Training in progress, epoch 30, checkpoint

checkpoint-15870/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-15870/adapter_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e201bed7a2af93e1b8c097bbfdedd4a71838ea6ba63c183835e310f6fe15f94a
+size 143269386

checkpoint-15870/added_tokens.json

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

checkpoint-15870/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2688f134e615dd7ea980ba86d26aa08eefd0600b23974937d47a72567c2c9905
-size 224537202
+oid sha256:a2604bfa8712d6882289a6261084814d2e643add1a5405e3f55fadeb7d1b4194
+size 286585234

checkpoint-15870/rng_state_0.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a91d64040875746c8f1e9e18f453c17a41265887bf1f113a03f5502ed03587aa
+oid sha256:edcd17ac6bbd900ce5180de62cf748560154e4cfff909c92023bbd9a65ef6ee4
 size 15024

checkpoint-15870/rng_state_1.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c9ca2e3f0ff118a2cd1fd52b036db7b7f922d496c659804b2b74633a12281d2f
+oid sha256:1fbb741c6ed167193bf6d4cd5f74d3ceb8b293c6f47dd96dc05dff8ee61b1f1a
 size 15024

checkpoint-15870/rng_state_2.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0f526babc31e69d83b58ac79479ba92a04cbb0d288f902bca896ee0177ad8a3b
+oid sha256:06bacda2ef28f8dd2fd44ac9381d068ee1e04dd078ba1e3c684c8162409f50e4
 size 15024

checkpoint-15870/rng_state_3.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9c941bebbb4708d9bda6374fcbea6487907e743a7b9c603f5a306bded01a922b
+oid sha256:a72d996e53fda0853d00d859d5827d2c7881171ac02c3d32135a0715d55b09cd
 size 15024

checkpoint-15870/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-15870/tokenizer.model

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

checkpoint-15870/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-15870/trainer_state.json

@@ -9,1581 +9,1409 @@
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.1890359168241966,
-      "grad_norm": 0.5107442140579224,
+      "epoch": 0.19,
       "learning_rate": 5e-05,
-      "loss": 2.0482,
+      "loss": 1.6355,
       "step": 100
     },
     {
-      "epoch": 0.3780718336483932,
-      "grad_norm": 0.5563424229621887,
+      "epoch": 0.38,
       "learning_rate": 5e-05,
-      "loss": 1.8835,
+      "loss": 1.4037,
       "step": 200
     },
     {
-      "epoch": 0.5671077504725898,
-      "grad_norm": 0.4735371172428131,
+      "epoch": 0.57,
       "learning_rate": 5e-05,
-      "loss": 1.8512,
+      "loss": 1.4183,
       "step": 300
     },
     {
-      "epoch": 0.7561436672967864,
-      "grad_norm": 0.4543495178222656,
+      "epoch": 0.76,
       "learning_rate": 5e-05,
-      "loss": 1.8489,
+      "loss": 1.3881,
       "step": 400
     },
     {
-      "epoch": 0.945179584120983,
-      "grad_norm": 0.5007112622261047,
+      "epoch": 0.95,
       "learning_rate": 5e-05,
-      "loss": 1.8439,
+      "loss": 1.3891,
       "step": 500
     },
     {
       "epoch": 1.0,
-      "eval_accuracy": 0.5733333333333334,
-      "eval_loss": 1.6194567680358887,
-      "eval_runtime": 5.4154,
-      "eval_samples_per_second": 92.33,
-      "eval_steps_per_second": 11.634,
+      "eval_accuracy": 0.6137948717948718,
+      "eval_loss": 1.3015451431274414,
+      "eval_runtime": 4.1821,
+      "eval_samples_per_second": 119.557,
+      "eval_steps_per_second": 15.064,
       "step": 529
     },
     {
       "epoch": 1.0,
-      "eval_exact_match": 17.0,
-      "eval_f1": 26.880952380952387,
+      "eval_exact_match": 23.2,
+      "eval_f1": 35.17014759120022,
       "step": 529
     },
     {
-      "epoch": 1.1342155009451795,
-      "grad_norm": 0.5181360244750977,
+      "epoch": 1.13,
       "learning_rate": 5e-05,
-      "loss": 1.8217,
+      "loss": 1.3808,
       "step": 600
     },
     {
-      "epoch": 1.3232514177693762,
-      "grad_norm": 1.1348918676376343,
+      "epoch": 1.32,
       "learning_rate": 5e-05,
-      "loss": 1.7866,
+      "loss": 1.348,
       "step": 700
     },
     {
-      "epoch": 1.5122873345935728,
-      "grad_norm": 0.661758303642273,
+      "epoch": 1.51,
       "learning_rate": 5e-05,
-      "loss": 1.7934,
+      "loss": 1.3627,
       "step": 800
     },
     {
-      "epoch": 1.7013232514177694,
-      "grad_norm": 0.5634772777557373,
+      "epoch": 1.7,
       "learning_rate": 5e-05,
-      "loss": 1.7731,
+      "loss": 1.3473,
       "step": 900
     },
     {
-      "epoch": 1.8903591682419658,
-      "grad_norm": 0.6698930263519287,
+      "epoch": 1.89,
       "learning_rate": 5e-05,
-      "loss": 1.7969,
+      "loss": 1.3633,
       "step": 1000
     },
     {
       "epoch": 2.0,
-      "eval_accuracy": 0.5751794871794872,
-      "eval_loss": 1.603736162185669,
-      "eval_runtime": 5.0986,
-      "eval_samples_per_second": 98.066,
-      "eval_steps_per_second": 12.356,
+      "eval_accuracy": 0.6165641025641025,
+      "eval_loss": 1.2854759693145752,
+      "eval_runtime": 4.1064,
+      "eval_samples_per_second": 121.761,
+      "eval_steps_per_second": 15.342,
       "step": 1058
     },
     {
       "epoch": 2.0,
-      "eval_exact_match": 17.6,
-      "eval_f1": 27.218095238095238,
+      "eval_exact_match": 25.8,
+      "eval_f1": 37.509047619047635,
       "step": 1058
     },
     {
-      "epoch": 2.0793950850661624,
-      "grad_norm": 0.6265193223953247,
+      "epoch": 2.08,
       "learning_rate": 5e-05,
-      "loss": 1.7416,
+      "loss": 1.3358,
       "step": 1100
     },
     {
-      "epoch": 2.268431001890359,
-      "grad_norm": 0.812364935874939,
+      "epoch": 2.27,
       "learning_rate": 5e-05,
-      "loss": 1.6957,
+      "loss": 1.2917,
       "step": 1200
     },
     {
-      "epoch": 2.4574669187145557,
-      "grad_norm": 1.0972926616668701,
+      "epoch": 2.46,
       "learning_rate": 5e-05,
-      "loss": 1.7033,
+      "loss": 1.2948,
       "step": 1300
     },
     {
-      "epoch": 2.6465028355387523,
-      "grad_norm": 1.0831658840179443,
+      "epoch": 2.65,
       "learning_rate": 5e-05,
-      "loss": 1.6935,
+      "loss": 1.3144,
       "step": 1400
     },
     {
-      "epoch": 2.835538752362949,
-      "grad_norm": 1.0022494792938232,
+      "epoch": 2.84,
       "learning_rate": 5e-05,
-      "loss": 1.6799,
+      "loss": 1.2929,
       "step": 1500
     },
     {
       "epoch": 3.0,
-      "eval_accuracy": 0.573025641025641,
-      "eval_loss": 1.6428505182266235,
-      "eval_runtime": 5.1488,
-      "eval_samples_per_second": 97.111,
-      "eval_steps_per_second": 12.236,
+      "eval_accuracy": 0.6176923076923077,
+      "eval_loss": 1.2953603267669678,
+      "eval_runtime": 4.0992,
+      "eval_samples_per_second": 121.974,
+      "eval_steps_per_second": 15.369,
       "step": 1587
     },
     {
       "epoch": 3.0,
-      "eval_exact_match": 20.2,
-      "eval_f1": 27.93809523809523,
+      "eval_exact_match": 25.0,
+      "eval_f1": 37.08253968253969,
       "step": 1587
     },
     {
-      "epoch": 3.0245746691871456,
-      "grad_norm": 0.8240587711334229,
+      "epoch": 3.02,
       "learning_rate": 5e-05,
-      "loss": 1.6618,
+      "loss": 1.2906,
       "step": 1600
     },
     {
-      "epoch": 3.213610586011342,
-      "grad_norm": 1.2994645833969116,
+      "epoch": 3.21,
       "learning_rate": 5e-05,
-      "loss": 1.5736,
+      "loss": 1.2216,
       "step": 1700
     },
     {
-      "epoch": 3.402646502835539,
-      "grad_norm": 1.0891159772872925,
+      "epoch": 3.4,
       "learning_rate": 5e-05,
-      "loss": 1.5798,
+      "loss": 1.2459,
       "step": 1800
     },
     {
-      "epoch": 3.5916824196597354,
-      "grad_norm": 1.3826475143432617,
+      "epoch": 3.59,
       "learning_rate": 5e-05,
-      "loss": 1.5957,
+      "loss": 1.2474,
       "step": 1900
     },
     {
-      "epoch": 3.780718336483932,
-      "grad_norm": 0.9901586174964905,
+      "epoch": 3.78,
       "learning_rate": 5e-05,
-      "loss": 1.6026,
+      "loss": 1.2518,
       "step": 2000
     },
     {
-      "epoch": 3.9697542533081287,
-      "grad_norm": 1.0607813596725464,
+      "epoch": 3.97,
       "learning_rate": 5e-05,
-      "loss": 1.6106,
+      "loss": 1.2361,
       "step": 2100
     },
     {
       "epoch": 4.0,
-      "eval_accuracy": 0.5708717948717948,
-      "eval_loss": 1.7143596410751343,
-      "eval_runtime": 5.3942,
-      "eval_samples_per_second": 92.693,
-      "eval_steps_per_second": 11.679,
+      "eval_accuracy": 0.6045128205128205,
+      "eval_loss": 1.3488503694534302,
+      "eval_runtime": 4.2015,
+      "eval_samples_per_second": 119.004,
+      "eval_steps_per_second": 14.995,
       "step": 2116
     },
     {
       "epoch": 4.0,
-      "eval_exact_match": 19.0,
-      "eval_f1": 27.71428571428571,
+      "eval_exact_match": 25.0,
+      "eval_f1": 36.82126984126986,
       "step": 2116
     },
     {
-      "epoch": 4.158790170132325,
-      "grad_norm": 1.3189178705215454,
+      "epoch": 4.16,
       "learning_rate": 5e-05,
-      "loss": 1.5029,
+      "loss": 1.1856,
       "step": 2200
     },
     {
-      "epoch": 4.3478260869565215,
-      "grad_norm": 1.062267541885376,
+      "epoch": 4.35,
       "learning_rate": 5e-05,
-      "loss": 1.51,
+      "loss": 1.1556,
       "step": 2300
     },
     {
-      "epoch": 4.536862003780718,
-      "grad_norm": 1.3810843229293823,
+      "epoch": 4.54,
       "learning_rate": 5e-05,
-      "loss": 1.4867,
+      "loss": 1.1828,
       "step": 2400
     },
     {
-      "epoch": 4.725897920604915,
-      "grad_norm": 1.1991653442382812,
+      "epoch": 4.73,
       "learning_rate": 5e-05,
-      "loss": 1.501,
+      "loss": 1.1687,
       "step": 2500
     },
     {
-      "epoch": 4.914933837429111,
-      "grad_norm": 1.310654878616333,
+      "epoch": 4.91,
       "learning_rate": 5e-05,
-      "loss": 1.5034,
+      "loss": 1.1856,
       "step": 2600
     },
     {
       "epoch": 5.0,
-      "eval_accuracy": 0.5681025641025641,
-      "eval_loss": 1.8328346014022827,
-      "eval_runtime": 5.7763,
-      "eval_samples_per_second": 86.561,
-      "eval_steps_per_second": 10.907,
+      "eval_accuracy": 0.6125128205128205,
+      "eval_loss": 1.3968186378479004,
+      "eval_runtime": 4.3081,
+      "eval_samples_per_second": 116.06,
+      "eval_steps_per_second": 14.624,
       "step": 2645
     },
     {
       "epoch": 5.0,
-      "eval_exact_match": 18.8,
-      "eval_f1": 26.6015873015873,
+      "eval_exact_match": 24.6,
+      "eval_f1": 37.079523809523806,
       "step": 2645
     },
     {
-      "epoch": 5.103969754253308,
-      "grad_norm": 1.1406631469726562,
+      "epoch": 5.1,
       "learning_rate": 5e-05,
-      "loss": 1.4605,
+      "loss": 1.1376,
       "step": 2700
     },
     {
-      "epoch": 5.293005671077505,
-      "grad_norm": 1.447221279144287,
+      "epoch": 5.29,
       "learning_rate": 5e-05,
-      "loss": 1.4204,
+      "loss": 1.1079,
       "step": 2800
     },
     {
-      "epoch": 5.482041587901701,
-      "grad_norm": 1.461799144744873,
+      "epoch": 5.48,
       "learning_rate": 5e-05,
-      "loss": 1.4233,
+      "loss": 1.0954,
       "step": 2900
     },
     {
-      "epoch": 5.671077504725898,
-      "grad_norm": 1.4184092283248901,
+      "epoch": 5.67,
       "learning_rate": 5e-05,
-      "loss": 1.4265,
+      "loss": 1.1317,
       "step": 3000
     },
     {
-      "epoch": 5.8601134215500945,
-      "grad_norm": 0.9850455522537231,
+      "epoch": 5.86,
       "learning_rate": 5e-05,
-      "loss": 1.4139,
+      "loss": 1.1098,
       "step": 3100
     },
     {
       "epoch": 6.0,
-      "eval_accuracy": 0.564974358974359,
-      "eval_loss": 1.931396484375,
-      "eval_runtime": 5.3358,
-      "eval_samples_per_second": 93.706,
-      "eval_steps_per_second": 11.807,
+      "eval_accuracy": 0.6114871794871795,
+      "eval_loss": 1.4720605611801147,
+      "eval_runtime": 4.5,
+      "eval_samples_per_second": 111.11,
+      "eval_steps_per_second": 14.0,
       "step": 3174
     },
     {
       "epoch": 6.0,
-      "eval_exact_match": 17.6,
-      "eval_f1": 25.464285714285715,
+      "eval_exact_match": 24.4,
+      "eval_f1": 35.78015873015873,
       "step": 3174
     },
     {
-      "epoch": 6.049149338374291,
-      "grad_norm": 1.2274116277694702,
+      "epoch": 6.05,
       "learning_rate": 5e-05,
-      "loss": 1.4159,
+      "loss": 1.0945,
       "step": 3200
     },
     {
-      "epoch": 6.238185255198488,
-      "grad_norm": 1.2107988595962524,
+      "epoch": 6.24,
       "learning_rate": 5e-05,
-      "loss": 1.3579,
+      "loss": 1.0309,
       "step": 3300
     },
     {
-      "epoch": 6.427221172022684,
-      "grad_norm": 1.2965952157974243,
+      "epoch": 6.43,
       "learning_rate": 5e-05,
-      "loss": 1.3485,
+      "loss": 1.0389,
       "step": 3400
     },
     {
-      "epoch": 6.616257088846881,
-      "grad_norm": 1.3035151958465576,
+      "epoch": 6.62,
       "learning_rate": 5e-05,
-      "loss": 1.36,
+      "loss": 1.0547,
       "step": 3500
     },
     {
-      "epoch": 6.805293005671078,
-      "grad_norm": 1.1840296983718872,
+      "epoch": 6.81,
       "learning_rate": 5e-05,
-      "loss": 1.3599,
+      "loss": 1.0618,
       "step": 3600
     },
     {
-      "epoch": 6.994328922495274,
-      "grad_norm": 1.780287265777588,
+      "epoch": 6.99,
       "learning_rate": 5e-05,
-      "loss": 1.355,
+      "loss": 1.0753,
       "step": 3700
     },
     {
       "epoch": 7.0,
-      "eval_accuracy": 0.565076923076923,
-      "eval_loss": 1.9659509658813477,
-      "eval_runtime": 5.3127,
-      "eval_samples_per_second": 94.114,
-      "eval_steps_per_second": 11.858,
+      "eval_accuracy": 0.6076410256410256,
+      "eval_loss": 1.5797967910766602,
+      "eval_runtime": 4.8128,
+      "eval_samples_per_second": 103.891,
+      "eval_steps_per_second": 13.09,
       "step": 3703
     },
     {
       "epoch": 7.0,
-      "eval_exact_match": 18.0,
-      "eval_f1": 26.546666666666663,
+      "eval_exact_match": 21.8,
+      "eval_f1": 33.15111111111111,
       "step": 3703
     },
     {
-      "epoch": 7.183364839319471,
-      "grad_norm": 1.1608421802520752,
+      "epoch": 7.18,
       "learning_rate": 5e-05,
-      "loss": 1.298,
+      "loss": 0.9727,
       "step": 3800
     },
     {
-      "epoch": 7.3724007561436675,
-      "grad_norm": 1.083270788192749,
+      "epoch": 7.37,
       "learning_rate": 5e-05,
-      "loss": 1.3088,
+      "loss": 0.9815,
       "step": 3900
     },
     {
-      "epoch": 7.561436672967864,
-      "grad_norm": 1.3489899635314941,
+      "epoch": 7.56,
       "learning_rate": 5e-05,
-      "loss": 1.299,
+      "loss": 0.9777,
       "step": 4000
     },
     {
-      "epoch": 7.750472589792061,
-      "grad_norm": 1.1663447618484497,
+      "epoch": 7.75,
       "learning_rate": 5e-05,
-      "loss": 1.2841,
+      "loss": 0.9984,
       "step": 4100
     },
     {
-      "epoch": 7.939508506616257,
-      "grad_norm": 1.9308583736419678,
+      "epoch": 7.94,
       "learning_rate": 5e-05,
-      "loss": 1.314,
+      "loss": 1.0048,
       "step": 4200
     },
     {
       "epoch": 8.0,
-      "eval_accuracy": 0.5616923076923077,
-      "eval_loss": 2.1164443492889404,
-      "eval_runtime": 5.2165,
-      "eval_samples_per_second": 95.85,
-      "eval_steps_per_second": 12.077,
+      "eval_accuracy": 0.6084102564102564,
+      "eval_loss": 1.6041829586029053,
+      "eval_runtime": 4.4087,
+      "eval_samples_per_second": 113.411,
+      "eval_steps_per_second": 14.29,
       "step": 4232
     },
     {
       "epoch": 8.0,
-      "eval_exact_match": 17.0,
-      "eval_f1": 24.659841269841277,
+      "eval_exact_match": 22.4,
+      "eval_f1": 35.09904761904763,
       "step": 4232
     },
     {
-      "epoch": 8.128544423440454,
-      "grad_norm": 1.109632134437561,
+      "epoch": 8.13,
       "learning_rate": 5e-05,
-      "loss": 1.2558,
+      "loss": 0.9568,
       "step": 4300
     },
     {
-      "epoch": 8.31758034026465,
-      "grad_norm": 1.1132491827011108,
+      "epoch": 8.32,
       "learning_rate": 5e-05,
-      "loss": 1.2264,
+      "loss": 0.9088,
       "step": 4400
     },
     {
-      "epoch": 8.506616257088847,
-      "grad_norm": 1.1017369031906128,
+      "epoch": 8.51,
       "learning_rate": 5e-05,
-      "loss": 1.2649,
+      "loss": 0.9341,
       "step": 4500
     },
     {
-      "epoch": 8.695652173913043,
-      "grad_norm": 1.0874963998794556,
+      "epoch": 8.7,
       "learning_rate": 5e-05,
-      "loss": 1.2512,
+      "loss": 0.9285,
       "step": 4600
     },
     {
-      "epoch": 8.88468809073724,
-      "grad_norm": 1.2617064714431763,
+      "epoch": 8.88,
       "learning_rate": 5e-05,
-      "loss": 1.261,
+      "loss": 0.9456,
       "step": 4700
     },
     {
       "epoch": 9.0,
-      "eval_accuracy": 0.5622051282051282,
-      "eval_loss": 2.1635327339172363,
-      "eval_runtime": 5.0832,
-      "eval_samples_per_second": 98.363,
-      "eval_steps_per_second": 12.394,
+      "eval_accuracy": 0.5976923076923077,
+      "eval_loss": 1.684310793876648,
+      "eval_runtime": 4.9249,
+      "eval_samples_per_second": 101.524,
+      "eval_steps_per_second": 12.792,
       "step": 4761
     },
     {
       "epoch": 9.0,
-      "eval_exact_match": 17.8,
-      "eval_f1": 25.451428571428576,
+      "eval_exact_match": 24.6,
+      "eval_f1": 36.32493506493508,
       "step": 4761
     },
     {
-      "epoch": 9.073724007561436,
-      "grad_norm": 1.2258027791976929,
+      "epoch": 9.07,
       "learning_rate": 5e-05,
-      "loss": 1.2298,
+      "loss": 0.912,
       "step": 4800
     },
     {
-      "epoch": 9.262759924385634,
-      "grad_norm": 1.8261148929595947,
+      "epoch": 9.26,
       "learning_rate": 5e-05,
-      "loss": 1.1748,
+      "loss": 0.8443,
       "step": 4900
     },
     {
-      "epoch": 9.45179584120983,
-      "grad_norm": 1.1221072673797607,
+      "epoch": 9.45,
       "learning_rate": 5e-05,
-      "loss": 1.19,
+      "loss": 0.8789,
       "step": 5000
     },
     {
-      "epoch": 9.640831758034027,
-      "grad_norm": 1.3320170640945435,
+      "epoch": 9.64,
       "learning_rate": 5e-05,
-      "loss": 1.1932,
+      "loss": 0.8865,
       "step": 5100
     },
     {
-      "epoch": 9.829867674858223,
-      "grad_norm": 1.194077730178833,
+      "epoch": 9.83,
       "learning_rate": 5e-05,
-      "loss": 1.217,
+      "loss": 0.8766,
       "step": 5200
     },
     {
       "epoch": 10.0,
-      "eval_accuracy": 0.5612820512820513,
-      "eval_loss": 2.244041919708252,
-      "eval_runtime": 5.1606,
-      "eval_samples_per_second": 96.889,
-      "eval_steps_per_second": 12.208,
+      "eval_accuracy": 0.6051282051282051,
+      "eval_loss": 1.7828805446624756,
+      "eval_runtime": 4.1983,
+      "eval_samples_per_second": 119.096,
+      "eval_steps_per_second": 15.006,
       "step": 5290
     },
     {
       "epoch": 10.0,
-      "eval_exact_match": 18.4,
-      "eval_f1": 26.067662337662348,
+      "eval_exact_match": 23.0,
+      "eval_f1": 34.336190476190495,
       "step": 5290
     },
     {
-      "epoch": 10.01890359168242,
-      "grad_norm": 1.1352814435958862,
+      "epoch": 10.02,
       "learning_rate": 5e-05,
-      "loss": 1.2093,
+      "loss": 0.8835,
       "step": 5300
     },
     {
-      "epoch": 10.207939508506616,
-      "grad_norm": 1.2741838693618774,
+      "epoch": 10.21,
       "learning_rate": 5e-05,
-      "loss": 1.1467,
+      "loss": 0.8095,
       "step": 5400
     },
     {
-      "epoch": 10.396975425330814,
-      "grad_norm": 1.6901249885559082,
+      "epoch": 10.4,
       "learning_rate": 5e-05,
-      "loss": 1.1288,
+      "loss": 0.8008,
       "step": 5500
     },
     {
-      "epoch": 10.58601134215501,
-      "grad_norm": 1.1944465637207031,
+      "epoch": 10.59,
       "learning_rate": 5e-05,
-      "loss": 1.1399,
+      "loss": 0.8208,
       "step": 5600
     },
     {
-      "epoch": 10.775047258979207,
-      "grad_norm": 1.2234572172164917,
+      "epoch": 10.78,
       "learning_rate": 5e-05,
-      "loss": 1.1678,
+      "loss": 0.8282,
       "step": 5700
     },
     {
-      "epoch": 10.964083175803403,
-      "grad_norm": 1.2781612873077393,
+      "epoch": 10.96,
       "learning_rate": 5e-05,
-      "loss": 1.146,
+      "loss": 0.8273,
       "step": 5800
     },
     {
       "epoch": 11.0,
-      "eval_accuracy": 0.5618461538461539,
-      "eval_loss": 2.296630859375,
-      "eval_runtime": 5.1078,
-      "eval_samples_per_second": 97.889,
-      "eval_steps_per_second": 12.334,
+      "eval_accuracy": 0.6043076923076923,
+      "eval_loss": 1.8059720993041992,
+      "eval_runtime": 4.2264,
+      "eval_samples_per_second": 118.304,
+      "eval_steps_per_second": 14.906,
       "step": 5819
     },
     {
       "epoch": 11.0,
-      "eval_exact_match": 16.8,
-      "eval_f1": 23.63305916305917,
+      "eval_exact_match": 21.6,
+      "eval_f1": 32.628095238095234,
       "step": 5819
     },
     {
-      "epoch": 11.1531190926276,
-      "grad_norm": 1.138975739479065,
+      "epoch": 11.15,
       "learning_rate": 5e-05,
-      "loss": 1.0842,
+      "loss": 0.7637,
       "step": 5900
     },
     {
-      "epoch": 11.342155009451796,
-      "grad_norm": 1.727989912033081,
+      "epoch": 11.34,
       "learning_rate": 5e-05,
-      "loss": 1.0806,
+      "loss": 0.7581,
       "step": 6000
     },
     {
-      "epoch": 11.531190926275993,
-      "grad_norm": 1.3104358911514282,
+      "epoch": 11.53,
       "learning_rate": 5e-05,
-      "loss": 1.0953,
+      "loss": 0.7571,
       "step": 6100
     },
     {
-      "epoch": 11.720226843100189,
-      "grad_norm": 1.3629850149154663,
+      "epoch": 11.72,
       "learning_rate": 5e-05,
-      "loss": 1.0945,
+      "loss": 0.7637,
       "step": 6200
     },
     {
-      "epoch": 11.909262759924385,
-      "grad_norm": 1.3276056051254272,
+      "epoch": 11.91,
       "learning_rate": 5e-05,
-      "loss": 1.1069,
+      "loss": 0.7755,
       "step": 6300
     },
     {
       "epoch": 12.0,
-      "eval_accuracy": 0.562,
-      "eval_loss": 2.3118155002593994,
-      "eval_runtime": 5.3837,
-      "eval_samples_per_second": 92.873,
-      "eval_steps_per_second": 11.702,
+      "eval_accuracy": 0.6018974358974359,
+      "eval_loss": 1.8729300498962402,
+      "eval_runtime": 4.2001,
+      "eval_samples_per_second": 119.045,
+      "eval_steps_per_second": 15.0,
       "step": 6348
     },
     {
       "epoch": 12.0,
-      "eval_exact_match": 17.4,
-      "eval_f1": 24.43924963924964,
+      "eval_exact_match": 16.8,
+      "eval_f1": 27.317351339157547,
       "step": 6348
     },
     {
-      "epoch": 12.098298676748582,
-      "grad_norm": 1.2608078718185425,
+      "epoch": 12.1,
       "learning_rate": 5e-05,
-      "loss": 1.0418,
+      "loss": 0.7386,
       "step": 6400
     },
     {
-      "epoch": 12.287334593572778,
-      "grad_norm": 1.4456945657730103,
+      "epoch": 12.29,
       "learning_rate": 5e-05,
-      "loss": 1.0258,
+      "loss": 0.6977,
       "step": 6500
     },
     {
-      "epoch": 12.476370510396976,
-      "grad_norm": 1.4195287227630615,
+      "epoch": 12.48,
       "learning_rate": 5e-05,
-      "loss": 1.0255,
+      "loss": 0.7185,
       "step": 6600
     },
     {
-      "epoch": 12.665406427221171,
-      "grad_norm": 1.2595598697662354,
+      "epoch": 12.67,
       "learning_rate": 5e-05,
-      "loss": 1.0511,
+      "loss": 0.7166,
       "step": 6700
     },
     {
-      "epoch": 12.854442344045369,
-      "grad_norm": 1.5901292562484741,
+      "epoch": 12.85,
       "learning_rate": 5e-05,
-      "loss": 1.045,
+      "loss": 0.715,
       "step": 6800
     },
     {
       "epoch": 13.0,
-      "eval_accuracy": 0.5614871794871795,
-      "eval_loss": 2.3580987453460693,
-      "eval_runtime": 5.3496,
-      "eval_samples_per_second": 93.466,
-      "eval_steps_per_second": 11.777,
+      "eval_accuracy": 0.6016923076923076,
+      "eval_loss": 1.9620097875595093,
+      "eval_runtime": 4.4055,
+      "eval_samples_per_second": 113.495,
+      "eval_steps_per_second": 14.3,
       "step": 6877
     },
     {
       "epoch": 13.0,
-      "eval_exact_match": 17.2,
-      "eval_f1": 24.350995670995676,
+      "eval_exact_match": 4.8,
+      "eval_f1": 9.356456790950029,
       "step": 6877
     },
     {
-      "epoch": 13.043478260869565,
-      "grad_norm": 1.471511721611023,
+      "epoch": 13.04,
       "learning_rate": 5e-05,
-      "loss": 1.0404,
+      "loss": 0.7032,
       "step": 6900
     },
     {
-      "epoch": 13.232514177693762,
-      "grad_norm": 1.2928757667541504,
+      "epoch": 13.23,
       "learning_rate": 5e-05,
-      "loss": 0.9573,
+      "loss": 0.6494,
       "step": 7000
     },
     {
-      "epoch": 13.421550094517958,
-      "grad_norm": 1.5545622110366821,
+      "epoch": 13.42,
       "learning_rate": 5e-05,
-      "loss": 0.9583,
+      "loss": 0.6584,
       "step": 7100
     },
     {
-      "epoch": 13.610586011342155,
-      "grad_norm": 1.6132512092590332,
+      "epoch": 13.61,
       "learning_rate": 5e-05,
-      "loss": 0.981,
+      "loss": 0.6681,
       "step": 7200
     },
     {
-      "epoch": 13.799621928166351,
-      "grad_norm": 1.7548742294311523,
+      "epoch": 13.8,
       "learning_rate": 5e-05,
-      "loss": 0.9999,
+      "loss": 0.6556,
       "step": 7300
     },
     {
-      "epoch": 13.988657844990549,
-      "grad_norm": 1.4219907522201538,
+      "epoch": 13.99,
       "learning_rate": 5e-05,
-      "loss": 1.0028,
+      "loss": 0.6804,
       "step": 7400
     },
     {
       "epoch": 14.0,
-      "eval_accuracy": 0.5618974358974359,
-      "eval_loss": 2.413728952407837,
-      "eval_runtime": 5.0464,
-      "eval_samples_per_second": 99.08,
-      "eval_steps_per_second": 12.484,
+      "eval_accuracy": 0.6009230769230769,
+      "eval_loss": 2.00297474861145,
+      "eval_runtime": 4.2009,
+      "eval_samples_per_second": 119.021,
+      "eval_steps_per_second": 14.997,
       "step": 7406
     },
     {
       "epoch": 14.0,
-      "eval_exact_match": 16.6,
-      "eval_f1": 24.201789321789327,
+      "eval_exact_match": 5.8,
+      "eval_f1": 13.996400415302853,
       "step": 7406
     },
     {
-      "epoch": 14.177693761814744,
-      "grad_norm": 1.5799965858459473,
+      "epoch": 14.18,
       "learning_rate": 5e-05,
-      "loss": 0.9096,
+      "loss": 0.5983,
       "step": 7500
     },
     {
-      "epoch": 14.366729678638942,
-      "grad_norm": 1.5961594581604004,
+      "epoch": 14.37,
       "learning_rate": 5e-05,
-      "loss": 0.9075,
+      "loss": 0.6056,
       "step": 7600
     },
     {
-      "epoch": 14.555765595463138,
-      "grad_norm": 1.5756512880325317,
+      "epoch": 14.56,
       "learning_rate": 5e-05,
-      "loss": 0.9222,
+      "loss": 0.6048,
       "step": 7700
     },
     {
-      "epoch": 14.744801512287335,
-      "grad_norm": 1.515251636505127,
+      "epoch": 14.74,
       "learning_rate": 5e-05,
-      "loss": 0.9445,
+      "loss": 0.6322,
       "step": 7800
     },
     {
-      "epoch": 14.93383742911153,
-      "grad_norm": 2.6804299354553223,
+      "epoch": 14.93,
       "learning_rate": 5e-05,
-      "loss": 0.9153,
+      "loss": 0.6277,
       "step": 7900
     },
     {
       "epoch": 15.0,
-      "eval_accuracy": 0.5596923076923077,
-      "eval_loss": 2.4756176471710205,
-      "eval_runtime": 5.1956,
-      "eval_samples_per_second": 96.236,
-      "eval_steps_per_second": 12.126,
+      "eval_accuracy": 0.5998461538461538,
+      "eval_loss": 2.0527751445770264,
+      "eval_runtime": 4.4287,
+      "eval_samples_per_second": 112.9,
+      "eval_steps_per_second": 14.225,
       "step": 7935
     },
     {
       "epoch": 15.0,
-      "eval_exact_match": 16.8,
-      "eval_f1": 24.146709956709966,
+      "eval_exact_match": 3.2,
+      "eval_f1": 8.78350649350649,
       "step": 7935
     },
     {
-      "epoch": 15.122873345935728,
-      "grad_norm": 1.7786214351654053,
+      "epoch": 15.12,
       "learning_rate": 5e-05,
-      "loss": 0.8874,
+      "loss": 0.5776,
       "step": 8000
     },
     {
-      "epoch": 15.311909262759924,
-      "grad_norm": 1.709060788154602,
+      "epoch": 15.31,
       "learning_rate": 5e-05,
-      "loss": 0.8538,
+      "loss": 0.5522,
       "step": 8100
     },
     {
-      "epoch": 15.500945179584122,
-      "grad_norm": 2.3245434761047363,
+      "epoch": 15.5,
       "learning_rate": 5e-05,
-      "loss": 0.8591,
+      "loss": 0.5735,
       "step": 8200
     },
     {
-      "epoch": 15.689981096408317,
-      "grad_norm": 1.6101887226104736,
+      "epoch": 15.69,
       "learning_rate": 5e-05,
-      "loss": 0.8593,
+      "loss": 0.5637,
       "step": 8300
     },
     {
-      "epoch": 15.879017013232515,
-      "grad_norm": 1.5917792320251465,
+      "epoch": 15.88,
       "learning_rate": 5e-05,
-      "loss": 0.8748,
+      "loss": 0.5733,
       "step": 8400
     },
     {
       "epoch": 16.0,
-      "eval_accuracy": 0.5604615384615385,
-      "eval_loss": 2.4434447288513184,
-      "eval_runtime": 5.1072,
-      "eval_samples_per_second": 97.901,
-      "eval_steps_per_second": 12.336,
+      "eval_accuracy": 0.6012307692307692,
+      "eval_loss": 2.047515392303467,
+      "eval_runtime": 4.1989,
+      "eval_samples_per_second": 119.079,
+      "eval_steps_per_second": 15.004,
       "step": 8464
     },
     {
       "epoch": 16.0,
-      "eval_exact_match": 15.6,
-      "eval_f1": 22.917027417027416,
+      "eval_exact_match": 0.6,
+      "eval_f1": 1.077443805758639,
       "step": 8464
     },
     {
-      "epoch": 16.068052930056712,
-      "grad_norm": 1.6529980897903442,
+      "epoch": 16.07,
       "learning_rate": 5e-05,
-      "loss": 0.8413,
+      "loss": 0.5601,
       "step": 8500
     },
     {
-      "epoch": 16.257088846880908,
-      "grad_norm": 1.931520938873291,
+      "epoch": 16.26,
       "learning_rate": 5e-05,
-      "loss": 0.7863,
+      "loss": 0.5126,
       "step": 8600
     },
     {
-      "epoch": 16.446124763705104,
-      "grad_norm": 1.8401894569396973,
+      "epoch": 16.45,
       "learning_rate": 5e-05,
-      "loss": 0.799,
+      "loss": 0.5217,
       "step": 8700
     },
     {
-      "epoch": 16.6351606805293,
-      "grad_norm": 1.7952015399932861,
+      "epoch": 16.64,
       "learning_rate": 5e-05,
-      "loss": 0.8238,
+      "loss": 0.5291,
       "step": 8800
     },
     {
-      "epoch": 16.8241965973535,
-      "grad_norm": 1.8028634786605835,
+      "epoch": 16.82,
       "learning_rate": 5e-05,
-      "loss": 0.807,
+      "loss": 0.5409,
       "step": 8900
     },
     {
       "epoch": 17.0,
-      "eval_accuracy": 0.5606153846153846,
-      "eval_loss": 2.4520959854125977,
-      "eval_runtime": 5.1769,
-      "eval_samples_per_second": 96.584,
-      "eval_steps_per_second": 12.17,
+      "eval_accuracy": 0.5748717948717949,
+      "eval_loss": 2.092020273208618,
+      "eval_runtime": 4.1044,
+      "eval_samples_per_second": 121.822,
+      "eval_steps_per_second": 15.35,
       "step": 8993
     },
     {
       "epoch": 17.0,
-      "eval_exact_match": 17.4,
-      "eval_f1": 25.104011544011545,
+      "eval_exact_match": 0.4,
+      "eval_f1": 2.398489535596529,
       "step": 8993
     },
     {
-      "epoch": 17.013232514177695,
-      "grad_norm": 1.7197688817977905,
+      "epoch": 17.01,
       "learning_rate": 5e-05,
-      "loss": 0.8249,
+      "loss": 0.5354,
       "step": 9000
     },
     {
-      "epoch": 17.20226843100189,
-      "grad_norm": 1.768463134765625,
+      "epoch": 17.2,
       "learning_rate": 5e-05,
-      "loss": 0.7304,
+      "loss": 0.4629,
       "step": 9100
     },
     {
-      "epoch": 17.391304347826086,
-      "grad_norm": 1.9696024656295776,
+      "epoch": 17.39,
       "learning_rate": 5e-05,
-      "loss": 0.7445,
+      "loss": 0.4795,
       "step": 9200
     },
     {
-      "epoch": 17.58034026465028,
-      "grad_norm": 2.093703269958496,
+      "epoch": 17.58,
       "learning_rate": 5e-05,
-      "loss": 0.755,
+      "loss": 0.4921,
       "step": 9300
     },
     {
-      "epoch": 17.76937618147448,
-      "grad_norm": 1.909550666809082,
+      "epoch": 17.77,
       "learning_rate": 5e-05,
-      "loss": 0.7698,
+      "loss": 0.4928,
       "step": 9400
     },
     {
-      "epoch": 17.958412098298677,
-      "grad_norm": 1.6835806369781494,
+      "epoch": 17.96,
       "learning_rate": 5e-05,
-      "loss": 0.7711,
+      "loss": 0.5024,
       "step": 9500
     },
     {
       "epoch": 18.0,
-      "eval_accuracy": 0.5603589743589743,
-      "eval_loss": 2.473491668701172,
-      "eval_runtime": 5.4257,
-      "eval_samples_per_second": 92.155,
-      "eval_steps_per_second": 11.611,
+      "eval_accuracy": 0.5986153846153847,
+      "eval_loss": 2.120666980743408,
+      "eval_runtime": 4.3114,
+      "eval_samples_per_second": 115.973,
+      "eval_steps_per_second": 14.613,
       "step": 9522
     },
     {
       "epoch": 18.0,
-      "eval_exact_match": 17.0,
-      "eval_f1": 24.074487734487732,
+      "eval_exact_match": 0.2,
+      "eval_f1": 0.6343183616587786,
       "step": 9522
     },
     {
-      "epoch": 18.147448015122873,
-      "grad_norm": 2.0483736991882324,
+      "epoch": 18.15,
       "learning_rate": 5e-05,
-      "loss": 0.7048,
+      "loss": 0.4539,
       "step": 9600
     },
     {
-      "epoch": 18.33648393194707,
-      "grad_norm": 1.987457275390625,
+      "epoch": 18.34,
       "learning_rate": 5e-05,
-      "loss": 0.6863,
+      "loss": 0.4345,
       "step": 9700
     },
     {
-      "epoch": 18.525519848771268,
-      "grad_norm": 2.1744563579559326,
+      "epoch": 18.53,
       "learning_rate": 5e-05,
-      "loss": 0.7045,
+      "loss": 0.4399,
       "step": 9800
     },
     {
-      "epoch": 18.714555765595463,
-      "grad_norm": 1.9871633052825928,
+      "epoch": 18.71,
       "learning_rate": 5e-05,
-      "loss": 0.7076,
+      "loss": 0.4581,
       "step": 9900
     },
     {
-      "epoch": 18.90359168241966,
-      "grad_norm": 1.8895001411437988,
+      "epoch": 18.9,
       "learning_rate": 5e-05,
-      "loss": 0.7202,
+      "loss": 0.4699,
       "step": 10000
     },
     {
       "epoch": 19.0,
-      "eval_accuracy": 0.5601538461538461,
-      "eval_loss": 2.5168700218200684,
-      "eval_runtime": 5.0434,
-      "eval_samples_per_second": 99.14,
-      "eval_steps_per_second": 12.492,
+      "eval_accuracy": 0.5992820512820513,
+      "eval_loss": 2.110786199569702,
+      "eval_runtime": 4.2008,
+      "eval_samples_per_second": 119.024,
+      "eval_steps_per_second": 14.997,
       "step": 10051
     },
     {
       "epoch": 19.0,
-      "eval_exact_match": 17.2,
-      "eval_f1": 24.260360750360757,
+      "eval_exact_match": 0.0,
+      "eval_f1": 0.006153846153846154,
       "step": 10051
     },
     {
-      "epoch": 19.092627599243855,
-      "grad_norm": 1.7914209365844727,
+      "epoch": 19.09,
       "learning_rate": 5e-05,
-      "loss": 0.6728,
+      "loss": 0.4363,
       "step": 10100
     },
     {
-      "epoch": 19.281663516068054,
-      "grad_norm": 2.198495388031006,
+      "epoch": 19.28,
       "learning_rate": 5e-05,
-      "loss": 0.6241,
+      "loss": 0.4021,
       "step": 10200
     },
     {
-      "epoch": 19.47069943289225,
-      "grad_norm": 1.81365966796875,
+      "epoch": 19.47,
       "learning_rate": 5e-05,
-      "loss": 0.6558,
+      "loss": 0.4161,
       "step": 10300
     },
     {
-      "epoch": 19.659735349716446,
-      "grad_norm": 1.6571800708770752,
+      "epoch": 19.66,
       "learning_rate": 5e-05,
-      "loss": 0.6652,
+      "loss": 0.425,
       "step": 10400
     },
     {
-      "epoch": 19.84877126654064,
-      "grad_norm": 1.6469954252243042,
+      "epoch": 19.85,
       "learning_rate": 5e-05,
-      "loss": 0.6637,
+      "loss": 0.4367,
       "step": 10500
     },
     {
       "epoch": 20.0,
-      "eval_accuracy": 0.5596410256410257,
-      "eval_loss": 2.516073703765869,
-      "eval_runtime": 5.8234,
-      "eval_samples_per_second": 85.86,
-      "eval_steps_per_second": 10.818,
+      "eval_accuracy": 0.6004615384615385,
+      "eval_loss": 2.1088976860046387,
+      "eval_runtime": 4.0958,
+      "eval_samples_per_second": 122.078,
+      "eval_steps_per_second": 15.382,
       "step": 10580
     },
     {
       "epoch": 20.0,
-      "eval_exact_match": 17.6,
-      "eval_f1": 25.225079365079367,
+      "eval_exact_match": 0.0,
+      "eval_f1": 0.04062256613300257,
       "step": 10580
     },
     {
-      "epoch": 20.03780718336484,
-      "grad_norm": 1.7114077806472778,
+      "epoch": 20.04,
       "learning_rate": 5e-05,
-      "loss": 0.66,
+      "loss": 0.4264,
       "step": 10600
     },
     {
-      "epoch": 20.226843100189036,
-      "grad_norm": 1.9962202310562134,
+      "epoch": 20.23,
       "learning_rate": 5e-05,
-      "loss": 0.5938,
+      "loss": 0.3746,
       "step": 10700
     },
     {
-      "epoch": 20.415879017013232,
-      "grad_norm": 2.397592067718506,
+      "epoch": 20.42,
       "learning_rate": 5e-05,
-      "loss": 0.6042,
+      "loss": 0.3834,
       "step": 10800
     },
     {
-      "epoch": 20.604914933837428,
-      "grad_norm": 2.0724213123321533,
+      "epoch": 20.6,
       "learning_rate": 5e-05,
-      "loss": 0.6124,
+      "loss": 0.4015,
       "step": 10900
     },
     {
-      "epoch": 20.793950850661627,
-      "grad_norm": 1.9385457038879395,
+      "epoch": 20.79,
       "learning_rate": 5e-05,
-      "loss": 0.6078,
+      "loss": 0.4284,
       "step": 11000
     },
     {
-      "epoch": 20.982986767485823,
-      "grad_norm": 2.034963607788086,
+      "epoch": 20.98,
       "learning_rate": 5e-05,
-      "loss": 0.6257,
+      "loss": 0.857,
       "step": 11100
     },
     {
       "epoch": 21.0,
-      "eval_accuracy": 0.5597948717948718,
-      "eval_loss": 2.5497183799743652,
-      "eval_runtime": 5.0442,
-      "eval_samples_per_second": 99.123,
-      "eval_steps_per_second": 12.489,
+      "eval_accuracy": 0.5983076923076923,
+      "eval_loss": 2.0215413570404053,
+      "eval_runtime": 4.0942,
+      "eval_samples_per_second": 122.125,
+      "eval_steps_per_second": 15.388,
       "step": 11109
     },
     {
       "epoch": 21.0,
-      "eval_exact_match": 17.4,
-      "eval_f1": 24.755238095238095,
+      "eval_exact_match": 0.0,
+      "eval_f1": 0.0,
       "step": 11109
     },
     {
-      "epoch": 21.17202268431002,
-      "grad_norm": 1.9768177270889282,
+      "epoch": 21.17,
       "learning_rate": 5e-05,
-      "loss": 0.5458,
+      "loss": 0.8884,
       "step": 11200
     },
     {
-      "epoch": 21.361058601134214,
-      "grad_norm": 2.14420747756958,
+      "epoch": 21.36,
       "learning_rate": 5e-05,
-      "loss": 0.558,
+      "loss": 1.4123,
       "step": 11300
     },
     {
-      "epoch": 21.550094517958414,
-      "grad_norm": 2.5267951488494873,
+      "epoch": 21.55,
       "learning_rate": 5e-05,
-      "loss": 0.5646,
+      "loss": 2.3033,
       "step": 11400
     },
     {
-      "epoch": 21.73913043478261,
-      "grad_norm": 2.114386558532715,
+      "epoch": 21.74,
       "learning_rate": 5e-05,
-      "loss": 0.572,
+      "loss": 2.1144,
       "step": 11500
     },
     {
-      "epoch": 21.928166351606805,
-      "grad_norm": 2.424217700958252,
+      "epoch": 21.93,
       "learning_rate": 5e-05,
-      "loss": 0.5928,
+      "loss": 3.7434,
       "step": 11600
     },
     {
       "epoch": 22.0,
-      "eval_accuracy": 0.5593846153846154,
-      "eval_loss": 2.579535722732544,
-      "eval_runtime": 5.7849,
-      "eval_samples_per_second": 86.433,
-      "eval_steps_per_second": 10.89,
+      "eval_accuracy": 0.22333333333333333,
+      "eval_loss": 10.118596076965332,
+      "eval_runtime": 4.3085,
+      "eval_samples_per_second": 116.049,
+      "eval_steps_per_second": 14.622,
       "step": 11638
     },
     {
       "epoch": 22.0,
-      "eval_exact_match": 16.8,
-      "eval_f1": 23.666507936507937,
+      "eval_exact_match": 0.0,
+      "eval_f1": 0.0,
       "step": 11638
     },
     {
-      "epoch": 22.117202268431,
-      "grad_norm": 1.7813427448272705,
+      "epoch": 22.12,
       "learning_rate": 5e-05,
-      "loss": 0.5342,
+      "loss": 9.1896,
       "step": 11700
     },
     {
-      "epoch": 22.3062381852552,
-      "grad_norm": 1.989479422569275,
+      "epoch": 22.31,
       "learning_rate": 5e-05,
-      "loss": 0.5157,
+      "loss": 9.6027,
       "step": 11800
     },
     {
-      "epoch": 22.495274102079396,
-      "grad_norm": 1.9694567918777466,
+      "epoch": 22.5,
       "learning_rate": 5e-05,
-      "loss": 0.5236,
+      "loss": 11.4681,
       "step": 11900
     },
     {
-      "epoch": 22.68431001890359,
-      "grad_norm": 2.0226142406463623,
+      "epoch": 22.68,
       "learning_rate": 5e-05,
-      "loss": 0.5315,
+      "loss": 8.4061,
       "step": 12000
     },
     {
-      "epoch": 22.873345935727787,
-      "grad_norm": 2.0204625129699707,
+      "epoch": 22.87,
       "learning_rate": 5e-05,
-      "loss": 0.5378,
+      "loss": 7.7259,
       "step": 12100
     },
     {
       "epoch": 23.0,
-      "eval_accuracy": 0.558974358974359,
-      "eval_loss": 2.5709009170532227,
-      "eval_runtime": 5.0788,
-      "eval_samples_per_second": 98.449,
-      "eval_steps_per_second": 12.405,
+      "eval_accuracy": 0.19856410256410256,
+      "eval_loss": 7.537852764129639,
+      "eval_runtime": 4.1062,
+      "eval_samples_per_second": 121.768,
+      "eval_steps_per_second": 15.343,
       "step": 12167
     },
     {
       "epoch": 23.0,
-      "eval_exact_match": 16.4,
-      "eval_f1": 23.69809523809524,
+      "eval_exact_match": 0.0,
+      "eval_f1": 0.08333333333333333,
       "step": 12167
     },
     {
-      "epoch": 23.062381852551987,
-      "grad_norm": 2.0633811950683594,
+      "epoch": 23.06,
       "learning_rate": 5e-05,
-      "loss": 0.5183,
+      "loss": 7.3869,
       "step": 12200
     },
     {
-      "epoch": 23.251417769376182,
-      "grad_norm": 2.3566079139709473,
+      "epoch": 23.25,
       "learning_rate": 5e-05,
-      "loss": 0.4761,
+      "loss": 7.3438,
       "step": 12300
     },
     {
-      "epoch": 23.440453686200378,
-      "grad_norm": 2.368450880050659,
+      "epoch": 23.44,
       "learning_rate": 5e-05,
-      "loss": 0.487,
+      "loss": 7.1672,
       "step": 12400
     },
     {
-      "epoch": 23.629489603024574,
-      "grad_norm": 2.1096372604370117,
+      "epoch": 23.63,
       "learning_rate": 5e-05,
-      "loss": 0.4945,
+      "loss": 7.5678,
       "step": 12500
     },
     {
-      "epoch": 23.81852551984877,
-      "grad_norm": 2.2136454582214355,
+      "epoch": 23.82,
       "learning_rate": 5e-05,
-      "loss": 0.5035,
+      "loss": 4.2204,
       "step": 12600
     },
     {
       "epoch": 24.0,
-      "eval_accuracy": 0.5586153846153846,
-      "eval_loss": 2.5898056030273438,
-      "eval_runtime": 5.7458,
-      "eval_samples_per_second": 87.021,
-      "eval_steps_per_second": 10.965,
+      "eval_accuracy": 0.5344615384615384,
+      "eval_loss": 2.1567773818969727,
+      "eval_runtime": 4.8187,
+      "eval_samples_per_second": 103.763,
+      "eval_steps_per_second": 13.074,
       "step": 12696
     },
     {
       "epoch": 24.0,
-      "eval_exact_match": 17.8,
-      "eval_f1": 24.89603174603175,
+      "eval_exact_match": 0.0,
+      "eval_f1": 0.1556396274796546,
       "step": 12696
     },
     {
-      "epoch": 24.00756143667297,
-      "grad_norm": 2.317958116531372,
+      "epoch": 24.01,
       "learning_rate": 5e-05,
-      "loss": 0.5106,
+      "loss": 2.243,
       "step": 12700
     },
     {
-      "epoch": 24.196597353497165,
-      "grad_norm": 2.0783536434173584,
+      "epoch": 24.2,
       "learning_rate": 5e-05,
-      "loss": 0.4404,
+      "loss": 1.4067,
       "step": 12800
     },
     {
-      "epoch": 24.38563327032136,
-      "grad_norm": 2.322791337966919,
+      "epoch": 24.39,
       "learning_rate": 5e-05,
-      "loss": 0.452,
+      "loss": 1.0065,
       "step": 12900
     },
     {
-      "epoch": 24.574669187145556,
-      "grad_norm": 1.9763044118881226,
+      "epoch": 24.57,
       "learning_rate": 5e-05,
-      "loss": 0.4617,
+      "loss": 0.8179,
       "step": 13000
     },
     {
-      "epoch": 24.763705103969755,
-      "grad_norm": 2.462392568588257,
+      "epoch": 24.76,
       "learning_rate": 5e-05,
-      "loss": 0.4731,
+      "loss": 0.7917,
       "step": 13100
     },
     {
-      "epoch": 24.95274102079395,
-      "grad_norm": 2.2400307655334473,
+      "epoch": 24.95,
       "learning_rate": 5e-05,
-      "loss": 0.4751,
+      "loss": 0.7385,
       "step": 13200
     },
     {
       "epoch": 25.0,
-      "eval_accuracy": 0.5594871794871795,
-      "eval_loss": 2.611313581466675,
-      "eval_runtime": 5.3564,
-      "eval_samples_per_second": 93.346,
-      "eval_steps_per_second": 11.762,
+      "eval_accuracy": 0.5963076923076923,
+      "eval_loss": 1.8228886127471924,
+      "eval_runtime": 4.8179,
+      "eval_samples_per_second": 103.781,
+      "eval_steps_per_second": 13.076,
       "step": 13225
     },
     {
       "epoch": 25.0,
-      "eval_exact_match": 15.8,
-      "eval_f1": 22.969047619047625,
+      "eval_exact_match": 5.2,
+      "eval_f1": 8.718379522680063,
       "step": 13225
     },
     {
-      "epoch": 25.141776937618147,
-      "grad_norm": 2.0931668281555176,
+      "epoch": 25.14,
       "learning_rate": 5e-05,
-      "loss": 0.4299,
+      "loss": 0.8862,
       "step": 13300
     },
     {
-      "epoch": 25.330812854442343,
-      "grad_norm": 2.7056140899658203,
+      "epoch": 25.33,
       "learning_rate": 5e-05,
-      "loss": 0.422,
+      "loss": 0.5963,
       "step": 13400
     },
     {
-      "epoch": 25.519848771266542,
-      "grad_norm": 1.9907532930374146,
+      "epoch": 25.52,
       "learning_rate": 5e-05,
-      "loss": 0.4374,
+      "loss": 2.7783,
       "step": 13500
     },
     {
-      "epoch": 25.708884688090738,
-      "grad_norm": 2.3209221363067627,
+      "epoch": 25.71,
       "learning_rate": 5e-05,
-      "loss": 0.4376,
+      "loss": 1.4347,
       "step": 13600
     },
     {
-      "epoch": 25.897920604914933,
-      "grad_norm": 3.2339212894439697,
+      "epoch": 25.9,
       "learning_rate": 5e-05,
-      "loss": 0.4432,
+      "loss": 1.1473,
       "step": 13700
     },
     {
       "epoch": 26.0,
-      "eval_accuracy": 0.5602051282051282,
-      "eval_loss": 2.6182327270507812,
-      "eval_runtime": 5.0321,
-      "eval_samples_per_second": 99.362,
-      "eval_steps_per_second": 12.52,
+      "eval_accuracy": 0.5788205128205128,
+      "eval_loss": 1.7569822072982788,
+      "eval_runtime": 4.1999,
+      "eval_samples_per_second": 119.049,
+      "eval_steps_per_second": 15.0,
       "step": 13754
     },
     {
       "epoch": 26.0,
-      "eval_exact_match": 15.4,
-      "eval_f1": 23.059047619047618,
+      "eval_exact_match": 0.0,
+      "eval_f1": 0.0,
       "step": 13754
     },
     {
-      "epoch": 26.08695652173913,
-      "grad_norm": 1.9935250282287598,
+      "epoch": 26.09,
       "learning_rate": 5e-05,
-      "loss": 0.4265,
+      "loss": 1.2078,
       "step": 13800
     },
     {
-      "epoch": 26.27599243856333,
-      "grad_norm": 2.5698649883270264,
+      "epoch": 26.28,
       "learning_rate": 5e-05,
-      "loss": 0.3949,
+      "loss": 1.2602,
       "step": 13900
     },
     {
-      "epoch": 26.465028355387524,
-      "grad_norm": 2.568392038345337,
+      "epoch": 26.47,
       "learning_rate": 5e-05,
-      "loss": 0.4041,
+      "loss": 2.3188,
       "step": 14000
     },
     {
-      "epoch": 26.65406427221172,
-      "grad_norm": 2.094651699066162,
+      "epoch": 26.65,
       "learning_rate": 5e-05,
-      "loss": 0.4128,
+      "loss": 2.2177,
       "step": 14100
     },
     {
-      "epoch": 26.843100189035916,
-      "grad_norm": 2.240649700164795,
+      "epoch": 26.84,
       "learning_rate": 5e-05,
-      "loss": 0.4191,
+      "loss": 2.0182,
       "step": 14200
     },
     {
       "epoch": 27.0,
-      "eval_accuracy": 0.56,
-      "eval_loss": 2.614851951599121,
-      "eval_runtime": 5.636,
-      "eval_samples_per_second": 88.715,
-      "eval_steps_per_second": 11.178,
+      "eval_accuracy": 0.5573333333333333,
+      "eval_loss": 1.729324221611023,
+      "eval_runtime": 4.5012,
+      "eval_samples_per_second": 111.082,
+      "eval_steps_per_second": 13.996,
       "step": 14283
     },
     {
       "epoch": 27.0,
-      "eval_exact_match": 17.2,
-      "eval_f1": 24.941948051948046,
+      "eval_exact_match": 0.0,
+      "eval_f1": 0.03798603689725877,
       "step": 14283
     },
     {
-      "epoch": 27.032136105860115,
-      "grad_norm": 1.8800567388534546,
+      "epoch": 27.03,
       "learning_rate": 5e-05,
-      "loss": 0.4135,
+      "loss": 1.4892,
       "step": 14300
     },
     {
-      "epoch": 27.22117202268431,
-      "grad_norm": 2.525048017501831,
+      "epoch": 27.22,
       "learning_rate": 5e-05,
-      "loss": 0.3717,
+      "loss": 1.6886,
       "step": 14400
     },
     {
-      "epoch": 27.410207939508506,
-      "grad_norm": 2.3803513050079346,
+      "epoch": 27.41,
       "learning_rate": 5e-05,
-      "loss": 0.3817,
+      "loss": 2.2033,
       "step": 14500
     },
     {
-      "epoch": 27.599243856332702,
-      "grad_norm": 2.369356155395508,
+      "epoch": 27.6,
       "learning_rate": 5e-05,
-      "loss": 0.388,
+      "loss": 2.5533,
       "step": 14600
     },
     {
-      "epoch": 27.7882797731569,
-      "grad_norm": 2.701702356338501,
+      "epoch": 27.79,
       "learning_rate": 5e-05,
-      "loss": 0.3943,
+      "loss": 2.5524,
       "step": 14700
     },
     {
-      "epoch": 27.977315689981097,
-      "grad_norm": 2.2755825519561768,
+      "epoch": 27.98,
       "learning_rate": 5e-05,
-      "loss": 0.4015,
+      "loss": 2.2707,
       "step": 14800
     },
     {
       "epoch": 28.0,
-      "eval_accuracy": 0.5592820512820513,
-      "eval_loss": 2.653672218322754,
-      "eval_runtime": 5.3543,
-      "eval_samples_per_second": 93.383,
-      "eval_steps_per_second": 11.766,
+      "eval_accuracy": 0.4955897435897436,
+      "eval_loss": 2.7017431259155273,
+      "eval_runtime": 4.2078,
+      "eval_samples_per_second": 118.828,
+      "eval_steps_per_second": 14.972,
       "step": 14812
     },
     {
       "epoch": 28.0,
-      "eval_exact_match": 17.4,
-      "eval_f1": 24.478888888888893,
+      "eval_exact_match": 0.0,
+      "eval_f1": 0.0059701492537313425,
       "step": 14812
     },
     {
-      "epoch": 28.166351606805293,
-      "grad_norm": 2.0458974838256836,
+      "epoch": 28.17,
       "learning_rate": 5e-05,
-      "loss": 0.3566,
+      "loss": 2.4127,
       "step": 14900
     },
     {
-      "epoch": 28.35538752362949,
-      "grad_norm": 2.040117025375366,
+      "epoch": 28.36,
       "learning_rate": 5e-05,
-      "loss": 0.3602,
+      "loss": 2.3233,
       "step": 15000
     },
     {
-      "epoch": 28.544423440453688,
-      "grad_norm": 1.947757363319397,
+      "epoch": 28.54,
       "learning_rate": 5e-05,
-      "loss": 0.3666,
+      "loss": 2.2381,
       "step": 15100
     },
     {
-      "epoch": 28.733459357277884,
-      "grad_norm": 2.0905520915985107,
+      "epoch": 28.73,
       "learning_rate": 5e-05,
-      "loss": 0.3747,
+      "loss": 2.9648,
       "step": 15200
     },
     {
-      "epoch": 28.92249527410208,
-      "grad_norm": 2.215851306915283,
+      "epoch": 28.92,
       "learning_rate": 5e-05,
-      "loss": 0.3798,
+      "loss": 4.1792,
       "step": 15300
     },
     {
       "epoch": 29.0,
-      "eval_accuracy": 0.5600512820512821,
-      "eval_loss": 2.6304192543029785,
-      "eval_runtime": 5.0359,
-      "eval_samples_per_second": 99.288,
-      "eval_steps_per_second": 12.51,
+      "eval_accuracy": 0.307025641025641,
+      "eval_loss": 5.828761577606201,
+      "eval_runtime": 4.2948,
+      "eval_samples_per_second": 116.419,
+      "eval_steps_per_second": 14.669,
       "step": 15341
     },
     {
       "epoch": 29.0,
-      "eval_exact_match": 16.2,
-      "eval_f1": 23.909062049062054,
+      "eval_exact_match": 0.0,
+      "eval_f1": 0.0,
       "step": 15341
     },
     {
-      "epoch": 29.111531190926275,
-      "grad_norm": 2.5050926208496094,
+      "epoch": 29.11,
       "learning_rate": 5e-05,
-      "loss": 0.3538,
+      "loss": 4.8652,
       "step": 15400
     },
     {
-      "epoch": 29.300567107750474,
-      "grad_norm": 2.0059566497802734,
+      "epoch": 29.3,
       "learning_rate": 5e-05,
-      "loss": 0.3365,
+      "loss": 5.5675,
       "step": 15500
     },
     {
-      "epoch": 29.48960302457467,
-      "grad_norm": 2.7678000926971436,
+      "epoch": 29.49,
       "learning_rate": 5e-05,
-      "loss": 0.3523,
+      "loss": 6.4337,
       "step": 15600
     },
     {
-      "epoch": 29.678638941398866,
-      "grad_norm": 2.6234450340270996,
+      "epoch": 29.68,
       "learning_rate": 5e-05,
-      "loss": 0.3552,
+      "loss": 7.5369,
       "step": 15700
     },
     {
-      "epoch": 29.86767485822306,
-      "grad_norm": 1.9958972930908203,
+      "epoch": 29.87,
       "learning_rate": 5e-05,
-      "loss": 0.3629,
+      "loss": 7.7703,
       "step": 15800
     },
     {
       "epoch": 30.0,
-      "eval_accuracy": 0.56,
-      "eval_loss": 2.6483516693115234,
-      "eval_runtime": 5.0621,
-      "eval_samples_per_second": 98.774,
-      "eval_steps_per_second": 12.446,
+      "eval_accuracy": 0.19215384615384615,
+      "eval_loss": 7.661909103393555,
+      "eval_runtime": 4.4106,
+      "eval_samples_per_second": 113.363,
+      "eval_steps_per_second": 14.284,
       "step": 15870
     },
     {
       "epoch": 30.0,
-      "eval_exact_match": 16.4,
-      "eval_f1": 24.021746031746037,
+      "eval_exact_match": 0.0,
+      "eval_f1": 0.0,
       "step": 15870
     }
   ],
   "logging_steps": 100,
   "max_steps": 26450,
-  "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": 9.776441934112358e+17,
-  "train_batch_size": 1,
+  "total_flos": 2.6803063082419487e+18,
   "trial_name": null,
   "trial_params": null
 }

checkpoint-15870/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f01d18f717251d8791982fc93ed5352f2d247add98e1102e2a52cc0a32128bb7
-size 5304
+oid sha256:391c04887a1ac777804ab56ff52563bcf9c4927713b353f37947aac9be091742
+size 4728