diff --git "a/checkpoint-6828/trainer_state.json" "b/checkpoint-6828/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-6828/trainer_state.json"
@@ -0,0 +1,4808 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.0,
+  "eval_steps": 500,
+  "global_step": 6828,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.005858230814294083,
+      "grad_norm": 14.95982551574707,
+      "learning_rate": 1.9978910369068542e-05,
+      "loss": 4.3468,
+      "step": 10
+    },
+    {
+      "epoch": 0.011716461628588167,
+      "grad_norm": 19.284658432006836,
+      "learning_rate": 1.9955477445811366e-05,
+      "loss": 4.6374,
+      "step": 20
+    },
+    {
+      "epoch": 0.01757469244288225,
+      "grad_norm": 11.212137222290039,
+      "learning_rate": 1.993204452255419e-05,
+      "loss": 2.4865,
+      "step": 30
+    },
+    {
+      "epoch": 0.023432923257176334,
+      "grad_norm": 6.0227203369140625,
+      "learning_rate": 1.9908611599297014e-05,
+      "loss": 2.662,
+      "step": 40
+    },
+    {
+      "epoch": 0.029291154071470416,
+      "grad_norm": 9.085187911987305,
+      "learning_rate": 1.988517867603984e-05,
+      "loss": 2.3138,
+      "step": 50
+    },
+    {
+      "epoch": 0.0351493848857645,
+      "grad_norm": 8.839755058288574,
+      "learning_rate": 1.9861745752782663e-05,
+      "loss": 1.9221,
+      "step": 60
+    },
+    {
+      "epoch": 0.041007615700058585,
+      "grad_norm": 5.003777027130127,
+      "learning_rate": 1.9838312829525487e-05,
+      "loss": 1.3857,
+      "step": 70
+    },
+    {
+      "epoch": 0.04686584651435267,
+      "grad_norm": 7.108287811279297,
+      "learning_rate": 1.981487990626831e-05,
+      "loss": 1.5308,
+      "step": 80
+    },
+    {
+      "epoch": 0.05272407732864675,
+      "grad_norm": 6.89133882522583,
+      "learning_rate": 1.9791446983011135e-05,
+      "loss": 1.3522,
+      "step": 90
+    },
+    {
+      "epoch": 0.05858230814294083,
+      "grad_norm": 6.8093695640563965,
+      "learning_rate": 1.9768014059753956e-05,
+      "loss": 1.0296,
+      "step": 100
+    },
+    {
+      "epoch": 0.06444053895723492,
+      "grad_norm": 2.8785605430603027,
+      "learning_rate": 1.974458113649678e-05,
+      "loss": 1.0909,
+      "step": 110
+    },
+    {
+      "epoch": 0.070298769771529,
+      "grad_norm": 3.344712018966675,
+      "learning_rate": 1.9721148213239604e-05,
+      "loss": 0.8385,
+      "step": 120
+    },
+    {
+      "epoch": 0.07615700058582309,
+      "grad_norm": 4.9901041984558105,
+      "learning_rate": 1.9697715289982428e-05,
+      "loss": 0.9958,
+      "step": 130
+    },
+    {
+      "epoch": 0.08201523140011717,
+      "grad_norm": 4.131304740905762,
+      "learning_rate": 1.967428236672525e-05,
+      "loss": 0.8819,
+      "step": 140
+    },
+    {
+      "epoch": 0.08787346221441125,
+      "grad_norm": 2.20668888092041,
+      "learning_rate": 1.9650849443468073e-05,
+      "loss": 0.7881,
+      "step": 150
+    },
+    {
+      "epoch": 0.09373169302870533,
+      "grad_norm": 4.06991720199585,
+      "learning_rate": 1.9627416520210897e-05,
+      "loss": 0.8286,
+      "step": 160
+    },
+    {
+      "epoch": 0.09958992384299942,
+      "grad_norm": 2.710232973098755,
+      "learning_rate": 1.960398359695372e-05,
+      "loss": 0.605,
+      "step": 170
+    },
+    {
+      "epoch": 0.1054481546572935,
+      "grad_norm": 1.2272921800613403,
+      "learning_rate": 1.9580550673696545e-05,
+      "loss": 0.5988,
+      "step": 180
+    },
+    {
+      "epoch": 0.11130638547158758,
+      "grad_norm": 1.7559036016464233,
+      "learning_rate": 1.955711775043937e-05,
+      "loss": 0.6549,
+      "step": 190
+    },
+    {
+      "epoch": 0.11716461628588166,
+      "grad_norm": 2.8257083892822266,
+      "learning_rate": 1.9533684827182193e-05,
+      "loss": 0.6415,
+      "step": 200
+    },
+    {
+      "epoch": 0.12302284710017575,
+      "grad_norm": 2.2019424438476562,
+      "learning_rate": 1.9510251903925017e-05,
+      "loss": 0.4694,
+      "step": 210
+    },
+    {
+      "epoch": 0.12888107791446984,
+      "grad_norm": 2.916757345199585,
+      "learning_rate": 1.948681898066784e-05,
+      "loss": 0.5989,
+      "step": 220
+    },
+    {
+      "epoch": 0.1347393087287639,
+      "grad_norm": 1.955367088317871,
+      "learning_rate": 1.9463386057410662e-05,
+      "loss": 0.478,
+      "step": 230
+    },
+    {
+      "epoch": 0.140597539543058,
+      "grad_norm": 1.6670275926589966,
+      "learning_rate": 1.9439953134153486e-05,
+      "loss": 0.4398,
+      "step": 240
+    },
+    {
+      "epoch": 0.14645577035735208,
+      "grad_norm": 1.7495224475860596,
+      "learning_rate": 1.941652021089631e-05,
+      "loss": 0.5017,
+      "step": 250
+    },
+    {
+      "epoch": 0.15231400117164617,
+      "grad_norm": 1.3516908884048462,
+      "learning_rate": 1.9393087287639135e-05,
+      "loss": 0.3734,
+      "step": 260
+    },
+    {
+      "epoch": 0.15817223198594024,
+      "grad_norm": 1.5071974992752075,
+      "learning_rate": 1.936965436438196e-05,
+      "loss": 0.4378,
+      "step": 270
+    },
+    {
+      "epoch": 0.16403046280023434,
+      "grad_norm": 1.428642749786377,
+      "learning_rate": 1.9346221441124783e-05,
+      "loss": 0.3708,
+      "step": 280
+    },
+    {
+      "epoch": 0.1698886936145284,
+      "grad_norm": 2.0682930946350098,
+      "learning_rate": 1.9322788517867607e-05,
+      "loss": 0.4321,
+      "step": 290
+    },
+    {
+      "epoch": 0.1757469244288225,
+      "grad_norm": 4.945940971374512,
+      "learning_rate": 1.9299355594610428e-05,
+      "loss": 0.4092,
+      "step": 300
+    },
+    {
+      "epoch": 0.18160515524311657,
+      "grad_norm": 1.2532398700714111,
+      "learning_rate": 1.9275922671353252e-05,
+      "loss": 0.2347,
+      "step": 310
+    },
+    {
+      "epoch": 0.18746338605741067,
+      "grad_norm": 1.3484373092651367,
+      "learning_rate": 1.9252489748096076e-05,
+      "loss": 0.2977,
+      "step": 320
+    },
+    {
+      "epoch": 0.19332161687170474,
+      "grad_norm": 1.7067067623138428,
+      "learning_rate": 1.92290568248389e-05,
+      "loss": 0.2155,
+      "step": 330
+    },
+    {
+      "epoch": 0.19917984768599883,
+      "grad_norm": 1.3422434329986572,
+      "learning_rate": 1.9205623901581724e-05,
+      "loss": 0.2571,
+      "step": 340
+    },
+    {
+      "epoch": 0.2050380785002929,
+      "grad_norm": 0.6787233948707581,
+      "learning_rate": 1.9182190978324548e-05,
+      "loss": 0.2459,
+      "step": 350
+    },
+    {
+      "epoch": 0.210896309314587,
+      "grad_norm": 0.7697986364364624,
+      "learning_rate": 1.915875805506737e-05,
+      "loss": 0.2171,
+      "step": 360
+    },
+    {
+      "epoch": 0.21675454012888107,
+      "grad_norm": 2.9185574054718018,
+      "learning_rate": 1.9135325131810193e-05,
+      "loss": 0.2731,
+      "step": 370
+    },
+    {
+      "epoch": 0.22261277094317516,
+      "grad_norm": 1.8088535070419312,
+      "learning_rate": 1.9111892208553017e-05,
+      "loss": 0.2691,
+      "step": 380
+    },
+    {
+      "epoch": 0.22847100175746923,
+      "grad_norm": 1.4816365242004395,
+      "learning_rate": 1.908845928529584e-05,
+      "loss": 0.1665,
+      "step": 390
+    },
+    {
+      "epoch": 0.23432923257176333,
+      "grad_norm": 3.9424285888671875,
+      "learning_rate": 1.9065026362038665e-05,
+      "loss": 0.1885,
+      "step": 400
+    },
+    {
+      "epoch": 0.2401874633860574,
+      "grad_norm": 1.9072023630142212,
+      "learning_rate": 1.904159343878149e-05,
+      "loss": 0.3134,
+      "step": 410
+    },
+    {
+      "epoch": 0.2460456942003515,
+      "grad_norm": 0.817992627620697,
+      "learning_rate": 1.9018160515524314e-05,
+      "loss": 0.1547,
+      "step": 420
+    },
+    {
+      "epoch": 0.2519039250146456,
+      "grad_norm": 2.727170944213867,
+      "learning_rate": 1.8994727592267138e-05,
+      "loss": 0.1292,
+      "step": 430
+    },
+    {
+      "epoch": 0.2577621558289397,
+      "grad_norm": 1.7762303352355957,
+      "learning_rate": 1.8971294669009962e-05,
+      "loss": 0.1353,
+      "step": 440
+    },
+    {
+      "epoch": 0.26362038664323373,
+      "grad_norm": 0.4511788785457611,
+      "learning_rate": 1.8947861745752786e-05,
+      "loss": 0.1999,
+      "step": 450
+    },
+    {
+      "epoch": 0.2694786174575278,
+      "grad_norm": 2.340090751647949,
+      "learning_rate": 1.892442882249561e-05,
+      "loss": 0.0877,
+      "step": 460
+    },
+    {
+      "epoch": 0.2753368482718219,
+      "grad_norm": 1.6077126264572144,
+      "learning_rate": 1.890099589923843e-05,
+      "loss": 0.178,
+      "step": 470
+    },
+    {
+      "epoch": 0.281195079086116,
+      "grad_norm": 1.6138532161712646,
+      "learning_rate": 1.8877562975981255e-05,
+      "loss": 0.2541,
+      "step": 480
+    },
+    {
+      "epoch": 0.28705330990041006,
+      "grad_norm": 3.209932327270508,
+      "learning_rate": 1.885413005272408e-05,
+      "loss": 0.108,
+      "step": 490
+    },
+    {
+      "epoch": 0.29291154071470415,
+      "grad_norm": 3.5678186416625977,
+      "learning_rate": 1.8830697129466903e-05,
+      "loss": 0.141,
+      "step": 500
+    },
+    {
+      "epoch": 0.29876977152899825,
+      "grad_norm": 3.03879976272583,
+      "learning_rate": 1.8807264206209724e-05,
+      "loss": 0.2155,
+      "step": 510
+    },
+    {
+      "epoch": 0.30462800234329235,
+      "grad_norm": 0.4257088899612427,
+      "learning_rate": 1.8783831282952548e-05,
+      "loss": 0.2603,
+      "step": 520
+    },
+    {
+      "epoch": 0.3104862331575864,
+      "grad_norm": 2.5923264026641846,
+      "learning_rate": 1.8760398359695372e-05,
+      "loss": 0.1271,
+      "step": 530
+    },
+    {
+      "epoch": 0.3163444639718805,
+      "grad_norm": 2.9671404361724854,
+      "learning_rate": 1.8736965436438196e-05,
+      "loss": 0.1323,
+      "step": 540
+    },
+    {
+      "epoch": 0.3222026947861746,
+      "grad_norm": 0.1397818624973297,
+      "learning_rate": 1.871353251318102e-05,
+      "loss": 0.0597,
+      "step": 550
+    },
+    {
+      "epoch": 0.3280609256004687,
+      "grad_norm": 0.5060864686965942,
+      "learning_rate": 1.8690099589923845e-05,
+      "loss": 0.2495,
+      "step": 560
+    },
+    {
+      "epoch": 0.3339191564147627,
+      "grad_norm": 1.6251068115234375,
+      "learning_rate": 1.866666666666667e-05,
+      "loss": 0.3011,
+      "step": 570
+    },
+    {
+      "epoch": 0.3397773872290568,
+      "grad_norm": 0.20143267512321472,
+      "learning_rate": 1.8643233743409493e-05,
+      "loss": 0.1201,
+      "step": 580
+    },
+    {
+      "epoch": 0.3456356180433509,
+      "grad_norm": 0.1996494084596634,
+      "learning_rate": 1.8619800820152317e-05,
+      "loss": 0.2238,
+      "step": 590
+    },
+    {
+      "epoch": 0.351493848857645,
+      "grad_norm": 2.6201045513153076,
+      "learning_rate": 1.859636789689514e-05,
+      "loss": 0.1274,
+      "step": 600
+    },
+    {
+      "epoch": 0.35735207967193905,
+      "grad_norm": 3.8342576026916504,
+      "learning_rate": 1.8572934973637965e-05,
+      "loss": 0.0662,
+      "step": 610
+    },
+    {
+      "epoch": 0.36321031048623315,
+      "grad_norm": 4.215113162994385,
+      "learning_rate": 1.8549502050380786e-05,
+      "loss": 0.0751,
+      "step": 620
+    },
+    {
+      "epoch": 0.36906854130052724,
+      "grad_norm": 6.333745002746582,
+      "learning_rate": 1.852606912712361e-05,
+      "loss": 0.1664,
+      "step": 630
+    },
+    {
+      "epoch": 0.37492677211482134,
+      "grad_norm": 6.037374973297119,
+      "learning_rate": 1.8502636203866434e-05,
+      "loss": 0.3169,
+      "step": 640
+    },
+    {
+      "epoch": 0.38078500292911543,
+      "grad_norm": 0.39565032720565796,
+      "learning_rate": 1.8479203280609258e-05,
+      "loss": 0.1296,
+      "step": 650
+    },
+    {
+      "epoch": 0.3866432337434095,
+      "grad_norm": 6.911872386932373,
+      "learning_rate": 1.8455770357352082e-05,
+      "loss": 0.256,
+      "step": 660
+    },
+    {
+      "epoch": 0.39250146455770357,
+      "grad_norm": 0.17452439665794373,
+      "learning_rate": 1.8432337434094903e-05,
+      "loss": 0.0724,
+      "step": 670
+    },
+    {
+      "epoch": 0.39835969537199767,
+      "grad_norm": 0.1822187453508377,
+      "learning_rate": 1.8408904510837727e-05,
+      "loss": 0.0972,
+      "step": 680
+    },
+    {
+      "epoch": 0.40421792618629176,
+      "grad_norm": 2.590902328491211,
+      "learning_rate": 1.838547158758055e-05,
+      "loss": 0.1851,
+      "step": 690
+    },
+    {
+      "epoch": 0.4100761570005858,
+      "grad_norm": 0.2342429906129837,
+      "learning_rate": 1.8362038664323375e-05,
+      "loss": 0.2552,
+      "step": 700
+    },
+    {
+      "epoch": 0.4159343878148799,
+      "grad_norm": 1.4866358041763306,
+      "learning_rate": 1.83386057410662e-05,
+      "loss": 0.131,
+      "step": 710
+    },
+    {
+      "epoch": 0.421792618629174,
+      "grad_norm": 3.6473488807678223,
+      "learning_rate": 1.8315172817809024e-05,
+      "loss": 0.2056,
+      "step": 720
+    },
+    {
+      "epoch": 0.4276508494434681,
+      "grad_norm": 4.957830429077148,
+      "learning_rate": 1.8291739894551848e-05,
+      "loss": 0.2499,
+      "step": 730
+    },
+    {
+      "epoch": 0.43350908025776214,
+      "grad_norm": 0.18394331634044647,
+      "learning_rate": 1.8268306971294672e-05,
+      "loss": 0.0795,
+      "step": 740
+    },
+    {
+      "epoch": 0.43936731107205623,
+      "grad_norm": 0.15296845138072968,
+      "learning_rate": 1.8244874048037493e-05,
+      "loss": 0.0301,
+      "step": 750
+    },
+    {
+      "epoch": 0.44522554188635033,
+      "grad_norm": 0.2405184507369995,
+      "learning_rate": 1.8221441124780317e-05,
+      "loss": 0.0557,
+      "step": 760
+    },
+    {
+      "epoch": 0.4510837727006444,
+      "grad_norm": 2.8683652877807617,
+      "learning_rate": 1.819800820152314e-05,
+      "loss": 0.2331,
+      "step": 770
+    },
+    {
+      "epoch": 0.45694200351493847,
+      "grad_norm": 0.8913503289222717,
+      "learning_rate": 1.8174575278265965e-05,
+      "loss": 0.2253,
+      "step": 780
+    },
+    {
+      "epoch": 0.46280023432923256,
+      "grad_norm": 0.7034773826599121,
+      "learning_rate": 1.815114235500879e-05,
+      "loss": 0.1177,
+      "step": 790
+    },
+    {
+      "epoch": 0.46865846514352666,
+      "grad_norm": 0.3686612844467163,
+      "learning_rate": 1.8127709431751613e-05,
+      "loss": 0.2178,
+      "step": 800
+    },
+    {
+      "epoch": 0.47451669595782076,
+      "grad_norm": 4.0067524909973145,
+      "learning_rate": 1.8104276508494437e-05,
+      "loss": 0.1373,
+      "step": 810
+    },
+    {
+      "epoch": 0.4803749267721148,
+      "grad_norm": 4.526524066925049,
+      "learning_rate": 1.808084358523726e-05,
+      "loss": 0.1769,
+      "step": 820
+    },
+    {
+      "epoch": 0.4862331575864089,
+      "grad_norm": 10.620451927185059,
+      "learning_rate": 1.8057410661980082e-05,
+      "loss": 0.0887,
+      "step": 830
+    },
+    {
+      "epoch": 0.492091388400703,
+      "grad_norm": 1.2064718008041382,
+      "learning_rate": 1.8033977738722906e-05,
+      "loss": 0.2784,
+      "step": 840
+    },
+    {
+      "epoch": 0.4979496192149971,
+      "grad_norm": 2.4158217906951904,
+      "learning_rate": 1.801054481546573e-05,
+      "loss": 0.1493,
+      "step": 850
+    },
+    {
+      "epoch": 0.5038078500292912,
+      "grad_norm": 4.48486328125,
+      "learning_rate": 1.7987111892208554e-05,
+      "loss": 0.0502,
+      "step": 860
+    },
+    {
+      "epoch": 0.5096660808435852,
+      "grad_norm": 0.5434872508049011,
+      "learning_rate": 1.796367896895138e-05,
+      "loss": 0.052,
+      "step": 870
+    },
+    {
+      "epoch": 0.5155243116578794,
+      "grad_norm": 0.11767154932022095,
+      "learning_rate": 1.7940246045694203e-05,
+      "loss": 0.1079,
+      "step": 880
+    },
+    {
+      "epoch": 0.5213825424721734,
+      "grad_norm": 0.14388304948806763,
+      "learning_rate": 1.7916813122437023e-05,
+      "loss": 0.1596,
+      "step": 890
+    },
+    {
+      "epoch": 0.5272407732864675,
+      "grad_norm": 4.8957905769348145,
+      "learning_rate": 1.7893380199179847e-05,
+      "loss": 0.1127,
+      "step": 900
+    },
+    {
+      "epoch": 0.5330990041007616,
+      "grad_norm": 0.09456570446491241,
+      "learning_rate": 1.786994727592267e-05,
+      "loss": 0.215,
+      "step": 910
+    },
+    {
+      "epoch": 0.5389572349150556,
+      "grad_norm": 0.1968098133802414,
+      "learning_rate": 1.7846514352665496e-05,
+      "loss": 0.1591,
+      "step": 920
+    },
+    {
+      "epoch": 0.5448154657293497,
+      "grad_norm": 0.3306093215942383,
+      "learning_rate": 1.782308142940832e-05,
+      "loss": 0.0166,
+      "step": 930
+    },
+    {
+      "epoch": 0.5506736965436438,
+      "grad_norm": 3.040661573410034,
+      "learning_rate": 1.7799648506151144e-05,
+      "loss": 0.0556,
+      "step": 940
+    },
+    {
+      "epoch": 0.5565319273579379,
+      "grad_norm": 0.1786162257194519,
+      "learning_rate": 1.7776215582893968e-05,
+      "loss": 0.15,
+      "step": 950
+    },
+    {
+      "epoch": 0.562390158172232,
+      "grad_norm": 3.1493492126464844,
+      "learning_rate": 1.7752782659636792e-05,
+      "loss": 0.3242,
+      "step": 960
+    },
+    {
+      "epoch": 0.5682483889865261,
+      "grad_norm": 0.3529878556728363,
+      "learning_rate": 1.7729349736379616e-05,
+      "loss": 0.1432,
+      "step": 970
+    },
+    {
+      "epoch": 0.5741066198008201,
+      "grad_norm": 3.4300451278686523,
+      "learning_rate": 1.770591681312244e-05,
+      "loss": 0.1302,
+      "step": 980
+    },
+    {
+      "epoch": 0.5799648506151143,
+      "grad_norm": 0.2585265636444092,
+      "learning_rate": 1.7682483889865264e-05,
+      "loss": 0.0252,
+      "step": 990
+    },
+    {
+      "epoch": 0.5858230814294083,
+      "grad_norm": 5.724643707275391,
+      "learning_rate": 1.7659050966608085e-05,
+      "loss": 0.1441,
+      "step": 1000
+    },
+    {
+      "epoch": 0.5916813122437024,
+      "grad_norm": 3.814067840576172,
+      "learning_rate": 1.763561804335091e-05,
+      "loss": 0.0298,
+      "step": 1010
+    },
+    {
+      "epoch": 0.5975395430579965,
+      "grad_norm": 0.9568992257118225,
+      "learning_rate": 1.7612185120093733e-05,
+      "loss": 0.375,
+      "step": 1020
+    },
+    {
+      "epoch": 0.6033977738722905,
+      "grad_norm": 0.07911839336156845,
+      "learning_rate": 1.7588752196836558e-05,
+      "loss": 0.1334,
+      "step": 1030
+    },
+    {
+      "epoch": 0.6092560046865847,
+      "grad_norm": 12.957563400268555,
+      "learning_rate": 1.7565319273579378e-05,
+      "loss": 0.2123,
+      "step": 1040
+    },
+    {
+      "epoch": 0.6151142355008787,
+      "grad_norm": 0.25080686807632446,
+      "learning_rate": 1.7541886350322202e-05,
+      "loss": 0.0201,
+      "step": 1050
+    },
+    {
+      "epoch": 0.6209724663151728,
+      "grad_norm": 0.17345930635929108,
+      "learning_rate": 1.7518453427065027e-05,
+      "loss": 0.108,
+      "step": 1060
+    },
+    {
+      "epoch": 0.6268306971294669,
+      "grad_norm": 5.377856731414795,
+      "learning_rate": 1.749502050380785e-05,
+      "loss": 0.1263,
+      "step": 1070
+    },
+    {
+      "epoch": 0.632688927943761,
+      "grad_norm": 0.24767795205116272,
+      "learning_rate": 1.7471587580550675e-05,
+      "loss": 0.0413,
+      "step": 1080
+    },
+    {
+      "epoch": 0.6385471587580551,
+      "grad_norm": 0.0758834108710289,
+      "learning_rate": 1.74481546572935e-05,
+      "loss": 0.017,
+      "step": 1090
+    },
+    {
+      "epoch": 0.6444053895723492,
+      "grad_norm": 4.1260905265808105,
+      "learning_rate": 1.7424721734036323e-05,
+      "loss": 0.061,
+      "step": 1100
+    },
+    {
+      "epoch": 0.6502636203866432,
+      "grad_norm": 0.050045400857925415,
+      "learning_rate": 1.7401288810779147e-05,
+      "loss": 0.0523,
+      "step": 1110
+    },
+    {
+      "epoch": 0.6561218512009374,
+      "grad_norm": 4.139227390289307,
+      "learning_rate": 1.737785588752197e-05,
+      "loss": 0.1138,
+      "step": 1120
+    },
+    {
+      "epoch": 0.6619800820152314,
+      "grad_norm": 2.0994224548339844,
+      "learning_rate": 1.7354422964264795e-05,
+      "loss": 0.2773,
+      "step": 1130
+    },
+    {
+      "epoch": 0.6678383128295254,
+      "grad_norm": 5.285941123962402,
+      "learning_rate": 1.733099004100762e-05,
+      "loss": 0.1247,
+      "step": 1140
+    },
+    {
+      "epoch": 0.6736965436438196,
+      "grad_norm": 0.09476584196090698,
+      "learning_rate": 1.730755711775044e-05,
+      "loss": 0.0738,
+      "step": 1150
+    },
+    {
+      "epoch": 0.6795547744581136,
+      "grad_norm": 0.10948991030454636,
+      "learning_rate": 1.7284124194493264e-05,
+      "loss": 0.1981,
+      "step": 1160
+    },
+    {
+      "epoch": 0.6854130052724078,
+      "grad_norm": 0.179415762424469,
+      "learning_rate": 1.726069127123609e-05,
+      "loss": 0.1038,
+      "step": 1170
+    },
+    {
+      "epoch": 0.6912712360867018,
+      "grad_norm": 4.8998894691467285,
+      "learning_rate": 1.7237258347978912e-05,
+      "loss": 0.1813,
+      "step": 1180
+    },
+    {
+      "epoch": 0.6971294669009959,
+      "grad_norm": 0.17422302067279816,
+      "learning_rate": 1.7213825424721737e-05,
+      "loss": 0.0713,
+      "step": 1190
+    },
+    {
+      "epoch": 0.70298769771529,
+      "grad_norm": 0.6925610899925232,
+      "learning_rate": 1.7190392501464557e-05,
+      "loss": 0.0414,
+      "step": 1200
+    },
+    {
+      "epoch": 0.7088459285295841,
+      "grad_norm": 0.05044275522232056,
+      "learning_rate": 1.716695957820738e-05,
+      "loss": 0.1019,
+      "step": 1210
+    },
+    {
+      "epoch": 0.7147041593438781,
+      "grad_norm": 0.14552544057369232,
+      "learning_rate": 1.7143526654950206e-05,
+      "loss": 0.1534,
+      "step": 1220
+    },
+    {
+      "epoch": 0.7205623901581723,
+      "grad_norm": 7.160550594329834,
+      "learning_rate": 1.712009373169303e-05,
+      "loss": 0.185,
+      "step": 1230
+    },
+    {
+      "epoch": 0.7264206209724663,
+      "grad_norm": 0.20127852261066437,
+      "learning_rate": 1.7096660808435854e-05,
+      "loss": 0.0505,
+      "step": 1240
+    },
+    {
+      "epoch": 0.7322788517867604,
+      "grad_norm": 0.07321985810995102,
+      "learning_rate": 1.7073227885178678e-05,
+      "loss": 0.0443,
+      "step": 1250
+    },
+    {
+      "epoch": 0.7381370826010545,
+      "grad_norm": 1.5325711965560913,
+      "learning_rate": 1.7049794961921502e-05,
+      "loss": 0.0673,
+      "step": 1260
+    },
+    {
+      "epoch": 0.7439953134153485,
+      "grad_norm": 1.1066352128982544,
+      "learning_rate": 1.7026362038664326e-05,
+      "loss": 0.1124,
+      "step": 1270
+    },
+    {
+      "epoch": 0.7498535442296427,
+      "grad_norm": 0.061127688735723495,
+      "learning_rate": 1.7002929115407147e-05,
+      "loss": 0.0795,
+      "step": 1280
+    },
+    {
+      "epoch": 0.7557117750439367,
+      "grad_norm": 0.2831653654575348,
+      "learning_rate": 1.697949619214997e-05,
+      "loss": 0.0952,
+      "step": 1290
+    },
+    {
+      "epoch": 0.7615700058582309,
+      "grad_norm": 0.04763650521636009,
+      "learning_rate": 1.6956063268892795e-05,
+      "loss": 0.227,
+      "step": 1300
+    },
+    {
+      "epoch": 0.7674282366725249,
+      "grad_norm": 0.043026477098464966,
+      "learning_rate": 1.693263034563562e-05,
+      "loss": 0.1314,
+      "step": 1310
+    },
+    {
+      "epoch": 0.773286467486819,
+      "grad_norm": 0.0663742944598198,
+      "learning_rate": 1.6909197422378443e-05,
+      "loss": 0.1627,
+      "step": 1320
+    },
+    {
+      "epoch": 0.7791446983011131,
+      "grad_norm": 0.09994468092918396,
+      "learning_rate": 1.6885764499121267e-05,
+      "loss": 0.0781,
+      "step": 1330
+    },
+    {
+      "epoch": 0.7850029291154071,
+      "grad_norm": 0.07375376671552658,
+      "learning_rate": 1.686233157586409e-05,
+      "loss": 0.1963,
+      "step": 1340
+    },
+    {
+      "epoch": 0.7908611599297012,
+      "grad_norm": 0.5165359973907471,
+      "learning_rate": 1.6838898652606916e-05,
+      "loss": 0.275,
+      "step": 1350
+    },
+    {
+      "epoch": 0.7967193907439953,
+      "grad_norm": 4.770512580871582,
+      "learning_rate": 1.681546572934974e-05,
+      "loss": 0.0923,
+      "step": 1360
+    },
+    {
+      "epoch": 0.8025776215582894,
+      "grad_norm": 0.10542096197605133,
+      "learning_rate": 1.679203280609256e-05,
+      "loss": 0.0993,
+      "step": 1370
+    },
+    {
+      "epoch": 0.8084358523725835,
+      "grad_norm": 0.08764227479696274,
+      "learning_rate": 1.6768599882835385e-05,
+      "loss": 0.0535,
+      "step": 1380
+    },
+    {
+      "epoch": 0.8142940831868776,
+      "grad_norm": 3.3957512378692627,
+      "learning_rate": 1.674516695957821e-05,
+      "loss": 0.0141,
+      "step": 1390
+    },
+    {
+      "epoch": 0.8201523140011716,
+      "grad_norm": 0.03514677286148071,
+      "learning_rate": 1.6721734036321033e-05,
+      "loss": 0.0481,
+      "step": 1400
+    },
+    {
+      "epoch": 0.8260105448154658,
+      "grad_norm": 0.054979559034109116,
+      "learning_rate": 1.6698301113063854e-05,
+      "loss": 0.0223,
+      "step": 1410
+    },
+    {
+      "epoch": 0.8318687756297598,
+      "grad_norm": 6.025042533874512,
+      "learning_rate": 1.6674868189806678e-05,
+      "loss": 0.2666,
+      "step": 1420
+    },
+    {
+      "epoch": 0.8377270064440538,
+      "grad_norm": 0.11764468997716904,
+      "learning_rate": 1.6651435266549502e-05,
+      "loss": 0.1323,
+      "step": 1430
+    },
+    {
+      "epoch": 0.843585237258348,
+      "grad_norm": 0.08743251860141754,
+      "learning_rate": 1.6628002343292326e-05,
+      "loss": 0.1267,
+      "step": 1440
+    },
+    {
+      "epoch": 0.849443468072642,
+      "grad_norm": 0.16297052800655365,
+      "learning_rate": 1.660456942003515e-05,
+      "loss": 0.1671,
+      "step": 1450
+    },
+    {
+      "epoch": 0.8553016988869362,
+      "grad_norm": 0.12175700068473816,
+      "learning_rate": 1.6581136496777974e-05,
+      "loss": 0.0084,
+      "step": 1460
+    },
+    {
+      "epoch": 0.8611599297012302,
+      "grad_norm": 4.576499938964844,
+      "learning_rate": 1.6557703573520798e-05,
+      "loss": 0.0398,
+      "step": 1470
+    },
+    {
+      "epoch": 0.8670181605155243,
+      "grad_norm": 1.2775115966796875,
+      "learning_rate": 1.6534270650263622e-05,
+      "loss": 0.0634,
+      "step": 1480
+    },
+    {
+      "epoch": 0.8728763913298184,
+      "grad_norm": 4.699284553527832,
+      "learning_rate": 1.6510837727006446e-05,
+      "loss": 0.0668,
+      "step": 1490
+    },
+    {
+      "epoch": 0.8787346221441125,
+      "grad_norm": 0.12817470729351044,
+      "learning_rate": 1.648740480374927e-05,
+      "loss": 0.0749,
+      "step": 1500
+    },
+    {
+      "epoch": 0.8845928529584065,
+      "grad_norm": 5.865419864654541,
+      "learning_rate": 1.6463971880492095e-05,
+      "loss": 0.115,
+      "step": 1510
+    },
+    {
+      "epoch": 0.8904510837727007,
+      "grad_norm": 3.5330662727355957,
+      "learning_rate": 1.644053895723492e-05,
+      "loss": 0.1223,
+      "step": 1520
+    },
+    {
+      "epoch": 0.8963093145869947,
+      "grad_norm": 0.0576438307762146,
+      "learning_rate": 1.641710603397774e-05,
+      "loss": 0.2236,
+      "step": 1530
+    },
+    {
+      "epoch": 0.9021675454012889,
+      "grad_norm": 0.03019624389708042,
+      "learning_rate": 1.6393673110720564e-05,
+      "loss": 0.062,
+      "step": 1540
+    },
+    {
+      "epoch": 0.9080257762155829,
+      "grad_norm": 0.11998365819454193,
+      "learning_rate": 1.6370240187463388e-05,
+      "loss": 0.1792,
+      "step": 1550
+    },
+    {
+      "epoch": 0.9138840070298769,
+      "grad_norm": 7.0185370445251465,
+      "learning_rate": 1.6346807264206212e-05,
+      "loss": 0.1184,
+      "step": 1560
+    },
+    {
+      "epoch": 0.9197422378441711,
+      "grad_norm": 0.26941293478012085,
+      "learning_rate": 1.6323374340949033e-05,
+      "loss": 0.0451,
+      "step": 1570
+    },
+    {
+      "epoch": 0.9256004686584651,
+      "grad_norm": 0.11105812340974808,
+      "learning_rate": 1.6299941417691857e-05,
+      "loss": 0.0957,
+      "step": 1580
+    },
+    {
+      "epoch": 0.9314586994727593,
+      "grad_norm": 7.551598072052002,
+      "learning_rate": 1.627650849443468e-05,
+      "loss": 0.063,
+      "step": 1590
+    },
+    {
+      "epoch": 0.9373169302870533,
+      "grad_norm": 0.03378705307841301,
+      "learning_rate": 1.6253075571177505e-05,
+      "loss": 0.0234,
+      "step": 1600
+    },
+    {
+      "epoch": 0.9431751611013474,
+      "grad_norm": 0.25736942887306213,
+      "learning_rate": 1.622964264792033e-05,
+      "loss": 0.041,
+      "step": 1610
+    },
+    {
+      "epoch": 0.9490333919156415,
+      "grad_norm": 0.14889341592788696,
+      "learning_rate": 1.6206209724663153e-05,
+      "loss": 0.0337,
+      "step": 1620
+    },
+    {
+      "epoch": 0.9548916227299356,
+      "grad_norm": 0.05913722887635231,
+      "learning_rate": 1.6182776801405977e-05,
+      "loss": 0.1896,
+      "step": 1630
+    },
+    {
+      "epoch": 0.9607498535442296,
+      "grad_norm": 0.3258120119571686,
+      "learning_rate": 1.61593438781488e-05,
+      "loss": 0.1096,
+      "step": 1640
+    },
+    {
+      "epoch": 0.9666080843585237,
+      "grad_norm": 0.05084146186709404,
+      "learning_rate": 1.6135910954891626e-05,
+      "loss": 0.1543,
+      "step": 1650
+    },
+    {
+      "epoch": 0.9724663151728178,
+      "grad_norm": 3.364567518234253,
+      "learning_rate": 1.611247803163445e-05,
+      "loss": 0.2953,
+      "step": 1660
+    },
+    {
+      "epoch": 0.9783245459871119,
+      "grad_norm": 0.07459431141614914,
+      "learning_rate": 1.608904510837727e-05,
+      "loss": 0.0135,
+      "step": 1670
+    },
+    {
+      "epoch": 0.984182776801406,
+      "grad_norm": 4.472078323364258,
+      "learning_rate": 1.6065612185120094e-05,
+      "loss": 0.1756,
+      "step": 1680
+    },
+    {
+      "epoch": 0.9900410076157,
+      "grad_norm": 0.0766572579741478,
+      "learning_rate": 1.604217926186292e-05,
+      "loss": 0.0382,
+      "step": 1690
+    },
+    {
+      "epoch": 0.9958992384299942,
+      "grad_norm": 4.629612445831299,
+      "learning_rate": 1.6018746338605743e-05,
+      "loss": 0.1475,
+      "step": 1700
+    },
+    {
+      "epoch": 1.0017574692442883,
+      "grad_norm": 2.2122583389282227,
+      "learning_rate": 1.5995313415348567e-05,
+      "loss": 0.007,
+      "step": 1710
+    },
+    {
+      "epoch": 1.0076157000585824,
+      "grad_norm": 2.1959640979766846,
+      "learning_rate": 1.597188049209139e-05,
+      "loss": 0.0108,
+      "step": 1720
+    },
+    {
+      "epoch": 1.0134739308728764,
+      "grad_norm": 6.177903652191162,
+      "learning_rate": 1.594844756883421e-05,
+      "loss": 0.1057,
+      "step": 1730
+    },
+    {
+      "epoch": 1.0193321616871704,
+      "grad_norm": 0.07769495248794556,
+      "learning_rate": 1.5925014645577036e-05,
+      "loss": 0.062,
+      "step": 1740
+    },
+    {
+      "epoch": 1.0251903925014645,
+      "grad_norm": 5.370426177978516,
+      "learning_rate": 1.590158172231986e-05,
+      "loss": 0.0702,
+      "step": 1750
+    },
+    {
+      "epoch": 1.0310486233157587,
+      "grad_norm": 0.06684985756874084,
+      "learning_rate": 1.5878148799062684e-05,
+      "loss": 0.1196,
+      "step": 1760
+    },
+    {
+      "epoch": 1.0369068541300528,
+      "grad_norm": 0.277066171169281,
+      "learning_rate": 1.5854715875805508e-05,
+      "loss": 0.0566,
+      "step": 1770
+    },
+    {
+      "epoch": 1.0427650849443468,
+      "grad_norm": 2.941352367401123,
+      "learning_rate": 1.5831282952548332e-05,
+      "loss": 0.069,
+      "step": 1780
+    },
+    {
+      "epoch": 1.0486233157586409,
+      "grad_norm": 0.050202783197164536,
+      "learning_rate": 1.5807850029291156e-05,
+      "loss": 0.0451,
+      "step": 1790
+    },
+    {
+      "epoch": 1.054481546572935,
+      "grad_norm": 0.04496518895030022,
+      "learning_rate": 1.5784417106033977e-05,
+      "loss": 0.1135,
+      "step": 1800
+    },
+    {
+      "epoch": 1.060339777387229,
+      "grad_norm": 7.685583114624023,
+      "learning_rate": 1.57609841827768e-05,
+      "loss": 0.2183,
+      "step": 1810
+    },
+    {
+      "epoch": 1.0661980082015232,
+      "grad_norm": 0.041232019662857056,
+      "learning_rate": 1.5737551259519625e-05,
+      "loss": 0.0681,
+      "step": 1820
+    },
+    {
+      "epoch": 1.0720562390158173,
+      "grad_norm": 0.11009787768125534,
+      "learning_rate": 1.571411833626245e-05,
+      "loss": 0.1064,
+      "step": 1830
+    },
+    {
+      "epoch": 1.0779144698301113,
+      "grad_norm": 10.8427095413208,
+      "learning_rate": 1.5690685413005274e-05,
+      "loss": 0.0494,
+      "step": 1840
+    },
+    {
+      "epoch": 1.0837727006444053,
+      "grad_norm": 0.10766121000051498,
+      "learning_rate": 1.5667252489748098e-05,
+      "loss": 0.161,
+      "step": 1850
+    },
+    {
+      "epoch": 1.0896309314586994,
+      "grad_norm": 6.553893566131592,
+      "learning_rate": 1.5643819566490922e-05,
+      "loss": 0.1559,
+      "step": 1860
+    },
+    {
+      "epoch": 1.0954891622729936,
+      "grad_norm": 10.493359565734863,
+      "learning_rate": 1.5620386643233746e-05,
+      "loss": 0.0245,
+      "step": 1870
+    },
+    {
+      "epoch": 1.1013473930872877,
+      "grad_norm": 8.9678373336792,
+      "learning_rate": 1.559695371997657e-05,
+      "loss": 0.1051,
+      "step": 1880
+    },
+    {
+      "epoch": 1.1072056239015817,
+      "grad_norm": 0.08524399250745773,
+      "learning_rate": 1.5573520796719394e-05,
+      "loss": 0.0044,
+      "step": 1890
+    },
+    {
+      "epoch": 1.1130638547158758,
+      "grad_norm": 0.027697084471583366,
+      "learning_rate": 1.5550087873462215e-05,
+      "loss": 0.0847,
+      "step": 1900
+    },
+    {
+      "epoch": 1.1189220855301698,
+      "grad_norm": 0.2592675983905792,
+      "learning_rate": 1.552665495020504e-05,
+      "loss": 0.0702,
+      "step": 1910
+    },
+    {
+      "epoch": 1.124780316344464,
+      "grad_norm": 0.033791959285736084,
+      "learning_rate": 1.5503222026947863e-05,
+      "loss": 0.1042,
+      "step": 1920
+    },
+    {
+      "epoch": 1.1306385471587581,
+      "grad_norm": 5.939085960388184,
+      "learning_rate": 1.5479789103690687e-05,
+      "loss": 0.0095,
+      "step": 1930
+    },
+    {
+      "epoch": 1.1364967779730522,
+      "grad_norm": 0.7582628726959229,
+      "learning_rate": 1.5456356180433508e-05,
+      "loss": 0.0844,
+      "step": 1940
+    },
+    {
+      "epoch": 1.1423550087873462,
+      "grad_norm": 0.34229347109794617,
+      "learning_rate": 1.5432923257176332e-05,
+      "loss": 0.0173,
+      "step": 1950
+    },
+    {
+      "epoch": 1.1482132396016402,
+      "grad_norm": 0.24234962463378906,
+      "learning_rate": 1.5409490333919156e-05,
+      "loss": 0.2988,
+      "step": 1960
+    },
+    {
+      "epoch": 1.1540714704159343,
+      "grad_norm": 0.21282519400119781,
+      "learning_rate": 1.538605741066198e-05,
+      "loss": 0.0115,
+      "step": 1970
+    },
+    {
+      "epoch": 1.1599297012302285,
+      "grad_norm": 0.2657891809940338,
+      "learning_rate": 1.5362624487404804e-05,
+      "loss": 0.1126,
+      "step": 1980
+    },
+    {
+      "epoch": 1.1657879320445226,
+      "grad_norm": 0.2175203263759613,
+      "learning_rate": 1.533919156414763e-05,
+      "loss": 0.2326,
+      "step": 1990
+    },
+    {
+      "epoch": 1.1716461628588166,
+      "grad_norm": 7.306258201599121,
+      "learning_rate": 1.5315758640890453e-05,
+      "loss": 0.1161,
+      "step": 2000
+    },
+    {
+      "epoch": 1.1775043936731107,
+      "grad_norm": 0.4325994849205017,
+      "learning_rate": 1.5292325717633277e-05,
+      "loss": 0.1118,
+      "step": 2010
+    },
+    {
+      "epoch": 1.1833626244874047,
+      "grad_norm": 0.03175567463040352,
+      "learning_rate": 1.52688927943761e-05,
+      "loss": 0.1036,
+      "step": 2020
+    },
+    {
+      "epoch": 1.189220855301699,
+      "grad_norm": 0.06606226414442062,
+      "learning_rate": 1.5245459871118923e-05,
+      "loss": 0.0048,
+      "step": 2030
+    },
+    {
+      "epoch": 1.195079086115993,
+      "grad_norm": 4.948124408721924,
+      "learning_rate": 1.5222026947861747e-05,
+      "loss": 0.1035,
+      "step": 2040
+    },
+    {
+      "epoch": 1.200937316930287,
+      "grad_norm": 0.05158844590187073,
+      "learning_rate": 1.5198594024604571e-05,
+      "loss": 0.2282,
+      "step": 2050
+    },
+    {
+      "epoch": 1.206795547744581,
+      "grad_norm": 5.9457106590271,
+      "learning_rate": 1.5175161101347396e-05,
+      "loss": 0.0251,
+      "step": 2060
+    },
+    {
+      "epoch": 1.2126537785588751,
+      "grad_norm": 0.41592299938201904,
+      "learning_rate": 1.5151728178090216e-05,
+      "loss": 0.1272,
+      "step": 2070
+    },
+    {
+      "epoch": 1.2185120093731694,
+      "grad_norm": 0.031298503279685974,
+      "learning_rate": 1.512829525483304e-05,
+      "loss": 0.0719,
+      "step": 2080
+    },
+    {
+      "epoch": 1.2243702401874634,
+      "grad_norm": 0.05066341161727905,
+      "learning_rate": 1.5104862331575865e-05,
+      "loss": 0.0489,
+      "step": 2090
+    },
+    {
+      "epoch": 1.2302284710017575,
+      "grad_norm": 0.06502586603164673,
+      "learning_rate": 1.5081429408318689e-05,
+      "loss": 0.0045,
+      "step": 2100
+    },
+    {
+      "epoch": 1.2360867018160515,
+      "grad_norm": 0.22800350189208984,
+      "learning_rate": 1.5057996485061513e-05,
+      "loss": 0.0287,
+      "step": 2110
+    },
+    {
+      "epoch": 1.2419449326303456,
+      "grad_norm": 0.05417035520076752,
+      "learning_rate": 1.5034563561804337e-05,
+      "loss": 0.0656,
+      "step": 2120
+    },
+    {
+      "epoch": 1.2478031634446398,
+      "grad_norm": 0.09729498624801636,
+      "learning_rate": 1.501113063854716e-05,
+      "loss": 0.1163,
+      "step": 2130
+    },
+    {
+      "epoch": 1.2536613942589339,
+      "grad_norm": 0.03419837728142738,
+      "learning_rate": 1.4987697715289983e-05,
+      "loss": 0.045,
+      "step": 2140
+    },
+    {
+      "epoch": 1.259519625073228,
+      "grad_norm": 35.93345642089844,
+      "learning_rate": 1.4964264792032807e-05,
+      "loss": 0.2297,
+      "step": 2150
+    },
+    {
+      "epoch": 1.265377855887522,
+      "grad_norm": 4.408044815063477,
+      "learning_rate": 1.4940831868775632e-05,
+      "loss": 0.0762,
+      "step": 2160
+    },
+    {
+      "epoch": 1.271236086701816,
+      "grad_norm": 0.2184348851442337,
+      "learning_rate": 1.4917398945518456e-05,
+      "loss": 0.042,
+      "step": 2170
+    },
+    {
+      "epoch": 1.2770943175161102,
+      "grad_norm": 0.036489009857177734,
+      "learning_rate": 1.489396602226128e-05,
+      "loss": 0.0239,
+      "step": 2180
+    },
+    {
+      "epoch": 1.2829525483304043,
+      "grad_norm": 0.10858841240406036,
+      "learning_rate": 1.4870533099004102e-05,
+      "loss": 0.0724,
+      "step": 2190
+    },
+    {
+      "epoch": 1.2888107791446983,
+      "grad_norm": 0.1485690176486969,
+      "learning_rate": 1.4847100175746925e-05,
+      "loss": 0.077,
+      "step": 2200
+    },
+    {
+      "epoch": 1.2946690099589924,
+      "grad_norm": 0.3935866951942444,
+      "learning_rate": 1.4823667252489749e-05,
+      "loss": 0.1539,
+      "step": 2210
+    },
+    {
+      "epoch": 1.3005272407732864,
+      "grad_norm": 0.10183463990688324,
+      "learning_rate": 1.4800234329232573e-05,
+      "loss": 0.1011,
+      "step": 2220
+    },
+    {
+      "epoch": 1.3063854715875807,
+      "grad_norm": 0.23003660142421722,
+      "learning_rate": 1.4776801405975395e-05,
+      "loss": 0.0398,
+      "step": 2230
+    },
+    {
+      "epoch": 1.3122437024018747,
+      "grad_norm": 1.0053473711013794,
+      "learning_rate": 1.475336848271822e-05,
+      "loss": 0.0114,
+      "step": 2240
+    },
+    {
+      "epoch": 1.3181019332161688,
+      "grad_norm": 0.0832851454615593,
+      "learning_rate": 1.4729935559461044e-05,
+      "loss": 0.0889,
+      "step": 2250
+    },
+    {
+      "epoch": 1.3239601640304628,
+      "grad_norm": 0.044370219111442566,
+      "learning_rate": 1.4706502636203868e-05,
+      "loss": 0.0047,
+      "step": 2260
+    },
+    {
+      "epoch": 1.3298183948447568,
+      "grad_norm": 2.207583427429199,
+      "learning_rate": 1.4683069712946692e-05,
+      "loss": 0.1405,
+      "step": 2270
+    },
+    {
+      "epoch": 1.335676625659051,
+      "grad_norm": 0.2738548219203949,
+      "learning_rate": 1.4659636789689516e-05,
+      "loss": 0.1053,
+      "step": 2280
+    },
+    {
+      "epoch": 1.341534856473345,
+      "grad_norm": 5.799202919006348,
+      "learning_rate": 1.4636203866432338e-05,
+      "loss": 0.1031,
+      "step": 2290
+    },
+    {
+      "epoch": 1.3473930872876392,
+      "grad_norm": 0.03329119086265564,
+      "learning_rate": 1.4612770943175162e-05,
+      "loss": 0.0167,
+      "step": 2300
+    },
+    {
+      "epoch": 1.3532513181019332,
+      "grad_norm": 0.10547787696123123,
+      "learning_rate": 1.4589338019917987e-05,
+      "loss": 0.0068,
+      "step": 2310
+    },
+    {
+      "epoch": 1.3591095489162273,
+      "grad_norm": 14.383464813232422,
+      "learning_rate": 1.456590509666081e-05,
+      "loss": 0.0457,
+      "step": 2320
+    },
+    {
+      "epoch": 1.3649677797305213,
+      "grad_norm": 1.1754510402679443,
+      "learning_rate": 1.4542472173403633e-05,
+      "loss": 0.1441,
+      "step": 2330
+    },
+    {
+      "epoch": 1.3708260105448153,
+      "grad_norm": 0.1931016892194748,
+      "learning_rate": 1.4519039250146455e-05,
+      "loss": 0.0026,
+      "step": 2340
+    },
+    {
+      "epoch": 1.3766842413591096,
+      "grad_norm": 0.036222219467163086,
+      "learning_rate": 1.449560632688928e-05,
+      "loss": 0.0575,
+      "step": 2350
+    },
+    {
+      "epoch": 1.3825424721734036,
+      "grad_norm": 0.05857452005147934,
+      "learning_rate": 1.4472173403632104e-05,
+      "loss": 0.0035,
+      "step": 2360
+    },
+    {
+      "epoch": 1.3884007029876977,
+      "grad_norm": 0.027153009548783302,
+      "learning_rate": 1.4448740480374928e-05,
+      "loss": 0.0071,
+      "step": 2370
+    },
+    {
+      "epoch": 1.3942589338019917,
+      "grad_norm": 0.030506083741784096,
+      "learning_rate": 1.4425307557117752e-05,
+      "loss": 0.0501,
+      "step": 2380
+    },
+    {
+      "epoch": 1.4001171646162858,
+      "grad_norm": 0.24087435007095337,
+      "learning_rate": 1.4401874633860576e-05,
+      "loss": 0.0457,
+      "step": 2390
+    },
+    {
+      "epoch": 1.40597539543058,
+      "grad_norm": 7.0238566398620605,
+      "learning_rate": 1.4378441710603398e-05,
+      "loss": 0.0856,
+      "step": 2400
+    },
+    {
+      "epoch": 1.411833626244874,
+      "grad_norm": 17.90679359436035,
+      "learning_rate": 1.4355008787346223e-05,
+      "loss": 0.1618,
+      "step": 2410
+    },
+    {
+      "epoch": 1.4176918570591681,
+      "grad_norm": 0.031517572700977325,
+      "learning_rate": 1.4331575864089047e-05,
+      "loss": 0.0033,
+      "step": 2420
+    },
+    {
+      "epoch": 1.4235500878734622,
+      "grad_norm": 0.8404099345207214,
+      "learning_rate": 1.430814294083187e-05,
+      "loss": 0.0529,
+      "step": 2430
+    },
+    {
+      "epoch": 1.4294083186877562,
+      "grad_norm": 0.01688993163406849,
+      "learning_rate": 1.4284710017574695e-05,
+      "loss": 0.0032,
+      "step": 2440
+    },
+    {
+      "epoch": 1.4352665495020505,
+      "grad_norm": 0.06315304338932037,
+      "learning_rate": 1.4261277094317519e-05,
+      "loss": 0.1193,
+      "step": 2450
+    },
+    {
+      "epoch": 1.4411247803163445,
+      "grad_norm": 0.04080997034907341,
+      "learning_rate": 1.423784417106034e-05,
+      "loss": 0.0028,
+      "step": 2460
+    },
+    {
+      "epoch": 1.4469830111306385,
+      "grad_norm": 0.023772750049829483,
+      "learning_rate": 1.4214411247803164e-05,
+      "loss": 0.0127,
+      "step": 2470
+    },
+    {
+      "epoch": 1.4528412419449326,
+      "grad_norm": 0.015330899506807327,
+      "learning_rate": 1.4190978324545988e-05,
+      "loss": 0.0706,
+      "step": 2480
+    },
+    {
+      "epoch": 1.4586994727592266,
+      "grad_norm": 0.4532860517501831,
+      "learning_rate": 1.4167545401288812e-05,
+      "loss": 0.0056,
+      "step": 2490
+    },
+    {
+      "epoch": 1.4645577035735209,
+      "grad_norm": 0.24234826862812042,
+      "learning_rate": 1.4144112478031635e-05,
+      "loss": 0.0024,
+      "step": 2500
+    },
+    {
+      "epoch": 1.470415934387815,
+      "grad_norm": 0.6973457932472229,
+      "learning_rate": 1.4120679554774459e-05,
+      "loss": 0.2314,
+      "step": 2510
+    },
+    {
+      "epoch": 1.476274165202109,
+      "grad_norm": 8.073694229125977,
+      "learning_rate": 1.4097246631517283e-05,
+      "loss": 0.0922,
+      "step": 2520
+    },
+    {
+      "epoch": 1.482132396016403,
+      "grad_norm": 0.03306486830115318,
+      "learning_rate": 1.4073813708260107e-05,
+      "loss": 0.1651,
+      "step": 2530
+    },
+    {
+      "epoch": 1.487990626830697,
+      "grad_norm": 0.10176576673984528,
+      "learning_rate": 1.4050380785002931e-05,
+      "loss": 0.0416,
+      "step": 2540
+    },
+    {
+      "epoch": 1.4938488576449913,
+      "grad_norm": 0.10366008430719376,
+      "learning_rate": 1.4026947861745755e-05,
+      "loss": 0.0615,
+      "step": 2550
+    },
+    {
+      "epoch": 1.4997070884592854,
+      "grad_norm": 0.21881893277168274,
+      "learning_rate": 1.4003514938488578e-05,
+      "loss": 0.0196,
+      "step": 2560
+    },
+    {
+      "epoch": 1.5055653192735794,
+      "grad_norm": 0.43505847454071045,
+      "learning_rate": 1.3980082015231402e-05,
+      "loss": 0.0996,
+      "step": 2570
+    },
+    {
+      "epoch": 1.5114235500878734,
+      "grad_norm": 7.067519187927246,
+      "learning_rate": 1.3956649091974226e-05,
+      "loss": 0.1073,
+      "step": 2580
+    },
+    {
+      "epoch": 1.5172817809021675,
+      "grad_norm": 0.022074054926633835,
+      "learning_rate": 1.3933216168717048e-05,
+      "loss": 0.0023,
+      "step": 2590
+    },
+    {
+      "epoch": 1.5231400117164617,
+      "grad_norm": 0.021333998069167137,
+      "learning_rate": 1.390978324545987e-05,
+      "loss": 0.0192,
+      "step": 2600
+    },
+    {
+      "epoch": 1.5289982425307556,
+      "grad_norm": 4.094031810760498,
+      "learning_rate": 1.3886350322202695e-05,
+      "loss": 0.1487,
+      "step": 2610
+    },
+    {
+      "epoch": 1.5348564733450498,
+      "grad_norm": 0.028797738254070282,
+      "learning_rate": 1.3862917398945519e-05,
+      "loss": 0.0181,
+      "step": 2620
+    },
+    {
+      "epoch": 1.5407147041593439,
+      "grad_norm": 0.37651917338371277,
+      "learning_rate": 1.3839484475688343e-05,
+      "loss": 0.0397,
+      "step": 2630
+    },
+    {
+      "epoch": 1.546572934973638,
+      "grad_norm": 0.04619598761200905,
+      "learning_rate": 1.3816051552431167e-05,
+      "loss": 0.0334,
+      "step": 2640
+    },
+    {
+      "epoch": 1.5524311657879322,
+      "grad_norm": 0.09797985851764679,
+      "learning_rate": 1.3792618629173991e-05,
+      "loss": 0.2177,
+      "step": 2650
+    },
+    {
+      "epoch": 1.558289396602226,
+      "grad_norm": 4.648836612701416,
+      "learning_rate": 1.3769185705916814e-05,
+      "loss": 0.1014,
+      "step": 2660
+    },
+    {
+      "epoch": 1.5641476274165202,
+      "grad_norm": 0.013686669990420341,
+      "learning_rate": 1.3745752782659638e-05,
+      "loss": 0.102,
+      "step": 2670
+    },
+    {
+      "epoch": 1.5700058582308143,
+      "grad_norm": 0.33198732137680054,
+      "learning_rate": 1.3722319859402462e-05,
+      "loss": 0.0181,
+      "step": 2680
+    },
+    {
+      "epoch": 1.5758640890451083,
+      "grad_norm": 9.264932632446289,
+      "learning_rate": 1.3698886936145286e-05,
+      "loss": 0.1374,
+      "step": 2690
+    },
+    {
+      "epoch": 1.5817223198594026,
+      "grad_norm": 4.808820724487305,
+      "learning_rate": 1.367545401288811e-05,
+      "loss": 0.0389,
+      "step": 2700
+    },
+    {
+      "epoch": 1.5875805506736964,
+      "grad_norm": 1.272582769393921,
+      "learning_rate": 1.3652021089630934e-05,
+      "loss": 0.0035,
+      "step": 2710
+    },
+    {
+      "epoch": 1.5934387814879907,
+      "grad_norm": 0.05750690773129463,
+      "learning_rate": 1.3628588166373755e-05,
+      "loss": 0.0981,
+      "step": 2720
+    },
+    {
+      "epoch": 1.5992970123022847,
+      "grad_norm": 0.174026757478714,
+      "learning_rate": 1.3605155243116579e-05,
+      "loss": 0.063,
+      "step": 2730
+    },
+    {
+      "epoch": 1.6051552431165788,
+      "grad_norm": 0.034066952764987946,
+      "learning_rate": 1.3581722319859403e-05,
+      "loss": 0.0291,
+      "step": 2740
+    },
+    {
+      "epoch": 1.611013473930873,
+      "grad_norm": 0.16694338619709015,
+      "learning_rate": 1.3558289396602227e-05,
+      "loss": 0.0909,
+      "step": 2750
+    },
+    {
+      "epoch": 1.6168717047451668,
+      "grad_norm": 1.0258171558380127,
+      "learning_rate": 1.353485647334505e-05,
+      "loss": 0.0612,
+      "step": 2760
+    },
+    {
+      "epoch": 1.622729935559461,
+      "grad_norm": 5.220030784606934,
+      "learning_rate": 1.3511423550087874e-05,
+      "loss": 0.1047,
+      "step": 2770
+    },
+    {
+      "epoch": 1.6285881663737551,
+      "grad_norm": 1.811539649963379,
+      "learning_rate": 1.3487990626830698e-05,
+      "loss": 0.1035,
+      "step": 2780
+    },
+    {
+      "epoch": 1.6344463971880492,
+      "grad_norm": 0.04276169091463089,
+      "learning_rate": 1.3464557703573522e-05,
+      "loss": 0.0646,
+      "step": 2790
+    },
+    {
+      "epoch": 1.6403046280023434,
+      "grad_norm": 0.025083236396312714,
+      "learning_rate": 1.3441124780316346e-05,
+      "loss": 0.0682,
+      "step": 2800
+    },
+    {
+      "epoch": 1.6461628588166373,
+      "grad_norm": 0.5909777879714966,
+      "learning_rate": 1.341769185705917e-05,
+      "loss": 0.0625,
+      "step": 2810
+    },
+    {
+      "epoch": 1.6520210896309315,
+      "grad_norm": 0.03190384432673454,
+      "learning_rate": 1.3394258933801994e-05,
+      "loss": 0.0549,
+      "step": 2820
+    },
+    {
+      "epoch": 1.6578793204452256,
+      "grad_norm": 0.008607199415564537,
+      "learning_rate": 1.3370826010544817e-05,
+      "loss": 0.1849,
+      "step": 2830
+    },
+    {
+      "epoch": 1.6637375512595196,
+      "grad_norm": 0.6105258464813232,
+      "learning_rate": 1.3347393087287641e-05,
+      "loss": 0.0044,
+      "step": 2840
+    },
+    {
+      "epoch": 1.6695957820738139,
+      "grad_norm": 0.008611824363470078,
+      "learning_rate": 1.3323960164030465e-05,
+      "loss": 0.0542,
+      "step": 2850
+    },
+    {
+      "epoch": 1.6754540128881077,
+      "grad_norm": 9.882821083068848,
+      "learning_rate": 1.3300527240773287e-05,
+      "loss": 0.0708,
+      "step": 2860
+    },
+    {
+      "epoch": 1.681312243702402,
+      "grad_norm": 0.09280505031347275,
+      "learning_rate": 1.327709431751611e-05,
+      "loss": 0.0021,
+      "step": 2870
+    },
+    {
+      "epoch": 1.687170474516696,
+      "grad_norm": 0.016819052398204803,
+      "learning_rate": 1.3253661394258934e-05,
+      "loss": 0.0595,
+      "step": 2880
+    },
+    {
+      "epoch": 1.69302870533099,
+      "grad_norm": 0.06223960965871811,
+      "learning_rate": 1.3230228471001758e-05,
+      "loss": 0.1019,
+      "step": 2890
+    },
+    {
+      "epoch": 1.698886936145284,
+      "grad_norm": 0.012930012308061123,
+      "learning_rate": 1.3206795547744582e-05,
+      "loss": 0.0211,
+      "step": 2900
+    },
+    {
+      "epoch": 1.7047451669595781,
+      "grad_norm": 10.481481552124023,
+      "learning_rate": 1.3183362624487406e-05,
+      "loss": 0.2043,
+      "step": 2910
+    },
+    {
+      "epoch": 1.7106033977738724,
+      "grad_norm": 0.48479416966438293,
+      "learning_rate": 1.315992970123023e-05,
+      "loss": 0.1616,
+      "step": 2920
+    },
+    {
+      "epoch": 1.7164616285881664,
+      "grad_norm": 0.01907259412109852,
+      "learning_rate": 1.3136496777973053e-05,
+      "loss": 0.0932,
+      "step": 2930
+    },
+    {
+      "epoch": 1.7223198594024605,
+      "grad_norm": 0.20193631947040558,
+      "learning_rate": 1.3113063854715877e-05,
+      "loss": 0.081,
+      "step": 2940
+    },
+    {
+      "epoch": 1.7281780902167545,
+      "grad_norm": 0.059374675154685974,
+      "learning_rate": 1.3089630931458701e-05,
+      "loss": 0.1085,
+      "step": 2950
+    },
+    {
+      "epoch": 1.7340363210310485,
+      "grad_norm": 0.023819535970687866,
+      "learning_rate": 1.3066198008201525e-05,
+      "loss": 0.0626,
+      "step": 2960
+    },
+    {
+      "epoch": 1.7398945518453428,
+      "grad_norm": 6.034724712371826,
+      "learning_rate": 1.304276508494435e-05,
+      "loss": 0.0466,
+      "step": 2970
+    },
+    {
+      "epoch": 1.7457527826596366,
+      "grad_norm": 0.017972761765122414,
+      "learning_rate": 1.3019332161687173e-05,
+      "loss": 0.0877,
+      "step": 2980
+    },
+    {
+      "epoch": 1.751611013473931,
+      "grad_norm": 0.4916331171989441,
+      "learning_rate": 1.2995899238429994e-05,
+      "loss": 0.1017,
+      "step": 2990
+    },
+    {
+      "epoch": 1.757469244288225,
+      "grad_norm": 0.07853368669748306,
+      "learning_rate": 1.2972466315172818e-05,
+      "loss": 0.0699,
+      "step": 3000
+    },
+    {
+      "epoch": 1.763327475102519,
+      "grad_norm": 0.03654066100716591,
+      "learning_rate": 1.2949033391915642e-05,
+      "loss": 0.0959,
+      "step": 3010
+    },
+    {
+      "epoch": 1.7691857059168132,
+      "grad_norm": 0.3065502345561981,
+      "learning_rate": 1.2925600468658466e-05,
+      "loss": 0.0642,
+      "step": 3020
+    },
+    {
+      "epoch": 1.775043936731107,
+      "grad_norm": 0.020343072712421417,
+      "learning_rate": 1.2902167545401289e-05,
+      "loss": 0.06,
+      "step": 3030
+    },
+    {
+      "epoch": 1.7809021675454013,
+      "grad_norm": 0.11528685688972473,
+      "learning_rate": 1.2878734622144113e-05,
+      "loss": 0.0314,
+      "step": 3040
+    },
+    {
+      "epoch": 1.7867603983596954,
+      "grad_norm": 0.08143705874681473,
+      "learning_rate": 1.2855301698886937e-05,
+      "loss": 0.0786,
+      "step": 3050
+    },
+    {
+      "epoch": 1.7926186291739894,
+      "grad_norm": 0.06304443627595901,
+      "learning_rate": 1.2831868775629761e-05,
+      "loss": 0.1096,
+      "step": 3060
+    },
+    {
+      "epoch": 1.7984768599882837,
+      "grad_norm": 0.020441118627786636,
+      "learning_rate": 1.2808435852372585e-05,
+      "loss": 0.1275,
+      "step": 3070
+    },
+    {
+      "epoch": 1.8043350908025775,
+      "grad_norm": 10.3707914352417,
+      "learning_rate": 1.278500292911541e-05,
+      "loss": 0.0262,
+      "step": 3080
+    },
+    {
+      "epoch": 1.8101933216168717,
+      "grad_norm": 0.03443164750933647,
+      "learning_rate": 1.2761570005858232e-05,
+      "loss": 0.0024,
+      "step": 3090
+    },
+    {
+      "epoch": 1.8160515524311658,
+      "grad_norm": 0.014218361116945744,
+      "learning_rate": 1.2738137082601056e-05,
+      "loss": 0.0672,
+      "step": 3100
+    },
+    {
+      "epoch": 1.8219097832454598,
+      "grad_norm": 0.12258031219244003,
+      "learning_rate": 1.271470415934388e-05,
+      "loss": 0.0822,
+      "step": 3110
+    },
+    {
+      "epoch": 1.827768014059754,
+      "grad_norm": 0.02281299978494644,
+      "learning_rate": 1.2691271236086702e-05,
+      "loss": 0.1317,
+      "step": 3120
+    },
+    {
+      "epoch": 1.833626244874048,
+      "grad_norm": 0.25066182017326355,
+      "learning_rate": 1.2667838312829525e-05,
+      "loss": 0.0247,
+      "step": 3130
+    },
+    {
+      "epoch": 1.8394844756883422,
+      "grad_norm": 0.01159273274242878,
+      "learning_rate": 1.2644405389572349e-05,
+      "loss": 0.0727,
+      "step": 3140
+    },
+    {
+      "epoch": 1.8453427065026362,
+      "grad_norm": 2.5836172103881836,
+      "learning_rate": 1.2620972466315173e-05,
+      "loss": 0.033,
+      "step": 3150
+    },
+    {
+      "epoch": 1.8512009373169303,
+      "grad_norm": 0.018062349408864975,
+      "learning_rate": 1.2597539543057997e-05,
+      "loss": 0.2145,
+      "step": 3160
+    },
+    {
+      "epoch": 1.8570591681312245,
+      "grad_norm": 6.357547283172607,
+      "learning_rate": 1.2574106619800821e-05,
+      "loss": 0.0635,
+      "step": 3170
+    },
+    {
+      "epoch": 1.8629173989455183,
+      "grad_norm": 0.021944306790828705,
+      "learning_rate": 1.2550673696543645e-05,
+      "loss": 0.0821,
+      "step": 3180
+    },
+    {
+      "epoch": 1.8687756297598126,
+      "grad_norm": 0.19342736899852753,
+      "learning_rate": 1.2527240773286468e-05,
+      "loss": 0.0994,
+      "step": 3190
+    },
+    {
+      "epoch": 1.8746338605741066,
+      "grad_norm": 0.1281069964170456,
+      "learning_rate": 1.2503807850029292e-05,
+      "loss": 0.2058,
+      "step": 3200
+    },
+    {
+      "epoch": 1.8804920913884007,
+      "grad_norm": 0.020071441307663918,
+      "learning_rate": 1.2480374926772116e-05,
+      "loss": 0.0817,
+      "step": 3210
+    },
+    {
+      "epoch": 1.886350322202695,
+      "grad_norm": 0.023471413180232048,
+      "learning_rate": 1.245694200351494e-05,
+      "loss": 0.0627,
+      "step": 3220
+    },
+    {
+      "epoch": 1.8922085530169888,
+      "grad_norm": 0.0315391905605793,
+      "learning_rate": 1.2433509080257764e-05,
+      "loss": 0.0035,
+      "step": 3230
+    },
+    {
+      "epoch": 1.898066783831283,
+      "grad_norm": 0.01387933362275362,
+      "learning_rate": 1.2410076157000588e-05,
+      "loss": 0.0019,
+      "step": 3240
+    },
+    {
+      "epoch": 1.903925014645577,
+      "grad_norm": 0.08367707580327988,
+      "learning_rate": 1.238664323374341e-05,
+      "loss": 0.0023,
+      "step": 3250
+    },
+    {
+      "epoch": 1.909783245459871,
+      "grad_norm": 0.020648911595344543,
+      "learning_rate": 1.2363210310486233e-05,
+      "loss": 0.0334,
+      "step": 3260
+    },
+    {
+      "epoch": 1.9156414762741654,
+      "grad_norm": 0.2322312295436859,
+      "learning_rate": 1.2339777387229057e-05,
+      "loss": 0.057,
+      "step": 3270
+    },
+    {
+      "epoch": 1.9214997070884592,
+      "grad_norm": 1.6584337949752808,
+      "learning_rate": 1.2316344463971882e-05,
+      "loss": 0.0028,
+      "step": 3280
+    },
+    {
+      "epoch": 1.9273579379027534,
+      "grad_norm": 0.12671466171741486,
+      "learning_rate": 1.2292911540714706e-05,
+      "loss": 0.1286,
+      "step": 3290
+    },
+    {
+      "epoch": 1.9332161687170475,
+      "grad_norm": 0.012327142059803009,
+      "learning_rate": 1.2269478617457528e-05,
+      "loss": 0.209,
+      "step": 3300
+    },
+    {
+      "epoch": 1.9390743995313415,
+      "grad_norm": 0.011200600303709507,
+      "learning_rate": 1.2246045694200352e-05,
+      "loss": 0.0742,
+      "step": 3310
+    },
+    {
+      "epoch": 1.9449326303456356,
+      "grad_norm": 0.019261429086327553,
+      "learning_rate": 1.2222612770943176e-05,
+      "loss": 0.0462,
+      "step": 3320
+    },
+    {
+      "epoch": 1.9507908611599296,
+      "grad_norm": 2.897488594055176,
+      "learning_rate": 1.2199179847686e-05,
+      "loss": 0.0962,
+      "step": 3330
+    },
+    {
+      "epoch": 1.9566490919742239,
+      "grad_norm": 0.047658707946538925,
+      "learning_rate": 1.2175746924428825e-05,
+      "loss": 0.0511,
+      "step": 3340
+    },
+    {
+      "epoch": 1.962507322788518,
+      "grad_norm": 0.07103870064020157,
+      "learning_rate": 1.2152314001171649e-05,
+      "loss": 0.0042,
+      "step": 3350
+    },
+    {
+      "epoch": 1.968365553602812,
+      "grad_norm": 0.06463142484426498,
+      "learning_rate": 1.2128881077914471e-05,
+      "loss": 0.0618,
+      "step": 3360
+    },
+    {
+      "epoch": 1.974223784417106,
+      "grad_norm": 0.04360285401344299,
+      "learning_rate": 1.2105448154657295e-05,
+      "loss": 0.0028,
+      "step": 3370
+    },
+    {
+      "epoch": 1.9800820152314,
+      "grad_norm": 0.06742165982723236,
+      "learning_rate": 1.2082015231400118e-05,
+      "loss": 0.0025,
+      "step": 3380
+    },
+    {
+      "epoch": 1.9859402460456943,
+      "grad_norm": 1.9260814189910889,
+      "learning_rate": 1.2058582308142942e-05,
+      "loss": 0.0641,
+      "step": 3390
+    },
+    {
+      "epoch": 1.9917984768599881,
+      "grad_norm": 9.20433521270752,
+      "learning_rate": 1.2035149384885764e-05,
+      "loss": 0.1534,
+      "step": 3400
+    },
+    {
+      "epoch": 1.9976567076742824,
+      "grad_norm": 0.06310843676328659,
+      "learning_rate": 1.2011716461628588e-05,
+      "loss": 0.0084,
+      "step": 3410
+    },
+    {
+      "epoch": 2.0035149384885766,
+      "grad_norm": 4.824597358703613,
+      "learning_rate": 1.1988283538371412e-05,
+      "loss": 0.1753,
+      "step": 3420
+    },
+    {
+      "epoch": 2.0093731693028705,
+      "grad_norm": 0.2831396162509918,
+      "learning_rate": 1.1964850615114236e-05,
+      "loss": 0.0487,
+      "step": 3430
+    },
+    {
+      "epoch": 2.0152314001171647,
+      "grad_norm": 0.027552679181098938,
+      "learning_rate": 1.194141769185706e-05,
+      "loss": 0.0609,
+      "step": 3440
+    },
+    {
+      "epoch": 2.0210896309314585,
+      "grad_norm": 0.059698984026908875,
+      "learning_rate": 1.1917984768599885e-05,
+      "loss": 0.0541,
+      "step": 3450
+    },
+    {
+      "epoch": 2.026947861745753,
+      "grad_norm": 0.024172687903046608,
+      "learning_rate": 1.1894551845342707e-05,
+      "loss": 0.0226,
+      "step": 3460
+    },
+    {
+      "epoch": 2.032806092560047,
+      "grad_norm": 0.02423352748155594,
+      "learning_rate": 1.1871118922085531e-05,
+      "loss": 0.0426,
+      "step": 3470
+    },
+    {
+      "epoch": 2.038664323374341,
+      "grad_norm": 0.01572337932884693,
+      "learning_rate": 1.1847685998828355e-05,
+      "loss": 0.0119,
+      "step": 3480
+    },
+    {
+      "epoch": 2.044522554188635,
+      "grad_norm": 0.01435055397450924,
+      "learning_rate": 1.182425307557118e-05,
+      "loss": 0.0529,
+      "step": 3490
+    },
+    {
+      "epoch": 2.050380785002929,
+      "grad_norm": 0.1196121945977211,
+      "learning_rate": 1.1800820152314004e-05,
+      "loss": 0.0428,
+      "step": 3500
+    },
+    {
+      "epoch": 2.0562390158172232,
+      "grad_norm": 0.014025365933775902,
+      "learning_rate": 1.1777387229056824e-05,
+      "loss": 0.0655,
+      "step": 3510
+    },
+    {
+      "epoch": 2.0620972466315175,
+      "grad_norm": 1.1602953672409058,
+      "learning_rate": 1.1753954305799648e-05,
+      "loss": 0.0953,
+      "step": 3520
+    },
+    {
+      "epoch": 2.0679554774458113,
+      "grad_norm": 0.04896058142185211,
+      "learning_rate": 1.1730521382542473e-05,
+      "loss": 0.0767,
+      "step": 3530
+    },
+    {
+      "epoch": 2.0738137082601056,
+      "grad_norm": 0.12855136394500732,
+      "learning_rate": 1.1707088459285297e-05,
+      "loss": 0.0127,
+      "step": 3540
+    },
+    {
+      "epoch": 2.0796719390743994,
+      "grad_norm": 0.19695617258548737,
+      "learning_rate": 1.168365553602812e-05,
+      "loss": 0.2379,
+      "step": 3550
+    },
+    {
+      "epoch": 2.0855301698886937,
+      "grad_norm": 1.3962332010269165,
+      "learning_rate": 1.1660222612770943e-05,
+      "loss": 0.1809,
+      "step": 3560
+    },
+    {
+      "epoch": 2.0913884007029875,
+      "grad_norm": 0.0690867155790329,
+      "learning_rate": 1.1636789689513767e-05,
+      "loss": 0.1417,
+      "step": 3570
+    },
+    {
+      "epoch": 2.0972466315172817,
+      "grad_norm": 0.20552076399326324,
+      "learning_rate": 1.1613356766256591e-05,
+      "loss": 0.0345,
+      "step": 3580
+    },
+    {
+      "epoch": 2.103104862331576,
+      "grad_norm": 0.014500975608825684,
+      "learning_rate": 1.1589923842999416e-05,
+      "loss": 0.034,
+      "step": 3590
+    },
+    {
+      "epoch": 2.10896309314587,
+      "grad_norm": 0.025813542306423187,
+      "learning_rate": 1.156649091974224e-05,
+      "loss": 0.1902,
+      "step": 3600
+    },
+    {
+      "epoch": 2.114821323960164,
+      "grad_norm": 0.09736531227827072,
+      "learning_rate": 1.1543057996485064e-05,
+      "loss": 0.0927,
+      "step": 3610
+    },
+    {
+      "epoch": 2.120679554774458,
+      "grad_norm": 3.991480588912964,
+      "learning_rate": 1.1519625073227886e-05,
+      "loss": 0.0578,
+      "step": 3620
+    },
+    {
+      "epoch": 2.126537785588752,
+      "grad_norm": 0.47018668055534363,
+      "learning_rate": 1.149619214997071e-05,
+      "loss": 0.0033,
+      "step": 3630
+    },
+    {
+      "epoch": 2.1323960164030464,
+      "grad_norm": 0.043486375361680984,
+      "learning_rate": 1.1472759226713533e-05,
+      "loss": 0.0472,
+      "step": 3640
+    },
+    {
+      "epoch": 2.1382542472173403,
+      "grad_norm": 0.12205738574266434,
+      "learning_rate": 1.1449326303456357e-05,
+      "loss": 0.0029,
+      "step": 3650
+    },
+    {
+      "epoch": 2.1441124780316345,
+      "grad_norm": 0.04534014314413071,
+      "learning_rate": 1.142589338019918e-05,
+      "loss": 0.1298,
+      "step": 3660
+    },
+    {
+      "epoch": 2.1499707088459283,
+      "grad_norm": 0.01751992292702198,
+      "learning_rate": 1.1402460456942003e-05,
+      "loss": 0.0116,
+      "step": 3670
+    },
+    {
+      "epoch": 2.1558289396602226,
+      "grad_norm": 0.017521362751722336,
+      "learning_rate": 1.1379027533684827e-05,
+      "loss": 0.1328,
+      "step": 3680
+    },
+    {
+      "epoch": 2.161687170474517,
+      "grad_norm": 0.015062491409480572,
+      "learning_rate": 1.1355594610427652e-05,
+      "loss": 0.0123,
+      "step": 3690
+    },
+    {
+      "epoch": 2.1675454012888107,
+      "grad_norm": 0.10153401643037796,
+      "learning_rate": 1.1332161687170476e-05,
+      "loss": 0.0029,
+      "step": 3700
+    },
+    {
+      "epoch": 2.173403632103105,
+      "grad_norm": 7.739234924316406,
+      "learning_rate": 1.13087287639133e-05,
+      "loss": 0.129,
+      "step": 3710
+    },
+    {
+      "epoch": 2.1792618629173988,
+      "grad_norm": 0.02731635794043541,
+      "learning_rate": 1.1285295840656122e-05,
+      "loss": 0.0144,
+      "step": 3720
+    },
+    {
+      "epoch": 2.185120093731693,
+      "grad_norm": 0.21970869600772858,
+      "learning_rate": 1.1261862917398946e-05,
+      "loss": 0.1084,
+      "step": 3730
+    },
+    {
+      "epoch": 2.1909783245459873,
+      "grad_norm": 0.04228947311639786,
+      "learning_rate": 1.123842999414177e-05,
+      "loss": 0.1032,
+      "step": 3740
+    },
+    {
+      "epoch": 2.196836555360281,
+      "grad_norm": 0.5683772563934326,
+      "learning_rate": 1.1214997070884595e-05,
+      "loss": 0.0031,
+      "step": 3750
+    },
+    {
+      "epoch": 2.2026947861745754,
+      "grad_norm": 0.01960350014269352,
+      "learning_rate": 1.1191564147627419e-05,
+      "loss": 0.0207,
+      "step": 3760
+    },
+    {
+      "epoch": 2.208553016988869,
+      "grad_norm": 0.22766079008579254,
+      "learning_rate": 1.116813122437024e-05,
+      "loss": 0.0379,
+      "step": 3770
+    },
+    {
+      "epoch": 2.2144112478031635,
+      "grad_norm": 0.08811239153146744,
+      "learning_rate": 1.1144698301113064e-05,
+      "loss": 0.0265,
+      "step": 3780
+    },
+    {
+      "epoch": 2.2202694786174577,
+      "grad_norm": 0.015844443812966347,
+      "learning_rate": 1.1121265377855888e-05,
+      "loss": 0.0542,
+      "step": 3790
+    },
+    {
+      "epoch": 2.2261277094317515,
+      "grad_norm": 0.022639505565166473,
+      "learning_rate": 1.1097832454598712e-05,
+      "loss": 0.0724,
+      "step": 3800
+    },
+    {
+      "epoch": 2.231985940246046,
+      "grad_norm": 0.011850577779114246,
+      "learning_rate": 1.1074399531341536e-05,
+      "loss": 0.0218,
+      "step": 3810
+    },
+    {
+      "epoch": 2.2378441710603396,
+      "grad_norm": 0.03309320658445358,
+      "learning_rate": 1.105096660808436e-05,
+      "loss": 0.004,
+      "step": 3820
+    },
+    {
+      "epoch": 2.243702401874634,
+      "grad_norm": 0.00993265025317669,
+      "learning_rate": 1.1027533684827182e-05,
+      "loss": 0.1141,
+      "step": 3830
+    },
+    {
+      "epoch": 2.249560632688928,
+      "grad_norm": 0.061374466866254807,
+      "learning_rate": 1.1004100761570007e-05,
+      "loss": 0.1819,
+      "step": 3840
+    },
+    {
+      "epoch": 2.255418863503222,
+      "grad_norm": 0.055793508887290955,
+      "learning_rate": 1.098066783831283e-05,
+      "loss": 0.0955,
+      "step": 3850
+    },
+    {
+      "epoch": 2.2612770943175162,
+      "grad_norm": 0.06129812076687813,
+      "learning_rate": 1.0957234915055655e-05,
+      "loss": 0.0022,
+      "step": 3860
+    },
+    {
+      "epoch": 2.26713532513181,
+      "grad_norm": 1.5925140380859375,
+      "learning_rate": 1.0933801991798479e-05,
+      "loss": 0.0715,
+      "step": 3870
+    },
+    {
+      "epoch": 2.2729935559461043,
+      "grad_norm": 0.6164373159408569,
+      "learning_rate": 1.0910369068541303e-05,
+      "loss": 0.0864,
+      "step": 3880
+    },
+    {
+      "epoch": 2.2788517867603986,
+      "grad_norm": 0.05251573026180267,
+      "learning_rate": 1.0886936145284125e-05,
+      "loss": 0.0227,
+      "step": 3890
+    },
+    {
+      "epoch": 2.2847100175746924,
+      "grad_norm": 5.049495220184326,
+      "learning_rate": 1.086350322202695e-05,
+      "loss": 0.1728,
+      "step": 3900
+    },
+    {
+      "epoch": 2.2905682483889866,
+      "grad_norm": 0.018607361242175102,
+      "learning_rate": 1.0840070298769772e-05,
+      "loss": 0.0381,
+      "step": 3910
+    },
+    {
+      "epoch": 2.2964264792032805,
+      "grad_norm": 0.013198847882449627,
+      "learning_rate": 1.0816637375512596e-05,
+      "loss": 0.2289,
+      "step": 3920
+    },
+    {
+      "epoch": 2.3022847100175747,
+      "grad_norm": 0.028414737433195114,
+      "learning_rate": 1.0793204452255418e-05,
+      "loss": 0.0066,
+      "step": 3930
+    },
+    {
+      "epoch": 2.3081429408318686,
+      "grad_norm": 0.02251887135207653,
+      "learning_rate": 1.0769771528998243e-05,
+      "loss": 0.0562,
+      "step": 3940
+    },
+    {
+      "epoch": 2.314001171646163,
+      "grad_norm": 0.03463216498494148,
+      "learning_rate": 1.0746338605741067e-05,
+      "loss": 0.0673,
+      "step": 3950
+    },
+    {
+      "epoch": 2.319859402460457,
+      "grad_norm": 0.028568778187036514,
+      "learning_rate": 1.072290568248389e-05,
+      "loss": 0.003,
+      "step": 3960
+    },
+    {
+      "epoch": 2.325717633274751,
+      "grad_norm": 0.012898040004074574,
+      "learning_rate": 1.0699472759226715e-05,
+      "loss": 0.005,
+      "step": 3970
+    },
+    {
+      "epoch": 2.331575864089045,
+      "grad_norm": 0.03388252109289169,
+      "learning_rate": 1.0676039835969539e-05,
+      "loss": 0.0102,
+      "step": 3980
+    },
+    {
+      "epoch": 2.3374340949033394,
+      "grad_norm": 0.02966475300490856,
+      "learning_rate": 1.0652606912712361e-05,
+      "loss": 0.0062,
+      "step": 3990
+    },
+    {
+      "epoch": 2.3432923257176332,
+      "grad_norm": 0.021042481064796448,
+      "learning_rate": 1.0629173989455186e-05,
+      "loss": 0.0849,
+      "step": 4000
+    },
+    {
+      "epoch": 2.3491505565319275,
+      "grad_norm": 0.25563105940818787,
+      "learning_rate": 1.060574106619801e-05,
+      "loss": 0.0025,
+      "step": 4010
+    },
+    {
+      "epoch": 2.3550087873462213,
+      "grad_norm": 0.010090864263474941,
+      "learning_rate": 1.0582308142940834e-05,
+      "loss": 0.0016,
+      "step": 4020
+    },
+    {
+      "epoch": 2.3608670181605156,
+      "grad_norm": 0.016884010285139084,
+      "learning_rate": 1.0558875219683658e-05,
+      "loss": 0.0906,
+      "step": 4030
+    },
+    {
+      "epoch": 2.3667252489748094,
+      "grad_norm": 0.011160296387970448,
+      "learning_rate": 1.0535442296426479e-05,
+      "loss": 0.0016,
+      "step": 4040
+    },
+    {
+      "epoch": 2.3725834797891037,
+      "grad_norm": 9.898134231567383,
+      "learning_rate": 1.0512009373169303e-05,
+      "loss": 0.1282,
+      "step": 4050
+    },
+    {
+      "epoch": 2.378441710603398,
+      "grad_norm": 4.420975685119629,
+      "learning_rate": 1.0488576449912127e-05,
+      "loss": 0.0056,
+      "step": 4060
+    },
+    {
+      "epoch": 2.3842999414176917,
+      "grad_norm": 0.08097882568836212,
+      "learning_rate": 1.0465143526654951e-05,
+      "loss": 0.0236,
+      "step": 4070
+    },
+    {
+      "epoch": 2.390158172231986,
+      "grad_norm": 0.014296771958470345,
+      "learning_rate": 1.0441710603397775e-05,
+      "loss": 0.0062,
+      "step": 4080
+    },
+    {
+      "epoch": 2.39601640304628,
+      "grad_norm": 0.08250059932470322,
+      "learning_rate": 1.0418277680140597e-05,
+      "loss": 0.0464,
+      "step": 4090
+    },
+    {
+      "epoch": 2.401874633860574,
+      "grad_norm": 0.09679897874593735,
+      "learning_rate": 1.0394844756883422e-05,
+      "loss": 0.0237,
+      "step": 4100
+    },
+    {
+      "epoch": 2.4077328646748684,
+      "grad_norm": 9.378355979919434,
+      "learning_rate": 1.0371411833626246e-05,
+      "loss": 0.0379,
+      "step": 4110
+    },
+    {
+      "epoch": 2.413591095489162,
+      "grad_norm": 11.158398628234863,
+      "learning_rate": 1.034797891036907e-05,
+      "loss": 0.1332,
+      "step": 4120
+    },
+    {
+      "epoch": 2.4194493263034564,
+      "grad_norm": 9.437854766845703,
+      "learning_rate": 1.0324545987111894e-05,
+      "loss": 0.2336,
+      "step": 4130
+    },
+    {
+      "epoch": 2.4253075571177503,
+      "grad_norm": 0.034454096108675,
+      "learning_rate": 1.0301113063854718e-05,
+      "loss": 0.0112,
+      "step": 4140
+    },
+    {
+      "epoch": 2.4311657879320445,
+      "grad_norm": 0.010446906089782715,
+      "learning_rate": 1.027768014059754e-05,
+      "loss": 0.0139,
+      "step": 4150
+    },
+    {
+      "epoch": 2.437024018746339,
+      "grad_norm": 0.008245312608778477,
+      "learning_rate": 1.0254247217340365e-05,
+      "loss": 0.0259,
+      "step": 4160
+    },
+    {
+      "epoch": 2.4428822495606326,
+      "grad_norm": 12.191228866577148,
+      "learning_rate": 1.0230814294083187e-05,
+      "loss": 0.0496,
+      "step": 4170
+    },
+    {
+      "epoch": 2.448740480374927,
+      "grad_norm": 1.1207407712936401,
+      "learning_rate": 1.0207381370826011e-05,
+      "loss": 0.1827,
+      "step": 4180
+    },
+    {
+      "epoch": 2.4545987111892207,
+      "grad_norm": 0.04254107177257538,
+      "learning_rate": 1.0183948447568834e-05,
+      "loss": 0.0021,
+      "step": 4190
+    },
+    {
+      "epoch": 2.460456942003515,
+      "grad_norm": 8.863370895385742,
+      "learning_rate": 1.0160515524311658e-05,
+      "loss": 0.0365,
+      "step": 4200
+    },
+    {
+      "epoch": 2.466315172817809,
+      "grad_norm": 0.019871840253472328,
+      "learning_rate": 1.0137082601054482e-05,
+      "loss": 0.0961,
+      "step": 4210
+    },
+    {
+      "epoch": 2.472173403632103,
+      "grad_norm": 12.787776947021484,
+      "learning_rate": 1.0113649677797306e-05,
+      "loss": 0.0048,
+      "step": 4220
+    },
+    {
+      "epoch": 2.4780316344463973,
+      "grad_norm": 0.010828366503119469,
+      "learning_rate": 1.009021675454013e-05,
+      "loss": 0.0633,
+      "step": 4230
+    },
+    {
+      "epoch": 2.483889865260691,
+      "grad_norm": 0.021939842030405998,
+      "learning_rate": 1.0066783831282954e-05,
+      "loss": 0.0236,
+      "step": 4240
+    },
+    {
+      "epoch": 2.4897480960749854,
+      "grad_norm": 0.9022541046142578,
+      "learning_rate": 1.0043350908025777e-05,
+      "loss": 0.0044,
+      "step": 4250
+    },
+    {
+      "epoch": 2.4956063268892796,
+      "grad_norm": 0.015246485359966755,
+      "learning_rate": 1.00199179847686e-05,
+      "loss": 0.0202,
+      "step": 4260
+    },
+    {
+      "epoch": 2.5014645577035735,
+      "grad_norm": 4.695196151733398,
+      "learning_rate": 9.996485061511425e-06,
+      "loss": 0.0345,
+      "step": 4270
+    },
+    {
+      "epoch": 2.5073227885178677,
+      "grad_norm": 0.3940305709838867,
+      "learning_rate": 9.973052138254247e-06,
+      "loss": 0.0334,
+      "step": 4280
+    },
+    {
+      "epoch": 2.5131810193321615,
+      "grad_norm": 0.010812806896865368,
+      "learning_rate": 9.949619214997071e-06,
+      "loss": 0.0723,
+      "step": 4290
+    },
+    {
+      "epoch": 2.519039250146456,
+      "grad_norm": 0.00904472079128027,
+      "learning_rate": 9.926186291739895e-06,
+      "loss": 0.1392,
+      "step": 4300
+    },
+    {
+      "epoch": 2.5248974809607496,
+      "grad_norm": 0.10065434128046036,
+      "learning_rate": 9.90275336848272e-06,
+      "loss": 0.0072,
+      "step": 4310
+    },
+    {
+      "epoch": 2.530755711775044,
+      "grad_norm": 0.02397744543850422,
+      "learning_rate": 9.879320445225544e-06,
+      "loss": 0.1593,
+      "step": 4320
+    },
+    {
+      "epoch": 2.536613942589338,
+      "grad_norm": 0.16617602109909058,
+      "learning_rate": 9.855887521968366e-06,
+      "loss": 0.0473,
+      "step": 4330
+    },
+    {
+      "epoch": 2.542472173403632,
+      "grad_norm": 0.03026733361184597,
+      "learning_rate": 9.83245459871119e-06,
+      "loss": 0.065,
+      "step": 4340
+    },
+    {
+      "epoch": 2.5483304042179262,
+      "grad_norm": 0.02081727422773838,
+      "learning_rate": 9.809021675454014e-06,
+      "loss": 0.0093,
+      "step": 4350
+    },
+    {
+      "epoch": 2.5541886350322205,
+      "grad_norm": 0.030087795108556747,
+      "learning_rate": 9.785588752196837e-06,
+      "loss": 0.1034,
+      "step": 4360
+    },
+    {
+      "epoch": 2.5600468658465143,
+      "grad_norm": 7.436207294464111,
+      "learning_rate": 9.76215582893966e-06,
+      "loss": 0.0905,
+      "step": 4370
+    },
+    {
+      "epoch": 2.5659050966608086,
+      "grad_norm": 0.014431145042181015,
+      "learning_rate": 9.738722905682485e-06,
+      "loss": 0.1189,
+      "step": 4380
+    },
+    {
+      "epoch": 2.5717633274751024,
+      "grad_norm": 13.210556983947754,
+      "learning_rate": 9.715289982425309e-06,
+      "loss": 0.0138,
+      "step": 4390
+    },
+    {
+      "epoch": 2.5776215582893967,
+      "grad_norm": 0.02586820349097252,
+      "learning_rate": 9.691857059168131e-06,
+      "loss": 0.0616,
+      "step": 4400
+    },
+    {
+      "epoch": 2.5834797891036905,
+      "grad_norm": 0.017524389550089836,
+      "learning_rate": 9.668424135910956e-06,
+      "loss": 0.0466,
+      "step": 4410
+    },
+    {
+      "epoch": 2.5893380199179847,
+      "grad_norm": 3.2110509872436523,
+      "learning_rate": 9.64499121265378e-06,
+      "loss": 0.007,
+      "step": 4420
+    },
+    {
+      "epoch": 2.595196250732279,
+      "grad_norm": 3.626349687576294,
+      "learning_rate": 9.621558289396604e-06,
+      "loss": 0.0712,
+      "step": 4430
+    },
+    {
+      "epoch": 2.601054481546573,
+      "grad_norm": 0.0480428971350193,
+      "learning_rate": 9.598125366139426e-06,
+      "loss": 0.0501,
+      "step": 4440
+    },
+    {
+      "epoch": 2.606912712360867,
+      "grad_norm": 0.1278507560491562,
+      "learning_rate": 9.57469244288225e-06,
+      "loss": 0.0185,
+      "step": 4450
+    },
+    {
+      "epoch": 2.6127709431751613,
+      "grad_norm": 0.01875203661620617,
+      "learning_rate": 9.551259519625073e-06,
+      "loss": 0.0111,
+      "step": 4460
+    },
+    {
+      "epoch": 2.618629173989455,
+      "grad_norm": 0.02753681316971779,
+      "learning_rate": 9.527826596367897e-06,
+      "loss": 0.0015,
+      "step": 4470
+    },
+    {
+      "epoch": 2.6244874048037494,
+      "grad_norm": 12.951713562011719,
+      "learning_rate": 9.504393673110721e-06,
+      "loss": 0.0405,
+      "step": 4480
+    },
+    {
+      "epoch": 2.6303456356180432,
+      "grad_norm": 0.025739021599292755,
+      "learning_rate": 9.480960749853545e-06,
+      "loss": 0.0115,
+      "step": 4490
+    },
+    {
+      "epoch": 2.6362038664323375,
+      "grad_norm": 0.6508920788764954,
+      "learning_rate": 9.45752782659637e-06,
+      "loss": 0.0777,
+      "step": 4500
+    },
+    {
+      "epoch": 2.6420620972466313,
+      "grad_norm": 0.0040559773333370686,
+      "learning_rate": 9.434094903339193e-06,
+      "loss": 0.0029,
+      "step": 4510
+    },
+    {
+      "epoch": 2.6479203280609256,
+      "grad_norm": 0.2040073573589325,
+      "learning_rate": 9.410661980082016e-06,
+      "loss": 0.0033,
+      "step": 4520
+    },
+    {
+      "epoch": 2.65377855887522,
+      "grad_norm": 0.14921385049819946,
+      "learning_rate": 9.38722905682484e-06,
+      "loss": 0.0383,
+      "step": 4530
+    },
+    {
+      "epoch": 2.6596367896895137,
+      "grad_norm": 0.017338193953037262,
+      "learning_rate": 9.363796133567662e-06,
+      "loss": 0.0802,
+      "step": 4540
+    },
+    {
+      "epoch": 2.665495020503808,
+      "grad_norm": 1.1883032321929932,
+      "learning_rate": 9.340363210310486e-06,
+      "loss": 0.0617,
+      "step": 4550
+    },
+    {
+      "epoch": 2.671353251318102,
+      "grad_norm": 0.0065250336192548275,
+      "learning_rate": 9.31693028705331e-06,
+      "loss": 0.0305,
+      "step": 4560
+    },
+    {
+      "epoch": 2.677211482132396,
+      "grad_norm": 4.130406379699707,
+      "learning_rate": 9.293497363796135e-06,
+      "loss": 0.0961,
+      "step": 4570
+    },
+    {
+      "epoch": 2.68306971294669,
+      "grad_norm": 0.02023070678114891,
+      "learning_rate": 9.270064440538959e-06,
+      "loss": 0.0086,
+      "step": 4580
+    },
+    {
+      "epoch": 2.688927943760984,
+      "grad_norm": 0.10395500808954239,
+      "learning_rate": 9.246631517281783e-06,
+      "loss": 0.0364,
+      "step": 4590
+    },
+    {
+      "epoch": 2.6947861745752784,
+      "grad_norm": 0.013902227394282818,
+      "learning_rate": 9.223198594024605e-06,
+      "loss": 0.01,
+      "step": 4600
+    },
+    {
+      "epoch": 2.700644405389572,
+      "grad_norm": 0.08112169802188873,
+      "learning_rate": 9.19976567076743e-06,
+      "loss": 0.0222,
+      "step": 4610
+    },
+    {
+      "epoch": 2.7065026362038664,
+      "grad_norm": 0.06331131607294083,
+      "learning_rate": 9.176332747510252e-06,
+      "loss": 0.0089,
+      "step": 4620
+    },
+    {
+      "epoch": 2.7123608670181607,
+      "grad_norm": 0.007761865854263306,
+      "learning_rate": 9.152899824253076e-06,
+      "loss": 0.0633,
+      "step": 4630
+    },
+    {
+      "epoch": 2.7182190978324545,
+      "grad_norm": 5.599218368530273,
+      "learning_rate": 9.1294669009959e-06,
+      "loss": 0.0516,
+      "step": 4640
+    },
+    {
+      "epoch": 2.724077328646749,
+      "grad_norm": 3.4084956645965576,
+      "learning_rate": 9.106033977738724e-06,
+      "loss": 0.0057,
+      "step": 4650
+    },
+    {
+      "epoch": 2.7299355594610426,
+      "grad_norm": 0.01933354139328003,
+      "learning_rate": 9.082601054481547e-06,
+      "loss": 0.0406,
+      "step": 4660
+    },
+    {
+      "epoch": 2.735793790275337,
+      "grad_norm": 0.033855751156806946,
+      "learning_rate": 9.05916813122437e-06,
+      "loss": 0.1619,
+      "step": 4670
+    },
+    {
+      "epoch": 2.7416520210896307,
+      "grad_norm": 0.009661266580224037,
+      "learning_rate": 9.035735207967195e-06,
+      "loss": 0.0012,
+      "step": 4680
+    },
+    {
+      "epoch": 2.747510251903925,
+      "grad_norm": 2.7071592807769775,
+      "learning_rate": 9.012302284710019e-06,
+      "loss": 0.0783,
+      "step": 4690
+    },
+    {
+      "epoch": 2.753368482718219,
+      "grad_norm": 0.045839034020900726,
+      "learning_rate": 8.988869361452841e-06,
+      "loss": 0.1256,
+      "step": 4700
+    },
+    {
+      "epoch": 2.759226713532513,
+      "grad_norm": 0.01187943760305643,
+      "learning_rate": 8.965436438195665e-06,
+      "loss": 0.0467,
+      "step": 4710
+    },
+    {
+      "epoch": 2.7650849443468073,
+      "grad_norm": 0.12289223819971085,
+      "learning_rate": 8.94200351493849e-06,
+      "loss": 0.224,
+      "step": 4720
+    },
+    {
+      "epoch": 2.7709431751611016,
+      "grad_norm": 0.24266257882118225,
+      "learning_rate": 8.918570591681312e-06,
+      "loss": 0.0033,
+      "step": 4730
+    },
+    {
+      "epoch": 2.7768014059753954,
+      "grad_norm": 0.05902494117617607,
+      "learning_rate": 8.895137668424136e-06,
+      "loss": 0.0786,
+      "step": 4740
+    },
+    {
+      "epoch": 2.7826596367896896,
+      "grad_norm": 0.03103451617062092,
+      "learning_rate": 8.87170474516696e-06,
+      "loss": 0.062,
+      "step": 4750
+    },
+    {
+      "epoch": 2.7885178676039835,
+      "grad_norm": 0.01481554750353098,
+      "learning_rate": 8.848271821909784e-06,
+      "loss": 0.0537,
+      "step": 4760
+    },
+    {
+      "epoch": 2.7943760984182777,
+      "grad_norm": 0.0064656296744942665,
+      "learning_rate": 8.824838898652608e-06,
+      "loss": 0.0799,
+      "step": 4770
+    },
+    {
+      "epoch": 2.8002343292325715,
+      "grad_norm": 0.018277239054441452,
+      "learning_rate": 8.801405975395433e-06,
+      "loss": 0.0438,
+      "step": 4780
+    },
+    {
+      "epoch": 2.806092560046866,
+      "grad_norm": 0.008237453177571297,
+      "learning_rate": 8.777973052138255e-06,
+      "loss": 0.0481,
+      "step": 4790
+    },
+    {
+      "epoch": 2.81195079086116,
+      "grad_norm": 0.06086944043636322,
+      "learning_rate": 8.754540128881079e-06,
+      "loss": 0.0578,
+      "step": 4800
+    },
+    {
+      "epoch": 2.817809021675454,
+      "grad_norm": 0.0078117563389241695,
+      "learning_rate": 8.731107205623902e-06,
+      "loss": 0.0054,
+      "step": 4810
+    },
+    {
+      "epoch": 2.823667252489748,
+      "grad_norm": 0.22204460203647614,
+      "learning_rate": 8.707674282366726e-06,
+      "loss": 0.0452,
+      "step": 4820
+    },
+    {
+      "epoch": 2.8295254833040424,
+      "grad_norm": 0.011765822768211365,
+      "learning_rate": 8.68424135910955e-06,
+      "loss": 0.1226,
+      "step": 4830
+    },
+    {
+      "epoch": 2.8353837141183362,
+      "grad_norm": 0.034977905452251434,
+      "learning_rate": 8.660808435852374e-06,
+      "loss": 0.0028,
+      "step": 4840
+    },
+    {
+      "epoch": 2.8412419449326305,
+      "grad_norm": 0.006177605129778385,
+      "learning_rate": 8.637375512595198e-06,
+      "loss": 0.0055,
+      "step": 4850
+    },
+    {
+      "epoch": 2.8471001757469243,
+      "grad_norm": 0.06611007452011108,
+      "learning_rate": 8.61394258933802e-06,
+      "loss": 0.0012,
+      "step": 4860
+    },
+    {
+      "epoch": 2.8529584065612186,
+      "grad_norm": 0.007338257972151041,
+      "learning_rate": 8.590509666080845e-06,
+      "loss": 0.0413,
+      "step": 4870
+    },
+    {
+      "epoch": 2.8588166373755124,
+      "grad_norm": 0.04989036172628403,
+      "learning_rate": 8.567076742823669e-06,
+      "loss": 0.0606,
+      "step": 4880
+    },
+    {
+      "epoch": 2.8646748681898067,
+      "grad_norm": 0.007559357676655054,
+      "learning_rate": 8.543643819566491e-06,
+      "loss": 0.1396,
+      "step": 4890
+    },
+    {
+      "epoch": 2.870533099004101,
+      "grad_norm": 0.05906621739268303,
+      "learning_rate": 8.520210896309315e-06,
+      "loss": 0.1917,
+      "step": 4900
+    },
+    {
+      "epoch": 2.8763913298183947,
+      "grad_norm": 0.012134838849306107,
+      "learning_rate": 8.49677797305214e-06,
+      "loss": 0.0011,
+      "step": 4910
+    },
+    {
+      "epoch": 2.882249560632689,
+      "grad_norm": 4.250200271606445,
+      "learning_rate": 8.473345049794962e-06,
+      "loss": 0.1104,
+      "step": 4920
+    },
+    {
+      "epoch": 2.8881077914469833,
+      "grad_norm": 0.15310168266296387,
+      "learning_rate": 8.449912126537786e-06,
+      "loss": 0.1596,
+      "step": 4930
+    },
+    {
+      "epoch": 2.893966022261277,
+      "grad_norm": 7.746487617492676,
+      "learning_rate": 8.42647920328061e-06,
+      "loss": 0.1139,
+      "step": 4940
+    },
+    {
+      "epoch": 2.899824253075571,
+      "grad_norm": 4.003668785095215,
+      "learning_rate": 8.403046280023434e-06,
+      "loss": 0.1337,
+      "step": 4950
+    },
+    {
+      "epoch": 2.905682483889865,
+      "grad_norm": 0.013791323639452457,
+      "learning_rate": 8.379613356766258e-06,
+      "loss": 0.0577,
+      "step": 4960
+    },
+    {
+      "epoch": 2.9115407147041594,
+      "grad_norm": 0.03509506955742836,
+      "learning_rate": 8.35618043350908e-06,
+      "loss": 0.0023,
+      "step": 4970
+    },
+    {
+      "epoch": 2.9173989455184532,
+      "grad_norm": 0.017392152920365334,
+      "learning_rate": 8.332747510251905e-06,
+      "loss": 0.0017,
+      "step": 4980
+    },
+    {
+      "epoch": 2.9232571763327475,
+      "grad_norm": 0.0742042288184166,
+      "learning_rate": 8.309314586994727e-06,
+      "loss": 0.0583,
+      "step": 4990
+    },
+    {
+      "epoch": 2.9291154071470418,
+      "grad_norm": 0.025125756859779358,
+      "learning_rate": 8.285881663737551e-06,
+      "loss": 0.0018,
+      "step": 5000
+    },
+    {
+      "epoch": 2.9349736379613356,
+      "grad_norm": 0.01341651938855648,
+      "learning_rate": 8.262448740480375e-06,
+      "loss": 0.0487,
+      "step": 5010
+    },
+    {
+      "epoch": 2.94083186877563,
+      "grad_norm": 0.024638604372739792,
+      "learning_rate": 8.2390158172232e-06,
+      "loss": 0.001,
+      "step": 5020
+    },
+    {
+      "epoch": 2.946690099589924,
+      "grad_norm": 0.038911301642656326,
+      "learning_rate": 8.215582893966024e-06,
+      "loss": 0.0506,
+      "step": 5030
+    },
+    {
+      "epoch": 2.952548330404218,
+      "grad_norm": 0.023850092664361,
+      "learning_rate": 8.192149970708848e-06,
+      "loss": 0.0145,
+      "step": 5040
+    },
+    {
+      "epoch": 2.9584065612185118,
+      "grad_norm": 0.012319614179432392,
+      "learning_rate": 8.16871704745167e-06,
+      "loss": 0.0683,
+      "step": 5050
+    },
+    {
+      "epoch": 2.964264792032806,
+      "grad_norm": 0.011331205256283283,
+      "learning_rate": 8.145284124194494e-06,
+      "loss": 0.001,
+      "step": 5060
+    },
+    {
+      "epoch": 2.9701230228471003,
+      "grad_norm": 0.003681463422253728,
+      "learning_rate": 8.121851200937317e-06,
+      "loss": 0.0044,
+      "step": 5070
+    },
+    {
+      "epoch": 2.975981253661394,
+      "grad_norm": 0.051831282675266266,
+      "learning_rate": 8.09841827768014e-06,
+      "loss": 0.0034,
+      "step": 5080
+    },
+    {
+      "epoch": 2.9818394844756884,
+      "grad_norm": 0.07008065283298492,
+      "learning_rate": 8.074985354422965e-06,
+      "loss": 0.1246,
+      "step": 5090
+    },
+    {
+      "epoch": 2.9876977152899826,
+      "grad_norm": 0.009249912574887276,
+      "learning_rate": 8.051552431165789e-06,
+      "loss": 0.0006,
+      "step": 5100
+    },
+    {
+      "epoch": 2.9935559461042764,
+      "grad_norm": 0.019453031942248344,
+      "learning_rate": 8.028119507908613e-06,
+      "loss": 0.0053,
+      "step": 5110
+    },
+    {
+      "epoch": 2.9994141769185707,
+      "grad_norm": 0.5509079098701477,
+      "learning_rate": 8.004686584651435e-06,
+      "loss": 0.0994,
+      "step": 5120
+    },
+    {
+      "epoch": 3.0052724077328645,
+      "grad_norm": 0.012448657304048538,
+      "learning_rate": 7.98125366139426e-06,
+      "loss": 0.0016,
+      "step": 5130
+    },
+    {
+      "epoch": 3.011130638547159,
+      "grad_norm": 6.02830171585083,
+      "learning_rate": 7.957820738137084e-06,
+      "loss": 0.0384,
+      "step": 5140
+    },
+    {
+      "epoch": 3.016988869361453,
+      "grad_norm": 0.013466407544910908,
+      "learning_rate": 7.934387814879906e-06,
+      "loss": 0.0013,
+      "step": 5150
+    },
+    {
+      "epoch": 3.022847100175747,
+      "grad_norm": 10.52933120727539,
+      "learning_rate": 7.91095489162273e-06,
+      "loss": 0.0748,
+      "step": 5160
+    },
+    {
+      "epoch": 3.028705330990041,
+      "grad_norm": 4.3404622077941895,
+      "learning_rate": 7.887521968365554e-06,
+      "loss": 0.1159,
+      "step": 5170
+    },
+    {
+      "epoch": 3.034563561804335,
+      "grad_norm": 0.016072312369942665,
+      "learning_rate": 7.864089045108378e-06,
+      "loss": 0.0696,
+      "step": 5180
+    },
+    {
+      "epoch": 3.040421792618629,
+      "grad_norm": 0.06083187833428383,
+      "learning_rate": 7.840656121851201e-06,
+      "loss": 0.0322,
+      "step": 5190
+    },
+    {
+      "epoch": 3.0462800234329235,
+      "grad_norm": 0.012007597833871841,
+      "learning_rate": 7.817223198594025e-06,
+      "loss": 0.1097,
+      "step": 5200
+    },
+    {
+      "epoch": 3.0521382542472173,
+      "grad_norm": 0.06487168371677399,
+      "learning_rate": 7.793790275336849e-06,
+      "loss": 0.0023,
+      "step": 5210
+    },
+    {
+      "epoch": 3.0579964850615116,
+      "grad_norm": 0.07004842162132263,
+      "learning_rate": 7.770357352079673e-06,
+      "loss": 0.0008,
+      "step": 5220
+    },
+    {
+      "epoch": 3.0638547158758054,
+      "grad_norm": 0.05898735299706459,
+      "learning_rate": 7.746924428822497e-06,
+      "loss": 0.0533,
+      "step": 5230
+    },
+    {
+      "epoch": 3.0697129466900996,
+      "grad_norm": 0.015656787902116776,
+      "learning_rate": 7.72349150556532e-06,
+      "loss": 0.0414,
+      "step": 5240
+    },
+    {
+      "epoch": 3.0755711775043935,
+      "grad_norm": 0.013448948040604591,
+      "learning_rate": 7.700058582308144e-06,
+      "loss": 0.0338,
+      "step": 5250
+    },
+    {
+      "epoch": 3.0814294083186877,
+      "grad_norm": 5.860596656799316,
+      "learning_rate": 7.676625659050966e-06,
+      "loss": 0.065,
+      "step": 5260
+    },
+    {
+      "epoch": 3.087287639132982,
+      "grad_norm": 0.013592137955129147,
+      "learning_rate": 7.65319273579379e-06,
+      "loss": 0.007,
+      "step": 5270
+    },
+    {
+      "epoch": 3.093145869947276,
+      "grad_norm": 20.216941833496094,
+      "learning_rate": 7.629759812536615e-06,
+      "loss": 0.1278,
+      "step": 5280
+    },
+    {
+      "epoch": 3.09900410076157,
+      "grad_norm": 0.07607584446668625,
+      "learning_rate": 7.606326889279439e-06,
+      "loss": 0.0025,
+      "step": 5290
+    },
+    {
+      "epoch": 3.104862331575864,
+      "grad_norm": 3.425032615661621,
+      "learning_rate": 7.582893966022262e-06,
+      "loss": 0.058,
+      "step": 5300
+    },
+    {
+      "epoch": 3.110720562390158,
+      "grad_norm": 0.026434851810336113,
+      "learning_rate": 7.559461042765086e-06,
+      "loss": 0.0361,
+      "step": 5310
+    },
+    {
+      "epoch": 3.1165787932044524,
+      "grad_norm": 0.047157056629657745,
+      "learning_rate": 7.5360281195079085e-06,
+      "loss": 0.0357,
+      "step": 5320
+    },
+    {
+      "epoch": 3.1224370240187462,
+      "grad_norm": 0.16366933286190033,
+      "learning_rate": 7.5125951962507326e-06,
+      "loss": 0.0015,
+      "step": 5330
+    },
+    {
+      "epoch": 3.1282952548330405,
+      "grad_norm": 0.013051210902631283,
+      "learning_rate": 7.489162272993557e-06,
+      "loss": 0.0354,
+      "step": 5340
+    },
+    {
+      "epoch": 3.1341534856473343,
+      "grad_norm": 0.05217352882027626,
+      "learning_rate": 7.46572934973638e-06,
+      "loss": 0.0009,
+      "step": 5350
+    },
+    {
+      "epoch": 3.1400117164616286,
+      "grad_norm": 7.535463333129883,
+      "learning_rate": 7.442296426479204e-06,
+      "loss": 0.0053,
+      "step": 5360
+    },
+    {
+      "epoch": 3.145869947275923,
+      "grad_norm": 0.7577245831489563,
+      "learning_rate": 7.418863503222028e-06,
+      "loss": 0.0021,
+      "step": 5370
+    },
+    {
+      "epoch": 3.1517281780902167,
+      "grad_norm": 0.09404406696557999,
+      "learning_rate": 7.395430579964851e-06,
+      "loss": 0.0646,
+      "step": 5380
+    },
+    {
+      "epoch": 3.157586408904511,
+      "grad_norm": 0.009803865104913712,
+      "learning_rate": 7.371997656707675e-06,
+      "loss": 0.1086,
+      "step": 5390
+    },
+    {
+      "epoch": 3.1634446397188047,
+      "grad_norm": 0.010212755762040615,
+      "learning_rate": 7.348564733450498e-06,
+      "loss": 0.0056,
+      "step": 5400
+    },
+    {
+      "epoch": 3.169302870533099,
+      "grad_norm": 0.04309909790754318,
+      "learning_rate": 7.325131810193322e-06,
+      "loss": 0.0217,
+      "step": 5410
+    },
+    {
+      "epoch": 3.1751611013473933,
+      "grad_norm": 2.965693712234497,
+      "learning_rate": 7.301698886936146e-06,
+      "loss": 0.0442,
+      "step": 5420
+    },
+    {
+      "epoch": 3.181019332161687,
+      "grad_norm": 0.6837348937988281,
+      "learning_rate": 7.2782659636789695e-06,
+      "loss": 0.1925,
+      "step": 5430
+    },
+    {
+      "epoch": 3.1868775629759813,
+      "grad_norm": 0.029974374920129776,
+      "learning_rate": 7.254833040421794e-06,
+      "loss": 0.001,
+      "step": 5440
+    },
+    {
+      "epoch": 3.192735793790275,
+      "grad_norm": 0.053507156670093536,
+      "learning_rate": 7.231400117164616e-06,
+      "loss": 0.0618,
+      "step": 5450
+    },
+    {
+      "epoch": 3.1985940246045694,
+      "grad_norm": 0.00892311055213213,
+      "learning_rate": 7.20796719390744e-06,
+      "loss": 0.0115,
+      "step": 5460
+    },
+    {
+      "epoch": 3.2044522554188637,
+      "grad_norm": 0.0072519052773714066,
+      "learning_rate": 7.184534270650264e-06,
+      "loss": 0.003,
+      "step": 5470
+    },
+    {
+      "epoch": 3.2103104862331575,
+      "grad_norm": 0.3369971215724945,
+      "learning_rate": 7.161101347393088e-06,
+      "loss": 0.0478,
+      "step": 5480
+    },
+    {
+      "epoch": 3.2161687170474518,
+      "grad_norm": 5.9024553298950195,
+      "learning_rate": 7.137668424135912e-06,
+      "loss": 0.0551,
+      "step": 5490
+    },
+    {
+      "epoch": 3.2220269478617456,
+      "grad_norm": 0.0728222206234932,
+      "learning_rate": 7.114235500878736e-06,
+      "loss": 0.0039,
+      "step": 5500
+    },
+    {
+      "epoch": 3.22788517867604,
+      "grad_norm": 0.012450517155230045,
+      "learning_rate": 7.090802577621558e-06,
+      "loss": 0.0358,
+      "step": 5510
+    },
+    {
+      "epoch": 3.233743409490334,
+      "grad_norm": 1.3134894371032715,
+      "learning_rate": 7.067369654364382e-06,
+      "loss": 0.066,
+      "step": 5520
+    },
+    {
+      "epoch": 3.239601640304628,
+      "grad_norm": 1.1952403783798218,
+      "learning_rate": 7.043936731107206e-06,
+      "loss": 0.0938,
+      "step": 5530
+    },
+    {
+      "epoch": 3.245459871118922,
+      "grad_norm": 0.003476966405287385,
+      "learning_rate": 7.02050380785003e-06,
+      "loss": 0.0918,
+      "step": 5540
+    },
+    {
+      "epoch": 3.251318101933216,
+      "grad_norm": 0.3250567317008972,
+      "learning_rate": 6.997070884592854e-06,
+      "loss": 0.0323,
+      "step": 5550
+    },
+    {
+      "epoch": 3.2571763327475103,
+      "grad_norm": 0.02057233452796936,
+      "learning_rate": 6.973637961335678e-06,
+      "loss": 0.0259,
+      "step": 5560
+    },
+    {
+      "epoch": 3.2630345635618045,
+      "grad_norm": 0.23570382595062256,
+      "learning_rate": 6.950205038078501e-06,
+      "loss": 0.0075,
+      "step": 5570
+    },
+    {
+      "epoch": 3.2688927943760984,
+      "grad_norm": 0.10391777008771896,
+      "learning_rate": 6.926772114821324e-06,
+      "loss": 0.0617,
+      "step": 5580
+    },
+    {
+      "epoch": 3.2747510251903926,
+      "grad_norm": 0.02281038463115692,
+      "learning_rate": 6.903339191564148e-06,
+      "loss": 0.1036,
+      "step": 5590
+    },
+    {
+      "epoch": 3.2806092560046864,
+      "grad_norm": 0.003879611613228917,
+      "learning_rate": 6.879906268306972e-06,
+      "loss": 0.0023,
+      "step": 5600
+    },
+    {
+      "epoch": 3.2864674868189807,
+      "grad_norm": 0.00758686289191246,
+      "learning_rate": 6.856473345049796e-06,
+      "loss": 0.058,
+      "step": 5610
+    },
+    {
+      "epoch": 3.2923257176332745,
+      "grad_norm": 1.52095365524292,
+      "learning_rate": 6.833040421792619e-06,
+      "loss": 0.0806,
+      "step": 5620
+    },
+    {
+      "epoch": 3.298183948447569,
+      "grad_norm": 0.0948934480547905,
+      "learning_rate": 6.809607498535443e-06,
+      "loss": 0.0012,
+      "step": 5630
+    },
+    {
+      "epoch": 3.304042179261863,
+      "grad_norm": 1.0891087055206299,
+      "learning_rate": 6.786174575278267e-06,
+      "loss": 0.0029,
+      "step": 5640
+    },
+    {
+      "epoch": 3.309900410076157,
+      "grad_norm": 0.1640109270811081,
+      "learning_rate": 6.76274165202109e-06,
+      "loss": 0.0035,
+      "step": 5650
+    },
+    {
+      "epoch": 3.315758640890451,
+      "grad_norm": 0.005726497154682875,
+      "learning_rate": 6.739308728763914e-06,
+      "loss": 0.0679,
+      "step": 5660
+    },
+    {
+      "epoch": 3.3216168717047454,
+      "grad_norm": 0.6655123233795166,
+      "learning_rate": 6.715875805506737e-06,
+      "loss": 0.0859,
+      "step": 5670
+    },
+    {
+      "epoch": 3.327475102519039,
+      "grad_norm": 0.015817932784557343,
+      "learning_rate": 6.692442882249561e-06,
+      "loss": 0.0406,
+      "step": 5680
+    },
+    {
+      "epoch": 3.3333333333333335,
+      "grad_norm": 0.7005910873413086,
+      "learning_rate": 6.669009958992385e-06,
+      "loss": 0.019,
+      "step": 5690
+    },
+    {
+      "epoch": 3.3391915641476273,
+      "grad_norm": 0.09578946232795715,
+      "learning_rate": 6.645577035735209e-06,
+      "loss": 0.0703,
+      "step": 5700
+    },
+    {
+      "epoch": 3.3450497949619216,
+      "grad_norm": 0.05586531385779381,
+      "learning_rate": 6.622144112478032e-06,
+      "loss": 0.0011,
+      "step": 5710
+    },
+    {
+      "epoch": 3.3509080257762154,
+      "grad_norm": 0.005320393946021795,
+      "learning_rate": 6.598711189220855e-06,
+      "loss": 0.0006,
+      "step": 5720
+    },
+    {
+      "epoch": 3.3567662565905096,
+      "grad_norm": 0.012893118895590305,
+      "learning_rate": 6.575278265963679e-06,
+      "loss": 0.0128,
+      "step": 5730
+    },
+    {
+      "epoch": 3.362624487404804,
+      "grad_norm": 0.014687235467135906,
+      "learning_rate": 6.5518453427065035e-06,
+      "loss": 0.0048,
+      "step": 5740
+    },
+    {
+      "epoch": 3.3684827182190977,
+      "grad_norm": 5.860557556152344,
+      "learning_rate": 6.528412419449327e-06,
+      "loss": 0.096,
+      "step": 5750
+    },
+    {
+      "epoch": 3.374340949033392,
+      "grad_norm": 3.1887881755828857,
+      "learning_rate": 6.504979496192151e-06,
+      "loss": 0.0686,
+      "step": 5760
+    },
+    {
+      "epoch": 3.380199179847686,
+      "grad_norm": 0.10010349005460739,
+      "learning_rate": 6.481546572934975e-06,
+      "loss": 0.0462,
+      "step": 5770
+    },
+    {
+      "epoch": 3.38605741066198,
+      "grad_norm": 0.0034016838762909174,
+      "learning_rate": 6.458113649677797e-06,
+      "loss": 0.0645,
+      "step": 5780
+    },
+    {
+      "epoch": 3.3919156414762743,
+      "grad_norm": 0.008133571594953537,
+      "learning_rate": 6.4346807264206215e-06,
+      "loss": 0.0044,
+      "step": 5790
+    },
+    {
+      "epoch": 3.397773872290568,
+      "grad_norm": 0.024280358105897903,
+      "learning_rate": 6.411247803163445e-06,
+      "loss": 0.0012,
+      "step": 5800
+    },
+    {
+      "epoch": 3.4036321031048624,
+      "grad_norm": 0.07370976358652115,
+      "learning_rate": 6.387814879906269e-06,
+      "loss": 0.002,
+      "step": 5810
+    },
+    {
+      "epoch": 3.4094903339191562,
+      "grad_norm": 0.002925069537013769,
+      "learning_rate": 6.364381956649093e-06,
+      "loss": 0.0899,
+      "step": 5820
+    },
+    {
+      "epoch": 3.4153485647334505,
+      "grad_norm": 0.0019779757130891085,
+      "learning_rate": 6.340949033391916e-06,
+      "loss": 0.0552,
+      "step": 5830
+    },
+    {
+      "epoch": 3.4212067955477448,
+      "grad_norm": 0.004934338852763176,
+      "learning_rate": 6.3175161101347395e-06,
+      "loss": 0.0536,
+      "step": 5840
+    },
+    {
+      "epoch": 3.4270650263620386,
+      "grad_norm": 0.014058866538107395,
+      "learning_rate": 6.294083186877563e-06,
+      "loss": 0.0165,
+      "step": 5850
+    },
+    {
+      "epoch": 3.432923257176333,
+      "grad_norm": 0.004640714265406132,
+      "learning_rate": 6.270650263620387e-06,
+      "loss": 0.0161,
+      "step": 5860
+    },
+    {
+      "epoch": 3.4387814879906267,
+      "grad_norm": 0.00481452327221632,
+      "learning_rate": 6.247217340363211e-06,
+      "loss": 0.0007,
+      "step": 5870
+    },
+    {
+      "epoch": 3.444639718804921,
+      "grad_norm": 0.0019116317853331566,
+      "learning_rate": 6.223784417106034e-06,
+      "loss": 0.0305,
+      "step": 5880
+    },
+    {
+      "epoch": 3.450497949619215,
+      "grad_norm": 0.003990430850535631,
+      "learning_rate": 6.200351493848858e-06,
+      "loss": 0.0041,
+      "step": 5890
+    },
+    {
+      "epoch": 3.456356180433509,
+      "grad_norm": 0.0065063601359725,
+      "learning_rate": 6.1769185705916825e-06,
+      "loss": 0.0761,
+      "step": 5900
+    },
+    {
+      "epoch": 3.4622144112478033,
+      "grad_norm": 0.11059655994176865,
+      "learning_rate": 6.153485647334505e-06,
+      "loss": 0.0466,
+      "step": 5910
+    },
+    {
+      "epoch": 3.468072642062097,
+      "grad_norm": 0.02467559650540352,
+      "learning_rate": 6.130052724077329e-06,
+      "loss": 0.0004,
+      "step": 5920
+    },
+    {
+      "epoch": 3.4739308728763914,
+      "grad_norm": 0.043093279004096985,
+      "learning_rate": 6.106619800820152e-06,
+      "loss": 0.0302,
+      "step": 5930
+    },
+    {
+      "epoch": 3.4797891036906856,
+      "grad_norm": 0.008807145990431309,
+      "learning_rate": 6.083186877562976e-06,
+      "loss": 0.019,
+      "step": 5940
+    },
+    {
+      "epoch": 3.4856473345049794,
+      "grad_norm": 0.0038015455938875675,
+      "learning_rate": 6.0597539543058005e-06,
+      "loss": 0.0585,
+      "step": 5950
+    },
+    {
+      "epoch": 3.4915055653192737,
+      "grad_norm": 0.00422382028773427,
+      "learning_rate": 6.036321031048624e-06,
+      "loss": 0.0962,
+      "step": 5960
+    },
+    {
+      "epoch": 3.4973637961335675,
+      "grad_norm": 0.016047485172748566,
+      "learning_rate": 6.012888107791447e-06,
+      "loss": 0.0006,
+      "step": 5970
+    },
+    {
+      "epoch": 3.503222026947862,
+      "grad_norm": 0.005743114277720451,
+      "learning_rate": 5.989455184534271e-06,
+      "loss": 0.0021,
+      "step": 5980
+    },
+    {
+      "epoch": 3.5090802577621556,
+      "grad_norm": 0.12893952429294586,
+      "learning_rate": 5.9660222612770944e-06,
+      "loss": 0.0011,
+      "step": 5990
+    },
+    {
+      "epoch": 3.51493848857645,
+      "grad_norm": 0.05099204555153847,
+      "learning_rate": 5.9425893380199186e-06,
+      "loss": 0.0009,
+      "step": 6000
+    },
+    {
+      "epoch": 3.520796719390744,
+      "grad_norm": 0.09570227563381195,
+      "learning_rate": 5.919156414762743e-06,
+      "loss": 0.0159,
+      "step": 6010
+    },
+    {
+      "epoch": 3.526654950205038,
+      "grad_norm": 0.055699292570352554,
+      "learning_rate": 5.895723491505566e-06,
+      "loss": 0.0012,
+      "step": 6020
+    },
+    {
+      "epoch": 3.532513181019332,
+      "grad_norm": 0.010486723855137825,
+      "learning_rate": 5.87229056824839e-06,
+      "loss": 0.0076,
+      "step": 6030
+    },
+    {
+      "epoch": 3.5383714118336265,
+      "grad_norm": 0.006424172315746546,
+      "learning_rate": 5.8488576449912125e-06,
+      "loss": 0.0119,
+      "step": 6040
+    },
+    {
+      "epoch": 3.5442296426479203,
+      "grad_norm": 0.042542118579149246,
+      "learning_rate": 5.825424721734037e-06,
+      "loss": 0.0117,
+      "step": 6050
+    },
+    {
+      "epoch": 3.5500878734622145,
+      "grad_norm": 0.009197155013680458,
+      "learning_rate": 5.801991798476861e-06,
+      "loss": 0.0016,
+      "step": 6060
+    },
+    {
+      "epoch": 3.5559461042765084,
+      "grad_norm": 0.020288530737161636,
+      "learning_rate": 5.778558875219684e-06,
+      "loss": 0.0012,
+      "step": 6070
+    },
+    {
+      "epoch": 3.5618043350908026,
+      "grad_norm": 14.225655555725098,
+      "learning_rate": 5.755125951962508e-06,
+      "loss": 0.0402,
+      "step": 6080
+    },
+    {
+      "epoch": 3.5676625659050965,
+      "grad_norm": 0.41536006331443787,
+      "learning_rate": 5.731693028705332e-06,
+      "loss": 0.0442,
+      "step": 6090
+    },
+    {
+      "epoch": 3.5735207967193907,
+      "grad_norm": 0.025049546733498573,
+      "learning_rate": 5.7082601054481555e-06,
+      "loss": 0.0009,
+      "step": 6100
+    },
+    {
+      "epoch": 3.579379027533685,
+      "grad_norm": 0.006835598032921553,
+      "learning_rate": 5.684827182190979e-06,
+      "loss": 0.18,
+      "step": 6110
+    },
+    {
+      "epoch": 3.585237258347979,
+      "grad_norm": 0.0030140692833811045,
+      "learning_rate": 5.661394258933802e-06,
+      "loss": 0.082,
+      "step": 6120
+    },
+    {
+      "epoch": 3.591095489162273,
+      "grad_norm": 0.008736114017665386,
+      "learning_rate": 5.637961335676626e-06,
+      "loss": 0.0187,
+      "step": 6130
+    },
+    {
+      "epoch": 3.5969537199765673,
+      "grad_norm": 0.053168874233961105,
+      "learning_rate": 5.61452841241945e-06,
+      "loss": 0.0209,
+      "step": 6140
+    },
+    {
+      "epoch": 3.602811950790861,
+      "grad_norm": 0.11160540580749512,
+      "learning_rate": 5.5910954891622735e-06,
+      "loss": 0.0193,
+      "step": 6150
+    },
+    {
+      "epoch": 3.6086701816051554,
+      "grad_norm": 0.0031351852230727673,
+      "learning_rate": 5.567662565905098e-06,
+      "loss": 0.0008,
+      "step": 6160
+    },
+    {
+      "epoch": 3.614528412419449,
+      "grad_norm": 10.10761547088623,
+      "learning_rate": 5.54422964264792e-06,
+      "loss": 0.0278,
+      "step": 6170
+    },
+    {
+      "epoch": 3.6203866432337435,
+      "grad_norm": 0.03536173328757286,
+      "learning_rate": 5.520796719390744e-06,
+      "loss": 0.032,
+      "step": 6180
+    },
+    {
+      "epoch": 3.6262448740480373,
+      "grad_norm": 0.05558720603585243,
+      "learning_rate": 5.497363796133568e-06,
+      "loss": 0.0622,
+      "step": 6190
+    },
+    {
+      "epoch": 3.6321031048623316,
+      "grad_norm": 0.011010909453034401,
+      "learning_rate": 5.4739308728763915e-06,
+      "loss": 0.0061,
+      "step": 6200
+    },
+    {
+      "epoch": 3.637961335676626,
+      "grad_norm": 0.03313250094652176,
+      "learning_rate": 5.450497949619216e-06,
+      "loss": 0.0008,
+      "step": 6210
+    },
+    {
+      "epoch": 3.6438195664909196,
+      "grad_norm": 5.195527076721191,
+      "learning_rate": 5.42706502636204e-06,
+      "loss": 0.2337,
+      "step": 6220
+    },
+    {
+      "epoch": 3.649677797305214,
+      "grad_norm": 0.05604048818349838,
+      "learning_rate": 5.403632103104863e-06,
+      "loss": 0.0463,
+      "step": 6230
+    },
+    {
+      "epoch": 3.655536028119508,
+      "grad_norm": 0.012128411792218685,
+      "learning_rate": 5.380199179847686e-06,
+      "loss": 0.0006,
+      "step": 6240
+    },
+    {
+      "epoch": 3.661394258933802,
+      "grad_norm": 5.831113338470459,
+      "learning_rate": 5.3567662565905095e-06,
+      "loss": 0.1197,
+      "step": 6250
+    },
+    {
+      "epoch": 3.667252489748096,
+      "grad_norm": 0.009865056723356247,
+      "learning_rate": 5.333333333333334e-06,
+      "loss": 0.1289,
+      "step": 6260
+    },
+    {
+      "epoch": 3.67311072056239,
+      "grad_norm": 0.011429395526647568,
+      "learning_rate": 5.309900410076158e-06,
+      "loss": 0.0004,
+      "step": 6270
+    },
+    {
+      "epoch": 3.6789689513766843,
+      "grad_norm": 0.01841702312231064,
+      "learning_rate": 5.286467486818981e-06,
+      "loss": 0.0037,
+      "step": 6280
+    },
+    {
+      "epoch": 3.684827182190978,
+      "grad_norm": 0.008382062427699566,
+      "learning_rate": 5.263034563561805e-06,
+      "loss": 0.0015,
+      "step": 6290
+    },
+    {
+      "epoch": 3.6906854130052724,
+      "grad_norm": 0.002653555478900671,
+      "learning_rate": 5.2396016403046276e-06,
+      "loss": 0.0074,
+      "step": 6300
+    },
+    {
+      "epoch": 3.6965436438195667,
+      "grad_norm": 0.02200056053698063,
+      "learning_rate": 5.216168717047452e-06,
+      "loss": 0.0071,
+      "step": 6310
+    },
+    {
+      "epoch": 3.7024018746338605,
+      "grad_norm": 0.12032772600650787,
+      "learning_rate": 5.192735793790276e-06,
+      "loss": 0.0742,
+      "step": 6320
+    },
+    {
+      "epoch": 3.7082601054481548,
+      "grad_norm": 0.03608386218547821,
+      "learning_rate": 5.169302870533099e-06,
+      "loss": 0.0713,
+      "step": 6330
+    },
+    {
+      "epoch": 3.7141183362624486,
+      "grad_norm": 0.004531078971922398,
+      "learning_rate": 5.145869947275923e-06,
+      "loss": 0.0423,
+      "step": 6340
+    },
+    {
+      "epoch": 3.719976567076743,
+      "grad_norm": 0.004161532036960125,
+      "learning_rate": 5.122437024018747e-06,
+      "loss": 0.0784,
+      "step": 6350
+    },
+    {
+      "epoch": 3.7258347978910367,
+      "grad_norm": 0.01157628558576107,
+      "learning_rate": 5.0990041007615706e-06,
+      "loss": 0.1958,
+      "step": 6360
+    },
+    {
+      "epoch": 3.731693028705331,
+      "grad_norm": 0.004814586602151394,
+      "learning_rate": 5.075571177504394e-06,
+      "loss": 0.0728,
+      "step": 6370
+    },
+    {
+      "epoch": 3.737551259519625,
+      "grad_norm": 0.061883583664894104,
+      "learning_rate": 5.052138254247217e-06,
+      "loss": 0.0012,
+      "step": 6380
+    },
+    {
+      "epoch": 3.743409490333919,
+      "grad_norm": 5.269415855407715,
+      "learning_rate": 5.028705330990041e-06,
+      "loss": 0.1278,
+      "step": 6390
+    },
+    {
+      "epoch": 3.7492677211482133,
+      "grad_norm": 0.005129367113113403,
+      "learning_rate": 5.005272407732865e-06,
+      "loss": 0.006,
+      "step": 6400
+    },
+    {
+      "epoch": 3.7551259519625075,
+      "grad_norm": 0.0043735988438129425,
+      "learning_rate": 4.981839484475689e-06,
+      "loss": 0.081,
+      "step": 6410
+    },
+    {
+      "epoch": 3.7609841827768014,
+      "grad_norm": 0.004041170235723257,
+      "learning_rate": 4.958406561218512e-06,
+      "loss": 0.0009,
+      "step": 6420
+    },
+    {
+      "epoch": 3.7668424135910956,
+      "grad_norm": 0.2732388377189636,
+      "learning_rate": 4.934973637961336e-06,
+      "loss": 0.0113,
+      "step": 6430
+    },
+    {
+      "epoch": 3.7727006444053894,
+      "grad_norm": 0.01091376505792141,
+      "learning_rate": 4.91154071470416e-06,
+      "loss": 0.0263,
+      "step": 6440
+    },
+    {
+      "epoch": 3.7785588752196837,
+      "grad_norm": 0.08126252889633179,
+      "learning_rate": 4.888107791446983e-06,
+      "loss": 0.0537,
+      "step": 6450
+    },
+    {
+      "epoch": 3.7844171060339775,
+      "grad_norm": 0.004293652717024088,
+      "learning_rate": 4.8646748681898075e-06,
+      "loss": 0.1306,
+      "step": 6460
+    },
+    {
+      "epoch": 3.790275336848272,
+      "grad_norm": 13.555880546569824,
+      "learning_rate": 4.841241944932631e-06,
+      "loss": 0.0568,
+      "step": 6470
+    },
+    {
+      "epoch": 3.796133567662566,
+      "grad_norm": 0.04533328115940094,
+      "learning_rate": 4.817809021675454e-06,
+      "loss": 0.0361,
+      "step": 6480
+    },
+    {
+      "epoch": 3.80199179847686,
+      "grad_norm": 0.0036789195146411657,
+      "learning_rate": 4.794376098418278e-06,
+      "loss": 0.0885,
+      "step": 6490
+    },
+    {
+      "epoch": 3.807850029291154,
+      "grad_norm": 0.015161854214966297,
+      "learning_rate": 4.770943175161102e-06,
+      "loss": 0.0025,
+      "step": 6500
+    },
+    {
+      "epoch": 3.8137082601054484,
+      "grad_norm": 0.03570953756570816,
+      "learning_rate": 4.7475102519039255e-06,
+      "loss": 0.0014,
+      "step": 6510
+    },
+    {
+      "epoch": 3.819566490919742,
+      "grad_norm": 0.029073569923639297,
+      "learning_rate": 4.724077328646749e-06,
+      "loss": 0.0007,
+      "step": 6520
+    },
+    {
+      "epoch": 3.8254247217340365,
+      "grad_norm": 0.001704677357338369,
+      "learning_rate": 4.700644405389573e-06,
+      "loss": 0.0637,
+      "step": 6530
+    },
+    {
+      "epoch": 3.8312829525483303,
+      "grad_norm": 0.014290403574705124,
+      "learning_rate": 4.677211482132397e-06,
+      "loss": 0.0005,
+      "step": 6540
+    },
+    {
+      "epoch": 3.8371411833626246,
+      "grad_norm": 0.0036969457287341356,
+      "learning_rate": 4.65377855887522e-06,
+      "loss": 0.0065,
+      "step": 6550
+    },
+    {
+      "epoch": 3.8429994141769184,
+      "grad_norm": 5.847663402557373,
+      "learning_rate": 4.6303456356180435e-06,
+      "loss": 0.1708,
+      "step": 6560
+    },
+    {
+      "epoch": 3.8488576449912126,
+      "grad_norm": 0.012637904845178127,
+      "learning_rate": 4.606912712360868e-06,
+      "loss": 0.0073,
+      "step": 6570
+    },
+    {
+      "epoch": 3.854715875805507,
+      "grad_norm": 0.0401185005903244,
+      "learning_rate": 4.583479789103691e-06,
+      "loss": 0.0516,
+      "step": 6580
+    },
+    {
+      "epoch": 3.8605741066198007,
+      "grad_norm": 0.030397135764360428,
+      "learning_rate": 4.560046865846515e-06,
+      "loss": 0.0711,
+      "step": 6590
+    },
+    {
+      "epoch": 3.866432337434095,
+      "grad_norm": 4.24653434753418,
+      "learning_rate": 4.536613942589338e-06,
+      "loss": 0.0038,
+      "step": 6600
+    },
+    {
+      "epoch": 3.8722905682483892,
+      "grad_norm": 0.017508642747998238,
+      "learning_rate": 4.5131810193321615e-06,
+      "loss": 0.0004,
+      "step": 6610
+    },
+    {
+      "epoch": 3.878148799062683,
+      "grad_norm": 7.2771124839782715,
+      "learning_rate": 4.489748096074986e-06,
+      "loss": 0.0742,
+      "step": 6620
+    },
+    {
+      "epoch": 3.884007029876977,
+      "grad_norm": 0.009751928970217705,
+      "learning_rate": 4.46631517281781e-06,
+      "loss": 0.0031,
+      "step": 6630
+    },
+    {
+      "epoch": 3.889865260691271,
+      "grad_norm": 0.006663624197244644,
+      "learning_rate": 4.442882249560633e-06,
+      "loss": 0.0506,
+      "step": 6640
+    },
+    {
+      "epoch": 3.8957234915055654,
+      "grad_norm": 0.09746979176998138,
+      "learning_rate": 4.419449326303456e-06,
+      "loss": 0.0009,
+      "step": 6650
+    },
+    {
+      "epoch": 3.9015817223198592,
+      "grad_norm": 0.021097734570503235,
+      "learning_rate": 4.3960164030462804e-06,
+      "loss": 0.0191,
+      "step": 6660
+    },
+    {
+      "epoch": 3.9074399531341535,
+      "grad_norm": 0.2939240634441376,
+      "learning_rate": 4.3725834797891045e-06,
+      "loss": 0.0042,
+      "step": 6670
+    },
+    {
+      "epoch": 3.9132981839484478,
+      "grad_norm": 11.390765190124512,
+      "learning_rate": 4.349150556531928e-06,
+      "loss": 0.0667,
+      "step": 6680
+    },
+    {
+      "epoch": 3.9191564147627416,
+      "grad_norm": 0.002888856688514352,
+      "learning_rate": 4.325717633274751e-06,
+      "loss": 0.0622,
+      "step": 6690
+    },
+    {
+      "epoch": 3.925014645577036,
+      "grad_norm": 0.0075246552005410194,
+      "learning_rate": 4.302284710017575e-06,
+      "loss": 0.0798,
+      "step": 6700
+    },
+    {
+      "epoch": 3.93087287639133,
+      "grad_norm": 0.03289024531841278,
+      "learning_rate": 4.2788517867603985e-06,
+      "loss": 0.0321,
+      "step": 6710
+    },
+    {
+      "epoch": 3.936731107205624,
+      "grad_norm": 0.013219290412962437,
+      "learning_rate": 4.2554188635032226e-06,
+      "loss": 0.0799,
+      "step": 6720
+    },
+    {
+      "epoch": 3.9425893380199177,
+      "grad_norm": 0.0025315419770777225,
+      "learning_rate": 4.231985940246046e-06,
+      "loss": 0.0069,
+      "step": 6730
+    },
+    {
+      "epoch": 3.948447568834212,
+      "grad_norm": 10.275527954101562,
+      "learning_rate": 4.20855301698887e-06,
+      "loss": 0.1077,
+      "step": 6740
+    },
+    {
+      "epoch": 3.9543057996485063,
+      "grad_norm": 0.09231194853782654,
+      "learning_rate": 4.185120093731693e-06,
+      "loss": 0.0109,
+      "step": 6750
+    },
+    {
+      "epoch": 3.9601640304628,
+      "grad_norm": 0.04757778346538544,
+      "learning_rate": 4.161687170474517e-06,
+      "loss": 0.0562,
+      "step": 6760
+    },
+    {
+      "epoch": 3.9660222612770943,
+      "grad_norm": 0.0062348204664886,
+      "learning_rate": 4.138254247217341e-06,
+      "loss": 0.0006,
+      "step": 6770
+    },
+    {
+      "epoch": 3.9718804920913886,
+      "grad_norm": 0.21855579316616058,
+      "learning_rate": 4.114821323960164e-06,
+      "loss": 0.0363,
+      "step": 6780
+    },
+    {
+      "epoch": 3.9777387229056824,
+      "grad_norm": 1.3651577234268188,
+      "learning_rate": 4.091388400702988e-06,
+      "loss": 0.0709,
+      "step": 6790
+    },
+    {
+      "epoch": 3.9835969537199767,
+      "grad_norm": 0.11818064749240875,
+      "learning_rate": 4.067955477445812e-06,
+      "loss": 0.0571,
+      "step": 6800
+    },
+    {
+      "epoch": 3.9894551845342705,
+      "grad_norm": 0.22804498672485352,
+      "learning_rate": 4.044522554188635e-06,
+      "loss": 0.1319,
+      "step": 6810
+    },
+    {
+      "epoch": 3.9953134153485648,
+      "grad_norm": 0.003853770438581705,
+      "learning_rate": 4.021089630931459e-06,
+      "loss": 0.0703,
+      "step": 6820
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 8535,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 7382675598336000.0,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}