diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,25804 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 7.081212657667626,
+  "eval_steps": 500,
+  "global_step": 36000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.00019670035160187848,
+      "grad_norm": 0.8963498473167419,
+      "learning_rate": 3.0257186081694406e-08,
+      "loss": 0.9638,
+      "step": 1
+    },
+    {
+      "epoch": 0.0019670035160187847,
+      "grad_norm": 2.033263683319092,
+      "learning_rate": 3.0257186081694407e-07,
+      "loss": 0.8112,
+      "step": 10
+    },
+    {
+      "epoch": 0.003934007032037569,
+      "grad_norm": 1.7033007144927979,
+      "learning_rate": 6.051437216338881e-07,
+      "loss": 0.7522,
+      "step": 20
+    },
+    {
+      "epoch": 0.005901010548056355,
+      "grad_norm": 1.5076667070388794,
+      "learning_rate": 9.077155824508321e-07,
+      "loss": 0.9369,
+      "step": 30
+    },
+    {
+      "epoch": 0.007868014064075139,
+      "grad_norm": 1.0427778959274292,
+      "learning_rate": 1.2102874432677763e-06,
+      "loss": 0.7922,
+      "step": 40
+    },
+    {
+      "epoch": 0.009835017580093924,
+      "grad_norm": 2.3299496173858643,
+      "learning_rate": 1.5128593040847204e-06,
+      "loss": 0.7739,
+      "step": 50
+    },
+    {
+      "epoch": 0.01180202109611271,
+      "grad_norm": 0.9595165252685547,
+      "learning_rate": 1.8154311649016642e-06,
+      "loss": 0.7738,
+      "step": 60
+    },
+    {
+      "epoch": 0.013769024612131494,
+      "grad_norm": 1.5153745412826538,
+      "learning_rate": 2.118003025718608e-06,
+      "loss": 0.7583,
+      "step": 70
+    },
+    {
+      "epoch": 0.015736028128150278,
+      "grad_norm": 1.1440826654434204,
+      "learning_rate": 2.4205748865355526e-06,
+      "loss": 0.7769,
+      "step": 80
+    },
+    {
+      "epoch": 0.017703031644169063,
+      "grad_norm": 2.1569674015045166,
+      "learning_rate": 2.7231467473524962e-06,
+      "loss": 0.7269,
+      "step": 90
+    },
+    {
+      "epoch": 0.019670035160187848,
+      "grad_norm": 1.6127907037734985,
+      "learning_rate": 3.0257186081694407e-06,
+      "loss": 0.6396,
+      "step": 100
+    },
+    {
+      "epoch": 0.021637038676206633,
+      "grad_norm": 1.2766520977020264,
+      "learning_rate": 3.3282904689863844e-06,
+      "loss": 0.6954,
+      "step": 110
+    },
+    {
+      "epoch": 0.02360404219222542,
+      "grad_norm": 1.3288867473602295,
+      "learning_rate": 3.6308623298033285e-06,
+      "loss": 0.7179,
+      "step": 120
+    },
+    {
+      "epoch": 0.025571045708244203,
+      "grad_norm": 0.6131573915481567,
+      "learning_rate": 3.933434190620273e-06,
+      "loss": 0.6522,
+      "step": 130
+    },
+    {
+      "epoch": 0.02753804922426299,
+      "grad_norm": 0.6934183239936829,
+      "learning_rate": 4.236006051437216e-06,
+      "loss": 0.6974,
+      "step": 140
+    },
+    {
+      "epoch": 0.029505052740281774,
+      "grad_norm": 0.6823475360870361,
+      "learning_rate": 4.53857791225416e-06,
+      "loss": 0.6047,
+      "step": 150
+    },
+    {
+      "epoch": 0.031472056256300555,
+      "grad_norm": 1.4353721141815186,
+      "learning_rate": 4.841149773071105e-06,
+      "loss": 0.5891,
+      "step": 160
+    },
+    {
+      "epoch": 0.03343905977231934,
+      "grad_norm": 0.7245773077011108,
+      "learning_rate": 5.143721633888049e-06,
+      "loss": 0.6713,
+      "step": 170
+    },
+    {
+      "epoch": 0.035406063288338126,
+      "grad_norm": 0.84307461977005,
+      "learning_rate": 5.4462934947049925e-06,
+      "loss": 0.5529,
+      "step": 180
+    },
+    {
+      "epoch": 0.03737306680435691,
+      "grad_norm": 0.6518082618713379,
+      "learning_rate": 5.7488653555219365e-06,
+      "loss": 0.6697,
+      "step": 190
+    },
+    {
+      "epoch": 0.039340070320375696,
+      "grad_norm": 0.724744439125061,
+      "learning_rate": 6.0514372163388815e-06,
+      "loss": 0.6472,
+      "step": 200
+    },
+    {
+      "epoch": 0.04130707383639448,
+      "grad_norm": 0.6548435091972351,
+      "learning_rate": 6.354009077155825e-06,
+      "loss": 0.5387,
+      "step": 210
+    },
+    {
+      "epoch": 0.043274077352413266,
+      "grad_norm": 0.372646689414978,
+      "learning_rate": 6.656580937972769e-06,
+      "loss": 0.565,
+      "step": 220
+    },
+    {
+      "epoch": 0.04524108086843205,
+      "grad_norm": 0.5441346168518066,
+      "learning_rate": 6.959152798789714e-06,
+      "loss": 0.5121,
+      "step": 230
+    },
+    {
+      "epoch": 0.04720808438445084,
+      "grad_norm": 0.9839174151420593,
+      "learning_rate": 7.261724659606657e-06,
+      "loss": 0.5849,
+      "step": 240
+    },
+    {
+      "epoch": 0.04917508790046962,
+      "grad_norm": 1.2899951934814453,
+      "learning_rate": 7.564296520423601e-06,
+      "loss": 0.5569,
+      "step": 250
+    },
+    {
+      "epoch": 0.05114209141648841,
+      "grad_norm": 0.5471493601799011,
+      "learning_rate": 7.866868381240546e-06,
+      "loss": 0.5766,
+      "step": 260
+    },
+    {
+      "epoch": 0.05310909493250719,
+      "grad_norm": 0.7858185768127441,
+      "learning_rate": 8.169440242057489e-06,
+      "loss": 0.4571,
+      "step": 270
+    },
+    {
+      "epoch": 0.05507609844852598,
+      "grad_norm": 0.7448890209197998,
+      "learning_rate": 8.472012102874432e-06,
+      "loss": 0.4962,
+      "step": 280
+    },
+    {
+      "epoch": 0.05704310196454476,
+      "grad_norm": 0.7200512886047363,
+      "learning_rate": 8.774583963691377e-06,
+      "loss": 0.56,
+      "step": 290
+    },
+    {
+      "epoch": 0.05901010548056355,
+      "grad_norm": 0.725430965423584,
+      "learning_rate": 9.07715582450832e-06,
+      "loss": 0.593,
+      "step": 300
+    },
+    {
+      "epoch": 0.06097710899658233,
+      "grad_norm": 0.9160416126251221,
+      "learning_rate": 9.379727685325265e-06,
+      "loss": 0.6143,
+      "step": 310
+    },
+    {
+      "epoch": 0.06294411251260111,
+      "grad_norm": 0.6902405619621277,
+      "learning_rate": 9.68229954614221e-06,
+      "loss": 0.6567,
+      "step": 320
+    },
+    {
+      "epoch": 0.0649111160286199,
+      "grad_norm": 0.47368934750556946,
+      "learning_rate": 9.984871406959154e-06,
+      "loss": 0.601,
+      "step": 330
+    },
+    {
+      "epoch": 0.06687811954463868,
+      "grad_norm": 0.9510943293571472,
+      "learning_rate": 1.0287443267776098e-05,
+      "loss": 0.6315,
+      "step": 340
+    },
+    {
+      "epoch": 0.06884512306065747,
+      "grad_norm": 0.5105342268943787,
+      "learning_rate": 1.059001512859304e-05,
+      "loss": 0.5069,
+      "step": 350
+    },
+    {
+      "epoch": 0.07081212657667625,
+      "grad_norm": 0.7906575202941895,
+      "learning_rate": 1.0892586989409985e-05,
+      "loss": 0.6008,
+      "step": 360
+    },
+    {
+      "epoch": 0.07277913009269504,
+      "grad_norm": 0.6110881567001343,
+      "learning_rate": 1.119515885022693e-05,
+      "loss": 0.6122,
+      "step": 370
+    },
+    {
+      "epoch": 0.07474613360871382,
+      "grad_norm": 0.6386378407478333,
+      "learning_rate": 1.1497730711043873e-05,
+      "loss": 0.6167,
+      "step": 380
+    },
+    {
+      "epoch": 0.07671313712473261,
+      "grad_norm": 0.7307829856872559,
+      "learning_rate": 1.1800302571860818e-05,
+      "loss": 0.6236,
+      "step": 390
+    },
+    {
+      "epoch": 0.07868014064075139,
+      "grad_norm": 0.8760218620300293,
+      "learning_rate": 1.2102874432677763e-05,
+      "loss": 0.5672,
+      "step": 400
+    },
+    {
+      "epoch": 0.08064714415677018,
+      "grad_norm": 0.4042556881904602,
+      "learning_rate": 1.2405446293494704e-05,
+      "loss": 0.4975,
+      "step": 410
+    },
+    {
+      "epoch": 0.08261414767278896,
+      "grad_norm": 0.4333750903606415,
+      "learning_rate": 1.270801815431165e-05,
+      "loss": 0.634,
+      "step": 420
+    },
+    {
+      "epoch": 0.08458115118880775,
+      "grad_norm": 0.36654746532440186,
+      "learning_rate": 1.3010590015128594e-05,
+      "loss": 0.6007,
+      "step": 430
+    },
+    {
+      "epoch": 0.08654815470482653,
+      "grad_norm": 0.7794342041015625,
+      "learning_rate": 1.3313161875945538e-05,
+      "loss": 0.6959,
+      "step": 440
+    },
+    {
+      "epoch": 0.08851515822084532,
+      "grad_norm": 1.202374815940857,
+      "learning_rate": 1.3615733736762482e-05,
+      "loss": 0.4908,
+      "step": 450
+    },
+    {
+      "epoch": 0.0904821617368641,
+      "grad_norm": 1.0065367221832275,
+      "learning_rate": 1.3918305597579427e-05,
+      "loss": 0.5246,
+      "step": 460
+    },
+    {
+      "epoch": 0.0924491652528829,
+      "grad_norm": 0.7841054797172546,
+      "learning_rate": 1.4220877458396369e-05,
+      "loss": 0.6303,
+      "step": 470
+    },
+    {
+      "epoch": 0.09441616876890167,
+      "grad_norm": 0.49154767394065857,
+      "learning_rate": 1.4523449319213314e-05,
+      "loss": 0.4197,
+      "step": 480
+    },
+    {
+      "epoch": 0.09638317228492047,
+      "grad_norm": 0.7281906604766846,
+      "learning_rate": 1.4826021180030259e-05,
+      "loss": 0.5574,
+      "step": 490
+    },
+    {
+      "epoch": 0.09835017580093924,
+      "grad_norm": 0.46101605892181396,
+      "learning_rate": 1.5128593040847202e-05,
+      "loss": 0.5686,
+      "step": 500
+    },
+    {
+      "epoch": 0.09835017580093924,
+      "eval_loss": 0.3269956707954407,
+      "eval_runtime": 8.9032,
+      "eval_samples_per_second": 5.616,
+      "eval_steps_per_second": 2.808,
+      "step": 500
+    },
+    {
+      "epoch": 0.10031717931695804,
+      "grad_norm": 0.7430837750434875,
+      "learning_rate": 1.5431164901664147e-05,
+      "loss": 0.6991,
+      "step": 510
+    },
+    {
+      "epoch": 0.10228418283297681,
+      "grad_norm": 0.4325253367424011,
+      "learning_rate": 1.5733736762481092e-05,
+      "loss": 0.6192,
+      "step": 520
+    },
+    {
+      "epoch": 0.10425118634899559,
+      "grad_norm": 0.45161041617393494,
+      "learning_rate": 1.6036308623298033e-05,
+      "loss": 0.5495,
+      "step": 530
+    },
+    {
+      "epoch": 0.10621818986501438,
+      "grad_norm": 0.43786877393722534,
+      "learning_rate": 1.6338880484114978e-05,
+      "loss": 0.5687,
+      "step": 540
+    },
+    {
+      "epoch": 0.10818519338103316,
+      "grad_norm": 0.5659717917442322,
+      "learning_rate": 1.6641452344931923e-05,
+      "loss": 0.5838,
+      "step": 550
+    },
+    {
+      "epoch": 0.11015219689705195,
+      "grad_norm": 0.6761313676834106,
+      "learning_rate": 1.6944024205748865e-05,
+      "loss": 0.5822,
+      "step": 560
+    },
+    {
+      "epoch": 0.11211920041307073,
+      "grad_norm": 0.4713389575481415,
+      "learning_rate": 1.724659606656581e-05,
+      "loss": 0.5271,
+      "step": 570
+    },
+    {
+      "epoch": 0.11408620392908952,
+      "grad_norm": 0.7082213759422302,
+      "learning_rate": 1.7549167927382755e-05,
+      "loss": 0.5033,
+      "step": 580
+    },
+    {
+      "epoch": 0.1160532074451083,
+      "grad_norm": 0.6817394495010376,
+      "learning_rate": 1.78517397881997e-05,
+      "loss": 0.6074,
+      "step": 590
+    },
+    {
+      "epoch": 0.1180202109611271,
+      "grad_norm": 0.5328401923179626,
+      "learning_rate": 1.815431164901664e-05,
+      "loss": 0.6559,
+      "step": 600
+    },
+    {
+      "epoch": 0.11998721447714587,
+      "grad_norm": 0.5534793138504028,
+      "learning_rate": 1.8456883509833586e-05,
+      "loss": 0.5362,
+      "step": 610
+    },
+    {
+      "epoch": 0.12195421799316467,
+      "grad_norm": 0.3978525996208191,
+      "learning_rate": 1.875945537065053e-05,
+      "loss": 0.5953,
+      "step": 620
+    },
+    {
+      "epoch": 0.12392122150918344,
+      "grad_norm": 0.7429622411727905,
+      "learning_rate": 1.9062027231467476e-05,
+      "loss": 0.5147,
+      "step": 630
+    },
+    {
+      "epoch": 0.12588822502520222,
+      "grad_norm": 0.4334476888179779,
+      "learning_rate": 1.936459909228442e-05,
+      "loss": 0.5867,
+      "step": 640
+    },
+    {
+      "epoch": 0.12785522854122103,
+      "grad_norm": 0.5612448453903198,
+      "learning_rate": 1.9667170953101362e-05,
+      "loss": 0.471,
+      "step": 650
+    },
+    {
+      "epoch": 0.1298222320572398,
+      "grad_norm": 0.6829410791397095,
+      "learning_rate": 1.9969742813918307e-05,
+      "loss": 0.4989,
+      "step": 660
+    },
+    {
+      "epoch": 0.13178923557325858,
+      "grad_norm": 0.42095333337783813,
+      "learning_rate": 1.9997248463725582e-05,
+      "loss": 0.5226,
+      "step": 670
+    },
+    {
+      "epoch": 0.13375623908927736,
+      "grad_norm": 0.4745030105113983,
+      "learning_rate": 1.999419120119845e-05,
+      "loss": 0.49,
+      "step": 680
+    },
+    {
+      "epoch": 0.13572324260529617,
+      "grad_norm": 0.6165034770965576,
+      "learning_rate": 1.9991133938671313e-05,
+      "loss": 0.5754,
+      "step": 690
+    },
+    {
+      "epoch": 0.13769024612131495,
+      "grad_norm": 0.5824708342552185,
+      "learning_rate": 1.9988076676144183e-05,
+      "loss": 0.442,
+      "step": 700
+    },
+    {
+      "epoch": 0.13965724963733372,
+      "grad_norm": 0.764433741569519,
+      "learning_rate": 1.998501941361705e-05,
+      "loss": 0.557,
+      "step": 710
+    },
+    {
+      "epoch": 0.1416242531533525,
+      "grad_norm": 0.5732194185256958,
+      "learning_rate": 1.9981962151089914e-05,
+      "loss": 0.5354,
+      "step": 720
+    },
+    {
+      "epoch": 0.1435912566693713,
+      "grad_norm": 0.4292770028114319,
+      "learning_rate": 1.9978904888562783e-05,
+      "loss": 0.5574,
+      "step": 730
+    },
+    {
+      "epoch": 0.1455582601853901,
+      "grad_norm": 0.9347184300422668,
+      "learning_rate": 1.997584762603565e-05,
+      "loss": 0.5045,
+      "step": 740
+    },
+    {
+      "epoch": 0.14752526370140887,
+      "grad_norm": 0.7119300365447998,
+      "learning_rate": 1.9972790363508515e-05,
+      "loss": 0.4441,
+      "step": 750
+    },
+    {
+      "epoch": 0.14949226721742764,
+      "grad_norm": 0.48641037940979004,
+      "learning_rate": 1.9969733100981384e-05,
+      "loss": 0.6329,
+      "step": 760
+    },
+    {
+      "epoch": 0.15145927073344645,
+      "grad_norm": 0.7848897576332092,
+      "learning_rate": 1.996667583845425e-05,
+      "loss": 0.4577,
+      "step": 770
+    },
+    {
+      "epoch": 0.15342627424946523,
+      "grad_norm": 0.6484516263008118,
+      "learning_rate": 1.996361857592712e-05,
+      "loss": 0.4841,
+      "step": 780
+    },
+    {
+      "epoch": 0.155393277765484,
+      "grad_norm": 0.7445515394210815,
+      "learning_rate": 1.9960561313399985e-05,
+      "loss": 0.4113,
+      "step": 790
+    },
+    {
+      "epoch": 0.15736028128150278,
+      "grad_norm": 0.6570308804512024,
+      "learning_rate": 1.995750405087285e-05,
+      "loss": 0.4707,
+      "step": 800
+    },
+    {
+      "epoch": 0.15932728479752156,
+      "grad_norm": 0.7139286994934082,
+      "learning_rate": 1.9954446788345716e-05,
+      "loss": 0.5357,
+      "step": 810
+    },
+    {
+      "epoch": 0.16129428831354037,
+      "grad_norm": 0.8385933637619019,
+      "learning_rate": 1.9951389525818582e-05,
+      "loss": 0.4862,
+      "step": 820
+    },
+    {
+      "epoch": 0.16326129182955915,
+      "grad_norm": 0.55597984790802,
+      "learning_rate": 1.994833226329145e-05,
+      "loss": 0.4843,
+      "step": 830
+    },
+    {
+      "epoch": 0.16522829534557792,
+      "grad_norm": 0.6878874897956848,
+      "learning_rate": 1.9945275000764317e-05,
+      "loss": 0.5163,
+      "step": 840
+    },
+    {
+      "epoch": 0.1671952988615967,
+      "grad_norm": 0.8103552460670471,
+      "learning_rate": 1.9942217738237183e-05,
+      "loss": 0.5898,
+      "step": 850
+    },
+    {
+      "epoch": 0.1691623023776155,
+      "grad_norm": 0.6601850390434265,
+      "learning_rate": 1.9939160475710052e-05,
+      "loss": 0.4824,
+      "step": 860
+    },
+    {
+      "epoch": 0.1711293058936343,
+      "grad_norm": 0.6179708242416382,
+      "learning_rate": 1.9936103213182918e-05,
+      "loss": 0.5301,
+      "step": 870
+    },
+    {
+      "epoch": 0.17309630940965307,
+      "grad_norm": 0.5425893664360046,
+      "learning_rate": 1.9933045950655784e-05,
+      "loss": 0.5353,
+      "step": 880
+    },
+    {
+      "epoch": 0.17506331292567184,
+      "grad_norm": 0.533669650554657,
+      "learning_rate": 1.9929988688128653e-05,
+      "loss": 0.4981,
+      "step": 890
+    },
+    {
+      "epoch": 0.17703031644169065,
+      "grad_norm": 0.5814224481582642,
+      "learning_rate": 1.992693142560152e-05,
+      "loss": 0.6519,
+      "step": 900
+    },
+    {
+      "epoch": 0.17899731995770943,
+      "grad_norm": 0.8669481873512268,
+      "learning_rate": 1.9923874163074388e-05,
+      "loss": 0.6702,
+      "step": 910
+    },
+    {
+      "epoch": 0.1809643234737282,
+      "grad_norm": 0.8260190486907959,
+      "learning_rate": 1.992081690054725e-05,
+      "loss": 0.551,
+      "step": 920
+    },
+    {
+      "epoch": 0.18293132698974698,
+      "grad_norm": 0.8739385008811951,
+      "learning_rate": 1.991775963802012e-05,
+      "loss": 0.4831,
+      "step": 930
+    },
+    {
+      "epoch": 0.1848983305057658,
+      "grad_norm": 1.0561296939849854,
+      "learning_rate": 1.9914702375492985e-05,
+      "loss": 0.5674,
+      "step": 940
+    },
+    {
+      "epoch": 0.18686533402178457,
+      "grad_norm": 0.6565110683441162,
+      "learning_rate": 1.991164511296585e-05,
+      "loss": 0.39,
+      "step": 950
+    },
+    {
+      "epoch": 0.18883233753780335,
+      "grad_norm": 0.45213982462882996,
+      "learning_rate": 1.990858785043872e-05,
+      "loss": 0.4733,
+      "step": 960
+    },
+    {
+      "epoch": 0.19079934105382212,
+      "grad_norm": 0.5731518268585205,
+      "learning_rate": 1.9905530587911586e-05,
+      "loss": 0.5276,
+      "step": 970
+    },
+    {
+      "epoch": 0.19276634456984093,
+      "grad_norm": 0.5559749603271484,
+      "learning_rate": 1.990247332538445e-05,
+      "loss": 0.5062,
+      "step": 980
+    },
+    {
+      "epoch": 0.1947333480858597,
+      "grad_norm": 0.7155632376670837,
+      "learning_rate": 1.989941606285732e-05,
+      "loss": 0.5511,
+      "step": 990
+    },
+    {
+      "epoch": 0.1967003516018785,
+      "grad_norm": 0.7516645193099976,
+      "learning_rate": 1.9896358800330186e-05,
+      "loss": 0.4663,
+      "step": 1000
+    },
+    {
+      "epoch": 0.1967003516018785,
+      "eval_loss": 0.2955791652202606,
+      "eval_runtime": 8.8701,
+      "eval_samples_per_second": 5.637,
+      "eval_steps_per_second": 2.818,
+      "step": 1000
+    },
+    {
+      "epoch": 0.19866735511789727,
+      "grad_norm": 0.6724827885627747,
+      "learning_rate": 1.9893301537803052e-05,
+      "loss": 0.3468,
+      "step": 1010
+    },
+    {
+      "epoch": 0.20063435863391607,
+      "grad_norm": 0.8730838298797607,
+      "learning_rate": 1.989024427527592e-05,
+      "loss": 0.4836,
+      "step": 1020
+    },
+    {
+      "epoch": 0.20260136214993485,
+      "grad_norm": 0.9460917711257935,
+      "learning_rate": 1.9887187012748784e-05,
+      "loss": 0.4645,
+      "step": 1030
+    },
+    {
+      "epoch": 0.20456836566595363,
+      "grad_norm": 1.7044711112976074,
+      "learning_rate": 1.9884129750221653e-05,
+      "loss": 0.6025,
+      "step": 1040
+    },
+    {
+      "epoch": 0.2065353691819724,
+      "grad_norm": 1.0980366468429565,
+      "learning_rate": 1.988107248769452e-05,
+      "loss": 0.5283,
+      "step": 1050
+    },
+    {
+      "epoch": 0.20850237269799118,
+      "grad_norm": 0.7067188024520874,
+      "learning_rate": 1.9878015225167388e-05,
+      "loss": 0.4825,
+      "step": 1060
+    },
+    {
+      "epoch": 0.21046937621401,
+      "grad_norm": 0.6638745665550232,
+      "learning_rate": 1.9874957962640254e-05,
+      "loss": 0.4808,
+      "step": 1070
+    },
+    {
+      "epoch": 0.21243637973002877,
+      "grad_norm": 0.6948477029800415,
+      "learning_rate": 1.987190070011312e-05,
+      "loss": 0.5871,
+      "step": 1080
+    },
+    {
+      "epoch": 0.21440338324604755,
+      "grad_norm": 0.7186503410339355,
+      "learning_rate": 1.986884343758599e-05,
+      "loss": 0.6654,
+      "step": 1090
+    },
+    {
+      "epoch": 0.21637038676206632,
+      "grad_norm": 0.737399160861969,
+      "learning_rate": 1.9865786175058854e-05,
+      "loss": 0.5239,
+      "step": 1100
+    },
+    {
+      "epoch": 0.21833739027808513,
+      "grad_norm": 0.7045579552650452,
+      "learning_rate": 1.986272891253172e-05,
+      "loss": 0.4644,
+      "step": 1110
+    },
+    {
+      "epoch": 0.2203043937941039,
+      "grad_norm": 0.6651056408882141,
+      "learning_rate": 1.985967165000459e-05,
+      "loss": 0.4719,
+      "step": 1120
+    },
+    {
+      "epoch": 0.2222713973101227,
+      "grad_norm": 0.6420923471450806,
+      "learning_rate": 1.9856614387477455e-05,
+      "loss": 0.4595,
+      "step": 1130
+    },
+    {
+      "epoch": 0.22423840082614147,
+      "grad_norm": 0.5954447984695435,
+      "learning_rate": 1.985355712495032e-05,
+      "loss": 0.4917,
+      "step": 1140
+    },
+    {
+      "epoch": 0.22620540434216027,
+      "grad_norm": 0.6256354451179504,
+      "learning_rate": 1.9850499862423187e-05,
+      "loss": 0.4765,
+      "step": 1150
+    },
+    {
+      "epoch": 0.22817240785817905,
+      "grad_norm": 1.5549144744873047,
+      "learning_rate": 1.9847442599896052e-05,
+      "loss": 0.62,
+      "step": 1160
+    },
+    {
+      "epoch": 0.23013941137419783,
+      "grad_norm": 1.0412925481796265,
+      "learning_rate": 1.984438533736892e-05,
+      "loss": 0.5452,
+      "step": 1170
+    },
+    {
+      "epoch": 0.2321064148902166,
+      "grad_norm": 0.7398366332054138,
+      "learning_rate": 1.9841328074841787e-05,
+      "loss": 0.6219,
+      "step": 1180
+    },
+    {
+      "epoch": 0.2340734184062354,
+      "grad_norm": 0.9944019317626953,
+      "learning_rate": 1.9838270812314657e-05,
+      "loss": 0.5009,
+      "step": 1190
+    },
+    {
+      "epoch": 0.2360404219222542,
+      "grad_norm": 0.9399623870849609,
+      "learning_rate": 1.9835213549787522e-05,
+      "loss": 0.4176,
+      "step": 1200
+    },
+    {
+      "epoch": 0.23800742543827297,
+      "grad_norm": 0.6136744022369385,
+      "learning_rate": 1.9832156287260388e-05,
+      "loss": 0.5097,
+      "step": 1210
+    },
+    {
+      "epoch": 0.23997442895429175,
+      "grad_norm": 0.657649576663971,
+      "learning_rate": 1.9829099024733257e-05,
+      "loss": 0.3794,
+      "step": 1220
+    },
+    {
+      "epoch": 0.24194143247031055,
+      "grad_norm": 0.6419724822044373,
+      "learning_rate": 1.9826041762206123e-05,
+      "loss": 0.6161,
+      "step": 1230
+    },
+    {
+      "epoch": 0.24390843598632933,
+      "grad_norm": 0.43958067893981934,
+      "learning_rate": 1.982298449967899e-05,
+      "loss": 0.5344,
+      "step": 1240
+    },
+    {
+      "epoch": 0.2458754395023481,
+      "grad_norm": 0.9898470044136047,
+      "learning_rate": 1.9819927237151858e-05,
+      "loss": 0.5375,
+      "step": 1250
+    },
+    {
+      "epoch": 0.2478424430183669,
+      "grad_norm": 1.0850481986999512,
+      "learning_rate": 1.981686997462472e-05,
+      "loss": 0.5602,
+      "step": 1260
+    },
+    {
+      "epoch": 0.2498094465343857,
+      "grad_norm": 2.405172109603882,
+      "learning_rate": 1.981381271209759e-05,
+      "loss": 0.4474,
+      "step": 1270
+    },
+    {
+      "epoch": 0.25177645005040444,
+      "grad_norm": 0.7804758548736572,
+      "learning_rate": 1.9810755449570455e-05,
+      "loss": 0.5501,
+      "step": 1280
+    },
+    {
+      "epoch": 0.25374345356642325,
+      "grad_norm": 0.706414520740509,
+      "learning_rate": 1.980769818704332e-05,
+      "loss": 0.5222,
+      "step": 1290
+    },
+    {
+      "epoch": 0.25571045708244206,
+      "grad_norm": 0.8361694812774658,
+      "learning_rate": 1.980464092451619e-05,
+      "loss": 0.559,
+      "step": 1300
+    },
+    {
+      "epoch": 0.2576774605984608,
+      "grad_norm": 0.8130835890769958,
+      "learning_rate": 1.9801583661989056e-05,
+      "loss": 0.5284,
+      "step": 1310
+    },
+    {
+      "epoch": 0.2596444641144796,
+      "grad_norm": 1.4580860137939453,
+      "learning_rate": 1.9798526399461925e-05,
+      "loss": 0.4438,
+      "step": 1320
+    },
+    {
+      "epoch": 0.26161146763049836,
+      "grad_norm": 0.7845149636268616,
+      "learning_rate": 1.979546913693479e-05,
+      "loss": 0.4895,
+      "step": 1330
+    },
+    {
+      "epoch": 0.26357847114651717,
+      "grad_norm": 0.851684033870697,
+      "learning_rate": 1.9792411874407657e-05,
+      "loss": 0.5788,
+      "step": 1340
+    },
+    {
+      "epoch": 0.265545474662536,
+      "grad_norm": 1.1903982162475586,
+      "learning_rate": 1.9789354611880526e-05,
+      "loss": 0.5113,
+      "step": 1350
+    },
+    {
+      "epoch": 0.2675124781785547,
+      "grad_norm": 0.595227837562561,
+      "learning_rate": 1.978629734935339e-05,
+      "loss": 0.4556,
+      "step": 1360
+    },
+    {
+      "epoch": 0.26947948169457353,
+      "grad_norm": 0.6648783087730408,
+      "learning_rate": 1.9783240086826257e-05,
+      "loss": 0.3519,
+      "step": 1370
+    },
+    {
+      "epoch": 0.27144648521059234,
+      "grad_norm": 0.5223883986473083,
+      "learning_rate": 1.9780182824299123e-05,
+      "loss": 0.6364,
+      "step": 1380
+    },
+    {
+      "epoch": 0.2734134887266111,
+      "grad_norm": 0.96187824010849,
+      "learning_rate": 1.977712556177199e-05,
+      "loss": 0.4932,
+      "step": 1390
+    },
+    {
+      "epoch": 0.2753804922426299,
+      "grad_norm": 0.8614581227302551,
+      "learning_rate": 1.9774068299244858e-05,
+      "loss": 0.5926,
+      "step": 1400
+    },
+    {
+      "epoch": 0.27734749575864864,
+      "grad_norm": 1.119659423828125,
+      "learning_rate": 1.9771011036717724e-05,
+      "loss": 0.3975,
+      "step": 1410
+    },
+    {
+      "epoch": 0.27931449927466745,
+      "grad_norm": 0.9885017275810242,
+      "learning_rate": 1.976795377419059e-05,
+      "loss": 0.5982,
+      "step": 1420
+    },
+    {
+      "epoch": 0.28128150279068626,
+      "grad_norm": 1.0573168992996216,
+      "learning_rate": 1.976489651166346e-05,
+      "loss": 0.5111,
+      "step": 1430
+    },
+    {
+      "epoch": 0.283248506306705,
+      "grad_norm": 0.9164770245552063,
+      "learning_rate": 1.9761839249136325e-05,
+      "loss": 0.6055,
+      "step": 1440
+    },
+    {
+      "epoch": 0.2852155098227238,
+      "grad_norm": 0.7315860986709595,
+      "learning_rate": 1.9758781986609194e-05,
+      "loss": 0.4542,
+      "step": 1450
+    },
+    {
+      "epoch": 0.2871825133387426,
+      "grad_norm": 0.679958701133728,
+      "learning_rate": 1.975572472408206e-05,
+      "loss": 0.5706,
+      "step": 1460
+    },
+    {
+      "epoch": 0.28914951685476137,
+      "grad_norm": 0.8225506544113159,
+      "learning_rate": 1.9752667461554925e-05,
+      "loss": 0.5128,
+      "step": 1470
+    },
+    {
+      "epoch": 0.2911165203707802,
+      "grad_norm": 0.3660297095775604,
+      "learning_rate": 1.974961019902779e-05,
+      "loss": 0.6373,
+      "step": 1480
+    },
+    {
+      "epoch": 0.2930835238867989,
+      "grad_norm": 0.6732741594314575,
+      "learning_rate": 1.9746552936500657e-05,
+      "loss": 0.577,
+      "step": 1490
+    },
+    {
+      "epoch": 0.29505052740281773,
+      "grad_norm": 0.38270995020866394,
+      "learning_rate": 1.9743495673973526e-05,
+      "loss": 0.5341,
+      "step": 1500
+    },
+    {
+      "epoch": 0.29505052740281773,
+      "eval_loss": 0.2840212285518646,
+      "eval_runtime": 8.8894,
+      "eval_samples_per_second": 5.625,
+      "eval_steps_per_second": 2.812,
+      "step": 1500
+    },
+    {
+      "epoch": 0.29701753091883654,
+      "grad_norm": 0.7969959378242493,
+      "learning_rate": 1.9740438411446392e-05,
+      "loss": 0.4869,
+      "step": 1510
+    },
+    {
+      "epoch": 0.2989845344348553,
+      "grad_norm": 0.6881989240646362,
+      "learning_rate": 1.9737381148919258e-05,
+      "loss": 0.4047,
+      "step": 1520
+    },
+    {
+      "epoch": 0.3009515379508741,
+      "grad_norm": 0.5518563389778137,
+      "learning_rate": 1.9734323886392127e-05,
+      "loss": 0.5494,
+      "step": 1530
+    },
+    {
+      "epoch": 0.3029185414668929,
+      "grad_norm": 0.6757585406303406,
+      "learning_rate": 1.9731266623864993e-05,
+      "loss": 0.6683,
+      "step": 1540
+    },
+    {
+      "epoch": 0.30488554498291165,
+      "grad_norm": 1.0032838582992554,
+      "learning_rate": 1.972820936133786e-05,
+      "loss": 0.442,
+      "step": 1550
+    },
+    {
+      "epoch": 0.30685254849893046,
+      "grad_norm": 0.6686198711395264,
+      "learning_rate": 1.9725152098810728e-05,
+      "loss": 0.4487,
+      "step": 1560
+    },
+    {
+      "epoch": 0.3088195520149492,
+      "grad_norm": 0.6934469938278198,
+      "learning_rate": 1.9722094836283593e-05,
+      "loss": 0.5372,
+      "step": 1570
+    },
+    {
+      "epoch": 0.310786555530968,
+      "grad_norm": 1.1455458402633667,
+      "learning_rate": 1.9719037573756463e-05,
+      "loss": 0.5002,
+      "step": 1580
+    },
+    {
+      "epoch": 0.3127535590469868,
+      "grad_norm": 0.7301696538925171,
+      "learning_rate": 1.9715980311229328e-05,
+      "loss": 0.4819,
+      "step": 1590
+    },
+    {
+      "epoch": 0.31472056256300557,
+      "grad_norm": 0.905255138874054,
+      "learning_rate": 1.9712923048702194e-05,
+      "loss": 0.5643,
+      "step": 1600
+    },
+    {
+      "epoch": 0.3166875660790244,
+      "grad_norm": 0.869118869304657,
+      "learning_rate": 1.970986578617506e-05,
+      "loss": 0.6479,
+      "step": 1610
+    },
+    {
+      "epoch": 0.3186545695950431,
+      "grad_norm": 0.6825863122940063,
+      "learning_rate": 1.9706808523647926e-05,
+      "loss": 0.4414,
+      "step": 1620
+    },
+    {
+      "epoch": 0.32062157311106193,
+      "grad_norm": 0.8974255323410034,
+      "learning_rate": 1.9703751261120795e-05,
+      "loss": 0.3885,
+      "step": 1630
+    },
+    {
+      "epoch": 0.32258857662708074,
+      "grad_norm": 0.6362448930740356,
+      "learning_rate": 1.970069399859366e-05,
+      "loss": 0.5692,
+      "step": 1640
+    },
+    {
+      "epoch": 0.3245555801430995,
+      "grad_norm": 0.6962127685546875,
+      "learning_rate": 1.9697636736066526e-05,
+      "loss": 0.5196,
+      "step": 1650
+    },
+    {
+      "epoch": 0.3265225836591183,
+      "grad_norm": 0.8557025790214539,
+      "learning_rate": 1.9694579473539396e-05,
+      "loss": 0.476,
+      "step": 1660
+    },
+    {
+      "epoch": 0.3284895871751371,
+      "grad_norm": 0.439887672662735,
+      "learning_rate": 1.969152221101226e-05,
+      "loss": 0.6215,
+      "step": 1670
+    },
+    {
+      "epoch": 0.33045659069115585,
+      "grad_norm": 0.5827410817146301,
+      "learning_rate": 1.9688464948485127e-05,
+      "loss": 0.4626,
+      "step": 1680
+    },
+    {
+      "epoch": 0.33242359420717466,
+      "grad_norm": 0.8372606635093689,
+      "learning_rate": 1.9685407685957996e-05,
+      "loss": 0.4751,
+      "step": 1690
+    },
+    {
+      "epoch": 0.3343905977231934,
+      "grad_norm": 0.7339947819709778,
+      "learning_rate": 1.9682350423430862e-05,
+      "loss": 0.5446,
+      "step": 1700
+    },
+    {
+      "epoch": 0.3363576012392122,
+      "grad_norm": 0.8041804432868958,
+      "learning_rate": 1.9679293160903728e-05,
+      "loss": 0.5248,
+      "step": 1710
+    },
+    {
+      "epoch": 0.338324604755231,
+      "grad_norm": 0.6950403451919556,
+      "learning_rate": 1.9676235898376594e-05,
+      "loss": 0.6077,
+      "step": 1720
+    },
+    {
+      "epoch": 0.34029160827124977,
+      "grad_norm": 0.8875169157981873,
+      "learning_rate": 1.9673178635849463e-05,
+      "loss": 0.5625,
+      "step": 1730
+    },
+    {
+      "epoch": 0.3422586117872686,
+      "grad_norm": 1.0996932983398438,
+      "learning_rate": 1.967012137332233e-05,
+      "loss": 0.5417,
+      "step": 1740
+    },
+    {
+      "epoch": 0.3442256153032874,
+      "grad_norm": 0.685312807559967,
+      "learning_rate": 1.9667064110795194e-05,
+      "loss": 0.4577,
+      "step": 1750
+    },
+    {
+      "epoch": 0.34619261881930613,
+      "grad_norm": 0.6270304322242737,
+      "learning_rate": 1.9664006848268063e-05,
+      "loss": 0.5995,
+      "step": 1760
+    },
+    {
+      "epoch": 0.34815962233532494,
+      "grad_norm": 0.49572035670280457,
+      "learning_rate": 1.966094958574093e-05,
+      "loss": 0.5602,
+      "step": 1770
+    },
+    {
+      "epoch": 0.3501266258513437,
+      "grad_norm": 0.8444635272026062,
+      "learning_rate": 1.9657892323213795e-05,
+      "loss": 0.536,
+      "step": 1780
+    },
+    {
+      "epoch": 0.3520936293673625,
+      "grad_norm": 0.7252330780029297,
+      "learning_rate": 1.9654835060686664e-05,
+      "loss": 0.5329,
+      "step": 1790
+    },
+    {
+      "epoch": 0.3540606328833813,
+      "grad_norm": 1.0123865604400635,
+      "learning_rate": 1.965177779815953e-05,
+      "loss": 0.5508,
+      "step": 1800
+    },
+    {
+      "epoch": 0.35602763639940005,
+      "grad_norm": 0.6840813159942627,
+      "learning_rate": 1.9648720535632396e-05,
+      "loss": 0.4869,
+      "step": 1810
+    },
+    {
+      "epoch": 0.35799463991541886,
+      "grad_norm": 0.9481569528579712,
+      "learning_rate": 1.964566327310526e-05,
+      "loss": 0.4472,
+      "step": 1820
+    },
+    {
+      "epoch": 0.35996164343143766,
+      "grad_norm": 0.7030127048492432,
+      "learning_rate": 1.964260601057813e-05,
+      "loss": 0.4563,
+      "step": 1830
+    },
+    {
+      "epoch": 0.3619286469474564,
+      "grad_norm": 1.2299069166183472,
+      "learning_rate": 1.9639548748050996e-05,
+      "loss": 0.3174,
+      "step": 1840
+    },
+    {
+      "epoch": 0.3638956504634752,
+      "grad_norm": 0.9066298007965088,
+      "learning_rate": 1.9636491485523862e-05,
+      "loss": 0.4359,
+      "step": 1850
+    },
+    {
+      "epoch": 0.36586265397949397,
+      "grad_norm": 0.9739953279495239,
+      "learning_rate": 1.963343422299673e-05,
+      "loss": 0.3874,
+      "step": 1860
+    },
+    {
+      "epoch": 0.3678296574955128,
+      "grad_norm": 0.6393256187438965,
+      "learning_rate": 1.9630376960469597e-05,
+      "loss": 0.4747,
+      "step": 1870
+    },
+    {
+      "epoch": 0.3697966610115316,
+      "grad_norm": 0.8060562014579773,
+      "learning_rate": 1.9627319697942463e-05,
+      "loss": 0.6424,
+      "step": 1880
+    },
+    {
+      "epoch": 0.37176366452755033,
+      "grad_norm": 0.8817802667617798,
+      "learning_rate": 1.9624262435415332e-05,
+      "loss": 0.4514,
+      "step": 1890
+    },
+    {
+      "epoch": 0.37373066804356914,
+      "grad_norm": 1.1933926343917847,
+      "learning_rate": 1.9621205172888198e-05,
+      "loss": 0.5605,
+      "step": 1900
+    },
+    {
+      "epoch": 0.3756976715595879,
+      "grad_norm": 1.1892311573028564,
+      "learning_rate": 1.9618147910361064e-05,
+      "loss": 0.4527,
+      "step": 1910
+    },
+    {
+      "epoch": 0.3776646750756067,
+      "grad_norm": 0.7793095111846924,
+      "learning_rate": 1.9615090647833933e-05,
+      "loss": 0.5498,
+      "step": 1920
+    },
+    {
+      "epoch": 0.3796316785916255,
+      "grad_norm": 0.4772995114326477,
+      "learning_rate": 1.9612033385306795e-05,
+      "loss": 0.4916,
+      "step": 1930
+    },
+    {
+      "epoch": 0.38159868210764425,
+      "grad_norm": 0.8411799669265747,
+      "learning_rate": 1.9608976122779664e-05,
+      "loss": 0.4298,
+      "step": 1940
+    },
+    {
+      "epoch": 0.38356568562366306,
+      "grad_norm": 1.2099697589874268,
+      "learning_rate": 1.960591886025253e-05,
+      "loss": 0.5165,
+      "step": 1950
+    },
+    {
+      "epoch": 0.38553268913968186,
+      "grad_norm": 1.0067557096481323,
+      "learning_rate": 1.96028615977254e-05,
+      "loss": 0.5543,
+      "step": 1960
+    },
+    {
+      "epoch": 0.3874996926557006,
+      "grad_norm": 1.0297846794128418,
+      "learning_rate": 1.9599804335198265e-05,
+      "loss": 0.4319,
+      "step": 1970
+    },
+    {
+      "epoch": 0.3894666961717194,
+      "grad_norm": 0.788569450378418,
+      "learning_rate": 1.959674707267113e-05,
+      "loss": 0.4775,
+      "step": 1980
+    },
+    {
+      "epoch": 0.39143369968773817,
+      "grad_norm": 0.9311039447784424,
+      "learning_rate": 1.9593689810144e-05,
+      "loss": 0.4415,
+      "step": 1990
+    },
+    {
+      "epoch": 0.393400703203757,
+      "grad_norm": 0.8891676068305969,
+      "learning_rate": 1.9590632547616866e-05,
+      "loss": 0.4075,
+      "step": 2000
+    },
+    {
+      "epoch": 0.393400703203757,
+      "eval_loss": 0.2848837673664093,
+      "eval_runtime": 8.8626,
+      "eval_samples_per_second": 5.642,
+      "eval_steps_per_second": 2.821,
+      "step": 2000
+    },
+    {
+      "epoch": 0.3953677067197758,
+      "grad_norm": 0.6306418180465698,
+      "learning_rate": 1.958757528508973e-05,
+      "loss": 0.5114,
+      "step": 2010
+    },
+    {
+      "epoch": 0.39733471023579453,
+      "grad_norm": 0.8960371613502502,
+      "learning_rate": 1.95845180225626e-05,
+      "loss": 0.386,
+      "step": 2020
+    },
+    {
+      "epoch": 0.39930171375181334,
+      "grad_norm": 1.0953959226608276,
+      "learning_rate": 1.9581460760035467e-05,
+      "loss": 0.5311,
+      "step": 2030
+    },
+    {
+      "epoch": 0.40126871726783214,
+      "grad_norm": 0.8647001385688782,
+      "learning_rate": 1.9578403497508332e-05,
+      "loss": 0.4544,
+      "step": 2040
+    },
+    {
+      "epoch": 0.4032357207838509,
+      "grad_norm": 0.9456301927566528,
+      "learning_rate": 1.9575346234981198e-05,
+      "loss": 0.456,
+      "step": 2050
+    },
+    {
+      "epoch": 0.4052027242998697,
+      "grad_norm": 0.7155416011810303,
+      "learning_rate": 1.9572288972454064e-05,
+      "loss": 0.5354,
+      "step": 2060
+    },
+    {
+      "epoch": 0.40716972781588845,
+      "grad_norm": 1.0676209926605225,
+      "learning_rate": 1.9569231709926933e-05,
+      "loss": 0.4509,
+      "step": 2070
+    },
+    {
+      "epoch": 0.40913673133190726,
+      "grad_norm": 1.194039225578308,
+      "learning_rate": 1.95661744473998e-05,
+      "loss": 0.6663,
+      "step": 2080
+    },
+    {
+      "epoch": 0.41110373484792606,
+      "grad_norm": 0.9243388175964355,
+      "learning_rate": 1.9563117184872668e-05,
+      "loss": 0.4106,
+      "step": 2090
+    },
+    {
+      "epoch": 0.4130707383639448,
+      "grad_norm": 0.9473809599876404,
+      "learning_rate": 1.9560059922345534e-05,
+      "loss": 0.4455,
+      "step": 2100
+    },
+    {
+      "epoch": 0.4150377418799636,
+      "grad_norm": 0.6198266744613647,
+      "learning_rate": 1.95570026598184e-05,
+      "loss": 0.4656,
+      "step": 2110
+    },
+    {
+      "epoch": 0.41700474539598237,
+      "grad_norm": 0.6981731057167053,
+      "learning_rate": 1.955394539729127e-05,
+      "loss": 0.4844,
+      "step": 2120
+    },
+    {
+      "epoch": 0.4189717489120012,
+      "grad_norm": 0.9984627366065979,
+      "learning_rate": 1.9550888134764134e-05,
+      "loss": 0.4839,
+      "step": 2130
+    },
+    {
+      "epoch": 0.42093875242802,
+      "grad_norm": 1.2150676250457764,
+      "learning_rate": 1.9547830872237e-05,
+      "loss": 0.5235,
+      "step": 2140
+    },
+    {
+      "epoch": 0.42290575594403873,
+      "grad_norm": 0.801630973815918,
+      "learning_rate": 1.954477360970987e-05,
+      "loss": 0.5282,
+      "step": 2150
+    },
+    {
+      "epoch": 0.42487275946005754,
+      "grad_norm": 1.599314570426941,
+      "learning_rate": 1.9541716347182732e-05,
+      "loss": 0.4939,
+      "step": 2160
+    },
+    {
+      "epoch": 0.42683976297607634,
+      "grad_norm": 0.885888397693634,
+      "learning_rate": 1.95386590846556e-05,
+      "loss": 0.5504,
+      "step": 2170
+    },
+    {
+      "epoch": 0.4288067664920951,
+      "grad_norm": 0.7052297592163086,
+      "learning_rate": 1.9535601822128467e-05,
+      "loss": 0.4576,
+      "step": 2180
+    },
+    {
+      "epoch": 0.4307737700081139,
+      "grad_norm": 0.635510265827179,
+      "learning_rate": 1.9532544559601333e-05,
+      "loss": 0.5811,
+      "step": 2190
+    },
+    {
+      "epoch": 0.43274077352413265,
+      "grad_norm": 0.619910478591919,
+      "learning_rate": 1.9529487297074202e-05,
+      "loss": 0.4898,
+      "step": 2200
+    },
+    {
+      "epoch": 0.43470777704015146,
+      "grad_norm": 0.8020810484886169,
+      "learning_rate": 1.9526430034547067e-05,
+      "loss": 0.4318,
+      "step": 2210
+    },
+    {
+      "epoch": 0.43667478055617026,
+      "grad_norm": 1.2118501663208008,
+      "learning_rate": 1.9523372772019937e-05,
+      "loss": 0.4156,
+      "step": 2220
+    },
+    {
+      "epoch": 0.438641784072189,
+      "grad_norm": 0.6296743154525757,
+      "learning_rate": 1.9520315509492802e-05,
+      "loss": 0.4371,
+      "step": 2230
+    },
+    {
+      "epoch": 0.4406087875882078,
+      "grad_norm": 0.7382553219795227,
+      "learning_rate": 1.9517258246965668e-05,
+      "loss": 0.4819,
+      "step": 2240
+    },
+    {
+      "epoch": 0.4425757911042266,
+      "grad_norm": 0.9509519338607788,
+      "learning_rate": 1.9514200984438537e-05,
+      "loss": 0.5008,
+      "step": 2250
+    },
+    {
+      "epoch": 0.4445427946202454,
+      "grad_norm": 1.098402976989746,
+      "learning_rate": 1.9511143721911403e-05,
+      "loss": 0.4785,
+      "step": 2260
+    },
+    {
+      "epoch": 0.4465097981362642,
+      "grad_norm": 0.6164669990539551,
+      "learning_rate": 1.950808645938427e-05,
+      "loss": 0.4082,
+      "step": 2270
+    },
+    {
+      "epoch": 0.44847680165228293,
+      "grad_norm": 0.9613497257232666,
+      "learning_rate": 1.9505029196857135e-05,
+      "loss": 0.3826,
+      "step": 2280
+    },
+    {
+      "epoch": 0.45044380516830174,
+      "grad_norm": 0.8639736175537109,
+      "learning_rate": 1.950197193433e-05,
+      "loss": 0.537,
+      "step": 2290
+    },
+    {
+      "epoch": 0.45241080868432054,
+      "grad_norm": 1.09634530544281,
+      "learning_rate": 1.949891467180287e-05,
+      "loss": 0.5444,
+      "step": 2300
+    },
+    {
+      "epoch": 0.4543778122003393,
+      "grad_norm": 0.7725170254707336,
+      "learning_rate": 1.9495857409275735e-05,
+      "loss": 0.3995,
+      "step": 2310
+    },
+    {
+      "epoch": 0.4563448157163581,
+      "grad_norm": 0.5570437908172607,
+      "learning_rate": 1.94928001467486e-05,
+      "loss": 0.4471,
+      "step": 2320
+    },
+    {
+      "epoch": 0.4583118192323769,
+      "grad_norm": 0.8947836756706238,
+      "learning_rate": 1.948974288422147e-05,
+      "loss": 0.4954,
+      "step": 2330
+    },
+    {
+      "epoch": 0.46027882274839566,
+      "grad_norm": 0.6316766738891602,
+      "learning_rate": 1.9486685621694336e-05,
+      "loss": 0.5009,
+      "step": 2340
+    },
+    {
+      "epoch": 0.46224582626441446,
+      "grad_norm": 0.6918854117393494,
+      "learning_rate": 1.9483628359167205e-05,
+      "loss": 0.5725,
+      "step": 2350
+    },
+    {
+      "epoch": 0.4642128297804332,
+      "grad_norm": 1.0366955995559692,
+      "learning_rate": 1.948057109664007e-05,
+      "loss": 0.5113,
+      "step": 2360
+    },
+    {
+      "epoch": 0.466179833296452,
+      "grad_norm": 1.386698603630066,
+      "learning_rate": 1.9477513834112937e-05,
+      "loss": 0.481,
+      "step": 2370
+    },
+    {
+      "epoch": 0.4681468368124708,
+      "grad_norm": 0.4483737349510193,
+      "learning_rate": 1.9474456571585806e-05,
+      "loss": 0.4917,
+      "step": 2380
+    },
+    {
+      "epoch": 0.4701138403284896,
+      "grad_norm": 0.7257867455482483,
+      "learning_rate": 1.947139930905867e-05,
+      "loss": 0.5398,
+      "step": 2390
+    },
+    {
+      "epoch": 0.4720808438445084,
+      "grad_norm": 1.0875515937805176,
+      "learning_rate": 1.9468342046531538e-05,
+      "loss": 0.3761,
+      "step": 2400
+    },
+    {
+      "epoch": 0.47404784736052713,
+      "grad_norm": 1.1179277896881104,
+      "learning_rate": 1.9465284784004403e-05,
+      "loss": 0.3777,
+      "step": 2410
+    },
+    {
+      "epoch": 0.47601485087654594,
+      "grad_norm": 1.132418155670166,
+      "learning_rate": 1.946222752147727e-05,
+      "loss": 0.4218,
+      "step": 2420
+    },
+    {
+      "epoch": 0.47798185439256474,
+      "grad_norm": 0.5543782114982605,
+      "learning_rate": 1.945917025895014e-05,
+      "loss": 0.3894,
+      "step": 2430
+    },
+    {
+      "epoch": 0.4799488579085835,
+      "grad_norm": 0.7127739191055298,
+      "learning_rate": 1.9456112996423004e-05,
+      "loss": 0.5318,
+      "step": 2440
+    },
+    {
+      "epoch": 0.4819158614246023,
+      "grad_norm": 0.6442409157752991,
+      "learning_rate": 1.945305573389587e-05,
+      "loss": 0.4879,
+      "step": 2450
+    },
+    {
+      "epoch": 0.4838828649406211,
+      "grad_norm": 0.7430324554443359,
+      "learning_rate": 1.944999847136874e-05,
+      "loss": 0.4242,
+      "step": 2460
+    },
+    {
+      "epoch": 0.48584986845663986,
+      "grad_norm": 1.3987452983856201,
+      "learning_rate": 1.9446941208841605e-05,
+      "loss": 0.5224,
+      "step": 2470
+    },
+    {
+      "epoch": 0.48781687197265866,
+      "grad_norm": 1.1393516063690186,
+      "learning_rate": 1.9443883946314474e-05,
+      "loss": 0.4801,
+      "step": 2480
+    },
+    {
+      "epoch": 0.4897838754886774,
+      "grad_norm": 1.146475911140442,
+      "learning_rate": 1.944082668378734e-05,
+      "loss": 0.4509,
+      "step": 2490
+    },
+    {
+      "epoch": 0.4917508790046962,
+      "grad_norm": 0.9160381555557251,
+      "learning_rate": 1.9437769421260206e-05,
+      "loss": 0.4539,
+      "step": 2500
+    },
+    {
+      "epoch": 0.4917508790046962,
+      "eval_loss": 0.27316009998321533,
+      "eval_runtime": 8.8733,
+      "eval_samples_per_second": 5.635,
+      "eval_steps_per_second": 2.817,
+      "step": 2500
+    },
+    {
+      "epoch": 0.493717882520715,
+      "grad_norm": 0.8414424657821655,
+      "learning_rate": 1.943471215873307e-05,
+      "loss": 0.4398,
+      "step": 2510
+    },
+    {
+      "epoch": 0.4956848860367338,
+      "grad_norm": 1.182061791419983,
+      "learning_rate": 1.9431654896205937e-05,
+      "loss": 0.4886,
+      "step": 2520
+    },
+    {
+      "epoch": 0.4976518895527526,
+      "grad_norm": 0.6704056262969971,
+      "learning_rate": 1.9428597633678806e-05,
+      "loss": 0.479,
+      "step": 2530
+    },
+    {
+      "epoch": 0.4996188930687714,
+      "grad_norm": 1.2297146320343018,
+      "learning_rate": 1.9425540371151672e-05,
+      "loss": 0.3661,
+      "step": 2540
+    },
+    {
+      "epoch": 0.5015858965847901,
+      "grad_norm": 0.8071584701538086,
+      "learning_rate": 1.9422483108624538e-05,
+      "loss": 0.4658,
+      "step": 2550
+    },
+    {
+      "epoch": 0.5035529001008089,
+      "grad_norm": 0.942072868347168,
+      "learning_rate": 1.9419425846097407e-05,
+      "loss": 0.4275,
+      "step": 2560
+    },
+    {
+      "epoch": 0.5055199036168277,
+      "grad_norm": 0.8354616761207581,
+      "learning_rate": 1.9416368583570273e-05,
+      "loss": 0.4297,
+      "step": 2570
+    },
+    {
+      "epoch": 0.5074869071328465,
+      "grad_norm": 0.7049144506454468,
+      "learning_rate": 1.941331132104314e-05,
+      "loss": 0.4462,
+      "step": 2580
+    },
+    {
+      "epoch": 0.5094539106488652,
+      "grad_norm": 0.7258726358413696,
+      "learning_rate": 1.9410254058516008e-05,
+      "loss": 0.6207,
+      "step": 2590
+    },
+    {
+      "epoch": 0.5114209141648841,
+      "grad_norm": 0.611747682094574,
+      "learning_rate": 1.9407196795988873e-05,
+      "loss": 0.3919,
+      "step": 2600
+    },
+    {
+      "epoch": 0.5133879176809029,
+      "grad_norm": 1.1328556537628174,
+      "learning_rate": 1.940413953346174e-05,
+      "loss": 0.4299,
+      "step": 2610
+    },
+    {
+      "epoch": 0.5153549211969216,
+      "grad_norm": 1.74727463722229,
+      "learning_rate": 1.9401082270934605e-05,
+      "loss": 0.5916,
+      "step": 2620
+    },
+    {
+      "epoch": 0.5173219247129405,
+      "grad_norm": 1.2672849893569946,
+      "learning_rate": 1.9398025008407474e-05,
+      "loss": 0.4927,
+      "step": 2630
+    },
+    {
+      "epoch": 0.5192889282289592,
+      "grad_norm": 1.1896312236785889,
+      "learning_rate": 1.939496774588034e-05,
+      "loss": 0.4934,
+      "step": 2640
+    },
+    {
+      "epoch": 0.521255931744978,
+      "grad_norm": 0.9513353109359741,
+      "learning_rate": 1.9391910483353206e-05,
+      "loss": 0.4687,
+      "step": 2650
+    },
+    {
+      "epoch": 0.5232229352609967,
+      "grad_norm": 1.061252474784851,
+      "learning_rate": 1.9388853220826075e-05,
+      "loss": 0.5134,
+      "step": 2660
+    },
+    {
+      "epoch": 0.5251899387770156,
+      "grad_norm": 0.498430997133255,
+      "learning_rate": 1.938579595829894e-05,
+      "loss": 0.4358,
+      "step": 2670
+    },
+    {
+      "epoch": 0.5271569422930343,
+      "grad_norm": 0.8934110999107361,
+      "learning_rate": 1.9382738695771806e-05,
+      "loss": 0.4829,
+      "step": 2680
+    },
+    {
+      "epoch": 0.5291239458090531,
+      "grad_norm": 0.6681143045425415,
+      "learning_rate": 1.9379681433244676e-05,
+      "loss": 0.5481,
+      "step": 2690
+    },
+    {
+      "epoch": 0.531090949325072,
+      "grad_norm": 0.9681861400604248,
+      "learning_rate": 1.937662417071754e-05,
+      "loss": 0.5768,
+      "step": 2700
+    },
+    {
+      "epoch": 0.5330579528410907,
+      "grad_norm": 0.7599331140518188,
+      "learning_rate": 1.9373566908190407e-05,
+      "loss": 0.4921,
+      "step": 2710
+    },
+    {
+      "epoch": 0.5350249563571094,
+      "grad_norm": 0.8120267391204834,
+      "learning_rate": 1.9370509645663276e-05,
+      "loss": 0.4022,
+      "step": 2720
+    },
+    {
+      "epoch": 0.5369919598731283,
+      "grad_norm": 1.1031180620193481,
+      "learning_rate": 1.9367452383136142e-05,
+      "loss": 0.4954,
+      "step": 2730
+    },
+    {
+      "epoch": 0.5389589633891471,
+      "grad_norm": 1.092254877090454,
+      "learning_rate": 1.9364395120609008e-05,
+      "loss": 0.4775,
+      "step": 2740
+    },
+    {
+      "epoch": 0.5409259669051658,
+      "grad_norm": 1.720622181892395,
+      "learning_rate": 1.9361337858081874e-05,
+      "loss": 0.5302,
+      "step": 2750
+    },
+    {
+      "epoch": 0.5428929704211847,
+      "grad_norm": 0.5872963070869446,
+      "learning_rate": 1.9358280595554743e-05,
+      "loss": 0.4608,
+      "step": 2760
+    },
+    {
+      "epoch": 0.5448599739372034,
+      "grad_norm": 1.1749262809753418,
+      "learning_rate": 1.935522333302761e-05,
+      "loss": 0.4118,
+      "step": 2770
+    },
+    {
+      "epoch": 0.5468269774532222,
+      "grad_norm": 1.0945734977722168,
+      "learning_rate": 1.9352166070500474e-05,
+      "loss": 0.3325,
+      "step": 2780
+    },
+    {
+      "epoch": 0.548793980969241,
+      "grad_norm": 1.1984425783157349,
+      "learning_rate": 1.9349108807973344e-05,
+      "loss": 0.4699,
+      "step": 2790
+    },
+    {
+      "epoch": 0.5507609844852598,
+      "grad_norm": 1.0518896579742432,
+      "learning_rate": 1.934605154544621e-05,
+      "loss": 0.5218,
+      "step": 2800
+    },
+    {
+      "epoch": 0.5527279880012785,
+      "grad_norm": 1.1263470649719238,
+      "learning_rate": 1.9342994282919075e-05,
+      "loss": 0.3322,
+      "step": 2810
+    },
+    {
+      "epoch": 0.5546949915172973,
+      "grad_norm": 0.9612912535667419,
+      "learning_rate": 1.9339937020391944e-05,
+      "loss": 0.486,
+      "step": 2820
+    },
+    {
+      "epoch": 0.5566619950333161,
+      "grad_norm": 0.9742056131362915,
+      "learning_rate": 1.933687975786481e-05,
+      "loss": 0.4987,
+      "step": 2830
+    },
+    {
+      "epoch": 0.5586289985493349,
+      "grad_norm": 1.2318615913391113,
+      "learning_rate": 1.9333822495337676e-05,
+      "loss": 0.5631,
+      "step": 2840
+    },
+    {
+      "epoch": 0.5605960020653536,
+      "grad_norm": 0.7405826449394226,
+      "learning_rate": 1.933076523281054e-05,
+      "loss": 0.427,
+      "step": 2850
+    },
+    {
+      "epoch": 0.5625630055813725,
+      "grad_norm": 1.1158024072647095,
+      "learning_rate": 1.932770797028341e-05,
+      "loss": 0.54,
+      "step": 2860
+    },
+    {
+      "epoch": 0.5645300090973913,
+      "grad_norm": 1.1322060823440552,
+      "learning_rate": 1.9324650707756277e-05,
+      "loss": 0.4481,
+      "step": 2870
+    },
+    {
+      "epoch": 0.56649701261341,
+      "grad_norm": 0.7638188004493713,
+      "learning_rate": 1.9321593445229142e-05,
+      "loss": 0.4681,
+      "step": 2880
+    },
+    {
+      "epoch": 0.5684640161294289,
+      "grad_norm": 0.7837921977043152,
+      "learning_rate": 1.931853618270201e-05,
+      "loss": 0.4704,
+      "step": 2890
+    },
+    {
+      "epoch": 0.5704310196454476,
+      "grad_norm": 0.8661313056945801,
+      "learning_rate": 1.9315478920174877e-05,
+      "loss": 0.4953,
+      "step": 2900
+    },
+    {
+      "epoch": 0.5723980231614664,
+      "grad_norm": 1.4302645921707153,
+      "learning_rate": 1.9312421657647743e-05,
+      "loss": 0.4387,
+      "step": 2910
+    },
+    {
+      "epoch": 0.5743650266774852,
+      "grad_norm": 1.5569978952407837,
+      "learning_rate": 1.9309364395120612e-05,
+      "loss": 0.4401,
+      "step": 2920
+    },
+    {
+      "epoch": 0.576332030193504,
+      "grad_norm": 1.0568249225616455,
+      "learning_rate": 1.9306307132593478e-05,
+      "loss": 0.5414,
+      "step": 2930
+    },
+    {
+      "epoch": 0.5782990337095227,
+      "grad_norm": 1.0306973457336426,
+      "learning_rate": 1.9303249870066344e-05,
+      "loss": 0.4364,
+      "step": 2940
+    },
+    {
+      "epoch": 0.5802660372255415,
+      "grad_norm": 1.2997857332229614,
+      "learning_rate": 1.930019260753921e-05,
+      "loss": 0.4826,
+      "step": 2950
+    },
+    {
+      "epoch": 0.5822330407415603,
+      "grad_norm": 1.0597947835922241,
+      "learning_rate": 1.9297135345012075e-05,
+      "loss": 0.4344,
+      "step": 2960
+    },
+    {
+      "epoch": 0.5842000442575791,
+      "grad_norm": 0.6990482211112976,
+      "learning_rate": 1.9294078082484945e-05,
+      "loss": 0.5371,
+      "step": 2970
+    },
+    {
+      "epoch": 0.5861670477735978,
+      "grad_norm": 1.2495508193969727,
+      "learning_rate": 1.929102081995781e-05,
+      "loss": 0.5635,
+      "step": 2980
+    },
+    {
+      "epoch": 0.5881340512896167,
+      "grad_norm": 1.3085463047027588,
+      "learning_rate": 1.928796355743068e-05,
+      "loss": 0.6532,
+      "step": 2990
+    },
+    {
+      "epoch": 0.5901010548056355,
+      "grad_norm": 0.8088992834091187,
+      "learning_rate": 1.9284906294903545e-05,
+      "loss": 0.4938,
+      "step": 3000
+    },
+    {
+      "epoch": 0.5901010548056355,
+      "eval_loss": 0.26935356855392456,
+      "eval_runtime": 8.8729,
+      "eval_samples_per_second": 5.635,
+      "eval_steps_per_second": 2.818,
+      "step": 3000
+    },
+    {
+      "epoch": 0.5920680583216542,
+      "grad_norm": 0.9487748742103577,
+      "learning_rate": 1.928184903237641e-05,
+      "loss": 0.4492,
+      "step": 3010
+    },
+    {
+      "epoch": 0.5940350618376731,
+      "grad_norm": 0.9440038800239563,
+      "learning_rate": 1.927879176984928e-05,
+      "loss": 0.4755,
+      "step": 3020
+    },
+    {
+      "epoch": 0.5960020653536918,
+      "grad_norm": 0.7290757298469543,
+      "learning_rate": 1.9275734507322146e-05,
+      "loss": 0.425,
+      "step": 3030
+    },
+    {
+      "epoch": 0.5979690688697106,
+      "grad_norm": 1.2127468585968018,
+      "learning_rate": 1.9272677244795012e-05,
+      "loss": 0.4825,
+      "step": 3040
+    },
+    {
+      "epoch": 0.5999360723857294,
+      "grad_norm": 1.1375706195831299,
+      "learning_rate": 1.926961998226788e-05,
+      "loss": 0.4468,
+      "step": 3050
+    },
+    {
+      "epoch": 0.6019030759017482,
+      "grad_norm": 0.8501954078674316,
+      "learning_rate": 1.9266562719740747e-05,
+      "loss": 0.4823,
+      "step": 3060
+    },
+    {
+      "epoch": 0.6038700794177669,
+      "grad_norm": 0.7859975099563599,
+      "learning_rate": 1.9263505457213612e-05,
+      "loss": 0.5721,
+      "step": 3070
+    },
+    {
+      "epoch": 0.6058370829337858,
+      "grad_norm": 0.6325510144233704,
+      "learning_rate": 1.9260448194686478e-05,
+      "loss": 0.3817,
+      "step": 3080
+    },
+    {
+      "epoch": 0.6078040864498045,
+      "grad_norm": 0.8559825420379639,
+      "learning_rate": 1.9257390932159344e-05,
+      "loss": 0.3648,
+      "step": 3090
+    },
+    {
+      "epoch": 0.6097710899658233,
+      "grad_norm": 1.0981616973876953,
+      "learning_rate": 1.9254333669632213e-05,
+      "loss": 0.5788,
+      "step": 3100
+    },
+    {
+      "epoch": 0.611738093481842,
+      "grad_norm": 0.7142055630683899,
+      "learning_rate": 1.925127640710508e-05,
+      "loss": 0.4691,
+      "step": 3110
+    },
+    {
+      "epoch": 0.6137050969978609,
+      "grad_norm": 0.9014882445335388,
+      "learning_rate": 1.9248219144577948e-05,
+      "loss": 0.4137,
+      "step": 3120
+    },
+    {
+      "epoch": 0.6156721005138797,
+      "grad_norm": 1.2298983335494995,
+      "learning_rate": 1.9245161882050814e-05,
+      "loss": 0.4547,
+      "step": 3130
+    },
+    {
+      "epoch": 0.6176391040298984,
+      "grad_norm": 0.9861557483673096,
+      "learning_rate": 1.924210461952368e-05,
+      "loss": 0.546,
+      "step": 3140
+    },
+    {
+      "epoch": 0.6196061075459173,
+      "grad_norm": 0.9207095503807068,
+      "learning_rate": 1.923904735699655e-05,
+      "loss": 0.6043,
+      "step": 3150
+    },
+    {
+      "epoch": 0.621573111061936,
+      "grad_norm": 0.9119741320610046,
+      "learning_rate": 1.9235990094469415e-05,
+      "loss": 0.3983,
+      "step": 3160
+    },
+    {
+      "epoch": 0.6235401145779548,
+      "grad_norm": 0.946865975856781,
+      "learning_rate": 1.923293283194228e-05,
+      "loss": 0.3886,
+      "step": 3170
+    },
+    {
+      "epoch": 0.6255071180939736,
+      "grad_norm": 0.8447843790054321,
+      "learning_rate": 1.9229875569415146e-05,
+      "loss": 0.4373,
+      "step": 3180
+    },
+    {
+      "epoch": 0.6274741216099924,
+      "grad_norm": 1.0305899381637573,
+      "learning_rate": 1.9226818306888012e-05,
+      "loss": 0.612,
+      "step": 3190
+    },
+    {
+      "epoch": 0.6294411251260111,
+      "grad_norm": 0.8156121373176575,
+      "learning_rate": 1.922376104436088e-05,
+      "loss": 0.3686,
+      "step": 3200
+    },
+    {
+      "epoch": 0.63140812864203,
+      "grad_norm": 0.884971559047699,
+      "learning_rate": 1.9220703781833747e-05,
+      "loss": 0.52,
+      "step": 3210
+    },
+    {
+      "epoch": 0.6333751321580487,
+      "grad_norm": 1.0552936792373657,
+      "learning_rate": 1.9217646519306613e-05,
+      "loss": 0.529,
+      "step": 3220
+    },
+    {
+      "epoch": 0.6353421356740675,
+      "grad_norm": 0.8704593777656555,
+      "learning_rate": 1.9214589256779482e-05,
+      "loss": 0.5397,
+      "step": 3230
+    },
+    {
+      "epoch": 0.6373091391900862,
+      "grad_norm": 1.075453281402588,
+      "learning_rate": 1.9211531994252348e-05,
+      "loss": 0.5448,
+      "step": 3240
+    },
+    {
+      "epoch": 0.6392761427061051,
+      "grad_norm": 0.6276763677597046,
+      "learning_rate": 1.9208474731725217e-05,
+      "loss": 0.4736,
+      "step": 3250
+    },
+    {
+      "epoch": 0.6412431462221239,
+      "grad_norm": 1.0386992692947388,
+      "learning_rate": 1.9205417469198083e-05,
+      "loss": 0.516,
+      "step": 3260
+    },
+    {
+      "epoch": 0.6432101497381426,
+      "grad_norm": 1.1383495330810547,
+      "learning_rate": 1.920236020667095e-05,
+      "loss": 0.5796,
+      "step": 3270
+    },
+    {
+      "epoch": 0.6451771532541615,
+      "grad_norm": 0.5847461819648743,
+      "learning_rate": 1.9199302944143818e-05,
+      "loss": 0.5272,
+      "step": 3280
+    },
+    {
+      "epoch": 0.6471441567701802,
+      "grad_norm": 0.7087602019309998,
+      "learning_rate": 1.919624568161668e-05,
+      "loss": 0.4371,
+      "step": 3290
+    },
+    {
+      "epoch": 0.649111160286199,
+      "grad_norm": 1.0593681335449219,
+      "learning_rate": 1.919318841908955e-05,
+      "loss": 0.4453,
+      "step": 3300
+    },
+    {
+      "epoch": 0.6510781638022178,
+      "grad_norm": 0.9373090863227844,
+      "learning_rate": 1.9190131156562415e-05,
+      "loss": 0.5262,
+      "step": 3310
+    },
+    {
+      "epoch": 0.6530451673182366,
+      "grad_norm": 1.0394222736358643,
+      "learning_rate": 1.918707389403528e-05,
+      "loss": 0.4508,
+      "step": 3320
+    },
+    {
+      "epoch": 0.6550121708342553,
+      "grad_norm": 1.1478387117385864,
+      "learning_rate": 1.918401663150815e-05,
+      "loss": 0.4628,
+      "step": 3330
+    },
+    {
+      "epoch": 0.6569791743502742,
+      "grad_norm": 1.0079622268676758,
+      "learning_rate": 1.9180959368981016e-05,
+      "loss": 0.5326,
+      "step": 3340
+    },
+    {
+      "epoch": 0.658946177866293,
+      "grad_norm": 1.0284887552261353,
+      "learning_rate": 1.917790210645388e-05,
+      "loss": 0.3955,
+      "step": 3350
+    },
+    {
+      "epoch": 0.6609131813823117,
+      "grad_norm": 0.9022873640060425,
+      "learning_rate": 1.917484484392675e-05,
+      "loss": 0.3859,
+      "step": 3360
+    },
+    {
+      "epoch": 0.6628801848983306,
+      "grad_norm": 1.5396238565444946,
+      "learning_rate": 1.9171787581399616e-05,
+      "loss": 0.5984,
+      "step": 3370
+    },
+    {
+      "epoch": 0.6648471884143493,
+      "grad_norm": 2.1132633686065674,
+      "learning_rate": 1.9168730318872485e-05,
+      "loss": 0.5249,
+      "step": 3380
+    },
+    {
+      "epoch": 0.6668141919303681,
+      "grad_norm": 0.8763299584388733,
+      "learning_rate": 1.916567305634535e-05,
+      "loss": 0.4281,
+      "step": 3390
+    },
+    {
+      "epoch": 0.6687811954463868,
+      "grad_norm": 1.1695775985717773,
+      "learning_rate": 1.9162615793818217e-05,
+      "loss": 0.4416,
+      "step": 3400
+    },
+    {
+      "epoch": 0.6707481989624057,
+      "grad_norm": 1.1734123229980469,
+      "learning_rate": 1.9159558531291083e-05,
+      "loss": 0.4649,
+      "step": 3410
+    },
+    {
+      "epoch": 0.6727152024784244,
+      "grad_norm": 1.1211220026016235,
+      "learning_rate": 1.915650126876395e-05,
+      "loss": 0.4709,
+      "step": 3420
+    },
+    {
+      "epoch": 0.6746822059944432,
+      "grad_norm": 0.763123631477356,
+      "learning_rate": 1.9153444006236818e-05,
+      "loss": 0.3934,
+      "step": 3430
+    },
+    {
+      "epoch": 0.676649209510462,
+      "grad_norm": 0.5031880140304565,
+      "learning_rate": 1.9150386743709684e-05,
+      "loss": 0.5802,
+      "step": 3440
+    },
+    {
+      "epoch": 0.6786162130264808,
+      "grad_norm": 1.1181432008743286,
+      "learning_rate": 1.914732948118255e-05,
+      "loss": 0.4726,
+      "step": 3450
+    },
+    {
+      "epoch": 0.6805832165424995,
+      "grad_norm": 1.1385880708694458,
+      "learning_rate": 1.914427221865542e-05,
+      "loss": 0.5769,
+      "step": 3460
+    },
+    {
+      "epoch": 0.6825502200585184,
+      "grad_norm": 1.035854697227478,
+      "learning_rate": 1.9141214956128284e-05,
+      "loss": 0.5229,
+      "step": 3470
+    },
+    {
+      "epoch": 0.6845172235745371,
+      "grad_norm": 0.8765487670898438,
+      "learning_rate": 1.913815769360115e-05,
+      "loss": 0.5184,
+      "step": 3480
+    },
+    {
+      "epoch": 0.6864842270905559,
+      "grad_norm": 0.7699954509735107,
+      "learning_rate": 1.913510043107402e-05,
+      "loss": 0.5459,
+      "step": 3490
+    },
+    {
+      "epoch": 0.6884512306065748,
+      "grad_norm": 1.094295859336853,
+      "learning_rate": 1.9132043168546885e-05,
+      "loss": 0.4648,
+      "step": 3500
+    },
+    {
+      "epoch": 0.6884512306065748,
+      "eval_loss": 0.26830142736434937,
+      "eval_runtime": 8.8663,
+      "eval_samples_per_second": 5.639,
+      "eval_steps_per_second": 2.82,
+      "step": 3500
+    },
+    {
+      "epoch": 0.6904182341225935,
+      "grad_norm": 0.8461592793464661,
+      "learning_rate": 1.9128985906019754e-05,
+      "loss": 0.4858,
+      "step": 3510
+    },
+    {
+      "epoch": 0.6923852376386123,
+      "grad_norm": 1.2828164100646973,
+      "learning_rate": 1.9125928643492617e-05,
+      "loss": 0.5672,
+      "step": 3520
+    },
+    {
+      "epoch": 0.694352241154631,
+      "grad_norm": 1.2198454141616821,
+      "learning_rate": 1.9122871380965486e-05,
+      "loss": 0.5289,
+      "step": 3530
+    },
+    {
+      "epoch": 0.6963192446706499,
+      "grad_norm": 1.0390441417694092,
+      "learning_rate": 1.911981411843835e-05,
+      "loss": 0.5492,
+      "step": 3540
+    },
+    {
+      "epoch": 0.6982862481866686,
+      "grad_norm": 1.178147792816162,
+      "learning_rate": 1.9116756855911217e-05,
+      "loss": 0.4086,
+      "step": 3550
+    },
+    {
+      "epoch": 0.7002532517026874,
+      "grad_norm": 1.009112000465393,
+      "learning_rate": 1.9113699593384086e-05,
+      "loss": 0.45,
+      "step": 3560
+    },
+    {
+      "epoch": 0.7022202552187062,
+      "grad_norm": 1.2733867168426514,
+      "learning_rate": 1.9110642330856952e-05,
+      "loss": 0.499,
+      "step": 3570
+    },
+    {
+      "epoch": 0.704187258734725,
+      "grad_norm": 1.1256855726242065,
+      "learning_rate": 1.9107585068329818e-05,
+      "loss": 0.4129,
+      "step": 3580
+    },
+    {
+      "epoch": 0.7061542622507437,
+      "grad_norm": 1.1128904819488525,
+      "learning_rate": 1.9104527805802687e-05,
+      "loss": 0.5093,
+      "step": 3590
+    },
+    {
+      "epoch": 0.7081212657667626,
+      "grad_norm": 0.9144822955131531,
+      "learning_rate": 1.9101470543275553e-05,
+      "loss": 0.3973,
+      "step": 3600
+    },
+    {
+      "epoch": 0.7100882692827813,
+      "grad_norm": 0.6767692565917969,
+      "learning_rate": 1.909841328074842e-05,
+      "loss": 0.4236,
+      "step": 3610
+    },
+    {
+      "epoch": 0.7120552727988001,
+      "grad_norm": 0.9001137614250183,
+      "learning_rate": 1.9095356018221288e-05,
+      "loss": 0.501,
+      "step": 3620
+    },
+    {
+      "epoch": 0.714022276314819,
+      "grad_norm": 0.9917561411857605,
+      "learning_rate": 1.909229875569415e-05,
+      "loss": 0.36,
+      "step": 3630
+    },
+    {
+      "epoch": 0.7159892798308377,
+      "grad_norm": 0.810536801815033,
+      "learning_rate": 1.908924149316702e-05,
+      "loss": 0.5213,
+      "step": 3640
+    },
+    {
+      "epoch": 0.7179562833468565,
+      "grad_norm": 1.5590412616729736,
+      "learning_rate": 1.9086184230639885e-05,
+      "loss": 0.6036,
+      "step": 3650
+    },
+    {
+      "epoch": 0.7199232868628753,
+      "grad_norm": 0.597394585609436,
+      "learning_rate": 1.9083126968112754e-05,
+      "loss": 0.4566,
+      "step": 3660
+    },
+    {
+      "epoch": 0.7218902903788941,
+      "grad_norm": 0.692649781703949,
+      "learning_rate": 1.908006970558562e-05,
+      "loss": 0.4622,
+      "step": 3670
+    },
+    {
+      "epoch": 0.7238572938949128,
+      "grad_norm": 0.9043506383895874,
+      "learning_rate": 1.9077012443058486e-05,
+      "loss": 0.4407,
+      "step": 3680
+    },
+    {
+      "epoch": 0.7258242974109316,
+      "grad_norm": 1.1840672492980957,
+      "learning_rate": 1.9073955180531355e-05,
+      "loss": 0.4605,
+      "step": 3690
+    },
+    {
+      "epoch": 0.7277913009269504,
+      "grad_norm": 0.9953239560127258,
+      "learning_rate": 1.907089791800422e-05,
+      "loss": 0.5095,
+      "step": 3700
+    },
+    {
+      "epoch": 0.7297583044429692,
+      "grad_norm": 0.6179831624031067,
+      "learning_rate": 1.9067840655477087e-05,
+      "loss": 0.4551,
+      "step": 3710
+    },
+    {
+      "epoch": 0.7317253079589879,
+      "grad_norm": 1.8376891613006592,
+      "learning_rate": 1.9064783392949956e-05,
+      "loss": 0.399,
+      "step": 3720
+    },
+    {
+      "epoch": 0.7336923114750068,
+      "grad_norm": 1.0583064556121826,
+      "learning_rate": 1.906172613042282e-05,
+      "loss": 0.465,
+      "step": 3730
+    },
+    {
+      "epoch": 0.7356593149910255,
+      "grad_norm": 1.0016286373138428,
+      "learning_rate": 1.9058668867895687e-05,
+      "loss": 0.5169,
+      "step": 3740
+    },
+    {
+      "epoch": 0.7376263185070443,
+      "grad_norm": 1.5237940549850464,
+      "learning_rate": 1.9055611605368553e-05,
+      "loss": 0.5857,
+      "step": 3750
+    },
+    {
+      "epoch": 0.7395933220230632,
+      "grad_norm": 1.1710821390151978,
+      "learning_rate": 1.905255434284142e-05,
+      "loss": 0.4861,
+      "step": 3760
+    },
+    {
+      "epoch": 0.7415603255390819,
+      "grad_norm": 0.7797861695289612,
+      "learning_rate": 1.9049497080314288e-05,
+      "loss": 0.5841,
+      "step": 3770
+    },
+    {
+      "epoch": 0.7435273290551007,
+      "grad_norm": 1.0958030223846436,
+      "learning_rate": 1.9046439817787154e-05,
+      "loss": 0.4966,
+      "step": 3780
+    },
+    {
+      "epoch": 0.7454943325711195,
+      "grad_norm": 0.7116280794143677,
+      "learning_rate": 1.9043382555260023e-05,
+      "loss": 0.4492,
+      "step": 3790
+    },
+    {
+      "epoch": 0.7474613360871383,
+      "grad_norm": 0.9613781571388245,
+      "learning_rate": 1.904032529273289e-05,
+      "loss": 0.5049,
+      "step": 3800
+    },
+    {
+      "epoch": 0.749428339603157,
+      "grad_norm": 0.9669978618621826,
+      "learning_rate": 1.9037268030205755e-05,
+      "loss": 0.5129,
+      "step": 3810
+    },
+    {
+      "epoch": 0.7513953431191758,
+      "grad_norm": 0.8262606263160706,
+      "learning_rate": 1.9034210767678624e-05,
+      "loss": 0.5172,
+      "step": 3820
+    },
+    {
+      "epoch": 0.7533623466351946,
+      "grad_norm": 2.477900743484497,
+      "learning_rate": 1.903115350515149e-05,
+      "loss": 0.537,
+      "step": 3830
+    },
+    {
+      "epoch": 0.7553293501512134,
+      "grad_norm": 0.6476783752441406,
+      "learning_rate": 1.9028096242624355e-05,
+      "loss": 0.5639,
+      "step": 3840
+    },
+    {
+      "epoch": 0.7572963536672321,
+      "grad_norm": 1.1109529733657837,
+      "learning_rate": 1.9025038980097224e-05,
+      "loss": 0.4391,
+      "step": 3850
+    },
+    {
+      "epoch": 0.759263357183251,
+      "grad_norm": 0.8402903079986572,
+      "learning_rate": 1.9021981717570087e-05,
+      "loss": 0.389,
+      "step": 3860
+    },
+    {
+      "epoch": 0.7612303606992697,
+      "grad_norm": 1.0447739362716675,
+      "learning_rate": 1.9018924455042956e-05,
+      "loss": 0.4631,
+      "step": 3870
+    },
+    {
+      "epoch": 0.7631973642152885,
+      "grad_norm": 1.4362361431121826,
+      "learning_rate": 1.9015867192515822e-05,
+      "loss": 0.4975,
+      "step": 3880
+    },
+    {
+      "epoch": 0.7651643677313074,
+      "grad_norm": 1.702034831047058,
+      "learning_rate": 1.9012809929988688e-05,
+      "loss": 0.5585,
+      "step": 3890
+    },
+    {
+      "epoch": 0.7671313712473261,
+      "grad_norm": 1.469128966331482,
+      "learning_rate": 1.9009752667461557e-05,
+      "loss": 0.5571,
+      "step": 3900
+    },
+    {
+      "epoch": 0.7690983747633449,
+      "grad_norm": 0.793484628200531,
+      "learning_rate": 1.9006695404934423e-05,
+      "loss": 0.4067,
+      "step": 3910
+    },
+    {
+      "epoch": 0.7710653782793637,
+      "grad_norm": 0.775711715221405,
+      "learning_rate": 1.900363814240729e-05,
+      "loss": 0.5249,
+      "step": 3920
+    },
+    {
+      "epoch": 0.7730323817953825,
+      "grad_norm": 0.528615415096283,
+      "learning_rate": 1.9000580879880157e-05,
+      "loss": 0.3714,
+      "step": 3930
+    },
+    {
+      "epoch": 0.7749993853114012,
+      "grad_norm": 0.9931712746620178,
+      "learning_rate": 1.8997523617353023e-05,
+      "loss": 0.4286,
+      "step": 3940
+    },
+    {
+      "epoch": 0.77696638882742,
+      "grad_norm": 1.9543815851211548,
+      "learning_rate": 1.8994466354825892e-05,
+      "loss": 0.4738,
+      "step": 3950
+    },
+    {
+      "epoch": 0.7789333923434388,
+      "grad_norm": 1.5863524675369263,
+      "learning_rate": 1.8991409092298758e-05,
+      "loss": 0.518,
+      "step": 3960
+    },
+    {
+      "epoch": 0.7809003958594576,
+      "grad_norm": 1.0554157495498657,
+      "learning_rate": 1.8988351829771624e-05,
+      "loss": 0.5735,
+      "step": 3970
+    },
+    {
+      "epoch": 0.7828673993754763,
+      "grad_norm": 1.5502032041549683,
+      "learning_rate": 1.898529456724449e-05,
+      "loss": 0.4925,
+      "step": 3980
+    },
+    {
+      "epoch": 0.7848344028914952,
+      "grad_norm": 0.9096987843513489,
+      "learning_rate": 1.8982237304717356e-05,
+      "loss": 0.2723,
+      "step": 3990
+    },
+    {
+      "epoch": 0.786801406407514,
+      "grad_norm": 1.4501546621322632,
+      "learning_rate": 1.8979180042190225e-05,
+      "loss": 0.4634,
+      "step": 4000
+    },
+    {
+      "epoch": 0.786801406407514,
+      "eval_loss": 0.2581852972507477,
+      "eval_runtime": 8.8766,
+      "eval_samples_per_second": 5.633,
+      "eval_steps_per_second": 2.816,
+      "step": 4000
+    },
+    {
+      "epoch": 0.7887684099235327,
+      "grad_norm": 0.8492615818977356,
+      "learning_rate": 1.897612277966309e-05,
+      "loss": 0.4806,
+      "step": 4010
+    },
+    {
+      "epoch": 0.7907354134395516,
+      "grad_norm": 1.0982277393341064,
+      "learning_rate": 1.8973065517135956e-05,
+      "loss": 0.481,
+      "step": 4020
+    },
+    {
+      "epoch": 0.7927024169555703,
+      "grad_norm": 1.1932698488235474,
+      "learning_rate": 1.8970008254608825e-05,
+      "loss": 0.4803,
+      "step": 4030
+    },
+    {
+      "epoch": 0.7946694204715891,
+      "grad_norm": 0.9135488867759705,
+      "learning_rate": 1.896695099208169e-05,
+      "loss": 0.5505,
+      "step": 4040
+    },
+    {
+      "epoch": 0.7966364239876079,
+      "grad_norm": 1.5870565176010132,
+      "learning_rate": 1.896389372955456e-05,
+      "loss": 0.4184,
+      "step": 4050
+    },
+    {
+      "epoch": 0.7986034275036267,
+      "grad_norm": 0.9987393021583557,
+      "learning_rate": 1.8960836467027426e-05,
+      "loss": 0.5769,
+      "step": 4060
+    },
+    {
+      "epoch": 0.8005704310196454,
+      "grad_norm": 1.0902693271636963,
+      "learning_rate": 1.8957779204500292e-05,
+      "loss": 0.4934,
+      "step": 4070
+    },
+    {
+      "epoch": 0.8025374345356643,
+      "grad_norm": 1.2201869487762451,
+      "learning_rate": 1.895472194197316e-05,
+      "loss": 0.4007,
+      "step": 4080
+    },
+    {
+      "epoch": 0.804504438051683,
+      "grad_norm": 1.425352692604065,
+      "learning_rate": 1.8951664679446023e-05,
+      "loss": 0.6055,
+      "step": 4090
+    },
+    {
+      "epoch": 0.8064714415677018,
+      "grad_norm": 1.107489824295044,
+      "learning_rate": 1.8948607416918893e-05,
+      "loss": 0.5402,
+      "step": 4100
+    },
+    {
+      "epoch": 0.8084384450837205,
+      "grad_norm": 0.8518027067184448,
+      "learning_rate": 1.894555015439176e-05,
+      "loss": 0.4802,
+      "step": 4110
+    },
+    {
+      "epoch": 0.8104054485997394,
+      "grad_norm": 0.9588760137557983,
+      "learning_rate": 1.8942492891864624e-05,
+      "loss": 0.4809,
+      "step": 4120
+    },
+    {
+      "epoch": 0.8123724521157581,
+      "grad_norm": 1.3297154903411865,
+      "learning_rate": 1.8939435629337493e-05,
+      "loss": 0.4471,
+      "step": 4130
+    },
+    {
+      "epoch": 0.8143394556317769,
+      "grad_norm": 0.9255673885345459,
+      "learning_rate": 1.893637836681036e-05,
+      "loss": 0.4714,
+      "step": 4140
+    },
+    {
+      "epoch": 0.8163064591477958,
+      "grad_norm": 1.3429300785064697,
+      "learning_rate": 1.8933321104283225e-05,
+      "loss": 0.5203,
+      "step": 4150
+    },
+    {
+      "epoch": 0.8182734626638145,
+      "grad_norm": 1.7732727527618408,
+      "learning_rate": 1.8930263841756094e-05,
+      "loss": 0.5918,
+      "step": 4160
+    },
+    {
+      "epoch": 0.8202404661798333,
+      "grad_norm": 1.1229453086853027,
+      "learning_rate": 1.892720657922896e-05,
+      "loss": 0.449,
+      "step": 4170
+    },
+    {
+      "epoch": 0.8222074696958521,
+      "grad_norm": 1.2160098552703857,
+      "learning_rate": 1.892414931670183e-05,
+      "loss": 0.4994,
+      "step": 4180
+    },
+    {
+      "epoch": 0.8241744732118709,
+      "grad_norm": 1.6124922037124634,
+      "learning_rate": 1.8921092054174695e-05,
+      "loss": 0.4356,
+      "step": 4190
+    },
+    {
+      "epoch": 0.8261414767278896,
+      "grad_norm": 1.0741393566131592,
+      "learning_rate": 1.891803479164756e-05,
+      "loss": 0.427,
+      "step": 4200
+    },
+    {
+      "epoch": 0.8281084802439085,
+      "grad_norm": 1.2140878438949585,
+      "learning_rate": 1.8914977529120426e-05,
+      "loss": 0.5503,
+      "step": 4210
+    },
+    {
+      "epoch": 0.8300754837599272,
+      "grad_norm": 1.016489028930664,
+      "learning_rate": 1.8911920266593292e-05,
+      "loss": 0.5485,
+      "step": 4220
+    },
+    {
+      "epoch": 0.832042487275946,
+      "grad_norm": 1.1552870273590088,
+      "learning_rate": 1.890886300406616e-05,
+      "loss": 0.4611,
+      "step": 4230
+    },
+    {
+      "epoch": 0.8340094907919647,
+      "grad_norm": 0.9830273985862732,
+      "learning_rate": 1.8905805741539027e-05,
+      "loss": 0.4794,
+      "step": 4240
+    },
+    {
+      "epoch": 0.8359764943079836,
+      "grad_norm": 0.9099911451339722,
+      "learning_rate": 1.8902748479011893e-05,
+      "loss": 0.4479,
+      "step": 4250
+    },
+    {
+      "epoch": 0.8379434978240023,
+      "grad_norm": 1.2491271495819092,
+      "learning_rate": 1.8899691216484762e-05,
+      "loss": 0.5845,
+      "step": 4260
+    },
+    {
+      "epoch": 0.8399105013400211,
+      "grad_norm": 1.144546627998352,
+      "learning_rate": 1.8896633953957628e-05,
+      "loss": 0.4979,
+      "step": 4270
+    },
+    {
+      "epoch": 0.84187750485604,
+      "grad_norm": 0.45869743824005127,
+      "learning_rate": 1.8893576691430494e-05,
+      "loss": 0.4119,
+      "step": 4280
+    },
+    {
+      "epoch": 0.8438445083720587,
+      "grad_norm": 1.6647731065750122,
+      "learning_rate": 1.8890519428903363e-05,
+      "loss": 0.5917,
+      "step": 4290
+    },
+    {
+      "epoch": 0.8458115118880775,
+      "grad_norm": 0.6388562321662903,
+      "learning_rate": 1.888746216637623e-05,
+      "loss": 0.534,
+      "step": 4300
+    },
+    {
+      "epoch": 0.8477785154040963,
+      "grad_norm": 1.499766230583191,
+      "learning_rate": 1.8884404903849094e-05,
+      "loss": 0.6057,
+      "step": 4310
+    },
+    {
+      "epoch": 0.8497455189201151,
+      "grad_norm": 1.1714287996292114,
+      "learning_rate": 1.888134764132196e-05,
+      "loss": 0.4296,
+      "step": 4320
+    },
+    {
+      "epoch": 0.8517125224361338,
+      "grad_norm": 0.9272406697273254,
+      "learning_rate": 1.887829037879483e-05,
+      "loss": 0.4614,
+      "step": 4330
+    },
+    {
+      "epoch": 0.8536795259521527,
+      "grad_norm": 0.9848095774650574,
+      "learning_rate": 1.8875233116267695e-05,
+      "loss": 0.4652,
+      "step": 4340
+    },
+    {
+      "epoch": 0.8556465294681714,
+      "grad_norm": 1.2984775304794312,
+      "learning_rate": 1.887217585374056e-05,
+      "loss": 0.4715,
+      "step": 4350
+    },
+    {
+      "epoch": 0.8576135329841902,
+      "grad_norm": 0.8881934285163879,
+      "learning_rate": 1.886911859121343e-05,
+      "loss": 0.3736,
+      "step": 4360
+    },
+    {
+      "epoch": 0.859580536500209,
+      "grad_norm": 1.8501014709472656,
+      "learning_rate": 1.8866061328686296e-05,
+      "loss": 0.4502,
+      "step": 4370
+    },
+    {
+      "epoch": 0.8615475400162278,
+      "grad_norm": 1.3918039798736572,
+      "learning_rate": 1.886300406615916e-05,
+      "loss": 0.5603,
+      "step": 4380
+    },
+    {
+      "epoch": 0.8635145435322465,
+      "grad_norm": 1.2267512083053589,
+      "learning_rate": 1.885994680363203e-05,
+      "loss": 0.4656,
+      "step": 4390
+    },
+    {
+      "epoch": 0.8654815470482653,
+      "grad_norm": 1.0257207155227661,
+      "learning_rate": 1.8856889541104896e-05,
+      "loss": 0.3862,
+      "step": 4400
+    },
+    {
+      "epoch": 0.8674485505642842,
+      "grad_norm": 0.5115770101547241,
+      "learning_rate": 1.8853832278577762e-05,
+      "loss": 0.4928,
+      "step": 4410
+    },
+    {
+      "epoch": 0.8694155540803029,
+      "grad_norm": 0.8817374110221863,
+      "learning_rate": 1.885077501605063e-05,
+      "loss": 0.4431,
+      "step": 4420
+    },
+    {
+      "epoch": 0.8713825575963217,
+      "grad_norm": 1.37067449092865,
+      "learning_rate": 1.8847717753523497e-05,
+      "loss": 0.5065,
+      "step": 4430
+    },
+    {
+      "epoch": 0.8733495611123405,
+      "grad_norm": 0.4619062840938568,
+      "learning_rate": 1.8844660490996363e-05,
+      "loss": 0.4052,
+      "step": 4440
+    },
+    {
+      "epoch": 0.8753165646283593,
+      "grad_norm": 0.8775585293769836,
+      "learning_rate": 1.884160322846923e-05,
+      "loss": 0.4538,
+      "step": 4450
+    },
+    {
+      "epoch": 0.877283568144378,
+      "grad_norm": 0.6405948996543884,
+      "learning_rate": 1.8838545965942098e-05,
+      "loss": 0.4881,
+      "step": 4460
+    },
+    {
+      "epoch": 0.8792505716603969,
+      "grad_norm": 1.11896812915802,
+      "learning_rate": 1.8835488703414964e-05,
+      "loss": 0.3966,
+      "step": 4470
+    },
+    {
+      "epoch": 0.8812175751764156,
+      "grad_norm": 1.493742823600769,
+      "learning_rate": 1.883243144088783e-05,
+      "loss": 0.4512,
+      "step": 4480
+    },
+    {
+      "epoch": 0.8831845786924344,
+      "grad_norm": 1.34097158908844,
+      "learning_rate": 1.88293741783607e-05,
+      "loss": 0.3983,
+      "step": 4490
+    },
+    {
+      "epoch": 0.8851515822084532,
+      "grad_norm": 1.2133510112762451,
+      "learning_rate": 1.8826316915833564e-05,
+      "loss": 0.4512,
+      "step": 4500
+    },
+    {
+      "epoch": 0.8851515822084532,
+      "eval_loss": 0.25500819087028503,
+      "eval_runtime": 8.9175,
+      "eval_samples_per_second": 5.607,
+      "eval_steps_per_second": 2.803,
+      "step": 4500
+    },
+    {
+      "epoch": 0.887118585724472,
+      "grad_norm": 0.8433915376663208,
+      "learning_rate": 1.882325965330643e-05,
+      "loss": 0.4379,
+      "step": 4510
+    },
+    {
+      "epoch": 0.8890855892404907,
+      "grad_norm": 1.048805832862854,
+      "learning_rate": 1.88202023907793e-05,
+      "loss": 0.3957,
+      "step": 4520
+    },
+    {
+      "epoch": 0.8910525927565095,
+      "grad_norm": 1.5707403421401978,
+      "learning_rate": 1.8817145128252165e-05,
+      "loss": 0.4594,
+      "step": 4530
+    },
+    {
+      "epoch": 0.8930195962725284,
+      "grad_norm": 1.2741488218307495,
+      "learning_rate": 1.881408786572503e-05,
+      "loss": 0.5692,
+      "step": 4540
+    },
+    {
+      "epoch": 0.8949865997885471,
+      "grad_norm": 0.6873851418495178,
+      "learning_rate": 1.8811030603197897e-05,
+      "loss": 0.3774,
+      "step": 4550
+    },
+    {
+      "epoch": 0.8969536033045659,
+      "grad_norm": 0.9358246922492981,
+      "learning_rate": 1.8807973340670766e-05,
+      "loss": 0.5593,
+      "step": 4560
+    },
+    {
+      "epoch": 0.8989206068205847,
+      "grad_norm": 1.3726028203964233,
+      "learning_rate": 1.880491607814363e-05,
+      "loss": 0.4497,
+      "step": 4570
+    },
+    {
+      "epoch": 0.9008876103366035,
+      "grad_norm": 0.9805541634559631,
+      "learning_rate": 1.8801858815616497e-05,
+      "loss": 0.3744,
+      "step": 4580
+    },
+    {
+      "epoch": 0.9028546138526222,
+      "grad_norm": 1.2838389873504639,
+      "learning_rate": 1.8798801553089367e-05,
+      "loss": 0.4415,
+      "step": 4590
+    },
+    {
+      "epoch": 0.9048216173686411,
+      "grad_norm": 0.782386064529419,
+      "learning_rate": 1.8795744290562232e-05,
+      "loss": 0.4138,
+      "step": 4600
+    },
+    {
+      "epoch": 0.9067886208846598,
+      "grad_norm": 1.1255232095718384,
+      "learning_rate": 1.8792687028035098e-05,
+      "loss": 0.5175,
+      "step": 4610
+    },
+    {
+      "epoch": 0.9087556244006786,
+      "grad_norm": 1.0745489597320557,
+      "learning_rate": 1.8789629765507967e-05,
+      "loss": 0.432,
+      "step": 4620
+    },
+    {
+      "epoch": 0.9107226279166974,
+      "grad_norm": 0.6713242530822754,
+      "learning_rate": 1.8786572502980833e-05,
+      "loss": 0.5976,
+      "step": 4630
+    },
+    {
+      "epoch": 0.9126896314327162,
+      "grad_norm": 1.3857566118240356,
+      "learning_rate": 1.87835152404537e-05,
+      "loss": 0.4972,
+      "step": 4640
+    },
+    {
+      "epoch": 0.914656634948735,
+      "grad_norm": 1.6660428047180176,
+      "learning_rate": 1.8780457977926565e-05,
+      "loss": 0.456,
+      "step": 4650
+    },
+    {
+      "epoch": 0.9166236384647538,
+      "grad_norm": 1.1957042217254639,
+      "learning_rate": 1.877740071539943e-05,
+      "loss": 0.5614,
+      "step": 4660
+    },
+    {
+      "epoch": 0.9185906419807726,
+      "grad_norm": 1.4520841836929321,
+      "learning_rate": 1.87743434528723e-05,
+      "loss": 0.5019,
+      "step": 4670
+    },
+    {
+      "epoch": 0.9205576454967913,
+      "grad_norm": 1.6292874813079834,
+      "learning_rate": 1.8771286190345165e-05,
+      "loss": 0.4422,
+      "step": 4680
+    },
+    {
+      "epoch": 0.9225246490128101,
+      "grad_norm": 1.2969595193862915,
+      "learning_rate": 1.8768228927818035e-05,
+      "loss": 0.4567,
+      "step": 4690
+    },
+    {
+      "epoch": 0.9244916525288289,
+      "grad_norm": 0.8778219223022461,
+      "learning_rate": 1.87651716652909e-05,
+      "loss": 0.6256,
+      "step": 4700
+    },
+    {
+      "epoch": 0.9264586560448477,
+      "grad_norm": 0.725437343120575,
+      "learning_rate": 1.8762114402763766e-05,
+      "loss": 0.5737,
+      "step": 4710
+    },
+    {
+      "epoch": 0.9284256595608664,
+      "grad_norm": 0.9920393228530884,
+      "learning_rate": 1.8759057140236635e-05,
+      "loss": 0.4377,
+      "step": 4720
+    },
+    {
+      "epoch": 0.9303926630768853,
+      "grad_norm": 1.587246298789978,
+      "learning_rate": 1.87559998777095e-05,
+      "loss": 0.4066,
+      "step": 4730
+    },
+    {
+      "epoch": 0.932359666592904,
+      "grad_norm": 1.5838128328323364,
+      "learning_rate": 1.8752942615182367e-05,
+      "loss": 0.4921,
+      "step": 4740
+    },
+    {
+      "epoch": 0.9343266701089228,
+      "grad_norm": 2.2074408531188965,
+      "learning_rate": 1.8749885352655236e-05,
+      "loss": 0.4415,
+      "step": 4750
+    },
+    {
+      "epoch": 0.9362936736249416,
+      "grad_norm": 0.7175712585449219,
+      "learning_rate": 1.8746828090128102e-05,
+      "loss": 0.5967,
+      "step": 4760
+    },
+    {
+      "epoch": 0.9382606771409604,
+      "grad_norm": 0.672035276889801,
+      "learning_rate": 1.8743770827600968e-05,
+      "loss": 0.4765,
+      "step": 4770
+    },
+    {
+      "epoch": 0.9402276806569791,
+      "grad_norm": 1.2136248350143433,
+      "learning_rate": 1.8740713565073833e-05,
+      "loss": 0.4918,
+      "step": 4780
+    },
+    {
+      "epoch": 0.942194684172998,
+      "grad_norm": 1.1846280097961426,
+      "learning_rate": 1.87376563025467e-05,
+      "loss": 0.5204,
+      "step": 4790
+    },
+    {
+      "epoch": 0.9441616876890168,
+      "grad_norm": 0.9919416904449463,
+      "learning_rate": 1.8734599040019568e-05,
+      "loss": 0.5251,
+      "step": 4800
+    },
+    {
+      "epoch": 0.9461286912050355,
+      "grad_norm": 0.9183461666107178,
+      "learning_rate": 1.8731541777492434e-05,
+      "loss": 0.3683,
+      "step": 4810
+    },
+    {
+      "epoch": 0.9480956947210543,
+      "grad_norm": 1.5398882627487183,
+      "learning_rate": 1.8728484514965303e-05,
+      "loss": 0.4656,
+      "step": 4820
+    },
+    {
+      "epoch": 0.9500626982370731,
+      "grad_norm": 2.7431869506835938,
+      "learning_rate": 1.872542725243817e-05,
+      "loss": 0.4459,
+      "step": 4830
+    },
+    {
+      "epoch": 0.9520297017530919,
+      "grad_norm": 1.4181733131408691,
+      "learning_rate": 1.8722369989911035e-05,
+      "loss": 0.4347,
+      "step": 4840
+    },
+    {
+      "epoch": 0.9539967052691106,
+      "grad_norm": 1.1322598457336426,
+      "learning_rate": 1.8719312727383904e-05,
+      "loss": 0.4761,
+      "step": 4850
+    },
+    {
+      "epoch": 0.9559637087851295,
+      "grad_norm": 1.0552661418914795,
+      "learning_rate": 1.871625546485677e-05,
+      "loss": 0.4593,
+      "step": 4860
+    },
+    {
+      "epoch": 0.9579307123011482,
+      "grad_norm": 1.7677165269851685,
+      "learning_rate": 1.8713198202329635e-05,
+      "loss": 0.516,
+      "step": 4870
+    },
+    {
+      "epoch": 0.959897715817167,
+      "grad_norm": 1.1124646663665771,
+      "learning_rate": 1.87101409398025e-05,
+      "loss": 0.4594,
+      "step": 4880
+    },
+    {
+      "epoch": 0.9618647193331858,
+      "grad_norm": 0.9659914970397949,
+      "learning_rate": 1.8707083677275367e-05,
+      "loss": 0.4844,
+      "step": 4890
+    },
+    {
+      "epoch": 0.9638317228492046,
+      "grad_norm": 0.9251281023025513,
+      "learning_rate": 1.8704026414748236e-05,
+      "loss": 0.5587,
+      "step": 4900
+    },
+    {
+      "epoch": 0.9657987263652233,
+      "grad_norm": 1.2297642230987549,
+      "learning_rate": 1.8700969152221102e-05,
+      "loss": 0.5739,
+      "step": 4910
+    },
+    {
+      "epoch": 0.9677657298812422,
+      "grad_norm": 1.329194188117981,
+      "learning_rate": 1.8697911889693968e-05,
+      "loss": 0.5016,
+      "step": 4920
+    },
+    {
+      "epoch": 0.969732733397261,
+      "grad_norm": 0.8818217515945435,
+      "learning_rate": 1.8694854627166837e-05,
+      "loss": 0.4428,
+      "step": 4930
+    },
+    {
+      "epoch": 0.9716997369132797,
+      "grad_norm": 1.1140741109848022,
+      "learning_rate": 1.8691797364639703e-05,
+      "loss": 0.5713,
+      "step": 4940
+    },
+    {
+      "epoch": 0.9736667404292986,
+      "grad_norm": 0.5966857075691223,
+      "learning_rate": 1.8688740102112572e-05,
+      "loss": 0.4876,
+      "step": 4950
+    },
+    {
+      "epoch": 0.9756337439453173,
+      "grad_norm": 0.8906331062316895,
+      "learning_rate": 1.8685682839585438e-05,
+      "loss": 0.4157,
+      "step": 4960
+    },
+    {
+      "epoch": 0.9776007474613361,
+      "grad_norm": 1.6256823539733887,
+      "learning_rate": 1.8682625577058303e-05,
+      "loss": 0.4524,
+      "step": 4970
+    },
+    {
+      "epoch": 0.9795677509773548,
+      "grad_norm": 1.4493746757507324,
+      "learning_rate": 1.8679568314531173e-05,
+      "loss": 0.5159,
+      "step": 4980
+    },
+    {
+      "epoch": 0.9815347544933737,
+      "grad_norm": 0.9830152988433838,
+      "learning_rate": 1.8676511052004035e-05,
+      "loss": 0.3848,
+      "step": 4990
+    },
+    {
+      "epoch": 0.9835017580093924,
+      "grad_norm": 1.2543774843215942,
+      "learning_rate": 1.8673453789476904e-05,
+      "loss": 0.5435,
+      "step": 5000
+    },
+    {
+      "epoch": 0.9835017580093924,
+      "eval_loss": 0.2498249113559723,
+      "eval_runtime": 8.8851,
+      "eval_samples_per_second": 5.627,
+      "eval_steps_per_second": 2.814,
+      "step": 5000
+    },
+    {
+      "epoch": 0.9854687615254112,
+      "grad_norm": 1.0389529466629028,
+      "learning_rate": 1.867039652694977e-05,
+      "loss": 0.4089,
+      "step": 5010
+    },
+    {
+      "epoch": 0.98743576504143,
+      "grad_norm": 0.9256637692451477,
+      "learning_rate": 1.8667339264422636e-05,
+      "loss": 0.5185,
+      "step": 5020
+    },
+    {
+      "epoch": 0.9894027685574488,
+      "grad_norm": 1.2138028144836426,
+      "learning_rate": 1.8664282001895505e-05,
+      "loss": 0.5416,
+      "step": 5030
+    },
+    {
+      "epoch": 0.9913697720734675,
+      "grad_norm": 1.4770766496658325,
+      "learning_rate": 1.866122473936837e-05,
+      "loss": 0.4611,
+      "step": 5040
+    },
+    {
+      "epoch": 0.9933367755894864,
+      "grad_norm": 0.783585250377655,
+      "learning_rate": 1.8658167476841236e-05,
+      "loss": 0.4212,
+      "step": 5050
+    },
+    {
+      "epoch": 0.9953037791055052,
+      "grad_norm": 1.2547729015350342,
+      "learning_rate": 1.8655110214314106e-05,
+      "loss": 0.5168,
+      "step": 5060
+    },
+    {
+      "epoch": 0.9972707826215239,
+      "grad_norm": 0.9369317293167114,
+      "learning_rate": 1.865205295178697e-05,
+      "loss": 0.5653,
+      "step": 5070
+    },
+    {
+      "epoch": 0.9992377861375428,
+      "grad_norm": 1.1755496263504028,
+      "learning_rate": 1.864899568925984e-05,
+      "loss": 0.5023,
+      "step": 5080
+    },
+    {
+      "epoch": 1.0012047896535614,
+      "grad_norm": 0.9298199415206909,
+      "learning_rate": 1.8645938426732706e-05,
+      "loss": 0.5254,
+      "step": 5090
+    },
+    {
+      "epoch": 1.0031717931695803,
+      "grad_norm": 1.0964889526367188,
+      "learning_rate": 1.8642881164205572e-05,
+      "loss": 0.367,
+      "step": 5100
+    },
+    {
+      "epoch": 1.0051387966855991,
+      "grad_norm": 1.8859641551971436,
+      "learning_rate": 1.8639823901678438e-05,
+      "loss": 0.4357,
+      "step": 5110
+    },
+    {
+      "epoch": 1.0071058002016178,
+      "grad_norm": 1.2474465370178223,
+      "learning_rate": 1.8636766639151304e-05,
+      "loss": 0.4071,
+      "step": 5120
+    },
+    {
+      "epoch": 1.0090728037176366,
+      "grad_norm": 0.820743203163147,
+      "learning_rate": 1.8633709376624173e-05,
+      "loss": 0.4397,
+      "step": 5130
+    },
+    {
+      "epoch": 1.0110398072336555,
+      "grad_norm": 1.2931946516036987,
+      "learning_rate": 1.863065211409704e-05,
+      "loss": 0.3876,
+      "step": 5140
+    },
+    {
+      "epoch": 1.0130068107496741,
+      "grad_norm": 0.6955686211585999,
+      "learning_rate": 1.8627594851569904e-05,
+      "loss": 0.408,
+      "step": 5150
+    },
+    {
+      "epoch": 1.014973814265693,
+      "grad_norm": 0.8109415769577026,
+      "learning_rate": 1.8624537589042773e-05,
+      "loss": 0.4783,
+      "step": 5160
+    },
+    {
+      "epoch": 1.0169408177817119,
+      "grad_norm": 0.7668949961662292,
+      "learning_rate": 1.862148032651564e-05,
+      "loss": 0.5177,
+      "step": 5170
+    },
+    {
+      "epoch": 1.0189078212977305,
+      "grad_norm": 1.3539754152297974,
+      "learning_rate": 1.8618423063988505e-05,
+      "loss": 0.4552,
+      "step": 5180
+    },
+    {
+      "epoch": 1.0208748248137494,
+      "grad_norm": 1.7623764276504517,
+      "learning_rate": 1.8615365801461374e-05,
+      "loss": 0.4712,
+      "step": 5190
+    },
+    {
+      "epoch": 1.0228418283297682,
+      "grad_norm": 1.3476308584213257,
+      "learning_rate": 1.861230853893424e-05,
+      "loss": 0.3688,
+      "step": 5200
+    },
+    {
+      "epoch": 1.0248088318457869,
+      "grad_norm": 0.7044321298599243,
+      "learning_rate": 1.860925127640711e-05,
+      "loss": 0.5159,
+      "step": 5210
+    },
+    {
+      "epoch": 1.0267758353618057,
+      "grad_norm": 1.0632188320159912,
+      "learning_rate": 1.860619401387997e-05,
+      "loss": 0.4336,
+      "step": 5220
+    },
+    {
+      "epoch": 1.0287428388778246,
+      "grad_norm": 1.7851991653442383,
+      "learning_rate": 1.860313675135284e-05,
+      "loss": 0.4671,
+      "step": 5230
+    },
+    {
+      "epoch": 1.0307098423938432,
+      "grad_norm": 0.7615410685539246,
+      "learning_rate": 1.8600079488825706e-05,
+      "loss": 0.4408,
+      "step": 5240
+    },
+    {
+      "epoch": 1.032676845909862,
+      "grad_norm": 1.042166829109192,
+      "learning_rate": 1.8597022226298572e-05,
+      "loss": 0.3965,
+      "step": 5250
+    },
+    {
+      "epoch": 1.034643849425881,
+      "grad_norm": 0.7487517595291138,
+      "learning_rate": 1.859396496377144e-05,
+      "loss": 0.3477,
+      "step": 5260
+    },
+    {
+      "epoch": 1.0366108529418996,
+      "grad_norm": 0.9106249213218689,
+      "learning_rate": 1.8590907701244307e-05,
+      "loss": 0.601,
+      "step": 5270
+    },
+    {
+      "epoch": 1.0385778564579184,
+      "grad_norm": 1.1695805788040161,
+      "learning_rate": 1.8587850438717173e-05,
+      "loss": 0.4192,
+      "step": 5280
+    },
+    {
+      "epoch": 1.0405448599739373,
+      "grad_norm": 0.398007333278656,
+      "learning_rate": 1.8584793176190042e-05,
+      "loss": 0.4124,
+      "step": 5290
+    },
+    {
+      "epoch": 1.042511863489956,
+      "grad_norm": 0.6286818981170654,
+      "learning_rate": 1.8581735913662908e-05,
+      "loss": 0.4796,
+      "step": 5300
+    },
+    {
+      "epoch": 1.0444788670059748,
+      "grad_norm": 1.6384111642837524,
+      "learning_rate": 1.8578678651135774e-05,
+      "loss": 0.4816,
+      "step": 5310
+    },
+    {
+      "epoch": 1.0464458705219934,
+      "grad_norm": 0.9090391397476196,
+      "learning_rate": 1.8575621388608643e-05,
+      "loss": 0.4333,
+      "step": 5320
+    },
+    {
+      "epoch": 1.0484128740380123,
+      "grad_norm": 1.0161601305007935,
+      "learning_rate": 1.857256412608151e-05,
+      "loss": 0.3375,
+      "step": 5330
+    },
+    {
+      "epoch": 1.0503798775540312,
+      "grad_norm": 1.1942650079727173,
+      "learning_rate": 1.8569506863554374e-05,
+      "loss": 0.411,
+      "step": 5340
+    },
+    {
+      "epoch": 1.0523468810700498,
+      "grad_norm": 1.3584073781967163,
+      "learning_rate": 1.856644960102724e-05,
+      "loss": 0.4614,
+      "step": 5350
+    },
+    {
+      "epoch": 1.0543138845860687,
+      "grad_norm": 1.0407729148864746,
+      "learning_rate": 1.856339233850011e-05,
+      "loss": 0.578,
+      "step": 5360
+    },
+    {
+      "epoch": 1.0562808881020875,
+      "grad_norm": 1.0869169235229492,
+      "learning_rate": 1.8560335075972975e-05,
+      "loss": 0.3789,
+      "step": 5370
+    },
+    {
+      "epoch": 1.0582478916181062,
+      "grad_norm": 0.9861720204353333,
+      "learning_rate": 1.855727781344584e-05,
+      "loss": 0.5299,
+      "step": 5380
+    },
+    {
+      "epoch": 1.060214895134125,
+      "grad_norm": 0.9632128477096558,
+      "learning_rate": 1.855422055091871e-05,
+      "loss": 0.4665,
+      "step": 5390
+    },
+    {
+      "epoch": 1.062181898650144,
+      "grad_norm": 1.5414925813674927,
+      "learning_rate": 1.8551163288391576e-05,
+      "loss": 0.4346,
+      "step": 5400
+    },
+    {
+      "epoch": 1.0641489021661625,
+      "grad_norm": 2.2143287658691406,
+      "learning_rate": 1.854810602586444e-05,
+      "loss": 0.3353,
+      "step": 5410
+    },
+    {
+      "epoch": 1.0661159056821814,
+      "grad_norm": 1.193596363067627,
+      "learning_rate": 1.854504876333731e-05,
+      "loss": 0.3978,
+      "step": 5420
+    },
+    {
+      "epoch": 1.0680829091982003,
+      "grad_norm": 1.4474225044250488,
+      "learning_rate": 1.8541991500810177e-05,
+      "loss": 0.3549,
+      "step": 5430
+    },
+    {
+      "epoch": 1.070049912714219,
+      "grad_norm": 0.8442851305007935,
+      "learning_rate": 1.8538934238283042e-05,
+      "loss": 0.4715,
+      "step": 5440
+    },
+    {
+      "epoch": 1.0720169162302378,
+      "grad_norm": 1.0033780336380005,
+      "learning_rate": 1.8535876975755908e-05,
+      "loss": 0.6038,
+      "step": 5450
+    },
+    {
+      "epoch": 1.0739839197462566,
+      "grad_norm": 0.9648568630218506,
+      "learning_rate": 1.8532819713228777e-05,
+      "loss": 0.385,
+      "step": 5460
+    },
+    {
+      "epoch": 1.0759509232622753,
+      "grad_norm": 1.1449640989303589,
+      "learning_rate": 1.8529762450701643e-05,
+      "loss": 0.4313,
+      "step": 5470
+    },
+    {
+      "epoch": 1.0779179267782941,
+      "grad_norm": 1.7114028930664062,
+      "learning_rate": 1.852670518817451e-05,
+      "loss": 0.4804,
+      "step": 5480
+    },
+    {
+      "epoch": 1.079884930294313,
+      "grad_norm": 1.1060335636138916,
+      "learning_rate": 1.8523647925647378e-05,
+      "loss": 0.3977,
+      "step": 5490
+    },
+    {
+      "epoch": 1.0818519338103316,
+      "grad_norm": 1.1767979860305786,
+      "learning_rate": 1.8520590663120244e-05,
+      "loss": 0.4773,
+      "step": 5500
+    },
+    {
+      "epoch": 1.0818519338103316,
+      "eval_loss": 0.25869181752204895,
+      "eval_runtime": 8.8591,
+      "eval_samples_per_second": 5.644,
+      "eval_steps_per_second": 2.822,
+      "step": 5500
+    },
+    {
+      "epoch": 1.0838189373263505,
+      "grad_norm": 1.1900354623794556,
+      "learning_rate": 1.851753340059311e-05,
+      "loss": 0.4315,
+      "step": 5510
+    },
+    {
+      "epoch": 1.0857859408423693,
+      "grad_norm": 1.4605563879013062,
+      "learning_rate": 1.851447613806598e-05,
+      "loss": 0.3617,
+      "step": 5520
+    },
+    {
+      "epoch": 1.087752944358388,
+      "grad_norm": 0.6865226030349731,
+      "learning_rate": 1.8511418875538845e-05,
+      "loss": 0.45,
+      "step": 5530
+    },
+    {
+      "epoch": 1.0897199478744068,
+      "grad_norm": 1.535517930984497,
+      "learning_rate": 1.850836161301171e-05,
+      "loss": 0.4856,
+      "step": 5540
+    },
+    {
+      "epoch": 1.0916869513904257,
+      "grad_norm": 1.234263300895691,
+      "learning_rate": 1.850530435048458e-05,
+      "loss": 0.4462,
+      "step": 5550
+    },
+    {
+      "epoch": 1.0936539549064443,
+      "grad_norm": 0.8305632472038269,
+      "learning_rate": 1.8502247087957442e-05,
+      "loss": 0.3267,
+      "step": 5560
+    },
+    {
+      "epoch": 1.0956209584224632,
+      "grad_norm": 0.8066553473472595,
+      "learning_rate": 1.849918982543031e-05,
+      "loss": 0.5352,
+      "step": 5570
+    },
+    {
+      "epoch": 1.097587961938482,
+      "grad_norm": 1.5289993286132812,
+      "learning_rate": 1.8496132562903177e-05,
+      "loss": 0.4365,
+      "step": 5580
+    },
+    {
+      "epoch": 1.0995549654545007,
+      "grad_norm": 1.027649164199829,
+      "learning_rate": 1.8493075300376046e-05,
+      "loss": 0.4864,
+      "step": 5590
+    },
+    {
+      "epoch": 1.1015219689705196,
+      "grad_norm": 0.9802488088607788,
+      "learning_rate": 1.8490018037848912e-05,
+      "loss": 0.4012,
+      "step": 5600
+    },
+    {
+      "epoch": 1.1034889724865384,
+      "grad_norm": 0.9458854794502258,
+      "learning_rate": 1.8486960775321778e-05,
+      "loss": 0.3817,
+      "step": 5610
+    },
+    {
+      "epoch": 1.105455976002557,
+      "grad_norm": 0.6298452615737915,
+      "learning_rate": 1.8483903512794647e-05,
+      "loss": 0.4332,
+      "step": 5620
+    },
+    {
+      "epoch": 1.107422979518576,
+      "grad_norm": 1.1540119647979736,
+      "learning_rate": 1.8480846250267512e-05,
+      "loss": 0.4173,
+      "step": 5630
+    },
+    {
+      "epoch": 1.1093899830345946,
+      "grad_norm": 1.4845781326293945,
+      "learning_rate": 1.8477788987740378e-05,
+      "loss": 0.3793,
+      "step": 5640
+    },
+    {
+      "epoch": 1.1113569865506134,
+      "grad_norm": 1.618861436843872,
+      "learning_rate": 1.8474731725213247e-05,
+      "loss": 0.4487,
+      "step": 5650
+    },
+    {
+      "epoch": 1.1133239900666323,
+      "grad_norm": 1.1488289833068848,
+      "learning_rate": 1.8471674462686113e-05,
+      "loss": 0.3696,
+      "step": 5660
+    },
+    {
+      "epoch": 1.115290993582651,
+      "grad_norm": 0.8757450580596924,
+      "learning_rate": 1.846861720015898e-05,
+      "loss": 0.4138,
+      "step": 5670
+    },
+    {
+      "epoch": 1.1172579970986698,
+      "grad_norm": 0.7664826512336731,
+      "learning_rate": 1.8465559937631845e-05,
+      "loss": 0.5212,
+      "step": 5680
+    },
+    {
+      "epoch": 1.1192250006146887,
+      "grad_norm": 0.9353389143943787,
+      "learning_rate": 1.846250267510471e-05,
+      "loss": 0.4855,
+      "step": 5690
+    },
+    {
+      "epoch": 1.1211920041307073,
+      "grad_norm": 1.3535192012786865,
+      "learning_rate": 1.845944541257758e-05,
+      "loss": 0.4871,
+      "step": 5700
+    },
+    {
+      "epoch": 1.1231590076467262,
+      "grad_norm": 0.7864957451820374,
+      "learning_rate": 1.8456388150050445e-05,
+      "loss": 0.4506,
+      "step": 5710
+    },
+    {
+      "epoch": 1.125126011162745,
+      "grad_norm": 1.1982388496398926,
+      "learning_rate": 1.8453330887523315e-05,
+      "loss": 0.4044,
+      "step": 5720
+    },
+    {
+      "epoch": 1.1270930146787637,
+      "grad_norm": 0.6584609746932983,
+      "learning_rate": 1.845027362499618e-05,
+      "loss": 0.4408,
+      "step": 5730
+    },
+    {
+      "epoch": 1.1290600181947825,
+      "grad_norm": 1.2081828117370605,
+      "learning_rate": 1.8447216362469046e-05,
+      "loss": 0.4274,
+      "step": 5740
+    },
+    {
+      "epoch": 1.1310270217108014,
+      "grad_norm": 1.024104356765747,
+      "learning_rate": 1.8444159099941915e-05,
+      "loss": 0.4062,
+      "step": 5750
+    },
+    {
+      "epoch": 1.13299402522682,
+      "grad_norm": 1.289136290550232,
+      "learning_rate": 1.844110183741478e-05,
+      "loss": 0.4041,
+      "step": 5760
+    },
+    {
+      "epoch": 1.1349610287428389,
+      "grad_norm": 1.1129745244979858,
+      "learning_rate": 1.8438044574887647e-05,
+      "loss": 0.3489,
+      "step": 5770
+    },
+    {
+      "epoch": 1.1369280322588577,
+      "grad_norm": 0.7395710945129395,
+      "learning_rate": 1.8434987312360516e-05,
+      "loss": 0.4426,
+      "step": 5780
+    },
+    {
+      "epoch": 1.1388950357748764,
+      "grad_norm": 0.6231014728546143,
+      "learning_rate": 1.843193004983338e-05,
+      "loss": 0.6161,
+      "step": 5790
+    },
+    {
+      "epoch": 1.1408620392908952,
+      "grad_norm": 1.7715225219726562,
+      "learning_rate": 1.8428872787306248e-05,
+      "loss": 0.4273,
+      "step": 5800
+    },
+    {
+      "epoch": 1.142829042806914,
+      "grad_norm": 0.9418803453445435,
+      "learning_rate": 1.8425815524779113e-05,
+      "loss": 0.4091,
+      "step": 5810
+    },
+    {
+      "epoch": 1.1447960463229327,
+      "grad_norm": 1.3185018301010132,
+      "learning_rate": 1.842275826225198e-05,
+      "loss": 0.4431,
+      "step": 5820
+    },
+    {
+      "epoch": 1.1467630498389516,
+      "grad_norm": 1.9988341331481934,
+      "learning_rate": 1.841970099972485e-05,
+      "loss": 0.4172,
+      "step": 5830
+    },
+    {
+      "epoch": 1.1487300533549702,
+      "grad_norm": 1.2586854696273804,
+      "learning_rate": 1.8416643737197714e-05,
+      "loss": 0.3204,
+      "step": 5840
+    },
+    {
+      "epoch": 1.150697056870989,
+      "grad_norm": 1.1469558477401733,
+      "learning_rate": 1.8413586474670583e-05,
+      "loss": 0.4413,
+      "step": 5850
+    },
+    {
+      "epoch": 1.152664060387008,
+      "grad_norm": 0.7499154806137085,
+      "learning_rate": 1.841052921214345e-05,
+      "loss": 0.4259,
+      "step": 5860
+    },
+    {
+      "epoch": 1.1546310639030266,
+      "grad_norm": 1.0914968252182007,
+      "learning_rate": 1.8407471949616315e-05,
+      "loss": 0.4374,
+      "step": 5870
+    },
+    {
+      "epoch": 1.1565980674190455,
+      "grad_norm": 1.812558889389038,
+      "learning_rate": 1.8404414687089184e-05,
+      "loss": 0.4405,
+      "step": 5880
+    },
+    {
+      "epoch": 1.1585650709350643,
+      "grad_norm": 0.8804789185523987,
+      "learning_rate": 1.840135742456205e-05,
+      "loss": 0.5325,
+      "step": 5890
+    },
+    {
+      "epoch": 1.160532074451083,
+      "grad_norm": 1.3734694719314575,
+      "learning_rate": 1.8398300162034916e-05,
+      "loss": 0.3723,
+      "step": 5900
+    },
+    {
+      "epoch": 1.1624990779671018,
+      "grad_norm": 1.260372519493103,
+      "learning_rate": 1.839524289950778e-05,
+      "loss": 0.4549,
+      "step": 5910
+    },
+    {
+      "epoch": 1.1644660814831207,
+      "grad_norm": 1.2938246726989746,
+      "learning_rate": 1.8392185636980647e-05,
+      "loss": 0.4422,
+      "step": 5920
+    },
+    {
+      "epoch": 1.1664330849991393,
+      "grad_norm": 1.4152106046676636,
+      "learning_rate": 1.8389128374453516e-05,
+      "loss": 0.3273,
+      "step": 5930
+    },
+    {
+      "epoch": 1.1684000885151582,
+      "grad_norm": 0.9573132395744324,
+      "learning_rate": 1.8386071111926382e-05,
+      "loss": 0.6119,
+      "step": 5940
+    },
+    {
+      "epoch": 1.170367092031177,
+      "grad_norm": 1.479049801826477,
+      "learning_rate": 1.8383013849399248e-05,
+      "loss": 0.5705,
+      "step": 5950
+    },
+    {
+      "epoch": 1.1723340955471957,
+      "grad_norm": 1.8711466789245605,
+      "learning_rate": 1.8379956586872117e-05,
+      "loss": 0.422,
+      "step": 5960
+    },
+    {
+      "epoch": 1.1743010990632146,
+      "grad_norm": 1.1066919565200806,
+      "learning_rate": 1.8376899324344983e-05,
+      "loss": 0.5835,
+      "step": 5970
+    },
+    {
+      "epoch": 1.1762681025792334,
+      "grad_norm": 1.9158978462219238,
+      "learning_rate": 1.8373842061817852e-05,
+      "loss": 0.3937,
+      "step": 5980
+    },
+    {
+      "epoch": 1.178235106095252,
+      "grad_norm": 1.785532832145691,
+      "learning_rate": 1.8370784799290718e-05,
+      "loss": 0.3908,
+      "step": 5990
+    },
+    {
+      "epoch": 1.180202109611271,
+      "grad_norm": 1.3090122938156128,
+      "learning_rate": 1.8367727536763584e-05,
+      "loss": 0.408,
+      "step": 6000
+    },
+    {
+      "epoch": 1.180202109611271,
+      "eval_loss": 0.2541753053665161,
+      "eval_runtime": 8.8707,
+      "eval_samples_per_second": 5.637,
+      "eval_steps_per_second": 2.818,
+      "step": 6000
+    },
+    {
+      "epoch": 1.1821691131272898,
+      "grad_norm": 1.0158838033676147,
+      "learning_rate": 1.836467027423645e-05,
+      "loss": 0.5237,
+      "step": 6010
+    },
+    {
+      "epoch": 1.1841361166433084,
+      "grad_norm": 1.9016451835632324,
+      "learning_rate": 1.8361613011709315e-05,
+      "loss": 0.419,
+      "step": 6020
+    },
+    {
+      "epoch": 1.1861031201593273,
+      "grad_norm": 0.9506327509880066,
+      "learning_rate": 1.8358555749182184e-05,
+      "loss": 0.4917,
+      "step": 6030
+    },
+    {
+      "epoch": 1.1880701236753461,
+      "grad_norm": 0.9373012781143188,
+      "learning_rate": 1.835549848665505e-05,
+      "loss": 0.3754,
+      "step": 6040
+    },
+    {
+      "epoch": 1.1900371271913648,
+      "grad_norm": 1.2526496648788452,
+      "learning_rate": 1.8352441224127916e-05,
+      "loss": 0.4037,
+      "step": 6050
+    },
+    {
+      "epoch": 1.1920041307073836,
+      "grad_norm": 1.1698423624038696,
+      "learning_rate": 1.8349383961600785e-05,
+      "loss": 0.5628,
+      "step": 6060
+    },
+    {
+      "epoch": 1.1939711342234025,
+      "grad_norm": 1.2731508016586304,
+      "learning_rate": 1.834632669907365e-05,
+      "loss": 0.4187,
+      "step": 6070
+    },
+    {
+      "epoch": 1.1959381377394211,
+      "grad_norm": 1.4930040836334229,
+      "learning_rate": 1.8343269436546517e-05,
+      "loss": 0.4448,
+      "step": 6080
+    },
+    {
+      "epoch": 1.19790514125544,
+      "grad_norm": 0.7274429202079773,
+      "learning_rate": 1.8340212174019386e-05,
+      "loss": 0.5058,
+      "step": 6090
+    },
+    {
+      "epoch": 1.1998721447714589,
+      "grad_norm": 0.5345088839530945,
+      "learning_rate": 1.833715491149225e-05,
+      "loss": 0.451,
+      "step": 6100
+    },
+    {
+      "epoch": 1.2018391482874775,
+      "grad_norm": 0.7829896807670593,
+      "learning_rate": 1.833409764896512e-05,
+      "loss": 0.3958,
+      "step": 6110
+    },
+    {
+      "epoch": 1.2038061518034964,
+      "grad_norm": 1.2394695281982422,
+      "learning_rate": 1.8331040386437986e-05,
+      "loss": 0.3365,
+      "step": 6120
+    },
+    {
+      "epoch": 1.2057731553195152,
+      "grad_norm": 0.8492904901504517,
+      "learning_rate": 1.8327983123910852e-05,
+      "loss": 0.414,
+      "step": 6130
+    },
+    {
+      "epoch": 1.2077401588355339,
+      "grad_norm": 1.0777946710586548,
+      "learning_rate": 1.8324925861383718e-05,
+      "loss": 0.4816,
+      "step": 6140
+    },
+    {
+      "epoch": 1.2097071623515527,
+      "grad_norm": 2.155651569366455,
+      "learning_rate": 1.8321868598856584e-05,
+      "loss": 0.5321,
+      "step": 6150
+    },
+    {
+      "epoch": 1.2116741658675716,
+      "grad_norm": 2.0705020427703857,
+      "learning_rate": 1.8318811336329453e-05,
+      "loss": 0.4464,
+      "step": 6160
+    },
+    {
+      "epoch": 1.2136411693835902,
+      "grad_norm": 1.2468976974487305,
+      "learning_rate": 1.831575407380232e-05,
+      "loss": 0.5157,
+      "step": 6170
+    },
+    {
+      "epoch": 1.215608172899609,
+      "grad_norm": 0.6635984778404236,
+      "learning_rate": 1.8312696811275184e-05,
+      "loss": 0.4052,
+      "step": 6180
+    },
+    {
+      "epoch": 1.217575176415628,
+      "grad_norm": 1.6718881130218506,
+      "learning_rate": 1.8309639548748054e-05,
+      "loss": 0.4765,
+      "step": 6190
+    },
+    {
+      "epoch": 1.2195421799316466,
+      "grad_norm": 1.2337743043899536,
+      "learning_rate": 1.830658228622092e-05,
+      "loss": 0.3758,
+      "step": 6200
+    },
+    {
+      "epoch": 1.2215091834476655,
+      "grad_norm": 1.3416141271591187,
+      "learning_rate": 1.8303525023693785e-05,
+      "loss": 0.4676,
+      "step": 6210
+    },
+    {
+      "epoch": 1.2234761869636843,
+      "grad_norm": 1.2023998498916626,
+      "learning_rate": 1.8300467761166654e-05,
+      "loss": 0.4034,
+      "step": 6220
+    },
+    {
+      "epoch": 1.225443190479703,
+      "grad_norm": 1.1398640871047974,
+      "learning_rate": 1.829741049863952e-05,
+      "loss": 0.5946,
+      "step": 6230
+    },
+    {
+      "epoch": 1.2274101939957218,
+      "grad_norm": 0.9996533989906311,
+      "learning_rate": 1.8294353236112386e-05,
+      "loss": 0.3983,
+      "step": 6240
+    },
+    {
+      "epoch": 1.2293771975117405,
+      "grad_norm": 1.0722390413284302,
+      "learning_rate": 1.8291295973585252e-05,
+      "loss": 0.4448,
+      "step": 6250
+    },
+    {
+      "epoch": 1.2313442010277593,
+      "grad_norm": 1.5299232006072998,
+      "learning_rate": 1.828823871105812e-05,
+      "loss": 0.5033,
+      "step": 6260
+    },
+    {
+      "epoch": 1.2333112045437782,
+      "grad_norm": 1.0260642766952515,
+      "learning_rate": 1.8285181448530987e-05,
+      "loss": 0.3642,
+      "step": 6270
+    },
+    {
+      "epoch": 1.2352782080597968,
+      "grad_norm": 1.560285210609436,
+      "learning_rate": 1.8282124186003852e-05,
+      "loss": 0.503,
+      "step": 6280
+    },
+    {
+      "epoch": 1.2372452115758157,
+      "grad_norm": 1.4115056991577148,
+      "learning_rate": 1.827906692347672e-05,
+      "loss": 0.3812,
+      "step": 6290
+    },
+    {
+      "epoch": 1.2392122150918345,
+      "grad_norm": 0.7947004437446594,
+      "learning_rate": 1.8276009660949587e-05,
+      "loss": 0.4024,
+      "step": 6300
+    },
+    {
+      "epoch": 1.2411792186078532,
+      "grad_norm": 0.9930680394172668,
+      "learning_rate": 1.8272952398422453e-05,
+      "loss": 0.441,
+      "step": 6310
+    },
+    {
+      "epoch": 1.243146222123872,
+      "grad_norm": 2.014035940170288,
+      "learning_rate": 1.8269895135895322e-05,
+      "loss": 0.4854,
+      "step": 6320
+    },
+    {
+      "epoch": 1.245113225639891,
+      "grad_norm": 1.7918622493743896,
+      "learning_rate": 1.8266837873368188e-05,
+      "loss": 0.4661,
+      "step": 6330
+    },
+    {
+      "epoch": 1.2470802291559095,
+      "grad_norm": 0.7752891182899475,
+      "learning_rate": 1.8263780610841054e-05,
+      "loss": 0.5,
+      "step": 6340
+    },
+    {
+      "epoch": 1.2490472326719284,
+      "grad_norm": 0.6974585056304932,
+      "learning_rate": 1.826072334831392e-05,
+      "loss": 0.4312,
+      "step": 6350
+    },
+    {
+      "epoch": 1.251014236187947,
+      "grad_norm": 1.6854921579360962,
+      "learning_rate": 1.825766608578679e-05,
+      "loss": 0.3643,
+      "step": 6360
+    },
+    {
+      "epoch": 1.252981239703966,
+      "grad_norm": 0.9491457343101501,
+      "learning_rate": 1.8254608823259655e-05,
+      "loss": 0.4874,
+      "step": 6370
+    },
+    {
+      "epoch": 1.2549482432199848,
+      "grad_norm": 1.1778289079666138,
+      "learning_rate": 1.825155156073252e-05,
+      "loss": 0.4696,
+      "step": 6380
+    },
+    {
+      "epoch": 1.2569152467360034,
+      "grad_norm": 3.334805488586426,
+      "learning_rate": 1.824849429820539e-05,
+      "loss": 0.475,
+      "step": 6390
+    },
+    {
+      "epoch": 1.2588822502520223,
+      "grad_norm": 0.7782668471336365,
+      "learning_rate": 1.8245437035678255e-05,
+      "loss": 0.3803,
+      "step": 6400
+    },
+    {
+      "epoch": 1.2608492537680411,
+      "grad_norm": 1.3739856481552124,
+      "learning_rate": 1.824237977315112e-05,
+      "loss": 0.4898,
+      "step": 6410
+    },
+    {
+      "epoch": 1.2628162572840598,
+      "grad_norm": 1.7571340799331665,
+      "learning_rate": 1.823932251062399e-05,
+      "loss": 0.5217,
+      "step": 6420
+    },
+    {
+      "epoch": 1.2647832608000786,
+      "grad_norm": 0.6563398241996765,
+      "learning_rate": 1.8236265248096856e-05,
+      "loss": 0.3598,
+      "step": 6430
+    },
+    {
+      "epoch": 1.2667502643160975,
+      "grad_norm": 0.8935515284538269,
+      "learning_rate": 1.8233207985569722e-05,
+      "loss": 0.4735,
+      "step": 6440
+    },
+    {
+      "epoch": 1.2687172678321161,
+      "grad_norm": 0.8654441833496094,
+      "learning_rate": 1.823015072304259e-05,
+      "loss": 0.4369,
+      "step": 6450
+    },
+    {
+      "epoch": 1.270684271348135,
+      "grad_norm": 1.5720266103744507,
+      "learning_rate": 1.8227093460515457e-05,
+      "loss": 0.3759,
+      "step": 6460
+    },
+    {
+      "epoch": 1.2726512748641539,
+      "grad_norm": 1.4234576225280762,
+      "learning_rate": 1.8224036197988323e-05,
+      "loss": 0.5163,
+      "step": 6470
+    },
+    {
+      "epoch": 1.2746182783801725,
+      "grad_norm": 1.3795866966247559,
+      "learning_rate": 1.822097893546119e-05,
+      "loss": 0.4891,
+      "step": 6480
+    },
+    {
+      "epoch": 1.2765852818961914,
+      "grad_norm": 2.2530128955841064,
+      "learning_rate": 1.8217921672934054e-05,
+      "loss": 0.5406,
+      "step": 6490
+    },
+    {
+      "epoch": 1.2785522854122102,
+      "grad_norm": 1.2608258724212646,
+      "learning_rate": 1.8214864410406923e-05,
+      "loss": 0.4731,
+      "step": 6500
+    },
+    {
+      "epoch": 1.2785522854122102,
+      "eval_loss": 0.2500273585319519,
+      "eval_runtime": 8.9303,
+      "eval_samples_per_second": 5.599,
+      "eval_steps_per_second": 2.799,
+      "step": 6500
+    },
+    {
+      "epoch": 1.2805192889282289,
+      "grad_norm": 1.2853199243545532,
+      "learning_rate": 1.821180714787979e-05,
+      "loss": 0.4657,
+      "step": 6510
+    },
+    {
+      "epoch": 1.2824862924442477,
+      "grad_norm": 0.6123810410499573,
+      "learning_rate": 1.8208749885352658e-05,
+      "loss": 0.6216,
+      "step": 6520
+    },
+    {
+      "epoch": 1.2844532959602666,
+      "grad_norm": 0.8398245573043823,
+      "learning_rate": 1.8205692622825524e-05,
+      "loss": 0.4216,
+      "step": 6530
+    },
+    {
+      "epoch": 1.2864202994762852,
+      "grad_norm": 1.0161961317062378,
+      "learning_rate": 1.820263536029839e-05,
+      "loss": 0.4325,
+      "step": 6540
+    },
+    {
+      "epoch": 1.288387302992304,
+      "grad_norm": 1.7545063495635986,
+      "learning_rate": 1.819957809777126e-05,
+      "loss": 0.5181,
+      "step": 6550
+    },
+    {
+      "epoch": 1.290354306508323,
+      "grad_norm": 0.9450379014015198,
+      "learning_rate": 1.8196520835244125e-05,
+      "loss": 0.4417,
+      "step": 6560
+    },
+    {
+      "epoch": 1.2923213100243416,
+      "grad_norm": 1.3141273260116577,
+      "learning_rate": 1.819346357271699e-05,
+      "loss": 0.5343,
+      "step": 6570
+    },
+    {
+      "epoch": 1.2942883135403604,
+      "grad_norm": 1.6922868490219116,
+      "learning_rate": 1.8190406310189856e-05,
+      "loss": 0.4544,
+      "step": 6580
+    },
+    {
+      "epoch": 1.2962553170563793,
+      "grad_norm": 1.2070684432983398,
+      "learning_rate": 1.8187349047662722e-05,
+      "loss": 0.3989,
+      "step": 6590
+    },
+    {
+      "epoch": 1.298222320572398,
+      "grad_norm": 1.0390480756759644,
+      "learning_rate": 1.818429178513559e-05,
+      "loss": 0.3849,
+      "step": 6600
+    },
+    {
+      "epoch": 1.3001893240884168,
+      "grad_norm": 1.3022695779800415,
+      "learning_rate": 1.8181234522608457e-05,
+      "loss": 0.3879,
+      "step": 6610
+    },
+    {
+      "epoch": 1.3021563276044357,
+      "grad_norm": 2.036789655685425,
+      "learning_rate": 1.8178177260081323e-05,
+      "loss": 0.4393,
+      "step": 6620
+    },
+    {
+      "epoch": 1.3041233311204543,
+      "grad_norm": 1.0028927326202393,
+      "learning_rate": 1.8175119997554192e-05,
+      "loss": 0.5197,
+      "step": 6630
+    },
+    {
+      "epoch": 1.3060903346364732,
+      "grad_norm": 0.9908479452133179,
+      "learning_rate": 1.8172062735027058e-05,
+      "loss": 0.3509,
+      "step": 6640
+    },
+    {
+      "epoch": 1.308057338152492,
+      "grad_norm": 1.617487907409668,
+      "learning_rate": 1.8169005472499927e-05,
+      "loss": 0.5052,
+      "step": 6650
+    },
+    {
+      "epoch": 1.3100243416685107,
+      "grad_norm": 1.229190468788147,
+      "learning_rate": 1.8165948209972793e-05,
+      "loss": 0.4564,
+      "step": 6660
+    },
+    {
+      "epoch": 1.3119913451845295,
+      "grad_norm": 1.5831772089004517,
+      "learning_rate": 1.816289094744566e-05,
+      "loss": 0.4283,
+      "step": 6670
+    },
+    {
+      "epoch": 1.3139583487005484,
+      "grad_norm": 1.4236372709274292,
+      "learning_rate": 1.8159833684918528e-05,
+      "loss": 0.4583,
+      "step": 6680
+    },
+    {
+      "epoch": 1.315925352216567,
+      "grad_norm": 0.9847359657287598,
+      "learning_rate": 1.815677642239139e-05,
+      "loss": 0.5584,
+      "step": 6690
+    },
+    {
+      "epoch": 1.317892355732586,
+      "grad_norm": 1.2194691896438599,
+      "learning_rate": 1.815371915986426e-05,
+      "loss": 0.4529,
+      "step": 6700
+    },
+    {
+      "epoch": 1.3198593592486048,
+      "grad_norm": 1.4143697023391724,
+      "learning_rate": 1.8150661897337125e-05,
+      "loss": 0.4783,
+      "step": 6710
+    },
+    {
+      "epoch": 1.3218263627646234,
+      "grad_norm": 1.3399523496627808,
+      "learning_rate": 1.814760463480999e-05,
+      "loss": 0.3955,
+      "step": 6720
+    },
+    {
+      "epoch": 1.3237933662806423,
+      "grad_norm": 0.7370299100875854,
+      "learning_rate": 1.814454737228286e-05,
+      "loss": 0.4647,
+      "step": 6730
+    },
+    {
+      "epoch": 1.3257603697966611,
+      "grad_norm": 1.4617048501968384,
+      "learning_rate": 1.8141490109755726e-05,
+      "loss": 0.4911,
+      "step": 6740
+    },
+    {
+      "epoch": 1.3277273733126798,
+      "grad_norm": 0.836281418800354,
+      "learning_rate": 1.813843284722859e-05,
+      "loss": 0.442,
+      "step": 6750
+    },
+    {
+      "epoch": 1.3296943768286986,
+      "grad_norm": 0.901801586151123,
+      "learning_rate": 1.813537558470146e-05,
+      "loss": 0.3387,
+      "step": 6760
+    },
+    {
+      "epoch": 1.3316613803447175,
+      "grad_norm": 1.461903691291809,
+      "learning_rate": 1.8132318322174326e-05,
+      "loss": 0.4815,
+      "step": 6770
+    },
+    {
+      "epoch": 1.3336283838607361,
+      "grad_norm": 1.2641518115997314,
+      "learning_rate": 1.8129261059647196e-05,
+      "loss": 0.4465,
+      "step": 6780
+    },
+    {
+      "epoch": 1.335595387376755,
+      "grad_norm": 2.127981662750244,
+      "learning_rate": 1.812620379712006e-05,
+      "loss": 0.4972,
+      "step": 6790
+    },
+    {
+      "epoch": 1.3375623908927738,
+      "grad_norm": 1.0643235445022583,
+      "learning_rate": 1.8123146534592927e-05,
+      "loss": 0.4823,
+      "step": 6800
+    },
+    {
+      "epoch": 1.3395293944087925,
+      "grad_norm": 1.392342448234558,
+      "learning_rate": 1.8120089272065793e-05,
+      "loss": 0.4488,
+      "step": 6810
+    },
+    {
+      "epoch": 1.3414963979248113,
+      "grad_norm": 1.372171401977539,
+      "learning_rate": 1.811703200953866e-05,
+      "loss": 0.5518,
+      "step": 6820
+    },
+    {
+      "epoch": 1.3434634014408302,
+      "grad_norm": 1.0841792821884155,
+      "learning_rate": 1.8113974747011528e-05,
+      "loss": 0.4059,
+      "step": 6830
+    },
+    {
+      "epoch": 1.3454304049568488,
+      "grad_norm": 1.0114047527313232,
+      "learning_rate": 1.8110917484484394e-05,
+      "loss": 0.4705,
+      "step": 6840
+    },
+    {
+      "epoch": 1.3473974084728677,
+      "grad_norm": 1.600369930267334,
+      "learning_rate": 1.810786022195726e-05,
+      "loss": 0.5064,
+      "step": 6850
+    },
+    {
+      "epoch": 1.3493644119888863,
+      "grad_norm": 1.3853987455368042,
+      "learning_rate": 1.810480295943013e-05,
+      "loss": 0.5352,
+      "step": 6860
+    },
+    {
+      "epoch": 1.3513314155049052,
+      "grad_norm": 1.627324104309082,
+      "learning_rate": 1.8101745696902994e-05,
+      "loss": 0.6512,
+      "step": 6870
+    },
+    {
+      "epoch": 1.353298419020924,
+      "grad_norm": 1.6808173656463623,
+      "learning_rate": 1.809868843437586e-05,
+      "loss": 0.4856,
+      "step": 6880
+    },
+    {
+      "epoch": 1.3552654225369427,
+      "grad_norm": 1.0722404718399048,
+      "learning_rate": 1.809563117184873e-05,
+      "loss": 0.4883,
+      "step": 6890
+    },
+    {
+      "epoch": 1.3572324260529616,
+      "grad_norm": 1.3000462055206299,
+      "learning_rate": 1.8092573909321595e-05,
+      "loss": 0.459,
+      "step": 6900
+    },
+    {
+      "epoch": 1.3591994295689804,
+      "grad_norm": 1.0217570066452026,
+      "learning_rate": 1.8089516646794464e-05,
+      "loss": 0.4908,
+      "step": 6910
+    },
+    {
+      "epoch": 1.361166433084999,
+      "grad_norm": 2.192474603652954,
+      "learning_rate": 1.8086459384267327e-05,
+      "loss": 0.5604,
+      "step": 6920
+    },
+    {
+      "epoch": 1.363133436601018,
+      "grad_norm": 1.4930495023727417,
+      "learning_rate": 1.8083402121740196e-05,
+      "loss": 0.412,
+      "step": 6930
+    },
+    {
+      "epoch": 1.3651004401170366,
+      "grad_norm": 0.9470030069351196,
+      "learning_rate": 1.808034485921306e-05,
+      "loss": 0.5397,
+      "step": 6940
+    },
+    {
+      "epoch": 1.3670674436330554,
+      "grad_norm": 1.439299464225769,
+      "learning_rate": 1.8077287596685927e-05,
+      "loss": 0.6071,
+      "step": 6950
+    },
+    {
+      "epoch": 1.3690344471490743,
+      "grad_norm": 1.0739001035690308,
+      "learning_rate": 1.8074230334158796e-05,
+      "loss": 0.4911,
+      "step": 6960
+    },
+    {
+      "epoch": 1.371001450665093,
+      "grad_norm": 2.3669240474700928,
+      "learning_rate": 1.8071173071631662e-05,
+      "loss": 0.5202,
+      "step": 6970
+    },
+    {
+      "epoch": 1.3729684541811118,
+      "grad_norm": 0.8354695439338684,
+      "learning_rate": 1.8068115809104528e-05,
+      "loss": 0.453,
+      "step": 6980
+    },
+    {
+      "epoch": 1.3749354576971307,
+      "grad_norm": 0.8748831748962402,
+      "learning_rate": 1.8065058546577397e-05,
+      "loss": 0.4235,
+      "step": 6990
+    },
+    {
+      "epoch": 1.3769024612131493,
+      "grad_norm": 0.8159099221229553,
+      "learning_rate": 1.8062001284050263e-05,
+      "loss": 0.4364,
+      "step": 7000
+    },
+    {
+      "epoch": 1.3769024612131493,
+      "eval_loss": 0.2508287727832794,
+      "eval_runtime": 8.8467,
+      "eval_samples_per_second": 5.652,
+      "eval_steps_per_second": 2.826,
+      "step": 7000
+    },
+    {
+      "epoch": 1.3788694647291682,
+      "grad_norm": 0.868539035320282,
+      "learning_rate": 1.805894402152313e-05,
+      "loss": 0.621,
+      "step": 7010
+    },
+    {
+      "epoch": 1.380836468245187,
+      "grad_norm": 1.5465404987335205,
+      "learning_rate": 1.8055886758995998e-05,
+      "loss": 0.402,
+      "step": 7020
+    },
+    {
+      "epoch": 1.3828034717612057,
+      "grad_norm": 2.190101146697998,
+      "learning_rate": 1.8052829496468864e-05,
+      "loss": 0.5353,
+      "step": 7030
+    },
+    {
+      "epoch": 1.3847704752772245,
+      "grad_norm": 0.7492843866348267,
+      "learning_rate": 1.804977223394173e-05,
+      "loss": 0.517,
+      "step": 7040
+    },
+    {
+      "epoch": 1.3867374787932434,
+      "grad_norm": 0.9948645234107971,
+      "learning_rate": 1.8046714971414595e-05,
+      "loss": 0.5211,
+      "step": 7050
+    },
+    {
+      "epoch": 1.388704482309262,
+      "grad_norm": 1.594095230102539,
+      "learning_rate": 1.8043657708887464e-05,
+      "loss": 0.4808,
+      "step": 7060
+    },
+    {
+      "epoch": 1.3906714858252809,
+      "grad_norm": 1.165153980255127,
+      "learning_rate": 1.804060044636033e-05,
+      "loss": 0.3884,
+      "step": 7070
+    },
+    {
+      "epoch": 1.3926384893412997,
+      "grad_norm": 2.061393976211548,
+      "learning_rate": 1.8037543183833196e-05,
+      "loss": 0.4339,
+      "step": 7080
+    },
+    {
+      "epoch": 1.3946054928573184,
+      "grad_norm": 1.4817818403244019,
+      "learning_rate": 1.8034485921306065e-05,
+      "loss": 0.4482,
+      "step": 7090
+    },
+    {
+      "epoch": 1.3965724963733372,
+      "grad_norm": 0.8791131377220154,
+      "learning_rate": 1.803142865877893e-05,
+      "loss": 0.4498,
+      "step": 7100
+    },
+    {
+      "epoch": 1.398539499889356,
+      "grad_norm": 0.8255666494369507,
+      "learning_rate": 1.8028371396251797e-05,
+      "loss": 0.4166,
+      "step": 7110
+    },
+    {
+      "epoch": 1.4005065034053747,
+      "grad_norm": 0.8668946027755737,
+      "learning_rate": 1.8025314133724666e-05,
+      "loss": 0.5698,
+      "step": 7120
+    },
+    {
+      "epoch": 1.4024735069213936,
+      "grad_norm": 1.2157158851623535,
+      "learning_rate": 1.802225687119753e-05,
+      "loss": 0.4634,
+      "step": 7130
+    },
+    {
+      "epoch": 1.4044405104374125,
+      "grad_norm": 1.54035484790802,
+      "learning_rate": 1.8019199608670397e-05,
+      "loss": 0.3969,
+      "step": 7140
+    },
+    {
+      "epoch": 1.406407513953431,
+      "grad_norm": 0.8776307702064514,
+      "learning_rate": 1.8016142346143263e-05,
+      "loss": 0.5061,
+      "step": 7150
+    },
+    {
+      "epoch": 1.40837451746945,
+      "grad_norm": 1.1032741069793701,
+      "learning_rate": 1.8013085083616132e-05,
+      "loss": 0.3971,
+      "step": 7160
+    },
+    {
+      "epoch": 1.4103415209854688,
+      "grad_norm": 1.0241215229034424,
+      "learning_rate": 1.8010027821088998e-05,
+      "loss": 0.4959,
+      "step": 7170
+    },
+    {
+      "epoch": 1.4123085245014875,
+      "grad_norm": 1.6800438165664673,
+      "learning_rate": 1.8006970558561864e-05,
+      "loss": 0.4738,
+      "step": 7180
+    },
+    {
+      "epoch": 1.4142755280175063,
+      "grad_norm": 1.5936535596847534,
+      "learning_rate": 1.8003913296034733e-05,
+      "loss": 0.4572,
+      "step": 7190
+    },
+    {
+      "epoch": 1.4162425315335252,
+      "grad_norm": 1.2243019342422485,
+      "learning_rate": 1.80008560335076e-05,
+      "loss": 0.3557,
+      "step": 7200
+    },
+    {
+      "epoch": 1.4182095350495438,
+      "grad_norm": 0.8471090197563171,
+      "learning_rate": 1.7997798770980465e-05,
+      "loss": 0.3765,
+      "step": 7210
+    },
+    {
+      "epoch": 1.4201765385655627,
+      "grad_norm": 1.2940151691436768,
+      "learning_rate": 1.7994741508453334e-05,
+      "loss": 0.5282,
+      "step": 7220
+    },
+    {
+      "epoch": 1.4221435420815816,
+      "grad_norm": 0.7135307192802429,
+      "learning_rate": 1.79916842459262e-05,
+      "loss": 0.4611,
+      "step": 7230
+    },
+    {
+      "epoch": 1.4241105455976002,
+      "grad_norm": 0.9937286972999573,
+      "learning_rate": 1.7988626983399065e-05,
+      "loss": 0.502,
+      "step": 7240
+    },
+    {
+      "epoch": 1.426077549113619,
+      "grad_norm": 1.5111737251281738,
+      "learning_rate": 1.7985569720871935e-05,
+      "loss": 0.4421,
+      "step": 7250
+    },
+    {
+      "epoch": 1.428044552629638,
+      "grad_norm": 1.29506516456604,
+      "learning_rate": 1.7982512458344797e-05,
+      "loss": 0.5282,
+      "step": 7260
+    },
+    {
+      "epoch": 1.4300115561456566,
+      "grad_norm": 1.1045817136764526,
+      "learning_rate": 1.7979455195817666e-05,
+      "loss": 0.4594,
+      "step": 7270
+    },
+    {
+      "epoch": 1.4319785596616754,
+      "grad_norm": 1.3081828355789185,
+      "learning_rate": 1.7976397933290532e-05,
+      "loss": 0.451,
+      "step": 7280
+    },
+    {
+      "epoch": 1.4339455631776943,
+      "grad_norm": 1.3445611000061035,
+      "learning_rate": 1.79733406707634e-05,
+      "loss": 0.4904,
+      "step": 7290
+    },
+    {
+      "epoch": 1.435912566693713,
+      "grad_norm": 1.8529108762741089,
+      "learning_rate": 1.7970283408236267e-05,
+      "loss": 0.537,
+      "step": 7300
+    },
+    {
+      "epoch": 1.4378795702097318,
+      "grad_norm": 1.6065847873687744,
+      "learning_rate": 1.7967226145709133e-05,
+      "loss": 0.3289,
+      "step": 7310
+    },
+    {
+      "epoch": 1.4398465737257506,
+      "grad_norm": 1.597464680671692,
+      "learning_rate": 1.7964168883182002e-05,
+      "loss": 0.4434,
+      "step": 7320
+    },
+    {
+      "epoch": 1.4418135772417693,
+      "grad_norm": 1.5186161994934082,
+      "learning_rate": 1.7961111620654868e-05,
+      "loss": 0.5233,
+      "step": 7330
+    },
+    {
+      "epoch": 1.4437805807577881,
+      "grad_norm": 2.497467279434204,
+      "learning_rate": 1.7958054358127733e-05,
+      "loss": 0.4332,
+      "step": 7340
+    },
+    {
+      "epoch": 1.445747584273807,
+      "grad_norm": 1.7080587148666382,
+      "learning_rate": 1.7954997095600602e-05,
+      "loss": 0.4665,
+      "step": 7350
+    },
+    {
+      "epoch": 1.4477145877898256,
+      "grad_norm": 1.3408483266830444,
+      "learning_rate": 1.7951939833073468e-05,
+      "loss": 0.4356,
+      "step": 7360
+    },
+    {
+      "epoch": 1.4496815913058445,
+      "grad_norm": 0.8652418255805969,
+      "learning_rate": 1.7948882570546334e-05,
+      "loss": 0.3355,
+      "step": 7370
+    },
+    {
+      "epoch": 1.4516485948218634,
+      "grad_norm": 1.6720373630523682,
+      "learning_rate": 1.79458253080192e-05,
+      "loss": 0.4702,
+      "step": 7380
+    },
+    {
+      "epoch": 1.453615598337882,
+      "grad_norm": 1.1792707443237305,
+      "learning_rate": 1.7942768045492066e-05,
+      "loss": 0.4056,
+      "step": 7390
+    },
+    {
+      "epoch": 1.4555826018539009,
+      "grad_norm": 0.5109983682632446,
+      "learning_rate": 1.7939710782964935e-05,
+      "loss": 0.4889,
+      "step": 7400
+    },
+    {
+      "epoch": 1.4575496053699197,
+      "grad_norm": 0.864371120929718,
+      "learning_rate": 1.79366535204378e-05,
+      "loss": 0.5084,
+      "step": 7410
+    },
+    {
+      "epoch": 1.4595166088859384,
+      "grad_norm": 1.4081306457519531,
+      "learning_rate": 1.793359625791067e-05,
+      "loss": 0.4884,
+      "step": 7420
+    },
+    {
+      "epoch": 1.4614836124019572,
+      "grad_norm": 0.5403172969818115,
+      "learning_rate": 1.7930538995383535e-05,
+      "loss": 0.4762,
+      "step": 7430
+    },
+    {
+      "epoch": 1.4634506159179759,
+      "grad_norm": 2.7623186111450195,
+      "learning_rate": 1.79274817328564e-05,
+      "loss": 0.4962,
+      "step": 7440
+    },
+    {
+      "epoch": 1.4654176194339947,
+      "grad_norm": 1.38148832321167,
+      "learning_rate": 1.792442447032927e-05,
+      "loss": 0.4259,
+      "step": 7450
+    },
+    {
+      "epoch": 1.4673846229500136,
+      "grad_norm": 1.2702479362487793,
+      "learning_rate": 1.7921367207802136e-05,
+      "loss": 0.4708,
+      "step": 7460
+    },
+    {
+      "epoch": 1.4693516264660322,
+      "grad_norm": 1.7170747518539429,
+      "learning_rate": 1.7918309945275002e-05,
+      "loss": 0.5886,
+      "step": 7470
+    },
+    {
+      "epoch": 1.471318629982051,
+      "grad_norm": 2.262479543685913,
+      "learning_rate": 1.791525268274787e-05,
+      "loss": 0.47,
+      "step": 7480
+    },
+    {
+      "epoch": 1.47328563349807,
+      "grad_norm": 0.977174699306488,
+      "learning_rate": 1.7912195420220734e-05,
+      "loss": 0.3943,
+      "step": 7490
+    },
+    {
+      "epoch": 1.4752526370140886,
+      "grad_norm": 1.1417875289916992,
+      "learning_rate": 1.7909138157693603e-05,
+      "loss": 0.3999,
+      "step": 7500
+    },
+    {
+      "epoch": 1.4752526370140886,
+      "eval_loss": 0.2466391772031784,
+      "eval_runtime": 8.9007,
+      "eval_samples_per_second": 5.618,
+      "eval_steps_per_second": 2.809,
+      "step": 7500
+    },
+    {
+      "epoch": 1.4772196405301075,
+      "grad_norm": 0.9850865602493286,
+      "learning_rate": 1.790608089516647e-05,
+      "loss": 0.4194,
+      "step": 7510
+    },
+    {
+      "epoch": 1.479186644046126,
+      "grad_norm": 0.9605159759521484,
+      "learning_rate": 1.7903023632639334e-05,
+      "loss": 0.5008,
+      "step": 7520
+    },
+    {
+      "epoch": 1.481153647562145,
+      "grad_norm": 0.9875272512435913,
+      "learning_rate": 1.7899966370112203e-05,
+      "loss": 0.384,
+      "step": 7530
+    },
+    {
+      "epoch": 1.4831206510781638,
+      "grad_norm": 1.9756124019622803,
+      "learning_rate": 1.789690910758507e-05,
+      "loss": 0.4649,
+      "step": 7540
+    },
+    {
+      "epoch": 1.4850876545941825,
+      "grad_norm": 1.7961291074752808,
+      "learning_rate": 1.789385184505794e-05,
+      "loss": 0.5256,
+      "step": 7550
+    },
+    {
+      "epoch": 1.4870546581102013,
+      "grad_norm": 1.253611445426941,
+      "learning_rate": 1.7890794582530804e-05,
+      "loss": 0.4242,
+      "step": 7560
+    },
+    {
+      "epoch": 1.4890216616262202,
+      "grad_norm": 1.7903774976730347,
+      "learning_rate": 1.788773732000367e-05,
+      "loss": 0.4745,
+      "step": 7570
+    },
+    {
+      "epoch": 1.4909886651422388,
+      "grad_norm": 2.0283164978027344,
+      "learning_rate": 1.788468005747654e-05,
+      "loss": 0.4065,
+      "step": 7580
+    },
+    {
+      "epoch": 1.4929556686582577,
+      "grad_norm": 1.0774911642074585,
+      "learning_rate": 1.7881622794949405e-05,
+      "loss": 0.5152,
+      "step": 7590
+    },
+    {
+      "epoch": 1.4949226721742765,
+      "grad_norm": 2.157869815826416,
+      "learning_rate": 1.787856553242227e-05,
+      "loss": 0.4008,
+      "step": 7600
+    },
+    {
+      "epoch": 1.4968896756902952,
+      "grad_norm": 1.0705121755599976,
+      "learning_rate": 1.7875508269895136e-05,
+      "loss": 0.4837,
+      "step": 7610
+    },
+    {
+      "epoch": 1.498856679206314,
+      "grad_norm": 1.2112722396850586,
+      "learning_rate": 1.7872451007368002e-05,
+      "loss": 0.4164,
+      "step": 7620
+    },
+    {
+      "epoch": 1.500823682722333,
+      "grad_norm": 0.9930263161659241,
+      "learning_rate": 1.786939374484087e-05,
+      "loss": 0.2991,
+      "step": 7630
+    },
+    {
+      "epoch": 1.5027906862383515,
+      "grad_norm": 1.176189661026001,
+      "learning_rate": 1.7866336482313737e-05,
+      "loss": 0.3959,
+      "step": 7640
+    },
+    {
+      "epoch": 1.5047576897543704,
+      "grad_norm": 2.2921009063720703,
+      "learning_rate": 1.7863279219786603e-05,
+      "loss": 0.4512,
+      "step": 7650
+    },
+    {
+      "epoch": 1.5067246932703893,
+      "grad_norm": 0.8029620051383972,
+      "learning_rate": 1.7860221957259472e-05,
+      "loss": 0.4313,
+      "step": 7660
+    },
+    {
+      "epoch": 1.508691696786408,
+      "grad_norm": 0.9565465450286865,
+      "learning_rate": 1.7857164694732338e-05,
+      "loss": 0.4258,
+      "step": 7670
+    },
+    {
+      "epoch": 1.5106587003024268,
+      "grad_norm": 1.5121804475784302,
+      "learning_rate": 1.7854107432205207e-05,
+      "loss": 0.4782,
+      "step": 7680
+    },
+    {
+      "epoch": 1.5126257038184456,
+      "grad_norm": 0.9973114132881165,
+      "learning_rate": 1.7851050169678073e-05,
+      "loss": 0.3971,
+      "step": 7690
+    },
+    {
+      "epoch": 1.5145927073344643,
+      "grad_norm": 0.9934736490249634,
+      "learning_rate": 1.784799290715094e-05,
+      "loss": 0.4563,
+      "step": 7700
+    },
+    {
+      "epoch": 1.5165597108504831,
+      "grad_norm": 1.394128441810608,
+      "learning_rate": 1.7844935644623804e-05,
+      "loss": 0.4807,
+      "step": 7710
+    },
+    {
+      "epoch": 1.518526714366502,
+      "grad_norm": 1.5065066814422607,
+      "learning_rate": 1.784187838209667e-05,
+      "loss": 0.4271,
+      "step": 7720
+    },
+    {
+      "epoch": 1.5204937178825206,
+      "grad_norm": 1.175439476966858,
+      "learning_rate": 1.783882111956954e-05,
+      "loss": 0.5912,
+      "step": 7730
+    },
+    {
+      "epoch": 1.5224607213985395,
+      "grad_norm": 1.7413098812103271,
+      "learning_rate": 1.7835763857042405e-05,
+      "loss": 0.4099,
+      "step": 7740
+    },
+    {
+      "epoch": 1.5244277249145584,
+      "grad_norm": 2.1757400035858154,
+      "learning_rate": 1.783270659451527e-05,
+      "loss": 0.4179,
+      "step": 7750
+    },
+    {
+      "epoch": 1.526394728430577,
+      "grad_norm": 1.3303672075271606,
+      "learning_rate": 1.782964933198814e-05,
+      "loss": 0.2962,
+      "step": 7760
+    },
+    {
+      "epoch": 1.5283617319465959,
+      "grad_norm": 1.1518796682357788,
+      "learning_rate": 1.7826592069461006e-05,
+      "loss": 0.3631,
+      "step": 7770
+    },
+    {
+      "epoch": 1.5303287354626147,
+      "grad_norm": 1.4014183282852173,
+      "learning_rate": 1.782353480693387e-05,
+      "loss": 0.407,
+      "step": 7780
+    },
+    {
+      "epoch": 1.5322957389786334,
+      "grad_norm": 1.4308451414108276,
+      "learning_rate": 1.782047754440674e-05,
+      "loss": 0.5349,
+      "step": 7790
+    },
+    {
+      "epoch": 1.5342627424946522,
+      "grad_norm": 2.0231456756591797,
+      "learning_rate": 1.7817420281879607e-05,
+      "loss": 0.4317,
+      "step": 7800
+    },
+    {
+      "epoch": 1.536229746010671,
+      "grad_norm": 0.7665293216705322,
+      "learning_rate": 1.7814363019352476e-05,
+      "loss": 0.602,
+      "step": 7810
+    },
+    {
+      "epoch": 1.5381967495266897,
+      "grad_norm": 1.1104438304901123,
+      "learning_rate": 1.781130575682534e-05,
+      "loss": 0.52,
+      "step": 7820
+    },
+    {
+      "epoch": 1.5401637530427086,
+      "grad_norm": 1.0240403413772583,
+      "learning_rate": 1.7808248494298207e-05,
+      "loss": 0.476,
+      "step": 7830
+    },
+    {
+      "epoch": 1.5421307565587274,
+      "grad_norm": 0.6431864500045776,
+      "learning_rate": 1.7805191231771073e-05,
+      "loss": 0.4132,
+      "step": 7840
+    },
+    {
+      "epoch": 1.544097760074746,
+      "grad_norm": 1.3472301959991455,
+      "learning_rate": 1.780213396924394e-05,
+      "loss": 0.5485,
+      "step": 7850
+    },
+    {
+      "epoch": 1.546064763590765,
+      "grad_norm": 1.5425872802734375,
+      "learning_rate": 1.7799076706716808e-05,
+      "loss": 0.4777,
+      "step": 7860
+    },
+    {
+      "epoch": 1.5480317671067838,
+      "grad_norm": 0.8339068293571472,
+      "learning_rate": 1.7796019444189674e-05,
+      "loss": 0.6139,
+      "step": 7870
+    },
+    {
+      "epoch": 1.5499987706228024,
+      "grad_norm": 1.1242626905441284,
+      "learning_rate": 1.779296218166254e-05,
+      "loss": 0.4117,
+      "step": 7880
+    },
+    {
+      "epoch": 1.5519657741388213,
+      "grad_norm": 0.9733056426048279,
+      "learning_rate": 1.778990491913541e-05,
+      "loss": 0.4415,
+      "step": 7890
+    },
+    {
+      "epoch": 1.5539327776548402,
+      "grad_norm": 1.3571527004241943,
+      "learning_rate": 1.7786847656608274e-05,
+      "loss": 0.4764,
+      "step": 7900
+    },
+    {
+      "epoch": 1.5558997811708588,
+      "grad_norm": 2.1205756664276123,
+      "learning_rate": 1.778379039408114e-05,
+      "loss": 0.4691,
+      "step": 7910
+    },
+    {
+      "epoch": 1.5578667846868777,
+      "grad_norm": 1.7528705596923828,
+      "learning_rate": 1.778073313155401e-05,
+      "loss": 0.4773,
+      "step": 7920
+    },
+    {
+      "epoch": 1.5598337882028965,
+      "grad_norm": 1.2486058473587036,
+      "learning_rate": 1.7777675869026875e-05,
+      "loss": 0.5109,
+      "step": 7930
+    },
+    {
+      "epoch": 1.5618007917189152,
+      "grad_norm": 1.1096512079238892,
+      "learning_rate": 1.777461860649974e-05,
+      "loss": 0.2977,
+      "step": 7940
+    },
+    {
+      "epoch": 1.563767795234934,
+      "grad_norm": 1.3998329639434814,
+      "learning_rate": 1.7771561343972607e-05,
+      "loss": 0.5104,
+      "step": 7950
+    },
+    {
+      "epoch": 1.565734798750953,
+      "grad_norm": 1.1133865118026733,
+      "learning_rate": 1.7768504081445476e-05,
+      "loss": 0.4394,
+      "step": 7960
+    },
+    {
+      "epoch": 1.5677018022669715,
+      "grad_norm": 1.2691013813018799,
+      "learning_rate": 1.776544681891834e-05,
+      "loss": 0.3867,
+      "step": 7970
+    },
+    {
+      "epoch": 1.5696688057829902,
+      "grad_norm": 0.6371937990188599,
+      "learning_rate": 1.7762389556391207e-05,
+      "loss": 0.4909,
+      "step": 7980
+    },
+    {
+      "epoch": 1.5716358092990093,
+      "grad_norm": 1.969839096069336,
+      "learning_rate": 1.7759332293864077e-05,
+      "loss": 0.4578,
+      "step": 7990
+    },
+    {
+      "epoch": 1.573602812815028,
+      "grad_norm": 1.014076590538025,
+      "learning_rate": 1.7756275031336942e-05,
+      "loss": 0.4606,
+      "step": 8000
+    },
+    {
+      "epoch": 1.573602812815028,
+      "eval_loss": 0.24194316565990448,
+      "eval_runtime": 8.8643,
+      "eval_samples_per_second": 5.641,
+      "eval_steps_per_second": 2.82,
+      "step": 8000
+    },
+    {
+      "epoch": 1.5755698163310465,
+      "grad_norm": 1.0506935119628906,
+      "learning_rate": 1.7753217768809808e-05,
+      "loss": 0.4692,
+      "step": 8010
+    },
+    {
+      "epoch": 1.5775368198470656,
+      "grad_norm": 0.9741007089614868,
+      "learning_rate": 1.7750160506282677e-05,
+      "loss": 0.4414,
+      "step": 8020
+    },
+    {
+      "epoch": 1.5795038233630843,
+      "grad_norm": 1.311547875404358,
+      "learning_rate": 1.7747103243755543e-05,
+      "loss": 0.5126,
+      "step": 8030
+    },
+    {
+      "epoch": 1.581470826879103,
+      "grad_norm": 1.2044017314910889,
+      "learning_rate": 1.774404598122841e-05,
+      "loss": 0.4694,
+      "step": 8040
+    },
+    {
+      "epoch": 1.583437830395122,
+      "grad_norm": 1.728625774383545,
+      "learning_rate": 1.7740988718701275e-05,
+      "loss": 0.3995,
+      "step": 8050
+    },
+    {
+      "epoch": 1.5854048339111406,
+      "grad_norm": 0.7783991694450378,
+      "learning_rate": 1.7737931456174144e-05,
+      "loss": 0.4051,
+      "step": 8060
+    },
+    {
+      "epoch": 1.5873718374271593,
+      "grad_norm": 1.6007112264633179,
+      "learning_rate": 1.773487419364701e-05,
+      "loss": 0.3564,
+      "step": 8070
+    },
+    {
+      "epoch": 1.5893388409431783,
+      "grad_norm": 1.422080159187317,
+      "learning_rate": 1.7731816931119875e-05,
+      "loss": 0.4121,
+      "step": 8080
+    },
+    {
+      "epoch": 1.591305844459197,
+      "grad_norm": 0.9315184950828552,
+      "learning_rate": 1.7728759668592745e-05,
+      "loss": 0.5115,
+      "step": 8090
+    },
+    {
+      "epoch": 1.5932728479752156,
+      "grad_norm": 1.5067553520202637,
+      "learning_rate": 1.772570240606561e-05,
+      "loss": 0.4785,
+      "step": 8100
+    },
+    {
+      "epoch": 1.5952398514912347,
+      "grad_norm": 0.9890210628509521,
+      "learning_rate": 1.7722645143538476e-05,
+      "loss": 0.3946,
+      "step": 8110
+    },
+    {
+      "epoch": 1.5972068550072533,
+      "grad_norm": 1.6322187185287476,
+      "learning_rate": 1.7719587881011345e-05,
+      "loss": 0.3755,
+      "step": 8120
+    },
+    {
+      "epoch": 1.599173858523272,
+      "grad_norm": 1.1822806596755981,
+      "learning_rate": 1.771653061848421e-05,
+      "loss": 0.3832,
+      "step": 8130
+    },
+    {
+      "epoch": 1.6011408620392908,
+      "grad_norm": 0.8035289645195007,
+      "learning_rate": 1.7713473355957077e-05,
+      "loss": 0.3896,
+      "step": 8140
+    },
+    {
+      "epoch": 1.6031078655553097,
+      "grad_norm": 1.2947710752487183,
+      "learning_rate": 1.7710416093429946e-05,
+      "loss": 0.4154,
+      "step": 8150
+    },
+    {
+      "epoch": 1.6050748690713283,
+      "grad_norm": 1.4459134340286255,
+      "learning_rate": 1.7707358830902812e-05,
+      "loss": 0.4079,
+      "step": 8160
+    },
+    {
+      "epoch": 1.6070418725873472,
+      "grad_norm": 2.4247255325317383,
+      "learning_rate": 1.7704301568375678e-05,
+      "loss": 0.4353,
+      "step": 8170
+    },
+    {
+      "epoch": 1.609008876103366,
+      "grad_norm": 1.7483638525009155,
+      "learning_rate": 1.7701244305848543e-05,
+      "loss": 0.4049,
+      "step": 8180
+    },
+    {
+      "epoch": 1.6109758796193847,
+      "grad_norm": 1.2426694631576538,
+      "learning_rate": 1.7698187043321413e-05,
+      "loss": 0.3751,
+      "step": 8190
+    },
+    {
+      "epoch": 1.6129428831354036,
+      "grad_norm": 1.3838658332824707,
+      "learning_rate": 1.7695129780794278e-05,
+      "loss": 0.3779,
+      "step": 8200
+    },
+    {
+      "epoch": 1.6149098866514224,
+      "grad_norm": 1.3950488567352295,
+      "learning_rate": 1.7692072518267144e-05,
+      "loss": 0.4883,
+      "step": 8210
+    },
+    {
+      "epoch": 1.616876890167441,
+      "grad_norm": 1.1213529109954834,
+      "learning_rate": 1.7689015255740013e-05,
+      "loss": 0.4647,
+      "step": 8220
+    },
+    {
+      "epoch": 1.61884389368346,
+      "grad_norm": 1.6724748611450195,
+      "learning_rate": 1.768595799321288e-05,
+      "loss": 0.6121,
+      "step": 8230
+    },
+    {
+      "epoch": 1.6208108971994788,
+      "grad_norm": 1.7976254224777222,
+      "learning_rate": 1.7682900730685745e-05,
+      "loss": 0.51,
+      "step": 8240
+    },
+    {
+      "epoch": 1.6227779007154974,
+      "grad_norm": 1.0217626094818115,
+      "learning_rate": 1.7679843468158614e-05,
+      "loss": 0.3698,
+      "step": 8250
+    },
+    {
+      "epoch": 1.6247449042315163,
+      "grad_norm": 2.1370279788970947,
+      "learning_rate": 1.767678620563148e-05,
+      "loss": 0.4984,
+      "step": 8260
+    },
+    {
+      "epoch": 1.6267119077475352,
+      "grad_norm": 1.4108999967575073,
+      "learning_rate": 1.7673728943104345e-05,
+      "loss": 0.507,
+      "step": 8270
+    },
+    {
+      "epoch": 1.6286789112635538,
+      "grad_norm": 0.9152798652648926,
+      "learning_rate": 1.767067168057721e-05,
+      "loss": 0.4259,
+      "step": 8280
+    },
+    {
+      "epoch": 1.6306459147795727,
+      "grad_norm": 1.3696630001068115,
+      "learning_rate": 1.7667614418050077e-05,
+      "loss": 0.482,
+      "step": 8290
+    },
+    {
+      "epoch": 1.6326129182955915,
+      "grad_norm": 1.8380405902862549,
+      "learning_rate": 1.7664557155522946e-05,
+      "loss": 0.3949,
+      "step": 8300
+    },
+    {
+      "epoch": 1.6345799218116102,
+      "grad_norm": 1.0962872505187988,
+      "learning_rate": 1.7661499892995812e-05,
+      "loss": 0.3266,
+      "step": 8310
+    },
+    {
+      "epoch": 1.636546925327629,
+      "grad_norm": 1.6513553857803345,
+      "learning_rate": 1.765844263046868e-05,
+      "loss": 0.3018,
+      "step": 8320
+    },
+    {
+      "epoch": 1.6385139288436479,
+      "grad_norm": 2.0764846801757812,
+      "learning_rate": 1.7655385367941547e-05,
+      "loss": 0.3792,
+      "step": 8330
+    },
+    {
+      "epoch": 1.6404809323596665,
+      "grad_norm": 1.0733838081359863,
+      "learning_rate": 1.7652328105414413e-05,
+      "loss": 0.5205,
+      "step": 8340
+    },
+    {
+      "epoch": 1.6424479358756854,
+      "grad_norm": 1.3473206758499146,
+      "learning_rate": 1.7649270842887282e-05,
+      "loss": 0.4039,
+      "step": 8350
+    },
+    {
+      "epoch": 1.6444149393917042,
+      "grad_norm": 1.290519118309021,
+      "learning_rate": 1.7646213580360148e-05,
+      "loss": 0.4938,
+      "step": 8360
+    },
+    {
+      "epoch": 1.6463819429077229,
+      "grad_norm": 1.4205468893051147,
+      "learning_rate": 1.7643156317833013e-05,
+      "loss": 0.5465,
+      "step": 8370
+    },
+    {
+      "epoch": 1.6483489464237417,
+      "grad_norm": 1.6838555335998535,
+      "learning_rate": 1.7640099055305883e-05,
+      "loss": 0.343,
+      "step": 8380
+    },
+    {
+      "epoch": 1.6503159499397606,
+      "grad_norm": 1.086775302886963,
+      "learning_rate": 1.7637041792778745e-05,
+      "loss": 0.6068,
+      "step": 8390
+    },
+    {
+      "epoch": 1.6522829534557792,
+      "grad_norm": 2.012615442276001,
+      "learning_rate": 1.7633984530251614e-05,
+      "loss": 0.3777,
+      "step": 8400
+    },
+    {
+      "epoch": 1.654249956971798,
+      "grad_norm": 1.1604315042495728,
+      "learning_rate": 1.763092726772448e-05,
+      "loss": 0.4205,
+      "step": 8410
+    },
+    {
+      "epoch": 1.656216960487817,
+      "grad_norm": 1.714273452758789,
+      "learning_rate": 1.7627870005197346e-05,
+      "loss": 0.4793,
+      "step": 8420
+    },
+    {
+      "epoch": 1.6581839640038356,
+      "grad_norm": 1.1449471712112427,
+      "learning_rate": 1.7624812742670215e-05,
+      "loss": 0.4119,
+      "step": 8430
+    },
+    {
+      "epoch": 1.6601509675198545,
+      "grad_norm": 1.364147663116455,
+      "learning_rate": 1.762175548014308e-05,
+      "loss": 0.4533,
+      "step": 8440
+    },
+    {
+      "epoch": 1.6621179710358733,
+      "grad_norm": 1.7727055549621582,
+      "learning_rate": 1.761869821761595e-05,
+      "loss": 0.4756,
+      "step": 8450
+    },
+    {
+      "epoch": 1.664084974551892,
+      "grad_norm": 1.1421890258789062,
+      "learning_rate": 1.7615640955088816e-05,
+      "loss": 0.4338,
+      "step": 8460
+    },
+    {
+      "epoch": 1.6660519780679108,
+      "grad_norm": 1.056373953819275,
+      "learning_rate": 1.761258369256168e-05,
+      "loss": 0.3724,
+      "step": 8470
+    },
+    {
+      "epoch": 1.6680189815839297,
+      "grad_norm": 2.1806955337524414,
+      "learning_rate": 1.760952643003455e-05,
+      "loss": 0.5298,
+      "step": 8480
+    },
+    {
+      "epoch": 1.6699859850999483,
+      "grad_norm": 1.1213524341583252,
+      "learning_rate": 1.7606469167507416e-05,
+      "loss": 0.4571,
+      "step": 8490
+    },
+    {
+      "epoch": 1.6719529886159672,
+      "grad_norm": 1.1251106262207031,
+      "learning_rate": 1.7603411904980282e-05,
+      "loss": 0.3943,
+      "step": 8500
+    },
+    {
+      "epoch": 1.6719529886159672,
+      "eval_loss": 0.24524246156215668,
+      "eval_runtime": 8.8832,
+      "eval_samples_per_second": 5.629,
+      "eval_steps_per_second": 2.814,
+      "step": 8500
+    },
+    {
+      "epoch": 1.673919992131986,
+      "grad_norm": 2.213970184326172,
+      "learning_rate": 1.7600354642453148e-05,
+      "loss": 0.4906,
+      "step": 8510
+    },
+    {
+      "epoch": 1.6758869956480047,
+      "grad_norm": 1.1807043552398682,
+      "learning_rate": 1.7597297379926014e-05,
+      "loss": 0.4446,
+      "step": 8520
+    },
+    {
+      "epoch": 1.6778539991640236,
+      "grad_norm": 1.5438140630722046,
+      "learning_rate": 1.7594240117398883e-05,
+      "loss": 0.4429,
+      "step": 8530
+    },
+    {
+      "epoch": 1.6798210026800424,
+      "grad_norm": 1.0600336790084839,
+      "learning_rate": 1.759118285487175e-05,
+      "loss": 0.5285,
+      "step": 8540
+    },
+    {
+      "epoch": 1.681788006196061,
+      "grad_norm": 0.9455369710922241,
+      "learning_rate": 1.7588125592344614e-05,
+      "loss": 0.4963,
+      "step": 8550
+    },
+    {
+      "epoch": 1.6837550097120797,
+      "grad_norm": 1.0323309898376465,
+      "learning_rate": 1.7585068329817484e-05,
+      "loss": 0.4413,
+      "step": 8560
+    },
+    {
+      "epoch": 1.6857220132280988,
+      "grad_norm": 1.9019440412521362,
+      "learning_rate": 1.758201106729035e-05,
+      "loss": 0.4316,
+      "step": 8570
+    },
+    {
+      "epoch": 1.6876890167441174,
+      "grad_norm": 1.4282838106155396,
+      "learning_rate": 1.757895380476322e-05,
+      "loss": 0.501,
+      "step": 8580
+    },
+    {
+      "epoch": 1.689656020260136,
+      "grad_norm": 1.0056709051132202,
+      "learning_rate": 1.7575896542236084e-05,
+      "loss": 0.4711,
+      "step": 8590
+    },
+    {
+      "epoch": 1.6916230237761551,
+      "grad_norm": 1.0339857339859009,
+      "learning_rate": 1.757283927970895e-05,
+      "loss": 0.5852,
+      "step": 8600
+    },
+    {
+      "epoch": 1.6935900272921738,
+      "grad_norm": 1.7100114822387695,
+      "learning_rate": 1.756978201718182e-05,
+      "loss": 0.3511,
+      "step": 8610
+    },
+    {
+      "epoch": 1.6955570308081924,
+      "grad_norm": 0.7833712697029114,
+      "learning_rate": 1.756672475465468e-05,
+      "loss": 0.5208,
+      "step": 8620
+    },
+    {
+      "epoch": 1.6975240343242115,
+      "grad_norm": 0.9590697884559631,
+      "learning_rate": 1.756366749212755e-05,
+      "loss": 0.4618,
+      "step": 8630
+    },
+    {
+      "epoch": 1.6994910378402301,
+      "grad_norm": 1.0845881700515747,
+      "learning_rate": 1.7560610229600417e-05,
+      "loss": 0.4621,
+      "step": 8640
+    },
+    {
+      "epoch": 1.7014580413562488,
+      "grad_norm": 1.404337763786316,
+      "learning_rate": 1.7557552967073282e-05,
+      "loss": 0.4138,
+      "step": 8650
+    },
+    {
+      "epoch": 1.7034250448722679,
+      "grad_norm": 0.9446793794631958,
+      "learning_rate": 1.755449570454615e-05,
+      "loss": 0.4518,
+      "step": 8660
+    },
+    {
+      "epoch": 1.7053920483882865,
+      "grad_norm": 2.2327349185943604,
+      "learning_rate": 1.7551438442019017e-05,
+      "loss": 0.3963,
+      "step": 8670
+    },
+    {
+      "epoch": 1.7073590519043051,
+      "grad_norm": 1.1843819618225098,
+      "learning_rate": 1.7548381179491883e-05,
+      "loss": 0.386,
+      "step": 8680
+    },
+    {
+      "epoch": 1.709326055420324,
+      "grad_norm": 0.536867082118988,
+      "learning_rate": 1.7545323916964752e-05,
+      "loss": 0.4608,
+      "step": 8690
+    },
+    {
+      "epoch": 1.7112930589363429,
+      "grad_norm": 1.349477767944336,
+      "learning_rate": 1.7542266654437618e-05,
+      "loss": 0.5574,
+      "step": 8700
+    },
+    {
+      "epoch": 1.7132600624523615,
+      "grad_norm": 0.8360427618026733,
+      "learning_rate": 1.7539209391910487e-05,
+      "loss": 0.4974,
+      "step": 8710
+    },
+    {
+      "epoch": 1.7152270659683804,
+      "grad_norm": 1.3188480138778687,
+      "learning_rate": 1.7536152129383353e-05,
+      "loss": 0.4416,
+      "step": 8720
+    },
+    {
+      "epoch": 1.7171940694843992,
+      "grad_norm": 0.8572363257408142,
+      "learning_rate": 1.753309486685622e-05,
+      "loss": 0.513,
+      "step": 8730
+    },
+    {
+      "epoch": 1.7191610730004179,
+      "grad_norm": 0.8236428499221802,
+      "learning_rate": 1.7530037604329084e-05,
+      "loss": 0.4828,
+      "step": 8740
+    },
+    {
+      "epoch": 1.7211280765164367,
+      "grad_norm": 1.0772844552993774,
+      "learning_rate": 1.752698034180195e-05,
+      "loss": 0.3824,
+      "step": 8750
+    },
+    {
+      "epoch": 1.7230950800324556,
+      "grad_norm": 1.132460355758667,
+      "learning_rate": 1.752392307927482e-05,
+      "loss": 0.4269,
+      "step": 8760
+    },
+    {
+      "epoch": 1.7250620835484742,
+      "grad_norm": 1.9588900804519653,
+      "learning_rate": 1.7520865816747685e-05,
+      "loss": 0.4036,
+      "step": 8770
+    },
+    {
+      "epoch": 1.727029087064493,
+      "grad_norm": 0.9417825937271118,
+      "learning_rate": 1.751780855422055e-05,
+      "loss": 0.4554,
+      "step": 8780
+    },
+    {
+      "epoch": 1.728996090580512,
+      "grad_norm": 0.6179748773574829,
+      "learning_rate": 1.751475129169342e-05,
+      "loss": 0.4016,
+      "step": 8790
+    },
+    {
+      "epoch": 1.7309630940965306,
+      "grad_norm": 1.306562900543213,
+      "learning_rate": 1.7511694029166286e-05,
+      "loss": 0.4883,
+      "step": 8800
+    },
+    {
+      "epoch": 1.7329300976125495,
+      "grad_norm": 1.5696890354156494,
+      "learning_rate": 1.7508636766639152e-05,
+      "loss": 0.4443,
+      "step": 8810
+    },
+    {
+      "epoch": 1.7348971011285683,
+      "grad_norm": 1.0068707466125488,
+      "learning_rate": 1.750557950411202e-05,
+      "loss": 0.4914,
+      "step": 8820
+    },
+    {
+      "epoch": 1.736864104644587,
+      "grad_norm": 1.382118821144104,
+      "learning_rate": 1.7502522241584887e-05,
+      "loss": 0.3365,
+      "step": 8830
+    },
+    {
+      "epoch": 1.7388311081606058,
+      "grad_norm": 2.175328493118286,
+      "learning_rate": 1.7499464979057756e-05,
+      "loss": 0.3458,
+      "step": 8840
+    },
+    {
+      "epoch": 1.7407981116766247,
+      "grad_norm": 1.2995758056640625,
+      "learning_rate": 1.7496407716530618e-05,
+      "loss": 0.5217,
+      "step": 8850
+    },
+    {
+      "epoch": 1.7427651151926433,
+      "grad_norm": 1.4120404720306396,
+      "learning_rate": 1.7493350454003487e-05,
+      "loss": 0.3349,
+      "step": 8860
+    },
+    {
+      "epoch": 1.7447321187086622,
+      "grad_norm": 1.544440507888794,
+      "learning_rate": 1.7490293191476353e-05,
+      "loss": 0.4898,
+      "step": 8870
+    },
+    {
+      "epoch": 1.746699122224681,
+      "grad_norm": 1.823754072189331,
+      "learning_rate": 1.748723592894922e-05,
+      "loss": 0.5335,
+      "step": 8880
+    },
+    {
+      "epoch": 1.7486661257406997,
+      "grad_norm": 2.340019464492798,
+      "learning_rate": 1.7484178666422088e-05,
+      "loss": 0.4089,
+      "step": 8890
+    },
+    {
+      "epoch": 1.7506331292567185,
+      "grad_norm": 1.16437828540802,
+      "learning_rate": 1.7481121403894954e-05,
+      "loss": 0.3566,
+      "step": 8900
+    },
+    {
+      "epoch": 1.7526001327727374,
+      "grad_norm": 0.9248781800270081,
+      "learning_rate": 1.747806414136782e-05,
+      "loss": 0.4785,
+      "step": 8910
+    },
+    {
+      "epoch": 1.754567136288756,
+      "grad_norm": 0.8662049770355225,
+      "learning_rate": 1.747500687884069e-05,
+      "loss": 0.4731,
+      "step": 8920
+    },
+    {
+      "epoch": 1.756534139804775,
+      "grad_norm": 2.055873394012451,
+      "learning_rate": 1.7471949616313555e-05,
+      "loss": 0.548,
+      "step": 8930
+    },
+    {
+      "epoch": 1.7585011433207938,
+      "grad_norm": 1.322381854057312,
+      "learning_rate": 1.746889235378642e-05,
+      "loss": 0.4159,
+      "step": 8940
+    },
+    {
+      "epoch": 1.7604681468368124,
+      "grad_norm": 0.811429500579834,
+      "learning_rate": 1.746583509125929e-05,
+      "loss": 0.4281,
+      "step": 8950
+    },
+    {
+      "epoch": 1.7624351503528313,
+      "grad_norm": 1.911391258239746,
+      "learning_rate": 1.7462777828732155e-05,
+      "loss": 0.4722,
+      "step": 8960
+    },
+    {
+      "epoch": 1.7644021538688501,
+      "grad_norm": 1.6919752359390259,
+      "learning_rate": 1.745972056620502e-05,
+      "loss": 0.4453,
+      "step": 8970
+    },
+    {
+      "epoch": 1.7663691573848688,
+      "grad_norm": 0.8237192630767822,
+      "learning_rate": 1.7456663303677887e-05,
+      "loss": 0.3616,
+      "step": 8980
+    },
+    {
+      "epoch": 1.7683361609008876,
+      "grad_norm": 1.7030389308929443,
+      "learning_rate": 1.7453606041150756e-05,
+      "loss": 0.4525,
+      "step": 8990
+    },
+    {
+      "epoch": 1.7703031644169065,
+      "grad_norm": 2.035853147506714,
+      "learning_rate": 1.7450548778623622e-05,
+      "loss": 0.4024,
+      "step": 9000
+    },
+    {
+      "epoch": 1.7703031644169065,
+      "eval_loss": 0.23471477627754211,
+      "eval_runtime": 8.9179,
+      "eval_samples_per_second": 5.607,
+      "eval_steps_per_second": 2.803,
+      "step": 9000
+    },
+    {
+      "epoch": 1.7722701679329251,
+      "grad_norm": 1.2445576190948486,
+      "learning_rate": 1.7447491516096488e-05,
+      "loss": 0.4392,
+      "step": 9010
+    },
+    {
+      "epoch": 1.774237171448944,
+      "grad_norm": 2.278787612915039,
+      "learning_rate": 1.7444434253569357e-05,
+      "loss": 0.4082,
+      "step": 9020
+    },
+    {
+      "epoch": 1.7762041749649629,
+      "grad_norm": 1.5828843116760254,
+      "learning_rate": 1.7441376991042223e-05,
+      "loss": 0.4268,
+      "step": 9030
+    },
+    {
+      "epoch": 1.7781711784809815,
+      "grad_norm": 1.3761073350906372,
+      "learning_rate": 1.743831972851509e-05,
+      "loss": 0.5517,
+      "step": 9040
+    },
+    {
+      "epoch": 1.7801381819970004,
+      "grad_norm": 1.5714308023452759,
+      "learning_rate": 1.7435262465987957e-05,
+      "loss": 0.3992,
+      "step": 9050
+    },
+    {
+      "epoch": 1.7821051855130192,
+      "grad_norm": 1.2077587842941284,
+      "learning_rate": 1.7432205203460823e-05,
+      "loss": 0.5408,
+      "step": 9060
+    },
+    {
+      "epoch": 1.7840721890290379,
+      "grad_norm": 1.1043856143951416,
+      "learning_rate": 1.742914794093369e-05,
+      "loss": 0.4638,
+      "step": 9070
+    },
+    {
+      "epoch": 1.7860391925450567,
+      "grad_norm": 2.701866388320923,
+      "learning_rate": 1.7426090678406555e-05,
+      "loss": 0.47,
+      "step": 9080
+    },
+    {
+      "epoch": 1.7880061960610756,
+      "grad_norm": 0.7272081971168518,
+      "learning_rate": 1.7423033415879424e-05,
+      "loss": 0.4197,
+      "step": 9090
+    },
+    {
+      "epoch": 1.7899731995770942,
+      "grad_norm": 1.856882929801941,
+      "learning_rate": 1.741997615335229e-05,
+      "loss": 0.4344,
+      "step": 9100
+    },
+    {
+      "epoch": 1.791940203093113,
+      "grad_norm": 1.0670031309127808,
+      "learning_rate": 1.7416918890825156e-05,
+      "loss": 0.453,
+      "step": 9110
+    },
+    {
+      "epoch": 1.793907206609132,
+      "grad_norm": 1.773953914642334,
+      "learning_rate": 1.7413861628298025e-05,
+      "loss": 0.4737,
+      "step": 9120
+    },
+    {
+      "epoch": 1.7958742101251506,
+      "grad_norm": 2.5238022804260254,
+      "learning_rate": 1.741080436577089e-05,
+      "loss": 0.4638,
+      "step": 9130
+    },
+    {
+      "epoch": 1.7978412136411692,
+      "grad_norm": 1.11234450340271,
+      "learning_rate": 1.7407747103243756e-05,
+      "loss": 0.3236,
+      "step": 9140
+    },
+    {
+      "epoch": 1.7998082171571883,
+      "grad_norm": 0.8358986973762512,
+      "learning_rate": 1.7404689840716625e-05,
+      "loss": 0.5479,
+      "step": 9150
+    },
+    {
+      "epoch": 1.801775220673207,
+      "grad_norm": 1.2683533430099487,
+      "learning_rate": 1.740163257818949e-05,
+      "loss": 0.4049,
+      "step": 9160
+    },
+    {
+      "epoch": 1.8037422241892256,
+      "grad_norm": 1.2132279872894287,
+      "learning_rate": 1.7398575315662357e-05,
+      "loss": 0.4431,
+      "step": 9170
+    },
+    {
+      "epoch": 1.8057092277052447,
+      "grad_norm": 2.021444082260132,
+      "learning_rate": 1.7395518053135226e-05,
+      "loss": 0.3557,
+      "step": 9180
+    },
+    {
+      "epoch": 1.8076762312212633,
+      "grad_norm": 2.0817456245422363,
+      "learning_rate": 1.739246079060809e-05,
+      "loss": 0.4853,
+      "step": 9190
+    },
+    {
+      "epoch": 1.809643234737282,
+      "grad_norm": 1.6060495376586914,
+      "learning_rate": 1.7389403528080958e-05,
+      "loss": 0.4487,
+      "step": 9200
+    },
+    {
+      "epoch": 1.811610238253301,
+      "grad_norm": 1.1672077178955078,
+      "learning_rate": 1.7386346265553823e-05,
+      "loss": 0.4644,
+      "step": 9210
+    },
+    {
+      "epoch": 1.8135772417693197,
+      "grad_norm": 1.7596189975738525,
+      "learning_rate": 1.7383289003026693e-05,
+      "loss": 0.4474,
+      "step": 9220
+    },
+    {
+      "epoch": 1.8155442452853383,
+      "grad_norm": 1.368772268295288,
+      "learning_rate": 1.738023174049956e-05,
+      "loss": 0.3707,
+      "step": 9230
+    },
+    {
+      "epoch": 1.8175112488013574,
+      "grad_norm": 1.8463257551193237,
+      "learning_rate": 1.7377174477972424e-05,
+      "loss": 0.426,
+      "step": 9240
+    },
+    {
+      "epoch": 1.819478252317376,
+      "grad_norm": 2.007481813430786,
+      "learning_rate": 1.7374117215445293e-05,
+      "loss": 0.4587,
+      "step": 9250
+    },
+    {
+      "epoch": 1.8214452558333947,
+      "grad_norm": 1.2451355457305908,
+      "learning_rate": 1.737105995291816e-05,
+      "loss": 0.3996,
+      "step": 9260
+    },
+    {
+      "epoch": 1.8234122593494135,
+      "grad_norm": 0.9296106696128845,
+      "learning_rate": 1.7368002690391025e-05,
+      "loss": 0.5172,
+      "step": 9270
+    },
+    {
+      "epoch": 1.8253792628654324,
+      "grad_norm": 1.6007474660873413,
+      "learning_rate": 1.7364945427863894e-05,
+      "loss": 0.4304,
+      "step": 9280
+    },
+    {
+      "epoch": 1.827346266381451,
+      "grad_norm": 1.849847674369812,
+      "learning_rate": 1.736188816533676e-05,
+      "loss": 0.4836,
+      "step": 9290
+    },
+    {
+      "epoch": 1.82931326989747,
+      "grad_norm": 1.7262097597122192,
+      "learning_rate": 1.7358830902809626e-05,
+      "loss": 0.4367,
+      "step": 9300
+    },
+    {
+      "epoch": 1.8312802734134888,
+      "grad_norm": 1.2491943836212158,
+      "learning_rate": 1.735577364028249e-05,
+      "loss": 0.4727,
+      "step": 9310
+    },
+    {
+      "epoch": 1.8332472769295074,
+      "grad_norm": 0.9708894491195679,
+      "learning_rate": 1.7352716377755357e-05,
+      "loss": 0.347,
+      "step": 9320
+    },
+    {
+      "epoch": 1.8352142804455263,
+      "grad_norm": 1.439257025718689,
+      "learning_rate": 1.7349659115228226e-05,
+      "loss": 0.5331,
+      "step": 9330
+    },
+    {
+      "epoch": 1.8371812839615451,
+      "grad_norm": 1.9630393981933594,
+      "learning_rate": 1.7346601852701092e-05,
+      "loss": 0.3824,
+      "step": 9340
+    },
+    {
+      "epoch": 1.8391482874775638,
+      "grad_norm": 0.876420259475708,
+      "learning_rate": 1.734354459017396e-05,
+      "loss": 0.3851,
+      "step": 9350
+    },
+    {
+      "epoch": 1.8411152909935826,
+      "grad_norm": 1.0410975217819214,
+      "learning_rate": 1.7340487327646827e-05,
+      "loss": 0.3504,
+      "step": 9360
+    },
+    {
+      "epoch": 1.8430822945096015,
+      "grad_norm": 1.2595709562301636,
+      "learning_rate": 1.7337430065119693e-05,
+      "loss": 0.4824,
+      "step": 9370
+    },
+    {
+      "epoch": 1.8450492980256201,
+      "grad_norm": 0.9842739701271057,
+      "learning_rate": 1.7334372802592562e-05,
+      "loss": 0.4738,
+      "step": 9380
+    },
+    {
+      "epoch": 1.847016301541639,
+      "grad_norm": 0.8385689854621887,
+      "learning_rate": 1.7331315540065428e-05,
+      "loss": 0.487,
+      "step": 9390
+    },
+    {
+      "epoch": 1.8489833050576578,
+      "grad_norm": 0.9482077360153198,
+      "learning_rate": 1.7328258277538294e-05,
+      "loss": 0.5305,
+      "step": 9400
+    },
+    {
+      "epoch": 1.8509503085736765,
+      "grad_norm": 2.2385172843933105,
+      "learning_rate": 1.732520101501116e-05,
+      "loss": 0.4707,
+      "step": 9410
+    },
+    {
+      "epoch": 1.8529173120896953,
+      "grad_norm": 1.1302319765090942,
+      "learning_rate": 1.7322143752484025e-05,
+      "loss": 0.4294,
+      "step": 9420
+    },
+    {
+      "epoch": 1.8548843156057142,
+      "grad_norm": 1.530410885810852,
+      "learning_rate": 1.7319086489956894e-05,
+      "loss": 0.5544,
+      "step": 9430
+    },
+    {
+      "epoch": 1.8568513191217328,
+      "grad_norm": 0.8635900020599365,
+      "learning_rate": 1.731602922742976e-05,
+      "loss": 0.4827,
+      "step": 9440
+    },
+    {
+      "epoch": 1.8588183226377517,
+      "grad_norm": 1.119480848312378,
+      "learning_rate": 1.7312971964902626e-05,
+      "loss": 0.3917,
+      "step": 9450
+    },
+    {
+      "epoch": 1.8607853261537706,
+      "grad_norm": 1.0276856422424316,
+      "learning_rate": 1.7309914702375495e-05,
+      "loss": 0.3687,
+      "step": 9460
+    },
+    {
+      "epoch": 1.8627523296697892,
+      "grad_norm": 1.7624456882476807,
+      "learning_rate": 1.730685743984836e-05,
+      "loss": 0.4276,
+      "step": 9470
+    },
+    {
+      "epoch": 1.864719333185808,
+      "grad_norm": 1.3257324695587158,
+      "learning_rate": 1.730380017732123e-05,
+      "loss": 0.3444,
+      "step": 9480
+    },
+    {
+      "epoch": 1.866686336701827,
+      "grad_norm": 1.4196683168411255,
+      "learning_rate": 1.7300742914794096e-05,
+      "loss": 0.3877,
+      "step": 9490
+    },
+    {
+      "epoch": 1.8686533402178456,
+      "grad_norm": 1.4340052604675293,
+      "learning_rate": 1.729768565226696e-05,
+      "loss": 0.3258,
+      "step": 9500
+    },
+    {
+      "epoch": 1.8686533402178456,
+      "eval_loss": 0.21660512685775757,
+      "eval_runtime": 8.8832,
+      "eval_samples_per_second": 5.629,
+      "eval_steps_per_second": 2.814,
+      "step": 9500
+    },
+    {
+      "epoch": 1.8706203437338644,
+      "grad_norm": 1.190011978149414,
+      "learning_rate": 1.729462838973983e-05,
+      "loss": 0.3835,
+      "step": 9510
+    },
+    {
+      "epoch": 1.8725873472498833,
+      "grad_norm": 1.5729475021362305,
+      "learning_rate": 1.7291571127212696e-05,
+      "loss": 0.3737,
+      "step": 9520
+    },
+    {
+      "epoch": 1.874554350765902,
+      "grad_norm": 1.9554622173309326,
+      "learning_rate": 1.7288513864685562e-05,
+      "loss": 0.5472,
+      "step": 9530
+    },
+    {
+      "epoch": 1.8765213542819208,
+      "grad_norm": 1.2588348388671875,
+      "learning_rate": 1.7285456602158428e-05,
+      "loss": 0.3402,
+      "step": 9540
+    },
+    {
+      "epoch": 1.8784883577979397,
+      "grad_norm": 1.588701844215393,
+      "learning_rate": 1.7282399339631294e-05,
+      "loss": 0.5021,
+      "step": 9550
+    },
+    {
+      "epoch": 1.8804553613139583,
+      "grad_norm": 1.0969479084014893,
+      "learning_rate": 1.7279342077104163e-05,
+      "loss": 0.4521,
+      "step": 9560
+    },
+    {
+      "epoch": 1.8824223648299772,
+      "grad_norm": 0.828027606010437,
+      "learning_rate": 1.727628481457703e-05,
+      "loss": 0.5221,
+      "step": 9570
+    },
+    {
+      "epoch": 1.884389368345996,
+      "grad_norm": 0.9304088950157166,
+      "learning_rate": 1.7273227552049895e-05,
+      "loss": 0.4416,
+      "step": 9580
+    },
+    {
+      "epoch": 1.8863563718620147,
+      "grad_norm": 1.506238341331482,
+      "learning_rate": 1.7270170289522764e-05,
+      "loss": 0.2849,
+      "step": 9590
+    },
+    {
+      "epoch": 1.8883233753780335,
+      "grad_norm": 2.2380335330963135,
+      "learning_rate": 1.726711302699563e-05,
+      "loss": 0.3606,
+      "step": 9600
+    },
+    {
+      "epoch": 1.8902903788940524,
+      "grad_norm": 1.978633165359497,
+      "learning_rate": 1.72640557644685e-05,
+      "loss": 0.4753,
+      "step": 9610
+    },
+    {
+      "epoch": 1.892257382410071,
+      "grad_norm": 2.0024452209472656,
+      "learning_rate": 1.7260998501941364e-05,
+      "loss": 0.43,
+      "step": 9620
+    },
+    {
+      "epoch": 1.8942243859260899,
+      "grad_norm": 1.901621699333191,
+      "learning_rate": 1.725794123941423e-05,
+      "loss": 0.3712,
+      "step": 9630
+    },
+    {
+      "epoch": 1.8961913894421087,
+      "grad_norm": 1.4384682178497314,
+      "learning_rate": 1.7254883976887096e-05,
+      "loss": 0.4753,
+      "step": 9640
+    },
+    {
+      "epoch": 1.8981583929581274,
+      "grad_norm": 1.5139740705490112,
+      "learning_rate": 1.7251826714359962e-05,
+      "loss": 0.5545,
+      "step": 9650
+    },
+    {
+      "epoch": 1.9001253964741462,
+      "grad_norm": 1.0212291479110718,
+      "learning_rate": 1.724876945183283e-05,
+      "loss": 0.3801,
+      "step": 9660
+    },
+    {
+      "epoch": 1.902092399990165,
+      "grad_norm": 0.6466912031173706,
+      "learning_rate": 1.7245712189305697e-05,
+      "loss": 0.4541,
+      "step": 9670
+    },
+    {
+      "epoch": 1.9040594035061837,
+      "grad_norm": 1.1486016511917114,
+      "learning_rate": 1.7242654926778562e-05,
+      "loss": 0.3891,
+      "step": 9680
+    },
+    {
+      "epoch": 1.9060264070222026,
+      "grad_norm": 1.5629327297210693,
+      "learning_rate": 1.723959766425143e-05,
+      "loss": 0.4214,
+      "step": 9690
+    },
+    {
+      "epoch": 1.9079934105382215,
+      "grad_norm": 1.9606152772903442,
+      "learning_rate": 1.7236540401724297e-05,
+      "loss": 0.4845,
+      "step": 9700
+    },
+    {
+      "epoch": 1.90996041405424,
+      "grad_norm": 1.438989281654358,
+      "learning_rate": 1.7233483139197163e-05,
+      "loss": 0.399,
+      "step": 9710
+    },
+    {
+      "epoch": 1.9119274175702587,
+      "grad_norm": 1.5821136236190796,
+      "learning_rate": 1.7230425876670032e-05,
+      "loss": 0.4228,
+      "step": 9720
+    },
+    {
+      "epoch": 1.9138944210862778,
+      "grad_norm": 1.7703495025634766,
+      "learning_rate": 1.7227368614142898e-05,
+      "loss": 0.5136,
+      "step": 9730
+    },
+    {
+      "epoch": 1.9158614246022965,
+      "grad_norm": 0.7840451598167419,
+      "learning_rate": 1.7224311351615767e-05,
+      "loss": 0.5638,
+      "step": 9740
+    },
+    {
+      "epoch": 1.917828428118315,
+      "grad_norm": 1.6046384572982788,
+      "learning_rate": 1.722125408908863e-05,
+      "loss": 0.5003,
+      "step": 9750
+    },
+    {
+      "epoch": 1.9197954316343342,
+      "grad_norm": 1.572587013244629,
+      "learning_rate": 1.72181968265615e-05,
+      "loss": 0.4326,
+      "step": 9760
+    },
+    {
+      "epoch": 1.9217624351503528,
+      "grad_norm": 1.9076180458068848,
+      "learning_rate": 1.7215139564034365e-05,
+      "loss": 0.4825,
+      "step": 9770
+    },
+    {
+      "epoch": 1.9237294386663715,
+      "grad_norm": 0.769214928150177,
+      "learning_rate": 1.721208230150723e-05,
+      "loss": 0.4463,
+      "step": 9780
+    },
+    {
+      "epoch": 1.9256964421823906,
+      "grad_norm": 1.7923251390457153,
+      "learning_rate": 1.72090250389801e-05,
+      "loss": 0.4338,
+      "step": 9790
+    },
+    {
+      "epoch": 1.9276634456984092,
+      "grad_norm": 1.6408928632736206,
+      "learning_rate": 1.7205967776452965e-05,
+      "loss": 0.534,
+      "step": 9800
+    },
+    {
+      "epoch": 1.9296304492144278,
+      "grad_norm": 0.9936132431030273,
+      "learning_rate": 1.720291051392583e-05,
+      "loss": 0.5031,
+      "step": 9810
+    },
+    {
+      "epoch": 1.931597452730447,
+      "grad_norm": 1.5383307933807373,
+      "learning_rate": 1.71998532513987e-05,
+      "loss": 0.4663,
+      "step": 9820
+    },
+    {
+      "epoch": 1.9335644562464656,
+      "grad_norm": 1.4885003566741943,
+      "learning_rate": 1.7196795988871566e-05,
+      "loss": 0.3868,
+      "step": 9830
+    },
+    {
+      "epoch": 1.9355314597624842,
+      "grad_norm": 1.8248869180679321,
+      "learning_rate": 1.7193738726344432e-05,
+      "loss": 0.6038,
+      "step": 9840
+    },
+    {
+      "epoch": 1.937498463278503,
+      "grad_norm": 0.8678923845291138,
+      "learning_rate": 1.71906814638173e-05,
+      "loss": 0.5078,
+      "step": 9850
+    },
+    {
+      "epoch": 1.939465466794522,
+      "grad_norm": 1.997922420501709,
+      "learning_rate": 1.7187624201290167e-05,
+      "loss": 0.4789,
+      "step": 9860
+    },
+    {
+      "epoch": 1.9414324703105406,
+      "grad_norm": 1.5485014915466309,
+      "learning_rate": 1.7184566938763033e-05,
+      "loss": 0.5152,
+      "step": 9870
+    },
+    {
+      "epoch": 1.9433994738265594,
+      "grad_norm": 1.3265610933303833,
+      "learning_rate": 1.71815096762359e-05,
+      "loss": 0.4607,
+      "step": 9880
+    },
+    {
+      "epoch": 1.9453664773425783,
+      "grad_norm": 0.9985576868057251,
+      "learning_rate": 1.7178452413708768e-05,
+      "loss": 0.4812,
+      "step": 9890
+    },
+    {
+      "epoch": 1.947333480858597,
+      "grad_norm": 1.0717341899871826,
+      "learning_rate": 1.7175395151181633e-05,
+      "loss": 0.3841,
+      "step": 9900
+    },
+    {
+      "epoch": 1.9493004843746158,
+      "grad_norm": 2.604396343231201,
+      "learning_rate": 1.71723378886545e-05,
+      "loss": 0.5168,
+      "step": 9910
+    },
+    {
+      "epoch": 1.9512674878906346,
+      "grad_norm": 1.9883261919021606,
+      "learning_rate": 1.7169280626127368e-05,
+      "loss": 0.4564,
+      "step": 9920
+    },
+    {
+      "epoch": 1.9532344914066533,
+      "grad_norm": 1.710569977760315,
+      "learning_rate": 1.7166223363600234e-05,
+      "loss": 0.4084,
+      "step": 9930
+    },
+    {
+      "epoch": 1.9552014949226721,
+      "grad_norm": 1.1840331554412842,
+      "learning_rate": 1.71631661010731e-05,
+      "loss": 0.517,
+      "step": 9940
+    },
+    {
+      "epoch": 1.957168498438691,
+      "grad_norm": 1.052003264427185,
+      "learning_rate": 1.716010883854597e-05,
+      "loss": 0.5016,
+      "step": 9950
+    },
+    {
+      "epoch": 1.9591355019547096,
+      "grad_norm": 2.5739831924438477,
+      "learning_rate": 1.7157051576018835e-05,
+      "loss": 0.4217,
+      "step": 9960
+    },
+    {
+      "epoch": 1.9611025054707285,
+      "grad_norm": 1.5411380529403687,
+      "learning_rate": 1.71539943134917e-05,
+      "loss": 0.3677,
+      "step": 9970
+    },
+    {
+      "epoch": 1.9630695089867474,
+      "grad_norm": 1.0582150220870972,
+      "learning_rate": 1.7150937050964566e-05,
+      "loss": 0.4218,
+      "step": 9980
+    },
+    {
+      "epoch": 1.965036512502766,
+      "grad_norm": 1.0257982015609741,
+      "learning_rate": 1.7147879788437432e-05,
+      "loss": 0.4642,
+      "step": 9990
+    },
+    {
+      "epoch": 1.9670035160187849,
+      "grad_norm": 1.5949499607086182,
+      "learning_rate": 1.71448225259103e-05,
+      "loss": 0.4337,
+      "step": 10000
+    },
+    {
+      "epoch": 1.9670035160187849,
+      "eval_loss": 0.2338702380657196,
+      "eval_runtime": 8.8698,
+      "eval_samples_per_second": 5.637,
+      "eval_steps_per_second": 2.819,
+      "step": 10000
+    },
+    {
+      "epoch": 1.9689705195348037,
+      "grad_norm": 1.4009411334991455,
+      "learning_rate": 1.7141765263383167e-05,
+      "loss": 0.433,
+      "step": 10010
+    },
+    {
+      "epoch": 1.9709375230508224,
+      "grad_norm": 1.0874663591384888,
+      "learning_rate": 1.7138708000856036e-05,
+      "loss": 0.3726,
+      "step": 10020
+    },
+    {
+      "epoch": 1.9729045265668412,
+      "grad_norm": 1.939491868019104,
+      "learning_rate": 1.7135650738328902e-05,
+      "loss": 0.479,
+      "step": 10030
+    },
+    {
+      "epoch": 1.97487153008286,
+      "grad_norm": 1.9109244346618652,
+      "learning_rate": 1.7132593475801768e-05,
+      "loss": 0.4814,
+      "step": 10040
+    },
+    {
+      "epoch": 1.9768385335988787,
+      "grad_norm": 1.6677470207214355,
+      "learning_rate": 1.7129536213274637e-05,
+      "loss": 0.3483,
+      "step": 10050
+    },
+    {
+      "epoch": 1.9788055371148976,
+      "grad_norm": 0.8189138770103455,
+      "learning_rate": 1.7126478950747503e-05,
+      "loss": 0.5348,
+      "step": 10060
+    },
+    {
+      "epoch": 1.9807725406309165,
+      "grad_norm": 2.4953317642211914,
+      "learning_rate": 1.712342168822037e-05,
+      "loss": 0.4601,
+      "step": 10070
+    },
+    {
+      "epoch": 1.982739544146935,
+      "grad_norm": 1.0296778678894043,
+      "learning_rate": 1.7120364425693238e-05,
+      "loss": 0.387,
+      "step": 10080
+    },
+    {
+      "epoch": 1.984706547662954,
+      "grad_norm": 1.6703166961669922,
+      "learning_rate": 1.71173071631661e-05,
+      "loss": 0.4055,
+      "step": 10090
+    },
+    {
+      "epoch": 1.9866735511789728,
+      "grad_norm": 1.5256836414337158,
+      "learning_rate": 1.711424990063897e-05,
+      "loss": 0.5053,
+      "step": 10100
+    },
+    {
+      "epoch": 1.9886405546949915,
+      "grad_norm": 1.264963984489441,
+      "learning_rate": 1.7111192638111835e-05,
+      "loss": 0.5051,
+      "step": 10110
+    },
+    {
+      "epoch": 1.9906075582110103,
+      "grad_norm": 0.8529186248779297,
+      "learning_rate": 1.71081353755847e-05,
+      "loss": 0.5238,
+      "step": 10120
+    },
+    {
+      "epoch": 1.9925745617270292,
+      "grad_norm": 1.695892333984375,
+      "learning_rate": 1.710507811305757e-05,
+      "loss": 0.3558,
+      "step": 10130
+    },
+    {
+      "epoch": 1.9945415652430478,
+      "grad_norm": 1.855906367301941,
+      "learning_rate": 1.7102020850530436e-05,
+      "loss": 0.4037,
+      "step": 10140
+    },
+    {
+      "epoch": 1.9965085687590667,
+      "grad_norm": 2.48172926902771,
+      "learning_rate": 1.7098963588003305e-05,
+      "loss": 0.5684,
+      "step": 10150
+    },
+    {
+      "epoch": 1.9984755722750855,
+      "grad_norm": 2.117180585861206,
+      "learning_rate": 1.709590632547617e-05,
+      "loss": 0.426,
+      "step": 10160
+    },
+    {
+      "epoch": 2.000442575791104,
+      "grad_norm": 1.000969409942627,
+      "learning_rate": 1.7092849062949036e-05,
+      "loss": 0.366,
+      "step": 10170
+    },
+    {
+      "epoch": 2.002409579307123,
+      "grad_norm": 1.5670902729034424,
+      "learning_rate": 1.7089791800421906e-05,
+      "loss": 0.3538,
+      "step": 10180
+    },
+    {
+      "epoch": 2.004376582823142,
+      "grad_norm": 2.4152426719665527,
+      "learning_rate": 1.708673453789477e-05,
+      "loss": 0.3628,
+      "step": 10190
+    },
+    {
+      "epoch": 2.0063435863391605,
+      "grad_norm": 1.6392732858657837,
+      "learning_rate": 1.7083677275367637e-05,
+      "loss": 0.4825,
+      "step": 10200
+    },
+    {
+      "epoch": 2.008310589855179,
+      "grad_norm": 1.2263078689575195,
+      "learning_rate": 1.7080620012840503e-05,
+      "loss": 0.3793,
+      "step": 10210
+    },
+    {
+      "epoch": 2.0102775933711983,
+      "grad_norm": 0.8084795475006104,
+      "learning_rate": 1.707756275031337e-05,
+      "loss": 0.4204,
+      "step": 10220
+    },
+    {
+      "epoch": 2.012244596887217,
+      "grad_norm": 1.0626640319824219,
+      "learning_rate": 1.7074505487786238e-05,
+      "loss": 0.4433,
+      "step": 10230
+    },
+    {
+      "epoch": 2.0142116004032355,
+      "grad_norm": 1.165838360786438,
+      "learning_rate": 1.7071448225259104e-05,
+      "loss": 0.4568,
+      "step": 10240
+    },
+    {
+      "epoch": 2.0161786039192546,
+      "grad_norm": 1.4117451906204224,
+      "learning_rate": 1.706839096273197e-05,
+      "loss": 0.4427,
+      "step": 10250
+    },
+    {
+      "epoch": 2.0181456074352733,
+      "grad_norm": 1.1329997777938843,
+      "learning_rate": 1.706533370020484e-05,
+      "loss": 0.4228,
+      "step": 10260
+    },
+    {
+      "epoch": 2.020112610951292,
+      "grad_norm": 2.336711883544922,
+      "learning_rate": 1.7062276437677704e-05,
+      "loss": 0.4538,
+      "step": 10270
+    },
+    {
+      "epoch": 2.022079614467311,
+      "grad_norm": 0.8653255701065063,
+      "learning_rate": 1.7059219175150574e-05,
+      "loss": 0.432,
+      "step": 10280
+    },
+    {
+      "epoch": 2.0240466179833296,
+      "grad_norm": 1.8122618198394775,
+      "learning_rate": 1.705616191262344e-05,
+      "loss": 0.4071,
+      "step": 10290
+    },
+    {
+      "epoch": 2.0260136214993483,
+      "grad_norm": 0.9676852226257324,
+      "learning_rate": 1.7053104650096305e-05,
+      "loss": 0.4196,
+      "step": 10300
+    },
+    {
+      "epoch": 2.0279806250153674,
+      "grad_norm": 0.8326351046562195,
+      "learning_rate": 1.7050047387569174e-05,
+      "loss": 0.41,
+      "step": 10310
+    },
+    {
+      "epoch": 2.029947628531386,
+      "grad_norm": 0.9962462782859802,
+      "learning_rate": 1.7046990125042037e-05,
+      "loss": 0.4066,
+      "step": 10320
+    },
+    {
+      "epoch": 2.0319146320474046,
+      "grad_norm": 2.096683979034424,
+      "learning_rate": 1.7043932862514906e-05,
+      "loss": 0.4446,
+      "step": 10330
+    },
+    {
+      "epoch": 2.0338816355634237,
+      "grad_norm": 1.3960990905761719,
+      "learning_rate": 1.704087559998777e-05,
+      "loss": 0.4261,
+      "step": 10340
+    },
+    {
+      "epoch": 2.0358486390794424,
+      "grad_norm": 1.7559967041015625,
+      "learning_rate": 1.7037818337460637e-05,
+      "loss": 0.4218,
+      "step": 10350
+    },
+    {
+      "epoch": 2.037815642595461,
+      "grad_norm": 1.368927240371704,
+      "learning_rate": 1.7034761074933507e-05,
+      "loss": 0.3458,
+      "step": 10360
+    },
+    {
+      "epoch": 2.03978264611148,
+      "grad_norm": 1.6407620906829834,
+      "learning_rate": 1.7031703812406372e-05,
+      "loss": 0.5405,
+      "step": 10370
+    },
+    {
+      "epoch": 2.0417496496274987,
+      "grad_norm": 1.106787085533142,
+      "learning_rate": 1.7028646549879238e-05,
+      "loss": 0.5497,
+      "step": 10380
+    },
+    {
+      "epoch": 2.0437166531435174,
+      "grad_norm": 1.9294019937515259,
+      "learning_rate": 1.7025589287352107e-05,
+      "loss": 0.5084,
+      "step": 10390
+    },
+    {
+      "epoch": 2.0456836566595364,
+      "grad_norm": 0.8011900186538696,
+      "learning_rate": 1.7022532024824973e-05,
+      "loss": 0.5933,
+      "step": 10400
+    },
+    {
+      "epoch": 2.047650660175555,
+      "grad_norm": 1.3763272762298584,
+      "learning_rate": 1.7019474762297842e-05,
+      "loss": 0.4236,
+      "step": 10410
+    },
+    {
+      "epoch": 2.0496176636915737,
+      "grad_norm": 1.301306128501892,
+      "learning_rate": 1.7016417499770708e-05,
+      "loss": 0.4095,
+      "step": 10420
+    },
+    {
+      "epoch": 2.051584667207593,
+      "grad_norm": 2.0290560722351074,
+      "learning_rate": 1.7013360237243574e-05,
+      "loss": 0.4528,
+      "step": 10430
+    },
+    {
+      "epoch": 2.0535516707236114,
+      "grad_norm": 0.6603031754493713,
+      "learning_rate": 1.701030297471644e-05,
+      "loss": 0.4511,
+      "step": 10440
+    },
+    {
+      "epoch": 2.05551867423963,
+      "grad_norm": 1.3809963464736938,
+      "learning_rate": 1.7007245712189305e-05,
+      "loss": 0.4434,
+      "step": 10450
+    },
+    {
+      "epoch": 2.057485677755649,
+      "grad_norm": 1.5114200115203857,
+      "learning_rate": 1.7004188449662174e-05,
+      "loss": 0.5448,
+      "step": 10460
+    },
+    {
+      "epoch": 2.059452681271668,
+      "grad_norm": 2.5838265419006348,
+      "learning_rate": 1.700113118713504e-05,
+      "loss": 0.4025,
+      "step": 10470
+    },
+    {
+      "epoch": 2.0614196847876864,
+      "grad_norm": 1.1007928848266602,
+      "learning_rate": 1.6998073924607906e-05,
+      "loss": 0.4732,
+      "step": 10480
+    },
+    {
+      "epoch": 2.0633866883037055,
+      "grad_norm": 1.4720772504806519,
+      "learning_rate": 1.6995016662080775e-05,
+      "loss": 0.4076,
+      "step": 10490
+    },
+    {
+      "epoch": 2.065353691819724,
+      "grad_norm": 1.3407565355300903,
+      "learning_rate": 1.699195939955364e-05,
+      "loss": 0.3955,
+      "step": 10500
+    },
+    {
+      "epoch": 2.065353691819724,
+      "eval_loss": 0.22142630815505981,
+      "eval_runtime": 8.865,
+      "eval_samples_per_second": 5.64,
+      "eval_steps_per_second": 2.82,
+      "step": 10500
+    },
+    {
+      "epoch": 2.067320695335743,
+      "grad_norm": 1.8334215879440308,
+      "learning_rate": 1.6988902137026507e-05,
+      "loss": 0.4093,
+      "step": 10510
+    },
+    {
+      "epoch": 2.069287698851762,
+      "grad_norm": 1.8577845096588135,
+      "learning_rate": 1.6985844874499376e-05,
+      "loss": 0.344,
+      "step": 10520
+    },
+    {
+      "epoch": 2.0712547023677805,
+      "grad_norm": 1.6269792318344116,
+      "learning_rate": 1.698278761197224e-05,
+      "loss": 0.3725,
+      "step": 10530
+    },
+    {
+      "epoch": 2.073221705883799,
+      "grad_norm": 2.4148001670837402,
+      "learning_rate": 1.697973034944511e-05,
+      "loss": 0.347,
+      "step": 10540
+    },
+    {
+      "epoch": 2.0751887093998183,
+      "grad_norm": 2.106750965118408,
+      "learning_rate": 1.6976673086917973e-05,
+      "loss": 0.4462,
+      "step": 10550
+    },
+    {
+      "epoch": 2.077155712915837,
+      "grad_norm": 1.6390737295150757,
+      "learning_rate": 1.6973615824390842e-05,
+      "loss": 0.44,
+      "step": 10560
+    },
+    {
+      "epoch": 2.0791227164318555,
+      "grad_norm": 1.192014455795288,
+      "learning_rate": 1.6970558561863708e-05,
+      "loss": 0.4548,
+      "step": 10570
+    },
+    {
+      "epoch": 2.0810897199478746,
+      "grad_norm": 0.9265616536140442,
+      "learning_rate": 1.6967501299336574e-05,
+      "loss": 0.5047,
+      "step": 10580
+    },
+    {
+      "epoch": 2.0830567234638933,
+      "grad_norm": 1.0966876745224,
+      "learning_rate": 1.6964444036809443e-05,
+      "loss": 0.3367,
+      "step": 10590
+    },
+    {
+      "epoch": 2.085023726979912,
+      "grad_norm": 1.5295296907424927,
+      "learning_rate": 1.696138677428231e-05,
+      "loss": 0.3238,
+      "step": 10600
+    },
+    {
+      "epoch": 2.0869907304959305,
+      "grad_norm": 1.5133509635925293,
+      "learning_rate": 1.6958329511755175e-05,
+      "loss": 0.5125,
+      "step": 10610
+    },
+    {
+      "epoch": 2.0889577340119496,
+      "grad_norm": 1.9635529518127441,
+      "learning_rate": 1.6955272249228044e-05,
+      "loss": 0.3324,
+      "step": 10620
+    },
+    {
+      "epoch": 2.0909247375279683,
+      "grad_norm": 1.3140554428100586,
+      "learning_rate": 1.695221498670091e-05,
+      "loss": 0.3871,
+      "step": 10630
+    },
+    {
+      "epoch": 2.092891741043987,
+      "grad_norm": 1.486924171447754,
+      "learning_rate": 1.6949157724173775e-05,
+      "loss": 0.3921,
+      "step": 10640
+    },
+    {
+      "epoch": 2.094858744560006,
+      "grad_norm": 1.5189197063446045,
+      "learning_rate": 1.6946100461646645e-05,
+      "loss": 0.3921,
+      "step": 10650
+    },
+    {
+      "epoch": 2.0968257480760246,
+      "grad_norm": 2.586416006088257,
+      "learning_rate": 1.694304319911951e-05,
+      "loss": 0.4417,
+      "step": 10660
+    },
+    {
+      "epoch": 2.0987927515920433,
+      "grad_norm": 1.2350314855575562,
+      "learning_rate": 1.6939985936592376e-05,
+      "loss": 0.4176,
+      "step": 10670
+    },
+    {
+      "epoch": 2.1007597551080623,
+      "grad_norm": 1.220737099647522,
+      "learning_rate": 1.6936928674065242e-05,
+      "loss": 0.4954,
+      "step": 10680
+    },
+    {
+      "epoch": 2.102726758624081,
+      "grad_norm": 0.970892608165741,
+      "learning_rate": 1.693387141153811e-05,
+      "loss": 0.3041,
+      "step": 10690
+    },
+    {
+      "epoch": 2.1046937621400996,
+      "grad_norm": 1.9684553146362305,
+      "learning_rate": 1.6930814149010977e-05,
+      "loss": 0.4289,
+      "step": 10700
+    },
+    {
+      "epoch": 2.1066607656561187,
+      "grad_norm": 1.4467488527297974,
+      "learning_rate": 1.6927756886483843e-05,
+      "loss": 0.4797,
+      "step": 10710
+    },
+    {
+      "epoch": 2.1086277691721373,
+      "grad_norm": 1.9098058938980103,
+      "learning_rate": 1.6924699623956712e-05,
+      "loss": 0.3886,
+      "step": 10720
+    },
+    {
+      "epoch": 2.110594772688156,
+      "grad_norm": 1.4749529361724854,
+      "learning_rate": 1.6921642361429578e-05,
+      "loss": 0.3511,
+      "step": 10730
+    },
+    {
+      "epoch": 2.112561776204175,
+      "grad_norm": 1.6586591005325317,
+      "learning_rate": 1.6918585098902443e-05,
+      "loss": 0.4309,
+      "step": 10740
+    },
+    {
+      "epoch": 2.1145287797201937,
+      "grad_norm": 1.833769679069519,
+      "learning_rate": 1.6915527836375313e-05,
+      "loss": 0.4512,
+      "step": 10750
+    },
+    {
+      "epoch": 2.1164957832362123,
+      "grad_norm": 0.7013012766838074,
+      "learning_rate": 1.6912470573848178e-05,
+      "loss": 0.3429,
+      "step": 10760
+    },
+    {
+      "epoch": 2.1184627867522314,
+      "grad_norm": 0.8748033046722412,
+      "learning_rate": 1.6909413311321044e-05,
+      "loss": 0.4688,
+      "step": 10770
+    },
+    {
+      "epoch": 2.12042979026825,
+      "grad_norm": 1.659879207611084,
+      "learning_rate": 1.690635604879391e-05,
+      "loss": 0.4047,
+      "step": 10780
+    },
+    {
+      "epoch": 2.1223967937842687,
+      "grad_norm": 1.2329697608947754,
+      "learning_rate": 1.690329878626678e-05,
+      "loss": 0.4941,
+      "step": 10790
+    },
+    {
+      "epoch": 2.124363797300288,
+      "grad_norm": 1.6511222124099731,
+      "learning_rate": 1.6900241523739645e-05,
+      "loss": 0.32,
+      "step": 10800
+    },
+    {
+      "epoch": 2.1263308008163064,
+      "grad_norm": 2.3250954151153564,
+      "learning_rate": 1.689718426121251e-05,
+      "loss": 0.4237,
+      "step": 10810
+    },
+    {
+      "epoch": 2.128297804332325,
+      "grad_norm": 1.3927966356277466,
+      "learning_rate": 1.689412699868538e-05,
+      "loss": 0.3225,
+      "step": 10820
+    },
+    {
+      "epoch": 2.130264807848344,
+      "grad_norm": 1.1719884872436523,
+      "learning_rate": 1.6891069736158246e-05,
+      "loss": 0.4178,
+      "step": 10830
+    },
+    {
+      "epoch": 2.132231811364363,
+      "grad_norm": 0.8177443742752075,
+      "learning_rate": 1.688801247363111e-05,
+      "loss": 0.455,
+      "step": 10840
+    },
+    {
+      "epoch": 2.1341988148803814,
+      "grad_norm": 2.0233986377716064,
+      "learning_rate": 1.688495521110398e-05,
+      "loss": 0.4306,
+      "step": 10850
+    },
+    {
+      "epoch": 2.1361658183964005,
+      "grad_norm": 2.01068377494812,
+      "learning_rate": 1.6881897948576846e-05,
+      "loss": 0.4518,
+      "step": 10860
+    },
+    {
+      "epoch": 2.138132821912419,
+      "grad_norm": 1.5686466693878174,
+      "learning_rate": 1.6878840686049712e-05,
+      "loss": 0.4131,
+      "step": 10870
+    },
+    {
+      "epoch": 2.140099825428438,
+      "grad_norm": 1.2859925031661987,
+      "learning_rate": 1.687578342352258e-05,
+      "loss": 0.4335,
+      "step": 10880
+    },
+    {
+      "epoch": 2.142066828944457,
+      "grad_norm": 1.372796893119812,
+      "learning_rate": 1.6872726160995444e-05,
+      "loss": 0.4711,
+      "step": 10890
+    },
+    {
+      "epoch": 2.1440338324604755,
+      "grad_norm": 1.5391151905059814,
+      "learning_rate": 1.6869668898468313e-05,
+      "loss": 0.4958,
+      "step": 10900
+    },
+    {
+      "epoch": 2.146000835976494,
+      "grad_norm": 1.3352559804916382,
+      "learning_rate": 1.686661163594118e-05,
+      "loss": 0.6165,
+      "step": 10910
+    },
+    {
+      "epoch": 2.1479678394925132,
+      "grad_norm": 2.093535900115967,
+      "learning_rate": 1.6863554373414048e-05,
+      "loss": 0.4023,
+      "step": 10920
+    },
+    {
+      "epoch": 2.149934843008532,
+      "grad_norm": 1.734489917755127,
+      "learning_rate": 1.6860497110886913e-05,
+      "loss": 0.4107,
+      "step": 10930
+    },
+    {
+      "epoch": 2.1519018465245505,
+      "grad_norm": 1.1061903238296509,
+      "learning_rate": 1.685743984835978e-05,
+      "loss": 0.387,
+      "step": 10940
+    },
+    {
+      "epoch": 2.1538688500405696,
+      "grad_norm": 0.8513095378875732,
+      "learning_rate": 1.685438258583265e-05,
+      "loss": 0.3447,
+      "step": 10950
+    },
+    {
+      "epoch": 2.1558358535565882,
+      "grad_norm": 1.3149404525756836,
+      "learning_rate": 1.6851325323305514e-05,
+      "loss": 0.4679,
+      "step": 10960
+    },
+    {
+      "epoch": 2.157802857072607,
+      "grad_norm": 0.6072118282318115,
+      "learning_rate": 1.684826806077838e-05,
+      "loss": 0.4193,
+      "step": 10970
+    },
+    {
+      "epoch": 2.159769860588626,
+      "grad_norm": 1.0672342777252197,
+      "learning_rate": 1.684521079825125e-05,
+      "loss": 0.4421,
+      "step": 10980
+    },
+    {
+      "epoch": 2.1617368641046446,
+      "grad_norm": 1.4868695735931396,
+      "learning_rate": 1.6842153535724115e-05,
+      "loss": 0.4104,
+      "step": 10990
+    },
+    {
+      "epoch": 2.1637038676206632,
+      "grad_norm": 2.028120279312134,
+      "learning_rate": 1.683909627319698e-05,
+      "loss": 0.3666,
+      "step": 11000
+    },
+    {
+      "epoch": 2.1637038676206632,
+      "eval_loss": 0.2037489265203476,
+      "eval_runtime": 8.8865,
+      "eval_samples_per_second": 5.627,
+      "eval_steps_per_second": 2.813,
+      "step": 11000
+    },
+    {
+      "epoch": 2.1656708711366823,
+      "grad_norm": 1.9844077825546265,
+      "learning_rate": 1.6836039010669846e-05,
+      "loss": 0.3107,
+      "step": 11010
+    },
+    {
+      "epoch": 2.167637874652701,
+      "grad_norm": 0.890990674495697,
+      "learning_rate": 1.6832981748142712e-05,
+      "loss": 0.4025,
+      "step": 11020
+    },
+    {
+      "epoch": 2.1696048781687196,
+      "grad_norm": 1.8403328657150269,
+      "learning_rate": 1.682992448561558e-05,
+      "loss": 0.3271,
+      "step": 11030
+    },
+    {
+      "epoch": 2.1715718816847387,
+      "grad_norm": 1.9692051410675049,
+      "learning_rate": 1.6826867223088447e-05,
+      "loss": 0.4018,
+      "step": 11040
+    },
+    {
+      "epoch": 2.1735388852007573,
+      "grad_norm": 1.9689651727676392,
+      "learning_rate": 1.6823809960561316e-05,
+      "loss": 0.4714,
+      "step": 11050
+    },
+    {
+      "epoch": 2.175505888716776,
+      "grad_norm": 1.6802412271499634,
+      "learning_rate": 1.6820752698034182e-05,
+      "loss": 0.4456,
+      "step": 11060
+    },
+    {
+      "epoch": 2.177472892232795,
+      "grad_norm": 0.762005090713501,
+      "learning_rate": 1.6817695435507048e-05,
+      "loss": 0.4683,
+      "step": 11070
+    },
+    {
+      "epoch": 2.1794398957488137,
+      "grad_norm": 1.079770803451538,
+      "learning_rate": 1.6814638172979917e-05,
+      "loss": 0.4442,
+      "step": 11080
+    },
+    {
+      "epoch": 2.1814068992648323,
+      "grad_norm": 1.6594971418380737,
+      "learning_rate": 1.6811580910452783e-05,
+      "loss": 0.3854,
+      "step": 11090
+    },
+    {
+      "epoch": 2.1833739027808514,
+      "grad_norm": 2.0914525985717773,
+      "learning_rate": 1.680852364792565e-05,
+      "loss": 0.4987,
+      "step": 11100
+    },
+    {
+      "epoch": 2.18534090629687,
+      "grad_norm": 0.9866094589233398,
+      "learning_rate": 1.6805466385398514e-05,
+      "loss": 0.4544,
+      "step": 11110
+    },
+    {
+      "epoch": 2.1873079098128887,
+      "grad_norm": 1.063025712966919,
+      "learning_rate": 1.680240912287138e-05,
+      "loss": 0.3954,
+      "step": 11120
+    },
+    {
+      "epoch": 2.189274913328908,
+      "grad_norm": 1.4042121171951294,
+      "learning_rate": 1.679935186034425e-05,
+      "loss": 0.5461,
+      "step": 11130
+    },
+    {
+      "epoch": 2.1912419168449264,
+      "grad_norm": 1.1642961502075195,
+      "learning_rate": 1.6796294597817115e-05,
+      "loss": 0.4287,
+      "step": 11140
+    },
+    {
+      "epoch": 2.193208920360945,
+      "grad_norm": 1.1392892599105835,
+      "learning_rate": 1.679323733528998e-05,
+      "loss": 0.4508,
+      "step": 11150
+    },
+    {
+      "epoch": 2.195175923876964,
+      "grad_norm": 1.2517368793487549,
+      "learning_rate": 1.679018007276285e-05,
+      "loss": 0.4513,
+      "step": 11160
+    },
+    {
+      "epoch": 2.197142927392983,
+      "grad_norm": 1.0235626697540283,
+      "learning_rate": 1.6787122810235716e-05,
+      "loss": 0.4612,
+      "step": 11170
+    },
+    {
+      "epoch": 2.1991099309090014,
+      "grad_norm": 0.9338393211364746,
+      "learning_rate": 1.6784065547708585e-05,
+      "loss": 0.4623,
+      "step": 11180
+    },
+    {
+      "epoch": 2.2010769344250205,
+      "grad_norm": 1.9318597316741943,
+      "learning_rate": 1.678100828518145e-05,
+      "loss": 0.5033,
+      "step": 11190
+    },
+    {
+      "epoch": 2.203043937941039,
+      "grad_norm": 1.3232470750808716,
+      "learning_rate": 1.6777951022654317e-05,
+      "loss": 0.5158,
+      "step": 11200
+    },
+    {
+      "epoch": 2.205010941457058,
+      "grad_norm": 1.7988661527633667,
+      "learning_rate": 1.6774893760127186e-05,
+      "loss": 0.5245,
+      "step": 11210
+    },
+    {
+      "epoch": 2.206977944973077,
+      "grad_norm": 1.9910075664520264,
+      "learning_rate": 1.677183649760005e-05,
+      "loss": 0.4869,
+      "step": 11220
+    },
+    {
+      "epoch": 2.2089449484890955,
+      "grad_norm": 0.9067610502243042,
+      "learning_rate": 1.6768779235072917e-05,
+      "loss": 0.4429,
+      "step": 11230
+    },
+    {
+      "epoch": 2.210911952005114,
+      "grad_norm": 1.7104099988937378,
+      "learning_rate": 1.6765721972545783e-05,
+      "loss": 0.3101,
+      "step": 11240
+    },
+    {
+      "epoch": 2.2128789555211332,
+      "grad_norm": 1.0749773979187012,
+      "learning_rate": 1.676266471001865e-05,
+      "loss": 0.6143,
+      "step": 11250
+    },
+    {
+      "epoch": 2.214845959037152,
+      "grad_norm": 1.470632791519165,
+      "learning_rate": 1.6759607447491518e-05,
+      "loss": 0.2978,
+      "step": 11260
+    },
+    {
+      "epoch": 2.2168129625531705,
+      "grad_norm": 1.5304147005081177,
+      "learning_rate": 1.6756550184964384e-05,
+      "loss": 0.4798,
+      "step": 11270
+    },
+    {
+      "epoch": 2.218779966069189,
+      "grad_norm": 1.8575870990753174,
+      "learning_rate": 1.675349292243725e-05,
+      "loss": 0.3762,
+      "step": 11280
+    },
+    {
+      "epoch": 2.2207469695852082,
+      "grad_norm": 1.5989304780960083,
+      "learning_rate": 1.675043565991012e-05,
+      "loss": 0.388,
+      "step": 11290
+    },
+    {
+      "epoch": 2.222713973101227,
+      "grad_norm": 0.9317789077758789,
+      "learning_rate": 1.6747378397382985e-05,
+      "loss": 0.5609,
+      "step": 11300
+    },
+    {
+      "epoch": 2.2246809766172455,
+      "grad_norm": 1.4716814756393433,
+      "learning_rate": 1.6744321134855854e-05,
+      "loss": 0.4376,
+      "step": 11310
+    },
+    {
+      "epoch": 2.2266479801332646,
+      "grad_norm": 0.7102442383766174,
+      "learning_rate": 1.674126387232872e-05,
+      "loss": 0.5373,
+      "step": 11320
+    },
+    {
+      "epoch": 2.2286149836492832,
+      "grad_norm": 0.704011857509613,
+      "learning_rate": 1.6738206609801585e-05,
+      "loss": 0.3751,
+      "step": 11330
+    },
+    {
+      "epoch": 2.230581987165302,
+      "grad_norm": 1.789819598197937,
+      "learning_rate": 1.673514934727445e-05,
+      "loss": 0.4251,
+      "step": 11340
+    },
+    {
+      "epoch": 2.232548990681321,
+      "grad_norm": 2.0482563972473145,
+      "learning_rate": 1.6732092084747317e-05,
+      "loss": 0.3117,
+      "step": 11350
+    },
+    {
+      "epoch": 2.2345159941973396,
+      "grad_norm": 1.1816494464874268,
+      "learning_rate": 1.6729034822220186e-05,
+      "loss": 0.4754,
+      "step": 11360
+    },
+    {
+      "epoch": 2.2364829977133582,
+      "grad_norm": 0.9137541055679321,
+      "learning_rate": 1.6725977559693052e-05,
+      "loss": 0.3389,
+      "step": 11370
+    },
+    {
+      "epoch": 2.2384500012293773,
+      "grad_norm": 3.108690023422241,
+      "learning_rate": 1.6722920297165918e-05,
+      "loss": 0.502,
+      "step": 11380
+    },
+    {
+      "epoch": 2.240417004745396,
+      "grad_norm": 1.4583312273025513,
+      "learning_rate": 1.6719863034638787e-05,
+      "loss": 0.3701,
+      "step": 11390
+    },
+    {
+      "epoch": 2.2423840082614146,
+      "grad_norm": 0.734485387802124,
+      "learning_rate": 1.6716805772111652e-05,
+      "loss": 0.4625,
+      "step": 11400
+    },
+    {
+      "epoch": 2.2443510117774337,
+      "grad_norm": 1.41990327835083,
+      "learning_rate": 1.6713748509584518e-05,
+      "loss": 0.4124,
+      "step": 11410
+    },
+    {
+      "epoch": 2.2463180152934523,
+      "grad_norm": 1.3609710931777954,
+      "learning_rate": 1.6710691247057387e-05,
+      "loss": 0.3965,
+      "step": 11420
+    },
+    {
+      "epoch": 2.248285018809471,
+      "grad_norm": 0.8547394871711731,
+      "learning_rate": 1.6707633984530253e-05,
+      "loss": 0.4872,
+      "step": 11430
+    },
+    {
+      "epoch": 2.25025202232549,
+      "grad_norm": 0.9560080170631409,
+      "learning_rate": 1.6704576722003122e-05,
+      "loss": 0.4059,
+      "step": 11440
+    },
+    {
+      "epoch": 2.2522190258415087,
+      "grad_norm": 1.0539902448654175,
+      "learning_rate": 1.6701519459475985e-05,
+      "loss": 0.3862,
+      "step": 11450
+    },
+    {
+      "epoch": 2.2541860293575273,
+      "grad_norm": 1.8969827890396118,
+      "learning_rate": 1.6698462196948854e-05,
+      "loss": 0.3707,
+      "step": 11460
+    },
+    {
+      "epoch": 2.2561530328735464,
+      "grad_norm": 1.145606517791748,
+      "learning_rate": 1.669540493442172e-05,
+      "loss": 0.3504,
+      "step": 11470
+    },
+    {
+      "epoch": 2.258120036389565,
+      "grad_norm": 1.8824901580810547,
+      "learning_rate": 1.6692347671894585e-05,
+      "loss": 0.3876,
+      "step": 11480
+    },
+    {
+      "epoch": 2.2600870399055837,
+      "grad_norm": 2.9286253452301025,
+      "learning_rate": 1.6689290409367455e-05,
+      "loss": 0.4352,
+      "step": 11490
+    },
+    {
+      "epoch": 2.2620540434216028,
+      "grad_norm": 1.540687918663025,
+      "learning_rate": 1.668623314684032e-05,
+      "loss": 0.4271,
+      "step": 11500
+    },
+    {
+      "epoch": 2.2620540434216028,
+      "eval_loss": 0.20973175764083862,
+      "eval_runtime": 8.8684,
+      "eval_samples_per_second": 5.638,
+      "eval_steps_per_second": 2.819,
+      "step": 11500
+    },
+    {
+      "epoch": 2.2640210469376214,
+      "grad_norm": 1.8869454860687256,
+      "learning_rate": 1.6683175884313186e-05,
+      "loss": 0.3653,
+      "step": 11510
+    },
+    {
+      "epoch": 2.26598805045364,
+      "grad_norm": 1.647462010383606,
+      "learning_rate": 1.6680118621786055e-05,
+      "loss": 0.3849,
+      "step": 11520
+    },
+    {
+      "epoch": 2.267955053969659,
+      "grad_norm": 1.2821617126464844,
+      "learning_rate": 1.667706135925892e-05,
+      "loss": 0.4158,
+      "step": 11530
+    },
+    {
+      "epoch": 2.2699220574856778,
+      "grad_norm": 0.9892310500144958,
+      "learning_rate": 1.6674004096731787e-05,
+      "loss": 0.4286,
+      "step": 11540
+    },
+    {
+      "epoch": 2.2718890610016964,
+      "grad_norm": 1.099701166152954,
+      "learning_rate": 1.6670946834204656e-05,
+      "loss": 0.5114,
+      "step": 11550
+    },
+    {
+      "epoch": 2.2738560645177155,
+      "grad_norm": 1.2315559387207031,
+      "learning_rate": 1.6667889571677522e-05,
+      "loss": 0.3438,
+      "step": 11560
+    },
+    {
+      "epoch": 2.275823068033734,
+      "grad_norm": 1.3679817914962769,
+      "learning_rate": 1.6664832309150388e-05,
+      "loss": 0.3972,
+      "step": 11570
+    },
+    {
+      "epoch": 2.2777900715497528,
+      "grad_norm": 1.3526530265808105,
+      "learning_rate": 1.6661775046623253e-05,
+      "loss": 0.5465,
+      "step": 11580
+    },
+    {
+      "epoch": 2.279757075065772,
+      "grad_norm": 2.072378396987915,
+      "learning_rate": 1.6658717784096123e-05,
+      "loss": 0.4384,
+      "step": 11590
+    },
+    {
+      "epoch": 2.2817240785817905,
+      "grad_norm": 2.052748918533325,
+      "learning_rate": 1.665566052156899e-05,
+      "loss": 0.4023,
+      "step": 11600
+    },
+    {
+      "epoch": 2.283691082097809,
+      "grad_norm": 2.0281856060028076,
+      "learning_rate": 1.6652603259041854e-05,
+      "loss": 0.3589,
+      "step": 11610
+    },
+    {
+      "epoch": 2.285658085613828,
+      "grad_norm": 1.2389588356018066,
+      "learning_rate": 1.6649545996514723e-05,
+      "loss": 0.3531,
+      "step": 11620
+    },
+    {
+      "epoch": 2.287625089129847,
+      "grad_norm": 1.8695019483566284,
+      "learning_rate": 1.664648873398759e-05,
+      "loss": 0.4811,
+      "step": 11630
+    },
+    {
+      "epoch": 2.2895920926458655,
+      "grad_norm": 1.843996524810791,
+      "learning_rate": 1.6643431471460455e-05,
+      "loss": 0.4894,
+      "step": 11640
+    },
+    {
+      "epoch": 2.2915590961618846,
+      "grad_norm": 1.7340086698532104,
+      "learning_rate": 1.6640374208933324e-05,
+      "loss": 0.3934,
+      "step": 11650
+    },
+    {
+      "epoch": 2.293526099677903,
+      "grad_norm": 0.9214049577713013,
+      "learning_rate": 1.663731694640619e-05,
+      "loss": 0.4839,
+      "step": 11660
+    },
+    {
+      "epoch": 2.295493103193922,
+      "grad_norm": 1.3762463331222534,
+      "learning_rate": 1.6634259683879056e-05,
+      "loss": 0.376,
+      "step": 11670
+    },
+    {
+      "epoch": 2.2974601067099405,
+      "grad_norm": 1.5327290296554565,
+      "learning_rate": 1.663120242135192e-05,
+      "loss": 0.4463,
+      "step": 11680
+    },
+    {
+      "epoch": 2.2994271102259596,
+      "grad_norm": 1.4228308200836182,
+      "learning_rate": 1.662814515882479e-05,
+      "loss": 0.4595,
+      "step": 11690
+    },
+    {
+      "epoch": 2.301394113741978,
+      "grad_norm": 0.9541878700256348,
+      "learning_rate": 1.6625087896297656e-05,
+      "loss": 0.4507,
+      "step": 11700
+    },
+    {
+      "epoch": 2.303361117257997,
+      "grad_norm": 1.2874113321304321,
+      "learning_rate": 1.6622030633770522e-05,
+      "loss": 0.4727,
+      "step": 11710
+    },
+    {
+      "epoch": 2.305328120774016,
+      "grad_norm": 1.3238129615783691,
+      "learning_rate": 1.661897337124339e-05,
+      "loss": 0.4588,
+      "step": 11720
+    },
+    {
+      "epoch": 2.3072951242900346,
+      "grad_norm": 1.3692721128463745,
+      "learning_rate": 1.6615916108716257e-05,
+      "loss": 0.4373,
+      "step": 11730
+    },
+    {
+      "epoch": 2.309262127806053,
+      "grad_norm": 2.104457139968872,
+      "learning_rate": 1.6612858846189123e-05,
+      "loss": 0.3013,
+      "step": 11740
+    },
+    {
+      "epoch": 2.3112291313220723,
+      "grad_norm": 1.6918872594833374,
+      "learning_rate": 1.6609801583661992e-05,
+      "loss": 0.3407,
+      "step": 11750
+    },
+    {
+      "epoch": 2.313196134838091,
+      "grad_norm": 1.5859813690185547,
+      "learning_rate": 1.6606744321134858e-05,
+      "loss": 0.2991,
+      "step": 11760
+    },
+    {
+      "epoch": 2.3151631383541096,
+      "grad_norm": 2.3670969009399414,
+      "learning_rate": 1.6603687058607723e-05,
+      "loss": 0.3871,
+      "step": 11770
+    },
+    {
+      "epoch": 2.3171301418701287,
+      "grad_norm": 1.1432586908340454,
+      "learning_rate": 1.6600629796080593e-05,
+      "loss": 0.4183,
+      "step": 11780
+    },
+    {
+      "epoch": 2.3190971453861473,
+      "grad_norm": 2.7241263389587402,
+      "learning_rate": 1.6597572533553455e-05,
+      "loss": 0.3574,
+      "step": 11790
+    },
+    {
+      "epoch": 2.321064148902166,
+      "grad_norm": 1.253374695777893,
+      "learning_rate": 1.6594515271026324e-05,
+      "loss": 0.4658,
+      "step": 11800
+    },
+    {
+      "epoch": 2.323031152418185,
+      "grad_norm": 1.5033408403396606,
+      "learning_rate": 1.659145800849919e-05,
+      "loss": 0.4508,
+      "step": 11810
+    },
+    {
+      "epoch": 2.3249981559342037,
+      "grad_norm": 2.561887502670288,
+      "learning_rate": 1.658840074597206e-05,
+      "loss": 0.4595,
+      "step": 11820
+    },
+    {
+      "epoch": 2.3269651594502223,
+      "grad_norm": 1.2708535194396973,
+      "learning_rate": 1.6585343483444925e-05,
+      "loss": 0.5024,
+      "step": 11830
+    },
+    {
+      "epoch": 2.3289321629662414,
+      "grad_norm": 1.9025499820709229,
+      "learning_rate": 1.658228622091779e-05,
+      "loss": 0.365,
+      "step": 11840
+    },
+    {
+      "epoch": 2.33089916648226,
+      "grad_norm": 1.0760164260864258,
+      "learning_rate": 1.657922895839066e-05,
+      "loss": 0.4055,
+      "step": 11850
+    },
+    {
+      "epoch": 2.3328661699982787,
+      "grad_norm": 0.8068252801895142,
+      "learning_rate": 1.6576171695863526e-05,
+      "loss": 0.4234,
+      "step": 11860
+    },
+    {
+      "epoch": 2.3348331735142978,
+      "grad_norm": 1.0687072277069092,
+      "learning_rate": 1.657311443333639e-05,
+      "loss": 0.5491,
+      "step": 11870
+    },
+    {
+      "epoch": 2.3368001770303164,
+      "grad_norm": 1.7262513637542725,
+      "learning_rate": 1.657005717080926e-05,
+      "loss": 0.4301,
+      "step": 11880
+    },
+    {
+      "epoch": 2.338767180546335,
+      "grad_norm": 1.9340697526931763,
+      "learning_rate": 1.6566999908282126e-05,
+      "loss": 0.4087,
+      "step": 11890
+    },
+    {
+      "epoch": 2.340734184062354,
+      "grad_norm": 1.6383976936340332,
+      "learning_rate": 1.6563942645754992e-05,
+      "loss": 0.3496,
+      "step": 11900
+    },
+    {
+      "epoch": 2.3427011875783728,
+      "grad_norm": 1.1806261539459229,
+      "learning_rate": 1.6560885383227858e-05,
+      "loss": 0.2965,
+      "step": 11910
+    },
+    {
+      "epoch": 2.3446681910943914,
+      "grad_norm": 1.8395899534225464,
+      "learning_rate": 1.6557828120700724e-05,
+      "loss": 0.5577,
+      "step": 11920
+    },
+    {
+      "epoch": 2.3466351946104105,
+      "grad_norm": 1.3617602586746216,
+      "learning_rate": 1.6554770858173593e-05,
+      "loss": 0.4576,
+      "step": 11930
+    },
+    {
+      "epoch": 2.348602198126429,
+      "grad_norm": 1.1605844497680664,
+      "learning_rate": 1.655171359564646e-05,
+      "loss": 0.3198,
+      "step": 11940
+    },
+    {
+      "epoch": 2.3505692016424478,
+      "grad_norm": 1.2354375123977661,
+      "learning_rate": 1.6548656333119328e-05,
+      "loss": 0.568,
+      "step": 11950
+    },
+    {
+      "epoch": 2.352536205158467,
+      "grad_norm": 1.2012954950332642,
+      "learning_rate": 1.6545599070592194e-05,
+      "loss": 0.3795,
+      "step": 11960
+    },
+    {
+      "epoch": 2.3545032086744855,
+      "grad_norm": 2.271904945373535,
+      "learning_rate": 1.654254180806506e-05,
+      "loss": 0.3787,
+      "step": 11970
+    },
+    {
+      "epoch": 2.356470212190504,
+      "grad_norm": 1.7770686149597168,
+      "learning_rate": 1.653948454553793e-05,
+      "loss": 0.3701,
+      "step": 11980
+    },
+    {
+      "epoch": 2.358437215706523,
+      "grad_norm": 1.3162378072738647,
+      "learning_rate": 1.6536427283010794e-05,
+      "loss": 0.3635,
+      "step": 11990
+    },
+    {
+      "epoch": 2.360404219222542,
+      "grad_norm": 0.8531973958015442,
+      "learning_rate": 1.653337002048366e-05,
+      "loss": 0.4221,
+      "step": 12000
+    },
+    {
+      "epoch": 2.360404219222542,
+      "eval_loss": 0.21400800347328186,
+      "eval_runtime": 8.8552,
+      "eval_samples_per_second": 5.646,
+      "eval_steps_per_second": 2.823,
+      "step": 12000
+    },
+    {
+      "epoch": 2.3623712227385605,
+      "grad_norm": 1.781295657157898,
+      "learning_rate": 1.653031275795653e-05,
+      "loss": 0.3358,
+      "step": 12010
+    },
+    {
+      "epoch": 2.3643382262545796,
+      "grad_norm": 2.028844118118286,
+      "learning_rate": 1.652725549542939e-05,
+      "loss": 0.4255,
+      "step": 12020
+    },
+    {
+      "epoch": 2.366305229770598,
+      "grad_norm": 2.3487181663513184,
+      "learning_rate": 1.652419823290226e-05,
+      "loss": 0.5234,
+      "step": 12030
+    },
+    {
+      "epoch": 2.368272233286617,
+      "grad_norm": 2.8350348472595215,
+      "learning_rate": 1.6521140970375127e-05,
+      "loss": 0.3041,
+      "step": 12040
+    },
+    {
+      "epoch": 2.370239236802636,
+      "grad_norm": 1.8248299360275269,
+      "learning_rate": 1.6518083707847992e-05,
+      "loss": 0.3711,
+      "step": 12050
+    },
+    {
+      "epoch": 2.3722062403186546,
+      "grad_norm": 1.7937493324279785,
+      "learning_rate": 1.651502644532086e-05,
+      "loss": 0.4739,
+      "step": 12060
+    },
+    {
+      "epoch": 2.374173243834673,
+      "grad_norm": 1.0475170612335205,
+      "learning_rate": 1.6511969182793727e-05,
+      "loss": 0.4552,
+      "step": 12070
+    },
+    {
+      "epoch": 2.3761402473506923,
+      "grad_norm": 1.3136638402938843,
+      "learning_rate": 1.6508911920266596e-05,
+      "loss": 0.3586,
+      "step": 12080
+    },
+    {
+      "epoch": 2.378107250866711,
+      "grad_norm": 1.4082086086273193,
+      "learning_rate": 1.6505854657739462e-05,
+      "loss": 0.4826,
+      "step": 12090
+    },
+    {
+      "epoch": 2.3800742543827296,
+      "grad_norm": 1.2185932397842407,
+      "learning_rate": 1.6502797395212328e-05,
+      "loss": 0.3852,
+      "step": 12100
+    },
+    {
+      "epoch": 2.3820412578987487,
+      "grad_norm": 2.0192642211914062,
+      "learning_rate": 1.6499740132685197e-05,
+      "loss": 0.4272,
+      "step": 12110
+    },
+    {
+      "epoch": 2.3840082614147673,
+      "grad_norm": 2.2088992595672607,
+      "learning_rate": 1.6496682870158063e-05,
+      "loss": 0.4102,
+      "step": 12120
+    },
+    {
+      "epoch": 2.385975264930786,
+      "grad_norm": 1.1546714305877686,
+      "learning_rate": 1.649362560763093e-05,
+      "loss": 0.4554,
+      "step": 12130
+    },
+    {
+      "epoch": 2.387942268446805,
+      "grad_norm": 0.7382022142410278,
+      "learning_rate": 1.6490568345103795e-05,
+      "loss": 0.3757,
+      "step": 12140
+    },
+    {
+      "epoch": 2.3899092719628237,
+      "grad_norm": 1.110977053642273,
+      "learning_rate": 1.648751108257666e-05,
+      "loss": 0.4407,
+      "step": 12150
+    },
+    {
+      "epoch": 2.3918762754788423,
+      "grad_norm": 1.028681755065918,
+      "learning_rate": 1.648445382004953e-05,
+      "loss": 0.3124,
+      "step": 12160
+    },
+    {
+      "epoch": 2.3938432789948614,
+      "grad_norm": 1.3879059553146362,
+      "learning_rate": 1.6481396557522395e-05,
+      "loss": 0.4393,
+      "step": 12170
+    },
+    {
+      "epoch": 2.39581028251088,
+      "grad_norm": 1.3907514810562134,
+      "learning_rate": 1.647833929499526e-05,
+      "loss": 0.4351,
+      "step": 12180
+    },
+    {
+      "epoch": 2.3977772860268987,
+      "grad_norm": 1.410379409790039,
+      "learning_rate": 1.647528203246813e-05,
+      "loss": 0.5897,
+      "step": 12190
+    },
+    {
+      "epoch": 2.3997442895429177,
+      "grad_norm": 2.0820980072021484,
+      "learning_rate": 1.6472224769940996e-05,
+      "loss": 0.4466,
+      "step": 12200
+    },
+    {
+      "epoch": 2.4017112930589364,
+      "grad_norm": 1.685351014137268,
+      "learning_rate": 1.6469167507413865e-05,
+      "loss": 0.4181,
+      "step": 12210
+    },
+    {
+      "epoch": 2.403678296574955,
+      "grad_norm": 2.2443206310272217,
+      "learning_rate": 1.646611024488673e-05,
+      "loss": 0.4247,
+      "step": 12220
+    },
+    {
+      "epoch": 2.405645300090974,
+      "grad_norm": 1.3944865465164185,
+      "learning_rate": 1.6463052982359597e-05,
+      "loss": 0.3066,
+      "step": 12230
+    },
+    {
+      "epoch": 2.4076123036069927,
+      "grad_norm": 1.7855195999145508,
+      "learning_rate": 1.6459995719832466e-05,
+      "loss": 0.5395,
+      "step": 12240
+    },
+    {
+      "epoch": 2.4095793071230114,
+      "grad_norm": 1.5307120084762573,
+      "learning_rate": 1.6456938457305328e-05,
+      "loss": 0.3447,
+      "step": 12250
+    },
+    {
+      "epoch": 2.4115463106390305,
+      "grad_norm": 2.71352219581604,
+      "learning_rate": 1.6453881194778197e-05,
+      "loss": 0.4779,
+      "step": 12260
+    },
+    {
+      "epoch": 2.413513314155049,
+      "grad_norm": 1.4388123750686646,
+      "learning_rate": 1.6450823932251063e-05,
+      "loss": 0.4417,
+      "step": 12270
+    },
+    {
+      "epoch": 2.4154803176710677,
+      "grad_norm": 1.343959093093872,
+      "learning_rate": 1.644776666972393e-05,
+      "loss": 0.5612,
+      "step": 12280
+    },
+    {
+      "epoch": 2.417447321187087,
+      "grad_norm": 1.7309019565582275,
+      "learning_rate": 1.6444709407196798e-05,
+      "loss": 0.5677,
+      "step": 12290
+    },
+    {
+      "epoch": 2.4194143247031055,
+      "grad_norm": 0.48004379868507385,
+      "learning_rate": 1.6441652144669664e-05,
+      "loss": 0.4653,
+      "step": 12300
+    },
+    {
+      "epoch": 2.421381328219124,
+      "grad_norm": 1.704228162765503,
+      "learning_rate": 1.643859488214253e-05,
+      "loss": 0.414,
+      "step": 12310
+    },
+    {
+      "epoch": 2.423348331735143,
+      "grad_norm": 1.2886383533477783,
+      "learning_rate": 1.64355376196154e-05,
+      "loss": 0.4166,
+      "step": 12320
+    },
+    {
+      "epoch": 2.425315335251162,
+      "grad_norm": 1.781337857246399,
+      "learning_rate": 1.6432480357088265e-05,
+      "loss": 0.3569,
+      "step": 12330
+    },
+    {
+      "epoch": 2.4272823387671805,
+      "grad_norm": 2.4359853267669678,
+      "learning_rate": 1.6429423094561134e-05,
+      "loss": 0.4891,
+      "step": 12340
+    },
+    {
+      "epoch": 2.4292493422831996,
+      "grad_norm": 1.3055243492126465,
+      "learning_rate": 1.6426365832034e-05,
+      "loss": 0.4029,
+      "step": 12350
+    },
+    {
+      "epoch": 2.431216345799218,
+      "grad_norm": 0.97089022397995,
+      "learning_rate": 1.6423308569506865e-05,
+      "loss": 0.4751,
+      "step": 12360
+    },
+    {
+      "epoch": 2.433183349315237,
+      "grad_norm": 0.9612852931022644,
+      "learning_rate": 1.642025130697973e-05,
+      "loss": 0.4231,
+      "step": 12370
+    },
+    {
+      "epoch": 2.435150352831256,
+      "grad_norm": 3.4028701782226562,
+      "learning_rate": 1.6417194044452597e-05,
+      "loss": 0.3678,
+      "step": 12380
+    },
+    {
+      "epoch": 2.4371173563472746,
+      "grad_norm": 1.2526423931121826,
+      "learning_rate": 1.6414136781925466e-05,
+      "loss": 0.5883,
+      "step": 12390
+    },
+    {
+      "epoch": 2.439084359863293,
+      "grad_norm": 1.2844873666763306,
+      "learning_rate": 1.6411079519398332e-05,
+      "loss": 0.4305,
+      "step": 12400
+    },
+    {
+      "epoch": 2.4410513633793123,
+      "grad_norm": 0.8970216512680054,
+      "learning_rate": 1.6408022256871198e-05,
+      "loss": 0.3743,
+      "step": 12410
+    },
+    {
+      "epoch": 2.443018366895331,
+      "grad_norm": 2.136035203933716,
+      "learning_rate": 1.6404964994344067e-05,
+      "loss": 0.5527,
+      "step": 12420
+    },
+    {
+      "epoch": 2.4449853704113496,
+      "grad_norm": 1.0382180213928223,
+      "learning_rate": 1.6401907731816933e-05,
+      "loss": 0.5142,
+      "step": 12430
+    },
+    {
+      "epoch": 2.4469523739273686,
+      "grad_norm": 1.2471837997436523,
+      "learning_rate": 1.63988504692898e-05,
+      "loss": 0.4031,
+      "step": 12440
+    },
+    {
+      "epoch": 2.4489193774433873,
+      "grad_norm": 1.7783029079437256,
+      "learning_rate": 1.6395793206762668e-05,
+      "loss": 0.5711,
+      "step": 12450
+    },
+    {
+      "epoch": 2.450886380959406,
+      "grad_norm": 2.7205777168273926,
+      "learning_rate": 1.6392735944235533e-05,
+      "loss": 0.5073,
+      "step": 12460
+    },
+    {
+      "epoch": 2.4528533844754246,
+      "grad_norm": 0.9302681088447571,
+      "learning_rate": 1.63896786817084e-05,
+      "loss": 0.3801,
+      "step": 12470
+    },
+    {
+      "epoch": 2.4548203879914436,
+      "grad_norm": 2.4945271015167236,
+      "learning_rate": 1.6386621419181265e-05,
+      "loss": 0.4482,
+      "step": 12480
+    },
+    {
+      "epoch": 2.4567873915074623,
+      "grad_norm": 1.396541953086853,
+      "learning_rate": 1.6383564156654134e-05,
+      "loss": 0.4727,
+      "step": 12490
+    },
+    {
+      "epoch": 2.458754395023481,
+      "grad_norm": 1.2617021799087524,
+      "learning_rate": 1.6380506894127e-05,
+      "loss": 0.5215,
+      "step": 12500
+    },
+    {
+      "epoch": 2.458754395023481,
+      "eval_loss": 0.20947669446468353,
+      "eval_runtime": 8.8627,
+      "eval_samples_per_second": 5.642,
+      "eval_steps_per_second": 2.821,
+      "step": 12500
+    },
+    {
+      "epoch": 2.4607213985395,
+      "grad_norm": 1.8948769569396973,
+      "learning_rate": 1.6377449631599866e-05,
+      "loss": 0.3883,
+      "step": 12510
+    },
+    {
+      "epoch": 2.4626884020555186,
+      "grad_norm": 2.941626787185669,
+      "learning_rate": 1.6374392369072735e-05,
+      "loss": 0.4237,
+      "step": 12520
+    },
+    {
+      "epoch": 2.4646554055715373,
+      "grad_norm": 0.9059364795684814,
+      "learning_rate": 1.63713351065456e-05,
+      "loss": 0.3822,
+      "step": 12530
+    },
+    {
+      "epoch": 2.4666224090875564,
+      "grad_norm": 1.0603015422821045,
+      "learning_rate": 1.6368277844018466e-05,
+      "loss": 0.3393,
+      "step": 12540
+    },
+    {
+      "epoch": 2.468589412603575,
+      "grad_norm": 2.579197645187378,
+      "learning_rate": 1.6365220581491335e-05,
+      "loss": 0.3045,
+      "step": 12550
+    },
+    {
+      "epoch": 2.4705564161195936,
+      "grad_norm": 1.5118027925491333,
+      "learning_rate": 1.63621633189642e-05,
+      "loss": 0.5017,
+      "step": 12560
+    },
+    {
+      "epoch": 2.4725234196356127,
+      "grad_norm": 0.8895286321640015,
+      "learning_rate": 1.6359106056437067e-05,
+      "loss": 0.3775,
+      "step": 12570
+    },
+    {
+      "epoch": 2.4744904231516314,
+      "grad_norm": 2.36152982711792,
+      "learning_rate": 1.6356048793909933e-05,
+      "loss": 0.3886,
+      "step": 12580
+    },
+    {
+      "epoch": 2.47645742666765,
+      "grad_norm": 0.6294612884521484,
+      "learning_rate": 1.63529915313828e-05,
+      "loss": 0.3861,
+      "step": 12590
+    },
+    {
+      "epoch": 2.478424430183669,
+      "grad_norm": 1.4751849174499512,
+      "learning_rate": 1.6349934268855668e-05,
+      "loss": 0.5084,
+      "step": 12600
+    },
+    {
+      "epoch": 2.4803914336996877,
+      "grad_norm": 1.883037805557251,
+      "learning_rate": 1.6346877006328534e-05,
+      "loss": 0.3831,
+      "step": 12610
+    },
+    {
+      "epoch": 2.4823584372157064,
+      "grad_norm": 0.8708747029304504,
+      "learning_rate": 1.6343819743801403e-05,
+      "loss": 0.489,
+      "step": 12620
+    },
+    {
+      "epoch": 2.4843254407317255,
+      "grad_norm": 1.8524725437164307,
+      "learning_rate": 1.634076248127427e-05,
+      "loss": 0.4233,
+      "step": 12630
+    },
+    {
+      "epoch": 2.486292444247744,
+      "grad_norm": 1.2213215827941895,
+      "learning_rate": 1.6337705218747134e-05,
+      "loss": 0.5392,
+      "step": 12640
+    },
+    {
+      "epoch": 2.4882594477637627,
+      "grad_norm": 2.4590892791748047,
+      "learning_rate": 1.6334647956220003e-05,
+      "loss": 0.3953,
+      "step": 12650
+    },
+    {
+      "epoch": 2.490226451279782,
+      "grad_norm": 1.7708888053894043,
+      "learning_rate": 1.633159069369287e-05,
+      "loss": 0.3311,
+      "step": 12660
+    },
+    {
+      "epoch": 2.4921934547958005,
+      "grad_norm": 2.163320541381836,
+      "learning_rate": 1.6328533431165735e-05,
+      "loss": 0.3635,
+      "step": 12670
+    },
+    {
+      "epoch": 2.494160458311819,
+      "grad_norm": 1.4477022886276245,
+      "learning_rate": 1.6325476168638604e-05,
+      "loss": 0.4243,
+      "step": 12680
+    },
+    {
+      "epoch": 2.496127461827838,
+      "grad_norm": 1.0409399271011353,
+      "learning_rate": 1.632241890611147e-05,
+      "loss": 0.4322,
+      "step": 12690
+    },
+    {
+      "epoch": 2.498094465343857,
+      "grad_norm": 1.2920570373535156,
+      "learning_rate": 1.6319361643584336e-05,
+      "loss": 0.3994,
+      "step": 12700
+    },
+    {
+      "epoch": 2.5000614688598755,
+      "grad_norm": 1.5400962829589844,
+      "learning_rate": 1.63163043810572e-05,
+      "loss": 0.299,
+      "step": 12710
+    },
+    {
+      "epoch": 2.502028472375894,
+      "grad_norm": 1.4039868116378784,
+      "learning_rate": 1.6313247118530067e-05,
+      "loss": 0.3499,
+      "step": 12720
+    },
+    {
+      "epoch": 2.503995475891913,
+      "grad_norm": 1.696679949760437,
+      "learning_rate": 1.6310189856002936e-05,
+      "loss": 0.5174,
+      "step": 12730
+    },
+    {
+      "epoch": 2.505962479407932,
+      "grad_norm": 1.6985901594161987,
+      "learning_rate": 1.6307132593475802e-05,
+      "loss": 0.4112,
+      "step": 12740
+    },
+    {
+      "epoch": 2.5079294829239505,
+      "grad_norm": 0.7867997288703918,
+      "learning_rate": 1.630407533094867e-05,
+      "loss": 0.4532,
+      "step": 12750
+    },
+    {
+      "epoch": 2.5098964864399695,
+      "grad_norm": 1.9344456195831299,
+      "learning_rate": 1.6301018068421537e-05,
+      "loss": 0.4035,
+      "step": 12760
+    },
+    {
+      "epoch": 2.511863489955988,
+      "grad_norm": 1.574959635734558,
+      "learning_rate": 1.6297960805894403e-05,
+      "loss": 0.3104,
+      "step": 12770
+    },
+    {
+      "epoch": 2.513830493472007,
+      "grad_norm": 1.628767967224121,
+      "learning_rate": 1.6294903543367272e-05,
+      "loss": 0.5006,
+      "step": 12780
+    },
+    {
+      "epoch": 2.515797496988026,
+      "grad_norm": 1.0452278852462769,
+      "learning_rate": 1.6291846280840138e-05,
+      "loss": 0.3795,
+      "step": 12790
+    },
+    {
+      "epoch": 2.5177645005040445,
+      "grad_norm": 1.067253589630127,
+      "learning_rate": 1.6288789018313004e-05,
+      "loss": 0.3938,
+      "step": 12800
+    },
+    {
+      "epoch": 2.519731504020063,
+      "grad_norm": 2.536316156387329,
+      "learning_rate": 1.628573175578587e-05,
+      "loss": 0.4488,
+      "step": 12810
+    },
+    {
+      "epoch": 2.5216985075360823,
+      "grad_norm": 2.0406346321105957,
+      "learning_rate": 1.6282674493258735e-05,
+      "loss": 0.3586,
+      "step": 12820
+    },
+    {
+      "epoch": 2.523665511052101,
+      "grad_norm": 2.524869680404663,
+      "learning_rate": 1.6279617230731604e-05,
+      "loss": 0.3906,
+      "step": 12830
+    },
+    {
+      "epoch": 2.5256325145681195,
+      "grad_norm": 1.8386890888214111,
+      "learning_rate": 1.627655996820447e-05,
+      "loss": 0.6009,
+      "step": 12840
+    },
+    {
+      "epoch": 2.5275995180841386,
+      "grad_norm": 0.31160733103752136,
+      "learning_rate": 1.6273502705677336e-05,
+      "loss": 0.4166,
+      "step": 12850
+    },
+    {
+      "epoch": 2.5295665216001573,
+      "grad_norm": 1.9621902704238892,
+      "learning_rate": 1.6270445443150205e-05,
+      "loss": 0.4007,
+      "step": 12860
+    },
+    {
+      "epoch": 2.531533525116176,
+      "grad_norm": 1.396183967590332,
+      "learning_rate": 1.626738818062307e-05,
+      "loss": 0.4049,
+      "step": 12870
+    },
+    {
+      "epoch": 2.533500528632195,
+      "grad_norm": 1.2113840579986572,
+      "learning_rate": 1.626433091809594e-05,
+      "loss": 0.4187,
+      "step": 12880
+    },
+    {
+      "epoch": 2.5354675321482136,
+      "grad_norm": 0.6130431294441223,
+      "learning_rate": 1.6261273655568806e-05,
+      "loss": 0.3547,
+      "step": 12890
+    },
+    {
+      "epoch": 2.5374345356642323,
+      "grad_norm": 1.4548547267913818,
+      "learning_rate": 1.625821639304167e-05,
+      "loss": 0.5581,
+      "step": 12900
+    },
+    {
+      "epoch": 2.5394015391802514,
+      "grad_norm": 1.243184208869934,
+      "learning_rate": 1.625515913051454e-05,
+      "loss": 0.3434,
+      "step": 12910
+    },
+    {
+      "epoch": 2.54136854269627,
+      "grad_norm": 1.3352422714233398,
+      "learning_rate": 1.6252101867987403e-05,
+      "loss": 0.3148,
+      "step": 12920
+    },
+    {
+      "epoch": 2.5433355462122886,
+      "grad_norm": 0.9811519980430603,
+      "learning_rate": 1.6249044605460272e-05,
+      "loss": 0.4075,
+      "step": 12930
+    },
+    {
+      "epoch": 2.5453025497283077,
+      "grad_norm": 1.1707277297973633,
+      "learning_rate": 1.6245987342933138e-05,
+      "loss": 0.4329,
+      "step": 12940
+    },
+    {
+      "epoch": 2.5472695532443264,
+      "grad_norm": 0.9138590693473816,
+      "learning_rate": 1.6242930080406004e-05,
+      "loss": 0.355,
+      "step": 12950
+    },
+    {
+      "epoch": 2.549236556760345,
+      "grad_norm": 1.6511414051055908,
+      "learning_rate": 1.6239872817878873e-05,
+      "loss": 0.3969,
+      "step": 12960
+    },
+    {
+      "epoch": 2.551203560276364,
+      "grad_norm": 1.2512931823730469,
+      "learning_rate": 1.623681555535174e-05,
+      "loss": 0.4485,
+      "step": 12970
+    },
+    {
+      "epoch": 2.5531705637923827,
+      "grad_norm": 2.979414701461792,
+      "learning_rate": 1.6233758292824605e-05,
+      "loss": 0.3468,
+      "step": 12980
+    },
+    {
+      "epoch": 2.5551375673084014,
+      "grad_norm": 2.5046603679656982,
+      "learning_rate": 1.6230701030297474e-05,
+      "loss": 0.438,
+      "step": 12990
+    },
+    {
+      "epoch": 2.5571045708244204,
+      "grad_norm": 1.1281431913375854,
+      "learning_rate": 1.622764376777034e-05,
+      "loss": 0.3836,
+      "step": 13000
+    },
+    {
+      "epoch": 2.5571045708244204,
+      "eval_loss": 0.20514748990535736,
+      "eval_runtime": 8.8701,
+      "eval_samples_per_second": 5.637,
+      "eval_steps_per_second": 2.818,
+      "step": 13000
+    },
+    {
+      "epoch": 2.559071574340439,
+      "grad_norm": 1.0448176860809326,
+      "learning_rate": 1.622458650524321e-05,
+      "loss": 0.5373,
+      "step": 13010
+    },
+    {
+      "epoch": 2.5610385778564577,
+      "grad_norm": 1.2854679822921753,
+      "learning_rate": 1.6221529242716074e-05,
+      "loss": 0.4245,
+      "step": 13020
+    },
+    {
+      "epoch": 2.563005581372477,
+      "grad_norm": 1.9112696647644043,
+      "learning_rate": 1.621847198018894e-05,
+      "loss": 0.4786,
+      "step": 13030
+    },
+    {
+      "epoch": 2.5649725848884954,
+      "grad_norm": 1.9691932201385498,
+      "learning_rate": 1.6215414717661806e-05,
+      "loss": 0.4054,
+      "step": 13040
+    },
+    {
+      "epoch": 2.566939588404514,
+      "grad_norm": 2.541759490966797,
+      "learning_rate": 1.6212357455134672e-05,
+      "loss": 0.4093,
+      "step": 13050
+    },
+    {
+      "epoch": 2.568906591920533,
+      "grad_norm": 1.623146414756775,
+      "learning_rate": 1.620930019260754e-05,
+      "loss": 0.4011,
+      "step": 13060
+    },
+    {
+      "epoch": 2.570873595436552,
+      "grad_norm": 1.0942410230636597,
+      "learning_rate": 1.6206242930080407e-05,
+      "loss": 0.3456,
+      "step": 13070
+    },
+    {
+      "epoch": 2.5728405989525704,
+      "grad_norm": 2.006178140640259,
+      "learning_rate": 1.6203185667553273e-05,
+      "loss": 0.4755,
+      "step": 13080
+    },
+    {
+      "epoch": 2.5748076024685895,
+      "grad_norm": 1.132165789604187,
+      "learning_rate": 1.6200128405026142e-05,
+      "loss": 0.4762,
+      "step": 13090
+    },
+    {
+      "epoch": 2.576774605984608,
+      "grad_norm": 0.8494846224784851,
+      "learning_rate": 1.6197071142499007e-05,
+      "loss": 0.5175,
+      "step": 13100
+    },
+    {
+      "epoch": 2.578741609500627,
+      "grad_norm": 1.8866459131240845,
+      "learning_rate": 1.6194013879971873e-05,
+      "loss": 0.5924,
+      "step": 13110
+    },
+    {
+      "epoch": 2.580708613016646,
+      "grad_norm": 0.9108813405036926,
+      "learning_rate": 1.6190956617444742e-05,
+      "loss": 0.4604,
+      "step": 13120
+    },
+    {
+      "epoch": 2.5826756165326645,
+      "grad_norm": 1.544240951538086,
+      "learning_rate": 1.6187899354917608e-05,
+      "loss": 0.4162,
+      "step": 13130
+    },
+    {
+      "epoch": 2.584642620048683,
+      "grad_norm": 2.693819046020508,
+      "learning_rate": 1.6184842092390477e-05,
+      "loss": 0.4684,
+      "step": 13140
+    },
+    {
+      "epoch": 2.5866096235647023,
+      "grad_norm": 2.237470865249634,
+      "learning_rate": 1.618178482986334e-05,
+      "loss": 0.3965,
+      "step": 13150
+    },
+    {
+      "epoch": 2.588576627080721,
+      "grad_norm": 1.9089361429214478,
+      "learning_rate": 1.617872756733621e-05,
+      "loss": 0.4684,
+      "step": 13160
+    },
+    {
+      "epoch": 2.5905436305967395,
+      "grad_norm": 1.8669962882995605,
+      "learning_rate": 1.6175670304809075e-05,
+      "loss": 0.5052,
+      "step": 13170
+    },
+    {
+      "epoch": 2.5925106341127586,
+      "grad_norm": 1.4717791080474854,
+      "learning_rate": 1.617261304228194e-05,
+      "loss": 0.4677,
+      "step": 13180
+    },
+    {
+      "epoch": 2.5944776376287773,
+      "grad_norm": 1.2365857362747192,
+      "learning_rate": 1.616955577975481e-05,
+      "loss": 0.516,
+      "step": 13190
+    },
+    {
+      "epoch": 2.596444641144796,
+      "grad_norm": 1.2148315906524658,
+      "learning_rate": 1.6166498517227675e-05,
+      "loss": 0.4646,
+      "step": 13200
+    },
+    {
+      "epoch": 2.598411644660815,
+      "grad_norm": 1.734046459197998,
+      "learning_rate": 1.616344125470054e-05,
+      "loss": 0.3972,
+      "step": 13210
+    },
+    {
+      "epoch": 2.6003786481768336,
+      "grad_norm": 1.1978328227996826,
+      "learning_rate": 1.616038399217341e-05,
+      "loss": 0.4358,
+      "step": 13220
+    },
+    {
+      "epoch": 2.6023456516928523,
+      "grad_norm": 1.2464817762374878,
+      "learning_rate": 1.6157326729646276e-05,
+      "loss": 0.3132,
+      "step": 13230
+    },
+    {
+      "epoch": 2.6043126552088713,
+      "grad_norm": 1.3837008476257324,
+      "learning_rate": 1.6154269467119142e-05,
+      "loss": 0.4221,
+      "step": 13240
+    },
+    {
+      "epoch": 2.60627965872489,
+      "grad_norm": 1.9459171295166016,
+      "learning_rate": 1.615121220459201e-05,
+      "loss": 0.3874,
+      "step": 13250
+    },
+    {
+      "epoch": 2.6082466622409086,
+      "grad_norm": 1.4665859937667847,
+      "learning_rate": 1.6148154942064877e-05,
+      "loss": 0.4835,
+      "step": 13260
+    },
+    {
+      "epoch": 2.6102136657569277,
+      "grad_norm": 1.916831612586975,
+      "learning_rate": 1.6145097679537743e-05,
+      "loss": 0.3224,
+      "step": 13270
+    },
+    {
+      "epoch": 2.6121806692729463,
+      "grad_norm": 1.1361554861068726,
+      "learning_rate": 1.614204041701061e-05,
+      "loss": 0.2798,
+      "step": 13280
+    },
+    {
+      "epoch": 2.614147672788965,
+      "grad_norm": 1.8515132665634155,
+      "learning_rate": 1.6138983154483478e-05,
+      "loss": 0.4076,
+      "step": 13290
+    },
+    {
+      "epoch": 2.616114676304984,
+      "grad_norm": 1.1301642656326294,
+      "learning_rate": 1.6135925891956343e-05,
+      "loss": 0.4611,
+      "step": 13300
+    },
+    {
+      "epoch": 2.6180816798210027,
+      "grad_norm": 1.3350282907485962,
+      "learning_rate": 1.613286862942921e-05,
+      "loss": 0.3452,
+      "step": 13310
+    },
+    {
+      "epoch": 2.6200486833370213,
+      "grad_norm": 3.7476720809936523,
+      "learning_rate": 1.6129811366902078e-05,
+      "loss": 0.3489,
+      "step": 13320
+    },
+    {
+      "epoch": 2.6220156868530404,
+      "grad_norm": 2.181448459625244,
+      "learning_rate": 1.6126754104374944e-05,
+      "loss": 0.4138,
+      "step": 13330
+    },
+    {
+      "epoch": 2.623982690369059,
+      "grad_norm": 0.9513285756111145,
+      "learning_rate": 1.612369684184781e-05,
+      "loss": 0.5012,
+      "step": 13340
+    },
+    {
+      "epoch": 2.6259496938850777,
+      "grad_norm": 1.0185880661010742,
+      "learning_rate": 1.612063957932068e-05,
+      "loss": 0.3593,
+      "step": 13350
+    },
+    {
+      "epoch": 2.627916697401097,
+      "grad_norm": 1.5732872486114502,
+      "learning_rate": 1.6117582316793545e-05,
+      "loss": 0.5159,
+      "step": 13360
+    },
+    {
+      "epoch": 2.6298837009171154,
+      "grad_norm": 0.9264469742774963,
+      "learning_rate": 1.611452505426641e-05,
+      "loss": 0.3632,
+      "step": 13370
+    },
+    {
+      "epoch": 2.631850704433134,
+      "grad_norm": 1.364571213722229,
+      "learning_rate": 1.6111467791739276e-05,
+      "loss": 0.3872,
+      "step": 13380
+    },
+    {
+      "epoch": 2.633817707949153,
+      "grad_norm": 0.6632816195487976,
+      "learning_rate": 1.6108410529212146e-05,
+      "loss": 0.3853,
+      "step": 13390
+    },
+    {
+      "epoch": 2.635784711465172,
+      "grad_norm": 1.6225327253341675,
+      "learning_rate": 1.610535326668501e-05,
+      "loss": 0.4046,
+      "step": 13400
+    },
+    {
+      "epoch": 2.6377517149811904,
+      "grad_norm": 1.5951011180877686,
+      "learning_rate": 1.6102296004157877e-05,
+      "loss": 0.5356,
+      "step": 13410
+    },
+    {
+      "epoch": 2.6397187184972095,
+      "grad_norm": 1.490448236465454,
+      "learning_rate": 1.6099238741630746e-05,
+      "loss": 0.3891,
+      "step": 13420
+    },
+    {
+      "epoch": 2.641685722013228,
+      "grad_norm": 0.9575764536857605,
+      "learning_rate": 1.6096181479103612e-05,
+      "loss": 0.5333,
+      "step": 13430
+    },
+    {
+      "epoch": 2.643652725529247,
+      "grad_norm": 1.7446562051773071,
+      "learning_rate": 1.6093124216576478e-05,
+      "loss": 0.4368,
+      "step": 13440
+    },
+    {
+      "epoch": 2.645619729045266,
+      "grad_norm": 2.8371479511260986,
+      "learning_rate": 1.6090066954049347e-05,
+      "loss": 0.4029,
+      "step": 13450
+    },
+    {
+      "epoch": 2.6475867325612845,
+      "grad_norm": 0.8692865371704102,
+      "learning_rate": 1.6087009691522213e-05,
+      "loss": 0.4666,
+      "step": 13460
+    },
+    {
+      "epoch": 2.649553736077303,
+      "grad_norm": 0.8409749269485474,
+      "learning_rate": 1.608395242899508e-05,
+      "loss": 0.3958,
+      "step": 13470
+    },
+    {
+      "epoch": 2.6515207395933222,
+      "grad_norm": 3.647979974746704,
+      "learning_rate": 1.6080895166467948e-05,
+      "loss": 0.3191,
+      "step": 13480
+    },
+    {
+      "epoch": 2.653487743109341,
+      "grad_norm": 0.7913485169410706,
+      "learning_rate": 1.607783790394081e-05,
+      "loss": 0.3484,
+      "step": 13490
+    },
+    {
+      "epoch": 2.6554547466253595,
+      "grad_norm": 1.9412989616394043,
+      "learning_rate": 1.607478064141368e-05,
+      "loss": 0.3757,
+      "step": 13500
+    },
+    {
+      "epoch": 2.6554547466253595,
+      "eval_loss": 0.1958555430173874,
+      "eval_runtime": 8.9086,
+      "eval_samples_per_second": 5.613,
+      "eval_steps_per_second": 2.806,
+      "step": 13500
+    },
+    {
+      "epoch": 2.6574217501413786,
+      "grad_norm": 3.0294971466064453,
+      "learning_rate": 1.6071723378886545e-05,
+      "loss": 0.4377,
+      "step": 13510
+    },
+    {
+      "epoch": 2.6593887536573972,
+      "grad_norm": 1.4375206232070923,
+      "learning_rate": 1.6068666116359414e-05,
+      "loss": 0.4421,
+      "step": 13520
+    },
+    {
+      "epoch": 2.661355757173416,
+      "grad_norm": 2.2551164627075195,
+      "learning_rate": 1.606560885383228e-05,
+      "loss": 0.3295,
+      "step": 13530
+    },
+    {
+      "epoch": 2.663322760689435,
+      "grad_norm": 0.9871407747268677,
+      "learning_rate": 1.6062551591305146e-05,
+      "loss": 0.4265,
+      "step": 13540
+    },
+    {
+      "epoch": 2.6652897642054536,
+      "grad_norm": 2.210333824157715,
+      "learning_rate": 1.6059494328778015e-05,
+      "loss": 0.3181,
+      "step": 13550
+    },
+    {
+      "epoch": 2.6672567677214722,
+      "grad_norm": 1.154691219329834,
+      "learning_rate": 1.605643706625088e-05,
+      "loss": 0.5528,
+      "step": 13560
+    },
+    {
+      "epoch": 2.6692237712374913,
+      "grad_norm": 1.9619114398956299,
+      "learning_rate": 1.6053379803723746e-05,
+      "loss": 0.512,
+      "step": 13570
+    },
+    {
+      "epoch": 2.67119077475351,
+      "grad_norm": 1.5608044862747192,
+      "learning_rate": 1.6050322541196616e-05,
+      "loss": 0.4688,
+      "step": 13580
+    },
+    {
+      "epoch": 2.6731577782695286,
+      "grad_norm": 1.3780293464660645,
+      "learning_rate": 1.604726527866948e-05,
+      "loss": 0.3468,
+      "step": 13590
+    },
+    {
+      "epoch": 2.6751247817855477,
+      "grad_norm": 1.2746591567993164,
+      "learning_rate": 1.6044208016142347e-05,
+      "loss": 0.5138,
+      "step": 13600
+    },
+    {
+      "epoch": 2.6770917853015663,
+      "grad_norm": 1.2642594575881958,
+      "learning_rate": 1.6041150753615213e-05,
+      "loss": 0.4725,
+      "step": 13610
+    },
+    {
+      "epoch": 2.679058788817585,
+      "grad_norm": 0.9786370992660522,
+      "learning_rate": 1.603809349108808e-05,
+      "loss": 0.4651,
+      "step": 13620
+    },
+    {
+      "epoch": 2.681025792333604,
+      "grad_norm": 0.7670680284500122,
+      "learning_rate": 1.6035036228560948e-05,
+      "loss": 0.4079,
+      "step": 13630
+    },
+    {
+      "epoch": 2.6829927958496227,
+      "grad_norm": 1.5032764673233032,
+      "learning_rate": 1.6031978966033814e-05,
+      "loss": 0.4281,
+      "step": 13640
+    },
+    {
+      "epoch": 2.6849597993656413,
+      "grad_norm": 0.8874984383583069,
+      "learning_rate": 1.6028921703506683e-05,
+      "loss": 0.5697,
+      "step": 13650
+    },
+    {
+      "epoch": 2.6869268028816604,
+      "grad_norm": 1.297289252281189,
+      "learning_rate": 1.602586444097955e-05,
+      "loss": 0.4133,
+      "step": 13660
+    },
+    {
+      "epoch": 2.688893806397679,
+      "grad_norm": 1.6247835159301758,
+      "learning_rate": 1.6022807178452414e-05,
+      "loss": 0.3557,
+      "step": 13670
+    },
+    {
+      "epoch": 2.6908608099136977,
+      "grad_norm": 1.0644588470458984,
+      "learning_rate": 1.6019749915925284e-05,
+      "loss": 0.342,
+      "step": 13680
+    },
+    {
+      "epoch": 2.6928278134297168,
+      "grad_norm": 1.2811824083328247,
+      "learning_rate": 1.601669265339815e-05,
+      "loss": 0.4606,
+      "step": 13690
+    },
+    {
+      "epoch": 2.6947948169457354,
+      "grad_norm": 0.8294884562492371,
+      "learning_rate": 1.6013635390871015e-05,
+      "loss": 0.4232,
+      "step": 13700
+    },
+    {
+      "epoch": 2.696761820461754,
+      "grad_norm": 1.221997618675232,
+      "learning_rate": 1.6010578128343884e-05,
+      "loss": 0.4509,
+      "step": 13710
+    },
+    {
+      "epoch": 2.6987288239777727,
+      "grad_norm": 1.3223415613174438,
+      "learning_rate": 1.6007520865816747e-05,
+      "loss": 0.4079,
+      "step": 13720
+    },
+    {
+      "epoch": 2.700695827493792,
+      "grad_norm": 1.2597076892852783,
+      "learning_rate": 1.6004463603289616e-05,
+      "loss": 0.3619,
+      "step": 13730
+    },
+    {
+      "epoch": 2.7026628310098104,
+      "grad_norm": 1.868239402770996,
+      "learning_rate": 1.600140634076248e-05,
+      "loss": 0.3647,
+      "step": 13740
+    },
+    {
+      "epoch": 2.704629834525829,
+      "grad_norm": 2.783144950866699,
+      "learning_rate": 1.5998349078235347e-05,
+      "loss": 0.3667,
+      "step": 13750
+    },
+    {
+      "epoch": 2.706596838041848,
+      "grad_norm": 1.2698179483413696,
+      "learning_rate": 1.5995291815708217e-05,
+      "loss": 0.3574,
+      "step": 13760
+    },
+    {
+      "epoch": 2.708563841557867,
+      "grad_norm": 1.1568933725357056,
+      "learning_rate": 1.5992234553181082e-05,
+      "loss": 0.3992,
+      "step": 13770
+    },
+    {
+      "epoch": 2.7105308450738854,
+      "grad_norm": 1.916214108467102,
+      "learning_rate": 1.598917729065395e-05,
+      "loss": 0.4522,
+      "step": 13780
+    },
+    {
+      "epoch": 2.7124978485899045,
+      "grad_norm": 1.5209614038467407,
+      "learning_rate": 1.5986120028126817e-05,
+      "loss": 0.4657,
+      "step": 13790
+    },
+    {
+      "epoch": 2.714464852105923,
+      "grad_norm": 1.1678006649017334,
+      "learning_rate": 1.5983062765599683e-05,
+      "loss": 0.3623,
+      "step": 13800
+    },
+    {
+      "epoch": 2.716431855621942,
+      "grad_norm": 1.426422357559204,
+      "learning_rate": 1.5980005503072552e-05,
+      "loss": 0.3642,
+      "step": 13810
+    },
+    {
+      "epoch": 2.718398859137961,
+      "grad_norm": 1.0496702194213867,
+      "learning_rate": 1.5976948240545418e-05,
+      "loss": 0.3638,
+      "step": 13820
+    },
+    {
+      "epoch": 2.7203658626539795,
+      "grad_norm": 1.1832960844039917,
+      "learning_rate": 1.5973890978018284e-05,
+      "loss": 0.4138,
+      "step": 13830
+    },
+    {
+      "epoch": 2.722332866169998,
+      "grad_norm": 1.9487724304199219,
+      "learning_rate": 1.597083371549115e-05,
+      "loss": 0.3491,
+      "step": 13840
+    },
+    {
+      "epoch": 2.7242998696860172,
+      "grad_norm": 2.9646143913269043,
+      "learning_rate": 1.5967776452964015e-05,
+      "loss": 0.3892,
+      "step": 13850
+    },
+    {
+      "epoch": 2.726266873202036,
+      "grad_norm": 2.2359533309936523,
+      "learning_rate": 1.5964719190436885e-05,
+      "loss": 0.4241,
+      "step": 13860
+    },
+    {
+      "epoch": 2.7282338767180545,
+      "grad_norm": 1.4290543794631958,
+      "learning_rate": 1.596166192790975e-05,
+      "loss": 0.4208,
+      "step": 13870
+    },
+    {
+      "epoch": 2.730200880234073,
+      "grad_norm": 0.8418980240821838,
+      "learning_rate": 1.5958604665382616e-05,
+      "loss": 0.4289,
+      "step": 13880
+    },
+    {
+      "epoch": 2.7321678837500922,
+      "grad_norm": 1.897002100944519,
+      "learning_rate": 1.5955547402855485e-05,
+      "loss": 0.3951,
+      "step": 13890
+    },
+    {
+      "epoch": 2.734134887266111,
+      "grad_norm": 1.466009497642517,
+      "learning_rate": 1.595249014032835e-05,
+      "loss": 0.4793,
+      "step": 13900
+    },
+    {
+      "epoch": 2.7361018907821295,
+      "grad_norm": 1.0053349733352661,
+      "learning_rate": 1.594943287780122e-05,
+      "loss": 0.3532,
+      "step": 13910
+    },
+    {
+      "epoch": 2.7380688942981486,
+      "grad_norm": 1.6650139093399048,
+      "learning_rate": 1.5946375615274086e-05,
+      "loss": 0.3795,
+      "step": 13920
+    },
+    {
+      "epoch": 2.7400358978141672,
+      "grad_norm": 0.7641739845275879,
+      "learning_rate": 1.5943318352746952e-05,
+      "loss": 0.4616,
+      "step": 13930
+    },
+    {
+      "epoch": 2.742002901330186,
+      "grad_norm": 1.5295542478561401,
+      "learning_rate": 1.5940261090219818e-05,
+      "loss": 0.3941,
+      "step": 13940
+    },
+    {
+      "epoch": 2.743969904846205,
+      "grad_norm": 1.3583757877349854,
+      "learning_rate": 1.5937203827692683e-05,
+      "loss": 0.4931,
+      "step": 13950
+    },
+    {
+      "epoch": 2.7459369083622236,
+      "grad_norm": 1.5385262966156006,
+      "learning_rate": 1.5934146565165552e-05,
+      "loss": 0.5179,
+      "step": 13960
+    },
+    {
+      "epoch": 2.7479039118782422,
+      "grad_norm": 1.2843902111053467,
+      "learning_rate": 1.5931089302638418e-05,
+      "loss": 0.3554,
+      "step": 13970
+    },
+    {
+      "epoch": 2.7498709153942613,
+      "grad_norm": 1.251584529876709,
+      "learning_rate": 1.5928032040111284e-05,
+      "loss": 0.3453,
+      "step": 13980
+    },
+    {
+      "epoch": 2.75183791891028,
+      "grad_norm": 1.3551993370056152,
+      "learning_rate": 1.5924974777584153e-05,
+      "loss": 0.3805,
+      "step": 13990
+    },
+    {
+      "epoch": 2.7538049224262986,
+      "grad_norm": 0.8944595456123352,
+      "learning_rate": 1.592191751505702e-05,
+      "loss": 0.4317,
+      "step": 14000
+    },
+    {
+      "epoch": 2.7538049224262986,
+      "eval_loss": 0.19482757151126862,
+      "eval_runtime": 8.8976,
+      "eval_samples_per_second": 5.62,
+      "eval_steps_per_second": 2.81,
+      "step": 14000
+    },
+    {
+      "epoch": 2.7557719259423177,
+      "grad_norm": 1.8329250812530518,
+      "learning_rate": 1.5918860252529885e-05,
+      "loss": 0.4319,
+      "step": 14010
+    },
+    {
+      "epoch": 2.7577389294583363,
+      "grad_norm": 0.7794898748397827,
+      "learning_rate": 1.5915802990002754e-05,
+      "loss": 0.4589,
+      "step": 14020
+    },
+    {
+      "epoch": 2.759705932974355,
+      "grad_norm": 2.663428544998169,
+      "learning_rate": 1.591274572747562e-05,
+      "loss": 0.35,
+      "step": 14030
+    },
+    {
+      "epoch": 2.761672936490374,
+      "grad_norm": 1.2760578393936157,
+      "learning_rate": 1.590968846494849e-05,
+      "loss": 0.3787,
+      "step": 14040
+    },
+    {
+      "epoch": 2.7636399400063927,
+      "grad_norm": 0.9405483603477478,
+      "learning_rate": 1.5906631202421355e-05,
+      "loss": 0.3884,
+      "step": 14050
+    },
+    {
+      "epoch": 2.7656069435224113,
+      "grad_norm": 0.6196711659431458,
+      "learning_rate": 1.590357393989422e-05,
+      "loss": 0.307,
+      "step": 14060
+    },
+    {
+      "epoch": 2.7675739470384304,
+      "grad_norm": 0.8994119763374329,
+      "learning_rate": 1.5900516677367086e-05,
+      "loss": 0.4833,
+      "step": 14070
+    },
+    {
+      "epoch": 2.769540950554449,
+      "grad_norm": 1.5671933889389038,
+      "learning_rate": 1.5897459414839952e-05,
+      "loss": 0.4233,
+      "step": 14080
+    },
+    {
+      "epoch": 2.7715079540704677,
+      "grad_norm": 1.0534205436706543,
+      "learning_rate": 1.589440215231282e-05,
+      "loss": 0.5757,
+      "step": 14090
+    },
+    {
+      "epoch": 2.7734749575864868,
+      "grad_norm": 1.036082148551941,
+      "learning_rate": 1.5891344889785687e-05,
+      "loss": 0.4192,
+      "step": 14100
+    },
+    {
+      "epoch": 2.7754419611025054,
+      "grad_norm": 1.4796607494354248,
+      "learning_rate": 1.5888287627258553e-05,
+      "loss": 0.3789,
+      "step": 14110
+    },
+    {
+      "epoch": 2.777408964618524,
+      "grad_norm": 0.9435361623764038,
+      "learning_rate": 1.5885230364731422e-05,
+      "loss": 0.4361,
+      "step": 14120
+    },
+    {
+      "epoch": 2.779375968134543,
+      "grad_norm": 1.5666536092758179,
+      "learning_rate": 1.5882173102204288e-05,
+      "loss": 0.3985,
+      "step": 14130
+    },
+    {
+      "epoch": 2.7813429716505618,
+      "grad_norm": 1.0621715784072876,
+      "learning_rate": 1.5879115839677153e-05,
+      "loss": 0.3678,
+      "step": 14140
+    },
+    {
+      "epoch": 2.7833099751665804,
+      "grad_norm": 1.2044368982315063,
+      "learning_rate": 1.5876058577150023e-05,
+      "loss": 0.393,
+      "step": 14150
+    },
+    {
+      "epoch": 2.7852769786825995,
+      "grad_norm": 1.5973824262619019,
+      "learning_rate": 1.587300131462289e-05,
+      "loss": 0.3445,
+      "step": 14160
+    },
+    {
+      "epoch": 2.787243982198618,
+      "grad_norm": 0.9761004447937012,
+      "learning_rate": 1.5869944052095754e-05,
+      "loss": 0.5057,
+      "step": 14170
+    },
+    {
+      "epoch": 2.7892109857146368,
+      "grad_norm": 1.2603461742401123,
+      "learning_rate": 1.586688678956862e-05,
+      "loss": 0.3153,
+      "step": 14180
+    },
+    {
+      "epoch": 2.791177989230656,
+      "grad_norm": 1.165386438369751,
+      "learning_rate": 1.586382952704149e-05,
+      "loss": 0.4833,
+      "step": 14190
+    },
+    {
+      "epoch": 2.7931449927466745,
+      "grad_norm": 2.618959903717041,
+      "learning_rate": 1.5860772264514355e-05,
+      "loss": 0.4365,
+      "step": 14200
+    },
+    {
+      "epoch": 2.795111996262693,
+      "grad_norm": 1.329698920249939,
+      "learning_rate": 1.585771500198722e-05,
+      "loss": 0.5081,
+      "step": 14210
+    },
+    {
+      "epoch": 2.797078999778712,
+      "grad_norm": 1.7404251098632812,
+      "learning_rate": 1.585465773946009e-05,
+      "loss": 0.3237,
+      "step": 14220
+    },
+    {
+      "epoch": 2.799046003294731,
+      "grad_norm": 1.8908653259277344,
+      "learning_rate": 1.5851600476932956e-05,
+      "loss": 0.3845,
+      "step": 14230
+    },
+    {
+      "epoch": 2.8010130068107495,
+      "grad_norm": 4.6009907722473145,
+      "learning_rate": 1.584854321440582e-05,
+      "loss": 0.3441,
+      "step": 14240
+    },
+    {
+      "epoch": 2.8029800103267686,
+      "grad_norm": 1.369461178779602,
+      "learning_rate": 1.584548595187869e-05,
+      "loss": 0.3429,
+      "step": 14250
+    },
+    {
+      "epoch": 2.804947013842787,
+      "grad_norm": 1.4235303401947021,
+      "learning_rate": 1.5842428689351556e-05,
+      "loss": 0.3876,
+      "step": 14260
+    },
+    {
+      "epoch": 2.806914017358806,
+      "grad_norm": 1.7543208599090576,
+      "learning_rate": 1.5839371426824422e-05,
+      "loss": 0.3113,
+      "step": 14270
+    },
+    {
+      "epoch": 2.808881020874825,
+      "grad_norm": 1.4244465827941895,
+      "learning_rate": 1.5836314164297288e-05,
+      "loss": 0.3924,
+      "step": 14280
+    },
+    {
+      "epoch": 2.8108480243908436,
+      "grad_norm": 1.5675772428512573,
+      "learning_rate": 1.5833256901770157e-05,
+      "loss": 0.5018,
+      "step": 14290
+    },
+    {
+      "epoch": 2.812815027906862,
+      "grad_norm": 1.6984508037567139,
+      "learning_rate": 1.5830199639243023e-05,
+      "loss": 0.4307,
+      "step": 14300
+    },
+    {
+      "epoch": 2.8147820314228813,
+      "grad_norm": 1.7523006200790405,
+      "learning_rate": 1.582714237671589e-05,
+      "loss": 0.3569,
+      "step": 14310
+    },
+    {
+      "epoch": 2.8167490349389,
+      "grad_norm": 1.2517403364181519,
+      "learning_rate": 1.5824085114188758e-05,
+      "loss": 0.4189,
+      "step": 14320
+    },
+    {
+      "epoch": 2.8187160384549186,
+      "grad_norm": 1.5489752292633057,
+      "learning_rate": 1.5821027851661624e-05,
+      "loss": 0.4768,
+      "step": 14330
+    },
+    {
+      "epoch": 2.8206830419709377,
+      "grad_norm": 1.4975192546844482,
+      "learning_rate": 1.581797058913449e-05,
+      "loss": 0.4721,
+      "step": 14340
+    },
+    {
+      "epoch": 2.8226500454869563,
+      "grad_norm": 1.262944221496582,
+      "learning_rate": 1.581491332660736e-05,
+      "loss": 0.4397,
+      "step": 14350
+    },
+    {
+      "epoch": 2.824617049002975,
+      "grad_norm": 1.9267690181732178,
+      "learning_rate": 1.5811856064080224e-05,
+      "loss": 0.3516,
+      "step": 14360
+    },
+    {
+      "epoch": 2.826584052518994,
+      "grad_norm": 2.1961848735809326,
+      "learning_rate": 1.580879880155309e-05,
+      "loss": 0.5012,
+      "step": 14370
+    },
+    {
+      "epoch": 2.8285510560350127,
+      "grad_norm": 2.383462905883789,
+      "learning_rate": 1.580574153902596e-05,
+      "loss": 0.4569,
+      "step": 14380
+    },
+    {
+      "epoch": 2.8305180595510313,
+      "grad_norm": 1.119138240814209,
+      "learning_rate": 1.5802684276498825e-05,
+      "loss": 0.4949,
+      "step": 14390
+    },
+    {
+      "epoch": 2.8324850630670504,
+      "grad_norm": 2.2834503650665283,
+      "learning_rate": 1.579962701397169e-05,
+      "loss": 0.3773,
+      "step": 14400
+    },
+    {
+      "epoch": 2.834452066583069,
+      "grad_norm": 1.1680762767791748,
+      "learning_rate": 1.5796569751444557e-05,
+      "loss": 0.3865,
+      "step": 14410
+    },
+    {
+      "epoch": 2.8364190700990877,
+      "grad_norm": 1.060563564300537,
+      "learning_rate": 1.5793512488917426e-05,
+      "loss": 0.3549,
+      "step": 14420
+    },
+    {
+      "epoch": 2.8383860736151068,
+      "grad_norm": 1.6449270248413086,
+      "learning_rate": 1.579045522639029e-05,
+      "loss": 0.3986,
+      "step": 14430
+    },
+    {
+      "epoch": 2.8403530771311254,
+      "grad_norm": 1.8094450235366821,
+      "learning_rate": 1.5787397963863157e-05,
+      "loss": 0.4567,
+      "step": 14440
+    },
+    {
+      "epoch": 2.842320080647144,
+      "grad_norm": 1.4255567789077759,
+      "learning_rate": 1.5784340701336026e-05,
+      "loss": 0.3615,
+      "step": 14450
+    },
+    {
+      "epoch": 2.844287084163163,
+      "grad_norm": 1.9378465414047241,
+      "learning_rate": 1.5781283438808892e-05,
+      "loss": 0.4763,
+      "step": 14460
+    },
+    {
+      "epoch": 2.8462540876791818,
+      "grad_norm": 1.6682242155075073,
+      "learning_rate": 1.5778226176281758e-05,
+      "loss": 0.391,
+      "step": 14470
+    },
+    {
+      "epoch": 2.8482210911952004,
+      "grad_norm": 1.55122709274292,
+      "learning_rate": 1.5775168913754627e-05,
+      "loss": 0.3561,
+      "step": 14480
+    },
+    {
+      "epoch": 2.8501880947112195,
+      "grad_norm": 1.6349620819091797,
+      "learning_rate": 1.5772111651227493e-05,
+      "loss": 0.4355,
+      "step": 14490
+    },
+    {
+      "epoch": 2.852155098227238,
+      "grad_norm": 1.3233345746994019,
+      "learning_rate": 1.576905438870036e-05,
+      "loss": 0.4785,
+      "step": 14500
+    },
+    {
+      "epoch": 2.852155098227238,
+      "eval_loss": 0.1918381303548813,
+      "eval_runtime": 8.8922,
+      "eval_samples_per_second": 5.623,
+      "eval_steps_per_second": 2.811,
+      "step": 14500
+    },
+    {
+      "epoch": 2.8541221017432568,
+      "grad_norm": 1.6185466051101685,
+      "learning_rate": 1.5765997126173224e-05,
+      "loss": 0.4041,
+      "step": 14510
+    },
+    {
+      "epoch": 2.856089105259276,
+      "grad_norm": 1.1693966388702393,
+      "learning_rate": 1.576293986364609e-05,
+      "loss": 0.4266,
+      "step": 14520
+    },
+    {
+      "epoch": 2.8580561087752945,
+      "grad_norm": 1.2927526235580444,
+      "learning_rate": 1.575988260111896e-05,
+      "loss": 0.4679,
+      "step": 14530
+    },
+    {
+      "epoch": 2.860023112291313,
+      "grad_norm": 1.0233153104782104,
+      "learning_rate": 1.5756825338591825e-05,
+      "loss": 0.4132,
+      "step": 14540
+    },
+    {
+      "epoch": 2.861990115807332,
+      "grad_norm": 1.5996166467666626,
+      "learning_rate": 1.5753768076064694e-05,
+      "loss": 0.5182,
+      "step": 14550
+    },
+    {
+      "epoch": 2.863957119323351,
+      "grad_norm": 1.4874346256256104,
+      "learning_rate": 1.575071081353756e-05,
+      "loss": 0.4805,
+      "step": 14560
+    },
+    {
+      "epoch": 2.8659241228393695,
+      "grad_norm": 0.8183672428131104,
+      "learning_rate": 1.5747653551010426e-05,
+      "loss": 0.3395,
+      "step": 14570
+    },
+    {
+      "epoch": 2.8678911263553886,
+      "grad_norm": 1.1641387939453125,
+      "learning_rate": 1.5744596288483295e-05,
+      "loss": 0.4027,
+      "step": 14580
+    },
+    {
+      "epoch": 2.869858129871407,
+      "grad_norm": 0.9957535862922668,
+      "learning_rate": 1.574153902595616e-05,
+      "loss": 0.4174,
+      "step": 14590
+    },
+    {
+      "epoch": 2.871825133387426,
+      "grad_norm": 2.292351007461548,
+      "learning_rate": 1.5738481763429027e-05,
+      "loss": 0.4162,
+      "step": 14600
+    },
+    {
+      "epoch": 2.873792136903445,
+      "grad_norm": 1.9051101207733154,
+      "learning_rate": 1.5735424500901896e-05,
+      "loss": 0.3827,
+      "step": 14610
+    },
+    {
+      "epoch": 2.8757591404194636,
+      "grad_norm": 2.993645668029785,
+      "learning_rate": 1.5732367238374758e-05,
+      "loss": 0.4401,
+      "step": 14620
+    },
+    {
+      "epoch": 2.877726143935482,
+      "grad_norm": 1.3731578588485718,
+      "learning_rate": 1.5729309975847627e-05,
+      "loss": 0.4085,
+      "step": 14630
+    },
+    {
+      "epoch": 2.8796931474515013,
+      "grad_norm": 1.1569373607635498,
+      "learning_rate": 1.5726252713320493e-05,
+      "loss": 0.4803,
+      "step": 14640
+    },
+    {
+      "epoch": 2.88166015096752,
+      "grad_norm": 1.196911334991455,
+      "learning_rate": 1.572319545079336e-05,
+      "loss": 0.5282,
+      "step": 14650
+    },
+    {
+      "epoch": 2.8836271544835386,
+      "grad_norm": 1.6104505062103271,
+      "learning_rate": 1.5720138188266228e-05,
+      "loss": 0.5026,
+      "step": 14660
+    },
+    {
+      "epoch": 2.8855941579995577,
+      "grad_norm": 1.437827229499817,
+      "learning_rate": 1.5717080925739094e-05,
+      "loss": 0.5583,
+      "step": 14670
+    },
+    {
+      "epoch": 2.8875611615155763,
+      "grad_norm": 1.507562518119812,
+      "learning_rate": 1.5714023663211963e-05,
+      "loss": 0.3675,
+      "step": 14680
+    },
+    {
+      "epoch": 2.889528165031595,
+      "grad_norm": 1.2080801725387573,
+      "learning_rate": 1.571096640068483e-05,
+      "loss": 0.4364,
+      "step": 14690
+    },
+    {
+      "epoch": 2.891495168547614,
+      "grad_norm": 1.4376025199890137,
+      "learning_rate": 1.5707909138157695e-05,
+      "loss": 0.3281,
+      "step": 14700
+    },
+    {
+      "epoch": 2.8934621720636327,
+      "grad_norm": 0.8100415468215942,
+      "learning_rate": 1.5704851875630564e-05,
+      "loss": 0.343,
+      "step": 14710
+    },
+    {
+      "epoch": 2.8954291755796513,
+      "grad_norm": 1.3493585586547852,
+      "learning_rate": 1.570179461310343e-05,
+      "loss": 0.5444,
+      "step": 14720
+    },
+    {
+      "epoch": 2.8973961790956704,
+      "grad_norm": 0.9113426208496094,
+      "learning_rate": 1.5698737350576295e-05,
+      "loss": 0.3933,
+      "step": 14730
+    },
+    {
+      "epoch": 2.899363182611689,
+      "grad_norm": 0.9956138134002686,
+      "learning_rate": 1.569568008804916e-05,
+      "loss": 0.3959,
+      "step": 14740
+    },
+    {
+      "epoch": 2.9013301861277077,
+      "grad_norm": 1.8397066593170166,
+      "learning_rate": 1.5692622825522027e-05,
+      "loss": 0.4507,
+      "step": 14750
+    },
+    {
+      "epoch": 2.9032971896437267,
+      "grad_norm": 1.2363187074661255,
+      "learning_rate": 1.5689565562994896e-05,
+      "loss": 0.4155,
+      "step": 14760
+    },
+    {
+      "epoch": 2.9052641931597454,
+      "grad_norm": 1.7232961654663086,
+      "learning_rate": 1.5686508300467762e-05,
+      "loss": 0.3256,
+      "step": 14770
+    },
+    {
+      "epoch": 2.907231196675764,
+      "grad_norm": 2.252438545227051,
+      "learning_rate": 1.5683451037940628e-05,
+      "loss": 0.6026,
+      "step": 14780
+    },
+    {
+      "epoch": 2.909198200191783,
+      "grad_norm": 3.091703414916992,
+      "learning_rate": 1.5680393775413497e-05,
+      "loss": 0.5315,
+      "step": 14790
+    },
+    {
+      "epoch": 2.9111652037078017,
+      "grad_norm": 0.9878594279289246,
+      "learning_rate": 1.5677336512886362e-05,
+      "loss": 0.4185,
+      "step": 14800
+    },
+    {
+      "epoch": 2.9131322072238204,
+      "grad_norm": 2.0725889205932617,
+      "learning_rate": 1.567427925035923e-05,
+      "loss": 0.4641,
+      "step": 14810
+    },
+    {
+      "epoch": 2.9150992107398395,
+      "grad_norm": 0.8847138285636902,
+      "learning_rate": 1.5671221987832097e-05,
+      "loss": 0.4097,
+      "step": 14820
+    },
+    {
+      "epoch": 2.917066214255858,
+      "grad_norm": 2.7409422397613525,
+      "learning_rate": 1.5668164725304963e-05,
+      "loss": 0.3762,
+      "step": 14830
+    },
+    {
+      "epoch": 2.9190332177718767,
+      "grad_norm": 1.3184597492218018,
+      "learning_rate": 1.5665107462777832e-05,
+      "loss": 0.3953,
+      "step": 14840
+    },
+    {
+      "epoch": 2.921000221287896,
+      "grad_norm": 1.068154215812683,
+      "learning_rate": 1.5662050200250695e-05,
+      "loss": 0.5554,
+      "step": 14850
+    },
+    {
+      "epoch": 2.9229672248039145,
+      "grad_norm": 1.1467418670654297,
+      "learning_rate": 1.5658992937723564e-05,
+      "loss": 0.4592,
+      "step": 14860
+    },
+    {
+      "epoch": 2.924934228319933,
+      "grad_norm": 1.156706690788269,
+      "learning_rate": 1.565593567519643e-05,
+      "loss": 0.3614,
+      "step": 14870
+    },
+    {
+      "epoch": 2.9269012318359517,
+      "grad_norm": 1.4045031070709229,
+      "learning_rate": 1.5652878412669295e-05,
+      "loss": 0.3408,
+      "step": 14880
+    },
+    {
+      "epoch": 2.928868235351971,
+      "grad_norm": 1.043555498123169,
+      "learning_rate": 1.5649821150142165e-05,
+      "loss": 0.3368,
+      "step": 14890
+    },
+    {
+      "epoch": 2.9308352388679895,
+      "grad_norm": 1.4246408939361572,
+      "learning_rate": 1.564676388761503e-05,
+      "loss": 0.3734,
+      "step": 14900
+    },
+    {
+      "epoch": 2.932802242384008,
+      "grad_norm": 1.4614734649658203,
+      "learning_rate": 1.5643706625087896e-05,
+      "loss": 0.4164,
+      "step": 14910
+    },
+    {
+      "epoch": 2.934769245900027,
+      "grad_norm": 1.4192919731140137,
+      "learning_rate": 1.5640649362560765e-05,
+      "loss": 0.487,
+      "step": 14920
+    },
+    {
+      "epoch": 2.936736249416046,
+      "grad_norm": 1.1312637329101562,
+      "learning_rate": 1.563759210003363e-05,
+      "loss": 0.4026,
+      "step": 14930
+    },
+    {
+      "epoch": 2.9387032529320645,
+      "grad_norm": 1.9668546915054321,
+      "learning_rate": 1.56345348375065e-05,
+      "loss": 0.4553,
+      "step": 14940
+    },
+    {
+      "epoch": 2.9406702564480836,
+      "grad_norm": 0.8810634613037109,
+      "learning_rate": 1.5631477574979366e-05,
+      "loss": 0.429,
+      "step": 14950
+    },
+    {
+      "epoch": 2.942637259964102,
+      "grad_norm": 1.1697512865066528,
+      "learning_rate": 1.5628420312452232e-05,
+      "loss": 0.3331,
+      "step": 14960
+    },
+    {
+      "epoch": 2.944604263480121,
+      "grad_norm": 0.7395417094230652,
+      "learning_rate": 1.5625363049925098e-05,
+      "loss": 0.3326,
+      "step": 14970
+    },
+    {
+      "epoch": 2.94657126699614,
+      "grad_norm": 1.3220033645629883,
+      "learning_rate": 1.5622305787397963e-05,
+      "loss": 0.5424,
+      "step": 14980
+    },
+    {
+      "epoch": 2.9485382705121586,
+      "grad_norm": 1.2082237005233765,
+      "learning_rate": 1.5619248524870833e-05,
+      "loss": 0.3298,
+      "step": 14990
+    },
+    {
+      "epoch": 2.950505274028177,
+      "grad_norm": 1.1049548387527466,
+      "learning_rate": 1.56161912623437e-05,
+      "loss": 0.4071,
+      "step": 15000
+    },
+    {
+      "epoch": 2.950505274028177,
+      "eval_loss": 0.19573713839054108,
+      "eval_runtime": 8.8769,
+      "eval_samples_per_second": 5.633,
+      "eval_steps_per_second": 2.816,
+      "step": 15000
+    },
+    {
+      "epoch": 2.952472277544196,
+      "grad_norm": 1.520330786705017,
+      "learning_rate": 1.5613133999816564e-05,
+      "loss": 0.4873,
+      "step": 15010
+    },
+    {
+      "epoch": 2.954439281060215,
+      "grad_norm": 2.7328922748565674,
+      "learning_rate": 1.5610076737289433e-05,
+      "loss": 0.3784,
+      "step": 15020
+    },
+    {
+      "epoch": 2.9564062845762336,
+      "grad_norm": 1.440152645111084,
+      "learning_rate": 1.56070194747623e-05,
+      "loss": 0.3922,
+      "step": 15030
+    },
+    {
+      "epoch": 2.958373288092252,
+      "grad_norm": 3.501024007797241,
+      "learning_rate": 1.5603962212235165e-05,
+      "loss": 0.3912,
+      "step": 15040
+    },
+    {
+      "epoch": 2.9603402916082713,
+      "grad_norm": 2.0499727725982666,
+      "learning_rate": 1.5600904949708034e-05,
+      "loss": 0.403,
+      "step": 15050
+    },
+    {
+      "epoch": 2.96230729512429,
+      "grad_norm": 1.093933343887329,
+      "learning_rate": 1.55978476871809e-05,
+      "loss": 0.4449,
+      "step": 15060
+    },
+    {
+      "epoch": 2.9642742986403086,
+      "grad_norm": 2.806871175765991,
+      "learning_rate": 1.559479042465377e-05,
+      "loss": 0.3951,
+      "step": 15070
+    },
+    {
+      "epoch": 2.9662413021563276,
+      "grad_norm": 1.279954433441162,
+      "learning_rate": 1.559173316212663e-05,
+      "loss": 0.3745,
+      "step": 15080
+    },
+    {
+      "epoch": 2.9682083056723463,
+      "grad_norm": 1.3171770572662354,
+      "learning_rate": 1.55886758995995e-05,
+      "loss": 0.6198,
+      "step": 15090
+    },
+    {
+      "epoch": 2.970175309188365,
+      "grad_norm": 1.5054320096969604,
+      "learning_rate": 1.5585618637072366e-05,
+      "loss": 0.4462,
+      "step": 15100
+    },
+    {
+      "epoch": 2.972142312704384,
+      "grad_norm": 2.4180641174316406,
+      "learning_rate": 1.5582561374545232e-05,
+      "loss": 0.3919,
+      "step": 15110
+    },
+    {
+      "epoch": 2.9741093162204026,
+      "grad_norm": 1.1554861068725586,
+      "learning_rate": 1.55795041120181e-05,
+      "loss": 0.4175,
+      "step": 15120
+    },
+    {
+      "epoch": 2.9760763197364213,
+      "grad_norm": 1.2998076677322388,
+      "learning_rate": 1.5576446849490967e-05,
+      "loss": 0.3055,
+      "step": 15130
+    },
+    {
+      "epoch": 2.9780433232524404,
+      "grad_norm": 1.6554224491119385,
+      "learning_rate": 1.5573389586963833e-05,
+      "loss": 0.3742,
+      "step": 15140
+    },
+    {
+      "epoch": 2.980010326768459,
+      "grad_norm": 1.5794579982757568,
+      "learning_rate": 1.5570332324436702e-05,
+      "loss": 0.3416,
+      "step": 15150
+    },
+    {
+      "epoch": 2.9819773302844776,
+      "grad_norm": 1.393416404724121,
+      "learning_rate": 1.5567275061909568e-05,
+      "loss": 0.3949,
+      "step": 15160
+    },
+    {
+      "epoch": 2.9839443338004967,
+      "grad_norm": 1.9517128467559814,
+      "learning_rate": 1.5564217799382434e-05,
+      "loss": 0.4206,
+      "step": 15170
+    },
+    {
+      "epoch": 2.9859113373165154,
+      "grad_norm": 1.381700873374939,
+      "learning_rate": 1.5561160536855303e-05,
+      "loss": 0.3732,
+      "step": 15180
+    },
+    {
+      "epoch": 2.987878340832534,
+      "grad_norm": 2.750070095062256,
+      "learning_rate": 1.555810327432817e-05,
+      "loss": 0.4142,
+      "step": 15190
+    },
+    {
+      "epoch": 2.989845344348553,
+      "grad_norm": 1.1984093189239502,
+      "learning_rate": 1.5555046011801034e-05,
+      "loss": 0.5847,
+      "step": 15200
+    },
+    {
+      "epoch": 2.9918123478645717,
+      "grad_norm": 0.7176037430763245,
+      "learning_rate": 1.55519887492739e-05,
+      "loss": 0.324,
+      "step": 15210
+    },
+    {
+      "epoch": 2.9937793513805904,
+      "grad_norm": 1.6049420833587646,
+      "learning_rate": 1.554893148674677e-05,
+      "loss": 0.4589,
+      "step": 15220
+    },
+    {
+      "epoch": 2.9957463548966095,
+      "grad_norm": 0.9500333070755005,
+      "learning_rate": 1.5545874224219635e-05,
+      "loss": 0.3222,
+      "step": 15230
+    },
+    {
+      "epoch": 2.997713358412628,
+      "grad_norm": 2.22035551071167,
+      "learning_rate": 1.55428169616925e-05,
+      "loss": 0.4337,
+      "step": 15240
+    },
+    {
+      "epoch": 2.9996803619286467,
+      "grad_norm": 0.9914708733558655,
+      "learning_rate": 1.553975969916537e-05,
+      "loss": 0.435,
+      "step": 15250
+    },
+    {
+      "epoch": 3.001647365444666,
+      "grad_norm": 1.351025938987732,
+      "learning_rate": 1.5536702436638236e-05,
+      "loss": 0.3849,
+      "step": 15260
+    },
+    {
+      "epoch": 3.0036143689606845,
+      "grad_norm": 1.5382455587387085,
+      "learning_rate": 1.55336451741111e-05,
+      "loss": 0.3485,
+      "step": 15270
+    },
+    {
+      "epoch": 3.005581372476703,
+      "grad_norm": 1.7761247158050537,
+      "learning_rate": 1.553058791158397e-05,
+      "loss": 0.3946,
+      "step": 15280
+    },
+    {
+      "epoch": 3.007548375992722,
+      "grad_norm": 1.7743333578109741,
+      "learning_rate": 1.5527530649056836e-05,
+      "loss": 0.4532,
+      "step": 15290
+    },
+    {
+      "epoch": 3.009515379508741,
+      "grad_norm": 1.4908400774002075,
+      "learning_rate": 1.5524473386529702e-05,
+      "loss": 0.3707,
+      "step": 15300
+    },
+    {
+      "epoch": 3.0114823830247595,
+      "grad_norm": 1.8430463075637817,
+      "learning_rate": 1.5521416124002568e-05,
+      "loss": 0.5074,
+      "step": 15310
+    },
+    {
+      "epoch": 3.0134493865407785,
+      "grad_norm": 1.3990598917007446,
+      "learning_rate": 1.5518358861475437e-05,
+      "loss": 0.5073,
+      "step": 15320
+    },
+    {
+      "epoch": 3.015416390056797,
+      "grad_norm": 1.564259648323059,
+      "learning_rate": 1.5515301598948303e-05,
+      "loss": 0.3635,
+      "step": 15330
+    },
+    {
+      "epoch": 3.017383393572816,
+      "grad_norm": 2.144291877746582,
+      "learning_rate": 1.551224433642117e-05,
+      "loss": 0.408,
+      "step": 15340
+    },
+    {
+      "epoch": 3.019350397088835,
+      "grad_norm": 0.6599649786949158,
+      "learning_rate": 1.5509187073894038e-05,
+      "loss": 0.4184,
+      "step": 15350
+    },
+    {
+      "epoch": 3.0213174006048535,
+      "grad_norm": 1.5132403373718262,
+      "learning_rate": 1.5506129811366904e-05,
+      "loss": 0.3938,
+      "step": 15360
+    },
+    {
+      "epoch": 3.023284404120872,
+      "grad_norm": 1.6984745264053345,
+      "learning_rate": 1.550307254883977e-05,
+      "loss": 0.5159,
+      "step": 15370
+    },
+    {
+      "epoch": 3.0252514076368913,
+      "grad_norm": 0.92393958568573,
+      "learning_rate": 1.550001528631264e-05,
+      "loss": 0.399,
+      "step": 15380
+    },
+    {
+      "epoch": 3.02721841115291,
+      "grad_norm": 1.139657974243164,
+      "learning_rate": 1.5496958023785504e-05,
+      "loss": 0.3417,
+      "step": 15390
+    },
+    {
+      "epoch": 3.0291854146689285,
+      "grad_norm": 0.8047592043876648,
+      "learning_rate": 1.549390076125837e-05,
+      "loss": 0.4166,
+      "step": 15400
+    },
+    {
+      "epoch": 3.0311524181849476,
+      "grad_norm": 2.0489370822906494,
+      "learning_rate": 1.549084349873124e-05,
+      "loss": 0.3989,
+      "step": 15410
+    },
+    {
+      "epoch": 3.0331194217009663,
+      "grad_norm": 1.3521877527236938,
+      "learning_rate": 1.5487786236204102e-05,
+      "loss": 0.3956,
+      "step": 15420
+    },
+    {
+      "epoch": 3.035086425216985,
+      "grad_norm": 1.4160988330841064,
+      "learning_rate": 1.548472897367697e-05,
+      "loss": 0.3439,
+      "step": 15430
+    },
+    {
+      "epoch": 3.037053428733004,
+      "grad_norm": 1.4726520776748657,
+      "learning_rate": 1.5481671711149837e-05,
+      "loss": 0.3935,
+      "step": 15440
+    },
+    {
+      "epoch": 3.0390204322490226,
+      "grad_norm": 0.754563570022583,
+      "learning_rate": 1.5478614448622702e-05,
+      "loss": 0.5188,
+      "step": 15450
+    },
+    {
+      "epoch": 3.0409874357650413,
+      "grad_norm": 3.205728769302368,
+      "learning_rate": 1.547555718609557e-05,
+      "loss": 0.3125,
+      "step": 15460
+    },
+    {
+      "epoch": 3.0429544392810604,
+      "grad_norm": 2.270024538040161,
+      "learning_rate": 1.5472499923568437e-05,
+      "loss": 0.512,
+      "step": 15470
+    },
+    {
+      "epoch": 3.044921442797079,
+      "grad_norm": 2.011733055114746,
+      "learning_rate": 1.5469442661041307e-05,
+      "loss": 0.3769,
+      "step": 15480
+    },
+    {
+      "epoch": 3.0468884463130976,
+      "grad_norm": 1.1816586256027222,
+      "learning_rate": 1.5466385398514172e-05,
+      "loss": 0.3847,
+      "step": 15490
+    },
+    {
+      "epoch": 3.0488554498291167,
+      "grad_norm": 1.7889635562896729,
+      "learning_rate": 1.5463328135987038e-05,
+      "loss": 0.3609,
+      "step": 15500
+    },
+    {
+      "epoch": 3.0488554498291167,
+      "eval_loss": 0.1888391375541687,
+      "eval_runtime": 8.8733,
+      "eval_samples_per_second": 5.635,
+      "eval_steps_per_second": 2.817,
+      "step": 15500
+    },
+    {
+      "epoch": 3.0508224533451354,
+      "grad_norm": 1.3694669008255005,
+      "learning_rate": 1.5460270873459907e-05,
+      "loss": 0.4962,
+      "step": 15510
+    },
+    {
+      "epoch": 3.052789456861154,
+      "grad_norm": 0.9643653035163879,
+      "learning_rate": 1.5457213610932773e-05,
+      "loss": 0.4182,
+      "step": 15520
+    },
+    {
+      "epoch": 3.054756460377173,
+      "grad_norm": 1.9267834424972534,
+      "learning_rate": 1.545415634840564e-05,
+      "loss": 0.4511,
+      "step": 15530
+    },
+    {
+      "epoch": 3.0567234638931917,
+      "grad_norm": 1.3095884323120117,
+      "learning_rate": 1.5451099085878505e-05,
+      "loss": 0.3976,
+      "step": 15540
+    },
+    {
+      "epoch": 3.0586904674092104,
+      "grad_norm": 0.8985733389854431,
+      "learning_rate": 1.544804182335137e-05,
+      "loss": 0.4533,
+      "step": 15550
+    },
+    {
+      "epoch": 3.0606574709252294,
+      "grad_norm": 1.9100348949432373,
+      "learning_rate": 1.544498456082424e-05,
+      "loss": 0.3699,
+      "step": 15560
+    },
+    {
+      "epoch": 3.062624474441248,
+      "grad_norm": 2.1582112312316895,
+      "learning_rate": 1.5441927298297105e-05,
+      "loss": 0.3652,
+      "step": 15570
+    },
+    {
+      "epoch": 3.0645914779572667,
+      "grad_norm": 0.8692134022712708,
+      "learning_rate": 1.543887003576997e-05,
+      "loss": 0.4368,
+      "step": 15580
+    },
+    {
+      "epoch": 3.066558481473286,
+      "grad_norm": 0.8661279082298279,
+      "learning_rate": 1.543581277324284e-05,
+      "loss": 0.3397,
+      "step": 15590
+    },
+    {
+      "epoch": 3.0685254849893044,
+      "grad_norm": 0.8110594153404236,
+      "learning_rate": 1.5432755510715706e-05,
+      "loss": 0.3904,
+      "step": 15600
+    },
+    {
+      "epoch": 3.070492488505323,
+      "grad_norm": 1.5086452960968018,
+      "learning_rate": 1.5429698248188575e-05,
+      "loss": 0.3595,
+      "step": 15610
+    },
+    {
+      "epoch": 3.072459492021342,
+      "grad_norm": 1.3329867124557495,
+      "learning_rate": 1.542664098566144e-05,
+      "loss": 0.4274,
+      "step": 15620
+    },
+    {
+      "epoch": 3.074426495537361,
+      "grad_norm": 1.3834015130996704,
+      "learning_rate": 1.5423583723134307e-05,
+      "loss": 0.3732,
+      "step": 15630
+    },
+    {
+      "epoch": 3.0763934990533794,
+      "grad_norm": 1.0691471099853516,
+      "learning_rate": 1.5420526460607173e-05,
+      "loss": 0.4107,
+      "step": 15640
+    },
+    {
+      "epoch": 3.0783605025693985,
+      "grad_norm": 2.663893938064575,
+      "learning_rate": 1.541746919808004e-05,
+      "loss": 0.4178,
+      "step": 15650
+    },
+    {
+      "epoch": 3.080327506085417,
+      "grad_norm": 1.1573967933654785,
+      "learning_rate": 1.5414411935552907e-05,
+      "loss": 0.3972,
+      "step": 15660
+    },
+    {
+      "epoch": 3.082294509601436,
+      "grad_norm": 1.3600716590881348,
+      "learning_rate": 1.5411354673025773e-05,
+      "loss": 0.361,
+      "step": 15670
+    },
+    {
+      "epoch": 3.084261513117455,
+      "grad_norm": 1.168461799621582,
+      "learning_rate": 1.540829741049864e-05,
+      "loss": 0.4059,
+      "step": 15680
+    },
+    {
+      "epoch": 3.0862285166334735,
+      "grad_norm": 1.0328609943389893,
+      "learning_rate": 1.5405240147971508e-05,
+      "loss": 0.491,
+      "step": 15690
+    },
+    {
+      "epoch": 3.088195520149492,
+      "grad_norm": 1.7007650136947632,
+      "learning_rate": 1.5402182885444374e-05,
+      "loss": 0.3999,
+      "step": 15700
+    },
+    {
+      "epoch": 3.0901625236655113,
+      "grad_norm": 1.2272205352783203,
+      "learning_rate": 1.539912562291724e-05,
+      "loss": 0.2953,
+      "step": 15710
+    },
+    {
+      "epoch": 3.09212952718153,
+      "grad_norm": 1.079796314239502,
+      "learning_rate": 1.539606836039011e-05,
+      "loss": 0.3475,
+      "step": 15720
+    },
+    {
+      "epoch": 3.0940965306975485,
+      "grad_norm": 0.5963640213012695,
+      "learning_rate": 1.5393011097862975e-05,
+      "loss": 0.4842,
+      "step": 15730
+    },
+    {
+      "epoch": 3.0960635342135676,
+      "grad_norm": 1.291596531867981,
+      "learning_rate": 1.5389953835335844e-05,
+      "loss": 0.4387,
+      "step": 15740
+    },
+    {
+      "epoch": 3.0980305377295863,
+      "grad_norm": 1.0582354068756104,
+      "learning_rate": 1.538689657280871e-05,
+      "loss": 0.3131,
+      "step": 15750
+    },
+    {
+      "epoch": 3.099997541245605,
+      "grad_norm": 1.327975869178772,
+      "learning_rate": 1.5383839310281575e-05,
+      "loss": 0.3869,
+      "step": 15760
+    },
+    {
+      "epoch": 3.101964544761624,
+      "grad_norm": 3.4642159938812256,
+      "learning_rate": 1.538078204775444e-05,
+      "loss": 0.4543,
+      "step": 15770
+    },
+    {
+      "epoch": 3.1039315482776426,
+      "grad_norm": 1.4730603694915771,
+      "learning_rate": 1.5377724785227307e-05,
+      "loss": 0.3647,
+      "step": 15780
+    },
+    {
+      "epoch": 3.1058985517936613,
+      "grad_norm": 3.070542573928833,
+      "learning_rate": 1.5374667522700176e-05,
+      "loss": 0.409,
+      "step": 15790
+    },
+    {
+      "epoch": 3.1078655553096803,
+      "grad_norm": 1.379279375076294,
+      "learning_rate": 1.5371610260173042e-05,
+      "loss": 0.2932,
+      "step": 15800
+    },
+    {
+      "epoch": 3.109832558825699,
+      "grad_norm": 1.639320969581604,
+      "learning_rate": 1.5368552997645908e-05,
+      "loss": 0.3491,
+      "step": 15810
+    },
+    {
+      "epoch": 3.1117995623417176,
+      "grad_norm": 1.58949875831604,
+      "learning_rate": 1.5365495735118777e-05,
+      "loss": 0.3182,
+      "step": 15820
+    },
+    {
+      "epoch": 3.1137665658577363,
+      "grad_norm": 1.9512661695480347,
+      "learning_rate": 1.5362438472591643e-05,
+      "loss": 0.342,
+      "step": 15830
+    },
+    {
+      "epoch": 3.1157335693737553,
+      "grad_norm": 1.656012773513794,
+      "learning_rate": 1.535938121006451e-05,
+      "loss": 0.409,
+      "step": 15840
+    },
+    {
+      "epoch": 3.117700572889774,
+      "grad_norm": 1.1831879615783691,
+      "learning_rate": 1.5356323947537378e-05,
+      "loss": 0.321,
+      "step": 15850
+    },
+    {
+      "epoch": 3.1196675764057926,
+      "grad_norm": 1.4055463075637817,
+      "learning_rate": 1.5353266685010243e-05,
+      "loss": 0.3609,
+      "step": 15860
+    },
+    {
+      "epoch": 3.1216345799218117,
+      "grad_norm": 1.3060204982757568,
+      "learning_rate": 1.535020942248311e-05,
+      "loss": 0.3488,
+      "step": 15870
+    },
+    {
+      "epoch": 3.1236015834378303,
+      "grad_norm": 1.2678416967391968,
+      "learning_rate": 1.5347152159955975e-05,
+      "loss": 0.3194,
+      "step": 15880
+    },
+    {
+      "epoch": 3.125568586953849,
+      "grad_norm": 0.9447997808456421,
+      "learning_rate": 1.5344094897428844e-05,
+      "loss": 0.4053,
+      "step": 15890
+    },
+    {
+      "epoch": 3.127535590469868,
+      "grad_norm": 1.1538068056106567,
+      "learning_rate": 1.534103763490171e-05,
+      "loss": 0.3742,
+      "step": 15900
+    },
+    {
+      "epoch": 3.1295025939858867,
+      "grad_norm": 1.4947891235351562,
+      "learning_rate": 1.5337980372374576e-05,
+      "loss": 0.3913,
+      "step": 15910
+    },
+    {
+      "epoch": 3.1314695975019053,
+      "grad_norm": 1.4566258192062378,
+      "learning_rate": 1.5334923109847445e-05,
+      "loss": 0.3162,
+      "step": 15920
+    },
+    {
+      "epoch": 3.1334366010179244,
+      "grad_norm": 1.3656127452850342,
+      "learning_rate": 1.533186584732031e-05,
+      "loss": 0.2902,
+      "step": 15930
+    },
+    {
+      "epoch": 3.135403604533943,
+      "grad_norm": 0.8082563877105713,
+      "learning_rate": 1.5328808584793176e-05,
+      "loss": 0.3822,
+      "step": 15940
+    },
+    {
+      "epoch": 3.1373706080499617,
+      "grad_norm": 0.9663358330726624,
+      "learning_rate": 1.5325751322266046e-05,
+      "loss": 0.3285,
+      "step": 15950
+    },
+    {
+      "epoch": 3.139337611565981,
+      "grad_norm": 1.7113450765609741,
+      "learning_rate": 1.532269405973891e-05,
+      "loss": 0.3592,
+      "step": 15960
+    },
+    {
+      "epoch": 3.1413046150819994,
+      "grad_norm": 2.219865322113037,
+      "learning_rate": 1.5319636797211777e-05,
+      "loss": 0.469,
+      "step": 15970
+    },
+    {
+      "epoch": 3.143271618598018,
+      "grad_norm": 1.3224714994430542,
+      "learning_rate": 1.5316579534684643e-05,
+      "loss": 0.4137,
+      "step": 15980
+    },
+    {
+      "epoch": 3.145238622114037,
+      "grad_norm": 2.301541328430176,
+      "learning_rate": 1.5313522272157512e-05,
+      "loss": 0.3628,
+      "step": 15990
+    },
+    {
+      "epoch": 3.147205625630056,
+      "grad_norm": 1.2435368299484253,
+      "learning_rate": 1.5310465009630378e-05,
+      "loss": 0.3875,
+      "step": 16000
+    },
+    {
+      "epoch": 3.147205625630056,
+      "eval_loss": 0.18709848821163177,
+      "eval_runtime": 8.8741,
+      "eval_samples_per_second": 5.634,
+      "eval_steps_per_second": 2.817,
+      "step": 16000
+    },
+    {
+      "epoch": 3.1491726291460744,
+      "grad_norm": 1.5594910383224487,
+      "learning_rate": 1.5307407747103244e-05,
+      "loss": 0.4676,
+      "step": 16010
+    },
+    {
+      "epoch": 3.1511396326620935,
+      "grad_norm": 2.1316165924072266,
+      "learning_rate": 1.5304350484576113e-05,
+      "loss": 0.4096,
+      "step": 16020
+    },
+    {
+      "epoch": 3.153106636178112,
+      "grad_norm": 1.002728819847107,
+      "learning_rate": 1.530129322204898e-05,
+      "loss": 0.3481,
+      "step": 16030
+    },
+    {
+      "epoch": 3.155073639694131,
+      "grad_norm": 1.6484594345092773,
+      "learning_rate": 1.5298235959521844e-05,
+      "loss": 0.4777,
+      "step": 16040
+    },
+    {
+      "epoch": 3.15704064321015,
+      "grad_norm": 1.0128759145736694,
+      "learning_rate": 1.5295178696994713e-05,
+      "loss": 0.3421,
+      "step": 16050
+    },
+    {
+      "epoch": 3.1590076467261685,
+      "grad_norm": 1.290736436843872,
+      "learning_rate": 1.529212143446758e-05,
+      "loss": 0.3759,
+      "step": 16060
+    },
+    {
+      "epoch": 3.160974650242187,
+      "grad_norm": 2.4858880043029785,
+      "learning_rate": 1.5289064171940445e-05,
+      "loss": 0.3818,
+      "step": 16070
+    },
+    {
+      "epoch": 3.1629416537582062,
+      "grad_norm": 1.9373151063919067,
+      "learning_rate": 1.5286006909413314e-05,
+      "loss": 0.364,
+      "step": 16080
+    },
+    {
+      "epoch": 3.164908657274225,
+      "grad_norm": 2.771146297454834,
+      "learning_rate": 1.528294964688618e-05,
+      "loss": 0.3584,
+      "step": 16090
+    },
+    {
+      "epoch": 3.1668756607902435,
+      "grad_norm": 1.1857792139053345,
+      "learning_rate": 1.5279892384359046e-05,
+      "loss": 0.4443,
+      "step": 16100
+    },
+    {
+      "epoch": 3.1688426643062626,
+      "grad_norm": 1.4357842206954956,
+      "learning_rate": 1.527683512183191e-05,
+      "loss": 0.3047,
+      "step": 16110
+    },
+    {
+      "epoch": 3.1708096678222812,
+      "grad_norm": 1.4366753101348877,
+      "learning_rate": 1.527377785930478e-05,
+      "loss": 0.4793,
+      "step": 16120
+    },
+    {
+      "epoch": 3.1727766713383,
+      "grad_norm": 1.1929391622543335,
+      "learning_rate": 1.5270720596777646e-05,
+      "loss": 0.4433,
+      "step": 16130
+    },
+    {
+      "epoch": 3.174743674854319,
+      "grad_norm": 1.8114970922470093,
+      "learning_rate": 1.5267663334250512e-05,
+      "loss": 0.3533,
+      "step": 16140
+    },
+    {
+      "epoch": 3.1767106783703376,
+      "grad_norm": 2.2348251342773438,
+      "learning_rate": 1.526460607172338e-05,
+      "loss": 0.3287,
+      "step": 16150
+    },
+    {
+      "epoch": 3.1786776818863562,
+      "grad_norm": 2.0907199382781982,
+      "learning_rate": 1.5261548809196247e-05,
+      "loss": 0.5437,
+      "step": 16160
+    },
+    {
+      "epoch": 3.1806446854023753,
+      "grad_norm": 1.5697036981582642,
+      "learning_rate": 1.5258491546669115e-05,
+      "loss": 0.3234,
+      "step": 16170
+    },
+    {
+      "epoch": 3.182611688918394,
+      "grad_norm": 2.318769931793213,
+      "learning_rate": 1.525543428414198e-05,
+      "loss": 0.3702,
+      "step": 16180
+    },
+    {
+      "epoch": 3.1845786924344126,
+      "grad_norm": 2.415572166442871,
+      "learning_rate": 1.5252377021614848e-05,
+      "loss": 0.4578,
+      "step": 16190
+    },
+    {
+      "epoch": 3.1865456959504317,
+      "grad_norm": 0.8824933767318726,
+      "learning_rate": 1.5249319759087715e-05,
+      "loss": 0.3238,
+      "step": 16200
+    },
+    {
+      "epoch": 3.1885126994664503,
+      "grad_norm": 2.243199348449707,
+      "learning_rate": 1.524626249656058e-05,
+      "loss": 0.3848,
+      "step": 16210
+    },
+    {
+      "epoch": 3.190479702982469,
+      "grad_norm": 1.778119683265686,
+      "learning_rate": 1.5243205234033447e-05,
+      "loss": 0.4075,
+      "step": 16220
+    },
+    {
+      "epoch": 3.192446706498488,
+      "grad_norm": 1.3491222858428955,
+      "learning_rate": 1.5240147971506314e-05,
+      "loss": 0.3983,
+      "step": 16230
+    },
+    {
+      "epoch": 3.1944137100145067,
+      "grad_norm": 1.729386329650879,
+      "learning_rate": 1.523709070897918e-05,
+      "loss": 0.3341,
+      "step": 16240
+    },
+    {
+      "epoch": 3.1963807135305253,
+      "grad_norm": 1.2898648977279663,
+      "learning_rate": 1.5234033446452048e-05,
+      "loss": 0.3996,
+      "step": 16250
+    },
+    {
+      "epoch": 3.1983477170465444,
+      "grad_norm": 1.3081196546554565,
+      "learning_rate": 1.5230976183924915e-05,
+      "loss": 0.5638,
+      "step": 16260
+    },
+    {
+      "epoch": 3.200314720562563,
+      "grad_norm": 1.233986258506775,
+      "learning_rate": 1.5227918921397783e-05,
+      "loss": 0.2604,
+      "step": 16270
+    },
+    {
+      "epoch": 3.2022817240785817,
+      "grad_norm": 1.209220290184021,
+      "learning_rate": 1.5224861658870648e-05,
+      "loss": 0.4057,
+      "step": 16280
+    },
+    {
+      "epoch": 3.2042487275946008,
+      "grad_norm": 1.7715145349502563,
+      "learning_rate": 1.5221804396343516e-05,
+      "loss": 0.5158,
+      "step": 16290
+    },
+    {
+      "epoch": 3.2062157311106194,
+      "grad_norm": 2.0262582302093506,
+      "learning_rate": 1.5218747133816383e-05,
+      "loss": 0.2559,
+      "step": 16300
+    },
+    {
+      "epoch": 3.208182734626638,
+      "grad_norm": 0.9263052940368652,
+      "learning_rate": 1.5215689871289249e-05,
+      "loss": 0.4124,
+      "step": 16310
+    },
+    {
+      "epoch": 3.2101497381426567,
+      "grad_norm": 0.5316019058227539,
+      "learning_rate": 1.5212632608762115e-05,
+      "loss": 0.3539,
+      "step": 16320
+    },
+    {
+      "epoch": 3.212116741658676,
+      "grad_norm": 1.329001784324646,
+      "learning_rate": 1.5209575346234982e-05,
+      "loss": 0.3256,
+      "step": 16330
+    },
+    {
+      "epoch": 3.2140837451746944,
+      "grad_norm": 2.238600254058838,
+      "learning_rate": 1.5206518083707848e-05,
+      "loss": 0.483,
+      "step": 16340
+    },
+    {
+      "epoch": 3.216050748690713,
+      "grad_norm": 1.8471485376358032,
+      "learning_rate": 1.5203460821180716e-05,
+      "loss": 0.4705,
+      "step": 16350
+    },
+    {
+      "epoch": 3.218017752206732,
+      "grad_norm": 1.351696252822876,
+      "learning_rate": 1.5200403558653583e-05,
+      "loss": 0.4801,
+      "step": 16360
+    },
+    {
+      "epoch": 3.219984755722751,
+      "grad_norm": 1.284036636352539,
+      "learning_rate": 1.5197346296126449e-05,
+      "loss": 0.397,
+      "step": 16370
+    },
+    {
+      "epoch": 3.2219517592387694,
+      "grad_norm": 1.4418704509735107,
+      "learning_rate": 1.5194289033599316e-05,
+      "loss": 0.4128,
+      "step": 16380
+    },
+    {
+      "epoch": 3.2239187627547885,
+      "grad_norm": 0.7759731411933899,
+      "learning_rate": 1.5191231771072184e-05,
+      "loss": 0.3414,
+      "step": 16390
+    },
+    {
+      "epoch": 3.225885766270807,
+      "grad_norm": 1.1827300786972046,
+      "learning_rate": 1.5188174508545051e-05,
+      "loss": 0.4214,
+      "step": 16400
+    },
+    {
+      "epoch": 3.227852769786826,
+      "grad_norm": 1.2514878511428833,
+      "learning_rate": 1.5185117246017917e-05,
+      "loss": 0.3382,
+      "step": 16410
+    },
+    {
+      "epoch": 3.229819773302845,
+      "grad_norm": 0.9900833964347839,
+      "learning_rate": 1.5182059983490785e-05,
+      "loss": 0.3742,
+      "step": 16420
+    },
+    {
+      "epoch": 3.2317867768188635,
+      "grad_norm": 0.8759545087814331,
+      "learning_rate": 1.5179002720963652e-05,
+      "loss": 0.4368,
+      "step": 16430
+    },
+    {
+      "epoch": 3.233753780334882,
+      "grad_norm": 2.74102520942688,
+      "learning_rate": 1.5175945458436516e-05,
+      "loss": 0.4181,
+      "step": 16440
+    },
+    {
+      "epoch": 3.2357207838509012,
+      "grad_norm": 1.1985174417495728,
+      "learning_rate": 1.5172888195909384e-05,
+      "loss": 0.3676,
+      "step": 16450
+    },
+    {
+      "epoch": 3.23768778736692,
+      "grad_norm": 1.1097427606582642,
+      "learning_rate": 1.5169830933382251e-05,
+      "loss": 0.3544,
+      "step": 16460
+    },
+    {
+      "epoch": 3.2396547908829385,
+      "grad_norm": 1.6733139753341675,
+      "learning_rate": 1.5166773670855117e-05,
+      "loss": 0.5471,
+      "step": 16470
+    },
+    {
+      "epoch": 3.2416217943989576,
+      "grad_norm": 0.8993484973907471,
+      "learning_rate": 1.5163716408327984e-05,
+      "loss": 0.4655,
+      "step": 16480
+    },
+    {
+      "epoch": 3.2435887979149762,
+      "grad_norm": 1.7301872968673706,
+      "learning_rate": 1.5160659145800852e-05,
+      "loss": 0.3459,
+      "step": 16490
+    },
+    {
+      "epoch": 3.245555801430995,
+      "grad_norm": 1.0023419857025146,
+      "learning_rate": 1.5157601883273718e-05,
+      "loss": 0.3609,
+      "step": 16500
+    },
+    {
+      "epoch": 3.245555801430995,
+      "eval_loss": 0.18538357317447662,
+      "eval_runtime": 8.898,
+      "eval_samples_per_second": 5.619,
+      "eval_steps_per_second": 2.81,
+      "step": 16500
+    },
+    {
+      "epoch": 3.247522804947014,
+      "grad_norm": 1.1783928871154785,
+      "learning_rate": 1.5154544620746585e-05,
+      "loss": 0.4928,
+      "step": 16510
+    },
+    {
+      "epoch": 3.2494898084630326,
+      "grad_norm": 1.1952968835830688,
+      "learning_rate": 1.5151487358219452e-05,
+      "loss": 0.5126,
+      "step": 16520
+    },
+    {
+      "epoch": 3.2514568119790512,
+      "grad_norm": 1.4290196895599365,
+      "learning_rate": 1.514843009569232e-05,
+      "loss": 0.4627,
+      "step": 16530
+    },
+    {
+      "epoch": 3.2534238154950703,
+      "grad_norm": 4.141489028930664,
+      "learning_rate": 1.5145372833165186e-05,
+      "loss": 0.4236,
+      "step": 16540
+    },
+    {
+      "epoch": 3.255390819011089,
+      "grad_norm": 3.053148031234741,
+      "learning_rate": 1.5142315570638051e-05,
+      "loss": 0.4596,
+      "step": 16550
+    },
+    {
+      "epoch": 3.2573578225271076,
+      "grad_norm": 1.426658034324646,
+      "learning_rate": 1.5139258308110917e-05,
+      "loss": 0.3954,
+      "step": 16560
+    },
+    {
+      "epoch": 3.2593248260431267,
+      "grad_norm": 4.583922386169434,
+      "learning_rate": 1.5136201045583785e-05,
+      "loss": 0.4003,
+      "step": 16570
+    },
+    {
+      "epoch": 3.2612918295591453,
+      "grad_norm": 0.754277765750885,
+      "learning_rate": 1.5133143783056652e-05,
+      "loss": 0.3205,
+      "step": 16580
+    },
+    {
+      "epoch": 3.263258833075164,
+      "grad_norm": 1.0575964450836182,
+      "learning_rate": 1.513008652052952e-05,
+      "loss": 0.3348,
+      "step": 16590
+    },
+    {
+      "epoch": 3.265225836591183,
+      "grad_norm": 1.7032275199890137,
+      "learning_rate": 1.5127029258002385e-05,
+      "loss": 0.4254,
+      "step": 16600
+    },
+    {
+      "epoch": 3.2671928401072017,
+      "grad_norm": 0.8603004813194275,
+      "learning_rate": 1.5123971995475253e-05,
+      "loss": 0.349,
+      "step": 16610
+    },
+    {
+      "epoch": 3.2691598436232203,
+      "grad_norm": 2.109483480453491,
+      "learning_rate": 1.512091473294812e-05,
+      "loss": 0.5604,
+      "step": 16620
+    },
+    {
+      "epoch": 3.2711268471392394,
+      "grad_norm": 2.1563918590545654,
+      "learning_rate": 1.5117857470420986e-05,
+      "loss": 0.516,
+      "step": 16630
+    },
+    {
+      "epoch": 3.273093850655258,
+      "grad_norm": 3.7578446865081787,
+      "learning_rate": 1.5114800207893854e-05,
+      "loss": 0.3768,
+      "step": 16640
+    },
+    {
+      "epoch": 3.2750608541712767,
+      "grad_norm": 1.2201038599014282,
+      "learning_rate": 1.5111742945366721e-05,
+      "loss": 0.3314,
+      "step": 16650
+    },
+    {
+      "epoch": 3.2770278576872958,
+      "grad_norm": 1.9109632968902588,
+      "learning_rate": 1.5108685682839585e-05,
+      "loss": 0.3612,
+      "step": 16660
+    },
+    {
+      "epoch": 3.2789948612033144,
+      "grad_norm": 1.967887043952942,
+      "learning_rate": 1.5105628420312453e-05,
+      "loss": 0.4012,
+      "step": 16670
+    },
+    {
+      "epoch": 3.280961864719333,
+      "grad_norm": 1.129434585571289,
+      "learning_rate": 1.510257115778532e-05,
+      "loss": 0.3515,
+      "step": 16680
+    },
+    {
+      "epoch": 3.282928868235352,
+      "grad_norm": 3.1458957195281982,
+      "learning_rate": 1.5099513895258186e-05,
+      "loss": 0.3917,
+      "step": 16690
+    },
+    {
+      "epoch": 3.2848958717513708,
+      "grad_norm": 1.294286847114563,
+      "learning_rate": 1.5096456632731053e-05,
+      "loss": 0.3951,
+      "step": 16700
+    },
+    {
+      "epoch": 3.2868628752673894,
+      "grad_norm": 1.4721623659133911,
+      "learning_rate": 1.5093399370203921e-05,
+      "loss": 0.3145,
+      "step": 16710
+    },
+    {
+      "epoch": 3.2888298787834085,
+      "grad_norm": 1.3779369592666626,
+      "learning_rate": 1.5090342107676788e-05,
+      "loss": 0.492,
+      "step": 16720
+    },
+    {
+      "epoch": 3.290796882299427,
+      "grad_norm": 2.5460314750671387,
+      "learning_rate": 1.5087284845149654e-05,
+      "loss": 0.4616,
+      "step": 16730
+    },
+    {
+      "epoch": 3.2927638858154458,
+      "grad_norm": 1.3833627700805664,
+      "learning_rate": 1.5084227582622522e-05,
+      "loss": 0.4766,
+      "step": 16740
+    },
+    {
+      "epoch": 3.294730889331465,
+      "grad_norm": 0.9380312561988831,
+      "learning_rate": 1.5081170320095389e-05,
+      "loss": 0.449,
+      "step": 16750
+    },
+    {
+      "epoch": 3.2966978928474835,
+      "grad_norm": 1.3430695533752441,
+      "learning_rate": 1.5078113057568255e-05,
+      "loss": 0.5169,
+      "step": 16760
+    },
+    {
+      "epoch": 3.298664896363502,
+      "grad_norm": 1.6764286756515503,
+      "learning_rate": 1.5075055795041122e-05,
+      "loss": 0.5024,
+      "step": 16770
+    },
+    {
+      "epoch": 3.300631899879521,
+      "grad_norm": 1.2297427654266357,
+      "learning_rate": 1.5071998532513988e-05,
+      "loss": 0.4217,
+      "step": 16780
+    },
+    {
+      "epoch": 3.30259890339554,
+      "grad_norm": 1.2263567447662354,
+      "learning_rate": 1.5068941269986854e-05,
+      "loss": 0.4577,
+      "step": 16790
+    },
+    {
+      "epoch": 3.3045659069115585,
+      "grad_norm": 1.4633382558822632,
+      "learning_rate": 1.5065884007459721e-05,
+      "loss": 0.3677,
+      "step": 16800
+    },
+    {
+      "epoch": 3.3065329104275776,
+      "grad_norm": 1.3397003412246704,
+      "learning_rate": 1.5062826744932589e-05,
+      "loss": 0.3227,
+      "step": 16810
+    },
+    {
+      "epoch": 3.308499913943596,
+      "grad_norm": 1.438306450843811,
+      "learning_rate": 1.5059769482405455e-05,
+      "loss": 0.3175,
+      "step": 16820
+    },
+    {
+      "epoch": 3.310466917459615,
+      "grad_norm": 1.435363531112671,
+      "learning_rate": 1.5056712219878322e-05,
+      "loss": 0.3392,
+      "step": 16830
+    },
+    {
+      "epoch": 3.312433920975634,
+      "grad_norm": 2.12419056892395,
+      "learning_rate": 1.505365495735119e-05,
+      "loss": 0.5618,
+      "step": 16840
+    },
+    {
+      "epoch": 3.3144009244916526,
+      "grad_norm": 1.4096325635910034,
+      "learning_rate": 1.5050597694824057e-05,
+      "loss": 0.3874,
+      "step": 16850
+    },
+    {
+      "epoch": 3.316367928007671,
+      "grad_norm": 2.04353404045105,
+      "learning_rate": 1.5047540432296923e-05,
+      "loss": 0.4306,
+      "step": 16860
+    },
+    {
+      "epoch": 3.3183349315236903,
+      "grad_norm": 1.5693720579147339,
+      "learning_rate": 1.504448316976979e-05,
+      "loss": 0.4537,
+      "step": 16870
+    },
+    {
+      "epoch": 3.320301935039709,
+      "grad_norm": 1.255326271057129,
+      "learning_rate": 1.5041425907242658e-05,
+      "loss": 0.4625,
+      "step": 16880
+    },
+    {
+      "epoch": 3.3222689385557276,
+      "grad_norm": 2.1881473064422607,
+      "learning_rate": 1.5038368644715522e-05,
+      "loss": 0.3993,
+      "step": 16890
+    },
+    {
+      "epoch": 3.3242359420717467,
+      "grad_norm": 1.0848050117492676,
+      "learning_rate": 1.503531138218839e-05,
+      "loss": 0.3404,
+      "step": 16900
+    },
+    {
+      "epoch": 3.3262029455877653,
+      "grad_norm": 1.1328253746032715,
+      "learning_rate": 1.5032254119661255e-05,
+      "loss": 0.367,
+      "step": 16910
+    },
+    {
+      "epoch": 3.328169949103784,
+      "grad_norm": 2.038686513900757,
+      "learning_rate": 1.5029196857134123e-05,
+      "loss": 0.4136,
+      "step": 16920
+    },
+    {
+      "epoch": 3.330136952619803,
+      "grad_norm": 1.5437724590301514,
+      "learning_rate": 1.502613959460699e-05,
+      "loss": 0.3227,
+      "step": 16930
+    },
+    {
+      "epoch": 3.3321039561358217,
+      "grad_norm": 1.6073527336120605,
+      "learning_rate": 1.5023082332079857e-05,
+      "loss": 0.4443,
+      "step": 16940
+    },
+    {
+      "epoch": 3.3340709596518403,
+      "grad_norm": 0.997283935546875,
+      "learning_rate": 1.5020025069552723e-05,
+      "loss": 0.5583,
+      "step": 16950
+    },
+    {
+      "epoch": 3.3360379631678594,
+      "grad_norm": 1.7358487844467163,
+      "learning_rate": 1.501696780702559e-05,
+      "loss": 0.4091,
+      "step": 16960
+    },
+    {
+      "epoch": 3.338004966683878,
+      "grad_norm": 1.2565847635269165,
+      "learning_rate": 1.5013910544498458e-05,
+      "loss": 0.4082,
+      "step": 16970
+    },
+    {
+      "epoch": 3.3399719701998967,
+      "grad_norm": 2.5975379943847656,
+      "learning_rate": 1.5010853281971326e-05,
+      "loss": 0.4562,
+      "step": 16980
+    },
+    {
+      "epoch": 3.3419389737159158,
+      "grad_norm": 0.9561290740966797,
+      "learning_rate": 1.5007796019444191e-05,
+      "loss": 0.4909,
+      "step": 16990
+    },
+    {
+      "epoch": 3.3439059772319344,
+      "grad_norm": 1.0989880561828613,
+      "learning_rate": 1.5004738756917057e-05,
+      "loss": 0.4527,
+      "step": 17000
+    },
+    {
+      "epoch": 3.3439059772319344,
+      "eval_loss": 0.17282529175281525,
+      "eval_runtime": 8.8713,
+      "eval_samples_per_second": 5.636,
+      "eval_steps_per_second": 2.818,
+      "step": 17000
+    },
+    {
+      "epoch": 3.345872980747953,
+      "grad_norm": 1.0213874578475952,
+      "learning_rate": 1.5001681494389923e-05,
+      "loss": 0.3839,
+      "step": 17010
+    },
+    {
+      "epoch": 3.347839984263972,
+      "grad_norm": 0.748192548751831,
+      "learning_rate": 1.499862423186279e-05,
+      "loss": 0.3286,
+      "step": 17020
+    },
+    {
+      "epoch": 3.3498069877799908,
+      "grad_norm": 2.626722574234009,
+      "learning_rate": 1.4995566969335658e-05,
+      "loss": 0.3294,
+      "step": 17030
+    },
+    {
+      "epoch": 3.3517739912960094,
+      "grad_norm": 2.035220146179199,
+      "learning_rate": 1.4992509706808524e-05,
+      "loss": 0.3095,
+      "step": 17040
+    },
+    {
+      "epoch": 3.3537409948120285,
+      "grad_norm": 0.9364410638809204,
+      "learning_rate": 1.4989452444281391e-05,
+      "loss": 0.5119,
+      "step": 17050
+    },
+    {
+      "epoch": 3.355707998328047,
+      "grad_norm": 1.4533653259277344,
+      "learning_rate": 1.4986395181754259e-05,
+      "loss": 0.392,
+      "step": 17060
+    },
+    {
+      "epoch": 3.3576750018440658,
+      "grad_norm": 0.9131273627281189,
+      "learning_rate": 1.4983337919227126e-05,
+      "loss": 0.4791,
+      "step": 17070
+    },
+    {
+      "epoch": 3.3596420053600844,
+      "grad_norm": 1.4228838682174683,
+      "learning_rate": 1.4980280656699992e-05,
+      "loss": 0.4465,
+      "step": 17080
+    },
+    {
+      "epoch": 3.3616090088761035,
+      "grad_norm": 1.7861515283584595,
+      "learning_rate": 1.497722339417286e-05,
+      "loss": 0.3783,
+      "step": 17090
+    },
+    {
+      "epoch": 3.363576012392122,
+      "grad_norm": 2.334275484085083,
+      "learning_rate": 1.4974166131645727e-05,
+      "loss": 0.4166,
+      "step": 17100
+    },
+    {
+      "epoch": 3.3655430159081408,
+      "grad_norm": 1.2254345417022705,
+      "learning_rate": 1.4971108869118594e-05,
+      "loss": 0.3472,
+      "step": 17110
+    },
+    {
+      "epoch": 3.36751001942416,
+      "grad_norm": 0.9154657125473022,
+      "learning_rate": 1.4968051606591458e-05,
+      "loss": 0.4692,
+      "step": 17120
+    },
+    {
+      "epoch": 3.3694770229401785,
+      "grad_norm": 1.3035657405853271,
+      "learning_rate": 1.4964994344064326e-05,
+      "loss": 0.3889,
+      "step": 17130
+    },
+    {
+      "epoch": 3.371444026456197,
+      "grad_norm": 1.2203218936920166,
+      "learning_rate": 1.4961937081537192e-05,
+      "loss": 0.3086,
+      "step": 17140
+    },
+    {
+      "epoch": 3.373411029972216,
+      "grad_norm": 0.7467970848083496,
+      "learning_rate": 1.4958879819010059e-05,
+      "loss": 0.4192,
+      "step": 17150
+    },
+    {
+      "epoch": 3.375378033488235,
+      "grad_norm": 1.9781105518341064,
+      "learning_rate": 1.4955822556482927e-05,
+      "loss": 0.3967,
+      "step": 17160
+    },
+    {
+      "epoch": 3.3773450370042535,
+      "grad_norm": 1.0278807878494263,
+      "learning_rate": 1.4952765293955792e-05,
+      "loss": 0.4609,
+      "step": 17170
+    },
+    {
+      "epoch": 3.3793120405202726,
+      "grad_norm": 1.5243288278579712,
+      "learning_rate": 1.494970803142866e-05,
+      "loss": 0.3332,
+      "step": 17180
+    },
+    {
+      "epoch": 3.381279044036291,
+      "grad_norm": 1.0082272291183472,
+      "learning_rate": 1.4946650768901527e-05,
+      "loss": 0.2864,
+      "step": 17190
+    },
+    {
+      "epoch": 3.38324604755231,
+      "grad_norm": 1.5018178224563599,
+      "learning_rate": 1.4943593506374395e-05,
+      "loss": 0.4893,
+      "step": 17200
+    },
+    {
+      "epoch": 3.385213051068329,
+      "grad_norm": 1.1987298727035522,
+      "learning_rate": 1.494053624384726e-05,
+      "loss": 0.2966,
+      "step": 17210
+    },
+    {
+      "epoch": 3.3871800545843476,
+      "grad_norm": 3.0867018699645996,
+      "learning_rate": 1.4937478981320128e-05,
+      "loss": 0.4841,
+      "step": 17220
+    },
+    {
+      "epoch": 3.389147058100366,
+      "grad_norm": 1.738324761390686,
+      "learning_rate": 1.4934421718792992e-05,
+      "loss": 0.3599,
+      "step": 17230
+    },
+    {
+      "epoch": 3.3911140616163853,
+      "grad_norm": 1.9667898416519165,
+      "learning_rate": 1.493136445626586e-05,
+      "loss": 0.3224,
+      "step": 17240
+    },
+    {
+      "epoch": 3.393081065132404,
+      "grad_norm": 1.2552376985549927,
+      "learning_rate": 1.4928307193738727e-05,
+      "loss": 0.4395,
+      "step": 17250
+    },
+    {
+      "epoch": 3.3950480686484226,
+      "grad_norm": 2.6338891983032227,
+      "learning_rate": 1.4925249931211595e-05,
+      "loss": 0.4038,
+      "step": 17260
+    },
+    {
+      "epoch": 3.3970150721644417,
+      "grad_norm": 1.44743013381958,
+      "learning_rate": 1.492219266868446e-05,
+      "loss": 0.2454,
+      "step": 17270
+    },
+    {
+      "epoch": 3.3989820756804603,
+      "grad_norm": 1.3355497121810913,
+      "learning_rate": 1.4919135406157328e-05,
+      "loss": 0.2837,
+      "step": 17280
+    },
+    {
+      "epoch": 3.400949079196479,
+      "grad_norm": 2.0968546867370605,
+      "learning_rate": 1.4916078143630195e-05,
+      "loss": 0.2952,
+      "step": 17290
+    },
+    {
+      "epoch": 3.402916082712498,
+      "grad_norm": 1.6755894422531128,
+      "learning_rate": 1.4913020881103061e-05,
+      "loss": 0.3963,
+      "step": 17300
+    },
+    {
+      "epoch": 3.4048830862285167,
+      "grad_norm": 2.201287031173706,
+      "learning_rate": 1.4909963618575929e-05,
+      "loss": 0.4035,
+      "step": 17310
+    },
+    {
+      "epoch": 3.4068500897445353,
+      "grad_norm": 1.1484105587005615,
+      "learning_rate": 1.4906906356048796e-05,
+      "loss": 0.4353,
+      "step": 17320
+    },
+    {
+      "epoch": 3.4088170932605544,
+      "grad_norm": 1.558431625366211,
+      "learning_rate": 1.4903849093521663e-05,
+      "loss": 0.4465,
+      "step": 17330
+    },
+    {
+      "epoch": 3.410784096776573,
+      "grad_norm": 1.5346964597702026,
+      "learning_rate": 1.4900791830994528e-05,
+      "loss": 0.4069,
+      "step": 17340
+    },
+    {
+      "epoch": 3.4127511002925917,
+      "grad_norm": 1.7767252922058105,
+      "learning_rate": 1.4897734568467395e-05,
+      "loss": 0.3883,
+      "step": 17350
+    },
+    {
+      "epoch": 3.4147181038086107,
+      "grad_norm": 0.967298686504364,
+      "learning_rate": 1.489467730594026e-05,
+      "loss": 0.3695,
+      "step": 17360
+    },
+    {
+      "epoch": 3.4166851073246294,
+      "grad_norm": 0.9392523765563965,
+      "learning_rate": 1.4891620043413128e-05,
+      "loss": 0.4026,
+      "step": 17370
+    },
+    {
+      "epoch": 3.418652110840648,
+      "grad_norm": 1.7647945880889893,
+      "learning_rate": 1.4888562780885996e-05,
+      "loss": 0.3074,
+      "step": 17380
+    },
+    {
+      "epoch": 3.420619114356667,
+      "grad_norm": 1.1813808679580688,
+      "learning_rate": 1.4885505518358863e-05,
+      "loss": 0.381,
+      "step": 17390
+    },
+    {
+      "epoch": 3.4225861178726857,
+      "grad_norm": 1.2232673168182373,
+      "learning_rate": 1.4882448255831729e-05,
+      "loss": 0.48,
+      "step": 17400
+    },
+    {
+      "epoch": 3.4245531213887044,
+      "grad_norm": 1.052878975868225,
+      "learning_rate": 1.4879390993304596e-05,
+      "loss": 0.4123,
+      "step": 17410
+    },
+    {
+      "epoch": 3.426520124904723,
+      "grad_norm": 1.2868292331695557,
+      "learning_rate": 1.4876333730777464e-05,
+      "loss": 0.3772,
+      "step": 17420
+    },
+    {
+      "epoch": 3.428487128420742,
+      "grad_norm": 1.3638333082199097,
+      "learning_rate": 1.487327646825033e-05,
+      "loss": 0.3819,
+      "step": 17430
+    },
+    {
+      "epoch": 3.4304541319367607,
+      "grad_norm": 2.3324482440948486,
+      "learning_rate": 1.4870219205723197e-05,
+      "loss": 0.4775,
+      "step": 17440
+    },
+    {
+      "epoch": 3.4324211354527794,
+      "grad_norm": 1.1716792583465576,
+      "learning_rate": 1.4867161943196065e-05,
+      "loss": 0.3989,
+      "step": 17450
+    },
+    {
+      "epoch": 3.4343881389687985,
+      "grad_norm": 1.5580689907073975,
+      "learning_rate": 1.4864104680668929e-05,
+      "loss": 0.4771,
+      "step": 17460
+    },
+    {
+      "epoch": 3.436355142484817,
+      "grad_norm": 1.3105270862579346,
+      "learning_rate": 1.4861047418141796e-05,
+      "loss": 0.378,
+      "step": 17470
+    },
+    {
+      "epoch": 3.4383221460008357,
+      "grad_norm": 1.6902211904525757,
+      "learning_rate": 1.4857990155614664e-05,
+      "loss": 0.4194,
+      "step": 17480
+    },
+    {
+      "epoch": 3.440289149516855,
+      "grad_norm": 0.9721977114677429,
+      "learning_rate": 1.485493289308753e-05,
+      "loss": 0.5298,
+      "step": 17490
+    },
+    {
+      "epoch": 3.4422561530328735,
+      "grad_norm": 1.1658124923706055,
+      "learning_rate": 1.4851875630560397e-05,
+      "loss": 0.3992,
+      "step": 17500
+    },
+    {
+      "epoch": 3.4422561530328735,
+      "eval_loss": 0.17830701172351837,
+      "eval_runtime": 8.8662,
+      "eval_samples_per_second": 5.639,
+      "eval_steps_per_second": 2.82,
+      "step": 17500
+    },
+    {
+      "epoch": 3.444223156548892,
+      "grad_norm": 0.9558307528495789,
+      "learning_rate": 1.4848818368033264e-05,
+      "loss": 0.4587,
+      "step": 17510
+    },
+    {
+      "epoch": 3.446190160064911,
+      "grad_norm": 1.4307557344436646,
+      "learning_rate": 1.4845761105506132e-05,
+      "loss": 0.3265,
+      "step": 17520
+    },
+    {
+      "epoch": 3.44815716358093,
+      "grad_norm": 1.1415700912475586,
+      "learning_rate": 1.4842703842978998e-05,
+      "loss": 0.3661,
+      "step": 17530
+    },
+    {
+      "epoch": 3.4501241670969485,
+      "grad_norm": 3.2358052730560303,
+      "learning_rate": 1.4839646580451865e-05,
+      "loss": 0.4427,
+      "step": 17540
+    },
+    {
+      "epoch": 3.4520911706129676,
+      "grad_norm": 1.6495846509933472,
+      "learning_rate": 1.4836589317924733e-05,
+      "loss": 0.3998,
+      "step": 17550
+    },
+    {
+      "epoch": 3.454058174128986,
+      "grad_norm": 1.502959966659546,
+      "learning_rate": 1.4833532055397598e-05,
+      "loss": 0.3918,
+      "step": 17560
+    },
+    {
+      "epoch": 3.456025177645005,
+      "grad_norm": 2.2317111492156982,
+      "learning_rate": 1.4830474792870464e-05,
+      "loss": 0.4154,
+      "step": 17570
+    },
+    {
+      "epoch": 3.457992181161024,
+      "grad_norm": 1.9064170122146606,
+      "learning_rate": 1.4827417530343332e-05,
+      "loss": 0.4326,
+      "step": 17580
+    },
+    {
+      "epoch": 3.4599591846770426,
+      "grad_norm": 1.7507431507110596,
+      "learning_rate": 1.4824360267816197e-05,
+      "loss": 0.4027,
+      "step": 17590
+    },
+    {
+      "epoch": 3.461926188193061,
+      "grad_norm": 1.6282731294631958,
+      "learning_rate": 1.4821303005289065e-05,
+      "loss": 0.386,
+      "step": 17600
+    },
+    {
+      "epoch": 3.4638931917090803,
+      "grad_norm": 1.266846776008606,
+      "learning_rate": 1.4818245742761932e-05,
+      "loss": 0.256,
+      "step": 17610
+    },
+    {
+      "epoch": 3.465860195225099,
+      "grad_norm": 2.4402332305908203,
+      "learning_rate": 1.4815188480234798e-05,
+      "loss": 0.4506,
+      "step": 17620
+    },
+    {
+      "epoch": 3.4678271987411176,
+      "grad_norm": 1.072805404663086,
+      "learning_rate": 1.4812131217707666e-05,
+      "loss": 0.298,
+      "step": 17630
+    },
+    {
+      "epoch": 3.4697942022571366,
+      "grad_norm": 2.0623440742492676,
+      "learning_rate": 1.4809073955180533e-05,
+      "loss": 0.4893,
+      "step": 17640
+    },
+    {
+      "epoch": 3.4717612057731553,
+      "grad_norm": 2.03657865524292,
+      "learning_rate": 1.48060166926534e-05,
+      "loss": 0.4633,
+      "step": 17650
+    },
+    {
+      "epoch": 3.473728209289174,
+      "grad_norm": 1.6767804622650146,
+      "learning_rate": 1.4802959430126266e-05,
+      "loss": 0.3956,
+      "step": 17660
+    },
+    {
+      "epoch": 3.475695212805193,
+      "grad_norm": 1.3445123434066772,
+      "learning_rate": 1.4799902167599134e-05,
+      "loss": 0.4233,
+      "step": 17670
+    },
+    {
+      "epoch": 3.4776622163212116,
+      "grad_norm": 0.9934619665145874,
+      "learning_rate": 1.4796844905071998e-05,
+      "loss": 0.4358,
+      "step": 17680
+    },
+    {
+      "epoch": 3.4796292198372303,
+      "grad_norm": 1.1431872844696045,
+      "learning_rate": 1.4793787642544865e-05,
+      "loss": 0.4817,
+      "step": 17690
+    },
+    {
+      "epoch": 3.4815962233532494,
+      "grad_norm": 2.3636295795440674,
+      "learning_rate": 1.4790730380017733e-05,
+      "loss": 0.3428,
+      "step": 17700
+    },
+    {
+      "epoch": 3.483563226869268,
+      "grad_norm": 1.1688228845596313,
+      "learning_rate": 1.47876731174906e-05,
+      "loss": 0.4124,
+      "step": 17710
+    },
+    {
+      "epoch": 3.4855302303852866,
+      "grad_norm": 1.8074513673782349,
+      "learning_rate": 1.4784615854963466e-05,
+      "loss": 0.4571,
+      "step": 17720
+    },
+    {
+      "epoch": 3.4874972339013057,
+      "grad_norm": 1.1620044708251953,
+      "learning_rate": 1.4781558592436334e-05,
+      "loss": 0.3975,
+      "step": 17730
+    },
+    {
+      "epoch": 3.4894642374173244,
+      "grad_norm": 2.0841927528381348,
+      "learning_rate": 1.4778501329909201e-05,
+      "loss": 0.3895,
+      "step": 17740
+    },
+    {
+      "epoch": 3.491431240933343,
+      "grad_norm": 2.7444779872894287,
+      "learning_rate": 1.4775444067382067e-05,
+      "loss": 0.4395,
+      "step": 17750
+    },
+    {
+      "epoch": 3.493398244449362,
+      "grad_norm": 2.4333293437957764,
+      "learning_rate": 1.4772386804854934e-05,
+      "loss": 0.3026,
+      "step": 17760
+    },
+    {
+      "epoch": 3.4953652479653807,
+      "grad_norm": 1.7708234786987305,
+      "learning_rate": 1.4769329542327802e-05,
+      "loss": 0.4695,
+      "step": 17770
+    },
+    {
+      "epoch": 3.4973322514813994,
+      "grad_norm": 1.2430254220962524,
+      "learning_rate": 1.476627227980067e-05,
+      "loss": 0.3391,
+      "step": 17780
+    },
+    {
+      "epoch": 3.4992992549974185,
+      "grad_norm": 1.132948875427246,
+      "learning_rate": 1.4763215017273535e-05,
+      "loss": 0.4641,
+      "step": 17790
+    },
+    {
+      "epoch": 3.501266258513437,
+      "grad_norm": 0.7967662811279297,
+      "learning_rate": 1.47601577547464e-05,
+      "loss": 0.4284,
+      "step": 17800
+    },
+    {
+      "epoch": 3.5032332620294557,
+      "grad_norm": 1.9069840908050537,
+      "learning_rate": 1.4757100492219267e-05,
+      "loss": 0.293,
+      "step": 17810
+    },
+    {
+      "epoch": 3.505200265545475,
+      "grad_norm": 1.1667743921279907,
+      "learning_rate": 1.4754043229692134e-05,
+      "loss": 0.4267,
+      "step": 17820
+    },
+    {
+      "epoch": 3.5071672690614935,
+      "grad_norm": 1.3001110553741455,
+      "learning_rate": 1.4750985967165002e-05,
+      "loss": 0.4152,
+      "step": 17830
+    },
+    {
+      "epoch": 3.509134272577512,
+      "grad_norm": 1.6212232112884521,
+      "learning_rate": 1.4747928704637869e-05,
+      "loss": 0.2784,
+      "step": 17840
+    },
+    {
+      "epoch": 3.511101276093531,
+      "grad_norm": 1.6824311017990112,
+      "learning_rate": 1.4744871442110735e-05,
+      "loss": 0.5409,
+      "step": 17850
+    },
+    {
+      "epoch": 3.51306827960955,
+      "grad_norm": 2.316366672515869,
+      "learning_rate": 1.4741814179583602e-05,
+      "loss": 0.3553,
+      "step": 17860
+    },
+    {
+      "epoch": 3.5150352831255685,
+      "grad_norm": 1.079383373260498,
+      "learning_rate": 1.473875691705647e-05,
+      "loss": 0.4142,
+      "step": 17870
+    },
+    {
+      "epoch": 3.5170022866415875,
+      "grad_norm": 1.539841651916504,
+      "learning_rate": 1.4735699654529335e-05,
+      "loss": 0.3635,
+      "step": 17880
+    },
+    {
+      "epoch": 3.518969290157606,
+      "grad_norm": 1.3688745498657227,
+      "learning_rate": 1.4732642392002203e-05,
+      "loss": 0.3125,
+      "step": 17890
+    },
+    {
+      "epoch": 3.520936293673625,
+      "grad_norm": 1.4701616764068604,
+      "learning_rate": 1.472958512947507e-05,
+      "loss": 0.3808,
+      "step": 17900
+    },
+    {
+      "epoch": 3.522903297189644,
+      "grad_norm": 1.2537261247634888,
+      "learning_rate": 1.4726527866947935e-05,
+      "loss": 0.3833,
+      "step": 17910
+    },
+    {
+      "epoch": 3.5248703007056625,
+      "grad_norm": 1.4147868156433105,
+      "learning_rate": 1.4723470604420802e-05,
+      "loss": 0.3304,
+      "step": 17920
+    },
+    {
+      "epoch": 3.526837304221681,
+      "grad_norm": 2.0437426567077637,
+      "learning_rate": 1.472041334189367e-05,
+      "loss": 0.4108,
+      "step": 17930
+    },
+    {
+      "epoch": 3.5288043077377003,
+      "grad_norm": 0.944835364818573,
+      "learning_rate": 1.4717356079366535e-05,
+      "loss": 0.4279,
+      "step": 17940
+    },
+    {
+      "epoch": 3.530771311253719,
+      "grad_norm": 0.7478554844856262,
+      "learning_rate": 1.4714298816839403e-05,
+      "loss": 0.4301,
+      "step": 17950
+    },
+    {
+      "epoch": 3.5327383147697375,
+      "grad_norm": 1.308043360710144,
+      "learning_rate": 1.471124155431227e-05,
+      "loss": 0.4849,
+      "step": 17960
+    },
+    {
+      "epoch": 3.5347053182857566,
+      "grad_norm": 0.990397572517395,
+      "learning_rate": 1.4708184291785138e-05,
+      "loss": 0.473,
+      "step": 17970
+    },
+    {
+      "epoch": 3.5366723218017753,
+      "grad_norm": 0.9142557978630066,
+      "learning_rate": 1.4705127029258003e-05,
+      "loss": 0.6008,
+      "step": 17980
+    },
+    {
+      "epoch": 3.538639325317794,
+      "grad_norm": 2.228318452835083,
+      "learning_rate": 1.4702069766730871e-05,
+      "loss": 0.4516,
+      "step": 17990
+    },
+    {
+      "epoch": 3.540606328833813,
+      "grad_norm": 1.151518702507019,
+      "learning_rate": 1.4699012504203738e-05,
+      "loss": 0.5171,
+      "step": 18000
+    },
+    {
+      "epoch": 3.540606328833813,
+      "eval_loss": 0.18341459333896637,
+      "eval_runtime": 8.9046,
+      "eval_samples_per_second": 5.615,
+      "eval_steps_per_second": 2.808,
+      "step": 18000
+    },
+    {
+      "epoch": 3.5425733323498316,
+      "grad_norm": 1.794045329093933,
+      "learning_rate": 1.4695955241676604e-05,
+      "loss": 0.3875,
+      "step": 18010
+    },
+    {
+      "epoch": 3.5445403358658503,
+      "grad_norm": 1.164614200592041,
+      "learning_rate": 1.469289797914947e-05,
+      "loss": 0.4417,
+      "step": 18020
+    },
+    {
+      "epoch": 3.5465073393818694,
+      "grad_norm": 1.6620136499404907,
+      "learning_rate": 1.4689840716622337e-05,
+      "loss": 0.3842,
+      "step": 18030
+    },
+    {
+      "epoch": 3.548474342897888,
+      "grad_norm": 1.4214859008789062,
+      "learning_rate": 1.4686783454095203e-05,
+      "loss": 0.4269,
+      "step": 18040
+    },
+    {
+      "epoch": 3.5504413464139066,
+      "grad_norm": 3.659273386001587,
+      "learning_rate": 1.468372619156807e-05,
+      "loss": 0.3654,
+      "step": 18050
+    },
+    {
+      "epoch": 3.5524083499299257,
+      "grad_norm": 1.5434625148773193,
+      "learning_rate": 1.4680668929040938e-05,
+      "loss": 0.4587,
+      "step": 18060
+    },
+    {
+      "epoch": 3.5543753534459444,
+      "grad_norm": 1.2292109727859497,
+      "learning_rate": 1.4677611666513804e-05,
+      "loss": 0.3767,
+      "step": 18070
+    },
+    {
+      "epoch": 3.556342356961963,
+      "grad_norm": 1.5507662296295166,
+      "learning_rate": 1.4674554403986671e-05,
+      "loss": 0.3647,
+      "step": 18080
+    },
+    {
+      "epoch": 3.558309360477982,
+      "grad_norm": 1.2681255340576172,
+      "learning_rate": 1.4671497141459539e-05,
+      "loss": 0.3738,
+      "step": 18090
+    },
+    {
+      "epoch": 3.5602763639940007,
+      "grad_norm": 1.5174590349197388,
+      "learning_rate": 1.4668439878932406e-05,
+      "loss": 0.3405,
+      "step": 18100
+    },
+    {
+      "epoch": 3.5622433675100194,
+      "grad_norm": 0.6839548945426941,
+      "learning_rate": 1.4665382616405272e-05,
+      "loss": 0.2135,
+      "step": 18110
+    },
+    {
+      "epoch": 3.5642103710260384,
+      "grad_norm": 2.3022007942199707,
+      "learning_rate": 1.466232535387814e-05,
+      "loss": 0.4563,
+      "step": 18120
+    },
+    {
+      "epoch": 3.566177374542057,
+      "grad_norm": 2.41715407371521,
+      "learning_rate": 1.4659268091351007e-05,
+      "loss": 0.3429,
+      "step": 18130
+    },
+    {
+      "epoch": 3.5681443780580757,
+      "grad_norm": 2.4926366806030273,
+      "learning_rate": 1.4656210828823871e-05,
+      "loss": 0.3883,
+      "step": 18140
+    },
+    {
+      "epoch": 3.570111381574095,
+      "grad_norm": 0.43145233392715454,
+      "learning_rate": 1.4653153566296739e-05,
+      "loss": 0.3409,
+      "step": 18150
+    },
+    {
+      "epoch": 3.5720783850901134,
+      "grad_norm": 1.6929571628570557,
+      "learning_rate": 1.4650096303769606e-05,
+      "loss": 0.3721,
+      "step": 18160
+    },
+    {
+      "epoch": 3.574045388606132,
+      "grad_norm": 1.570227026939392,
+      "learning_rate": 1.4647039041242472e-05,
+      "loss": 0.4791,
+      "step": 18170
+    },
+    {
+      "epoch": 3.576012392122151,
+      "grad_norm": 0.941781222820282,
+      "learning_rate": 1.464398177871534e-05,
+      "loss": 0.4605,
+      "step": 18180
+    },
+    {
+      "epoch": 3.57797939563817,
+      "grad_norm": 1.1423379182815552,
+      "learning_rate": 1.4640924516188207e-05,
+      "loss": 0.3748,
+      "step": 18190
+    },
+    {
+      "epoch": 3.5799463991541884,
+      "grad_norm": 1.7628474235534668,
+      "learning_rate": 1.4637867253661073e-05,
+      "loss": 0.4167,
+      "step": 18200
+    },
+    {
+      "epoch": 3.5819134026702075,
+      "grad_norm": 1.6465951204299927,
+      "learning_rate": 1.463480999113394e-05,
+      "loss": 0.4413,
+      "step": 18210
+    },
+    {
+      "epoch": 3.583880406186226,
+      "grad_norm": 1.6730773448944092,
+      "learning_rate": 1.4631752728606807e-05,
+      "loss": 0.3454,
+      "step": 18220
+    },
+    {
+      "epoch": 3.585847409702245,
+      "grad_norm": 1.4015109539031982,
+      "learning_rate": 1.4628695466079675e-05,
+      "loss": 0.4662,
+      "step": 18230
+    },
+    {
+      "epoch": 3.587814413218264,
+      "grad_norm": 1.1885541677474976,
+      "learning_rate": 1.462563820355254e-05,
+      "loss": 0.3946,
+      "step": 18240
+    },
+    {
+      "epoch": 3.5897814167342825,
+      "grad_norm": 0.859453022480011,
+      "learning_rate": 1.4622580941025407e-05,
+      "loss": 0.4901,
+      "step": 18250
+    },
+    {
+      "epoch": 3.591748420250301,
+      "grad_norm": 2.0021779537200928,
+      "learning_rate": 1.4619523678498272e-05,
+      "loss": 0.3715,
+      "step": 18260
+    },
+    {
+      "epoch": 3.5937154237663202,
+      "grad_norm": 1.107502818107605,
+      "learning_rate": 1.461646641597114e-05,
+      "loss": 0.4452,
+      "step": 18270
+    },
+    {
+      "epoch": 3.595682427282339,
+      "grad_norm": 1.370747447013855,
+      "learning_rate": 1.4613409153444007e-05,
+      "loss": 0.4037,
+      "step": 18280
+    },
+    {
+      "epoch": 3.5976494307983575,
+      "grad_norm": 2.8257832527160645,
+      "learning_rate": 1.4610351890916875e-05,
+      "loss": 0.3248,
+      "step": 18290
+    },
+    {
+      "epoch": 3.5996164343143766,
+      "grad_norm": 2.032862901687622,
+      "learning_rate": 1.460729462838974e-05,
+      "loss": 0.4332,
+      "step": 18300
+    },
+    {
+      "epoch": 3.6015834378303953,
+      "grad_norm": 1.1066280603408813,
+      "learning_rate": 1.4604237365862608e-05,
+      "loss": 0.4214,
+      "step": 18310
+    },
+    {
+      "epoch": 3.603550441346414,
+      "grad_norm": 0.9145089983940125,
+      "learning_rate": 1.4601180103335475e-05,
+      "loss": 0.4127,
+      "step": 18320
+    },
+    {
+      "epoch": 3.6055174448624325,
+      "grad_norm": 1.2561908960342407,
+      "learning_rate": 1.4598122840808341e-05,
+      "loss": 0.3641,
+      "step": 18330
+    },
+    {
+      "epoch": 3.6074844483784516,
+      "grad_norm": 1.9700093269348145,
+      "learning_rate": 1.4595065578281209e-05,
+      "loss": 0.348,
+      "step": 18340
+    },
+    {
+      "epoch": 3.6094514518944703,
+      "grad_norm": 1.7856909036636353,
+      "learning_rate": 1.4592008315754076e-05,
+      "loss": 0.401,
+      "step": 18350
+    },
+    {
+      "epoch": 3.611418455410489,
+      "grad_norm": 0.8354535102844238,
+      "learning_rate": 1.458895105322694e-05,
+      "loss": 0.3844,
+      "step": 18360
+    },
+    {
+      "epoch": 3.613385458926508,
+      "grad_norm": 1.3362191915512085,
+      "learning_rate": 1.4585893790699808e-05,
+      "loss": 0.3867,
+      "step": 18370
+    },
+    {
+      "epoch": 3.6153524624425266,
+      "grad_norm": 1.0379217863082886,
+      "learning_rate": 1.4582836528172675e-05,
+      "loss": 0.424,
+      "step": 18380
+    },
+    {
+      "epoch": 3.6173194659585453,
+      "grad_norm": 2.015969753265381,
+      "learning_rate": 1.4579779265645541e-05,
+      "loss": 0.4422,
+      "step": 18390
+    },
+    {
+      "epoch": 3.6192864694745643,
+      "grad_norm": 1.8602677583694458,
+      "learning_rate": 1.4576722003118408e-05,
+      "loss": 0.2997,
+      "step": 18400
+    },
+    {
+      "epoch": 3.621253472990583,
+      "grad_norm": 1.1193724870681763,
+      "learning_rate": 1.4573664740591276e-05,
+      "loss": 0.3575,
+      "step": 18410
+    },
+    {
+      "epoch": 3.6232204765066016,
+      "grad_norm": 2.074537992477417,
+      "learning_rate": 1.4570607478064143e-05,
+      "loss": 0.4189,
+      "step": 18420
+    },
+    {
+      "epoch": 3.6251874800226207,
+      "grad_norm": 2.3834993839263916,
+      "learning_rate": 1.456755021553701e-05,
+      "loss": 0.4037,
+      "step": 18430
+    },
+    {
+      "epoch": 3.6271544835386393,
+      "grad_norm": 1.3813104629516602,
+      "learning_rate": 1.4564492953009877e-05,
+      "loss": 0.451,
+      "step": 18440
+    },
+    {
+      "epoch": 3.629121487054658,
+      "grad_norm": 1.1835546493530273,
+      "learning_rate": 1.4561435690482744e-05,
+      "loss": 0.3707,
+      "step": 18450
+    },
+    {
+      "epoch": 3.6310884905706766,
+      "grad_norm": 0.9354956150054932,
+      "learning_rate": 1.455837842795561e-05,
+      "loss": 0.3163,
+      "step": 18460
+    },
+    {
+      "epoch": 3.6330554940866957,
+      "grad_norm": 2.104048490524292,
+      "learning_rate": 1.4555321165428477e-05,
+      "loss": 0.2469,
+      "step": 18470
+    },
+    {
+      "epoch": 3.6350224976027143,
+      "grad_norm": 1.4287692308425903,
+      "learning_rate": 1.4552263902901343e-05,
+      "loss": 0.4124,
+      "step": 18480
+    },
+    {
+      "epoch": 3.636989501118733,
+      "grad_norm": 1.1007471084594727,
+      "learning_rate": 1.4549206640374209e-05,
+      "loss": 0.3916,
+      "step": 18490
+    },
+    {
+      "epoch": 3.638956504634752,
+      "grad_norm": 2.9746992588043213,
+      "learning_rate": 1.4546149377847076e-05,
+      "loss": 0.4604,
+      "step": 18500
+    },
+    {
+      "epoch": 3.638956504634752,
+      "eval_loss": 0.18906159698963165,
+      "eval_runtime": 8.8794,
+      "eval_samples_per_second": 5.631,
+      "eval_steps_per_second": 2.816,
+      "step": 18500
+    },
+    {
+      "epoch": 3.6409235081507707,
+      "grad_norm": 1.591049313545227,
+      "learning_rate": 1.4543092115319944e-05,
+      "loss": 0.4539,
+      "step": 18510
+    },
+    {
+      "epoch": 3.6428905116667893,
+      "grad_norm": 1.869845986366272,
+      "learning_rate": 1.454003485279281e-05,
+      "loss": 0.4262,
+      "step": 18520
+    },
+    {
+      "epoch": 3.6448575151828084,
+      "grad_norm": 2.037968873977661,
+      "learning_rate": 1.4536977590265677e-05,
+      "loss": 0.435,
+      "step": 18530
+    },
+    {
+      "epoch": 3.646824518698827,
+      "grad_norm": 1.494908094406128,
+      "learning_rate": 1.4533920327738545e-05,
+      "loss": 0.3022,
+      "step": 18540
+    },
+    {
+      "epoch": 3.6487915222148457,
+      "grad_norm": 1.9231337308883667,
+      "learning_rate": 1.4530863065211412e-05,
+      "loss": 0.3857,
+      "step": 18550
+    },
+    {
+      "epoch": 3.650758525730865,
+      "grad_norm": 1.4916620254516602,
+      "learning_rate": 1.4527805802684278e-05,
+      "loss": 0.4307,
+      "step": 18560
+    },
+    {
+      "epoch": 3.6527255292468834,
+      "grad_norm": 1.293621301651001,
+      "learning_rate": 1.4524748540157145e-05,
+      "loss": 0.3926,
+      "step": 18570
+    },
+    {
+      "epoch": 3.654692532762902,
+      "grad_norm": 1.8174000978469849,
+      "learning_rate": 1.4521691277630013e-05,
+      "loss": 0.4001,
+      "step": 18580
+    },
+    {
+      "epoch": 3.656659536278921,
+      "grad_norm": 1.4056103229522705,
+      "learning_rate": 1.4518634015102877e-05,
+      "loss": 0.4308,
+      "step": 18590
+    },
+    {
+      "epoch": 3.65862653979494,
+      "grad_norm": 1.1704431772232056,
+      "learning_rate": 1.4515576752575744e-05,
+      "loss": 0.3702,
+      "step": 18600
+    },
+    {
+      "epoch": 3.6605935433109584,
+      "grad_norm": 2.3189661502838135,
+      "learning_rate": 1.4512519490048612e-05,
+      "loss": 0.3239,
+      "step": 18610
+    },
+    {
+      "epoch": 3.6625605468269775,
+      "grad_norm": 2.15673828125,
+      "learning_rate": 1.4509462227521478e-05,
+      "loss": 0.3059,
+      "step": 18620
+    },
+    {
+      "epoch": 3.664527550342996,
+      "grad_norm": 2.0306997299194336,
+      "learning_rate": 1.4506404964994345e-05,
+      "loss": 0.3696,
+      "step": 18630
+    },
+    {
+      "epoch": 3.666494553859015,
+      "grad_norm": 1.7248976230621338,
+      "learning_rate": 1.4503347702467213e-05,
+      "loss": 0.3235,
+      "step": 18640
+    },
+    {
+      "epoch": 3.668461557375034,
+      "grad_norm": 1.1028352975845337,
+      "learning_rate": 1.4500290439940078e-05,
+      "loss": 0.407,
+      "step": 18650
+    },
+    {
+      "epoch": 3.6704285608910525,
+      "grad_norm": 1.5350819826126099,
+      "learning_rate": 1.4497233177412946e-05,
+      "loss": 0.3412,
+      "step": 18660
+    },
+    {
+      "epoch": 3.672395564407071,
+      "grad_norm": 1.2633845806121826,
+      "learning_rate": 1.4494175914885813e-05,
+      "loss": 0.4628,
+      "step": 18670
+    },
+    {
+      "epoch": 3.6743625679230902,
+      "grad_norm": 0.8121323585510254,
+      "learning_rate": 1.449111865235868e-05,
+      "loss": 0.3931,
+      "step": 18680
+    },
+    {
+      "epoch": 3.676329571439109,
+      "grad_norm": 2.546295404434204,
+      "learning_rate": 1.4488061389831546e-05,
+      "loss": 0.3877,
+      "step": 18690
+    },
+    {
+      "epoch": 3.6782965749551275,
+      "grad_norm": 1.443408489227295,
+      "learning_rate": 1.4485004127304412e-05,
+      "loss": 0.3466,
+      "step": 18700
+    },
+    {
+      "epoch": 3.6802635784711466,
+      "grad_norm": 2.1941025257110596,
+      "learning_rate": 1.4481946864777278e-05,
+      "loss": 0.317,
+      "step": 18710
+    },
+    {
+      "epoch": 3.6822305819871652,
+      "grad_norm": 1.4300298690795898,
+      "learning_rate": 1.4478889602250146e-05,
+      "loss": 0.515,
+      "step": 18720
+    },
+    {
+      "epoch": 3.684197585503184,
+      "grad_norm": 1.361330270767212,
+      "learning_rate": 1.4475832339723013e-05,
+      "loss": 0.3974,
+      "step": 18730
+    },
+    {
+      "epoch": 3.686164589019203,
+      "grad_norm": 1.9021443128585815,
+      "learning_rate": 1.447277507719588e-05,
+      "loss": 0.3661,
+      "step": 18740
+    },
+    {
+      "epoch": 3.6881315925352216,
+      "grad_norm": 1.5542855262756348,
+      "learning_rate": 1.4469717814668746e-05,
+      "loss": 0.4889,
+      "step": 18750
+    },
+    {
+      "epoch": 3.6900985960512402,
+      "grad_norm": 1.2739965915679932,
+      "learning_rate": 1.4466660552141614e-05,
+      "loss": 0.4009,
+      "step": 18760
+    },
+    {
+      "epoch": 3.6920655995672593,
+      "grad_norm": 2.3175907135009766,
+      "learning_rate": 1.4463603289614481e-05,
+      "loss": 0.4487,
+      "step": 18770
+    },
+    {
+      "epoch": 3.694032603083278,
+      "grad_norm": 4.204349994659424,
+      "learning_rate": 1.4460546027087347e-05,
+      "loss": 0.3568,
+      "step": 18780
+    },
+    {
+      "epoch": 3.6959996065992966,
+      "grad_norm": 1.1765432357788086,
+      "learning_rate": 1.4457488764560214e-05,
+      "loss": 0.3488,
+      "step": 18790
+    },
+    {
+      "epoch": 3.6979666101153157,
+      "grad_norm": 1.1111738681793213,
+      "learning_rate": 1.4454431502033082e-05,
+      "loss": 0.3493,
+      "step": 18800
+    },
+    {
+      "epoch": 3.6999336136313343,
+      "grad_norm": 0.8277744650840759,
+      "learning_rate": 1.445137423950595e-05,
+      "loss": 0.3763,
+      "step": 18810
+    },
+    {
+      "epoch": 3.701900617147353,
+      "grad_norm": 1.207329511642456,
+      "learning_rate": 1.4448316976978813e-05,
+      "loss": 0.4516,
+      "step": 18820
+    },
+    {
+      "epoch": 3.703867620663372,
+      "grad_norm": 1.5290664434432983,
+      "learning_rate": 1.4445259714451681e-05,
+      "loss": 0.3653,
+      "step": 18830
+    },
+    {
+      "epoch": 3.7058346241793907,
+      "grad_norm": 1.277742862701416,
+      "learning_rate": 1.4442202451924547e-05,
+      "loss": 0.471,
+      "step": 18840
+    },
+    {
+      "epoch": 3.7078016276954093,
+      "grad_norm": 1.5165798664093018,
+      "learning_rate": 1.4439145189397414e-05,
+      "loss": 0.2837,
+      "step": 18850
+    },
+    {
+      "epoch": 3.7097686312114284,
+      "grad_norm": 1.6007874011993408,
+      "learning_rate": 1.4436087926870282e-05,
+      "loss": 0.4247,
+      "step": 18860
+    },
+    {
+      "epoch": 3.711735634727447,
+      "grad_norm": 1.5536600351333618,
+      "learning_rate": 1.4433030664343149e-05,
+      "loss": 0.3162,
+      "step": 18870
+    },
+    {
+      "epoch": 3.7137026382434657,
+      "grad_norm": 1.0041977167129517,
+      "learning_rate": 1.4429973401816015e-05,
+      "loss": 0.4459,
+      "step": 18880
+    },
+    {
+      "epoch": 3.7156696417594848,
+      "grad_norm": 1.6984635591506958,
+      "learning_rate": 1.4426916139288882e-05,
+      "loss": 0.3537,
+      "step": 18890
+    },
+    {
+      "epoch": 3.7176366452755034,
+      "grad_norm": 1.9706878662109375,
+      "learning_rate": 1.442385887676175e-05,
+      "loss": 0.4942,
+      "step": 18900
+    },
+    {
+      "epoch": 3.719603648791522,
+      "grad_norm": 0.7841013669967651,
+      "learning_rate": 1.4420801614234616e-05,
+      "loss": 0.3564,
+      "step": 18910
+    },
+    {
+      "epoch": 3.721570652307541,
+      "grad_norm": 1.4458509683609009,
+      "learning_rate": 1.4417744351707483e-05,
+      "loss": 0.3577,
+      "step": 18920
+    },
+    {
+      "epoch": 3.7235376558235598,
+      "grad_norm": 2.219909906387329,
+      "learning_rate": 1.4414687089180349e-05,
+      "loss": 0.5433,
+      "step": 18930
+    },
+    {
+      "epoch": 3.7255046593395784,
+      "grad_norm": 0.8798016905784607,
+      "learning_rate": 1.4411629826653215e-05,
+      "loss": 0.3507,
+      "step": 18940
+    },
+    {
+      "epoch": 3.7274716628555975,
+      "grad_norm": 1.1821753978729248,
+      "learning_rate": 1.4408572564126082e-05,
+      "loss": 0.4787,
+      "step": 18950
+    },
+    {
+      "epoch": 3.729438666371616,
+      "grad_norm": 0.8908816576004028,
+      "learning_rate": 1.440551530159895e-05,
+      "loss": 0.358,
+      "step": 18960
+    },
+    {
+      "epoch": 3.731405669887635,
+      "grad_norm": 0.9876505732536316,
+      "learning_rate": 1.4402458039071815e-05,
+      "loss": 0.4758,
+      "step": 18970
+    },
+    {
+      "epoch": 3.733372673403654,
+      "grad_norm": 0.9503434896469116,
+      "learning_rate": 1.4399400776544683e-05,
+      "loss": 0.3929,
+      "step": 18980
+    },
+    {
+      "epoch": 3.7353396769196725,
+      "grad_norm": 0.9545804262161255,
+      "learning_rate": 1.439634351401755e-05,
+      "loss": 0.4286,
+      "step": 18990
+    },
+    {
+      "epoch": 3.737306680435691,
+      "grad_norm": 2.3993990421295166,
+      "learning_rate": 1.4393286251490418e-05,
+      "loss": 0.3785,
+      "step": 19000
+    },
+    {
+      "epoch": 3.737306680435691,
+      "eval_loss": 0.17065879702568054,
+      "eval_runtime": 8.8951,
+      "eval_samples_per_second": 5.621,
+      "eval_steps_per_second": 2.811,
+      "step": 19000
+    },
+    {
+      "epoch": 3.7392736839517102,
+      "grad_norm": 2.2503209114074707,
+      "learning_rate": 1.4390228988963284e-05,
+      "loss": 0.4456,
+      "step": 19010
+    },
+    {
+      "epoch": 3.741240687467729,
+      "grad_norm": 1.7953013181686401,
+      "learning_rate": 1.4387171726436151e-05,
+      "loss": 0.3437,
+      "step": 19020
+    },
+    {
+      "epoch": 3.7432076909837475,
+      "grad_norm": 3.9343481063842773,
+      "learning_rate": 1.4384114463909019e-05,
+      "loss": 0.4451,
+      "step": 19030
+    },
+    {
+      "epoch": 3.7451746944997666,
+      "grad_norm": 1.1897847652435303,
+      "learning_rate": 1.4381057201381883e-05,
+      "loss": 0.4459,
+      "step": 19040
+    },
+    {
+      "epoch": 3.7471416980157852,
+      "grad_norm": 1.6288838386535645,
+      "learning_rate": 1.437799993885475e-05,
+      "loss": 0.4209,
+      "step": 19050
+    },
+    {
+      "epoch": 3.749108701531804,
+      "grad_norm": 1.8541220426559448,
+      "learning_rate": 1.4374942676327618e-05,
+      "loss": 0.4737,
+      "step": 19060
+    },
+    {
+      "epoch": 3.751075705047823,
+      "grad_norm": 1.3628418445587158,
+      "learning_rate": 1.4371885413800483e-05,
+      "loss": 0.4486,
+      "step": 19070
+    },
+    {
+      "epoch": 3.7530427085638416,
+      "grad_norm": 1.4050499200820923,
+      "learning_rate": 1.436882815127335e-05,
+      "loss": 0.39,
+      "step": 19080
+    },
+    {
+      "epoch": 3.7550097120798602,
+      "grad_norm": 0.8367191553115845,
+      "learning_rate": 1.4365770888746218e-05,
+      "loss": 0.3072,
+      "step": 19090
+    },
+    {
+      "epoch": 3.7569767155958793,
+      "grad_norm": 1.374824047088623,
+      "learning_rate": 1.4362713626219084e-05,
+      "loss": 0.3958,
+      "step": 19100
+    },
+    {
+      "epoch": 3.758943719111898,
+      "grad_norm": 1.165436029434204,
+      "learning_rate": 1.4359656363691952e-05,
+      "loss": 0.4409,
+      "step": 19110
+    },
+    {
+      "epoch": 3.7609107226279166,
+      "grad_norm": 1.6920998096466064,
+      "learning_rate": 1.4356599101164819e-05,
+      "loss": 0.4824,
+      "step": 19120
+    },
+    {
+      "epoch": 3.7628777261439357,
+      "grad_norm": 0.9310535788536072,
+      "learning_rate": 1.4353541838637686e-05,
+      "loss": 0.4144,
+      "step": 19130
+    },
+    {
+      "epoch": 3.7648447296599543,
+      "grad_norm": 1.4429125785827637,
+      "learning_rate": 1.4350484576110552e-05,
+      "loss": 0.2829,
+      "step": 19140
+    },
+    {
+      "epoch": 3.766811733175973,
+      "grad_norm": 1.7485815286636353,
+      "learning_rate": 1.434742731358342e-05,
+      "loss": 0.5212,
+      "step": 19150
+    },
+    {
+      "epoch": 3.768778736691992,
+      "grad_norm": 1.283400535583496,
+      "learning_rate": 1.4344370051056284e-05,
+      "loss": 0.4345,
+      "step": 19160
+    },
+    {
+      "epoch": 3.7707457402080107,
+      "grad_norm": 1.8467475175857544,
+      "learning_rate": 1.4341312788529151e-05,
+      "loss": 0.2751,
+      "step": 19170
+    },
+    {
+      "epoch": 3.7727127437240293,
+      "grad_norm": 1.880601406097412,
+      "learning_rate": 1.4338255526002019e-05,
+      "loss": 0.347,
+      "step": 19180
+    },
+    {
+      "epoch": 3.7746797472400484,
+      "grad_norm": 2.1124298572540283,
+      "learning_rate": 1.4335198263474886e-05,
+      "loss": 0.5249,
+      "step": 19190
+    },
+    {
+      "epoch": 3.776646750756067,
+      "grad_norm": 1.1820361614227295,
+      "learning_rate": 1.4332141000947752e-05,
+      "loss": 0.3955,
+      "step": 19200
+    },
+    {
+      "epoch": 3.7786137542720857,
+      "grad_norm": 1.8350051641464233,
+      "learning_rate": 1.432908373842062e-05,
+      "loss": 0.5027,
+      "step": 19210
+    },
+    {
+      "epoch": 3.7805807577881048,
+      "grad_norm": 1.0418702363967896,
+      "learning_rate": 1.4326026475893487e-05,
+      "loss": 0.4091,
+      "step": 19220
+    },
+    {
+      "epoch": 3.7825477613041234,
+      "grad_norm": 1.2310411930084229,
+      "learning_rate": 1.4322969213366353e-05,
+      "loss": 0.3877,
+      "step": 19230
+    },
+    {
+      "epoch": 3.784514764820142,
+      "grad_norm": 1.0506771802902222,
+      "learning_rate": 1.431991195083922e-05,
+      "loss": 0.3961,
+      "step": 19240
+    },
+    {
+      "epoch": 3.786481768336161,
+      "grad_norm": 1.3733229637145996,
+      "learning_rate": 1.4316854688312088e-05,
+      "loss": 0.3588,
+      "step": 19250
+    },
+    {
+      "epoch": 3.7884487718521798,
+      "grad_norm": 1.6931480169296265,
+      "learning_rate": 1.4313797425784955e-05,
+      "loss": 0.4726,
+      "step": 19260
+    },
+    {
+      "epoch": 3.7904157753681984,
+      "grad_norm": 1.191991925239563,
+      "learning_rate": 1.431074016325782e-05,
+      "loss": 0.3923,
+      "step": 19270
+    },
+    {
+      "epoch": 3.7923827788842175,
+      "grad_norm": 2.010301351547241,
+      "learning_rate": 1.4307682900730687e-05,
+      "loss": 0.4324,
+      "step": 19280
+    },
+    {
+      "epoch": 3.794349782400236,
+      "grad_norm": 0.7601318359375,
+      "learning_rate": 1.4304625638203552e-05,
+      "loss": 0.5484,
+      "step": 19290
+    },
+    {
+      "epoch": 3.7963167859162548,
+      "grad_norm": 1.0302248001098633,
+      "learning_rate": 1.430156837567642e-05,
+      "loss": 0.4386,
+      "step": 19300
+    },
+    {
+      "epoch": 3.798283789432274,
+      "grad_norm": 1.992654800415039,
+      "learning_rate": 1.4298511113149287e-05,
+      "loss": 0.34,
+      "step": 19310
+    },
+    {
+      "epoch": 3.8002507929482925,
+      "grad_norm": 1.9000381231307983,
+      "learning_rate": 1.4295453850622155e-05,
+      "loss": 0.5453,
+      "step": 19320
+    },
+    {
+      "epoch": 3.802217796464311,
+      "grad_norm": 1.3942879438400269,
+      "learning_rate": 1.429239658809502e-05,
+      "loss": 0.4681,
+      "step": 19330
+    },
+    {
+      "epoch": 3.80418479998033,
+      "grad_norm": 0.8642085790634155,
+      "learning_rate": 1.4289339325567888e-05,
+      "loss": 0.3664,
+      "step": 19340
+    },
+    {
+      "epoch": 3.806151803496349,
+      "grad_norm": 1.3701379299163818,
+      "learning_rate": 1.4286282063040756e-05,
+      "loss": 0.3545,
+      "step": 19350
+    },
+    {
+      "epoch": 3.8081188070123675,
+      "grad_norm": 1.2130693197250366,
+      "learning_rate": 1.4283224800513621e-05,
+      "loss": 0.353,
+      "step": 19360
+    },
+    {
+      "epoch": 3.8100858105283866,
+      "grad_norm": 1.9979584217071533,
+      "learning_rate": 1.4280167537986489e-05,
+      "loss": 0.4111,
+      "step": 19370
+    },
+    {
+      "epoch": 3.812052814044405,
+      "grad_norm": 1.0861486196517944,
+      "learning_rate": 1.4277110275459355e-05,
+      "loss": 0.3785,
+      "step": 19380
+    },
+    {
+      "epoch": 3.814019817560424,
+      "grad_norm": 1.5842684507369995,
+      "learning_rate": 1.427405301293222e-05,
+      "loss": 0.3934,
+      "step": 19390
+    },
+    {
+      "epoch": 3.815986821076443,
+      "grad_norm": 1.0454131364822388,
+      "learning_rate": 1.4270995750405088e-05,
+      "loss": 0.4427,
+      "step": 19400
+    },
+    {
+      "epoch": 3.8179538245924616,
+      "grad_norm": 0.670005202293396,
+      "learning_rate": 1.4267938487877955e-05,
+      "loss": 0.4881,
+      "step": 19410
+    },
+    {
+      "epoch": 3.81992082810848,
+      "grad_norm": 0.9665763974189758,
+      "learning_rate": 1.4264881225350821e-05,
+      "loss": 0.3345,
+      "step": 19420
+    },
+    {
+      "epoch": 3.8218878316244993,
+      "grad_norm": 0.7225205302238464,
+      "learning_rate": 1.4261823962823689e-05,
+      "loss": 0.4946,
+      "step": 19430
+    },
+    {
+      "epoch": 3.823854835140518,
+      "grad_norm": 1.0303906202316284,
+      "learning_rate": 1.4258766700296556e-05,
+      "loss": 0.4415,
+      "step": 19440
+    },
+    {
+      "epoch": 3.8258218386565366,
+      "grad_norm": 1.3895282745361328,
+      "learning_rate": 1.4255709437769424e-05,
+      "loss": 0.3707,
+      "step": 19450
+    },
+    {
+      "epoch": 3.827788842172555,
+      "grad_norm": 1.1078003644943237,
+      "learning_rate": 1.425265217524229e-05,
+      "loss": 0.4026,
+      "step": 19460
+    },
+    {
+      "epoch": 3.8297558456885743,
+      "grad_norm": 2.3170745372772217,
+      "learning_rate": 1.4249594912715157e-05,
+      "loss": 0.341,
+      "step": 19470
+    },
+    {
+      "epoch": 3.831722849204593,
+      "grad_norm": 1.1369905471801758,
+      "learning_rate": 1.4246537650188024e-05,
+      "loss": 0.4851,
+      "step": 19480
+    },
+    {
+      "epoch": 3.8336898527206116,
+      "grad_norm": 1.0877054929733276,
+      "learning_rate": 1.424348038766089e-05,
+      "loss": 0.4159,
+      "step": 19490
+    },
+    {
+      "epoch": 3.8356568562366307,
+      "grad_norm": 1.8775924444198608,
+      "learning_rate": 1.4240423125133756e-05,
+      "loss": 0.3514,
+      "step": 19500
+    },
+    {
+      "epoch": 3.8356568562366307,
+      "eval_loss": 0.18426425755023956,
+      "eval_runtime": 8.8833,
+      "eval_samples_per_second": 5.629,
+      "eval_steps_per_second": 2.814,
+      "step": 19500
+    },
+    {
+      "epoch": 3.8376238597526493,
+      "grad_norm": 1.6352936029434204,
+      "learning_rate": 1.4237365862606623e-05,
+      "loss": 0.6251,
+      "step": 19510
+    },
+    {
+      "epoch": 3.839590863268668,
+      "grad_norm": 1.4518588781356812,
+      "learning_rate": 1.4234308600079489e-05,
+      "loss": 0.3573,
+      "step": 19520
+    },
+    {
+      "epoch": 3.841557866784687,
+      "grad_norm": 1.4944310188293457,
+      "learning_rate": 1.4231251337552357e-05,
+      "loss": 0.4555,
+      "step": 19530
+    },
+    {
+      "epoch": 3.8435248703007057,
+      "grad_norm": 1.3529949188232422,
+      "learning_rate": 1.4228194075025224e-05,
+      "loss": 0.3766,
+      "step": 19540
+    },
+    {
+      "epoch": 3.8454918738167243,
+      "grad_norm": 1.6837838888168335,
+      "learning_rate": 1.422513681249809e-05,
+      "loss": 0.4189,
+      "step": 19550
+    },
+    {
+      "epoch": 3.8474588773327434,
+      "grad_norm": 0.719254732131958,
+      "learning_rate": 1.4222079549970957e-05,
+      "loss": 0.4939,
+      "step": 19560
+    },
+    {
+      "epoch": 3.849425880848762,
+      "grad_norm": 1.1821671724319458,
+      "learning_rate": 1.4219022287443825e-05,
+      "loss": 0.4553,
+      "step": 19570
+    },
+    {
+      "epoch": 3.8513928843647807,
+      "grad_norm": 1.264253854751587,
+      "learning_rate": 1.4215965024916692e-05,
+      "loss": 0.52,
+      "step": 19580
+    },
+    {
+      "epoch": 3.8533598878807998,
+      "grad_norm": 1.4270673990249634,
+      "learning_rate": 1.4212907762389558e-05,
+      "loss": 0.3172,
+      "step": 19590
+    },
+    {
+      "epoch": 3.8553268913968184,
+      "grad_norm": 1.0974931716918945,
+      "learning_rate": 1.4209850499862425e-05,
+      "loss": 0.4648,
+      "step": 19600
+    },
+    {
+      "epoch": 3.857293894912837,
+      "grad_norm": 1.9445582628250122,
+      "learning_rate": 1.420679323733529e-05,
+      "loss": 0.3901,
+      "step": 19610
+    },
+    {
+      "epoch": 3.8592608984288557,
+      "grad_norm": 1.3331164121627808,
+      "learning_rate": 1.4203735974808157e-05,
+      "loss": 0.394,
+      "step": 19620
+    },
+    {
+      "epoch": 3.8612279019448748,
+      "grad_norm": 1.6156160831451416,
+      "learning_rate": 1.4200678712281024e-05,
+      "loss": 0.352,
+      "step": 19630
+    },
+    {
+      "epoch": 3.8631949054608934,
+      "grad_norm": 1.9054970741271973,
+      "learning_rate": 1.4197621449753892e-05,
+      "loss": 0.455,
+      "step": 19640
+    },
+    {
+      "epoch": 3.865161908976912,
+      "grad_norm": 1.2336102724075317,
+      "learning_rate": 1.4194564187226758e-05,
+      "loss": 0.3731,
+      "step": 19650
+    },
+    {
+      "epoch": 3.867128912492931,
+      "grad_norm": 1.1675928831100464,
+      "learning_rate": 1.4191506924699625e-05,
+      "loss": 0.3594,
+      "step": 19660
+    },
+    {
+      "epoch": 3.8690959160089498,
+      "grad_norm": 1.6791987419128418,
+      "learning_rate": 1.4188449662172493e-05,
+      "loss": 0.3734,
+      "step": 19670
+    },
+    {
+      "epoch": 3.8710629195249684,
+      "grad_norm": 1.1804096698760986,
+      "learning_rate": 1.4185392399645358e-05,
+      "loss": 0.3333,
+      "step": 19680
+    },
+    {
+      "epoch": 3.8730299230409875,
+      "grad_norm": 0.658993661403656,
+      "learning_rate": 1.4182335137118226e-05,
+      "loss": 0.27,
+      "step": 19690
+    },
+    {
+      "epoch": 3.874996926557006,
+      "grad_norm": 1.942895770072937,
+      "learning_rate": 1.4179277874591093e-05,
+      "loss": 0.2867,
+      "step": 19700
+    },
+    {
+      "epoch": 3.8769639300730248,
+      "grad_norm": 2.464463233947754,
+      "learning_rate": 1.4176220612063961e-05,
+      "loss": 0.4725,
+      "step": 19710
+    },
+    {
+      "epoch": 3.878930933589044,
+      "grad_norm": 1.6073675155639648,
+      "learning_rate": 1.4173163349536825e-05,
+      "loss": 0.3144,
+      "step": 19720
+    },
+    {
+      "epoch": 3.8808979371050625,
+      "grad_norm": 1.7023011445999146,
+      "learning_rate": 1.4170106087009692e-05,
+      "loss": 0.3217,
+      "step": 19730
+    },
+    {
+      "epoch": 3.882864940621081,
+      "grad_norm": 3.136507749557495,
+      "learning_rate": 1.4167048824482558e-05,
+      "loss": 0.4172,
+      "step": 19740
+    },
+    {
+      "epoch": 3.8848319441371,
+      "grad_norm": 1.6217586994171143,
+      "learning_rate": 1.4163991561955426e-05,
+      "loss": 0.5113,
+      "step": 19750
+    },
+    {
+      "epoch": 3.886798947653119,
+      "grad_norm": 1.7035020589828491,
+      "learning_rate": 1.4160934299428293e-05,
+      "loss": 0.379,
+      "step": 19760
+    },
+    {
+      "epoch": 3.8887659511691375,
+      "grad_norm": 1.4003608226776123,
+      "learning_rate": 1.415787703690116e-05,
+      "loss": 0.3828,
+      "step": 19770
+    },
+    {
+      "epoch": 3.8907329546851566,
+      "grad_norm": 1.6719714403152466,
+      "learning_rate": 1.4154819774374026e-05,
+      "loss": 0.3379,
+      "step": 19780
+    },
+    {
+      "epoch": 3.892699958201175,
+      "grad_norm": 1.173722743988037,
+      "learning_rate": 1.4151762511846894e-05,
+      "loss": 0.3918,
+      "step": 19790
+    },
+    {
+      "epoch": 3.894666961717194,
+      "grad_norm": 0.8642592430114746,
+      "learning_rate": 1.4148705249319761e-05,
+      "loss": 0.4654,
+      "step": 19800
+    },
+    {
+      "epoch": 3.896633965233213,
+      "grad_norm": 2.1731534004211426,
+      "learning_rate": 1.4145647986792627e-05,
+      "loss": 0.4032,
+      "step": 19810
+    },
+    {
+      "epoch": 3.8986009687492316,
+      "grad_norm": 1.903075933456421,
+      "learning_rate": 1.4142590724265495e-05,
+      "loss": 0.3913,
+      "step": 19820
+    },
+    {
+      "epoch": 3.90056797226525,
+      "grad_norm": 2.3391592502593994,
+      "learning_rate": 1.4139533461738362e-05,
+      "loss": 0.3946,
+      "step": 19830
+    },
+    {
+      "epoch": 3.9025349757812693,
+      "grad_norm": 1.1577038764953613,
+      "learning_rate": 1.4136476199211226e-05,
+      "loss": 0.4448,
+      "step": 19840
+    },
+    {
+      "epoch": 3.904501979297288,
+      "grad_norm": 1.8592039346694946,
+      "learning_rate": 1.4133418936684094e-05,
+      "loss": 0.3715,
+      "step": 19850
+    },
+    {
+      "epoch": 3.9064689828133066,
+      "grad_norm": 1.1040771007537842,
+      "learning_rate": 1.4130361674156961e-05,
+      "loss": 0.3806,
+      "step": 19860
+    },
+    {
+      "epoch": 3.9084359863293257,
+      "grad_norm": 1.3957312107086182,
+      "learning_rate": 1.4127304411629827e-05,
+      "loss": 0.431,
+      "step": 19870
+    },
+    {
+      "epoch": 3.9104029898453443,
+      "grad_norm": 1.2022626399993896,
+      "learning_rate": 1.4124247149102694e-05,
+      "loss": 0.2885,
+      "step": 19880
+    },
+    {
+      "epoch": 3.912369993361363,
+      "grad_norm": 2.126249074935913,
+      "learning_rate": 1.4121189886575562e-05,
+      "loss": 0.45,
+      "step": 19890
+    },
+    {
+      "epoch": 3.914336996877382,
+      "grad_norm": 2.9966373443603516,
+      "learning_rate": 1.411813262404843e-05,
+      "loss": 0.5084,
+      "step": 19900
+    },
+    {
+      "epoch": 3.9163040003934007,
+      "grad_norm": 0.9211567640304565,
+      "learning_rate": 1.4115075361521295e-05,
+      "loss": 0.4613,
+      "step": 19910
+    },
+    {
+      "epoch": 3.9182710039094193,
+      "grad_norm": 1.7329440116882324,
+      "learning_rate": 1.4112018098994163e-05,
+      "loss": 0.3291,
+      "step": 19920
+    },
+    {
+      "epoch": 3.9202380074254384,
+      "grad_norm": 2.0354349613189697,
+      "learning_rate": 1.410896083646703e-05,
+      "loss": 0.4231,
+      "step": 19930
+    },
+    {
+      "epoch": 3.922205010941457,
+      "grad_norm": 19.046695709228516,
+      "learning_rate": 1.4105903573939896e-05,
+      "loss": 0.4676,
+      "step": 19940
+    },
+    {
+      "epoch": 3.9241720144574757,
+      "grad_norm": 1.4789849519729614,
+      "learning_rate": 1.4102846311412762e-05,
+      "loss": 0.5275,
+      "step": 19950
+    },
+    {
+      "epoch": 3.9261390179734947,
+      "grad_norm": 0.89300936460495,
+      "learning_rate": 1.4099789048885627e-05,
+      "loss": 0.4568,
+      "step": 19960
+    },
+    {
+      "epoch": 3.9281060214895134,
+      "grad_norm": 1.1245640516281128,
+      "learning_rate": 1.4096731786358495e-05,
+      "loss": 0.4998,
+      "step": 19970
+    },
+    {
+      "epoch": 3.930073025005532,
+      "grad_norm": 1.9542558193206787,
+      "learning_rate": 1.4093674523831362e-05,
+      "loss": 0.4368,
+      "step": 19980
+    },
+    {
+      "epoch": 3.932040028521551,
+      "grad_norm": 1.956928014755249,
+      "learning_rate": 1.409061726130423e-05,
+      "loss": 0.4154,
+      "step": 19990
+    },
+    {
+      "epoch": 3.9340070320375697,
+      "grad_norm": 2.864295721054077,
+      "learning_rate": 1.4087559998777096e-05,
+      "loss": 0.3798,
+      "step": 20000
+    },
+    {
+      "epoch": 3.9340070320375697,
+      "eval_loss": 0.16804826259613037,
+      "eval_runtime": 8.8889,
+      "eval_samples_per_second": 5.625,
+      "eval_steps_per_second": 2.812,
+      "step": 20000
+    },
+    {
+      "epoch": 3.9359740355535884,
+      "grad_norm": 1.1259634494781494,
+      "learning_rate": 1.4084502736249963e-05,
+      "loss": 0.3126,
+      "step": 20010
+    },
+    {
+      "epoch": 3.9379410390696075,
+      "grad_norm": 1.4045112133026123,
+      "learning_rate": 1.408144547372283e-05,
+      "loss": 0.3549,
+      "step": 20020
+    },
+    {
+      "epoch": 3.939908042585626,
+      "grad_norm": 1.2622108459472656,
+      "learning_rate": 1.4078388211195698e-05,
+      "loss": 0.3867,
+      "step": 20030
+    },
+    {
+      "epoch": 3.9418750461016447,
+      "grad_norm": 1.6983309984207153,
+      "learning_rate": 1.4075330948668564e-05,
+      "loss": 0.3751,
+      "step": 20040
+    },
+    {
+      "epoch": 3.943842049617664,
+      "grad_norm": 1.3586803674697876,
+      "learning_rate": 1.4072273686141431e-05,
+      "loss": 0.6043,
+      "step": 20050
+    },
+    {
+      "epoch": 3.9458090531336825,
+      "grad_norm": 0.9066451191902161,
+      "learning_rate": 1.4069216423614295e-05,
+      "loss": 0.3718,
+      "step": 20060
+    },
+    {
+      "epoch": 3.947776056649701,
+      "grad_norm": 1.4924407005310059,
+      "learning_rate": 1.4066159161087163e-05,
+      "loss": 0.3764,
+      "step": 20070
+    },
+    {
+      "epoch": 3.94974306016572,
+      "grad_norm": 1.226970911026001,
+      "learning_rate": 1.406310189856003e-05,
+      "loss": 0.3908,
+      "step": 20080
+    },
+    {
+      "epoch": 3.951710063681739,
+      "grad_norm": 0.9943327307701111,
+      "learning_rate": 1.4060044636032896e-05,
+      "loss": 0.4514,
+      "step": 20090
+    },
+    {
+      "epoch": 3.9536770671977575,
+      "grad_norm": 1.371333360671997,
+      "learning_rate": 1.4056987373505763e-05,
+      "loss": 0.4506,
+      "step": 20100
+    },
+    {
+      "epoch": 3.9556440707137766,
+      "grad_norm": 1.717349648475647,
+      "learning_rate": 1.4053930110978631e-05,
+      "loss": 0.3844,
+      "step": 20110
+    },
+    {
+      "epoch": 3.957611074229795,
+      "grad_norm": 2.133634090423584,
+      "learning_rate": 1.4050872848451498e-05,
+      "loss": 0.4074,
+      "step": 20120
+    },
+    {
+      "epoch": 3.959578077745814,
+      "grad_norm": 1.379530906677246,
+      "learning_rate": 1.4047815585924364e-05,
+      "loss": 0.4095,
+      "step": 20130
+    },
+    {
+      "epoch": 3.961545081261833,
+      "grad_norm": 1.0096360445022583,
+      "learning_rate": 1.4044758323397232e-05,
+      "loss": 0.3714,
+      "step": 20140
+    },
+    {
+      "epoch": 3.9635120847778516,
+      "grad_norm": 2.1940183639526367,
+      "learning_rate": 1.4041701060870099e-05,
+      "loss": 0.4436,
+      "step": 20150
+    },
+    {
+      "epoch": 3.96547908829387,
+      "grad_norm": 1.4452601671218872,
+      "learning_rate": 1.4038643798342967e-05,
+      "loss": 0.4699,
+      "step": 20160
+    },
+    {
+      "epoch": 3.9674460918098893,
+      "grad_norm": 1.9705860614776611,
+      "learning_rate": 1.4035586535815832e-05,
+      "loss": 0.386,
+      "step": 20170
+    },
+    {
+      "epoch": 3.969413095325908,
+      "grad_norm": 1.5604002475738525,
+      "learning_rate": 1.4032529273288698e-05,
+      "loss": 0.4451,
+      "step": 20180
+    },
+    {
+      "epoch": 3.9713800988419266,
+      "grad_norm": 1.4916459321975708,
+      "learning_rate": 1.4029472010761564e-05,
+      "loss": 0.3558,
+      "step": 20190
+    },
+    {
+      "epoch": 3.9733471023579456,
+      "grad_norm": 2.563976287841797,
+      "learning_rate": 1.4026414748234431e-05,
+      "loss": 0.3958,
+      "step": 20200
+    },
+    {
+      "epoch": 3.9753141058739643,
+      "grad_norm": 2.008441209793091,
+      "learning_rate": 1.4023357485707299e-05,
+      "loss": 0.4369,
+      "step": 20210
+    },
+    {
+      "epoch": 3.977281109389983,
+      "grad_norm": 1.1274158954620361,
+      "learning_rate": 1.4020300223180165e-05,
+      "loss": 0.2473,
+      "step": 20220
+    },
+    {
+      "epoch": 3.979248112906002,
+      "grad_norm": 1.1995351314544678,
+      "learning_rate": 1.4017242960653032e-05,
+      "loss": 0.4072,
+      "step": 20230
+    },
+    {
+      "epoch": 3.9812151164220206,
+      "grad_norm": 1.7753493785858154,
+      "learning_rate": 1.40141856981259e-05,
+      "loss": 0.3449,
+      "step": 20240
+    },
+    {
+      "epoch": 3.9831821199380393,
+      "grad_norm": 2.2701807022094727,
+      "learning_rate": 1.4011128435598767e-05,
+      "loss": 0.4204,
+      "step": 20250
+    },
+    {
+      "epoch": 3.9851491234540584,
+      "grad_norm": 1.2807681560516357,
+      "learning_rate": 1.4008071173071633e-05,
+      "loss": 0.2937,
+      "step": 20260
+    },
+    {
+      "epoch": 3.987116126970077,
+      "grad_norm": 0.9695661664009094,
+      "learning_rate": 1.40050139105445e-05,
+      "loss": 0.382,
+      "step": 20270
+    },
+    {
+      "epoch": 3.9890831304860956,
+      "grad_norm": 2.4322593212127686,
+      "learning_rate": 1.4001956648017368e-05,
+      "loss": 0.4999,
+      "step": 20280
+    },
+    {
+      "epoch": 3.9910501340021147,
+      "grad_norm": 1.0660431385040283,
+      "learning_rate": 1.3998899385490232e-05,
+      "loss": 0.406,
+      "step": 20290
+    },
+    {
+      "epoch": 3.9930171375181334,
+      "grad_norm": 1.3648877143859863,
+      "learning_rate": 1.39958421229631e-05,
+      "loss": 0.4379,
+      "step": 20300
+    },
+    {
+      "epoch": 3.994984141034152,
+      "grad_norm": 1.5317339897155762,
+      "learning_rate": 1.3992784860435967e-05,
+      "loss": 0.4288,
+      "step": 20310
+    },
+    {
+      "epoch": 3.996951144550171,
+      "grad_norm": 1.7145333290100098,
+      "learning_rate": 1.3989727597908833e-05,
+      "loss": 0.3317,
+      "step": 20320
+    },
+    {
+      "epoch": 3.9989181480661897,
+      "grad_norm": 3.5187952518463135,
+      "learning_rate": 1.39866703353817e-05,
+      "loss": 0.4305,
+      "step": 20330
+    },
+    {
+      "epoch": 4.000885151582208,
+      "grad_norm": 0.9262216091156006,
+      "learning_rate": 1.3983613072854568e-05,
+      "loss": 0.3561,
+      "step": 20340
+    },
+    {
+      "epoch": 4.0028521550982274,
+      "grad_norm": 1.3652830123901367,
+      "learning_rate": 1.3980555810327433e-05,
+      "loss": 0.4874,
+      "step": 20350
+    },
+    {
+      "epoch": 4.004819158614246,
+      "grad_norm": 0.9129854440689087,
+      "learning_rate": 1.39774985478003e-05,
+      "loss": 0.3942,
+      "step": 20360
+    },
+    {
+      "epoch": 4.006786162130265,
+      "grad_norm": 1.0334409475326538,
+      "learning_rate": 1.3974441285273168e-05,
+      "loss": 0.3191,
+      "step": 20370
+    },
+    {
+      "epoch": 4.008753165646284,
+      "grad_norm": 1.292149543762207,
+      "learning_rate": 1.3971384022746036e-05,
+      "loss": 0.4056,
+      "step": 20380
+    },
+    {
+      "epoch": 4.010720169162302,
+      "grad_norm": 1.2655421495437622,
+      "learning_rate": 1.3968326760218902e-05,
+      "loss": 0.2615,
+      "step": 20390
+    },
+    {
+      "epoch": 4.012687172678321,
+      "grad_norm": 3.0423643589019775,
+      "learning_rate": 1.3965269497691767e-05,
+      "loss": 0.3888,
+      "step": 20400
+    },
+    {
+      "epoch": 4.01465417619434,
+      "grad_norm": 1.6518824100494385,
+      "learning_rate": 1.3962212235164633e-05,
+      "loss": 0.3371,
+      "step": 20410
+    },
+    {
+      "epoch": 4.016621179710358,
+      "grad_norm": 1.5549308061599731,
+      "learning_rate": 1.39591549726375e-05,
+      "loss": 0.4344,
+      "step": 20420
+    },
+    {
+      "epoch": 4.0185881832263775,
+      "grad_norm": 1.3199830055236816,
+      "learning_rate": 1.3956097710110368e-05,
+      "loss": 0.4384,
+      "step": 20430
+    },
+    {
+      "epoch": 4.0205551867423965,
+      "grad_norm": 1.8251094818115234,
+      "learning_rate": 1.3953040447583235e-05,
+      "loss": 0.3351,
+      "step": 20440
+    },
+    {
+      "epoch": 4.022522190258415,
+      "grad_norm": 1.6318509578704834,
+      "learning_rate": 1.3949983185056101e-05,
+      "loss": 0.4346,
+      "step": 20450
+    },
+    {
+      "epoch": 4.024489193774434,
+      "grad_norm": 1.016904354095459,
+      "learning_rate": 1.3946925922528969e-05,
+      "loss": 0.3728,
+      "step": 20460
+    },
+    {
+      "epoch": 4.026456197290453,
+      "grad_norm": 1.3480511903762817,
+      "learning_rate": 1.3943868660001836e-05,
+      "loss": 0.2728,
+      "step": 20470
+    },
+    {
+      "epoch": 4.028423200806471,
+      "grad_norm": 1.0706782341003418,
+      "learning_rate": 1.3940811397474702e-05,
+      "loss": 0.4899,
+      "step": 20480
+    },
+    {
+      "epoch": 4.03039020432249,
+      "grad_norm": 2.8634934425354004,
+      "learning_rate": 1.393775413494757e-05,
+      "loss": 0.4906,
+      "step": 20490
+    },
+    {
+      "epoch": 4.032357207838509,
+      "grad_norm": 1.0132216215133667,
+      "learning_rate": 1.3934696872420437e-05,
+      "loss": 0.3943,
+      "step": 20500
+    },
+    {
+      "epoch": 4.032357207838509,
+      "eval_loss": 0.16937392950057983,
+      "eval_runtime": 8.8749,
+      "eval_samples_per_second": 5.634,
+      "eval_steps_per_second": 2.817,
+      "step": 20500
+    },
+    {
+      "epoch": 4.0343242113545275,
+      "grad_norm": 1.4437650442123413,
+      "learning_rate": 1.3931639609893304e-05,
+      "loss": 0.3685,
+      "step": 20510
+    },
+    {
+      "epoch": 4.0362912148705465,
+      "grad_norm": 1.7009004354476929,
+      "learning_rate": 1.3928582347366168e-05,
+      "loss": 0.4131,
+      "step": 20520
+    },
+    {
+      "epoch": 4.038258218386566,
+      "grad_norm": 1.1112629175186157,
+      "learning_rate": 1.3925525084839036e-05,
+      "loss": 0.4157,
+      "step": 20530
+    },
+    {
+      "epoch": 4.040225221902584,
+      "grad_norm": 1.0249930620193481,
+      "learning_rate": 1.3922467822311902e-05,
+      "loss": 0.4676,
+      "step": 20540
+    },
+    {
+      "epoch": 4.042192225418603,
+      "grad_norm": 1.9239535331726074,
+      "learning_rate": 1.391941055978477e-05,
+      "loss": 0.3815,
+      "step": 20550
+    },
+    {
+      "epoch": 4.044159228934622,
+      "grad_norm": 1.6564403772354126,
+      "learning_rate": 1.3916353297257637e-05,
+      "loss": 0.3594,
+      "step": 20560
+    },
+    {
+      "epoch": 4.04612623245064,
+      "grad_norm": 1.2980719804763794,
+      "learning_rate": 1.3913296034730504e-05,
+      "loss": 0.2283,
+      "step": 20570
+    },
+    {
+      "epoch": 4.048093235966659,
+      "grad_norm": 0.801551878452301,
+      "learning_rate": 1.391023877220337e-05,
+      "loss": 0.3371,
+      "step": 20580
+    },
+    {
+      "epoch": 4.050060239482678,
+      "grad_norm": 2.342517852783203,
+      "learning_rate": 1.3907181509676237e-05,
+      "loss": 0.3808,
+      "step": 20590
+    },
+    {
+      "epoch": 4.0520272429986965,
+      "grad_norm": 2.5508110523223877,
+      "learning_rate": 1.3904124247149105e-05,
+      "loss": 0.4242,
+      "step": 20600
+    },
+    {
+      "epoch": 4.053994246514716,
+      "grad_norm": 1.4144970178604126,
+      "learning_rate": 1.390106698462197e-05,
+      "loss": 0.3226,
+      "step": 20610
+    },
+    {
+      "epoch": 4.055961250030735,
+      "grad_norm": 1.0900161266326904,
+      "learning_rate": 1.3898009722094838e-05,
+      "loss": 0.283,
+      "step": 20620
+    },
+    {
+      "epoch": 4.057928253546753,
+      "grad_norm": 1.0960556268692017,
+      "learning_rate": 1.3894952459567704e-05,
+      "loss": 0.3846,
+      "step": 20630
+    },
+    {
+      "epoch": 4.059895257062772,
+      "grad_norm": 1.9161323308944702,
+      "learning_rate": 1.389189519704057e-05,
+      "loss": 0.5165,
+      "step": 20640
+    },
+    {
+      "epoch": 4.061862260578791,
+      "grad_norm": 0.9415732026100159,
+      "learning_rate": 1.3888837934513437e-05,
+      "loss": 0.4541,
+      "step": 20650
+    },
+    {
+      "epoch": 4.063829264094809,
+      "grad_norm": 1.3521498441696167,
+      "learning_rate": 1.3885780671986305e-05,
+      "loss": 0.3138,
+      "step": 20660
+    },
+    {
+      "epoch": 4.065796267610828,
+      "grad_norm": 1.2440180778503418,
+      "learning_rate": 1.388272340945917e-05,
+      "loss": 0.4493,
+      "step": 20670
+    },
+    {
+      "epoch": 4.067763271126847,
+      "grad_norm": 1.1224850416183472,
+      "learning_rate": 1.3879666146932038e-05,
+      "loss": 0.3012,
+      "step": 20680
+    },
+    {
+      "epoch": 4.069730274642866,
+      "grad_norm": 1.1821485757827759,
+      "learning_rate": 1.3876608884404905e-05,
+      "loss": 0.4457,
+      "step": 20690
+    },
+    {
+      "epoch": 4.071697278158885,
+      "grad_norm": 1.7971569299697876,
+      "learning_rate": 1.3873551621877773e-05,
+      "loss": 0.3576,
+      "step": 20700
+    },
+    {
+      "epoch": 4.073664281674904,
+      "grad_norm": 1.944004774093628,
+      "learning_rate": 1.3870494359350639e-05,
+      "loss": 0.4334,
+      "step": 20710
+    },
+    {
+      "epoch": 4.075631285190922,
+      "grad_norm": 0.9380494952201843,
+      "learning_rate": 1.3867437096823506e-05,
+      "loss": 0.3865,
+      "step": 20720
+    },
+    {
+      "epoch": 4.077598288706941,
+      "grad_norm": 0.9753469824790955,
+      "learning_rate": 1.3864379834296374e-05,
+      "loss": 0.3999,
+      "step": 20730
+    },
+    {
+      "epoch": 4.07956529222296,
+      "grad_norm": 1.8330798149108887,
+      "learning_rate": 1.3861322571769238e-05,
+      "loss": 0.3551,
+      "step": 20740
+    },
+    {
+      "epoch": 4.081532295738978,
+      "grad_norm": 1.0402212142944336,
+      "learning_rate": 1.3858265309242105e-05,
+      "loss": 0.3628,
+      "step": 20750
+    },
+    {
+      "epoch": 4.083499299254997,
+      "grad_norm": 0.5755205154418945,
+      "learning_rate": 1.3855208046714973e-05,
+      "loss": 0.4447,
+      "step": 20760
+    },
+    {
+      "epoch": 4.0854663027710165,
+      "grad_norm": 1.3143962621688843,
+      "learning_rate": 1.3852150784187838e-05,
+      "loss": 0.3869,
+      "step": 20770
+    },
+    {
+      "epoch": 4.087433306287035,
+      "grad_norm": 1.2326987981796265,
+      "learning_rate": 1.3849093521660706e-05,
+      "loss": 0.377,
+      "step": 20780
+    },
+    {
+      "epoch": 4.089400309803054,
+      "grad_norm": 2.1556975841522217,
+      "learning_rate": 1.3846036259133573e-05,
+      "loss": 0.4335,
+      "step": 20790
+    },
+    {
+      "epoch": 4.091367313319073,
+      "grad_norm": 1.4288907051086426,
+      "learning_rate": 1.3842978996606439e-05,
+      "loss": 0.243,
+      "step": 20800
+    },
+    {
+      "epoch": 4.093334316835091,
+      "grad_norm": 1.218528151512146,
+      "learning_rate": 1.3839921734079307e-05,
+      "loss": 0.4571,
+      "step": 20810
+    },
+    {
+      "epoch": 4.09530132035111,
+      "grad_norm": 0.7718478441238403,
+      "learning_rate": 1.3836864471552174e-05,
+      "loss": 0.3369,
+      "step": 20820
+    },
+    {
+      "epoch": 4.097268323867129,
+      "grad_norm": 1.171607494354248,
+      "learning_rate": 1.3833807209025041e-05,
+      "loss": 0.3596,
+      "step": 20830
+    },
+    {
+      "epoch": 4.099235327383147,
+      "grad_norm": 0.9774153232574463,
+      "learning_rate": 1.3830749946497907e-05,
+      "loss": 0.2873,
+      "step": 20840
+    },
+    {
+      "epoch": 4.1012023308991665,
+      "grad_norm": 2.146676540374756,
+      "learning_rate": 1.3827692683970775e-05,
+      "loss": 0.3206,
+      "step": 20850
+    },
+    {
+      "epoch": 4.103169334415186,
+      "grad_norm": 1.7797375917434692,
+      "learning_rate": 1.3824635421443639e-05,
+      "loss": 0.2919,
+      "step": 20860
+    },
+    {
+      "epoch": 4.105136337931204,
+      "grad_norm": 1.1288076639175415,
+      "learning_rate": 1.3821578158916506e-05,
+      "loss": 0.3807,
+      "step": 20870
+    },
+    {
+      "epoch": 4.107103341447223,
+      "grad_norm": 1.2551835775375366,
+      "learning_rate": 1.3818520896389374e-05,
+      "loss": 0.4071,
+      "step": 20880
+    },
+    {
+      "epoch": 4.109070344963242,
+      "grad_norm": 1.2019129991531372,
+      "learning_rate": 1.3815463633862241e-05,
+      "loss": 0.377,
+      "step": 20890
+    },
+    {
+      "epoch": 4.11103734847926,
+      "grad_norm": 2.886444091796875,
+      "learning_rate": 1.3812406371335107e-05,
+      "loss": 0.4303,
+      "step": 20900
+    },
+    {
+      "epoch": 4.113004351995279,
+      "grad_norm": 2.8083629608154297,
+      "learning_rate": 1.3809349108807974e-05,
+      "loss": 0.3066,
+      "step": 20910
+    },
+    {
+      "epoch": 4.114971355511298,
+      "grad_norm": 1.7566733360290527,
+      "learning_rate": 1.3806291846280842e-05,
+      "loss": 0.4121,
+      "step": 20920
+    },
+    {
+      "epoch": 4.1169383590273165,
+      "grad_norm": 1.5556622743606567,
+      "learning_rate": 1.3803234583753708e-05,
+      "loss": 0.3334,
+      "step": 20930
+    },
+    {
+      "epoch": 4.118905362543336,
+      "grad_norm": 0.984484851360321,
+      "learning_rate": 1.3800177321226575e-05,
+      "loss": 0.3508,
+      "step": 20940
+    },
+    {
+      "epoch": 4.120872366059355,
+      "grad_norm": 1.2625389099121094,
+      "learning_rate": 1.3797120058699443e-05,
+      "loss": 0.2859,
+      "step": 20950
+    },
+    {
+      "epoch": 4.122839369575373,
+      "grad_norm": 2.6469364166259766,
+      "learning_rate": 1.379406279617231e-05,
+      "loss": 0.331,
+      "step": 20960
+    },
+    {
+      "epoch": 4.124806373091392,
+      "grad_norm": 0.9096398949623108,
+      "learning_rate": 1.3791005533645174e-05,
+      "loss": 0.3237,
+      "step": 20970
+    },
+    {
+      "epoch": 4.126773376607411,
+      "grad_norm": 1.5150424242019653,
+      "learning_rate": 1.3787948271118042e-05,
+      "loss": 0.3276,
+      "step": 20980
+    },
+    {
+      "epoch": 4.128740380123429,
+      "grad_norm": 1.083422064781189,
+      "learning_rate": 1.3784891008590907e-05,
+      "loss": 0.3369,
+      "step": 20990
+    },
+    {
+      "epoch": 4.130707383639448,
+      "grad_norm": 1.2050446271896362,
+      "learning_rate": 1.3781833746063775e-05,
+      "loss": 0.2082,
+      "step": 21000
+    },
+    {
+      "epoch": 4.130707383639448,
+      "eval_loss": 0.15738588571548462,
+      "eval_runtime": 8.8899,
+      "eval_samples_per_second": 5.624,
+      "eval_steps_per_second": 2.812,
+      "step": 21000
+    },
+    {
+      "epoch": 4.132674387155467,
+      "grad_norm": 0.7613298892974854,
+      "learning_rate": 1.3778776483536642e-05,
+      "loss": 0.4471,
+      "step": 21010
+    },
+    {
+      "epoch": 4.134641390671486,
+      "grad_norm": 0.7388508915901184,
+      "learning_rate": 1.377571922100951e-05,
+      "loss": 0.4445,
+      "step": 21020
+    },
+    {
+      "epoch": 4.136608394187505,
+      "grad_norm": 1.4384886026382446,
+      "learning_rate": 1.3772661958482376e-05,
+      "loss": 0.3718,
+      "step": 21030
+    },
+    {
+      "epoch": 4.138575397703524,
+      "grad_norm": 1.3622539043426514,
+      "learning_rate": 1.3769604695955243e-05,
+      "loss": 0.3525,
+      "step": 21040
+    },
+    {
+      "epoch": 4.140542401219542,
+      "grad_norm": 1.0458835363388062,
+      "learning_rate": 1.376654743342811e-05,
+      "loss": 0.3866,
+      "step": 21050
+    },
+    {
+      "epoch": 4.142509404735561,
+      "grad_norm": 3.151491641998291,
+      "learning_rate": 1.3763490170900976e-05,
+      "loss": 0.3231,
+      "step": 21060
+    },
+    {
+      "epoch": 4.14447640825158,
+      "grad_norm": 1.1052993535995483,
+      "learning_rate": 1.3760432908373844e-05,
+      "loss": 0.3444,
+      "step": 21070
+    },
+    {
+      "epoch": 4.146443411767598,
+      "grad_norm": 0.982151210308075,
+      "learning_rate": 1.375737564584671e-05,
+      "loss": 0.3724,
+      "step": 21080
+    },
+    {
+      "epoch": 4.148410415283617,
+      "grad_norm": 1.6932227611541748,
+      "learning_rate": 1.3754318383319575e-05,
+      "loss": 0.4338,
+      "step": 21090
+    },
+    {
+      "epoch": 4.1503774187996365,
+      "grad_norm": 0.9574220180511475,
+      "learning_rate": 1.3751261120792443e-05,
+      "loss": 0.4565,
+      "step": 21100
+    },
+    {
+      "epoch": 4.152344422315655,
+      "grad_norm": 1.4933550357818604,
+      "learning_rate": 1.374820385826531e-05,
+      "loss": 0.4857,
+      "step": 21110
+    },
+    {
+      "epoch": 4.154311425831674,
+      "grad_norm": 3.230536699295044,
+      "learning_rate": 1.3745146595738176e-05,
+      "loss": 0.3114,
+      "step": 21120
+    },
+    {
+      "epoch": 4.156278429347693,
+      "grad_norm": 0.8398426175117493,
+      "learning_rate": 1.3742089333211044e-05,
+      "loss": 0.5186,
+      "step": 21130
+    },
+    {
+      "epoch": 4.158245432863711,
+      "grad_norm": 1.4111764430999756,
+      "learning_rate": 1.3739032070683911e-05,
+      "loss": 0.3243,
+      "step": 21140
+    },
+    {
+      "epoch": 4.16021243637973,
+      "grad_norm": 0.9417469501495361,
+      "learning_rate": 1.3735974808156779e-05,
+      "loss": 0.2852,
+      "step": 21150
+    },
+    {
+      "epoch": 4.162179439895749,
+      "grad_norm": 0.43298131227493286,
+      "learning_rate": 1.3732917545629644e-05,
+      "loss": 0.313,
+      "step": 21160
+    },
+    {
+      "epoch": 4.164146443411767,
+      "grad_norm": 1.1655223369598389,
+      "learning_rate": 1.3729860283102512e-05,
+      "loss": 0.422,
+      "step": 21170
+    },
+    {
+      "epoch": 4.1661134469277865,
+      "grad_norm": 3.525562047958374,
+      "learning_rate": 1.372680302057538e-05,
+      "loss": 0.3823,
+      "step": 21180
+    },
+    {
+      "epoch": 4.168080450443806,
+      "grad_norm": 1.0082656145095825,
+      "learning_rate": 1.3723745758048245e-05,
+      "loss": 0.3787,
+      "step": 21190
+    },
+    {
+      "epoch": 4.170047453959824,
+      "grad_norm": 1.1629912853240967,
+      "learning_rate": 1.372068849552111e-05,
+      "loss": 0.4133,
+      "step": 21200
+    },
+    {
+      "epoch": 4.172014457475843,
+      "grad_norm": 0.9328269958496094,
+      "learning_rate": 1.3717631232993978e-05,
+      "loss": 0.4058,
+      "step": 21210
+    },
+    {
+      "epoch": 4.173981460991861,
+      "grad_norm": 1.1281689405441284,
+      "learning_rate": 1.3714573970466844e-05,
+      "loss": 0.4844,
+      "step": 21220
+    },
+    {
+      "epoch": 4.17594846450788,
+      "grad_norm": 0.7831975221633911,
+      "learning_rate": 1.3711516707939712e-05,
+      "loss": 0.4359,
+      "step": 21230
+    },
+    {
+      "epoch": 4.177915468023899,
+      "grad_norm": 1.9914780855178833,
+      "learning_rate": 1.3708459445412579e-05,
+      "loss": 0.3736,
+      "step": 21240
+    },
+    {
+      "epoch": 4.179882471539917,
+      "grad_norm": 1.0189566612243652,
+      "learning_rate": 1.3705402182885445e-05,
+      "loss": 0.3837,
+      "step": 21250
+    },
+    {
+      "epoch": 4.1818494750559365,
+      "grad_norm": 1.1360441446304321,
+      "learning_rate": 1.3702344920358312e-05,
+      "loss": 0.3335,
+      "step": 21260
+    },
+    {
+      "epoch": 4.183816478571956,
+      "grad_norm": 1.208349585533142,
+      "learning_rate": 1.369928765783118e-05,
+      "loss": 0.387,
+      "step": 21270
+    },
+    {
+      "epoch": 4.185783482087974,
+      "grad_norm": 1.4240858554840088,
+      "learning_rate": 1.3696230395304047e-05,
+      "loss": 0.4941,
+      "step": 21280
+    },
+    {
+      "epoch": 4.187750485603993,
+      "grad_norm": 1.1946526765823364,
+      "learning_rate": 1.3693173132776913e-05,
+      "loss": 0.3432,
+      "step": 21290
+    },
+    {
+      "epoch": 4.189717489120012,
+      "grad_norm": 1.1512346267700195,
+      "learning_rate": 1.369011587024978e-05,
+      "loss": 0.3967,
+      "step": 21300
+    },
+    {
+      "epoch": 4.19168449263603,
+      "grad_norm": 1.373772144317627,
+      "learning_rate": 1.3687058607722645e-05,
+      "loss": 0.503,
+      "step": 21310
+    },
+    {
+      "epoch": 4.193651496152049,
+      "grad_norm": 1.0407246351242065,
+      "learning_rate": 1.3684001345195512e-05,
+      "loss": 0.4068,
+      "step": 21320
+    },
+    {
+      "epoch": 4.195618499668068,
+      "grad_norm": 1.2824262380599976,
+      "learning_rate": 1.368094408266838e-05,
+      "loss": 0.3464,
+      "step": 21330
+    },
+    {
+      "epoch": 4.1975855031840865,
+      "grad_norm": 1.0151809453964233,
+      "learning_rate": 1.3677886820141247e-05,
+      "loss": 0.4902,
+      "step": 21340
+    },
+    {
+      "epoch": 4.199552506700106,
+      "grad_norm": 0.9720301032066345,
+      "learning_rate": 1.3674829557614113e-05,
+      "loss": 0.3562,
+      "step": 21350
+    },
+    {
+      "epoch": 4.201519510216125,
+      "grad_norm": 1.015008807182312,
+      "learning_rate": 1.367177229508698e-05,
+      "loss": 0.4697,
+      "step": 21360
+    },
+    {
+      "epoch": 4.203486513732143,
+      "grad_norm": 1.029425859451294,
+      "learning_rate": 1.3668715032559848e-05,
+      "loss": 0.4514,
+      "step": 21370
+    },
+    {
+      "epoch": 4.205453517248162,
+      "grad_norm": 1.1204736232757568,
+      "learning_rate": 1.3665657770032713e-05,
+      "loss": 0.3916,
+      "step": 21380
+    },
+    {
+      "epoch": 4.207420520764181,
+      "grad_norm": 1.102508544921875,
+      "learning_rate": 1.3662600507505581e-05,
+      "loss": 0.4383,
+      "step": 21390
+    },
+    {
+      "epoch": 4.209387524280199,
+      "grad_norm": 0.8906152248382568,
+      "learning_rate": 1.3659543244978448e-05,
+      "loss": 0.4674,
+      "step": 21400
+    },
+    {
+      "epoch": 4.211354527796218,
+      "grad_norm": 1.9512590169906616,
+      "learning_rate": 1.3656485982451316e-05,
+      "loss": 0.3862,
+      "step": 21410
+    },
+    {
+      "epoch": 4.213321531312237,
+      "grad_norm": 1.4732189178466797,
+      "learning_rate": 1.365342871992418e-05,
+      "loss": 0.3456,
+      "step": 21420
+    },
+    {
+      "epoch": 4.215288534828256,
+      "grad_norm": 1.1794780492782593,
+      "learning_rate": 1.3650371457397047e-05,
+      "loss": 0.33,
+      "step": 21430
+    },
+    {
+      "epoch": 4.217255538344275,
+      "grad_norm": 2.1888999938964844,
+      "learning_rate": 1.3647314194869913e-05,
+      "loss": 0.3866,
+      "step": 21440
+    },
+    {
+      "epoch": 4.219222541860294,
+      "grad_norm": 1.7608472108840942,
+      "learning_rate": 1.364425693234278e-05,
+      "loss": 0.3101,
+      "step": 21450
+    },
+    {
+      "epoch": 4.221189545376312,
+      "grad_norm": 2.152985095977783,
+      "learning_rate": 1.3641199669815648e-05,
+      "loss": 0.4059,
+      "step": 21460
+    },
+    {
+      "epoch": 4.223156548892331,
+      "grad_norm": 1.4136911630630493,
+      "learning_rate": 1.3638142407288516e-05,
+      "loss": 0.4259,
+      "step": 21470
+    },
+    {
+      "epoch": 4.22512355240835,
+      "grad_norm": 1.411493182182312,
+      "learning_rate": 1.3635085144761381e-05,
+      "loss": 0.4125,
+      "step": 21480
+    },
+    {
+      "epoch": 4.227090555924368,
+      "grad_norm": 1.3379024267196655,
+      "learning_rate": 1.3632027882234249e-05,
+      "loss": 0.4389,
+      "step": 21490
+    },
+    {
+      "epoch": 4.229057559440387,
+      "grad_norm": 0.7551414966583252,
+      "learning_rate": 1.3628970619707116e-05,
+      "loss": 0.3998,
+      "step": 21500
+    },
+    {
+      "epoch": 4.229057559440387,
+      "eval_loss": 0.16503483057022095,
+      "eval_runtime": 8.8997,
+      "eval_samples_per_second": 5.618,
+      "eval_steps_per_second": 2.809,
+      "step": 21500
+    },
+    {
+      "epoch": 4.2310245629564065,
+      "grad_norm": 1.2745224237442017,
+      "learning_rate": 1.3625913357179982e-05,
+      "loss": 0.3906,
+      "step": 21510
+    },
+    {
+      "epoch": 4.232991566472425,
+      "grad_norm": 3.2287611961364746,
+      "learning_rate": 1.362285609465285e-05,
+      "loss": 0.4169,
+      "step": 21520
+    },
+    {
+      "epoch": 4.234958569988444,
+      "grad_norm": 1.4224720001220703,
+      "learning_rate": 1.3619798832125717e-05,
+      "loss": 0.2925,
+      "step": 21530
+    },
+    {
+      "epoch": 4.236925573504463,
+      "grad_norm": 2.0160717964172363,
+      "learning_rate": 1.3616741569598581e-05,
+      "loss": 0.3777,
+      "step": 21540
+    },
+    {
+      "epoch": 4.238892577020481,
+      "grad_norm": 0.9775928854942322,
+      "learning_rate": 1.3613684307071449e-05,
+      "loss": 0.3709,
+      "step": 21550
+    },
+    {
+      "epoch": 4.2408595805365,
+      "grad_norm": 1.825934886932373,
+      "learning_rate": 1.3610627044544316e-05,
+      "loss": 0.4786,
+      "step": 21560
+    },
+    {
+      "epoch": 4.242826584052519,
+      "grad_norm": 3.071707248687744,
+      "learning_rate": 1.3607569782017182e-05,
+      "loss": 0.2793,
+      "step": 21570
+    },
+    {
+      "epoch": 4.244793587568537,
+      "grad_norm": 1.1243879795074463,
+      "learning_rate": 1.360451251949005e-05,
+      "loss": 0.249,
+      "step": 21580
+    },
+    {
+      "epoch": 4.2467605910845565,
+      "grad_norm": 0.7699998021125793,
+      "learning_rate": 1.3601455256962917e-05,
+      "loss": 0.3444,
+      "step": 21590
+    },
+    {
+      "epoch": 4.248727594600576,
+      "grad_norm": 1.3177696466445923,
+      "learning_rate": 1.3598397994435784e-05,
+      "loss": 0.4581,
+      "step": 21600
+    },
+    {
+      "epoch": 4.250694598116594,
+      "grad_norm": 1.4589474201202393,
+      "learning_rate": 1.359534073190865e-05,
+      "loss": 0.5666,
+      "step": 21610
+    },
+    {
+      "epoch": 4.252661601632613,
+      "grad_norm": 0.6593105792999268,
+      "learning_rate": 1.3592283469381518e-05,
+      "loss": 0.4002,
+      "step": 21620
+    },
+    {
+      "epoch": 4.254628605148632,
+      "grad_norm": 2.0840675830841064,
+      "learning_rate": 1.3589226206854385e-05,
+      "loss": 0.4659,
+      "step": 21630
+    },
+    {
+      "epoch": 4.25659560866465,
+      "grad_norm": 1.280366063117981,
+      "learning_rate": 1.358616894432725e-05,
+      "loss": 0.3954,
+      "step": 21640
+    },
+    {
+      "epoch": 4.258562612180669,
+      "grad_norm": 1.2666908502578735,
+      "learning_rate": 1.3583111681800117e-05,
+      "loss": 0.4636,
+      "step": 21650
+    },
+    {
+      "epoch": 4.260529615696688,
+      "grad_norm": 1.0792936086654663,
+      "learning_rate": 1.3580054419272984e-05,
+      "loss": 0.4839,
+      "step": 21660
+    },
+    {
+      "epoch": 4.2624966192127065,
+      "grad_norm": 0.5781777501106262,
+      "learning_rate": 1.357699715674585e-05,
+      "loss": 0.3682,
+      "step": 21670
+    },
+    {
+      "epoch": 4.264463622728726,
+      "grad_norm": 2.4557440280914307,
+      "learning_rate": 1.3573939894218717e-05,
+      "loss": 0.3156,
+      "step": 21680
+    },
+    {
+      "epoch": 4.266430626244745,
+      "grad_norm": 2.1623849868774414,
+      "learning_rate": 1.3570882631691585e-05,
+      "loss": 0.3706,
+      "step": 21690
+    },
+    {
+      "epoch": 4.268397629760763,
+      "grad_norm": 1.0489895343780518,
+      "learning_rate": 1.356782536916445e-05,
+      "loss": 0.426,
+      "step": 21700
+    },
+    {
+      "epoch": 4.270364633276782,
+      "grad_norm": 0.9245631694793701,
+      "learning_rate": 1.3564768106637318e-05,
+      "loss": 0.3085,
+      "step": 21710
+    },
+    {
+      "epoch": 4.272331636792801,
+      "grad_norm": 2.06693696975708,
+      "learning_rate": 1.3561710844110185e-05,
+      "loss": 0.3848,
+      "step": 21720
+    },
+    {
+      "epoch": 4.274298640308819,
+      "grad_norm": 1.9542118310928345,
+      "learning_rate": 1.3558653581583053e-05,
+      "loss": 0.4136,
+      "step": 21730
+    },
+    {
+      "epoch": 4.276265643824838,
+      "grad_norm": 0.6859905123710632,
+      "learning_rate": 1.3555596319055919e-05,
+      "loss": 0.357,
+      "step": 21740
+    },
+    {
+      "epoch": 4.278232647340857,
+      "grad_norm": 1.3181674480438232,
+      "learning_rate": 1.3552539056528786e-05,
+      "loss": 0.3272,
+      "step": 21750
+    },
+    {
+      "epoch": 4.280199650856876,
+      "grad_norm": 2.021536350250244,
+      "learning_rate": 1.354948179400165e-05,
+      "loss": 0.424,
+      "step": 21760
+    },
+    {
+      "epoch": 4.282166654372895,
+      "grad_norm": 1.2987251281738281,
+      "learning_rate": 1.3546424531474518e-05,
+      "loss": 0.327,
+      "step": 21770
+    },
+    {
+      "epoch": 4.284133657888914,
+      "grad_norm": 1.4116606712341309,
+      "learning_rate": 1.3543367268947385e-05,
+      "loss": 0.3835,
+      "step": 21780
+    },
+    {
+      "epoch": 4.286100661404932,
+      "grad_norm": 1.1729151010513306,
+      "learning_rate": 1.3540310006420253e-05,
+      "loss": 0.3466,
+      "step": 21790
+    },
+    {
+      "epoch": 4.288067664920951,
+      "grad_norm": 1.3461573123931885,
+      "learning_rate": 1.3537252743893118e-05,
+      "loss": 0.3492,
+      "step": 21800
+    },
+    {
+      "epoch": 4.29003466843697,
+      "grad_norm": 1.5087522268295288,
+      "learning_rate": 1.3534195481365986e-05,
+      "loss": 0.3127,
+      "step": 21810
+    },
+    {
+      "epoch": 4.292001671952988,
+      "grad_norm": 0.7613654136657715,
+      "learning_rate": 1.3531138218838853e-05,
+      "loss": 0.4316,
+      "step": 21820
+    },
+    {
+      "epoch": 4.293968675469007,
+      "grad_norm": 2.289320230484009,
+      "learning_rate": 1.352808095631172e-05,
+      "loss": 0.3419,
+      "step": 21830
+    },
+    {
+      "epoch": 4.2959356789850265,
+      "grad_norm": 1.3401827812194824,
+      "learning_rate": 1.3525023693784587e-05,
+      "loss": 0.4404,
+      "step": 21840
+    },
+    {
+      "epoch": 4.297902682501045,
+      "grad_norm": 1.1575355529785156,
+      "learning_rate": 1.3521966431257454e-05,
+      "loss": 0.3049,
+      "step": 21850
+    },
+    {
+      "epoch": 4.299869686017064,
+      "grad_norm": 1.0929820537567139,
+      "learning_rate": 1.3518909168730322e-05,
+      "loss": 0.4202,
+      "step": 21860
+    },
+    {
+      "epoch": 4.301836689533083,
+      "grad_norm": 1.3887335062026978,
+      "learning_rate": 1.3515851906203187e-05,
+      "loss": 0.4188,
+      "step": 21870
+    },
+    {
+      "epoch": 4.303803693049101,
+      "grad_norm": 4.001399040222168,
+      "learning_rate": 1.3512794643676053e-05,
+      "loss": 0.3887,
+      "step": 21880
+    },
+    {
+      "epoch": 4.30577069656512,
+      "grad_norm": 2.886561155319214,
+      "learning_rate": 1.3509737381148919e-05,
+      "loss": 0.4775,
+      "step": 21890
+    },
+    {
+      "epoch": 4.307737700081139,
+      "grad_norm": 1.860848307609558,
+      "learning_rate": 1.3506680118621786e-05,
+      "loss": 0.4234,
+      "step": 21900
+    },
+    {
+      "epoch": 4.309704703597157,
+      "grad_norm": 3.108816146850586,
+      "learning_rate": 1.3503622856094654e-05,
+      "loss": 0.4309,
+      "step": 21910
+    },
+    {
+      "epoch": 4.3116717071131765,
+      "grad_norm": 2.461402654647827,
+      "learning_rate": 1.3500565593567521e-05,
+      "loss": 0.2674,
+      "step": 21920
+    },
+    {
+      "epoch": 4.313638710629196,
+      "grad_norm": 0.9968726634979248,
+      "learning_rate": 1.3497508331040387e-05,
+      "loss": 0.5311,
+      "step": 21930
+    },
+    {
+      "epoch": 4.315605714145214,
+      "grad_norm": 1.9840035438537598,
+      "learning_rate": 1.3494451068513255e-05,
+      "loss": 0.448,
+      "step": 21940
+    },
+    {
+      "epoch": 4.317572717661233,
+      "grad_norm": 1.257011890411377,
+      "learning_rate": 1.3491393805986122e-05,
+      "loss": 0.3892,
+      "step": 21950
+    },
+    {
+      "epoch": 4.319539721177252,
+      "grad_norm": 1.2027013301849365,
+      "learning_rate": 1.3488336543458988e-05,
+      "loss": 0.3691,
+      "step": 21960
+    },
+    {
+      "epoch": 4.32150672469327,
+      "grad_norm": 1.6953173875808716,
+      "learning_rate": 1.3485279280931855e-05,
+      "loss": 0.306,
+      "step": 21970
+    },
+    {
+      "epoch": 4.323473728209289,
+      "grad_norm": 1.4018330574035645,
+      "learning_rate": 1.3482222018404723e-05,
+      "loss": 0.4501,
+      "step": 21980
+    },
+    {
+      "epoch": 4.325440731725308,
+      "grad_norm": 1.5370357036590576,
+      "learning_rate": 1.3479164755877587e-05,
+      "loss": 0.3796,
+      "step": 21990
+    },
+    {
+      "epoch": 4.3274077352413265,
+      "grad_norm": 1.5908989906311035,
+      "learning_rate": 1.3476107493350454e-05,
+      "loss": 0.3807,
+      "step": 22000
+    },
+    {
+      "epoch": 4.3274077352413265,
+      "eval_loss": 0.16674765944480896,
+      "eval_runtime": 8.8966,
+      "eval_samples_per_second": 5.62,
+      "eval_steps_per_second": 2.81,
+      "step": 22000
+    },
+    {
+      "epoch": 4.329374738757346,
+      "grad_norm": 1.3100334405899048,
+      "learning_rate": 1.3473050230823322e-05,
+      "loss": 0.2926,
+      "step": 22010
+    },
+    {
+      "epoch": 4.331341742273365,
+      "grad_norm": 1.27474844455719,
+      "learning_rate": 1.3469992968296188e-05,
+      "loss": 0.3087,
+      "step": 22020
+    },
+    {
+      "epoch": 4.333308745789383,
+      "grad_norm": 2.377283811569214,
+      "learning_rate": 1.3466935705769055e-05,
+      "loss": 0.3764,
+      "step": 22030
+    },
+    {
+      "epoch": 4.335275749305402,
+      "grad_norm": 0.9264887571334839,
+      "learning_rate": 1.3463878443241923e-05,
+      "loss": 0.2962,
+      "step": 22040
+    },
+    {
+      "epoch": 4.337242752821421,
+      "grad_norm": 1.8624486923217773,
+      "learning_rate": 1.346082118071479e-05,
+      "loss": 0.3453,
+      "step": 22050
+    },
+    {
+      "epoch": 4.339209756337439,
+      "grad_norm": 2.4104392528533936,
+      "learning_rate": 1.3457763918187656e-05,
+      "loss": 0.3549,
+      "step": 22060
+    },
+    {
+      "epoch": 4.341176759853458,
+      "grad_norm": 1.0291612148284912,
+      "learning_rate": 1.3454706655660523e-05,
+      "loss": 0.3597,
+      "step": 22070
+    },
+    {
+      "epoch": 4.343143763369477,
+      "grad_norm": 1.2357598543167114,
+      "learning_rate": 1.345164939313339e-05,
+      "loss": 0.5817,
+      "step": 22080
+    },
+    {
+      "epoch": 4.345110766885496,
+      "grad_norm": 1.3580961227416992,
+      "learning_rate": 1.3448592130606257e-05,
+      "loss": 0.2892,
+      "step": 22090
+    },
+    {
+      "epoch": 4.347077770401515,
+      "grad_norm": 0.9541229605674744,
+      "learning_rate": 1.3445534868079122e-05,
+      "loss": 0.3649,
+      "step": 22100
+    },
+    {
+      "epoch": 4.349044773917534,
+      "grad_norm": 1.903846025466919,
+      "learning_rate": 1.344247760555199e-05,
+      "loss": 0.2888,
+      "step": 22110
+    },
+    {
+      "epoch": 4.351011777433552,
+      "grad_norm": 1.6936200857162476,
+      "learning_rate": 1.3439420343024856e-05,
+      "loss": 0.4391,
+      "step": 22120
+    },
+    {
+      "epoch": 4.352978780949571,
+      "grad_norm": 1.6819368600845337,
+      "learning_rate": 1.3436363080497723e-05,
+      "loss": 0.2995,
+      "step": 22130
+    },
+    {
+      "epoch": 4.35494578446559,
+      "grad_norm": 1.0277513265609741,
+      "learning_rate": 1.343330581797059e-05,
+      "loss": 0.4502,
+      "step": 22140
+    },
+    {
+      "epoch": 4.356912787981608,
+      "grad_norm": 1.3117585182189941,
+      "learning_rate": 1.3430248555443456e-05,
+      "loss": 0.4085,
+      "step": 22150
+    },
+    {
+      "epoch": 4.358879791497627,
+      "grad_norm": 2.2798779010772705,
+      "learning_rate": 1.3427191292916324e-05,
+      "loss": 0.2803,
+      "step": 22160
+    },
+    {
+      "epoch": 4.3608467950136465,
+      "grad_norm": 0.8277081251144409,
+      "learning_rate": 1.3424134030389191e-05,
+      "loss": 0.3755,
+      "step": 22170
+    },
+    {
+      "epoch": 4.362813798529665,
+      "grad_norm": 1.4714564085006714,
+      "learning_rate": 1.3421076767862059e-05,
+      "loss": 0.446,
+      "step": 22180
+    },
+    {
+      "epoch": 4.364780802045684,
+      "grad_norm": 1.0377657413482666,
+      "learning_rate": 1.3418019505334924e-05,
+      "loss": 0.4361,
+      "step": 22190
+    },
+    {
+      "epoch": 4.366747805561703,
+      "grad_norm": 1.56504487991333,
+      "learning_rate": 1.3414962242807792e-05,
+      "loss": 0.4736,
+      "step": 22200
+    },
+    {
+      "epoch": 4.368714809077721,
+      "grad_norm": 1.3587467670440674,
+      "learning_rate": 1.341190498028066e-05,
+      "loss": 0.4469,
+      "step": 22210
+    },
+    {
+      "epoch": 4.37068181259374,
+      "grad_norm": 2.5480990409851074,
+      "learning_rate": 1.3408847717753524e-05,
+      "loss": 0.395,
+      "step": 22220
+    },
+    {
+      "epoch": 4.372648816109759,
+      "grad_norm": 0.9612480401992798,
+      "learning_rate": 1.3405790455226391e-05,
+      "loss": 0.2845,
+      "step": 22230
+    },
+    {
+      "epoch": 4.374615819625777,
+      "grad_norm": 3.587682008743286,
+      "learning_rate": 1.3402733192699258e-05,
+      "loss": 0.4049,
+      "step": 22240
+    },
+    {
+      "epoch": 4.3765828231417965,
+      "grad_norm": 1.1484252214431763,
+      "learning_rate": 1.3399675930172124e-05,
+      "loss": 0.4146,
+      "step": 22250
+    },
+    {
+      "epoch": 4.378549826657816,
+      "grad_norm": 1.5784882307052612,
+      "learning_rate": 1.3396618667644992e-05,
+      "loss": 0.3146,
+      "step": 22260
+    },
+    {
+      "epoch": 4.380516830173834,
+      "grad_norm": 0.8098098039627075,
+      "learning_rate": 1.339356140511786e-05,
+      "loss": 0.3316,
+      "step": 22270
+    },
+    {
+      "epoch": 4.382483833689853,
+      "grad_norm": 1.1505993604660034,
+      "learning_rate": 1.3390504142590725e-05,
+      "loss": 0.3938,
+      "step": 22280
+    },
+    {
+      "epoch": 4.384450837205872,
+      "grad_norm": 1.157180905342102,
+      "learning_rate": 1.3387446880063592e-05,
+      "loss": 0.3029,
+      "step": 22290
+    },
+    {
+      "epoch": 4.38641784072189,
+      "grad_norm": 0.7741907835006714,
+      "learning_rate": 1.338438961753646e-05,
+      "loss": 0.4229,
+      "step": 22300
+    },
+    {
+      "epoch": 4.388384844237909,
+      "grad_norm": 2.2423791885375977,
+      "learning_rate": 1.3381332355009327e-05,
+      "loss": 0.3072,
+      "step": 22310
+    },
+    {
+      "epoch": 4.390351847753928,
+      "grad_norm": 1.6292169094085693,
+      "learning_rate": 1.3378275092482193e-05,
+      "loss": 0.4627,
+      "step": 22320
+    },
+    {
+      "epoch": 4.3923188512699465,
+      "grad_norm": 1.3712037801742554,
+      "learning_rate": 1.3375217829955059e-05,
+      "loss": 0.3798,
+      "step": 22330
+    },
+    {
+      "epoch": 4.394285854785966,
+      "grad_norm": 1.2356741428375244,
+      "learning_rate": 1.3372160567427925e-05,
+      "loss": 0.4468,
+      "step": 22340
+    },
+    {
+      "epoch": 4.396252858301985,
+      "grad_norm": 1.029561996459961,
+      "learning_rate": 1.3369103304900792e-05,
+      "loss": 0.3884,
+      "step": 22350
+    },
+    {
+      "epoch": 4.398219861818003,
+      "grad_norm": 1.0780653953552246,
+      "learning_rate": 1.336604604237366e-05,
+      "loss": 0.3785,
+      "step": 22360
+    },
+    {
+      "epoch": 4.400186865334022,
+      "grad_norm": 1.105497121810913,
+      "learning_rate": 1.3362988779846527e-05,
+      "loss": 0.373,
+      "step": 22370
+    },
+    {
+      "epoch": 4.402153868850041,
+      "grad_norm": 1.8972887992858887,
+      "learning_rate": 1.3359931517319393e-05,
+      "loss": 0.3213,
+      "step": 22380
+    },
+    {
+      "epoch": 4.404120872366059,
+      "grad_norm": 1.4248191118240356,
+      "learning_rate": 1.335687425479226e-05,
+      "loss": 0.2402,
+      "step": 22390
+    },
+    {
+      "epoch": 4.406087875882078,
+      "grad_norm": 2.0353078842163086,
+      "learning_rate": 1.3353816992265128e-05,
+      "loss": 0.331,
+      "step": 22400
+    },
+    {
+      "epoch": 4.408054879398097,
+      "grad_norm": 1.0217580795288086,
+      "learning_rate": 1.3350759729737994e-05,
+      "loss": 0.4146,
+      "step": 22410
+    },
+    {
+      "epoch": 4.410021882914116,
+      "grad_norm": 3.259878635406494,
+      "learning_rate": 1.3347702467210861e-05,
+      "loss": 0.435,
+      "step": 22420
+    },
+    {
+      "epoch": 4.411988886430135,
+      "grad_norm": 2.661207437515259,
+      "learning_rate": 1.3344645204683729e-05,
+      "loss": 0.3198,
+      "step": 22430
+    },
+    {
+      "epoch": 4.413955889946154,
+      "grad_norm": 1.1849759817123413,
+      "learning_rate": 1.3341587942156593e-05,
+      "loss": 0.5211,
+      "step": 22440
+    },
+    {
+      "epoch": 4.415922893462172,
+      "grad_norm": 1.2727833986282349,
+      "learning_rate": 1.333853067962946e-05,
+      "loss": 0.3356,
+      "step": 22450
+    },
+    {
+      "epoch": 4.417889896978191,
+      "grad_norm": 1.231123685836792,
+      "learning_rate": 1.3335473417102328e-05,
+      "loss": 0.367,
+      "step": 22460
+    },
+    {
+      "epoch": 4.41985690049421,
+      "grad_norm": 2.1626877784729004,
+      "learning_rate": 1.3332416154575193e-05,
+      "loss": 0.438,
+      "step": 22470
+    },
+    {
+      "epoch": 4.421823904010228,
+      "grad_norm": 1.1750102043151855,
+      "learning_rate": 1.332935889204806e-05,
+      "loss": 0.382,
+      "step": 22480
+    },
+    {
+      "epoch": 4.423790907526247,
+      "grad_norm": 2.559382438659668,
+      "learning_rate": 1.3326301629520928e-05,
+      "loss": 0.3909,
+      "step": 22490
+    },
+    {
+      "epoch": 4.4257579110422665,
+      "grad_norm": 1.5452768802642822,
+      "learning_rate": 1.3323244366993796e-05,
+      "loss": 0.4676,
+      "step": 22500
+    },
+    {
+      "epoch": 4.4257579110422665,
+      "eval_loss": 0.16149091720581055,
+      "eval_runtime": 8.8776,
+      "eval_samples_per_second": 5.632,
+      "eval_steps_per_second": 2.816,
+      "step": 22500
+    },
+    {
+      "epoch": 4.427724914558285,
+      "grad_norm": 1.9061115980148315,
+      "learning_rate": 1.3320187104466662e-05,
+      "loss": 0.4678,
+      "step": 22510
+    },
+    {
+      "epoch": 4.429691918074304,
+      "grad_norm": 2.9274566173553467,
+      "learning_rate": 1.3317129841939529e-05,
+      "loss": 0.3,
+      "step": 22520
+    },
+    {
+      "epoch": 4.431658921590322,
+      "grad_norm": 1.1326109170913696,
+      "learning_rate": 1.3314072579412397e-05,
+      "loss": 0.4364,
+      "step": 22530
+    },
+    {
+      "epoch": 4.433625925106341,
+      "grad_norm": 0.9025644659996033,
+      "learning_rate": 1.3311015316885262e-05,
+      "loss": 0.3303,
+      "step": 22540
+    },
+    {
+      "epoch": 4.43559292862236,
+      "grad_norm": 1.5909408330917358,
+      "learning_rate": 1.330795805435813e-05,
+      "loss": 0.3363,
+      "step": 22550
+    },
+    {
+      "epoch": 4.437559932138378,
+      "grad_norm": 0.7686471939086914,
+      "learning_rate": 1.3304900791830996e-05,
+      "loss": 0.4155,
+      "step": 22560
+    },
+    {
+      "epoch": 4.439526935654397,
+      "grad_norm": 0.9131650328636169,
+      "learning_rate": 1.3301843529303861e-05,
+      "loss": 0.483,
+      "step": 22570
+    },
+    {
+      "epoch": 4.4414939391704165,
+      "grad_norm": 0.9710274338722229,
+      "learning_rate": 1.3298786266776729e-05,
+      "loss": 0.3863,
+      "step": 22580
+    },
+    {
+      "epoch": 4.443460942686435,
+      "grad_norm": 2.079197645187378,
+      "learning_rate": 1.3295729004249596e-05,
+      "loss": 0.4701,
+      "step": 22590
+    },
+    {
+      "epoch": 4.445427946202454,
+      "grad_norm": 1.4030396938323975,
+      "learning_rate": 1.3292671741722462e-05,
+      "loss": 0.413,
+      "step": 22600
+    },
+    {
+      "epoch": 4.447394949718473,
+      "grad_norm": 1.82746160030365,
+      "learning_rate": 1.328961447919533e-05,
+      "loss": 0.5094,
+      "step": 22610
+    },
+    {
+      "epoch": 4.449361953234491,
+      "grad_norm": 1.0365689992904663,
+      "learning_rate": 1.3286557216668197e-05,
+      "loss": 0.3536,
+      "step": 22620
+    },
+    {
+      "epoch": 4.45132895675051,
+      "grad_norm": 1.287792682647705,
+      "learning_rate": 1.3283499954141064e-05,
+      "loss": 0.3274,
+      "step": 22630
+    },
+    {
+      "epoch": 4.453295960266529,
+      "grad_norm": 2.0417988300323486,
+      "learning_rate": 1.328044269161393e-05,
+      "loss": 0.432,
+      "step": 22640
+    },
+    {
+      "epoch": 4.455262963782547,
+      "grad_norm": 1.6449002027511597,
+      "learning_rate": 1.3277385429086798e-05,
+      "loss": 0.3048,
+      "step": 22650
+    },
+    {
+      "epoch": 4.4572299672985665,
+      "grad_norm": 1.3553321361541748,
+      "learning_rate": 1.3274328166559665e-05,
+      "loss": 0.4551,
+      "step": 22660
+    },
+    {
+      "epoch": 4.4591969708145855,
+      "grad_norm": 0.7964408993721008,
+      "learning_rate": 1.327127090403253e-05,
+      "loss": 0.4138,
+      "step": 22670
+    },
+    {
+      "epoch": 4.461163974330604,
+      "grad_norm": 2.022167682647705,
+      "learning_rate": 1.3268213641505397e-05,
+      "loss": 0.3209,
+      "step": 22680
+    },
+    {
+      "epoch": 4.463130977846623,
+      "grad_norm": 1.58821702003479,
+      "learning_rate": 1.3265156378978264e-05,
+      "loss": 0.4652,
+      "step": 22690
+    },
+    {
+      "epoch": 4.465097981362642,
+      "grad_norm": 2.6645796298980713,
+      "learning_rate": 1.326209911645113e-05,
+      "loss": 0.4101,
+      "step": 22700
+    },
+    {
+      "epoch": 4.46706498487866,
+      "grad_norm": 0.9394934177398682,
+      "learning_rate": 1.3259041853923997e-05,
+      "loss": 0.3394,
+      "step": 22710
+    },
+    {
+      "epoch": 4.469031988394679,
+      "grad_norm": 1.1719261407852173,
+      "learning_rate": 1.3255984591396865e-05,
+      "loss": 0.3804,
+      "step": 22720
+    },
+    {
+      "epoch": 4.470998991910698,
+      "grad_norm": 1.7606682777404785,
+      "learning_rate": 1.325292732886973e-05,
+      "loss": 0.4182,
+      "step": 22730
+    },
+    {
+      "epoch": 4.4729659954267165,
+      "grad_norm": 1.0615359544754028,
+      "learning_rate": 1.3249870066342598e-05,
+      "loss": 0.3121,
+      "step": 22740
+    },
+    {
+      "epoch": 4.4749329989427356,
+      "grad_norm": 1.0118271112442017,
+      "learning_rate": 1.3246812803815466e-05,
+      "loss": 0.4423,
+      "step": 22750
+    },
+    {
+      "epoch": 4.476900002458755,
+      "grad_norm": 1.467221736907959,
+      "learning_rate": 1.3243755541288333e-05,
+      "loss": 0.4035,
+      "step": 22760
+    },
+    {
+      "epoch": 4.478867005974773,
+      "grad_norm": 0.9635478854179382,
+      "learning_rate": 1.3240698278761199e-05,
+      "loss": 0.4424,
+      "step": 22770
+    },
+    {
+      "epoch": 4.480834009490792,
+      "grad_norm": 2.118645429611206,
+      "learning_rate": 1.3237641016234065e-05,
+      "loss": 0.467,
+      "step": 22780
+    },
+    {
+      "epoch": 4.482801013006811,
+      "grad_norm": 1.9423701763153076,
+      "learning_rate": 1.323458375370693e-05,
+      "loss": 0.3502,
+      "step": 22790
+    },
+    {
+      "epoch": 4.484768016522829,
+      "grad_norm": 2.8893489837646484,
+      "learning_rate": 1.3231526491179798e-05,
+      "loss": 0.376,
+      "step": 22800
+    },
+    {
+      "epoch": 4.486735020038848,
+      "grad_norm": 2.3034305572509766,
+      "learning_rate": 1.3228469228652665e-05,
+      "loss": 0.3,
+      "step": 22810
+    },
+    {
+      "epoch": 4.488702023554867,
+      "grad_norm": 2.976353168487549,
+      "learning_rate": 1.3225411966125533e-05,
+      "loss": 0.3781,
+      "step": 22820
+    },
+    {
+      "epoch": 4.4906690270708856,
+      "grad_norm": 1.0142730474472046,
+      "learning_rate": 1.3222354703598399e-05,
+      "loss": 0.4124,
+      "step": 22830
+    },
+    {
+      "epoch": 4.492636030586905,
+      "grad_norm": 2.81199312210083,
+      "learning_rate": 1.3219297441071266e-05,
+      "loss": 0.3022,
+      "step": 22840
+    },
+    {
+      "epoch": 4.494603034102924,
+      "grad_norm": 1.7177281379699707,
+      "learning_rate": 1.3216240178544134e-05,
+      "loss": 0.2978,
+      "step": 22850
+    },
+    {
+      "epoch": 4.496570037618942,
+      "grad_norm": 1.6933962106704712,
+      "learning_rate": 1.3213182916017e-05,
+      "loss": 0.4975,
+      "step": 22860
+    },
+    {
+      "epoch": 4.498537041134961,
+      "grad_norm": 1.574341893196106,
+      "learning_rate": 1.3210125653489867e-05,
+      "loss": 0.3842,
+      "step": 22870
+    },
+    {
+      "epoch": 4.50050404465098,
+      "grad_norm": 1.6971651315689087,
+      "learning_rate": 1.3207068390962734e-05,
+      "loss": 0.3666,
+      "step": 22880
+    },
+    {
+      "epoch": 4.502471048166998,
+      "grad_norm": 1.8739854097366333,
+      "learning_rate": 1.3204011128435598e-05,
+      "loss": 0.4354,
+      "step": 22890
+    },
+    {
+      "epoch": 4.504438051683017,
+      "grad_norm": 1.7573387622833252,
+      "learning_rate": 1.3200953865908466e-05,
+      "loss": 0.3319,
+      "step": 22900
+    },
+    {
+      "epoch": 4.5064050551990364,
+      "grad_norm": 1.734623670578003,
+      "learning_rate": 1.3197896603381333e-05,
+      "loss": 0.3508,
+      "step": 22910
+    },
+    {
+      "epoch": 4.508372058715055,
+      "grad_norm": 1.4408247470855713,
+      "learning_rate": 1.3194839340854199e-05,
+      "loss": 0.398,
+      "step": 22920
+    },
+    {
+      "epoch": 4.510339062231074,
+      "grad_norm": 1.2463972568511963,
+      "learning_rate": 1.3191782078327067e-05,
+      "loss": 0.4155,
+      "step": 22930
+    },
+    {
+      "epoch": 4.512306065747093,
+      "grad_norm": 1.2701466083526611,
+      "learning_rate": 1.3188724815799934e-05,
+      "loss": 0.3238,
+      "step": 22940
+    },
+    {
+      "epoch": 4.514273069263111,
+      "grad_norm": 2.2108230590820312,
+      "learning_rate": 1.3185667553272802e-05,
+      "loss": 0.3183,
+      "step": 22950
+    },
+    {
+      "epoch": 4.51624007277913,
+      "grad_norm": 1.100644588470459,
+      "learning_rate": 1.3182610290745667e-05,
+      "loss": 0.4386,
+      "step": 22960
+    },
+    {
+      "epoch": 4.518207076295149,
+      "grad_norm": 1.379320740699768,
+      "learning_rate": 1.3179553028218535e-05,
+      "loss": 0.3571,
+      "step": 22970
+    },
+    {
+      "epoch": 4.520174079811167,
+      "grad_norm": 0.7771784663200378,
+      "learning_rate": 1.3176495765691402e-05,
+      "loss": 0.5327,
+      "step": 22980
+    },
+    {
+      "epoch": 4.5221410833271865,
+      "grad_norm": 1.1977689266204834,
+      "learning_rate": 1.3173438503164268e-05,
+      "loss": 0.3398,
+      "step": 22990
+    },
+    {
+      "epoch": 4.5241080868432055,
+      "grad_norm": 0.870847225189209,
+      "learning_rate": 1.3170381240637135e-05,
+      "loss": 0.4439,
+      "step": 23000
+    },
+    {
+      "epoch": 4.5241080868432055,
+      "eval_loss": 0.17048148810863495,
+      "eval_runtime": 8.9021,
+      "eval_samples_per_second": 5.617,
+      "eval_steps_per_second": 2.808,
+      "step": 23000
+    },
+    {
+      "epoch": 4.526075090359224,
+      "grad_norm": 1.433722734451294,
+      "learning_rate": 1.316732397811e-05,
+      "loss": 0.4748,
+      "step": 23010
+    },
+    {
+      "epoch": 4.528042093875243,
+      "grad_norm": 1.5698235034942627,
+      "learning_rate": 1.3164266715582867e-05,
+      "loss": 0.3629,
+      "step": 23020
+    },
+    {
+      "epoch": 4.530009097391262,
+      "grad_norm": 1.342296838760376,
+      "learning_rate": 1.3161209453055735e-05,
+      "loss": 0.4292,
+      "step": 23030
+    },
+    {
+      "epoch": 4.53197610090728,
+      "grad_norm": 1.4127466678619385,
+      "learning_rate": 1.3158152190528602e-05,
+      "loss": 0.3382,
+      "step": 23040
+    },
+    {
+      "epoch": 4.533943104423299,
+      "grad_norm": 0.9064677357673645,
+      "learning_rate": 1.3155094928001468e-05,
+      "loss": 0.3593,
+      "step": 23050
+    },
+    {
+      "epoch": 4.535910107939318,
+      "grad_norm": 1.4167490005493164,
+      "learning_rate": 1.3152037665474335e-05,
+      "loss": 0.3732,
+      "step": 23060
+    },
+    {
+      "epoch": 4.5378771114553365,
+      "grad_norm": 1.2913470268249512,
+      "learning_rate": 1.3148980402947203e-05,
+      "loss": 0.3262,
+      "step": 23070
+    },
+    {
+      "epoch": 4.5398441149713555,
+      "grad_norm": 1.5516128540039062,
+      "learning_rate": 1.3145923140420068e-05,
+      "loss": 0.381,
+      "step": 23080
+    },
+    {
+      "epoch": 4.541811118487375,
+      "grad_norm": 1.0643260478973389,
+      "learning_rate": 1.3142865877892936e-05,
+      "loss": 0.5163,
+      "step": 23090
+    },
+    {
+      "epoch": 4.543778122003393,
+      "grad_norm": 1.4565191268920898,
+      "learning_rate": 1.3139808615365803e-05,
+      "loss": 0.3783,
+      "step": 23100
+    },
+    {
+      "epoch": 4.545745125519412,
+      "grad_norm": 1.2845790386199951,
+      "learning_rate": 1.3136751352838671e-05,
+      "loss": 0.2615,
+      "step": 23110
+    },
+    {
+      "epoch": 4.547712129035431,
+      "grad_norm": 1.2871668338775635,
+      "learning_rate": 1.3133694090311535e-05,
+      "loss": 0.3775,
+      "step": 23120
+    },
+    {
+      "epoch": 4.549679132551449,
+      "grad_norm": 1.8691515922546387,
+      "learning_rate": 1.3130636827784402e-05,
+      "loss": 0.3405,
+      "step": 23130
+    },
+    {
+      "epoch": 4.551646136067468,
+      "grad_norm": 0.7955536842346191,
+      "learning_rate": 1.3127579565257268e-05,
+      "loss": 0.3757,
+      "step": 23140
+    },
+    {
+      "epoch": 4.553613139583487,
+      "grad_norm": 1.201189398765564,
+      "learning_rate": 1.3124522302730136e-05,
+      "loss": 0.4091,
+      "step": 23150
+    },
+    {
+      "epoch": 4.5555801430995055,
+      "grad_norm": 2.295210361480713,
+      "learning_rate": 1.3121465040203003e-05,
+      "loss": 0.3536,
+      "step": 23160
+    },
+    {
+      "epoch": 4.557547146615525,
+      "grad_norm": 3.4921929836273193,
+      "learning_rate": 1.311840777767587e-05,
+      "loss": 0.3792,
+      "step": 23170
+    },
+    {
+      "epoch": 4.559514150131544,
+      "grad_norm": 1.5654789209365845,
+      "learning_rate": 1.3115350515148736e-05,
+      "loss": 0.407,
+      "step": 23180
+    },
+    {
+      "epoch": 4.561481153647562,
+      "grad_norm": 1.4418089389801025,
+      "learning_rate": 1.3112293252621604e-05,
+      "loss": 0.3842,
+      "step": 23190
+    },
+    {
+      "epoch": 4.563448157163581,
+      "grad_norm": 1.6266974210739136,
+      "learning_rate": 1.3109235990094471e-05,
+      "loss": 0.3039,
+      "step": 23200
+    },
+    {
+      "epoch": 4.5654151606796,
+      "grad_norm": 2.377856731414795,
+      "learning_rate": 1.3106178727567337e-05,
+      "loss": 0.4083,
+      "step": 23210
+    },
+    {
+      "epoch": 4.567382164195618,
+      "grad_norm": 0.9909720420837402,
+      "learning_rate": 1.3103121465040205e-05,
+      "loss": 0.3155,
+      "step": 23220
+    },
+    {
+      "epoch": 4.569349167711637,
+      "grad_norm": 1.2801272869110107,
+      "learning_rate": 1.310006420251307e-05,
+      "loss": 0.4249,
+      "step": 23230
+    },
+    {
+      "epoch": 4.571316171227656,
+      "grad_norm": 1.7023869752883911,
+      "learning_rate": 1.3097006939985936e-05,
+      "loss": 0.2412,
+      "step": 23240
+    },
+    {
+      "epoch": 4.573283174743675,
+      "grad_norm": 1.0783114433288574,
+      "learning_rate": 1.3093949677458804e-05,
+      "loss": 0.3342,
+      "step": 23250
+    },
+    {
+      "epoch": 4.575250178259694,
+      "grad_norm": 2.0209755897521973,
+      "learning_rate": 1.3090892414931671e-05,
+      "loss": 0.3702,
+      "step": 23260
+    },
+    {
+      "epoch": 4.577217181775713,
+      "grad_norm": 0.7944973111152649,
+      "learning_rate": 1.3087835152404537e-05,
+      "loss": 0.3751,
+      "step": 23270
+    },
+    {
+      "epoch": 4.579184185291731,
+      "grad_norm": 1.268553614616394,
+      "learning_rate": 1.3084777889877404e-05,
+      "loss": 0.3624,
+      "step": 23280
+    },
+    {
+      "epoch": 4.58115118880775,
+      "grad_norm": 0.9479203224182129,
+      "learning_rate": 1.3081720627350272e-05,
+      "loss": 0.3788,
+      "step": 23290
+    },
+    {
+      "epoch": 4.583118192323769,
+      "grad_norm": 1.9021347761154175,
+      "learning_rate": 1.307866336482314e-05,
+      "loss": 0.4087,
+      "step": 23300
+    },
+    {
+      "epoch": 4.585085195839787,
+      "grad_norm": 0.9418231844902039,
+      "learning_rate": 1.3075606102296005e-05,
+      "loss": 0.2752,
+      "step": 23310
+    },
+    {
+      "epoch": 4.587052199355806,
+      "grad_norm": 1.3885140419006348,
+      "learning_rate": 1.3072548839768873e-05,
+      "loss": 0.3861,
+      "step": 23320
+    },
+    {
+      "epoch": 4.5890192028718255,
+      "grad_norm": 1.2173035144805908,
+      "learning_rate": 1.306949157724174e-05,
+      "loss": 0.5022,
+      "step": 23330
+    },
+    {
+      "epoch": 4.590986206387844,
+      "grad_norm": 0.7699891328811646,
+      "learning_rate": 1.3066434314714606e-05,
+      "loss": 0.3706,
+      "step": 23340
+    },
+    {
+      "epoch": 4.592953209903863,
+      "grad_norm": 1.2280246019363403,
+      "learning_rate": 1.3063377052187472e-05,
+      "loss": 0.4889,
+      "step": 23350
+    },
+    {
+      "epoch": 4.594920213419881,
+      "grad_norm": 3.758246660232544,
+      "learning_rate": 1.3060319789660339e-05,
+      "loss": 0.3761,
+      "step": 23360
+    },
+    {
+      "epoch": 4.5968872169359,
+      "grad_norm": 1.604286551475525,
+      "learning_rate": 1.3057262527133205e-05,
+      "loss": 0.3536,
+      "step": 23370
+    },
+    {
+      "epoch": 4.598854220451919,
+      "grad_norm": 1.3617031574249268,
+      "learning_rate": 1.3054205264606072e-05,
+      "loss": 0.3312,
+      "step": 23380
+    },
+    {
+      "epoch": 4.600821223967937,
+      "grad_norm": 1.016905665397644,
+      "learning_rate": 1.305114800207894e-05,
+      "loss": 0.3307,
+      "step": 23390
+    },
+    {
+      "epoch": 4.602788227483956,
+      "grad_norm": 0.9720826148986816,
+      "learning_rate": 1.3048090739551806e-05,
+      "loss": 0.4432,
+      "step": 23400
+    },
+    {
+      "epoch": 4.6047552309999755,
+      "grad_norm": 1.73250150680542,
+      "learning_rate": 1.3045033477024673e-05,
+      "loss": 0.3725,
+      "step": 23410
+    },
+    {
+      "epoch": 4.606722234515994,
+      "grad_norm": 1.7776602506637573,
+      "learning_rate": 1.304197621449754e-05,
+      "loss": 0.4515,
+      "step": 23420
+    },
+    {
+      "epoch": 4.608689238032013,
+      "grad_norm": 1.3988317251205444,
+      "learning_rate": 1.3038918951970408e-05,
+      "loss": 0.357,
+      "step": 23430
+    },
+    {
+      "epoch": 4.610656241548032,
+      "grad_norm": 1.8836336135864258,
+      "learning_rate": 1.3035861689443274e-05,
+      "loss": 0.3908,
+      "step": 23440
+    },
+    {
+      "epoch": 4.61262324506405,
+      "grad_norm": 4.226807117462158,
+      "learning_rate": 1.3032804426916141e-05,
+      "loss": 0.3504,
+      "step": 23450
+    },
+    {
+      "epoch": 4.614590248580069,
+      "grad_norm": 1.609248399734497,
+      "learning_rate": 1.3029747164389005e-05,
+      "loss": 0.4287,
+      "step": 23460
+    },
+    {
+      "epoch": 4.616557252096088,
+      "grad_norm": 1.37553071975708,
+      "learning_rate": 1.3026689901861873e-05,
+      "loss": 0.3076,
+      "step": 23470
+    },
+    {
+      "epoch": 4.618524255612106,
+      "grad_norm": 1.6155554056167603,
+      "learning_rate": 1.302363263933474e-05,
+      "loss": 0.398,
+      "step": 23480
+    },
+    {
+      "epoch": 4.6204912591281255,
+      "grad_norm": 1.0259311199188232,
+      "learning_rate": 1.3020575376807608e-05,
+      "loss": 0.4076,
+      "step": 23490
+    },
+    {
+      "epoch": 4.622458262644145,
+      "grad_norm": 1.4233862161636353,
+      "learning_rate": 1.3017518114280474e-05,
+      "loss": 0.2926,
+      "step": 23500
+    },
+    {
+      "epoch": 4.622458262644145,
+      "eval_loss": 0.15736329555511475,
+      "eval_runtime": 8.8744,
+      "eval_samples_per_second": 5.634,
+      "eval_steps_per_second": 2.817,
+      "step": 23500
+    },
+    {
+      "epoch": 4.624425266160163,
+      "grad_norm": 1.5899595022201538,
+      "learning_rate": 1.3014460851753341e-05,
+      "loss": 0.4433,
+      "step": 23510
+    },
+    {
+      "epoch": 4.626392269676182,
+      "grad_norm": 1.366363286972046,
+      "learning_rate": 1.3011403589226208e-05,
+      "loss": 0.3417,
+      "step": 23520
+    },
+    {
+      "epoch": 4.628359273192201,
+      "grad_norm": 1.814328908920288,
+      "learning_rate": 1.3008346326699074e-05,
+      "loss": 0.3231,
+      "step": 23530
+    },
+    {
+      "epoch": 4.630326276708219,
+      "grad_norm": 1.5949935913085938,
+      "learning_rate": 1.3005289064171942e-05,
+      "loss": 0.3993,
+      "step": 23540
+    },
+    {
+      "epoch": 4.632293280224238,
+      "grad_norm": 0.9953024387359619,
+      "learning_rate": 1.300223180164481e-05,
+      "loss": 0.2743,
+      "step": 23550
+    },
+    {
+      "epoch": 4.634260283740257,
+      "grad_norm": 1.2114768028259277,
+      "learning_rate": 1.2999174539117677e-05,
+      "loss": 0.5201,
+      "step": 23560
+    },
+    {
+      "epoch": 4.6362272872562755,
+      "grad_norm": 1.964851975440979,
+      "learning_rate": 1.299611727659054e-05,
+      "loss": 0.5065,
+      "step": 23570
+    },
+    {
+      "epoch": 4.638194290772295,
+      "grad_norm": 1.2670104503631592,
+      "learning_rate": 1.2993060014063408e-05,
+      "loss": 0.4855,
+      "step": 23580
+    },
+    {
+      "epoch": 4.640161294288314,
+      "grad_norm": 0.9536296129226685,
+      "learning_rate": 1.2990002751536274e-05,
+      "loss": 0.4132,
+      "step": 23590
+    },
+    {
+      "epoch": 4.642128297804332,
+      "grad_norm": 2.4617717266082764,
+      "learning_rate": 1.2986945489009141e-05,
+      "loss": 0.3878,
+      "step": 23600
+    },
+    {
+      "epoch": 4.644095301320351,
+      "grad_norm": 2.0903079509735107,
+      "learning_rate": 1.2983888226482009e-05,
+      "loss": 0.3278,
+      "step": 23610
+    },
+    {
+      "epoch": 4.64606230483637,
+      "grad_norm": 1.6212421655654907,
+      "learning_rate": 1.2980830963954876e-05,
+      "loss": 0.4473,
+      "step": 23620
+    },
+    {
+      "epoch": 4.648029308352388,
+      "grad_norm": 1.0318902730941772,
+      "learning_rate": 1.2977773701427742e-05,
+      "loss": 0.2512,
+      "step": 23630
+    },
+    {
+      "epoch": 4.649996311868407,
+      "grad_norm": 0.8654146194458008,
+      "learning_rate": 1.297471643890061e-05,
+      "loss": 0.4187,
+      "step": 23640
+    },
+    {
+      "epoch": 4.651963315384426,
+      "grad_norm": 1.1589908599853516,
+      "learning_rate": 1.2971659176373477e-05,
+      "loss": 0.2749,
+      "step": 23650
+    },
+    {
+      "epoch": 4.653930318900445,
+      "grad_norm": 0.9683002829551697,
+      "learning_rate": 1.2968601913846343e-05,
+      "loss": 0.3428,
+      "step": 23660
+    },
+    {
+      "epoch": 4.655897322416464,
+      "grad_norm": 2.9495432376861572,
+      "learning_rate": 1.296554465131921e-05,
+      "loss": 0.3887,
+      "step": 23670
+    },
+    {
+      "epoch": 4.657864325932483,
+      "grad_norm": 0.9485315084457397,
+      "learning_rate": 1.2962487388792078e-05,
+      "loss": 0.2573,
+      "step": 23680
+    },
+    {
+      "epoch": 4.659831329448501,
+      "grad_norm": 0.9480960965156555,
+      "learning_rate": 1.2959430126264942e-05,
+      "loss": 0.3191,
+      "step": 23690
+    },
+    {
+      "epoch": 4.66179833296452,
+      "grad_norm": 2.1239635944366455,
+      "learning_rate": 1.295637286373781e-05,
+      "loss": 0.4233,
+      "step": 23700
+    },
+    {
+      "epoch": 4.663765336480539,
+      "grad_norm": 1.388307809829712,
+      "learning_rate": 1.2953315601210677e-05,
+      "loss": 0.3491,
+      "step": 23710
+    },
+    {
+      "epoch": 4.665732339996557,
+      "grad_norm": 1.6451371908187866,
+      "learning_rate": 1.2950258338683543e-05,
+      "loss": 0.4707,
+      "step": 23720
+    },
+    {
+      "epoch": 4.667699343512576,
+      "grad_norm": 1.3180400133132935,
+      "learning_rate": 1.294720107615641e-05,
+      "loss": 0.2803,
+      "step": 23730
+    },
+    {
+      "epoch": 4.6696663470285955,
+      "grad_norm": 2.698408365249634,
+      "learning_rate": 1.2944143813629278e-05,
+      "loss": 0.3303,
+      "step": 23740
+    },
+    {
+      "epoch": 4.671633350544614,
+      "grad_norm": 1.5707173347473145,
+      "learning_rate": 1.2941086551102145e-05,
+      "loss": 0.3554,
+      "step": 23750
+    },
+    {
+      "epoch": 4.673600354060633,
+      "grad_norm": 2.129290819168091,
+      "learning_rate": 1.293802928857501e-05,
+      "loss": 0.4207,
+      "step": 23760
+    },
+    {
+      "epoch": 4.675567357576652,
+      "grad_norm": 1.163686752319336,
+      "learning_rate": 1.2934972026047878e-05,
+      "loss": 0.3788,
+      "step": 23770
+    },
+    {
+      "epoch": 4.67753436109267,
+      "grad_norm": 1.2933921813964844,
+      "learning_rate": 1.2931914763520746e-05,
+      "loss": 0.396,
+      "step": 23780
+    },
+    {
+      "epoch": 4.679501364608689,
+      "grad_norm": 0.8011611700057983,
+      "learning_rate": 1.2928857500993612e-05,
+      "loss": 0.3598,
+      "step": 23790
+    },
+    {
+      "epoch": 4.681468368124708,
+      "grad_norm": 0.9323627352714539,
+      "learning_rate": 1.2925800238466477e-05,
+      "loss": 0.3895,
+      "step": 23800
+    },
+    {
+      "epoch": 4.683435371640726,
+      "grad_norm": 1.582582950592041,
+      "learning_rate": 1.2922742975939345e-05,
+      "loss": 0.4277,
+      "step": 23810
+    },
+    {
+      "epoch": 4.6854023751567455,
+      "grad_norm": 1.743823528289795,
+      "learning_rate": 1.291968571341221e-05,
+      "loss": 0.3476,
+      "step": 23820
+    },
+    {
+      "epoch": 4.687369378672765,
+      "grad_norm": 1.2965720891952515,
+      "learning_rate": 1.2916628450885078e-05,
+      "loss": 0.36,
+      "step": 23830
+    },
+    {
+      "epoch": 4.689336382188783,
+      "grad_norm": 1.0615127086639404,
+      "learning_rate": 1.2913571188357946e-05,
+      "loss": 0.4398,
+      "step": 23840
+    },
+    {
+      "epoch": 4.691303385704802,
+      "grad_norm": 1.2573604583740234,
+      "learning_rate": 1.2910513925830811e-05,
+      "loss": 0.4662,
+      "step": 23850
+    },
+    {
+      "epoch": 4.693270389220821,
+      "grad_norm": 1.6875718832015991,
+      "learning_rate": 1.2907456663303679e-05,
+      "loss": 0.2507,
+      "step": 23860
+    },
+    {
+      "epoch": 4.695237392736839,
+      "grad_norm": 1.415879249572754,
+      "learning_rate": 1.2904399400776546e-05,
+      "loss": 0.233,
+      "step": 23870
+    },
+    {
+      "epoch": 4.697204396252858,
+      "grad_norm": 2.006418466567993,
+      "learning_rate": 1.2901342138249414e-05,
+      "loss": 0.3111,
+      "step": 23880
+    },
+    {
+      "epoch": 4.699171399768877,
+      "grad_norm": 1.177172064781189,
+      "learning_rate": 1.289828487572228e-05,
+      "loss": 0.4526,
+      "step": 23890
+    },
+    {
+      "epoch": 4.7011384032848955,
+      "grad_norm": 1.4981369972229004,
+      "learning_rate": 1.2895227613195147e-05,
+      "loss": 0.3032,
+      "step": 23900
+    },
+    {
+      "epoch": 4.703105406800915,
+      "grad_norm": 1.1772596836090088,
+      "learning_rate": 1.2892170350668011e-05,
+      "loss": 0.2929,
+      "step": 23910
+    },
+    {
+      "epoch": 4.705072410316934,
+      "grad_norm": 1.217176914215088,
+      "learning_rate": 1.2889113088140879e-05,
+      "loss": 0.3669,
+      "step": 23920
+    },
+    {
+      "epoch": 4.707039413832952,
+      "grad_norm": 1.7104442119598389,
+      "learning_rate": 1.2886055825613746e-05,
+      "loss": 0.4283,
+      "step": 23930
+    },
+    {
+      "epoch": 4.709006417348971,
+      "grad_norm": 1.0308455228805542,
+      "learning_rate": 1.2882998563086613e-05,
+      "loss": 0.3678,
+      "step": 23940
+    },
+    {
+      "epoch": 4.71097342086499,
+      "grad_norm": 1.3772929906845093,
+      "learning_rate": 1.287994130055948e-05,
+      "loss": 0.2768,
+      "step": 23950
+    },
+    {
+      "epoch": 4.712940424381008,
+      "grad_norm": 1.864748477935791,
+      "learning_rate": 1.2876884038032347e-05,
+      "loss": 0.446,
+      "step": 23960
+    },
+    {
+      "epoch": 4.714907427897027,
+      "grad_norm": 1.0032296180725098,
+      "learning_rate": 1.2873826775505214e-05,
+      "loss": 0.4478,
+      "step": 23970
+    },
+    {
+      "epoch": 4.716874431413046,
+      "grad_norm": 3.7188913822174072,
+      "learning_rate": 1.287076951297808e-05,
+      "loss": 0.3125,
+      "step": 23980
+    },
+    {
+      "epoch": 4.718841434929065,
+      "grad_norm": 0.7291481494903564,
+      "learning_rate": 1.2867712250450947e-05,
+      "loss": 0.4573,
+      "step": 23990
+    },
+    {
+      "epoch": 4.720808438445084,
+      "grad_norm": 3.503469944000244,
+      "learning_rate": 1.2864654987923815e-05,
+      "loss": 0.5222,
+      "step": 24000
+    },
+    {
+      "epoch": 4.720808438445084,
+      "eval_loss": 0.15961362421512604,
+      "eval_runtime": 8.865,
+      "eval_samples_per_second": 5.64,
+      "eval_steps_per_second": 2.82,
+      "step": 24000
+    },
+    {
+      "epoch": 4.722775441961103,
+      "grad_norm": 1.0029255151748657,
+      "learning_rate": 1.2861597725396682e-05,
+      "loss": 0.4331,
+      "step": 24010
+    },
+    {
+      "epoch": 4.724742445477121,
+      "grad_norm": 1.118600606918335,
+      "learning_rate": 1.2858540462869548e-05,
+      "loss": 0.4526,
+      "step": 24020
+    },
+    {
+      "epoch": 4.72670944899314,
+      "grad_norm": 1.1678026914596558,
+      "learning_rate": 1.2855483200342414e-05,
+      "loss": 0.2711,
+      "step": 24030
+    },
+    {
+      "epoch": 4.728676452509159,
+      "grad_norm": 1.6292004585266113,
+      "learning_rate": 1.285242593781528e-05,
+      "loss": 0.3166,
+      "step": 24040
+    },
+    {
+      "epoch": 4.730643456025177,
+      "grad_norm": 0.6910290122032166,
+      "learning_rate": 1.2849368675288147e-05,
+      "loss": 0.4412,
+      "step": 24050
+    },
+    {
+      "epoch": 4.732610459541196,
+      "grad_norm": 1.3265618085861206,
+      "learning_rate": 1.2846311412761015e-05,
+      "loss": 0.3109,
+      "step": 24060
+    },
+    {
+      "epoch": 4.7345774630572155,
+      "grad_norm": 1.3849608898162842,
+      "learning_rate": 1.2843254150233882e-05,
+      "loss": 0.368,
+      "step": 24070
+    },
+    {
+      "epoch": 4.736544466573234,
+      "grad_norm": 0.561913013458252,
+      "learning_rate": 1.2840196887706748e-05,
+      "loss": 0.3611,
+      "step": 24080
+    },
+    {
+      "epoch": 4.738511470089253,
+      "grad_norm": 1.4088836908340454,
+      "learning_rate": 1.2837139625179615e-05,
+      "loss": 0.4978,
+      "step": 24090
+    },
+    {
+      "epoch": 4.740478473605272,
+      "grad_norm": 1.5019394159317017,
+      "learning_rate": 1.2834082362652483e-05,
+      "loss": 0.2134,
+      "step": 24100
+    },
+    {
+      "epoch": 4.74244547712129,
+      "grad_norm": 1.1798714399337769,
+      "learning_rate": 1.2831025100125349e-05,
+      "loss": 0.529,
+      "step": 24110
+    },
+    {
+      "epoch": 4.744412480637309,
+      "grad_norm": 0.8506179451942444,
+      "learning_rate": 1.2827967837598216e-05,
+      "loss": 0.3883,
+      "step": 24120
+    },
+    {
+      "epoch": 4.746379484153328,
+      "grad_norm": 1.204187273979187,
+      "learning_rate": 1.2824910575071084e-05,
+      "loss": 0.493,
+      "step": 24130
+    },
+    {
+      "epoch": 4.748346487669346,
+      "grad_norm": 1.6845051050186157,
+      "learning_rate": 1.2821853312543948e-05,
+      "loss": 0.4189,
+      "step": 24140
+    },
+    {
+      "epoch": 4.7503134911853655,
+      "grad_norm": 2.0980777740478516,
+      "learning_rate": 1.2818796050016815e-05,
+      "loss": 0.4742,
+      "step": 24150
+    },
+    {
+      "epoch": 4.752280494701385,
+      "grad_norm": 2.7500083446502686,
+      "learning_rate": 1.2815738787489683e-05,
+      "loss": 0.3244,
+      "step": 24160
+    },
+    {
+      "epoch": 4.754247498217403,
+      "grad_norm": 1.2423878908157349,
+      "learning_rate": 1.2812681524962548e-05,
+      "loss": 0.3138,
+      "step": 24170
+    },
+    {
+      "epoch": 4.756214501733422,
+      "grad_norm": 1.9693214893341064,
+      "learning_rate": 1.2809624262435416e-05,
+      "loss": 0.4574,
+      "step": 24180
+    },
+    {
+      "epoch": 4.758181505249441,
+      "grad_norm": 1.148080587387085,
+      "learning_rate": 1.2806566999908283e-05,
+      "loss": 0.4755,
+      "step": 24190
+    },
+    {
+      "epoch": 4.760148508765459,
+      "grad_norm": 1.0471198558807373,
+      "learning_rate": 1.280350973738115e-05,
+      "loss": 0.3744,
+      "step": 24200
+    },
+    {
+      "epoch": 4.762115512281478,
+      "grad_norm": 4.0511698722839355,
+      "learning_rate": 1.2800452474854017e-05,
+      "loss": 0.3998,
+      "step": 24210
+    },
+    {
+      "epoch": 4.764082515797497,
+      "grad_norm": 0.9432234168052673,
+      "learning_rate": 1.2797395212326884e-05,
+      "loss": 0.2854,
+      "step": 24220
+    },
+    {
+      "epoch": 4.7660495193135155,
+      "grad_norm": 1.4072970151901245,
+      "learning_rate": 1.2794337949799752e-05,
+      "loss": 0.3818,
+      "step": 24230
+    },
+    {
+      "epoch": 4.768016522829535,
+      "grad_norm": 1.4117906093597412,
+      "learning_rate": 1.2791280687272617e-05,
+      "loss": 0.3295,
+      "step": 24240
+    },
+    {
+      "epoch": 4.769983526345554,
+      "grad_norm": 1.4891633987426758,
+      "learning_rate": 1.2788223424745483e-05,
+      "loss": 0.5198,
+      "step": 24250
+    },
+    {
+      "epoch": 4.771950529861572,
+      "grad_norm": 1.3312366008758545,
+      "learning_rate": 1.278516616221835e-05,
+      "loss": 0.3685,
+      "step": 24260
+    },
+    {
+      "epoch": 4.773917533377591,
+      "grad_norm": 1.0341721773147583,
+      "learning_rate": 1.2782108899691216e-05,
+      "loss": 0.3587,
+      "step": 24270
+    },
+    {
+      "epoch": 4.77588453689361,
+      "grad_norm": 1.457330346107483,
+      "learning_rate": 1.2779051637164084e-05,
+      "loss": 0.4167,
+      "step": 24280
+    },
+    {
+      "epoch": 4.777851540409628,
+      "grad_norm": 1.134305715560913,
+      "learning_rate": 1.2775994374636951e-05,
+      "loss": 0.2711,
+      "step": 24290
+    },
+    {
+      "epoch": 4.779818543925647,
+      "grad_norm": 1.0785402059555054,
+      "learning_rate": 1.2772937112109817e-05,
+      "loss": 0.4465,
+      "step": 24300
+    },
+    {
+      "epoch": 4.781785547441666,
+      "grad_norm": 1.2902367115020752,
+      "learning_rate": 1.2769879849582685e-05,
+      "loss": 0.4601,
+      "step": 24310
+    },
+    {
+      "epoch": 4.783752550957685,
+      "grad_norm": 1.6840447187423706,
+      "learning_rate": 1.2766822587055552e-05,
+      "loss": 0.3442,
+      "step": 24320
+    },
+    {
+      "epoch": 4.785719554473704,
+      "grad_norm": 1.1976608037948608,
+      "learning_rate": 1.276376532452842e-05,
+      "loss": 0.2902,
+      "step": 24330
+    },
+    {
+      "epoch": 4.787686557989723,
+      "grad_norm": 2.4353525638580322,
+      "learning_rate": 1.2760708062001285e-05,
+      "loss": 0.3511,
+      "step": 24340
+    },
+    {
+      "epoch": 4.789653561505741,
+      "grad_norm": 1.9608737230300903,
+      "learning_rate": 1.2757650799474153e-05,
+      "loss": 0.334,
+      "step": 24350
+    },
+    {
+      "epoch": 4.79162056502176,
+      "grad_norm": 1.517730951309204,
+      "learning_rate": 1.275459353694702e-05,
+      "loss": 0.246,
+      "step": 24360
+    },
+    {
+      "epoch": 4.793587568537779,
+      "grad_norm": 3.3855648040771484,
+      "learning_rate": 1.2751536274419884e-05,
+      "loss": 0.4411,
+      "step": 24370
+    },
+    {
+      "epoch": 4.795554572053797,
+      "grad_norm": 1.0907025337219238,
+      "learning_rate": 1.2748479011892752e-05,
+      "loss": 0.4161,
+      "step": 24380
+    },
+    {
+      "epoch": 4.797521575569816,
+      "grad_norm": 2.2040274143218994,
+      "learning_rate": 1.274542174936562e-05,
+      "loss": 0.4809,
+      "step": 24390
+    },
+    {
+      "epoch": 4.7994885790858355,
+      "grad_norm": 1.5226056575775146,
+      "learning_rate": 1.2742364486838485e-05,
+      "loss": 0.4644,
+      "step": 24400
+    },
+    {
+      "epoch": 4.801455582601854,
+      "grad_norm": 1.3238670825958252,
+      "learning_rate": 1.2739307224311352e-05,
+      "loss": 0.3859,
+      "step": 24410
+    },
+    {
+      "epoch": 4.803422586117873,
+      "grad_norm": 1.0836786031723022,
+      "learning_rate": 1.273624996178422e-05,
+      "loss": 0.4176,
+      "step": 24420
+    },
+    {
+      "epoch": 4.805389589633892,
+      "grad_norm": 1.9304059743881226,
+      "learning_rate": 1.2733192699257086e-05,
+      "loss": 0.3487,
+      "step": 24430
+    },
+    {
+      "epoch": 4.80735659314991,
+      "grad_norm": 3.5189664363861084,
+      "learning_rate": 1.2730135436729953e-05,
+      "loss": 0.4214,
+      "step": 24440
+    },
+    {
+      "epoch": 4.809323596665929,
+      "grad_norm": 1.5368452072143555,
+      "learning_rate": 1.272707817420282e-05,
+      "loss": 0.5255,
+      "step": 24450
+    },
+    {
+      "epoch": 4.811290600181948,
+      "grad_norm": 2.091585159301758,
+      "learning_rate": 1.2724020911675688e-05,
+      "loss": 0.331,
+      "step": 24460
+    },
+    {
+      "epoch": 4.813257603697966,
+      "grad_norm": 2.5375471115112305,
+      "learning_rate": 1.2720963649148554e-05,
+      "loss": 0.3701,
+      "step": 24470
+    },
+    {
+      "epoch": 4.8152246072139855,
+      "grad_norm": 2.3465840816497803,
+      "learning_rate": 1.271790638662142e-05,
+      "loss": 0.4327,
+      "step": 24480
+    },
+    {
+      "epoch": 4.817191610730005,
+      "grad_norm": 1.6373897790908813,
+      "learning_rate": 1.2714849124094285e-05,
+      "loss": 0.4704,
+      "step": 24490
+    },
+    {
+      "epoch": 4.819158614246023,
+      "grad_norm": 1.7311253547668457,
+      "learning_rate": 1.2711791861567153e-05,
+      "loss": 0.4387,
+      "step": 24500
+    },
+    {
+      "epoch": 4.819158614246023,
+      "eval_loss": 0.1739385724067688,
+      "eval_runtime": 8.8844,
+      "eval_samples_per_second": 5.628,
+      "eval_steps_per_second": 2.814,
+      "step": 24500
+    },
+    {
+      "epoch": 4.821125617762042,
+      "grad_norm": 1.9144916534423828,
+      "learning_rate": 1.270873459904002e-05,
+      "loss": 0.3141,
+      "step": 24510
+    },
+    {
+      "epoch": 4.823092621278061,
+      "grad_norm": 1.6824641227722168,
+      "learning_rate": 1.2705677336512888e-05,
+      "loss": 0.4319,
+      "step": 24520
+    },
+    {
+      "epoch": 4.825059624794079,
+      "grad_norm": 1.2311413288116455,
+      "learning_rate": 1.2702620073985754e-05,
+      "loss": 0.338,
+      "step": 24530
+    },
+    {
+      "epoch": 4.827026628310098,
+      "grad_norm": 1.8054691553115845,
+      "learning_rate": 1.2699562811458621e-05,
+      "loss": 0.4764,
+      "step": 24540
+    },
+    {
+      "epoch": 4.828993631826117,
+      "grad_norm": 2.1235504150390625,
+      "learning_rate": 1.2696505548931489e-05,
+      "loss": 0.3935,
+      "step": 24550
+    },
+    {
+      "epoch": 4.8309606353421355,
+      "grad_norm": 1.0977386236190796,
+      "learning_rate": 1.2693448286404354e-05,
+      "loss": 0.3601,
+      "step": 24560
+    },
+    {
+      "epoch": 4.832927638858155,
+      "grad_norm": 0.9977229833602905,
+      "learning_rate": 1.2690391023877222e-05,
+      "loss": 0.4251,
+      "step": 24570
+    },
+    {
+      "epoch": 4.834894642374174,
+      "grad_norm": 1.337618350982666,
+      "learning_rate": 1.268733376135009e-05,
+      "loss": 0.3121,
+      "step": 24580
+    },
+    {
+      "epoch": 4.836861645890192,
+      "grad_norm": 1.5812876224517822,
+      "learning_rate": 1.2684276498822953e-05,
+      "loss": 0.3465,
+      "step": 24590
+    },
+    {
+      "epoch": 4.838828649406211,
+      "grad_norm": 1.1964857578277588,
+      "learning_rate": 1.2681219236295821e-05,
+      "loss": 0.3805,
+      "step": 24600
+    },
+    {
+      "epoch": 4.84079565292223,
+      "grad_norm": 1.448060393333435,
+      "learning_rate": 1.2678161973768688e-05,
+      "loss": 0.3723,
+      "step": 24610
+    },
+    {
+      "epoch": 4.842762656438248,
+      "grad_norm": 1.366903305053711,
+      "learning_rate": 1.2675104711241554e-05,
+      "loss": 0.4319,
+      "step": 24620
+    },
+    {
+      "epoch": 4.844729659954267,
+      "grad_norm": 1.6274155378341675,
+      "learning_rate": 1.2672047448714422e-05,
+      "loss": 0.371,
+      "step": 24630
+    },
+    {
+      "epoch": 4.846696663470286,
+      "grad_norm": 2.308397054672241,
+      "learning_rate": 1.2668990186187289e-05,
+      "loss": 0.4486,
+      "step": 24640
+    },
+    {
+      "epoch": 4.848663666986305,
+      "grad_norm": 0.8979236483573914,
+      "learning_rate": 1.2665932923660157e-05,
+      "loss": 0.3777,
+      "step": 24650
+    },
+    {
+      "epoch": 4.850630670502324,
+      "grad_norm": 1.4962645769119263,
+      "learning_rate": 1.2662875661133022e-05,
+      "loss": 0.4427,
+      "step": 24660
+    },
+    {
+      "epoch": 4.852597674018343,
+      "grad_norm": 1.5937800407409668,
+      "learning_rate": 1.265981839860589e-05,
+      "loss": 0.4127,
+      "step": 24670
+    },
+    {
+      "epoch": 4.854564677534361,
+      "grad_norm": 0.9097000360488892,
+      "learning_rate": 1.2656761136078757e-05,
+      "loss": 0.4768,
+      "step": 24680
+    },
+    {
+      "epoch": 4.85653168105038,
+      "grad_norm": 1.4313491582870483,
+      "learning_rate": 1.2653703873551623e-05,
+      "loss": 0.5025,
+      "step": 24690
+    },
+    {
+      "epoch": 4.858498684566399,
+      "grad_norm": 1.1728724241256714,
+      "learning_rate": 1.265064661102449e-05,
+      "loss": 0.3964,
+      "step": 24700
+    },
+    {
+      "epoch": 4.860465688082417,
+      "grad_norm": 2.7858197689056396,
+      "learning_rate": 1.2647589348497356e-05,
+      "loss": 0.395,
+      "step": 24710
+    },
+    {
+      "epoch": 4.862432691598436,
+      "grad_norm": 0.7792349457740784,
+      "learning_rate": 1.2644532085970222e-05,
+      "loss": 0.4374,
+      "step": 24720
+    },
+    {
+      "epoch": 4.8643996951144555,
+      "grad_norm": 1.8455607891082764,
+      "learning_rate": 1.264147482344309e-05,
+      "loss": 0.4001,
+      "step": 24730
+    },
+    {
+      "epoch": 4.866366698630474,
+      "grad_norm": 2.4060239791870117,
+      "learning_rate": 1.2638417560915957e-05,
+      "loss": 0.4162,
+      "step": 24740
+    },
+    {
+      "epoch": 4.868333702146493,
+      "grad_norm": 1.1814616918563843,
+      "learning_rate": 1.2635360298388823e-05,
+      "loss": 0.5639,
+      "step": 24750
+    },
+    {
+      "epoch": 4.870300705662512,
+      "grad_norm": 1.329742670059204,
+      "learning_rate": 1.263230303586169e-05,
+      "loss": 0.5668,
+      "step": 24760
+    },
+    {
+      "epoch": 4.87226770917853,
+      "grad_norm": 1.1229251623153687,
+      "learning_rate": 1.2629245773334558e-05,
+      "loss": 0.5126,
+      "step": 24770
+    },
+    {
+      "epoch": 4.874234712694549,
+      "grad_norm": 1.2327066659927368,
+      "learning_rate": 1.2626188510807425e-05,
+      "loss": 0.3026,
+      "step": 24780
+    },
+    {
+      "epoch": 4.876201716210568,
+      "grad_norm": 1.2386301755905151,
+      "learning_rate": 1.2623131248280291e-05,
+      "loss": 0.3811,
+      "step": 24790
+    },
+    {
+      "epoch": 4.878168719726586,
+      "grad_norm": 1.8087528944015503,
+      "learning_rate": 1.2620073985753158e-05,
+      "loss": 0.3896,
+      "step": 24800
+    },
+    {
+      "epoch": 4.8801357232426055,
+      "grad_norm": 1.0072718858718872,
+      "learning_rate": 1.2617016723226026e-05,
+      "loss": 0.2802,
+      "step": 24810
+    },
+    {
+      "epoch": 4.8821027267586246,
+      "grad_norm": 1.5886342525482178,
+      "learning_rate": 1.261395946069889e-05,
+      "loss": 0.4388,
+      "step": 24820
+    },
+    {
+      "epoch": 4.884069730274643,
+      "grad_norm": 1.1532931327819824,
+      "learning_rate": 1.2610902198171757e-05,
+      "loss": 0.4262,
+      "step": 24830
+    },
+    {
+      "epoch": 4.886036733790662,
+      "grad_norm": 1.3830076456069946,
+      "learning_rate": 1.2607844935644625e-05,
+      "loss": 0.3209,
+      "step": 24840
+    },
+    {
+      "epoch": 4.888003737306681,
+      "grad_norm": 1.9003983736038208,
+      "learning_rate": 1.260478767311749e-05,
+      "loss": 0.3566,
+      "step": 24850
+    },
+    {
+      "epoch": 4.889970740822699,
+      "grad_norm": 1.8158513307571411,
+      "learning_rate": 1.2601730410590358e-05,
+      "loss": 0.3776,
+      "step": 24860
+    },
+    {
+      "epoch": 4.891937744338718,
+      "grad_norm": 1.6109884977340698,
+      "learning_rate": 1.2598673148063226e-05,
+      "loss": 0.3846,
+      "step": 24870
+    },
+    {
+      "epoch": 4.893904747854737,
+      "grad_norm": 1.3823764324188232,
+      "learning_rate": 1.2595615885536091e-05,
+      "loss": 0.3041,
+      "step": 24880
+    },
+    {
+      "epoch": 4.8958717513707555,
+      "grad_norm": 2.2418899536132812,
+      "learning_rate": 1.2592558623008959e-05,
+      "loss": 0.3792,
+      "step": 24890
+    },
+    {
+      "epoch": 4.897838754886775,
+      "grad_norm": 0.8362561464309692,
+      "learning_rate": 1.2589501360481826e-05,
+      "loss": 0.3726,
+      "step": 24900
+    },
+    {
+      "epoch": 4.899805758402794,
+      "grad_norm": 1.6469014883041382,
+      "learning_rate": 1.2586444097954694e-05,
+      "loss": 0.4788,
+      "step": 24910
+    },
+    {
+      "epoch": 4.901772761918812,
+      "grad_norm": 1.3245086669921875,
+      "learning_rate": 1.258338683542756e-05,
+      "loss": 0.4149,
+      "step": 24920
+    },
+    {
+      "epoch": 4.903739765434831,
+      "grad_norm": 0.9041739702224731,
+      "learning_rate": 1.2580329572900425e-05,
+      "loss": 0.3088,
+      "step": 24930
+    },
+    {
+      "epoch": 4.905706768950849,
+      "grad_norm": 1.0792338848114014,
+      "learning_rate": 1.2577272310373291e-05,
+      "loss": 0.3612,
+      "step": 24940
+    },
+    {
+      "epoch": 4.907673772466868,
+      "grad_norm": 2.269197940826416,
+      "learning_rate": 1.2574215047846159e-05,
+      "loss": 0.454,
+      "step": 24950
+    },
+    {
+      "epoch": 4.909640775982887,
+      "grad_norm": 1.8159810304641724,
+      "learning_rate": 1.2571157785319026e-05,
+      "loss": 0.3962,
+      "step": 24960
+    },
+    {
+      "epoch": 4.9116077794989055,
+      "grad_norm": 1.9534015655517578,
+      "learning_rate": 1.2568100522791894e-05,
+      "loss": 0.3694,
+      "step": 24970
+    },
+    {
+      "epoch": 4.913574783014925,
+      "grad_norm": 2.8406848907470703,
+      "learning_rate": 1.256504326026476e-05,
+      "loss": 0.3452,
+      "step": 24980
+    },
+    {
+      "epoch": 4.915541786530944,
+      "grad_norm": 3.0141844749450684,
+      "learning_rate": 1.2561985997737627e-05,
+      "loss": 0.4919,
+      "step": 24990
+    },
+    {
+      "epoch": 4.917508790046962,
+      "grad_norm": 0.9889708161354065,
+      "learning_rate": 1.2558928735210494e-05,
+      "loss": 0.487,
+      "step": 25000
+    },
+    {
+      "epoch": 4.917508790046962,
+      "eval_loss": 0.16260863840579987,
+      "eval_runtime": 8.8961,
+      "eval_samples_per_second": 5.62,
+      "eval_steps_per_second": 2.81,
+      "step": 25000
+    },
+    {
+      "epoch": 4.919475793562981,
+      "grad_norm": 0.8428569436073303,
+      "learning_rate": 1.255587147268336e-05,
+      "loss": 0.4327,
+      "step": 25010
+    },
+    {
+      "epoch": 4.921442797079,
+      "grad_norm": 1.1603949069976807,
+      "learning_rate": 1.2552814210156228e-05,
+      "loss": 0.3681,
+      "step": 25020
+    },
+    {
+      "epoch": 4.923409800595018,
+      "grad_norm": 1.6248499155044556,
+      "learning_rate": 1.2549756947629095e-05,
+      "loss": 0.331,
+      "step": 25030
+    },
+    {
+      "epoch": 4.925376804111037,
+      "grad_norm": 1.5447056293487549,
+      "learning_rate": 1.2546699685101963e-05,
+      "loss": 0.4105,
+      "step": 25040
+    },
+    {
+      "epoch": 4.927343807627056,
+      "grad_norm": 0.8895068168640137,
+      "learning_rate": 1.2543642422574827e-05,
+      "loss": 0.3633,
+      "step": 25050
+    },
+    {
+      "epoch": 4.929310811143075,
+      "grad_norm": 2.2197225093841553,
+      "learning_rate": 1.2540585160047694e-05,
+      "loss": 0.4538,
+      "step": 25060
+    },
+    {
+      "epoch": 4.931277814659094,
+      "grad_norm": 0.859494686126709,
+      "learning_rate": 1.253752789752056e-05,
+      "loss": 0.502,
+      "step": 25070
+    },
+    {
+      "epoch": 4.933244818175113,
+      "grad_norm": 1.645679235458374,
+      "learning_rate": 1.2534470634993427e-05,
+      "loss": 0.4424,
+      "step": 25080
+    },
+    {
+      "epoch": 4.935211821691131,
+      "grad_norm": 2.0506739616394043,
+      "learning_rate": 1.2531413372466295e-05,
+      "loss": 0.4194,
+      "step": 25090
+    },
+    {
+      "epoch": 4.93717882520715,
+      "grad_norm": 0.7906273603439331,
+      "learning_rate": 1.2528356109939162e-05,
+      "loss": 0.3206,
+      "step": 25100
+    },
+    {
+      "epoch": 4.939145828723169,
+      "grad_norm": 1.0917754173278809,
+      "learning_rate": 1.2525298847412028e-05,
+      "loss": 0.3554,
+      "step": 25110
+    },
+    {
+      "epoch": 4.941112832239187,
+      "grad_norm": 2.7056996822357178,
+      "learning_rate": 1.2522241584884896e-05,
+      "loss": 0.3399,
+      "step": 25120
+    },
+    {
+      "epoch": 4.943079835755206,
+      "grad_norm": 0.8214200735092163,
+      "learning_rate": 1.2519184322357763e-05,
+      "loss": 0.4405,
+      "step": 25130
+    },
+    {
+      "epoch": 4.9450468392712255,
+      "grad_norm": 2.3560619354248047,
+      "learning_rate": 1.2516127059830629e-05,
+      "loss": 0.3374,
+      "step": 25140
+    },
+    {
+      "epoch": 4.947013842787244,
+      "grad_norm": 0.8455098271369934,
+      "learning_rate": 1.2513069797303496e-05,
+      "loss": 0.4155,
+      "step": 25150
+    },
+    {
+      "epoch": 4.948980846303263,
+      "grad_norm": 2.703481912612915,
+      "learning_rate": 1.2510012534776362e-05,
+      "loss": 0.4248,
+      "step": 25160
+    },
+    {
+      "epoch": 4.950947849819282,
+      "grad_norm": 1.9255882501602173,
+      "learning_rate": 1.2506955272249228e-05,
+      "loss": 0.2887,
+      "step": 25170
+    },
+    {
+      "epoch": 4.9529148533353,
+      "grad_norm": 0.7499654293060303,
+      "learning_rate": 1.2503898009722095e-05,
+      "loss": 0.4598,
+      "step": 25180
+    },
+    {
+      "epoch": 4.954881856851319,
+      "grad_norm": 1.097519874572754,
+      "learning_rate": 1.2500840747194963e-05,
+      "loss": 0.3877,
+      "step": 25190
+    },
+    {
+      "epoch": 4.956848860367338,
+      "grad_norm": 1.5604249238967896,
+      "learning_rate": 1.2497783484667829e-05,
+      "loss": 0.4214,
+      "step": 25200
+    },
+    {
+      "epoch": 4.958815863883356,
+      "grad_norm": 1.6495370864868164,
+      "learning_rate": 1.2494726222140696e-05,
+      "loss": 0.3071,
+      "step": 25210
+    },
+    {
+      "epoch": 4.9607828673993755,
+      "grad_norm": 1.7267638444900513,
+      "learning_rate": 1.2491668959613563e-05,
+      "loss": 0.3833,
+      "step": 25220
+    },
+    {
+      "epoch": 4.9627498709153945,
+      "grad_norm": 0.581771731376648,
+      "learning_rate": 1.2488611697086431e-05,
+      "loss": 0.324,
+      "step": 25230
+    },
+    {
+      "epoch": 4.964716874431413,
+      "grad_norm": 1.5528723001480103,
+      "learning_rate": 1.2485554434559297e-05,
+      "loss": 0.3987,
+      "step": 25240
+    },
+    {
+      "epoch": 4.966683877947432,
+      "grad_norm": 1.1603542566299438,
+      "learning_rate": 1.2482497172032164e-05,
+      "loss": 0.2939,
+      "step": 25250
+    },
+    {
+      "epoch": 4.968650881463451,
+      "grad_norm": 0.8113834261894226,
+      "learning_rate": 1.2479439909505032e-05,
+      "loss": 0.3677,
+      "step": 25260
+    },
+    {
+      "epoch": 4.970617884979469,
+      "grad_norm": 1.3022969961166382,
+      "learning_rate": 1.2476382646977896e-05,
+      "loss": 0.4211,
+      "step": 25270
+    },
+    {
+      "epoch": 4.972584888495488,
+      "grad_norm": 1.1333060264587402,
+      "learning_rate": 1.2473325384450763e-05,
+      "loss": 0.3824,
+      "step": 25280
+    },
+    {
+      "epoch": 4.974551892011507,
+      "grad_norm": 1.6616308689117432,
+      "learning_rate": 1.247026812192363e-05,
+      "loss": 0.4234,
+      "step": 25290
+    },
+    {
+      "epoch": 4.9765188955275255,
+      "grad_norm": 1.6122316122055054,
+      "learning_rate": 1.2467210859396496e-05,
+      "loss": 0.5025,
+      "step": 25300
+    },
+    {
+      "epoch": 4.9784858990435445,
+      "grad_norm": 1.1408931016921997,
+      "learning_rate": 1.2464153596869364e-05,
+      "loss": 0.4242,
+      "step": 25310
+    },
+    {
+      "epoch": 4.980452902559564,
+      "grad_norm": 0.966662585735321,
+      "learning_rate": 1.2461096334342231e-05,
+      "loss": 0.4244,
+      "step": 25320
+    },
+    {
+      "epoch": 4.982419906075582,
+      "grad_norm": 1.6419168710708618,
+      "learning_rate": 1.2458039071815097e-05,
+      "loss": 0.3898,
+      "step": 25330
+    },
+    {
+      "epoch": 4.984386909591601,
+      "grad_norm": 1.3502774238586426,
+      "learning_rate": 1.2454981809287965e-05,
+      "loss": 0.3736,
+      "step": 25340
+    },
+    {
+      "epoch": 4.98635391310762,
+      "grad_norm": 1.3421791791915894,
+      "learning_rate": 1.2451924546760832e-05,
+      "loss": 0.4539,
+      "step": 25350
+    },
+    {
+      "epoch": 4.988320916623638,
+      "grad_norm": 0.8233484625816345,
+      "learning_rate": 1.24488672842337e-05,
+      "loss": 0.4242,
+      "step": 25360
+    },
+    {
+      "epoch": 4.990287920139657,
+      "grad_norm": 1.437248706817627,
+      "learning_rate": 1.2445810021706565e-05,
+      "loss": 0.3697,
+      "step": 25370
+    },
+    {
+      "epoch": 4.992254923655676,
+      "grad_norm": 0.8841282725334167,
+      "learning_rate": 1.2442752759179433e-05,
+      "loss": 0.4423,
+      "step": 25380
+    },
+    {
+      "epoch": 4.9942219271716946,
+      "grad_norm": 0.6750190258026123,
+      "learning_rate": 1.2439695496652297e-05,
+      "loss": 0.5867,
+      "step": 25390
+    },
+    {
+      "epoch": 4.996188930687714,
+      "grad_norm": 1.2948499917984009,
+      "learning_rate": 1.2436638234125164e-05,
+      "loss": 0.4803,
+      "step": 25400
+    },
+    {
+      "epoch": 4.998155934203733,
+      "grad_norm": 1.183539867401123,
+      "learning_rate": 1.2433580971598032e-05,
+      "loss": 0.4717,
+      "step": 25410
+    },
+    {
+      "epoch": 5.000122937719751,
+      "grad_norm": 1.8746291399002075,
+      "learning_rate": 1.24305237090709e-05,
+      "loss": 0.4304,
+      "step": 25420
+    },
+    {
+      "epoch": 5.00208994123577,
+      "grad_norm": 1.3940856456756592,
+      "learning_rate": 1.2427466446543765e-05,
+      "loss": 0.368,
+      "step": 25430
+    },
+    {
+      "epoch": 5.004056944751789,
+      "grad_norm": 1.1577160358428955,
+      "learning_rate": 1.2424409184016633e-05,
+      "loss": 0.2709,
+      "step": 25440
+    },
+    {
+      "epoch": 5.006023948267807,
+      "grad_norm": 0.8124995827674866,
+      "learning_rate": 1.24213519214895e-05,
+      "loss": 0.4429,
+      "step": 25450
+    },
+    {
+      "epoch": 5.007990951783826,
+      "grad_norm": 0.6613554358482361,
+      "learning_rate": 1.2418294658962366e-05,
+      "loss": 0.3882,
+      "step": 25460
+    },
+    {
+      "epoch": 5.0099579552998454,
+      "grad_norm": 1.7531118392944336,
+      "learning_rate": 1.2415237396435233e-05,
+      "loss": 0.3169,
+      "step": 25470
+    },
+    {
+      "epoch": 5.011924958815864,
+      "grad_norm": 1.7000898122787476,
+      "learning_rate": 1.24121801339081e-05,
+      "loss": 0.3254,
+      "step": 25480
+    },
+    {
+      "epoch": 5.013891962331883,
+      "grad_norm": 1.7405487298965454,
+      "learning_rate": 1.2409122871380968e-05,
+      "loss": 0.3976,
+      "step": 25490
+    },
+    {
+      "epoch": 5.015858965847902,
+      "grad_norm": 1.5716254711151123,
+      "learning_rate": 1.2406065608853832e-05,
+      "loss": 0.3814,
+      "step": 25500
+    },
+    {
+      "epoch": 5.015858965847902,
+      "eval_loss": 0.15875256061553955,
+      "eval_runtime": 8.886,
+      "eval_samples_per_second": 5.627,
+      "eval_steps_per_second": 2.813,
+      "step": 25500
+    },
+    {
+      "epoch": 5.01782596936392,
+      "grad_norm": 1.5281723737716675,
+      "learning_rate": 1.24030083463267e-05,
+      "loss": 0.3889,
+      "step": 25510
+    },
+    {
+      "epoch": 5.019792972879939,
+      "grad_norm": 0.9451437592506409,
+      "learning_rate": 1.2399951083799566e-05,
+      "loss": 0.2675,
+      "step": 25520
+    },
+    {
+      "epoch": 5.021759976395958,
+      "grad_norm": 1.0830464363098145,
+      "learning_rate": 1.2396893821272433e-05,
+      "loss": 0.4978,
+      "step": 25530
+    },
+    {
+      "epoch": 5.023726979911976,
+      "grad_norm": 1.3201954364776611,
+      "learning_rate": 1.23938365587453e-05,
+      "loss": 0.3073,
+      "step": 25540
+    },
+    {
+      "epoch": 5.0256939834279954,
+      "grad_norm": 1.1136363744735718,
+      "learning_rate": 1.2390779296218168e-05,
+      "loss": 0.3773,
+      "step": 25550
+    },
+    {
+      "epoch": 5.0276609869440145,
+      "grad_norm": 1.7179425954818726,
+      "learning_rate": 1.2387722033691034e-05,
+      "loss": 0.463,
+      "step": 25560
+    },
+    {
+      "epoch": 5.029627990460033,
+      "grad_norm": 1.7579281330108643,
+      "learning_rate": 1.2384664771163901e-05,
+      "loss": 0.2775,
+      "step": 25570
+    },
+    {
+      "epoch": 5.031594993976052,
+      "grad_norm": 1.5655035972595215,
+      "learning_rate": 1.2381607508636769e-05,
+      "loss": 0.4461,
+      "step": 25580
+    },
+    {
+      "epoch": 5.033561997492071,
+      "grad_norm": 1.0195614099502563,
+      "learning_rate": 1.2378550246109635e-05,
+      "loss": 0.4831,
+      "step": 25590
+    },
+    {
+      "epoch": 5.035529001008089,
+      "grad_norm": 1.006157398223877,
+      "learning_rate": 1.2375492983582502e-05,
+      "loss": 0.3355,
+      "step": 25600
+    },
+    {
+      "epoch": 5.037496004524108,
+      "grad_norm": 0.8941965103149414,
+      "learning_rate": 1.2372435721055368e-05,
+      "loss": 0.3134,
+      "step": 25610
+    },
+    {
+      "epoch": 5.039463008040127,
+      "grad_norm": 1.5068615674972534,
+      "learning_rate": 1.2369378458528234e-05,
+      "loss": 0.3168,
+      "step": 25620
+    },
+    {
+      "epoch": 5.0414300115561455,
+      "grad_norm": 1.5665245056152344,
+      "learning_rate": 1.2366321196001101e-05,
+      "loss": 0.438,
+      "step": 25630
+    },
+    {
+      "epoch": 5.0433970150721645,
+      "grad_norm": 1.9946467876434326,
+      "learning_rate": 1.2363263933473969e-05,
+      "loss": 0.313,
+      "step": 25640
+    },
+    {
+      "epoch": 5.045364018588184,
+      "grad_norm": 0.8916930556297302,
+      "learning_rate": 1.2360206670946834e-05,
+      "loss": 0.3134,
+      "step": 25650
+    },
+    {
+      "epoch": 5.047331022104202,
+      "grad_norm": 1.5052319765090942,
+      "learning_rate": 1.2357149408419702e-05,
+      "loss": 0.234,
+      "step": 25660
+    },
+    {
+      "epoch": 5.049298025620221,
+      "grad_norm": 1.262294888496399,
+      "learning_rate": 1.235409214589257e-05,
+      "loss": 0.4548,
+      "step": 25670
+    },
+    {
+      "epoch": 5.05126502913624,
+      "grad_norm": 1.7379635572433472,
+      "learning_rate": 1.2351034883365437e-05,
+      "loss": 0.3546,
+      "step": 25680
+    },
+    {
+      "epoch": 5.053232032652258,
+      "grad_norm": 0.9460011124610901,
+      "learning_rate": 1.2347977620838302e-05,
+      "loss": 0.3477,
+      "step": 25690
+    },
+    {
+      "epoch": 5.055199036168277,
+      "grad_norm": 2.176176071166992,
+      "learning_rate": 1.234492035831117e-05,
+      "loss": 0.3871,
+      "step": 25700
+    },
+    {
+      "epoch": 5.057166039684296,
+      "grad_norm": 1.6649911403656006,
+      "learning_rate": 1.2341863095784037e-05,
+      "loss": 0.472,
+      "step": 25710
+    },
+    {
+      "epoch": 5.0591330432003145,
+      "grad_norm": 1.012190580368042,
+      "learning_rate": 1.2338805833256903e-05,
+      "loss": 0.4348,
+      "step": 25720
+    },
+    {
+      "epoch": 5.061100046716334,
+      "grad_norm": 1.0883797407150269,
+      "learning_rate": 1.2335748570729769e-05,
+      "loss": 0.3849,
+      "step": 25730
+    },
+    {
+      "epoch": 5.063067050232353,
+      "grad_norm": 1.2795614004135132,
+      "learning_rate": 1.2332691308202636e-05,
+      "loss": 0.4584,
+      "step": 25740
+    },
+    {
+      "epoch": 5.065034053748371,
+      "grad_norm": 0.5763323903083801,
+      "learning_rate": 1.2329634045675502e-05,
+      "loss": 0.3416,
+      "step": 25750
+    },
+    {
+      "epoch": 5.06700105726439,
+      "grad_norm": 2.406430959701538,
+      "learning_rate": 1.232657678314837e-05,
+      "loss": 0.2789,
+      "step": 25760
+    },
+    {
+      "epoch": 5.068968060780409,
+      "grad_norm": 1.3768939971923828,
+      "learning_rate": 1.2323519520621237e-05,
+      "loss": 0.3298,
+      "step": 25770
+    },
+    {
+      "epoch": 5.070935064296427,
+      "grad_norm": 1.4049491882324219,
+      "learning_rate": 1.2320462258094103e-05,
+      "loss": 0.3892,
+      "step": 25780
+    },
+    {
+      "epoch": 5.072902067812446,
+      "grad_norm": 0.5275664329528809,
+      "learning_rate": 1.231740499556697e-05,
+      "loss": 0.3585,
+      "step": 25790
+    },
+    {
+      "epoch": 5.074869071328465,
+      "grad_norm": 3.447392702102661,
+      "learning_rate": 1.2314347733039838e-05,
+      "loss": 0.4339,
+      "step": 25800
+    },
+    {
+      "epoch": 5.076836074844484,
+      "grad_norm": 0.7431132793426514,
+      "learning_rate": 1.2311290470512705e-05,
+      "loss": 0.4617,
+      "step": 25810
+    },
+    {
+      "epoch": 5.078803078360503,
+      "grad_norm": 0.9460954666137695,
+      "learning_rate": 1.2308233207985571e-05,
+      "loss": 0.3782,
+      "step": 25820
+    },
+    {
+      "epoch": 5.080770081876521,
+      "grad_norm": 1.6559048891067505,
+      "learning_rate": 1.2305175945458439e-05,
+      "loss": 0.4792,
+      "step": 25830
+    },
+    {
+      "epoch": 5.08273708539254,
+      "grad_norm": 1.7823586463928223,
+      "learning_rate": 1.2302118682931303e-05,
+      "loss": 0.3103,
+      "step": 25840
+    },
+    {
+      "epoch": 5.084704088908559,
+      "grad_norm": 0.5632954835891724,
+      "learning_rate": 1.229906142040417e-05,
+      "loss": 0.2107,
+      "step": 25850
+    },
+    {
+      "epoch": 5.086671092424577,
+      "grad_norm": 1.380471110343933,
+      "learning_rate": 1.2296004157877038e-05,
+      "loss": 0.4044,
+      "step": 25860
+    },
+    {
+      "epoch": 5.088638095940596,
+      "grad_norm": 1.372969388961792,
+      "learning_rate": 1.2292946895349903e-05,
+      "loss": 0.3801,
+      "step": 25870
+    },
+    {
+      "epoch": 5.090605099456615,
+      "grad_norm": 1.3248234987258911,
+      "learning_rate": 1.2289889632822771e-05,
+      "loss": 0.4301,
+      "step": 25880
+    },
+    {
+      "epoch": 5.092572102972634,
+      "grad_norm": 1.0714397430419922,
+      "learning_rate": 1.2286832370295638e-05,
+      "loss": 0.4881,
+      "step": 25890
+    },
+    {
+      "epoch": 5.094539106488653,
+      "grad_norm": 0.6456104516983032,
+      "learning_rate": 1.2283775107768506e-05,
+      "loss": 0.4173,
+      "step": 25900
+    },
+    {
+      "epoch": 5.096506110004672,
+      "grad_norm": 1.3249002695083618,
+      "learning_rate": 1.2280717845241372e-05,
+      "loss": 0.283,
+      "step": 25910
+    },
+    {
+      "epoch": 5.09847311352069,
+      "grad_norm": 0.9373102188110352,
+      "learning_rate": 1.2277660582714239e-05,
+      "loss": 0.3163,
+      "step": 25920
+    },
+    {
+      "epoch": 5.100440117036709,
+      "grad_norm": 0.5609747171401978,
+      "learning_rate": 1.2274603320187107e-05,
+      "loss": 0.4604,
+      "step": 25930
+    },
+    {
+      "epoch": 5.102407120552728,
+      "grad_norm": 0.5619985461235046,
+      "learning_rate": 1.2271546057659974e-05,
+      "loss": 0.2962,
+      "step": 25940
+    },
+    {
+      "epoch": 5.104374124068746,
+      "grad_norm": 1.2047351598739624,
+      "learning_rate": 1.2268488795132838e-05,
+      "loss": 0.3264,
+      "step": 25950
+    },
+    {
+      "epoch": 5.106341127584765,
+      "grad_norm": 0.7514748573303223,
+      "learning_rate": 1.2265431532605706e-05,
+      "loss": 0.392,
+      "step": 25960
+    },
+    {
+      "epoch": 5.1083081311007845,
+      "grad_norm": 1.2202684879302979,
+      "learning_rate": 1.2262374270078571e-05,
+      "loss": 0.4654,
+      "step": 25970
+    },
+    {
+      "epoch": 5.110275134616803,
+      "grad_norm": 1.5096663236618042,
+      "learning_rate": 1.2259317007551439e-05,
+      "loss": 0.3987,
+      "step": 25980
+    },
+    {
+      "epoch": 5.112242138132822,
+      "grad_norm": 2.5822386741638184,
+      "learning_rate": 1.2256259745024306e-05,
+      "loss": 0.4966,
+      "step": 25990
+    },
+    {
+      "epoch": 5.114209141648841,
+      "grad_norm": 1.6036911010742188,
+      "learning_rate": 1.2253202482497172e-05,
+      "loss": 0.3473,
+      "step": 26000
+    },
+    {
+      "epoch": 5.114209141648841,
+      "eval_loss": 0.1639460325241089,
+      "eval_runtime": 8.8667,
+      "eval_samples_per_second": 5.639,
+      "eval_steps_per_second": 2.82,
+      "step": 26000
+    },
+    {
+      "epoch": 5.116176145164859,
+      "grad_norm": 5.09474515914917,
+      "learning_rate": 1.225014521997004e-05,
+      "loss": 0.3937,
+      "step": 26010
+    },
+    {
+      "epoch": 5.118143148680878,
+      "grad_norm": 0.8852512240409851,
+      "learning_rate": 1.2247087957442907e-05,
+      "loss": 0.3332,
+      "step": 26020
+    },
+    {
+      "epoch": 5.120110152196897,
+      "grad_norm": 1.108298420906067,
+      "learning_rate": 1.2244030694915774e-05,
+      "loss": 0.2213,
+      "step": 26030
+    },
+    {
+      "epoch": 5.122077155712915,
+      "grad_norm": 1.3808269500732422,
+      "learning_rate": 1.224097343238864e-05,
+      "loss": 0.3315,
+      "step": 26040
+    },
+    {
+      "epoch": 5.1240441592289345,
+      "grad_norm": 1.3293715715408325,
+      "learning_rate": 1.2237916169861508e-05,
+      "loss": 0.3907,
+      "step": 26050
+    },
+    {
+      "epoch": 5.126011162744954,
+      "grad_norm": 2.228424310684204,
+      "learning_rate": 1.2234858907334375e-05,
+      "loss": 0.3814,
+      "step": 26060
+    },
+    {
+      "epoch": 5.127978166260972,
+      "grad_norm": 0.8705452680587769,
+      "learning_rate": 1.223180164480724e-05,
+      "loss": 0.3779,
+      "step": 26070
+    },
+    {
+      "epoch": 5.129945169776991,
+      "grad_norm": 1.1205730438232422,
+      "learning_rate": 1.2228744382280107e-05,
+      "loss": 0.3116,
+      "step": 26080
+    },
+    {
+      "epoch": 5.13191217329301,
+      "grad_norm": 3.4011454582214355,
+      "learning_rate": 1.2225687119752974e-05,
+      "loss": 0.4678,
+      "step": 26090
+    },
+    {
+      "epoch": 5.133879176809028,
+      "grad_norm": 1.2198089361190796,
+      "learning_rate": 1.222262985722584e-05,
+      "loss": 0.5545,
+      "step": 26100
+    },
+    {
+      "epoch": 5.135846180325047,
+      "grad_norm": 1.411157488822937,
+      "learning_rate": 1.2219572594698707e-05,
+      "loss": 0.3571,
+      "step": 26110
+    },
+    {
+      "epoch": 5.137813183841066,
+      "grad_norm": 1.0183827877044678,
+      "learning_rate": 1.2216515332171575e-05,
+      "loss": 0.2842,
+      "step": 26120
+    },
+    {
+      "epoch": 5.1397801873570845,
+      "grad_norm": 0.849287211894989,
+      "learning_rate": 1.221345806964444e-05,
+      "loss": 0.3608,
+      "step": 26130
+    },
+    {
+      "epoch": 5.141747190873104,
+      "grad_norm": 1.7928563356399536,
+      "learning_rate": 1.2210400807117308e-05,
+      "loss": 0.3096,
+      "step": 26140
+    },
+    {
+      "epoch": 5.143714194389123,
+      "grad_norm": 0.9198914170265198,
+      "learning_rate": 1.2207343544590176e-05,
+      "loss": 0.2994,
+      "step": 26150
+    },
+    {
+      "epoch": 5.145681197905141,
+      "grad_norm": 1.2599544525146484,
+      "learning_rate": 1.2204286282063043e-05,
+      "loss": 0.2084,
+      "step": 26160
+    },
+    {
+      "epoch": 5.14764820142116,
+      "grad_norm": 0.6974878311157227,
+      "learning_rate": 1.2201229019535909e-05,
+      "loss": 0.4318,
+      "step": 26170
+    },
+    {
+      "epoch": 5.149615204937179,
+      "grad_norm": 1.316336750984192,
+      "learning_rate": 1.2198171757008775e-05,
+      "loss": 0.382,
+      "step": 26180
+    },
+    {
+      "epoch": 5.151582208453197,
+      "grad_norm": 1.2267037630081177,
+      "learning_rate": 1.219511449448164e-05,
+      "loss": 0.3681,
+      "step": 26190
+    },
+    {
+      "epoch": 5.153549211969216,
+      "grad_norm": 3.4800543785095215,
+      "learning_rate": 1.2192057231954508e-05,
+      "loss": 0.4206,
+      "step": 26200
+    },
+    {
+      "epoch": 5.155516215485235,
+      "grad_norm": 0.8750397562980652,
+      "learning_rate": 1.2188999969427375e-05,
+      "loss": 0.4165,
+      "step": 26210
+    },
+    {
+      "epoch": 5.157483219001254,
+      "grad_norm": 1.767290711402893,
+      "learning_rate": 1.2185942706900243e-05,
+      "loss": 0.3753,
+      "step": 26220
+    },
+    {
+      "epoch": 5.159450222517273,
+      "grad_norm": 1.1647064685821533,
+      "learning_rate": 1.2182885444373109e-05,
+      "loss": 0.3647,
+      "step": 26230
+    },
+    {
+      "epoch": 5.161417226033292,
+      "grad_norm": 1.799381971359253,
+      "learning_rate": 1.2179828181845976e-05,
+      "loss": 0.3751,
+      "step": 26240
+    },
+    {
+      "epoch": 5.16338422954931,
+      "grad_norm": 1.3957469463348389,
+      "learning_rate": 1.2176770919318844e-05,
+      "loss": 0.2916,
+      "step": 26250
+    },
+    {
+      "epoch": 5.165351233065329,
+      "grad_norm": 1.7819017171859741,
+      "learning_rate": 1.217371365679171e-05,
+      "loss": 0.2783,
+      "step": 26260
+    },
+    {
+      "epoch": 5.167318236581348,
+      "grad_norm": 1.2645076513290405,
+      "learning_rate": 1.2170656394264577e-05,
+      "loss": 0.4173,
+      "step": 26270
+    },
+    {
+      "epoch": 5.169285240097366,
+      "grad_norm": 1.433780550956726,
+      "learning_rate": 1.2167599131737444e-05,
+      "loss": 0.463,
+      "step": 26280
+    },
+    {
+      "epoch": 5.171252243613385,
+      "grad_norm": 1.155297040939331,
+      "learning_rate": 1.2164541869210308e-05,
+      "loss": 0.3507,
+      "step": 26290
+    },
+    {
+      "epoch": 5.1732192471294045,
+      "grad_norm": 2.542121648788452,
+      "learning_rate": 1.2161484606683176e-05,
+      "loss": 0.3733,
+      "step": 26300
+    },
+    {
+      "epoch": 5.175186250645423,
+      "grad_norm": 1.2050806283950806,
+      "learning_rate": 1.2158427344156043e-05,
+      "loss": 0.3418,
+      "step": 26310
+    },
+    {
+      "epoch": 5.177153254161442,
+      "grad_norm": 0.7434633374214172,
+      "learning_rate": 1.215537008162891e-05,
+      "loss": 0.3878,
+      "step": 26320
+    },
+    {
+      "epoch": 5.179120257677461,
+      "grad_norm": 1.3844777345657349,
+      "learning_rate": 1.2152312819101777e-05,
+      "loss": 0.3923,
+      "step": 26330
+    },
+    {
+      "epoch": 5.181087261193479,
+      "grad_norm": 1.2223122119903564,
+      "learning_rate": 1.2149255556574644e-05,
+      "loss": 0.3741,
+      "step": 26340
+    },
+    {
+      "epoch": 5.183054264709498,
+      "grad_norm": 1.156844139099121,
+      "learning_rate": 1.2146198294047512e-05,
+      "loss": 0.2896,
+      "step": 26350
+    },
+    {
+      "epoch": 5.185021268225517,
+      "grad_norm": 0.49281927943229675,
+      "learning_rate": 1.2143141031520377e-05,
+      "loss": 0.3394,
+      "step": 26360
+    },
+    {
+      "epoch": 5.186988271741535,
+      "grad_norm": 2.524413585662842,
+      "learning_rate": 1.2140083768993245e-05,
+      "loss": 0.4893,
+      "step": 26370
+    },
+    {
+      "epoch": 5.1889552752575545,
+      "grad_norm": 3.383070230484009,
+      "learning_rate": 1.2137026506466112e-05,
+      "loss": 0.3611,
+      "step": 26380
+    },
+    {
+      "epoch": 5.190922278773574,
+      "grad_norm": 1.3388442993164062,
+      "learning_rate": 1.2133969243938978e-05,
+      "loss": 0.3775,
+      "step": 26390
+    },
+    {
+      "epoch": 5.192889282289592,
+      "grad_norm": 1.3757628202438354,
+      "learning_rate": 1.2130911981411846e-05,
+      "loss": 0.3374,
+      "step": 26400
+    },
+    {
+      "epoch": 5.194856285805611,
+      "grad_norm": 1.150669813156128,
+      "learning_rate": 1.2127854718884711e-05,
+      "loss": 0.4746,
+      "step": 26410
+    },
+    {
+      "epoch": 5.19682328932163,
+      "grad_norm": 0.8864215612411499,
+      "learning_rate": 1.2124797456357577e-05,
+      "loss": 0.3137,
+      "step": 26420
+    },
+    {
+      "epoch": 5.198790292837648,
+      "grad_norm": 3.9217827320098877,
+      "learning_rate": 1.2121740193830445e-05,
+      "loss": 0.3472,
+      "step": 26430
+    },
+    {
+      "epoch": 5.200757296353667,
+      "grad_norm": 1.4964346885681152,
+      "learning_rate": 1.2118682931303312e-05,
+      "loss": 0.4938,
+      "step": 26440
+    },
+    {
+      "epoch": 5.202724299869686,
+      "grad_norm": 0.8897117972373962,
+      "learning_rate": 1.2115625668776178e-05,
+      "loss": 0.2581,
+      "step": 26450
+    },
+    {
+      "epoch": 5.2046913033857045,
+      "grad_norm": 1.115395426750183,
+      "learning_rate": 1.2112568406249045e-05,
+      "loss": 0.4467,
+      "step": 26460
+    },
+    {
+      "epoch": 5.206658306901724,
+      "grad_norm": 1.5817797183990479,
+      "learning_rate": 1.2109511143721913e-05,
+      "loss": 0.4972,
+      "step": 26470
+    },
+    {
+      "epoch": 5.208625310417743,
+      "grad_norm": 1.0178269147872925,
+      "learning_rate": 1.210645388119478e-05,
+      "loss": 0.4131,
+      "step": 26480
+    },
+    {
+      "epoch": 5.210592313933761,
+      "grad_norm": 2.237933874130249,
+      "learning_rate": 1.2103396618667646e-05,
+      "loss": 0.2859,
+      "step": 26490
+    },
+    {
+      "epoch": 5.21255931744978,
+      "grad_norm": 1.0409702062606812,
+      "learning_rate": 1.2100339356140513e-05,
+      "loss": 0.2925,
+      "step": 26500
+    },
+    {
+      "epoch": 5.21255931744978,
+      "eval_loss": 0.1575675904750824,
+      "eval_runtime": 8.8783,
+      "eval_samples_per_second": 5.632,
+      "eval_steps_per_second": 2.816,
+      "step": 26500
+    },
+    {
+      "epoch": 5.214526320965799,
+      "grad_norm": 1.6005821228027344,
+      "learning_rate": 1.2097282093613381e-05,
+      "loss": 0.3946,
+      "step": 26510
+    },
+    {
+      "epoch": 5.216493324481817,
+      "grad_norm": 1.4268238544464111,
+      "learning_rate": 1.2094224831086245e-05,
+      "loss": 0.3797,
+      "step": 26520
+    },
+    {
+      "epoch": 5.218460327997836,
+      "grad_norm": 1.2413865327835083,
+      "learning_rate": 1.2091167568559113e-05,
+      "loss": 0.319,
+      "step": 26530
+    },
+    {
+      "epoch": 5.220427331513855,
+      "grad_norm": 0.78890460729599,
+      "learning_rate": 1.208811030603198e-05,
+      "loss": 0.3216,
+      "step": 26540
+    },
+    {
+      "epoch": 5.222394335029874,
+      "grad_norm": 0.7617972493171692,
+      "learning_rate": 1.2085053043504846e-05,
+      "loss": 0.4423,
+      "step": 26550
+    },
+    {
+      "epoch": 5.224361338545893,
+      "grad_norm": 1.5445433855056763,
+      "learning_rate": 1.2081995780977713e-05,
+      "loss": 0.2114,
+      "step": 26560
+    },
+    {
+      "epoch": 5.226328342061912,
+      "grad_norm": 1.269335150718689,
+      "learning_rate": 1.207893851845058e-05,
+      "loss": 0.4142,
+      "step": 26570
+    },
+    {
+      "epoch": 5.22829534557793,
+      "grad_norm": 1.3220189809799194,
+      "learning_rate": 1.2075881255923446e-05,
+      "loss": 0.2565,
+      "step": 26580
+    },
+    {
+      "epoch": 5.230262349093949,
+      "grad_norm": 1.7140494585037231,
+      "learning_rate": 1.2072823993396314e-05,
+      "loss": 0.2602,
+      "step": 26590
+    },
+    {
+      "epoch": 5.232229352609968,
+      "grad_norm": 1.852797508239746,
+      "learning_rate": 1.2069766730869181e-05,
+      "loss": 0.24,
+      "step": 26600
+    },
+    {
+      "epoch": 5.234196356125986,
+      "grad_norm": 1.7020344734191895,
+      "learning_rate": 1.2066709468342049e-05,
+      "loss": 0.2881,
+      "step": 26610
+    },
+    {
+      "epoch": 5.236163359642005,
+      "grad_norm": 0.5673180818557739,
+      "learning_rate": 1.2063652205814915e-05,
+      "loss": 0.3332,
+      "step": 26620
+    },
+    {
+      "epoch": 5.2381303631580245,
+      "grad_norm": 0.687757670879364,
+      "learning_rate": 1.206059494328778e-05,
+      "loss": 0.3021,
+      "step": 26630
+    },
+    {
+      "epoch": 5.240097366674043,
+      "grad_norm": 1.2419289350509644,
+      "learning_rate": 1.2057537680760646e-05,
+      "loss": 0.3224,
+      "step": 26640
+    },
+    {
+      "epoch": 5.242064370190062,
+      "grad_norm": 1.7298426628112793,
+      "learning_rate": 1.2054480418233514e-05,
+      "loss": 0.3101,
+      "step": 26650
+    },
+    {
+      "epoch": 5.244031373706081,
+      "grad_norm": 3.0599703788757324,
+      "learning_rate": 1.2051423155706381e-05,
+      "loss": 0.2711,
+      "step": 26660
+    },
+    {
+      "epoch": 5.245998377222099,
+      "grad_norm": 0.3817830979824066,
+      "learning_rate": 1.2048365893179249e-05,
+      "loss": 0.4047,
+      "step": 26670
+    },
+    {
+      "epoch": 5.247965380738118,
+      "grad_norm": 1.0103836059570312,
+      "learning_rate": 1.2045308630652114e-05,
+      "loss": 0.2704,
+      "step": 26680
+    },
+    {
+      "epoch": 5.249932384254137,
+      "grad_norm": 1.6104187965393066,
+      "learning_rate": 1.2042251368124982e-05,
+      "loss": 0.2766,
+      "step": 26690
+    },
+    {
+      "epoch": 5.251899387770155,
+      "grad_norm": 0.8950489163398743,
+      "learning_rate": 1.203919410559785e-05,
+      "loss": 0.458,
+      "step": 26700
+    },
+    {
+      "epoch": 5.2538663912861745,
+      "grad_norm": 1.5272265672683716,
+      "learning_rate": 1.2036136843070715e-05,
+      "loss": 0.3857,
+      "step": 26710
+    },
+    {
+      "epoch": 5.255833394802194,
+      "grad_norm": 1.1579011678695679,
+      "learning_rate": 1.2033079580543583e-05,
+      "loss": 0.4661,
+      "step": 26720
+    },
+    {
+      "epoch": 5.257800398318212,
+      "grad_norm": 1.5543522834777832,
+      "learning_rate": 1.203002231801645e-05,
+      "loss": 0.3932,
+      "step": 26730
+    },
+    {
+      "epoch": 5.259767401834231,
+      "grad_norm": 1.385425090789795,
+      "learning_rate": 1.2026965055489318e-05,
+      "loss": 0.368,
+      "step": 26740
+    },
+    {
+      "epoch": 5.26173440535025,
+      "grad_norm": 0.7192743420600891,
+      "learning_rate": 1.2023907792962182e-05,
+      "loss": 0.3624,
+      "step": 26750
+    },
+    {
+      "epoch": 5.263701408866268,
+      "grad_norm": 0.9118638634681702,
+      "learning_rate": 1.2020850530435049e-05,
+      "loss": 0.4483,
+      "step": 26760
+    },
+    {
+      "epoch": 5.265668412382287,
+      "grad_norm": 1.5351526737213135,
+      "learning_rate": 1.2017793267907915e-05,
+      "loss": 0.3773,
+      "step": 26770
+    },
+    {
+      "epoch": 5.267635415898306,
+      "grad_norm": 1.1610013246536255,
+      "learning_rate": 1.2014736005380782e-05,
+      "loss": 0.4611,
+      "step": 26780
+    },
+    {
+      "epoch": 5.2696024194143245,
+      "grad_norm": 1.0738345384597778,
+      "learning_rate": 1.201167874285365e-05,
+      "loss": 0.3372,
+      "step": 26790
+    },
+    {
+      "epoch": 5.271569422930344,
+      "grad_norm": 1.3111422061920166,
+      "learning_rate": 1.2008621480326517e-05,
+      "loss": 0.3991,
+      "step": 26800
+    },
+    {
+      "epoch": 5.273536426446363,
+      "grad_norm": 1.3614267110824585,
+      "learning_rate": 1.2005564217799383e-05,
+      "loss": 0.4385,
+      "step": 26810
+    },
+    {
+      "epoch": 5.275503429962381,
+      "grad_norm": 1.3455390930175781,
+      "learning_rate": 1.200250695527225e-05,
+      "loss": 0.3435,
+      "step": 26820
+    },
+    {
+      "epoch": 5.2774704334784,
+      "grad_norm": 1.029335618019104,
+      "learning_rate": 1.1999449692745118e-05,
+      "loss": 0.3084,
+      "step": 26830
+    },
+    {
+      "epoch": 5.279437436994419,
+      "grad_norm": 1.7655671834945679,
+      "learning_rate": 1.1996392430217984e-05,
+      "loss": 0.3199,
+      "step": 26840
+    },
+    {
+      "epoch": 5.281404440510437,
+      "grad_norm": 1.5547866821289062,
+      "learning_rate": 1.1993335167690851e-05,
+      "loss": 0.3297,
+      "step": 26850
+    },
+    {
+      "epoch": 5.283371444026456,
+      "grad_norm": 2.986433506011963,
+      "learning_rate": 1.1990277905163717e-05,
+      "loss": 0.3683,
+      "step": 26860
+    },
+    {
+      "epoch": 5.285338447542475,
+      "grad_norm": 2.3336493968963623,
+      "learning_rate": 1.1987220642636583e-05,
+      "loss": 0.3083,
+      "step": 26870
+    },
+    {
+      "epoch": 5.287305451058494,
+      "grad_norm": 1.0171363353729248,
+      "learning_rate": 1.198416338010945e-05,
+      "loss": 0.427,
+      "step": 26880
+    },
+    {
+      "epoch": 5.289272454574513,
+      "grad_norm": 1.2654080390930176,
+      "learning_rate": 1.1981106117582318e-05,
+      "loss": 0.431,
+      "step": 26890
+    },
+    {
+      "epoch": 5.291239458090532,
+      "grad_norm": 1.0362647771835327,
+      "learning_rate": 1.1978048855055184e-05,
+      "loss": 0.3448,
+      "step": 26900
+    },
+    {
+      "epoch": 5.29320646160655,
+      "grad_norm": 0.8070294260978699,
+      "learning_rate": 1.1974991592528051e-05,
+      "loss": 0.3733,
+      "step": 26910
+    },
+    {
+      "epoch": 5.295173465122569,
+      "grad_norm": 1.697971224784851,
+      "learning_rate": 1.1971934330000919e-05,
+      "loss": 0.4335,
+      "step": 26920
+    },
+    {
+      "epoch": 5.297140468638588,
+      "grad_norm": 1.8027409315109253,
+      "learning_rate": 1.1968877067473786e-05,
+      "loss": 0.3142,
+      "step": 26930
+    },
+    {
+      "epoch": 5.299107472154606,
+      "grad_norm": 1.0393763780593872,
+      "learning_rate": 1.1965819804946652e-05,
+      "loss": 0.403,
+      "step": 26940
+    },
+    {
+      "epoch": 5.301074475670625,
+      "grad_norm": 0.7654443383216858,
+      "learning_rate": 1.196276254241952e-05,
+      "loss": 0.4825,
+      "step": 26950
+    },
+    {
+      "epoch": 5.3030414791866445,
+      "grad_norm": 1.463889479637146,
+      "learning_rate": 1.1959705279892387e-05,
+      "loss": 0.4558,
+      "step": 26960
+    },
+    {
+      "epoch": 5.305008482702663,
+      "grad_norm": 1.0815491676330566,
+      "learning_rate": 1.195664801736525e-05,
+      "loss": 0.2217,
+      "step": 26970
+    },
+    {
+      "epoch": 5.306975486218682,
+      "grad_norm": 0.9593465924263,
+      "learning_rate": 1.1953590754838118e-05,
+      "loss": 0.3366,
+      "step": 26980
+    },
+    {
+      "epoch": 5.308942489734701,
+      "grad_norm": 1.3412492275238037,
+      "learning_rate": 1.1950533492310986e-05,
+      "loss": 0.2579,
+      "step": 26990
+    },
+    {
+      "epoch": 5.310909493250719,
+      "grad_norm": 1.191043496131897,
+      "learning_rate": 1.1947476229783852e-05,
+      "loss": 0.4985,
+      "step": 27000
+    },
+    {
+      "epoch": 5.310909493250719,
+      "eval_loss": 0.15775032341480255,
+      "eval_runtime": 8.861,
+      "eval_samples_per_second": 5.643,
+      "eval_steps_per_second": 2.821,
+      "step": 27000
+    },
+    {
+      "epoch": 5.312876496766738,
+      "grad_norm": 2.1759085655212402,
+      "learning_rate": 1.1944418967256719e-05,
+      "loss": 0.366,
+      "step": 27010
+    },
+    {
+      "epoch": 5.314843500282757,
+      "grad_norm": 1.6948908567428589,
+      "learning_rate": 1.1941361704729586e-05,
+      "loss": 0.414,
+      "step": 27020
+    },
+    {
+      "epoch": 5.316810503798775,
+      "grad_norm": 1.4809417724609375,
+      "learning_rate": 1.1938304442202452e-05,
+      "loss": 0.3949,
+      "step": 27030
+    },
+    {
+      "epoch": 5.3187775073147945,
+      "grad_norm": 2.090308904647827,
+      "learning_rate": 1.193524717967532e-05,
+      "loss": 0.4441,
+      "step": 27040
+    },
+    {
+      "epoch": 5.320744510830814,
+      "grad_norm": 1.194943904876709,
+      "learning_rate": 1.1932189917148187e-05,
+      "loss": 0.4599,
+      "step": 27050
+    },
+    {
+      "epoch": 5.322711514346832,
+      "grad_norm": 1.152958869934082,
+      "learning_rate": 1.1929132654621055e-05,
+      "loss": 0.4161,
+      "step": 27060
+    },
+    {
+      "epoch": 5.324678517862851,
+      "grad_norm": 1.3262512683868408,
+      "learning_rate": 1.192607539209392e-05,
+      "loss": 0.4106,
+      "step": 27070
+    },
+    {
+      "epoch": 5.32664552137887,
+      "grad_norm": 1.7198249101638794,
+      "learning_rate": 1.1923018129566788e-05,
+      "loss": 0.2689,
+      "step": 27080
+    },
+    {
+      "epoch": 5.328612524894888,
+      "grad_norm": 0.76631760597229,
+      "learning_rate": 1.1919960867039652e-05,
+      "loss": 0.4833,
+      "step": 27090
+    },
+    {
+      "epoch": 5.330579528410907,
+      "grad_norm": 1.0496758222579956,
+      "learning_rate": 1.191690360451252e-05,
+      "loss": 0.4318,
+      "step": 27100
+    },
+    {
+      "epoch": 5.332546531926925,
+      "grad_norm": 0.9311888217926025,
+      "learning_rate": 1.1913846341985387e-05,
+      "loss": 0.4997,
+      "step": 27110
+    },
+    {
+      "epoch": 5.3345135354429445,
+      "grad_norm": 1.919349193572998,
+      "learning_rate": 1.1910789079458254e-05,
+      "loss": 0.4288,
+      "step": 27120
+    },
+    {
+      "epoch": 5.336480538958964,
+      "grad_norm": 1.031426191329956,
+      "learning_rate": 1.190773181693112e-05,
+      "loss": 0.3159,
+      "step": 27130
+    },
+    {
+      "epoch": 5.338447542474982,
+      "grad_norm": 0.9320018291473389,
+      "learning_rate": 1.1904674554403988e-05,
+      "loss": 0.4373,
+      "step": 27140
+    },
+    {
+      "epoch": 5.340414545991001,
+      "grad_norm": 0.9622688293457031,
+      "learning_rate": 1.1901617291876855e-05,
+      "loss": 0.4539,
+      "step": 27150
+    },
+    {
+      "epoch": 5.34238154950702,
+      "grad_norm": 1.055091142654419,
+      "learning_rate": 1.1898560029349721e-05,
+      "loss": 0.556,
+      "step": 27160
+    },
+    {
+      "epoch": 5.344348553023038,
+      "grad_norm": 1.1777045726776123,
+      "learning_rate": 1.1895502766822588e-05,
+      "loss": 0.2429,
+      "step": 27170
+    },
+    {
+      "epoch": 5.346315556539057,
+      "grad_norm": 1.031323790550232,
+      "learning_rate": 1.1892445504295456e-05,
+      "loss": 0.2711,
+      "step": 27180
+    },
+    {
+      "epoch": 5.348282560055076,
+      "grad_norm": 1.3598463535308838,
+      "learning_rate": 1.1889388241768323e-05,
+      "loss": 0.376,
+      "step": 27190
+    },
+    {
+      "epoch": 5.3502495635710945,
+      "grad_norm": 0.9220485091209412,
+      "learning_rate": 1.1886330979241187e-05,
+      "loss": 0.4076,
+      "step": 27200
+    },
+    {
+      "epoch": 5.352216567087114,
+      "grad_norm": 1.2869757413864136,
+      "learning_rate": 1.1883273716714055e-05,
+      "loss": 0.2967,
+      "step": 27210
+    },
+    {
+      "epoch": 5.354183570603133,
+      "grad_norm": 1.306017279624939,
+      "learning_rate": 1.188021645418692e-05,
+      "loss": 0.3353,
+      "step": 27220
+    },
+    {
+      "epoch": 5.356150574119151,
+      "grad_norm": 1.5376003980636597,
+      "learning_rate": 1.1877159191659788e-05,
+      "loss": 0.3431,
+      "step": 27230
+    },
+    {
+      "epoch": 5.35811757763517,
+      "grad_norm": 1.538486361503601,
+      "learning_rate": 1.1874101929132656e-05,
+      "loss": 0.3858,
+      "step": 27240
+    },
+    {
+      "epoch": 5.360084581151189,
+      "grad_norm": 1.009808897972107,
+      "learning_rate": 1.1871044666605523e-05,
+      "loss": 0.3684,
+      "step": 27250
+    },
+    {
+      "epoch": 5.362051584667207,
+      "grad_norm": 1.3101840019226074,
+      "learning_rate": 1.1867987404078389e-05,
+      "loss": 0.3092,
+      "step": 27260
+    },
+    {
+      "epoch": 5.364018588183226,
+      "grad_norm": 1.3292027711868286,
+      "learning_rate": 1.1864930141551256e-05,
+      "loss": 0.4574,
+      "step": 27270
+    },
+    {
+      "epoch": 5.365985591699245,
+      "grad_norm": 2.2095041275024414,
+      "learning_rate": 1.1861872879024124e-05,
+      "loss": 0.3182,
+      "step": 27280
+    },
+    {
+      "epoch": 5.367952595215264,
+      "grad_norm": 0.6917831301689148,
+      "learning_rate": 1.185881561649699e-05,
+      "loss": 0.2448,
+      "step": 27290
+    },
+    {
+      "epoch": 5.369919598731283,
+      "grad_norm": 0.4980989396572113,
+      "learning_rate": 1.1855758353969857e-05,
+      "loss": 0.3643,
+      "step": 27300
+    },
+    {
+      "epoch": 5.371886602247302,
+      "grad_norm": 1.0535064935684204,
+      "learning_rate": 1.1852701091442723e-05,
+      "loss": 0.482,
+      "step": 27310
+    },
+    {
+      "epoch": 5.37385360576332,
+      "grad_norm": 1.0042133331298828,
+      "learning_rate": 1.1849643828915589e-05,
+      "loss": 0.3711,
+      "step": 27320
+    },
+    {
+      "epoch": 5.375820609279339,
+      "grad_norm": 1.6627103090286255,
+      "learning_rate": 1.1846586566388456e-05,
+      "loss": 0.4113,
+      "step": 27330
+    },
+    {
+      "epoch": 5.377787612795358,
+      "grad_norm": 1.3288803100585938,
+      "learning_rate": 1.1843529303861324e-05,
+      "loss": 0.4166,
+      "step": 27340
+    },
+    {
+      "epoch": 5.379754616311376,
+      "grad_norm": 2.352654457092285,
+      "learning_rate": 1.184047204133419e-05,
+      "loss": 0.4147,
+      "step": 27350
+    },
+    {
+      "epoch": 5.381721619827395,
+      "grad_norm": 1.5574378967285156,
+      "learning_rate": 1.1837414778807057e-05,
+      "loss": 0.4605,
+      "step": 27360
+    },
+    {
+      "epoch": 5.3836886233434145,
+      "grad_norm": 0.9702771902084351,
+      "learning_rate": 1.1834357516279924e-05,
+      "loss": 0.5087,
+      "step": 27370
+    },
+    {
+      "epoch": 5.385655626859433,
+      "grad_norm": 0.9326369762420654,
+      "learning_rate": 1.1831300253752792e-05,
+      "loss": 0.3388,
+      "step": 27380
+    },
+    {
+      "epoch": 5.387622630375452,
+      "grad_norm": 0.9466789364814758,
+      "learning_rate": 1.1828242991225658e-05,
+      "loss": 0.3118,
+      "step": 27390
+    },
+    {
+      "epoch": 5.389589633891471,
+      "grad_norm": 1.6036527156829834,
+      "learning_rate": 1.1825185728698525e-05,
+      "loss": 0.3969,
+      "step": 27400
+    },
+    {
+      "epoch": 5.391556637407489,
+      "grad_norm": 1.7829012870788574,
+      "learning_rate": 1.1822128466171392e-05,
+      "loss": 0.3626,
+      "step": 27410
+    },
+    {
+      "epoch": 5.393523640923508,
+      "grad_norm": 1.2935879230499268,
+      "learning_rate": 1.1819071203644258e-05,
+      "loss": 0.4981,
+      "step": 27420
+    },
+    {
+      "epoch": 5.395490644439527,
+      "grad_norm": 2.1549735069274902,
+      "learning_rate": 1.1816013941117124e-05,
+      "loss": 0.3817,
+      "step": 27430
+    },
+    {
+      "epoch": 5.397457647955545,
+      "grad_norm": 0.5140044689178467,
+      "learning_rate": 1.1812956678589991e-05,
+      "loss": 0.2148,
+      "step": 27440
+    },
+    {
+      "epoch": 5.3994246514715645,
+      "grad_norm": 2.3934736251831055,
+      "learning_rate": 1.1809899416062857e-05,
+      "loss": 0.382,
+      "step": 27450
+    },
+    {
+      "epoch": 5.401391654987584,
+      "grad_norm": 1.491032361984253,
+      "learning_rate": 1.1806842153535725e-05,
+      "loss": 0.2914,
+      "step": 27460
+    },
+    {
+      "epoch": 5.403358658503602,
+      "grad_norm": 1.6353907585144043,
+      "learning_rate": 1.1803784891008592e-05,
+      "loss": 0.2502,
+      "step": 27470
+    },
+    {
+      "epoch": 5.405325662019621,
+      "grad_norm": 1.7032148838043213,
+      "learning_rate": 1.1800727628481458e-05,
+      "loss": 0.5484,
+      "step": 27480
+    },
+    {
+      "epoch": 5.40729266553564,
+      "grad_norm": 1.4025053977966309,
+      "learning_rate": 1.1797670365954325e-05,
+      "loss": 0.3196,
+      "step": 27490
+    },
+    {
+      "epoch": 5.409259669051658,
+      "grad_norm": 1.2315196990966797,
+      "learning_rate": 1.1794613103427193e-05,
+      "loss": 0.4845,
+      "step": 27500
+    },
+    {
+      "epoch": 5.409259669051658,
+      "eval_loss": 0.16736486554145813,
+      "eval_runtime": 8.8667,
+      "eval_samples_per_second": 5.639,
+      "eval_steps_per_second": 2.82,
+      "step": 27500
+    },
+    {
+      "epoch": 5.411226672567677,
+      "grad_norm": 1.790334701538086,
+      "learning_rate": 1.179155584090006e-05,
+      "loss": 0.3443,
+      "step": 27510
+    },
+    {
+      "epoch": 5.413193676083696,
+      "grad_norm": 0.7707914113998413,
+      "learning_rate": 1.1788498578372926e-05,
+      "loss": 0.3989,
+      "step": 27520
+    },
+    {
+      "epoch": 5.4151606795997145,
+      "grad_norm": 1.0538116693496704,
+      "learning_rate": 1.1785441315845794e-05,
+      "loss": 0.4151,
+      "step": 27530
+    },
+    {
+      "epoch": 5.417127683115734,
+      "grad_norm": 1.2277323007583618,
+      "learning_rate": 1.1782384053318658e-05,
+      "loss": 0.4971,
+      "step": 27540
+    },
+    {
+      "epoch": 5.419094686631753,
+      "grad_norm": 2.6161787509918213,
+      "learning_rate": 1.1779326790791525e-05,
+      "loss": 0.3905,
+      "step": 27550
+    },
+    {
+      "epoch": 5.421061690147771,
+      "grad_norm": 1.2206039428710938,
+      "learning_rate": 1.1776269528264393e-05,
+      "loss": 0.3759,
+      "step": 27560
+    },
+    {
+      "epoch": 5.42302869366379,
+      "grad_norm": 1.2194130420684814,
+      "learning_rate": 1.177321226573726e-05,
+      "loss": 0.4211,
+      "step": 27570
+    },
+    {
+      "epoch": 5.424995697179809,
+      "grad_norm": 1.6475505828857422,
+      "learning_rate": 1.1770155003210126e-05,
+      "loss": 0.4336,
+      "step": 27580
+    },
+    {
+      "epoch": 5.426962700695827,
+      "grad_norm": 1.873953938484192,
+      "learning_rate": 1.1767097740682993e-05,
+      "loss": 0.3902,
+      "step": 27590
+    },
+    {
+      "epoch": 5.428929704211846,
+      "grad_norm": 0.45552027225494385,
+      "learning_rate": 1.1764040478155861e-05,
+      "loss": 0.3605,
+      "step": 27600
+    },
+    {
+      "epoch": 5.430896707727865,
+      "grad_norm": 1.3343397378921509,
+      "learning_rate": 1.1760983215628727e-05,
+      "loss": 0.3379,
+      "step": 27610
+    },
+    {
+      "epoch": 5.432863711243884,
+      "grad_norm": 1.5617072582244873,
+      "learning_rate": 1.1757925953101594e-05,
+      "loss": 0.3483,
+      "step": 27620
+    },
+    {
+      "epoch": 5.434830714759903,
+      "grad_norm": 0.8592610359191895,
+      "learning_rate": 1.1754868690574462e-05,
+      "loss": 0.3791,
+      "step": 27630
+    },
+    {
+      "epoch": 5.436797718275922,
+      "grad_norm": 2.699916362762451,
+      "learning_rate": 1.1751811428047329e-05,
+      "loss": 0.391,
+      "step": 27640
+    },
+    {
+      "epoch": 5.43876472179194,
+      "grad_norm": 1.036454677581787,
+      "learning_rate": 1.1748754165520193e-05,
+      "loss": 0.4393,
+      "step": 27650
+    },
+    {
+      "epoch": 5.440731725307959,
+      "grad_norm": 1.018223524093628,
+      "learning_rate": 1.174569690299306e-05,
+      "loss": 0.3592,
+      "step": 27660
+    },
+    {
+      "epoch": 5.442698728823978,
+      "grad_norm": 0.7438509464263916,
+      "learning_rate": 1.1742639640465926e-05,
+      "loss": 0.3895,
+      "step": 27670
+    },
+    {
+      "epoch": 5.444665732339996,
+      "grad_norm": 1.229840874671936,
+      "learning_rate": 1.1739582377938794e-05,
+      "loss": 0.4443,
+      "step": 27680
+    },
+    {
+      "epoch": 5.446632735856015,
+      "grad_norm": 1.2053213119506836,
+      "learning_rate": 1.1736525115411661e-05,
+      "loss": 0.3692,
+      "step": 27690
+    },
+    {
+      "epoch": 5.4485997393720345,
+      "grad_norm": 1.5629843473434448,
+      "learning_rate": 1.1733467852884529e-05,
+      "loss": 0.4367,
+      "step": 27700
+    },
+    {
+      "epoch": 5.450566742888053,
+      "grad_norm": 1.0979888439178467,
+      "learning_rate": 1.1730410590357395e-05,
+      "loss": 0.3375,
+      "step": 27710
+    },
+    {
+      "epoch": 5.452533746404072,
+      "grad_norm": 0.9558001160621643,
+      "learning_rate": 1.1727353327830262e-05,
+      "loss": 0.4945,
+      "step": 27720
+    },
+    {
+      "epoch": 5.454500749920091,
+      "grad_norm": 3.2173619270324707,
+      "learning_rate": 1.172429606530313e-05,
+      "loss": 0.4662,
+      "step": 27730
+    },
+    {
+      "epoch": 5.456467753436109,
+      "grad_norm": 1.0213227272033691,
+      "learning_rate": 1.1721238802775995e-05,
+      "loss": 0.3302,
+      "step": 27740
+    },
+    {
+      "epoch": 5.458434756952128,
+      "grad_norm": 1.0318022966384888,
+      "learning_rate": 1.1718181540248863e-05,
+      "loss": 0.4005,
+      "step": 27750
+    },
+    {
+      "epoch": 5.460401760468147,
+      "grad_norm": 1.205824613571167,
+      "learning_rate": 1.171512427772173e-05,
+      "loss": 0.3246,
+      "step": 27760
+    },
+    {
+      "epoch": 5.462368763984165,
+      "grad_norm": 1.5735535621643066,
+      "learning_rate": 1.1712067015194594e-05,
+      "loss": 0.4247,
+      "step": 27770
+    },
+    {
+      "epoch": 5.4643357675001845,
+      "grad_norm": 2.0467681884765625,
+      "learning_rate": 1.1709009752667462e-05,
+      "loss": 0.4287,
+      "step": 27780
+    },
+    {
+      "epoch": 5.4663027710162035,
+      "grad_norm": 0.7347807288169861,
+      "learning_rate": 1.170595249014033e-05,
+      "loss": 0.4136,
+      "step": 27790
+    },
+    {
+      "epoch": 5.468269774532222,
+      "grad_norm": 1.6695806980133057,
+      "learning_rate": 1.1702895227613195e-05,
+      "loss": 0.3893,
+      "step": 27800
+    },
+    {
+      "epoch": 5.470236778048241,
+      "grad_norm": 1.408737063407898,
+      "learning_rate": 1.1699837965086063e-05,
+      "loss": 0.3863,
+      "step": 27810
+    },
+    {
+      "epoch": 5.47220378156426,
+      "grad_norm": 2.0042405128479004,
+      "learning_rate": 1.169678070255893e-05,
+      "loss": 0.3712,
+      "step": 27820
+    },
+    {
+      "epoch": 5.474170785080278,
+      "grad_norm": 1.8257968425750732,
+      "learning_rate": 1.1693723440031797e-05,
+      "loss": 0.3769,
+      "step": 27830
+    },
+    {
+      "epoch": 5.476137788596297,
+      "grad_norm": 1.2176882028579712,
+      "learning_rate": 1.1690666177504663e-05,
+      "loss": 0.4373,
+      "step": 27840
+    },
+    {
+      "epoch": 5.478104792112316,
+      "grad_norm": 2.558542013168335,
+      "learning_rate": 1.168760891497753e-05,
+      "loss": 0.3779,
+      "step": 27850
+    },
+    {
+      "epoch": 5.4800717956283345,
+      "grad_norm": 1.4738448858261108,
+      "learning_rate": 1.1684551652450398e-05,
+      "loss": 0.3334,
+      "step": 27860
+    },
+    {
+      "epoch": 5.4820387991443535,
+      "grad_norm": 1.189650058746338,
+      "learning_rate": 1.1681494389923264e-05,
+      "loss": 0.4015,
+      "step": 27870
+    },
+    {
+      "epoch": 5.484005802660373,
+      "grad_norm": 2.8915460109710693,
+      "learning_rate": 1.167843712739613e-05,
+      "loss": 0.3709,
+      "step": 27880
+    },
+    {
+      "epoch": 5.485972806176391,
+      "grad_norm": 1.3140373229980469,
+      "learning_rate": 1.1675379864868997e-05,
+      "loss": 0.3003,
+      "step": 27890
+    },
+    {
+      "epoch": 5.48793980969241,
+      "grad_norm": 2.5268616676330566,
+      "learning_rate": 1.1672322602341863e-05,
+      "loss": 0.3986,
+      "step": 27900
+    },
+    {
+      "epoch": 5.489906813208429,
+      "grad_norm": 1.513447880744934,
+      "learning_rate": 1.166926533981473e-05,
+      "loss": 0.3804,
+      "step": 27910
+    },
+    {
+      "epoch": 5.491873816724447,
+      "grad_norm": 1.0257861614227295,
+      "learning_rate": 1.1666208077287598e-05,
+      "loss": 0.3423,
+      "step": 27920
+    },
+    {
+      "epoch": 5.493840820240466,
+      "grad_norm": 3.507874011993408,
+      "learning_rate": 1.1663150814760464e-05,
+      "loss": 0.3601,
+      "step": 27930
+    },
+    {
+      "epoch": 5.495807823756485,
+      "grad_norm": 0.7899544835090637,
+      "learning_rate": 1.1660093552233331e-05,
+      "loss": 0.3116,
+      "step": 27940
+    },
+    {
+      "epoch": 5.4977748272725036,
+      "grad_norm": 0.9232675433158875,
+      "learning_rate": 1.1657036289706199e-05,
+      "loss": 0.3821,
+      "step": 27950
+    },
+    {
+      "epoch": 5.499741830788523,
+      "grad_norm": 1.5610233545303345,
+      "learning_rate": 1.1653979027179066e-05,
+      "loss": 0.2642,
+      "step": 27960
+    },
+    {
+      "epoch": 5.501708834304541,
+      "grad_norm": 1.6391479969024658,
+      "learning_rate": 1.1650921764651932e-05,
+      "loss": 0.3912,
+      "step": 27970
+    },
+    {
+      "epoch": 5.50367583782056,
+      "grad_norm": 0.9958533644676208,
+      "learning_rate": 1.16478645021248e-05,
+      "loss": 0.3767,
+      "step": 27980
+    },
+    {
+      "epoch": 5.505642841336579,
+      "grad_norm": 1.2634178400039673,
+      "learning_rate": 1.1644807239597663e-05,
+      "loss": 0.3419,
+      "step": 27990
+    },
+    {
+      "epoch": 5.507609844852597,
+      "grad_norm": 1.0763895511627197,
+      "learning_rate": 1.1641749977070531e-05,
+      "loss": 0.3763,
+      "step": 28000
+    },
+    {
+      "epoch": 5.507609844852597,
+      "eval_loss": 0.1603037267923355,
+      "eval_runtime": 8.8616,
+      "eval_samples_per_second": 5.642,
+      "eval_steps_per_second": 2.821,
+      "step": 28000
+    },
+    {
+      "epoch": 5.509576848368616,
+      "grad_norm": 2.0165274143218994,
+      "learning_rate": 1.1638692714543398e-05,
+      "loss": 0.3444,
+      "step": 28010
+    },
+    {
+      "epoch": 5.511543851884635,
+      "grad_norm": 1.7106163501739502,
+      "learning_rate": 1.1635635452016266e-05,
+      "loss": 0.4143,
+      "step": 28020
+    },
+    {
+      "epoch": 5.5135108554006536,
+      "grad_norm": 1.8056857585906982,
+      "learning_rate": 1.1632578189489132e-05,
+      "loss": 0.3066,
+      "step": 28030
+    },
+    {
+      "epoch": 5.515477858916673,
+      "grad_norm": 0.7835657596588135,
+      "learning_rate": 1.1629520926961999e-05,
+      "loss": 0.383,
+      "step": 28040
+    },
+    {
+      "epoch": 5.517444862432692,
+      "grad_norm": 0.7902368307113647,
+      "learning_rate": 1.1626463664434867e-05,
+      "loss": 0.568,
+      "step": 28050
+    },
+    {
+      "epoch": 5.51941186594871,
+      "grad_norm": 1.101770281791687,
+      "learning_rate": 1.1623406401907732e-05,
+      "loss": 0.2871,
+      "step": 28060
+    },
+    {
+      "epoch": 5.521378869464729,
+      "grad_norm": 1.5315871238708496,
+      "learning_rate": 1.16203491393806e-05,
+      "loss": 0.4072,
+      "step": 28070
+    },
+    {
+      "epoch": 5.523345872980748,
+      "grad_norm": 1.7434234619140625,
+      "learning_rate": 1.1617291876853467e-05,
+      "loss": 0.4428,
+      "step": 28080
+    },
+    {
+      "epoch": 5.525312876496766,
+      "grad_norm": 0.9678106904029846,
+      "learning_rate": 1.1614234614326335e-05,
+      "loss": 0.3465,
+      "step": 28090
+    },
+    {
+      "epoch": 5.527279880012785,
+      "grad_norm": 1.8099299669265747,
+      "learning_rate": 1.16111773517992e-05,
+      "loss": 0.429,
+      "step": 28100
+    },
+    {
+      "epoch": 5.5292468835288044,
+      "grad_norm": 1.2735828161239624,
+      "learning_rate": 1.1608120089272066e-05,
+      "loss": 0.3512,
+      "step": 28110
+    },
+    {
+      "epoch": 5.531213887044823,
+      "grad_norm": 1.0866531133651733,
+      "learning_rate": 1.1605062826744932e-05,
+      "loss": 0.3188,
+      "step": 28120
+    },
+    {
+      "epoch": 5.533180890560842,
+      "grad_norm": 1.9226776361465454,
+      "learning_rate": 1.16020055642178e-05,
+      "loss": 0.3112,
+      "step": 28130
+    },
+    {
+      "epoch": 5.535147894076861,
+      "grad_norm": 0.8396310806274414,
+      "learning_rate": 1.1598948301690667e-05,
+      "loss": 0.2962,
+      "step": 28140
+    },
+    {
+      "epoch": 5.537114897592879,
+      "grad_norm": 1.6284130811691284,
+      "learning_rate": 1.1595891039163535e-05,
+      "loss": 0.2717,
+      "step": 28150
+    },
+    {
+      "epoch": 5.539081901108898,
+      "grad_norm": 0.705360472202301,
+      "learning_rate": 1.15928337766364e-05,
+      "loss": 0.3184,
+      "step": 28160
+    },
+    {
+      "epoch": 5.541048904624917,
+      "grad_norm": 2.09700345993042,
+      "learning_rate": 1.1589776514109268e-05,
+      "loss": 0.4465,
+      "step": 28170
+    },
+    {
+      "epoch": 5.543015908140935,
+      "grad_norm": 0.7859101891517639,
+      "learning_rate": 1.1586719251582135e-05,
+      "loss": 0.355,
+      "step": 28180
+    },
+    {
+      "epoch": 5.5449829116569544,
+      "grad_norm": 2.0662646293640137,
+      "learning_rate": 1.1583661989055001e-05,
+      "loss": 0.4825,
+      "step": 28190
+    },
+    {
+      "epoch": 5.5469499151729735,
+      "grad_norm": 1.131251335144043,
+      "learning_rate": 1.1580604726527869e-05,
+      "loss": 0.3399,
+      "step": 28200
+    },
+    {
+      "epoch": 5.548916918688992,
+      "grad_norm": 1.2447280883789062,
+      "learning_rate": 1.1577547464000736e-05,
+      "loss": 0.4206,
+      "step": 28210
+    },
+    {
+      "epoch": 5.550883922205011,
+      "grad_norm": 2.063737154006958,
+      "learning_rate": 1.15744902014736e-05,
+      "loss": 0.4005,
+      "step": 28220
+    },
+    {
+      "epoch": 5.55285092572103,
+      "grad_norm": 1.5366238355636597,
+      "learning_rate": 1.1571432938946468e-05,
+      "loss": 0.378,
+      "step": 28230
+    },
+    {
+      "epoch": 5.554817929237048,
+      "grad_norm": 1.0414916276931763,
+      "learning_rate": 1.1568375676419335e-05,
+      "loss": 0.3657,
+      "step": 28240
+    },
+    {
+      "epoch": 5.556784932753067,
+      "grad_norm": 0.9085996150970459,
+      "learning_rate": 1.15653184138922e-05,
+      "loss": 0.4206,
+      "step": 28250
+    },
+    {
+      "epoch": 5.558751936269086,
+      "grad_norm": 1.2168952226638794,
+      "learning_rate": 1.1562261151365068e-05,
+      "loss": 0.2791,
+      "step": 28260
+    },
+    {
+      "epoch": 5.5607189397851045,
+      "grad_norm": 2.0721793174743652,
+      "learning_rate": 1.1559203888837936e-05,
+      "loss": 0.4165,
+      "step": 28270
+    },
+    {
+      "epoch": 5.5626859433011235,
+      "grad_norm": 1.996752142906189,
+      "learning_rate": 1.1556146626310803e-05,
+      "loss": 0.3526,
+      "step": 28280
+    },
+    {
+      "epoch": 5.564652946817143,
+      "grad_norm": 0.9958781003952026,
+      "learning_rate": 1.1553089363783669e-05,
+      "loss": 0.2766,
+      "step": 28290
+    },
+    {
+      "epoch": 5.566619950333161,
+      "grad_norm": 1.2039257287979126,
+      "learning_rate": 1.1550032101256536e-05,
+      "loss": 0.3814,
+      "step": 28300
+    },
+    {
+      "epoch": 5.56858695384918,
+      "grad_norm": 0.8608440160751343,
+      "learning_rate": 1.1546974838729404e-05,
+      "loss": 0.3909,
+      "step": 28310
+    },
+    {
+      "epoch": 5.570553957365199,
+      "grad_norm": 2.0225956439971924,
+      "learning_rate": 1.154391757620227e-05,
+      "loss": 0.5571,
+      "step": 28320
+    },
+    {
+      "epoch": 5.572520960881217,
+      "grad_norm": 1.0121216773986816,
+      "learning_rate": 1.1540860313675135e-05,
+      "loss": 0.3789,
+      "step": 28330
+    },
+    {
+      "epoch": 5.574487964397236,
+      "grad_norm": 0.8687868714332581,
+      "learning_rate": 1.1537803051148003e-05,
+      "loss": 0.2573,
+      "step": 28340
+    },
+    {
+      "epoch": 5.576454967913255,
+      "grad_norm": 1.3339784145355225,
+      "learning_rate": 1.1534745788620869e-05,
+      "loss": 0.4801,
+      "step": 28350
+    },
+    {
+      "epoch": 5.5784219714292735,
+      "grad_norm": 1.0207067728042603,
+      "learning_rate": 1.1531688526093736e-05,
+      "loss": 0.4482,
+      "step": 28360
+    },
+    {
+      "epoch": 5.580388974945293,
+      "grad_norm": 1.1519733667373657,
+      "learning_rate": 1.1528631263566604e-05,
+      "loss": 0.4116,
+      "step": 28370
+    },
+    {
+      "epoch": 5.582355978461312,
+      "grad_norm": 2.023810386657715,
+      "learning_rate": 1.152557400103947e-05,
+      "loss": 0.3789,
+      "step": 28380
+    },
+    {
+      "epoch": 5.58432298197733,
+      "grad_norm": 1.0238547325134277,
+      "learning_rate": 1.1522516738512337e-05,
+      "loss": 0.3157,
+      "step": 28390
+    },
+    {
+      "epoch": 5.586289985493349,
+      "grad_norm": 0.9904107451438904,
+      "learning_rate": 1.1519459475985204e-05,
+      "loss": 0.4655,
+      "step": 28400
+    },
+    {
+      "epoch": 5.588256989009368,
+      "grad_norm": 1.5001986026763916,
+      "learning_rate": 1.1516402213458072e-05,
+      "loss": 0.3845,
+      "step": 28410
+    },
+    {
+      "epoch": 5.590223992525386,
+      "grad_norm": 1.6891433000564575,
+      "learning_rate": 1.1513344950930938e-05,
+      "loss": 0.4138,
+      "step": 28420
+    },
+    {
+      "epoch": 5.592190996041405,
+      "grad_norm": 2.148615598678589,
+      "learning_rate": 1.1510287688403805e-05,
+      "loss": 0.3071,
+      "step": 28430
+    },
+    {
+      "epoch": 5.594157999557424,
+      "grad_norm": 0.9641100764274597,
+      "learning_rate": 1.1507230425876673e-05,
+      "loss": 0.3471,
+      "step": 28440
+    },
+    {
+      "epoch": 5.596125003073443,
+      "grad_norm": 0.9903495907783508,
+      "learning_rate": 1.1504173163349537e-05,
+      "loss": 0.3853,
+      "step": 28450
+    },
+    {
+      "epoch": 5.598092006589462,
+      "grad_norm": 1.5597717761993408,
+      "learning_rate": 1.1501115900822404e-05,
+      "loss": 0.3705,
+      "step": 28460
+    },
+    {
+      "epoch": 5.600059010105481,
+      "grad_norm": 1.0873743295669556,
+      "learning_rate": 1.1498058638295272e-05,
+      "loss": 0.3527,
+      "step": 28470
+    },
+    {
+      "epoch": 5.602026013621499,
+      "grad_norm": 1.7205636501312256,
+      "learning_rate": 1.1495001375768137e-05,
+      "loss": 0.311,
+      "step": 28480
+    },
+    {
+      "epoch": 5.603993017137518,
+      "grad_norm": 2.3332083225250244,
+      "learning_rate": 1.1491944113241005e-05,
+      "loss": 0.3583,
+      "step": 28490
+    },
+    {
+      "epoch": 5.605960020653537,
+      "grad_norm": 1.5018432140350342,
+      "learning_rate": 1.1488886850713872e-05,
+      "loss": 0.2985,
+      "step": 28500
+    },
+    {
+      "epoch": 5.605960020653537,
+      "eval_loss": 0.15660564601421356,
+      "eval_runtime": 8.8811,
+      "eval_samples_per_second": 5.63,
+      "eval_steps_per_second": 2.815,
+      "step": 28500
+    },
+    {
+      "epoch": 5.607927024169555,
+      "grad_norm": 0.7703841924667358,
+      "learning_rate": 1.1485829588186738e-05,
+      "loss": 0.486,
+      "step": 28510
+    },
+    {
+      "epoch": 5.609894027685574,
+      "grad_norm": 2.638970375061035,
+      "learning_rate": 1.1482772325659606e-05,
+      "loss": 0.3021,
+      "step": 28520
+    },
+    {
+      "epoch": 5.6118610312015935,
+      "grad_norm": 0.8547009229660034,
+      "learning_rate": 1.1479715063132473e-05,
+      "loss": 0.3226,
+      "step": 28530
+    },
+    {
+      "epoch": 5.613828034717612,
+      "grad_norm": 1.1358122825622559,
+      "learning_rate": 1.147665780060534e-05,
+      "loss": 0.4475,
+      "step": 28540
+    },
+    {
+      "epoch": 5.615795038233631,
+      "grad_norm": 1.3914819955825806,
+      "learning_rate": 1.1473600538078206e-05,
+      "loss": 0.3538,
+      "step": 28550
+    },
+    {
+      "epoch": 5.61776204174965,
+      "grad_norm": 1.1753898859024048,
+      "learning_rate": 1.1470543275551072e-05,
+      "loss": 0.2922,
+      "step": 28560
+    },
+    {
+      "epoch": 5.619729045265668,
+      "grad_norm": 1.4788683652877808,
+      "learning_rate": 1.1467486013023938e-05,
+      "loss": 0.3902,
+      "step": 28570
+    },
+    {
+      "epoch": 5.621696048781687,
+      "grad_norm": 1.8976927995681763,
+      "learning_rate": 1.1464428750496805e-05,
+      "loss": 0.468,
+      "step": 28580
+    },
+    {
+      "epoch": 5.623663052297706,
+      "grad_norm": 0.8130624294281006,
+      "learning_rate": 1.1461371487969673e-05,
+      "loss": 0.4195,
+      "step": 28590
+    },
+    {
+      "epoch": 5.625630055813724,
+      "grad_norm": 1.275107741355896,
+      "learning_rate": 1.145831422544254e-05,
+      "loss": 0.3311,
+      "step": 28600
+    },
+    {
+      "epoch": 5.6275970593297435,
+      "grad_norm": 1.0753655433654785,
+      "learning_rate": 1.1455256962915406e-05,
+      "loss": 0.3383,
+      "step": 28610
+    },
+    {
+      "epoch": 5.629564062845763,
+      "grad_norm": 0.929724931716919,
+      "learning_rate": 1.1452199700388274e-05,
+      "loss": 0.4467,
+      "step": 28620
+    },
+    {
+      "epoch": 5.631531066361781,
+      "grad_norm": 1.3278186321258545,
+      "learning_rate": 1.1449142437861141e-05,
+      "loss": 0.3923,
+      "step": 28630
+    },
+    {
+      "epoch": 5.6334980698778,
+      "grad_norm": 1.7642334699630737,
+      "learning_rate": 1.1446085175334007e-05,
+      "loss": 0.3315,
+      "step": 28640
+    },
+    {
+      "epoch": 5.635465073393819,
+      "grad_norm": 2.041316032409668,
+      "learning_rate": 1.1443027912806874e-05,
+      "loss": 0.3939,
+      "step": 28650
+    },
+    {
+      "epoch": 5.637432076909837,
+      "grad_norm": 1.0631572008132935,
+      "learning_rate": 1.1439970650279742e-05,
+      "loss": 0.3989,
+      "step": 28660
+    },
+    {
+      "epoch": 5.639399080425856,
+      "grad_norm": 0.8136534094810486,
+      "learning_rate": 1.1436913387752606e-05,
+      "loss": 0.3533,
+      "step": 28670
+    },
+    {
+      "epoch": 5.641366083941875,
+      "grad_norm": 0.8498640656471252,
+      "learning_rate": 1.1433856125225473e-05,
+      "loss": 0.3885,
+      "step": 28680
+    },
+    {
+      "epoch": 5.6433330874578935,
+      "grad_norm": 1.0519620180130005,
+      "learning_rate": 1.143079886269834e-05,
+      "loss": 0.3664,
+      "step": 28690
+    },
+    {
+      "epoch": 5.645300090973913,
+      "grad_norm": 2.0294244289398193,
+      "learning_rate": 1.1427741600171207e-05,
+      "loss": 0.2631,
+      "step": 28700
+    },
+    {
+      "epoch": 5.647267094489932,
+      "grad_norm": 1.1428825855255127,
+      "learning_rate": 1.1424684337644074e-05,
+      "loss": 0.3553,
+      "step": 28710
+    },
+    {
+      "epoch": 5.64923409800595,
+      "grad_norm": 1.2105053663253784,
+      "learning_rate": 1.1421627075116941e-05,
+      "loss": 0.2884,
+      "step": 28720
+    },
+    {
+      "epoch": 5.651201101521969,
+      "grad_norm": 0.7727178931236267,
+      "learning_rate": 1.1418569812589809e-05,
+      "loss": 0.462,
+      "step": 28730
+    },
+    {
+      "epoch": 5.653168105037988,
+      "grad_norm": 0.9051704406738281,
+      "learning_rate": 1.1415512550062675e-05,
+      "loss": 0.3031,
+      "step": 28740
+    },
+    {
+      "epoch": 5.655135108554006,
+      "grad_norm": 0.8865196704864502,
+      "learning_rate": 1.1412455287535542e-05,
+      "loss": 0.4039,
+      "step": 28750
+    },
+    {
+      "epoch": 5.657102112070025,
+      "grad_norm": 1.78319251537323,
+      "learning_rate": 1.140939802500841e-05,
+      "loss": 0.3758,
+      "step": 28760
+    },
+    {
+      "epoch": 5.659069115586044,
+      "grad_norm": 1.8657971620559692,
+      "learning_rate": 1.1406340762481275e-05,
+      "loss": 0.3919,
+      "step": 28770
+    },
+    {
+      "epoch": 5.661036119102063,
+      "grad_norm": 1.4746910333633423,
+      "learning_rate": 1.1403283499954143e-05,
+      "loss": 0.4362,
+      "step": 28780
+    },
+    {
+      "epoch": 5.663003122618082,
+      "grad_norm": 1.1907963752746582,
+      "learning_rate": 1.1400226237427007e-05,
+      "loss": 0.3671,
+      "step": 28790
+    },
+    {
+      "epoch": 5.664970126134101,
+      "grad_norm": 0.8082676529884338,
+      "learning_rate": 1.1397168974899874e-05,
+      "loss": 0.4336,
+      "step": 28800
+    },
+    {
+      "epoch": 5.666937129650119,
+      "grad_norm": 0.8014049530029297,
+      "learning_rate": 1.1394111712372742e-05,
+      "loss": 0.3373,
+      "step": 28810
+    },
+    {
+      "epoch": 5.668904133166138,
+      "grad_norm": 1.8428163528442383,
+      "learning_rate": 1.139105444984561e-05,
+      "loss": 0.3643,
+      "step": 28820
+    },
+    {
+      "epoch": 5.670871136682157,
+      "grad_norm": 0.9919217228889465,
+      "learning_rate": 1.1387997187318475e-05,
+      "loss": 0.4063,
+      "step": 28830
+    },
+    {
+      "epoch": 5.672838140198175,
+      "grad_norm": 0.7838166952133179,
+      "learning_rate": 1.1384939924791343e-05,
+      "loss": 0.395,
+      "step": 28840
+    },
+    {
+      "epoch": 5.674805143714194,
+      "grad_norm": 0.8631680011749268,
+      "learning_rate": 1.138188266226421e-05,
+      "loss": 0.3382,
+      "step": 28850
+    },
+    {
+      "epoch": 5.6767721472302135,
+      "grad_norm": 0.9368540644645691,
+      "learning_rate": 1.1378825399737078e-05,
+      "loss": 0.2909,
+      "step": 28860
+    },
+    {
+      "epoch": 5.678739150746232,
+      "grad_norm": 1.333701252937317,
+      "learning_rate": 1.1375768137209943e-05,
+      "loss": 0.3619,
+      "step": 28870
+    },
+    {
+      "epoch": 5.680706154262251,
+      "grad_norm": 1.3428051471710205,
+      "learning_rate": 1.1372710874682811e-05,
+      "loss": 0.37,
+      "step": 28880
+    },
+    {
+      "epoch": 5.68267315777827,
+      "grad_norm": 1.2829957008361816,
+      "learning_rate": 1.1369653612155678e-05,
+      "loss": 0.3244,
+      "step": 28890
+    },
+    {
+      "epoch": 5.684640161294288,
+      "grad_norm": 2.1554901599884033,
+      "learning_rate": 1.1366596349628542e-05,
+      "loss": 0.2984,
+      "step": 28900
+    },
+    {
+      "epoch": 5.686607164810307,
+      "grad_norm": 2.656132936477661,
+      "learning_rate": 1.136353908710141e-05,
+      "loss": 0.4132,
+      "step": 28910
+    },
+    {
+      "epoch": 5.688574168326326,
+      "grad_norm": 0.8451870679855347,
+      "learning_rate": 1.1360481824574276e-05,
+      "loss": 0.4124,
+      "step": 28920
+    },
+    {
+      "epoch": 5.690541171842344,
+      "grad_norm": 0.7421035170555115,
+      "learning_rate": 1.1357424562047143e-05,
+      "loss": 0.3635,
+      "step": 28930
+    },
+    {
+      "epoch": 5.6925081753583635,
+      "grad_norm": 2.9646100997924805,
+      "learning_rate": 1.135436729952001e-05,
+      "loss": 0.4154,
+      "step": 28940
+    },
+    {
+      "epoch": 5.694475178874383,
+      "grad_norm": 1.0777528285980225,
+      "learning_rate": 1.1351310036992878e-05,
+      "loss": 0.4434,
+      "step": 28950
+    },
+    {
+      "epoch": 5.696442182390401,
+      "grad_norm": 1.156477689743042,
+      "learning_rate": 1.1348252774465744e-05,
+      "loss": 0.3711,
+      "step": 28960
+    },
+    {
+      "epoch": 5.69840918590642,
+      "grad_norm": 2.4916858673095703,
+      "learning_rate": 1.1345195511938611e-05,
+      "loss": 0.3715,
+      "step": 28970
+    },
+    {
+      "epoch": 5.700376189422439,
+      "grad_norm": 1.9382354021072388,
+      "learning_rate": 1.1342138249411479e-05,
+      "loss": 0.3031,
+      "step": 28980
+    },
+    {
+      "epoch": 5.702343192938457,
+      "grad_norm": 0.9780879020690918,
+      "learning_rate": 1.1339080986884346e-05,
+      "loss": 0.4033,
+      "step": 28990
+    },
+    {
+      "epoch": 5.704310196454476,
+      "grad_norm": 0.7445433139801025,
+      "learning_rate": 1.1336023724357212e-05,
+      "loss": 0.3011,
+      "step": 29000
+    },
+    {
+      "epoch": 5.704310196454476,
+      "eval_loss": 0.15423063933849335,
+      "eval_runtime": 8.8909,
+      "eval_samples_per_second": 5.624,
+      "eval_steps_per_second": 2.812,
+      "step": 29000
+    },
+    {
+      "epoch": 5.706277199970495,
+      "grad_norm": 0.8199791312217712,
+      "learning_rate": 1.1332966461830078e-05,
+      "loss": 0.5285,
+      "step": 29010
+    },
+    {
+      "epoch": 5.7082442034865135,
+      "grad_norm": 1.2353311777114868,
+      "learning_rate": 1.1329909199302944e-05,
+      "loss": 0.4156,
+      "step": 29020
+    },
+    {
+      "epoch": 5.710211207002533,
+      "grad_norm": 0.6677939295768738,
+      "learning_rate": 1.1326851936775811e-05,
+      "loss": 0.3689,
+      "step": 29030
+    },
+    {
+      "epoch": 5.712178210518552,
+      "grad_norm": 0.8684714436531067,
+      "learning_rate": 1.1323794674248679e-05,
+      "loss": 0.2528,
+      "step": 29040
+    },
+    {
+      "epoch": 5.71414521403457,
+      "grad_norm": 1.1042633056640625,
+      "learning_rate": 1.1320737411721544e-05,
+      "loss": 0.4604,
+      "step": 29050
+    },
+    {
+      "epoch": 5.716112217550589,
+      "grad_norm": 1.7766307592391968,
+      "learning_rate": 1.1317680149194412e-05,
+      "loss": 0.3185,
+      "step": 29060
+    },
+    {
+      "epoch": 5.718079221066608,
+      "grad_norm": 1.5943020582199097,
+      "learning_rate": 1.131462288666728e-05,
+      "loss": 0.4921,
+      "step": 29070
+    },
+    {
+      "epoch": 5.720046224582626,
+      "grad_norm": 0.9290481209754944,
+      "learning_rate": 1.1311565624140147e-05,
+      "loss": 0.3609,
+      "step": 29080
+    },
+    {
+      "epoch": 5.722013228098645,
+      "grad_norm": 1.2227500677108765,
+      "learning_rate": 1.1308508361613013e-05,
+      "loss": 0.2812,
+      "step": 29090
+    },
+    {
+      "epoch": 5.723980231614664,
+      "grad_norm": 1.03496515750885,
+      "learning_rate": 1.130545109908588e-05,
+      "loss": 0.4757,
+      "step": 29100
+    },
+    {
+      "epoch": 5.725947235130683,
+      "grad_norm": 1.3645904064178467,
+      "learning_rate": 1.1302393836558747e-05,
+      "loss": 0.5169,
+      "step": 29110
+    },
+    {
+      "epoch": 5.727914238646702,
+      "grad_norm": 1.264628529548645,
+      "learning_rate": 1.1299336574031615e-05,
+      "loss": 0.4321,
+      "step": 29120
+    },
+    {
+      "epoch": 5.729881242162721,
+      "grad_norm": 1.108580231666565,
+      "learning_rate": 1.1296279311504479e-05,
+      "loss": 0.3469,
+      "step": 29130
+    },
+    {
+      "epoch": 5.731848245678739,
+      "grad_norm": 1.319706678390503,
+      "learning_rate": 1.1293222048977347e-05,
+      "loss": 0.4453,
+      "step": 29140
+    },
+    {
+      "epoch": 5.733815249194758,
+      "grad_norm": 1.1196752786636353,
+      "learning_rate": 1.1290164786450212e-05,
+      "loss": 0.5532,
+      "step": 29150
+    },
+    {
+      "epoch": 5.735782252710777,
+      "grad_norm": 1.2848409414291382,
+      "learning_rate": 1.128710752392308e-05,
+      "loss": 0.4147,
+      "step": 29160
+    },
+    {
+      "epoch": 5.737749256226795,
+      "grad_norm": 1.723487377166748,
+      "learning_rate": 1.1284050261395947e-05,
+      "loss": 0.3738,
+      "step": 29170
+    },
+    {
+      "epoch": 5.739716259742814,
+      "grad_norm": 0.617231547832489,
+      "learning_rate": 1.1280992998868813e-05,
+      "loss": 0.368,
+      "step": 29180
+    },
+    {
+      "epoch": 5.7416832632588335,
+      "grad_norm": 1.5116612911224365,
+      "learning_rate": 1.127793573634168e-05,
+      "loss": 0.4737,
+      "step": 29190
+    },
+    {
+      "epoch": 5.743650266774852,
+      "grad_norm": 1.343389868736267,
+      "learning_rate": 1.1274878473814548e-05,
+      "loss": 0.3062,
+      "step": 29200
+    },
+    {
+      "epoch": 5.745617270290871,
+      "grad_norm": 0.6800724864006042,
+      "learning_rate": 1.1271821211287415e-05,
+      "loss": 0.4265,
+      "step": 29210
+    },
+    {
+      "epoch": 5.74758427380689,
+      "grad_norm": 1.4235820770263672,
+      "learning_rate": 1.1268763948760281e-05,
+      "loss": 0.4251,
+      "step": 29220
+    },
+    {
+      "epoch": 5.749551277322908,
+      "grad_norm": 1.607358455657959,
+      "learning_rate": 1.1265706686233149e-05,
+      "loss": 0.3046,
+      "step": 29230
+    },
+    {
+      "epoch": 5.751518280838927,
+      "grad_norm": 1.0604151487350464,
+      "learning_rate": 1.1262649423706013e-05,
+      "loss": 0.2405,
+      "step": 29240
+    },
+    {
+      "epoch": 5.753485284354946,
+      "grad_norm": 1.6413496732711792,
+      "learning_rate": 1.125959216117888e-05,
+      "loss": 0.3785,
+      "step": 29250
+    },
+    {
+      "epoch": 5.755452287870964,
+      "grad_norm": 2.484588146209717,
+      "learning_rate": 1.1256534898651748e-05,
+      "loss": 0.4283,
+      "step": 29260
+    },
+    {
+      "epoch": 5.7574192913869835,
+      "grad_norm": 0.934377133846283,
+      "learning_rate": 1.1253477636124615e-05,
+      "loss": 0.4663,
+      "step": 29270
+    },
+    {
+      "epoch": 5.759386294903003,
+      "grad_norm": 0.6259771585464478,
+      "learning_rate": 1.1250420373597481e-05,
+      "loss": 0.4307,
+      "step": 29280
+    },
+    {
+      "epoch": 5.761353298419021,
+      "grad_norm": 1.0241243839263916,
+      "learning_rate": 1.1247363111070348e-05,
+      "loss": 0.3361,
+      "step": 29290
+    },
+    {
+      "epoch": 5.76332030193504,
+      "grad_norm": 1.6852381229400635,
+      "learning_rate": 1.1244305848543216e-05,
+      "loss": 0.3302,
+      "step": 29300
+    },
+    {
+      "epoch": 5.765287305451059,
+      "grad_norm": 1.2226417064666748,
+      "learning_rate": 1.1241248586016082e-05,
+      "loss": 0.2649,
+      "step": 29310
+    },
+    {
+      "epoch": 5.767254308967077,
+      "grad_norm": 0.9367130398750305,
+      "learning_rate": 1.1238191323488949e-05,
+      "loss": 0.3448,
+      "step": 29320
+    },
+    {
+      "epoch": 5.769221312483096,
+      "grad_norm": 1.9569463729858398,
+      "learning_rate": 1.1235134060961817e-05,
+      "loss": 0.291,
+      "step": 29330
+    },
+    {
+      "epoch": 5.771188315999115,
+      "grad_norm": 1.3722015619277954,
+      "learning_rate": 1.1232076798434684e-05,
+      "loss": 0.3712,
+      "step": 29340
+    },
+    {
+      "epoch": 5.7731553195151335,
+      "grad_norm": 1.0473368167877197,
+      "learning_rate": 1.1229019535907548e-05,
+      "loss": 0.3014,
+      "step": 29350
+    },
+    {
+      "epoch": 5.775122323031153,
+      "grad_norm": 1.4684929847717285,
+      "learning_rate": 1.1225962273380416e-05,
+      "loss": 0.279,
+      "step": 29360
+    },
+    {
+      "epoch": 5.777089326547172,
+      "grad_norm": 1.4670180082321167,
+      "learning_rate": 1.1222905010853281e-05,
+      "loss": 0.4775,
+      "step": 29370
+    },
+    {
+      "epoch": 5.77905633006319,
+      "grad_norm": 1.68386709690094,
+      "learning_rate": 1.1219847748326149e-05,
+      "loss": 0.3514,
+      "step": 29380
+    },
+    {
+      "epoch": 5.781023333579209,
+      "grad_norm": 1.5234280824661255,
+      "learning_rate": 1.1216790485799016e-05,
+      "loss": 0.5099,
+      "step": 29390
+    },
+    {
+      "epoch": 5.782990337095228,
+      "grad_norm": 2.241706609725952,
+      "learning_rate": 1.1213733223271884e-05,
+      "loss": 0.4748,
+      "step": 29400
+    },
+    {
+      "epoch": 5.784957340611246,
+      "grad_norm": 0.9543262124061584,
+      "learning_rate": 1.121067596074475e-05,
+      "loss": 0.3328,
+      "step": 29410
+    },
+    {
+      "epoch": 5.786924344127265,
+      "grad_norm": 0.9000210165977478,
+      "learning_rate": 1.1207618698217617e-05,
+      "loss": 0.4743,
+      "step": 29420
+    },
+    {
+      "epoch": 5.788891347643284,
+      "grad_norm": 1.6912353038787842,
+      "learning_rate": 1.1204561435690485e-05,
+      "loss": 0.4446,
+      "step": 29430
+    },
+    {
+      "epoch": 5.790858351159303,
+      "grad_norm": 1.8903340101242065,
+      "learning_rate": 1.120150417316335e-05,
+      "loss": 0.3844,
+      "step": 29440
+    },
+    {
+      "epoch": 5.792825354675322,
+      "grad_norm": 1.5042668581008911,
+      "learning_rate": 1.1198446910636218e-05,
+      "loss": 0.5255,
+      "step": 29450
+    },
+    {
+      "epoch": 5.794792358191341,
+      "grad_norm": 1.3050025701522827,
+      "learning_rate": 1.1195389648109085e-05,
+      "loss": 0.391,
+      "step": 29460
+    },
+    {
+      "epoch": 5.796759361707359,
+      "grad_norm": 1.804996371269226,
+      "learning_rate": 1.119233238558195e-05,
+      "loss": 0.4965,
+      "step": 29470
+    },
+    {
+      "epoch": 5.798726365223378,
+      "grad_norm": 1.802411437034607,
+      "learning_rate": 1.1189275123054817e-05,
+      "loss": 0.3778,
+      "step": 29480
+    },
+    {
+      "epoch": 5.800693368739397,
+      "grad_norm": 1.2030383348464966,
+      "learning_rate": 1.1186217860527684e-05,
+      "loss": 0.3617,
+      "step": 29490
+    },
+    {
+      "epoch": 5.802660372255415,
+      "grad_norm": 1.6366369724273682,
+      "learning_rate": 1.118316059800055e-05,
+      "loss": 0.3232,
+      "step": 29500
+    },
+    {
+      "epoch": 5.802660372255415,
+      "eval_loss": 0.1477995663881302,
+      "eval_runtime": 8.8723,
+      "eval_samples_per_second": 5.635,
+      "eval_steps_per_second": 2.818,
+      "step": 29500
+    },
+    {
+      "epoch": 5.804627375771434,
+      "grad_norm": 0.9681687951087952,
+      "learning_rate": 1.1180103335473418e-05,
+      "loss": 0.37,
+      "step": 29510
+    },
+    {
+      "epoch": 5.806594379287453,
+      "grad_norm": 1.039945363998413,
+      "learning_rate": 1.1177046072946285e-05,
+      "loss": 0.3007,
+      "step": 29520
+    },
+    {
+      "epoch": 5.808561382803472,
+      "grad_norm": 1.011750340461731,
+      "learning_rate": 1.1173988810419152e-05,
+      "loss": 0.362,
+      "step": 29530
+    },
+    {
+      "epoch": 5.810528386319491,
+      "grad_norm": 1.4740638732910156,
+      "learning_rate": 1.1170931547892018e-05,
+      "loss": 0.4319,
+      "step": 29540
+    },
+    {
+      "epoch": 5.812495389835509,
+      "grad_norm": 0.6471105813980103,
+      "learning_rate": 1.1167874285364886e-05,
+      "loss": 0.3625,
+      "step": 29550
+    },
+    {
+      "epoch": 5.814462393351528,
+      "grad_norm": 0.8530930876731873,
+      "learning_rate": 1.1164817022837753e-05,
+      "loss": 0.2879,
+      "step": 29560
+    },
+    {
+      "epoch": 5.816429396867547,
+      "grad_norm": 2.395834445953369,
+      "learning_rate": 1.1161759760310619e-05,
+      "loss": 0.3562,
+      "step": 29570
+    },
+    {
+      "epoch": 5.818396400383565,
+      "grad_norm": 1.0060678720474243,
+      "learning_rate": 1.1158702497783485e-05,
+      "loss": 0.3603,
+      "step": 29580
+    },
+    {
+      "epoch": 5.820363403899584,
+      "grad_norm": 0.6827914714813232,
+      "learning_rate": 1.1155645235256352e-05,
+      "loss": 0.3915,
+      "step": 29590
+    },
+    {
+      "epoch": 5.8223304074156035,
+      "grad_norm": 1.3076701164245605,
+      "learning_rate": 1.1152587972729218e-05,
+      "loss": 0.3659,
+      "step": 29600
+    },
+    {
+      "epoch": 5.824297410931622,
+      "grad_norm": 1.4336459636688232,
+      "learning_rate": 1.1149530710202085e-05,
+      "loss": 0.3897,
+      "step": 29610
+    },
+    {
+      "epoch": 5.826264414447641,
+      "grad_norm": 2.1561131477355957,
+      "learning_rate": 1.1146473447674953e-05,
+      "loss": 0.5855,
+      "step": 29620
+    },
+    {
+      "epoch": 5.82823141796366,
+      "grad_norm": 1.179270625114441,
+      "learning_rate": 1.1143416185147819e-05,
+      "loss": 0.3869,
+      "step": 29630
+    },
+    {
+      "epoch": 5.830198421479678,
+      "grad_norm": 1.5720471143722534,
+      "learning_rate": 1.1140358922620686e-05,
+      "loss": 0.4048,
+      "step": 29640
+    },
+    {
+      "epoch": 5.832165424995697,
+      "grad_norm": 2.540081024169922,
+      "learning_rate": 1.1137301660093554e-05,
+      "loss": 0.398,
+      "step": 29650
+    },
+    {
+      "epoch": 5.834132428511716,
+      "grad_norm": 0.8215223550796509,
+      "learning_rate": 1.1134244397566421e-05,
+      "loss": 0.3492,
+      "step": 29660
+    },
+    {
+      "epoch": 5.836099432027734,
+      "grad_norm": 1.6602963209152222,
+      "learning_rate": 1.1131187135039287e-05,
+      "loss": 0.4209,
+      "step": 29670
+    },
+    {
+      "epoch": 5.8380664355437535,
+      "grad_norm": 1.6773931980133057,
+      "learning_rate": 1.1128129872512154e-05,
+      "loss": 0.3097,
+      "step": 29680
+    },
+    {
+      "epoch": 5.840033439059773,
+      "grad_norm": 1.333003044128418,
+      "learning_rate": 1.1125072609985018e-05,
+      "loss": 0.3265,
+      "step": 29690
+    },
+    {
+      "epoch": 5.842000442575791,
+      "grad_norm": 2.068171262741089,
+      "learning_rate": 1.1122015347457886e-05,
+      "loss": 0.26,
+      "step": 29700
+    },
+    {
+      "epoch": 5.84396744609181,
+      "grad_norm": 0.8129364252090454,
+      "learning_rate": 1.1118958084930753e-05,
+      "loss": 0.2799,
+      "step": 29710
+    },
+    {
+      "epoch": 5.845934449607829,
+      "grad_norm": 0.8760831952095032,
+      "learning_rate": 1.1115900822403621e-05,
+      "loss": 0.5451,
+      "step": 29720
+    },
+    {
+      "epoch": 5.847901453123847,
+      "grad_norm": 1.9979091882705688,
+      "learning_rate": 1.1112843559876487e-05,
+      "loss": 0.3071,
+      "step": 29730
+    },
+    {
+      "epoch": 5.849868456639866,
+      "grad_norm": 2.7981607913970947,
+      "learning_rate": 1.1109786297349354e-05,
+      "loss": 0.404,
+      "step": 29740
+    },
+    {
+      "epoch": 5.851835460155885,
+      "grad_norm": 0.3538930416107178,
+      "learning_rate": 1.1106729034822222e-05,
+      "loss": 0.3743,
+      "step": 29750
+    },
+    {
+      "epoch": 5.8538024636719035,
+      "grad_norm": 0.823993980884552,
+      "learning_rate": 1.1103671772295087e-05,
+      "loss": 0.298,
+      "step": 29760
+    },
+    {
+      "epoch": 5.855769467187923,
+      "grad_norm": 0.8794128894805908,
+      "learning_rate": 1.1100614509767955e-05,
+      "loss": 0.3656,
+      "step": 29770
+    },
+    {
+      "epoch": 5.857736470703942,
+      "grad_norm": 0.6292213797569275,
+      "learning_rate": 1.1097557247240822e-05,
+      "loss": 0.2991,
+      "step": 29780
+    },
+    {
+      "epoch": 5.85970347421996,
+      "grad_norm": 1.622949481010437,
+      "learning_rate": 1.109449998471369e-05,
+      "loss": 0.4715,
+      "step": 29790
+    },
+    {
+      "epoch": 5.861670477735979,
+      "grad_norm": 0.844419538974762,
+      "learning_rate": 1.1091442722186556e-05,
+      "loss": 0.3423,
+      "step": 29800
+    },
+    {
+      "epoch": 5.863637481251998,
+      "grad_norm": 0.8577235341072083,
+      "learning_rate": 1.1088385459659421e-05,
+      "loss": 0.3279,
+      "step": 29810
+    },
+    {
+      "epoch": 5.865604484768016,
+      "grad_norm": 1.3123620748519897,
+      "learning_rate": 1.1085328197132287e-05,
+      "loss": 0.3767,
+      "step": 29820
+    },
+    {
+      "epoch": 5.867571488284035,
+      "grad_norm": 1.5416537523269653,
+      "learning_rate": 1.1082270934605155e-05,
+      "loss": 0.3926,
+      "step": 29830
+    },
+    {
+      "epoch": 5.869538491800054,
+      "grad_norm": 1.229455590248108,
+      "learning_rate": 1.1079213672078022e-05,
+      "loss": 0.3863,
+      "step": 29840
+    },
+    {
+      "epoch": 5.871505495316073,
+      "grad_norm": 1.6253336668014526,
+      "learning_rate": 1.107615640955089e-05,
+      "loss": 0.4067,
+      "step": 29850
+    },
+    {
+      "epoch": 5.873472498832092,
+      "grad_norm": 1.830370306968689,
+      "learning_rate": 1.1073099147023755e-05,
+      "loss": 0.3894,
+      "step": 29860
+    },
+    {
+      "epoch": 5.875439502348111,
+      "grad_norm": 1.366328239440918,
+      "learning_rate": 1.1070041884496623e-05,
+      "loss": 0.2714,
+      "step": 29870
+    },
+    {
+      "epoch": 5.877406505864129,
+      "grad_norm": 0.6919177770614624,
+      "learning_rate": 1.106698462196949e-05,
+      "loss": 0.3454,
+      "step": 29880
+    },
+    {
+      "epoch": 5.879373509380148,
+      "grad_norm": 0.7361364364624023,
+      "learning_rate": 1.1063927359442356e-05,
+      "loss": 0.4565,
+      "step": 29890
+    },
+    {
+      "epoch": 5.881340512896167,
+      "grad_norm": 0.7951186299324036,
+      "learning_rate": 1.1060870096915224e-05,
+      "loss": 0.4556,
+      "step": 29900
+    },
+    {
+      "epoch": 5.883307516412185,
+      "grad_norm": 2.3393678665161133,
+      "learning_rate": 1.1057812834388091e-05,
+      "loss": 0.4803,
+      "step": 29910
+    },
+    {
+      "epoch": 5.885274519928204,
+      "grad_norm": 0.8017082810401917,
+      "learning_rate": 1.1054755571860955e-05,
+      "loss": 0.4325,
+      "step": 29920
+    },
+    {
+      "epoch": 5.8872415234442235,
+      "grad_norm": 2.544628143310547,
+      "learning_rate": 1.1051698309333823e-05,
+      "loss": 0.3766,
+      "step": 29930
+    },
+    {
+      "epoch": 5.889208526960242,
+      "grad_norm": 0.9725555777549744,
+      "learning_rate": 1.104864104680669e-05,
+      "loss": 0.3983,
+      "step": 29940
+    },
+    {
+      "epoch": 5.891175530476261,
+      "grad_norm": 0.8429176807403564,
+      "learning_rate": 1.1045583784279556e-05,
+      "loss": 0.3185,
+      "step": 29950
+    },
+    {
+      "epoch": 5.89314253399228,
+      "grad_norm": 1.1367381811141968,
+      "learning_rate": 1.1042526521752423e-05,
+      "loss": 0.2792,
+      "step": 29960
+    },
+    {
+      "epoch": 5.895109537508298,
+      "grad_norm": 1.9774765968322754,
+      "learning_rate": 1.103946925922529e-05,
+      "loss": 0.4607,
+      "step": 29970
+    },
+    {
+      "epoch": 5.897076541024317,
+      "grad_norm": 0.982170045375824,
+      "learning_rate": 1.1036411996698158e-05,
+      "loss": 0.3693,
+      "step": 29980
+    },
+    {
+      "epoch": 5.899043544540336,
+      "grad_norm": 0.7595378160476685,
+      "learning_rate": 1.1033354734171024e-05,
+      "loss": 0.2639,
+      "step": 29990
+    },
+    {
+      "epoch": 5.901010548056354,
+      "grad_norm": 1.9132200479507446,
+      "learning_rate": 1.1030297471643891e-05,
+      "loss": 0.2905,
+      "step": 30000
+    },
+    {
+      "epoch": 5.901010548056354,
+      "eval_loss": 0.1469283252954483,
+      "eval_runtime": 8.9116,
+      "eval_samples_per_second": 5.611,
+      "eval_steps_per_second": 2.805,
+      "step": 30000
+    },
+    {
+      "epoch": 5.9029775515723735,
+      "grad_norm": 1.2966291904449463,
+      "learning_rate": 1.1027240209116759e-05,
+      "loss": 0.2171,
+      "step": 30010
+    },
+    {
+      "epoch": 5.9049445550883926,
+      "grad_norm": 1.0475329160690308,
+      "learning_rate": 1.1024182946589625e-05,
+      "loss": 0.3754,
+      "step": 30020
+    },
+    {
+      "epoch": 5.906911558604411,
+      "grad_norm": 1.6951876878738403,
+      "learning_rate": 1.102112568406249e-05,
+      "loss": 0.4807,
+      "step": 30030
+    },
+    {
+      "epoch": 5.90887856212043,
+      "grad_norm": 1.9632104635238647,
+      "learning_rate": 1.1018068421535358e-05,
+      "loss": 0.339,
+      "step": 30040
+    },
+    {
+      "epoch": 5.910845565636449,
+      "grad_norm": 1.9631885290145874,
+      "learning_rate": 1.1015011159008224e-05,
+      "loss": 0.3866,
+      "step": 30050
+    },
+    {
+      "epoch": 5.912812569152467,
+      "grad_norm": 1.97544264793396,
+      "learning_rate": 1.1011953896481091e-05,
+      "loss": 0.3408,
+      "step": 30060
+    },
+    {
+      "epoch": 5.914779572668486,
+      "grad_norm": 1.4105381965637207,
+      "learning_rate": 1.1008896633953959e-05,
+      "loss": 0.4061,
+      "step": 30070
+    },
+    {
+      "epoch": 5.916746576184505,
+      "grad_norm": 1.1220061779022217,
+      "learning_rate": 1.1005839371426824e-05,
+      "loss": 0.4103,
+      "step": 30080
+    },
+    {
+      "epoch": 5.9187135797005235,
+      "grad_norm": 0.9747940897941589,
+      "learning_rate": 1.1002782108899692e-05,
+      "loss": 0.3795,
+      "step": 30090
+    },
+    {
+      "epoch": 5.920680583216543,
+      "grad_norm": 1.0602132081985474,
+      "learning_rate": 1.099972484637256e-05,
+      "loss": 0.3018,
+      "step": 30100
+    },
+    {
+      "epoch": 5.922647586732562,
+      "grad_norm": 1.2885088920593262,
+      "learning_rate": 1.0996667583845427e-05,
+      "loss": 0.4286,
+      "step": 30110
+    },
+    {
+      "epoch": 5.92461459024858,
+      "grad_norm": 0.9011586904525757,
+      "learning_rate": 1.0993610321318293e-05,
+      "loss": 0.393,
+      "step": 30120
+    },
+    {
+      "epoch": 5.926581593764599,
+      "grad_norm": 1.460485816001892,
+      "learning_rate": 1.099055305879116e-05,
+      "loss": 0.3156,
+      "step": 30130
+    },
+    {
+      "epoch": 5.928548597280618,
+      "grad_norm": 0.9864498972892761,
+      "learning_rate": 1.0987495796264028e-05,
+      "loss": 0.5503,
+      "step": 30140
+    },
+    {
+      "epoch": 5.930515600796636,
+      "grad_norm": 1.3364967107772827,
+      "learning_rate": 1.0984438533736892e-05,
+      "loss": 0.3924,
+      "step": 30150
+    },
+    {
+      "epoch": 5.932482604312655,
+      "grad_norm": 1.594724416732788,
+      "learning_rate": 1.098138127120976e-05,
+      "loss": 0.3058,
+      "step": 30160
+    },
+    {
+      "epoch": 5.934449607828674,
+      "grad_norm": 1.7341972589492798,
+      "learning_rate": 1.0978324008682627e-05,
+      "loss": 0.5161,
+      "step": 30170
+    },
+    {
+      "epoch": 5.936416611344693,
+      "grad_norm": 1.722612977027893,
+      "learning_rate": 1.0975266746155492e-05,
+      "loss": 0.4114,
+      "step": 30180
+    },
+    {
+      "epoch": 5.938383614860712,
+      "grad_norm": 2.482858419418335,
+      "learning_rate": 1.097220948362836e-05,
+      "loss": 0.3087,
+      "step": 30190
+    },
+    {
+      "epoch": 5.940350618376731,
+      "grad_norm": 1.0486516952514648,
+      "learning_rate": 1.0969152221101227e-05,
+      "loss": 0.3556,
+      "step": 30200
+    },
+    {
+      "epoch": 5.942317621892749,
+      "grad_norm": 1.137136697769165,
+      "learning_rate": 1.0966094958574093e-05,
+      "loss": 0.4288,
+      "step": 30210
+    },
+    {
+      "epoch": 5.944284625408768,
+      "grad_norm": 1.8614591360092163,
+      "learning_rate": 1.096303769604696e-05,
+      "loss": 0.3394,
+      "step": 30220
+    },
+    {
+      "epoch": 5.946251628924787,
+      "grad_norm": 1.0514382123947144,
+      "learning_rate": 1.0959980433519828e-05,
+      "loss": 0.4297,
+      "step": 30230
+    },
+    {
+      "epoch": 5.948218632440805,
+      "grad_norm": 0.6810349822044373,
+      "learning_rate": 1.0956923170992696e-05,
+      "loss": 0.3954,
+      "step": 30240
+    },
+    {
+      "epoch": 5.950185635956824,
+      "grad_norm": 1.3290073871612549,
+      "learning_rate": 1.0953865908465561e-05,
+      "loss": 0.3973,
+      "step": 30250
+    },
+    {
+      "epoch": 5.952152639472843,
+      "grad_norm": 1.5532108545303345,
+      "learning_rate": 1.0950808645938427e-05,
+      "loss": 0.3683,
+      "step": 30260
+    },
+    {
+      "epoch": 5.954119642988862,
+      "grad_norm": 0.6228435635566711,
+      "learning_rate": 1.0947751383411293e-05,
+      "loss": 0.4073,
+      "step": 30270
+    },
+    {
+      "epoch": 5.956086646504881,
+      "grad_norm": 1.2247480154037476,
+      "learning_rate": 1.094469412088416e-05,
+      "loss": 0.4175,
+      "step": 30280
+    },
+    {
+      "epoch": 5.958053650020899,
+      "grad_norm": 1.8101277351379395,
+      "learning_rate": 1.0941636858357028e-05,
+      "loss": 0.3097,
+      "step": 30290
+    },
+    {
+      "epoch": 5.960020653536918,
+      "grad_norm": 0.718777596950531,
+      "learning_rate": 1.0938579595829895e-05,
+      "loss": 0.2533,
+      "step": 30300
+    },
+    {
+      "epoch": 5.961987657052937,
+      "grad_norm": 0.6376405954360962,
+      "learning_rate": 1.0935522333302761e-05,
+      "loss": 0.3094,
+      "step": 30310
+    },
+    {
+      "epoch": 5.963954660568955,
+      "grad_norm": 1.3167697191238403,
+      "learning_rate": 1.0932465070775629e-05,
+      "loss": 0.3132,
+      "step": 30320
+    },
+    {
+      "epoch": 5.965921664084974,
+      "grad_norm": 1.6995993852615356,
+      "learning_rate": 1.0929407808248496e-05,
+      "loss": 0.3559,
+      "step": 30330
+    },
+    {
+      "epoch": 5.9678886676009935,
+      "grad_norm": 1.4039970636367798,
+      "learning_rate": 1.0926350545721362e-05,
+      "loss": 0.3164,
+      "step": 30340
+    },
+    {
+      "epoch": 5.969855671117012,
+      "grad_norm": 2.015141487121582,
+      "learning_rate": 1.092329328319423e-05,
+      "loss": 0.3403,
+      "step": 30350
+    },
+    {
+      "epoch": 5.971822674633031,
+      "grad_norm": 1.9489538669586182,
+      "learning_rate": 1.0920236020667097e-05,
+      "loss": 0.4236,
+      "step": 30360
+    },
+    {
+      "epoch": 5.97378967814905,
+      "grad_norm": 1.7877446413040161,
+      "learning_rate": 1.091717875813996e-05,
+      "loss": 0.3764,
+      "step": 30370
+    },
+    {
+      "epoch": 5.975756681665068,
+      "grad_norm": 1.0314801931381226,
+      "learning_rate": 1.0914121495612828e-05,
+      "loss": 0.3748,
+      "step": 30380
+    },
+    {
+      "epoch": 5.977723685181087,
+      "grad_norm": 0.8923133611679077,
+      "learning_rate": 1.0911064233085696e-05,
+      "loss": 0.3876,
+      "step": 30390
+    },
+    {
+      "epoch": 5.979690688697106,
+      "grad_norm": 0.8820732831954956,
+      "learning_rate": 1.0908006970558562e-05,
+      "loss": 0.3788,
+      "step": 30400
+    },
+    {
+      "epoch": 5.981657692213124,
+      "grad_norm": 2.1888816356658936,
+      "learning_rate": 1.0904949708031429e-05,
+      "loss": 0.4475,
+      "step": 30410
+    },
+    {
+      "epoch": 5.9836246957291435,
+      "grad_norm": 1.369350790977478,
+      "learning_rate": 1.0901892445504297e-05,
+      "loss": 0.4265,
+      "step": 30420
+    },
+    {
+      "epoch": 5.9855916992451625,
+      "grad_norm": 0.6697467565536499,
+      "learning_rate": 1.0898835182977164e-05,
+      "loss": 0.4157,
+      "step": 30430
+    },
+    {
+      "epoch": 5.987558702761181,
+      "grad_norm": 1.720997929573059,
+      "learning_rate": 1.089577792045003e-05,
+      "loss": 0.4813,
+      "step": 30440
+    },
+    {
+      "epoch": 5.9895257062772,
+      "grad_norm": 0.8081413507461548,
+      "learning_rate": 1.0892720657922897e-05,
+      "loss": 0.4658,
+      "step": 30450
+    },
+    {
+      "epoch": 5.991492709793219,
+      "grad_norm": 1.599493384361267,
+      "learning_rate": 1.0889663395395765e-05,
+      "loss": 0.3787,
+      "step": 30460
+    },
+    {
+      "epoch": 5.993459713309237,
+      "grad_norm": 1.8009963035583496,
+      "learning_rate": 1.088660613286863e-05,
+      "loss": 0.3958,
+      "step": 30470
+    },
+    {
+      "epoch": 5.995426716825256,
+      "grad_norm": 1.6714438199996948,
+      "learning_rate": 1.0883548870341498e-05,
+      "loss": 0.3767,
+      "step": 30480
+    },
+    {
+      "epoch": 5.997393720341275,
+      "grad_norm": 0.8772445321083069,
+      "learning_rate": 1.0880491607814364e-05,
+      "loss": 0.3406,
+      "step": 30490
+    },
+    {
+      "epoch": 5.9993607238572935,
+      "grad_norm": 1.883542776107788,
+      "learning_rate": 1.087743434528723e-05,
+      "loss": 0.2961,
+      "step": 30500
+    },
+    {
+      "epoch": 5.9993607238572935,
+      "eval_loss": 0.14722052216529846,
+      "eval_runtime": 8.9135,
+      "eval_samples_per_second": 5.609,
+      "eval_steps_per_second": 2.805,
+      "step": 30500
+    },
+    {
+      "epoch": 6.0013277273733125,
+      "grad_norm": 2.056985378265381,
+      "learning_rate": 1.0874377082760097e-05,
+      "loss": 0.3812,
+      "step": 30510
+    },
+    {
+      "epoch": 6.003294730889332,
+      "grad_norm": 1.3648536205291748,
+      "learning_rate": 1.0871319820232964e-05,
+      "loss": 0.3523,
+      "step": 30520
+    },
+    {
+      "epoch": 6.00526173440535,
+      "grad_norm": 1.4668800830841064,
+      "learning_rate": 1.086826255770583e-05,
+      "loss": 0.3414,
+      "step": 30530
+    },
+    {
+      "epoch": 6.007228737921369,
+      "grad_norm": 1.6990594863891602,
+      "learning_rate": 1.0865205295178698e-05,
+      "loss": 0.4042,
+      "step": 30540
+    },
+    {
+      "epoch": 6.009195741437388,
+      "grad_norm": 1.3983603715896606,
+      "learning_rate": 1.0862148032651565e-05,
+      "loss": 0.3461,
+      "step": 30550
+    },
+    {
+      "epoch": 6.011162744953406,
+      "grad_norm": 0.9408298134803772,
+      "learning_rate": 1.0859090770124433e-05,
+      "loss": 0.2446,
+      "step": 30560
+    },
+    {
+      "epoch": 6.013129748469425,
+      "grad_norm": 0.7286332845687866,
+      "learning_rate": 1.0856033507597298e-05,
+      "loss": 0.4312,
+      "step": 30570
+    },
+    {
+      "epoch": 6.015096751985444,
+      "grad_norm": 0.896022617816925,
+      "learning_rate": 1.0852976245070166e-05,
+      "loss": 0.297,
+      "step": 30580
+    },
+    {
+      "epoch": 6.0170637555014626,
+      "grad_norm": 1.477001667022705,
+      "learning_rate": 1.0849918982543033e-05,
+      "loss": 0.4265,
+      "step": 30590
+    },
+    {
+      "epoch": 6.019030759017482,
+      "grad_norm": 1.0923590660095215,
+      "learning_rate": 1.0846861720015897e-05,
+      "loss": 0.3933,
+      "step": 30600
+    },
+    {
+      "epoch": 6.020997762533501,
+      "grad_norm": 0.9560948610305786,
+      "learning_rate": 1.0843804457488765e-05,
+      "loss": 0.3813,
+      "step": 30610
+    },
+    {
+      "epoch": 6.022964766049519,
+      "grad_norm": 1.1864969730377197,
+      "learning_rate": 1.0840747194961632e-05,
+      "loss": 0.3759,
+      "step": 30620
+    },
+    {
+      "epoch": 6.024931769565538,
+      "grad_norm": 0.9665082097053528,
+      "learning_rate": 1.0837689932434498e-05,
+      "loss": 0.3564,
+      "step": 30630
+    },
+    {
+      "epoch": 6.026898773081557,
+      "grad_norm": 1.4572361707687378,
+      "learning_rate": 1.0834632669907366e-05,
+      "loss": 0.3752,
+      "step": 30640
+    },
+    {
+      "epoch": 6.028865776597575,
+      "grad_norm": 1.0030657052993774,
+      "learning_rate": 1.0831575407380233e-05,
+      "loss": 0.2882,
+      "step": 30650
+    },
+    {
+      "epoch": 6.030832780113594,
+      "grad_norm": 0.9930024147033691,
+      "learning_rate": 1.0828518144853099e-05,
+      "loss": 0.3293,
+      "step": 30660
+    },
+    {
+      "epoch": 6.032799783629613,
+      "grad_norm": 2.801129102706909,
+      "learning_rate": 1.0825460882325966e-05,
+      "loss": 0.4869,
+      "step": 30670
+    },
+    {
+      "epoch": 6.034766787145632,
+      "grad_norm": 1.1921310424804688,
+      "learning_rate": 1.0822403619798834e-05,
+      "loss": 0.3916,
+      "step": 30680
+    },
+    {
+      "epoch": 6.036733790661651,
+      "grad_norm": 0.4918254017829895,
+      "learning_rate": 1.0819346357271701e-05,
+      "loss": 0.3047,
+      "step": 30690
+    },
+    {
+      "epoch": 6.03870079417767,
+      "grad_norm": 0.8728923797607422,
+      "learning_rate": 1.0816289094744567e-05,
+      "loss": 0.3435,
+      "step": 30700
+    },
+    {
+      "epoch": 6.040667797693688,
+      "grad_norm": 1.234466791152954,
+      "learning_rate": 1.0813231832217433e-05,
+      "loss": 0.5229,
+      "step": 30710
+    },
+    {
+      "epoch": 6.042634801209707,
+      "grad_norm": 4.090738296508789,
+      "learning_rate": 1.0810174569690299e-05,
+      "loss": 0.3178,
+      "step": 30720
+    },
+    {
+      "epoch": 6.044601804725726,
+      "grad_norm": 2.3472020626068115,
+      "learning_rate": 1.0807117307163166e-05,
+      "loss": 0.305,
+      "step": 30730
+    },
+    {
+      "epoch": 6.046568808241744,
+      "grad_norm": 1.0088740587234497,
+      "learning_rate": 1.0804060044636034e-05,
+      "loss": 0.3439,
+      "step": 30740
+    },
+    {
+      "epoch": 6.0485358117577634,
+      "grad_norm": 1.0725480318069458,
+      "learning_rate": 1.0801002782108901e-05,
+      "loss": 0.3794,
+      "step": 30750
+    },
+    {
+      "epoch": 6.0505028152737825,
+      "grad_norm": 0.7405242919921875,
+      "learning_rate": 1.0797945519581767e-05,
+      "loss": 0.3402,
+      "step": 30760
+    },
+    {
+      "epoch": 6.052469818789801,
+      "grad_norm": 1.314536690711975,
+      "learning_rate": 1.0794888257054634e-05,
+      "loss": 0.3956,
+      "step": 30770
+    },
+    {
+      "epoch": 6.05443682230582,
+      "grad_norm": 0.7027092576026917,
+      "learning_rate": 1.0791830994527502e-05,
+      "loss": 0.3186,
+      "step": 30780
+    },
+    {
+      "epoch": 6.056403825821839,
+      "grad_norm": 1.0569257736206055,
+      "learning_rate": 1.0788773732000368e-05,
+      "loss": 0.4329,
+      "step": 30790
+    },
+    {
+      "epoch": 6.058370829337857,
+      "grad_norm": 1.2603479623794556,
+      "learning_rate": 1.0785716469473235e-05,
+      "loss": 0.3964,
+      "step": 30800
+    },
+    {
+      "epoch": 6.060337832853876,
+      "grad_norm": 1.2180461883544922,
+      "learning_rate": 1.0782659206946103e-05,
+      "loss": 0.3594,
+      "step": 30810
+    },
+    {
+      "epoch": 6.062304836369895,
+      "grad_norm": 1.306077003479004,
+      "learning_rate": 1.077960194441897e-05,
+      "loss": 0.4708,
+      "step": 30820
+    },
+    {
+      "epoch": 6.0642718398859135,
+      "grad_norm": 1.0077178478240967,
+      "learning_rate": 1.0776544681891834e-05,
+      "loss": 0.3159,
+      "step": 30830
+    },
+    {
+      "epoch": 6.0662388434019325,
+      "grad_norm": 0.9312611818313599,
+      "learning_rate": 1.0773487419364702e-05,
+      "loss": 0.3526,
+      "step": 30840
+    },
+    {
+      "epoch": 6.068205846917952,
+      "grad_norm": 0.8618937134742737,
+      "learning_rate": 1.0770430156837567e-05,
+      "loss": 0.3417,
+      "step": 30850
+    },
+    {
+      "epoch": 6.07017285043397,
+      "grad_norm": 0.825440526008606,
+      "learning_rate": 1.0767372894310435e-05,
+      "loss": 0.3364,
+      "step": 30860
+    },
+    {
+      "epoch": 6.072139853949989,
+      "grad_norm": 0.8122813105583191,
+      "learning_rate": 1.0764315631783302e-05,
+      "loss": 0.421,
+      "step": 30870
+    },
+    {
+      "epoch": 6.074106857466008,
+      "grad_norm": 1.007061243057251,
+      "learning_rate": 1.076125836925617e-05,
+      "loss": 0.341,
+      "step": 30880
+    },
+    {
+      "epoch": 6.076073860982026,
+      "grad_norm": 0.4585067629814148,
+      "learning_rate": 1.0758201106729035e-05,
+      "loss": 0.3691,
+      "step": 30890
+    },
+    {
+      "epoch": 6.078040864498045,
+      "grad_norm": 0.7221927046775818,
+      "learning_rate": 1.0755143844201903e-05,
+      "loss": 0.3342,
+      "step": 30900
+    },
+    {
+      "epoch": 6.080007868014064,
+      "grad_norm": 1.0165441036224365,
+      "learning_rate": 1.075208658167477e-05,
+      "loss": 0.477,
+      "step": 30910
+    },
+    {
+      "epoch": 6.0819748715300825,
+      "grad_norm": 1.446653962135315,
+      "learning_rate": 1.0749029319147636e-05,
+      "loss": 0.3424,
+      "step": 30920
+    },
+    {
+      "epoch": 6.083941875046102,
+      "grad_norm": 0.7181447148323059,
+      "learning_rate": 1.0745972056620504e-05,
+      "loss": 0.2247,
+      "step": 30930
+    },
+    {
+      "epoch": 6.085908878562121,
+      "grad_norm": 0.5309480428695679,
+      "learning_rate": 1.074291479409337e-05,
+      "loss": 0.4787,
+      "step": 30940
+    },
+    {
+      "epoch": 6.087875882078139,
+      "grad_norm": 0.8217592835426331,
+      "learning_rate": 1.0739857531566235e-05,
+      "loss": 0.3421,
+      "step": 30950
+    },
+    {
+      "epoch": 6.089842885594158,
+      "grad_norm": 1.1308156251907349,
+      "learning_rate": 1.0736800269039103e-05,
+      "loss": 0.3263,
+      "step": 30960
+    },
+    {
+      "epoch": 6.091809889110177,
+      "grad_norm": 0.877742350101471,
+      "learning_rate": 1.073374300651197e-05,
+      "loss": 0.3078,
+      "step": 30970
+    },
+    {
+      "epoch": 6.093776892626195,
+      "grad_norm": 1.2607346773147583,
+      "learning_rate": 1.0730685743984836e-05,
+      "loss": 0.51,
+      "step": 30980
+    },
+    {
+      "epoch": 6.095743896142214,
+      "grad_norm": 1.0867124795913696,
+      "learning_rate": 1.0727628481457703e-05,
+      "loss": 0.2541,
+      "step": 30990
+    },
+    {
+      "epoch": 6.097710899658233,
+      "grad_norm": 1.2764980792999268,
+      "learning_rate": 1.0724571218930571e-05,
+      "loss": 0.3536,
+      "step": 31000
+    },
+    {
+      "epoch": 6.097710899658233,
+      "eval_loss": 0.14491592347621918,
+      "eval_runtime": 8.8633,
+      "eval_samples_per_second": 5.641,
+      "eval_steps_per_second": 2.821,
+      "step": 31000
+    },
+    {
+      "epoch": 6.099677903174252,
+      "grad_norm": 0.7683907747268677,
+      "learning_rate": 1.0721513956403438e-05,
+      "loss": 0.3945,
+      "step": 31010
+    },
+    {
+      "epoch": 6.101644906690271,
+      "grad_norm": 1.1609355211257935,
+      "learning_rate": 1.0718456693876304e-05,
+      "loss": 0.337,
+      "step": 31020
+    },
+    {
+      "epoch": 6.10361191020629,
+      "grad_norm": 0.5849171876907349,
+      "learning_rate": 1.0715399431349172e-05,
+      "loss": 0.3,
+      "step": 31030
+    },
+    {
+      "epoch": 6.105578913722308,
+      "grad_norm": 1.1877692937850952,
+      "learning_rate": 1.0712342168822039e-05,
+      "loss": 0.3381,
+      "step": 31040
+    },
+    {
+      "epoch": 6.107545917238327,
+      "grad_norm": 0.8901480436325073,
+      "learning_rate": 1.0709284906294903e-05,
+      "loss": 0.4137,
+      "step": 31050
+    },
+    {
+      "epoch": 6.109512920754346,
+      "grad_norm": 1.0427314043045044,
+      "learning_rate": 1.070622764376777e-05,
+      "loss": 0.2717,
+      "step": 31060
+    },
+    {
+      "epoch": 6.111479924270364,
+      "grad_norm": 1.543684959411621,
+      "learning_rate": 1.0703170381240638e-05,
+      "loss": 0.3706,
+      "step": 31070
+    },
+    {
+      "epoch": 6.113446927786383,
+      "grad_norm": 1.26493501663208,
+      "learning_rate": 1.0700113118713504e-05,
+      "loss": 0.3729,
+      "step": 31080
+    },
+    {
+      "epoch": 6.1154139313024025,
+      "grad_norm": 1.4996311664581299,
+      "learning_rate": 1.0697055856186371e-05,
+      "loss": 0.328,
+      "step": 31090
+    },
+    {
+      "epoch": 6.117380934818421,
+      "grad_norm": 1.4721986055374146,
+      "learning_rate": 1.0693998593659239e-05,
+      "loss": 0.3299,
+      "step": 31100
+    },
+    {
+      "epoch": 6.11934793833444,
+      "grad_norm": 3.207667350769043,
+      "learning_rate": 1.0690941331132105e-05,
+      "loss": 0.2632,
+      "step": 31110
+    },
+    {
+      "epoch": 6.121314941850459,
+      "grad_norm": 0.6640446782112122,
+      "learning_rate": 1.0687884068604972e-05,
+      "loss": 0.3463,
+      "step": 31120
+    },
+    {
+      "epoch": 6.123281945366477,
+      "grad_norm": 1.1663861274719238,
+      "learning_rate": 1.068482680607784e-05,
+      "loss": 0.3626,
+      "step": 31130
+    },
+    {
+      "epoch": 6.125248948882496,
+      "grad_norm": 1.0878609418869019,
+      "learning_rate": 1.0681769543550707e-05,
+      "loss": 0.3306,
+      "step": 31140
+    },
+    {
+      "epoch": 6.127215952398515,
+      "grad_norm": 1.0775517225265503,
+      "learning_rate": 1.0678712281023573e-05,
+      "loss": 0.3593,
+      "step": 31150
+    },
+    {
+      "epoch": 6.129182955914533,
+      "grad_norm": 1.3499873876571655,
+      "learning_rate": 1.067565501849644e-05,
+      "loss": 0.303,
+      "step": 31160
+    },
+    {
+      "epoch": 6.1311499594305525,
+      "grad_norm": 1.3294696807861328,
+      "learning_rate": 1.0672597755969304e-05,
+      "loss": 0.365,
+      "step": 31170
+    },
+    {
+      "epoch": 6.133116962946572,
+      "grad_norm": 0.8494240641593933,
+      "learning_rate": 1.0669540493442172e-05,
+      "loss": 0.3471,
+      "step": 31180
+    },
+    {
+      "epoch": 6.13508396646259,
+      "grad_norm": 2.2927815914154053,
+      "learning_rate": 1.066648323091504e-05,
+      "loss": 0.3805,
+      "step": 31190
+    },
+    {
+      "epoch": 6.137050969978609,
+      "grad_norm": 1.4000989198684692,
+      "learning_rate": 1.0663425968387907e-05,
+      "loss": 0.2334,
+      "step": 31200
+    },
+    {
+      "epoch": 6.139017973494628,
+      "grad_norm": 0.920910656452179,
+      "learning_rate": 1.0660368705860773e-05,
+      "loss": 0.2703,
+      "step": 31210
+    },
+    {
+      "epoch": 6.140984977010646,
+      "grad_norm": 0.863723874092102,
+      "learning_rate": 1.065731144333364e-05,
+      "loss": 0.4328,
+      "step": 31220
+    },
+    {
+      "epoch": 6.142951980526665,
+      "grad_norm": 1.2679826021194458,
+      "learning_rate": 1.0654254180806508e-05,
+      "loss": 0.3987,
+      "step": 31230
+    },
+    {
+      "epoch": 6.144918984042684,
+      "grad_norm": 1.067123293876648,
+      "learning_rate": 1.0651196918279373e-05,
+      "loss": 0.3339,
+      "step": 31240
+    },
+    {
+      "epoch": 6.1468859875587025,
+      "grad_norm": 1.1733587980270386,
+      "learning_rate": 1.064813965575224e-05,
+      "loss": 0.2214,
+      "step": 31250
+    },
+    {
+      "epoch": 6.148852991074722,
+      "grad_norm": 1.0510685443878174,
+      "learning_rate": 1.0645082393225108e-05,
+      "loss": 0.3718,
+      "step": 31260
+    },
+    {
+      "epoch": 6.150819994590741,
+      "grad_norm": 1.1614587306976318,
+      "learning_rate": 1.0642025130697976e-05,
+      "loss": 0.3971,
+      "step": 31270
+    },
+    {
+      "epoch": 6.152786998106759,
+      "grad_norm": 1.8867449760437012,
+      "learning_rate": 1.063896786817084e-05,
+      "loss": 0.3666,
+      "step": 31280
+    },
+    {
+      "epoch": 6.154754001622778,
+      "grad_norm": 1.8797495365142822,
+      "learning_rate": 1.0635910605643707e-05,
+      "loss": 0.3763,
+      "step": 31290
+    },
+    {
+      "epoch": 6.156721005138797,
+      "grad_norm": 1.5775450468063354,
+      "learning_rate": 1.0632853343116573e-05,
+      "loss": 0.4743,
+      "step": 31300
+    },
+    {
+      "epoch": 6.158688008654815,
+      "grad_norm": 1.456451416015625,
+      "learning_rate": 1.062979608058944e-05,
+      "loss": 0.4399,
+      "step": 31310
+    },
+    {
+      "epoch": 6.160655012170834,
+      "grad_norm": 1.745969533920288,
+      "learning_rate": 1.0626738818062308e-05,
+      "loss": 0.2838,
+      "step": 31320
+    },
+    {
+      "epoch": 6.162622015686853,
+      "grad_norm": 1.225696325302124,
+      "learning_rate": 1.0623681555535175e-05,
+      "loss": 0.3019,
+      "step": 31330
+    },
+    {
+      "epoch": 6.164589019202872,
+      "grad_norm": 1.4007424116134644,
+      "learning_rate": 1.0620624293008041e-05,
+      "loss": 0.4124,
+      "step": 31340
+    },
+    {
+      "epoch": 6.166556022718891,
+      "grad_norm": 0.6728320717811584,
+      "learning_rate": 1.0617567030480909e-05,
+      "loss": 0.4112,
+      "step": 31350
+    },
+    {
+      "epoch": 6.16852302623491,
+      "grad_norm": 0.8944595456123352,
+      "learning_rate": 1.0614509767953776e-05,
+      "loss": 0.3456,
+      "step": 31360
+    },
+    {
+      "epoch": 6.170490029750928,
+      "grad_norm": 1.4572259187698364,
+      "learning_rate": 1.0611452505426642e-05,
+      "loss": 0.4435,
+      "step": 31370
+    },
+    {
+      "epoch": 6.172457033266947,
+      "grad_norm": 2.9878594875335693,
+      "learning_rate": 1.060839524289951e-05,
+      "loss": 0.4064,
+      "step": 31380
+    },
+    {
+      "epoch": 6.174424036782966,
+      "grad_norm": 0.9966301321983337,
+      "learning_rate": 1.0605337980372375e-05,
+      "loss": 0.3786,
+      "step": 31390
+    },
+    {
+      "epoch": 6.176391040298984,
+      "grad_norm": 0.930111289024353,
+      "learning_rate": 1.0602280717845241e-05,
+      "loss": 0.4371,
+      "step": 31400
+    },
+    {
+      "epoch": 6.178358043815003,
+      "grad_norm": 1.0499768257141113,
+      "learning_rate": 1.0599223455318108e-05,
+      "loss": 0.3302,
+      "step": 31410
+    },
+    {
+      "epoch": 6.1803250473310225,
+      "grad_norm": 1.316510796546936,
+      "learning_rate": 1.0596166192790976e-05,
+      "loss": 0.3829,
+      "step": 31420
+    },
+    {
+      "epoch": 6.182292050847041,
+      "grad_norm": 1.1173617839813232,
+      "learning_rate": 1.0593108930263842e-05,
+      "loss": 0.4753,
+      "step": 31430
+    },
+    {
+      "epoch": 6.18425905436306,
+      "grad_norm": 1.035115122795105,
+      "learning_rate": 1.059005166773671e-05,
+      "loss": 0.3719,
+      "step": 31440
+    },
+    {
+      "epoch": 6.186226057879079,
+      "grad_norm": 1.8233667612075806,
+      "learning_rate": 1.0586994405209577e-05,
+      "loss": 0.4091,
+      "step": 31450
+    },
+    {
+      "epoch": 6.188193061395097,
+      "grad_norm": 1.4853065013885498,
+      "learning_rate": 1.0583937142682444e-05,
+      "loss": 0.3037,
+      "step": 31460
+    },
+    {
+      "epoch": 6.190160064911116,
+      "grad_norm": 1.442393183708191,
+      "learning_rate": 1.058087988015531e-05,
+      "loss": 0.4173,
+      "step": 31470
+    },
+    {
+      "epoch": 6.192127068427135,
+      "grad_norm": 0.843182384967804,
+      "learning_rate": 1.0577822617628177e-05,
+      "loss": 0.4116,
+      "step": 31480
+    },
+    {
+      "epoch": 6.194094071943153,
+      "grad_norm": 1.1940809488296509,
+      "learning_rate": 1.0574765355101045e-05,
+      "loss": 0.3755,
+      "step": 31490
+    },
+    {
+      "epoch": 6.1960610754591725,
+      "grad_norm": 1.3931796550750732,
+      "learning_rate": 1.057170809257391e-05,
+      "loss": 0.279,
+      "step": 31500
+    },
+    {
+      "epoch": 6.1960610754591725,
+      "eval_loss": 0.14727434515953064,
+      "eval_runtime": 8.9002,
+      "eval_samples_per_second": 5.618,
+      "eval_steps_per_second": 2.809,
+      "step": 31500
+    },
+    {
+      "epoch": 6.198028078975192,
+      "grad_norm": 1.3128204345703125,
+      "learning_rate": 1.0568650830046776e-05,
+      "loss": 0.3738,
+      "step": 31510
+    },
+    {
+      "epoch": 6.19999508249121,
+      "grad_norm": 1.0435333251953125,
+      "learning_rate": 1.0565593567519644e-05,
+      "loss": 0.3621,
+      "step": 31520
+    },
+    {
+      "epoch": 6.201962086007229,
+      "grad_norm": 1.0718704462051392,
+      "learning_rate": 1.056253630499251e-05,
+      "loss": 0.4154,
+      "step": 31530
+    },
+    {
+      "epoch": 6.203929089523248,
+      "grad_norm": 1.5530494451522827,
+      "learning_rate": 1.0559479042465377e-05,
+      "loss": 0.3527,
+      "step": 31540
+    },
+    {
+      "epoch": 6.205896093039266,
+      "grad_norm": 1.1111003160476685,
+      "learning_rate": 1.0556421779938245e-05,
+      "loss": 0.3764,
+      "step": 31550
+    },
+    {
+      "epoch": 6.207863096555285,
+      "grad_norm": 1.1496018171310425,
+      "learning_rate": 1.055336451741111e-05,
+      "loss": 0.2756,
+      "step": 31560
+    },
+    {
+      "epoch": 6.209830100071304,
+      "grad_norm": 1.0827871561050415,
+      "learning_rate": 1.0550307254883978e-05,
+      "loss": 0.2736,
+      "step": 31570
+    },
+    {
+      "epoch": 6.2117971035873225,
+      "grad_norm": 1.5268453359603882,
+      "learning_rate": 1.0547249992356845e-05,
+      "loss": 0.4035,
+      "step": 31580
+    },
+    {
+      "epoch": 6.213764107103342,
+      "grad_norm": 2.096116781234741,
+      "learning_rate": 1.0544192729829713e-05,
+      "loss": 0.3846,
+      "step": 31590
+    },
+    {
+      "epoch": 6.215731110619361,
+      "grad_norm": 1.0222452878952026,
+      "learning_rate": 1.0541135467302579e-05,
+      "loss": 0.4076,
+      "step": 31600
+    },
+    {
+      "epoch": 6.217698114135379,
+      "grad_norm": 1.360022783279419,
+      "learning_rate": 1.0538078204775446e-05,
+      "loss": 0.3092,
+      "step": 31610
+    },
+    {
+      "epoch": 6.219665117651398,
+      "grad_norm": 0.7893637418746948,
+      "learning_rate": 1.053502094224831e-05,
+      "loss": 0.4069,
+      "step": 31620
+    },
+    {
+      "epoch": 6.221632121167417,
+      "grad_norm": 1.6887398958206177,
+      "learning_rate": 1.0531963679721178e-05,
+      "loss": 0.4277,
+      "step": 31630
+    },
+    {
+      "epoch": 6.223599124683435,
+      "grad_norm": 1.5129915475845337,
+      "learning_rate": 1.0528906417194045e-05,
+      "loss": 0.3649,
+      "step": 31640
+    },
+    {
+      "epoch": 6.225566128199454,
+      "grad_norm": 1.1816556453704834,
+      "learning_rate": 1.0525849154666913e-05,
+      "loss": 0.3348,
+      "step": 31650
+    },
+    {
+      "epoch": 6.2275331317154725,
+      "grad_norm": 1.3360129594802856,
+      "learning_rate": 1.0522791892139778e-05,
+      "loss": 0.3609,
+      "step": 31660
+    },
+    {
+      "epoch": 6.229500135231492,
+      "grad_norm": 1.489035964012146,
+      "learning_rate": 1.0519734629612646e-05,
+      "loss": 0.3642,
+      "step": 31670
+    },
+    {
+      "epoch": 6.231467138747511,
+      "grad_norm": 1.5423457622528076,
+      "learning_rate": 1.0516677367085513e-05,
+      "loss": 0.3332,
+      "step": 31680
+    },
+    {
+      "epoch": 6.233434142263529,
+      "grad_norm": 1.2122647762298584,
+      "learning_rate": 1.0513620104558379e-05,
+      "loss": 0.4714,
+      "step": 31690
+    },
+    {
+      "epoch": 6.235401145779548,
+      "grad_norm": 1.450034737586975,
+      "learning_rate": 1.0510562842031247e-05,
+      "loss": 0.3323,
+      "step": 31700
+    },
+    {
+      "epoch": 6.237368149295567,
+      "grad_norm": 1.2099430561065674,
+      "learning_rate": 1.0507505579504114e-05,
+      "loss": 0.4134,
+      "step": 31710
+    },
+    {
+      "epoch": 6.239335152811585,
+      "grad_norm": 2.170869827270508,
+      "learning_rate": 1.0504448316976981e-05,
+      "loss": 0.4279,
+      "step": 31720
+    },
+    {
+      "epoch": 6.241302156327604,
+      "grad_norm": 0.630581796169281,
+      "learning_rate": 1.0501391054449846e-05,
+      "loss": 0.5135,
+      "step": 31730
+    },
+    {
+      "epoch": 6.243269159843623,
+      "grad_norm": 2.2780323028564453,
+      "learning_rate": 1.0498333791922713e-05,
+      "loss": 0.423,
+      "step": 31740
+    },
+    {
+      "epoch": 6.245236163359642,
+      "grad_norm": 1.5196157693862915,
+      "learning_rate": 1.0495276529395579e-05,
+      "loss": 0.2993,
+      "step": 31750
+    },
+    {
+      "epoch": 6.247203166875661,
+      "grad_norm": 2.440230369567871,
+      "learning_rate": 1.0492219266868446e-05,
+      "loss": 0.3687,
+      "step": 31760
+    },
+    {
+      "epoch": 6.24917017039168,
+      "grad_norm": 0.8297176957130432,
+      "learning_rate": 1.0489162004341314e-05,
+      "loss": 0.4382,
+      "step": 31770
+    },
+    {
+      "epoch": 6.251137173907698,
+      "grad_norm": 1.7238305807113647,
+      "learning_rate": 1.0486104741814181e-05,
+      "loss": 0.3213,
+      "step": 31780
+    },
+    {
+      "epoch": 6.253104177423717,
+      "grad_norm": 1.177493691444397,
+      "learning_rate": 1.0483047479287047e-05,
+      "loss": 0.2217,
+      "step": 31790
+    },
+    {
+      "epoch": 6.255071180939736,
+      "grad_norm": 1.8509219884872437,
+      "learning_rate": 1.0479990216759914e-05,
+      "loss": 0.3354,
+      "step": 31800
+    },
+    {
+      "epoch": 6.257038184455754,
+      "grad_norm": 0.8892121315002441,
+      "learning_rate": 1.0476932954232782e-05,
+      "loss": 0.3079,
+      "step": 31810
+    },
+    {
+      "epoch": 6.259005187971773,
+      "grad_norm": 1.7575033903121948,
+      "learning_rate": 1.0473875691705648e-05,
+      "loss": 0.3453,
+      "step": 31820
+    },
+    {
+      "epoch": 6.2609721914877925,
+      "grad_norm": 0.984065055847168,
+      "learning_rate": 1.0470818429178515e-05,
+      "loss": 0.2972,
+      "step": 31830
+    },
+    {
+      "epoch": 6.262939195003811,
+      "grad_norm": 1.5092118978500366,
+      "learning_rate": 1.0467761166651383e-05,
+      "loss": 0.4259,
+      "step": 31840
+    },
+    {
+      "epoch": 6.26490619851983,
+      "grad_norm": 1.1990478038787842,
+      "learning_rate": 1.0464703904124247e-05,
+      "loss": 0.3105,
+      "step": 31850
+    },
+    {
+      "epoch": 6.266873202035849,
+      "grad_norm": 1.0917842388153076,
+      "learning_rate": 1.0461646641597114e-05,
+      "loss": 0.3538,
+      "step": 31860
+    },
+    {
+      "epoch": 6.268840205551867,
+      "grad_norm": 1.0836409330368042,
+      "learning_rate": 1.0458589379069982e-05,
+      "loss": 0.2392,
+      "step": 31870
+    },
+    {
+      "epoch": 6.270807209067886,
+      "grad_norm": 1.336677074432373,
+      "learning_rate": 1.0455532116542847e-05,
+      "loss": 0.3993,
+      "step": 31880
+    },
+    {
+      "epoch": 6.272774212583905,
+      "grad_norm": 0.9778781533241272,
+      "learning_rate": 1.0452474854015715e-05,
+      "loss": 0.2921,
+      "step": 31890
+    },
+    {
+      "epoch": 6.274741216099923,
+      "grad_norm": 1.270330786705017,
+      "learning_rate": 1.0449417591488582e-05,
+      "loss": 0.4029,
+      "step": 31900
+    },
+    {
+      "epoch": 6.2767082196159425,
+      "grad_norm": 1.2101235389709473,
+      "learning_rate": 1.044636032896145e-05,
+      "loss": 0.3091,
+      "step": 31910
+    },
+    {
+      "epoch": 6.278675223131962,
+      "grad_norm": 1.7291163206100464,
+      "learning_rate": 1.0443303066434316e-05,
+      "loss": 0.4171,
+      "step": 31920
+    },
+    {
+      "epoch": 6.28064222664798,
+      "grad_norm": 1.4795582294464111,
+      "learning_rate": 1.0440245803907183e-05,
+      "loss": 0.3782,
+      "step": 31930
+    },
+    {
+      "epoch": 6.282609230163999,
+      "grad_norm": 1.6377198696136475,
+      "learning_rate": 1.043718854138005e-05,
+      "loss": 0.3424,
+      "step": 31940
+    },
+    {
+      "epoch": 6.284576233680018,
+      "grad_norm": 1.3972163200378418,
+      "learning_rate": 1.0434131278852916e-05,
+      "loss": 0.3734,
+      "step": 31950
+    },
+    {
+      "epoch": 6.286543237196036,
+      "grad_norm": 1.3955159187316895,
+      "learning_rate": 1.0431074016325782e-05,
+      "loss": 0.4133,
+      "step": 31960
+    },
+    {
+      "epoch": 6.288510240712055,
+      "grad_norm": 1.2619258165359497,
+      "learning_rate": 1.0428016753798648e-05,
+      "loss": 0.3035,
+      "step": 31970
+    },
+    {
+      "epoch": 6.290477244228074,
+      "grad_norm": 1.8464090824127197,
+      "learning_rate": 1.0424959491271515e-05,
+      "loss": 0.4578,
+      "step": 31980
+    },
+    {
+      "epoch": 6.2924442477440925,
+      "grad_norm": 1.2729978561401367,
+      "learning_rate": 1.0421902228744383e-05,
+      "loss": 0.4224,
+      "step": 31990
+    },
+    {
+      "epoch": 6.294411251260112,
+      "grad_norm": 0.8992209434509277,
+      "learning_rate": 1.041884496621725e-05,
+      "loss": 0.3831,
+      "step": 32000
+    },
+    {
+      "epoch": 6.294411251260112,
+      "eval_loss": 0.14667941629886627,
+      "eval_runtime": 8.9005,
+      "eval_samples_per_second": 5.618,
+      "eval_steps_per_second": 2.809,
+      "step": 32000
+    },
+    {
+      "epoch": 6.296378254776131,
+      "grad_norm": 0.7840712070465088,
+      "learning_rate": 1.0415787703690116e-05,
+      "loss": 0.3726,
+      "step": 32010
+    },
+    {
+      "epoch": 6.298345258292149,
+      "grad_norm": 0.7266148924827576,
+      "learning_rate": 1.0412730441162984e-05,
+      "loss": 0.2885,
+      "step": 32020
+    },
+    {
+      "epoch": 6.300312261808168,
+      "grad_norm": 1.2583742141723633,
+      "learning_rate": 1.0409673178635851e-05,
+      "loss": 0.2949,
+      "step": 32030
+    },
+    {
+      "epoch": 6.302279265324187,
+      "grad_norm": 1.4370315074920654,
+      "learning_rate": 1.0406615916108717e-05,
+      "loss": 0.3672,
+      "step": 32040
+    },
+    {
+      "epoch": 6.304246268840205,
+      "grad_norm": 0.9335483908653259,
+      "learning_rate": 1.0403558653581584e-05,
+      "loss": 0.3721,
+      "step": 32050
+    },
+    {
+      "epoch": 6.306213272356224,
+      "grad_norm": 1.3395602703094482,
+      "learning_rate": 1.0400501391054452e-05,
+      "loss": 0.3631,
+      "step": 32060
+    },
+    {
+      "epoch": 6.308180275872243,
+      "grad_norm": 1.8281170129776,
+      "learning_rate": 1.0397444128527316e-05,
+      "loss": 0.3413,
+      "step": 32070
+    },
+    {
+      "epoch": 6.310147279388262,
+      "grad_norm": 1.3462482690811157,
+      "learning_rate": 1.0394386866000183e-05,
+      "loss": 0.3088,
+      "step": 32080
+    },
+    {
+      "epoch": 6.312114282904281,
+      "grad_norm": 1.1540286540985107,
+      "learning_rate": 1.039132960347305e-05,
+      "loss": 0.42,
+      "step": 32090
+    },
+    {
+      "epoch": 6.3140812864203,
+      "grad_norm": 1.4234508275985718,
+      "learning_rate": 1.0388272340945917e-05,
+      "loss": 0.3489,
+      "step": 32100
+    },
+    {
+      "epoch": 6.316048289936318,
+      "grad_norm": 1.2908326387405396,
+      "learning_rate": 1.0385215078418784e-05,
+      "loss": 0.4033,
+      "step": 32110
+    },
+    {
+      "epoch": 6.318015293452337,
+      "grad_norm": 0.784662127494812,
+      "learning_rate": 1.0382157815891652e-05,
+      "loss": 0.2647,
+      "step": 32120
+    },
+    {
+      "epoch": 6.319982296968356,
+      "grad_norm": 1.8527098894119263,
+      "learning_rate": 1.0379100553364519e-05,
+      "loss": 0.3639,
+      "step": 32130
+    },
+    {
+      "epoch": 6.321949300484374,
+      "grad_norm": 1.9791474342346191,
+      "learning_rate": 1.0376043290837385e-05,
+      "loss": 0.4099,
+      "step": 32140
+    },
+    {
+      "epoch": 6.323916304000393,
+      "grad_norm": 1.3550399541854858,
+      "learning_rate": 1.0372986028310252e-05,
+      "loss": 0.3554,
+      "step": 32150
+    },
+    {
+      "epoch": 6.3258833075164125,
+      "grad_norm": 0.712272047996521,
+      "learning_rate": 1.036992876578312e-05,
+      "loss": 0.2878,
+      "step": 32160
+    },
+    {
+      "epoch": 6.327850311032431,
+      "grad_norm": 1.3497095108032227,
+      "learning_rate": 1.0366871503255986e-05,
+      "loss": 0.3627,
+      "step": 32170
+    },
+    {
+      "epoch": 6.32981731454845,
+      "grad_norm": 1.3138824701309204,
+      "learning_rate": 1.0363814240728853e-05,
+      "loss": 0.3123,
+      "step": 32180
+    },
+    {
+      "epoch": 6.331784318064469,
+      "grad_norm": 1.375718593597412,
+      "learning_rate": 1.0360756978201719e-05,
+      "loss": 0.3333,
+      "step": 32190
+    },
+    {
+      "epoch": 6.333751321580487,
+      "grad_norm": 1.069199800491333,
+      "learning_rate": 1.0357699715674585e-05,
+      "loss": 0.294,
+      "step": 32200
+    },
+    {
+      "epoch": 6.335718325096506,
+      "grad_norm": 1.2835943698883057,
+      "learning_rate": 1.0354642453147452e-05,
+      "loss": 0.3857,
+      "step": 32210
+    },
+    {
+      "epoch": 6.337685328612525,
+      "grad_norm": 1.1084645986557007,
+      "learning_rate": 1.035158519062032e-05,
+      "loss": 0.3547,
+      "step": 32220
+    },
+    {
+      "epoch": 6.339652332128543,
+      "grad_norm": 0.8232660293579102,
+      "learning_rate": 1.0348527928093185e-05,
+      "loss": 0.2852,
+      "step": 32230
+    },
+    {
+      "epoch": 6.3416193356445625,
+      "grad_norm": 1.066155195236206,
+      "learning_rate": 1.0345470665566053e-05,
+      "loss": 0.391,
+      "step": 32240
+    },
+    {
+      "epoch": 6.343586339160582,
+      "grad_norm": 1.3041104078292847,
+      "learning_rate": 1.034241340303892e-05,
+      "loss": 0.3893,
+      "step": 32250
+    },
+    {
+      "epoch": 6.3455533426766,
+      "grad_norm": 0.8700298070907593,
+      "learning_rate": 1.0339356140511788e-05,
+      "loss": 0.4521,
+      "step": 32260
+    },
+    {
+      "epoch": 6.347520346192619,
+      "grad_norm": 0.884760320186615,
+      "learning_rate": 1.0336298877984653e-05,
+      "loss": 0.3217,
+      "step": 32270
+    },
+    {
+      "epoch": 6.349487349708638,
+      "grad_norm": 1.4652529954910278,
+      "learning_rate": 1.0333241615457521e-05,
+      "loss": 0.3204,
+      "step": 32280
+    },
+    {
+      "epoch": 6.351454353224656,
+      "grad_norm": 1.0134848356246948,
+      "learning_rate": 1.0330184352930388e-05,
+      "loss": 0.3291,
+      "step": 32290
+    },
+    {
+      "epoch": 6.353421356740675,
+      "grad_norm": 1.1539208889007568,
+      "learning_rate": 1.0327127090403252e-05,
+      "loss": 0.3689,
+      "step": 32300
+    },
+    {
+      "epoch": 6.355388360256694,
+      "grad_norm": 1.202022910118103,
+      "learning_rate": 1.032406982787612e-05,
+      "loss": 0.3629,
+      "step": 32310
+    },
+    {
+      "epoch": 6.3573553637727125,
+      "grad_norm": 0.8208584785461426,
+      "learning_rate": 1.0321012565348987e-05,
+      "loss": 0.329,
+      "step": 32320
+    },
+    {
+      "epoch": 6.359322367288732,
+      "grad_norm": 0.9570119380950928,
+      "learning_rate": 1.0317955302821853e-05,
+      "loss": 0.3857,
+      "step": 32330
+    },
+    {
+      "epoch": 6.361289370804751,
+      "grad_norm": 0.9772025942802429,
+      "learning_rate": 1.031489804029472e-05,
+      "loss": 0.5343,
+      "step": 32340
+    },
+    {
+      "epoch": 6.363256374320769,
+      "grad_norm": 0.41304048895835876,
+      "learning_rate": 1.0311840777767588e-05,
+      "loss": 0.323,
+      "step": 32350
+    },
+    {
+      "epoch": 6.365223377836788,
+      "grad_norm": 1.1537532806396484,
+      "learning_rate": 1.0308783515240454e-05,
+      "loss": 0.3416,
+      "step": 32360
+    },
+    {
+      "epoch": 6.367190381352807,
+      "grad_norm": 1.2086857557296753,
+      "learning_rate": 1.0305726252713321e-05,
+      "loss": 0.3707,
+      "step": 32370
+    },
+    {
+      "epoch": 6.369157384868825,
+      "grad_norm": 0.5529403686523438,
+      "learning_rate": 1.0302668990186189e-05,
+      "loss": 0.3776,
+      "step": 32380
+    },
+    {
+      "epoch": 6.371124388384844,
+      "grad_norm": 0.9833685159683228,
+      "learning_rate": 1.0299611727659056e-05,
+      "loss": 0.4551,
+      "step": 32390
+    },
+    {
+      "epoch": 6.373091391900863,
+      "grad_norm": 1.5811911821365356,
+      "learning_rate": 1.0296554465131922e-05,
+      "loss": 0.5041,
+      "step": 32400
+    },
+    {
+      "epoch": 6.375058395416882,
+      "grad_norm": 0.6909576654434204,
+      "learning_rate": 1.0293497202604788e-05,
+      "loss": 0.3439,
+      "step": 32410
+    },
+    {
+      "epoch": 6.377025398932901,
+      "grad_norm": 0.8619405031204224,
+      "learning_rate": 1.0290439940077654e-05,
+      "loss": 0.3578,
+      "step": 32420
+    },
+    {
+      "epoch": 6.37899240244892,
+      "grad_norm": 0.9947187900543213,
+      "learning_rate": 1.0287382677550521e-05,
+      "loss": 0.4153,
+      "step": 32430
+    },
+    {
+      "epoch": 6.380959405964938,
+      "grad_norm": 0.8003689050674438,
+      "learning_rate": 1.0284325415023389e-05,
+      "loss": 0.2718,
+      "step": 32440
+    },
+    {
+      "epoch": 6.382926409480957,
+      "grad_norm": 1.235228419303894,
+      "learning_rate": 1.0281268152496256e-05,
+      "loss": 0.5116,
+      "step": 32450
+    },
+    {
+      "epoch": 6.384893412996976,
+      "grad_norm": 1.0889111757278442,
+      "learning_rate": 1.0278210889969122e-05,
+      "loss": 0.4442,
+      "step": 32460
+    },
+    {
+      "epoch": 6.386860416512994,
+      "grad_norm": 1.042790174484253,
+      "learning_rate": 1.027515362744199e-05,
+      "loss": 0.276,
+      "step": 32470
+    },
+    {
+      "epoch": 6.388827420029013,
+      "grad_norm": 1.6201059818267822,
+      "learning_rate": 1.0272096364914857e-05,
+      "loss": 0.3747,
+      "step": 32480
+    },
+    {
+      "epoch": 6.3907944235450325,
+      "grad_norm": 1.099861741065979,
+      "learning_rate": 1.0269039102387723e-05,
+      "loss": 0.4556,
+      "step": 32490
+    },
+    {
+      "epoch": 6.392761427061051,
+      "grad_norm": 1.1103860139846802,
+      "learning_rate": 1.026598183986059e-05,
+      "loss": 0.2511,
+      "step": 32500
+    },
+    {
+      "epoch": 6.392761427061051,
+      "eval_loss": 0.14782226085662842,
+      "eval_runtime": 8.8932,
+      "eval_samples_per_second": 5.622,
+      "eval_steps_per_second": 2.811,
+      "step": 32500
+    },
+    {
+      "epoch": 6.39472843057707,
+      "grad_norm": 0.9129806160926819,
+      "learning_rate": 1.0262924577333458e-05,
+      "loss": 0.4684,
+      "step": 32510
+    },
+    {
+      "epoch": 6.396695434093089,
+      "grad_norm": 1.1795682907104492,
+      "learning_rate": 1.0259867314806325e-05,
+      "loss": 0.3383,
+      "step": 32520
+    },
+    {
+      "epoch": 6.398662437609107,
+      "grad_norm": 2.0722591876983643,
+      "learning_rate": 1.0256810052279189e-05,
+      "loss": 0.3128,
+      "step": 32530
+    },
+    {
+      "epoch": 6.400629441125126,
+      "grad_norm": 1.4256749153137207,
+      "learning_rate": 1.0253752789752057e-05,
+      "loss": 0.3892,
+      "step": 32540
+    },
+    {
+      "epoch": 6.402596444641145,
+      "grad_norm": 2.753934383392334,
+      "learning_rate": 1.0250695527224922e-05,
+      "loss": 0.4021,
+      "step": 32550
+    },
+    {
+      "epoch": 6.404563448157163,
+      "grad_norm": 1.4296075105667114,
+      "learning_rate": 1.024763826469779e-05,
+      "loss": 0.3808,
+      "step": 32560
+    },
+    {
+      "epoch": 6.4065304516731825,
+      "grad_norm": 0.9655913710594177,
+      "learning_rate": 1.0244581002170657e-05,
+      "loss": 0.4817,
+      "step": 32570
+    },
+    {
+      "epoch": 6.4084974551892016,
+      "grad_norm": 1.0583401918411255,
+      "learning_rate": 1.0241523739643525e-05,
+      "loss": 0.4399,
+      "step": 32580
+    },
+    {
+      "epoch": 6.41046445870522,
+      "grad_norm": 0.9479387998580933,
+      "learning_rate": 1.023846647711639e-05,
+      "loss": 0.3651,
+      "step": 32590
+    },
+    {
+      "epoch": 6.412431462221239,
+      "grad_norm": 0.7584673166275024,
+      "learning_rate": 1.0235409214589258e-05,
+      "loss": 0.3063,
+      "step": 32600
+    },
+    {
+      "epoch": 6.414398465737258,
+      "grad_norm": 1.1318553686141968,
+      "learning_rate": 1.0232351952062125e-05,
+      "loss": 0.426,
+      "step": 32610
+    },
+    {
+      "epoch": 6.416365469253276,
+      "grad_norm": 1.6314092874526978,
+      "learning_rate": 1.0229294689534991e-05,
+      "loss": 0.3312,
+      "step": 32620
+    },
+    {
+      "epoch": 6.418332472769295,
+      "grad_norm": 0.7870709300041199,
+      "learning_rate": 1.0226237427007859e-05,
+      "loss": 0.3737,
+      "step": 32630
+    },
+    {
+      "epoch": 6.420299476285313,
+      "grad_norm": 0.8929570913314819,
+      "learning_rate": 1.0223180164480724e-05,
+      "loss": 0.2973,
+      "step": 32640
+    },
+    {
+      "epoch": 6.4222664798013325,
+      "grad_norm": 1.1901662349700928,
+      "learning_rate": 1.022012290195359e-05,
+      "loss": 0.3487,
+      "step": 32650
+    },
+    {
+      "epoch": 6.424233483317352,
+      "grad_norm": 1.270427942276001,
+      "learning_rate": 1.0217065639426458e-05,
+      "loss": 0.4018,
+      "step": 32660
+    },
+    {
+      "epoch": 6.42620048683337,
+      "grad_norm": 1.4906806945800781,
+      "learning_rate": 1.0214008376899325e-05,
+      "loss": 0.3262,
+      "step": 32670
+    },
+    {
+      "epoch": 6.428167490349389,
+      "grad_norm": 1.2974570989608765,
+      "learning_rate": 1.0210951114372191e-05,
+      "loss": 0.3769,
+      "step": 32680
+    },
+    {
+      "epoch": 6.430134493865408,
+      "grad_norm": 1.0146774053573608,
+      "learning_rate": 1.0207893851845058e-05,
+      "loss": 0.4504,
+      "step": 32690
+    },
+    {
+      "epoch": 6.432101497381426,
+      "grad_norm": 1.3171179294586182,
+      "learning_rate": 1.0204836589317926e-05,
+      "loss": 0.3756,
+      "step": 32700
+    },
+    {
+      "epoch": 6.434068500897445,
+      "grad_norm": 1.1145819425582886,
+      "learning_rate": 1.0201779326790793e-05,
+      "loss": 0.3819,
+      "step": 32710
+    },
+    {
+      "epoch": 6.436035504413464,
+      "grad_norm": 1.547440528869629,
+      "learning_rate": 1.019872206426366e-05,
+      "loss": 0.4455,
+      "step": 32720
+    },
+    {
+      "epoch": 6.4380025079294825,
+      "grad_norm": 1.3117337226867676,
+      "learning_rate": 1.0195664801736527e-05,
+      "loss": 0.453,
+      "step": 32730
+    },
+    {
+      "epoch": 6.439969511445502,
+      "grad_norm": 1.771700382232666,
+      "learning_rate": 1.0192607539209394e-05,
+      "loss": 0.3808,
+      "step": 32740
+    },
+    {
+      "epoch": 6.441936514961521,
+      "grad_norm": 0.7972196936607361,
+      "learning_rate": 1.0189550276682258e-05,
+      "loss": 0.5139,
+      "step": 32750
+    },
+    {
+      "epoch": 6.443903518477539,
+      "grad_norm": 0.4077063798904419,
+      "learning_rate": 1.0186493014155126e-05,
+      "loss": 0.3401,
+      "step": 32760
+    },
+    {
+      "epoch": 6.445870521993558,
+      "grad_norm": 1.2201696634292603,
+      "learning_rate": 1.0183435751627993e-05,
+      "loss": 0.3978,
+      "step": 32770
+    },
+    {
+      "epoch": 6.447837525509577,
+      "grad_norm": 2.2601189613342285,
+      "learning_rate": 1.0180378489100859e-05,
+      "loss": 0.3514,
+      "step": 32780
+    },
+    {
+      "epoch": 6.449804529025595,
+      "grad_norm": 1.1869385242462158,
+      "learning_rate": 1.0177321226573726e-05,
+      "loss": 0.4924,
+      "step": 32790
+    },
+    {
+      "epoch": 6.451771532541614,
+      "grad_norm": 1.4132533073425293,
+      "learning_rate": 1.0174263964046594e-05,
+      "loss": 0.3294,
+      "step": 32800
+    },
+    {
+      "epoch": 6.453738536057633,
+      "grad_norm": 1.091038465499878,
+      "learning_rate": 1.017120670151946e-05,
+      "loss": 0.4039,
+      "step": 32810
+    },
+    {
+      "epoch": 6.455705539573652,
+      "grad_norm": 1.458566665649414,
+      "learning_rate": 1.0168149438992327e-05,
+      "loss": 0.3343,
+      "step": 32820
+    },
+    {
+      "epoch": 6.457672543089671,
+      "grad_norm": 1.7340314388275146,
+      "learning_rate": 1.0165092176465195e-05,
+      "loss": 0.3251,
+      "step": 32830
+    },
+    {
+      "epoch": 6.45963954660569,
+      "grad_norm": 2.443197250366211,
+      "learning_rate": 1.0162034913938062e-05,
+      "loss": 0.4722,
+      "step": 32840
+    },
+    {
+      "epoch": 6.461606550121708,
+      "grad_norm": 0.9471251368522644,
+      "learning_rate": 1.0158977651410928e-05,
+      "loss": 0.3322,
+      "step": 32850
+    },
+    {
+      "epoch": 6.463573553637727,
+      "grad_norm": 1.2571531534194946,
+      "learning_rate": 1.0155920388883795e-05,
+      "loss": 0.3339,
+      "step": 32860
+    },
+    {
+      "epoch": 6.465540557153746,
+      "grad_norm": 0.7914329171180725,
+      "learning_rate": 1.015286312635666e-05,
+      "loss": 0.3534,
+      "step": 32870
+    },
+    {
+      "epoch": 6.467507560669764,
+      "grad_norm": 1.7630119323730469,
+      "learning_rate": 1.0149805863829527e-05,
+      "loss": 0.4638,
+      "step": 32880
+    },
+    {
+      "epoch": 6.469474564185783,
+      "grad_norm": 1.0704189538955688,
+      "learning_rate": 1.0146748601302394e-05,
+      "loss": 0.341,
+      "step": 32890
+    },
+    {
+      "epoch": 6.4714415677018025,
+      "grad_norm": 1.5141534805297852,
+      "learning_rate": 1.0143691338775262e-05,
+      "loss": 0.4852,
+      "step": 32900
+    },
+    {
+      "epoch": 6.473408571217821,
+      "grad_norm": 0.8697926998138428,
+      "learning_rate": 1.0140634076248128e-05,
+      "loss": 0.3165,
+      "step": 32910
+    },
+    {
+      "epoch": 6.47537557473384,
+      "grad_norm": 1.7349079847335815,
+      "learning_rate": 1.0137576813720995e-05,
+      "loss": 0.2807,
+      "step": 32920
+    },
+    {
+      "epoch": 6.477342578249859,
+      "grad_norm": 1.5268747806549072,
+      "learning_rate": 1.0134519551193863e-05,
+      "loss": 0.3529,
+      "step": 32930
+    },
+    {
+      "epoch": 6.479309581765877,
+      "grad_norm": 1.3544410467147827,
+      "learning_rate": 1.0131462288666728e-05,
+      "loss": 0.2781,
+      "step": 32940
+    },
+    {
+      "epoch": 6.481276585281896,
+      "grad_norm": 1.6031782627105713,
+      "learning_rate": 1.0128405026139596e-05,
+      "loss": 0.3772,
+      "step": 32950
+    },
+    {
+      "epoch": 6.483243588797915,
+      "grad_norm": 0.9160840511322021,
+      "learning_rate": 1.0125347763612463e-05,
+      "loss": 0.2645,
+      "step": 32960
+    },
+    {
+      "epoch": 6.485210592313933,
+      "grad_norm": 1.133995532989502,
+      "learning_rate": 1.012229050108533e-05,
+      "loss": 0.2998,
+      "step": 32970
+    },
+    {
+      "epoch": 6.4871775958299525,
+      "grad_norm": 2.93790864944458,
+      "learning_rate": 1.0119233238558195e-05,
+      "loss": 0.3928,
+      "step": 32980
+    },
+    {
+      "epoch": 6.4891445993459715,
+      "grad_norm": 2.064584255218506,
+      "learning_rate": 1.0116175976031062e-05,
+      "loss": 0.3689,
+      "step": 32990
+    },
+    {
+      "epoch": 6.49111160286199,
+      "grad_norm": 0.8051519393920898,
+      "learning_rate": 1.0113118713503928e-05,
+      "loss": 0.4315,
+      "step": 33000
+    },
+    {
+      "epoch": 6.49111160286199,
+      "eval_loss": 0.14803524315357208,
+      "eval_runtime": 8.8266,
+      "eval_samples_per_second": 5.665,
+      "eval_steps_per_second": 2.832,
+      "step": 33000
+    },
+    {
+      "epoch": 6.493078606378009,
+      "grad_norm": 1.1119964122772217,
+      "learning_rate": 1.0110061450976796e-05,
+      "loss": 0.4261,
+      "step": 33010
+    },
+    {
+      "epoch": 6.495045609894028,
+      "grad_norm": 0.7975828051567078,
+      "learning_rate": 1.0107004188449663e-05,
+      "loss": 0.3573,
+      "step": 33020
+    },
+    {
+      "epoch": 6.497012613410046,
+      "grad_norm": 1.5630384683609009,
+      "learning_rate": 1.010394692592253e-05,
+      "loss": 0.3167,
+      "step": 33030
+    },
+    {
+      "epoch": 6.498979616926065,
+      "grad_norm": 0.8238933086395264,
+      "learning_rate": 1.0100889663395396e-05,
+      "loss": 0.4177,
+      "step": 33040
+    },
+    {
+      "epoch": 6.500946620442084,
+      "grad_norm": 0.8332470655441284,
+      "learning_rate": 1.0097832400868264e-05,
+      "loss": 0.3705,
+      "step": 33050
+    },
+    {
+      "epoch": 6.5029136239581025,
+      "grad_norm": 0.9948000907897949,
+      "learning_rate": 1.0094775138341131e-05,
+      "loss": 0.4374,
+      "step": 33060
+    },
+    {
+      "epoch": 6.5048806274741215,
+      "grad_norm": 0.8258544206619263,
+      "learning_rate": 1.0091717875813997e-05,
+      "loss": 0.3398,
+      "step": 33070
+    },
+    {
+      "epoch": 6.506847630990141,
+      "grad_norm": 0.8997102379798889,
+      "learning_rate": 1.0088660613286864e-05,
+      "loss": 0.2978,
+      "step": 33080
+    },
+    {
+      "epoch": 6.508814634506159,
+      "grad_norm": 1.6043155193328857,
+      "learning_rate": 1.008560335075973e-05,
+      "loss": 0.4263,
+      "step": 33090
+    },
+    {
+      "epoch": 6.510781638022178,
+      "grad_norm": 1.8473858833312988,
+      "learning_rate": 1.0082546088232596e-05,
+      "loss": 0.3009,
+      "step": 33100
+    },
+    {
+      "epoch": 6.512748641538197,
+      "grad_norm": 1.3721176385879517,
+      "learning_rate": 1.0079488825705463e-05,
+      "loss": 0.331,
+      "step": 33110
+    },
+    {
+      "epoch": 6.514715645054215,
+      "grad_norm": 1.10801100730896,
+      "learning_rate": 1.0076431563178331e-05,
+      "loss": 0.334,
+      "step": 33120
+    },
+    {
+      "epoch": 6.516682648570234,
+      "grad_norm": 0.6907325983047485,
+      "learning_rate": 1.0073374300651197e-05,
+      "loss": 0.346,
+      "step": 33130
+    },
+    {
+      "epoch": 6.518649652086253,
+      "grad_norm": 0.8262765407562256,
+      "learning_rate": 1.0070317038124064e-05,
+      "loss": 0.3339,
+      "step": 33140
+    },
+    {
+      "epoch": 6.5206166556022715,
+      "grad_norm": 1.0150346755981445,
+      "learning_rate": 1.0067259775596932e-05,
+      "loss": 0.317,
+      "step": 33150
+    },
+    {
+      "epoch": 6.522583659118291,
+      "grad_norm": 2.1446454524993896,
+      "learning_rate": 1.00642025130698e-05,
+      "loss": 0.2435,
+      "step": 33160
+    },
+    {
+      "epoch": 6.52455066263431,
+      "grad_norm": 1.5295718908309937,
+      "learning_rate": 1.0061145250542665e-05,
+      "loss": 0.3208,
+      "step": 33170
+    },
+    {
+      "epoch": 6.526517666150328,
+      "grad_norm": 0.38532695174217224,
+      "learning_rate": 1.0058087988015532e-05,
+      "loss": 0.322,
+      "step": 33180
+    },
+    {
+      "epoch": 6.528484669666347,
+      "grad_norm": 0.6402590274810791,
+      "learning_rate": 1.00550307254884e-05,
+      "loss": 0.3332,
+      "step": 33190
+    },
+    {
+      "epoch": 6.530451673182366,
+      "grad_norm": 1.3206236362457275,
+      "learning_rate": 1.0051973462961266e-05,
+      "loss": 0.3526,
+      "step": 33200
+    },
+    {
+      "epoch": 6.532418676698384,
+      "grad_norm": 1.180548071861267,
+      "learning_rate": 1.0048916200434131e-05,
+      "loss": 0.3406,
+      "step": 33210
+    },
+    {
+      "epoch": 6.534385680214403,
+      "grad_norm": 0.8427443504333496,
+      "learning_rate": 1.0045858937906999e-05,
+      "loss": 0.3626,
+      "step": 33220
+    },
+    {
+      "epoch": 6.536352683730422,
+      "grad_norm": 1.1290903091430664,
+      "learning_rate": 1.0042801675379865e-05,
+      "loss": 0.3294,
+      "step": 33230
+    },
+    {
+      "epoch": 6.538319687246441,
+      "grad_norm": 1.4042856693267822,
+      "learning_rate": 1.0039744412852732e-05,
+      "loss": 0.351,
+      "step": 33240
+    },
+    {
+      "epoch": 6.54028669076246,
+      "grad_norm": 1.821936011314392,
+      "learning_rate": 1.00366871503256e-05,
+      "loss": 0.3147,
+      "step": 33250
+    },
+    {
+      "epoch": 6.542253694278479,
+      "grad_norm": 1.4279309511184692,
+      "learning_rate": 1.0033629887798465e-05,
+      "loss": 0.3773,
+      "step": 33260
+    },
+    {
+      "epoch": 6.544220697794497,
+      "grad_norm": 0.9054422378540039,
+      "learning_rate": 1.0030572625271333e-05,
+      "loss": 0.3617,
+      "step": 33270
+    },
+    {
+      "epoch": 6.546187701310516,
+      "grad_norm": 1.0631873607635498,
+      "learning_rate": 1.00275153627442e-05,
+      "loss": 0.4193,
+      "step": 33280
+    },
+    {
+      "epoch": 6.548154704826535,
+      "grad_norm": 1.4749603271484375,
+      "learning_rate": 1.0024458100217068e-05,
+      "loss": 0.3965,
+      "step": 33290
+    },
+    {
+      "epoch": 6.550121708342553,
+      "grad_norm": 1.0832397937774658,
+      "learning_rate": 1.0021400837689934e-05,
+      "loss": 0.2609,
+      "step": 33300
+    },
+    {
+      "epoch": 6.5520887118585724,
+      "grad_norm": 2.431015729904175,
+      "learning_rate": 1.0018343575162801e-05,
+      "loss": 0.3519,
+      "step": 33310
+    },
+    {
+      "epoch": 6.5540557153745915,
+      "grad_norm": 1.274034023284912,
+      "learning_rate": 1.0015286312635665e-05,
+      "loss": 0.3999,
+      "step": 33320
+    },
+    {
+      "epoch": 6.55602271889061,
+      "grad_norm": 2.3737590312957764,
+      "learning_rate": 1.0012229050108533e-05,
+      "loss": 0.4696,
+      "step": 33330
+    },
+    {
+      "epoch": 6.557989722406629,
+      "grad_norm": 2.1949424743652344,
+      "learning_rate": 1.00091717875814e-05,
+      "loss": 0.3498,
+      "step": 33340
+    },
+    {
+      "epoch": 6.559956725922648,
+      "grad_norm": 1.2474371194839478,
+      "learning_rate": 1.0006114525054268e-05,
+      "loss": 0.505,
+      "step": 33350
+    },
+    {
+      "epoch": 6.561923729438666,
+      "grad_norm": 1.3272656202316284,
+      "learning_rate": 1.0003057262527133e-05,
+      "loss": 0.4109,
+      "step": 33360
+    },
+    {
+      "epoch": 6.563890732954685,
+      "grad_norm": 0.6079081296920776,
+      "learning_rate": 1e-05,
+      "loss": 0.2928,
+      "step": 33370
+    },
+    {
+      "epoch": 6.565857736470704,
+      "grad_norm": 0.39416030049324036,
+      "learning_rate": 9.996942737472868e-06,
+      "loss": 0.4067,
+      "step": 33380
+    },
+    {
+      "epoch": 6.5678247399867224,
+      "grad_norm": 1.118470549583435,
+      "learning_rate": 9.993885474945734e-06,
+      "loss": 0.3543,
+      "step": 33390
+    },
+    {
+      "epoch": 6.5697917435027415,
+      "grad_norm": 1.9530739784240723,
+      "learning_rate": 9.9908282124186e-06,
+      "loss": 0.4074,
+      "step": 33400
+    },
+    {
+      "epoch": 6.571758747018761,
+      "grad_norm": 0.9212591052055359,
+      "learning_rate": 9.987770949891467e-06,
+      "loss": 0.4043,
+      "step": 33410
+    },
+    {
+      "epoch": 6.573725750534779,
+      "grad_norm": 1.6362104415893555,
+      "learning_rate": 9.984713687364335e-06,
+      "loss": 0.3953,
+      "step": 33420
+    },
+    {
+      "epoch": 6.575692754050798,
+      "grad_norm": 0.9912499189376831,
+      "learning_rate": 9.981656424837202e-06,
+      "loss": 0.2861,
+      "step": 33430
+    },
+    {
+      "epoch": 6.577659757566817,
+      "grad_norm": 1.6725244522094727,
+      "learning_rate": 9.978599162310068e-06,
+      "loss": 0.3207,
+      "step": 33440
+    },
+    {
+      "epoch": 6.579626761082835,
+      "grad_norm": 0.7188600301742554,
+      "learning_rate": 9.975541899782936e-06,
+      "loss": 0.2644,
+      "step": 33450
+    },
+    {
+      "epoch": 6.581593764598854,
+      "grad_norm": 2.0234286785125732,
+      "learning_rate": 9.972484637255801e-06,
+      "loss": 0.3419,
+      "step": 33460
+    },
+    {
+      "epoch": 6.583560768114873,
+      "grad_norm": 0.6678216457366943,
+      "learning_rate": 9.969427374728669e-06,
+      "loss": 0.374,
+      "step": 33470
+    },
+    {
+      "epoch": 6.5855277716308915,
+      "grad_norm": 0.9582223892211914,
+      "learning_rate": 9.966370112201536e-06,
+      "loss": 0.3289,
+      "step": 33480
+    },
+    {
+      "epoch": 6.587494775146911,
+      "grad_norm": 1.9727556705474854,
+      "learning_rate": 9.963312849674402e-06,
+      "loss": 0.3535,
+      "step": 33490
+    },
+    {
+      "epoch": 6.58946177866293,
+      "grad_norm": 1.2631657123565674,
+      "learning_rate": 9.96025558714727e-06,
+      "loss": 0.3904,
+      "step": 33500
+    },
+    {
+      "epoch": 6.58946177866293,
+      "eval_loss": 0.1484050750732422,
+      "eval_runtime": 8.8797,
+      "eval_samples_per_second": 5.631,
+      "eval_steps_per_second": 2.815,
+      "step": 33500
+    },
+    {
+      "epoch": 6.591428782178948,
+      "grad_norm": 0.944599449634552,
+      "learning_rate": 9.957198324620137e-06,
+      "loss": 0.3904,
+      "step": 33510
+    },
+    {
+      "epoch": 6.593395785694967,
+      "grad_norm": 1.7309553623199463,
+      "learning_rate": 9.954141062093003e-06,
+      "loss": 0.4003,
+      "step": 33520
+    },
+    {
+      "epoch": 6.595362789210986,
+      "grad_norm": 1.5342662334442139,
+      "learning_rate": 9.951083799565869e-06,
+      "loss": 0.3995,
+      "step": 33530
+    },
+    {
+      "epoch": 6.597329792727004,
+      "grad_norm": 0.9480270743370056,
+      "learning_rate": 9.948026537038736e-06,
+      "loss": 0.2778,
+      "step": 33540
+    },
+    {
+      "epoch": 6.599296796243023,
+      "grad_norm": 0.864125669002533,
+      "learning_rate": 9.944969274511603e-06,
+      "loss": 0.4614,
+      "step": 33550
+    },
+    {
+      "epoch": 6.601263799759042,
+      "grad_norm": 3.487790107727051,
+      "learning_rate": 9.941912011984471e-06,
+      "loss": 0.4029,
+      "step": 33560
+    },
+    {
+      "epoch": 6.603230803275061,
+      "grad_norm": 2.3038623332977295,
+      "learning_rate": 9.938854749457337e-06,
+      "loss": 0.3582,
+      "step": 33570
+    },
+    {
+      "epoch": 6.60519780679108,
+      "grad_norm": 2.7051186561584473,
+      "learning_rate": 9.935797486930202e-06,
+      "loss": 0.3567,
+      "step": 33580
+    },
+    {
+      "epoch": 6.607164810307099,
+      "grad_norm": 2.1781363487243652,
+      "learning_rate": 9.93274022440307e-06,
+      "loss": 0.4223,
+      "step": 33590
+    },
+    {
+      "epoch": 6.609131813823117,
+      "grad_norm": 0.9622666239738464,
+      "learning_rate": 9.929682961875937e-06,
+      "loss": 0.3935,
+      "step": 33600
+    },
+    {
+      "epoch": 6.611098817339136,
+      "grad_norm": 0.6262819170951843,
+      "learning_rate": 9.926625699348805e-06,
+      "loss": 0.2501,
+      "step": 33610
+    },
+    {
+      "epoch": 6.613065820855155,
+      "grad_norm": 1.4548559188842773,
+      "learning_rate": 9.92356843682167e-06,
+      "loss": 0.5423,
+      "step": 33620
+    },
+    {
+      "epoch": 6.615032824371173,
+      "grad_norm": 0.7134609222412109,
+      "learning_rate": 9.920511174294536e-06,
+      "loss": 0.3038,
+      "step": 33630
+    },
+    {
+      "epoch": 6.616999827887192,
+      "grad_norm": 0.9675575494766235,
+      "learning_rate": 9.917453911767404e-06,
+      "loss": 0.313,
+      "step": 33640
+    },
+    {
+      "epoch": 6.6189668314032115,
+      "grad_norm": 1.1765183210372925,
+      "learning_rate": 9.914396649240271e-06,
+      "loss": 0.4338,
+      "step": 33650
+    },
+    {
+      "epoch": 6.62093383491923,
+      "grad_norm": 1.1487053632736206,
+      "learning_rate": 9.911339386713137e-06,
+      "loss": 0.4628,
+      "step": 33660
+    },
+    {
+      "epoch": 6.622900838435249,
+      "grad_norm": 0.7740342020988464,
+      "learning_rate": 9.908282124186005e-06,
+      "loss": 0.3036,
+      "step": 33670
+    },
+    {
+      "epoch": 6.624867841951268,
+      "grad_norm": 0.717975914478302,
+      "learning_rate": 9.905224861658872e-06,
+      "loss": 0.4258,
+      "step": 33680
+    },
+    {
+      "epoch": 6.626834845467286,
+      "grad_norm": 1.3622957468032837,
+      "learning_rate": 9.902167599131738e-06,
+      "loss": 0.5035,
+      "step": 33690
+    },
+    {
+      "epoch": 6.628801848983305,
+      "grad_norm": 0.8520593643188477,
+      "learning_rate": 9.899110336604605e-06,
+      "loss": 0.4059,
+      "step": 33700
+    },
+    {
+      "epoch": 6.630768852499324,
+      "grad_norm": 1.1923712491989136,
+      "learning_rate": 9.896053074077471e-06,
+      "loss": 0.3483,
+      "step": 33710
+    },
+    {
+      "epoch": 6.632735856015342,
+      "grad_norm": 1.4621663093566895,
+      "learning_rate": 9.892995811550339e-06,
+      "loss": 0.3298,
+      "step": 33720
+    },
+    {
+      "epoch": 6.6347028595313615,
+      "grad_norm": 1.36235511302948,
+      "learning_rate": 9.889938549023206e-06,
+      "loss": 0.3386,
+      "step": 33730
+    },
+    {
+      "epoch": 6.636669863047381,
+      "grad_norm": 0.9748831987380981,
+      "learning_rate": 9.886881286496072e-06,
+      "loss": 0.3864,
+      "step": 33740
+    },
+    {
+      "epoch": 6.638636866563399,
+      "grad_norm": 0.5554082989692688,
+      "learning_rate": 9.88382402396894e-06,
+      "loss": 0.3939,
+      "step": 33750
+    },
+    {
+      "epoch": 6.640603870079418,
+      "grad_norm": 1.192552924156189,
+      "learning_rate": 9.880766761441805e-06,
+      "loss": 0.4679,
+      "step": 33760
+    },
+    {
+      "epoch": 6.642570873595437,
+      "grad_norm": 1.8474180698394775,
+      "learning_rate": 9.877709498914673e-06,
+      "loss": 0.4699,
+      "step": 33770
+    },
+    {
+      "epoch": 6.644537877111455,
+      "grad_norm": 1.1618289947509766,
+      "learning_rate": 9.87465223638754e-06,
+      "loss": 0.3393,
+      "step": 33780
+    },
+    {
+      "epoch": 6.646504880627474,
+      "grad_norm": 0.9800117015838623,
+      "learning_rate": 9.871594973860406e-06,
+      "loss": 0.3497,
+      "step": 33790
+    },
+    {
+      "epoch": 6.648471884143493,
+      "grad_norm": 1.332341194152832,
+      "learning_rate": 9.868537711333273e-06,
+      "loss": 0.4472,
+      "step": 33800
+    },
+    {
+      "epoch": 6.6504388876595115,
+      "grad_norm": 1.6299222707748413,
+      "learning_rate": 9.865480448806139e-06,
+      "loss": 0.4534,
+      "step": 33810
+    },
+    {
+      "epoch": 6.652405891175531,
+      "grad_norm": 0.9234393239021301,
+      "learning_rate": 9.862423186279007e-06,
+      "loss": 0.3266,
+      "step": 33820
+    },
+    {
+      "epoch": 6.65437289469155,
+      "grad_norm": 0.9883418679237366,
+      "learning_rate": 9.859365923751874e-06,
+      "loss": 0.2851,
+      "step": 33830
+    },
+    {
+      "epoch": 6.656339898207568,
+      "grad_norm": 1.3818767070770264,
+      "learning_rate": 9.85630866122474e-06,
+      "loss": 0.3963,
+      "step": 33840
+    },
+    {
+      "epoch": 6.658306901723587,
+      "grad_norm": 1.418997049331665,
+      "learning_rate": 9.853251398697607e-06,
+      "loss": 0.2881,
+      "step": 33850
+    },
+    {
+      "epoch": 6.660273905239606,
+      "grad_norm": 1.2970802783966064,
+      "learning_rate": 9.850194136170473e-06,
+      "loss": 0.2917,
+      "step": 33860
+    },
+    {
+      "epoch": 6.662240908755624,
+      "grad_norm": 1.2719851732254028,
+      "learning_rate": 9.84713687364334e-06,
+      "loss": 0.5373,
+      "step": 33870
+    },
+    {
+      "epoch": 6.664207912271643,
+      "grad_norm": 0.9323065280914307,
+      "learning_rate": 9.844079611116208e-06,
+      "loss": 0.3705,
+      "step": 33880
+    },
+    {
+      "epoch": 6.666174915787662,
+      "grad_norm": 1.5898349285125732,
+      "learning_rate": 9.841022348589074e-06,
+      "loss": 0.4218,
+      "step": 33890
+    },
+    {
+      "epoch": 6.668141919303681,
+      "grad_norm": 1.950779676437378,
+      "learning_rate": 9.837965086061941e-06,
+      "loss": 0.4501,
+      "step": 33900
+    },
+    {
+      "epoch": 6.6701089228197,
+      "grad_norm": 1.2006927728652954,
+      "learning_rate": 9.834907823534807e-06,
+      "loss": 0.2791,
+      "step": 33910
+    },
+    {
+      "epoch": 6.672075926335719,
+      "grad_norm": 4.144277572631836,
+      "learning_rate": 9.831850561007675e-06,
+      "loss": 0.3079,
+      "step": 33920
+    },
+    {
+      "epoch": 6.674042929851737,
+      "grad_norm": 0.7587095499038696,
+      "learning_rate": 9.828793298480542e-06,
+      "loss": 0.2736,
+      "step": 33930
+    },
+    {
+      "epoch": 6.676009933367756,
+      "grad_norm": 1.2794342041015625,
+      "learning_rate": 9.825736035953408e-06,
+      "loss": 0.3481,
+      "step": 33940
+    },
+    {
+      "epoch": 6.677976936883775,
+      "grad_norm": 1.8106298446655273,
+      "learning_rate": 9.822678773426275e-06,
+      "loss": 0.3745,
+      "step": 33950
+    },
+    {
+      "epoch": 6.679943940399793,
+      "grad_norm": 1.3771394491195679,
+      "learning_rate": 9.819621510899143e-06,
+      "loss": 0.3761,
+      "step": 33960
+    },
+    {
+      "epoch": 6.681910943915812,
+      "grad_norm": 2.139343023300171,
+      "learning_rate": 9.816564248372008e-06,
+      "loss": 0.3894,
+      "step": 33970
+    },
+    {
+      "epoch": 6.6838779474318315,
+      "grad_norm": 1.1938695907592773,
+      "learning_rate": 9.813506985844874e-06,
+      "loss": 0.4358,
+      "step": 33980
+    },
+    {
+      "epoch": 6.68584495094785,
+      "grad_norm": 1.9572737216949463,
+      "learning_rate": 9.810449723317742e-06,
+      "loss": 0.3695,
+      "step": 33990
+    },
+    {
+      "epoch": 6.687811954463869,
+      "grad_norm": 1.5220359563827515,
+      "learning_rate": 9.80739246079061e-06,
+      "loss": 0.4523,
+      "step": 34000
+    },
+    {
+      "epoch": 6.687811954463869,
+      "eval_loss": 0.14722880721092224,
+      "eval_runtime": 8.8579,
+      "eval_samples_per_second": 5.645,
+      "eval_steps_per_second": 2.822,
+      "step": 34000
+    },
+    {
+      "epoch": 6.689778957979888,
+      "grad_norm": 1.234450101852417,
+      "learning_rate": 9.804335198263477e-06,
+      "loss": 0.4261,
+      "step": 34010
+    },
+    {
+      "epoch": 6.691745961495906,
+      "grad_norm": 3.4291653633117676,
+      "learning_rate": 9.801277935736342e-06,
+      "loss": 0.4619,
+      "step": 34020
+    },
+    {
+      "epoch": 6.693712965011925,
+      "grad_norm": 1.2263903617858887,
+      "learning_rate": 9.798220673209208e-06,
+      "loss": 0.4102,
+      "step": 34030
+    },
+    {
+      "epoch": 6.695679968527944,
+      "grad_norm": 1.5767312049865723,
+      "learning_rate": 9.795163410682076e-06,
+      "loss": 0.4185,
+      "step": 34040
+    },
+    {
+      "epoch": 6.697646972043962,
+      "grad_norm": 1.0134525299072266,
+      "learning_rate": 9.792106148154943e-06,
+      "loss": 0.5154,
+      "step": 34050
+    },
+    {
+      "epoch": 6.6996139755599815,
+      "grad_norm": 1.2162753343582153,
+      "learning_rate": 9.78904888562781e-06,
+      "loss": 0.3059,
+      "step": 34060
+    },
+    {
+      "epoch": 6.701580979076001,
+      "grad_norm": 0.9273087382316589,
+      "learning_rate": 9.785991623100676e-06,
+      "loss": 0.5046,
+      "step": 34070
+    },
+    {
+      "epoch": 6.703547982592019,
+      "grad_norm": 1.6081489324569702,
+      "learning_rate": 9.782934360573542e-06,
+      "loss": 0.3274,
+      "step": 34080
+    },
+    {
+      "epoch": 6.705514986108038,
+      "grad_norm": 1.1477094888687134,
+      "learning_rate": 9.77987709804641e-06,
+      "loss": 0.3135,
+      "step": 34090
+    },
+    {
+      "epoch": 6.707481989624057,
+      "grad_norm": 1.948773980140686,
+      "learning_rate": 9.776819835519277e-06,
+      "loss": 0.3611,
+      "step": 34100
+    },
+    {
+      "epoch": 6.709448993140075,
+      "grad_norm": 0.7589294910430908,
+      "learning_rate": 9.773762572992143e-06,
+      "loss": 0.4675,
+      "step": 34110
+    },
+    {
+      "epoch": 6.711415996656094,
+      "grad_norm": 1.0926454067230225,
+      "learning_rate": 9.77070531046501e-06,
+      "loss": 0.4786,
+      "step": 34120
+    },
+    {
+      "epoch": 6.713383000172112,
+      "grad_norm": 1.9458638429641724,
+      "learning_rate": 9.767648047937878e-06,
+      "loss": 0.391,
+      "step": 34130
+    },
+    {
+      "epoch": 6.7153500036881315,
+      "grad_norm": 1.7874311208724976,
+      "learning_rate": 9.764590785410744e-06,
+      "loss": 0.3388,
+      "step": 34140
+    },
+    {
+      "epoch": 6.717317007204151,
+      "grad_norm": 1.0775543451309204,
+      "learning_rate": 9.761533522883611e-06,
+      "loss": 0.4601,
+      "step": 34150
+    },
+    {
+      "epoch": 6.719284010720169,
+      "grad_norm": 1.0132941007614136,
+      "learning_rate": 9.758476260356477e-06,
+      "loss": 0.3121,
+      "step": 34160
+    },
+    {
+      "epoch": 6.721251014236188,
+      "grad_norm": 0.6694428324699402,
+      "learning_rate": 9.755418997829344e-06,
+      "loss": 0.4602,
+      "step": 34170
+    },
+    {
+      "epoch": 6.723218017752207,
+      "grad_norm": 4.015652656555176,
+      "learning_rate": 9.752361735302212e-06,
+      "loss": 0.3528,
+      "step": 34180
+    },
+    {
+      "epoch": 6.725185021268225,
+      "grad_norm": 1.2746211290359497,
+      "learning_rate": 9.74930447277508e-06,
+      "loss": 0.2661,
+      "step": 34190
+    },
+    {
+      "epoch": 6.727152024784244,
+      "grad_norm": 0.7422463893890381,
+      "learning_rate": 9.746247210247945e-06,
+      "loss": 0.2808,
+      "step": 34200
+    },
+    {
+      "epoch": 6.729119028300263,
+      "grad_norm": 1.6404222249984741,
+      "learning_rate": 9.743189947720811e-06,
+      "loss": 0.2404,
+      "step": 34210
+    },
+    {
+      "epoch": 6.7310860318162815,
+      "grad_norm": 1.4843297004699707,
+      "learning_rate": 9.740132685193678e-06,
+      "loss": 0.4713,
+      "step": 34220
+    },
+    {
+      "epoch": 6.733053035332301,
+      "grad_norm": 0.8812487125396729,
+      "learning_rate": 9.737075422666546e-06,
+      "loss": 0.3954,
+      "step": 34230
+    },
+    {
+      "epoch": 6.73502003884832,
+      "grad_norm": 1.0680122375488281,
+      "learning_rate": 9.734018160139412e-06,
+      "loss": 0.3453,
+      "step": 34240
+    },
+    {
+      "epoch": 6.736987042364338,
+      "grad_norm": 0.6341516971588135,
+      "learning_rate": 9.730960897612279e-06,
+      "loss": 0.3423,
+      "step": 34250
+    },
+    {
+      "epoch": 6.738954045880357,
+      "grad_norm": 1.3379762172698975,
+      "learning_rate": 9.727903635085145e-06,
+      "loss": 0.3971,
+      "step": 34260
+    },
+    {
+      "epoch": 6.740921049396376,
+      "grad_norm": 0.9502032399177551,
+      "learning_rate": 9.724846372558012e-06,
+      "loss": 0.3367,
+      "step": 34270
+    },
+    {
+      "epoch": 6.742888052912394,
+      "grad_norm": 1.1735727787017822,
+      "learning_rate": 9.72178911003088e-06,
+      "loss": 0.3365,
+      "step": 34280
+    },
+    {
+      "epoch": 6.744855056428413,
+      "grad_norm": 1.3326324224472046,
+      "learning_rate": 9.718731847503746e-06,
+      "loss": 0.2607,
+      "step": 34290
+    },
+    {
+      "epoch": 6.746822059944432,
+      "grad_norm": 1.5648033618927002,
+      "learning_rate": 9.715674584976613e-06,
+      "loss": 0.3734,
+      "step": 34300
+    },
+    {
+      "epoch": 6.748789063460451,
+      "grad_norm": 1.3950005769729614,
+      "learning_rate": 9.712617322449479e-06,
+      "loss": 0.3373,
+      "step": 34310
+    },
+    {
+      "epoch": 6.75075606697647,
+      "grad_norm": 1.2418285608291626,
+      "learning_rate": 9.709560059922346e-06,
+      "loss": 0.4781,
+      "step": 34320
+    },
+    {
+      "epoch": 6.752723070492489,
+      "grad_norm": 1.5094794034957886,
+      "learning_rate": 9.706502797395214e-06,
+      "loss": 0.3345,
+      "step": 34330
+    },
+    {
+      "epoch": 6.754690074008507,
+      "grad_norm": 1.1631951332092285,
+      "learning_rate": 9.70344553486808e-06,
+      "loss": 0.2247,
+      "step": 34340
+    },
+    {
+      "epoch": 6.756657077524526,
+      "grad_norm": 2.2930450439453125,
+      "learning_rate": 9.700388272340947e-06,
+      "loss": 0.5198,
+      "step": 34350
+    },
+    {
+      "epoch": 6.758624081040545,
+      "grad_norm": 1.347464919090271,
+      "learning_rate": 9.697331009813814e-06,
+      "loss": 0.2888,
+      "step": 34360
+    },
+    {
+      "epoch": 6.760591084556563,
+      "grad_norm": 1.6012616157531738,
+      "learning_rate": 9.69427374728668e-06,
+      "loss": 0.4417,
+      "step": 34370
+    },
+    {
+      "epoch": 6.762558088072582,
+      "grad_norm": 1.6863954067230225,
+      "learning_rate": 9.691216484759548e-06,
+      "loss": 0.2953,
+      "step": 34380
+    },
+    {
+      "epoch": 6.7645250915886015,
+      "grad_norm": 1.4628210067749023,
+      "learning_rate": 9.688159222232413e-06,
+      "loss": 0.36,
+      "step": 34390
+    },
+    {
+      "epoch": 6.76649209510462,
+      "grad_norm": 1.8912512063980103,
+      "learning_rate": 9.685101959705281e-06,
+      "loss": 0.3969,
+      "step": 34400
+    },
+    {
+      "epoch": 6.768459098620639,
+      "grad_norm": 1.2532269954681396,
+      "learning_rate": 9.682044697178148e-06,
+      "loss": 0.4475,
+      "step": 34410
+    },
+    {
+      "epoch": 6.770426102136658,
+      "grad_norm": 1.3984140157699585,
+      "learning_rate": 9.678987434651014e-06,
+      "loss": 0.3566,
+      "step": 34420
+    },
+    {
+      "epoch": 6.772393105652676,
+      "grad_norm": 0.7679744958877563,
+      "learning_rate": 9.67593017212388e-06,
+      "loss": 0.3454,
+      "step": 34430
+    },
+    {
+      "epoch": 6.774360109168695,
+      "grad_norm": 0.7441784739494324,
+      "learning_rate": 9.672872909596747e-06,
+      "loss": 0.3793,
+      "step": 34440
+    },
+    {
+      "epoch": 6.776327112684714,
+      "grad_norm": 1.1137535572052002,
+      "learning_rate": 9.669815647069615e-06,
+      "loss": 0.4104,
+      "step": 34450
+    },
+    {
+      "epoch": 6.778294116200732,
+      "grad_norm": 8.478157997131348,
+      "learning_rate": 9.666758384542482e-06,
+      "loss": 0.337,
+      "step": 34460
+    },
+    {
+      "epoch": 6.7802611197167515,
+      "grad_norm": 0.7802620530128479,
+      "learning_rate": 9.663701122015348e-06,
+      "loss": 0.4132,
+      "step": 34470
+    },
+    {
+      "epoch": 6.782228123232771,
+      "grad_norm": 1.2110645771026611,
+      "learning_rate": 9.660643859488214e-06,
+      "loss": 0.3649,
+      "step": 34480
+    },
+    {
+      "epoch": 6.784195126748789,
+      "grad_norm": 0.8571938872337341,
+      "learning_rate": 9.657586596961081e-06,
+      "loss": 0.3435,
+      "step": 34490
+    },
+    {
+      "epoch": 6.786162130264808,
+      "grad_norm": 0.9900618195533752,
+      "learning_rate": 9.654529334433949e-06,
+      "loss": 0.3434,
+      "step": 34500
+    },
+    {
+      "epoch": 6.786162130264808,
+      "eval_loss": 0.14516228437423706,
+      "eval_runtime": 8.8825,
+      "eval_samples_per_second": 5.629,
+      "eval_steps_per_second": 2.815,
+      "step": 34500
+    },
+    {
+      "epoch": 6.788129133780827,
+      "grad_norm": 1.6814920902252197,
+      "learning_rate": 9.651472071906816e-06,
+      "loss": 0.4392,
+      "step": 34510
+    },
+    {
+      "epoch": 6.790096137296845,
+      "grad_norm": 2.54838490486145,
+      "learning_rate": 9.648414809379682e-06,
+      "loss": 0.2895,
+      "step": 34520
+    },
+    {
+      "epoch": 6.792063140812864,
+      "grad_norm": 1.4882313013076782,
+      "learning_rate": 9.64535754685255e-06,
+      "loss": 0.4699,
+      "step": 34530
+    },
+    {
+      "epoch": 6.794030144328883,
+      "grad_norm": 1.2992831468582153,
+      "learning_rate": 9.642300284325415e-06,
+      "loss": 0.3009,
+      "step": 34540
+    },
+    {
+      "epoch": 6.7959971478449015,
+      "grad_norm": 1.3498175144195557,
+      "learning_rate": 9.639243021798283e-06,
+      "loss": 0.5047,
+      "step": 34550
+    },
+    {
+      "epoch": 6.797964151360921,
+      "grad_norm": 0.9251207113265991,
+      "learning_rate": 9.636185759271149e-06,
+      "loss": 0.3042,
+      "step": 34560
+    },
+    {
+      "epoch": 6.79993115487694,
+      "grad_norm": 1.6040048599243164,
+      "learning_rate": 9.633128496744016e-06,
+      "loss": 0.3682,
+      "step": 34570
+    },
+    {
+      "epoch": 6.801898158392958,
+      "grad_norm": 0.9295186996459961,
+      "learning_rate": 9.630071234216884e-06,
+      "loss": 0.2468,
+      "step": 34580
+    },
+    {
+      "epoch": 6.803865161908977,
+      "grad_norm": 1.454932451248169,
+      "learning_rate": 9.62701397168975e-06,
+      "loss": 0.4661,
+      "step": 34590
+    },
+    {
+      "epoch": 6.805832165424996,
+      "grad_norm": 0.9594427347183228,
+      "learning_rate": 9.623956709162617e-06,
+      "loss": 0.3591,
+      "step": 34600
+    },
+    {
+      "epoch": 6.807799168941014,
+      "grad_norm": 1.4732391834259033,
+      "learning_rate": 9.620899446635483e-06,
+      "loss": 0.3109,
+      "step": 34610
+    },
+    {
+      "epoch": 6.809766172457033,
+      "grad_norm": 1.8341666460037231,
+      "learning_rate": 9.61784218410835e-06,
+      "loss": 0.2608,
+      "step": 34620
+    },
+    {
+      "epoch": 6.811733175973052,
+      "grad_norm": 1.3303409814834595,
+      "learning_rate": 9.614784921581218e-06,
+      "loss": 0.3969,
+      "step": 34630
+    },
+    {
+      "epoch": 6.813700179489071,
+      "grad_norm": 2.279348850250244,
+      "learning_rate": 9.611727659054085e-06,
+      "loss": 0.3592,
+      "step": 34640
+    },
+    {
+      "epoch": 6.81566718300509,
+      "grad_norm": 1.6453670263290405,
+      "learning_rate": 9.60867039652695e-06,
+      "loss": 0.4212,
+      "step": 34650
+    },
+    {
+      "epoch": 6.817634186521109,
+      "grad_norm": 1.5812371969223022,
+      "learning_rate": 9.605613133999817e-06,
+      "loss": 0.3665,
+      "step": 34660
+    },
+    {
+      "epoch": 6.819601190037127,
+      "grad_norm": 0.9130092263221741,
+      "learning_rate": 9.602555871472684e-06,
+      "loss": 0.362,
+      "step": 34670
+    },
+    {
+      "epoch": 6.821568193553146,
+      "grad_norm": 0.9470576047897339,
+      "learning_rate": 9.599498608945552e-06,
+      "loss": 0.2942,
+      "step": 34680
+    },
+    {
+      "epoch": 6.823535197069165,
+      "grad_norm": 0.9792674779891968,
+      "learning_rate": 9.596441346418417e-06,
+      "loss": 0.388,
+      "step": 34690
+    },
+    {
+      "epoch": 6.825502200585183,
+      "grad_norm": 1.5535203218460083,
+      "learning_rate": 9.593384083891285e-06,
+      "loss": 0.321,
+      "step": 34700
+    },
+    {
+      "epoch": 6.827469204101202,
+      "grad_norm": 1.3171846866607666,
+      "learning_rate": 9.59032682136415e-06,
+      "loss": 0.3896,
+      "step": 34710
+    },
+    {
+      "epoch": 6.8294362076172215,
+      "grad_norm": 1.0758510828018188,
+      "learning_rate": 9.587269558837018e-06,
+      "loss": 0.3583,
+      "step": 34720
+    },
+    {
+      "epoch": 6.83140321113324,
+      "grad_norm": 0.9125159978866577,
+      "learning_rate": 9.584212296309886e-06,
+      "loss": 0.3526,
+      "step": 34730
+    },
+    {
+      "epoch": 6.833370214649259,
+      "grad_norm": 1.0016282796859741,
+      "learning_rate": 9.581155033782751e-06,
+      "loss": 0.2809,
+      "step": 34740
+    },
+    {
+      "epoch": 6.835337218165278,
+      "grad_norm": 1.18870210647583,
+      "learning_rate": 9.578097771255619e-06,
+      "loss": 0.4357,
+      "step": 34750
+    },
+    {
+      "epoch": 6.837304221681296,
+      "grad_norm": 1.2393519878387451,
+      "learning_rate": 9.575040508728485e-06,
+      "loss": 0.2914,
+      "step": 34760
+    },
+    {
+      "epoch": 6.839271225197315,
+      "grad_norm": 0.9247721433639526,
+      "learning_rate": 9.571983246201352e-06,
+      "loss": 0.3452,
+      "step": 34770
+    },
+    {
+      "epoch": 6.841238228713334,
+      "grad_norm": 0.6819230318069458,
+      "learning_rate": 9.56892598367422e-06,
+      "loss": 0.428,
+      "step": 34780
+    },
+    {
+      "epoch": 6.843205232229352,
+      "grad_norm": 0.8827270269393921,
+      "learning_rate": 9.565868721147085e-06,
+      "loss": 0.4428,
+      "step": 34790
+    },
+    {
+      "epoch": 6.8451722357453715,
+      "grad_norm": 1.2155219316482544,
+      "learning_rate": 9.562811458619953e-06,
+      "loss": 0.3088,
+      "step": 34800
+    },
+    {
+      "epoch": 6.847139239261391,
+      "grad_norm": 1.844770908355713,
+      "learning_rate": 9.55975419609282e-06,
+      "loss": 0.3482,
+      "step": 34810
+    },
+    {
+      "epoch": 6.849106242777409,
+      "grad_norm": 1.582555890083313,
+      "learning_rate": 9.556696933565686e-06,
+      "loss": 0.3297,
+      "step": 34820
+    },
+    {
+      "epoch": 6.851073246293428,
+      "grad_norm": 0.9563121795654297,
+      "learning_rate": 9.553639671038552e-06,
+      "loss": 0.4846,
+      "step": 34830
+    },
+    {
+      "epoch": 6.853040249809446,
+      "grad_norm": 1.4238522052764893,
+      "learning_rate": 9.55058240851142e-06,
+      "loss": 0.3762,
+      "step": 34840
+    },
+    {
+      "epoch": 6.855007253325465,
+      "grad_norm": 1.6920660734176636,
+      "learning_rate": 9.547525145984287e-06,
+      "loss": 0.4887,
+      "step": 34850
+    },
+    {
+      "epoch": 6.856974256841484,
+      "grad_norm": 1.3875815868377686,
+      "learning_rate": 9.544467883457154e-06,
+      "loss": 0.5668,
+      "step": 34860
+    },
+    {
+      "epoch": 6.858941260357502,
+      "grad_norm": 0.7671670913696289,
+      "learning_rate": 9.54141062093002e-06,
+      "loss": 0.3215,
+      "step": 34870
+    },
+    {
+      "epoch": 6.8609082638735215,
+      "grad_norm": 0.8364396691322327,
+      "learning_rate": 9.538353358402886e-06,
+      "loss": 0.3085,
+      "step": 34880
+    },
+    {
+      "epoch": 6.862875267389541,
+      "grad_norm": 1.3606170415878296,
+      "learning_rate": 9.535296095875753e-06,
+      "loss": 0.2713,
+      "step": 34890
+    },
+    {
+      "epoch": 6.864842270905559,
+      "grad_norm": 1.0086907148361206,
+      "learning_rate": 9.53223883334862e-06,
+      "loss": 0.3078,
+      "step": 34900
+    },
+    {
+      "epoch": 6.866809274421578,
+      "grad_norm": 1.5053181648254395,
+      "learning_rate": 9.529181570821488e-06,
+      "loss": 0.4035,
+      "step": 34910
+    },
+    {
+      "epoch": 6.868776277937597,
+      "grad_norm": 0.9070050716400146,
+      "learning_rate": 9.526124308294354e-06,
+      "loss": 0.343,
+      "step": 34920
+    },
+    {
+      "epoch": 6.870743281453615,
+      "grad_norm": 1.2644065618515015,
+      "learning_rate": 9.52306704576722e-06,
+      "loss": 0.2343,
+      "step": 34930
+    },
+    {
+      "epoch": 6.872710284969634,
+      "grad_norm": 0.9814834594726562,
+      "learning_rate": 9.520009783240087e-06,
+      "loss": 0.393,
+      "step": 34940
+    },
+    {
+      "epoch": 6.874677288485653,
+      "grad_norm": 1.202989101409912,
+      "learning_rate": 9.516952520712955e-06,
+      "loss": 0.4422,
+      "step": 34950
+    },
+    {
+      "epoch": 6.8766442920016715,
+      "grad_norm": 2.017094612121582,
+      "learning_rate": 9.51389525818582e-06,
+      "loss": 0.3723,
+      "step": 34960
+    },
+    {
+      "epoch": 6.878611295517691,
+      "grad_norm": 1.194692850112915,
+      "learning_rate": 9.510837995658688e-06,
+      "loss": 0.3942,
+      "step": 34970
+    },
+    {
+      "epoch": 6.88057829903371,
+      "grad_norm": 1.6615229845046997,
+      "learning_rate": 9.507780733131555e-06,
+      "loss": 0.3137,
+      "step": 34980
+    },
+    {
+      "epoch": 6.882545302549728,
+      "grad_norm": 0.8167328238487244,
+      "learning_rate": 9.504723470604421e-06,
+      "loss": 0.3133,
+      "step": 34990
+    },
+    {
+      "epoch": 6.884512306065747,
+      "grad_norm": 0.8303702473640442,
+      "learning_rate": 9.501666208077289e-06,
+      "loss": 0.3374,
+      "step": 35000
+    },
+    {
+      "epoch": 6.884512306065747,
+      "eval_loss": 0.14276579022407532,
+      "eval_runtime": 8.8481,
+      "eval_samples_per_second": 5.651,
+      "eval_steps_per_second": 2.825,
+      "step": 35000
+    },
+    {
+      "epoch": 6.886479309581766,
+      "grad_norm": 1.779717206954956,
+      "learning_rate": 9.498608945550154e-06,
+      "loss": 0.4124,
+      "step": 35010
+    },
+    {
+      "epoch": 6.888446313097784,
+      "grad_norm": 1.0585066080093384,
+      "learning_rate": 9.495551683023022e-06,
+      "loss": 0.3539,
+      "step": 35020
+    },
+    {
+      "epoch": 6.890413316613803,
+      "grad_norm": 2.3185861110687256,
+      "learning_rate": 9.49249442049589e-06,
+      "loss": 0.3541,
+      "step": 35030
+    },
+    {
+      "epoch": 6.892380320129822,
+      "grad_norm": 1.2141361236572266,
+      "learning_rate": 9.489437157968757e-06,
+      "loss": 0.3086,
+      "step": 35040
+    },
+    {
+      "epoch": 6.894347323645841,
+      "grad_norm": 2.0195775032043457,
+      "learning_rate": 9.486379895441623e-06,
+      "loss": 0.4795,
+      "step": 35050
+    },
+    {
+      "epoch": 6.89631432716186,
+      "grad_norm": 1.4040886163711548,
+      "learning_rate": 9.483322632914488e-06,
+      "loss": 0.3061,
+      "step": 35060
+    },
+    {
+      "epoch": 6.898281330677879,
+      "grad_norm": 0.7142741084098816,
+      "learning_rate": 9.480265370387356e-06,
+      "loss": 0.278,
+      "step": 35070
+    },
+    {
+      "epoch": 6.900248334193897,
+      "grad_norm": 1.6775517463684082,
+      "learning_rate": 9.477208107860223e-06,
+      "loss": 0.3032,
+      "step": 35080
+    },
+    {
+      "epoch": 6.902215337709916,
+      "grad_norm": 1.7707325220108032,
+      "learning_rate": 9.474150845333089e-06,
+      "loss": 0.4228,
+      "step": 35090
+    },
+    {
+      "epoch": 6.904182341225935,
+      "grad_norm": 1.264266848564148,
+      "learning_rate": 9.471093582805957e-06,
+      "loss": 0.4042,
+      "step": 35100
+    },
+    {
+      "epoch": 6.906149344741953,
+      "grad_norm": 2.7818541526794434,
+      "learning_rate": 9.468036320278822e-06,
+      "loss": 0.4654,
+      "step": 35110
+    },
+    {
+      "epoch": 6.908116348257972,
+      "grad_norm": 1.321373701095581,
+      "learning_rate": 9.46497905775169e-06,
+      "loss": 0.4151,
+      "step": 35120
+    },
+    {
+      "epoch": 6.9100833517739915,
+      "grad_norm": 0.5779895186424255,
+      "learning_rate": 9.461921795224557e-06,
+      "loss": 0.3116,
+      "step": 35130
+    },
+    {
+      "epoch": 6.91205035529001,
+      "grad_norm": 2.980159044265747,
+      "learning_rate": 9.458864532697423e-06,
+      "loss": 0.3251,
+      "step": 35140
+    },
+    {
+      "epoch": 6.914017358806029,
+      "grad_norm": 0.6384599804878235,
+      "learning_rate": 9.45580727017029e-06,
+      "loss": 0.3456,
+      "step": 35150
+    },
+    {
+      "epoch": 6.915984362322048,
+      "grad_norm": 0.8337164521217346,
+      "learning_rate": 9.452750007643156e-06,
+      "loss": 0.3414,
+      "step": 35160
+    },
+    {
+      "epoch": 6.917951365838066,
+      "grad_norm": 0.8615796566009521,
+      "learning_rate": 9.449692745116024e-06,
+      "loss": 0.3019,
+      "step": 35170
+    },
+    {
+      "epoch": 6.919918369354085,
+      "grad_norm": 1.3931223154067993,
+      "learning_rate": 9.446635482588891e-06,
+      "loss": 0.2995,
+      "step": 35180
+    },
+    {
+      "epoch": 6.921885372870104,
+      "grad_norm": 0.6666834354400635,
+      "learning_rate": 9.443578220061757e-06,
+      "loss": 0.2891,
+      "step": 35190
+    },
+    {
+      "epoch": 6.923852376386122,
+      "grad_norm": 2.166940927505493,
+      "learning_rate": 9.440520957534625e-06,
+      "loss": 0.4111,
+      "step": 35200
+    },
+    {
+      "epoch": 6.9258193799021415,
+      "grad_norm": 1.2376841306686401,
+      "learning_rate": 9.437463695007492e-06,
+      "loss": 0.4181,
+      "step": 35210
+    },
+    {
+      "epoch": 6.9277863834181606,
+      "grad_norm": 1.8692290782928467,
+      "learning_rate": 9.434406432480358e-06,
+      "loss": 0.3888,
+      "step": 35220
+    },
+    {
+      "epoch": 6.929753386934179,
+      "grad_norm": 1.0836957693099976,
+      "learning_rate": 9.431349169953225e-06,
+      "loss": 0.4205,
+      "step": 35230
+    },
+    {
+      "epoch": 6.931720390450198,
+      "grad_norm": 1.6166654825210571,
+      "learning_rate": 9.428291907426091e-06,
+      "loss": 0.4591,
+      "step": 35240
+    },
+    {
+      "epoch": 6.933687393966217,
+      "grad_norm": 0.9476025104522705,
+      "learning_rate": 9.425234644898958e-06,
+      "loss": 0.3028,
+      "step": 35250
+    },
+    {
+      "epoch": 6.935654397482235,
+      "grad_norm": 1.6330329179763794,
+      "learning_rate": 9.422177382371826e-06,
+      "loss": 0.3836,
+      "step": 35260
+    },
+    {
+      "epoch": 6.937621400998254,
+      "grad_norm": 1.3539565801620483,
+      "learning_rate": 9.419120119844692e-06,
+      "loss": 0.3086,
+      "step": 35270
+    },
+    {
+      "epoch": 6.939588404514273,
+      "grad_norm": 2.6535205841064453,
+      "learning_rate": 9.416062857317558e-06,
+      "loss": 0.3566,
+      "step": 35280
+    },
+    {
+      "epoch": 6.9415554080302915,
+      "grad_norm": 0.5615221858024597,
+      "learning_rate": 9.413005594790425e-06,
+      "loss": 0.3996,
+      "step": 35290
+    },
+    {
+      "epoch": 6.943522411546311,
+      "grad_norm": 1.2192131280899048,
+      "learning_rate": 9.409948332263292e-06,
+      "loss": 0.4368,
+      "step": 35300
+    },
+    {
+      "epoch": 6.94548941506233,
+      "grad_norm": 3.1623566150665283,
+      "learning_rate": 9.40689106973616e-06,
+      "loss": 0.2673,
+      "step": 35310
+    },
+    {
+      "epoch": 6.947456418578348,
+      "grad_norm": 0.8841261267662048,
+      "learning_rate": 9.403833807209026e-06,
+      "loss": 0.3659,
+      "step": 35320
+    },
+    {
+      "epoch": 6.949423422094367,
+      "grad_norm": 1.4013664722442627,
+      "learning_rate": 9.400776544681891e-06,
+      "loss": 0.5024,
+      "step": 35330
+    },
+    {
+      "epoch": 6.951390425610386,
+      "grad_norm": 1.311947226524353,
+      "learning_rate": 9.397719282154759e-06,
+      "loss": 0.4321,
+      "step": 35340
+    },
+    {
+      "epoch": 6.953357429126404,
+      "grad_norm": 3.628854274749756,
+      "learning_rate": 9.394662019627626e-06,
+      "loss": 0.3072,
+      "step": 35350
+    },
+    {
+      "epoch": 6.955324432642423,
+      "grad_norm": 1.5111637115478516,
+      "learning_rate": 9.391604757100494e-06,
+      "loss": 0.3502,
+      "step": 35360
+    },
+    {
+      "epoch": 6.957291436158442,
+      "grad_norm": 1.093699336051941,
+      "learning_rate": 9.38854749457336e-06,
+      "loss": 0.5111,
+      "step": 35370
+    },
+    {
+      "epoch": 6.959258439674461,
+      "grad_norm": 1.2105228900909424,
+      "learning_rate": 9.385490232046227e-06,
+      "loss": 0.5082,
+      "step": 35380
+    },
+    {
+      "epoch": 6.96122544319048,
+      "grad_norm": 0.9326910376548767,
+      "learning_rate": 9.382432969519093e-06,
+      "loss": 0.263,
+      "step": 35390
+    },
+    {
+      "epoch": 6.963192446706499,
+      "grad_norm": 2.3566551208496094,
+      "learning_rate": 9.37937570699196e-06,
+      "loss": 0.3236,
+      "step": 35400
+    },
+    {
+      "epoch": 6.965159450222517,
+      "grad_norm": 0.6763259768486023,
+      "learning_rate": 9.376318444464826e-06,
+      "loss": 0.2762,
+      "step": 35410
+    },
+    {
+      "epoch": 6.967126453738536,
+      "grad_norm": 1.633030891418457,
+      "learning_rate": 9.373261181937694e-06,
+      "loss": 0.4472,
+      "step": 35420
+    },
+    {
+      "epoch": 6.969093457254555,
+      "grad_norm": 2.4877285957336426,
+      "learning_rate": 9.370203919410561e-06,
+      "loss": 0.2424,
+      "step": 35430
+    },
+    {
+      "epoch": 6.971060460770573,
+      "grad_norm": 0.9804292917251587,
+      "learning_rate": 9.367146656883427e-06,
+      "loss": 0.3773,
+      "step": 35440
+    },
+    {
+      "epoch": 6.973027464286592,
+      "grad_norm": 1.5808591842651367,
+      "learning_rate": 9.364089394356294e-06,
+      "loss": 0.3392,
+      "step": 35450
+    },
+    {
+      "epoch": 6.9749944678026115,
+      "grad_norm": 1.5255887508392334,
+      "learning_rate": 9.36103213182916e-06,
+      "loss": 0.4773,
+      "step": 35460
+    },
+    {
+      "epoch": 6.97696147131863,
+      "grad_norm": 0.9557861685752869,
+      "learning_rate": 9.357974869302028e-06,
+      "loss": 0.3166,
+      "step": 35470
+    },
+    {
+      "epoch": 6.978928474834649,
+      "grad_norm": 1.788199782371521,
+      "learning_rate": 9.354917606774895e-06,
+      "loss": 0.4094,
+      "step": 35480
+    },
+    {
+      "epoch": 6.980895478350668,
+      "grad_norm": 0.9672784805297852,
+      "learning_rate": 9.351860344247763e-06,
+      "loss": 0.3216,
+      "step": 35490
+    },
+    {
+      "epoch": 6.982862481866686,
+      "grad_norm": 1.6953893899917603,
+      "learning_rate": 9.348803081720628e-06,
+      "loss": 0.5089,
+      "step": 35500
+    },
+    {
+      "epoch": 6.982862481866686,
+      "eval_loss": 0.14116249978542328,
+      "eval_runtime": 8.8785,
+      "eval_samples_per_second": 5.632,
+      "eval_steps_per_second": 2.816,
+      "step": 35500
+    },
+    {
+      "epoch": 6.984829485382705,
+      "grad_norm": 0.8304757475852966,
+      "learning_rate": 9.345745819193494e-06,
+      "loss": 0.2787,
+      "step": 35510
+    },
+    {
+      "epoch": 6.986796488898724,
+      "grad_norm": 0.7213247418403625,
+      "learning_rate": 9.342688556666362e-06,
+      "loss": 0.4252,
+      "step": 35520
+    },
+    {
+      "epoch": 6.988763492414742,
+      "grad_norm": 2.7171456813812256,
+      "learning_rate": 9.339631294139229e-06,
+      "loss": 0.5031,
+      "step": 35530
+    },
+    {
+      "epoch": 6.9907304959307615,
+      "grad_norm": 1.4926707744598389,
+      "learning_rate": 9.336574031612095e-06,
+      "loss": 0.3693,
+      "step": 35540
+    },
+    {
+      "epoch": 6.9926974994467805,
+      "grad_norm": 1.0265848636627197,
+      "learning_rate": 9.333516769084962e-06,
+      "loss": 0.3407,
+      "step": 35550
+    },
+    {
+      "epoch": 6.994664502962799,
+      "grad_norm": 1.8866267204284668,
+      "learning_rate": 9.330459506557828e-06,
+      "loss": 0.3216,
+      "step": 35560
+    },
+    {
+      "epoch": 6.996631506478818,
+      "grad_norm": 3.09433913230896,
+      "learning_rate": 9.327402244030696e-06,
+      "loss": 0.3974,
+      "step": 35570
+    },
+    {
+      "epoch": 6.998598509994837,
+      "grad_norm": 2.6518099308013916,
+      "learning_rate": 9.324344981503563e-06,
+      "loss": 0.2656,
+      "step": 35580
+    },
+    {
+      "epoch": 7.000565513510855,
+      "grad_norm": 0.38323965668678284,
+      "learning_rate": 9.321287718976429e-06,
+      "loss": 0.3545,
+      "step": 35590
+    },
+    {
+      "epoch": 7.002532517026874,
+      "grad_norm": 0.8496131896972656,
+      "learning_rate": 9.318230456449296e-06,
+      "loss": 0.2795,
+      "step": 35600
+    },
+    {
+      "epoch": 7.004499520542893,
+      "grad_norm": 1.9276858568191528,
+      "learning_rate": 9.315173193922162e-06,
+      "loss": 0.3878,
+      "step": 35610
+    },
+    {
+      "epoch": 7.0064665240589115,
+      "grad_norm": 0.9172375798225403,
+      "learning_rate": 9.31211593139503e-06,
+      "loss": 0.3019,
+      "step": 35620
+    },
+    {
+      "epoch": 7.0084335275749305,
+      "grad_norm": 1.2532397508621216,
+      "learning_rate": 9.309058668867897e-06,
+      "loss": 0.37,
+      "step": 35630
+    },
+    {
+      "epoch": 7.01040053109095,
+      "grad_norm": 1.1102087497711182,
+      "learning_rate": 9.306001406340763e-06,
+      "loss": 0.3783,
+      "step": 35640
+    },
+    {
+      "epoch": 7.012367534606968,
+      "grad_norm": 1.4500218629837036,
+      "learning_rate": 9.30294414381363e-06,
+      "loss": 0.4186,
+      "step": 35650
+    },
+    {
+      "epoch": 7.014334538122987,
+      "grad_norm": 1.0918534994125366,
+      "learning_rate": 9.299886881286498e-06,
+      "loss": 0.4503,
+      "step": 35660
+    },
+    {
+      "epoch": 7.016301541639006,
+      "grad_norm": 0.8374407291412354,
+      "learning_rate": 9.296829618759364e-06,
+      "loss": 0.2792,
+      "step": 35670
+    },
+    {
+      "epoch": 7.018268545155024,
+      "grad_norm": 0.924603283405304,
+      "learning_rate": 9.293772356232231e-06,
+      "loss": 0.4561,
+      "step": 35680
+    },
+    {
+      "epoch": 7.020235548671043,
+      "grad_norm": 0.7381618618965149,
+      "learning_rate": 9.290715093705097e-06,
+      "loss": 0.3591,
+      "step": 35690
+    },
+    {
+      "epoch": 7.022202552187062,
+      "grad_norm": 1.1523455381393433,
+      "learning_rate": 9.287657831177964e-06,
+      "loss": 0.2453,
+      "step": 35700
+    },
+    {
+      "epoch": 7.0241695557030805,
+      "grad_norm": 1.0780432224273682,
+      "learning_rate": 9.284600568650832e-06,
+      "loss": 0.3303,
+      "step": 35710
+    },
+    {
+      "epoch": 7.0261365592191,
+      "grad_norm": 0.5242741703987122,
+      "learning_rate": 9.281543306123697e-06,
+      "loss": 0.1654,
+      "step": 35720
+    },
+    {
+      "epoch": 7.028103562735119,
+      "grad_norm": 0.6786366701126099,
+      "learning_rate": 9.278486043596563e-06,
+      "loss": 0.3589,
+      "step": 35730
+    },
+    {
+      "epoch": 7.030070566251137,
+      "grad_norm": 0.8653206825256348,
+      "learning_rate": 9.27542878106943e-06,
+      "loss": 0.3865,
+      "step": 35740
+    },
+    {
+      "epoch": 7.032037569767156,
+      "grad_norm": 1.2041491270065308,
+      "learning_rate": 9.272371518542298e-06,
+      "loss": 0.3778,
+      "step": 35750
+    },
+    {
+      "epoch": 7.034004573283175,
+      "grad_norm": 0.9845544695854187,
+      "learning_rate": 9.269314256015166e-06,
+      "loss": 0.3827,
+      "step": 35760
+    },
+    {
+      "epoch": 7.035971576799193,
+      "grad_norm": 0.5570571422576904,
+      "learning_rate": 9.266256993488031e-06,
+      "loss": 0.3254,
+      "step": 35770
+    },
+    {
+      "epoch": 7.037938580315212,
+      "grad_norm": 0.9653236269950867,
+      "learning_rate": 9.263199730960897e-06,
+      "loss": 0.4093,
+      "step": 35780
+    },
+    {
+      "epoch": 7.039905583831231,
+      "grad_norm": 1.872336745262146,
+      "learning_rate": 9.260142468433765e-06,
+      "loss": 0.3385,
+      "step": 35790
+    },
+    {
+      "epoch": 7.04187258734725,
+      "grad_norm": 1.1234791278839111,
+      "learning_rate": 9.257085205906632e-06,
+      "loss": 0.3464,
+      "step": 35800
+    },
+    {
+      "epoch": 7.043839590863269,
+      "grad_norm": 1.0836743116378784,
+      "learning_rate": 9.2540279433795e-06,
+      "loss": 0.3793,
+      "step": 35810
+    },
+    {
+      "epoch": 7.045806594379288,
+      "grad_norm": 0.8646442890167236,
+      "learning_rate": 9.250970680852365e-06,
+      "loss": 0.3513,
+      "step": 35820
+    },
+    {
+      "epoch": 7.047773597895306,
+      "grad_norm": 0.9303305745124817,
+      "learning_rate": 9.247913418325233e-06,
+      "loss": 0.4035,
+      "step": 35830
+    },
+    {
+      "epoch": 7.049740601411325,
+      "grad_norm": 1.3007382154464722,
+      "learning_rate": 9.244856155798099e-06,
+      "loss": 0.3699,
+      "step": 35840
+    },
+    {
+      "epoch": 7.051707604927344,
+      "grad_norm": 1.8843724727630615,
+      "learning_rate": 9.241798893270966e-06,
+      "loss": 0.3185,
+      "step": 35850
+    },
+    {
+      "epoch": 7.053674608443362,
+      "grad_norm": 1.665401577949524,
+      "learning_rate": 9.238741630743832e-06,
+      "loss": 0.3567,
+      "step": 35860
+    },
+    {
+      "epoch": 7.055641611959381,
+      "grad_norm": 0.9232697486877441,
+      "learning_rate": 9.2356843682167e-06,
+      "loss": 0.3165,
+      "step": 35870
+    },
+    {
+      "epoch": 7.0576086154754005,
+      "grad_norm": 1.5284390449523926,
+      "learning_rate": 9.232627105689567e-06,
+      "loss": 0.2676,
+      "step": 35880
+    },
+    {
+      "epoch": 7.059575618991419,
+      "grad_norm": 0.810941219329834,
+      "learning_rate": 9.229569843162434e-06,
+      "loss": 0.4656,
+      "step": 35890
+    },
+    {
+      "epoch": 7.061542622507438,
+      "grad_norm": 0.7167800068855286,
+      "learning_rate": 9.2265125806353e-06,
+      "loss": 0.3008,
+      "step": 35900
+    },
+    {
+      "epoch": 7.063509626023457,
+      "grad_norm": 0.9715979099273682,
+      "learning_rate": 9.223455318108166e-06,
+      "loss": 0.3422,
+      "step": 35910
+    },
+    {
+      "epoch": 7.065476629539475,
+      "grad_norm": 2.699817657470703,
+      "learning_rate": 9.220398055581033e-06,
+      "loss": 0.35,
+      "step": 35920
+    },
+    {
+      "epoch": 7.067443633055494,
+      "grad_norm": 1.0953922271728516,
+      "learning_rate": 9.2173407930539e-06,
+      "loss": 0.3478,
+      "step": 35930
+    },
+    {
+      "epoch": 7.069410636571513,
+      "grad_norm": 0.5451570749282837,
+      "learning_rate": 9.214283530526768e-06,
+      "loss": 0.3762,
+      "step": 35940
+    },
+    {
+      "epoch": 7.0713776400875314,
+      "grad_norm": 1.066961407661438,
+      "learning_rate": 9.211226267999634e-06,
+      "loss": 0.3103,
+      "step": 35950
+    },
+    {
+      "epoch": 7.0733446436035505,
+      "grad_norm": 1.1819201707839966,
+      "learning_rate": 9.2081690054725e-06,
+      "loss": 0.3374,
+      "step": 35960
+    },
+    {
+      "epoch": 7.07531164711957,
+      "grad_norm": 0.8258270621299744,
+      "learning_rate": 9.205111742945367e-06,
+      "loss": 0.3513,
+      "step": 35970
+    },
+    {
+      "epoch": 7.077278650635588,
+      "grad_norm": 1.4958642721176147,
+      "learning_rate": 9.202054480418235e-06,
+      "loss": 0.3744,
+      "step": 35980
+    },
+    {
+      "epoch": 7.079245654151607,
+      "grad_norm": 0.4954798221588135,
+      "learning_rate": 9.1989972178911e-06,
+      "loss": 0.3777,
+      "step": 35990
+    },
+    {
+      "epoch": 7.081212657667626,
+      "grad_norm": 0.4899837076663971,
+      "learning_rate": 9.195939955363968e-06,
+      "loss": 0.243,
+      "step": 36000
+    },
+    {
+      "epoch": 7.081212657667626,
+      "eval_loss": 0.14319463074207306,
+      "eval_runtime": 8.8442,
+      "eval_samples_per_second": 5.653,
+      "eval_steps_per_second": 2.827,
+      "step": 36000
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 66079,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 13,
+  "save_steps": 4000,
+  "total_flos": 1.9414670099987497e+18,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}