Upload folder using huggingface_hub

.gitattributes

@@ -34,3 +34,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
 200M/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+400M/tokenizer.json filter=lfs diff=lfs merge=lfs -text

400M/chat_template.jinja

@@ -0,0 +1 @@
+{% if messages[0]['role'] == 'system' %}{% set loop_messages = messages[1:] %}{% set system_message = messages[0]['content'] %}{% else %}{% set loop_messages = messages %}{% endif %}{% if system_message is defined %}{{ system_message }}{% endif %}{% for message in loop_messages %}{% set content = message['content'] %}{% if message['role'] == 'user' %}{{ content }}{% elif message['role'] == 'assistant' %}{{ content }}{% endif %}{% endfor %}

400M/config.json

@@ -0,0 +1,170 @@
+{
+  "architectures": [
+    "Gemma4UnifiedForConditionalGeneration"
+  ],
+  "audio_config": {
+    "_name_or_path": "",
+    "architectures": null,
+    "audio_embed_dim": 640,
+    "chunk_size_feed_forward": 0,
+    "dtype": "bfloat16",
+    "id2label": {
+      "0": "LABEL_0",
+      "1": "LABEL_1"
+    },
+    "initializer_range": 0.02,
+    "is_encoder_decoder": false,
+    "label2id": {
+      "LABEL_0": 0,
+      "LABEL_1": 1
+    },
+    "model_type": "gemma4_unified_audio",
+    "output_attentions": false,
+    "output_hidden_states": false,
+    "problem_type": null,
+    "return_dict": true,
+    "rms_norm_eps": 1e-06
+  },
+  "audio_token_id": 258881,
+  "boa_token_id": 256000,
+  "boi_token_id": 255999,
+  "bos_token_id": 2,
+  "dtype": "bfloat16",
+  "eoa_token_index": 258883,
+  "eoi_token_id": 258882,
+  "eos_token_id": 1,
+  "hidden_size": 3840,
+  "image_token_id": 258880,
+  "initializer_range": 0.02,
+  "model_type": "gemma4_unified",
+  "pad_token_id": 0,
+  "text_config": {
+    "attention_bias": false,
+    "attention_dropout": 0.0,
+    "attention_k_eq_v": true,
+    "bos_token_id": 2,
+    "dtype": "bfloat16",
+    "enable_moe_block": false,
+    "eos_token_id": 1,
+    "final_logit_softcapping": 30.0,
+    "global_head_dim": 512,
+    "head_dim": 256,
+    "hidden_activation": "gelu_pytorch_tanh",
+    "hidden_size": 3840,
+    "hidden_size_per_layer_input": 0,
+    "initializer_range": 0.02,
+    "intermediate_size": 15360,
+    "layer_types": [
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "full_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "full_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "full_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "full_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "full_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "full_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "full_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "full_attention"
+    ],
+    "max_position_embeddings": 262144,
+    "model_type": "gemma4_unified_text",
+    "moe_intermediate_size": null,
+    "num_attention_heads": 16,
+    "num_experts": null,
+    "num_global_key_value_heads": 1,
+    "num_hidden_layers": 48,
+    "num_key_value_heads": 8,
+    "num_kv_shared_layers": 0,
+    "pad_token_id": 0,
+    "rms_norm_eps": 1e-06,
+    "rope_parameters": {
+      "full_attention": {
+        "partial_rotary_factor": 0.25,
+        "rope_theta": 1000000.0,
+        "rope_type": "proportional"
+      },
+      "sliding_attention": {
+        "rope_theta": 10000.0,
+        "rope_type": "default"
+      }
+    },
+    "sliding_window": 1024,
+    "tie_word_embeddings": true,
+    "top_k_experts": null,
+    "use_bidirectional_attention": "vision",
+    "use_cache": false,
+    "use_double_wide_mlp": false,
+    "vocab_size": 262144,
+    "vocab_size_per_layer_input": 262144
+  },
+  "tie_word_embeddings": true,
+  "transformers_version": "5.10.0",
+  "use_cache": false,
+  "video_token_id": 258884,
+  "vision_config": {
+    "_name_or_path": "",
+    "architectures": null,
+    "chunk_size_feed_forward": 0,
+    "dtype": "bfloat16",
+    "id2label": {
+      "0": "LABEL_0",
+      "1": "LABEL_1"
+    },
+    "initializer_range": 0.02,
+    "is_encoder_decoder": false,
+    "label2id": {
+      "LABEL_0": 0,
+      "LABEL_1": 1
+    },
+    "mm_embed_dim": 3840,
+    "mm_posemb_size": 1120,
+    "model_type": "gemma4_unified_vision",
+    "num_soft_tokens": 280,
+    "output_attentions": false,
+    "output_hidden_states": false,
+    "output_proj_dims": 3840,
+    "patch_size": 16,
+    "pooling_kernel_size": 3,
+    "problem_type": null,
+    "return_dict": true,
+    "rms_norm_eps": 1e-06
+  }
+}

400M/generation_config.json

@@ -0,0 +1,17 @@
+{
+  "bos_token_id": 2,
+  "do_sample": true,
+  "eos_token_id": [
+    1,
+    1
+  ],
+  "pad_token_id": 0,
+  "suppress_tokens": [
+    258883,
+    258882
+  ],
+  "temperature": 1.0,
+  "top_k": 64,
+  "top_p": 0.95,
+  "transformers_version": "5.10.0"
+}

400M/latest

@@ -0,0 +1 @@
+global_step4070

400M/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7787fa691cfd93da552f852c4ba2434f67b39d96c2d5da03e24e8ed72ab327aa
+size 23919549408

400M/processor_config.json

@@ -0,0 +1,62 @@
+{
+  "audio_ms_per_token": 40,
+  "audio_seq_length": 750,
+  "feature_extractor": {
+    "audio_samples_per_token": 640,
+    "feature_extractor_type": "Gemma4UnifiedAudioFeatureExtractor",
+    "feature_size": 640,
+    "padding_side": "right",
+    "padding_value": 0.0,
+    "return_attention_mask": true,
+    "sampling_rate": 16000
+  },
+  "image_processor": {
+    "do_convert_rgb": true,
+    "do_normalize": false,
+    "do_rescale": true,
+    "do_resize": true,
+    "image_mean": [
+      0.0,
+      0.0,
+      0.0
+    ],
+    "image_processor_type": "Gemma4UnifiedImageProcessor",
+    "image_std": [
+      1.0,
+      1.0,
+      1.0
+    ],
+    "max_soft_tokens": 280,
+    "patch_size": 16,
+    "pooling_kernel_size": 3,
+    "resample": 3,
+    "rescale_factor": 0.00392156862745098
+  },
+  "image_seq_length": 280,
+  "processor_class": "Gemma4UnifiedProcessor",
+  "video_processor": {
+    "do_convert_rgb": true,
+    "do_normalize": true,
+    "do_rescale": true,
+    "do_resize": true,
+    "do_sample_frames": true,
+    "image_mean": [
+      0.0,
+      0.0,
+      0.0
+    ],
+    "image_std": [
+      1.0,
+      1.0,
+      1.0
+    ],
+    "max_soft_tokens": 70,
+    "num_frames": 32,
+    "patch_size": 16,
+    "pooling_kernel_size": 3,
+    "resample": 3,
+    "rescale_factor": 0.00392156862745098,
+    "return_metadata": false,
+    "video_processor_type": "Gemma4UnifiedVideoProcessor"
+  }
+}

400M/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4d7dcf54a8dbd1fff243640ddc034149fbe7c90ed70b8a95a6bb093e989a7577
+size 14469

400M/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f2a2b0b9d330140afd742a57df45106276d425d38a78e53b0fd43dba9559f8d4
+size 14469

400M/rng_state_10.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2ce222f8f11fd62fd6cb82264500aefbc9e7ef16907ed3e4aa3a8a99c21d95d3
+size 14476

400M/rng_state_11.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d07144bfc4ed675dda70ab66923d17d6f9233560b2db09689cbe1f262c590c88
+size 14476

400M/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f5db634f6ed5151d3c898fbe3c365f950516c404bed926de16b253b26e63c7cf
+size 14469

400M/rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ad774eade44525098fff2d5910e3d16565fca4c2af29ed0bc4d9315a8ac9607a
+size 14469

400M/rng_state_4.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ebdee1b37be2238764bcfeed9dad268e938248152b0fba60eaf89adda99eac0f
+size 14469

400M/rng_state_5.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6046157bd3d69a7cf6ee508b7359b30cb71488a11b11b07a614041b06813c553
+size 14469

400M/rng_state_6.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ebc9300f101a324d0361f01aec1d57e26b57a703b4e536f4fe982c8ba57a9144
+size 14469

400M/rng_state_7.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:357df7b609d6ffd7f22b998281f7856a711ee83505e5ed2835959554b0185df7
+size 14469

400M/rng_state_8.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4fb33667d752a94ade6841f62b8b8fcc22256b4712bd568b141bc690d4a42e11
+size 14469

400M/rng_state_9.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:354a7a38e97ba11b8ad29bf4d00d74f92295b32124cff0ef46baa862aa9b0f2f
+size 14469

400M/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c373bca4c9a6ef963a59ee82f3b16a9a90b89b812375170f487261fe4db9e20a
+size 1465

400M/tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:12bac982b793c44b03d52a250a9f0d0b666813da566b910c24a6da0695fd11e6
+size 32170070

400M/tokenizer_config.json

@@ -0,0 +1,56 @@
+{
+  "audio_token": "<|audio|>",
+  "backend": "tokenizers",
+  "boa_token": "<|audio>",
+  "boi_token": "<|image>",
+  "bos_token": "<bos>",
+  "eoa_token": "<audio|>",
+  "eoc_token": "<channel|>",
+  "eoi_token": "<image|>",
+  "eos_token": "<eos>",
+  "eot_token": "<turn|>",
+  "escape_token": "<|\"|>",
+  "etc_token": "<tool_call|>",
+  "etd_token": "<tool|>",
+  "etr_token": "<tool_response|>",
+  "extra_special_tokens": [
+    "<|video|>"
+  ],
+  "image_token": "<|image|>",
+  "is_local": true,
+  "local_files_only": true,
+  "mask_token": "<mask>",
+  "model_max_length": 1000000000000000019884624838656,
+  "model_specific_special_tokens": {
+    "audio_token": "<|audio|>",
+    "boa_token": "<|audio>",
+    "boi_token": "<|image>",
+    "eoa_token": "<audio|>",
+    "eoc_token": "<channel|>",
+    "eoi_token": "<image|>",
+    "eot_token": "<turn|>",
+    "escape_token": "<|\"|>",
+    "etc_token": "<tool_call|>",
+    "etd_token": "<tool|>",
+    "etr_token": "<tool_response|>",
+    "image_token": "<|image|>",
+    "soc_token": "<|channel>",
+    "sot_token": "<|turn>",
+    "stc_token": "<|tool_call>",
+    "std_token": "<|tool>",
+    "str_token": "<|tool_response>",
+    "think_token": "<|think|>"
+  },
+  "pad_token": "<pad>",
+  "padding_side": "right",
+  "processor_class": "Gemma4UnifiedProcessor",
+  "soc_token": "<|channel>",
+  "sot_token": "<|turn>",
+  "split_special_tokens": false,
+  "stc_token": "<|tool_call>",
+  "std_token": "<|tool>",
+  "str_token": "<|tool_response>",
+  "think_token": "<|think|>",
+  "tokenizer_class": "GemmaTokenizer",
+  "unk_token": "<unk>"
+}

400M/trainer_state.json

@@ -0,0 +1,2883 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.25358255451713396,
+  "eval_steps": 500,
+  "global_step": 4070,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0006230529595015577,
+      "grad_norm": 63.15375442917419,
+      "learning_rate": 2.910112359550562e-06,
+      "loss": 1.372981071472168,
+      "step": 10
+    },
+    {
+      "epoch": 0.0012461059190031153,
+      "grad_norm": 63.39470944322245,
+      "learning_rate": 3.921348314606742e-06,
+      "loss": 0.8310403823852539,
+      "step": 20
+    },
+    {
+      "epoch": 0.001869158878504673,
+      "grad_norm": 11.96310343076372,
+      "learning_rate": 4.932584269662922e-06,
+      "loss": 0.7284352779388428,
+      "step": 30
+    },
+    {
+      "epoch": 0.0024922118380062306,
+      "grad_norm": 159.35287956464992,
+      "learning_rate": 5.943820224719102e-06,
+      "loss": 0.6918130874633789,
+      "step": 40
+    },
+    {
+      "epoch": 0.003115264797507788,
+      "grad_norm": 140.43690439070613,
+      "learning_rate": 6.955056179775282e-06,
+      "loss": 0.6914069652557373,
+      "step": 50
+    },
+    {
+      "epoch": 0.003738317757009346,
+      "grad_norm": 20.04034887568955,
+      "learning_rate": 7.966292134831462e-06,
+      "loss": 0.6684707641601563,
+      "step": 60
+    },
+    {
+      "epoch": 0.004361370716510903,
+      "grad_norm": 146.62158267929294,
+      "learning_rate": 8.977528089887641e-06,
+      "loss": 0.6722880363464355,
+      "step": 70
+    },
+    {
+      "epoch": 0.004984423676012461,
+      "grad_norm": 48.33263649592231,
+      "learning_rate": 9.988764044943822e-06,
+      "loss": 0.7312785625457764,
+      "step": 80
+    },
+    {
+      "epoch": 0.005607476635514018,
+      "grad_norm": 96.97131898292652,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 0.7157273292541504,
+      "step": 90
+    },
+    {
+      "epoch": 0.006230529595015576,
+      "grad_norm": 95.5113244321525,
+      "learning_rate": 1.2011235955056182e-05,
+      "loss": 0.7347418785095214,
+      "step": 100
+    },
+    {
+      "epoch": 0.006853582554517134,
+      "grad_norm": 17.213242453649954,
+      "learning_rate": 1.3022471910112362e-05,
+      "loss": 0.7522170066833496,
+      "step": 110
+    },
+    {
+      "epoch": 0.007476635514018692,
+      "grad_norm": 31.300913667908624,
+      "learning_rate": 1.403370786516854e-05,
+      "loss": 0.7286031246185303,
+      "step": 120
+    },
+    {
+      "epoch": 0.00809968847352025,
+      "grad_norm": 69.2417895023605,
+      "learning_rate": 1.504494382022472e-05,
+      "loss": 0.6891061305999756,
+      "step": 130
+    },
+    {
+      "epoch": 0.008722741433021807,
+      "grad_norm": 20.324536486476994,
+      "learning_rate": 1.60561797752809e-05,
+      "loss": 1.3122027397155762,
+      "step": 140
+    },
+    {
+      "epoch": 0.009345794392523364,
+      "grad_norm": 4.902224214231,
+      "learning_rate": 1.706741573033708e-05,
+      "loss": 0.7708694458007812,
+      "step": 150
+    },
+    {
+      "epoch": 0.009968847352024923,
+      "grad_norm": 14.386566577478696,
+      "learning_rate": 1.8078651685393256e-05,
+      "loss": 0.7179630279541016,
+      "step": 160
+    },
+    {
+      "epoch": 0.01059190031152648,
+      "grad_norm": 7.998488720594685,
+      "learning_rate": 1.908988764044944e-05,
+      "loss": 0.684850025177002,
+      "step": 170
+    },
+    {
+      "epoch": 0.011214953271028037,
+      "grad_norm": 10.20788593785828,
+      "learning_rate": 2e-05,
+      "loss": 0.7212257385253906,
+      "step": 180
+    },
+    {
+      "epoch": 0.011838006230529595,
+      "grad_norm": 15.093739760534373,
+      "learning_rate": 2e-05,
+      "loss": 0.7274382591247559,
+      "step": 190
+    },
+    {
+      "epoch": 0.012461059190031152,
+      "grad_norm": 7.624987181272481,
+      "learning_rate": 2e-05,
+      "loss": 0.6997462272644043,
+      "step": 200
+    },
+    {
+      "epoch": 0.013084112149532711,
+      "grad_norm": 38.20889799071737,
+      "learning_rate": 2e-05,
+      "loss": 0.8538092613220215,
+      "step": 210
+    },
+    {
+      "epoch": 0.013707165109034268,
+      "grad_norm": 5.610917778335814,
+      "learning_rate": 2e-05,
+      "loss": 0.7119334697723388,
+      "step": 220
+    },
+    {
+      "epoch": 0.014330218068535825,
+      "grad_norm": 7.071002934278071,
+      "learning_rate": 2e-05,
+      "loss": 0.6945069313049317,
+      "step": 230
+    },
+    {
+      "epoch": 0.014953271028037384,
+      "grad_norm": 1.9779972566834714,
+      "learning_rate": 2e-05,
+      "loss": 0.6839393615722656,
+      "step": 240
+    },
+    {
+      "epoch": 0.01557632398753894,
+      "grad_norm": 4.542653465163337,
+      "learning_rate": 2e-05,
+      "loss": 0.6552486419677734,
+      "step": 250
+    },
+    {
+      "epoch": 0.0161993769470405,
+      "grad_norm": 4.1326478566474645,
+      "learning_rate": 2e-05,
+      "loss": 0.7250523567199707,
+      "step": 260
+    },
+    {
+      "epoch": 0.016822429906542057,
+      "grad_norm": 2.71061736504502,
+      "learning_rate": 2e-05,
+      "loss": 0.6365773677825928,
+      "step": 270
+    },
+    {
+      "epoch": 0.017445482866043614,
+      "grad_norm": 2.4835202986671043,
+      "learning_rate": 2e-05,
+      "loss": 0.6632792472839355,
+      "step": 280
+    },
+    {
+      "epoch": 0.01806853582554517,
+      "grad_norm": 22.09800173446232,
+      "learning_rate": 2e-05,
+      "loss": 0.663812780380249,
+      "step": 290
+    },
+    {
+      "epoch": 0.018691588785046728,
+      "grad_norm": 15.489771776735218,
+      "learning_rate": 2e-05,
+      "loss": 0.650740385055542,
+      "step": 300
+    },
+    {
+      "epoch": 0.019314641744548288,
+      "grad_norm": 1.8136045854235476,
+      "learning_rate": 2e-05,
+      "loss": 0.6493139266967773,
+      "step": 310
+    },
+    {
+      "epoch": 0.019937694704049845,
+      "grad_norm": 1.4242810637346537,
+      "learning_rate": 2e-05,
+      "loss": 0.6703683376312256,
+      "step": 320
+    },
+    {
+      "epoch": 0.020560747663551402,
+      "grad_norm": 2.01645414429854,
+      "learning_rate": 2e-05,
+      "loss": 0.6692399501800537,
+      "step": 330
+    },
+    {
+      "epoch": 0.02118380062305296,
+      "grad_norm": 1.3748624267038327,
+      "learning_rate": 2e-05,
+      "loss": 0.6543716430664063,
+      "step": 340
+    },
+    {
+      "epoch": 0.021806853582554516,
+      "grad_norm": 3.7632290163066595,
+      "learning_rate": 2e-05,
+      "loss": 0.6510223388671875,
+      "step": 350
+    },
+    {
+      "epoch": 0.022429906542056073,
+      "grad_norm": 3.069062402265582,
+      "learning_rate": 2e-05,
+      "loss": 0.6401685237884521,
+      "step": 360
+    },
+    {
+      "epoch": 0.023052959501557634,
+      "grad_norm": 3.9242151311515756,
+      "learning_rate": 2e-05,
+      "loss": 0.650885534286499,
+      "step": 370
+    },
+    {
+      "epoch": 0.02367601246105919,
+      "grad_norm": 2.5645994240054146,
+      "learning_rate": 2e-05,
+      "loss": 0.6438897132873536,
+      "step": 380
+    },
+    {
+      "epoch": 0.024299065420560748,
+      "grad_norm": 3.3264331329128587,
+      "learning_rate": 2e-05,
+      "loss": 0.6713788032531738,
+      "step": 390
+    },
+    {
+      "epoch": 0.024922118380062305,
+      "grad_norm": 5.952680558627565,
+      "learning_rate": 2e-05,
+      "loss": 0.695007610321045,
+      "step": 400
+    },
+    {
+      "epoch": 0.02554517133956386,
+      "grad_norm": 7.2640539794724965,
+      "learning_rate": 2e-05,
+      "loss": 0.6785341262817383,
+      "step": 410
+    },
+    {
+      "epoch": 0.026168224299065422,
+      "grad_norm": 12.065849294819978,
+      "learning_rate": 2e-05,
+      "loss": 0.6561185836791992,
+      "step": 420
+    },
+    {
+      "epoch": 0.02679127725856698,
+      "grad_norm": 2.9882502235892647,
+      "learning_rate": 2e-05,
+      "loss": 0.6439939498901367,
+      "step": 430
+    },
+    {
+      "epoch": 0.027414330218068536,
+      "grad_norm": 2.919046299610538,
+      "learning_rate": 2e-05,
+      "loss": 0.6663602352142334,
+      "step": 440
+    },
+    {
+      "epoch": 0.028037383177570093,
+      "grad_norm": 4.189627201982762,
+      "learning_rate": 2e-05,
+      "loss": 0.6693576812744141,
+      "step": 450
+    },
+    {
+      "epoch": 0.02866043613707165,
+      "grad_norm": 7.890660674041811,
+      "learning_rate": 2e-05,
+      "loss": 0.6684383392333985,
+      "step": 460
+    },
+    {
+      "epoch": 0.029283489096573207,
+      "grad_norm": 82.91478815091692,
+      "learning_rate": 2e-05,
+      "loss": 0.7290635585784913,
+      "step": 470
+    },
+    {
+      "epoch": 0.029906542056074768,
+      "grad_norm": 3.333359402201608,
+      "learning_rate": 2e-05,
+      "loss": 0.6626537322998047,
+      "step": 480
+    },
+    {
+      "epoch": 0.030529595015576325,
+      "grad_norm": 2.4351561694941313,
+      "learning_rate": 2e-05,
+      "loss": 0.6529546260833741,
+      "step": 490
+    },
+    {
+      "epoch": 0.03115264797507788,
+      "grad_norm": 3.5168056840413233,
+      "learning_rate": 2e-05,
+      "loss": 0.6785277843475341,
+      "step": 500
+    },
+    {
+      "epoch": 0.03177570093457944,
+      "grad_norm": 64.52415387460657,
+      "learning_rate": 2e-05,
+      "loss": 0.6317290306091309,
+      "step": 510
+    },
+    {
+      "epoch": 0.032398753894081,
+      "grad_norm": 1.9567345855965799,
+      "learning_rate": 2e-05,
+      "loss": 0.6504751205444336,
+      "step": 520
+    },
+    {
+      "epoch": 0.03302180685358255,
+      "grad_norm": 2.4483916233070815,
+      "learning_rate": 2e-05,
+      "loss": 0.6714927673339843,
+      "step": 530
+    },
+    {
+      "epoch": 0.03364485981308411,
+      "grad_norm": 6.84493838576302,
+      "learning_rate": 2e-05,
+      "loss": 0.6702317714691162,
+      "step": 540
+    },
+    {
+      "epoch": 0.03426791277258567,
+      "grad_norm": 1.6302467458157637,
+      "learning_rate": 2e-05,
+      "loss": 0.6749137878417969,
+      "step": 550
+    },
+    {
+      "epoch": 0.03489096573208723,
+      "grad_norm": 1.3830723147413793,
+      "learning_rate": 2e-05,
+      "loss": 0.6726428985595703,
+      "step": 560
+    },
+    {
+      "epoch": 0.03551401869158879,
+      "grad_norm": 1.816565044561178,
+      "learning_rate": 2e-05,
+      "loss": 0.6633450984954834,
+      "step": 570
+    },
+    {
+      "epoch": 0.03613707165109034,
+      "grad_norm": 1.4154803760531598,
+      "learning_rate": 2e-05,
+      "loss": 0.6670496940612793,
+      "step": 580
+    },
+    {
+      "epoch": 0.0367601246105919,
+      "grad_norm": 1.441388864695392,
+      "learning_rate": 2e-05,
+      "loss": 0.66356520652771,
+      "step": 590
+    },
+    {
+      "epoch": 0.037383177570093455,
+      "grad_norm": 5.297501644827849,
+      "learning_rate": 2e-05,
+      "loss": 0.6568848609924316,
+      "step": 600
+    },
+    {
+      "epoch": 0.038006230529595016,
+      "grad_norm": 1.3268959282636765,
+      "learning_rate": 2e-05,
+      "loss": 0.6364224433898926,
+      "step": 610
+    },
+    {
+      "epoch": 0.038629283489096576,
+      "grad_norm": 3.1357427218379725,
+      "learning_rate": 2e-05,
+      "loss": 0.6608361721038818,
+      "step": 620
+    },
+    {
+      "epoch": 0.03925233644859813,
+      "grad_norm": 3.2870153380544496,
+      "learning_rate": 2e-05,
+      "loss": 0.7057211875915528,
+      "step": 630
+    },
+    {
+      "epoch": 0.03987538940809969,
+      "grad_norm": 2.197431012725625,
+      "learning_rate": 2e-05,
+      "loss": 0.6897117614746093,
+      "step": 640
+    },
+    {
+      "epoch": 0.040498442367601244,
+      "grad_norm": 8.502549125517568,
+      "learning_rate": 2e-05,
+      "loss": 0.6920805931091308,
+      "step": 650
+    },
+    {
+      "epoch": 0.041121495327102804,
+      "grad_norm": 2.462320802295259,
+      "learning_rate": 2e-05,
+      "loss": 0.7063193321228027,
+      "step": 660
+    },
+    {
+      "epoch": 0.041744548286604365,
+      "grad_norm": 5.683201827218238,
+      "learning_rate": 2e-05,
+      "loss": 0.6920748710632324,
+      "step": 670
+    },
+    {
+      "epoch": 0.04236760124610592,
+      "grad_norm": 3.027115327083748,
+      "learning_rate": 2e-05,
+      "loss": 0.6580988883972168,
+      "step": 680
+    },
+    {
+      "epoch": 0.04299065420560748,
+      "grad_norm": 9.069631138625143,
+      "learning_rate": 2e-05,
+      "loss": 0.6772819519042969,
+      "step": 690
+    },
+    {
+      "epoch": 0.04361370716510903,
+      "grad_norm": 3.5222303907627026,
+      "learning_rate": 2e-05,
+      "loss": 0.6700127601623536,
+      "step": 700
+    },
+    {
+      "epoch": 0.04423676012461059,
+      "grad_norm": 2.539625946046831,
+      "learning_rate": 2e-05,
+      "loss": 0.6825683116912842,
+      "step": 710
+    },
+    {
+      "epoch": 0.044859813084112146,
+      "grad_norm": 7.530973621569606,
+      "learning_rate": 2e-05,
+      "loss": 0.6668391227722168,
+      "step": 720
+    },
+    {
+      "epoch": 0.04548286604361371,
+      "grad_norm": 3.10826200653366,
+      "learning_rate": 2e-05,
+      "loss": 0.6900657176971435,
+      "step": 730
+    },
+    {
+      "epoch": 0.04610591900311527,
+      "grad_norm": 3.021206626575616,
+      "learning_rate": 2e-05,
+      "loss": 0.666568374633789,
+      "step": 740
+    },
+    {
+      "epoch": 0.04672897196261682,
+      "grad_norm": 3.4063444253551793,
+      "learning_rate": 2e-05,
+      "loss": 0.6786983489990235,
+      "step": 750
+    },
+    {
+      "epoch": 0.04735202492211838,
+      "grad_norm": 5.674285755423112,
+      "learning_rate": 2e-05,
+      "loss": 0.6631364822387695,
+      "step": 760
+    },
+    {
+      "epoch": 0.047975077881619935,
+      "grad_norm": 5.447979039303706,
+      "learning_rate": 2e-05,
+      "loss": 0.6851268768310547,
+      "step": 770
+    },
+    {
+      "epoch": 0.048598130841121495,
+      "grad_norm": 2.0548995784325275,
+      "learning_rate": 2e-05,
+      "loss": 0.6662421703338623,
+      "step": 780
+    },
+    {
+      "epoch": 0.049221183800623056,
+      "grad_norm": 1.2276649895732483,
+      "learning_rate": 2e-05,
+      "loss": 0.6671768188476562,
+      "step": 790
+    },
+    {
+      "epoch": 0.04984423676012461,
+      "grad_norm": 1.5987199277283939,
+      "learning_rate": 2e-05,
+      "loss": 0.6489299774169922,
+      "step": 800
+    },
+    {
+      "epoch": 0.05046728971962617,
+      "grad_norm": 22.9031846877308,
+      "learning_rate": 2e-05,
+      "loss": 0.6675164699554443,
+      "step": 810
+    },
+    {
+      "epoch": 0.05109034267912772,
+      "grad_norm": 2.9034364102338985,
+      "learning_rate": 2e-05,
+      "loss": 0.6826793670654296,
+      "step": 820
+    },
+    {
+      "epoch": 0.051713395638629284,
+      "grad_norm": 1.8432534502921918,
+      "learning_rate": 2e-05,
+      "loss": 0.6763291835784913,
+      "step": 830
+    },
+    {
+      "epoch": 0.052336448598130844,
+      "grad_norm": 2.0417799918298596,
+      "learning_rate": 2e-05,
+      "loss": 0.6497424125671387,
+      "step": 840
+    },
+    {
+      "epoch": 0.0529595015576324,
+      "grad_norm": 22.32540564160995,
+      "learning_rate": 2e-05,
+      "loss": 0.6354857444763183,
+      "step": 850
+    },
+    {
+      "epoch": 0.05358255451713396,
+      "grad_norm": 6.229762616389806,
+      "learning_rate": 2e-05,
+      "loss": 0.657370662689209,
+      "step": 860
+    },
+    {
+      "epoch": 0.05420560747663551,
+      "grad_norm": 3.23998905516123,
+      "learning_rate": 2e-05,
+      "loss": 0.6524142265319824,
+      "step": 870
+    },
+    {
+      "epoch": 0.05482866043613707,
+      "grad_norm": 2.8784021094456604,
+      "learning_rate": 2e-05,
+      "loss": 0.653789472579956,
+      "step": 880
+    },
+    {
+      "epoch": 0.05545171339563863,
+      "grad_norm": 1.40857663500608,
+      "learning_rate": 2e-05,
+      "loss": 0.6703225612640381,
+      "step": 890
+    },
+    {
+      "epoch": 0.056074766355140186,
+      "grad_norm": 5.080917667561178,
+      "learning_rate": 2e-05,
+      "loss": 0.6439192771911622,
+      "step": 900
+    },
+    {
+      "epoch": 0.05669781931464175,
+      "grad_norm": 4.5764665849587445,
+      "learning_rate": 2e-05,
+      "loss": 0.6634146690368652,
+      "step": 910
+    },
+    {
+      "epoch": 0.0573208722741433,
+      "grad_norm": 3.2037710370417023,
+      "learning_rate": 2e-05,
+      "loss": 0.6824374198913574,
+      "step": 920
+    },
+    {
+      "epoch": 0.05794392523364486,
+      "grad_norm": 6.053033149131505,
+      "learning_rate": 2e-05,
+      "loss": 0.6423258781433105,
+      "step": 930
+    },
+    {
+      "epoch": 0.058566978193146414,
+      "grad_norm": 13.436486964861611,
+      "learning_rate": 2e-05,
+      "loss": 0.6521054744720459,
+      "step": 940
+    },
+    {
+      "epoch": 0.059190031152647975,
+      "grad_norm": 1.4820242941537698,
+      "learning_rate": 2e-05,
+      "loss": 0.6490562438964844,
+      "step": 950
+    },
+    {
+      "epoch": 0.059813084112149535,
+      "grad_norm": 3.970984796336633,
+      "learning_rate": 2e-05,
+      "loss": 0.6425233364105225,
+      "step": 960
+    },
+    {
+      "epoch": 0.06043613707165109,
+      "grad_norm": 2.606688835969579,
+      "learning_rate": 2e-05,
+      "loss": 0.6780929088592529,
+      "step": 970
+    },
+    {
+      "epoch": 0.06105919003115265,
+      "grad_norm": 1.6441407042964533,
+      "learning_rate": 2e-05,
+      "loss": 0.6381833553314209,
+      "step": 980
+    },
+    {
+      "epoch": 0.0616822429906542,
+      "grad_norm": 1.9338014114509088,
+      "learning_rate": 2e-05,
+      "loss": 0.671638298034668,
+      "step": 990
+    },
+    {
+      "epoch": 0.06230529595015576,
+      "grad_norm": 2.39864419114088,
+      "learning_rate": 2e-05,
+      "loss": 0.659054183959961,
+      "step": 1000
+    },
+    {
+      "epoch": 0.06292834890965732,
+      "grad_norm": 4.683446585267705,
+      "learning_rate": 2e-05,
+      "loss": 0.6786028861999511,
+      "step": 1010
+    },
+    {
+      "epoch": 0.06355140186915888,
+      "grad_norm": 2.719894383949491,
+      "learning_rate": 2e-05,
+      "loss": 0.6658808708190918,
+      "step": 1020
+    },
+    {
+      "epoch": 0.06417445482866044,
+      "grad_norm": 1.1588652494157246,
+      "learning_rate": 2e-05,
+      "loss": 0.6696397304534912,
+      "step": 1030
+    },
+    {
+      "epoch": 0.064797507788162,
+      "grad_norm": 1.0813640353054965,
+      "learning_rate": 2e-05,
+      "loss": 0.6676293849945069,
+      "step": 1040
+    },
+    {
+      "epoch": 0.06542056074766354,
+      "grad_norm": 1.2271064002015928,
+      "learning_rate": 2e-05,
+      "loss": 0.6389595985412597,
+      "step": 1050
+    },
+    {
+      "epoch": 0.0660436137071651,
+      "grad_norm": 2.8431458204110074,
+      "learning_rate": 2e-05,
+      "loss": 0.6555405616760254,
+      "step": 1060
+    },
+    {
+      "epoch": 0.06666666666666667,
+      "grad_norm": 3.2188265816780692,
+      "learning_rate": 2e-05,
+      "loss": 0.6418774604797364,
+      "step": 1070
+    },
+    {
+      "epoch": 0.06728971962616823,
+      "grad_norm": 9.465426552560904,
+      "learning_rate": 2e-05,
+      "loss": 0.6477942943572998,
+      "step": 1080
+    },
+    {
+      "epoch": 0.06791277258566979,
+      "grad_norm": 1.1426314291875073,
+      "learning_rate": 2e-05,
+      "loss": 0.6486814022064209,
+      "step": 1090
+    },
+    {
+      "epoch": 0.06853582554517133,
+      "grad_norm": 1.3614517863008955,
+      "learning_rate": 2e-05,
+      "loss": 0.6399056911468506,
+      "step": 1100
+    },
+    {
+      "epoch": 0.0691588785046729,
+      "grad_norm": 1.616958777603617,
+      "learning_rate": 2e-05,
+      "loss": 0.625498104095459,
+      "step": 1110
+    },
+    {
+      "epoch": 0.06978193146417445,
+      "grad_norm": 1.700623675066532,
+      "learning_rate": 2e-05,
+      "loss": 0.620833158493042,
+      "step": 1120
+    },
+    {
+      "epoch": 0.07040498442367601,
+      "grad_norm": 2.5588029176909948,
+      "learning_rate": 2e-05,
+      "loss": 0.6351099967956543,
+      "step": 1130
+    },
+    {
+      "epoch": 0.07102803738317758,
+      "grad_norm": 6.556254605023658,
+      "learning_rate": 2e-05,
+      "loss": 0.6419850826263428,
+      "step": 1140
+    },
+    {
+      "epoch": 0.07165109034267912,
+      "grad_norm": 2.4751544762969946,
+      "learning_rate": 2e-05,
+      "loss": 0.6722092628479004,
+      "step": 1150
+    },
+    {
+      "epoch": 0.07227414330218068,
+      "grad_norm": 1.5727047413222857,
+      "learning_rate": 2e-05,
+      "loss": 0.6558211326599122,
+      "step": 1160
+    },
+    {
+      "epoch": 0.07289719626168224,
+      "grad_norm": 9.438322770095112,
+      "learning_rate": 2e-05,
+      "loss": 0.6519765853881836,
+      "step": 1170
+    },
+    {
+      "epoch": 0.0735202492211838,
+      "grad_norm": 2.839029932122009,
+      "learning_rate": 2e-05,
+      "loss": 0.6472910881042481,
+      "step": 1180
+    },
+    {
+      "epoch": 0.07414330218068536,
+      "grad_norm": 1.3804888984788395,
+      "learning_rate": 2e-05,
+      "loss": 0.6604276180267334,
+      "step": 1190
+    },
+    {
+      "epoch": 0.07476635514018691,
+      "grad_norm": 1.0934274555239194,
+      "learning_rate": 2e-05,
+      "loss": 0.6465075969696045,
+      "step": 1200
+    },
+    {
+      "epoch": 0.07538940809968847,
+      "grad_norm": 1.8841069578655136,
+      "learning_rate": 2e-05,
+      "loss": 0.6400872230529785,
+      "step": 1210
+    },
+    {
+      "epoch": 0.07601246105919003,
+      "grad_norm": 16.12011257195063,
+      "learning_rate": 2e-05,
+      "loss": 0.655942440032959,
+      "step": 1220
+    },
+    {
+      "epoch": 0.07663551401869159,
+      "grad_norm": 1.9392045965213456,
+      "learning_rate": 2e-05,
+      "loss": 0.6418064117431641,
+      "step": 1230
+    },
+    {
+      "epoch": 0.07725856697819315,
+      "grad_norm": 1.2879388752480592,
+      "learning_rate": 2e-05,
+      "loss": 0.6534547328948974,
+      "step": 1240
+    },
+    {
+      "epoch": 0.0778816199376947,
+      "grad_norm": 1.3491269021694952,
+      "learning_rate": 2e-05,
+      "loss": 0.6325421333312988,
+      "step": 1250
+    },
+    {
+      "epoch": 0.07850467289719626,
+      "grad_norm": 1.9538179100083624,
+      "learning_rate": 2e-05,
+      "loss": 0.6555644989013671,
+      "step": 1260
+    },
+    {
+      "epoch": 0.07912772585669782,
+      "grad_norm": 1.686205721804821,
+      "learning_rate": 2e-05,
+      "loss": 0.6385934352874756,
+      "step": 1270
+    },
+    {
+      "epoch": 0.07975077881619938,
+      "grad_norm": 1.4326411502688534,
+      "learning_rate": 2e-05,
+      "loss": 0.6342977046966553,
+      "step": 1280
+    },
+    {
+      "epoch": 0.08037383177570094,
+      "grad_norm": 4.860024808022459,
+      "learning_rate": 2e-05,
+      "loss": 0.6609707832336426,
+      "step": 1290
+    },
+    {
+      "epoch": 0.08099688473520249,
+      "grad_norm": 1.2371686758942715,
+      "learning_rate": 2e-05,
+      "loss": 0.6476799964904785,
+      "step": 1300
+    },
+    {
+      "epoch": 0.08161993769470405,
+      "grad_norm": 1.1597203515801051,
+      "learning_rate": 2e-05,
+      "loss": 0.6473023891448975,
+      "step": 1310
+    },
+    {
+      "epoch": 0.08224299065420561,
+      "grad_norm": 1.8273978345410087,
+      "learning_rate": 2e-05,
+      "loss": 0.6425920009613038,
+      "step": 1320
+    },
+    {
+      "epoch": 0.08286604361370717,
+      "grad_norm": 1.6088638146203655,
+      "learning_rate": 2e-05,
+      "loss": 0.6325952053070069,
+      "step": 1330
+    },
+    {
+      "epoch": 0.08348909657320873,
+      "grad_norm": 1.0395526162178266,
+      "learning_rate": 2e-05,
+      "loss": 0.6280010223388672,
+      "step": 1340
+    },
+    {
+      "epoch": 0.08411214953271028,
+      "grad_norm": 1.1776111039430426,
+      "learning_rate": 2e-05,
+      "loss": 0.6420888900756836,
+      "step": 1350
+    },
+    {
+      "epoch": 0.08473520249221184,
+      "grad_norm": 1.144362156326936,
+      "learning_rate": 2e-05,
+      "loss": 0.6447340965270996,
+      "step": 1360
+    },
+    {
+      "epoch": 0.0853582554517134,
+      "grad_norm": 1.2899619747320272,
+      "learning_rate": 2e-05,
+      "loss": 0.6319891929626464,
+      "step": 1370
+    },
+    {
+      "epoch": 0.08598130841121496,
+      "grad_norm": 1.1503180888416502,
+      "learning_rate": 2e-05,
+      "loss": 0.6263866424560547,
+      "step": 1380
+    },
+    {
+      "epoch": 0.08660436137071652,
+      "grad_norm": 0.9492449615285058,
+      "learning_rate": 2e-05,
+      "loss": 0.6156892776489258,
+      "step": 1390
+    },
+    {
+      "epoch": 0.08722741433021806,
+      "grad_norm": 0.9840501517202644,
+      "learning_rate": 2e-05,
+      "loss": 0.6201627731323243,
+      "step": 1400
+    },
+    {
+      "epoch": 0.08785046728971962,
+      "grad_norm": 0.9663965412569069,
+      "learning_rate": 2e-05,
+      "loss": 0.618841552734375,
+      "step": 1410
+    },
+    {
+      "epoch": 0.08847352024922119,
+      "grad_norm": 1.2024904063187882,
+      "learning_rate": 2e-05,
+      "loss": 0.6406625270843506,
+      "step": 1420
+    },
+    {
+      "epoch": 0.08909657320872275,
+      "grad_norm": 0.9573647223711624,
+      "learning_rate": 2e-05,
+      "loss": 0.6550570487976074,
+      "step": 1430
+    },
+    {
+      "epoch": 0.08971962616822429,
+      "grad_norm": 1.0307197954823704,
+      "learning_rate": 2e-05,
+      "loss": 0.6338574886322021,
+      "step": 1440
+    },
+    {
+      "epoch": 0.09034267912772585,
+      "grad_norm": 2.305281565846579,
+      "learning_rate": 2e-05,
+      "loss": 0.6363538265228271,
+      "step": 1450
+    },
+    {
+      "epoch": 0.09096573208722741,
+      "grad_norm": 1.4667417153142859,
+      "learning_rate": 2e-05,
+      "loss": 0.6235406875610352,
+      "step": 1460
+    },
+    {
+      "epoch": 0.09158878504672897,
+      "grad_norm": 2.9584530551772437,
+      "learning_rate": 2e-05,
+      "loss": 0.6403841018676758,
+      "step": 1470
+    },
+    {
+      "epoch": 0.09221183800623053,
+      "grad_norm": 1.112253956223209,
+      "learning_rate": 2e-05,
+      "loss": 0.6414080619812011,
+      "step": 1480
+    },
+    {
+      "epoch": 0.09283489096573208,
+      "grad_norm": 1.197396557670252,
+      "learning_rate": 2e-05,
+      "loss": 0.6136775970458984,
+      "step": 1490
+    },
+    {
+      "epoch": 0.09345794392523364,
+      "grad_norm": 0.930725260856421,
+      "learning_rate": 2e-05,
+      "loss": 0.6281547546386719,
+      "step": 1500
+    },
+    {
+      "epoch": 0.0940809968847352,
+      "grad_norm": 2.246282067394849,
+      "learning_rate": 2e-05,
+      "loss": 0.6232186317443847,
+      "step": 1510
+    },
+    {
+      "epoch": 0.09470404984423676,
+      "grad_norm": 1.811695916729897,
+      "learning_rate": 2e-05,
+      "loss": 0.6236707210540772,
+      "step": 1520
+    },
+    {
+      "epoch": 0.09532710280373832,
+      "grad_norm": 1.8453018493553404,
+      "learning_rate": 2e-05,
+      "loss": 0.6396072387695313,
+      "step": 1530
+    },
+    {
+      "epoch": 0.09595015576323987,
+      "grad_norm": 1.4810091845213549,
+      "learning_rate": 2e-05,
+      "loss": 0.6346172332763672,
+      "step": 1540
+    },
+    {
+      "epoch": 0.09657320872274143,
+      "grad_norm": 1.1558342150707985,
+      "learning_rate": 2e-05,
+      "loss": 0.6340350151062012,
+      "step": 1550
+    },
+    {
+      "epoch": 0.09719626168224299,
+      "grad_norm": 1.1277930144534245,
+      "learning_rate": 2e-05,
+      "loss": 0.6178334712982178,
+      "step": 1560
+    },
+    {
+      "epoch": 0.09781931464174455,
+      "grad_norm": 1.6697265802855836,
+      "learning_rate": 2e-05,
+      "loss": 0.6273056030273437,
+      "step": 1570
+    },
+    {
+      "epoch": 0.09844236760124611,
+      "grad_norm": 1.1996090977492178,
+      "learning_rate": 2e-05,
+      "loss": 0.6388472557067871,
+      "step": 1580
+    },
+    {
+      "epoch": 0.09906542056074766,
+      "grad_norm": 4.193851985080715,
+      "learning_rate": 2e-05,
+      "loss": 0.6388599395751953,
+      "step": 1590
+    },
+    {
+      "epoch": 0.09968847352024922,
+      "grad_norm": 1.3073390628519994,
+      "learning_rate": 2e-05,
+      "loss": 0.6429434776306152,
+      "step": 1600
+    },
+    {
+      "epoch": 0.10031152647975078,
+      "grad_norm": 1.5006350300856774,
+      "learning_rate": 2e-05,
+      "loss": 0.6440896034240723,
+      "step": 1610
+    },
+    {
+      "epoch": 0.10093457943925234,
+      "grad_norm": 1.2275982378982346,
+      "learning_rate": 2e-05,
+      "loss": 0.6285616874694824,
+      "step": 1620
+    },
+    {
+      "epoch": 0.1015576323987539,
+      "grad_norm": 1.1626753929589895,
+      "learning_rate": 2e-05,
+      "loss": 0.6467322826385498,
+      "step": 1630
+    },
+    {
+      "epoch": 0.10218068535825545,
+      "grad_norm": 1.8546526164062445,
+      "learning_rate": 2e-05,
+      "loss": 0.6216934204101563,
+      "step": 1640
+    },
+    {
+      "epoch": 0.102803738317757,
+      "grad_norm": 4.204227006721846,
+      "learning_rate": 2e-05,
+      "loss": 0.6037077903747559,
+      "step": 1650
+    },
+    {
+      "epoch": 0.10342679127725857,
+      "grad_norm": 2.4575617089230892,
+      "learning_rate": 2e-05,
+      "loss": 0.6549998760223389,
+      "step": 1660
+    },
+    {
+      "epoch": 0.10404984423676013,
+      "grad_norm": 0.9831225747692173,
+      "learning_rate": 2e-05,
+      "loss": 0.6362233638763428,
+      "step": 1670
+    },
+    {
+      "epoch": 0.10467289719626169,
+      "grad_norm": 37.336116705288426,
+      "learning_rate": 2e-05,
+      "loss": 0.6452502250671387,
+      "step": 1680
+    },
+    {
+      "epoch": 0.10529595015576323,
+      "grad_norm": 10.151830562748597,
+      "learning_rate": 2e-05,
+      "loss": 0.6224545478820801,
+      "step": 1690
+    },
+    {
+      "epoch": 0.1059190031152648,
+      "grad_norm": 1.0474870590574359,
+      "learning_rate": 2e-05,
+      "loss": 0.6610437393188476,
+      "step": 1700
+    },
+    {
+      "epoch": 0.10654205607476636,
+      "grad_norm": 3.461782785767983,
+      "learning_rate": 2e-05,
+      "loss": 0.62818922996521,
+      "step": 1710
+    },
+    {
+      "epoch": 0.10716510903426792,
+      "grad_norm": 1.0436258381858183,
+      "learning_rate": 2e-05,
+      "loss": 0.6486543178558349,
+      "step": 1720
+    },
+    {
+      "epoch": 0.10778816199376948,
+      "grad_norm": 3.870674655703194,
+      "learning_rate": 2e-05,
+      "loss": 0.6329143524169922,
+      "step": 1730
+    },
+    {
+      "epoch": 0.10841121495327102,
+      "grad_norm": 0.8873679451298351,
+      "learning_rate": 2e-05,
+      "loss": 0.6266915798187256,
+      "step": 1740
+    },
+    {
+      "epoch": 0.10903426791277258,
+      "grad_norm": 1.934393771485732,
+      "learning_rate": 2e-05,
+      "loss": 0.6206174373626709,
+      "step": 1750
+    },
+    {
+      "epoch": 0.10965732087227414,
+      "grad_norm": 0.9788497030191609,
+      "learning_rate": 2e-05,
+      "loss": 0.617664098739624,
+      "step": 1760
+    },
+    {
+      "epoch": 0.1102803738317757,
+      "grad_norm": 1.3294647839474172,
+      "learning_rate": 2e-05,
+      "loss": 0.6125317573547363,
+      "step": 1770
+    },
+    {
+      "epoch": 0.11090342679127727,
+      "grad_norm": 0.9114337186956374,
+      "learning_rate": 2e-05,
+      "loss": 0.6136180877685546,
+      "step": 1780
+    },
+    {
+      "epoch": 0.11152647975077881,
+      "grad_norm": 0.9168910070389206,
+      "learning_rate": 2e-05,
+      "loss": 0.6254988670349121,
+      "step": 1790
+    },
+    {
+      "epoch": 0.11214953271028037,
+      "grad_norm": 1.2163942422286602,
+      "learning_rate": 2e-05,
+      "loss": 0.6339495658874512,
+      "step": 1800
+    },
+    {
+      "epoch": 0.11277258566978193,
+      "grad_norm": 1.0031026506964655,
+      "learning_rate": 2e-05,
+      "loss": 0.6163856506347656,
+      "step": 1810
+    },
+    {
+      "epoch": 0.1133956386292835,
+      "grad_norm": 1.4487390671485545,
+      "learning_rate": 2e-05,
+      "loss": 0.6172023296356202,
+      "step": 1820
+    },
+    {
+      "epoch": 0.11401869158878504,
+      "grad_norm": 0.971335799491775,
+      "learning_rate": 2e-05,
+      "loss": 0.603542423248291,
+      "step": 1830
+    },
+    {
+      "epoch": 0.1146417445482866,
+      "grad_norm": 1.0645808741308516,
+      "learning_rate": 2e-05,
+      "loss": 0.6314529418945313,
+      "step": 1840
+    },
+    {
+      "epoch": 0.11526479750778816,
+      "grad_norm": 1.1564622773995759,
+      "learning_rate": 2e-05,
+      "loss": 0.6358022212982177,
+      "step": 1850
+    },
+    {
+      "epoch": 0.11588785046728972,
+      "grad_norm": 6.250954738770854,
+      "learning_rate": 2e-05,
+      "loss": 0.6266725063323975,
+      "step": 1860
+    },
+    {
+      "epoch": 0.11651090342679128,
+      "grad_norm": 2.485950620772887,
+      "learning_rate": 2e-05,
+      "loss": 0.6347331047058106,
+      "step": 1870
+    },
+    {
+      "epoch": 0.11713395638629283,
+      "grad_norm": 1.0491315047430319,
+      "learning_rate": 2e-05,
+      "loss": 0.6353681087493896,
+      "step": 1880
+    },
+    {
+      "epoch": 0.11775700934579439,
+      "grad_norm": 0.9344118553673864,
+      "learning_rate": 2e-05,
+      "loss": 0.6243095397949219,
+      "step": 1890
+    },
+    {
+      "epoch": 0.11838006230529595,
+      "grad_norm": 1.166751507604473,
+      "learning_rate": 2e-05,
+      "loss": 0.624453353881836,
+      "step": 1900
+    },
+    {
+      "epoch": 0.11900311526479751,
+      "grad_norm": 1.1609173008743447,
+      "learning_rate": 2e-05,
+      "loss": 0.6156741142272949,
+      "step": 1910
+    },
+    {
+      "epoch": 0.11962616822429907,
+      "grad_norm": 1.2650824767917728,
+      "learning_rate": 2e-05,
+      "loss": 0.6225091457366944,
+      "step": 1920
+    },
+    {
+      "epoch": 0.12024922118380062,
+      "grad_norm": 1.323039674561904,
+      "learning_rate": 2e-05,
+      "loss": 0.6126754760742188,
+      "step": 1930
+    },
+    {
+      "epoch": 0.12087227414330218,
+      "grad_norm": 10.576149324131405,
+      "learning_rate": 2e-05,
+      "loss": 0.6002795219421386,
+      "step": 1940
+    },
+    {
+      "epoch": 0.12149532710280374,
+      "grad_norm": 1.1408899621175108,
+      "learning_rate": 2e-05,
+      "loss": 0.6268309593200684,
+      "step": 1950
+    },
+    {
+      "epoch": 0.1221183800623053,
+      "grad_norm": 1.04557318312074,
+      "learning_rate": 2e-05,
+      "loss": 0.6239619255065918,
+      "step": 1960
+    },
+    {
+      "epoch": 0.12274143302180686,
+      "grad_norm": 1.2159263913891547,
+      "learning_rate": 2e-05,
+      "loss": 0.6297917842864991,
+      "step": 1970
+    },
+    {
+      "epoch": 0.1233644859813084,
+      "grad_norm": 2.784941022644952,
+      "learning_rate": 2e-05,
+      "loss": 0.6152359008789062,
+      "step": 1980
+    },
+    {
+      "epoch": 0.12398753894080997,
+      "grad_norm": 0.9318872113204291,
+      "learning_rate": 2e-05,
+      "loss": 0.6123355388641357,
+      "step": 1990
+    },
+    {
+      "epoch": 0.12461059190031153,
+      "grad_norm": 1.6289119810248258,
+      "learning_rate": 2e-05,
+      "loss": 0.6148693084716796,
+      "step": 2000
+    },
+    {
+      "epoch": 0.1252336448598131,
+      "grad_norm": 6.458481915206926,
+      "learning_rate": 2e-05,
+      "loss": 0.6158293724060059,
+      "step": 2010
+    },
+    {
+      "epoch": 0.12585669781931463,
+      "grad_norm": 1.9163812666195066,
+      "learning_rate": 2e-05,
+      "loss": 0.6355224132537842,
+      "step": 2020
+    },
+    {
+      "epoch": 0.1264797507788162,
+      "grad_norm": 5.359098164590273,
+      "learning_rate": 2e-05,
+      "loss": 0.6285907745361328,
+      "step": 2030
+    },
+    {
+      "epoch": 0.12710280373831775,
+      "grad_norm": 1.1658843339568825,
+      "learning_rate": 2e-05,
+      "loss": 0.6075748443603516,
+      "step": 2040
+    },
+    {
+      "epoch": 0.1277258566978193,
+      "grad_norm": 1.0287109496616869,
+      "learning_rate": 2e-05,
+      "loss": 0.6184503078460694,
+      "step": 2050
+    },
+    {
+      "epoch": 0.12834890965732088,
+      "grad_norm": 1.6661818568533469,
+      "learning_rate": 2e-05,
+      "loss": 0.6156071662902832,
+      "step": 2060
+    },
+    {
+      "epoch": 0.12897196261682242,
+      "grad_norm": 2.7553866859395897,
+      "learning_rate": 2e-05,
+      "loss": 0.6271682262420655,
+      "step": 2070
+    },
+    {
+      "epoch": 0.129595015576324,
+      "grad_norm": 1.8819696133295345,
+      "learning_rate": 2e-05,
+      "loss": 0.6274081707000733,
+      "step": 2080
+    },
+    {
+      "epoch": 0.13021806853582554,
+      "grad_norm": 1.8333992311174614,
+      "learning_rate": 2e-05,
+      "loss": 0.6532745361328125,
+      "step": 2090
+    },
+    {
+      "epoch": 0.1308411214953271,
+      "grad_norm": 3.989954274295162,
+      "learning_rate": 2e-05,
+      "loss": 0.6255852699279785,
+      "step": 2100
+    },
+    {
+      "epoch": 0.13146417445482866,
+      "grad_norm": 4.791554748258434,
+      "learning_rate": 2e-05,
+      "loss": 0.6459884643554688,
+      "step": 2110
+    },
+    {
+      "epoch": 0.1320872274143302,
+      "grad_norm": 3.122715592121081,
+      "learning_rate": 2e-05,
+      "loss": 0.6504601955413818,
+      "step": 2120
+    },
+    {
+      "epoch": 0.13271028037383178,
+      "grad_norm": 1.3817520678395894,
+      "learning_rate": 2e-05,
+      "loss": 0.6090775966644287,
+      "step": 2130
+    },
+    {
+      "epoch": 0.13333333333333333,
+      "grad_norm": 1.5428641889518173,
+      "learning_rate": 2e-05,
+      "loss": 0.6260268211364746,
+      "step": 2140
+    },
+    {
+      "epoch": 0.13395638629283488,
+      "grad_norm": 1.0628268769141096,
+      "learning_rate": 2e-05,
+      "loss": 0.6195857048034668,
+      "step": 2150
+    },
+    {
+      "epoch": 0.13457943925233645,
+      "grad_norm": 1.8898718261172782,
+      "learning_rate": 2e-05,
+      "loss": 0.6365998744964599,
+      "step": 2160
+    },
+    {
+      "epoch": 0.135202492211838,
+      "grad_norm": 1.1331876920736932,
+      "learning_rate": 2e-05,
+      "loss": 0.6129371643066406,
+      "step": 2170
+    },
+    {
+      "epoch": 0.13582554517133957,
+      "grad_norm": 1.000166605046901,
+      "learning_rate": 2e-05,
+      "loss": 0.6213845252990723,
+      "step": 2180
+    },
+    {
+      "epoch": 0.13644859813084112,
+      "grad_norm": 1.0776344449307316,
+      "learning_rate": 2e-05,
+      "loss": 0.6136645317077637,
+      "step": 2190
+    },
+    {
+      "epoch": 0.13707165109034267,
+      "grad_norm": 0.9643079095713766,
+      "learning_rate": 2e-05,
+      "loss": 0.6012526512145996,
+      "step": 2200
+    },
+    {
+      "epoch": 0.13769470404984424,
+      "grad_norm": 0.8630389459211013,
+      "learning_rate": 2e-05,
+      "loss": 0.6241889476776123,
+      "step": 2210
+    },
+    {
+      "epoch": 0.1383177570093458,
+      "grad_norm": 0.9201314480587546,
+      "learning_rate": 2e-05,
+      "loss": 0.6349925994873047,
+      "step": 2220
+    },
+    {
+      "epoch": 0.13894080996884736,
+      "grad_norm": 1.1117228903286944,
+      "learning_rate": 2e-05,
+      "loss": 0.6251606941223145,
+      "step": 2230
+    },
+    {
+      "epoch": 0.1395638629283489,
+      "grad_norm": 1.4765753842124631,
+      "learning_rate": 2e-05,
+      "loss": 0.643409252166748,
+      "step": 2240
+    },
+    {
+      "epoch": 0.14018691588785046,
+      "grad_norm": 1.5194553627864302,
+      "learning_rate": 2e-05,
+      "loss": 0.6377650737762451,
+      "step": 2250
+    },
+    {
+      "epoch": 0.14080996884735203,
+      "grad_norm": 0.9307058039922883,
+      "learning_rate": 2e-05,
+      "loss": 0.6198421478271484,
+      "step": 2260
+    },
+    {
+      "epoch": 0.14143302180685358,
+      "grad_norm": 0.8248507354046317,
+      "learning_rate": 2e-05,
+      "loss": 0.6291728496551514,
+      "step": 2270
+    },
+    {
+      "epoch": 0.14205607476635515,
+      "grad_norm": 1.0517232210797287,
+      "learning_rate": 2e-05,
+      "loss": 0.610629940032959,
+      "step": 2280
+    },
+    {
+      "epoch": 0.1426791277258567,
+      "grad_norm": 0.7991464577811347,
+      "learning_rate": 2e-05,
+      "loss": 0.6206608772277832,
+      "step": 2290
+    },
+    {
+      "epoch": 0.14330218068535824,
+      "grad_norm": 1.0486150512063594,
+      "learning_rate": 2e-05,
+      "loss": 0.621973991394043,
+      "step": 2300
+    },
+    {
+      "epoch": 0.14392523364485982,
+      "grad_norm": 1.4649856376560535,
+      "learning_rate": 2e-05,
+      "loss": 0.6150485038757324,
+      "step": 2310
+    },
+    {
+      "epoch": 0.14454828660436136,
+      "grad_norm": 0.9220144729217414,
+      "learning_rate": 2e-05,
+      "loss": 0.6255329132080079,
+      "step": 2320
+    },
+    {
+      "epoch": 0.14517133956386294,
+      "grad_norm": 1.9234796847600562,
+      "learning_rate": 2e-05,
+      "loss": 0.6152030944824218,
+      "step": 2330
+    },
+    {
+      "epoch": 0.14579439252336449,
+      "grad_norm": 2.278734085313781,
+      "learning_rate": 2e-05,
+      "loss": 0.6301046371459961,
+      "step": 2340
+    },
+    {
+      "epoch": 0.14641744548286603,
+      "grad_norm": 0.912994756687639,
+      "learning_rate": 2e-05,
+      "loss": 0.6168749809265137,
+      "step": 2350
+    },
+    {
+      "epoch": 0.1470404984423676,
+      "grad_norm": 1.512514282445954,
+      "learning_rate": 2e-05,
+      "loss": 0.6138379573822021,
+      "step": 2360
+    },
+    {
+      "epoch": 0.14766355140186915,
+      "grad_norm": 0.8488271053398221,
+      "learning_rate": 2e-05,
+      "loss": 0.6176482200622558,
+      "step": 2370
+    },
+    {
+      "epoch": 0.14828660436137073,
+      "grad_norm": 1.1260154406535807,
+      "learning_rate": 2e-05,
+      "loss": 0.6312280178070069,
+      "step": 2380
+    },
+    {
+      "epoch": 0.14890965732087227,
+      "grad_norm": 1.0249329213490168,
+      "learning_rate": 2e-05,
+      "loss": 0.6207536220550537,
+      "step": 2390
+    },
+    {
+      "epoch": 0.14953271028037382,
+      "grad_norm": 6.830352047529763,
+      "learning_rate": 2e-05,
+      "loss": 0.6195199489593506,
+      "step": 2400
+    },
+    {
+      "epoch": 0.1501557632398754,
+      "grad_norm": 0.8212162064758552,
+      "learning_rate": 2e-05,
+      "loss": 0.6240365505218506,
+      "step": 2410
+    },
+    {
+      "epoch": 0.15077881619937694,
+      "grad_norm": 0.990394552773978,
+      "learning_rate": 2e-05,
+      "loss": 0.616001033782959,
+      "step": 2420
+    },
+    {
+      "epoch": 0.15140186915887852,
+      "grad_norm": 1.0409093552353237,
+      "learning_rate": 2e-05,
+      "loss": 0.624701976776123,
+      "step": 2430
+    },
+    {
+      "epoch": 0.15202492211838006,
+      "grad_norm": 0.7417001141637177,
+      "learning_rate": 2e-05,
+      "loss": 0.6140687942504883,
+      "step": 2440
+    },
+    {
+      "epoch": 0.1526479750778816,
+      "grad_norm": 0.8594235791688132,
+      "learning_rate": 2e-05,
+      "loss": 0.6134128570556641,
+      "step": 2450
+    },
+    {
+      "epoch": 0.15327102803738318,
+      "grad_norm": 1.1569208903092205,
+      "learning_rate": 2e-05,
+      "loss": 0.6117560386657714,
+      "step": 2460
+    },
+    {
+      "epoch": 0.15389408099688473,
+      "grad_norm": 1.69967871711498,
+      "learning_rate": 2e-05,
+      "loss": 0.6019431114196777,
+      "step": 2470
+    },
+    {
+      "epoch": 0.1545171339563863,
+      "grad_norm": 0.8548931092288313,
+      "learning_rate": 2e-05,
+      "loss": 0.6015212059020996,
+      "step": 2480
+    },
+    {
+      "epoch": 0.15514018691588785,
+      "grad_norm": 0.886887658669317,
+      "learning_rate": 2e-05,
+      "loss": 0.6151418209075927,
+      "step": 2490
+    },
+    {
+      "epoch": 0.1557632398753894,
+      "grad_norm": 0.8126022995500624,
+      "learning_rate": 2e-05,
+      "loss": 0.6098021984100341,
+      "step": 2500
+    },
+    {
+      "epoch": 0.15638629283489097,
+      "grad_norm": 5.8688610503250995,
+      "learning_rate": 2e-05,
+      "loss": 0.6503901481628418,
+      "step": 2510
+    },
+    {
+      "epoch": 0.15700934579439252,
+      "grad_norm": 1.5154051735521508,
+      "learning_rate": 2e-05,
+      "loss": 0.596190357208252,
+      "step": 2520
+    },
+    {
+      "epoch": 0.1576323987538941,
+      "grad_norm": 0.7654243639322663,
+      "learning_rate": 2e-05,
+      "loss": 0.6025626182556152,
+      "step": 2530
+    },
+    {
+      "epoch": 0.15825545171339564,
+      "grad_norm": 1.1132944369687552,
+      "learning_rate": 2e-05,
+      "loss": 0.6203357219696045,
+      "step": 2540
+    },
+    {
+      "epoch": 0.1588785046728972,
+      "grad_norm": 0.7579451495884797,
+      "learning_rate": 2e-05,
+      "loss": 0.6061652660369873,
+      "step": 2550
+    },
+    {
+      "epoch": 0.15950155763239876,
+      "grad_norm": 0.776258460799822,
+      "learning_rate": 2e-05,
+      "loss": 0.6222086906433105,
+      "step": 2560
+    },
+    {
+      "epoch": 0.1601246105919003,
+      "grad_norm": 0.8583483750630405,
+      "learning_rate": 2e-05,
+      "loss": 0.6153072357177735,
+      "step": 2570
+    },
+    {
+      "epoch": 0.16074766355140188,
+      "grad_norm": 2.0037458535906816,
+      "learning_rate": 2e-05,
+      "loss": 0.6055281639099122,
+      "step": 2580
+    },
+    {
+      "epoch": 0.16137071651090343,
+      "grad_norm": 1.097944434413077,
+      "learning_rate": 2e-05,
+      "loss": 0.6079090118408204,
+      "step": 2590
+    },
+    {
+      "epoch": 0.16199376947040497,
+      "grad_norm": 1.2751424275914838,
+      "learning_rate": 2e-05,
+      "loss": 0.6183807373046875,
+      "step": 2600
+    },
+    {
+      "epoch": 0.16261682242990655,
+      "grad_norm": 1.0063327265274151,
+      "learning_rate": 2e-05,
+      "loss": 0.602623701095581,
+      "step": 2610
+    },
+    {
+      "epoch": 0.1632398753894081,
+      "grad_norm": 1.259540123034618,
+      "learning_rate": 2e-05,
+      "loss": 0.6367355346679687,
+      "step": 2620
+    },
+    {
+      "epoch": 0.16386292834890967,
+      "grad_norm": 104.4583162475991,
+      "learning_rate": 2e-05,
+      "loss": 0.6318365097045898,
+      "step": 2630
+    },
+    {
+      "epoch": 0.16448598130841122,
+      "grad_norm": 1.427123342296283,
+      "learning_rate": 2e-05,
+      "loss": 0.6292930603027344,
+      "step": 2640
+    },
+    {
+      "epoch": 0.16510903426791276,
+      "grad_norm": 0.8650086206351112,
+      "learning_rate": 2e-05,
+      "loss": 0.61443190574646,
+      "step": 2650
+    },
+    {
+      "epoch": 0.16573208722741434,
+      "grad_norm": 3.1916398257250504,
+      "learning_rate": 2e-05,
+      "loss": 0.6197348594665527,
+      "step": 2660
+    },
+    {
+      "epoch": 0.16635514018691588,
+      "grad_norm": 0.7986041860851832,
+      "learning_rate": 2e-05,
+      "loss": 0.614453649520874,
+      "step": 2670
+    },
+    {
+      "epoch": 0.16697819314641746,
+      "grad_norm": 1.3201508165010398,
+      "learning_rate": 2e-05,
+      "loss": 0.6107942581176757,
+      "step": 2680
+    },
+    {
+      "epoch": 0.167601246105919,
+      "grad_norm": 1.665074918696496,
+      "learning_rate": 2e-05,
+      "loss": 0.613221263885498,
+      "step": 2690
+    },
+    {
+      "epoch": 0.16822429906542055,
+      "grad_norm": 0.8620318209601655,
+      "learning_rate": 2e-05,
+      "loss": 0.6344241142272949,
+      "step": 2700
+    },
+    {
+      "epoch": 0.16884735202492213,
+      "grad_norm": 1.3442702571461933,
+      "learning_rate": 2e-05,
+      "loss": 0.6164389610290527,
+      "step": 2710
+    },
+    {
+      "epoch": 0.16947040498442367,
+      "grad_norm": 0.9260701589715262,
+      "learning_rate": 2e-05,
+      "loss": 0.613405704498291,
+      "step": 2720
+    },
+    {
+      "epoch": 0.17009345794392525,
+      "grad_norm": 1.099544477746399,
+      "learning_rate": 2e-05,
+      "loss": 0.6274864196777343,
+      "step": 2730
+    },
+    {
+      "epoch": 0.1707165109034268,
+      "grad_norm": 0.8643537360903935,
+      "learning_rate": 2e-05,
+      "loss": 0.6000308513641357,
+      "step": 2740
+    },
+    {
+      "epoch": 0.17133956386292834,
+      "grad_norm": 1.3017161161080681,
+      "learning_rate": 2e-05,
+      "loss": 0.5971711635589599,
+      "step": 2750
+    },
+    {
+      "epoch": 0.17196261682242991,
+      "grad_norm": 1.1902319884643848,
+      "learning_rate": 2e-05,
+      "loss": 0.6154521942138672,
+      "step": 2760
+    },
+    {
+      "epoch": 0.17258566978193146,
+      "grad_norm": 1.0493409764115966,
+      "learning_rate": 2e-05,
+      "loss": 0.6070003032684326,
+      "step": 2770
+    },
+    {
+      "epoch": 0.17320872274143304,
+      "grad_norm": 0.9839248624483965,
+      "learning_rate": 2e-05,
+      "loss": 0.622803020477295,
+      "step": 2780
+    },
+    {
+      "epoch": 0.17383177570093458,
+      "grad_norm": 0.987506757652346,
+      "learning_rate": 2e-05,
+      "loss": 0.6237383842468261,
+      "step": 2790
+    },
+    {
+      "epoch": 0.17445482866043613,
+      "grad_norm": 0.9556442579019893,
+      "learning_rate": 2e-05,
+      "loss": 0.6238723754882812,
+      "step": 2800
+    },
+    {
+      "epoch": 0.1750778816199377,
+      "grad_norm": 1.3015249537711309,
+      "learning_rate": 2e-05,
+      "loss": 0.6240334987640381,
+      "step": 2810
+    },
+    {
+      "epoch": 0.17570093457943925,
+      "grad_norm": 0.8689079178190325,
+      "learning_rate": 2e-05,
+      "loss": 0.6084201335906982,
+      "step": 2820
+    },
+    {
+      "epoch": 0.1763239875389408,
+      "grad_norm": 1.1340919350819536,
+      "learning_rate": 2e-05,
+      "loss": 0.5996224403381347,
+      "step": 2830
+    },
+    {
+      "epoch": 0.17694704049844237,
+      "grad_norm": 0.9795705202791282,
+      "learning_rate": 2e-05,
+      "loss": 0.6038512706756591,
+      "step": 2840
+    },
+    {
+      "epoch": 0.17757009345794392,
+      "grad_norm": 2.6606846257663945,
+      "learning_rate": 2e-05,
+      "loss": 0.5987706661224366,
+      "step": 2850
+    },
+    {
+      "epoch": 0.1781931464174455,
+      "grad_norm": 19.68930944533334,
+      "learning_rate": 2e-05,
+      "loss": 0.5907588958740234,
+      "step": 2860
+    },
+    {
+      "epoch": 0.17881619937694704,
+      "grad_norm": 1.8719423141945688,
+      "learning_rate": 2e-05,
+      "loss": 0.6066802024841309,
+      "step": 2870
+    },
+    {
+      "epoch": 0.17943925233644858,
+      "grad_norm": 2.216098319853744,
+      "learning_rate": 2e-05,
+      "loss": 0.6057388305664062,
+      "step": 2880
+    },
+    {
+      "epoch": 0.18006230529595016,
+      "grad_norm": 1.3808655703948696,
+      "learning_rate": 2e-05,
+      "loss": 0.6170113563537598,
+      "step": 2890
+    },
+    {
+      "epoch": 0.1806853582554517,
+      "grad_norm": 1.4359607237161436,
+      "learning_rate": 2e-05,
+      "loss": 0.6162316799163818,
+      "step": 2900
+    },
+    {
+      "epoch": 0.18130841121495328,
+      "grad_norm": 1.37873153836489,
+      "learning_rate": 2e-05,
+      "loss": 0.6307910919189453,
+      "step": 2910
+    },
+    {
+      "epoch": 0.18193146417445483,
+      "grad_norm": 0.9692476440693383,
+      "learning_rate": 2e-05,
+      "loss": 0.6094634056091308,
+      "step": 2920
+    },
+    {
+      "epoch": 0.18255451713395637,
+      "grad_norm": 3.5356799501610623,
+      "learning_rate": 2e-05,
+      "loss": 0.6067126750946045,
+      "step": 2930
+    },
+    {
+      "epoch": 0.18317757009345795,
+      "grad_norm": 2.188926708701314,
+      "learning_rate": 2e-05,
+      "loss": 0.6112101554870606,
+      "step": 2940
+    },
+    {
+      "epoch": 0.1838006230529595,
+      "grad_norm": 1.1365629203544558,
+      "learning_rate": 2e-05,
+      "loss": 0.6269072532653809,
+      "step": 2950
+    },
+    {
+      "epoch": 0.18442367601246107,
+      "grad_norm": 1.4340837125205705,
+      "learning_rate": 2e-05,
+      "loss": 0.6226873397827148,
+      "step": 2960
+    },
+    {
+      "epoch": 0.18504672897196262,
+      "grad_norm": 1.0717789116911174,
+      "learning_rate": 2e-05,
+      "loss": 0.6069749355316162,
+      "step": 2970
+    },
+    {
+      "epoch": 0.18566978193146416,
+      "grad_norm": 0.9718344540330416,
+      "learning_rate": 2e-05,
+      "loss": 0.603231143951416,
+      "step": 2980
+    },
+    {
+      "epoch": 0.18629283489096574,
+      "grad_norm": 0.9596508610300603,
+      "learning_rate": 2e-05,
+      "loss": 0.6163008689880372,
+      "step": 2990
+    },
+    {
+      "epoch": 0.18691588785046728,
+      "grad_norm": 0.9810718421072577,
+      "learning_rate": 2e-05,
+      "loss": 0.6064743041992188,
+      "step": 3000
+    },
+    {
+      "epoch": 0.18753894080996886,
+      "grad_norm": 1.0834670742216737,
+      "learning_rate": 2e-05,
+      "loss": 0.6198514938354492,
+      "step": 3010
+    },
+    {
+      "epoch": 0.1881619937694704,
+      "grad_norm": 0.8463762143478956,
+      "learning_rate": 2e-05,
+      "loss": 0.6061748504638672,
+      "step": 3020
+    },
+    {
+      "epoch": 0.18878504672897195,
+      "grad_norm": 0.9813823798184725,
+      "learning_rate": 2e-05,
+      "loss": 0.6404414176940918,
+      "step": 3030
+    },
+    {
+      "epoch": 0.18940809968847352,
+      "grad_norm": 1.0712788072261463,
+      "learning_rate": 2e-05,
+      "loss": 0.604254674911499,
+      "step": 3040
+    },
+    {
+      "epoch": 0.19003115264797507,
+      "grad_norm": 1.2909589733701912,
+      "learning_rate": 2e-05,
+      "loss": 0.6029664516448975,
+      "step": 3050
+    },
+    {
+      "epoch": 0.19065420560747665,
+      "grad_norm": 1.1908100552696426,
+      "learning_rate": 2e-05,
+      "loss": 0.5972621440887451,
+      "step": 3060
+    },
+    {
+      "epoch": 0.1912772585669782,
+      "grad_norm": 1.4739572911129615,
+      "learning_rate": 2e-05,
+      "loss": 0.6041380882263183,
+      "step": 3070
+    },
+    {
+      "epoch": 0.19190031152647974,
+      "grad_norm": 1.128814468184035,
+      "learning_rate": 2e-05,
+      "loss": 0.6038087844848633,
+      "step": 3080
+    },
+    {
+      "epoch": 0.1925233644859813,
+      "grad_norm": 1.7651612345255372,
+      "learning_rate": 2e-05,
+      "loss": 0.5975883483886719,
+      "step": 3090
+    },
+    {
+      "epoch": 0.19314641744548286,
+      "grad_norm": 0.9461212924061347,
+      "learning_rate": 2e-05,
+      "loss": 0.6105434417724609,
+      "step": 3100
+    },
+    {
+      "epoch": 0.19376947040498443,
+      "grad_norm": 2.2891880654075445,
+      "learning_rate": 2e-05,
+      "loss": 0.6082868099212646,
+      "step": 3110
+    },
+    {
+      "epoch": 0.19439252336448598,
+      "grad_norm": 1.8448080043361226,
+      "learning_rate": 2e-05,
+      "loss": 0.5853926658630371,
+      "step": 3120
+    },
+    {
+      "epoch": 0.19501557632398753,
+      "grad_norm": 1.4989368303434212,
+      "learning_rate": 2e-05,
+      "loss": 0.5968837738037109,
+      "step": 3130
+    },
+    {
+      "epoch": 0.1956386292834891,
+      "grad_norm": 1.249953695303028,
+      "learning_rate": 2e-05,
+      "loss": 0.6129058361053467,
+      "step": 3140
+    },
+    {
+      "epoch": 0.19626168224299065,
+      "grad_norm": 1.4591880703396227,
+      "learning_rate": 2e-05,
+      "loss": 0.6002510070800782,
+      "step": 3150
+    },
+    {
+      "epoch": 0.19688473520249222,
+      "grad_norm": 1.4423488433612786,
+      "learning_rate": 2e-05,
+      "loss": 0.6035333156585694,
+      "step": 3160
+    },
+    {
+      "epoch": 0.19750778816199377,
+      "grad_norm": 1.3947351389055,
+      "learning_rate": 2e-05,
+      "loss": 0.6291537284851074,
+      "step": 3170
+    },
+    {
+      "epoch": 0.19813084112149532,
+      "grad_norm": 4.10240563769065,
+      "learning_rate": 2e-05,
+      "loss": 0.5963770866394043,
+      "step": 3180
+    },
+    {
+      "epoch": 0.1987538940809969,
+      "grad_norm": 1.7244210112082368,
+      "learning_rate": 2e-05,
+      "loss": 0.6121903896331787,
+      "step": 3190
+    },
+    {
+      "epoch": 0.19937694704049844,
+      "grad_norm": 2.0848711542832348,
+      "learning_rate": 2e-05,
+      "loss": 0.6024190425872803,
+      "step": 3200
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.0499244367033858,
+      "learning_rate": 2e-05,
+      "loss": 0.6077515602111816,
+      "step": 3210
+    },
+    {
+      "epoch": 0.20062305295950156,
+      "grad_norm": 6.079414127216597,
+      "learning_rate": 2e-05,
+      "loss": 0.6281554222106933,
+      "step": 3220
+    },
+    {
+      "epoch": 0.2012461059190031,
+      "grad_norm": 1.5639320177345721,
+      "learning_rate": 2e-05,
+      "loss": 0.6229495525360107,
+      "step": 3230
+    },
+    {
+      "epoch": 0.20186915887850468,
+      "grad_norm": 1.2999103502580012,
+      "learning_rate": 2e-05,
+      "loss": 0.6261126518249511,
+      "step": 3240
+    },
+    {
+      "epoch": 0.20249221183800623,
+      "grad_norm": 0.9483827187157255,
+      "learning_rate": 2e-05,
+      "loss": 0.6157403945922851,
+      "step": 3250
+    },
+    {
+      "epoch": 0.2031152647975078,
+      "grad_norm": 2.6334832064179547,
+      "learning_rate": 2e-05,
+      "loss": 0.6289525508880616,
+      "step": 3260
+    },
+    {
+      "epoch": 0.20373831775700935,
+      "grad_norm": 5.007587221569407,
+      "learning_rate": 2e-05,
+      "loss": 0.6042305946350097,
+      "step": 3270
+    },
+    {
+      "epoch": 0.2043613707165109,
+      "grad_norm": 1.2041073716260875,
+      "learning_rate": 2e-05,
+      "loss": 0.6058319091796875,
+      "step": 3280
+    },
+    {
+      "epoch": 0.20498442367601247,
+      "grad_norm": 1.766204737783734,
+      "learning_rate": 2e-05,
+      "loss": 0.6154224395751953,
+      "step": 3290
+    },
+    {
+      "epoch": 0.205607476635514,
+      "grad_norm": 0.8740746341333337,
+      "learning_rate": 2e-05,
+      "loss": 0.6213559627532959,
+      "step": 3300
+    },
+    {
+      "epoch": 0.2062305295950156,
+      "grad_norm": 1.04279039502041,
+      "learning_rate": 2e-05,
+      "loss": 0.6154450416564942,
+      "step": 3310
+    },
+    {
+      "epoch": 0.20685358255451713,
+      "grad_norm": 2.92186700598731,
+      "learning_rate": 2e-05,
+      "loss": 0.6027130126953125,
+      "step": 3320
+    },
+    {
+      "epoch": 0.20747663551401868,
+      "grad_norm": 1.3480157636105494,
+      "learning_rate": 2e-05,
+      "loss": 0.5983580589294434,
+      "step": 3330
+    },
+    {
+      "epoch": 0.20809968847352026,
+      "grad_norm": 2.9836349102977056,
+      "learning_rate": 2e-05,
+      "loss": 0.6069539546966553,
+      "step": 3340
+    },
+    {
+      "epoch": 0.2087227414330218,
+      "grad_norm": 1.1252906183894023,
+      "learning_rate": 2e-05,
+      "loss": 0.6007477760314941,
+      "step": 3350
+    },
+    {
+      "epoch": 0.20934579439252338,
+      "grad_norm": 2.2587668816699344,
+      "learning_rate": 2e-05,
+      "loss": 0.6101480007171631,
+      "step": 3360
+    },
+    {
+      "epoch": 0.20996884735202492,
+      "grad_norm": 0.9360361723449176,
+      "learning_rate": 2e-05,
+      "loss": 0.6228442192077637,
+      "step": 3370
+    },
+    {
+      "epoch": 0.21059190031152647,
+      "grad_norm": 0.9728125708602748,
+      "learning_rate": 2e-05,
+      "loss": 0.592249059677124,
+      "step": 3380
+    },
+    {
+      "epoch": 0.21121495327102804,
+      "grad_norm": 1.4927763717256455,
+      "learning_rate": 2e-05,
+      "loss": 0.6324718475341797,
+      "step": 3390
+    },
+    {
+      "epoch": 0.2118380062305296,
+      "grad_norm": 2.499834746076374,
+      "learning_rate": 2e-05,
+      "loss": 0.6269383430480957,
+      "step": 3400
+    },
+    {
+      "epoch": 0.21246105919003117,
+      "grad_norm": 2.705969032439363,
+      "learning_rate": 2e-05,
+      "loss": 0.5976783275604248,
+      "step": 3410
+    },
+    {
+      "epoch": 0.2130841121495327,
+      "grad_norm": 1.1283976867593206,
+      "learning_rate": 2e-05,
+      "loss": 0.6239967346191406,
+      "step": 3420
+    },
+    {
+      "epoch": 0.21370716510903426,
+      "grad_norm": 1.3467908684248446,
+      "learning_rate": 2e-05,
+      "loss": 0.6075163841247558,
+      "step": 3430
+    },
+    {
+      "epoch": 0.21433021806853583,
+      "grad_norm": 0.9478567100662277,
+      "learning_rate": 2e-05,
+      "loss": 0.5992548942565918,
+      "step": 3440
+    },
+    {
+      "epoch": 0.21495327102803738,
+      "grad_norm": 0.993132373114707,
+      "learning_rate": 2e-05,
+      "loss": 0.6148913383483887,
+      "step": 3450
+    },
+    {
+      "epoch": 0.21557632398753895,
+      "grad_norm": 3.910982900878596,
+      "learning_rate": 2e-05,
+      "loss": 0.5927858352661133,
+      "step": 3460
+    },
+    {
+      "epoch": 0.2161993769470405,
+      "grad_norm": 6.009374682670098,
+      "learning_rate": 2e-05,
+      "loss": 0.6289549827575683,
+      "step": 3470
+    },
+    {
+      "epoch": 0.21682242990654205,
+      "grad_norm": 0.9663269424414038,
+      "learning_rate": 2e-05,
+      "loss": 0.6224092006683349,
+      "step": 3480
+    },
+    {
+      "epoch": 0.21744548286604362,
+      "grad_norm": 1.6365724772727905,
+      "learning_rate": 2e-05,
+      "loss": 0.6078380584716797,
+      "step": 3490
+    },
+    {
+      "epoch": 0.21806853582554517,
+      "grad_norm": 1.229574083958739,
+      "learning_rate": 2e-05,
+      "loss": 0.5792657375335694,
+      "step": 3500
+    },
+    {
+      "epoch": 0.21869158878504674,
+      "grad_norm": 2.992165227161805,
+      "learning_rate": 2e-05,
+      "loss": 0.621655797958374,
+      "step": 3510
+    },
+    {
+      "epoch": 0.2193146417445483,
+      "grad_norm": 4.148526182581887,
+      "learning_rate": 2e-05,
+      "loss": 0.6216963768005371,
+      "step": 3520
+    },
+    {
+      "epoch": 0.21993769470404984,
+      "grad_norm": 10.67368274230513,
+      "learning_rate": 2e-05,
+      "loss": 0.6301459789276123,
+      "step": 3530
+    },
+    {
+      "epoch": 0.2205607476635514,
+      "grad_norm": 11.941995595215085,
+      "learning_rate": 2e-05,
+      "loss": 0.6267009258270264,
+      "step": 3540
+    },
+    {
+      "epoch": 0.22118380062305296,
+      "grad_norm": 2.6774563166793413,
+      "learning_rate": 2e-05,
+      "loss": 0.6055225372314453,
+      "step": 3550
+    },
+    {
+      "epoch": 0.22180685358255453,
+      "grad_norm": 1.3212737377183037,
+      "learning_rate": 2e-05,
+      "loss": 0.6109843254089355,
+      "step": 3560
+    },
+    {
+      "epoch": 0.22242990654205608,
+      "grad_norm": 8.076756246749495,
+      "learning_rate": 2e-05,
+      "loss": 0.5983676910400391,
+      "step": 3570
+    },
+    {
+      "epoch": 0.22305295950155762,
+      "grad_norm": 2.8551680592494164,
+      "learning_rate": 2e-05,
+      "loss": 0.6131735801696777,
+      "step": 3580
+    },
+    {
+      "epoch": 0.2236760124610592,
+      "grad_norm": 2.2225395947886226,
+      "learning_rate": 2e-05,
+      "loss": 0.632578182220459,
+      "step": 3590
+    },
+    {
+      "epoch": 0.22429906542056074,
+      "grad_norm": 5.825227047417447,
+      "learning_rate": 2e-05,
+      "loss": 0.664147138595581,
+      "step": 3600
+    },
+    {
+      "epoch": 0.22492211838006232,
+      "grad_norm": 2.440874666575554,
+      "learning_rate": 2e-05,
+      "loss": 0.5942914485931396,
+      "step": 3610
+    },
+    {
+      "epoch": 0.22554517133956387,
+      "grad_norm": 3.1694509305743948,
+      "learning_rate": 2e-05,
+      "loss": 0.6079733848571778,
+      "step": 3620
+    },
+    {
+      "epoch": 0.2261682242990654,
+      "grad_norm": 1.5756407496084817,
+      "learning_rate": 2e-05,
+      "loss": 0.6157325744628906,
+      "step": 3630
+    },
+    {
+      "epoch": 0.226791277258567,
+      "grad_norm": 3.416018110861746,
+      "learning_rate": 2e-05,
+      "loss": 0.6079260826110839,
+      "step": 3640
+    },
+    {
+      "epoch": 0.22741433021806853,
+      "grad_norm": 5.52741589135722,
+      "learning_rate": 2e-05,
+      "loss": 0.6063628673553467,
+      "step": 3650
+    },
+    {
+      "epoch": 0.22803738317757008,
+      "grad_norm": 1.861439779936567,
+      "learning_rate": 2e-05,
+      "loss": 0.5976187705993652,
+      "step": 3660
+    },
+    {
+      "epoch": 0.22866043613707165,
+      "grad_norm": 1.8689453992844312,
+      "learning_rate": 2e-05,
+      "loss": 0.6108988761901856,
+      "step": 3670
+    },
+    {
+      "epoch": 0.2292834890965732,
+      "grad_norm": 1.4407378081205044,
+      "learning_rate": 2e-05,
+      "loss": 0.6123188018798829,
+      "step": 3680
+    },
+    {
+      "epoch": 0.22990654205607478,
+      "grad_norm": 1.6523402550965685,
+      "learning_rate": 2e-05,
+      "loss": 0.6129428386688233,
+      "step": 3690
+    },
+    {
+      "epoch": 0.23052959501557632,
+      "grad_norm": 0.8575852264373719,
+      "learning_rate": 2e-05,
+      "loss": 0.6136985778808594,
+      "step": 3700
+    },
+    {
+      "epoch": 0.23115264797507787,
+      "grad_norm": 1.281763383798869,
+      "learning_rate": 2e-05,
+      "loss": 0.5970061302185059,
+      "step": 3710
+    },
+    {
+      "epoch": 0.23177570093457944,
+      "grad_norm": 1.5925361439188803,
+      "learning_rate": 2e-05,
+      "loss": 0.5993070602416992,
+      "step": 3720
+    },
+    {
+      "epoch": 0.232398753894081,
+      "grad_norm": 2.8022563271278855,
+      "learning_rate": 2e-05,
+      "loss": 0.596066665649414,
+      "step": 3730
+    },
+    {
+      "epoch": 0.23302180685358256,
+      "grad_norm": 1.262749907659991,
+      "learning_rate": 2e-05,
+      "loss": 0.6160148620605469,
+      "step": 3740
+    },
+    {
+      "epoch": 0.2336448598130841,
+      "grad_norm": 2.320187663651848,
+      "learning_rate": 2e-05,
+      "loss": 0.6163660049438476,
+      "step": 3750
+    },
+    {
+      "epoch": 0.23426791277258566,
+      "grad_norm": 10.566365917195634,
+      "learning_rate": 2e-05,
+      "loss": 0.6310874938964843,
+      "step": 3760
+    },
+    {
+      "epoch": 0.23489096573208723,
+      "grad_norm": 12.914122608869691,
+      "learning_rate": 2e-05,
+      "loss": 0.5889072418212891,
+      "step": 3770
+    },
+    {
+      "epoch": 0.23551401869158878,
+      "grad_norm": 0.9254758767576047,
+      "learning_rate": 2e-05,
+      "loss": 0.6076817989349366,
+      "step": 3780
+    },
+    {
+      "epoch": 0.23613707165109035,
+      "grad_norm": 0.9401442841255053,
+      "learning_rate": 2e-05,
+      "loss": 0.6412858009338379,
+      "step": 3790
+    },
+    {
+      "epoch": 0.2367601246105919,
+      "grad_norm": 0.9047476364645441,
+      "learning_rate": 2e-05,
+      "loss": 0.6125865459442139,
+      "step": 3800
+    },
+    {
+      "epoch": 0.23738317757009345,
+      "grad_norm": 1.7144124522782018,
+      "learning_rate": 2e-05,
+      "loss": 0.6077367305755615,
+      "step": 3810
+    },
+    {
+      "epoch": 0.23800623052959502,
+      "grad_norm": 1.0136460689732938,
+      "learning_rate": 2e-05,
+      "loss": 0.596423864364624,
+      "step": 3820
+    },
+    {
+      "epoch": 0.23862928348909657,
+      "grad_norm": 0.8685269175550879,
+      "learning_rate": 2e-05,
+      "loss": 0.5939150810241699,
+      "step": 3830
+    },
+    {
+      "epoch": 0.23925233644859814,
+      "grad_norm": 0.7917949384811468,
+      "learning_rate": 2e-05,
+      "loss": 0.5833038330078125,
+      "step": 3840
+    },
+    {
+      "epoch": 0.2398753894080997,
+      "grad_norm": 1.577708167866883,
+      "learning_rate": 2e-05,
+      "loss": 0.6014552116394043,
+      "step": 3850
+    },
+    {
+      "epoch": 0.24049844236760123,
+      "grad_norm": 0.8039608094716413,
+      "learning_rate": 2e-05,
+      "loss": 0.5982142448425293,
+      "step": 3860
+    },
+    {
+      "epoch": 0.2411214953271028,
+      "grad_norm": 0.811526983917987,
+      "learning_rate": 2e-05,
+      "loss": 0.5976861000061036,
+      "step": 3870
+    },
+    {
+      "epoch": 0.24174454828660435,
+      "grad_norm": 1.6709117069682082,
+      "learning_rate": 2e-05,
+      "loss": 0.6228075981140136,
+      "step": 3880
+    },
+    {
+      "epoch": 0.24236760124610593,
+      "grad_norm": 2.219833264360674,
+      "learning_rate": 2e-05,
+      "loss": 0.5801692962646484,
+      "step": 3890
+    },
+    {
+      "epoch": 0.24299065420560748,
+      "grad_norm": 1.1198032739357877,
+      "learning_rate": 2e-05,
+      "loss": 0.6063448905944824,
+      "step": 3900
+    },
+    {
+      "epoch": 0.24361370716510902,
+      "grad_norm": 0.8368802907802825,
+      "learning_rate": 2e-05,
+      "loss": 0.600613784790039,
+      "step": 3910
+    },
+    {
+      "epoch": 0.2442367601246106,
+      "grad_norm": 1.2126989742636782,
+      "learning_rate": 2e-05,
+      "loss": 0.5789689540863037,
+      "step": 3920
+    },
+    {
+      "epoch": 0.24485981308411214,
+      "grad_norm": 1.3615959301231297,
+      "learning_rate": 2e-05,
+      "loss": 0.5999377250671387,
+      "step": 3930
+    },
+    {
+      "epoch": 0.24548286604361372,
+      "grad_norm": 2.1023053247644103,
+      "learning_rate": 2e-05,
+      "loss": 0.6115778923034668,
+      "step": 3940
+    },
+    {
+      "epoch": 0.24610591900311526,
+      "grad_norm": 4.193609729089036,
+      "learning_rate": 2e-05,
+      "loss": 0.6208731651306152,
+      "step": 3950
+    },
+    {
+      "epoch": 0.2467289719626168,
+      "grad_norm": 1.6804215889163965,
+      "learning_rate": 2e-05,
+      "loss": 0.6064939022064209,
+      "step": 3960
+    },
+    {
+      "epoch": 0.24735202492211839,
+      "grad_norm": 4.912295554467544,
+      "learning_rate": 2e-05,
+      "loss": 0.5894756317138672,
+      "step": 3970
+    },
+    {
+      "epoch": 0.24797507788161993,
+      "grad_norm": 1.1404451491077938,
+      "learning_rate": 2e-05,
+      "loss": 0.5897003173828125,
+      "step": 3980
+    },
+    {
+      "epoch": 0.2485981308411215,
+      "grad_norm": 1.7114106724658036,
+      "learning_rate": 2e-05,
+      "loss": 0.59234619140625,
+      "step": 3990
+    },
+    {
+      "epoch": 0.24922118380062305,
+      "grad_norm": 2.5665067626456093,
+      "learning_rate": 2e-05,
+      "loss": 0.6199214935302735,
+      "step": 4000
+    },
+    {
+      "epoch": 0.2498442367601246,
+      "grad_norm": 1.7306229008409815,
+      "learning_rate": 2e-05,
+      "loss": 0.6039542198181153,
+      "step": 4010
+    },
+    {
+      "epoch": 0.2504672897196262,
+      "grad_norm": 1.995363481901881,
+      "learning_rate": 2e-05,
+      "loss": 0.5996993064880372,
+      "step": 4020
+    },
+    {
+      "epoch": 0.2510903426791277,
+      "grad_norm": 3.882130945374297,
+      "learning_rate": 2e-05,
+      "loss": 0.6014129161834717,
+      "step": 4030
+    },
+    {
+      "epoch": 0.25171339563862927,
+      "grad_norm": 1.1173712086171077,
+      "learning_rate": 2e-05,
+      "loss": 0.6095175743103027,
+      "step": 4040
+    },
+    {
+      "epoch": 0.2523364485981308,
+      "grad_norm": 1.2005317520120649,
+      "learning_rate": 2e-05,
+      "loss": 0.6037531852722168,
+      "step": 4050
+    },
+    {
+      "epoch": 0.2529595015576324,
+      "grad_norm": 1.0072130454113009,
+      "learning_rate": 2e-05,
+      "loss": 0.5903209686279297,
+      "step": 4060
+    },
+    {
+      "epoch": 0.25358255451713396,
+      "grad_norm": 1.294749504835451,
+      "learning_rate": 2e-05,
+      "loss": 0.5879500865936279,
+      "step": 4070
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 5935,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 2035,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1927226987642880.0,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}

400M/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:aff51fc471427e3e0a6d8c88d0bfeec583ddb6b2423d3b49640f42c3d156eb52
+size 7313