Training in progress, epoch 42, checkpoint

checkpoint-8062/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-8062/adapter_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:80e16408a9a5e0da9815fb829801d18a45e32b68e9c75b892a351c5b1b4b2d95
+size 143269386

checkpoint-8062/added_tokens.json

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

checkpoint-8062/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6a96e56a4f4e6baa99ebf2428c80843c7a46e791710907e8bebe1b4cf9263488
-size 224537202
+oid sha256:3e83d63874c7899647669037516db40e07cac5f2ac0bc36ed31620bc48036e66
+size 286585234

checkpoint-8062/rng_state_0.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:5f8d73123255eb09500c770746e4f3aab43eda528bf72a657523e21a6ef35cfb
+oid sha256:d74144de1c01009da204c2e53c691471c2f1bbdb8656c1e1045b96fb1c74f3e2
 size 15024

checkpoint-8062/rng_state_1.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a94fb6728fca3f4d4348d170df360f3bd8fcc2bdb4cccbcdf9d11a14b280e280
+oid sha256:9a4804c0c8ea62e1805f735d7c948d4673fbe03428358f5bcb6de96a95eb598f
 size 15024

checkpoint-8062/rng_state_2.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:eb72d4b88a93891eec31e77831e68e49b3a0a97b717fe079aedad2c21c28daae
+oid sha256:96d294f9dfcb36132c39bf9c8594dc77d18bdcead0c85272f313e3f6fa66090e
 size 15024

checkpoint-8062/rng_state_3.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:cc0b6c8c0e6f293d3679d25d43e766519d549c10558c25c9a3660ada4ebd4013
+oid sha256:1e42491973d49d72c8fbb1b85f85909cc8bf79eae21a3553ade3bf64715e1850
 size 15024

checkpoint-8062/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-8062/tokenizer.model

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

checkpoint-8062/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-8062/trainer_state.json

@@ -9,1230 +9,1136 @@
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.5333333333333333,
-      "grad_norm": 0.64143306016922,
+      "epoch": 0.53,
       "learning_rate": 0.0001,
-      "loss": 1.7779,
+      "loss": 1.598,
       "step": 100
     },
     {
-      "epoch": 0.9973333333333333,
-      "eval_accuracy": 0.5734871794871795,
-      "eval_loss": 1.6191272735595703,
-      "eval_runtime": 5.393,
-      "eval_samples_per_second": 92.713,
-      "eval_steps_per_second": 11.682,
+      "epoch": 1.0,
+      "eval_accuracy": 0.6146666666666667,
+      "eval_loss": 1.2691738605499268,
+      "eval_runtime": 4.3939,
+      "eval_samples_per_second": 113.793,
+      "eval_steps_per_second": 14.338,
       "step": 187
     },
     {
-      "epoch": 0.9973333333333333,
-      "eval_exact_match": 18.0,
-      "eval_f1": 28.280476190476193,
+      "epoch": 1.0,
+      "eval_exact_match": 24.8,
+      "eval_f1": 38.10857142857143,
       "step": 187
     },
     {
-      "epoch": 1.0666666666666667,
-      "grad_norm": 0.7118939161300659,
+      "epoch": 1.07,
       "learning_rate": 0.0001,
-      "loss": 1.6349,
+      "loss": 1.2894,
       "step": 200
     },
     {
       "epoch": 1.6,
-      "grad_norm": 1.0694799423217773,
       "learning_rate": 0.0001,
-      "loss": 1.4887,
+      "loss": 1.1923,
       "step": 300
     },
     {
       "epoch": 2.0,
-      "eval_accuracy": 0.5735897435897436,
-      "eval_loss": 1.6172032356262207,
-      "eval_runtime": 5.0425,
-      "eval_samples_per_second": 99.158,
-      "eval_steps_per_second": 12.494,
+      "eval_accuracy": 0.6175897435897436,
+      "eval_loss": 1.2732893228530884,
+      "eval_runtime": 4.5482,
+      "eval_samples_per_second": 109.934,
+      "eval_steps_per_second": 13.852,
       "step": 375
     },
     {
       "epoch": 2.0,
-      "eval_exact_match": 19.8,
-      "eval_f1": 28.77095238095238,
+      "eval_exact_match": 25.6,
+      "eval_f1": 39.04746031746033,
       "step": 375
     },
     {
-      "epoch": 2.1333333333333333,
-      "grad_norm": 1.28153395652771,
+      "epoch": 2.13,
       "learning_rate": 0.0001,
-      "loss": 1.4304,
+      "loss": 1.1174,
       "step": 400
     },
     {
-      "epoch": 2.6666666666666665,
-      "grad_norm": 1.3190444707870483,
+      "epoch": 2.67,
       "learning_rate": 0.0001,
-      "loss": 1.2346,
+      "loss": 0.9732,
       "step": 500
     },
     {
-      "epoch": 2.997333333333333,
-      "eval_accuracy": 0.5706153846153846,
-      "eval_loss": 1.6950472593307495,
-      "eval_runtime": 5.1981,
-      "eval_samples_per_second": 96.188,
-      "eval_steps_per_second": 12.12,
+      "epoch": 3.0,
+      "eval_accuracy": 0.6135897435897436,
+      "eval_loss": 1.33958101272583,
+      "eval_runtime": 4.0988,
+      "eval_samples_per_second": 121.986,
+      "eval_steps_per_second": 15.37,
       "step": 562
     },
     {
-      "epoch": 2.997333333333333,
-      "eval_exact_match": 19.4,
-      "eval_f1": 28.241904761904756,
+      "epoch": 3.0,
+      "eval_exact_match": 25.4,
+      "eval_f1": 37.65698412698413,
       "step": 562
     },
     {
       "epoch": 3.2,
-      "grad_norm": 1.5040416717529297,
       "learning_rate": 0.0001,
-      "loss": 1.1583,
+      "loss": 0.9111,
       "step": 600
     },
     {
-      "epoch": 3.7333333333333334,
-      "grad_norm": 1.512290358543396,
+      "epoch": 3.73,
       "learning_rate": 0.0001,
-      "loss": 1.0104,
+      "loss": 0.7763,
       "step": 700
     },
     {
       "epoch": 4.0,
-      "eval_accuracy": 0.5655897435897436,
-      "eval_loss": 1.7944799661636353,
-      "eval_runtime": 5.3441,
-      "eval_samples_per_second": 93.561,
-      "eval_steps_per_second": 11.789,
+      "eval_accuracy": 0.6104102564102564,
+      "eval_loss": 1.4358161687850952,
+      "eval_runtime": 4.4028,
+      "eval_samples_per_second": 113.565,
+      "eval_steps_per_second": 14.309,
       "step": 750
     },
     {
       "epoch": 4.0,
-      "eval_exact_match": 15.8,
-      "eval_f1": 23.639523809523823,
+      "eval_exact_match": 25.2,
+      "eval_f1": 37.8809523809524,
       "step": 750
     },
     {
-      "epoch": 4.266666666666667,
-      "grad_norm": 1.698282241821289,
+      "epoch": 4.27,
       "learning_rate": 0.0001,
-      "loss": 0.9275,
+      "loss": 0.7118,
       "step": 800
     },
     {
       "epoch": 4.8,
-      "grad_norm": 1.6007401943206787,
       "learning_rate": 0.0001,
-      "loss": 0.843,
+      "loss": 0.6498,
       "step": 900
     },
     {
-      "epoch": 4.997333333333334,
-      "eval_accuracy": 0.5632820512820513,
-      "eval_loss": 1.9219905138015747,
-      "eval_runtime": 5.2659,
-      "eval_samples_per_second": 94.951,
-      "eval_steps_per_second": 11.964,
+      "epoch": 5.0,
+      "eval_accuracy": 0.6051794871794872,
+      "eval_loss": 1.5630384683609009,
+      "eval_runtime": 4.1079,
+      "eval_samples_per_second": 121.717,
+      "eval_steps_per_second": 15.336,
       "step": 937
     },
     {
-      "epoch": 4.997333333333334,
-      "eval_exact_match": 15.8,
-      "eval_f1": 24.631904761904778,
+      "epoch": 5.0,
+      "eval_exact_match": 23.4,
+      "eval_f1": 35.55111111111113,
       "step": 937
     },
     {
-      "epoch": 5.333333333333333,
-      "grad_norm": 1.5375592708587646,
+      "epoch": 5.33,
       "learning_rate": 0.0001,
-      "loss": 0.7516,
+      "loss": 0.5911,
       "step": 1000
     },
     {
-      "epoch": 5.866666666666667,
-      "grad_norm": 1.718989372253418,
+      "epoch": 5.87,
       "learning_rate": 0.0001,
-      "loss": 0.7199,
+      "loss": 0.57,
       "step": 1100
     },
     {
       "epoch": 6.0,
-      "eval_accuracy": 0.5616923076923077,
-      "eval_loss": 2.016322612762451,
-      "eval_runtime": 5.116,
-      "eval_samples_per_second": 97.734,
-      "eval_steps_per_second": 12.314,
+      "eval_accuracy": 0.6031282051282051,
+      "eval_loss": 1.6599360704421997,
+      "eval_runtime": 4.2005,
+      "eval_samples_per_second": 119.033,
+      "eval_steps_per_second": 14.998,
       "step": 1125
     },
     {
       "epoch": 6.0,
-      "eval_exact_match": 14.4,
-      "eval_f1": 22.39285714285715,
+      "eval_exact_match": 22.6,
+      "eval_f1": 34.44650793650795,
       "step": 1125
     },
     {
       "epoch": 6.4,
-      "grad_norm": 1.8337936401367188,
       "learning_rate": 0.0001,
-      "loss": 0.6378,
+      "loss": 0.517,
       "step": 1200
     },
     {
-      "epoch": 6.933333333333334,
-      "grad_norm": 1.9019217491149902,
+      "epoch": 6.93,
       "learning_rate": 0.0001,
-      "loss": 0.6432,
+      "loss": 0.5253,
       "step": 1300
     },
     {
-      "epoch": 6.997333333333334,
-      "eval_accuracy": 0.5616410256410257,
-      "eval_loss": 2.1283841133117676,
-      "eval_runtime": 5.1352,
-      "eval_samples_per_second": 97.367,
-      "eval_steps_per_second": 12.268,
+      "epoch": 7.0,
+      "eval_accuracy": 0.6027179487179487,
+      "eval_loss": 1.7479698657989502,
+      "eval_runtime": 4.4276,
+      "eval_samples_per_second": 112.929,
+      "eval_steps_per_second": 14.229,
       "step": 1312
     },
     {
-      "epoch": 6.997333333333334,
-      "eval_exact_match": 13.8,
-      "eval_f1": 22.042857142857148,
+      "epoch": 7.0,
+      "eval_exact_match": 22.2,
+      "eval_f1": 33.87303030303032,
       "step": 1312
     },
     {
-      "epoch": 7.466666666666667,
-      "grad_norm": 1.246907114982605,
+      "epoch": 7.47,
       "learning_rate": 0.0001,
-      "loss": 0.578,
+      "loss": 0.4784,
       "step": 1400
     },
     {
       "epoch": 8.0,
-      "grad_norm": 1.6256760358810425,
       "learning_rate": 0.0001,
-      "loss": 0.596,
+      "loss": 0.4958,
       "step": 1500
     },
     {
       "epoch": 8.0,
-      "eval_accuracy": 0.5603589743589743,
-      "eval_loss": 2.170966386795044,
-      "eval_runtime": 5.0636,
-      "eval_samples_per_second": 98.744,
-      "eval_steps_per_second": 12.442,
+      "eval_accuracy": 0.602051282051282,
+      "eval_loss": 1.8059930801391602,
+      "eval_runtime": 4.1898,
+      "eval_samples_per_second": 119.337,
+      "eval_steps_per_second": 15.036,
       "step": 1500
     },
     {
       "epoch": 8.0,
-      "eval_exact_match": 14.4,
-      "eval_f1": 21.71636363636364,
+      "eval_exact_match": 21.8,
+      "eval_f1": 32.95365079365081,
       "step": 1500
     },
     {
-      "epoch": 8.533333333333333,
-      "grad_norm": 1.3106721639633179,
+      "epoch": 8.53,
       "learning_rate": 0.0001,
-      "loss": 0.543,
+      "loss": 0.4521,
       "step": 1600
     },
     {
-      "epoch": 8.997333333333334,
-      "eval_accuracy": 0.5595384615384615,
-      "eval_loss": 2.262348175048828,
-      "eval_runtime": 5.3198,
-      "eval_samples_per_second": 93.989,
-      "eval_steps_per_second": 11.843,
+      "epoch": 9.0,
+      "eval_accuracy": 0.6013333333333334,
+      "eval_loss": 1.8598920106887817,
+      "eval_runtime": 4.9021,
+      "eval_samples_per_second": 101.996,
+      "eval_steps_per_second": 12.852,
       "step": 1687
     },
     {
-      "epoch": 8.997333333333334,
-      "eval_exact_match": 16.0,
-      "eval_f1": 24.34507936507938,
+      "epoch": 9.0,
+      "eval_exact_match": 22.6,
+      "eval_f1": 33.89858585858587,
       "step": 1687
     },
     {
-      "epoch": 9.066666666666666,
-      "grad_norm": 1.0499460697174072,
+      "epoch": 9.07,
       "learning_rate": 0.0001,
-      "loss": 0.5644,
+      "loss": 0.4694,
       "step": 1700
     },
     {
       "epoch": 9.6,
-      "grad_norm": 0.9217801690101624,
       "learning_rate": 0.0001,
-      "loss": 0.5314,
+      "loss": 0.443,
       "step": 1800
     },
     {
       "epoch": 10.0,
-      "eval_accuracy": 0.5583589743589744,
-      "eval_loss": 2.268125057220459,
-      "eval_runtime": 5.457,
-      "eval_samples_per_second": 91.626,
-      "eval_steps_per_second": 11.545,
+      "eval_accuracy": 0.6013333333333334,
+      "eval_loss": 1.9468085765838623,
+      "eval_runtime": 4.4033,
+      "eval_samples_per_second": 113.552,
+      "eval_steps_per_second": 14.308,
       "step": 1875
     },
     {
       "epoch": 10.0,
-      "eval_exact_match": 15.4,
-      "eval_f1": 23.46095238095238,
+      "eval_exact_match": 21.8,
+      "eval_f1": 32.7704761904762,
       "step": 1875
     },
     {
-      "epoch": 10.133333333333333,
-      "grad_norm": 0.969529926776886,
+      "epoch": 10.13,
       "learning_rate": 0.0001,
-      "loss": 0.534,
+      "loss": 0.4513,
       "step": 1900
     },
     {
-      "epoch": 10.666666666666666,
-      "grad_norm": 0.9540187120437622,
+      "epoch": 10.67,
       "learning_rate": 0.0001,
-      "loss": 0.5171,
+      "loss": 0.439,
       "step": 2000
     },
     {
-      "epoch": 10.997333333333334,
-      "eval_accuracy": 0.5584615384615385,
-      "eval_loss": 2.3312978744506836,
-      "eval_runtime": 5.2187,
-      "eval_samples_per_second": 95.81,
-      "eval_steps_per_second": 12.072,
+      "epoch": 11.0,
+      "eval_accuracy": 0.6014871794871794,
+      "eval_loss": 1.9500343799591064,
+      "eval_runtime": 4.0917,
+      "eval_samples_per_second": 122.197,
+      "eval_steps_per_second": 15.397,
       "step": 2062
     },
     {
-      "epoch": 10.997333333333334,
-      "eval_exact_match": 16.2,
-      "eval_f1": 23.968888888888895,
+      "epoch": 11.0,
+      "eval_exact_match": 22.4,
+      "eval_f1": 33.80857142857144,
       "step": 2062
     },
     {
       "epoch": 11.2,
-      "grad_norm": 0.8946124911308289,
       "learning_rate": 0.0001,
-      "loss": 0.5208,
+      "loss": 0.4372,
       "step": 2100
     },
     {
-      "epoch": 11.733333333333333,
-      "grad_norm": 0.8362706899642944,
+      "epoch": 11.73,
       "learning_rate": 0.0001,
-      "loss": 0.512,
+      "loss": 0.433,
       "step": 2200
     },
     {
       "epoch": 12.0,
-      "eval_accuracy": 0.559025641025641,
-      "eval_loss": 2.364483594894409,
-      "eval_runtime": 5.0957,
-      "eval_samples_per_second": 98.122,
-      "eval_steps_per_second": 12.363,
+      "eval_accuracy": 0.6021025641025641,
+      "eval_loss": 1.9103901386260986,
+      "eval_runtime": 4.4069,
+      "eval_samples_per_second": 113.459,
+      "eval_steps_per_second": 14.296,
       "step": 2250
     },
     {
       "epoch": 12.0,
-      "eval_exact_match": 16.0,
-      "eval_f1": 24.015714285714285,
+      "eval_exact_match": 21.4,
+      "eval_f1": 33.23952380952382,
       "step": 2250
     },
     {
-      "epoch": 12.266666666666667,
-      "grad_norm": 0.7387551665306091,
+      "epoch": 12.27,
       "learning_rate": 0.0001,
-      "loss": 0.5038,
+      "loss": 0.4301,
       "step": 2300
     },
     {
       "epoch": 12.8,
-      "grad_norm": 0.8488346338272095,
       "learning_rate": 0.0001,
-      "loss": 0.503,
+      "loss": 0.4323,
       "step": 2400
     },
     {
-      "epoch": 12.997333333333334,
-      "eval_accuracy": 0.5588205128205128,
-      "eval_loss": 2.3642191886901855,
-      "eval_runtime": 5.4807,
-      "eval_samples_per_second": 91.23,
-      "eval_steps_per_second": 11.495,
+      "epoch": 13.0,
+      "eval_accuracy": 0.6001025641025641,
+      "eval_loss": 2.0078930854797363,
+      "eval_runtime": 4.1244,
+      "eval_samples_per_second": 121.229,
+      "eval_steps_per_second": 15.275,
       "step": 2437
     },
     {
-      "epoch": 12.997333333333334,
-      "eval_exact_match": 16.0,
-      "eval_f1": 23.92238095238096,
+      "epoch": 13.0,
+      "eval_exact_match": 21.6,
+      "eval_f1": 33.54033189033191,
       "step": 2437
     },
     {
-      "epoch": 13.333333333333334,
-      "grad_norm": 0.9431668519973755,
+      "epoch": 13.33,
       "learning_rate": 0.0001,
-      "loss": 0.4941,
+      "loss": 0.4231,
       "step": 2500
     },
     {
-      "epoch": 13.866666666666667,
-      "grad_norm": 0.8775368928909302,
+      "epoch": 13.87,
       "learning_rate": 0.0001,
-      "loss": 0.4996,
+      "loss": 0.4281,
       "step": 2600
     },
     {
       "epoch": 14.0,
-      "eval_accuracy": 0.5582051282051282,
-      "eval_loss": 2.436706304550171,
-      "eval_runtime": 5.1472,
-      "eval_samples_per_second": 97.141,
-      "eval_steps_per_second": 12.24,
+      "eval_accuracy": 0.6007692307692307,
+      "eval_loss": 1.9881398677825928,
+      "eval_runtime": 4.4077,
+      "eval_samples_per_second": 113.439,
+      "eval_steps_per_second": 14.293,
       "step": 2625
     },
     {
       "epoch": 14.0,
-      "eval_exact_match": 15.4,
-      "eval_f1": 24.02761904761905,
+      "eval_exact_match": 23.0,
+      "eval_f1": 34.730952380952395,
       "step": 2625
     },
     {
       "epoch": 14.4,
-      "grad_norm": 0.6873505115509033,
       "learning_rate": 0.0001,
-      "loss": 0.4835,
+      "loss": 0.4184,
       "step": 2700
     },
     {
-      "epoch": 14.933333333333334,
-      "grad_norm": 0.7216406464576721,
+      "epoch": 14.93,
       "learning_rate": 0.0001,
-      "loss": 0.4966,
+      "loss": 0.4277,
       "step": 2800
     },
     {
-      "epoch": 14.997333333333334,
-      "eval_accuracy": 0.5585128205128205,
-      "eval_loss": 2.4315266609191895,
-      "eval_runtime": 5.7766,
-      "eval_samples_per_second": 86.556,
-      "eval_steps_per_second": 10.906,
+      "epoch": 15.0,
+      "eval_accuracy": 0.6005128205128205,
+      "eval_loss": 2.0305135250091553,
+      "eval_runtime": 4.4217,
+      "eval_samples_per_second": 113.079,
+      "eval_steps_per_second": 14.248,
       "step": 2812
     },
     {
-      "epoch": 14.997333333333334,
-      "eval_exact_match": 15.2,
-      "eval_f1": 23.853333333333342,
+      "epoch": 15.0,
+      "eval_exact_match": 21.2,
+      "eval_f1": 33.533492063492076,
       "step": 2812
     },
     {
-      "epoch": 15.466666666666667,
-      "grad_norm": 0.7105251550674438,
+      "epoch": 15.47,
       "learning_rate": 0.0001,
-      "loss": 0.476,
+      "loss": 0.4105,
       "step": 2900
     },
     {
       "epoch": 16.0,
-      "grad_norm": 0.6648848056793213,
       "learning_rate": 0.0001,
-      "loss": 0.4967,
+      "loss": 0.4298,
       "step": 3000
     },
     {
       "epoch": 16.0,
-      "eval_accuracy": 0.5587179487179487,
-      "eval_loss": 2.417990207672119,
-      "eval_runtime": 5.286,
-      "eval_samples_per_second": 94.589,
-      "eval_steps_per_second": 11.918,
+      "eval_accuracy": 0.6004615384615385,
+      "eval_loss": 2.0478391647338867,
+      "eval_runtime": 4.1977,
+      "eval_samples_per_second": 119.112,
+      "eval_steps_per_second": 15.008,
       "step": 3000
     },
     {
       "epoch": 16.0,
-      "eval_exact_match": 14.8,
-      "eval_f1": 23.16634920634922,
+      "eval_exact_match": 21.6,
+      "eval_f1": 33.046031746031765,
       "step": 3000
     },
     {
-      "epoch": 16.533333333333335,
-      "grad_norm": 0.7454655766487122,
+      "epoch": 16.53,
       "learning_rate": 0.0001,
-      "loss": 0.4692,
+      "loss": 0.4082,
       "step": 3100
     },
     {
-      "epoch": 16.997333333333334,
-      "eval_accuracy": 0.559076923076923,
-      "eval_loss": 2.4195001125335693,
-      "eval_runtime": 5.3355,
-      "eval_samples_per_second": 93.712,
-      "eval_steps_per_second": 11.808,
+      "epoch": 17.0,
+      "eval_accuracy": 0.6006666666666667,
+      "eval_loss": 2.053948163986206,
+      "eval_runtime": 4.2197,
+      "eval_samples_per_second": 118.491,
+      "eval_steps_per_second": 14.93,
       "step": 3187
     },
     {
-      "epoch": 16.997333333333334,
-      "eval_exact_match": 15.6,
-      "eval_f1": 24.129523809523818,
+      "epoch": 17.0,
+      "eval_exact_match": 21.8,
+      "eval_f1": 33.470476190476205,
       "step": 3187
     },
     {
-      "epoch": 17.066666666666666,
-      "grad_norm": 1.0016876459121704,
+      "epoch": 17.07,
       "learning_rate": 0.0001,
-      "loss": 0.4901,
+      "loss": 0.4223,
       "step": 3200
     },
     {
       "epoch": 17.6,
-      "grad_norm": 1.0129618644714355,
       "learning_rate": 0.0001,
-      "loss": 0.4738,
+      "loss": 0.411,
       "step": 3300
     },
     {
       "epoch": 18.0,
-      "eval_accuracy": 0.5584102564102564,
-      "eval_loss": 2.449228048324585,
-      "eval_runtime": 5.0748,
-      "eval_samples_per_second": 98.526,
-      "eval_steps_per_second": 12.414,
+      "eval_accuracy": 0.6004615384615385,
+      "eval_loss": 2.031390905380249,
+      "eval_runtime": 4.461,
+      "eval_samples_per_second": 112.081,
+      "eval_steps_per_second": 14.122,
       "step": 3375
     },
     {
       "epoch": 18.0,
-      "eval_exact_match": 14.8,
-      "eval_f1": 23.08238095238097,
+      "eval_exact_match": 21.6,
+      "eval_f1": 32.86714285714286,
       "step": 3375
     },
     {
-      "epoch": 18.133333333333333,
-      "grad_norm": 0.6989188194274902,
+      "epoch": 18.13,
       "learning_rate": 0.0001,
-      "loss": 0.4822,
+      "loss": 0.4178,
       "step": 3400
     },
     {
-      "epoch": 18.666666666666668,
-      "grad_norm": 0.730583667755127,
+      "epoch": 18.67,
       "learning_rate": 0.0001,
-      "loss": 0.4732,
+      "loss": 0.4113,
       "step": 3500
     },
     {
-      "epoch": 18.997333333333334,
-      "eval_accuracy": 0.5583076923076923,
-      "eval_loss": 2.410703182220459,
-      "eval_runtime": 5.3514,
-      "eval_samples_per_second": 93.434,
-      "eval_steps_per_second": 11.773,
+      "epoch": 19.0,
+      "eval_accuracy": 0.6011282051282051,
+      "eval_loss": 2.0367722511291504,
+      "eval_runtime": 4.8193,
+      "eval_samples_per_second": 103.749,
+      "eval_steps_per_second": 13.072,
       "step": 3562
     },
     {
-      "epoch": 18.997333333333334,
-      "eval_exact_match": 15.4,
-      "eval_f1": 22.446825396825407,
+      "epoch": 19.0,
+      "eval_exact_match": 22.4,
+      "eval_f1": 33.90603174603176,
       "step": 3562
     },
     {
       "epoch": 19.2,
-      "grad_norm": 0.6478275060653687,
       "learning_rate": 0.0001,
-      "loss": 0.4765,
+      "loss": 0.4111,
       "step": 3600
     },
     {
-      "epoch": 19.733333333333334,
-      "grad_norm": 1.6843740940093994,
+      "epoch": 19.73,
       "learning_rate": 0.0001,
-      "loss": 0.4735,
+      "loss": 0.4121,
       "step": 3700
     },
     {
       "epoch": 20.0,
-      "eval_accuracy": 0.5587692307692308,
-      "eval_loss": 2.4269626140594482,
-      "eval_runtime": 5.7798,
-      "eval_samples_per_second": 86.508,
-      "eval_steps_per_second": 10.9,
+      "eval_accuracy": 0.6016923076923076,
+      "eval_loss": 2.102205514907837,
+      "eval_runtime": 4.1919,
+      "eval_samples_per_second": 119.279,
+      "eval_steps_per_second": 15.029,
       "step": 3750
     },
     {
       "epoch": 20.0,
-      "eval_exact_match": 16.0,
-      "eval_f1": 23.713174603174604,
+      "eval_exact_match": 20.6,
+      "eval_f1": 32.77507936507938,
       "step": 3750
     },
     {
-      "epoch": 20.266666666666666,
-      "grad_norm": 0.6934134364128113,
+      "epoch": 20.27,
       "learning_rate": 0.0001,
-      "loss": 0.471,
+      "loss": 0.41,
       "step": 3800
     },
     {
       "epoch": 20.8,
-      "grad_norm": 0.7414541244506836,
       "learning_rate": 0.0001,
-      "loss": 0.4753,
+      "loss": 0.414,
       "step": 3900
     },
     {
-      "epoch": 20.997333333333334,
-      "eval_accuracy": 0.5592820512820513,
-      "eval_loss": 2.4946200847625732,
-      "eval_runtime": 5.5089,
-      "eval_samples_per_second": 90.762,
-      "eval_steps_per_second": 11.436,
+      "epoch": 21.0,
+      "eval_accuracy": 0.6006666666666667,
+      "eval_loss": 2.0511531829833984,
+      "eval_runtime": 4.5241,
+      "eval_samples_per_second": 110.519,
+      "eval_steps_per_second": 13.925,
       "step": 3937
     },
     {
-      "epoch": 20.997333333333334,
-      "eval_exact_match": 16.4,
-      "eval_f1": 24.212222222222234,
+      "epoch": 21.0,
+      "eval_exact_match": 21.2,
+      "eval_f1": 33.08360195360196,
       "step": 3937
     },
     {
-      "epoch": 21.333333333333332,
-      "grad_norm": 0.6189724206924438,
+      "epoch": 21.33,
       "learning_rate": 0.0001,
-      "loss": 0.465,
+      "loss": 0.4081,
       "step": 4000
     },
     {
-      "epoch": 21.866666666666667,
-      "grad_norm": 0.724271297454834,
+      "epoch": 21.87,
       "learning_rate": 0.0001,
-      "loss": 0.4758,
+      "loss": 0.4163,
       "step": 4100
     },
     {
       "epoch": 22.0,
-      "eval_accuracy": 0.5568205128205128,
-      "eval_loss": 2.4646029472351074,
-      "eval_runtime": 4.9867,
-      "eval_samples_per_second": 100.268,
-      "eval_steps_per_second": 12.634,
+      "eval_accuracy": 0.6015897435897436,
+      "eval_loss": 2.114720106124878,
+      "eval_runtime": 4.8019,
+      "eval_samples_per_second": 104.126,
+      "eval_steps_per_second": 13.12,
       "step": 4125
     },
     {
       "epoch": 22.0,
-      "eval_exact_match": 14.0,
-      "eval_f1": 21.840793650793653,
+      "eval_exact_match": 22.8,
+      "eval_f1": 34.690000000000005,
       "step": 4125
     },
     {
       "epoch": 22.4,
-      "grad_norm": 0.8124653100967407,
       "learning_rate": 0.0001,
-      "loss": 0.4629,
+      "loss": 0.402,
       "step": 4200
     },
     {
-      "epoch": 22.933333333333334,
-      "grad_norm": 1.1330875158309937,
+      "epoch": 22.93,
       "learning_rate": 0.0001,
-      "loss": 0.4769,
+      "loss": 0.4172,
       "step": 4300
     },
     {
-      "epoch": 22.997333333333334,
-      "eval_accuracy": 0.5572307692307692,
-      "eval_loss": 2.4494264125823975,
-      "eval_runtime": 5.7914,
-      "eval_samples_per_second": 86.335,
-      "eval_steps_per_second": 10.878,
+      "epoch": 23.0,
+      "eval_accuracy": 0.6007179487179487,
+      "eval_loss": 2.094172954559326,
+      "eval_runtime": 4.8193,
+      "eval_samples_per_second": 103.75,
+      "eval_steps_per_second": 13.073,
       "step": 4312
     },
     {
-      "epoch": 22.997333333333334,
-      "eval_exact_match": 14.8,
-      "eval_f1": 23.25746031746033,
+      "epoch": 23.0,
+      "eval_exact_match": 21.6,
+      "eval_f1": 33.82884004884006,
       "step": 4312
     },
     {
-      "epoch": 23.466666666666665,
-      "grad_norm": 1.0306299924850464,
+      "epoch": 23.47,
       "learning_rate": 0.0001,
-      "loss": 0.4616,
+      "loss": 0.4018,
       "step": 4400
     },
     {
       "epoch": 24.0,
-      "grad_norm": 0.5557557940483093,
       "learning_rate": 0.0001,
-      "loss": 0.4766,
+      "loss": 0.4156,
       "step": 4500
     },
     {
       "epoch": 24.0,
-      "eval_accuracy": 0.5583076923076923,
-      "eval_loss": 2.464630365371704,
-      "eval_runtime": 5.0419,
-      "eval_samples_per_second": 99.17,
-      "eval_steps_per_second": 12.495,
+      "eval_accuracy": 0.6007692307692307,
+      "eval_loss": 2.1200733184814453,
+      "eval_runtime": 4.0917,
+      "eval_samples_per_second": 122.197,
+      "eval_steps_per_second": 15.397,
       "step": 4500
     },
     {
       "epoch": 24.0,
-      "eval_exact_match": 15.0,
-      "eval_f1": 23.326507936507948,
+      "eval_exact_match": 22.2,
+      "eval_f1": 35.02444444444445,
       "step": 4500
     },
     {
-      "epoch": 24.533333333333335,
-      "grad_norm": 0.6188230514526367,
+      "epoch": 24.53,
       "learning_rate": 0.0001,
-      "loss": 0.4588,
+      "loss": 0.3997,
       "step": 4600
     },
     {
-      "epoch": 24.997333333333334,
-      "eval_accuracy": 0.5575897435897436,
-      "eval_loss": 2.4749364852905273,
-      "eval_runtime": 5.8046,
-      "eval_samples_per_second": 86.139,
-      "eval_steps_per_second": 10.854,
+      "epoch": 25.0,
+      "eval_accuracy": 0.601025641025641,
+      "eval_loss": 2.065951347351074,
+      "eval_runtime": 4.4048,
+      "eval_samples_per_second": 113.512,
+      "eval_steps_per_second": 14.302,
       "step": 4687
     },
     {
-      "epoch": 24.997333333333334,
-      "eval_exact_match": 15.4,
-      "eval_f1": 23.709696969696974,
+      "epoch": 25.0,
+      "eval_exact_match": 22.0,
+      "eval_f1": 33.90190476190476,
       "step": 4687
     },
     {
-      "epoch": 25.066666666666666,
-      "grad_norm": 0.5854879021644592,
+      "epoch": 25.07,
       "learning_rate": 0.0001,
-      "loss": 0.47,
+      "loss": 0.4127,
       "step": 4700
     },
     {
       "epoch": 25.6,
-      "grad_norm": 0.5842415690422058,
       "learning_rate": 0.0001,
-      "loss": 0.4596,
+      "loss": 0.3994,
       "step": 4800
     },
     {
       "epoch": 26.0,
-      "eval_accuracy": 0.5573333333333333,
-      "eval_loss": 2.4922566413879395,
-      "eval_runtime": 5.1065,
-      "eval_samples_per_second": 97.914,
-      "eval_steps_per_second": 12.337,
+      "eval_accuracy": 0.6005641025641025,
+      "eval_loss": 2.0832056999206543,
+      "eval_runtime": 4.587,
+      "eval_samples_per_second": 109.003,
+      "eval_steps_per_second": 13.734,
       "step": 4875
     },
     {
       "epoch": 26.0,
-      "eval_exact_match": 14.0,
-      "eval_f1": 22.40301587301588,
+      "eval_exact_match": 22.0,
+      "eval_f1": 33.120952380952374,
       "step": 4875
     },
     {
-      "epoch": 26.133333333333333,
-      "grad_norm": 0.736111044883728,
+      "epoch": 26.13,
       "learning_rate": 0.0001,
-      "loss": 0.4699,
+      "loss": 0.4109,
       "step": 4900
     },
     {
-      "epoch": 26.666666666666668,
-      "grad_norm": 0.6232179403305054,
+      "epoch": 26.67,
       "learning_rate": 0.0001,
-      "loss": 0.4618,
+      "loss": 0.4032,
       "step": 5000
     },
     {
-      "epoch": 26.997333333333334,
-      "eval_accuracy": 0.558,
-      "eval_loss": 2.454937696456909,
-      "eval_runtime": 5.1785,
-      "eval_samples_per_second": 96.553,
-      "eval_steps_per_second": 12.166,
+      "epoch": 27.0,
+      "eval_accuracy": 0.6003076923076923,
+      "eval_loss": 2.1423139572143555,
+      "eval_runtime": 4.0933,
+      "eval_samples_per_second": 122.15,
+      "eval_steps_per_second": 15.391,
       "step": 5062
     },
     {
-      "epoch": 26.997333333333334,
-      "eval_exact_match": 14.4,
-      "eval_f1": 23.143809523809526,
+      "epoch": 27.0,
+      "eval_exact_match": 22.4,
+      "eval_f1": 33.54269841269841,
       "step": 5062
     },
     {
       "epoch": 27.2,
-      "grad_norm": 0.6592839956283569,
       "learning_rate": 0.0001,
-      "loss": 0.4626,
+      "loss": 0.4064,
       "step": 5100
     },
     {
-      "epoch": 27.733333333333334,
-      "grad_norm": 0.5580378174781799,
+      "epoch": 27.73,
       "learning_rate": 0.0001,
-      "loss": 0.4646,
+      "loss": 0.4058,
       "step": 5200
     },
     {
       "epoch": 28.0,
-      "eval_accuracy": 0.5573333333333333,
-      "eval_loss": 2.475144147872925,
-      "eval_runtime": 5.7088,
-      "eval_samples_per_second": 87.584,
-      "eval_steps_per_second": 11.036,
+      "eval_accuracy": 0.6014871794871794,
+      "eval_loss": 2.1000020503997803,
+      "eval_runtime": 4.1975,
+      "eval_samples_per_second": 119.119,
+      "eval_steps_per_second": 15.009,
       "step": 5250
     },
     {
       "epoch": 28.0,
-      "eval_exact_match": 14.8,
-      "eval_f1": 22.921428571428585,
+      "eval_exact_match": 21.8,
+      "eval_f1": 33.625555555555565,
       "step": 5250
     },
     {
-      "epoch": 28.266666666666666,
-      "grad_norm": 0.6161842346191406,
+      "epoch": 28.27,
       "learning_rate": 0.0001,
-      "loss": 0.4611,
+      "loss": 0.4026,
       "step": 5300
     },
     {
       "epoch": 28.8,
-      "grad_norm": 0.5740762948989868,
       "learning_rate": 0.0001,
-      "loss": 0.4645,
+      "loss": 0.4065,
       "step": 5400
     },
     {
-      "epoch": 28.997333333333334,
-      "eval_accuracy": 0.5574871794871795,
-      "eval_loss": 2.5007574558258057,
-      "eval_runtime": 5.7875,
-      "eval_samples_per_second": 86.393,
-      "eval_steps_per_second": 10.886,
+      "epoch": 29.0,
+      "eval_accuracy": 0.6009230769230769,
+      "eval_loss": 2.1064789295196533,
+      "eval_runtime": 4.3996,
+      "eval_samples_per_second": 113.646,
+      "eval_steps_per_second": 14.319,
       "step": 5437
     },
     {
-      "epoch": 28.997333333333334,
-      "eval_exact_match": 15.2,
-      "eval_f1": 23.13380952380953,
+      "epoch": 29.0,
+      "eval_exact_match": 20.2,
+      "eval_f1": 32.787330447330454,
       "step": 5437
     },
     {
-      "epoch": 29.333333333333332,
-      "grad_norm": 1.0054398775100708,
+      "epoch": 29.33,
       "learning_rate": 0.0001,
-      "loss": 0.4562,
+      "loss": 0.4021,
       "step": 5500
     },
     {
-      "epoch": 29.866666666666667,
-      "grad_norm": 0.8751959800720215,
+      "epoch": 29.87,
       "learning_rate": 0.0001,
-      "loss": 0.4669,
+      "loss": 0.4068,
       "step": 5600
     },
     {
       "epoch": 30.0,
-      "eval_accuracy": 0.5561025641025641,
-      "eval_loss": 2.482387065887451,
-      "eval_runtime": 5.1096,
-      "eval_samples_per_second": 97.855,
-      "eval_steps_per_second": 12.33,
+      "eval_accuracy": 0.6006153846153847,
+      "eval_loss": 2.138920307159424,
+      "eval_runtime": 4.8054,
+      "eval_samples_per_second": 104.049,
+      "eval_steps_per_second": 13.11,
       "step": 5625
     },
     {
       "epoch": 30.0,
-      "eval_exact_match": 15.6,
-      "eval_f1": 23.701428571428576,
+      "eval_exact_match": 22.2,
+      "eval_f1": 34.06333333333334,
       "step": 5625
     },
     {
       "epoch": 30.4,
-      "grad_norm": 0.6013503670692444,
       "learning_rate": 0.0001,
-      "loss": 0.4536,
+      "loss": 0.3984,
       "step": 5700
     },
     {
-      "epoch": 30.933333333333334,
-      "grad_norm": 0.559660792350769,
+      "epoch": 30.93,
       "learning_rate": 0.0001,
-      "loss": 0.4675,
+      "loss": 0.4091,
       "step": 5800
     },
     {
-      "epoch": 30.997333333333334,
-      "eval_accuracy": 0.5568205128205128,
-      "eval_loss": 2.4686777591705322,
-      "eval_runtime": 5.4297,
-      "eval_samples_per_second": 92.086,
-      "eval_steps_per_second": 11.603,
+      "epoch": 31.0,
+      "eval_accuracy": 0.6004615384615385,
+      "eval_loss": 2.1240592002868652,
+      "eval_runtime": 4.7802,
+      "eval_samples_per_second": 104.598,
+      "eval_steps_per_second": 13.179,
       "step": 5812
     },
     {
-      "epoch": 30.997333333333334,
-      "eval_exact_match": 16.8,
-      "eval_f1": 25.139682539682543,
+      "epoch": 31.0,
+      "eval_exact_match": 19.6,
+      "eval_f1": 32.183333333333344,
       "step": 5812
     },
     {
-      "epoch": 31.466666666666665,
-      "grad_norm": 0.8210071921348572,
+      "epoch": 31.47,
       "learning_rate": 0.0001,
-      "loss": 0.4503,
+      "loss": 0.398,
       "step": 5900
     },
     {
       "epoch": 32.0,
-      "grad_norm": 0.7878498435020447,
       "learning_rate": 0.0001,
-      "loss": 0.4701,
+      "loss": 0.4103,
       "step": 6000
     },
     {
       "epoch": 32.0,
-      "eval_accuracy": 0.556923076923077,
-      "eval_loss": 2.4814717769622803,
-      "eval_runtime": 4.9983,
-      "eval_samples_per_second": 100.033,
-      "eval_steps_per_second": 12.604,
+      "eval_accuracy": 0.601025641025641,
+      "eval_loss": 2.1240859031677246,
+      "eval_runtime": 4.2046,
+      "eval_samples_per_second": 118.918,
+      "eval_steps_per_second": 14.984,
       "step": 6000
     },
     {
       "epoch": 32.0,
-      "eval_exact_match": 16.2,
-      "eval_f1": 24.101428571428585,
+      "eval_exact_match": 14.6,
+      "eval_f1": 24.464480519480528,
       "step": 6000
     },
     {
-      "epoch": 32.53333333333333,
-      "grad_norm": 0.6601159572601318,
+      "epoch": 32.53,
       "learning_rate": 0.0001,
-      "loss": 0.4504,
+      "loss": 0.3959,
       "step": 6100
     },
     {
-      "epoch": 32.99733333333333,
-      "eval_accuracy": 0.5563589743589744,
-      "eval_loss": 2.513813018798828,
-      "eval_runtime": 5.3528,
-      "eval_samples_per_second": 93.409,
-      "eval_steps_per_second": 11.769,
+      "epoch": 33.0,
+      "eval_accuracy": 0.6021025641025641,
+      "eval_loss": 2.1205942630767822,
+      "eval_runtime": 4.0969,
+      "eval_samples_per_second": 122.043,
+      "eval_steps_per_second": 15.377,
       "step": 6187
     },
     {
-      "epoch": 32.99733333333333,
-      "eval_exact_match": 15.6,
-      "eval_f1": 23.765714285714292,
+      "epoch": 33.0,
+      "eval_exact_match": 5.2,
+      "eval_f1": 9.914294948434568,
       "step": 6187
     },
     {
-      "epoch": 33.06666666666667,
-      "grad_norm": 0.5784469842910767,
+      "epoch": 33.07,
       "learning_rate": 0.0001,
-      "loss": 0.4622,
+      "loss": 0.4074,
       "step": 6200
     },
     {
       "epoch": 33.6,
-      "grad_norm": 0.6005637645721436,
       "learning_rate": 0.0001,
-      "loss": 0.4522,
+      "loss": 0.3974,
       "step": 6300
     },
     {
       "epoch": 34.0,
-      "eval_accuracy": 0.5565128205128205,
-      "eval_loss": 2.504781723022461,
-      "eval_runtime": 5.7399,
-      "eval_samples_per_second": 87.109,
-      "eval_steps_per_second": 10.976,
+      "eval_accuracy": 0.6017435897435898,
+      "eval_loss": 2.106126308441162,
+      "eval_runtime": 4.417,
+      "eval_samples_per_second": 113.199,
+      "eval_steps_per_second": 14.263,
       "step": 6375
     },
     {
       "epoch": 34.0,
-      "eval_exact_match": 14.2,
-      "eval_f1": 23.09952380952382,
+      "eval_exact_match": 11.0,
+      "eval_f1": 16.971111111111124,
       "step": 6375
     },
     {
-      "epoch": 34.13333333333333,
-      "grad_norm": 0.5809851288795471,
+      "epoch": 34.13,
       "learning_rate": 0.0001,
-      "loss": 0.4607,
+      "loss": 0.4037,
       "step": 6400
     },
     {
-      "epoch": 34.666666666666664,
-      "grad_norm": 0.6082825064659119,
+      "epoch": 34.67,
       "learning_rate": 0.0001,
-      "loss": 0.4526,
+      "loss": 0.3983,
       "step": 6500
     },
     {
-      "epoch": 34.99733333333333,
-      "eval_accuracy": 0.5563589743589744,
-      "eval_loss": 2.529914617538452,
-      "eval_runtime": 5.3389,
-      "eval_samples_per_second": 93.652,
-      "eval_steps_per_second": 11.8,
+      "epoch": 35.0,
+      "eval_accuracy": 0.6012820512820513,
+      "eval_loss": 2.1041135787963867,
+      "eval_runtime": 4.1955,
+      "eval_samples_per_second": 119.176,
+      "eval_steps_per_second": 15.016,
       "step": 6562
     },
     {
-      "epoch": 34.99733333333333,
-      "eval_exact_match": 15.4,
-      "eval_f1": 23.87190476190477,
+      "epoch": 35.0,
+      "eval_exact_match": 7.8,
+      "eval_f1": 12.592722468631077,
       "step": 6562
     },
     {
       "epoch": 35.2,
-      "grad_norm": 0.5042786002159119,
       "learning_rate": 0.0001,
-      "loss": 0.4545,
+      "loss": 0.4027,
       "step": 6600
     },
     {
-      "epoch": 35.733333333333334,
-      "grad_norm": 0.8007566928863525,
+      "epoch": 35.73,
       "learning_rate": 0.0001,
-      "loss": 0.4558,
+      "loss": 0.4034,
       "step": 6700
     },
     {
       "epoch": 36.0,
-      "eval_accuracy": 0.5582564102564103,
-      "eval_loss": 2.485551357269287,
-      "eval_runtime": 5.7241,
-      "eval_samples_per_second": 87.351,
-      "eval_steps_per_second": 11.006,
+      "eval_accuracy": 0.6017435897435898,
+      "eval_loss": 2.0842514038085938,
+      "eval_runtime": 4.4017,
+      "eval_samples_per_second": 113.592,
+      "eval_steps_per_second": 14.313,
       "step": 6750
     },
     {
       "epoch": 36.0,
-      "eval_exact_match": 15.8,
-      "eval_f1": 23.60142857142858,
+      "eval_exact_match": 0.4,
+      "eval_f1": 1.2985909338645831,
       "step": 6750
     },
     {
-      "epoch": 36.266666666666666,
-      "grad_norm": 0.4920075535774231,
+      "epoch": 36.27,
       "learning_rate": 0.0001,
-      "loss": 0.452,
+      "loss": 0.4007,
       "step": 6800
     },
     {
       "epoch": 36.8,
-      "grad_norm": 0.5003876686096191,
       "learning_rate": 0.0001,
-      "loss": 0.458,
+      "loss": 0.4035,
       "step": 6900
     },
     {
-      "epoch": 36.99733333333333,
-      "eval_accuracy": 0.5572820512820513,
-      "eval_loss": 2.516226291656494,
-      "eval_runtime": 5.7691,
-      "eval_samples_per_second": 86.669,
-      "eval_steps_per_second": 10.92,
+      "epoch": 37.0,
+      "eval_accuracy": 0.6034871794871794,
+      "eval_loss": 2.083721876144409,
+      "eval_runtime": 4.2147,
+      "eval_samples_per_second": 118.633,
+      "eval_steps_per_second": 14.948,
       "step": 6937
     },
     {
-      "epoch": 36.99733333333333,
-      "eval_exact_match": 15.0,
-      "eval_f1": 22.680317460317468,
+      "epoch": 37.0,
+      "eval_exact_match": 13.2,
+      "eval_f1": 19.512857142857147,
       "step": 6937
     },
     {
-      "epoch": 37.333333333333336,
-      "grad_norm": 0.5616186261177063,
+      "epoch": 37.33,
       "learning_rate": 0.0001,
-      "loss": 0.4512,
+      "loss": 0.4285,
       "step": 7000
     },
     {
-      "epoch": 37.86666666666667,
-      "grad_norm": 0.5125617384910583,
+      "epoch": 37.87,
       "learning_rate": 0.0001,
-      "loss": 0.4597,
+      "loss": 0.4013,
       "step": 7100
     },
     {
       "epoch": 38.0,
-      "eval_accuracy": 0.555025641025641,
-      "eval_loss": 2.5178050994873047,
-      "eval_runtime": 4.9917,
-      "eval_samples_per_second": 100.167,
-      "eval_steps_per_second": 12.621,
+      "eval_accuracy": 0.6014871794871794,
+      "eval_loss": 2.1708192825317383,
+      "eval_runtime": 4.195,
+      "eval_samples_per_second": 119.189,
+      "eval_steps_per_second": 15.018,
       "step": 7125
     },
     {
       "epoch": 38.0,
-      "eval_exact_match": 14.2,
-      "eval_f1": 21.111904761904768,
+      "eval_exact_match": 0.2,
+      "eval_f1": 2.2475702075702073,
       "step": 7125
     },
     {
       "epoch": 38.4,
-      "grad_norm": 0.5520429015159607,
       "learning_rate": 0.0001,
-      "loss": 0.4463,
+      "loss": 0.3944,
       "step": 7200
     },
     {
-      "epoch": 38.93333333333333,
-      "grad_norm": 0.630486249923706,
+      "epoch": 38.93,
       "learning_rate": 0.0001,
-      "loss": 0.4607,
+      "loss": 0.4063,
       "step": 7300
     },
     {
-      "epoch": 38.99733333333333,
-      "eval_accuracy": 0.5562051282051282,
-      "eval_loss": 2.5462260246276855,
-      "eval_runtime": 5.7594,
-      "eval_samples_per_second": 86.815,
-      "eval_steps_per_second": 10.939,
+      "epoch": 39.0,
+      "eval_accuracy": 0.602,
+      "eval_loss": 2.094609260559082,
+      "eval_runtime": 4.1702,
+      "eval_samples_per_second": 119.897,
+      "eval_steps_per_second": 15.107,
       "step": 7312
     },
     {
-      "epoch": 38.99733333333333,
-      "eval_exact_match": 14.8,
-      "eval_f1": 22.822857142857156,
+      "epoch": 39.0,
+      "eval_exact_match": 0.2,
+      "eval_f1": 1.0749958319523536,
       "step": 7312
     },
     {
-      "epoch": 39.46666666666667,
-      "grad_norm": 0.5777439475059509,
+      "epoch": 39.47,
       "learning_rate": 0.0001,
-      "loss": 0.4447,
+      "loss": 0.393,
       "step": 7400
     },
     {
       "epoch": 40.0,
-      "grad_norm": 0.5794367790222168,
       "learning_rate": 0.0001,
-      "loss": 0.4614,
+      "loss": 0.4049,
       "step": 7500
     },
     {
       "epoch": 40.0,
-      "eval_accuracy": 0.5564615384615385,
-      "eval_loss": 2.5492753982543945,
-      "eval_runtime": 4.9903,
-      "eval_samples_per_second": 100.194,
-      "eval_steps_per_second": 12.624,
+      "eval_accuracy": 0.601948717948718,
+      "eval_loss": 2.167102575302124,
+      "eval_runtime": 4.0944,
+      "eval_samples_per_second": 122.119,
+      "eval_steps_per_second": 15.387,
       "step": 7500
     },
     {
       "epoch": 40.0,
-      "eval_exact_match": 15.0,
-      "eval_f1": 22.4652380952381,
+      "eval_exact_match": 0.4,
+      "eval_f1": 0.6425925925925926,
       "step": 7500
     },
     {
-      "epoch": 40.53333333333333,
-      "grad_norm": 0.5546026229858398,
+      "epoch": 40.53,
       "learning_rate": 0.0001,
-      "loss": 0.4449,
+      "loss": 0.391,
       "step": 7600
     },
     {
-      "epoch": 40.99733333333333,
-      "eval_accuracy": 0.5558461538461539,
-      "eval_loss": 2.564469814300537,
-      "eval_runtime": 5.0211,
-      "eval_samples_per_second": 99.581,
-      "eval_steps_per_second": 12.547,
+      "epoch": 41.0,
+      "eval_accuracy": 0.6026153846153847,
+      "eval_loss": 2.1508407592773438,
+      "eval_runtime": 4.1993,
+      "eval_samples_per_second": 119.068,
+      "eval_steps_per_second": 15.003,
       "step": 7687
     },
     {
-      "epoch": 40.99733333333333,
-      "eval_exact_match": 15.8,
-      "eval_f1": 23.358571428571434,
+      "epoch": 41.0,
+      "eval_exact_match": 0.0,
+      "eval_f1": 0.15599864052666793,
       "step": 7687
     },
     {
-      "epoch": 41.06666666666667,
-      "grad_norm": 0.5399795770645142,
+      "epoch": 41.07,
       "learning_rate": 0.0001,
-      "loss": 0.4561,
+      "loss": 0.4032,
       "step": 7700
     },
     {
       "epoch": 41.6,
-      "grad_norm": 0.5286694765090942,
       "learning_rate": 0.0001,
-      "loss": 0.4472,
+      "loss": 0.3913,
       "step": 7800
     },
     {
       "epoch": 42.0,
-      "eval_accuracy": 0.5548717948717948,
-      "eval_loss": 2.538248062133789,
-      "eval_runtime": 4.9877,
-      "eval_samples_per_second": 100.246,
-      "eval_steps_per_second": 12.631,
+      "eval_accuracy": 0.5998461538461538,
+      "eval_loss": 2.2061829566955566,
+      "eval_runtime": 4.4055,
+      "eval_samples_per_second": 113.493,
+      "eval_steps_per_second": 14.3,
       "step": 7875
     },
     {
       "epoch": 42.0,
-      "eval_exact_match": 14.2,
-      "eval_f1": 22.103809523809534,
+      "eval_exact_match": 0.0,
+      "eval_f1": 0.25829123249594316,
       "step": 7875
     },
     {
-      "epoch": 42.13333333333333,
-      "grad_norm": 0.7045392990112305,
+      "epoch": 42.13,
       "learning_rate": 0.0001,
-      "loss": 0.4588,
+      "loss": 0.4,
       "step": 7900
     },
     {
-      "epoch": 42.666666666666664,
-      "grad_norm": 0.7843219637870789,
+      "epoch": 42.67,
       "learning_rate": 0.0001,
-      "loss": 0.4549,
+      "loss": 0.3945,
       "step": 8000
     },
     {
-      "epoch": 42.99733333333333,
-      "eval_accuracy": 0.554974358974359,
-      "eval_loss": 2.55190110206604,
-      "eval_runtime": 5.0951,
-      "eval_samples_per_second": 98.134,
-      "eval_steps_per_second": 12.365,
+      "epoch": 43.0,
+      "eval_accuracy": 0.6011794871794872,
+      "eval_loss": 2.2214462757110596,
+      "eval_runtime": 4.0938,
+      "eval_samples_per_second": 122.135,
+      "eval_steps_per_second": 15.389,
       "step": 8062
     },
     {
-      "epoch": 42.99733333333333,
-      "eval_exact_match": 14.6,
-      "eval_f1": 22.485714285714288,
+      "epoch": 43.0,
+      "eval_exact_match": 0.0,
+      "eval_f1": 1.966911267657296,
       "step": 8062
     }
   ],
   "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": 1.476975114972037e+17,
-  "train_batch_size": 2,
+  "total_flos": 3.008590038507192e+17,
   "trial_name": null,
   "trial_params": null
 }

checkpoint-8062/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d667d5e0e40fe951f9fe0d2cd7fb4682a608b09af78a113a245307211adedfbc
-size 5304
+oid sha256:dfdad39e2291c20f000713e6b820dbeb747acc6873c9eec22bd69c854755aa38
+size 4728