Training in progress, epoch 44, checkpoint

checkpoint-8250/adapter_config.json

@@ -1,6 +1,6 @@
 {
   "auto_mapping": null,
-  "base_model_name_or_path": "Qwen/Qwen1.5-4B",
+  "base_model_name_or_path": "meta-llama/Llama-2-7b-hf",
   "bias": "none",
   "fan_in_fan_out": false,
   "inference_mode": true,

checkpoint-8250/adapter_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7a96244627816f928d7c132e82d058eb6d73b14ac8c86342253507a33df1dfc7
+size 143269386

checkpoint-8250/added_tokens.json

@@ -1,5 +1,5 @@
 {
-  "<|endoftext|>": 151643,
-  "<|im_end|>": 151645,
-  "<|im_start|>": 151644
+  "</s>": 2,
+  "<s>": 1,
+  "<unk>": 0
 }

checkpoint-8250/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:19708dde545c388377c1d3aafee974f3d0bb6f4144ba6b4c728bd851853316f6
-size 224537202
+oid sha256:8c3ff2f797a413d6261518ff6ff9ab6d2d8dfa36a47ab54ab6d0a3d82604ae3a
+size 286585234

checkpoint-8250/rng_state_0.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:bc7424e34bc979a100015d374398d9f1fbcd1dcab130b46d3837ac466883aee3
+oid sha256:a9d9906f96dcc02e3e458da008e1e2acc4a428845c078a8622d08437fd3e9d2d
 size 15024

checkpoint-8250/rng_state_1.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:19982a3ef4b68452f82302dded9850b6c1cdeca19989d5bddb96d6d645c37d45
+oid sha256:4484c0afb9b6f44be4c2cf2c46284c3052b677969550d9c9492ab7eca3cfe940
 size 15024

checkpoint-8250/rng_state_2.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6a2c3ef66b816120954de7d85a900e16f0b75fcb86b2de36fb89d5bf2687200a
+oid sha256:b239c9ceb1ba75da949ca0c5efc88b194128b631c1c6b8fdf321e30106ed8245
 size 15024

checkpoint-8250/rng_state_3.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9d704f2f8cf806676a82db2ce66c73c1fc402b64de15ccc231ace6726dea9c79
+oid sha256:3296632cd7773d7a9c14f242298e61032aaf2986da2d815034f6e6736d0372bb
 size 15024

checkpoint-8250/special_tokens_map.json

@@ -1,14 +1,6 @@
 {
-  "additional_special_tokens": [
-    "<|im_start|>",
-    "<|im_end|>"
-  ],
-  "eos_token": {
-    "content": "<|endoftext|>",
-    "lstrip": false,
-    "normalized": false,
-    "rstrip": false,
-    "single_word": false
-  },
-  "pad_token": "<|endoftext|>"
+  "bos_token": "<s>",
+  "eos_token": "</s>",
+  "pad_token": "</s>",
+  "unk_token": "<unk>"
 }

checkpoint-8250/tokenizer.model

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347
+size 499723

checkpoint-8250/tokenizer_config.json

@@ -1,24 +1,23 @@
 {
-  "add_prefix_space": false,
   "added_tokens_decoder": {
-    "151643": {
-      "content": "<|endoftext|>",
+    "0": {
+      "content": "<unk>",
       "lstrip": false,
       "normalized": false,
       "rstrip": false,
       "single_word": false,
       "special": true
     },
-    "151644": {
-      "content": "<|im_start|>",
+    "1": {
+      "content": "<s>",
       "lstrip": false,
       "normalized": false,
       "rstrip": false,
       "single_word": false,
       "special": true
     },
-    "151645": {
-      "content": "<|im_end|>",
+    "2": {
+      "content": "</s>",
       "lstrip": false,
       "normalized": false,
       "rstrip": false,
@@ -26,18 +25,16 @@
       "special": true
     }
   },
-  "additional_special_tokens": [
-    "<|im_start|>",
-    "<|im_end|>"
-  ],
-  "bos_token": null,
-  "chat_template": "{% for message in messages %}{% if loop.first and messages[0]['role'] != 'system' %}{{ '<|im_start|>system\nYou are a helpful assistant<|im_end|>\n' }}{% endif %}{{'<|im_start|>' + message['role'] + '\n' + message['content'] + '<|im_end|>' + '\n'}}{% endfor %}{% if add_generation_prompt %}{{ '<|im_start|>assistant\n' }}{% endif %}",
+  "additional_special_tokens": [],
+  "bos_token": "<s>",
   "clean_up_tokenization_spaces": false,
-  "eos_token": "<|endoftext|>",
-  "errors": "replace",
-  "model_max_length": 32768,
-  "pad_token": "<|endoftext|>",
-  "split_special_tokens": false,
-  "tokenizer_class": "Qwen2Tokenizer",
-  "unk_token": null
+  "eos_token": "</s>",
+  "legacy": false,
+  "model_max_length": 1000000000000000019884624838656,
+  "pad_token": "</s>",
+  "padding_side": "left",
+  "sp_model_kwargs": {},
+  "tokenizer_class": "LlamaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": true
 }

checkpoint-8250/trainer_state.json

@@ -9,1259 +9,1163 @@
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.5333333333333333,
-      "grad_norm": 0.9283596873283386,
+      "epoch": 0.53,
       "learning_rate": 3e-05,
-      "loss": 1.963,
+      "loss": 2.0369,
       "step": 100
     },
     {
-      "epoch": 0.9973333333333333,
-      "eval_accuracy": 0.5695384615384615,
-      "eval_loss": 1.6438876390457153,
-      "eval_runtime": 5.0534,
-      "eval_samples_per_second": 98.943,
-      "eval_steps_per_second": 12.467,
+      "epoch": 1.0,
+      "eval_accuracy": 0.6128205128205129,
+      "eval_loss": 1.2953312397003174,
+      "eval_runtime": 4.1114,
+      "eval_samples_per_second": 121.614,
+      "eval_steps_per_second": 15.323,
       "step": 187
     },
     {
-      "epoch": 0.9973333333333333,
-      "eval_exact_match": 16.2,
-      "eval_f1": 25.888571428571442,
+      "epoch": 1.0,
+      "eval_exact_match": 27.2,
+      "eval_f1": 37.44218938149973,
       "step": 187
     },
     {
-      "epoch": 1.0666666666666667,
-      "grad_norm": 1.0407154560089111,
+      "epoch": 1.07,
       "learning_rate": 3e-05,
-      "loss": 1.6668,
+      "loss": 1.3264,
       "step": 200
     },
     {
       "epoch": 1.6,
-      "grad_norm": 1.1281720399856567,
       "learning_rate": 3e-05,
-      "loss": 1.6099,
+      "loss": 1.2821,
       "step": 300
     },
     {
       "epoch": 2.0,
-      "eval_accuracy": 0.5733333333333334,
-      "eval_loss": 1.618270754814148,
-      "eval_runtime": 5.1983,
-      "eval_samples_per_second": 96.186,
-      "eval_steps_per_second": 12.119,
+      "eval_accuracy": 0.6146153846153846,
+      "eval_loss": 1.2741013765335083,
+      "eval_runtime": 4.8302,
+      "eval_samples_per_second": 103.516,
+      "eval_steps_per_second": 13.043,
       "step": 375
     },
     {
       "epoch": 2.0,
-      "eval_exact_match": 16.6,
-      "eval_f1": 26.287619047619057,
+      "eval_exact_match": 25.6,
+      "eval_f1": 38.026666666666664,
       "step": 375
     },
     {
-      "epoch": 2.1333333333333333,
-      "grad_norm": 1.2508225440979004,
+      "epoch": 2.13,
       "learning_rate": 3e-05,
-      "loss": 1.5866,
+      "loss": 1.2348,
       "step": 400
     },
     {
-      "epoch": 2.6666666666666665,
-      "grad_norm": 1.3197834491729736,
+      "epoch": 2.67,
       "learning_rate": 3e-05,
-      "loss": 1.524,
+      "loss": 1.1987,
       "step": 500
     },
     {
-      "epoch": 2.997333333333333,
-      "eval_accuracy": 0.5743589743589743,
-      "eval_loss": 1.6164497137069702,
-      "eval_runtime": 5.0495,
-      "eval_samples_per_second": 99.019,
-      "eval_steps_per_second": 12.476,
+      "epoch": 3.0,
+      "eval_accuracy": 0.6161538461538462,
+      "eval_loss": 1.2714948654174805,
+      "eval_runtime": 4.868,
+      "eval_samples_per_second": 102.712,
+      "eval_steps_per_second": 12.942,
       "step": 562
     },
     {
-      "epoch": 2.997333333333333,
-      "eval_exact_match": 17.8,
-      "eval_f1": 27.35428571428571,
+      "epoch": 3.0,
+      "eval_exact_match": 28.0,
+      "eval_f1": 40.6962091503268,
       "step": 562
     },
     {
       "epoch": 3.2,
-      "grad_norm": 1.6320394277572632,
       "learning_rate": 3e-05,
-      "loss": 1.4752,
+      "loss": 1.1537,
       "step": 600
     },
     {
-      "epoch": 3.7333333333333334,
-      "grad_norm": 1.7101725339889526,
+      "epoch": 3.73,
       "learning_rate": 3e-05,
-      "loss": 1.3938,
+      "loss": 1.066,
       "step": 700
     },
     {
       "epoch": 4.0,
-      "eval_accuracy": 0.5728717948717948,
-      "eval_loss": 1.6376241445541382,
-      "eval_runtime": 4.9662,
-      "eval_samples_per_second": 100.68,
-      "eval_steps_per_second": 12.686,
+      "eval_accuracy": 0.6151282051282051,
+      "eval_loss": 1.3011157512664795,
+      "eval_runtime": 4.407,
+      "eval_samples_per_second": 113.455,
+      "eval_steps_per_second": 14.295,
       "step": 750
     },
     {
       "epoch": 4.0,
-      "eval_exact_match": 18.0,
-      "eval_f1": 27.498571428571427,
+      "eval_exact_match": 26.2,
+      "eval_f1": 38.19190476190477,
       "step": 750
     },
     {
-      "epoch": 4.266666666666667,
-      "grad_norm": 2.1727664470672607,
+      "epoch": 4.27,
       "learning_rate": 3e-05,
-      "loss": 1.342,
+      "loss": 1.001,
       "step": 800
     },
     {
       "epoch": 4.8,
-      "grad_norm": 2.341384172439575,
       "learning_rate": 3e-05,
-      "loss": 1.2685,
+      "loss": 0.9381,
       "step": 900
     },
     {
-      "epoch": 4.997333333333334,
-      "eval_accuracy": 0.5699487179487179,
-      "eval_loss": 1.684618353843689,
-      "eval_runtime": 5.5457,
-      "eval_samples_per_second": 90.16,
-      "eval_steps_per_second": 11.36,
+      "epoch": 5.0,
+      "eval_accuracy": 0.6125641025641025,
+      "eval_loss": 1.3728467226028442,
+      "eval_runtime": 5.153,
+      "eval_samples_per_second": 97.03,
+      "eval_steps_per_second": 12.226,
       "step": 937
     },
     {
-      "epoch": 4.997333333333334,
-      "eval_exact_match": 19.2,
-      "eval_f1": 28.190952380952375,
+      "epoch": 5.0,
+      "eval_exact_match": 24.6,
+      "eval_f1": 37.08666666666668,
       "step": 937
     },
     {
-      "epoch": 5.333333333333333,
-      "grad_norm": 2.5624172687530518,
+      "epoch": 5.33,
       "learning_rate": 3e-05,
-      "loss": 1.1847,
+      "loss": 0.8667,
       "step": 1000
     },
     {
-      "epoch": 5.866666666666667,
-      "grad_norm": 2.8159019947052,
+      "epoch": 5.87,
       "learning_rate": 3e-05,
-      "loss": 1.1591,
+      "loss": 0.8238,
       "step": 1100
     },
     {
       "epoch": 6.0,
-      "eval_accuracy": 0.5672820512820513,
-      "eval_loss": 1.7546838521957397,
-      "eval_runtime": 5.0186,
-      "eval_samples_per_second": 99.629,
-      "eval_steps_per_second": 12.553,
+      "eval_accuracy": 0.6090769230769231,
+      "eval_loss": 1.4598820209503174,
+      "eval_runtime": 4.0994,
+      "eval_samples_per_second": 121.97,
+      "eval_steps_per_second": 15.368,
       "step": 1125
     },
     {
       "epoch": 6.0,
-      "eval_exact_match": 16.6,
-      "eval_f1": 25.054761904761914,
+      "eval_exact_match": 23.0,
+      "eval_f1": 35.324444444444445,
       "step": 1125
     },
     {
       "epoch": 6.4,
-      "grad_norm": 3.056166410446167,
       "learning_rate": 3e-05,
-      "loss": 1.0486,
+      "loss": 0.7461,
       "step": 1200
     },
     {
-      "epoch": 6.933333333333334,
-      "grad_norm": 2.873281240463257,
+      "epoch": 6.93,
       "learning_rate": 3e-05,
-      "loss": 1.0444,
+      "loss": 0.7289,
       "step": 1300
     },
     {
-      "epoch": 6.997333333333334,
-      "eval_accuracy": 0.5643076923076923,
-      "eval_loss": 1.8394943475723267,
-      "eval_runtime": 5.5931,
-      "eval_samples_per_second": 89.396,
-      "eval_steps_per_second": 11.264,
+      "epoch": 7.0,
+      "eval_accuracy": 0.6064102564102564,
+      "eval_loss": 1.5455493927001953,
+      "eval_runtime": 4.1683,
+      "eval_samples_per_second": 119.952,
+      "eval_steps_per_second": 15.114,
       "step": 1312
     },
     {
-      "epoch": 6.997333333333334,
-      "eval_exact_match": 15.8,
-      "eval_f1": 23.925714285714296,
+      "epoch": 7.0,
+      "eval_exact_match": 21.8,
+      "eval_f1": 34.57222222222222,
       "step": 1312
     },
     {
-      "epoch": 7.466666666666667,
-      "grad_norm": 3.069322109222412,
+      "epoch": 7.47,
       "learning_rate": 3e-05,
-      "loss": 0.9374,
+      "loss": 0.6464,
       "step": 1400
     },
     {
       "epoch": 8.0,
-      "grad_norm": 3.2732129096984863,
       "learning_rate": 3e-05,
-      "loss": 0.9535,
+      "loss": 0.6559,
       "step": 1500
     },
     {
       "epoch": 8.0,
-      "eval_accuracy": 0.5612820512820513,
-      "eval_loss": 1.9008216857910156,
-      "eval_runtime": 5.1357,
-      "eval_samples_per_second": 97.358,
-      "eval_steps_per_second": 12.267,
+      "eval_accuracy": 0.6025641025641025,
+      "eval_loss": 1.6359158754348755,
+      "eval_runtime": 4.3161,
+      "eval_samples_per_second": 115.844,
+      "eval_steps_per_second": 14.596,
       "step": 1500
     },
     {
       "epoch": 8.0,
-      "eval_exact_match": 15.2,
-      "eval_f1": 23.12063492063493,
+      "eval_exact_match": 20.6,
+      "eval_f1": 33.70507936507938,
       "step": 1500
     },
     {
-      "epoch": 8.533333333333333,
-      "grad_norm": 3.382610321044922,
+      "epoch": 8.53,
       "learning_rate": 3e-05,
-      "loss": 0.8235,
+      "loss": 0.5733,
       "step": 1600
     },
     {
-      "epoch": 8.997333333333334,
-      "eval_accuracy": 0.5592307692307692,
-      "eval_loss": 2.026776075363159,
-      "eval_runtime": 5.5056,
-      "eval_samples_per_second": 90.817,
-      "eval_steps_per_second": 11.443,
+      "epoch": 9.0,
+      "eval_accuracy": 0.6006153846153847,
+      "eval_loss": 1.7149169445037842,
+      "eval_runtime": 4.1084,
+      "eval_samples_per_second": 121.702,
+      "eval_steps_per_second": 15.334,
       "step": 1687
     },
     {
-      "epoch": 8.997333333333334,
-      "eval_exact_match": 13.8,
-      "eval_f1": 21.42285714285715,
+      "epoch": 9.0,
+      "eval_exact_match": 22.0,
+      "eval_f1": 33.980476190476196,
       "step": 1687
     },
     {
-      "epoch": 9.066666666666666,
-      "grad_norm": 4.363819122314453,
+      "epoch": 9.07,
       "learning_rate": 3e-05,
-      "loss": 0.8547,
+      "loss": 0.5859,
       "step": 1700
     },
     {
       "epoch": 9.6,
-      "grad_norm": 3.685805320739746,
       "learning_rate": 3e-05,
-      "loss": 0.7635,
+      "loss": 0.5336,
       "step": 1800
     },
     {
       "epoch": 10.0,
-      "eval_accuracy": 0.5567692307692308,
-      "eval_loss": 2.0937435626983643,
-      "eval_runtime": 5.235,
-      "eval_samples_per_second": 95.512,
-      "eval_steps_per_second": 12.034,
+      "eval_accuracy": 0.5989230769230769,
+      "eval_loss": 1.8006359338760376,
+      "eval_runtime": 4.4039,
+      "eval_samples_per_second": 113.535,
+      "eval_steps_per_second": 14.305,
       "step": 1875
     },
     {
       "epoch": 10.0,
-      "eval_exact_match": 15.2,
-      "eval_f1": 22.031904761904762,
+      "eval_exact_match": 20.4,
+      "eval_f1": 32.68793650793652,
       "step": 1875
     },
     {
-      "epoch": 10.133333333333333,
-      "grad_norm": 3.5318357944488525,
+      "epoch": 10.13,
       "learning_rate": 3e-05,
-      "loss": 0.7554,
+      "loss": 0.5379,
       "step": 1900
     },
     {
-      "epoch": 10.666666666666666,
-      "grad_norm": 3.6454901695251465,
+      "epoch": 10.67,
       "learning_rate": 3e-05,
-      "loss": 0.6978,
+      "loss": 0.5116,
       "step": 2000
     },
     {
-      "epoch": 10.997333333333334,
-      "eval_accuracy": 0.556974358974359,
-      "eval_loss": 2.1557741165161133,
-      "eval_runtime": 5.1556,
-      "eval_samples_per_second": 96.981,
-      "eval_steps_per_second": 12.22,
+      "epoch": 11.0,
+      "eval_accuracy": 0.5982051282051282,
+      "eval_loss": 1.8851306438446045,
+      "eval_runtime": 4.8059,
+      "eval_samples_per_second": 104.038,
+      "eval_steps_per_second": 13.109,
       "step": 2062
     },
     {
-      "epoch": 10.997333333333334,
-      "eval_exact_match": 15.0,
-      "eval_f1": 23.010000000000012,
+      "epoch": 11.0,
+      "eval_exact_match": 19.8,
+      "eval_f1": 31.627460317460333,
       "step": 2062
     },
     {
       "epoch": 11.2,
-      "grad_norm": 3.658707618713379,
       "learning_rate": 3e-05,
-      "loss": 0.6969,
+      "loss": 0.5025,
       "step": 2100
     },
     {
-      "epoch": 11.733333333333333,
-      "grad_norm": 3.7350716590881348,
+      "epoch": 11.73,
       "learning_rate": 3e-05,
-      "loss": 0.6615,
+      "loss": 0.4934,
       "step": 2200
     },
     {
       "epoch": 12.0,
-      "eval_accuracy": 0.5551794871794872,
-      "eval_loss": 2.239973545074463,
-      "eval_runtime": 4.9026,
-      "eval_samples_per_second": 101.986,
-      "eval_steps_per_second": 12.85,
+      "eval_accuracy": 0.5982051282051282,
+      "eval_loss": 1.9262274503707886,
+      "eval_runtime": 4.2024,
+      "eval_samples_per_second": 118.98,
+      "eval_steps_per_second": 14.992,
       "step": 2250
     },
     {
       "epoch": 12.0,
-      "eval_exact_match": 15.6,
-      "eval_f1": 23.333492063492066,
+      "eval_exact_match": 19.2,
+      "eval_f1": 30.531746031746042,
       "step": 2250
     },
     {
-      "epoch": 12.266666666666667,
-      "grad_norm": 4.4434709548950195,
+      "epoch": 12.27,
       "learning_rate": 3e-05,
-      "loss": 0.632,
+      "loss": 0.4827,
       "step": 2300
     },
     {
       "epoch": 12.8,
-      "grad_norm": 3.4198718070983887,
       "learning_rate": 3e-05,
-      "loss": 0.6262,
+      "loss": 0.4823,
       "step": 2400
     },
     {
-      "epoch": 12.997333333333334,
-      "eval_accuracy": 0.5556410256410257,
-      "eval_loss": 2.268662452697754,
-      "eval_runtime": 5.327,
-      "eval_samples_per_second": 93.862,
-      "eval_steps_per_second": 11.827,
+      "epoch": 13.0,
+      "eval_accuracy": 0.5974358974358974,
+      "eval_loss": 1.9412546157836914,
+      "eval_runtime": 4.4921,
+      "eval_samples_per_second": 111.306,
+      "eval_steps_per_second": 14.025,
       "step": 2437
     },
     {
-      "epoch": 12.997333333333334,
-      "eval_exact_match": 14.4,
-      "eval_f1": 22.583492063492077,
+      "epoch": 13.0,
+      "eval_exact_match": 18.8,
+      "eval_f1": 30.277619047619062,
       "step": 2437
     },
     {
-      "epoch": 13.333333333333334,
-      "grad_norm": 3.158496618270874,
+      "epoch": 13.33,
       "learning_rate": 3e-05,
-      "loss": 0.5973,
+      "loss": 0.467,
       "step": 2500
     },
     {
-      "epoch": 13.866666666666667,
-      "grad_norm": 3.5720713138580322,
+      "epoch": 13.87,
       "learning_rate": 3e-05,
-      "loss": 0.5958,
+      "loss": 0.47,
       "step": 2600
     },
     {
       "epoch": 14.0,
-      "eval_accuracy": 0.5536923076923077,
-      "eval_loss": 2.358220100402832,
-      "eval_runtime": 5.236,
-      "eval_samples_per_second": 95.493,
-      "eval_steps_per_second": 12.032,
+      "eval_accuracy": 0.5967179487179487,
+      "eval_loss": 2.01212739944458,
+      "eval_runtime": 4.0908,
+      "eval_samples_per_second": 122.226,
+      "eval_steps_per_second": 15.4,
       "step": 2625
     },
     {
       "epoch": 14.0,
-      "eval_exact_match": 14.6,
-      "eval_f1": 21.79904761904763,
+      "eval_exact_match": 18.8,
+      "eval_f1": 30.478571428571446,
       "step": 2625
     },
     {
       "epoch": 14.4,
-      "grad_norm": 5.126227855682373,
       "learning_rate": 3e-05,
-      "loss": 0.5692,
+      "loss": 0.454,
       "step": 2700
     },
     {
-      "epoch": 14.933333333333334,
-      "grad_norm": 3.3391811847686768,
+      "epoch": 14.93,
       "learning_rate": 3e-05,
-      "loss": 0.5778,
+      "loss": 0.4661,
       "step": 2800
     },
     {
-      "epoch": 14.997333333333334,
-      "eval_accuracy": 0.5533846153846154,
-      "eval_loss": 2.395972728729248,
-      "eval_runtime": 5.1135,
-      "eval_samples_per_second": 97.78,
-      "eval_steps_per_second": 12.32,
+      "epoch": 15.0,
+      "eval_accuracy": 0.5968205128205128,
+      "eval_loss": 2.0249853134155273,
+      "eval_runtime": 4.0875,
+      "eval_samples_per_second": 122.323,
+      "eval_steps_per_second": 15.413,
       "step": 2812
     },
     {
-      "epoch": 14.997333333333334,
-      "eval_exact_match": 14.6,
-      "eval_f1": 22.537301587301595,
+      "epoch": 15.0,
+      "eval_exact_match": 20.6,
+      "eval_f1": 32.526031746031755,
       "step": 2812
     },
     {
-      "epoch": 15.466666666666667,
-      "grad_norm": 2.8773677349090576,
+      "epoch": 15.47,
       "learning_rate": 3e-05,
-      "loss": 0.5461,
+      "loss": 0.4423,
       "step": 2900
     },
     {
       "epoch": 16.0,
-      "grad_norm": 3.580195903778076,
       "learning_rate": 3e-05,
-      "loss": 0.5661,
+      "loss": 0.462,
       "step": 3000
     },
     {
       "epoch": 16.0,
-      "eval_accuracy": 0.5533846153846154,
-      "eval_loss": 2.4321775436401367,
-      "eval_runtime": 5.0238,
-      "eval_samples_per_second": 99.527,
-      "eval_steps_per_second": 12.54,
+      "eval_accuracy": 0.598974358974359,
+      "eval_loss": 1.9804810285568237,
+      "eval_runtime": 4.0893,
+      "eval_samples_per_second": 122.269,
+      "eval_steps_per_second": 15.406,
       "step": 3000
     },
     {
       "epoch": 16.0,
-      "eval_exact_match": 14.0,
-      "eval_f1": 22.300808080808093,
+      "eval_exact_match": 19.8,
+      "eval_f1": 31.533809523809534,
       "step": 3000
     },
     {
-      "epoch": 16.533333333333335,
-      "grad_norm": 3.6829302310943604,
+      "epoch": 16.53,
       "learning_rate": 3e-05,
-      "loss": 0.5277,
+      "loss": 0.4357,
       "step": 3100
     },
     {
-      "epoch": 16.997333333333334,
-      "eval_accuracy": 0.5515384615384615,
-      "eval_loss": 2.482792615890503,
-      "eval_runtime": 5.3006,
-      "eval_samples_per_second": 94.328,
-      "eval_steps_per_second": 11.885,
+      "epoch": 17.0,
+      "eval_accuracy": 0.5975897435897436,
+      "eval_loss": 2.0656166076660156,
+      "eval_runtime": 4.1949,
+      "eval_samples_per_second": 119.193,
+      "eval_steps_per_second": 15.018,
       "step": 3187
     },
     {
-      "epoch": 16.997333333333334,
-      "eval_exact_match": 13.4,
-      "eval_f1": 21.439682539682543,
+      "epoch": 17.0,
+      "eval_exact_match": 18.8,
+      "eval_f1": 31.295238095238105,
       "step": 3187
     },
     {
-      "epoch": 17.066666666666666,
-      "grad_norm": 3.1766672134399414,
+      "epoch": 17.07,
       "learning_rate": 3e-05,
-      "loss": 0.5475,
+      "loss": 0.4511,
       "step": 3200
     },
     {
       "epoch": 17.6,
-      "grad_norm": 3.070977210998535,
       "learning_rate": 3e-05,
-      "loss": 0.5211,
+      "loss": 0.4348,
       "step": 3300
     },
     {
       "epoch": 18.0,
-      "eval_accuracy": 0.5516410256410257,
-      "eval_loss": 2.510619640350342,
-      "eval_runtime": 5.0519,
-      "eval_samples_per_second": 98.973,
-      "eval_steps_per_second": 12.471,
+      "eval_accuracy": 0.5978974358974359,
+      "eval_loss": 2.030846118927002,
+      "eval_runtime": 4.1974,
+      "eval_samples_per_second": 119.122,
+      "eval_steps_per_second": 15.009,
       "step": 3375
     },
     {
       "epoch": 18.0,
-      "eval_exact_match": 12.6,
-      "eval_f1": 20.136349206349216,
+      "eval_exact_match": 19.6,
+      "eval_f1": 31.137619047619065,
       "step": 3375
     },
     {
-      "epoch": 18.133333333333333,
-      "grad_norm": 4.728194713592529,
+      "epoch": 18.13,
       "learning_rate": 3e-05,
-      "loss": 0.5344,
+      "loss": 0.4436,
       "step": 3400
     },
     {
-      "epoch": 18.666666666666668,
-      "grad_norm": 3.1506965160369873,
+      "epoch": 18.67,
       "learning_rate": 3e-05,
-      "loss": 0.5189,
+      "loss": 0.4331,
       "step": 3500
     },
     {
-      "epoch": 18.997333333333334,
-      "eval_accuracy": 0.5514871794871795,
-      "eval_loss": 2.5705809593200684,
-      "eval_runtime": 4.9081,
-      "eval_samples_per_second": 101.872,
-      "eval_steps_per_second": 12.836,
+      "epoch": 19.0,
+      "eval_accuracy": 0.598974358974359,
+      "eval_loss": 2.062859058380127,
+      "eval_runtime": 4.7948,
+      "eval_samples_per_second": 104.279,
+      "eval_steps_per_second": 13.139,
       "step": 3562
     },
     {
-      "epoch": 18.997333333333334,
-      "eval_exact_match": 13.8,
-      "eval_f1": 21.51634920634921,
+      "epoch": 19.0,
+      "eval_exact_match": 20.6,
+      "eval_f1": 32.42380952380953,
       "step": 3562
     },
     {
       "epoch": 19.2,
-      "grad_norm": 2.7202513217926025,
       "learning_rate": 3e-05,
-      "loss": 0.5198,
+      "loss": 0.4338,
       "step": 3600
     },
     {
-      "epoch": 19.733333333333334,
-      "grad_norm": 2.581169366836548,
+      "epoch": 19.73,
       "learning_rate": 3e-05,
-      "loss": 0.5166,
+      "loss": 0.4341,
       "step": 3700
     },
     {
       "epoch": 20.0,
-      "eval_accuracy": 0.5526153846153846,
-      "eval_loss": 2.542161226272583,
-      "eval_runtime": 5.0196,
-      "eval_samples_per_second": 99.609,
-      "eval_steps_per_second": 12.551,
+      "eval_accuracy": 0.5983076923076923,
+      "eval_loss": 2.0815415382385254,
+      "eval_runtime": 4.4041,
+      "eval_samples_per_second": 113.531,
+      "eval_steps_per_second": 14.305,
       "step": 3750
     },
     {
       "epoch": 20.0,
-      "eval_exact_match": 14.4,
-      "eval_f1": 21.563492063492074,
+      "eval_exact_match": 21.6,
+      "eval_f1": 33.345238095238116,
       "step": 3750
     },
     {
-      "epoch": 20.266666666666666,
-      "grad_norm": 2.6741623878479004,
+      "epoch": 20.27,
       "learning_rate": 3e-05,
-      "loss": 0.5112,
+      "loss": 0.4316,
       "step": 3800
     },
     {
       "epoch": 20.8,
-      "grad_norm": 3.241607666015625,
       "learning_rate": 3e-05,
-      "loss": 0.5132,
+      "loss": 0.434,
       "step": 3900
     },
     {
-      "epoch": 20.997333333333334,
-      "eval_accuracy": 0.550923076923077,
-      "eval_loss": 2.594764232635498,
-      "eval_runtime": 5.2553,
-      "eval_samples_per_second": 95.142,
-      "eval_steps_per_second": 11.988,
+      "epoch": 21.0,
+      "eval_accuracy": 0.5968205128205128,
+      "eval_loss": 2.1252710819244385,
+      "eval_runtime": 4.3531,
+      "eval_samples_per_second": 114.862,
+      "eval_steps_per_second": 14.473,
       "step": 3937
     },
     {
-      "epoch": 20.997333333333334,
-      "eval_exact_match": 13.4,
-      "eval_f1": 21.25015873015874,
+      "epoch": 21.0,
+      "eval_exact_match": 21.4,
+      "eval_f1": 32.32285714285715,
       "step": 3937
     },
     {
-      "epoch": 21.333333333333332,
-      "grad_norm": 2.6324737071990967,
+      "epoch": 21.33,
       "learning_rate": 3e-05,
-      "loss": 0.5011,
+      "loss": 0.4255,
       "step": 4000
     },
     {
-      "epoch": 21.866666666666667,
-      "grad_norm": 3.572909116744995,
+      "epoch": 21.87,
       "learning_rate": 3e-05,
-      "loss": 0.5115,
+      "loss": 0.4335,
       "step": 4100
     },
     {
       "epoch": 22.0,
-      "eval_accuracy": 0.5512307692307692,
-      "eval_loss": 2.604835033416748,
-      "eval_runtime": 5.1325,
-      "eval_samples_per_second": 97.419,
-      "eval_steps_per_second": 12.275,
+      "eval_accuracy": 0.5971282051282051,
+      "eval_loss": 2.178868293762207,
+      "eval_runtime": 4.1999,
+      "eval_samples_per_second": 119.05,
+      "eval_steps_per_second": 15.0,
       "step": 4125
     },
     {
       "epoch": 22.0,
-      "eval_exact_match": 14.2,
-      "eval_f1": 21.673968253968265,
+      "eval_exact_match": 21.2,
+      "eval_f1": 31.35571428571428,
       "step": 4125
     },
     {
       "epoch": 22.4,
-      "grad_norm": 2.303550958633423,
       "learning_rate": 3e-05,
-      "loss": 0.4973,
+      "loss": 0.4201,
       "step": 4200
     },
     {
-      "epoch": 22.933333333333334,
-      "grad_norm": 2.5230953693389893,
+      "epoch": 22.93,
       "learning_rate": 3e-05,
-      "loss": 0.5083,
+      "loss": 0.4346,
       "step": 4300
     },
     {
-      "epoch": 22.997333333333334,
-      "eval_accuracy": 0.5520512820512821,
-      "eval_loss": 2.5811188220977783,
-      "eval_runtime": 5.2251,
-      "eval_samples_per_second": 95.692,
-      "eval_steps_per_second": 12.057,
+      "epoch": 23.0,
+      "eval_accuracy": 0.5952307692307692,
+      "eval_loss": 2.145519971847534,
+      "eval_runtime": 4.0955,
+      "eval_samples_per_second": 122.084,
+      "eval_steps_per_second": 15.383,
       "step": 4312
     },
     {
-      "epoch": 22.997333333333334,
-      "eval_exact_match": 14.6,
-      "eval_f1": 22.511111111111124,
+      "epoch": 23.0,
+      "eval_exact_match": 21.0,
+      "eval_f1": 32.04714285714286,
       "step": 4312
     },
     {
-      "epoch": 23.466666666666665,
-      "grad_norm": 2.1872711181640625,
+      "epoch": 23.47,
       "learning_rate": 3e-05,
-      "loss": 0.49,
+      "loss": 0.4177,
       "step": 4400
     },
     {
       "epoch": 24.0,
-      "grad_norm": 2.1276657581329346,
       "learning_rate": 3e-05,
-      "loss": 0.5081,
+      "loss": 0.4326,
       "step": 4500
     },
     {
       "epoch": 24.0,
-      "eval_accuracy": 0.5513333333333333,
-      "eval_loss": 2.5661673545837402,
-      "eval_runtime": 5.1065,
-      "eval_samples_per_second": 97.915,
-      "eval_steps_per_second": 12.337,
+      "eval_accuracy": 0.5971282051282051,
+      "eval_loss": 2.1990349292755127,
+      "eval_runtime": 4.0925,
+      "eval_samples_per_second": 122.176,
+      "eval_steps_per_second": 15.394,
       "step": 4500
     },
     {
       "epoch": 24.0,
-      "eval_exact_match": 15.2,
-      "eval_f1": 23.028547008547015,
+      "eval_exact_match": 19.6,
+      "eval_f1": 30.60428571428572,
       "step": 4500
     },
     {
-      "epoch": 24.533333333333335,
-      "grad_norm": 3.943235158920288,
+      "epoch": 24.53,
       "learning_rate": 3e-05,
-      "loss": 0.4862,
+      "loss": 0.4139,
       "step": 4600
     },
     {
-      "epoch": 24.997333333333334,
-      "eval_accuracy": 0.5522051282051282,
-      "eval_loss": 2.6429293155670166,
-      "eval_runtime": 4.8982,
-      "eval_samples_per_second": 102.078,
-      "eval_steps_per_second": 12.862,
+      "epoch": 25.0,
+      "eval_accuracy": 0.5976410256410256,
+      "eval_loss": 2.1889517307281494,
+      "eval_runtime": 4.1957,
+      "eval_samples_per_second": 119.17,
+      "eval_steps_per_second": 15.015,
       "step": 4687
     },
     {
-      "epoch": 24.997333333333334,
-      "eval_exact_match": 13.8,
-      "eval_f1": 21.402380952380966,
+      "epoch": 25.0,
+      "eval_exact_match": 20.4,
+      "eval_f1": 31.052554112554116,
       "step": 4687
     },
     {
-      "epoch": 25.066666666666666,
-      "grad_norm": 4.37394380569458,
+      "epoch": 25.07,
       "learning_rate": 3e-05,
-      "loss": 0.4998,
+      "loss": 0.4268,
       "step": 4700
     },
     {
       "epoch": 25.6,
-      "grad_norm": 2.133417844772339,
       "learning_rate": 3e-05,
-      "loss": 0.4845,
+      "loss": 0.4139,
       "step": 4800
     },
     {
       "epoch": 26.0,
-      "eval_accuracy": 0.5533846153846154,
-      "eval_loss": 2.6020326614379883,
-      "eval_runtime": 5.0467,
-      "eval_samples_per_second": 99.075,
-      "eval_steps_per_second": 12.483,
+      "eval_accuracy": 0.5967692307692307,
+      "eval_loss": 2.1938648223876953,
+      "eval_runtime": 4.417,
+      "eval_samples_per_second": 113.198,
+      "eval_steps_per_second": 14.263,
       "step": 4875
     },
     {
       "epoch": 26.0,
-      "eval_exact_match": 13.8,
-      "eval_f1": 22.005714285714294,
+      "eval_exact_match": 20.4,
+      "eval_f1": 32.143030303030315,
       "step": 4875
     },
     {
-      "epoch": 26.133333333333333,
-      "grad_norm": 2.307835102081299,
+      "epoch": 26.13,
       "learning_rate": 3e-05,
-      "loss": 0.4967,
+      "loss": 0.4222,
       "step": 4900
     },
     {
-      "epoch": 26.666666666666668,
-      "grad_norm": 2.352783203125,
+      "epoch": 26.67,
       "learning_rate": 3e-05,
-      "loss": 0.4869,
+      "loss": 0.4162,
       "step": 5000
     },
     {
-      "epoch": 26.997333333333334,
-      "eval_accuracy": 0.5521538461538461,
-      "eval_loss": 2.633915424346924,
-      "eval_runtime": 5.2178,
-      "eval_samples_per_second": 95.826,
-      "eval_steps_per_second": 12.074,
+      "epoch": 27.0,
+      "eval_accuracy": 0.5964615384615385,
+      "eval_loss": 2.218991279602051,
+      "eval_runtime": 4.2318,
+      "eval_samples_per_second": 118.152,
+      "eval_steps_per_second": 14.887,
       "step": 5062
     },
     {
-      "epoch": 26.997333333333334,
-      "eval_exact_match": 13.6,
-      "eval_f1": 21.69952380952382,
+      "epoch": 27.0,
+      "eval_exact_match": 21.2,
+      "eval_f1": 31.969523809523817,
       "step": 5062
     },
     {
       "epoch": 27.2,
-      "grad_norm": 3.5862042903900146,
       "learning_rate": 3e-05,
-      "loss": 0.4863,
+      "loss": 0.419,
       "step": 5100
     },
     {
-      "epoch": 27.733333333333334,
-      "grad_norm": 2.1928396224975586,
+      "epoch": 27.73,
       "learning_rate": 3e-05,
-      "loss": 0.4862,
+      "loss": 0.4177,
       "step": 5200
     },
     {
       "epoch": 28.0,
-      "eval_accuracy": 0.5524102564102564,
-      "eval_loss": 2.6161999702453613,
-      "eval_runtime": 5.0462,
-      "eval_samples_per_second": 99.085,
-      "eval_steps_per_second": 12.485,
+      "eval_accuracy": 0.5954871794871794,
+      "eval_loss": 2.2781052589416504,
+      "eval_runtime": 4.1923,
+      "eval_samples_per_second": 119.265,
+      "eval_steps_per_second": 15.027,
       "step": 5250
     },
     {
       "epoch": 28.0,
-      "eval_exact_match": 14.8,
-      "eval_f1": 22.126190476190487,
+      "eval_exact_match": 19.8,
+      "eval_f1": 30.879220779220788,
       "step": 5250
     },
     {
-      "epoch": 28.266666666666666,
-      "grad_norm": 2.0402398109436035,
+      "epoch": 28.27,
       "learning_rate": 3e-05,
-      "loss": 0.4836,
+      "loss": 0.4143,
       "step": 5300
     },
     {
       "epoch": 28.8,
-      "grad_norm": 2.1713831424713135,
       "learning_rate": 3e-05,
-      "loss": 0.4856,
+      "loss": 0.4173,
       "step": 5400
     },
     {
-      "epoch": 28.997333333333334,
-      "eval_accuracy": 0.5525641025641026,
-      "eval_loss": 2.6764330863952637,
-      "eval_runtime": 5.9146,
-      "eval_samples_per_second": 84.537,
-      "eval_steps_per_second": 10.652,
+      "epoch": 29.0,
+      "eval_accuracy": 0.5975897435897436,
+      "eval_loss": 2.2680609226226807,
+      "eval_runtime": 4.401,
+      "eval_samples_per_second": 113.61,
+      "eval_steps_per_second": 14.315,
       "step": 5437
     },
     {
-      "epoch": 28.997333333333334,
-      "eval_exact_match": 14.2,
-      "eval_f1": 21.448412698412714,
+      "epoch": 29.0,
+      "eval_exact_match": 20.6,
+      "eval_f1": 31.83809523809525,
       "step": 5437
     },
     {
-      "epoch": 29.333333333333332,
-      "grad_norm": 2.2155110836029053,
+      "epoch": 29.33,
       "learning_rate": 3e-05,
-      "loss": 0.4786,
+      "loss": 0.4117,
       "step": 5500
     },
     {
-      "epoch": 29.866666666666667,
-      "grad_norm": 1.7778940200805664,
+      "epoch": 29.87,
       "learning_rate": 3e-05,
-      "loss": 0.4871,
+      "loss": 0.4187,
       "step": 5600
     },
     {
       "epoch": 30.0,
-      "eval_accuracy": 0.5525641025641026,
-      "eval_loss": 2.6703062057495117,
-      "eval_runtime": 4.8938,
-      "eval_samples_per_second": 102.171,
-      "eval_steps_per_second": 12.874,
+      "eval_accuracy": 0.5958974358974359,
+      "eval_loss": 2.2996103763580322,
+      "eval_runtime": 4.199,
+      "eval_samples_per_second": 119.077,
+      "eval_steps_per_second": 15.004,
       "step": 5625
     },
     {
       "epoch": 30.0,
-      "eval_exact_match": 14.2,
-      "eval_f1": 20.916825396825406,
+      "eval_exact_match": 20.4,
+      "eval_f1": 32.61430014430016,
       "step": 5625
     },
     {
       "epoch": 30.4,
-      "grad_norm": 2.07055401802063,
       "learning_rate": 3e-05,
-      "loss": 0.475,
+      "loss": 0.408,
       "step": 5700
     },
     {
-      "epoch": 30.933333333333334,
-      "grad_norm": 1.788883924484253,
+      "epoch": 30.93,
       "learning_rate": 3e-05,
-      "loss": 0.4863,
+      "loss": 0.4199,
       "step": 5800
     },
     {
-      "epoch": 30.997333333333334,
-      "eval_accuracy": 0.5533333333333333,
-      "eval_loss": 2.6787095069885254,
-      "eval_runtime": 5.0298,
-      "eval_samples_per_second": 99.407,
-      "eval_steps_per_second": 12.525,
+      "epoch": 31.0,
+      "eval_accuracy": 0.5981025641025641,
+      "eval_loss": 2.239492177963257,
+      "eval_runtime": 4.8298,
+      "eval_samples_per_second": 103.523,
+      "eval_steps_per_second": 13.044,
       "step": 5812
     },
     {
-      "epoch": 30.997333333333334,
-      "eval_exact_match": 14.0,
-      "eval_f1": 21.74063492063493,
+      "epoch": 31.0,
+      "eval_exact_match": 20.2,
+      "eval_f1": 31.465887445887457,
       "step": 5812
     },
     {
-      "epoch": 31.466666666666665,
-      "grad_norm": 1.6789844036102295,
+      "epoch": 31.47,
       "learning_rate": 3e-05,
-      "loss": 0.4693,
+      "loss": 0.4055,
       "step": 5900
     },
     {
       "epoch": 32.0,
-      "grad_norm": 1.622942566871643,
       "learning_rate": 3e-05,
-      "loss": 0.4884,
+      "loss": 0.4213,
       "step": 6000
     },
     {
       "epoch": 32.0,
-      "eval_accuracy": 0.5528205128205128,
-      "eval_loss": 2.6847567558288574,
-      "eval_runtime": 5.2277,
-      "eval_samples_per_second": 95.644,
-      "eval_steps_per_second": 12.051,
+      "eval_accuracy": 0.5957435897435898,
+      "eval_loss": 2.299053907394409,
+      "eval_runtime": 4.0871,
+      "eval_samples_per_second": 122.335,
+      "eval_steps_per_second": 15.414,
       "step": 6000
     },
     {
       "epoch": 32.0,
-      "eval_exact_match": 14.2,
-      "eval_f1": 21.513968253968265,
+      "eval_exact_match": 20.0,
+      "eval_f1": 31.56428571428572,
       "step": 6000
     },
     {
-      "epoch": 32.53333333333333,
-      "grad_norm": 1.7828670740127563,
+      "epoch": 32.53,
       "learning_rate": 3e-05,
-      "loss": 0.467,
+      "loss": 0.4015,
       "step": 6100
     },
     {
-      "epoch": 32.99733333333333,
-      "eval_accuracy": 0.5531282051282052,
-      "eval_loss": 2.668905735015869,
-      "eval_runtime": 5.7429,
-      "eval_samples_per_second": 87.064,
-      "eval_steps_per_second": 10.97,
+      "epoch": 33.0,
+      "eval_accuracy": 0.5952307692307692,
+      "eval_loss": 2.3223156929016113,
+      "eval_runtime": 4.1951,
+      "eval_samples_per_second": 119.187,
+      "eval_steps_per_second": 15.018,
       "step": 6187
     },
     {
-      "epoch": 32.99733333333333,
-      "eval_exact_match": 14.8,
-      "eval_f1": 22.606190476190484,
+      "epoch": 33.0,
+      "eval_exact_match": 21.2,
+      "eval_f1": 32.325714285714284,
       "step": 6187
     },
     {
-      "epoch": 33.06666666666667,
-      "grad_norm": 1.8942480087280273,
+      "epoch": 33.07,
       "learning_rate": 3e-05,
-      "loss": 0.4805,
+      "loss": 0.4161,
       "step": 6200
     },
     {
       "epoch": 33.6,
-      "grad_norm": 1.859069585800171,
       "learning_rate": 3e-05,
-      "loss": 0.4694,
+      "loss": 0.4058,
       "step": 6300
     },
     {
       "epoch": 34.0,
-      "eval_accuracy": 0.5524615384615384,
-      "eval_loss": 2.701338768005371,
-      "eval_runtime": 5.1485,
-      "eval_samples_per_second": 97.115,
-      "eval_steps_per_second": 12.237,
+      "eval_accuracy": 0.5956923076923077,
+      "eval_loss": 2.3266167640686035,
+      "eval_runtime": 4.0888,
+      "eval_samples_per_second": 122.286,
+      "eval_steps_per_second": 15.408,
       "step": 6375
     },
     {
       "epoch": 34.0,
-      "eval_exact_match": 13.4,
-      "eval_f1": 21.174920634920642,
+      "eval_exact_match": 20.2,
+      "eval_f1": 31.896666666666672,
       "step": 6375
     },
     {
-      "epoch": 34.13333333333333,
-      "grad_norm": 1.8238064050674438,
+      "epoch": 34.13,
       "learning_rate": 3e-05,
-      "loss": 0.476,
+      "loss": 0.4118,
       "step": 6400
     },
     {
-      "epoch": 34.666666666666664,
-      "grad_norm": 1.784118413925171,
+      "epoch": 34.67,
       "learning_rate": 3e-05,
-      "loss": 0.4712,
+      "loss": 0.4056,
       "step": 6500
     },
     {
-      "epoch": 34.99733333333333,
-      "eval_accuracy": 0.5521025641025641,
-      "eval_loss": 2.706479072570801,
-      "eval_runtime": 5.7012,
-      "eval_samples_per_second": 87.7,
-      "eval_steps_per_second": 11.05,
+      "epoch": 35.0,
+      "eval_accuracy": 0.5946153846153847,
+      "eval_loss": 2.3779330253601074,
+      "eval_runtime": 4.1964,
+      "eval_samples_per_second": 119.15,
+      "eval_steps_per_second": 15.013,
       "step": 6562
     },
     {
-      "epoch": 34.99733333333333,
-      "eval_exact_match": 13.2,
-      "eval_f1": 21.548253968253984,
+      "epoch": 35.0,
+      "eval_exact_match": 20.6,
+      "eval_f1": 32.23207792207793,
       "step": 6562
     },
     {
       "epoch": 35.2,
-      "grad_norm": 1.6702320575714111,
       "learning_rate": 3e-05,
-      "loss": 0.4711,
+      "loss": 0.4103,
       "step": 6600
     },
     {
-      "epoch": 35.733333333333334,
-      "grad_norm": 2.1222314834594727,
+      "epoch": 35.73,
       "learning_rate": 3e-05,
-      "loss": 0.4733,
+      "loss": 0.4078,
       "step": 6700
     },
     {
       "epoch": 36.0,
-      "eval_accuracy": 0.5522564102564103,
-      "eval_loss": 2.670701742172241,
-      "eval_runtime": 5.2238,
-      "eval_samples_per_second": 95.716,
-      "eval_steps_per_second": 12.06,
+      "eval_accuracy": 0.5950769230769231,
+      "eval_loss": 2.345280408859253,
+      "eval_runtime": 5.1209,
+      "eval_samples_per_second": 97.64,
+      "eval_steps_per_second": 12.303,
       "step": 6750
     },
     {
       "epoch": 36.0,
-      "eval_exact_match": 12.8,
-      "eval_f1": 20.31333333333334,
+      "eval_exact_match": 21.0,
+      "eval_f1": 32.03683982683983,
       "step": 6750
     },
     {
-      "epoch": 36.266666666666666,
-      "grad_norm": 1.5269900560379028,
+      "epoch": 36.27,
       "learning_rate": 3e-05,
-      "loss": 0.4678,
+      "loss": 0.4048,
       "step": 6800
     },
     {
       "epoch": 36.8,
-      "grad_norm": 1.7864423990249634,
       "learning_rate": 3e-05,
-      "loss": 0.4752,
+      "loss": 0.4097,
       "step": 6900
     },
     {
-      "epoch": 36.99733333333333,
-      "eval_accuracy": 0.5531794871794872,
-      "eval_loss": 2.675729513168335,
-      "eval_runtime": 6.0665,
-      "eval_samples_per_second": 82.42,
-      "eval_steps_per_second": 10.385,
+      "epoch": 37.0,
+      "eval_accuracy": 0.5965128205128205,
+      "eval_loss": 2.3379180431365967,
+      "eval_runtime": 4.1922,
+      "eval_samples_per_second": 119.268,
+      "eval_steps_per_second": 15.028,
       "step": 6937
     },
     {
-      "epoch": 36.99733333333333,
-      "eval_exact_match": 14.4,
-      "eval_f1": 22.22396825396826,
+      "epoch": 37.0,
+      "eval_exact_match": 20.6,
+      "eval_f1": 32.22937950937952,
       "step": 6937
     },
     {
-      "epoch": 37.333333333333336,
-      "grad_norm": 2.6186816692352295,
+      "epoch": 37.33,
       "learning_rate": 3e-05,
-      "loss": 0.4671,
+      "loss": 0.4059,
       "step": 7000
     },
     {
-      "epoch": 37.86666666666667,
-      "grad_norm": 2.1805689334869385,
+      "epoch": 37.87,
       "learning_rate": 3e-05,
-      "loss": 0.4744,
+      "loss": 0.4105,
       "step": 7100
     },
     {
       "epoch": 38.0,
-      "eval_accuracy": 0.5534358974358974,
-      "eval_loss": 2.701582431793213,
-      "eval_runtime": 4.964,
-      "eval_samples_per_second": 100.725,
-      "eval_steps_per_second": 12.691,
+      "eval_accuracy": 0.5969230769230769,
+      "eval_loss": 2.362399101257324,
+      "eval_runtime": 4.1963,
+      "eval_samples_per_second": 119.153,
+      "eval_steps_per_second": 15.013,
       "step": 7125
     },
     {
       "epoch": 38.0,
-      "eval_exact_match": 13.6,
-      "eval_f1": 21.16952380952382,
+      "eval_exact_match": 20.6,
+      "eval_f1": 31.836363636363643,
       "step": 7125
     },
     {
       "epoch": 38.4,
-      "grad_norm": 1.9267176389694214,
       "learning_rate": 3e-05,
-      "loss": 0.4611,
+      "loss": 0.4013,
       "step": 7200
     },
     {
-      "epoch": 38.93333333333333,
-      "grad_norm": 1.8288675546646118,
+      "epoch": 38.93,
       "learning_rate": 3e-05,
-      "loss": 0.4759,
+      "loss": 0.4116,
       "step": 7300
     },
     {
-      "epoch": 38.99733333333333,
-      "eval_accuracy": 0.5525641025641026,
-      "eval_loss": 2.726337432861328,
-      "eval_runtime": 5.6276,
-      "eval_samples_per_second": 88.847,
-      "eval_steps_per_second": 11.195,
+      "epoch": 39.0,
+      "eval_accuracy": 0.5961538461538461,
+      "eval_loss": 2.3846352100372314,
+      "eval_runtime": 4.0932,
+      "eval_samples_per_second": 122.154,
+      "eval_steps_per_second": 15.391,
       "step": 7312
     },
     {
-      "epoch": 38.99733333333333,
-      "eval_exact_match": 12.2,
-      "eval_f1": 19.783968253968265,
+      "epoch": 39.0,
+      "eval_exact_match": 20.0,
+      "eval_f1": 31.287619047619053,
       "step": 7312
     },
     {
-      "epoch": 39.46666666666667,
-      "grad_norm": 1.7793669700622559,
+      "epoch": 39.47,
       "learning_rate": 3e-05,
-      "loss": 0.4589,
+      "loss": 0.3993,
       "step": 7400
     },
     {
       "epoch": 40.0,
-      "grad_norm": 1.6059751510620117,
       "learning_rate": 3e-05,
-      "loss": 0.4759,
+      "loss": 0.4121,
       "step": 7500
     },
     {
       "epoch": 40.0,
-      "eval_accuracy": 0.5525128205128205,
-      "eval_loss": 2.736008405685425,
-      "eval_runtime": 5.6736,
-      "eval_samples_per_second": 88.128,
-      "eval_steps_per_second": 11.104,
+      "eval_accuracy": 0.5944615384615385,
+      "eval_loss": 2.3748481273651123,
+      "eval_runtime": 4.9211,
+      "eval_samples_per_second": 101.603,
+      "eval_steps_per_second": 12.802,
       "step": 7500
     },
     {
       "epoch": 40.0,
-      "eval_exact_match": 12.4,
-      "eval_f1": 20.80158730158731,
+      "eval_exact_match": 20.4,
+      "eval_f1": 32.21826839826841,
       "step": 7500
     },
     {
-      "epoch": 40.53333333333333,
-      "grad_norm": 1.7500640153884888,
+      "epoch": 40.53,
       "learning_rate": 3e-05,
-      "loss": 0.4569,
+      "loss": 0.3973,
       "step": 7600
     },
     {
-      "epoch": 40.99733333333333,
-      "eval_accuracy": 0.5524102564102564,
-      "eval_loss": 2.757992744445801,
-      "eval_runtime": 5.7367,
-      "eval_samples_per_second": 87.158,
-      "eval_steps_per_second": 10.982,
+      "epoch": 41.0,
+      "eval_accuracy": 0.5955897435897436,
+      "eval_loss": 2.3797402381896973,
+      "eval_runtime": 4.0885,
+      "eval_samples_per_second": 122.293,
+      "eval_steps_per_second": 15.409,
       "step": 7687
     },
     {
-      "epoch": 40.99733333333333,
-      "eval_exact_match": 12.8,
-      "eval_f1": 20.586666666666677,
+      "epoch": 41.0,
+      "eval_exact_match": 21.8,
+      "eval_f1": 33.78445887445889,
       "step": 7687
     },
     {
-      "epoch": 41.06666666666667,
-      "grad_norm": 3.21926212310791,
+      "epoch": 41.07,
       "learning_rate": 3e-05,
-      "loss": 0.469,
+      "loss": 0.4092,
       "step": 7700
     },
     {
       "epoch": 41.6,
-      "grad_norm": 1.447327971458435,
       "learning_rate": 3e-05,
-      "loss": 0.4585,
+      "loss": 0.3985,
       "step": 7800
     },
     {
       "epoch": 42.0,
-      "eval_accuracy": 0.5521025641025641,
-      "eval_loss": 2.7459263801574707,
-      "eval_runtime": 5.947,
-      "eval_samples_per_second": 84.076,
-      "eval_steps_per_second": 10.594,
+      "eval_accuracy": 0.5966666666666667,
+      "eval_loss": 2.359941005706787,
+      "eval_runtime": 4.319,
+      "eval_samples_per_second": 115.769,
+      "eval_steps_per_second": 14.587,
       "step": 7875
     },
     {
       "epoch": 42.0,
-      "eval_exact_match": 12.4,
-      "eval_f1": 19.576522366522376,
+      "eval_exact_match": 22.6,
+      "eval_f1": 34.189696969696975,
       "step": 7875
     },
     {
-      "epoch": 42.13333333333333,
-      "grad_norm": 1.6787067651748657,
+      "epoch": 42.13,
       "learning_rate": 3e-05,
-      "loss": 0.4688,
+      "loss": 0.4075,
       "step": 7900
     },
     {
-      "epoch": 42.666666666666664,
-      "grad_norm": 1.491296648979187,
+      "epoch": 42.67,
       "learning_rate": 3e-05,
-      "loss": 0.4602,
+      "loss": 0.4014,
       "step": 8000
     },
     {
-      "epoch": 42.99733333333333,
-      "eval_accuracy": 0.5521538461538461,
-      "eval_loss": 2.7965104579925537,
-      "eval_runtime": 4.909,
-      "eval_samples_per_second": 101.854,
-      "eval_steps_per_second": 12.834,
+      "epoch": 43.0,
+      "eval_accuracy": 0.5970769230769231,
+      "eval_loss": 2.347489356994629,
+      "eval_runtime": 4.0917,
+      "eval_samples_per_second": 122.197,
+      "eval_steps_per_second": 15.397,
       "step": 8062
     },
     {
-      "epoch": 42.99733333333333,
-      "eval_exact_match": 13.0,
-      "eval_f1": 20.853015873015874,
+      "epoch": 43.0,
+      "eval_exact_match": 21.4,
+      "eval_f1": 33.01255411255412,
       "step": 8062
     },
     {
       "epoch": 43.2,
-      "grad_norm": 1.7218726873397827,
       "learning_rate": 3e-05,
-      "loss": 0.4635,
+      "loss": 0.4032,
       "step": 8100
     },
     {
-      "epoch": 43.733333333333334,
-      "grad_norm": 1.5168484449386597,
+      "epoch": 43.73,
       "learning_rate": 3e-05,
-      "loss": 0.4631,
+      "loss": 0.4032,
       "step": 8200
     },
     {
       "epoch": 44.0,
-      "eval_accuracy": 0.5515897435897436,
-      "eval_loss": 2.799527645111084,
-      "eval_runtime": 5.0542,
-      "eval_samples_per_second": 98.928,
-      "eval_steps_per_second": 12.465,
+      "eval_accuracy": 0.5987179487179487,
+      "eval_loss": 2.3937387466430664,
+      "eval_runtime": 4.3012,
+      "eval_samples_per_second": 116.246,
+      "eval_steps_per_second": 14.647,
       "step": 8250
     },
     {
       "epoch": 44.0,
-      "eval_exact_match": 13.2,
-      "eval_f1": 20.908253968253973,
+      "eval_exact_match": 20.6,
+      "eval_f1": 31.90666666666667,
       "step": 8250
     }
   ],
   "logging_steps": 100,
   "max_steps": 9350,
-  "num_input_tokens_seen": 0,
   "num_train_epochs": 50,
   "save_steps": 500,
-  "stateful_callbacks": {
-    "TrainerControl": {
-      "args": {
-        "should_epoch_stop": false,
-        "should_evaluate": false,
-        "should_log": false,
-        "should_save": true,
-        "should_training_stop": false
-      },
-      "attributes": {}
-    }
-  },
-  "total_flos": 1.5113268177246618e+17,
-  "train_batch_size": 2,
+  "total_flos": 3.0785373765999e+17,
   "trial_name": null,
   "trial_params": null
 }

checkpoint-8250/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:1a175de730fa6917b3a611bb9e53766640b7a5e4a426f825f2a602cf98661e47
-size 5304
+oid sha256:52c53f755ad4d3669bc1e92a0dd04ac5e6ef02b39406da28fac147ebfc373982
+size 4728