62'}
+
 34%|███▍      | 2470/7308 [30:15<52:56,  1.52it/s]
 34%|███▍      | 2471/7308 [30:16<52:55,  1.52it/s]
 34%|███▍      | 2472/7308 [30:17<52:53,  1.52it/s]
 34%|███▍      | 2473/7308 [30:17<52:52,  1.52it/s]
 34%|███▍      | 2474/7308 [30:18<52:51,  1.52it/s]
 34%|███▍      | 2475/7308 [30:18<52:49,  1.52it/s]
 34%|███▍      | 2476/7308 [30:19<52:49,  1.52it/s]
 34%|███▍      | 2477/7308 [30:20<52:47,  1.53it/s]
 34%|███▍      | 2478/7308 [30:20<52:48,  1.52it/s]
 34%|███▍      | 2479/7308 [30:21<52:48,  1.52it/s]
 34%|███▍      | 2480/7308 [30:22<52:47,  1.52it/s]
                                                   
{'loss': '3.284', 'grad_norm': '0.4244', 'learning_rate': '0.0006', 'epoch': '68.9'}
+
 34%|███▍      | 2480/7308 [30:22<52:47,  1.52it/s]
 34%|███▍      | 2481/7308 [30:22<52:47,  1.52it/s]
 34%|███▍      | 2482/7308 [30:23<52:49,  1.52it/s]
 34%|███▍      | 2483/7308 [30:24<52:52,  1.52it/s]
 34%|███▍      | 2484/7308 [30:24<43:04,  1.87it/s]
 34%|███▍      | 2485/7308 [30:25<47:36,  1.69it/s]
 34%|███▍      | 2486/7308 [30:25<49:08,  1.64it/s]
 34%|███▍      | 2487/7308 [30:26<50:13,  1.60it/s]
 34%|███▍      | 2488/7308 [30:27<50:58,  1.58it/s]
 34%|███▍      | 2489/7308 [30:27<51:28,  1.56it/s]
 34%|███▍      | 2490/7308 [30:28<51:48,  1.55it/s]
                                                   
{'loss': '3.234', 'grad_norm': '0.5264', 'learning_rate': '0.0006', 'epoch': '69.17'}
+
 34%|███▍      | 2490/7308 [30:28<51:48,  1.55it/s]
 34%|███▍      | 2491/7308 [30:29<52:04,  1.54it/s]
 34%|███▍      | 2492/7308 [30:29<52:15,  1.54it/s]
 34%|███▍      | 2493/7308 [30:30<52:24,  1.53it/s]
 34%|███▍      | 2494/7308 [30:31<52:27,  1.53it/s]
 34%|███▍      | 2495/7308 [30:31<52:29,  1.53it/s]
 34%|███▍      | 2496/7308 [30:32<52:30,  1.53it/s]
 34%|███▍      | 2497/7308 [30:33<52:30,  1.53it/s]
 34%|███▍      | 2498/7308 [30:33<52:31,  1.53it/s]
 34%|███▍      | 2499/7308 [30:34<52:32,  1.53it/s]
 34%|███▍      | 2500/7308 [30:35<52:30,  1.53it/s]
                                                   
{'loss': '3.206', 'grad_norm': '0.4841', 'learning_rate': '0.0006', 'epoch': '69.45'}
+
 34%|███▍      | 2500/7308 [30:35<52:30,  1.53it/s]
 34%|███▍      | 2501/7308 [30:35<52:32,  1.52it/s]
 34%|███▍      | 2502/7308 [30:36<52:32,  1.52it/s]
 34%|███▍      | 2503/7308 [30:37<52:33,  1.52it/s]
 34%|███▍      | 2504/7308 [30:37<52:31,  1.52it/s]
 34%|███▍      | 2505/7308 [30:38<52:29,  1.53it/s]
 34%|███▍      | 2506/7308 [30:38<52:27,  1.53it/s]
 34%|███▍      | 2507/7308 [30:39<52:28,  1.53it/s]
 34%|███▍      | 2508/7308 [30:40<52:27,  1.53it/s]
 34%|███▍      | 2509/7308 [30:40<52:28,  1.52it/s]
 34%|███▍      | 2510/7308 [30:41<52:25,  1.53it/s]
                                                   
{'loss': '3.249', 'grad_norm': '0.406', 'learning_rate': '0.0006', 'epoch': '69.73'}
+
 34%|███▍      | 2510/7308 [30:41<52:25,  1.53it/s]
 34%|███▍      | 2511/7308 [30:42<52:23,  1.53it/s]
 34%|███▍      | 2512/7308 [30:42<52:26,  1.52it/s]
 34%|███▍      | 2513/7308 [30:43<52:24,  1.52it/s]
 34%|███▍      | 2514/7308 [30:44<52:23,  1.53it/s]
 34%|███▍      | 2515/7308 [30:44<52:22,  1.53it/s]
 34%|███▍      | 2516/7308 [30:45<52:22,  1.53it/s]
 34%|███▍      | 2517/7308 [30:46<52:23,  1.52it/s]
 34%|███▍      | 2518/7308 [30:46<52:23,  1.52it/s]
 34%|███▍      | 2519/7308 [30:47<52:18,  1.53it/s]
 34%|███▍      | 2520/7308 [30:47<42:37,  1.87it/s]
                                                   
{'loss': '3.287', 'grad_norm': '0.4944', 'learning_rate': '0.0006', 'epoch': '70'}
+
 34%|███▍      | 2520/7308 [30:47<42:37,  1.87it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.78it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.85it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.66it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.10it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.79it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.59it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.48it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.40it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.35it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.31it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.30it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.29it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.28it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.27it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.26it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.24it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.24it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.23it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s]
                                                   
+
                                               
{'eval_loss': '4.597', 'eval_runtime': '3.244', 'eval_samples_per_second': '752.6', 'eval_steps_per_second': '12.02', 'epoch': '70'}
+
 34%|███▍      | 2520/7308 [30:51<42:37,  1.87it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s][eval] epoch 70.0 step 2520 loss 4.5970 (best 4.1514 @ ep 34.0, 18/20 since best)
+
+
                                               
 34%|███▍      | 2521/7308 [30:51<2:05:02,  1.57s/it]
 35%|███▍      | 2522/7308 [30:52<1:43:13,  1.29s/it]
 35%|███▍      | 2523/7308 [30:53<1:27:54,  1.10s/it]
 35%|███▍      | 2524/7308 [30:53<1:17:10,  1.03it/s]
 35%|███▍      | 2525/7308 [30:54<1:09:41,  1.14it/s]
 35%|███▍      | 2526/7308 [30:55<1:04:26,  1.24it/s]
 35%|███▍      | 2527/7308 [30:55<1:00:46,  1.31it/s]
 35%|███▍      | 2528/7308 [30:56<58:12,  1.37it/s]  
 35%|███▍      | 2529/7308 [30:56<56:25,  1.41it/s]
 35%|███▍      | 2530/7308 [30:57<55:09,  1.44it/s]
                                                   
{'loss': '3.181', 'grad_norm': '0.562', 'learning_rate': '0.0006', 'epoch': '70.28'}
+
 35%|███▍      | 2530/7308 [30:57<55:09,  1.44it/s]
 35%|███▍      | 2531/7308 [30:58<54:16,  1.47it/s]
 35%|███▍      | 2532/7308 [30:58<53:35,  1.49it/s]
 35%|███▍      | 2533/7308 [30:59<53:08,  1.50it/s]
 35%|███▍      | 2534/7308 [31:00<52:52,  1.51it/s]
 35%|███▍      | 2535/7308 [31:00<52:39,  1.51it/s]
 35%|███▍      | 2536/7308 [31:01<52:28,  1.52it/s]
 35%|███▍      | 2537/7308 [31:02<52:23,  1.52it/s]
 35%|███▍      | 2538/7308 [31:02<52:15,  1.52it/s]
 35%|███▍      | 2539/7308 [31:03<52:12,  1.52it/s]
 35%|███▍      | 2540/7308 [31:04<52:09,  1.52it/s]
                                                   
{'loss': '3.221', 'grad_norm': '0.468', 'learning_rate': '0.0006', 'epoch': '70.57'}
+
 35%|███▍      | 2540/7308 [31:04<52:09,  1.52it/s]
 35%|███▍      | 2541/7308 [31:04<52:09,  1.52it/s]
 35%|███▍      | 2542/7308 [31:05<52:07,  1.52it/s]
 35%|███▍      | 2543/7308 [31:06<52:07,  1.52it/s]
 35%|███▍      | 2544/7308 [31:06<52:10,  1.52it/s]
 35%|███▍      | 2545/7308 [31:07<52:08,  1.52it/s]
 35%|███▍      | 2546/7308 [31:08<52:06,  1.52it/s]
 35%|███▍      | 2547/7308 [31:08<52:03,  1.52it/s]
 35%|███▍      | 2548/7308 [31:09<52:01,  1.53it/s]
 35%|███▍      | 2549/7308 [31:10<52:00,  1.52it/s]
 35%|███▍      | 2550/7308 [31:10<51:59,  1.53it/s]
                                                   
{'loss': '3.254', 'grad_norm': '0.4603', 'learning_rate': '0.0006', 'epoch': '70.85'}
+
 35%|███▍      | 2550/7308 [31:10<51:59,  1.53it/s]
 35%|███▍      | 2551/7308 [31:11<51:57,  1.53it/s]
 35%|███▍      | 2552/7308 [31:12<51:56,  1.53it/s]
 35%|███▍      | 2553/7308 [31:12<51:58,  1.52it/s]
 35%|███▍      | 2554/7308 [31:13<51:58,  1.52it/s]
 35%|███▍      | 2555/7308 [31:14<51:55,  1.53it/s]
 35%|███▍      | 2556/7308 [31:14<42:19,  1.87it/s]
 35%|███▍      | 2557/7308 [31:15<47:15,  1.68it/s]
 35%|███▌      | 2558/7308 [31:15<48:39,  1.63it/s]
 35%|███▌      | 2559/7308 [31:16<49:39,  1.59it/s]
 35%|███▌      | 2560/7308 [31:17<50:23,  1.57it/s]
                                                   
{'loss': '3.224', 'grad_norm': '0.6187', 'learning_rate': '0.0006', 'epoch': '71.11'}
+
 35%|███▌      | 2560/7308 [31:17<50:23,  1.57it/s]
 35%|███▌      | 2561/7308 [31:17<50:50,  1.56it/s]
 35%|███▌      | 2562/7308 [31:18<51:16,  1.54it/s]
 35%|███▌      | 2563/7308 [31:18<51:23,  1.54it/s]
 35%|███▌      | 2564/7308 [31:19<51:31,  1.53it/s]
 35%|███▌      | 2565/7308 [31:20<51:39,  1.53it/s]
 35%|███▌      | 2566/7308 [31:20<51:45,  1.53it/s]
 35%|███▌      | 2567/7308 [31:21<51:49,  1.52it/s]
 35%|███▌      | 2568/7308 [31:22<51:51,  1.52it/s]
 35%|███▌      | 2569/7308 [31:22<51:51,  1.52it/s]
 35%|███▌      | 2570/7308 [31:23<51:48,  1.52it/s]
                                                   
{'loss': '3.194', 'grad_norm': '0.5944', 'learning_rate': '0.0006', 'epoch': '71.4'}
+
 35%|███▌      | 2570/7308 [31:23<51:48,  1.52it/s]
 35%|███▌      | 2571/7308 [31:24<51:48,  1.52it/s]
 35%|███▌      | 2572/7308 [31:24<51:49,  1.52it/s]
 35%|███▌      | 2573/7308 [31:25<51:50,  1.52it/s]
 35%|███▌      | 2574/7308 [31:26<51:51,  1.52it/s]
 35%|███▌      | 2575/7308 [31:26<51:51,  1.52it/s]
 35%|███▌      | 2576/7308 [31:27<51:50,  1.52it/s]
 35%|███▌      | 2577/7308 [31:28<51:48,  1.52it/s]
 35%|███▌      | 2578/7308 [31:28<51:46,  1.52it/s]
 35%|███▌      | 2579/7308 [31:29<51:46,  1.52it/s]
 35%|███▌      | 2580/7308 [31:30<51:48,  1.52it/s]
                                                   
{'loss': '3.228', 'grad_norm': '0.4772', 'learning_rate': '0.0006', 'epoch': '71.68'}
+
 35%|███▌      | 2580/7308 [31:30<51:48,  1.52it/s]
 35%|███▌      | 2581/7308 [31:30<51:49,  1.52it/s]
 35%|███▌      | 2582/7308 [31:31<51:48,  1.52it/s]
 35%|███▌      | 2583/7308 [31:32<51:47,  1.52it/s]
 35%|███▌      | 2584/7308 [31:32<51:45,  1.52it/s]
 35%|███▌      | 2585/7308 [31:33<51:42,  1.52it/s]
 35%|███▌      | 2586/7308 [31:34<51:40,  1.52it/s]
 35%|███▌      | 2587/7308 [31:34<51:42,  1.52it/s]
 35%|███▌      | 2588/7308 [31:35<51:42,  1.52it/s]
 35%|███▌      | 2589/7308 [31:36<51:41,  1.52it/s]
 35%|███▌      | 2590/7308 [31:36<51:42,  1.52it/s]
                                                   
{'loss': '3.25', 'grad_norm': '0.4043', 'learning_rate': '0.0006', 'epoch': '71.96'}
+
 35%|███▌      | 2590/7308 [31:36<51:42,  1.52it/s]
 35%|███▌      | 2591/7308 [31:37<51:36,  1.52it/s]
 35%|███▌      | 2592/7308 [31:37<42:04,  1.87it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.70it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.89it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.66it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.10it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.77it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.58it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.45it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.39it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.32it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.31it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.27it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.27it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.26it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.25it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.25it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.25it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.25it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.25it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.25it/s]
                                                   
+
                                               
{'eval_loss': '4.636', 'eval_runtime': '3.257', 'eval_samples_per_second': '749.5', 'eval_steps_per_second': '11.97', 'epoch': '72'}
+
 35%|███▌      | 2592/7308 [31:40<42:04,  1.87it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.25it/s][eval] epoch 72.0 step 2592 loss 4.6362 (best 4.1514 @ ep 34.0, 19/20 since best)
+
+
                                               
 35%|███▌      | 2593/7308 [31:41<2:04:01,  1.58s/it]
 35%|███▌      | 2594/7308 [31:42<1:42:13,  1.30s/it]
 36%|███▌      | 2595/7308 [31:42<1:27:01,  1.11s/it]
 36%|███▌      | 2596/7308 [31:43<1:16:22,  1.03it/s]
 36%|███▌      | 2597/7308 [31:44<1:08:53,  1.14it/s]
 36%|███▌      | 2598/7308 [31:44<1:03:38,  1.23it/s]
 36%|███▌      | 2599/7308 [31:45<59:57,  1.31it/s]  
 36%|███▌      | 2600/7308 [31:46<57:25,  1.37it/s]
                                                   
{'loss': '3.157', 'grad_norm': '0.4899', 'learning_rate': '0.0006', 'epoch': '72.23'}
+
 36%|███▌      | 2600/7308 [31:46<57:25,  1.37it/s]
 36%|███▌      | 2601/7308 [31:46<55:39,  1.41it/s]
 36%|███▌      | 2602/7308 [31:47<54:23,  1.44it/s]
 36%|███▌      | 2603/7308 [31:48<53:31,  1.47it/s]
 36%|███▌      | 2604/7308 [31:48<52:50,  1.48it/s]
 36%|███▌      | 2605/7308 [31:49<52:25,  1.50it/s]
 36%|███▌      | 2606/7308 [31:50<52:07,  1.50it/s]
 36%|███▌      | 2607/7308 [31:50<51:55,  1.51it/s]
 36%|███▌      | 2608/7308 [31:51<51:44,  1.51it/s]
 36%|███▌      | 2609/7308 [31:52<51:35,  1.52it/s]
 36%|███▌      | 2610/7308 [31:52<51:30,  1.52it/s]
                                                   
{'loss': '3.182', 'grad_norm': '0.5409', 'learning_rate': '0.0006', 'epoch': '72.51'}
+
 36%|███▌      | 2610/7308 [31:52<51:30,  1.52it/s]
 36%|███▌      | 2611/7308 [31:53<51:29,  1.52it/s]
 36%|███▌      | 2612/7308 [31:54<51:26,  1.52it/s]
 36%|███▌      | 2613/7308 [31:54<51:22,  1.52it/s]
 36%|███▌      | 2614/7308 [31:55<51:19,  1.52it/s]
 36%|███▌      | 2615/7308 [31:56<51:19,  1.52it/s]
 36%|███▌      | 2616/7308 [31:56<51:20,  1.52it/s]
 36%|███▌      | 2617/7308 [31:57<51:18,  1.52it/s]
 36%|███▌      | 2618/7308 [31:58<51:16,  1.52it/s]
 36%|███▌      | 2619/7308 [31:58<51:14,  1.53it/s]
 36%|███▌      | 2620/7308 [31:59<51:14,  1.52it/s]
                                                   
{'loss': '3.232', 'grad_norm': '0.4399', 'learning_rate': '0.0006', 'epoch': '72.79'}
+
 36%|███▌      | 2620/7308 [31:59<51:14,  1.52it/s]
 36%|███▌      | 2621/7308 [32:00<51:15,  1.52it/s]
 36%|███▌      | 2622/7308 [32:00<51:13,  1.52it/s]
 36%|███▌      | 2623/7308 [32:01<51:11,  1.53it/s]
 36%|███▌      | 2624/7308 [32:01<51:10,  1.53it/s]
 36%|███▌      | 2625/7308 [32:02<51:10,  1.53it/s]
 36%|███▌      | 2626/7308 [32:03<51:12,  1.52it/s]
 36%|███▌      | 2627/7308 [32:03<51:09,  1.52it/s]
 36%|███▌      | 2628/7308 [32:04<41:41,  1.87it/s]
 36%|███▌      | 2629/7308 [32:04<46:51,  1.66it/s]
 36%|███▌      | 2630/7308 [32:05<48:06,  1.62it/s]
                                                   
{'loss': '3.226', 'grad_norm': '0.5044', 'learning_rate': '0.0006', 'epoch': '73.06'}
+
 36%|███▌      | 2630/7308 [32:05<48:06,  1.62it/s]
 36%|███▌      | 2631/7308 [32:06<49:01,  1.59it/s]
 36%|███▌      | 2632/7308 [32:06<49:37,  1.57it/s]
 36%|███▌      | 2633/7308 [32:07<50:01,  1.56it/s]
 36%|███▌      | 2634/7308 [32:08<50:21,  1.55it/s]
 36%|███▌      | 2635/7308 [32:08<50:35,  1.54it/s]
 36%|███▌      | 2636/7308 [32:09<50:43,  1.54it/s]
 36%|███▌      | 2637/7308 [32:10<50:48,  1.53it/s]
 36%|███▌      | 2638/7308 [32:10<50:53,  1.53it/s]
 36%|███▌      | 2639/7308 [32:11<50:56,  1.53it/s]
 36%|███▌      | 2640/7308 [32:12<50:56,  1.53it/s]
                                                   
{'loss': '3.148', 'grad_norm': '0.4615', 'learning_rate': '0.0006', 'epoch': '73.34'}
+
 36%|███▌      | 2640/7308 [32:12<50:56,  1.53it/s]
 36%|███▌      | 2641/7308 [32:12<50:58,  1.53it/s]
 36%|███▌      | 2642/7308 [32:13<50:57,  1.53it/s]
 36%|███▌      | 2643/7308 [32:14<50:57,  1.53it/s]
 36%|███▌      | 2644/7308 [32:14<50:59,  1.52it/s]
 36%|███▌      | 2645/7308 [32:15<51:01,  1.52it/s]
 36%|███▌      | 2646/7308 [32:16<51:00,  1.52it/s]
 36%|███▌      | 2647/7308 [32:16<51:02,  1.52it/s]
 36%|███▌      | 2648/7308 [32:17<50:59,  1.52it/s]
 36%|███▌      | 2649/7308 [32:18<50:56,  1.52it/s]
 36%|███▋      | 2650/7308 [32:18<50:56,  1.52it/s]
                                                   
{'loss': '3.189', 'grad_norm': '0.4378', 'learning_rate': '0.0006', 'epoch': '73.62'}
+
 36%|███▋      | 2650/7308 [32:18<50:56,  1.52it/s]
 36%|███▋      | 2651/7308 [32:19<50:56,  1.52it/s]
 36%|███▋      | 2652/7308 [32:20<50:55,  1.52it/s]
 36%|███▋      | 2653/7308 [32:20<50:53,  1.52it/s]
 36%|███▋      | 2654/7308 [32:21<50:50,  1.53it/s]
 36%|███▋      | 2655/7308 [32:22<50:51,  1.53it/s]
 36%|███▋      | 2656/7308 [32:22<50:49,  1.53it/s]
 36%|███▋      | 2657/7308 [32:23<50:50,  1.52it/s]
 36%|███▋      | 2658/7308 [32:23<50:49,  1.52it/s]
 36%|███▋      | 2659/7308 [32:24<50:49,  1.52it/s]
 36%|███▋      | 2660/7308 [32:25<50:47,  1.52it/s]
                                                   
{'loss': '3.224', 'grad_norm': '0.4416', 'learning_rate': '0.0006', 'epoch': '73.9'}
+
 36%|███▋      | 2660/7308 [32:25<50:47,  1.52it/s]
 36%|███▋      | 2661/7308 [32:25<50:47,  1.52it/s]
 36%|███▋      | 2662/7308 [32:26<50:47,  1.52it/s]
 36%|███▋      | 2663/7308 [32:27<50:45,  1.53it/s]
 36%|███▋      | 2664/7308 [32:27<41:21,  1.87it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 18.73it/s]
+
 13%|█▎        | 5/39 [00:00<00:02, 14.87it/s]
+
 18%|█▊        | 7/39 [00:00<00:02, 13.64it/s]
+
 23%|██▎       | 9/39 [00:00<00:02, 13.10it/s]
+
 28%|██▊       | 11/39 [00:00<00:02, 12.78it/s]
+
 33%|███▎      | 13/39 [00:00<00:02, 12.59it/s]
+
 38%|███▊      | 15/39 [00:01<00:01, 12.47it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 12.41it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 12.36it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 12.32it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 12.31it/s]
+
 64%|██████▍   | 25/39 [00:01<00:01, 12.29it/s]
+
 69%|██████▉   | 27/39 [00:02<00:00, 12.28it/s]
+
 74%|███████▍  | 29/39 [00:02<00:00, 12.28it/s]
+
 79%|███████▉  | 31/39 [00:02<00:00, 12.26it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 12.25it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 12.24it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 12.23it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s]
                                                   
+
                                               
{'eval_loss': '4.664', 'eval_runtime': '3.243', 'eval_samples_per_second': '752.7', 'eval_steps_per_second': '12.03', 'epoch': '74'}
+
 36%|███▋      | 2664/7308 [32:30<41:21,  1.87it/s]
+
100%|██████████| 39/39 [00:03<00:00, 12.23it/s][eval] epoch 74.0 step 2664 loss 4.6636 (best 4.1514 @ ep 34.0, 20/20 since best)
+
+
                                               [transformers] 
+
+Training completed. Do not forget to share your model on huggingface.co/models =)
+
+
+
                                                   
{'train_runtime': '1952', 'train_samples_per_second': '1879', 'train_steps_per_second': '3.744', 'train_loss': '4.316', 'epoch': '74'}
+
 36%|███▋      | 2664/7308 [32:30<41:21,  1.87it/s]
 36%|███▋      | 2664/7308 [32:30<56:40,  1.37it/s]
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+wandb: 
+wandb: 🚀 View run t2a_37M_C1_rand15 at: https://wandb.ai/slowrun-wankers/data-aug-pretraining/runs/n0ad7csv
+wandb: Find logs at: wandb/run-20260818_193241-n0ad7csv/logs
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+[rank0]:[W818 20:05:59.641763054 ProcessGroupNCCL.cpp:1575] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator())
diff --git a/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-1656/chat_template.jinja b/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-1656/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..699ff8df401fe4788525e9c1f9b86a99eadd6230
--- /dev/null
+++ b/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-1656/chat_template.jinja
@@ -0,0 +1,85 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0].role == 'system' %}
+        {{- messages[0].content + '\n\n' }}
+    {%- endif %}
+    {{- "# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0].content + '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
+{%- for message in messages[::-1] %}
+    {%- set index = (messages|length - 1) - loop.index0 %}
+    {%- if ns.multi_step_tool and message.role == "user" and not(message.content.startswith('<tool_response>') and message.content.endswith('</tool_response>')) %}
+        {%- set ns.multi_step_tool = false %}
+        {%- set ns.last_query_index = index %}
+    {%- endif %}
+{%- endfor %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {%- set content = message.content %}
+        {%- set reasoning_content = '' %}
+        {%- if message.reasoning_content is defined and message.reasoning_content is not none %}
+            {%- set reasoning_content = message.reasoning_content %}
+        {%- else %}
+            {%- if '</think>' in message.content %}
+                {%- set content = message.content.split('</think>')[-1].lstrip('\n') %}
+                {%- set reasoning_content = message.content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
+            {%- endif %}
+        {%- endif %}
+        {%- if loop.index0 > ns.last_query_index %}
+            {%- if loop.last or (not loop.last and reasoning_content) %}
+                {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content.strip('\n') + '\n</think>\n\n' + content.lstrip('\n') }}
+            {%- else %}
+                {{- '<|im_start|>' + message.role + '\n' + content }}
+            {%- endif %}
+        {%- else %}
+            {{- '<|im_start|>' + message.role + '\n' + content }}
+        {%- endif %}
+        {%- if message.tool_calls %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if (loop.first and content) or (not loop.first) %}
+                    {{- '\n' }}
+                {%- endif %}
+                {%- if tool_call.function %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {{- '<tool_call>\n{"name": "' }}
+                {{- tool_call.name }}
+                {{- '", "arguments": ' }}
+                {%- if tool_call.arguments is string %}
+                    {{- tool_call.arguments }}
+                {%- else %}
+                    {{- tool_call.arguments | tojson }}
+                {%- endif %}
+                {{- '}\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.first or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+    {%- if enable_thinking is defined and enable_thinking is false %}
+        {{- '<think>\n\n</think>\n\n' }}
+    {%- endif %}
+{%- endif %}
\ No newline at end of file
diff --git a/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-1656/config.json b/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-1656/config.json
new file mode 100644
index 0000000000000000000000000000000000000000..a3474619f10931027097cc05d837296fa4f33a6c
--- /dev/null
+++ b/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-1656/config.json
@@ -0,0 +1,32 @@
+{
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": null,
+  "dtype": "float32",
+  "eos_token_id": 151645,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 512,
+  "initializer_range": 0.02,
+  "intermediate_size": 1536,
+  "max_position_embeddings": 2048,
+  "mlp_bias": false,
+  "model_type": "llama",
+  "num_attention_heads": 4,
+  "num_hidden_layers": 20,
+  "num_key_value_heads": 4,
+  "pad_token_id": 151645,
+  "pretraining_tp": 1,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 10000.0,
+    "rope_type": "default"
+  },
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": false,
+  "vocab_size": 151676
+}
diff --git a/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-1656/generation_config.json b/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-1656/generation_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..8c4242f32d40df256857f87264f646130f684e33
--- /dev/null
+++ b/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-1656/generation_config.json
@@ -0,0 +1,11 @@
+{
+  "_from_model_config": true,
+  "eos_token_id": [
+    151645
+  ],
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "transformers_version": "5.15.0",
+  "use_cache": true
+}
diff --git a/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-1656/tokenizer_config.json b/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-1656/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..3d43c97a447f4b25a8c36be5873f7b622935fa90
--- /dev/null
+++ b/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-1656/tokenizer_config.json
@@ -0,0 +1,24 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|l2r_pred|>",
+    "<|r2l_pred|>",
+    "<|next_1_pred|>",
+    "<|next_2_pred|>",
+    "<|next_3_pred|>",
+    "<|next_4_pred|>",
+    "<|next_5_pred|>"
+  ],
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 131072,
+  "pad_token": "<|im_end|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}
diff --git a/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-1656/trainer_state.json b/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-1656/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..f2da5b8d21680d2cb4c2244419a76050990167bc
--- /dev/null
+++ b/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-1656/trainer_state.json
@@ -0,0 +1,1380 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 46.0,
+  "eval_steps": 72,
+  "global_step": 1656,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.028268551236749116,
+      "grad_norm": 2.8743574619293213,
+      "learning_rate": 0.0,
+      "loss": 12.01491928100586,
+      "step": 1
+    },
+    {
+      "epoch": 0.2826855123674912,
+      "grad_norm": 1.7668960094451904,
+      "learning_rate": 5.399999999999999e-05,
+      "loss": 11.665293375651041,
+      "step": 10
+    },
+    {
+      "epoch": 0.5653710247349824,
+      "grad_norm": 1.683153510093689,
+      "learning_rate": 0.00011399999999999999,
+      "loss": 10.979911804199219,
+      "step": 20
+    },
+    {
+      "epoch": 0.8480565371024735,
+      "grad_norm": 1.643569827079773,
+      "learning_rate": 0.00017399999999999997,
+      "loss": 10.204737854003906,
+      "step": 30
+    },
+    {
+      "epoch": 1.1130742049469964,
+      "grad_norm": 1.4890183210372925,
+      "learning_rate": 0.000234,
+      "loss": 9.147047424316407,
+      "step": 40
+    },
+    {
+      "epoch": 1.3957597173144876,
+      "grad_norm": 0.7586965560913086,
+      "learning_rate": 0.000294,
+      "loss": 8.18013916015625,
+      "step": 50
+    },
+    {
+      "epoch": 1.6784452296819787,
+      "grad_norm": 0.18824999034404755,
+      "learning_rate": 0.00035399999999999993,
+      "loss": 7.701556396484375,
+      "step": 60
+    },
+    {
+      "epoch": 1.96113074204947,
+      "grad_norm": 1.0514609813690186,
+      "learning_rate": 0.0004139999999999999,
+      "loss": 7.584004211425781,
+      "step": 70
+    },
+    {
+      "epoch": 2.0,
+      "eval_loss": 7.5267229080200195,
+      "eval_runtime": 26.9297,
+      "eval_samples_per_second": 90.643,
+      "eval_steps_per_second": 1.448,
+      "step": 72
+    },
+    {
+      "epoch": 2.2261484098939928,
+      "grad_norm": 0.21495477855205536,
+      "learning_rate": 0.000474,
+      "loss": 7.535646057128906,
+      "step": 80
+    },
+    {
+      "epoch": 2.508833922261484,
+      "grad_norm": 0.24516992270946503,
+      "learning_rate": 0.000534,
+      "loss": 7.455622863769531,
+      "step": 90
+    },
+    {
+      "epoch": 2.791519434628975,
+      "grad_norm": 0.40658488869667053,
+      "learning_rate": 0.0005939999999999999,
+      "loss": 7.425553894042968,
+      "step": 100
+    },
+    {
+      "epoch": 3.056537102473498,
+      "grad_norm": 0.26207974553108215,
+      "learning_rate": 0.0006,
+      "loss": 7.397167205810547,
+      "step": 110
+    },
+    {
+      "epoch": 3.3392226148409896,
+      "grad_norm": 0.20743456482887268,
+      "learning_rate": 0.0006,
+      "loss": 7.374625396728516,
+      "step": 120
+    },
+    {
+      "epoch": 3.6219081272084805,
+      "grad_norm": 0.2363196462392807,
+      "learning_rate": 0.0006,
+      "loss": 7.339272308349609,
+      "step": 130
+    },
+    {
+      "epoch": 3.904593639575972,
+      "grad_norm": 0.3136215806007385,
+      "learning_rate": 0.0006,
+      "loss": 7.3050788879394535,
+      "step": 140
+    },
+    {
+      "epoch": 4.0,
+      "eval_loss": 7.140003204345703,
+      "eval_runtime": 2.6268,
+      "eval_samples_per_second": 929.268,
+      "eval_steps_per_second": 14.847,
+      "step": 144
+    },
+    {
+      "epoch": 4.169611307420495,
+      "grad_norm": 0.21707509458065033,
+      "learning_rate": 0.0006,
+      "loss": 7.264811706542969,
+      "step": 150
+    },
+    {
+      "epoch": 4.4522968197879855,
+      "grad_norm": 0.4698176085948944,
+      "learning_rate": 0.0006,
+      "loss": 7.201763916015625,
+      "step": 160
+    },
+    {
+      "epoch": 4.734982332155477,
+      "grad_norm": 0.2652647793292999,
+      "learning_rate": 0.0006,
+      "loss": 7.178302001953125,
+      "step": 170
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 0.5432197451591492,
+      "learning_rate": 0.0006,
+      "loss": 7.121822357177734,
+      "step": 180
+    },
+    {
+      "epoch": 5.282685512367491,
+      "grad_norm": 0.6569910645484924,
+      "learning_rate": 0.0006,
+      "loss": 7.087571716308593,
+      "step": 190
+    },
+    {
+      "epoch": 5.565371024734983,
+      "grad_norm": 0.3058389723300934,
+      "learning_rate": 0.0006,
+      "loss": 6.976187133789063,
+      "step": 200
+    },
+    {
+      "epoch": 5.848056537102473,
+      "grad_norm": 0.38990965485572815,
+      "learning_rate": 0.0006,
+      "loss": 6.889680480957031,
+      "step": 210
+    },
+    {
+      "epoch": 6.0,
+      "eval_loss": 6.4830546379089355,
+      "eval_runtime": 2.6356,
+      "eval_samples_per_second": 926.16,
+      "eval_steps_per_second": 14.797,
+      "step": 216
+    },
+    {
+      "epoch": 6.113074204946996,
+      "grad_norm": 0.3219214975833893,
+      "learning_rate": 0.0006,
+      "loss": 6.754783630371094,
+      "step": 220
+    },
+    {
+      "epoch": 6.395759717314488,
+      "grad_norm": 0.728987455368042,
+      "learning_rate": 0.0006,
+      "loss": 6.695416259765625,
+      "step": 230
+    },
+    {
+      "epoch": 6.678445229681979,
+      "grad_norm": 0.2898261547088623,
+      "learning_rate": 0.0006,
+      "loss": 6.652455139160156,
+      "step": 240
+    },
+    {
+      "epoch": 6.96113074204947,
+      "grad_norm": 0.3531045913696289,
+      "learning_rate": 0.0006,
+      "loss": 6.578369140625,
+      "step": 250
+    },
+    {
+      "epoch": 7.226148409893993,
+      "grad_norm": 0.4957226514816284,
+      "learning_rate": 0.0006,
+      "loss": 6.493442535400391,
+      "step": 260
+    },
+    {
+      "epoch": 7.508833922261484,
+      "grad_norm": 0.39571669697761536,
+      "learning_rate": 0.0006,
+      "loss": 6.4140464782714846,
+      "step": 270
+    },
+    {
+      "epoch": 7.791519434628976,
+      "grad_norm": 0.9383025169372559,
+      "learning_rate": 0.0006,
+      "loss": 6.340756225585937,
+      "step": 280
+    },
+    {
+      "epoch": 8.0,
+      "eval_loss": 5.9631195068359375,
+      "eval_runtime": 2.6383,
+      "eval_samples_per_second": 925.215,
+      "eval_steps_per_second": 14.782,
+      "step": 288
+    },
+    {
+      "epoch": 8.056537102473499,
+      "grad_norm": 0.32992589473724365,
+      "learning_rate": 0.0006,
+      "loss": 6.30542221069336,
+      "step": 290
+    },
+    {
+      "epoch": 8.33922261484099,
+      "grad_norm": 0.3678796589374542,
+      "learning_rate": 0.0006,
+      "loss": 6.269955444335937,
+      "step": 300
+    },
+    {
+      "epoch": 8.621908127208481,
+      "grad_norm": 0.30860036611557007,
+      "learning_rate": 0.0006,
+      "loss": 6.206460952758789,
+      "step": 310
+    },
+    {
+      "epoch": 8.904593639575971,
+      "grad_norm": 0.5080128908157349,
+      "learning_rate": 0.0006,
+      "loss": 6.158573150634766,
+      "step": 320
+    },
+    {
+      "epoch": 9.169611307420494,
+      "grad_norm": 0.39322611689567566,
+      "learning_rate": 0.0006,
+      "loss": 6.152021408081055,
+      "step": 330
+    },
+    {
+      "epoch": 9.452296819787986,
+      "grad_norm": 0.43705448508262634,
+      "learning_rate": 0.0006,
+      "loss": 6.120379638671875,
+      "step": 340
+    },
+    {
+      "epoch": 9.734982332155477,
+      "grad_norm": 0.32063648104667664,
+      "learning_rate": 0.0006,
+      "loss": 6.070568466186524,
+      "step": 350
+    },
+    {
+      "epoch": 10.0,
+      "grad_norm": 0.3635176718235016,
+      "learning_rate": 0.0006,
+      "loss": 6.018914794921875,
+      "step": 360
+    },
+    {
+      "epoch": 10.0,
+      "eval_loss": 5.580841541290283,
+      "eval_runtime": 2.6427,
+      "eval_samples_per_second": 923.662,
+      "eval_steps_per_second": 14.757,
+      "step": 360
+    },
+    {
+      "epoch": 10.282685512367491,
+      "grad_norm": 0.3551004230976105,
+      "learning_rate": 0.0006,
+      "loss": 5.904553985595703,
+      "step": 370
+    },
+    {
+      "epoch": 10.565371024734983,
+      "grad_norm": 0.41268590092658997,
+      "learning_rate": 0.0006,
+      "loss": 5.919625091552734,
+      "step": 380
+    },
+    {
+      "epoch": 10.848056537102474,
+      "grad_norm": 0.42684611678123474,
+      "learning_rate": 0.0006,
+      "loss": 5.900517272949219,
+      "step": 390
+    },
+    {
+      "epoch": 11.113074204946997,
+      "grad_norm": 0.3290698528289795,
+      "learning_rate": 0.0006,
+      "loss": 5.839530181884766,
+      "step": 400
+    },
+    {
+      "epoch": 11.395759717314487,
+      "grad_norm": 0.2853550612926483,
+      "learning_rate": 0.0006,
+      "loss": 5.740743637084961,
+      "step": 410
+    },
+    {
+      "epoch": 11.678445229681978,
+      "grad_norm": 0.2584449052810669,
+      "learning_rate": 0.0006,
+      "loss": 5.709418487548828,
+      "step": 420
+    },
+    {
+      "epoch": 11.96113074204947,
+      "grad_norm": 0.4823988080024719,
+      "learning_rate": 0.0006,
+      "loss": 5.725716018676758,
+      "step": 430
+    },
+    {
+      "epoch": 12.0,
+      "eval_loss": 5.295837879180908,
+      "eval_runtime": 2.6337,
+      "eval_samples_per_second": 926.826,
+      "eval_steps_per_second": 14.808,
+      "step": 432
+    },
+    {
+      "epoch": 12.226148409893993,
+      "grad_norm": 0.44408848881721497,
+      "learning_rate": 0.0006,
+      "loss": 5.723075103759766,
+      "step": 440
+    },
+    {
+      "epoch": 12.508833922261484,
+      "grad_norm": 0.20247244834899902,
+      "learning_rate": 0.0006,
+      "loss": 5.718096542358398,
+      "step": 450
+    },
+    {
+      "epoch": 12.791519434628976,
+      "grad_norm": 0.20365497469902039,
+      "learning_rate": 0.0006,
+      "loss": 5.621192169189453,
+      "step": 460
+    },
+    {
+      "epoch": 13.056537102473499,
+      "grad_norm": 0.765959620475769,
+      "learning_rate": 0.0006,
+      "loss": 5.588929748535156,
+      "step": 470
+    },
+    {
+      "epoch": 13.33922261484099,
+      "grad_norm": 0.2936733663082123,
+      "learning_rate": 0.0006,
+      "loss": 5.610209274291992,
+      "step": 480
+    },
+    {
+      "epoch": 13.621908127208481,
+      "grad_norm": 0.21337978541851044,
+      "learning_rate": 0.0006,
+      "loss": 5.541879272460937,
+      "step": 490
+    },
+    {
+      "epoch": 13.904593639575971,
+      "grad_norm": 0.2346913367509842,
+      "learning_rate": 0.0006,
+      "loss": 5.512181091308594,
+      "step": 500
+    },
+    {
+      "epoch": 14.0,
+      "eval_loss": 5.028252124786377,
+      "eval_runtime": 2.6358,
+      "eval_samples_per_second": 926.103,
+      "eval_steps_per_second": 14.796,
+      "step": 504
+    },
+    {
+      "epoch": 14.169611307420494,
+      "grad_norm": 0.5524874925613403,
+      "learning_rate": 0.0006,
+      "loss": 5.540120697021484,
+      "step": 510
+    },
+    {
+      "epoch": 14.452296819787986,
+      "grad_norm": 0.4901306927204132,
+      "learning_rate": 0.0006,
+      "loss": 5.4401905059814455,
+      "step": 520
+    },
+    {
+      "epoch": 14.734982332155477,
+      "grad_norm": 0.23014582693576813,
+      "learning_rate": 0.0006,
+      "loss": 5.469002532958984,
+      "step": 530
+    },
+    {
+      "epoch": 15.0,
+      "grad_norm": 0.36113497614860535,
+      "learning_rate": 0.0006,
+      "loss": 5.424568939208984,
+      "step": 540
+    },
+    {
+      "epoch": 15.282685512367491,
+      "grad_norm": 0.2621628940105438,
+      "learning_rate": 0.0006,
+      "loss": 5.292803955078125,
+      "step": 550
+    },
+    {
+      "epoch": 15.565371024734983,
+      "grad_norm": 0.35952049493789673,
+      "learning_rate": 0.0006,
+      "loss": 5.256277084350586,
+      "step": 560
+    },
+    {
+      "epoch": 15.848056537102474,
+      "grad_norm": 0.25577273964881897,
+      "learning_rate": 0.0006,
+      "loss": 5.302001190185547,
+      "step": 570
+    },
+    {
+      "epoch": 16.0,
+      "eval_loss": 4.781620502471924,
+      "eval_runtime": 2.6375,
+      "eval_samples_per_second": 925.491,
+      "eval_steps_per_second": 14.787,
+      "step": 576
+    },
+    {
+      "epoch": 16.113074204946997,
+      "grad_norm": 0.4443848431110382,
+      "learning_rate": 0.0006,
+      "loss": 5.251607131958008,
+      "step": 580
+    },
+    {
+      "epoch": 16.395759717314487,
+      "grad_norm": 0.3939669728279114,
+      "learning_rate": 0.0006,
+      "loss": 5.268299865722656,
+      "step": 590
+    },
+    {
+      "epoch": 16.67844522968198,
+      "grad_norm": 0.3370627164840698,
+      "learning_rate": 0.0006,
+      "loss": 5.190369033813477,
+      "step": 600
+    },
+    {
+      "epoch": 16.96113074204947,
+      "grad_norm": 0.2633129954338074,
+      "learning_rate": 0.0006,
+      "loss": 5.186720275878907,
+      "step": 610
+    },
+    {
+      "epoch": 17.226148409893995,
+      "grad_norm": 0.34745559096336365,
+      "learning_rate": 0.0006,
+      "loss": 5.142190933227539,
+      "step": 620
+    },
+    {
+      "epoch": 17.508833922261484,
+      "grad_norm": 0.5286869406700134,
+      "learning_rate": 0.0006,
+      "loss": 5.100742340087891,
+      "step": 630
+    },
+    {
+      "epoch": 17.791519434628974,
+      "grad_norm": 0.30838146805763245,
+      "learning_rate": 0.0006,
+      "loss": 5.1460929870605465,
+      "step": 640
+    },
+    {
+      "epoch": 18.0,
+      "eval_loss": 4.597140789031982,
+      "eval_runtime": 2.6379,
+      "eval_samples_per_second": 925.351,
+      "eval_steps_per_second": 14.784,
+      "step": 648
+    },
+    {
+      "epoch": 18.0565371024735,
+      "grad_norm": 0.29841697216033936,
+      "learning_rate": 0.0006,
+      "loss": 5.1088401794433596,
+      "step": 650
+    },
+    {
+      "epoch": 18.33922261484099,
+      "grad_norm": 0.2715815603733063,
+      "learning_rate": 0.0006,
+      "loss": 5.097176361083984,
+      "step": 660
+    },
+    {
+      "epoch": 18.62190812720848,
+      "grad_norm": 0.49782630801200867,
+      "learning_rate": 0.0006,
+      "loss": 5.0622211456298825,
+      "step": 670
+    },
+    {
+      "epoch": 18.90459363957597,
+      "grad_norm": 0.41143688559532166,
+      "learning_rate": 0.0006,
+      "loss": 5.04049072265625,
+      "step": 680
+    },
+    {
+      "epoch": 19.169611307420496,
+      "grad_norm": 0.3029995560646057,
+      "learning_rate": 0.0006,
+      "loss": 5.027577590942383,
+      "step": 690
+    },
+    {
+      "epoch": 19.452296819787986,
+      "grad_norm": 0.2995366156101227,
+      "learning_rate": 0.0006,
+      "loss": 4.936068725585938,
+      "step": 700
+    },
+    {
+      "epoch": 19.73498233215548,
+      "grad_norm": 0.29094353318214417,
+      "learning_rate": 0.0006,
+      "loss": 5.024449157714844,
+      "step": 710
+    },
+    {
+      "epoch": 20.0,
+      "grad_norm": 0.2722618877887726,
+      "learning_rate": 0.0006,
+      "loss": 4.928033447265625,
+      "step": 720
+    },
+    {
+      "epoch": 20.0,
+      "eval_loss": 4.465268611907959,
+      "eval_runtime": 2.6372,
+      "eval_samples_per_second": 925.607,
+      "eval_steps_per_second": 14.788,
+      "step": 720
+    },
+    {
+      "epoch": 20.28268551236749,
+      "grad_norm": 0.5062562823295593,
+      "learning_rate": 0.0006,
+      "loss": 4.853726577758789,
+      "step": 730
+    },
+    {
+      "epoch": 20.565371024734983,
+      "grad_norm": 0.31436464190483093,
+      "learning_rate": 0.0006,
+      "loss": 4.985739135742188,
+      "step": 740
+    },
+    {
+      "epoch": 20.848056537102472,
+      "grad_norm": 0.22794955968856812,
+      "learning_rate": 0.0006,
+      "loss": 4.8645683288574215,
+      "step": 750
+    },
+    {
+      "epoch": 21.113074204946997,
+      "grad_norm": 0.2349792867898941,
+      "learning_rate": 0.0006,
+      "loss": 4.868588638305664,
+      "step": 760
+    },
+    {
+      "epoch": 21.395759717314487,
+      "grad_norm": 0.2699965834617615,
+      "learning_rate": 0.0006,
+      "loss": 4.8230735778808596,
+      "step": 770
+    },
+    {
+      "epoch": 21.67844522968198,
+      "grad_norm": 0.276645690202713,
+      "learning_rate": 0.0006,
+      "loss": 4.829259872436523,
+      "step": 780
+    },
+    {
+      "epoch": 21.96113074204947,
+      "grad_norm": 0.3184371590614319,
+      "learning_rate": 0.0006,
+      "loss": 4.8284660339355465,
+      "step": 790
+    },
+    {
+      "epoch": 22.0,
+      "eval_loss": 4.376378059387207,
+      "eval_runtime": 2.6348,
+      "eval_samples_per_second": 926.438,
+      "eval_steps_per_second": 14.802,
+      "step": 792
+    },
+    {
+      "epoch": 22.226148409893995,
+      "grad_norm": 0.4015864431858063,
+      "learning_rate": 0.0006,
+      "loss": 4.80762939453125,
+      "step": 800
+    },
+    {
+      "epoch": 22.508833922261484,
+      "grad_norm": 0.28217560052871704,
+      "learning_rate": 0.0006,
+      "loss": 4.7932586669921875,
+      "step": 810
+    },
+    {
+      "epoch": 22.791519434628974,
+      "grad_norm": 0.29987338185310364,
+      "learning_rate": 0.0006,
+      "loss": 4.878922653198242,
+      "step": 820
+    },
+    {
+      "epoch": 23.0565371024735,
+      "grad_norm": 0.2588404715061188,
+      "learning_rate": 0.0006,
+      "loss": 4.8302764892578125,
+      "step": 830
+    },
+    {
+      "epoch": 23.33922261484099,
+      "grad_norm": 0.28714779019355774,
+      "learning_rate": 0.0006,
+      "loss": 4.786470794677735,
+      "step": 840
+    },
+    {
+      "epoch": 23.62190812720848,
+      "grad_norm": 0.4411408007144928,
+      "learning_rate": 0.0006,
+      "loss": 4.767354583740234,
+      "step": 850
+    },
+    {
+      "epoch": 23.90459363957597,
+      "grad_norm": 0.2867676615715027,
+      "learning_rate": 0.0006,
+      "loss": 4.723580169677734,
+      "step": 860
+    },
+    {
+      "epoch": 24.0,
+      "eval_loss": 4.321147918701172,
+      "eval_runtime": 2.6353,
+      "eval_samples_per_second": 926.256,
+      "eval_steps_per_second": 14.799,
+      "step": 864
+    },
+    {
+      "epoch": 24.169611307420496,
+      "grad_norm": 0.3366245627403259,
+      "learning_rate": 0.0006,
+      "loss": 4.788224792480468,
+      "step": 870
+    },
+    {
+      "epoch": 24.452296819787986,
+      "grad_norm": 0.2279486358165741,
+      "learning_rate": 0.0006,
+      "loss": 4.629734039306641,
+      "step": 880
+    },
+    {
+      "epoch": 24.73498233215548,
+      "grad_norm": 0.2645115554332733,
+      "learning_rate": 0.0006,
+      "loss": 4.6986228942871096,
+      "step": 890
+    },
+    {
+      "epoch": 25.0,
+      "grad_norm": 0.2536068558692932,
+      "learning_rate": 0.0006,
+      "loss": 4.722436141967774,
+      "step": 900
+    },
+    {
+      "epoch": 25.28268551236749,
+      "grad_norm": 0.25411075353622437,
+      "learning_rate": 0.0006,
+      "loss": 4.7429862976074215,
+      "step": 910
+    },
+    {
+      "epoch": 25.565371024734983,
+      "grad_norm": 0.30635619163513184,
+      "learning_rate": 0.0006,
+      "loss": 4.686801910400391,
+      "step": 920
+    },
+    {
+      "epoch": 25.848056537102472,
+      "grad_norm": 0.2899259626865387,
+      "learning_rate": 0.0006,
+      "loss": 4.634278106689453,
+      "step": 930
+    },
+    {
+      "epoch": 26.0,
+      "eval_loss": 4.253746032714844,
+      "eval_runtime": 2.6227,
+      "eval_samples_per_second": 930.724,
+      "eval_steps_per_second": 14.87,
+      "step": 936
+    },
+    {
+      "epoch": 26.113074204946997,
+      "grad_norm": 0.3404591381549835,
+      "learning_rate": 0.0006,
+      "loss": 4.719232559204102,
+      "step": 940
+    },
+    {
+      "epoch": 26.395759717314487,
+      "grad_norm": 0.35413897037506104,
+      "learning_rate": 0.0006,
+      "loss": 4.709876632690429,
+      "step": 950
+    },
+    {
+      "epoch": 26.67844522968198,
+      "grad_norm": 0.2728523015975952,
+      "learning_rate": 0.0006,
+      "loss": 4.671334075927734,
+      "step": 960
+    },
+    {
+      "epoch": 26.96113074204947,
+      "grad_norm": 0.2241905927658081,
+      "learning_rate": 0.0006,
+      "loss": 4.648395156860351,
+      "step": 970
+    },
+    {
+      "epoch": 27.226148409893995,
+      "grad_norm": 0.3640039563179016,
+      "learning_rate": 0.0006,
+      "loss": 4.583827209472656,
+      "step": 980
+    },
+    {
+      "epoch": 27.508833922261484,
+      "grad_norm": 0.3236243426799774,
+      "learning_rate": 0.0006,
+      "loss": 4.625511932373047,
+      "step": 990
+    },
+    {
+      "epoch": 27.791519434628974,
+      "grad_norm": 0.2396886944770813,
+      "learning_rate": 0.0006,
+      "loss": 4.667061996459961,
+      "step": 1000
+    },
+    {
+      "epoch": 28.0,
+      "eval_loss": 4.206930160522461,
+      "eval_runtime": 2.6122,
+      "eval_samples_per_second": 934.47,
+      "eval_steps_per_second": 14.93,
+      "step": 1008
+    },
+    {
+      "epoch": 28.0565371024735,
+      "grad_norm": 0.3273698091506958,
+      "learning_rate": 0.0006,
+      "loss": 4.687345123291015,
+      "step": 1010
+    },
+    {
+      "epoch": 28.33922261484099,
+      "grad_norm": 0.29029396176338196,
+      "learning_rate": 0.0006,
+      "loss": 4.61729621887207,
+      "step": 1020
+    },
+    {
+      "epoch": 28.62190812720848,
+      "grad_norm": 0.21701481938362122,
+      "learning_rate": 0.0006,
+      "loss": 4.595256042480469,
+      "step": 1030
+    },
+    {
+      "epoch": 28.90459363957597,
+      "grad_norm": 0.25574585795402527,
+      "learning_rate": 0.0006,
+      "loss": 4.626471710205078,
+      "step": 1040
+    },
+    {
+      "epoch": 29.169611307420496,
+      "grad_norm": 0.40239572525024414,
+      "learning_rate": 0.0006,
+      "loss": 4.5872650146484375,
+      "step": 1050
+    },
+    {
+      "epoch": 29.452296819787986,
+      "grad_norm": 3.627098560333252,
+      "learning_rate": 0.0006,
+      "loss": 4.643867492675781,
+      "step": 1060
+    },
+    {
+      "epoch": 29.73498233215548,
+      "grad_norm": 1.5268198251724243,
+      "learning_rate": 0.0006,
+      "loss": 4.966595077514649,
+      "step": 1070
+    },
+    {
+      "epoch": 30.0,
+      "grad_norm": 0.9516384601593018,
+      "learning_rate": 0.0006,
+      "loss": 5.058966827392578,
+      "step": 1080
+    },
+    {
+      "epoch": 30.0,
+      "eval_loss": 4.588338851928711,
+      "eval_runtime": 2.621,
+      "eval_samples_per_second": 931.312,
+      "eval_steps_per_second": 14.88,
+      "step": 1080
+    },
+    {
+      "epoch": 30.28268551236749,
+      "grad_norm": 0.36072757840156555,
+      "learning_rate": 0.0006,
+      "loss": 4.98731689453125,
+      "step": 1090
+    },
+    {
+      "epoch": 30.565371024734983,
+      "grad_norm": 0.20920471847057343,
+      "learning_rate": 0.0006,
+      "loss": 4.8701927185058596,
+      "step": 1100
+    },
+    {
+      "epoch": 30.848056537102472,
+      "grad_norm": 0.16687443852424622,
+      "learning_rate": 0.0006,
+      "loss": 4.767285919189453,
+      "step": 1110
+    },
+    {
+      "epoch": 31.113074204946997,
+      "grad_norm": 0.15980704128742218,
+      "learning_rate": 0.0006,
+      "loss": 4.667778778076172,
+      "step": 1120
+    },
+    {
+      "epoch": 31.395759717314487,
+      "grad_norm": 0.16218802332878113,
+      "learning_rate": 0.0006,
+      "loss": 4.629835891723633,
+      "step": 1130
+    },
+    {
+      "epoch": 31.67844522968198,
+      "grad_norm": 0.1536215841770172,
+      "learning_rate": 0.0006,
+      "loss": 4.584923171997071,
+      "step": 1140
+    },
+    {
+      "epoch": 31.96113074204947,
+      "grad_norm": 0.15973758697509766,
+      "learning_rate": 0.0006,
+      "loss": 4.47484130859375,
+      "step": 1150
+    },
+    {
+      "epoch": 32.0,
+      "eval_loss": 4.178291320800781,
+      "eval_runtime": 2.6181,
+      "eval_samples_per_second": 932.353,
+      "eval_steps_per_second": 14.896,
+      "step": 1152
+    },
+    {
+      "epoch": 32.226148409893995,
+      "grad_norm": 0.16165708005428314,
+      "learning_rate": 0.0006,
+      "loss": 4.634153747558594,
+      "step": 1160
+    },
+    {
+      "epoch": 32.50883392226148,
+      "grad_norm": 0.15012520551681519,
+      "learning_rate": 0.0006,
+      "loss": 4.523527145385742,
+      "step": 1170
+    },
+    {
+      "epoch": 32.791519434628974,
+      "grad_norm": 0.1723557859659195,
+      "learning_rate": 0.0006,
+      "loss": 4.559682846069336,
+      "step": 1180
+    },
+    {
+      "epoch": 33.0565371024735,
+      "grad_norm": 0.19103454053401947,
+      "learning_rate": 0.0006,
+      "loss": 4.569684600830078,
+      "step": 1190
+    },
+    {
+      "epoch": 33.33922261484099,
+      "grad_norm": 0.17424911260604858,
+      "learning_rate": 0.0006,
+      "loss": 4.506781387329101,
+      "step": 1200
+    },
+    {
+      "epoch": 33.62190812720848,
+      "grad_norm": 0.17364977300167084,
+      "learning_rate": 0.0006,
+      "loss": 4.504320526123047,
+      "step": 1210
+    },
+    {
+      "epoch": 33.90459363957597,
+      "grad_norm": 0.18329985439777374,
+      "learning_rate": 0.0006,
+      "loss": 4.473563385009766,
+      "step": 1220
+    },
+    {
+      "epoch": 34.0,
+      "eval_loss": 4.137629985809326,
+      "eval_runtime": 2.6229,
+      "eval_samples_per_second": 930.662,
+      "eval_steps_per_second": 14.869,
+      "step": 1224
+    },
+    {
+      "epoch": 34.169611307420496,
+      "grad_norm": 0.19849638640880585,
+      "learning_rate": 0.0006,
+      "loss": 4.478599548339844,
+      "step": 1230
+    },
+    {
+      "epoch": 34.45229681978799,
+      "grad_norm": 0.17951421439647675,
+      "learning_rate": 0.0006,
+      "loss": 4.4578857421875,
+      "step": 1240
+    },
+    {
+      "epoch": 34.734982332155475,
+      "grad_norm": 0.23478449881076813,
+      "learning_rate": 0.0006,
+      "loss": 4.493223571777344,
+      "step": 1250
+    },
+    {
+      "epoch": 35.0,
+      "grad_norm": 0.2571250796318054,
+      "learning_rate": 0.0006,
+      "loss": 4.492740631103516,
+      "step": 1260
+    },
+    {
+      "epoch": 35.28268551236749,
+      "grad_norm": 0.22227154672145844,
+      "learning_rate": 0.0006,
+      "loss": 4.475664520263672,
+      "step": 1270
+    },
+    {
+      "epoch": 35.56537102473498,
+      "grad_norm": 0.2148682028055191,
+      "learning_rate": 0.0006,
+      "loss": 4.479212951660156,
+      "step": 1280
+    },
+    {
+      "epoch": 35.84805653710247,
+      "grad_norm": 0.18800148367881775,
+      "learning_rate": 0.0006,
+      "loss": 4.4515525817871096,
+      "step": 1290
+    },
+    {
+      "epoch": 36.0,
+      "eval_loss": 4.116716384887695,
+      "eval_runtime": 2.6283,
+      "eval_samples_per_second": 928.742,
+      "eval_steps_per_second": 14.839,
+      "step": 1296
+    },
+    {
+      "epoch": 36.113074204947,
+      "grad_norm": 0.19955147802829742,
+      "learning_rate": 0.0006,
+      "loss": 4.411654281616211,
+      "step": 1300
+    },
+    {
+      "epoch": 36.39575971731449,
+      "grad_norm": 0.23287619650363922,
+      "learning_rate": 0.0006,
+      "loss": 4.455777740478515,
+      "step": 1310
+    },
+    {
+      "epoch": 36.67844522968198,
+      "grad_norm": 0.23932035267353058,
+      "learning_rate": 0.0006,
+      "loss": 4.485030746459961,
+      "step": 1320
+    },
+    {
+      "epoch": 36.96113074204947,
+      "grad_norm": 0.22151590883731842,
+      "learning_rate": 0.0006,
+      "loss": 4.436020278930664,
+      "step": 1330
+    },
+    {
+      "epoch": 37.226148409893995,
+      "grad_norm": 0.20594088733196259,
+      "learning_rate": 0.0006,
+      "loss": 4.394052505493164,
+      "step": 1340
+    },
+    {
+      "epoch": 37.50883392226148,
+      "grad_norm": 0.22881275415420532,
+      "learning_rate": 0.0006,
+      "loss": 4.4315132141113285,
+      "step": 1350
+    },
+    {
+      "epoch": 37.791519434628974,
+      "grad_norm": 0.19765672087669373,
+      "learning_rate": 0.0006,
+      "loss": 4.4137004852294925,
+      "step": 1360
+    },
+    {
+      "epoch": 38.0,
+      "eval_loss": 4.104861736297607,
+      "eval_runtime": 2.6218,
+      "eval_samples_per_second": 931.054,
+      "eval_steps_per_second": 14.875,
+      "step": 1368
+    },
+    {
+      "epoch": 38.0565371024735,
+      "grad_norm": 0.24705128371715546,
+      "learning_rate": 0.0006,
+      "loss": 4.406088638305664,
+      "step": 1370
+    },
+    {
+      "epoch": 38.33922261484099,
+      "grad_norm": 0.24569973349571228,
+      "learning_rate": 0.0006,
+      "loss": 4.330841445922852,
+      "step": 1380
+    },
+    {
+      "epoch": 38.62190812720848,
+      "grad_norm": 0.2524004280567169,
+      "learning_rate": 0.0006,
+      "loss": 4.441641235351563,
+      "step": 1390
+    },
+    {
+      "epoch": 38.90459363957597,
+      "grad_norm": 0.21748925745487213,
+      "learning_rate": 0.0006,
+      "loss": 4.38212890625,
+      "step": 1400
+    },
+    {
+      "epoch": 39.169611307420496,
+      "grad_norm": 0.27604401111602783,
+      "learning_rate": 0.0006,
+      "loss": 4.347838592529297,
+      "step": 1410
+    },
+    {
+      "epoch": 39.45229681978799,
+      "grad_norm": 0.23354436457157135,
+      "learning_rate": 0.0006,
+      "loss": 4.404919815063477,
+      "step": 1420
+    },
+    {
+      "epoch": 39.734982332155475,
+      "grad_norm": 0.20389200747013092,
+      "learning_rate": 0.0006,
+      "loss": 4.422263717651367,
+      "step": 1430
+    },
+    {
+      "epoch": 40.0,
+      "grad_norm": 0.28286007046699524,
+      "learning_rate": 0.0006,
+      "loss": 4.357413482666016,
+      "step": 1440
+    },
+    {
+      "epoch": 40.0,
+      "eval_loss": 4.098140716552734,
+      "eval_runtime": 2.6257,
+      "eval_samples_per_second": 929.654,
+      "eval_steps_per_second": 14.853,
+      "step": 1440
+    },
+    {
+      "epoch": 40.28268551236749,
+      "grad_norm": 0.2097036987543106,
+      "learning_rate": 0.0006,
+      "loss": 4.338772583007812,
+      "step": 1450
+    },
+    {
+      "epoch": 40.56537102473498,
+      "grad_norm": 0.2298099249601364,
+      "learning_rate": 0.0006,
+      "loss": 4.375312042236328,
+      "step": 1460
+    },
+    {
+      "epoch": 40.84805653710247,
+      "grad_norm": 0.2528385519981384,
+      "learning_rate": 0.0006,
+      "loss": 4.364345932006836,
+      "step": 1470
+    },
+    {
+      "epoch": 41.113074204947,
+      "grad_norm": 0.287425696849823,
+      "learning_rate": 0.0006,
+      "loss": 4.349925231933594,
+      "step": 1480
+    },
+    {
+      "epoch": 41.39575971731449,
+      "grad_norm": 0.21734286844730377,
+      "learning_rate": 0.0006,
+      "loss": 4.352121734619141,
+      "step": 1490
+    },
+    {
+      "epoch": 41.67844522968198,
+      "grad_norm": 0.23842498660087585,
+      "learning_rate": 0.0006,
+      "loss": 4.297610855102539,
+      "step": 1500
+    },
+    {
+      "epoch": 41.96113074204947,
+      "grad_norm": 0.23788051307201385,
+      "learning_rate": 0.0006,
+      "loss": 4.320751953125,
+      "step": 1510
+    },
+    {
+      "epoch": 42.0,
+      "eval_loss": 4.085766792297363,
+      "eval_runtime": 2.6137,
+      "eval_samples_per_second": 933.928,
+      "eval_steps_per_second": 14.921,
+      "step": 1512
+    },
+    {
+      "epoch": 42.226148409893995,
+      "grad_norm": 0.3015732169151306,
+      "learning_rate": 0.0006,
+      "loss": 4.363670349121094,
+      "step": 1520
+    },
+    {
+      "epoch": 42.50883392226148,
+      "grad_norm": 0.3243810832500458,
+      "learning_rate": 0.0006,
+      "loss": 4.35053596496582,
+      "step": 1530
+    },
+    {
+      "epoch": 42.791519434628974,
+      "grad_norm": 0.23295873403549194,
+      "learning_rate": 0.0006,
+      "loss": 4.3441162109375,
+      "step": 1540
+    },
+    {
+      "epoch": 43.0565371024735,
+      "grad_norm": 0.26591989398002625,
+      "learning_rate": 0.0006,
+      "loss": 4.339892578125,
+      "step": 1550
+    },
+    {
+      "epoch": 43.33922261484099,
+      "grad_norm": 0.2581478953361511,
+      "learning_rate": 0.0006,
+      "loss": 4.3083984375,
+      "step": 1560
+    },
+    {
+      "epoch": 43.62190812720848,
+      "grad_norm": 0.3253999352455139,
+      "learning_rate": 0.0006,
+      "loss": 4.316517639160156,
+      "step": 1570
+    },
+    {
+      "epoch": 43.90459363957597,
+      "grad_norm": 0.22917279601097107,
+      "learning_rate": 0.0006,
+      "loss": 4.337581634521484,
+      "step": 1580
+    },
+    {
+      "epoch": 44.0,
+      "eval_loss": 4.083998203277588,
+      "eval_runtime": 2.6287,
+      "eval_samples_per_second": 928.59,
+      "eval_steps_per_second": 14.836,
+      "step": 1584
+    },
+    {
+      "epoch": 44.169611307420496,
+      "grad_norm": 0.2844901382923126,
+      "learning_rate": 0.0006,
+      "loss": 4.253636169433594,
+      "step": 1590
+    },
+    {
+      "epoch": 44.45229681978799,
+      "grad_norm": 0.24951669573783875,
+      "learning_rate": 0.0006,
+      "loss": 4.200857543945313,
+      "step": 1600
+    },
+    {
+      "epoch": 44.734982332155475,
+      "grad_norm": 0.23017176985740662,
+      "learning_rate": 0.0006,
+      "loss": 4.3106330871582035,
+      "step": 1610
+    },
+    {
+      "epoch": 45.0,
+      "grad_norm": 0.27370086312294006,
+      "learning_rate": 0.0006,
+      "loss": 4.201045989990234,
+      "step": 1620
+    },
+    {
+      "epoch": 45.28268551236749,
+      "grad_norm": 0.3193263113498688,
+      "learning_rate": 0.0006,
+      "loss": 4.2449493408203125,
+      "step": 1630
+    },
+    {
+      "epoch": 45.56537102473498,
+      "grad_norm": 0.26069021224975586,
+      "learning_rate": 0.0006,
+      "loss": 4.214767074584961,
+      "step": 1640
+    },
+    {
+      "epoch": 45.84805653710247,
+      "grad_norm": 0.28351539373397827,
+      "learning_rate": 0.0006,
+      "loss": 4.223235321044922,
+      "step": 1650
+    },
+    {
+      "epoch": 46.0,
+      "eval_loss": 4.079798698425293,
+      "eval_runtime": 2.6355,
+      "eval_samples_per_second": 926.207,
+      "eval_steps_per_second": 14.798,
+      "step": 1656
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 7308,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 203,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": false,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 6.97995222029697e+17,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/runs/t2a/t2a_37M_C2_r2l50_i5exp/train.log b/runs/t2a/t2a_37M_C2_r2l50_i5exp/train.log
new file mode 100644
index 0000000000000000000000000000000000000000..6a4eb7ed13ea78280ab3955a527369e0d94efe7e
--- /dev/null
+++ b/runs/t2a/t2a_37M_C2_r2l50_i5exp/train.log
@@ -0,0 +1,2652 @@
+W0818 20:06:03.301000 3018287 torch/distributed/run.py:851] 
+W0818 20:06:03.301000 3018287 torch/distributed/run.py:851] *****************************************
+W0818 20:06:03.301000 3018287 torch/distributed/run.py:851] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+W0818 20:06:03.301000 3018287 torch/distributed/run.py:851] *****************************************
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] loading configuration file config.json from cache at /home/ubuntu/.cache/huggingface/hub/models--Qwen--Qwen3-Embedding-0.6B/snapshots/97b0c614be4d77ee51c0cef4e5f07c00f9eb65b3/config.json
+EOS token: <|im_end|>
+[transformers] Model config Qwen3Config {
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.15.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151669
+}
+
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+EOS token: <|im_end|>
+[transformers] Generate config GenerationConfig {
+  "eos_token_id": 151645,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 151645,
+  "use_cache": true
+}
+
+total params: 145836544
+[dataset] loaded packed cache: 18063 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+total params: 145836544
+[dataset] loaded packed cache: 18063 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+total params: 145836544
+[dataset] loaded packed cache: 18063 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+total params: 145836544
+[dataset] loaded packed cache: 18063 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+total params: 145836544
+[dataset] loaded packed cache: 18063 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+total params: 145836544
+[dataset] loaded packed cache: 18063 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+total params: 145836544
+[dataset] loaded packed cache: 18063 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+total params: 145836544
+[dataset] loaded packed cache: 18063 train / 2441 eval blocks
+[transformers] PyTorch: setting up devices
+[transformers] Setting TF32 in CUDA backends to speedup torch compile, you won't see any improvement otherwise.
+[transformers] ***** Running training *****
+[transformers]   Num examples = 18,063
+[transformers]   Num Epochs = 203
+[transformers]   Num update steps per epoch = 36
+[transformers]   Instantaneous batch size per device = 8
+[transformers]   Total train batch size (w. parallel, distributed & accumulation) = 512
+[transformers]   Gradient Accumulation steps = 8
+[transformers]   Total optimization steps = 7,308
+[transformers]   Number of trainable parameters = 145,836,544
+wandb: WARNING Changes to your `wandb` environment variables will be ignored because your `wandb` session has already started. For more information on how to modify your settings with `wandb.init()` arguments, please refer to https://wandb.me/wandb-init.
+wandb: setting up run oo3nkqdx
+wandb: Tracking run with wandb version 0.28.2
+wandb: Run data is saved locally in /home/ubuntu/workspace/data-augmentations-for-pretraining/wandb/run-20260818_200623-oo3nkqdx
+wandb: Run `wandb offline` to turn off syncing.
+wandb: Syncing run t2a_37M_C2_r2l50_i5exp
+wandb: ⭐️ View project at https://wandb.ai/slowrun-wankers/data-aug-pretraining
+wandb: 🚀 View run at https://wandb.ai/slowrun-wankers/data-aug-pretraining/runs/oo3nkqdx
+
  0%|          | 0/7308 [00:00<?, ?it/s]wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+[rank5]:W0818 20:06:46.719000 3018479 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+[rank3]:W0818 20:06:46.828000 3018477 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+[rank2]:W0818 20:06:46.866000 3018476 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+[rank6]:W0818 20:06:46.887000 3018480 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+[rank7]:W0818 20:06:46.893000 3018481 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+[rank1]:W0818 20:06:46.944000 3018475 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+[rank4]:W0818 20:06:48.053000 3018478 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+[rank0]:W0818 20:06:48.701000 3018474 torch/_logging/_internal.py:1279] [0/0] Profiler record function <class 'torch.autograd.profiler.record_function'> will be ignored
+
  0%|          | 1/7308 [00:40<82:48:03, 40.79s/it]
                                                   
{'loss': '12.01', 'grad_norm': '2.874', 'learning_rate': '0', 'epoch': '0.02827'}
+
  0%|          | 1/7308 [00:40<82:48:03, 40.79s/it]
  0%|          | 2/7308 [00:41<34:53:11, 17.19s/it]
  0%|          | 3/7308 [00:42<19:32:15,  9.63s/it]
  0%|          | 4/7308 [00:42<12:19:41,  6.08s/it]
  0%|          | 5/7308 [00:43<8:20:35,  4.11s/it] 
  0%|          | 6/7308 [00:43<5:56:28,  2.93s/it]
  0%|          | 7/7308 [00:44<4:25:01,  2.18s/it]
  0%|          | 8/7308 [00:45<3:25:04,  1.69s/it]
  0%|          | 9/7308 [00:45<2:44:55,  1.36s/it]
  0%|          | 10/7308 [00:46<2:17:41,  1.13s/it]
                                                   
{'loss': '11.67', 'grad_norm': '1.767', 'learning_rate': '5.4e-05', 'epoch': '0.2827'}
+
  0%|          | 10/7308 [00:46<2:17:41,  1.13s/it]
  0%|          | 11/7308 [00:47<1:59:00,  1.02it/s]
  0%|          | 12/7308 [00:47<1:46:08,  1.15it/s]
  0%|          | 13/7308 [00:48<1:37:11,  1.25it/s]
  0%|          | 14/7308 [00:49<1:30:58,  1.34it/s]
  0%|          | 15/7308 [00:49<1:26:38,  1.40it/s]
  0%|          | 16/7308 [00:50<1:23:37,  1.45it/s]
  0%|          | 17/7308 [00:50<1:21:30,  1.49it/s]
  0%|          | 18/7308 [00:51<1:20:01,  1.52it/s]
  0%|          | 19/7308 [00:52<1:18:59,  1.54it/s]
  0%|          | 20/7308 [00:52<1:18:14,  1.55it/s]
                                                   
{'loss': '10.98', 'grad_norm': '1.683', 'learning_rate': '0.000114', 'epoch': '0.5654'}
+
  0%|          | 20/7308 [00:52<1:18:14,  1.55it/s]
  0%|          | 21/7308 [00:53<1:17:43,  1.56it/s]
  0%|          | 22/7308 [00:54<1:17:19,  1.57it/s]
  0%|          | 23/7308 [00:54<1:17:02,  1.58it/s]
  0%|          | 24/7308 [00:55<1:16:52,  1.58it/s]
  0%|          | 25/7308 [00:55<1:16:43,  1.58it/s]
  0%|          | 26/7308 [00:56<1:16:34,  1.58it/s]
  0%|          | 27/7308 [00:57<1:16:29,  1.59it/s]
  0%|          | 28/7308 [00:57<1:16:24,  1.59it/s]
  0%|          | 29/7308 [00:58<1:16:21,  1.59it/s]
  0%|          | 30/7308 [00:59<1:16:18,  1.59it/s]
                                                   
{'loss': '10.2', 'grad_norm': '1.644', 'learning_rate': '0.000174', 'epoch': '0.8481'}
+
  0%|          | 30/7308 [00:59<1:16:18,  1.59it/s]
  0%|          | 31/7308 [00:59<1:16:21,  1.59it/s]
  0%|          | 32/7308 [01:00<1:16:17,  1.59it/s]
  0%|          | 33/7308 [01:00<1:16:12,  1.59it/s]
  0%|          | 34/7308 [01:01<1:16:07,  1.59it/s]
  0%|          | 35/7308 [01:02<1:16:03,  1.59it/s]
  0%|          | 36/7308 [01:02<1:01:58,  1.96it/s]
  1%|          | 37/7308 [01:03<1:09:53,  1.73it/s]
  1%|          | 38/7308 [01:03<1:11:43,  1.69it/s]
  1%|          | 39/7308 [01:04<1:13:00,  1.66it/s]
  1%|          | 40/7308 [01:05<1:13:55,  1.64it/s]
                                                   
{'loss': '9.147', 'grad_norm': '1.489', 'learning_rate': '0.000234', 'epoch': '1.113'}
+
  1%|          | 40/7308 [01:05<1:13:55,  1.64it/s]
  1%|          | 41/7308 [01:05<1:14:40,  1.62it/s]
  1%|          | 42/7308 [01:06<1:15:05,  1.61it/s]
  1%|          | 43/7308 [01:06<1:15:20,  1.61it/s]
  1%|          | 44/7308 [01:07<1:15:30,  1.60it/s]
  1%|          | 45/7308 [01:08<1:15:38,  1.60it/s]
  1%|          | 46/7308 [01:08<1:15:43,  1.60it/s]
  1%|          | 47/7308 [01:09<1:15:47,  1.60it/s]
  1%|          | 48/7308 [01:10<1:15:49,  1.60it/s]
  1%|          | 49/7308 [01:10<1:15:52,  1.59it/s]
  1%|          | 50/7308 [01:11<1:15:55,  1.59it/s]
                                                   
{'loss': '8.18', 'grad_norm': '0.7587', 'learning_rate': '0.000294', 'epoch': '1.396'}
+
  1%|          | 50/7308 [01:11<1:15:55,  1.59it/s]
  1%|          | 51/7308 [01:12<1:15:57,  1.59it/s]
  1%|          | 52/7308 [01:12<1:15:56,  1.59it/s]
  1%|          | 53/7308 [01:13<1:15:58,  1.59it/s]
  1%|          | 54/7308 [01:13<1:15:59,  1.59it/s]
  1%|          | 55/7308 [01:14<1:15:57,  1.59it/s]
  1%|          | 56/7308 [01:15<1:15:58,  1.59it/s]
  1%|          | 57/7308 [01:15<1:16:01,  1.59it/s]
  1%|          | 58/7308 [01:16<1:16:01,  1.59it/s]
  1%|          | 59/7308 [01:17<1:16:00,  1.59it/s]
  1%|          | 60/7308 [01:17<1:15:59,  1.59it/s]
                                                   
{'loss': '7.702', 'grad_norm': '0.1882', 'learning_rate': '0.000354', 'epoch': '1.678'}
+
  1%|          | 60/7308 [01:17<1:15:59,  1.59it/s]
  1%|          | 61/7308 [01:18<1:15:59,  1.59it/s]
  1%|          | 62/7308 [01:18<1:15:58,  1.59it/s]
  1%|          | 63/7308 [01:19<1:15:58,  1.59it/s]
  1%|          | 64/7308 [01:20<1:15:57,  1.59it/s]
  1%|          | 65/7308 [01:20<1:15:56,  1.59it/s]
  1%|          | 66/7308 [01:21<1:15:55,  1.59it/s]
  1%|          | 67/7308 [01:22<1:15:59,  1.59it/s]
  1%|          | 68/7308 [01:22<1:15:58,  1.59it/s]
  1%|          | 69/7308 [01:23<1:15:58,  1.59it/s]
  1%|          | 70/7308 [01:23<1:15:56,  1.59it/s]
                                                   
{'loss': '7.584', 'grad_norm': '1.051', 'learning_rate': '0.000414', 'epoch': '1.961'}
+
  1%|          | 70/7308 [01:23<1:15:56,  1.59it/s]
  1%|          | 71/7308 [01:24<1:15:56,  1.59it/s]
  1%|          | 72/7308 [01:24<1:01:52,  1.95it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: [wandb.login()] Using explicit session credentials for https://api.wandb.ai.
+wandb: Appending key for api.wandb.ai to your netrc file: /home/ubuntu/.netrc
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+wandb: Currently logged in as: michaelckj (slowrun-wankers) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
 10%|█         | 4/39 [00:00<00:01, 25.30it/s]
+
 18%|█▊        | 7/39 [00:00<00:01, 18.81it/s]
+
 23%|██▎       | 9/39 [00:00<00:01, 17.49it/s]
+
 28%|██▊       | 11/39 [00:00<00:01, 16.64it/s]
+
 33%|███▎      | 13/39 [00:00<00:01, 16.20it/s]
+
 38%|███▊      | 15/39 [00:00<00:01, 15.87it/s]
+
 44%|████▎     | 17/39 [00:01<00:01, 15.66it/s]
+
 49%|████▊     | 19/39 [00:01<00:01, 15.48it/s]
+
 54%|█████▍    | 21/39 [00:01<00:01, 15.42it/s]
+
 59%|█████▉    | 23/39 [00:01<00:01, 15.35it/s]
+
 64%|██████▍   | 25/39 [00:01<00:00, 15.31it/s]
+
 69%|██████▉   | 27/39 [00:01<00:00, 15.30it/s]
+
 74%|███████▍  | 29/39 [00:01<00:00, 15.28it/s]
+
 79%|███████▉  | 31/39 [00:01<00:00, 15.26it/s]
+
 85%|████████▍ | 33/39 [00:02<00:00, 15.26it/s]
+
 90%|████████▉ | 35/39 [00:02<00:00, 15.25it/s]
+
 95%|█████████▍| 37/39 [00:02<00:00, 15.24it/s]
+
100%|██████████| 39/39 [00:02<00:00, 15.25it/s]
                                                   
+
                                               
{'eval_loss': '7.527', 'eval_runtime': '26.93', 'eval_samples_per_second': '90.64', 'eval_steps_per_second': '1.448', 'epoch': '2'}
+
  1%|          | 72/7308 [01:51<1:01:52,  1.95it/s]
+
100%|██████████| 39/39 [00:02<00:00, 15.25it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-72
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-72/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-72/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.73it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.73it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-72/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-72/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-72/tokenizer_config.json
+[eval] epoch 2.0 step 72 loss 7.5267 (best 7.5267 @ ep 2.0, 0/20 since best)
+
+
                                               
  1%|          | 73/7308 [01:54<18:32:43,  9.23s/it]
  1%|          | 74/7308 [01:55<13:21:37,  6.65s/it]
  1%|          | 75/7308 [01:55<9:43:56,  4.84s/it] 
  1%|          | 76/7308 [01:56<7:11:32,  3.58s/it]
  1%|          | 77/7308 [01:56<5:24:54,  2.70s/it]
  1%|          | 78/7308 [01:57<4:10:13,  2.08s/it]
  1%|          | 79/7308 [01:58<3:18:02,  1.64s/it]
  1%|          | 80/7308 [01:58<2:41:25,  1.34s/it]
                                                   
{'loss': '7.536', 'grad_norm': '0.215', 'learning_rate': '0.000474', 'epoch': '2.226'}
+
  1%|          | 80/7308 [01:58<2:41:25,  1.34s/it]
  1%|          | 81/7308 [01:59<2:15:50,  1.13s/it]
  1%|          | 82/7308 [02:00<1:57:56,  1.02it/s]
  1%|          | 83/7308 [02:00<1:45:23,  1.14it/s]
  1%|          | 84/7308 [02:01<1:36:35,  1.25it/s]
  1%|          | 85/7308 [02:01<1:30:24,  1.33it/s]
  1%|          | 86/7308 [02:02<1:26:06,  1.40it/s]
  1%|          | 87/7308 [02:03<1:23:04,  1.45it/s]
  1%|          | 88/7308 [02:03<1:20:56,  1.49it/s]
  1%|          | 89/7308 [02:04<1:19:27,  1.51it/s]
  1%|          | 90/7308 [02:05<1:18:26,  1.53it/s]
                                                   
{'loss': '7.456', 'grad_norm': '0.2452', 'learning_rate': '0.000534', 'epoch': '2.509'}
+
  1%|          | 90/7308 [02:05<1:18:26,  1.53it/s]
  1%|          | 91/7308 [02:05<1:17:42,  1.55it/s]
  1%|▏         | 92/7308 [02:06<1:17:13,  1.56it/s]
  1%|▏         | 93/7308 [02:07<1:16:52,  1.56it/s]
  1%|▏         | 94/7308 [02:07<1:16:37,  1.57it/s]
  1%|▏         | 95/7308 [02:08<1:16:37,  1.57it/s]
  1%|▏         | 96/7308 [02:08<1:16:24,  1.57it/s]
  1%|▏         | 97/7308 [02:09<1:16:15,  1.58it/s]
  1%|▏         | 98/7308 [02:10<1:16:09,  1.58it/s]
  1%|▏         | 99/7308 [02:10<1:16:05,  1.58it/s]
  1%|▏         | 100/7308 [02:11<1:16:02,  1.58it/s]
                                                    
{'loss': '7.426', 'grad_norm': '0.4066', 'learning_rate': '0.000594', 'epoch': '2.792'}
+
  1%|▏         | 100/7308 [02:11<1:16:02,  1.58it/s]
  1%|▏         | 101/7308 [02:12<1:16:00,  1.58it/s]
  1%|▏         | 102/7308 [02:12<1:15:57,  1.58it/s]
  1%|▏         | 103/7308 [02:13<1:15:59,  1.58it/s]
  1%|▏         | 104/7308 [02:13<1:15:58,  1.58it/s]
  1%|▏         | 105/7308 [02:14<1:15:57,  1.58it/s]
  1%|▏         | 106/7308 [02:15<1:15:52,  1.58it/s]
  1%|▏         | 107/7308 [02:15<1:15:46,  1.58it/s]
  1%|▏         | 108/7308 [02:16<1:01:45,  1.94it/s]
  1%|▏         | 109/7308 [02:16<1:09:08,  1.74it/s]
  2%|▏         | 110/7308 [02:17<1:11:07,  1.69it/s]
                                                    
{'loss': '7.397', 'grad_norm': '0.2621', 'learning_rate': '0.0006', 'epoch': '3.057'}
+
  2%|▏         | 110/7308 [02:17<1:11:07,  1.69it/s]
  2%|▏         | 111/7308 [02:18<1:12:32,  1.65it/s]
  2%|▏         | 112/7308 [02:18<1:13:37,  1.63it/s]
  2%|▏         | 113/7308 [02:19<1:14:15,  1.61it/s]
  2%|▏         | 114/7308 [02:20<1:14:43,  1.60it/s]
  2%|▏         | 115/7308 [02:20<1:15:03,  1.60it/s]
  2%|▏         | 116/7308 [02:21<1:15:16,  1.59it/s]
  2%|▏         | 117/7308 [02:21<1:15:26,  1.59it/s]
  2%|▏         | 118/7308 [02:22<1:15:32,  1.59it/s]
  2%|▏         | 119/7308 [02:23<1:15:35,  1.59it/s]
  2%|▏         | 120/7308 [02:23<1:15:40,  1.58it/s]
                                                    
{'loss': '7.375', 'grad_norm': '0.2074', 'learning_rate': '0.0006', 'epoch': '3.339'}
+
  2%|▏         | 120/7308 [02:23<1:15:40,  1.58it/s]
  2%|▏         | 121/7308 [02:24<1:15:43,  1.58it/s]
  2%|▏         | 122/7308 [02:25<1:15:44,  1.58it/s]
  2%|▏         | 123/7308 [02:25<1:15:43,  1.58it/s]
  2%|▏         | 124/7308 [02:26<1:15:44,  1.58it/s]
  2%|▏         | 125/7308 [02:26<1:15:48,  1.58it/s]
  2%|▏         | 126/7308 [02:27<1:15:46,  1.58it/s]
  2%|▏         | 127/7308 [02:28<1:15:45,  1.58it/s]
  2%|▏         | 128/7308 [02:28<1:15:47,  1.58it/s]
  2%|▏         | 129/7308 [02:29<1:15:46,  1.58it/s]
  2%|▏         | 130/7308 [02:30<1:15:44,  1.58it/s]
                                                    
{'loss': '7.339', 'grad_norm': '0.2363', 'learning_rate': '0.0006', 'epoch': '3.622'}
+
  2%|▏         | 130/7308 [02:30<1:15:44,  1.58it/s]
  2%|▏         | 131/7308 [02:30<1:15:47,  1.58it/s]
  2%|▏         | 132/7308 [02:31<1:15:46,  1.58it/s]
  2%|▏         | 133/7308 [02:32<1:15:43,  1.58it/s]
  2%|▏         | 134/7308 [02:32<1:15:42,  1.58it/s]
  2%|▏         | 135/7308 [02:33<1:15:41,  1.58it/s]
  2%|▏         | 136/7308 [02:33<1:15:42,  1.58it/s]
  2%|▏         | 137/7308 [02:34<1:15:41,  1.58it/s]
  2%|▏         | 138/7308 [02:35<1:15:41,  1.58it/s]
  2%|▏         | 139/7308 [02:35<1:15:40,  1.58it/s]
  2%|▏         | 140/7308 [02:36<1:15:38,  1.58it/s]
                                                    
{'loss': '7.305', 'grad_norm': '0.3136', 'learning_rate': '0.0006', 'epoch': '3.905'}
+
  2%|▏         | 140/7308 [02:36<1:15:38,  1.58it/s]
  2%|▏         | 141/7308 [02:37<1:15:41,  1.58it/s]
  2%|▏         | 142/7308 [02:37<1:15:40,  1.58it/s]
  2%|▏         | 143/7308 [02:38<1:15:33,  1.58it/s]
  2%|▏         | 144/7308 [02:38<1:01:33,  1.94it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 22.34it/s]
+
 15%|█▌        | 6/39 [00:00<00:01, 17.46it/s]
+
 21%|██        | 8/39 [00:00<00:01, 16.58it/s]
+
 26%|██▌       | 10/39 [00:00<00:01, 16.10it/s]
+
 31%|███       | 12/39 [00:00<00:01, 15.79it/s]
+
 36%|███▌      | 14/39 [00:00<00:01, 15.61it/s]
+
 41%|████      | 16/39 [00:00<00:01, 15.48it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 15.41it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 15.34it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 15.31it/s]
+
 62%|██████▏   | 24/39 [00:01<00:00, 15.29it/s]
+
 67%|██████▋   | 26/39 [00:01<00:00, 15.26it/s]
+
 72%|███████▏  | 28/39 [00:01<00:00, 15.24it/s]
+
 77%|███████▋  | 30/39 [00:01<00:00, 15.24it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 15.22it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 15.22it/s]
+
 92%|█████████▏| 36/39 [00:02<00:00, 15.19it/s]
+
 97%|█████████▋| 38/39 [00:02<00:00, 15.12it/s]
                                                    
+
                                               
{'eval_loss': '7.14', 'eval_runtime': '2.627', 'eval_samples_per_second': '929.3', 'eval_steps_per_second': '14.85', 'epoch': '4'}
+
  2%|▏         | 144/7308 [02:41<1:01:33,  1.94it/s]
+
100%|██████████| 39/39 [00:02<00:00, 15.12it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-144
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-144/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-144/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.89it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.89it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-144/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-144/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-144/tokenizer_config.json
+[eval] epoch 4.0 step 144 loss 7.1400 (best 7.1400 @ ep 4.0, 0/20 since best)
+
+
                                               
  2%|▏         | 145/7308 [02:43<3:54:16,  1.96s/it]
  2%|▏         | 146/7308 [02:44<3:06:54,  1.57s/it]
  2%|▏         | 147/7308 [02:45<2:33:34,  1.29s/it]
  2%|▏         | 148/7308 [02:45<2:10:09,  1.09s/it]
  2%|▏         | 149/7308 [02:46<1:53:45,  1.05it/s]
  2%|▏         | 150/7308 [02:47<1:42:18,  1.17it/s]
                                                    
{'loss': '7.265', 'grad_norm': '0.2171', 'learning_rate': '0.0006', 'epoch': '4.17'}
+
  2%|▏         | 150/7308 [02:47<1:42:18,  1.17it/s]
  2%|▏         | 151/7308 [02:47<1:34:18,  1.26it/s]
  2%|▏         | 152/7308 [02:48<1:28:39,  1.35it/s]
  2%|▏         | 153/7308 [02:49<1:24:42,  1.41it/s]
  2%|▏         | 154/7308 [02:49<1:21:55,  1.46it/s]
  2%|▏         | 155/7308 [02:50<1:19:59,  1.49it/s]
  2%|▏         | 156/7308 [02:50<1:18:38,  1.52it/s]
  2%|▏         | 157/7308 [02:51<1:17:40,  1.53it/s]
  2%|▏         | 158/7308 [02:52<1:17:00,  1.55it/s]
  2%|▏         | 159/7308 [02:52<1:16:31,  1.56it/s]
  2%|▏         | 160/7308 [02:53<1:16:12,  1.56it/s]
                                                    
{'loss': '7.202', 'grad_norm': '0.4698', 'learning_rate': '0.0006', 'epoch': '4.452'}
+
  2%|▏         | 160/7308 [02:53<1:16:12,  1.56it/s]
  2%|▏         | 161/7308 [02:54<1:15:58,  1.57it/s]
  2%|▏         | 162/7308 [02:54<1:15:48,  1.57it/s]
  2%|▏         | 163/7308 [02:55<1:15:41,  1.57it/s]
  2%|▏         | 164/7308 [02:56<1:15:35,  1.58it/s]
  2%|▏         | 165/7308 [02:56<1:15:29,  1.58it/s]
  2%|▏         | 166/7308 [02:57<1:15:26,  1.58it/s]
  2%|▏         | 167/7308 [02:57<1:15:28,  1.58it/s]
  2%|▏         | 168/7308 [02:58<1:15:25,  1.58it/s]
  2%|▏         | 169/7308 [02:59<1:15:24,  1.58it/s]
  2%|▏         | 170/7308 [02:59<1:15:22,  1.58it/s]
                                                    
{'loss': '7.178', 'grad_norm': '0.2653', 'learning_rate': '0.0006', 'epoch': '4.735'}
+
  2%|▏         | 170/7308 [02:59<1:15:22,  1.58it/s]
  2%|▏         | 171/7308 [03:00<1:15:21,  1.58it/s]
  2%|▏         | 172/7308 [03:01<1:15:24,  1.58it/s]
  2%|▏         | 173/7308 [03:01<1:15:21,  1.58it/s]
  2%|▏         | 174/7308 [03:02<1:30:58,  1.31it/s]
  2%|▏         | 175/7308 [03:03<1:26:15,  1.38it/s]
  2%|▏         | 176/7308 [03:04<1:22:57,  1.43it/s]
  2%|▏         | 177/7308 [03:04<1:20:38,  1.47it/s]
  2%|▏         | 178/7308 [03:05<1:18:57,  1.50it/s]
  2%|▏         | 179/7308 [03:05<1:17:45,  1.53it/s]
  2%|▏         | 180/7308 [03:06<1:03:03,  1.88it/s]
                                                    
{'loss': '7.122', 'grad_norm': '0.5432', 'learning_rate': '0.0006', 'epoch': '5'}
+
  2%|▏         | 180/7308 [03:06<1:03:03,  1.88it/s]
  2%|▏         | 181/7308 [03:06<1:09:38,  1.71it/s]
  2%|▏         | 182/7308 [03:07<1:11:17,  1.67it/s]
  3%|▎         | 183/7308 [03:08<1:12:24,  1.64it/s]
  3%|▎         | 184/7308 [03:08<1:13:12,  1.62it/s]
  3%|▎         | 185/7308 [03:09<1:13:44,  1.61it/s]
  3%|▎         | 186/7308 [03:10<1:14:08,  1.60it/s]
  3%|▎         | 187/7308 [03:10<1:14:22,  1.60it/s]
  3%|▎         | 188/7308 [03:11<1:14:33,  1.59it/s]
  3%|▎         | 189/7308 [03:11<1:14:40,  1.59it/s]
  3%|▎         | 190/7308 [03:12<1:14:44,  1.59it/s]
                                                    
{'loss': '7.088', 'grad_norm': '0.657', 'learning_rate': '0.0006', 'epoch': '5.283'}
+
  3%|▎         | 190/7308 [03:12<1:14:44,  1.59it/s]
  3%|▎         | 191/7308 [03:13<1:14:49,  1.59it/s]
  3%|▎         | 192/7308 [03:14<1:30:29,  1.31it/s]
  3%|▎         | 193/7308 [03:14<1:25:49,  1.38it/s]
  3%|▎         | 194/7308 [03:15<1:22:34,  1.44it/s]
  3%|▎         | 195/7308 [03:16<1:20:16,  1.48it/s]
  3%|▎         | 196/7308 [03:16<1:18:40,  1.51it/s]
  3%|▎         | 197/7308 [03:17<1:17:35,  1.53it/s]
  3%|▎         | 198/7308 [03:18<1:16:50,  1.54it/s]
  3%|▎         | 199/7308 [03:18<1:16:17,  1.55it/s]
  3%|▎         | 200/7308 [03:19<1:15:52,  1.56it/s]
                                                    
{'loss': '6.976', 'grad_norm': '0.3058', 'learning_rate': '0.0006', 'epoch': '5.565'}
+
  3%|▎         | 200/7308 [03:19<1:15:52,  1.56it/s]
  3%|▎         | 201/7308 [03:19<1:15:38,  1.57it/s]
  3%|▎         | 202/7308 [03:20<1:15:24,  1.57it/s]
  3%|▎         | 203/7308 [03:21<1:15:15,  1.57it/s]
  3%|▎         | 204/7308 [03:21<1:15:09,  1.58it/s]
  3%|▎         | 205/7308 [03:22<1:15:04,  1.58it/s]
  3%|▎         | 206/7308 [03:23<1:15:00,  1.58it/s]
  3%|▎         | 207/7308 [03:23<1:14:59,  1.58it/s]
  3%|▎         | 208/7308 [03:24<1:14:57,  1.58it/s]
  3%|▎         | 209/7308 [03:25<1:14:55,  1.58it/s]
  3%|▎         | 210/7308 [03:25<1:14:53,  1.58it/s]
                                                    
{'loss': '6.89', 'grad_norm': '0.3899', 'learning_rate': '0.0006', 'epoch': '5.848'}
+
  3%|▎         | 210/7308 [03:25<1:14:53,  1.58it/s]
  3%|▎         | 211/7308 [03:26<1:14:53,  1.58it/s]
  3%|▎         | 212/7308 [03:26<1:14:53,  1.58it/s]
  3%|▎         | 213/7308 [03:27<1:14:54,  1.58it/s]
  3%|▎         | 214/7308 [03:28<1:14:55,  1.58it/s]
  3%|▎         | 215/7308 [03:29<1:40:40,  1.17it/s]
  3%|▎         | 216/7308 [03:29<1:19:04,  1.49it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 23.43it/s]
+
 15%|█▌        | 6/39 [00:00<00:01, 17.71it/s]
+
 21%|██        | 8/39 [00:00<00:01, 16.73it/s]
+
 26%|██▌       | 10/39 [00:00<00:01, 16.14it/s]
+
 31%|███       | 12/39 [00:00<00:01, 15.81it/s]
+
 36%|███▌      | 14/39 [00:00<00:01, 15.59it/s]
+
 41%|████      | 16/39 [00:00<00:01, 15.49it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 15.38it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 15.34it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 15.27it/s]
+
 62%|██████▏   | 24/39 [00:01<00:00, 15.30it/s]
+
 67%|██████▋   | 26/39 [00:01<00:00, 15.27it/s]
+
 72%|███████▏  | 28/39 [00:01<00:00, 15.26it/s]
+
 77%|███████▋  | 30/39 [00:01<00:00, 15.25it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 15.24it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 15.24it/s]
+
 92%|█████████▏| 36/39 [00:02<00:00, 15.23it/s]
+
 97%|█████████▋| 38/39 [00:02<00:00, 15.23it/s]
                                                    
+
                                               
{'eval_loss': '6.483', 'eval_runtime': '2.636', 'eval_samples_per_second': '926.2', 'eval_steps_per_second': '14.8', 'epoch': '6'}
+
  3%|▎         | 216/7308 [03:32<1:19:04,  1.49it/s]
+
100%|██████████| 39/39 [00:02<00:00, 15.23it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-216
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-216/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-216/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.94it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.93it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-216/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-216/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-216/tokenizer_config.json
+[eval] epoch 6.0 step 216 loss 6.4831 (best 6.4831 @ ep 6.0, 0/20 since best)
+
+
                                               
  3%|▎         | 217/7308 [03:35<4:04:10,  2.07s/it]
  3%|▎         | 218/7308 [03:35<3:13:19,  1.64s/it]
  3%|▎         | 219/7308 [03:36<2:37:44,  1.34s/it]
  3%|▎         | 220/7308 [03:37<2:28:40,  1.26s/it]
                                                    
{'loss': '6.755', 'grad_norm': '0.3219', 'learning_rate': '0.0006', 'epoch': '6.113'}
+
  3%|▎         | 220/7308 [03:37<2:28:40,  1.26s/it]
  3%|▎         | 221/7308 [03:38<2:06:28,  1.07s/it]
  3%|▎         | 222/7308 [03:38<1:50:57,  1.06it/s]
  3%|▎         | 223/7308 [03:39<1:40:05,  1.18it/s]
  3%|▎         | 224/7308 [03:40<1:32:29,  1.28it/s]
  3%|▎         | 225/7308 [03:41<1:43:27,  1.14it/s]
  3%|▎         | 226/7308 [03:41<1:34:48,  1.24it/s]
  3%|▎         | 227/7308 [03:42<1:28:47,  1.33it/s]
  3%|▎         | 228/7308 [03:43<1:24:33,  1.40it/s]
  3%|▎         | 229/7308 [03:43<1:21:36,  1.45it/s]
  3%|▎         | 230/7308 [03:44<1:19:31,  1.48it/s]
                                                    
{'loss': '6.695', 'grad_norm': '0.729', 'learning_rate': '0.0006', 'epoch': '6.396'}
+
  3%|▎         | 230/7308 [03:44<1:19:31,  1.48it/s]
  3%|▎         | 231/7308 [03:44<1:18:03,  1.51it/s]
  3%|▎         | 232/7308 [03:45<1:17:00,  1.53it/s]
  3%|▎         | 233/7308 [03:46<1:16:16,  1.55it/s]
  3%|▎         | 234/7308 [03:46<1:15:45,  1.56it/s]
  3%|▎         | 235/7308 [03:47<1:31:37,  1.29it/s]
  3%|▎         | 236/7308 [03:48<1:26:27,  1.36it/s]
  3%|▎         | 237/7308 [03:49<1:22:53,  1.42it/s]
  3%|▎         | 238/7308 [03:49<1:20:25,  1.47it/s]
  3%|▎         | 239/7308 [03:50<1:18:39,  1.50it/s]
  3%|▎         | 240/7308 [03:51<1:17:25,  1.52it/s]
                                                    
{'loss': '6.652', 'grad_norm': '0.2898', 'learning_rate': '0.0006', 'epoch': '6.678'}
+
  3%|▎         | 240/7308 [03:51<1:17:25,  1.52it/s]
  3%|▎         | 241/7308 [03:51<1:16:36,  1.54it/s]
  3%|▎         | 242/7308 [03:52<1:15:58,  1.55it/s]
  3%|▎         | 243/7308 [03:52<1:15:32,  1.56it/s]
  3%|▎         | 244/7308 [03:53<1:15:18,  1.56it/s]
  3%|▎         | 245/7308 [03:54<1:15:03,  1.57it/s]
  3%|▎         | 246/7308 [03:54<1:14:53,  1.57it/s]
  3%|▎         | 247/7308 [03:55<1:14:46,  1.57it/s]
  3%|▎         | 248/7308 [03:56<1:14:41,  1.58it/s]
  3%|▎         | 249/7308 [03:56<1:14:37,  1.58it/s]
  3%|▎         | 250/7308 [03:57<1:14:36,  1.58it/s]
                                                    
{'loss': '6.578', 'grad_norm': '0.3531', 'learning_rate': '0.0006', 'epoch': '6.961'}
+
  3%|▎         | 250/7308 [03:57<1:14:36,  1.58it/s]
  3%|▎         | 251/7308 [03:58<1:14:31,  1.58it/s]
  3%|▎         | 252/7308 [03:58<1:00:43,  1.94it/s]
  3%|▎         | 253/7308 [03:58<1:07:13,  1.75it/s]
  3%|▎         | 254/7308 [03:59<1:09:22,  1.69it/s]
  3%|▎         | 255/7308 [04:00<1:10:55,  1.66it/s]
  4%|▎         | 256/7308 [04:00<1:11:59,  1.63it/s]
  4%|▎         | 257/7308 [04:01<1:12:45,  1.62it/s]
  4%|▎         | 258/7308 [04:02<1:13:15,  1.60it/s]
  4%|▎         | 259/7308 [04:02<1:13:35,  1.60it/s]
  4%|▎         | 260/7308 [04:03<1:13:50,  1.59it/s]
                                                    
{'loss': '6.493', 'grad_norm': '0.4957', 'learning_rate': '0.0006', 'epoch': '7.226'}
+
  4%|▎         | 260/7308 [04:03<1:13:50,  1.59it/s]
  4%|▎         | 261/7308 [04:04<1:14:01,  1.59it/s]
  4%|▎         | 262/7308 [04:04<1:14:07,  1.58it/s]
  4%|▎         | 263/7308 [04:05<1:14:14,  1.58it/s]
  4%|▎         | 264/7308 [04:05<1:14:17,  1.58it/s]
  4%|▎         | 265/7308 [04:06<1:14:18,  1.58it/s]
  4%|▎         | 266/7308 [04:07<1:14:20,  1.58it/s]
  4%|▎         | 267/7308 [04:07<1:14:22,  1.58it/s]
  4%|▎         | 268/7308 [04:08<1:14:21,  1.58it/s]
  4%|▎         | 269/7308 [04:09<1:14:21,  1.58it/s]
  4%|▎         | 270/7308 [04:09<1:14:19,  1.58it/s]
                                                    
{'loss': '6.414', 'grad_norm': '0.3957', 'learning_rate': '0.0006', 'epoch': '7.509'}
+
  4%|▎         | 270/7308 [04:09<1:14:19,  1.58it/s]
  4%|▎         | 271/7308 [04:10<1:14:20,  1.58it/s]
  4%|▎         | 272/7308 [04:11<1:14:21,  1.58it/s]
  4%|▎         | 273/7308 [04:11<1:14:19,  1.58it/s]
  4%|▎         | 274/7308 [04:12<1:14:16,  1.58it/s]
  4%|▍         | 275/7308 [04:12<1:14:12,  1.58it/s]
  4%|▍         | 276/7308 [04:13<1:14:11,  1.58it/s]
  4%|▍         | 277/7308 [04:14<1:14:20,  1.58it/s]
  4%|▍         | 278/7308 [04:14<1:14:20,  1.58it/s]
  4%|▍         | 279/7308 [04:15<1:14:19,  1.58it/s]
  4%|▍         | 280/7308 [04:16<1:14:18,  1.58it/s]
                                                    
{'loss': '6.341', 'grad_norm': '0.9383', 'learning_rate': '0.0006', 'epoch': '7.792'}
+
  4%|▍         | 280/7308 [04:16<1:14:18,  1.58it/s]
  4%|▍         | 281/7308 [04:16<1:14:21,  1.58it/s]
  4%|▍         | 282/7308 [04:17<1:14:23,  1.57it/s]
  4%|▍         | 283/7308 [04:18<1:14:22,  1.57it/s]
  4%|▍         | 284/7308 [04:18<1:14:19,  1.58it/s]
  4%|▍         | 285/7308 [04:19<1:14:19,  1.57it/s]
  4%|▍         | 286/7308 [04:19<1:14:18,  1.58it/s]
  4%|▍         | 287/7308 [04:20<1:14:11,  1.58it/s]
  4%|▍         | 288/7308 [04:20<1:00:26,  1.94it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 23.15it/s]
+
 15%|█▌        | 6/39 [00:00<00:01, 17.67it/s]
+
 21%|██        | 8/39 [00:00<00:01, 16.70it/s]
+
 26%|██▌       | 10/39 [00:00<00:01, 16.14it/s]
+
 31%|███       | 12/39 [00:00<00:01, 15.82it/s]
+
 36%|███▌      | 14/39 [00:00<00:01, 15.59it/s]
+
 41%|████      | 16/39 [00:00<00:01, 15.47it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 15.36it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 15.32it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 15.26it/s]
+
 62%|██████▏   | 24/39 [00:01<00:00, 15.26it/s]
+
 67%|██████▋   | 26/39 [00:01<00:00, 15.25it/s]
+
 72%|███████▏  | 28/39 [00:01<00:00, 15.25it/s]
+
 77%|███████▋  | 30/39 [00:01<00:00, 15.24it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 15.23it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 15.22it/s]
+
 92%|█████████▏| 36/39 [00:02<00:00, 15.22it/s]
+
 97%|█████████▋| 38/39 [00:02<00:00, 15.22it/s]
                                                    
+
                                               
{'eval_loss': '5.963', 'eval_runtime': '2.638', 'eval_samples_per_second': '925.2', 'eval_steps_per_second': '14.78', 'epoch': '8'}
+
  4%|▍         | 288/7308 [04:23<1:00:26,  1.94it/s]
+
100%|██████████| 39/39 [00:02<00:00, 15.22it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-288
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-288/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-288/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.96it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.96it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-288/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-288/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-288/tokenizer_config.json
+[eval] epoch 8.0 step 288 loss 5.9631 (best 5.9631 @ ep 8.0, 0/20 since best)
+
+
                                               
  4%|▍         | 289/7308 [04:26<3:49:38,  1.96s/it]
  4%|▍         | 290/7308 [04:26<3:02:58,  1.56s/it]
                                                    
{'loss': '6.305', 'grad_norm': '0.3299', 'learning_rate': '0.0006', 'epoch': '8.057'}
+
  4%|▍         | 290/7308 [04:26<3:02:58,  1.56s/it]
  4%|▍         | 291/7308 [04:27<2:30:19,  1.29s/it]
  4%|▍         | 292/7308 [04:28<2:07:27,  1.09s/it]
  4%|▍         | 293/7308 [04:28<1:51:26,  1.05it/s]
  4%|▍         | 294/7308 [04:29<1:40:13,  1.17it/s]
  4%|▍         | 295/7308 [04:29<1:32:24,  1.26it/s]
  4%|▍         | 296/7308 [04:30<1:26:56,  1.34it/s]
  4%|▍         | 297/7308 [04:31<1:23:05,  1.41it/s]
  4%|▍         | 298/7308 [04:31<1:20:25,  1.45it/s]
  4%|▍         | 299/7308 [04:32<1:18:35,  1.49it/s]
  4%|▍         | 300/7308 [04:33<1:17:11,  1.51it/s]
                                                    
{'loss': '6.27', 'grad_norm': '0.3679', 'learning_rate': '0.0006', 'epoch': '8.339'}
+
  4%|▍         | 300/7308 [04:33<1:17:11,  1.51it/s]
  4%|▍         | 301/7308 [04:33<1:16:13,  1.53it/s]
  4%|▍         | 302/7308 [04:34<1:15:32,  1.55it/s]
  4%|▍         | 303/7308 [04:35<1:15:03,  1.56it/s]
  4%|▍         | 304/7308 [04:35<1:14:41,  1.56it/s]
  4%|▍         | 305/7308 [04:36<1:14:26,  1.57it/s]
  4%|▍         | 306/7308 [04:36<1:14:16,  1.57it/s]
  4%|▍         | 307/7308 [04:37<1:14:11,  1.57it/s]
  4%|▍         | 308/7308 [04:38<1:14:04,  1.58it/s]
  4%|▍         | 309/7308 [04:38<1:13:58,  1.58it/s]
  4%|▍         | 310/7308 [04:39<1:13:56,  1.58it/s]
                                                    
{'loss': '6.206', 'grad_norm': '0.3086', 'learning_rate': '0.0006', 'epoch': '8.622'}
+
  4%|▍         | 310/7308 [04:39<1:13:56,  1.58it/s]
  4%|▍         | 311/7308 [04:40<1:13:55,  1.58it/s]
  4%|▍         | 312/7308 [04:40<1:13:52,  1.58it/s]
  4%|▍         | 313/7308 [04:41<1:13:51,  1.58it/s]
  4%|▍         | 314/7308 [04:41<1:13:52,  1.58it/s]
  4%|▍         | 315/7308 [04:42<1:13:50,  1.58it/s]
  4%|▍         | 316/7308 [04:43<1:13:52,  1.58it/s]
  4%|▍         | 317/7308 [04:43<1:13:51,  1.58it/s]
  4%|▍         | 318/7308 [04:44<1:13:51,  1.58it/s]
  4%|▍         | 319/7308 [04:45<1:13:50,  1.58it/s]
  4%|▍         | 320/7308 [04:45<1:13:49,  1.58it/s]
                                                    
{'loss': '6.159', 'grad_norm': '0.508', 'learning_rate': '0.0006', 'epoch': '8.905'}
+
  4%|▍         | 320/7308 [04:45<1:13:49,  1.58it/s]
  4%|▍         | 321/7308 [04:46<1:13:49,  1.58it/s]
  4%|▍         | 322/7308 [04:47<1:13:52,  1.58it/s]
  4%|▍         | 323/7308 [04:47<1:13:47,  1.58it/s]
  4%|▍         | 324/7308 [04:47<1:00:08,  1.94it/s]
  4%|▍         | 325/7308 [04:48<1:07:29,  1.72it/s]
  4%|▍         | 326/7308 [04:49<1:09:21,  1.68it/s]
  4%|▍         | 327/7308 [04:49<1:10:39,  1.65it/s]
  4%|▍         | 328/7308 [04:50<1:11:32,  1.63it/s]
  5%|▍         | 329/7308 [04:51<1:12:10,  1.61it/s]
  5%|▍         | 330/7308 [04:51<1:12:44,  1.60it/s]
                                                    
{'loss': '6.152', 'grad_norm': '0.3932', 'learning_rate': '0.0006', 'epoch': '9.17'}
+
  5%|▍         | 330/7308 [04:51<1:12:44,  1.60it/s]
  5%|▍         | 331/7308 [04:52<1:13:05,  1.59it/s]
  5%|▍         | 332/7308 [04:53<1:13:16,  1.59it/s]
  5%|▍         | 333/7308 [04:53<1:13:23,  1.58it/s]
  5%|▍         | 334/7308 [04:54<1:13:27,  1.58it/s]
  5%|▍         | 335/7308 [04:54<1:13:30,  1.58it/s]
  5%|▍         | 336/7308 [04:55<1:13:32,  1.58it/s]
  5%|▍         | 337/7308 [04:56<1:13:33,  1.58it/s]
  5%|▍         | 338/7308 [04:56<1:13:38,  1.58it/s]
  5%|▍         | 339/7308 [04:57<1:13:37,  1.58it/s]
  5%|▍         | 340/7308 [04:58<1:13:36,  1.58it/s]
                                                    
{'loss': '6.12', 'grad_norm': '0.4371', 'learning_rate': '0.0006', 'epoch': '9.452'}
+
  5%|▍         | 340/7308 [04:58<1:13:36,  1.58it/s]
  5%|▍         | 341/7308 [04:58<1:13:36,  1.58it/s]
  5%|▍         | 342/7308 [04:59<1:13:34,  1.58it/s]
  5%|▍         | 343/7308 [05:00<1:13:35,  1.58it/s]
  5%|▍         | 344/7308 [05:00<1:13:32,  1.58it/s]
  5%|▍         | 345/7308 [05:01<1:13:32,  1.58it/s]
  5%|▍         | 346/7308 [05:01<1:13:31,  1.58it/s]
  5%|▍         | 347/7308 [05:02<1:13:35,  1.58it/s]
  5%|▍         | 348/7308 [05:03<1:13:35,  1.58it/s]
  5%|▍         | 349/7308 [05:03<1:13:33,  1.58it/s]
  5%|▍         | 350/7308 [05:04<1:13:32,  1.58it/s]
                                                    
{'loss': '6.071', 'grad_norm': '0.3206', 'learning_rate': '0.0006', 'epoch': '9.735'}
+
  5%|▍         | 350/7308 [05:04<1:13:32,  1.58it/s]
  5%|▍         | 351/7308 [05:05<1:13:31,  1.58it/s]
  5%|▍         | 352/7308 [05:05<1:13:31,  1.58it/s]
  5%|▍         | 353/7308 [05:06<1:13:29,  1.58it/s]
  5%|▍         | 354/7308 [05:07<1:13:30,  1.58it/s]
  5%|▍         | 355/7308 [05:07<1:13:29,  1.58it/s]
  5%|▍         | 356/7308 [05:08<1:13:29,  1.58it/s]
  5%|▍         | 357/7308 [05:08<1:13:28,  1.58it/s]
  5%|▍         | 358/7308 [05:09<1:13:30,  1.58it/s]
  5%|▍         | 359/7308 [05:10<1:13:25,  1.58it/s]
  5%|▍         | 360/7308 [05:10<59:50,  1.94it/s]  
                                                  
{'loss': '6.019', 'grad_norm': '0.3635', 'learning_rate': '0.0006', 'epoch': '10'}
+
  5%|▍         | 360/7308 [05:10<59:50,  1.94it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 23.47it/s]
+
 15%|█▌        | 6/39 [00:00<00:01, 17.74it/s]
+
 21%|██        | 8/39 [00:00<00:01, 16.73it/s]
+
 26%|██▌       | 10/39 [00:00<00:01, 16.17it/s]
+
 31%|███       | 12/39 [00:00<00:01, 15.82it/s]
+
 36%|███▌      | 14/39 [00:00<00:01, 15.58it/s]
+
 41%|████      | 16/39 [00:00<00:01, 15.46it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 15.23it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 15.21it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 15.18it/s]
+
 62%|██████▏   | 24/39 [00:01<00:00, 15.21it/s]
+
 67%|██████▋   | 26/39 [00:01<00:00, 15.21it/s]
+
 72%|███████▏  | 28/39 [00:01<00:00, 15.22it/s]
+
 77%|███████▋  | 30/39 [00:01<00:00, 15.22it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 15.19it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 15.20it/s]
+
 92%|█████████▏| 36/39 [00:02<00:00, 15.21it/s]
+
 97%|█████████▋| 38/39 [00:02<00:00, 15.21it/s]
                                                  
+
                                               
{'eval_loss': '5.581', 'eval_runtime': '2.643', 'eval_samples_per_second': '923.7', 'eval_steps_per_second': '14.76', 'epoch': '10'}
+
  5%|▍         | 360/7308 [05:13<59:50,  1.94it/s]
+
100%|██████████| 39/39 [00:02<00:00, 15.21it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-360
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-360/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-360/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.95it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.95it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-360/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-360/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-360/tokenizer_config.json
+[eval] epoch 10.0 step 360 loss 5.5808 (best 5.5808 @ ep 10.0, 0/20 since best)
+
+
                                               
  5%|▍         | 361/7308 [05:15<3:47:13,  1.96s/it]
  5%|▍         | 362/7308 [05:16<3:01:03,  1.56s/it]
  5%|▍         | 363/7308 [05:17<2:28:44,  1.28s/it]
  5%|▍         | 364/7308 [05:17<2:06:06,  1.09s/it]
  5%|▍         | 365/7308 [05:18<1:50:16,  1.05it/s]
  5%|▌         | 366/7308 [05:18<1:39:10,  1.17it/s]
  5%|▌         | 367/7308 [05:19<1:31:24,  1.27it/s]
  5%|▌         | 368/7308 [05:20<1:25:57,  1.35it/s]
  5%|▌         | 369/7308 [05:20<1:22:09,  1.41it/s]
  5%|▌         | 370/7308 [05:21<1:19:28,  1.45it/s]
                                                    
{'loss': '5.905', 'grad_norm': '0.3551', 'learning_rate': '0.0006', 'epoch': '10.28'}
+
  5%|▌         | 370/7308 [05:21<1:19:28,  1.45it/s]
  5%|▌         | 371/7308 [05:22<1:17:38,  1.49it/s]
  5%|▌         | 372/7308 [05:22<1:16:20,  1.51it/s]
  5%|▌         | 373/7308 [05:23<1:15:27,  1.53it/s]
  5%|▌         | 374/7308 [05:24<1:14:46,  1.55it/s]
  5%|▌         | 375/7308 [05:24<1:14:18,  1.55it/s]
  5%|▌         | 376/7308 [05:25<1:13:58,  1.56it/s]
  5%|▌         | 377/7308 [05:25<1:13:44,  1.57it/s]
  5%|▌         | 378/7308 [05:26<1:13:35,  1.57it/s]
  5%|▌         | 379/7308 [05:27<1:13:30,  1.57it/s]
  5%|▌         | 380/7308 [05:27<1:13:25,  1.57it/s]
                                                    
{'loss': '5.92', 'grad_norm': '0.4127', 'learning_rate': '0.0006', 'epoch': '10.57'}
+
  5%|▌         | 380/7308 [05:27<1:13:25,  1.57it/s]
  5%|▌         | 381/7308 [05:28<1:13:21,  1.57it/s]
  5%|▌         | 382/7308 [05:29<1:13:17,  1.57it/s]
  5%|▌         | 383/7308 [05:29<1:13:13,  1.58it/s]
  5%|▌         | 384/7308 [05:30<1:13:10,  1.58it/s]
  5%|▌         | 385/7308 [05:31<1:13:09,  1.58it/s]
  5%|▌         | 386/7308 [05:31<1:13:09,  1.58it/s]
  5%|▌         | 387/7308 [05:32<1:13:07,  1.58it/s]
  5%|▌         | 388/7308 [05:32<1:13:05,  1.58it/s]
  5%|▌         | 389/7308 [05:33<1:13:07,  1.58it/s]
  5%|▌         | 390/7308 [05:34<1:13:05,  1.58it/s]
                                                    
{'loss': '5.901', 'grad_norm': '0.4268', 'learning_rate': '0.0006', 'epoch': '10.85'}
+
  5%|▌         | 390/7308 [05:34<1:13:05,  1.58it/s]
  5%|▌         | 391/7308 [05:34<1:13:05,  1.58it/s]
  5%|▌         | 392/7308 [05:35<1:13:03,  1.58it/s]
  5%|▌         | 393/7308 [05:36<1:13:01,  1.58it/s]
  5%|▌         | 394/7308 [05:36<1:13:02,  1.58it/s]
  5%|▌         | 395/7308 [05:37<1:12:59,  1.58it/s]
  5%|▌         | 396/7308 [05:37<59:27,  1.94it/s]  
  5%|▌         | 397/7308 [05:38<1:06:54,  1.72it/s]
  5%|▌         | 398/7308 [05:38<1:08:42,  1.68it/s]
  5%|▌         | 399/7308 [05:39<1:09:58,  1.65it/s]
  5%|▌         | 400/7308 [05:40<1:10:51,  1.62it/s]
                                                    
{'loss': '5.84', 'grad_norm': '0.3291', 'learning_rate': '0.0006', 'epoch': '11.11'}
+
  5%|▌         | 400/7308 [05:40<1:10:51,  1.62it/s]
  5%|▌         | 401/7308 [05:40<1:11:28,  1.61it/s]
  6%|▌         | 402/7308 [05:41<1:11:53,  1.60it/s]
  6%|▌         | 403/7308 [05:42<1:12:10,  1.59it/s]
  6%|▌         | 404/7308 [05:42<1:12:22,  1.59it/s]
  6%|▌         | 405/7308 [05:43<1:12:31,  1.59it/s]
  6%|▌         | 406/7308 [05:44<1:12:38,  1.58it/s]
  6%|▌         | 407/7308 [05:44<1:12:42,  1.58it/s]
  6%|▌         | 408/7308 [05:45<1:12:44,  1.58it/s]
  6%|▌         | 409/7308 [05:45<1:12:44,  1.58it/s]
  6%|▌         | 410/7308 [05:46<1:12:45,  1.58it/s]
                                                    
{'loss': '5.741', 'grad_norm': '0.2854', 'learning_rate': '0.0006', 'epoch': '11.4'}
+
  6%|▌         | 410/7308 [05:46<1:12:45,  1.58it/s]
  6%|▌         | 411/7308 [05:47<1:12:48,  1.58it/s]
  6%|▌         | 412/7308 [05:47<1:12:47,  1.58it/s]
  6%|▌         | 413/7308 [05:48<1:12:47,  1.58it/s]
  6%|▌         | 414/7308 [05:49<1:12:47,  1.58it/s]
  6%|▌         | 415/7308 [05:49<1:12:45,  1.58it/s]
  6%|▌         | 416/7308 [05:50<1:12:44,  1.58it/s]
  6%|▌         | 417/7308 [05:50<1:12:43,  1.58it/s]
  6%|▌         | 418/7308 [05:51<1:12:42,  1.58it/s]
  6%|▌         | 419/7308 [05:52<1:12:42,  1.58it/s]
  6%|▌         | 420/7308 [05:52<1:12:42,  1.58it/s]
                                                    
{'loss': '5.709', 'grad_norm': '0.2584', 'learning_rate': '0.0006', 'epoch': '11.68'}
+
  6%|▌         | 420/7308 [05:52<1:12:42,  1.58it/s]
  6%|▌         | 421/7308 [05:53<1:12:44,  1.58it/s]
  6%|▌         | 422/7308 [05:54<1:12:47,  1.58it/s]
  6%|▌         | 423/7308 [05:54<1:12:45,  1.58it/s]
  6%|▌         | 424/7308 [05:55<1:12:43,  1.58it/s]
  6%|▌         | 425/7308 [05:56<1:12:42,  1.58it/s]
  6%|▌         | 426/7308 [05:56<1:12:45,  1.58it/s]
  6%|▌         | 427/7308 [05:57<1:12:42,  1.58it/s]
  6%|▌         | 428/7308 [05:57<1:12:41,  1.58it/s]
  6%|▌         | 429/7308 [05:58<1:12:40,  1.58it/s]
  6%|▌         | 430/7308 [05:59<1:12:39,  1.58it/s]
                                                    
{'loss': '5.726', 'grad_norm': '0.4824', 'learning_rate': '0.0006', 'epoch': '11.96'}
+
  6%|▌         | 430/7308 [05:59<1:12:39,  1.58it/s]
  6%|▌         | 431/7308 [05:59<1:12:36,  1.58it/s]
  6%|▌         | 432/7308 [06:00<59:10,  1.94it/s]  [transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 22.73it/s]
+
 15%|█▌        | 6/39 [00:00<00:01, 17.61it/s]
+
 21%|██        | 8/39 [00:00<00:01, 16.65it/s]
+
 26%|██▌       | 10/39 [00:00<00:01, 16.14it/s]
+
 31%|███       | 12/39 [00:00<00:01, 15.81it/s]
+
 36%|███▌      | 14/39 [00:00<00:01, 15.62it/s]
+
 41%|████      | 16/39 [00:00<00:01, 15.47it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 15.40it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 15.33it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 15.30it/s]
+
 62%|██████▏   | 24/39 [00:01<00:00, 15.29it/s]
+
 67%|██████▋   | 26/39 [00:01<00:00, 15.28it/s]
+
 72%|███████▏  | 28/39 [00:01<00:00, 15.26it/s]
+
 77%|███████▋  | 30/39 [00:01<00:00, 15.25it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 15.23it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 15.24it/s]
+
 92%|█████████▏| 36/39 [00:02<00:00, 15.23it/s]
+
 97%|█████████▋| 38/39 [00:02<00:00, 15.23it/s]
                                                  
+
                                               
{'eval_loss': '5.296', 'eval_runtime': '2.634', 'eval_samples_per_second': '926.8', 'eval_steps_per_second': '14.81', 'epoch': '12'}
+
  6%|▌         | 432/7308 [06:02<59:10,  1.94it/s]
+
100%|██████████| 39/39 [00:02<00:00, 15.23it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-432
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-432/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-432/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.73it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.73it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-432/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-432/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-432/tokenizer_config.json
+[eval] epoch 12.0 step 432 loss 5.2958 (best 5.2958 @ ep 12.0, 0/20 since best)
+
+
                                               
  6%|▌         | 433/7308 [06:05<3:47:17,  1.98s/it]
  6%|▌         | 434/7308 [06:06<3:00:50,  1.58s/it]
  6%|▌         | 435/7308 [06:06<2:28:20,  1.29s/it]
  6%|▌         | 436/7308 [06:07<2:05:35,  1.10s/it]
  6%|▌         | 437/7308 [06:08<1:49:39,  1.04it/s]
  6%|▌         | 438/7308 [06:08<1:38:30,  1.16it/s]
  6%|▌         | 439/7308 [06:09<1:30:43,  1.26it/s]
  6%|▌         | 440/7308 [06:09<1:25:13,  1.34it/s]
                                                    
{'loss': '5.723', 'grad_norm': '0.4441', 'learning_rate': '0.0006', 'epoch': '12.23'}
+
  6%|▌         | 440/7308 [06:09<1:25:13,  1.34it/s]
  6%|▌         | 441/7308 [06:10<1:21:25,  1.41it/s]
  6%|▌         | 442/7308 [06:11<1:18:44,  1.45it/s]
  6%|▌         | 443/7308 [06:11<1:16:50,  1.49it/s]
  6%|▌         | 444/7308 [06:12<1:15:31,  1.51it/s]
  6%|▌         | 445/7308 [06:13<1:14:38,  1.53it/s]
  6%|▌         | 446/7308 [06:13<1:13:58,  1.55it/s]
  6%|▌         | 447/7308 [06:14<1:13:33,  1.55it/s]
  6%|▌         | 448/7308 [06:15<1:13:15,  1.56it/s]
  6%|▌         | 449/7308 [06:15<1:13:01,  1.57it/s]
  6%|▌         | 450/7308 [06:16<1:12:52,  1.57it/s]
                                                    
{'loss': '5.718', 'grad_norm': '0.2025', 'learning_rate': '0.0006', 'epoch': '12.51'}
+
  6%|▌         | 450/7308 [06:16<1:12:52,  1.57it/s]
  6%|▌         | 451/7308 [06:16<1:12:49,  1.57it/s]
  6%|▌         | 452/7308 [06:17<1:12:42,  1.57it/s]
  6%|▌         | 453/7308 [06:18<1:12:38,  1.57it/s]
  6%|▌         | 454/7308 [06:18<1:12:34,  1.57it/s]
  6%|▌         | 455/7308 [06:19<1:12:31,  1.57it/s]
  6%|▌         | 456/7308 [06:20<1:12:30,  1.57it/s]
  6%|▋         | 457/7308 [06:20<1:12:27,  1.58it/s]
  6%|▋         | 458/7308 [06:21<1:12:24,  1.58it/s]
  6%|▋         | 459/7308 [06:21<1:12:23,  1.58it/s]
  6%|▋         | 460/7308 [06:22<1:12:21,  1.58it/s]
                                                    
{'loss': '5.621', 'grad_norm': '0.2037', 'learning_rate': '0.0006', 'epoch': '12.79'}
+
  6%|▋         | 460/7308 [06:22<1:12:21,  1.58it/s]
  6%|▋         | 461/7308 [06:23<1:12:23,  1.58it/s]
  6%|▋         | 462/7308 [06:23<1:12:21,  1.58it/s]
  6%|▋         | 463/7308 [06:24<1:12:21,  1.58it/s]
  6%|▋         | 464/7308 [06:25<1:12:21,  1.58it/s]
  6%|▋         | 465/7308 [06:25<1:12:19,  1.58it/s]
  6%|▋         | 466/7308 [06:26<1:12:19,  1.58it/s]
  6%|▋         | 467/7308 [06:27<1:12:16,  1.58it/s]
  6%|▋         | 468/7308 [06:27<58:53,  1.94it/s]  
  6%|▋         | 469/7308 [06:28<1:06:11,  1.72it/s]
  6%|▋         | 470/7308 [06:28<1:08:00,  1.68it/s]
                                                    
{'loss': '5.589', 'grad_norm': '0.766', 'learning_rate': '0.0006', 'epoch': '13.06'}
+
  6%|▋         | 470/7308 [06:28<1:08:00,  1.68it/s]
  6%|▋         | 471/7308 [06:29<1:09:19,  1.64it/s]
  6%|▋         | 472/7308 [06:29<1:10:10,  1.62it/s]
  6%|▋         | 473/7308 [06:30<1:10:45,  1.61it/s]
  6%|▋         | 474/7308 [06:31<1:11:09,  1.60it/s]
  6%|▋         | 475/7308 [06:31<1:11:27,  1.59it/s]
  7%|▋         | 476/7308 [06:32<1:11:39,  1.59it/s]
  7%|▋         | 477/7308 [06:33<1:11:47,  1.59it/s]
  7%|▋         | 478/7308 [06:33<1:11:53,  1.58it/s]
  7%|▋         | 479/7308 [06:34<1:11:56,  1.58it/s]
  7%|▋         | 480/7308 [06:35<1:12:01,  1.58it/s]
                                                    
{'loss': '5.61', 'grad_norm': '0.2937', 'learning_rate': '0.0006', 'epoch': '13.34'}
+
  7%|▋         | 480/7308 [06:35<1:12:01,  1.58it/s]
  7%|▋         | 481/7308 [06:35<1:12:04,  1.58it/s]
  7%|▋         | 482/7308 [06:36<1:12:03,  1.58it/s]
  7%|▋         | 483/7308 [06:36<1:12:05,  1.58it/s]
  7%|▋         | 484/7308 [06:37<1:12:04,  1.58it/s]
  7%|▋         | 485/7308 [06:38<1:12:03,  1.58it/s]
  7%|▋         | 486/7308 [06:38<1:12:01,  1.58it/s]
  7%|▋         | 487/7308 [06:39<1:12:02,  1.58it/s]
  7%|▋         | 488/7308 [06:40<1:12:00,  1.58it/s]
  7%|▋         | 489/7308 [06:40<1:12:00,  1.58it/s]
  7%|▋         | 490/7308 [06:41<1:12:00,  1.58it/s]
                                                    
{'loss': '5.542', 'grad_norm': '0.2134', 'learning_rate': '0.0006', 'epoch': '13.62'}
+
  7%|▋         | 490/7308 [06:41<1:12:00,  1.58it/s]
  7%|▋         | 491/7308 [06:41<1:12:00,  1.58it/s]
  7%|▋         | 492/7308 [06:42<1:11:59,  1.58it/s]
  7%|▋         | 493/7308 [06:43<1:11:58,  1.58it/s]
  7%|▋         | 494/7308 [06:43<1:11:59,  1.58it/s]
  7%|▋         | 495/7308 [06:44<1:11:56,  1.58it/s]
  7%|▋         | 496/7308 [06:45<1:11:57,  1.58it/s]
  7%|▋         | 497/7308 [06:45<1:11:57,  1.58it/s]
  7%|▋         | 498/7308 [06:46<1:11:56,  1.58it/s]
  7%|▋         | 499/7308 [06:47<1:11:59,  1.58it/s]
  7%|▋         | 500/7308 [06:47<1:11:59,  1.58it/s]
                                                    
{'loss': '5.512', 'grad_norm': '0.2347', 'learning_rate': '0.0006', 'epoch': '13.9'}
+
  7%|▋         | 500/7308 [06:47<1:11:59,  1.58it/s]
  7%|▋         | 501/7308 [06:48<1:11:56,  1.58it/s]
  7%|▋         | 502/7308 [06:48<1:11:55,  1.58it/s]
  7%|▋         | 503/7308 [06:49<1:11:52,  1.58it/s]
  7%|▋         | 504/7308 [06:49<58:34,  1.94it/s]  [transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 22.77it/s]
+
 15%|█▌        | 6/39 [00:00<00:01, 17.59it/s]
+
 21%|██        | 8/39 [00:00<00:01, 16.66it/s]
+
 26%|██▌       | 10/39 [00:00<00:01, 16.12it/s]
+
 31%|███       | 12/39 [00:00<00:01, 15.82it/s]
+
 36%|███▌      | 14/39 [00:00<00:01, 15.61it/s]
+
 41%|████      | 16/39 [00:00<00:01, 15.49it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 15.38it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 15.33it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 15.28it/s]
+
 62%|██████▏   | 24/39 [00:01<00:00, 15.28it/s]
+
 67%|██████▋   | 26/39 [00:01<00:00, 15.26it/s]
+
 72%|███████▏  | 28/39 [00:01<00:00, 15.26it/s]
+
 77%|███████▋  | 30/39 [00:01<00:00, 15.23it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 15.24it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 15.21it/s]
+
 92%|█████████▏| 36/39 [00:02<00:00, 15.23it/s]
+
 97%|█████████▋| 38/39 [00:02<00:00, 15.22it/s]
                                                  
+
                                               
{'eval_loss': '5.028', 'eval_runtime': '2.636', 'eval_samples_per_second': '926.1', 'eval_steps_per_second': '14.8', 'epoch': '14'}
+
  7%|▋         | 504/7308 [06:52<58:34,  1.94it/s]
+
100%|██████████| 39/39 [00:02<00:00, 15.22it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-504
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-504/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-504/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  2.11it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  2.11it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-504/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-504/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-504/tokenizer_config.json
+[eval] epoch 14.0 step 504 loss 5.0283 (best 5.0283 @ ep 14.0, 0/20 since best)
+
+
                                               
  7%|▋         | 505/7308 [06:55<3:41:23,  1.95s/it]
  7%|▋         | 506/7308 [06:55<2:56:29,  1.56s/it]
  7%|▋         | 507/7308 [06:56<2:25:03,  1.28s/it]
  7%|▋         | 508/7308 [06:57<2:03:03,  1.09s/it]
  7%|▋         | 509/7308 [06:57<1:47:39,  1.05it/s]
  7%|▋         | 510/7308 [06:58<1:36:52,  1.17it/s]
                                                    
{'loss': '5.54', 'grad_norm': '0.5525', 'learning_rate': '0.0006', 'epoch': '14.17'}
+
  7%|▋         | 510/7308 [06:58<1:36:52,  1.17it/s]
  7%|▋         | 511/7308 [06:58<1:29:20,  1.27it/s]
  7%|▋         | 512/7308 [06:59<1:24:03,  1.35it/s]
  7%|▋         | 513/7308 [07:00<1:20:20,  1.41it/s]
  7%|▋         | 514/7308 [07:00<1:17:43,  1.46it/s]
  7%|▋         | 515/7308 [07:01<1:15:54,  1.49it/s]
  7%|▋         | 516/7308 [07:02<1:14:38,  1.52it/s]
  7%|▋         | 517/7308 [07:02<1:13:43,  1.54it/s]
  7%|▋         | 518/7308 [07:03<1:13:06,  1.55it/s]
  7%|▋         | 519/7308 [07:04<1:12:39,  1.56it/s]
  7%|▋         | 520/7308 [07:04<1:12:21,  1.56it/s]
                                                    
{'loss': '5.44', 'grad_norm': '0.4901', 'learning_rate': '0.0006', 'epoch': '14.45'}
+
  7%|▋         | 520/7308 [07:04<1:12:21,  1.56it/s]
  7%|▋         | 521/7308 [07:05<1:12:11,  1.57it/s]
  7%|▋         | 522/7308 [07:05<1:12:00,  1.57it/s]
  7%|▋         | 523/7308 [07:06<1:11:54,  1.57it/s]
  7%|▋         | 524/7308 [07:07<1:11:51,  1.57it/s]
  7%|▋         | 525/7308 [07:07<1:11:47,  1.57it/s]
  7%|▋         | 526/7308 [07:08<1:11:43,  1.58it/s]
  7%|▋         | 527/7308 [07:09<1:11:41,  1.58it/s]
  7%|▋         | 528/7308 [07:09<1:11:39,  1.58it/s]
  7%|▋         | 529/7308 [07:10<1:11:38,  1.58it/s]
  7%|▋         | 530/7308 [07:10<1:11:37,  1.58it/s]
                                                    
{'loss': '5.469', 'grad_norm': '0.2301', 'learning_rate': '0.0006', 'epoch': '14.73'}
+
  7%|▋         | 530/7308 [07:10<1:11:37,  1.58it/s]
  7%|▋         | 531/7308 [07:11<1:11:37,  1.58it/s]
  7%|▋         | 532/7308 [07:12<1:11:34,  1.58it/s]
  7%|▋         | 533/7308 [07:12<1:11:33,  1.58it/s]
  7%|▋         | 534/7308 [07:13<1:11:34,  1.58it/s]
  7%|▋         | 535/7308 [07:14<1:11:32,  1.58it/s]
  7%|▋         | 536/7308 [07:14<1:11:31,  1.58it/s]
  7%|▋         | 537/7308 [07:15<1:11:33,  1.58it/s]
  7%|▋         | 538/7308 [07:16<1:11:32,  1.58it/s]
  7%|▋         | 539/7308 [07:16<1:11:29,  1.58it/s]
  7%|▋         | 540/7308 [07:16<58:17,  1.94it/s]  
                                                  
{'loss': '5.425', 'grad_norm': '0.3611', 'learning_rate': '0.0006', 'epoch': '15'}
+
  7%|▋         | 540/7308 [07:16<58:17,  1.94it/s]
  7%|▋         | 541/7308 [07:17<1:04:38,  1.74it/s]
  7%|▋         | 542/7308 [07:18<1:06:39,  1.69it/s]
  7%|▋         | 543/7308 [07:18<1:08:04,  1.66it/s]
  7%|▋         | 544/7308 [07:19<1:09:04,  1.63it/s]
  7%|▋         | 545/7308 [07:20<1:09:45,  1.62it/s]
  7%|▋         | 546/7308 [07:20<1:10:15,  1.60it/s]
  7%|▋         | 547/7308 [07:21<1:10:35,  1.60it/s]
  7%|▋         | 548/7308 [07:22<1:10:49,  1.59it/s]
  8%|▊         | 549/7308 [07:22<1:10:58,  1.59it/s]
  8%|▊         | 550/7308 [07:23<1:11:06,  1.58it/s]
                                                    
{'loss': '5.293', 'grad_norm': '0.2622', 'learning_rate': '0.0006', 'epoch': '15.28'}
+
  8%|▊         | 550/7308 [07:23<1:11:06,  1.58it/s]
  8%|▊         | 551/7308 [07:23<1:11:11,  1.58it/s]
  8%|▊         | 552/7308 [07:24<1:11:13,  1.58it/s]
  8%|▊         | 553/7308 [07:25<1:11:15,  1.58it/s]
  8%|▊         | 554/7308 [07:25<1:11:17,  1.58it/s]
  8%|▊         | 555/7308 [07:26<1:11:17,  1.58it/s]
  8%|▊         | 556/7308 [07:27<1:11:18,  1.58it/s]
  8%|▊         | 557/7308 [07:27<1:11:18,  1.58it/s]
  8%|▊         | 558/7308 [07:28<1:11:16,  1.58it/s]
  8%|▊         | 559/7308 [07:29<1:11:16,  1.58it/s]
  8%|▊         | 560/7308 [07:29<1:11:17,  1.58it/s]
                                                    
{'loss': '5.256', 'grad_norm': '0.3595', 'learning_rate': '0.0006', 'epoch': '15.57'}
+
  8%|▊         | 560/7308 [07:29<1:11:17,  1.58it/s]
  8%|▊         | 561/7308 [07:30<1:11:18,  1.58it/s]
  8%|▊         | 562/7308 [07:30<1:11:19,  1.58it/s]
  8%|▊         | 563/7308 [07:31<1:11:19,  1.58it/s]
  8%|▊         | 564/7308 [07:32<1:11:17,  1.58it/s]
  8%|▊         | 565/7308 [07:32<1:11:16,  1.58it/s]
  8%|▊         | 566/7308 [07:33<1:11:15,  1.58it/s]
  8%|▊         | 567/7308 [07:34<1:11:14,  1.58it/s]
  8%|▊         | 568/7308 [07:34<1:11:13,  1.58it/s]
  8%|▊         | 569/7308 [07:35<1:11:13,  1.58it/s]
  8%|▊         | 570/7308 [07:36<1:11:12,  1.58it/s]
                                                    
{'loss': '5.302', 'grad_norm': '0.2558', 'learning_rate': '0.0006', 'epoch': '15.85'}
+
  8%|▊         | 570/7308 [07:36<1:11:12,  1.58it/s]
  8%|▊         | 571/7308 [07:36<1:11:12,  1.58it/s]
  8%|▊         | 572/7308 [07:37<1:11:13,  1.58it/s]
  8%|▊         | 573/7308 [07:37<1:11:12,  1.58it/s]
  8%|▊         | 574/7308 [07:38<1:11:12,  1.58it/s]
  8%|▊         | 575/7308 [07:39<1:11:07,  1.58it/s]
  8%|▊         | 576/7308 [07:39<57:56,  1.94it/s]  [transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 22.65it/s]
+
 15%|█▌        | 6/39 [00:00<00:01, 17.56it/s]
+
 21%|██        | 8/39 [00:00<00:01, 16.63it/s]
+
 26%|██▌       | 10/39 [00:00<00:01, 16.10it/s]
+
 31%|███       | 12/39 [00:00<00:01, 15.78it/s]
+
 36%|███▌      | 14/39 [00:00<00:01, 15.59it/s]
+
 41%|████      | 16/39 [00:00<00:01, 15.46it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 15.37it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 15.31it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 15.27it/s]
+
 62%|██████▏   | 24/39 [00:01<00:00, 15.26it/s]
+
 67%|██████▋   | 26/39 [00:01<00:00, 15.25it/s]
+
 72%|███████▏  | 28/39 [00:01<00:00, 15.25it/s]
+
 77%|███████▋  | 30/39 [00:01<00:00, 15.24it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 15.23it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 15.23it/s]
+
 92%|█████████▏| 36/39 [00:02<00:00, 15.23it/s]
+
 97%|█████████▋| 38/39 [00:02<00:00, 15.21it/s]
                                                  
+
                                               
{'eval_loss': '4.782', 'eval_runtime': '2.638', 'eval_samples_per_second': '925.5', 'eval_steps_per_second': '14.79', 'epoch': '16'}
+
  8%|▊         | 576/7308 [07:42<57:56,  1.94it/s]
+
100%|██████████| 39/39 [00:02<00:00, 15.21it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-576
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-576/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-576/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.96it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.96it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-576/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-576/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-576/tokenizer_config.json
+[eval] epoch 16.0 step 576 loss 4.7816 (best 4.7816 @ ep 16.0, 0/20 since best)
+
+
                                               
  8%|▊         | 577/7308 [07:44<3:40:03,  1.96s/it]
  8%|▊         | 578/7308 [07:45<2:55:21,  1.56s/it]
  8%|▊         | 579/7308 [07:46<2:24:05,  1.28s/it]
  8%|▊         | 580/7308 [07:46<2:02:13,  1.09s/it]
                                                    
{'loss': '5.252', 'grad_norm': '0.4444', 'learning_rate': '0.0006', 'epoch': '16.11'}
+
  8%|▊         | 580/7308 [07:46<2:02:13,  1.09s/it]
  8%|▊         | 581/7308 [07:47<1:46:54,  1.05it/s]
  8%|▊         | 582/7308 [07:47<1:36:10,  1.17it/s]
  8%|▊         | 583/7308 [07:48<1:28:41,  1.26it/s]
  8%|▊         | 584/7308 [07:49<1:23:25,  1.34it/s]
  8%|▊         | 585/7308 [07:49<1:19:44,  1.41it/s]
  8%|▊         | 586/7308 [07:50<1:17:08,  1.45it/s]
  8%|▊         | 587/7308 [07:51<1:15:19,  1.49it/s]
  8%|▊         | 588/7308 [07:51<1:14:02,  1.51it/s]
  8%|▊         | 589/7308 [07:52<1:13:09,  1.53it/s]
  8%|▊         | 590/7308 [07:53<1:12:31,  1.54it/s]
                                                    
{'loss': '5.268', 'grad_norm': '0.394', 'learning_rate': '0.0006', 'epoch': '16.4'}
+
  8%|▊         | 590/7308 [07:53<1:12:31,  1.54it/s]
  8%|▊         | 591/7308 [07:53<1:12:06,  1.55it/s]
  8%|▊         | 592/7308 [07:54<1:11:47,  1.56it/s]
  8%|▊         | 593/7308 [07:54<1:11:33,  1.56it/s]
  8%|▊         | 594/7308 [07:55<1:11:24,  1.57it/s]
  8%|▊         | 595/7308 [07:56<1:11:17,  1.57it/s]
  8%|▊         | 596/7308 [07:56<1:11:14,  1.57it/s]
  8%|▊         | 597/7308 [07:57<1:11:09,  1.57it/s]
  8%|▊         | 598/7308 [07:58<1:11:08,  1.57it/s]
  8%|▊         | 599/7308 [07:58<1:11:06,  1.57it/s]
  8%|▊         | 600/7308 [07:59<1:11:06,  1.57it/s]
                                                    
{'loss': '5.19', 'grad_norm': '0.3371', 'learning_rate': '0.0006', 'epoch': '16.68'}
+
  8%|▊         | 600/7308 [07:59<1:11:06,  1.57it/s]
  8%|▊         | 601/7308 [08:00<1:11:03,  1.57it/s]
  8%|▊         | 602/7308 [08:00<1:11:01,  1.57it/s]
  8%|▊         | 603/7308 [08:01<1:10:59,  1.57it/s]
  8%|▊         | 604/7308 [08:01<1:10:56,  1.57it/s]
  8%|▊         | 605/7308 [08:02<1:10:56,  1.57it/s]
  8%|▊         | 606/7308 [08:03<1:10:51,  1.58it/s]
  8%|▊         | 607/7308 [08:03<1:10:50,  1.58it/s]
  8%|▊         | 608/7308 [08:04<1:10:47,  1.58it/s]
  8%|▊         | 609/7308 [08:05<1:10:46,  1.58it/s]
  8%|▊         | 610/7308 [08:05<1:10:45,  1.58it/s]
                                                    
{'loss': '5.187', 'grad_norm': '0.2633', 'learning_rate': '0.0006', 'epoch': '16.96'}
+
  8%|▊         | 610/7308 [08:05<1:10:45,  1.58it/s]
  8%|▊         | 611/7308 [08:06<1:10:43,  1.58it/s]
  8%|▊         | 612/7308 [08:06<57:38,  1.94it/s]  
  8%|▊         | 613/7308 [08:07<1:04:18,  1.74it/s]
  8%|▊         | 614/7308 [08:07<1:06:11,  1.69it/s]
  8%|▊         | 615/7308 [08:08<1:07:34,  1.65it/s]
  8%|▊         | 616/7308 [08:09<1:08:29,  1.63it/s]
  8%|▊         | 617/7308 [08:09<1:09:08,  1.61it/s]
  8%|▊         | 618/7308 [08:10<1:09:35,  1.60it/s]
  8%|▊         | 619/7308 [08:11<1:09:53,  1.59it/s]
  8%|▊         | 620/7308 [08:11<1:10:07,  1.59it/s]
                                                    
{'loss': '5.142', 'grad_norm': '0.3475', 'learning_rate': '0.0006', 'epoch': '17.23'}
+
  8%|▊         | 620/7308 [08:11<1:10:07,  1.59it/s]
  8%|▊         | 621/7308 [08:12<1:10:16,  1.59it/s]
  9%|▊         | 622/7308 [08:13<1:10:20,  1.58it/s]
  9%|▊         | 623/7308 [08:13<1:10:24,  1.58it/s]
  9%|▊         | 624/7308 [08:14<1:10:26,  1.58it/s]
  9%|▊         | 625/7308 [08:14<1:10:27,  1.58it/s]
  9%|▊         | 626/7308 [08:15<1:10:30,  1.58it/s]
  9%|▊         | 627/7308 [08:16<1:10:29,  1.58it/s]
  9%|▊         | 628/7308 [08:16<1:10:30,  1.58it/s]
  9%|▊         | 629/7308 [08:17<1:10:31,  1.58it/s]
  9%|▊         | 630/7308 [08:18<1:10:30,  1.58it/s]
                                                    
{'loss': '5.101', 'grad_norm': '0.5287', 'learning_rate': '0.0006', 'epoch': '17.51'}
+
  9%|▊         | 630/7308 [08:18<1:10:30,  1.58it/s]
  9%|▊         | 631/7308 [08:18<1:10:32,  1.58it/s]
  9%|▊         | 632/7308 [08:19<1:10:31,  1.58it/s]
  9%|▊         | 633/7308 [08:19<1:10:31,  1.58it/s]
  9%|▊         | 634/7308 [08:20<1:10:32,  1.58it/s]
  9%|▊         | 635/7308 [08:21<1:10:30,  1.58it/s]
  9%|▊         | 636/7308 [08:21<1:10:30,  1.58it/s]
  9%|▊         | 637/7308 [08:22<1:10:29,  1.58it/s]
  9%|▊         | 638/7308 [08:23<1:10:28,  1.58it/s]
  9%|▊         | 639/7308 [08:23<1:10:28,  1.58it/s]
  9%|▉         | 640/7308 [08:24<1:10:27,  1.58it/s]
                                                    
{'loss': '5.146', 'grad_norm': '0.3084', 'learning_rate': '0.0006', 'epoch': '17.79'}
+
  9%|▉         | 640/7308 [08:24<1:10:27,  1.58it/s]
  9%|▉         | 641/7308 [08:25<1:10:27,  1.58it/s]
  9%|▉         | 642/7308 [08:25<1:10:25,  1.58it/s]
  9%|▉         | 643/7308 [08:26<1:10:24,  1.58it/s]
  9%|▉         | 644/7308 [08:26<1:10:24,  1.58it/s]
  9%|▉         | 645/7308 [08:27<1:10:23,  1.58it/s]
  9%|▉         | 646/7308 [08:28<1:10:22,  1.58it/s]
  9%|▉         | 647/7308 [08:28<1:10:18,  1.58it/s]
  9%|▉         | 648/7308 [08:29<57:19,  1.94it/s]  [transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 22.78it/s]
+
 15%|█▌        | 6/39 [00:00<00:01, 17.59it/s]
+
 21%|██        | 8/39 [00:00<00:01, 16.65it/s]
+
 26%|██▌       | 10/39 [00:00<00:01, 16.12it/s]
+
 31%|███       | 12/39 [00:00<00:01, 15.80it/s]
+
 36%|███▌      | 14/39 [00:00<00:01, 15.60it/s]
+
 41%|████      | 16/39 [00:00<00:01, 15.47it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 15.38it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 15.32it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 15.28it/s]
+
 62%|██████▏   | 24/39 [00:01<00:00, 15.28it/s]
+
 67%|██████▋   | 26/39 [00:01<00:00, 15.25it/s]
+
 72%|███████▏  | 28/39 [00:01<00:00, 15.22it/s]
+
 77%|███████▋  | 30/39 [00:01<00:00, 15.21it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 15.22it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 15.21it/s]
+
 92%|█████████▏| 36/39 [00:02<00:00, 15.22it/s]
+
 97%|█████████▋| 38/39 [00:02<00:00, 15.22it/s]
                                                  
+
                                               
{'eval_loss': '4.597', 'eval_runtime': '2.638', 'eval_samples_per_second': '925.4', 'eval_steps_per_second': '14.78', 'epoch': '18'}
+
  9%|▉         | 648/7308 [08:31<57:19,  1.94it/s]
+
100%|██████████| 39/39 [00:02<00:00, 15.22it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-648
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-648/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-648/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  2.17it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  2.16it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-648/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-648/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-648/tokenizer_config.json
+[eval] epoch 18.0 step 648 loss 4.5971 (best 4.5971 @ ep 18.0, 0/20 since best)
+
+
                                               
  9%|▉         | 649/7308 [08:34<3:36:33,  1.95s/it]
  9%|▉         | 650/7308 [08:35<2:52:39,  1.56s/it]
                                                    
{'loss': '5.109', 'grad_norm': '0.2984', 'learning_rate': '0.0006', 'epoch': '18.06'}
+
  9%|▉         | 650/7308 [08:35<2:52:39,  1.56s/it]
  9%|▉         | 651/7308 [08:35<2:21:55,  1.28s/it]
  9%|▉         | 652/7308 [08:36<2:00:24,  1.09s/it]
  9%|▉         | 653/7308 [08:36<1:45:22,  1.05it/s]
  9%|▉         | 654/7308 [08:37<1:34:49,  1.17it/s]
  9%|▉         | 655/7308 [08:38<1:27:26,  1.27it/s]
  9%|▉         | 656/7308 [08:38<1:22:16,  1.35it/s]
  9%|▉         | 657/7308 [08:39<1:18:38,  1.41it/s]
  9%|▉         | 658/7308 [08:40<1:16:07,  1.46it/s]
  9%|▉         | 659/7308 [08:40<1:14:20,  1.49it/s]
  9%|▉         | 660/7308 [08:41<1:13:05,  1.52it/s]
                                                    
{'loss': '5.097', 'grad_norm': '0.2716', 'learning_rate': '0.0006', 'epoch': '18.34'}
+
  9%|▉         | 660/7308 [08:41<1:13:05,  1.52it/s]
  9%|▉         | 661/7308 [08:41<1:12:13,  1.53it/s]
  9%|▉         | 662/7308 [08:42<1:11:36,  1.55it/s]
  9%|▉         | 663/7308 [08:43<1:11:10,  1.56it/s]
  9%|▉         | 664/7308 [08:43<1:10:51,  1.56it/s]
  9%|▉         | 665/7308 [08:44<1:10:39,  1.57it/s]
  9%|▉         | 666/7308 [08:45<1:10:30,  1.57it/s]
  9%|▉         | 667/7308 [08:45<1:10:23,  1.57it/s]
  9%|▉         | 668/7308 [08:46<1:10:19,  1.57it/s]
  9%|▉         | 669/7308 [08:47<1:10:18,  1.57it/s]
  9%|▉         | 670/7308 [08:47<1:10:14,  1.58it/s]
                                                    
{'loss': '5.062', 'grad_norm': '0.4978', 'learning_rate': '0.0006', 'epoch': '18.62'}
+
  9%|▉         | 670/7308 [08:47<1:10:14,  1.58it/s]
  9%|▉         | 671/7308 [08:48<1:10:14,  1.57it/s]
  9%|▉         | 672/7308 [08:48<1:10:10,  1.58it/s]
  9%|▉         | 673/7308 [08:49<1:10:07,  1.58it/s]
  9%|▉         | 674/7308 [08:50<1:10:08,  1.58it/s]
  9%|▉         | 675/7308 [08:50<1:10:06,  1.58it/s]
  9%|▉         | 676/7308 [08:51<1:10:03,  1.58it/s]
  9%|▉         | 677/7308 [08:52<1:10:01,  1.58it/s]
  9%|▉         | 678/7308 [08:52<1:10:01,  1.58it/s]
  9%|▉         | 679/7308 [08:53<1:10:01,  1.58it/s]
  9%|▉         | 680/7308 [08:54<1:10:07,  1.58it/s]
                                                    
{'loss': '5.04', 'grad_norm': '0.4114', 'learning_rate': '0.0006', 'epoch': '18.9'}
+
  9%|▉         | 680/7308 [08:54<1:10:07,  1.58it/s]
  9%|▉         | 681/7308 [08:54<1:10:07,  1.58it/s]
  9%|▉         | 682/7308 [08:55<1:10:09,  1.57it/s]
  9%|▉         | 683/7308 [08:55<1:10:07,  1.57it/s]
  9%|▉         | 684/7308 [08:56<57:11,  1.93it/s]  
  9%|▉         | 685/7308 [08:56<1:03:45,  1.73it/s]
  9%|▉         | 686/7308 [08:57<1:05:37,  1.68it/s]
  9%|▉         | 687/7308 [08:58<1:06:59,  1.65it/s]
  9%|▉         | 688/7308 [08:58<1:07:54,  1.62it/s]
  9%|▉         | 689/7308 [08:59<1:08:34,  1.61it/s]
  9%|▉         | 690/7308 [09:00<1:08:59,  1.60it/s]
                                                    
{'loss': '5.028', 'grad_norm': '0.303', 'learning_rate': '0.0006', 'epoch': '19.17'}
+
  9%|▉         | 690/7308 [09:00<1:08:59,  1.60it/s]
  9%|▉         | 691/7308 [09:00<1:09:15,  1.59it/s]
  9%|▉         | 692/7308 [09:01<1:09:25,  1.59it/s]
  9%|▉         | 693/7308 [09:01<1:09:33,  1.58it/s]
  9%|▉         | 694/7308 [09:02<1:09:37,  1.58it/s]
 10%|▉         | 695/7308 [09:03<1:09:42,  1.58it/s]
 10%|▉         | 696/7308 [09:03<1:09:43,  1.58it/s]
 10%|▉         | 697/7308 [09:04<1:09:43,  1.58it/s]
 10%|▉         | 698/7308 [09:05<1:09:44,  1.58it/s]
 10%|▉         | 699/7308 [09:05<1:09:47,  1.58it/s]
 10%|▉         | 700/7308 [09:06<1:09:49,  1.58it/s]
                                                    
{'loss': '4.936', 'grad_norm': '0.2995', 'learning_rate': '0.0006', 'epoch': '19.45'}
+
 10%|▉         | 700/7308 [09:06<1:09:49,  1.58it/s]
 10%|▉         | 701/7308 [09:07<1:09:51,  1.58it/s]
 10%|▉         | 702/7308 [09:07<1:09:51,  1.58it/s]
 10%|▉         | 703/7308 [09:08<1:09:51,  1.58it/s]
 10%|▉         | 704/7308 [09:08<1:09:51,  1.58it/s]
 10%|▉         | 705/7308 [09:09<1:09:50,  1.58it/s]
 10%|▉         | 706/7308 [09:10<1:09:48,  1.58it/s]
 10%|▉         | 707/7308 [09:10<1:09:50,  1.58it/s]
 10%|▉         | 708/7308 [09:11<1:09:47,  1.58it/s]
 10%|▉         | 709/7308 [09:12<1:09:44,  1.58it/s]
 10%|▉         | 710/7308 [09:12<1:09:44,  1.58it/s]
                                                    
{'loss': '5.024', 'grad_norm': '0.2909', 'learning_rate': '0.0006', 'epoch': '19.73'}
+
 10%|▉         | 710/7308 [09:12<1:09:44,  1.58it/s]
 10%|▉         | 711/7308 [09:13<1:09:43,  1.58it/s]
 10%|▉         | 712/7308 [09:14<1:09:42,  1.58it/s]
 10%|▉         | 713/7308 [09:14<1:09:41,  1.58it/s]
 10%|▉         | 714/7308 [09:15<1:09:38,  1.58it/s]
 10%|▉         | 715/7308 [09:15<1:09:37,  1.58it/s]
 10%|▉         | 716/7308 [09:16<1:09:36,  1.58it/s]
 10%|▉         | 717/7308 [09:17<1:09:38,  1.58it/s]
 10%|▉         | 718/7308 [09:17<1:09:37,  1.58it/s]
 10%|▉         | 719/7308 [09:18<1:09:37,  1.58it/s]
 10%|▉         | 720/7308 [09:18<56:46,  1.93it/s]  
                                                  
{'loss': '4.928', 'grad_norm': '0.2723', 'learning_rate': '0.0006', 'epoch': '20'}
+
 10%|▉         | 720/7308 [09:18<56:46,  1.93it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 22.82it/s]
+
 15%|█▌        | 6/39 [00:00<00:01, 17.57it/s]
+
 21%|██        | 8/39 [00:00<00:01, 16.61it/s]
+
 26%|██▌       | 10/39 [00:00<00:01, 16.10it/s]
+
 31%|███       | 12/39 [00:00<00:01, 15.79it/s]
+
 36%|███▌      | 14/39 [00:00<00:01, 15.58it/s]
+
 41%|████      | 16/39 [00:00<00:01, 15.46it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 15.37it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 15.32it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 15.27it/s]
+
 62%|██████▏   | 24/39 [00:01<00:00, 15.29it/s]
+
 67%|██████▋   | 26/39 [00:01<00:00, 15.27it/s]
+
 72%|███████▏  | 28/39 [00:01<00:00, 15.26it/s]
+
 77%|███████▋  | 30/39 [00:01<00:00, 15.24it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 15.23it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 15.21it/s]
+
 92%|█████████▏| 36/39 [00:02<00:00, 15.22it/s]
+
 97%|█████████▋| 38/39 [00:02<00:00, 15.20it/s]
                                                  
+
                                               
{'eval_loss': '4.465', 'eval_runtime': '2.637', 'eval_samples_per_second': '925.6', 'eval_steps_per_second': '14.79', 'epoch': '20'}
+
 10%|▉         | 720/7308 [09:21<56:46,  1.93it/s]
+
100%|██████████| 39/39 [00:02<00:00, 15.20it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-720
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-720/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-720/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.95it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.95it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-720/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-720/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-720/tokenizer_config.json
+[eval] epoch 20.0 step 720 loss 4.4653 (best 4.4653 @ ep 20.0, 0/20 since best)
+
+
                                               
 10%|▉         | 721/7308 [09:24<3:35:35,  1.96s/it]
 10%|▉         | 722/7308 [09:24<2:51:44,  1.56s/it]
 10%|▉         | 723/7308 [09:25<2:21:03,  1.29s/it]
 10%|▉         | 724/7308 [09:25<1:59:34,  1.09s/it]
 10%|▉         | 725/7308 [09:26<1:44:32,  1.05it/s]
 10%|▉         | 726/7308 [09:27<1:34:03,  1.17it/s]
 10%|▉         | 727/7308 [09:27<1:26:40,  1.27it/s]
 10%|▉         | 728/7308 [09:28<1:21:31,  1.35it/s]
 10%|▉         | 729/7308 [09:29<1:17:54,  1.41it/s]
 10%|▉         | 730/7308 [09:29<1:15:20,  1.46it/s]
                                                    
{'loss': '4.854', 'grad_norm': '0.5063', 'learning_rate': '0.0006', 'epoch': '20.28'}
+
 10%|▉         | 730/7308 [09:29<1:15:20,  1.46it/s]
 10%|█         | 731/7308 [09:30<1:13:37,  1.49it/s]
 10%|█         | 732/7308 [09:31<1:12:24,  1.51it/s]
 10%|█         | 733/7308 [09:31<1:11:30,  1.53it/s]
 10%|█         | 734/7308 [09:32<1:10:54,  1.55it/s]
 10%|█         | 735/7308 [09:32<1:10:28,  1.55it/s]
 10%|█         | 736/7308 [09:33<1:10:10,  1.56it/s]
 10%|█         | 737/7308 [09:34<1:09:55,  1.57it/s]
 10%|█         | 738/7308 [09:34<1:09:46,  1.57it/s]
 10%|█         | 739/7308 [09:35<1:09:38,  1.57it/s]
 10%|█         | 740/7308 [09:36<1:09:33,  1.57it/s]
                                                    
{'loss': '4.986', 'grad_norm': '0.3144', 'learning_rate': '0.0006', 'epoch': '20.57'}
+
 10%|█         | 740/7308 [09:36<1:09:33,  1.57it/s]
 10%|█         | 741/7308 [09:36<1:09:30,  1.57it/s]
 10%|█         | 742/7308 [09:37<1:09:26,  1.58it/s]
 10%|█         | 743/7308 [09:38<1:09:23,  1.58it/s]
 10%|█         | 744/7308 [09:38<1:09:20,  1.58it/s]
 10%|█         | 745/7308 [09:39<1:09:19,  1.58it/s]
 10%|█         | 746/7308 [09:39<1:09:18,  1.58it/s]
 10%|█         | 747/7308 [09:40<1:09:18,  1.58it/s]
 10%|█         | 748/7308 [09:41<1:09:18,  1.58it/s]
 10%|█         | 749/7308 [09:41<1:09:18,  1.58it/s]
 10%|█         | 750/7308 [09:42<1:09:17,  1.58it/s]
                                                    
{'loss': '4.865', 'grad_norm': '0.2279', 'learning_rate': '0.0006', 'epoch': '20.85'}
+
 10%|█         | 750/7308 [09:42<1:09:17,  1.58it/s]
 10%|█         | 751/7308 [09:43<1:09:16,  1.58it/s]
 10%|█         | 752/7308 [09:43<1:09:15,  1.58it/s]
 10%|█         | 753/7308 [09:44<1:09:14,  1.58it/s]
 10%|█         | 754/7308 [09:44<1:09:15,  1.58it/s]
 10%|█         | 755/7308 [09:45<1:09:14,  1.58it/s]
 10%|█         | 756/7308 [09:45<56:26,  1.93it/s]  
 10%|█         | 757/7308 [09:46<1:03:25,  1.72it/s]
 10%|█         | 758/7308 [09:47<1:05:11,  1.67it/s]
 10%|█         | 759/7308 [09:47<1:06:23,  1.64it/s]
 10%|█         | 760/7308 [09:48<1:07:13,  1.62it/s]
                                                    
{'loss': '4.869', 'grad_norm': '0.235', 'learning_rate': '0.0006', 'epoch': '21.11'}
+
 10%|█         | 760/7308 [09:48<1:07:13,  1.62it/s]
 10%|█         | 761/7308 [09:49<1:07:51,  1.61it/s]
 10%|█         | 762/7308 [09:49<1:08:14,  1.60it/s]
 10%|█         | 763/7308 [09:50<1:08:30,  1.59it/s]
 10%|█         | 764/7308 [09:51<1:08:41,  1.59it/s]
 10%|█         | 765/7308 [09:51<1:08:48,  1.58it/s]
 10%|█         | 766/7308 [09:52<1:08:56,  1.58it/s]
 10%|█         | 767/7308 [09:52<1:09:02,  1.58it/s]
 11%|█         | 768/7308 [09:53<1:09:02,  1.58it/s]
 11%|█         | 769/7308 [09:54<1:09:05,  1.58it/s]
 11%|█         | 770/7308 [09:54<1:09:03,  1.58it/s]
                                                    
{'loss': '4.823', 'grad_norm': '0.27', 'learning_rate': '0.0006', 'epoch': '21.4'}
+
 11%|█         | 770/7308 [09:54<1:09:03,  1.58it/s]
 11%|█         | 771/7308 [09:55<1:09:06,  1.58it/s]
 11%|█         | 772/7308 [09:56<1:09:04,  1.58it/s]
 11%|█         | 773/7308 [09:56<1:09:03,  1.58it/s]
 11%|█         | 774/7308 [09:57<1:09:02,  1.58it/s]
 11%|█         | 775/7308 [09:58<1:09:03,  1.58it/s]
 11%|█         | 776/7308 [09:58<1:09:01,  1.58it/s]
 11%|█         | 777/7308 [09:59<1:09:03,  1.58it/s]
 11%|█         | 778/7308 [09:59<1:09:02,  1.58it/s]
 11%|█         | 779/7308 [10:00<1:09:01,  1.58it/s]
 11%|█         | 780/7308 [10:01<1:08:59,  1.58it/s]
                                                    
{'loss': '4.829', 'grad_norm': '0.2766', 'learning_rate': '0.0006', 'epoch': '21.68'}
+
 11%|█         | 780/7308 [10:01<1:08:59,  1.58it/s]
 11%|█         | 781/7308 [10:01<1:08:59,  1.58it/s]
 11%|█         | 782/7308 [10:02<1:08:58,  1.58it/s]
 11%|█         | 783/7308 [10:03<1:08:58,  1.58it/s]
 11%|█         | 784/7308 [10:03<1:08:57,  1.58it/s]
 11%|█         | 785/7308 [10:04<1:08:59,  1.58it/s]
 11%|█         | 786/7308 [10:04<1:08:57,  1.58it/s]
 11%|█         | 787/7308 [10:05<1:08:55,  1.58it/s]
 11%|█         | 788/7308 [10:06<1:08:55,  1.58it/s]
 11%|█         | 789/7308 [10:06<1:08:54,  1.58it/s]
 11%|█         | 790/7308 [10:07<1:08:53,  1.58it/s]
                                                    
{'loss': '4.828', 'grad_norm': '0.3184', 'learning_rate': '0.0006', 'epoch': '21.96'}
+
 11%|█         | 790/7308 [10:07<1:08:53,  1.58it/s]
 11%|█         | 791/7308 [10:08<1:08:52,  1.58it/s]
 11%|█         | 792/7308 [10:08<56:08,  1.93it/s]  [transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 22.75it/s]
+
 15%|█▌        | 6/39 [00:00<00:01, 17.58it/s]
+
 21%|██        | 8/39 [00:00<00:01, 16.64it/s]
+
 26%|██▌       | 10/39 [00:00<00:01, 16.12it/s]
+
 31%|███       | 12/39 [00:00<00:01, 15.80it/s]
+
 36%|███▌      | 14/39 [00:00<00:01, 15.60it/s]
+
 41%|████      | 16/39 [00:00<00:01, 15.47it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 15.39it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 15.33it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 15.29it/s]
+
 62%|██████▏   | 24/39 [00:01<00:00, 15.30it/s]
+
 67%|██████▋   | 26/39 [00:01<00:00, 15.28it/s]
+
 72%|███████▏  | 28/39 [00:01<00:00, 15.25it/s]
+
 77%|███████▋  | 30/39 [00:01<00:00, 15.25it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 15.24it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 15.24it/s]
+
 92%|█████████▏| 36/39 [00:02<00:00, 15.23it/s]
+
 97%|█████████▋| 38/39 [00:02<00:00, 15.23it/s]
                                                  
+
                                               
{'eval_loss': '4.376', 'eval_runtime': '2.635', 'eval_samples_per_second': '926.4', 'eval_steps_per_second': '14.8', 'epoch': '22'}
+
 11%|█         | 792/7308 [10:11<56:08,  1.93it/s]
+
100%|██████████| 39/39 [00:02<00:00, 15.23it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-792
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-792/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-792/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  2.18it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  2.18it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-792/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-792/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-792/tokenizer_config.json
+[eval] epoch 22.0 step 792 loss 4.3764 (best 4.3764 @ ep 22.0, 0/20 since best)
+
+
                                               
 11%|█         | 793/7308 [10:13<3:31:14,  1.95s/it]
 11%|█         | 794/7308 [10:14<2:48:30,  1.55s/it]
 11%|█         | 795/7308 [10:14<2:18:32,  1.28s/it]
 11%|█         | 796/7308 [10:15<1:57:36,  1.08s/it]
 11%|█         | 797/7308 [10:16<1:42:56,  1.05it/s]
 11%|█         | 798/7308 [10:16<1:32:40,  1.17it/s]
 11%|█         | 799/7308 [10:17<1:25:28,  1.27it/s]
 11%|█         | 800/7308 [10:18<1:20:26,  1.35it/s]
                                                    
{'loss': '4.808', 'grad_norm': '0.4016', 'learning_rate': '0.0006', 'epoch': '22.23'}
+
 11%|█         | 800/7308 [10:18<1:20:26,  1.35it/s]
 11%|█         | 801/7308 [10:18<1:16:57,  1.41it/s]
 11%|█         | 802/7308 [10:19<1:14:29,  1.46it/s]
 11%|█         | 803/7308 [10:20<1:12:46,  1.49it/s]
 11%|█         | 804/7308 [10:20<1:11:32,  1.52it/s]
 11%|█         | 805/7308 [10:21<1:10:41,  1.53it/s]
 11%|█         | 806/7308 [10:21<1:10:05,  1.55it/s]
 11%|█         | 807/7308 [10:22<1:09:41,  1.55it/s]
 11%|█         | 808/7308 [10:23<1:09:22,  1.56it/s]
 11%|█         | 809/7308 [10:23<1:09:09,  1.57it/s]
 11%|█         | 810/7308 [10:24<1:09:00,  1.57it/s]
                                                    
{'loss': '4.793', 'grad_norm': '0.2822', 'learning_rate': '0.0006', 'epoch': '22.51'}
+
 11%|█         | 810/7308 [10:24<1:09:00,  1.57it/s]
 11%|█         | 811/7308 [10:25<1:08:55,  1.57it/s]
 11%|█         | 812/7308 [10:25<1:08:48,  1.57it/s]
 11%|█         | 813/7308 [10:26<1:08:45,  1.57it/s]
 11%|█         | 814/7308 [10:26<1:08:42,  1.58it/s]
 11%|█         | 815/7308 [10:27<1:08:39,  1.58it/s]
 11%|█         | 816/7308 [10:28<1:08:37,  1.58it/s]
 11%|█         | 817/7308 [10:28<1:08:38,  1.58it/s]
 11%|█         | 818/7308 [10:29<1:08:37,  1.58it/s]
 11%|█         | 819/7308 [10:30<1:08:35,  1.58it/s]
 11%|█         | 820/7308 [10:30<1:08:34,  1.58it/s]
                                                    
{'loss': '4.879', 'grad_norm': '0.2999', 'learning_rate': '0.0006', 'epoch': '22.79'}
+
 11%|█         | 820/7308 [10:30<1:08:34,  1.58it/s]
 11%|█         | 821/7308 [10:31<1:08:33,  1.58it/s]
 11%|█         | 822/7308 [10:32<1:08:31,  1.58it/s]
 11%|█▏        | 823/7308 [10:32<1:08:29,  1.58it/s]
 11%|█▏        | 824/7308 [10:33<1:08:30,  1.58it/s]
 11%|█▏        | 825/7308 [10:33<1:08:30,  1.58it/s]
 11%|█▏        | 826/7308 [10:34<1:08:31,  1.58it/s]
 11%|█▏        | 827/7308 [10:35<1:08:28,  1.58it/s]
 11%|█▏        | 828/7308 [10:35<55:49,  1.93it/s]  
 11%|█▏        | 829/7308 [10:36<1:01:51,  1.75it/s]
 11%|█▏        | 830/7308 [10:36<1:03:49,  1.69it/s]
                                                    
{'loss': '4.83', 'grad_norm': '0.2588', 'learning_rate': '0.0006', 'epoch': '23.06'}
+
 11%|█▏        | 830/7308 [10:36<1:03:49,  1.69it/s]
 11%|█▏        | 831/7308 [10:37<1:05:14,  1.65it/s]
 11%|█▏        | 832/7308 [10:38<1:06:11,  1.63it/s]
 11%|█▏        | 833/7308 [10:38<1:06:50,  1.61it/s]
 11%|█▏        | 834/7308 [10:39<1:07:17,  1.60it/s]
 11%|█▏        | 835/7308 [10:39<1:07:38,  1.59it/s]
 11%|█▏        | 836/7308 [10:40<1:07:51,  1.59it/s]
 11%|█▏        | 837/7308 [10:41<1:07:59,  1.59it/s]
 11%|█▏        | 838/7308 [10:41<1:08:07,  1.58it/s]
 11%|█▏        | 839/7308 [10:42<1:08:10,  1.58it/s]
 11%|█▏        | 840/7308 [10:43<1:08:12,  1.58it/s]
                                                    
{'loss': '4.786', 'grad_norm': '0.2871', 'learning_rate': '0.0006', 'epoch': '23.34'}
+
 11%|█▏        | 840/7308 [10:43<1:08:12,  1.58it/s]
 12%|█▏        | 841/7308 [10:43<1:08:15,  1.58it/s]
 12%|█▏        | 842/7308 [10:44<1:08:15,  1.58it/s]
 12%|█▏        | 843/7308 [10:45<1:08:15,  1.58it/s]
 12%|█▏        | 844/7308 [10:45<1:08:16,  1.58it/s]
 12%|█▏        | 845/7308 [10:46<1:08:16,  1.58it/s]
 12%|█▏        | 846/7308 [10:46<1:08:18,  1.58it/s]
 12%|█▏        | 847/7308 [10:47<1:08:17,  1.58it/s]
 12%|█▏        | 848/7308 [10:48<1:08:15,  1.58it/s]
 12%|█▏        | 849/7308 [10:48<1:08:14,  1.58it/s]
 12%|█▏        | 850/7308 [10:49<1:08:14,  1.58it/s]
                                                    
{'loss': '4.767', 'grad_norm': '0.4411', 'learning_rate': '0.0006', 'epoch': '23.62'}
+
 12%|█▏        | 850/7308 [10:49<1:08:14,  1.58it/s]
 12%|█▏        | 851/7308 [10:50<1:08:15,  1.58it/s]
 12%|█▏        | 852/7308 [10:50<1:08:13,  1.58it/s]
 12%|█▏        | 853/7308 [10:51<1:08:12,  1.58it/s]
 12%|█▏        | 854/7308 [10:52<1:08:13,  1.58it/s]
 12%|█▏        | 855/7308 [10:52<1:08:14,  1.58it/s]
 12%|█▏        | 856/7308 [10:53<1:08:12,  1.58it/s]
 12%|█▏        | 857/7308 [10:53<1:08:11,  1.58it/s]
 12%|█▏        | 858/7308 [10:54<1:08:10,  1.58it/s]
 12%|█▏        | 859/7308 [10:55<1:08:10,  1.58it/s]
 12%|█▏        | 860/7308 [10:55<1:08:08,  1.58it/s]
                                                    
{'loss': '4.724', 'grad_norm': '0.2868', 'learning_rate': '0.0006', 'epoch': '23.9'}
+
 12%|█▏        | 860/7308 [10:55<1:08:08,  1.58it/s]
 12%|█▏        | 861/7308 [10:56<1:08:08,  1.58it/s]
 12%|█▏        | 862/7308 [10:57<1:08:09,  1.58it/s]
 12%|█▏        | 863/7308 [10:57<1:08:08,  1.58it/s]
 12%|█▏        | 864/7308 [10:57<55:30,  1.93it/s]  [transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 22.80it/s]
+
 15%|█▌        | 6/39 [00:00<00:01, 17.59it/s]
+
 21%|██        | 8/39 [00:00<00:01, 16.62it/s]
+
 26%|██▌       | 10/39 [00:00<00:01, 16.12it/s]
+
 31%|███       | 12/39 [00:00<00:01, 15.80it/s]
+
 36%|███▌      | 14/39 [00:00<00:01, 15.61it/s]
+
 41%|████      | 16/39 [00:00<00:01, 15.46it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 15.38it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 15.30it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 15.27it/s]
+
 62%|██████▏   | 24/39 [00:01<00:00, 15.26it/s]
+
 67%|██████▋   | 26/39 [00:01<00:00, 15.25it/s]
+
 72%|███████▏  | 28/39 [00:01<00:00, 15.25it/s]
+
 77%|███████▋  | 30/39 [00:01<00:00, 15.22it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 15.23it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 15.22it/s]
+
 92%|█████████▏| 36/39 [00:02<00:00, 15.21it/s]
+
 97%|█████████▋| 38/39 [00:02<00:00, 15.21it/s]
                                                  
+
                                               
{'eval_loss': '4.321', 'eval_runtime': '2.635', 'eval_samples_per_second': '926.3', 'eval_steps_per_second': '14.8', 'epoch': '24'}
+
 12%|█▏        | 864/7308 [11:00<55:30,  1.93it/s]
+
100%|██████████| 39/39 [00:02<00:00, 15.21it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-864
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-864/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-864/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.96it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.96it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-864/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-864/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-864/tokenizer_config.json
+[eval] epoch 24.0 step 864 loss 4.3211 (best 4.3211 @ ep 24.0, 0/20 since best)
+
+
                                               
 12%|█▏        | 865/7308 [11:03<3:30:46,  1.96s/it]
 12%|█▏        | 866/7308 [11:03<2:47:54,  1.56s/it]
 12%|█▏        | 867/7308 [11:04<2:17:54,  1.28s/it]
 12%|█▏        | 868/7308 [11:05<1:56:55,  1.09s/it]
 12%|█▏        | 869/7308 [11:05<1:42:13,  1.05it/s]
 12%|█▏        | 870/7308 [11:06<1:31:55,  1.17it/s]
                                                    
{'loss': '4.788', 'grad_norm': '0.3366', 'learning_rate': '0.0006', 'epoch': '24.17'}
+
 12%|█▏        | 870/7308 [11:06<1:31:55,  1.17it/s]
 12%|█▏        | 871/7308 [11:07<1:24:44,  1.27it/s]
 12%|█▏        | 872/7308 [11:07<1:19:41,  1.35it/s]
 12%|█▏        | 873/7308 [11:08<1:16:11,  1.41it/s]
 12%|█▏        | 874/7308 [11:09<1:13:41,  1.46it/s]
 12%|█▏        | 875/7308 [11:09<1:11:58,  1.49it/s]
 12%|█▏        | 876/7308 [11:10<1:10:46,  1.51it/s]
 12%|█▏        | 877/7308 [11:10<1:09:54,  1.53it/s]
 12%|█▏        | 878/7308 [11:11<1:09:19,  1.55it/s]
 12%|█▏        | 879/7308 [11:12<1:08:55,  1.55it/s]
 12%|█▏        | 880/7308 [11:12<1:08:45,  1.56it/s]
                                                    
{'loss': '4.63', 'grad_norm': '0.2279', 'learning_rate': '0.0006', 'epoch': '24.45'}
+
 12%|█▏        | 880/7308 [11:12<1:08:45,  1.56it/s]
 12%|█▏        | 881/7308 [11:13<1:08:30,  1.56it/s]
 12%|█▏        | 882/7308 [11:14<1:08:18,  1.57it/s]
 12%|█▏        | 883/7308 [11:14<1:08:11,  1.57it/s]
 12%|█▏        | 884/7308 [11:15<1:08:07,  1.57it/s]
 12%|█▏        | 885/7308 [11:16<1:08:02,  1.57it/s]
 12%|█▏        | 886/7308 [11:16<1:07:58,  1.57it/s]
 12%|█▏        | 887/7308 [11:17<1:07:56,  1.58it/s]
 12%|█▏        | 888/7308 [11:17<1:07:54,  1.58it/s]
 12%|█▏        | 889/7308 [11:18<1:07:51,  1.58it/s]
 12%|█▏        | 890/7308 [11:19<1:07:50,  1.58it/s]
                                                    
{'loss': '4.699', 'grad_norm': '0.2645', 'learning_rate': '0.0006', 'epoch': '24.73'}
+
 12%|█▏        | 890/7308 [11:19<1:07:50,  1.58it/s]
 12%|█▏        | 891/7308 [11:19<1:07:50,  1.58it/s]
 12%|█▏        | 892/7308 [11:20<1:07:47,  1.58it/s]
 12%|█▏        | 893/7308 [11:21<1:07:47,  1.58it/s]
 12%|█▏        | 894/7308 [11:21<1:07:48,  1.58it/s]
 12%|█▏        | 895/7308 [11:22<1:07:46,  1.58it/s]
 12%|█▏        | 896/7308 [11:22<1:07:45,  1.58it/s]
 12%|█▏        | 897/7308 [11:23<1:07:45,  1.58it/s]
 12%|█▏        | 898/7308 [11:24<1:07:45,  1.58it/s]
 12%|█▏        | 899/7308 [11:24<1:07:42,  1.58it/s]
 12%|█▏        | 900/7308 [11:25<55:10,  1.94it/s]  
                                                  
{'loss': '4.722', 'grad_norm': '0.2536', 'learning_rate': '0.0006', 'epoch': '25'}
+
 12%|█▏        | 900/7308 [11:25<55:10,  1.94it/s]
 12%|█▏        | 901/7308 [11:25<1:01:41,  1.73it/s]
 12%|█▏        | 902/7308 [11:26<1:03:28,  1.68it/s]
 12%|█▏        | 903/7308 [11:27<1:04:45,  1.65it/s]
 12%|█▏        | 904/7308 [11:27<1:05:36,  1.63it/s]
 12%|█▏        | 905/7308 [11:28<1:06:12,  1.61it/s]
 12%|█▏        | 906/7308 [11:29<1:06:38,  1.60it/s]
 12%|█▏        | 907/7308 [11:29<1:06:58,  1.59it/s]
 12%|█▏        | 908/7308 [11:30<1:07:11,  1.59it/s]
 12%|█▏        | 909/7308 [11:30<1:07:19,  1.58it/s]
 12%|█▏        | 910/7308 [11:31<1:07:23,  1.58it/s]
                                                    
{'loss': '4.743', 'grad_norm': '0.2541', 'learning_rate': '0.0006', 'epoch': '25.28'}
+
 12%|█▏        | 910/7308 [11:31<1:07:23,  1.58it/s]
 12%|█▏        | 911/7308 [11:32<1:07:27,  1.58it/s]
 12%|█▏        | 912/7308 [11:32<1:07:29,  1.58it/s]
 12%|█▏        | 913/7308 [11:33<1:07:32,  1.58it/s]
 13%|█▎        | 914/7308 [11:34<1:07:33,  1.58it/s]
 13%|█▎        | 915/7308 [11:34<1:07:32,  1.58it/s]
 13%|█▎        | 916/7308 [11:35<1:07:32,  1.58it/s]
 13%|█▎        | 917/7308 [11:35<1:07:32,  1.58it/s]
 13%|█▎        | 918/7308 [11:36<1:07:31,  1.58it/s]
 13%|█▎        | 919/7308 [11:37<1:07:30,  1.58it/s]
 13%|█▎        | 920/7308 [11:37<1:07:29,  1.58it/s]
                                                    
{'loss': '4.687', 'grad_norm': '0.3064', 'learning_rate': '0.0006', 'epoch': '25.57'}
+
 13%|█▎        | 920/7308 [11:37<1:07:29,  1.58it/s]
 13%|█▎        | 921/7308 [11:38<1:07:31,  1.58it/s]
 13%|█▎        | 922/7308 [11:39<1:07:29,  1.58it/s]
 13%|█▎        | 923/7308 [11:39<1:07:28,  1.58it/s]
 13%|█▎        | 924/7308 [11:40<1:07:26,  1.58it/s]
 13%|█▎        | 925/7308 [11:41<1:07:25,  1.58it/s]
 13%|█▎        | 926/7308 [11:41<1:07:28,  1.58it/s]
 13%|█▎        | 927/7308 [11:42<1:07:29,  1.58it/s]
 13%|█▎        | 928/7308 [11:42<1:07:27,  1.58it/s]
 13%|█▎        | 929/7308 [11:43<1:07:27,  1.58it/s]
 13%|█▎        | 930/7308 [11:44<1:07:25,  1.58it/s]
                                                    
{'loss': '4.634', 'grad_norm': '0.2899', 'learning_rate': '0.0006', 'epoch': '25.85'}
+
 13%|█▎        | 930/7308 [11:44<1:07:25,  1.58it/s]
 13%|█▎        | 931/7308 [11:44<1:07:24,  1.58it/s]
 13%|█▎        | 932/7308 [11:45<1:07:24,  1.58it/s]
 13%|█▎        | 933/7308 [11:46<1:07:23,  1.58it/s]
 13%|█▎        | 934/7308 [11:46<1:07:22,  1.58it/s]
 13%|█▎        | 935/7308 [11:47<1:07:20,  1.58it/s]
 13%|█▎        | 936/7308 [11:47<54:52,  1.94it/s]  [transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 23.26it/s]
+
 15%|█▌        | 6/39 [00:00<00:01, 17.66it/s]
+
 21%|██        | 8/39 [00:00<00:01, 16.71it/s]
+
 26%|██▌       | 10/39 [00:00<00:01, 16.14it/s]
+
 31%|███       | 12/39 [00:00<00:01, 15.81it/s]
+
 36%|███▌      | 14/39 [00:00<00:01, 15.60it/s]
+
 41%|████      | 16/39 [00:00<00:01, 15.47it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 15.36it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 15.31it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 15.26it/s]
+
 62%|██████▏   | 24/39 [00:01<00:00, 15.28it/s]
+
 67%|██████▋   | 26/39 [00:01<00:00, 15.25it/s]
+
 72%|███████▏  | 28/39 [00:01<00:00, 15.24it/s]
+
 77%|███████▋  | 30/39 [00:01<00:00, 15.23it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 15.23it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 15.22it/s]
+
 92%|█████████▏| 36/39 [00:02<00:00, 15.22it/s]
+
 97%|█████████▋| 38/39 [00:02<00:00, 15.21it/s]
                                                  
+
                                               
{'eval_loss': '4.254', 'eval_runtime': '2.623', 'eval_samples_per_second': '930.7', 'eval_steps_per_second': '14.87', 'epoch': '26'}
+
 13%|█▎        | 936/7308 [11:50<54:52,  1.94it/s]
+
100%|██████████| 39/39 [00:02<00:00, 15.21it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-936
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-936/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-936/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.71it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.71it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-936/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-936/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-936/tokenizer_config.json
+[eval] epoch 26.0 step 936 loss 4.2537 (best 4.2537 @ ep 26.0, 0/20 since best)
+
+
                                               
 13%|█▎        | 937/7308 [11:53<3:30:56,  1.99s/it]
 13%|█▎        | 938/7308 [11:53<2:47:49,  1.58s/it]
 13%|█▎        | 939/7308 [11:54<2:17:37,  1.30s/it]
 13%|█▎        | 940/7308 [11:54<1:56:30,  1.10s/it]
                                                    
{'loss': '4.719', 'grad_norm': '0.3405', 'learning_rate': '0.0006', 'epoch': '26.11'}
+
 13%|█▎        | 940/7308 [11:54<1:56:30,  1.10s/it]
 13%|█▎        | 941/7308 [11:55<1:41:44,  1.04it/s]
 13%|█▎        | 942/7308 [11:56<1:31:23,  1.16it/s]
 13%|█▎        | 943/7308 [11:56<1:24:09,  1.26it/s]
 13%|█▎        | 944/7308 [11:57<1:19:04,  1.34it/s]
 13%|█▎        | 945/7308 [11:58<1:15:30,  1.40it/s]
 13%|█▎        | 946/7308 [11:58<1:13:00,  1.45it/s]
 13%|█▎        | 947/7308 [11:59<1:11:15,  1.49it/s]
 13%|█▎        | 948/7308 [12:00<1:10:02,  1.51it/s]
 13%|█▎        | 949/7308 [12:00<1:09:10,  1.53it/s]
 13%|█▎        | 950/7308 [12:01<1:08:33,  1.55it/s]
                                                    
{'loss': '4.71', 'grad_norm': '0.3541', 'learning_rate': '0.0006', 'epoch': '26.4'}
+
 13%|█▎        | 950/7308 [12:01<1:08:33,  1.55it/s]
 13%|█▎        | 951/7308 [12:01<1:08:09,  1.55it/s]
 13%|█▎        | 952/7308 [12:02<1:07:53,  1.56it/s]
 13%|█▎        | 953/7308 [12:03<1:07:40,  1.57it/s]
 13%|█▎        | 954/7308 [12:03<1:07:33,  1.57it/s]
 13%|█▎        | 955/7308 [12:04<1:07:26,  1.57it/s]
 13%|█▎        | 956/7308 [12:05<1:07:19,  1.57it/s]
 13%|█▎        | 957/7308 [12:05<1:07:18,  1.57it/s]
 13%|█▎        | 958/7308 [12:06<1:07:13,  1.57it/s]
 13%|█▎        | 959/7308 [12:07<1:07:12,  1.57it/s]
 13%|█▎        | 960/7308 [12:07<1:07:11,  1.57it/s]
                                                    
{'loss': '4.671', 'grad_norm': '0.2729', 'learning_rate': '0.0006', 'epoch': '26.68'}
+
 13%|█▎        | 960/7308 [12:07<1:07:11,  1.57it/s]
 13%|█▎        | 961/7308 [12:08<1:07:09,  1.58it/s]
 13%|█▎        | 962/7308 [12:08<1:07:07,  1.58it/s]
 13%|█▎        | 963/7308 [12:09<1:07:06,  1.58it/s]
 13%|█▎        | 964/7308 [12:10<1:07:05,  1.58it/s]
 13%|█▎        | 965/7308 [12:10<1:07:03,  1.58it/s]
 13%|█▎        | 966/7308 [12:11<1:07:02,  1.58it/s]
 13%|█▎        | 967/7308 [12:12<1:07:02,  1.58it/s]
 13%|█▎        | 968/7308 [12:12<1:07:03,  1.58it/s]
 13%|█▎        | 969/7308 [12:13<1:07:00,  1.58it/s]
 13%|█▎        | 970/7308 [12:13<1:06:59,  1.58it/s]
                                                    
{'loss': '4.648', 'grad_norm': '0.2242', 'learning_rate': '0.0006', 'epoch': '26.96'}
+
 13%|█▎        | 970/7308 [12:14<1:06:59,  1.58it/s]
 13%|█▎        | 971/7308 [12:14<1:07:05,  1.57it/s]
 13%|█▎        | 972/7308 [12:14<54:40,  1.93it/s]  
 13%|█▎        | 973/7308 [12:15<1:01:27,  1.72it/s]
 13%|█▎        | 974/7308 [12:16<1:03:05,  1.67it/s]
 13%|█▎        | 975/7308 [12:16<1:04:14,  1.64it/s]
 13%|█▎        | 976/7308 [12:17<1:05:03,  1.62it/s]
 13%|█▎        | 977/7308 [12:18<1:05:36,  1.61it/s]
 13%|█▎        | 978/7308 [12:18<1:05:58,  1.60it/s]
 13%|█▎        | 979/7308 [12:19<1:06:13,  1.59it/s]
 13%|█▎        | 980/7308 [12:20<1:06:24,  1.59it/s]
                                                    
{'loss': '4.584', 'grad_norm': '0.364', 'learning_rate': '0.0006', 'epoch': '27.23'}
+
 13%|█▎        | 980/7308 [12:20<1:06:24,  1.59it/s]
 13%|█▎        | 981/7308 [12:20<1:06:33,  1.58it/s]
 13%|█▎        | 982/7308 [12:21<1:06:37,  1.58it/s]
 13%|█▎        | 983/7308 [12:21<1:06:39,  1.58it/s]
 13%|█▎        | 984/7308 [12:22<1:06:42,  1.58it/s]
 13%|█▎        | 985/7308 [12:23<1:06:46,  1.58it/s]
 13%|█▎        | 986/7308 [12:23<1:06:46,  1.58it/s]
 14%|█▎        | 987/7308 [12:24<1:06:46,  1.58it/s]
 14%|█▎        | 988/7308 [12:25<1:06:45,  1.58it/s]
 14%|█▎        | 989/7308 [12:25<1:06:45,  1.58it/s]
 14%|█▎        | 990/7308 [12:26<1:06:45,  1.58it/s]
                                                    
{'loss': '4.626', 'grad_norm': '0.3236', 'learning_rate': '0.0006', 'epoch': '27.51'}
+
 14%|█▎        | 990/7308 [12:26<1:06:45,  1.58it/s]
 14%|█▎        | 991/7308 [12:27<1:06:46,  1.58it/s]
 14%|█▎        | 992/7308 [12:27<1:06:46,  1.58it/s]
 14%|█▎        | 993/7308 [12:28<1:06:44,  1.58it/s]
 14%|█▎        | 994/7308 [12:28<1:06:43,  1.58it/s]
 14%|█▎        | 995/7308 [12:29<1:06:42,  1.58it/s]
 14%|█▎        | 996/7308 [12:30<1:06:41,  1.58it/s]
 14%|█▎        | 997/7308 [12:30<1:06:40,  1.58it/s]
 14%|█▎        | 998/7308 [12:31<1:06:40,  1.58it/s]
 14%|█▎        | 999/7308 [12:32<1:06:40,  1.58it/s]
 14%|█▎        | 1000/7308 [12:32<1:06:40,  1.58it/s]
                                                     
{'loss': '4.667', 'grad_norm': '0.2397', 'learning_rate': '0.0006', 'epoch': '27.79'}
+
 14%|█▎        | 1000/7308 [12:32<1:06:40,  1.58it/s]
 14%|█▎        | 1001/7308 [12:33<1:06:42,  1.58it/s]
 14%|█▎        | 1002/7308 [12:34<1:06:39,  1.58it/s]
 14%|█▎        | 1003/7308 [12:34<1:06:39,  1.58it/s]
 14%|█▎        | 1004/7308 [12:35<1:06:38,  1.58it/s]
 14%|█▍        | 1005/7308 [12:35<1:06:36,  1.58it/s]
 14%|█▍        | 1006/7308 [12:36<1:06:37,  1.58it/s]
 14%|█▍        | 1007/7308 [12:37<1:06:35,  1.58it/s]
 14%|█▍        | 1008/7308 [12:37<54:16,  1.93it/s]  [transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 24.10it/s]
+
 15%|█▌        | 6/39 [00:00<00:01, 17.88it/s]
+
 21%|██        | 8/39 [00:00<00:01, 16.74it/s]
+
 26%|██▌       | 10/39 [00:00<00:01, 16.22it/s]
+
 31%|███       | 12/39 [00:00<00:01, 15.86it/s]
+
 36%|███▌      | 14/39 [00:00<00:01, 15.64it/s]
+
 41%|████      | 16/39 [00:00<00:01, 15.50it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 15.40it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 15.33it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 15.28it/s]
+
 62%|██████▏   | 24/39 [00:01<00:00, 15.27it/s]
+
 67%|██████▋   | 26/39 [00:01<00:00, 15.25it/s]
+
 72%|███████▏  | 28/39 [00:01<00:00, 15.24it/s]
+
 77%|███████▋  | 30/39 [00:01<00:00, 15.22it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 15.23it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 15.23it/s]
+
 92%|█████████▏| 36/39 [00:02<00:00, 15.23it/s]
+
 97%|█████████▋| 38/39 [00:02<00:00, 15.23it/s]
                                                   
+
                                               
{'eval_loss': '4.207', 'eval_runtime': '2.612', 'eval_samples_per_second': '934.5', 'eval_steps_per_second': '14.93', 'epoch': '28'}
+
 14%|█▍        | 1008/7308 [12:40<54:16,  1.93it/s]
+
100%|██████████| 39/39 [00:02<00:00, 15.23it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-1008
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-1008/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-1008/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.95it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.95it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-1008/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-1008/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-1008/tokenizer_config.json
+[eval] epoch 28.0 step 1008 loss 4.2069 (best 4.2069 @ ep 28.0, 0/20 since best)
+
+
                                               
 14%|█▍        | 1009/7308 [12:42<3:25:43,  1.96s/it]
 14%|█▍        | 1010/7308 [12:43<2:43:58,  1.56s/it]
                                                     
{'loss': '4.687', 'grad_norm': '0.3274', 'learning_rate': '0.0006', 'epoch': '28.06'}
+
 14%|█▍        | 1010/7308 [12:43<2:43:58,  1.56s/it]
 14%|█▍        | 1011/7308 [12:44<2:14:43,  1.28s/it]
 14%|█▍        | 1012/7308 [12:44<1:54:16,  1.09s/it]
 14%|█▍        | 1013/7308 [12:45<1:39:56,  1.05it/s]
 14%|█▍        | 1014/7308 [12:45<1:29:54,  1.17it/s]
 14%|█▍        | 1015/7308 [12:46<1:22:52,  1.27it/s]
 14%|█▍        | 1016/7308 [12:47<1:17:57,  1.35it/s]
 14%|█▍        | 1017/7308 [12:47<1:14:29,  1.41it/s]
 14%|█▍        | 1018/7308 [12:48<1:12:07,  1.45it/s]
 14%|█▍        | 1019/7308 [12:49<1:10:25,  1.49it/s]
 14%|█▍        | 1020/7308 [12:49<1:09:13,  1.51it/s]
                                                     
{'loss': '4.617', 'grad_norm': '0.2903', 'learning_rate': '0.0006', 'epoch': '28.34'}
+
 14%|█▍        | 1020/7308 [12:49<1:09:13,  1.51it/s]
 14%|█▍        | 1021/7308 [12:50<1:08:25,  1.53it/s]
 14%|█▍        | 1022/7308 [12:50<1:07:51,  1.54it/s]
 14%|█▍        | 1023/7308 [12:51<1:07:24,  1.55it/s]
 14%|█▍        | 1024/7308 [12:52<1:07:05,  1.56it/s]
 14%|█▍        | 1025/7308 [12:52<1:06:51,  1.57it/s]
 14%|█▍        | 1026/7308 [12:53<1:06:42,  1.57it/s]
 14%|█▍        | 1027/7308 [12:54<1:06:35,  1.57it/s]
 14%|█▍        | 1028/7308 [12:54<1:06:30,  1.57it/s]
 14%|█▍        | 1029/7308 [12:55<1:06:26,  1.58it/s]
 14%|█▍        | 1030/7308 [12:56<1:06:23,  1.58it/s]
                                                     
{'loss': '4.595', 'grad_norm': '0.217', 'learning_rate': '0.0006', 'epoch': '28.62'}
+
 14%|█▍        | 1030/7308 [12:56<1:06:23,  1.58it/s]
 14%|█▍        | 1031/7308 [12:56<1:06:23,  1.58it/s]
 14%|█▍        | 1032/7308 [12:57<1:06:21,  1.58it/s]
 14%|█▍        | 1033/7308 [12:57<1:06:19,  1.58it/s]
 14%|█▍        | 1034/7308 [12:58<1:06:16,  1.58it/s]
 14%|█▍        | 1035/7308 [12:59<1:06:16,  1.58it/s]
 14%|█▍        | 1036/7308 [12:59<1:06:14,  1.58it/s]
 14%|█▍        | 1037/7308 [13:00<1:06:13,  1.58it/s]
 14%|█▍        | 1038/7308 [13:01<1:06:12,  1.58it/s]
 14%|█▍        | 1039/7308 [13:01<1:06:13,  1.58it/s]
 14%|█▍        | 1040/7308 [13:02<1:06:13,  1.58it/s]
                                                     
{'loss': '4.626', 'grad_norm': '0.2557', 'learning_rate': '0.0006', 'epoch': '28.9'}
+
 14%|█▍        | 1040/7308 [13:02<1:06:13,  1.58it/s]
 14%|█▍        | 1041/7308 [13:03<1:06:13,  1.58it/s]
 14%|█▍        | 1042/7308 [13:03<1:06:14,  1.58it/s]
 14%|█▍        | 1043/7308 [13:04<1:06:10,  1.58it/s]
 14%|█▍        | 1044/7308 [13:04<53:55,  1.94it/s]  
 14%|█▍        | 1045/7308 [13:05<1:00:44,  1.72it/s]
 14%|█▍        | 1046/7308 [13:05<1:02:21,  1.67it/s]
 14%|█▍        | 1047/7308 [13:06<1:03:27,  1.64it/s]
 14%|█▍        | 1048/7308 [13:07<1:04:18,  1.62it/s]
 14%|█▍        | 1049/7308 [13:07<1:04:49,  1.61it/s]
 14%|█▍        | 1050/7308 [13:08<1:05:12,  1.60it/s]
                                                     
{'loss': '4.587', 'grad_norm': '0.4024', 'learning_rate': '0.0006', 'epoch': '29.17'}
+
 14%|█▍        | 1050/7308 [13:08<1:05:12,  1.60it/s]
 14%|█▍        | 1051/7308 [13:09<1:05:29,  1.59it/s]
 14%|█▍        | 1052/7308 [13:09<1:05:39,  1.59it/s]
 14%|█▍        | 1053/7308 [13:10<1:05:46,  1.59it/s]
 14%|█▍        | 1054/7308 [13:10<1:05:51,  1.58it/s]
 14%|█▍        | 1055/7308 [13:11<1:05:54,  1.58it/s]
 14%|█▍        | 1056/7308 [13:12<1:05:57,  1.58it/s]
 14%|█▍        | 1057/7308 [13:12<1:05:58,  1.58it/s]
 14%|█▍        | 1058/7308 [13:13<1:05:58,  1.58it/s]
 14%|█▍        | 1059/7308 [13:14<1:06:00,  1.58it/s]
 15%|█▍        | 1060/7308 [13:14<1:06:00,  1.58it/s]
                                                     
{'loss': '4.644', 'grad_norm': '3.627', 'learning_rate': '0.0006', 'epoch': '29.45'}
+
 15%|█▍        | 1060/7308 [13:14<1:06:00,  1.58it/s]
 15%|█▍        | 1061/7308 [13:15<1:06:00,  1.58it/s]
 15%|█▍        | 1062/7308 [13:16<1:05:59,  1.58it/s]
 15%|█▍        | 1063/7308 [13:16<1:06:00,  1.58it/s]
 15%|█▍        | 1064/7308 [13:17<1:05:58,  1.58it/s]
 15%|█▍        | 1065/7308 [13:17<1:05:59,  1.58it/s]
 15%|█▍        | 1066/7308 [13:18<1:05:57,  1.58it/s]
 15%|█▍        | 1067/7308 [13:19<1:05:57,  1.58it/s]
 15%|█▍        | 1068/7308 [13:19<1:05:57,  1.58it/s]
 15%|█▍        | 1069/7308 [13:20<1:05:57,  1.58it/s]
 15%|█▍        | 1070/7308 [13:21<1:05:56,  1.58it/s]
                                                     
{'loss': '4.967', 'grad_norm': '1.527', 'learning_rate': '0.0006', 'epoch': '29.73'}
+
 15%|█▍        | 1070/7308 [13:21<1:05:56,  1.58it/s]
 15%|█▍        | 1071/7308 [13:21<1:05:55,  1.58it/s]
 15%|█▍        | 1072/7308 [13:22<1:05:54,  1.58it/s]
 15%|█▍        | 1073/7308 [13:23<1:05:55,  1.58it/s]
 15%|█▍        | 1074/7308 [13:23<1:05:54,  1.58it/s]
 15%|█▍        | 1075/7308 [13:24<1:05:53,  1.58it/s]
 15%|█▍        | 1076/7308 [13:24<1:05:52,  1.58it/s]
 15%|█▍        | 1077/7308 [13:25<1:05:51,  1.58it/s]
 15%|█▍        | 1078/7308 [13:26<1:05:49,  1.58it/s]
 15%|█▍        | 1079/7308 [13:26<1:05:48,  1.58it/s]
 15%|█▍        | 1080/7308 [13:27<53:38,  1.94it/s]  
                                                   
{'loss': '5.059', 'grad_norm': '0.9516', 'learning_rate': '0.0006', 'epoch': '30'}
+
 15%|█▍        | 1080/7308 [13:27<53:38,  1.94it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 23.39it/s]
+
 15%|█▌        | 6/39 [00:00<00:01, 17.74it/s]
+
 21%|██        | 8/39 [00:00<00:01, 16.66it/s]
+
 26%|██▌       | 10/39 [00:00<00:01, 16.13it/s]
+
 31%|███       | 12/39 [00:00<00:01, 15.81it/s]
+
 36%|███▌      | 14/39 [00:00<00:01, 15.61it/s]
+
 41%|████      | 16/39 [00:00<00:01, 15.48it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 15.39it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 15.31it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 15.29it/s]
+
 62%|██████▏   | 24/39 [00:01<00:00, 15.29it/s]
+
 67%|██████▋   | 26/39 [00:01<00:00, 15.28it/s]
+
 72%|███████▏  | 28/39 [00:01<00:00, 15.24it/s]
+
 77%|███████▋  | 30/39 [00:01<00:00, 15.24it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 15.22it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 15.22it/s]
+
 92%|█████████▏| 36/39 [00:02<00:00, 15.21it/s]
+
 97%|█████████▋| 38/39 [00:02<00:00, 15.22it/s]
                                                   
+
                                               
{'eval_loss': '4.588', 'eval_runtime': '2.621', 'eval_samples_per_second': '931.3', 'eval_steps_per_second': '14.88', 'epoch': '30'}
+
 15%|█▍        | 1080/7308 [13:29<53:38,  1.94it/s]
+
100%|██████████| 39/39 [00:02<00:00, 15.22it/s][eval] epoch 30.0 step 1080 loss 4.5883 (best 4.2069 @ ep 28.0, 1/20 since best)
+
+
                                               
 15%|█▍        | 1081/7308 [13:30<2:20:59,  1.36s/it]
 15%|█▍        | 1082/7308 [13:31<1:58:24,  1.14s/it]
 15%|█▍        | 1083/7308 [13:31<1:42:35,  1.01it/s]
 15%|█▍        | 1084/7308 [13:32<1:31:30,  1.13it/s]
 15%|█▍        | 1085/7308 [13:32<1:23:46,  1.24it/s]
 15%|█▍        | 1086/7308 [13:33<1:18:20,  1.32it/s]
 15%|█▍        | 1087/7308 [13:34<1:14:32,  1.39it/s]
 15%|█▍        | 1088/7308 [13:34<1:11:52,  1.44it/s]
 15%|█▍        | 1089/7308 [13:35<1:10:01,  1.48it/s]
 15%|█▍        | 1090/7308 [13:36<1:08:42,  1.51it/s]
                                                     
{'loss': '4.987', 'grad_norm': '0.3607', 'learning_rate': '0.0006', 'epoch': '30.28'}
+
 15%|█▍        | 1090/7308 [13:36<1:08:42,  1.51it/s]
 15%|█▍        | 1091/7308 [13:36<1:07:59,  1.52it/s]
 15%|█▍        | 1092/7308 [13:37<1:07:19,  1.54it/s]
 15%|█▍        | 1093/7308 [13:38<1:06:50,  1.55it/s]
 15%|█▍        | 1094/7308 [13:38<1:06:29,  1.56it/s]
 15%|█▍        | 1095/7308 [13:39<1:06:16,  1.56it/s]
 15%|█▍        | 1096/7308 [13:39<1:06:09,  1.57it/s]
 15%|█▌        | 1097/7308 [13:40<1:06:02,  1.57it/s]
 15%|█▌        | 1098/7308 [13:41<1:05:54,  1.57it/s]
 15%|█▌        | 1099/7308 [13:41<1:05:49,  1.57it/s]
 15%|█▌        | 1100/7308 [13:42<1:05:45,  1.57it/s]
                                                     
{'loss': '4.87', 'grad_norm': '0.2092', 'learning_rate': '0.0006', 'epoch': '30.57'}
+
 15%|█▌        | 1100/7308 [13:42<1:05:45,  1.57it/s]
 15%|█▌        | 1101/7308 [13:43<1:05:44,  1.57it/s]
 15%|█▌        | 1102/7308 [13:43<1:05:42,  1.57it/s]
 15%|█▌        | 1103/7308 [13:44<1:05:40,  1.57it/s]
 15%|█▌        | 1104/7308 [13:45<1:05:40,  1.57it/s]
 15%|█▌        | 1105/7308 [13:45<1:05:38,  1.57it/s]
 15%|█▌        | 1106/7308 [13:46<1:05:36,  1.58it/s]
 15%|█▌        | 1107/7308 [13:46<1:05:35,  1.58it/s]
 15%|█▌        | 1108/7308 [13:47<1:05:33,  1.58it/s]
 15%|█▌        | 1109/7308 [13:48<1:05:33,  1.58it/s]
 15%|█▌        | 1110/7308 [13:48<1:05:32,  1.58it/s]
                                                     
{'loss': '4.767', 'grad_norm': '0.1669', 'learning_rate': '0.0006', 'epoch': '30.85'}
+
 15%|█▌        | 1110/7308 [13:48<1:05:32,  1.58it/s]
 15%|█▌        | 1111/7308 [13:49<1:05:32,  1.58it/s]
 15%|█▌        | 1112/7308 [13:50<1:05:31,  1.58it/s]
 15%|█▌        | 1113/7308 [13:50<1:05:31,  1.58it/s]
 15%|█▌        | 1114/7308 [13:51<1:05:29,  1.58it/s]
 15%|█▌        | 1115/7308 [13:51<1:05:26,  1.58it/s]
 15%|█▌        | 1116/7308 [13:52<53:20,  1.93it/s]  
 15%|█▌        | 1117/7308 [13:52<59:53,  1.72it/s]
 15%|█▌        | 1118/7308 [13:53<1:01:34,  1.68it/s]
 15%|█▌        | 1119/7308 [13:54<1:02:45,  1.64it/s]
 15%|█▌        | 1120/7308 [13:54<1:03:35,  1.62it/s]
                                                     
{'loss': '4.668', 'grad_norm': '0.1598', 'learning_rate': '0.0006', 'epoch': '31.11'}
+
 15%|█▌        | 1120/7308 [13:54<1:03:35,  1.62it/s]
 15%|█▌        | 1121/7308 [13:55<1:04:10,  1.61it/s]
 15%|█▌        | 1122/7308 [13:56<1:04:33,  1.60it/s]
 15%|█▌        | 1123/7308 [13:56<1:04:50,  1.59it/s]
 15%|█▌        | 1124/7308 [13:57<1:05:01,  1.58it/s]
 15%|█▌        | 1125/7308 [13:58<1:05:09,  1.58it/s]
 15%|█▌        | 1126/7308 [13:58<1:05:14,  1.58it/s]
 15%|█▌        | 1127/7308 [13:59<1:05:17,  1.58it/s]
 15%|█▌        | 1128/7308 [13:59<1:05:19,  1.58it/s]
 15%|█▌        | 1129/7308 [14:00<1:05:21,  1.58it/s]
 15%|█▌        | 1130/7308 [14:01<1:05:24,  1.57it/s]
                                                     
{'loss': '4.63', 'grad_norm': '0.1622', 'learning_rate': '0.0006', 'epoch': '31.4'}
+
 15%|█▌        | 1130/7308 [14:01<1:05:24,  1.57it/s]
 15%|█▌        | 1131/7308 [14:01<1:05:24,  1.57it/s]
 15%|█▌        | 1132/7308 [14:02<1:05:24,  1.57it/s]
 16%|█▌        | 1133/7308 [14:03<1:05:22,  1.57it/s]
 16%|█▌        | 1134/7308 [14:03<1:05:23,  1.57it/s]
 16%|█▌        | 1135/7308 [14:04<1:05:23,  1.57it/s]
 16%|█▌        | 1136/7308 [14:05<1:05:22,  1.57it/s]
 16%|█▌        | 1137/7308 [14:05<1:05:21,  1.57it/s]
 16%|█▌        | 1138/7308 [14:06<1:05:19,  1.57it/s]
 16%|█▌        | 1139/7308 [14:06<1:05:20,  1.57it/s]
 16%|█▌        | 1140/7308 [14:07<1:05:18,  1.57it/s]
                                                     
{'loss': '4.585', 'grad_norm': '0.1536', 'learning_rate': '0.0006', 'epoch': '31.68'}
+
 16%|█▌        | 1140/7308 [14:07<1:05:18,  1.57it/s]
 16%|█▌        | 1141/7308 [14:08<1:05:20,  1.57it/s]
 16%|█▌        | 1142/7308 [14:08<1:05:19,  1.57it/s]
 16%|█▌        | 1143/7308 [14:09<1:05:18,  1.57it/s]
 16%|█▌        | 1144/7308 [14:10<1:05:17,  1.57it/s]
 16%|█▌        | 1145/7308 [14:10<1:05:16,  1.57it/s]
 16%|█▌        | 1146/7308 [14:11<1:05:16,  1.57it/s]
 16%|█▌        | 1147/7308 [14:12<1:05:15,  1.57it/s]
 16%|█▌        | 1148/7308 [14:12<1:05:15,  1.57it/s]
 16%|█▌        | 1149/7308 [14:13<1:05:14,  1.57it/s]
 16%|█▌        | 1150/7308 [14:13<1:05:14,  1.57it/s]
                                                     
{'loss': '4.475', 'grad_norm': '0.1597', 'learning_rate': '0.0006', 'epoch': '31.96'}
+
 16%|█▌        | 1150/7308 [14:13<1:05:14,  1.57it/s]
 16%|█▌        | 1151/7308 [14:14<1:05:11,  1.57it/s]
 16%|█▌        | 1152/7308 [14:14<53:06,  1.93it/s]  [transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 22.21it/s]
+
 15%|█▌        | 6/39 [00:00<00:01, 17.46it/s]
+
 21%|██        | 8/39 [00:00<00:01, 16.59it/s]
+
 26%|██▌       | 10/39 [00:00<00:01, 16.09it/s]
+
 31%|███       | 12/39 [00:00<00:01, 15.80it/s]
+
 36%|███▌      | 14/39 [00:00<00:01, 15.58it/s]
+
 41%|████      | 16/39 [00:00<00:01, 15.47it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 15.37it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 15.33it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 15.28it/s]
+
 62%|██████▏   | 24/39 [00:01<00:00, 15.29it/s]
+
 67%|██████▋   | 26/39 [00:01<00:00, 15.26it/s]
+
 72%|███████▏  | 28/39 [00:01<00:00, 15.25it/s]
+
 77%|███████▋  | 30/39 [00:01<00:00, 15.23it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 15.23it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 15.22it/s]
+
 92%|█████████▏| 36/39 [00:02<00:00, 15.23it/s]
+
 97%|█████████▋| 38/39 [00:02<00:00, 15.22it/s]
                                                   
+
                                               
{'eval_loss': '4.178', 'eval_runtime': '2.618', 'eval_samples_per_second': '932.4', 'eval_steps_per_second': '14.9', 'epoch': '32'}
+
 16%|█▌        | 1152/7308 [14:17<53:06,  1.93it/s]
+
100%|██████████| 39/39 [00:02<00:00, 15.22it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-1152
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-1152/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-1152/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.96it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.95it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-1152/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-1152/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-1152/tokenizer_config.json
+[eval] epoch 32.0 step 1152 loss 4.1783 (best 4.1783 @ ep 32.0, 0/20 since best)
+
+
                                               
 16%|█▌        | 1153/7308 [14:20<3:21:09,  1.96s/it]
 16%|█▌        | 1154/7308 [14:20<2:40:18,  1.56s/it]
 16%|█▌        | 1155/7308 [14:21<2:11:43,  1.28s/it]
 16%|█▌        | 1156/7308 [14:22<1:51:43,  1.09s/it]
 16%|█▌        | 1157/7308 [14:22<1:37:44,  1.05it/s]
 16%|█▌        | 1158/7308 [14:23<1:27:58,  1.17it/s]
 16%|█▌        | 1159/7308 [14:23<1:21:07,  1.26it/s]
 16%|█▌        | 1160/7308 [14:24<1:16:17,  1.34it/s]
                                                     
{'loss': '4.634', 'grad_norm': '0.1617', 'learning_rate': '0.0006', 'epoch': '32.23'}
+
 16%|█▌        | 1160/7308 [14:24<1:16:17,  1.34it/s]
 16%|█▌        | 1161/7308 [14:25<1:12:57,  1.40it/s]
 16%|█▌        | 1162/7308 [14:25<1:10:39,  1.45it/s]
 16%|█▌        | 1163/7308 [14:26<1:08:58,  1.48it/s]
 16%|█▌        | 1164/7308 [14:27<1:07:47,  1.51it/s]
 16%|█▌        | 1165/7308 [14:27<1:06:57,  1.53it/s]
 16%|█▌        | 1166/7308 [14:28<1:06:24,  1.54it/s]
 16%|█▌        | 1167/7308 [14:29<1:05:58,  1.55it/s]
 16%|█▌        | 1168/7308 [14:29<1:05:40,  1.56it/s]
 16%|█▌        | 1169/7308 [14:30<1:05:27,  1.56it/s]
 16%|█▌        | 1170/7308 [14:30<1:05:19,  1.57it/s]
                                                     
{'loss': '4.524', 'grad_norm': '0.1501', 'learning_rate': '0.0006', 'epoch': '32.51'}
+
 16%|█▌        | 1170/7308 [14:30<1:05:19,  1.57it/s]
 16%|█▌        | 1171/7308 [14:31<1:05:14,  1.57it/s]
 16%|█▌        | 1172/7308 [14:32<1:05:09,  1.57it/s]
 16%|█▌        | 1173/7308 [14:32<1:05:06,  1.57it/s]
 16%|█▌        | 1174/7308 [14:33<1:05:03,  1.57it/s]
 16%|█▌        | 1175/7308 [14:34<1:05:02,  1.57it/s]
 16%|█▌        | 1176/7308 [14:34<1:05:00,  1.57it/s]
 16%|█▌        | 1177/7308 [14:35<1:04:58,  1.57it/s]
 16%|█▌        | 1178/7308 [14:36<1:04:56,  1.57it/s]
 16%|█▌        | 1179/7308 [14:36<1:04:56,  1.57it/s]
 16%|█▌        | 1180/7308 [14:37<1:04:56,  1.57it/s]
                                                     
{'loss': '4.56', 'grad_norm': '0.1724', 'learning_rate': '0.0006', 'epoch': '32.79'}
+
 16%|█▌        | 1180/7308 [14:37<1:04:56,  1.57it/s]
 16%|█▌        | 1181/7308 [14:37<1:04:56,  1.57it/s]
 16%|█▌        | 1182/7308 [14:38<1:04:56,  1.57it/s]
 16%|█▌        | 1183/7308 [14:39<1:04:54,  1.57it/s]
 16%|█▌        | 1184/7308 [14:39<1:04:53,  1.57it/s]
 16%|█▌        | 1185/7308 [14:40<1:04:52,  1.57it/s]
 16%|█▌        | 1186/7308 [14:41<1:04:53,  1.57it/s]
 16%|█▌        | 1187/7308 [14:41<1:04:46,  1.57it/s]
 16%|█▋        | 1188/7308 [14:41<52:47,  1.93it/s]  
 16%|█▋        | 1189/7308 [14:42<59:22,  1.72it/s]
 16%|█▋        | 1190/7308 [14:43<1:00:59,  1.67it/s]
                                                     
{'loss': '4.57', 'grad_norm': '0.191', 'learning_rate': '0.0006', 'epoch': '33.06'}
+
 16%|█▋        | 1190/7308 [14:43<1:00:59,  1.67it/s]
 16%|█▋        | 1191/7308 [14:43<1:02:07,  1.64it/s]
 16%|█▋        | 1192/7308 [14:44<1:02:53,  1.62it/s]
 16%|█▋        | 1193/7308 [14:45<1:03:26,  1.61it/s]
 16%|█▋        | 1194/7308 [14:45<1:03:50,  1.60it/s]
 16%|█▋        | 1195/7308 [14:46<1:04:07,  1.59it/s]
 16%|█▋        | 1196/7308 [14:47<1:04:20,  1.58it/s]
 16%|█▋        | 1197/7308 [14:47<1:04:25,  1.58it/s]
 16%|█▋        | 1198/7308 [14:48<1:04:29,  1.58it/s]
 16%|█▋        | 1199/7308 [14:49<1:04:32,  1.58it/s]
 16%|█▋        | 1200/7308 [14:49<1:04:34,  1.58it/s]
                                                     
{'loss': '4.507', 'grad_norm': '0.1742', 'learning_rate': '0.0006', 'epoch': '33.34'}
+
 16%|█▋        | 1200/7308 [14:49<1:04:34,  1.58it/s]
 16%|█▋        | 1201/7308 [14:50<1:04:35,  1.58it/s]
 16%|█▋        | 1202/7308 [14:50<1:04:36,  1.57it/s]
 16%|█▋        | 1203/7308 [14:51<1:04:37,  1.57it/s]
 16%|█▋        | 1204/7308 [14:52<1:04:37,  1.57it/s]
 16%|█▋        | 1205/7308 [14:52<1:04:36,  1.57it/s]
 17%|█▋        | 1206/7308 [14:53<1:04:37,  1.57it/s]
 17%|█▋        | 1207/7308 [14:54<1:04:36,  1.57it/s]
 17%|█▋        | 1208/7308 [14:54<1:04:36,  1.57it/s]
 17%|█▋        | 1209/7308 [14:55<1:04:35,  1.57it/s]
 17%|█▋        | 1210/7308 [14:56<1:04:34,  1.57it/s]
                                                     
{'loss': '4.504', 'grad_norm': '0.1736', 'learning_rate': '0.0006', 'epoch': '33.62'}
+
 17%|█▋        | 1210/7308 [14:56<1:04:34,  1.57it/s]
 17%|█▋        | 1211/7308 [14:56<1:04:35,  1.57it/s]
 17%|█▋        | 1212/7308 [14:57<1:04:33,  1.57it/s]
 17%|█▋        | 1213/7308 [14:57<1:04:33,  1.57it/s]
 17%|█▋        | 1214/7308 [14:58<1:04:33,  1.57it/s]
 17%|█▋        | 1215/7308 [14:59<1:04:33,  1.57it/s]
 17%|█▋        | 1216/7308 [14:59<1:04:31,  1.57it/s]
 17%|█▋        | 1217/7308 [15:00<1:04:31,  1.57it/s]
 17%|█▋        | 1218/7308 [15:01<1:04:30,  1.57it/s]
 17%|█▋        | 1219/7308 [15:01<1:04:29,  1.57it/s]
 17%|█▋        | 1220/7308 [15:02<1:04:28,  1.57it/s]
                                                     
{'loss': '4.474', 'grad_norm': '0.1833', 'learning_rate': '0.0006', 'epoch': '33.9'}
+
 17%|█▋        | 1220/7308 [15:02<1:04:28,  1.57it/s]
 17%|█▋        | 1221/7308 [15:03<1:04:30,  1.57it/s]
 17%|█▋        | 1222/7308 [15:03<1:04:30,  1.57it/s]
 17%|█▋        | 1223/7308 [15:04<1:04:24,  1.57it/s]
 17%|█▋        | 1224/7308 [15:04<52:27,  1.93it/s]  [transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 22.68it/s]
+
 15%|█▌        | 6/39 [00:00<00:01, 17.57it/s]
+
 21%|██        | 8/39 [00:00<00:01, 16.61it/s]
+
 26%|██▌       | 10/39 [00:00<00:01, 16.11it/s]
+
 31%|███       | 12/39 [00:00<00:01, 15.79it/s]
+
 36%|███▌      | 14/39 [00:00<00:01, 15.60it/s]
+
 41%|████      | 16/39 [00:00<00:01, 15.45it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 15.37it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 15.31it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 15.28it/s]
+
 62%|██████▏   | 24/39 [00:01<00:00, 15.26it/s]
+
 67%|██████▋   | 26/39 [00:01<00:00, 15.25it/s]
+
 72%|███████▏  | 28/39 [00:01<00:00, 15.24it/s]
+
 77%|███████▋  | 30/39 [00:01<00:00, 15.24it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 15.22it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 15.22it/s]
+
 92%|█████████▏| 36/39 [00:02<00:00, 15.21it/s]
+
 97%|█████████▋| 38/39 [00:02<00:00, 15.22it/s]
                                                   
+
                                               
{'eval_loss': '4.138', 'eval_runtime': '2.623', 'eval_samples_per_second': '930.7', 'eval_steps_per_second': '14.87', 'epoch': '34'}
+
 17%|█▋        | 1224/7308 [15:07<52:27,  1.93it/s]
+
100%|██████████| 39/39 [00:02<00:00, 15.22it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-1224
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-1224/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-1224/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.95it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.94it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-1224/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-1224/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-1224/tokenizer_config.json
+[eval] epoch 34.0 step 1224 loss 4.1376 (best 4.1376 @ ep 34.0, 0/20 since best)
+
+
                                               
 17%|█▋        | 1225/7308 [15:09<3:19:04,  1.96s/it]
 17%|█▋        | 1226/7308 [15:10<2:38:37,  1.56s/it]
 17%|█▋        | 1227/7308 [15:11<2:10:20,  1.29s/it]
 17%|█▋        | 1228/7308 [15:11<1:50:33,  1.09s/it]
 17%|█▋        | 1229/7308 [15:12<1:36:41,  1.05it/s]
 17%|█▋        | 1230/7308 [15:13<1:26:59,  1.16it/s]
                                                     
{'loss': '4.479', 'grad_norm': '0.1985', 'learning_rate': '0.0006', 'epoch': '34.17'}
+
 17%|█▋        | 1230/7308 [15:13<1:26:59,  1.16it/s]
 17%|█▋        | 1231/7308 [15:13<1:20:12,  1.26it/s]
 17%|█▋        | 1232/7308 [15:14<1:15:26,  1.34it/s]
 17%|█▋        | 1233/7308 [15:14<1:12:05,  1.40it/s]
 17%|█▋        | 1234/7308 [15:15<1:09:44,  1.45it/s]
 17%|█▋        | 1235/7308 [15:16<1:18:40,  1.29it/s]
 17%|█▋        | 1236/7308 [15:17<1:14:20,  1.36it/s]
 17%|█▋        | 1237/7308 [15:18<1:21:45,  1.24it/s]
 17%|█▋        | 1238/7308 [15:18<1:16:28,  1.32it/s]
 17%|█▋        | 1239/7308 [15:19<1:12:49,  1.39it/s]
 17%|█▋        | 1240/7308 [15:20<1:10:14,  1.44it/s]
                                                     
{'loss': '4.458', 'grad_norm': '0.1795', 'learning_rate': '0.0006', 'epoch': '34.45'}
+
 17%|█▋        | 1240/7308 [15:20<1:10:14,  1.44it/s]
 17%|█▋        | 1241/7308 [15:20<1:08:26,  1.48it/s]
 17%|█▋        | 1242/7308 [15:21<1:07:09,  1.51it/s]
 17%|█▋        | 1243/7308 [15:22<1:06:17,  1.52it/s]
 17%|█▋        | 1244/7308 [15:22<1:05:39,  1.54it/s]
 17%|█▋        | 1245/7308 [15:23<1:05:12,  1.55it/s]
 17%|█▋        | 1246/7308 [15:23<1:04:54,  1.56it/s]
 17%|█▋        | 1247/7308 [15:24<1:04:41,  1.56it/s]
 17%|█▋        | 1248/7308 [15:25<1:04:31,  1.57it/s]
 17%|█▋        | 1249/7308 [15:25<1:04:24,  1.57it/s]
 17%|█▋        | 1250/7308 [15:26<1:04:18,  1.57it/s]
                                                     
{'loss': '4.493', 'grad_norm': '0.2348', 'learning_rate': '0.0006', 'epoch': '34.73'}
+
 17%|█▋        | 1250/7308 [15:26<1:04:18,  1.57it/s]
 17%|█▋        | 1251/7308 [15:27<1:04:17,  1.57it/s]
 17%|█▋        | 1252/7308 [15:27<1:04:14,  1.57it/s]
 17%|█▋        | 1253/7308 [15:28<1:04:12,  1.57it/s]
 17%|█▋        | 1254/7308 [15:29<1:04:10,  1.57it/s]
 17%|█▋        | 1255/7308 [15:29<1:04:09,  1.57it/s]
 17%|█▋        | 1256/7308 [15:30<1:04:04,  1.57it/s]
 17%|█▋        | 1257/7308 [15:30<1:04:03,  1.57it/s]
 17%|█▋        | 1258/7308 [15:31<1:04:00,  1.58it/s]
 17%|█▋        | 1259/7308 [15:32<1:03:56,  1.58it/s]
 17%|█▋        | 1260/7308 [15:32<52:06,  1.93it/s]  
                                                   
{'loss': '4.493', 'grad_norm': '0.2571', 'learning_rate': '0.0006', 'epoch': '35'}
+
 17%|█▋        | 1260/7308 [15:32<52:06,  1.93it/s]
 17%|█▋        | 1261/7308 [15:33<58:40,  1.72it/s]
 17%|█▋        | 1262/7308 [15:33<1:00:17,  1.67it/s]
 17%|█▋        | 1263/7308 [15:34<1:01:23,  1.64it/s]
 17%|█▋        | 1264/7308 [15:35<1:02:08,  1.62it/s]
 17%|█▋        | 1265/7308 [15:35<1:02:38,  1.61it/s]
 17%|█▋        | 1266/7308 [15:36<1:03:00,  1.60it/s]
 17%|█▋        | 1267/7308 [15:36<1:03:25,  1.59it/s]
 17%|█▋        | 1268/7308 [15:37<1:03:33,  1.58it/s]
 17%|█▋        | 1269/7308 [15:38<1:03:39,  1.58it/s]
 17%|█▋        | 1270/7308 [15:38<1:03:42,  1.58it/s]
                                                     
{'loss': '4.476', 'grad_norm': '0.2223', 'learning_rate': '0.0006', 'epoch': '35.28'}
+
 17%|█▋        | 1270/7308 [15:38<1:03:42,  1.58it/s]
 17%|█▋        | 1271/7308 [15:39<1:03:46,  1.58it/s]
 17%|█▋        | 1272/7308 [15:40<1:03:47,  1.58it/s]
 17%|█▋        | 1273/7308 [15:40<1:03:47,  1.58it/s]
 17%|█▋        | 1274/7308 [15:41<1:03:48,  1.58it/s]
 17%|█▋        | 1275/7308 [15:42<1:03:48,  1.58it/s]
 17%|█▋        | 1276/7308 [15:42<1:03:47,  1.58it/s]
 17%|█▋        | 1277/7308 [15:43<1:03:47,  1.58it/s]
 17%|█▋        | 1278/7308 [15:43<1:03:46,  1.58it/s]
 18%|█▊        | 1279/7308 [15:44<1:03:46,  1.58it/s]
 18%|█▊        | 1280/7308 [15:45<1:03:46,  1.58it/s]
                                                     
{'loss': '4.479', 'grad_norm': '0.2149', 'learning_rate': '0.0006', 'epoch': '35.57'}
+
 18%|█▊        | 1280/7308 [15:45<1:03:46,  1.58it/s]
 18%|█▊        | 1281/7308 [15:45<1:03:46,  1.58it/s]
 18%|█▊        | 1282/7308 [15:46<1:03:45,  1.58it/s]
 18%|█▊        | 1283/7308 [15:47<1:03:55,  1.57it/s]
 18%|█▊        | 1284/7308 [15:47<1:03:52,  1.57it/s]
 18%|█▊        | 1285/7308 [15:48<1:03:49,  1.57it/s]
 18%|█▊        | 1286/7308 [15:49<1:03:46,  1.57it/s]
 18%|█▊        | 1287/7308 [15:49<1:03:44,  1.57it/s]
 18%|█▊        | 1288/7308 [15:50<1:03:42,  1.57it/s]
 18%|█▊        | 1289/7308 [15:50<1:03:41,  1.57it/s]
 18%|█▊        | 1290/7308 [15:51<1:03:41,  1.57it/s]
                                                     
{'loss': '4.452', 'grad_norm': '0.188', 'learning_rate': '0.0006', 'epoch': '35.85'}
+
 18%|█▊        | 1290/7308 [15:51<1:03:41,  1.57it/s]
 18%|█▊        | 1291/7308 [15:52<1:03:41,  1.57it/s]
 18%|█▊        | 1292/7308 [15:52<1:03:40,  1.57it/s]
 18%|█▊        | 1293/7308 [15:53<1:14:06,  1.35it/s]
 18%|█▊        | 1294/7308 [15:54<1:10:56,  1.41it/s]
 18%|█▊        | 1295/7308 [15:55<1:08:41,  1.46it/s]
 18%|█▊        | 1296/7308 [15:55<55:23,  1.81it/s]  [transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 23.25it/s]
+
 15%|█▌        | 6/39 [00:00<00:01, 17.70it/s]
+
 21%|██        | 8/39 [00:00<00:01, 16.69it/s]
+
 26%|██▌       | 10/39 [00:00<00:01, 16.14it/s]
+
 31%|███       | 12/39 [00:00<00:01, 15.82it/s]
+
 36%|███▌      | 14/39 [00:00<00:01, 15.60it/s]
+
 41%|████      | 16/39 [00:00<00:01, 15.48it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 15.38it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 15.32it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 15.26it/s]
+
 62%|██████▏   | 24/39 [00:01<00:00, 15.26it/s]
+
 67%|██████▋   | 26/39 [00:01<00:00, 15.25it/s]
+
 72%|███████▏  | 28/39 [00:01<00:00, 15.24it/s]
+
 77%|███████▋  | 30/39 [00:01<00:00, 15.23it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 15.23it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 15.21it/s]
+
 92%|█████████▏| 36/39 [00:02<00:00, 15.22it/s]
+
 97%|█████████▋| 38/39 [00:02<00:00, 15.22it/s]
                                                   
+
                                               
{'eval_loss': '4.117', 'eval_runtime': '2.628', 'eval_samples_per_second': '928.7', 'eval_steps_per_second': '14.84', 'epoch': '36'}
+
 18%|█▊        | 1296/7308 [15:57<55:23,  1.81it/s]
+
100%|██████████| 39/39 [00:02<00:00, 15.22it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-1296
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-1296/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-1296/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  2.19it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  2.19it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-1296/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-1296/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-1296/tokenizer_config.json
+[eval] epoch 36.0 step 1296 loss 4.1167 (best 4.1167 @ ep 36.0, 0/20 since best)
+
+
                                               
 18%|█▊        | 1297/7308 [16:00<3:17:34,  1.97s/it]
 18%|█▊        | 1298/7308 [16:01<2:37:21,  1.57s/it]
 18%|█▊        | 1299/7308 [16:01<2:09:12,  1.29s/it]
 18%|█▊        | 1300/7308 [16:02<1:49:29,  1.09s/it]
                                                     
{'loss': '4.412', 'grad_norm': '0.1996', 'learning_rate': '0.0006', 'epoch': '36.11'}
+
 18%|█▊        | 1300/7308 [16:02<1:49:29,  1.09s/it]
 18%|█▊        | 1301/7308 [16:03<1:35:43,  1.05it/s]
 18%|█▊        | 1302/7308 [16:03<1:26:04,  1.16it/s]
 18%|█▊        | 1303/7308 [16:04<1:19:19,  1.26it/s]
 18%|█▊        | 1304/7308 [16:05<1:14:35,  1.34it/s]
 18%|█▊        | 1305/7308 [16:05<1:11:16,  1.40it/s]
 18%|█▊        | 1306/7308 [16:06<1:08:57,  1.45it/s]
 18%|█▊        | 1307/7308 [16:06<1:07:19,  1.49it/s]
 18%|█▊        | 1308/7308 [16:07<1:06:10,  1.51it/s]
 18%|█▊        | 1309/7308 [16:08<1:05:21,  1.53it/s]
 18%|█▊        | 1310/7308 [16:08<1:04:47,  1.54it/s]
                                                     
{'loss': '4.456', 'grad_norm': '0.2329', 'learning_rate': '0.0006', 'epoch': '36.4'}
+
 18%|█▊        | 1310/7308 [16:08<1:04:47,  1.54it/s]
 18%|█▊        | 1311/7308 [16:09<1:14:52,  1.33it/s]
 18%|█▊        | 1312/7308 [16:10<1:11:26,  1.40it/s]
 18%|█▊        | 1313/7308 [16:11<1:09:02,  1.45it/s]
 18%|█▊        | 1314/7308 [16:11<1:07:20,  1.48it/s]
 18%|█▊        | 1315/7308 [16:12<1:16:41,  1.30it/s]
 18%|█▊        | 1316/7308 [16:13<1:12:42,  1.37it/s]
 18%|█▊        | 1317/7308 [16:14<1:09:54,  1.43it/s]
 18%|█▊        | 1318/7308 [16:14<1:07:56,  1.47it/s]
 18%|█▊        | 1319/7308 [16:15<1:06:36,  1.50it/s]
 18%|█▊        | 1320/7308 [16:15<1:05:38,  1.52it/s]
                                                     
{'loss': '4.485', 'grad_norm': '0.2393', 'learning_rate': '0.0006', 'epoch': '36.68'}
+
 18%|█▊        | 1320/7308 [16:15<1:05:38,  1.52it/s]
 18%|█▊        | 1321/7308 [16:16<1:04:58,  1.54it/s]
 18%|█▊        | 1322/7308 [16:17<1:04:28,  1.55it/s]
 18%|█▊        | 1323/7308 [16:18<1:14:27,  1.34it/s]
 18%|█▊        | 1324/7308 [16:18<1:11:07,  1.40it/s]
 18%|█▊        | 1325/7308 [16:19<1:08:46,  1.45it/s]
 18%|█▊        | 1326/7308 [16:20<1:07:06,  1.49it/s]
 18%|█▊        | 1327/7308 [16:20<1:05:57,  1.51it/s]
 18%|█▊        | 1328/7308 [16:21<1:05:09,  1.53it/s]
 18%|█▊        | 1329/7308 [16:22<1:04:36,  1.54it/s]
 18%|█▊        | 1330/7308 [16:22<1:04:15,  1.55it/s]
                                                     
{'loss': '4.436', 'grad_norm': '0.2215', 'learning_rate': '0.0006', 'epoch': '36.96'}
+
 18%|█▊        | 1330/7308 [16:22<1:04:15,  1.55it/s]
 18%|█▊        | 1331/7308 [16:23<1:03:54,  1.56it/s]
 18%|█▊        | 1332/7308 [16:23<51:59,  1.92it/s]  
 18%|█▊        | 1333/7308 [16:24<57:46,  1.72it/s]
 18%|█▊        | 1334/7308 [16:24<59:25,  1.68it/s]
 18%|█▊        | 1335/7308 [16:25<1:11:04,  1.40it/s]
 18%|█▊        | 1336/7308 [16:26<1:08:41,  1.45it/s]
 18%|█▊        | 1337/7308 [16:27<1:07:02,  1.48it/s]
 18%|█▊        | 1338/7308 [16:27<1:05:52,  1.51it/s]
 18%|█▊        | 1339/7308 [16:28<1:05:04,  1.53it/s]
 18%|█▊        | 1340/7308 [16:29<1:04:37,  1.54it/s]
                                                     
{'loss': '4.394', 'grad_norm': '0.2059', 'learning_rate': '0.0006', 'epoch': '37.23'}
+
 18%|█▊        | 1340/7308 [16:29<1:04:37,  1.54it/s]
 18%|█▊        | 1341/7308 [16:29<1:04:13,  1.55it/s]
 18%|█▊        | 1342/7308 [16:30<1:03:57,  1.55it/s]
 18%|█▊        | 1343/7308 [16:30<1:03:41,  1.56it/s]
 18%|█▊        | 1344/7308 [16:31<1:03:31,  1.56it/s]
 18%|█▊        | 1345/7308 [16:32<1:03:23,  1.57it/s]
 18%|█▊        | 1346/7308 [16:32<1:03:18,  1.57it/s]
 18%|█▊        | 1347/7308 [16:33<1:03:14,  1.57it/s]
 18%|█▊        | 1348/7308 [16:34<1:03:12,  1.57it/s]
 18%|█▊        | 1349/7308 [16:34<1:03:09,  1.57it/s]
 18%|█▊        | 1350/7308 [16:35<1:03:06,  1.57it/s]
                                                     
{'loss': '4.432', 'grad_norm': '0.2288', 'learning_rate': '0.0006', 'epoch': '37.51'}
+
 18%|█▊        | 1350/7308 [16:35<1:03:06,  1.57it/s]
 18%|█▊        | 1351/7308 [16:36<1:03:06,  1.57it/s]
 19%|█▊        | 1352/7308 [16:36<1:03:07,  1.57it/s]
 19%|█▊        | 1353/7308 [16:37<1:03:05,  1.57it/s]
 19%|█▊        | 1354/7308 [16:37<1:03:02,  1.57it/s]
 19%|█▊        | 1355/7308 [16:38<1:03:02,  1.57it/s]
 19%|█▊        | 1356/7308 [16:39<1:02:59,  1.57it/s]
 19%|█▊        | 1357/7308 [16:39<1:02:56,  1.58it/s]
 19%|█▊        | 1358/7308 [16:40<1:02:56,  1.58it/s]
 19%|█▊        | 1359/7308 [16:41<1:02:54,  1.58it/s]
 19%|█▊        | 1360/7308 [16:41<1:03:03,  1.57it/s]
                                                     
{'loss': '4.414', 'grad_norm': '0.1977', 'learning_rate': '0.0006', 'epoch': '37.79'}
+
 19%|█▊        | 1360/7308 [16:41<1:03:03,  1.57it/s]
 19%|█▊        | 1361/7308 [16:42<1:03:01,  1.57it/s]
 19%|█▊        | 1362/7308 [16:43<1:02:59,  1.57it/s]
 19%|█▊        | 1363/7308 [16:43<1:02:58,  1.57it/s]
 19%|█▊        | 1364/7308 [16:44<1:02:59,  1.57it/s]
 19%|█▊        | 1365/7308 [16:44<1:02:58,  1.57it/s]
 19%|█▊        | 1366/7308 [16:45<1:02:58,  1.57it/s]
 19%|█▊        | 1367/7308 [16:46<1:02:51,  1.58it/s]
 19%|█▊        | 1368/7308 [16:46<51:13,  1.93it/s]  [transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 23.27it/s]
+
 15%|█▌        | 6/39 [00:00<00:01, 17.68it/s]
+
 21%|██        | 8/39 [00:00<00:01, 16.70it/s]
+
 26%|██▌       | 10/39 [00:00<00:01, 16.13it/s]
+
 31%|███       | 12/39 [00:00<00:01, 15.82it/s]
+
 36%|███▌      | 14/39 [00:00<00:01, 15.57it/s]
+
 41%|████      | 16/39 [00:00<00:01, 15.45it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 15.34it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 15.30it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 15.25it/s]
+
 62%|██████▏   | 24/39 [00:01<00:00, 15.28it/s]
+
 67%|██████▋   | 26/39 [00:01<00:00, 15.25it/s]
+
 72%|███████▏  | 28/39 [00:01<00:00, 15.24it/s]
+
 77%|███████▋  | 30/39 [00:01<00:00, 15.22it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 15.23it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 15.22it/s]
+
 92%|█████████▏| 36/39 [00:02<00:00, 15.23it/s]
+
 97%|█████████▋| 38/39 [00:02<00:00, 15.22it/s]
                                                   
+
                                               
{'eval_loss': '4.105', 'eval_runtime': '2.622', 'eval_samples_per_second': '931.1', 'eval_steps_per_second': '14.88', 'epoch': '38'}
+
 19%|█▊        | 1368/7308 [16:49<51:13,  1.93it/s]
+
100%|██████████| 39/39 [00:02<00:00, 15.22it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-1368
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-1368/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-1368/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.73it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.72it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-1368/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-1368/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-1368/tokenizer_config.json
+[eval] epoch 38.0 step 1368 loss 4.1049 (best 4.1049 @ ep 38.0, 0/20 since best)
+
+
                                               
 19%|█▊        | 1369/7308 [16:51<3:16:03,  1.98s/it]
 19%|█▊        | 1370/7308 [16:52<2:36:01,  1.58s/it]
                                                     
{'loss': '4.406', 'grad_norm': '0.2471', 'learning_rate': '0.0006', 'epoch': '38.06'}
+
 19%|█▊        | 1370/7308 [16:52<2:36:01,  1.58s/it]
 19%|█▉        | 1371/7308 [16:53<2:08:01,  1.29s/it]
 19%|█▉        | 1372/7308 [16:53<1:48:26,  1.10s/it]
 19%|█▉        | 1373/7308 [16:54<1:34:45,  1.04it/s]
 19%|█▉        | 1374/7308 [16:55<1:25:08,  1.16it/s]
 19%|█▉        | 1375/7308 [16:55<1:18:24,  1.26it/s]
 19%|█▉        | 1376/7308 [16:56<1:13:41,  1.34it/s]
 19%|█▉        | 1377/7308 [16:56<1:10:24,  1.40it/s]
 19%|█▉        | 1378/7308 [16:57<1:08:09,  1.45it/s]
 19%|█▉        | 1379/7308 [16:58<1:06:31,  1.49it/s]
 19%|█▉        | 1380/7308 [16:58<1:05:21,  1.51it/s]
                                                     
{'loss': '4.331', 'grad_norm': '0.2457', 'learning_rate': '0.0006', 'epoch': '38.34'}
+
 19%|█▉        | 1380/7308 [16:58<1:05:21,  1.51it/s]
 19%|█▉        | 1381/7308 [16:59<1:04:32,  1.53it/s]
 19%|█▉        | 1382/7308 [17:00<1:03:56,  1.54it/s]
 19%|█▉        | 1383/7308 [17:00<1:03:32,  1.55it/s]
 19%|█▉        | 1384/7308 [17:01<1:03:14,  1.56it/s]
 19%|█▉        | 1385/7308 [17:01<1:03:00,  1.57it/s]
 19%|█▉        | 1386/7308 [17:02<1:02:52,  1.57it/s]
 19%|█▉        | 1387/7308 [17:03<1:02:45,  1.57it/s]
 19%|█▉        | 1388/7308 [17:03<1:02:40,  1.57it/s]
 19%|█▉        | 1389/7308 [17:04<1:02:35,  1.58it/s]
 19%|█▉        | 1390/7308 [17:05<1:02:34,  1.58it/s]
                                                     
{'loss': '4.442', 'grad_norm': '0.2524', 'learning_rate': '0.0006', 'epoch': '38.62'}
+
 19%|█▉        | 1390/7308 [17:05<1:02:34,  1.58it/s]
 19%|█▉        | 1391/7308 [17:05<1:02:35,  1.58it/s]
 19%|█▉        | 1392/7308 [17:06<1:02:31,  1.58it/s]
 19%|█▉        | 1393/7308 [17:07<1:02:31,  1.58it/s]
 19%|█▉        | 1394/7308 [17:07<1:02:30,  1.58it/s]
 19%|█▉        | 1395/7308 [17:08<1:02:27,  1.58it/s]
 19%|█▉        | 1396/7308 [17:08<1:02:26,  1.58it/s]
 19%|█▉        | 1397/7308 [17:09<1:02:27,  1.58it/s]
 19%|█▉        | 1398/7308 [17:10<1:02:26,  1.58it/s]
 19%|█▉        | 1399/7308 [17:10<1:02:24,  1.58it/s]
 19%|█▉        | 1400/7308 [17:11<1:02:24,  1.58it/s]
                                                     
{'loss': '4.382', 'grad_norm': '0.2175', 'learning_rate': '0.0006', 'epoch': '38.9'}
+
 19%|█▉        | 1400/7308 [17:11<1:02:24,  1.58it/s]
 19%|█▉        | 1401/7308 [17:12<1:02:24,  1.58it/s]
 19%|█▉        | 1402/7308 [17:12<1:02:23,  1.58it/s]
 19%|█▉        | 1403/7308 [17:13<1:02:20,  1.58it/s]
 19%|█▉        | 1404/7308 [17:13<50:48,  1.94it/s]  
 19%|█▉        | 1405/7308 [17:14<57:10,  1.72it/s]
 19%|█▉        | 1406/7308 [17:15<58:42,  1.68it/s]
 19%|█▉        | 1407/7308 [17:15<59:47,  1.65it/s]
 19%|█▉        | 1408/7308 [17:16<1:00:34,  1.62it/s]
 19%|█▉        | 1409/7308 [17:16<1:01:06,  1.61it/s]
 19%|█▉        | 1410/7308 [17:17<1:01:30,  1.60it/s]
                                                     
{'loss': '4.348', 'grad_norm': '0.276', 'learning_rate': '0.0006', 'epoch': '39.17'}
+
 19%|█▉        | 1410/7308 [17:17<1:01:30,  1.60it/s]
 19%|█▉        | 1411/7308 [17:18<1:01:47,  1.59it/s]
 19%|█▉        | 1412/7308 [17:18<1:01:59,  1.59it/s]
 19%|█▉        | 1413/7308 [17:19<1:02:07,  1.58it/s]
 19%|█▉        | 1414/7308 [17:20<1:02:13,  1.58it/s]
 19%|█▉        | 1415/7308 [17:20<1:02:16,  1.58it/s]
 19%|█▉        | 1416/7308 [17:21<1:02:18,  1.58it/s]
 19%|█▉        | 1417/7308 [17:21<1:02:20,  1.57it/s]
 19%|█▉        | 1418/7308 [17:22<1:02:21,  1.57it/s]
 19%|█▉        | 1419/7308 [17:23<1:02:20,  1.57it/s]
 19%|█▉        | 1420/7308 [17:23<1:02:20,  1.57it/s]
                                                     
{'loss': '4.405', 'grad_norm': '0.2335', 'learning_rate': '0.0006', 'epoch': '39.45'}
+
 19%|█▉        | 1420/7308 [17:23<1:02:20,  1.57it/s]
 19%|█▉        | 1421/7308 [17:24<1:02:20,  1.57it/s]
 19%|█▉        | 1422/7308 [17:25<1:02:20,  1.57it/s]
 19%|█▉        | 1423/7308 [17:25<1:02:20,  1.57it/s]
 19%|█▉        | 1424/7308 [17:26<1:02:19,  1.57it/s]
 19%|█▉        | 1425/7308 [17:27<1:02:19,  1.57it/s]
 20%|█▉        | 1426/7308 [17:27<1:02:20,  1.57it/s]
 20%|█▉        | 1427/7308 [17:28<1:02:18,  1.57it/s]
 20%|█▉        | 1428/7308 [17:28<1:02:17,  1.57it/s]
 20%|█▉        | 1429/7308 [17:29<1:02:17,  1.57it/s]
 20%|█▉        | 1430/7308 [17:30<1:02:18,  1.57it/s]
                                                     
{'loss': '4.422', 'grad_norm': '0.2039', 'learning_rate': '0.0006', 'epoch': '39.73'}
+
 20%|█▉        | 1430/7308 [17:30<1:02:18,  1.57it/s]
 20%|█▉        | 1431/7308 [17:30<1:02:20,  1.57it/s]
 20%|█▉        | 1432/7308 [17:31<1:02:19,  1.57it/s]
 20%|█▉        | 1433/7308 [17:32<1:02:16,  1.57it/s]
 20%|█▉        | 1434/7308 [17:32<1:02:14,  1.57it/s]
 20%|█▉        | 1435/7308 [17:33<1:02:13,  1.57it/s]
 20%|█▉        | 1436/7308 [17:34<1:02:13,  1.57it/s]
 20%|█▉        | 1437/7308 [17:34<1:02:12,  1.57it/s]
 20%|█▉        | 1438/7308 [17:35<1:02:12,  1.57it/s]
 20%|█▉        | 1439/7308 [17:35<1:02:07,  1.57it/s]
 20%|█▉        | 1440/7308 [17:36<50:36,  1.93it/s]  
                                                   
{'loss': '4.357', 'grad_norm': '0.2829', 'learning_rate': '0.0006', 'epoch': '40'}
+
 20%|█▉        | 1440/7308 [17:36<50:36,  1.93it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 22.72it/s]
+
 15%|█▌        | 6/39 [00:00<00:01, 17.58it/s]
+
 21%|██        | 8/39 [00:00<00:01, 16.64it/s]
+
 26%|██▌       | 10/39 [00:00<00:01, 16.11it/s]
+
 31%|███       | 12/39 [00:00<00:01, 15.78it/s]
+
 36%|███▌      | 14/39 [00:00<00:01, 15.59it/s]
+
 41%|████      | 16/39 [00:00<00:01, 15.46it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 15.37it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 15.32it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 15.28it/s]
+
 62%|██████▏   | 24/39 [00:01<00:00, 15.28it/s]
+
 67%|██████▋   | 26/39 [00:01<00:00, 15.27it/s]
+
 72%|███████▏  | 28/39 [00:01<00:00, 15.24it/s]
+
 77%|███████▋  | 30/39 [00:01<00:00, 15.24it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 15.23it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 15.23it/s]
+
 92%|█████████▏| 36/39 [00:02<00:00, 15.21it/s]
+
 97%|█████████▋| 38/39 [00:02<00:00, 15.22it/s]
                                                   
+
                                               
{'eval_loss': '4.098', 'eval_runtime': '2.626', 'eval_samples_per_second': '929.7', 'eval_steps_per_second': '14.85', 'epoch': '40'}
+
 20%|█▉        | 1440/7308 [17:38<50:36,  1.93it/s]
+
100%|██████████| 39/39 [00:02<00:00, 15.22it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-1440
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-1440/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-1440/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  2.17it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  2.17it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-1440/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-1440/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-1440/tokenizer_config.json
+[eval] epoch 40.0 step 1440 loss 4.0981 (best 4.0981 @ ep 40.0, 0/20 since best)
+
+
                                               
 20%|█▉        | 1441/7308 [17:41<3:10:22,  1.95s/it]
 20%|█▉        | 1442/7308 [17:42<2:31:52,  1.55s/it]
 20%|█▉        | 1443/7308 [17:42<2:04:53,  1.28s/it]
 20%|█▉        | 1444/7308 [17:43<1:46:02,  1.09s/it]
 20%|█▉        | 1445/7308 [17:44<1:32:50,  1.05it/s]
 20%|█▉        | 1446/7308 [17:44<1:23:35,  1.17it/s]
 20%|█▉        | 1447/7308 [17:45<1:17:07,  1.27it/s]
 20%|█▉        | 1448/7308 [17:45<1:12:36,  1.35it/s]
 20%|█▉        | 1449/7308 [17:46<1:09:26,  1.41it/s]
 20%|█▉        | 1450/7308 [17:47<1:07:12,  1.45it/s]
                                                     
{'loss': '4.339', 'grad_norm': '0.2097', 'learning_rate': '0.0006', 'epoch': '40.28'}
+
 20%|█▉        | 1450/7308 [17:47<1:07:12,  1.45it/s]
 20%|█▉        | 1451/7308 [17:47<1:05:39,  1.49it/s]
 20%|█▉        | 1452/7308 [17:48<1:04:34,  1.51it/s]
 20%|█▉        | 1453/7308 [17:49<1:03:46,  1.53it/s]
 20%|█▉        | 1454/7308 [17:49<1:03:12,  1.54it/s]
 20%|█▉        | 1455/7308 [17:50<1:02:49,  1.55it/s]
 20%|█▉        | 1456/7308 [17:51<1:02:33,  1.56it/s]
 20%|█▉        | 1457/7308 [17:51<1:02:20,  1.56it/s]
 20%|█▉        | 1458/7308 [17:52<1:02:11,  1.57it/s]
 20%|█▉        | 1459/7308 [17:52<1:02:05,  1.57it/s]
 20%|█▉        | 1460/7308 [17:53<1:02:02,  1.57it/s]
                                                     
{'loss': '4.375', 'grad_norm': '0.2298', 'learning_rate': '0.0006', 'epoch': '40.57'}
+
 20%|█▉        | 1460/7308 [17:53<1:02:02,  1.57it/s]
 20%|█▉        | 1461/7308 [17:54<1:01:58,  1.57it/s]
 20%|██        | 1462/7308 [17:54<1:01:56,  1.57it/s]
 20%|██        | 1463/7308 [17:55<1:01:53,  1.57it/s]
 20%|██        | 1464/7308 [17:56<1:01:52,  1.57it/s]
 20%|██        | 1465/7308 [17:56<1:01:50,  1.57it/s]
 20%|██        | 1466/7308 [17:57<1:01:49,  1.57it/s]
 20%|██        | 1467/7308 [17:58<1:01:50,  1.57it/s]
 20%|██        | 1468/7308 [17:58<1:01:49,  1.57it/s]
 20%|██        | 1469/7308 [17:59<1:01:47,  1.57it/s]
 20%|██        | 1470/7308 [17:59<1:01:46,  1.58it/s]
                                                     
{'loss': '4.364', 'grad_norm': '0.2528', 'learning_rate': '0.0006', 'epoch': '40.85'}
+
 20%|██        | 1470/7308 [17:59<1:01:46,  1.58it/s]
 20%|██        | 1471/7308 [18:00<1:01:47,  1.57it/s]
 20%|██        | 1472/7308 [18:01<1:01:44,  1.58it/s]
 20%|██        | 1473/7308 [18:01<1:01:43,  1.58it/s]
 20%|██        | 1474/7308 [18:02<1:01:44,  1.57it/s]
 20%|██        | 1475/7308 [18:03<1:01:39,  1.58it/s]
 20%|██        | 1476/7308 [18:03<50:15,  1.93it/s]  
 20%|██        | 1477/7308 [18:04<55:40,  1.75it/s]
 20%|██        | 1478/7308 [18:04<57:27,  1.69it/s]
 20%|██        | 1479/7308 [18:05<58:40,  1.66it/s]
 20%|██        | 1480/7308 [18:05<59:33,  1.63it/s]
                                                   
{'loss': '4.35', 'grad_norm': '0.2874', 'learning_rate': '0.0006', 'epoch': '41.11'}
+
 20%|██        | 1480/7308 [18:05<59:33,  1.63it/s]
 20%|██        | 1481/7308 [18:06<1:00:11,  1.61it/s]
 20%|██        | 1482/7308 [18:07<1:00:38,  1.60it/s]
 20%|██        | 1483/7308 [18:07<1:00:55,  1.59it/s]
 20%|██        | 1484/7308 [18:08<1:01:07,  1.59it/s]
 20%|██        | 1485/7308 [18:09<1:01:18,  1.58it/s]
 20%|██        | 1486/7308 [18:09<1:01:22,  1.58it/s]
 20%|██        | 1487/7308 [18:10<1:01:27,  1.58it/s]
 20%|██        | 1488/7308 [18:11<1:01:29,  1.58it/s]
 20%|██        | 1489/7308 [18:11<1:01:30,  1.58it/s]
 20%|██        | 1490/7308 [18:12<1:01:30,  1.58it/s]
                                                     
{'loss': '4.352', 'grad_norm': '0.2173', 'learning_rate': '0.0006', 'epoch': '41.4'}
+
 20%|██        | 1490/7308 [18:12<1:01:30,  1.58it/s]
 20%|██        | 1491/7308 [18:12<1:01:31,  1.58it/s]
 20%|██        | 1492/7308 [18:13<1:01:31,  1.58it/s]
 20%|██        | 1493/7308 [18:14<1:01:30,  1.58it/s]
 20%|██        | 1494/7308 [18:14<1:01:30,  1.58it/s]
 20%|██        | 1495/7308 [18:15<1:01:29,  1.58it/s]
 20%|██        | 1496/7308 [18:16<1:01:28,  1.58it/s]
 20%|██        | 1497/7308 [18:16<1:01:27,  1.58it/s]
 20%|██        | 1498/7308 [18:17<1:01:32,  1.57it/s]
 21%|██        | 1499/7308 [18:18<1:01:31,  1.57it/s]
 21%|██        | 1500/7308 [18:18<1:01:29,  1.57it/s]
                                                     
{'loss': '4.298', 'grad_norm': '0.2384', 'learning_rate': '0.0006', 'epoch': '41.68'}
+
 21%|██        | 1500/7308 [18:18<1:01:29,  1.57it/s]
 21%|██        | 1501/7308 [18:19<1:01:29,  1.57it/s]
 21%|██        | 1502/7308 [18:19<1:01:30,  1.57it/s]
 21%|██        | 1503/7308 [18:20<1:01:28,  1.57it/s]
 21%|██        | 1504/7308 [18:21<1:01:27,  1.57it/s]
 21%|██        | 1505/7308 [18:21<1:01:27,  1.57it/s]
 21%|██        | 1506/7308 [18:22<1:01:24,  1.57it/s]
 21%|██        | 1507/7308 [18:23<1:01:24,  1.57it/s]
 21%|██        | 1508/7308 [18:23<1:01:23,  1.57it/s]
 21%|██        | 1509/7308 [18:24<1:01:22,  1.57it/s]
 21%|██        | 1510/7308 [18:25<1:01:23,  1.57it/s]
                                                     
{'loss': '4.321', 'grad_norm': '0.2379', 'learning_rate': '0.0006', 'epoch': '41.96'}
+
 21%|██        | 1510/7308 [18:25<1:01:23,  1.57it/s]
 21%|██        | 1511/7308 [18:25<1:01:21,  1.57it/s]
 21%|██        | 1512/7308 [18:25<49:59,  1.93it/s]  [transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 23.02it/s]
+
 15%|█▌        | 6/39 [00:00<00:01, 17.66it/s]
+
 21%|██        | 8/39 [00:00<00:01, 16.66it/s]
+
 26%|██▌       | 10/39 [00:00<00:01, 16.13it/s]
+
 31%|███       | 12/39 [00:00<00:01, 15.81it/s]
+
 36%|███▌      | 14/39 [00:00<00:01, 15.60it/s]
+
 41%|████      | 16/39 [00:00<00:01, 15.45it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 15.36it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 15.30it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 15.27it/s]
+
 62%|██████▏   | 24/39 [00:01<00:00, 15.26it/s]
+
 67%|██████▋   | 26/39 [00:01<00:00, 15.25it/s]
+
 72%|███████▏  | 28/39 [00:01<00:00, 15.24it/s]
+
 77%|███████▋  | 30/39 [00:01<00:00, 15.24it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 15.23it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 15.22it/s]
+
 92%|█████████▏| 36/39 [00:02<00:00, 15.22it/s]
+
 97%|█████████▋| 38/39 [00:02<00:00, 15.21it/s]
                                                   
+
                                               
{'eval_loss': '4.086', 'eval_runtime': '2.614', 'eval_samples_per_second': '933.9', 'eval_steps_per_second': '14.92', 'epoch': '42'}
+
 21%|██        | 1512/7308 [18:28<49:59,  1.93it/s]
+
100%|██████████| 39/39 [00:02<00:00, 15.21it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-1512
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-1512/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-1512/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.95it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.95it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-1512/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-1512/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-1512/tokenizer_config.json
+[eval] epoch 42.0 step 1512 loss 4.0858 (best 4.0858 @ ep 42.0, 0/20 since best)
+
+
                                               
 21%|██        | 1513/7308 [18:31<3:09:25,  1.96s/it]
 21%|██        | 1514/7308 [18:31<2:30:57,  1.56s/it]
 21%|██        | 1515/7308 [18:32<2:04:03,  1.28s/it]
 21%|██        | 1516/7308 [18:33<1:45:13,  1.09s/it]
 21%|██        | 1517/7308 [18:33<1:32:01,  1.05it/s]
 21%|██        | 1518/7308 [18:34<1:22:48,  1.17it/s]
 21%|██        | 1519/7308 [18:35<1:16:20,  1.26it/s]
 21%|██        | 1520/7308 [18:35<1:11:49,  1.34it/s]
                                                     
{'loss': '4.364', 'grad_norm': '0.3016', 'learning_rate': '0.0006', 'epoch': '42.23'}
+
 21%|██        | 1520/7308 [18:35<1:11:49,  1.34it/s]
 21%|██        | 1521/7308 [18:36<1:08:41,  1.40it/s]
 21%|██        | 1522/7308 [18:36<1:06:26,  1.45it/s]
 21%|██        | 1523/7308 [18:37<1:04:54,  1.49it/s]
 21%|██        | 1524/7308 [18:38<1:03:48,  1.51it/s]
 21%|██        | 1525/7308 [18:38<1:03:02,  1.53it/s]
 21%|██        | 1526/7308 [18:39<1:02:29,  1.54it/s]
 21%|██        | 1527/7308 [18:40<1:02:06,  1.55it/s]
 21%|██        | 1528/7308 [18:40<1:01:50,  1.56it/s]
 21%|██        | 1529/7308 [18:41<1:01:38,  1.56it/s]
 21%|██        | 1530/7308 [18:42<1:01:30,  1.57it/s]
                                                     
{'loss': '4.351', 'grad_norm': '0.3244', 'learning_rate': '0.0006', 'epoch': '42.51'}
+
 21%|██        | 1530/7308 [18:42<1:01:30,  1.57it/s]
 21%|██        | 1531/7308 [18:42<1:01:26,  1.57it/s]
 21%|██        | 1532/7308 [18:43<1:01:22,  1.57it/s]
 21%|██        | 1533/7308 [18:43<1:01:18,  1.57it/s]
 21%|██        | 1534/7308 [18:44<1:01:14,  1.57it/s]
 21%|██        | 1535/7308 [18:45<1:01:11,  1.57it/s]
 21%|██        | 1536/7308 [18:45<1:01:08,  1.57it/s]
 21%|██        | 1537/7308 [18:46<1:01:05,  1.57it/s]
 21%|██        | 1538/7308 [18:47<1:01:05,  1.57it/s]
 21%|██        | 1539/7308 [18:47<1:01:04,  1.57it/s]
 21%|██        | 1540/7308 [18:48<1:01:02,  1.57it/s]
                                                     
{'loss': '4.344', 'grad_norm': '0.233', 'learning_rate': '0.0006', 'epoch': '42.79'}
+
 21%|██        | 1540/7308 [18:48<1:01:02,  1.57it/s]
 21%|██        | 1541/7308 [18:48<1:01:02,  1.57it/s]
 21%|██        | 1542/7308 [18:49<1:01:00,  1.58it/s]
 21%|██        | 1543/7308 [18:50<1:00:59,  1.58it/s]
 21%|██        | 1544/7308 [18:50<1:00:58,  1.58it/s]
 21%|██        | 1545/7308 [18:51<1:00:57,  1.58it/s]
 21%|██        | 1546/7308 [18:52<1:00:57,  1.58it/s]
 21%|██        | 1547/7308 [18:52<1:00:54,  1.58it/s]
 21%|██        | 1548/7308 [18:53<49:36,  1.93it/s]  
 21%|██        | 1549/7308 [18:53<55:24,  1.73it/s]
 21%|██        | 1550/7308 [18:54<57:04,  1.68it/s]
                                                   
{'loss': '4.34', 'grad_norm': '0.2659', 'learning_rate': '0.0006', 'epoch': '43.06'}
+
 21%|██        | 1550/7308 [18:54<57:04,  1.68it/s]
 21%|██        | 1551/7308 [18:55<58:15,  1.65it/s]
 21%|██        | 1552/7308 [18:55<59:04,  1.62it/s]
 21%|██▏       | 1553/7308 [18:56<59:37,  1.61it/s]
 21%|██▏       | 1554/7308 [18:56<1:00:01,  1.60it/s]
 21%|██▏       | 1555/7308 [18:57<1:00:18,  1.59it/s]
 21%|██▏       | 1556/7308 [18:58<1:00:29,  1.58it/s]
 21%|██▏       | 1557/7308 [18:58<1:00:36,  1.58it/s]
 21%|██▏       | 1558/7308 [18:59<1:00:40,  1.58it/s]
 21%|██▏       | 1559/7308 [19:00<1:00:43,  1.58it/s]
 21%|██▏       | 1560/7308 [19:00<1:00:46,  1.58it/s]
                                                     
{'loss': '4.308', 'grad_norm': '0.2581', 'learning_rate': '0.0006', 'epoch': '43.34'}
+
 21%|██▏       | 1560/7308 [19:00<1:00:46,  1.58it/s]
 21%|██▏       | 1561/7308 [19:01<1:00:49,  1.57it/s]
 21%|██▏       | 1562/7308 [19:02<1:00:50,  1.57it/s]
 21%|██▏       | 1563/7308 [19:02<1:00:49,  1.57it/s]
 21%|██▏       | 1564/7308 [19:03<1:00:49,  1.57it/s]
 21%|██▏       | 1565/7308 [19:03<1:00:50,  1.57it/s]
 21%|██▏       | 1566/7308 [19:04<1:00:49,  1.57it/s]
 21%|██▏       | 1567/7308 [19:05<1:00:48,  1.57it/s]
 21%|██▏       | 1568/7308 [19:05<1:00:48,  1.57it/s]
 21%|██▏       | 1569/7308 [19:06<1:00:47,  1.57it/s]
 21%|██▏       | 1570/7308 [19:07<1:00:47,  1.57it/s]
                                                     
{'loss': '4.317', 'grad_norm': '0.3254', 'learning_rate': '0.0006', 'epoch': '43.62'}
+
 21%|██▏       | 1570/7308 [19:07<1:00:47,  1.57it/s]
 21%|██▏       | 1571/7308 [19:07<1:00:47,  1.57it/s]
 22%|██▏       | 1572/7308 [19:08<1:00:46,  1.57it/s]
 22%|██▏       | 1573/7308 [19:09<1:00:44,  1.57it/s]
 22%|██▏       | 1574/7308 [19:09<1:00:43,  1.57it/s]
 22%|██▏       | 1575/7308 [19:10<1:00:43,  1.57it/s]
 22%|██▏       | 1576/7308 [19:10<1:00:42,  1.57it/s]
 22%|██▏       | 1577/7308 [19:11<1:00:42,  1.57it/s]
 22%|██▏       | 1578/7308 [19:12<1:00:43,  1.57it/s]
 22%|██▏       | 1579/7308 [19:12<1:00:42,  1.57it/s]
 22%|██▏       | 1580/7308 [19:13<1:00:40,  1.57it/s]
                                                     
{'loss': '4.338', 'grad_norm': '0.2292', 'learning_rate': '0.0006', 'epoch': '43.9'}
+
 22%|██▏       | 1580/7308 [19:13<1:00:40,  1.57it/s]
 22%|██▏       | 1581/7308 [19:14<1:00:39,  1.57it/s]
 22%|██▏       | 1582/7308 [19:14<1:00:39,  1.57it/s]
 22%|██▏       | 1583/7308 [19:15<1:00:35,  1.57it/s]
 22%|██▏       | 1584/7308 [19:15<49:22,  1.93it/s]  [transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 22.76it/s]
+
 15%|█▌        | 6/39 [00:00<00:01, 17.61it/s]
+
 21%|██        | 8/39 [00:00<00:01, 16.65it/s]
+
 26%|██▌       | 10/39 [00:00<00:01, 16.13it/s]
+
 31%|███       | 12/39 [00:00<00:01, 15.82it/s]
+
 36%|███▌      | 14/39 [00:00<00:01, 15.60it/s]
+
 41%|████      | 16/39 [00:00<00:01, 15.45it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 15.39it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 15.34it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 15.29it/s]
+
 62%|██████▏   | 24/39 [00:01<00:00, 15.28it/s]
+
 67%|██████▋   | 26/39 [00:01<00:00, 15.25it/s]
+
 72%|███████▏  | 28/39 [00:01<00:00, 15.24it/s]
+
 77%|███████▋  | 30/39 [00:01<00:00, 15.23it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 15.23it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 15.20it/s]
+
 92%|█████████▏| 36/39 [00:02<00:00, 15.19it/s]
+
 97%|█████████▋| 38/39 [00:02<00:00, 15.20it/s]
                                                   
+
                                               
{'eval_loss': '4.084', 'eval_runtime': '2.629', 'eval_samples_per_second': '928.6', 'eval_steps_per_second': '14.84', 'epoch': '44'}
+
 22%|██▏       | 1584/7308 [19:18<49:22,  1.93it/s]
+
100%|██████████| 39/39 [00:02<00:00, 15.20it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-1584
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-1584/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-1584/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.96it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  1.96it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-1584/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-1584/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-1584/tokenizer_config.json
+[eval] epoch 44.0 step 1584 loss 4.0840 (best 4.0840 @ ep 44.0, 0/20 since best)
+
+
                                               
 22%|██▏       | 1585/7308 [19:20<3:07:07,  1.96s/it]
 22%|██▏       | 1586/7308 [19:21<2:29:07,  1.56s/it]
 22%|██▏       | 1587/7308 [19:22<2:02:31,  1.29s/it]
 22%|██▏       | 1588/7308 [19:22<1:43:55,  1.09s/it]
 22%|██▏       | 1589/7308 [19:23<1:30:54,  1.05it/s]
 22%|██▏       | 1590/7308 [19:24<1:21:46,  1.17it/s]
                                                     
{'loss': '4.254', 'grad_norm': '0.2845', 'learning_rate': '0.0006', 'epoch': '44.17'}
+
 22%|██▏       | 1590/7308 [19:24<1:21:46,  1.17it/s]
 22%|██▏       | 1591/7308 [19:24<1:15:24,  1.26it/s]
 22%|██▏       | 1592/7308 [19:25<1:10:55,  1.34it/s]
 22%|██▏       | 1593/7308 [19:26<1:07:47,  1.41it/s]
 22%|██▏       | 1594/7308 [19:26<1:05:33,  1.45it/s]
 22%|██▏       | 1595/7308 [19:27<1:04:02,  1.49it/s]
 22%|██▏       | 1596/7308 [19:27<1:02:57,  1.51it/s]
 22%|██▏       | 1597/7308 [19:28<1:02:11,  1.53it/s]
 22%|██▏       | 1598/7308 [19:29<1:01:38,  1.54it/s]
 22%|██▏       | 1599/7308 [19:29<1:01:16,  1.55it/s]
 22%|██▏       | 1600/7308 [19:30<1:01:00,  1.56it/s]
                                                     
{'loss': '4.201', 'grad_norm': '0.2495', 'learning_rate': '0.0006', 'epoch': '44.45'}
+
 22%|██▏       | 1600/7308 [19:30<1:01:00,  1.56it/s]
 22%|██▏       | 1601/7308 [19:31<1:00:54,  1.56it/s]
 22%|██▏       | 1602/7308 [19:31<1:00:45,  1.57it/s]
 22%|██▏       | 1603/7308 [19:32<1:00:38,  1.57it/s]
 22%|██▏       | 1604/7308 [19:33<1:00:33,  1.57it/s]
 22%|██▏       | 1605/7308 [19:33<1:00:29,  1.57it/s]
 22%|██▏       | 1606/7308 [19:34<1:00:26,  1.57it/s]
 22%|██▏       | 1607/7308 [19:34<1:00:23,  1.57it/s]
 22%|██▏       | 1608/7308 [19:35<1:00:21,  1.57it/s]
 22%|██▏       | 1609/7308 [19:36<1:00:19,  1.57it/s]
 22%|██▏       | 1610/7308 [19:36<1:00:19,  1.57it/s]
                                                     
{'loss': '4.311', 'grad_norm': '0.2302', 'learning_rate': '0.0006', 'epoch': '44.73'}
+
 22%|██▏       | 1610/7308 [19:36<1:00:19,  1.57it/s]
 22%|██▏       | 1611/7308 [19:37<1:00:19,  1.57it/s]
 22%|██▏       | 1612/7308 [19:38<1:00:19,  1.57it/s]
 22%|██▏       | 1613/7308 [19:38<1:00:18,  1.57it/s]
 22%|██▏       | 1614/7308 [19:39<1:00:17,  1.57it/s]
 22%|██▏       | 1615/7308 [19:40<1:00:15,  1.57it/s]
 22%|██▏       | 1616/7308 [19:40<1:00:14,  1.57it/s]
 22%|██▏       | 1617/7308 [19:41<1:00:13,  1.57it/s]
 22%|██▏       | 1618/7308 [19:41<1:00:14,  1.57it/s]
 22%|██▏       | 1619/7308 [19:42<1:00:09,  1.58it/s]
 22%|██▏       | 1620/7308 [19:42<49:00,  1.93it/s]  
                                                   
{'loss': '4.201', 'grad_norm': '0.2737', 'learning_rate': '0.0006', 'epoch': '45'}
+
 22%|██▏       | 1620/7308 [19:42<49:00,  1.93it/s]
 22%|██▏       | 1621/7308 [19:43<54:26,  1.74it/s]
 22%|██▏       | 1622/7308 [19:44<56:10,  1.69it/s]
 22%|██▏       | 1623/7308 [19:44<57:22,  1.65it/s]
 22%|██▏       | 1624/7308 [19:45<58:12,  1.63it/s]
 22%|██▏       | 1625/7308 [19:46<58:48,  1.61it/s]
 22%|██▏       | 1626/7308 [19:46<59:13,  1.60it/s]
 22%|██▏       | 1627/7308 [19:47<59:31,  1.59it/s]
 22%|██▏       | 1628/7308 [19:47<59:42,  1.59it/s]
 22%|██▏       | 1629/7308 [19:48<59:50,  1.58it/s]
 22%|██▏       | 1630/7308 [19:49<59:56,  1.58it/s]
                                                   
{'loss': '4.245', 'grad_norm': '0.3193', 'learning_rate': '0.0006', 'epoch': '45.28'}
+
 22%|██▏       | 1630/7308 [19:49<59:56,  1.58it/s]
 22%|██▏       | 1631/7308 [19:49<1:00:01,  1.58it/s]
 22%|██▏       | 1632/7308 [19:50<1:00:04,  1.57it/s]
 22%|██▏       | 1633/7308 [19:51<1:00:04,  1.57it/s]
 22%|██▏       | 1634/7308 [19:51<1:00:04,  1.57it/s]
 22%|██▏       | 1635/7308 [19:52<1:00:06,  1.57it/s]
 22%|██▏       | 1636/7308 [19:53<1:00:06,  1.57it/s]
 22%|██▏       | 1637/7308 [19:53<1:00:06,  1.57it/s]
 22%|██▏       | 1638/7308 [19:54<1:00:06,  1.57it/s]
 22%|██▏       | 1639/7308 [19:54<1:00:05,  1.57it/s]
 22%|██▏       | 1640/7308 [19:55<1:00:04,  1.57it/s]
                                                     
{'loss': '4.215', 'grad_norm': '0.2607', 'learning_rate': '0.0006', 'epoch': '45.57'}
+
 22%|██▏       | 1640/7308 [19:55<1:00:04,  1.57it/s]
 22%|██▏       | 1641/7308 [19:56<1:00:05,  1.57it/s]
 22%|██▏       | 1642/7308 [19:56<1:00:04,  1.57it/s]
 22%|██▏       | 1643/7308 [19:57<1:00:03,  1.57it/s]
 22%|██▏       | 1644/7308 [19:58<1:00:01,  1.57it/s]
 23%|██▎       | 1645/7308 [19:58<1:00:00,  1.57it/s]
 23%|██▎       | 1646/7308 [19:59<1:00:00,  1.57it/s]
 23%|██▎       | 1647/7308 [20:00<59:58,  1.57it/s]  
 23%|██▎       | 1648/7308 [20:00<59:56,  1.57it/s]
 23%|██▎       | 1649/7308 [20:01<59:57,  1.57it/s]
 23%|██▎       | 1650/7308 [20:01<59:57,  1.57it/s]
                                                   
{'loss': '4.223', 'grad_norm': '0.2835', 'learning_rate': '0.0006', 'epoch': '45.85'}
+
 23%|██▎       | 1650/7308 [20:01<59:57,  1.57it/s]
 23%|██▎       | 1651/7308 [20:02<59:57,  1.57it/s]
 23%|██▎       | 1652/7308 [20:03<59:55,  1.57it/s]
 23%|██▎       | 1653/7308 [20:03<59:53,  1.57it/s]
 23%|██▎       | 1654/7308 [20:04<59:50,  1.57it/s]
 23%|██▎       | 1655/7308 [20:05<59:46,  1.58it/s]
 23%|██▎       | 1656/7308 [20:05<48:42,  1.93it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 23.79it/s]
+
 15%|█▌        | 6/39 [00:00<00:01, 17.84it/s]
+
 21%|██        | 8/39 [00:00<00:01, 16.78it/s]
+
 26%|██▌       | 10/39 [00:00<00:01, 16.22it/s]
+
 31%|███       | 12/39 [00:00<00:01, 15.87it/s]
+
 36%|███▌      | 14/39 [00:00<00:01, 15.66it/s]
+
 41%|████      | 16/39 [00:00<00:01, 15.50it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 15.41it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 15.34it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 15.30it/s]
+
 62%|██████▏   | 24/39 [00:01<00:00, 15.29it/s]
+
 67%|██████▋   | 26/39 [00:01<00:00, 15.27it/s]
+
 72%|███████▏  | 28/39 [00:01<00:00, 15.22it/s]
+
 77%|███████▋  | 30/39 [00:01<00:00, 15.25it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 15.24it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 15.24it/s]
+
 92%|█████████▏| 36/39 [00:02<00:00, 15.22it/s]
+
 97%|█████████▋| 38/39 [00:02<00:00, 15.22it/s]
                                                   
+
                                               
{'eval_loss': '4.08', 'eval_runtime': '2.635', 'eval_samples_per_second': '926.2', 'eval_steps_per_second': '14.8', 'epoch': '46'}
+
 23%|██▎       | 1656/7308 [20:07<48:42,  1.93it/s]
+
100%|██████████| 39/39 [00:02<00:00, 15.22it/s][transformers] Saving model checkpoint to /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-1656
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-1656/config.json
+[transformers] Configuration saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-1656/generation_config.json
+
+
+
Writing model shards:   0%|          | 0/1 [00:00<?, ?it/s]
+
+
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  2.16it/s]
Writing model shards: 100%|██████████| 1/1 [00:00<00:00,  2.16it/s]
+[transformers] Model weights saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-1656/model.safetensors
+[transformers] chat template saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-1656/chat_template.jinja
+[transformers] tokenizer config file saved in /home/ubuntu/workspace/data-augmentations-for-pretraining/runs/t2a/t2a_37M_C2_r2l50_i5exp/checkpoint-1656/tokenizer_config.json
+[eval] epoch 46.0 step 1656 loss 4.0798 (best 4.0798 @ ep 46.0, 0/20 since best)
+
+
                                               
 23%|██▎       | 1657/7308 [20:10<3:03:50,  1.95s/it]
 23%|██▎       | 1658/7308 [20:11<2:26:33,  1.56s/it]
 23%|██▎       | 1659/7308 [20:11<2:00:27,  1.28s/it]
 23%|██▎       | 1660/7308 [20:12<1:42:12,  1.09s/it]
                                                     
{'loss': '4.26', 'grad_norm': '0.2795', 'learning_rate': '0.0006', 'epoch': '46.11'}
+
 23%|██▎       | 1660/7308 [20:12<1:42:12,  1.09s/it]
 23%|██▎       | 1661/7308 [20:13<1:29:28,  1.05it/s]
 23%|██▎       | 1662/7308 [20:13<1:20:31,  1.17it/s]
 23%|██▎       | 1663/7308 [20:14<1:14:15,  1.27it/s]
 23%|██▎       | 1664/7308 [20:15<1:09:52,  1.35it/s]
 23%|██▎       | 1665/7308 [20:15<1:06:48,  1.41it/s]
 23%|██▎       | 1666/7308 [20:16<1:04:38,  1.45it/s]
 23%|██▎       | 1667/7308 [20:16<1:03:09,  1.49it/s]
 23%|██▎       | 1668/7308 [20:17<1:02:08,  1.51it/s]
 23%|██▎       | 1669/7308 [20:18<1:01:21,  1.53it/s]
 23%|██▎       | 1670/7308 [20:18<1:00:47,  1.55it/s]
                                                     
{'loss': '4.222', 'grad_norm': '0.2544', 'learning_rate': '0.0006', 'epoch': '46.4'}
+
 23%|██▎       | 1670/7308 [20:18<1:00:47,  1.55it/s]
 23%|██▎       | 1671/7308 [20:19<1:00:27,  1.55it/s]
 23%|██▎       | 1672/7308 [20:20<1:00:11,  1.56it/s]
 23%|██▎       | 1673/7308 [20:20<59:58,  1.57it/s]  
 23%|██▎       | 1674/7308 [20:21<59:49,  1.57it/s]
 23%|██▎       | 1675/7308 [20:22<59:44,  1.57it/s]
 23%|██▎       | 1676/7308 [20:22<59:40,  1.57it/s]
 23%|██▎       | 1677/7308 [20:23<59:40,  1.57it/s]
 23%|██▎       | 1678/7308 [20:23<59:42,  1.57it/s]
 23%|██▎       | 1679/7308 [20:24<59:41,  1.57it/s]
 23%|██▎       | 1680/7308 [20:25<59:41,  1.57it/s]
                                                   
{'loss': '4.202', 'grad_norm': '0.2423', 'learning_rate': '0.0006', 'epoch': '46.68'}
+
 23%|██▎       | 1680/7308 [20:25<59:41,  1.57it/s]
 23%|██▎       | 1681/7308 [20:25<59:40,  1.57it/s]
 23%|██▎       | 1682/7308 [20:26<59:38,  1.57it/s]
 23%|██▎       | 1683/7308 [20:27<59:38,  1.57it/s]
 23%|██▎       | 1684/7308 [20:27<59:37,  1.57it/s]
 23%|██▎       | 1685/7308 [20:28<59:37,  1.57it/s]
 23%|██▎       | 1686/7308 [20:29<59:34,  1.57it/s]
 23%|██▎       | 1687/7308 [20:29<59:32,  1.57it/s]
 23%|██▎       | 1688/7308 [20:30<59:29,  1.57it/s]
 23%|██▎       | 1689/7308 [20:30<59:27,  1.58it/s]
 23%|██▎       | 1690/7308 [20:31<59:26,  1.58it/s]
                                                   
{'loss': '4.268', 'grad_norm': '0.2288', 'learning_rate': '0.0006', 'epoch': '46.96'}
+
 23%|██▎       | 1690/7308 [20:31<59:26,  1.58it/s]
 23%|██▎       | 1691/7308 [20:32<59:24,  1.58it/s]
 23%|██▎       | 1692/7308 [20:32<48:24,  1.93it/s]
 23%|██▎       | 1693/7308 [20:33<54:25,  1.72it/s]
 23%|██▎       | 1694/7308 [20:33<55:55,  1.67it/s]
 23%|██▎       | 1695/7308 [20:34<56:55,  1.64it/s]
 23%|██▎       | 1696/7308 [20:35<57:40,  1.62it/s]
 23%|██▎       | 1697/7308 [20:35<58:10,  1.61it/s]
 23%|██▎       | 1698/7308 [20:36<58:33,  1.60it/s]
 23%|██▎       | 1699/7308 [20:37<58:47,  1.59it/s]
 23%|██▎       | 1700/7308 [20:37<58:59,  1.58it/s]
                                                   
{'loss': '4.213', 'grad_norm': '0.242', 'learning_rate': '0.0006', 'epoch': '47.23'}
+
 23%|██▎       | 1700/7308 [20:37<58:59,  1.58it/s]
 23%|██▎       | 1701/7308 [20:38<59:05,  1.58it/s]
 23%|██▎       | 1702/7308 [20:38<59:08,  1.58it/s]
 23%|██▎       | 1703/7308 [20:39<59:10,  1.58it/s]
 23%|██▎       | 1704/7308 [20:40<59:14,  1.58it/s]
 23%|██▎       | 1705/7308 [20:40<59:16,  1.58it/s]
 23%|██▎       | 1706/7308 [20:41<59:16,  1.58it/s]
 23%|██▎       | 1707/7308 [20:42<59:13,  1.58it/s]
 23%|██▎       | 1708/7308 [20:42<59:11,  1.58it/s]
 23%|██▎       | 1709/7308 [20:43<59:11,  1.58it/s]
 23%|██▎       | 1710/7308 [20:44<59:14,  1.58it/s]
                                                   
{'loss': '4.262', 'grad_norm': '0.2477', 'learning_rate': '0.0006', 'epoch': '47.51'}
+
 23%|██▎       | 1710/7308 [20:44<59:14,  1.58it/s]
 23%|██▎       | 1711/7308 [20:44<59:13,  1.57it/s]
 23%|██▎       | 1712/7308 [20:45<59:12,  1.58it/s]
 23%|██▎       | 1713/7308 [20:45<59:10,  1.58it/s]
 23%|██▎       | 1714/7308 [20:46<59:11,  1.58it/s]
 23%|██▎       | 1715/7308 [20:47<59:09,  1.58it/s]
 23%|██▎       | 1716/7308 [20:47<59:08,  1.58it/s]
 23%|██▎       | 1717/7308 [20:48<59:08,  1.58it/s]
 24%|██▎       | 1718/7308 [20:49<59:09,  1.58it/s]
 24%|██▎       | 1719/7308 [20:49<59:08,  1.57it/s]
 24%|██▎       | 1720/7308 [20:50<59:07,  1.58it/s]
                                                   
{'loss': '4.285', 'grad_norm': '0.2603', 'learning_rate': '0.0006', 'epoch': '47.79'}
+
 24%|██▎       | 1720/7308 [20:50<59:07,  1.58it/s]
 24%|██▎       | 1721/7308 [20:50<59:08,  1.57it/s]
 24%|██▎       | 1722/7308 [20:51<59:06,  1.58it/s]
 24%|██▎       | 1723/7308 [20:52<59:04,  1.58it/s]
 24%|██▎       | 1724/7308 [20:52<59:03,  1.58it/s]
 24%|██▎       | 1725/7308 [20:53<59:04,  1.58it/s]
 24%|██▎       | 1726/7308 [20:54<59:04,  1.57it/s]
 24%|██▎       | 1727/7308 [20:54<59:02,  1.58it/s]
 24%|██▎       | 1728/7308 [20:55<48:06,  1.93it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 23.20it/s]
+
 15%|█▌        | 6/39 [00:00<00:01, 17.73it/s]
+
 21%|██        | 8/39 [00:00<00:01, 16.69it/s]
+
 26%|██▌       | 10/39 [00:00<00:01, 16.17it/s]
+
 31%|███       | 12/39 [00:00<00:01, 15.83it/s]
+
 36%|███▌      | 14/39 [00:00<00:01, 15.63it/s]
+
 41%|████      | 16/39 [00:00<00:01, 15.48it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 15.39it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 15.32it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 15.29it/s]
+
 62%|██████▏   | 24/39 [00:01<00:00, 15.29it/s]
+
 67%|██████▋   | 26/39 [00:01<00:00, 15.26it/s]
+
 72%|███████▏  | 28/39 [00:01<00:00, 15.24it/s]
+
 77%|███████▋  | 30/39 [00:01<00:00, 15.23it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 15.22it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 15.21it/s]
+
 92%|█████████▏| 36/39 [00:02<00:00, 15.20it/s]
+
 97%|█████████▋| 38/39 [00:02<00:00, 15.21it/s]
                                                   
+
                                               
{'eval_loss': '4.085', 'eval_runtime': '2.638', 'eval_samples_per_second': '925.4', 'eval_steps_per_second': '14.79', 'epoch': '48'}
+
 24%|██▎       | 1728/7308 [20:57<48:06,  1.93it/s]
+
100%|██████████| 39/39 [00:02<00:00, 15.21it/s][eval] epoch 48.0 step 1728 loss 4.0853 (best 4.0798 @ ep 46.0, 1/20 since best)
+
+
                                               
 24%|██▎       | 1729/7308 [20:58<2:07:09,  1.37s/it]
 24%|██▎       | 1730/7308 [20:59<1:46:42,  1.15s/it]
                                                     
{'loss': '4.193', 'grad_norm': '0.3359', 'learning_rate': '0.0006', 'epoch': '48.06'}
+
 24%|██▎       | 1730/7308 [20:59<1:46:42,  1.15s/it]
 24%|██▎       | 1731/7308 [20:59<1:32:22,  1.01it/s]
 24%|██▎       | 1732/7308 [21:00<1:22:23,  1.13it/s]
 24%|██▎       | 1733/7308 [21:00<1:15:19,  1.23it/s]
 24%|██▎       | 1734/7308 [21:01<1:10:25,  1.32it/s]
 24%|██▎       | 1735/7308 [21:02<1:07:02,  1.39it/s]
 24%|██▍       | 1736/7308 [21:02<1:04:35,  1.44it/s]
 24%|██▍       | 1737/7308 [21:03<1:02:54,  1.48it/s]
 24%|██▍       | 1738/7308 [21:04<1:01:43,  1.50it/s]
 24%|██▍       | 1739/7308 [21:04<1:00:56,  1.52it/s]
 24%|██▍       | 1740/7308 [21:05<1:00:20,  1.54it/s]
                                                     
{'loss': '4.159', 'grad_norm': '0.3325', 'learning_rate': '0.0006', 'epoch': '48.34'}
+
 24%|██▍       | 1740/7308 [21:05<1:00:20,  1.54it/s]
 24%|██▍       | 1741/7308 [21:06<59:54,  1.55it/s]  
 24%|██▍       | 1742/7308 [21:06<59:35,  1.56it/s]
 24%|██▍       | 1743/7308 [21:07<59:24,  1.56it/s]
 24%|██▍       | 1744/7308 [21:07<59:13,  1.57it/s]
 24%|██▍       | 1745/7308 [21:08<59:06,  1.57it/s]
 24%|██▍       | 1746/7308 [21:09<59:00,  1.57it/s]
 24%|██▍       | 1747/7308 [21:09<58:57,  1.57it/s]
 24%|██▍       | 1748/7308 [21:10<58:54,  1.57it/s]
 24%|██▍       | 1749/7308 [21:11<58:50,  1.57it/s]
 24%|██▍       | 1750/7308 [21:11<58:51,  1.57it/s]
                                                   
{'loss': '4.183', 'grad_norm': '0.275', 'learning_rate': '0.0006', 'epoch': '48.62'}
+
 24%|██▍       | 1750/7308 [21:11<58:51,  1.57it/s]
 24%|██▍       | 1751/7308 [21:12<58:52,  1.57it/s]
 24%|██▍       | 1752/7308 [21:13<58:48,  1.57it/s]
 24%|██▍       | 1753/7308 [21:13<58:47,  1.57it/s]
 24%|██▍       | 1754/7308 [21:14<58:46,  1.58it/s]
 24%|██▍       | 1755/7308 [21:14<58:44,  1.58it/s]
 24%|██▍       | 1756/7308 [21:15<58:43,  1.58it/s]
 24%|██▍       | 1757/7308 [21:16<58:44,  1.57it/s]
 24%|██▍       | 1758/7308 [21:16<58:43,  1.58it/s]
 24%|██▍       | 1759/7308 [21:17<58:46,  1.57it/s]
 24%|██▍       | 1760/7308 [21:18<58:44,  1.57it/s]
                                                   
{'loss': '4.294', 'grad_norm': '0.2519', 'learning_rate': '0.0006', 'epoch': '48.9'}
+
 24%|██▍       | 1760/7308 [21:18<58:44,  1.57it/s]
 24%|██▍       | 1761/7308 [21:18<58:42,  1.57it/s]
 24%|██▍       | 1762/7308 [21:19<58:41,  1.57it/s]
 24%|██▍       | 1763/7308 [21:19<58:38,  1.58it/s]
 24%|██▍       | 1764/7308 [21:20<47:46,  1.93it/s]
 24%|██▍       | 1765/7308 [21:20<53:16,  1.73it/s]
 24%|██▍       | 1766/7308 [21:21<54:51,  1.68it/s]
 24%|██▍       | 1767/7308 [21:22<55:57,  1.65it/s]
 24%|██▍       | 1768/7308 [21:22<56:44,  1.63it/s]
 24%|██▍       | 1769/7308 [21:23<57:16,  1.61it/s]
 24%|██▍       | 1770/7308 [21:24<57:38,  1.60it/s]
                                                   
{'loss': '4.248', 'grad_norm': '0.2601', 'learning_rate': '0.0006', 'epoch': '49.17'}
+
 24%|██▍       | 1770/7308 [21:24<57:38,  1.60it/s]
 24%|██▍       | 1771/7308 [21:24<57:57,  1.59it/s]
 24%|██▍       | 1772/7308 [21:25<58:09,  1.59it/s]
 24%|██▍       | 1773/7308 [21:26<58:16,  1.58it/s]
 24%|██▍       | 1774/7308 [21:26<58:19,  1.58it/s]
 24%|██▍       | 1775/7308 [21:27<58:23,  1.58it/s]
 24%|██▍       | 1776/7308 [21:27<58:26,  1.58it/s]
 24%|██▍       | 1777/7308 [21:28<58:27,  1.58it/s]
 24%|██▍       | 1778/7308 [21:29<58:26,  1.58it/s]
 24%|██▍       | 1779/7308 [21:29<58:26,  1.58it/s]
 24%|██▍       | 1780/7308 [21:30<58:26,  1.58it/s]
                                                   
{'loss': '4.191', 'grad_norm': '0.2374', 'learning_rate': '0.0006', 'epoch': '49.45'}
+
 24%|██▍       | 1780/7308 [21:30<58:26,  1.58it/s]
 24%|██▍       | 1781/7308 [21:31<58:28,  1.58it/s]
 24%|██▍       | 1782/7308 [21:31<58:26,  1.58it/s]
 24%|██▍       | 1783/7308 [21:32<58:26,  1.58it/s]
 24%|██▍       | 1784/7308 [21:33<58:24,  1.58it/s]
 24%|██▍       | 1785/7308 [21:33<58:23,  1.58it/s]
 24%|██▍       | 1786/7308 [21:34<58:23,  1.58it/s]
 24%|██▍       | 1787/7308 [21:34<58:22,  1.58it/s]
 24%|██▍       | 1788/7308 [21:35<58:22,  1.58it/s]
 24%|██▍       | 1789/7308 [21:36<58:22,  1.58it/s]
 24%|██▍       | 1790/7308 [21:36<58:22,  1.58it/s]
                                                   
{'loss': '4.223', 'grad_norm': '0.2589', 'learning_rate': '0.0006', 'epoch': '49.73'}
+
 24%|██▍       | 1790/7308 [21:36<58:22,  1.58it/s]
 25%|██▍       | 1791/7308 [21:37<58:32,  1.57it/s]
 25%|██▍       | 1792/7308 [21:38<58:27,  1.57it/s]
 25%|██▍       | 1793/7308 [21:38<58:24,  1.57it/s]
 25%|██▍       | 1794/7308 [21:39<58:21,  1.57it/s]
 25%|██▍       | 1795/7308 [21:39<58:19,  1.58it/s]
 25%|██▍       | 1796/7308 [21:40<58:20,  1.57it/s]
 25%|██▍       | 1797/7308 [21:41<58:19,  1.57it/s]
 25%|██▍       | 1798/7308 [21:41<58:19,  1.57it/s]
 25%|██▍       | 1799/7308 [21:42<58:16,  1.58it/s]
 25%|██▍       | 1800/7308 [21:42<47:29,  1.93it/s]
                                                   
{'loss': '4.204', 'grad_norm': '0.2985', 'learning_rate': '0.0006', 'epoch': '50'}
+
 25%|██▍       | 1800/7308 [21:42<47:29,  1.93it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 23.18it/s]
+
 15%|█▌        | 6/39 [00:00<00:01, 17.69it/s]
+
 21%|██        | 8/39 [00:00<00:01, 16.71it/s]
+
 26%|██▌       | 10/39 [00:00<00:01, 16.15it/s]
+
 31%|███       | 12/39 [00:00<00:01, 15.83it/s]
+
 36%|███▌      | 14/39 [00:00<00:01, 15.62it/s]
+
 41%|████      | 16/39 [00:00<00:01, 15.49it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 15.38it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 15.34it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 15.28it/s]
+
 62%|██████▏   | 24/39 [00:01<00:00, 15.28it/s]
+
 67%|██████▋   | 26/39 [00:01<00:00, 15.27it/s]
+
 72%|███████▏  | 28/39 [00:01<00:00, 15.25it/s]
+
 77%|███████▋  | 30/39 [00:01<00:00, 15.24it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 15.23it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 15.23it/s]
+
 92%|█████████▏| 36/39 [00:02<00:00, 15.20it/s]
+
 97%|█████████▋| 38/39 [00:02<00:00, 15.21it/s]
                                                   
+
                                               
{'eval_loss': '4.085', 'eval_runtime': '2.621', 'eval_samples_per_second': '931.3', 'eval_steps_per_second': '14.88', 'epoch': '50'}
+
 25%|██▍       | 1800/7308 [21:45<47:29,  1.93it/s]
+
100%|██████████| 39/39 [00:02<00:00, 15.21it/s][eval] epoch 50.0 step 1800 loss 4.0851 (best 4.0798 @ ep 46.0, 2/20 since best)
+
+
                                               
 25%|██▍       | 1801/7308 [21:46<2:05:16,  1.36s/it]
 25%|██▍       | 1802/7308 [21:46<1:45:09,  1.15s/it]
 25%|██▍       | 1803/7308 [21:47<1:31:03,  1.01it/s]
 25%|██▍       | 1804/7308 [21:48<1:21:11,  1.13it/s]
 25%|██▍       | 1805/7308 [21:48<1:14:17,  1.23it/s]
 25%|██▍       | 1806/7308 [21:49<1:09:27,  1.32it/s]
 25%|██▍       | 1807/7308 [21:49<1:06:02,  1.39it/s]
 25%|██▍       | 1808/7308 [21:50<1:03:40,  1.44it/s]
 25%|██▍       | 1809/7308 [21:51<1:02:00,  1.48it/s]
 25%|██▍       | 1810/7308 [21:51<1:00:50,  1.51it/s]
                                                     
{'loss': '4.141', 'grad_norm': '0.3257', 'learning_rate': '0.0006', 'epoch': '50.28'}
+
 25%|██▍       | 1810/7308 [21:51<1:00:50,  1.51it/s]
 25%|██▍       | 1811/7308 [21:52<1:00:02,  1.53it/s]
 25%|██▍       | 1812/7308 [21:53<59:27,  1.54it/s]  
 25%|██▍       | 1813/7308 [21:53<59:05,  1.55it/s]
 25%|██▍       | 1814/7308 [21:54<58:46,  1.56it/s]
 25%|██▍       | 1815/7308 [21:55<58:34,  1.56it/s]
 25%|██▍       | 1816/7308 [21:55<58:25,  1.57it/s]
 25%|██▍       | 1817/7308 [21:56<58:19,  1.57it/s]
 25%|██▍       | 1818/7308 [21:56<58:14,  1.57it/s]
 25%|██▍       | 1819/7308 [21:57<58:11,  1.57it/s]
 25%|██▍       | 1820/7308 [21:58<58:07,  1.57it/s]
                                                   
{'loss': '4.079', 'grad_norm': '0.2715', 'learning_rate': '0.0006', 'epoch': '50.57'}
+
 25%|██▍       | 1820/7308 [21:58<58:07,  1.57it/s]
 25%|██▍       | 1821/7308 [21:58<58:06,  1.57it/s]
 25%|██▍       | 1822/7308 [21:59<58:03,  1.57it/s]
 25%|██▍       | 1823/7308 [22:00<58:01,  1.58it/s]
 25%|██▍       | 1824/7308 [22:00<58:00,  1.58it/s]
 25%|██▍       | 1825/7308 [22:01<57:59,  1.58it/s]
 25%|██▍       | 1826/7308 [22:01<57:58,  1.58it/s]
 25%|██▌       | 1827/7308 [22:02<57:58,  1.58it/s]
 25%|██▌       | 1828/7308 [22:03<57:58,  1.58it/s]
 25%|██▌       | 1829/7308 [22:03<57:56,  1.58it/s]
 25%|██▌       | 1830/7308 [22:04<57:56,  1.58it/s]
                                                   
{'loss': '4.187', 'grad_norm': '0.2401', 'learning_rate': '0.0006', 'epoch': '50.85'}
+
 25%|██▌       | 1830/7308 [22:04<57:56,  1.58it/s]
 25%|██▌       | 1831/7308 [22:05<57:55,  1.58it/s]
 25%|██▌       | 1832/7308 [22:05<57:55,  1.58it/s]
 25%|██▌       | 1833/7308 [22:06<57:54,  1.58it/s]
 25%|██▌       | 1834/7308 [22:07<57:55,  1.57it/s]
 25%|██▌       | 1835/7308 [22:07<57:54,  1.58it/s]
 25%|██▌       | 1836/7308 [22:07<47:11,  1.93it/s]
 25%|██▌       | 1837/7308 [22:08<52:36,  1.73it/s]
 25%|██▌       | 1838/7308 [22:09<54:10,  1.68it/s]
 25%|██▌       | 1839/7308 [22:09<55:16,  1.65it/s]
 25%|██▌       | 1840/7308 [22:10<56:01,  1.63it/s]
                                                   
{'loss': '4.125', 'grad_norm': '0.2969', 'learning_rate': '0.0006', 'epoch': '51.11'}
+
 25%|██▌       | 1840/7308 [22:10<56:01,  1.63it/s]
 25%|██▌       | 1841/7308 [22:11<56:35,  1.61it/s]
 25%|██▌       | 1842/7308 [22:11<56:58,  1.60it/s]
 25%|██▌       | 1843/7308 [22:12<57:11,  1.59it/s]
 25%|██▌       | 1844/7308 [22:13<57:23,  1.59it/s]
 25%|██▌       | 1845/7308 [22:13<57:29,  1.58it/s]
 25%|██▌       | 1846/7308 [22:14<57:33,  1.58it/s]
 25%|██▌       | 1847/7308 [22:15<57:35,  1.58it/s]
 25%|██▌       | 1848/7308 [22:15<57:38,  1.58it/s]
 25%|██▌       | 1849/7308 [22:16<57:41,  1.58it/s]
 25%|██▌       | 1850/7308 [22:16<57:41,  1.58it/s]
                                                   
{'loss': '4.12', 'grad_norm': '0.268', 'learning_rate': '0.0006', 'epoch': '51.4'}
+
 25%|██▌       | 1850/7308 [22:16<57:41,  1.58it/s]
 25%|██▌       | 1851/7308 [22:17<57:42,  1.58it/s]
 25%|██▌       | 1852/7308 [22:18<57:43,  1.58it/s]
 25%|██▌       | 1853/7308 [22:18<57:43,  1.57it/s]
 25%|██▌       | 1854/7308 [22:19<57:42,  1.58it/s]
 25%|██▌       | 1855/7308 [22:20<57:41,  1.58it/s]
 25%|██▌       | 1856/7308 [22:20<57:39,  1.58it/s]
 25%|██▌       | 1857/7308 [22:21<57:38,  1.58it/s]
 25%|██▌       | 1858/7308 [22:21<57:37,  1.58it/s]
 25%|██▌       | 1859/7308 [22:22<57:37,  1.58it/s]
 25%|██▌       | 1860/7308 [22:23<57:37,  1.58it/s]
                                                   
{'loss': '4.171', 'grad_norm': '0.3077', 'learning_rate': '0.0006', 'epoch': '51.68'}
+
 25%|██▌       | 1860/7308 [22:23<57:37,  1.58it/s]
 25%|██▌       | 1861/7308 [22:23<57:38,  1.57it/s]
 25%|██▌       | 1862/7308 [22:24<57:37,  1.57it/s]
 25%|██▌       | 1863/7308 [22:25<57:35,  1.58it/s]
 26%|██▌       | 1864/7308 [22:25<57:35,  1.58it/s]
 26%|██▌       | 1865/7308 [22:26<57:36,  1.57it/s]
 26%|██▌       | 1866/7308 [22:27<57:33,  1.58it/s]
 26%|██▌       | 1867/7308 [22:27<57:34,  1.58it/s]
 26%|██▌       | 1868/7308 [22:28<57:33,  1.58it/s]
 26%|██▌       | 1869/7308 [22:28<57:32,  1.58it/s]
 26%|██▌       | 1870/7308 [22:29<57:31,  1.58it/s]
                                                   
{'loss': '4.159', 'grad_norm': '0.2589', 'learning_rate': '0.0006', 'epoch': '51.96'}
+
 26%|██▌       | 1870/7308 [22:29<57:31,  1.58it/s]
 26%|██▌       | 1871/7308 [22:30<57:31,  1.58it/s]
 26%|██▌       | 1872/7308 [22:30<46:52,  1.93it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 22.58it/s]
+
 15%|█▌        | 6/39 [00:00<00:01, 17.56it/s]
+
 21%|██        | 8/39 [00:00<00:01, 16.62it/s]
+
 26%|██▌       | 10/39 [00:00<00:01, 16.11it/s]
+
 31%|███       | 12/39 [00:00<00:01, 15.77it/s]
+
 36%|███▌      | 14/39 [00:00<00:01, 15.59it/s]
+
 41%|████      | 16/39 [00:00<00:01, 15.46it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 15.38it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 15.33it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 15.30it/s]
+
 62%|██████▏   | 24/39 [00:01<00:00, 15.31it/s]
+
 67%|██████▋   | 26/39 [00:01<00:00, 15.29it/s]
+
 72%|███████▏  | 28/39 [00:01<00:00, 15.25it/s]
+
 77%|███████▋  | 30/39 [00:01<00:00, 15.25it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 15.23it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 15.22it/s]
+
 92%|█████████▏| 36/39 [00:02<00:00, 15.21it/s]
+
 97%|█████████▋| 38/39 [00:02<00:00, 15.22it/s]
                                                   
+
                                               
{'eval_loss': '4.091', 'eval_runtime': '2.635', 'eval_samples_per_second': '926.2', 'eval_steps_per_second': '14.8', 'epoch': '52'}
+
 26%|██▌       | 1872/7308 [22:33<46:52,  1.93it/s]
+
100%|██████████| 39/39 [00:02<00:00, 15.22it/s][eval] epoch 52.0 step 1872 loss 4.0905 (best 4.0798 @ ep 46.0, 3/20 since best)
+
+
                                               
 26%|██▌       | 1873/7308 [22:33<2:03:41,  1.37s/it]
 26%|██▌       | 1874/7308 [22:34<1:43:48,  1.15s/it]
 26%|██▌       | 1875/7308 [22:35<1:29:52,  1.01it/s]
 26%|██▌       | 1876/7308 [22:35<1:20:08,  1.13it/s]
 26%|██▌       | 1877/7308 [22:36<1:13:18,  1.23it/s]
 26%|██▌       | 1878/7308 [22:37<1:08:32,  1.32it/s]
 26%|██▌       | 1879/7308 [22:37<1:05:11,  1.39it/s]
 26%|██▌       | 1880/7308 [22:38<1:02:50,  1.44it/s]
                                                     
{'loss': '4.153', 'grad_norm': '0.2844', 'learning_rate': '0.0006', 'epoch': '52.23'}
+
 26%|██▌       | 1880/7308 [22:38<1:02:50,  1.44it/s]
 26%|██▌       | 1881/7308 [22:38<1:01:12,  1.48it/s]
 26%|██▌       | 1882/7308 [22:39<1:00:02,  1.51it/s]
 26%|██▌       | 1883/7308 [22:40<59:13,  1.53it/s]  
 26%|██▌       | 1884/7308 [22:40<58:38,  1.54it/s]
 26%|██▌       | 1885/7308 [22:41<58:16,  1.55it/s]
 26%|██▌       | 1886/7308 [22:42<57:58,  1.56it/s]
 26%|██▌       | 1887/7308 [22:42<57:46,  1.56it/s]
 26%|██▌       | 1888/7308 [22:43<57:36,  1.57it/s]
 26%|██▌       | 1889/7308 [22:43<57:31,  1.57it/s]
 26%|██▌       | 1890/7308 [22:44<57:27,  1.57it/s]
                                                   
{'loss': '4.143', 'grad_norm': '0.2508', 'learning_rate': '0.0006', 'epoch': '52.51'}
+
 26%|██▌       | 1890/7308 [22:44<57:27,  1.57it/s]
 26%|██▌       | 1891/7308 [22:45<57:23,  1.57it/s]
 26%|██▌       | 1892/7308 [22:45<57:20,  1.57it/s]
 26%|██▌       | 1893/7308 [22:46<57:20,  1.57it/s]
 26%|██▌       | 1894/7308 [22:47<57:19,  1.57it/s]
 26%|██▌       | 1895/7308 [22:47<57:18,  1.57it/s]
 26%|██▌       | 1896/7308 [22:48<57:16,  1.58it/s]
 26%|██▌       | 1897/7308 [22:49<57:14,  1.58it/s]
 26%|██▌       | 1898/7308 [22:49<57:14,  1.58it/s]
 26%|██▌       | 1899/7308 [22:50<57:12,  1.58it/s]
 26%|██▌       | 1900/7308 [22:50<57:10,  1.58it/s]
                                                   
{'loss': '4.185', 'grad_norm': '0.2494', 'learning_rate': '0.0006', 'epoch': '52.79'}
+
 26%|██▌       | 1900/7308 [22:50<57:10,  1.58it/s]
 26%|██▌       | 1901/7308 [22:51<57:13,  1.58it/s]
 26%|██▌       | 1902/7308 [22:52<57:11,  1.58it/s]
 26%|██▌       | 1903/7308 [22:52<57:10,  1.58it/s]
 26%|██▌       | 1904/7308 [22:53<57:09,  1.58it/s]
 26%|██▌       | 1905/7308 [22:54<57:07,  1.58it/s]
 26%|██▌       | 1906/7308 [22:54<57:07,  1.58it/s]
 26%|██▌       | 1907/7308 [22:55<57:05,  1.58it/s]
 26%|██▌       | 1908/7308 [22:55<46:31,  1.93it/s]
 26%|██▌       | 1909/7308 [22:56<52:17,  1.72it/s]
 26%|██▌       | 1910/7308 [22:57<53:43,  1.67it/s]
                                                   
{'loss': '4.092', 'grad_norm': '0.246', 'learning_rate': '0.0006', 'epoch': '53.06'}
+
 26%|██▌       | 1910/7308 [22:57<53:43,  1.67it/s]
 26%|██▌       | 1911/7308 [22:57<54:47,  1.64it/s]
 26%|██▌       | 1912/7308 [22:58<55:27,  1.62it/s]
 26%|██▌       | 1913/7308 [22:58<55:56,  1.61it/s]
 26%|██▌       | 1914/7308 [22:59<56:16,  1.60it/s]
 26%|██▌       | 1915/7308 [23:00<56:28,  1.59it/s]
 26%|██▌       | 1916/7308 [23:00<56:38,  1.59it/s]
 26%|██▌       | 1917/7308 [23:01<56:45,  1.58it/s]
 26%|██▌       | 1918/7308 [23:02<56:48,  1.58it/s]
 26%|██▋       | 1919/7308 [23:02<56:50,  1.58it/s]
 26%|██▋       | 1920/7308 [23:03<56:52,  1.58it/s]
                                                   
{'loss': '4.073', 'grad_norm': '0.2782', 'learning_rate': '0.0006', 'epoch': '53.34'}
+
 26%|██▋       | 1920/7308 [23:03<56:52,  1.58it/s]
 26%|██▋       | 1921/7308 [23:03<56:54,  1.58it/s]
 26%|██▋       | 1922/7308 [23:04<56:54,  1.58it/s]
 26%|██▋       | 1923/7308 [23:05<56:54,  1.58it/s]
 26%|██▋       | 1924/7308 [23:05<56:54,  1.58it/s]
 26%|██▋       | 1925/7308 [23:06<56:54,  1.58it/s]
 26%|██▋       | 1926/7308 [23:07<56:55,  1.58it/s]
 26%|██▋       | 1927/7308 [23:07<56:55,  1.58it/s]
 26%|██▋       | 1928/7308 [23:08<56:54,  1.58it/s]
 26%|██▋       | 1929/7308 [23:09<56:55,  1.58it/s]
 26%|██▋       | 1930/7308 [23:09<56:53,  1.58it/s]
                                                   
{'loss': '4.094', 'grad_norm': '0.2792', 'learning_rate': '0.0006', 'epoch': '53.62'}
+
 26%|██▋       | 1930/7308 [23:09<56:53,  1.58it/s]
 26%|██▋       | 1931/7308 [23:10<56:53,  1.58it/s]
 26%|██▋       | 1932/7308 [23:10<56:52,  1.58it/s]
 26%|██▋       | 1933/7308 [23:11<56:51,  1.58it/s]
 26%|██▋       | 1934/7308 [23:12<56:49,  1.58it/s]
 26%|██▋       | 1935/7308 [23:12<56:50,  1.58it/s]
 26%|██▋       | 1936/7308 [23:13<56:50,  1.58it/s]
 27%|██▋       | 1937/7308 [23:14<56:48,  1.58it/s]
 27%|██▋       | 1938/7308 [23:14<56:47,  1.58it/s]
 27%|██▋       | 1939/7308 [23:15<56:46,  1.58it/s]
 27%|██▋       | 1940/7308 [23:16<56:46,  1.58it/s]
                                                   
{'loss': '4.111', 'grad_norm': '0.2478', 'learning_rate': '0.0006', 'epoch': '53.9'}
+
 27%|██▋       | 1940/7308 [23:16<56:46,  1.58it/s]
 27%|██▋       | 1941/7308 [23:16<56:47,  1.58it/s]
 27%|██▋       | 1942/7308 [23:17<56:47,  1.57it/s]
 27%|██▋       | 1943/7308 [23:17<56:45,  1.58it/s]
 27%|██▋       | 1944/7308 [23:18<46:15,  1.93it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 22.96it/s]
+
 15%|█▌        | 6/39 [00:00<00:01, 17.61it/s]
+
 21%|██        | 8/39 [00:00<00:01, 16.65it/s]
+
 26%|██▌       | 10/39 [00:00<00:01, 16.10it/s]
+
 31%|███       | 12/39 [00:00<00:01, 15.79it/s]
+
 36%|███▌      | 14/39 [00:00<00:01, 15.58it/s]
+
 41%|████      | 16/39 [00:00<00:01, 15.45it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 15.34it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 15.28it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 15.24it/s]
+
 62%|██████▏   | 24/39 [00:01<00:00, 15.23it/s]
+
 67%|██████▋   | 26/39 [00:01<00:00, 15.23it/s]
+
 72%|███████▏  | 28/39 [00:01<00:00, 15.22it/s]
+
 77%|███████▋  | 30/39 [00:01<00:00, 15.21it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 15.22it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 15.20it/s]
+
 92%|█████████▏| 36/39 [00:02<00:00, 15.21it/s]
+
 97%|█████████▋| 38/39 [00:02<00:00, 15.21it/s]
                                                   
+
                                               
{'eval_loss': '4.102', 'eval_runtime': '2.626', 'eval_samples_per_second': '929.7', 'eval_steps_per_second': '14.85', 'epoch': '54'}
+
 27%|██▋       | 1944/7308 [23:20<46:15,  1.93it/s]
+
100%|██████████| 39/39 [00:02<00:00, 15.21it/s][eval] epoch 54.0 step 1944 loss 4.1025 (best 4.0798 @ ep 46.0, 4/20 since best)
+
+
                                               
 27%|██▋       | 1945/7308 [23:21<2:02:24,  1.37s/it]
 27%|██▋       | 1946/7308 [23:22<1:42:42,  1.15s/it]
 27%|██▋       | 1947/7308 [23:22<1:28:52,  1.01it/s]
 27%|██▋       | 1948/7308 [23:23<1:19:11,  1.13it/s]
 27%|██▋       | 1949/7308 [23:24<1:12:26,  1.23it/s]
 27%|██▋       | 1950/7308 [23:24<1:07:42,  1.32it/s]
                                                     
{'loss': '4.104', 'grad_norm': '0.4483', 'learning_rate': '0.0006', 'epoch': '54.17'}
+
 27%|██▋       | 1950/7308 [23:24<1:07:42,  1.32it/s]
 27%|██▋       | 1951/7308 [23:25<1:04:24,  1.39it/s]
 27%|██▋       | 1952/7308 [23:26<1:02:03,  1.44it/s]
 27%|██▋       | 1953/7308 [23:26<1:00:26,  1.48it/s]
 27%|██▋       | 1954/7308 [23:27<59:17,  1.50it/s]  
 27%|██▋       | 1955/7308 [23:27<58:32,  1.52it/s]
 27%|██▋       | 1956/7308 [23:28<57:57,  1.54it/s]
 27%|██▋       | 1957/7308 [23:29<57:33,  1.55it/s]
 27%|██▋       | 1958/7308 [23:29<57:16,  1.56it/s]
 27%|██▋       | 1959/7308 [23:30<57:03,  1.56it/s]
 27%|██▋       | 1960/7308 [23:31<56:55,  1.57it/s]
                                                   
{'loss': '4.152', 'grad_norm': '0.2986', 'learning_rate': '0.0006', 'epoch': '54.45'}
+
 27%|██▋       | 1960/7308 [23:31<56:55,  1.57it/s]
 27%|██▋       | 1961/7308 [23:31<56:49,  1.57it/s]
 27%|██▋       | 1962/7308 [23:32<56:43,  1.57it/s]
 27%|██▋       | 1963/7308 [23:32<56:39,  1.57it/s]
 27%|██▋       | 1964/7308 [23:33<56:36,  1.57it/s]
 27%|██▋       | 1965/7308 [23:34<56:34,  1.57it/s]
 27%|██▋       | 1966/7308 [23:34<56:32,  1.57it/s]
 27%|██▋       | 1967/7308 [23:35<56:31,  1.58it/s]
 27%|██▋       | 1968/7308 [23:36<56:32,  1.57it/s]
 27%|██▋       | 1969/7308 [23:36<56:30,  1.57it/s]
 27%|██▋       | 1970/7308 [23:37<56:30,  1.57it/s]
                                                   
{'loss': '4.082', 'grad_norm': '0.3224', 'learning_rate': '0.0006', 'epoch': '54.73'}
+
 27%|██▋       | 1970/7308 [23:37<56:30,  1.57it/s]
 27%|██▋       | 1971/7308 [23:38<56:28,  1.57it/s]
 27%|██▋       | 1972/7308 [23:38<56:28,  1.57it/s]
 27%|██▋       | 1973/7308 [23:39<56:27,  1.57it/s]
 27%|██▋       | 1974/7308 [23:39<56:27,  1.57it/s]
 27%|██▋       | 1975/7308 [23:40<56:25,  1.58it/s]
 27%|██▋       | 1976/7308 [23:41<56:24,  1.58it/s]
 27%|██▋       | 1977/7308 [23:41<56:23,  1.58it/s]
 27%|██▋       | 1978/7308 [23:42<56:23,  1.58it/s]
 27%|██▋       | 1979/7308 [23:43<56:21,  1.58it/s]
 27%|██▋       | 1980/7308 [23:43<45:54,  1.93it/s]
                                                   
{'loss': '4.174', 'grad_norm': '0.3145', 'learning_rate': '0.0006', 'epoch': '55'}
+
 27%|██▋       | 1980/7308 [23:43<45:54,  1.93it/s]
 27%|██▋       | 1981/7308 [23:44<51:42,  1.72it/s]
 27%|██▋       | 1982/7308 [23:44<53:04,  1.67it/s]
 27%|██▋       | 1983/7308 [23:45<54:02,  1.64it/s]
 27%|██▋       | 1984/7308 [23:46<54:42,  1.62it/s]
 27%|██▋       | 1985/7308 [23:46<55:11,  1.61it/s]
 27%|██▋       | 1986/7308 [23:47<55:31,  1.60it/s]
 27%|██▋       | 1987/7308 [23:47<55:45,  1.59it/s]
 27%|██▋       | 1988/7308 [23:48<55:53,  1.59it/s]
 27%|██▋       | 1989/7308 [23:49<55:59,  1.58it/s]
 27%|██▋       | 1990/7308 [23:49<56:03,  1.58it/s]
                                                   
{'loss': '4.051', 'grad_norm': '0.2843', 'learning_rate': '0.0006', 'epoch': '55.28'}
+
 27%|██▋       | 1990/7308 [23:49<56:03,  1.58it/s]
 27%|██▋       | 1991/7308 [23:50<56:07,  1.58it/s]
 27%|██▋       | 1992/7308 [23:51<56:08,  1.58it/s]
 27%|██▋       | 1993/7308 [23:51<56:09,  1.58it/s]
 27%|██▋       | 1994/7308 [23:52<56:12,  1.58it/s]
 27%|██▋       | 1995/7308 [23:53<56:13,  1.57it/s]
 27%|██▋       | 1996/7308 [23:53<56:14,  1.57it/s]
 27%|██▋       | 1997/7308 [23:54<56:14,  1.57it/s]
 27%|██▋       | 1998/7308 [23:54<56:15,  1.57it/s]
 27%|██▋       | 1999/7308 [23:55<56:15,  1.57it/s]
 27%|██▋       | 2000/7308 [23:56<56:15,  1.57it/s]
                                                   
{'loss': '4.071', 'grad_norm': '0.2741', 'learning_rate': '0.0006', 'epoch': '55.57'}
+
 27%|██▋       | 2000/7308 [23:56<56:15,  1.57it/s]
 27%|██▋       | 2001/7308 [23:56<56:15,  1.57it/s]
 27%|██▋       | 2002/7308 [23:57<56:15,  1.57it/s]
 27%|██▋       | 2003/7308 [23:58<56:14,  1.57it/s]
 27%|██▋       | 2004/7308 [23:58<56:14,  1.57it/s]
 27%|██▋       | 2005/7308 [23:59<56:12,  1.57it/s]
 27%|██▋       | 2006/7308 [24:00<56:11,  1.57it/s]
 27%|██▋       | 2007/7308 [24:00<56:14,  1.57it/s]
 27%|██▋       | 2008/7308 [24:01<56:12,  1.57it/s]
 27%|██▋       | 2009/7308 [24:01<56:11,  1.57it/s]
 28%|██▊       | 2010/7308 [24:02<56:09,  1.57it/s]
                                                   
{'loss': '4.098', 'grad_norm': '0.2774', 'learning_rate': '0.0006', 'epoch': '55.85'}
+
 28%|██▊       | 2010/7308 [24:02<56:09,  1.57it/s]
 28%|██▊       | 2011/7308 [24:03<56:08,  1.57it/s]
 28%|██▊       | 2012/7308 [24:03<56:08,  1.57it/s]
 28%|██▊       | 2013/7308 [24:04<56:08,  1.57it/s]
 28%|██▊       | 2014/7308 [24:05<56:05,  1.57it/s]
 28%|██▊       | 2015/7308 [24:05<56:01,  1.57it/s]
 28%|██▊       | 2016/7308 [24:05<45:38,  1.93it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 23.43it/s]
+
 15%|█▌        | 6/39 [00:00<00:01, 17.77it/s]
+
 21%|██        | 8/39 [00:00<00:01, 16.72it/s]
+
 26%|██▌       | 10/39 [00:00<00:01, 16.17it/s]
+
 31%|███       | 12/39 [00:00<00:01, 15.81it/s]
+
 36%|███▌      | 14/39 [00:00<00:01, 15.60it/s]
+
 41%|████      | 16/39 [00:00<00:01, 15.45it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 15.37it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 15.29it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 15.25it/s]
+
 62%|██████▏   | 24/39 [00:01<00:00, 15.24it/s]
+
 67%|██████▋   | 26/39 [00:01<00:00, 15.24it/s]
+
 72%|███████▏  | 28/39 [00:01<00:00, 15.22it/s]
+
 77%|███████▋  | 30/39 [00:01<00:00, 15.22it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 15.21it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 15.22it/s]
+
 92%|█████████▏| 36/39 [00:02<00:00, 15.19it/s]
+
 97%|█████████▋| 38/39 [00:02<00:00, 15.17it/s]
                                                   
+
                                               
{'eval_loss': '4.11', 'eval_runtime': '2.627', 'eval_samples_per_second': '929.2', 'eval_steps_per_second': '14.85', 'epoch': '56'}
+
 28%|██▊       | 2016/7308 [24:08<45:38,  1.93it/s]
+
100%|██████████| 39/39 [00:02<00:00, 15.17it/s][eval] epoch 56.0 step 2016 loss 4.1095 (best 4.0798 @ ep 46.0, 5/20 since best)
+
+
                                               
 28%|██▊       | 2017/7308 [24:09<2:00:25,  1.37s/it]
 28%|██▊       | 2018/7308 [24:09<1:41:05,  1.15s/it]
 28%|██▊       | 2019/7308 [24:10<1:27:33,  1.01it/s]
 28%|██▊       | 2020/7308 [24:11<1:18:05,  1.13it/s]
                                                     
{'loss': '4.095', 'grad_norm': '0.2809', 'learning_rate': '0.0006', 'epoch': '56.11'}
+
 28%|██▊       | 2020/7308 [24:11<1:18:05,  1.13it/s]
 28%|██▊       | 2021/7308 [24:11<1:11:29,  1.23it/s]
 28%|██▊       | 2022/7308 [24:12<1:06:51,  1.32it/s]
 28%|██▊       | 2023/7308 [24:13<1:03:35,  1.39it/s]
 28%|██▊       | 2024/7308 [24:13<1:01:18,  1.44it/s]
 28%|██▊       | 2025/7308 [24:14<59:43,  1.47it/s]  
 28%|██▊       | 2026/7308 [24:15<58:35,  1.50it/s]
 28%|██▊       | 2027/7308 [24:15<57:48,  1.52it/s]
 28%|██▊       | 2028/7308 [24:16<57:14,  1.54it/s]
 28%|██▊       | 2029/7308 [24:16<56:49,  1.55it/s]
 28%|██▊       | 2030/7308 [24:17<56:33,  1.56it/s]
                                                   
{'loss': '3.994', 'grad_norm': '0.2864', 'learning_rate': '0.0006', 'epoch': '56.4'}
+
 28%|██▊       | 2030/7308 [24:17<56:33,  1.56it/s]
 28%|██▊       | 2031/7308 [24:18<56:21,  1.56it/s]
 28%|██▊       | 2032/7308 [24:18<56:14,  1.56it/s]
 28%|██▊       | 2033/7308 [24:19<56:07,  1.57it/s]
 28%|██▊       | 2034/7308 [24:20<56:02,  1.57it/s]
 28%|██▊       | 2035/7308 [24:20<55:59,  1.57it/s]
 28%|██▊       | 2036/7308 [24:21<55:57,  1.57it/s]
 28%|██▊       | 2037/7308 [24:22<55:55,  1.57it/s]
 28%|██▊       | 2038/7308 [24:22<55:53,  1.57it/s]
 28%|██▊       | 2039/7308 [24:23<55:51,  1.57it/s]
 28%|██▊       | 2040/7308 [24:23<55:50,  1.57it/s]
                                                   
{'loss': '4.089', 'grad_norm': '0.2964', 'learning_rate': '0.0006', 'epoch': '56.68'}
+
 28%|██▊       | 2040/7308 [24:23<55:50,  1.57it/s]
 28%|██▊       | 2041/7308 [24:24<55:53,  1.57it/s]
 28%|██▊       | 2042/7308 [24:25<55:51,  1.57it/s]
 28%|██▊       | 2043/7308 [24:25<55:50,  1.57it/s]
 28%|██▊       | 2044/7308 [24:26<55:48,  1.57it/s]
 28%|██▊       | 2045/7308 [24:27<55:47,  1.57it/s]
 28%|██▊       | 2046/7308 [24:27<55:46,  1.57it/s]
 28%|██▊       | 2047/7308 [24:28<55:45,  1.57it/s]
 28%|██▊       | 2048/7308 [24:29<55:45,  1.57it/s]
 28%|██▊       | 2049/7308 [24:29<55:44,  1.57it/s]
 28%|██▊       | 2050/7308 [24:30<55:42,  1.57it/s]
                                                   
{'loss': '4.07', 'grad_norm': '0.3101', 'learning_rate': '0.0006', 'epoch': '56.96'}
+
 28%|██▊       | 2050/7308 [24:30<55:42,  1.57it/s]
 28%|██▊       | 2051/7308 [24:30<55:39,  1.57it/s]
 28%|██▊       | 2052/7308 [24:31<45:21,  1.93it/s]
 28%|██▊       | 2053/7308 [24:31<50:34,  1.73it/s]
 28%|██▊       | 2054/7308 [24:32<52:03,  1.68it/s]
 28%|██▊       | 2055/7308 [24:33<53:05,  1.65it/s]
 28%|██▊       | 2056/7308 [24:33<53:48,  1.63it/s]
 28%|██▊       | 2057/7308 [24:34<54:20,  1.61it/s]
 28%|██▊       | 2058/7308 [24:35<54:40,  1.60it/s]
 28%|██▊       | 2059/7308 [24:35<54:55,  1.59it/s]
 28%|██▊       | 2060/7308 [24:36<55:05,  1.59it/s]
                                                   
{'loss': '4.136', 'grad_norm': '0.2918', 'learning_rate': '0.0006', 'epoch': '57.23'}
+
 28%|██▊       | 2060/7308 [24:36<55:05,  1.59it/s]
 28%|██▊       | 2061/7308 [24:36<55:13,  1.58it/s]
 28%|██▊       | 2062/7308 [24:37<55:19,  1.58it/s]
 28%|██▊       | 2063/7308 [24:38<55:21,  1.58it/s]
 28%|██▊       | 2064/7308 [24:38<55:22,  1.58it/s]
 28%|██▊       | 2065/7308 [24:39<55:24,  1.58it/s]
 28%|██▊       | 2066/7308 [24:40<55:24,  1.58it/s]
 28%|██▊       | 2067/7308 [24:40<55:24,  1.58it/s]
 28%|██▊       | 2068/7308 [24:41<55:23,  1.58it/s]
 28%|██▊       | 2069/7308 [24:42<55:24,  1.58it/s]
 28%|██▊       | 2070/7308 [24:42<55:23,  1.58it/s]
                                                   
{'loss': '4.063', 'grad_norm': '0.3186', 'learning_rate': '0.0006', 'epoch': '57.51'}
+
 28%|██▊       | 2070/7308 [24:42<55:23,  1.58it/s]
 28%|██▊       | 2071/7308 [24:43<55:23,  1.58it/s]
 28%|██▊       | 2072/7308 [24:43<55:23,  1.58it/s]
 28%|██▊       | 2073/7308 [24:44<55:24,  1.57it/s]
 28%|██▊       | 2074/7308 [24:45<55:22,  1.58it/s]
 28%|██▊       | 2075/7308 [24:45<55:21,  1.58it/s]
 28%|██▊       | 2076/7308 [24:46<55:20,  1.58it/s]
 28%|██▊       | 2077/7308 [24:47<55:18,  1.58it/s]
 28%|██▊       | 2078/7308 [24:47<55:18,  1.58it/s]
 28%|██▊       | 2079/7308 [24:48<55:17,  1.58it/s]
 28%|██▊       | 2080/7308 [24:49<55:17,  1.58it/s]
                                                   
{'loss': '4.122', 'grad_norm': '0.2541', 'learning_rate': '0.0006', 'epoch': '57.79'}
+
 28%|██▊       | 2080/7308 [24:49<55:17,  1.58it/s]
 28%|██▊       | 2081/7308 [24:49<55:18,  1.57it/s]
 28%|██▊       | 2082/7308 [24:50<55:18,  1.57it/s]
 29%|██▊       | 2083/7308 [24:50<55:17,  1.57it/s]
 29%|██▊       | 2084/7308 [24:51<55:15,  1.58it/s]
 29%|██▊       | 2085/7308 [24:52<55:16,  1.58it/s]
 29%|██▊       | 2086/7308 [24:52<55:15,  1.58it/s]
 29%|██▊       | 2087/7308 [24:53<55:12,  1.58it/s]
 29%|██▊       | 2088/7308 [24:53<44:59,  1.93it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 22.83it/s]
+
 15%|█▌        | 6/39 [00:00<00:01, 17.61it/s]
+
 21%|██        | 8/39 [00:00<00:01, 16.65it/s]
+
 26%|██▌       | 10/39 [00:00<00:01, 16.12it/s]
+
 31%|███       | 12/39 [00:00<00:01, 15.79it/s]
+
 36%|███▌      | 14/39 [00:00<00:01, 15.58it/s]
+
 41%|████      | 16/39 [00:00<00:01, 15.45it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 15.36it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 15.31it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 15.26it/s]
+
 62%|██████▏   | 24/39 [00:01<00:00, 15.25it/s]
+
 67%|██████▋   | 26/39 [00:01<00:00, 15.24it/s]
+
 72%|███████▏  | 28/39 [00:01<00:00, 15.24it/s]
+
 77%|███████▋  | 30/39 [00:01<00:00, 15.23it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 15.24it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 15.23it/s]
+
 92%|█████████▏| 36/39 [00:02<00:00, 15.24it/s]
+
 97%|█████████▋| 38/39 [00:02<00:00, 15.22it/s]
                                                   
+
                                               
{'eval_loss': '4.112', 'eval_runtime': '2.624', 'eval_samples_per_second': '930.2', 'eval_steps_per_second': '14.86', 'epoch': '58'}
+
 29%|██▊       | 2088/7308 [24:56<44:59,  1.93it/s]
+
100%|██████████| 39/39 [00:02<00:00, 15.22it/s][eval] epoch 58.0 step 2088 loss 4.1123 (best 4.0798 @ ep 46.0, 6/20 since best)
+
+
                                               
 29%|██▊       | 2089/7308 [24:57<1:59:02,  1.37s/it]
 29%|██▊       | 2090/7308 [24:57<1:39:52,  1.15s/it]
                                                     
{'loss': '4.03', 'grad_norm': '0.2785', 'learning_rate': '0.0006', 'epoch': '58.06'}
+
 29%|██▊       | 2090/7308 [24:57<1:39:52,  1.15s/it]
 29%|██▊       | 2091/7308 [24:58<1:26:27,  1.01it/s]
 29%|██▊       | 2092/7308 [24:58<1:17:03,  1.13it/s]
 29%|██▊       | 2093/7308 [24:59<1:10:30,  1.23it/s]
 29%|██▊       | 2094/7308 [25:00<1:05:54,  1.32it/s]
 29%|██▊       | 2095/7308 [25:00<1:02:39,  1.39it/s]
 29%|██▊       | 2096/7308 [25:01<1:00:22,  1.44it/s]
 29%|██▊       | 2097/7308 [25:02<58:47,  1.48it/s]  
 29%|██▊       | 2098/7308 [25:02<57:41,  1.51it/s]
 29%|██▊       | 2099/7308 [25:03<56:54,  1.53it/s]
 29%|██▊       | 2100/7308 [25:04<56:20,  1.54it/s]
                                                   
{'loss': '4.032', 'grad_norm': '0.2863', 'learning_rate': '0.0006', 'epoch': '58.34'}
+
 29%|██▊       | 2100/7308 [25:04<56:20,  1.54it/s]
 29%|██▊       | 2101/7308 [25:04<55:59,  1.55it/s]
 29%|██▉       | 2102/7308 [25:05<55:41,  1.56it/s]
 29%|██▉       | 2103/7308 [25:05<55:29,  1.56it/s]
 29%|██▉       | 2104/7308 [25:06<55:19,  1.57it/s]
 29%|██▉       | 2105/7308 [25:07<55:13,  1.57it/s]
 29%|██▉       | 2106/7308 [25:07<55:08,  1.57it/s]
 29%|██▉       | 2107/7308 [25:08<55:04,  1.57it/s]
 29%|██▉       | 2108/7308 [25:09<55:02,  1.57it/s]
 29%|██▉       | 2109/7308 [25:09<55:00,  1.58it/s]
 29%|██▉       | 2110/7308 [25:10<54:59,  1.58it/s]
                                                   
{'loss': '4.057', 'grad_norm': '0.2741', 'learning_rate': '0.0006', 'epoch': '58.62'}
+
 29%|██▉       | 2110/7308 [25:10<54:59,  1.58it/s]
 29%|██▉       | 2111/7308 [25:11<54:58,  1.58it/s]
 29%|██▉       | 2112/7308 [25:11<54:56,  1.58it/s]
 29%|██▉       | 2113/7308 [25:12<54:56,  1.58it/s]
 29%|██▉       | 2114/7308 [25:12<54:55,  1.58it/s]
 29%|██▉       | 2115/7308 [25:13<54:54,  1.58it/s]
 29%|██▉       | 2116/7308 [25:14<54:53,  1.58it/s]
 29%|██▉       | 2117/7308 [25:14<54:51,  1.58it/s]
 29%|██▉       | 2118/7308 [25:15<54:51,  1.58it/s]
 29%|██▉       | 2119/7308 [25:16<54:49,  1.58it/s]
 29%|██▉       | 2120/7308 [25:16<54:48,  1.58it/s]
                                                   
{'loss': '4.068', 'grad_norm': '0.3107', 'learning_rate': '0.0006', 'epoch': '58.9'}
+
 29%|██▉       | 2120/7308 [25:16<54:48,  1.58it/s]
 29%|██▉       | 2121/7308 [25:17<54:49,  1.58it/s]
 29%|██▉       | 2122/7308 [25:18<54:48,  1.58it/s]
 29%|██▉       | 2123/7308 [25:18<54:46,  1.58it/s]
 29%|██▉       | 2124/7308 [25:18<44:37,  1.94it/s]
 29%|██▉       | 2125/7308 [25:19<50:08,  1.72it/s]
 29%|██▉       | 2126/7308 [25:20<51:32,  1.68it/s]
 29%|██▉       | 2127/7308 [25:20<52:29,  1.65it/s]
 29%|██▉       | 2128/7308 [25:21<53:10,  1.62it/s]
 29%|██▉       | 2129/7308 [25:22<53:38,  1.61it/s]
 29%|██▉       | 2130/7308 [25:22<53:57,  1.60it/s]
                                                   
{'loss': '4.011', 'grad_norm': '0.3925', 'learning_rate': '0.0006', 'epoch': '59.17'}
+
 29%|██▉       | 2130/7308 [25:22<53:57,  1.60it/s]
 29%|██▉       | 2131/7308 [25:23<54:12,  1.59it/s]
 29%|██▉       | 2132/7308 [25:24<54:21,  1.59it/s]
 29%|██▉       | 2133/7308 [25:24<54:28,  1.58it/s]
 29%|██▉       | 2134/7308 [25:25<54:31,  1.58it/s]
 29%|██▉       | 2135/7308 [25:25<54:33,  1.58it/s]
 29%|██▉       | 2136/7308 [25:26<54:35,  1.58it/s]
 29%|██▉       | 2137/7308 [25:27<54:38,  1.58it/s]
 29%|██▉       | 2138/7308 [25:27<54:39,  1.58it/s]
 29%|██▉       | 2139/7308 [25:28<54:38,  1.58it/s]
 29%|██▉       | 2140/7308 [25:29<54:38,  1.58it/s]
                                                   
{'loss': '4.001', 'grad_norm': '0.259', 'learning_rate': '0.0006', 'epoch': '59.45'}
+
 29%|██▉       | 2140/7308 [25:29<54:38,  1.58it/s]
 29%|██▉       | 2141/7308 [25:29<54:39,  1.58it/s]
 29%|██▉       | 2142/7308 [25:30<54:38,  1.58it/s]
 29%|██▉       | 2143/7308 [25:31<54:39,  1.57it/s]
 29%|██▉       | 2144/7308 [25:31<54:39,  1.57it/s]
 29%|██▉       | 2145/7308 [25:32<54:38,  1.57it/s]
 29%|██▉       | 2146/7308 [25:32<54:36,  1.58it/s]
 29%|██▉       | 2147/7308 [25:33<54:37,  1.57it/s]
 29%|██▉       | 2148/7308 [25:34<54:36,  1.57it/s]
 29%|██▉       | 2149/7308 [25:34<54:36,  1.57it/s]
 29%|██▉       | 2150/7308 [25:35<54:34,  1.58it/s]
                                                   
{'loss': '4.056', 'grad_norm': '0.268', 'learning_rate': '0.0006', 'epoch': '59.73'}
+
 29%|██▉       | 2150/7308 [25:35<54:34,  1.58it/s]
 29%|██▉       | 2151/7308 [25:36<54:34,  1.58it/s]
 29%|██▉       | 2152/7308 [25:36<54:33,  1.58it/s]
 29%|██▉       | 2153/7308 [25:37<54:35,  1.57it/s]
 29%|██▉       | 2154/7308 [25:38<54:33,  1.57it/s]
 29%|██▉       | 2155/7308 [25:38<54:31,  1.57it/s]
 30%|██▉       | 2156/7308 [25:39<54:30,  1.58it/s]
 30%|██▉       | 2157/7308 [25:39<54:29,  1.58it/s]
 30%|██▉       | 2158/7308 [25:40<54:28,  1.58it/s]
 30%|██▉       | 2159/7308 [25:41<54:29,  1.58it/s]
 30%|██▉       | 2160/7308 [25:41<44:23,  1.93it/s]
                                                   
{'loss': '4.152', 'grad_norm': '0.3184', 'learning_rate': '0.0006', 'epoch': '60'}
+
 30%|██▉       | 2160/7308 [25:41<44:23,  1.93it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 23.24it/s]
+
 15%|█▌        | 6/39 [00:00<00:01, 17.70it/s]
+
 21%|██        | 8/39 [00:00<00:01, 16.69it/s]
+
 26%|██▌       | 10/39 [00:00<00:01, 16.16it/s]
+
 31%|███       | 12/39 [00:00<00:01, 15.80it/s]
+
 36%|███▌      | 14/39 [00:00<00:01, 15.61it/s]
+
 41%|████      | 16/39 [00:00<00:01, 15.45it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 15.35it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 15.28it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 15.26it/s]
+
 62%|██████▏   | 24/39 [00:01<00:00, 15.24it/s]
+
 67%|██████▋   | 26/39 [00:01<00:00, 15.24it/s]
+
 72%|███████▏  | 28/39 [00:01<00:00, 15.22it/s]
+
 77%|███████▋  | 30/39 [00:01<00:00, 15.22it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 15.22it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 15.23it/s]
+
 92%|█████████▏| 36/39 [00:02<00:00, 15.22it/s]
+
 97%|█████████▋| 38/39 [00:02<00:00, 15.23it/s]
                                                   
+
                                               
{'eval_loss': '4.131', 'eval_runtime': '2.624', 'eval_samples_per_second': '930.3', 'eval_steps_per_second': '14.86', 'epoch': '60'}
+
 30%|██▉       | 2160/7308 [25:44<44:23,  1.93it/s]
+
100%|██████████| 39/39 [00:02<00:00, 15.23it/s][eval] epoch 60.0 step 2160 loss 4.1313 (best 4.0798 @ ep 46.0, 7/20 since best)
+
+
                                               
 30%|██▉       | 2161/7308 [25:44<1:57:32,  1.37s/it]
 30%|██▉       | 2162/7308 [25:45<1:38:35,  1.15s/it]
 30%|██▉       | 2163/7308 [25:46<1:25:19,  1.00it/s]
 30%|██▉       | 2164/7308 [25:46<1:16:02,  1.13it/s]
 30%|██▉       | 2165/7308 [25:47<1:09:38,  1.23it/s]
 30%|██▉       | 2166/7308 [25:47<1:05:06,  1.32it/s]
 30%|██▉       | 2167/7308 [25:48<1:01:54,  1.38it/s]
 30%|██▉       | 2168/7308 [25:49<59:41,  1.44it/s]  
 30%|██▉       | 2169/7308 [25:49<58:05,  1.47it/s]
 30%|██▉       | 2170/7308 [25:50<56:59,  1.50it/s]
                                                   
{'loss': '3.945', 'grad_norm': '0.2936', 'learning_rate': '0.0006', 'epoch': '60.28'}
+
 30%|██▉       | 2170/7308 [25:50<56:59,  1.50it/s]
 30%|██▉       | 2171/7308 [25:51<56:14,  1.52it/s]
 30%|██▉       | 2172/7308 [25:51<55:41,  1.54it/s]
 30%|██▉       | 2173/7308 [25:52<55:17,  1.55it/s]
 30%|██▉       | 2174/7308 [25:53<55:01,  1.56it/s]
 30%|██▉       | 2175/7308 [25:53<54:49,  1.56it/s]
 30%|██▉       | 2176/7308 [25:54<54:41,  1.56it/s]
 30%|██▉       | 2177/7308 [25:54<54:36,  1.57it/s]
 30%|██▉       | 2178/7308 [25:55<54:30,  1.57it/s]
 30%|██▉       | 2179/7308 [25:56<54:27,  1.57it/s]
 30%|██▉       | 2180/7308 [25:56<54:23,  1.57it/s]
                                                   
{'loss': '4.013', 'grad_norm': '0.2576', 'learning_rate': '0.0006', 'epoch': '60.57'}
+
 30%|██▉       | 2180/7308 [25:56<54:23,  1.57it/s]
 30%|██▉       | 2181/7308 [25:57<54:21,  1.57it/s]
 30%|██▉       | 2182/7308 [25:58<54:21,  1.57it/s]
 30%|██▉       | 2183/7308 [25:58<54:17,  1.57it/s]
 30%|██▉       | 2184/7308 [25:59<54:15,  1.57it/s]
 30%|██▉       | 2185/7308 [26:00<54:16,  1.57it/s]
 30%|██▉       | 2186/7308 [26:00<54:19,  1.57it/s]
 30%|██▉       | 2187/7308 [26:01<54:20,  1.57it/s]
 30%|██▉       | 2188/7308 [26:01<54:20,  1.57it/s]
 30%|██▉       | 2189/7308 [26:02<54:17,  1.57it/s]
 30%|██▉       | 2190/7308 [26:03<54:13,  1.57it/s]
                                                   
{'loss': '4.109', 'grad_norm': '0.3057', 'learning_rate': '0.0006', 'epoch': '60.85'}
+
 30%|██▉       | 2190/7308 [26:03<54:13,  1.57it/s]
 30%|██▉       | 2191/7308 [26:03<54:12,  1.57it/s]
 30%|██▉       | 2192/7308 [26:04<54:14,  1.57it/s]
 30%|███       | 2193/7308 [26:05<54:12,  1.57it/s]
 30%|███       | 2194/7308 [26:05<54:11,  1.57it/s]
 30%|███       | 2195/7308 [26:06<54:07,  1.57it/s]
 30%|███       | 2196/7308 [26:06<44:05,  1.93it/s]
 30%|███       | 2197/7308 [26:07<49:36,  1.72it/s]
 30%|███       | 2198/7308 [26:08<50:56,  1.67it/s]
 30%|███       | 2199/7308 [26:08<51:50,  1.64it/s]
 30%|███       | 2200/7308 [26:09<52:30,  1.62it/s]
                                                   
{'loss': '4.019', 'grad_norm': '0.3318', 'learning_rate': '0.0006', 'epoch': '61.11'}
+
 30%|███       | 2200/7308 [26:09<52:30,  1.62it/s]
 30%|███       | 2201/7308 [26:09<52:59,  1.61it/s]
 30%|███       | 2202/7308 [26:10<53:18,  1.60it/s]
 30%|███       | 2203/7308 [26:11<53:29,  1.59it/s]
 30%|███       | 2204/7308 [26:11<53:39,  1.59it/s]
 30%|███       | 2205/7308 [26:12<53:44,  1.58it/s]
 30%|███       | 2206/7308 [26:13<53:47,  1.58it/s]
 30%|███       | 2207/7308 [26:13<53:49,  1.58it/s]
 30%|███       | 2208/7308 [26:14<53:51,  1.58it/s]
 30%|███       | 2209/7308 [26:15<53:53,  1.58it/s]
 30%|███       | 2210/7308 [26:16<1:02:51,  1.35it/s]
                                                     
{'loss': '3.991', 'grad_norm': '0.2926', 'learning_rate': '0.0006', 'epoch': '61.4'}
+
 30%|███       | 2210/7308 [26:16<1:02:51,  1.35it/s]
 30%|███       | 2211/7308 [26:16<1:00:10,  1.41it/s]
 30%|███       | 2212/7308 [26:17<58:19,  1.46it/s]  
 30%|███       | 2213/7308 [26:17<56:58,  1.49it/s]
 30%|███       | 2214/7308 [26:18<56:01,  1.52it/s]
 30%|███       | 2215/7308 [26:19<55:21,  1.53it/s]
 30%|███       | 2216/7308 [26:19<54:55,  1.55it/s]
 30%|███       | 2217/7308 [26:20<54:38,  1.55it/s]
 30%|███       | 2218/7308 [26:21<54:26,  1.56it/s]
 30%|███       | 2219/7308 [26:21<54:18,  1.56it/s]
 30%|███       | 2220/7308 [26:22<54:11,  1.56it/s]
                                                   
{'loss': '4.022', 'grad_norm': '0.2667', 'learning_rate': '0.0006', 'epoch': '61.68'}
+
 30%|███       | 2220/7308 [26:22<54:11,  1.56it/s]
 30%|███       | 2221/7308 [26:22<54:05,  1.57it/s]
 30%|███       | 2222/7308 [26:23<54:05,  1.57it/s]
 30%|███       | 2223/7308 [26:24<54:00,  1.57it/s]
 30%|███       | 2224/7308 [26:24<53:57,  1.57it/s]
 30%|███       | 2225/7308 [26:25<53:56,  1.57it/s]
 30%|███       | 2226/7308 [26:26<53:54,  1.57it/s]
 30%|███       | 2227/7308 [26:26<53:52,  1.57it/s]
 30%|███       | 2228/7308 [26:27<53:51,  1.57it/s]
 31%|███       | 2229/7308 [26:28<53:50,  1.57it/s]
 31%|███       | 2230/7308 [26:28<53:50,  1.57it/s]
                                                   
{'loss': '4.072', 'grad_norm': '0.2423', 'learning_rate': '0.0006', 'epoch': '61.96'}
+
 31%|███       | 2230/7308 [26:28<53:50,  1.57it/s]
 31%|███       | 2231/7308 [26:29<53:46,  1.57it/s]
 31%|███       | 2232/7308 [26:29<43:49,  1.93it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 23.18it/s]
+
 15%|█▌        | 6/39 [00:00<00:01, 17.70it/s]
+
 21%|██        | 8/39 [00:00<00:01, 16.72it/s]
+
 26%|██▌       | 10/39 [00:00<00:01, 16.15it/s]
+
 31%|███       | 12/39 [00:00<00:01, 15.83it/s]
+
 36%|███▌      | 14/39 [00:00<00:01, 15.61it/s]
+
 41%|████      | 16/39 [00:00<00:01, 15.48it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 15.38it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 15.33it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 15.28it/s]
+
 62%|██████▏   | 24/39 [00:01<00:00, 15.28it/s]
+
 67%|██████▋   | 26/39 [00:01<00:00, 15.26it/s]
+
 72%|███████▏  | 28/39 [00:01<00:00, 15.25it/s]
+
 77%|███████▋  | 30/39 [00:01<00:00, 15.23it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 15.24it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 15.21it/s]
+
 92%|█████████▏| 36/39 [00:02<00:00, 15.22it/s]
+
 97%|█████████▋| 38/39 [00:02<00:00, 15.20it/s]
                                                   
+
                                               
{'eval_loss': '4.144', 'eval_runtime': '2.621', 'eval_samples_per_second': '931.3', 'eval_steps_per_second': '14.88', 'epoch': '62'}
+
 31%|███       | 2232/7308 [26:32<43:49,  1.93it/s]
+
100%|██████████| 39/39 [00:02<00:00, 15.20it/s][eval] epoch 62.0 step 2232 loss 4.1438 (best 4.0798 @ ep 46.0, 8/20 since best)
+
+
                                               
 31%|███       | 2233/7308 [26:32<1:55:43,  1.37s/it]
 31%|███       | 2234/7308 [26:33<1:37:08,  1.15s/it]
 31%|███       | 2235/7308 [26:34<1:24:07,  1.01it/s]
 31%|███       | 2236/7308 [26:34<1:14:59,  1.13it/s]
 31%|███       | 2237/7308 [26:35<1:08:37,  1.23it/s]
 31%|███       | 2238/7308 [26:36<1:04:09,  1.32it/s]
 31%|███       | 2239/7308 [26:36<1:01:00,  1.38it/s]
 31%|███       | 2240/7308 [26:37<58:49,  1.44it/s]  
                                                   
{'loss': '4.066', 'grad_norm': '0.3651', 'learning_rate': '0.0006', 'epoch': '62.23'}
+
 31%|███       | 2240/7308 [26:37<58:49,  1.44it/s]
 31%|███       | 2241/7308 [26:38<57:17,  1.47it/s]
 31%|███       | 2242/7308 [26:38<56:12,  1.50it/s]
 31%|███       | 2243/7308 [26:39<55:26,  1.52it/s]
 31%|███       | 2244/7308 [26:39<54:53,  1.54it/s]
 31%|███       | 2245/7308 [26:40<54:30,  1.55it/s]
 31%|███       | 2246/7308 [26:41<54:14,  1.56it/s]
 31%|███       | 2247/7308 [26:41<54:03,  1.56it/s]
 31%|███       | 2248/7308 [26:42<53:58,  1.56it/s]
 31%|███       | 2249/7308 [26:43<53:51,  1.57it/s]
 31%|███       | 2250/7308 [26:43<53:46,  1.57it/s]
                                                   
{'loss': '3.898', 'grad_norm': '0.2886', 'learning_rate': '0.0006', 'epoch': '62.51'}
+
 31%|███       | 2250/7308 [26:43<53:46,  1.57it/s]
 31%|███       | 2251/7308 [26:44<53:44,  1.57it/s]
 31%|███       | 2252/7308 [26:45<53:40,  1.57it/s]
 31%|███       | 2253/7308 [26:45<53:39,  1.57it/s]
 31%|███       | 2254/7308 [26:46<53:37,  1.57it/s]
 31%|███       | 2255/7308 [26:46<53:38,  1.57it/s]
 31%|███       | 2256/7308 [26:47<53:37,  1.57it/s]
 31%|███       | 2257/7308 [26:48<53:35,  1.57it/s]
 31%|███       | 2258/7308 [26:48<53:35,  1.57it/s]
 31%|███       | 2259/7308 [26:49<53:34,  1.57it/s]
 31%|███       | 2260/7308 [26:50<53:33,  1.57it/s]
                                                   
{'loss': '3.966', 'grad_norm': '0.3169', 'learning_rate': '0.0006', 'epoch': '62.79'}
+
 31%|███       | 2260/7308 [26:50<53:33,  1.57it/s]
 31%|███       | 2261/7308 [26:50<53:31,  1.57it/s]
 31%|███       | 2262/7308 [26:51<53:29,  1.57it/s]
 31%|███       | 2263/7308 [26:52<53:29,  1.57it/s]
 31%|███       | 2264/7308 [26:52<53:27,  1.57it/s]
 31%|███       | 2265/7308 [26:53<53:26,  1.57it/s]
 31%|███       | 2266/7308 [26:53<53:24,  1.57it/s]
 31%|███       | 2267/7308 [26:54<53:20,  1.58it/s]
 31%|███       | 2268/7308 [26:54<43:28,  1.93it/s]
 31%|███       | 2269/7308 [26:55<48:51,  1.72it/s]
 31%|███       | 2270/7308 [26:56<50:11,  1.67it/s]
                                                   
{'loss': '4.063', 'grad_norm': '0.2929', 'learning_rate': '0.0006', 'epoch': '63.06'}
+
 31%|███       | 2270/7308 [26:56<50:11,  1.67it/s]
 31%|███       | 2271/7308 [26:56<51:08,  1.64it/s]
 31%|███       | 2272/7308 [26:57<51:47,  1.62it/s]
 31%|███       | 2273/7308 [26:58<52:14,  1.61it/s]
 31%|███       | 2274/7308 [26:58<52:32,  1.60it/s]
 31%|███       | 2275/7308 [26:59<52:44,  1.59it/s]
 31%|███       | 2276/7308 [27:00<52:54,  1.59it/s]
 31%|███       | 2277/7308 [27:00<52:59,  1.58it/s]
 31%|███       | 2278/7308 [27:01<53:03,  1.58it/s]
 31%|███       | 2279/7308 [27:01<53:07,  1.58it/s]
 31%|███       | 2280/7308 [27:02<53:07,  1.58it/s]
                                                   
{'loss': '4.068', 'grad_norm': '0.3391', 'learning_rate': '0.0006', 'epoch': '63.34'}
+
 31%|███       | 2280/7308 [27:02<53:07,  1.58it/s]
 31%|███       | 2281/7308 [27:03<53:10,  1.58it/s]
 31%|███       | 2282/7308 [27:03<53:09,  1.58it/s]
 31%|███       | 2283/7308 [27:04<53:08,  1.58it/s]
 31%|███▏      | 2284/7308 [27:05<53:08,  1.58it/s]
 31%|███▏      | 2285/7308 [27:05<53:08,  1.58it/s]
 31%|███▏      | 2286/7308 [27:06<53:07,  1.58it/s]
 31%|███▏      | 2287/7308 [27:06<53:09,  1.57it/s]
 31%|███▏      | 2288/7308 [27:07<53:09,  1.57it/s]
 31%|███▏      | 2289/7308 [27:08<53:07,  1.57it/s]
 31%|███▏      | 2290/7308 [27:08<53:06,  1.57it/s]
                                                   
{'loss': '4.03', 'grad_norm': '0.2859', 'learning_rate': '0.0006', 'epoch': '63.62'}
+
 31%|███▏      | 2290/7308 [27:08<53:06,  1.57it/s]
 31%|███▏      | 2291/7308 [27:09<53:06,  1.57it/s]
 31%|███▏      | 2292/7308 [27:10<53:05,  1.57it/s]
 31%|███▏      | 2293/7308 [27:10<53:03,  1.58it/s]
 31%|███▏      | 2294/7308 [27:11<53:03,  1.58it/s]
 31%|███▏      | 2295/7308 [27:12<53:01,  1.58it/s]
 31%|███▏      | 2296/7308 [27:12<53:01,  1.58it/s]
 31%|███▏      | 2297/7308 [27:13<53:01,  1.57it/s]
 31%|███▏      | 2298/7308 [27:13<53:00,  1.58it/s]
 31%|███▏      | 2299/7308 [27:14<53:00,  1.57it/s]
 31%|███▏      | 2300/7308 [27:15<52:59,  1.58it/s]
                                                   
{'loss': '3.946', 'grad_norm': '0.2611', 'learning_rate': '0.0006', 'epoch': '63.9'}
+
 31%|███▏      | 2300/7308 [27:15<52:59,  1.58it/s]
 31%|███▏      | 2301/7308 [27:15<53:00,  1.57it/s]
 31%|███▏      | 2302/7308 [27:16<52:58,  1.57it/s]
 32%|███▏      | 2303/7308 [27:17<52:57,  1.58it/s]
 32%|███▏      | 2304/7308 [27:17<43:09,  1.93it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 22.77it/s]
+
 15%|█▌        | 6/39 [00:00<00:01, 17.60it/s]
+
 21%|██        | 8/39 [00:00<00:01, 16.61it/s]
+
 26%|██▌       | 10/39 [00:00<00:01, 16.11it/s]
+
 31%|███       | 12/39 [00:00<00:01, 15.76it/s]
+
 36%|███▌      | 14/39 [00:00<00:01, 15.58it/s]
+
 41%|████      | 16/39 [00:00<00:01, 15.43it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 15.33it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 15.28it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 15.26it/s]
+
 62%|██████▏   | 24/39 [00:01<00:00, 15.24it/s]
+
 67%|██████▋   | 26/39 [00:01<00:00, 15.23it/s]
+
 72%|███████▏  | 28/39 [00:01<00:00, 15.23it/s]
+
 77%|███████▋  | 30/39 [00:01<00:00, 15.22it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 15.22it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 15.22it/s]
+
 92%|█████████▏| 36/39 [00:02<00:00, 15.22it/s]
+
 97%|█████████▋| 38/39 [00:02<00:00, 15.22it/s]
                                                   
+
                                               
{'eval_loss': '4.149', 'eval_runtime': '2.639', 'eval_samples_per_second': '924.9', 'eval_steps_per_second': '14.78', 'epoch': '64'}
+
 32%|███▏      | 2304/7308 [27:20<43:09,  1.93it/s]
+
100%|██████████| 39/39 [00:02<00:00, 15.22it/s][eval] epoch 64.0 step 2304 loss 4.1494 (best 4.0798 @ ep 46.0, 9/20 since best)
+
+
                                               
 32%|███▏      | 2305/7308 [27:20<1:54:32,  1.37s/it]
 32%|███▏      | 2306/7308 [27:21<1:36:01,  1.15s/it]
 32%|███▏      | 2307/7308 [27:22<1:23:04,  1.00it/s]
 32%|███▏      | 2308/7308 [27:22<1:14:00,  1.13it/s]
 32%|███▏      | 2309/7308 [27:23<1:07:39,  1.23it/s]
 32%|███▏      | 2310/7308 [27:23<1:03:13,  1.32it/s]
                                                     
{'loss': '3.943', 'grad_norm': '0.3498', 'learning_rate': '0.0006', 'epoch': '64.17'}
+
 32%|███▏      | 2310/7308 [27:23<1:03:13,  1.32it/s]
 32%|███▏      | 2311/7308 [27:24<1:00:07,  1.39it/s]
 32%|███▏      | 2312/7308 [27:25<57:55,  1.44it/s]  
 32%|███▏      | 2313/7308 [27:25<56:24,  1.48it/s]
 32%|███▏      | 2314/7308 [27:26<55:18,  1.50it/s]
 32%|███▏      | 2315/7308 [27:27<54:33,  1.53it/s]
 32%|███▏      | 2316/7308 [27:27<54:01,  1.54it/s]
 32%|███▏      | 2317/7308 [27:28<53:39,  1.55it/s]
 32%|███▏      | 2318/7308 [27:29<53:23,  1.56it/s]
 32%|███▏      | 2319/7308 [27:29<53:11,  1.56it/s]
 32%|███▏      | 2320/7308 [27:30<53:03,  1.57it/s]
                                                   
{'loss': '3.966', 'grad_norm': '0.3261', 'learning_rate': '0.0006', 'epoch': '64.45'}
+
 32%|███▏      | 2320/7308 [27:30<53:03,  1.57it/s]
 32%|███▏      | 2321/7308 [27:30<53:00,  1.57it/s]
 32%|███▏      | 2322/7308 [27:31<52:55,  1.57it/s]
 32%|███▏      | 2323/7308 [27:32<52:51,  1.57it/s]
 32%|███▏      | 2324/7308 [27:32<52:48,  1.57it/s]
 32%|███▏      | 2325/7308 [27:33<52:47,  1.57it/s]
 32%|███▏      | 2326/7308 [27:34<52:46,  1.57it/s]
 32%|███▏      | 2327/7308 [27:34<52:44,  1.57it/s]
 32%|███▏      | 2328/7308 [27:35<52:42,  1.57it/s]
 32%|███▏      | 2329/7308 [27:35<52:41,  1.57it/s]
 32%|███▏      | 2330/7308 [27:36<52:40,  1.58it/s]
                                                   
{'loss': '3.978', 'grad_norm': '0.273', 'learning_rate': '0.0006', 'epoch': '64.73'}
+
 32%|███▏      | 2330/7308 [27:36<52:40,  1.58it/s]
 32%|███▏      | 2331/7308 [27:37<52:40,  1.57it/s]
 32%|███▏      | 2332/7308 [27:37<52:40,  1.57it/s]
 32%|███▏      | 2333/7308 [27:38<1:01:10,  1.36it/s]
 32%|███▏      | 2334/7308 [27:39<58:34,  1.42it/s]  
 32%|███▏      | 2335/7308 [27:40<56:45,  1.46it/s]
 32%|███▏      | 2336/7308 [27:40<55:30,  1.49it/s]
 32%|███▏      | 2337/7308 [27:41<54:37,  1.52it/s]
 32%|███▏      | 2338/7308 [27:42<53:59,  1.53it/s]
 32%|███▏      | 2339/7308 [27:42<53:31,  1.55it/s]
 32%|███▏      | 2340/7308 [27:42<43:30,  1.90it/s]
                                                   
{'loss': '4.033', 'grad_norm': '0.3528', 'learning_rate': '0.0006', 'epoch': '65'}
+
 32%|███▏      | 2340/7308 [27:42<43:30,  1.90it/s]
 32%|███▏      | 2341/7308 [27:43<48:38,  1.70it/s]
 32%|███▏      | 2342/7308 [27:44<49:49,  1.66it/s]
 32%|███▏      | 2343/7308 [27:44<50:36,  1.63it/s]
 32%|███▏      | 2344/7308 [27:45<51:10,  1.62it/s]
 32%|███▏      | 2345/7308 [27:46<51:34,  1.60it/s]
 32%|███▏      | 2346/7308 [27:46<51:50,  1.60it/s]
 32%|███▏      | 2347/7308 [27:47<52:02,  1.59it/s]
 32%|███▏      | 2348/7308 [27:48<52:11,  1.58it/s]
 32%|███▏      | 2349/7308 [27:48<52:17,  1.58it/s]
 32%|███▏      | 2350/7308 [27:49<52:19,  1.58it/s]
                                                   
{'loss': '3.936', 'grad_norm': '0.3049', 'learning_rate': '0.0006', 'epoch': '65.28'}
+
 32%|███▏      | 2350/7308 [27:49<52:19,  1.58it/s]
 32%|███▏      | 2351/7308 [27:50<52:21,  1.58it/s]
 32%|███▏      | 2352/7308 [27:50<52:22,  1.58it/s]
 32%|███▏      | 2353/7308 [27:51<52:24,  1.58it/s]
 32%|███▏      | 2354/7308 [27:51<52:24,  1.58it/s]
 32%|███▏      | 2355/7308 [27:52<1:01:12,  1.35it/s]
 32%|███▏      | 2356/7308 [27:53<58:32,  1.41it/s]  
 32%|███▏      | 2357/7308 [27:54<56:41,  1.46it/s]
 32%|███▏      | 2358/7308 [27:54<55:22,  1.49it/s]
 32%|███▏      | 2359/7308 [27:55<54:27,  1.51it/s]
 32%|███▏      | 2360/7308 [27:56<53:49,  1.53it/s]
                                                   
{'loss': '4.019', 'grad_norm': '0.3349', 'learning_rate': '0.0006', 'epoch': '65.57'}
+
 32%|███▏      | 2360/7308 [27:56<53:49,  1.53it/s]
 32%|███▏      | 2361/7308 [27:56<53:22,  1.54it/s]
 32%|███▏      | 2362/7308 [27:57<53:04,  1.55it/s]
 32%|███▏      | 2363/7308 [27:57<52:51,  1.56it/s]
 32%|███▏      | 2364/7308 [27:58<52:41,  1.56it/s]
 32%|███▏      | 2365/7308 [27:59<52:36,  1.57it/s]
 32%|███▏      | 2366/7308 [27:59<52:29,  1.57it/s]
 32%|███▏      | 2367/7308 [28:00<52:24,  1.57it/s]
 32%|███▏      | 2368/7308 [28:01<52:20,  1.57it/s]
 32%|███▏      | 2369/7308 [28:01<52:18,  1.57it/s]
 32%|███▏      | 2370/7308 [28:02<52:16,  1.57it/s]
                                                   
{'loss': '3.974', 'grad_norm': '0.278', 'learning_rate': '0.0006', 'epoch': '65.85'}
+
 32%|███▏      | 2370/7308 [28:02<52:16,  1.57it/s]
 32%|███▏      | 2371/7308 [28:03<52:16,  1.57it/s]
 32%|███▏      | 2372/7308 [28:03<52:15,  1.57it/s]
 32%|███▏      | 2373/7308 [28:04<52:14,  1.57it/s]
 32%|███▏      | 2374/7308 [28:04<52:13,  1.57it/s]
 32%|███▏      | 2375/7308 [28:05<52:10,  1.58it/s]
 33%|███▎      | 2376/7308 [28:05<42:30,  1.93it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 23.11it/s]
+
 15%|█▌        | 6/39 [00:00<00:01, 17.69it/s]
+
 21%|██        | 8/39 [00:00<00:01, 16.70it/s]
+
 26%|██▌       | 10/39 [00:00<00:01, 16.16it/s]
+
 31%|███       | 12/39 [00:00<00:01, 15.84it/s]
+
 36%|███▌      | 14/39 [00:00<00:01, 15.63it/s]
+
 41%|████      | 16/39 [00:00<00:01, 15.50it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 15.40it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 15.33it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 15.28it/s]
+
 62%|██████▏   | 24/39 [00:01<00:00, 15.26it/s]
+
 67%|██████▋   | 26/39 [00:01<00:00, 15.28it/s]
+
 72%|███████▏  | 28/39 [00:01<00:00, 15.24it/s]
+
 77%|███████▋  | 30/39 [00:01<00:00, 15.24it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 15.23it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 15.22it/s]
+
 92%|█████████▏| 36/39 [00:02<00:00, 15.21it/s]
+
 97%|█████████▋| 38/39 [00:02<00:00, 15.21it/s]
                                                   
+
                                               
{'eval_loss': '4.161', 'eval_runtime': '2.621', 'eval_samples_per_second': '931.4', 'eval_steps_per_second': '14.88', 'epoch': '66'}
+
 33%|███▎      | 2376/7308 [28:08<42:30,  1.93it/s]
+
100%|██████████| 39/39 [00:02<00:00, 15.21it/s][eval] epoch 66.0 step 2376 loss 4.1611 (best 4.0798 @ ep 46.0, 10/20 since best)
+
+
                                               
 33%|███▎      | 2377/7308 [28:09<1:52:22,  1.37s/it]
 33%|███▎      | 2378/7308 [28:09<1:34:17,  1.15s/it]
 33%|███▎      | 2379/7308 [28:10<1:21:38,  1.01it/s]
 33%|███▎      | 2380/7308 [28:11<1:12:46,  1.13it/s]
                                                     
{'loss': '4.036', 'grad_norm': '0.4387', 'learning_rate': '0.0006', 'epoch': '66.11'}
+
 33%|███▎      | 2380/7308 [28:11<1:12:46,  1.13it/s]
 33%|███▎      | 2381/7308 [28:11<1:06:34,  1.23it/s]
 33%|███▎      | 2382/7308 [28:12<1:02:14,  1.32it/s]
 33%|███▎      | 2383/7308 [28:13<1:07:36,  1.21it/s]
 33%|███▎      | 2384/7308 [28:13<1:02:57,  1.30it/s]
 33%|███▎      | 2385/7308 [28:14<59:40,  1.37it/s]  
 33%|███▎      | 2386/7308 [28:15<57:24,  1.43it/s]
 33%|███▎      | 2387/7308 [28:15<55:47,  1.47it/s]
 33%|███▎      | 2388/7308 [28:16<54:38,  1.50it/s]
 33%|███▎      | 2389/7308 [28:17<53:50,  1.52it/s]
 33%|███▎      | 2390/7308 [28:17<53:16,  1.54it/s]
                                                   
{'loss': '4.012', 'grad_norm': '0.3254', 'learning_rate': '0.0006', 'epoch': '66.4'}
+
 33%|███▎      | 2390/7308 [28:17<53:16,  1.54it/s]
 33%|███▎      | 2391/7308 [28:18<52:54,  1.55it/s]
 33%|███▎      | 2392/7308 [28:19<52:38,  1.56it/s]
 33%|███▎      | 2393/7308 [28:19<52:26,  1.56it/s]
 33%|███▎      | 2394/7308 [28:20<52:16,  1.57it/s]
 33%|███▎      | 2395/7308 [28:20<52:10,  1.57it/s]
 33%|███▎      | 2396/7308 [28:21<52:05,  1.57it/s]
 33%|███▎      | 2397/7308 [28:22<52:03,  1.57it/s]
 33%|███▎      | 2398/7308 [28:22<52:00,  1.57it/s]
 33%|███▎      | 2399/7308 [28:23<51:58,  1.57it/s]
 33%|███▎      | 2400/7308 [28:24<51:57,  1.57it/s]
                                                   
{'loss': '3.926', 'grad_norm': '0.2673', 'learning_rate': '0.0006', 'epoch': '66.68'}
+
 33%|███▎      | 2400/7308 [28:24<51:57,  1.57it/s]
 33%|███▎      | 2401/7308 [28:24<51:56,  1.57it/s]
 33%|███▎      | 2402/7308 [28:25<51:55,  1.57it/s]
 33%|███▎      | 2403/7308 [28:26<51:53,  1.58it/s]
 33%|███▎      | 2404/7308 [28:26<51:53,  1.58it/s]
 33%|███▎      | 2405/7308 [28:27<51:51,  1.58it/s]
 33%|███▎      | 2406/7308 [28:27<51:50,  1.58it/s]
 33%|███▎      | 2407/7308 [28:28<51:49,  1.58it/s]
 33%|███▎      | 2408/7308 [28:29<51:51,  1.57it/s]
 33%|███▎      | 2409/7308 [28:29<51:50,  1.58it/s]
 33%|███▎      | 2410/7308 [28:30<51:49,  1.58it/s]
                                                   
{'loss': '3.965', 'grad_norm': '0.3218', 'learning_rate': '0.0006', 'epoch': '66.96'}
+
 33%|███▎      | 2410/7308 [28:30<51:49,  1.58it/s]
 33%|███▎      | 2411/7308 [28:31<1:00:11,  1.36it/s]
 33%|███▎      | 2412/7308 [28:31<48:04,  1.70it/s]  
 33%|███▎      | 2413/7308 [28:32<51:33,  1.58it/s]
 33%|███▎      | 2414/7308 [28:33<51:36,  1.58it/s]
 33%|███▎      | 2415/7308 [28:33<51:39,  1.58it/s]
 33%|███▎      | 2416/7308 [28:34<51:39,  1.58it/s]
 33%|███▎      | 2417/7308 [28:34<51:40,  1.58it/s]
 33%|███▎      | 2418/7308 [28:35<51:41,  1.58it/s]
 33%|███▎      | 2419/7308 [28:36<51:40,  1.58it/s]
 33%|███▎      | 2420/7308 [28:36<51:42,  1.58it/s]
                                                   
{'loss': '3.921', 'grad_norm': '0.3208', 'learning_rate': '0.0006', 'epoch': '67.23'}
+
 33%|███▎      | 2420/7308 [28:36<51:42,  1.58it/s]
 33%|███▎      | 2421/7308 [28:37<51:43,  1.57it/s]
 33%|███▎      | 2422/7308 [28:38<51:41,  1.58it/s]
 33%|███▎      | 2423/7308 [28:38<51:39,  1.58it/s]
 33%|███▎      | 2424/7308 [28:39<51:39,  1.58it/s]
 33%|███▎      | 2425/7308 [28:40<51:40,  1.57it/s]
 33%|███▎      | 2426/7308 [28:40<51:38,  1.58it/s]
 33%|███▎      | 2427/7308 [28:41<51:38,  1.58it/s]
 33%|███▎      | 2428/7308 [28:41<51:37,  1.58it/s]
 33%|███▎      | 2429/7308 [28:42<51:37,  1.58it/s]
 33%|███▎      | 2430/7308 [28:43<51:35,  1.58it/s]
                                                   
{'loss': '3.934', 'grad_norm': '0.3138', 'learning_rate': '0.0006', 'epoch': '67.51'}
+
 33%|███▎      | 2430/7308 [28:43<51:35,  1.58it/s]
 33%|███▎      | 2431/7308 [28:43<51:36,  1.57it/s]
 33%|███▎      | 2432/7308 [28:44<51:35,  1.58it/s]
 33%|███▎      | 2433/7308 [28:45<51:34,  1.58it/s]
 33%|███▎      | 2434/7308 [28:45<51:33,  1.58it/s]
 33%|███▎      | 2435/7308 [28:46<51:33,  1.58it/s]
 33%|███▎      | 2436/7308 [28:47<51:31,  1.58it/s]
 33%|███▎      | 2437/7308 [28:47<51:32,  1.58it/s]
 33%|███▎      | 2438/7308 [28:48<51:31,  1.58it/s]
 33%|███▎      | 2439/7308 [28:48<51:31,  1.58it/s]
 33%|███▎      | 2440/7308 [28:49<51:33,  1.57it/s]
                                                   
{'loss': '3.999', 'grad_norm': '0.262', 'learning_rate': '0.0006', 'epoch': '67.79'}
+
 33%|███▎      | 2440/7308 [28:49<51:33,  1.57it/s]
 33%|███▎      | 2441/7308 [28:50<51:32,  1.57it/s]
 33%|███▎      | 2442/7308 [28:50<51:30,  1.57it/s]
 33%|███▎      | 2443/7308 [28:51<51:28,  1.58it/s]
 33%|███▎      | 2444/7308 [28:52<59:54,  1.35it/s]
 33%|███▎      | 2445/7308 [28:53<57:20,  1.41it/s]
 33%|███▎      | 2446/7308 [28:53<55:33,  1.46it/s]
 33%|███▎      | 2447/7308 [28:54<54:16,  1.49it/s]
 33%|███▎      | 2448/7308 [28:54<43:52,  1.85it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 23.01it/s]
+
 15%|█▌        | 6/39 [00:00<00:01, 17.65it/s]
+
 21%|██        | 8/39 [00:00<00:01, 16.63it/s]
+
 26%|██▌       | 10/39 [00:00<00:01, 16.13it/s]
+
 31%|███       | 12/39 [00:00<00:01, 15.77it/s]
+
 36%|███▌      | 14/39 [00:00<00:01, 15.58it/s]
+
 41%|████      | 16/39 [00:00<00:01, 15.44it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 15.37it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 15.30it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 15.27it/s]
+
 62%|██████▏   | 24/39 [00:01<00:00, 15.25it/s]
+
 67%|██████▋   | 26/39 [00:01<00:00, 15.24it/s]
+
 72%|███████▏  | 28/39 [00:01<00:00, 15.24it/s]
+
 77%|███████▋  | 30/39 [00:01<00:00, 15.23it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 15.23it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 15.23it/s]
+
 92%|█████████▏| 36/39 [00:02<00:00, 15.21it/s]
+
 97%|█████████▋| 38/39 [00:02<00:00, 15.22it/s]
                                                   
+
                                               
{'eval_loss': '4.179', 'eval_runtime': '2.625', 'eval_samples_per_second': '929.8', 'eval_steps_per_second': '14.86', 'epoch': '68'}
+
 33%|███▎      | 2448/7308 [28:57<43:52,  1.85it/s]
+
100%|██████████| 39/39 [00:02<00:00, 15.22it/s][eval] epoch 68.0 step 2448 loss 4.1789 (best 4.0798 @ ep 46.0, 11/20 since best)
+
+
                                               
 34%|███▎      | 2449/7308 [28:57<1:51:30,  1.38s/it]
 34%|███▎      | 2450/7308 [28:58<1:41:45,  1.26s/it]
                                                     
{'loss': '3.945', 'grad_norm': '0.3138', 'learning_rate': '0.0006', 'epoch': '68.06'}
+
 34%|███▎      | 2450/7308 [28:58<1:41:45,  1.26s/it]
 34%|███▎      | 2451/7308 [28:59<1:26:38,  1.07s/it]
 34%|███▎      | 2452/7308 [29:00<1:16:01,  1.06it/s]
 34%|███▎      | 2453/7308 [29:00<1:08:37,  1.18it/s]
 34%|███▎      | 2454/7308 [29:01<1:03:25,  1.28it/s]
 34%|███▎      | 2455/7308 [29:02<59:47,  1.35it/s]  
 34%|███▎      | 2456/7308 [29:03<1:05:36,  1.23it/s]
 34%|███▎      | 2457/7308 [29:03<1:01:18,  1.32it/s]
 34%|███▎      | 2458/7308 [29:04<58:16,  1.39it/s]  
 34%|███▎      | 2459/7308 [29:04<56:09,  1.44it/s]
 34%|███▎      | 2460/7308 [29:05<54:41,  1.48it/s]
                                                   
{'loss': '3.943', 'grad_norm': '0.3047', 'learning_rate': '0.0006', 'epoch': '68.34'}
+
 34%|███▎      | 2460/7308 [29:05<54:41,  1.48it/s]
 34%|███▎      | 2461/7308 [29:06<53:44,  1.50it/s]
 34%|███▎      | 2462/7308 [29:06<53:00,  1.52it/s]
 34%|███▎      | 2463/7308 [29:07<52:27,  1.54it/s]
 34%|███▎      | 2464/7308 [29:08<52:04,  1.55it/s]
 34%|███▎      | 2465/7308 [29:08<51:52,  1.56it/s]
 34%|███▎      | 2466/7308 [29:09<51:40,  1.56it/s]
 34%|███▍      | 2467/7308 [29:10<51:30,  1.57it/s]
 34%|███▍      | 2468/7308 [29:10<51:24,  1.57it/s]
 34%|███▍      | 2469/7308 [29:11<51:20,  1.57it/s]
 34%|███▍      | 2470/7308 [29:11<51:17,  1.57it/s]
                                                   
{'loss': '3.92', 'grad_norm': '0.2843', 'learning_rate': '0.0006', 'epoch': '68.62'}
+
 34%|███▍      | 2470/7308 [29:11<51:17,  1.57it/s]
 34%|███▍      | 2471/7308 [29:12<51:14,  1.57it/s]
 34%|███▍      | 2472/7308 [29:13<51:13,  1.57it/s]
 34%|███▍      | 2473/7308 [29:13<51:11,  1.57it/s]
 34%|███▍      | 2474/7308 [29:14<51:09,  1.57it/s]
 34%|███▍      | 2475/7308 [29:15<51:07,  1.58it/s]
 34%|███▍      | 2476/7308 [29:15<51:07,  1.58it/s]
 34%|███▍      | 2477/7308 [29:16<51:09,  1.57it/s]
 34%|███▍      | 2478/7308 [29:17<51:07,  1.57it/s]
 34%|███▍      | 2479/7308 [29:17<51:04,  1.58it/s]
 34%|███▍      | 2480/7308 [29:18<51:03,  1.58it/s]
                                                   
{'loss': '3.984', 'grad_norm': '0.2883', 'learning_rate': '0.0006', 'epoch': '68.9'}
+
 34%|███▍      | 2480/7308 [29:18<51:03,  1.58it/s]
 34%|███▍      | 2481/7308 [29:18<51:04,  1.58it/s]
 34%|███▍      | 2482/7308 [29:19<51:03,  1.58it/s]
 34%|███▍      | 2483/7308 [29:20<51:02,  1.58it/s]
 34%|███▍      | 2484/7308 [29:20<41:34,  1.93it/s]
 34%|███▍      | 2485/7308 [29:21<46:12,  1.74it/s]
 34%|███▍      | 2486/7308 [29:21<47:38,  1.69it/s]
 34%|███▍      | 2487/7308 [29:22<48:38,  1.65it/s]
 34%|███▍      | 2488/7308 [29:23<49:19,  1.63it/s]
 34%|███▍      | 2489/7308 [29:23<49:48,  1.61it/s]
 34%|███▍      | 2490/7308 [29:24<50:07,  1.60it/s]
                                                   
{'loss': '3.851', 'grad_norm': '0.4276', 'learning_rate': '0.0006', 'epoch': '69.17'}
+
 34%|███▍      | 2490/7308 [29:24<50:07,  1.60it/s]
 34%|███▍      | 2491/7308 [29:24<50:22,  1.59it/s]
 34%|███▍      | 2492/7308 [29:25<50:31,  1.59it/s]
 34%|███▍      | 2493/7308 [29:26<50:38,  1.58it/s]
 34%|███▍      | 2494/7308 [29:26<50:42,  1.58it/s]
 34%|███▍      | 2495/7308 [29:27<50:45,  1.58it/s]
 34%|███▍      | 2496/7308 [29:28<50:48,  1.58it/s]
 34%|███▍      | 2497/7308 [29:28<50:48,  1.58it/s]
 34%|███▍      | 2498/7308 [29:29<50:48,  1.58it/s]
 34%|███▍      | 2499/7308 [29:30<50:48,  1.58it/s]
 34%|███▍      | 2500/7308 [29:30<50:48,  1.58it/s]
                                                   
{'loss': '3.893', 'grad_norm': '0.3209', 'learning_rate': '0.0006', 'epoch': '69.45'}
+
 34%|███▍      | 2500/7308 [29:30<50:48,  1.58it/s]
 34%|███▍      | 2501/7308 [29:31<50:52,  1.57it/s]
 34%|███▍      | 2502/7308 [29:31<50:50,  1.58it/s]
 34%|███▍      | 2503/7308 [29:32<50:48,  1.58it/s]
 34%|███▍      | 2504/7308 [29:33<50:47,  1.58it/s]
 34%|███▍      | 2505/7308 [29:33<50:47,  1.58it/s]
 34%|███▍      | 2506/7308 [29:34<50:45,  1.58it/s]
 34%|███▍      | 2507/7308 [29:35<50:45,  1.58it/s]
 34%|███▍      | 2508/7308 [29:35<50:44,  1.58it/s]
 34%|███▍      | 2509/7308 [29:36<50:45,  1.58it/s]
 34%|███▍      | 2510/7308 [29:37<50:43,  1.58it/s]
                                                   
{'loss': '3.91', 'grad_norm': '0.2753', 'learning_rate': '0.0006', 'epoch': '69.73'}
+
 34%|███▍      | 2510/7308 [29:37<50:43,  1.58it/s]
 34%|███▍      | 2511/7308 [29:37<50:45,  1.58it/s]
 34%|███▍      | 2512/7308 [29:38<50:44,  1.58it/s]
 34%|███▍      | 2513/7308 [29:38<50:44,  1.58it/s]
 34%|███▍      | 2514/7308 [29:39<50:43,  1.58it/s]
 34%|███▍      | 2515/7308 [29:40<50:43,  1.57it/s]
 34%|███▍      | 2516/7308 [29:40<50:42,  1.58it/s]
 34%|███▍      | 2517/7308 [29:41<50:43,  1.57it/s]
 34%|███▍      | 2518/7308 [29:42<50:41,  1.57it/s]
 34%|███▍      | 2519/7308 [29:42<50:39,  1.58it/s]
 34%|███▍      | 2520/7308 [29:42<41:16,  1.93it/s]
                                                   
{'loss': '3.988', 'grad_norm': '0.3538', 'learning_rate': '0.0006', 'epoch': '70'}
+
 34%|███▍      | 2520/7308 [29:42<41:16,  1.93it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 23.72it/s]
+
 15%|█▌        | 6/39 [00:00<00:01, 17.82it/s]
+
 21%|██        | 8/39 [00:00<00:01, 16.78it/s]
+
 26%|██▌       | 10/39 [00:00<00:01, 16.22it/s]
+
 31%|███       | 12/39 [00:00<00:01, 15.86it/s]
+
 36%|███▌      | 14/39 [00:00<00:01, 15.64it/s]
+
 41%|████      | 16/39 [00:00<00:01, 15.49it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 15.42it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 15.34it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 15.31it/s]
+
 62%|██████▏   | 24/39 [00:01<00:00, 15.30it/s]
+
 67%|██████▋   | 26/39 [00:01<00:00, 15.28it/s]
+
 72%|███████▏  | 28/39 [00:01<00:00, 15.25it/s]
+
 77%|███████▋  | 30/39 [00:01<00:00, 15.24it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 15.22it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 15.23it/s]
+
 92%|█████████▏| 36/39 [00:02<00:00, 15.21it/s]
+
 97%|█████████▋| 38/39 [00:02<00:00, 15.21it/s]
                                                   
+
                                               
{'eval_loss': '4.193', 'eval_runtime': '2.616', 'eval_samples_per_second': '933', 'eval_steps_per_second': '14.91', 'epoch': '70'}
+
 34%|███▍      | 2520/7308 [29:45<41:16,  1.93it/s]
+
100%|██████████| 39/39 [00:02<00:00, 15.21it/s][eval] epoch 70.0 step 2520 loss 4.1926 (best 4.0798 @ ep 46.0, 12/20 since best)
+
+
                                               
 34%|███▍      | 2521/7308 [29:46<1:48:39,  1.36s/it]
 35%|███▍      | 2522/7308 [29:46<1:31:13,  1.14s/it]
 35%|███▍      | 2523/7308 [29:47<1:19:03,  1.01it/s]
 35%|███▍      | 2524/7308 [29:48<1:10:29,  1.13it/s]
 35%|███▍      | 2525/7308 [29:48<1:04:35,  1.23it/s]
 35%|███▍      | 2526/7308 [29:49<1:00:22,  1.32it/s]
 35%|███▍      | 2527/7308 [29:50<57:25,  1.39it/s]  
 35%|███▍      | 2528/7308 [29:50<55:22,  1.44it/s]
 35%|███▍      | 2529/7308 [29:51<53:55,  1.48it/s]
 35%|███▍      | 2530/7308 [29:52<52:53,  1.51it/s]
                                                   
{'loss': '3.842', 'grad_norm': '0.3079', 'learning_rate': '0.0006', 'epoch': '70.28'}
+
 35%|███▍      | 2530/7308 [29:52<52:53,  1.51it/s]
 35%|███▍      | 2531/7308 [29:52<52:11,  1.53it/s]
 35%|███▍      | 2532/7308 [29:53<51:41,  1.54it/s]
 35%|███▍      | 2533/7308 [29:53<51:20,  1.55it/s]
 35%|███▍      | 2534/7308 [29:54<51:04,  1.56it/s]
 35%|███▍      | 2535/7308 [29:55<50:53,  1.56it/s]
 35%|███▍      | 2536/7308 [29:55<50:45,  1.57it/s]
 35%|███▍      | 2537/7308 [29:56<50:38,  1.57it/s]
 35%|███▍      | 2538/7308 [29:57<50:35,  1.57it/s]
 35%|███▍      | 2539/7308 [29:57<50:33,  1.57it/s]
 35%|███▍      | 2540/7308 [29:58<50:30,  1.57it/s]
                                                   
{'loss': '3.88', 'grad_norm': '0.3375', 'learning_rate': '0.0006', 'epoch': '70.57'}
+
 35%|███▍      | 2540/7308 [29:58<50:30,  1.57it/s]
 35%|███▍      | 2541/7308 [29:59<50:30,  1.57it/s]
 35%|███▍      | 2542/7308 [29:59<50:28,  1.57it/s]
 35%|███▍      | 2543/7308 [30:00<50:26,  1.57it/s]
 35%|███▍      | 2544/7308 [30:00<50:24,  1.58it/s]
 35%|███▍      | 2545/7308 [30:01<50:22,  1.58it/s]
 35%|███▍      | 2546/7308 [30:02<50:22,  1.58it/s]
 35%|███▍      | 2547/7308 [30:02<50:22,  1.58it/s]
 35%|███▍      | 2548/7308 [30:03<50:21,  1.58it/s]
 35%|███▍      | 2549/7308 [30:04<50:20,  1.58it/s]
 35%|███▍      | 2550/7308 [30:04<50:19,  1.58it/s]
                                                   
{'loss': '3.926', 'grad_norm': '0.2851', 'learning_rate': '0.0006', 'epoch': '70.85'}
+
 35%|███▍      | 2550/7308 [30:04<50:19,  1.58it/s]
 35%|███▍      | 2551/7308 [30:05<50:21,  1.57it/s]
 35%|███▍      | 2552/7308 [30:05<50:20,  1.57it/s]
 35%|███▍      | 2553/7308 [30:06<50:18,  1.58it/s]
 35%|███▍      | 2554/7308 [30:07<50:18,  1.58it/s]
 35%|███▍      | 2555/7308 [30:07<50:18,  1.57it/s]
 35%|███▍      | 2556/7308 [30:08<40:59,  1.93it/s]
 35%|███▍      | 2557/7308 [30:08<45:47,  1.73it/s]
 35%|███▌      | 2558/7308 [30:09<47:09,  1.68it/s]
 35%|███▌      | 2559/7308 [30:10<48:04,  1.65it/s]
 35%|███▌      | 2560/7308 [30:10<48:41,  1.63it/s]
                                                   
{'loss': '3.928', 'grad_norm': '0.3893', 'learning_rate': '0.0006', 'epoch': '71.11'}
+
 35%|███▌      | 2560/7308 [30:10<48:41,  1.63it/s]
 35%|███▌      | 2561/7308 [30:11<49:10,  1.61it/s]
 35%|███▌      | 2562/7308 [30:12<49:28,  1.60it/s]
 35%|███▌      | 2563/7308 [30:12<49:40,  1.59it/s]
 35%|███▌      | 2564/7308 [30:13<49:49,  1.59it/s]
 35%|███▌      | 2565/7308 [30:13<49:54,  1.58it/s]
 35%|███▌      | 2566/7308 [30:14<49:58,  1.58it/s]
 35%|███▌      | 2567/7308 [30:15<50:00,  1.58it/s]
 35%|███▌      | 2568/7308 [30:15<50:02,  1.58it/s]
 35%|███▌      | 2569/7308 [30:16<50:02,  1.58it/s]
 35%|███▌      | 2570/7308 [30:17<50:02,  1.58it/s]
                                                   
{'loss': '3.926', 'grad_norm': '0.3607', 'learning_rate': '0.0006', 'epoch': '71.4'}
+
 35%|███▌      | 2570/7308 [30:17<50:02,  1.58it/s]
 35%|███▌      | 2571/7308 [30:17<50:05,  1.58it/s]
 35%|███▌      | 2572/7308 [30:18<50:05,  1.58it/s]
 35%|███▌      | 2573/7308 [30:19<50:04,  1.58it/s]
 35%|███▌      | 2574/7308 [30:19<50:03,  1.58it/s]
 35%|███▌      | 2575/7308 [30:20<50:03,  1.58it/s]
 35%|███▌      | 2576/7308 [30:20<50:03,  1.58it/s]
 35%|███▌      | 2577/7308 [30:21<50:02,  1.58it/s]
 35%|███▌      | 2578/7308 [30:22<50:02,  1.58it/s]
 35%|███▌      | 2579/7308 [30:22<50:02,  1.57it/s]
 35%|███▌      | 2580/7308 [30:23<50:02,  1.57it/s]
                                                   
{'loss': '3.926', 'grad_norm': '0.3587', 'learning_rate': '0.0006', 'epoch': '71.68'}
+
 35%|███▌      | 2580/7308 [30:23<50:02,  1.57it/s]
 35%|███▌      | 2581/7308 [30:24<50:02,  1.57it/s]
 35%|███▌      | 2582/7308 [30:24<50:00,  1.58it/s]
 35%|███▌      | 2583/7308 [30:25<49:58,  1.58it/s]
 35%|███▌      | 2584/7308 [30:26<49:58,  1.58it/s]
 35%|███▌      | 2585/7308 [30:26<49:58,  1.58it/s]
 35%|███▌      | 2586/7308 [30:27<49:58,  1.57it/s]
 35%|███▌      | 2587/7308 [30:27<50:01,  1.57it/s]
 35%|███▌      | 2588/7308 [30:28<50:01,  1.57it/s]
 35%|███▌      | 2589/7308 [30:29<50:00,  1.57it/s]
 35%|███▌      | 2590/7308 [30:29<50:02,  1.57it/s]
                                                   
{'loss': '3.934', 'grad_norm': '0.2713', 'learning_rate': '0.0006', 'epoch': '71.96'}
+
 35%|███▌      | 2590/7308 [30:29<50:02,  1.57it/s]
 35%|███▌      | 2591/7308 [30:30<49:58,  1.57it/s]
 35%|███▌      | 2592/7308 [30:30<40:43,  1.93it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 23.65it/s]
+
 15%|█▌        | 6/39 [00:00<00:01, 17.77it/s]
+
 21%|██        | 8/39 [00:00<00:01, 16.76it/s]
+
 26%|██▌       | 10/39 [00:00<00:01, 16.15it/s]
+
 31%|███       | 12/39 [00:00<00:01, 15.84it/s]
+
 36%|███▌      | 14/39 [00:00<00:01, 15.60it/s]
+
 41%|████      | 16/39 [00:00<00:01, 15.48it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 15.36it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 15.27it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 15.23it/s]
+
 62%|██████▏   | 24/39 [00:01<00:00, 15.26it/s]
+
 67%|██████▋   | 26/39 [00:01<00:00, 15.24it/s]
+
 72%|███████▏  | 28/39 [00:01<00:00, 15.24it/s]
+
 77%|███████▋  | 30/39 [00:01<00:00, 15.23it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 15.23it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 15.20it/s]
+
 92%|█████████▏| 36/39 [00:02<00:00, 15.22it/s]
+
 97%|█████████▋| 38/39 [00:02<00:00, 15.21it/s]
                                                   
+
                                               
{'eval_loss': '4.205', 'eval_runtime': '2.614', 'eval_samples_per_second': '933.8', 'eval_steps_per_second': '14.92', 'epoch': '72'}
+
 35%|███▌      | 2592/7308 [30:33<40:43,  1.93it/s]
+
100%|██████████| 39/39 [00:02<00:00, 15.21it/s][eval] epoch 72.0 step 2592 loss 4.2052 (best 4.0798 @ ep 46.0, 13/20 since best)
+
+
                                               
 35%|███▌      | 2593/7308 [30:34<1:46:41,  1.36s/it]
 35%|███▌      | 2594/7308 [30:34<1:29:37,  1.14s/it]
 36%|███▌      | 2595/7308 [30:35<1:17:41,  1.01it/s]
 36%|███▌      | 2596/7308 [30:35<1:09:20,  1.13it/s]
 36%|███▌      | 2597/7308 [30:36<1:03:29,  1.24it/s]
 36%|███▌      | 2598/7308 [30:37<59:23,  1.32it/s]  
 36%|███▌      | 2599/7308 [30:37<56:31,  1.39it/s]
 36%|███▌      | 2600/7308 [30:38<54:31,  1.44it/s]
                                                   
{'loss': '3.848', 'grad_norm': '0.3249', 'learning_rate': '0.0006', 'epoch': '72.23'}
+
 36%|███▌      | 2600/7308 [30:38<54:31,  1.44it/s]
 36%|███▌      | 2601/7308 [30:39<53:07,  1.48it/s]
 36%|███▌      | 2602/7308 [30:39<52:07,  1.50it/s]
 36%|███▌      | 2603/7308 [30:40<51:25,  1.52it/s]
 36%|███▌      | 2604/7308 [30:41<50:54,  1.54it/s]
 36%|███▌      | 2605/7308 [30:41<50:33,  1.55it/s]
 36%|███▌      | 2606/7308 [30:42<50:18,  1.56it/s]
 36%|███▌      | 2607/7308 [30:42<50:08,  1.56it/s]
 36%|███▌      | 2608/7308 [30:43<50:02,  1.57it/s]
 36%|███▌      | 2609/7308 [30:44<49:56,  1.57it/s]
 36%|███▌      | 2610/7308 [30:44<49:51,  1.57it/s]
                                                   
{'loss': '3.838', 'grad_norm': '0.3555', 'learning_rate': '0.0006', 'epoch': '72.51'}
+
 36%|███▌      | 2610/7308 [30:44<49:51,  1.57it/s]
 36%|███▌      | 2611/7308 [30:45<49:50,  1.57it/s]
 36%|███▌      | 2612/7308 [30:46<49:48,  1.57it/s]
 36%|███▌      | 2613/7308 [30:46<49:47,  1.57it/s]
 36%|███▌      | 2614/7308 [30:47<49:45,  1.57it/s]
 36%|███▌      | 2615/7308 [30:47<49:43,  1.57it/s]
 36%|███▌      | 2616/7308 [30:48<49:43,  1.57it/s]
 36%|███▌      | 2617/7308 [30:49<49:41,  1.57it/s]
 36%|███▌      | 2618/7308 [30:49<49:39,  1.57it/s]
 36%|███▌      | 2619/7308 [30:50<49:39,  1.57it/s]
 36%|███▌      | 2620/7308 [30:51<49:38,  1.57it/s]
                                                   
{'loss': '3.961', 'grad_norm': '0.3163', 'learning_rate': '0.0006', 'epoch': '72.79'}
+
 36%|███▌      | 2620/7308 [30:51<49:38,  1.57it/s]
 36%|███▌      | 2621/7308 [30:51<49:38,  1.57it/s]
 36%|███▌      | 2622/7308 [30:52<49:36,  1.57it/s]
 36%|███▌      | 2623/7308 [30:53<49:35,  1.57it/s]
 36%|███▌      | 2624/7308 [30:53<49:36,  1.57it/s]
 36%|███▌      | 2625/7308 [30:54<49:34,  1.57it/s]
 36%|███▌      | 2626/7308 [30:54<49:32,  1.58it/s]
 36%|███▌      | 2627/7308 [30:55<49:30,  1.58it/s]
 36%|███▌      | 2628/7308 [30:55<40:20,  1.93it/s]
 36%|███▌      | 2629/7308 [30:56<44:35,  1.75it/s]
 36%|███▌      | 2630/7308 [30:57<46:02,  1.69it/s]
                                                   
{'loss': '3.938', 'grad_norm': '0.3189', 'learning_rate': '0.0006', 'epoch': '73.06'}
+
 36%|███▌      | 2630/7308 [30:57<46:02,  1.69it/s]
 36%|███▌      | 2631/7308 [30:57<47:07,  1.65it/s]
 36%|███▌      | 2632/7308 [30:58<47:48,  1.63it/s]
 36%|███▌      | 2633/7308 [30:59<48:18,  1.61it/s]
 36%|███▌      | 2634/7308 [30:59<48:37,  1.60it/s]
 36%|███▌      | 2635/7308 [31:00<48:52,  1.59it/s]
 36%|███▌      | 2636/7308 [31:01<49:01,  1.59it/s]
 36%|███▌      | 2637/7308 [31:01<49:08,  1.58it/s]
 36%|███▌      | 2638/7308 [31:02<49:12,  1.58it/s]
 36%|███▌      | 2639/7308 [31:02<49:14,  1.58it/s]
 36%|███▌      | 2640/7308 [31:03<49:18,  1.58it/s]
                                                   
{'loss': '3.843', 'grad_norm': '0.3396', 'learning_rate': '0.0006', 'epoch': '73.34'}
+
 36%|███▌      | 2640/7308 [31:03<49:18,  1.58it/s]
 36%|███▌      | 2641/7308 [31:04<49:19,  1.58it/s]
 36%|███▌      | 2642/7308 [31:04<49:19,  1.58it/s]
 36%|███▌      | 2643/7308 [31:05<49:20,  1.58it/s]
 36%|███▌      | 2644/7308 [31:06<49:19,  1.58it/s]
 36%|███▌      | 2645/7308 [31:06<49:18,  1.58it/s]
 36%|███▌      | 2646/7308 [31:07<49:18,  1.58it/s]
 36%|███▌      | 2647/7308 [31:07<49:18,  1.58it/s]
 36%|███▌      | 2648/7308 [31:08<49:17,  1.58it/s]
 36%|███▌      | 2649/7308 [31:09<49:17,  1.58it/s]
 36%|███▋      | 2650/7308 [31:09<49:17,  1.57it/s]
                                                   
{'loss': '3.908', 'grad_norm': '0.2926', 'learning_rate': '0.0006', 'epoch': '73.62'}
+
 36%|███▋      | 2650/7308 [31:09<49:17,  1.57it/s]
 36%|███▋      | 2651/7308 [31:10<49:17,  1.57it/s]
 36%|███▋      | 2652/7308 [31:11<49:17,  1.57it/s]
 36%|███▋      | 2653/7308 [31:11<49:16,  1.57it/s]
 36%|███▋      | 2654/7308 [31:12<49:14,  1.58it/s]
 36%|███▋      | 2655/7308 [31:13<49:13,  1.58it/s]
 36%|███▋      | 2656/7308 [31:13<49:13,  1.58it/s]
 36%|███▋      | 2657/7308 [31:14<49:12,  1.58it/s]
 36%|███▋      | 2658/7308 [31:14<49:11,  1.58it/s]
 36%|███▋      | 2659/7308 [31:15<49:11,  1.58it/s]
 36%|███▋      | 2660/7308 [31:16<49:11,  1.57it/s]
                                                   
{'loss': '3.913', 'grad_norm': '0.3085', 'learning_rate': '0.0006', 'epoch': '73.9'}
+
 36%|███▋      | 2660/7308 [31:16<49:11,  1.57it/s]
 36%|███▋      | 2661/7308 [31:16<49:10,  1.57it/s]
 36%|███▋      | 2662/7308 [31:17<49:09,  1.57it/s]
 36%|███▋      | 2663/7308 [31:18<49:07,  1.58it/s]
 36%|███▋      | 2664/7308 [31:18<40:02,  1.93it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 23.43it/s]
+
 15%|█▌        | 6/39 [00:00<00:01, 17.74it/s]
+
 21%|██        | 8/39 [00:00<00:01, 16.73it/s]
+
 26%|██▌       | 10/39 [00:00<00:01, 16.15it/s]
+
 31%|███       | 12/39 [00:00<00:01, 15.83it/s]
+
 36%|███▌      | 14/39 [00:00<00:01, 15.60it/s]
+
 41%|████      | 16/39 [00:00<00:01, 15.47it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 15.37it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 15.32it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 15.23it/s]
+
 62%|██████▏   | 24/39 [00:01<00:00, 15.25it/s]
+
 67%|██████▋   | 26/39 [00:01<00:00, 15.24it/s]
+
 72%|███████▏  | 28/39 [00:01<00:00, 15.23it/s]
+
 77%|███████▋  | 30/39 [00:01<00:00, 15.23it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 15.21it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 15.23it/s]
+
 92%|█████████▏| 36/39 [00:02<00:00, 15.23it/s]
+
 97%|█████████▋| 38/39 [00:02<00:00, 15.23it/s]
                                                   
+
                                               
{'eval_loss': '4.225', 'eval_runtime': '2.622', 'eval_samples_per_second': '931', 'eval_steps_per_second': '14.87', 'epoch': '74'}
+
 36%|███▋      | 2664/7308 [31:21<40:02,  1.93it/s]
+
100%|██████████| 39/39 [00:02<00:00, 15.23it/s][eval] epoch 74.0 step 2664 loss 4.2251 (best 4.0798 @ ep 46.0, 14/20 since best)
+
+
                                               
 36%|███▋      | 2665/7308 [31:21<1:45:07,  1.36s/it]
 36%|███▋      | 2666/7308 [31:22<1:28:18,  1.14s/it]
 36%|███▋      | 2667/7308 [31:22<1:16:31,  1.01it/s]
 37%|███▋      | 2668/7308 [31:23<1:08:17,  1.13it/s]
 37%|███▋      | 2669/7308 [31:24<1:02:31,  1.24it/s]
 37%|███▋      | 2670/7308 [31:24<58:29,  1.32it/s]  
                                                   
{'loss': '3.874', 'grad_norm': '0.3065', 'learning_rate': '0.0006', 'epoch': '74.17'}
+
 37%|███▋      | 2670/7308 [31:24<58:29,  1.32it/s]
 37%|███▋      | 2671/7308 [31:25<55:39,  1.39it/s]
 37%|███▋      | 2672/7308 [31:26<53:41,  1.44it/s]
 37%|███▋      | 2673/7308 [31:26<52:18,  1.48it/s]
 37%|███▋      | 2674/7308 [31:27<51:20,  1.50it/s]
 37%|███▋      | 2675/7308 [31:28<50:38,  1.52it/s]
 37%|███▋      | 2676/7308 [31:28<50:08,  1.54it/s]
 37%|███▋      | 2677/7308 [31:29<49:47,  1.55it/s]
 37%|███▋      | 2678/7308 [31:29<49:32,  1.56it/s]
 37%|███▋      | 2679/7308 [31:30<49:22,  1.56it/s]
 37%|███▋      | 2680/7308 [31:31<49:15,  1.57it/s]
                                                   
{'loss': '3.917', 'grad_norm': '0.3225', 'learning_rate': '0.0006', 'epoch': '74.45'}
+
 37%|███▋      | 2680/7308 [31:31<49:15,  1.57it/s]
 37%|███▋      | 2681/7308 [31:31<49:09,  1.57it/s]
 37%|███▋      | 2682/7308 [31:32<49:05,  1.57it/s]
 37%|███▋      | 2683/7308 [31:33<49:02,  1.57it/s]
 37%|███▋      | 2684/7308 [31:33<48:59,  1.57it/s]
 37%|███▋      | 2685/7308 [31:34<48:57,  1.57it/s]
 37%|███▋      | 2686/7308 [31:35<48:55,  1.57it/s]
 37%|███▋      | 2687/7308 [31:35<48:55,  1.57it/s]
 37%|███▋      | 2688/7308 [31:36<48:53,  1.57it/s]
 37%|███▋      | 2689/7308 [31:36<48:52,  1.58it/s]
 37%|███▋      | 2690/7308 [31:37<48:52,  1.57it/s]
                                                   
{'loss': '3.81', 'grad_norm': '0.3246', 'learning_rate': '0.0006', 'epoch': '74.73'}
+
 37%|███▋      | 2690/7308 [31:37<48:52,  1.57it/s]
 37%|███▋      | 2691/7308 [31:38<48:56,  1.57it/s]
 37%|███▋      | 2692/7308 [31:38<48:53,  1.57it/s]
 37%|███▋      | 2693/7308 [31:39<48:51,  1.57it/s]
 37%|███▋      | 2694/7308 [31:40<48:50,  1.57it/s]
 37%|███▋      | 2695/7308 [31:40<48:48,  1.57it/s]
 37%|███▋      | 2696/7308 [31:41<48:47,  1.58it/s]
 37%|███▋      | 2697/7308 [31:42<48:46,  1.58it/s]
 37%|███▋      | 2698/7308 [31:42<48:45,  1.58it/s]
 37%|███▋      | 2699/7308 [31:43<48:43,  1.58it/s]
 37%|███▋      | 2700/7308 [31:43<39:42,  1.93it/s]
                                                   
{'loss': '3.976', 'grad_norm': '0.3591', 'learning_rate': '0.0006', 'epoch': '75'}
+
 37%|███▋      | 2700/7308 [31:43<39:42,  1.93it/s]
 37%|███▋      | 2701/7308 [31:44<44:06,  1.74it/s]
 37%|███▋      | 2702/7308 [31:44<45:29,  1.69it/s]
 37%|███▋      | 2703/7308 [31:45<46:27,  1.65it/s]
 37%|███▋      | 2704/7308 [31:46<47:06,  1.63it/s]
 37%|███▋      | 2705/7308 [31:46<47:34,  1.61it/s]
 37%|███▋      | 2706/7308 [31:47<47:53,  1.60it/s]
 37%|███▋      | 2707/7308 [31:48<48:07,  1.59it/s]
 37%|███▋      | 2708/7308 [31:48<48:16,  1.59it/s]
 37%|███▋      | 2709/7308 [31:49<48:22,  1.58it/s]
 37%|███▋      | 2710/7308 [31:49<48:27,  1.58it/s]
                                                   
{'loss': '3.836', 'grad_norm': '0.3047', 'learning_rate': '0.0006', 'epoch': '75.28'}
+
 37%|███▋      | 2710/7308 [31:49<48:27,  1.58it/s]
 37%|███▋      | 2711/7308 [31:50<48:30,  1.58it/s]
 37%|███▋      | 2712/7308 [31:51<48:31,  1.58it/s]
 37%|███▋      | 2713/7308 [31:51<48:32,  1.58it/s]
 37%|███▋      | 2714/7308 [31:52<48:33,  1.58it/s]
 37%|███▋      | 2715/7308 [31:53<48:33,  1.58it/s]
 37%|███▋      | 2716/7308 [31:53<48:34,  1.58it/s]
 37%|███▋      | 2717/7308 [31:54<48:33,  1.58it/s]
 37%|███▋      | 2718/7308 [31:55<48:33,  1.58it/s]
 37%|███▋      | 2719/7308 [31:55<48:32,  1.58it/s]
 37%|███▋      | 2720/7308 [31:56<48:31,  1.58it/s]
                                                   
{'loss': '3.787', 'grad_norm': '0.368', 'learning_rate': '0.0006', 'epoch': '75.57'}
+
 37%|███▋      | 2720/7308 [31:56<48:31,  1.58it/s]
 37%|███▋      | 2721/7308 [31:56<48:31,  1.58it/s]
 37%|███▋      | 2722/7308 [31:57<48:30,  1.58it/s]
 37%|███▋      | 2723/7308 [31:58<48:28,  1.58it/s]
 37%|███▋      | 2724/7308 [31:58<48:27,  1.58it/s]
 37%|███▋      | 2725/7308 [31:59<48:27,  1.58it/s]
 37%|███▋      | 2726/7308 [32:00<48:26,  1.58it/s]
 37%|███▋      | 2727/7308 [32:00<48:25,  1.58it/s]
 37%|███▋      | 2728/7308 [32:01<48:25,  1.58it/s]
 37%|███▋      | 2729/7308 [32:02<48:25,  1.58it/s]
 37%|███▋      | 2730/7308 [32:02<48:23,  1.58it/s]
                                                   
{'loss': '3.92', 'grad_norm': '0.2986', 'learning_rate': '0.0006', 'epoch': '75.85'}
+
 37%|███▋      | 2730/7308 [32:02<48:23,  1.58it/s]
 37%|███▋      | 2731/7308 [32:03<48:24,  1.58it/s]
 37%|███▋      | 2732/7308 [32:03<48:23,  1.58it/s]
 37%|███▋      | 2733/7308 [32:04<48:22,  1.58it/s]
 37%|███▋      | 2734/7308 [32:05<48:22,  1.58it/s]
 37%|███▋      | 2735/7308 [32:05<48:21,  1.58it/s]
 37%|███▋      | 2736/7308 [32:06<39:25,  1.93it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 23.14it/s]
+
 15%|█▌        | 6/39 [00:00<00:01, 17.64it/s]
+
 21%|██        | 8/39 [00:00<00:01, 16.69it/s]
+
 26%|██▌       | 10/39 [00:00<00:01, 16.11it/s]
+
 31%|███       | 12/39 [00:00<00:01, 15.79it/s]
+
 36%|███▌      | 14/39 [00:00<00:01, 15.57it/s]
+
 41%|████      | 16/39 [00:00<00:01, 15.46it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 15.35it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 15.31it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 15.24it/s]
+
 62%|██████▏   | 24/39 [00:01<00:00, 15.26it/s]
+
 67%|██████▋   | 26/39 [00:01<00:00, 15.24it/s]
+
 72%|███████▏  | 28/39 [00:01<00:00, 15.24it/s]
+
 77%|███████▋  | 30/39 [00:01<00:00, 15.22it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 15.23it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 15.22it/s]
+
 92%|█████████▏| 36/39 [00:02<00:00, 15.23it/s]
+
 97%|█████████▋| 38/39 [00:02<00:00, 15.23it/s]
                                                   
+
                                               
{'eval_loss': '4.246', 'eval_runtime': '2.619', 'eval_samples_per_second': '932.2', 'eval_steps_per_second': '14.89', 'epoch': '76'}
+
 37%|███▋      | 2736/7308 [32:08<39:25,  1.93it/s]
+
100%|██████████| 39/39 [00:02<00:00, 15.23it/s][eval] epoch 76.0 step 2736 loss 4.2464 (best 4.0798 @ ep 46.0, 15/20 since best)
+
+
                                               
 37%|███▋      | 2737/7308 [32:09<1:44:10,  1.37s/it]
 37%|███▋      | 2738/7308 [32:10<1:27:23,  1.15s/it]
 37%|███▋      | 2739/7308 [32:10<1:15:42,  1.01it/s]
 37%|███▋      | 2740/7308 [32:11<1:07:28,  1.13it/s]
                                                     
{'loss': '3.895', 'grad_norm': '0.3891', 'learning_rate': '0.0006', 'epoch': '76.11'}
+
 37%|███▋      | 2740/7308 [32:11<1:07:28,  1.13it/s]
 38%|███▊      | 2741/7308 [32:11<1:01:42,  1.23it/s]
 38%|███▊      | 2742/7308 [32:12<57:39,  1.32it/s]  
 38%|███▊      | 2743/7308 [32:13<54:49,  1.39it/s]
 38%|███▊      | 2744/7308 [32:13<52:52,  1.44it/s]
 38%|███▊      | 2745/7308 [32:14<51:28,  1.48it/s]
 38%|███▊      | 2746/7308 [32:15<50:29,  1.51it/s]
 38%|███▊      | 2747/7308 [32:15<49:48,  1.53it/s]
 38%|███▊      | 2748/7308 [32:16<49:19,  1.54it/s]
 38%|███▊      | 2749/7308 [32:17<48:59,  1.55it/s]
 38%|███▊      | 2750/7308 [32:17<48:45,  1.56it/s]
                                                   
{'loss': '3.893', 'grad_norm': '0.3009', 'learning_rate': '0.0006', 'epoch': '76.4'}
+
 38%|███▊      | 2750/7308 [32:17<48:45,  1.56it/s]
 38%|███▊      | 2751/7308 [32:18<48:35,  1.56it/s]
 38%|███▊      | 2752/7308 [32:18<48:26,  1.57it/s]
 38%|███▊      | 2753/7308 [32:19<48:21,  1.57it/s]
 38%|███▊      | 2754/7308 [32:20<48:17,  1.57it/s]
 38%|███▊      | 2755/7308 [32:20<48:14,  1.57it/s]
 38%|███▊      | 2756/7308 [32:21<48:12,  1.57it/s]
 38%|███▊      | 2757/7308 [32:22<48:09,  1.57it/s]
 38%|███▊      | 2758/7308 [32:22<48:08,  1.57it/s]
 38%|███▊      | 2759/7308 [32:23<48:08,  1.58it/s]
 38%|███▊      | 2760/7308 [32:24<48:06,  1.58it/s]
                                                   
{'loss': '3.859', 'grad_norm': '0.2946', 'learning_rate': '0.0006', 'epoch': '76.68'}
+
 38%|███▊      | 2760/7308 [32:24<48:06,  1.58it/s]
 38%|███▊      | 2761/7308 [32:24<48:05,  1.58it/s]
 38%|███▊      | 2762/7308 [32:25<48:04,  1.58it/s]
 38%|███▊      | 2763/7308 [32:25<48:04,  1.58it/s]
 38%|███▊      | 2764/7308 [32:26<48:04,  1.58it/s]
 38%|███▊      | 2765/7308 [32:27<48:03,  1.58it/s]
 38%|███▊      | 2766/7308 [32:27<48:02,  1.58it/s]
 38%|███▊      | 2767/7308 [32:28<48:01,  1.58it/s]
 38%|███▊      | 2768/7308 [32:29<48:00,  1.58it/s]
 38%|███▊      | 2769/7308 [32:29<48:02,  1.57it/s]
 38%|███▊      | 2770/7308 [32:30<48:00,  1.58it/s]
                                                   
{'loss': '3.87', 'grad_norm': '0.2939', 'learning_rate': '0.0006', 'epoch': '76.96'}
+
 38%|███▊      | 2770/7308 [32:30<48:00,  1.58it/s]
 38%|███▊      | 2771/7308 [32:30<47:58,  1.58it/s]
 38%|███▊      | 2772/7308 [32:31<39:05,  1.93it/s]
 38%|███▊      | 2773/7308 [32:31<43:57,  1.72it/s]
 38%|███▊      | 2774/7308 [32:32<45:08,  1.67it/s]
 38%|███▊      | 2775/7308 [32:33<45:59,  1.64it/s]
 38%|███▊      | 2776/7308 [32:33<46:33,  1.62it/s]
 38%|███▊      | 2777/7308 [32:34<46:57,  1.61it/s]
 38%|███▊      | 2778/7308 [32:35<47:14,  1.60it/s]
 38%|███▊      | 2779/7308 [32:35<47:25,  1.59it/s]
 38%|███▊      | 2780/7308 [32:36<47:32,  1.59it/s]
                                                   
{'loss': '3.843', 'grad_norm': '0.3621', 'learning_rate': '0.0006', 'epoch': '77.23'}
+
 38%|███▊      | 2780/7308 [32:36<47:32,  1.59it/s]
 38%|███▊      | 2781/7308 [32:37<47:39,  1.58it/s]
 38%|███▊      | 2782/7308 [32:37<47:42,  1.58it/s]
 38%|███▊      | 2783/7308 [32:38<47:46,  1.58it/s]
 38%|███▊      | 2784/7308 [32:38<47:46,  1.58it/s]
 38%|███▊      | 2785/7308 [32:39<47:47,  1.58it/s]
 38%|███▊      | 2786/7308 [32:40<47:47,  1.58it/s]
 38%|███▊      | 2787/7308 [32:40<47:47,  1.58it/s]
 38%|███▊      | 2788/7308 [32:41<47:47,  1.58it/s]
 38%|███▊      | 2789/7308 [32:42<47:46,  1.58it/s]
 38%|███▊      | 2790/7308 [32:42<47:46,  1.58it/s]
                                                   
{'loss': '3.955', 'grad_norm': '0.3496', 'learning_rate': '0.0006', 'epoch': '77.51'}
+
 38%|███▊      | 2790/7308 [32:42<47:46,  1.58it/s]
 38%|███▊      | 2791/7308 [32:43<47:46,  1.58it/s]
 38%|███▊      | 2792/7308 [32:44<47:45,  1.58it/s]
 38%|███▊      | 2793/7308 [32:44<47:45,  1.58it/s]
 38%|███▊      | 2794/7308 [32:45<47:45,  1.58it/s]
 38%|███▊      | 2795/7308 [32:45<47:43,  1.58it/s]
 38%|███▊      | 2796/7308 [32:46<47:42,  1.58it/s]
 38%|███▊      | 2797/7308 [32:47<47:43,  1.58it/s]
 38%|███▊      | 2798/7308 [32:47<47:42,  1.58it/s]
 38%|███▊      | 2799/7308 [32:48<47:42,  1.58it/s]
 38%|███▊      | 2800/7308 [32:49<47:40,  1.58it/s]
                                                   
{'loss': '3.868', 'grad_norm': '0.3154', 'learning_rate': '0.0006', 'epoch': '77.79'}
+
 38%|███▊      | 2800/7308 [32:49<47:40,  1.58it/s]
 38%|███▊      | 2801/7308 [32:49<47:41,  1.58it/s]
 38%|███▊      | 2802/7308 [32:50<47:39,  1.58it/s]
 38%|███▊      | 2803/7308 [32:51<47:39,  1.58it/s]
 38%|███▊      | 2804/7308 [32:51<47:39,  1.58it/s]
 38%|███▊      | 2805/7308 [32:52<47:38,  1.58it/s]
 38%|███▊      | 2806/7308 [32:52<47:37,  1.58it/s]
 38%|███▊      | 2807/7308 [32:53<47:35,  1.58it/s]
 38%|███▊      | 2808/7308 [32:53<38:46,  1.93it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 22.81it/s]
+
 15%|█▌        | 6/39 [00:00<00:01, 17.61it/s]
+
 21%|██        | 8/39 [00:00<00:01, 16.64it/s]
+
 26%|██▌       | 10/39 [00:00<00:01, 16.10it/s]
+
 31%|███       | 12/39 [00:00<00:01, 15.78it/s]
+
 36%|███▌      | 14/39 [00:00<00:01, 15.59it/s]
+
 41%|████      | 16/39 [00:00<00:01, 15.44it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 15.38it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 15.31it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 15.25it/s]
+
 62%|██████▏   | 24/39 [00:01<00:00, 15.25it/s]
+
 67%|██████▋   | 26/39 [00:01<00:00, 15.24it/s]
+
 72%|███████▏  | 28/39 [00:01<00:00, 15.23it/s]
+
 77%|███████▋  | 30/39 [00:01<00:00, 15.23it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 15.22it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 15.23it/s]
+
 92%|█████████▏| 36/39 [00:02<00:00, 15.21it/s]
+
 97%|█████████▋| 38/39 [00:02<00:00, 15.22it/s]
                                                   
+
                                               
{'eval_loss': '4.247', 'eval_runtime': '2.638', 'eval_samples_per_second': '925.3', 'eval_steps_per_second': '14.78', 'epoch': '78'}
+
 38%|███▊      | 2808/7308 [32:56<38:46,  1.93it/s]
+
100%|██████████| 39/39 [00:02<00:00, 15.22it/s][eval] epoch 78.0 step 2808 loss 4.2475 (best 4.0798 @ ep 46.0, 16/20 since best)
+
+
                                               
 38%|███▊      | 2809/7308 [32:57<1:42:43,  1.37s/it]
 38%|███▊      | 2810/7308 [32:57<1:26:11,  1.15s/it]
                                                     
{'loss': '3.761', 'grad_norm': '0.4166', 'learning_rate': '0.0006', 'epoch': '78.06'}
+
 38%|███▊      | 2810/7308 [32:57<1:26:11,  1.15s/it]
 38%|███▊      | 2811/7308 [32:58<1:14:37,  1.00it/s]
 38%|███▊      | 2812/7308 [32:59<1:06:29,  1.13it/s]
 38%|███▊      | 2813/7308 [32:59<1:00:47,  1.23it/s]
 39%|███▊      | 2814/7308 [33:00<56:47,  1.32it/s]  
 39%|███▊      | 2815/7308 [33:00<54:00,  1.39it/s]
 39%|███▊      | 2816/7308 [33:01<52:03,  1.44it/s]
 39%|███▊      | 2817/7308 [33:02<50:41,  1.48it/s]
 39%|███▊      | 2818/7308 [33:02<49:41,  1.51it/s]
 39%|███▊      | 2819/7308 [33:03<49:00,  1.53it/s]
 39%|███▊      | 2820/7308 [33:04<48:31,  1.54it/s]
                                                   
{'loss': '3.835', 'grad_norm': '0.324', 'learning_rate': '0.0006', 'epoch': '78.34'}
+
 39%|███▊      | 2820/7308 [33:04<48:31,  1.54it/s]
 39%|███▊      | 2821/7308 [33:04<48:12,  1.55it/s]
 39%|███▊      | 2822/7308 [33:05<47:56,  1.56it/s]
 39%|███▊      | 2823/7308 [33:06<47:45,  1.56it/s]
 39%|███▊      | 2824/7308 [33:06<47:38,  1.57it/s]
 39%|███▊      | 2825/7308 [33:07<47:33,  1.57it/s]
 39%|███▊      | 2826/7308 [33:07<47:29,  1.57it/s]
 39%|███▊      | 2827/7308 [33:08<47:26,  1.57it/s]
 39%|███▊      | 2828/7308 [33:09<47:24,  1.57it/s]
 39%|███▊      | 2829/7308 [33:09<47:22,  1.58it/s]
 39%|███▊      | 2830/7308 [33:10<47:20,  1.58it/s]
                                                   
{'loss': '3.918', 'grad_norm': '0.3009', 'learning_rate': '0.0006', 'epoch': '78.62'}
+
 39%|███▊      | 2830/7308 [33:10<47:20,  1.58it/s]
 39%|███▊      | 2831/7308 [33:11<47:18,  1.58it/s]
 39%|███▉      | 2832/7308 [33:11<47:18,  1.58it/s]
 39%|███▉      | 2833/7308 [33:12<47:17,  1.58it/s]
 39%|███▉      | 2834/7308 [33:13<47:15,  1.58it/s]
 39%|███▉      | 2835/7308 [33:13<47:15,  1.58it/s]
 39%|███▉      | 2836/7308 [33:14<47:14,  1.58it/s]
 39%|███▉      | 2837/7308 [33:14<47:13,  1.58it/s]
 39%|███▉      | 2838/7308 [33:15<47:13,  1.58it/s]
 39%|███▉      | 2839/7308 [33:16<47:12,  1.58it/s]
 39%|███▉      | 2840/7308 [33:16<47:11,  1.58it/s]
                                                   
{'loss': '3.907', 'grad_norm': '0.3151', 'learning_rate': '0.0006', 'epoch': '78.9'}
+
 39%|███▉      | 2840/7308 [33:16<47:11,  1.58it/s]
 39%|███▉      | 2841/7308 [33:17<47:10,  1.58it/s]
 39%|███▉      | 2842/7308 [33:18<47:10,  1.58it/s]
 39%|███▉      | 2843/7308 [33:18<47:08,  1.58it/s]
 39%|███▉      | 2844/7308 [33:18<38:25,  1.94it/s]
 39%|███▉      | 2845/7308 [33:19<42:53,  1.73it/s]
 39%|███▉      | 2846/7308 [33:20<44:09,  1.68it/s]
 39%|███▉      | 2847/7308 [33:20<45:02,  1.65it/s]
 39%|███▉      | 2848/7308 [33:21<45:39,  1.63it/s]
 39%|███▉      | 2849/7308 [33:22<46:05,  1.61it/s]
 39%|███▉      | 2850/7308 [33:22<46:24,  1.60it/s]
                                                   
{'loss': '3.834', 'grad_norm': '0.3585', 'learning_rate': '0.0006', 'epoch': '79.17'}
+
 39%|███▉      | 2850/7308 [33:22<46:24,  1.60it/s]
 39%|███▉      | 2851/7308 [33:23<46:37,  1.59it/s]
 39%|███▉      | 2852/7308 [33:24<46:45,  1.59it/s]
 39%|███▉      | 2853/7308 [33:24<46:50,  1.59it/s]
 39%|███▉      | 2854/7308 [33:25<46:55,  1.58it/s]
 39%|███▉      | 2855/7308 [33:26<46:57,  1.58it/s]
 39%|███▉      | 2856/7308 [33:26<46:59,  1.58it/s]
 39%|███▉      | 2857/7308 [33:27<47:00,  1.58it/s]
 39%|███▉      | 2858/7308 [33:27<47:00,  1.58it/s]
 39%|███▉      | 2859/7308 [33:28<47:03,  1.58it/s]
 39%|███▉      | 2860/7308 [33:29<47:02,  1.58it/s]
                                                   
{'loss': '3.817', 'grad_norm': '0.3458', 'learning_rate': '0.0006', 'epoch': '79.45'}
+
 39%|███▉      | 2860/7308 [33:29<47:02,  1.58it/s]
 39%|███▉      | 2861/7308 [33:29<47:02,  1.58it/s]
 39%|███▉      | 2862/7308 [33:30<47:03,  1.57it/s]
 39%|███▉      | 2863/7308 [33:31<47:02,  1.58it/s]
 39%|███▉      | 2864/7308 [33:31<47:01,  1.58it/s]
 39%|███▉      | 2865/7308 [33:32<47:00,  1.58it/s]
 39%|███▉      | 2866/7308 [33:32<46:58,  1.58it/s]
 39%|███▉      | 2867/7308 [33:33<46:57,  1.58it/s]
 39%|███▉      | 2868/7308 [33:34<46:57,  1.58it/s]
 39%|███▉      | 2869/7308 [33:34<46:56,  1.58it/s]
 39%|███▉      | 2870/7308 [33:35<46:55,  1.58it/s]
                                                   
{'loss': '3.839', 'grad_norm': '0.2807', 'learning_rate': '0.0006', 'epoch': '79.73'}
+
 39%|███▉      | 2870/7308 [33:35<46:55,  1.58it/s]
 39%|███▉      | 2871/7308 [33:36<46:55,  1.58it/s]
 39%|███▉      | 2872/7308 [33:36<46:54,  1.58it/s]
 39%|███▉      | 2873/7308 [33:37<46:54,  1.58it/s]
 39%|███▉      | 2874/7308 [33:38<46:53,  1.58it/s]
 39%|███▉      | 2875/7308 [33:38<46:53,  1.58it/s]
 39%|███▉      | 2876/7308 [33:39<46:52,  1.58it/s]
 39%|███▉      | 2877/7308 [33:39<46:51,  1.58it/s]
 39%|███▉      | 2878/7308 [33:40<46:51,  1.58it/s]
 39%|███▉      | 2879/7308 [33:41<46:49,  1.58it/s]
 39%|███▉      | 2880/7308 [33:41<38:10,  1.93it/s]
                                                   
{'loss': '3.872', 'grad_norm': '0.3789', 'learning_rate': '0.0006', 'epoch': '80'}
+
 39%|███▉      | 2880/7308 [33:41<38:10,  1.93it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 22.79it/s]
+
 15%|█▌        | 6/39 [00:00<00:01, 17.55it/s]
+
 21%|██        | 8/39 [00:00<00:01, 16.64it/s]
+
 26%|██▌       | 10/39 [00:00<00:01, 16.10it/s]
+
 31%|███       | 12/39 [00:00<00:01, 15.79it/s]
+
 36%|███▌      | 14/39 [00:00<00:01, 15.61it/s]
+
 41%|████      | 16/39 [00:00<00:01, 15.48it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 15.39it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 15.33it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 15.28it/s]
+
 62%|██████▏   | 24/39 [00:01<00:00, 15.27it/s]
+
 67%|██████▋   | 26/39 [00:01<00:00, 15.25it/s]
+
 72%|███████▏  | 28/39 [00:01<00:00, 15.24it/s]
+
 77%|███████▋  | 30/39 [00:01<00:00, 15.24it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 15.23it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 15.22it/s]
+
 92%|█████████▏| 36/39 [00:02<00:00, 15.22it/s]
+
 97%|█████████▋| 38/39 [00:02<00:00, 15.21it/s]
                                                   
+
                                               
{'eval_loss': '4.28', 'eval_runtime': '2.629', 'eval_samples_per_second': '928.6', 'eval_steps_per_second': '14.84', 'epoch': '80'}
+
 39%|███▉      | 2880/7308 [33:44<38:10,  1.93it/s]
+
100%|██████████| 39/39 [00:02<00:00, 15.21it/s][eval] epoch 80.0 step 2880 loss 4.2801 (best 4.0798 @ ep 46.0, 17/20 since best)
+
+
                                               
 39%|███▉      | 2881/7308 [33:44<1:41:10,  1.37s/it]
 39%|███▉      | 2882/7308 [33:45<1:24:49,  1.15s/it]
 39%|███▉      | 2883/7308 [33:46<1:13:24,  1.00it/s]
 39%|███▉      | 2884/7308 [33:46<1:05:24,  1.13it/s]
 39%|███▉      | 2885/7308 [33:47<59:48,  1.23it/s]  
 39%|███▉      | 2886/7308 [33:48<55:53,  1.32it/s]
 40%|███▉      | 2887/7308 [33:48<53:08,  1.39it/s]
 40%|███▉      | 2888/7308 [33:49<51:12,  1.44it/s]
 40%|███▉      | 2889/7308 [33:49<49:50,  1.48it/s]
 40%|███▉      | 2890/7308 [33:50<48:52,  1.51it/s]
                                                   
{'loss': '3.851', 'grad_norm': '0.3181', 'learning_rate': '0.0006', 'epoch': '80.28'}
+
 40%|███▉      | 2890/7308 [33:50<48:52,  1.51it/s]
 40%|███▉      | 2891/7308 [33:51<48:13,  1.53it/s]
 40%|███▉      | 2892/7308 [33:51<47:44,  1.54it/s]
 40%|███▉      | 2893/7308 [33:52<47:24,  1.55it/s]
 40%|███▉      | 2894/7308 [33:53<47:11,  1.56it/s]
 40%|███▉      | 2895/7308 [33:53<47:00,  1.56it/s]
 40%|███▉      | 2896/7308 [33:54<46:52,  1.57it/s]
 40%|███▉      | 2897/7308 [33:55<46:47,  1.57it/s]
 40%|███▉      | 2898/7308 [33:55<46:42,  1.57it/s]
 40%|███▉      | 2899/7308 [33:56<46:39,  1.57it/s]
 40%|███▉      | 2900/7308 [33:56<46:37,  1.58it/s]
                                                   
{'loss': '3.894', 'grad_norm': '0.3478', 'learning_rate': '0.0006', 'epoch': '80.57'}
+
 40%|███▉      | 2900/7308 [33:56<46:37,  1.58it/s]
 40%|███▉      | 2901/7308 [33:57<46:37,  1.58it/s]
 40%|███▉      | 2902/7308 [33:58<46:35,  1.58it/s]
 40%|███▉      | 2903/7308 [33:58<46:33,  1.58it/s]
 40%|███▉      | 2904/7308 [33:59<46:34,  1.58it/s]
 40%|███▉      | 2905/7308 [34:00<46:32,  1.58it/s]
 40%|███▉      | 2906/7308 [34:00<46:31,  1.58it/s]
 40%|███▉      | 2907/7308 [34:01<46:30,  1.58it/s]
 40%|███▉      | 2908/7308 [34:01<46:28,  1.58it/s]
 40%|███▉      | 2909/7308 [34:02<46:27,  1.58it/s]
 40%|███▉      | 2910/7308 [34:03<46:27,  1.58it/s]
                                                   
{'loss': '3.831', 'grad_norm': '0.2945', 'learning_rate': '0.0006', 'epoch': '80.85'}
+
 40%|███▉      | 2910/7308 [34:03<46:27,  1.58it/s]
 40%|███▉      | 2911/7308 [34:03<46:27,  1.58it/s]
 40%|███▉      | 2912/7308 [34:04<46:27,  1.58it/s]
 40%|███▉      | 2913/7308 [34:05<46:26,  1.58it/s]
 40%|███▉      | 2914/7308 [34:05<46:26,  1.58it/s]
 40%|███▉      | 2915/7308 [34:06<46:24,  1.58it/s]
 40%|███▉      | 2916/7308 [34:06<37:49,  1.94it/s]
 40%|███▉      | 2917/7308 [34:07<42:09,  1.74it/s]
 40%|███▉      | 2918/7308 [34:08<43:25,  1.68it/s]
 40%|███▉      | 2919/7308 [34:08<44:19,  1.65it/s]
 40%|███▉      | 2920/7308 [34:09<44:55,  1.63it/s]
                                                   
{'loss': '3.815', 'grad_norm': '0.3634', 'learning_rate': '0.0006', 'epoch': '81.11'}
+
 40%|███▉      | 2920/7308 [34:09<44:55,  1.63it/s]
 40%|███▉      | 2921/7308 [34:09<45:22,  1.61it/s]
 40%|███▉      | 2922/7308 [34:10<45:39,  1.60it/s]
 40%|███▉      | 2923/7308 [34:11<45:52,  1.59it/s]
 40%|████      | 2924/7308 [34:11<45:59,  1.59it/s]
 40%|████      | 2925/7308 [34:12<46:05,  1.59it/s]
 40%|████      | 2926/7308 [34:13<46:09,  1.58it/s]
 40%|████      | 2927/7308 [34:13<46:14,  1.58it/s]
 40%|████      | 2928/7308 [34:14<46:15,  1.58it/s]
 40%|████      | 2929/7308 [34:14<46:15,  1.58it/s]
 40%|████      | 2930/7308 [34:15<46:15,  1.58it/s]
                                                   
{'loss': '3.841', 'grad_norm': '0.3312', 'learning_rate': '0.0006', 'epoch': '81.4'}
+
 40%|████      | 2930/7308 [34:15<46:15,  1.58it/s]
 40%|████      | 2931/7308 [34:16<46:15,  1.58it/s]
 40%|████      | 2932/7308 [34:16<46:15,  1.58it/s]
 40%|████      | 2933/7308 [34:17<46:14,  1.58it/s]
 40%|████      | 2934/7308 [34:18<46:14,  1.58it/s]
 40%|████      | 2935/7308 [34:18<46:14,  1.58it/s]
 40%|████      | 2936/7308 [34:19<46:16,  1.57it/s]
 40%|████      | 2937/7308 [34:20<46:15,  1.57it/s]
 40%|████      | 2938/7308 [34:20<46:14,  1.58it/s]
 40%|████      | 2939/7308 [34:21<46:13,  1.58it/s]
 40%|████      | 2940/7308 [34:21<46:12,  1.58it/s]
                                                   
{'loss': '3.875', 'grad_norm': '0.3289', 'learning_rate': '0.0006', 'epoch': '81.68'}
+
 40%|████      | 2940/7308 [34:21<46:12,  1.58it/s]
 40%|████      | 2941/7308 [34:22<46:13,  1.57it/s]
 40%|████      | 2942/7308 [34:23<46:11,  1.58it/s]
 40%|████      | 2943/7308 [34:23<46:11,  1.57it/s]
 40%|████      | 2944/7308 [34:24<46:10,  1.57it/s]
 40%|████      | 2945/7308 [34:25<46:09,  1.58it/s]
 40%|████      | 2946/7308 [34:25<46:07,  1.58it/s]
 40%|████      | 2947/7308 [34:26<46:07,  1.58it/s]
 40%|████      | 2948/7308 [34:27<46:07,  1.58it/s]
 40%|████      | 2949/7308 [34:27<46:07,  1.58it/s]
 40%|████      | 2950/7308 [34:28<46:05,  1.58it/s]
                                                   
{'loss': '3.867', 'grad_norm': '0.3395', 'learning_rate': '0.0006', 'epoch': '81.96'}
+
 40%|████      | 2950/7308 [34:28<46:05,  1.58it/s]
 40%|████      | 2951/7308 [34:28<46:04,  1.58it/s]
 40%|████      | 2952/7308 [34:29<37:33,  1.93it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 23.77it/s]
+
 15%|█▌        | 6/39 [00:00<00:01, 17.84it/s]
+
 21%|██        | 8/39 [00:00<00:01, 16.70it/s]
+
 26%|██▌       | 10/39 [00:00<00:01, 16.11it/s]
+
 31%|███       | 12/39 [00:00<00:01, 15.80it/s]
+
 36%|███▌      | 14/39 [00:00<00:01, 15.58it/s]
+
 41%|████      | 16/39 [00:00<00:01, 15.44it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 15.37it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 15.32it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 15.27it/s]
+
 62%|██████▏   | 24/39 [00:01<00:00, 15.28it/s]
+
 67%|██████▋   | 26/39 [00:01<00:00, 15.26it/s]
+
 72%|███████▏  | 28/39 [00:01<00:00, 15.25it/s]
+
 77%|███████▋  | 30/39 [00:01<00:00, 15.24it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 15.21it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 15.23it/s]
+
 92%|█████████▏| 36/39 [00:02<00:00, 15.22it/s]
+
 97%|█████████▋| 38/39 [00:02<00:00, 15.22it/s]
                                                   
+
                                               
{'eval_loss': '4.288', 'eval_runtime': '2.617', 'eval_samples_per_second': '932.8', 'eval_steps_per_second': '14.9', 'epoch': '82'}
+
 40%|████      | 2952/7308 [34:31<37:33,  1.93it/s]
+
100%|██████████| 39/39 [00:02<00:00, 15.22it/s][eval] epoch 82.0 step 2952 loss 4.2878 (best 4.0798 @ ep 46.0, 18/20 since best)
+
+
                                               
 40%|████      | 2953/7308 [34:32<1:39:14,  1.37s/it]
 40%|████      | 2954/7308 [34:33<1:23:15,  1.15s/it]
 40%|████      | 2955/7308 [34:33<1:12:05,  1.01it/s]
 40%|████      | 2956/7308 [34:34<1:04:15,  1.13it/s]
 40%|████      | 2957/7308 [34:35<58:47,  1.23it/s]  
 40%|████      | 2958/7308 [34:35<54:56,  1.32it/s]
 40%|████      | 2959/7308 [34:36<52:14,  1.39it/s]
 41%|████      | 2960/7308 [34:36<50:22,  1.44it/s]
                                                   
{'loss': '3.828', 'grad_norm': '0.3637', 'learning_rate': '0.0006', 'epoch': '82.23'}
+
 41%|████      | 2960/7308 [34:36<50:22,  1.44it/s]
 41%|████      | 2961/7308 [34:37<49:05,  1.48it/s]
 41%|████      | 2962/7308 [34:38<48:08,  1.50it/s]
 41%|████      | 2963/7308 [34:38<47:28,  1.53it/s]
 41%|████      | 2964/7308 [34:39<47:01,  1.54it/s]
 41%|████      | 2965/7308 [34:40<46:40,  1.55it/s]
 41%|████      | 2966/7308 [34:40<46:26,  1.56it/s]
 41%|████      | 2967/7308 [34:41<46:16,  1.56it/s]
 41%|████      | 2968/7308 [34:42<46:08,  1.57it/s]
 41%|████      | 2969/7308 [34:42<46:03,  1.57it/s]
 41%|████      | 2970/7308 [34:43<45:58,  1.57it/s]
                                                   
{'loss': '3.876', 'grad_norm': '0.3166', 'learning_rate': '0.0006', 'epoch': '82.51'}
+
 41%|████      | 2970/7308 [34:43<45:58,  1.57it/s]
 41%|████      | 2971/7308 [34:43<45:57,  1.57it/s]
 41%|████      | 2972/7308 [34:44<45:56,  1.57it/s]
 41%|████      | 2973/7308 [34:45<45:54,  1.57it/s]
 41%|████      | 2974/7308 [34:45<45:52,  1.57it/s]
 41%|████      | 2975/7308 [34:46<45:51,  1.57it/s]
 41%|████      | 2976/7308 [34:47<45:50,  1.57it/s]
 41%|████      | 2977/7308 [34:47<45:48,  1.58it/s]
 41%|████      | 2978/7308 [34:48<45:48,  1.58it/s]
 41%|████      | 2979/7308 [34:49<45:48,  1.58it/s]
 41%|████      | 2980/7308 [34:49<45:47,  1.58it/s]
                                                   
{'loss': '3.838', 'grad_norm': '0.3504', 'learning_rate': '0.0006', 'epoch': '82.79'}
+
 41%|████      | 2980/7308 [34:49<45:47,  1.58it/s]
 41%|████      | 2981/7308 [34:50<45:46,  1.58it/s]
 41%|████      | 2982/7308 [34:50<45:45,  1.58it/s]
 41%|████      | 2983/7308 [34:51<45:45,  1.58it/s]
 41%|████      | 2984/7308 [34:52<45:43,  1.58it/s]
 41%|████      | 2985/7308 [34:52<45:42,  1.58it/s]
 41%|████      | 2986/7308 [34:53<45:42,  1.58it/s]
 41%|████      | 2987/7308 [34:54<45:41,  1.58it/s]
 41%|████      | 2988/7308 [34:54<37:13,  1.93it/s]
 41%|████      | 2989/7308 [34:55<41:30,  1.73it/s]
 41%|████      | 2990/7308 [34:55<42:45,  1.68it/s]
                                                   
{'loss': '3.853', 'grad_norm': '0.3712', 'learning_rate': '0.0006', 'epoch': '83.06'}
+
 41%|████      | 2990/7308 [34:55<42:45,  1.68it/s]
 41%|████      | 2991/7308 [34:56<43:37,  1.65it/s]
 41%|████      | 2992/7308 [34:56<44:13,  1.63it/s]
 41%|████      | 2993/7308 [34:57<44:39,  1.61it/s]
 41%|████      | 2994/7308 [34:58<44:56,  1.60it/s]
 41%|████      | 2995/7308 [34:58<45:09,  1.59it/s]
 41%|████      | 2996/7308 [34:59<45:17,  1.59it/s]
 41%|████      | 2997/7308 [35:00<45:22,  1.58it/s]
 41%|████      | 2998/7308 [35:00<45:26,  1.58it/s]
 41%|████      | 2999/7308 [35:01<45:28,  1.58it/s]
 41%|████      | 3000/7308 [35:02<45:29,  1.58it/s]
                                                   
{'loss': '3.755', 'grad_norm': '0.453', 'learning_rate': '0.0006', 'epoch': '83.34'}
+
 41%|████      | 3000/7308 [35:02<45:29,  1.58it/s]
 41%|████      | 3001/7308 [35:02<45:30,  1.58it/s]
 41%|████      | 3002/7308 [35:03<45:30,  1.58it/s]
 41%|████      | 3003/7308 [35:03<45:30,  1.58it/s]
 41%|████      | 3004/7308 [35:04<45:30,  1.58it/s]
 41%|████      | 3005/7308 [35:05<45:29,  1.58it/s]
 41%|████      | 3006/7308 [35:05<45:28,  1.58it/s]
 41%|████      | 3007/7308 [35:06<45:28,  1.58it/s]
 41%|████      | 3008/7308 [35:07<45:27,  1.58it/s]
 41%|████      | 3009/7308 [35:07<45:27,  1.58it/s]
 41%|████      | 3010/7308 [35:08<45:25,  1.58it/s]
                                                   
{'loss': '3.811', 'grad_norm': '0.3487', 'learning_rate': '0.0006', 'epoch': '83.62'}
+
 41%|████      | 3010/7308 [35:08<45:25,  1.58it/s]
 41%|████      | 3011/7308 [35:09<45:28,  1.58it/s]
 41%|████      | 3012/7308 [35:09<45:28,  1.57it/s]
 41%|████      | 3013/7308 [35:10<45:26,  1.58it/s]
 41%|████      | 3014/7308 [35:10<45:25,  1.58it/s]
 41%|████▏     | 3015/7308 [35:11<45:25,  1.58it/s]
 41%|████▏     | 3016/7308 [35:12<45:23,  1.58it/s]
 41%|████▏     | 3017/7308 [35:12<45:22,  1.58it/s]
 41%|████▏     | 3018/7308 [35:13<45:22,  1.58it/s]
 41%|████▏     | 3019/7308 [35:14<45:20,  1.58it/s]
 41%|████▏     | 3020/7308 [35:14<45:20,  1.58it/s]
                                                   
{'loss': '3.879', 'grad_norm': '0.3054', 'learning_rate': '0.0006', 'epoch': '83.9'}
+
 41%|████▏     | 3020/7308 [35:14<45:20,  1.58it/s]
 41%|████▏     | 3021/7308 [35:15<45:21,  1.58it/s]
 41%|████▏     | 3022/7308 [35:16<45:20,  1.58it/s]
 41%|████▏     | 3023/7308 [35:16<45:17,  1.58it/s]
 41%|████▏     | 3024/7308 [35:16<36:54,  1.93it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 23.10it/s]
+
 15%|█▌        | 6/39 [00:00<00:01, 17.66it/s]
+
 21%|██        | 8/39 [00:00<00:01, 16.70it/s]
+
 26%|██▌       | 10/39 [00:00<00:01, 16.16it/s]
+
 31%|███       | 12/39 [00:00<00:01, 15.83it/s]
+
 36%|███▌      | 14/39 [00:00<00:01, 15.62it/s]
+
 41%|████      | 16/39 [00:00<00:01, 15.49it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 15.39it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 15.34it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 15.28it/s]
+
 62%|██████▏   | 24/39 [00:01<00:00, 15.28it/s]
+
 67%|██████▋   | 26/39 [00:01<00:00, 15.26it/s]
+
 72%|███████▏  | 28/39 [00:01<00:00, 15.25it/s]
+
 77%|███████▋  | 30/39 [00:01<00:00, 15.23it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 15.24it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 15.22it/s]
+
 92%|█████████▏| 36/39 [00:02<00:00, 15.23it/s]
+
 97%|█████████▋| 38/39 [00:02<00:00, 15.22it/s]
                                                   
+
                                               
{'eval_loss': '4.301', 'eval_runtime': '2.621', 'eval_samples_per_second': '931.3', 'eval_steps_per_second': '14.88', 'epoch': '84'}
+
 41%|████▏     | 3024/7308 [35:19<36:54,  1.93it/s]
+
100%|██████████| 39/39 [00:02<00:00, 15.22it/s][eval] epoch 84.0 step 3024 loss 4.3011 (best 4.0798 @ ep 46.0, 19/20 since best)
+
+
                                               
 41%|████▏     | 3025/7308 [35:20<1:36:57,  1.36s/it]
 41%|████▏     | 3026/7308 [35:20<1:21:26,  1.14s/it]
 41%|████▏     | 3027/7308 [35:21<1:10:35,  1.01it/s]
 41%|████▏     | 3028/7308 [35:22<1:02:58,  1.13it/s]
 41%|████▏     | 3029/7308 [35:22<57:38,  1.24it/s]  
 41%|████▏     | 3030/7308 [35:23<53:54,  1.32it/s]
                                                   
{'loss': '3.887', 'grad_norm': '0.3738', 'learning_rate': '0.0006', 'epoch': '84.17'}
+
 41%|████▏     | 3030/7308 [35:23<53:54,  1.32it/s]
 41%|████▏     | 3031/7308 [35:24<51:17,  1.39it/s]
 41%|████▏     | 3032/7308 [35:24<49:28,  1.44it/s]
 42%|████▏     | 3033/7308 [35:25<48:11,  1.48it/s]
 42%|████▏     | 3034/7308 [35:25<47:16,  1.51it/s]
 42%|████▏     | 3035/7308 [35:26<46:38,  1.53it/s]
 42%|████▏     | 3036/7308 [35:27<46:11,  1.54it/s]
 42%|████▏     | 3037/7308 [35:27<45:52,  1.55it/s]
 42%|████▏     | 3038/7308 [35:28<45:40,  1.56it/s]
 42%|████▏     | 3039/7308 [35:29<45:30,  1.56it/s]
 42%|████▏     | 3040/7308 [35:29<45:23,  1.57it/s]
                                                   
{'loss': '3.76', 'grad_norm': '0.4268', 'learning_rate': '0.0006', 'epoch': '84.45'}
+
 42%|████▏     | 3040/7308 [35:29<45:23,  1.57it/s]
 42%|████▏     | 3041/7308 [35:30<45:18,  1.57it/s]
 42%|████▏     | 3042/7308 [35:31<45:15,  1.57it/s]
 42%|████▏     | 3043/7308 [35:31<45:12,  1.57it/s]
 42%|████▏     | 3044/7308 [35:32<45:11,  1.57it/s]
 42%|████▏     | 3045/7308 [35:32<45:09,  1.57it/s]
 42%|████▏     | 3046/7308 [35:33<45:07,  1.57it/s]
 42%|████▏     | 3047/7308 [35:34<45:05,  1.57it/s]
 42%|████▏     | 3048/7308 [35:34<45:04,  1.57it/s]
 42%|████▏     | 3049/7308 [35:35<45:03,  1.58it/s]
 42%|████▏     | 3050/7308 [35:36<45:02,  1.58it/s]
                                                   
{'loss': '3.789', 'grad_norm': '0.3369', 'learning_rate': '0.0006', 'epoch': '84.73'}
+
 42%|████▏     | 3050/7308 [35:36<45:02,  1.58it/s]
 42%|████▏     | 3051/7308 [35:36<45:03,  1.57it/s]
 42%|████▏     | 3052/7308 [35:37<45:02,  1.58it/s]
 42%|████▏     | 3053/7308 [35:37<45:01,  1.58it/s]
 42%|████▏     | 3054/7308 [35:38<45:00,  1.58it/s]
 42%|████▏     | 3055/7308 [35:39<45:00,  1.58it/s]
 42%|████▏     | 3056/7308 [35:39<44:59,  1.58it/s]
 42%|████▏     | 3057/7308 [35:40<44:57,  1.58it/s]
 42%|████▏     | 3058/7308 [35:41<44:56,  1.58it/s]
 42%|████▏     | 3059/7308 [35:41<44:55,  1.58it/s]
 42%|████▏     | 3060/7308 [35:42<36:36,  1.93it/s]
                                                   
{'loss': '3.821', 'grad_norm': '0.3798', 'learning_rate': '0.0006', 'epoch': '85'}
+
 42%|████▏     | 3060/7308 [35:42<36:36,  1.93it/s]
 42%|████▏     | 3061/7308 [35:42<40:38,  1.74it/s]
 42%|████▏     | 3062/7308 [35:43<41:54,  1.69it/s]
 42%|████▏     | 3063/7308 [35:44<42:47,  1.65it/s]
 42%|████▏     | 3064/7308 [35:44<43:24,  1.63it/s]
 42%|████▏     | 3065/7308 [35:45<43:51,  1.61it/s]
 42%|████▏     | 3066/7308 [35:45<44:09,  1.60it/s]
 42%|████▏     | 3067/7308 [35:46<44:21,  1.59it/s]
 42%|████▏     | 3068/7308 [35:47<44:30,  1.59it/s]
 42%|████▏     | 3069/7308 [35:47<44:36,  1.58it/s]
 42%|████▏     | 3070/7308 [35:48<44:41,  1.58it/s]
                                                   
{'loss': '3.759', 'grad_norm': '0.3441', 'learning_rate': '0.0006', 'epoch': '85.28'}
+
 42%|████▏     | 3070/7308 [35:48<44:41,  1.58it/s]
 42%|████▏     | 3071/7308 [35:49<44:46,  1.58it/s]
 42%|████▏     | 3072/7308 [35:49<44:47,  1.58it/s]
 42%|████▏     | 3073/7308 [35:50<44:48,  1.58it/s]
 42%|████▏     | 3074/7308 [35:51<44:49,  1.57it/s]
 42%|████▏     | 3075/7308 [35:51<44:50,  1.57it/s]
 42%|████▏     | 3076/7308 [35:52<44:50,  1.57it/s]
 42%|████▏     | 3077/7308 [35:52<44:50,  1.57it/s]
 42%|████▏     | 3078/7308 [35:53<44:49,  1.57it/s]
 42%|████▏     | 3079/7308 [35:54<44:49,  1.57it/s]
 42%|████▏     | 3080/7308 [35:54<44:49,  1.57it/s]
                                                   
{'loss': '3.717', 'grad_norm': '0.3191', 'learning_rate': '0.0006', 'epoch': '85.57'}
+
 42%|████▏     | 3080/7308 [35:54<44:49,  1.57it/s]
 42%|████▏     | 3081/7308 [35:55<44:49,  1.57it/s]
 42%|████▏     | 3082/7308 [35:56<44:48,  1.57it/s]
 42%|████▏     | 3083/7308 [35:56<44:48,  1.57it/s]
 42%|████▏     | 3084/7308 [35:57<44:47,  1.57it/s]
 42%|████▏     | 3085/7308 [35:58<44:46,  1.57it/s]
 42%|████▏     | 3086/7308 [35:58<44:46,  1.57it/s]
 42%|████▏     | 3087/7308 [35:59<44:45,  1.57it/s]
 42%|████▏     | 3088/7308 [35:59<44:44,  1.57it/s]
 42%|████▏     | 3089/7308 [36:00<44:43,  1.57it/s]
 42%|████▏     | 3090/7308 [36:01<44:42,  1.57it/s]
                                                   
{'loss': '3.837', 'grad_norm': '0.3044', 'learning_rate': '0.0006', 'epoch': '85.85'}
+
 42%|████▏     | 3090/7308 [36:01<44:42,  1.57it/s]
 42%|████▏     | 3091/7308 [36:01<44:42,  1.57it/s]
 42%|████▏     | 3092/7308 [36:02<44:41,  1.57it/s]
 42%|████▏     | 3093/7308 [36:03<44:41,  1.57it/s]
 42%|████▏     | 3094/7308 [36:03<44:41,  1.57it/s]
 42%|████▏     | 3095/7308 [36:04<44:38,  1.57it/s]
 42%|████▏     | 3096/7308 [36:04<36:22,  1.93it/s][transformers] 
+***** Running Evaluation *****
+[transformers]   Num examples = 2441
+[transformers]   Batch size = 8
+
+
  0%|          | 0/39 [00:00<?, ?it/s]
+
  8%|▊         | 3/39 [00:00<00:01, 23.32it/s]
+
 15%|█▌        | 6/39 [00:00<00:01, 17.72it/s]
+
 21%|██        | 8/39 [00:00<00:01, 16.70it/s]
+
 26%|██▌       | 10/39 [00:00<00:01, 16.18it/s]
+
 31%|███       | 12/39 [00:00<00:01, 15.81it/s]
+
 36%|███▌      | 14/39 [00:00<00:01, 15.64it/s]
+
 41%|████      | 16/39 [00:00<00:01, 15.48it/s]
+
 46%|████▌     | 18/39 [00:01<00:01, 15.39it/s]
+
 51%|█████▏    | 20/39 [00:01<00:01, 15.32it/s]
+
 56%|█████▋    | 22/39 [00:01<00:01, 15.29it/s]
+
 62%|██████▏   | 24/39 [00:01<00:00, 15.27it/s]
+
 67%|██████▋   | 26/39 [00:01<00:00, 15.26it/s]
+
 72%|███████▏  | 28/39 [00:01<00:00, 15.25it/s]
+
 77%|███████▋  | 30/39 [00:01<00:00, 15.24it/s]
+
 82%|████████▏ | 32/39 [00:02<00:00, 15.22it/s]
+
 87%|████████▋ | 34/39 [00:02<00:00, 15.21it/s]
+
 92%|█████████▏| 36/39 [00:02<00:00, 15.21it/s]
+
 97%|█████████▋| 38/39 [00:02<00:00, 15.22it/s]
                                                   
+
                                               
{'eval_loss': '4.309', 'eval_runtime': '2.622', 'eval_samples_per_second': '931', 'eval_steps_per_second': '14.87', 'epoch': '86'}
+
 42%|████▏     | 3096/7308 [36:07<36:22,  1.93it/s]
+
100%|██████████| 39/39 [00:02<00:00, 15.22it/s][eval] epoch 86.0 step 3096 loss 4.3091 (best 4.0798 @ ep 46.0, 20/20 since best)
+
+
                                               [transformers] 
+
+Training completed. Do not forget to share your model on huggingface.co/models =)
+
+
+
                                                   
{'train_runtime': '2169', 'train_samples_per_second': '1691', 'train_steps_per_second': '3.369', 'train_loss': '4.705', 'epoch': '86'}
+
 42%|████▏     | 3096/7308 [36:07<36:22,  1.93it/s]
 42%|████▏     | 3096/7308 [36:07<49:08,  1.43it/s]
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+wandb: 
+wandb: 🚀 View run t2a_37M_C2_r2l50_i5exp at: https://wandb.ai/slowrun-wankers/data-aug-pretraining/runs/oo3nkqdx
+wandb: Find logs at: wandb/run-20260818_200623-oo3nkqdx/logs
+/home/ubuntu/.local/share/uv/python/cpython-3.12.13-linux-x86_64-gnu/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 8 leaked semaphore objects to clean up at shutdown
+  warnings.warn('resource_tracker: There appear to be %d '
+[rank0]:[W818 20:43:19.867262035 ProcessGroupNCCL.cpp:1575] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator())