Training in progress, epoch 40, checkpoint

checkpoint-14001/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-14001/adapter_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:16af4f5ebc5f205491ed8ffb298bb3c4ccac1270a9b2a542459b9b457bdf1680
+size 143269386

checkpoint-14001/added_tokens.json

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

checkpoint-14001/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:844bd9d7c7d3ac9b34ad483dad4662e2510acdbccf9c9239cda3df5c0cfdfdca
-size 224537202
+oid sha256:53ac20c2d22ce80f20470c42d7e020e3961d63a35dc9195dd38257f2b7c17b83
+size 286585234

checkpoint-14001/rng_state_0.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e1b7fe752f2fb4d2a3b432b6ff1b3f6613d18a37b4cdef6b4367ce9d501d0d7b
+oid sha256:10f45d1a88177042dad8cc5c97bdc15b6c6da271061703b6a4642817d64ab64a
 size 15024

checkpoint-14001/rng_state_1.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c1a0051c84f7d4247328888d069b339c538cc724478361053b768ced66c3b41b
+oid sha256:fddbae0e1ea4aacb779108ee4c65459118211a43b7126091ee4d79c8774736d6
 size 15024

checkpoint-14001/rng_state_2.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9d4419d264a1ed3208a9531d1c8c959cde68b94344bdc11642d60a300953bf89
+oid sha256:6c5677425651a1000a7858b7161f7cb78357e57bb2c17bfff2dd4be1f6bbae6b
 size 15024

checkpoint-14001/rng_state_3.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:1b29d90244292201cce6270d11e57c6213bcb14c08d6437606abeb4117f8f1a9
+oid sha256:9b9768bc619485fb491274f54d3b18451f2bda41b018164a4a0278274fb89550
 size 15024

checkpoint-14001/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-14001/tokenizer.model

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

checkpoint-14001/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-14001/trainer_state.json

@@ -9,1661 +9,1429 @@
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.29282576866764276,
-      "grad_norm": 0.2411503791809082,
+      "epoch": 0.29,
       "learning_rate": 0.0001,
-      "loss": 2.0009,
+      "loss": 1.4842,
       "step": 100
     },
     {
-      "epoch": 0.5856515373352855,
-      "grad_norm": 0.22286373376846313,
+      "epoch": 0.59,
       "learning_rate": 0.0001,
-      "loss": 1.9538,
+      "loss": 1.4016,
       "step": 200
     },
     {
-      "epoch": 0.8784773060029283,
-      "grad_norm": 0.24079065024852753,
+      "epoch": 0.88,
       "learning_rate": 0.0001,
-      "loss": 1.9571,
+      "loss": 1.3892,
       "step": 300
     },
     {
-      "epoch": 0.9985358711566618,
-      "eval_accuracy": 0.45379487179487177,
-      "eval_loss": 3.9512319564819336,
-      "eval_runtime": 5.0647,
-      "eval_samples_per_second": 98.722,
-      "eval_steps_per_second": 12.439,
+      "epoch": 1.0,
+      "eval_accuracy": 0.4543589743589744,
+      "eval_loss": 3.4055869579315186,
+      "eval_runtime": 4.1688,
+      "eval_samples_per_second": 119.938,
+      "eval_steps_per_second": 15.112,
       "step": 341
     },
     {
-      "epoch": 0.9985358711566618,
-      "eval_bleu": 0.28767194245703276,
-      "eval_exact_match": 0.03,
-      "eval_prefix_exact_match": 0.042,
+      "epoch": 1.0,
+      "eval_bleu": 0.09711501224106496,
+      "eval_exact_match": 0.0,
+      "eval_prefix_exact_match": 0.054,
       "step": 341
     },
     {
-      "epoch": 1.171303074670571,
-      "grad_norm": 0.33546799421310425,
+      "epoch": 1.17,
       "learning_rate": 0.0001,
-      "loss": 1.9255,
+      "loss": 1.3608,
       "step": 400
     },
     {
-      "epoch": 1.4641288433382138,
-      "grad_norm": 0.3515666723251343,
+      "epoch": 1.46,
       "learning_rate": 0.0001,
-      "loss": 1.8684,
+      "loss": 1.3456,
       "step": 500
     },
     {
-      "epoch": 1.7569546120058566,
-      "grad_norm": 0.35733896493911743,
+      "epoch": 1.76,
       "learning_rate": 0.0001,
-      "loss": 1.8819,
+      "loss": 1.3499,
       "step": 600
     },
     {
       "epoch": 2.0,
-      "eval_accuracy": 0.4482564102564103,
-      "eval_loss": 4.112804412841797,
-      "eval_runtime": 5.0848,
-      "eval_samples_per_second": 98.333,
-      "eval_steps_per_second": 12.39,
+      "eval_accuracy": 0.4576923076923077,
+      "eval_loss": 3.453113317489624,
+      "eval_runtime": 4.4031,
+      "eval_samples_per_second": 113.556,
+      "eval_steps_per_second": 14.308,
       "step": 683
     },
     {
       "epoch": 2.0,
-      "eval_bleu": 0.24521038067633488,
-      "eval_exact_match": 0.02,
-      "eval_prefix_exact_match": 0.044,
+      "eval_bleu": 0.1225593574040313,
+      "eval_exact_match": 0.0,
+      "eval_prefix_exact_match": 0.056,
       "step": 683
     },
     {
-      "epoch": 2.049780380673499,
-      "grad_norm": 0.3784458339214325,
+      "epoch": 2.05,
       "learning_rate": 0.0001,
-      "loss": 1.8613,
+      "loss": 1.3238,
       "step": 700
     },
     {
-      "epoch": 2.342606149341142,
-      "grad_norm": 0.46702849864959717,
+      "epoch": 2.34,
       "learning_rate": 0.0001,
-      "loss": 1.743,
+      "loss": 1.2486,
       "step": 800
     },
     {
-      "epoch": 2.6354319180087846,
-      "grad_norm": 0.5156407356262207,
+      "epoch": 2.64,
       "learning_rate": 0.0001,
-      "loss": 1.7653,
+      "loss": 1.2452,
       "step": 900
     },
     {
-      "epoch": 2.9282576866764276,
-      "grad_norm": 0.4755038619041443,
+      "epoch": 2.93,
       "learning_rate": 0.0001,
-      "loss": 1.7702,
+      "loss": 1.2427,
       "step": 1000
     },
     {
-      "epoch": 2.998535871156662,
-      "eval_accuracy": 0.446051282051282,
-      "eval_loss": 4.327652454376221,
-      "eval_runtime": 5.1641,
-      "eval_samples_per_second": 96.822,
-      "eval_steps_per_second": 12.2,
+      "epoch": 3.0,
+      "eval_accuracy": 0.4584102564102564,
+      "eval_loss": 3.671137809753418,
+      "eval_runtime": 4.4094,
+      "eval_samples_per_second": 113.393,
+      "eval_steps_per_second": 14.288,
       "step": 1024
     },
     {
-      "epoch": 2.998535871156662,
-      "eval_bleu": 0.2598440457703149,
-      "eval_exact_match": 0.018,
-      "eval_prefix_exact_match": 0.034,
+      "epoch": 3.0,
+      "eval_bleu": 0.13165889774889974,
+      "eval_exact_match": 0.0,
+      "eval_prefix_exact_match": 0.056,
       "step": 1024
     },
     {
-      "epoch": 3.22108345534407,
-      "grad_norm": 0.5288150906562805,
+      "epoch": 3.22,
       "learning_rate": 0.0001,
-      "loss": 1.6325,
+      "loss": 1.1507,
       "step": 1100
     },
     {
-      "epoch": 3.513909224011713,
-      "grad_norm": 0.5784010887145996,
+      "epoch": 3.51,
       "learning_rate": 0.0001,
-      "loss": 1.5896,
+      "loss": 1.1218,
       "step": 1200
     },
     {
-      "epoch": 3.8067349926793557,
-      "grad_norm": 0.6458429098129272,
+      "epoch": 3.81,
       "learning_rate": 0.0001,
-      "loss": 1.6163,
+      "loss": 1.1231,
       "step": 1300
     },
     {
       "epoch": 4.0,
-      "eval_accuracy": 0.44235897435897437,
-      "eval_loss": 4.584905624389648,
-      "eval_runtime": 4.9243,
-      "eval_samples_per_second": 101.538,
-      "eval_steps_per_second": 12.794,
+      "eval_accuracy": 0.45697435897435895,
+      "eval_loss": 3.7999510765075684,
+      "eval_runtime": 4.0886,
+      "eval_samples_per_second": 122.29,
+      "eval_steps_per_second": 15.409,
       "step": 1366
     },
     {
       "epoch": 4.0,
-      "eval_bleu": 0.27967222841565076,
-      "eval_exact_match": 0.01,
-      "eval_prefix_exact_match": 0.038,
+      "eval_bleu": 0.15413978034957934,
+      "eval_exact_match": 0.0,
+      "eval_prefix_exact_match": 0.054,
       "step": 1366
     },
     {
-      "epoch": 4.099560761346998,
-      "grad_norm": 0.732841432094574,
+      "epoch": 4.1,
       "learning_rate": 0.0001,
-      "loss": 1.5441,
+      "loss": 1.0823,
       "step": 1400
     },
     {
-      "epoch": 4.392386530014641,
-      "grad_norm": 0.7372816205024719,
+      "epoch": 4.39,
       "learning_rate": 0.0001,
-      "loss": 1.4142,
+      "loss": 0.997,
       "step": 1500
     },
     {
-      "epoch": 4.685212298682284,
-      "grad_norm": 0.8235049247741699,
+      "epoch": 4.69,
       "learning_rate": 0.0001,
-      "loss": 1.4198,
+      "loss": 1.0024,
       "step": 1600
     },
     {
-      "epoch": 4.978038067349927,
-      "grad_norm": 0.7765743732452393,
+      "epoch": 4.98,
       "learning_rate": 0.0001,
-      "loss": 1.4427,
+      "loss": 0.995,
       "step": 1700
     },
     {
-      "epoch": 4.998535871156662,
-      "eval_accuracy": 0.43861538461538463,
-      "eval_loss": 4.850330829620361,
-      "eval_runtime": 4.9395,
-      "eval_samples_per_second": 101.224,
-      "eval_steps_per_second": 12.754,
+      "epoch": 5.0,
+      "eval_accuracy": 0.4552307692307692,
+      "eval_loss": 3.953216314315796,
+      "eval_runtime": 4.9824,
+      "eval_samples_per_second": 100.354,
+      "eval_steps_per_second": 12.645,
       "step": 1707
     },
     {
-      "epoch": 4.998535871156662,
-      "eval_bleu": 0.2170008640509292,
-      "eval_exact_match": 0.006,
-      "eval_prefix_exact_match": 0.032,
+      "epoch": 5.0,
+      "eval_bleu": 0.1611344512465791,
+      "eval_exact_match": 0.0,
+      "eval_prefix_exact_match": 0.046,
       "step": 1707
     },
     {
-      "epoch": 5.270863836017569,
-      "grad_norm": 0.9119400978088379,
+      "epoch": 5.27,
       "learning_rate": 0.0001,
-      "loss": 1.2469,
+      "loss": 0.8495,
       "step": 1800
     },
     {
-      "epoch": 5.563689604685212,
-      "grad_norm": 0.8966185450553894,
+      "epoch": 5.56,
       "learning_rate": 0.0001,
-      "loss": 1.2294,
+      "loss": 0.854,
       "step": 1900
     },
     {
-      "epoch": 5.856515373352855,
-      "grad_norm": 0.9629669189453125,
+      "epoch": 5.86,
       "learning_rate": 0.0001,
-      "loss": 1.2498,
+      "loss": 0.8693,
       "step": 2000
     },
     {
       "epoch": 6.0,
-      "eval_accuracy": 0.4348717948717949,
-      "eval_loss": 5.09260892868042,
-      "eval_runtime": 4.9572,
-      "eval_samples_per_second": 100.863,
-      "eval_steps_per_second": 12.709,
+      "eval_accuracy": 0.45261538461538464,
+      "eval_loss": 4.0766072273254395,
+      "eval_runtime": 4.8095,
+      "eval_samples_per_second": 103.961,
+      "eval_steps_per_second": 13.099,
       "step": 2049
     },
     {
       "epoch": 6.0,
-      "eval_bleu": 0.20196479624851718,
-      "eval_exact_match": 0.002,
-      "eval_prefix_exact_match": 0.03,
+      "eval_bleu": 0.1489904300924264,
+      "eval_exact_match": 0.0,
+      "eval_prefix_exact_match": 0.044,
       "step": 2049
     },
     {
-      "epoch": 6.149341142020498,
-      "grad_norm": 1.0808535814285278,
+      "epoch": 6.15,
       "learning_rate": 0.0001,
-      "loss": 1.1238,
+      "loss": 0.7917,
       "step": 2100
     },
     {
-      "epoch": 6.44216691068814,
-      "grad_norm": 1.1639196872711182,
+      "epoch": 6.44,
       "learning_rate": 0.0001,
-      "loss": 1.0301,
+      "loss": 0.7061,
       "step": 2200
     },
     {
-      "epoch": 6.734992679355783,
-      "grad_norm": 1.0662291049957275,
+      "epoch": 6.73,
       "learning_rate": 0.0001,
-      "loss": 1.0655,
+      "loss": 0.7302,
       "step": 2300
     },
     {
-      "epoch": 6.998535871156662,
-      "eval_accuracy": 0.4326153846153846,
-      "eval_loss": 5.270845890045166,
-      "eval_runtime": 5.089,
-      "eval_samples_per_second": 98.251,
-      "eval_steps_per_second": 12.38,
+      "epoch": 7.0,
+      "eval_accuracy": 0.4501025641025641,
+      "eval_loss": 4.171727180480957,
+      "eval_runtime": 4.2169,
+      "eval_samples_per_second": 118.57,
+      "eval_steps_per_second": 14.94,
       "step": 2390
     },
     {
-      "epoch": 6.998535871156662,
-      "eval_bleu": 0.17780932376070374,
-      "eval_exact_match": 0.004,
-      "eval_prefix_exact_match": 0.024,
+      "epoch": 7.0,
       "step": 2390
     },
     {
-      "epoch": 7.027818448023426,
-      "grad_norm": 1.6462695598602295,
+      "epoch": 7.03,
       "learning_rate": 0.0001,
-      "loss": 1.0358,
+      "loss": 0.7227,
       "step": 2400
     },
     {
-      "epoch": 7.320644216691068,
-      "grad_norm": 1.2542779445648193,
+      "epoch": 7.32,
       "learning_rate": 0.0001,
-      "loss": 0.8413,
+      "loss": 0.573,
       "step": 2500
     },
     {
-      "epoch": 7.613469985358711,
-      "grad_norm": 1.334762454032898,
+      "epoch": 7.61,
       "learning_rate": 0.0001,
-      "loss": 0.8654,
+      "loss": 0.6036,
       "step": 2600
     },
     {
-      "epoch": 7.906295754026354,
-      "grad_norm": 1.2401787042617798,
+      "epoch": 7.91,
       "learning_rate": 0.0001,
-      "loss": 0.8733,
+      "loss": 0.6033,
       "step": 2700
     },
     {
       "epoch": 8.0,
-      "eval_accuracy": 0.4316923076923077,
-      "eval_loss": 5.402440547943115,
-      "eval_runtime": 4.9383,
-      "eval_samples_per_second": 101.25,
-      "eval_steps_per_second": 12.758,
+      "eval_accuracy": 0.448,
+      "eval_loss": 4.277780055999756,
+      "eval_runtime": 4.0947,
+      "eval_samples_per_second": 122.11,
+      "eval_steps_per_second": 15.386,
       "step": 2732
     },
     {
       "epoch": 8.0,
-      "eval_bleu": 0.21418356891762502,
-      "eval_exact_match": 0.002,
-      "eval_prefix_exact_match": 0.026,
       "step": 2732
     },
     {
-      "epoch": 8.199121522693996,
-      "grad_norm": 1.255996584892273,
+      "epoch": 8.2,
       "learning_rate": 0.0001,
-      "loss": 0.7302,
+      "loss": 0.4945,
       "step": 2800
     },
     {
-      "epoch": 8.49194729136164,
-      "grad_norm": 1.2836525440216064,
+      "epoch": 8.49,
       "learning_rate": 0.0001,
-      "loss": 0.6875,
+      "loss": 0.4718,
       "step": 2900
     },
     {
-      "epoch": 8.784773060029282,
-      "grad_norm": 1.375715732574463,
+      "epoch": 8.78,
       "learning_rate": 0.0001,
-      "loss": 0.7219,
+      "loss": 0.4825,
       "step": 3000
     },
     {
-      "epoch": 8.998535871156662,
-      "eval_accuracy": 0.42938461538461536,
-      "eval_loss": 5.534832000732422,
-      "eval_runtime": 5.2363,
-      "eval_samples_per_second": 95.487,
-      "eval_steps_per_second": 12.031,
+      "epoch": 9.0,
+      "eval_accuracy": 0.4462051282051282,
+      "eval_loss": 4.34151554107666,
+      "eval_runtime": 4.8113,
+      "eval_samples_per_second": 103.921,
+      "eval_steps_per_second": 13.094,
       "step": 3073
     },
     {
-      "epoch": 8.998535871156662,
-      "eval_bleu": 0.20459866130314297,
-      "eval_exact_match": 0.0,
-      "eval_prefix_exact_match": 0.024,
+      "epoch": 9.0,
       "step": 3073
     },
     {
-      "epoch": 9.077598828696924,
-      "grad_norm": 1.481998324394226,
+      "epoch": 9.08,
       "learning_rate": 0.0001,
-      "loss": 0.6722,
+      "loss": 0.4538,
       "step": 3100
     },
     {
-      "epoch": 9.370424597364568,
-      "grad_norm": 1.4086748361587524,
+      "epoch": 9.37,
       "learning_rate": 0.0001,
-      "loss": 0.5489,
+      "loss": 0.3609,
       "step": 3200
     },
     {
-      "epoch": 9.66325036603221,
-      "grad_norm": 1.4381046295166016,
+      "epoch": 9.66,
       "learning_rate": 0.0001,
-      "loss": 0.565,
+      "loss": 0.3769,
       "step": 3300
     },
     {
-      "epoch": 9.956076134699854,
-      "grad_norm": 1.7364269495010376,
+      "epoch": 9.96,
       "learning_rate": 0.0001,
-      "loss": 0.5932,
+      "loss": 0.387,
       "step": 3400
     },
     {
       "epoch": 10.0,
-      "eval_accuracy": 0.42605128205128207,
-      "eval_loss": 5.768980503082275,
-      "eval_runtime": 5.0821,
-      "eval_samples_per_second": 98.384,
-      "eval_steps_per_second": 12.396,
+      "eval_accuracy": 0.4463076923076923,
+      "eval_loss": 4.413115978240967,
+      "eval_runtime": 4.2874,
+      "eval_samples_per_second": 116.62,
+      "eval_steps_per_second": 14.694,
       "step": 3415
     },
     {
       "epoch": 10.0,
-      "eval_bleu": 0.19515580420261575,
-      "eval_exact_match": 0.002,
-      "eval_prefix_exact_match": 0.026,
       "step": 3415
     },
     {
-      "epoch": 10.248901903367496,
-      "grad_norm": 1.599753737449646,
+      "epoch": 10.25,
       "learning_rate": 0.0001,
-      "loss": 0.4581,
+      "loss": 0.2911,
       "step": 3500
     },
     {
-      "epoch": 10.541727672035138,
-      "grad_norm": 1.5595277547836304,
+      "epoch": 10.54,
       "learning_rate": 0.0001,
-      "loss": 0.4519,
+      "loss": 0.2881,
       "step": 3600
     },
     {
-      "epoch": 10.834553440702782,
-      "grad_norm": 1.6201436519622803,
+      "epoch": 10.83,
       "learning_rate": 0.0001,
-      "loss": 0.4719,
+      "loss": 0.2933,
       "step": 3700
     },
     {
-      "epoch": 10.998535871156662,
-      "eval_accuracy": 0.42538461538461536,
-      "eval_loss": 5.89434289932251,
-      "eval_runtime": 5.0717,
-      "eval_samples_per_second": 98.586,
-      "eval_steps_per_second": 12.422,
+      "epoch": 11.0,
+      "eval_accuracy": 0.44343589743589745,
+      "eval_loss": 4.490556716918945,
+      "eval_runtime": 4.2824,
+      "eval_samples_per_second": 116.757,
+      "eval_steps_per_second": 14.711,
       "step": 3756
     },
     {
-      "epoch": 10.998535871156662,
-      "eval_bleu": 0.2074466521823303,
-      "eval_exact_match": 0.0,
-      "eval_prefix_exact_match": 0.02,
+      "epoch": 11.0,
       "step": 3756
     },
     {
-      "epoch": 11.127379209370424,
-      "grad_norm": 1.6209241151809692,
+      "epoch": 11.13,
       "learning_rate": 0.0001,
-      "loss": 0.4263,
+      "loss": 0.2653,
       "step": 3800
     },
     {
-      "epoch": 11.420204978038067,
-      "grad_norm": 1.5148394107818604,
+      "epoch": 11.42,
       "learning_rate": 0.0001,
-      "loss": 0.3625,
+      "loss": 0.2181,
       "step": 3900
     },
     {
-      "epoch": 11.71303074670571,
-      "grad_norm": 1.6862399578094482,
+      "epoch": 11.71,
       "learning_rate": 0.0001,
-      "loss": 0.3838,
+      "loss": 0.2344,
       "step": 4000
     },
     {
       "epoch": 12.0,
-      "eval_accuracy": 0.4247179487179487,
-      "eval_loss": 6.019149303436279,
-      "eval_runtime": 5.3028,
-      "eval_samples_per_second": 94.291,
-      "eval_steps_per_second": 11.881,
+      "eval_accuracy": 0.44251282051282054,
+      "eval_loss": 4.651719570159912,
+      "eval_runtime": 4.0941,
+      "eval_samples_per_second": 122.126,
+      "eval_steps_per_second": 15.388,
       "step": 4098
     },
     {
       "epoch": 12.0,
-      "eval_bleu": 0.16189430124724263,
-      "eval_exact_match": 0.002,
-      "eval_prefix_exact_match": 0.028,
       "step": 4098
     },
     {
-      "epoch": 12.005856515373353,
-      "grad_norm": 1.3296407461166382,
+      "epoch": 12.01,
       "learning_rate": 0.0001,
-      "loss": 0.3952,
+      "loss": 0.2414,
       "step": 4100
     },
     {
-      "epoch": 12.298682284040996,
-      "grad_norm": 1.3439353704452515,
+      "epoch": 12.3,
       "learning_rate": 0.0001,
-      "loss": 0.2901,
+      "loss": 0.173,
       "step": 4200
     },
     {
-      "epoch": 12.591508052708638,
-      "grad_norm": 1.5329118967056274,
+      "epoch": 12.59,
       "learning_rate": 0.0001,
-      "loss": 0.3114,
+      "loss": 0.183,
       "step": 4300
     },
     {
-      "epoch": 12.88433382137628,
-      "grad_norm": 1.649775743484497,
+      "epoch": 12.88,
       "learning_rate": 0.0001,
-      "loss": 0.329,
+      "loss": 0.1919,
       "step": 4400
     },
     {
-      "epoch": 12.998535871156662,
-      "eval_accuracy": 0.42456410256410254,
-      "eval_loss": 6.104373455047607,
-      "eval_runtime": 5.9944,
-      "eval_samples_per_second": 83.411,
-      "eval_steps_per_second": 10.51,
+      "epoch": 13.0,
+      "eval_accuracy": 0.44076923076923075,
+      "eval_loss": 4.751525402069092,
+      "eval_runtime": 4.0902,
+      "eval_samples_per_second": 122.244,
+      "eval_steps_per_second": 15.403,
       "step": 4439
     },
     {
-      "epoch": 12.998535871156662,
-      "eval_bleu": 0.17170793542335236,
-      "eval_exact_match": 0.002,
-      "eval_prefix_exact_match": 0.028,
+      "epoch": 13.0,
       "step": 4439
     },
     {
-      "epoch": 13.177159590043924,
-      "grad_norm": 1.5262281894683838,
+      "epoch": 13.18,
       "learning_rate": 0.0001,
-      "loss": 0.2758,
+      "loss": 0.1661,
       "step": 4500
     },
     {
-      "epoch": 13.469985358711567,
-      "grad_norm": 1.5962131023406982,
+      "epoch": 13.47,
       "learning_rate": 0.0001,
-      "loss": 0.2594,
+      "loss": 0.1501,
       "step": 4600
     },
     {
-      "epoch": 13.762811127379209,
-      "grad_norm": 1.4258382320404053,
+      "epoch": 13.76,
       "learning_rate": 0.0001,
-      "loss": 0.2742,
+      "loss": 0.1581,
       "step": 4700
     },
     {
       "epoch": 14.0,
-      "eval_accuracy": 0.4216410256410256,
-      "eval_loss": 6.146524906158447,
-      "eval_runtime": 4.9576,
-      "eval_samples_per_second": 100.855,
-      "eval_steps_per_second": 12.708,
+      "eval_accuracy": 0.442051282051282,
+      "eval_loss": 4.7323455810546875,
+      "eval_runtime": 4.8128,
+      "eval_samples_per_second": 103.89,
+      "eval_steps_per_second": 13.09,
       "step": 4781
     },
     {
       "epoch": 14.0,
-      "eval_bleu": 0.20004000747911055,
-      "eval_exact_match": 0.002,
-      "eval_prefix_exact_match": 0.026,
+      "eval_bleu": 0.01991302915032707,
+      "eval_exact_match": 0.0,
+      "eval_prefix_exact_match": 0.0,
       "step": 4781
     },
     {
-      "epoch": 14.055636896046853,
-      "grad_norm": 1.388717532157898,
+      "epoch": 14.06,
       "learning_rate": 0.0001,
-      "loss": 0.269,
+      "loss": 0.1563,
       "step": 4800
     },
     {
-      "epoch": 14.348462664714495,
-      "grad_norm": 1.2370668649673462,
+      "epoch": 14.35,
       "learning_rate": 0.0001,
-      "loss": 0.2226,
+      "loss": 0.13,
       "step": 4900
     },
     {
-      "epoch": 14.641288433382137,
-      "grad_norm": 1.3778610229492188,
+      "epoch": 14.64,
       "learning_rate": 0.0001,
-      "loss": 0.2334,
+      "loss": 0.1388,
       "step": 5000
     },
     {
-      "epoch": 14.93411420204978,
-      "grad_norm": 1.6371277570724487,
+      "epoch": 14.93,
       "learning_rate": 0.0001,
-      "loss": 0.2432,
+      "loss": 0.1429,
       "step": 5100
     },
     {
-      "epoch": 14.998535871156662,
-      "eval_accuracy": 0.4226666666666667,
-      "eval_loss": 6.3254475593566895,
-      "eval_runtime": 5.2196,
-      "eval_samples_per_second": 95.794,
-      "eval_steps_per_second": 12.07,
+      "epoch": 15.0,
+      "eval_accuracy": 0.44066666666666665,
+      "eval_loss": 4.810059547424316,
+      "eval_runtime": 5.0086,
+      "eval_samples_per_second": 99.829,
+      "eval_steps_per_second": 12.578,
       "step": 5122
     },
     {
-      "epoch": 14.998535871156662,
-      "eval_bleu": 0.16083749853030369,
+      "epoch": 15.0,
+      "eval_bleu": 0.007355658423274405,
       "eval_exact_match": 0.0,
-      "eval_prefix_exact_match": 0.024,
+      "eval_prefix_exact_match": 0.002,
       "step": 5122
     },
     {
-      "epoch": 15.226939970717423,
-      "grad_norm": 1.2761411666870117,
+      "epoch": 15.23,
       "learning_rate": 0.0001,
-      "loss": 0.2066,
+      "loss": 0.1228,
       "step": 5200
     },
     {
-      "epoch": 15.519765739385067,
-      "grad_norm": 1.6878156661987305,
+      "epoch": 15.52,
       "learning_rate": 0.0001,
-      "loss": 0.2044,
+      "loss": 0.1233,
       "step": 5300
     },
     {
-      "epoch": 15.812591508052709,
-      "grad_norm": 1.4964863061904907,
+      "epoch": 15.81,
       "learning_rate": 0.0001,
-      "loss": 0.2158,
+      "loss": 0.1279,
       "step": 5400
     },
     {
       "epoch": 16.0,
-      "eval_accuracy": 0.4227692307692308,
-      "eval_loss": 6.441042423248291,
-      "eval_runtime": 4.9987,
-      "eval_samples_per_second": 100.025,
-      "eval_steps_per_second": 12.603,
+      "eval_accuracy": 0.44061538461538463,
+      "eval_loss": 4.84822940826416,
+      "eval_runtime": 4.1962,
+      "eval_samples_per_second": 119.156,
+      "eval_steps_per_second": 15.014,
       "step": 5464
     },
     {
       "epoch": 16.0,
-      "eval_bleu": 0.1349628182313855,
-      "eval_exact_match": 0.002,
-      "eval_prefix_exact_match": 0.026,
+      "eval_bleu": 0.020518239891916933,
+      "eval_exact_match": 0.0,
+      "eval_prefix_exact_match": 0.0,
       "step": 5464
     },
     {
-      "epoch": 16.105417276720353,
-      "grad_norm": 0.9965784549713135,
+      "epoch": 16.11,
       "learning_rate": 0.0001,
-      "loss": 0.2038,
+      "loss": 0.1248,
       "step": 5500
     },
     {
-      "epoch": 16.398243045387993,
-      "grad_norm": 1.2694511413574219,
+      "epoch": 16.4,
       "learning_rate": 0.0001,
-      "loss": 0.1857,
+      "loss": 0.1118,
       "step": 5600
     },
     {
-      "epoch": 16.691068814055637,
-      "grad_norm": 1.436286211013794,
+      "epoch": 16.69,
       "learning_rate": 0.0001,
-      "loss": 0.1944,
+      "loss": 0.1172,
       "step": 5700
     },
     {
-      "epoch": 16.98389458272328,
-      "grad_norm": 1.526081919670105,
+      "epoch": 16.98,
       "learning_rate": 0.0001,
-      "loss": 0.2013,
+      "loss": 0.1231,
       "step": 5800
     },
     {
-      "epoch": 16.998535871156662,
-      "eval_accuracy": 0.4214871794871795,
-      "eval_loss": 6.39243745803833,
-      "eval_runtime": 5.1033,
-      "eval_samples_per_second": 97.975,
-      "eval_steps_per_second": 12.345,
+      "epoch": 17.0,
+      "eval_accuracy": 0.4410769230769231,
+      "eval_loss": 4.973475933074951,
+      "eval_runtime": 4.4042,
+      "eval_samples_per_second": 113.529,
+      "eval_steps_per_second": 14.305,
       "step": 5805
     },
     {
-      "epoch": 16.998535871156662,
-      "eval_bleu": 0.18366001556911046,
+      "epoch": 17.0,
+      "eval_bleu": 0.025070041424787165,
       "eval_exact_match": 0.0,
-      "eval_prefix_exact_match": 0.018,
+      "eval_prefix_exact_match": 0.0,
       "step": 5805
     },
     {
-      "epoch": 17.27672035139092,
-      "grad_norm": 1.1292760372161865,
+      "epoch": 17.28,
       "learning_rate": 0.0001,
-      "loss": 0.1694,
+      "loss": 0.1068,
       "step": 5900
     },
     {
-      "epoch": 17.569546120058565,
-      "grad_norm": 1.440766453742981,
+      "epoch": 17.57,
       "learning_rate": 0.0001,
-      "loss": 0.1782,
+      "loss": 0.111,
       "step": 6000
     },
     {
-      "epoch": 17.86237188872621,
-      "grad_norm": 1.2640533447265625,
+      "epoch": 17.86,
       "learning_rate": 0.0001,
-      "loss": 0.1851,
+      "loss": 0.1145,
       "step": 6100
     },
     {
       "epoch": 18.0,
-      "eval_accuracy": 0.42005128205128206,
-      "eval_loss": 6.5217366218566895,
-      "eval_runtime": 5.0655,
-      "eval_samples_per_second": 98.707,
-      "eval_steps_per_second": 12.437,
+      "eval_accuracy": 0.44153846153846155,
+      "eval_loss": 5.012055397033691,
+      "eval_runtime": 4.1955,
+      "eval_samples_per_second": 119.174,
+      "eval_steps_per_second": 15.016,
       "step": 6147
     },
     {
       "epoch": 18.0,
-      "eval_bleu": 0.14616910828479238,
+      "eval_bleu": 0.0018720973379033757,
       "eval_exact_match": 0.0,
-      "eval_prefix_exact_match": 0.026,
+      "eval_prefix_exact_match": 0.0,
       "step": 6147
     },
     {
-      "epoch": 18.15519765739385,
-      "grad_norm": 1.430617094039917,
+      "epoch": 18.16,
       "learning_rate": 0.0001,
-      "loss": 0.1717,
+      "loss": 0.1093,
       "step": 6200
     },
     {
-      "epoch": 18.448023426061493,
-      "grad_norm": 1.1885381937026978,
+      "epoch": 18.45,
       "learning_rate": 0.0001,
-      "loss": 0.1656,
+      "loss": 0.1051,
       "step": 6300
     },
     {
-      "epoch": 18.740849194729137,
-      "grad_norm": 1.64545738697052,
+      "epoch": 18.74,
       "learning_rate": 0.0001,
-      "loss": 0.1721,
+      "loss": 0.1087,
       "step": 6400
     },
     {
-      "epoch": 18.998535871156662,
-      "eval_accuracy": 0.4208717948717949,
-      "eval_loss": 6.557307720184326,
-      "eval_runtime": 5.092,
-      "eval_samples_per_second": 98.193,
-      "eval_steps_per_second": 12.372,
+      "epoch": 19.0,
+      "eval_accuracy": 0.43943589743589745,
+      "eval_loss": 4.983563423156738,
+      "eval_runtime": 4.0932,
+      "eval_samples_per_second": 122.154,
+      "eval_steps_per_second": 15.391,
       "step": 6488
     },
     {
-      "epoch": 18.998535871156662,
-      "eval_bleu": 0.16916503510298256,
+      "epoch": 19.0,
+      "eval_bleu": 0.0014273377715875865,
       "eval_exact_match": 0.0,
-      "eval_prefix_exact_match": 0.02,
+      "eval_prefix_exact_match": 0.0,
       "step": 6488
     },
     {
-      "epoch": 19.03367496339678,
-      "grad_norm": 1.2168452739715576,
+      "epoch": 19.03,
       "learning_rate": 0.0001,
-      "loss": 0.1743,
+      "loss": 0.1115,
       "step": 6500
     },
     {
-      "epoch": 19.32650073206442,
-      "grad_norm": 1.2320917844772339,
+      "epoch": 19.33,
       "learning_rate": 0.0001,
-      "loss": 0.152,
+      "loss": 0.1001,
       "step": 6600
     },
     {
-      "epoch": 19.619326500732065,
-      "grad_norm": 1.1541388034820557,
+      "epoch": 19.62,
       "learning_rate": 0.0001,
-      "loss": 0.1619,
+      "loss": 0.1049,
       "step": 6700
     },
     {
-      "epoch": 19.91215226939971,
-      "grad_norm": 1.2821346521377563,
+      "epoch": 19.91,
       "learning_rate": 0.0001,
-      "loss": 0.1676,
+      "loss": 0.1084,
       "step": 6800
     },
     {
       "epoch": 20.0,
-      "eval_accuracy": 0.42138461538461536,
-      "eval_loss": 6.566097736358643,
-      "eval_runtime": 5.0056,
-      "eval_samples_per_second": 99.888,
-      "eval_steps_per_second": 12.586,
+      "eval_accuracy": 0.4388205128205128,
+      "eval_loss": 5.117064952850342,
+      "eval_runtime": 4.0908,
+      "eval_samples_per_second": 122.227,
+      "eval_steps_per_second": 15.401,
       "step": 6830
     },
     {
       "epoch": 20.0,
-      "eval_bleu": 0.21213643830050385,
+      "eval_bleu": 0.004914730802266865,
       "eval_exact_match": 0.0,
-      "eval_prefix_exact_match": 0.026,
+      "eval_prefix_exact_match": 0.0,
       "step": 6830
     },
     {
-      "epoch": 20.20497803806735,
-      "grad_norm": 1.1950445175170898,
+      "epoch": 20.2,
       "learning_rate": 0.0001,
-      "loss": 0.1533,
+      "loss": 0.102,
       "step": 6900
     },
     {
-      "epoch": 20.497803806734993,
-      "grad_norm": 1.253625750541687,
+      "epoch": 20.5,
       "learning_rate": 0.0001,
-      "loss": 0.1522,
+      "loss": 0.1014,
       "step": 7000
     },
     {
-      "epoch": 20.790629575402637,
-      "grad_norm": 1.193354606628418,
+      "epoch": 20.79,
       "learning_rate": 0.0001,
-      "loss": 0.1579,
+      "loss": 0.1069,
       "step": 7100
     },
     {
-      "epoch": 20.998535871156662,
-      "eval_accuracy": 0.42133333333333334,
-      "eval_loss": 6.566288948059082,
-      "eval_runtime": 5.2039,
-      "eval_samples_per_second": 96.082,
-      "eval_steps_per_second": 12.106,
+      "epoch": 21.0,
+      "eval_accuracy": 0.44051282051282054,
+      "eval_loss": 5.011991500854492,
+      "eval_runtime": 4.1961,
+      "eval_samples_per_second": 119.159,
+      "eval_steps_per_second": 15.014,
       "step": 7171
     },
     {
-      "epoch": 20.998535871156662,
-      "eval_bleu": 0.16340256740662695,
-      "eval_exact_match": 0.0,
-      "eval_prefix_exact_match": 0.022,
+      "epoch": 21.0,
       "step": 7171
     },
     {
-      "epoch": 21.083455344070277,
-      "grad_norm": 1.0565904378890991,
+      "epoch": 21.08,
       "learning_rate": 0.0001,
-      "loss": 0.1575,
+      "loss": 0.1075,
       "step": 7200
     },
     {
-      "epoch": 21.37628111273792,
-      "grad_norm": 1.1256818771362305,
+      "epoch": 21.38,
       "learning_rate": 0.0001,
-      "loss": 0.1459,
+      "loss": 0.0985,
       "step": 7300
     },
     {
-      "epoch": 21.669106881405565,
-      "grad_norm": 1.4385403394699097,
+      "epoch": 21.67,
       "learning_rate": 0.0001,
-      "loss": 0.1511,
+      "loss": 0.1032,
       "step": 7400
     },
     {
-      "epoch": 21.961932650073205,
-      "grad_norm": 1.1152088642120361,
+      "epoch": 21.96,
       "learning_rate": 0.0001,
-      "loss": 0.1575,
+      "loss": 0.1075,
       "step": 7500
     },
     {
       "epoch": 22.0,
-      "eval_accuracy": 0.42015384615384616,
-      "eval_loss": 6.625944137573242,
-      "eval_runtime": 5.0735,
-      "eval_samples_per_second": 98.552,
-      "eval_steps_per_second": 12.418,
+      "eval_accuracy": 0.44,
+      "eval_loss": 5.234306812286377,
+      "eval_runtime": 4.4185,
+      "eval_samples_per_second": 113.162,
+      "eval_steps_per_second": 14.258,
       "step": 7513
     },
     {
       "epoch": 22.0,
-      "eval_bleu": 0.19792591482411717,
-      "eval_exact_match": 0.0,
-      "eval_prefix_exact_match": 0.022,
       "step": 7513
     },
     {
-      "epoch": 22.25475841874085,
-      "grad_norm": 1.1440153121948242,
+      "epoch": 22.25,
       "learning_rate": 0.0001,
-      "loss": 0.1414,
+      "loss": 0.0959,
       "step": 7600
     },
     {
-      "epoch": 22.547584187408493,
-      "grad_norm": 1.1086493730545044,
+      "epoch": 22.55,
       "learning_rate": 0.0001,
-      "loss": 0.1435,
+      "loss": 0.0998,
       "step": 7700
     },
     {
-      "epoch": 22.840409956076133,
-      "grad_norm": 1.2570273876190186,
+      "epoch": 22.84,
       "learning_rate": 0.0001,
-      "loss": 0.15,
+      "loss": 0.1024,
       "step": 7800
     },
     {
-      "epoch": 22.998535871156662,
-      "eval_accuracy": 0.42143589743589743,
-      "eval_loss": 6.595519065856934,
-      "eval_runtime": 5.2943,
-      "eval_samples_per_second": 94.442,
-      "eval_steps_per_second": 11.9,
+      "epoch": 23.0,
+      "eval_accuracy": 0.4408717948717949,
+      "eval_loss": 5.150094985961914,
+      "eval_runtime": 4.2006,
+      "eval_samples_per_second": 119.032,
+      "eval_steps_per_second": 14.998,
       "step": 7854
     },
     {
-      "epoch": 22.998535871156662,
-      "eval_bleu": 0.17402208428859392,
-      "eval_exact_match": 0.002,
-      "eval_prefix_exact_match": 0.028,
+      "epoch": 23.0,
+      "eval_bleu": 0.008507706871345696,
+      "eval_exact_match": 0.0,
+      "eval_prefix_exact_match": 0.002,
       "step": 7854
     },
     {
-      "epoch": 23.133235724743777,
-      "grad_norm": 1.070988655090332,
+      "epoch": 23.13,
       "learning_rate": 0.0001,
-      "loss": 0.1455,
+      "loss": 0.0997,
       "step": 7900
     },
     {
-      "epoch": 23.42606149341142,
-      "grad_norm": 1.021813154220581,
+      "epoch": 23.43,
       "learning_rate": 0.0001,
-      "loss": 0.138,
+      "loss": 0.0946,
       "step": 8000
     },
     {
-      "epoch": 23.718887262079065,
-      "grad_norm": 1.2333232164382935,
+      "epoch": 23.72,
       "learning_rate": 0.0001,
-      "loss": 0.1427,
+      "loss": 0.0981,
       "step": 8100
     },
     {
       "epoch": 24.0,
-      "eval_accuracy": 0.4216410256410256,
-      "eval_loss": 6.62971830368042,
-      "eval_runtime": 4.9115,
-      "eval_samples_per_second": 101.802,
-      "eval_steps_per_second": 12.827,
+      "eval_accuracy": 0.44025641025641027,
+      "eval_loss": 5.080117225646973,
+      "eval_runtime": 4.1942,
+      "eval_samples_per_second": 119.212,
+      "eval_steps_per_second": 15.021,
       "step": 8196
     },
     {
       "epoch": 24.0,
-      "eval_bleu": 0.1531918581052099,
-      "eval_exact_match": 0.0,
-      "eval_prefix_exact_match": 0.028,
       "step": 8196
     },
     {
-      "epoch": 24.011713030746705,
-      "grad_norm": 0.9755038022994995,
+      "epoch": 24.01,
       "learning_rate": 0.0001,
-      "loss": 0.1488,
+      "loss": 0.101,
       "step": 8200
     },
     {
-      "epoch": 24.30453879941435,
-      "grad_norm": 1.2587764263153076,
+      "epoch": 24.3,
       "learning_rate": 0.0001,
-      "loss": 0.1328,
+      "loss": 0.0923,
       "step": 8300
     },
     {
-      "epoch": 24.597364568081993,
-      "grad_norm": 0.9521226286888123,
+      "epoch": 24.6,
       "learning_rate": 0.0001,
-      "loss": 0.1384,
+      "loss": 0.095,
       "step": 8400
     },
     {
-      "epoch": 24.890190336749633,
-      "grad_norm": 1.2158619165420532,
+      "epoch": 24.89,
       "learning_rate": 0.0001,
-      "loss": 0.145,
+      "loss": 0.097,
       "step": 8500
     },
     {
-      "epoch": 24.998535871156662,
-      "eval_accuracy": 0.4226666666666667,
-      "eval_loss": 6.57574462890625,
-      "eval_runtime": 5.0443,
-      "eval_samples_per_second": 99.122,
-      "eval_steps_per_second": 12.489,
+      "epoch": 25.0,
+      "eval_accuracy": 0.44158974358974357,
+      "eval_loss": 5.103724956512451,
+      "eval_runtime": 4.307,
+      "eval_samples_per_second": 116.09,
+      "eval_steps_per_second": 14.627,
       "step": 8537
     },
     {
-      "epoch": 24.998535871156662,
-      "eval_bleu": 0.17454308300545518,
-      "eval_exact_match": 0.0,
-      "eval_prefix_exact_match": 0.022,
+      "epoch": 25.0,
       "step": 8537
     },
     {
-      "epoch": 25.183016105417277,
-      "grad_norm": 1.0446624755859375,
+      "epoch": 25.18,
       "learning_rate": 0.0001,
-      "loss": 0.1357,
+      "loss": 0.0925,
       "step": 8600
     },
     {
-      "epoch": 25.47584187408492,
-      "grad_norm": 1.07607901096344,
+      "epoch": 25.48,
       "learning_rate": 0.0001,
-      "loss": 0.1341,
+      "loss": 0.0926,
       "step": 8700
     },
     {
-      "epoch": 25.76866764275256,
-      "grad_norm": 1.0376451015472412,
+      "epoch": 25.77,
       "learning_rate": 0.0001,
-      "loss": 0.1393,
+      "loss": 0.0963,
       "step": 8800
     },
     {
       "epoch": 26.0,
-      "eval_accuracy": 0.42133333333333334,
-      "eval_loss": 6.567473411560059,
-      "eval_runtime": 5.3766,
-      "eval_samples_per_second": 92.995,
-      "eval_steps_per_second": 11.717,
+      "eval_accuracy": 0.4397948717948718,
+      "eval_loss": 5.206360340118408,
+      "eval_runtime": 4.8198,
+      "eval_samples_per_second": 103.739,
+      "eval_steps_per_second": 13.071,
       "step": 8879
     },
     {
       "epoch": 26.0,
-      "eval_bleu": 0.183512215739427,
-      "eval_exact_match": 0.0,
-      "eval_prefix_exact_match": 0.026,
       "step": 8879
     },
     {
-      "epoch": 26.061493411420205,
-      "grad_norm": 0.8689504861831665,
+      "epoch": 26.06,
       "learning_rate": 0.0001,
-      "loss": 0.1415,
+      "loss": 0.0966,
       "step": 8900
     },
     {
-      "epoch": 26.35431918008785,
-      "grad_norm": 0.8533344864845276,
+      "epoch": 26.35,
       "learning_rate": 0.0001,
-      "loss": 0.1292,
+      "loss": 0.0909,
       "step": 9000
     },
     {
-      "epoch": 26.64714494875549,
-      "grad_norm": 1.0425468683242798,
+      "epoch": 26.65,
       "learning_rate": 0.0001,
-      "loss": 0.1348,
+      "loss": 0.0951,
       "step": 9100
     },
     {
-      "epoch": 26.939970717423133,
-      "grad_norm": 1.2276948690414429,
+      "epoch": 26.94,
       "learning_rate": 0.0001,
-      "loss": 0.1405,
+      "loss": 0.0983,
       "step": 9200
     },
     {
-      "epoch": 26.998535871156662,
-      "eval_accuracy": 0.42133333333333334,
-      "eval_loss": 6.664970874786377,
-      "eval_runtime": 5.6177,
-      "eval_samples_per_second": 89.004,
-      "eval_steps_per_second": 11.215,
+      "epoch": 27.0,
+      "eval_accuracy": 0.4413846153846154,
+      "eval_loss": 5.066434383392334,
+      "eval_runtime": 4.1912,
+      "eval_samples_per_second": 119.298,
+      "eval_steps_per_second": 15.032,
       "step": 9220
     },
     {
-      "epoch": 26.998535871156662,
-      "eval_bleu": 0.19671240169357163,
-      "eval_exact_match": 0.002,
-      "eval_prefix_exact_match": 0.026,
+      "epoch": 27.0,
       "step": 9220
     },
     {
-      "epoch": 27.232796486090777,
-      "grad_norm": 1.0451931953430176,
+      "epoch": 27.23,
       "learning_rate": 0.0001,
-      "loss": 0.1282,
+      "loss": 0.092,
       "step": 9300
     },
     {
-      "epoch": 27.525622254758417,
-      "grad_norm": 1.0023565292358398,
+      "epoch": 27.53,
       "learning_rate": 0.0001,
-      "loss": 0.1307,
+      "loss": 0.0935,
       "step": 9400
     },
     {
-      "epoch": 27.81844802342606,
-      "grad_norm": 0.9287552833557129,
+      "epoch": 27.82,
       "learning_rate": 0.0001,
-      "loss": 0.1365,
+      "loss": 0.0969,
       "step": 9500
     },
     {
       "epoch": 28.0,
-      "eval_accuracy": 0.420974358974359,
-      "eval_loss": 6.642728328704834,
-      "eval_runtime": 4.9204,
-      "eval_samples_per_second": 101.618,
-      "eval_steps_per_second": 12.804,
+      "eval_accuracy": 0.440974358974359,
+      "eval_loss": 5.255864143371582,
+      "eval_runtime": 4.1948,
+      "eval_samples_per_second": 119.194,
+      "eval_steps_per_second": 15.018,
       "step": 9562
     },
     {
       "epoch": 28.0,
-      "eval_bleu": 0.17149559728372338,
-      "eval_exact_match": 0.0,
-      "eval_prefix_exact_match": 0.028,
       "step": 9562
     },
     {
-      "epoch": 28.111273792093705,
-      "grad_norm": 0.7279195785522461,
+      "epoch": 28.11,
       "learning_rate": 0.0001,
-      "loss": 0.133,
+      "loss": 0.0947,
       "step": 9600
     },
     {
-      "epoch": 28.404099560761345,
-      "grad_norm": 0.9348853230476379,
+      "epoch": 28.4,
       "learning_rate": 0.0001,
-      "loss": 0.1268,
+      "loss": 0.0915,
       "step": 9700
     },
     {
-      "epoch": 28.69692532942899,
-      "grad_norm": 0.9402393102645874,
+      "epoch": 28.7,
       "learning_rate": 0.0001,
-      "loss": 0.1323,
+      "loss": 0.0948,
       "step": 9800
     },
     {
-      "epoch": 28.989751098096633,
-      "grad_norm": 0.9316591620445251,
+      "epoch": 28.99,
       "learning_rate": 0.0001,
-      "loss": 0.1372,
+      "loss": 0.0966,
       "step": 9900
     },
     {
-      "epoch": 28.998535871156662,
-      "eval_accuracy": 0.4208717948717949,
-      "eval_loss": 6.548105239868164,
-      "eval_runtime": 4.8211,
-      "eval_samples_per_second": 103.71,
-      "eval_steps_per_second": 13.067,
+      "epoch": 29.0,
+      "eval_accuracy": 0.44035897435897436,
+      "eval_loss": 5.196049690246582,
+      "eval_runtime": 5.1571,
+      "eval_samples_per_second": 96.953,
+      "eval_steps_per_second": 12.216,
       "step": 9903
     },
     {
-      "epoch": 28.998535871156662,
-      "eval_bleu": 0.16884007365872924,
-      "eval_exact_match": 0.0,
-      "eval_prefix_exact_match": 0.02,
+      "epoch": 29.0,
       "step": 9903
     },
     {
-      "epoch": 29.282576866764277,
-      "grad_norm": 0.9301176071166992,
+      "epoch": 29.28,
       "learning_rate": 0.0001,
-      "loss": 0.1234,
+      "loss": 0.0884,
       "step": 10000
     },
     {
-      "epoch": 29.575402635431917,
-      "grad_norm": 0.8069124221801758,
+      "epoch": 29.58,
       "learning_rate": 0.0001,
-      "loss": 0.1287,
+      "loss": 0.0913,
       "step": 10100
     },
     {
-      "epoch": 29.86822840409956,
-      "grad_norm": 1.0093735456466675,
+      "epoch": 29.87,
       "learning_rate": 0.0001,
-      "loss": 0.134,
+      "loss": 0.0954,
       "step": 10200
     },
     {
       "epoch": 30.0,
-      "eval_accuracy": 0.41994871794871796,
-      "eval_loss": 6.661746025085449,
-      "eval_runtime": 4.8287,
-      "eval_samples_per_second": 103.548,
-      "eval_steps_per_second": 13.047,
+      "eval_accuracy": 0.43958974358974356,
+      "eval_loss": 5.223780632019043,
+      "eval_runtime": 4.8025,
+      "eval_samples_per_second": 104.112,
+      "eval_steps_per_second": 13.118,
       "step": 10245
     },
     {
       "epoch": 30.0,
-      "eval_bleu": 0.19612712097933296,
-      "eval_exact_match": 0.0,
-      "eval_prefix_exact_match": 0.028,
       "step": 10245
     },
     {
-      "epoch": 30.161054172767205,
-      "grad_norm": 0.965752899646759,
+      "epoch": 30.16,
       "learning_rate": 0.0001,
-      "loss": 0.1268,
+      "loss": 0.0913,
       "step": 10300
     },
     {
-      "epoch": 30.453879941434845,
-      "grad_norm": 0.9576678276062012,
+      "epoch": 30.45,
       "learning_rate": 0.0001,
-      "loss": 0.1241,
+      "loss": 0.0901,
       "step": 10400
     },
     {
-      "epoch": 30.74670571010249,
-      "grad_norm": 0.8335288763046265,
+      "epoch": 30.75,
       "learning_rate": 0.0001,
-      "loss": 0.1287,
+      "loss": 0.0931,
       "step": 10500
     },
     {
-      "epoch": 30.998535871156662,
-      "eval_accuracy": 0.4206666666666667,
-      "eval_loss": 6.624051094055176,
-      "eval_runtime": 5.0567,
-      "eval_samples_per_second": 98.879,
-      "eval_steps_per_second": 12.459,
+      "epoch": 31.0,
+      "eval_accuracy": 0.4402051282051282,
+      "eval_loss": 5.219459056854248,
+      "eval_runtime": 4.2447,
+      "eval_samples_per_second": 117.794,
+      "eval_steps_per_second": 14.842,
       "step": 10586
     },
     {
-      "epoch": 30.998535871156662,
-      "eval_bleu": 0.17063063536886222,
-      "eval_exact_match": 0.0,
-      "eval_prefix_exact_match": 0.026,
+      "epoch": 31.0,
       "step": 10586
     },
     {
-      "epoch": 31.039531478770133,
-      "grad_norm": 0.731296718120575,
+      "epoch": 31.04,
       "learning_rate": 0.0001,
-      "loss": 0.1327,
+      "loss": 0.0939,
       "step": 10600
     },
     {
-      "epoch": 31.332357247437773,
-      "grad_norm": 0.9108264446258545,
+      "epoch": 31.33,
       "learning_rate": 0.0001,
-      "loss": 0.1204,
+      "loss": 0.0879,
       "step": 10700
     },
     {
-      "epoch": 31.625183016105417,
-      "grad_norm": 0.9855822324752808,
+      "epoch": 31.63,
       "learning_rate": 0.0001,
-      "loss": 0.1259,
+      "loss": 0.0902,
       "step": 10800
     },
     {
-      "epoch": 31.91800878477306,
-      "grad_norm": 1.1106101274490356,
+      "epoch": 31.92,
       "learning_rate": 0.0001,
-      "loss": 0.1305,
+      "loss": 0.0923,
       "step": 10900
     },
     {
       "epoch": 32.0,
-      "eval_accuracy": 0.41994871794871796,
-      "eval_loss": 6.609373092651367,
-      "eval_runtime": 5.0029,
-      "eval_samples_per_second": 99.941,
-      "eval_steps_per_second": 12.593,
+      "eval_accuracy": 0.44066666666666665,
+      "eval_loss": 5.287062644958496,
+      "eval_runtime": 4.1005,
+      "eval_samples_per_second": 121.936,
+      "eval_steps_per_second": 15.364,
       "step": 10928
     },
     {
       "epoch": 32.0,
-      "eval_bleu": 0.1600123189131272,
-      "eval_exact_match": 0.002,
-      "eval_prefix_exact_match": 0.03,
       "step": 10928
     },
     {
-      "epoch": 32.210834553440705,
-      "grad_norm": 0.8511720895767212,
+      "epoch": 32.21,
       "learning_rate": 0.0001,
-      "loss": 0.1216,
+      "loss": 0.0881,
       "step": 11000
     },
     {
-      "epoch": 32.503660322108345,
-      "grad_norm": 1.0120123624801636,
+      "epoch": 32.5,
       "learning_rate": 0.0001,
-      "loss": 0.1225,
+      "loss": 0.0891,
       "step": 11100
     },
     {
-      "epoch": 32.796486090775986,
-      "grad_norm": 0.9636625051498413,
+      "epoch": 32.8,
       "learning_rate": 0.0001,
-      "loss": 0.1274,
+      "loss": 0.0911,
       "step": 11200
     },
     {
-      "epoch": 32.99853587115666,
-      "eval_accuracy": 0.4165128205128205,
-      "eval_loss": 6.682253360748291,
-      "eval_runtime": 5.2415,
-      "eval_samples_per_second": 95.392,
-      "eval_steps_per_second": 12.019,
+      "epoch": 33.0,
+      "eval_accuracy": 0.4391794871794872,
+      "eval_loss": 5.320067405700684,
+      "eval_runtime": 4.1991,
+      "eval_samples_per_second": 119.074,
+      "eval_steps_per_second": 15.003,
       "step": 11269
     },
     {
-      "epoch": 32.99853587115666,
-      "eval_bleu": 0.21084378912778606,
-      "eval_exact_match": 0.002,
-      "eval_prefix_exact_match": 0.022,
+      "epoch": 33.0,
       "step": 11269
     },
     {
-      "epoch": 33.08931185944363,
-      "grad_norm": 0.991923451423645,
+      "epoch": 33.09,
       "learning_rate": 0.0001,
-      "loss": 0.1271,
+      "loss": 0.0915,
       "step": 11300
     },
     {
-      "epoch": 33.38213762811127,
-      "grad_norm": 0.8520097136497498,
+      "epoch": 33.38,
       "learning_rate": 0.0001,
-      "loss": 0.1195,
+      "loss": 0.0873,
       "step": 11400
     },
     {
-      "epoch": 33.674963396778914,
-      "grad_norm": 1.1480833292007446,
+      "epoch": 33.67,
       "learning_rate": 0.0001,
-      "loss": 0.1242,
+      "loss": 0.0898,
       "step": 11500
     },
     {
-      "epoch": 33.96778916544656,
-      "grad_norm": 0.8655959367752075,
+      "epoch": 33.97,
       "learning_rate": 0.0001,
-      "loss": 0.1296,
+      "loss": 0.0934,
       "step": 11600
     },
     {
       "epoch": 34.0,
-      "eval_accuracy": 0.4195384615384615,
-      "eval_loss": 6.621041774749756,
-      "eval_runtime": 5.1712,
-      "eval_samples_per_second": 96.689,
-      "eval_steps_per_second": 12.183,
+      "eval_accuracy": 0.4387179487179487,
+      "eval_loss": 5.362847328186035,
+      "eval_runtime": 4.1975,
+      "eval_samples_per_second": 119.119,
+      "eval_steps_per_second": 15.009,
       "step": 11611
     },
     {
       "epoch": 34.0,
-      "eval_bleu": 0.2088991994544206,
-      "eval_exact_match": 0.0,
-      "eval_prefix_exact_match": 0.02,
       "step": 11611
     },
     {
-      "epoch": 34.2606149341142,
-      "grad_norm": 0.7514725923538208,
+      "epoch": 34.26,
       "learning_rate": 0.0001,
-      "loss": 0.1199,
+      "loss": 0.086,
       "step": 11700
     },
     {
-      "epoch": 34.55344070278184,
-      "grad_norm": 0.9057220220565796,
+      "epoch": 34.55,
       "learning_rate": 0.0001,
-      "loss": 0.121,
+      "loss": 0.0883,
       "step": 11800
     },
     {
-      "epoch": 34.84626647144949,
-      "grad_norm": 0.8322867155075073,
+      "epoch": 34.85,
       "learning_rate": 0.0001,
-      "loss": 0.1271,
+      "loss": 0.091,
       "step": 11900
     },
     {
-      "epoch": 34.99853587115666,
-      "eval_accuracy": 0.41846153846153844,
-      "eval_loss": 6.704195976257324,
-      "eval_runtime": 5.1841,
-      "eval_samples_per_second": 96.448,
-      "eval_steps_per_second": 12.153,
+      "epoch": 35.0,
+      "eval_accuracy": 0.438974358974359,
+      "eval_loss": 5.319708347320557,
+      "eval_runtime": 4.201,
+      "eval_samples_per_second": 119.019,
+      "eval_steps_per_second": 14.996,
       "step": 11952
     },
     {
-      "epoch": 34.99853587115666,
-      "eval_bleu": 0.17311282679501735,
-      "eval_exact_match": 0.0,
-      "eval_prefix_exact_match": 0.022,
+      "epoch": 35.0,
       "step": 11952
     },
     {
-      "epoch": 35.13909224011713,
-      "grad_norm": 0.9303777813911438,
+      "epoch": 35.14,
       "learning_rate": 0.0001,
-      "loss": 0.1231,
+      "loss": 0.0897,
       "step": 12000
     },
     {
-      "epoch": 35.43191800878477,
-      "grad_norm": 0.8847092986106873,
+      "epoch": 35.43,
       "learning_rate": 0.0001,
-      "loss": 0.1201,
+      "loss": 0.0876,
       "step": 12100
     },
     {
-      "epoch": 35.72474377745242,
-      "grad_norm": 1.0509458780288696,
+      "epoch": 35.72,
       "learning_rate": 0.0001,
-      "loss": 0.1239,
+      "loss": 0.0902,
       "step": 12200
     },
     {
       "epoch": 36.0,
-      "eval_accuracy": 0.42035897435897435,
-      "eval_loss": 6.601570129394531,
-      "eval_runtime": 4.9227,
-      "eval_samples_per_second": 101.571,
-      "eval_steps_per_second": 12.798,
+      "eval_accuracy": 0.4391282051282051,
+      "eval_loss": 5.186809539794922,
+      "eval_runtime": 4.2,
+      "eval_samples_per_second": 119.046,
+      "eval_steps_per_second": 15.0,
       "step": 12294
     },
     {
       "epoch": 36.0,
-      "eval_bleu": 0.19246146326233543,
-      "eval_exact_match": 0.0,
-      "eval_prefix_exact_match": 0.022,
       "step": 12294
     },
     {
-      "epoch": 36.01756954612006,
-      "grad_norm": 1.0885076522827148,
+      "epoch": 36.02,
       "learning_rate": 0.0001,
-      "loss": 0.1262,
+      "loss": 0.0931,
       "step": 12300
     },
     {
-      "epoch": 36.3103953147877,
-      "grad_norm": 0.8121168613433838,
+      "epoch": 36.31,
       "learning_rate": 0.0001,
-      "loss": 0.1153,
+      "loss": 0.0857,
       "step": 12400
     },
     {
-      "epoch": 36.603221083455345,
-      "grad_norm": 0.7476085424423218,
+      "epoch": 36.6,
       "learning_rate": 0.0001,
-      "loss": 0.1212,
+      "loss": 0.0894,
       "step": 12500
     },
     {
-      "epoch": 36.896046852122986,
-      "grad_norm": 1.0950144529342651,
+      "epoch": 36.9,
       "learning_rate": 0.0001,
-      "loss": 0.1263,
+      "loss": 0.0916,
       "step": 12600
     },
     {
-      "epoch": 36.99853587115666,
-      "eval_accuracy": 0.4195384615384615,
-      "eval_loss": 6.573636531829834,
-      "eval_runtime": 4.8706,
-      "eval_samples_per_second": 102.657,
-      "eval_steps_per_second": 12.935,
+      "epoch": 37.0,
+      "eval_accuracy": 0.44235897435897437,
+      "eval_loss": 5.122660160064697,
+      "eval_runtime": 4.4215,
+      "eval_samples_per_second": 113.083,
+      "eval_steps_per_second": 14.248,
       "step": 12635
     },
     {
-      "epoch": 36.99853587115666,
-      "eval_bleu": 0.11460835767059586,
-      "eval_exact_match": 0.0,
-      "eval_prefix_exact_match": 0.022,
+      "epoch": 37.0,
       "step": 12635
     },
     {
-      "epoch": 37.18887262079063,
-      "grad_norm": 1.199575662612915,
+      "epoch": 37.19,
       "learning_rate": 0.0001,
-      "loss": 0.1187,
+      "loss": 0.0894,
       "step": 12700
     },
     {
-      "epoch": 37.48169838945827,
-      "grad_norm": 0.6631502509117126,
+      "epoch": 37.48,
       "learning_rate": 0.0001,
-      "loss": 0.1189,
+      "loss": 0.0888,
       "step": 12800
     },
     {
-      "epoch": 37.774524158125914,
-      "grad_norm": 1.0126887559890747,
+      "epoch": 37.77,
       "learning_rate": 0.0001,
-      "loss": 0.1234,
+      "loss": 0.0905,
       "step": 12900
     },
     {
       "epoch": 38.0,
-      "eval_accuracy": 0.4169230769230769,
-      "eval_loss": 6.609395503997803,
-      "eval_runtime": 5.0108,
-      "eval_samples_per_second": 99.784,
-      "eval_steps_per_second": 12.573,
+      "eval_accuracy": 0.43666666666666665,
+      "eval_loss": 5.221393585205078,
+      "eval_runtime": 4.6657,
+      "eval_samples_per_second": 107.164,
+      "eval_steps_per_second": 13.503,
       "step": 12977
     },
     {
       "epoch": 38.0,
-      "eval_bleu": 0.13428219944335265,
-      "eval_exact_match": 0.002,
-      "eval_prefix_exact_match": 0.024,
       "step": 12977
     },
     {
-      "epoch": 38.06734992679356,
-      "grad_norm": 0.7301933169364929,
+      "epoch": 38.07,
       "learning_rate": 0.0001,
-      "loss": 0.1234,
+      "loss": 0.0906,
       "step": 13000
     },
     {
-      "epoch": 38.3601756954612,
-      "grad_norm": 0.7770780920982361,
+      "epoch": 38.36,
       "learning_rate": 0.0001,
-      "loss": 0.1163,
+      "loss": 0.0857,
       "step": 13100
     },
     {
-      "epoch": 38.65300146412884,
-      "grad_norm": 0.768286406993866,
+      "epoch": 38.65,
       "learning_rate": 0.0001,
-      "loss": 0.1195,
+      "loss": 0.0884,
       "step": 13200
     },
     {
-      "epoch": 38.94582723279649,
-      "grad_norm": 0.8422756195068359,
+      "epoch": 38.95,
       "learning_rate": 0.0001,
-      "loss": 0.1236,
+      "loss": 0.0907,
       "step": 13300
     },
     {
-      "epoch": 38.99853587115666,
-      "eval_accuracy": 0.41512820512820514,
-      "eval_loss": 6.6394548416137695,
-      "eval_runtime": 4.9439,
-      "eval_samples_per_second": 101.135,
-      "eval_steps_per_second": 12.743,
+      "epoch": 39.0,
+      "eval_accuracy": 0.4411794871794872,
+      "eval_loss": 5.241151332855225,
+      "eval_runtime": 4.1927,
+      "eval_samples_per_second": 119.255,
+      "eval_steps_per_second": 15.026,
       "step": 13318
     },
     {
-      "epoch": 38.99853587115666,
-      "eval_bleu": 0.17940902220327,
+      "epoch": 39.0,
+      "eval_bleu": 0.05134364405270662,
       "eval_exact_match": 0.0,
-      "eval_prefix_exact_match": 0.022,
+      "eval_prefix_exact_match": 0.006,
       "step": 13318
     },
     {
-      "epoch": 39.23865300146413,
-      "grad_norm": 0.6590162515640259,
+      "epoch": 39.24,
       "learning_rate": 0.0001,
-      "loss": 0.1166,
+      "loss": 0.0862,
       "step": 13400
     },
     {
-      "epoch": 39.53147877013177,
-      "grad_norm": 0.6215115785598755,
+      "epoch": 39.53,
       "learning_rate": 0.0001,
-      "loss": 0.1181,
+      "loss": 0.0864,
       "step": 13500
     },
     {
-      "epoch": 39.82430453879942,
-      "grad_norm": 1.042077660560608,
+      "epoch": 39.82,
       "learning_rate": 0.0001,
-      "loss": 0.1211,
+      "loss": 0.0883,
       "step": 13600
     },
     {
       "epoch": 40.0,
-      "eval_accuracy": 0.41317948717948716,
-      "eval_loss": 6.6604390144348145,
-      "eval_runtime": 4.9405,
-      "eval_samples_per_second": 101.205,
-      "eval_steps_per_second": 12.752,
+      "eval_accuracy": 0.43948717948717947,
+      "eval_loss": 5.301468372344971,
+      "eval_runtime": 4.0986,
+      "eval_samples_per_second": 121.992,
+      "eval_steps_per_second": 15.371,
       "step": 13660
     },
     {
       "epoch": 40.0,
-      "eval_bleu": 0.18494582369403326,
-      "eval_exact_match": 0.0,
-      "eval_prefix_exact_match": 0.02,
       "step": 13660
     },
     {
-      "epoch": 40.11713030746706,
-      "grad_norm": 0.6930384039878845,
+      "epoch": 40.12,
       "learning_rate": 0.0001,
-      "loss": 0.1194,
+      "loss": 0.0864,
       "step": 13700
     },
     {
-      "epoch": 40.4099560761347,
-      "grad_norm": 0.8434835076332092,
+      "epoch": 40.41,
       "learning_rate": 0.0001,
-      "loss": 0.1151,
+      "loss": 0.0838,
       "step": 13800
     },
     {
-      "epoch": 40.702781844802345,
-      "grad_norm": 0.6417421102523804,
+      "epoch": 40.7,
       "learning_rate": 0.0001,
-      "loss": 0.1196,
+      "loss": 0.0863,
       "step": 13900
     },
     {
-      "epoch": 40.995607613469986,
-      "grad_norm": 0.7239630222320557,
+      "epoch": 41.0,
       "learning_rate": 0.0001,
-      "loss": 0.1235,
+      "loss": 0.0892,
       "step": 14000
     },
     {
-      "epoch": 40.99853587115666,
-      "eval_accuracy": 0.41717948717948716,
-      "eval_loss": 6.709766864776611,
-      "eval_runtime": 5.1408,
-      "eval_samples_per_second": 97.262,
-      "eval_steps_per_second": 12.255,
+      "epoch": 41.0,
+      "eval_accuracy": 0.43923076923076926,
+      "eval_loss": 5.2815632820129395,
+      "eval_runtime": 4.4744,
+      "eval_samples_per_second": 111.747,
+      "eval_steps_per_second": 14.08,
       "step": 14001
     },
     {
-      "epoch": 40.99853587115666,
-      "eval_bleu": 0.17857535751716297,
-      "eval_exact_match": 0.0,
-      "eval_prefix_exact_match": 0.02,
+      "epoch": 41.0,
       "step": 14001
     }
   ],
   "logging_steps": 100,
   "max_steps": 17050,
-  "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": 1.2023289469479158e+18,
-  "train_batch_size": 1,
+  "total_flos": 2.7727205637879235e+18,
   "trial_name": null,
   "trial_params": null
 }

checkpoint-14001/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:7e63e74298c414f7364874e60281a380506bddfac466373e2426a740b4eb5594
-size 5304
+oid sha256:3c60050af374bb2df141f4673c8141d59d83a5bd892d35ffa56ce8af555309ff
+size 4728