End of training

README.md

@@ -4,6 +4,7 @@ license: apache-2.0
 base_model: Qwen/Qwen2.5-7B-Instruct
 tags:
 - llama-factory
+- full
 - generated_from_trainer
 model-index:
 - name: global_batchsize_1024_laradjusted4
@@ -15,7 +16,7 @@ should probably proofread and complete it, then remove this comment. -->
 
 # global_batchsize_1024_laradjusted4
 
-This model is a fine-tuned version of [Qwen/Qwen2.5-7B-Instruct](https://huggingface.co/Qwen/Qwen2.5-7B-Instruct) on an unknown dataset.
+This model is a fine-tuned version of [Qwen/Qwen2.5-7B-Instruct](https://huggingface.co/Qwen/Qwen2.5-7B-Instruct) on the open-thoughts/OpenThoughts-114k dataset.
 
 ## Model description
 

all_results.json

@@ -0,0 +1,8 @@
+{
+    "epoch": 2.9943851768669285,
+    "total_flos": 4607052149424128.0,
+    "train_loss": 0.4725978755736136,
+    "train_runtime": 115789.2905,
+    "train_samples_per_second": 2.953,
+    "train_steps_per_second": 0.003
+}

train_results.json

@@ -0,0 +1,8 @@
+{
+    "epoch": 2.9943851768669285,
+    "total_flos": 4607052149424128.0,
+    "train_loss": 0.4725978755736136,
+    "train_runtime": 115789.2905,
+    "train_samples_per_second": 2.953,
+    "train_steps_per_second": 0.003
+}

trainer_state.json

@@ -0,0 +1,2373 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.9943851768669285,
+  "eval_steps": 500,
+  "global_step": 333,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.008983717012914094,
+      "grad_norm": 5.839086702019255,
+      "learning_rate": 4.705882352941177e-06,
+      "loss": 0.851,
+      "step": 1
+    },
+    {
+      "epoch": 0.017967434025828188,
+      "grad_norm": 5.796604506548498,
+      "learning_rate": 9.411764705882354e-06,
+      "loss": 0.8589,
+      "step": 2
+    },
+    {
+      "epoch": 0.02695115103874228,
+      "grad_norm": 4.339124227516403,
+      "learning_rate": 1.4117647058823532e-05,
+      "loss": 0.8175,
+      "step": 3
+    },
+    {
+      "epoch": 0.035934868051656375,
+      "grad_norm": 2.117331281643237,
+      "learning_rate": 1.8823529411764708e-05,
+      "loss": 0.7666,
+      "step": 4
+    },
+    {
+      "epoch": 0.044918585064570464,
+      "grad_norm": 6.08947531782942,
+      "learning_rate": 2.3529411764705884e-05,
+      "loss": 0.7997,
+      "step": 5
+    },
+    {
+      "epoch": 0.05390230207748456,
+      "grad_norm": 11.437654369979397,
+      "learning_rate": 2.8235294117647063e-05,
+      "loss": 0.8658,
+      "step": 6
+    },
+    {
+      "epoch": 0.06288601909039865,
+      "grad_norm": 6.230659224656504,
+      "learning_rate": 3.294117647058824e-05,
+      "loss": 0.7742,
+      "step": 7
+    },
+    {
+      "epoch": 0.07186973610331275,
+      "grad_norm": 4.258899179934184,
+      "learning_rate": 3.7647058823529415e-05,
+      "loss": 0.7478,
+      "step": 8
+    },
+    {
+      "epoch": 0.08085345311622684,
+      "grad_norm": 3.163611777765137,
+      "learning_rate": 4.235294117647059e-05,
+      "loss": 0.7167,
+      "step": 9
+    },
+    {
+      "epoch": 0.08983717012914093,
+      "grad_norm": 2.4646084609310077,
+      "learning_rate": 4.705882352941177e-05,
+      "loss": 0.6839,
+      "step": 10
+    },
+    {
+      "epoch": 0.09882088714205503,
+      "grad_norm": 2.1835077872968514,
+      "learning_rate": 5.176470588235295e-05,
+      "loss": 0.6594,
+      "step": 11
+    },
+    {
+      "epoch": 0.10780460415496912,
+      "grad_norm": 1.956338905834903,
+      "learning_rate": 5.6470588235294126e-05,
+      "loss": 0.6447,
+      "step": 12
+    },
+    {
+      "epoch": 0.11678832116788321,
+      "grad_norm": 1.2530266966782506,
+      "learning_rate": 6.117647058823529e-05,
+      "loss": 0.6225,
+      "step": 13
+    },
+    {
+      "epoch": 0.1257720381807973,
+      "grad_norm": 2.746116145636045,
+      "learning_rate": 6.588235294117648e-05,
+      "loss": 0.6409,
+      "step": 14
+    },
+    {
+      "epoch": 0.13475575519371139,
+      "grad_norm": 1.836377286540359,
+      "learning_rate": 7.058823529411765e-05,
+      "loss": 0.6265,
+      "step": 15
+    },
+    {
+      "epoch": 0.1437394722066255,
+      "grad_norm": 2.81566991788529,
+      "learning_rate": 7.529411764705883e-05,
+      "loss": 0.6182,
+      "step": 16
+    },
+    {
+      "epoch": 0.1527231892195396,
+      "grad_norm": 1.9484267685166876,
+      "learning_rate": 8e-05,
+      "loss": 0.6088,
+      "step": 17
+    },
+    {
+      "epoch": 0.16170690623245368,
+      "grad_norm": 1.4245865160208993,
+      "learning_rate": 8.470588235294118e-05,
+      "loss": 0.5934,
+      "step": 18
+    },
+    {
+      "epoch": 0.17069062324536777,
+      "grad_norm": 2.326255459435556,
+      "learning_rate": 8.941176470588236e-05,
+      "loss": 0.6025,
+      "step": 19
+    },
+    {
+      "epoch": 0.17967434025828186,
+      "grad_norm": 1.5418881568160665,
+      "learning_rate": 9.411764705882353e-05,
+      "loss": 0.5952,
+      "step": 20
+    },
+    {
+      "epoch": 0.18865805727119594,
+      "grad_norm": 1.8971846222744186,
+      "learning_rate": 9.882352941176472e-05,
+      "loss": 0.5812,
+      "step": 21
+    },
+    {
+      "epoch": 0.19764177428411006,
+      "grad_norm": 1.200246618699023,
+      "learning_rate": 0.0001035294117647059,
+      "loss": 0.5796,
+      "step": 22
+    },
+    {
+      "epoch": 0.20662549129702415,
+      "grad_norm": 1.541626574943513,
+      "learning_rate": 0.00010823529411764708,
+      "loss": 0.5735,
+      "step": 23
+    },
+    {
+      "epoch": 0.21560920830993824,
+      "grad_norm": 2.8830857335375746,
+      "learning_rate": 0.00011294117647058825,
+      "loss": 0.5932,
+      "step": 24
+    },
+    {
+      "epoch": 0.22459292532285233,
+      "grad_norm": 1.2631947275776292,
+      "learning_rate": 0.00011764705882352943,
+      "loss": 0.5768,
+      "step": 25
+    },
+    {
+      "epoch": 0.23357664233576642,
+      "grad_norm": 3.7808866136207353,
+      "learning_rate": 0.00012235294117647058,
+      "loss": 0.6315,
+      "step": 26
+    },
+    {
+      "epoch": 0.2425603593486805,
+      "grad_norm": 2.3059028840452847,
+      "learning_rate": 0.00012705882352941178,
+      "loss": 0.6138,
+      "step": 27
+    },
+    {
+      "epoch": 0.2515440763615946,
+      "grad_norm": 3.060917512545427,
+      "learning_rate": 0.00013176470588235296,
+      "loss": 0.5971,
+      "step": 28
+    },
+    {
+      "epoch": 0.2605277933745087,
+      "grad_norm": 2.2144909817302425,
+      "learning_rate": 0.00013647058823529413,
+      "loss": 0.6019,
+      "step": 29
+    },
+    {
+      "epoch": 0.26951151038742277,
+      "grad_norm": 10.523000576466588,
+      "learning_rate": 0.0001411764705882353,
+      "loss": 0.6902,
+      "step": 30
+    },
+    {
+      "epoch": 0.2784952274003369,
+      "grad_norm": 3733.897541065198,
+      "learning_rate": 0.00014588235294117648,
+      "loss": 2.1691,
+      "step": 31
+    },
+    {
+      "epoch": 0.287478944413251,
+      "grad_norm": 152.85200430325233,
+      "learning_rate": 0.00015058823529411766,
+      "loss": 2.3024,
+      "step": 32
+    },
+    {
+      "epoch": 0.29646266142616506,
+      "grad_norm": 5.795238259929523,
+      "learning_rate": 0.00015529411764705884,
+      "loss": 0.9312,
+      "step": 33
+    },
+    {
+      "epoch": 0.3054463784390792,
+      "grad_norm": 2.5238561333157286,
+      "learning_rate": 0.00016,
+      "loss": 0.7014,
+      "step": 34
+    },
+    {
+      "epoch": 0.31443009545199324,
+      "grad_norm": 3.227790922816581,
+      "learning_rate": 0.0001599955841596387,
+      "loss": 0.6914,
+      "step": 35
+    },
+    {
+      "epoch": 0.32341381246490736,
+      "grad_norm": 2.590941539337386,
+      "learning_rate": 0.00015998233712604597,
+      "loss": 0.716,
+      "step": 36
+    },
+    {
+      "epoch": 0.3323975294778215,
+      "grad_norm": 9.464462354232134,
+      "learning_rate": 0.00015996026036164143,
+      "loss": 1.163,
+      "step": 37
+    },
+    {
+      "epoch": 0.34138124649073553,
+      "grad_norm": 9.4056400087482,
+      "learning_rate": 0.00015992935630361176,
+      "loss": 0.9691,
+      "step": 38
+    },
+    {
+      "epoch": 0.35036496350364965,
+      "grad_norm": 3.229822785794876,
+      "learning_rate": 0.00015988962836364164,
+      "loss": 0.8032,
+      "step": 39
+    },
+    {
+      "epoch": 0.3593486805165637,
+      "grad_norm": 2.4264007116828594,
+      "learning_rate": 0.00015984108092753708,
+      "loss": 0.7292,
+      "step": 40
+    },
+    {
+      "epoch": 0.36833239752947783,
+      "grad_norm": 5.595688226616298,
+      "learning_rate": 0.0001597837193547413,
+      "loss": 0.681,
+      "step": 41
+    },
+    {
+      "epoch": 0.3773161145423919,
+      "grad_norm": 1.6476213464085256,
+      "learning_rate": 0.00015971754997774298,
+      "loss": 0.676,
+      "step": 42
+    },
+    {
+      "epoch": 0.386299831555306,
+      "grad_norm": 11.92511194308152,
+      "learning_rate": 0.0001596425801013773,
+      "loss": 0.6598,
+      "step": 43
+    },
+    {
+      "epoch": 0.3952835485682201,
+      "grad_norm": 3.9213311829331734,
+      "learning_rate": 0.00015955881800201933,
+      "loss": 0.7401,
+      "step": 44
+    },
+    {
+      "epoch": 0.4042672655811342,
+      "grad_norm": 2.0354638536769767,
+      "learning_rate": 0.00015946627292667064,
+      "loss": 0.6612,
+      "step": 45
+    },
+    {
+      "epoch": 0.4132509825940483,
+      "grad_norm": 1.886963956827035,
+      "learning_rate": 0.00015936495509193816,
+      "loss": 0.6393,
+      "step": 46
+    },
+    {
+      "epoch": 0.42223469960696236,
+      "grad_norm": 1.6161025496299593,
+      "learning_rate": 0.00015925487568290645,
+      "loss": 0.6135,
+      "step": 47
+    },
+    {
+      "epoch": 0.4312184166198765,
+      "grad_norm": 1.3094812966970897,
+      "learning_rate": 0.00015913604685190304,
+      "loss": 0.5968,
+      "step": 48
+    },
+    {
+      "epoch": 0.4402021336327906,
+      "grad_norm": 1.0856830435724436,
+      "learning_rate": 0.00015900848171715654,
+      "loss": 0.5947,
+      "step": 49
+    },
+    {
+      "epoch": 0.44918585064570465,
+      "grad_norm": 3.133660662863731,
+      "learning_rate": 0.00015887219436134874,
+      "loss": 0.5986,
+      "step": 50
+    },
+    {
+      "epoch": 0.45816956765861877,
+      "grad_norm": 1.629031015427086,
+      "learning_rate": 0.00015872719983005986,
+      "loss": 0.6136,
+      "step": 51
+    },
+    {
+      "epoch": 0.46715328467153283,
+      "grad_norm": 1.2979437140842698,
+      "learning_rate": 0.0001585735141301074,
+      "loss": 0.6008,
+      "step": 52
+    },
+    {
+      "epoch": 0.47613700168444695,
+      "grad_norm": 2.022435241771787,
+      "learning_rate": 0.00015841115422777935,
+      "loss": 0.599,
+      "step": 53
+    },
+    {
+      "epoch": 0.485120718697361,
+      "grad_norm": 1.0176598957138148,
+      "learning_rate": 0.0001582401380469609,
+      "loss": 0.5911,
+      "step": 54
+    },
+    {
+      "epoch": 0.4941044357102751,
+      "grad_norm": 1.340446281009079,
+      "learning_rate": 0.00015806048446715594,
+      "loss": 0.6018,
+      "step": 55
+    },
+    {
+      "epoch": 0.5030881527231892,
+      "grad_norm": 1.1781671057543985,
+      "learning_rate": 0.00015787221332140268,
+      "loss": 0.5775,
+      "step": 56
+    },
+    {
+      "epoch": 0.5120718697361033,
+      "grad_norm": 0.8746372899225068,
+      "learning_rate": 0.00015767534539408418,
+      "loss": 0.5764,
+      "step": 57
+    },
+    {
+      "epoch": 0.5210555867490174,
+      "grad_norm": 0.8619016762831145,
+      "learning_rate": 0.00015746990241863394,
+      "loss": 0.5621,
+      "step": 58
+    },
+    {
+      "epoch": 0.5300393037619315,
+      "grad_norm": 0.8908249631898915,
+      "learning_rate": 0.0001572559070751366,
+      "loss": 0.5626,
+      "step": 59
+    },
+    {
+      "epoch": 0.5390230207748455,
+      "grad_norm": 0.9941730669724362,
+      "learning_rate": 0.00015703338298782396,
+      "loss": 0.5587,
+      "step": 60
+    },
+    {
+      "epoch": 0.5480067377877597,
+      "grad_norm": 1.075188528305704,
+      "learning_rate": 0.0001568023547224672,
+      "loss": 0.5524,
+      "step": 61
+    },
+    {
+      "epoch": 0.5569904548006738,
+      "grad_norm": 0.8331471695655838,
+      "learning_rate": 0.00015656284778366479,
+      "loss": 0.5383,
+      "step": 62
+    },
+    {
+      "epoch": 0.5659741718135879,
+      "grad_norm": 0.7654032131505463,
+      "learning_rate": 0.00015631488861202688,
+      "loss": 0.538,
+      "step": 63
+    },
+    {
+      "epoch": 0.574957888826502,
+      "grad_norm": 0.8860510131733896,
+      "learning_rate": 0.00015605850458125646,
+      "loss": 0.531,
+      "step": 64
+    },
+    {
+      "epoch": 0.583941605839416,
+      "grad_norm": 0.6831724579134854,
+      "learning_rate": 0.0001557937239951273,
+      "loss": 0.5314,
+      "step": 65
+    },
+    {
+      "epoch": 0.5929253228523301,
+      "grad_norm": 0.6778833496972871,
+      "learning_rate": 0.00015552057608435936,
+      "loss": 0.5291,
+      "step": 66
+    },
+    {
+      "epoch": 0.6019090398652442,
+      "grad_norm": 0.7986870903265276,
+      "learning_rate": 0.00015523909100339186,
+      "loss": 0.531,
+      "step": 67
+    },
+    {
+      "epoch": 0.6108927568781584,
+      "grad_norm": 0.7775755149998839,
+      "learning_rate": 0.0001549492998270544,
+      "loss": 0.5297,
+      "step": 68
+    },
+    {
+      "epoch": 0.6198764738910725,
+      "grad_norm": 0.4942435514859441,
+      "learning_rate": 0.00015465123454713622,
+      "loss": 0.5166,
+      "step": 69
+    },
+    {
+      "epoch": 0.6288601909039865,
+      "grad_norm": 0.5343768774089758,
+      "learning_rate": 0.0001543449280688547,
+      "loss": 0.5149,
+      "step": 70
+    },
+    {
+      "epoch": 0.6378439079169006,
+      "grad_norm": 0.5061205195595769,
+      "learning_rate": 0.00015403041420722257,
+      "loss": 0.515,
+      "step": 71
+    },
+    {
+      "epoch": 0.6468276249298147,
+      "grad_norm": 0.46967339805307057,
+      "learning_rate": 0.00015370772768331497,
+      "loss": 0.5026,
+      "step": 72
+    },
+    {
+      "epoch": 0.6558113419427288,
+      "grad_norm": 0.4141303041477235,
+      "learning_rate": 0.00015337690412043625,
+      "loss": 0.5137,
+      "step": 73
+    },
+    {
+      "epoch": 0.664795058955643,
+      "grad_norm": 0.45219045262379115,
+      "learning_rate": 0.0001530379800401875,
+      "loss": 0.5059,
+      "step": 74
+    },
+    {
+      "epoch": 0.673778775968557,
+      "grad_norm": 0.3940676548155913,
+      "learning_rate": 0.0001526909928584345,
+      "loss": 0.5036,
+      "step": 75
+    },
+    {
+      "epoch": 0.6827624929814711,
+      "grad_norm": 0.5141391546743134,
+      "learning_rate": 0.00015233598088117733,
+      "loss": 0.5163,
+      "step": 76
+    },
+    {
+      "epoch": 0.6917462099943852,
+      "grad_norm": 0.6219020292862112,
+      "learning_rate": 0.00015197298330032146,
+      "loss": 0.502,
+      "step": 77
+    },
+    {
+      "epoch": 0.7007299270072993,
+      "grad_norm": 0.5118275087823705,
+      "learning_rate": 0.000151602040189351,
+      "loss": 0.5015,
+      "step": 78
+    },
+    {
+      "epoch": 0.7097136440202133,
+      "grad_norm": 0.6008412729706559,
+      "learning_rate": 0.00015122319249890513,
+      "loss": 0.4927,
+      "step": 79
+    },
+    {
+      "epoch": 0.7186973610331274,
+      "grad_norm": 0.9550462875054769,
+      "learning_rate": 0.00015083648205225679,
+      "loss": 0.5102,
+      "step": 80
+    },
+    {
+      "epoch": 0.7276810780460415,
+      "grad_norm": 1.4871365184883178,
+      "learning_rate": 0.000150441951540696,
+      "loss": 0.5042,
+      "step": 81
+    },
+    {
+      "epoch": 0.7366647950589557,
+      "grad_norm": 0.4668622982117298,
+      "learning_rate": 0.00015003964451881666,
+      "loss": 0.481,
+      "step": 82
+    },
+    {
+      "epoch": 0.7456485120718698,
+      "grad_norm": 0.9148709606250665,
+      "learning_rate": 0.0001496296053997084,
+      "loss": 0.5007,
+      "step": 83
+    },
+    {
+      "epoch": 0.7546322290847838,
+      "grad_norm": 1.279060221546284,
+      "learning_rate": 0.00014921187945005348,
+      "loss": 0.4993,
+      "step": 84
+    },
+    {
+      "epoch": 0.7636159460976979,
+      "grad_norm": 0.616746320687276,
+      "learning_rate": 0.00014878651278512966,
+      "loss": 0.4903,
+      "step": 85
+    },
+    {
+      "epoch": 0.772599663110612,
+      "grad_norm": 0.9313866786660758,
+      "learning_rate": 0.0001483535523637191,
+      "loss": 0.4951,
+      "step": 86
+    },
+    {
+      "epoch": 0.7815833801235261,
+      "grad_norm": 0.6267949258981602,
+      "learning_rate": 0.0001479130459829244,
+      "loss": 0.4849,
+      "step": 87
+    },
+    {
+      "epoch": 0.7905670971364402,
+      "grad_norm": 0.7584425730651909,
+      "learning_rate": 0.00014746504227289192,
+      "loss": 0.4986,
+      "step": 88
+    },
+    {
+      "epoch": 0.7995508141493542,
+      "grad_norm": 0.5681276511497075,
+      "learning_rate": 0.00014700959069144332,
+      "loss": 0.4785,
+      "step": 89
+    },
+    {
+      "epoch": 0.8085345311622684,
+      "grad_norm": 0.5906735363635705,
+      "learning_rate": 0.0001465467415186155,
+      "loss": 0.4907,
+      "step": 90
+    },
+    {
+      "epoch": 0.8175182481751825,
+      "grad_norm": 0.7053181757428566,
+      "learning_rate": 0.0001460765458511099,
+      "loss": 0.4912,
+      "step": 91
+    },
+    {
+      "epoch": 0.8265019651880966,
+      "grad_norm": 0.6751343616215291,
+      "learning_rate": 0.00014559905559665169,
+      "loss": 0.4844,
+      "step": 92
+    },
+    {
+      "epoch": 0.8354856822010107,
+      "grad_norm": 0.6487198602578013,
+      "learning_rate": 0.00014511432346825932,
+      "loss": 0.4801,
+      "step": 93
+    },
+    {
+      "epoch": 0.8444693992139247,
+      "grad_norm": 0.323668455749265,
+      "learning_rate": 0.0001446224029784252,
+      "loss": 0.4793,
+      "step": 94
+    },
+    {
+      "epoch": 0.8534531162268388,
+      "grad_norm": 0.5861018094486455,
+      "learning_rate": 0.0001441233484332082,
+      "loss": 0.4822,
+      "step": 95
+    },
+    {
+      "epoch": 0.862436833239753,
+      "grad_norm": 0.4370056395937098,
+      "learning_rate": 0.00014361721492623834,
+      "loss": 0.4878,
+      "step": 96
+    },
+    {
+      "epoch": 0.8714205502526671,
+      "grad_norm": 0.5421346987411235,
+      "learning_rate": 0.00014310405833263486,
+      "loss": 0.4872,
+      "step": 97
+    },
+    {
+      "epoch": 0.8804042672655812,
+      "grad_norm": 0.5238036281016801,
+      "learning_rate": 0.00014258393530283773,
+      "loss": 0.47,
+      "step": 98
+    },
+    {
+      "epoch": 0.8893879842784952,
+      "grad_norm": 0.557528609344473,
+      "learning_rate": 0.0001420569032563536,
+      "loss": 0.4859,
+      "step": 99
+    },
+    {
+      "epoch": 0.8983717012914093,
+      "grad_norm": 0.7040485273548837,
+      "learning_rate": 0.00014152302037541707,
+      "loss": 0.4747,
+      "step": 100
+    },
+    {
+      "epoch": 0.9073554183043234,
+      "grad_norm": 0.6427583637387018,
+      "learning_rate": 0.00014098234559856748,
+      "loss": 0.487,
+      "step": 101
+    },
+    {
+      "epoch": 0.9163391353172375,
+      "grad_norm": 0.5447220351132417,
+      "learning_rate": 0.00014043493861414235,
+      "loss": 0.4721,
+      "step": 102
+    },
+    {
+      "epoch": 0.9253228523301515,
+      "grad_norm": 0.5093866707759475,
+      "learning_rate": 0.0001398808598536881,
+      "loss": 0.4838,
+      "step": 103
+    },
+    {
+      "epoch": 0.9343065693430657,
+      "grad_norm": 0.5848991205095823,
+      "learning_rate": 0.00013932017048528858,
+      "loss": 0.4732,
+      "step": 104
+    },
+    {
+      "epoch": 0.9432902863559798,
+      "grad_norm": 0.46406379014333665,
+      "learning_rate": 0.00013875293240681237,
+      "loss": 0.4829,
+      "step": 105
+    },
+    {
+      "epoch": 0.9522740033688939,
+      "grad_norm": 0.3855555200229203,
+      "learning_rate": 0.00013817920823907946,
+      "loss": 0.4699,
+      "step": 106
+    },
+    {
+      "epoch": 0.961257720381808,
+      "grad_norm": 0.47252048758301546,
+      "learning_rate": 0.0001375990613189483,
+      "loss": 0.4709,
+      "step": 107
+    },
+    {
+      "epoch": 0.970241437394722,
+      "grad_norm": 0.5128650342882652,
+      "learning_rate": 0.00013701255569232356,
+      "loss": 0.4625,
+      "step": 108
+    },
+    {
+      "epoch": 0.9792251544076361,
+      "grad_norm": 0.40600999436688917,
+      "learning_rate": 0.0001364197561070856,
+      "loss": 0.4696,
+      "step": 109
+    },
+    {
+      "epoch": 0.9882088714205502,
+      "grad_norm": 0.44134356238524264,
+      "learning_rate": 0.00013582072800594284,
+      "loss": 0.4775,
+      "step": 110
+    },
+    {
+      "epoch": 0.9971925884334644,
+      "grad_norm": 0.5024734834453151,
+      "learning_rate": 0.00013521553751920693,
+      "loss": 0.4707,
+      "step": 111
+    },
+    {
+      "epoch": 1.0075800112296462,
+      "grad_norm": 1.0362269730450384,
+      "learning_rate": 0.00013460425145749236,
+      "loss": 0.8319,
+      "step": 112
+    },
+    {
+      "epoch": 1.0165637282425604,
+      "grad_norm": 1.7796385504660885,
+      "learning_rate": 0.00013398693730434062,
+      "loss": 0.4451,
+      "step": 113
+    },
+    {
+      "epoch": 1.0255474452554745,
+      "grad_norm": 0.7356380006169851,
+      "learning_rate": 0.00013336366320877062,
+      "loss": 0.448,
+      "step": 114
+    },
+    {
+      "epoch": 1.0345311622683886,
+      "grad_norm": 1.7391096880778294,
+      "learning_rate": 0.00013273449797775503,
+      "loss": 0.4648,
+      "step": 115
+    },
+    {
+      "epoch": 1.0435148792813027,
+      "grad_norm": 0.973846535405518,
+      "learning_rate": 0.00013209951106862437,
+      "loss": 0.4656,
+      "step": 116
+    },
+    {
+      "epoch": 1.0524985962942168,
+      "grad_norm": 1.292877541779147,
+      "learning_rate": 0.00013145877258139918,
+      "loss": 0.457,
+      "step": 117
+    },
+    {
+      "epoch": 1.0614823133071307,
+      "grad_norm": 0.9804792453442043,
+      "learning_rate": 0.0001308123532510513,
+      "loss": 0.465,
+      "step": 118
+    },
+    {
+      "epoch": 1.0704660303200448,
+      "grad_norm": 0.7853353924269662,
+      "learning_rate": 0.00013016032443969493,
+      "loss": 0.4457,
+      "step": 119
+    },
+    {
+      "epoch": 1.079449747332959,
+      "grad_norm": 0.7657346345603476,
+      "learning_rate": 0.0001295027581287087,
+      "loss": 0.4472,
+      "step": 120
+    },
+    {
+      "epoch": 1.088433464345873,
+      "grad_norm": 0.5359112242814267,
+      "learning_rate": 0.00012883972691078892,
+      "loss": 0.4509,
+      "step": 121
+    },
+    {
+      "epoch": 1.0974171813587872,
+      "grad_norm": 0.5597649131000553,
+      "learning_rate": 0.00012817130398193595,
+      "loss": 0.4389,
+      "step": 122
+    },
+    {
+      "epoch": 1.1064008983717013,
+      "grad_norm": 0.44706566167796036,
+      "learning_rate": 0.00012749756313337353,
+      "loss": 0.4552,
+      "step": 123
+    },
+    {
+      "epoch": 1.1153846153846154,
+      "grad_norm": 0.43781190235972983,
+      "learning_rate": 0.00012681857874340235,
+      "loss": 0.431,
+      "step": 124
+    },
+    {
+      "epoch": 1.1243683323975295,
+      "grad_norm": 0.3759045435946263,
+      "learning_rate": 0.0001261344257691894,
+      "loss": 0.4577,
+      "step": 125
+    },
+    {
+      "epoch": 1.1333520494104437,
+      "grad_norm": 0.30888079178718136,
+      "learning_rate": 0.00012544517973849247,
+      "loss": 0.4267,
+      "step": 126
+    },
+    {
+      "epoch": 1.1423357664233578,
+      "grad_norm": 0.33375312345057856,
+      "learning_rate": 0.00012475091674132266,
+      "loss": 0.4422,
+      "step": 127
+    },
+    {
+      "epoch": 1.1513194834362717,
+      "grad_norm": 0.2801704573825818,
+      "learning_rate": 0.00012405171342154409,
+      "loss": 0.431,
+      "step": 128
+    },
+    {
+      "epoch": 1.1603032004491858,
+      "grad_norm": 0.2923926642463168,
+      "learning_rate": 0.00012334764696841273,
+      "loss": 0.4439,
+      "step": 129
+    },
+    {
+      "epoch": 1.1692869174621,
+      "grad_norm": 0.2515690098567012,
+      "learning_rate": 0.0001226387951080551,
+      "loss": 0.4351,
+      "step": 130
+    },
+    {
+      "epoch": 1.178270634475014,
+      "grad_norm": 0.23889820398885311,
+      "learning_rate": 0.0001219252360948876,
+      "loss": 0.4355,
+      "step": 131
+    },
+    {
+      "epoch": 1.1872543514879281,
+      "grad_norm": 0.25178540985599773,
+      "learning_rate": 0.00012120704870297748,
+      "loss": 0.4339,
+      "step": 132
+    },
+    {
+      "epoch": 1.1962380685008422,
+      "grad_norm": 0.21429460359169059,
+      "learning_rate": 0.00012048431221734653,
+      "loss": 0.4342,
+      "step": 133
+    },
+    {
+      "epoch": 1.2052217855137564,
+      "grad_norm": 0.24272912960518966,
+      "learning_rate": 0.00011975710642521837,
+      "loss": 0.4296,
+      "step": 134
+    },
+    {
+      "epoch": 1.2142055025266705,
+      "grad_norm": 0.2302992707934062,
+      "learning_rate": 0.0001190255116072102,
+      "loss": 0.4305,
+      "step": 135
+    },
+    {
+      "epoch": 1.2231892195395846,
+      "grad_norm": 0.19950449025995803,
+      "learning_rate": 0.00011828960852847013,
+      "loss": 0.4302,
+      "step": 136
+    },
+    {
+      "epoch": 1.2321729365524985,
+      "grad_norm": 0.26523954308583547,
+      "learning_rate": 0.00011754947842976111,
+      "loss": 0.4336,
+      "step": 137
+    },
+    {
+      "epoch": 1.2411566535654126,
+      "grad_norm": 0.30033806349738085,
+      "learning_rate": 0.00011680520301849219,
+      "loss": 0.4306,
+      "step": 138
+    },
+    {
+      "epoch": 1.2501403705783267,
+      "grad_norm": 0.31461278548219257,
+      "learning_rate": 0.00011605686445969838,
+      "loss": 0.4303,
+      "step": 139
+    },
+    {
+      "epoch": 1.2591240875912408,
+      "grad_norm": 0.3198850103038025,
+      "learning_rate": 0.00011530454536697001,
+      "loss": 0.429,
+      "step": 140
+    },
+    {
+      "epoch": 1.268107804604155,
+      "grad_norm": 0.3187477850631323,
+      "learning_rate": 0.00011454832879333245,
+      "loss": 0.4311,
+      "step": 141
+    },
+    {
+      "epoch": 1.277091521617069,
+      "grad_norm": 0.21115367860711695,
+      "learning_rate": 0.00011378829822207734,
+      "loss": 0.4281,
+      "step": 142
+    },
+    {
+      "epoch": 1.2860752386299832,
+      "grad_norm": 0.24800096621587547,
+      "learning_rate": 0.00011302453755754653,
+      "loss": 0.4303,
+      "step": 143
+    },
+    {
+      "epoch": 1.2950589556428973,
+      "grad_norm": 0.22634162707057526,
+      "learning_rate": 0.00011225713111586919,
+      "loss": 0.4354,
+      "step": 144
+    },
+    {
+      "epoch": 1.3040426726558114,
+      "grad_norm": 0.22574772625477368,
+      "learning_rate": 0.00011148616361565386,
+      "loss": 0.4176,
+      "step": 145
+    },
+    {
+      "epoch": 1.3130263896687255,
+      "grad_norm": 0.24036094076866749,
+      "learning_rate": 0.00011071172016863557,
+      "loss": 0.4299,
+      "step": 146
+    },
+    {
+      "epoch": 1.3220101066816397,
+      "grad_norm": 0.28738541137546303,
+      "learning_rate": 0.00010993388627028018,
+      "loss": 0.4304,
+      "step": 147
+    },
+    {
+      "epoch": 1.3309938236945535,
+      "grad_norm": 0.2609472165218046,
+      "learning_rate": 0.0001091527477903457,
+      "loss": 0.4332,
+      "step": 148
+    },
+    {
+      "epoch": 1.3399775407074677,
+      "grad_norm": 0.26038582479198685,
+      "learning_rate": 0.00010836839096340285,
+      "loss": 0.4209,
+      "step": 149
+    },
+    {
+      "epoch": 1.3489612577203818,
+      "grad_norm": 0.2316024258260121,
+      "learning_rate": 0.00010758090237931497,
+      "loss": 0.4283,
+      "step": 150
+    },
+    {
+      "epoch": 1.357944974733296,
+      "grad_norm": 0.2098606063881437,
+      "learning_rate": 0.0001067903689736789,
+      "loss": 0.4263,
+      "step": 151
+    },
+    {
+      "epoch": 1.36692869174621,
+      "grad_norm": 0.24683787256011072,
+      "learning_rate": 0.00010599687801822763,
+      "loss": 0.4141,
+      "step": 152
+    },
+    {
+      "epoch": 1.3759124087591241,
+      "grad_norm": 0.24229393658697693,
+      "learning_rate": 0.00010520051711119585,
+      "loss": 0.4202,
+      "step": 153
+    },
+    {
+      "epoch": 1.3848961257720382,
+      "grad_norm": 0.20606020348790752,
+      "learning_rate": 0.00010440137416764941,
+      "loss": 0.4227,
+      "step": 154
+    },
+    {
+      "epoch": 1.3938798427849521,
+      "grad_norm": 0.2804421841986466,
+      "learning_rate": 0.00010359953740977998,
+      "loss": 0.4163,
+      "step": 155
+    },
+    {
+      "epoch": 1.4028635597978663,
+      "grad_norm": 0.23544388812205802,
+      "learning_rate": 0.00010279509535716549,
+      "loss": 0.4141,
+      "step": 156
+    },
+    {
+      "epoch": 1.4118472768107804,
+      "grad_norm": 0.19406146485494044,
+      "learning_rate": 0.00010198813681699804,
+      "loss": 0.4276,
+      "step": 157
+    },
+    {
+      "epoch": 1.4208309938236945,
+      "grad_norm": 0.24473989108124805,
+      "learning_rate": 0.00010117875087427998,
+      "loss": 0.4168,
+      "step": 158
+    },
+    {
+      "epoch": 1.4298147108366086,
+      "grad_norm": 0.29142953549457795,
+      "learning_rate": 0.00010036702688198906,
+      "loss": 0.4309,
+      "step": 159
+    },
+    {
+      "epoch": 1.4387984278495227,
+      "grad_norm": 0.6467486970522354,
+      "learning_rate": 9.955305445121455e-05,
+      "loss": 0.4269,
+      "step": 160
+    },
+    {
+      "epoch": 1.4477821448624368,
+      "grad_norm": 0.3055063735132388,
+      "learning_rate": 9.87369234412642e-05,
+      "loss": 0.4284,
+      "step": 161
+    },
+    {
+      "epoch": 1.456765861875351,
+      "grad_norm": 0.24616493986031823,
+      "learning_rate": 9.79187239497444e-05,
+      "loss": 0.4333,
+      "step": 162
+    },
+    {
+      "epoch": 1.465749578888265,
+      "grad_norm": 0.2384894678445937,
+      "learning_rate": 9.709854630261359e-05,
+      "loss": 0.427,
+      "step": 163
+    },
+    {
+      "epoch": 1.4747332959011792,
+      "grad_norm": 0.24871583396920668,
+      "learning_rate": 9.627648104421073e-05,
+      "loss": 0.4221,
+      "step": 164
+    },
+    {
+      "epoch": 1.4837170129140933,
+      "grad_norm": 0.22412182435439584,
+      "learning_rate": 9.545261892725947e-05,
+      "loss": 0.4247,
+      "step": 165
+    },
+    {
+      "epoch": 1.4927007299270074,
+      "grad_norm": 0.2487283398786264,
+      "learning_rate": 9.462705090284956e-05,
+      "loss": 0.4202,
+      "step": 166
+    },
+    {
+      "epoch": 1.5016844469399215,
+      "grad_norm": 0.29923719674921373,
+      "learning_rate": 9.379986811039604e-05,
+      "loss": 0.4271,
+      "step": 167
+    },
+    {
+      "epoch": 1.5106681639528357,
+      "grad_norm": 1.3405716596708455,
+      "learning_rate": 9.297116186757798e-05,
+      "loss": 0.4289,
+      "step": 168
+    },
+    {
+      "epoch": 1.5196518809657495,
+      "grad_norm": 0.38664890635523597,
+      "learning_rate": 9.214102366025724e-05,
+      "loss": 0.4186,
+      "step": 169
+    },
+    {
+      "epoch": 1.5286355979786637,
+      "grad_norm": 0.4762048641691243,
+      "learning_rate": 9.130954513237883e-05,
+      "loss": 0.4295,
+      "step": 170
+    },
+    {
+      "epoch": 1.5376193149915778,
+      "grad_norm": 0.4899933718508197,
+      "learning_rate": 9.047681807585381e-05,
+      "loss": 0.4238,
+      "step": 171
+    },
+    {
+      "epoch": 1.546603032004492,
+      "grad_norm": 0.4626782645201007,
+      "learning_rate": 8.964293442042585e-05,
+      "loss": 0.427,
+      "step": 172
+    },
+    {
+      "epoch": 1.5555867490174058,
+      "grad_norm": 0.47452287145204264,
+      "learning_rate": 8.880798622352253e-05,
+      "loss": 0.4221,
+      "step": 173
+    },
+    {
+      "epoch": 1.56457046603032,
+      "grad_norm": 0.41668257298151723,
+      "learning_rate": 8.797206566009252e-05,
+      "loss": 0.4157,
+      "step": 174
+    },
+    {
+      "epoch": 1.573554183043234,
+      "grad_norm": 0.2988059378537405,
+      "learning_rate": 8.713526501242991e-05,
+      "loss": 0.4287,
+      "step": 175
+    },
+    {
+      "epoch": 1.5825379000561481,
+      "grad_norm": 0.449887308091825,
+      "learning_rate": 8.629767665998652e-05,
+      "loss": 0.428,
+      "step": 176
+    },
+    {
+      "epoch": 1.5915216170690623,
+      "grad_norm": 0.4683956205571529,
+      "learning_rate": 8.54593930691737e-05,
+      "loss": 0.419,
+      "step": 177
+    },
+    {
+      "epoch": 1.6005053340819764,
+      "grad_norm": 0.2834519006631767,
+      "learning_rate": 8.462050678315429e-05,
+      "loss": 0.4149,
+      "step": 178
+    },
+    {
+      "epoch": 1.6094890510948905,
+      "grad_norm": 0.34501079837011184,
+      "learning_rate": 8.37811104116263e-05,
+      "loss": 0.4229,
+      "step": 179
+    },
+    {
+      "epoch": 1.6184727681078046,
+      "grad_norm": 0.3768651775591254,
+      "learning_rate": 8.294129662059918e-05,
+      "loss": 0.4203,
+      "step": 180
+    },
+    {
+      "epoch": 1.6274564851207187,
+      "grad_norm": 0.3767673855555118,
+      "learning_rate": 8.210115812216378e-05,
+      "loss": 0.4236,
+      "step": 181
+    },
+    {
+      "epoch": 1.6364402021336328,
+      "grad_norm": 0.4116044954095348,
+      "learning_rate": 8.126078766425732e-05,
+      "loss": 0.4168,
+      "step": 182
+    },
+    {
+      "epoch": 1.645423919146547,
+      "grad_norm": 0.24536748096315872,
+      "learning_rate": 8.04202780204245e-05,
+      "loss": 0.4112,
+      "step": 183
+    },
+    {
+      "epoch": 1.654407636159461,
+      "grad_norm": 0.36986369671919106,
+      "learning_rate": 7.957972197957554e-05,
+      "loss": 0.4119,
+      "step": 184
+    },
+    {
+      "epoch": 1.6633913531723752,
+      "grad_norm": 0.36469684298934046,
+      "learning_rate": 7.87392123357427e-05,
+      "loss": 0.4208,
+      "step": 185
+    },
+    {
+      "epoch": 1.6723750701852893,
+      "grad_norm": 0.3062531478790991,
+      "learning_rate": 7.789884187783625e-05,
+      "loss": 0.4199,
+      "step": 186
+    },
+    {
+      "epoch": 1.6813587871982034,
+      "grad_norm": 0.2245212911708979,
+      "learning_rate": 7.705870337940085e-05,
+      "loss": 0.419,
+      "step": 187
+    },
+    {
+      "epoch": 1.6903425042111173,
+      "grad_norm": 0.2815117246044071,
+      "learning_rate": 7.621888958837372e-05,
+      "loss": 0.4156,
+      "step": 188
+    },
+    {
+      "epoch": 1.6993262212240314,
+      "grad_norm": 0.25130028675485033,
+      "learning_rate": 7.537949321684574e-05,
+      "loss": 0.4227,
+      "step": 189
+    },
+    {
+      "epoch": 1.7083099382369455,
+      "grad_norm": 0.24156587078920358,
+      "learning_rate": 7.454060693082634e-05,
+      "loss": 0.4128,
+      "step": 190
+    },
+    {
+      "epoch": 1.7172936552498597,
+      "grad_norm": 0.24268807529448605,
+      "learning_rate": 7.37023233400135e-05,
+      "loss": 0.4225,
+      "step": 191
+    },
+    {
+      "epoch": 1.7262773722627736,
+      "grad_norm": 0.34956262423936846,
+      "learning_rate": 7.286473498757011e-05,
+      "loss": 0.4224,
+      "step": 192
+    },
+    {
+      "epoch": 1.7352610892756877,
+      "grad_norm": 0.2608226148563663,
+      "learning_rate": 7.20279343399075e-05,
+      "loss": 0.4222,
+      "step": 193
+    },
+    {
+      "epoch": 1.7442448062886018,
+      "grad_norm": 0.23482325735709875,
+      "learning_rate": 7.11920137764775e-05,
+      "loss": 0.4131,
+      "step": 194
+    },
+    {
+      "epoch": 1.753228523301516,
+      "grad_norm": 0.22733452979745924,
+      "learning_rate": 7.035706557957416e-05,
+      "loss": 0.4234,
+      "step": 195
+    },
+    {
+      "epoch": 1.76221224031443,
+      "grad_norm": 0.31731646035712113,
+      "learning_rate": 6.952318192414623e-05,
+      "loss": 0.4274,
+      "step": 196
+    },
+    {
+      "epoch": 1.7711959573273441,
+      "grad_norm": 0.2180643349500018,
+      "learning_rate": 6.869045486762122e-05,
+      "loss": 0.4138,
+      "step": 197
+    },
+    {
+      "epoch": 1.7801796743402583,
+      "grad_norm": 0.24249072075661643,
+      "learning_rate": 6.78589763397428e-05,
+      "loss": 0.4073,
+      "step": 198
+    },
+    {
+      "epoch": 1.7891633913531724,
+      "grad_norm": 0.30886511591698557,
+      "learning_rate": 6.702883813242205e-05,
+      "loss": 0.4113,
+      "step": 199
+    },
+    {
+      "epoch": 1.7981471083660865,
+      "grad_norm": 0.20119667545730122,
+      "learning_rate": 6.620013188960398e-05,
+      "loss": 0.4129,
+      "step": 200
+    },
+    {
+      "epoch": 1.8071308253790006,
+      "grad_norm": 0.24435752019081544,
+      "learning_rate": 6.537294909715048e-05,
+      "loss": 0.4163,
+      "step": 201
+    },
+    {
+      "epoch": 1.8161145423919147,
+      "grad_norm": 0.2791733867754067,
+      "learning_rate": 6.454738107274056e-05,
+      "loss": 0.4292,
+      "step": 202
+    },
+    {
+      "epoch": 1.8250982594048288,
+      "grad_norm": 0.18260363770459387,
+      "learning_rate": 6.372351895578931e-05,
+      "loss": 0.4134,
+      "step": 203
+    },
+    {
+      "epoch": 1.834081976417743,
+      "grad_norm": 0.21731481060715926,
+      "learning_rate": 6.290145369738644e-05,
+      "loss": 0.4078,
+      "step": 204
+    },
+    {
+      "epoch": 1.843065693430657,
+      "grad_norm": 0.21873653534247425,
+      "learning_rate": 6.208127605025564e-05,
+      "loss": 0.4193,
+      "step": 205
+    },
+    {
+      "epoch": 1.8520494104435712,
+      "grad_norm": 0.22981600238165162,
+      "learning_rate": 6.126307655873584e-05,
+      "loss": 0.4076,
+      "step": 206
+    },
+    {
+      "epoch": 1.861033127456485,
+      "grad_norm": 0.20566956828097804,
+      "learning_rate": 6.0446945548785506e-05,
+      "loss": 0.4151,
+      "step": 207
+    },
+    {
+      "epoch": 1.8700168444693992,
+      "grad_norm": 0.2119512773707407,
+      "learning_rate": 5.9632973118010964e-05,
+      "loss": 0.4248,
+      "step": 208
+    },
+    {
+      "epoch": 1.8790005614823133,
+      "grad_norm": 0.20654052551807822,
+      "learning_rate": 5.882124912572005e-05,
+      "loss": 0.4125,
+      "step": 209
+    },
+    {
+      "epoch": 1.8879842784952274,
+      "grad_norm": 0.23846846179660958,
+      "learning_rate": 5.801186318300195e-05,
+      "loss": 0.4122,
+      "step": 210
+    },
+    {
+      "epoch": 1.8969679955081415,
+      "grad_norm": 0.17491624726700092,
+      "learning_rate": 5.720490464283452e-05,
+      "loss": 0.4124,
+      "step": 211
+    },
+    {
+      "epoch": 1.9059517125210554,
+      "grad_norm": 0.22200392942766323,
+      "learning_rate": 5.6400462590220024e-05,
+      "loss": 0.4213,
+      "step": 212
+    },
+    {
+      "epoch": 1.9149354295339696,
+      "grad_norm": 0.23091340411259711,
+      "learning_rate": 5.559862583235059e-05,
+      "loss": 0.4032,
+      "step": 213
+    },
+    {
+      "epoch": 1.9239191465468837,
+      "grad_norm": 0.1665570698719248,
+      "learning_rate": 5.4799482888804174e-05,
+      "loss": 0.4214,
+      "step": 214
+    },
+    {
+      "epoch": 1.9329028635597978,
+      "grad_norm": 0.22812374386475856,
+      "learning_rate": 5.400312198177239e-05,
+      "loss": 0.4116,
+      "step": 215
+    },
+    {
+      "epoch": 1.941886580572712,
+      "grad_norm": 0.19255968363783327,
+      "learning_rate": 5.320963102632111e-05,
+      "loss": 0.4153,
+      "step": 216
+    },
+    {
+      "epoch": 1.950870297585626,
+      "grad_norm": 0.20277918206935358,
+      "learning_rate": 5.2419097620685035e-05,
+      "loss": 0.4047,
+      "step": 217
+    },
+    {
+      "epoch": 1.9598540145985401,
+      "grad_norm": 0.23199199574198273,
+      "learning_rate": 5.163160903659715e-05,
+      "loss": 0.4234,
+      "step": 218
+    },
+    {
+      "epoch": 1.9688377316114543,
+      "grad_norm": 0.20877288575006245,
+      "learning_rate": 5.08472522096543e-05,
+      "loss": 0.4143,
+      "step": 219
+    },
+    {
+      "epoch": 1.9778214486243684,
+      "grad_norm": 0.1718225040486589,
+      "learning_rate": 5.006611372971982e-05,
+      "loss": 0.4106,
+      "step": 220
+    },
+    {
+      "epoch": 1.9868051656372825,
+      "grad_norm": 0.21428383606764617,
+      "learning_rate": 4.928827983136444e-05,
+      "loss": 0.3995,
+      "step": 221
+    },
+    {
+      "epoch": 1.9957888826501966,
+      "grad_norm": 0.18488960532491983,
+      "learning_rate": 4.8513836384346175e-05,
+      "loss": 0.4229,
+      "step": 222
+    },
+    {
+      "epoch": 2.0061763054463784,
+      "grad_norm": 0.44509287430368083,
+      "learning_rate": 4.774286888413083e-05,
+      "loss": 0.72,
+      "step": 223
+    },
+    {
+      "epoch": 2.0151600224592925,
+      "grad_norm": 0.349453446774281,
+      "learning_rate": 4.697546244245351e-05,
+      "loss": 0.3579,
+      "step": 224
+    },
+    {
+      "epoch": 2.0241437394722066,
+      "grad_norm": 0.2983670407361141,
+      "learning_rate": 4.6211701777922695e-05,
+      "loss": 0.3791,
+      "step": 225
+    },
+    {
+      "epoch": 2.0331274564851207,
+      "grad_norm": 0.35065646157559205,
+      "learning_rate": 4.545167120666758e-05,
+      "loss": 0.3665,
+      "step": 226
+    },
+    {
+      "epoch": 2.042111173498035,
+      "grad_norm": 0.2928621163403631,
+      "learning_rate": 4.469545463303e-05,
+      "loss": 0.3715,
+      "step": 227
+    },
+    {
+      "epoch": 2.051094890510949,
+      "grad_norm": 0.2954327190991827,
+      "learning_rate": 4.394313554030163e-05,
+      "loss": 0.3641,
+      "step": 228
+    },
+    {
+      "epoch": 2.060078607523863,
+      "grad_norm": 0.29422712186180044,
+      "learning_rate": 4.3194796981507835e-05,
+      "loss": 0.3781,
+      "step": 229
+    },
+    {
+      "epoch": 2.069062324536777,
+      "grad_norm": 0.3071711408148744,
+      "learning_rate": 4.245052157023891e-05,
+      "loss": 0.3645,
+      "step": 230
+    },
+    {
+      "epoch": 2.0780460415496913,
+      "grad_norm": 0.23769683777534895,
+      "learning_rate": 4.1710391471529894e-05,
+      "loss": 0.3654,
+      "step": 231
+    },
+    {
+      "epoch": 2.0870297585626054,
+      "grad_norm": 0.28768316660855947,
+      "learning_rate": 4.0974488392789824e-05,
+      "loss": 0.3681,
+      "step": 232
+    },
+    {
+      "epoch": 2.0960134755755195,
+      "grad_norm": 0.22020511686667768,
+      "learning_rate": 4.0242893574781645e-05,
+      "loss": 0.3544,
+      "step": 233
+    },
+    {
+      "epoch": 2.1049971925884337,
+      "grad_norm": 0.240285704968683,
+      "learning_rate": 3.951568778265347e-05,
+      "loss": 0.3685,
+      "step": 234
+    },
+    {
+      "epoch": 2.1139809096013478,
+      "grad_norm": 0.2146794751583024,
+      "learning_rate": 3.879295129702254e-05,
+      "loss": 0.367,
+      "step": 235
+    },
+    {
+      "epoch": 2.1229646266142614,
+      "grad_norm": 0.21870712204994086,
+      "learning_rate": 3.807476390511243e-05,
+      "loss": 0.3627,
+      "step": 236
+    },
+    {
+      "epoch": 2.1319483436271756,
+      "grad_norm": 0.22541727469315342,
+      "learning_rate": 3.7361204891944926e-05,
+      "loss": 0.371,
+      "step": 237
+    },
+    {
+      "epoch": 2.1409320606400897,
+      "grad_norm": 0.19265971553935168,
+      "learning_rate": 3.66523530315873e-05,
+      "loss": 0.3604,
+      "step": 238
+    },
+    {
+      "epoch": 2.149915777653004,
+      "grad_norm": 0.23010447898223835,
+      "learning_rate": 3.594828657845594e-05,
+      "loss": 0.3624,
+      "step": 239
+    },
+    {
+      "epoch": 2.158899494665918,
+      "grad_norm": 0.17138567803268676,
+      "learning_rate": 3.524908325867735e-05,
+      "loss": 0.3639,
+      "step": 240
+    },
+    {
+      "epoch": 2.167883211678832,
+      "grad_norm": 0.1951941414970527,
+      "learning_rate": 3.455482026150755e-05,
+      "loss": 0.3544,
+      "step": 241
+    },
+    {
+      "epoch": 2.176866928691746,
+      "grad_norm": 0.18352754967573434,
+      "learning_rate": 3.3865574230810636e-05,
+      "loss": 0.3773,
+      "step": 242
+    },
+    {
+      "epoch": 2.1858506457046603,
+      "grad_norm": 0.21536452648283833,
+      "learning_rate": 3.318142125659765e-05,
+      "loss": 0.3825,
+      "step": 243
+    },
+    {
+      "epoch": 2.1948343627175744,
+      "grad_norm": 0.1829109115920877,
+      "learning_rate": 3.250243686662651e-05,
+      "loss": 0.3579,
+      "step": 244
+    },
+    {
+      "epoch": 2.2038180797304885,
+      "grad_norm": 0.18214718381067138,
+      "learning_rate": 3.182869601806406e-05,
+      "loss": 0.3626,
+      "step": 245
+    },
+    {
+      "epoch": 2.2128017967434026,
+      "grad_norm": 0.1825125643696691,
+      "learning_rate": 3.11602730892111e-05,
+      "loss": 0.3573,
+      "step": 246
+    },
+    {
+      "epoch": 2.2217855137563167,
+      "grad_norm": 0.18441564836737437,
+      "learning_rate": 3.0497241871291333e-05,
+      "loss": 0.3667,
+      "step": 247
+    },
+    {
+      "epoch": 2.230769230769231,
+      "grad_norm": 0.18116751476899712,
+      "learning_rate": 2.9839675560305088e-05,
+      "loss": 0.3539,
+      "step": 248
+    },
+    {
+      "epoch": 2.239752947782145,
+      "grad_norm": 0.18390826080033681,
+      "learning_rate": 2.918764674894875e-05,
+      "loss": 0.3618,
+      "step": 249
+    },
+    {
+      "epoch": 2.248736664795059,
+      "grad_norm": 0.14612480392036745,
+      "learning_rate": 2.8541227418600858e-05,
+      "loss": 0.3624,
+      "step": 250
+    },
+    {
+      "epoch": 2.257720381807973,
+      "grad_norm": 0.17070194783528053,
+      "learning_rate": 2.7900488931375666e-05,
+      "loss": 0.3664,
+      "step": 251
+    },
+    {
+      "epoch": 2.2667040988208873,
+      "grad_norm": 0.1354347326780045,
+      "learning_rate": 2.7265502022244993e-05,
+      "loss": 0.372,
+      "step": 252
+    },
+    {
+      "epoch": 2.2756878158338014,
+      "grad_norm": 0.14413453810840812,
+      "learning_rate": 2.6636336791229393e-05,
+      "loss": 0.3601,
+      "step": 253
+    },
+    {
+      "epoch": 2.2846715328467155,
+      "grad_norm": 0.15479999812641926,
+      "learning_rate": 2.6013062695659397e-05,
+      "loss": 0.3779,
+      "step": 254
+    },
+    {
+      "epoch": 2.293655249859629,
+      "grad_norm": 0.13022418657134416,
+      "learning_rate": 2.539574854250769e-05,
+      "loss": 0.3628,
+      "step": 255
+    },
+    {
+      "epoch": 2.3026389668725433,
+      "grad_norm": 0.14668634420749707,
+      "learning_rate": 2.4784462480793084e-05,
+      "loss": 0.3706,
+      "step": 256
+    },
+    {
+      "epoch": 2.3116226838854574,
+      "grad_norm": 0.12905147875548487,
+      "learning_rate": 2.4179271994057184e-05,
+      "loss": 0.3702,
+      "step": 257
+    },
+    {
+      "epoch": 2.3206064008983716,
+      "grad_norm": 0.1447256804180326,
+      "learning_rate": 2.3580243892914424e-05,
+      "loss": 0.3573,
+      "step": 258
+    },
+    {
+      "epoch": 2.3295901179112857,
+      "grad_norm": 0.11748170585015046,
+      "learning_rate": 2.2987444307676478e-05,
+      "loss": 0.3758,
+      "step": 259
+    },
+    {
+      "epoch": 2.3385738349242,
+      "grad_norm": 0.12882383618208246,
+      "learning_rate": 2.24009386810517e-05,
+      "loss": 0.3511,
+      "step": 260
+    },
+    {
+      "epoch": 2.347557551937114,
+      "grad_norm": 0.12865217121068986,
+      "learning_rate": 2.1820791760920547e-05,
+      "loss": 0.3716,
+      "step": 261
+    },
+    {
+      "epoch": 2.356541268950028,
+      "grad_norm": 0.12081698628061978,
+      "learning_rate": 2.124706759318766e-05,
+      "loss": 0.3683,
+      "step": 262
+    },
+    {
+      "epoch": 2.365524985962942,
+      "grad_norm": 0.12215193992569423,
+      "learning_rate": 2.0679829514711436e-05,
+      "loss": 0.3646,
+      "step": 263
+    },
+    {
+      "epoch": 2.3745087029758563,
+      "grad_norm": 0.11354398681818435,
+      "learning_rate": 2.0119140146311905e-05,
+      "loss": 0.3686,
+      "step": 264
+    },
+    {
+      "epoch": 2.3834924199887704,
+      "grad_norm": 0.12582747852018303,
+      "learning_rate": 1.9565061385857663e-05,
+      "loss": 0.3682,
+      "step": 265
+    },
+    {
+      "epoch": 2.3924761370016845,
+      "grad_norm": 0.11257895081104091,
+      "learning_rate": 1.9017654401432546e-05,
+      "loss": 0.3717,
+      "step": 266
+    },
+    {
+      "epoch": 2.4014598540145986,
+      "grad_norm": 0.12305919426390005,
+      "learning_rate": 1.8476979624582937e-05,
+      "loss": 0.3581,
+      "step": 267
+    },
+    {
+      "epoch": 2.4104435710275127,
+      "grad_norm": 0.11203368697453076,
+      "learning_rate": 1.794309674364641e-05,
+      "loss": 0.37,
+      "step": 268
+    },
+    {
+      "epoch": 2.419427288040427,
+      "grad_norm": 0.12463695189114064,
+      "learning_rate": 1.7416064697162287e-05,
+      "loss": 0.3736,
+      "step": 269
+    },
+    {
+      "epoch": 2.428411005053341,
+      "grad_norm": 0.10917678973672461,
+      "learning_rate": 1.6895941667365168e-05,
+      "loss": 0.3535,
+      "step": 270
+    },
+    {
+      "epoch": 2.437394722066255,
+      "grad_norm": 0.11281664768314911,
+      "learning_rate": 1.6382785073761703e-05,
+      "loss": 0.3571,
+      "step": 271
+    },
+    {
+      "epoch": 2.446378439079169,
+      "grad_norm": 0.13666809456134899,
+      "learning_rate": 1.587665156679185e-05,
+      "loss": 0.3658,
+      "step": 272
+    },
+    {
+      "epoch": 2.4553621560920833,
+      "grad_norm": 0.10808808623606102,
+      "learning_rate": 1.537759702157482e-05,
+      "loss": 0.358,
+      "step": 273
+    },
+    {
+      "epoch": 2.464345873104997,
+      "grad_norm": 0.1116710945877027,
+      "learning_rate": 1.4885676531740694e-05,
+      "loss": 0.3758,
+      "step": 274
+    },
+    {
+      "epoch": 2.473329590117911,
+      "grad_norm": 0.13267100870029716,
+      "learning_rate": 1.4400944403348321e-05,
+      "loss": 0.366,
+      "step": 275
+    },
+    {
+      "epoch": 2.482313307130825,
+      "grad_norm": 0.11602385380874608,
+      "learning_rate": 1.392345414889011e-05,
+      "loss": 0.3702,
+      "step": 276
+    },
+    {
+      "epoch": 2.4912970241437393,
+      "grad_norm": 0.1145440180474532,
+      "learning_rate": 1.345325848138451e-05,
+      "loss": 0.3511,
+      "step": 277
+    },
+    {
+      "epoch": 2.5002807411566534,
+      "grad_norm": 0.11526664415049362,
+      "learning_rate": 1.2990409308556679e-05,
+      "loss": 0.3516,
+      "step": 278
+    },
+    {
+      "epoch": 2.5092644581695676,
+      "grad_norm": 0.11013047619147111,
+      "learning_rate": 1.2534957727108083e-05,
+      "loss": 0.3633,
+      "step": 279
+    },
+    {
+      "epoch": 2.5182481751824817,
+      "grad_norm": 0.10354704499832852,
+      "learning_rate": 1.208695401707562e-05,
+      "loss": 0.3733,
+      "step": 280
+    },
+    {
+      "epoch": 2.527231892195396,
+      "grad_norm": 0.10236481273298736,
+      "learning_rate": 1.1646447636280916e-05,
+      "loss": 0.3555,
+      "step": 281
+    },
+    {
+      "epoch": 2.53621560920831,
+      "grad_norm": 0.11167580752401748,
+      "learning_rate": 1.1213487214870366e-05,
+      "loss": 0.359,
+      "step": 282
+    },
+    {
+      "epoch": 2.545199326221224,
+      "grad_norm": 0.1099233626090954,
+      "learning_rate": 1.0788120549946535e-05,
+      "loss": 0.3744,
+      "step": 283
+    },
+    {
+      "epoch": 2.554183043234138,
+      "grad_norm": 0.11082716206350583,
+      "learning_rate": 1.0370394600291633e-05,
+      "loss": 0.3508,
+      "step": 284
+    },
+    {
+      "epoch": 2.5631667602470523,
+      "grad_norm": 0.12103690964444253,
+      "learning_rate": 9.960355481183357e-06,
+      "loss": 0.3789,
+      "step": 285
+    },
+    {
+      "epoch": 2.5721504772599664,
+      "grad_norm": 0.11920362715834164,
+      "learning_rate": 9.55804845930401e-06,
+      "loss": 0.3667,
+      "step": 286
+    },
+    {
+      "epoch": 2.5811341942728805,
+      "grad_norm": 0.1044793820973793,
+      "learning_rate": 9.163517947743218e-06,
+      "loss": 0.368,
+      "step": 287
+    },
+    {
+      "epoch": 2.5901179112857946,
+      "grad_norm": 0.1079810919458012,
+      "learning_rate": 8.776807501094891e-06,
+      "loss": 0.3568,
+      "step": 288
+    },
+    {
+      "epoch": 2.5991016282987087,
+      "grad_norm": 0.10388558819512615,
+      "learning_rate": 8.397959810648992e-06,
+      "loss": 0.3564,
+      "step": 289
+    },
+    {
+      "epoch": 2.608085345311623,
+      "grad_norm": 0.10325472697547591,
+      "learning_rate": 8.027016699678581e-06,
+      "loss": 0.3726,
+      "step": 290
+    },
+    {
+      "epoch": 2.6170690623245365,
+      "grad_norm": 0.1027618868655184,
+      "learning_rate": 7.664019118822676e-06,
+      "loss": 0.367,
+      "step": 291
+    },
+    {
+      "epoch": 2.626052779337451,
+      "grad_norm": 0.0929701185172587,
+      "learning_rate": 7.30900714156551e-06,
+      "loss": 0.3525,
+      "step": 292
+    },
+    {
+      "epoch": 2.6350364963503647,
+      "grad_norm": 0.10035396955714077,
+      "learning_rate": 6.962019959812516e-06,
+      "loss": 0.3699,
+      "step": 293
+    },
+    {
+      "epoch": 2.6440202133632793,
+      "grad_norm": 0.0928922534561677,
+      "learning_rate": 6.623095879563774e-06,
+      "loss": 0.3665,
+      "step": 294
+    },
+    {
+      "epoch": 2.653003930376193,
+      "grad_norm": 0.10258049793200555,
+      "learning_rate": 6.29227231668506e-06,
+      "loss": 0.3583,
+      "step": 295
+    },
+    {
+      "epoch": 2.661987647389107,
+      "grad_norm": 0.0944834724986228,
+      "learning_rate": 5.9695857927774396e-06,
+      "loss": 0.3679,
+      "step": 296
+    },
+    {
+      "epoch": 2.670971364402021,
+      "grad_norm": 0.0924161403478989,
+      "learning_rate": 5.655071931145317e-06,
+      "loss": 0.3653,
+      "step": 297
+    },
+    {
+      "epoch": 2.6799550814149353,
+      "grad_norm": 0.09826749411539404,
+      "learning_rate": 5.3487654528638025e-06,
+      "loss": 0.3646,
+      "step": 298
+    },
+    {
+      "epoch": 2.6889387984278494,
+      "grad_norm": 0.09353683680477741,
+      "learning_rate": 5.050700172945631e-06,
+      "loss": 0.3577,
+      "step": 299
+    },
+    {
+      "epoch": 2.6979225154407636,
+      "grad_norm": 0.08959975875667178,
+      "learning_rate": 4.760908996608145e-06,
+      "loss": 0.3564,
+      "step": 300
+    },
+    {
+      "epoch": 2.7069062324536777,
+      "grad_norm": 0.08693868685516433,
+      "learning_rate": 4.47942391564066e-06,
+      "loss": 0.3654,
+      "step": 301
+    },
+    {
+      "epoch": 2.715889949466592,
+      "grad_norm": 0.09893623632113224,
+      "learning_rate": 4.206276004872729e-06,
+      "loss": 0.372,
+      "step": 302
+    },
+    {
+      "epoch": 2.724873666479506,
+      "grad_norm": 0.09139811989819925,
+      "learning_rate": 3.941495418743557e-06,
+      "loss": 0.3588,
+      "step": 303
+    },
+    {
+      "epoch": 2.73385738349242,
+      "grad_norm": 0.09200970263158591,
+      "learning_rate": 3.685111387973139e-06,
+      "loss": 0.3545,
+      "step": 304
+    },
+    {
+      "epoch": 2.742841100505334,
+      "grad_norm": 0.0872066817053106,
+      "learning_rate": 3.4371522163352444e-06,
+      "loss": 0.3634,
+      "step": 305
+    },
+    {
+      "epoch": 2.7518248175182483,
+      "grad_norm": 0.08567103650346183,
+      "learning_rate": 3.1976452775328302e-06,
+      "loss": 0.3557,
+      "step": 306
+    },
+    {
+      "epoch": 2.7608085345311624,
+      "grad_norm": 0.08300411963401971,
+      "learning_rate": 2.9666170121760698e-06,
+      "loss": 0.3652,
+      "step": 307
+    },
+    {
+      "epoch": 2.7697922515440765,
+      "grad_norm": 0.08826810677038714,
+      "learning_rate": 2.7440929248634216e-06,
+      "loss": 0.3493,
+      "step": 308
+    },
+    {
+      "epoch": 2.7787759685569906,
+      "grad_norm": 0.08440475179999454,
+      "learning_rate": 2.5300975813660467e-06,
+      "loss": 0.3799,
+      "step": 309
+    },
+    {
+      "epoch": 2.7877596855699043,
+      "grad_norm": 0.08633683276388147,
+      "learning_rate": 2.324654605915839e-06,
+      "loss": 0.3485,
+      "step": 310
+    },
+    {
+      "epoch": 2.796743402582819,
+      "grad_norm": 0.08317447366792373,
+      "learning_rate": 2.127786678597348e-06,
+      "loss": 0.3685,
+      "step": 311
+    },
+    {
+      "epoch": 2.8057271195957325,
+      "grad_norm": 0.0873801185267239,
+      "learning_rate": 1.9395155328440605e-06,
+      "loss": 0.3665,
+      "step": 312
+    },
+    {
+      "epoch": 2.814710836608647,
+      "grad_norm": 0.08702648151496879,
+      "learning_rate": 1.7598619530390904e-06,
+      "loss": 0.3597,
+      "step": 313
+    },
+    {
+      "epoch": 2.8236945536215607,
+      "grad_norm": 0.08722398379689802,
+      "learning_rate": 1.5888457722206529e-06,
+      "loss": 0.3651,
+      "step": 314
+    },
+    {
+      "epoch": 2.832678270634475,
+      "grad_norm": 0.08744721551504023,
+      "learning_rate": 1.4264858698926022e-06,
+      "loss": 0.3596,
+      "step": 315
+    },
+    {
+      "epoch": 2.841661987647389,
+      "grad_norm": 0.0830808255878642,
+      "learning_rate": 1.2728001699401583e-06,
+      "loss": 0.3632,
+      "step": 316
+    },
+    {
+      "epoch": 2.850645704660303,
+      "grad_norm": 0.07917152492798647,
+      "learning_rate": 1.1278056386512514e-06,
+      "loss": 0.3698,
+      "step": 317
+    },
+    {
+      "epoch": 2.859629421673217,
+      "grad_norm": 0.07773500081958919,
+      "learning_rate": 9.915182828434688e-07,
+      "loss": 0.3634,
+      "step": 318
+    },
+    {
+      "epoch": 2.8686131386861313,
+      "grad_norm": 0.08350629325245962,
+      "learning_rate": 8.639531480969787e-07,
+      "loss": 0.3589,
+      "step": 319
+    },
+    {
+      "epoch": 2.8775968556990454,
+      "grad_norm": 0.08414289148049224,
+      "learning_rate": 7.451243170935396e-07,
+      "loss": 0.3684,
+      "step": 320
+    },
+    {
+      "epoch": 2.8865805727119596,
+      "grad_norm": 0.07899075375043686,
+      "learning_rate": 6.350449080618593e-07,
+      "loss": 0.3617,
+      "step": 321
+    },
+    {
+      "epoch": 2.8955642897248737,
+      "grad_norm": 0.07798394987626306,
+      "learning_rate": 5.337270733293665e-07,
+      "loss": 0.3613,
+      "step": 322
+    },
+    {
+      "epoch": 2.904548006737788,
+      "grad_norm": 0.0828825070825642,
+      "learning_rate": 4.4118199798066195e-07,
+      "loss": 0.3654,
+      "step": 323
+    },
+    {
+      "epoch": 2.913531723750702,
+      "grad_norm": 0.08209736577715379,
+      "learning_rate": 3.5741989862271955e-07,
+      "loss": 0.3592,
+      "step": 324
+    },
+    {
+      "epoch": 2.922515440763616,
+      "grad_norm": 0.08335581990528856,
+      "learning_rate": 2.824500222570148e-07,
+      "loss": 0.3694,
+      "step": 325
+    },
+    {
+      "epoch": 2.93149915777653,
+      "grad_norm": 0.07573943820594226,
+      "learning_rate": 2.162806452587063e-07,
+      "loss": 0.3545,
+      "step": 326
+    },
+    {
+      "epoch": 2.9404828747894443,
+      "grad_norm": 0.0776837963138844,
+      "learning_rate": 1.5891907246292192e-07,
+      "loss": 0.3521,
+      "step": 327
+    },
+    {
+      "epoch": 2.9494665918023584,
+      "grad_norm": 0.07757662948055022,
+      "learning_rate": 1.103716363583729e-07,
+      "loss": 0.3645,
+      "step": 328
+    },
+    {
+      "epoch": 2.958450308815272,
+      "grad_norm": 0.0806492488802,
+      "learning_rate": 7.064369638825064e-08,
+      "loss": 0.36,
+      "step": 329
+    },
+    {
+      "epoch": 2.9674340258281866,
+      "grad_norm": 0.07961635351123682,
+      "learning_rate": 3.973963835858463e-08,
+      "loss": 0.363,
+      "step": 330
+    },
+    {
+      "epoch": 2.9764177428411003,
+      "grad_norm": 0.08063307866994539,
+      "learning_rate": 1.766287395404298e-08,
+      "loss": 0.3659,
+      "step": 331
+    },
+    {
+      "epoch": 2.985401459854015,
+      "grad_norm": 0.07899457396008377,
+      "learning_rate": 4.415840361300383e-09,
+      "loss": 0.3593,
+      "step": 332
+    },
+    {
+      "epoch": 2.9943851768669285,
+      "grad_norm": 0.08099286982566073,
+      "learning_rate": 0.0,
+      "loss": 0.3626,
+      "step": 333
+    },
+    {
+      "epoch": 2.9943851768669285,
+      "step": 333,
+      "total_flos": 4607052149424128.0,
+      "train_loss": 0.4725978755736136,
+      "train_runtime": 115789.2905,
+      "train_samples_per_second": 2.953,
+      "train_steps_per_second": 0.003
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 333,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 4607052149424128.0,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}