diff --git "a/checkpoint-10000/trainer_state.json" "b/checkpoint-10000/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-10000/trainer_state.json"
@@ -0,0 +1,4704 @@
+{
+  "best_metric": 0.009820309467613697,
+  "best_model_checkpoint": "/workspace/previous_works/RadFM/output/RadFM-Llama3-8B-pretrain-0002-embed_tokens-depth32-lora-10ep/checkpoint-10000",
+  "epoch": 2.0951183741881416,
+  "eval_steps": 10000,
+  "global_step": 10000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0031426775612822125,
+      "grad_norm": 38.333740234375,
+      "learning_rate": 3.488372093023256e-06,
+      "loss": 2.6324,
+      "step": 15
+    },
+    {
+      "epoch": 0.006285355122564425,
+      "grad_norm": 23.8914794921875,
+      "learning_rate": 6.976744186046512e-06,
+      "loss": 2.3565,
+      "step": 30
+    },
+    {
+      "epoch": 0.009428032683846637,
+      "grad_norm": 6.890503883361816,
+      "learning_rate": 1.0465116279069768e-05,
+      "loss": 1.8897,
+      "step": 45
+    },
+    {
+      "epoch": 0.01257071024512885,
+      "grad_norm": 3.9464468955993652,
+      "learning_rate": 1.3953488372093024e-05,
+      "loss": 1.3707,
+      "step": 60
+    },
+    {
+      "epoch": 0.01571338780641106,
+      "grad_norm": 4.443431854248047,
+      "learning_rate": 1.744186046511628e-05,
+      "loss": 1.055,
+      "step": 75
+    },
+    {
+      "epoch": 0.018856065367693273,
+      "grad_norm": 3.5747361183166504,
+      "learning_rate": 2.0930232558139536e-05,
+      "loss": 0.9048,
+      "step": 90
+    },
+    {
+      "epoch": 0.02199874292897549,
+      "grad_norm": 4.540731430053711,
+      "learning_rate": 2.441860465116279e-05,
+      "loss": 0.9143,
+      "step": 105
+    },
+    {
+      "epoch": 0.0251414204902577,
+      "grad_norm": 4.121450424194336,
+      "learning_rate": 2.7906976744186048e-05,
+      "loss": 0.7641,
+      "step": 120
+    },
+    {
+      "epoch": 0.028284098051539912,
+      "grad_norm": 3.1179299354553223,
+      "learning_rate": 3.13953488372093e-05,
+      "loss": 0.7784,
+      "step": 135
+    },
+    {
+      "epoch": 0.03142677561282212,
+      "grad_norm": 2.9703869819641113,
+      "learning_rate": 3.488372093023256e-05,
+      "loss": 0.7299,
+      "step": 150
+    },
+    {
+      "epoch": 0.034569453174104335,
+      "grad_norm": 2.706854820251465,
+      "learning_rate": 3.837209302325582e-05,
+      "loss": 0.6778,
+      "step": 165
+    },
+    {
+      "epoch": 0.03771213073538655,
+      "grad_norm": 3.361267328262329,
+      "learning_rate": 4.186046511627907e-05,
+      "loss": 0.7222,
+      "step": 180
+    },
+    {
+      "epoch": 0.04085480829666876,
+      "grad_norm": 4.040229797363281,
+      "learning_rate": 4.5348837209302326e-05,
+      "loss": 0.6684,
+      "step": 195
+    },
+    {
+      "epoch": 0.04399748585795098,
+      "grad_norm": 2.817627429962158,
+      "learning_rate": 4.883720930232558e-05,
+      "loss": 0.7458,
+      "step": 210
+    },
+    {
+      "epoch": 0.04714016341923319,
+      "grad_norm": 2.8800182342529297,
+      "learning_rate": 5.232558139534884e-05,
+      "loss": 0.6338,
+      "step": 225
+    },
+    {
+      "epoch": 0.0502828409805154,
+      "grad_norm": 2.436993360519409,
+      "learning_rate": 5.5813953488372095e-05,
+      "loss": 0.6299,
+      "step": 240
+    },
+    {
+      "epoch": 0.05342551854179761,
+      "grad_norm": 3.5814456939697266,
+      "learning_rate": 5.9302325581395356e-05,
+      "loss": 0.5728,
+      "step": 255
+    },
+    {
+      "epoch": 0.056568196103079824,
+      "grad_norm": 2.8744938373565674,
+      "learning_rate": 6.27906976744186e-05,
+      "loss": 0.59,
+      "step": 270
+    },
+    {
+      "epoch": 0.059710873664362035,
+      "grad_norm": 2.679749011993408,
+      "learning_rate": 6.627906976744186e-05,
+      "loss": 0.6016,
+      "step": 285
+    },
+    {
+      "epoch": 0.06285355122564425,
+      "grad_norm": 3.1333463191986084,
+      "learning_rate": 6.976744186046513e-05,
+      "loss": 0.6569,
+      "step": 300
+    },
+    {
+      "epoch": 0.06599622878692646,
+      "grad_norm": 2.2865939140319824,
+      "learning_rate": 7.325581395348837e-05,
+      "loss": 0.6385,
+      "step": 315
+    },
+    {
+      "epoch": 0.06913890634820867,
+      "grad_norm": 2.9787251949310303,
+      "learning_rate": 7.674418604651163e-05,
+      "loss": 0.6307,
+      "step": 330
+    },
+    {
+      "epoch": 0.07228158390949088,
+      "grad_norm": 2.078509569168091,
+      "learning_rate": 8.023255813953489e-05,
+      "loss": 0.5454,
+      "step": 345
+    },
+    {
+      "epoch": 0.0754242614707731,
+      "grad_norm": 2.6606740951538086,
+      "learning_rate": 8.372093023255814e-05,
+      "loss": 0.6211,
+      "step": 360
+    },
+    {
+      "epoch": 0.0785669390320553,
+      "grad_norm": 1.9346429109573364,
+      "learning_rate": 8.72093023255814e-05,
+      "loss": 0.5954,
+      "step": 375
+    },
+    {
+      "epoch": 0.08170961659333752,
+      "grad_norm": 2.2432360649108887,
+      "learning_rate": 9.069767441860465e-05,
+      "loss": 0.5385,
+      "step": 390
+    },
+    {
+      "epoch": 0.08485229415461974,
+      "grad_norm": 2.1645498275756836,
+      "learning_rate": 9.418604651162792e-05,
+      "loss": 0.592,
+      "step": 405
+    },
+    {
+      "epoch": 0.08799497171590195,
+      "grad_norm": 2.1806533336639404,
+      "learning_rate": 9.767441860465116e-05,
+      "loss": 0.5372,
+      "step": 420
+    },
+    {
+      "epoch": 0.09113764927718417,
+      "grad_norm": 2.445610761642456,
+      "learning_rate": 9.999996802299678e-05,
+      "loss": 0.6487,
+      "step": 435
+    },
+    {
+      "epoch": 0.09428032683846638,
+      "grad_norm": 2.3592734336853027,
+      "learning_rate": 9.999948836876656e-05,
+      "loss": 0.5957,
+      "step": 450
+    },
+    {
+      "epoch": 0.09742300439974859,
+      "grad_norm": 2.3027069568634033,
+      "learning_rate": 9.999843313485898e-05,
+      "loss": 0.5835,
+      "step": 465
+    },
+    {
+      "epoch": 0.1005656819610308,
+      "grad_norm": 2.6429057121276855,
+      "learning_rate": 9.999680233342161e-05,
+      "loss": 0.592,
+      "step": 480
+    },
+    {
+      "epoch": 0.10370835952231301,
+      "grad_norm": 2.0832202434539795,
+      "learning_rate": 9.999459598322778e-05,
+      "loss": 0.6203,
+      "step": 495
+    },
+    {
+      "epoch": 0.10685103708359522,
+      "grad_norm": 2.481870412826538,
+      "learning_rate": 9.999181410967633e-05,
+      "loss": 0.5428,
+      "step": 510
+    },
+    {
+      "epoch": 0.10999371464487744,
+      "grad_norm": 1.9621151685714722,
+      "learning_rate": 9.99884567447914e-05,
+      "loss": 0.6101,
+      "step": 525
+    },
+    {
+      "epoch": 0.11313639220615965,
+      "grad_norm": 2.8833186626434326,
+      "learning_rate": 9.998452392722198e-05,
+      "loss": 0.5577,
+      "step": 540
+    },
+    {
+      "epoch": 0.11627906976744186,
+      "grad_norm": 2.4447429180145264,
+      "learning_rate": 9.998001570224158e-05,
+      "loss": 0.566,
+      "step": 555
+    },
+    {
+      "epoch": 0.11942174732872407,
+      "grad_norm": 2.141496419906616,
+      "learning_rate": 9.997493212174753e-05,
+      "loss": 0.6211,
+      "step": 570
+    },
+    {
+      "epoch": 0.12256442489000628,
+      "grad_norm": 2.389796495437622,
+      "learning_rate": 9.996927324426057e-05,
+      "loss": 0.5937,
+      "step": 585
+    },
+    {
+      "epoch": 0.1257071024512885,
+      "grad_norm": 2.1194262504577637,
+      "learning_rate": 9.996303913492408e-05,
+      "loss": 0.5847,
+      "step": 600
+    },
+    {
+      "epoch": 0.12884978001257072,
+      "grad_norm": 1.7767274379730225,
+      "learning_rate": 9.99562298655033e-05,
+      "loss": 0.518,
+      "step": 615
+    },
+    {
+      "epoch": 0.13199245757385292,
+      "grad_norm": 2.0348453521728516,
+      "learning_rate": 9.994884551438458e-05,
+      "loss": 0.5941,
+      "step": 630
+    },
+    {
+      "epoch": 0.13513513513513514,
+      "grad_norm": 1.443819284439087,
+      "learning_rate": 9.994088616657444e-05,
+      "loss": 0.5022,
+      "step": 645
+    },
+    {
+      "epoch": 0.13827781269641734,
+      "grad_norm": 2.1748251914978027,
+      "learning_rate": 9.993235191369861e-05,
+      "loss": 0.5369,
+      "step": 660
+    },
+    {
+      "epoch": 0.14142049025769957,
+      "grad_norm": 1.9295774698257446,
+      "learning_rate": 9.99232428540009e-05,
+      "loss": 0.607,
+      "step": 675
+    },
+    {
+      "epoch": 0.14456316781898176,
+      "grad_norm": 1.7530088424682617,
+      "learning_rate": 9.991355909234224e-05,
+      "loss": 0.5417,
+      "step": 690
+    },
+    {
+      "epoch": 0.147705845380264,
+      "grad_norm": 10.02226448059082,
+      "learning_rate": 9.990330074019925e-05,
+      "loss": 0.5901,
+      "step": 705
+    },
+    {
+      "epoch": 0.1508485229415462,
+      "grad_norm": 1.3864644765853882,
+      "learning_rate": 9.989246791566314e-05,
+      "loss": 0.678,
+      "step": 720
+    },
+    {
+      "epoch": 0.1539912005028284,
+      "grad_norm": 1.6103929281234741,
+      "learning_rate": 9.988106074343823e-05,
+      "loss": 0.4741,
+      "step": 735
+    },
+    {
+      "epoch": 0.1571338780641106,
+      "grad_norm": 1.5933347940444946,
+      "learning_rate": 9.986907935484064e-05,
+      "loss": 0.5391,
+      "step": 750
+    },
+    {
+      "epoch": 0.16027655562539284,
+      "grad_norm": 1.5971338748931885,
+      "learning_rate": 9.985652388779663e-05,
+      "loss": 0.5782,
+      "step": 765
+    },
+    {
+      "epoch": 0.16341923318667503,
+      "grad_norm": 1.559793472290039,
+      "learning_rate": 9.984339448684113e-05,
+      "loss": 0.5227,
+      "step": 780
+    },
+    {
+      "epoch": 0.16656191074795726,
+      "grad_norm": 1.3077164888381958,
+      "learning_rate": 9.982969130311597e-05,
+      "loss": 0.5203,
+      "step": 795
+    },
+    {
+      "epoch": 0.16970458830923948,
+      "grad_norm": 1.6828336715698242,
+      "learning_rate": 9.98154144943683e-05,
+      "loss": 0.5471,
+      "step": 810
+    },
+    {
+      "epoch": 0.17284726587052168,
+      "grad_norm": 1.387099266052246,
+      "learning_rate": 9.98005642249486e-05,
+      "loss": 0.5399,
+      "step": 825
+    },
+    {
+      "epoch": 0.1759899434318039,
+      "grad_norm": 1.723253607749939,
+      "learning_rate": 9.978514066580886e-05,
+      "loss": 0.5606,
+      "step": 840
+    },
+    {
+      "epoch": 0.1791326209930861,
+      "grad_norm": 1.22931706905365,
+      "learning_rate": 9.976914399450068e-05,
+      "loss": 0.5024,
+      "step": 855
+    },
+    {
+      "epoch": 0.18227529855436833,
+      "grad_norm": 1.4278538227081299,
+      "learning_rate": 9.97525743951731e-05,
+      "loss": 0.5983,
+      "step": 870
+    },
+    {
+      "epoch": 0.18541797611565053,
+      "grad_norm": 1.4029372930526733,
+      "learning_rate": 9.973543205857057e-05,
+      "loss": 0.5699,
+      "step": 885
+    },
+    {
+      "epoch": 0.18856065367693275,
+      "grad_norm": 1.3018133640289307,
+      "learning_rate": 9.971771718203072e-05,
+      "loss": 0.4936,
+      "step": 900
+    },
+    {
+      "epoch": 0.19170333123821495,
+      "grad_norm": 1.3082265853881836,
+      "learning_rate": 9.969942996948209e-05,
+      "loss": 0.5025,
+      "step": 915
+    },
+    {
+      "epoch": 0.19484600879949718,
+      "grad_norm": 1.2923167943954468,
+      "learning_rate": 9.968057063144182e-05,
+      "loss": 0.5779,
+      "step": 930
+    },
+    {
+      "epoch": 0.19798868636077938,
+      "grad_norm": 1.2902971506118774,
+      "learning_rate": 9.966113938501313e-05,
+      "loss": 0.5373,
+      "step": 945
+    },
+    {
+      "epoch": 0.2011313639220616,
+      "grad_norm": 1.391560673713684,
+      "learning_rate": 9.964113645388293e-05,
+      "loss": 0.5858,
+      "step": 960
+    },
+    {
+      "epoch": 0.2042740414833438,
+      "grad_norm": 1.3245513439178467,
+      "learning_rate": 9.96205620683192e-05,
+      "loss": 0.6043,
+      "step": 975
+    },
+    {
+      "epoch": 0.20741671904462602,
+      "grad_norm": 1.4998241662979126,
+      "learning_rate": 9.95994164651683e-05,
+      "loss": 0.5785,
+      "step": 990
+    },
+    {
+      "epoch": 0.21055939660590822,
+      "grad_norm": 1.090804934501648,
+      "learning_rate": 9.957769988785236e-05,
+      "loss": 0.6439,
+      "step": 1005
+    },
+    {
+      "epoch": 0.21370207416719045,
+      "grad_norm": 1.1564654111862183,
+      "learning_rate": 9.955541258636631e-05,
+      "loss": 0.5091,
+      "step": 1020
+    },
+    {
+      "epoch": 0.21684475172847265,
+      "grad_norm": 1.1778066158294678,
+      "learning_rate": 9.953255481727513e-05,
+      "loss": 0.5456,
+      "step": 1035
+    },
+    {
+      "epoch": 0.21998742928975487,
+      "grad_norm": 1.3568626642227173,
+      "learning_rate": 9.950912684371088e-05,
+      "loss": 0.5208,
+      "step": 1050
+    },
+    {
+      "epoch": 0.2231301068510371,
+      "grad_norm": 1.804425597190857,
+      "learning_rate": 9.948512893536961e-05,
+      "loss": 0.4956,
+      "step": 1065
+    },
+    {
+      "epoch": 0.2262727844123193,
+      "grad_norm": 1.226159930229187,
+      "learning_rate": 9.946056136850833e-05,
+      "loss": 0.5812,
+      "step": 1080
+    },
+    {
+      "epoch": 0.22941546197360152,
+      "grad_norm": 1.1530790328979492,
+      "learning_rate": 9.943542442594177e-05,
+      "loss": 0.4742,
+      "step": 1095
+    },
+    {
+      "epoch": 0.23255813953488372,
+      "grad_norm": 1.390417218208313,
+      "learning_rate": 9.940971839703916e-05,
+      "loss": 0.619,
+      "step": 1110
+    },
+    {
+      "epoch": 0.23570081709616594,
+      "grad_norm": 1.4010789394378662,
+      "learning_rate": 9.938344357772087e-05,
+      "loss": 0.6086,
+      "step": 1125
+    },
+    {
+      "epoch": 0.23884349465744814,
+      "grad_norm": 1.6488044261932373,
+      "learning_rate": 9.935660027045506e-05,
+      "loss": 0.551,
+      "step": 1140
+    },
+    {
+      "epoch": 0.24198617221873037,
+      "grad_norm": 1.0560044050216675,
+      "learning_rate": 9.932918878425412e-05,
+      "loss": 0.532,
+      "step": 1155
+    },
+    {
+      "epoch": 0.24512884978001256,
+      "grad_norm": 1.0651888847351074,
+      "learning_rate": 9.930120943467117e-05,
+      "loss": 0.5012,
+      "step": 1170
+    },
+    {
+      "epoch": 0.2482715273412948,
+      "grad_norm": 1.0553079843521118,
+      "learning_rate": 9.927266254379642e-05,
+      "loss": 0.5576,
+      "step": 1185
+    },
+    {
+      "epoch": 0.251414204902577,
+      "grad_norm": 1.007480263710022,
+      "learning_rate": 9.924354844025339e-05,
+      "loss": 0.4839,
+      "step": 1200
+    },
+    {
+      "epoch": 0.2545568824638592,
+      "grad_norm": 1.0924334526062012,
+      "learning_rate": 9.921386745919528e-05,
+      "loss": 0.595,
+      "step": 1215
+    },
+    {
+      "epoch": 0.25769956002514144,
+      "grad_norm": 1.3309390544891357,
+      "learning_rate": 9.918361994230097e-05,
+      "loss": 0.5224,
+      "step": 1230
+    },
+    {
+      "epoch": 0.2608422375864236,
+      "grad_norm": 0.9702763557434082,
+      "learning_rate": 9.915280623777114e-05,
+      "loss": 0.4871,
+      "step": 1245
+    },
+    {
+      "epoch": 0.26398491514770583,
+      "grad_norm": 1.0511876344680786,
+      "learning_rate": 9.912142670032427e-05,
+      "loss": 0.5861,
+      "step": 1260
+    },
+    {
+      "epoch": 0.26712759270898806,
+      "grad_norm": 1.396050214767456,
+      "learning_rate": 9.908948169119251e-05,
+      "loss": 0.4651,
+      "step": 1275
+    },
+    {
+      "epoch": 0.2702702702702703,
+      "grad_norm": 0.985396683216095,
+      "learning_rate": 9.905697157811761e-05,
+      "loss": 0.4302,
+      "step": 1290
+    },
+    {
+      "epoch": 0.27341294783155246,
+      "grad_norm": 0.9169828295707703,
+      "learning_rate": 9.902389673534659e-05,
+      "loss": 0.5212,
+      "step": 1305
+    },
+    {
+      "epoch": 0.2765556253928347,
+      "grad_norm": 0.9107710123062134,
+      "learning_rate": 9.899025754362751e-05,
+      "loss": 0.4941,
+      "step": 1320
+    },
+    {
+      "epoch": 0.2796983029541169,
+      "grad_norm": 0.9720286726951599,
+      "learning_rate": 9.8956054390205e-05,
+      "loss": 0.5169,
+      "step": 1335
+    },
+    {
+      "epoch": 0.28284098051539913,
+      "grad_norm": 1.1490366458892822,
+      "learning_rate": 9.892128766881596e-05,
+      "loss": 0.4973,
+      "step": 1350
+    },
+    {
+      "epoch": 0.28598365807668136,
+      "grad_norm": 1.2628952264785767,
+      "learning_rate": 9.888595777968479e-05,
+      "loss": 0.5194,
+      "step": 1365
+    },
+    {
+      "epoch": 0.2891263356379635,
+      "grad_norm": 1.1610651016235352,
+      "learning_rate": 9.885006512951897e-05,
+      "loss": 0.4994,
+      "step": 1380
+    },
+    {
+      "epoch": 0.29226901319924575,
+      "grad_norm": 1.054768681526184,
+      "learning_rate": 9.881361013150436e-05,
+      "loss": 0.4664,
+      "step": 1395
+    },
+    {
+      "epoch": 0.295411690760528,
+      "grad_norm": 1.0745666027069092,
+      "learning_rate": 9.877659320530037e-05,
+      "loss": 0.5306,
+      "step": 1410
+    },
+    {
+      "epoch": 0.2985543683218102,
+      "grad_norm": 1.3258591890335083,
+      "learning_rate": 9.873901477703516e-05,
+      "loss": 0.5076,
+      "step": 1425
+    },
+    {
+      "epoch": 0.3016970458830924,
+      "grad_norm": 1.222783088684082,
+      "learning_rate": 9.870087527930077e-05,
+      "loss": 0.4581,
+      "step": 1440
+    },
+    {
+      "epoch": 0.3048397234443746,
+      "grad_norm": 0.9374076724052429,
+      "learning_rate": 9.866217515114805e-05,
+      "loss": 0.4643,
+      "step": 1455
+    },
+    {
+      "epoch": 0.3079824010056568,
+      "grad_norm": 1.3485162258148193,
+      "learning_rate": 9.862291483808173e-05,
+      "loss": 0.5551,
+      "step": 1470
+    },
+    {
+      "epoch": 0.31112507856693905,
+      "grad_norm": 0.9162548780441284,
+      "learning_rate": 9.858309479205519e-05,
+      "loss": 0.5592,
+      "step": 1485
+    },
+    {
+      "epoch": 0.3142677561282212,
+      "grad_norm": 1.1385138034820557,
+      "learning_rate": 9.854271547146531e-05,
+      "loss": 0.477,
+      "step": 1500
+    },
+    {
+      "epoch": 0.31741043368950345,
+      "grad_norm": 1.0023164749145508,
+      "learning_rate": 9.850177734114718e-05,
+      "loss": 0.4972,
+      "step": 1515
+    },
+    {
+      "epoch": 0.32055311125078567,
+      "grad_norm": 2.540215492248535,
+      "learning_rate": 9.846028087236873e-05,
+      "loss": 0.5007,
+      "step": 1530
+    },
+    {
+      "epoch": 0.3236957888120679,
+      "grad_norm": 1.2012773752212524,
+      "learning_rate": 9.841822654282533e-05,
+      "loss": 0.5481,
+      "step": 1545
+    },
+    {
+      "epoch": 0.32683846637335007,
+      "grad_norm": 0.9517608284950256,
+      "learning_rate": 9.837561483663429e-05,
+      "loss": 0.567,
+      "step": 1560
+    },
+    {
+      "epoch": 0.3299811439346323,
+      "grad_norm": 1.0308321714401245,
+      "learning_rate": 9.833244624432927e-05,
+      "loss": 0.4856,
+      "step": 1575
+    },
+    {
+      "epoch": 0.3331238214959145,
+      "grad_norm": 1.118574857711792,
+      "learning_rate": 9.828872126285465e-05,
+      "loss": 0.465,
+      "step": 1590
+    },
+    {
+      "epoch": 0.33626649905719674,
+      "grad_norm": 1.0821537971496582,
+      "learning_rate": 9.824444039555977e-05,
+      "loss": 0.4394,
+      "step": 1605
+    },
+    {
+      "epoch": 0.33940917661847897,
+      "grad_norm": 0.8795451521873474,
+      "learning_rate": 9.81996041521932e-05,
+      "loss": 0.4383,
+      "step": 1620
+    },
+    {
+      "epoch": 0.34255185417976114,
+      "grad_norm": 1.1455141305923462,
+      "learning_rate": 9.815421304889687e-05,
+      "loss": 0.4805,
+      "step": 1635
+    },
+    {
+      "epoch": 0.34569453174104336,
+      "grad_norm": 1.1445369720458984,
+      "learning_rate": 9.81082676082e-05,
+      "loss": 0.5315,
+      "step": 1650
+    },
+    {
+      "epoch": 0.3488372093023256,
+      "grad_norm": 1.0800312757492065,
+      "learning_rate": 9.806176835901328e-05,
+      "loss": 0.5205,
+      "step": 1665
+    },
+    {
+      "epoch": 0.3519798868636078,
+      "grad_norm": 0.7038319706916809,
+      "learning_rate": 9.801471583662263e-05,
+      "loss": 0.515,
+      "step": 1680
+    },
+    {
+      "epoch": 0.35512256442489,
+      "grad_norm": 0.9790651202201843,
+      "learning_rate": 9.796711058268313e-05,
+      "loss": 0.504,
+      "step": 1695
+    },
+    {
+      "epoch": 0.3582652419861722,
+      "grad_norm": 1.1764894723892212,
+      "learning_rate": 9.791895314521267e-05,
+      "loss": 0.4806,
+      "step": 1710
+    },
+    {
+      "epoch": 0.36140791954745444,
+      "grad_norm": 0.9900022745132446,
+      "learning_rate": 9.787024407858582e-05,
+      "loss": 0.5358,
+      "step": 1725
+    },
+    {
+      "epoch": 0.36455059710873666,
+      "grad_norm": 0.8621386289596558,
+      "learning_rate": 9.782098394352725e-05,
+      "loss": 0.5494,
+      "step": 1740
+    },
+    {
+      "epoch": 0.36769327467001883,
+      "grad_norm": 0.8717844486236572,
+      "learning_rate": 9.777117330710547e-05,
+      "loss": 0.4967,
+      "step": 1755
+    },
+    {
+      "epoch": 0.37083595223130106,
+      "grad_norm": 0.9800569415092468,
+      "learning_rate": 9.772081274272611e-05,
+      "loss": 0.4538,
+      "step": 1770
+    },
+    {
+      "epoch": 0.3739786297925833,
+      "grad_norm": 0.9540134072303772,
+      "learning_rate": 9.766990283012544e-05,
+      "loss": 0.5149,
+      "step": 1785
+    },
+    {
+      "epoch": 0.3771213073538655,
+      "grad_norm": 1.0856047868728638,
+      "learning_rate": 9.761844415536372e-05,
+      "loss": 0.5042,
+      "step": 1800
+    },
+    {
+      "epoch": 0.3802639849151477,
+      "grad_norm": 1.0914040803909302,
+      "learning_rate": 9.756643731081833e-05,
+      "loss": 0.5059,
+      "step": 1815
+    },
+    {
+      "epoch": 0.3834066624764299,
+      "grad_norm": 1.2371134757995605,
+      "learning_rate": 9.751388289517704e-05,
+      "loss": 0.4506,
+      "step": 1830
+    },
+    {
+      "epoch": 0.38654934003771213,
+      "grad_norm": 1.0402591228485107,
+      "learning_rate": 9.746078151343116e-05,
+      "loss": 0.5535,
+      "step": 1845
+    },
+    {
+      "epoch": 0.38969201759899436,
+      "grad_norm": 0.6260209083557129,
+      "learning_rate": 9.740713377686843e-05,
+      "loss": 0.4436,
+      "step": 1860
+    },
+    {
+      "epoch": 0.3928346951602766,
+      "grad_norm": 0.9588780999183655,
+      "learning_rate": 9.735294030306611e-05,
+      "loss": 0.5573,
+      "step": 1875
+    },
+    {
+      "epoch": 0.39597737272155875,
+      "grad_norm": 1.0838474035263062,
+      "learning_rate": 9.729820171588384e-05,
+      "loss": 0.4627,
+      "step": 1890
+    },
+    {
+      "epoch": 0.399120050282841,
+      "grad_norm": 1.0682798624038696,
+      "learning_rate": 9.724291864545643e-05,
+      "loss": 0.4893,
+      "step": 1905
+    },
+    {
+      "epoch": 0.4022627278441232,
+      "grad_norm": 0.9129301309585571,
+      "learning_rate": 9.718709172818661e-05,
+      "loss": 0.4898,
+      "step": 1920
+    },
+    {
+      "epoch": 0.40540540540540543,
+      "grad_norm": 1.0116883516311646,
+      "learning_rate": 9.713072160673777e-05,
+      "loss": 0.4615,
+      "step": 1935
+    },
+    {
+      "epoch": 0.4085480829666876,
+      "grad_norm": 1.057822823524475,
+      "learning_rate": 9.707380893002646e-05,
+      "loss": 0.4899,
+      "step": 1950
+    },
+    {
+      "epoch": 0.4116907605279698,
+      "grad_norm": 0.6419869661331177,
+      "learning_rate": 9.7016354353215e-05,
+      "loss": 0.4348,
+      "step": 1965
+    },
+    {
+      "epoch": 0.41483343808925205,
+      "grad_norm": 0.961713433265686,
+      "learning_rate": 9.695835853770387e-05,
+      "loss": 0.4921,
+      "step": 1980
+    },
+    {
+      "epoch": 0.4179761156505343,
+      "grad_norm": 0.9473373889923096,
+      "learning_rate": 9.689982215112417e-05,
+      "loss": 0.4926,
+      "step": 1995
+    },
+    {
+      "epoch": 0.42111879321181644,
+      "grad_norm": 1.2034335136413574,
+      "learning_rate": 9.684074586732987e-05,
+      "loss": 0.5042,
+      "step": 2010
+    },
+    {
+      "epoch": 0.42426147077309867,
+      "grad_norm": 0.9373855590820312,
+      "learning_rate": 9.678113036639014e-05,
+      "loss": 0.5076,
+      "step": 2025
+    },
+    {
+      "epoch": 0.4274041483343809,
+      "grad_norm": 1.016756296157837,
+      "learning_rate": 9.672097633458136e-05,
+      "loss": 0.4805,
+      "step": 2040
+    },
+    {
+      "epoch": 0.4305468258956631,
+      "grad_norm": 0.7454690337181091,
+      "learning_rate": 9.666028446437942e-05,
+      "loss": 0.5382,
+      "step": 2055
+    },
+    {
+      "epoch": 0.4336895034569453,
+      "grad_norm": 0.8196286559104919,
+      "learning_rate": 9.659905545445159e-05,
+      "loss": 0.4613,
+      "step": 2070
+    },
+    {
+      "epoch": 0.4368321810182275,
+      "grad_norm": 0.9132091403007507,
+      "learning_rate": 9.653729000964857e-05,
+      "loss": 0.4595,
+      "step": 2085
+    },
+    {
+      "epoch": 0.43997485857950974,
+      "grad_norm": 0.8063992857933044,
+      "learning_rate": 9.647498884099633e-05,
+      "loss": 0.4139,
+      "step": 2100
+    },
+    {
+      "epoch": 0.44311753614079197,
+      "grad_norm": 0.9756997227668762,
+      "learning_rate": 9.641215266568794e-05,
+      "loss": 0.3941,
+      "step": 2115
+    },
+    {
+      "epoch": 0.4462602137020742,
+      "grad_norm": 0.6542510390281677,
+      "learning_rate": 9.634878220707531e-05,
+      "loss": 0.4768,
+      "step": 2130
+    },
+    {
+      "epoch": 0.44940289126335636,
+      "grad_norm": 0.9039008617401123,
+      "learning_rate": 9.628487819466086e-05,
+      "loss": 0.4248,
+      "step": 2145
+    },
+    {
+      "epoch": 0.4525455688246386,
+      "grad_norm": 1.1151047945022583,
+      "learning_rate": 9.622044136408914e-05,
+      "loss": 0.5041,
+      "step": 2160
+    },
+    {
+      "epoch": 0.4556882463859208,
+      "grad_norm": 0.8580663800239563,
+      "learning_rate": 9.615547245713836e-05,
+      "loss": 0.4766,
+      "step": 2175
+    },
+    {
+      "epoch": 0.45883092394720304,
+      "grad_norm": 0.9799042344093323,
+      "learning_rate": 9.608997222171178e-05,
+      "loss": 0.4714,
+      "step": 2190
+    },
+    {
+      "epoch": 0.4619736015084852,
+      "grad_norm": 0.8485172986984253,
+      "learning_rate": 9.602394141182927e-05,
+      "loss": 0.4556,
+      "step": 2205
+    },
+    {
+      "epoch": 0.46511627906976744,
+      "grad_norm": 0.9632934927940369,
+      "learning_rate": 9.595738078761837e-05,
+      "loss": 0.4791,
+      "step": 2220
+    },
+    {
+      "epoch": 0.46825895663104966,
+      "grad_norm": 0.8843478560447693,
+      "learning_rate": 9.589029111530586e-05,
+      "loss": 0.4603,
+      "step": 2235
+    },
+    {
+      "epoch": 0.4714016341923319,
+      "grad_norm": 1.1230348348617554,
+      "learning_rate": 9.582267316720861e-05,
+      "loss": 0.491,
+      "step": 2250
+    },
+    {
+      "epoch": 0.47454431175361406,
+      "grad_norm": 0.8234013915061951,
+      "learning_rate": 9.575452772172495e-05,
+      "loss": 0.44,
+      "step": 2265
+    },
+    {
+      "epoch": 0.4776869893148963,
+      "grad_norm": 0.6838919520378113,
+      "learning_rate": 9.568585556332559e-05,
+      "loss": 0.4456,
+      "step": 2280
+    },
+    {
+      "epoch": 0.4808296668761785,
+      "grad_norm": 0.8424423336982727,
+      "learning_rate": 9.561665748254456e-05,
+      "loss": 0.4556,
+      "step": 2295
+    },
+    {
+      "epoch": 0.48397234443746073,
+      "grad_norm": 0.6735498905181885,
+      "learning_rate": 9.554693427597024e-05,
+      "loss": 0.5184,
+      "step": 2310
+    },
+    {
+      "epoch": 0.4871150219987429,
+      "grad_norm": 0.8868768811225891,
+      "learning_rate": 9.5476686746236e-05,
+      "loss": 0.5403,
+      "step": 2325
+    },
+    {
+      "epoch": 0.49025769956002513,
+      "grad_norm": 0.9957670569419861,
+      "learning_rate": 9.540591570201116e-05,
+      "loss": 0.4997,
+      "step": 2340
+    },
+    {
+      "epoch": 0.49340037712130735,
+      "grad_norm": 0.76320481300354,
+      "learning_rate": 9.533462195799157e-05,
+      "loss": 0.4534,
+      "step": 2355
+    },
+    {
+      "epoch": 0.4965430546825896,
+      "grad_norm": 0.8841500282287598,
+      "learning_rate": 9.526280633489018e-05,
+      "loss": 0.4724,
+      "step": 2370
+    },
+    {
+      "epoch": 0.4996857322438718,
+      "grad_norm": 0.8852142095565796,
+      "learning_rate": 9.519046965942776e-05,
+      "loss": 0.4655,
+      "step": 2385
+    },
+    {
+      "epoch": 0.502828409805154,
+      "grad_norm": 0.839430570602417,
+      "learning_rate": 9.511761276432321e-05,
+      "loss": 0.4386,
+      "step": 2400
+    },
+    {
+      "epoch": 0.5059710873664363,
+      "grad_norm": 0.7581266760826111,
+      "learning_rate": 9.50442364882841e-05,
+      "loss": 0.4774,
+      "step": 2415
+    },
+    {
+      "epoch": 0.5091137649277184,
+      "grad_norm": 0.8754017949104309,
+      "learning_rate": 9.497034167599691e-05,
+      "loss": 0.4744,
+      "step": 2430
+    },
+    {
+      "epoch": 0.5122564424890006,
+      "grad_norm": 0.9099476337432861,
+      "learning_rate": 9.48959291781174e-05,
+      "loss": 0.4292,
+      "step": 2445
+    },
+    {
+      "epoch": 0.5153991200502829,
+      "grad_norm": 0.9721155166625977,
+      "learning_rate": 9.482099985126079e-05,
+      "loss": 0.4137,
+      "step": 2460
+    },
+    {
+      "epoch": 0.518541797611565,
+      "grad_norm": 0.8385334014892578,
+      "learning_rate": 9.474555455799181e-05,
+      "loss": 0.471,
+      "step": 2475
+    },
+    {
+      "epoch": 0.5216844751728472,
+      "grad_norm": 0.9853966236114502,
+      "learning_rate": 9.466959416681495e-05,
+      "loss": 0.4233,
+      "step": 2490
+    },
+    {
+      "epoch": 0.5248271527341295,
+      "grad_norm": 1.1044224500656128,
+      "learning_rate": 9.459311955216428e-05,
+      "loss": 0.5188,
+      "step": 2505
+    },
+    {
+      "epoch": 0.5279698302954117,
+      "grad_norm": 0.870677649974823,
+      "learning_rate": 9.451613159439349e-05,
+      "loss": 0.4676,
+      "step": 2520
+    },
+    {
+      "epoch": 0.531112507856694,
+      "grad_norm": 0.8571140170097351,
+      "learning_rate": 9.443863117976573e-05,
+      "loss": 0.4863,
+      "step": 2535
+    },
+    {
+      "epoch": 0.5342551854179761,
+      "grad_norm": 1.0573495626449585,
+      "learning_rate": 9.436061920044341e-05,
+      "loss": 0.5057,
+      "step": 2550
+    },
+    {
+      "epoch": 0.5373978629792583,
+      "grad_norm": 0.9805963635444641,
+      "learning_rate": 9.42820965544779e-05,
+      "loss": 0.468,
+      "step": 2565
+    },
+    {
+      "epoch": 0.5405405405405406,
+      "grad_norm": 0.8198602199554443,
+      "learning_rate": 9.420306414579925e-05,
+      "loss": 0.5054,
+      "step": 2580
+    },
+    {
+      "epoch": 0.5436832181018227,
+      "grad_norm": 0.9718137979507446,
+      "learning_rate": 9.412352288420572e-05,
+      "loss": 0.4824,
+      "step": 2595
+    },
+    {
+      "epoch": 0.5468258956631049,
+      "grad_norm": 1.0223153829574585,
+      "learning_rate": 9.404347368535337e-05,
+      "loss": 0.4502,
+      "step": 2610
+    },
+    {
+      "epoch": 0.5499685732243872,
+      "grad_norm": 0.9398010969161987,
+      "learning_rate": 9.396291747074547e-05,
+      "loss": 0.4761,
+      "step": 2625
+    },
+    {
+      "epoch": 0.5531112507856694,
+      "grad_norm": 0.9091777801513672,
+      "learning_rate": 9.38818551677219e-05,
+      "loss": 0.4033,
+      "step": 2640
+    },
+    {
+      "epoch": 0.5562539283469516,
+      "grad_norm": 1.06580650806427,
+      "learning_rate": 9.380028770944849e-05,
+      "loss": 0.4052,
+      "step": 2655
+    },
+    {
+      "epoch": 0.5593966059082338,
+      "grad_norm": 0.7236329913139343,
+      "learning_rate": 9.371821603490627e-05,
+      "loss": 0.4677,
+      "step": 2670
+    },
+    {
+      "epoch": 0.562539283469516,
+      "grad_norm": 0.8263210654258728,
+      "learning_rate": 9.363564108888069e-05,
+      "loss": 0.4576,
+      "step": 2685
+    },
+    {
+      "epoch": 0.5656819610307983,
+      "grad_norm": 1.022448182106018,
+      "learning_rate": 9.355256382195068e-05,
+      "loss": 0.4963,
+      "step": 2700
+    },
+    {
+      "epoch": 0.5688246385920804,
+      "grad_norm": 0.9639766812324524,
+      "learning_rate": 9.346898519047775e-05,
+      "loss": 0.4113,
+      "step": 2715
+    },
+    {
+      "epoch": 0.5719673161533627,
+      "grad_norm": 1.1044561862945557,
+      "learning_rate": 9.338490615659499e-05,
+      "loss": 0.5023,
+      "step": 2730
+    },
+    {
+      "epoch": 0.5751099937146449,
+      "grad_norm": 0.8272239565849304,
+      "learning_rate": 9.330032768819596e-05,
+      "loss": 0.4699,
+      "step": 2745
+    },
+    {
+      "epoch": 0.578252671275927,
+      "grad_norm": 0.7692523002624512,
+      "learning_rate": 9.321525075892356e-05,
+      "loss": 0.4292,
+      "step": 2760
+    },
+    {
+      "epoch": 0.5813953488372093,
+      "grad_norm": 0.9032982587814331,
+      "learning_rate": 9.312967634815888e-05,
+      "loss": 0.4432,
+      "step": 2775
+    },
+    {
+      "epoch": 0.5845380263984915,
+      "grad_norm": 0.7676737904548645,
+      "learning_rate": 9.304360544100982e-05,
+      "loss": 0.4311,
+      "step": 2790
+    },
+    {
+      "epoch": 0.5876807039597737,
+      "grad_norm": 0.9019532799720764,
+      "learning_rate": 9.29570390282998e-05,
+      "loss": 0.4464,
+      "step": 2805
+    },
+    {
+      "epoch": 0.590823381521056,
+      "grad_norm": 0.9738386869430542,
+      "learning_rate": 9.286997810655638e-05,
+      "loss": 0.5019,
+      "step": 2820
+    },
+    {
+      "epoch": 0.5939660590823381,
+      "grad_norm": 0.7886769771575928,
+      "learning_rate": 9.278242367799978e-05,
+      "loss": 0.4919,
+      "step": 2835
+    },
+    {
+      "epoch": 0.5971087366436204,
+      "grad_norm": 0.9002622365951538,
+      "learning_rate": 9.269437675053129e-05,
+      "loss": 0.4695,
+      "step": 2850
+    },
+    {
+      "epoch": 0.6002514142049026,
+      "grad_norm": 0.7023227214813232,
+      "learning_rate": 9.260583833772172e-05,
+      "loss": 0.4338,
+      "step": 2865
+    },
+    {
+      "epoch": 0.6033940917661847,
+      "grad_norm": 0.9442479014396667,
+      "learning_rate": 9.251680945879975e-05,
+      "loss": 0.4907,
+      "step": 2880
+    },
+    {
+      "epoch": 0.606536769327467,
+      "grad_norm": 0.6304488778114319,
+      "learning_rate": 9.24272911386401e-05,
+      "loss": 0.4612,
+      "step": 2895
+    },
+    {
+      "epoch": 0.6096794468887492,
+      "grad_norm": 0.731960117816925,
+      "learning_rate": 9.233728440775185e-05,
+      "loss": 0.4207,
+      "step": 2910
+    },
+    {
+      "epoch": 0.6128221244500315,
+      "grad_norm": 1.083849549293518,
+      "learning_rate": 9.224679030226648e-05,
+      "loss": 0.4775,
+      "step": 2925
+    },
+    {
+      "epoch": 0.6159648020113137,
+      "grad_norm": 0.6792687177658081,
+      "learning_rate": 9.215580986392607e-05,
+      "loss": 0.4708,
+      "step": 2940
+    },
+    {
+      "epoch": 0.6191074795725958,
+      "grad_norm": 0.7582160830497742,
+      "learning_rate": 9.20643441400711e-05,
+      "loss": 0.4352,
+      "step": 2955
+    },
+    {
+      "epoch": 0.6222501571338781,
+      "grad_norm": 0.7785065174102783,
+      "learning_rate": 9.197239418362862e-05,
+      "loss": 0.4199,
+      "step": 2970
+    },
+    {
+      "epoch": 0.6253928346951603,
+      "grad_norm": 0.9076778292655945,
+      "learning_rate": 9.187996105309995e-05,
+      "loss": 0.4937,
+      "step": 2985
+    },
+    {
+      "epoch": 0.6285355122564424,
+      "grad_norm": 0.9189762473106384,
+      "learning_rate": 9.178704581254865e-05,
+      "loss": 0.4553,
+      "step": 3000
+    },
+    {
+      "epoch": 0.6316781898177247,
+      "grad_norm": 0.8485803008079529,
+      "learning_rate": 9.169364953158812e-05,
+      "loss": 0.4799,
+      "step": 3015
+    },
+    {
+      "epoch": 0.6348208673790069,
+      "grad_norm": 0.8296557068824768,
+      "learning_rate": 9.15997732853694e-05,
+      "loss": 0.4799,
+      "step": 3030
+    },
+    {
+      "epoch": 0.6379635449402892,
+      "grad_norm": 0.9346463680267334,
+      "learning_rate": 9.150541815456874e-05,
+      "loss": 0.4707,
+      "step": 3045
+    },
+    {
+      "epoch": 0.6411062225015713,
+      "grad_norm": 1.0045510530471802,
+      "learning_rate": 9.141058522537515e-05,
+      "loss": 0.5216,
+      "step": 3060
+    },
+    {
+      "epoch": 0.6442489000628535,
+      "grad_norm": 0.5840141773223877,
+      "learning_rate": 9.131527558947796e-05,
+      "loss": 0.429,
+      "step": 3075
+    },
+    {
+      "epoch": 0.6473915776241358,
+      "grad_norm": 0.8743481040000916,
+      "learning_rate": 9.121949034405417e-05,
+      "loss": 0.4734,
+      "step": 3090
+    },
+    {
+      "epoch": 0.650534255185418,
+      "grad_norm": 0.9631288051605225,
+      "learning_rate": 9.112323059175588e-05,
+      "loss": 0.4856,
+      "step": 3105
+    },
+    {
+      "epoch": 0.6536769327467001,
+      "grad_norm": 0.7583104372024536,
+      "learning_rate": 9.102649744069758e-05,
+      "loss": 0.4428,
+      "step": 3120
+    },
+    {
+      "epoch": 0.6568196103079824,
+      "grad_norm": 0.9227087497711182,
+      "learning_rate": 9.092929200444337e-05,
+      "loss": 0.4622,
+      "step": 3135
+    },
+    {
+      "epoch": 0.6599622878692646,
+      "grad_norm": 0.720124363899231,
+      "learning_rate": 9.083161540199417e-05,
+      "loss": 0.4136,
+      "step": 3150
+    },
+    {
+      "epoch": 0.6631049654305469,
+      "grad_norm": 0.6481117010116577,
+      "learning_rate": 9.073346875777487e-05,
+      "loss": 0.5445,
+      "step": 3165
+    },
+    {
+      "epoch": 0.666247642991829,
+      "grad_norm": 0.6970652937889099,
+      "learning_rate": 9.063485320162126e-05,
+      "loss": 0.4247,
+      "step": 3180
+    },
+    {
+      "epoch": 0.6693903205531112,
+      "grad_norm": 0.5132230520248413,
+      "learning_rate": 9.053576986876718e-05,
+      "loss": 0.4415,
+      "step": 3195
+    },
+    {
+      "epoch": 0.6725329981143935,
+      "grad_norm": 0.7673790454864502,
+      "learning_rate": 9.043621989983135e-05,
+      "loss": 0.5188,
+      "step": 3210
+    },
+    {
+      "epoch": 0.6756756756756757,
+      "grad_norm": 0.8441967368125916,
+      "learning_rate": 9.033620444080428e-05,
+      "loss": 0.4343,
+      "step": 3225
+    },
+    {
+      "epoch": 0.6788183532369579,
+      "grad_norm": 0.8746171593666077,
+      "learning_rate": 9.023572464303506e-05,
+      "loss": 0.4114,
+      "step": 3240
+    },
+    {
+      "epoch": 0.6819610307982401,
+      "grad_norm": 0.7494221925735474,
+      "learning_rate": 9.013478166321812e-05,
+      "loss": 0.4334,
+      "step": 3255
+    },
+    {
+      "epoch": 0.6851037083595223,
+      "grad_norm": 0.7263948917388916,
+      "learning_rate": 9.00333766633799e-05,
+      "loss": 0.4322,
+      "step": 3270
+    },
+    {
+      "epoch": 0.6882463859208046,
+      "grad_norm": 0.852172315120697,
+      "learning_rate": 8.99315108108655e-05,
+      "loss": 0.4506,
+      "step": 3285
+    },
+    {
+      "epoch": 0.6913890634820867,
+      "grad_norm": 0.7959320545196533,
+      "learning_rate": 8.98291852783252e-05,
+      "loss": 0.4456,
+      "step": 3300
+    },
+    {
+      "epoch": 0.6945317410433689,
+      "grad_norm": 0.5918748378753662,
+      "learning_rate": 8.9726401243701e-05,
+      "loss": 0.4181,
+      "step": 3315
+    },
+    {
+      "epoch": 0.6976744186046512,
+      "grad_norm": 0.9726805090904236,
+      "learning_rate": 8.962315989021304e-05,
+      "loss": 0.4964,
+      "step": 3330
+    },
+    {
+      "epoch": 0.7008170961659334,
+      "grad_norm": 0.8826568126678467,
+      "learning_rate": 8.951946240634596e-05,
+      "loss": 0.4702,
+      "step": 3345
+    },
+    {
+      "epoch": 0.7039597737272156,
+      "grad_norm": 0.7354099154472351,
+      "learning_rate": 8.941530998583527e-05,
+      "loss": 0.4258,
+      "step": 3360
+    },
+    {
+      "epoch": 0.7071024512884978,
+      "grad_norm": 0.9217835664749146,
+      "learning_rate": 8.931070382765359e-05,
+      "loss": 0.5185,
+      "step": 3375
+    },
+    {
+      "epoch": 0.71024512884978,
+      "grad_norm": 0.7444872260093689,
+      "learning_rate": 8.920564513599679e-05,
+      "loss": 0.4534,
+      "step": 3390
+    },
+    {
+      "epoch": 0.7133878064110623,
+      "grad_norm": 0.7847276926040649,
+      "learning_rate": 8.910013512027022e-05,
+      "loss": 0.4232,
+      "step": 3405
+    },
+    {
+      "epoch": 0.7165304839723444,
+      "grad_norm": 0.8024355173110962,
+      "learning_rate": 8.899417499507471e-05,
+      "loss": 0.4579,
+      "step": 3420
+    },
+    {
+      "epoch": 0.7196731615336267,
+      "grad_norm": 0.7088613510131836,
+      "learning_rate": 8.888776598019266e-05,
+      "loss": 0.4437,
+      "step": 3435
+    },
+    {
+      "epoch": 0.7228158390949089,
+      "grad_norm": 0.6009235382080078,
+      "learning_rate": 8.87809093005739e-05,
+      "loss": 0.397,
+      "step": 3450
+    },
+    {
+      "epoch": 0.725958516656191,
+      "grad_norm": 0.8743120431900024,
+      "learning_rate": 8.867360618632172e-05,
+      "loss": 0.5056,
+      "step": 3465
+    },
+    {
+      "epoch": 0.7291011942174733,
+      "grad_norm": 0.899148166179657,
+      "learning_rate": 8.856585787267856e-05,
+      "loss": 0.4521,
+      "step": 3480
+    },
+    {
+      "epoch": 0.7322438717787555,
+      "grad_norm": 0.8690171837806702,
+      "learning_rate": 8.845766560001193e-05,
+      "loss": 0.4708,
+      "step": 3495
+    },
+    {
+      "epoch": 0.7353865493400377,
+      "grad_norm": 0.9699186682701111,
+      "learning_rate": 8.834903061380002e-05,
+      "loss": 0.4534,
+      "step": 3510
+    },
+    {
+      "epoch": 0.73852922690132,
+      "grad_norm": 0.8577262163162231,
+      "learning_rate": 8.823995416461744e-05,
+      "loss": 0.4096,
+      "step": 3525
+    },
+    {
+      "epoch": 0.7416719044626021,
+      "grad_norm": 0.7458922266960144,
+      "learning_rate": 8.81304375081208e-05,
+      "loss": 0.46,
+      "step": 3540
+    },
+    {
+      "epoch": 0.7448145820238844,
+      "grad_norm": 0.7347140908241272,
+      "learning_rate": 8.802048190503423e-05,
+      "loss": 0.4684,
+      "step": 3555
+    },
+    {
+      "epoch": 0.7479572595851666,
+      "grad_norm": 0.7161451578140259,
+      "learning_rate": 8.79100886211349e-05,
+      "loss": 0.4715,
+      "step": 3570
+    },
+    {
+      "epoch": 0.7510999371464487,
+      "grad_norm": 0.8321588039398193,
+      "learning_rate": 8.779925892723842e-05,
+      "loss": 0.3598,
+      "step": 3585
+    },
+    {
+      "epoch": 0.754242614707731,
+      "grad_norm": 0.9462142586708069,
+      "learning_rate": 8.768799409918423e-05,
+      "loss": 0.4404,
+      "step": 3600
+    },
+    {
+      "epoch": 0.7573852922690132,
+      "grad_norm": 0.6842710971832275,
+      "learning_rate": 8.75762954178209e-05,
+      "loss": 0.4648,
+      "step": 3615
+    },
+    {
+      "epoch": 0.7605279698302954,
+      "grad_norm": 0.8573241829872131,
+      "learning_rate": 8.746416416899145e-05,
+      "loss": 0.4592,
+      "step": 3630
+    },
+    {
+      "epoch": 0.7636706473915776,
+      "grad_norm": 0.751291811466217,
+      "learning_rate": 8.735160164351841e-05,
+      "loss": 0.5319,
+      "step": 3645
+    },
+    {
+      "epoch": 0.7668133249528598,
+      "grad_norm": 0.731086790561676,
+      "learning_rate": 8.72386091371891e-05,
+      "loss": 0.4629,
+      "step": 3660
+    },
+    {
+      "epoch": 0.7699560025141421,
+      "grad_norm": 0.9289976358413696,
+      "learning_rate": 8.712518795074063e-05,
+      "loss": 0.4427,
+      "step": 3675
+    },
+    {
+      "epoch": 0.7730986800754243,
+      "grad_norm": 0.7036064267158508,
+      "learning_rate": 8.701133938984496e-05,
+      "loss": 0.4679,
+      "step": 3690
+    },
+    {
+      "epoch": 0.7762413576367064,
+      "grad_norm": 0.778161346912384,
+      "learning_rate": 8.689706476509385e-05,
+      "loss": 0.4489,
+      "step": 3705
+    },
+    {
+      "epoch": 0.7793840351979887,
+      "grad_norm": 0.8694556951522827,
+      "learning_rate": 8.678236539198382e-05,
+      "loss": 0.4048,
+      "step": 3720
+    },
+    {
+      "epoch": 0.7825267127592709,
+      "grad_norm": 0.5768362283706665,
+      "learning_rate": 8.666724259090092e-05,
+      "loss": 0.4434,
+      "step": 3735
+    },
+    {
+      "epoch": 0.7856693903205532,
+      "grad_norm": 0.604917585849762,
+      "learning_rate": 8.655169768710562e-05,
+      "loss": 0.4669,
+      "step": 3750
+    },
+    {
+      "epoch": 0.7888120678818353,
+      "grad_norm": 0.833985447883606,
+      "learning_rate": 8.643573201071748e-05,
+      "loss": 0.4267,
+      "step": 3765
+    },
+    {
+      "epoch": 0.7919547454431175,
+      "grad_norm": 0.7951568365097046,
+      "learning_rate": 8.631934689669992e-05,
+      "loss": 0.4028,
+      "step": 3780
+    },
+    {
+      "epoch": 0.7950974230043998,
+      "grad_norm": 0.7703410983085632,
+      "learning_rate": 8.620254368484474e-05,
+      "loss": 0.4153,
+      "step": 3795
+    },
+    {
+      "epoch": 0.798240100565682,
+      "grad_norm": 0.8545910716056824,
+      "learning_rate": 8.608532371975684e-05,
+      "loss": 0.4949,
+      "step": 3810
+    },
+    {
+      "epoch": 0.8013827781269641,
+      "grad_norm": 0.8206099271774292,
+      "learning_rate": 8.59676883508386e-05,
+      "loss": 0.4714,
+      "step": 3825
+    },
+    {
+      "epoch": 0.8045254556882464,
+      "grad_norm": 0.7841479182243347,
+      "learning_rate": 8.584963893227442e-05,
+      "loss": 0.4888,
+      "step": 3840
+    },
+    {
+      "epoch": 0.8076681332495286,
+      "grad_norm": 0.7417731285095215,
+      "learning_rate": 8.573117682301514e-05,
+      "loss": 0.4951,
+      "step": 3855
+    },
+    {
+      "epoch": 0.8108108108108109,
+      "grad_norm": 0.9013925194740295,
+      "learning_rate": 8.561230338676239e-05,
+      "loss": 0.4542,
+      "step": 3870
+    },
+    {
+      "epoch": 0.813953488372093,
+      "grad_norm": 1.2146642208099365,
+      "learning_rate": 8.549301999195283e-05,
+      "loss": 0.4606,
+      "step": 3885
+    },
+    {
+      "epoch": 0.8170961659333752,
+      "grad_norm": 0.8740483522415161,
+      "learning_rate": 8.537332801174245e-05,
+      "loss": 0.4562,
+      "step": 3900
+    },
+    {
+      "epoch": 0.8202388434946575,
+      "grad_norm": 0.7769590020179749,
+      "learning_rate": 8.525322882399082e-05,
+      "loss": 0.4385,
+      "step": 3915
+    },
+    {
+      "epoch": 0.8233815210559396,
+      "grad_norm": 0.7966271042823792,
+      "learning_rate": 8.513272381124511e-05,
+      "loss": 0.4011,
+      "step": 3930
+    },
+    {
+      "epoch": 0.8265241986172219,
+      "grad_norm": 0.6132526397705078,
+      "learning_rate": 8.501181436072422e-05,
+      "loss": 0.393,
+      "step": 3945
+    },
+    {
+      "epoch": 0.8296668761785041,
+      "grad_norm": 0.6438138484954834,
+      "learning_rate": 8.489050186430285e-05,
+      "loss": 0.4226,
+      "step": 3960
+    },
+    {
+      "epoch": 0.8328095537397863,
+      "grad_norm": 0.8362025022506714,
+      "learning_rate": 8.476878771849545e-05,
+      "loss": 0.4216,
+      "step": 3975
+    },
+    {
+      "epoch": 0.8359522313010685,
+      "grad_norm": 0.770706057548523,
+      "learning_rate": 8.464667332444012e-05,
+      "loss": 0.4278,
+      "step": 3990
+    },
+    {
+      "epoch": 0.8390949088623507,
+      "grad_norm": 0.8944802284240723,
+      "learning_rate": 8.452416008788254e-05,
+      "loss": 0.4609,
+      "step": 4005
+    },
+    {
+      "epoch": 0.8422375864236329,
+      "grad_norm": 0.9292035102844238,
+      "learning_rate": 8.440124941915972e-05,
+      "loss": 0.4124,
+      "step": 4020
+    },
+    {
+      "epoch": 0.8453802639849152,
+      "grad_norm": 0.6450730562210083,
+      "learning_rate": 8.427794273318377e-05,
+      "loss": 0.4124,
+      "step": 4035
+    },
+    {
+      "epoch": 0.8485229415461973,
+      "grad_norm": 1.0732468366622925,
+      "learning_rate": 8.415424144942569e-05,
+      "loss": 0.4678,
+      "step": 4050
+    },
+    {
+      "epoch": 0.8516656191074796,
+      "grad_norm": 0.900360107421875,
+      "learning_rate": 8.403014699189892e-05,
+      "loss": 0.4299,
+      "step": 4065
+    },
+    {
+      "epoch": 0.8548082966687618,
+      "grad_norm": 0.7163972854614258,
+      "learning_rate": 8.39056607891431e-05,
+      "loss": 0.4651,
+      "step": 4080
+    },
+    {
+      "epoch": 0.857950974230044,
+      "grad_norm": 0.6078224182128906,
+      "learning_rate": 8.378078427420739e-05,
+      "loss": 0.4612,
+      "step": 4095
+    },
+    {
+      "epoch": 0.8610936517913262,
+      "grad_norm": 0.7975668907165527,
+      "learning_rate": 8.365551888463423e-05,
+      "loss": 0.4521,
+      "step": 4110
+    },
+    {
+      "epoch": 0.8642363293526084,
+      "grad_norm": 0.7620348930358887,
+      "learning_rate": 8.352986606244262e-05,
+      "loss": 0.4527,
+      "step": 4125
+    },
+    {
+      "epoch": 0.8673790069138906,
+      "grad_norm": 0.7811437249183655,
+      "learning_rate": 8.340382725411155e-05,
+      "loss": 0.4639,
+      "step": 4140
+    },
+    {
+      "epoch": 0.8705216844751729,
+      "grad_norm": 0.46538805961608887,
+      "learning_rate": 8.327740391056343e-05,
+      "loss": 0.3793,
+      "step": 4155
+    },
+    {
+      "epoch": 0.873664362036455,
+      "grad_norm": 0.893225371837616,
+      "learning_rate": 8.315059748714728e-05,
+      "loss": 0.4824,
+      "step": 4170
+    },
+    {
+      "epoch": 0.8768070395977373,
+      "grad_norm": 0.8325145244598389,
+      "learning_rate": 8.302340944362205e-05,
+      "loss": 0.4623,
+      "step": 4185
+    },
+    {
+      "epoch": 0.8799497171590195,
+      "grad_norm": 0.7328510880470276,
+      "learning_rate": 8.289584124413978e-05,
+      "loss": 0.4075,
+      "step": 4200
+    },
+    {
+      "epoch": 0.8830923947203017,
+      "grad_norm": 0.35754507780075073,
+      "learning_rate": 8.276789435722875e-05,
+      "loss": 0.3328,
+      "step": 4215
+    },
+    {
+      "epoch": 0.8862350722815839,
+      "grad_norm": 0.78349369764328,
+      "learning_rate": 8.263957025577663e-05,
+      "loss": 0.4962,
+      "step": 4230
+    },
+    {
+      "epoch": 0.8893777498428661,
+      "grad_norm": 0.644481360912323,
+      "learning_rate": 8.251087041701339e-05,
+      "loss": 0.3977,
+      "step": 4245
+    },
+    {
+      "epoch": 0.8925204274041484,
+      "grad_norm": 0.618881344795227,
+      "learning_rate": 8.238179632249443e-05,
+      "loss": 0.3967,
+      "step": 4260
+    },
+    {
+      "epoch": 0.8956631049654306,
+      "grad_norm": 0.7603642344474792,
+      "learning_rate": 8.22523494580835e-05,
+      "loss": 0.4413,
+      "step": 4275
+    },
+    {
+      "epoch": 0.8988057825267127,
+      "grad_norm": 0.6301630735397339,
+      "learning_rate": 8.212253131393549e-05,
+      "loss": 0.4333,
+      "step": 4290
+    },
+    {
+      "epoch": 0.901948460087995,
+      "grad_norm": 0.7729358077049255,
+      "learning_rate": 8.199234338447942e-05,
+      "loss": 0.4633,
+      "step": 4305
+    },
+    {
+      "epoch": 0.9050911376492772,
+      "grad_norm": 0.9121199250221252,
+      "learning_rate": 8.186178716840118e-05,
+      "loss": 0.4411,
+      "step": 4320
+    },
+    {
+      "epoch": 0.9082338152105593,
+      "grad_norm": 0.5462374091148376,
+      "learning_rate": 8.17308641686262e-05,
+      "loss": 0.4659,
+      "step": 4335
+    },
+    {
+      "epoch": 0.9113764927718416,
+      "grad_norm": 0.7599003911018372,
+      "learning_rate": 8.15995758923023e-05,
+      "loss": 0.4015,
+      "step": 4350
+    },
+    {
+      "epoch": 0.9145191703331238,
+      "grad_norm": 0.8557884693145752,
+      "learning_rate": 8.14679238507822e-05,
+      "loss": 0.4574,
+      "step": 4365
+    },
+    {
+      "epoch": 0.9176618478944061,
+      "grad_norm": 0.7987812757492065,
+      "learning_rate": 8.133590955960619e-05,
+      "loss": 0.4501,
+      "step": 4380
+    },
+    {
+      "epoch": 0.9208045254556882,
+      "grad_norm": 0.8603717088699341,
+      "learning_rate": 8.120353453848471e-05,
+      "loss": 0.4201,
+      "step": 4395
+    },
+    {
+      "epoch": 0.9239472030169704,
+      "grad_norm": 0.7066472768783569,
+      "learning_rate": 8.107080031128078e-05,
+      "loss": 0.4035,
+      "step": 4410
+    },
+    {
+      "epoch": 0.9270898805782527,
+      "grad_norm": 0.6430373191833496,
+      "learning_rate": 8.09377084059925e-05,
+      "loss": 0.4141,
+      "step": 4425
+    },
+    {
+      "epoch": 0.9302325581395349,
+      "grad_norm": 0.6911259889602661,
+      "learning_rate": 8.080426035473549e-05,
+      "loss": 0.4431,
+      "step": 4440
+    },
+    {
+      "epoch": 0.933375235700817,
+      "grad_norm": 0.8445611000061035,
+      "learning_rate": 8.067045769372515e-05,
+      "loss": 0.4469,
+      "step": 4455
+    },
+    {
+      "epoch": 0.9365179132620993,
+      "grad_norm": 0.9317618012428284,
+      "learning_rate": 8.053630196325914e-05,
+      "loss": 0.4051,
+      "step": 4470
+    },
+    {
+      "epoch": 0.9396605908233815,
+      "grad_norm": 0.8286532163619995,
+      "learning_rate": 8.040179470769946e-05,
+      "loss": 0.4158,
+      "step": 4485
+    },
+    {
+      "epoch": 0.9428032683846638,
+      "grad_norm": 0.7000495195388794,
+      "learning_rate": 8.026693747545486e-05,
+      "loss": 0.4202,
+      "step": 4500
+    },
+    {
+      "epoch": 0.9459459459459459,
+      "grad_norm": 0.8104173541069031,
+      "learning_rate": 8.013173181896283e-05,
+      "loss": 0.4369,
+      "step": 4515
+    },
+    {
+      "epoch": 0.9490886235072281,
+      "grad_norm": 0.864750862121582,
+      "learning_rate": 7.999617929467187e-05,
+      "loss": 0.4152,
+      "step": 4530
+    },
+    {
+      "epoch": 0.9522313010685104,
+      "grad_norm": 0.7788864970207214,
+      "learning_rate": 7.98602814630235e-05,
+      "loss": 0.492,
+      "step": 4545
+    },
+    {
+      "epoch": 0.9553739786297926,
+      "grad_norm": 0.707156777381897,
+      "learning_rate": 7.972403988843435e-05,
+      "loss": 0.4105,
+      "step": 4560
+    },
+    {
+      "epoch": 0.9585166561910748,
+      "grad_norm": 0.8454593420028687,
+      "learning_rate": 7.958745613927809e-05,
+      "loss": 0.4622,
+      "step": 4575
+    },
+    {
+      "epoch": 0.961659333752357,
+      "grad_norm": 0.8026373982429504,
+      "learning_rate": 7.945053178786744e-05,
+      "loss": 0.4236,
+      "step": 4590
+    },
+    {
+      "epoch": 0.9648020113136392,
+      "grad_norm": 0.786409318447113,
+      "learning_rate": 7.931326841043596e-05,
+      "loss": 0.4677,
+      "step": 4605
+    },
+    {
+      "epoch": 0.9679446888749215,
+      "grad_norm": 0.5381405353546143,
+      "learning_rate": 7.917566758712005e-05,
+      "loss": 0.443,
+      "step": 4620
+    },
+    {
+      "epoch": 0.9710873664362036,
+      "grad_norm": 0.6609058380126953,
+      "learning_rate": 7.903773090194069e-05,
+      "loss": 0.4573,
+      "step": 4635
+    },
+    {
+      "epoch": 0.9742300439974858,
+      "grad_norm": 0.7192760705947876,
+      "learning_rate": 7.889945994278514e-05,
+      "loss": 0.4387,
+      "step": 4650
+    },
+    {
+      "epoch": 0.9773727215587681,
+      "grad_norm": 0.7502164244651794,
+      "learning_rate": 7.87608563013888e-05,
+      "loss": 0.399,
+      "step": 4665
+    },
+    {
+      "epoch": 0.9805153991200503,
+      "grad_norm": 0.7829092144966125,
+      "learning_rate": 7.86219215733168e-05,
+      "loss": 0.3705,
+      "step": 4680
+    },
+    {
+      "epoch": 0.9836580766813325,
+      "grad_norm": 0.791359007358551,
+      "learning_rate": 7.848265735794558e-05,
+      "loss": 0.4434,
+      "step": 4695
+    },
+    {
+      "epoch": 0.9868007542426147,
+      "grad_norm": 0.7627493739128113,
+      "learning_rate": 7.834306525844461e-05,
+      "loss": 0.4496,
+      "step": 4710
+    },
+    {
+      "epoch": 0.9899434318038969,
+      "grad_norm": 0.679959237575531,
+      "learning_rate": 7.820314688175784e-05,
+      "loss": 0.4815,
+      "step": 4725
+    },
+    {
+      "epoch": 0.9930861093651792,
+      "grad_norm": 0.8766529560089111,
+      "learning_rate": 7.806290383858523e-05,
+      "loss": 0.4704,
+      "step": 4740
+    },
+    {
+      "epoch": 0.9962287869264613,
+      "grad_norm": 1.1642574071884155,
+      "learning_rate": 7.792233774336423e-05,
+      "loss": 0.4974,
+      "step": 4755
+    },
+    {
+      "epoch": 0.9993714644877436,
+      "grad_norm": 0.7194317579269409,
+      "learning_rate": 7.778145021425114e-05,
+      "loss": 0.4423,
+      "step": 4770
+    },
+    {
+      "epoch": 1.0025141420490258,
+      "grad_norm": 0.7814803719520569,
+      "learning_rate": 7.764024287310252e-05,
+      "loss": 0.4194,
+      "step": 4785
+    },
+    {
+      "epoch": 1.005656819610308,
+      "grad_norm": 0.8891781568527222,
+      "learning_rate": 7.749871734545652e-05,
+      "loss": 0.3977,
+      "step": 4800
+    },
+    {
+      "epoch": 1.0087994971715901,
+      "grad_norm": 0.7444355487823486,
+      "learning_rate": 7.735687526051418e-05,
+      "loss": 0.3924,
+      "step": 4815
+    },
+    {
+      "epoch": 1.0119421747328725,
+      "grad_norm": 0.9248786568641663,
+      "learning_rate": 7.721471825112062e-05,
+      "loss": 0.4273,
+      "step": 4830
+    },
+    {
+      "epoch": 1.0150848522941547,
+      "grad_norm": 0.6513450741767883,
+      "learning_rate": 7.70722479537463e-05,
+      "loss": 0.3909,
+      "step": 4845
+    },
+    {
+      "epoch": 1.0182275298554369,
+      "grad_norm": 0.8597205877304077,
+      "learning_rate": 7.692946600846818e-05,
+      "loss": 0.4027,
+      "step": 4860
+    },
+    {
+      "epoch": 1.021370207416719,
+      "grad_norm": 0.9086320996284485,
+      "learning_rate": 7.678637405895076e-05,
+      "loss": 0.4225,
+      "step": 4875
+    },
+    {
+      "epoch": 1.0245128849780012,
+      "grad_norm": 0.8219915628433228,
+      "learning_rate": 7.66429737524273e-05,
+      "loss": 0.4055,
+      "step": 4890
+    },
+    {
+      "epoch": 1.0276555625392834,
+      "grad_norm": 0.9232605695724487,
+      "learning_rate": 7.649926673968069e-05,
+      "loss": 0.3801,
+      "step": 4905
+    },
+    {
+      "epoch": 1.0307982401005658,
+      "grad_norm": 0.8866775035858154,
+      "learning_rate": 7.635525467502462e-05,
+      "loss": 0.3887,
+      "step": 4920
+    },
+    {
+      "epoch": 1.033940917661848,
+      "grad_norm": 0.6395006775856018,
+      "learning_rate": 7.62109392162844e-05,
+      "loss": 0.4018,
+      "step": 4935
+    },
+    {
+      "epoch": 1.03708359522313,
+      "grad_norm": 0.8276055455207825,
+      "learning_rate": 7.60663220247779e-05,
+      "loss": 0.3875,
+      "step": 4950
+    },
+    {
+      "epoch": 1.0402262727844123,
+      "grad_norm": 0.8251763582229614,
+      "learning_rate": 7.592140476529652e-05,
+      "loss": 0.3912,
+      "step": 4965
+    },
+    {
+      "epoch": 1.0433689503456944,
+      "grad_norm": 0.8321304321289062,
+      "learning_rate": 7.577618910608591e-05,
+      "loss": 0.4317,
+      "step": 4980
+    },
+    {
+      "epoch": 1.0465116279069768,
+      "grad_norm": 0.6474670171737671,
+      "learning_rate": 7.56306767188268e-05,
+      "loss": 0.4594,
+      "step": 4995
+    },
+    {
+      "epoch": 1.049654305468259,
+      "grad_norm": 0.6989348530769348,
+      "learning_rate": 7.548486927861582e-05,
+      "loss": 0.3744,
+      "step": 5010
+    },
+    {
+      "epoch": 1.0527969830295412,
+      "grad_norm": 0.8184515237808228,
+      "learning_rate": 7.533876846394613e-05,
+      "loss": 0.3364,
+      "step": 5025
+    },
+    {
+      "epoch": 1.0559396605908233,
+      "grad_norm": 0.7965102195739746,
+      "learning_rate": 7.519237595668811e-05,
+      "loss": 0.3934,
+      "step": 5040
+    },
+    {
+      "epoch": 1.0590823381521055,
+      "grad_norm": 0.731299638748169,
+      "learning_rate": 7.504569344207007e-05,
+      "loss": 0.4161,
+      "step": 5055
+    },
+    {
+      "epoch": 1.062225015713388,
+      "grad_norm": 0.9074578881263733,
+      "learning_rate": 7.489872260865877e-05,
+      "loss": 0.4103,
+      "step": 5070
+    },
+    {
+      "epoch": 1.06536769327467,
+      "grad_norm": 0.8735909461975098,
+      "learning_rate": 7.475146514834001e-05,
+      "loss": 0.3686,
+      "step": 5085
+    },
+    {
+      "epoch": 1.0685103708359522,
+      "grad_norm": 0.7814076542854309,
+      "learning_rate": 7.460392275629918e-05,
+      "loss": 0.3943,
+      "step": 5100
+    },
+    {
+      "epoch": 1.0716530483972344,
+      "grad_norm": 0.8307476043701172,
+      "learning_rate": 7.445609713100171e-05,
+      "loss": 0.3999,
+      "step": 5115
+    },
+    {
+      "epoch": 1.0747957259585166,
+      "grad_norm": 0.7908287048339844,
+      "learning_rate": 7.430798997417353e-05,
+      "loss": 0.4104,
+      "step": 5130
+    },
+    {
+      "epoch": 1.077938403519799,
+      "grad_norm": 0.8598707914352417,
+      "learning_rate": 7.415960299078143e-05,
+      "loss": 0.3976,
+      "step": 5145
+    },
+    {
+      "epoch": 1.0810810810810811,
+      "grad_norm": 0.5163241028785706,
+      "learning_rate": 7.40109378890136e-05,
+      "loss": 0.3506,
+      "step": 5160
+    },
+    {
+      "epoch": 1.0842237586423633,
+      "grad_norm": 0.8642787933349609,
+      "learning_rate": 7.386199638025973e-05,
+      "loss": 0.31,
+      "step": 5175
+    },
+    {
+      "epoch": 1.0873664362036455,
+      "grad_norm": 0.7603743076324463,
+      "learning_rate": 7.371278017909148e-05,
+      "loss": 0.4695,
+      "step": 5190
+    },
+    {
+      "epoch": 1.0905091137649277,
+      "grad_norm": 0.7949853539466858,
+      "learning_rate": 7.356329100324273e-05,
+      "loss": 0.4076,
+      "step": 5205
+    },
+    {
+      "epoch": 1.0936517913262098,
+      "grad_norm": 0.8560110926628113,
+      "learning_rate": 7.341353057358966e-05,
+      "loss": 0.3833,
+      "step": 5220
+    },
+    {
+      "epoch": 1.0967944688874922,
+      "grad_norm": 0.632763147354126,
+      "learning_rate": 7.326350061413114e-05,
+      "loss": 0.4128,
+      "step": 5235
+    },
+    {
+      "epoch": 1.0999371464487744,
+      "grad_norm": 0.9416031837463379,
+      "learning_rate": 7.311320285196875e-05,
+      "loss": 0.3665,
+      "step": 5250
+    },
+    {
+      "epoch": 1.1030798240100566,
+      "grad_norm": 0.6195524334907532,
+      "learning_rate": 7.296263901728694e-05,
+      "loss": 0.362,
+      "step": 5265
+    },
+    {
+      "epoch": 1.1062225015713387,
+      "grad_norm": 0.8545498251914978,
+      "learning_rate": 7.281181084333311e-05,
+      "loss": 0.361,
+      "step": 5280
+    },
+    {
+      "epoch": 1.109365179132621,
+      "grad_norm": 0.75226229429245,
+      "learning_rate": 7.26607200663977e-05,
+      "loss": 0.3948,
+      "step": 5295
+    },
+    {
+      "epoch": 1.1125078566939033,
+      "grad_norm": 0.877756655216217,
+      "learning_rate": 7.250936842579407e-05,
+      "loss": 0.4061,
+      "step": 5310
+    },
+    {
+      "epoch": 1.1156505342551855,
+      "grad_norm": 0.5953283309936523,
+      "learning_rate": 7.235775766383862e-05,
+      "loss": 0.3273,
+      "step": 5325
+    },
+    {
+      "epoch": 1.1187932118164676,
+      "grad_norm": 0.8206706643104553,
+      "learning_rate": 7.220588952583071e-05,
+      "loss": 0.3757,
+      "step": 5340
+    },
+    {
+      "epoch": 1.1219358893777498,
+      "grad_norm": 0.7466344237327576,
+      "learning_rate": 7.205376576003247e-05,
+      "loss": 0.3892,
+      "step": 5355
+    },
+    {
+      "epoch": 1.125078566939032,
+      "grad_norm": 0.8034494519233704,
+      "learning_rate": 7.190138811764882e-05,
+      "loss": 0.4043,
+      "step": 5370
+    },
+    {
+      "epoch": 1.1282212445003144,
+      "grad_norm": 0.9050668478012085,
+      "learning_rate": 7.174875835280716e-05,
+      "loss": 0.3812,
+      "step": 5385
+    },
+    {
+      "epoch": 1.1313639220615965,
+      "grad_norm": 0.8540876507759094,
+      "learning_rate": 7.159587822253733e-05,
+      "loss": 0.3645,
+      "step": 5400
+    },
+    {
+      "epoch": 1.1345065996228787,
+      "grad_norm": 0.7688354849815369,
+      "learning_rate": 7.14427494867512e-05,
+      "loss": 0.3683,
+      "step": 5415
+    },
+    {
+      "epoch": 1.1376492771841609,
+      "grad_norm": 0.6950829029083252,
+      "learning_rate": 7.128937390822261e-05,
+      "loss": 0.3347,
+      "step": 5430
+    },
+    {
+      "epoch": 1.140791954745443,
+      "grad_norm": 0.8212427496910095,
+      "learning_rate": 7.113575325256694e-05,
+      "loss": 0.3775,
+      "step": 5445
+    },
+    {
+      "epoch": 1.1439346323067254,
+      "grad_norm": 0.8312988877296448,
+      "learning_rate": 7.098188928822084e-05,
+      "loss": 0.4325,
+      "step": 5460
+    },
+    {
+      "epoch": 1.1470773098680076,
+      "grad_norm": 0.9646623134613037,
+      "learning_rate": 7.082778378642184e-05,
+      "loss": 0.3898,
+      "step": 5475
+    },
+    {
+      "epoch": 1.1502199874292898,
+      "grad_norm": 0.8333424925804138,
+      "learning_rate": 7.0673438521188e-05,
+      "loss": 0.4068,
+      "step": 5490
+    },
+    {
+      "epoch": 1.153362664990572,
+      "grad_norm": 0.918892502784729,
+      "learning_rate": 7.051885526929747e-05,
+      "loss": 0.3968,
+      "step": 5505
+    },
+    {
+      "epoch": 1.156505342551854,
+      "grad_norm": 0.5460782647132874,
+      "learning_rate": 7.0364035810268e-05,
+      "loss": 0.3672,
+      "step": 5520
+    },
+    {
+      "epoch": 1.1596480201131363,
+      "grad_norm": 0.876811683177948,
+      "learning_rate": 7.020898192633655e-05,
+      "loss": 0.408,
+      "step": 5535
+    },
+    {
+      "epoch": 1.1627906976744187,
+      "grad_norm": 0.6740222573280334,
+      "learning_rate": 7.005369540243864e-05,
+      "loss": 0.2995,
+      "step": 5550
+    },
+    {
+      "epoch": 1.1659333752357008,
+      "grad_norm": 0.8702965378761292,
+      "learning_rate": 6.989817802618792e-05,
+      "loss": 0.3307,
+      "step": 5565
+    },
+    {
+      "epoch": 1.169076052796983,
+      "grad_norm": 0.8837511539459229,
+      "learning_rate": 6.974243158785554e-05,
+      "loss": 0.3864,
+      "step": 5580
+    },
+    {
+      "epoch": 1.1722187303582652,
+      "grad_norm": 0.4050454795360565,
+      "learning_rate": 6.958645788034952e-05,
+      "loss": 0.3525,
+      "step": 5595
+    },
+    {
+      "epoch": 1.1753614079195476,
+      "grad_norm": 0.8361005187034607,
+      "learning_rate": 6.943025869919418e-05,
+      "loss": 0.3747,
+      "step": 5610
+    },
+    {
+      "epoch": 1.1785040854808297,
+      "grad_norm": 0.841556191444397,
+      "learning_rate": 6.92738358425094e-05,
+      "loss": 0.406,
+      "step": 5625
+    },
+    {
+      "epoch": 1.181646763042112,
+      "grad_norm": 0.629443883895874,
+      "learning_rate": 6.911719111098996e-05,
+      "loss": 0.4175,
+      "step": 5640
+    },
+    {
+      "epoch": 1.184789440603394,
+      "grad_norm": 0.7146449685096741,
+      "learning_rate": 6.896032630788476e-05,
+      "loss": 0.3511,
+      "step": 5655
+    },
+    {
+      "epoch": 1.1879321181646763,
+      "grad_norm": 0.8358393311500549,
+      "learning_rate": 6.880324323897617e-05,
+      "loss": 0.3851,
+      "step": 5670
+    },
+    {
+      "epoch": 1.1910747957259584,
+      "grad_norm": 0.742857813835144,
+      "learning_rate": 6.864594371255913e-05,
+      "loss": 0.3821,
+      "step": 5685
+    },
+    {
+      "epoch": 1.1942174732872408,
+      "grad_norm": 0.7099196910858154,
+      "learning_rate": 6.848842953942036e-05,
+      "loss": 0.3789,
+      "step": 5700
+    },
+    {
+      "epoch": 1.197360150848523,
+      "grad_norm": 0.754542350769043,
+      "learning_rate": 6.83307025328176e-05,
+      "loss": 0.3472,
+      "step": 5715
+    },
+    {
+      "epoch": 1.2005028284098052,
+      "grad_norm": 0.7466986775398254,
+      "learning_rate": 6.817276450845856e-05,
+      "loss": 0.3393,
+      "step": 5730
+    },
+    {
+      "epoch": 1.2036455059710873,
+      "grad_norm": 0.7026840448379517,
+      "learning_rate": 6.801461728448022e-05,
+      "loss": 0.3891,
+      "step": 5745
+    },
+    {
+      "epoch": 1.2067881835323695,
+      "grad_norm": 1.1348669528961182,
+      "learning_rate": 6.785626268142777e-05,
+      "loss": 0.3802,
+      "step": 5760
+    },
+    {
+      "epoch": 1.2099308610936519,
+      "grad_norm": 0.7511578798294067,
+      "learning_rate": 6.769770252223369e-05,
+      "loss": 0.4252,
+      "step": 5775
+    },
+    {
+      "epoch": 1.213073538654934,
+      "grad_norm": 0.8412914276123047,
+      "learning_rate": 6.753893863219675e-05,
+      "loss": 0.3813,
+      "step": 5790
+    },
+    {
+      "epoch": 1.2162162162162162,
+      "grad_norm": 0.8765383958816528,
+      "learning_rate": 6.737997283896103e-05,
+      "loss": 0.3712,
+      "step": 5805
+    },
+    {
+      "epoch": 1.2193588937774984,
+      "grad_norm": 0.7843053340911865,
+      "learning_rate": 6.722080697249487e-05,
+      "loss": 0.3776,
+      "step": 5820
+    },
+    {
+      "epoch": 1.2225015713387806,
+      "grad_norm": 1.0745536088943481,
+      "learning_rate": 6.706144286506978e-05,
+      "loss": 0.3499,
+      "step": 5835
+    },
+    {
+      "epoch": 1.2256442489000627,
+      "grad_norm": 0.7722020745277405,
+      "learning_rate": 6.690188235123934e-05,
+      "loss": 0.4211,
+      "step": 5850
+    },
+    {
+      "epoch": 1.2287869264613451,
+      "grad_norm": 0.9631087183952332,
+      "learning_rate": 6.674212726781814e-05,
+      "loss": 0.3772,
+      "step": 5865
+    },
+    {
+      "epoch": 1.2319296040226273,
+      "grad_norm": 0.8981698751449585,
+      "learning_rate": 6.65821794538606e-05,
+      "loss": 0.4598,
+      "step": 5880
+    },
+    {
+      "epoch": 1.2350722815839095,
+      "grad_norm": 0.778362512588501,
+      "learning_rate": 6.642204075063974e-05,
+      "loss": 0.4179,
+      "step": 5895
+    },
+    {
+      "epoch": 1.2382149591451916,
+      "grad_norm": 0.8421118259429932,
+      "learning_rate": 6.626171300162615e-05,
+      "loss": 0.3583,
+      "step": 5910
+    },
+    {
+      "epoch": 1.241357636706474,
+      "grad_norm": 1.0227240324020386,
+      "learning_rate": 6.610119805246653e-05,
+      "loss": 0.3919,
+      "step": 5925
+    },
+    {
+      "epoch": 1.2445003142677562,
+      "grad_norm": 0.5748106837272644,
+      "learning_rate": 6.594049775096268e-05,
+      "loss": 0.3571,
+      "step": 5940
+    },
+    {
+      "epoch": 1.2476429918290384,
+      "grad_norm": 0.6924661993980408,
+      "learning_rate": 6.577961394705008e-05,
+      "loss": 0.3812,
+      "step": 5955
+    },
+    {
+      "epoch": 1.2507856693903205,
+      "grad_norm": 0.7702043056488037,
+      "learning_rate": 6.561854849277664e-05,
+      "loss": 0.331,
+      "step": 5970
+    },
+    {
+      "epoch": 1.2539283469516027,
+      "grad_norm": 0.6666329503059387,
+      "learning_rate": 6.545730324228136e-05,
+      "loss": 0.3266,
+      "step": 5985
+    },
+    {
+      "epoch": 1.2570710245128849,
+      "grad_norm": 0.9120034575462341,
+      "learning_rate": 6.529588005177305e-05,
+      "loss": 0.4188,
+      "step": 6000
+    },
+    {
+      "epoch": 1.260213702074167,
+      "grad_norm": 0.7251651287078857,
+      "learning_rate": 6.513428077950886e-05,
+      "loss": 0.4067,
+      "step": 6015
+    },
+    {
+      "epoch": 1.2633563796354494,
+      "grad_norm": 0.6845729947090149,
+      "learning_rate": 6.497250728577296e-05,
+      "loss": 0.4266,
+      "step": 6030
+    },
+    {
+      "epoch": 1.2664990571967316,
+      "grad_norm": 0.7530787587165833,
+      "learning_rate": 6.481056143285512e-05,
+      "loss": 0.3302,
+      "step": 6045
+    },
+    {
+      "epoch": 1.2696417347580138,
+      "grad_norm": 0.7474608421325684,
+      "learning_rate": 6.464844508502927e-05,
+      "loss": 0.4305,
+      "step": 6060
+    },
+    {
+      "epoch": 1.2727844123192962,
+      "grad_norm": 0.8672669529914856,
+      "learning_rate": 6.448616010853199e-05,
+      "loss": 0.4267,
+      "step": 6075
+    },
+    {
+      "epoch": 1.2759270898805783,
+      "grad_norm": 0.7703887224197388,
+      "learning_rate": 6.432370837154109e-05,
+      "loss": 0.3531,
+      "step": 6090
+    },
+    {
+      "epoch": 1.2790697674418605,
+      "grad_norm": 0.7432886958122253,
+      "learning_rate": 6.416109174415406e-05,
+      "loss": 0.3189,
+      "step": 6105
+    },
+    {
+      "epoch": 1.2822124450031427,
+      "grad_norm": 0.9600912928581238,
+      "learning_rate": 6.399831209836659e-05,
+      "loss": 0.4036,
+      "step": 6120
+    },
+    {
+      "epoch": 1.2853551225644249,
+      "grad_norm": 0.7727882862091064,
+      "learning_rate": 6.383537130805098e-05,
+      "loss": 0.3857,
+      "step": 6135
+    },
+    {
+      "epoch": 1.288497800125707,
+      "grad_norm": 0.7628008723258972,
+      "learning_rate": 6.367227124893455e-05,
+      "loss": 0.4229,
+      "step": 6150
+    },
+    {
+      "epoch": 1.2916404776869892,
+      "grad_norm": 0.9682219624519348,
+      "learning_rate": 6.350901379857814e-05,
+      "loss": 0.3544,
+      "step": 6165
+    },
+    {
+      "epoch": 1.2947831552482716,
+      "grad_norm": 0.7553837895393372,
+      "learning_rate": 6.334560083635434e-05,
+      "loss": 0.3968,
+      "step": 6180
+    },
+    {
+      "epoch": 1.2979258328095538,
+      "grad_norm": 0.7951422333717346,
+      "learning_rate": 6.318203424342605e-05,
+      "loss": 0.2946,
+      "step": 6195
+    },
+    {
+      "epoch": 1.301068510370836,
+      "grad_norm": 0.9351706504821777,
+      "learning_rate": 6.301831590272465e-05,
+      "loss": 0.4203,
+      "step": 6210
+    },
+    {
+      "epoch": 1.304211187932118,
+      "grad_norm": 0.8283166289329529,
+      "learning_rate": 6.28544476989284e-05,
+      "loss": 0.4166,
+      "step": 6225
+    },
+    {
+      "epoch": 1.3073538654934005,
+      "grad_norm": 0.7889246940612793,
+      "learning_rate": 6.269043151844081e-05,
+      "loss": 0.4084,
+      "step": 6240
+    },
+    {
+      "epoch": 1.3104965430546827,
+      "grad_norm": 0.7893148064613342,
+      "learning_rate": 6.252626924936876e-05,
+      "loss": 0.3327,
+      "step": 6255
+    },
+    {
+      "epoch": 1.3136392206159648,
+      "grad_norm": 0.9599968194961548,
+      "learning_rate": 6.236196278150092e-05,
+      "loss": 0.3987,
+      "step": 6270
+    },
+    {
+      "epoch": 1.316781898177247,
+      "grad_norm": 0.7326962351799011,
+      "learning_rate": 6.219751400628593e-05,
+      "loss": 0.3872,
+      "step": 6285
+    },
+    {
+      "epoch": 1.3199245757385292,
+      "grad_norm": 0.7666275501251221,
+      "learning_rate": 6.203292481681061e-05,
+      "loss": 0.2906,
+      "step": 6300
+    },
+    {
+      "epoch": 1.3230672532998113,
+      "grad_norm": 0.7648006081581116,
+      "learning_rate": 6.186819710777819e-05,
+      "loss": 0.4077,
+      "step": 6315
+    },
+    {
+      "epoch": 1.3262099308610937,
+      "grad_norm": 0.8993086218833923,
+      "learning_rate": 6.170333277548653e-05,
+      "loss": 0.3334,
+      "step": 6330
+    },
+    {
+      "epoch": 1.329352608422376,
+      "grad_norm": 0.8966405987739563,
+      "learning_rate": 6.153833371780622e-05,
+      "loss": 0.3772,
+      "step": 6345
+    },
+    {
+      "epoch": 1.332495285983658,
+      "grad_norm": 0.955697774887085,
+      "learning_rate": 6.137320183415877e-05,
+      "loss": 0.3652,
+      "step": 6360
+    },
+    {
+      "epoch": 1.3356379635449402,
+      "grad_norm": 0.913931667804718,
+      "learning_rate": 6.120793902549478e-05,
+      "loss": 0.3943,
+      "step": 6375
+    },
+    {
+      "epoch": 1.3387806411062226,
+      "grad_norm": 0.471160352230072,
+      "learning_rate": 6.1042547194272e-05,
+      "loss": 0.3656,
+      "step": 6390
+    },
+    {
+      "epoch": 1.3419233186675048,
+      "grad_norm": 0.7883521914482117,
+      "learning_rate": 6.0877028244433444e-05,
+      "loss": 0.3494,
+      "step": 6405
+    },
+    {
+      "epoch": 1.345065996228787,
+      "grad_norm": 0.8015203475952148,
+      "learning_rate": 6.071138408138547e-05,
+      "loss": 0.3498,
+      "step": 6420
+    },
+    {
+      "epoch": 1.3482086737900691,
+      "grad_norm": 0.8431302905082703,
+      "learning_rate": 6.0545616611975886e-05,
+      "loss": 0.3726,
+      "step": 6435
+    },
+    {
+      "epoch": 1.3513513513513513,
+      "grad_norm": 0.6410717964172363,
+      "learning_rate": 6.0379727744471936e-05,
+      "loss": 0.3793,
+      "step": 6450
+    },
+    {
+      "epoch": 1.3544940289126335,
+      "grad_norm": 0.8410218358039856,
+      "learning_rate": 6.021371938853839e-05,
+      "loss": 0.4294,
+      "step": 6465
+    },
+    {
+      "epoch": 1.3576367064739157,
+      "grad_norm": 0.622178852558136,
+      "learning_rate": 6.004759345521552e-05,
+      "loss": 0.3373,
+      "step": 6480
+    },
+    {
+      "epoch": 1.360779384035198,
+      "grad_norm": 0.8277848362922668,
+      "learning_rate": 5.988135185689712e-05,
+      "loss": 0.3796,
+      "step": 6495
+    },
+    {
+      "epoch": 1.3639220615964802,
+      "grad_norm": 0.799150824546814,
+      "learning_rate": 5.9714996507308465e-05,
+      "loss": 0.3361,
+      "step": 6510
+    },
+    {
+      "epoch": 1.3670647391577624,
+      "grad_norm": 0.8518102765083313,
+      "learning_rate": 5.954852932148433e-05,
+      "loss": 0.3913,
+      "step": 6525
+    },
+    {
+      "epoch": 1.3702074167190446,
+      "grad_norm": 0.7465687990188599,
+      "learning_rate": 5.9381952215746905e-05,
+      "loss": 0.3546,
+      "step": 6540
+    },
+    {
+      "epoch": 1.373350094280327,
+      "grad_norm": 0.7342978119850159,
+      "learning_rate": 5.921526710768376e-05,
+      "loss": 0.3832,
+      "step": 6555
+    },
+    {
+      "epoch": 1.3764927718416091,
+      "grad_norm": 0.6754856109619141,
+      "learning_rate": 5.9048475916125723e-05,
+      "loss": 0.4051,
+      "step": 6570
+    },
+    {
+      "epoch": 1.3796354494028913,
+      "grad_norm": 0.6392863988876343,
+      "learning_rate": 5.888158056112486e-05,
+      "loss": 0.3828,
+      "step": 6585
+    },
+    {
+      "epoch": 1.3827781269641735,
+      "grad_norm": 0.897132933139801,
+      "learning_rate": 5.871458296393231e-05,
+      "loss": 0.405,
+      "step": 6600
+    },
+    {
+      "epoch": 1.3859208045254556,
+      "grad_norm": 0.7124328017234802,
+      "learning_rate": 5.854748504697624e-05,
+      "loss": 0.3712,
+      "step": 6615
+    },
+    {
+      "epoch": 1.3890634820867378,
+      "grad_norm": 0.8436194062232971,
+      "learning_rate": 5.8380288733839585e-05,
+      "loss": 0.3773,
+      "step": 6630
+    },
+    {
+      "epoch": 1.3922061596480202,
+      "grad_norm": 0.780944287776947,
+      "learning_rate": 5.8212995949238083e-05,
+      "loss": 0.3529,
+      "step": 6645
+    },
+    {
+      "epoch": 1.3953488372093024,
+      "grad_norm": 1.0335406064987183,
+      "learning_rate": 5.804560861899795e-05,
+      "loss": 0.4262,
+      "step": 6660
+    },
+    {
+      "epoch": 1.3984915147705845,
+      "grad_norm": 0.7593971490859985,
+      "learning_rate": 5.7878128670033826e-05,
+      "loss": 0.4079,
+      "step": 6675
+    },
+    {
+      "epoch": 1.4016341923318667,
+      "grad_norm": 0.7240027189254761,
+      "learning_rate": 5.7710558030326545e-05,
+      "loss": 0.3835,
+      "step": 6690
+    },
+    {
+      "epoch": 1.404776869893149,
+      "grad_norm": 1.530868411064148,
+      "learning_rate": 5.754289862890093e-05,
+      "loss": 0.4294,
+      "step": 6705
+    },
+    {
+      "epoch": 1.4079195474544313,
+      "grad_norm": 0.6043078899383545,
+      "learning_rate": 5.7375152395803624e-05,
+      "loss": 0.3343,
+      "step": 6720
+    },
+    {
+      "epoch": 1.4110622250157134,
+      "grad_norm": 0.8058659434318542,
+      "learning_rate": 5.720732126208082e-05,
+      "loss": 0.4533,
+      "step": 6735
+    },
+    {
+      "epoch": 1.4142049025769956,
+      "grad_norm": 0.7185141444206238,
+      "learning_rate": 5.7039407159756106e-05,
+      "loss": 0.42,
+      "step": 6750
+    },
+    {
+      "epoch": 1.4173475801382778,
+      "grad_norm": 1.0086369514465332,
+      "learning_rate": 5.687141202180817e-05,
+      "loss": 0.3701,
+      "step": 6765
+    },
+    {
+      "epoch": 1.42049025769956,
+      "grad_norm": 1.0289742946624756,
+      "learning_rate": 5.67033377821485e-05,
+      "loss": 0.4565,
+      "step": 6780
+    },
+    {
+      "epoch": 1.4236329352608421,
+      "grad_norm": 1.1389039754867554,
+      "learning_rate": 5.6535186375599266e-05,
+      "loss": 0.3555,
+      "step": 6795
+    },
+    {
+      "epoch": 1.4267756128221245,
+      "grad_norm": 0.887610673904419,
+      "learning_rate": 5.636695973787093e-05,
+      "loss": 0.368,
+      "step": 6810
+    },
+    {
+      "epoch": 1.4299182903834067,
+      "grad_norm": 0.9625629186630249,
+      "learning_rate": 5.619865980553994e-05,
+      "loss": 0.3962,
+      "step": 6825
+    },
+    {
+      "epoch": 1.4330609679446888,
+      "grad_norm": 0.8793766498565674,
+      "learning_rate": 5.6030288516026564e-05,
+      "loss": 0.3979,
+      "step": 6840
+    },
+    {
+      "epoch": 1.436203645505971,
+      "grad_norm": 0.7626388669013977,
+      "learning_rate": 5.586184780757251e-05,
+      "loss": 0.345,
+      "step": 6855
+    },
+    {
+      "epoch": 1.4393463230672534,
+      "grad_norm": 1.109713077545166,
+      "learning_rate": 5.5693339619218534e-05,
+      "loss": 0.4446,
+      "step": 6870
+    },
+    {
+      "epoch": 1.4424890006285356,
+      "grad_norm": 0.9758956432342529,
+      "learning_rate": 5.552476589078231e-05,
+      "loss": 0.401,
+      "step": 6885
+    },
+    {
+      "epoch": 1.4456316781898177,
+      "grad_norm": 0.923329770565033,
+      "learning_rate": 5.5356128562835904e-05,
+      "loss": 0.385,
+      "step": 6900
+    },
+    {
+      "epoch": 1.4487743557511,
+      "grad_norm": 0.7539265155792236,
+      "learning_rate": 5.518742957668359e-05,
+      "loss": 0.3274,
+      "step": 6915
+    },
+    {
+      "epoch": 1.451917033312382,
+      "grad_norm": 0.8187793493270874,
+      "learning_rate": 5.5018670874339386e-05,
+      "loss": 0.3677,
+      "step": 6930
+    },
+    {
+      "epoch": 1.4550597108736643,
+      "grad_norm": 0.9522603750228882,
+      "learning_rate": 5.484985439850473e-05,
+      "loss": 0.3319,
+      "step": 6945
+    },
+    {
+      "epoch": 1.4582023884349467,
+      "grad_norm": 0.8808611631393433,
+      "learning_rate": 5.468098209254622e-05,
+      "loss": 0.4311,
+      "step": 6960
+    },
+    {
+      "epoch": 1.4613450659962288,
+      "grad_norm": 0.6949836611747742,
+      "learning_rate": 5.4512055900473035e-05,
+      "loss": 0.3679,
+      "step": 6975
+    },
+    {
+      "epoch": 1.464487743557511,
+      "grad_norm": 0.783545196056366,
+      "learning_rate": 5.434307776691479e-05,
+      "loss": 0.3552,
+      "step": 6990
+    },
+    {
+      "epoch": 1.4676304211187932,
+      "grad_norm": 0.8342312574386597,
+      "learning_rate": 5.417404963709894e-05,
+      "loss": 0.3755,
+      "step": 7005
+    },
+    {
+      "epoch": 1.4707730986800756,
+      "grad_norm": 0.7615540027618408,
+      "learning_rate": 5.400497345682857e-05,
+      "loss": 0.3605,
+      "step": 7020
+    },
+    {
+      "epoch": 1.4739157762413577,
+      "grad_norm": 0.8944594860076904,
+      "learning_rate": 5.3835851172459794e-05,
+      "loss": 0.3948,
+      "step": 7035
+    },
+    {
+      "epoch": 1.47705845380264,
+      "grad_norm": 0.8412215113639832,
+      "learning_rate": 5.36666847308796e-05,
+      "loss": 0.3658,
+      "step": 7050
+    },
+    {
+      "epoch": 1.480201131363922,
+      "grad_norm": 0.8457724452018738,
+      "learning_rate": 5.34974760794832e-05,
+      "loss": 0.4327,
+      "step": 7065
+    },
+    {
+      "epoch": 1.4833438089252042,
+      "grad_norm": 0.7231891751289368,
+      "learning_rate": 5.332822716615172e-05,
+      "loss": 0.3489,
+      "step": 7080
+    },
+    {
+      "epoch": 1.4864864864864864,
+      "grad_norm": 0.8975026607513428,
+      "learning_rate": 5.315893993922986e-05,
+      "loss": 0.331,
+      "step": 7095
+    },
+    {
+      "epoch": 1.4896291640477686,
+      "grad_norm": 0.871842086315155,
+      "learning_rate": 5.2989616347503244e-05,
+      "loss": 0.4056,
+      "step": 7110
+    },
+    {
+      "epoch": 1.492771841609051,
+      "grad_norm": 0.5846161246299744,
+      "learning_rate": 5.282025834017623e-05,
+      "loss": 0.381,
+      "step": 7125
+    },
+    {
+      "epoch": 1.4959145191703331,
+      "grad_norm": 0.6650387644767761,
+      "learning_rate": 5.265086786684929e-05,
+      "loss": 0.34,
+      "step": 7140
+    },
+    {
+      "epoch": 1.4990571967316153,
+      "grad_norm": 0.862241804599762,
+      "learning_rate": 5.2481446877496665e-05,
+      "loss": 0.354,
+      "step": 7155
+    },
+    {
+      "epoch": 1.5021998742928977,
+      "grad_norm": 0.8328828811645508,
+      "learning_rate": 5.231199732244386e-05,
+      "loss": 0.3772,
+      "step": 7170
+    },
+    {
+      "epoch": 1.5053425518541799,
+      "grad_norm": 0.5438669323921204,
+      "learning_rate": 5.214252115234527e-05,
+      "loss": 0.3493,
+      "step": 7185
+    },
+    {
+      "epoch": 1.508485229415462,
+      "grad_norm": 0.7722681760787964,
+      "learning_rate": 5.197302031816165e-05,
+      "loss": 0.3494,
+      "step": 7200
+    },
+    {
+      "epoch": 1.5116279069767442,
+      "grad_norm": 0.9693325161933899,
+      "learning_rate": 5.180349677113762e-05,
+      "loss": 0.3512,
+      "step": 7215
+    },
+    {
+      "epoch": 1.5147705845380264,
+      "grad_norm": 1.0208348035812378,
+      "learning_rate": 5.163395246277938e-05,
+      "loss": 0.2772,
+      "step": 7230
+    },
+    {
+      "epoch": 1.5179132620993085,
+      "grad_norm": 0.8255509734153748,
+      "learning_rate": 5.1464389344832024e-05,
+      "loss": 0.3491,
+      "step": 7245
+    },
+    {
+      "epoch": 1.5210559396605907,
+      "grad_norm": 0.723574697971344,
+      "learning_rate": 5.1294809369257244e-05,
+      "loss": 0.3894,
+      "step": 7260
+    },
+    {
+      "epoch": 1.5241986172218729,
+      "grad_norm": 0.8955418467521667,
+      "learning_rate": 5.112521448821076e-05,
+      "loss": 0.3722,
+      "step": 7275
+    },
+    {
+      "epoch": 1.5273412947831553,
+      "grad_norm": 0.9446234703063965,
+      "learning_rate": 5.0955606654019895e-05,
+      "loss": 0.3602,
+      "step": 7290
+    },
+    {
+      "epoch": 1.5304839723444374,
+      "grad_norm": 0.7256786227226257,
+      "learning_rate": 5.078598781916107e-05,
+      "loss": 0.3488,
+      "step": 7305
+    },
+    {
+      "epoch": 1.5336266499057196,
+      "grad_norm": 0.775834858417511,
+      "learning_rate": 5.0616359936237355e-05,
+      "loss": 0.3983,
+      "step": 7320
+    },
+    {
+      "epoch": 1.536769327467002,
+      "grad_norm": 0.7684575915336609,
+      "learning_rate": 5.044672495795598e-05,
+      "loss": 0.3992,
+      "step": 7335
+    },
+    {
+      "epoch": 1.5399120050282842,
+      "grad_norm": 0.7569010853767395,
+      "learning_rate": 5.0277084837105826e-05,
+      "loss": 0.352,
+      "step": 7350
+    },
+    {
+      "epoch": 1.5430546825895664,
+      "grad_norm": 0.7330282926559448,
+      "learning_rate": 5.010744152653501e-05,
+      "loss": 0.3486,
+      "step": 7365
+    },
+    {
+      "epoch": 1.5461973601508485,
+      "grad_norm": 0.8921106457710266,
+      "learning_rate": 4.993779697912837e-05,
+      "loss": 0.3107,
+      "step": 7380
+    },
+    {
+      "epoch": 1.5493400377121307,
+      "grad_norm": 0.7190592288970947,
+      "learning_rate": 4.976815314778493e-05,
+      "loss": 0.3429,
+      "step": 7395
+    },
+    {
+      "epoch": 1.5524827152734129,
+      "grad_norm": 0.8145999312400818,
+      "learning_rate": 4.9598511985395535e-05,
+      "loss": 0.3455,
+      "step": 7410
+    },
+    {
+      "epoch": 1.555625392834695,
+      "grad_norm": 0.7628950476646423,
+      "learning_rate": 4.942887544482029e-05,
+      "loss": 0.3362,
+      "step": 7425
+    },
+    {
+      "epoch": 1.5587680703959774,
+      "grad_norm": 0.5859194993972778,
+      "learning_rate": 4.925924547886603e-05,
+      "loss": 0.3723,
+      "step": 7440
+    },
+    {
+      "epoch": 1.5619107479572596,
+      "grad_norm": 0.7906526327133179,
+      "learning_rate": 4.9089624040264013e-05,
+      "loss": 0.3511,
+      "step": 7455
+    },
+    {
+      "epoch": 1.5650534255185418,
+      "grad_norm": 0.7591722011566162,
+      "learning_rate": 4.892001308164727e-05,
+      "loss": 0.4439,
+      "step": 7470
+    },
+    {
+      "epoch": 1.5681961030798242,
+      "grad_norm": 0.9237760901451111,
+      "learning_rate": 4.875041455552817e-05,
+      "loss": 0.3638,
+      "step": 7485
+    },
+    {
+      "epoch": 1.5713387806411063,
+      "grad_norm": 0.734752893447876,
+      "learning_rate": 4.858083041427599e-05,
+      "loss": 0.4047,
+      "step": 7500
+    },
+    {
+      "epoch": 1.5744814582023885,
+      "grad_norm": 0.676703155040741,
+      "learning_rate": 4.8411262610094445e-05,
+      "loss": 0.3566,
+      "step": 7515
+    },
+    {
+      "epoch": 1.5776241357636707,
+      "grad_norm": 0.8751126527786255,
+      "learning_rate": 4.824171309499913e-05,
+      "loss": 0.3743,
+      "step": 7530
+    },
+    {
+      "epoch": 1.5807668133249528,
+      "grad_norm": 0.6884835958480835,
+      "learning_rate": 4.807218382079511e-05,
+      "loss": 0.3821,
+      "step": 7545
+    },
+    {
+      "epoch": 1.583909490886235,
+      "grad_norm": 0.8230961561203003,
+      "learning_rate": 4.790267673905447e-05,
+      "loss": 0.3193,
+      "step": 7560
+    },
+    {
+      "epoch": 1.5870521684475172,
+      "grad_norm": 0.8046270608901978,
+      "learning_rate": 4.7733193801093803e-05,
+      "loss": 0.3714,
+      "step": 7575
+    },
+    {
+      "epoch": 1.5901948460087993,
+      "grad_norm": 0.895897626876831,
+      "learning_rate": 4.756373695795177e-05,
+      "loss": 0.386,
+      "step": 7590
+    },
+    {
+      "epoch": 1.5933375235700817,
+      "grad_norm": 0.8858537077903748,
+      "learning_rate": 4.7394308160366617e-05,
+      "loss": 0.3755,
+      "step": 7605
+    },
+    {
+      "epoch": 1.596480201131364,
+      "grad_norm": 0.6874979138374329,
+      "learning_rate": 4.722490935875377e-05,
+      "loss": 0.3547,
+      "step": 7620
+    },
+    {
+      "epoch": 1.5996228786926463,
+      "grad_norm": 0.8027022480964661,
+      "learning_rate": 4.705554250318335e-05,
+      "loss": 0.3702,
+      "step": 7635
+    },
+    {
+      "epoch": 1.6027655562539285,
+      "grad_norm": 0.9383290410041809,
+      "learning_rate": 4.688620954335766e-05,
+      "loss": 0.4038,
+      "step": 7650
+    },
+    {
+      "epoch": 1.6059082338152106,
+      "grad_norm": 0.8475779294967651,
+      "learning_rate": 4.671691242858891e-05,
+      "loss": 0.3257,
+      "step": 7665
+    },
+    {
+      "epoch": 1.6090509113764928,
+      "grad_norm": 0.702893853187561,
+      "learning_rate": 4.654765310777659e-05,
+      "loss": 0.3642,
+      "step": 7680
+    },
+    {
+      "epoch": 1.612193588937775,
+      "grad_norm": 0.7762289047241211,
+      "learning_rate": 4.6378433529385157e-05,
+      "loss": 0.3859,
+      "step": 7695
+    },
+    {
+      "epoch": 1.6153362664990571,
+      "grad_norm": 0.7309826016426086,
+      "learning_rate": 4.620925564142151e-05,
+      "loss": 0.3427,
+      "step": 7710
+    },
+    {
+      "epoch": 1.6184789440603393,
+      "grad_norm": 0.655974805355072,
+      "learning_rate": 4.60401213914127e-05,
+      "loss": 0.3893,
+      "step": 7725
+    },
+    {
+      "epoch": 1.6216216216216215,
+      "grad_norm": 0.7434260845184326,
+      "learning_rate": 4.5871032726383386e-05,
+      "loss": 0.3528,
+      "step": 7740
+    },
+    {
+      "epoch": 1.6247642991829039,
+      "grad_norm": 0.981696605682373,
+      "learning_rate": 4.570199159283345e-05,
+      "loss": 0.3792,
+      "step": 7755
+    },
+    {
+      "epoch": 1.627906976744186,
+      "grad_norm": 0.5884058475494385,
+      "learning_rate": 4.553299993671567e-05,
+      "loss": 0.3082,
+      "step": 7770
+    },
+    {
+      "epoch": 1.6310496543054682,
+      "grad_norm": 0.9349349737167358,
+      "learning_rate": 4.536405970341317e-05,
+      "loss": 0.3736,
+      "step": 7785
+    },
+    {
+      "epoch": 1.6341923318667506,
+      "grad_norm": 0.8422302603721619,
+      "learning_rate": 4.519517283771717e-05,
+      "loss": 0.3897,
+      "step": 7800
+    },
+    {
+      "epoch": 1.6373350094280328,
+      "grad_norm": 0.7569222450256348,
+      "learning_rate": 4.502634128380448e-05,
+      "loss": 0.3581,
+      "step": 7815
+    },
+    {
+      "epoch": 1.640477686989315,
+      "grad_norm": 0.8034069538116455,
+      "learning_rate": 4.4857566985215276e-05,
+      "loss": 0.3542,
+      "step": 7830
+    },
+    {
+      "epoch": 1.6436203645505971,
+      "grad_norm": 0.5547857284545898,
+      "learning_rate": 4.4688851884830516e-05,
+      "loss": 0.3089,
+      "step": 7845
+    },
+    {
+      "epoch": 1.6467630421118793,
+      "grad_norm": 0.8145669102668762,
+      "learning_rate": 4.452019792484975e-05,
+      "loss": 0.3391,
+      "step": 7860
+    },
+    {
+      "epoch": 1.6499057196731615,
+      "grad_norm": 0.672332227230072,
+      "learning_rate": 4.4351607046768704e-05,
+      "loss": 0.3866,
+      "step": 7875
+    },
+    {
+      "epoch": 1.6530483972344436,
+      "grad_norm": 0.7952318787574768,
+      "learning_rate": 4.418308119135686e-05,
+      "loss": 0.4221,
+      "step": 7890
+    },
+    {
+      "epoch": 1.6561910747957258,
+      "grad_norm": 0.7489158511161804,
+      "learning_rate": 4.401462229863526e-05,
+      "loss": 0.3687,
+      "step": 7905
+    },
+    {
+      "epoch": 1.6593337523570082,
+      "grad_norm": 0.8457122445106506,
+      "learning_rate": 4.3846232307854e-05,
+      "loss": 0.3888,
+      "step": 7920
+    },
+    {
+      "epoch": 1.6624764299182904,
+      "grad_norm": 0.7040199637413025,
+      "learning_rate": 4.36779131574701e-05,
+      "loss": 0.3437,
+      "step": 7935
+    },
+    {
+      "epoch": 1.6656191074795728,
+      "grad_norm": 1.0369516611099243,
+      "learning_rate": 4.3509666785125005e-05,
+      "loss": 0.3557,
+      "step": 7950
+    },
+    {
+      "epoch": 1.668761785040855,
+      "grad_norm": 0.7418217062950134,
+      "learning_rate": 4.334149512762238e-05,
+      "loss": 0.351,
+      "step": 7965
+    },
+    {
+      "epoch": 1.671904462602137,
+      "grad_norm": 0.6527841687202454,
+      "learning_rate": 4.3173400120905824e-05,
+      "loss": 0.3286,
+      "step": 7980
+    },
+    {
+      "epoch": 1.6750471401634193,
+      "grad_norm": 0.9062017798423767,
+      "learning_rate": 4.3005383700036525e-05,
+      "loss": 0.3828,
+      "step": 7995
+    },
+    {
+      "epoch": 1.6781898177247014,
+      "grad_norm": 0.6981047987937927,
+      "learning_rate": 4.283744779917102e-05,
+      "loss": 0.3689,
+      "step": 8010
+    },
+    {
+      "epoch": 1.6813324952859836,
+      "grad_norm": 0.8865767121315002,
+      "learning_rate": 4.26695943515389e-05,
+      "loss": 0.3912,
+      "step": 8025
+    },
+    {
+      "epoch": 1.6844751728472658,
+      "grad_norm": 0.5835604667663574,
+      "learning_rate": 4.250182528942065e-05,
+      "loss": 0.317,
+      "step": 8040
+    },
+    {
+      "epoch": 1.687617850408548,
+      "grad_norm": 0.869529128074646,
+      "learning_rate": 4.233414254412525e-05,
+      "loss": 0.4031,
+      "step": 8055
+    },
+    {
+      "epoch": 1.6907605279698303,
+      "grad_norm": 0.7666299939155579,
+      "learning_rate": 4.216654804596808e-05,
+      "loss": 0.3635,
+      "step": 8070
+    },
+    {
+      "epoch": 1.6939032055311125,
+      "grad_norm": 0.6868289709091187,
+      "learning_rate": 4.199904372424858e-05,
+      "loss": 0.3554,
+      "step": 8085
+    },
+    {
+      "epoch": 1.6970458830923947,
+      "grad_norm": 0.7406291961669922,
+      "learning_rate": 4.183163150722822e-05,
+      "loss": 0.3216,
+      "step": 8100
+    },
+    {
+      "epoch": 1.700188560653677,
+      "grad_norm": 0.7962248921394348,
+      "learning_rate": 4.166431332210807e-05,
+      "loss": 0.3398,
+      "step": 8115
+    },
+    {
+      "epoch": 1.7033312382149592,
+      "grad_norm": 1.02495276927948,
+      "learning_rate": 4.149709109500678e-05,
+      "loss": 0.3817,
+      "step": 8130
+    },
+    {
+      "epoch": 1.7064739157762414,
+      "grad_norm": 0.7741113305091858,
+      "learning_rate": 4.13299667509384e-05,
+      "loss": 0.4072,
+      "step": 8145
+    },
+    {
+      "epoch": 1.7096165933375236,
+      "grad_norm": 0.7952526807785034,
+      "learning_rate": 4.1162942213790086e-05,
+      "loss": 0.3441,
+      "step": 8160
+    },
+    {
+      "epoch": 1.7127592708988058,
+      "grad_norm": 0.7849689722061157,
+      "learning_rate": 4.0996019406300126e-05,
+      "loss": 0.3417,
+      "step": 8175
+    },
+    {
+      "epoch": 1.715901948460088,
+      "grad_norm": 0.7431788444519043,
+      "learning_rate": 4.082920025003567e-05,
+      "loss": 0.3995,
+      "step": 8190
+    },
+    {
+      "epoch": 1.71904462602137,
+      "grad_norm": 0.7709872126579285,
+      "learning_rate": 4.0662486665370734e-05,
+      "loss": 0.4069,
+      "step": 8205
+    },
+    {
+      "epoch": 1.7221873035826523,
+      "grad_norm": 0.6013693809509277,
+      "learning_rate": 4.049588057146394e-05,
+      "loss": 0.3877,
+      "step": 8220
+    },
+    {
+      "epoch": 1.7253299811439347,
+      "grad_norm": 0.7985032796859741,
+      "learning_rate": 4.032938388623657e-05,
+      "loss": 0.3407,
+      "step": 8235
+    },
+    {
+      "epoch": 1.7284726587052168,
+      "grad_norm": 0.6259362101554871,
+      "learning_rate": 4.01629985263504e-05,
+      "loss": 0.3167,
+      "step": 8250
+    },
+    {
+      "epoch": 1.7316153362664992,
+      "grad_norm": 0.7632457613945007,
+      "learning_rate": 3.999672640718567e-05,
+      "loss": 0.365,
+      "step": 8265
+    },
+    {
+      "epoch": 1.7347580138277814,
+      "grad_norm": 0.9532593488693237,
+      "learning_rate": 3.983056944281901e-05,
+      "loss": 0.427,
+      "step": 8280
+    },
+    {
+      "epoch": 1.7379006913890636,
+      "grad_norm": 0.7168596386909485,
+      "learning_rate": 3.966452954600142e-05,
+      "loss": 0.3776,
+      "step": 8295
+    },
+    {
+      "epoch": 1.7410433689503457,
+      "grad_norm": 0.753966748714447,
+      "learning_rate": 3.94986086281363e-05,
+      "loss": 0.3792,
+      "step": 8310
+    },
+    {
+      "epoch": 1.744186046511628,
+      "grad_norm": 0.38063740730285645,
+      "learning_rate": 3.933280859925734e-05,
+      "loss": 0.3499,
+      "step": 8325
+    },
+    {
+      "epoch": 1.74732872407291,
+      "grad_norm": 0.8001086711883545,
+      "learning_rate": 3.916713136800659e-05,
+      "loss": 0.3491,
+      "step": 8340
+    },
+    {
+      "epoch": 1.7504714016341922,
+      "grad_norm": 0.7394033074378967,
+      "learning_rate": 3.900157884161255e-05,
+      "loss": 0.3383,
+      "step": 8355
+    },
+    {
+      "epoch": 1.7536140791954744,
+      "grad_norm": 0.7337818741798401,
+      "learning_rate": 3.8836152925868114e-05,
+      "loss": 0.3705,
+      "step": 8370
+    },
+    {
+      "epoch": 1.7567567567567568,
+      "grad_norm": 0.7671971917152405,
+      "learning_rate": 3.867085552510864e-05,
+      "loss": 0.3125,
+      "step": 8385
+    },
+    {
+      "epoch": 1.759899434318039,
+      "grad_norm": 0.8018542528152466,
+      "learning_rate": 3.850568854219011e-05,
+      "loss": 0.3678,
+      "step": 8400
+    },
+    {
+      "epoch": 1.7630421118793211,
+      "grad_norm": 0.8364083766937256,
+      "learning_rate": 3.834065387846718e-05,
+      "loss": 0.4179,
+      "step": 8415
+    },
+    {
+      "epoch": 1.7661847894406035,
+      "grad_norm": 0.8526837825775146,
+      "learning_rate": 3.817575343377122e-05,
+      "loss": 0.3881,
+      "step": 8430
+    },
+    {
+      "epoch": 1.7693274670018857,
+      "grad_norm": 0.6416676640510559,
+      "learning_rate": 3.8010989106388554e-05,
+      "loss": 0.3099,
+      "step": 8445
+    },
+    {
+      "epoch": 1.7724701445631679,
+      "grad_norm": 0.7990739941596985,
+      "learning_rate": 3.784636279303858e-05,
+      "loss": 0.3598,
+      "step": 8460
+    },
+    {
+      "epoch": 1.77561282212445,
+      "grad_norm": 0.8872657418251038,
+      "learning_rate": 3.76818763888519e-05,
+      "loss": 0.3882,
+      "step": 8475
+    },
+    {
+      "epoch": 1.7787554996857322,
+      "grad_norm": 0.8712546229362488,
+      "learning_rate": 3.7517531787348484e-05,
+      "loss": 0.3773,
+      "step": 8490
+    },
+    {
+      "epoch": 1.7818981772470144,
+      "grad_norm": 0.7423908710479736,
+      "learning_rate": 3.735333088041596e-05,
+      "loss": 0.3777,
+      "step": 8505
+    },
+    {
+      "epoch": 1.7850408548082966,
+      "grad_norm": 0.9166727066040039,
+      "learning_rate": 3.718927555828779e-05,
+      "loss": 0.4059,
+      "step": 8520
+    },
+    {
+      "epoch": 1.7881835323695787,
+      "grad_norm": 0.7207896113395691,
+      "learning_rate": 3.702536770952148e-05,
+      "loss": 0.3754,
+      "step": 8535
+    },
+    {
+      "epoch": 1.7913262099308611,
+      "grad_norm": 0.844727635383606,
+      "learning_rate": 3.6861609220976846e-05,
+      "loss": 0.3328,
+      "step": 8550
+    },
+    {
+      "epoch": 1.7944688874921433,
+      "grad_norm": 0.7674320340156555,
+      "learning_rate": 3.6698001977794366e-05,
+      "loss": 0.3806,
+      "step": 8565
+    },
+    {
+      "epoch": 1.7976115650534257,
+      "grad_norm": 0.6307094693183899,
+      "learning_rate": 3.6534547863373394e-05,
+      "loss": 0.3694,
+      "step": 8580
+    },
+    {
+      "epoch": 1.8007542426147078,
+      "grad_norm": 0.767432451248169,
+      "learning_rate": 3.63712487593505e-05,
+      "loss": 0.4028,
+      "step": 8595
+    },
+    {
+      "epoch": 1.80389692017599,
+      "grad_norm": 0.8937990665435791,
+      "learning_rate": 3.6208106545577824e-05,
+      "loss": 0.3372,
+      "step": 8610
+    },
+    {
+      "epoch": 1.8070395977372722,
+      "grad_norm": 0.590930163860321,
+      "learning_rate": 3.604512310010146e-05,
+      "loss": 0.3684,
+      "step": 8625
+    },
+    {
+      "epoch": 1.8101822752985544,
+      "grad_norm": 0.8184636831283569,
+      "learning_rate": 3.58823002991398e-05,
+      "loss": 0.373,
+      "step": 8640
+    },
+    {
+      "epoch": 1.8133249528598365,
+      "grad_norm": 0.9741955399513245,
+      "learning_rate": 3.5719640017061885e-05,
+      "loss": 0.3374,
+      "step": 8655
+    },
+    {
+      "epoch": 1.8164676304211187,
+      "grad_norm": 1.0014973878860474,
+      "learning_rate": 3.555714412636595e-05,
+      "loss": 0.3848,
+      "step": 8670
+    },
+    {
+      "epoch": 1.8196103079824009,
+      "grad_norm": 0.6335365772247314,
+      "learning_rate": 3.53948144976578e-05,
+      "loss": 0.3689,
+      "step": 8685
+    },
+    {
+      "epoch": 1.8227529855436833,
+      "grad_norm": 0.5687909722328186,
+      "learning_rate": 3.523265299962924e-05,
+      "loss": 0.4178,
+      "step": 8700
+    },
+    {
+      "epoch": 1.8258956631049654,
+      "grad_norm": 0.8622750043869019,
+      "learning_rate": 3.507066149903662e-05,
+      "loss": 0.3899,
+      "step": 8715
+    },
+    {
+      "epoch": 1.8290383406662476,
+      "grad_norm": 0.7984293699264526,
+      "learning_rate": 3.490884186067935e-05,
+      "loss": 0.4353,
+      "step": 8730
+    },
+    {
+      "epoch": 1.83218101822753,
+      "grad_norm": 0.7962972521781921,
+      "learning_rate": 3.474719594737842e-05,
+      "loss": 0.3324,
+      "step": 8745
+    },
+    {
+      "epoch": 1.8353236957888122,
+      "grad_norm": 0.7194257974624634,
+      "learning_rate": 3.4585725619954864e-05,
+      "loss": 0.3765,
+      "step": 8760
+    },
+    {
+      "epoch": 1.8384663733500943,
+      "grad_norm": 0.6931387782096863,
+      "learning_rate": 3.442443273720853e-05,
+      "loss": 0.3183,
+      "step": 8775
+    },
+    {
+      "epoch": 1.8416090509113765,
+      "grad_norm": 0.7540430426597595,
+      "learning_rate": 3.426331915589651e-05,
+      "loss": 0.3975,
+      "step": 8790
+    },
+    {
+      "epoch": 1.8447517284726587,
+      "grad_norm": 0.7310993671417236,
+      "learning_rate": 3.410238673071185e-05,
+      "loss": 0.3975,
+      "step": 8805
+    },
+    {
+      "epoch": 1.8478944060339408,
+      "grad_norm": 0.7351768612861633,
+      "learning_rate": 3.394163731426216e-05,
+      "loss": 0.3558,
+      "step": 8820
+    },
+    {
+      "epoch": 1.851037083595223,
+      "grad_norm": 0.7860934138298035,
+      "learning_rate": 3.378107275704834e-05,
+      "loss": 0.3601,
+      "step": 8835
+    },
+    {
+      "epoch": 1.8541797611565052,
+      "grad_norm": 0.6049594283103943,
+      "learning_rate": 3.362069490744322e-05,
+      "loss": 0.3692,
+      "step": 8850
+    },
+    {
+      "epoch": 1.8573224387177876,
+      "grad_norm": 0.9184178709983826,
+      "learning_rate": 3.346050561167029e-05,
+      "loss": 0.3518,
+      "step": 8865
+    },
+    {
+      "epoch": 1.8604651162790697,
+      "grad_norm": 0.7558075189590454,
+      "learning_rate": 3.3300506713782495e-05,
+      "loss": 0.3587,
+      "step": 8880
+    },
+    {
+      "epoch": 1.8636077938403521,
+      "grad_norm": 0.7545658349990845,
+      "learning_rate": 3.314070005564097e-05,
+      "loss": 0.3679,
+      "step": 8895
+    },
+    {
+      "epoch": 1.8667504714016343,
+      "grad_norm": 0.9135695695877075,
+      "learning_rate": 3.2981087476893853e-05,
+      "loss": 0.3725,
+      "step": 8910
+    },
+    {
+      "epoch": 1.8698931489629165,
+      "grad_norm": 0.9788998961448669,
+      "learning_rate": 3.2821670814955026e-05,
+      "loss": 0.3149,
+      "step": 8925
+    },
+    {
+      "epoch": 1.8730358265241986,
+      "grad_norm": 0.7953155636787415,
+      "learning_rate": 3.266245190498311e-05,
+      "loss": 0.3461,
+      "step": 8940
+    },
+    {
+      "epoch": 1.8761785040854808,
+      "grad_norm": 0.9166163802146912,
+      "learning_rate": 3.250343257986027e-05,
+      "loss": 0.3866,
+      "step": 8955
+    },
+    {
+      "epoch": 1.879321181646763,
+      "grad_norm": 0.9379754066467285,
+      "learning_rate": 3.2344614670171025e-05,
+      "loss": 0.3928,
+      "step": 8970
+    },
+    {
+      "epoch": 1.8824638592080452,
+      "grad_norm": 0.8782539963722229,
+      "learning_rate": 3.2186000004181314e-05,
+      "loss": 0.3959,
+      "step": 8985
+    },
+    {
+      "epoch": 1.8856065367693273,
+      "grad_norm": 0.7237117886543274,
+      "learning_rate": 3.2027590407817407e-05,
+      "loss": 0.3458,
+      "step": 9000
+    },
+    {
+      "epoch": 1.8887492143306097,
+      "grad_norm": 0.8787809014320374,
+      "learning_rate": 3.186938770464486e-05,
+      "loss": 0.4081,
+      "step": 9015
+    },
+    {
+      "epoch": 1.8918918918918919,
+      "grad_norm": 0.7628602981567383,
+      "learning_rate": 3.1711393715847476e-05,
+      "loss": 0.3928,
+      "step": 9030
+    },
+    {
+      "epoch": 1.895034569453174,
+      "grad_norm": 0.9172194600105286,
+      "learning_rate": 3.15536102602065e-05,
+      "loss": 0.3777,
+      "step": 9045
+    },
+    {
+      "epoch": 1.8981772470144564,
+      "grad_norm": 0.8413445353507996,
+      "learning_rate": 3.13960391540795e-05,
+      "loss": 0.36,
+      "step": 9060
+    },
+    {
+      "epoch": 1.9013199245757386,
+      "grad_norm": 0.9793257117271423,
+      "learning_rate": 3.1238682211379586e-05,
+      "loss": 0.3801,
+      "step": 9075
+    },
+    {
+      "epoch": 1.9044626021370208,
+      "grad_norm": 0.7620652318000793,
+      "learning_rate": 3.1081541243554427e-05,
+      "loss": 0.3689,
+      "step": 9090
+    },
+    {
+      "epoch": 1.907605279698303,
+      "grad_norm": 0.8353012800216675,
+      "learning_rate": 3.092461805956551e-05,
+      "loss": 0.3961,
+      "step": 9105
+    },
+    {
+      "epoch": 1.9107479572595851,
+      "grad_norm": 0.8704758882522583,
+      "learning_rate": 3.0767914465867246e-05,
+      "loss": 0.3168,
+      "step": 9120
+    },
+    {
+      "epoch": 1.9138906348208673,
+      "grad_norm": 0.6754759550094604,
+      "learning_rate": 3.061143226638611e-05,
+      "loss": 0.3407,
+      "step": 9135
+    },
+    {
+      "epoch": 1.9170333123821495,
+      "grad_norm": 0.9682889580726624,
+      "learning_rate": 3.0455173262500093e-05,
+      "loss": 0.4251,
+      "step": 9150
+    },
+    {
+      "epoch": 1.9201759899434316,
+      "grad_norm": 0.8114556670188904,
+      "learning_rate": 3.0299139253017695e-05,
+      "loss": 0.3397,
+      "step": 9165
+    },
+    {
+      "epoch": 1.923318667504714,
+      "grad_norm": 0.8123522996902466,
+      "learning_rate": 3.014333203415741e-05,
+      "loss": 0.3372,
+      "step": 9180
+    },
+    {
+      "epoch": 1.9264613450659962,
+      "grad_norm": 0.6080268025398254,
+      "learning_rate": 2.9987753399526934e-05,
+      "loss": 0.3506,
+      "step": 9195
+    },
+    {
+      "epoch": 1.9296040226272786,
+      "grad_norm": 0.8804168701171875,
+      "learning_rate": 2.9832405140102637e-05,
+      "loss": 0.3689,
+      "step": 9210
+    },
+    {
+      "epoch": 1.9327467001885608,
+      "grad_norm": 0.8579033613204956,
+      "learning_rate": 2.9677289044208833e-05,
+      "loss": 0.3875,
+      "step": 9225
+    },
+    {
+      "epoch": 1.935889377749843,
+      "grad_norm": 0.9520317316055298,
+      "learning_rate": 2.952240689749722e-05,
+      "loss": 0.422,
+      "step": 9240
+    },
+    {
+      "epoch": 1.939032055311125,
+      "grad_norm": 0.9517824053764343,
+      "learning_rate": 2.9367760482926393e-05,
+      "loss": 0.3917,
+      "step": 9255
+    },
+    {
+      "epoch": 1.9421747328724073,
+      "grad_norm": 0.8813058733940125,
+      "learning_rate": 2.921335158074122e-05,
+      "loss": 0.3551,
+      "step": 9270
+    },
+    {
+      "epoch": 1.9453174104336894,
+      "grad_norm": 0.8402652144432068,
+      "learning_rate": 2.905918196845242e-05,
+      "loss": 0.3468,
+      "step": 9285
+    },
+    {
+      "epoch": 1.9484600879949716,
+      "grad_norm": 0.855032205581665,
+      "learning_rate": 2.8905253420816035e-05,
+      "loss": 0.3534,
+      "step": 9300
+    },
+    {
+      "epoch": 1.9516027655562538,
+      "grad_norm": 0.7760915756225586,
+      "learning_rate": 2.875156770981311e-05,
+      "loss": 0.348,
+      "step": 9315
+    },
+    {
+      "epoch": 1.9547454431175362,
+      "grad_norm": 0.946934163570404,
+      "learning_rate": 2.8598126604629195e-05,
+      "loss": 0.3556,
+      "step": 9330
+    },
+    {
+      "epoch": 1.9578881206788183,
+      "grad_norm": 0.7589976191520691,
+      "learning_rate": 2.844493187163395e-05,
+      "loss": 0.3944,
+      "step": 9345
+    },
+    {
+      "epoch": 1.9610307982401005,
+      "grad_norm": 0.8831868171691895,
+      "learning_rate": 2.8291985274360983e-05,
+      "loss": 0.3192,
+      "step": 9360
+    },
+    {
+      "epoch": 1.964173475801383,
+      "grad_norm": 0.8260477781295776,
+      "learning_rate": 2.8139288573487337e-05,
+      "loss": 0.3476,
+      "step": 9375
+    },
+    {
+      "epoch": 1.967316153362665,
+      "grad_norm": 0.9583712816238403,
+      "learning_rate": 2.7986843526813343e-05,
+      "loss": 0.3112,
+      "step": 9390
+    },
+    {
+      "epoch": 1.9704588309239472,
+      "grad_norm": 0.8534590005874634,
+      "learning_rate": 2.783465188924239e-05,
+      "loss": 0.3738,
+      "step": 9405
+    },
+    {
+      "epoch": 1.9736015084852294,
+      "grad_norm": 0.8562766909599304,
+      "learning_rate": 2.7682715412760696e-05,
+      "loss": 0.3831,
+      "step": 9420
+    },
+    {
+      "epoch": 1.9767441860465116,
+      "grad_norm": 0.649868905544281,
+      "learning_rate": 2.7531035846417107e-05,
+      "loss": 0.379,
+      "step": 9435
+    },
+    {
+      "epoch": 1.9798868636077938,
+      "grad_norm": 0.7702896595001221,
+      "learning_rate": 2.7379614936302982e-05,
+      "loss": 0.3617,
+      "step": 9450
+    },
+    {
+      "epoch": 1.983029541169076,
+      "grad_norm": 0.9378584623336792,
+      "learning_rate": 2.7228454425532157e-05,
+      "loss": 0.3681,
+      "step": 9465
+    },
+    {
+      "epoch": 1.9861722187303583,
+      "grad_norm": 1.0069222450256348,
+      "learning_rate": 2.7077556054220804e-05,
+      "loss": 0.3356,
+      "step": 9480
+    },
+    {
+      "epoch": 1.9893148962916405,
+      "grad_norm": 0.9345496892929077,
+      "learning_rate": 2.6926921559467412e-05,
+      "loss": 0.3974,
+      "step": 9495
+    },
+    {
+      "epoch": 1.9924575738529227,
+      "grad_norm": 0.8090453147888184,
+      "learning_rate": 2.6776552675332768e-05,
+      "loss": 0.3397,
+      "step": 9510
+    },
+    {
+      "epoch": 1.995600251414205,
+      "grad_norm": 0.647416353225708,
+      "learning_rate": 2.6626451132820085e-05,
+      "loss": 0.3259,
+      "step": 9525
+    },
+    {
+      "epoch": 1.9987429289754872,
+      "grad_norm": 0.7810280323028564,
+      "learning_rate": 2.6476618659855023e-05,
+      "loss": 0.3234,
+      "step": 9540
+    },
+    {
+      "epoch": 2.0018856065367694,
+      "grad_norm": 0.7231355309486389,
+      "learning_rate": 2.6327056981265708e-05,
+      "loss": 0.3276,
+      "step": 9555
+    },
+    {
+      "epoch": 2.0050282840980516,
+      "grad_norm": 0.7072864174842834,
+      "learning_rate": 2.6177767818763062e-05,
+      "loss": 0.2683,
+      "step": 9570
+    },
+    {
+      "epoch": 2.0081709616593337,
+      "grad_norm": 0.8502817749977112,
+      "learning_rate": 2.6028752890920783e-05,
+      "loss": 0.2844,
+      "step": 9585
+    },
+    {
+      "epoch": 2.011313639220616,
+      "grad_norm": 0.6001257300376892,
+      "learning_rate": 2.5880013913155743e-05,
+      "loss": 0.2582,
+      "step": 9600
+    },
+    {
+      "epoch": 2.014456316781898,
+      "grad_norm": 1.037467360496521,
+      "learning_rate": 2.5731552597708086e-05,
+      "loss": 0.2666,
+      "step": 9615
+    },
+    {
+      "epoch": 2.0175989943431802,
+      "grad_norm": 0.990047812461853,
+      "learning_rate": 2.5583370653621652e-05,
+      "loss": 0.3042,
+      "step": 9630
+    },
+    {
+      "epoch": 2.0207416719044624,
+      "grad_norm": 1.0518317222595215,
+      "learning_rate": 2.5435469786724204e-05,
+      "loss": 0.2543,
+      "step": 9645
+    },
+    {
+      "epoch": 2.023884349465745,
+      "grad_norm": 1.225774884223938,
+      "learning_rate": 2.528785169960779e-05,
+      "loss": 0.3183,
+      "step": 9660
+    },
+    {
+      "epoch": 2.027027027027027,
+      "grad_norm": 0.9525572061538696,
+      "learning_rate": 2.5140518091609256e-05,
+      "loss": 0.3426,
+      "step": 9675
+    },
+    {
+      "epoch": 2.0301697045883094,
+      "grad_norm": 1.0750566720962524,
+      "learning_rate": 2.4993470658790573e-05,
+      "loss": 0.3172,
+      "step": 9690
+    },
+    {
+      "epoch": 2.0333123821495915,
+      "grad_norm": 0.8268773555755615,
+      "learning_rate": 2.484671109391933e-05,
+      "loss": 0.31,
+      "step": 9705
+    },
+    {
+      "epoch": 2.0364550597108737,
+      "grad_norm": 0.679678201675415,
+      "learning_rate": 2.470024108644925e-05,
+      "loss": 0.2868,
+      "step": 9720
+    },
+    {
+      "epoch": 2.039597737272156,
+      "grad_norm": 0.997440755367279,
+      "learning_rate": 2.4554062322500797e-05,
+      "loss": 0.3291,
+      "step": 9735
+    },
+    {
+      "epoch": 2.042740414833438,
+      "grad_norm": 0.9968817830085754,
+      "learning_rate": 2.4408176484841732e-05,
+      "loss": 0.2664,
+      "step": 9750
+    },
+    {
+      "epoch": 2.04588309239472,
+      "grad_norm": 1.0939124822616577,
+      "learning_rate": 2.4262585252867686e-05,
+      "loss": 0.2895,
+      "step": 9765
+    },
+    {
+      "epoch": 2.0490257699560024,
+      "grad_norm": 1.0220900774002075,
+      "learning_rate": 2.4117290302582872e-05,
+      "loss": 0.3191,
+      "step": 9780
+    },
+    {
+      "epoch": 2.0521684475172846,
+      "grad_norm": 0.635898768901825,
+      "learning_rate": 2.397229330658084e-05,
+      "loss": 0.307,
+      "step": 9795
+    },
+    {
+      "epoch": 2.0553111250785667,
+      "grad_norm": 1.112257719039917,
+      "learning_rate": 2.382759593402517e-05,
+      "loss": 0.2748,
+      "step": 9810
+    },
+    {
+      "epoch": 2.0584538026398493,
+      "grad_norm": 0.9440275430679321,
+      "learning_rate": 2.3683199850630213e-05,
+      "loss": 0.2893,
+      "step": 9825
+    },
+    {
+      "epoch": 2.0615964802011315,
+      "grad_norm": 1.2118226289749146,
+      "learning_rate": 2.3539106718642034e-05,
+      "loss": 0.2791,
+      "step": 9840
+    },
+    {
+      "epoch": 2.0647391577624137,
+      "grad_norm": 1.1374374628067017,
+      "learning_rate": 2.339531819681914e-05,
+      "loss": 0.2777,
+      "step": 9855
+    },
+    {
+      "epoch": 2.067881835323696,
+      "grad_norm": 0.6932136416435242,
+      "learning_rate": 2.3251835940413517e-05,
+      "loss": 0.2828,
+      "step": 9870
+    },
+    {
+      "epoch": 2.071024512884978,
+      "grad_norm": 1.0308489799499512,
+      "learning_rate": 2.310866160115146e-05,
+      "loss": 0.2947,
+      "step": 9885
+    },
+    {
+      "epoch": 2.07416719044626,
+      "grad_norm": 1.063235878944397,
+      "learning_rate": 2.2965796827214665e-05,
+      "loss": 0.3204,
+      "step": 9900
+    },
+    {
+      "epoch": 2.0773098680075424,
+      "grad_norm": 1.1612193584442139,
+      "learning_rate": 2.282324326322115e-05,
+      "loss": 0.2976,
+      "step": 9915
+    },
+    {
+      "epoch": 2.0804525455688245,
+      "grad_norm": 0.8928938508033752,
+      "learning_rate": 2.2681002550206355e-05,
+      "loss": 0.2921,
+      "step": 9930
+    },
+    {
+      "epoch": 2.0835952231301067,
+      "grad_norm": 1.066124677658081,
+      "learning_rate": 2.253907632560439e-05,
+      "loss": 0.298,
+      "step": 9945
+    },
+    {
+      "epoch": 2.086737900691389,
+      "grad_norm": 0.8713576197624207,
+      "learning_rate": 2.2397466223228947e-05,
+      "loss": 0.275,
+      "step": 9960
+    },
+    {
+      "epoch": 2.0898805782526715,
+      "grad_norm": 1.1056296825408936,
+      "learning_rate": 2.2256173873254643e-05,
+      "loss": 0.3266,
+      "step": 9975
+    },
+    {
+      "epoch": 2.0930232558139537,
+      "grad_norm": 0.9172502160072327,
+      "learning_rate": 2.211520090219821e-05,
+      "loss": 0.2731,
+      "step": 9990
+    },
+    {
+      "epoch": 2.0951183741881416,
+      "eval_accuracy": 0.009820309467613697,
+      "eval_loss": 0.4190310835838318,
+      "eval_runtime": 424.9528,
+      "eval_samples_per_second": 11.26,
+      "eval_steps_per_second": 2.817,
+      "step": 10000
+    }
+  ],
+  "logging_steps": 15,
+  "max_steps": 14319,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 10000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.9082208625284874e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}