diff --git a/.gitattributes b/.gitattributes
index a6344aac8c09253b3b630fb776ae94478aa0275b..c68e018dd309b4926322e1acfd559e2eaed4bd45 100644
--- a/.gitattributes
+++ b/.gitattributes
@@ -33,3 +33,15 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
*.zip filter=lfs diff=lfs merge=lfs -text
*.zst filter=lfs diff=lfs merge=lfs -text
*tfevents* filter=lfs diff=lfs merge=lfs -text
+qwen2.5-1.5b-instruct/checkpoint-142/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+qwen2.5-1.5b-instruct/checkpoint-213/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+qwen2.5-1.5b-instruct/checkpoint-71/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+qwen2.5-1.5b-instruct/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+qwen2.5-3b-instruct/checkpoint-142/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+qwen2.5-3b-instruct/checkpoint-213/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+qwen2.5-3b-instruct/checkpoint-71/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+qwen2.5-3b-instruct/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+qwen2.5-7b-instruct/checkpoint-142/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+qwen2.5-7b-instruct/checkpoint-213/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+qwen2.5-7b-instruct/checkpoint-71/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+qwen2.5-7b-instruct/tokenizer.json filter=lfs diff=lfs merge=lfs -text
diff --git a/qwen2.5-1.5b-instruct/README.md b/qwen2.5-1.5b-instruct/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..f44d3f62e28ffb6a458d18b69d4035231e6bd823
--- /dev/null
+++ b/qwen2.5-1.5b-instruct/README.md
@@ -0,0 +1,69 @@
+---
+library_name: transformers
+license: other
+base_model: Qwen/Qwen2.5-1.5B-Instruct
+tags:
+- llama-factory
+- full
+- generated_from_trainer
+model-index:
+- name: qwen2.5-1.5b-instruct
+ results: []
+---
+
+
+
+# qwen2.5-1.5b-instruct
+
+This model is a fine-tuned version of [Qwen/Qwen2.5-1.5B-Instruct](https://huggingface.co/Qwen/Qwen2.5-1.5B-Instruct) on the wealth_train_20k_no_trace_generation dataset.
+It achieves the following results on the evaluation set:
+- Loss: 2.2276
+
+## Model description
+
+More information needed
+
+## Intended uses & limitations
+
+More information needed
+
+## Training and evaluation data
+
+More information needed
+
+## Training procedure
+
+### Training hyperparameters
+
+The following hyperparameters were used during training:
+- learning_rate: 1e-05
+- train_batch_size: 1
+- eval_batch_size: 1
+- seed: 42
+- distributed_type: multi-GPU
+- num_devices: 4
+- gradient_accumulation_steps: 128
+- total_train_batch_size: 512
+- total_eval_batch_size: 4
+- optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
+- lr_scheduler_type: cosine
+- lr_scheduler_warmup_steps: 0.1
+- num_epochs: 3
+
+### Training results
+
+| Training Loss | Epoch | Step | Validation Loss |
+|:-------------:|:------:|:----:|:---------------:|
+| 0.4606 | 0.7111 | 50 | 0.4878 |
+| 0.4223 | 1.4124 | 100 | 0.4897 |
+| 0.3581 | 2.1138 | 150 | 0.4923 |
+| 0.3697 | 2.8249 | 200 | 0.4990 |
+
+
+### Framework versions
+
+- Transformers 5.2.0
+- Pytorch 2.11.0+cu128
+- Datasets 4.0.0
+- Tokenizers 0.22.2
diff --git a/qwen2.5-1.5b-instruct/all_results.json b/qwen2.5-1.5b-instruct/all_results.json
new file mode 100644
index 0000000000000000000000000000000000000000..8aadaff452c83c6e2fcb9dbe7bb4f720325498f7
--- /dev/null
+++ b/qwen2.5-1.5b-instruct/all_results.json
@@ -0,0 +1,12 @@
+{
+ "epoch": 3.0,
+ "eval_loss": 2.227557897567749,
+ "eval_runtime": 8.5719,
+ "eval_samples_per_second": 23.332,
+ "eval_steps_per_second": 5.833,
+ "total_flos": 4.868007593928294e+16,
+ "train_loss": 0.0,
+ "train_runtime": 3.3055,
+ "train_samples_per_second": 16336.51,
+ "train_steps_per_second": 32.673
+}
\ No newline at end of file
diff --git a/qwen2.5-1.5b-instruct/chat_template.jinja b/qwen2.5-1.5b-instruct/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..bdf7919a96cfe43d50914a007b9c0877bd0ec27e
--- /dev/null
+++ b/qwen2.5-1.5b-instruct/chat_template.jinja
@@ -0,0 +1,54 @@
+{%- if tools %}
+ {{- '<|im_start|>system\n' }}
+ {%- if messages[0]['role'] == 'system' %}
+ {{- messages[0]['content'] }}
+ {%- else %}
+ {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+ {%- endif %}
+ {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within XML tags:\n" }}
+ {%- for tool in tools %}
+ {{- "\n" }}
+ {{- tool | tojson }}
+ {%- endfor %}
+ {{- "\n\n\nFor each function call, return a json object with function name and arguments within XML tags:\n\n{\"name\": , \"arguments\": }\n<|im_end|>\n" }}
+{%- else %}
+ {%- if messages[0]['role'] == 'system' %}
+ {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+ {%- else %}
+ {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+ {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+ {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+ {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+ {%- elif message.role == "assistant" %}
+ {{- '<|im_start|>' + message.role }}
+ {%- if message.content %}
+ {{- '\n' + message.content }}
+ {%- endif %}
+ {%- for tool_call in message.tool_calls %}
+ {%- if tool_call.function is defined %}
+ {%- set tool_call = tool_call.function %}
+ {%- endif %}
+ {{- '\n\n{"name": "' }}
+ {{- tool_call.name }}
+ {{- '", "arguments": ' }}
+ {{- tool_call.arguments | tojson }}
+ {{- '}\n' }}
+ {%- endfor %}
+ {{- '<|im_end|>\n' }}
+ {%- elif message.role == "tool" %}
+ {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+ {{- '<|im_start|>user' }}
+ {%- endif %}
+ {{- '\n\n' }}
+ {{- message.content }}
+ {{- '\n' }}
+ {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+ {{- '<|im_end|>\n' }}
+ {%- endif %}
+ {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+ {{- '<|im_start|>assistant\n' }}
+{%- endif %}
diff --git a/qwen2.5-1.5b-instruct/checkpoint-142/chat_template.jinja b/qwen2.5-1.5b-instruct/checkpoint-142/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..bdf7919a96cfe43d50914a007b9c0877bd0ec27e
--- /dev/null
+++ b/qwen2.5-1.5b-instruct/checkpoint-142/chat_template.jinja
@@ -0,0 +1,54 @@
+{%- if tools %}
+ {{- '<|im_start|>system\n' }}
+ {%- if messages[0]['role'] == 'system' %}
+ {{- messages[0]['content'] }}
+ {%- else %}
+ {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+ {%- endif %}
+ {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within XML tags:\n" }}
+ {%- for tool in tools %}
+ {{- "\n" }}
+ {{- tool | tojson }}
+ {%- endfor %}
+ {{- "\n\n\nFor each function call, return a json object with function name and arguments within XML tags:\n\n{\"name\": , \"arguments\": }\n<|im_end|>\n" }}
+{%- else %}
+ {%- if messages[0]['role'] == 'system' %}
+ {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+ {%- else %}
+ {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+ {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+ {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+ {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+ {%- elif message.role == "assistant" %}
+ {{- '<|im_start|>' + message.role }}
+ {%- if message.content %}
+ {{- '\n' + message.content }}
+ {%- endif %}
+ {%- for tool_call in message.tool_calls %}
+ {%- if tool_call.function is defined %}
+ {%- set tool_call = tool_call.function %}
+ {%- endif %}
+ {{- '\n\n{"name": "' }}
+ {{- tool_call.name }}
+ {{- '", "arguments": ' }}
+ {{- tool_call.arguments | tojson }}
+ {{- '}\n' }}
+ {%- endfor %}
+ {{- '<|im_end|>\n' }}
+ {%- elif message.role == "tool" %}
+ {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+ {{- '<|im_start|>user' }}
+ {%- endif %}
+ {{- '\n\n' }}
+ {{- message.content }}
+ {{- '\n' }}
+ {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+ {{- '<|im_end|>\n' }}
+ {%- endif %}
+ {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+ {{- '<|im_start|>assistant\n' }}
+{%- endif %}
diff --git a/qwen2.5-1.5b-instruct/checkpoint-142/config.json b/qwen2.5-1.5b-instruct/checkpoint-142/config.json
new file mode 100644
index 0000000000000000000000000000000000000000..499065fd744a64fc8ad2884da26e85a6e99d468a
--- /dev/null
+++ b/qwen2.5-1.5b-instruct/checkpoint-142/config.json
@@ -0,0 +1,61 @@
+{
+ "architectures": [
+ "Qwen2ForCausalLM"
+ ],
+ "attention_dropout": 0.0,
+ "bos_token_id": null,
+ "dtype": "float32",
+ "eos_token_id": 151645,
+ "hidden_act": "silu",
+ "hidden_size": 1536,
+ "initializer_range": 0.02,
+ "intermediate_size": 8960,
+ "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": 21,
+ "model_type": "qwen2",
+ "num_attention_heads": 12,
+ "num_hidden_layers": 28,
+ "num_key_value_heads": 2,
+ "pad_token_id": 151643,
+ "rms_norm_eps": 1e-06,
+ "rope_parameters": {
+ "rope_theta": 1000000.0,
+ "rope_type": "default"
+ },
+ "sliding_window": null,
+ "tie_word_embeddings": true,
+ "transformers_version": "5.2.0",
+ "use_cache": false,
+ "use_sliding_window": false,
+ "vocab_size": 151936
+}
diff --git a/qwen2.5-1.5b-instruct/checkpoint-142/generation_config.json b/qwen2.5-1.5b-instruct/checkpoint-142/generation_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..c5b4e11411fed3e7a5e6557165f888a7b0b1248b
--- /dev/null
+++ b/qwen2.5-1.5b-instruct/checkpoint-142/generation_config.json
@@ -0,0 +1,13 @@
+{
+ "do_sample": true,
+ "eos_token_id": [
+ 151645,
+ 151643
+ ],
+ "pad_token_id": 151643,
+ "repetition_penalty": 1.1,
+ "temperature": 0.7,
+ "top_k": 20,
+ "top_p": 0.8,
+ "transformers_version": "5.2.0"
+}
diff --git a/qwen2.5-1.5b-instruct/checkpoint-142/model.safetensors b/qwen2.5-1.5b-instruct/checkpoint-142/model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..314213a8a15aea2230ca3748b264a2606470c082
--- /dev/null
+++ b/qwen2.5-1.5b-instruct/checkpoint-142/model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bc6bd3d3331854fe5c3d0bfaf5d76bd7b7ce0405c8328bb4a339f3a8c29f6862
+size 6174895536
diff --git a/qwen2.5-1.5b-instruct/checkpoint-142/optimizer.pt b/qwen2.5-1.5b-instruct/checkpoint-142/optimizer.pt
new file mode 100644
index 0000000000000000000000000000000000000000..e47d01f2ef8c0fb0bc0f08d9ba86b0c63d4cbb2b
--- /dev/null
+++ b/qwen2.5-1.5b-instruct/checkpoint-142/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d04877e8d3ab5a714fd42cd86c246c65807c6408e227a22cc3adc02fa6899ff0
+size 12350013801
diff --git a/qwen2.5-1.5b-instruct/checkpoint-142/rng_state_0.pth b/qwen2.5-1.5b-instruct/checkpoint-142/rng_state_0.pth
new file mode 100644
index 0000000000000000000000000000000000000000..870021e3fa5ac35c2f711adf0c93a556ab4842da
--- /dev/null
+++ b/qwen2.5-1.5b-instruct/checkpoint-142/rng_state_0.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:95e5fc2074c0df31522a514f862c86cb00d71c946a7f15cc9ec0e53a69fb28a7
+size 14917
diff --git a/qwen2.5-1.5b-instruct/checkpoint-142/rng_state_1.pth b/qwen2.5-1.5b-instruct/checkpoint-142/rng_state_1.pth
new file mode 100644
index 0000000000000000000000000000000000000000..21f20da5eb1da017f08aaa88bd19cf24d40e3fbf
--- /dev/null
+++ b/qwen2.5-1.5b-instruct/checkpoint-142/rng_state_1.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0e7153eae67b6c9232a41bc996a2bf5b83229b8c7230d61911ac0fd40e64154e
+size 14917
diff --git a/qwen2.5-1.5b-instruct/checkpoint-142/scheduler.pt b/qwen2.5-1.5b-instruct/checkpoint-142/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..2967e7243507a115f9424b1b386bf3657f047c75
--- /dev/null
+++ b/qwen2.5-1.5b-instruct/checkpoint-142/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:922deb71175797a96c4bddc5e1cef1c097df38c0dfa52334e97c89e0f573f2bb
+size 1465
diff --git a/qwen2.5-1.5b-instruct/checkpoint-142/tokenizer.json b/qwen2.5-1.5b-instruct/checkpoint-142/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..34510ff0037cd50428af467a17ead5a96140a32c
--- /dev/null
+++ b/qwen2.5-1.5b-instruct/checkpoint-142/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
+size 11421892
diff --git a/qwen2.5-1.5b-instruct/checkpoint-142/tokenizer_config.json b/qwen2.5-1.5b-instruct/checkpoint-142/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..145e2c7a68b109ae0253dda29fc2e66cb01b4b85
--- /dev/null
+++ b/qwen2.5-1.5b-instruct/checkpoint-142/tokenizer_config.json
@@ -0,0 +1,30 @@
+{
+ "add_prefix_space": false,
+ "backend": "tokenizers",
+ "bos_token": null,
+ "clean_up_tokenization_spaces": false,
+ "eos_token": "<|im_end|>",
+ "errors": "replace",
+ "extra_special_tokens": [
+ "<|im_start|>",
+ "<|im_end|>",
+ "<|object_ref_start|>",
+ "<|object_ref_end|>",
+ "<|box_start|>",
+ "<|box_end|>",
+ "<|quad_start|>",
+ "<|quad_end|>",
+ "<|vision_start|>",
+ "<|vision_end|>",
+ "<|vision_pad|>",
+ "<|image_pad|>",
+ "<|video_pad|>"
+ ],
+ "is_local": false,
+ "model_max_length": 131072,
+ "pad_token": "<|endoftext|>",
+ "padding_side": "right",
+ "split_special_tokens": false,
+ "tokenizer_class": "Qwen2Tokenizer",
+ "unk_token": null
+}
diff --git a/qwen2.5-1.5b-instruct/checkpoint-142/trainer_state.json b/qwen2.5-1.5b-instruct/checkpoint-142/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..f77901042cde7419208c200b7953efcb7a243b09
--- /dev/null
+++ b/qwen2.5-1.5b-instruct/checkpoint-142/trainer_state.json
@@ -0,0 +1,1044 @@
+{
+ "best_global_step": null,
+ "best_metric": null,
+ "best_model_checkpoint": null,
+ "epoch": 2.0,
+ "eval_steps": 50,
+ "global_step": 142,
+ "is_hyper_param_search": false,
+ "is_local_process_zero": true,
+ "is_world_process_zero": true,
+ "log_history": [
+ {
+ "epoch": 0.014222222222222223,
+ "grad_norm": 0.10009177774190903,
+ "learning_rate": 0.0,
+ "loss": 0.704937219619751,
+ "step": 1
+ },
+ {
+ "epoch": 0.028444444444444446,
+ "grad_norm": 0.10451912879943848,
+ "learning_rate": 4.5454545454545457e-07,
+ "loss": 0.7318670749664307,
+ "step": 2
+ },
+ {
+ "epoch": 0.042666666666666665,
+ "grad_norm": 0.07928935438394547,
+ "learning_rate": 9.090909090909091e-07,
+ "loss": 0.6017104387283325,
+ "step": 3
+ },
+ {
+ "epoch": 0.05688888888888889,
+ "grad_norm": 0.048069290816783905,
+ "learning_rate": 1.3636363636363636e-06,
+ "loss": 0.6029032468795776,
+ "step": 4
+ },
+ {
+ "epoch": 0.07111111111111111,
+ "grad_norm": 0.02866927720606327,
+ "learning_rate": 1.8181818181818183e-06,
+ "loss": 0.5051919221878052,
+ "step": 5
+ },
+ {
+ "epoch": 0.08533333333333333,
+ "grad_norm": 0.022201435640454292,
+ "learning_rate": 2.2727272727272728e-06,
+ "loss": 0.5942293405532837,
+ "step": 6
+ },
+ {
+ "epoch": 0.09955555555555555,
+ "grad_norm": 0.018263086676597595,
+ "learning_rate": 2.7272727272727272e-06,
+ "loss": 0.5781338214874268,
+ "step": 7
+ },
+ {
+ "epoch": 0.11377777777777778,
+ "grad_norm": 0.01827927492558956,
+ "learning_rate": 3.181818181818182e-06,
+ "loss": 0.5180214047431946,
+ "step": 8
+ },
+ {
+ "epoch": 0.128,
+ "grad_norm": 0.014687157236039639,
+ "learning_rate": 3.6363636363636366e-06,
+ "loss": 0.5420116186141968,
+ "step": 9
+ },
+ {
+ "epoch": 0.14222222222222222,
+ "grad_norm": 0.012042663060128689,
+ "learning_rate": 4.0909090909090915e-06,
+ "loss": 0.5345908403396606,
+ "step": 10
+ },
+ {
+ "epoch": 0.15644444444444444,
+ "grad_norm": 0.012934447266161442,
+ "learning_rate": 4.5454545454545455e-06,
+ "loss": 0.46689414978027344,
+ "step": 11
+ },
+ {
+ "epoch": 0.17066666666666666,
+ "grad_norm": 0.012295619584619999,
+ "learning_rate": 5e-06,
+ "loss": 0.5062131881713867,
+ "step": 12
+ },
+ {
+ "epoch": 0.18488888888888888,
+ "grad_norm": 0.011856353841722012,
+ "learning_rate": 5.4545454545454545e-06,
+ "loss": 0.5063096880912781,
+ "step": 13
+ },
+ {
+ "epoch": 0.1991111111111111,
+ "grad_norm": 0.009451798163354397,
+ "learning_rate": 5.90909090909091e-06,
+ "loss": 0.5130602121353149,
+ "step": 14
+ },
+ {
+ "epoch": 0.21333333333333335,
+ "grad_norm": 0.01111016795039177,
+ "learning_rate": 6.363636363636364e-06,
+ "loss": 0.5016045570373535,
+ "step": 15
+ },
+ {
+ "epoch": 0.22755555555555557,
+ "grad_norm": 0.010735263116657734,
+ "learning_rate": 6.818181818181818e-06,
+ "loss": 0.49827253818511963,
+ "step": 16
+ },
+ {
+ "epoch": 0.24177777777777779,
+ "grad_norm": 0.010710126720368862,
+ "learning_rate": 7.272727272727273e-06,
+ "loss": 0.479993999004364,
+ "step": 17
+ },
+ {
+ "epoch": 0.256,
+ "grad_norm": 0.009600787423551083,
+ "learning_rate": 7.727272727272727e-06,
+ "loss": 0.4925128221511841,
+ "step": 18
+ },
+ {
+ "epoch": 0.2702222222222222,
+ "grad_norm": 0.010486789979040623,
+ "learning_rate": 8.181818181818183e-06,
+ "loss": 0.5188871622085571,
+ "step": 19
+ },
+ {
+ "epoch": 0.28444444444444444,
+ "grad_norm": 0.013396799564361572,
+ "learning_rate": 8.636363636363637e-06,
+ "loss": 0.5192517042160034,
+ "step": 20
+ },
+ {
+ "epoch": 0.2986666666666667,
+ "grad_norm": 0.009299715980887413,
+ "learning_rate": 9.090909090909091e-06,
+ "loss": 0.4580157399177551,
+ "step": 21
+ },
+ {
+ "epoch": 0.3128888888888889,
+ "grad_norm": 0.009691477753221989,
+ "learning_rate": 9.545454545454547e-06,
+ "loss": 0.49455106258392334,
+ "step": 22
+ },
+ {
+ "epoch": 0.32711111111111113,
+ "grad_norm": 0.008896945044398308,
+ "learning_rate": 1e-05,
+ "loss": 0.48672014474868774,
+ "step": 23
+ },
+ {
+ "epoch": 0.3413333333333333,
+ "grad_norm": 0.009234798140823841,
+ "learning_rate": 9.999323662872998e-06,
+ "loss": 0.5052967071533203,
+ "step": 24
+ },
+ {
+ "epoch": 0.35555555555555557,
+ "grad_norm": 0.009167873300611973,
+ "learning_rate": 9.99729483446475e-06,
+ "loss": 0.4740435779094696,
+ "step": 25
+ },
+ {
+ "epoch": 0.36977777777777776,
+ "grad_norm": 0.009001555852591991,
+ "learning_rate": 9.993914063644053e-06,
+ "loss": 0.47315946221351624,
+ "step": 26
+ },
+ {
+ "epoch": 0.384,
+ "grad_norm": 0.00915393978357315,
+ "learning_rate": 9.989182265027232e-06,
+ "loss": 0.4557237923145294,
+ "step": 27
+ },
+ {
+ "epoch": 0.3982222222222222,
+ "grad_norm": 0.008505849167704582,
+ "learning_rate": 9.98310071873072e-06,
+ "loss": 0.46317678689956665,
+ "step": 28
+ },
+ {
+ "epoch": 0.41244444444444445,
+ "grad_norm": 0.008668503724038601,
+ "learning_rate": 9.975671070024741e-06,
+ "loss": 0.4533114433288574,
+ "step": 29
+ },
+ {
+ "epoch": 0.4266666666666667,
+ "grad_norm": 0.009284070692956448,
+ "learning_rate": 9.966895328888195e-06,
+ "loss": 0.5624005794525146,
+ "step": 30
+ },
+ {
+ "epoch": 0.4408888888888889,
+ "grad_norm": 0.00935022346675396,
+ "learning_rate": 9.956775869464901e-06,
+ "loss": 0.4420597553253174,
+ "step": 31
+ },
+ {
+ "epoch": 0.45511111111111113,
+ "grad_norm": 0.008682326413691044,
+ "learning_rate": 9.945315429421307e-06,
+ "loss": 0.4745720624923706,
+ "step": 32
+ },
+ {
+ "epoch": 0.4693333333333333,
+ "grad_norm": 0.007816251367330551,
+ "learning_rate": 9.932517109205849e-06,
+ "loss": 0.439736545085907,
+ "step": 33
+ },
+ {
+ "epoch": 0.48355555555555557,
+ "grad_norm": 0.009315130300819874,
+ "learning_rate": 9.918384371210178e-06,
+ "loss": 0.5357447862625122,
+ "step": 34
+ },
+ {
+ "epoch": 0.49777777777777776,
+ "grad_norm": 0.009141312912106514,
+ "learning_rate": 9.902921038832456e-06,
+ "loss": 0.519463062286377,
+ "step": 35
+ },
+ {
+ "epoch": 0.512,
+ "grad_norm": 0.008826862089335918,
+ "learning_rate": 9.886131295443003e-06,
+ "loss": 0.48377716541290283,
+ "step": 36
+ },
+ {
+ "epoch": 0.5262222222222223,
+ "grad_norm": 0.008700878359377384,
+ "learning_rate": 9.868019683252543e-06,
+ "loss": 0.48436683416366577,
+ "step": 37
+ },
+ {
+ "epoch": 0.5404444444444444,
+ "grad_norm": 0.011412105523049831,
+ "learning_rate": 9.848591102083375e-06,
+ "loss": 0.4949312210083008,
+ "step": 38
+ },
+ {
+ "epoch": 0.5546666666666666,
+ "grad_norm": 0.009214675053954124,
+ "learning_rate": 9.82785080804381e-06,
+ "loss": 0.4901079535484314,
+ "step": 39
+ },
+ {
+ "epoch": 0.5688888888888889,
+ "grad_norm": 0.01163570862263441,
+ "learning_rate": 9.805804412106197e-06,
+ "loss": 0.6070311665534973,
+ "step": 40
+ },
+ {
+ "epoch": 0.5831111111111111,
+ "grad_norm": 0.009325762279331684,
+ "learning_rate": 9.782457878588977e-06,
+ "loss": 0.48412975668907166,
+ "step": 41
+ },
+ {
+ "epoch": 0.5973333333333334,
+ "grad_norm": 0.008266489021480083,
+ "learning_rate": 9.75781752354311e-06,
+ "loss": 0.4440661072731018,
+ "step": 42
+ },
+ {
+ "epoch": 0.6115555555555555,
+ "grad_norm": 0.00843313243240118,
+ "learning_rate": 9.731890013043367e-06,
+ "loss": 0.502442479133606,
+ "step": 43
+ },
+ {
+ "epoch": 0.6257777777777778,
+ "grad_norm": 0.00934070535004139,
+ "learning_rate": 9.704682361384941e-06,
+ "loss": 0.47924768924713135,
+ "step": 44
+ },
+ {
+ "epoch": 0.64,
+ "grad_norm": 0.010360248386859894,
+ "learning_rate": 9.676201929185809e-06,
+ "loss": 0.4161364436149597,
+ "step": 45
+ },
+ {
+ "epoch": 0.6542222222222223,
+ "grad_norm": 0.0094041982665658,
+ "learning_rate": 9.646456421395447e-06,
+ "loss": 0.48306071758270264,
+ "step": 46
+ },
+ {
+ "epoch": 0.6684444444444444,
+ "grad_norm": 0.009070150554180145,
+ "learning_rate": 9.615453885210368e-06,
+ "loss": 0.4774249792098999,
+ "step": 47
+ },
+ {
+ "epoch": 0.6826666666666666,
+ "grad_norm": 0.009771023876965046,
+ "learning_rate": 9.583202707897075e-06,
+ "loss": 0.4948921203613281,
+ "step": 48
+ },
+ {
+ "epoch": 0.6968888888888889,
+ "grad_norm": 0.008749130181968212,
+ "learning_rate": 9.549711614523007e-06,
+ "loss": 0.4825429916381836,
+ "step": 49
+ },
+ {
+ "epoch": 0.7111111111111111,
+ "grad_norm": 0.007630312815308571,
+ "learning_rate": 9.514989665596114e-06,
+ "loss": 0.46055319905281067,
+ "step": 50
+ },
+ {
+ "epoch": 0.7111111111111111,
+ "eval_loss": 0.48781517148017883,
+ "eval_runtime": 4.5903,
+ "eval_samples_per_second": 43.57,
+ "eval_steps_per_second": 21.785,
+ "step": 50
+ },
+ {
+ "epoch": 0.7253333333333334,
+ "grad_norm": 0.008408056572079659,
+ "learning_rate": 9.479046254613673e-06,
+ "loss": 0.4626312255859375,
+ "step": 51
+ },
+ {
+ "epoch": 0.7395555555555555,
+ "grad_norm": 0.007699670735746622,
+ "learning_rate": 9.441891105521005e-06,
+ "loss": 0.4499070644378662,
+ "step": 52
+ },
+ {
+ "epoch": 0.7537777777777778,
+ "grad_norm": 0.009070169180631638,
+ "learning_rate": 9.40353427008083e-06,
+ "loss": 0.4825657606124878,
+ "step": 53
+ },
+ {
+ "epoch": 0.768,
+ "grad_norm": 0.007915646769106388,
+ "learning_rate": 9.3639861251539e-06,
+ "loss": 0.4534468650817871,
+ "step": 54
+ },
+ {
+ "epoch": 0.7822222222222223,
+ "grad_norm": 0.011336776427924633,
+ "learning_rate": 9.323257369891702e-06,
+ "loss": 0.4988631010055542,
+ "step": 55
+ },
+ {
+ "epoch": 0.7964444444444444,
+ "grad_norm": 0.00761911366134882,
+ "learning_rate": 9.281359022841966e-06,
+ "loss": 0.4371437728404999,
+ "step": 56
+ },
+ {
+ "epoch": 0.8106666666666666,
+ "grad_norm": 0.009316994808614254,
+ "learning_rate": 9.238302418967757e-06,
+ "loss": 0.4956932067871094,
+ "step": 57
+ },
+ {
+ "epoch": 0.8248888888888889,
+ "grad_norm": 0.009821794927120209,
+ "learning_rate": 9.194099206580981e-06,
+ "loss": 0.5392974615097046,
+ "step": 58
+ },
+ {
+ "epoch": 0.8391111111111111,
+ "grad_norm": 0.008422416634857655,
+ "learning_rate": 9.14876134419111e-06,
+ "loss": 0.47983357310295105,
+ "step": 59
+ },
+ {
+ "epoch": 0.8533333333333334,
+ "grad_norm": 0.00837004091590643,
+ "learning_rate": 9.102301097269974e-06,
+ "loss": 0.5183510780334473,
+ "step": 60
+ },
+ {
+ "epoch": 0.8675555555555555,
+ "grad_norm": 0.008596297353506088,
+ "learning_rate": 9.05473103493355e-06,
+ "loss": 0.47292786836624146,
+ "step": 61
+ },
+ {
+ "epoch": 0.8817777777777778,
+ "grad_norm": 0.009059637784957886,
+ "learning_rate": 9.006064026541549e-06,
+ "loss": 0.5188573598861694,
+ "step": 62
+ },
+ {
+ "epoch": 0.896,
+ "grad_norm": 0.007816917262971401,
+ "learning_rate": 8.956313238215824e-06,
+ "loss": 0.45641395449638367,
+ "step": 63
+ },
+ {
+ "epoch": 0.9102222222222223,
+ "grad_norm": 0.007985202595591545,
+ "learning_rate": 8.905492129278478e-06,
+ "loss": 0.4586361050605774,
+ "step": 64
+ },
+ {
+ "epoch": 0.9244444444444444,
+ "grad_norm": 0.007170964498072863,
+ "learning_rate": 8.85361444861063e-06,
+ "loss": 0.4394378066062927,
+ "step": 65
+ },
+ {
+ "epoch": 0.9386666666666666,
+ "grad_norm": 0.00909162126481533,
+ "learning_rate": 8.800694230932885e-06,
+ "loss": 0.509820818901062,
+ "step": 66
+ },
+ {
+ "epoch": 0.9528888888888889,
+ "grad_norm": 0.00786044355481863,
+ "learning_rate": 8.74674579300843e-06,
+ "loss": 0.43584880232810974,
+ "step": 67
+ },
+ {
+ "epoch": 0.9671111111111111,
+ "grad_norm": 0.00832363311201334,
+ "learning_rate": 8.691783729769874e-06,
+ "loss": 0.4317135214805603,
+ "step": 68
+ },
+ {
+ "epoch": 0.9813333333333333,
+ "grad_norm": 0.008376957848668098,
+ "learning_rate": 8.635822910370793e-06,
+ "loss": 0.49933016300201416,
+ "step": 69
+ },
+ {
+ "epoch": 0.9955555555555555,
+ "grad_norm": 0.00835558120161295,
+ "learning_rate": 8.578878474163115e-06,
+ "loss": 0.46570393443107605,
+ "step": 70
+ },
+ {
+ "epoch": 1.0,
+ "grad_norm": 0.013256405480206013,
+ "learning_rate": 8.520965826601394e-06,
+ "loss": 0.45107918977737427,
+ "step": 71
+ },
+ {
+ "epoch": 1.0142222222222221,
+ "grad_norm": 0.008708071894943714,
+ "learning_rate": 8.462100635075097e-06,
+ "loss": 0.47125527262687683,
+ "step": 72
+ },
+ {
+ "epoch": 1.0284444444444445,
+ "grad_norm": 0.008086243644356728,
+ "learning_rate": 8.40229882467003e-06,
+ "loss": 0.4322492480278015,
+ "step": 73
+ },
+ {
+ "epoch": 1.0426666666666666,
+ "grad_norm": 0.0073950947262346745,
+ "learning_rate": 8.341576573860049e-06,
+ "loss": 0.38362830877304077,
+ "step": 74
+ },
+ {
+ "epoch": 1.056888888888889,
+ "grad_norm": 0.0076777078211307526,
+ "learning_rate": 8.279950310130218e-06,
+ "loss": 0.3878062665462494,
+ "step": 75
+ },
+ {
+ "epoch": 1.0711111111111111,
+ "grad_norm": 0.007898081094026566,
+ "learning_rate": 8.2174367055326e-06,
+ "loss": 0.3833811283111572,
+ "step": 76
+ },
+ {
+ "epoch": 1.0853333333333333,
+ "grad_norm": 0.007348628249019384,
+ "learning_rate": 8.154052672175888e-06,
+ "loss": 0.3845909833908081,
+ "step": 77
+ },
+ {
+ "epoch": 1.0995555555555556,
+ "grad_norm": 0.008542248979210854,
+ "learning_rate": 8.08981535765009e-06,
+ "loss": 0.44259265065193176,
+ "step": 78
+ },
+ {
+ "epoch": 1.1137777777777778,
+ "grad_norm": 0.007998447865247726,
+ "learning_rate": 8.024742140387506e-06,
+ "loss": 0.38503497838974,
+ "step": 79
+ },
+ {
+ "epoch": 1.1280000000000001,
+ "grad_norm": 0.008362009190022945,
+ "learning_rate": 7.95885062496126e-06,
+ "loss": 0.42608046531677246,
+ "step": 80
+ },
+ {
+ "epoch": 1.1422222222222222,
+ "grad_norm": 0.007770548574626446,
+ "learning_rate": 7.892158637322647e-06,
+ "loss": 0.38991105556488037,
+ "step": 81
+ },
+ {
+ "epoch": 1.1564444444444444,
+ "grad_norm": 0.007649772334843874,
+ "learning_rate": 7.824684219978591e-06,
+ "loss": 0.36854323744773865,
+ "step": 82
+ },
+ {
+ "epoch": 1.1706666666666667,
+ "grad_norm": 0.008056909777224064,
+ "learning_rate": 7.756445627110523e-06,
+ "loss": 0.4200327396392822,
+ "step": 83
+ },
+ {
+ "epoch": 1.1848888888888889,
+ "grad_norm": 0.0089965695515275,
+ "learning_rate": 7.68746131963598e-06,
+ "loss": 0.4378758668899536,
+ "step": 84
+ },
+ {
+ "epoch": 1.199111111111111,
+ "grad_norm": 0.007884320802986622,
+ "learning_rate": 7.6177499602143e-06,
+ "loss": 0.3808164894580841,
+ "step": 85
+ },
+ {
+ "epoch": 1.2133333333333334,
+ "grad_norm": 0.009615465067327023,
+ "learning_rate": 7.547330408197695e-06,
+ "loss": 0.4453170895576477,
+ "step": 86
+ },
+ {
+ "epoch": 1.2275555555555555,
+ "grad_norm": 0.00891174841672182,
+ "learning_rate": 7.476221714529167e-06,
+ "loss": 0.4456181526184082,
+ "step": 87
+ },
+ {
+ "epoch": 1.2417777777777779,
+ "grad_norm": 0.008833018131554127,
+ "learning_rate": 7.404443116588548e-06,
+ "loss": 0.43805548548698425,
+ "step": 88
+ },
+ {
+ "epoch": 1.256,
+ "grad_norm": 0.008171891793608665,
+ "learning_rate": 7.332014032988123e-06,
+ "loss": 0.4144285321235657,
+ "step": 89
+ },
+ {
+ "epoch": 1.2702222222222221,
+ "grad_norm": 0.009090668521821499,
+ "learning_rate": 7.2589540583192165e-06,
+ "loss": 0.38928651809692383,
+ "step": 90
+ },
+ {
+ "epoch": 1.2844444444444445,
+ "grad_norm": 0.008077106438577175,
+ "learning_rate": 7.185282957851175e-06,
+ "loss": 0.387723445892334,
+ "step": 91
+ },
+ {
+ "epoch": 1.2986666666666666,
+ "grad_norm": 0.008803839795291424,
+ "learning_rate": 7.111020662184174e-06,
+ "loss": 0.4158015549182892,
+ "step": 92
+ },
+ {
+ "epoch": 1.3128888888888888,
+ "grad_norm": 0.008249706588685513,
+ "learning_rate": 7.036187261857289e-06,
+ "loss": 0.39764612913131714,
+ "step": 93
+ },
+ {
+ "epoch": 1.3271111111111111,
+ "grad_norm": 0.008549178950488567,
+ "learning_rate": 6.960803001913315e-06,
+ "loss": 0.38216379284858704,
+ "step": 94
+ },
+ {
+ "epoch": 1.3413333333333333,
+ "grad_norm": 0.007755788508802652,
+ "learning_rate": 6.884888276421766e-06,
+ "loss": 0.413602352142334,
+ "step": 95
+ },
+ {
+ "epoch": 1.3555555555555556,
+ "grad_norm": 0.008742124773561954,
+ "learning_rate": 6.8084636229615786e-06,
+ "loss": 0.41756486892700195,
+ "step": 96
+ },
+ {
+ "epoch": 1.3697777777777778,
+ "grad_norm": 0.008473540656268597,
+ "learning_rate": 6.731549717064975e-06,
+ "loss": 0.4028819799423218,
+ "step": 97
+ },
+ {
+ "epoch": 1.384,
+ "grad_norm": 0.008585155010223389,
+ "learning_rate": 6.654167366624009e-06,
+ "loss": 0.37123921513557434,
+ "step": 98
+ },
+ {
+ "epoch": 1.3982222222222223,
+ "grad_norm": 0.00882649328559637,
+ "learning_rate": 6.576337506261314e-06,
+ "loss": 0.39443182945251465,
+ "step": 99
+ },
+ {
+ "epoch": 1.4124444444444444,
+ "grad_norm": 0.008174613118171692,
+ "learning_rate": 6.498081191666549e-06,
+ "loss": 0.42232874035835266,
+ "step": 100
+ },
+ {
+ "epoch": 1.4124444444444444,
+ "eval_loss": 0.4897068440914154,
+ "eval_runtime": 5.0728,
+ "eval_samples_per_second": 39.426,
+ "eval_steps_per_second": 19.713,
+ "step": 100
+ },
+ {
+ "epoch": 1.4266666666666667,
+ "grad_norm": 0.00753920478746295,
+ "learning_rate": 6.419419593900109e-06,
+ "loss": 0.38927626609802246,
+ "step": 101
+ },
+ {
+ "epoch": 1.4408888888888889,
+ "grad_norm": 0.007967391051352024,
+ "learning_rate": 6.340373993665607e-06,
+ "loss": 0.36544570326805115,
+ "step": 102
+ },
+ {
+ "epoch": 1.455111111111111,
+ "grad_norm": 0.008992813527584076,
+ "learning_rate": 6.2609657755527135e-06,
+ "loss": 0.4033282995223999,
+ "step": 103
+ },
+ {
+ "epoch": 1.4693333333333334,
+ "grad_norm": 0.007906445302069187,
+ "learning_rate": 6.1812164222518626e-06,
+ "loss": 0.40876060724258423,
+ "step": 104
+ },
+ {
+ "epoch": 1.4835555555555555,
+ "grad_norm": 0.00823493767529726,
+ "learning_rate": 6.101147508742456e-06,
+ "loss": 0.3994408845901489,
+ "step": 105
+ },
+ {
+ "epoch": 1.4977777777777779,
+ "grad_norm": 0.009032925590872765,
+ "learning_rate": 6.020780696456059e-06,
+ "loss": 0.4667425751686096,
+ "step": 106
+ },
+ {
+ "epoch": 1.512,
+ "grad_norm": 0.007943308912217617,
+ "learning_rate": 5.940137727416247e-06,
+ "loss": 0.4092869460582733,
+ "step": 107
+ },
+ {
+ "epoch": 1.5262222222222221,
+ "grad_norm": 0.008353099226951599,
+ "learning_rate": 5.859240418356614e-06,
+ "loss": 0.39832064509391785,
+ "step": 108
+ },
+ {
+ "epoch": 1.5404444444444443,
+ "grad_norm": 0.008241581730544567,
+ "learning_rate": 5.778110654818602e-06,
+ "loss": 0.4114166498184204,
+ "step": 109
+ },
+ {
+ "epoch": 1.5546666666666666,
+ "grad_norm": 0.007937194779515266,
+ "learning_rate": 5.696770385230679e-06,
+ "loss": 0.4360125660896301,
+ "step": 110
+ },
+ {
+ "epoch": 1.568888888888889,
+ "grad_norm": 0.008319812826812267,
+ "learning_rate": 5.615241614970546e-06,
+ "loss": 0.4153256118297577,
+ "step": 111
+ },
+ {
+ "epoch": 1.5831111111111111,
+ "grad_norm": 0.008011740632355213,
+ "learning_rate": 5.533546400411899e-06,
+ "loss": 0.3954656720161438,
+ "step": 112
+ },
+ {
+ "epoch": 1.5973333333333333,
+ "grad_norm": 0.008084246888756752,
+ "learning_rate": 5.4517068429574215e-06,
+ "loss": 0.4052407741546631,
+ "step": 113
+ },
+ {
+ "epoch": 1.6115555555555554,
+ "grad_norm": 0.007481739390641451,
+ "learning_rate": 5.3697450830595775e-06,
+ "loss": 0.4017290472984314,
+ "step": 114
+ },
+ {
+ "epoch": 1.6257777777777778,
+ "grad_norm": 0.008050751872360706,
+ "learning_rate": 5.287683294230855e-06,
+ "loss": 0.38467150926589966,
+ "step": 115
+ },
+ {
+ "epoch": 1.6400000000000001,
+ "grad_norm": 0.0077187251299619675,
+ "learning_rate": 5.20554367704505e-06,
+ "loss": 0.395123690366745,
+ "step": 116
+ },
+ {
+ "epoch": 1.6542222222222223,
+ "grad_norm": 0.008423185907304287,
+ "learning_rate": 5.123348453131242e-06,
+ "loss": 0.39796996116638184,
+ "step": 117
+ },
+ {
+ "epoch": 1.6684444444444444,
+ "grad_norm": 0.009052549488842487,
+ "learning_rate": 5.041119859162068e-06,
+ "loss": 0.4162113070487976,
+ "step": 118
+ },
+ {
+ "epoch": 1.6826666666666665,
+ "grad_norm": 0.008246202953159809,
+ "learning_rate": 4.958880140837934e-06,
+ "loss": 0.42128854990005493,
+ "step": 119
+ },
+ {
+ "epoch": 1.696888888888889,
+ "grad_norm": 0.00833312701433897,
+ "learning_rate": 4.876651546868759e-06,
+ "loss": 0.4024372696876526,
+ "step": 120
+ },
+ {
+ "epoch": 1.7111111111111112,
+ "grad_norm": 0.008461604826152325,
+ "learning_rate": 4.794456322954953e-06,
+ "loss": 0.46918508410453796,
+ "step": 121
+ },
+ {
+ "epoch": 1.7253333333333334,
+ "grad_norm": 0.00825911108404398,
+ "learning_rate": 4.7123167057691446e-06,
+ "loss": 0.41863948106765747,
+ "step": 122
+ },
+ {
+ "epoch": 1.7395555555555555,
+ "grad_norm": 0.009252563118934631,
+ "learning_rate": 4.630254916940424e-06,
+ "loss": 0.44644808769226074,
+ "step": 123
+ },
+ {
+ "epoch": 1.7537777777777777,
+ "grad_norm": 0.007843364961445332,
+ "learning_rate": 4.548293157042581e-06,
+ "loss": 0.42086148262023926,
+ "step": 124
+ },
+ {
+ "epoch": 1.768,
+ "grad_norm": 0.007915403693914413,
+ "learning_rate": 4.466453599588103e-06,
+ "loss": 0.4266040027141571,
+ "step": 125
+ },
+ {
+ "epoch": 1.7822222222222224,
+ "grad_norm": 0.007302967831492424,
+ "learning_rate": 4.384758385029457e-06,
+ "loss": 0.3592030704021454,
+ "step": 126
+ },
+ {
+ "epoch": 1.7964444444444445,
+ "grad_norm": 0.00869770348072052,
+ "learning_rate": 4.3032296147693225e-06,
+ "loss": 0.42212432622909546,
+ "step": 127
+ },
+ {
+ "epoch": 1.8106666666666666,
+ "grad_norm": 0.008091778494417667,
+ "learning_rate": 4.2218893451814005e-06,
+ "loss": 0.37964335083961487,
+ "step": 128
+ },
+ {
+ "epoch": 1.8248888888888888,
+ "grad_norm": 0.0076453504152596,
+ "learning_rate": 4.140759581643386e-06,
+ "loss": 0.37467944622039795,
+ "step": 129
+ },
+ {
+ "epoch": 1.8391111111111111,
+ "grad_norm": 0.007656064350157976,
+ "learning_rate": 4.059862272583755e-06,
+ "loss": 0.393369197845459,
+ "step": 130
+ },
+ {
+ "epoch": 1.8533333333333335,
+ "grad_norm": 0.008399168029427528,
+ "learning_rate": 3.979219303543942e-06,
+ "loss": 0.394899845123291,
+ "step": 131
+ },
+ {
+ "epoch": 1.8675555555555556,
+ "grad_norm": 0.008833736181259155,
+ "learning_rate": 3.898852491257547e-06,
+ "loss": 0.3997966945171356,
+ "step": 132
+ },
+ {
+ "epoch": 1.8817777777777778,
+ "grad_norm": 0.008364683948457241,
+ "learning_rate": 3.818783577748138e-06,
+ "loss": 0.44034862518310547,
+ "step": 133
+ },
+ {
+ "epoch": 1.896,
+ "grad_norm": 0.008851971477270126,
+ "learning_rate": 3.7390342244472886e-06,
+ "loss": 0.4251548647880554,
+ "step": 134
+ },
+ {
+ "epoch": 1.9102222222222223,
+ "grad_norm": 0.007921694777905941,
+ "learning_rate": 3.659626006334395e-06,
+ "loss": 0.3972621560096741,
+ "step": 135
+ },
+ {
+ "epoch": 1.9244444444444444,
+ "grad_norm": 0.008714950643479824,
+ "learning_rate": 3.5805804060998926e-06,
+ "loss": 0.4237692356109619,
+ "step": 136
+ },
+ {
+ "epoch": 1.9386666666666668,
+ "grad_norm": 0.008546526543796062,
+ "learning_rate": 3.501918808333453e-06,
+ "loss": 0.5015424489974976,
+ "step": 137
+ },
+ {
+ "epoch": 1.952888888888889,
+ "grad_norm": 0.008581602945923805,
+ "learning_rate": 3.4236624937386874e-06,
+ "loss": 0.4541424512863159,
+ "step": 138
+ },
+ {
+ "epoch": 1.967111111111111,
+ "grad_norm": 0.008913579396903515,
+ "learning_rate": 3.3458326333759927e-06,
+ "loss": 0.42448267340660095,
+ "step": 139
+ },
+ {
+ "epoch": 1.9813333333333332,
+ "grad_norm": 0.008540489710867405,
+ "learning_rate": 3.268450282935026e-06,
+ "loss": 0.4514390528202057,
+ "step": 140
+ },
+ {
+ "epoch": 1.9955555555555555,
+ "grad_norm": 0.007863311097025871,
+ "learning_rate": 3.1915363770384223e-06,
+ "loss": 0.38879817724227905,
+ "step": 141
+ },
+ {
+ "epoch": 2.0,
+ "grad_norm": 0.013070300221443176,
+ "learning_rate": 3.1151117235782346e-06,
+ "loss": 0.3734951913356781,
+ "step": 142
+ }
+ ],
+ "logging_steps": 1,
+ "max_steps": 213,
+ "num_input_tokens_seen": 0,
+ "num_train_epochs": 3,
+ "save_steps": 500,
+ "stateful_callbacks": {
+ "TrainerControl": {
+ "args": {
+ "should_epoch_stop": false,
+ "should_evaluate": false,
+ "should_log": false,
+ "should_save": true,
+ "should_training_stop": false
+ },
+ "attributes": {}
+ }
+ },
+ "total_flos": 3.2453383931559936e+16,
+ "train_batch_size": 1,
+ "trial_name": null,
+ "trial_params": null
+}
diff --git a/qwen2.5-1.5b-instruct/checkpoint-142/training_args.bin b/qwen2.5-1.5b-instruct/checkpoint-142/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..1c92bb7fdd3ef14952372d452eaf2c7d0ade0ccf
--- /dev/null
+++ b/qwen2.5-1.5b-instruct/checkpoint-142/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:841e251fafc8c15cdc7b32aa47df63c6ea8f5d8eb459fbecdbb2a9f5f74be033
+size 5649
diff --git a/qwen2.5-1.5b-instruct/checkpoint-213/chat_template.jinja b/qwen2.5-1.5b-instruct/checkpoint-213/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..bdf7919a96cfe43d50914a007b9c0877bd0ec27e
--- /dev/null
+++ b/qwen2.5-1.5b-instruct/checkpoint-213/chat_template.jinja
@@ -0,0 +1,54 @@
+{%- if tools %}
+ {{- '<|im_start|>system\n' }}
+ {%- if messages[0]['role'] == 'system' %}
+ {{- messages[0]['content'] }}
+ {%- else %}
+ {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+ {%- endif %}
+ {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within XML tags:\n" }}
+ {%- for tool in tools %}
+ {{- "\n" }}
+ {{- tool | tojson }}
+ {%- endfor %}
+ {{- "\n\n\nFor each function call, return a json object with function name and arguments within XML tags:\n\n{\"name\": , \"arguments\": }\n<|im_end|>\n" }}
+{%- else %}
+ {%- if messages[0]['role'] == 'system' %}
+ {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+ {%- else %}
+ {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+ {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+ {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+ {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+ {%- elif message.role == "assistant" %}
+ {{- '<|im_start|>' + message.role }}
+ {%- if message.content %}
+ {{- '\n' + message.content }}
+ {%- endif %}
+ {%- for tool_call in message.tool_calls %}
+ {%- if tool_call.function is defined %}
+ {%- set tool_call = tool_call.function %}
+ {%- endif %}
+ {{- '\n\n{"name": "' }}
+ {{- tool_call.name }}
+ {{- '", "arguments": ' }}
+ {{- tool_call.arguments | tojson }}
+ {{- '}\n' }}
+ {%- endfor %}
+ {{- '<|im_end|>\n' }}
+ {%- elif message.role == "tool" %}
+ {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+ {{- '<|im_start|>user' }}
+ {%- endif %}
+ {{- '\n\n' }}
+ {{- message.content }}
+ {{- '\n' }}
+ {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+ {{- '<|im_end|>\n' }}
+ {%- endif %}
+ {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+ {{- '<|im_start|>assistant\n' }}
+{%- endif %}
diff --git a/qwen2.5-1.5b-instruct/checkpoint-213/config.json b/qwen2.5-1.5b-instruct/checkpoint-213/config.json
new file mode 100644
index 0000000000000000000000000000000000000000..499065fd744a64fc8ad2884da26e85a6e99d468a
--- /dev/null
+++ b/qwen2.5-1.5b-instruct/checkpoint-213/config.json
@@ -0,0 +1,61 @@
+{
+ "architectures": [
+ "Qwen2ForCausalLM"
+ ],
+ "attention_dropout": 0.0,
+ "bos_token_id": null,
+ "dtype": "float32",
+ "eos_token_id": 151645,
+ "hidden_act": "silu",
+ "hidden_size": 1536,
+ "initializer_range": 0.02,
+ "intermediate_size": 8960,
+ "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": 21,
+ "model_type": "qwen2",
+ "num_attention_heads": 12,
+ "num_hidden_layers": 28,
+ "num_key_value_heads": 2,
+ "pad_token_id": 151643,
+ "rms_norm_eps": 1e-06,
+ "rope_parameters": {
+ "rope_theta": 1000000.0,
+ "rope_type": "default"
+ },
+ "sliding_window": null,
+ "tie_word_embeddings": true,
+ "transformers_version": "5.2.0",
+ "use_cache": false,
+ "use_sliding_window": false,
+ "vocab_size": 151936
+}
diff --git a/qwen2.5-1.5b-instruct/checkpoint-213/generation_config.json b/qwen2.5-1.5b-instruct/checkpoint-213/generation_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..c5b4e11411fed3e7a5e6557165f888a7b0b1248b
--- /dev/null
+++ b/qwen2.5-1.5b-instruct/checkpoint-213/generation_config.json
@@ -0,0 +1,13 @@
+{
+ "do_sample": true,
+ "eos_token_id": [
+ 151645,
+ 151643
+ ],
+ "pad_token_id": 151643,
+ "repetition_penalty": 1.1,
+ "temperature": 0.7,
+ "top_k": 20,
+ "top_p": 0.8,
+ "transformers_version": "5.2.0"
+}
diff --git a/qwen2.5-1.5b-instruct/checkpoint-213/model.safetensors b/qwen2.5-1.5b-instruct/checkpoint-213/model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..7aa60a284dae84c00e62cae176fd1986a775b887
--- /dev/null
+++ b/qwen2.5-1.5b-instruct/checkpoint-213/model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3434dfcf0343459c122b117ee99c0d83e3e6ac7f15a7c1e7cac6d3772e907708
+size 6174895536
diff --git a/qwen2.5-1.5b-instruct/checkpoint-213/optimizer.pt b/qwen2.5-1.5b-instruct/checkpoint-213/optimizer.pt
new file mode 100644
index 0000000000000000000000000000000000000000..b751586242d8605e7ff725e5f282ea33b3bc7c94
--- /dev/null
+++ b/qwen2.5-1.5b-instruct/checkpoint-213/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fb7627525a73024b21445c887cc027ecedceadcd017c7a082f40d703cd60fd23
+size 12350013801
diff --git a/qwen2.5-1.5b-instruct/checkpoint-213/rng_state_0.pth b/qwen2.5-1.5b-instruct/checkpoint-213/rng_state_0.pth
new file mode 100644
index 0000000000000000000000000000000000000000..870021e3fa5ac35c2f711adf0c93a556ab4842da
--- /dev/null
+++ b/qwen2.5-1.5b-instruct/checkpoint-213/rng_state_0.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:95e5fc2074c0df31522a514f862c86cb00d71c946a7f15cc9ec0e53a69fb28a7
+size 14917
diff --git a/qwen2.5-1.5b-instruct/checkpoint-213/rng_state_1.pth b/qwen2.5-1.5b-instruct/checkpoint-213/rng_state_1.pth
new file mode 100644
index 0000000000000000000000000000000000000000..21f20da5eb1da017f08aaa88bd19cf24d40e3fbf
--- /dev/null
+++ b/qwen2.5-1.5b-instruct/checkpoint-213/rng_state_1.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0e7153eae67b6c9232a41bc996a2bf5b83229b8c7230d61911ac0fd40e64154e
+size 14917
diff --git a/qwen2.5-1.5b-instruct/checkpoint-213/scheduler.pt b/qwen2.5-1.5b-instruct/checkpoint-213/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..39cf87775c3a71d10e7311e0cab769e112f357ba
--- /dev/null
+++ b/qwen2.5-1.5b-instruct/checkpoint-213/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c38a67b98c17be81d4c6d00b60029cdedd43e9aba35ca28ec28ef946c400f268
+size 1465
diff --git a/qwen2.5-1.5b-instruct/checkpoint-213/tokenizer.json b/qwen2.5-1.5b-instruct/checkpoint-213/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..34510ff0037cd50428af467a17ead5a96140a32c
--- /dev/null
+++ b/qwen2.5-1.5b-instruct/checkpoint-213/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
+size 11421892
diff --git a/qwen2.5-1.5b-instruct/checkpoint-213/tokenizer_config.json b/qwen2.5-1.5b-instruct/checkpoint-213/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..145e2c7a68b109ae0253dda29fc2e66cb01b4b85
--- /dev/null
+++ b/qwen2.5-1.5b-instruct/checkpoint-213/tokenizer_config.json
@@ -0,0 +1,30 @@
+{
+ "add_prefix_space": false,
+ "backend": "tokenizers",
+ "bos_token": null,
+ "clean_up_tokenization_spaces": false,
+ "eos_token": "<|im_end|>",
+ "errors": "replace",
+ "extra_special_tokens": [
+ "<|im_start|>",
+ "<|im_end|>",
+ "<|object_ref_start|>",
+ "<|object_ref_end|>",
+ "<|box_start|>",
+ "<|box_end|>",
+ "<|quad_start|>",
+ "<|quad_end|>",
+ "<|vision_start|>",
+ "<|vision_end|>",
+ "<|vision_pad|>",
+ "<|image_pad|>",
+ "<|video_pad|>"
+ ],
+ "is_local": false,
+ "model_max_length": 131072,
+ "pad_token": "<|endoftext|>",
+ "padding_side": "right",
+ "split_special_tokens": false,
+ "tokenizer_class": "Qwen2Tokenizer",
+ "unk_token": null
+}
diff --git a/qwen2.5-1.5b-instruct/checkpoint-213/trainer_state.json b/qwen2.5-1.5b-instruct/checkpoint-213/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..869e6ee60a1e1afe0eb4e4a8afc7dbada5790675
--- /dev/null
+++ b/qwen2.5-1.5b-instruct/checkpoint-213/trainer_state.json
@@ -0,0 +1,1557 @@
+{
+ "best_global_step": null,
+ "best_metric": null,
+ "best_model_checkpoint": null,
+ "epoch": 3.0,
+ "eval_steps": 50,
+ "global_step": 213,
+ "is_hyper_param_search": false,
+ "is_local_process_zero": true,
+ "is_world_process_zero": true,
+ "log_history": [
+ {
+ "epoch": 0.014222222222222223,
+ "grad_norm": 0.10009177774190903,
+ "learning_rate": 0.0,
+ "loss": 0.704937219619751,
+ "step": 1
+ },
+ {
+ "epoch": 0.028444444444444446,
+ "grad_norm": 0.10451912879943848,
+ "learning_rate": 4.5454545454545457e-07,
+ "loss": 0.7318670749664307,
+ "step": 2
+ },
+ {
+ "epoch": 0.042666666666666665,
+ "grad_norm": 0.07928935438394547,
+ "learning_rate": 9.090909090909091e-07,
+ "loss": 0.6017104387283325,
+ "step": 3
+ },
+ {
+ "epoch": 0.05688888888888889,
+ "grad_norm": 0.048069290816783905,
+ "learning_rate": 1.3636363636363636e-06,
+ "loss": 0.6029032468795776,
+ "step": 4
+ },
+ {
+ "epoch": 0.07111111111111111,
+ "grad_norm": 0.02866927720606327,
+ "learning_rate": 1.8181818181818183e-06,
+ "loss": 0.5051919221878052,
+ "step": 5
+ },
+ {
+ "epoch": 0.08533333333333333,
+ "grad_norm": 0.022201435640454292,
+ "learning_rate": 2.2727272727272728e-06,
+ "loss": 0.5942293405532837,
+ "step": 6
+ },
+ {
+ "epoch": 0.09955555555555555,
+ "grad_norm": 0.018263086676597595,
+ "learning_rate": 2.7272727272727272e-06,
+ "loss": 0.5781338214874268,
+ "step": 7
+ },
+ {
+ "epoch": 0.11377777777777778,
+ "grad_norm": 0.01827927492558956,
+ "learning_rate": 3.181818181818182e-06,
+ "loss": 0.5180214047431946,
+ "step": 8
+ },
+ {
+ "epoch": 0.128,
+ "grad_norm": 0.014687157236039639,
+ "learning_rate": 3.6363636363636366e-06,
+ "loss": 0.5420116186141968,
+ "step": 9
+ },
+ {
+ "epoch": 0.14222222222222222,
+ "grad_norm": 0.012042663060128689,
+ "learning_rate": 4.0909090909090915e-06,
+ "loss": 0.5345908403396606,
+ "step": 10
+ },
+ {
+ "epoch": 0.15644444444444444,
+ "grad_norm": 0.012934447266161442,
+ "learning_rate": 4.5454545454545455e-06,
+ "loss": 0.46689414978027344,
+ "step": 11
+ },
+ {
+ "epoch": 0.17066666666666666,
+ "grad_norm": 0.012295619584619999,
+ "learning_rate": 5e-06,
+ "loss": 0.5062131881713867,
+ "step": 12
+ },
+ {
+ "epoch": 0.18488888888888888,
+ "grad_norm": 0.011856353841722012,
+ "learning_rate": 5.4545454545454545e-06,
+ "loss": 0.5063096880912781,
+ "step": 13
+ },
+ {
+ "epoch": 0.1991111111111111,
+ "grad_norm": 0.009451798163354397,
+ "learning_rate": 5.90909090909091e-06,
+ "loss": 0.5130602121353149,
+ "step": 14
+ },
+ {
+ "epoch": 0.21333333333333335,
+ "grad_norm": 0.01111016795039177,
+ "learning_rate": 6.363636363636364e-06,
+ "loss": 0.5016045570373535,
+ "step": 15
+ },
+ {
+ "epoch": 0.22755555555555557,
+ "grad_norm": 0.010735263116657734,
+ "learning_rate": 6.818181818181818e-06,
+ "loss": 0.49827253818511963,
+ "step": 16
+ },
+ {
+ "epoch": 0.24177777777777779,
+ "grad_norm": 0.010710126720368862,
+ "learning_rate": 7.272727272727273e-06,
+ "loss": 0.479993999004364,
+ "step": 17
+ },
+ {
+ "epoch": 0.256,
+ "grad_norm": 0.009600787423551083,
+ "learning_rate": 7.727272727272727e-06,
+ "loss": 0.4925128221511841,
+ "step": 18
+ },
+ {
+ "epoch": 0.2702222222222222,
+ "grad_norm": 0.010486789979040623,
+ "learning_rate": 8.181818181818183e-06,
+ "loss": 0.5188871622085571,
+ "step": 19
+ },
+ {
+ "epoch": 0.28444444444444444,
+ "grad_norm": 0.013396799564361572,
+ "learning_rate": 8.636363636363637e-06,
+ "loss": 0.5192517042160034,
+ "step": 20
+ },
+ {
+ "epoch": 0.2986666666666667,
+ "grad_norm": 0.009299715980887413,
+ "learning_rate": 9.090909090909091e-06,
+ "loss": 0.4580157399177551,
+ "step": 21
+ },
+ {
+ "epoch": 0.3128888888888889,
+ "grad_norm": 0.009691477753221989,
+ "learning_rate": 9.545454545454547e-06,
+ "loss": 0.49455106258392334,
+ "step": 22
+ },
+ {
+ "epoch": 0.32711111111111113,
+ "grad_norm": 0.008896945044398308,
+ "learning_rate": 1e-05,
+ "loss": 0.48672014474868774,
+ "step": 23
+ },
+ {
+ "epoch": 0.3413333333333333,
+ "grad_norm": 0.009234798140823841,
+ "learning_rate": 9.999323662872998e-06,
+ "loss": 0.5052967071533203,
+ "step": 24
+ },
+ {
+ "epoch": 0.35555555555555557,
+ "grad_norm": 0.009167873300611973,
+ "learning_rate": 9.99729483446475e-06,
+ "loss": 0.4740435779094696,
+ "step": 25
+ },
+ {
+ "epoch": 0.36977777777777776,
+ "grad_norm": 0.009001555852591991,
+ "learning_rate": 9.993914063644053e-06,
+ "loss": 0.47315946221351624,
+ "step": 26
+ },
+ {
+ "epoch": 0.384,
+ "grad_norm": 0.00915393978357315,
+ "learning_rate": 9.989182265027232e-06,
+ "loss": 0.4557237923145294,
+ "step": 27
+ },
+ {
+ "epoch": 0.3982222222222222,
+ "grad_norm": 0.008505849167704582,
+ "learning_rate": 9.98310071873072e-06,
+ "loss": 0.46317678689956665,
+ "step": 28
+ },
+ {
+ "epoch": 0.41244444444444445,
+ "grad_norm": 0.008668503724038601,
+ "learning_rate": 9.975671070024741e-06,
+ "loss": 0.4533114433288574,
+ "step": 29
+ },
+ {
+ "epoch": 0.4266666666666667,
+ "grad_norm": 0.009284070692956448,
+ "learning_rate": 9.966895328888195e-06,
+ "loss": 0.5624005794525146,
+ "step": 30
+ },
+ {
+ "epoch": 0.4408888888888889,
+ "grad_norm": 0.00935022346675396,
+ "learning_rate": 9.956775869464901e-06,
+ "loss": 0.4420597553253174,
+ "step": 31
+ },
+ {
+ "epoch": 0.45511111111111113,
+ "grad_norm": 0.008682326413691044,
+ "learning_rate": 9.945315429421307e-06,
+ "loss": 0.4745720624923706,
+ "step": 32
+ },
+ {
+ "epoch": 0.4693333333333333,
+ "grad_norm": 0.007816251367330551,
+ "learning_rate": 9.932517109205849e-06,
+ "loss": 0.439736545085907,
+ "step": 33
+ },
+ {
+ "epoch": 0.48355555555555557,
+ "grad_norm": 0.009315130300819874,
+ "learning_rate": 9.918384371210178e-06,
+ "loss": 0.5357447862625122,
+ "step": 34
+ },
+ {
+ "epoch": 0.49777777777777776,
+ "grad_norm": 0.009141312912106514,
+ "learning_rate": 9.902921038832456e-06,
+ "loss": 0.519463062286377,
+ "step": 35
+ },
+ {
+ "epoch": 0.512,
+ "grad_norm": 0.008826862089335918,
+ "learning_rate": 9.886131295443003e-06,
+ "loss": 0.48377716541290283,
+ "step": 36
+ },
+ {
+ "epoch": 0.5262222222222223,
+ "grad_norm": 0.008700878359377384,
+ "learning_rate": 9.868019683252543e-06,
+ "loss": 0.48436683416366577,
+ "step": 37
+ },
+ {
+ "epoch": 0.5404444444444444,
+ "grad_norm": 0.011412105523049831,
+ "learning_rate": 9.848591102083375e-06,
+ "loss": 0.4949312210083008,
+ "step": 38
+ },
+ {
+ "epoch": 0.5546666666666666,
+ "grad_norm": 0.009214675053954124,
+ "learning_rate": 9.82785080804381e-06,
+ "loss": 0.4901079535484314,
+ "step": 39
+ },
+ {
+ "epoch": 0.5688888888888889,
+ "grad_norm": 0.01163570862263441,
+ "learning_rate": 9.805804412106197e-06,
+ "loss": 0.6070311665534973,
+ "step": 40
+ },
+ {
+ "epoch": 0.5831111111111111,
+ "grad_norm": 0.009325762279331684,
+ "learning_rate": 9.782457878588977e-06,
+ "loss": 0.48412975668907166,
+ "step": 41
+ },
+ {
+ "epoch": 0.5973333333333334,
+ "grad_norm": 0.008266489021480083,
+ "learning_rate": 9.75781752354311e-06,
+ "loss": 0.4440661072731018,
+ "step": 42
+ },
+ {
+ "epoch": 0.6115555555555555,
+ "grad_norm": 0.00843313243240118,
+ "learning_rate": 9.731890013043367e-06,
+ "loss": 0.502442479133606,
+ "step": 43
+ },
+ {
+ "epoch": 0.6257777777777778,
+ "grad_norm": 0.00934070535004139,
+ "learning_rate": 9.704682361384941e-06,
+ "loss": 0.47924768924713135,
+ "step": 44
+ },
+ {
+ "epoch": 0.64,
+ "grad_norm": 0.010360248386859894,
+ "learning_rate": 9.676201929185809e-06,
+ "loss": 0.4161364436149597,
+ "step": 45
+ },
+ {
+ "epoch": 0.6542222222222223,
+ "grad_norm": 0.0094041982665658,
+ "learning_rate": 9.646456421395447e-06,
+ "loss": 0.48306071758270264,
+ "step": 46
+ },
+ {
+ "epoch": 0.6684444444444444,
+ "grad_norm": 0.009070150554180145,
+ "learning_rate": 9.615453885210368e-06,
+ "loss": 0.4774249792098999,
+ "step": 47
+ },
+ {
+ "epoch": 0.6826666666666666,
+ "grad_norm": 0.009771023876965046,
+ "learning_rate": 9.583202707897075e-06,
+ "loss": 0.4948921203613281,
+ "step": 48
+ },
+ {
+ "epoch": 0.6968888888888889,
+ "grad_norm": 0.008749130181968212,
+ "learning_rate": 9.549711614523007e-06,
+ "loss": 0.4825429916381836,
+ "step": 49
+ },
+ {
+ "epoch": 0.7111111111111111,
+ "grad_norm": 0.007630312815308571,
+ "learning_rate": 9.514989665596114e-06,
+ "loss": 0.46055319905281067,
+ "step": 50
+ },
+ {
+ "epoch": 0.7111111111111111,
+ "eval_loss": 0.48781517148017883,
+ "eval_runtime": 4.5903,
+ "eval_samples_per_second": 43.57,
+ "eval_steps_per_second": 21.785,
+ "step": 50
+ },
+ {
+ "epoch": 0.7253333333333334,
+ "grad_norm": 0.008408056572079659,
+ "learning_rate": 9.479046254613673e-06,
+ "loss": 0.4626312255859375,
+ "step": 51
+ },
+ {
+ "epoch": 0.7395555555555555,
+ "grad_norm": 0.007699670735746622,
+ "learning_rate": 9.441891105521005e-06,
+ "loss": 0.4499070644378662,
+ "step": 52
+ },
+ {
+ "epoch": 0.7537777777777778,
+ "grad_norm": 0.009070169180631638,
+ "learning_rate": 9.40353427008083e-06,
+ "loss": 0.4825657606124878,
+ "step": 53
+ },
+ {
+ "epoch": 0.768,
+ "grad_norm": 0.007915646769106388,
+ "learning_rate": 9.3639861251539e-06,
+ "loss": 0.4534468650817871,
+ "step": 54
+ },
+ {
+ "epoch": 0.7822222222222223,
+ "grad_norm": 0.011336776427924633,
+ "learning_rate": 9.323257369891702e-06,
+ "loss": 0.4988631010055542,
+ "step": 55
+ },
+ {
+ "epoch": 0.7964444444444444,
+ "grad_norm": 0.00761911366134882,
+ "learning_rate": 9.281359022841966e-06,
+ "loss": 0.4371437728404999,
+ "step": 56
+ },
+ {
+ "epoch": 0.8106666666666666,
+ "grad_norm": 0.009316994808614254,
+ "learning_rate": 9.238302418967757e-06,
+ "loss": 0.4956932067871094,
+ "step": 57
+ },
+ {
+ "epoch": 0.8248888888888889,
+ "grad_norm": 0.009821794927120209,
+ "learning_rate": 9.194099206580981e-06,
+ "loss": 0.5392974615097046,
+ "step": 58
+ },
+ {
+ "epoch": 0.8391111111111111,
+ "grad_norm": 0.008422416634857655,
+ "learning_rate": 9.14876134419111e-06,
+ "loss": 0.47983357310295105,
+ "step": 59
+ },
+ {
+ "epoch": 0.8533333333333334,
+ "grad_norm": 0.00837004091590643,
+ "learning_rate": 9.102301097269974e-06,
+ "loss": 0.5183510780334473,
+ "step": 60
+ },
+ {
+ "epoch": 0.8675555555555555,
+ "grad_norm": 0.008596297353506088,
+ "learning_rate": 9.05473103493355e-06,
+ "loss": 0.47292786836624146,
+ "step": 61
+ },
+ {
+ "epoch": 0.8817777777777778,
+ "grad_norm": 0.009059637784957886,
+ "learning_rate": 9.006064026541549e-06,
+ "loss": 0.5188573598861694,
+ "step": 62
+ },
+ {
+ "epoch": 0.896,
+ "grad_norm": 0.007816917262971401,
+ "learning_rate": 8.956313238215824e-06,
+ "loss": 0.45641395449638367,
+ "step": 63
+ },
+ {
+ "epoch": 0.9102222222222223,
+ "grad_norm": 0.007985202595591545,
+ "learning_rate": 8.905492129278478e-06,
+ "loss": 0.4586361050605774,
+ "step": 64
+ },
+ {
+ "epoch": 0.9244444444444444,
+ "grad_norm": 0.007170964498072863,
+ "learning_rate": 8.85361444861063e-06,
+ "loss": 0.4394378066062927,
+ "step": 65
+ },
+ {
+ "epoch": 0.9386666666666666,
+ "grad_norm": 0.00909162126481533,
+ "learning_rate": 8.800694230932885e-06,
+ "loss": 0.509820818901062,
+ "step": 66
+ },
+ {
+ "epoch": 0.9528888888888889,
+ "grad_norm": 0.00786044355481863,
+ "learning_rate": 8.74674579300843e-06,
+ "loss": 0.43584880232810974,
+ "step": 67
+ },
+ {
+ "epoch": 0.9671111111111111,
+ "grad_norm": 0.00832363311201334,
+ "learning_rate": 8.691783729769874e-06,
+ "loss": 0.4317135214805603,
+ "step": 68
+ },
+ {
+ "epoch": 0.9813333333333333,
+ "grad_norm": 0.008376957848668098,
+ "learning_rate": 8.635822910370793e-06,
+ "loss": 0.49933016300201416,
+ "step": 69
+ },
+ {
+ "epoch": 0.9955555555555555,
+ "grad_norm": 0.00835558120161295,
+ "learning_rate": 8.578878474163115e-06,
+ "loss": 0.46570393443107605,
+ "step": 70
+ },
+ {
+ "epoch": 1.0,
+ "grad_norm": 0.013256405480206013,
+ "learning_rate": 8.520965826601394e-06,
+ "loss": 0.45107918977737427,
+ "step": 71
+ },
+ {
+ "epoch": 1.0142222222222221,
+ "grad_norm": 0.008708071894943714,
+ "learning_rate": 8.462100635075097e-06,
+ "loss": 0.47125527262687683,
+ "step": 72
+ },
+ {
+ "epoch": 1.0284444444444445,
+ "grad_norm": 0.008086243644356728,
+ "learning_rate": 8.40229882467003e-06,
+ "loss": 0.4322492480278015,
+ "step": 73
+ },
+ {
+ "epoch": 1.0426666666666666,
+ "grad_norm": 0.0073950947262346745,
+ "learning_rate": 8.341576573860049e-06,
+ "loss": 0.38362830877304077,
+ "step": 74
+ },
+ {
+ "epoch": 1.056888888888889,
+ "grad_norm": 0.0076777078211307526,
+ "learning_rate": 8.279950310130218e-06,
+ "loss": 0.3878062665462494,
+ "step": 75
+ },
+ {
+ "epoch": 1.0711111111111111,
+ "grad_norm": 0.007898081094026566,
+ "learning_rate": 8.2174367055326e-06,
+ "loss": 0.3833811283111572,
+ "step": 76
+ },
+ {
+ "epoch": 1.0853333333333333,
+ "grad_norm": 0.007348628249019384,
+ "learning_rate": 8.154052672175888e-06,
+ "loss": 0.3845909833908081,
+ "step": 77
+ },
+ {
+ "epoch": 1.0995555555555556,
+ "grad_norm": 0.008542248979210854,
+ "learning_rate": 8.08981535765009e-06,
+ "loss": 0.44259265065193176,
+ "step": 78
+ },
+ {
+ "epoch": 1.1137777777777778,
+ "grad_norm": 0.007998447865247726,
+ "learning_rate": 8.024742140387506e-06,
+ "loss": 0.38503497838974,
+ "step": 79
+ },
+ {
+ "epoch": 1.1280000000000001,
+ "grad_norm": 0.008362009190022945,
+ "learning_rate": 7.95885062496126e-06,
+ "loss": 0.42608046531677246,
+ "step": 80
+ },
+ {
+ "epoch": 1.1422222222222222,
+ "grad_norm": 0.007770548574626446,
+ "learning_rate": 7.892158637322647e-06,
+ "loss": 0.38991105556488037,
+ "step": 81
+ },
+ {
+ "epoch": 1.1564444444444444,
+ "grad_norm": 0.007649772334843874,
+ "learning_rate": 7.824684219978591e-06,
+ "loss": 0.36854323744773865,
+ "step": 82
+ },
+ {
+ "epoch": 1.1706666666666667,
+ "grad_norm": 0.008056909777224064,
+ "learning_rate": 7.756445627110523e-06,
+ "loss": 0.4200327396392822,
+ "step": 83
+ },
+ {
+ "epoch": 1.1848888888888889,
+ "grad_norm": 0.0089965695515275,
+ "learning_rate": 7.68746131963598e-06,
+ "loss": 0.4378758668899536,
+ "step": 84
+ },
+ {
+ "epoch": 1.199111111111111,
+ "grad_norm": 0.007884320802986622,
+ "learning_rate": 7.6177499602143e-06,
+ "loss": 0.3808164894580841,
+ "step": 85
+ },
+ {
+ "epoch": 1.2133333333333334,
+ "grad_norm": 0.009615465067327023,
+ "learning_rate": 7.547330408197695e-06,
+ "loss": 0.4453170895576477,
+ "step": 86
+ },
+ {
+ "epoch": 1.2275555555555555,
+ "grad_norm": 0.00891174841672182,
+ "learning_rate": 7.476221714529167e-06,
+ "loss": 0.4456181526184082,
+ "step": 87
+ },
+ {
+ "epoch": 1.2417777777777779,
+ "grad_norm": 0.008833018131554127,
+ "learning_rate": 7.404443116588548e-06,
+ "loss": 0.43805548548698425,
+ "step": 88
+ },
+ {
+ "epoch": 1.256,
+ "grad_norm": 0.008171891793608665,
+ "learning_rate": 7.332014032988123e-06,
+ "loss": 0.4144285321235657,
+ "step": 89
+ },
+ {
+ "epoch": 1.2702222222222221,
+ "grad_norm": 0.009090668521821499,
+ "learning_rate": 7.2589540583192165e-06,
+ "loss": 0.38928651809692383,
+ "step": 90
+ },
+ {
+ "epoch": 1.2844444444444445,
+ "grad_norm": 0.008077106438577175,
+ "learning_rate": 7.185282957851175e-06,
+ "loss": 0.387723445892334,
+ "step": 91
+ },
+ {
+ "epoch": 1.2986666666666666,
+ "grad_norm": 0.008803839795291424,
+ "learning_rate": 7.111020662184174e-06,
+ "loss": 0.4158015549182892,
+ "step": 92
+ },
+ {
+ "epoch": 1.3128888888888888,
+ "grad_norm": 0.008249706588685513,
+ "learning_rate": 7.036187261857289e-06,
+ "loss": 0.39764612913131714,
+ "step": 93
+ },
+ {
+ "epoch": 1.3271111111111111,
+ "grad_norm": 0.008549178950488567,
+ "learning_rate": 6.960803001913315e-06,
+ "loss": 0.38216379284858704,
+ "step": 94
+ },
+ {
+ "epoch": 1.3413333333333333,
+ "grad_norm": 0.007755788508802652,
+ "learning_rate": 6.884888276421766e-06,
+ "loss": 0.413602352142334,
+ "step": 95
+ },
+ {
+ "epoch": 1.3555555555555556,
+ "grad_norm": 0.008742124773561954,
+ "learning_rate": 6.8084636229615786e-06,
+ "loss": 0.41756486892700195,
+ "step": 96
+ },
+ {
+ "epoch": 1.3697777777777778,
+ "grad_norm": 0.008473540656268597,
+ "learning_rate": 6.731549717064975e-06,
+ "loss": 0.4028819799423218,
+ "step": 97
+ },
+ {
+ "epoch": 1.384,
+ "grad_norm": 0.008585155010223389,
+ "learning_rate": 6.654167366624009e-06,
+ "loss": 0.37123921513557434,
+ "step": 98
+ },
+ {
+ "epoch": 1.3982222222222223,
+ "grad_norm": 0.00882649328559637,
+ "learning_rate": 6.576337506261314e-06,
+ "loss": 0.39443182945251465,
+ "step": 99
+ },
+ {
+ "epoch": 1.4124444444444444,
+ "grad_norm": 0.008174613118171692,
+ "learning_rate": 6.498081191666549e-06,
+ "loss": 0.42232874035835266,
+ "step": 100
+ },
+ {
+ "epoch": 1.4124444444444444,
+ "eval_loss": 0.4897068440914154,
+ "eval_runtime": 5.0728,
+ "eval_samples_per_second": 39.426,
+ "eval_steps_per_second": 19.713,
+ "step": 100
+ },
+ {
+ "epoch": 1.4266666666666667,
+ "grad_norm": 0.00753920478746295,
+ "learning_rate": 6.419419593900109e-06,
+ "loss": 0.38927626609802246,
+ "step": 101
+ },
+ {
+ "epoch": 1.4408888888888889,
+ "grad_norm": 0.007967391051352024,
+ "learning_rate": 6.340373993665607e-06,
+ "loss": 0.36544570326805115,
+ "step": 102
+ },
+ {
+ "epoch": 1.455111111111111,
+ "grad_norm": 0.008992813527584076,
+ "learning_rate": 6.2609657755527135e-06,
+ "loss": 0.4033282995223999,
+ "step": 103
+ },
+ {
+ "epoch": 1.4693333333333334,
+ "grad_norm": 0.007906445302069187,
+ "learning_rate": 6.1812164222518626e-06,
+ "loss": 0.40876060724258423,
+ "step": 104
+ },
+ {
+ "epoch": 1.4835555555555555,
+ "grad_norm": 0.00823493767529726,
+ "learning_rate": 6.101147508742456e-06,
+ "loss": 0.3994408845901489,
+ "step": 105
+ },
+ {
+ "epoch": 1.4977777777777779,
+ "grad_norm": 0.009032925590872765,
+ "learning_rate": 6.020780696456059e-06,
+ "loss": 0.4667425751686096,
+ "step": 106
+ },
+ {
+ "epoch": 1.512,
+ "grad_norm": 0.007943308912217617,
+ "learning_rate": 5.940137727416247e-06,
+ "loss": 0.4092869460582733,
+ "step": 107
+ },
+ {
+ "epoch": 1.5262222222222221,
+ "grad_norm": 0.008353099226951599,
+ "learning_rate": 5.859240418356614e-06,
+ "loss": 0.39832064509391785,
+ "step": 108
+ },
+ {
+ "epoch": 1.5404444444444443,
+ "grad_norm": 0.008241581730544567,
+ "learning_rate": 5.778110654818602e-06,
+ "loss": 0.4114166498184204,
+ "step": 109
+ },
+ {
+ "epoch": 1.5546666666666666,
+ "grad_norm": 0.007937194779515266,
+ "learning_rate": 5.696770385230679e-06,
+ "loss": 0.4360125660896301,
+ "step": 110
+ },
+ {
+ "epoch": 1.568888888888889,
+ "grad_norm": 0.008319812826812267,
+ "learning_rate": 5.615241614970546e-06,
+ "loss": 0.4153256118297577,
+ "step": 111
+ },
+ {
+ "epoch": 1.5831111111111111,
+ "grad_norm": 0.008011740632355213,
+ "learning_rate": 5.533546400411899e-06,
+ "loss": 0.3954656720161438,
+ "step": 112
+ },
+ {
+ "epoch": 1.5973333333333333,
+ "grad_norm": 0.008084246888756752,
+ "learning_rate": 5.4517068429574215e-06,
+ "loss": 0.4052407741546631,
+ "step": 113
+ },
+ {
+ "epoch": 1.6115555555555554,
+ "grad_norm": 0.007481739390641451,
+ "learning_rate": 5.3697450830595775e-06,
+ "loss": 0.4017290472984314,
+ "step": 114
+ },
+ {
+ "epoch": 1.6257777777777778,
+ "grad_norm": 0.008050751872360706,
+ "learning_rate": 5.287683294230855e-06,
+ "loss": 0.38467150926589966,
+ "step": 115
+ },
+ {
+ "epoch": 1.6400000000000001,
+ "grad_norm": 0.0077187251299619675,
+ "learning_rate": 5.20554367704505e-06,
+ "loss": 0.395123690366745,
+ "step": 116
+ },
+ {
+ "epoch": 1.6542222222222223,
+ "grad_norm": 0.008423185907304287,
+ "learning_rate": 5.123348453131242e-06,
+ "loss": 0.39796996116638184,
+ "step": 117
+ },
+ {
+ "epoch": 1.6684444444444444,
+ "grad_norm": 0.009052549488842487,
+ "learning_rate": 5.041119859162068e-06,
+ "loss": 0.4162113070487976,
+ "step": 118
+ },
+ {
+ "epoch": 1.6826666666666665,
+ "grad_norm": 0.008246202953159809,
+ "learning_rate": 4.958880140837934e-06,
+ "loss": 0.42128854990005493,
+ "step": 119
+ },
+ {
+ "epoch": 1.696888888888889,
+ "grad_norm": 0.00833312701433897,
+ "learning_rate": 4.876651546868759e-06,
+ "loss": 0.4024372696876526,
+ "step": 120
+ },
+ {
+ "epoch": 1.7111111111111112,
+ "grad_norm": 0.008461604826152325,
+ "learning_rate": 4.794456322954953e-06,
+ "loss": 0.46918508410453796,
+ "step": 121
+ },
+ {
+ "epoch": 1.7253333333333334,
+ "grad_norm": 0.00825911108404398,
+ "learning_rate": 4.7123167057691446e-06,
+ "loss": 0.41863948106765747,
+ "step": 122
+ },
+ {
+ "epoch": 1.7395555555555555,
+ "grad_norm": 0.009252563118934631,
+ "learning_rate": 4.630254916940424e-06,
+ "loss": 0.44644808769226074,
+ "step": 123
+ },
+ {
+ "epoch": 1.7537777777777777,
+ "grad_norm": 0.007843364961445332,
+ "learning_rate": 4.548293157042581e-06,
+ "loss": 0.42086148262023926,
+ "step": 124
+ },
+ {
+ "epoch": 1.768,
+ "grad_norm": 0.007915403693914413,
+ "learning_rate": 4.466453599588103e-06,
+ "loss": 0.4266040027141571,
+ "step": 125
+ },
+ {
+ "epoch": 1.7822222222222224,
+ "grad_norm": 0.007302967831492424,
+ "learning_rate": 4.384758385029457e-06,
+ "loss": 0.3592030704021454,
+ "step": 126
+ },
+ {
+ "epoch": 1.7964444444444445,
+ "grad_norm": 0.00869770348072052,
+ "learning_rate": 4.3032296147693225e-06,
+ "loss": 0.42212432622909546,
+ "step": 127
+ },
+ {
+ "epoch": 1.8106666666666666,
+ "grad_norm": 0.008091778494417667,
+ "learning_rate": 4.2218893451814005e-06,
+ "loss": 0.37964335083961487,
+ "step": 128
+ },
+ {
+ "epoch": 1.8248888888888888,
+ "grad_norm": 0.0076453504152596,
+ "learning_rate": 4.140759581643386e-06,
+ "loss": 0.37467944622039795,
+ "step": 129
+ },
+ {
+ "epoch": 1.8391111111111111,
+ "grad_norm": 0.007656064350157976,
+ "learning_rate": 4.059862272583755e-06,
+ "loss": 0.393369197845459,
+ "step": 130
+ },
+ {
+ "epoch": 1.8533333333333335,
+ "grad_norm": 0.008399168029427528,
+ "learning_rate": 3.979219303543942e-06,
+ "loss": 0.394899845123291,
+ "step": 131
+ },
+ {
+ "epoch": 1.8675555555555556,
+ "grad_norm": 0.008833736181259155,
+ "learning_rate": 3.898852491257547e-06,
+ "loss": 0.3997966945171356,
+ "step": 132
+ },
+ {
+ "epoch": 1.8817777777777778,
+ "grad_norm": 0.008364683948457241,
+ "learning_rate": 3.818783577748138e-06,
+ "loss": 0.44034862518310547,
+ "step": 133
+ },
+ {
+ "epoch": 1.896,
+ "grad_norm": 0.008851971477270126,
+ "learning_rate": 3.7390342244472886e-06,
+ "loss": 0.4251548647880554,
+ "step": 134
+ },
+ {
+ "epoch": 1.9102222222222223,
+ "grad_norm": 0.007921694777905941,
+ "learning_rate": 3.659626006334395e-06,
+ "loss": 0.3972621560096741,
+ "step": 135
+ },
+ {
+ "epoch": 1.9244444444444444,
+ "grad_norm": 0.008714950643479824,
+ "learning_rate": 3.5805804060998926e-06,
+ "loss": 0.4237692356109619,
+ "step": 136
+ },
+ {
+ "epoch": 1.9386666666666668,
+ "grad_norm": 0.008546526543796062,
+ "learning_rate": 3.501918808333453e-06,
+ "loss": 0.5015424489974976,
+ "step": 137
+ },
+ {
+ "epoch": 1.952888888888889,
+ "grad_norm": 0.008581602945923805,
+ "learning_rate": 3.4236624937386874e-06,
+ "loss": 0.4541424512863159,
+ "step": 138
+ },
+ {
+ "epoch": 1.967111111111111,
+ "grad_norm": 0.008913579396903515,
+ "learning_rate": 3.3458326333759927e-06,
+ "loss": 0.42448267340660095,
+ "step": 139
+ },
+ {
+ "epoch": 1.9813333333333332,
+ "grad_norm": 0.008540489710867405,
+ "learning_rate": 3.268450282935026e-06,
+ "loss": 0.4514390528202057,
+ "step": 140
+ },
+ {
+ "epoch": 1.9955555555555555,
+ "grad_norm": 0.007863311097025871,
+ "learning_rate": 3.1915363770384223e-06,
+ "loss": 0.38879817724227905,
+ "step": 141
+ },
+ {
+ "epoch": 2.0,
+ "grad_norm": 0.013070300221443176,
+ "learning_rate": 3.1151117235782346e-06,
+ "loss": 0.3734951913356781,
+ "step": 142
+ },
+ {
+ "epoch": 2.014222222222222,
+ "grad_norm": 0.00914511177688837,
+ "learning_rate": 3.0391969980866874e-06,
+ "loss": 0.3868422210216522,
+ "step": 143
+ },
+ {
+ "epoch": 2.0284444444444443,
+ "grad_norm": 0.008015231229364872,
+ "learning_rate": 2.963812738142713e-06,
+ "loss": 0.3441769480705261,
+ "step": 144
+ },
+ {
+ "epoch": 2.042666666666667,
+ "grad_norm": 0.008334742859005928,
+ "learning_rate": 2.8889793378158284e-06,
+ "loss": 0.3698546290397644,
+ "step": 145
+ },
+ {
+ "epoch": 2.056888888888889,
+ "grad_norm": 0.0077324071899056435,
+ "learning_rate": 2.814717042148827e-06,
+ "loss": 0.3752461075782776,
+ "step": 146
+ },
+ {
+ "epoch": 2.071111111111111,
+ "grad_norm": 0.007629899773746729,
+ "learning_rate": 2.7410459416807856e-06,
+ "loss": 0.3570592999458313,
+ "step": 147
+ },
+ {
+ "epoch": 2.0853333333333333,
+ "grad_norm": 0.008662652224302292,
+ "learning_rate": 2.6679859670118785e-06,
+ "loss": 0.3812662363052368,
+ "step": 148
+ },
+ {
+ "epoch": 2.0995555555555554,
+ "grad_norm": 0.007828332483768463,
+ "learning_rate": 2.5955568834114523e-06,
+ "loss": 0.3505663275718689,
+ "step": 149
+ },
+ {
+ "epoch": 2.113777777777778,
+ "grad_norm": 0.0074827345088124275,
+ "learning_rate": 2.523778285470835e-06,
+ "loss": 0.35809028148651123,
+ "step": 150
+ },
+ {
+ "epoch": 2.113777777777778,
+ "eval_loss": 0.49233150482177734,
+ "eval_runtime": 4.1869,
+ "eval_samples_per_second": 47.768,
+ "eval_steps_per_second": 23.884,
+ "step": 150
+ },
+ {
+ "epoch": 2.128,
+ "grad_norm": 0.007365878205746412,
+ "learning_rate": 2.452669591802307e-06,
+ "loss": 0.32126089930534363,
+ "step": 151
+ },
+ {
+ "epoch": 2.1422222222222222,
+ "grad_norm": 0.007244541309773922,
+ "learning_rate": 2.3822500397857016e-06,
+ "loss": 0.3627392649650574,
+ "step": 152
+ },
+ {
+ "epoch": 2.1564444444444444,
+ "grad_norm": 0.00791338924318552,
+ "learning_rate": 2.3125386803640188e-06,
+ "loss": 0.36303800344467163,
+ "step": 153
+ },
+ {
+ "epoch": 2.1706666666666665,
+ "grad_norm": 0.007537501864135265,
+ "learning_rate": 2.243554372889479e-06,
+ "loss": 0.3294636011123657,
+ "step": 154
+ },
+ {
+ "epoch": 2.1848888888888887,
+ "grad_norm": 0.008005894720554352,
+ "learning_rate": 2.1753157800214107e-06,
+ "loss": 0.35944664478302,
+ "step": 155
+ },
+ {
+ "epoch": 2.1991111111111112,
+ "grad_norm": 0.008430157788097858,
+ "learning_rate": 2.1078413626773547e-06,
+ "loss": 0.35921794176101685,
+ "step": 156
+ },
+ {
+ "epoch": 2.2133333333333334,
+ "grad_norm": 0.00803217850625515,
+ "learning_rate": 2.0411493750387423e-06,
+ "loss": 0.3335600197315216,
+ "step": 157
+ },
+ {
+ "epoch": 2.2275555555555555,
+ "grad_norm": 0.007777590304613113,
+ "learning_rate": 1.9752578596124955e-06,
+ "loss": 0.3621842861175537,
+ "step": 158
+ },
+ {
+ "epoch": 2.2417777777777776,
+ "grad_norm": 0.007962247356772423,
+ "learning_rate": 1.9101846423499113e-06,
+ "loss": 0.34454649686813354,
+ "step": 159
+ },
+ {
+ "epoch": 2.2560000000000002,
+ "grad_norm": 0.007974164560437202,
+ "learning_rate": 1.8459473278241125e-06,
+ "loss": 0.3537226915359497,
+ "step": 160
+ },
+ {
+ "epoch": 2.2702222222222224,
+ "grad_norm": 0.007959051057696342,
+ "learning_rate": 1.7825632944674016e-06,
+ "loss": 0.3517017662525177,
+ "step": 161
+ },
+ {
+ "epoch": 2.2844444444444445,
+ "grad_norm": 0.008884078823029995,
+ "learning_rate": 1.7200496898697832e-06,
+ "loss": 0.34244149923324585,
+ "step": 162
+ },
+ {
+ "epoch": 2.2986666666666666,
+ "grad_norm": 0.008463534526526928,
+ "learning_rate": 1.6584234261399535e-06,
+ "loss": 0.35110828280448914,
+ "step": 163
+ },
+ {
+ "epoch": 2.3128888888888888,
+ "grad_norm": 0.008407258428633213,
+ "learning_rate": 1.5977011753299726e-06,
+ "loss": 0.34295642375946045,
+ "step": 164
+ },
+ {
+ "epoch": 2.327111111111111,
+ "grad_norm": 0.008790143765509129,
+ "learning_rate": 1.5378993649249053e-06,
+ "loss": 0.4045950770378113,
+ "step": 165
+ },
+ {
+ "epoch": 2.3413333333333335,
+ "grad_norm": 0.008037876337766647,
+ "learning_rate": 1.4790341733986085e-06,
+ "loss": 0.3649445176124573,
+ "step": 166
+ },
+ {
+ "epoch": 2.3555555555555556,
+ "grad_norm": 0.007316106464713812,
+ "learning_rate": 1.4211215258368866e-06,
+ "loss": 0.3201165795326233,
+ "step": 167
+ },
+ {
+ "epoch": 2.3697777777777778,
+ "grad_norm": 0.00920423399657011,
+ "learning_rate": 1.3641770896292083e-06,
+ "loss": 0.345939040184021,
+ "step": 168
+ },
+ {
+ "epoch": 2.384,
+ "grad_norm": 0.008216304704546928,
+ "learning_rate": 1.3082162702301276e-06,
+ "loss": 0.35247528553009033,
+ "step": 169
+ },
+ {
+ "epoch": 2.398222222222222,
+ "grad_norm": 0.008257976733148098,
+ "learning_rate": 1.2532542069915722e-06,
+ "loss": 0.40231946110725403,
+ "step": 170
+ },
+ {
+ "epoch": 2.4124444444444446,
+ "grad_norm": 0.008689168840646744,
+ "learning_rate": 1.1993057690671174e-06,
+ "loss": 0.3605418801307678,
+ "step": 171
+ },
+ {
+ "epoch": 2.4266666666666667,
+ "grad_norm": 0.008199715055525303,
+ "learning_rate": 1.1463855513893695e-06,
+ "loss": 0.382945716381073,
+ "step": 172
+ },
+ {
+ "epoch": 2.440888888888889,
+ "grad_norm": 0.00789699237793684,
+ "learning_rate": 1.0945078707215224e-06,
+ "loss": 0.36510083079338074,
+ "step": 173
+ },
+ {
+ "epoch": 2.455111111111111,
+ "grad_norm": 0.008184579201042652,
+ "learning_rate": 1.0436867617841768e-06,
+ "loss": 0.38933348655700684,
+ "step": 174
+ },
+ {
+ "epoch": 2.469333333333333,
+ "grad_norm": 0.007531109265983105,
+ "learning_rate": 9.939359734584552e-07,
+ "loss": 0.3710438013076782,
+ "step": 175
+ },
+ {
+ "epoch": 2.4835555555555557,
+ "grad_norm": 0.008019962348043919,
+ "learning_rate": 9.452689650664515e-07,
+ "loss": 0.36227256059646606,
+ "step": 176
+ },
+ {
+ "epoch": 2.497777777777778,
+ "grad_norm": 0.007935193367302418,
+ "learning_rate": 8.976989027300265e-07,
+ "loss": 0.36987176537513733,
+ "step": 177
+ },
+ {
+ "epoch": 2.512,
+ "grad_norm": 0.007968267425894737,
+ "learning_rate": 8.512386558088919e-07,
+ "loss": 0.35104453563690186,
+ "step": 178
+ },
+ {
+ "epoch": 2.526222222222222,
+ "grad_norm": 0.008082241751253605,
+ "learning_rate": 8.059007934190194e-07,
+ "loss": 0.3681498169898987,
+ "step": 179
+ },
+ {
+ "epoch": 2.5404444444444443,
+ "grad_norm": 0.00833115540444851,
+ "learning_rate": 7.61697581032243e-07,
+ "loss": 0.353618860244751,
+ "step": 180
+ },
+ {
+ "epoch": 2.554666666666667,
+ "grad_norm": 0.008011356927454472,
+ "learning_rate": 7.186409771580355e-07,
+ "loss": 0.35408419370651245,
+ "step": 181
+ },
+ {
+ "epoch": 2.568888888888889,
+ "grad_norm": 0.00763704115524888,
+ "learning_rate": 6.76742630108298e-07,
+ "loss": 0.3726443648338318,
+ "step": 182
+ },
+ {
+ "epoch": 2.583111111111111,
+ "grad_norm": 0.007385968696326017,
+ "learning_rate": 6.360138748461015e-07,
+ "loss": 0.3561905026435852,
+ "step": 183
+ },
+ {
+ "epoch": 2.5973333333333333,
+ "grad_norm": 0.00736213056370616,
+ "learning_rate": 5.964657299191712e-07,
+ "loss": 0.3148964047431946,
+ "step": 184
+ },
+ {
+ "epoch": 2.6115555555555554,
+ "grad_norm": 0.008179187774658203,
+ "learning_rate": 5.581088944789953e-07,
+ "loss": 0.3561857342720032,
+ "step": 185
+ },
+ {
+ "epoch": 2.6257777777777775,
+ "grad_norm": 0.008100884966552258,
+ "learning_rate": 5.20953745386329e-07,
+ "loss": 0.3500719964504242,
+ "step": 186
+ },
+ {
+ "epoch": 2.64,
+ "grad_norm": 0.007978768087923527,
+ "learning_rate": 4.850103344038853e-07,
+ "loss": 0.3637757897377014,
+ "step": 187
+ },
+ {
+ "epoch": 2.6542222222222223,
+ "grad_norm": 0.007483206689357758,
+ "learning_rate": 4.502883854769935e-07,
+ "loss": 0.32206273078918457,
+ "step": 188
+ },
+ {
+ "epoch": 2.6684444444444444,
+ "grad_norm": 0.007899755612015724,
+ "learning_rate": 4.167972921029262e-07,
+ "loss": 0.33746200799942017,
+ "step": 189
+ },
+ {
+ "epoch": 2.6826666666666665,
+ "grad_norm": 0.008544563315808773,
+ "learning_rate": 3.8454611478963235e-07,
+ "loss": 0.38060814142227173,
+ "step": 190
+ },
+ {
+ "epoch": 2.696888888888889,
+ "grad_norm": 0.00889600533992052,
+ "learning_rate": 3.535435786045538e-07,
+ "loss": 0.3551645278930664,
+ "step": 191
+ },
+ {
+ "epoch": 2.7111111111111112,
+ "grad_norm": 0.007875156588852406,
+ "learning_rate": 3.237980708141919e-07,
+ "loss": 0.35464680194854736,
+ "step": 192
+ },
+ {
+ "epoch": 2.7253333333333334,
+ "grad_norm": 0.008730563335120678,
+ "learning_rate": 2.9531763861505967e-07,
+ "loss": 0.3980952501296997,
+ "step": 193
+ },
+ {
+ "epoch": 2.7395555555555555,
+ "grad_norm": 0.008214394561946392,
+ "learning_rate": 2.681099869566328e-07,
+ "loss": 0.33311402797698975,
+ "step": 194
+ },
+ {
+ "epoch": 2.7537777777777777,
+ "grad_norm": 0.007815193384885788,
+ "learning_rate": 2.4218247645689306e-07,
+ "loss": 0.36939185857772827,
+ "step": 195
+ },
+ {
+ "epoch": 2.768,
+ "grad_norm": 0.008210226893424988,
+ "learning_rate": 2.175421214110235e-07,
+ "loss": 0.3736470937728882,
+ "step": 196
+ },
+ {
+ "epoch": 2.7822222222222224,
+ "grad_norm": 0.007955148816108704,
+ "learning_rate": 1.941955878938029e-07,
+ "loss": 0.35068604350090027,
+ "step": 197
+ },
+ {
+ "epoch": 2.7964444444444445,
+ "grad_norm": 0.008125124499201775,
+ "learning_rate": 1.721491919561913e-07,
+ "loss": 0.3597073554992676,
+ "step": 198
+ },
+ {
+ "epoch": 2.8106666666666666,
+ "grad_norm": 0.007504038512706757,
+ "learning_rate": 1.514088979166256e-07,
+ "loss": 0.3420804440975189,
+ "step": 199
+ },
+ {
+ "epoch": 2.824888888888889,
+ "grad_norm": 0.007838167250156403,
+ "learning_rate": 1.3198031674745814e-07,
+ "loss": 0.3697103261947632,
+ "step": 200
+ },
+ {
+ "epoch": 2.824888888888889,
+ "eval_loss": 0.49903449416160583,
+ "eval_runtime": 4.213,
+ "eval_samples_per_second": 47.473,
+ "eval_steps_per_second": 23.736,
+ "step": 200
+ },
+ {
+ "epoch": 2.8391111111111114,
+ "grad_norm": 0.008086729794740677,
+ "learning_rate": 1.138687045569975e-07,
+ "loss": 0.35418957471847534,
+ "step": 201
+ },
+ {
+ "epoch": 2.8533333333333335,
+ "grad_norm": 0.007895248010754585,
+ "learning_rate": 9.70789611675449e-08,
+ "loss": 0.3809206783771515,
+ "step": 202
+ },
+ {
+ "epoch": 2.8675555555555556,
+ "grad_norm": 0.007653921376913786,
+ "learning_rate": 8.161562878982399e-08,
+ "loss": 0.33729785680770874,
+ "step": 203
+ },
+ {
+ "epoch": 2.8817777777777778,
+ "grad_norm": 0.00890001654624939,
+ "learning_rate": 6.74828907941516e-08,
+ "loss": 0.3682692050933838,
+ "step": 204
+ },
+ {
+ "epoch": 2.896,
+ "grad_norm": 0.008669495582580566,
+ "learning_rate": 5.468457057869358e-08,
+ "loss": 0.40091854333877563,
+ "step": 205
+ },
+ {
+ "epoch": 2.910222222222222,
+ "grad_norm": 0.00770574901252985,
+ "learning_rate": 4.322413053509944e-08,
+ "loss": 0.338018000125885,
+ "step": 206
+ },
+ {
+ "epoch": 2.924444444444444,
+ "grad_norm": 0.008314141072332859,
+ "learning_rate": 3.31046711118066e-08,
+ "loss": 0.3696953058242798,
+ "step": 207
+ },
+ {
+ "epoch": 2.9386666666666668,
+ "grad_norm": 0.007442758418619633,
+ "learning_rate": 2.4328929975260262e-08,
+ "loss": 0.3460949659347534,
+ "step": 208
+ },
+ {
+ "epoch": 2.952888888888889,
+ "grad_norm": 0.008314011618494987,
+ "learning_rate": 1.6899281269279756e-08,
+ "loss": 0.35688620805740356,
+ "step": 209
+ },
+ {
+ "epoch": 2.967111111111111,
+ "grad_norm": 0.007803616113960743,
+ "learning_rate": 1.0817734972768946e-08,
+ "loss": 0.39211517572402954,
+ "step": 210
+ },
+ {
+ "epoch": 2.981333333333333,
+ "grad_norm": 0.008699736557900906,
+ "learning_rate": 6.085936355947897e-09,
+ "loss": 0.38052815198898315,
+ "step": 211
+ },
+ {
+ "epoch": 2.9955555555555557,
+ "grad_norm": 0.007957417517900467,
+ "learning_rate": 2.7051655352494654e-09,
+ "loss": 0.3612944185733795,
+ "step": 212
+ },
+ {
+ "epoch": 3.0,
+ "grad_norm": 0.014429433271288872,
+ "learning_rate": 6.763371270035457e-10,
+ "loss": 0.5174569487571716,
+ "step": 213
+ }
+ ],
+ "logging_steps": 1,
+ "max_steps": 213,
+ "num_input_tokens_seen": 0,
+ "num_train_epochs": 3,
+ "save_steps": 500,
+ "stateful_callbacks": {
+ "TrainerControl": {
+ "args": {
+ "should_epoch_stop": false,
+ "should_evaluate": false,
+ "should_log": false,
+ "should_save": true,
+ "should_training_stop": true
+ },
+ "attributes": {}
+ }
+ },
+ "total_flos": 4.868007593928294e+16,
+ "train_batch_size": 1,
+ "trial_name": null,
+ "trial_params": null
+}
diff --git a/qwen2.5-1.5b-instruct/checkpoint-213/training_args.bin b/qwen2.5-1.5b-instruct/checkpoint-213/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..1c92bb7fdd3ef14952372d452eaf2c7d0ade0ccf
--- /dev/null
+++ b/qwen2.5-1.5b-instruct/checkpoint-213/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:841e251fafc8c15cdc7b32aa47df63c6ea8f5d8eb459fbecdbb2a9f5f74be033
+size 5649
diff --git a/qwen2.5-1.5b-instruct/checkpoint-71/chat_template.jinja b/qwen2.5-1.5b-instruct/checkpoint-71/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..bdf7919a96cfe43d50914a007b9c0877bd0ec27e
--- /dev/null
+++ b/qwen2.5-1.5b-instruct/checkpoint-71/chat_template.jinja
@@ -0,0 +1,54 @@
+{%- if tools %}
+ {{- '<|im_start|>system\n' }}
+ {%- if messages[0]['role'] == 'system' %}
+ {{- messages[0]['content'] }}
+ {%- else %}
+ {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+ {%- endif %}
+ {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within XML tags:\n" }}
+ {%- for tool in tools %}
+ {{- "\n" }}
+ {{- tool | tojson }}
+ {%- endfor %}
+ {{- "\n\n\nFor each function call, return a json object with function name and arguments within XML tags:\n\n{\"name\": , \"arguments\": }\n<|im_end|>\n" }}
+{%- else %}
+ {%- if messages[0]['role'] == 'system' %}
+ {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+ {%- else %}
+ {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+ {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+ {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+ {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+ {%- elif message.role == "assistant" %}
+ {{- '<|im_start|>' + message.role }}
+ {%- if message.content %}
+ {{- '\n' + message.content }}
+ {%- endif %}
+ {%- for tool_call in message.tool_calls %}
+ {%- if tool_call.function is defined %}
+ {%- set tool_call = tool_call.function %}
+ {%- endif %}
+ {{- '\n\n{"name": "' }}
+ {{- tool_call.name }}
+ {{- '", "arguments": ' }}
+ {{- tool_call.arguments | tojson }}
+ {{- '}\n' }}
+ {%- endfor %}
+ {{- '<|im_end|>\n' }}
+ {%- elif message.role == "tool" %}
+ {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+ {{- '<|im_start|>user' }}
+ {%- endif %}
+ {{- '\n\n' }}
+ {{- message.content }}
+ {{- '\n' }}
+ {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+ {{- '<|im_end|>\n' }}
+ {%- endif %}
+ {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+ {{- '<|im_start|>assistant\n' }}
+{%- endif %}
diff --git a/qwen2.5-1.5b-instruct/checkpoint-71/config.json b/qwen2.5-1.5b-instruct/checkpoint-71/config.json
new file mode 100644
index 0000000000000000000000000000000000000000..499065fd744a64fc8ad2884da26e85a6e99d468a
--- /dev/null
+++ b/qwen2.5-1.5b-instruct/checkpoint-71/config.json
@@ -0,0 +1,61 @@
+{
+ "architectures": [
+ "Qwen2ForCausalLM"
+ ],
+ "attention_dropout": 0.0,
+ "bos_token_id": null,
+ "dtype": "float32",
+ "eos_token_id": 151645,
+ "hidden_act": "silu",
+ "hidden_size": 1536,
+ "initializer_range": 0.02,
+ "intermediate_size": 8960,
+ "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": 21,
+ "model_type": "qwen2",
+ "num_attention_heads": 12,
+ "num_hidden_layers": 28,
+ "num_key_value_heads": 2,
+ "pad_token_id": 151643,
+ "rms_norm_eps": 1e-06,
+ "rope_parameters": {
+ "rope_theta": 1000000.0,
+ "rope_type": "default"
+ },
+ "sliding_window": null,
+ "tie_word_embeddings": true,
+ "transformers_version": "5.2.0",
+ "use_cache": false,
+ "use_sliding_window": false,
+ "vocab_size": 151936
+}
diff --git a/qwen2.5-1.5b-instruct/checkpoint-71/generation_config.json b/qwen2.5-1.5b-instruct/checkpoint-71/generation_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..c5b4e11411fed3e7a5e6557165f888a7b0b1248b
--- /dev/null
+++ b/qwen2.5-1.5b-instruct/checkpoint-71/generation_config.json
@@ -0,0 +1,13 @@
+{
+ "do_sample": true,
+ "eos_token_id": [
+ 151645,
+ 151643
+ ],
+ "pad_token_id": 151643,
+ "repetition_penalty": 1.1,
+ "temperature": 0.7,
+ "top_k": 20,
+ "top_p": 0.8,
+ "transformers_version": "5.2.0"
+}
diff --git a/qwen2.5-1.5b-instruct/checkpoint-71/model.safetensors b/qwen2.5-1.5b-instruct/checkpoint-71/model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..20d371620def832a0c73d75561c750a08c592f95
--- /dev/null
+++ b/qwen2.5-1.5b-instruct/checkpoint-71/model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:92be2f2ded78014cb66d8fc7f35af7bb3ee591fde9c188d1a92c3fb3b4f47e82
+size 6174895536
diff --git a/qwen2.5-1.5b-instruct/checkpoint-71/optimizer.pt b/qwen2.5-1.5b-instruct/checkpoint-71/optimizer.pt
new file mode 100644
index 0000000000000000000000000000000000000000..0e63806a2b84e719c268a192e7950043916f2d9c
--- /dev/null
+++ b/qwen2.5-1.5b-instruct/checkpoint-71/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4966e88c4e62b8a810fac60d88bf48b3ce297ab0e3be902890e475ee0ebc3231
+size 12350013801
diff --git a/qwen2.5-1.5b-instruct/checkpoint-71/rng_state_0.pth b/qwen2.5-1.5b-instruct/checkpoint-71/rng_state_0.pth
new file mode 100644
index 0000000000000000000000000000000000000000..870021e3fa5ac35c2f711adf0c93a556ab4842da
--- /dev/null
+++ b/qwen2.5-1.5b-instruct/checkpoint-71/rng_state_0.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:95e5fc2074c0df31522a514f862c86cb00d71c946a7f15cc9ec0e53a69fb28a7
+size 14917
diff --git a/qwen2.5-1.5b-instruct/checkpoint-71/rng_state_1.pth b/qwen2.5-1.5b-instruct/checkpoint-71/rng_state_1.pth
new file mode 100644
index 0000000000000000000000000000000000000000..21f20da5eb1da017f08aaa88bd19cf24d40e3fbf
--- /dev/null
+++ b/qwen2.5-1.5b-instruct/checkpoint-71/rng_state_1.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0e7153eae67b6c9232a41bc996a2bf5b83229b8c7230d61911ac0fd40e64154e
+size 14917
diff --git a/qwen2.5-1.5b-instruct/checkpoint-71/scheduler.pt b/qwen2.5-1.5b-instruct/checkpoint-71/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..c3be1d29bde40f225713ff5084d3d9d69858fcfd
--- /dev/null
+++ b/qwen2.5-1.5b-instruct/checkpoint-71/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9c96c28f12671e2f5db493417daefb3ff0b1bc36602843e0d7ef6df3649c27b3
+size 1465
diff --git a/qwen2.5-1.5b-instruct/checkpoint-71/tokenizer.json b/qwen2.5-1.5b-instruct/checkpoint-71/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..34510ff0037cd50428af467a17ead5a96140a32c
--- /dev/null
+++ b/qwen2.5-1.5b-instruct/checkpoint-71/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
+size 11421892
diff --git a/qwen2.5-1.5b-instruct/checkpoint-71/tokenizer_config.json b/qwen2.5-1.5b-instruct/checkpoint-71/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..145e2c7a68b109ae0253dda29fc2e66cb01b4b85
--- /dev/null
+++ b/qwen2.5-1.5b-instruct/checkpoint-71/tokenizer_config.json
@@ -0,0 +1,30 @@
+{
+ "add_prefix_space": false,
+ "backend": "tokenizers",
+ "bos_token": null,
+ "clean_up_tokenization_spaces": false,
+ "eos_token": "<|im_end|>",
+ "errors": "replace",
+ "extra_special_tokens": [
+ "<|im_start|>",
+ "<|im_end|>",
+ "<|object_ref_start|>",
+ "<|object_ref_end|>",
+ "<|box_start|>",
+ "<|box_end|>",
+ "<|quad_start|>",
+ "<|quad_end|>",
+ "<|vision_start|>",
+ "<|vision_end|>",
+ "<|vision_pad|>",
+ "<|image_pad|>",
+ "<|video_pad|>"
+ ],
+ "is_local": false,
+ "model_max_length": 131072,
+ "pad_token": "<|endoftext|>",
+ "padding_side": "right",
+ "split_special_tokens": false,
+ "tokenizer_class": "Qwen2Tokenizer",
+ "unk_token": null
+}
diff --git a/qwen2.5-1.5b-instruct/checkpoint-71/trainer_state.json b/qwen2.5-1.5b-instruct/checkpoint-71/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..687b4b7649e817208b278b188954f73d3c8fc531
--- /dev/null
+++ b/qwen2.5-1.5b-instruct/checkpoint-71/trainer_state.json
@@ -0,0 +1,539 @@
+{
+ "best_global_step": null,
+ "best_metric": null,
+ "best_model_checkpoint": null,
+ "epoch": 1.0,
+ "eval_steps": 50,
+ "global_step": 71,
+ "is_hyper_param_search": false,
+ "is_local_process_zero": true,
+ "is_world_process_zero": true,
+ "log_history": [
+ {
+ "epoch": 0.014222222222222223,
+ "grad_norm": 0.10009177774190903,
+ "learning_rate": 0.0,
+ "loss": 0.704937219619751,
+ "step": 1
+ },
+ {
+ "epoch": 0.028444444444444446,
+ "grad_norm": 0.10451912879943848,
+ "learning_rate": 4.5454545454545457e-07,
+ "loss": 0.7318670749664307,
+ "step": 2
+ },
+ {
+ "epoch": 0.042666666666666665,
+ "grad_norm": 0.07928935438394547,
+ "learning_rate": 9.090909090909091e-07,
+ "loss": 0.6017104387283325,
+ "step": 3
+ },
+ {
+ "epoch": 0.05688888888888889,
+ "grad_norm": 0.048069290816783905,
+ "learning_rate": 1.3636363636363636e-06,
+ "loss": 0.6029032468795776,
+ "step": 4
+ },
+ {
+ "epoch": 0.07111111111111111,
+ "grad_norm": 0.02866927720606327,
+ "learning_rate": 1.8181818181818183e-06,
+ "loss": 0.5051919221878052,
+ "step": 5
+ },
+ {
+ "epoch": 0.08533333333333333,
+ "grad_norm": 0.022201435640454292,
+ "learning_rate": 2.2727272727272728e-06,
+ "loss": 0.5942293405532837,
+ "step": 6
+ },
+ {
+ "epoch": 0.09955555555555555,
+ "grad_norm": 0.018263086676597595,
+ "learning_rate": 2.7272727272727272e-06,
+ "loss": 0.5781338214874268,
+ "step": 7
+ },
+ {
+ "epoch": 0.11377777777777778,
+ "grad_norm": 0.01827927492558956,
+ "learning_rate": 3.181818181818182e-06,
+ "loss": 0.5180214047431946,
+ "step": 8
+ },
+ {
+ "epoch": 0.128,
+ "grad_norm": 0.014687157236039639,
+ "learning_rate": 3.6363636363636366e-06,
+ "loss": 0.5420116186141968,
+ "step": 9
+ },
+ {
+ "epoch": 0.14222222222222222,
+ "grad_norm": 0.012042663060128689,
+ "learning_rate": 4.0909090909090915e-06,
+ "loss": 0.5345908403396606,
+ "step": 10
+ },
+ {
+ "epoch": 0.15644444444444444,
+ "grad_norm": 0.012934447266161442,
+ "learning_rate": 4.5454545454545455e-06,
+ "loss": 0.46689414978027344,
+ "step": 11
+ },
+ {
+ "epoch": 0.17066666666666666,
+ "grad_norm": 0.012295619584619999,
+ "learning_rate": 5e-06,
+ "loss": 0.5062131881713867,
+ "step": 12
+ },
+ {
+ "epoch": 0.18488888888888888,
+ "grad_norm": 0.011856353841722012,
+ "learning_rate": 5.4545454545454545e-06,
+ "loss": 0.5063096880912781,
+ "step": 13
+ },
+ {
+ "epoch": 0.1991111111111111,
+ "grad_norm": 0.009451798163354397,
+ "learning_rate": 5.90909090909091e-06,
+ "loss": 0.5130602121353149,
+ "step": 14
+ },
+ {
+ "epoch": 0.21333333333333335,
+ "grad_norm": 0.01111016795039177,
+ "learning_rate": 6.363636363636364e-06,
+ "loss": 0.5016045570373535,
+ "step": 15
+ },
+ {
+ "epoch": 0.22755555555555557,
+ "grad_norm": 0.010735263116657734,
+ "learning_rate": 6.818181818181818e-06,
+ "loss": 0.49827253818511963,
+ "step": 16
+ },
+ {
+ "epoch": 0.24177777777777779,
+ "grad_norm": 0.010710126720368862,
+ "learning_rate": 7.272727272727273e-06,
+ "loss": 0.479993999004364,
+ "step": 17
+ },
+ {
+ "epoch": 0.256,
+ "grad_norm": 0.009600787423551083,
+ "learning_rate": 7.727272727272727e-06,
+ "loss": 0.4925128221511841,
+ "step": 18
+ },
+ {
+ "epoch": 0.2702222222222222,
+ "grad_norm": 0.010486789979040623,
+ "learning_rate": 8.181818181818183e-06,
+ "loss": 0.5188871622085571,
+ "step": 19
+ },
+ {
+ "epoch": 0.28444444444444444,
+ "grad_norm": 0.013396799564361572,
+ "learning_rate": 8.636363636363637e-06,
+ "loss": 0.5192517042160034,
+ "step": 20
+ },
+ {
+ "epoch": 0.2986666666666667,
+ "grad_norm": 0.009299715980887413,
+ "learning_rate": 9.090909090909091e-06,
+ "loss": 0.4580157399177551,
+ "step": 21
+ },
+ {
+ "epoch": 0.3128888888888889,
+ "grad_norm": 0.009691477753221989,
+ "learning_rate": 9.545454545454547e-06,
+ "loss": 0.49455106258392334,
+ "step": 22
+ },
+ {
+ "epoch": 0.32711111111111113,
+ "grad_norm": 0.008896945044398308,
+ "learning_rate": 1e-05,
+ "loss": 0.48672014474868774,
+ "step": 23
+ },
+ {
+ "epoch": 0.3413333333333333,
+ "grad_norm": 0.009234798140823841,
+ "learning_rate": 9.999323662872998e-06,
+ "loss": 0.5052967071533203,
+ "step": 24
+ },
+ {
+ "epoch": 0.35555555555555557,
+ "grad_norm": 0.009167873300611973,
+ "learning_rate": 9.99729483446475e-06,
+ "loss": 0.4740435779094696,
+ "step": 25
+ },
+ {
+ "epoch": 0.36977777777777776,
+ "grad_norm": 0.009001555852591991,
+ "learning_rate": 9.993914063644053e-06,
+ "loss": 0.47315946221351624,
+ "step": 26
+ },
+ {
+ "epoch": 0.384,
+ "grad_norm": 0.00915393978357315,
+ "learning_rate": 9.989182265027232e-06,
+ "loss": 0.4557237923145294,
+ "step": 27
+ },
+ {
+ "epoch": 0.3982222222222222,
+ "grad_norm": 0.008505849167704582,
+ "learning_rate": 9.98310071873072e-06,
+ "loss": 0.46317678689956665,
+ "step": 28
+ },
+ {
+ "epoch": 0.41244444444444445,
+ "grad_norm": 0.008668503724038601,
+ "learning_rate": 9.975671070024741e-06,
+ "loss": 0.4533114433288574,
+ "step": 29
+ },
+ {
+ "epoch": 0.4266666666666667,
+ "grad_norm": 0.009284070692956448,
+ "learning_rate": 9.966895328888195e-06,
+ "loss": 0.5624005794525146,
+ "step": 30
+ },
+ {
+ "epoch": 0.4408888888888889,
+ "grad_norm": 0.00935022346675396,
+ "learning_rate": 9.956775869464901e-06,
+ "loss": 0.4420597553253174,
+ "step": 31
+ },
+ {
+ "epoch": 0.45511111111111113,
+ "grad_norm": 0.008682326413691044,
+ "learning_rate": 9.945315429421307e-06,
+ "loss": 0.4745720624923706,
+ "step": 32
+ },
+ {
+ "epoch": 0.4693333333333333,
+ "grad_norm": 0.007816251367330551,
+ "learning_rate": 9.932517109205849e-06,
+ "loss": 0.439736545085907,
+ "step": 33
+ },
+ {
+ "epoch": 0.48355555555555557,
+ "grad_norm": 0.009315130300819874,
+ "learning_rate": 9.918384371210178e-06,
+ "loss": 0.5357447862625122,
+ "step": 34
+ },
+ {
+ "epoch": 0.49777777777777776,
+ "grad_norm": 0.009141312912106514,
+ "learning_rate": 9.902921038832456e-06,
+ "loss": 0.519463062286377,
+ "step": 35
+ },
+ {
+ "epoch": 0.512,
+ "grad_norm": 0.008826862089335918,
+ "learning_rate": 9.886131295443003e-06,
+ "loss": 0.48377716541290283,
+ "step": 36
+ },
+ {
+ "epoch": 0.5262222222222223,
+ "grad_norm": 0.008700878359377384,
+ "learning_rate": 9.868019683252543e-06,
+ "loss": 0.48436683416366577,
+ "step": 37
+ },
+ {
+ "epoch": 0.5404444444444444,
+ "grad_norm": 0.011412105523049831,
+ "learning_rate": 9.848591102083375e-06,
+ "loss": 0.4949312210083008,
+ "step": 38
+ },
+ {
+ "epoch": 0.5546666666666666,
+ "grad_norm": 0.009214675053954124,
+ "learning_rate": 9.82785080804381e-06,
+ "loss": 0.4901079535484314,
+ "step": 39
+ },
+ {
+ "epoch": 0.5688888888888889,
+ "grad_norm": 0.01163570862263441,
+ "learning_rate": 9.805804412106197e-06,
+ "loss": 0.6070311665534973,
+ "step": 40
+ },
+ {
+ "epoch": 0.5831111111111111,
+ "grad_norm": 0.009325762279331684,
+ "learning_rate": 9.782457878588977e-06,
+ "loss": 0.48412975668907166,
+ "step": 41
+ },
+ {
+ "epoch": 0.5973333333333334,
+ "grad_norm": 0.008266489021480083,
+ "learning_rate": 9.75781752354311e-06,
+ "loss": 0.4440661072731018,
+ "step": 42
+ },
+ {
+ "epoch": 0.6115555555555555,
+ "grad_norm": 0.00843313243240118,
+ "learning_rate": 9.731890013043367e-06,
+ "loss": 0.502442479133606,
+ "step": 43
+ },
+ {
+ "epoch": 0.6257777777777778,
+ "grad_norm": 0.00934070535004139,
+ "learning_rate": 9.704682361384941e-06,
+ "loss": 0.47924768924713135,
+ "step": 44
+ },
+ {
+ "epoch": 0.64,
+ "grad_norm": 0.010360248386859894,
+ "learning_rate": 9.676201929185809e-06,
+ "loss": 0.4161364436149597,
+ "step": 45
+ },
+ {
+ "epoch": 0.6542222222222223,
+ "grad_norm": 0.0094041982665658,
+ "learning_rate": 9.646456421395447e-06,
+ "loss": 0.48306071758270264,
+ "step": 46
+ },
+ {
+ "epoch": 0.6684444444444444,
+ "grad_norm": 0.009070150554180145,
+ "learning_rate": 9.615453885210368e-06,
+ "loss": 0.4774249792098999,
+ "step": 47
+ },
+ {
+ "epoch": 0.6826666666666666,
+ "grad_norm": 0.009771023876965046,
+ "learning_rate": 9.583202707897075e-06,
+ "loss": 0.4948921203613281,
+ "step": 48
+ },
+ {
+ "epoch": 0.6968888888888889,
+ "grad_norm": 0.008749130181968212,
+ "learning_rate": 9.549711614523007e-06,
+ "loss": 0.4825429916381836,
+ "step": 49
+ },
+ {
+ "epoch": 0.7111111111111111,
+ "grad_norm": 0.007630312815308571,
+ "learning_rate": 9.514989665596114e-06,
+ "loss": 0.46055319905281067,
+ "step": 50
+ },
+ {
+ "epoch": 0.7111111111111111,
+ "eval_loss": 0.48781517148017883,
+ "eval_runtime": 4.5903,
+ "eval_samples_per_second": 43.57,
+ "eval_steps_per_second": 21.785,
+ "step": 50
+ },
+ {
+ "epoch": 0.7253333333333334,
+ "grad_norm": 0.008408056572079659,
+ "learning_rate": 9.479046254613673e-06,
+ "loss": 0.4626312255859375,
+ "step": 51
+ },
+ {
+ "epoch": 0.7395555555555555,
+ "grad_norm": 0.007699670735746622,
+ "learning_rate": 9.441891105521005e-06,
+ "loss": 0.4499070644378662,
+ "step": 52
+ },
+ {
+ "epoch": 0.7537777777777778,
+ "grad_norm": 0.009070169180631638,
+ "learning_rate": 9.40353427008083e-06,
+ "loss": 0.4825657606124878,
+ "step": 53
+ },
+ {
+ "epoch": 0.768,
+ "grad_norm": 0.007915646769106388,
+ "learning_rate": 9.3639861251539e-06,
+ "loss": 0.4534468650817871,
+ "step": 54
+ },
+ {
+ "epoch": 0.7822222222222223,
+ "grad_norm": 0.011336776427924633,
+ "learning_rate": 9.323257369891702e-06,
+ "loss": 0.4988631010055542,
+ "step": 55
+ },
+ {
+ "epoch": 0.7964444444444444,
+ "grad_norm": 0.00761911366134882,
+ "learning_rate": 9.281359022841966e-06,
+ "loss": 0.4371437728404999,
+ "step": 56
+ },
+ {
+ "epoch": 0.8106666666666666,
+ "grad_norm": 0.009316994808614254,
+ "learning_rate": 9.238302418967757e-06,
+ "loss": 0.4956932067871094,
+ "step": 57
+ },
+ {
+ "epoch": 0.8248888888888889,
+ "grad_norm": 0.009821794927120209,
+ "learning_rate": 9.194099206580981e-06,
+ "loss": 0.5392974615097046,
+ "step": 58
+ },
+ {
+ "epoch": 0.8391111111111111,
+ "grad_norm": 0.008422416634857655,
+ "learning_rate": 9.14876134419111e-06,
+ "loss": 0.47983357310295105,
+ "step": 59
+ },
+ {
+ "epoch": 0.8533333333333334,
+ "grad_norm": 0.00837004091590643,
+ "learning_rate": 9.102301097269974e-06,
+ "loss": 0.5183510780334473,
+ "step": 60
+ },
+ {
+ "epoch": 0.8675555555555555,
+ "grad_norm": 0.008596297353506088,
+ "learning_rate": 9.05473103493355e-06,
+ "loss": 0.47292786836624146,
+ "step": 61
+ },
+ {
+ "epoch": 0.8817777777777778,
+ "grad_norm": 0.009059637784957886,
+ "learning_rate": 9.006064026541549e-06,
+ "loss": 0.5188573598861694,
+ "step": 62
+ },
+ {
+ "epoch": 0.896,
+ "grad_norm": 0.007816917262971401,
+ "learning_rate": 8.956313238215824e-06,
+ "loss": 0.45641395449638367,
+ "step": 63
+ },
+ {
+ "epoch": 0.9102222222222223,
+ "grad_norm": 0.007985202595591545,
+ "learning_rate": 8.905492129278478e-06,
+ "loss": 0.4586361050605774,
+ "step": 64
+ },
+ {
+ "epoch": 0.9244444444444444,
+ "grad_norm": 0.007170964498072863,
+ "learning_rate": 8.85361444861063e-06,
+ "loss": 0.4394378066062927,
+ "step": 65
+ },
+ {
+ "epoch": 0.9386666666666666,
+ "grad_norm": 0.00909162126481533,
+ "learning_rate": 8.800694230932885e-06,
+ "loss": 0.509820818901062,
+ "step": 66
+ },
+ {
+ "epoch": 0.9528888888888889,
+ "grad_norm": 0.00786044355481863,
+ "learning_rate": 8.74674579300843e-06,
+ "loss": 0.43584880232810974,
+ "step": 67
+ },
+ {
+ "epoch": 0.9671111111111111,
+ "grad_norm": 0.00832363311201334,
+ "learning_rate": 8.691783729769874e-06,
+ "loss": 0.4317135214805603,
+ "step": 68
+ },
+ {
+ "epoch": 0.9813333333333333,
+ "grad_norm": 0.008376957848668098,
+ "learning_rate": 8.635822910370793e-06,
+ "loss": 0.49933016300201416,
+ "step": 69
+ },
+ {
+ "epoch": 0.9955555555555555,
+ "grad_norm": 0.00835558120161295,
+ "learning_rate": 8.578878474163115e-06,
+ "loss": 0.46570393443107605,
+ "step": 70
+ },
+ {
+ "epoch": 1.0,
+ "grad_norm": 0.013256405480206013,
+ "learning_rate": 8.520965826601394e-06,
+ "loss": 0.45107918977737427,
+ "step": 71
+ }
+ ],
+ "logging_steps": 1,
+ "max_steps": 213,
+ "num_input_tokens_seen": 0,
+ "num_train_epochs": 3,
+ "save_steps": 500,
+ "stateful_callbacks": {
+ "TrainerControl": {
+ "args": {
+ "should_epoch_stop": false,
+ "should_evaluate": false,
+ "should_log": false,
+ "should_save": true,
+ "should_training_stop": false
+ },
+ "attributes": {}
+ }
+ },
+ "total_flos": 1.6226691949002752e+16,
+ "train_batch_size": 1,
+ "trial_name": null,
+ "trial_params": null
+}
diff --git a/qwen2.5-1.5b-instruct/checkpoint-71/training_args.bin b/qwen2.5-1.5b-instruct/checkpoint-71/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..1c92bb7fdd3ef14952372d452eaf2c7d0ade0ccf
--- /dev/null
+++ b/qwen2.5-1.5b-instruct/checkpoint-71/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:841e251fafc8c15cdc7b32aa47df63c6ea8f5d8eb459fbecdbb2a9f5f74be033
+size 5649
diff --git a/qwen2.5-1.5b-instruct/config.json b/qwen2.5-1.5b-instruct/config.json
new file mode 100644
index 0000000000000000000000000000000000000000..499065fd744a64fc8ad2884da26e85a6e99d468a
--- /dev/null
+++ b/qwen2.5-1.5b-instruct/config.json
@@ -0,0 +1,61 @@
+{
+ "architectures": [
+ "Qwen2ForCausalLM"
+ ],
+ "attention_dropout": 0.0,
+ "bos_token_id": null,
+ "dtype": "float32",
+ "eos_token_id": 151645,
+ "hidden_act": "silu",
+ "hidden_size": 1536,
+ "initializer_range": 0.02,
+ "intermediate_size": 8960,
+ "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": 21,
+ "model_type": "qwen2",
+ "num_attention_heads": 12,
+ "num_hidden_layers": 28,
+ "num_key_value_heads": 2,
+ "pad_token_id": 151643,
+ "rms_norm_eps": 1e-06,
+ "rope_parameters": {
+ "rope_theta": 1000000.0,
+ "rope_type": "default"
+ },
+ "sliding_window": null,
+ "tie_word_embeddings": true,
+ "transformers_version": "5.2.0",
+ "use_cache": false,
+ "use_sliding_window": false,
+ "vocab_size": 151936
+}
diff --git a/qwen2.5-1.5b-instruct/eval_results.json b/qwen2.5-1.5b-instruct/eval_results.json
new file mode 100644
index 0000000000000000000000000000000000000000..326987bc25a5067a8e850c4872e5766e3d196142
--- /dev/null
+++ b/qwen2.5-1.5b-instruct/eval_results.json
@@ -0,0 +1,7 @@
+{
+ "epoch": 3.0,
+ "eval_loss": 2.227557897567749,
+ "eval_runtime": 8.5719,
+ "eval_samples_per_second": 23.332,
+ "eval_steps_per_second": 5.833
+}
\ No newline at end of file
diff --git a/qwen2.5-1.5b-instruct/generation_config.json b/qwen2.5-1.5b-instruct/generation_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..c5b4e11411fed3e7a5e6557165f888a7b0b1248b
--- /dev/null
+++ b/qwen2.5-1.5b-instruct/generation_config.json
@@ -0,0 +1,13 @@
+{
+ "do_sample": true,
+ "eos_token_id": [
+ 151645,
+ 151643
+ ],
+ "pad_token_id": 151643,
+ "repetition_penalty": 1.1,
+ "temperature": 0.7,
+ "top_k": 20,
+ "top_p": 0.8,
+ "transformers_version": "5.2.0"
+}
diff --git a/qwen2.5-1.5b-instruct/model.safetensors b/qwen2.5-1.5b-instruct/model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..7aa60a284dae84c00e62cae176fd1986a775b887
--- /dev/null
+++ b/qwen2.5-1.5b-instruct/model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3434dfcf0343459c122b117ee99c0d83e3e6ac7f15a7c1e7cac6d3772e907708
+size 6174895536
diff --git a/qwen2.5-1.5b-instruct/tokenizer.json b/qwen2.5-1.5b-instruct/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..34510ff0037cd50428af467a17ead5a96140a32c
--- /dev/null
+++ b/qwen2.5-1.5b-instruct/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
+size 11421892
diff --git a/qwen2.5-1.5b-instruct/tokenizer_config.json b/qwen2.5-1.5b-instruct/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..145e2c7a68b109ae0253dda29fc2e66cb01b4b85
--- /dev/null
+++ b/qwen2.5-1.5b-instruct/tokenizer_config.json
@@ -0,0 +1,30 @@
+{
+ "add_prefix_space": false,
+ "backend": "tokenizers",
+ "bos_token": null,
+ "clean_up_tokenization_spaces": false,
+ "eos_token": "<|im_end|>",
+ "errors": "replace",
+ "extra_special_tokens": [
+ "<|im_start|>",
+ "<|im_end|>",
+ "<|object_ref_start|>",
+ "<|object_ref_end|>",
+ "<|box_start|>",
+ "<|box_end|>",
+ "<|quad_start|>",
+ "<|quad_end|>",
+ "<|vision_start|>",
+ "<|vision_end|>",
+ "<|vision_pad|>",
+ "<|image_pad|>",
+ "<|video_pad|>"
+ ],
+ "is_local": false,
+ "model_max_length": 131072,
+ "pad_token": "<|endoftext|>",
+ "padding_side": "right",
+ "split_special_tokens": false,
+ "tokenizer_class": "Qwen2Tokenizer",
+ "unk_token": null
+}
diff --git a/qwen2.5-1.5b-instruct/train_results.json b/qwen2.5-1.5b-instruct/train_results.json
new file mode 100644
index 0000000000000000000000000000000000000000..72255549457d9a89c06b802b7256f4338c8c6fc9
--- /dev/null
+++ b/qwen2.5-1.5b-instruct/train_results.json
@@ -0,0 +1,8 @@
+{
+ "epoch": 3.0,
+ "total_flos": 4.868007593928294e+16,
+ "train_loss": 0.0,
+ "train_runtime": 3.3055,
+ "train_samples_per_second": 16336.51,
+ "train_steps_per_second": 32.673
+}
\ No newline at end of file
diff --git a/qwen2.5-1.5b-instruct/trainer_log.jsonl b/qwen2.5-1.5b-instruct/trainer_log.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..63baf68915ac0173fb14442fd0a8098c81b01ef1
--- /dev/null
+++ b/qwen2.5-1.5b-instruct/trainer_log.jsonl
@@ -0,0 +1,221 @@
+{"current_steps": 1, "total_steps": 213, "loss": 0.704937219619751, "lr": 0.0, "epoch": 0.014222222222222223, "percentage": 0.47, "elapsed_time": "0:00:31", "remaining_time": "1:49:53"}
+{"current_steps": 2, "total_steps": 213, "loss": 0.7318670749664307, "lr": 4.5454545454545457e-07, "epoch": 0.028444444444444446, "percentage": 0.94, "elapsed_time": "0:01:01", "remaining_time": "1:47:27"}
+{"current_steps": 1, "total_steps": 213, "loss": 0.704937219619751, "lr": 0.0, "epoch": 0.014222222222222223, "percentage": 0.47, "elapsed_time": "0:00:30", "remaining_time": "1:46:40"}
+{"current_steps": 2, "total_steps": 213, "loss": 0.7318670749664307, "lr": 4.5454545454545457e-07, "epoch": 0.028444444444444446, "percentage": 0.94, "elapsed_time": "0:00:59", "remaining_time": "1:44:08"}
+{"current_steps": 3, "total_steps": 213, "loss": 0.6017104387283325, "lr": 9.090909090909091e-07, "epoch": 0.042666666666666665, "percentage": 1.41, "elapsed_time": "0:01:20", "remaining_time": "1:33:32"}
+{"current_steps": 4, "total_steps": 213, "loss": 0.6029032468795776, "lr": 1.3636363636363636e-06, "epoch": 0.05688888888888889, "percentage": 1.88, "elapsed_time": "0:01:38", "remaining_time": "1:25:37"}
+{"current_steps": 5, "total_steps": 213, "loss": 0.5051919221878052, "lr": 1.8181818181818183e-06, "epoch": 0.07111111111111111, "percentage": 2.35, "elapsed_time": "0:01:56", "remaining_time": "1:20:46"}
+{"current_steps": 6, "total_steps": 213, "loss": 0.5942293405532837, "lr": 2.2727272727272728e-06, "epoch": 0.08533333333333333, "percentage": 2.82, "elapsed_time": "0:02:14", "remaining_time": "1:17:23"}
+{"current_steps": 7, "total_steps": 213, "loss": 0.5781338214874268, "lr": 2.7272727272727272e-06, "epoch": 0.09955555555555555, "percentage": 3.29, "elapsed_time": "0:02:32", "remaining_time": "1:14:53"}
+{"current_steps": 8, "total_steps": 213, "loss": 0.5180214047431946, "lr": 3.181818181818182e-06, "epoch": 0.11377777777777778, "percentage": 3.76, "elapsed_time": "0:02:50", "remaining_time": "1:13:01"}
+{"current_steps": 9, "total_steps": 213, "loss": 0.5420116186141968, "lr": 3.6363636363636366e-06, "epoch": 0.128, "percentage": 4.23, "elapsed_time": "0:03:09", "remaining_time": "1:11:24"}
+{"current_steps": 10, "total_steps": 213, "loss": 0.5345908403396606, "lr": 4.0909090909090915e-06, "epoch": 0.14222222222222222, "percentage": 4.69, "elapsed_time": "0:03:27", "remaining_time": "1:10:10"}
+{"current_steps": 11, "total_steps": 213, "loss": 0.46689414978027344, "lr": 4.5454545454545455e-06, "epoch": 0.15644444444444444, "percentage": 5.16, "elapsed_time": "0:03:45", "remaining_time": "1:09:04"}
+{"current_steps": 12, "total_steps": 213, "loss": 0.5062131881713867, "lr": 5e-06, "epoch": 0.17066666666666666, "percentage": 5.63, "elapsed_time": "0:04:04", "remaining_time": "1:08:07"}
+{"current_steps": 13, "total_steps": 213, "loss": 0.5063096880912781, "lr": 5.4545454545454545e-06, "epoch": 0.18488888888888888, "percentage": 6.1, "elapsed_time": "0:04:22", "remaining_time": "1:07:17"}
+{"current_steps": 14, "total_steps": 213, "loss": 0.5130602121353149, "lr": 5.90909090909091e-06, "epoch": 0.1991111111111111, "percentage": 6.57, "elapsed_time": "0:04:40", "remaining_time": "1:06:29"}
+{"current_steps": 15, "total_steps": 213, "loss": 0.5016045570373535, "lr": 6.363636363636364e-06, "epoch": 0.21333333333333335, "percentage": 7.04, "elapsed_time": "0:04:58", "remaining_time": "1:05:44"}
+{"current_steps": 16, "total_steps": 213, "loss": 0.49827253818511963, "lr": 6.818181818181818e-06, "epoch": 0.22755555555555557, "percentage": 7.51, "elapsed_time": "0:05:17", "remaining_time": "1:05:03"}
+{"current_steps": 17, "total_steps": 213, "loss": 0.479993999004364, "lr": 7.272727272727273e-06, "epoch": 0.24177777777777779, "percentage": 7.98, "elapsed_time": "0:05:35", "remaining_time": "1:04:27"}
+{"current_steps": 18, "total_steps": 213, "loss": 0.4925128221511841, "lr": 7.727272727272727e-06, "epoch": 0.256, "percentage": 8.45, "elapsed_time": "0:05:53", "remaining_time": "1:03:51"}
+{"current_steps": 19, "total_steps": 213, "loss": 0.5188871622085571, "lr": 8.181818181818183e-06, "epoch": 0.2702222222222222, "percentage": 8.92, "elapsed_time": "0:06:12", "remaining_time": "1:03:18"}
+{"current_steps": 20, "total_steps": 213, "loss": 0.5192517042160034, "lr": 8.636363636363637e-06, "epoch": 0.28444444444444444, "percentage": 9.39, "elapsed_time": "0:06:30", "remaining_time": "1:02:47"}
+{"current_steps": 21, "total_steps": 213, "loss": 0.4580157399177551, "lr": 9.090909090909091e-06, "epoch": 0.2986666666666667, "percentage": 9.86, "elapsed_time": "0:06:48", "remaining_time": "1:02:17"}
+{"current_steps": 22, "total_steps": 213, "loss": 0.49455106258392334, "lr": 9.545454545454547e-06, "epoch": 0.3128888888888889, "percentage": 10.33, "elapsed_time": "0:07:07", "remaining_time": "1:01:47"}
+{"current_steps": 23, "total_steps": 213, "loss": 0.48672014474868774, "lr": 1e-05, "epoch": 0.32711111111111113, "percentage": 10.8, "elapsed_time": "0:07:25", "remaining_time": "1:01:19"}
+{"current_steps": 24, "total_steps": 213, "loss": 0.5052967071533203, "lr": 9.999323662872998e-06, "epoch": 0.3413333333333333, "percentage": 11.27, "elapsed_time": "0:07:43", "remaining_time": "1:00:50"}
+{"current_steps": 25, "total_steps": 213, "loss": 0.4740435779094696, "lr": 9.99729483446475e-06, "epoch": 0.35555555555555557, "percentage": 11.74, "elapsed_time": "0:08:01", "remaining_time": "1:00:24"}
+{"current_steps": 26, "total_steps": 213, "loss": 0.47315946221351624, "lr": 9.993914063644053e-06, "epoch": 0.36977777777777776, "percentage": 12.21, "elapsed_time": "0:08:20", "remaining_time": "0:59:57"}
+{"current_steps": 27, "total_steps": 213, "loss": 0.4557237923145294, "lr": 9.989182265027232e-06, "epoch": 0.384, "percentage": 12.68, "elapsed_time": "0:08:38", "remaining_time": "0:59:31"}
+{"current_steps": 28, "total_steps": 213, "loss": 0.46317678689956665, "lr": 9.98310071873072e-06, "epoch": 0.3982222222222222, "percentage": 13.15, "elapsed_time": "0:08:57", "remaining_time": "0:59:09"}
+{"current_steps": 29, "total_steps": 213, "loss": 0.4533114433288574, "lr": 9.975671070024741e-06, "epoch": 0.41244444444444445, "percentage": 13.62, "elapsed_time": "0:09:15", "remaining_time": "0:58:46"}
+{"current_steps": 30, "total_steps": 213, "loss": 0.5624005794525146, "lr": 9.966895328888195e-06, "epoch": 0.4266666666666667, "percentage": 14.08, "elapsed_time": "0:09:34", "remaining_time": "0:58:22"}
+{"current_steps": 31, "total_steps": 213, "loss": 0.4420597553253174, "lr": 9.956775869464901e-06, "epoch": 0.4408888888888889, "percentage": 14.55, "elapsed_time": "0:09:52", "remaining_time": "0:57:59"}
+{"current_steps": 32, "total_steps": 213, "loss": 0.4745720624923706, "lr": 9.945315429421307e-06, "epoch": 0.45511111111111113, "percentage": 15.02, "elapsed_time": "0:10:10", "remaining_time": "0:57:35"}
+{"current_steps": 33, "total_steps": 213, "loss": 0.439736545085907, "lr": 9.932517109205849e-06, "epoch": 0.4693333333333333, "percentage": 15.49, "elapsed_time": "0:10:29", "remaining_time": "0:57:12"}
+{"current_steps": 34, "total_steps": 213, "loss": 0.5357447862625122, "lr": 9.918384371210178e-06, "epoch": 0.48355555555555557, "percentage": 15.96, "elapsed_time": "0:10:47", "remaining_time": "0:56:49"}
+{"current_steps": 35, "total_steps": 213, "loss": 0.519463062286377, "lr": 9.902921038832456e-06, "epoch": 0.49777777777777776, "percentage": 16.43, "elapsed_time": "0:11:05", "remaining_time": "0:56:26"}
+{"current_steps": 36, "total_steps": 213, "loss": 0.48377716541290283, "lr": 9.886131295443003e-06, "epoch": 0.512, "percentage": 16.9, "elapsed_time": "0:11:24", "remaining_time": "0:56:03"}
+{"current_steps": 37, "total_steps": 213, "loss": 0.48436683416366577, "lr": 9.868019683252543e-06, "epoch": 0.5262222222222223, "percentage": 17.37, "elapsed_time": "0:11:42", "remaining_time": "0:55:41"}
+{"current_steps": 38, "total_steps": 213, "loss": 0.4949312210083008, "lr": 9.848591102083375e-06, "epoch": 0.5404444444444444, "percentage": 17.84, "elapsed_time": "0:12:00", "remaining_time": "0:55:18"}
+{"current_steps": 39, "total_steps": 213, "loss": 0.4901079535484314, "lr": 9.82785080804381e-06, "epoch": 0.5546666666666666, "percentage": 18.31, "elapsed_time": "0:12:18", "remaining_time": "0:54:56"}
+{"current_steps": 40, "total_steps": 213, "loss": 0.6070311665534973, "lr": 9.805804412106197e-06, "epoch": 0.5688888888888889, "percentage": 18.78, "elapsed_time": "0:12:37", "remaining_time": "0:54:34"}
+{"current_steps": 41, "total_steps": 213, "loss": 0.48412975668907166, "lr": 9.782457878588977e-06, "epoch": 0.5831111111111111, "percentage": 19.25, "elapsed_time": "0:12:55", "remaining_time": "0:54:13"}
+{"current_steps": 42, "total_steps": 213, "loss": 0.4440661072731018, "lr": 9.75781752354311e-06, "epoch": 0.5973333333333334, "percentage": 19.72, "elapsed_time": "0:13:13", "remaining_time": "0:53:51"}
+{"current_steps": 43, "total_steps": 213, "loss": 0.502442479133606, "lr": 9.731890013043367e-06, "epoch": 0.6115555555555555, "percentage": 20.19, "elapsed_time": "0:13:31", "remaining_time": "0:53:29"}
+{"current_steps": 44, "total_steps": 213, "loss": 0.47924768924713135, "lr": 9.704682361384941e-06, "epoch": 0.6257777777777778, "percentage": 20.66, "elapsed_time": "0:13:50", "remaining_time": "0:53:08"}
+{"current_steps": 45, "total_steps": 213, "loss": 0.4161364436149597, "lr": 9.676201929185809e-06, "epoch": 0.64, "percentage": 21.13, "elapsed_time": "0:14:08", "remaining_time": "0:52:48"}
+{"current_steps": 46, "total_steps": 213, "loss": 0.48306071758270264, "lr": 9.646456421395447e-06, "epoch": 0.6542222222222223, "percentage": 21.6, "elapsed_time": "0:14:27", "remaining_time": "0:52:27"}
+{"current_steps": 47, "total_steps": 213, "loss": 0.4774249792098999, "lr": 9.615453885210368e-06, "epoch": 0.6684444444444444, "percentage": 22.07, "elapsed_time": "0:14:45", "remaining_time": "0:52:07"}
+{"current_steps": 48, "total_steps": 213, "loss": 0.4948921203613281, "lr": 9.583202707897075e-06, "epoch": 0.6826666666666666, "percentage": 22.54, "elapsed_time": "0:15:03", "remaining_time": "0:51:46"}
+{"current_steps": 49, "total_steps": 213, "loss": 0.4825429916381836, "lr": 9.549711614523007e-06, "epoch": 0.6968888888888889, "percentage": 23.0, "elapsed_time": "0:15:21", "remaining_time": "0:51:25"}
+{"current_steps": 50, "total_steps": 213, "loss": 0.46055319905281067, "lr": 9.514989665596114e-06, "epoch": 0.7111111111111111, "percentage": 23.47, "elapsed_time": "0:15:40", "remaining_time": "0:51:05"}
+{"current_steps": 50, "total_steps": 213, "eval_loss": 0.48781517148017883, "epoch": 0.7111111111111111, "percentage": 23.47, "elapsed_time": "0:15:44", "remaining_time": "0:51:20"}
+{"current_steps": 51, "total_steps": 213, "loss": 0.4626312255859375, "lr": 9.479046254613673e-06, "epoch": 0.7253333333333334, "percentage": 23.94, "elapsed_time": "0:16:03", "remaining_time": "0:50:59"}
+{"current_steps": 52, "total_steps": 213, "loss": 0.4499070644378662, "lr": 9.441891105521005e-06, "epoch": 0.7395555555555555, "percentage": 24.41, "elapsed_time": "0:16:21", "remaining_time": "0:50:38"}
+{"current_steps": 53, "total_steps": 213, "loss": 0.4825657606124878, "lr": 9.40353427008083e-06, "epoch": 0.7537777777777778, "percentage": 24.88, "elapsed_time": "0:16:39", "remaining_time": "0:50:17"}
+{"current_steps": 54, "total_steps": 213, "loss": 0.4534468650817871, "lr": 9.3639861251539e-06, "epoch": 0.768, "percentage": 25.35, "elapsed_time": "0:16:58", "remaining_time": "0:49:57"}
+{"current_steps": 55, "total_steps": 213, "loss": 0.4988631010055542, "lr": 9.323257369891702e-06, "epoch": 0.7822222222222223, "percentage": 25.82, "elapsed_time": "0:17:16", "remaining_time": "0:49:37"}
+{"current_steps": 56, "total_steps": 213, "loss": 0.4371437728404999, "lr": 9.281359022841966e-06, "epoch": 0.7964444444444444, "percentage": 26.29, "elapsed_time": "0:17:34", "remaining_time": "0:49:17"}
+{"current_steps": 57, "total_steps": 213, "loss": 0.4956932067871094, "lr": 9.238302418967757e-06, "epoch": 0.8106666666666666, "percentage": 26.76, "elapsed_time": "0:17:53", "remaining_time": "0:48:58"}
+{"current_steps": 58, "total_steps": 213, "loss": 0.5392974615097046, "lr": 9.194099206580981e-06, "epoch": 0.8248888888888889, "percentage": 27.23, "elapsed_time": "0:18:12", "remaining_time": "0:48:38"}
+{"current_steps": 59, "total_steps": 213, "loss": 0.47983357310295105, "lr": 9.14876134419111e-06, "epoch": 0.8391111111111111, "percentage": 27.7, "elapsed_time": "0:18:30", "remaining_time": "0:48:19"}
+{"current_steps": 60, "total_steps": 213, "loss": 0.5183510780334473, "lr": 9.102301097269974e-06, "epoch": 0.8533333333333334, "percentage": 28.17, "elapsed_time": "0:18:49", "remaining_time": "0:47:59"}
+{"current_steps": 61, "total_steps": 213, "loss": 0.47292786836624146, "lr": 9.05473103493355e-06, "epoch": 0.8675555555555555, "percentage": 28.64, "elapsed_time": "0:19:07", "remaining_time": "0:47:39"}
+{"current_steps": 62, "total_steps": 213, "loss": 0.5188573598861694, "lr": 9.006064026541549e-06, "epoch": 0.8817777777777778, "percentage": 29.11, "elapsed_time": "0:19:26", "remaining_time": "0:47:19"}
+{"current_steps": 63, "total_steps": 213, "loss": 0.45641395449638367, "lr": 8.956313238215824e-06, "epoch": 0.896, "percentage": 29.58, "elapsed_time": "0:19:44", "remaining_time": "0:46:59"}
+{"current_steps": 64, "total_steps": 213, "loss": 0.4586361050605774, "lr": 8.905492129278478e-06, "epoch": 0.9102222222222223, "percentage": 30.05, "elapsed_time": "0:20:02", "remaining_time": "0:46:39"}
+{"current_steps": 65, "total_steps": 213, "loss": 0.4394378066062927, "lr": 8.85361444861063e-06, "epoch": 0.9244444444444444, "percentage": 30.52, "elapsed_time": "0:20:21", "remaining_time": "0:46:21"}
+{"current_steps": 66, "total_steps": 213, "loss": 0.509820818901062, "lr": 8.800694230932885e-06, "epoch": 0.9386666666666666, "percentage": 30.99, "elapsed_time": "0:20:39", "remaining_time": "0:46:00"}
+{"current_steps": 67, "total_steps": 213, "loss": 0.43584880232810974, "lr": 8.74674579300843e-06, "epoch": 0.9528888888888889, "percentage": 31.46, "elapsed_time": "0:20:57", "remaining_time": "0:45:41"}
+{"current_steps": 68, "total_steps": 213, "loss": 0.4317135214805603, "lr": 8.691783729769874e-06, "epoch": 0.9671111111111111, "percentage": 31.92, "elapsed_time": "0:21:16", "remaining_time": "0:45:21"}
+{"current_steps": 69, "total_steps": 213, "loss": 0.49933016300201416, "lr": 8.635822910370793e-06, "epoch": 0.9813333333333333, "percentage": 32.39, "elapsed_time": "0:21:34", "remaining_time": "0:45:02"}
+{"current_steps": 70, "total_steps": 213, "loss": 0.46570393443107605, "lr": 8.578878474163115e-06, "epoch": 0.9955555555555555, "percentage": 32.86, "elapsed_time": "0:21:52", "remaining_time": "0:44:42"}
+{"current_steps": 71, "total_steps": 213, "loss": 0.45107918977737427, "lr": 8.520965826601394e-06, "epoch": 1.0, "percentage": 33.33, "elapsed_time": "0:21:58", "remaining_time": "0:43:57"}
+{"current_steps": 72, "total_steps": 213, "loss": 0.47125527262687683, "lr": 8.462100635075097e-06, "epoch": 1.0142222222222221, "percentage": 33.8, "elapsed_time": "0:23:11", "remaining_time": "0:45:24"}
+{"current_steps": 73, "total_steps": 213, "loss": 0.4322492480278015, "lr": 8.40229882467003e-06, "epoch": 1.0284444444444445, "percentage": 34.27, "elapsed_time": "0:23:29", "remaining_time": "0:45:03"}
+{"current_steps": 74, "total_steps": 213, "loss": 0.38362830877304077, "lr": 8.341576573860049e-06, "epoch": 1.0426666666666666, "percentage": 34.74, "elapsed_time": "0:23:48", "remaining_time": "0:44:42"}
+{"current_steps": 75, "total_steps": 213, "loss": 0.3878062665462494, "lr": 8.279950310130218e-06, "epoch": 1.056888888888889, "percentage": 35.21, "elapsed_time": "0:24:06", "remaining_time": "0:44:21"}
+{"current_steps": 76, "total_steps": 213, "loss": 0.3833811283111572, "lr": 8.2174367055326e-06, "epoch": 1.0711111111111111, "percentage": 35.68, "elapsed_time": "0:24:24", "remaining_time": "0:44:00"}
+{"current_steps": 77, "total_steps": 213, "loss": 0.3845909833908081, "lr": 8.154052672175888e-06, "epoch": 1.0853333333333333, "percentage": 36.15, "elapsed_time": "0:24:43", "remaining_time": "0:43:39"}
+{"current_steps": 78, "total_steps": 213, "loss": 0.44259265065193176, "lr": 8.08981535765009e-06, "epoch": 1.0995555555555556, "percentage": 36.62, "elapsed_time": "0:25:01", "remaining_time": "0:43:18"}
+{"current_steps": 79, "total_steps": 213, "loss": 0.38503497838974, "lr": 8.024742140387506e-06, "epoch": 1.1137777777777778, "percentage": 37.09, "elapsed_time": "0:25:19", "remaining_time": "0:42:57"}
+{"current_steps": 80, "total_steps": 213, "loss": 0.42608046531677246, "lr": 7.95885062496126e-06, "epoch": 1.1280000000000001, "percentage": 37.56, "elapsed_time": "0:25:37", "remaining_time": "0:42:36"}
+{"current_steps": 81, "total_steps": 213, "loss": 0.38991105556488037, "lr": 7.892158637322647e-06, "epoch": 1.1422222222222222, "percentage": 38.03, "elapsed_time": "0:25:56", "remaining_time": "0:42:15"}
+{"current_steps": 82, "total_steps": 213, "loss": 0.36854323744773865, "lr": 7.824684219978591e-06, "epoch": 1.1564444444444444, "percentage": 38.5, "elapsed_time": "0:26:14", "remaining_time": "0:41:55"}
+{"current_steps": 83, "total_steps": 213, "loss": 0.4200327396392822, "lr": 7.756445627110523e-06, "epoch": 1.1706666666666667, "percentage": 38.97, "elapsed_time": "0:26:32", "remaining_time": "0:41:34"}
+{"current_steps": 84, "total_steps": 213, "loss": 0.4378758668899536, "lr": 7.68746131963598e-06, "epoch": 1.1848888888888889, "percentage": 39.44, "elapsed_time": "0:26:51", "remaining_time": "0:41:14"}
+{"current_steps": 85, "total_steps": 213, "loss": 0.3808164894580841, "lr": 7.6177499602143e-06, "epoch": 1.199111111111111, "percentage": 39.91, "elapsed_time": "0:27:09", "remaining_time": "0:40:53"}
+{"current_steps": 86, "total_steps": 213, "loss": 0.4453170895576477, "lr": 7.547330408197695e-06, "epoch": 1.2133333333333334, "percentage": 40.38, "elapsed_time": "0:27:27", "remaining_time": "0:40:33"}
+{"current_steps": 87, "total_steps": 213, "loss": 0.4456181526184082, "lr": 7.476221714529167e-06, "epoch": 1.2275555555555555, "percentage": 40.85, "elapsed_time": "0:27:46", "remaining_time": "0:40:13"}
+{"current_steps": 88, "total_steps": 213, "loss": 0.43805548548698425, "lr": 7.404443116588548e-06, "epoch": 1.2417777777777779, "percentage": 41.31, "elapsed_time": "0:28:04", "remaining_time": "0:39:52"}
+{"current_steps": 89, "total_steps": 213, "loss": 0.4144285321235657, "lr": 7.332014032988123e-06, "epoch": 1.256, "percentage": 41.78, "elapsed_time": "0:28:23", "remaining_time": "0:39:33"}
+{"current_steps": 90, "total_steps": 213, "loss": 0.38928651809692383, "lr": 7.2589540583192165e-06, "epoch": 1.2702222222222221, "percentage": 42.25, "elapsed_time": "0:28:41", "remaining_time": "0:39:12"}
+{"current_steps": 91, "total_steps": 213, "loss": 0.387723445892334, "lr": 7.185282957851175e-06, "epoch": 1.2844444444444445, "percentage": 42.72, "elapsed_time": "0:28:59", "remaining_time": "0:38:52"}
+{"current_steps": 92, "total_steps": 213, "loss": 0.4158015549182892, "lr": 7.111020662184174e-06, "epoch": 1.2986666666666666, "percentage": 43.19, "elapsed_time": "0:29:18", "remaining_time": "0:38:32"}
+{"current_steps": 93, "total_steps": 213, "loss": 0.39764612913131714, "lr": 7.036187261857289e-06, "epoch": 1.3128888888888888, "percentage": 43.66, "elapsed_time": "0:29:36", "remaining_time": "0:38:12"}
+{"current_steps": 94, "total_steps": 213, "loss": 0.38216379284858704, "lr": 6.960803001913315e-06, "epoch": 1.3271111111111111, "percentage": 44.13, "elapsed_time": "0:29:55", "remaining_time": "0:37:52"}
+{"current_steps": 95, "total_steps": 213, "loss": 0.413602352142334, "lr": 6.884888276421766e-06, "epoch": 1.3413333333333333, "percentage": 44.6, "elapsed_time": "0:30:13", "remaining_time": "0:37:32"}
+{"current_steps": 96, "total_steps": 213, "loss": 0.41756486892700195, "lr": 6.8084636229615786e-06, "epoch": 1.3555555555555556, "percentage": 45.07, "elapsed_time": "0:30:31", "remaining_time": "0:37:12"}
+{"current_steps": 97, "total_steps": 213, "loss": 0.4028819799423218, "lr": 6.731549717064975e-06, "epoch": 1.3697777777777778, "percentage": 45.54, "elapsed_time": "0:30:49", "remaining_time": "0:36:52"}
+{"current_steps": 98, "total_steps": 213, "loss": 0.37123921513557434, "lr": 6.654167366624009e-06, "epoch": 1.384, "percentage": 46.01, "elapsed_time": "0:31:08", "remaining_time": "0:36:32"}
+{"current_steps": 99, "total_steps": 213, "loss": 0.39443182945251465, "lr": 6.576337506261314e-06, "epoch": 1.3982222222222223, "percentage": 46.48, "elapsed_time": "0:31:26", "remaining_time": "0:36:12"}
+{"current_steps": 100, "total_steps": 213, "loss": 0.42232874035835266, "lr": 6.498081191666549e-06, "epoch": 1.4124444444444444, "percentage": 46.95, "elapsed_time": "0:31:44", "remaining_time": "0:35:52"}
+{"current_steps": 100, "total_steps": 213, "eval_loss": 0.4897068440914154, "epoch": 1.4124444444444444, "percentage": 46.95, "elapsed_time": "0:31:49", "remaining_time": "0:35:57"}
+{"current_steps": 101, "total_steps": 213, "loss": 0.38927626609802246, "lr": 6.419419593900109e-06, "epoch": 1.4266666666666667, "percentage": 47.42, "elapsed_time": "0:32:08", "remaining_time": "0:35:38"}
+{"current_steps": 102, "total_steps": 213, "loss": 0.36544570326805115, "lr": 6.340373993665607e-06, "epoch": 1.4408888888888889, "percentage": 47.89, "elapsed_time": "0:32:26", "remaining_time": "0:35:18"}
+{"current_steps": 103, "total_steps": 213, "loss": 0.4033282995223999, "lr": 6.2609657755527135e-06, "epoch": 1.455111111111111, "percentage": 48.36, "elapsed_time": "0:32:44", "remaining_time": "0:34:58"}
+{"current_steps": 104, "total_steps": 213, "loss": 0.40876060724258423, "lr": 6.1812164222518626e-06, "epoch": 1.4693333333333334, "percentage": 48.83, "elapsed_time": "0:33:03", "remaining_time": "0:34:38"}
+{"current_steps": 105, "total_steps": 213, "loss": 0.3994408845901489, "lr": 6.101147508742456e-06, "epoch": 1.4835555555555555, "percentage": 49.3, "elapsed_time": "0:33:21", "remaining_time": "0:34:18"}
+{"current_steps": 106, "total_steps": 213, "loss": 0.4667425751686096, "lr": 6.020780696456059e-06, "epoch": 1.4977777777777779, "percentage": 49.77, "elapsed_time": "0:33:39", "remaining_time": "0:33:58"}
+{"current_steps": 107, "total_steps": 213, "loss": 0.4092869460582733, "lr": 5.940137727416247e-06, "epoch": 1.512, "percentage": 50.23, "elapsed_time": "0:33:58", "remaining_time": "0:33:39"}
+{"current_steps": 108, "total_steps": 213, "loss": 0.39832064509391785, "lr": 5.859240418356614e-06, "epoch": 1.5262222222222221, "percentage": 50.7, "elapsed_time": "0:34:16", "remaining_time": "0:33:19"}
+{"current_steps": 109, "total_steps": 213, "loss": 0.4114166498184204, "lr": 5.778110654818602e-06, "epoch": 1.5404444444444443, "percentage": 51.17, "elapsed_time": "0:34:34", "remaining_time": "0:32:59"}
+{"current_steps": 110, "total_steps": 213, "loss": 0.4360125660896301, "lr": 5.696770385230679e-06, "epoch": 1.5546666666666666, "percentage": 51.64, "elapsed_time": "0:34:52", "remaining_time": "0:32:39"}
+{"current_steps": 111, "total_steps": 213, "loss": 0.4153256118297577, "lr": 5.615241614970546e-06, "epoch": 1.568888888888889, "percentage": 52.11, "elapsed_time": "0:35:11", "remaining_time": "0:32:20"}
+{"current_steps": 112, "total_steps": 213, "loss": 0.3954656720161438, "lr": 5.533546400411899e-06, "epoch": 1.5831111111111111, "percentage": 52.58, "elapsed_time": "0:35:29", "remaining_time": "0:32:00"}
+{"current_steps": 113, "total_steps": 213, "loss": 0.4052407741546631, "lr": 5.4517068429574215e-06, "epoch": 1.5973333333333333, "percentage": 53.05, "elapsed_time": "0:35:48", "remaining_time": "0:31:40"}
+{"current_steps": 114, "total_steps": 213, "loss": 0.4017290472984314, "lr": 5.3697450830595775e-06, "epoch": 1.6115555555555554, "percentage": 53.52, "elapsed_time": "0:36:06", "remaining_time": "0:31:21"}
+{"current_steps": 115, "total_steps": 213, "loss": 0.38467150926589966, "lr": 5.287683294230855e-06, "epoch": 1.6257777777777778, "percentage": 53.99, "elapsed_time": "0:36:24", "remaining_time": "0:31:01"}
+{"current_steps": 116, "total_steps": 213, "loss": 0.395123690366745, "lr": 5.20554367704505e-06, "epoch": 1.6400000000000001, "percentage": 54.46, "elapsed_time": "0:36:43", "remaining_time": "0:30:42"}
+{"current_steps": 117, "total_steps": 213, "loss": 0.39796996116638184, "lr": 5.123348453131242e-06, "epoch": 1.6542222222222223, "percentage": 54.93, "elapsed_time": "0:37:01", "remaining_time": "0:30:23"}
+{"current_steps": 118, "total_steps": 213, "loss": 0.4162113070487976, "lr": 5.041119859162068e-06, "epoch": 1.6684444444444444, "percentage": 55.4, "elapsed_time": "0:37:20", "remaining_time": "0:30:03"}
+{"current_steps": 119, "total_steps": 213, "loss": 0.42128854990005493, "lr": 4.958880140837934e-06, "epoch": 1.6826666666666665, "percentage": 55.87, "elapsed_time": "0:37:39", "remaining_time": "0:29:44"}
+{"current_steps": 120, "total_steps": 213, "loss": 0.4024372696876526, "lr": 4.876651546868759e-06, "epoch": 1.696888888888889, "percentage": 56.34, "elapsed_time": "0:37:57", "remaining_time": "0:29:24"}
+{"current_steps": 121, "total_steps": 213, "loss": 0.46918508410453796, "lr": 4.794456322954953e-06, "epoch": 1.7111111111111112, "percentage": 56.81, "elapsed_time": "0:38:15", "remaining_time": "0:29:05"}
+{"current_steps": 122, "total_steps": 213, "loss": 0.41863948106765747, "lr": 4.7123167057691446e-06, "epoch": 1.7253333333333334, "percentage": 57.28, "elapsed_time": "0:38:33", "remaining_time": "0:28:45"}
+{"current_steps": 123, "total_steps": 213, "loss": 0.44644808769226074, "lr": 4.630254916940424e-06, "epoch": 1.7395555555555555, "percentage": 57.75, "elapsed_time": "0:38:52", "remaining_time": "0:28:26"}
+{"current_steps": 124, "total_steps": 213, "loss": 0.42086148262023926, "lr": 4.548293157042581e-06, "epoch": 1.7537777777777777, "percentage": 58.22, "elapsed_time": "0:39:10", "remaining_time": "0:28:06"}
+{"current_steps": 125, "total_steps": 213, "loss": 0.4266040027141571, "lr": 4.466453599588103e-06, "epoch": 1.768, "percentage": 58.69, "elapsed_time": "0:39:28", "remaining_time": "0:27:47"}
+{"current_steps": 126, "total_steps": 213, "loss": 0.3592030704021454, "lr": 4.384758385029457e-06, "epoch": 1.7822222222222224, "percentage": 59.15, "elapsed_time": "0:39:47", "remaining_time": "0:27:28"}
+{"current_steps": 127, "total_steps": 213, "loss": 0.42212432622909546, "lr": 4.3032296147693225e-06, "epoch": 1.7964444444444445, "percentage": 59.62, "elapsed_time": "0:40:05", "remaining_time": "0:27:08"}
+{"current_steps": 128, "total_steps": 213, "loss": 0.37964335083961487, "lr": 4.2218893451814005e-06, "epoch": 1.8106666666666666, "percentage": 60.09, "elapsed_time": "0:40:23", "remaining_time": "0:26:49"}
+{"current_steps": 129, "total_steps": 213, "loss": 0.37467944622039795, "lr": 4.140759581643386e-06, "epoch": 1.8248888888888888, "percentage": 60.56, "elapsed_time": "0:40:42", "remaining_time": "0:26:30"}
+{"current_steps": 130, "total_steps": 213, "loss": 0.393369197845459, "lr": 4.059862272583755e-06, "epoch": 1.8391111111111111, "percentage": 61.03, "elapsed_time": "0:41:00", "remaining_time": "0:26:11"}
+{"current_steps": 131, "total_steps": 213, "loss": 0.394899845123291, "lr": 3.979219303543942e-06, "epoch": 1.8533333333333335, "percentage": 61.5, "elapsed_time": "0:41:19", "remaining_time": "0:25:51"}
+{"current_steps": 132, "total_steps": 213, "loss": 0.3997966945171356, "lr": 3.898852491257547e-06, "epoch": 1.8675555555555556, "percentage": 61.97, "elapsed_time": "0:41:37", "remaining_time": "0:25:32"}
+{"current_steps": 133, "total_steps": 213, "loss": 0.44034862518310547, "lr": 3.818783577748138e-06, "epoch": 1.8817777777777778, "percentage": 62.44, "elapsed_time": "0:41:55", "remaining_time": "0:25:13"}
+{"current_steps": 134, "total_steps": 213, "loss": 0.4251548647880554, "lr": 3.7390342244472886e-06, "epoch": 1.896, "percentage": 62.91, "elapsed_time": "0:42:13", "remaining_time": "0:24:53"}
+{"current_steps": 135, "total_steps": 213, "loss": 0.3972621560096741, "lr": 3.659626006334395e-06, "epoch": 1.9102222222222223, "percentage": 63.38, "elapsed_time": "0:42:32", "remaining_time": "0:24:34"}
+{"current_steps": 136, "total_steps": 213, "loss": 0.4237692356109619, "lr": 3.5805804060998926e-06, "epoch": 1.9244444444444444, "percentage": 63.85, "elapsed_time": "0:42:50", "remaining_time": "0:24:15"}
+{"current_steps": 137, "total_steps": 213, "loss": 0.5015424489974976, "lr": 3.501918808333453e-06, "epoch": 1.9386666666666668, "percentage": 64.32, "elapsed_time": "0:43:08", "remaining_time": "0:23:56"}
+{"current_steps": 138, "total_steps": 213, "loss": 0.4541424512863159, "lr": 3.4236624937386874e-06, "epoch": 1.952888888888889, "percentage": 64.79, "elapsed_time": "0:43:26", "remaining_time": "0:23:36"}
+{"current_steps": 139, "total_steps": 213, "loss": 0.42448267340660095, "lr": 3.3458326333759927e-06, "epoch": 1.967111111111111, "percentage": 65.26, "elapsed_time": "0:43:45", "remaining_time": "0:23:17"}
+{"current_steps": 140, "total_steps": 213, "loss": 0.4514390528202057, "lr": 3.268450282935026e-06, "epoch": 1.9813333333333332, "percentage": 65.73, "elapsed_time": "0:44:03", "remaining_time": "0:22:58"}
+{"current_steps": 141, "total_steps": 213, "loss": 0.38879817724227905, "lr": 3.1915363770384223e-06, "epoch": 1.9955555555555555, "percentage": 66.2, "elapsed_time": "0:44:21", "remaining_time": "0:22:39"}
+{"current_steps": 142, "total_steps": 213, "loss": 0.3734951913356781, "lr": 3.1151117235782346e-06, "epoch": 2.0, "percentage": 66.67, "elapsed_time": "0:44:27", "remaining_time": "0:22:13"}
+{"current_steps": 143, "total_steps": 213, "loss": 0.3868422210216522, "lr": 3.0391969980866874e-06, "epoch": 2.014222222222222, "percentage": 67.14, "elapsed_time": "0:45:37", "remaining_time": "0:22:20"}
+{"current_steps": 144, "total_steps": 213, "loss": 0.3441769480705261, "lr": 2.963812738142713e-06, "epoch": 2.0284444444444443, "percentage": 67.61, "elapsed_time": "0:45:56", "remaining_time": "0:22:00"}
+{"current_steps": 145, "total_steps": 213, "loss": 0.3698546290397644, "lr": 2.8889793378158284e-06, "epoch": 2.042666666666667, "percentage": 68.08, "elapsed_time": "0:46:14", "remaining_time": "0:21:41"}
+{"current_steps": 146, "total_steps": 213, "loss": 0.3752461075782776, "lr": 2.814717042148827e-06, "epoch": 2.056888888888889, "percentage": 68.54, "elapsed_time": "0:46:32", "remaining_time": "0:21:21"}
+{"current_steps": 147, "total_steps": 213, "loss": 0.3570592999458313, "lr": 2.7410459416807856e-06, "epoch": 2.071111111111111, "percentage": 69.01, "elapsed_time": "0:46:51", "remaining_time": "0:21:02"}
+{"current_steps": 148, "total_steps": 213, "loss": 0.3812662363052368, "lr": 2.6679859670118785e-06, "epoch": 2.0853333333333333, "percentage": 69.48, "elapsed_time": "0:47:09", "remaining_time": "0:20:42"}
+{"current_steps": 149, "total_steps": 213, "loss": 0.3505663275718689, "lr": 2.5955568834114523e-06, "epoch": 2.0995555555555554, "percentage": 69.95, "elapsed_time": "0:47:28", "remaining_time": "0:20:23"}
+{"current_steps": 150, "total_steps": 213, "loss": 0.35809028148651123, "lr": 2.523778285470835e-06, "epoch": 2.113777777777778, "percentage": 70.42, "elapsed_time": "0:47:46", "remaining_time": "0:20:04"}
+{"current_steps": 150, "total_steps": 213, "eval_loss": 0.49233150482177734, "epoch": 2.113777777777778, "percentage": 70.42, "elapsed_time": "0:47:50", "remaining_time": "0:20:05"}
+{"current_steps": 151, "total_steps": 213, "loss": 0.32126089930534363, "lr": 2.452669591802307e-06, "epoch": 2.128, "percentage": 70.89, "elapsed_time": "0:48:09", "remaining_time": "0:19:46"}
+{"current_steps": 152, "total_steps": 213, "loss": 0.3627392649650574, "lr": 2.3822500397857016e-06, "epoch": 2.1422222222222222, "percentage": 71.36, "elapsed_time": "0:48:27", "remaining_time": "0:19:26"}
+{"current_steps": 153, "total_steps": 213, "loss": 0.36303800344467163, "lr": 2.3125386803640188e-06, "epoch": 2.1564444444444444, "percentage": 71.83, "elapsed_time": "0:48:46", "remaining_time": "0:19:07"}
+{"current_steps": 154, "total_steps": 213, "loss": 0.3294636011123657, "lr": 2.243554372889479e-06, "epoch": 2.1706666666666665, "percentage": 72.3, "elapsed_time": "0:49:04", "remaining_time": "0:18:48"}
+{"current_steps": 155, "total_steps": 213, "loss": 0.35944664478302, "lr": 2.1753157800214107e-06, "epoch": 2.1848888888888887, "percentage": 72.77, "elapsed_time": "0:49:22", "remaining_time": "0:18:28"}
+{"current_steps": 156, "total_steps": 213, "loss": 0.35921794176101685, "lr": 2.1078413626773547e-06, "epoch": 2.1991111111111112, "percentage": 73.24, "elapsed_time": "0:49:41", "remaining_time": "0:18:09"}
+{"current_steps": 157, "total_steps": 213, "loss": 0.3335600197315216, "lr": 2.0411493750387423e-06, "epoch": 2.2133333333333334, "percentage": 73.71, "elapsed_time": "0:49:59", "remaining_time": "0:17:49"}
+{"current_steps": 158, "total_steps": 213, "loss": 0.3621842861175537, "lr": 1.9752578596124955e-06, "epoch": 2.2275555555555555, "percentage": 74.18, "elapsed_time": "0:50:17", "remaining_time": "0:17:30"}
+{"current_steps": 159, "total_steps": 213, "loss": 0.34454649686813354, "lr": 1.9101846423499113e-06, "epoch": 2.2417777777777776, "percentage": 74.65, "elapsed_time": "0:50:36", "remaining_time": "0:17:11"}
+{"current_steps": 160, "total_steps": 213, "loss": 0.3537226915359497, "lr": 1.8459473278241125e-06, "epoch": 2.2560000000000002, "percentage": 75.12, "elapsed_time": "0:50:54", "remaining_time": "0:16:51"}
+{"current_steps": 161, "total_steps": 213, "loss": 0.3517017662525177, "lr": 1.7825632944674016e-06, "epoch": 2.2702222222222224, "percentage": 75.59, "elapsed_time": "0:51:12", "remaining_time": "0:16:32"}
+{"current_steps": 162, "total_steps": 213, "loss": 0.34244149923324585, "lr": 1.7200496898697832e-06, "epoch": 2.2844444444444445, "percentage": 76.06, "elapsed_time": "0:51:31", "remaining_time": "0:16:13"}
+{"current_steps": 163, "total_steps": 213, "loss": 0.35110828280448914, "lr": 1.6584234261399535e-06, "epoch": 2.2986666666666666, "percentage": 76.53, "elapsed_time": "0:51:49", "remaining_time": "0:15:53"}
+{"current_steps": 164, "total_steps": 213, "loss": 0.34295642375946045, "lr": 1.5977011753299726e-06, "epoch": 2.3128888888888888, "percentage": 77.0, "elapsed_time": "0:52:07", "remaining_time": "0:15:34"}
+{"current_steps": 165, "total_steps": 213, "loss": 0.4045950770378113, "lr": 1.5378993649249053e-06, "epoch": 2.327111111111111, "percentage": 77.46, "elapsed_time": "0:52:26", "remaining_time": "0:15:15"}
+{"current_steps": 166, "total_steps": 213, "loss": 0.3649445176124573, "lr": 1.4790341733986085e-06, "epoch": 2.3413333333333335, "percentage": 77.93, "elapsed_time": "0:52:44", "remaining_time": "0:14:55"}
+{"current_steps": 167, "total_steps": 213, "loss": 0.3201165795326233, "lr": 1.4211215258368866e-06, "epoch": 2.3555555555555556, "percentage": 78.4, "elapsed_time": "0:53:03", "remaining_time": "0:14:36"}
+{"current_steps": 168, "total_steps": 213, "loss": 0.345939040184021, "lr": 1.3641770896292083e-06, "epoch": 2.3697777777777778, "percentage": 78.87, "elapsed_time": "0:53:21", "remaining_time": "0:14:17"}
+{"current_steps": 169, "total_steps": 213, "loss": 0.35247528553009033, "lr": 1.3082162702301276e-06, "epoch": 2.384, "percentage": 79.34, "elapsed_time": "0:53:39", "remaining_time": "0:13:58"}
+{"current_steps": 170, "total_steps": 213, "loss": 0.40231946110725403, "lr": 1.2532542069915722e-06, "epoch": 2.398222222222222, "percentage": 79.81, "elapsed_time": "0:53:58", "remaining_time": "0:13:39"}
+{"current_steps": 171, "total_steps": 213, "loss": 0.3605418801307678, "lr": 1.1993057690671174e-06, "epoch": 2.4124444444444446, "percentage": 80.28, "elapsed_time": "0:54:16", "remaining_time": "0:13:19"}
+{"current_steps": 172, "total_steps": 213, "loss": 0.382945716381073, "lr": 1.1463855513893695e-06, "epoch": 2.4266666666666667, "percentage": 80.75, "elapsed_time": "0:54:35", "remaining_time": "0:13:00"}
+{"current_steps": 173, "total_steps": 213, "loss": 0.36510083079338074, "lr": 1.0945078707215224e-06, "epoch": 2.440888888888889, "percentage": 81.22, "elapsed_time": "0:54:53", "remaining_time": "0:12:41"}
+{"current_steps": 174, "total_steps": 213, "loss": 0.38933348655700684, "lr": 1.0436867617841768e-06, "epoch": 2.455111111111111, "percentage": 81.69, "elapsed_time": "0:55:11", "remaining_time": "0:12:22"}
+{"current_steps": 175, "total_steps": 213, "loss": 0.3710438013076782, "lr": 9.939359734584552e-07, "epoch": 2.469333333333333, "percentage": 82.16, "elapsed_time": "0:55:29", "remaining_time": "0:12:03"}
+{"current_steps": 176, "total_steps": 213, "loss": 0.36227256059646606, "lr": 9.452689650664515e-07, "epoch": 2.4835555555555557, "percentage": 82.63, "elapsed_time": "0:55:48", "remaining_time": "0:11:43"}
+{"current_steps": 177, "total_steps": 213, "loss": 0.36987176537513733, "lr": 8.976989027300265e-07, "epoch": 2.497777777777778, "percentage": 83.1, "elapsed_time": "0:56:06", "remaining_time": "0:11:24"}
+{"current_steps": 178, "total_steps": 213, "loss": 0.35104453563690186, "lr": 8.512386558088919e-07, "epoch": 2.512, "percentage": 83.57, "elapsed_time": "0:56:25", "remaining_time": "0:11:05"}
+{"current_steps": 179, "total_steps": 213, "loss": 0.3681498169898987, "lr": 8.059007934190194e-07, "epoch": 2.526222222222222, "percentage": 84.04, "elapsed_time": "0:56:43", "remaining_time": "0:10:46"}
+{"current_steps": 180, "total_steps": 213, "loss": 0.353618860244751, "lr": 7.61697581032243e-07, "epoch": 2.5404444444444443, "percentage": 84.51, "elapsed_time": "0:57:01", "remaining_time": "0:10:27"}
+{"current_steps": 181, "total_steps": 213, "loss": 0.35408419370651245, "lr": 7.186409771580355e-07, "epoch": 2.554666666666667, "percentage": 84.98, "elapsed_time": "0:57:20", "remaining_time": "0:10:08"}
+{"current_steps": 182, "total_steps": 213, "loss": 0.3726443648338318, "lr": 6.76742630108298e-07, "epoch": 2.568888888888889, "percentage": 85.45, "elapsed_time": "0:57:38", "remaining_time": "0:09:49"}
+{"current_steps": 183, "total_steps": 213, "loss": 0.3561905026435852, "lr": 6.360138748461015e-07, "epoch": 2.583111111111111, "percentage": 85.92, "elapsed_time": "0:57:56", "remaining_time": "0:09:29"}
+{"current_steps": 184, "total_steps": 213, "loss": 0.3148964047431946, "lr": 5.964657299191712e-07, "epoch": 2.5973333333333333, "percentage": 86.38, "elapsed_time": "0:58:14", "remaining_time": "0:09:10"}
+{"current_steps": 185, "total_steps": 213, "loss": 0.3561857342720032, "lr": 5.581088944789953e-07, "epoch": 2.6115555555555554, "percentage": 86.85, "elapsed_time": "0:58:33", "remaining_time": "0:08:51"}
+{"current_steps": 186, "total_steps": 213, "loss": 0.3500719964504242, "lr": 5.20953745386329e-07, "epoch": 2.6257777777777775, "percentage": 87.32, "elapsed_time": "0:58:51", "remaining_time": "0:08:32"}
+{"current_steps": 187, "total_steps": 213, "loss": 0.3637757897377014, "lr": 4.850103344038853e-07, "epoch": 2.64, "percentage": 87.79, "elapsed_time": "0:59:09", "remaining_time": "0:08:13"}
+{"current_steps": 188, "total_steps": 213, "loss": 0.32206273078918457, "lr": 4.502883854769935e-07, "epoch": 2.6542222222222223, "percentage": 88.26, "elapsed_time": "0:59:28", "remaining_time": "0:07:54"}
+{"current_steps": 189, "total_steps": 213, "loss": 0.33746200799942017, "lr": 4.167972921029262e-07, "epoch": 2.6684444444444444, "percentage": 88.73, "elapsed_time": "0:59:46", "remaining_time": "0:07:35"}
+{"current_steps": 190, "total_steps": 213, "loss": 0.38060814142227173, "lr": 3.8454611478963235e-07, "epoch": 2.6826666666666665, "percentage": 89.2, "elapsed_time": "1:00:04", "remaining_time": "0:07:16"}
+{"current_steps": 191, "total_steps": 213, "loss": 0.3551645278930664, "lr": 3.535435786045538e-07, "epoch": 2.696888888888889, "percentage": 89.67, "elapsed_time": "1:00:23", "remaining_time": "0:06:57"}
+{"current_steps": 192, "total_steps": 213, "loss": 0.35464680194854736, "lr": 3.237980708141919e-07, "epoch": 2.7111111111111112, "percentage": 90.14, "elapsed_time": "1:00:41", "remaining_time": "0:06:38"}
+{"current_steps": 193, "total_steps": 213, "loss": 0.3980952501296997, "lr": 2.9531763861505967e-07, "epoch": 2.7253333333333334, "percentage": 90.61, "elapsed_time": "1:00:59", "remaining_time": "0:06:19"}
+{"current_steps": 194, "total_steps": 213, "loss": 0.33311402797698975, "lr": 2.681099869566328e-07, "epoch": 2.7395555555555555, "percentage": 91.08, "elapsed_time": "1:01:18", "remaining_time": "0:06:00"}
+{"current_steps": 195, "total_steps": 213, "loss": 0.36939185857772827, "lr": 2.4218247645689306e-07, "epoch": 2.7537777777777777, "percentage": 91.55, "elapsed_time": "1:01:36", "remaining_time": "0:05:41"}
+{"current_steps": 196, "total_steps": 213, "loss": 0.3736470937728882, "lr": 2.175421214110235e-07, "epoch": 2.768, "percentage": 92.02, "elapsed_time": "1:01:54", "remaining_time": "0:05:22"}
+{"current_steps": 197, "total_steps": 213, "loss": 0.35068604350090027, "lr": 1.941955878938029e-07, "epoch": 2.7822222222222224, "percentage": 92.49, "elapsed_time": "1:02:13", "remaining_time": "0:05:03"}
+{"current_steps": 198, "total_steps": 213, "loss": 0.3597073554992676, "lr": 1.721491919561913e-07, "epoch": 2.7964444444444445, "percentage": 92.96, "elapsed_time": "1:02:31", "remaining_time": "0:04:44"}
+{"current_steps": 199, "total_steps": 213, "loss": 0.3420804440975189, "lr": 1.514088979166256e-07, "epoch": 2.8106666666666666, "percentage": 93.43, "elapsed_time": "1:02:49", "remaining_time": "0:04:25"}
+{"current_steps": 200, "total_steps": 213, "loss": 0.3697103261947632, "lr": 1.3198031674745814e-07, "epoch": 2.824888888888889, "percentage": 93.9, "elapsed_time": "1:03:08", "remaining_time": "0:04:06"}
+{"current_steps": 200, "total_steps": 213, "eval_loss": 0.49903449416160583, "epoch": 2.824888888888889, "percentage": 93.9, "elapsed_time": "1:03:12", "remaining_time": "0:04:06"}
+{"current_steps": 201, "total_steps": 213, "loss": 0.35418957471847534, "lr": 1.138687045569975e-07, "epoch": 2.8391111111111114, "percentage": 94.37, "elapsed_time": "1:03:30", "remaining_time": "0:03:47"}
+{"current_steps": 202, "total_steps": 213, "loss": 0.3809206783771515, "lr": 9.70789611675449e-08, "epoch": 2.8533333333333335, "percentage": 94.84, "elapsed_time": "1:03:48", "remaining_time": "0:03:28"}
+{"current_steps": 203, "total_steps": 213, "loss": 0.33729785680770874, "lr": 8.161562878982399e-08, "epoch": 2.8675555555555556, "percentage": 95.31, "elapsed_time": "1:04:07", "remaining_time": "0:03:09"}
+{"current_steps": 204, "total_steps": 213, "loss": 0.3682692050933838, "lr": 6.74828907941516e-08, "epoch": 2.8817777777777778, "percentage": 95.77, "elapsed_time": "1:04:25", "remaining_time": "0:02:50"}
+{"current_steps": 205, "total_steps": 213, "loss": 0.40091854333877563, "lr": 5.468457057869358e-08, "epoch": 2.896, "percentage": 96.24, "elapsed_time": "1:04:43", "remaining_time": "0:02:31"}
+{"current_steps": 206, "total_steps": 213, "loss": 0.338018000125885, "lr": 4.322413053509944e-08, "epoch": 2.910222222222222, "percentage": 96.71, "elapsed_time": "1:05:02", "remaining_time": "0:02:12"}
+{"current_steps": 207, "total_steps": 213, "loss": 0.3696953058242798, "lr": 3.31046711118066e-08, "epoch": 2.924444444444444, "percentage": 97.18, "elapsed_time": "1:05:20", "remaining_time": "0:01:53"}
+{"current_steps": 208, "total_steps": 213, "loss": 0.3460949659347534, "lr": 2.4328929975260262e-08, "epoch": 2.9386666666666668, "percentage": 97.65, "elapsed_time": "1:05:39", "remaining_time": "0:01:34"}
+{"current_steps": 209, "total_steps": 213, "loss": 0.35688620805740356, "lr": 1.6899281269279756e-08, "epoch": 2.952888888888889, "percentage": 98.12, "elapsed_time": "1:05:57", "remaining_time": "0:01:15"}
+{"current_steps": 210, "total_steps": 213, "loss": 0.39211517572402954, "lr": 1.0817734972768946e-08, "epoch": 2.967111111111111, "percentage": 98.59, "elapsed_time": "1:06:15", "remaining_time": "0:00:56"}
+{"current_steps": 211, "total_steps": 213, "loss": 0.38052815198898315, "lr": 6.085936355947897e-09, "epoch": 2.981333333333333, "percentage": 99.06, "elapsed_time": "1:06:34", "remaining_time": "0:00:37"}
+{"current_steps": 212, "total_steps": 213, "loss": 0.3612944185733795, "lr": 2.7051655352494654e-09, "epoch": 2.9955555555555557, "percentage": 99.53, "elapsed_time": "1:06:52", "remaining_time": "0:00:18"}
+{"current_steps": 213, "total_steps": 213, "loss": 0.5174569487571716, "lr": 6.763371270035457e-10, "epoch": 3.0, "percentage": 100.0, "elapsed_time": "1:06:58", "remaining_time": "0:00:00"}
+{"current_steps": 213, "total_steps": 213, "epoch": 3.0, "percentage": 100.0, "elapsed_time": "1:07:47", "remaining_time": "0:00:00"}
+{"current_steps": 213, "total_steps": 108, "epoch": 3.0, "percentage": 197.22, "elapsed_time": "0:00:00", "remaining_time": "0:00:00"}
diff --git a/qwen2.5-1.5b-instruct/trainer_state.json b/qwen2.5-1.5b-instruct/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..2fb010ee3da364bf5e21ecc6769d96b9bac09e9a
--- /dev/null
+++ b/qwen2.5-1.5b-instruct/trainer_state.json
@@ -0,0 +1,1566 @@
+{
+ "best_global_step": null,
+ "best_metric": null,
+ "best_model_checkpoint": null,
+ "epoch": 3.0,
+ "eval_steps": 50,
+ "global_step": 213,
+ "is_hyper_param_search": false,
+ "is_local_process_zero": true,
+ "is_world_process_zero": true,
+ "log_history": [
+ {
+ "epoch": 0.014222222222222223,
+ "grad_norm": 0.10009177774190903,
+ "learning_rate": 0.0,
+ "loss": 0.704937219619751,
+ "step": 1
+ },
+ {
+ "epoch": 0.028444444444444446,
+ "grad_norm": 0.10451912879943848,
+ "learning_rate": 4.5454545454545457e-07,
+ "loss": 0.7318670749664307,
+ "step": 2
+ },
+ {
+ "epoch": 0.042666666666666665,
+ "grad_norm": 0.07928935438394547,
+ "learning_rate": 9.090909090909091e-07,
+ "loss": 0.6017104387283325,
+ "step": 3
+ },
+ {
+ "epoch": 0.05688888888888889,
+ "grad_norm": 0.048069290816783905,
+ "learning_rate": 1.3636363636363636e-06,
+ "loss": 0.6029032468795776,
+ "step": 4
+ },
+ {
+ "epoch": 0.07111111111111111,
+ "grad_norm": 0.02866927720606327,
+ "learning_rate": 1.8181818181818183e-06,
+ "loss": 0.5051919221878052,
+ "step": 5
+ },
+ {
+ "epoch": 0.08533333333333333,
+ "grad_norm": 0.022201435640454292,
+ "learning_rate": 2.2727272727272728e-06,
+ "loss": 0.5942293405532837,
+ "step": 6
+ },
+ {
+ "epoch": 0.09955555555555555,
+ "grad_norm": 0.018263086676597595,
+ "learning_rate": 2.7272727272727272e-06,
+ "loss": 0.5781338214874268,
+ "step": 7
+ },
+ {
+ "epoch": 0.11377777777777778,
+ "grad_norm": 0.01827927492558956,
+ "learning_rate": 3.181818181818182e-06,
+ "loss": 0.5180214047431946,
+ "step": 8
+ },
+ {
+ "epoch": 0.128,
+ "grad_norm": 0.014687157236039639,
+ "learning_rate": 3.6363636363636366e-06,
+ "loss": 0.5420116186141968,
+ "step": 9
+ },
+ {
+ "epoch": 0.14222222222222222,
+ "grad_norm": 0.012042663060128689,
+ "learning_rate": 4.0909090909090915e-06,
+ "loss": 0.5345908403396606,
+ "step": 10
+ },
+ {
+ "epoch": 0.15644444444444444,
+ "grad_norm": 0.012934447266161442,
+ "learning_rate": 4.5454545454545455e-06,
+ "loss": 0.46689414978027344,
+ "step": 11
+ },
+ {
+ "epoch": 0.17066666666666666,
+ "grad_norm": 0.012295619584619999,
+ "learning_rate": 5e-06,
+ "loss": 0.5062131881713867,
+ "step": 12
+ },
+ {
+ "epoch": 0.18488888888888888,
+ "grad_norm": 0.011856353841722012,
+ "learning_rate": 5.4545454545454545e-06,
+ "loss": 0.5063096880912781,
+ "step": 13
+ },
+ {
+ "epoch": 0.1991111111111111,
+ "grad_norm": 0.009451798163354397,
+ "learning_rate": 5.90909090909091e-06,
+ "loss": 0.5130602121353149,
+ "step": 14
+ },
+ {
+ "epoch": 0.21333333333333335,
+ "grad_norm": 0.01111016795039177,
+ "learning_rate": 6.363636363636364e-06,
+ "loss": 0.5016045570373535,
+ "step": 15
+ },
+ {
+ "epoch": 0.22755555555555557,
+ "grad_norm": 0.010735263116657734,
+ "learning_rate": 6.818181818181818e-06,
+ "loss": 0.49827253818511963,
+ "step": 16
+ },
+ {
+ "epoch": 0.24177777777777779,
+ "grad_norm": 0.010710126720368862,
+ "learning_rate": 7.272727272727273e-06,
+ "loss": 0.479993999004364,
+ "step": 17
+ },
+ {
+ "epoch": 0.256,
+ "grad_norm": 0.009600787423551083,
+ "learning_rate": 7.727272727272727e-06,
+ "loss": 0.4925128221511841,
+ "step": 18
+ },
+ {
+ "epoch": 0.2702222222222222,
+ "grad_norm": 0.010486789979040623,
+ "learning_rate": 8.181818181818183e-06,
+ "loss": 0.5188871622085571,
+ "step": 19
+ },
+ {
+ "epoch": 0.28444444444444444,
+ "grad_norm": 0.013396799564361572,
+ "learning_rate": 8.636363636363637e-06,
+ "loss": 0.5192517042160034,
+ "step": 20
+ },
+ {
+ "epoch": 0.2986666666666667,
+ "grad_norm": 0.009299715980887413,
+ "learning_rate": 9.090909090909091e-06,
+ "loss": 0.4580157399177551,
+ "step": 21
+ },
+ {
+ "epoch": 0.3128888888888889,
+ "grad_norm": 0.009691477753221989,
+ "learning_rate": 9.545454545454547e-06,
+ "loss": 0.49455106258392334,
+ "step": 22
+ },
+ {
+ "epoch": 0.32711111111111113,
+ "grad_norm": 0.008896945044398308,
+ "learning_rate": 1e-05,
+ "loss": 0.48672014474868774,
+ "step": 23
+ },
+ {
+ "epoch": 0.3413333333333333,
+ "grad_norm": 0.009234798140823841,
+ "learning_rate": 9.999323662872998e-06,
+ "loss": 0.5052967071533203,
+ "step": 24
+ },
+ {
+ "epoch": 0.35555555555555557,
+ "grad_norm": 0.009167873300611973,
+ "learning_rate": 9.99729483446475e-06,
+ "loss": 0.4740435779094696,
+ "step": 25
+ },
+ {
+ "epoch": 0.36977777777777776,
+ "grad_norm": 0.009001555852591991,
+ "learning_rate": 9.993914063644053e-06,
+ "loss": 0.47315946221351624,
+ "step": 26
+ },
+ {
+ "epoch": 0.384,
+ "grad_norm": 0.00915393978357315,
+ "learning_rate": 9.989182265027232e-06,
+ "loss": 0.4557237923145294,
+ "step": 27
+ },
+ {
+ "epoch": 0.3982222222222222,
+ "grad_norm": 0.008505849167704582,
+ "learning_rate": 9.98310071873072e-06,
+ "loss": 0.46317678689956665,
+ "step": 28
+ },
+ {
+ "epoch": 0.41244444444444445,
+ "grad_norm": 0.008668503724038601,
+ "learning_rate": 9.975671070024741e-06,
+ "loss": 0.4533114433288574,
+ "step": 29
+ },
+ {
+ "epoch": 0.4266666666666667,
+ "grad_norm": 0.009284070692956448,
+ "learning_rate": 9.966895328888195e-06,
+ "loss": 0.5624005794525146,
+ "step": 30
+ },
+ {
+ "epoch": 0.4408888888888889,
+ "grad_norm": 0.00935022346675396,
+ "learning_rate": 9.956775869464901e-06,
+ "loss": 0.4420597553253174,
+ "step": 31
+ },
+ {
+ "epoch": 0.45511111111111113,
+ "grad_norm": 0.008682326413691044,
+ "learning_rate": 9.945315429421307e-06,
+ "loss": 0.4745720624923706,
+ "step": 32
+ },
+ {
+ "epoch": 0.4693333333333333,
+ "grad_norm": 0.007816251367330551,
+ "learning_rate": 9.932517109205849e-06,
+ "loss": 0.439736545085907,
+ "step": 33
+ },
+ {
+ "epoch": 0.48355555555555557,
+ "grad_norm": 0.009315130300819874,
+ "learning_rate": 9.918384371210178e-06,
+ "loss": 0.5357447862625122,
+ "step": 34
+ },
+ {
+ "epoch": 0.49777777777777776,
+ "grad_norm": 0.009141312912106514,
+ "learning_rate": 9.902921038832456e-06,
+ "loss": 0.519463062286377,
+ "step": 35
+ },
+ {
+ "epoch": 0.512,
+ "grad_norm": 0.008826862089335918,
+ "learning_rate": 9.886131295443003e-06,
+ "loss": 0.48377716541290283,
+ "step": 36
+ },
+ {
+ "epoch": 0.5262222222222223,
+ "grad_norm": 0.008700878359377384,
+ "learning_rate": 9.868019683252543e-06,
+ "loss": 0.48436683416366577,
+ "step": 37
+ },
+ {
+ "epoch": 0.5404444444444444,
+ "grad_norm": 0.011412105523049831,
+ "learning_rate": 9.848591102083375e-06,
+ "loss": 0.4949312210083008,
+ "step": 38
+ },
+ {
+ "epoch": 0.5546666666666666,
+ "grad_norm": 0.009214675053954124,
+ "learning_rate": 9.82785080804381e-06,
+ "loss": 0.4901079535484314,
+ "step": 39
+ },
+ {
+ "epoch": 0.5688888888888889,
+ "grad_norm": 0.01163570862263441,
+ "learning_rate": 9.805804412106197e-06,
+ "loss": 0.6070311665534973,
+ "step": 40
+ },
+ {
+ "epoch": 0.5831111111111111,
+ "grad_norm": 0.009325762279331684,
+ "learning_rate": 9.782457878588977e-06,
+ "loss": 0.48412975668907166,
+ "step": 41
+ },
+ {
+ "epoch": 0.5973333333333334,
+ "grad_norm": 0.008266489021480083,
+ "learning_rate": 9.75781752354311e-06,
+ "loss": 0.4440661072731018,
+ "step": 42
+ },
+ {
+ "epoch": 0.6115555555555555,
+ "grad_norm": 0.00843313243240118,
+ "learning_rate": 9.731890013043367e-06,
+ "loss": 0.502442479133606,
+ "step": 43
+ },
+ {
+ "epoch": 0.6257777777777778,
+ "grad_norm": 0.00934070535004139,
+ "learning_rate": 9.704682361384941e-06,
+ "loss": 0.47924768924713135,
+ "step": 44
+ },
+ {
+ "epoch": 0.64,
+ "grad_norm": 0.010360248386859894,
+ "learning_rate": 9.676201929185809e-06,
+ "loss": 0.4161364436149597,
+ "step": 45
+ },
+ {
+ "epoch": 0.6542222222222223,
+ "grad_norm": 0.0094041982665658,
+ "learning_rate": 9.646456421395447e-06,
+ "loss": 0.48306071758270264,
+ "step": 46
+ },
+ {
+ "epoch": 0.6684444444444444,
+ "grad_norm": 0.009070150554180145,
+ "learning_rate": 9.615453885210368e-06,
+ "loss": 0.4774249792098999,
+ "step": 47
+ },
+ {
+ "epoch": 0.6826666666666666,
+ "grad_norm": 0.009771023876965046,
+ "learning_rate": 9.583202707897075e-06,
+ "loss": 0.4948921203613281,
+ "step": 48
+ },
+ {
+ "epoch": 0.6968888888888889,
+ "grad_norm": 0.008749130181968212,
+ "learning_rate": 9.549711614523007e-06,
+ "loss": 0.4825429916381836,
+ "step": 49
+ },
+ {
+ "epoch": 0.7111111111111111,
+ "grad_norm": 0.007630312815308571,
+ "learning_rate": 9.514989665596114e-06,
+ "loss": 0.46055319905281067,
+ "step": 50
+ },
+ {
+ "epoch": 0.7111111111111111,
+ "eval_loss": 0.48781517148017883,
+ "eval_runtime": 4.5903,
+ "eval_samples_per_second": 43.57,
+ "eval_steps_per_second": 21.785,
+ "step": 50
+ },
+ {
+ "epoch": 0.7253333333333334,
+ "grad_norm": 0.008408056572079659,
+ "learning_rate": 9.479046254613673e-06,
+ "loss": 0.4626312255859375,
+ "step": 51
+ },
+ {
+ "epoch": 0.7395555555555555,
+ "grad_norm": 0.007699670735746622,
+ "learning_rate": 9.441891105521005e-06,
+ "loss": 0.4499070644378662,
+ "step": 52
+ },
+ {
+ "epoch": 0.7537777777777778,
+ "grad_norm": 0.009070169180631638,
+ "learning_rate": 9.40353427008083e-06,
+ "loss": 0.4825657606124878,
+ "step": 53
+ },
+ {
+ "epoch": 0.768,
+ "grad_norm": 0.007915646769106388,
+ "learning_rate": 9.3639861251539e-06,
+ "loss": 0.4534468650817871,
+ "step": 54
+ },
+ {
+ "epoch": 0.7822222222222223,
+ "grad_norm": 0.011336776427924633,
+ "learning_rate": 9.323257369891702e-06,
+ "loss": 0.4988631010055542,
+ "step": 55
+ },
+ {
+ "epoch": 0.7964444444444444,
+ "grad_norm": 0.00761911366134882,
+ "learning_rate": 9.281359022841966e-06,
+ "loss": 0.4371437728404999,
+ "step": 56
+ },
+ {
+ "epoch": 0.8106666666666666,
+ "grad_norm": 0.009316994808614254,
+ "learning_rate": 9.238302418967757e-06,
+ "loss": 0.4956932067871094,
+ "step": 57
+ },
+ {
+ "epoch": 0.8248888888888889,
+ "grad_norm": 0.009821794927120209,
+ "learning_rate": 9.194099206580981e-06,
+ "loss": 0.5392974615097046,
+ "step": 58
+ },
+ {
+ "epoch": 0.8391111111111111,
+ "grad_norm": 0.008422416634857655,
+ "learning_rate": 9.14876134419111e-06,
+ "loss": 0.47983357310295105,
+ "step": 59
+ },
+ {
+ "epoch": 0.8533333333333334,
+ "grad_norm": 0.00837004091590643,
+ "learning_rate": 9.102301097269974e-06,
+ "loss": 0.5183510780334473,
+ "step": 60
+ },
+ {
+ "epoch": 0.8675555555555555,
+ "grad_norm": 0.008596297353506088,
+ "learning_rate": 9.05473103493355e-06,
+ "loss": 0.47292786836624146,
+ "step": 61
+ },
+ {
+ "epoch": 0.8817777777777778,
+ "grad_norm": 0.009059637784957886,
+ "learning_rate": 9.006064026541549e-06,
+ "loss": 0.5188573598861694,
+ "step": 62
+ },
+ {
+ "epoch": 0.896,
+ "grad_norm": 0.007816917262971401,
+ "learning_rate": 8.956313238215824e-06,
+ "loss": 0.45641395449638367,
+ "step": 63
+ },
+ {
+ "epoch": 0.9102222222222223,
+ "grad_norm": 0.007985202595591545,
+ "learning_rate": 8.905492129278478e-06,
+ "loss": 0.4586361050605774,
+ "step": 64
+ },
+ {
+ "epoch": 0.9244444444444444,
+ "grad_norm": 0.007170964498072863,
+ "learning_rate": 8.85361444861063e-06,
+ "loss": 0.4394378066062927,
+ "step": 65
+ },
+ {
+ "epoch": 0.9386666666666666,
+ "grad_norm": 0.00909162126481533,
+ "learning_rate": 8.800694230932885e-06,
+ "loss": 0.509820818901062,
+ "step": 66
+ },
+ {
+ "epoch": 0.9528888888888889,
+ "grad_norm": 0.00786044355481863,
+ "learning_rate": 8.74674579300843e-06,
+ "loss": 0.43584880232810974,
+ "step": 67
+ },
+ {
+ "epoch": 0.9671111111111111,
+ "grad_norm": 0.00832363311201334,
+ "learning_rate": 8.691783729769874e-06,
+ "loss": 0.4317135214805603,
+ "step": 68
+ },
+ {
+ "epoch": 0.9813333333333333,
+ "grad_norm": 0.008376957848668098,
+ "learning_rate": 8.635822910370793e-06,
+ "loss": 0.49933016300201416,
+ "step": 69
+ },
+ {
+ "epoch": 0.9955555555555555,
+ "grad_norm": 0.00835558120161295,
+ "learning_rate": 8.578878474163115e-06,
+ "loss": 0.46570393443107605,
+ "step": 70
+ },
+ {
+ "epoch": 1.0,
+ "grad_norm": 0.013256405480206013,
+ "learning_rate": 8.520965826601394e-06,
+ "loss": 0.45107918977737427,
+ "step": 71
+ },
+ {
+ "epoch": 1.0142222222222221,
+ "grad_norm": 0.008708071894943714,
+ "learning_rate": 8.462100635075097e-06,
+ "loss": 0.47125527262687683,
+ "step": 72
+ },
+ {
+ "epoch": 1.0284444444444445,
+ "grad_norm": 0.008086243644356728,
+ "learning_rate": 8.40229882467003e-06,
+ "loss": 0.4322492480278015,
+ "step": 73
+ },
+ {
+ "epoch": 1.0426666666666666,
+ "grad_norm": 0.0073950947262346745,
+ "learning_rate": 8.341576573860049e-06,
+ "loss": 0.38362830877304077,
+ "step": 74
+ },
+ {
+ "epoch": 1.056888888888889,
+ "grad_norm": 0.0076777078211307526,
+ "learning_rate": 8.279950310130218e-06,
+ "loss": 0.3878062665462494,
+ "step": 75
+ },
+ {
+ "epoch": 1.0711111111111111,
+ "grad_norm": 0.007898081094026566,
+ "learning_rate": 8.2174367055326e-06,
+ "loss": 0.3833811283111572,
+ "step": 76
+ },
+ {
+ "epoch": 1.0853333333333333,
+ "grad_norm": 0.007348628249019384,
+ "learning_rate": 8.154052672175888e-06,
+ "loss": 0.3845909833908081,
+ "step": 77
+ },
+ {
+ "epoch": 1.0995555555555556,
+ "grad_norm": 0.008542248979210854,
+ "learning_rate": 8.08981535765009e-06,
+ "loss": 0.44259265065193176,
+ "step": 78
+ },
+ {
+ "epoch": 1.1137777777777778,
+ "grad_norm": 0.007998447865247726,
+ "learning_rate": 8.024742140387506e-06,
+ "loss": 0.38503497838974,
+ "step": 79
+ },
+ {
+ "epoch": 1.1280000000000001,
+ "grad_norm": 0.008362009190022945,
+ "learning_rate": 7.95885062496126e-06,
+ "loss": 0.42608046531677246,
+ "step": 80
+ },
+ {
+ "epoch": 1.1422222222222222,
+ "grad_norm": 0.007770548574626446,
+ "learning_rate": 7.892158637322647e-06,
+ "loss": 0.38991105556488037,
+ "step": 81
+ },
+ {
+ "epoch": 1.1564444444444444,
+ "grad_norm": 0.007649772334843874,
+ "learning_rate": 7.824684219978591e-06,
+ "loss": 0.36854323744773865,
+ "step": 82
+ },
+ {
+ "epoch": 1.1706666666666667,
+ "grad_norm": 0.008056909777224064,
+ "learning_rate": 7.756445627110523e-06,
+ "loss": 0.4200327396392822,
+ "step": 83
+ },
+ {
+ "epoch": 1.1848888888888889,
+ "grad_norm": 0.0089965695515275,
+ "learning_rate": 7.68746131963598e-06,
+ "loss": 0.4378758668899536,
+ "step": 84
+ },
+ {
+ "epoch": 1.199111111111111,
+ "grad_norm": 0.007884320802986622,
+ "learning_rate": 7.6177499602143e-06,
+ "loss": 0.3808164894580841,
+ "step": 85
+ },
+ {
+ "epoch": 1.2133333333333334,
+ "grad_norm": 0.009615465067327023,
+ "learning_rate": 7.547330408197695e-06,
+ "loss": 0.4453170895576477,
+ "step": 86
+ },
+ {
+ "epoch": 1.2275555555555555,
+ "grad_norm": 0.00891174841672182,
+ "learning_rate": 7.476221714529167e-06,
+ "loss": 0.4456181526184082,
+ "step": 87
+ },
+ {
+ "epoch": 1.2417777777777779,
+ "grad_norm": 0.008833018131554127,
+ "learning_rate": 7.404443116588548e-06,
+ "loss": 0.43805548548698425,
+ "step": 88
+ },
+ {
+ "epoch": 1.256,
+ "grad_norm": 0.008171891793608665,
+ "learning_rate": 7.332014032988123e-06,
+ "loss": 0.4144285321235657,
+ "step": 89
+ },
+ {
+ "epoch": 1.2702222222222221,
+ "grad_norm": 0.009090668521821499,
+ "learning_rate": 7.2589540583192165e-06,
+ "loss": 0.38928651809692383,
+ "step": 90
+ },
+ {
+ "epoch": 1.2844444444444445,
+ "grad_norm": 0.008077106438577175,
+ "learning_rate": 7.185282957851175e-06,
+ "loss": 0.387723445892334,
+ "step": 91
+ },
+ {
+ "epoch": 1.2986666666666666,
+ "grad_norm": 0.008803839795291424,
+ "learning_rate": 7.111020662184174e-06,
+ "loss": 0.4158015549182892,
+ "step": 92
+ },
+ {
+ "epoch": 1.3128888888888888,
+ "grad_norm": 0.008249706588685513,
+ "learning_rate": 7.036187261857289e-06,
+ "loss": 0.39764612913131714,
+ "step": 93
+ },
+ {
+ "epoch": 1.3271111111111111,
+ "grad_norm": 0.008549178950488567,
+ "learning_rate": 6.960803001913315e-06,
+ "loss": 0.38216379284858704,
+ "step": 94
+ },
+ {
+ "epoch": 1.3413333333333333,
+ "grad_norm": 0.007755788508802652,
+ "learning_rate": 6.884888276421766e-06,
+ "loss": 0.413602352142334,
+ "step": 95
+ },
+ {
+ "epoch": 1.3555555555555556,
+ "grad_norm": 0.008742124773561954,
+ "learning_rate": 6.8084636229615786e-06,
+ "loss": 0.41756486892700195,
+ "step": 96
+ },
+ {
+ "epoch": 1.3697777777777778,
+ "grad_norm": 0.008473540656268597,
+ "learning_rate": 6.731549717064975e-06,
+ "loss": 0.4028819799423218,
+ "step": 97
+ },
+ {
+ "epoch": 1.384,
+ "grad_norm": 0.008585155010223389,
+ "learning_rate": 6.654167366624009e-06,
+ "loss": 0.37123921513557434,
+ "step": 98
+ },
+ {
+ "epoch": 1.3982222222222223,
+ "grad_norm": 0.00882649328559637,
+ "learning_rate": 6.576337506261314e-06,
+ "loss": 0.39443182945251465,
+ "step": 99
+ },
+ {
+ "epoch": 1.4124444444444444,
+ "grad_norm": 0.008174613118171692,
+ "learning_rate": 6.498081191666549e-06,
+ "loss": 0.42232874035835266,
+ "step": 100
+ },
+ {
+ "epoch": 1.4124444444444444,
+ "eval_loss": 0.4897068440914154,
+ "eval_runtime": 5.0728,
+ "eval_samples_per_second": 39.426,
+ "eval_steps_per_second": 19.713,
+ "step": 100
+ },
+ {
+ "epoch": 1.4266666666666667,
+ "grad_norm": 0.00753920478746295,
+ "learning_rate": 6.419419593900109e-06,
+ "loss": 0.38927626609802246,
+ "step": 101
+ },
+ {
+ "epoch": 1.4408888888888889,
+ "grad_norm": 0.007967391051352024,
+ "learning_rate": 6.340373993665607e-06,
+ "loss": 0.36544570326805115,
+ "step": 102
+ },
+ {
+ "epoch": 1.455111111111111,
+ "grad_norm": 0.008992813527584076,
+ "learning_rate": 6.2609657755527135e-06,
+ "loss": 0.4033282995223999,
+ "step": 103
+ },
+ {
+ "epoch": 1.4693333333333334,
+ "grad_norm": 0.007906445302069187,
+ "learning_rate": 6.1812164222518626e-06,
+ "loss": 0.40876060724258423,
+ "step": 104
+ },
+ {
+ "epoch": 1.4835555555555555,
+ "grad_norm": 0.00823493767529726,
+ "learning_rate": 6.101147508742456e-06,
+ "loss": 0.3994408845901489,
+ "step": 105
+ },
+ {
+ "epoch": 1.4977777777777779,
+ "grad_norm": 0.009032925590872765,
+ "learning_rate": 6.020780696456059e-06,
+ "loss": 0.4667425751686096,
+ "step": 106
+ },
+ {
+ "epoch": 1.512,
+ "grad_norm": 0.007943308912217617,
+ "learning_rate": 5.940137727416247e-06,
+ "loss": 0.4092869460582733,
+ "step": 107
+ },
+ {
+ "epoch": 1.5262222222222221,
+ "grad_norm": 0.008353099226951599,
+ "learning_rate": 5.859240418356614e-06,
+ "loss": 0.39832064509391785,
+ "step": 108
+ },
+ {
+ "epoch": 1.5404444444444443,
+ "grad_norm": 0.008241581730544567,
+ "learning_rate": 5.778110654818602e-06,
+ "loss": 0.4114166498184204,
+ "step": 109
+ },
+ {
+ "epoch": 1.5546666666666666,
+ "grad_norm": 0.007937194779515266,
+ "learning_rate": 5.696770385230679e-06,
+ "loss": 0.4360125660896301,
+ "step": 110
+ },
+ {
+ "epoch": 1.568888888888889,
+ "grad_norm": 0.008319812826812267,
+ "learning_rate": 5.615241614970546e-06,
+ "loss": 0.4153256118297577,
+ "step": 111
+ },
+ {
+ "epoch": 1.5831111111111111,
+ "grad_norm": 0.008011740632355213,
+ "learning_rate": 5.533546400411899e-06,
+ "loss": 0.3954656720161438,
+ "step": 112
+ },
+ {
+ "epoch": 1.5973333333333333,
+ "grad_norm": 0.008084246888756752,
+ "learning_rate": 5.4517068429574215e-06,
+ "loss": 0.4052407741546631,
+ "step": 113
+ },
+ {
+ "epoch": 1.6115555555555554,
+ "grad_norm": 0.007481739390641451,
+ "learning_rate": 5.3697450830595775e-06,
+ "loss": 0.4017290472984314,
+ "step": 114
+ },
+ {
+ "epoch": 1.6257777777777778,
+ "grad_norm": 0.008050751872360706,
+ "learning_rate": 5.287683294230855e-06,
+ "loss": 0.38467150926589966,
+ "step": 115
+ },
+ {
+ "epoch": 1.6400000000000001,
+ "grad_norm": 0.0077187251299619675,
+ "learning_rate": 5.20554367704505e-06,
+ "loss": 0.395123690366745,
+ "step": 116
+ },
+ {
+ "epoch": 1.6542222222222223,
+ "grad_norm": 0.008423185907304287,
+ "learning_rate": 5.123348453131242e-06,
+ "loss": 0.39796996116638184,
+ "step": 117
+ },
+ {
+ "epoch": 1.6684444444444444,
+ "grad_norm": 0.009052549488842487,
+ "learning_rate": 5.041119859162068e-06,
+ "loss": 0.4162113070487976,
+ "step": 118
+ },
+ {
+ "epoch": 1.6826666666666665,
+ "grad_norm": 0.008246202953159809,
+ "learning_rate": 4.958880140837934e-06,
+ "loss": 0.42128854990005493,
+ "step": 119
+ },
+ {
+ "epoch": 1.696888888888889,
+ "grad_norm": 0.00833312701433897,
+ "learning_rate": 4.876651546868759e-06,
+ "loss": 0.4024372696876526,
+ "step": 120
+ },
+ {
+ "epoch": 1.7111111111111112,
+ "grad_norm": 0.008461604826152325,
+ "learning_rate": 4.794456322954953e-06,
+ "loss": 0.46918508410453796,
+ "step": 121
+ },
+ {
+ "epoch": 1.7253333333333334,
+ "grad_norm": 0.00825911108404398,
+ "learning_rate": 4.7123167057691446e-06,
+ "loss": 0.41863948106765747,
+ "step": 122
+ },
+ {
+ "epoch": 1.7395555555555555,
+ "grad_norm": 0.009252563118934631,
+ "learning_rate": 4.630254916940424e-06,
+ "loss": 0.44644808769226074,
+ "step": 123
+ },
+ {
+ "epoch": 1.7537777777777777,
+ "grad_norm": 0.007843364961445332,
+ "learning_rate": 4.548293157042581e-06,
+ "loss": 0.42086148262023926,
+ "step": 124
+ },
+ {
+ "epoch": 1.768,
+ "grad_norm": 0.007915403693914413,
+ "learning_rate": 4.466453599588103e-06,
+ "loss": 0.4266040027141571,
+ "step": 125
+ },
+ {
+ "epoch": 1.7822222222222224,
+ "grad_norm": 0.007302967831492424,
+ "learning_rate": 4.384758385029457e-06,
+ "loss": 0.3592030704021454,
+ "step": 126
+ },
+ {
+ "epoch": 1.7964444444444445,
+ "grad_norm": 0.00869770348072052,
+ "learning_rate": 4.3032296147693225e-06,
+ "loss": 0.42212432622909546,
+ "step": 127
+ },
+ {
+ "epoch": 1.8106666666666666,
+ "grad_norm": 0.008091778494417667,
+ "learning_rate": 4.2218893451814005e-06,
+ "loss": 0.37964335083961487,
+ "step": 128
+ },
+ {
+ "epoch": 1.8248888888888888,
+ "grad_norm": 0.0076453504152596,
+ "learning_rate": 4.140759581643386e-06,
+ "loss": 0.37467944622039795,
+ "step": 129
+ },
+ {
+ "epoch": 1.8391111111111111,
+ "grad_norm": 0.007656064350157976,
+ "learning_rate": 4.059862272583755e-06,
+ "loss": 0.393369197845459,
+ "step": 130
+ },
+ {
+ "epoch": 1.8533333333333335,
+ "grad_norm": 0.008399168029427528,
+ "learning_rate": 3.979219303543942e-06,
+ "loss": 0.394899845123291,
+ "step": 131
+ },
+ {
+ "epoch": 1.8675555555555556,
+ "grad_norm": 0.008833736181259155,
+ "learning_rate": 3.898852491257547e-06,
+ "loss": 0.3997966945171356,
+ "step": 132
+ },
+ {
+ "epoch": 1.8817777777777778,
+ "grad_norm": 0.008364683948457241,
+ "learning_rate": 3.818783577748138e-06,
+ "loss": 0.44034862518310547,
+ "step": 133
+ },
+ {
+ "epoch": 1.896,
+ "grad_norm": 0.008851971477270126,
+ "learning_rate": 3.7390342244472886e-06,
+ "loss": 0.4251548647880554,
+ "step": 134
+ },
+ {
+ "epoch": 1.9102222222222223,
+ "grad_norm": 0.007921694777905941,
+ "learning_rate": 3.659626006334395e-06,
+ "loss": 0.3972621560096741,
+ "step": 135
+ },
+ {
+ "epoch": 1.9244444444444444,
+ "grad_norm": 0.008714950643479824,
+ "learning_rate": 3.5805804060998926e-06,
+ "loss": 0.4237692356109619,
+ "step": 136
+ },
+ {
+ "epoch": 1.9386666666666668,
+ "grad_norm": 0.008546526543796062,
+ "learning_rate": 3.501918808333453e-06,
+ "loss": 0.5015424489974976,
+ "step": 137
+ },
+ {
+ "epoch": 1.952888888888889,
+ "grad_norm": 0.008581602945923805,
+ "learning_rate": 3.4236624937386874e-06,
+ "loss": 0.4541424512863159,
+ "step": 138
+ },
+ {
+ "epoch": 1.967111111111111,
+ "grad_norm": 0.008913579396903515,
+ "learning_rate": 3.3458326333759927e-06,
+ "loss": 0.42448267340660095,
+ "step": 139
+ },
+ {
+ "epoch": 1.9813333333333332,
+ "grad_norm": 0.008540489710867405,
+ "learning_rate": 3.268450282935026e-06,
+ "loss": 0.4514390528202057,
+ "step": 140
+ },
+ {
+ "epoch": 1.9955555555555555,
+ "grad_norm": 0.007863311097025871,
+ "learning_rate": 3.1915363770384223e-06,
+ "loss": 0.38879817724227905,
+ "step": 141
+ },
+ {
+ "epoch": 2.0,
+ "grad_norm": 0.013070300221443176,
+ "learning_rate": 3.1151117235782346e-06,
+ "loss": 0.3734951913356781,
+ "step": 142
+ },
+ {
+ "epoch": 2.014222222222222,
+ "grad_norm": 0.00914511177688837,
+ "learning_rate": 3.0391969980866874e-06,
+ "loss": 0.3868422210216522,
+ "step": 143
+ },
+ {
+ "epoch": 2.0284444444444443,
+ "grad_norm": 0.008015231229364872,
+ "learning_rate": 2.963812738142713e-06,
+ "loss": 0.3441769480705261,
+ "step": 144
+ },
+ {
+ "epoch": 2.042666666666667,
+ "grad_norm": 0.008334742859005928,
+ "learning_rate": 2.8889793378158284e-06,
+ "loss": 0.3698546290397644,
+ "step": 145
+ },
+ {
+ "epoch": 2.056888888888889,
+ "grad_norm": 0.0077324071899056435,
+ "learning_rate": 2.814717042148827e-06,
+ "loss": 0.3752461075782776,
+ "step": 146
+ },
+ {
+ "epoch": 2.071111111111111,
+ "grad_norm": 0.007629899773746729,
+ "learning_rate": 2.7410459416807856e-06,
+ "loss": 0.3570592999458313,
+ "step": 147
+ },
+ {
+ "epoch": 2.0853333333333333,
+ "grad_norm": 0.008662652224302292,
+ "learning_rate": 2.6679859670118785e-06,
+ "loss": 0.3812662363052368,
+ "step": 148
+ },
+ {
+ "epoch": 2.0995555555555554,
+ "grad_norm": 0.007828332483768463,
+ "learning_rate": 2.5955568834114523e-06,
+ "loss": 0.3505663275718689,
+ "step": 149
+ },
+ {
+ "epoch": 2.113777777777778,
+ "grad_norm": 0.0074827345088124275,
+ "learning_rate": 2.523778285470835e-06,
+ "loss": 0.35809028148651123,
+ "step": 150
+ },
+ {
+ "epoch": 2.113777777777778,
+ "eval_loss": 0.49233150482177734,
+ "eval_runtime": 4.1869,
+ "eval_samples_per_second": 47.768,
+ "eval_steps_per_second": 23.884,
+ "step": 150
+ },
+ {
+ "epoch": 2.128,
+ "grad_norm": 0.007365878205746412,
+ "learning_rate": 2.452669591802307e-06,
+ "loss": 0.32126089930534363,
+ "step": 151
+ },
+ {
+ "epoch": 2.1422222222222222,
+ "grad_norm": 0.007244541309773922,
+ "learning_rate": 2.3822500397857016e-06,
+ "loss": 0.3627392649650574,
+ "step": 152
+ },
+ {
+ "epoch": 2.1564444444444444,
+ "grad_norm": 0.00791338924318552,
+ "learning_rate": 2.3125386803640188e-06,
+ "loss": 0.36303800344467163,
+ "step": 153
+ },
+ {
+ "epoch": 2.1706666666666665,
+ "grad_norm": 0.007537501864135265,
+ "learning_rate": 2.243554372889479e-06,
+ "loss": 0.3294636011123657,
+ "step": 154
+ },
+ {
+ "epoch": 2.1848888888888887,
+ "grad_norm": 0.008005894720554352,
+ "learning_rate": 2.1753157800214107e-06,
+ "loss": 0.35944664478302,
+ "step": 155
+ },
+ {
+ "epoch": 2.1991111111111112,
+ "grad_norm": 0.008430157788097858,
+ "learning_rate": 2.1078413626773547e-06,
+ "loss": 0.35921794176101685,
+ "step": 156
+ },
+ {
+ "epoch": 2.2133333333333334,
+ "grad_norm": 0.00803217850625515,
+ "learning_rate": 2.0411493750387423e-06,
+ "loss": 0.3335600197315216,
+ "step": 157
+ },
+ {
+ "epoch": 2.2275555555555555,
+ "grad_norm": 0.007777590304613113,
+ "learning_rate": 1.9752578596124955e-06,
+ "loss": 0.3621842861175537,
+ "step": 158
+ },
+ {
+ "epoch": 2.2417777777777776,
+ "grad_norm": 0.007962247356772423,
+ "learning_rate": 1.9101846423499113e-06,
+ "loss": 0.34454649686813354,
+ "step": 159
+ },
+ {
+ "epoch": 2.2560000000000002,
+ "grad_norm": 0.007974164560437202,
+ "learning_rate": 1.8459473278241125e-06,
+ "loss": 0.3537226915359497,
+ "step": 160
+ },
+ {
+ "epoch": 2.2702222222222224,
+ "grad_norm": 0.007959051057696342,
+ "learning_rate": 1.7825632944674016e-06,
+ "loss": 0.3517017662525177,
+ "step": 161
+ },
+ {
+ "epoch": 2.2844444444444445,
+ "grad_norm": 0.008884078823029995,
+ "learning_rate": 1.7200496898697832e-06,
+ "loss": 0.34244149923324585,
+ "step": 162
+ },
+ {
+ "epoch": 2.2986666666666666,
+ "grad_norm": 0.008463534526526928,
+ "learning_rate": 1.6584234261399535e-06,
+ "loss": 0.35110828280448914,
+ "step": 163
+ },
+ {
+ "epoch": 2.3128888888888888,
+ "grad_norm": 0.008407258428633213,
+ "learning_rate": 1.5977011753299726e-06,
+ "loss": 0.34295642375946045,
+ "step": 164
+ },
+ {
+ "epoch": 2.327111111111111,
+ "grad_norm": 0.008790143765509129,
+ "learning_rate": 1.5378993649249053e-06,
+ "loss": 0.4045950770378113,
+ "step": 165
+ },
+ {
+ "epoch": 2.3413333333333335,
+ "grad_norm": 0.008037876337766647,
+ "learning_rate": 1.4790341733986085e-06,
+ "loss": 0.3649445176124573,
+ "step": 166
+ },
+ {
+ "epoch": 2.3555555555555556,
+ "grad_norm": 0.007316106464713812,
+ "learning_rate": 1.4211215258368866e-06,
+ "loss": 0.3201165795326233,
+ "step": 167
+ },
+ {
+ "epoch": 2.3697777777777778,
+ "grad_norm": 0.00920423399657011,
+ "learning_rate": 1.3641770896292083e-06,
+ "loss": 0.345939040184021,
+ "step": 168
+ },
+ {
+ "epoch": 2.384,
+ "grad_norm": 0.008216304704546928,
+ "learning_rate": 1.3082162702301276e-06,
+ "loss": 0.35247528553009033,
+ "step": 169
+ },
+ {
+ "epoch": 2.398222222222222,
+ "grad_norm": 0.008257976733148098,
+ "learning_rate": 1.2532542069915722e-06,
+ "loss": 0.40231946110725403,
+ "step": 170
+ },
+ {
+ "epoch": 2.4124444444444446,
+ "grad_norm": 0.008689168840646744,
+ "learning_rate": 1.1993057690671174e-06,
+ "loss": 0.3605418801307678,
+ "step": 171
+ },
+ {
+ "epoch": 2.4266666666666667,
+ "grad_norm": 0.008199715055525303,
+ "learning_rate": 1.1463855513893695e-06,
+ "loss": 0.382945716381073,
+ "step": 172
+ },
+ {
+ "epoch": 2.440888888888889,
+ "grad_norm": 0.00789699237793684,
+ "learning_rate": 1.0945078707215224e-06,
+ "loss": 0.36510083079338074,
+ "step": 173
+ },
+ {
+ "epoch": 2.455111111111111,
+ "grad_norm": 0.008184579201042652,
+ "learning_rate": 1.0436867617841768e-06,
+ "loss": 0.38933348655700684,
+ "step": 174
+ },
+ {
+ "epoch": 2.469333333333333,
+ "grad_norm": 0.007531109265983105,
+ "learning_rate": 9.939359734584552e-07,
+ "loss": 0.3710438013076782,
+ "step": 175
+ },
+ {
+ "epoch": 2.4835555555555557,
+ "grad_norm": 0.008019962348043919,
+ "learning_rate": 9.452689650664515e-07,
+ "loss": 0.36227256059646606,
+ "step": 176
+ },
+ {
+ "epoch": 2.497777777777778,
+ "grad_norm": 0.007935193367302418,
+ "learning_rate": 8.976989027300265e-07,
+ "loss": 0.36987176537513733,
+ "step": 177
+ },
+ {
+ "epoch": 2.512,
+ "grad_norm": 0.007968267425894737,
+ "learning_rate": 8.512386558088919e-07,
+ "loss": 0.35104453563690186,
+ "step": 178
+ },
+ {
+ "epoch": 2.526222222222222,
+ "grad_norm": 0.008082241751253605,
+ "learning_rate": 8.059007934190194e-07,
+ "loss": 0.3681498169898987,
+ "step": 179
+ },
+ {
+ "epoch": 2.5404444444444443,
+ "grad_norm": 0.00833115540444851,
+ "learning_rate": 7.61697581032243e-07,
+ "loss": 0.353618860244751,
+ "step": 180
+ },
+ {
+ "epoch": 2.554666666666667,
+ "grad_norm": 0.008011356927454472,
+ "learning_rate": 7.186409771580355e-07,
+ "loss": 0.35408419370651245,
+ "step": 181
+ },
+ {
+ "epoch": 2.568888888888889,
+ "grad_norm": 0.00763704115524888,
+ "learning_rate": 6.76742630108298e-07,
+ "loss": 0.3726443648338318,
+ "step": 182
+ },
+ {
+ "epoch": 2.583111111111111,
+ "grad_norm": 0.007385968696326017,
+ "learning_rate": 6.360138748461015e-07,
+ "loss": 0.3561905026435852,
+ "step": 183
+ },
+ {
+ "epoch": 2.5973333333333333,
+ "grad_norm": 0.00736213056370616,
+ "learning_rate": 5.964657299191712e-07,
+ "loss": 0.3148964047431946,
+ "step": 184
+ },
+ {
+ "epoch": 2.6115555555555554,
+ "grad_norm": 0.008179187774658203,
+ "learning_rate": 5.581088944789953e-07,
+ "loss": 0.3561857342720032,
+ "step": 185
+ },
+ {
+ "epoch": 2.6257777777777775,
+ "grad_norm": 0.008100884966552258,
+ "learning_rate": 5.20953745386329e-07,
+ "loss": 0.3500719964504242,
+ "step": 186
+ },
+ {
+ "epoch": 2.64,
+ "grad_norm": 0.007978768087923527,
+ "learning_rate": 4.850103344038853e-07,
+ "loss": 0.3637757897377014,
+ "step": 187
+ },
+ {
+ "epoch": 2.6542222222222223,
+ "grad_norm": 0.007483206689357758,
+ "learning_rate": 4.502883854769935e-07,
+ "loss": 0.32206273078918457,
+ "step": 188
+ },
+ {
+ "epoch": 2.6684444444444444,
+ "grad_norm": 0.007899755612015724,
+ "learning_rate": 4.167972921029262e-07,
+ "loss": 0.33746200799942017,
+ "step": 189
+ },
+ {
+ "epoch": 2.6826666666666665,
+ "grad_norm": 0.008544563315808773,
+ "learning_rate": 3.8454611478963235e-07,
+ "loss": 0.38060814142227173,
+ "step": 190
+ },
+ {
+ "epoch": 2.696888888888889,
+ "grad_norm": 0.00889600533992052,
+ "learning_rate": 3.535435786045538e-07,
+ "loss": 0.3551645278930664,
+ "step": 191
+ },
+ {
+ "epoch": 2.7111111111111112,
+ "grad_norm": 0.007875156588852406,
+ "learning_rate": 3.237980708141919e-07,
+ "loss": 0.35464680194854736,
+ "step": 192
+ },
+ {
+ "epoch": 2.7253333333333334,
+ "grad_norm": 0.008730563335120678,
+ "learning_rate": 2.9531763861505967e-07,
+ "loss": 0.3980952501296997,
+ "step": 193
+ },
+ {
+ "epoch": 2.7395555555555555,
+ "grad_norm": 0.008214394561946392,
+ "learning_rate": 2.681099869566328e-07,
+ "loss": 0.33311402797698975,
+ "step": 194
+ },
+ {
+ "epoch": 2.7537777777777777,
+ "grad_norm": 0.007815193384885788,
+ "learning_rate": 2.4218247645689306e-07,
+ "loss": 0.36939185857772827,
+ "step": 195
+ },
+ {
+ "epoch": 2.768,
+ "grad_norm": 0.008210226893424988,
+ "learning_rate": 2.175421214110235e-07,
+ "loss": 0.3736470937728882,
+ "step": 196
+ },
+ {
+ "epoch": 2.7822222222222224,
+ "grad_norm": 0.007955148816108704,
+ "learning_rate": 1.941955878938029e-07,
+ "loss": 0.35068604350090027,
+ "step": 197
+ },
+ {
+ "epoch": 2.7964444444444445,
+ "grad_norm": 0.008125124499201775,
+ "learning_rate": 1.721491919561913e-07,
+ "loss": 0.3597073554992676,
+ "step": 198
+ },
+ {
+ "epoch": 2.8106666666666666,
+ "grad_norm": 0.007504038512706757,
+ "learning_rate": 1.514088979166256e-07,
+ "loss": 0.3420804440975189,
+ "step": 199
+ },
+ {
+ "epoch": 2.824888888888889,
+ "grad_norm": 0.007838167250156403,
+ "learning_rate": 1.3198031674745814e-07,
+ "loss": 0.3697103261947632,
+ "step": 200
+ },
+ {
+ "epoch": 2.824888888888889,
+ "eval_loss": 0.49903449416160583,
+ "eval_runtime": 4.213,
+ "eval_samples_per_second": 47.473,
+ "eval_steps_per_second": 23.736,
+ "step": 200
+ },
+ {
+ "epoch": 2.8391111111111114,
+ "grad_norm": 0.008086729794740677,
+ "learning_rate": 1.138687045569975e-07,
+ "loss": 0.35418957471847534,
+ "step": 201
+ },
+ {
+ "epoch": 2.8533333333333335,
+ "grad_norm": 0.007895248010754585,
+ "learning_rate": 9.70789611675449e-08,
+ "loss": 0.3809206783771515,
+ "step": 202
+ },
+ {
+ "epoch": 2.8675555555555556,
+ "grad_norm": 0.007653921376913786,
+ "learning_rate": 8.161562878982399e-08,
+ "loss": 0.33729785680770874,
+ "step": 203
+ },
+ {
+ "epoch": 2.8817777777777778,
+ "grad_norm": 0.00890001654624939,
+ "learning_rate": 6.74828907941516e-08,
+ "loss": 0.3682692050933838,
+ "step": 204
+ },
+ {
+ "epoch": 2.896,
+ "grad_norm": 0.008669495582580566,
+ "learning_rate": 5.468457057869358e-08,
+ "loss": 0.40091854333877563,
+ "step": 205
+ },
+ {
+ "epoch": 2.910222222222222,
+ "grad_norm": 0.00770574901252985,
+ "learning_rate": 4.322413053509944e-08,
+ "loss": 0.338018000125885,
+ "step": 206
+ },
+ {
+ "epoch": 2.924444444444444,
+ "grad_norm": 0.008314141072332859,
+ "learning_rate": 3.31046711118066e-08,
+ "loss": 0.3696953058242798,
+ "step": 207
+ },
+ {
+ "epoch": 2.9386666666666668,
+ "grad_norm": 0.007442758418619633,
+ "learning_rate": 2.4328929975260262e-08,
+ "loss": 0.3460949659347534,
+ "step": 208
+ },
+ {
+ "epoch": 2.952888888888889,
+ "grad_norm": 0.008314011618494987,
+ "learning_rate": 1.6899281269279756e-08,
+ "loss": 0.35688620805740356,
+ "step": 209
+ },
+ {
+ "epoch": 2.967111111111111,
+ "grad_norm": 0.007803616113960743,
+ "learning_rate": 1.0817734972768946e-08,
+ "loss": 0.39211517572402954,
+ "step": 210
+ },
+ {
+ "epoch": 2.981333333333333,
+ "grad_norm": 0.008699736557900906,
+ "learning_rate": 6.085936355947897e-09,
+ "loss": 0.38052815198898315,
+ "step": 211
+ },
+ {
+ "epoch": 2.9955555555555557,
+ "grad_norm": 0.007957417517900467,
+ "learning_rate": 2.7051655352494654e-09,
+ "loss": 0.3612944185733795,
+ "step": 212
+ },
+ {
+ "epoch": 3.0,
+ "grad_norm": 0.014429433271288872,
+ "learning_rate": 6.763371270035457e-10,
+ "loss": 0.5174569487571716,
+ "step": 213
+ },
+ {
+ "epoch": 3.0,
+ "step": 213,
+ "total_flos": 4.868007593928294e+16,
+ "train_loss": 0.0,
+ "train_runtime": 3.3055,
+ "train_samples_per_second": 16336.51,
+ "train_steps_per_second": 32.673
+ }
+ ],
+ "logging_steps": 1,
+ "max_steps": 108,
+ "num_input_tokens_seen": 0,
+ "num_train_epochs": 3,
+ "save_steps": 500,
+ "stateful_callbacks": {
+ "TrainerControl": {
+ "args": {
+ "should_epoch_stop": false,
+ "should_evaluate": false,
+ "should_log": false,
+ "should_save": true,
+ "should_training_stop": true
+ },
+ "attributes": {}
+ }
+ },
+ "total_flos": 4.868007593928294e+16,
+ "train_batch_size": 1,
+ "trial_name": null,
+ "trial_params": null
+}
diff --git a/qwen2.5-1.5b-instruct/training_args.bin b/qwen2.5-1.5b-instruct/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..06a3f8bda7300a19edb6bd2d3d3f004fc404cee0
--- /dev/null
+++ b/qwen2.5-1.5b-instruct/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:62ad4295000dff7a1bf02ff3b916927f3c891fffd6b84367629501b0265c42b0
+size 5713
diff --git a/qwen2.5-1.5b-instruct/training_eval_loss.png b/qwen2.5-1.5b-instruct/training_eval_loss.png
new file mode 100644
index 0000000000000000000000000000000000000000..3df5f72616ba045a1db34ef0ab62c13b33733bd3
Binary files /dev/null and b/qwen2.5-1.5b-instruct/training_eval_loss.png differ
diff --git a/qwen2.5-1.5b-instruct/training_loss.png b/qwen2.5-1.5b-instruct/training_loss.png
new file mode 100644
index 0000000000000000000000000000000000000000..56fea257d0e2265da049c34284a9ac529c7d4a6c
Binary files /dev/null and b/qwen2.5-1.5b-instruct/training_loss.png differ
diff --git a/qwen2.5-3b-instruct/README.md b/qwen2.5-3b-instruct/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..32617082bb01da37a9d6147e23a914cd7d3cd5a8
--- /dev/null
+++ b/qwen2.5-3b-instruct/README.md
@@ -0,0 +1,69 @@
+---
+library_name: transformers
+license: other
+base_model: Qwen/Qwen2.5-3B-Instruct
+tags:
+- llama-factory
+- full
+- generated_from_trainer
+model-index:
+- name: qwen2.5-3b-instruct
+ results: []
+---
+
+
+
+# qwen2.5-3b-instruct
+
+This model is a fine-tuned version of [Qwen/Qwen2.5-3B-Instruct](https://huggingface.co/Qwen/Qwen2.5-3B-Instruct) on the codealpaca_train_20k_no_trace_generation dataset.
+It achieves the following results on the evaluation set:
+- Loss: 0.4735
+
+## Model description
+
+More information needed
+
+## Intended uses & limitations
+
+More information needed
+
+## Training and evaluation data
+
+More information needed
+
+## Training procedure
+
+### Training hyperparameters
+
+The following hyperparameters were used during training:
+- learning_rate: 1e-05
+- train_batch_size: 1
+- eval_batch_size: 1
+- seed: 42
+- distributed_type: multi-GPU
+- num_devices: 2
+- gradient_accumulation_steps: 128
+- total_train_batch_size: 256
+- total_eval_batch_size: 2
+- optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
+- lr_scheduler_type: cosine
+- lr_scheduler_warmup_steps: 0.1
+- num_epochs: 3
+
+### Training results
+
+| Training Loss | Epoch | Step | Validation Loss |
+|:-------------:|:------:|:----:|:---------------:|
+| 0.4377 | 0.7111 | 50 | 0.4685 |
+| 0.4175 | 1.4124 | 100 | 0.4669 |
+| 0.3545 | 2.1138 | 150 | 0.4679 |
+| 0.3659 | 2.8249 | 200 | 0.4734 |
+
+
+### Framework versions
+
+- Transformers 5.2.0
+- Pytorch 2.11.0+cu128
+- Datasets 4.0.0
+- Tokenizers 0.22.2
diff --git a/qwen2.5-3b-instruct/all_results.json b/qwen2.5-3b-instruct/all_results.json
new file mode 100644
index 0000000000000000000000000000000000000000..b3776ba2af0b77c8126c38a71040037ed03c82dc
--- /dev/null
+++ b/qwen2.5-3b-instruct/all_results.json
@@ -0,0 +1,12 @@
+{
+ "epoch": 3.0,
+ "eval_loss": 0.47346630692481995,
+ "eval_runtime": 5.3972,
+ "eval_samples_per_second": 37.056,
+ "eval_steps_per_second": 18.528,
+ "total_flos": 1.0308479847759872e+17,
+ "train_loss": 0.4208552807989255,
+ "train_runtime": 5691.1072,
+ "train_samples_per_second": 9.488,
+ "train_steps_per_second": 0.037
+}
\ No newline at end of file
diff --git a/qwen2.5-3b-instruct/chat_template.jinja b/qwen2.5-3b-instruct/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..bdf7919a96cfe43d50914a007b9c0877bd0ec27e
--- /dev/null
+++ b/qwen2.5-3b-instruct/chat_template.jinja
@@ -0,0 +1,54 @@
+{%- if tools %}
+ {{- '<|im_start|>system\n' }}
+ {%- if messages[0]['role'] == 'system' %}
+ {{- messages[0]['content'] }}
+ {%- else %}
+ {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+ {%- endif %}
+ {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within XML tags:\n" }}
+ {%- for tool in tools %}
+ {{- "\n" }}
+ {{- tool | tojson }}
+ {%- endfor %}
+ {{- "\n\n\nFor each function call, return a json object with function name and arguments within XML tags:\n\n{\"name\": , \"arguments\": }\n<|im_end|>\n" }}
+{%- else %}
+ {%- if messages[0]['role'] == 'system' %}
+ {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+ {%- else %}
+ {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+ {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+ {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+ {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+ {%- elif message.role == "assistant" %}
+ {{- '<|im_start|>' + message.role }}
+ {%- if message.content %}
+ {{- '\n' + message.content }}
+ {%- endif %}
+ {%- for tool_call in message.tool_calls %}
+ {%- if tool_call.function is defined %}
+ {%- set tool_call = tool_call.function %}
+ {%- endif %}
+ {{- '\n\n{"name": "' }}
+ {{- tool_call.name }}
+ {{- '", "arguments": ' }}
+ {{- tool_call.arguments | tojson }}
+ {{- '}\n' }}
+ {%- endfor %}
+ {{- '<|im_end|>\n' }}
+ {%- elif message.role == "tool" %}
+ {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+ {{- '<|im_start|>user' }}
+ {%- endif %}
+ {{- '\n\n' }}
+ {{- message.content }}
+ {{- '\n' }}
+ {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+ {{- '<|im_end|>\n' }}
+ {%- endif %}
+ {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+ {{- '<|im_start|>assistant\n' }}
+{%- endif %}
diff --git a/qwen2.5-3b-instruct/checkpoint-142/chat_template.jinja b/qwen2.5-3b-instruct/checkpoint-142/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..bdf7919a96cfe43d50914a007b9c0877bd0ec27e
--- /dev/null
+++ b/qwen2.5-3b-instruct/checkpoint-142/chat_template.jinja
@@ -0,0 +1,54 @@
+{%- if tools %}
+ {{- '<|im_start|>system\n' }}
+ {%- if messages[0]['role'] == 'system' %}
+ {{- messages[0]['content'] }}
+ {%- else %}
+ {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+ {%- endif %}
+ {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within XML tags:\n" }}
+ {%- for tool in tools %}
+ {{- "\n" }}
+ {{- tool | tojson }}
+ {%- endfor %}
+ {{- "\n\n\nFor each function call, return a json object with function name and arguments within XML tags:\n\n{\"name\": , \"arguments\": }\n<|im_end|>\n" }}
+{%- else %}
+ {%- if messages[0]['role'] == 'system' %}
+ {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+ {%- else %}
+ {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+ {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+ {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+ {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+ {%- elif message.role == "assistant" %}
+ {{- '<|im_start|>' + message.role }}
+ {%- if message.content %}
+ {{- '\n' + message.content }}
+ {%- endif %}
+ {%- for tool_call in message.tool_calls %}
+ {%- if tool_call.function is defined %}
+ {%- set tool_call = tool_call.function %}
+ {%- endif %}
+ {{- '\n\n{"name": "' }}
+ {{- tool_call.name }}
+ {{- '", "arguments": ' }}
+ {{- tool_call.arguments | tojson }}
+ {{- '}\n' }}
+ {%- endfor %}
+ {{- '<|im_end|>\n' }}
+ {%- elif message.role == "tool" %}
+ {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+ {{- '<|im_start|>user' }}
+ {%- endif %}
+ {{- '\n\n' }}
+ {{- message.content }}
+ {{- '\n' }}
+ {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+ {{- '<|im_end|>\n' }}
+ {%- endif %}
+ {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+ {{- '<|im_start|>assistant\n' }}
+{%- endif %}
diff --git a/qwen2.5-3b-instruct/checkpoint-142/config.json b/qwen2.5-3b-instruct/checkpoint-142/config.json
new file mode 100644
index 0000000000000000000000000000000000000000..4814292863d1e5aa2f4fda1db18f4082cd61aa93
--- /dev/null
+++ b/qwen2.5-3b-instruct/checkpoint-142/config.json
@@ -0,0 +1,69 @@
+{
+ "architectures": [
+ "Qwen2ForCausalLM"
+ ],
+ "attention_dropout": 0.0,
+ "bos_token_id": null,
+ "dtype": "float32",
+ "eos_token_id": 151645,
+ "hidden_act": "silu",
+ "hidden_size": 2048,
+ "initializer_range": 0.02,
+ "intermediate_size": 11008,
+ "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",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention"
+ ],
+ "max_position_embeddings": 32768,
+ "max_window_layers": 70,
+ "model_type": "qwen2",
+ "num_attention_heads": 16,
+ "num_hidden_layers": 36,
+ "num_key_value_heads": 2,
+ "pad_token_id": 151643,
+ "rms_norm_eps": 1e-06,
+ "rope_parameters": {
+ "rope_theta": 1000000.0,
+ "rope_type": "default"
+ },
+ "sliding_window": null,
+ "tie_word_embeddings": true,
+ "transformers_version": "5.2.0",
+ "use_cache": false,
+ "use_sliding_window": false,
+ "vocab_size": 151936
+}
diff --git a/qwen2.5-3b-instruct/checkpoint-142/generation_config.json b/qwen2.5-3b-instruct/checkpoint-142/generation_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..47205fbf3386a215e7c2b9d62df6576f49b77e7b
--- /dev/null
+++ b/qwen2.5-3b-instruct/checkpoint-142/generation_config.json
@@ -0,0 +1,13 @@
+{
+ "do_sample": true,
+ "eos_token_id": [
+ 151645,
+ 151643
+ ],
+ "pad_token_id": 151643,
+ "repetition_penalty": 1.05,
+ "temperature": 0.7,
+ "top_k": 20,
+ "top_p": 0.8,
+ "transformers_version": "5.2.0"
+}
diff --git a/qwen2.5-3b-instruct/checkpoint-142/model.safetensors b/qwen2.5-3b-instruct/checkpoint-142/model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..d419ec4c7927d64dd5385507230e6a7294bab4b1
--- /dev/null
+++ b/qwen2.5-3b-instruct/checkpoint-142/model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fb28124271916a8f61e67cc847008a817d0c91ed5d7ff60ab3a28e35167059d1
+size 12343804296
diff --git a/qwen2.5-3b-instruct/checkpoint-142/optimizer.pt b/qwen2.5-3b-instruct/checkpoint-142/optimizer.pt
new file mode 100644
index 0000000000000000000000000000000000000000..14b7789a7b4b65fa7ab1a5f7b0cac94ef2f3cd3b
--- /dev/null
+++ b/qwen2.5-3b-instruct/checkpoint-142/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:37bc79e044eba5af801a7839ce87841678598ab3fbc20a690e7eaaa62b958952
+size 24687895753
diff --git a/qwen2.5-3b-instruct/checkpoint-142/rng_state_0.pth b/qwen2.5-3b-instruct/checkpoint-142/rng_state_0.pth
new file mode 100644
index 0000000000000000000000000000000000000000..870021e3fa5ac35c2f711adf0c93a556ab4842da
--- /dev/null
+++ b/qwen2.5-3b-instruct/checkpoint-142/rng_state_0.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:95e5fc2074c0df31522a514f862c86cb00d71c946a7f15cc9ec0e53a69fb28a7
+size 14917
diff --git a/qwen2.5-3b-instruct/checkpoint-142/rng_state_1.pth b/qwen2.5-3b-instruct/checkpoint-142/rng_state_1.pth
new file mode 100644
index 0000000000000000000000000000000000000000..21f20da5eb1da017f08aaa88bd19cf24d40e3fbf
--- /dev/null
+++ b/qwen2.5-3b-instruct/checkpoint-142/rng_state_1.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0e7153eae67b6c9232a41bc996a2bf5b83229b8c7230d61911ac0fd40e64154e
+size 14917
diff --git a/qwen2.5-3b-instruct/checkpoint-142/scheduler.pt b/qwen2.5-3b-instruct/checkpoint-142/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..2967e7243507a115f9424b1b386bf3657f047c75
--- /dev/null
+++ b/qwen2.5-3b-instruct/checkpoint-142/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:922deb71175797a96c4bddc5e1cef1c097df38c0dfa52334e97c89e0f573f2bb
+size 1465
diff --git a/qwen2.5-3b-instruct/checkpoint-142/tokenizer.json b/qwen2.5-3b-instruct/checkpoint-142/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..34510ff0037cd50428af467a17ead5a96140a32c
--- /dev/null
+++ b/qwen2.5-3b-instruct/checkpoint-142/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
+size 11421892
diff --git a/qwen2.5-3b-instruct/checkpoint-142/tokenizer_config.json b/qwen2.5-3b-instruct/checkpoint-142/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..145e2c7a68b109ae0253dda29fc2e66cb01b4b85
--- /dev/null
+++ b/qwen2.5-3b-instruct/checkpoint-142/tokenizer_config.json
@@ -0,0 +1,30 @@
+{
+ "add_prefix_space": false,
+ "backend": "tokenizers",
+ "bos_token": null,
+ "clean_up_tokenization_spaces": false,
+ "eos_token": "<|im_end|>",
+ "errors": "replace",
+ "extra_special_tokens": [
+ "<|im_start|>",
+ "<|im_end|>",
+ "<|object_ref_start|>",
+ "<|object_ref_end|>",
+ "<|box_start|>",
+ "<|box_end|>",
+ "<|quad_start|>",
+ "<|quad_end|>",
+ "<|vision_start|>",
+ "<|vision_end|>",
+ "<|vision_pad|>",
+ "<|image_pad|>",
+ "<|video_pad|>"
+ ],
+ "is_local": false,
+ "model_max_length": 131072,
+ "pad_token": "<|endoftext|>",
+ "padding_side": "right",
+ "split_special_tokens": false,
+ "tokenizer_class": "Qwen2Tokenizer",
+ "unk_token": null
+}
diff --git a/qwen2.5-3b-instruct/checkpoint-142/trainer_state.json b/qwen2.5-3b-instruct/checkpoint-142/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..18aeff067b2f4a5efe14ff0ec92e295d4ebd1a12
--- /dev/null
+++ b/qwen2.5-3b-instruct/checkpoint-142/trainer_state.json
@@ -0,0 +1,1044 @@
+{
+ "best_global_step": null,
+ "best_metric": null,
+ "best_model_checkpoint": null,
+ "epoch": 2.0,
+ "eval_steps": 50,
+ "global_step": 142,
+ "is_hyper_param_search": false,
+ "is_local_process_zero": true,
+ "is_world_process_zero": true,
+ "log_history": [
+ {
+ "epoch": 0.014222222222222223,
+ "grad_norm": 0.253393292427063,
+ "learning_rate": 0.0,
+ "loss": 1.19514799118042,
+ "step": 1
+ },
+ {
+ "epoch": 0.028444444444444446,
+ "grad_norm": 0.2720104455947876,
+ "learning_rate": 4.5454545454545457e-07,
+ "loss": 1.2398133277893066,
+ "step": 2
+ },
+ {
+ "epoch": 0.042666666666666665,
+ "grad_norm": 0.22159169614315033,
+ "learning_rate": 9.090909090909091e-07,
+ "loss": 0.9782462120056152,
+ "step": 3
+ },
+ {
+ "epoch": 0.05688888888888889,
+ "grad_norm": 0.17487271130084991,
+ "learning_rate": 1.3636363636363636e-06,
+ "loss": 0.8072947263717651,
+ "step": 4
+ },
+ {
+ "epoch": 0.07111111111111111,
+ "grad_norm": 0.08528934419155121,
+ "learning_rate": 1.8181818181818183e-06,
+ "loss": 0.5301008224487305,
+ "step": 5
+ },
+ {
+ "epoch": 0.08533333333333333,
+ "grad_norm": 0.08074624091386795,
+ "learning_rate": 2.2727272727272728e-06,
+ "loss": 0.6367583274841309,
+ "step": 6
+ },
+ {
+ "epoch": 0.09955555555555555,
+ "grad_norm": 0.05584918335080147,
+ "learning_rate": 2.7272727272727272e-06,
+ "loss": 0.6089775562286377,
+ "step": 7
+ },
+ {
+ "epoch": 0.11377777777777778,
+ "grad_norm": 0.032228983938694,
+ "learning_rate": 3.181818181818182e-06,
+ "loss": 0.5260579586029053,
+ "step": 8
+ },
+ {
+ "epoch": 0.128,
+ "grad_norm": 0.026230908930301666,
+ "learning_rate": 3.6363636363636366e-06,
+ "loss": 0.5446298122406006,
+ "step": 9
+ },
+ {
+ "epoch": 0.14222222222222222,
+ "grad_norm": 0.019648801535367966,
+ "learning_rate": 4.0909090909090915e-06,
+ "loss": 0.5338459610939026,
+ "step": 10
+ },
+ {
+ "epoch": 0.15644444444444444,
+ "grad_norm": 0.014007463119924068,
+ "learning_rate": 4.5454545454545455e-06,
+ "loss": 0.45045405626296997,
+ "step": 11
+ },
+ {
+ "epoch": 0.17066666666666666,
+ "grad_norm": 0.013295263983309269,
+ "learning_rate": 5e-06,
+ "loss": 0.49363279342651367,
+ "step": 12
+ },
+ {
+ "epoch": 0.18488888888888888,
+ "grad_norm": 0.01435786858201027,
+ "learning_rate": 5.4545454545454545e-06,
+ "loss": 0.48757249116897583,
+ "step": 13
+ },
+ {
+ "epoch": 0.1991111111111111,
+ "grad_norm": 0.015238163061439991,
+ "learning_rate": 5.90909090909091e-06,
+ "loss": 0.5002430081367493,
+ "step": 14
+ },
+ {
+ "epoch": 0.21333333333333335,
+ "grad_norm": 0.014418123289942741,
+ "learning_rate": 6.363636363636364e-06,
+ "loss": 0.4964277148246765,
+ "step": 15
+ },
+ {
+ "epoch": 0.22755555555555557,
+ "grad_norm": 0.014306087978184223,
+ "learning_rate": 6.818181818181818e-06,
+ "loss": 0.4818333387374878,
+ "step": 16
+ },
+ {
+ "epoch": 0.24177777777777779,
+ "grad_norm": 0.009899069555103779,
+ "learning_rate": 7.272727272727273e-06,
+ "loss": 0.4609573483467102,
+ "step": 17
+ },
+ {
+ "epoch": 0.256,
+ "grad_norm": 0.010643995366990566,
+ "learning_rate": 7.727272727272727e-06,
+ "loss": 0.47600606083869934,
+ "step": 18
+ },
+ {
+ "epoch": 0.2702222222222222,
+ "grad_norm": 0.012857547029852867,
+ "learning_rate": 8.181818181818183e-06,
+ "loss": 0.4962907135486603,
+ "step": 19
+ },
+ {
+ "epoch": 0.28444444444444444,
+ "grad_norm": 0.01670556701719761,
+ "learning_rate": 8.636363636363637e-06,
+ "loss": 0.5028363466262817,
+ "step": 20
+ },
+ {
+ "epoch": 0.2986666666666667,
+ "grad_norm": 0.010331347584724426,
+ "learning_rate": 9.090909090909091e-06,
+ "loss": 0.44582226872444153,
+ "step": 21
+ },
+ {
+ "epoch": 0.3128888888888889,
+ "grad_norm": 0.010525904595851898,
+ "learning_rate": 9.545454545454547e-06,
+ "loss": 0.4757377505302429,
+ "step": 22
+ },
+ {
+ "epoch": 0.32711111111111113,
+ "grad_norm": 0.009544270113110542,
+ "learning_rate": 1e-05,
+ "loss": 0.4707476496696472,
+ "step": 23
+ },
+ {
+ "epoch": 0.3413333333333333,
+ "grad_norm": 0.009730499237775803,
+ "learning_rate": 9.999323662872998e-06,
+ "loss": 0.49240708351135254,
+ "step": 24
+ },
+ {
+ "epoch": 0.35555555555555557,
+ "grad_norm": 0.008985932916402817,
+ "learning_rate": 9.99729483446475e-06,
+ "loss": 0.450992614030838,
+ "step": 25
+ },
+ {
+ "epoch": 0.36977777777777776,
+ "grad_norm": 0.01072725746780634,
+ "learning_rate": 9.993914063644053e-06,
+ "loss": 0.4514142870903015,
+ "step": 26
+ },
+ {
+ "epoch": 0.384,
+ "grad_norm": 0.01058228686451912,
+ "learning_rate": 9.989182265027232e-06,
+ "loss": 0.44021710753440857,
+ "step": 27
+ },
+ {
+ "epoch": 0.3982222222222222,
+ "grad_norm": 0.009333685040473938,
+ "learning_rate": 9.98310071873072e-06,
+ "loss": 0.4491778016090393,
+ "step": 28
+ },
+ {
+ "epoch": 0.41244444444444445,
+ "grad_norm": 0.009250632487237453,
+ "learning_rate": 9.975671070024741e-06,
+ "loss": 0.4285920560359955,
+ "step": 29
+ },
+ {
+ "epoch": 0.4266666666666667,
+ "grad_norm": 0.010210932232439518,
+ "learning_rate": 9.966895328888195e-06,
+ "loss": 0.5484102368354797,
+ "step": 30
+ },
+ {
+ "epoch": 0.4408888888888889,
+ "grad_norm": 0.008848125115036964,
+ "learning_rate": 9.956775869464901e-06,
+ "loss": 0.4242202043533325,
+ "step": 31
+ },
+ {
+ "epoch": 0.45511111111111113,
+ "grad_norm": 0.00905672088265419,
+ "learning_rate": 9.945315429421307e-06,
+ "loss": 0.4533390402793884,
+ "step": 32
+ },
+ {
+ "epoch": 0.4693333333333333,
+ "grad_norm": 0.008254914544522762,
+ "learning_rate": 9.932517109205849e-06,
+ "loss": 0.4243648052215576,
+ "step": 33
+ },
+ {
+ "epoch": 0.48355555555555557,
+ "grad_norm": 0.009192033670842648,
+ "learning_rate": 9.918384371210178e-06,
+ "loss": 0.5120309591293335,
+ "step": 34
+ },
+ {
+ "epoch": 0.49777777777777776,
+ "grad_norm": 0.009636987932026386,
+ "learning_rate": 9.902921038832456e-06,
+ "loss": 0.4871100187301636,
+ "step": 35
+ },
+ {
+ "epoch": 0.512,
+ "grad_norm": 0.008572430349886417,
+ "learning_rate": 9.886131295443003e-06,
+ "loss": 0.46277713775634766,
+ "step": 36
+ },
+ {
+ "epoch": 0.5262222222222223,
+ "grad_norm": 0.008587818592786789,
+ "learning_rate": 9.868019683252543e-06,
+ "loss": 0.45996755361557007,
+ "step": 37
+ },
+ {
+ "epoch": 0.5404444444444444,
+ "grad_norm": 0.010775095783174038,
+ "learning_rate": 9.848591102083375e-06,
+ "loss": 0.47378021478652954,
+ "step": 38
+ },
+ {
+ "epoch": 0.5546666666666666,
+ "grad_norm": 0.00995537731796503,
+ "learning_rate": 9.82785080804381e-06,
+ "loss": 0.4684975743293762,
+ "step": 39
+ },
+ {
+ "epoch": 0.5688888888888889,
+ "grad_norm": 0.009079818613827229,
+ "learning_rate": 9.805804412106197e-06,
+ "loss": 0.5784010887145996,
+ "step": 40
+ },
+ {
+ "epoch": 0.5831111111111111,
+ "grad_norm": 0.008139098063111305,
+ "learning_rate": 9.782457878588977e-06,
+ "loss": 0.46650874614715576,
+ "step": 41
+ },
+ {
+ "epoch": 0.5973333333333334,
+ "grad_norm": 0.007847482338547707,
+ "learning_rate": 9.75781752354311e-06,
+ "loss": 0.42804157733917236,
+ "step": 42
+ },
+ {
+ "epoch": 0.6115555555555555,
+ "grad_norm": 0.008446140214800835,
+ "learning_rate": 9.731890013043367e-06,
+ "loss": 0.48224180936813354,
+ "step": 43
+ },
+ {
+ "epoch": 0.6257777777777778,
+ "grad_norm": 0.008632696233689785,
+ "learning_rate": 9.704682361384941e-06,
+ "loss": 0.4561079144477844,
+ "step": 44
+ },
+ {
+ "epoch": 0.64,
+ "grad_norm": 0.008637293241918087,
+ "learning_rate": 9.676201929185809e-06,
+ "loss": 0.4007866382598877,
+ "step": 45
+ },
+ {
+ "epoch": 0.6542222222222223,
+ "grad_norm": 0.008761093020439148,
+ "learning_rate": 9.646456421395447e-06,
+ "loss": 0.4587845802307129,
+ "step": 46
+ },
+ {
+ "epoch": 0.6684444444444444,
+ "grad_norm": 0.008726399391889572,
+ "learning_rate": 9.615453885210368e-06,
+ "loss": 0.4571158289909363,
+ "step": 47
+ },
+ {
+ "epoch": 0.6826666666666666,
+ "grad_norm": 0.008282204158604145,
+ "learning_rate": 9.583202707897075e-06,
+ "loss": 0.4724605679512024,
+ "step": 48
+ },
+ {
+ "epoch": 0.6968888888888889,
+ "grad_norm": 0.008231760933995247,
+ "learning_rate": 9.549711614523007e-06,
+ "loss": 0.4600468575954437,
+ "step": 49
+ },
+ {
+ "epoch": 0.7111111111111111,
+ "grad_norm": 0.00751511100679636,
+ "learning_rate": 9.514989665596114e-06,
+ "loss": 0.4377005994319916,
+ "step": 50
+ },
+ {
+ "epoch": 0.7111111111111111,
+ "eval_loss": 0.4685295820236206,
+ "eval_runtime": 5.4716,
+ "eval_samples_per_second": 36.553,
+ "eval_steps_per_second": 18.276,
+ "step": 50
+ },
+ {
+ "epoch": 0.7253333333333334,
+ "grad_norm": 0.008261662907898426,
+ "learning_rate": 9.479046254613673e-06,
+ "loss": 0.4415626525878906,
+ "step": 51
+ },
+ {
+ "epoch": 0.7395555555555555,
+ "grad_norm": 0.007395230233669281,
+ "learning_rate": 9.441891105521005e-06,
+ "loss": 0.42561036348342896,
+ "step": 52
+ },
+ {
+ "epoch": 0.7537777777777778,
+ "grad_norm": 0.008451394736766815,
+ "learning_rate": 9.40353427008083e-06,
+ "loss": 0.4630648195743561,
+ "step": 53
+ },
+ {
+ "epoch": 0.768,
+ "grad_norm": 0.007475187536329031,
+ "learning_rate": 9.3639861251539e-06,
+ "loss": 0.43418699502944946,
+ "step": 54
+ },
+ {
+ "epoch": 0.7822222222222223,
+ "grad_norm": 0.00859637837857008,
+ "learning_rate": 9.323257369891702e-06,
+ "loss": 0.47063255310058594,
+ "step": 55
+ },
+ {
+ "epoch": 0.7964444444444444,
+ "grad_norm": 0.007597765419632196,
+ "learning_rate": 9.281359022841966e-06,
+ "loss": 0.41920918226242065,
+ "step": 56
+ },
+ {
+ "epoch": 0.8106666666666666,
+ "grad_norm": 0.018636908382177353,
+ "learning_rate": 9.238302418967757e-06,
+ "loss": 0.4757845401763916,
+ "step": 57
+ },
+ {
+ "epoch": 0.8248888888888889,
+ "grad_norm": 0.008423218503594398,
+ "learning_rate": 9.194099206580981e-06,
+ "loss": 0.5198032855987549,
+ "step": 58
+ },
+ {
+ "epoch": 0.8391111111111111,
+ "grad_norm": 0.008149159140884876,
+ "learning_rate": 9.14876134419111e-06,
+ "loss": 0.46035224199295044,
+ "step": 59
+ },
+ {
+ "epoch": 0.8533333333333334,
+ "grad_norm": 0.00819750688970089,
+ "learning_rate": 9.102301097269974e-06,
+ "loss": 0.4945712983608246,
+ "step": 60
+ },
+ {
+ "epoch": 0.8675555555555555,
+ "grad_norm": 0.00855066254734993,
+ "learning_rate": 9.05473103493355e-06,
+ "loss": 0.4513372778892517,
+ "step": 61
+ },
+ {
+ "epoch": 0.8817777777777778,
+ "grad_norm": 0.008660132065415382,
+ "learning_rate": 9.006064026541549e-06,
+ "loss": 0.49865084886550903,
+ "step": 62
+ },
+ {
+ "epoch": 0.896,
+ "grad_norm": 0.007741494104266167,
+ "learning_rate": 8.956313238215824e-06,
+ "loss": 0.4319220185279846,
+ "step": 63
+ },
+ {
+ "epoch": 0.9102222222222223,
+ "grad_norm": 0.007900395430624485,
+ "learning_rate": 8.905492129278478e-06,
+ "loss": 0.43760013580322266,
+ "step": 64
+ },
+ {
+ "epoch": 0.9244444444444444,
+ "grad_norm": 0.007346940692514181,
+ "learning_rate": 8.85361444861063e-06,
+ "loss": 0.41593095660209656,
+ "step": 65
+ },
+ {
+ "epoch": 0.9386666666666666,
+ "grad_norm": 0.00902250874787569,
+ "learning_rate": 8.800694230932885e-06,
+ "loss": 0.48464590311050415,
+ "step": 66
+ },
+ {
+ "epoch": 0.9528888888888889,
+ "grad_norm": 0.007672084029763937,
+ "learning_rate": 8.74674579300843e-06,
+ "loss": 0.41884416341781616,
+ "step": 67
+ },
+ {
+ "epoch": 0.9671111111111111,
+ "grad_norm": 0.008009645156562328,
+ "learning_rate": 8.691783729769874e-06,
+ "loss": 0.40650737285614014,
+ "step": 68
+ },
+ {
+ "epoch": 0.9813333333333333,
+ "grad_norm": 0.008077424019575119,
+ "learning_rate": 8.635822910370793e-06,
+ "loss": 0.4726918637752533,
+ "step": 69
+ },
+ {
+ "epoch": 0.9955555555555555,
+ "grad_norm": 0.007576521951705217,
+ "learning_rate": 8.578878474163115e-06,
+ "loss": 0.4404842257499695,
+ "step": 70
+ },
+ {
+ "epoch": 1.0,
+ "grad_norm": 0.012638773769140244,
+ "learning_rate": 8.520965826601394e-06,
+ "loss": 0.42633137106895447,
+ "step": 71
+ },
+ {
+ "epoch": 1.0142222222222221,
+ "grad_norm": 0.008662556298077106,
+ "learning_rate": 8.462100635075097e-06,
+ "loss": 0.46000930666923523,
+ "step": 72
+ },
+ {
+ "epoch": 1.0284444444444445,
+ "grad_norm": 0.008270082995295525,
+ "learning_rate": 8.40229882467003e-06,
+ "loss": 0.42150819301605225,
+ "step": 73
+ },
+ {
+ "epoch": 1.0426666666666666,
+ "grad_norm": 0.007625850848853588,
+ "learning_rate": 8.341576573860049e-06,
+ "loss": 0.37313175201416016,
+ "step": 74
+ },
+ {
+ "epoch": 1.056888888888889,
+ "grad_norm": 0.00790152233093977,
+ "learning_rate": 8.279950310130218e-06,
+ "loss": 0.382789671421051,
+ "step": 75
+ },
+ {
+ "epoch": 1.0711111111111111,
+ "grad_norm": 0.007986027747392654,
+ "learning_rate": 8.2174367055326e-06,
+ "loss": 0.3734534978866577,
+ "step": 76
+ },
+ {
+ "epoch": 1.0853333333333333,
+ "grad_norm": 0.007174879312515259,
+ "learning_rate": 8.154052672175888e-06,
+ "loss": 0.37693876028060913,
+ "step": 77
+ },
+ {
+ "epoch": 1.0995555555555556,
+ "grad_norm": 0.008486706763505936,
+ "learning_rate": 8.08981535765009e-06,
+ "loss": 0.4265296459197998,
+ "step": 78
+ },
+ {
+ "epoch": 1.1137777777777778,
+ "grad_norm": 0.007609546184539795,
+ "learning_rate": 8.024742140387506e-06,
+ "loss": 0.3812956213951111,
+ "step": 79
+ },
+ {
+ "epoch": 1.1280000000000001,
+ "grad_norm": 0.007912754081189632,
+ "learning_rate": 7.95885062496126e-06,
+ "loss": 0.4131423234939575,
+ "step": 80
+ },
+ {
+ "epoch": 1.1422222222222222,
+ "grad_norm": 0.007641648408025503,
+ "learning_rate": 7.892158637322647e-06,
+ "loss": 0.37861311435699463,
+ "step": 81
+ },
+ {
+ "epoch": 1.1564444444444444,
+ "grad_norm": 0.007570190355181694,
+ "learning_rate": 7.824684219978591e-06,
+ "loss": 0.3589654564857483,
+ "step": 82
+ },
+ {
+ "epoch": 1.1706666666666667,
+ "grad_norm": 0.007670491933822632,
+ "learning_rate": 7.756445627110523e-06,
+ "loss": 0.40794360637664795,
+ "step": 83
+ },
+ {
+ "epoch": 1.1848888888888889,
+ "grad_norm": 0.009116973727941513,
+ "learning_rate": 7.68746131963598e-06,
+ "loss": 0.4278006851673126,
+ "step": 84
+ },
+ {
+ "epoch": 1.199111111111111,
+ "grad_norm": 0.007852817885577679,
+ "learning_rate": 7.6177499602143e-06,
+ "loss": 0.371395081281662,
+ "step": 85
+ },
+ {
+ "epoch": 1.2133333333333334,
+ "grad_norm": 0.00870604906231165,
+ "learning_rate": 7.547330408197695e-06,
+ "loss": 0.4305034279823303,
+ "step": 86
+ },
+ {
+ "epoch": 1.2275555555555555,
+ "grad_norm": 0.008404702879488468,
+ "learning_rate": 7.476221714529167e-06,
+ "loss": 0.4348181188106537,
+ "step": 87
+ },
+ {
+ "epoch": 1.2417777777777779,
+ "grad_norm": 0.008913768455386162,
+ "learning_rate": 7.404443116588548e-06,
+ "loss": 0.4309384822845459,
+ "step": 88
+ },
+ {
+ "epoch": 1.256,
+ "grad_norm": 0.007682010997086763,
+ "learning_rate": 7.332014032988123e-06,
+ "loss": 0.4107038676738739,
+ "step": 89
+ },
+ {
+ "epoch": 1.2702222222222221,
+ "grad_norm": 0.00787457637488842,
+ "learning_rate": 7.2589540583192165e-06,
+ "loss": 0.37457963824272156,
+ "step": 90
+ },
+ {
+ "epoch": 1.2844444444444445,
+ "grad_norm": 0.007768277078866959,
+ "learning_rate": 7.185282957851175e-06,
+ "loss": 0.37697723507881165,
+ "step": 91
+ },
+ {
+ "epoch": 1.2986666666666666,
+ "grad_norm": 0.008521760813891888,
+ "learning_rate": 7.111020662184174e-06,
+ "loss": 0.40353477001190186,
+ "step": 92
+ },
+ {
+ "epoch": 1.3128888888888888,
+ "grad_norm": 0.0077499281615018845,
+ "learning_rate": 7.036187261857289e-06,
+ "loss": 0.37755411863327026,
+ "step": 93
+ },
+ {
+ "epoch": 1.3271111111111111,
+ "grad_norm": 0.00870734266936779,
+ "learning_rate": 6.960803001913315e-06,
+ "loss": 0.3729894757270813,
+ "step": 94
+ },
+ {
+ "epoch": 1.3413333333333333,
+ "grad_norm": 0.007563227321952581,
+ "learning_rate": 6.884888276421766e-06,
+ "loss": 0.4091319441795349,
+ "step": 95
+ },
+ {
+ "epoch": 1.3555555555555556,
+ "grad_norm": 0.00803642813116312,
+ "learning_rate": 6.8084636229615786e-06,
+ "loss": 0.4048840403556824,
+ "step": 96
+ },
+ {
+ "epoch": 1.3697777777777778,
+ "grad_norm": 0.00779891200363636,
+ "learning_rate": 6.731549717064975e-06,
+ "loss": 0.39409685134887695,
+ "step": 97
+ },
+ {
+ "epoch": 1.384,
+ "grad_norm": 0.008260682225227356,
+ "learning_rate": 6.654167366624009e-06,
+ "loss": 0.36237582564353943,
+ "step": 98
+ },
+ {
+ "epoch": 1.3982222222222223,
+ "grad_norm": 0.008373094722628593,
+ "learning_rate": 6.576337506261314e-06,
+ "loss": 0.38763654232025146,
+ "step": 99
+ },
+ {
+ "epoch": 1.4124444444444444,
+ "grad_norm": 0.007797387894243002,
+ "learning_rate": 6.498081191666549e-06,
+ "loss": 0.41751331090927124,
+ "step": 100
+ },
+ {
+ "epoch": 1.4124444444444444,
+ "eval_loss": 0.4669477939605713,
+ "eval_runtime": 5.2715,
+ "eval_samples_per_second": 37.94,
+ "eval_steps_per_second": 18.97,
+ "step": 100
+ },
+ {
+ "epoch": 1.4266666666666667,
+ "grad_norm": 0.007338334806263447,
+ "learning_rate": 6.419419593900109e-06,
+ "loss": 0.38504236936569214,
+ "step": 101
+ },
+ {
+ "epoch": 1.4408888888888889,
+ "grad_norm": 0.007528901565819979,
+ "learning_rate": 6.340373993665607e-06,
+ "loss": 0.36031031608581543,
+ "step": 102
+ },
+ {
+ "epoch": 1.455111111111111,
+ "grad_norm": 0.007573080714792013,
+ "learning_rate": 6.2609657755527135e-06,
+ "loss": 0.3863949179649353,
+ "step": 103
+ },
+ {
+ "epoch": 1.4693333333333334,
+ "grad_norm": 0.008099780417978764,
+ "learning_rate": 6.1812164222518626e-06,
+ "loss": 0.3955462574958801,
+ "step": 104
+ },
+ {
+ "epoch": 1.4835555555555555,
+ "grad_norm": 0.008717137388885021,
+ "learning_rate": 6.101147508742456e-06,
+ "loss": 0.39310169219970703,
+ "step": 105
+ },
+ {
+ "epoch": 1.4977777777777779,
+ "grad_norm": 0.008383029140532017,
+ "learning_rate": 6.020780696456059e-06,
+ "loss": 0.4533539414405823,
+ "step": 106
+ },
+ {
+ "epoch": 1.512,
+ "grad_norm": 0.007702656090259552,
+ "learning_rate": 5.940137727416247e-06,
+ "loss": 0.3985057473182678,
+ "step": 107
+ },
+ {
+ "epoch": 1.5262222222222221,
+ "grad_norm": 0.00808175653219223,
+ "learning_rate": 5.859240418356614e-06,
+ "loss": 0.3878968358039856,
+ "step": 108
+ },
+ {
+ "epoch": 1.5404444444444443,
+ "grad_norm": 0.008781371638178825,
+ "learning_rate": 5.778110654818602e-06,
+ "loss": 0.39648741483688354,
+ "step": 109
+ },
+ {
+ "epoch": 1.5546666666666666,
+ "grad_norm": 0.008896375074982643,
+ "learning_rate": 5.696770385230679e-06,
+ "loss": 0.4319821000099182,
+ "step": 110
+ },
+ {
+ "epoch": 1.568888888888889,
+ "grad_norm": 0.008202116936445236,
+ "learning_rate": 5.615241614970546e-06,
+ "loss": 0.4078192710876465,
+ "step": 111
+ },
+ {
+ "epoch": 1.5831111111111111,
+ "grad_norm": 0.007843846455216408,
+ "learning_rate": 5.533546400411899e-06,
+ "loss": 0.38504159450531006,
+ "step": 112
+ },
+ {
+ "epoch": 1.5973333333333333,
+ "grad_norm": 0.00823921337723732,
+ "learning_rate": 5.4517068429574215e-06,
+ "loss": 0.3971819579601288,
+ "step": 113
+ },
+ {
+ "epoch": 1.6115555555555554,
+ "grad_norm": 0.007411513011902571,
+ "learning_rate": 5.3697450830595775e-06,
+ "loss": 0.3886791169643402,
+ "step": 114
+ },
+ {
+ "epoch": 1.6257777777777778,
+ "grad_norm": 0.007902945391833782,
+ "learning_rate": 5.287683294230855e-06,
+ "loss": 0.37610527873039246,
+ "step": 115
+ },
+ {
+ "epoch": 1.6400000000000001,
+ "grad_norm": 0.007778776343911886,
+ "learning_rate": 5.20554367704505e-06,
+ "loss": 0.3846814036369324,
+ "step": 116
+ },
+ {
+ "epoch": 1.6542222222222223,
+ "grad_norm": 0.008897176012396812,
+ "learning_rate": 5.123348453131242e-06,
+ "loss": 0.3917155861854553,
+ "step": 117
+ },
+ {
+ "epoch": 1.6684444444444444,
+ "grad_norm": 0.008548559620976448,
+ "learning_rate": 5.041119859162068e-06,
+ "loss": 0.39533206820487976,
+ "step": 118
+ },
+ {
+ "epoch": 1.6826666666666665,
+ "grad_norm": 0.008137239143252373,
+ "learning_rate": 4.958880140837934e-06,
+ "loss": 0.4113273620605469,
+ "step": 119
+ },
+ {
+ "epoch": 1.696888888888889,
+ "grad_norm": 0.008478965610265732,
+ "learning_rate": 4.876651546868759e-06,
+ "loss": 0.39543214440345764,
+ "step": 120
+ },
+ {
+ "epoch": 1.7111111111111112,
+ "grad_norm": 0.00835413858294487,
+ "learning_rate": 4.794456322954953e-06,
+ "loss": 0.4548013508319855,
+ "step": 121
+ },
+ {
+ "epoch": 1.7253333333333334,
+ "grad_norm": 0.00832394976168871,
+ "learning_rate": 4.7123167057691446e-06,
+ "loss": 0.4111294150352478,
+ "step": 122
+ },
+ {
+ "epoch": 1.7395555555555555,
+ "grad_norm": 0.008472571149468422,
+ "learning_rate": 4.630254916940424e-06,
+ "loss": 0.43050098419189453,
+ "step": 123
+ },
+ {
+ "epoch": 1.7537777777777777,
+ "grad_norm": 0.00816201139241457,
+ "learning_rate": 4.548293157042581e-06,
+ "loss": 0.41356489062309265,
+ "step": 124
+ },
+ {
+ "epoch": 1.768,
+ "grad_norm": 0.008180899545550346,
+ "learning_rate": 4.466453599588103e-06,
+ "loss": 0.41799914836883545,
+ "step": 125
+ },
+ {
+ "epoch": 1.7822222222222224,
+ "grad_norm": 0.007402091287076473,
+ "learning_rate": 4.384758385029457e-06,
+ "loss": 0.3510154187679291,
+ "step": 126
+ },
+ {
+ "epoch": 1.7964444444444445,
+ "grad_norm": 0.008476709015667439,
+ "learning_rate": 4.3032296147693225e-06,
+ "loss": 0.4098949432373047,
+ "step": 127
+ },
+ {
+ "epoch": 1.8106666666666666,
+ "grad_norm": 0.008135325275361538,
+ "learning_rate": 4.2218893451814005e-06,
+ "loss": 0.36957240104675293,
+ "step": 128
+ },
+ {
+ "epoch": 1.8248888888888888,
+ "grad_norm": 0.007964376360177994,
+ "learning_rate": 4.140759581643386e-06,
+ "loss": 0.3535369038581848,
+ "step": 129
+ },
+ {
+ "epoch": 1.8391111111111111,
+ "grad_norm": 0.00742813665419817,
+ "learning_rate": 4.059862272583755e-06,
+ "loss": 0.38296306133270264,
+ "step": 130
+ },
+ {
+ "epoch": 1.8533333333333335,
+ "grad_norm": 0.007901367731392384,
+ "learning_rate": 3.979219303543942e-06,
+ "loss": 0.38752996921539307,
+ "step": 131
+ },
+ {
+ "epoch": 1.8675555555555556,
+ "grad_norm": 0.008205283433198929,
+ "learning_rate": 3.898852491257547e-06,
+ "loss": 0.3867356479167938,
+ "step": 132
+ },
+ {
+ "epoch": 1.8817777777777778,
+ "grad_norm": 0.008251599036157131,
+ "learning_rate": 3.818783577748138e-06,
+ "loss": 0.42934614419937134,
+ "step": 133
+ },
+ {
+ "epoch": 1.896,
+ "grad_norm": 0.008710331283509731,
+ "learning_rate": 3.7390342244472886e-06,
+ "loss": 0.4096015691757202,
+ "step": 134
+ },
+ {
+ "epoch": 1.9102222222222223,
+ "grad_norm": 0.008127489127218723,
+ "learning_rate": 3.659626006334395e-06,
+ "loss": 0.3849722743034363,
+ "step": 135
+ },
+ {
+ "epoch": 1.9244444444444444,
+ "grad_norm": 0.008900986984372139,
+ "learning_rate": 3.5805804060998926e-06,
+ "loss": 0.415260910987854,
+ "step": 136
+ },
+ {
+ "epoch": 1.9386666666666668,
+ "grad_norm": 0.00907993782311678,
+ "learning_rate": 3.501918808333453e-06,
+ "loss": 0.48382842540740967,
+ "step": 137
+ },
+ {
+ "epoch": 1.952888888888889,
+ "grad_norm": 0.008263039402663708,
+ "learning_rate": 3.4236624937386874e-06,
+ "loss": 0.4434107542037964,
+ "step": 138
+ },
+ {
+ "epoch": 1.967111111111111,
+ "grad_norm": 0.008353665471076965,
+ "learning_rate": 3.3458326333759927e-06,
+ "loss": 0.41760051250457764,
+ "step": 139
+ },
+ {
+ "epoch": 1.9813333333333332,
+ "grad_norm": 0.008556668646633625,
+ "learning_rate": 3.268450282935026e-06,
+ "loss": 0.43642479181289673,
+ "step": 140
+ },
+ {
+ "epoch": 1.9955555555555555,
+ "grad_norm": 0.008040979504585266,
+ "learning_rate": 3.1915363770384223e-06,
+ "loss": 0.38327687978744507,
+ "step": 141
+ },
+ {
+ "epoch": 2.0,
+ "grad_norm": 0.012571632862091064,
+ "learning_rate": 3.1151117235782346e-06,
+ "loss": 0.3751964867115021,
+ "step": 142
+ }
+ ],
+ "logging_steps": 1,
+ "max_steps": 213,
+ "num_input_tokens_seen": 0,
+ "num_train_epochs": 3,
+ "save_steps": 500,
+ "stateful_callbacks": {
+ "TrainerControl": {
+ "args": {
+ "should_epoch_stop": false,
+ "should_evaluate": false,
+ "should_log": false,
+ "should_save": true,
+ "should_training_stop": false
+ },
+ "attributes": {}
+ }
+ },
+ "total_flos": 6.872319897947341e+16,
+ "train_batch_size": 1,
+ "trial_name": null,
+ "trial_params": null
+}
diff --git a/qwen2.5-3b-instruct/checkpoint-142/training_args.bin b/qwen2.5-3b-instruct/checkpoint-142/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..bd95d1f874b25af121b3c8254bd05a3eb95ffd5f
--- /dev/null
+++ b/qwen2.5-3b-instruct/checkpoint-142/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c4f1847ec07466eeb896ef40e20d9241e19882e62efb61a14f1e7c6a96737fa0
+size 5649
diff --git a/qwen2.5-3b-instruct/checkpoint-213/chat_template.jinja b/qwen2.5-3b-instruct/checkpoint-213/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..bdf7919a96cfe43d50914a007b9c0877bd0ec27e
--- /dev/null
+++ b/qwen2.5-3b-instruct/checkpoint-213/chat_template.jinja
@@ -0,0 +1,54 @@
+{%- if tools %}
+ {{- '<|im_start|>system\n' }}
+ {%- if messages[0]['role'] == 'system' %}
+ {{- messages[0]['content'] }}
+ {%- else %}
+ {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+ {%- endif %}
+ {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within XML tags:\n" }}
+ {%- for tool in tools %}
+ {{- "\n" }}
+ {{- tool | tojson }}
+ {%- endfor %}
+ {{- "\n\n\nFor each function call, return a json object with function name and arguments within XML tags:\n\n{\"name\": , \"arguments\": }\n<|im_end|>\n" }}
+{%- else %}
+ {%- if messages[0]['role'] == 'system' %}
+ {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+ {%- else %}
+ {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+ {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+ {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+ {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+ {%- elif message.role == "assistant" %}
+ {{- '<|im_start|>' + message.role }}
+ {%- if message.content %}
+ {{- '\n' + message.content }}
+ {%- endif %}
+ {%- for tool_call in message.tool_calls %}
+ {%- if tool_call.function is defined %}
+ {%- set tool_call = tool_call.function %}
+ {%- endif %}
+ {{- '\n\n{"name": "' }}
+ {{- tool_call.name }}
+ {{- '", "arguments": ' }}
+ {{- tool_call.arguments | tojson }}
+ {{- '}\n' }}
+ {%- endfor %}
+ {{- '<|im_end|>\n' }}
+ {%- elif message.role == "tool" %}
+ {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+ {{- '<|im_start|>user' }}
+ {%- endif %}
+ {{- '\n\n' }}
+ {{- message.content }}
+ {{- '\n' }}
+ {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+ {{- '<|im_end|>\n' }}
+ {%- endif %}
+ {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+ {{- '<|im_start|>assistant\n' }}
+{%- endif %}
diff --git a/qwen2.5-3b-instruct/checkpoint-213/config.json b/qwen2.5-3b-instruct/checkpoint-213/config.json
new file mode 100644
index 0000000000000000000000000000000000000000..4814292863d1e5aa2f4fda1db18f4082cd61aa93
--- /dev/null
+++ b/qwen2.5-3b-instruct/checkpoint-213/config.json
@@ -0,0 +1,69 @@
+{
+ "architectures": [
+ "Qwen2ForCausalLM"
+ ],
+ "attention_dropout": 0.0,
+ "bos_token_id": null,
+ "dtype": "float32",
+ "eos_token_id": 151645,
+ "hidden_act": "silu",
+ "hidden_size": 2048,
+ "initializer_range": 0.02,
+ "intermediate_size": 11008,
+ "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",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention"
+ ],
+ "max_position_embeddings": 32768,
+ "max_window_layers": 70,
+ "model_type": "qwen2",
+ "num_attention_heads": 16,
+ "num_hidden_layers": 36,
+ "num_key_value_heads": 2,
+ "pad_token_id": 151643,
+ "rms_norm_eps": 1e-06,
+ "rope_parameters": {
+ "rope_theta": 1000000.0,
+ "rope_type": "default"
+ },
+ "sliding_window": null,
+ "tie_word_embeddings": true,
+ "transformers_version": "5.2.0",
+ "use_cache": false,
+ "use_sliding_window": false,
+ "vocab_size": 151936
+}
diff --git a/qwen2.5-3b-instruct/checkpoint-213/generation_config.json b/qwen2.5-3b-instruct/checkpoint-213/generation_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..47205fbf3386a215e7c2b9d62df6576f49b77e7b
--- /dev/null
+++ b/qwen2.5-3b-instruct/checkpoint-213/generation_config.json
@@ -0,0 +1,13 @@
+{
+ "do_sample": true,
+ "eos_token_id": [
+ 151645,
+ 151643
+ ],
+ "pad_token_id": 151643,
+ "repetition_penalty": 1.05,
+ "temperature": 0.7,
+ "top_k": 20,
+ "top_p": 0.8,
+ "transformers_version": "5.2.0"
+}
diff --git a/qwen2.5-3b-instruct/checkpoint-213/model.safetensors b/qwen2.5-3b-instruct/checkpoint-213/model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..49702ec9826e3cd7c26c2b828a2fab8d09ce50c2
--- /dev/null
+++ b/qwen2.5-3b-instruct/checkpoint-213/model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:19b1d375c807d8900f5c4a64fceb62f4bd9a2bf032bd23ad921474f05998e666
+size 12343804296
diff --git a/qwen2.5-3b-instruct/checkpoint-213/optimizer.pt b/qwen2.5-3b-instruct/checkpoint-213/optimizer.pt
new file mode 100644
index 0000000000000000000000000000000000000000..a2886551acd89d167eaa0e0cf6a66e7ca55e1d29
--- /dev/null
+++ b/qwen2.5-3b-instruct/checkpoint-213/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7cc3519c3a0eb0528eac1418b4111030795fe60c38b6f0ece577dd75bea3893d
+size 24687895753
diff --git a/qwen2.5-3b-instruct/checkpoint-213/rng_state_0.pth b/qwen2.5-3b-instruct/checkpoint-213/rng_state_0.pth
new file mode 100644
index 0000000000000000000000000000000000000000..870021e3fa5ac35c2f711adf0c93a556ab4842da
--- /dev/null
+++ b/qwen2.5-3b-instruct/checkpoint-213/rng_state_0.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:95e5fc2074c0df31522a514f862c86cb00d71c946a7f15cc9ec0e53a69fb28a7
+size 14917
diff --git a/qwen2.5-3b-instruct/checkpoint-213/rng_state_1.pth b/qwen2.5-3b-instruct/checkpoint-213/rng_state_1.pth
new file mode 100644
index 0000000000000000000000000000000000000000..21f20da5eb1da017f08aaa88bd19cf24d40e3fbf
--- /dev/null
+++ b/qwen2.5-3b-instruct/checkpoint-213/rng_state_1.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0e7153eae67b6c9232a41bc996a2bf5b83229b8c7230d61911ac0fd40e64154e
+size 14917
diff --git a/qwen2.5-3b-instruct/checkpoint-213/scheduler.pt b/qwen2.5-3b-instruct/checkpoint-213/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..39cf87775c3a71d10e7311e0cab769e112f357ba
--- /dev/null
+++ b/qwen2.5-3b-instruct/checkpoint-213/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c38a67b98c17be81d4c6d00b60029cdedd43e9aba35ca28ec28ef946c400f268
+size 1465
diff --git a/qwen2.5-3b-instruct/checkpoint-213/tokenizer.json b/qwen2.5-3b-instruct/checkpoint-213/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..34510ff0037cd50428af467a17ead5a96140a32c
--- /dev/null
+++ b/qwen2.5-3b-instruct/checkpoint-213/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
+size 11421892
diff --git a/qwen2.5-3b-instruct/checkpoint-213/tokenizer_config.json b/qwen2.5-3b-instruct/checkpoint-213/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..145e2c7a68b109ae0253dda29fc2e66cb01b4b85
--- /dev/null
+++ b/qwen2.5-3b-instruct/checkpoint-213/tokenizer_config.json
@@ -0,0 +1,30 @@
+{
+ "add_prefix_space": false,
+ "backend": "tokenizers",
+ "bos_token": null,
+ "clean_up_tokenization_spaces": false,
+ "eos_token": "<|im_end|>",
+ "errors": "replace",
+ "extra_special_tokens": [
+ "<|im_start|>",
+ "<|im_end|>",
+ "<|object_ref_start|>",
+ "<|object_ref_end|>",
+ "<|box_start|>",
+ "<|box_end|>",
+ "<|quad_start|>",
+ "<|quad_end|>",
+ "<|vision_start|>",
+ "<|vision_end|>",
+ "<|vision_pad|>",
+ "<|image_pad|>",
+ "<|video_pad|>"
+ ],
+ "is_local": false,
+ "model_max_length": 131072,
+ "pad_token": "<|endoftext|>",
+ "padding_side": "right",
+ "split_special_tokens": false,
+ "tokenizer_class": "Qwen2Tokenizer",
+ "unk_token": null
+}
diff --git a/qwen2.5-3b-instruct/checkpoint-213/trainer_state.json b/qwen2.5-3b-instruct/checkpoint-213/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..70c08b83401f592bdda84d11df52e26dcb06a24c
--- /dev/null
+++ b/qwen2.5-3b-instruct/checkpoint-213/trainer_state.json
@@ -0,0 +1,1557 @@
+{
+ "best_global_step": null,
+ "best_metric": null,
+ "best_model_checkpoint": null,
+ "epoch": 3.0,
+ "eval_steps": 50,
+ "global_step": 213,
+ "is_hyper_param_search": false,
+ "is_local_process_zero": true,
+ "is_world_process_zero": true,
+ "log_history": [
+ {
+ "epoch": 0.014222222222222223,
+ "grad_norm": 0.253393292427063,
+ "learning_rate": 0.0,
+ "loss": 1.19514799118042,
+ "step": 1
+ },
+ {
+ "epoch": 0.028444444444444446,
+ "grad_norm": 0.2720104455947876,
+ "learning_rate": 4.5454545454545457e-07,
+ "loss": 1.2398133277893066,
+ "step": 2
+ },
+ {
+ "epoch": 0.042666666666666665,
+ "grad_norm": 0.22159169614315033,
+ "learning_rate": 9.090909090909091e-07,
+ "loss": 0.9782462120056152,
+ "step": 3
+ },
+ {
+ "epoch": 0.05688888888888889,
+ "grad_norm": 0.17487271130084991,
+ "learning_rate": 1.3636363636363636e-06,
+ "loss": 0.8072947263717651,
+ "step": 4
+ },
+ {
+ "epoch": 0.07111111111111111,
+ "grad_norm": 0.08528934419155121,
+ "learning_rate": 1.8181818181818183e-06,
+ "loss": 0.5301008224487305,
+ "step": 5
+ },
+ {
+ "epoch": 0.08533333333333333,
+ "grad_norm": 0.08074624091386795,
+ "learning_rate": 2.2727272727272728e-06,
+ "loss": 0.6367583274841309,
+ "step": 6
+ },
+ {
+ "epoch": 0.09955555555555555,
+ "grad_norm": 0.05584918335080147,
+ "learning_rate": 2.7272727272727272e-06,
+ "loss": 0.6089775562286377,
+ "step": 7
+ },
+ {
+ "epoch": 0.11377777777777778,
+ "grad_norm": 0.032228983938694,
+ "learning_rate": 3.181818181818182e-06,
+ "loss": 0.5260579586029053,
+ "step": 8
+ },
+ {
+ "epoch": 0.128,
+ "grad_norm": 0.026230908930301666,
+ "learning_rate": 3.6363636363636366e-06,
+ "loss": 0.5446298122406006,
+ "step": 9
+ },
+ {
+ "epoch": 0.14222222222222222,
+ "grad_norm": 0.019648801535367966,
+ "learning_rate": 4.0909090909090915e-06,
+ "loss": 0.5338459610939026,
+ "step": 10
+ },
+ {
+ "epoch": 0.15644444444444444,
+ "grad_norm": 0.014007463119924068,
+ "learning_rate": 4.5454545454545455e-06,
+ "loss": 0.45045405626296997,
+ "step": 11
+ },
+ {
+ "epoch": 0.17066666666666666,
+ "grad_norm": 0.013295263983309269,
+ "learning_rate": 5e-06,
+ "loss": 0.49363279342651367,
+ "step": 12
+ },
+ {
+ "epoch": 0.18488888888888888,
+ "grad_norm": 0.01435786858201027,
+ "learning_rate": 5.4545454545454545e-06,
+ "loss": 0.48757249116897583,
+ "step": 13
+ },
+ {
+ "epoch": 0.1991111111111111,
+ "grad_norm": 0.015238163061439991,
+ "learning_rate": 5.90909090909091e-06,
+ "loss": 0.5002430081367493,
+ "step": 14
+ },
+ {
+ "epoch": 0.21333333333333335,
+ "grad_norm": 0.014418123289942741,
+ "learning_rate": 6.363636363636364e-06,
+ "loss": 0.4964277148246765,
+ "step": 15
+ },
+ {
+ "epoch": 0.22755555555555557,
+ "grad_norm": 0.014306087978184223,
+ "learning_rate": 6.818181818181818e-06,
+ "loss": 0.4818333387374878,
+ "step": 16
+ },
+ {
+ "epoch": 0.24177777777777779,
+ "grad_norm": 0.009899069555103779,
+ "learning_rate": 7.272727272727273e-06,
+ "loss": 0.4609573483467102,
+ "step": 17
+ },
+ {
+ "epoch": 0.256,
+ "grad_norm": 0.010643995366990566,
+ "learning_rate": 7.727272727272727e-06,
+ "loss": 0.47600606083869934,
+ "step": 18
+ },
+ {
+ "epoch": 0.2702222222222222,
+ "grad_norm": 0.012857547029852867,
+ "learning_rate": 8.181818181818183e-06,
+ "loss": 0.4962907135486603,
+ "step": 19
+ },
+ {
+ "epoch": 0.28444444444444444,
+ "grad_norm": 0.01670556701719761,
+ "learning_rate": 8.636363636363637e-06,
+ "loss": 0.5028363466262817,
+ "step": 20
+ },
+ {
+ "epoch": 0.2986666666666667,
+ "grad_norm": 0.010331347584724426,
+ "learning_rate": 9.090909090909091e-06,
+ "loss": 0.44582226872444153,
+ "step": 21
+ },
+ {
+ "epoch": 0.3128888888888889,
+ "grad_norm": 0.010525904595851898,
+ "learning_rate": 9.545454545454547e-06,
+ "loss": 0.4757377505302429,
+ "step": 22
+ },
+ {
+ "epoch": 0.32711111111111113,
+ "grad_norm": 0.009544270113110542,
+ "learning_rate": 1e-05,
+ "loss": 0.4707476496696472,
+ "step": 23
+ },
+ {
+ "epoch": 0.3413333333333333,
+ "grad_norm": 0.009730499237775803,
+ "learning_rate": 9.999323662872998e-06,
+ "loss": 0.49240708351135254,
+ "step": 24
+ },
+ {
+ "epoch": 0.35555555555555557,
+ "grad_norm": 0.008985932916402817,
+ "learning_rate": 9.99729483446475e-06,
+ "loss": 0.450992614030838,
+ "step": 25
+ },
+ {
+ "epoch": 0.36977777777777776,
+ "grad_norm": 0.01072725746780634,
+ "learning_rate": 9.993914063644053e-06,
+ "loss": 0.4514142870903015,
+ "step": 26
+ },
+ {
+ "epoch": 0.384,
+ "grad_norm": 0.01058228686451912,
+ "learning_rate": 9.989182265027232e-06,
+ "loss": 0.44021710753440857,
+ "step": 27
+ },
+ {
+ "epoch": 0.3982222222222222,
+ "grad_norm": 0.009333685040473938,
+ "learning_rate": 9.98310071873072e-06,
+ "loss": 0.4491778016090393,
+ "step": 28
+ },
+ {
+ "epoch": 0.41244444444444445,
+ "grad_norm": 0.009250632487237453,
+ "learning_rate": 9.975671070024741e-06,
+ "loss": 0.4285920560359955,
+ "step": 29
+ },
+ {
+ "epoch": 0.4266666666666667,
+ "grad_norm": 0.010210932232439518,
+ "learning_rate": 9.966895328888195e-06,
+ "loss": 0.5484102368354797,
+ "step": 30
+ },
+ {
+ "epoch": 0.4408888888888889,
+ "grad_norm": 0.008848125115036964,
+ "learning_rate": 9.956775869464901e-06,
+ "loss": 0.4242202043533325,
+ "step": 31
+ },
+ {
+ "epoch": 0.45511111111111113,
+ "grad_norm": 0.00905672088265419,
+ "learning_rate": 9.945315429421307e-06,
+ "loss": 0.4533390402793884,
+ "step": 32
+ },
+ {
+ "epoch": 0.4693333333333333,
+ "grad_norm": 0.008254914544522762,
+ "learning_rate": 9.932517109205849e-06,
+ "loss": 0.4243648052215576,
+ "step": 33
+ },
+ {
+ "epoch": 0.48355555555555557,
+ "grad_norm": 0.009192033670842648,
+ "learning_rate": 9.918384371210178e-06,
+ "loss": 0.5120309591293335,
+ "step": 34
+ },
+ {
+ "epoch": 0.49777777777777776,
+ "grad_norm": 0.009636987932026386,
+ "learning_rate": 9.902921038832456e-06,
+ "loss": 0.4871100187301636,
+ "step": 35
+ },
+ {
+ "epoch": 0.512,
+ "grad_norm": 0.008572430349886417,
+ "learning_rate": 9.886131295443003e-06,
+ "loss": 0.46277713775634766,
+ "step": 36
+ },
+ {
+ "epoch": 0.5262222222222223,
+ "grad_norm": 0.008587818592786789,
+ "learning_rate": 9.868019683252543e-06,
+ "loss": 0.45996755361557007,
+ "step": 37
+ },
+ {
+ "epoch": 0.5404444444444444,
+ "grad_norm": 0.010775095783174038,
+ "learning_rate": 9.848591102083375e-06,
+ "loss": 0.47378021478652954,
+ "step": 38
+ },
+ {
+ "epoch": 0.5546666666666666,
+ "grad_norm": 0.00995537731796503,
+ "learning_rate": 9.82785080804381e-06,
+ "loss": 0.4684975743293762,
+ "step": 39
+ },
+ {
+ "epoch": 0.5688888888888889,
+ "grad_norm": 0.009079818613827229,
+ "learning_rate": 9.805804412106197e-06,
+ "loss": 0.5784010887145996,
+ "step": 40
+ },
+ {
+ "epoch": 0.5831111111111111,
+ "grad_norm": 0.008139098063111305,
+ "learning_rate": 9.782457878588977e-06,
+ "loss": 0.46650874614715576,
+ "step": 41
+ },
+ {
+ "epoch": 0.5973333333333334,
+ "grad_norm": 0.007847482338547707,
+ "learning_rate": 9.75781752354311e-06,
+ "loss": 0.42804157733917236,
+ "step": 42
+ },
+ {
+ "epoch": 0.6115555555555555,
+ "grad_norm": 0.008446140214800835,
+ "learning_rate": 9.731890013043367e-06,
+ "loss": 0.48224180936813354,
+ "step": 43
+ },
+ {
+ "epoch": 0.6257777777777778,
+ "grad_norm": 0.008632696233689785,
+ "learning_rate": 9.704682361384941e-06,
+ "loss": 0.4561079144477844,
+ "step": 44
+ },
+ {
+ "epoch": 0.64,
+ "grad_norm": 0.008637293241918087,
+ "learning_rate": 9.676201929185809e-06,
+ "loss": 0.4007866382598877,
+ "step": 45
+ },
+ {
+ "epoch": 0.6542222222222223,
+ "grad_norm": 0.008761093020439148,
+ "learning_rate": 9.646456421395447e-06,
+ "loss": 0.4587845802307129,
+ "step": 46
+ },
+ {
+ "epoch": 0.6684444444444444,
+ "grad_norm": 0.008726399391889572,
+ "learning_rate": 9.615453885210368e-06,
+ "loss": 0.4571158289909363,
+ "step": 47
+ },
+ {
+ "epoch": 0.6826666666666666,
+ "grad_norm": 0.008282204158604145,
+ "learning_rate": 9.583202707897075e-06,
+ "loss": 0.4724605679512024,
+ "step": 48
+ },
+ {
+ "epoch": 0.6968888888888889,
+ "grad_norm": 0.008231760933995247,
+ "learning_rate": 9.549711614523007e-06,
+ "loss": 0.4600468575954437,
+ "step": 49
+ },
+ {
+ "epoch": 0.7111111111111111,
+ "grad_norm": 0.00751511100679636,
+ "learning_rate": 9.514989665596114e-06,
+ "loss": 0.4377005994319916,
+ "step": 50
+ },
+ {
+ "epoch": 0.7111111111111111,
+ "eval_loss": 0.4685295820236206,
+ "eval_runtime": 5.4716,
+ "eval_samples_per_second": 36.553,
+ "eval_steps_per_second": 18.276,
+ "step": 50
+ },
+ {
+ "epoch": 0.7253333333333334,
+ "grad_norm": 0.008261662907898426,
+ "learning_rate": 9.479046254613673e-06,
+ "loss": 0.4415626525878906,
+ "step": 51
+ },
+ {
+ "epoch": 0.7395555555555555,
+ "grad_norm": 0.007395230233669281,
+ "learning_rate": 9.441891105521005e-06,
+ "loss": 0.42561036348342896,
+ "step": 52
+ },
+ {
+ "epoch": 0.7537777777777778,
+ "grad_norm": 0.008451394736766815,
+ "learning_rate": 9.40353427008083e-06,
+ "loss": 0.4630648195743561,
+ "step": 53
+ },
+ {
+ "epoch": 0.768,
+ "grad_norm": 0.007475187536329031,
+ "learning_rate": 9.3639861251539e-06,
+ "loss": 0.43418699502944946,
+ "step": 54
+ },
+ {
+ "epoch": 0.7822222222222223,
+ "grad_norm": 0.00859637837857008,
+ "learning_rate": 9.323257369891702e-06,
+ "loss": 0.47063255310058594,
+ "step": 55
+ },
+ {
+ "epoch": 0.7964444444444444,
+ "grad_norm": 0.007597765419632196,
+ "learning_rate": 9.281359022841966e-06,
+ "loss": 0.41920918226242065,
+ "step": 56
+ },
+ {
+ "epoch": 0.8106666666666666,
+ "grad_norm": 0.018636908382177353,
+ "learning_rate": 9.238302418967757e-06,
+ "loss": 0.4757845401763916,
+ "step": 57
+ },
+ {
+ "epoch": 0.8248888888888889,
+ "grad_norm": 0.008423218503594398,
+ "learning_rate": 9.194099206580981e-06,
+ "loss": 0.5198032855987549,
+ "step": 58
+ },
+ {
+ "epoch": 0.8391111111111111,
+ "grad_norm": 0.008149159140884876,
+ "learning_rate": 9.14876134419111e-06,
+ "loss": 0.46035224199295044,
+ "step": 59
+ },
+ {
+ "epoch": 0.8533333333333334,
+ "grad_norm": 0.00819750688970089,
+ "learning_rate": 9.102301097269974e-06,
+ "loss": 0.4945712983608246,
+ "step": 60
+ },
+ {
+ "epoch": 0.8675555555555555,
+ "grad_norm": 0.00855066254734993,
+ "learning_rate": 9.05473103493355e-06,
+ "loss": 0.4513372778892517,
+ "step": 61
+ },
+ {
+ "epoch": 0.8817777777777778,
+ "grad_norm": 0.008660132065415382,
+ "learning_rate": 9.006064026541549e-06,
+ "loss": 0.49865084886550903,
+ "step": 62
+ },
+ {
+ "epoch": 0.896,
+ "grad_norm": 0.007741494104266167,
+ "learning_rate": 8.956313238215824e-06,
+ "loss": 0.4319220185279846,
+ "step": 63
+ },
+ {
+ "epoch": 0.9102222222222223,
+ "grad_norm": 0.007900395430624485,
+ "learning_rate": 8.905492129278478e-06,
+ "loss": 0.43760013580322266,
+ "step": 64
+ },
+ {
+ "epoch": 0.9244444444444444,
+ "grad_norm": 0.007346940692514181,
+ "learning_rate": 8.85361444861063e-06,
+ "loss": 0.41593095660209656,
+ "step": 65
+ },
+ {
+ "epoch": 0.9386666666666666,
+ "grad_norm": 0.00902250874787569,
+ "learning_rate": 8.800694230932885e-06,
+ "loss": 0.48464590311050415,
+ "step": 66
+ },
+ {
+ "epoch": 0.9528888888888889,
+ "grad_norm": 0.007672084029763937,
+ "learning_rate": 8.74674579300843e-06,
+ "loss": 0.41884416341781616,
+ "step": 67
+ },
+ {
+ "epoch": 0.9671111111111111,
+ "grad_norm": 0.008009645156562328,
+ "learning_rate": 8.691783729769874e-06,
+ "loss": 0.40650737285614014,
+ "step": 68
+ },
+ {
+ "epoch": 0.9813333333333333,
+ "grad_norm": 0.008077424019575119,
+ "learning_rate": 8.635822910370793e-06,
+ "loss": 0.4726918637752533,
+ "step": 69
+ },
+ {
+ "epoch": 0.9955555555555555,
+ "grad_norm": 0.007576521951705217,
+ "learning_rate": 8.578878474163115e-06,
+ "loss": 0.4404842257499695,
+ "step": 70
+ },
+ {
+ "epoch": 1.0,
+ "grad_norm": 0.012638773769140244,
+ "learning_rate": 8.520965826601394e-06,
+ "loss": 0.42633137106895447,
+ "step": 71
+ },
+ {
+ "epoch": 1.0142222222222221,
+ "grad_norm": 0.008662556298077106,
+ "learning_rate": 8.462100635075097e-06,
+ "loss": 0.46000930666923523,
+ "step": 72
+ },
+ {
+ "epoch": 1.0284444444444445,
+ "grad_norm": 0.008270082995295525,
+ "learning_rate": 8.40229882467003e-06,
+ "loss": 0.42150819301605225,
+ "step": 73
+ },
+ {
+ "epoch": 1.0426666666666666,
+ "grad_norm": 0.007625850848853588,
+ "learning_rate": 8.341576573860049e-06,
+ "loss": 0.37313175201416016,
+ "step": 74
+ },
+ {
+ "epoch": 1.056888888888889,
+ "grad_norm": 0.00790152233093977,
+ "learning_rate": 8.279950310130218e-06,
+ "loss": 0.382789671421051,
+ "step": 75
+ },
+ {
+ "epoch": 1.0711111111111111,
+ "grad_norm": 0.007986027747392654,
+ "learning_rate": 8.2174367055326e-06,
+ "loss": 0.3734534978866577,
+ "step": 76
+ },
+ {
+ "epoch": 1.0853333333333333,
+ "grad_norm": 0.007174879312515259,
+ "learning_rate": 8.154052672175888e-06,
+ "loss": 0.37693876028060913,
+ "step": 77
+ },
+ {
+ "epoch": 1.0995555555555556,
+ "grad_norm": 0.008486706763505936,
+ "learning_rate": 8.08981535765009e-06,
+ "loss": 0.4265296459197998,
+ "step": 78
+ },
+ {
+ "epoch": 1.1137777777777778,
+ "grad_norm": 0.007609546184539795,
+ "learning_rate": 8.024742140387506e-06,
+ "loss": 0.3812956213951111,
+ "step": 79
+ },
+ {
+ "epoch": 1.1280000000000001,
+ "grad_norm": 0.007912754081189632,
+ "learning_rate": 7.95885062496126e-06,
+ "loss": 0.4131423234939575,
+ "step": 80
+ },
+ {
+ "epoch": 1.1422222222222222,
+ "grad_norm": 0.007641648408025503,
+ "learning_rate": 7.892158637322647e-06,
+ "loss": 0.37861311435699463,
+ "step": 81
+ },
+ {
+ "epoch": 1.1564444444444444,
+ "grad_norm": 0.007570190355181694,
+ "learning_rate": 7.824684219978591e-06,
+ "loss": 0.3589654564857483,
+ "step": 82
+ },
+ {
+ "epoch": 1.1706666666666667,
+ "grad_norm": 0.007670491933822632,
+ "learning_rate": 7.756445627110523e-06,
+ "loss": 0.40794360637664795,
+ "step": 83
+ },
+ {
+ "epoch": 1.1848888888888889,
+ "grad_norm": 0.009116973727941513,
+ "learning_rate": 7.68746131963598e-06,
+ "loss": 0.4278006851673126,
+ "step": 84
+ },
+ {
+ "epoch": 1.199111111111111,
+ "grad_norm": 0.007852817885577679,
+ "learning_rate": 7.6177499602143e-06,
+ "loss": 0.371395081281662,
+ "step": 85
+ },
+ {
+ "epoch": 1.2133333333333334,
+ "grad_norm": 0.00870604906231165,
+ "learning_rate": 7.547330408197695e-06,
+ "loss": 0.4305034279823303,
+ "step": 86
+ },
+ {
+ "epoch": 1.2275555555555555,
+ "grad_norm": 0.008404702879488468,
+ "learning_rate": 7.476221714529167e-06,
+ "loss": 0.4348181188106537,
+ "step": 87
+ },
+ {
+ "epoch": 1.2417777777777779,
+ "grad_norm": 0.008913768455386162,
+ "learning_rate": 7.404443116588548e-06,
+ "loss": 0.4309384822845459,
+ "step": 88
+ },
+ {
+ "epoch": 1.256,
+ "grad_norm": 0.007682010997086763,
+ "learning_rate": 7.332014032988123e-06,
+ "loss": 0.4107038676738739,
+ "step": 89
+ },
+ {
+ "epoch": 1.2702222222222221,
+ "grad_norm": 0.00787457637488842,
+ "learning_rate": 7.2589540583192165e-06,
+ "loss": 0.37457963824272156,
+ "step": 90
+ },
+ {
+ "epoch": 1.2844444444444445,
+ "grad_norm": 0.007768277078866959,
+ "learning_rate": 7.185282957851175e-06,
+ "loss": 0.37697723507881165,
+ "step": 91
+ },
+ {
+ "epoch": 1.2986666666666666,
+ "grad_norm": 0.008521760813891888,
+ "learning_rate": 7.111020662184174e-06,
+ "loss": 0.40353477001190186,
+ "step": 92
+ },
+ {
+ "epoch": 1.3128888888888888,
+ "grad_norm": 0.0077499281615018845,
+ "learning_rate": 7.036187261857289e-06,
+ "loss": 0.37755411863327026,
+ "step": 93
+ },
+ {
+ "epoch": 1.3271111111111111,
+ "grad_norm": 0.00870734266936779,
+ "learning_rate": 6.960803001913315e-06,
+ "loss": 0.3729894757270813,
+ "step": 94
+ },
+ {
+ "epoch": 1.3413333333333333,
+ "grad_norm": 0.007563227321952581,
+ "learning_rate": 6.884888276421766e-06,
+ "loss": 0.4091319441795349,
+ "step": 95
+ },
+ {
+ "epoch": 1.3555555555555556,
+ "grad_norm": 0.00803642813116312,
+ "learning_rate": 6.8084636229615786e-06,
+ "loss": 0.4048840403556824,
+ "step": 96
+ },
+ {
+ "epoch": 1.3697777777777778,
+ "grad_norm": 0.00779891200363636,
+ "learning_rate": 6.731549717064975e-06,
+ "loss": 0.39409685134887695,
+ "step": 97
+ },
+ {
+ "epoch": 1.384,
+ "grad_norm": 0.008260682225227356,
+ "learning_rate": 6.654167366624009e-06,
+ "loss": 0.36237582564353943,
+ "step": 98
+ },
+ {
+ "epoch": 1.3982222222222223,
+ "grad_norm": 0.008373094722628593,
+ "learning_rate": 6.576337506261314e-06,
+ "loss": 0.38763654232025146,
+ "step": 99
+ },
+ {
+ "epoch": 1.4124444444444444,
+ "grad_norm": 0.007797387894243002,
+ "learning_rate": 6.498081191666549e-06,
+ "loss": 0.41751331090927124,
+ "step": 100
+ },
+ {
+ "epoch": 1.4124444444444444,
+ "eval_loss": 0.4669477939605713,
+ "eval_runtime": 5.2715,
+ "eval_samples_per_second": 37.94,
+ "eval_steps_per_second": 18.97,
+ "step": 100
+ },
+ {
+ "epoch": 1.4266666666666667,
+ "grad_norm": 0.007338334806263447,
+ "learning_rate": 6.419419593900109e-06,
+ "loss": 0.38504236936569214,
+ "step": 101
+ },
+ {
+ "epoch": 1.4408888888888889,
+ "grad_norm": 0.007528901565819979,
+ "learning_rate": 6.340373993665607e-06,
+ "loss": 0.36031031608581543,
+ "step": 102
+ },
+ {
+ "epoch": 1.455111111111111,
+ "grad_norm": 0.007573080714792013,
+ "learning_rate": 6.2609657755527135e-06,
+ "loss": 0.3863949179649353,
+ "step": 103
+ },
+ {
+ "epoch": 1.4693333333333334,
+ "grad_norm": 0.008099780417978764,
+ "learning_rate": 6.1812164222518626e-06,
+ "loss": 0.3955462574958801,
+ "step": 104
+ },
+ {
+ "epoch": 1.4835555555555555,
+ "grad_norm": 0.008717137388885021,
+ "learning_rate": 6.101147508742456e-06,
+ "loss": 0.39310169219970703,
+ "step": 105
+ },
+ {
+ "epoch": 1.4977777777777779,
+ "grad_norm": 0.008383029140532017,
+ "learning_rate": 6.020780696456059e-06,
+ "loss": 0.4533539414405823,
+ "step": 106
+ },
+ {
+ "epoch": 1.512,
+ "grad_norm": 0.007702656090259552,
+ "learning_rate": 5.940137727416247e-06,
+ "loss": 0.3985057473182678,
+ "step": 107
+ },
+ {
+ "epoch": 1.5262222222222221,
+ "grad_norm": 0.00808175653219223,
+ "learning_rate": 5.859240418356614e-06,
+ "loss": 0.3878968358039856,
+ "step": 108
+ },
+ {
+ "epoch": 1.5404444444444443,
+ "grad_norm": 0.008781371638178825,
+ "learning_rate": 5.778110654818602e-06,
+ "loss": 0.39648741483688354,
+ "step": 109
+ },
+ {
+ "epoch": 1.5546666666666666,
+ "grad_norm": 0.008896375074982643,
+ "learning_rate": 5.696770385230679e-06,
+ "loss": 0.4319821000099182,
+ "step": 110
+ },
+ {
+ "epoch": 1.568888888888889,
+ "grad_norm": 0.008202116936445236,
+ "learning_rate": 5.615241614970546e-06,
+ "loss": 0.4078192710876465,
+ "step": 111
+ },
+ {
+ "epoch": 1.5831111111111111,
+ "grad_norm": 0.007843846455216408,
+ "learning_rate": 5.533546400411899e-06,
+ "loss": 0.38504159450531006,
+ "step": 112
+ },
+ {
+ "epoch": 1.5973333333333333,
+ "grad_norm": 0.00823921337723732,
+ "learning_rate": 5.4517068429574215e-06,
+ "loss": 0.3971819579601288,
+ "step": 113
+ },
+ {
+ "epoch": 1.6115555555555554,
+ "grad_norm": 0.007411513011902571,
+ "learning_rate": 5.3697450830595775e-06,
+ "loss": 0.3886791169643402,
+ "step": 114
+ },
+ {
+ "epoch": 1.6257777777777778,
+ "grad_norm": 0.007902945391833782,
+ "learning_rate": 5.287683294230855e-06,
+ "loss": 0.37610527873039246,
+ "step": 115
+ },
+ {
+ "epoch": 1.6400000000000001,
+ "grad_norm": 0.007778776343911886,
+ "learning_rate": 5.20554367704505e-06,
+ "loss": 0.3846814036369324,
+ "step": 116
+ },
+ {
+ "epoch": 1.6542222222222223,
+ "grad_norm": 0.008897176012396812,
+ "learning_rate": 5.123348453131242e-06,
+ "loss": 0.3917155861854553,
+ "step": 117
+ },
+ {
+ "epoch": 1.6684444444444444,
+ "grad_norm": 0.008548559620976448,
+ "learning_rate": 5.041119859162068e-06,
+ "loss": 0.39533206820487976,
+ "step": 118
+ },
+ {
+ "epoch": 1.6826666666666665,
+ "grad_norm": 0.008137239143252373,
+ "learning_rate": 4.958880140837934e-06,
+ "loss": 0.4113273620605469,
+ "step": 119
+ },
+ {
+ "epoch": 1.696888888888889,
+ "grad_norm": 0.008478965610265732,
+ "learning_rate": 4.876651546868759e-06,
+ "loss": 0.39543214440345764,
+ "step": 120
+ },
+ {
+ "epoch": 1.7111111111111112,
+ "grad_norm": 0.00835413858294487,
+ "learning_rate": 4.794456322954953e-06,
+ "loss": 0.4548013508319855,
+ "step": 121
+ },
+ {
+ "epoch": 1.7253333333333334,
+ "grad_norm": 0.00832394976168871,
+ "learning_rate": 4.7123167057691446e-06,
+ "loss": 0.4111294150352478,
+ "step": 122
+ },
+ {
+ "epoch": 1.7395555555555555,
+ "grad_norm": 0.008472571149468422,
+ "learning_rate": 4.630254916940424e-06,
+ "loss": 0.43050098419189453,
+ "step": 123
+ },
+ {
+ "epoch": 1.7537777777777777,
+ "grad_norm": 0.00816201139241457,
+ "learning_rate": 4.548293157042581e-06,
+ "loss": 0.41356489062309265,
+ "step": 124
+ },
+ {
+ "epoch": 1.768,
+ "grad_norm": 0.008180899545550346,
+ "learning_rate": 4.466453599588103e-06,
+ "loss": 0.41799914836883545,
+ "step": 125
+ },
+ {
+ "epoch": 1.7822222222222224,
+ "grad_norm": 0.007402091287076473,
+ "learning_rate": 4.384758385029457e-06,
+ "loss": 0.3510154187679291,
+ "step": 126
+ },
+ {
+ "epoch": 1.7964444444444445,
+ "grad_norm": 0.008476709015667439,
+ "learning_rate": 4.3032296147693225e-06,
+ "loss": 0.4098949432373047,
+ "step": 127
+ },
+ {
+ "epoch": 1.8106666666666666,
+ "grad_norm": 0.008135325275361538,
+ "learning_rate": 4.2218893451814005e-06,
+ "loss": 0.36957240104675293,
+ "step": 128
+ },
+ {
+ "epoch": 1.8248888888888888,
+ "grad_norm": 0.007964376360177994,
+ "learning_rate": 4.140759581643386e-06,
+ "loss": 0.3535369038581848,
+ "step": 129
+ },
+ {
+ "epoch": 1.8391111111111111,
+ "grad_norm": 0.00742813665419817,
+ "learning_rate": 4.059862272583755e-06,
+ "loss": 0.38296306133270264,
+ "step": 130
+ },
+ {
+ "epoch": 1.8533333333333335,
+ "grad_norm": 0.007901367731392384,
+ "learning_rate": 3.979219303543942e-06,
+ "loss": 0.38752996921539307,
+ "step": 131
+ },
+ {
+ "epoch": 1.8675555555555556,
+ "grad_norm": 0.008205283433198929,
+ "learning_rate": 3.898852491257547e-06,
+ "loss": 0.3867356479167938,
+ "step": 132
+ },
+ {
+ "epoch": 1.8817777777777778,
+ "grad_norm": 0.008251599036157131,
+ "learning_rate": 3.818783577748138e-06,
+ "loss": 0.42934614419937134,
+ "step": 133
+ },
+ {
+ "epoch": 1.896,
+ "grad_norm": 0.008710331283509731,
+ "learning_rate": 3.7390342244472886e-06,
+ "loss": 0.4096015691757202,
+ "step": 134
+ },
+ {
+ "epoch": 1.9102222222222223,
+ "grad_norm": 0.008127489127218723,
+ "learning_rate": 3.659626006334395e-06,
+ "loss": 0.3849722743034363,
+ "step": 135
+ },
+ {
+ "epoch": 1.9244444444444444,
+ "grad_norm": 0.008900986984372139,
+ "learning_rate": 3.5805804060998926e-06,
+ "loss": 0.415260910987854,
+ "step": 136
+ },
+ {
+ "epoch": 1.9386666666666668,
+ "grad_norm": 0.00907993782311678,
+ "learning_rate": 3.501918808333453e-06,
+ "loss": 0.48382842540740967,
+ "step": 137
+ },
+ {
+ "epoch": 1.952888888888889,
+ "grad_norm": 0.008263039402663708,
+ "learning_rate": 3.4236624937386874e-06,
+ "loss": 0.4434107542037964,
+ "step": 138
+ },
+ {
+ "epoch": 1.967111111111111,
+ "grad_norm": 0.008353665471076965,
+ "learning_rate": 3.3458326333759927e-06,
+ "loss": 0.41760051250457764,
+ "step": 139
+ },
+ {
+ "epoch": 1.9813333333333332,
+ "grad_norm": 0.008556668646633625,
+ "learning_rate": 3.268450282935026e-06,
+ "loss": 0.43642479181289673,
+ "step": 140
+ },
+ {
+ "epoch": 1.9955555555555555,
+ "grad_norm": 0.008040979504585266,
+ "learning_rate": 3.1915363770384223e-06,
+ "loss": 0.38327687978744507,
+ "step": 141
+ },
+ {
+ "epoch": 2.0,
+ "grad_norm": 0.012571632862091064,
+ "learning_rate": 3.1151117235782346e-06,
+ "loss": 0.3751964867115021,
+ "step": 142
+ },
+ {
+ "epoch": 2.014222222222222,
+ "grad_norm": 0.008403269574046135,
+ "learning_rate": 3.0391969980866874e-06,
+ "loss": 0.3831713795661926,
+ "step": 143
+ },
+ {
+ "epoch": 2.0284444444444443,
+ "grad_norm": 0.008236616849899292,
+ "learning_rate": 2.963812738142713e-06,
+ "loss": 0.338630735874176,
+ "step": 144
+ },
+ {
+ "epoch": 2.042666666666667,
+ "grad_norm": 0.008433214388787746,
+ "learning_rate": 2.8889793378158284e-06,
+ "loss": 0.3684050440788269,
+ "step": 145
+ },
+ {
+ "epoch": 2.056888888888889,
+ "grad_norm": 0.008040281943976879,
+ "learning_rate": 2.814717042148827e-06,
+ "loss": 0.37092137336730957,
+ "step": 146
+ },
+ {
+ "epoch": 2.071111111111111,
+ "grad_norm": 0.008197156712412834,
+ "learning_rate": 2.7410459416807856e-06,
+ "loss": 0.34995853900909424,
+ "step": 147
+ },
+ {
+ "epoch": 2.0853333333333333,
+ "grad_norm": 0.009072407148778439,
+ "learning_rate": 2.6679859670118785e-06,
+ "loss": 0.37355348467826843,
+ "step": 148
+ },
+ {
+ "epoch": 2.0995555555555554,
+ "grad_norm": 0.00807118322700262,
+ "learning_rate": 2.5955568834114523e-06,
+ "loss": 0.3448713421821594,
+ "step": 149
+ },
+ {
+ "epoch": 2.113777777777778,
+ "grad_norm": 0.00770068122074008,
+ "learning_rate": 2.523778285470835e-06,
+ "loss": 0.3545028865337372,
+ "step": 150
+ },
+ {
+ "epoch": 2.113777777777778,
+ "eval_loss": 0.4679063558578491,
+ "eval_runtime": 5.4161,
+ "eval_samples_per_second": 36.927,
+ "eval_steps_per_second": 18.463,
+ "step": 150
+ },
+ {
+ "epoch": 2.128,
+ "grad_norm": 0.007919526658952236,
+ "learning_rate": 2.452669591802307e-06,
+ "loss": 0.32551515102386475,
+ "step": 151
+ },
+ {
+ "epoch": 2.1422222222222222,
+ "grad_norm": 0.007321993354707956,
+ "learning_rate": 2.3822500397857016e-06,
+ "loss": 0.35764873027801514,
+ "step": 152
+ },
+ {
+ "epoch": 2.1564444444444444,
+ "grad_norm": 0.0076365480199456215,
+ "learning_rate": 2.3125386803640188e-06,
+ "loss": 0.35795581340789795,
+ "step": 153
+ },
+ {
+ "epoch": 2.1706666666666665,
+ "grad_norm": 0.007587647996842861,
+ "learning_rate": 2.243554372889479e-06,
+ "loss": 0.32651567459106445,
+ "step": 154
+ },
+ {
+ "epoch": 2.1848888888888887,
+ "grad_norm": 0.007799312938004732,
+ "learning_rate": 2.1753157800214107e-06,
+ "loss": 0.3542096018791199,
+ "step": 155
+ },
+ {
+ "epoch": 2.1991111111111112,
+ "grad_norm": 0.007910694926977158,
+ "learning_rate": 2.1078413626773547e-06,
+ "loss": 0.3560444116592407,
+ "step": 156
+ },
+ {
+ "epoch": 2.2133333333333334,
+ "grad_norm": 0.008300457149744034,
+ "learning_rate": 2.0411493750387423e-06,
+ "loss": 0.3283962607383728,
+ "step": 157
+ },
+ {
+ "epoch": 2.2275555555555555,
+ "grad_norm": 0.007787623442709446,
+ "learning_rate": 1.9752578596124955e-06,
+ "loss": 0.362970769405365,
+ "step": 158
+ },
+ {
+ "epoch": 2.2417777777777776,
+ "grad_norm": 0.007931886240839958,
+ "learning_rate": 1.9101846423499113e-06,
+ "loss": 0.3372136354446411,
+ "step": 159
+ },
+ {
+ "epoch": 2.2560000000000002,
+ "grad_norm": 0.008211000822484493,
+ "learning_rate": 1.8459473278241125e-06,
+ "loss": 0.343697726726532,
+ "step": 160
+ },
+ {
+ "epoch": 2.2702222222222224,
+ "grad_norm": 0.008033247664570808,
+ "learning_rate": 1.7825632944674016e-06,
+ "loss": 0.3472204804420471,
+ "step": 161
+ },
+ {
+ "epoch": 2.2844444444444445,
+ "grad_norm": 0.008930593729019165,
+ "learning_rate": 1.7200496898697832e-06,
+ "loss": 0.34057295322418213,
+ "step": 162
+ },
+ {
+ "epoch": 2.2986666666666666,
+ "grad_norm": 0.007960367947816849,
+ "learning_rate": 1.6584234261399535e-06,
+ "loss": 0.3398140072822571,
+ "step": 163
+ },
+ {
+ "epoch": 2.3128888888888888,
+ "grad_norm": 0.008388614282011986,
+ "learning_rate": 1.5977011753299726e-06,
+ "loss": 0.3423030376434326,
+ "step": 164
+ },
+ {
+ "epoch": 2.327111111111111,
+ "grad_norm": 0.009143203496932983,
+ "learning_rate": 1.5378993649249053e-06,
+ "loss": 0.4059848189353943,
+ "step": 165
+ },
+ {
+ "epoch": 2.3413333333333335,
+ "grad_norm": 0.008042566478252411,
+ "learning_rate": 1.4790341733986085e-06,
+ "loss": 0.35659873485565186,
+ "step": 166
+ },
+ {
+ "epoch": 2.3555555555555556,
+ "grad_norm": 0.007340035866945982,
+ "learning_rate": 1.4211215258368866e-06,
+ "loss": 0.320560097694397,
+ "step": 167
+ },
+ {
+ "epoch": 2.3697777777777778,
+ "grad_norm": 0.008758354000747204,
+ "learning_rate": 1.3641770896292083e-06,
+ "loss": 0.3432486057281494,
+ "step": 168
+ },
+ {
+ "epoch": 2.384,
+ "grad_norm": 0.008217270486056805,
+ "learning_rate": 1.3082162702301276e-06,
+ "loss": 0.3448861241340637,
+ "step": 169
+ },
+ {
+ "epoch": 2.398222222222222,
+ "grad_norm": 0.008302765898406506,
+ "learning_rate": 1.2532542069915722e-06,
+ "loss": 0.4012100100517273,
+ "step": 170
+ },
+ {
+ "epoch": 2.4124444444444446,
+ "grad_norm": 0.0086008096113801,
+ "learning_rate": 1.1993057690671174e-06,
+ "loss": 0.35434189438819885,
+ "step": 171
+ },
+ {
+ "epoch": 2.4266666666666667,
+ "grad_norm": 0.009528086520731449,
+ "learning_rate": 1.1463855513893695e-06,
+ "loss": 0.37489616870880127,
+ "step": 172
+ },
+ {
+ "epoch": 2.440888888888889,
+ "grad_norm": 0.008264975622296333,
+ "learning_rate": 1.0945078707215224e-06,
+ "loss": 0.3599584698677063,
+ "step": 173
+ },
+ {
+ "epoch": 2.455111111111111,
+ "grad_norm": 0.008293098770081997,
+ "learning_rate": 1.0436867617841768e-06,
+ "loss": 0.38492339849472046,
+ "step": 174
+ },
+ {
+ "epoch": 2.469333333333333,
+ "grad_norm": 0.007827190682291985,
+ "learning_rate": 9.939359734584552e-07,
+ "loss": 0.36968979239463806,
+ "step": 175
+ },
+ {
+ "epoch": 2.4835555555555557,
+ "grad_norm": 0.008390667848289013,
+ "learning_rate": 9.452689650664515e-07,
+ "loss": 0.3575425148010254,
+ "step": 176
+ },
+ {
+ "epoch": 2.497777777777778,
+ "grad_norm": 0.007562275975942612,
+ "learning_rate": 8.976989027300265e-07,
+ "loss": 0.35382080078125,
+ "step": 177
+ },
+ {
+ "epoch": 2.512,
+ "grad_norm": 0.007893560454249382,
+ "learning_rate": 8.512386558088919e-07,
+ "loss": 0.35292625427246094,
+ "step": 178
+ },
+ {
+ "epoch": 2.526222222222222,
+ "grad_norm": 0.007973878644406796,
+ "learning_rate": 8.059007934190194e-07,
+ "loss": 0.3658143877983093,
+ "step": 179
+ },
+ {
+ "epoch": 2.5404444444444443,
+ "grad_norm": 0.008489987812936306,
+ "learning_rate": 7.61697581032243e-07,
+ "loss": 0.34945034980773926,
+ "step": 180
+ },
+ {
+ "epoch": 2.554666666666667,
+ "grad_norm": 0.007993645034730434,
+ "learning_rate": 7.186409771580355e-07,
+ "loss": 0.3492674231529236,
+ "step": 181
+ },
+ {
+ "epoch": 2.568888888888889,
+ "grad_norm": 0.007816205732524395,
+ "learning_rate": 6.76742630108298e-07,
+ "loss": 0.35949787497520447,
+ "step": 182
+ },
+ {
+ "epoch": 2.583111111111111,
+ "grad_norm": 0.00765154417604208,
+ "learning_rate": 6.360138748461015e-07,
+ "loss": 0.35065189003944397,
+ "step": 183
+ },
+ {
+ "epoch": 2.5973333333333333,
+ "grad_norm": 0.007598839234560728,
+ "learning_rate": 5.964657299191712e-07,
+ "loss": 0.31525540351867676,
+ "step": 184
+ },
+ {
+ "epoch": 2.6115555555555554,
+ "grad_norm": 0.00796365924179554,
+ "learning_rate": 5.581088944789953e-07,
+ "loss": 0.34680360555648804,
+ "step": 185
+ },
+ {
+ "epoch": 2.6257777777777775,
+ "grad_norm": 0.008079800754785538,
+ "learning_rate": 5.20953745386329e-07,
+ "loss": 0.3423649072647095,
+ "step": 186
+ },
+ {
+ "epoch": 2.64,
+ "grad_norm": 0.008171727880835533,
+ "learning_rate": 4.850103344038853e-07,
+ "loss": 0.3583834767341614,
+ "step": 187
+ },
+ {
+ "epoch": 2.6542222222222223,
+ "grad_norm": 0.007568752858787775,
+ "learning_rate": 4.502883854769935e-07,
+ "loss": 0.3272634744644165,
+ "step": 188
+ },
+ {
+ "epoch": 2.6684444444444444,
+ "grad_norm": 0.00814193394035101,
+ "learning_rate": 4.167972921029262e-07,
+ "loss": 0.3352956175804138,
+ "step": 189
+ },
+ {
+ "epoch": 2.6826666666666665,
+ "grad_norm": 0.008751407265663147,
+ "learning_rate": 3.8454611478963235e-07,
+ "loss": 0.3704286515712738,
+ "step": 190
+ },
+ {
+ "epoch": 2.696888888888889,
+ "grad_norm": 0.007807895075529814,
+ "learning_rate": 3.535435786045538e-07,
+ "loss": 0.34566229581832886,
+ "step": 191
+ },
+ {
+ "epoch": 2.7111111111111112,
+ "grad_norm": 0.007984541356563568,
+ "learning_rate": 3.237980708141919e-07,
+ "loss": 0.35052379965782166,
+ "step": 192
+ },
+ {
+ "epoch": 2.7253333333333334,
+ "grad_norm": 0.009312397800385952,
+ "learning_rate": 2.9531763861505967e-07,
+ "loss": 0.38708382844924927,
+ "step": 193
+ },
+ {
+ "epoch": 2.7395555555555555,
+ "grad_norm": 0.008055887185037136,
+ "learning_rate": 2.681099869566328e-07,
+ "loss": 0.3316305875778198,
+ "step": 194
+ },
+ {
+ "epoch": 2.7537777777777777,
+ "grad_norm": 0.008029562421143055,
+ "learning_rate": 2.4218247645689306e-07,
+ "loss": 0.3650358319282532,
+ "step": 195
+ },
+ {
+ "epoch": 2.768,
+ "grad_norm": 0.00823246967047453,
+ "learning_rate": 2.175421214110235e-07,
+ "loss": 0.36560434103012085,
+ "step": 196
+ },
+ {
+ "epoch": 2.7822222222222224,
+ "grad_norm": 0.008692881092429161,
+ "learning_rate": 1.941955878938029e-07,
+ "loss": 0.3463239371776581,
+ "step": 197
+ },
+ {
+ "epoch": 2.7964444444444445,
+ "grad_norm": 0.008173434063792229,
+ "learning_rate": 1.721491919561913e-07,
+ "loss": 0.35629892349243164,
+ "step": 198
+ },
+ {
+ "epoch": 2.8106666666666666,
+ "grad_norm": 0.00767622422426939,
+ "learning_rate": 1.514088979166256e-07,
+ "loss": 0.33887457847595215,
+ "step": 199
+ },
+ {
+ "epoch": 2.824888888888889,
+ "grad_norm": 0.007811338175088167,
+ "learning_rate": 1.3198031674745814e-07,
+ "loss": 0.36586421728134155,
+ "step": 200
+ },
+ {
+ "epoch": 2.824888888888889,
+ "eval_loss": 0.4734371602535248,
+ "eval_runtime": 7.1506,
+ "eval_samples_per_second": 27.97,
+ "eval_steps_per_second": 13.985,
+ "step": 200
+ },
+ {
+ "epoch": 2.8391111111111114,
+ "grad_norm": 0.008272381499409676,
+ "learning_rate": 1.138687045569975e-07,
+ "loss": 0.3514959216117859,
+ "step": 201
+ },
+ {
+ "epoch": 2.8533333333333335,
+ "grad_norm": 0.008001282811164856,
+ "learning_rate": 9.70789611675449e-08,
+ "loss": 0.37950894236564636,
+ "step": 202
+ },
+ {
+ "epoch": 2.8675555555555556,
+ "grad_norm": 0.007416888140141964,
+ "learning_rate": 8.161562878982399e-08,
+ "loss": 0.3352017402648926,
+ "step": 203
+ },
+ {
+ "epoch": 2.8817777777777778,
+ "grad_norm": 0.008772965520620346,
+ "learning_rate": 6.74828907941516e-08,
+ "loss": 0.36504414677619934,
+ "step": 204
+ },
+ {
+ "epoch": 2.896,
+ "grad_norm": 0.008397468365728855,
+ "learning_rate": 5.468457057869358e-08,
+ "loss": 0.3987666964530945,
+ "step": 205
+ },
+ {
+ "epoch": 2.910222222222222,
+ "grad_norm": 0.00775982066988945,
+ "learning_rate": 4.322413053509944e-08,
+ "loss": 0.3345492482185364,
+ "step": 206
+ },
+ {
+ "epoch": 2.924444444444444,
+ "grad_norm": 0.008166249841451645,
+ "learning_rate": 3.31046711118066e-08,
+ "loss": 0.367769718170166,
+ "step": 207
+ },
+ {
+ "epoch": 2.9386666666666668,
+ "grad_norm": 0.007682916708290577,
+ "learning_rate": 2.4328929975260262e-08,
+ "loss": 0.34441590309143066,
+ "step": 208
+ },
+ {
+ "epoch": 2.952888888888889,
+ "grad_norm": 0.008051120676100254,
+ "learning_rate": 1.6899281269279756e-08,
+ "loss": 0.3472161889076233,
+ "step": 209
+ },
+ {
+ "epoch": 2.967111111111111,
+ "grad_norm": 0.0077741509303450584,
+ "learning_rate": 1.0817734972768946e-08,
+ "loss": 0.3906354308128357,
+ "step": 210
+ },
+ {
+ "epoch": 2.981333333333333,
+ "grad_norm": 0.008870133198797703,
+ "learning_rate": 6.085936355947897e-09,
+ "loss": 0.37977275252342224,
+ "step": 211
+ },
+ {
+ "epoch": 2.9955555555555557,
+ "grad_norm": 0.008871600031852722,
+ "learning_rate": 2.7051655352494654e-09,
+ "loss": 0.3533574342727661,
+ "step": 212
+ },
+ {
+ "epoch": 3.0,
+ "grad_norm": 0.014481575228273869,
+ "learning_rate": 6.763371270035457e-10,
+ "loss": 0.5198727250099182,
+ "step": 213
+ }
+ ],
+ "logging_steps": 1,
+ "max_steps": 213,
+ "num_input_tokens_seen": 0,
+ "num_train_epochs": 3,
+ "save_steps": 500,
+ "stateful_callbacks": {
+ "TrainerControl": {
+ "args": {
+ "should_epoch_stop": false,
+ "should_evaluate": false,
+ "should_log": false,
+ "should_save": true,
+ "should_training_stop": true
+ },
+ "attributes": {}
+ }
+ },
+ "total_flos": 1.0308479847759872e+17,
+ "train_batch_size": 1,
+ "trial_name": null,
+ "trial_params": null
+}
diff --git a/qwen2.5-3b-instruct/checkpoint-213/training_args.bin b/qwen2.5-3b-instruct/checkpoint-213/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..bd95d1f874b25af121b3c8254bd05a3eb95ffd5f
--- /dev/null
+++ b/qwen2.5-3b-instruct/checkpoint-213/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c4f1847ec07466eeb896ef40e20d9241e19882e62efb61a14f1e7c6a96737fa0
+size 5649
diff --git a/qwen2.5-3b-instruct/checkpoint-71/chat_template.jinja b/qwen2.5-3b-instruct/checkpoint-71/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..bdf7919a96cfe43d50914a007b9c0877bd0ec27e
--- /dev/null
+++ b/qwen2.5-3b-instruct/checkpoint-71/chat_template.jinja
@@ -0,0 +1,54 @@
+{%- if tools %}
+ {{- '<|im_start|>system\n' }}
+ {%- if messages[0]['role'] == 'system' %}
+ {{- messages[0]['content'] }}
+ {%- else %}
+ {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+ {%- endif %}
+ {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within XML tags:\n" }}
+ {%- for tool in tools %}
+ {{- "\n" }}
+ {{- tool | tojson }}
+ {%- endfor %}
+ {{- "\n\n\nFor each function call, return a json object with function name and arguments within XML tags:\n\n{\"name\": , \"arguments\": }\n<|im_end|>\n" }}
+{%- else %}
+ {%- if messages[0]['role'] == 'system' %}
+ {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+ {%- else %}
+ {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+ {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+ {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+ {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+ {%- elif message.role == "assistant" %}
+ {{- '<|im_start|>' + message.role }}
+ {%- if message.content %}
+ {{- '\n' + message.content }}
+ {%- endif %}
+ {%- for tool_call in message.tool_calls %}
+ {%- if tool_call.function is defined %}
+ {%- set tool_call = tool_call.function %}
+ {%- endif %}
+ {{- '\n\n{"name": "' }}
+ {{- tool_call.name }}
+ {{- '", "arguments": ' }}
+ {{- tool_call.arguments | tojson }}
+ {{- '}\n' }}
+ {%- endfor %}
+ {{- '<|im_end|>\n' }}
+ {%- elif message.role == "tool" %}
+ {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+ {{- '<|im_start|>user' }}
+ {%- endif %}
+ {{- '\n\n' }}
+ {{- message.content }}
+ {{- '\n' }}
+ {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+ {{- '<|im_end|>\n' }}
+ {%- endif %}
+ {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+ {{- '<|im_start|>assistant\n' }}
+{%- endif %}
diff --git a/qwen2.5-3b-instruct/checkpoint-71/config.json b/qwen2.5-3b-instruct/checkpoint-71/config.json
new file mode 100644
index 0000000000000000000000000000000000000000..4814292863d1e5aa2f4fda1db18f4082cd61aa93
--- /dev/null
+++ b/qwen2.5-3b-instruct/checkpoint-71/config.json
@@ -0,0 +1,69 @@
+{
+ "architectures": [
+ "Qwen2ForCausalLM"
+ ],
+ "attention_dropout": 0.0,
+ "bos_token_id": null,
+ "dtype": "float32",
+ "eos_token_id": 151645,
+ "hidden_act": "silu",
+ "hidden_size": 2048,
+ "initializer_range": 0.02,
+ "intermediate_size": 11008,
+ "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",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention"
+ ],
+ "max_position_embeddings": 32768,
+ "max_window_layers": 70,
+ "model_type": "qwen2",
+ "num_attention_heads": 16,
+ "num_hidden_layers": 36,
+ "num_key_value_heads": 2,
+ "pad_token_id": 151643,
+ "rms_norm_eps": 1e-06,
+ "rope_parameters": {
+ "rope_theta": 1000000.0,
+ "rope_type": "default"
+ },
+ "sliding_window": null,
+ "tie_word_embeddings": true,
+ "transformers_version": "5.2.0",
+ "use_cache": false,
+ "use_sliding_window": false,
+ "vocab_size": 151936
+}
diff --git a/qwen2.5-3b-instruct/checkpoint-71/generation_config.json b/qwen2.5-3b-instruct/checkpoint-71/generation_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..47205fbf3386a215e7c2b9d62df6576f49b77e7b
--- /dev/null
+++ b/qwen2.5-3b-instruct/checkpoint-71/generation_config.json
@@ -0,0 +1,13 @@
+{
+ "do_sample": true,
+ "eos_token_id": [
+ 151645,
+ 151643
+ ],
+ "pad_token_id": 151643,
+ "repetition_penalty": 1.05,
+ "temperature": 0.7,
+ "top_k": 20,
+ "top_p": 0.8,
+ "transformers_version": "5.2.0"
+}
diff --git a/qwen2.5-3b-instruct/checkpoint-71/model.safetensors b/qwen2.5-3b-instruct/checkpoint-71/model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..048d6c8710fbf1866d07cb849cae02ce8df08cdf
--- /dev/null
+++ b/qwen2.5-3b-instruct/checkpoint-71/model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:460aaa82fb068d71ab3eadbede8571df310bb265f89edf5c5ff5e94f055f3b4f
+size 12343804296
diff --git a/qwen2.5-3b-instruct/checkpoint-71/optimizer.pt b/qwen2.5-3b-instruct/checkpoint-71/optimizer.pt
new file mode 100644
index 0000000000000000000000000000000000000000..bf24bfe43a31df6b62e83b124f38870b3baea657
--- /dev/null
+++ b/qwen2.5-3b-instruct/checkpoint-71/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:06cc4269b1425a73aa69d87e67116cb9b9e3c657e7e9532b78b8790639f32a7d
+size 24687895753
diff --git a/qwen2.5-3b-instruct/checkpoint-71/rng_state_0.pth b/qwen2.5-3b-instruct/checkpoint-71/rng_state_0.pth
new file mode 100644
index 0000000000000000000000000000000000000000..870021e3fa5ac35c2f711adf0c93a556ab4842da
--- /dev/null
+++ b/qwen2.5-3b-instruct/checkpoint-71/rng_state_0.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:95e5fc2074c0df31522a514f862c86cb00d71c946a7f15cc9ec0e53a69fb28a7
+size 14917
diff --git a/qwen2.5-3b-instruct/checkpoint-71/rng_state_1.pth b/qwen2.5-3b-instruct/checkpoint-71/rng_state_1.pth
new file mode 100644
index 0000000000000000000000000000000000000000..21f20da5eb1da017f08aaa88bd19cf24d40e3fbf
--- /dev/null
+++ b/qwen2.5-3b-instruct/checkpoint-71/rng_state_1.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0e7153eae67b6c9232a41bc996a2bf5b83229b8c7230d61911ac0fd40e64154e
+size 14917
diff --git a/qwen2.5-3b-instruct/checkpoint-71/scheduler.pt b/qwen2.5-3b-instruct/checkpoint-71/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..c3be1d29bde40f225713ff5084d3d9d69858fcfd
--- /dev/null
+++ b/qwen2.5-3b-instruct/checkpoint-71/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9c96c28f12671e2f5db493417daefb3ff0b1bc36602843e0d7ef6df3649c27b3
+size 1465
diff --git a/qwen2.5-3b-instruct/checkpoint-71/tokenizer.json b/qwen2.5-3b-instruct/checkpoint-71/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..34510ff0037cd50428af467a17ead5a96140a32c
--- /dev/null
+++ b/qwen2.5-3b-instruct/checkpoint-71/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
+size 11421892
diff --git a/qwen2.5-3b-instruct/checkpoint-71/tokenizer_config.json b/qwen2.5-3b-instruct/checkpoint-71/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..145e2c7a68b109ae0253dda29fc2e66cb01b4b85
--- /dev/null
+++ b/qwen2.5-3b-instruct/checkpoint-71/tokenizer_config.json
@@ -0,0 +1,30 @@
+{
+ "add_prefix_space": false,
+ "backend": "tokenizers",
+ "bos_token": null,
+ "clean_up_tokenization_spaces": false,
+ "eos_token": "<|im_end|>",
+ "errors": "replace",
+ "extra_special_tokens": [
+ "<|im_start|>",
+ "<|im_end|>",
+ "<|object_ref_start|>",
+ "<|object_ref_end|>",
+ "<|box_start|>",
+ "<|box_end|>",
+ "<|quad_start|>",
+ "<|quad_end|>",
+ "<|vision_start|>",
+ "<|vision_end|>",
+ "<|vision_pad|>",
+ "<|image_pad|>",
+ "<|video_pad|>"
+ ],
+ "is_local": false,
+ "model_max_length": 131072,
+ "pad_token": "<|endoftext|>",
+ "padding_side": "right",
+ "split_special_tokens": false,
+ "tokenizer_class": "Qwen2Tokenizer",
+ "unk_token": null
+}
diff --git a/qwen2.5-3b-instruct/checkpoint-71/trainer_state.json b/qwen2.5-3b-instruct/checkpoint-71/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..876eae4b217247a77cdfea80554924101a2d27c9
--- /dev/null
+++ b/qwen2.5-3b-instruct/checkpoint-71/trainer_state.json
@@ -0,0 +1,539 @@
+{
+ "best_global_step": null,
+ "best_metric": null,
+ "best_model_checkpoint": null,
+ "epoch": 1.0,
+ "eval_steps": 50,
+ "global_step": 71,
+ "is_hyper_param_search": false,
+ "is_local_process_zero": true,
+ "is_world_process_zero": true,
+ "log_history": [
+ {
+ "epoch": 0.014222222222222223,
+ "grad_norm": 0.253393292427063,
+ "learning_rate": 0.0,
+ "loss": 1.19514799118042,
+ "step": 1
+ },
+ {
+ "epoch": 0.028444444444444446,
+ "grad_norm": 0.2720104455947876,
+ "learning_rate": 4.5454545454545457e-07,
+ "loss": 1.2398133277893066,
+ "step": 2
+ },
+ {
+ "epoch": 0.042666666666666665,
+ "grad_norm": 0.22159169614315033,
+ "learning_rate": 9.090909090909091e-07,
+ "loss": 0.9782462120056152,
+ "step": 3
+ },
+ {
+ "epoch": 0.05688888888888889,
+ "grad_norm": 0.17487271130084991,
+ "learning_rate": 1.3636363636363636e-06,
+ "loss": 0.8072947263717651,
+ "step": 4
+ },
+ {
+ "epoch": 0.07111111111111111,
+ "grad_norm": 0.08528934419155121,
+ "learning_rate": 1.8181818181818183e-06,
+ "loss": 0.5301008224487305,
+ "step": 5
+ },
+ {
+ "epoch": 0.08533333333333333,
+ "grad_norm": 0.08074624091386795,
+ "learning_rate": 2.2727272727272728e-06,
+ "loss": 0.6367583274841309,
+ "step": 6
+ },
+ {
+ "epoch": 0.09955555555555555,
+ "grad_norm": 0.05584918335080147,
+ "learning_rate": 2.7272727272727272e-06,
+ "loss": 0.6089775562286377,
+ "step": 7
+ },
+ {
+ "epoch": 0.11377777777777778,
+ "grad_norm": 0.032228983938694,
+ "learning_rate": 3.181818181818182e-06,
+ "loss": 0.5260579586029053,
+ "step": 8
+ },
+ {
+ "epoch": 0.128,
+ "grad_norm": 0.026230908930301666,
+ "learning_rate": 3.6363636363636366e-06,
+ "loss": 0.5446298122406006,
+ "step": 9
+ },
+ {
+ "epoch": 0.14222222222222222,
+ "grad_norm": 0.019648801535367966,
+ "learning_rate": 4.0909090909090915e-06,
+ "loss": 0.5338459610939026,
+ "step": 10
+ },
+ {
+ "epoch": 0.15644444444444444,
+ "grad_norm": 0.014007463119924068,
+ "learning_rate": 4.5454545454545455e-06,
+ "loss": 0.45045405626296997,
+ "step": 11
+ },
+ {
+ "epoch": 0.17066666666666666,
+ "grad_norm": 0.013295263983309269,
+ "learning_rate": 5e-06,
+ "loss": 0.49363279342651367,
+ "step": 12
+ },
+ {
+ "epoch": 0.18488888888888888,
+ "grad_norm": 0.01435786858201027,
+ "learning_rate": 5.4545454545454545e-06,
+ "loss": 0.48757249116897583,
+ "step": 13
+ },
+ {
+ "epoch": 0.1991111111111111,
+ "grad_norm": 0.015238163061439991,
+ "learning_rate": 5.90909090909091e-06,
+ "loss": 0.5002430081367493,
+ "step": 14
+ },
+ {
+ "epoch": 0.21333333333333335,
+ "grad_norm": 0.014418123289942741,
+ "learning_rate": 6.363636363636364e-06,
+ "loss": 0.4964277148246765,
+ "step": 15
+ },
+ {
+ "epoch": 0.22755555555555557,
+ "grad_norm": 0.014306087978184223,
+ "learning_rate": 6.818181818181818e-06,
+ "loss": 0.4818333387374878,
+ "step": 16
+ },
+ {
+ "epoch": 0.24177777777777779,
+ "grad_norm": 0.009899069555103779,
+ "learning_rate": 7.272727272727273e-06,
+ "loss": 0.4609573483467102,
+ "step": 17
+ },
+ {
+ "epoch": 0.256,
+ "grad_norm": 0.010643995366990566,
+ "learning_rate": 7.727272727272727e-06,
+ "loss": 0.47600606083869934,
+ "step": 18
+ },
+ {
+ "epoch": 0.2702222222222222,
+ "grad_norm": 0.012857547029852867,
+ "learning_rate": 8.181818181818183e-06,
+ "loss": 0.4962907135486603,
+ "step": 19
+ },
+ {
+ "epoch": 0.28444444444444444,
+ "grad_norm": 0.01670556701719761,
+ "learning_rate": 8.636363636363637e-06,
+ "loss": 0.5028363466262817,
+ "step": 20
+ },
+ {
+ "epoch": 0.2986666666666667,
+ "grad_norm": 0.010331347584724426,
+ "learning_rate": 9.090909090909091e-06,
+ "loss": 0.44582226872444153,
+ "step": 21
+ },
+ {
+ "epoch": 0.3128888888888889,
+ "grad_norm": 0.010525904595851898,
+ "learning_rate": 9.545454545454547e-06,
+ "loss": 0.4757377505302429,
+ "step": 22
+ },
+ {
+ "epoch": 0.32711111111111113,
+ "grad_norm": 0.009544270113110542,
+ "learning_rate": 1e-05,
+ "loss": 0.4707476496696472,
+ "step": 23
+ },
+ {
+ "epoch": 0.3413333333333333,
+ "grad_norm": 0.009730499237775803,
+ "learning_rate": 9.999323662872998e-06,
+ "loss": 0.49240708351135254,
+ "step": 24
+ },
+ {
+ "epoch": 0.35555555555555557,
+ "grad_norm": 0.008985932916402817,
+ "learning_rate": 9.99729483446475e-06,
+ "loss": 0.450992614030838,
+ "step": 25
+ },
+ {
+ "epoch": 0.36977777777777776,
+ "grad_norm": 0.01072725746780634,
+ "learning_rate": 9.993914063644053e-06,
+ "loss": 0.4514142870903015,
+ "step": 26
+ },
+ {
+ "epoch": 0.384,
+ "grad_norm": 0.01058228686451912,
+ "learning_rate": 9.989182265027232e-06,
+ "loss": 0.44021710753440857,
+ "step": 27
+ },
+ {
+ "epoch": 0.3982222222222222,
+ "grad_norm": 0.009333685040473938,
+ "learning_rate": 9.98310071873072e-06,
+ "loss": 0.4491778016090393,
+ "step": 28
+ },
+ {
+ "epoch": 0.41244444444444445,
+ "grad_norm": 0.009250632487237453,
+ "learning_rate": 9.975671070024741e-06,
+ "loss": 0.4285920560359955,
+ "step": 29
+ },
+ {
+ "epoch": 0.4266666666666667,
+ "grad_norm": 0.010210932232439518,
+ "learning_rate": 9.966895328888195e-06,
+ "loss": 0.5484102368354797,
+ "step": 30
+ },
+ {
+ "epoch": 0.4408888888888889,
+ "grad_norm": 0.008848125115036964,
+ "learning_rate": 9.956775869464901e-06,
+ "loss": 0.4242202043533325,
+ "step": 31
+ },
+ {
+ "epoch": 0.45511111111111113,
+ "grad_norm": 0.00905672088265419,
+ "learning_rate": 9.945315429421307e-06,
+ "loss": 0.4533390402793884,
+ "step": 32
+ },
+ {
+ "epoch": 0.4693333333333333,
+ "grad_norm": 0.008254914544522762,
+ "learning_rate": 9.932517109205849e-06,
+ "loss": 0.4243648052215576,
+ "step": 33
+ },
+ {
+ "epoch": 0.48355555555555557,
+ "grad_norm": 0.009192033670842648,
+ "learning_rate": 9.918384371210178e-06,
+ "loss": 0.5120309591293335,
+ "step": 34
+ },
+ {
+ "epoch": 0.49777777777777776,
+ "grad_norm": 0.009636987932026386,
+ "learning_rate": 9.902921038832456e-06,
+ "loss": 0.4871100187301636,
+ "step": 35
+ },
+ {
+ "epoch": 0.512,
+ "grad_norm": 0.008572430349886417,
+ "learning_rate": 9.886131295443003e-06,
+ "loss": 0.46277713775634766,
+ "step": 36
+ },
+ {
+ "epoch": 0.5262222222222223,
+ "grad_norm": 0.008587818592786789,
+ "learning_rate": 9.868019683252543e-06,
+ "loss": 0.45996755361557007,
+ "step": 37
+ },
+ {
+ "epoch": 0.5404444444444444,
+ "grad_norm": 0.010775095783174038,
+ "learning_rate": 9.848591102083375e-06,
+ "loss": 0.47378021478652954,
+ "step": 38
+ },
+ {
+ "epoch": 0.5546666666666666,
+ "grad_norm": 0.00995537731796503,
+ "learning_rate": 9.82785080804381e-06,
+ "loss": 0.4684975743293762,
+ "step": 39
+ },
+ {
+ "epoch": 0.5688888888888889,
+ "grad_norm": 0.009079818613827229,
+ "learning_rate": 9.805804412106197e-06,
+ "loss": 0.5784010887145996,
+ "step": 40
+ },
+ {
+ "epoch": 0.5831111111111111,
+ "grad_norm": 0.008139098063111305,
+ "learning_rate": 9.782457878588977e-06,
+ "loss": 0.46650874614715576,
+ "step": 41
+ },
+ {
+ "epoch": 0.5973333333333334,
+ "grad_norm": 0.007847482338547707,
+ "learning_rate": 9.75781752354311e-06,
+ "loss": 0.42804157733917236,
+ "step": 42
+ },
+ {
+ "epoch": 0.6115555555555555,
+ "grad_norm": 0.008446140214800835,
+ "learning_rate": 9.731890013043367e-06,
+ "loss": 0.48224180936813354,
+ "step": 43
+ },
+ {
+ "epoch": 0.6257777777777778,
+ "grad_norm": 0.008632696233689785,
+ "learning_rate": 9.704682361384941e-06,
+ "loss": 0.4561079144477844,
+ "step": 44
+ },
+ {
+ "epoch": 0.64,
+ "grad_norm": 0.008637293241918087,
+ "learning_rate": 9.676201929185809e-06,
+ "loss": 0.4007866382598877,
+ "step": 45
+ },
+ {
+ "epoch": 0.6542222222222223,
+ "grad_norm": 0.008761093020439148,
+ "learning_rate": 9.646456421395447e-06,
+ "loss": 0.4587845802307129,
+ "step": 46
+ },
+ {
+ "epoch": 0.6684444444444444,
+ "grad_norm": 0.008726399391889572,
+ "learning_rate": 9.615453885210368e-06,
+ "loss": 0.4571158289909363,
+ "step": 47
+ },
+ {
+ "epoch": 0.6826666666666666,
+ "grad_norm": 0.008282204158604145,
+ "learning_rate": 9.583202707897075e-06,
+ "loss": 0.4724605679512024,
+ "step": 48
+ },
+ {
+ "epoch": 0.6968888888888889,
+ "grad_norm": 0.008231760933995247,
+ "learning_rate": 9.549711614523007e-06,
+ "loss": 0.4600468575954437,
+ "step": 49
+ },
+ {
+ "epoch": 0.7111111111111111,
+ "grad_norm": 0.00751511100679636,
+ "learning_rate": 9.514989665596114e-06,
+ "loss": 0.4377005994319916,
+ "step": 50
+ },
+ {
+ "epoch": 0.7111111111111111,
+ "eval_loss": 0.4685295820236206,
+ "eval_runtime": 5.4716,
+ "eval_samples_per_second": 36.553,
+ "eval_steps_per_second": 18.276,
+ "step": 50
+ },
+ {
+ "epoch": 0.7253333333333334,
+ "grad_norm": 0.008261662907898426,
+ "learning_rate": 9.479046254613673e-06,
+ "loss": 0.4415626525878906,
+ "step": 51
+ },
+ {
+ "epoch": 0.7395555555555555,
+ "grad_norm": 0.007395230233669281,
+ "learning_rate": 9.441891105521005e-06,
+ "loss": 0.42561036348342896,
+ "step": 52
+ },
+ {
+ "epoch": 0.7537777777777778,
+ "grad_norm": 0.008451394736766815,
+ "learning_rate": 9.40353427008083e-06,
+ "loss": 0.4630648195743561,
+ "step": 53
+ },
+ {
+ "epoch": 0.768,
+ "grad_norm": 0.007475187536329031,
+ "learning_rate": 9.3639861251539e-06,
+ "loss": 0.43418699502944946,
+ "step": 54
+ },
+ {
+ "epoch": 0.7822222222222223,
+ "grad_norm": 0.00859637837857008,
+ "learning_rate": 9.323257369891702e-06,
+ "loss": 0.47063255310058594,
+ "step": 55
+ },
+ {
+ "epoch": 0.7964444444444444,
+ "grad_norm": 0.007597765419632196,
+ "learning_rate": 9.281359022841966e-06,
+ "loss": 0.41920918226242065,
+ "step": 56
+ },
+ {
+ "epoch": 0.8106666666666666,
+ "grad_norm": 0.018636908382177353,
+ "learning_rate": 9.238302418967757e-06,
+ "loss": 0.4757845401763916,
+ "step": 57
+ },
+ {
+ "epoch": 0.8248888888888889,
+ "grad_norm": 0.008423218503594398,
+ "learning_rate": 9.194099206580981e-06,
+ "loss": 0.5198032855987549,
+ "step": 58
+ },
+ {
+ "epoch": 0.8391111111111111,
+ "grad_norm": 0.008149159140884876,
+ "learning_rate": 9.14876134419111e-06,
+ "loss": 0.46035224199295044,
+ "step": 59
+ },
+ {
+ "epoch": 0.8533333333333334,
+ "grad_norm": 0.00819750688970089,
+ "learning_rate": 9.102301097269974e-06,
+ "loss": 0.4945712983608246,
+ "step": 60
+ },
+ {
+ "epoch": 0.8675555555555555,
+ "grad_norm": 0.00855066254734993,
+ "learning_rate": 9.05473103493355e-06,
+ "loss": 0.4513372778892517,
+ "step": 61
+ },
+ {
+ "epoch": 0.8817777777777778,
+ "grad_norm": 0.008660132065415382,
+ "learning_rate": 9.006064026541549e-06,
+ "loss": 0.49865084886550903,
+ "step": 62
+ },
+ {
+ "epoch": 0.896,
+ "grad_norm": 0.007741494104266167,
+ "learning_rate": 8.956313238215824e-06,
+ "loss": 0.4319220185279846,
+ "step": 63
+ },
+ {
+ "epoch": 0.9102222222222223,
+ "grad_norm": 0.007900395430624485,
+ "learning_rate": 8.905492129278478e-06,
+ "loss": 0.43760013580322266,
+ "step": 64
+ },
+ {
+ "epoch": 0.9244444444444444,
+ "grad_norm": 0.007346940692514181,
+ "learning_rate": 8.85361444861063e-06,
+ "loss": 0.41593095660209656,
+ "step": 65
+ },
+ {
+ "epoch": 0.9386666666666666,
+ "grad_norm": 0.00902250874787569,
+ "learning_rate": 8.800694230932885e-06,
+ "loss": 0.48464590311050415,
+ "step": 66
+ },
+ {
+ "epoch": 0.9528888888888889,
+ "grad_norm": 0.007672084029763937,
+ "learning_rate": 8.74674579300843e-06,
+ "loss": 0.41884416341781616,
+ "step": 67
+ },
+ {
+ "epoch": 0.9671111111111111,
+ "grad_norm": 0.008009645156562328,
+ "learning_rate": 8.691783729769874e-06,
+ "loss": 0.40650737285614014,
+ "step": 68
+ },
+ {
+ "epoch": 0.9813333333333333,
+ "grad_norm": 0.008077424019575119,
+ "learning_rate": 8.635822910370793e-06,
+ "loss": 0.4726918637752533,
+ "step": 69
+ },
+ {
+ "epoch": 0.9955555555555555,
+ "grad_norm": 0.007576521951705217,
+ "learning_rate": 8.578878474163115e-06,
+ "loss": 0.4404842257499695,
+ "step": 70
+ },
+ {
+ "epoch": 1.0,
+ "grad_norm": 0.012638773769140244,
+ "learning_rate": 8.520965826601394e-06,
+ "loss": 0.42633137106895447,
+ "step": 71
+ }
+ ],
+ "logging_steps": 1,
+ "max_steps": 213,
+ "num_input_tokens_seen": 0,
+ "num_train_epochs": 3,
+ "save_steps": 500,
+ "stateful_callbacks": {
+ "TrainerControl": {
+ "args": {
+ "should_epoch_stop": false,
+ "should_evaluate": false,
+ "should_log": false,
+ "should_save": true,
+ "should_training_stop": false
+ },
+ "attributes": {}
+ }
+ },
+ "total_flos": 3.4361599565234176e+16,
+ "train_batch_size": 1,
+ "trial_name": null,
+ "trial_params": null
+}
diff --git a/qwen2.5-3b-instruct/checkpoint-71/training_args.bin b/qwen2.5-3b-instruct/checkpoint-71/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..bd95d1f874b25af121b3c8254bd05a3eb95ffd5f
--- /dev/null
+++ b/qwen2.5-3b-instruct/checkpoint-71/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c4f1847ec07466eeb896ef40e20d9241e19882e62efb61a14f1e7c6a96737fa0
+size 5649
diff --git a/qwen2.5-3b-instruct/config.json b/qwen2.5-3b-instruct/config.json
new file mode 100644
index 0000000000000000000000000000000000000000..4814292863d1e5aa2f4fda1db18f4082cd61aa93
--- /dev/null
+++ b/qwen2.5-3b-instruct/config.json
@@ -0,0 +1,69 @@
+{
+ "architectures": [
+ "Qwen2ForCausalLM"
+ ],
+ "attention_dropout": 0.0,
+ "bos_token_id": null,
+ "dtype": "float32",
+ "eos_token_id": 151645,
+ "hidden_act": "silu",
+ "hidden_size": 2048,
+ "initializer_range": 0.02,
+ "intermediate_size": 11008,
+ "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",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention"
+ ],
+ "max_position_embeddings": 32768,
+ "max_window_layers": 70,
+ "model_type": "qwen2",
+ "num_attention_heads": 16,
+ "num_hidden_layers": 36,
+ "num_key_value_heads": 2,
+ "pad_token_id": 151643,
+ "rms_norm_eps": 1e-06,
+ "rope_parameters": {
+ "rope_theta": 1000000.0,
+ "rope_type": "default"
+ },
+ "sliding_window": null,
+ "tie_word_embeddings": true,
+ "transformers_version": "5.2.0",
+ "use_cache": false,
+ "use_sliding_window": false,
+ "vocab_size": 151936
+}
diff --git a/qwen2.5-3b-instruct/eval_results.json b/qwen2.5-3b-instruct/eval_results.json
new file mode 100644
index 0000000000000000000000000000000000000000..9392d6b20aa7ead5a19def379d210e531ca8b099
--- /dev/null
+++ b/qwen2.5-3b-instruct/eval_results.json
@@ -0,0 +1,7 @@
+{
+ "epoch": 3.0,
+ "eval_loss": 0.47346630692481995,
+ "eval_runtime": 5.3972,
+ "eval_samples_per_second": 37.056,
+ "eval_steps_per_second": 18.528
+}
\ No newline at end of file
diff --git a/qwen2.5-3b-instruct/generation_config.json b/qwen2.5-3b-instruct/generation_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..47205fbf3386a215e7c2b9d62df6576f49b77e7b
--- /dev/null
+++ b/qwen2.5-3b-instruct/generation_config.json
@@ -0,0 +1,13 @@
+{
+ "do_sample": true,
+ "eos_token_id": [
+ 151645,
+ 151643
+ ],
+ "pad_token_id": 151643,
+ "repetition_penalty": 1.05,
+ "temperature": 0.7,
+ "top_k": 20,
+ "top_p": 0.8,
+ "transformers_version": "5.2.0"
+}
diff --git a/qwen2.5-3b-instruct/model.safetensors b/qwen2.5-3b-instruct/model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..49702ec9826e3cd7c26c2b828a2fab8d09ce50c2
--- /dev/null
+++ b/qwen2.5-3b-instruct/model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:19b1d375c807d8900f5c4a64fceb62f4bd9a2bf032bd23ad921474f05998e666
+size 12343804296
diff --git a/qwen2.5-3b-instruct/tokenizer.json b/qwen2.5-3b-instruct/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..34510ff0037cd50428af467a17ead5a96140a32c
--- /dev/null
+++ b/qwen2.5-3b-instruct/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
+size 11421892
diff --git a/qwen2.5-3b-instruct/tokenizer_config.json b/qwen2.5-3b-instruct/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..145e2c7a68b109ae0253dda29fc2e66cb01b4b85
--- /dev/null
+++ b/qwen2.5-3b-instruct/tokenizer_config.json
@@ -0,0 +1,30 @@
+{
+ "add_prefix_space": false,
+ "backend": "tokenizers",
+ "bos_token": null,
+ "clean_up_tokenization_spaces": false,
+ "eos_token": "<|im_end|>",
+ "errors": "replace",
+ "extra_special_tokens": [
+ "<|im_start|>",
+ "<|im_end|>",
+ "<|object_ref_start|>",
+ "<|object_ref_end|>",
+ "<|box_start|>",
+ "<|box_end|>",
+ "<|quad_start|>",
+ "<|quad_end|>",
+ "<|vision_start|>",
+ "<|vision_end|>",
+ "<|vision_pad|>",
+ "<|image_pad|>",
+ "<|video_pad|>"
+ ],
+ "is_local": false,
+ "model_max_length": 131072,
+ "pad_token": "<|endoftext|>",
+ "padding_side": "right",
+ "split_special_tokens": false,
+ "tokenizer_class": "Qwen2Tokenizer",
+ "unk_token": null
+}
diff --git a/qwen2.5-3b-instruct/train_results.json b/qwen2.5-3b-instruct/train_results.json
new file mode 100644
index 0000000000000000000000000000000000000000..1cf4fcdebbd4f6fb0e9fa1045f65cd044ae2dd5a
--- /dev/null
+++ b/qwen2.5-3b-instruct/train_results.json
@@ -0,0 +1,8 @@
+{
+ "epoch": 3.0,
+ "total_flos": 1.0308479847759872e+17,
+ "train_loss": 0.4208552807989255,
+ "train_runtime": 5691.1072,
+ "train_samples_per_second": 9.488,
+ "train_steps_per_second": 0.037
+}
\ No newline at end of file
diff --git a/qwen2.5-3b-instruct/trainer_log.jsonl b/qwen2.5-3b-instruct/trainer_log.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..b91abf21dea109ec2eb5157d90febdfb4d6e9a5c
--- /dev/null
+++ b/qwen2.5-3b-instruct/trainer_log.jsonl
@@ -0,0 +1,218 @@
+{"current_steps": 1, "total_steps": 213, "loss": 1.19514799118042, "lr": 0.0, "epoch": 0.014222222222222223, "percentage": 0.47, "elapsed_time": "0:00:39", "remaining_time": "2:20:58"}
+{"current_steps": 2, "total_steps": 213, "loss": 1.2398133277893066, "lr": 4.5454545454545457e-07, "epoch": 0.028444444444444446, "percentage": 0.94, "elapsed_time": "0:01:18", "remaining_time": "2:17:27"}
+{"current_steps": 3, "total_steps": 213, "loss": 0.9782462120056152, "lr": 9.090909090909091e-07, "epoch": 0.042666666666666665, "percentage": 1.41, "elapsed_time": "0:01:56", "remaining_time": "2:15:44"}
+{"current_steps": 4, "total_steps": 213, "loss": 0.8072947263717651, "lr": 1.3636363636363636e-06, "epoch": 0.05688888888888889, "percentage": 1.88, "elapsed_time": "0:02:24", "remaining_time": "2:06:00"}
+{"current_steps": 5, "total_steps": 213, "loss": 0.5301008224487305, "lr": 1.8181818181818183e-06, "epoch": 0.07111111111111111, "percentage": 2.35, "elapsed_time": "0:02:49", "remaining_time": "1:57:35"}
+{"current_steps": 6, "total_steps": 213, "loss": 0.6367583274841309, "lr": 2.2727272727272728e-06, "epoch": 0.08533333333333333, "percentage": 2.82, "elapsed_time": "0:03:14", "remaining_time": "1:51:42"}
+{"current_steps": 7, "total_steps": 213, "loss": 0.6089775562286377, "lr": 2.7272727272727272e-06, "epoch": 0.09955555555555555, "percentage": 3.29, "elapsed_time": "0:03:38", "remaining_time": "1:47:15"}
+{"current_steps": 8, "total_steps": 213, "loss": 0.5260579586029053, "lr": 3.181818181818182e-06, "epoch": 0.11377777777777778, "percentage": 3.76, "elapsed_time": "0:04:03", "remaining_time": "1:43:55"}
+{"current_steps": 9, "total_steps": 213, "loss": 0.5446298122406006, "lr": 3.6363636363636366e-06, "epoch": 0.128, "percentage": 4.23, "elapsed_time": "0:04:28", "remaining_time": "1:41:18"}
+{"current_steps": 10, "total_steps": 213, "loss": 0.5338459610939026, "lr": 4.0909090909090915e-06, "epoch": 0.14222222222222222, "percentage": 4.69, "elapsed_time": "0:04:52", "remaining_time": "1:39:07"}
+{"current_steps": 11, "total_steps": 213, "loss": 0.45045405626296997, "lr": 4.5454545454545455e-06, "epoch": 0.15644444444444444, "percentage": 5.16, "elapsed_time": "0:05:17", "remaining_time": "1:37:17"}
+{"current_steps": 12, "total_steps": 213, "loss": 0.49363279342651367, "lr": 5e-06, "epoch": 0.17066666666666666, "percentage": 5.63, "elapsed_time": "0:05:42", "remaining_time": "1:35:39"}
+{"current_steps": 13, "total_steps": 213, "loss": 0.48757249116897583, "lr": 5.4545454545454545e-06, "epoch": 0.18488888888888888, "percentage": 6.1, "elapsed_time": "0:06:07", "remaining_time": "1:34:18"}
+{"current_steps": 14, "total_steps": 213, "loss": 0.5002430081367493, "lr": 5.90909090909091e-06, "epoch": 0.1991111111111111, "percentage": 6.57, "elapsed_time": "0:06:32", "remaining_time": "1:33:01"}
+{"current_steps": 15, "total_steps": 213, "loss": 0.4964277148246765, "lr": 6.363636363636364e-06, "epoch": 0.21333333333333335, "percentage": 7.04, "elapsed_time": "0:06:57", "remaining_time": "1:31:47"}
+{"current_steps": 16, "total_steps": 213, "loss": 0.4818333387374878, "lr": 6.818181818181818e-06, "epoch": 0.22755555555555557, "percentage": 7.51, "elapsed_time": "0:07:22", "remaining_time": "1:30:42"}
+{"current_steps": 17, "total_steps": 213, "loss": 0.4609573483467102, "lr": 7.272727272727273e-06, "epoch": 0.24177777777777779, "percentage": 7.98, "elapsed_time": "0:07:47", "remaining_time": "1:29:45"}
+{"current_steps": 18, "total_steps": 213, "loss": 0.47600606083869934, "lr": 7.727272727272727e-06, "epoch": 0.256, "percentage": 8.45, "elapsed_time": "0:08:11", "remaining_time": "1:28:48"}
+{"current_steps": 19, "total_steps": 213, "loss": 0.4962907135486603, "lr": 8.181818181818183e-06, "epoch": 0.2702222222222222, "percentage": 8.92, "elapsed_time": "0:08:36", "remaining_time": "1:27:57"}
+{"current_steps": 20, "total_steps": 213, "loss": 0.5028363466262817, "lr": 8.636363636363637e-06, "epoch": 0.28444444444444444, "percentage": 9.39, "elapsed_time": "0:09:01", "remaining_time": "1:27:08"}
+{"current_steps": 21, "total_steps": 213, "loss": 0.44582226872444153, "lr": 9.090909090909091e-06, "epoch": 0.2986666666666667, "percentage": 9.86, "elapsed_time": "0:09:27", "remaining_time": "1:26:24"}
+{"current_steps": 22, "total_steps": 213, "loss": 0.4757377505302429, "lr": 9.545454545454547e-06, "epoch": 0.3128888888888889, "percentage": 10.33, "elapsed_time": "0:09:52", "remaining_time": "1:25:40"}
+{"current_steps": 23, "total_steps": 213, "loss": 0.4707476496696472, "lr": 1e-05, "epoch": 0.32711111111111113, "percentage": 10.8, "elapsed_time": "0:10:16", "remaining_time": "1:24:55"}
+{"current_steps": 24, "total_steps": 213, "loss": 0.49240708351135254, "lr": 9.999323662872998e-06, "epoch": 0.3413333333333333, "percentage": 11.27, "elapsed_time": "0:10:41", "remaining_time": "1:24:12"}
+{"current_steps": 25, "total_steps": 213, "loss": 0.450992614030838, "lr": 9.99729483446475e-06, "epoch": 0.35555555555555557, "percentage": 11.74, "elapsed_time": "0:11:06", "remaining_time": "1:23:33"}
+{"current_steps": 26, "total_steps": 213, "loss": 0.4514142870903015, "lr": 9.993914063644053e-06, "epoch": 0.36977777777777776, "percentage": 12.21, "elapsed_time": "0:11:31", "remaining_time": "1:22:53"}
+{"current_steps": 27, "total_steps": 213, "loss": 0.44021710753440857, "lr": 9.989182265027232e-06, "epoch": 0.384, "percentage": 12.68, "elapsed_time": "0:11:56", "remaining_time": "1:22:16"}
+{"current_steps": 28, "total_steps": 213, "loss": 0.4491778016090393, "lr": 9.98310071873072e-06, "epoch": 0.3982222222222222, "percentage": 13.15, "elapsed_time": "0:12:21", "remaining_time": "1:21:39"}
+{"current_steps": 29, "total_steps": 213, "loss": 0.4285920560359955, "lr": 9.975671070024741e-06, "epoch": 0.41244444444444445, "percentage": 13.62, "elapsed_time": "0:12:46", "remaining_time": "1:21:05"}
+{"current_steps": 30, "total_steps": 213, "loss": 0.5484102368354797, "lr": 9.966895328888195e-06, "epoch": 0.4266666666666667, "percentage": 14.08, "elapsed_time": "0:13:12", "remaining_time": "1:20:31"}
+{"current_steps": 31, "total_steps": 213, "loss": 0.4242202043533325, "lr": 9.956775869464901e-06, "epoch": 0.4408888888888889, "percentage": 14.55, "elapsed_time": "0:13:37", "remaining_time": "1:19:57"}
+{"current_steps": 32, "total_steps": 213, "loss": 0.4533390402793884, "lr": 9.945315429421307e-06, "epoch": 0.45511111111111113, "percentage": 15.02, "elapsed_time": "0:14:02", "remaining_time": "1:19:23"}
+{"current_steps": 33, "total_steps": 213, "loss": 0.4243648052215576, "lr": 9.932517109205849e-06, "epoch": 0.4693333333333333, "percentage": 15.49, "elapsed_time": "0:14:27", "remaining_time": "1:18:49"}
+{"current_steps": 34, "total_steps": 213, "loss": 0.5120309591293335, "lr": 9.918384371210178e-06, "epoch": 0.48355555555555557, "percentage": 15.96, "elapsed_time": "0:14:52", "remaining_time": "1:18:16"}
+{"current_steps": 35, "total_steps": 213, "loss": 0.4871100187301636, "lr": 9.902921038832456e-06, "epoch": 0.49777777777777776, "percentage": 16.43, "elapsed_time": "0:15:16", "remaining_time": "1:17:42"}
+{"current_steps": 36, "total_steps": 213, "loss": 0.46277713775634766, "lr": 9.886131295443003e-06, "epoch": 0.512, "percentage": 16.9, "elapsed_time": "0:15:41", "remaining_time": "1:17:08"}
+{"current_steps": 37, "total_steps": 213, "loss": 0.45996755361557007, "lr": 9.868019683252543e-06, "epoch": 0.5262222222222223, "percentage": 17.37, "elapsed_time": "0:16:06", "remaining_time": "1:16:35"}
+{"current_steps": 38, "total_steps": 213, "loss": 0.47378021478652954, "lr": 9.848591102083375e-06, "epoch": 0.5404444444444444, "percentage": 17.84, "elapsed_time": "0:16:30", "remaining_time": "1:16:02"}
+{"current_steps": 39, "total_steps": 213, "loss": 0.4684975743293762, "lr": 9.82785080804381e-06, "epoch": 0.5546666666666666, "percentage": 18.31, "elapsed_time": "0:16:55", "remaining_time": "1:15:31"}
+{"current_steps": 40, "total_steps": 213, "loss": 0.5784010887145996, "lr": 9.805804412106197e-06, "epoch": 0.5688888888888889, "percentage": 18.78, "elapsed_time": "0:17:20", "remaining_time": "1:14:59"}
+{"current_steps": 41, "total_steps": 213, "loss": 0.46650874614715576, "lr": 9.782457878588977e-06, "epoch": 0.5831111111111111, "percentage": 19.25, "elapsed_time": "0:17:45", "remaining_time": "1:14:29"}
+{"current_steps": 42, "total_steps": 213, "loss": 0.42804157733917236, "lr": 9.75781752354311e-06, "epoch": 0.5973333333333334, "percentage": 19.72, "elapsed_time": "0:18:10", "remaining_time": "1:13:58"}
+{"current_steps": 43, "total_steps": 213, "loss": 0.48224180936813354, "lr": 9.731890013043367e-06, "epoch": 0.6115555555555555, "percentage": 20.19, "elapsed_time": "0:18:35", "remaining_time": "1:13:28"}
+{"current_steps": 44, "total_steps": 213, "loss": 0.4561079144477844, "lr": 9.704682361384941e-06, "epoch": 0.6257777777777778, "percentage": 20.66, "elapsed_time": "0:19:00", "remaining_time": "1:12:59"}
+{"current_steps": 45, "total_steps": 213, "loss": 0.4007866382598877, "lr": 9.676201929185809e-06, "epoch": 0.64, "percentage": 21.13, "elapsed_time": "0:19:25", "remaining_time": "1:12:30"}
+{"current_steps": 46, "total_steps": 213, "loss": 0.4587845802307129, "lr": 9.646456421395447e-06, "epoch": 0.6542222222222223, "percentage": 21.6, "elapsed_time": "0:19:50", "remaining_time": "1:12:00"}
+{"current_steps": 47, "total_steps": 213, "loss": 0.4571158289909363, "lr": 9.615453885210368e-06, "epoch": 0.6684444444444444, "percentage": 22.07, "elapsed_time": "0:20:15", "remaining_time": "1:11:32"}
+{"current_steps": 48, "total_steps": 213, "loss": 0.4724605679512024, "lr": 9.583202707897075e-06, "epoch": 0.6826666666666666, "percentage": 22.54, "elapsed_time": "0:20:40", "remaining_time": "1:11:03"}
+{"current_steps": 49, "total_steps": 213, "loss": 0.4600468575954437, "lr": 9.549711614523007e-06, "epoch": 0.6968888888888889, "percentage": 23.0, "elapsed_time": "0:21:05", "remaining_time": "1:10:34"}
+{"current_steps": 50, "total_steps": 213, "loss": 0.4377005994319916, "lr": 9.514989665596114e-06, "epoch": 0.7111111111111111, "percentage": 23.47, "elapsed_time": "0:21:29", "remaining_time": "1:10:05"}
+{"current_steps": 50, "total_steps": 213, "eval_loss": 0.4685295820236206, "epoch": 0.7111111111111111, "percentage": 23.47, "elapsed_time": "0:21:36", "remaining_time": "1:10:26"}
+{"current_steps": 51, "total_steps": 213, "loss": 0.4415626525878906, "lr": 9.479046254613673e-06, "epoch": 0.7253333333333334, "percentage": 23.94, "elapsed_time": "0:22:01", "remaining_time": "1:09:57"}
+{"current_steps": 52, "total_steps": 213, "loss": 0.42561036348342896, "lr": 9.441891105521005e-06, "epoch": 0.7395555555555555, "percentage": 24.41, "elapsed_time": "0:22:26", "remaining_time": "1:09:28"}
+{"current_steps": 53, "total_steps": 213, "loss": 0.4630648195743561, "lr": 9.40353427008083e-06, "epoch": 0.7537777777777778, "percentage": 24.88, "elapsed_time": "0:22:51", "remaining_time": "1:08:59"}
+{"current_steps": 54, "total_steps": 213, "loss": 0.43418699502944946, "lr": 9.3639861251539e-06, "epoch": 0.768, "percentage": 25.35, "elapsed_time": "0:23:16", "remaining_time": "1:08:30"}
+{"current_steps": 55, "total_steps": 213, "loss": 0.47063255310058594, "lr": 9.323257369891702e-06, "epoch": 0.7822222222222223, "percentage": 25.82, "elapsed_time": "0:23:40", "remaining_time": "1:08:01"}
+{"current_steps": 56, "total_steps": 213, "loss": 0.41920918226242065, "lr": 9.281359022841966e-06, "epoch": 0.7964444444444444, "percentage": 26.29, "elapsed_time": "0:24:05", "remaining_time": "1:07:33"}
+{"current_steps": 57, "total_steps": 213, "loss": 0.4757845401763916, "lr": 9.238302418967757e-06, "epoch": 0.8106666666666666, "percentage": 26.76, "elapsed_time": "0:24:31", "remaining_time": "1:07:06"}
+{"current_steps": 58, "total_steps": 213, "loss": 0.5198032855987549, "lr": 9.194099206580981e-06, "epoch": 0.8248888888888889, "percentage": 27.23, "elapsed_time": "0:24:56", "remaining_time": "1:06:39"}
+{"current_steps": 59, "total_steps": 213, "loss": 0.46035224199295044, "lr": 9.14876134419111e-06, "epoch": 0.8391111111111111, "percentage": 27.7, "elapsed_time": "0:25:21", "remaining_time": "1:06:11"}
+{"current_steps": 60, "total_steps": 213, "loss": 0.4945712983608246, "lr": 9.102301097269974e-06, "epoch": 0.8533333333333334, "percentage": 28.17, "elapsed_time": "0:25:46", "remaining_time": "1:05:43"}
+{"current_steps": 61, "total_steps": 213, "loss": 0.4513372778892517, "lr": 9.05473103493355e-06, "epoch": 0.8675555555555555, "percentage": 28.64, "elapsed_time": "0:26:11", "remaining_time": "1:05:15"}
+{"current_steps": 62, "total_steps": 213, "loss": 0.49865084886550903, "lr": 9.006064026541549e-06, "epoch": 0.8817777777777778, "percentage": 29.11, "elapsed_time": "0:26:36", "remaining_time": "1:04:47"}
+{"current_steps": 63, "total_steps": 213, "loss": 0.4319220185279846, "lr": 8.956313238215824e-06, "epoch": 0.896, "percentage": 29.58, "elapsed_time": "0:27:01", "remaining_time": "1:04:20"}
+{"current_steps": 64, "total_steps": 213, "loss": 0.43760013580322266, "lr": 8.905492129278478e-06, "epoch": 0.9102222222222223, "percentage": 30.05, "elapsed_time": "0:27:26", "remaining_time": "1:03:52"}
+{"current_steps": 65, "total_steps": 213, "loss": 0.41593095660209656, "lr": 8.85361444861063e-06, "epoch": 0.9244444444444444, "percentage": 30.52, "elapsed_time": "0:27:52", "remaining_time": "1:03:27"}
+{"current_steps": 66, "total_steps": 213, "loss": 0.48464590311050415, "lr": 8.800694230932885e-06, "epoch": 0.9386666666666666, "percentage": 30.99, "elapsed_time": "0:28:17", "remaining_time": "1:03:00"}
+{"current_steps": 67, "total_steps": 213, "loss": 0.41884416341781616, "lr": 8.74674579300843e-06, "epoch": 0.9528888888888889, "percentage": 31.46, "elapsed_time": "0:28:42", "remaining_time": "1:02:33"}
+{"current_steps": 68, "total_steps": 213, "loss": 0.40650737285614014, "lr": 8.691783729769874e-06, "epoch": 0.9671111111111111, "percentage": 31.92, "elapsed_time": "0:29:07", "remaining_time": "1:02:05"}
+{"current_steps": 69, "total_steps": 213, "loss": 0.4726918637752533, "lr": 8.635822910370793e-06, "epoch": 0.9813333333333333, "percentage": 32.39, "elapsed_time": "0:29:32", "remaining_time": "1:01:38"}
+{"current_steps": 70, "total_steps": 213, "loss": 0.4404842257499695, "lr": 8.578878474163115e-06, "epoch": 0.9955555555555555, "percentage": 32.86, "elapsed_time": "0:29:57", "remaining_time": "1:01:11"}
+{"current_steps": 71, "total_steps": 213, "loss": 0.42633137106895447, "lr": 8.520965826601394e-06, "epoch": 1.0, "percentage": 33.33, "elapsed_time": "0:30:04", "remaining_time": "1:00:09"}
+{"current_steps": 72, "total_steps": 213, "loss": 0.46000930666923523, "lr": 8.462100635075097e-06, "epoch": 1.0142222222222221, "percentage": 33.8, "elapsed_time": "0:32:29", "remaining_time": "1:03:37"}
+{"current_steps": 73, "total_steps": 213, "loss": 0.42150819301605225, "lr": 8.40229882467003e-06, "epoch": 1.0284444444444445, "percentage": 34.27, "elapsed_time": "0:32:54", "remaining_time": "1:03:07"}
+{"current_steps": 74, "total_steps": 213, "loss": 0.37313175201416016, "lr": 8.341576573860049e-06, "epoch": 1.0426666666666666, "percentage": 34.74, "elapsed_time": "0:33:20", "remaining_time": "1:02:36"}
+{"current_steps": 75, "total_steps": 213, "loss": 0.382789671421051, "lr": 8.279950310130218e-06, "epoch": 1.056888888888889, "percentage": 35.21, "elapsed_time": "0:33:44", "remaining_time": "1:02:05"}
+{"current_steps": 76, "total_steps": 213, "loss": 0.3734534978866577, "lr": 8.2174367055326e-06, "epoch": 1.0711111111111111, "percentage": 35.68, "elapsed_time": "0:34:10", "remaining_time": "1:01:35"}
+{"current_steps": 77, "total_steps": 213, "loss": 0.37693876028060913, "lr": 8.154052672175888e-06, "epoch": 1.0853333333333333, "percentage": 36.15, "elapsed_time": "0:34:35", "remaining_time": "1:01:05"}
+{"current_steps": 78, "total_steps": 213, "loss": 0.4265296459197998, "lr": 8.08981535765009e-06, "epoch": 1.0995555555555556, "percentage": 36.62, "elapsed_time": "0:35:00", "remaining_time": "1:00:34"}
+{"current_steps": 79, "total_steps": 213, "loss": 0.3812956213951111, "lr": 8.024742140387506e-06, "epoch": 1.1137777777777778, "percentage": 37.09, "elapsed_time": "0:35:25", "remaining_time": "1:00:04"}
+{"current_steps": 80, "total_steps": 213, "loss": 0.4131423234939575, "lr": 7.95885062496126e-06, "epoch": 1.1280000000000001, "percentage": 37.56, "elapsed_time": "0:35:49", "remaining_time": "0:59:34"}
+{"current_steps": 81, "total_steps": 213, "loss": 0.37861311435699463, "lr": 7.892158637322647e-06, "epoch": 1.1422222222222222, "percentage": 38.03, "elapsed_time": "0:36:14", "remaining_time": "0:59:04"}
+{"current_steps": 82, "total_steps": 213, "loss": 0.3589654564857483, "lr": 7.824684219978591e-06, "epoch": 1.1564444444444444, "percentage": 38.5, "elapsed_time": "0:36:40", "remaining_time": "0:58:34"}
+{"current_steps": 83, "total_steps": 213, "loss": 0.40794360637664795, "lr": 7.756445627110523e-06, "epoch": 1.1706666666666667, "percentage": 38.97, "elapsed_time": "0:37:04", "remaining_time": "0:58:04"}
+{"current_steps": 84, "total_steps": 213, "loss": 0.4278006851673126, "lr": 7.68746131963598e-06, "epoch": 1.1848888888888889, "percentage": 39.44, "elapsed_time": "0:37:29", "remaining_time": "0:57:34"}
+{"current_steps": 85, "total_steps": 213, "loss": 0.371395081281662, "lr": 7.6177499602143e-06, "epoch": 1.199111111111111, "percentage": 39.91, "elapsed_time": "0:37:54", "remaining_time": "0:57:05"}
+{"current_steps": 86, "total_steps": 213, "loss": 0.4305034279823303, "lr": 7.547330408197695e-06, "epoch": 1.2133333333333334, "percentage": 40.38, "elapsed_time": "0:38:19", "remaining_time": "0:56:36"}
+{"current_steps": 87, "total_steps": 213, "loss": 0.4348181188106537, "lr": 7.476221714529167e-06, "epoch": 1.2275555555555555, "percentage": 40.85, "elapsed_time": "0:38:44", "remaining_time": "0:56:06"}
+{"current_steps": 88, "total_steps": 213, "loss": 0.4309384822845459, "lr": 7.404443116588548e-06, "epoch": 1.2417777777777779, "percentage": 41.31, "elapsed_time": "0:39:09", "remaining_time": "0:55:37"}
+{"current_steps": 89, "total_steps": 213, "loss": 0.4107038676738739, "lr": 7.332014032988123e-06, "epoch": 1.256, "percentage": 41.78, "elapsed_time": "0:39:34", "remaining_time": "0:55:08"}
+{"current_steps": 90, "total_steps": 213, "loss": 0.37457963824272156, "lr": 7.2589540583192165e-06, "epoch": 1.2702222222222221, "percentage": 42.25, "elapsed_time": "0:39:59", "remaining_time": "0:54:39"}
+{"current_steps": 91, "total_steps": 213, "loss": 0.37697723507881165, "lr": 7.185282957851175e-06, "epoch": 1.2844444444444445, "percentage": 42.72, "elapsed_time": "0:40:24", "remaining_time": "0:54:09"}
+{"current_steps": 92, "total_steps": 213, "loss": 0.40353477001190186, "lr": 7.111020662184174e-06, "epoch": 1.2986666666666666, "percentage": 43.19, "elapsed_time": "0:40:49", "remaining_time": "0:53:41"}
+{"current_steps": 93, "total_steps": 213, "loss": 0.37755411863327026, "lr": 7.036187261857289e-06, "epoch": 1.3128888888888888, "percentage": 43.66, "elapsed_time": "0:41:14", "remaining_time": "0:53:12"}
+{"current_steps": 94, "total_steps": 213, "loss": 0.3729894757270813, "lr": 6.960803001913315e-06, "epoch": 1.3271111111111111, "percentage": 44.13, "elapsed_time": "0:41:39", "remaining_time": "0:52:43"}
+{"current_steps": 95, "total_steps": 213, "loss": 0.4091319441795349, "lr": 6.884888276421766e-06, "epoch": 1.3413333333333333, "percentage": 44.6, "elapsed_time": "0:42:04", "remaining_time": "0:52:15"}
+{"current_steps": 96, "total_steps": 213, "loss": 0.4048840403556824, "lr": 6.8084636229615786e-06, "epoch": 1.3555555555555556, "percentage": 45.07, "elapsed_time": "0:42:28", "remaining_time": "0:51:46"}
+{"current_steps": 97, "total_steps": 213, "loss": 0.39409685134887695, "lr": 6.731549717064975e-06, "epoch": 1.3697777777777778, "percentage": 45.54, "elapsed_time": "0:42:53", "remaining_time": "0:51:17"}
+{"current_steps": 98, "total_steps": 213, "loss": 0.36237582564353943, "lr": 6.654167366624009e-06, "epoch": 1.384, "percentage": 46.01, "elapsed_time": "0:43:18", "remaining_time": "0:50:49"}
+{"current_steps": 99, "total_steps": 213, "loss": 0.38763654232025146, "lr": 6.576337506261314e-06, "epoch": 1.3982222222222223, "percentage": 46.48, "elapsed_time": "0:43:43", "remaining_time": "0:50:20"}
+{"current_steps": 100, "total_steps": 213, "loss": 0.41751331090927124, "lr": 6.498081191666549e-06, "epoch": 1.4124444444444444, "percentage": 46.95, "elapsed_time": "0:44:08", "remaining_time": "0:49:52"}
+{"current_steps": 100, "total_steps": 213, "eval_loss": 0.4669477939605713, "epoch": 1.4124444444444444, "percentage": 46.95, "elapsed_time": "0:44:13", "remaining_time": "0:49:58"}
+{"current_steps": 101, "total_steps": 213, "loss": 0.38504236936569214, "lr": 6.419419593900109e-06, "epoch": 1.4266666666666667, "percentage": 47.42, "elapsed_time": "0:44:38", "remaining_time": "0:49:30"}
+{"current_steps": 102, "total_steps": 213, "loss": 0.36031031608581543, "lr": 6.340373993665607e-06, "epoch": 1.4408888888888889, "percentage": 47.89, "elapsed_time": "0:45:03", "remaining_time": "0:49:01"}
+{"current_steps": 103, "total_steps": 213, "loss": 0.3863949179649353, "lr": 6.2609657755527135e-06, "epoch": 1.455111111111111, "percentage": 48.36, "elapsed_time": "0:45:28", "remaining_time": "0:48:34"}
+{"current_steps": 104, "total_steps": 213, "loss": 0.3955462574958801, "lr": 6.1812164222518626e-06, "epoch": 1.4693333333333334, "percentage": 48.83, "elapsed_time": "0:45:53", "remaining_time": "0:48:05"}
+{"current_steps": 105, "total_steps": 213, "loss": 0.39310169219970703, "lr": 6.101147508742456e-06, "epoch": 1.4835555555555555, "percentage": 49.3, "elapsed_time": "0:46:18", "remaining_time": "0:47:37"}
+{"current_steps": 106, "total_steps": 213, "loss": 0.4533539414405823, "lr": 6.020780696456059e-06, "epoch": 1.4977777777777779, "percentage": 49.77, "elapsed_time": "0:46:43", "remaining_time": "0:47:09"}
+{"current_steps": 107, "total_steps": 213, "loss": 0.3985057473182678, "lr": 5.940137727416247e-06, "epoch": 1.512, "percentage": 50.23, "elapsed_time": "0:47:08", "remaining_time": "0:46:41"}
+{"current_steps": 108, "total_steps": 213, "loss": 0.3878968358039856, "lr": 5.859240418356614e-06, "epoch": 1.5262222222222221, "percentage": 50.7, "elapsed_time": "0:47:33", "remaining_time": "0:46:14"}
+{"current_steps": 109, "total_steps": 213, "loss": 0.39648741483688354, "lr": 5.778110654818602e-06, "epoch": 1.5404444444444443, "percentage": 51.17, "elapsed_time": "0:47:58", "remaining_time": "0:45:46"}
+{"current_steps": 110, "total_steps": 213, "loss": 0.4319821000099182, "lr": 5.696770385230679e-06, "epoch": 1.5546666666666666, "percentage": 51.64, "elapsed_time": "0:48:23", "remaining_time": "0:45:18"}
+{"current_steps": 111, "total_steps": 213, "loss": 0.4078192710876465, "lr": 5.615241614970546e-06, "epoch": 1.568888888888889, "percentage": 52.11, "elapsed_time": "0:48:47", "remaining_time": "0:44:50"}
+{"current_steps": 112, "total_steps": 213, "loss": 0.38504159450531006, "lr": 5.533546400411899e-06, "epoch": 1.5831111111111111, "percentage": 52.58, "elapsed_time": "0:49:12", "remaining_time": "0:44:22"}
+{"current_steps": 113, "total_steps": 213, "loss": 0.3971819579601288, "lr": 5.4517068429574215e-06, "epoch": 1.5973333333333333, "percentage": 53.05, "elapsed_time": "0:49:37", "remaining_time": "0:43:55"}
+{"current_steps": 114, "total_steps": 213, "loss": 0.3886791169643402, "lr": 5.3697450830595775e-06, "epoch": 1.6115555555555554, "percentage": 53.52, "elapsed_time": "0:50:02", "remaining_time": "0:43:27"}
+{"current_steps": 115, "total_steps": 213, "loss": 0.37610527873039246, "lr": 5.287683294230855e-06, "epoch": 1.6257777777777778, "percentage": 53.99, "elapsed_time": "0:50:27", "remaining_time": "0:43:00"}
+{"current_steps": 116, "total_steps": 213, "loss": 0.3846814036369324, "lr": 5.20554367704505e-06, "epoch": 1.6400000000000001, "percentage": 54.46, "elapsed_time": "0:50:52", "remaining_time": "0:42:32"}
+{"current_steps": 117, "total_steps": 213, "loss": 0.3917155861854553, "lr": 5.123348453131242e-06, "epoch": 1.6542222222222223, "percentage": 54.93, "elapsed_time": "0:51:17", "remaining_time": "0:42:05"}
+{"current_steps": 118, "total_steps": 213, "loss": 0.39533206820487976, "lr": 5.041119859162068e-06, "epoch": 1.6684444444444444, "percentage": 55.4, "elapsed_time": "0:51:42", "remaining_time": "0:41:37"}
+{"current_steps": 119, "total_steps": 213, "loss": 0.4113273620605469, "lr": 4.958880140837934e-06, "epoch": 1.6826666666666665, "percentage": 55.87, "elapsed_time": "0:52:07", "remaining_time": "0:41:10"}
+{"current_steps": 120, "total_steps": 213, "loss": 0.39543214440345764, "lr": 4.876651546868759e-06, "epoch": 1.696888888888889, "percentage": 56.34, "elapsed_time": "0:52:32", "remaining_time": "0:40:43"}
+{"current_steps": 121, "total_steps": 213, "loss": 0.4548013508319855, "lr": 4.794456322954953e-06, "epoch": 1.7111111111111112, "percentage": 56.81, "elapsed_time": "0:52:57", "remaining_time": "0:40:15"}
+{"current_steps": 122, "total_steps": 213, "loss": 0.4111294150352478, "lr": 4.7123167057691446e-06, "epoch": 1.7253333333333334, "percentage": 57.28, "elapsed_time": "0:53:22", "remaining_time": "0:39:48"}
+{"current_steps": 123, "total_steps": 213, "loss": 0.43050098419189453, "lr": 4.630254916940424e-06, "epoch": 1.7395555555555555, "percentage": 57.75, "elapsed_time": "0:53:48", "remaining_time": "0:39:22"}
+{"current_steps": 124, "total_steps": 213, "loss": 0.41356489062309265, "lr": 4.548293157042581e-06, "epoch": 1.7537777777777777, "percentage": 58.22, "elapsed_time": "0:54:13", "remaining_time": "0:38:54"}
+{"current_steps": 125, "total_steps": 213, "loss": 0.41799914836883545, "lr": 4.466453599588103e-06, "epoch": 1.768, "percentage": 58.69, "elapsed_time": "0:54:37", "remaining_time": "0:38:27"}
+{"current_steps": 126, "total_steps": 213, "loss": 0.3510154187679291, "lr": 4.384758385029457e-06, "epoch": 1.7822222222222224, "percentage": 59.15, "elapsed_time": "0:55:03", "remaining_time": "0:38:00"}
+{"current_steps": 127, "total_steps": 213, "loss": 0.4098949432373047, "lr": 4.3032296147693225e-06, "epoch": 1.7964444444444445, "percentage": 59.62, "elapsed_time": "0:55:28", "remaining_time": "0:37:33"}
+{"current_steps": 128, "total_steps": 213, "loss": 0.36957240104675293, "lr": 4.2218893451814005e-06, "epoch": 1.8106666666666666, "percentage": 60.09, "elapsed_time": "0:55:53", "remaining_time": "0:37:06"}
+{"current_steps": 129, "total_steps": 213, "loss": 0.3535369038581848, "lr": 4.140759581643386e-06, "epoch": 1.8248888888888888, "percentage": 60.56, "elapsed_time": "0:56:19", "remaining_time": "0:36:40"}
+{"current_steps": 130, "total_steps": 213, "loss": 0.38296306133270264, "lr": 4.059862272583755e-06, "epoch": 1.8391111111111111, "percentage": 61.03, "elapsed_time": "0:56:43", "remaining_time": "0:36:13"}
+{"current_steps": 131, "total_steps": 213, "loss": 0.38752996921539307, "lr": 3.979219303543942e-06, "epoch": 1.8533333333333335, "percentage": 61.5, "elapsed_time": "0:57:08", "remaining_time": "0:35:46"}
+{"current_steps": 132, "total_steps": 213, "loss": 0.3867356479167938, "lr": 3.898852491257547e-06, "epoch": 1.8675555555555556, "percentage": 61.97, "elapsed_time": "0:57:33", "remaining_time": "0:35:19"}
+{"current_steps": 133, "total_steps": 213, "loss": 0.42934614419937134, "lr": 3.818783577748138e-06, "epoch": 1.8817777777777778, "percentage": 62.44, "elapsed_time": "0:57:58", "remaining_time": "0:34:52"}
+{"current_steps": 134, "total_steps": 213, "loss": 0.4096015691757202, "lr": 3.7390342244472886e-06, "epoch": 1.896, "percentage": 62.91, "elapsed_time": "0:58:23", "remaining_time": "0:34:25"}
+{"current_steps": 135, "total_steps": 213, "loss": 0.3849722743034363, "lr": 3.659626006334395e-06, "epoch": 1.9102222222222223, "percentage": 63.38, "elapsed_time": "0:58:47", "remaining_time": "0:33:58"}
+{"current_steps": 136, "total_steps": 213, "loss": 0.415260910987854, "lr": 3.5805804060998926e-06, "epoch": 1.9244444444444444, "percentage": 63.85, "elapsed_time": "0:59:12", "remaining_time": "0:33:31"}
+{"current_steps": 137, "total_steps": 213, "loss": 0.48382842540740967, "lr": 3.501918808333453e-06, "epoch": 1.9386666666666668, "percentage": 64.32, "elapsed_time": "0:59:37", "remaining_time": "0:33:04"}
+{"current_steps": 138, "total_steps": 213, "loss": 0.4434107542037964, "lr": 3.4236624937386874e-06, "epoch": 1.952888888888889, "percentage": 64.79, "elapsed_time": "1:00:02", "remaining_time": "0:32:37"}
+{"current_steps": 139, "total_steps": 213, "loss": 0.41760051250457764, "lr": 3.3458326333759927e-06, "epoch": 1.967111111111111, "percentage": 65.26, "elapsed_time": "1:00:27", "remaining_time": "0:32:11"}
+{"current_steps": 140, "total_steps": 213, "loss": 0.43642479181289673, "lr": 3.268450282935026e-06, "epoch": 1.9813333333333332, "percentage": 65.73, "elapsed_time": "1:00:51", "remaining_time": "0:31:44"}
+{"current_steps": 141, "total_steps": 213, "loss": 0.38327687978744507, "lr": 3.1915363770384223e-06, "epoch": 1.9955555555555555, "percentage": 66.2, "elapsed_time": "1:01:17", "remaining_time": "0:31:17"}
+{"current_steps": 142, "total_steps": 213, "loss": 0.3751964867115021, "lr": 3.1151117235782346e-06, "epoch": 2.0, "percentage": 66.67, "elapsed_time": "1:01:25", "remaining_time": "0:30:42"}
+{"current_steps": 143, "total_steps": 213, "loss": 0.3831713795661926, "lr": 3.0391969980866874e-06, "epoch": 2.014222222222222, "percentage": 67.14, "elapsed_time": "1:03:45", "remaining_time": "0:31:12"}
+{"current_steps": 144, "total_steps": 213, "loss": 0.338630735874176, "lr": 2.963812738142713e-06, "epoch": 2.0284444444444443, "percentage": 67.61, "elapsed_time": "1:04:10", "remaining_time": "0:30:45"}
+{"current_steps": 145, "total_steps": 213, "loss": 0.3684050440788269, "lr": 2.8889793378158284e-06, "epoch": 2.042666666666667, "percentage": 68.08, "elapsed_time": "1:04:35", "remaining_time": "0:30:17"}
+{"current_steps": 146, "total_steps": 213, "loss": 0.37092137336730957, "lr": 2.814717042148827e-06, "epoch": 2.056888888888889, "percentage": 68.54, "elapsed_time": "1:05:00", "remaining_time": "0:29:50"}
+{"current_steps": 147, "total_steps": 213, "loss": 0.34995853900909424, "lr": 2.7410459416807856e-06, "epoch": 2.071111111111111, "percentage": 69.01, "elapsed_time": "1:05:26", "remaining_time": "0:29:22"}
+{"current_steps": 148, "total_steps": 213, "loss": 0.37355348467826843, "lr": 2.6679859670118785e-06, "epoch": 2.0853333333333333, "percentage": 69.48, "elapsed_time": "1:05:51", "remaining_time": "0:28:55"}
+{"current_steps": 149, "total_steps": 213, "loss": 0.3448713421821594, "lr": 2.5955568834114523e-06, "epoch": 2.0995555555555554, "percentage": 69.95, "elapsed_time": "1:06:16", "remaining_time": "0:28:28"}
+{"current_steps": 150, "total_steps": 213, "loss": 0.3545028865337372, "lr": 2.523778285470835e-06, "epoch": 2.113777777777778, "percentage": 70.42, "elapsed_time": "1:06:41", "remaining_time": "0:28:00"}
+{"current_steps": 150, "total_steps": 213, "eval_loss": 0.4679063558578491, "epoch": 2.113777777777778, "percentage": 70.42, "elapsed_time": "1:06:46", "remaining_time": "0:28:02"}
+{"current_steps": 151, "total_steps": 213, "loss": 0.32551515102386475, "lr": 2.452669591802307e-06, "epoch": 2.128, "percentage": 70.89, "elapsed_time": "1:07:12", "remaining_time": "0:27:35"}
+{"current_steps": 152, "total_steps": 213, "loss": 0.35764873027801514, "lr": 2.3822500397857016e-06, "epoch": 2.1422222222222222, "percentage": 71.36, "elapsed_time": "1:07:37", "remaining_time": "0:27:08"}
+{"current_steps": 153, "total_steps": 213, "loss": 0.35795581340789795, "lr": 2.3125386803640188e-06, "epoch": 2.1564444444444444, "percentage": 71.83, "elapsed_time": "1:08:02", "remaining_time": "0:26:40"}
+{"current_steps": 154, "total_steps": 213, "loss": 0.32651567459106445, "lr": 2.243554372889479e-06, "epoch": 2.1706666666666665, "percentage": 72.3, "elapsed_time": "1:08:27", "remaining_time": "0:26:13"}
+{"current_steps": 155, "total_steps": 213, "loss": 0.3542096018791199, "lr": 2.1753157800214107e-06, "epoch": 2.1848888888888887, "percentage": 72.77, "elapsed_time": "1:08:52", "remaining_time": "0:25:46"}
+{"current_steps": 156, "total_steps": 213, "loss": 0.3560444116592407, "lr": 2.1078413626773547e-06, "epoch": 2.1991111111111112, "percentage": 73.24, "elapsed_time": "1:09:17", "remaining_time": "0:25:18"}
+{"current_steps": 157, "total_steps": 213, "loss": 0.3283962607383728, "lr": 2.0411493750387423e-06, "epoch": 2.2133333333333334, "percentage": 73.71, "elapsed_time": "1:09:41", "remaining_time": "0:24:51"}
+{"current_steps": 158, "total_steps": 213, "loss": 0.362970769405365, "lr": 1.9752578596124955e-06, "epoch": 2.2275555555555555, "percentage": 74.18, "elapsed_time": "1:10:07", "remaining_time": "0:24:24"}
+{"current_steps": 159, "total_steps": 213, "loss": 0.3372136354446411, "lr": 1.9101846423499113e-06, "epoch": 2.2417777777777776, "percentage": 74.65, "elapsed_time": "1:10:32", "remaining_time": "0:23:57"}
+{"current_steps": 160, "total_steps": 213, "loss": 0.343697726726532, "lr": 1.8459473278241125e-06, "epoch": 2.2560000000000002, "percentage": 75.12, "elapsed_time": "1:10:57", "remaining_time": "0:23:30"}
+{"current_steps": 161, "total_steps": 213, "loss": 0.3472204804420471, "lr": 1.7825632944674016e-06, "epoch": 2.2702222222222224, "percentage": 75.59, "elapsed_time": "1:11:22", "remaining_time": "0:23:03"}
+{"current_steps": 162, "total_steps": 213, "loss": 0.34057295322418213, "lr": 1.7200496898697832e-06, "epoch": 2.2844444444444445, "percentage": 76.06, "elapsed_time": "1:11:46", "remaining_time": "0:22:35"}
+{"current_steps": 163, "total_steps": 213, "loss": 0.3398140072822571, "lr": 1.6584234261399535e-06, "epoch": 2.2986666666666666, "percentage": 76.53, "elapsed_time": "1:12:11", "remaining_time": "0:22:08"}
+{"current_steps": 164, "total_steps": 213, "loss": 0.3423030376434326, "lr": 1.5977011753299726e-06, "epoch": 2.3128888888888888, "percentage": 77.0, "elapsed_time": "1:12:36", "remaining_time": "0:21:41"}
+{"current_steps": 165, "total_steps": 213, "loss": 0.4059848189353943, "lr": 1.5378993649249053e-06, "epoch": 2.327111111111111, "percentage": 77.46, "elapsed_time": "1:13:01", "remaining_time": "0:21:14"}
+{"current_steps": 166, "total_steps": 213, "loss": 0.35659873485565186, "lr": 1.4790341733986085e-06, "epoch": 2.3413333333333335, "percentage": 77.93, "elapsed_time": "1:13:26", "remaining_time": "0:20:47"}
+{"current_steps": 167, "total_steps": 213, "loss": 0.320560097694397, "lr": 1.4211215258368866e-06, "epoch": 2.3555555555555556, "percentage": 78.4, "elapsed_time": "1:13:51", "remaining_time": "0:20:20"}
+{"current_steps": 168, "total_steps": 213, "loss": 0.3432486057281494, "lr": 1.3641770896292083e-06, "epoch": 2.3697777777777778, "percentage": 78.87, "elapsed_time": "1:14:16", "remaining_time": "0:19:53"}
+{"current_steps": 169, "total_steps": 213, "loss": 0.3448861241340637, "lr": 1.3082162702301276e-06, "epoch": 2.384, "percentage": 79.34, "elapsed_time": "1:14:41", "remaining_time": "0:19:26"}
+{"current_steps": 170, "total_steps": 213, "loss": 0.4012100100517273, "lr": 1.2532542069915722e-06, "epoch": 2.398222222222222, "percentage": 79.81, "elapsed_time": "1:15:05", "remaining_time": "0:18:59"}
+{"current_steps": 171, "total_steps": 213, "loss": 0.35434189438819885, "lr": 1.1993057690671174e-06, "epoch": 2.4124444444444446, "percentage": 80.28, "elapsed_time": "1:15:30", "remaining_time": "0:18:32"}
+{"current_steps": 172, "total_steps": 213, "loss": 0.37489616870880127, "lr": 1.1463855513893695e-06, "epoch": 2.4266666666666667, "percentage": 80.75, "elapsed_time": "1:15:56", "remaining_time": "0:18:06"}
+{"current_steps": 173, "total_steps": 213, "loss": 0.3599584698677063, "lr": 1.0945078707215224e-06, "epoch": 2.440888888888889, "percentage": 81.22, "elapsed_time": "1:16:21", "remaining_time": "0:17:39"}
+{"current_steps": 174, "total_steps": 213, "loss": 0.38492339849472046, "lr": 1.0436867617841768e-06, "epoch": 2.455111111111111, "percentage": 81.69, "elapsed_time": "1:16:46", "remaining_time": "0:17:12"}
+{"current_steps": 175, "total_steps": 213, "loss": 0.36968979239463806, "lr": 9.939359734584552e-07, "epoch": 2.469333333333333, "percentage": 82.16, "elapsed_time": "1:17:11", "remaining_time": "0:16:45"}
+{"current_steps": 176, "total_steps": 213, "loss": 0.3575425148010254, "lr": 9.452689650664515e-07, "epoch": 2.4835555555555557, "percentage": 82.63, "elapsed_time": "1:17:36", "remaining_time": "0:16:18"}
+{"current_steps": 177, "total_steps": 213, "loss": 0.35382080078125, "lr": 8.976989027300265e-07, "epoch": 2.497777777777778, "percentage": 83.1, "elapsed_time": "1:18:01", "remaining_time": "0:15:52"}
+{"current_steps": 178, "total_steps": 213, "loss": 0.35292625427246094, "lr": 8.512386558088919e-07, "epoch": 2.512, "percentage": 83.57, "elapsed_time": "1:18:26", "remaining_time": "0:15:25"}
+{"current_steps": 179, "total_steps": 213, "loss": 0.3658143877983093, "lr": 8.059007934190194e-07, "epoch": 2.526222222222222, "percentage": 84.04, "elapsed_time": "1:18:51", "remaining_time": "0:14:58"}
+{"current_steps": 180, "total_steps": 213, "loss": 0.34945034980773926, "lr": 7.61697581032243e-07, "epoch": 2.5404444444444443, "percentage": 84.51, "elapsed_time": "1:19:16", "remaining_time": "0:14:32"}
+{"current_steps": 181, "total_steps": 213, "loss": 0.3492674231529236, "lr": 7.186409771580355e-07, "epoch": 2.554666666666667, "percentage": 84.98, "elapsed_time": "1:19:41", "remaining_time": "0:14:05"}
+{"current_steps": 182, "total_steps": 213, "loss": 0.35949787497520447, "lr": 6.76742630108298e-07, "epoch": 2.568888888888889, "percentage": 85.45, "elapsed_time": "1:20:06", "remaining_time": "0:13:38"}
+{"current_steps": 183, "total_steps": 213, "loss": 0.35065189003944397, "lr": 6.360138748461015e-07, "epoch": 2.583111111111111, "percentage": 85.92, "elapsed_time": "1:20:31", "remaining_time": "0:13:12"}
+{"current_steps": 184, "total_steps": 213, "loss": 0.31525540351867676, "lr": 5.964657299191712e-07, "epoch": 2.5973333333333333, "percentage": 86.38, "elapsed_time": "1:20:56", "remaining_time": "0:12:45"}
+{"current_steps": 185, "total_steps": 213, "loss": 0.34680360555648804, "lr": 5.581088944789953e-07, "epoch": 2.6115555555555554, "percentage": 86.85, "elapsed_time": "1:21:21", "remaining_time": "0:12:18"}
+{"current_steps": 186, "total_steps": 213, "loss": 0.3423649072647095, "lr": 5.20953745386329e-07, "epoch": 2.6257777777777775, "percentage": 87.32, "elapsed_time": "1:21:46", "remaining_time": "0:11:52"}
+{"current_steps": 187, "total_steps": 213, "loss": 0.3583834767341614, "lr": 4.850103344038853e-07, "epoch": 2.64, "percentage": 87.79, "elapsed_time": "1:22:10", "remaining_time": "0:11:25"}
+{"current_steps": 188, "total_steps": 213, "loss": 0.3272634744644165, "lr": 4.502883854769935e-07, "epoch": 2.6542222222222223, "percentage": 88.26, "elapsed_time": "1:22:35", "remaining_time": "0:10:58"}
+{"current_steps": 189, "total_steps": 213, "loss": 0.3352956175804138, "lr": 4.167972921029262e-07, "epoch": 2.6684444444444444, "percentage": 88.73, "elapsed_time": "1:23:00", "remaining_time": "0:10:32"}
+{"current_steps": 190, "total_steps": 213, "loss": 0.3704286515712738, "lr": 3.8454611478963235e-07, "epoch": 2.6826666666666665, "percentage": 89.2, "elapsed_time": "1:23:25", "remaining_time": "0:10:05"}
+{"current_steps": 191, "total_steps": 213, "loss": 0.34566229581832886, "lr": 3.535435786045538e-07, "epoch": 2.696888888888889, "percentage": 89.67, "elapsed_time": "1:23:50", "remaining_time": "0:09:39"}
+{"current_steps": 192, "total_steps": 213, "loss": 0.35052379965782166, "lr": 3.237980708141919e-07, "epoch": 2.7111111111111112, "percentage": 90.14, "elapsed_time": "1:24:15", "remaining_time": "0:09:12"}
+{"current_steps": 193, "total_steps": 213, "loss": 0.38708382844924927, "lr": 2.9531763861505967e-07, "epoch": 2.7253333333333334, "percentage": 90.61, "elapsed_time": "1:24:40", "remaining_time": "0:08:46"}
+{"current_steps": 194, "total_steps": 213, "loss": 0.3316305875778198, "lr": 2.681099869566328e-07, "epoch": 2.7395555555555555, "percentage": 91.08, "elapsed_time": "1:25:05", "remaining_time": "0:08:19"}
+{"current_steps": 195, "total_steps": 213, "loss": 0.3650358319282532, "lr": 2.4218247645689306e-07, "epoch": 2.7537777777777777, "percentage": 91.55, "elapsed_time": "1:25:29", "remaining_time": "0:07:53"}
+{"current_steps": 196, "total_steps": 213, "loss": 0.36560434103012085, "lr": 2.175421214110235e-07, "epoch": 2.768, "percentage": 92.02, "elapsed_time": "1:25:54", "remaining_time": "0:07:27"}
+{"current_steps": 197, "total_steps": 213, "loss": 0.3463239371776581, "lr": 1.941955878938029e-07, "epoch": 2.7822222222222224, "percentage": 92.49, "elapsed_time": "1:26:19", "remaining_time": "0:07:00"}
+{"current_steps": 198, "total_steps": 213, "loss": 0.35629892349243164, "lr": 1.721491919561913e-07, "epoch": 2.7964444444444445, "percentage": 92.96, "elapsed_time": "1:26:44", "remaining_time": "0:06:34"}
+{"current_steps": 199, "total_steps": 213, "loss": 0.33887457847595215, "lr": 1.514088979166256e-07, "epoch": 2.8106666666666666, "percentage": 93.43, "elapsed_time": "1:27:09", "remaining_time": "0:06:07"}
+{"current_steps": 200, "total_steps": 213, "loss": 0.36586421728134155, "lr": 1.3198031674745814e-07, "epoch": 2.824888888888889, "percentage": 93.9, "elapsed_time": "1:27:34", "remaining_time": "0:05:41"}
+{"current_steps": 200, "total_steps": 213, "eval_loss": 0.4734371602535248, "epoch": 2.824888888888889, "percentage": 93.9, "elapsed_time": "1:27:41", "remaining_time": "0:05:42"}
+{"current_steps": 201, "total_steps": 213, "loss": 0.3514959216117859, "lr": 1.138687045569975e-07, "epoch": 2.8391111111111114, "percentage": 94.37, "elapsed_time": "1:28:06", "remaining_time": "0:05:15"}
+{"current_steps": 202, "total_steps": 213, "loss": 0.37950894236564636, "lr": 9.70789611675449e-08, "epoch": 2.8533333333333335, "percentage": 94.84, "elapsed_time": "1:28:31", "remaining_time": "0:04:49"}
+{"current_steps": 203, "total_steps": 213, "loss": 0.3352017402648926, "lr": 8.161562878982399e-08, "epoch": 2.8675555555555556, "percentage": 95.31, "elapsed_time": "1:28:56", "remaining_time": "0:04:22"}
+{"current_steps": 204, "total_steps": 213, "loss": 0.36504414677619934, "lr": 6.74828907941516e-08, "epoch": 2.8817777777777778, "percentage": 95.77, "elapsed_time": "1:29:21", "remaining_time": "0:03:56"}
+{"current_steps": 205, "total_steps": 213, "loss": 0.3987666964530945, "lr": 5.468457057869358e-08, "epoch": 2.896, "percentage": 96.24, "elapsed_time": "1:29:47", "remaining_time": "0:03:30"}
+{"current_steps": 206, "total_steps": 213, "loss": 0.3345492482185364, "lr": 4.322413053509944e-08, "epoch": 2.910222222222222, "percentage": 96.71, "elapsed_time": "1:30:12", "remaining_time": "0:03:03"}
+{"current_steps": 207, "total_steps": 213, "loss": 0.367769718170166, "lr": 3.31046711118066e-08, "epoch": 2.924444444444444, "percentage": 97.18, "elapsed_time": "1:30:37", "remaining_time": "0:02:37"}
+{"current_steps": 208, "total_steps": 213, "loss": 0.34441590309143066, "lr": 2.4328929975260262e-08, "epoch": 2.9386666666666668, "percentage": 97.65, "elapsed_time": "1:31:02", "remaining_time": "0:02:11"}
+{"current_steps": 209, "total_steps": 213, "loss": 0.3472161889076233, "lr": 1.6899281269279756e-08, "epoch": 2.952888888888889, "percentage": 98.12, "elapsed_time": "1:31:27", "remaining_time": "0:01:45"}
+{"current_steps": 210, "total_steps": 213, "loss": 0.3906354308128357, "lr": 1.0817734972768946e-08, "epoch": 2.967111111111111, "percentage": 98.59, "elapsed_time": "1:31:52", "remaining_time": "0:01:18"}
+{"current_steps": 211, "total_steps": 213, "loss": 0.37977275252342224, "lr": 6.085936355947897e-09, "epoch": 2.981333333333333, "percentage": 99.06, "elapsed_time": "1:32:17", "remaining_time": "0:00:52"}
+{"current_steps": 212, "total_steps": 213, "loss": 0.3533574342727661, "lr": 2.7051655352494654e-09, "epoch": 2.9955555555555557, "percentage": 99.53, "elapsed_time": "1:32:42", "remaining_time": "0:00:26"}
+{"current_steps": 213, "total_steps": 213, "loss": 0.5198727250099182, "lr": 6.763371270035457e-10, "epoch": 3.0, "percentage": 100.0, "elapsed_time": "1:32:50", "remaining_time": "0:00:00"}
+{"current_steps": 213, "total_steps": 213, "epoch": 3.0, "percentage": 100.0, "elapsed_time": "1:34:47", "remaining_time": "0:00:00"}
diff --git a/qwen2.5-3b-instruct/trainer_state.json b/qwen2.5-3b-instruct/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..18f8093fdcfc75343a1441a98431a57f51bf256e
--- /dev/null
+++ b/qwen2.5-3b-instruct/trainer_state.json
@@ -0,0 +1,1566 @@
+{
+ "best_global_step": null,
+ "best_metric": null,
+ "best_model_checkpoint": null,
+ "epoch": 3.0,
+ "eval_steps": 50,
+ "global_step": 213,
+ "is_hyper_param_search": false,
+ "is_local_process_zero": true,
+ "is_world_process_zero": true,
+ "log_history": [
+ {
+ "epoch": 0.014222222222222223,
+ "grad_norm": 0.253393292427063,
+ "learning_rate": 0.0,
+ "loss": 1.19514799118042,
+ "step": 1
+ },
+ {
+ "epoch": 0.028444444444444446,
+ "grad_norm": 0.2720104455947876,
+ "learning_rate": 4.5454545454545457e-07,
+ "loss": 1.2398133277893066,
+ "step": 2
+ },
+ {
+ "epoch": 0.042666666666666665,
+ "grad_norm": 0.22159169614315033,
+ "learning_rate": 9.090909090909091e-07,
+ "loss": 0.9782462120056152,
+ "step": 3
+ },
+ {
+ "epoch": 0.05688888888888889,
+ "grad_norm": 0.17487271130084991,
+ "learning_rate": 1.3636363636363636e-06,
+ "loss": 0.8072947263717651,
+ "step": 4
+ },
+ {
+ "epoch": 0.07111111111111111,
+ "grad_norm": 0.08528934419155121,
+ "learning_rate": 1.8181818181818183e-06,
+ "loss": 0.5301008224487305,
+ "step": 5
+ },
+ {
+ "epoch": 0.08533333333333333,
+ "grad_norm": 0.08074624091386795,
+ "learning_rate": 2.2727272727272728e-06,
+ "loss": 0.6367583274841309,
+ "step": 6
+ },
+ {
+ "epoch": 0.09955555555555555,
+ "grad_norm": 0.05584918335080147,
+ "learning_rate": 2.7272727272727272e-06,
+ "loss": 0.6089775562286377,
+ "step": 7
+ },
+ {
+ "epoch": 0.11377777777777778,
+ "grad_norm": 0.032228983938694,
+ "learning_rate": 3.181818181818182e-06,
+ "loss": 0.5260579586029053,
+ "step": 8
+ },
+ {
+ "epoch": 0.128,
+ "grad_norm": 0.026230908930301666,
+ "learning_rate": 3.6363636363636366e-06,
+ "loss": 0.5446298122406006,
+ "step": 9
+ },
+ {
+ "epoch": 0.14222222222222222,
+ "grad_norm": 0.019648801535367966,
+ "learning_rate": 4.0909090909090915e-06,
+ "loss": 0.5338459610939026,
+ "step": 10
+ },
+ {
+ "epoch": 0.15644444444444444,
+ "grad_norm": 0.014007463119924068,
+ "learning_rate": 4.5454545454545455e-06,
+ "loss": 0.45045405626296997,
+ "step": 11
+ },
+ {
+ "epoch": 0.17066666666666666,
+ "grad_norm": 0.013295263983309269,
+ "learning_rate": 5e-06,
+ "loss": 0.49363279342651367,
+ "step": 12
+ },
+ {
+ "epoch": 0.18488888888888888,
+ "grad_norm": 0.01435786858201027,
+ "learning_rate": 5.4545454545454545e-06,
+ "loss": 0.48757249116897583,
+ "step": 13
+ },
+ {
+ "epoch": 0.1991111111111111,
+ "grad_norm": 0.015238163061439991,
+ "learning_rate": 5.90909090909091e-06,
+ "loss": 0.5002430081367493,
+ "step": 14
+ },
+ {
+ "epoch": 0.21333333333333335,
+ "grad_norm": 0.014418123289942741,
+ "learning_rate": 6.363636363636364e-06,
+ "loss": 0.4964277148246765,
+ "step": 15
+ },
+ {
+ "epoch": 0.22755555555555557,
+ "grad_norm": 0.014306087978184223,
+ "learning_rate": 6.818181818181818e-06,
+ "loss": 0.4818333387374878,
+ "step": 16
+ },
+ {
+ "epoch": 0.24177777777777779,
+ "grad_norm": 0.009899069555103779,
+ "learning_rate": 7.272727272727273e-06,
+ "loss": 0.4609573483467102,
+ "step": 17
+ },
+ {
+ "epoch": 0.256,
+ "grad_norm": 0.010643995366990566,
+ "learning_rate": 7.727272727272727e-06,
+ "loss": 0.47600606083869934,
+ "step": 18
+ },
+ {
+ "epoch": 0.2702222222222222,
+ "grad_norm": 0.012857547029852867,
+ "learning_rate": 8.181818181818183e-06,
+ "loss": 0.4962907135486603,
+ "step": 19
+ },
+ {
+ "epoch": 0.28444444444444444,
+ "grad_norm": 0.01670556701719761,
+ "learning_rate": 8.636363636363637e-06,
+ "loss": 0.5028363466262817,
+ "step": 20
+ },
+ {
+ "epoch": 0.2986666666666667,
+ "grad_norm": 0.010331347584724426,
+ "learning_rate": 9.090909090909091e-06,
+ "loss": 0.44582226872444153,
+ "step": 21
+ },
+ {
+ "epoch": 0.3128888888888889,
+ "grad_norm": 0.010525904595851898,
+ "learning_rate": 9.545454545454547e-06,
+ "loss": 0.4757377505302429,
+ "step": 22
+ },
+ {
+ "epoch": 0.32711111111111113,
+ "grad_norm": 0.009544270113110542,
+ "learning_rate": 1e-05,
+ "loss": 0.4707476496696472,
+ "step": 23
+ },
+ {
+ "epoch": 0.3413333333333333,
+ "grad_norm": 0.009730499237775803,
+ "learning_rate": 9.999323662872998e-06,
+ "loss": 0.49240708351135254,
+ "step": 24
+ },
+ {
+ "epoch": 0.35555555555555557,
+ "grad_norm": 0.008985932916402817,
+ "learning_rate": 9.99729483446475e-06,
+ "loss": 0.450992614030838,
+ "step": 25
+ },
+ {
+ "epoch": 0.36977777777777776,
+ "grad_norm": 0.01072725746780634,
+ "learning_rate": 9.993914063644053e-06,
+ "loss": 0.4514142870903015,
+ "step": 26
+ },
+ {
+ "epoch": 0.384,
+ "grad_norm": 0.01058228686451912,
+ "learning_rate": 9.989182265027232e-06,
+ "loss": 0.44021710753440857,
+ "step": 27
+ },
+ {
+ "epoch": 0.3982222222222222,
+ "grad_norm": 0.009333685040473938,
+ "learning_rate": 9.98310071873072e-06,
+ "loss": 0.4491778016090393,
+ "step": 28
+ },
+ {
+ "epoch": 0.41244444444444445,
+ "grad_norm": 0.009250632487237453,
+ "learning_rate": 9.975671070024741e-06,
+ "loss": 0.4285920560359955,
+ "step": 29
+ },
+ {
+ "epoch": 0.4266666666666667,
+ "grad_norm": 0.010210932232439518,
+ "learning_rate": 9.966895328888195e-06,
+ "loss": 0.5484102368354797,
+ "step": 30
+ },
+ {
+ "epoch": 0.4408888888888889,
+ "grad_norm": 0.008848125115036964,
+ "learning_rate": 9.956775869464901e-06,
+ "loss": 0.4242202043533325,
+ "step": 31
+ },
+ {
+ "epoch": 0.45511111111111113,
+ "grad_norm": 0.00905672088265419,
+ "learning_rate": 9.945315429421307e-06,
+ "loss": 0.4533390402793884,
+ "step": 32
+ },
+ {
+ "epoch": 0.4693333333333333,
+ "grad_norm": 0.008254914544522762,
+ "learning_rate": 9.932517109205849e-06,
+ "loss": 0.4243648052215576,
+ "step": 33
+ },
+ {
+ "epoch": 0.48355555555555557,
+ "grad_norm": 0.009192033670842648,
+ "learning_rate": 9.918384371210178e-06,
+ "loss": 0.5120309591293335,
+ "step": 34
+ },
+ {
+ "epoch": 0.49777777777777776,
+ "grad_norm": 0.009636987932026386,
+ "learning_rate": 9.902921038832456e-06,
+ "loss": 0.4871100187301636,
+ "step": 35
+ },
+ {
+ "epoch": 0.512,
+ "grad_norm": 0.008572430349886417,
+ "learning_rate": 9.886131295443003e-06,
+ "loss": 0.46277713775634766,
+ "step": 36
+ },
+ {
+ "epoch": 0.5262222222222223,
+ "grad_norm": 0.008587818592786789,
+ "learning_rate": 9.868019683252543e-06,
+ "loss": 0.45996755361557007,
+ "step": 37
+ },
+ {
+ "epoch": 0.5404444444444444,
+ "grad_norm": 0.010775095783174038,
+ "learning_rate": 9.848591102083375e-06,
+ "loss": 0.47378021478652954,
+ "step": 38
+ },
+ {
+ "epoch": 0.5546666666666666,
+ "grad_norm": 0.00995537731796503,
+ "learning_rate": 9.82785080804381e-06,
+ "loss": 0.4684975743293762,
+ "step": 39
+ },
+ {
+ "epoch": 0.5688888888888889,
+ "grad_norm": 0.009079818613827229,
+ "learning_rate": 9.805804412106197e-06,
+ "loss": 0.5784010887145996,
+ "step": 40
+ },
+ {
+ "epoch": 0.5831111111111111,
+ "grad_norm": 0.008139098063111305,
+ "learning_rate": 9.782457878588977e-06,
+ "loss": 0.46650874614715576,
+ "step": 41
+ },
+ {
+ "epoch": 0.5973333333333334,
+ "grad_norm": 0.007847482338547707,
+ "learning_rate": 9.75781752354311e-06,
+ "loss": 0.42804157733917236,
+ "step": 42
+ },
+ {
+ "epoch": 0.6115555555555555,
+ "grad_norm": 0.008446140214800835,
+ "learning_rate": 9.731890013043367e-06,
+ "loss": 0.48224180936813354,
+ "step": 43
+ },
+ {
+ "epoch": 0.6257777777777778,
+ "grad_norm": 0.008632696233689785,
+ "learning_rate": 9.704682361384941e-06,
+ "loss": 0.4561079144477844,
+ "step": 44
+ },
+ {
+ "epoch": 0.64,
+ "grad_norm": 0.008637293241918087,
+ "learning_rate": 9.676201929185809e-06,
+ "loss": 0.4007866382598877,
+ "step": 45
+ },
+ {
+ "epoch": 0.6542222222222223,
+ "grad_norm": 0.008761093020439148,
+ "learning_rate": 9.646456421395447e-06,
+ "loss": 0.4587845802307129,
+ "step": 46
+ },
+ {
+ "epoch": 0.6684444444444444,
+ "grad_norm": 0.008726399391889572,
+ "learning_rate": 9.615453885210368e-06,
+ "loss": 0.4571158289909363,
+ "step": 47
+ },
+ {
+ "epoch": 0.6826666666666666,
+ "grad_norm": 0.008282204158604145,
+ "learning_rate": 9.583202707897075e-06,
+ "loss": 0.4724605679512024,
+ "step": 48
+ },
+ {
+ "epoch": 0.6968888888888889,
+ "grad_norm": 0.008231760933995247,
+ "learning_rate": 9.549711614523007e-06,
+ "loss": 0.4600468575954437,
+ "step": 49
+ },
+ {
+ "epoch": 0.7111111111111111,
+ "grad_norm": 0.00751511100679636,
+ "learning_rate": 9.514989665596114e-06,
+ "loss": 0.4377005994319916,
+ "step": 50
+ },
+ {
+ "epoch": 0.7111111111111111,
+ "eval_loss": 0.4685295820236206,
+ "eval_runtime": 5.4716,
+ "eval_samples_per_second": 36.553,
+ "eval_steps_per_second": 18.276,
+ "step": 50
+ },
+ {
+ "epoch": 0.7253333333333334,
+ "grad_norm": 0.008261662907898426,
+ "learning_rate": 9.479046254613673e-06,
+ "loss": 0.4415626525878906,
+ "step": 51
+ },
+ {
+ "epoch": 0.7395555555555555,
+ "grad_norm": 0.007395230233669281,
+ "learning_rate": 9.441891105521005e-06,
+ "loss": 0.42561036348342896,
+ "step": 52
+ },
+ {
+ "epoch": 0.7537777777777778,
+ "grad_norm": 0.008451394736766815,
+ "learning_rate": 9.40353427008083e-06,
+ "loss": 0.4630648195743561,
+ "step": 53
+ },
+ {
+ "epoch": 0.768,
+ "grad_norm": 0.007475187536329031,
+ "learning_rate": 9.3639861251539e-06,
+ "loss": 0.43418699502944946,
+ "step": 54
+ },
+ {
+ "epoch": 0.7822222222222223,
+ "grad_norm": 0.00859637837857008,
+ "learning_rate": 9.323257369891702e-06,
+ "loss": 0.47063255310058594,
+ "step": 55
+ },
+ {
+ "epoch": 0.7964444444444444,
+ "grad_norm": 0.007597765419632196,
+ "learning_rate": 9.281359022841966e-06,
+ "loss": 0.41920918226242065,
+ "step": 56
+ },
+ {
+ "epoch": 0.8106666666666666,
+ "grad_norm": 0.018636908382177353,
+ "learning_rate": 9.238302418967757e-06,
+ "loss": 0.4757845401763916,
+ "step": 57
+ },
+ {
+ "epoch": 0.8248888888888889,
+ "grad_norm": 0.008423218503594398,
+ "learning_rate": 9.194099206580981e-06,
+ "loss": 0.5198032855987549,
+ "step": 58
+ },
+ {
+ "epoch": 0.8391111111111111,
+ "grad_norm": 0.008149159140884876,
+ "learning_rate": 9.14876134419111e-06,
+ "loss": 0.46035224199295044,
+ "step": 59
+ },
+ {
+ "epoch": 0.8533333333333334,
+ "grad_norm": 0.00819750688970089,
+ "learning_rate": 9.102301097269974e-06,
+ "loss": 0.4945712983608246,
+ "step": 60
+ },
+ {
+ "epoch": 0.8675555555555555,
+ "grad_norm": 0.00855066254734993,
+ "learning_rate": 9.05473103493355e-06,
+ "loss": 0.4513372778892517,
+ "step": 61
+ },
+ {
+ "epoch": 0.8817777777777778,
+ "grad_norm": 0.008660132065415382,
+ "learning_rate": 9.006064026541549e-06,
+ "loss": 0.49865084886550903,
+ "step": 62
+ },
+ {
+ "epoch": 0.896,
+ "grad_norm": 0.007741494104266167,
+ "learning_rate": 8.956313238215824e-06,
+ "loss": 0.4319220185279846,
+ "step": 63
+ },
+ {
+ "epoch": 0.9102222222222223,
+ "grad_norm": 0.007900395430624485,
+ "learning_rate": 8.905492129278478e-06,
+ "loss": 0.43760013580322266,
+ "step": 64
+ },
+ {
+ "epoch": 0.9244444444444444,
+ "grad_norm": 0.007346940692514181,
+ "learning_rate": 8.85361444861063e-06,
+ "loss": 0.41593095660209656,
+ "step": 65
+ },
+ {
+ "epoch": 0.9386666666666666,
+ "grad_norm": 0.00902250874787569,
+ "learning_rate": 8.800694230932885e-06,
+ "loss": 0.48464590311050415,
+ "step": 66
+ },
+ {
+ "epoch": 0.9528888888888889,
+ "grad_norm": 0.007672084029763937,
+ "learning_rate": 8.74674579300843e-06,
+ "loss": 0.41884416341781616,
+ "step": 67
+ },
+ {
+ "epoch": 0.9671111111111111,
+ "grad_norm": 0.008009645156562328,
+ "learning_rate": 8.691783729769874e-06,
+ "loss": 0.40650737285614014,
+ "step": 68
+ },
+ {
+ "epoch": 0.9813333333333333,
+ "grad_norm": 0.008077424019575119,
+ "learning_rate": 8.635822910370793e-06,
+ "loss": 0.4726918637752533,
+ "step": 69
+ },
+ {
+ "epoch": 0.9955555555555555,
+ "grad_norm": 0.007576521951705217,
+ "learning_rate": 8.578878474163115e-06,
+ "loss": 0.4404842257499695,
+ "step": 70
+ },
+ {
+ "epoch": 1.0,
+ "grad_norm": 0.012638773769140244,
+ "learning_rate": 8.520965826601394e-06,
+ "loss": 0.42633137106895447,
+ "step": 71
+ },
+ {
+ "epoch": 1.0142222222222221,
+ "grad_norm": 0.008662556298077106,
+ "learning_rate": 8.462100635075097e-06,
+ "loss": 0.46000930666923523,
+ "step": 72
+ },
+ {
+ "epoch": 1.0284444444444445,
+ "grad_norm": 0.008270082995295525,
+ "learning_rate": 8.40229882467003e-06,
+ "loss": 0.42150819301605225,
+ "step": 73
+ },
+ {
+ "epoch": 1.0426666666666666,
+ "grad_norm": 0.007625850848853588,
+ "learning_rate": 8.341576573860049e-06,
+ "loss": 0.37313175201416016,
+ "step": 74
+ },
+ {
+ "epoch": 1.056888888888889,
+ "grad_norm": 0.00790152233093977,
+ "learning_rate": 8.279950310130218e-06,
+ "loss": 0.382789671421051,
+ "step": 75
+ },
+ {
+ "epoch": 1.0711111111111111,
+ "grad_norm": 0.007986027747392654,
+ "learning_rate": 8.2174367055326e-06,
+ "loss": 0.3734534978866577,
+ "step": 76
+ },
+ {
+ "epoch": 1.0853333333333333,
+ "grad_norm": 0.007174879312515259,
+ "learning_rate": 8.154052672175888e-06,
+ "loss": 0.37693876028060913,
+ "step": 77
+ },
+ {
+ "epoch": 1.0995555555555556,
+ "grad_norm": 0.008486706763505936,
+ "learning_rate": 8.08981535765009e-06,
+ "loss": 0.4265296459197998,
+ "step": 78
+ },
+ {
+ "epoch": 1.1137777777777778,
+ "grad_norm": 0.007609546184539795,
+ "learning_rate": 8.024742140387506e-06,
+ "loss": 0.3812956213951111,
+ "step": 79
+ },
+ {
+ "epoch": 1.1280000000000001,
+ "grad_norm": 0.007912754081189632,
+ "learning_rate": 7.95885062496126e-06,
+ "loss": 0.4131423234939575,
+ "step": 80
+ },
+ {
+ "epoch": 1.1422222222222222,
+ "grad_norm": 0.007641648408025503,
+ "learning_rate": 7.892158637322647e-06,
+ "loss": 0.37861311435699463,
+ "step": 81
+ },
+ {
+ "epoch": 1.1564444444444444,
+ "grad_norm": 0.007570190355181694,
+ "learning_rate": 7.824684219978591e-06,
+ "loss": 0.3589654564857483,
+ "step": 82
+ },
+ {
+ "epoch": 1.1706666666666667,
+ "grad_norm": 0.007670491933822632,
+ "learning_rate": 7.756445627110523e-06,
+ "loss": 0.40794360637664795,
+ "step": 83
+ },
+ {
+ "epoch": 1.1848888888888889,
+ "grad_norm": 0.009116973727941513,
+ "learning_rate": 7.68746131963598e-06,
+ "loss": 0.4278006851673126,
+ "step": 84
+ },
+ {
+ "epoch": 1.199111111111111,
+ "grad_norm": 0.007852817885577679,
+ "learning_rate": 7.6177499602143e-06,
+ "loss": 0.371395081281662,
+ "step": 85
+ },
+ {
+ "epoch": 1.2133333333333334,
+ "grad_norm": 0.00870604906231165,
+ "learning_rate": 7.547330408197695e-06,
+ "loss": 0.4305034279823303,
+ "step": 86
+ },
+ {
+ "epoch": 1.2275555555555555,
+ "grad_norm": 0.008404702879488468,
+ "learning_rate": 7.476221714529167e-06,
+ "loss": 0.4348181188106537,
+ "step": 87
+ },
+ {
+ "epoch": 1.2417777777777779,
+ "grad_norm": 0.008913768455386162,
+ "learning_rate": 7.404443116588548e-06,
+ "loss": 0.4309384822845459,
+ "step": 88
+ },
+ {
+ "epoch": 1.256,
+ "grad_norm": 0.007682010997086763,
+ "learning_rate": 7.332014032988123e-06,
+ "loss": 0.4107038676738739,
+ "step": 89
+ },
+ {
+ "epoch": 1.2702222222222221,
+ "grad_norm": 0.00787457637488842,
+ "learning_rate": 7.2589540583192165e-06,
+ "loss": 0.37457963824272156,
+ "step": 90
+ },
+ {
+ "epoch": 1.2844444444444445,
+ "grad_norm": 0.007768277078866959,
+ "learning_rate": 7.185282957851175e-06,
+ "loss": 0.37697723507881165,
+ "step": 91
+ },
+ {
+ "epoch": 1.2986666666666666,
+ "grad_norm": 0.008521760813891888,
+ "learning_rate": 7.111020662184174e-06,
+ "loss": 0.40353477001190186,
+ "step": 92
+ },
+ {
+ "epoch": 1.3128888888888888,
+ "grad_norm": 0.0077499281615018845,
+ "learning_rate": 7.036187261857289e-06,
+ "loss": 0.37755411863327026,
+ "step": 93
+ },
+ {
+ "epoch": 1.3271111111111111,
+ "grad_norm": 0.00870734266936779,
+ "learning_rate": 6.960803001913315e-06,
+ "loss": 0.3729894757270813,
+ "step": 94
+ },
+ {
+ "epoch": 1.3413333333333333,
+ "grad_norm": 0.007563227321952581,
+ "learning_rate": 6.884888276421766e-06,
+ "loss": 0.4091319441795349,
+ "step": 95
+ },
+ {
+ "epoch": 1.3555555555555556,
+ "grad_norm": 0.00803642813116312,
+ "learning_rate": 6.8084636229615786e-06,
+ "loss": 0.4048840403556824,
+ "step": 96
+ },
+ {
+ "epoch": 1.3697777777777778,
+ "grad_norm": 0.00779891200363636,
+ "learning_rate": 6.731549717064975e-06,
+ "loss": 0.39409685134887695,
+ "step": 97
+ },
+ {
+ "epoch": 1.384,
+ "grad_norm": 0.008260682225227356,
+ "learning_rate": 6.654167366624009e-06,
+ "loss": 0.36237582564353943,
+ "step": 98
+ },
+ {
+ "epoch": 1.3982222222222223,
+ "grad_norm": 0.008373094722628593,
+ "learning_rate": 6.576337506261314e-06,
+ "loss": 0.38763654232025146,
+ "step": 99
+ },
+ {
+ "epoch": 1.4124444444444444,
+ "grad_norm": 0.007797387894243002,
+ "learning_rate": 6.498081191666549e-06,
+ "loss": 0.41751331090927124,
+ "step": 100
+ },
+ {
+ "epoch": 1.4124444444444444,
+ "eval_loss": 0.4669477939605713,
+ "eval_runtime": 5.2715,
+ "eval_samples_per_second": 37.94,
+ "eval_steps_per_second": 18.97,
+ "step": 100
+ },
+ {
+ "epoch": 1.4266666666666667,
+ "grad_norm": 0.007338334806263447,
+ "learning_rate": 6.419419593900109e-06,
+ "loss": 0.38504236936569214,
+ "step": 101
+ },
+ {
+ "epoch": 1.4408888888888889,
+ "grad_norm": 0.007528901565819979,
+ "learning_rate": 6.340373993665607e-06,
+ "loss": 0.36031031608581543,
+ "step": 102
+ },
+ {
+ "epoch": 1.455111111111111,
+ "grad_norm": 0.007573080714792013,
+ "learning_rate": 6.2609657755527135e-06,
+ "loss": 0.3863949179649353,
+ "step": 103
+ },
+ {
+ "epoch": 1.4693333333333334,
+ "grad_norm": 0.008099780417978764,
+ "learning_rate": 6.1812164222518626e-06,
+ "loss": 0.3955462574958801,
+ "step": 104
+ },
+ {
+ "epoch": 1.4835555555555555,
+ "grad_norm": 0.008717137388885021,
+ "learning_rate": 6.101147508742456e-06,
+ "loss": 0.39310169219970703,
+ "step": 105
+ },
+ {
+ "epoch": 1.4977777777777779,
+ "grad_norm": 0.008383029140532017,
+ "learning_rate": 6.020780696456059e-06,
+ "loss": 0.4533539414405823,
+ "step": 106
+ },
+ {
+ "epoch": 1.512,
+ "grad_norm": 0.007702656090259552,
+ "learning_rate": 5.940137727416247e-06,
+ "loss": 0.3985057473182678,
+ "step": 107
+ },
+ {
+ "epoch": 1.5262222222222221,
+ "grad_norm": 0.00808175653219223,
+ "learning_rate": 5.859240418356614e-06,
+ "loss": 0.3878968358039856,
+ "step": 108
+ },
+ {
+ "epoch": 1.5404444444444443,
+ "grad_norm": 0.008781371638178825,
+ "learning_rate": 5.778110654818602e-06,
+ "loss": 0.39648741483688354,
+ "step": 109
+ },
+ {
+ "epoch": 1.5546666666666666,
+ "grad_norm": 0.008896375074982643,
+ "learning_rate": 5.696770385230679e-06,
+ "loss": 0.4319821000099182,
+ "step": 110
+ },
+ {
+ "epoch": 1.568888888888889,
+ "grad_norm": 0.008202116936445236,
+ "learning_rate": 5.615241614970546e-06,
+ "loss": 0.4078192710876465,
+ "step": 111
+ },
+ {
+ "epoch": 1.5831111111111111,
+ "grad_norm": 0.007843846455216408,
+ "learning_rate": 5.533546400411899e-06,
+ "loss": 0.38504159450531006,
+ "step": 112
+ },
+ {
+ "epoch": 1.5973333333333333,
+ "grad_norm": 0.00823921337723732,
+ "learning_rate": 5.4517068429574215e-06,
+ "loss": 0.3971819579601288,
+ "step": 113
+ },
+ {
+ "epoch": 1.6115555555555554,
+ "grad_norm": 0.007411513011902571,
+ "learning_rate": 5.3697450830595775e-06,
+ "loss": 0.3886791169643402,
+ "step": 114
+ },
+ {
+ "epoch": 1.6257777777777778,
+ "grad_norm": 0.007902945391833782,
+ "learning_rate": 5.287683294230855e-06,
+ "loss": 0.37610527873039246,
+ "step": 115
+ },
+ {
+ "epoch": 1.6400000000000001,
+ "grad_norm": 0.007778776343911886,
+ "learning_rate": 5.20554367704505e-06,
+ "loss": 0.3846814036369324,
+ "step": 116
+ },
+ {
+ "epoch": 1.6542222222222223,
+ "grad_norm": 0.008897176012396812,
+ "learning_rate": 5.123348453131242e-06,
+ "loss": 0.3917155861854553,
+ "step": 117
+ },
+ {
+ "epoch": 1.6684444444444444,
+ "grad_norm": 0.008548559620976448,
+ "learning_rate": 5.041119859162068e-06,
+ "loss": 0.39533206820487976,
+ "step": 118
+ },
+ {
+ "epoch": 1.6826666666666665,
+ "grad_norm": 0.008137239143252373,
+ "learning_rate": 4.958880140837934e-06,
+ "loss": 0.4113273620605469,
+ "step": 119
+ },
+ {
+ "epoch": 1.696888888888889,
+ "grad_norm": 0.008478965610265732,
+ "learning_rate": 4.876651546868759e-06,
+ "loss": 0.39543214440345764,
+ "step": 120
+ },
+ {
+ "epoch": 1.7111111111111112,
+ "grad_norm": 0.00835413858294487,
+ "learning_rate": 4.794456322954953e-06,
+ "loss": 0.4548013508319855,
+ "step": 121
+ },
+ {
+ "epoch": 1.7253333333333334,
+ "grad_norm": 0.00832394976168871,
+ "learning_rate": 4.7123167057691446e-06,
+ "loss": 0.4111294150352478,
+ "step": 122
+ },
+ {
+ "epoch": 1.7395555555555555,
+ "grad_norm": 0.008472571149468422,
+ "learning_rate": 4.630254916940424e-06,
+ "loss": 0.43050098419189453,
+ "step": 123
+ },
+ {
+ "epoch": 1.7537777777777777,
+ "grad_norm": 0.00816201139241457,
+ "learning_rate": 4.548293157042581e-06,
+ "loss": 0.41356489062309265,
+ "step": 124
+ },
+ {
+ "epoch": 1.768,
+ "grad_norm": 0.008180899545550346,
+ "learning_rate": 4.466453599588103e-06,
+ "loss": 0.41799914836883545,
+ "step": 125
+ },
+ {
+ "epoch": 1.7822222222222224,
+ "grad_norm": 0.007402091287076473,
+ "learning_rate": 4.384758385029457e-06,
+ "loss": 0.3510154187679291,
+ "step": 126
+ },
+ {
+ "epoch": 1.7964444444444445,
+ "grad_norm": 0.008476709015667439,
+ "learning_rate": 4.3032296147693225e-06,
+ "loss": 0.4098949432373047,
+ "step": 127
+ },
+ {
+ "epoch": 1.8106666666666666,
+ "grad_norm": 0.008135325275361538,
+ "learning_rate": 4.2218893451814005e-06,
+ "loss": 0.36957240104675293,
+ "step": 128
+ },
+ {
+ "epoch": 1.8248888888888888,
+ "grad_norm": 0.007964376360177994,
+ "learning_rate": 4.140759581643386e-06,
+ "loss": 0.3535369038581848,
+ "step": 129
+ },
+ {
+ "epoch": 1.8391111111111111,
+ "grad_norm": 0.00742813665419817,
+ "learning_rate": 4.059862272583755e-06,
+ "loss": 0.38296306133270264,
+ "step": 130
+ },
+ {
+ "epoch": 1.8533333333333335,
+ "grad_norm": 0.007901367731392384,
+ "learning_rate": 3.979219303543942e-06,
+ "loss": 0.38752996921539307,
+ "step": 131
+ },
+ {
+ "epoch": 1.8675555555555556,
+ "grad_norm": 0.008205283433198929,
+ "learning_rate": 3.898852491257547e-06,
+ "loss": 0.3867356479167938,
+ "step": 132
+ },
+ {
+ "epoch": 1.8817777777777778,
+ "grad_norm": 0.008251599036157131,
+ "learning_rate": 3.818783577748138e-06,
+ "loss": 0.42934614419937134,
+ "step": 133
+ },
+ {
+ "epoch": 1.896,
+ "grad_norm": 0.008710331283509731,
+ "learning_rate": 3.7390342244472886e-06,
+ "loss": 0.4096015691757202,
+ "step": 134
+ },
+ {
+ "epoch": 1.9102222222222223,
+ "grad_norm": 0.008127489127218723,
+ "learning_rate": 3.659626006334395e-06,
+ "loss": 0.3849722743034363,
+ "step": 135
+ },
+ {
+ "epoch": 1.9244444444444444,
+ "grad_norm": 0.008900986984372139,
+ "learning_rate": 3.5805804060998926e-06,
+ "loss": 0.415260910987854,
+ "step": 136
+ },
+ {
+ "epoch": 1.9386666666666668,
+ "grad_norm": 0.00907993782311678,
+ "learning_rate": 3.501918808333453e-06,
+ "loss": 0.48382842540740967,
+ "step": 137
+ },
+ {
+ "epoch": 1.952888888888889,
+ "grad_norm": 0.008263039402663708,
+ "learning_rate": 3.4236624937386874e-06,
+ "loss": 0.4434107542037964,
+ "step": 138
+ },
+ {
+ "epoch": 1.967111111111111,
+ "grad_norm": 0.008353665471076965,
+ "learning_rate": 3.3458326333759927e-06,
+ "loss": 0.41760051250457764,
+ "step": 139
+ },
+ {
+ "epoch": 1.9813333333333332,
+ "grad_norm": 0.008556668646633625,
+ "learning_rate": 3.268450282935026e-06,
+ "loss": 0.43642479181289673,
+ "step": 140
+ },
+ {
+ "epoch": 1.9955555555555555,
+ "grad_norm": 0.008040979504585266,
+ "learning_rate": 3.1915363770384223e-06,
+ "loss": 0.38327687978744507,
+ "step": 141
+ },
+ {
+ "epoch": 2.0,
+ "grad_norm": 0.012571632862091064,
+ "learning_rate": 3.1151117235782346e-06,
+ "loss": 0.3751964867115021,
+ "step": 142
+ },
+ {
+ "epoch": 2.014222222222222,
+ "grad_norm": 0.008403269574046135,
+ "learning_rate": 3.0391969980866874e-06,
+ "loss": 0.3831713795661926,
+ "step": 143
+ },
+ {
+ "epoch": 2.0284444444444443,
+ "grad_norm": 0.008236616849899292,
+ "learning_rate": 2.963812738142713e-06,
+ "loss": 0.338630735874176,
+ "step": 144
+ },
+ {
+ "epoch": 2.042666666666667,
+ "grad_norm": 0.008433214388787746,
+ "learning_rate": 2.8889793378158284e-06,
+ "loss": 0.3684050440788269,
+ "step": 145
+ },
+ {
+ "epoch": 2.056888888888889,
+ "grad_norm": 0.008040281943976879,
+ "learning_rate": 2.814717042148827e-06,
+ "loss": 0.37092137336730957,
+ "step": 146
+ },
+ {
+ "epoch": 2.071111111111111,
+ "grad_norm": 0.008197156712412834,
+ "learning_rate": 2.7410459416807856e-06,
+ "loss": 0.34995853900909424,
+ "step": 147
+ },
+ {
+ "epoch": 2.0853333333333333,
+ "grad_norm": 0.009072407148778439,
+ "learning_rate": 2.6679859670118785e-06,
+ "loss": 0.37355348467826843,
+ "step": 148
+ },
+ {
+ "epoch": 2.0995555555555554,
+ "grad_norm": 0.00807118322700262,
+ "learning_rate": 2.5955568834114523e-06,
+ "loss": 0.3448713421821594,
+ "step": 149
+ },
+ {
+ "epoch": 2.113777777777778,
+ "grad_norm": 0.00770068122074008,
+ "learning_rate": 2.523778285470835e-06,
+ "loss": 0.3545028865337372,
+ "step": 150
+ },
+ {
+ "epoch": 2.113777777777778,
+ "eval_loss": 0.4679063558578491,
+ "eval_runtime": 5.4161,
+ "eval_samples_per_second": 36.927,
+ "eval_steps_per_second": 18.463,
+ "step": 150
+ },
+ {
+ "epoch": 2.128,
+ "grad_norm": 0.007919526658952236,
+ "learning_rate": 2.452669591802307e-06,
+ "loss": 0.32551515102386475,
+ "step": 151
+ },
+ {
+ "epoch": 2.1422222222222222,
+ "grad_norm": 0.007321993354707956,
+ "learning_rate": 2.3822500397857016e-06,
+ "loss": 0.35764873027801514,
+ "step": 152
+ },
+ {
+ "epoch": 2.1564444444444444,
+ "grad_norm": 0.0076365480199456215,
+ "learning_rate": 2.3125386803640188e-06,
+ "loss": 0.35795581340789795,
+ "step": 153
+ },
+ {
+ "epoch": 2.1706666666666665,
+ "grad_norm": 0.007587647996842861,
+ "learning_rate": 2.243554372889479e-06,
+ "loss": 0.32651567459106445,
+ "step": 154
+ },
+ {
+ "epoch": 2.1848888888888887,
+ "grad_norm": 0.007799312938004732,
+ "learning_rate": 2.1753157800214107e-06,
+ "loss": 0.3542096018791199,
+ "step": 155
+ },
+ {
+ "epoch": 2.1991111111111112,
+ "grad_norm": 0.007910694926977158,
+ "learning_rate": 2.1078413626773547e-06,
+ "loss": 0.3560444116592407,
+ "step": 156
+ },
+ {
+ "epoch": 2.2133333333333334,
+ "grad_norm": 0.008300457149744034,
+ "learning_rate": 2.0411493750387423e-06,
+ "loss": 0.3283962607383728,
+ "step": 157
+ },
+ {
+ "epoch": 2.2275555555555555,
+ "grad_norm": 0.007787623442709446,
+ "learning_rate": 1.9752578596124955e-06,
+ "loss": 0.362970769405365,
+ "step": 158
+ },
+ {
+ "epoch": 2.2417777777777776,
+ "grad_norm": 0.007931886240839958,
+ "learning_rate": 1.9101846423499113e-06,
+ "loss": 0.3372136354446411,
+ "step": 159
+ },
+ {
+ "epoch": 2.2560000000000002,
+ "grad_norm": 0.008211000822484493,
+ "learning_rate": 1.8459473278241125e-06,
+ "loss": 0.343697726726532,
+ "step": 160
+ },
+ {
+ "epoch": 2.2702222222222224,
+ "grad_norm": 0.008033247664570808,
+ "learning_rate": 1.7825632944674016e-06,
+ "loss": 0.3472204804420471,
+ "step": 161
+ },
+ {
+ "epoch": 2.2844444444444445,
+ "grad_norm": 0.008930593729019165,
+ "learning_rate": 1.7200496898697832e-06,
+ "loss": 0.34057295322418213,
+ "step": 162
+ },
+ {
+ "epoch": 2.2986666666666666,
+ "grad_norm": 0.007960367947816849,
+ "learning_rate": 1.6584234261399535e-06,
+ "loss": 0.3398140072822571,
+ "step": 163
+ },
+ {
+ "epoch": 2.3128888888888888,
+ "grad_norm": 0.008388614282011986,
+ "learning_rate": 1.5977011753299726e-06,
+ "loss": 0.3423030376434326,
+ "step": 164
+ },
+ {
+ "epoch": 2.327111111111111,
+ "grad_norm": 0.009143203496932983,
+ "learning_rate": 1.5378993649249053e-06,
+ "loss": 0.4059848189353943,
+ "step": 165
+ },
+ {
+ "epoch": 2.3413333333333335,
+ "grad_norm": 0.008042566478252411,
+ "learning_rate": 1.4790341733986085e-06,
+ "loss": 0.35659873485565186,
+ "step": 166
+ },
+ {
+ "epoch": 2.3555555555555556,
+ "grad_norm": 0.007340035866945982,
+ "learning_rate": 1.4211215258368866e-06,
+ "loss": 0.320560097694397,
+ "step": 167
+ },
+ {
+ "epoch": 2.3697777777777778,
+ "grad_norm": 0.008758354000747204,
+ "learning_rate": 1.3641770896292083e-06,
+ "loss": 0.3432486057281494,
+ "step": 168
+ },
+ {
+ "epoch": 2.384,
+ "grad_norm": 0.008217270486056805,
+ "learning_rate": 1.3082162702301276e-06,
+ "loss": 0.3448861241340637,
+ "step": 169
+ },
+ {
+ "epoch": 2.398222222222222,
+ "grad_norm": 0.008302765898406506,
+ "learning_rate": 1.2532542069915722e-06,
+ "loss": 0.4012100100517273,
+ "step": 170
+ },
+ {
+ "epoch": 2.4124444444444446,
+ "grad_norm": 0.0086008096113801,
+ "learning_rate": 1.1993057690671174e-06,
+ "loss": 0.35434189438819885,
+ "step": 171
+ },
+ {
+ "epoch": 2.4266666666666667,
+ "grad_norm": 0.009528086520731449,
+ "learning_rate": 1.1463855513893695e-06,
+ "loss": 0.37489616870880127,
+ "step": 172
+ },
+ {
+ "epoch": 2.440888888888889,
+ "grad_norm": 0.008264975622296333,
+ "learning_rate": 1.0945078707215224e-06,
+ "loss": 0.3599584698677063,
+ "step": 173
+ },
+ {
+ "epoch": 2.455111111111111,
+ "grad_norm": 0.008293098770081997,
+ "learning_rate": 1.0436867617841768e-06,
+ "loss": 0.38492339849472046,
+ "step": 174
+ },
+ {
+ "epoch": 2.469333333333333,
+ "grad_norm": 0.007827190682291985,
+ "learning_rate": 9.939359734584552e-07,
+ "loss": 0.36968979239463806,
+ "step": 175
+ },
+ {
+ "epoch": 2.4835555555555557,
+ "grad_norm": 0.008390667848289013,
+ "learning_rate": 9.452689650664515e-07,
+ "loss": 0.3575425148010254,
+ "step": 176
+ },
+ {
+ "epoch": 2.497777777777778,
+ "grad_norm": 0.007562275975942612,
+ "learning_rate": 8.976989027300265e-07,
+ "loss": 0.35382080078125,
+ "step": 177
+ },
+ {
+ "epoch": 2.512,
+ "grad_norm": 0.007893560454249382,
+ "learning_rate": 8.512386558088919e-07,
+ "loss": 0.35292625427246094,
+ "step": 178
+ },
+ {
+ "epoch": 2.526222222222222,
+ "grad_norm": 0.007973878644406796,
+ "learning_rate": 8.059007934190194e-07,
+ "loss": 0.3658143877983093,
+ "step": 179
+ },
+ {
+ "epoch": 2.5404444444444443,
+ "grad_norm": 0.008489987812936306,
+ "learning_rate": 7.61697581032243e-07,
+ "loss": 0.34945034980773926,
+ "step": 180
+ },
+ {
+ "epoch": 2.554666666666667,
+ "grad_norm": 0.007993645034730434,
+ "learning_rate": 7.186409771580355e-07,
+ "loss": 0.3492674231529236,
+ "step": 181
+ },
+ {
+ "epoch": 2.568888888888889,
+ "grad_norm": 0.007816205732524395,
+ "learning_rate": 6.76742630108298e-07,
+ "loss": 0.35949787497520447,
+ "step": 182
+ },
+ {
+ "epoch": 2.583111111111111,
+ "grad_norm": 0.00765154417604208,
+ "learning_rate": 6.360138748461015e-07,
+ "loss": 0.35065189003944397,
+ "step": 183
+ },
+ {
+ "epoch": 2.5973333333333333,
+ "grad_norm": 0.007598839234560728,
+ "learning_rate": 5.964657299191712e-07,
+ "loss": 0.31525540351867676,
+ "step": 184
+ },
+ {
+ "epoch": 2.6115555555555554,
+ "grad_norm": 0.00796365924179554,
+ "learning_rate": 5.581088944789953e-07,
+ "loss": 0.34680360555648804,
+ "step": 185
+ },
+ {
+ "epoch": 2.6257777777777775,
+ "grad_norm": 0.008079800754785538,
+ "learning_rate": 5.20953745386329e-07,
+ "loss": 0.3423649072647095,
+ "step": 186
+ },
+ {
+ "epoch": 2.64,
+ "grad_norm": 0.008171727880835533,
+ "learning_rate": 4.850103344038853e-07,
+ "loss": 0.3583834767341614,
+ "step": 187
+ },
+ {
+ "epoch": 2.6542222222222223,
+ "grad_norm": 0.007568752858787775,
+ "learning_rate": 4.502883854769935e-07,
+ "loss": 0.3272634744644165,
+ "step": 188
+ },
+ {
+ "epoch": 2.6684444444444444,
+ "grad_norm": 0.00814193394035101,
+ "learning_rate": 4.167972921029262e-07,
+ "loss": 0.3352956175804138,
+ "step": 189
+ },
+ {
+ "epoch": 2.6826666666666665,
+ "grad_norm": 0.008751407265663147,
+ "learning_rate": 3.8454611478963235e-07,
+ "loss": 0.3704286515712738,
+ "step": 190
+ },
+ {
+ "epoch": 2.696888888888889,
+ "grad_norm": 0.007807895075529814,
+ "learning_rate": 3.535435786045538e-07,
+ "loss": 0.34566229581832886,
+ "step": 191
+ },
+ {
+ "epoch": 2.7111111111111112,
+ "grad_norm": 0.007984541356563568,
+ "learning_rate": 3.237980708141919e-07,
+ "loss": 0.35052379965782166,
+ "step": 192
+ },
+ {
+ "epoch": 2.7253333333333334,
+ "grad_norm": 0.009312397800385952,
+ "learning_rate": 2.9531763861505967e-07,
+ "loss": 0.38708382844924927,
+ "step": 193
+ },
+ {
+ "epoch": 2.7395555555555555,
+ "grad_norm": 0.008055887185037136,
+ "learning_rate": 2.681099869566328e-07,
+ "loss": 0.3316305875778198,
+ "step": 194
+ },
+ {
+ "epoch": 2.7537777777777777,
+ "grad_norm": 0.008029562421143055,
+ "learning_rate": 2.4218247645689306e-07,
+ "loss": 0.3650358319282532,
+ "step": 195
+ },
+ {
+ "epoch": 2.768,
+ "grad_norm": 0.00823246967047453,
+ "learning_rate": 2.175421214110235e-07,
+ "loss": 0.36560434103012085,
+ "step": 196
+ },
+ {
+ "epoch": 2.7822222222222224,
+ "grad_norm": 0.008692881092429161,
+ "learning_rate": 1.941955878938029e-07,
+ "loss": 0.3463239371776581,
+ "step": 197
+ },
+ {
+ "epoch": 2.7964444444444445,
+ "grad_norm": 0.008173434063792229,
+ "learning_rate": 1.721491919561913e-07,
+ "loss": 0.35629892349243164,
+ "step": 198
+ },
+ {
+ "epoch": 2.8106666666666666,
+ "grad_norm": 0.00767622422426939,
+ "learning_rate": 1.514088979166256e-07,
+ "loss": 0.33887457847595215,
+ "step": 199
+ },
+ {
+ "epoch": 2.824888888888889,
+ "grad_norm": 0.007811338175088167,
+ "learning_rate": 1.3198031674745814e-07,
+ "loss": 0.36586421728134155,
+ "step": 200
+ },
+ {
+ "epoch": 2.824888888888889,
+ "eval_loss": 0.4734371602535248,
+ "eval_runtime": 7.1506,
+ "eval_samples_per_second": 27.97,
+ "eval_steps_per_second": 13.985,
+ "step": 200
+ },
+ {
+ "epoch": 2.8391111111111114,
+ "grad_norm": 0.008272381499409676,
+ "learning_rate": 1.138687045569975e-07,
+ "loss": 0.3514959216117859,
+ "step": 201
+ },
+ {
+ "epoch": 2.8533333333333335,
+ "grad_norm": 0.008001282811164856,
+ "learning_rate": 9.70789611675449e-08,
+ "loss": 0.37950894236564636,
+ "step": 202
+ },
+ {
+ "epoch": 2.8675555555555556,
+ "grad_norm": 0.007416888140141964,
+ "learning_rate": 8.161562878982399e-08,
+ "loss": 0.3352017402648926,
+ "step": 203
+ },
+ {
+ "epoch": 2.8817777777777778,
+ "grad_norm": 0.008772965520620346,
+ "learning_rate": 6.74828907941516e-08,
+ "loss": 0.36504414677619934,
+ "step": 204
+ },
+ {
+ "epoch": 2.896,
+ "grad_norm": 0.008397468365728855,
+ "learning_rate": 5.468457057869358e-08,
+ "loss": 0.3987666964530945,
+ "step": 205
+ },
+ {
+ "epoch": 2.910222222222222,
+ "grad_norm": 0.00775982066988945,
+ "learning_rate": 4.322413053509944e-08,
+ "loss": 0.3345492482185364,
+ "step": 206
+ },
+ {
+ "epoch": 2.924444444444444,
+ "grad_norm": 0.008166249841451645,
+ "learning_rate": 3.31046711118066e-08,
+ "loss": 0.367769718170166,
+ "step": 207
+ },
+ {
+ "epoch": 2.9386666666666668,
+ "grad_norm": 0.007682916708290577,
+ "learning_rate": 2.4328929975260262e-08,
+ "loss": 0.34441590309143066,
+ "step": 208
+ },
+ {
+ "epoch": 2.952888888888889,
+ "grad_norm": 0.008051120676100254,
+ "learning_rate": 1.6899281269279756e-08,
+ "loss": 0.3472161889076233,
+ "step": 209
+ },
+ {
+ "epoch": 2.967111111111111,
+ "grad_norm": 0.0077741509303450584,
+ "learning_rate": 1.0817734972768946e-08,
+ "loss": 0.3906354308128357,
+ "step": 210
+ },
+ {
+ "epoch": 2.981333333333333,
+ "grad_norm": 0.008870133198797703,
+ "learning_rate": 6.085936355947897e-09,
+ "loss": 0.37977275252342224,
+ "step": 211
+ },
+ {
+ "epoch": 2.9955555555555557,
+ "grad_norm": 0.008871600031852722,
+ "learning_rate": 2.7051655352494654e-09,
+ "loss": 0.3533574342727661,
+ "step": 212
+ },
+ {
+ "epoch": 3.0,
+ "grad_norm": 0.014481575228273869,
+ "learning_rate": 6.763371270035457e-10,
+ "loss": 0.5198727250099182,
+ "step": 213
+ },
+ {
+ "epoch": 3.0,
+ "step": 213,
+ "total_flos": 1.0308479847759872e+17,
+ "train_loss": 0.4208552807989255,
+ "train_runtime": 5691.1072,
+ "train_samples_per_second": 9.488,
+ "train_steps_per_second": 0.037
+ }
+ ],
+ "logging_steps": 1,
+ "max_steps": 213,
+ "num_input_tokens_seen": 0,
+ "num_train_epochs": 3,
+ "save_steps": 500,
+ "stateful_callbacks": {
+ "TrainerControl": {
+ "args": {
+ "should_epoch_stop": false,
+ "should_evaluate": false,
+ "should_log": false,
+ "should_save": true,
+ "should_training_stop": true
+ },
+ "attributes": {}
+ }
+ },
+ "total_flos": 1.0308479847759872e+17,
+ "train_batch_size": 1,
+ "trial_name": null,
+ "trial_params": null
+}
diff --git a/qwen2.5-3b-instruct/training_args.bin b/qwen2.5-3b-instruct/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..bd95d1f874b25af121b3c8254bd05a3eb95ffd5f
--- /dev/null
+++ b/qwen2.5-3b-instruct/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c4f1847ec07466eeb896ef40e20d9241e19882e62efb61a14f1e7c6a96737fa0
+size 5649
diff --git a/qwen2.5-3b-instruct/training_eval_loss.png b/qwen2.5-3b-instruct/training_eval_loss.png
new file mode 100644
index 0000000000000000000000000000000000000000..031d405eca3cbba9e1458ddbb33b42e78d47c2d1
Binary files /dev/null and b/qwen2.5-3b-instruct/training_eval_loss.png differ
diff --git a/qwen2.5-3b-instruct/training_loss.png b/qwen2.5-3b-instruct/training_loss.png
new file mode 100644
index 0000000000000000000000000000000000000000..7be8b1427b4555ea729dbcb47f2695fc1c31613c
Binary files /dev/null and b/qwen2.5-3b-instruct/training_loss.png differ
diff --git a/qwen2.5-7b-instruct/README.md b/qwen2.5-7b-instruct/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..76711f9dd6d64f9fea6d5aeabcd37d1ee8fb4cb3
--- /dev/null
+++ b/qwen2.5-7b-instruct/README.md
@@ -0,0 +1,69 @@
+---
+library_name: transformers
+license: other
+base_model: Qwen/Qwen2.5-3B-Instruct
+tags:
+- llama-factory
+- full
+- generated_from_trainer
+model-index:
+- name: qwen2.5-7b-instruct
+ results: []
+---
+
+
+
+# qwen2.5-7b-instruct
+
+This model is a fine-tuned version of [Qwen/Qwen2.5-3B-Instruct](https://huggingface.co/Qwen/Qwen2.5-3B-Instruct) on the codealpaca_train_20k_no_trace_generation dataset.
+It achieves the following results on the evaluation set:
+- Loss: nan
+
+## Model description
+
+More information needed
+
+## Intended uses & limitations
+
+More information needed
+
+## Training and evaluation data
+
+More information needed
+
+## Training procedure
+
+### Training hyperparameters
+
+The following hyperparameters were used during training:
+- learning_rate: 1e-05
+- train_batch_size: 1
+- eval_batch_size: 1
+- seed: 42
+- distributed_type: multi-GPU
+- num_devices: 4
+- gradient_accumulation_steps: 64
+- total_train_batch_size: 256
+- total_eval_batch_size: 4
+- optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
+- lr_scheduler_type: cosine
+- lr_scheduler_warmup_steps: 0.1
+- num_epochs: 3
+
+### Training results
+
+| Training Loss | Epoch | Step | Validation Loss |
+|:-------------:|:------:|:----:|:---------------:|
+| 0.4456 | 0.7111 | 50 | 0.4940 |
+| 3.3619 | 1.4124 | 100 | 3.6941 |
+| 3.3854 | 2.1138 | 150 | 3.6941 |
+| 0.0 | 2.8249 | 200 | nan |
+
+
+### Framework versions
+
+- Transformers 5.2.0
+- Pytorch 2.11.0+cu128
+- Datasets 4.0.0
+- Tokenizers 0.22.2
diff --git a/qwen2.5-7b-instruct/all_results.json b/qwen2.5-7b-instruct/all_results.json
new file mode 100644
index 0000000000000000000000000000000000000000..d273245f739e1c598e140de0b8cf7878da80fdbd
--- /dev/null
+++ b/qwen2.5-7b-instruct/all_results.json
@@ -0,0 +1,12 @@
+{
+ "epoch": 3.0,
+ "eval_loss": NaN,
+ "eval_runtime": 5.4995,
+ "eval_samples_per_second": 36.367,
+ "eval_steps_per_second": 9.092,
+ "total_flos": 1.0308479847759872e+17,
+ "train_loss": 1.8930653027245696,
+ "train_runtime": 5673.0769,
+ "train_samples_per_second": 9.519,
+ "train_steps_per_second": 0.038
+}
\ No newline at end of file
diff --git a/qwen2.5-7b-instruct/chat_template.jinja b/qwen2.5-7b-instruct/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..bdf7919a96cfe43d50914a007b9c0877bd0ec27e
--- /dev/null
+++ b/qwen2.5-7b-instruct/chat_template.jinja
@@ -0,0 +1,54 @@
+{%- if tools %}
+ {{- '<|im_start|>system\n' }}
+ {%- if messages[0]['role'] == 'system' %}
+ {{- messages[0]['content'] }}
+ {%- else %}
+ {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+ {%- endif %}
+ {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within XML tags:\n" }}
+ {%- for tool in tools %}
+ {{- "\n" }}
+ {{- tool | tojson }}
+ {%- endfor %}
+ {{- "\n\n\nFor each function call, return a json object with function name and arguments within XML tags:\n\n{\"name\": , \"arguments\": }\n<|im_end|>\n" }}
+{%- else %}
+ {%- if messages[0]['role'] == 'system' %}
+ {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+ {%- else %}
+ {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+ {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+ {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+ {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+ {%- elif message.role == "assistant" %}
+ {{- '<|im_start|>' + message.role }}
+ {%- if message.content %}
+ {{- '\n' + message.content }}
+ {%- endif %}
+ {%- for tool_call in message.tool_calls %}
+ {%- if tool_call.function is defined %}
+ {%- set tool_call = tool_call.function %}
+ {%- endif %}
+ {{- '\n\n{"name": "' }}
+ {{- tool_call.name }}
+ {{- '", "arguments": ' }}
+ {{- tool_call.arguments | tojson }}
+ {{- '}\n' }}
+ {%- endfor %}
+ {{- '<|im_end|>\n' }}
+ {%- elif message.role == "tool" %}
+ {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+ {{- '<|im_start|>user' }}
+ {%- endif %}
+ {{- '\n\n' }}
+ {{- message.content }}
+ {{- '\n' }}
+ {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+ {{- '<|im_end|>\n' }}
+ {%- endif %}
+ {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+ {{- '<|im_start|>assistant\n' }}
+{%- endif %}
diff --git a/qwen2.5-7b-instruct/checkpoint-142/chat_template.jinja b/qwen2.5-7b-instruct/checkpoint-142/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..bdf7919a96cfe43d50914a007b9c0877bd0ec27e
--- /dev/null
+++ b/qwen2.5-7b-instruct/checkpoint-142/chat_template.jinja
@@ -0,0 +1,54 @@
+{%- if tools %}
+ {{- '<|im_start|>system\n' }}
+ {%- if messages[0]['role'] == 'system' %}
+ {{- messages[0]['content'] }}
+ {%- else %}
+ {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+ {%- endif %}
+ {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within XML tags:\n" }}
+ {%- for tool in tools %}
+ {{- "\n" }}
+ {{- tool | tojson }}
+ {%- endfor %}
+ {{- "\n\n\nFor each function call, return a json object with function name and arguments within XML tags:\n\n{\"name\": , \"arguments\": }\n<|im_end|>\n" }}
+{%- else %}
+ {%- if messages[0]['role'] == 'system' %}
+ {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+ {%- else %}
+ {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+ {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+ {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+ {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+ {%- elif message.role == "assistant" %}
+ {{- '<|im_start|>' + message.role }}
+ {%- if message.content %}
+ {{- '\n' + message.content }}
+ {%- endif %}
+ {%- for tool_call in message.tool_calls %}
+ {%- if tool_call.function is defined %}
+ {%- set tool_call = tool_call.function %}
+ {%- endif %}
+ {{- '\n\n{"name": "' }}
+ {{- tool_call.name }}
+ {{- '", "arguments": ' }}
+ {{- tool_call.arguments | tojson }}
+ {{- '}\n' }}
+ {%- endfor %}
+ {{- '<|im_end|>\n' }}
+ {%- elif message.role == "tool" %}
+ {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+ {{- '<|im_start|>user' }}
+ {%- endif %}
+ {{- '\n\n' }}
+ {{- message.content }}
+ {{- '\n' }}
+ {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+ {{- '<|im_end|>\n' }}
+ {%- endif %}
+ {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+ {{- '<|im_start|>assistant\n' }}
+{%- endif %}
diff --git a/qwen2.5-7b-instruct/checkpoint-142/config.json b/qwen2.5-7b-instruct/checkpoint-142/config.json
new file mode 100644
index 0000000000000000000000000000000000000000..4814292863d1e5aa2f4fda1db18f4082cd61aa93
--- /dev/null
+++ b/qwen2.5-7b-instruct/checkpoint-142/config.json
@@ -0,0 +1,69 @@
+{
+ "architectures": [
+ "Qwen2ForCausalLM"
+ ],
+ "attention_dropout": 0.0,
+ "bos_token_id": null,
+ "dtype": "float32",
+ "eos_token_id": 151645,
+ "hidden_act": "silu",
+ "hidden_size": 2048,
+ "initializer_range": 0.02,
+ "intermediate_size": 11008,
+ "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",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention"
+ ],
+ "max_position_embeddings": 32768,
+ "max_window_layers": 70,
+ "model_type": "qwen2",
+ "num_attention_heads": 16,
+ "num_hidden_layers": 36,
+ "num_key_value_heads": 2,
+ "pad_token_id": 151643,
+ "rms_norm_eps": 1e-06,
+ "rope_parameters": {
+ "rope_theta": 1000000.0,
+ "rope_type": "default"
+ },
+ "sliding_window": null,
+ "tie_word_embeddings": true,
+ "transformers_version": "5.2.0",
+ "use_cache": false,
+ "use_sliding_window": false,
+ "vocab_size": 151936
+}
diff --git a/qwen2.5-7b-instruct/checkpoint-142/generation_config.json b/qwen2.5-7b-instruct/checkpoint-142/generation_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..47205fbf3386a215e7c2b9d62df6576f49b77e7b
--- /dev/null
+++ b/qwen2.5-7b-instruct/checkpoint-142/generation_config.json
@@ -0,0 +1,13 @@
+{
+ "do_sample": true,
+ "eos_token_id": [
+ 151645,
+ 151643
+ ],
+ "pad_token_id": 151643,
+ "repetition_penalty": 1.05,
+ "temperature": 0.7,
+ "top_k": 20,
+ "top_p": 0.8,
+ "transformers_version": "5.2.0"
+}
diff --git a/qwen2.5-7b-instruct/checkpoint-142/model.safetensors b/qwen2.5-7b-instruct/checkpoint-142/model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..4258d4aead6ee04138d405e8db8d01afc1c850ea
--- /dev/null
+++ b/qwen2.5-7b-instruct/checkpoint-142/model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:59f31e42298cae8a9b58a6536fb393388ff79c3c4f1c61e1c776f9275582c85e
+size 12343804296
diff --git a/qwen2.5-7b-instruct/checkpoint-142/optimizer.pt b/qwen2.5-7b-instruct/checkpoint-142/optimizer.pt
new file mode 100644
index 0000000000000000000000000000000000000000..a9404028d52612f7404bedd9024c61146b4ae2a4
--- /dev/null
+++ b/qwen2.5-7b-instruct/checkpoint-142/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5265939003c98970f6f968bcd1b0840f5579907da0669669d9b4fe6e10e0ee53
+size 24687895753
diff --git a/qwen2.5-7b-instruct/checkpoint-142/rng_state_0.pth b/qwen2.5-7b-instruct/checkpoint-142/rng_state_0.pth
new file mode 100644
index 0000000000000000000000000000000000000000..b2e0e7337f62c934e90c141589ca185ff78fae36
--- /dev/null
+++ b/qwen2.5-7b-instruct/checkpoint-142/rng_state_0.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2fe6b873f1da64bb980afe982fc1c6b6bc461ad8c33073d856804323f64b26d9
+size 15429
diff --git a/qwen2.5-7b-instruct/checkpoint-142/rng_state_1.pth b/qwen2.5-7b-instruct/checkpoint-142/rng_state_1.pth
new file mode 100644
index 0000000000000000000000000000000000000000..204fc909b58b8ae0fab9c8b0f41ea5cc8f3277f8
--- /dev/null
+++ b/qwen2.5-7b-instruct/checkpoint-142/rng_state_1.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:13db4dd10b22b5464f5ef0b539152b46390eae6a827b355108cb54b1af2eb830
+size 15429
diff --git a/qwen2.5-7b-instruct/checkpoint-142/rng_state_2.pth b/qwen2.5-7b-instruct/checkpoint-142/rng_state_2.pth
new file mode 100644
index 0000000000000000000000000000000000000000..0c3095386194e2f629b38c7957abe0b9ab74aa8b
--- /dev/null
+++ b/qwen2.5-7b-instruct/checkpoint-142/rng_state_2.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:06e293f13fb6e873e8081636b6b07115410b397c55f1ed08b8772b059c8eb74e
+size 15429
diff --git a/qwen2.5-7b-instruct/checkpoint-142/rng_state_3.pth b/qwen2.5-7b-instruct/checkpoint-142/rng_state_3.pth
new file mode 100644
index 0000000000000000000000000000000000000000..fdf7f18d5d86f0d363c9ca44b9ad169d469e47f7
--- /dev/null
+++ b/qwen2.5-7b-instruct/checkpoint-142/rng_state_3.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c595c077664f94b7f91088561951be20e39483e01865774a83bc46f3971a3152
+size 15429
diff --git a/qwen2.5-7b-instruct/checkpoint-142/scheduler.pt b/qwen2.5-7b-instruct/checkpoint-142/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..2967e7243507a115f9424b1b386bf3657f047c75
--- /dev/null
+++ b/qwen2.5-7b-instruct/checkpoint-142/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:922deb71175797a96c4bddc5e1cef1c097df38c0dfa52334e97c89e0f573f2bb
+size 1465
diff --git a/qwen2.5-7b-instruct/checkpoint-142/tokenizer.json b/qwen2.5-7b-instruct/checkpoint-142/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..34510ff0037cd50428af467a17ead5a96140a32c
--- /dev/null
+++ b/qwen2.5-7b-instruct/checkpoint-142/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
+size 11421892
diff --git a/qwen2.5-7b-instruct/checkpoint-142/tokenizer_config.json b/qwen2.5-7b-instruct/checkpoint-142/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..145e2c7a68b109ae0253dda29fc2e66cb01b4b85
--- /dev/null
+++ b/qwen2.5-7b-instruct/checkpoint-142/tokenizer_config.json
@@ -0,0 +1,30 @@
+{
+ "add_prefix_space": false,
+ "backend": "tokenizers",
+ "bos_token": null,
+ "clean_up_tokenization_spaces": false,
+ "eos_token": "<|im_end|>",
+ "errors": "replace",
+ "extra_special_tokens": [
+ "<|im_start|>",
+ "<|im_end|>",
+ "<|object_ref_start|>",
+ "<|object_ref_end|>",
+ "<|box_start|>",
+ "<|box_end|>",
+ "<|quad_start|>",
+ "<|quad_end|>",
+ "<|vision_start|>",
+ "<|vision_end|>",
+ "<|vision_pad|>",
+ "<|image_pad|>",
+ "<|video_pad|>"
+ ],
+ "is_local": false,
+ "model_max_length": 131072,
+ "pad_token": "<|endoftext|>",
+ "padding_side": "right",
+ "split_special_tokens": false,
+ "tokenizer_class": "Qwen2Tokenizer",
+ "unk_token": null
+}
diff --git a/qwen2.5-7b-instruct/checkpoint-142/trainer_state.json b/qwen2.5-7b-instruct/checkpoint-142/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..f9e66c71b853427f09f8249c7bf69b34922dfefb
--- /dev/null
+++ b/qwen2.5-7b-instruct/checkpoint-142/trainer_state.json
@@ -0,0 +1,1044 @@
+{
+ "best_global_step": null,
+ "best_metric": null,
+ "best_model_checkpoint": null,
+ "epoch": 2.0,
+ "eval_steps": 50,
+ "global_step": 142,
+ "is_hyper_param_search": false,
+ "is_local_process_zero": true,
+ "is_world_process_zero": true,
+ "log_history": [
+ {
+ "epoch": 0.014222222222222223,
+ "grad_norm": 0.506786584854126,
+ "learning_rate": 0.0,
+ "loss": 1.1951476335525513,
+ "step": 1
+ },
+ {
+ "epoch": 0.028444444444444446,
+ "grad_norm": 0.5440636277198792,
+ "learning_rate": 4.5454545454545457e-07,
+ "loss": 1.2398298978805542,
+ "step": 2
+ },
+ {
+ "epoch": 0.042666666666666665,
+ "grad_norm": 0.4429919421672821,
+ "learning_rate": 9.090909090909091e-07,
+ "loss": 0.9777733087539673,
+ "step": 3
+ },
+ {
+ "epoch": 0.05688888888888889,
+ "grad_norm": 0.34922653436660767,
+ "learning_rate": 1.3636363636363636e-06,
+ "loss": 0.8064044713973999,
+ "step": 4
+ },
+ {
+ "epoch": 0.07111111111111111,
+ "grad_norm": 0.16975589096546173,
+ "learning_rate": 1.8181818181818183e-06,
+ "loss": 0.5294505953788757,
+ "step": 5
+ },
+ {
+ "epoch": 0.08533333333333333,
+ "grad_norm": 0.16188450157642365,
+ "learning_rate": 2.2727272727272728e-06,
+ "loss": 0.636877179145813,
+ "step": 6
+ },
+ {
+ "epoch": 0.09955555555555555,
+ "grad_norm": 0.11110807955265045,
+ "learning_rate": 2.7272727272727272e-06,
+ "loss": 0.6087832450866699,
+ "step": 7
+ },
+ {
+ "epoch": 0.11377777777777778,
+ "grad_norm": 0.0641367956995964,
+ "learning_rate": 3.181818181818182e-06,
+ "loss": 0.5258303880691528,
+ "step": 8
+ },
+ {
+ "epoch": 0.128,
+ "grad_norm": 0.052072592079639435,
+ "learning_rate": 3.6363636363636366e-06,
+ "loss": 0.5444644689559937,
+ "step": 9
+ },
+ {
+ "epoch": 0.14222222222222222,
+ "grad_norm": 0.0392303504049778,
+ "learning_rate": 4.0909090909090915e-06,
+ "loss": 0.5337975025177002,
+ "step": 10
+ },
+ {
+ "epoch": 0.15644444444444444,
+ "grad_norm": 0.028075236827135086,
+ "learning_rate": 4.5454545454545455e-06,
+ "loss": 0.45043277740478516,
+ "step": 11
+ },
+ {
+ "epoch": 0.17066666666666666,
+ "grad_norm": 0.02663014456629753,
+ "learning_rate": 5e-06,
+ "loss": 0.49355342984199524,
+ "step": 12
+ },
+ {
+ "epoch": 0.18488888888888888,
+ "grad_norm": 0.028838366270065308,
+ "learning_rate": 5.4545454545454545e-06,
+ "loss": 0.487549751996994,
+ "step": 13
+ },
+ {
+ "epoch": 0.1991111111111111,
+ "grad_norm": 0.03057004138827324,
+ "learning_rate": 5.90909090909091e-06,
+ "loss": 0.500217616558075,
+ "step": 14
+ },
+ {
+ "epoch": 0.21333333333333335,
+ "grad_norm": 0.028651155531406403,
+ "learning_rate": 6.363636363636364e-06,
+ "loss": 0.49628522992134094,
+ "step": 15
+ },
+ {
+ "epoch": 0.22755555555555557,
+ "grad_norm": 0.028336627408862114,
+ "learning_rate": 6.818181818181818e-06,
+ "loss": 0.48170676827430725,
+ "step": 16
+ },
+ {
+ "epoch": 0.24177777777777779,
+ "grad_norm": 0.019824286922812462,
+ "learning_rate": 7.272727272727273e-06,
+ "loss": 0.4609394073486328,
+ "step": 17
+ },
+ {
+ "epoch": 0.256,
+ "grad_norm": 0.02125639095902443,
+ "learning_rate": 7.727272727272727e-06,
+ "loss": 0.4759177565574646,
+ "step": 18
+ },
+ {
+ "epoch": 0.2702222222222222,
+ "grad_norm": 0.025788351893424988,
+ "learning_rate": 8.181818181818183e-06,
+ "loss": 0.4962843060493469,
+ "step": 19
+ },
+ {
+ "epoch": 0.28444444444444444,
+ "grad_norm": 0.033310867846012115,
+ "learning_rate": 8.636363636363637e-06,
+ "loss": 0.5028114318847656,
+ "step": 20
+ },
+ {
+ "epoch": 0.2986666666666667,
+ "grad_norm": 0.020624130964279175,
+ "learning_rate": 9.090909090909091e-06,
+ "loss": 0.44575220346450806,
+ "step": 21
+ },
+ {
+ "epoch": 0.3128888888888889,
+ "grad_norm": 0.020986203104257584,
+ "learning_rate": 9.545454545454547e-06,
+ "loss": 0.4755765199661255,
+ "step": 22
+ },
+ {
+ "epoch": 0.32711111111111113,
+ "grad_norm": 2526660.75,
+ "learning_rate": 1e-05,
+ "loss": 0.4706365466117859,
+ "step": 23
+ },
+ {
+ "epoch": 0.3413333333333333,
+ "grad_norm": 0.021011805161833763,
+ "learning_rate": 9.999323662872998e-06,
+ "loss": 0.4945886731147766,
+ "step": 24
+ },
+ {
+ "epoch": 0.35555555555555557,
+ "grad_norm": 0.020573964342474937,
+ "learning_rate": 9.99729483446475e-06,
+ "loss": 0.4544343054294586,
+ "step": 25
+ },
+ {
+ "epoch": 0.36977777777777776,
+ "grad_norm": 0.021116767078638077,
+ "learning_rate": 9.993914063644053e-06,
+ "loss": 0.4556806683540344,
+ "step": 26
+ },
+ {
+ "epoch": 0.384,
+ "grad_norm": 0.019978955388069153,
+ "learning_rate": 9.989182265027232e-06,
+ "loss": 0.4431076645851135,
+ "step": 27
+ },
+ {
+ "epoch": 0.3982222222222222,
+ "grad_norm": 0.019631512463092804,
+ "learning_rate": 9.98310071873072e-06,
+ "loss": 0.4541151523590088,
+ "step": 28
+ },
+ {
+ "epoch": 0.41244444444444445,
+ "grad_norm": 0.01977457106113434,
+ "learning_rate": 9.975671070024741e-06,
+ "loss": 0.4352927803993225,
+ "step": 29
+ },
+ {
+ "epoch": 0.4266666666666667,
+ "grad_norm": 0.022281618788838387,
+ "learning_rate": 9.966895328888195e-06,
+ "loss": 0.5559179186820984,
+ "step": 30
+ },
+ {
+ "epoch": 0.4408888888888889,
+ "grad_norm": 0.018352897837758064,
+ "learning_rate": 9.956775869464901e-06,
+ "loss": 0.42942947149276733,
+ "step": 31
+ },
+ {
+ "epoch": 0.45511111111111113,
+ "grad_norm": 0.019121287390589714,
+ "learning_rate": 9.945315429421307e-06,
+ "loss": 0.45916861295700073,
+ "step": 32
+ },
+ {
+ "epoch": 0.4693333333333333,
+ "grad_norm": 0.01712602935731411,
+ "learning_rate": 9.932517109205849e-06,
+ "loss": 0.4312085807323456,
+ "step": 33
+ },
+ {
+ "epoch": 0.48355555555555557,
+ "grad_norm": 0.01892935112118721,
+ "learning_rate": 9.918384371210178e-06,
+ "loss": 0.5169366598129272,
+ "step": 34
+ },
+ {
+ "epoch": 0.49777777777777776,
+ "grad_norm": 0.02034984715282917,
+ "learning_rate": 9.902921038832456e-06,
+ "loss": 0.49438145756721497,
+ "step": 35
+ },
+ {
+ "epoch": 0.512,
+ "grad_norm": 0.017887447029352188,
+ "learning_rate": 9.886131295443003e-06,
+ "loss": 0.4683716893196106,
+ "step": 36
+ },
+ {
+ "epoch": 0.5262222222222223,
+ "grad_norm": 0.01887686364352703,
+ "learning_rate": 9.868019683252543e-06,
+ "loss": 0.46688079833984375,
+ "step": 37
+ },
+ {
+ "epoch": 0.5404444444444444,
+ "grad_norm": 0.019363289698958397,
+ "learning_rate": 9.848591102083375e-06,
+ "loss": 0.4786224067211151,
+ "step": 38
+ },
+ {
+ "epoch": 0.5546666666666666,
+ "grad_norm": 0.0201108418405056,
+ "learning_rate": 9.82785080804381e-06,
+ "loss": 0.47358086705207825,
+ "step": 39
+ },
+ {
+ "epoch": 0.5688888888888889,
+ "grad_norm": 0.019631700590252876,
+ "learning_rate": 9.805804412106197e-06,
+ "loss": 0.585416316986084,
+ "step": 40
+ },
+ {
+ "epoch": 0.5831111111111111,
+ "grad_norm": 0.01689487136900425,
+ "learning_rate": 9.782457878588977e-06,
+ "loss": 0.47205525636672974,
+ "step": 41
+ },
+ {
+ "epoch": 0.5973333333333334,
+ "grad_norm": 0.01628926210105419,
+ "learning_rate": 9.75781752354311e-06,
+ "loss": 0.4348371624946594,
+ "step": 42
+ },
+ {
+ "epoch": 0.6115555555555555,
+ "grad_norm": 0.01758437603712082,
+ "learning_rate": 9.731890013043367e-06,
+ "loss": 0.4891083538532257,
+ "step": 43
+ },
+ {
+ "epoch": 0.6257777777777778,
+ "grad_norm": 0.017766334116458893,
+ "learning_rate": 9.704682361384941e-06,
+ "loss": 0.46330946683883667,
+ "step": 44
+ },
+ {
+ "epoch": 0.64,
+ "grad_norm": 0.017524050548672676,
+ "learning_rate": 9.676201929185809e-06,
+ "loss": 0.4083091914653778,
+ "step": 45
+ },
+ {
+ "epoch": 0.6542222222222223,
+ "grad_norm": 0.017291702330112457,
+ "learning_rate": 9.646456421395447e-06,
+ "loss": 0.4653710126876831,
+ "step": 46
+ },
+ {
+ "epoch": 0.6684444444444444,
+ "grad_norm": 0.017993608489632607,
+ "learning_rate": 9.615453885210368e-06,
+ "loss": 0.4626290202140808,
+ "step": 47
+ },
+ {
+ "epoch": 0.6826666666666666,
+ "grad_norm": 0.017247265204787254,
+ "learning_rate": 9.583202707897075e-06,
+ "loss": 0.47964128851890564,
+ "step": 48
+ },
+ {
+ "epoch": 0.6968888888888889,
+ "grad_norm": 0.017244094982743263,
+ "learning_rate": 9.549711614523007e-06,
+ "loss": 0.46718651056289673,
+ "step": 49
+ },
+ {
+ "epoch": 0.7111111111111111,
+ "grad_norm": 0.01574530452489853,
+ "learning_rate": 9.514989665596114e-06,
+ "loss": 0.4456287622451782,
+ "step": 50
+ },
+ {
+ "epoch": 0.7111111111111111,
+ "eval_loss": 0.4940151274204254,
+ "eval_runtime": 6.2328,
+ "eval_samples_per_second": 32.088,
+ "eval_steps_per_second": 8.022,
+ "step": 50
+ },
+ {
+ "epoch": 0.7253333333333334,
+ "grad_norm": 0.01783670485019684,
+ "learning_rate": 9.479046254613673e-06,
+ "loss": 0.45046865940093994,
+ "step": 51
+ },
+ {
+ "epoch": 0.7395555555555555,
+ "grad_norm": 0.015281656756997108,
+ "learning_rate": 9.441891105521005e-06,
+ "loss": 0.43093934655189514,
+ "step": 52
+ },
+ {
+ "epoch": 0.7537777777777778,
+ "grad_norm": 0.017853204160928726,
+ "learning_rate": 9.40353427008083e-06,
+ "loss": 0.4706975519657135,
+ "step": 53
+ },
+ {
+ "epoch": 0.768,
+ "grad_norm": 0.01542010996490717,
+ "learning_rate": 9.3639861251539e-06,
+ "loss": 0.4405261278152466,
+ "step": 54
+ },
+ {
+ "epoch": 0.7822222222222223,
+ "grad_norm": 0.017832862213253975,
+ "learning_rate": 9.323257369891702e-06,
+ "loss": 0.47680750489234924,
+ "step": 55
+ },
+ {
+ "epoch": 0.7964444444444444,
+ "grad_norm": 0.015628881752490997,
+ "learning_rate": 9.281359022841966e-06,
+ "loss": 0.42392900586128235,
+ "step": 56
+ },
+ {
+ "epoch": 0.8106666666666666,
+ "grad_norm": 0.01757017895579338,
+ "learning_rate": 9.238302418967757e-06,
+ "loss": 0.48122283816337585,
+ "step": 57
+ },
+ {
+ "epoch": 0.8248888888888889,
+ "grad_norm": 0.017186852172017097,
+ "learning_rate": 9.194099206580981e-06,
+ "loss": 0.5255264043807983,
+ "step": 58
+ },
+ {
+ "epoch": 0.8391111111111111,
+ "grad_norm": 0.017161328345537186,
+ "learning_rate": 9.14876134419111e-06,
+ "loss": 0.4682884216308594,
+ "step": 59
+ },
+ {
+ "epoch": 0.8533333333333334,
+ "grad_norm": 0.01665889099240303,
+ "learning_rate": 9.102301097269974e-06,
+ "loss": 0.5003230571746826,
+ "step": 60
+ },
+ {
+ "epoch": 0.8675555555555555,
+ "grad_norm": 0.02079184167087078,
+ "learning_rate": 9.05473103493355e-06,
+ "loss": 0.45744988322257996,
+ "step": 61
+ },
+ {
+ "epoch": 0.8817777777777778,
+ "grad_norm": 0.018035434186458588,
+ "learning_rate": 9.006064026541549e-06,
+ "loss": 0.5037047863006592,
+ "step": 62
+ },
+ {
+ "epoch": 0.896,
+ "grad_norm": 0.015980787575244904,
+ "learning_rate": 8.956313238215824e-06,
+ "loss": 0.4382043778896332,
+ "step": 63
+ },
+ {
+ "epoch": 0.9102222222222223,
+ "grad_norm": 0.016865955665707588,
+ "learning_rate": 8.905492129278478e-06,
+ "loss": 0.44455963373184204,
+ "step": 64
+ },
+ {
+ "epoch": 0.9244444444444444,
+ "grad_norm": 0.014777022413909435,
+ "learning_rate": 8.85361444861063e-06,
+ "loss": 0.4215100109577179,
+ "step": 65
+ },
+ {
+ "epoch": 0.9386666666666666,
+ "grad_norm": 0.01814866065979004,
+ "learning_rate": 8.800694230932885e-06,
+ "loss": 0.48813939094543457,
+ "step": 66
+ },
+ {
+ "epoch": 0.9528888888888889,
+ "grad_norm": 0.013353748247027397,
+ "learning_rate": 8.74674579300843e-06,
+ "loss": 3.263723850250244,
+ "step": 67
+ },
+ {
+ "epoch": 0.9671111111111111,
+ "grad_norm": 0.014444677159190178,
+ "learning_rate": 8.691783729769874e-06,
+ "loss": 2.8557608127593994,
+ "step": 68
+ },
+ {
+ "epoch": 0.9813333333333333,
+ "grad_norm": 0.01388335507363081,
+ "learning_rate": 8.635822910370793e-06,
+ "loss": 3.485170602798462,
+ "step": 69
+ },
+ {
+ "epoch": 0.9955555555555555,
+ "grad_norm": 0.01357923075556755,
+ "learning_rate": 8.578878474163115e-06,
+ "loss": 3.1958985328674316,
+ "step": 70
+ },
+ {
+ "epoch": 1.0,
+ "grad_norm": 0.02305089496076107,
+ "learning_rate": 8.520965826601394e-06,
+ "loss": 2.678990364074707,
+ "step": 71
+ },
+ {
+ "epoch": 1.0142222222222221,
+ "grad_norm": 0.014260848984122276,
+ "learning_rate": 8.462100635075097e-06,
+ "loss": 3.4466443061828613,
+ "step": 72
+ },
+ {
+ "epoch": 1.0284444444444445,
+ "grad_norm": 0.014910527504980564,
+ "learning_rate": 8.40229882467003e-06,
+ "loss": 3.4718263149261475,
+ "step": 73
+ },
+ {
+ "epoch": 1.0426666666666666,
+ "grad_norm": 0.013167285360395908,
+ "learning_rate": 8.341576573860049e-06,
+ "loss": 3.6029202938079834,
+ "step": 74
+ },
+ {
+ "epoch": 1.056888888888889,
+ "grad_norm": 0.012767510488629341,
+ "learning_rate": 8.279950310130218e-06,
+ "loss": 4.018604755401611,
+ "step": 75
+ },
+ {
+ "epoch": 1.0711111111111111,
+ "grad_norm": 0.013733700849115849,
+ "learning_rate": 8.2174367055326e-06,
+ "loss": 3.0605146884918213,
+ "step": 76
+ },
+ {
+ "epoch": 1.0853333333333333,
+ "grad_norm": 0.011987799778580666,
+ "learning_rate": 8.154052672175888e-06,
+ "loss": 3.556879997253418,
+ "step": 77
+ },
+ {
+ "epoch": 1.0995555555555556,
+ "grad_norm": 0.014038470573723316,
+ "learning_rate": 8.08981535765009e-06,
+ "loss": 3.7346572875976562,
+ "step": 78
+ },
+ {
+ "epoch": 1.1137777777777778,
+ "grad_norm": 0.013301494531333447,
+ "learning_rate": 8.024742140387506e-06,
+ "loss": 3.2394087314605713,
+ "step": 79
+ },
+ {
+ "epoch": 1.1280000000000001,
+ "grad_norm": 0.013333727605640888,
+ "learning_rate": 7.95885062496126e-06,
+ "loss": 3.625601053237915,
+ "step": 80
+ },
+ {
+ "epoch": 1.1422222222222222,
+ "grad_norm": 0.013422077521681786,
+ "learning_rate": 7.892158637322647e-06,
+ "loss": 3.3379783630371094,
+ "step": 81
+ },
+ {
+ "epoch": 1.1564444444444444,
+ "grad_norm": 0.012368597090244293,
+ "learning_rate": 7.824684219978591e-06,
+ "loss": 3.502805471420288,
+ "step": 82
+ },
+ {
+ "epoch": 1.1706666666666667,
+ "grad_norm": 0.013034222647547722,
+ "learning_rate": 7.756445627110523e-06,
+ "loss": 3.362370729446411,
+ "step": 83
+ },
+ {
+ "epoch": 1.1848888888888889,
+ "grad_norm": 0.014766803942620754,
+ "learning_rate": 7.68746131963598e-06,
+ "loss": 3.1864588260650635,
+ "step": 84
+ },
+ {
+ "epoch": 1.199111111111111,
+ "grad_norm": 0.012399553321301937,
+ "learning_rate": 7.6177499602143e-06,
+ "loss": 3.4393932819366455,
+ "step": 85
+ },
+ {
+ "epoch": 1.2133333333333334,
+ "grad_norm": 0.014413993805646896,
+ "learning_rate": 7.547330408197695e-06,
+ "loss": 3.76263689994812,
+ "step": 86
+ },
+ {
+ "epoch": 1.2275555555555555,
+ "grad_norm": 0.014122135005891323,
+ "learning_rate": 7.476221714529167e-06,
+ "loss": 3.3016631603240967,
+ "step": 87
+ },
+ {
+ "epoch": 1.2417777777777779,
+ "grad_norm": 0.015857728198170662,
+ "learning_rate": 7.404443116588548e-06,
+ "loss": 2.8077316284179688,
+ "step": 88
+ },
+ {
+ "epoch": 1.256,
+ "grad_norm": 0.013196472078561783,
+ "learning_rate": 7.332014032988123e-06,
+ "loss": 3.496107816696167,
+ "step": 89
+ },
+ {
+ "epoch": 1.2702222222222221,
+ "grad_norm": 0.01415536180138588,
+ "learning_rate": 7.2589540583192165e-06,
+ "loss": 3.4046881198883057,
+ "step": 90
+ },
+ {
+ "epoch": 1.2844444444444445,
+ "grad_norm": 0.01354147121310234,
+ "learning_rate": 7.185282957851175e-06,
+ "loss": 3.3288354873657227,
+ "step": 91
+ },
+ {
+ "epoch": 1.2986666666666666,
+ "grad_norm": 0.013861758634448051,
+ "learning_rate": 7.111020662184174e-06,
+ "loss": 3.573462963104248,
+ "step": 92
+ },
+ {
+ "epoch": 1.3128888888888888,
+ "grad_norm": 0.01368795894086361,
+ "learning_rate": 7.036187261857289e-06,
+ "loss": 3.4570071697235107,
+ "step": 93
+ },
+ {
+ "epoch": 1.3271111111111111,
+ "grad_norm": 0.014616986736655235,
+ "learning_rate": 6.960803001913315e-06,
+ "loss": 3.511754274368286,
+ "step": 94
+ },
+ {
+ "epoch": 1.3413333333333333,
+ "grad_norm": 0.01352851465344429,
+ "learning_rate": 6.884888276421766e-06,
+ "loss": 4.017252445220947,
+ "step": 95
+ },
+ {
+ "epoch": 1.3555555555555556,
+ "grad_norm": 0.013078316114842892,
+ "learning_rate": 6.8084636229615786e-06,
+ "loss": 3.5670371055603027,
+ "step": 96
+ },
+ {
+ "epoch": 1.3697777777777778,
+ "grad_norm": 0.012880469672381878,
+ "learning_rate": 6.731549717064975e-06,
+ "loss": 4.055681228637695,
+ "step": 97
+ },
+ {
+ "epoch": 1.384,
+ "grad_norm": 0.013766918331384659,
+ "learning_rate": 6.654167366624009e-06,
+ "loss": 3.303616762161255,
+ "step": 98
+ },
+ {
+ "epoch": 1.3982222222222223,
+ "grad_norm": 0.015087485313415527,
+ "learning_rate": 6.576337506261314e-06,
+ "loss": 2.8152995109558105,
+ "step": 99
+ },
+ {
+ "epoch": 1.4124444444444444,
+ "grad_norm": 0.013361976481974125,
+ "learning_rate": 6.498081191666549e-06,
+ "loss": 3.3619225025177,
+ "step": 100
+ },
+ {
+ "epoch": 1.4124444444444444,
+ "eval_loss": 3.694061279296875,
+ "eval_runtime": 5.8211,
+ "eval_samples_per_second": 34.358,
+ "eval_steps_per_second": 8.589,
+ "step": 100
+ },
+ {
+ "epoch": 1.4266666666666667,
+ "grad_norm": 0.013033254072070122,
+ "learning_rate": 6.419419593900109e-06,
+ "loss": 3.0373668670654297,
+ "step": 101
+ },
+ {
+ "epoch": 1.4408888888888889,
+ "grad_norm": 0.012533189728856087,
+ "learning_rate": 6.340373993665607e-06,
+ "loss": 3.642394542694092,
+ "step": 102
+ },
+ {
+ "epoch": 1.455111111111111,
+ "grad_norm": 0.013018937781453133,
+ "learning_rate": 6.2609657755527135e-06,
+ "loss": 3.3127541542053223,
+ "step": 103
+ },
+ {
+ "epoch": 1.4693333333333334,
+ "grad_norm": 0.014089098200201988,
+ "learning_rate": 6.1812164222518626e-06,
+ "loss": 3.500901937484741,
+ "step": 104
+ },
+ {
+ "epoch": 1.4835555555555555,
+ "grad_norm": 0.0140894316136837,
+ "learning_rate": 6.101147508742456e-06,
+ "loss": 3.4911036491394043,
+ "step": 105
+ },
+ {
+ "epoch": 1.4977777777777779,
+ "grad_norm": 0.014050478115677834,
+ "learning_rate": 6.020780696456059e-06,
+ "loss": 3.375718116760254,
+ "step": 106
+ },
+ {
+ "epoch": 1.512,
+ "grad_norm": 0.013563934713602066,
+ "learning_rate": 5.940137727416247e-06,
+ "loss": 3.340306282043457,
+ "step": 107
+ },
+ {
+ "epoch": 1.5262222222222221,
+ "grad_norm": 0.013330376707017422,
+ "learning_rate": 5.859240418356614e-06,
+ "loss": 3.5164387226104736,
+ "step": 108
+ },
+ {
+ "epoch": 1.5404444444444443,
+ "grad_norm": 0.013517254032194614,
+ "learning_rate": 5.778110654818602e-06,
+ "loss": 3.955772876739502,
+ "step": 109
+ },
+ {
+ "epoch": 1.5546666666666666,
+ "grad_norm": 0.014757441356778145,
+ "learning_rate": 5.696770385230679e-06,
+ "loss": 3.425647020339966,
+ "step": 110
+ },
+ {
+ "epoch": 1.568888888888889,
+ "grad_norm": 0.014494120143353939,
+ "learning_rate": 5.615241614970546e-06,
+ "loss": 3.219924211502075,
+ "step": 111
+ },
+ {
+ "epoch": 1.5831111111111111,
+ "grad_norm": 0.012901407666504383,
+ "learning_rate": 5.533546400411899e-06,
+ "loss": 3.3812978267669678,
+ "step": 112
+ },
+ {
+ "epoch": 1.5973333333333333,
+ "grad_norm": 0.013786875642836094,
+ "learning_rate": 5.4517068429574215e-06,
+ "loss": 3.7506139278411865,
+ "step": 113
+ },
+ {
+ "epoch": 1.6115555555555554,
+ "grad_norm": 0.012589235790073872,
+ "learning_rate": 5.3697450830595775e-06,
+ "loss": 3.259803295135498,
+ "step": 114
+ },
+ {
+ "epoch": 1.6257777777777778,
+ "grad_norm": 0.01369628868997097,
+ "learning_rate": 5.287683294230855e-06,
+ "loss": 2.8727362155914307,
+ "step": 115
+ },
+ {
+ "epoch": 1.6400000000000001,
+ "grad_norm": 0.012632690370082855,
+ "learning_rate": 5.20554367704505e-06,
+ "loss": 3.628926992416382,
+ "step": 116
+ },
+ {
+ "epoch": 1.6542222222222223,
+ "grad_norm": 0.014506797306239605,
+ "learning_rate": 5.123348453131242e-06,
+ "loss": 3.236464023590088,
+ "step": 117
+ },
+ {
+ "epoch": 1.6684444444444444,
+ "grad_norm": 0.014436502009630203,
+ "learning_rate": 5.041119859162068e-06,
+ "loss": 2.888354778289795,
+ "step": 118
+ },
+ {
+ "epoch": 1.6826666666666665,
+ "grad_norm": 0.014416320249438286,
+ "learning_rate": 4.958880140837934e-06,
+ "loss": 3.105433702468872,
+ "step": 119
+ },
+ {
+ "epoch": 1.696888888888889,
+ "grad_norm": 0.014613285660743713,
+ "learning_rate": 4.876651546868759e-06,
+ "loss": 3.2962746620178223,
+ "step": 120
+ },
+ {
+ "epoch": 1.7111111111111112,
+ "grad_norm": 0.014369661919772625,
+ "learning_rate": 4.794456322954953e-06,
+ "loss": 3.451850414276123,
+ "step": 121
+ },
+ {
+ "epoch": 1.7253333333333334,
+ "grad_norm": 0.013792592100799084,
+ "learning_rate": 4.7123167057691446e-06,
+ "loss": 3.6925525665283203,
+ "step": 122
+ },
+ {
+ "epoch": 1.7395555555555555,
+ "grad_norm": 0.013583365827798843,
+ "learning_rate": 4.630254916940424e-06,
+ "loss": 4.191713333129883,
+ "step": 123
+ },
+ {
+ "epoch": 1.7537777777777777,
+ "grad_norm": 0.014320321381092072,
+ "learning_rate": 4.548293157042581e-06,
+ "loss": 2.8300790786743164,
+ "step": 124
+ },
+ {
+ "epoch": 1.768,
+ "grad_norm": 0.013430479913949966,
+ "learning_rate": 4.466453599588103e-06,
+ "loss": 3.5787062644958496,
+ "step": 125
+ },
+ {
+ "epoch": 1.7822222222222224,
+ "grad_norm": 0.012813941575586796,
+ "learning_rate": 4.384758385029457e-06,
+ "loss": 2.687720537185669,
+ "step": 126
+ },
+ {
+ "epoch": 1.7964444444444445,
+ "grad_norm": 0.014274616725742817,
+ "learning_rate": 4.3032296147693225e-06,
+ "loss": 3.221151113510132,
+ "step": 127
+ },
+ {
+ "epoch": 1.8106666666666666,
+ "grad_norm": 0.013514580205082893,
+ "learning_rate": 4.2218893451814005e-06,
+ "loss": 3.415364980697632,
+ "step": 128
+ },
+ {
+ "epoch": 1.8248888888888888,
+ "grad_norm": 0.01293711643666029,
+ "learning_rate": 4.140759581643386e-06,
+ "loss": 3.2137668132781982,
+ "step": 129
+ },
+ {
+ "epoch": 1.8391111111111111,
+ "grad_norm": 0.013326463289558887,
+ "learning_rate": 4.059862272583755e-06,
+ "loss": 2.874941349029541,
+ "step": 130
+ },
+ {
+ "epoch": 1.8533333333333335,
+ "grad_norm": 0.013346531428396702,
+ "learning_rate": 3.979219303543942e-06,
+ "loss": 3.6437442302703857,
+ "step": 131
+ },
+ {
+ "epoch": 1.8675555555555556,
+ "grad_norm": 0.013448791578412056,
+ "learning_rate": 3.898852491257547e-06,
+ "loss": 3.262843370437622,
+ "step": 132
+ },
+ {
+ "epoch": 1.8817777777777778,
+ "grad_norm": 0.013963771052658558,
+ "learning_rate": 3.818783577748138e-06,
+ "loss": 2.763110637664795,
+ "step": 133
+ },
+ {
+ "epoch": 1.896,
+ "grad_norm": 0.01534951850771904,
+ "learning_rate": 3.7390342244472886e-06,
+ "loss": 2.978278160095215,
+ "step": 134
+ },
+ {
+ "epoch": 1.9102222222222223,
+ "grad_norm": 0.014479058794677258,
+ "learning_rate": 3.659626006334395e-06,
+ "loss": 2.9288432598114014,
+ "step": 135
+ },
+ {
+ "epoch": 1.9244444444444444,
+ "grad_norm": 0.01563713327050209,
+ "learning_rate": 3.5805804060998926e-06,
+ "loss": 3.158463954925537,
+ "step": 136
+ },
+ {
+ "epoch": 1.9386666666666668,
+ "grad_norm": 0.01603933610022068,
+ "learning_rate": 3.501918808333453e-06,
+ "loss": 2.699950933456421,
+ "step": 137
+ },
+ {
+ "epoch": 1.952888888888889,
+ "grad_norm": 0.014355695806443691,
+ "learning_rate": 3.4236624937386874e-06,
+ "loss": 3.840054512023926,
+ "step": 138
+ },
+ {
+ "epoch": 1.967111111111111,
+ "grad_norm": 0.013930619694292545,
+ "learning_rate": 3.3458326333759927e-06,
+ "loss": 2.972839832305908,
+ "step": 139
+ },
+ {
+ "epoch": 1.9813333333333332,
+ "grad_norm": 0.015130864456295967,
+ "learning_rate": 3.268450282935026e-06,
+ "loss": 2.8588995933532715,
+ "step": 140
+ },
+ {
+ "epoch": 1.9955555555555555,
+ "grad_norm": 0.013637659139931202,
+ "learning_rate": 3.1915363770384223e-06,
+ "loss": 3.6230926513671875,
+ "step": 141
+ },
+ {
+ "epoch": 2.0,
+ "grad_norm": 0.023099087178707123,
+ "learning_rate": 3.1151117235782346e-06,
+ "loss": 2.1066107749938965,
+ "step": 142
+ }
+ ],
+ "logging_steps": 1,
+ "max_steps": 213,
+ "num_input_tokens_seen": 0,
+ "num_train_epochs": 3,
+ "save_steps": 500,
+ "stateful_callbacks": {
+ "TrainerControl": {
+ "args": {
+ "should_epoch_stop": false,
+ "should_evaluate": false,
+ "should_log": false,
+ "should_save": true,
+ "should_training_stop": false
+ },
+ "attributes": {}
+ }
+ },
+ "total_flos": 6.872319887881011e+16,
+ "train_batch_size": 1,
+ "trial_name": null,
+ "trial_params": null
+}
diff --git a/qwen2.5-7b-instruct/checkpoint-142/training_args.bin b/qwen2.5-7b-instruct/checkpoint-142/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..b5415efad8d753f3c17d7314f34dccfa7e73e9e0
--- /dev/null
+++ b/qwen2.5-7b-instruct/checkpoint-142/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:62861a63c10b8e91d966486b438a4ca616ac892c434ce240cbba36083500e44e
+size 5649
diff --git a/qwen2.5-7b-instruct/checkpoint-213/chat_template.jinja b/qwen2.5-7b-instruct/checkpoint-213/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..bdf7919a96cfe43d50914a007b9c0877bd0ec27e
--- /dev/null
+++ b/qwen2.5-7b-instruct/checkpoint-213/chat_template.jinja
@@ -0,0 +1,54 @@
+{%- if tools %}
+ {{- '<|im_start|>system\n' }}
+ {%- if messages[0]['role'] == 'system' %}
+ {{- messages[0]['content'] }}
+ {%- else %}
+ {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+ {%- endif %}
+ {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within XML tags:\n" }}
+ {%- for tool in tools %}
+ {{- "\n" }}
+ {{- tool | tojson }}
+ {%- endfor %}
+ {{- "\n\n\nFor each function call, return a json object with function name and arguments within XML tags:\n\n{\"name\": , \"arguments\": }\n<|im_end|>\n" }}
+{%- else %}
+ {%- if messages[0]['role'] == 'system' %}
+ {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+ {%- else %}
+ {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+ {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+ {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+ {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+ {%- elif message.role == "assistant" %}
+ {{- '<|im_start|>' + message.role }}
+ {%- if message.content %}
+ {{- '\n' + message.content }}
+ {%- endif %}
+ {%- for tool_call in message.tool_calls %}
+ {%- if tool_call.function is defined %}
+ {%- set tool_call = tool_call.function %}
+ {%- endif %}
+ {{- '\n\n{"name": "' }}
+ {{- tool_call.name }}
+ {{- '", "arguments": ' }}
+ {{- tool_call.arguments | tojson }}
+ {{- '}\n' }}
+ {%- endfor %}
+ {{- '<|im_end|>\n' }}
+ {%- elif message.role == "tool" %}
+ {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+ {{- '<|im_start|>user' }}
+ {%- endif %}
+ {{- '\n\n' }}
+ {{- message.content }}
+ {{- '\n' }}
+ {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+ {{- '<|im_end|>\n' }}
+ {%- endif %}
+ {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+ {{- '<|im_start|>assistant\n' }}
+{%- endif %}
diff --git a/qwen2.5-7b-instruct/checkpoint-213/config.json b/qwen2.5-7b-instruct/checkpoint-213/config.json
new file mode 100644
index 0000000000000000000000000000000000000000..4814292863d1e5aa2f4fda1db18f4082cd61aa93
--- /dev/null
+++ b/qwen2.5-7b-instruct/checkpoint-213/config.json
@@ -0,0 +1,69 @@
+{
+ "architectures": [
+ "Qwen2ForCausalLM"
+ ],
+ "attention_dropout": 0.0,
+ "bos_token_id": null,
+ "dtype": "float32",
+ "eos_token_id": 151645,
+ "hidden_act": "silu",
+ "hidden_size": 2048,
+ "initializer_range": 0.02,
+ "intermediate_size": 11008,
+ "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",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention"
+ ],
+ "max_position_embeddings": 32768,
+ "max_window_layers": 70,
+ "model_type": "qwen2",
+ "num_attention_heads": 16,
+ "num_hidden_layers": 36,
+ "num_key_value_heads": 2,
+ "pad_token_id": 151643,
+ "rms_norm_eps": 1e-06,
+ "rope_parameters": {
+ "rope_theta": 1000000.0,
+ "rope_type": "default"
+ },
+ "sliding_window": null,
+ "tie_word_embeddings": true,
+ "transformers_version": "5.2.0",
+ "use_cache": false,
+ "use_sliding_window": false,
+ "vocab_size": 151936
+}
diff --git a/qwen2.5-7b-instruct/checkpoint-213/generation_config.json b/qwen2.5-7b-instruct/checkpoint-213/generation_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..47205fbf3386a215e7c2b9d62df6576f49b77e7b
--- /dev/null
+++ b/qwen2.5-7b-instruct/checkpoint-213/generation_config.json
@@ -0,0 +1,13 @@
+{
+ "do_sample": true,
+ "eos_token_id": [
+ 151645,
+ 151643
+ ],
+ "pad_token_id": 151643,
+ "repetition_penalty": 1.05,
+ "temperature": 0.7,
+ "top_k": 20,
+ "top_p": 0.8,
+ "transformers_version": "5.2.0"
+}
diff --git a/qwen2.5-7b-instruct/checkpoint-213/model.safetensors b/qwen2.5-7b-instruct/checkpoint-213/model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..61029f7ead30eb65ec9a23e4c34c1f73a7a1af78
--- /dev/null
+++ b/qwen2.5-7b-instruct/checkpoint-213/model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:98f699d765f656b4c23b83714137d84fa3f23ec986eebd693963185de38bd7b3
+size 12343804296
diff --git a/qwen2.5-7b-instruct/checkpoint-213/optimizer.pt b/qwen2.5-7b-instruct/checkpoint-213/optimizer.pt
new file mode 100644
index 0000000000000000000000000000000000000000..df0df705bd8518865b93b8627f287868a5d48bcf
--- /dev/null
+++ b/qwen2.5-7b-instruct/checkpoint-213/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d68954404a5cef07e6b2ac96fbd50ea0346578c2e1c979c2ac2024b50f67d399
+size 24687895753
diff --git a/qwen2.5-7b-instruct/checkpoint-213/rng_state_0.pth b/qwen2.5-7b-instruct/checkpoint-213/rng_state_0.pth
new file mode 100644
index 0000000000000000000000000000000000000000..b2e0e7337f62c934e90c141589ca185ff78fae36
--- /dev/null
+++ b/qwen2.5-7b-instruct/checkpoint-213/rng_state_0.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2fe6b873f1da64bb980afe982fc1c6b6bc461ad8c33073d856804323f64b26d9
+size 15429
diff --git a/qwen2.5-7b-instruct/checkpoint-213/rng_state_1.pth b/qwen2.5-7b-instruct/checkpoint-213/rng_state_1.pth
new file mode 100644
index 0000000000000000000000000000000000000000..204fc909b58b8ae0fab9c8b0f41ea5cc8f3277f8
--- /dev/null
+++ b/qwen2.5-7b-instruct/checkpoint-213/rng_state_1.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:13db4dd10b22b5464f5ef0b539152b46390eae6a827b355108cb54b1af2eb830
+size 15429
diff --git a/qwen2.5-7b-instruct/checkpoint-213/rng_state_2.pth b/qwen2.5-7b-instruct/checkpoint-213/rng_state_2.pth
new file mode 100644
index 0000000000000000000000000000000000000000..0c3095386194e2f629b38c7957abe0b9ab74aa8b
--- /dev/null
+++ b/qwen2.5-7b-instruct/checkpoint-213/rng_state_2.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:06e293f13fb6e873e8081636b6b07115410b397c55f1ed08b8772b059c8eb74e
+size 15429
diff --git a/qwen2.5-7b-instruct/checkpoint-213/rng_state_3.pth b/qwen2.5-7b-instruct/checkpoint-213/rng_state_3.pth
new file mode 100644
index 0000000000000000000000000000000000000000..fdf7f18d5d86f0d363c9ca44b9ad169d469e47f7
--- /dev/null
+++ b/qwen2.5-7b-instruct/checkpoint-213/rng_state_3.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c595c077664f94b7f91088561951be20e39483e01865774a83bc46f3971a3152
+size 15429
diff --git a/qwen2.5-7b-instruct/checkpoint-213/scheduler.pt b/qwen2.5-7b-instruct/checkpoint-213/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..39cf87775c3a71d10e7311e0cab769e112f357ba
--- /dev/null
+++ b/qwen2.5-7b-instruct/checkpoint-213/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c38a67b98c17be81d4c6d00b60029cdedd43e9aba35ca28ec28ef946c400f268
+size 1465
diff --git a/qwen2.5-7b-instruct/checkpoint-213/tokenizer.json b/qwen2.5-7b-instruct/checkpoint-213/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..34510ff0037cd50428af467a17ead5a96140a32c
--- /dev/null
+++ b/qwen2.5-7b-instruct/checkpoint-213/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
+size 11421892
diff --git a/qwen2.5-7b-instruct/checkpoint-213/tokenizer_config.json b/qwen2.5-7b-instruct/checkpoint-213/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..145e2c7a68b109ae0253dda29fc2e66cb01b4b85
--- /dev/null
+++ b/qwen2.5-7b-instruct/checkpoint-213/tokenizer_config.json
@@ -0,0 +1,30 @@
+{
+ "add_prefix_space": false,
+ "backend": "tokenizers",
+ "bos_token": null,
+ "clean_up_tokenization_spaces": false,
+ "eos_token": "<|im_end|>",
+ "errors": "replace",
+ "extra_special_tokens": [
+ "<|im_start|>",
+ "<|im_end|>",
+ "<|object_ref_start|>",
+ "<|object_ref_end|>",
+ "<|box_start|>",
+ "<|box_end|>",
+ "<|quad_start|>",
+ "<|quad_end|>",
+ "<|vision_start|>",
+ "<|vision_end|>",
+ "<|vision_pad|>",
+ "<|image_pad|>",
+ "<|video_pad|>"
+ ],
+ "is_local": false,
+ "model_max_length": 131072,
+ "pad_token": "<|endoftext|>",
+ "padding_side": "right",
+ "split_special_tokens": false,
+ "tokenizer_class": "Qwen2Tokenizer",
+ "unk_token": null
+}
diff --git a/qwen2.5-7b-instruct/checkpoint-213/trainer_state.json b/qwen2.5-7b-instruct/checkpoint-213/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..e1da6da263198be5b9335c0789415a66dc04a2c5
--- /dev/null
+++ b/qwen2.5-7b-instruct/checkpoint-213/trainer_state.json
@@ -0,0 +1,1557 @@
+{
+ "best_global_step": null,
+ "best_metric": null,
+ "best_model_checkpoint": null,
+ "epoch": 3.0,
+ "eval_steps": 50,
+ "global_step": 213,
+ "is_hyper_param_search": false,
+ "is_local_process_zero": true,
+ "is_world_process_zero": true,
+ "log_history": [
+ {
+ "epoch": 0.014222222222222223,
+ "grad_norm": 0.506786584854126,
+ "learning_rate": 0.0,
+ "loss": 1.1951476335525513,
+ "step": 1
+ },
+ {
+ "epoch": 0.028444444444444446,
+ "grad_norm": 0.5440636277198792,
+ "learning_rate": 4.5454545454545457e-07,
+ "loss": 1.2398298978805542,
+ "step": 2
+ },
+ {
+ "epoch": 0.042666666666666665,
+ "grad_norm": 0.4429919421672821,
+ "learning_rate": 9.090909090909091e-07,
+ "loss": 0.9777733087539673,
+ "step": 3
+ },
+ {
+ "epoch": 0.05688888888888889,
+ "grad_norm": 0.34922653436660767,
+ "learning_rate": 1.3636363636363636e-06,
+ "loss": 0.8064044713973999,
+ "step": 4
+ },
+ {
+ "epoch": 0.07111111111111111,
+ "grad_norm": 0.16975589096546173,
+ "learning_rate": 1.8181818181818183e-06,
+ "loss": 0.5294505953788757,
+ "step": 5
+ },
+ {
+ "epoch": 0.08533333333333333,
+ "grad_norm": 0.16188450157642365,
+ "learning_rate": 2.2727272727272728e-06,
+ "loss": 0.636877179145813,
+ "step": 6
+ },
+ {
+ "epoch": 0.09955555555555555,
+ "grad_norm": 0.11110807955265045,
+ "learning_rate": 2.7272727272727272e-06,
+ "loss": 0.6087832450866699,
+ "step": 7
+ },
+ {
+ "epoch": 0.11377777777777778,
+ "grad_norm": 0.0641367956995964,
+ "learning_rate": 3.181818181818182e-06,
+ "loss": 0.5258303880691528,
+ "step": 8
+ },
+ {
+ "epoch": 0.128,
+ "grad_norm": 0.052072592079639435,
+ "learning_rate": 3.6363636363636366e-06,
+ "loss": 0.5444644689559937,
+ "step": 9
+ },
+ {
+ "epoch": 0.14222222222222222,
+ "grad_norm": 0.0392303504049778,
+ "learning_rate": 4.0909090909090915e-06,
+ "loss": 0.5337975025177002,
+ "step": 10
+ },
+ {
+ "epoch": 0.15644444444444444,
+ "grad_norm": 0.028075236827135086,
+ "learning_rate": 4.5454545454545455e-06,
+ "loss": 0.45043277740478516,
+ "step": 11
+ },
+ {
+ "epoch": 0.17066666666666666,
+ "grad_norm": 0.02663014456629753,
+ "learning_rate": 5e-06,
+ "loss": 0.49355342984199524,
+ "step": 12
+ },
+ {
+ "epoch": 0.18488888888888888,
+ "grad_norm": 0.028838366270065308,
+ "learning_rate": 5.4545454545454545e-06,
+ "loss": 0.487549751996994,
+ "step": 13
+ },
+ {
+ "epoch": 0.1991111111111111,
+ "grad_norm": 0.03057004138827324,
+ "learning_rate": 5.90909090909091e-06,
+ "loss": 0.500217616558075,
+ "step": 14
+ },
+ {
+ "epoch": 0.21333333333333335,
+ "grad_norm": 0.028651155531406403,
+ "learning_rate": 6.363636363636364e-06,
+ "loss": 0.49628522992134094,
+ "step": 15
+ },
+ {
+ "epoch": 0.22755555555555557,
+ "grad_norm": 0.028336627408862114,
+ "learning_rate": 6.818181818181818e-06,
+ "loss": 0.48170676827430725,
+ "step": 16
+ },
+ {
+ "epoch": 0.24177777777777779,
+ "grad_norm": 0.019824286922812462,
+ "learning_rate": 7.272727272727273e-06,
+ "loss": 0.4609394073486328,
+ "step": 17
+ },
+ {
+ "epoch": 0.256,
+ "grad_norm": 0.02125639095902443,
+ "learning_rate": 7.727272727272727e-06,
+ "loss": 0.4759177565574646,
+ "step": 18
+ },
+ {
+ "epoch": 0.2702222222222222,
+ "grad_norm": 0.025788351893424988,
+ "learning_rate": 8.181818181818183e-06,
+ "loss": 0.4962843060493469,
+ "step": 19
+ },
+ {
+ "epoch": 0.28444444444444444,
+ "grad_norm": 0.033310867846012115,
+ "learning_rate": 8.636363636363637e-06,
+ "loss": 0.5028114318847656,
+ "step": 20
+ },
+ {
+ "epoch": 0.2986666666666667,
+ "grad_norm": 0.020624130964279175,
+ "learning_rate": 9.090909090909091e-06,
+ "loss": 0.44575220346450806,
+ "step": 21
+ },
+ {
+ "epoch": 0.3128888888888889,
+ "grad_norm": 0.020986203104257584,
+ "learning_rate": 9.545454545454547e-06,
+ "loss": 0.4755765199661255,
+ "step": 22
+ },
+ {
+ "epoch": 0.32711111111111113,
+ "grad_norm": 2526660.75,
+ "learning_rate": 1e-05,
+ "loss": 0.4706365466117859,
+ "step": 23
+ },
+ {
+ "epoch": 0.3413333333333333,
+ "grad_norm": 0.021011805161833763,
+ "learning_rate": 9.999323662872998e-06,
+ "loss": 0.4945886731147766,
+ "step": 24
+ },
+ {
+ "epoch": 0.35555555555555557,
+ "grad_norm": 0.020573964342474937,
+ "learning_rate": 9.99729483446475e-06,
+ "loss": 0.4544343054294586,
+ "step": 25
+ },
+ {
+ "epoch": 0.36977777777777776,
+ "grad_norm": 0.021116767078638077,
+ "learning_rate": 9.993914063644053e-06,
+ "loss": 0.4556806683540344,
+ "step": 26
+ },
+ {
+ "epoch": 0.384,
+ "grad_norm": 0.019978955388069153,
+ "learning_rate": 9.989182265027232e-06,
+ "loss": 0.4431076645851135,
+ "step": 27
+ },
+ {
+ "epoch": 0.3982222222222222,
+ "grad_norm": 0.019631512463092804,
+ "learning_rate": 9.98310071873072e-06,
+ "loss": 0.4541151523590088,
+ "step": 28
+ },
+ {
+ "epoch": 0.41244444444444445,
+ "grad_norm": 0.01977457106113434,
+ "learning_rate": 9.975671070024741e-06,
+ "loss": 0.4352927803993225,
+ "step": 29
+ },
+ {
+ "epoch": 0.4266666666666667,
+ "grad_norm": 0.022281618788838387,
+ "learning_rate": 9.966895328888195e-06,
+ "loss": 0.5559179186820984,
+ "step": 30
+ },
+ {
+ "epoch": 0.4408888888888889,
+ "grad_norm": 0.018352897837758064,
+ "learning_rate": 9.956775869464901e-06,
+ "loss": 0.42942947149276733,
+ "step": 31
+ },
+ {
+ "epoch": 0.45511111111111113,
+ "grad_norm": 0.019121287390589714,
+ "learning_rate": 9.945315429421307e-06,
+ "loss": 0.45916861295700073,
+ "step": 32
+ },
+ {
+ "epoch": 0.4693333333333333,
+ "grad_norm": 0.01712602935731411,
+ "learning_rate": 9.932517109205849e-06,
+ "loss": 0.4312085807323456,
+ "step": 33
+ },
+ {
+ "epoch": 0.48355555555555557,
+ "grad_norm": 0.01892935112118721,
+ "learning_rate": 9.918384371210178e-06,
+ "loss": 0.5169366598129272,
+ "step": 34
+ },
+ {
+ "epoch": 0.49777777777777776,
+ "grad_norm": 0.02034984715282917,
+ "learning_rate": 9.902921038832456e-06,
+ "loss": 0.49438145756721497,
+ "step": 35
+ },
+ {
+ "epoch": 0.512,
+ "grad_norm": 0.017887447029352188,
+ "learning_rate": 9.886131295443003e-06,
+ "loss": 0.4683716893196106,
+ "step": 36
+ },
+ {
+ "epoch": 0.5262222222222223,
+ "grad_norm": 0.01887686364352703,
+ "learning_rate": 9.868019683252543e-06,
+ "loss": 0.46688079833984375,
+ "step": 37
+ },
+ {
+ "epoch": 0.5404444444444444,
+ "grad_norm": 0.019363289698958397,
+ "learning_rate": 9.848591102083375e-06,
+ "loss": 0.4786224067211151,
+ "step": 38
+ },
+ {
+ "epoch": 0.5546666666666666,
+ "grad_norm": 0.0201108418405056,
+ "learning_rate": 9.82785080804381e-06,
+ "loss": 0.47358086705207825,
+ "step": 39
+ },
+ {
+ "epoch": 0.5688888888888889,
+ "grad_norm": 0.019631700590252876,
+ "learning_rate": 9.805804412106197e-06,
+ "loss": 0.585416316986084,
+ "step": 40
+ },
+ {
+ "epoch": 0.5831111111111111,
+ "grad_norm": 0.01689487136900425,
+ "learning_rate": 9.782457878588977e-06,
+ "loss": 0.47205525636672974,
+ "step": 41
+ },
+ {
+ "epoch": 0.5973333333333334,
+ "grad_norm": 0.01628926210105419,
+ "learning_rate": 9.75781752354311e-06,
+ "loss": 0.4348371624946594,
+ "step": 42
+ },
+ {
+ "epoch": 0.6115555555555555,
+ "grad_norm": 0.01758437603712082,
+ "learning_rate": 9.731890013043367e-06,
+ "loss": 0.4891083538532257,
+ "step": 43
+ },
+ {
+ "epoch": 0.6257777777777778,
+ "grad_norm": 0.017766334116458893,
+ "learning_rate": 9.704682361384941e-06,
+ "loss": 0.46330946683883667,
+ "step": 44
+ },
+ {
+ "epoch": 0.64,
+ "grad_norm": 0.017524050548672676,
+ "learning_rate": 9.676201929185809e-06,
+ "loss": 0.4083091914653778,
+ "step": 45
+ },
+ {
+ "epoch": 0.6542222222222223,
+ "grad_norm": 0.017291702330112457,
+ "learning_rate": 9.646456421395447e-06,
+ "loss": 0.4653710126876831,
+ "step": 46
+ },
+ {
+ "epoch": 0.6684444444444444,
+ "grad_norm": 0.017993608489632607,
+ "learning_rate": 9.615453885210368e-06,
+ "loss": 0.4626290202140808,
+ "step": 47
+ },
+ {
+ "epoch": 0.6826666666666666,
+ "grad_norm": 0.017247265204787254,
+ "learning_rate": 9.583202707897075e-06,
+ "loss": 0.47964128851890564,
+ "step": 48
+ },
+ {
+ "epoch": 0.6968888888888889,
+ "grad_norm": 0.017244094982743263,
+ "learning_rate": 9.549711614523007e-06,
+ "loss": 0.46718651056289673,
+ "step": 49
+ },
+ {
+ "epoch": 0.7111111111111111,
+ "grad_norm": 0.01574530452489853,
+ "learning_rate": 9.514989665596114e-06,
+ "loss": 0.4456287622451782,
+ "step": 50
+ },
+ {
+ "epoch": 0.7111111111111111,
+ "eval_loss": 0.4940151274204254,
+ "eval_runtime": 6.2328,
+ "eval_samples_per_second": 32.088,
+ "eval_steps_per_second": 8.022,
+ "step": 50
+ },
+ {
+ "epoch": 0.7253333333333334,
+ "grad_norm": 0.01783670485019684,
+ "learning_rate": 9.479046254613673e-06,
+ "loss": 0.45046865940093994,
+ "step": 51
+ },
+ {
+ "epoch": 0.7395555555555555,
+ "grad_norm": 0.015281656756997108,
+ "learning_rate": 9.441891105521005e-06,
+ "loss": 0.43093934655189514,
+ "step": 52
+ },
+ {
+ "epoch": 0.7537777777777778,
+ "grad_norm": 0.017853204160928726,
+ "learning_rate": 9.40353427008083e-06,
+ "loss": 0.4706975519657135,
+ "step": 53
+ },
+ {
+ "epoch": 0.768,
+ "grad_norm": 0.01542010996490717,
+ "learning_rate": 9.3639861251539e-06,
+ "loss": 0.4405261278152466,
+ "step": 54
+ },
+ {
+ "epoch": 0.7822222222222223,
+ "grad_norm": 0.017832862213253975,
+ "learning_rate": 9.323257369891702e-06,
+ "loss": 0.47680750489234924,
+ "step": 55
+ },
+ {
+ "epoch": 0.7964444444444444,
+ "grad_norm": 0.015628881752490997,
+ "learning_rate": 9.281359022841966e-06,
+ "loss": 0.42392900586128235,
+ "step": 56
+ },
+ {
+ "epoch": 0.8106666666666666,
+ "grad_norm": 0.01757017895579338,
+ "learning_rate": 9.238302418967757e-06,
+ "loss": 0.48122283816337585,
+ "step": 57
+ },
+ {
+ "epoch": 0.8248888888888889,
+ "grad_norm": 0.017186852172017097,
+ "learning_rate": 9.194099206580981e-06,
+ "loss": 0.5255264043807983,
+ "step": 58
+ },
+ {
+ "epoch": 0.8391111111111111,
+ "grad_norm": 0.017161328345537186,
+ "learning_rate": 9.14876134419111e-06,
+ "loss": 0.4682884216308594,
+ "step": 59
+ },
+ {
+ "epoch": 0.8533333333333334,
+ "grad_norm": 0.01665889099240303,
+ "learning_rate": 9.102301097269974e-06,
+ "loss": 0.5003230571746826,
+ "step": 60
+ },
+ {
+ "epoch": 0.8675555555555555,
+ "grad_norm": 0.02079184167087078,
+ "learning_rate": 9.05473103493355e-06,
+ "loss": 0.45744988322257996,
+ "step": 61
+ },
+ {
+ "epoch": 0.8817777777777778,
+ "grad_norm": 0.018035434186458588,
+ "learning_rate": 9.006064026541549e-06,
+ "loss": 0.5037047863006592,
+ "step": 62
+ },
+ {
+ "epoch": 0.896,
+ "grad_norm": 0.015980787575244904,
+ "learning_rate": 8.956313238215824e-06,
+ "loss": 0.4382043778896332,
+ "step": 63
+ },
+ {
+ "epoch": 0.9102222222222223,
+ "grad_norm": 0.016865955665707588,
+ "learning_rate": 8.905492129278478e-06,
+ "loss": 0.44455963373184204,
+ "step": 64
+ },
+ {
+ "epoch": 0.9244444444444444,
+ "grad_norm": 0.014777022413909435,
+ "learning_rate": 8.85361444861063e-06,
+ "loss": 0.4215100109577179,
+ "step": 65
+ },
+ {
+ "epoch": 0.9386666666666666,
+ "grad_norm": 0.01814866065979004,
+ "learning_rate": 8.800694230932885e-06,
+ "loss": 0.48813939094543457,
+ "step": 66
+ },
+ {
+ "epoch": 0.9528888888888889,
+ "grad_norm": 0.013353748247027397,
+ "learning_rate": 8.74674579300843e-06,
+ "loss": 3.263723850250244,
+ "step": 67
+ },
+ {
+ "epoch": 0.9671111111111111,
+ "grad_norm": 0.014444677159190178,
+ "learning_rate": 8.691783729769874e-06,
+ "loss": 2.8557608127593994,
+ "step": 68
+ },
+ {
+ "epoch": 0.9813333333333333,
+ "grad_norm": 0.01388335507363081,
+ "learning_rate": 8.635822910370793e-06,
+ "loss": 3.485170602798462,
+ "step": 69
+ },
+ {
+ "epoch": 0.9955555555555555,
+ "grad_norm": 0.01357923075556755,
+ "learning_rate": 8.578878474163115e-06,
+ "loss": 3.1958985328674316,
+ "step": 70
+ },
+ {
+ "epoch": 1.0,
+ "grad_norm": 0.02305089496076107,
+ "learning_rate": 8.520965826601394e-06,
+ "loss": 2.678990364074707,
+ "step": 71
+ },
+ {
+ "epoch": 1.0142222222222221,
+ "grad_norm": 0.014260848984122276,
+ "learning_rate": 8.462100635075097e-06,
+ "loss": 3.4466443061828613,
+ "step": 72
+ },
+ {
+ "epoch": 1.0284444444444445,
+ "grad_norm": 0.014910527504980564,
+ "learning_rate": 8.40229882467003e-06,
+ "loss": 3.4718263149261475,
+ "step": 73
+ },
+ {
+ "epoch": 1.0426666666666666,
+ "grad_norm": 0.013167285360395908,
+ "learning_rate": 8.341576573860049e-06,
+ "loss": 3.6029202938079834,
+ "step": 74
+ },
+ {
+ "epoch": 1.056888888888889,
+ "grad_norm": 0.012767510488629341,
+ "learning_rate": 8.279950310130218e-06,
+ "loss": 4.018604755401611,
+ "step": 75
+ },
+ {
+ "epoch": 1.0711111111111111,
+ "grad_norm": 0.013733700849115849,
+ "learning_rate": 8.2174367055326e-06,
+ "loss": 3.0605146884918213,
+ "step": 76
+ },
+ {
+ "epoch": 1.0853333333333333,
+ "grad_norm": 0.011987799778580666,
+ "learning_rate": 8.154052672175888e-06,
+ "loss": 3.556879997253418,
+ "step": 77
+ },
+ {
+ "epoch": 1.0995555555555556,
+ "grad_norm": 0.014038470573723316,
+ "learning_rate": 8.08981535765009e-06,
+ "loss": 3.7346572875976562,
+ "step": 78
+ },
+ {
+ "epoch": 1.1137777777777778,
+ "grad_norm": 0.013301494531333447,
+ "learning_rate": 8.024742140387506e-06,
+ "loss": 3.2394087314605713,
+ "step": 79
+ },
+ {
+ "epoch": 1.1280000000000001,
+ "grad_norm": 0.013333727605640888,
+ "learning_rate": 7.95885062496126e-06,
+ "loss": 3.625601053237915,
+ "step": 80
+ },
+ {
+ "epoch": 1.1422222222222222,
+ "grad_norm": 0.013422077521681786,
+ "learning_rate": 7.892158637322647e-06,
+ "loss": 3.3379783630371094,
+ "step": 81
+ },
+ {
+ "epoch": 1.1564444444444444,
+ "grad_norm": 0.012368597090244293,
+ "learning_rate": 7.824684219978591e-06,
+ "loss": 3.502805471420288,
+ "step": 82
+ },
+ {
+ "epoch": 1.1706666666666667,
+ "grad_norm": 0.013034222647547722,
+ "learning_rate": 7.756445627110523e-06,
+ "loss": 3.362370729446411,
+ "step": 83
+ },
+ {
+ "epoch": 1.1848888888888889,
+ "grad_norm": 0.014766803942620754,
+ "learning_rate": 7.68746131963598e-06,
+ "loss": 3.1864588260650635,
+ "step": 84
+ },
+ {
+ "epoch": 1.199111111111111,
+ "grad_norm": 0.012399553321301937,
+ "learning_rate": 7.6177499602143e-06,
+ "loss": 3.4393932819366455,
+ "step": 85
+ },
+ {
+ "epoch": 1.2133333333333334,
+ "grad_norm": 0.014413993805646896,
+ "learning_rate": 7.547330408197695e-06,
+ "loss": 3.76263689994812,
+ "step": 86
+ },
+ {
+ "epoch": 1.2275555555555555,
+ "grad_norm": 0.014122135005891323,
+ "learning_rate": 7.476221714529167e-06,
+ "loss": 3.3016631603240967,
+ "step": 87
+ },
+ {
+ "epoch": 1.2417777777777779,
+ "grad_norm": 0.015857728198170662,
+ "learning_rate": 7.404443116588548e-06,
+ "loss": 2.8077316284179688,
+ "step": 88
+ },
+ {
+ "epoch": 1.256,
+ "grad_norm": 0.013196472078561783,
+ "learning_rate": 7.332014032988123e-06,
+ "loss": 3.496107816696167,
+ "step": 89
+ },
+ {
+ "epoch": 1.2702222222222221,
+ "grad_norm": 0.01415536180138588,
+ "learning_rate": 7.2589540583192165e-06,
+ "loss": 3.4046881198883057,
+ "step": 90
+ },
+ {
+ "epoch": 1.2844444444444445,
+ "grad_norm": 0.01354147121310234,
+ "learning_rate": 7.185282957851175e-06,
+ "loss": 3.3288354873657227,
+ "step": 91
+ },
+ {
+ "epoch": 1.2986666666666666,
+ "grad_norm": 0.013861758634448051,
+ "learning_rate": 7.111020662184174e-06,
+ "loss": 3.573462963104248,
+ "step": 92
+ },
+ {
+ "epoch": 1.3128888888888888,
+ "grad_norm": 0.01368795894086361,
+ "learning_rate": 7.036187261857289e-06,
+ "loss": 3.4570071697235107,
+ "step": 93
+ },
+ {
+ "epoch": 1.3271111111111111,
+ "grad_norm": 0.014616986736655235,
+ "learning_rate": 6.960803001913315e-06,
+ "loss": 3.511754274368286,
+ "step": 94
+ },
+ {
+ "epoch": 1.3413333333333333,
+ "grad_norm": 0.01352851465344429,
+ "learning_rate": 6.884888276421766e-06,
+ "loss": 4.017252445220947,
+ "step": 95
+ },
+ {
+ "epoch": 1.3555555555555556,
+ "grad_norm": 0.013078316114842892,
+ "learning_rate": 6.8084636229615786e-06,
+ "loss": 3.5670371055603027,
+ "step": 96
+ },
+ {
+ "epoch": 1.3697777777777778,
+ "grad_norm": 0.012880469672381878,
+ "learning_rate": 6.731549717064975e-06,
+ "loss": 4.055681228637695,
+ "step": 97
+ },
+ {
+ "epoch": 1.384,
+ "grad_norm": 0.013766918331384659,
+ "learning_rate": 6.654167366624009e-06,
+ "loss": 3.303616762161255,
+ "step": 98
+ },
+ {
+ "epoch": 1.3982222222222223,
+ "grad_norm": 0.015087485313415527,
+ "learning_rate": 6.576337506261314e-06,
+ "loss": 2.8152995109558105,
+ "step": 99
+ },
+ {
+ "epoch": 1.4124444444444444,
+ "grad_norm": 0.013361976481974125,
+ "learning_rate": 6.498081191666549e-06,
+ "loss": 3.3619225025177,
+ "step": 100
+ },
+ {
+ "epoch": 1.4124444444444444,
+ "eval_loss": 3.694061279296875,
+ "eval_runtime": 5.8211,
+ "eval_samples_per_second": 34.358,
+ "eval_steps_per_second": 8.589,
+ "step": 100
+ },
+ {
+ "epoch": 1.4266666666666667,
+ "grad_norm": 0.013033254072070122,
+ "learning_rate": 6.419419593900109e-06,
+ "loss": 3.0373668670654297,
+ "step": 101
+ },
+ {
+ "epoch": 1.4408888888888889,
+ "grad_norm": 0.012533189728856087,
+ "learning_rate": 6.340373993665607e-06,
+ "loss": 3.642394542694092,
+ "step": 102
+ },
+ {
+ "epoch": 1.455111111111111,
+ "grad_norm": 0.013018937781453133,
+ "learning_rate": 6.2609657755527135e-06,
+ "loss": 3.3127541542053223,
+ "step": 103
+ },
+ {
+ "epoch": 1.4693333333333334,
+ "grad_norm": 0.014089098200201988,
+ "learning_rate": 6.1812164222518626e-06,
+ "loss": 3.500901937484741,
+ "step": 104
+ },
+ {
+ "epoch": 1.4835555555555555,
+ "grad_norm": 0.0140894316136837,
+ "learning_rate": 6.101147508742456e-06,
+ "loss": 3.4911036491394043,
+ "step": 105
+ },
+ {
+ "epoch": 1.4977777777777779,
+ "grad_norm": 0.014050478115677834,
+ "learning_rate": 6.020780696456059e-06,
+ "loss": 3.375718116760254,
+ "step": 106
+ },
+ {
+ "epoch": 1.512,
+ "grad_norm": 0.013563934713602066,
+ "learning_rate": 5.940137727416247e-06,
+ "loss": 3.340306282043457,
+ "step": 107
+ },
+ {
+ "epoch": 1.5262222222222221,
+ "grad_norm": 0.013330376707017422,
+ "learning_rate": 5.859240418356614e-06,
+ "loss": 3.5164387226104736,
+ "step": 108
+ },
+ {
+ "epoch": 1.5404444444444443,
+ "grad_norm": 0.013517254032194614,
+ "learning_rate": 5.778110654818602e-06,
+ "loss": 3.955772876739502,
+ "step": 109
+ },
+ {
+ "epoch": 1.5546666666666666,
+ "grad_norm": 0.014757441356778145,
+ "learning_rate": 5.696770385230679e-06,
+ "loss": 3.425647020339966,
+ "step": 110
+ },
+ {
+ "epoch": 1.568888888888889,
+ "grad_norm": 0.014494120143353939,
+ "learning_rate": 5.615241614970546e-06,
+ "loss": 3.219924211502075,
+ "step": 111
+ },
+ {
+ "epoch": 1.5831111111111111,
+ "grad_norm": 0.012901407666504383,
+ "learning_rate": 5.533546400411899e-06,
+ "loss": 3.3812978267669678,
+ "step": 112
+ },
+ {
+ "epoch": 1.5973333333333333,
+ "grad_norm": 0.013786875642836094,
+ "learning_rate": 5.4517068429574215e-06,
+ "loss": 3.7506139278411865,
+ "step": 113
+ },
+ {
+ "epoch": 1.6115555555555554,
+ "grad_norm": 0.012589235790073872,
+ "learning_rate": 5.3697450830595775e-06,
+ "loss": 3.259803295135498,
+ "step": 114
+ },
+ {
+ "epoch": 1.6257777777777778,
+ "grad_norm": 0.01369628868997097,
+ "learning_rate": 5.287683294230855e-06,
+ "loss": 2.8727362155914307,
+ "step": 115
+ },
+ {
+ "epoch": 1.6400000000000001,
+ "grad_norm": 0.012632690370082855,
+ "learning_rate": 5.20554367704505e-06,
+ "loss": 3.628926992416382,
+ "step": 116
+ },
+ {
+ "epoch": 1.6542222222222223,
+ "grad_norm": 0.014506797306239605,
+ "learning_rate": 5.123348453131242e-06,
+ "loss": 3.236464023590088,
+ "step": 117
+ },
+ {
+ "epoch": 1.6684444444444444,
+ "grad_norm": 0.014436502009630203,
+ "learning_rate": 5.041119859162068e-06,
+ "loss": 2.888354778289795,
+ "step": 118
+ },
+ {
+ "epoch": 1.6826666666666665,
+ "grad_norm": 0.014416320249438286,
+ "learning_rate": 4.958880140837934e-06,
+ "loss": 3.105433702468872,
+ "step": 119
+ },
+ {
+ "epoch": 1.696888888888889,
+ "grad_norm": 0.014613285660743713,
+ "learning_rate": 4.876651546868759e-06,
+ "loss": 3.2962746620178223,
+ "step": 120
+ },
+ {
+ "epoch": 1.7111111111111112,
+ "grad_norm": 0.014369661919772625,
+ "learning_rate": 4.794456322954953e-06,
+ "loss": 3.451850414276123,
+ "step": 121
+ },
+ {
+ "epoch": 1.7253333333333334,
+ "grad_norm": 0.013792592100799084,
+ "learning_rate": 4.7123167057691446e-06,
+ "loss": 3.6925525665283203,
+ "step": 122
+ },
+ {
+ "epoch": 1.7395555555555555,
+ "grad_norm": 0.013583365827798843,
+ "learning_rate": 4.630254916940424e-06,
+ "loss": 4.191713333129883,
+ "step": 123
+ },
+ {
+ "epoch": 1.7537777777777777,
+ "grad_norm": 0.014320321381092072,
+ "learning_rate": 4.548293157042581e-06,
+ "loss": 2.8300790786743164,
+ "step": 124
+ },
+ {
+ "epoch": 1.768,
+ "grad_norm": 0.013430479913949966,
+ "learning_rate": 4.466453599588103e-06,
+ "loss": 3.5787062644958496,
+ "step": 125
+ },
+ {
+ "epoch": 1.7822222222222224,
+ "grad_norm": 0.012813941575586796,
+ "learning_rate": 4.384758385029457e-06,
+ "loss": 2.687720537185669,
+ "step": 126
+ },
+ {
+ "epoch": 1.7964444444444445,
+ "grad_norm": 0.014274616725742817,
+ "learning_rate": 4.3032296147693225e-06,
+ "loss": 3.221151113510132,
+ "step": 127
+ },
+ {
+ "epoch": 1.8106666666666666,
+ "grad_norm": 0.013514580205082893,
+ "learning_rate": 4.2218893451814005e-06,
+ "loss": 3.415364980697632,
+ "step": 128
+ },
+ {
+ "epoch": 1.8248888888888888,
+ "grad_norm": 0.01293711643666029,
+ "learning_rate": 4.140759581643386e-06,
+ "loss": 3.2137668132781982,
+ "step": 129
+ },
+ {
+ "epoch": 1.8391111111111111,
+ "grad_norm": 0.013326463289558887,
+ "learning_rate": 4.059862272583755e-06,
+ "loss": 2.874941349029541,
+ "step": 130
+ },
+ {
+ "epoch": 1.8533333333333335,
+ "grad_norm": 0.013346531428396702,
+ "learning_rate": 3.979219303543942e-06,
+ "loss": 3.6437442302703857,
+ "step": 131
+ },
+ {
+ "epoch": 1.8675555555555556,
+ "grad_norm": 0.013448791578412056,
+ "learning_rate": 3.898852491257547e-06,
+ "loss": 3.262843370437622,
+ "step": 132
+ },
+ {
+ "epoch": 1.8817777777777778,
+ "grad_norm": 0.013963771052658558,
+ "learning_rate": 3.818783577748138e-06,
+ "loss": 2.763110637664795,
+ "step": 133
+ },
+ {
+ "epoch": 1.896,
+ "grad_norm": 0.01534951850771904,
+ "learning_rate": 3.7390342244472886e-06,
+ "loss": 2.978278160095215,
+ "step": 134
+ },
+ {
+ "epoch": 1.9102222222222223,
+ "grad_norm": 0.014479058794677258,
+ "learning_rate": 3.659626006334395e-06,
+ "loss": 2.9288432598114014,
+ "step": 135
+ },
+ {
+ "epoch": 1.9244444444444444,
+ "grad_norm": 0.01563713327050209,
+ "learning_rate": 3.5805804060998926e-06,
+ "loss": 3.158463954925537,
+ "step": 136
+ },
+ {
+ "epoch": 1.9386666666666668,
+ "grad_norm": 0.01603933610022068,
+ "learning_rate": 3.501918808333453e-06,
+ "loss": 2.699950933456421,
+ "step": 137
+ },
+ {
+ "epoch": 1.952888888888889,
+ "grad_norm": 0.014355695806443691,
+ "learning_rate": 3.4236624937386874e-06,
+ "loss": 3.840054512023926,
+ "step": 138
+ },
+ {
+ "epoch": 1.967111111111111,
+ "grad_norm": 0.013930619694292545,
+ "learning_rate": 3.3458326333759927e-06,
+ "loss": 2.972839832305908,
+ "step": 139
+ },
+ {
+ "epoch": 1.9813333333333332,
+ "grad_norm": 0.015130864456295967,
+ "learning_rate": 3.268450282935026e-06,
+ "loss": 2.8588995933532715,
+ "step": 140
+ },
+ {
+ "epoch": 1.9955555555555555,
+ "grad_norm": 0.013637659139931202,
+ "learning_rate": 3.1915363770384223e-06,
+ "loss": 3.6230926513671875,
+ "step": 141
+ },
+ {
+ "epoch": 2.0,
+ "grad_norm": 0.023099087178707123,
+ "learning_rate": 3.1151117235782346e-06,
+ "loss": 2.1066107749938965,
+ "step": 142
+ },
+ {
+ "epoch": 2.014222222222222,
+ "grad_norm": 0.014238209463655949,
+ "learning_rate": 3.0391969980866874e-06,
+ "loss": 3.128911256790161,
+ "step": 143
+ },
+ {
+ "epoch": 2.0284444444444443,
+ "grad_norm": 0.012874635867774487,
+ "learning_rate": 2.963812738142713e-06,
+ "loss": 3.2379772663116455,
+ "step": 144
+ },
+ {
+ "epoch": 2.042666666666667,
+ "grad_norm": 0.013502821326255798,
+ "learning_rate": 2.8889793378158284e-06,
+ "loss": 3.2959132194519043,
+ "step": 145
+ },
+ {
+ "epoch": 2.056888888888889,
+ "grad_norm": 0.013256440870463848,
+ "learning_rate": 2.814717042148827e-06,
+ "loss": 3.3406944274902344,
+ "step": 146
+ },
+ {
+ "epoch": 2.071111111111111,
+ "grad_norm": 0.013458102941513062,
+ "learning_rate": 2.7410459416807856e-06,
+ "loss": 3.3720455169677734,
+ "step": 147
+ },
+ {
+ "epoch": 2.0853333333333333,
+ "grad_norm": 0.01476226095110178,
+ "learning_rate": 2.6679859670118785e-06,
+ "loss": 3.153254508972168,
+ "step": 148
+ },
+ {
+ "epoch": 2.0995555555555554,
+ "grad_norm": 0.013766537420451641,
+ "learning_rate": 2.5955568834114523e-06,
+ "loss": 2.9255833625793457,
+ "step": 149
+ },
+ {
+ "epoch": 2.113777777777778,
+ "grad_norm": 0.012672817334532738,
+ "learning_rate": 2.523778285470835e-06,
+ "loss": 3.385373115539551,
+ "step": 150
+ },
+ {
+ "epoch": 2.113777777777778,
+ "eval_loss": 3.694100856781006,
+ "eval_runtime": 6.2397,
+ "eval_samples_per_second": 32.053,
+ "eval_steps_per_second": 8.013,
+ "step": 150
+ },
+ {
+ "epoch": 2.128,
+ "grad_norm": 0.013193093240261078,
+ "learning_rate": 2.452669591802307e-06,
+ "loss": 2.953336238861084,
+ "step": 151
+ },
+ {
+ "epoch": 2.1422222222222222,
+ "grad_norm": 0.012676793150603771,
+ "learning_rate": 2.3822500397857016e-06,
+ "loss": 3.4797470569610596,
+ "step": 152
+ },
+ {
+ "epoch": 2.1564444444444444,
+ "grad_norm": 0.013178958557546139,
+ "learning_rate": 2.3125386803640188e-06,
+ "loss": 3.766719341278076,
+ "step": 153
+ },
+ {
+ "epoch": 2.1706666666666665,
+ "grad_norm": 0.013044048100709915,
+ "learning_rate": 2.243554372889479e-06,
+ "loss": 3.105588436126709,
+ "step": 154
+ },
+ {
+ "epoch": 2.1848888888888887,
+ "grad_norm": 0.01274812500923872,
+ "learning_rate": 2.1753157800214107e-06,
+ "loss": 3.7258148193359375,
+ "step": 155
+ },
+ {
+ "epoch": 2.1991111111111112,
+ "grad_norm": 0.014226537197828293,
+ "learning_rate": 2.1078413626773547e-06,
+ "loss": 3.2563636302948,
+ "step": 156
+ },
+ {
+ "epoch": 2.2133333333333334,
+ "grad_norm": 0.013875821605324745,
+ "learning_rate": 2.0411493750387423e-06,
+ "loss": 3.6285147666931152,
+ "step": 157
+ },
+ {
+ "epoch": 2.2275555555555555,
+ "grad_norm": 0.013328024186193943,
+ "learning_rate": 1.9752578596124955e-06,
+ "loss": 3.4809606075286865,
+ "step": 158
+ },
+ {
+ "epoch": 2.2417777777777776,
+ "grad_norm": 0.014153145253658295,
+ "learning_rate": 1.9101846423499113e-06,
+ "loss": 3.331500768661499,
+ "step": 159
+ },
+ {
+ "epoch": 2.2560000000000002,
+ "grad_norm": 0.013629083521664143,
+ "learning_rate": 1.8459473278241125e-06,
+ "loss": 3.103165626525879,
+ "step": 160
+ },
+ {
+ "epoch": 2.2702222222222224,
+ "grad_norm": 0.013135570101439953,
+ "learning_rate": 1.7825632944674016e-06,
+ "loss": 2.949009656906128,
+ "step": 161
+ },
+ {
+ "epoch": 2.2844444444444445,
+ "grad_norm": 0.015113947913050652,
+ "learning_rate": 1.7200496898697832e-06,
+ "loss": 3.8803815841674805,
+ "step": 162
+ },
+ {
+ "epoch": 2.2986666666666666,
+ "grad_norm": 0.01285882480442524,
+ "learning_rate": 1.6584234261399535e-06,
+ "loss": 3.3277387619018555,
+ "step": 163
+ },
+ {
+ "epoch": 2.3128888888888888,
+ "grad_norm": 0.013268567621707916,
+ "learning_rate": 1.5977011753299726e-06,
+ "loss": 3.04113507270813,
+ "step": 164
+ },
+ {
+ "epoch": 2.327111111111111,
+ "grad_norm": 0.015572321601212025,
+ "learning_rate": 1.5378993649249053e-06,
+ "loss": 3.1088926792144775,
+ "step": 165
+ },
+ {
+ "epoch": 2.3413333333333335,
+ "grad_norm": 0.013958935625851154,
+ "learning_rate": 1.4790341733986085e-06,
+ "loss": 2.9704766273498535,
+ "step": 166
+ },
+ {
+ "epoch": 2.3555555555555556,
+ "grad_norm": 0.01262224093079567,
+ "learning_rate": 1.4211215258368866e-06,
+ "loss": 2.666156053543091,
+ "step": 167
+ },
+ {
+ "epoch": 2.3697777777777778,
+ "grad_norm": 0.01445838063955307,
+ "learning_rate": 1.3641770896292083e-06,
+ "loss": 3.5968339443206787,
+ "step": 168
+ },
+ {
+ "epoch": 2.384,
+ "grad_norm": 0.013657037168741226,
+ "learning_rate": 1.3082162702301276e-06,
+ "loss": 3.576735496520996,
+ "step": 169
+ },
+ {
+ "epoch": 2.398222222222222,
+ "grad_norm": 0.014326483011245728,
+ "learning_rate": 1.2532542069915722e-06,
+ "loss": 3.0840680599212646,
+ "step": 170
+ },
+ {
+ "epoch": 2.4124444444444446,
+ "grad_norm": 0.014063039794564247,
+ "learning_rate": 1.1993057690671174e-06,
+ "loss": 3.7334625720977783,
+ "step": 171
+ },
+ {
+ "epoch": 2.4266666666666667,
+ "grad_norm": 0.012410550378262997,
+ "learning_rate": 1.1463855513893695e-06,
+ "loss": 4.4254841804504395,
+ "step": 172
+ },
+ {
+ "epoch": 2.440888888888889,
+ "grad_norm": 0.012715302407741547,
+ "learning_rate": 1.0945078707215224e-06,
+ "loss": 3.775759696960449,
+ "step": 173
+ },
+ {
+ "epoch": 2.455111111111111,
+ "grad_norm": 0.014594604261219501,
+ "learning_rate": 1.0436867617841768e-06,
+ "loss": 2.9598851203918457,
+ "step": 174
+ },
+ {
+ "epoch": 2.469333333333333,
+ "grad_norm": 0.013803915120661259,
+ "learning_rate": 9.939359734584552e-07,
+ "loss": 2.9934566020965576,
+ "step": 175
+ },
+ {
+ "epoch": 2.4835555555555557,
+ "grad_norm": 0.013807227835059166,
+ "learning_rate": 9.452689650664515e-07,
+ "loss": 3.3066673278808594,
+ "step": 176
+ },
+ {
+ "epoch": 2.497777777777778,
+ "grad_norm": 0.013583092018961906,
+ "learning_rate": 8.976989027300265e-07,
+ "loss": 2.8772518634796143,
+ "step": 177
+ },
+ {
+ "epoch": 2.512,
+ "grad_norm": NaN,
+ "learning_rate": 8.512386558088919e-07,
+ "loss": 0.2714032232761383,
+ "step": 178
+ },
+ {
+ "epoch": 2.526222222222222,
+ "grad_norm": NaN,
+ "learning_rate": 8.059007934190194e-07,
+ "loss": 0.0,
+ "step": 179
+ },
+ {
+ "epoch": 2.5404444444444443,
+ "grad_norm": NaN,
+ "learning_rate": 7.61697581032243e-07,
+ "loss": 0.0,
+ "step": 180
+ },
+ {
+ "epoch": 2.554666666666667,
+ "grad_norm": NaN,
+ "learning_rate": 7.186409771580355e-07,
+ "loss": 0.0,
+ "step": 181
+ },
+ {
+ "epoch": 2.568888888888889,
+ "grad_norm": NaN,
+ "learning_rate": 6.76742630108298e-07,
+ "loss": 0.0,
+ "step": 182
+ },
+ {
+ "epoch": 2.583111111111111,
+ "grad_norm": NaN,
+ "learning_rate": 6.360138748461015e-07,
+ "loss": 0.0,
+ "step": 183
+ },
+ {
+ "epoch": 2.5973333333333333,
+ "grad_norm": NaN,
+ "learning_rate": 5.964657299191712e-07,
+ "loss": 0.0,
+ "step": 184
+ },
+ {
+ "epoch": 2.6115555555555554,
+ "grad_norm": NaN,
+ "learning_rate": 5.581088944789953e-07,
+ "loss": 0.0,
+ "step": 185
+ },
+ {
+ "epoch": 2.6257777777777775,
+ "grad_norm": NaN,
+ "learning_rate": 5.20953745386329e-07,
+ "loss": 0.0,
+ "step": 186
+ },
+ {
+ "epoch": 2.64,
+ "grad_norm": NaN,
+ "learning_rate": 4.850103344038853e-07,
+ "loss": 0.0,
+ "step": 187
+ },
+ {
+ "epoch": 2.6542222222222223,
+ "grad_norm": NaN,
+ "learning_rate": 4.502883854769935e-07,
+ "loss": 0.0,
+ "step": 188
+ },
+ {
+ "epoch": 2.6684444444444444,
+ "grad_norm": NaN,
+ "learning_rate": 4.167972921029262e-07,
+ "loss": 0.0,
+ "step": 189
+ },
+ {
+ "epoch": 2.6826666666666665,
+ "grad_norm": NaN,
+ "learning_rate": 3.8454611478963235e-07,
+ "loss": 0.0,
+ "step": 190
+ },
+ {
+ "epoch": 2.696888888888889,
+ "grad_norm": NaN,
+ "learning_rate": 3.535435786045538e-07,
+ "loss": 0.0,
+ "step": 191
+ },
+ {
+ "epoch": 2.7111111111111112,
+ "grad_norm": NaN,
+ "learning_rate": 3.237980708141919e-07,
+ "loss": 0.0,
+ "step": 192
+ },
+ {
+ "epoch": 2.7253333333333334,
+ "grad_norm": NaN,
+ "learning_rate": 2.9531763861505967e-07,
+ "loss": 0.0,
+ "step": 193
+ },
+ {
+ "epoch": 2.7395555555555555,
+ "grad_norm": NaN,
+ "learning_rate": 2.681099869566328e-07,
+ "loss": 0.0,
+ "step": 194
+ },
+ {
+ "epoch": 2.7537777777777777,
+ "grad_norm": NaN,
+ "learning_rate": 2.4218247645689306e-07,
+ "loss": 0.0,
+ "step": 195
+ },
+ {
+ "epoch": 2.768,
+ "grad_norm": NaN,
+ "learning_rate": 2.175421214110235e-07,
+ "loss": 0.0,
+ "step": 196
+ },
+ {
+ "epoch": 2.7822222222222224,
+ "grad_norm": NaN,
+ "learning_rate": 1.941955878938029e-07,
+ "loss": 0.0,
+ "step": 197
+ },
+ {
+ "epoch": 2.7964444444444445,
+ "grad_norm": NaN,
+ "learning_rate": 1.721491919561913e-07,
+ "loss": 0.0,
+ "step": 198
+ },
+ {
+ "epoch": 2.8106666666666666,
+ "grad_norm": NaN,
+ "learning_rate": 1.514088979166256e-07,
+ "loss": 0.0,
+ "step": 199
+ },
+ {
+ "epoch": 2.824888888888889,
+ "grad_norm": NaN,
+ "learning_rate": 1.3198031674745814e-07,
+ "loss": 0.0,
+ "step": 200
+ },
+ {
+ "epoch": 2.824888888888889,
+ "eval_loss": NaN,
+ "eval_runtime": 5.5777,
+ "eval_samples_per_second": 35.857,
+ "eval_steps_per_second": 8.964,
+ "step": 200
+ },
+ {
+ "epoch": 2.8391111111111114,
+ "grad_norm": NaN,
+ "learning_rate": 1.138687045569975e-07,
+ "loss": 0.0,
+ "step": 201
+ },
+ {
+ "epoch": 2.8533333333333335,
+ "grad_norm": NaN,
+ "learning_rate": 9.70789611675449e-08,
+ "loss": 0.0,
+ "step": 202
+ },
+ {
+ "epoch": 2.8675555555555556,
+ "grad_norm": NaN,
+ "learning_rate": 8.161562878982399e-08,
+ "loss": 0.0,
+ "step": 203
+ },
+ {
+ "epoch": 2.8817777777777778,
+ "grad_norm": NaN,
+ "learning_rate": 6.74828907941516e-08,
+ "loss": 0.0,
+ "step": 204
+ },
+ {
+ "epoch": 2.896,
+ "grad_norm": NaN,
+ "learning_rate": 5.468457057869358e-08,
+ "loss": 0.0,
+ "step": 205
+ },
+ {
+ "epoch": 2.910222222222222,
+ "grad_norm": NaN,
+ "learning_rate": 4.322413053509944e-08,
+ "loss": 0.0,
+ "step": 206
+ },
+ {
+ "epoch": 2.924444444444444,
+ "grad_norm": NaN,
+ "learning_rate": 3.31046711118066e-08,
+ "loss": 0.0,
+ "step": 207
+ },
+ {
+ "epoch": 2.9386666666666668,
+ "grad_norm": NaN,
+ "learning_rate": 2.4328929975260262e-08,
+ "loss": 0.0,
+ "step": 208
+ },
+ {
+ "epoch": 2.952888888888889,
+ "grad_norm": NaN,
+ "learning_rate": 1.6899281269279756e-08,
+ "loss": 0.0,
+ "step": 209
+ },
+ {
+ "epoch": 2.967111111111111,
+ "grad_norm": NaN,
+ "learning_rate": 1.0817734972768946e-08,
+ "loss": 0.0,
+ "step": 210
+ },
+ {
+ "epoch": 2.981333333333333,
+ "grad_norm": NaN,
+ "learning_rate": 6.085936355947897e-09,
+ "loss": 0.0,
+ "step": 211
+ },
+ {
+ "epoch": 2.9955555555555557,
+ "grad_norm": NaN,
+ "learning_rate": 2.7051655352494654e-09,
+ "loss": 0.0,
+ "step": 212
+ },
+ {
+ "epoch": 3.0,
+ "grad_norm": NaN,
+ "learning_rate": 6.763371270035457e-10,
+ "loss": 0.0,
+ "step": 213
+ }
+ ],
+ "logging_steps": 1,
+ "max_steps": 213,
+ "num_input_tokens_seen": 0,
+ "num_train_epochs": 3,
+ "save_steps": 500,
+ "stateful_callbacks": {
+ "TrainerControl": {
+ "args": {
+ "should_epoch_stop": false,
+ "should_evaluate": false,
+ "should_log": false,
+ "should_save": true,
+ "should_training_stop": true
+ },
+ "attributes": {}
+ }
+ },
+ "total_flos": 1.0308479847759872e+17,
+ "train_batch_size": 1,
+ "trial_name": null,
+ "trial_params": null
+}
diff --git a/qwen2.5-7b-instruct/checkpoint-213/training_args.bin b/qwen2.5-7b-instruct/checkpoint-213/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..b5415efad8d753f3c17d7314f34dccfa7e73e9e0
--- /dev/null
+++ b/qwen2.5-7b-instruct/checkpoint-213/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:62861a63c10b8e91d966486b438a4ca616ac892c434ce240cbba36083500e44e
+size 5649
diff --git a/qwen2.5-7b-instruct/checkpoint-71/chat_template.jinja b/qwen2.5-7b-instruct/checkpoint-71/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..bdf7919a96cfe43d50914a007b9c0877bd0ec27e
--- /dev/null
+++ b/qwen2.5-7b-instruct/checkpoint-71/chat_template.jinja
@@ -0,0 +1,54 @@
+{%- if tools %}
+ {{- '<|im_start|>system\n' }}
+ {%- if messages[0]['role'] == 'system' %}
+ {{- messages[0]['content'] }}
+ {%- else %}
+ {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+ {%- endif %}
+ {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within XML tags:\n" }}
+ {%- for tool in tools %}
+ {{- "\n" }}
+ {{- tool | tojson }}
+ {%- endfor %}
+ {{- "\n\n\nFor each function call, return a json object with function name and arguments within XML tags:\n\n{\"name\": , \"arguments\": }\n<|im_end|>\n" }}
+{%- else %}
+ {%- if messages[0]['role'] == 'system' %}
+ {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+ {%- else %}
+ {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+ {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+ {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+ {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+ {%- elif message.role == "assistant" %}
+ {{- '<|im_start|>' + message.role }}
+ {%- if message.content %}
+ {{- '\n' + message.content }}
+ {%- endif %}
+ {%- for tool_call in message.tool_calls %}
+ {%- if tool_call.function is defined %}
+ {%- set tool_call = tool_call.function %}
+ {%- endif %}
+ {{- '\n\n{"name": "' }}
+ {{- tool_call.name }}
+ {{- '", "arguments": ' }}
+ {{- tool_call.arguments | tojson }}
+ {{- '}\n' }}
+ {%- endfor %}
+ {{- '<|im_end|>\n' }}
+ {%- elif message.role == "tool" %}
+ {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+ {{- '<|im_start|>user' }}
+ {%- endif %}
+ {{- '\n\n' }}
+ {{- message.content }}
+ {{- '\n' }}
+ {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+ {{- '<|im_end|>\n' }}
+ {%- endif %}
+ {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+ {{- '<|im_start|>assistant\n' }}
+{%- endif %}
diff --git a/qwen2.5-7b-instruct/checkpoint-71/config.json b/qwen2.5-7b-instruct/checkpoint-71/config.json
new file mode 100644
index 0000000000000000000000000000000000000000..4814292863d1e5aa2f4fda1db18f4082cd61aa93
--- /dev/null
+++ b/qwen2.5-7b-instruct/checkpoint-71/config.json
@@ -0,0 +1,69 @@
+{
+ "architectures": [
+ "Qwen2ForCausalLM"
+ ],
+ "attention_dropout": 0.0,
+ "bos_token_id": null,
+ "dtype": "float32",
+ "eos_token_id": 151645,
+ "hidden_act": "silu",
+ "hidden_size": 2048,
+ "initializer_range": 0.02,
+ "intermediate_size": 11008,
+ "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",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention"
+ ],
+ "max_position_embeddings": 32768,
+ "max_window_layers": 70,
+ "model_type": "qwen2",
+ "num_attention_heads": 16,
+ "num_hidden_layers": 36,
+ "num_key_value_heads": 2,
+ "pad_token_id": 151643,
+ "rms_norm_eps": 1e-06,
+ "rope_parameters": {
+ "rope_theta": 1000000.0,
+ "rope_type": "default"
+ },
+ "sliding_window": null,
+ "tie_word_embeddings": true,
+ "transformers_version": "5.2.0",
+ "use_cache": false,
+ "use_sliding_window": false,
+ "vocab_size": 151936
+}
diff --git a/qwen2.5-7b-instruct/checkpoint-71/generation_config.json b/qwen2.5-7b-instruct/checkpoint-71/generation_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..47205fbf3386a215e7c2b9d62df6576f49b77e7b
--- /dev/null
+++ b/qwen2.5-7b-instruct/checkpoint-71/generation_config.json
@@ -0,0 +1,13 @@
+{
+ "do_sample": true,
+ "eos_token_id": [
+ 151645,
+ 151643
+ ],
+ "pad_token_id": 151643,
+ "repetition_penalty": 1.05,
+ "temperature": 0.7,
+ "top_k": 20,
+ "top_p": 0.8,
+ "transformers_version": "5.2.0"
+}
diff --git a/qwen2.5-7b-instruct/checkpoint-71/model.safetensors b/qwen2.5-7b-instruct/checkpoint-71/model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..81c8e9ad3007433383808b7ce6f310262a18b82e
--- /dev/null
+++ b/qwen2.5-7b-instruct/checkpoint-71/model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:004fc5534629a92f8fe3bcc90cb82a0516862b1f6a63604c41b52eab09a8e137
+size 12343804296
diff --git a/qwen2.5-7b-instruct/checkpoint-71/optimizer.pt b/qwen2.5-7b-instruct/checkpoint-71/optimizer.pt
new file mode 100644
index 0000000000000000000000000000000000000000..f7f6d8c9146642bbadcfcd09d8955aaaba1f688e
--- /dev/null
+++ b/qwen2.5-7b-instruct/checkpoint-71/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7f225576c667cad1ed41491c5636668335118e5a5349fcc0ec3b5ce4c3e260f6
+size 24687895753
diff --git a/qwen2.5-7b-instruct/checkpoint-71/rng_state_0.pth b/qwen2.5-7b-instruct/checkpoint-71/rng_state_0.pth
new file mode 100644
index 0000000000000000000000000000000000000000..b2e0e7337f62c934e90c141589ca185ff78fae36
--- /dev/null
+++ b/qwen2.5-7b-instruct/checkpoint-71/rng_state_0.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2fe6b873f1da64bb980afe982fc1c6b6bc461ad8c33073d856804323f64b26d9
+size 15429
diff --git a/qwen2.5-7b-instruct/checkpoint-71/rng_state_1.pth b/qwen2.5-7b-instruct/checkpoint-71/rng_state_1.pth
new file mode 100644
index 0000000000000000000000000000000000000000..204fc909b58b8ae0fab9c8b0f41ea5cc8f3277f8
--- /dev/null
+++ b/qwen2.5-7b-instruct/checkpoint-71/rng_state_1.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:13db4dd10b22b5464f5ef0b539152b46390eae6a827b355108cb54b1af2eb830
+size 15429
diff --git a/qwen2.5-7b-instruct/checkpoint-71/rng_state_2.pth b/qwen2.5-7b-instruct/checkpoint-71/rng_state_2.pth
new file mode 100644
index 0000000000000000000000000000000000000000..0c3095386194e2f629b38c7957abe0b9ab74aa8b
--- /dev/null
+++ b/qwen2.5-7b-instruct/checkpoint-71/rng_state_2.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:06e293f13fb6e873e8081636b6b07115410b397c55f1ed08b8772b059c8eb74e
+size 15429
diff --git a/qwen2.5-7b-instruct/checkpoint-71/rng_state_3.pth b/qwen2.5-7b-instruct/checkpoint-71/rng_state_3.pth
new file mode 100644
index 0000000000000000000000000000000000000000..fdf7f18d5d86f0d363c9ca44b9ad169d469e47f7
--- /dev/null
+++ b/qwen2.5-7b-instruct/checkpoint-71/rng_state_3.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c595c077664f94b7f91088561951be20e39483e01865774a83bc46f3971a3152
+size 15429
diff --git a/qwen2.5-7b-instruct/checkpoint-71/scheduler.pt b/qwen2.5-7b-instruct/checkpoint-71/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..c3be1d29bde40f225713ff5084d3d9d69858fcfd
--- /dev/null
+++ b/qwen2.5-7b-instruct/checkpoint-71/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9c96c28f12671e2f5db493417daefb3ff0b1bc36602843e0d7ef6df3649c27b3
+size 1465
diff --git a/qwen2.5-7b-instruct/checkpoint-71/tokenizer.json b/qwen2.5-7b-instruct/checkpoint-71/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..34510ff0037cd50428af467a17ead5a96140a32c
--- /dev/null
+++ b/qwen2.5-7b-instruct/checkpoint-71/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
+size 11421892
diff --git a/qwen2.5-7b-instruct/checkpoint-71/tokenizer_config.json b/qwen2.5-7b-instruct/checkpoint-71/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..145e2c7a68b109ae0253dda29fc2e66cb01b4b85
--- /dev/null
+++ b/qwen2.5-7b-instruct/checkpoint-71/tokenizer_config.json
@@ -0,0 +1,30 @@
+{
+ "add_prefix_space": false,
+ "backend": "tokenizers",
+ "bos_token": null,
+ "clean_up_tokenization_spaces": false,
+ "eos_token": "<|im_end|>",
+ "errors": "replace",
+ "extra_special_tokens": [
+ "<|im_start|>",
+ "<|im_end|>",
+ "<|object_ref_start|>",
+ "<|object_ref_end|>",
+ "<|box_start|>",
+ "<|box_end|>",
+ "<|quad_start|>",
+ "<|quad_end|>",
+ "<|vision_start|>",
+ "<|vision_end|>",
+ "<|vision_pad|>",
+ "<|image_pad|>",
+ "<|video_pad|>"
+ ],
+ "is_local": false,
+ "model_max_length": 131072,
+ "pad_token": "<|endoftext|>",
+ "padding_side": "right",
+ "split_special_tokens": false,
+ "tokenizer_class": "Qwen2Tokenizer",
+ "unk_token": null
+}
diff --git a/qwen2.5-7b-instruct/checkpoint-71/trainer_state.json b/qwen2.5-7b-instruct/checkpoint-71/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..f42ece582e3a29204fb995b56dc60a07c902f99d
--- /dev/null
+++ b/qwen2.5-7b-instruct/checkpoint-71/trainer_state.json
@@ -0,0 +1,539 @@
+{
+ "best_global_step": null,
+ "best_metric": null,
+ "best_model_checkpoint": null,
+ "epoch": 1.0,
+ "eval_steps": 50,
+ "global_step": 71,
+ "is_hyper_param_search": false,
+ "is_local_process_zero": true,
+ "is_world_process_zero": true,
+ "log_history": [
+ {
+ "epoch": 0.014222222222222223,
+ "grad_norm": 0.506786584854126,
+ "learning_rate": 0.0,
+ "loss": 1.1951476335525513,
+ "step": 1
+ },
+ {
+ "epoch": 0.028444444444444446,
+ "grad_norm": 0.5440636277198792,
+ "learning_rate": 4.5454545454545457e-07,
+ "loss": 1.2398298978805542,
+ "step": 2
+ },
+ {
+ "epoch": 0.042666666666666665,
+ "grad_norm": 0.4429919421672821,
+ "learning_rate": 9.090909090909091e-07,
+ "loss": 0.9777733087539673,
+ "step": 3
+ },
+ {
+ "epoch": 0.05688888888888889,
+ "grad_norm": 0.34922653436660767,
+ "learning_rate": 1.3636363636363636e-06,
+ "loss": 0.8064044713973999,
+ "step": 4
+ },
+ {
+ "epoch": 0.07111111111111111,
+ "grad_norm": 0.16975589096546173,
+ "learning_rate": 1.8181818181818183e-06,
+ "loss": 0.5294505953788757,
+ "step": 5
+ },
+ {
+ "epoch": 0.08533333333333333,
+ "grad_norm": 0.16188450157642365,
+ "learning_rate": 2.2727272727272728e-06,
+ "loss": 0.636877179145813,
+ "step": 6
+ },
+ {
+ "epoch": 0.09955555555555555,
+ "grad_norm": 0.11110807955265045,
+ "learning_rate": 2.7272727272727272e-06,
+ "loss": 0.6087832450866699,
+ "step": 7
+ },
+ {
+ "epoch": 0.11377777777777778,
+ "grad_norm": 0.0641367956995964,
+ "learning_rate": 3.181818181818182e-06,
+ "loss": 0.5258303880691528,
+ "step": 8
+ },
+ {
+ "epoch": 0.128,
+ "grad_norm": 0.052072592079639435,
+ "learning_rate": 3.6363636363636366e-06,
+ "loss": 0.5444644689559937,
+ "step": 9
+ },
+ {
+ "epoch": 0.14222222222222222,
+ "grad_norm": 0.0392303504049778,
+ "learning_rate": 4.0909090909090915e-06,
+ "loss": 0.5337975025177002,
+ "step": 10
+ },
+ {
+ "epoch": 0.15644444444444444,
+ "grad_norm": 0.028075236827135086,
+ "learning_rate": 4.5454545454545455e-06,
+ "loss": 0.45043277740478516,
+ "step": 11
+ },
+ {
+ "epoch": 0.17066666666666666,
+ "grad_norm": 0.02663014456629753,
+ "learning_rate": 5e-06,
+ "loss": 0.49355342984199524,
+ "step": 12
+ },
+ {
+ "epoch": 0.18488888888888888,
+ "grad_norm": 0.028838366270065308,
+ "learning_rate": 5.4545454545454545e-06,
+ "loss": 0.487549751996994,
+ "step": 13
+ },
+ {
+ "epoch": 0.1991111111111111,
+ "grad_norm": 0.03057004138827324,
+ "learning_rate": 5.90909090909091e-06,
+ "loss": 0.500217616558075,
+ "step": 14
+ },
+ {
+ "epoch": 0.21333333333333335,
+ "grad_norm": 0.028651155531406403,
+ "learning_rate": 6.363636363636364e-06,
+ "loss": 0.49628522992134094,
+ "step": 15
+ },
+ {
+ "epoch": 0.22755555555555557,
+ "grad_norm": 0.028336627408862114,
+ "learning_rate": 6.818181818181818e-06,
+ "loss": 0.48170676827430725,
+ "step": 16
+ },
+ {
+ "epoch": 0.24177777777777779,
+ "grad_norm": 0.019824286922812462,
+ "learning_rate": 7.272727272727273e-06,
+ "loss": 0.4609394073486328,
+ "step": 17
+ },
+ {
+ "epoch": 0.256,
+ "grad_norm": 0.02125639095902443,
+ "learning_rate": 7.727272727272727e-06,
+ "loss": 0.4759177565574646,
+ "step": 18
+ },
+ {
+ "epoch": 0.2702222222222222,
+ "grad_norm": 0.025788351893424988,
+ "learning_rate": 8.181818181818183e-06,
+ "loss": 0.4962843060493469,
+ "step": 19
+ },
+ {
+ "epoch": 0.28444444444444444,
+ "grad_norm": 0.033310867846012115,
+ "learning_rate": 8.636363636363637e-06,
+ "loss": 0.5028114318847656,
+ "step": 20
+ },
+ {
+ "epoch": 0.2986666666666667,
+ "grad_norm": 0.020624130964279175,
+ "learning_rate": 9.090909090909091e-06,
+ "loss": 0.44575220346450806,
+ "step": 21
+ },
+ {
+ "epoch": 0.3128888888888889,
+ "grad_norm": 0.020986203104257584,
+ "learning_rate": 9.545454545454547e-06,
+ "loss": 0.4755765199661255,
+ "step": 22
+ },
+ {
+ "epoch": 0.32711111111111113,
+ "grad_norm": 2526660.75,
+ "learning_rate": 1e-05,
+ "loss": 0.4706365466117859,
+ "step": 23
+ },
+ {
+ "epoch": 0.3413333333333333,
+ "grad_norm": 0.021011805161833763,
+ "learning_rate": 9.999323662872998e-06,
+ "loss": 0.4945886731147766,
+ "step": 24
+ },
+ {
+ "epoch": 0.35555555555555557,
+ "grad_norm": 0.020573964342474937,
+ "learning_rate": 9.99729483446475e-06,
+ "loss": 0.4544343054294586,
+ "step": 25
+ },
+ {
+ "epoch": 0.36977777777777776,
+ "grad_norm": 0.021116767078638077,
+ "learning_rate": 9.993914063644053e-06,
+ "loss": 0.4556806683540344,
+ "step": 26
+ },
+ {
+ "epoch": 0.384,
+ "grad_norm": 0.019978955388069153,
+ "learning_rate": 9.989182265027232e-06,
+ "loss": 0.4431076645851135,
+ "step": 27
+ },
+ {
+ "epoch": 0.3982222222222222,
+ "grad_norm": 0.019631512463092804,
+ "learning_rate": 9.98310071873072e-06,
+ "loss": 0.4541151523590088,
+ "step": 28
+ },
+ {
+ "epoch": 0.41244444444444445,
+ "grad_norm": 0.01977457106113434,
+ "learning_rate": 9.975671070024741e-06,
+ "loss": 0.4352927803993225,
+ "step": 29
+ },
+ {
+ "epoch": 0.4266666666666667,
+ "grad_norm": 0.022281618788838387,
+ "learning_rate": 9.966895328888195e-06,
+ "loss": 0.5559179186820984,
+ "step": 30
+ },
+ {
+ "epoch": 0.4408888888888889,
+ "grad_norm": 0.018352897837758064,
+ "learning_rate": 9.956775869464901e-06,
+ "loss": 0.42942947149276733,
+ "step": 31
+ },
+ {
+ "epoch": 0.45511111111111113,
+ "grad_norm": 0.019121287390589714,
+ "learning_rate": 9.945315429421307e-06,
+ "loss": 0.45916861295700073,
+ "step": 32
+ },
+ {
+ "epoch": 0.4693333333333333,
+ "grad_norm": 0.01712602935731411,
+ "learning_rate": 9.932517109205849e-06,
+ "loss": 0.4312085807323456,
+ "step": 33
+ },
+ {
+ "epoch": 0.48355555555555557,
+ "grad_norm": 0.01892935112118721,
+ "learning_rate": 9.918384371210178e-06,
+ "loss": 0.5169366598129272,
+ "step": 34
+ },
+ {
+ "epoch": 0.49777777777777776,
+ "grad_norm": 0.02034984715282917,
+ "learning_rate": 9.902921038832456e-06,
+ "loss": 0.49438145756721497,
+ "step": 35
+ },
+ {
+ "epoch": 0.512,
+ "grad_norm": 0.017887447029352188,
+ "learning_rate": 9.886131295443003e-06,
+ "loss": 0.4683716893196106,
+ "step": 36
+ },
+ {
+ "epoch": 0.5262222222222223,
+ "grad_norm": 0.01887686364352703,
+ "learning_rate": 9.868019683252543e-06,
+ "loss": 0.46688079833984375,
+ "step": 37
+ },
+ {
+ "epoch": 0.5404444444444444,
+ "grad_norm": 0.019363289698958397,
+ "learning_rate": 9.848591102083375e-06,
+ "loss": 0.4786224067211151,
+ "step": 38
+ },
+ {
+ "epoch": 0.5546666666666666,
+ "grad_norm": 0.0201108418405056,
+ "learning_rate": 9.82785080804381e-06,
+ "loss": 0.47358086705207825,
+ "step": 39
+ },
+ {
+ "epoch": 0.5688888888888889,
+ "grad_norm": 0.019631700590252876,
+ "learning_rate": 9.805804412106197e-06,
+ "loss": 0.585416316986084,
+ "step": 40
+ },
+ {
+ "epoch": 0.5831111111111111,
+ "grad_norm": 0.01689487136900425,
+ "learning_rate": 9.782457878588977e-06,
+ "loss": 0.47205525636672974,
+ "step": 41
+ },
+ {
+ "epoch": 0.5973333333333334,
+ "grad_norm": 0.01628926210105419,
+ "learning_rate": 9.75781752354311e-06,
+ "loss": 0.4348371624946594,
+ "step": 42
+ },
+ {
+ "epoch": 0.6115555555555555,
+ "grad_norm": 0.01758437603712082,
+ "learning_rate": 9.731890013043367e-06,
+ "loss": 0.4891083538532257,
+ "step": 43
+ },
+ {
+ "epoch": 0.6257777777777778,
+ "grad_norm": 0.017766334116458893,
+ "learning_rate": 9.704682361384941e-06,
+ "loss": 0.46330946683883667,
+ "step": 44
+ },
+ {
+ "epoch": 0.64,
+ "grad_norm": 0.017524050548672676,
+ "learning_rate": 9.676201929185809e-06,
+ "loss": 0.4083091914653778,
+ "step": 45
+ },
+ {
+ "epoch": 0.6542222222222223,
+ "grad_norm": 0.017291702330112457,
+ "learning_rate": 9.646456421395447e-06,
+ "loss": 0.4653710126876831,
+ "step": 46
+ },
+ {
+ "epoch": 0.6684444444444444,
+ "grad_norm": 0.017993608489632607,
+ "learning_rate": 9.615453885210368e-06,
+ "loss": 0.4626290202140808,
+ "step": 47
+ },
+ {
+ "epoch": 0.6826666666666666,
+ "grad_norm": 0.017247265204787254,
+ "learning_rate": 9.583202707897075e-06,
+ "loss": 0.47964128851890564,
+ "step": 48
+ },
+ {
+ "epoch": 0.6968888888888889,
+ "grad_norm": 0.017244094982743263,
+ "learning_rate": 9.549711614523007e-06,
+ "loss": 0.46718651056289673,
+ "step": 49
+ },
+ {
+ "epoch": 0.7111111111111111,
+ "grad_norm": 0.01574530452489853,
+ "learning_rate": 9.514989665596114e-06,
+ "loss": 0.4456287622451782,
+ "step": 50
+ },
+ {
+ "epoch": 0.7111111111111111,
+ "eval_loss": 0.4940151274204254,
+ "eval_runtime": 6.2328,
+ "eval_samples_per_second": 32.088,
+ "eval_steps_per_second": 8.022,
+ "step": 50
+ },
+ {
+ "epoch": 0.7253333333333334,
+ "grad_norm": 0.01783670485019684,
+ "learning_rate": 9.479046254613673e-06,
+ "loss": 0.45046865940093994,
+ "step": 51
+ },
+ {
+ "epoch": 0.7395555555555555,
+ "grad_norm": 0.015281656756997108,
+ "learning_rate": 9.441891105521005e-06,
+ "loss": 0.43093934655189514,
+ "step": 52
+ },
+ {
+ "epoch": 0.7537777777777778,
+ "grad_norm": 0.017853204160928726,
+ "learning_rate": 9.40353427008083e-06,
+ "loss": 0.4706975519657135,
+ "step": 53
+ },
+ {
+ "epoch": 0.768,
+ "grad_norm": 0.01542010996490717,
+ "learning_rate": 9.3639861251539e-06,
+ "loss": 0.4405261278152466,
+ "step": 54
+ },
+ {
+ "epoch": 0.7822222222222223,
+ "grad_norm": 0.017832862213253975,
+ "learning_rate": 9.323257369891702e-06,
+ "loss": 0.47680750489234924,
+ "step": 55
+ },
+ {
+ "epoch": 0.7964444444444444,
+ "grad_norm": 0.015628881752490997,
+ "learning_rate": 9.281359022841966e-06,
+ "loss": 0.42392900586128235,
+ "step": 56
+ },
+ {
+ "epoch": 0.8106666666666666,
+ "grad_norm": 0.01757017895579338,
+ "learning_rate": 9.238302418967757e-06,
+ "loss": 0.48122283816337585,
+ "step": 57
+ },
+ {
+ "epoch": 0.8248888888888889,
+ "grad_norm": 0.017186852172017097,
+ "learning_rate": 9.194099206580981e-06,
+ "loss": 0.5255264043807983,
+ "step": 58
+ },
+ {
+ "epoch": 0.8391111111111111,
+ "grad_norm": 0.017161328345537186,
+ "learning_rate": 9.14876134419111e-06,
+ "loss": 0.4682884216308594,
+ "step": 59
+ },
+ {
+ "epoch": 0.8533333333333334,
+ "grad_norm": 0.01665889099240303,
+ "learning_rate": 9.102301097269974e-06,
+ "loss": 0.5003230571746826,
+ "step": 60
+ },
+ {
+ "epoch": 0.8675555555555555,
+ "grad_norm": 0.02079184167087078,
+ "learning_rate": 9.05473103493355e-06,
+ "loss": 0.45744988322257996,
+ "step": 61
+ },
+ {
+ "epoch": 0.8817777777777778,
+ "grad_norm": 0.018035434186458588,
+ "learning_rate": 9.006064026541549e-06,
+ "loss": 0.5037047863006592,
+ "step": 62
+ },
+ {
+ "epoch": 0.896,
+ "grad_norm": 0.015980787575244904,
+ "learning_rate": 8.956313238215824e-06,
+ "loss": 0.4382043778896332,
+ "step": 63
+ },
+ {
+ "epoch": 0.9102222222222223,
+ "grad_norm": 0.016865955665707588,
+ "learning_rate": 8.905492129278478e-06,
+ "loss": 0.44455963373184204,
+ "step": 64
+ },
+ {
+ "epoch": 0.9244444444444444,
+ "grad_norm": 0.014777022413909435,
+ "learning_rate": 8.85361444861063e-06,
+ "loss": 0.4215100109577179,
+ "step": 65
+ },
+ {
+ "epoch": 0.9386666666666666,
+ "grad_norm": 0.01814866065979004,
+ "learning_rate": 8.800694230932885e-06,
+ "loss": 0.48813939094543457,
+ "step": 66
+ },
+ {
+ "epoch": 0.9528888888888889,
+ "grad_norm": 0.013353748247027397,
+ "learning_rate": 8.74674579300843e-06,
+ "loss": 3.263723850250244,
+ "step": 67
+ },
+ {
+ "epoch": 0.9671111111111111,
+ "grad_norm": 0.014444677159190178,
+ "learning_rate": 8.691783729769874e-06,
+ "loss": 2.8557608127593994,
+ "step": 68
+ },
+ {
+ "epoch": 0.9813333333333333,
+ "grad_norm": 0.01388335507363081,
+ "learning_rate": 8.635822910370793e-06,
+ "loss": 3.485170602798462,
+ "step": 69
+ },
+ {
+ "epoch": 0.9955555555555555,
+ "grad_norm": 0.01357923075556755,
+ "learning_rate": 8.578878474163115e-06,
+ "loss": 3.1958985328674316,
+ "step": 70
+ },
+ {
+ "epoch": 1.0,
+ "grad_norm": 0.02305089496076107,
+ "learning_rate": 8.520965826601394e-06,
+ "loss": 2.678990364074707,
+ "step": 71
+ }
+ ],
+ "logging_steps": 1,
+ "max_steps": 213,
+ "num_input_tokens_seen": 0,
+ "num_train_epochs": 3,
+ "save_steps": 500,
+ "stateful_callbacks": {
+ "TrainerControl": {
+ "args": {
+ "should_epoch_stop": false,
+ "should_evaluate": false,
+ "should_log": false,
+ "should_save": true,
+ "should_training_stop": false
+ },
+ "attributes": {}
+ }
+ },
+ "total_flos": 3.436159949812531e+16,
+ "train_batch_size": 1,
+ "trial_name": null,
+ "trial_params": null
+}
diff --git a/qwen2.5-7b-instruct/checkpoint-71/training_args.bin b/qwen2.5-7b-instruct/checkpoint-71/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..b5415efad8d753f3c17d7314f34dccfa7e73e9e0
--- /dev/null
+++ b/qwen2.5-7b-instruct/checkpoint-71/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:62861a63c10b8e91d966486b438a4ca616ac892c434ce240cbba36083500e44e
+size 5649
diff --git a/qwen2.5-7b-instruct/config.json b/qwen2.5-7b-instruct/config.json
new file mode 100644
index 0000000000000000000000000000000000000000..4814292863d1e5aa2f4fda1db18f4082cd61aa93
--- /dev/null
+++ b/qwen2.5-7b-instruct/config.json
@@ -0,0 +1,69 @@
+{
+ "architectures": [
+ "Qwen2ForCausalLM"
+ ],
+ "attention_dropout": 0.0,
+ "bos_token_id": null,
+ "dtype": "float32",
+ "eos_token_id": 151645,
+ "hidden_act": "silu",
+ "hidden_size": 2048,
+ "initializer_range": 0.02,
+ "intermediate_size": 11008,
+ "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",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention",
+ "full_attention"
+ ],
+ "max_position_embeddings": 32768,
+ "max_window_layers": 70,
+ "model_type": "qwen2",
+ "num_attention_heads": 16,
+ "num_hidden_layers": 36,
+ "num_key_value_heads": 2,
+ "pad_token_id": 151643,
+ "rms_norm_eps": 1e-06,
+ "rope_parameters": {
+ "rope_theta": 1000000.0,
+ "rope_type": "default"
+ },
+ "sliding_window": null,
+ "tie_word_embeddings": true,
+ "transformers_version": "5.2.0",
+ "use_cache": false,
+ "use_sliding_window": false,
+ "vocab_size": 151936
+}
diff --git a/qwen2.5-7b-instruct/eval_results.json b/qwen2.5-7b-instruct/eval_results.json
new file mode 100644
index 0000000000000000000000000000000000000000..cb74902e5d769edef250d8f0f13d3ef873d8b685
--- /dev/null
+++ b/qwen2.5-7b-instruct/eval_results.json
@@ -0,0 +1,7 @@
+{
+ "epoch": 3.0,
+ "eval_loss": NaN,
+ "eval_runtime": 5.4995,
+ "eval_samples_per_second": 36.367,
+ "eval_steps_per_second": 9.092
+}
\ No newline at end of file
diff --git a/qwen2.5-7b-instruct/generation_config.json b/qwen2.5-7b-instruct/generation_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..47205fbf3386a215e7c2b9d62df6576f49b77e7b
--- /dev/null
+++ b/qwen2.5-7b-instruct/generation_config.json
@@ -0,0 +1,13 @@
+{
+ "do_sample": true,
+ "eos_token_id": [
+ 151645,
+ 151643
+ ],
+ "pad_token_id": 151643,
+ "repetition_penalty": 1.05,
+ "temperature": 0.7,
+ "top_k": 20,
+ "top_p": 0.8,
+ "transformers_version": "5.2.0"
+}
diff --git a/qwen2.5-7b-instruct/model.safetensors b/qwen2.5-7b-instruct/model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..61029f7ead30eb65ec9a23e4c34c1f73a7a1af78
--- /dev/null
+++ b/qwen2.5-7b-instruct/model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:98f699d765f656b4c23b83714137d84fa3f23ec986eebd693963185de38bd7b3
+size 12343804296
diff --git a/qwen2.5-7b-instruct/tokenizer.json b/qwen2.5-7b-instruct/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..34510ff0037cd50428af467a17ead5a96140a32c
--- /dev/null
+++ b/qwen2.5-7b-instruct/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
+size 11421892
diff --git a/qwen2.5-7b-instruct/tokenizer_config.json b/qwen2.5-7b-instruct/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..145e2c7a68b109ae0253dda29fc2e66cb01b4b85
--- /dev/null
+++ b/qwen2.5-7b-instruct/tokenizer_config.json
@@ -0,0 +1,30 @@
+{
+ "add_prefix_space": false,
+ "backend": "tokenizers",
+ "bos_token": null,
+ "clean_up_tokenization_spaces": false,
+ "eos_token": "<|im_end|>",
+ "errors": "replace",
+ "extra_special_tokens": [
+ "<|im_start|>",
+ "<|im_end|>",
+ "<|object_ref_start|>",
+ "<|object_ref_end|>",
+ "<|box_start|>",
+ "<|box_end|>",
+ "<|quad_start|>",
+ "<|quad_end|>",
+ "<|vision_start|>",
+ "<|vision_end|>",
+ "<|vision_pad|>",
+ "<|image_pad|>",
+ "<|video_pad|>"
+ ],
+ "is_local": false,
+ "model_max_length": 131072,
+ "pad_token": "<|endoftext|>",
+ "padding_side": "right",
+ "split_special_tokens": false,
+ "tokenizer_class": "Qwen2Tokenizer",
+ "unk_token": null
+}
diff --git a/qwen2.5-7b-instruct/train_results.json b/qwen2.5-7b-instruct/train_results.json
new file mode 100644
index 0000000000000000000000000000000000000000..10ad1760a170d0b32e3d60c9f1d06e33a034a4e6
--- /dev/null
+++ b/qwen2.5-7b-instruct/train_results.json
@@ -0,0 +1,8 @@
+{
+ "epoch": 3.0,
+ "total_flos": 1.0308479847759872e+17,
+ "train_loss": 1.8930653027245696,
+ "train_runtime": 5673.0769,
+ "train_samples_per_second": 9.519,
+ "train_steps_per_second": 0.038
+}
\ No newline at end of file
diff --git a/qwen2.5-7b-instruct/trainer_log.jsonl b/qwen2.5-7b-instruct/trainer_log.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..0a1d6fa436adc06fead8f098eab2b80f550e4a8a
--- /dev/null
+++ b/qwen2.5-7b-instruct/trainer_log.jsonl
@@ -0,0 +1,218 @@
+{"current_steps": 1, "total_steps": 213, "loss": 1.1951476335525513, "lr": 0.0, "epoch": 0.014222222222222223, "percentage": 0.47, "elapsed_time": "0:00:40", "remaining_time": "2:24:49"}
+{"current_steps": 2, "total_steps": 213, "loss": 1.2398298978805542, "lr": 4.5454545454545457e-07, "epoch": 0.028444444444444446, "percentage": 0.94, "elapsed_time": "0:01:18", "remaining_time": "2:18:11"}
+{"current_steps": 3, "total_steps": 213, "loss": 0.9777733087539673, "lr": 9.090909090909091e-07, "epoch": 0.042666666666666665, "percentage": 1.41, "elapsed_time": "0:01:56", "remaining_time": "2:15:53"}
+{"current_steps": 4, "total_steps": 213, "loss": 0.8064044713973999, "lr": 1.3636363636363636e-06, "epoch": 0.05688888888888889, "percentage": 1.88, "elapsed_time": "0:02:34", "remaining_time": "2:14:23"}
+{"current_steps": 5, "total_steps": 213, "loss": 0.5294505953788757, "lr": 1.8181818181818183e-06, "epoch": 0.07111111111111111, "percentage": 2.35, "elapsed_time": "0:03:09", "remaining_time": "2:11:38"}
+{"current_steps": 6, "total_steps": 213, "loss": 0.636877179145813, "lr": 2.2727272727272728e-06, "epoch": 0.08533333333333333, "percentage": 2.82, "elapsed_time": "0:03:33", "remaining_time": "2:02:48"}
+{"current_steps": 7, "total_steps": 213, "loss": 0.6087832450866699, "lr": 2.7272727272727272e-06, "epoch": 0.09955555555555555, "percentage": 3.29, "elapsed_time": "0:03:57", "remaining_time": "1:56:26"}
+{"current_steps": 8, "total_steps": 213, "loss": 0.5258303880691528, "lr": 3.181818181818182e-06, "epoch": 0.11377777777777778, "percentage": 3.76, "elapsed_time": "0:04:20", "remaining_time": "1:51:27"}
+{"current_steps": 9, "total_steps": 213, "loss": 0.5444644689559937, "lr": 3.6363636363636366e-06, "epoch": 0.128, "percentage": 4.23, "elapsed_time": "0:04:44", "remaining_time": "1:47:37"}
+{"current_steps": 10, "total_steps": 213, "loss": 0.5337975025177002, "lr": 4.0909090909090915e-06, "epoch": 0.14222222222222222, "percentage": 4.69, "elapsed_time": "0:05:08", "remaining_time": "1:44:27"}
+{"current_steps": 11, "total_steps": 213, "loss": 0.45043277740478516, "lr": 4.5454545454545455e-06, "epoch": 0.15644444444444444, "percentage": 5.16, "elapsed_time": "0:05:32", "remaining_time": "1:41:48"}
+{"current_steps": 12, "total_steps": 213, "loss": 0.49355342984199524, "lr": 5e-06, "epoch": 0.17066666666666666, "percentage": 5.63, "elapsed_time": "0:05:56", "remaining_time": "1:39:32"}
+{"current_steps": 13, "total_steps": 213, "loss": 0.487549751996994, "lr": 5.4545454545454545e-06, "epoch": 0.18488888888888888, "percentage": 6.1, "elapsed_time": "0:06:20", "remaining_time": "1:37:28"}
+{"current_steps": 14, "total_steps": 213, "loss": 0.500217616558075, "lr": 5.90909090909091e-06, "epoch": 0.1991111111111111, "percentage": 6.57, "elapsed_time": "0:06:43", "remaining_time": "1:35:41"}
+{"current_steps": 15, "total_steps": 213, "loss": 0.49628522992134094, "lr": 6.363636363636364e-06, "epoch": 0.21333333333333335, "percentage": 7.04, "elapsed_time": "0:07:07", "remaining_time": "1:34:03"}
+{"current_steps": 16, "total_steps": 213, "loss": 0.48170676827430725, "lr": 6.818181818181818e-06, "epoch": 0.22755555555555557, "percentage": 7.51, "elapsed_time": "0:07:31", "remaining_time": "1:32:38"}
+{"current_steps": 17, "total_steps": 213, "loss": 0.4609394073486328, "lr": 7.272727272727273e-06, "epoch": 0.24177777777777779, "percentage": 7.98, "elapsed_time": "0:07:55", "remaining_time": "1:31:23"}
+{"current_steps": 18, "total_steps": 213, "loss": 0.4759177565574646, "lr": 7.727272727272727e-06, "epoch": 0.256, "percentage": 8.45, "elapsed_time": "0:08:19", "remaining_time": "1:30:06"}
+{"current_steps": 19, "total_steps": 213, "loss": 0.4962843060493469, "lr": 8.181818181818183e-06, "epoch": 0.2702222222222222, "percentage": 8.92, "elapsed_time": "0:08:43", "remaining_time": "1:29:00"}
+{"current_steps": 20, "total_steps": 213, "loss": 0.5028114318847656, "lr": 8.636363636363637e-06, "epoch": 0.28444444444444444, "percentage": 9.39, "elapsed_time": "0:09:06", "remaining_time": "1:27:55"}
+{"current_steps": 21, "total_steps": 213, "loss": 0.44575220346450806, "lr": 9.090909090909091e-06, "epoch": 0.2986666666666667, "percentage": 9.86, "elapsed_time": "0:09:30", "remaining_time": "1:26:55"}
+{"current_steps": 22, "total_steps": 213, "loss": 0.4755765199661255, "lr": 9.545454545454547e-06, "epoch": 0.3128888888888889, "percentage": 10.33, "elapsed_time": "0:09:54", "remaining_time": "1:25:59"}
+{"current_steps": 23, "total_steps": 213, "loss": 0.4706365466117859, "lr": 1e-05, "epoch": 0.32711111111111113, "percentage": 10.8, "elapsed_time": "0:10:17", "remaining_time": "1:25:01"}
+{"current_steps": 24, "total_steps": 213, "loss": 0.4945886731147766, "lr": 9.999323662872998e-06, "epoch": 0.3413333333333333, "percentage": 11.27, "elapsed_time": "0:10:41", "remaining_time": "1:24:11"}
+{"current_steps": 25, "total_steps": 213, "loss": 0.4544343054294586, "lr": 9.99729483446475e-06, "epoch": 0.35555555555555557, "percentage": 11.74, "elapsed_time": "0:11:05", "remaining_time": "1:23:22"}
+{"current_steps": 26, "total_steps": 213, "loss": 0.4556806683540344, "lr": 9.993914063644053e-06, "epoch": 0.36977777777777776, "percentage": 12.21, "elapsed_time": "0:11:29", "remaining_time": "1:22:37"}
+{"current_steps": 27, "total_steps": 213, "loss": 0.4431076645851135, "lr": 9.989182265027232e-06, "epoch": 0.384, "percentage": 12.68, "elapsed_time": "0:11:53", "remaining_time": "1:21:53"}
+{"current_steps": 28, "total_steps": 213, "loss": 0.4541151523590088, "lr": 9.98310071873072e-06, "epoch": 0.3982222222222222, "percentage": 13.15, "elapsed_time": "0:12:16", "remaining_time": "1:21:08"}
+{"current_steps": 29, "total_steps": 213, "loss": 0.4352927803993225, "lr": 9.975671070024741e-06, "epoch": 0.41244444444444445, "percentage": 13.62, "elapsed_time": "0:12:40", "remaining_time": "1:20:26"}
+{"current_steps": 30, "total_steps": 213, "loss": 0.5559179186820984, "lr": 9.966895328888195e-06, "epoch": 0.4266666666666667, "percentage": 14.08, "elapsed_time": "0:13:04", "remaining_time": "1:19:47"}
+{"current_steps": 31, "total_steps": 213, "loss": 0.42942947149276733, "lr": 9.956775869464901e-06, "epoch": 0.4408888888888889, "percentage": 14.55, "elapsed_time": "0:13:28", "remaining_time": "1:19:07"}
+{"current_steps": 32, "total_steps": 213, "loss": 0.45916861295700073, "lr": 9.945315429421307e-06, "epoch": 0.45511111111111113, "percentage": 15.02, "elapsed_time": "0:13:52", "remaining_time": "1:18:27"}
+{"current_steps": 33, "total_steps": 213, "loss": 0.4312085807323456, "lr": 9.932517109205849e-06, "epoch": 0.4693333333333333, "percentage": 15.49, "elapsed_time": "0:14:16", "remaining_time": "1:17:50"}
+{"current_steps": 34, "total_steps": 213, "loss": 0.5169366598129272, "lr": 9.918384371210178e-06, "epoch": 0.48355555555555557, "percentage": 15.96, "elapsed_time": "0:14:40", "remaining_time": "1:17:12"}
+{"current_steps": 35, "total_steps": 213, "loss": 0.49438145756721497, "lr": 9.902921038832456e-06, "epoch": 0.49777777777777776, "percentage": 16.43, "elapsed_time": "0:15:03", "remaining_time": "1:16:37"}
+{"current_steps": 36, "total_steps": 213, "loss": 0.4683716893196106, "lr": 9.886131295443003e-06, "epoch": 0.512, "percentage": 16.9, "elapsed_time": "0:15:27", "remaining_time": "1:16:01"}
+{"current_steps": 37, "total_steps": 213, "loss": 0.46688079833984375, "lr": 9.868019683252543e-06, "epoch": 0.5262222222222223, "percentage": 17.37, "elapsed_time": "0:15:51", "remaining_time": "1:15:24"}
+{"current_steps": 38, "total_steps": 213, "loss": 0.4786224067211151, "lr": 9.848591102083375e-06, "epoch": 0.5404444444444444, "percentage": 17.84, "elapsed_time": "0:16:14", "remaining_time": "1:14:49"}
+{"current_steps": 39, "total_steps": 213, "loss": 0.47358086705207825, "lr": 9.82785080804381e-06, "epoch": 0.5546666666666666, "percentage": 18.31, "elapsed_time": "0:16:38", "remaining_time": "1:14:15"}
+{"current_steps": 40, "total_steps": 213, "loss": 0.585416316986084, "lr": 9.805804412106197e-06, "epoch": 0.5688888888888889, "percentage": 18.78, "elapsed_time": "0:17:02", "remaining_time": "1:13:42"}
+{"current_steps": 41, "total_steps": 213, "loss": 0.47205525636672974, "lr": 9.782457878588977e-06, "epoch": 0.5831111111111111, "percentage": 19.25, "elapsed_time": "0:17:26", "remaining_time": "1:13:11"}
+{"current_steps": 42, "total_steps": 213, "loss": 0.4348371624946594, "lr": 9.75781752354311e-06, "epoch": 0.5973333333333334, "percentage": 19.72, "elapsed_time": "0:17:50", "remaining_time": "1:12:37"}
+{"current_steps": 43, "total_steps": 213, "loss": 0.4891083538532257, "lr": 9.731890013043367e-06, "epoch": 0.6115555555555555, "percentage": 20.19, "elapsed_time": "0:18:14", "remaining_time": "1:12:05"}
+{"current_steps": 44, "total_steps": 213, "loss": 0.46330946683883667, "lr": 9.704682361384941e-06, "epoch": 0.6257777777777778, "percentage": 20.66, "elapsed_time": "0:18:37", "remaining_time": "1:11:34"}
+{"current_steps": 45, "total_steps": 213, "loss": 0.4083091914653778, "lr": 9.676201929185809e-06, "epoch": 0.64, "percentage": 21.13, "elapsed_time": "0:19:01", "remaining_time": "1:11:02"}
+{"current_steps": 46, "total_steps": 213, "loss": 0.4653710126876831, "lr": 9.646456421395447e-06, "epoch": 0.6542222222222223, "percentage": 21.6, "elapsed_time": "0:19:25", "remaining_time": "1:10:32"}
+{"current_steps": 47, "total_steps": 213, "loss": 0.4626290202140808, "lr": 9.615453885210368e-06, "epoch": 0.6684444444444444, "percentage": 22.07, "elapsed_time": "0:19:49", "remaining_time": "1:10:02"}
+{"current_steps": 48, "total_steps": 213, "loss": 0.47964128851890564, "lr": 9.583202707897075e-06, "epoch": 0.6826666666666666, "percentage": 22.54, "elapsed_time": "0:20:13", "remaining_time": "1:09:32"}
+{"current_steps": 49, "total_steps": 213, "loss": 0.46718651056289673, "lr": 9.549711614523007e-06, "epoch": 0.6968888888888889, "percentage": 23.0, "elapsed_time": "0:20:37", "remaining_time": "1:09:01"}
+{"current_steps": 50, "total_steps": 213, "loss": 0.4456287622451782, "lr": 9.514989665596114e-06, "epoch": 0.7111111111111111, "percentage": 23.47, "elapsed_time": "0:21:01", "remaining_time": "1:08:31"}
+{"current_steps": 50, "total_steps": 213, "eval_loss": 0.4940151274204254, "epoch": 0.7111111111111111, "percentage": 23.47, "elapsed_time": "0:21:08", "remaining_time": "1:08:54"}
+{"current_steps": 51, "total_steps": 213, "loss": 0.45046865940093994, "lr": 9.479046254613673e-06, "epoch": 0.7253333333333334, "percentage": 23.94, "elapsed_time": "0:21:32", "remaining_time": "1:08:24"}
+{"current_steps": 52, "total_steps": 213, "loss": 0.43093934655189514, "lr": 9.441891105521005e-06, "epoch": 0.7395555555555555, "percentage": 24.41, "elapsed_time": "0:21:56", "remaining_time": "1:07:56"}
+{"current_steps": 53, "total_steps": 213, "loss": 0.4706975519657135, "lr": 9.40353427008083e-06, "epoch": 0.7537777777777778, "percentage": 24.88, "elapsed_time": "0:22:20", "remaining_time": "1:07:25"}
+{"current_steps": 54, "total_steps": 213, "loss": 0.4405261278152466, "lr": 9.3639861251539e-06, "epoch": 0.768, "percentage": 25.35, "elapsed_time": "0:22:44", "remaining_time": "1:06:57"}
+{"current_steps": 55, "total_steps": 213, "loss": 0.47680750489234924, "lr": 9.323257369891702e-06, "epoch": 0.7822222222222223, "percentage": 25.82, "elapsed_time": "0:23:09", "remaining_time": "1:06:31"}
+{"current_steps": 56, "total_steps": 213, "loss": 0.42392900586128235, "lr": 9.281359022841966e-06, "epoch": 0.7964444444444444, "percentage": 26.29, "elapsed_time": "0:23:33", "remaining_time": "1:06:02"}
+{"current_steps": 57, "total_steps": 213, "loss": 0.48122283816337585, "lr": 9.238302418967757e-06, "epoch": 0.8106666666666666, "percentage": 26.76, "elapsed_time": "0:23:57", "remaining_time": "1:05:33"}
+{"current_steps": 58, "total_steps": 213, "loss": 0.5255264043807983, "lr": 9.194099206580981e-06, "epoch": 0.8248888888888889, "percentage": 27.23, "elapsed_time": "0:24:21", "remaining_time": "1:05:05"}
+{"current_steps": 59, "total_steps": 213, "loss": 0.4682884216308594, "lr": 9.14876134419111e-06, "epoch": 0.8391111111111111, "percentage": 27.7, "elapsed_time": "0:24:45", "remaining_time": "1:04:36"}
+{"current_steps": 60, "total_steps": 213, "loss": 0.5003230571746826, "lr": 9.102301097269974e-06, "epoch": 0.8533333333333334, "percentage": 28.17, "elapsed_time": "0:25:09", "remaining_time": "1:04:08"}
+{"current_steps": 61, "total_steps": 213, "loss": 0.45744988322257996, "lr": 9.05473103493355e-06, "epoch": 0.8675555555555555, "percentage": 28.64, "elapsed_time": "0:25:32", "remaining_time": "1:03:39"}
+{"current_steps": 62, "total_steps": 213, "loss": 0.5037047863006592, "lr": 9.006064026541549e-06, "epoch": 0.8817777777777778, "percentage": 29.11, "elapsed_time": "0:25:56", "remaining_time": "1:03:11"}
+{"current_steps": 63, "total_steps": 213, "loss": 0.4382043778896332, "lr": 8.956313238215824e-06, "epoch": 0.896, "percentage": 29.58, "elapsed_time": "0:26:20", "remaining_time": "1:02:42"}
+{"current_steps": 64, "total_steps": 213, "loss": 0.44455963373184204, "lr": 8.905492129278478e-06, "epoch": 0.9102222222222223, "percentage": 30.05, "elapsed_time": "0:26:43", "remaining_time": "1:02:14"}
+{"current_steps": 65, "total_steps": 213, "loss": 0.4215100109577179, "lr": 8.85361444861063e-06, "epoch": 0.9244444444444444, "percentage": 30.52, "elapsed_time": "0:27:08", "remaining_time": "1:01:47"}
+{"current_steps": 66, "total_steps": 213, "loss": 0.48813939094543457, "lr": 8.800694230932885e-06, "epoch": 0.9386666666666666, "percentage": 30.99, "elapsed_time": "0:27:32", "remaining_time": "1:01:19"}
+{"current_steps": 67, "total_steps": 213, "loss": 3.263723850250244, "lr": 8.74674579300843e-06, "epoch": 0.9528888888888889, "percentage": 31.46, "elapsed_time": "0:27:56", "remaining_time": "1:00:52"}
+{"current_steps": 68, "total_steps": 213, "loss": 2.8557608127593994, "lr": 8.691783729769874e-06, "epoch": 0.9671111111111111, "percentage": 31.92, "elapsed_time": "0:28:19", "remaining_time": "1:00:24"}
+{"current_steps": 69, "total_steps": 213, "loss": 3.485170602798462, "lr": 8.635822910370793e-06, "epoch": 0.9813333333333333, "percentage": 32.39, "elapsed_time": "0:28:43", "remaining_time": "0:59:57"}
+{"current_steps": 70, "total_steps": 213, "loss": 3.1958985328674316, "lr": 8.578878474163115e-06, "epoch": 0.9955555555555555, "percentage": 32.86, "elapsed_time": "0:29:07", "remaining_time": "0:59:30"}
+{"current_steps": 71, "total_steps": 213, "loss": 2.678990364074707, "lr": 8.520965826601394e-06, "epoch": 1.0, "percentage": 33.33, "elapsed_time": "0:29:15", "remaining_time": "0:58:30"}
+{"current_steps": 72, "total_steps": 213, "loss": 3.4466443061828613, "lr": 8.462100635075097e-06, "epoch": 1.0142222222222221, "percentage": 33.8, "elapsed_time": "0:32:38", "remaining_time": "1:03:55"}
+{"current_steps": 73, "total_steps": 213, "loss": 3.4718263149261475, "lr": 8.40229882467003e-06, "epoch": 1.0284444444444445, "percentage": 34.27, "elapsed_time": "0:33:02", "remaining_time": "1:03:21"}
+{"current_steps": 74, "total_steps": 213, "loss": 3.6029202938079834, "lr": 8.341576573860049e-06, "epoch": 1.0426666666666666, "percentage": 34.74, "elapsed_time": "0:33:26", "remaining_time": "1:02:48"}
+{"current_steps": 75, "total_steps": 213, "loss": 4.018604755401611, "lr": 8.279950310130218e-06, "epoch": 1.056888888888889, "percentage": 35.21, "elapsed_time": "0:33:49", "remaining_time": "1:02:14"}
+{"current_steps": 76, "total_steps": 213, "loss": 3.0605146884918213, "lr": 8.2174367055326e-06, "epoch": 1.0711111111111111, "percentage": 35.68, "elapsed_time": "0:34:13", "remaining_time": "1:01:41"}
+{"current_steps": 77, "total_steps": 213, "loss": 3.556879997253418, "lr": 8.154052672175888e-06, "epoch": 1.0853333333333333, "percentage": 36.15, "elapsed_time": "0:34:37", "remaining_time": "1:01:09"}
+{"current_steps": 78, "total_steps": 213, "loss": 3.7346572875976562, "lr": 8.08981535765009e-06, "epoch": 1.0995555555555556, "percentage": 36.62, "elapsed_time": "0:35:01", "remaining_time": "1:00:37"}
+{"current_steps": 79, "total_steps": 213, "loss": 3.2394087314605713, "lr": 8.024742140387506e-06, "epoch": 1.1137777777777778, "percentage": 37.09, "elapsed_time": "0:35:25", "remaining_time": "1:00:05"}
+{"current_steps": 80, "total_steps": 213, "loss": 3.625601053237915, "lr": 7.95885062496126e-06, "epoch": 1.1280000000000001, "percentage": 37.56, "elapsed_time": "0:35:49", "remaining_time": "0:59:32"}
+{"current_steps": 81, "total_steps": 213, "loss": 3.3379783630371094, "lr": 7.892158637322647e-06, "epoch": 1.1422222222222222, "percentage": 38.03, "elapsed_time": "0:36:12", "remaining_time": "0:59:01"}
+{"current_steps": 82, "total_steps": 213, "loss": 3.502805471420288, "lr": 7.824684219978591e-06, "epoch": 1.1564444444444444, "percentage": 38.5, "elapsed_time": "0:36:36", "remaining_time": "0:58:29"}
+{"current_steps": 83, "total_steps": 213, "loss": 3.362370729446411, "lr": 7.756445627110523e-06, "epoch": 1.1706666666666667, "percentage": 38.97, "elapsed_time": "0:37:00", "remaining_time": "0:57:58"}
+{"current_steps": 84, "total_steps": 213, "loss": 3.1864588260650635, "lr": 7.68746131963598e-06, "epoch": 1.1848888888888889, "percentage": 39.44, "elapsed_time": "0:37:24", "remaining_time": "0:57:27"}
+{"current_steps": 85, "total_steps": 213, "loss": 3.4393932819366455, "lr": 7.6177499602143e-06, "epoch": 1.199111111111111, "percentage": 39.91, "elapsed_time": "0:37:48", "remaining_time": "0:56:56"}
+{"current_steps": 86, "total_steps": 213, "loss": 3.76263689994812, "lr": 7.547330408197695e-06, "epoch": 1.2133333333333334, "percentage": 40.38, "elapsed_time": "0:38:12", "remaining_time": "0:56:25"}
+{"current_steps": 87, "total_steps": 213, "loss": 3.3016631603240967, "lr": 7.476221714529167e-06, "epoch": 1.2275555555555555, "percentage": 40.85, "elapsed_time": "0:38:36", "remaining_time": "0:55:55"}
+{"current_steps": 88, "total_steps": 213, "loss": 2.8077316284179688, "lr": 7.404443116588548e-06, "epoch": 1.2417777777777779, "percentage": 41.31, "elapsed_time": "0:39:00", "remaining_time": "0:55:24"}
+{"current_steps": 89, "total_steps": 213, "loss": 3.496107816696167, "lr": 7.332014032988123e-06, "epoch": 1.256, "percentage": 41.78, "elapsed_time": "0:39:24", "remaining_time": "0:54:54"}
+{"current_steps": 90, "total_steps": 213, "loss": 3.4046881198883057, "lr": 7.2589540583192165e-06, "epoch": 1.2702222222222221, "percentage": 42.25, "elapsed_time": "0:39:48", "remaining_time": "0:54:23"}
+{"current_steps": 91, "total_steps": 213, "loss": 3.3288354873657227, "lr": 7.185282957851175e-06, "epoch": 1.2844444444444445, "percentage": 42.72, "elapsed_time": "0:40:12", "remaining_time": "0:53:53"}
+{"current_steps": 92, "total_steps": 213, "loss": 3.573462963104248, "lr": 7.111020662184174e-06, "epoch": 1.2986666666666666, "percentage": 43.19, "elapsed_time": "0:40:36", "remaining_time": "0:53:24"}
+{"current_steps": 93, "total_steps": 213, "loss": 3.4570071697235107, "lr": 7.036187261857289e-06, "epoch": 1.3128888888888888, "percentage": 43.66, "elapsed_time": "0:41:00", "remaining_time": "0:52:54"}
+{"current_steps": 94, "total_steps": 213, "loss": 3.511754274368286, "lr": 6.960803001913315e-06, "epoch": 1.3271111111111111, "percentage": 44.13, "elapsed_time": "0:41:24", "remaining_time": "0:52:24"}
+{"current_steps": 95, "total_steps": 213, "loss": 4.017252445220947, "lr": 6.884888276421766e-06, "epoch": 1.3413333333333333, "percentage": 44.6, "elapsed_time": "0:41:48", "remaining_time": "0:51:55"}
+{"current_steps": 96, "total_steps": 213, "loss": 3.5670371055603027, "lr": 6.8084636229615786e-06, "epoch": 1.3555555555555556, "percentage": 45.07, "elapsed_time": "0:42:11", "remaining_time": "0:51:25"}
+{"current_steps": 97, "total_steps": 213, "loss": 4.055681228637695, "lr": 6.731549717064975e-06, "epoch": 1.3697777777777778, "percentage": 45.54, "elapsed_time": "0:42:35", "remaining_time": "0:50:56"}
+{"current_steps": 98, "total_steps": 213, "loss": 3.303616762161255, "lr": 6.654167366624009e-06, "epoch": 1.384, "percentage": 46.01, "elapsed_time": "0:42:59", "remaining_time": "0:50:27"}
+{"current_steps": 99, "total_steps": 213, "loss": 2.8152995109558105, "lr": 6.576337506261314e-06, "epoch": 1.3982222222222223, "percentage": 46.48, "elapsed_time": "0:43:23", "remaining_time": "0:49:57"}
+{"current_steps": 100, "total_steps": 213, "loss": 3.3619225025177, "lr": 6.498081191666549e-06, "epoch": 1.4124444444444444, "percentage": 46.95, "elapsed_time": "0:43:46", "remaining_time": "0:49:28"}
+{"current_steps": 100, "total_steps": 213, "eval_loss": 3.694061279296875, "epoch": 1.4124444444444444, "percentage": 46.95, "elapsed_time": "0:43:52", "remaining_time": "0:49:35"}
+{"current_steps": 101, "total_steps": 213, "loss": 3.0373668670654297, "lr": 6.419419593900109e-06, "epoch": 1.4266666666666667, "percentage": 47.42, "elapsed_time": "0:44:16", "remaining_time": "0:49:05"}
+{"current_steps": 102, "total_steps": 213, "loss": 3.642394542694092, "lr": 6.340373993665607e-06, "epoch": 1.4408888888888889, "percentage": 47.89, "elapsed_time": "0:44:40", "remaining_time": "0:48:36"}
+{"current_steps": 103, "total_steps": 213, "loss": 3.3127541542053223, "lr": 6.2609657755527135e-06, "epoch": 1.455111111111111, "percentage": 48.36, "elapsed_time": "0:45:04", "remaining_time": "0:48:07"}
+{"current_steps": 104, "total_steps": 213, "loss": 3.500901937484741, "lr": 6.1812164222518626e-06, "epoch": 1.4693333333333334, "percentage": 48.83, "elapsed_time": "0:45:28", "remaining_time": "0:47:39"}
+{"current_steps": 105, "total_steps": 213, "loss": 3.4911036491394043, "lr": 6.101147508742456e-06, "epoch": 1.4835555555555555, "percentage": 49.3, "elapsed_time": "0:45:51", "remaining_time": "0:47:10"}
+{"current_steps": 106, "total_steps": 213, "loss": 3.375718116760254, "lr": 6.020780696456059e-06, "epoch": 1.4977777777777779, "percentage": 49.77, "elapsed_time": "0:46:15", "remaining_time": "0:46:42"}
+{"current_steps": 107, "total_steps": 213, "loss": 3.340306282043457, "lr": 5.940137727416247e-06, "epoch": 1.512, "percentage": 50.23, "elapsed_time": "0:46:39", "remaining_time": "0:46:13"}
+{"current_steps": 108, "total_steps": 213, "loss": 3.5164387226104736, "lr": 5.859240418356614e-06, "epoch": 1.5262222222222221, "percentage": 50.7, "elapsed_time": "0:47:03", "remaining_time": "0:45:45"}
+{"current_steps": 109, "total_steps": 213, "loss": 3.955772876739502, "lr": 5.778110654818602e-06, "epoch": 1.5404444444444443, "percentage": 51.17, "elapsed_time": "0:47:27", "remaining_time": "0:45:16"}
+{"current_steps": 110, "total_steps": 213, "loss": 3.425647020339966, "lr": 5.696770385230679e-06, "epoch": 1.5546666666666666, "percentage": 51.64, "elapsed_time": "0:47:50", "remaining_time": "0:44:48"}
+{"current_steps": 111, "total_steps": 213, "loss": 3.219924211502075, "lr": 5.615241614970546e-06, "epoch": 1.568888888888889, "percentage": 52.11, "elapsed_time": "0:48:15", "remaining_time": "0:44:20"}
+{"current_steps": 112, "total_steps": 213, "loss": 3.3812978267669678, "lr": 5.533546400411899e-06, "epoch": 1.5831111111111111, "percentage": 52.58, "elapsed_time": "0:48:39", "remaining_time": "0:43:52"}
+{"current_steps": 113, "total_steps": 213, "loss": 3.7506139278411865, "lr": 5.4517068429574215e-06, "epoch": 1.5973333333333333, "percentage": 53.05, "elapsed_time": "0:49:03", "remaining_time": "0:43:24"}
+{"current_steps": 114, "total_steps": 213, "loss": 3.259803295135498, "lr": 5.3697450830595775e-06, "epoch": 1.6115555555555554, "percentage": 53.52, "elapsed_time": "0:49:27", "remaining_time": "0:42:56"}
+{"current_steps": 115, "total_steps": 213, "loss": 2.8727362155914307, "lr": 5.287683294230855e-06, "epoch": 1.6257777777777778, "percentage": 53.99, "elapsed_time": "0:49:50", "remaining_time": "0:42:28"}
+{"current_steps": 116, "total_steps": 213, "loss": 3.628926992416382, "lr": 5.20554367704505e-06, "epoch": 1.6400000000000001, "percentage": 54.46, "elapsed_time": "0:50:14", "remaining_time": "0:42:00"}
+{"current_steps": 117, "total_steps": 213, "loss": 3.236464023590088, "lr": 5.123348453131242e-06, "epoch": 1.6542222222222223, "percentage": 54.93, "elapsed_time": "0:50:38", "remaining_time": "0:41:32"}
+{"current_steps": 118, "total_steps": 213, "loss": 2.888354778289795, "lr": 5.041119859162068e-06, "epoch": 1.6684444444444444, "percentage": 55.4, "elapsed_time": "0:51:02", "remaining_time": "0:41:05"}
+{"current_steps": 119, "total_steps": 213, "loss": 3.105433702468872, "lr": 4.958880140837934e-06, "epoch": 1.6826666666666665, "percentage": 55.87, "elapsed_time": "0:51:25", "remaining_time": "0:40:37"}
+{"current_steps": 120, "total_steps": 213, "loss": 3.2962746620178223, "lr": 4.876651546868759e-06, "epoch": 1.696888888888889, "percentage": 56.34, "elapsed_time": "0:51:49", "remaining_time": "0:40:09"}
+{"current_steps": 121, "total_steps": 213, "loss": 3.451850414276123, "lr": 4.794456322954953e-06, "epoch": 1.7111111111111112, "percentage": 56.81, "elapsed_time": "0:52:13", "remaining_time": "0:39:42"}
+{"current_steps": 122, "total_steps": 213, "loss": 3.6925525665283203, "lr": 4.7123167057691446e-06, "epoch": 1.7253333333333334, "percentage": 57.28, "elapsed_time": "0:52:37", "remaining_time": "0:39:14"}
+{"current_steps": 123, "total_steps": 213, "loss": 4.191713333129883, "lr": 4.630254916940424e-06, "epoch": 1.7395555555555555, "percentage": 57.75, "elapsed_time": "0:53:00", "remaining_time": "0:38:47"}
+{"current_steps": 124, "total_steps": 213, "loss": 2.8300790786743164, "lr": 4.548293157042581e-06, "epoch": 1.7537777777777777, "percentage": 58.22, "elapsed_time": "0:53:24", "remaining_time": "0:38:20"}
+{"current_steps": 125, "total_steps": 213, "loss": 3.5787062644958496, "lr": 4.466453599588103e-06, "epoch": 1.768, "percentage": 58.69, "elapsed_time": "0:53:48", "remaining_time": "0:37:52"}
+{"current_steps": 126, "total_steps": 213, "loss": 2.687720537185669, "lr": 4.384758385029457e-06, "epoch": 1.7822222222222224, "percentage": 59.15, "elapsed_time": "0:54:12", "remaining_time": "0:37:25"}
+{"current_steps": 127, "total_steps": 213, "loss": 3.221151113510132, "lr": 4.3032296147693225e-06, "epoch": 1.7964444444444445, "percentage": 59.62, "elapsed_time": "0:54:36", "remaining_time": "0:36:58"}
+{"current_steps": 128, "total_steps": 213, "loss": 3.415364980697632, "lr": 4.2218893451814005e-06, "epoch": 1.8106666666666666, "percentage": 60.09, "elapsed_time": "0:55:00", "remaining_time": "0:36:31"}
+{"current_steps": 129, "total_steps": 213, "loss": 3.2137668132781982, "lr": 4.140759581643386e-06, "epoch": 1.8248888888888888, "percentage": 60.56, "elapsed_time": "0:55:24", "remaining_time": "0:36:04"}
+{"current_steps": 130, "total_steps": 213, "loss": 2.874941349029541, "lr": 4.059862272583755e-06, "epoch": 1.8391111111111111, "percentage": 61.03, "elapsed_time": "0:55:47", "remaining_time": "0:35:37"}
+{"current_steps": 131, "total_steps": 213, "loss": 3.6437442302703857, "lr": 3.979219303543942e-06, "epoch": 1.8533333333333335, "percentage": 61.5, "elapsed_time": "0:56:11", "remaining_time": "0:35:10"}
+{"current_steps": 132, "total_steps": 213, "loss": 3.262843370437622, "lr": 3.898852491257547e-06, "epoch": 1.8675555555555556, "percentage": 61.97, "elapsed_time": "0:56:35", "remaining_time": "0:34:43"}
+{"current_steps": 133, "total_steps": 213, "loss": 2.763110637664795, "lr": 3.818783577748138e-06, "epoch": 1.8817777777777778, "percentage": 62.44, "elapsed_time": "0:56:59", "remaining_time": "0:34:16"}
+{"current_steps": 134, "total_steps": 213, "loss": 2.978278160095215, "lr": 3.7390342244472886e-06, "epoch": 1.896, "percentage": 62.91, "elapsed_time": "0:57:23", "remaining_time": "0:33:49"}
+{"current_steps": 135, "total_steps": 213, "loss": 2.9288432598114014, "lr": 3.659626006334395e-06, "epoch": 1.9102222222222223, "percentage": 63.38, "elapsed_time": "0:57:46", "remaining_time": "0:33:23"}
+{"current_steps": 136, "total_steps": 213, "loss": 3.158463954925537, "lr": 3.5805804060998926e-06, "epoch": 1.9244444444444444, "percentage": 63.85, "elapsed_time": "0:58:10", "remaining_time": "0:32:56"}
+{"current_steps": 137, "total_steps": 213, "loss": 2.699950933456421, "lr": 3.501918808333453e-06, "epoch": 1.9386666666666668, "percentage": 64.32, "elapsed_time": "0:58:34", "remaining_time": "0:32:29"}
+{"current_steps": 138, "total_steps": 213, "loss": 3.840054512023926, "lr": 3.4236624937386874e-06, "epoch": 1.952888888888889, "percentage": 64.79, "elapsed_time": "0:58:58", "remaining_time": "0:32:03"}
+{"current_steps": 139, "total_steps": 213, "loss": 2.972839832305908, "lr": 3.3458326333759927e-06, "epoch": 1.967111111111111, "percentage": 65.26, "elapsed_time": "0:59:22", "remaining_time": "0:31:36"}
+{"current_steps": 140, "total_steps": 213, "loss": 2.8588995933532715, "lr": 3.268450282935026e-06, "epoch": 1.9813333333333332, "percentage": 65.73, "elapsed_time": "0:59:46", "remaining_time": "0:31:09"}
+{"current_steps": 141, "total_steps": 213, "loss": 3.6230926513671875, "lr": 3.1915363770384223e-06, "epoch": 1.9955555555555555, "percentage": 66.2, "elapsed_time": "1:00:09", "remaining_time": "0:30:43"}
+{"current_steps": 142, "total_steps": 213, "loss": 2.1066107749938965, "lr": 3.1151117235782346e-06, "epoch": 2.0, "percentage": 66.67, "elapsed_time": "1:00:17", "remaining_time": "0:30:08"}
+{"current_steps": 143, "total_steps": 213, "loss": 3.128911256790161, "lr": 3.0391969980866874e-06, "epoch": 2.014222222222222, "percentage": 67.14, "elapsed_time": "1:03:44", "remaining_time": "0:31:12"}
+{"current_steps": 144, "total_steps": 213, "loss": 3.2379772663116455, "lr": 2.963812738142713e-06, "epoch": 2.0284444444444443, "percentage": 67.61, "elapsed_time": "1:04:09", "remaining_time": "0:30:44"}
+{"current_steps": 145, "total_steps": 213, "loss": 3.2959132194519043, "lr": 2.8889793378158284e-06, "epoch": 2.042666666666667, "percentage": 68.08, "elapsed_time": "1:04:32", "remaining_time": "0:30:16"}
+{"current_steps": 146, "total_steps": 213, "loss": 3.3406944274902344, "lr": 2.814717042148827e-06, "epoch": 2.056888888888889, "percentage": 68.54, "elapsed_time": "1:04:56", "remaining_time": "0:29:48"}
+{"current_steps": 147, "total_steps": 213, "loss": 3.3720455169677734, "lr": 2.7410459416807856e-06, "epoch": 2.071111111111111, "percentage": 69.01, "elapsed_time": "1:05:20", "remaining_time": "0:29:20"}
+{"current_steps": 148, "total_steps": 213, "loss": 3.153254508972168, "lr": 2.6679859670118785e-06, "epoch": 2.0853333333333333, "percentage": 69.48, "elapsed_time": "1:05:44", "remaining_time": "0:28:52"}
+{"current_steps": 149, "total_steps": 213, "loss": 2.9255833625793457, "lr": 2.5955568834114523e-06, "epoch": 2.0995555555555554, "percentage": 69.95, "elapsed_time": "1:06:08", "remaining_time": "0:28:24"}
+{"current_steps": 150, "total_steps": 213, "loss": 3.385373115539551, "lr": 2.523778285470835e-06, "epoch": 2.113777777777778, "percentage": 70.42, "elapsed_time": "1:06:31", "remaining_time": "0:27:56"}
+{"current_steps": 150, "total_steps": 213, "eval_loss": 3.694100856781006, "epoch": 2.113777777777778, "percentage": 70.42, "elapsed_time": "1:06:38", "remaining_time": "0:27:59"}
+{"current_steps": 151, "total_steps": 213, "loss": 2.953336238861084, "lr": 2.452669591802307e-06, "epoch": 2.128, "percentage": 70.89, "elapsed_time": "1:07:01", "remaining_time": "0:27:31"}
+{"current_steps": 152, "total_steps": 213, "loss": 3.4797470569610596, "lr": 2.3822500397857016e-06, "epoch": 2.1422222222222222, "percentage": 71.36, "elapsed_time": "1:07:25", "remaining_time": "0:27:03"}
+{"current_steps": 153, "total_steps": 213, "loss": 3.766719341278076, "lr": 2.3125386803640188e-06, "epoch": 2.1564444444444444, "percentage": 71.83, "elapsed_time": "1:07:49", "remaining_time": "0:26:35"}
+{"current_steps": 154, "total_steps": 213, "loss": 3.105588436126709, "lr": 2.243554372889479e-06, "epoch": 2.1706666666666665, "percentage": 72.3, "elapsed_time": "1:08:13", "remaining_time": "0:26:08"}
+{"current_steps": 155, "total_steps": 213, "loss": 3.7258148193359375, "lr": 2.1753157800214107e-06, "epoch": 2.1848888888888887, "percentage": 72.77, "elapsed_time": "1:08:36", "remaining_time": "0:25:40"}
+{"current_steps": 156, "total_steps": 213, "loss": 3.2563636302948, "lr": 2.1078413626773547e-06, "epoch": 2.1991111111111112, "percentage": 73.24, "elapsed_time": "1:09:00", "remaining_time": "0:25:12"}
+{"current_steps": 157, "total_steps": 213, "loss": 3.6285147666931152, "lr": 2.0411493750387423e-06, "epoch": 2.2133333333333334, "percentage": 73.71, "elapsed_time": "1:09:24", "remaining_time": "0:24:45"}
+{"current_steps": 158, "total_steps": 213, "loss": 3.4809606075286865, "lr": 1.9752578596124955e-06, "epoch": 2.2275555555555555, "percentage": 74.18, "elapsed_time": "1:09:48", "remaining_time": "0:24:17"}
+{"current_steps": 159, "total_steps": 213, "loss": 3.331500768661499, "lr": 1.9101846423499113e-06, "epoch": 2.2417777777777776, "percentage": 74.65, "elapsed_time": "1:10:11", "remaining_time": "0:23:50"}
+{"current_steps": 160, "total_steps": 213, "loss": 3.103165626525879, "lr": 1.8459473278241125e-06, "epoch": 2.2560000000000002, "percentage": 75.12, "elapsed_time": "1:10:35", "remaining_time": "0:23:23"}
+{"current_steps": 161, "total_steps": 213, "loss": 2.949009656906128, "lr": 1.7825632944674016e-06, "epoch": 2.2702222222222224, "percentage": 75.59, "elapsed_time": "1:10:59", "remaining_time": "0:22:55"}
+{"current_steps": 162, "total_steps": 213, "loss": 3.8803815841674805, "lr": 1.7200496898697832e-06, "epoch": 2.2844444444444445, "percentage": 76.06, "elapsed_time": "1:11:23", "remaining_time": "0:22:28"}
+{"current_steps": 163, "total_steps": 213, "loss": 3.3277387619018555, "lr": 1.6584234261399535e-06, "epoch": 2.2986666666666666, "percentage": 76.53, "elapsed_time": "1:11:47", "remaining_time": "0:22:01"}
+{"current_steps": 164, "total_steps": 213, "loss": 3.04113507270813, "lr": 1.5977011753299726e-06, "epoch": 2.3128888888888888, "percentage": 77.0, "elapsed_time": "1:12:10", "remaining_time": "0:21:34"}
+{"current_steps": 165, "total_steps": 213, "loss": 3.1088926792144775, "lr": 1.5378993649249053e-06, "epoch": 2.327111111111111, "percentage": 77.46, "elapsed_time": "1:12:34", "remaining_time": "0:21:06"}
+{"current_steps": 166, "total_steps": 213, "loss": 2.9704766273498535, "lr": 1.4790341733986085e-06, "epoch": 2.3413333333333335, "percentage": 77.93, "elapsed_time": "1:12:58", "remaining_time": "0:20:39"}
+{"current_steps": 167, "total_steps": 213, "loss": 2.666156053543091, "lr": 1.4211215258368866e-06, "epoch": 2.3555555555555556, "percentage": 78.4, "elapsed_time": "1:13:22", "remaining_time": "0:20:12"}
+{"current_steps": 168, "total_steps": 213, "loss": 3.5968339443206787, "lr": 1.3641770896292083e-06, "epoch": 2.3697777777777778, "percentage": 78.87, "elapsed_time": "1:13:46", "remaining_time": "0:19:45"}
+{"current_steps": 169, "total_steps": 213, "loss": 3.576735496520996, "lr": 1.3082162702301276e-06, "epoch": 2.384, "percentage": 79.34, "elapsed_time": "1:14:10", "remaining_time": "0:19:18"}
+{"current_steps": 170, "total_steps": 213, "loss": 3.0840680599212646, "lr": 1.2532542069915722e-06, "epoch": 2.398222222222222, "percentage": 79.81, "elapsed_time": "1:14:34", "remaining_time": "0:18:51"}
+{"current_steps": 171, "total_steps": 213, "loss": 3.7334625720977783, "lr": 1.1993057690671174e-06, "epoch": 2.4124444444444446, "percentage": 80.28, "elapsed_time": "1:14:58", "remaining_time": "0:18:24"}
+{"current_steps": 172, "total_steps": 213, "loss": 4.4254841804504395, "lr": 1.1463855513893695e-06, "epoch": 2.4266666666666667, "percentage": 80.75, "elapsed_time": "1:15:23", "remaining_time": "0:17:58"}
+{"current_steps": 173, "total_steps": 213, "loss": 3.775759696960449, "lr": 1.0945078707215224e-06, "epoch": 2.440888888888889, "percentage": 81.22, "elapsed_time": "1:15:46", "remaining_time": "0:17:31"}
+{"current_steps": 174, "total_steps": 213, "loss": 2.9598851203918457, "lr": 1.0436867617841768e-06, "epoch": 2.455111111111111, "percentage": 81.69, "elapsed_time": "1:16:10", "remaining_time": "0:17:04"}
+{"current_steps": 175, "total_steps": 213, "loss": 2.9934566020965576, "lr": 9.939359734584552e-07, "epoch": 2.469333333333333, "percentage": 82.16, "elapsed_time": "1:16:34", "remaining_time": "0:16:37"}
+{"current_steps": 176, "total_steps": 213, "loss": 3.3066673278808594, "lr": 9.452689650664515e-07, "epoch": 2.4835555555555557, "percentage": 82.63, "elapsed_time": "1:16:58", "remaining_time": "0:16:10"}
+{"current_steps": 177, "total_steps": 213, "loss": 2.8772518634796143, "lr": 8.976989027300265e-07, "epoch": 2.497777777777778, "percentage": 83.1, "elapsed_time": "1:17:22", "remaining_time": "0:15:44"}
+{"current_steps": 178, "total_steps": 213, "loss": 0.2714032232761383, "lr": 8.512386558088919e-07, "epoch": 2.512, "percentage": 83.57, "elapsed_time": "1:17:45", "remaining_time": "0:15:17"}
+{"current_steps": 179, "total_steps": 213, "loss": 0.0, "lr": 8.059007934190194e-07, "epoch": 2.526222222222222, "percentage": 84.04, "elapsed_time": "1:18:09", "remaining_time": "0:14:50"}
+{"current_steps": 180, "total_steps": 213, "loss": 0.0, "lr": 7.61697581032243e-07, "epoch": 2.5404444444444443, "percentage": 84.51, "elapsed_time": "1:18:33", "remaining_time": "0:14:24"}
+{"current_steps": 181, "total_steps": 213, "loss": 0.0, "lr": 7.186409771580355e-07, "epoch": 2.554666666666667, "percentage": 84.98, "elapsed_time": "1:18:57", "remaining_time": "0:13:57"}
+{"current_steps": 182, "total_steps": 213, "loss": 0.0, "lr": 6.76742630108298e-07, "epoch": 2.568888888888889, "percentage": 85.45, "elapsed_time": "1:19:21", "remaining_time": "0:13:30"}
+{"current_steps": 183, "total_steps": 213, "loss": 0.0, "lr": 6.360138748461015e-07, "epoch": 2.583111111111111, "percentage": 85.92, "elapsed_time": "1:19:44", "remaining_time": "0:13:04"}
+{"current_steps": 184, "total_steps": 213, "loss": 0.0, "lr": 5.964657299191712e-07, "epoch": 2.5973333333333333, "percentage": 86.38, "elapsed_time": "1:20:08", "remaining_time": "0:12:37"}
+{"current_steps": 185, "total_steps": 213, "loss": 0.0, "lr": 5.581088944789953e-07, "epoch": 2.6115555555555554, "percentage": 86.85, "elapsed_time": "1:20:32", "remaining_time": "0:12:11"}
+{"current_steps": 186, "total_steps": 213, "loss": 0.0, "lr": 5.20953745386329e-07, "epoch": 2.6257777777777775, "percentage": 87.32, "elapsed_time": "1:20:56", "remaining_time": "0:11:44"}
+{"current_steps": 187, "total_steps": 213, "loss": 0.0, "lr": 4.850103344038853e-07, "epoch": 2.64, "percentage": 87.79, "elapsed_time": "1:21:20", "remaining_time": "0:11:18"}
+{"current_steps": 188, "total_steps": 213, "loss": 0.0, "lr": 4.502883854769935e-07, "epoch": 2.6542222222222223, "percentage": 88.26, "elapsed_time": "1:21:43", "remaining_time": "0:10:52"}
+{"current_steps": 189, "total_steps": 213, "loss": 0.0, "lr": 4.167972921029262e-07, "epoch": 2.6684444444444444, "percentage": 88.73, "elapsed_time": "1:22:07", "remaining_time": "0:10:25"}
+{"current_steps": 190, "total_steps": 213, "loss": 0.0, "lr": 3.8454611478963235e-07, "epoch": 2.6826666666666665, "percentage": 89.2, "elapsed_time": "1:22:31", "remaining_time": "0:09:59"}
+{"current_steps": 191, "total_steps": 213, "loss": 0.0, "lr": 3.535435786045538e-07, "epoch": 2.696888888888889, "percentage": 89.67, "elapsed_time": "1:22:55", "remaining_time": "0:09:33"}
+{"current_steps": 192, "total_steps": 213, "loss": 0.0, "lr": 3.237980708141919e-07, "epoch": 2.7111111111111112, "percentage": 90.14, "elapsed_time": "1:23:18", "remaining_time": "0:09:06"}
+{"current_steps": 193, "total_steps": 213, "loss": 0.0, "lr": 2.9531763861505967e-07, "epoch": 2.7253333333333334, "percentage": 90.61, "elapsed_time": "1:23:42", "remaining_time": "0:08:40"}
+{"current_steps": 194, "total_steps": 213, "loss": 0.0, "lr": 2.681099869566328e-07, "epoch": 2.7395555555555555, "percentage": 91.08, "elapsed_time": "1:24:06", "remaining_time": "0:08:14"}
+{"current_steps": 195, "total_steps": 213, "loss": 0.0, "lr": 2.4218247645689306e-07, "epoch": 2.7537777777777777, "percentage": 91.55, "elapsed_time": "1:24:30", "remaining_time": "0:07:48"}
+{"current_steps": 196, "total_steps": 213, "loss": 0.0, "lr": 2.175421214110235e-07, "epoch": 2.768, "percentage": 92.02, "elapsed_time": "1:24:53", "remaining_time": "0:07:21"}
+{"current_steps": 197, "total_steps": 213, "loss": 0.0, "lr": 1.941955878938029e-07, "epoch": 2.7822222222222224, "percentage": 92.49, "elapsed_time": "1:25:17", "remaining_time": "0:06:55"}
+{"current_steps": 198, "total_steps": 213, "loss": 0.0, "lr": 1.721491919561913e-07, "epoch": 2.7964444444444445, "percentage": 92.96, "elapsed_time": "1:25:40", "remaining_time": "0:06:29"}
+{"current_steps": 199, "total_steps": 213, "loss": 0.0, "lr": 1.514088979166256e-07, "epoch": 2.8106666666666666, "percentage": 93.43, "elapsed_time": "1:26:04", "remaining_time": "0:06:03"}
+{"current_steps": 200, "total_steps": 213, "loss": 0.0, "lr": 1.3198031674745814e-07, "epoch": 2.824888888888889, "percentage": 93.9, "elapsed_time": "1:26:28", "remaining_time": "0:05:37"}
+{"current_steps": 200, "total_steps": 213, "eval_loss": NaN, "epoch": 2.824888888888889, "percentage": 93.9, "elapsed_time": "1:26:34", "remaining_time": "0:05:37"}
+{"current_steps": 201, "total_steps": 213, "loss": 0.0, "lr": 1.138687045569975e-07, "epoch": 2.8391111111111114, "percentage": 94.37, "elapsed_time": "1:26:57", "remaining_time": "0:05:11"}
+{"current_steps": 202, "total_steps": 213, "loss": 0.0, "lr": 9.70789611675449e-08, "epoch": 2.8533333333333335, "percentage": 94.84, "elapsed_time": "1:27:21", "remaining_time": "0:04:45"}
+{"current_steps": 203, "total_steps": 213, "loss": 0.0, "lr": 8.161562878982399e-08, "epoch": 2.8675555555555556, "percentage": 95.31, "elapsed_time": "1:27:45", "remaining_time": "0:04:19"}
+{"current_steps": 204, "total_steps": 213, "loss": 0.0, "lr": 6.74828907941516e-08, "epoch": 2.8817777777777778, "percentage": 95.77, "elapsed_time": "1:28:09", "remaining_time": "0:03:53"}
+{"current_steps": 205, "total_steps": 213, "loss": 0.0, "lr": 5.468457057869358e-08, "epoch": 2.896, "percentage": 96.24, "elapsed_time": "1:28:32", "remaining_time": "0:03:27"}
+{"current_steps": 206, "total_steps": 213, "loss": 0.0, "lr": 4.322413053509944e-08, "epoch": 2.910222222222222, "percentage": 96.71, "elapsed_time": "1:28:56", "remaining_time": "0:03:01"}
+{"current_steps": 207, "total_steps": 213, "loss": 0.0, "lr": 3.31046711118066e-08, "epoch": 2.924444444444444, "percentage": 97.18, "elapsed_time": "1:29:20", "remaining_time": "0:02:35"}
+{"current_steps": 208, "total_steps": 213, "loss": 0.0, "lr": 2.4328929975260262e-08, "epoch": 2.9386666666666668, "percentage": 97.65, "elapsed_time": "1:29:44", "remaining_time": "0:02:09"}
+{"current_steps": 209, "total_steps": 213, "loss": 0.0, "lr": 1.6899281269279756e-08, "epoch": 2.952888888888889, "percentage": 98.12, "elapsed_time": "1:30:08", "remaining_time": "0:01:43"}
+{"current_steps": 210, "total_steps": 213, "loss": 0.0, "lr": 1.0817734972768946e-08, "epoch": 2.967111111111111, "percentage": 98.59, "elapsed_time": "1:30:31", "remaining_time": "0:01:17"}
+{"current_steps": 211, "total_steps": 213, "loss": 0.0, "lr": 6.085936355947897e-09, "epoch": 2.981333333333333, "percentage": 99.06, "elapsed_time": "1:30:55", "remaining_time": "0:00:51"}
+{"current_steps": 212, "total_steps": 213, "loss": 0.0, "lr": 2.7051655352494654e-09, "epoch": 2.9955555555555557, "percentage": 99.53, "elapsed_time": "1:31:19", "remaining_time": "0:00:25"}
+{"current_steps": 213, "total_steps": 213, "loss": 0.0, "lr": 6.763371270035457e-10, "epoch": 3.0, "percentage": 100.0, "elapsed_time": "1:31:27", "remaining_time": "0:00:00"}
+{"current_steps": 213, "total_steps": 213, "epoch": 3.0, "percentage": 100.0, "elapsed_time": "1:34:30", "remaining_time": "0:00:00"}
diff --git a/qwen2.5-7b-instruct/trainer_state.json b/qwen2.5-7b-instruct/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..9554d054dc2eb90858d42ba3eea9fc6b593ef2ec
--- /dev/null
+++ b/qwen2.5-7b-instruct/trainer_state.json
@@ -0,0 +1,1566 @@
+{
+ "best_global_step": null,
+ "best_metric": null,
+ "best_model_checkpoint": null,
+ "epoch": 3.0,
+ "eval_steps": 50,
+ "global_step": 213,
+ "is_hyper_param_search": false,
+ "is_local_process_zero": true,
+ "is_world_process_zero": true,
+ "log_history": [
+ {
+ "epoch": 0.014222222222222223,
+ "grad_norm": 0.506786584854126,
+ "learning_rate": 0.0,
+ "loss": 1.1951476335525513,
+ "step": 1
+ },
+ {
+ "epoch": 0.028444444444444446,
+ "grad_norm": 0.5440636277198792,
+ "learning_rate": 4.5454545454545457e-07,
+ "loss": 1.2398298978805542,
+ "step": 2
+ },
+ {
+ "epoch": 0.042666666666666665,
+ "grad_norm": 0.4429919421672821,
+ "learning_rate": 9.090909090909091e-07,
+ "loss": 0.9777733087539673,
+ "step": 3
+ },
+ {
+ "epoch": 0.05688888888888889,
+ "grad_norm": 0.34922653436660767,
+ "learning_rate": 1.3636363636363636e-06,
+ "loss": 0.8064044713973999,
+ "step": 4
+ },
+ {
+ "epoch": 0.07111111111111111,
+ "grad_norm": 0.16975589096546173,
+ "learning_rate": 1.8181818181818183e-06,
+ "loss": 0.5294505953788757,
+ "step": 5
+ },
+ {
+ "epoch": 0.08533333333333333,
+ "grad_norm": 0.16188450157642365,
+ "learning_rate": 2.2727272727272728e-06,
+ "loss": 0.636877179145813,
+ "step": 6
+ },
+ {
+ "epoch": 0.09955555555555555,
+ "grad_norm": 0.11110807955265045,
+ "learning_rate": 2.7272727272727272e-06,
+ "loss": 0.6087832450866699,
+ "step": 7
+ },
+ {
+ "epoch": 0.11377777777777778,
+ "grad_norm": 0.0641367956995964,
+ "learning_rate": 3.181818181818182e-06,
+ "loss": 0.5258303880691528,
+ "step": 8
+ },
+ {
+ "epoch": 0.128,
+ "grad_norm": 0.052072592079639435,
+ "learning_rate": 3.6363636363636366e-06,
+ "loss": 0.5444644689559937,
+ "step": 9
+ },
+ {
+ "epoch": 0.14222222222222222,
+ "grad_norm": 0.0392303504049778,
+ "learning_rate": 4.0909090909090915e-06,
+ "loss": 0.5337975025177002,
+ "step": 10
+ },
+ {
+ "epoch": 0.15644444444444444,
+ "grad_norm": 0.028075236827135086,
+ "learning_rate": 4.5454545454545455e-06,
+ "loss": 0.45043277740478516,
+ "step": 11
+ },
+ {
+ "epoch": 0.17066666666666666,
+ "grad_norm": 0.02663014456629753,
+ "learning_rate": 5e-06,
+ "loss": 0.49355342984199524,
+ "step": 12
+ },
+ {
+ "epoch": 0.18488888888888888,
+ "grad_norm": 0.028838366270065308,
+ "learning_rate": 5.4545454545454545e-06,
+ "loss": 0.487549751996994,
+ "step": 13
+ },
+ {
+ "epoch": 0.1991111111111111,
+ "grad_norm": 0.03057004138827324,
+ "learning_rate": 5.90909090909091e-06,
+ "loss": 0.500217616558075,
+ "step": 14
+ },
+ {
+ "epoch": 0.21333333333333335,
+ "grad_norm": 0.028651155531406403,
+ "learning_rate": 6.363636363636364e-06,
+ "loss": 0.49628522992134094,
+ "step": 15
+ },
+ {
+ "epoch": 0.22755555555555557,
+ "grad_norm": 0.028336627408862114,
+ "learning_rate": 6.818181818181818e-06,
+ "loss": 0.48170676827430725,
+ "step": 16
+ },
+ {
+ "epoch": 0.24177777777777779,
+ "grad_norm": 0.019824286922812462,
+ "learning_rate": 7.272727272727273e-06,
+ "loss": 0.4609394073486328,
+ "step": 17
+ },
+ {
+ "epoch": 0.256,
+ "grad_norm": 0.02125639095902443,
+ "learning_rate": 7.727272727272727e-06,
+ "loss": 0.4759177565574646,
+ "step": 18
+ },
+ {
+ "epoch": 0.2702222222222222,
+ "grad_norm": 0.025788351893424988,
+ "learning_rate": 8.181818181818183e-06,
+ "loss": 0.4962843060493469,
+ "step": 19
+ },
+ {
+ "epoch": 0.28444444444444444,
+ "grad_norm": 0.033310867846012115,
+ "learning_rate": 8.636363636363637e-06,
+ "loss": 0.5028114318847656,
+ "step": 20
+ },
+ {
+ "epoch": 0.2986666666666667,
+ "grad_norm": 0.020624130964279175,
+ "learning_rate": 9.090909090909091e-06,
+ "loss": 0.44575220346450806,
+ "step": 21
+ },
+ {
+ "epoch": 0.3128888888888889,
+ "grad_norm": 0.020986203104257584,
+ "learning_rate": 9.545454545454547e-06,
+ "loss": 0.4755765199661255,
+ "step": 22
+ },
+ {
+ "epoch": 0.32711111111111113,
+ "grad_norm": 2526660.75,
+ "learning_rate": 1e-05,
+ "loss": 0.4706365466117859,
+ "step": 23
+ },
+ {
+ "epoch": 0.3413333333333333,
+ "grad_norm": 0.021011805161833763,
+ "learning_rate": 9.999323662872998e-06,
+ "loss": 0.4945886731147766,
+ "step": 24
+ },
+ {
+ "epoch": 0.35555555555555557,
+ "grad_norm": 0.020573964342474937,
+ "learning_rate": 9.99729483446475e-06,
+ "loss": 0.4544343054294586,
+ "step": 25
+ },
+ {
+ "epoch": 0.36977777777777776,
+ "grad_norm": 0.021116767078638077,
+ "learning_rate": 9.993914063644053e-06,
+ "loss": 0.4556806683540344,
+ "step": 26
+ },
+ {
+ "epoch": 0.384,
+ "grad_norm": 0.019978955388069153,
+ "learning_rate": 9.989182265027232e-06,
+ "loss": 0.4431076645851135,
+ "step": 27
+ },
+ {
+ "epoch": 0.3982222222222222,
+ "grad_norm": 0.019631512463092804,
+ "learning_rate": 9.98310071873072e-06,
+ "loss": 0.4541151523590088,
+ "step": 28
+ },
+ {
+ "epoch": 0.41244444444444445,
+ "grad_norm": 0.01977457106113434,
+ "learning_rate": 9.975671070024741e-06,
+ "loss": 0.4352927803993225,
+ "step": 29
+ },
+ {
+ "epoch": 0.4266666666666667,
+ "grad_norm": 0.022281618788838387,
+ "learning_rate": 9.966895328888195e-06,
+ "loss": 0.5559179186820984,
+ "step": 30
+ },
+ {
+ "epoch": 0.4408888888888889,
+ "grad_norm": 0.018352897837758064,
+ "learning_rate": 9.956775869464901e-06,
+ "loss": 0.42942947149276733,
+ "step": 31
+ },
+ {
+ "epoch": 0.45511111111111113,
+ "grad_norm": 0.019121287390589714,
+ "learning_rate": 9.945315429421307e-06,
+ "loss": 0.45916861295700073,
+ "step": 32
+ },
+ {
+ "epoch": 0.4693333333333333,
+ "grad_norm": 0.01712602935731411,
+ "learning_rate": 9.932517109205849e-06,
+ "loss": 0.4312085807323456,
+ "step": 33
+ },
+ {
+ "epoch": 0.48355555555555557,
+ "grad_norm": 0.01892935112118721,
+ "learning_rate": 9.918384371210178e-06,
+ "loss": 0.5169366598129272,
+ "step": 34
+ },
+ {
+ "epoch": 0.49777777777777776,
+ "grad_norm": 0.02034984715282917,
+ "learning_rate": 9.902921038832456e-06,
+ "loss": 0.49438145756721497,
+ "step": 35
+ },
+ {
+ "epoch": 0.512,
+ "grad_norm": 0.017887447029352188,
+ "learning_rate": 9.886131295443003e-06,
+ "loss": 0.4683716893196106,
+ "step": 36
+ },
+ {
+ "epoch": 0.5262222222222223,
+ "grad_norm": 0.01887686364352703,
+ "learning_rate": 9.868019683252543e-06,
+ "loss": 0.46688079833984375,
+ "step": 37
+ },
+ {
+ "epoch": 0.5404444444444444,
+ "grad_norm": 0.019363289698958397,
+ "learning_rate": 9.848591102083375e-06,
+ "loss": 0.4786224067211151,
+ "step": 38
+ },
+ {
+ "epoch": 0.5546666666666666,
+ "grad_norm": 0.0201108418405056,
+ "learning_rate": 9.82785080804381e-06,
+ "loss": 0.47358086705207825,
+ "step": 39
+ },
+ {
+ "epoch": 0.5688888888888889,
+ "grad_norm": 0.019631700590252876,
+ "learning_rate": 9.805804412106197e-06,
+ "loss": 0.585416316986084,
+ "step": 40
+ },
+ {
+ "epoch": 0.5831111111111111,
+ "grad_norm": 0.01689487136900425,
+ "learning_rate": 9.782457878588977e-06,
+ "loss": 0.47205525636672974,
+ "step": 41
+ },
+ {
+ "epoch": 0.5973333333333334,
+ "grad_norm": 0.01628926210105419,
+ "learning_rate": 9.75781752354311e-06,
+ "loss": 0.4348371624946594,
+ "step": 42
+ },
+ {
+ "epoch": 0.6115555555555555,
+ "grad_norm": 0.01758437603712082,
+ "learning_rate": 9.731890013043367e-06,
+ "loss": 0.4891083538532257,
+ "step": 43
+ },
+ {
+ "epoch": 0.6257777777777778,
+ "grad_norm": 0.017766334116458893,
+ "learning_rate": 9.704682361384941e-06,
+ "loss": 0.46330946683883667,
+ "step": 44
+ },
+ {
+ "epoch": 0.64,
+ "grad_norm": 0.017524050548672676,
+ "learning_rate": 9.676201929185809e-06,
+ "loss": 0.4083091914653778,
+ "step": 45
+ },
+ {
+ "epoch": 0.6542222222222223,
+ "grad_norm": 0.017291702330112457,
+ "learning_rate": 9.646456421395447e-06,
+ "loss": 0.4653710126876831,
+ "step": 46
+ },
+ {
+ "epoch": 0.6684444444444444,
+ "grad_norm": 0.017993608489632607,
+ "learning_rate": 9.615453885210368e-06,
+ "loss": 0.4626290202140808,
+ "step": 47
+ },
+ {
+ "epoch": 0.6826666666666666,
+ "grad_norm": 0.017247265204787254,
+ "learning_rate": 9.583202707897075e-06,
+ "loss": 0.47964128851890564,
+ "step": 48
+ },
+ {
+ "epoch": 0.6968888888888889,
+ "grad_norm": 0.017244094982743263,
+ "learning_rate": 9.549711614523007e-06,
+ "loss": 0.46718651056289673,
+ "step": 49
+ },
+ {
+ "epoch": 0.7111111111111111,
+ "grad_norm": 0.01574530452489853,
+ "learning_rate": 9.514989665596114e-06,
+ "loss": 0.4456287622451782,
+ "step": 50
+ },
+ {
+ "epoch": 0.7111111111111111,
+ "eval_loss": 0.4940151274204254,
+ "eval_runtime": 6.2328,
+ "eval_samples_per_second": 32.088,
+ "eval_steps_per_second": 8.022,
+ "step": 50
+ },
+ {
+ "epoch": 0.7253333333333334,
+ "grad_norm": 0.01783670485019684,
+ "learning_rate": 9.479046254613673e-06,
+ "loss": 0.45046865940093994,
+ "step": 51
+ },
+ {
+ "epoch": 0.7395555555555555,
+ "grad_norm": 0.015281656756997108,
+ "learning_rate": 9.441891105521005e-06,
+ "loss": 0.43093934655189514,
+ "step": 52
+ },
+ {
+ "epoch": 0.7537777777777778,
+ "grad_norm": 0.017853204160928726,
+ "learning_rate": 9.40353427008083e-06,
+ "loss": 0.4706975519657135,
+ "step": 53
+ },
+ {
+ "epoch": 0.768,
+ "grad_norm": 0.01542010996490717,
+ "learning_rate": 9.3639861251539e-06,
+ "loss": 0.4405261278152466,
+ "step": 54
+ },
+ {
+ "epoch": 0.7822222222222223,
+ "grad_norm": 0.017832862213253975,
+ "learning_rate": 9.323257369891702e-06,
+ "loss": 0.47680750489234924,
+ "step": 55
+ },
+ {
+ "epoch": 0.7964444444444444,
+ "grad_norm": 0.015628881752490997,
+ "learning_rate": 9.281359022841966e-06,
+ "loss": 0.42392900586128235,
+ "step": 56
+ },
+ {
+ "epoch": 0.8106666666666666,
+ "grad_norm": 0.01757017895579338,
+ "learning_rate": 9.238302418967757e-06,
+ "loss": 0.48122283816337585,
+ "step": 57
+ },
+ {
+ "epoch": 0.8248888888888889,
+ "grad_norm": 0.017186852172017097,
+ "learning_rate": 9.194099206580981e-06,
+ "loss": 0.5255264043807983,
+ "step": 58
+ },
+ {
+ "epoch": 0.8391111111111111,
+ "grad_norm": 0.017161328345537186,
+ "learning_rate": 9.14876134419111e-06,
+ "loss": 0.4682884216308594,
+ "step": 59
+ },
+ {
+ "epoch": 0.8533333333333334,
+ "grad_norm": 0.01665889099240303,
+ "learning_rate": 9.102301097269974e-06,
+ "loss": 0.5003230571746826,
+ "step": 60
+ },
+ {
+ "epoch": 0.8675555555555555,
+ "grad_norm": 0.02079184167087078,
+ "learning_rate": 9.05473103493355e-06,
+ "loss": 0.45744988322257996,
+ "step": 61
+ },
+ {
+ "epoch": 0.8817777777777778,
+ "grad_norm": 0.018035434186458588,
+ "learning_rate": 9.006064026541549e-06,
+ "loss": 0.5037047863006592,
+ "step": 62
+ },
+ {
+ "epoch": 0.896,
+ "grad_norm": 0.015980787575244904,
+ "learning_rate": 8.956313238215824e-06,
+ "loss": 0.4382043778896332,
+ "step": 63
+ },
+ {
+ "epoch": 0.9102222222222223,
+ "grad_norm": 0.016865955665707588,
+ "learning_rate": 8.905492129278478e-06,
+ "loss": 0.44455963373184204,
+ "step": 64
+ },
+ {
+ "epoch": 0.9244444444444444,
+ "grad_norm": 0.014777022413909435,
+ "learning_rate": 8.85361444861063e-06,
+ "loss": 0.4215100109577179,
+ "step": 65
+ },
+ {
+ "epoch": 0.9386666666666666,
+ "grad_norm": 0.01814866065979004,
+ "learning_rate": 8.800694230932885e-06,
+ "loss": 0.48813939094543457,
+ "step": 66
+ },
+ {
+ "epoch": 0.9528888888888889,
+ "grad_norm": 0.013353748247027397,
+ "learning_rate": 8.74674579300843e-06,
+ "loss": 3.263723850250244,
+ "step": 67
+ },
+ {
+ "epoch": 0.9671111111111111,
+ "grad_norm": 0.014444677159190178,
+ "learning_rate": 8.691783729769874e-06,
+ "loss": 2.8557608127593994,
+ "step": 68
+ },
+ {
+ "epoch": 0.9813333333333333,
+ "grad_norm": 0.01388335507363081,
+ "learning_rate": 8.635822910370793e-06,
+ "loss": 3.485170602798462,
+ "step": 69
+ },
+ {
+ "epoch": 0.9955555555555555,
+ "grad_norm": 0.01357923075556755,
+ "learning_rate": 8.578878474163115e-06,
+ "loss": 3.1958985328674316,
+ "step": 70
+ },
+ {
+ "epoch": 1.0,
+ "grad_norm": 0.02305089496076107,
+ "learning_rate": 8.520965826601394e-06,
+ "loss": 2.678990364074707,
+ "step": 71
+ },
+ {
+ "epoch": 1.0142222222222221,
+ "grad_norm": 0.014260848984122276,
+ "learning_rate": 8.462100635075097e-06,
+ "loss": 3.4466443061828613,
+ "step": 72
+ },
+ {
+ "epoch": 1.0284444444444445,
+ "grad_norm": 0.014910527504980564,
+ "learning_rate": 8.40229882467003e-06,
+ "loss": 3.4718263149261475,
+ "step": 73
+ },
+ {
+ "epoch": 1.0426666666666666,
+ "grad_norm": 0.013167285360395908,
+ "learning_rate": 8.341576573860049e-06,
+ "loss": 3.6029202938079834,
+ "step": 74
+ },
+ {
+ "epoch": 1.056888888888889,
+ "grad_norm": 0.012767510488629341,
+ "learning_rate": 8.279950310130218e-06,
+ "loss": 4.018604755401611,
+ "step": 75
+ },
+ {
+ "epoch": 1.0711111111111111,
+ "grad_norm": 0.013733700849115849,
+ "learning_rate": 8.2174367055326e-06,
+ "loss": 3.0605146884918213,
+ "step": 76
+ },
+ {
+ "epoch": 1.0853333333333333,
+ "grad_norm": 0.011987799778580666,
+ "learning_rate": 8.154052672175888e-06,
+ "loss": 3.556879997253418,
+ "step": 77
+ },
+ {
+ "epoch": 1.0995555555555556,
+ "grad_norm": 0.014038470573723316,
+ "learning_rate": 8.08981535765009e-06,
+ "loss": 3.7346572875976562,
+ "step": 78
+ },
+ {
+ "epoch": 1.1137777777777778,
+ "grad_norm": 0.013301494531333447,
+ "learning_rate": 8.024742140387506e-06,
+ "loss": 3.2394087314605713,
+ "step": 79
+ },
+ {
+ "epoch": 1.1280000000000001,
+ "grad_norm": 0.013333727605640888,
+ "learning_rate": 7.95885062496126e-06,
+ "loss": 3.625601053237915,
+ "step": 80
+ },
+ {
+ "epoch": 1.1422222222222222,
+ "grad_norm": 0.013422077521681786,
+ "learning_rate": 7.892158637322647e-06,
+ "loss": 3.3379783630371094,
+ "step": 81
+ },
+ {
+ "epoch": 1.1564444444444444,
+ "grad_norm": 0.012368597090244293,
+ "learning_rate": 7.824684219978591e-06,
+ "loss": 3.502805471420288,
+ "step": 82
+ },
+ {
+ "epoch": 1.1706666666666667,
+ "grad_norm": 0.013034222647547722,
+ "learning_rate": 7.756445627110523e-06,
+ "loss": 3.362370729446411,
+ "step": 83
+ },
+ {
+ "epoch": 1.1848888888888889,
+ "grad_norm": 0.014766803942620754,
+ "learning_rate": 7.68746131963598e-06,
+ "loss": 3.1864588260650635,
+ "step": 84
+ },
+ {
+ "epoch": 1.199111111111111,
+ "grad_norm": 0.012399553321301937,
+ "learning_rate": 7.6177499602143e-06,
+ "loss": 3.4393932819366455,
+ "step": 85
+ },
+ {
+ "epoch": 1.2133333333333334,
+ "grad_norm": 0.014413993805646896,
+ "learning_rate": 7.547330408197695e-06,
+ "loss": 3.76263689994812,
+ "step": 86
+ },
+ {
+ "epoch": 1.2275555555555555,
+ "grad_norm": 0.014122135005891323,
+ "learning_rate": 7.476221714529167e-06,
+ "loss": 3.3016631603240967,
+ "step": 87
+ },
+ {
+ "epoch": 1.2417777777777779,
+ "grad_norm": 0.015857728198170662,
+ "learning_rate": 7.404443116588548e-06,
+ "loss": 2.8077316284179688,
+ "step": 88
+ },
+ {
+ "epoch": 1.256,
+ "grad_norm": 0.013196472078561783,
+ "learning_rate": 7.332014032988123e-06,
+ "loss": 3.496107816696167,
+ "step": 89
+ },
+ {
+ "epoch": 1.2702222222222221,
+ "grad_norm": 0.01415536180138588,
+ "learning_rate": 7.2589540583192165e-06,
+ "loss": 3.4046881198883057,
+ "step": 90
+ },
+ {
+ "epoch": 1.2844444444444445,
+ "grad_norm": 0.01354147121310234,
+ "learning_rate": 7.185282957851175e-06,
+ "loss": 3.3288354873657227,
+ "step": 91
+ },
+ {
+ "epoch": 1.2986666666666666,
+ "grad_norm": 0.013861758634448051,
+ "learning_rate": 7.111020662184174e-06,
+ "loss": 3.573462963104248,
+ "step": 92
+ },
+ {
+ "epoch": 1.3128888888888888,
+ "grad_norm": 0.01368795894086361,
+ "learning_rate": 7.036187261857289e-06,
+ "loss": 3.4570071697235107,
+ "step": 93
+ },
+ {
+ "epoch": 1.3271111111111111,
+ "grad_norm": 0.014616986736655235,
+ "learning_rate": 6.960803001913315e-06,
+ "loss": 3.511754274368286,
+ "step": 94
+ },
+ {
+ "epoch": 1.3413333333333333,
+ "grad_norm": 0.01352851465344429,
+ "learning_rate": 6.884888276421766e-06,
+ "loss": 4.017252445220947,
+ "step": 95
+ },
+ {
+ "epoch": 1.3555555555555556,
+ "grad_norm": 0.013078316114842892,
+ "learning_rate": 6.8084636229615786e-06,
+ "loss": 3.5670371055603027,
+ "step": 96
+ },
+ {
+ "epoch": 1.3697777777777778,
+ "grad_norm": 0.012880469672381878,
+ "learning_rate": 6.731549717064975e-06,
+ "loss": 4.055681228637695,
+ "step": 97
+ },
+ {
+ "epoch": 1.384,
+ "grad_norm": 0.013766918331384659,
+ "learning_rate": 6.654167366624009e-06,
+ "loss": 3.303616762161255,
+ "step": 98
+ },
+ {
+ "epoch": 1.3982222222222223,
+ "grad_norm": 0.015087485313415527,
+ "learning_rate": 6.576337506261314e-06,
+ "loss": 2.8152995109558105,
+ "step": 99
+ },
+ {
+ "epoch": 1.4124444444444444,
+ "grad_norm": 0.013361976481974125,
+ "learning_rate": 6.498081191666549e-06,
+ "loss": 3.3619225025177,
+ "step": 100
+ },
+ {
+ "epoch": 1.4124444444444444,
+ "eval_loss": 3.694061279296875,
+ "eval_runtime": 5.8211,
+ "eval_samples_per_second": 34.358,
+ "eval_steps_per_second": 8.589,
+ "step": 100
+ },
+ {
+ "epoch": 1.4266666666666667,
+ "grad_norm": 0.013033254072070122,
+ "learning_rate": 6.419419593900109e-06,
+ "loss": 3.0373668670654297,
+ "step": 101
+ },
+ {
+ "epoch": 1.4408888888888889,
+ "grad_norm": 0.012533189728856087,
+ "learning_rate": 6.340373993665607e-06,
+ "loss": 3.642394542694092,
+ "step": 102
+ },
+ {
+ "epoch": 1.455111111111111,
+ "grad_norm": 0.013018937781453133,
+ "learning_rate": 6.2609657755527135e-06,
+ "loss": 3.3127541542053223,
+ "step": 103
+ },
+ {
+ "epoch": 1.4693333333333334,
+ "grad_norm": 0.014089098200201988,
+ "learning_rate": 6.1812164222518626e-06,
+ "loss": 3.500901937484741,
+ "step": 104
+ },
+ {
+ "epoch": 1.4835555555555555,
+ "grad_norm": 0.0140894316136837,
+ "learning_rate": 6.101147508742456e-06,
+ "loss": 3.4911036491394043,
+ "step": 105
+ },
+ {
+ "epoch": 1.4977777777777779,
+ "grad_norm": 0.014050478115677834,
+ "learning_rate": 6.020780696456059e-06,
+ "loss": 3.375718116760254,
+ "step": 106
+ },
+ {
+ "epoch": 1.512,
+ "grad_norm": 0.013563934713602066,
+ "learning_rate": 5.940137727416247e-06,
+ "loss": 3.340306282043457,
+ "step": 107
+ },
+ {
+ "epoch": 1.5262222222222221,
+ "grad_norm": 0.013330376707017422,
+ "learning_rate": 5.859240418356614e-06,
+ "loss": 3.5164387226104736,
+ "step": 108
+ },
+ {
+ "epoch": 1.5404444444444443,
+ "grad_norm": 0.013517254032194614,
+ "learning_rate": 5.778110654818602e-06,
+ "loss": 3.955772876739502,
+ "step": 109
+ },
+ {
+ "epoch": 1.5546666666666666,
+ "grad_norm": 0.014757441356778145,
+ "learning_rate": 5.696770385230679e-06,
+ "loss": 3.425647020339966,
+ "step": 110
+ },
+ {
+ "epoch": 1.568888888888889,
+ "grad_norm": 0.014494120143353939,
+ "learning_rate": 5.615241614970546e-06,
+ "loss": 3.219924211502075,
+ "step": 111
+ },
+ {
+ "epoch": 1.5831111111111111,
+ "grad_norm": 0.012901407666504383,
+ "learning_rate": 5.533546400411899e-06,
+ "loss": 3.3812978267669678,
+ "step": 112
+ },
+ {
+ "epoch": 1.5973333333333333,
+ "grad_norm": 0.013786875642836094,
+ "learning_rate": 5.4517068429574215e-06,
+ "loss": 3.7506139278411865,
+ "step": 113
+ },
+ {
+ "epoch": 1.6115555555555554,
+ "grad_norm": 0.012589235790073872,
+ "learning_rate": 5.3697450830595775e-06,
+ "loss": 3.259803295135498,
+ "step": 114
+ },
+ {
+ "epoch": 1.6257777777777778,
+ "grad_norm": 0.01369628868997097,
+ "learning_rate": 5.287683294230855e-06,
+ "loss": 2.8727362155914307,
+ "step": 115
+ },
+ {
+ "epoch": 1.6400000000000001,
+ "grad_norm": 0.012632690370082855,
+ "learning_rate": 5.20554367704505e-06,
+ "loss": 3.628926992416382,
+ "step": 116
+ },
+ {
+ "epoch": 1.6542222222222223,
+ "grad_norm": 0.014506797306239605,
+ "learning_rate": 5.123348453131242e-06,
+ "loss": 3.236464023590088,
+ "step": 117
+ },
+ {
+ "epoch": 1.6684444444444444,
+ "grad_norm": 0.014436502009630203,
+ "learning_rate": 5.041119859162068e-06,
+ "loss": 2.888354778289795,
+ "step": 118
+ },
+ {
+ "epoch": 1.6826666666666665,
+ "grad_norm": 0.014416320249438286,
+ "learning_rate": 4.958880140837934e-06,
+ "loss": 3.105433702468872,
+ "step": 119
+ },
+ {
+ "epoch": 1.696888888888889,
+ "grad_norm": 0.014613285660743713,
+ "learning_rate": 4.876651546868759e-06,
+ "loss": 3.2962746620178223,
+ "step": 120
+ },
+ {
+ "epoch": 1.7111111111111112,
+ "grad_norm": 0.014369661919772625,
+ "learning_rate": 4.794456322954953e-06,
+ "loss": 3.451850414276123,
+ "step": 121
+ },
+ {
+ "epoch": 1.7253333333333334,
+ "grad_norm": 0.013792592100799084,
+ "learning_rate": 4.7123167057691446e-06,
+ "loss": 3.6925525665283203,
+ "step": 122
+ },
+ {
+ "epoch": 1.7395555555555555,
+ "grad_norm": 0.013583365827798843,
+ "learning_rate": 4.630254916940424e-06,
+ "loss": 4.191713333129883,
+ "step": 123
+ },
+ {
+ "epoch": 1.7537777777777777,
+ "grad_norm": 0.014320321381092072,
+ "learning_rate": 4.548293157042581e-06,
+ "loss": 2.8300790786743164,
+ "step": 124
+ },
+ {
+ "epoch": 1.768,
+ "grad_norm": 0.013430479913949966,
+ "learning_rate": 4.466453599588103e-06,
+ "loss": 3.5787062644958496,
+ "step": 125
+ },
+ {
+ "epoch": 1.7822222222222224,
+ "grad_norm": 0.012813941575586796,
+ "learning_rate": 4.384758385029457e-06,
+ "loss": 2.687720537185669,
+ "step": 126
+ },
+ {
+ "epoch": 1.7964444444444445,
+ "grad_norm": 0.014274616725742817,
+ "learning_rate": 4.3032296147693225e-06,
+ "loss": 3.221151113510132,
+ "step": 127
+ },
+ {
+ "epoch": 1.8106666666666666,
+ "grad_norm": 0.013514580205082893,
+ "learning_rate": 4.2218893451814005e-06,
+ "loss": 3.415364980697632,
+ "step": 128
+ },
+ {
+ "epoch": 1.8248888888888888,
+ "grad_norm": 0.01293711643666029,
+ "learning_rate": 4.140759581643386e-06,
+ "loss": 3.2137668132781982,
+ "step": 129
+ },
+ {
+ "epoch": 1.8391111111111111,
+ "grad_norm": 0.013326463289558887,
+ "learning_rate": 4.059862272583755e-06,
+ "loss": 2.874941349029541,
+ "step": 130
+ },
+ {
+ "epoch": 1.8533333333333335,
+ "grad_norm": 0.013346531428396702,
+ "learning_rate": 3.979219303543942e-06,
+ "loss": 3.6437442302703857,
+ "step": 131
+ },
+ {
+ "epoch": 1.8675555555555556,
+ "grad_norm": 0.013448791578412056,
+ "learning_rate": 3.898852491257547e-06,
+ "loss": 3.262843370437622,
+ "step": 132
+ },
+ {
+ "epoch": 1.8817777777777778,
+ "grad_norm": 0.013963771052658558,
+ "learning_rate": 3.818783577748138e-06,
+ "loss": 2.763110637664795,
+ "step": 133
+ },
+ {
+ "epoch": 1.896,
+ "grad_norm": 0.01534951850771904,
+ "learning_rate": 3.7390342244472886e-06,
+ "loss": 2.978278160095215,
+ "step": 134
+ },
+ {
+ "epoch": 1.9102222222222223,
+ "grad_norm": 0.014479058794677258,
+ "learning_rate": 3.659626006334395e-06,
+ "loss": 2.9288432598114014,
+ "step": 135
+ },
+ {
+ "epoch": 1.9244444444444444,
+ "grad_norm": 0.01563713327050209,
+ "learning_rate": 3.5805804060998926e-06,
+ "loss": 3.158463954925537,
+ "step": 136
+ },
+ {
+ "epoch": 1.9386666666666668,
+ "grad_norm": 0.01603933610022068,
+ "learning_rate": 3.501918808333453e-06,
+ "loss": 2.699950933456421,
+ "step": 137
+ },
+ {
+ "epoch": 1.952888888888889,
+ "grad_norm": 0.014355695806443691,
+ "learning_rate": 3.4236624937386874e-06,
+ "loss": 3.840054512023926,
+ "step": 138
+ },
+ {
+ "epoch": 1.967111111111111,
+ "grad_norm": 0.013930619694292545,
+ "learning_rate": 3.3458326333759927e-06,
+ "loss": 2.972839832305908,
+ "step": 139
+ },
+ {
+ "epoch": 1.9813333333333332,
+ "grad_norm": 0.015130864456295967,
+ "learning_rate": 3.268450282935026e-06,
+ "loss": 2.8588995933532715,
+ "step": 140
+ },
+ {
+ "epoch": 1.9955555555555555,
+ "grad_norm": 0.013637659139931202,
+ "learning_rate": 3.1915363770384223e-06,
+ "loss": 3.6230926513671875,
+ "step": 141
+ },
+ {
+ "epoch": 2.0,
+ "grad_norm": 0.023099087178707123,
+ "learning_rate": 3.1151117235782346e-06,
+ "loss": 2.1066107749938965,
+ "step": 142
+ },
+ {
+ "epoch": 2.014222222222222,
+ "grad_norm": 0.014238209463655949,
+ "learning_rate": 3.0391969980866874e-06,
+ "loss": 3.128911256790161,
+ "step": 143
+ },
+ {
+ "epoch": 2.0284444444444443,
+ "grad_norm": 0.012874635867774487,
+ "learning_rate": 2.963812738142713e-06,
+ "loss": 3.2379772663116455,
+ "step": 144
+ },
+ {
+ "epoch": 2.042666666666667,
+ "grad_norm": 0.013502821326255798,
+ "learning_rate": 2.8889793378158284e-06,
+ "loss": 3.2959132194519043,
+ "step": 145
+ },
+ {
+ "epoch": 2.056888888888889,
+ "grad_norm": 0.013256440870463848,
+ "learning_rate": 2.814717042148827e-06,
+ "loss": 3.3406944274902344,
+ "step": 146
+ },
+ {
+ "epoch": 2.071111111111111,
+ "grad_norm": 0.013458102941513062,
+ "learning_rate": 2.7410459416807856e-06,
+ "loss": 3.3720455169677734,
+ "step": 147
+ },
+ {
+ "epoch": 2.0853333333333333,
+ "grad_norm": 0.01476226095110178,
+ "learning_rate": 2.6679859670118785e-06,
+ "loss": 3.153254508972168,
+ "step": 148
+ },
+ {
+ "epoch": 2.0995555555555554,
+ "grad_norm": 0.013766537420451641,
+ "learning_rate": 2.5955568834114523e-06,
+ "loss": 2.9255833625793457,
+ "step": 149
+ },
+ {
+ "epoch": 2.113777777777778,
+ "grad_norm": 0.012672817334532738,
+ "learning_rate": 2.523778285470835e-06,
+ "loss": 3.385373115539551,
+ "step": 150
+ },
+ {
+ "epoch": 2.113777777777778,
+ "eval_loss": 3.694100856781006,
+ "eval_runtime": 6.2397,
+ "eval_samples_per_second": 32.053,
+ "eval_steps_per_second": 8.013,
+ "step": 150
+ },
+ {
+ "epoch": 2.128,
+ "grad_norm": 0.013193093240261078,
+ "learning_rate": 2.452669591802307e-06,
+ "loss": 2.953336238861084,
+ "step": 151
+ },
+ {
+ "epoch": 2.1422222222222222,
+ "grad_norm": 0.012676793150603771,
+ "learning_rate": 2.3822500397857016e-06,
+ "loss": 3.4797470569610596,
+ "step": 152
+ },
+ {
+ "epoch": 2.1564444444444444,
+ "grad_norm": 0.013178958557546139,
+ "learning_rate": 2.3125386803640188e-06,
+ "loss": 3.766719341278076,
+ "step": 153
+ },
+ {
+ "epoch": 2.1706666666666665,
+ "grad_norm": 0.013044048100709915,
+ "learning_rate": 2.243554372889479e-06,
+ "loss": 3.105588436126709,
+ "step": 154
+ },
+ {
+ "epoch": 2.1848888888888887,
+ "grad_norm": 0.01274812500923872,
+ "learning_rate": 2.1753157800214107e-06,
+ "loss": 3.7258148193359375,
+ "step": 155
+ },
+ {
+ "epoch": 2.1991111111111112,
+ "grad_norm": 0.014226537197828293,
+ "learning_rate": 2.1078413626773547e-06,
+ "loss": 3.2563636302948,
+ "step": 156
+ },
+ {
+ "epoch": 2.2133333333333334,
+ "grad_norm": 0.013875821605324745,
+ "learning_rate": 2.0411493750387423e-06,
+ "loss": 3.6285147666931152,
+ "step": 157
+ },
+ {
+ "epoch": 2.2275555555555555,
+ "grad_norm": 0.013328024186193943,
+ "learning_rate": 1.9752578596124955e-06,
+ "loss": 3.4809606075286865,
+ "step": 158
+ },
+ {
+ "epoch": 2.2417777777777776,
+ "grad_norm": 0.014153145253658295,
+ "learning_rate": 1.9101846423499113e-06,
+ "loss": 3.331500768661499,
+ "step": 159
+ },
+ {
+ "epoch": 2.2560000000000002,
+ "grad_norm": 0.013629083521664143,
+ "learning_rate": 1.8459473278241125e-06,
+ "loss": 3.103165626525879,
+ "step": 160
+ },
+ {
+ "epoch": 2.2702222222222224,
+ "grad_norm": 0.013135570101439953,
+ "learning_rate": 1.7825632944674016e-06,
+ "loss": 2.949009656906128,
+ "step": 161
+ },
+ {
+ "epoch": 2.2844444444444445,
+ "grad_norm": 0.015113947913050652,
+ "learning_rate": 1.7200496898697832e-06,
+ "loss": 3.8803815841674805,
+ "step": 162
+ },
+ {
+ "epoch": 2.2986666666666666,
+ "grad_norm": 0.01285882480442524,
+ "learning_rate": 1.6584234261399535e-06,
+ "loss": 3.3277387619018555,
+ "step": 163
+ },
+ {
+ "epoch": 2.3128888888888888,
+ "grad_norm": 0.013268567621707916,
+ "learning_rate": 1.5977011753299726e-06,
+ "loss": 3.04113507270813,
+ "step": 164
+ },
+ {
+ "epoch": 2.327111111111111,
+ "grad_norm": 0.015572321601212025,
+ "learning_rate": 1.5378993649249053e-06,
+ "loss": 3.1088926792144775,
+ "step": 165
+ },
+ {
+ "epoch": 2.3413333333333335,
+ "grad_norm": 0.013958935625851154,
+ "learning_rate": 1.4790341733986085e-06,
+ "loss": 2.9704766273498535,
+ "step": 166
+ },
+ {
+ "epoch": 2.3555555555555556,
+ "grad_norm": 0.01262224093079567,
+ "learning_rate": 1.4211215258368866e-06,
+ "loss": 2.666156053543091,
+ "step": 167
+ },
+ {
+ "epoch": 2.3697777777777778,
+ "grad_norm": 0.01445838063955307,
+ "learning_rate": 1.3641770896292083e-06,
+ "loss": 3.5968339443206787,
+ "step": 168
+ },
+ {
+ "epoch": 2.384,
+ "grad_norm": 0.013657037168741226,
+ "learning_rate": 1.3082162702301276e-06,
+ "loss": 3.576735496520996,
+ "step": 169
+ },
+ {
+ "epoch": 2.398222222222222,
+ "grad_norm": 0.014326483011245728,
+ "learning_rate": 1.2532542069915722e-06,
+ "loss": 3.0840680599212646,
+ "step": 170
+ },
+ {
+ "epoch": 2.4124444444444446,
+ "grad_norm": 0.014063039794564247,
+ "learning_rate": 1.1993057690671174e-06,
+ "loss": 3.7334625720977783,
+ "step": 171
+ },
+ {
+ "epoch": 2.4266666666666667,
+ "grad_norm": 0.012410550378262997,
+ "learning_rate": 1.1463855513893695e-06,
+ "loss": 4.4254841804504395,
+ "step": 172
+ },
+ {
+ "epoch": 2.440888888888889,
+ "grad_norm": 0.012715302407741547,
+ "learning_rate": 1.0945078707215224e-06,
+ "loss": 3.775759696960449,
+ "step": 173
+ },
+ {
+ "epoch": 2.455111111111111,
+ "grad_norm": 0.014594604261219501,
+ "learning_rate": 1.0436867617841768e-06,
+ "loss": 2.9598851203918457,
+ "step": 174
+ },
+ {
+ "epoch": 2.469333333333333,
+ "grad_norm": 0.013803915120661259,
+ "learning_rate": 9.939359734584552e-07,
+ "loss": 2.9934566020965576,
+ "step": 175
+ },
+ {
+ "epoch": 2.4835555555555557,
+ "grad_norm": 0.013807227835059166,
+ "learning_rate": 9.452689650664515e-07,
+ "loss": 3.3066673278808594,
+ "step": 176
+ },
+ {
+ "epoch": 2.497777777777778,
+ "grad_norm": 0.013583092018961906,
+ "learning_rate": 8.976989027300265e-07,
+ "loss": 2.8772518634796143,
+ "step": 177
+ },
+ {
+ "epoch": 2.512,
+ "grad_norm": NaN,
+ "learning_rate": 8.512386558088919e-07,
+ "loss": 0.2714032232761383,
+ "step": 178
+ },
+ {
+ "epoch": 2.526222222222222,
+ "grad_norm": NaN,
+ "learning_rate": 8.059007934190194e-07,
+ "loss": 0.0,
+ "step": 179
+ },
+ {
+ "epoch": 2.5404444444444443,
+ "grad_norm": NaN,
+ "learning_rate": 7.61697581032243e-07,
+ "loss": 0.0,
+ "step": 180
+ },
+ {
+ "epoch": 2.554666666666667,
+ "grad_norm": NaN,
+ "learning_rate": 7.186409771580355e-07,
+ "loss": 0.0,
+ "step": 181
+ },
+ {
+ "epoch": 2.568888888888889,
+ "grad_norm": NaN,
+ "learning_rate": 6.76742630108298e-07,
+ "loss": 0.0,
+ "step": 182
+ },
+ {
+ "epoch": 2.583111111111111,
+ "grad_norm": NaN,
+ "learning_rate": 6.360138748461015e-07,
+ "loss": 0.0,
+ "step": 183
+ },
+ {
+ "epoch": 2.5973333333333333,
+ "grad_norm": NaN,
+ "learning_rate": 5.964657299191712e-07,
+ "loss": 0.0,
+ "step": 184
+ },
+ {
+ "epoch": 2.6115555555555554,
+ "grad_norm": NaN,
+ "learning_rate": 5.581088944789953e-07,
+ "loss": 0.0,
+ "step": 185
+ },
+ {
+ "epoch": 2.6257777777777775,
+ "grad_norm": NaN,
+ "learning_rate": 5.20953745386329e-07,
+ "loss": 0.0,
+ "step": 186
+ },
+ {
+ "epoch": 2.64,
+ "grad_norm": NaN,
+ "learning_rate": 4.850103344038853e-07,
+ "loss": 0.0,
+ "step": 187
+ },
+ {
+ "epoch": 2.6542222222222223,
+ "grad_norm": NaN,
+ "learning_rate": 4.502883854769935e-07,
+ "loss": 0.0,
+ "step": 188
+ },
+ {
+ "epoch": 2.6684444444444444,
+ "grad_norm": NaN,
+ "learning_rate": 4.167972921029262e-07,
+ "loss": 0.0,
+ "step": 189
+ },
+ {
+ "epoch": 2.6826666666666665,
+ "grad_norm": NaN,
+ "learning_rate": 3.8454611478963235e-07,
+ "loss": 0.0,
+ "step": 190
+ },
+ {
+ "epoch": 2.696888888888889,
+ "grad_norm": NaN,
+ "learning_rate": 3.535435786045538e-07,
+ "loss": 0.0,
+ "step": 191
+ },
+ {
+ "epoch": 2.7111111111111112,
+ "grad_norm": NaN,
+ "learning_rate": 3.237980708141919e-07,
+ "loss": 0.0,
+ "step": 192
+ },
+ {
+ "epoch": 2.7253333333333334,
+ "grad_norm": NaN,
+ "learning_rate": 2.9531763861505967e-07,
+ "loss": 0.0,
+ "step": 193
+ },
+ {
+ "epoch": 2.7395555555555555,
+ "grad_norm": NaN,
+ "learning_rate": 2.681099869566328e-07,
+ "loss": 0.0,
+ "step": 194
+ },
+ {
+ "epoch": 2.7537777777777777,
+ "grad_norm": NaN,
+ "learning_rate": 2.4218247645689306e-07,
+ "loss": 0.0,
+ "step": 195
+ },
+ {
+ "epoch": 2.768,
+ "grad_norm": NaN,
+ "learning_rate": 2.175421214110235e-07,
+ "loss": 0.0,
+ "step": 196
+ },
+ {
+ "epoch": 2.7822222222222224,
+ "grad_norm": NaN,
+ "learning_rate": 1.941955878938029e-07,
+ "loss": 0.0,
+ "step": 197
+ },
+ {
+ "epoch": 2.7964444444444445,
+ "grad_norm": NaN,
+ "learning_rate": 1.721491919561913e-07,
+ "loss": 0.0,
+ "step": 198
+ },
+ {
+ "epoch": 2.8106666666666666,
+ "grad_norm": NaN,
+ "learning_rate": 1.514088979166256e-07,
+ "loss": 0.0,
+ "step": 199
+ },
+ {
+ "epoch": 2.824888888888889,
+ "grad_norm": NaN,
+ "learning_rate": 1.3198031674745814e-07,
+ "loss": 0.0,
+ "step": 200
+ },
+ {
+ "epoch": 2.824888888888889,
+ "eval_loss": NaN,
+ "eval_runtime": 5.5777,
+ "eval_samples_per_second": 35.857,
+ "eval_steps_per_second": 8.964,
+ "step": 200
+ },
+ {
+ "epoch": 2.8391111111111114,
+ "grad_norm": NaN,
+ "learning_rate": 1.138687045569975e-07,
+ "loss": 0.0,
+ "step": 201
+ },
+ {
+ "epoch": 2.8533333333333335,
+ "grad_norm": NaN,
+ "learning_rate": 9.70789611675449e-08,
+ "loss": 0.0,
+ "step": 202
+ },
+ {
+ "epoch": 2.8675555555555556,
+ "grad_norm": NaN,
+ "learning_rate": 8.161562878982399e-08,
+ "loss": 0.0,
+ "step": 203
+ },
+ {
+ "epoch": 2.8817777777777778,
+ "grad_norm": NaN,
+ "learning_rate": 6.74828907941516e-08,
+ "loss": 0.0,
+ "step": 204
+ },
+ {
+ "epoch": 2.896,
+ "grad_norm": NaN,
+ "learning_rate": 5.468457057869358e-08,
+ "loss": 0.0,
+ "step": 205
+ },
+ {
+ "epoch": 2.910222222222222,
+ "grad_norm": NaN,
+ "learning_rate": 4.322413053509944e-08,
+ "loss": 0.0,
+ "step": 206
+ },
+ {
+ "epoch": 2.924444444444444,
+ "grad_norm": NaN,
+ "learning_rate": 3.31046711118066e-08,
+ "loss": 0.0,
+ "step": 207
+ },
+ {
+ "epoch": 2.9386666666666668,
+ "grad_norm": NaN,
+ "learning_rate": 2.4328929975260262e-08,
+ "loss": 0.0,
+ "step": 208
+ },
+ {
+ "epoch": 2.952888888888889,
+ "grad_norm": NaN,
+ "learning_rate": 1.6899281269279756e-08,
+ "loss": 0.0,
+ "step": 209
+ },
+ {
+ "epoch": 2.967111111111111,
+ "grad_norm": NaN,
+ "learning_rate": 1.0817734972768946e-08,
+ "loss": 0.0,
+ "step": 210
+ },
+ {
+ "epoch": 2.981333333333333,
+ "grad_norm": NaN,
+ "learning_rate": 6.085936355947897e-09,
+ "loss": 0.0,
+ "step": 211
+ },
+ {
+ "epoch": 2.9955555555555557,
+ "grad_norm": NaN,
+ "learning_rate": 2.7051655352494654e-09,
+ "loss": 0.0,
+ "step": 212
+ },
+ {
+ "epoch": 3.0,
+ "grad_norm": NaN,
+ "learning_rate": 6.763371270035457e-10,
+ "loss": 0.0,
+ "step": 213
+ },
+ {
+ "epoch": 3.0,
+ "step": 213,
+ "total_flos": 1.0308479847759872e+17,
+ "train_loss": 1.8930653027245696,
+ "train_runtime": 5673.0769,
+ "train_samples_per_second": 9.519,
+ "train_steps_per_second": 0.038
+ }
+ ],
+ "logging_steps": 1,
+ "max_steps": 213,
+ "num_input_tokens_seen": 0,
+ "num_train_epochs": 3,
+ "save_steps": 500,
+ "stateful_callbacks": {
+ "TrainerControl": {
+ "args": {
+ "should_epoch_stop": false,
+ "should_evaluate": false,
+ "should_log": false,
+ "should_save": true,
+ "should_training_stop": true
+ },
+ "attributes": {}
+ }
+ },
+ "total_flos": 1.0308479847759872e+17,
+ "train_batch_size": 1,
+ "trial_name": null,
+ "trial_params": null
+}
diff --git a/qwen2.5-7b-instruct/training_args.bin b/qwen2.5-7b-instruct/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..b5415efad8d753f3c17d7314f34dccfa7e73e9e0
--- /dev/null
+++ b/qwen2.5-7b-instruct/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:62861a63c10b8e91d966486b438a4ca616ac892c434ce240cbba36083500e44e
+size 5649
diff --git a/qwen2.5-7b-instruct/training_eval_loss.png b/qwen2.5-7b-instruct/training_eval_loss.png
new file mode 100644
index 0000000000000000000000000000000000000000..1504f05248a069ef05c3464b7a91f99071b6cbdd
Binary files /dev/null and b/qwen2.5-7b-instruct/training_eval_loss.png differ
diff --git a/qwen2.5-7b-instruct/training_loss.png b/qwen2.5-7b-instruct/training_loss.png
new file mode 100644
index 0000000000000000000000000000000000000000..e677a3e03f895f26ee360e937862ae248c2aa53c
Binary files /dev/null and b/qwen2.5-7b-instruct/training_loss.png differ