Training in progress, epoch 27, checkpoint

checkpoint-14283/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-14283/adapter_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6e0a8a2319286bf4f2f3d5601f23391e4245f29d35aa8976cbca3ec48bd32667
+size 143269386

checkpoint-14283/added_tokens.json

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

checkpoint-14283/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a270f27895311ee1e6348c9ff9ea2080d4b94f758e7f1c4eae913938384c8daf
-size 224537202
+oid sha256:6b3dcbc2a9c8ff60ddadd734c13ed7187e8eda433bde60ab5a3cbf07f0421bb2
+size 286585234

checkpoint-14283/rng_state_0.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:32791bf743836d95eb258ce04a89820271d883776f43aa14374300ae54f5e065
+oid sha256:733b90f1fef7fd1f9170e3a64484d7f4803363c6f470df931dbcdf1664e23833
 size 15024

checkpoint-14283/rng_state_1.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:1f774f2b68d7957c7bf38eee0ba7d1463c800c8271ee964c044a159b52b28a9c
+oid sha256:46ee7d347a0587b31d910be5e43b53d8670561ebc51510567d151f531e9a82fc
 size 15024

checkpoint-14283/rng_state_2.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:5c49ffae27a223164491b8d03a3f4e3cb372f0931b5df78a3e2586f0f839ab9e
+oid sha256:5d6ebc234b53abf07e6cfd962c346610080d7ba233c08f099ca323aaef595a63
 size 15024

checkpoint-14283/rng_state_3.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:93d27186fa85476b327f8a3654f9b60939e37c7ca5fc0650b5870545341906a4
+oid sha256:fbb2e645e7e31cbb1870a5ded7a5ade77b3f21bd63a76e5e800ee2a0da920b1e
 size 15024

checkpoint-14283/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-14283/tokenizer.model

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

checkpoint-14283/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-14283/trainer_state.json

@@ -9,97 +9,87 @@
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.1890359168241966,
-      "grad_norm": 0.3393945097923279,
+      "epoch": 0.19,
       "learning_rate": 0.0003,
-      "loss": 1.951,
+      "loss": 1.4758,
       "step": 100
     },
     {
-      "epoch": 0.3780718336483932,
-      "grad_norm": 8017.20361328125,
+      "epoch": 0.38,
       "learning_rate": 0.0003,
-      "loss": 2.0285,
+      "loss": 1.3867,
       "step": 200
     },
     {
-      "epoch": 0.5671077504725898,
-      "grad_norm": 9146.7236328125,
+      "epoch": 0.57,
       "learning_rate": 0.0003,
-      "loss": 6.5404,
+      "loss": 1.4064,
       "step": 300
     },
     {
-      "epoch": 0.7561436672967864,
-      "grad_norm": 574.934326171875,
+      "epoch": 0.76,
       "learning_rate": 0.0003,
-      "loss": 7.7357,
+      "loss": 1.3789,
       "step": 400
     },
     {
-      "epoch": 0.945179584120983,
-      "grad_norm": 25128.55859375,
+      "epoch": 0.95,
       "learning_rate": 0.0003,
-      "loss": 6.3771,
+      "loss": 1.3822,
       "step": 500
     },
     {
       "epoch": 1.0,
-      "eval_accuracy": 0.24974358974358973,
-      "eval_loss": 10.19681453704834,
-      "eval_runtime": 5.1408,
-      "eval_samples_per_second": 97.26,
-      "eval_steps_per_second": 12.255,
+      "eval_accuracy": 0.6171794871794872,
+      "eval_loss": 1.2976911067962646,
+      "eval_runtime": 4.224,
+      "eval_samples_per_second": 118.372,
+      "eval_steps_per_second": 14.915,
       "step": 529
     },
     {
       "epoch": 1.0,
-      "eval_exact_match": 0.0,
-      "eval_f1": 0.0,
+      "eval_exact_match": 0.2,
+      "eval_f1": 2.031878046337681,
       "step": 529
     },
     {
-      "epoch": 1.1342155009451795,
-      "grad_norm": 1332.8885498046875,
+      "epoch": 1.13,
       "learning_rate": 0.0003,
-      "loss": 11.0858,
+      "loss": 1.3048,
       "step": 600
     },
     {
-      "epoch": 1.3232514177693762,
-      "grad_norm": 3666.881591796875,
+      "epoch": 1.32,
       "learning_rate": 0.0003,
-      "loss": 9.483,
+      "loss": 1.2524,
       "step": 700
     },
     {
-      "epoch": 1.5122873345935728,
-      "grad_norm": 356847.0,
+      "epoch": 1.51,
       "learning_rate": 0.0003,
-      "loss": 8.069,
+      "loss": 1.3009,
       "step": 800
     },
     {
-      "epoch": 1.7013232514177694,
-      "grad_norm": 1061.877685546875,
+      "epoch": 1.7,
       "learning_rate": 0.0003,
-      "loss": 8.7493,
+      "loss": 1.268,
       "step": 900
     },
     {
-      "epoch": 1.8903591682419658,
-      "grad_norm": 46670.3046875,
+      "epoch": 1.89,
       "learning_rate": 0.0003,
-      "loss": 8.3415,
+      "loss": 1.2744,
       "step": 1000
     },
     {
       "epoch": 2.0,
-      "eval_accuracy": 0.20507692307692307,
-      "eval_loss": 7.504091262817383,
-      "eval_runtime": 5.2437,
-      "eval_samples_per_second": 95.352,
-      "eval_steps_per_second": 12.014,
+      "eval_accuracy": 0.6032307692307692,
+      "eval_loss": 1.374481201171875,
+      "eval_runtime": 4.4141,
+      "eval_samples_per_second": 113.274,
+      "eval_steps_per_second": 14.272,
       "step": 1058
     },
     {
@@ -109,204 +99,183 @@
       "step": 1058
     },
     {
-      "epoch": 2.0793950850661624,
-      "grad_norm": 2.0727381706237793,
+      "epoch": 2.08,
       "learning_rate": 0.0003,
-      "loss": 8.3511,
+      "loss": 1.2158,
       "step": 1100
     },
     {
-      "epoch": 2.268431001890359,
-      "grad_norm": 1.8185893297195435,
+      "epoch": 2.27,
       "learning_rate": 0.0003,
-      "loss": 7.3065,
+      "loss": 1.1066,
       "step": 1200
     },
     {
-      "epoch": 2.4574669187145557,
-      "grad_norm": 1620.93701171875,
+      "epoch": 2.46,
       "learning_rate": 0.0003,
-      "loss": 7.1302,
+      "loss": 1.1365,
       "step": 1300
     },
     {
-      "epoch": 2.6465028355387523,
-      "grad_norm": 1.0691380500793457,
+      "epoch": 2.65,
       "learning_rate": 0.0003,
-      "loss": 7.4423,
+      "loss": 1.1486,
       "step": 1400
     },
     {
-      "epoch": 2.835538752362949,
-      "grad_norm": 1.0605100393295288,
+      "epoch": 2.84,
       "learning_rate": 0.0003,
-      "loss": 7.6715,
+      "loss": 1.1768,
       "step": 1500
     },
     {
       "epoch": 3.0,
-      "eval_accuracy": 0.28507692307692306,
-      "eval_loss": 7.084368705749512,
-      "eval_runtime": 5.493,
-      "eval_samples_per_second": 91.026,
-      "eval_steps_per_second": 11.469,
+      "eval_accuracy": 0.6157435897435898,
+      "eval_loss": 1.331903100013733,
+      "eval_runtime": 4.8466,
+      "eval_samples_per_second": 103.164,
+      "eval_steps_per_second": 12.999,
       "step": 1587
     },
     {
       "epoch": 3.0,
       "eval_exact_match": 0.0,
-      "eval_f1": 0.08,
+      "eval_f1": 0.0,
       "step": 1587
     },
     {
-      "epoch": 3.0245746691871456,
-      "grad_norm": 1967676800.0,
+      "epoch": 3.02,
       "learning_rate": 0.0003,
-      "loss": 7.8617,
+      "loss": 1.1769,
       "step": 1600
     },
     {
-      "epoch": 3.213610586011342,
-      "grad_norm": 83918778793984.0,
+      "epoch": 3.21,
       "learning_rate": 0.0003,
-      "loss": 8.083,
+      "loss": 0.9541,
       "step": 1700
     },
     {
-      "epoch": 3.402646502835539,
-      "grad_norm": 12349449.0,
+      "epoch": 3.4,
       "learning_rate": 0.0003,
-      "loss": 8.0382,
+      "loss": 1.2939,
       "step": 1800
     },
     {
-      "epoch": 3.5916824196597354,
-      "grad_norm": 52939337728.0,
+      "epoch": 3.59,
       "learning_rate": 0.0003,
-      "loss": 7.8761,
+      "loss": 0.9852,
       "step": 1900
     },
     {
-      "epoch": 3.780718336483932,
-      "grad_norm": 1381182.25,
+      "epoch": 3.78,
       "learning_rate": 0.0003,
-      "loss": 7.4665,
+      "loss": 0.9538,
       "step": 2000
     },
     {
-      "epoch": 3.9697542533081287,
-      "grad_norm": 105.80550384521484,
+      "epoch": 3.97,
       "learning_rate": 0.0003,
-      "loss": 7.8182,
+      "loss": 0.9247,
       "step": 2100
     },
     {
       "epoch": 4.0,
-      "eval_accuracy": 0.23933333333333334,
-      "eval_loss": 7.05845308303833,
-      "eval_runtime": 5.0891,
-      "eval_samples_per_second": 98.248,
-      "eval_steps_per_second": 12.379,
+      "eval_accuracy": 0.6101538461538462,
+      "eval_loss": 1.4367107152938843,
+      "eval_runtime": 4.4146,
+      "eval_samples_per_second": 113.259,
+      "eval_steps_per_second": 14.271,
       "step": 2116
     },
     {
       "epoch": 4.0,
-      "eval_exact_match": 0.0,
-      "eval_f1": 0.0,
+      "eval_exact_match": 6.6,
+      "eval_f1": 9.897142857142855,
       "step": 2116
     },
     {
-      "epoch": 4.158790170132325,
-      "grad_norm": 16.110862731933594,
+      "epoch": 4.16,
       "learning_rate": 0.0003,
-      "loss": 7.9728,
+      "loss": 1.1334,
       "step": 2200
     },
     {
-      "epoch": 4.3478260869565215,
-      "grad_norm": 4.362427234649658,
+      "epoch": 4.35,
       "learning_rate": 0.0003,
-      "loss": 7.2875,
+      "loss": 1.5175,
       "step": 2300
     },
     {
-      "epoch": 4.536862003780718,
-      "grad_norm": 18039.095703125,
+      "epoch": 4.54,
       "learning_rate": 0.0003,
-      "loss": 7.3972,
+      "loss": 4.2745,
       "step": 2400
     },
     {
-      "epoch": 4.725897920604915,
-      "grad_norm": 2.3812594051160474e+17,
+      "epoch": 4.73,
       "learning_rate": 0.0003,
-      "loss": 7.4631,
+      "loss": 2.5667,
       "step": 2500
     },
     {
-      "epoch": 4.914933837429111,
-      "grad_norm": 3.9549436569213867,
+      "epoch": 4.91,
       "learning_rate": 0.0003,
-      "loss": 7.3322,
+      "loss": 1.1836,
       "step": 2600
     },
     {
       "epoch": 5.0,
-      "eval_accuracy": 0.22430769230769232,
-      "eval_loss": 6.72421932220459,
-      "eval_runtime": 5.1689,
-      "eval_samples_per_second": 96.731,
-      "eval_steps_per_second": 12.188,
+      "eval_accuracy": 0.556923076923077,
+      "eval_loss": 1.9167817831039429,
+      "eval_runtime": 4.8432,
+      "eval_samples_per_second": 103.238,
+      "eval_steps_per_second": 13.008,
       "step": 2645
     },
     {
       "epoch": 5.0,
       "eval_exact_match": 0.0,
-      "eval_f1": 0.0,
+      "eval_f1": 0.2264729350912329,
       "step": 2645
     },
     {
-      "epoch": 5.103969754253308,
-      "grad_norm": 120114.2421875,
+      "epoch": 5.1,
       "learning_rate": 0.0003,
-      "loss": 7.2428,
+      "loss": 1.7226,
       "step": 2700
     },
     {
-      "epoch": 5.293005671077505,
-      "grad_norm": 391.66119384765625,
+      "epoch": 5.29,
       "learning_rate": 0.0003,
-      "loss": 7.2495,
+      "loss": 2.1466,
       "step": 2800
     },
     {
-      "epoch": 5.482041587901701,
-      "grad_norm": 599371136.0,
+      "epoch": 5.48,
       "learning_rate": 0.0003,
-      "loss": 7.5915,
+      "loss": 1.83,
       "step": 2900
     },
     {
-      "epoch": 5.671077504725898,
-      "grad_norm": 5585.97314453125,
+      "epoch": 5.67,
       "learning_rate": 0.0003,
-      "loss": 7.3014,
+      "loss": 1.8789,
       "step": 3000
     },
     {
-      "epoch": 5.8601134215500945,
-      "grad_norm": 268239548448768.0,
+      "epoch": 5.86,
       "learning_rate": 0.0003,
-      "loss": 7.259,
+      "loss": 2.035,
       "step": 3100
     },
     {
       "epoch": 6.0,
-      "eval_accuracy": 0.24984615384615386,
-      "eval_loss": 6.531904220581055,
-      "eval_runtime": 5.0344,
-      "eval_samples_per_second": 99.317,
-      "eval_steps_per_second": 12.514,
+      "eval_accuracy": 0.5377435897435897,
+      "eval_loss": 2.0794410705566406,
+      "eval_runtime": 4.1284,
+      "eval_samples_per_second": 121.112,
+      "eval_steps_per_second": 15.26,
       "step": 3174
     },
     {
@@ -316,54 +285,48 @@
       "step": 3174
     },
     {
-      "epoch": 6.049149338374291,
-      "grad_norm": 21535.734375,
+      "epoch": 6.05,
       "learning_rate": 0.0003,
-      "loss": 7.5829,
+      "loss": 2.2251,
       "step": 3200
     },
     {
-      "epoch": 6.238185255198488,
-      "grad_norm": 40871530496.0,
+      "epoch": 6.24,
       "learning_rate": 0.0003,
-      "loss": 8.6364,
+      "loss": 2.1457,
       "step": 3300
     },
     {
-      "epoch": 6.427221172022684,
-      "grad_norm": 13.242510795593262,
+      "epoch": 6.43,
       "learning_rate": 0.0003,
-      "loss": 7.6599,
+      "loss": 2.0764,
       "step": 3400
     },
     {
-      "epoch": 6.616257088846881,
-      "grad_norm": 3.0841150283813477,
+      "epoch": 6.62,
       "learning_rate": 0.0003,
-      "loss": 7.5796,
+      "loss": 2.3695,
       "step": 3500
     },
     {
-      "epoch": 6.805293005671078,
-      "grad_norm": 1.6124571561813354,
+      "epoch": 6.81,
       "learning_rate": 0.0003,
-      "loss": 7.3499,
+      "loss": 2.2254,
       "step": 3600
     },
     {
-      "epoch": 6.994328922495274,
-      "grad_norm": 1.8358014822006226,
+      "epoch": 6.99,
       "learning_rate": 0.0003,
-      "loss": 7.2206,
+      "loss": 3.7483,
       "step": 3700
     },
     {
       "epoch": 7.0,
-      "eval_accuracy": 0.22430769230769232,
-      "eval_loss": 6.724715709686279,
-      "eval_runtime": 5.0476,
-      "eval_samples_per_second": 99.056,
-      "eval_steps_per_second": 12.481,
+      "eval_accuracy": 0.48805128205128206,
+      "eval_loss": 2.6722612380981445,
+      "eval_runtime": 4.2095,
+      "eval_samples_per_second": 118.778,
+      "eval_steps_per_second": 14.966,
       "step": 3703
     },
     {
@@ -373,47 +336,42 @@
       "step": 3703
     },
     {
-      "epoch": 7.183364839319471,
-      "grad_norm": 5.786651134490967,
+      "epoch": 7.18,
       "learning_rate": 0.0003,
-      "loss": 7.2316,
+      "loss": 6.5533,
       "step": 3800
     },
     {
-      "epoch": 7.3724007561436675,
-      "grad_norm": 3.7182490825653076,
+      "epoch": 7.37,
       "learning_rate": 0.0003,
-      "loss": 7.2591,
+      "loss": 6.708,
       "step": 3900
     },
     {
-      "epoch": 7.561436672967864,
-      "grad_norm": 31.241104125976562,
+      "epoch": 7.56,
       "learning_rate": 0.0003,
-      "loss": 7.2557,
+      "loss": 7.4301,
       "step": 4000
     },
     {
-      "epoch": 7.750472589792061,
-      "grad_norm": 896688896.0,
+      "epoch": 7.75,
       "learning_rate": 0.0003,
-      "loss": 8.4436,
+      "loss": 8.6731,
       "step": 4100
     },
     {
-      "epoch": 7.939508506616257,
-      "grad_norm": 1.1150538843415904e+19,
+      "epoch": 7.94,
       "learning_rate": 0.0003,
-      "loss": 8.7572,
+      "loss": 7.127,
       "step": 4200
     },
     {
       "epoch": 8.0,
-      "eval_accuracy": 0.26184615384615384,
-      "eval_loss": 6.479637622833252,
-      "eval_runtime": 5.1473,
-      "eval_samples_per_second": 97.138,
-      "eval_steps_per_second": 12.239,
+      "eval_accuracy": 0.1922051282051282,
+      "eval_loss": 7.040984153747559,
+      "eval_runtime": 4.8262,
+      "eval_samples_per_second": 103.6,
+      "eval_steps_per_second": 13.054,
       "step": 4232
     },
     {
@@ -423,47 +381,42 @@
       "step": 4232
     },
     {
-      "epoch": 8.128544423440454,
-      "grad_norm": 7.2513813972473145,
+      "epoch": 8.13,
       "learning_rate": 0.0003,
-      "loss": 8.1381,
+      "loss": 7.8796,
       "step": 4300
     },
     {
-      "epoch": 8.31758034026465,
-      "grad_norm": 14.863628387451172,
+      "epoch": 8.32,
       "learning_rate": 0.0003,
-      "loss": 7.3335,
+      "loss": 8.0716,
       "step": 4400
     },
     {
-      "epoch": 8.506616257088847,
-      "grad_norm": 5.358551979064941,
+      "epoch": 8.51,
       "learning_rate": 0.0003,
-      "loss": 8.3921,
+      "loss": 8.2395,
       "step": 4500
     },
     {
-      "epoch": 8.695652173913043,
-      "grad_norm": 113.71702575683594,
+      "epoch": 8.7,
       "learning_rate": 0.0003,
-      "loss": 7.221,
+      "loss": 7.9157,
       "step": 4600
     },
     {
-      "epoch": 8.88468809073724,
-      "grad_norm": 2.3997488021850586,
+      "epoch": 8.88,
       "learning_rate": 0.0003,
-      "loss": 8.4835,
+      "loss": 7.5321,
       "step": 4700
     },
     {
       "epoch": 9.0,
-      "eval_accuracy": 0.26184615384615384,
-      "eval_loss": 6.614655017852783,
-      "eval_runtime": 5.026,
-      "eval_samples_per_second": 99.483,
-      "eval_steps_per_second": 12.535,
+      "eval_accuracy": 0.19405128205128205,
+      "eval_loss": 6.64878511428833,
+      "eval_runtime": 4.104,
+      "eval_samples_per_second": 121.831,
+      "eval_steps_per_second": 15.351,
       "step": 4761
     },
     {
@@ -473,47 +426,42 @@
       "step": 4761
     },
     {
-      "epoch": 9.073724007561436,
-      "grad_norm": 7.822358131408691,
+      "epoch": 9.07,
       "learning_rate": 0.0003,
-      "loss": 7.2836,
+      "loss": 7.2432,
       "step": 4800
     },
     {
-      "epoch": 9.262759924385634,
-      "grad_norm": 89279359352832.0,
+      "epoch": 9.26,
       "learning_rate": 0.0003,
-      "loss": 7.2712,
+      "loss": 7.4268,
       "step": 4900
     },
     {
-      "epoch": 9.45179584120983,
-      "grad_norm": 656.6845703125,
+      "epoch": 9.45,
       "learning_rate": 0.0003,
-      "loss": 7.2623,
+      "loss": 7.7517,
       "step": 5000
     },
     {
-      "epoch": 9.640831758034027,
-      "grad_norm": 13841700.0,
+      "epoch": 9.64,
       "learning_rate": 0.0003,
-      "loss": 7.4981,
+      "loss": 7.4674,
       "step": 5100
     },
     {
-      "epoch": 9.829867674858223,
-      "grad_norm": 768177.9375,
+      "epoch": 9.83,
       "learning_rate": 0.0003,
-      "loss": 7.6044,
+      "loss": 7.3806,
       "step": 5200
     },
     {
       "epoch": 10.0,
-      "eval_accuracy": 0.2508717948717949,
-      "eval_loss": 8.167869567871094,
-      "eval_runtime": 5.1342,
-      "eval_samples_per_second": 97.387,
-      "eval_steps_per_second": 12.271,
+      "eval_accuracy": 0.2196923076923077,
+      "eval_loss": 6.842738151550293,
+      "eval_runtime": 4.8429,
+      "eval_samples_per_second": 103.244,
+      "eval_steps_per_second": 13.009,
       "step": 5290
     },
     {
@@ -523,54 +471,48 @@
       "step": 5290
     },
     {
-      "epoch": 10.01890359168242,
-      "grad_norm": 389293.03125,
+      "epoch": 10.02,
       "learning_rate": 0.0003,
-      "loss": 7.9142,
+      "loss": 7.4491,
       "step": 5300
     },
     {
-      "epoch": 10.207939508506616,
-      "grad_norm": 3129429.25,
+      "epoch": 10.21,
       "learning_rate": 0.0003,
-      "loss": 8.998,
+      "loss": 7.4851,
       "step": 5400
     },
     {
-      "epoch": 10.396975425330814,
-      "grad_norm": 6832450.5,
+      "epoch": 10.4,
       "learning_rate": 0.0003,
-      "loss": 8.9548,
+      "loss": 7.5747,
       "step": 5500
     },
     {
-      "epoch": 10.58601134215501,
-      "grad_norm": 24994486.0,
+      "epoch": 10.59,
       "learning_rate": 0.0003,
-      "loss": 9.3085,
+      "loss": 7.7881,
       "step": 5600
     },
     {
-      "epoch": 10.775047258979207,
-      "grad_norm": 10634.3076171875,
+      "epoch": 10.78,
       "learning_rate": 0.0003,
-      "loss": 9.3052,
+      "loss": 7.8095,
       "step": 5700
     },
     {
-      "epoch": 10.964083175803403,
-      "grad_norm": 37.100101470947266,
+      "epoch": 10.96,
       "learning_rate": 0.0003,
-      "loss": 10.0839,
+      "loss": 7.8159,
       "step": 5800
     },
     {
       "epoch": 11.0,
-      "eval_accuracy": 0.2362051282051282,
-      "eval_loss": 6.820600509643555,
-      "eval_runtime": 5.0262,
-      "eval_samples_per_second": 99.478,
-      "eval_steps_per_second": 12.534,
+      "eval_accuracy": 0.2196923076923077,
+      "eval_loss": 6.883613586425781,
+      "eval_runtime": 4.1075,
+      "eval_samples_per_second": 121.727,
+      "eval_steps_per_second": 15.338,
       "step": 5819
     },
     {
@@ -580,47 +522,42 @@
       "step": 5819
     },
     {
-      "epoch": 11.1531190926276,
-      "grad_norm": 319.2048645019531,
+      "epoch": 11.15,
       "learning_rate": 0.0003,
-      "loss": 7.4258,
+      "loss": 7.7222,
       "step": 5900
     },
     {
-      "epoch": 11.342155009451796,
-      "grad_norm": 72.28472900390625,
+      "epoch": 11.34,
       "learning_rate": 0.0003,
-      "loss": 7.2609,
+      "loss": 7.6203,
       "step": 6000
     },
     {
-      "epoch": 11.531190926275993,
-      "grad_norm": 11.851283073425293,
+      "epoch": 11.53,
       "learning_rate": 0.0003,
-      "loss": 7.5323,
+      "loss": 7.8391,
       "step": 6100
     },
     {
-      "epoch": 11.720226843100189,
-      "grad_norm": 6.574771404266357,
+      "epoch": 11.72,
       "learning_rate": 0.0003,
-      "loss": 7.3028,
+      "loss": 7.8888,
       "step": 6200
     },
     {
-      "epoch": 11.909262759924385,
-      "grad_norm": 4.578986644744873,
+      "epoch": 11.91,
       "learning_rate": 0.0003,
-      "loss": 7.2957,
+      "loss": 7.975,
       "step": 6300
     },
     {
       "epoch": 12.0,
-      "eval_accuracy": 0.26184615384615384,
-      "eval_loss": 6.52316427230835,
-      "eval_runtime": 5.0329,
-      "eval_samples_per_second": 99.346,
-      "eval_steps_per_second": 12.518,
+      "eval_accuracy": 0.2196923076923077,
+      "eval_loss": 6.876262187957764,
+      "eval_runtime": 4.4184,
+      "eval_samples_per_second": 113.162,
+      "eval_steps_per_second": 14.258,
       "step": 6348
     },
     {
@@ -630,47 +567,42 @@
       "step": 6348
     },
     {
-      "epoch": 12.098298676748582,
-      "grad_norm": 21.886821746826172,
+      "epoch": 12.1,
       "learning_rate": 0.0003,
-      "loss": 7.189,
+      "loss": 7.9981,
       "step": 6400
     },
     {
-      "epoch": 12.287334593572778,
-      "grad_norm": 102.8978500366211,
+      "epoch": 12.29,
       "learning_rate": 0.0003,
-      "loss": 7.2076,
+      "loss": 7.9944,
       "step": 6500
     },
     {
-      "epoch": 12.476370510396976,
-      "grad_norm": 4.47935676574707,
+      "epoch": 12.48,
       "learning_rate": 0.0003,
-      "loss": 7.2369,
+      "loss": 8.0191,
       "step": 6600
     },
     {
-      "epoch": 12.665406427221171,
-      "grad_norm": 259.2627868652344,
+      "epoch": 12.67,
       "learning_rate": 0.0003,
-      "loss": 7.3119,
+      "loss": 8.0091,
       "step": 6700
     },
     {
-      "epoch": 12.854442344045369,
-      "grad_norm": 473.829345703125,
+      "epoch": 12.85,
       "learning_rate": 0.0003,
-      "loss": 7.3484,
+      "loss": 7.9902,
       "step": 6800
     },
     {
       "epoch": 13.0,
-      "eval_accuracy": 0.26184615384615384,
-      "eval_loss": 6.655651569366455,
-      "eval_runtime": 5.7482,
-      "eval_samples_per_second": 86.984,
-      "eval_steps_per_second": 10.96,
+      "eval_accuracy": 0.2196923076923077,
+      "eval_loss": 6.872635841369629,
+      "eval_runtime": 4.943,
+      "eval_samples_per_second": 101.152,
+      "eval_steps_per_second": 12.745,
       "step": 6877
     },
     {
@@ -680,54 +612,48 @@
       "step": 6877
     },
     {
-      "epoch": 13.043478260869565,
-      "grad_norm": 150.06381225585938,
+      "epoch": 13.04,
       "learning_rate": 0.0003,
-      "loss": 7.2438,
+      "loss": 8.0141,
       "step": 6900
     },
     {
-      "epoch": 13.232514177693762,
-      "grad_norm": 1546.5465087890625,
+      "epoch": 13.23,
       "learning_rate": 0.0003,
-      "loss": 7.2503,
+      "loss": 8.0041,
       "step": 7000
     },
     {
-      "epoch": 13.421550094517958,
-      "grad_norm": 18108432.0,
+      "epoch": 13.42,
       "learning_rate": 0.0003,
-      "loss": 7.5797,
+      "loss": 7.9641,
       "step": 7100
     },
     {
-      "epoch": 13.610586011342155,
-      "grad_norm": 379725545472.0,
+      "epoch": 13.61,
       "learning_rate": 0.0003,
-      "loss": 8.6953,
+      "loss": 7.9734,
       "step": 7200
     },
     {
-      "epoch": 13.799621928166351,
-      "grad_norm": 14.351493835449219,
+      "epoch": 13.8,
       "learning_rate": 0.0003,
-      "loss": 7.5947,
+      "loss": 7.9893,
       "step": 7300
     },
     {
-      "epoch": 13.988657844990549,
-      "grad_norm": 3.4325509071350098,
+      "epoch": 13.99,
       "learning_rate": 0.0003,
-      "loss": 8.2657,
+      "loss": 7.8585,
       "step": 7400
     },
     {
       "epoch": 14.0,
-      "eval_accuracy": 0.2362051282051282,
-      "eval_loss": 6.780774116516113,
-      "eval_runtime": 5.7058,
-      "eval_samples_per_second": 87.629,
-      "eval_steps_per_second": 11.041,
+      "eval_accuracy": 0.2194871794871795,
+      "eval_loss": 6.823603630065918,
+      "eval_runtime": 4.416,
+      "eval_samples_per_second": 113.224,
+      "eval_steps_per_second": 14.266,
       "step": 7406
     },
     {
@@ -737,47 +663,42 @@
       "step": 7406
     },
     {
-      "epoch": 14.177693761814744,
-      "grad_norm": 3408474406912.0,
+      "epoch": 14.18,
       "learning_rate": 0.0003,
-      "loss": 7.5487,
+      "loss": 7.3595,
       "step": 7500
     },
     {
-      "epoch": 14.366729678638942,
-      "grad_norm": 766890.1875,
+      "epoch": 14.37,
       "learning_rate": 0.0003,
-      "loss": 7.4765,
+      "loss": 7.3486,
       "step": 7600
     },
     {
-      "epoch": 14.555765595463138,
-      "grad_norm": 4749101.0,
+      "epoch": 14.56,
       "learning_rate": 0.0003,
-      "loss": 7.4927,
+      "loss": 7.335,
       "step": 7700
     },
     {
-      "epoch": 14.744801512287335,
-      "grad_norm": 21509612.0,
+      "epoch": 14.74,
       "learning_rate": 0.0003,
-      "loss": 7.4864,
+      "loss": 7.3724,
       "step": 7800
     },
     {
-      "epoch": 14.93383742911153,
-      "grad_norm": 1.0536716136244838e+18,
+      "epoch": 14.93,
       "learning_rate": 0.0003,
-      "loss": 7.4391,
+      "loss": 7.3449,
       "step": 7900
     },
     {
       "epoch": 15.0,
-      "eval_accuracy": 0.1995897435897436,
-      "eval_loss": NaN,
-      "eval_runtime": 5.788,
-      "eval_samples_per_second": 86.386,
-      "eval_steps_per_second": 10.885,
+      "eval_accuracy": 0.1922051282051282,
+      "eval_loss": 7.199692249298096,
+      "eval_runtime": 4.4181,
+      "eval_samples_per_second": 113.17,
+      "eval_steps_per_second": 14.259,
       "step": 7935
     },
     {
@@ -787,47 +708,42 @@
       "step": 7935
     },
     {
-      "epoch": 15.122873345935728,
-      "grad_norm": NaN,
+      "epoch": 15.12,
       "learning_rate": 0.0003,
-      "loss": 602.3478,
+      "loss": 7.3378,
       "step": 8000
     },
     {
-      "epoch": 15.311909262759924,
-      "grad_norm": NaN,
+      "epoch": 15.31,
       "learning_rate": 0.0003,
-      "loss": 0.0,
+      "loss": 7.3525,
       "step": 8100
     },
     {
-      "epoch": 15.500945179584122,
-      "grad_norm": NaN,
+      "epoch": 15.5,
       "learning_rate": 0.0003,
-      "loss": 0.0,
+      "loss": 7.3645,
       "step": 8200
     },
     {
-      "epoch": 15.689981096408317,
-      "grad_norm": NaN,
+      "epoch": 15.69,
       "learning_rate": 0.0003,
-      "loss": 0.0,
+      "loss": 7.3455,
       "step": 8300
     },
     {
-      "epoch": 15.879017013232515,
-      "grad_norm": NaN,
+      "epoch": 15.88,
       "learning_rate": 0.0003,
-      "loss": 0.0,
+      "loss": 7.3133,
       "step": 8400
     },
     {
       "epoch": 16.0,
-      "eval_accuracy": 0.1995897435897436,
-      "eval_loss": NaN,
-      "eval_runtime": 5.1552,
-      "eval_samples_per_second": 96.99,
-      "eval_steps_per_second": 12.221,
+      "eval_accuracy": 0.18692307692307691,
+      "eval_loss": 6.745450496673584,
+      "eval_runtime": 4.218,
+      "eval_samples_per_second": 118.541,
+      "eval_steps_per_second": 14.936,
       "step": 8464
     },
     {
@@ -837,47 +753,42 @@
       "step": 8464
     },
     {
-      "epoch": 16.068052930056712,
-      "grad_norm": NaN,
+      "epoch": 16.07,
       "learning_rate": 0.0003,
-      "loss": 0.0,
+      "loss": 7.3153,
       "step": 8500
     },
     {
-      "epoch": 16.257088846880908,
-      "grad_norm": NaN,
+      "epoch": 16.26,
       "learning_rate": 0.0003,
-      "loss": 0.0,
+      "loss": 7.3123,
       "step": 8600
     },
     {
-      "epoch": 16.446124763705104,
-      "grad_norm": NaN,
+      "epoch": 16.45,
       "learning_rate": 0.0003,
-      "loss": 0.0,
+      "loss": 7.3081,
       "step": 8700
     },
     {
-      "epoch": 16.6351606805293,
-      "grad_norm": NaN,
+      "epoch": 16.64,
       "learning_rate": 0.0003,
-      "loss": 0.0,
+      "loss": 7.3069,
       "step": 8800
     },
     {
-      "epoch": 16.8241965973535,
-      "grad_norm": NaN,
+      "epoch": 16.82,
       "learning_rate": 0.0003,
-      "loss": 0.0,
+      "loss": 7.305,
       "step": 8900
     },
     {
       "epoch": 17.0,
-      "eval_accuracy": 0.1995897435897436,
-      "eval_loss": NaN,
-      "eval_runtime": 5.2516,
-      "eval_samples_per_second": 95.209,
-      "eval_steps_per_second": 11.996,
+      "eval_accuracy": 0.18692307692307691,
+      "eval_loss": 6.745449066162109,
+      "eval_runtime": 4.2111,
+      "eval_samples_per_second": 118.735,
+      "eval_steps_per_second": 14.961,
       "step": 8993
     },
     {
@@ -887,54 +798,48 @@
       "step": 8993
     },
     {
-      "epoch": 17.013232514177695,
-      "grad_norm": NaN,
+      "epoch": 17.01,
       "learning_rate": 0.0003,
-      "loss": 0.0,
+      "loss": 7.3114,
       "step": 9000
     },
     {
-      "epoch": 17.20226843100189,
-      "grad_norm": NaN,
+      "epoch": 17.2,
       "learning_rate": 0.0003,
-      "loss": 0.0,
+      "loss": 7.3226,
       "step": 9100
     },
     {
-      "epoch": 17.391304347826086,
-      "grad_norm": NaN,
+      "epoch": 17.39,
       "learning_rate": 0.0003,
-      "loss": 0.0,
+      "loss": 7.3234,
       "step": 9200
     },
     {
-      "epoch": 17.58034026465028,
-      "grad_norm": NaN,
+      "epoch": 17.58,
       "learning_rate": 0.0003,
-      "loss": 0.0,
+      "loss": 7.3929,
       "step": 9300
     },
     {
-      "epoch": 17.76937618147448,
-      "grad_norm": NaN,
+      "epoch": 17.77,
       "learning_rate": 0.0003,
-      "loss": 0.0,
+      "loss": 7.5523,
       "step": 9400
     },
     {
-      "epoch": 17.958412098298677,
-      "grad_norm": NaN,
+      "epoch": 17.96,
       "learning_rate": 0.0003,
-      "loss": 0.0,
+      "loss": 7.7463,
       "step": 9500
     },
     {
       "epoch": 18.0,
-      "eval_accuracy": 0.1995897435897436,
-      "eval_loss": NaN,
-      "eval_runtime": 5.7505,
-      "eval_samples_per_second": 86.949,
-      "eval_steps_per_second": 10.956,
+      "eval_accuracy": 0.18697435897435896,
+      "eval_loss": 8.831918716430664,
+      "eval_runtime": 4.1023,
+      "eval_samples_per_second": 121.881,
+      "eval_steps_per_second": 15.357,
       "step": 9522
     },
     {
@@ -944,47 +849,42 @@
       "step": 9522
     },
     {
-      "epoch": 18.147448015122873,
-      "grad_norm": NaN,
+      "epoch": 18.15,
       "learning_rate": 0.0003,
-      "loss": 0.0,
+      "loss": 9.7591,
       "step": 9600
     },
     {
-      "epoch": 18.33648393194707,
-      "grad_norm": NaN,
+      "epoch": 18.34,
       "learning_rate": 0.0003,
-      "loss": 0.0,
+      "loss": 10.4962,
       "step": 9700
     },
     {
-      "epoch": 18.525519848771268,
-      "grad_norm": NaN,
+      "epoch": 18.53,
       "learning_rate": 0.0003,
-      "loss": 0.0,
+      "loss": 9.9571,
       "step": 9800
     },
     {
-      "epoch": 18.714555765595463,
-      "grad_norm": NaN,
+      "epoch": 18.71,
       "learning_rate": 0.0003,
-      "loss": 0.0,
+      "loss": 9.9661,
       "step": 9900
     },
     {
-      "epoch": 18.90359168241966,
-      "grad_norm": NaN,
+      "epoch": 18.9,
       "learning_rate": 0.0003,
-      "loss": 0.0,
+      "loss": 9.9696,
       "step": 10000
     },
     {
       "epoch": 19.0,
-      "eval_accuracy": 0.1995897435897436,
-      "eval_loss": NaN,
-      "eval_runtime": 5.3461,
-      "eval_samples_per_second": 93.526,
-      "eval_steps_per_second": 11.784,
+      "eval_accuracy": 0.16923076923076924,
+      "eval_loss": 10.070154190063477,
+      "eval_runtime": 4.1088,
+      "eval_samples_per_second": 121.69,
+      "eval_steps_per_second": 15.333,
       "step": 10051
     },
     {
@@ -994,47 +894,42 @@
       "step": 10051
     },
     {
-      "epoch": 19.092627599243855,
-      "grad_norm": NaN,
+      "epoch": 19.09,
       "learning_rate": 0.0003,
-      "loss": 0.0,
+      "loss": 9.9832,
       "step": 10100
     },
     {
-      "epoch": 19.281663516068054,
-      "grad_norm": NaN,
+      "epoch": 19.28,
       "learning_rate": 0.0003,
-      "loss": 0.0,
+      "loss": 9.9764,
       "step": 10200
     },
     {
-      "epoch": 19.47069943289225,
-      "grad_norm": NaN,
+      "epoch": 19.47,
       "learning_rate": 0.0003,
-      "loss": 0.0,
+      "loss": 9.967,
       "step": 10300
     },
     {
-      "epoch": 19.659735349716446,
-      "grad_norm": NaN,
+      "epoch": 19.66,
       "learning_rate": 0.0003,
-      "loss": 0.0,
+      "loss": 9.9739,
       "step": 10400
     },
     {
-      "epoch": 19.84877126654064,
-      "grad_norm": NaN,
+      "epoch": 19.85,
       "learning_rate": 0.0003,
-      "loss": 0.0,
+      "loss": 9.9845,
       "step": 10500
     },
     {
       "epoch": 20.0,
-      "eval_accuracy": 0.1995897435897436,
-      "eval_loss": NaN,
-      "eval_runtime": 5.0526,
-      "eval_samples_per_second": 98.959,
-      "eval_steps_per_second": 12.469,
+      "eval_accuracy": 0.16923076923076924,
+      "eval_loss": 10.070154190063477,
+      "eval_runtime": 4.4134,
+      "eval_samples_per_second": 113.292,
+      "eval_steps_per_second": 14.275,
       "step": 10580
     },
     {
@@ -1044,54 +939,48 @@
       "step": 10580
     },
     {
-      "epoch": 20.03780718336484,
-      "grad_norm": NaN,
+      "epoch": 20.04,
       "learning_rate": 0.0003,
-      "loss": 0.0,
+      "loss": 9.9691,
       "step": 10600
     },
     {
-      "epoch": 20.226843100189036,
-      "grad_norm": NaN,
+      "epoch": 20.23,
       "learning_rate": 0.0003,
-      "loss": 0.0,
+      "loss": 9.9408,
       "step": 10700
     },
     {
-      "epoch": 20.415879017013232,
-      "grad_norm": NaN,
+      "epoch": 20.42,
       "learning_rate": 0.0003,
-      "loss": 0.0,
+      "loss": 9.9893,
       "step": 10800
     },
     {
-      "epoch": 20.604914933837428,
-      "grad_norm": NaN,
+      "epoch": 20.6,
       "learning_rate": 0.0003,
-      "loss": 0.0,
+      "loss": 9.9797,
       "step": 10900
     },
     {
-      "epoch": 20.793950850661627,
-      "grad_norm": NaN,
+      "epoch": 20.79,
       "learning_rate": 0.0003,
-      "loss": 0.0,
+      "loss": 9.9908,
       "step": 11000
     },
     {
-      "epoch": 20.982986767485823,
-      "grad_norm": NaN,
+      "epoch": 20.98,
       "learning_rate": 0.0003,
-      "loss": 0.0,
+      "loss": 9.9502,
       "step": 11100
     },
     {
       "epoch": 21.0,
-      "eval_accuracy": 0.1995897435897436,
-      "eval_loss": NaN,
-      "eval_runtime": 5.7608,
-      "eval_samples_per_second": 86.793,
-      "eval_steps_per_second": 10.936,
+      "eval_accuracy": 0.16923076923076924,
+      "eval_loss": 10.070154190063477,
+      "eval_runtime": 4.2664,
+      "eval_samples_per_second": 117.194,
+      "eval_steps_per_second": 14.766,
       "step": 11109
     },
     {
@@ -1101,47 +990,42 @@
       "step": 11109
     },
     {
-      "epoch": 21.17202268431002,
-      "grad_norm": NaN,
+      "epoch": 21.17,
       "learning_rate": 0.0003,
-      "loss": 0.0,
+      "loss": 9.9688,
       "step": 11200
     },
     {
-      "epoch": 21.361058601134214,
-      "grad_norm": NaN,
+      "epoch": 21.36,
       "learning_rate": 0.0003,
-      "loss": 0.0,
+      "loss": 9.9395,
       "step": 11300
     },
     {
-      "epoch": 21.550094517958414,
-      "grad_norm": NaN,
+      "epoch": 21.55,
       "learning_rate": 0.0003,
-      "loss": 0.0,
+      "loss": 9.9855,
       "step": 11400
     },
     {
-      "epoch": 21.73913043478261,
-      "grad_norm": NaN,
+      "epoch": 21.74,
       "learning_rate": 0.0003,
-      "loss": 0.0,
+      "loss": 9.9895,
       "step": 11500
     },
     {
-      "epoch": 21.928166351606805,
-      "grad_norm": NaN,
+      "epoch": 21.93,
       "learning_rate": 0.0003,
-      "loss": 0.0,
+      "loss": 9.9726,
       "step": 11600
     },
     {
       "epoch": 22.0,
-      "eval_accuracy": 0.1995897435897436,
-      "eval_loss": NaN,
-      "eval_runtime": 5.0399,
-      "eval_samples_per_second": 99.209,
-      "eval_steps_per_second": 12.5,
+      "eval_accuracy": 0.16923076923076924,
+      "eval_loss": 10.070154190063477,
+      "eval_runtime": 4.1678,
+      "eval_samples_per_second": 119.966,
+      "eval_steps_per_second": 15.116,
       "step": 11638
     },
     {
@@ -1151,47 +1035,42 @@
       "step": 11638
     },
     {
-      "epoch": 22.117202268431,
-      "grad_norm": NaN,
+      "epoch": 22.12,
       "learning_rate": 0.0003,
-      "loss": 0.0,
+      "loss": 9.9604,
       "step": 11700
     },
     {
-      "epoch": 22.3062381852552,
-      "grad_norm": NaN,
+      "epoch": 22.31,
       "learning_rate": 0.0003,
-      "loss": 0.0,
+      "loss": 9.9835,
       "step": 11800
     },
     {
-      "epoch": 22.495274102079396,
-      "grad_norm": NaN,
+      "epoch": 22.5,
       "learning_rate": 0.0003,
-      "loss": 0.0,
+      "loss": 9.9707,
       "step": 11900
     },
     {
-      "epoch": 22.68431001890359,
-      "grad_norm": NaN,
+      "epoch": 22.68,
       "learning_rate": 0.0003,
-      "loss": 0.0,
+      "loss": 9.9732,
       "step": 12000
     },
     {
-      "epoch": 22.873345935727787,
-      "grad_norm": NaN,
+      "epoch": 22.87,
       "learning_rate": 0.0003,
-      "loss": 0.0,
+      "loss": 9.9648,
       "step": 12100
     },
     {
       "epoch": 23.0,
-      "eval_accuracy": 0.1995897435897436,
-      "eval_loss": NaN,
-      "eval_runtime": 5.2777,
-      "eval_samples_per_second": 94.738,
-      "eval_steps_per_second": 11.937,
+      "eval_accuracy": 0.16923076923076924,
+      "eval_loss": 10.070154190063477,
+      "eval_runtime": 4.8312,
+      "eval_samples_per_second": 103.494,
+      "eval_steps_per_second": 13.04,
       "step": 12167
     },
     {
@@ -1201,47 +1080,42 @@
       "step": 12167
     },
     {
-      "epoch": 23.062381852551987,
-      "grad_norm": NaN,
+      "epoch": 23.06,
       "learning_rate": 0.0003,
-      "loss": 0.0,
+      "loss": 9.9607,
       "step": 12200
     },
     {
-      "epoch": 23.251417769376182,
-      "grad_norm": NaN,
+      "epoch": 23.25,
       "learning_rate": 0.0003,
-      "loss": 0.0,
+      "loss": 10.0081,
       "step": 12300
     },
     {
-      "epoch": 23.440453686200378,
-      "grad_norm": NaN,
+      "epoch": 23.44,
       "learning_rate": 0.0003,
-      "loss": 0.0,
+      "loss": 9.9695,
       "step": 12400
     },
     {
-      "epoch": 23.629489603024574,
-      "grad_norm": NaN,
+      "epoch": 23.63,
       "learning_rate": 0.0003,
-      "loss": 0.0,
+      "loss": 9.9583,
       "step": 12500
     },
     {
-      "epoch": 23.81852551984877,
-      "grad_norm": NaN,
+      "epoch": 23.82,
       "learning_rate": 0.0003,
-      "loss": 0.0,
+      "loss": 9.9579,
       "step": 12600
     },
     {
       "epoch": 24.0,
-      "eval_accuracy": 0.1995897435897436,
-      "eval_loss": NaN,
-      "eval_runtime": 5.0268,
-      "eval_samples_per_second": 99.467,
-      "eval_steps_per_second": 12.533,
+      "eval_accuracy": 0.16923076923076924,
+      "eval_loss": 10.070154190063477,
+      "eval_runtime": 4.8233,
+      "eval_samples_per_second": 103.664,
+      "eval_steps_per_second": 13.062,
       "step": 12696
     },
     {
@@ -1251,54 +1125,48 @@
       "step": 12696
     },
     {
-      "epoch": 24.00756143667297,
-      "grad_norm": NaN,
+      "epoch": 24.01,
       "learning_rate": 0.0003,
-      "loss": 0.0,
+      "loss": 9.9819,
       "step": 12700
     },
     {
-      "epoch": 24.196597353497165,
-      "grad_norm": NaN,
+      "epoch": 24.2,
       "learning_rate": 0.0003,
-      "loss": 0.0,
+      "loss": 9.9625,
       "step": 12800
     },
     {
-      "epoch": 24.38563327032136,
-      "grad_norm": NaN,
+      "epoch": 24.39,
       "learning_rate": 0.0003,
-      "loss": 0.0,
+      "loss": 9.9902,
       "step": 12900
     },
     {
-      "epoch": 24.574669187145556,
-      "grad_norm": NaN,
+      "epoch": 24.57,
       "learning_rate": 0.0003,
-      "loss": 0.0,
+      "loss": 9.9718,
       "step": 13000
     },
     {
-      "epoch": 24.763705103969755,
-      "grad_norm": NaN,
+      "epoch": 24.76,
       "learning_rate": 0.0003,
-      "loss": 0.0,
+      "loss": 9.9811,
       "step": 13100
     },
     {
-      "epoch": 24.95274102079395,
-      "grad_norm": NaN,
+      "epoch": 24.95,
       "learning_rate": 0.0003,
-      "loss": 0.0,
+      "loss": 9.9519,
       "step": 13200
     },
     {
       "epoch": 25.0,
-      "eval_accuracy": 0.1995897435897436,
-      "eval_loss": NaN,
-      "eval_runtime": 5.3305,
-      "eval_samples_per_second": 93.8,
-      "eval_steps_per_second": 11.819,
+      "eval_accuracy": 0.16923076923076924,
+      "eval_loss": 10.070154190063477,
+      "eval_runtime": 4.8319,
+      "eval_samples_per_second": 103.478,
+      "eval_steps_per_second": 13.038,
       "step": 13225
     },
     {
@@ -1308,47 +1176,42 @@
       "step": 13225
     },
     {
-      "epoch": 25.141776937618147,
-      "grad_norm": NaN,
+      "epoch": 25.14,
       "learning_rate": 0.0003,
-      "loss": 0.0,
+      "loss": 9.9708,
       "step": 13300
     },
     {
-      "epoch": 25.330812854442343,
-      "grad_norm": NaN,
+      "epoch": 25.33,
       "learning_rate": 0.0003,
-      "loss": 0.0,
+      "loss": 9.9703,
       "step": 13400
     },
     {
-      "epoch": 25.519848771266542,
-      "grad_norm": NaN,
+      "epoch": 25.52,
       "learning_rate": 0.0003,
-      "loss": 0.0,
+      "loss": 9.9751,
       "step": 13500
     },
     {
-      "epoch": 25.708884688090738,
-      "grad_norm": NaN,
+      "epoch": 25.71,
       "learning_rate": 0.0003,
-      "loss": 0.0,
+      "loss": 9.9741,
       "step": 13600
     },
     {
-      "epoch": 25.897920604914933,
-      "grad_norm": NaN,
+      "epoch": 25.9,
       "learning_rate": 0.0003,
-      "loss": 0.0,
+      "loss": 9.9849,
       "step": 13700
     },
     {
       "epoch": 26.0,
-      "eval_accuracy": 0.1995897435897436,
-      "eval_loss": NaN,
-      "eval_runtime": 5.173,
-      "eval_samples_per_second": 96.656,
-      "eval_steps_per_second": 12.179,
+      "eval_accuracy": 0.16923076923076924,
+      "eval_loss": 10.070154190063477,
+      "eval_runtime": 4.1084,
+      "eval_samples_per_second": 121.703,
+      "eval_steps_per_second": 15.335,
       "step": 13754
     },
     {
@@ -1358,47 +1221,42 @@
       "step": 13754
     },
     {
-      "epoch": 26.08695652173913,
-      "grad_norm": NaN,
+      "epoch": 26.09,
       "learning_rate": 0.0003,
-      "loss": 0.0,
+      "loss": 9.9706,
       "step": 13800
     },
     {
-      "epoch": 26.27599243856333,
-      "grad_norm": NaN,
+      "epoch": 26.28,
       "learning_rate": 0.0003,
-      "loss": 0.0,
+      "loss": 9.9562,
       "step": 13900
     },
     {
-      "epoch": 26.465028355387524,
-      "grad_norm": NaN,
+      "epoch": 26.47,
       "learning_rate": 0.0003,
-      "loss": 0.0,
+      "loss": 9.9702,
       "step": 14000
     },
     {
-      "epoch": 26.65406427221172,
-      "grad_norm": NaN,
+      "epoch": 26.65,
       "learning_rate": 0.0003,
-      "loss": 0.0,
+      "loss": 9.9922,
       "step": 14100
     },
     {
-      "epoch": 26.843100189035916,
-      "grad_norm": NaN,
+      "epoch": 26.84,
       "learning_rate": 0.0003,
-      "loss": 0.0,
+      "loss": 9.9591,
       "step": 14200
     },
     {
       "epoch": 27.0,
-      "eval_accuracy": 0.1995897435897436,
-      "eval_loss": NaN,
-      "eval_runtime": 5.2291,
-      "eval_samples_per_second": 95.618,
-      "eval_steps_per_second": 12.048,
+      "eval_accuracy": 0.16923076923076924,
+      "eval_loss": 10.070154190063477,
+      "eval_runtime": 4.4181,
+      "eval_samples_per_second": 113.171,
+      "eval_steps_per_second": 14.26,
       "step": 14283
     },
     {
@@ -1410,23 +1268,9 @@
   ],
   "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": 1.1626185967072707e+18,
-  "train_batch_size": 2,
+  "total_flos": 2.412230972633252e+18,
   "trial_name": null,
   "trial_params": null
 }

checkpoint-14283/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:186f3ae462881ca0bb3a6b5295a97a245e0a78172be9c14948a2d07a12aa4241
-size 5304
+oid sha256:b59ca157797b57867ac15a1da8ca60d782441a1506bb7cb43703c7748b856900
+size 4728