diff --git "a/checkpoint-11000/trainer_state.json" "b/checkpoint-11000/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-11000/trainer_state.json"
@@ -0,0 +1,7734 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.4079098937667989,
+  "eval_steps": 500,
+  "global_step": 11000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.01278772378516624,
+      "grad_norm": 83.28244018554688,
+      "learning_rate": 1.3499999999999998e-05,
+      "loss": 16.4089,
+      "step": 10
+    },
+    {
+      "epoch": 0.02557544757033248,
+      "grad_norm": 23.9244327545166,
+      "learning_rate": 2.8499999999999998e-05,
+      "loss": 12.7299,
+      "step": 20
+    },
+    {
+      "epoch": 0.03836317135549872,
+      "grad_norm": 9.106985092163086,
+      "learning_rate": 4.3499999999999993e-05,
+      "loss": 12.0294,
+      "step": 30
+    },
+    {
+      "epoch": 0.05115089514066496,
+      "grad_norm": 26.396886825561523,
+      "learning_rate": 5.85e-05,
+      "loss": 10.0869,
+      "step": 40
+    },
+    {
+      "epoch": 0.0639386189258312,
+      "grad_norm": 1.9576493501663208,
+      "learning_rate": 7.35e-05,
+      "loss": 8.6136,
+      "step": 50
+    },
+    {
+      "epoch": 0.07672634271099744,
+      "grad_norm": 1.866382360458374,
+      "learning_rate": 8.849999999999998e-05,
+      "loss": 8.4649,
+      "step": 60
+    },
+    {
+      "epoch": 0.08951406649616368,
+      "grad_norm": 2.786794662475586,
+      "learning_rate": 0.00010349999999999998,
+      "loss": 8.4592,
+      "step": 70
+    },
+    {
+      "epoch": 0.10230179028132992,
+      "grad_norm": 3.3617353439331055,
+      "learning_rate": 0.0001185,
+      "loss": 8.4249,
+      "step": 80
+    },
+    {
+      "epoch": 0.11508951406649616,
+      "grad_norm": 6.848003387451172,
+      "learning_rate": 0.0001335,
+      "loss": 8.2559,
+      "step": 90
+    },
+    {
+      "epoch": 0.1278772378516624,
+      "grad_norm": 7.901203632354736,
+      "learning_rate": 0.00014849999999999998,
+      "loss": 8.1613,
+      "step": 100
+    },
+    {
+      "epoch": 0.14066496163682865,
+      "grad_norm": 18.165496826171875,
+      "learning_rate": 0.0001635,
+      "loss": 7.7662,
+      "step": 110
+    },
+    {
+      "epoch": 0.1534526854219949,
+      "grad_norm": 8.8899507522583,
+      "learning_rate": 0.00017849999999999997,
+      "loss": 7.5454,
+      "step": 120
+    },
+    {
+      "epoch": 0.16624040920716113,
+      "grad_norm": 23.676000595092773,
+      "learning_rate": 0.0001935,
+      "loss": 7.7264,
+      "step": 130
+    },
+    {
+      "epoch": 0.17902813299232737,
+      "grad_norm": 20.63345718383789,
+      "learning_rate": 0.00020849999999999997,
+      "loss": 7.8592,
+      "step": 140
+    },
+    {
+      "epoch": 0.1918158567774936,
+      "grad_norm": 15.780885696411133,
+      "learning_rate": 0.00022349999999999998,
+      "loss": 7.5941,
+      "step": 150
+    },
+    {
+      "epoch": 0.20460358056265984,
+      "grad_norm": 60.87010955810547,
+      "learning_rate": 0.0002385,
+      "loss": 7.4103,
+      "step": 160
+    },
+    {
+      "epoch": 0.21739130434782608,
+      "grad_norm": 1284.3363037109375,
+      "learning_rate": 0.0002535,
+      "loss": 7.6876,
+      "step": 170
+    },
+    {
+      "epoch": 0.23017902813299232,
+      "grad_norm": 12.15369701385498,
+      "learning_rate": 0.00026849999999999997,
+      "loss": 7.3406,
+      "step": 180
+    },
+    {
+      "epoch": 0.24296675191815856,
+      "grad_norm": 155.99559020996094,
+      "learning_rate": 0.00028349999999999995,
+      "loss": 6.9872,
+      "step": 190
+    },
+    {
+      "epoch": 0.2557544757033248,
+      "grad_norm": 30.107227325439453,
+      "learning_rate": 0.0002985,
+      "loss": 7.1682,
+      "step": 200
+    },
+    {
+      "epoch": 0.26854219948849106,
+      "grad_norm": 4.686079502105713,
+      "learning_rate": 0.0002999997484922834,
+      "loss": 7.0612,
+      "step": 210
+    },
+    {
+      "epoch": 0.2813299232736573,
+      "grad_norm": 5.214747905731201,
+      "learning_rate": 0.0002999988790839756,
+      "loss": 6.7361,
+      "step": 220
+    },
+    {
+      "epoch": 0.29411764705882354,
+      "grad_norm": 2.2351861000061035,
+      "learning_rate": 0.00029999738867364154,
+      "loss": 6.625,
+      "step": 230
+    },
+    {
+      "epoch": 0.3069053708439898,
+      "grad_norm": 1.2042266130447388,
+      "learning_rate": 0.0002999952772674517,
+      "loss": 6.5884,
+      "step": 240
+    },
+    {
+      "epoch": 0.319693094629156,
+      "grad_norm": 15.000553131103516,
+      "learning_rate": 0.00029999254487414717,
+      "loss": 6.6537,
+      "step": 250
+    },
+    {
+      "epoch": 0.33248081841432225,
+      "grad_norm": 0.5562590956687927,
+      "learning_rate": 0.0002999891915050404,
+      "loss": 6.5825,
+      "step": 260
+    },
+    {
+      "epoch": 0.3452685421994885,
+      "grad_norm": 0.9475784301757812,
+      "learning_rate": 0.00029998521717401446,
+      "loss": 6.5398,
+      "step": 270
+    },
+    {
+      "epoch": 0.35805626598465473,
+      "grad_norm": 3.0192008018493652,
+      "learning_rate": 0.0002999806218975231,
+      "loss": 6.537,
+      "step": 280
+    },
+    {
+      "epoch": 0.37084398976982097,
+      "grad_norm": 0.6399412751197815,
+      "learning_rate": 0.0002999754056945911,
+      "loss": 6.5786,
+      "step": 290
+    },
+    {
+      "epoch": 0.3836317135549872,
+      "grad_norm": 48.2303466796875,
+      "learning_rate": 0.00029996956858681373,
+      "loss": 6.6661,
+      "step": 300
+    },
+    {
+      "epoch": 0.39641943734015345,
+      "grad_norm": 6.753386497497559,
+      "learning_rate": 0.00029996311059835696,
+      "loss": 6.5767,
+      "step": 310
+    },
+    {
+      "epoch": 0.4092071611253197,
+      "grad_norm": 0.7255383729934692,
+      "learning_rate": 0.000299956031755957,
+      "loss": 6.5298,
+      "step": 320
+    },
+    {
+      "epoch": 0.4219948849104859,
+      "grad_norm": 19.085235595703125,
+      "learning_rate": 0.0002999483320889207,
+      "loss": 6.6483,
+      "step": 330
+    },
+    {
+      "epoch": 0.43478260869565216,
+      "grad_norm": 7.604534149169922,
+      "learning_rate": 0.0002999400116291249,
+      "loss": 6.5788,
+      "step": 340
+    },
+    {
+      "epoch": 0.4475703324808184,
+      "grad_norm": 8.935670852661133,
+      "learning_rate": 0.0002999310704110168,
+      "loss": 6.5955,
+      "step": 350
+    },
+    {
+      "epoch": 0.46035805626598464,
+      "grad_norm": 5.167853832244873,
+      "learning_rate": 0.0002999215084716133,
+      "loss": 6.5713,
+      "step": 360
+    },
+    {
+      "epoch": 0.4731457800511509,
+      "grad_norm": 7.617280006408691,
+      "learning_rate": 0.00029991132585050146,
+      "loss": 6.5118,
+      "step": 370
+    },
+    {
+      "epoch": 0.4859335038363171,
+      "grad_norm": 1.1345467567443848,
+      "learning_rate": 0.00029990052258983764,
+      "loss": 6.532,
+      "step": 380
+    },
+    {
+      "epoch": 0.49872122762148335,
+      "grad_norm": 62.21474838256836,
+      "learning_rate": 0.0002998890987343478,
+      "loss": 6.8089,
+      "step": 390
+    },
+    {
+      "epoch": 0.5115089514066496,
+      "grad_norm": 18.671003341674805,
+      "learning_rate": 0.0002998770543313273,
+      "loss": 7.5951,
+      "step": 400
+    },
+    {
+      "epoch": 0.5242966751918159,
+      "grad_norm": 51.78537368774414,
+      "learning_rate": 0.0002998643894306404,
+      "loss": 7.2577,
+      "step": 410
+    },
+    {
+      "epoch": 0.5370843989769821,
+      "grad_norm": 3.8242099285125732,
+      "learning_rate": 0.00029985110408472046,
+      "loss": 6.9031,
+      "step": 420
+    },
+    {
+      "epoch": 0.5498721227621484,
+      "grad_norm": 0.8091524839401245,
+      "learning_rate": 0.0002998371983485693,
+      "loss": 6.513,
+      "step": 430
+    },
+    {
+      "epoch": 0.5626598465473146,
+      "grad_norm": 1.8598475456237793,
+      "learning_rate": 0.00029982267227975735,
+      "loss": 6.5526,
+      "step": 440
+    },
+    {
+      "epoch": 0.5754475703324808,
+      "grad_norm": 0.7141775488853455,
+      "learning_rate": 0.0002998075259384232,
+      "loss": 6.5129,
+      "step": 450
+    },
+    {
+      "epoch": 0.5882352941176471,
+      "grad_norm": 0.31634974479675293,
+      "learning_rate": 0.0002997917593872733,
+      "loss": 6.4868,
+      "step": 460
+    },
+    {
+      "epoch": 0.6010230179028133,
+      "grad_norm": 0.5466232299804688,
+      "learning_rate": 0.000299775372691582,
+      "loss": 6.4654,
+      "step": 470
+    },
+    {
+      "epoch": 0.6138107416879796,
+      "grad_norm": 0.4127891957759857,
+      "learning_rate": 0.0002997583659191908,
+      "loss": 6.454,
+      "step": 480
+    },
+    {
+      "epoch": 0.6265984654731458,
+      "grad_norm": 3.9770538806915283,
+      "learning_rate": 0.0002997407391405085,
+      "loss": 6.4464,
+      "step": 490
+    },
+    {
+      "epoch": 0.639386189258312,
+      "grad_norm": 0.9428657293319702,
+      "learning_rate": 0.00029972249242851093,
+      "loss": 6.3952,
+      "step": 500
+    },
+    {
+      "epoch": 0.6521739130434783,
+      "grad_norm": 0.7024242281913757,
+      "learning_rate": 0.0002997036258587401,
+      "loss": 6.3843,
+      "step": 510
+    },
+    {
+      "epoch": 0.6649616368286445,
+      "grad_norm": 0.7226349115371704,
+      "learning_rate": 0.0002996841395093045,
+      "loss": 6.3729,
+      "step": 520
+    },
+    {
+      "epoch": 0.6777493606138107,
+      "grad_norm": 1.3316643238067627,
+      "learning_rate": 0.0002996640334608785,
+      "loss": 6.3582,
+      "step": 530
+    },
+    {
+      "epoch": 0.690537084398977,
+      "grad_norm": 1.465959906578064,
+      "learning_rate": 0.000299643307796702,
+      "loss": 6.4178,
+      "step": 540
+    },
+    {
+      "epoch": 0.7033248081841432,
+      "grad_norm": 1.373779058456421,
+      "learning_rate": 0.00029962196260258024,
+      "loss": 6.3698,
+      "step": 550
+    },
+    {
+      "epoch": 0.7161125319693095,
+      "grad_norm": 1.058214545249939,
+      "learning_rate": 0.0002995999979668833,
+      "loss": 6.3461,
+      "step": 560
+    },
+    {
+      "epoch": 0.7289002557544757,
+      "grad_norm": 1.4381204843521118,
+      "learning_rate": 0.00029957741398054557,
+      "loss": 6.3417,
+      "step": 570
+    },
+    {
+      "epoch": 0.7416879795396419,
+      "grad_norm": 7.174375057220459,
+      "learning_rate": 0.0002995542107370659,
+      "loss": 6.4027,
+      "step": 580
+    },
+    {
+      "epoch": 0.7544757033248082,
+      "grad_norm": 1.1935175657272339,
+      "learning_rate": 0.0002995303883325067,
+      "loss": 6.4132,
+      "step": 590
+    },
+    {
+      "epoch": 0.7672634271099744,
+      "grad_norm": 0.6950662732124329,
+      "learning_rate": 0.0002995059468654939,
+      "loss": 6.3146,
+      "step": 600
+    },
+    {
+      "epoch": 0.7800511508951407,
+      "grad_norm": 4.71625280380249,
+      "learning_rate": 0.0002994808864372161,
+      "loss": 6.3259,
+      "step": 610
+    },
+    {
+      "epoch": 0.7928388746803069,
+      "grad_norm": 0.7315058708190918,
+      "learning_rate": 0.00029945520715142455,
+      "loss": 6.5004,
+      "step": 620
+    },
+    {
+      "epoch": 0.8056265984654731,
+      "grad_norm": 1.352614164352417,
+      "learning_rate": 0.00029942890911443284,
+      "loss": 6.3314,
+      "step": 630
+    },
+    {
+      "epoch": 0.8184143222506394,
+      "grad_norm": 1.55237877368927,
+      "learning_rate": 0.0002994019924351159,
+      "loss": 6.291,
+      "step": 640
+    },
+    {
+      "epoch": 0.8312020460358056,
+      "grad_norm": 1.7966560125350952,
+      "learning_rate": 0.0002993744572249099,
+      "loss": 6.3086,
+      "step": 650
+    },
+    {
+      "epoch": 0.8439897698209718,
+      "grad_norm": 0.62131667137146,
+      "learning_rate": 0.00029934630359781196,
+      "loss": 6.307,
+      "step": 660
+    },
+    {
+      "epoch": 0.8567774936061381,
+      "grad_norm": 1.4420467615127563,
+      "learning_rate": 0.00029931753167037933,
+      "loss": 6.295,
+      "step": 670
+    },
+    {
+      "epoch": 0.8695652173913043,
+      "grad_norm": 0.6070270538330078,
+      "learning_rate": 0.000299288141561729,
+      "loss": 6.2912,
+      "step": 680
+    },
+    {
+      "epoch": 0.8823529411764706,
+      "grad_norm": 0.7900067567825317,
+      "learning_rate": 0.00029925813339353737,
+      "loss": 6.2815,
+      "step": 690
+    },
+    {
+      "epoch": 0.8951406649616368,
+      "grad_norm": 0.3804131746292114,
+      "learning_rate": 0.0002992275072900396,
+      "loss": 6.2626,
+      "step": 700
+    },
+    {
+      "epoch": 0.907928388746803,
+      "grad_norm": 0.6143403053283691,
+      "learning_rate": 0.00029919626337802913,
+      "loss": 6.2581,
+      "step": 710
+    },
+    {
+      "epoch": 0.9207161125319693,
+      "grad_norm": 0.6377658247947693,
+      "learning_rate": 0.00029916440178685713,
+      "loss": 6.2671,
+      "step": 720
+    },
+    {
+      "epoch": 0.9335038363171355,
+      "grad_norm": 0.5696841478347778,
+      "learning_rate": 0.00029913192264843195,
+      "loss": 6.2527,
+      "step": 730
+    },
+    {
+      "epoch": 0.9462915601023018,
+      "grad_norm": 0.39249590039253235,
+      "learning_rate": 0.00029909882609721876,
+      "loss": 6.2406,
+      "step": 740
+    },
+    {
+      "epoch": 0.959079283887468,
+      "grad_norm": 0.6684032082557678,
+      "learning_rate": 0.0002990651122702387,
+      "loss": 6.2444,
+      "step": 750
+    },
+    {
+      "epoch": 0.9718670076726342,
+      "grad_norm": 0.7203615307807922,
+      "learning_rate": 0.00029903078130706846,
+      "loss": 6.2242,
+      "step": 760
+    },
+    {
+      "epoch": 0.9846547314578005,
+      "grad_norm": 1.0230414867401123,
+      "learning_rate": 0.00029899583334983985,
+      "loss": 6.1915,
+      "step": 770
+    },
+    {
+      "epoch": 0.9974424552429667,
+      "grad_norm": 1.015184760093689,
+      "learning_rate": 0.00029896026854323894,
+      "loss": 6.2248,
+      "step": 780
+    },
+    {
+      "epoch": 1.010230179028133,
+      "grad_norm": 0.6776460409164429,
+      "learning_rate": 0.0002989240870345056,
+      "loss": 6.2067,
+      "step": 790
+    },
+    {
+      "epoch": 1.0230179028132993,
+      "grad_norm": 0.4673589766025543,
+      "learning_rate": 0.0002988872889734328,
+      "loss": 6.1699,
+      "step": 800
+    },
+    {
+      "epoch": 1.0358056265984654,
+      "grad_norm": 0.4850061535835266,
+      "learning_rate": 0.0002988498745123663,
+      "loss": 6.1736,
+      "step": 810
+    },
+    {
+      "epoch": 1.0485933503836318,
+      "grad_norm": 0.6966080665588379,
+      "learning_rate": 0.0002988118438062036,
+      "loss": 6.1996,
+      "step": 820
+    },
+    {
+      "epoch": 1.061381074168798,
+      "grad_norm": 0.7484626770019531,
+      "learning_rate": 0.0002987731970123934,
+      "loss": 6.1527,
+      "step": 830
+    },
+    {
+      "epoch": 1.0741687979539642,
+      "grad_norm": 2.176851511001587,
+      "learning_rate": 0.0002987339342909352,
+      "loss": 6.1356,
+      "step": 840
+    },
+    {
+      "epoch": 1.0869565217391304,
+      "grad_norm": 0.634508490562439,
+      "learning_rate": 0.0002986940558043785,
+      "loss": 6.142,
+      "step": 850
+    },
+    {
+      "epoch": 1.0997442455242967,
+      "grad_norm": 0.6721930503845215,
+      "learning_rate": 0.00029865356171782196,
+      "loss": 6.1289,
+      "step": 860
+    },
+    {
+      "epoch": 1.1125319693094629,
+      "grad_norm": 0.5172439813613892,
+      "learning_rate": 0.0002986124521989128,
+      "loss": 6.1097,
+      "step": 870
+    },
+    {
+      "epoch": 1.1253196930946292,
+      "grad_norm": 0.609614372253418,
+      "learning_rate": 0.00029857072741784634,
+      "loss": 6.1196,
+      "step": 880
+    },
+    {
+      "epoch": 1.1381074168797953,
+      "grad_norm": 0.7491389513015747,
+      "learning_rate": 0.0002985283875473651,
+      "loss": 6.0896,
+      "step": 890
+    },
+    {
+      "epoch": 1.1508951406649617,
+      "grad_norm": 0.5810283422470093,
+      "learning_rate": 0.0002984854327627578,
+      "loss": 6.097,
+      "step": 900
+    },
+    {
+      "epoch": 1.1636828644501278,
+      "grad_norm": 0.691870391368866,
+      "learning_rate": 0.00029844186324185943,
+      "loss": 6.0979,
+      "step": 910
+    },
+    {
+      "epoch": 1.1764705882352942,
+      "grad_norm": 0.5209296345710754,
+      "learning_rate": 0.00029839767916504955,
+      "loss": 6.068,
+      "step": 920
+    },
+    {
+      "epoch": 1.1892583120204603,
+      "grad_norm": 1.1897755861282349,
+      "learning_rate": 0.0002983528807152523,
+      "loss": 6.0626,
+      "step": 930
+    },
+    {
+      "epoch": 1.2020460358056266,
+      "grad_norm": 0.6294477581977844,
+      "learning_rate": 0.0002983074680779352,
+      "loss": 6.0441,
+      "step": 940
+    },
+    {
+      "epoch": 1.2148337595907928,
+      "grad_norm": 0.7823428511619568,
+      "learning_rate": 0.0002982614414411086,
+      "loss": 6.0634,
+      "step": 950
+    },
+    {
+      "epoch": 1.227621483375959,
+      "grad_norm": 0.641527533531189,
+      "learning_rate": 0.00029821480099532485,
+      "loss": 6.0567,
+      "step": 960
+    },
+    {
+      "epoch": 1.2404092071611252,
+      "grad_norm": 1.0938187837600708,
+      "learning_rate": 0.00029816754693367745,
+      "loss": 6.0302,
+      "step": 970
+    },
+    {
+      "epoch": 1.2531969309462916,
+      "grad_norm": 2.3455729484558105,
+      "learning_rate": 0.0002981196794518003,
+      "loss": 6.0441,
+      "step": 980
+    },
+    {
+      "epoch": 1.265984654731458,
+      "grad_norm": 0.5875425934791565,
+      "learning_rate": 0.0002980711987478669,
+      "loss": 6.0277,
+      "step": 990
+    },
+    {
+      "epoch": 1.278772378516624,
+      "grad_norm": 1.0881823301315308,
+      "learning_rate": 0.0002980221050225896,
+      "loss": 6.0174,
+      "step": 1000
+    },
+    {
+      "epoch": 1.2915601023017902,
+      "grad_norm": 0.8842185735702515,
+      "learning_rate": 0.0002979723984792185,
+      "loss": 5.9959,
+      "step": 1010
+    },
+    {
+      "epoch": 1.3043478260869565,
+      "grad_norm": 0.7018517255783081,
+      "learning_rate": 0.0002979220793235409,
+      "loss": 6.0133,
+      "step": 1020
+    },
+    {
+      "epoch": 1.317135549872123,
+      "grad_norm": 1.0132428407669067,
+      "learning_rate": 0.0002978711477638804,
+      "loss": 5.9918,
+      "step": 1030
+    },
+    {
+      "epoch": 1.329923273657289,
+      "grad_norm": 0.8519907593727112,
+      "learning_rate": 0.00029781960401109597,
+      "loss": 6.0053,
+      "step": 1040
+    },
+    {
+      "epoch": 1.3427109974424551,
+      "grad_norm": 1.0255190134048462,
+      "learning_rate": 0.00029776744827858087,
+      "loss": 5.999,
+      "step": 1050
+    },
+    {
+      "epoch": 1.3554987212276215,
+      "grad_norm": 0.6434299945831299,
+      "learning_rate": 0.0002977146807822623,
+      "loss": 5.9846,
+      "step": 1060
+    },
+    {
+      "epoch": 1.3682864450127878,
+      "grad_norm": 0.6655330657958984,
+      "learning_rate": 0.0002976613017406,
+      "loss": 5.9611,
+      "step": 1070
+    },
+    {
+      "epoch": 1.381074168797954,
+      "grad_norm": 0.7621287703514099,
+      "learning_rate": 0.00029760731137458554,
+      "loss": 5.9486,
+      "step": 1080
+    },
+    {
+      "epoch": 1.39386189258312,
+      "grad_norm": 0.8872424960136414,
+      "learning_rate": 0.00029755270990774145,
+      "loss": 5.9409,
+      "step": 1090
+    },
+    {
+      "epoch": 1.4066496163682864,
+      "grad_norm": 0.978261411190033,
+      "learning_rate": 0.0002974974975661203,
+      "loss": 5.9434,
+      "step": 1100
+    },
+    {
+      "epoch": 1.4194373401534528,
+      "grad_norm": 0.9707337617874146,
+      "learning_rate": 0.0002974416745783035,
+      "loss": 5.9248,
+      "step": 1110
+    },
+    {
+      "epoch": 1.432225063938619,
+      "grad_norm": 1.0900273323059082,
+      "learning_rate": 0.0002973852411754008,
+      "loss": 5.9105,
+      "step": 1120
+    },
+    {
+      "epoch": 1.445012787723785,
+      "grad_norm": 3.00311017036438,
+      "learning_rate": 0.00029732819759104896,
+      "loss": 5.922,
+      "step": 1130
+    },
+    {
+      "epoch": 1.4578005115089514,
+      "grad_norm": 0.9423596858978271,
+      "learning_rate": 0.00029727054406141097,
+      "loss": 5.9285,
+      "step": 1140
+    },
+    {
+      "epoch": 1.4705882352941178,
+      "grad_norm": 0.7384818196296692,
+      "learning_rate": 0.000297212280825175,
+      "loss": 5.9222,
+      "step": 1150
+    },
+    {
+      "epoch": 1.4833759590792839,
+      "grad_norm": 1.1171971559524536,
+      "learning_rate": 0.0002971534081235535,
+      "loss": 5.9156,
+      "step": 1160
+    },
+    {
+      "epoch": 1.49616368286445,
+      "grad_norm": 0.7531259059906006,
+      "learning_rate": 0.00029709392620028195,
+      "loss": 5.9004,
+      "step": 1170
+    },
+    {
+      "epoch": 1.5089514066496164,
+      "grad_norm": 0.9441642761230469,
+      "learning_rate": 0.0002970338353016182,
+      "loss": 5.9076,
+      "step": 1180
+    },
+    {
+      "epoch": 1.5217391304347827,
+      "grad_norm": 1.2140872478485107,
+      "learning_rate": 0.00029697313567634133,
+      "loss": 5.8894,
+      "step": 1190
+    },
+    {
+      "epoch": 1.5345268542199488,
+      "grad_norm": 1.1109689474105835,
+      "learning_rate": 0.0002969118275757504,
+      "loss": 5.8663,
+      "step": 1200
+    },
+    {
+      "epoch": 1.547314578005115,
+      "grad_norm": 0.8246760964393616,
+      "learning_rate": 0.00029684991125366364,
+      "loss": 5.8749,
+      "step": 1210
+    },
+    {
+      "epoch": 1.5601023017902813,
+      "grad_norm": 0.744144856929779,
+      "learning_rate": 0.0002967873869664175,
+      "loss": 5.8384,
+      "step": 1220
+    },
+    {
+      "epoch": 1.5728900255754477,
+      "grad_norm": 1.149007797241211,
+      "learning_rate": 0.00029672425497286514,
+      "loss": 5.8582,
+      "step": 1230
+    },
+    {
+      "epoch": 1.5856777493606138,
+      "grad_norm": 0.7142267823219299,
+      "learning_rate": 0.00029666051553437595,
+      "loss": 5.8428,
+      "step": 1240
+    },
+    {
+      "epoch": 1.59846547314578,
+      "grad_norm": 1.0163673162460327,
+      "learning_rate": 0.000296596168914834,
+      "loss": 5.8399,
+      "step": 1250
+    },
+    {
+      "epoch": 1.6112531969309463,
+      "grad_norm": 1.5262479782104492,
+      "learning_rate": 0.0002965312153806371,
+      "loss": 5.8286,
+      "step": 1260
+    },
+    {
+      "epoch": 1.6240409207161126,
+      "grad_norm": 1.150258183479309,
+      "learning_rate": 0.0002964656552006959,
+      "loss": 5.8294,
+      "step": 1270
+    },
+    {
+      "epoch": 1.6368286445012787,
+      "grad_norm": 3.7115478515625,
+      "learning_rate": 0.0002963994886464323,
+      "loss": 5.8069,
+      "step": 1280
+    },
+    {
+      "epoch": 1.6496163682864449,
+      "grad_norm": 0.9176708459854126,
+      "learning_rate": 0.00029633271599177895,
+      "loss": 5.8503,
+      "step": 1290
+    },
+    {
+      "epoch": 1.6624040920716112,
+      "grad_norm": 0.7877741456031799,
+      "learning_rate": 0.00029626533751317744,
+      "loss": 5.8067,
+      "step": 1300
+    },
+    {
+      "epoch": 1.6751918158567776,
+      "grad_norm": 0.8420455455780029,
+      "learning_rate": 0.0002961973534895778,
+      "loss": 5.8071,
+      "step": 1310
+    },
+    {
+      "epoch": 1.6879795396419437,
+      "grad_norm": 1.0373153686523438,
+      "learning_rate": 0.0002961287642024368,
+      "loss": 5.8232,
+      "step": 1320
+    },
+    {
+      "epoch": 1.7007672634271098,
+      "grad_norm": 0.7778970003128052,
+      "learning_rate": 0.0002960595699357172,
+      "loss": 5.8029,
+      "step": 1330
+    },
+    {
+      "epoch": 1.7135549872122762,
+      "grad_norm": 0.77557772397995,
+      "learning_rate": 0.00029598977097588635,
+      "loss": 5.8092,
+      "step": 1340
+    },
+    {
+      "epoch": 1.7263427109974425,
+      "grad_norm": 1.099355697631836,
+      "learning_rate": 0.000295919367611915,
+      "loss": 5.8076,
+      "step": 1350
+    },
+    {
+      "epoch": 1.7391304347826086,
+      "grad_norm": 1.891553521156311,
+      "learning_rate": 0.00029584836013527623,
+      "loss": 5.7752,
+      "step": 1360
+    },
+    {
+      "epoch": 1.7519181585677748,
+      "grad_norm": 1.6321626901626587,
+      "learning_rate": 0.00029577674883994414,
+      "loss": 5.7776,
+      "step": 1370
+    },
+    {
+      "epoch": 1.7647058823529411,
+      "grad_norm": 1.969916582107544,
+      "learning_rate": 0.0002957045340223927,
+      "loss": 5.7882,
+      "step": 1380
+    },
+    {
+      "epoch": 1.7774936061381075,
+      "grad_norm": 0.7964078187942505,
+      "learning_rate": 0.0002956317159815944,
+      "loss": 5.7876,
+      "step": 1390
+    },
+    {
+      "epoch": 1.7902813299232738,
+      "grad_norm": 1.2642195224761963,
+      "learning_rate": 0.00029555829501901927,
+      "loss": 5.7629,
+      "step": 1400
+    },
+    {
+      "epoch": 1.80306905370844,
+      "grad_norm": 0.8739522099494934,
+      "learning_rate": 0.0002954842714386333,
+      "loss": 5.7302,
+      "step": 1410
+    },
+    {
+      "epoch": 1.815856777493606,
+      "grad_norm": 1.0974680185317993,
+      "learning_rate": 0.00029540964554689735,
+      "loss": 5.7658,
+      "step": 1420
+    },
+    {
+      "epoch": 1.8286445012787724,
+      "grad_norm": 1.1861488819122314,
+      "learning_rate": 0.000295334417652766,
+      "loss": 5.7617,
+      "step": 1430
+    },
+    {
+      "epoch": 1.8414322250639388,
+      "grad_norm": 1.473907470703125,
+      "learning_rate": 0.0002952585880676861,
+      "loss": 5.7425,
+      "step": 1440
+    },
+    {
+      "epoch": 1.854219948849105,
+      "grad_norm": 1.195039987564087,
+      "learning_rate": 0.0002951821571055954,
+      "loss": 5.7414,
+      "step": 1450
+    },
+    {
+      "epoch": 1.867007672634271,
+      "grad_norm": 1.2760419845581055,
+      "learning_rate": 0.00029510512508292155,
+      "loss": 5.7445,
+      "step": 1460
+    },
+    {
+      "epoch": 1.8797953964194374,
+      "grad_norm": 2.53118896484375,
+      "learning_rate": 0.0002950274923185806,
+      "loss": 5.7394,
+      "step": 1470
+    },
+    {
+      "epoch": 1.8925831202046037,
+      "grad_norm": 1.0948902368545532,
+      "learning_rate": 0.00029494925913397553,
+      "loss": 5.7318,
+      "step": 1480
+    },
+    {
+      "epoch": 1.9053708439897699,
+      "grad_norm": 1.546405553817749,
+      "learning_rate": 0.00029487042585299526,
+      "loss": 5.7352,
+      "step": 1490
+    },
+    {
+      "epoch": 1.918158567774936,
+      "grad_norm": 70.01959228515625,
+      "learning_rate": 0.0002947909928020131,
+      "loss": 5.9204,
+      "step": 1500
+    },
+    {
+      "epoch": 1.9309462915601023,
+      "grad_norm": 1859.7359619140625,
+      "learning_rate": 0.0002947109603098853,
+      "loss": 5.9433,
+      "step": 1510
+    },
+    {
+      "epoch": 1.9437340153452687,
+      "grad_norm": 75.3654556274414,
+      "learning_rate": 0.0002946303287079501,
+      "loss": 7.0599,
+      "step": 1520
+    },
+    {
+      "epoch": 1.9565217391304348,
+      "grad_norm": 1592.2181396484375,
+      "learning_rate": 0.00029454909833002583,
+      "loss": 8.1567,
+      "step": 1530
+    },
+    {
+      "epoch": 1.969309462915601,
+      "grad_norm": 2557.483154296875,
+      "learning_rate": 0.0002944672695124098,
+      "loss": 9.1084,
+      "step": 1540
+    },
+    {
+      "epoch": 1.9820971867007673,
+      "grad_norm": 564582.5625,
+      "learning_rate": 0.00029438484259387707,
+      "loss": 11.1384,
+      "step": 1550
+    },
+    {
+      "epoch": 1.9948849104859336,
+      "grad_norm": 83855.9765625,
+      "learning_rate": 0.0002943018179156787,
+      "loss": 15.2769,
+      "step": 1560
+    },
+    {
+      "epoch": 2.0076726342710995,
+      "grad_norm": 17.882341384887695,
+      "learning_rate": 0.0002942181958215405,
+      "loss": 10.5658,
+      "step": 1570
+    },
+    {
+      "epoch": 2.020460358056266,
+      "grad_norm": 7614.091796875,
+      "learning_rate": 0.0002941339766576618,
+      "loss": 8.1135,
+      "step": 1580
+    },
+    {
+      "epoch": 2.0332480818414322,
+      "grad_norm": 479.05926513671875,
+      "learning_rate": 0.0002940491607727135,
+      "loss": 18.0011,
+      "step": 1590
+    },
+    {
+      "epoch": 2.0460358056265986,
+      "grad_norm": 233.18682861328125,
+      "learning_rate": 0.00029396374851783723,
+      "loss": 8.1656,
+      "step": 1600
+    },
+    {
+      "epoch": 2.0588235294117645,
+      "grad_norm": 7.251436710357666,
+      "learning_rate": 0.0002938777402466435,
+      "loss": 9.6865,
+      "step": 1610
+    },
+    {
+      "epoch": 2.071611253196931,
+      "grad_norm": 13.384018898010254,
+      "learning_rate": 0.0002937911363152104,
+      "loss": 7.5508,
+      "step": 1620
+    },
+    {
+      "epoch": 2.084398976982097,
+      "grad_norm": 64.90263366699219,
+      "learning_rate": 0.00029370393708208205,
+      "loss": 6.7978,
+      "step": 1630
+    },
+    {
+      "epoch": 2.0971867007672635,
+      "grad_norm": 10.841496467590332,
+      "learning_rate": 0.00029361614290826705,
+      "loss": 6.589,
+      "step": 1640
+    },
+    {
+      "epoch": 2.10997442455243,
+      "grad_norm": 31.3606014251709,
+      "learning_rate": 0.00029352775415723733,
+      "loss": 6.5118,
+      "step": 1650
+    },
+    {
+      "epoch": 2.122762148337596,
+      "grad_norm": 35.56670379638672,
+      "learning_rate": 0.0002934387711949262,
+      "loss": 6.3938,
+      "step": 1660
+    },
+    {
+      "epoch": 2.135549872122762,
+      "grad_norm": 9.298479080200195,
+      "learning_rate": 0.00029334919438972707,
+      "loss": 6.3621,
+      "step": 1670
+    },
+    {
+      "epoch": 2.1483375959079285,
+      "grad_norm": 31.193031311035156,
+      "learning_rate": 0.0002932590241124918,
+      "loss": 6.3468,
+      "step": 1680
+    },
+    {
+      "epoch": 2.1611253196930944,
+      "grad_norm": 63.973716735839844,
+      "learning_rate": 0.0002931682607365296,
+      "loss": 6.358,
+      "step": 1690
+    },
+    {
+      "epoch": 2.1739130434782608,
+      "grad_norm": 14.983797073364258,
+      "learning_rate": 0.0002930769046376046,
+      "loss": 6.3489,
+      "step": 1700
+    },
+    {
+      "epoch": 2.186700767263427,
+      "grad_norm": 2612.127685546875,
+      "learning_rate": 0.00029298495619393527,
+      "loss": 6.3614,
+      "step": 1710
+    },
+    {
+      "epoch": 2.1994884910485935,
+      "grad_norm": 1995.3865966796875,
+      "learning_rate": 0.0002928924157861922,
+      "loss": 6.3641,
+      "step": 1720
+    },
+    {
+      "epoch": 2.21227621483376,
+      "grad_norm": 9.419896125793457,
+      "learning_rate": 0.0002927992837974968,
+      "loss": 6.3839,
+      "step": 1730
+    },
+    {
+      "epoch": 2.2250639386189257,
+      "grad_norm": 12.694928169250488,
+      "learning_rate": 0.00029270556061341966,
+      "loss": 6.395,
+      "step": 1740
+    },
+    {
+      "epoch": 2.237851662404092,
+      "grad_norm": 1460.1500244140625,
+      "learning_rate": 0.0002926112466219789,
+      "loss": 6.6829,
+      "step": 1750
+    },
+    {
+      "epoch": 2.2506393861892584,
+      "grad_norm": 11581.7744140625,
+      "learning_rate": 0.0002925163422136386,
+      "loss": 7.3051,
+      "step": 1760
+    },
+    {
+      "epoch": 2.2634271099744243,
+      "grad_norm": 1636.6549072265625,
+      "learning_rate": 0.0002924208477813074,
+      "loss": 7.1446,
+      "step": 1770
+    },
+    {
+      "epoch": 2.2762148337595907,
+      "grad_norm": 62.05595016479492,
+      "learning_rate": 0.0002923247637203362,
+      "loss": 6.7255,
+      "step": 1780
+    },
+    {
+      "epoch": 2.289002557544757,
+      "grad_norm": 1366.1842041015625,
+      "learning_rate": 0.00029222809042851755,
+      "loss": 7.1959,
+      "step": 1790
+    },
+    {
+      "epoch": 2.3017902813299234,
+      "grad_norm": 8.141697883605957,
+      "learning_rate": 0.0002921308283060831,
+      "loss": 6.8231,
+      "step": 1800
+    },
+    {
+      "epoch": 2.3145780051150897,
+      "grad_norm": 20.41695785522461,
+      "learning_rate": 0.0002920329777557023,
+      "loss": 6.6307,
+      "step": 1810
+    },
+    {
+      "epoch": 2.3273657289002556,
+      "grad_norm": 5.036224365234375,
+      "learning_rate": 0.0002919345391824809,
+      "loss": 6.3748,
+      "step": 1820
+    },
+    {
+      "epoch": 2.340153452685422,
+      "grad_norm": 1.9425593614578247,
+      "learning_rate": 0.00029183551299395873,
+      "loss": 6.3822,
+      "step": 1830
+    },
+    {
+      "epoch": 2.3529411764705883,
+      "grad_norm": 17.239046096801758,
+      "learning_rate": 0.00029173589960010887,
+      "loss": 6.4479,
+      "step": 1840
+    },
+    {
+      "epoch": 2.3657289002557547,
+      "grad_norm": 12.49085521697998,
+      "learning_rate": 0.00029163569941333513,
+      "loss": 6.3155,
+      "step": 1850
+    },
+    {
+      "epoch": 2.3785166240409206,
+      "grad_norm": 1.438551664352417,
+      "learning_rate": 0.0002915349128484706,
+      "loss": 6.3356,
+      "step": 1860
+    },
+    {
+      "epoch": 2.391304347826087,
+      "grad_norm": 10.845519065856934,
+      "learning_rate": 0.0002914335403227763,
+      "loss": 6.3907,
+      "step": 1870
+    },
+    {
+      "epoch": 2.4040920716112533,
+      "grad_norm": 1.9113539457321167,
+      "learning_rate": 0.0002913315822559389,
+      "loss": 6.2959,
+      "step": 1880
+    },
+    {
+      "epoch": 2.4168797953964196,
+      "grad_norm": 1.137809157371521,
+      "learning_rate": 0.0002912290390700694,
+      "loss": 6.2159,
+      "step": 1890
+    },
+    {
+      "epoch": 2.4296675191815855,
+      "grad_norm": 6.75790548324585,
+      "learning_rate": 0.0002911259111897011,
+      "loss": 6.1437,
+      "step": 1900
+    },
+    {
+      "epoch": 2.442455242966752,
+      "grad_norm": 13.505289077758789,
+      "learning_rate": 0.000291022199041788,
+      "loss": 6.1807,
+      "step": 1910
+    },
+    {
+      "epoch": 2.455242966751918,
+      "grad_norm": 1.1891614198684692,
+      "learning_rate": 0.000290917903055703,
+      "loss": 6.0738,
+      "step": 1920
+    },
+    {
+      "epoch": 2.4680306905370846,
+      "grad_norm": 9.028098106384277,
+      "learning_rate": 0.0002908130236632361,
+      "loss": 6.0556,
+      "step": 1930
+    },
+    {
+      "epoch": 2.4808184143222505,
+      "grad_norm": 41.855743408203125,
+      "learning_rate": 0.0002907075612985928,
+      "loss": 6.0713,
+      "step": 1940
+    },
+    {
+      "epoch": 2.493606138107417,
+      "grad_norm": 5.515649795532227,
+      "learning_rate": 0.00029060151639839174,
+      "loss": 6.0796,
+      "step": 1950
+    },
+    {
+      "epoch": 2.506393861892583,
+      "grad_norm": 740.7005004882812,
+      "learning_rate": 0.0002904948894016637,
+      "loss": 6.7849,
+      "step": 1960
+    },
+    {
+      "epoch": 2.5191815856777495,
+      "grad_norm": 100.34882354736328,
+      "learning_rate": 0.0002903876807498491,
+      "loss": 7.0059,
+      "step": 1970
+    },
+    {
+      "epoch": 2.531969309462916,
+      "grad_norm": 105.65509033203125,
+      "learning_rate": 0.0002902798908867966,
+      "loss": 6.3334,
+      "step": 1980
+    },
+    {
+      "epoch": 2.544757033248082,
+      "grad_norm": 20.092199325561523,
+      "learning_rate": 0.0002901715202587609,
+      "loss": 6.376,
+      "step": 1990
+    },
+    {
+      "epoch": 2.557544757033248,
+      "grad_norm": 4.902982711791992,
+      "learning_rate": 0.0002900625693144013,
+      "loss": 6.3327,
+      "step": 2000
+    },
+    {
+      "epoch": 2.5703324808184145,
+      "grad_norm": 4.835528373718262,
+      "learning_rate": 0.00028995303850477957,
+      "loss": 6.2468,
+      "step": 2010
+    },
+    {
+      "epoch": 2.5831202046035804,
+      "grad_norm": 3.2727978229522705,
+      "learning_rate": 0.00028984292828335807,
+      "loss": 6.1977,
+      "step": 2020
+    },
+    {
+      "epoch": 2.5959079283887467,
+      "grad_norm": 0.9278408288955688,
+      "learning_rate": 0.00028973223910599803,
+      "loss": 6.0873,
+      "step": 2030
+    },
+    {
+      "epoch": 2.608695652173913,
+      "grad_norm": 0.6826826333999634,
+      "learning_rate": 0.0002896209714309577,
+      "loss": 5.9931,
+      "step": 2040
+    },
+    {
+      "epoch": 2.6214833759590794,
+      "grad_norm": 0.9041231274604797,
+      "learning_rate": 0.0002895091257188899,
+      "loss": 5.9545,
+      "step": 2050
+    },
+    {
+      "epoch": 2.634271099744246,
+      "grad_norm": 0.9247362017631531,
+      "learning_rate": 0.00028939670243284115,
+      "loss": 5.9662,
+      "step": 2060
+    },
+    {
+      "epoch": 2.6470588235294117,
+      "grad_norm": 1.4400997161865234,
+      "learning_rate": 0.00028928370203824865,
+      "loss": 5.9395,
+      "step": 2070
+    },
+    {
+      "epoch": 2.659846547314578,
+      "grad_norm": 1.0552617311477661,
+      "learning_rate": 0.00028917012500293915,
+      "loss": 5.9204,
+      "step": 2080
+    },
+    {
+      "epoch": 2.6726342710997444,
+      "grad_norm": 0.743399977684021,
+      "learning_rate": 0.0002890559717971266,
+      "loss": 5.9061,
+      "step": 2090
+    },
+    {
+      "epoch": 2.6854219948849103,
+      "grad_norm": 49.769935607910156,
+      "learning_rate": 0.0002889412428934104,
+      "loss": 6.8681,
+      "step": 2100
+    },
+    {
+      "epoch": 2.6982097186700766,
+      "grad_norm": 1.0962203741073608,
+      "learning_rate": 0.0002888259387667732,
+      "loss": 6.0047,
+      "step": 2110
+    },
+    {
+      "epoch": 2.710997442455243,
+      "grad_norm": 0.7588474154472351,
+      "learning_rate": 0.0002887100598945793,
+      "loss": 5.946,
+      "step": 2120
+    },
+    {
+      "epoch": 2.7237851662404093,
+      "grad_norm": 1.135327935218811,
+      "learning_rate": 0.0002885936067565723,
+      "loss": 5.917,
+      "step": 2130
+    },
+    {
+      "epoch": 2.7365728900255757,
+      "grad_norm": 1.0815144777297974,
+      "learning_rate": 0.00028847657983487344,
+      "loss": 5.8928,
+      "step": 2140
+    },
+    {
+      "epoch": 2.7493606138107416,
+      "grad_norm": 0.9049263000488281,
+      "learning_rate": 0.0002883589796139794,
+      "loss": 5.8849,
+      "step": 2150
+    },
+    {
+      "epoch": 2.762148337595908,
+      "grad_norm": 1.9964706897735596,
+      "learning_rate": 0.00028824080658076024,
+      "loss": 5.9063,
+      "step": 2160
+    },
+    {
+      "epoch": 2.7749360613810743,
+      "grad_norm": 1.1962535381317139,
+      "learning_rate": 0.00028812206122445766,
+      "loss": 5.8739,
+      "step": 2170
+    },
+    {
+      "epoch": 2.78772378516624,
+      "grad_norm": 0.9967917203903198,
+      "learning_rate": 0.0002880027440366826,
+      "loss": 5.8599,
+      "step": 2180
+    },
+    {
+      "epoch": 2.8005115089514065,
+      "grad_norm": 1.099144697189331,
+      "learning_rate": 0.00028788285551141366,
+      "loss": 5.8559,
+      "step": 2190
+    },
+    {
+      "epoch": 2.813299232736573,
+      "grad_norm": 0.7207736968994141,
+      "learning_rate": 0.00028776239614499447,
+      "loss": 5.8369,
+      "step": 2200
+    },
+    {
+      "epoch": 2.8260869565217392,
+      "grad_norm": 1.000334620475769,
+      "learning_rate": 0.0002876413664361323,
+      "loss": 5.8644,
+      "step": 2210
+    },
+    {
+      "epoch": 2.8388746803069056,
+      "grad_norm": 0.778367280960083,
+      "learning_rate": 0.0002875197668858956,
+      "loss": 5.8342,
+      "step": 2220
+    },
+    {
+      "epoch": 2.8516624040920715,
+      "grad_norm": 1.511242389678955,
+      "learning_rate": 0.0002873975979977117,
+      "loss": 5.8202,
+      "step": 2230
+    },
+    {
+      "epoch": 2.864450127877238,
+      "grad_norm": 1.0907540321350098,
+      "learning_rate": 0.00028727486027736536,
+      "loss": 5.8303,
+      "step": 2240
+    },
+    {
+      "epoch": 2.877237851662404,
+      "grad_norm": 1.5989691019058228,
+      "learning_rate": 0.00028715155423299617,
+      "loss": 5.8212,
+      "step": 2250
+    },
+    {
+      "epoch": 2.89002557544757,
+      "grad_norm": 1.2920646667480469,
+      "learning_rate": 0.0002870276803750967,
+      "loss": 5.825,
+      "step": 2260
+    },
+    {
+      "epoch": 2.9028132992327365,
+      "grad_norm": 0.9930233955383301,
+      "learning_rate": 0.0002869032392165101,
+      "loss": 5.8205,
+      "step": 2270
+    },
+    {
+      "epoch": 2.915601023017903,
+      "grad_norm": 0.9936062693595886,
+      "learning_rate": 0.0002867782312724284,
+      "loss": 5.7916,
+      "step": 2280
+    },
+    {
+      "epoch": 2.928388746803069,
+      "grad_norm": 2.00677490234375,
+      "learning_rate": 0.00028665265706039,
+      "loss": 5.8037,
+      "step": 2290
+    },
+    {
+      "epoch": 2.9411764705882355,
+      "grad_norm": 1.0021674633026123,
+      "learning_rate": 0.00028652651710027775,
+      "loss": 5.8161,
+      "step": 2300
+    },
+    {
+      "epoch": 2.9539641943734014,
+      "grad_norm": 0.845365583896637,
+      "learning_rate": 0.0002863998119143166,
+      "loss": 5.787,
+      "step": 2310
+    },
+    {
+      "epoch": 2.9667519181585678,
+      "grad_norm": 0.663759708404541,
+      "learning_rate": 0.0002862725420270718,
+      "loss": 5.8021,
+      "step": 2320
+    },
+    {
+      "epoch": 2.979539641943734,
+      "grad_norm": 1.7215557098388672,
+      "learning_rate": 0.0002861447079654462,
+      "loss": 5.7939,
+      "step": 2330
+    },
+    {
+      "epoch": 2.9923273657289,
+      "grad_norm": 1.1704254150390625,
+      "learning_rate": 0.00028601631025867855,
+      "loss": 5.7715,
+      "step": 2340
+    },
+    {
+      "epoch": 3.0051150895140664,
+      "grad_norm": 0.7022088170051575,
+      "learning_rate": 0.00028588734943834113,
+      "loss": 5.7745,
+      "step": 2350
+    },
+    {
+      "epoch": 3.0179028132992327,
+      "grad_norm": 1.17882239818573,
+      "learning_rate": 0.0002857578260383374,
+      "loss": 5.7668,
+      "step": 2360
+    },
+    {
+      "epoch": 3.030690537084399,
+      "grad_norm": 1.0343165397644043,
+      "learning_rate": 0.00028562774059489996,
+      "loss": 5.7837,
+      "step": 2370
+    },
+    {
+      "epoch": 3.0434782608695654,
+      "grad_norm": 1.3794867992401123,
+      "learning_rate": 0.00028549709364658843,
+      "loss": 5.7698,
+      "step": 2380
+    },
+    {
+      "epoch": 3.0562659846547313,
+      "grad_norm": 1.120381474494934,
+      "learning_rate": 0.00028536588573428685,
+      "loss": 5.7644,
+      "step": 2390
+    },
+    {
+      "epoch": 3.0690537084398977,
+      "grad_norm": 1.0266337394714355,
+      "learning_rate": 0.0002852341174012019,
+      "loss": 5.7424,
+      "step": 2400
+    },
+    {
+      "epoch": 3.081841432225064,
+      "grad_norm": 1.448861002922058,
+      "learning_rate": 0.0002851017891928604,
+      "loss": 5.7292,
+      "step": 2410
+    },
+    {
+      "epoch": 3.0946291560102304,
+      "grad_norm": 1.5596977472305298,
+      "learning_rate": 0.0002849689016571068,
+      "loss": 5.7575,
+      "step": 2420
+    },
+    {
+      "epoch": 3.1074168797953963,
+      "grad_norm": 0.8733810186386108,
+      "learning_rate": 0.00028483545534410156,
+      "loss": 5.7362,
+      "step": 2430
+    },
+    {
+      "epoch": 3.1202046035805626,
+      "grad_norm": 9.138484954833984,
+      "learning_rate": 0.0002847014508063184,
+      "loss": 5.7518,
+      "step": 2440
+    },
+    {
+      "epoch": 3.132992327365729,
+      "grad_norm": 7.065587520599365,
+      "learning_rate": 0.0002845668885985419,
+      "loss": 5.9074,
+      "step": 2450
+    },
+    {
+      "epoch": 3.1457800511508953,
+      "grad_norm": 2.038409471511841,
+      "learning_rate": 0.0002844317692778657,
+      "loss": 5.7704,
+      "step": 2460
+    },
+    {
+      "epoch": 3.1585677749360612,
+      "grad_norm": 0.9123558402061462,
+      "learning_rate": 0.00028429609340368975,
+      "loss": 5.7661,
+      "step": 2470
+    },
+    {
+      "epoch": 3.1713554987212276,
+      "grad_norm": 0.7301475405693054,
+      "learning_rate": 0.0002841598615377182,
+      "loss": 5.7469,
+      "step": 2480
+    },
+    {
+      "epoch": 3.184143222506394,
+      "grad_norm": 0.8775326609611511,
+      "learning_rate": 0.000284023074243957,
+      "loss": 5.7485,
+      "step": 2490
+    },
+    {
+      "epoch": 3.1969309462915603,
+      "grad_norm": 1.6849805116653442,
+      "learning_rate": 0.00028388573208871175,
+      "loss": 5.7296,
+      "step": 2500
+    },
+    {
+      "epoch": 3.209718670076726,
+      "grad_norm": 0.8062297701835632,
+      "learning_rate": 0.00028374783564058493,
+      "loss": 5.7297,
+      "step": 2510
+    },
+    {
+      "epoch": 3.2225063938618925,
+      "grad_norm": 0.9257404804229736,
+      "learning_rate": 0.00028360938547047405,
+      "loss": 5.7582,
+      "step": 2520
+    },
+    {
+      "epoch": 3.235294117647059,
+      "grad_norm": 1.384611964225769,
+      "learning_rate": 0.00028347038215156893,
+      "loss": 5.7018,
+      "step": 2530
+    },
+    {
+      "epoch": 3.2480818414322252,
+      "grad_norm": 1.1805202960968018,
+      "learning_rate": 0.0002833308262593495,
+      "loss": 5.7122,
+      "step": 2540
+    },
+    {
+      "epoch": 3.260869565217391,
+      "grad_norm": 1.3140634298324585,
+      "learning_rate": 0.00028319071837158336,
+      "loss": 5.7173,
+      "step": 2550
+    },
+    {
+      "epoch": 3.2736572890025575,
+      "grad_norm": 0.8321253061294556,
+      "learning_rate": 0.00028305005906832333,
+      "loss": 5.7095,
+      "step": 2560
+    },
+    {
+      "epoch": 3.286445012787724,
+      "grad_norm": 0.866754412651062,
+      "learning_rate": 0.0002829088489319053,
+      "loss": 5.7009,
+      "step": 2570
+    },
+    {
+      "epoch": 3.29923273657289,
+      "grad_norm": 0.97830730676651,
+      "learning_rate": 0.00028276708854694545,
+      "loss": 5.6795,
+      "step": 2580
+    },
+    {
+      "epoch": 3.312020460358056,
+      "grad_norm": 1.2616524696350098,
+      "learning_rate": 0.00028262477850033807,
+      "loss": 5.7157,
+      "step": 2590
+    },
+    {
+      "epoch": 3.3248081841432224,
+      "grad_norm": 0.9064726829528809,
+      "learning_rate": 0.00028248191938125305,
+      "loss": 5.7089,
+      "step": 2600
+    },
+    {
+      "epoch": 3.337595907928389,
+      "grad_norm": 0.9063658714294434,
+      "learning_rate": 0.0002823385117811335,
+      "loss": 5.9595,
+      "step": 2610
+    },
+    {
+      "epoch": 3.350383631713555,
+      "grad_norm": 1.100963830947876,
+      "learning_rate": 0.00028219455629369334,
+      "loss": 5.6872,
+      "step": 2620
+    },
+    {
+      "epoch": 3.363171355498721,
+      "grad_norm": 1.2329927682876587,
+      "learning_rate": 0.0002820500535149146,
+      "loss": 5.7029,
+      "step": 2630
+    },
+    {
+      "epoch": 3.3759590792838874,
+      "grad_norm": 0.8984890580177307,
+      "learning_rate": 0.00028190500404304524,
+      "loss": 5.7381,
+      "step": 2640
+    },
+    {
+      "epoch": 3.3887468030690537,
+      "grad_norm": 0.696719765663147,
+      "learning_rate": 0.0002817594084785965,
+      "loss": 5.6881,
+      "step": 2650
+    },
+    {
+      "epoch": 3.40153452685422,
+      "grad_norm": 1.0669254064559937,
+      "learning_rate": 0.0002816132674243405,
+      "loss": 5.6737,
+      "step": 2660
+    },
+    {
+      "epoch": 3.414322250639386,
+      "grad_norm": 0.8603677153587341,
+      "learning_rate": 0.0002814665814853077,
+      "loss": 5.6669,
+      "step": 2670
+    },
+    {
+      "epoch": 3.4271099744245523,
+      "grad_norm": 0.902407169342041,
+      "learning_rate": 0.00028131935126878444,
+      "loss": 5.6782,
+      "step": 2680
+    },
+    {
+      "epoch": 3.4398976982097187,
+      "grad_norm": 0.8212502598762512,
+      "learning_rate": 0.00028117157738431027,
+      "loss": 5.6825,
+      "step": 2690
+    },
+    {
+      "epoch": 3.452685421994885,
+      "grad_norm": 0.7769032120704651,
+      "learning_rate": 0.0002810232604436758,
+      "loss": 5.6798,
+      "step": 2700
+    },
+    {
+      "epoch": 3.4654731457800514,
+      "grad_norm": 1.140012264251709,
+      "learning_rate": 0.0002808744010609196,
+      "loss": 5.6552,
+      "step": 2710
+    },
+    {
+      "epoch": 3.4782608695652173,
+      "grad_norm": 1.1231127977371216,
+      "learning_rate": 0.00028072499985232614,
+      "loss": 5.6731,
+      "step": 2720
+    },
+    {
+      "epoch": 3.4910485933503836,
+      "grad_norm": 0.6917468309402466,
+      "learning_rate": 0.0002805750574364231,
+      "loss": 5.6562,
+      "step": 2730
+    },
+    {
+      "epoch": 3.50383631713555,
+      "grad_norm": 1.1626530885696411,
+      "learning_rate": 0.00028042457443397883,
+      "loss": 5.6433,
+      "step": 2740
+    },
+    {
+      "epoch": 3.516624040920716,
+      "grad_norm": 1.6136029958724976,
+      "learning_rate": 0.0002802735514679995,
+      "loss": 5.6741,
+      "step": 2750
+    },
+    {
+      "epoch": 3.5294117647058822,
+      "grad_norm": 0.9433214068412781,
+      "learning_rate": 0.00028012198916372704,
+      "loss": 5.6615,
+      "step": 2760
+    },
+    {
+      "epoch": 3.5421994884910486,
+      "grad_norm": 0.8657507300376892,
+      "learning_rate": 0.000279969888148636,
+      "loss": 5.6417,
+      "step": 2770
+    },
+    {
+      "epoch": 3.554987212276215,
+      "grad_norm": 1.2951968908309937,
+      "learning_rate": 0.0002798172490524315,
+      "loss": 5.6563,
+      "step": 2780
+    },
+    {
+      "epoch": 3.5677749360613813,
+      "grad_norm": 0.9073137640953064,
+      "learning_rate": 0.000279664072507046,
+      "loss": 5.6356,
+      "step": 2790
+    },
+    {
+      "epoch": 3.580562659846547,
+      "grad_norm": 1.0318324565887451,
+      "learning_rate": 0.00027951035914663736,
+      "loss": 5.6399,
+      "step": 2800
+    },
+    {
+      "epoch": 3.5933503836317136,
+      "grad_norm": 1.2848615646362305,
+      "learning_rate": 0.00027935610960758557,
+      "loss": 5.6362,
+      "step": 2810
+    },
+    {
+      "epoch": 3.60613810741688,
+      "grad_norm": 1.8240704536437988,
+      "learning_rate": 0.0002792013245284907,
+      "loss": 5.631,
+      "step": 2820
+    },
+    {
+      "epoch": 3.618925831202046,
+      "grad_norm": 3.091918468475342,
+      "learning_rate": 0.00027904600455016984,
+      "loss": 5.6453,
+      "step": 2830
+    },
+    {
+      "epoch": 3.631713554987212,
+      "grad_norm": 0.8699560165405273,
+      "learning_rate": 0.00027889015031565457,
+      "loss": 5.6166,
+      "step": 2840
+    },
+    {
+      "epoch": 3.6445012787723785,
+      "grad_norm": 0.9239150285720825,
+      "learning_rate": 0.0002787337624701883,
+      "loss": 5.618,
+      "step": 2850
+    },
+    {
+      "epoch": 3.657289002557545,
+      "grad_norm": 0.8173996806144714,
+      "learning_rate": 0.00027857684166122384,
+      "loss": 5.6296,
+      "step": 2860
+    },
+    {
+      "epoch": 3.670076726342711,
+      "grad_norm": 0.7985453009605408,
+      "learning_rate": 0.0002784193885384201,
+      "loss": 5.6385,
+      "step": 2870
+    },
+    {
+      "epoch": 3.682864450127877,
+      "grad_norm": 1.2572569847106934,
+      "learning_rate": 0.0002782614037536402,
+      "loss": 5.6183,
+      "step": 2880
+    },
+    {
+      "epoch": 3.6956521739130435,
+      "grad_norm": 1.2069624662399292,
+      "learning_rate": 0.000278102887960948,
+      "loss": 5.5862,
+      "step": 2890
+    },
+    {
+      "epoch": 3.70843989769821,
+      "grad_norm": 0.8699544072151184,
+      "learning_rate": 0.000277943841816606,
+      "loss": 5.6158,
+      "step": 2900
+    },
+    {
+      "epoch": 3.7212276214833757,
+      "grad_norm": 1.2560055255889893,
+      "learning_rate": 0.00027778426597907237,
+      "loss": 5.6094,
+      "step": 2910
+    },
+    {
+      "epoch": 3.734015345268542,
+      "grad_norm": 1.0291788578033447,
+      "learning_rate": 0.0002776241611089981,
+      "loss": 5.6025,
+      "step": 2920
+    },
+    {
+      "epoch": 3.7468030690537084,
+      "grad_norm": 0.9255660772323608,
+      "learning_rate": 0.0002774635278692244,
+      "loss": 5.6193,
+      "step": 2930
+    },
+    {
+      "epoch": 3.7595907928388748,
+      "grad_norm": 0.9527194499969482,
+      "learning_rate": 0.0002773023669247801,
+      "loss": 5.609,
+      "step": 2940
+    },
+    {
+      "epoch": 3.772378516624041,
+      "grad_norm": 1.1681740283966064,
+      "learning_rate": 0.00027714067894287874,
+      "loss": 5.5862,
+      "step": 2950
+    },
+    {
+      "epoch": 3.785166240409207,
+      "grad_norm": 0.9430054426193237,
+      "learning_rate": 0.0002769784645929156,
+      "loss": 5.601,
+      "step": 2960
+    },
+    {
+      "epoch": 3.7979539641943734,
+      "grad_norm": 0.7085330486297607,
+      "learning_rate": 0.0002768157245464655,
+      "loss": 5.5931,
+      "step": 2970
+    },
+    {
+      "epoch": 3.8107416879795397,
+      "grad_norm": 0.92620849609375,
+      "learning_rate": 0.00027665245947727926,
+      "loss": 5.584,
+      "step": 2980
+    },
+    {
+      "epoch": 3.8235294117647056,
+      "grad_norm": 1.2099533081054688,
+      "learning_rate": 0.00027648867006128166,
+      "loss": 5.5962,
+      "step": 2990
+    },
+    {
+      "epoch": 3.836317135549872,
+      "grad_norm": 0.8536057472229004,
+      "learning_rate": 0.00027632435697656804,
+      "loss": 5.575,
+      "step": 3000
+    },
+    {
+      "epoch": 3.8491048593350383,
+      "grad_norm": 0.6810599565505981,
+      "learning_rate": 0.000276159520903402,
+      "loss": 5.575,
+      "step": 3010
+    },
+    {
+      "epoch": 3.8618925831202047,
+      "grad_norm": 1.2196543216705322,
+      "learning_rate": 0.00027599416252421213,
+      "loss": 5.6085,
+      "step": 3020
+    },
+    {
+      "epoch": 3.874680306905371,
+      "grad_norm": 1.5269405841827393,
+      "learning_rate": 0.00027582828252358953,
+      "loss": 5.599,
+      "step": 3030
+    },
+    {
+      "epoch": 3.887468030690537,
+      "grad_norm": 0.9655811786651611,
+      "learning_rate": 0.0002756618815882847,
+      "loss": 5.5786,
+      "step": 3040
+    },
+    {
+      "epoch": 3.9002557544757033,
+      "grad_norm": 0.9123175144195557,
+      "learning_rate": 0.000275494960407205,
+      "loss": 5.5982,
+      "step": 3050
+    },
+    {
+      "epoch": 3.9130434782608696,
+      "grad_norm": 1.5311001539230347,
+      "learning_rate": 0.0002753275196714114,
+      "loss": 5.5947,
+      "step": 3060
+    },
+    {
+      "epoch": 3.9258312020460355,
+      "grad_norm": 0.829192578792572,
+      "learning_rate": 0.0002751595600741161,
+      "loss": 5.574,
+      "step": 3070
+    },
+    {
+      "epoch": 3.938618925831202,
+      "grad_norm": 1.4503875970840454,
+      "learning_rate": 0.0002749910823106793,
+      "loss": 5.5739,
+      "step": 3080
+    },
+    {
+      "epoch": 3.9514066496163682,
+      "grad_norm": 1.0710264444351196,
+      "learning_rate": 0.0002748220870786064,
+      "loss": 5.583,
+      "step": 3090
+    },
+    {
+      "epoch": 3.9641943734015346,
+      "grad_norm": 1.2536656856536865,
+      "learning_rate": 0.00027465257507754517,
+      "loss": 5.5767,
+      "step": 3100
+    },
+    {
+      "epoch": 3.976982097186701,
+      "grad_norm": 0.6235175728797913,
+      "learning_rate": 0.0002744825470092828,
+      "loss": 5.5572,
+      "step": 3110
+    },
+    {
+      "epoch": 3.9897698209718673,
+      "grad_norm": 0.9897062182426453,
+      "learning_rate": 0.0002743120035777431,
+      "loss": 5.5586,
+      "step": 3120
+    },
+    {
+      "epoch": 4.002557544757034,
+      "grad_norm": 1.1203209161758423,
+      "learning_rate": 0.00027414094548898347,
+      "loss": 5.5683,
+      "step": 3130
+    },
+    {
+      "epoch": 4.015345268542199,
+      "grad_norm": 0.6553176045417786,
+      "learning_rate": 0.0002739693734511919,
+      "loss": 5.5741,
+      "step": 3140
+    },
+    {
+      "epoch": 4.028132992327365,
+      "grad_norm": 0.6654773950576782,
+      "learning_rate": 0.00027379728817468436,
+      "loss": 5.537,
+      "step": 3150
+    },
+    {
+      "epoch": 4.040920716112532,
+      "grad_norm": 1.263575792312622,
+      "learning_rate": 0.0002736246903719015,
+      "loss": 5.5378,
+      "step": 3160
+    },
+    {
+      "epoch": 4.053708439897698,
+      "grad_norm": 1.274593710899353,
+      "learning_rate": 0.0002734515807574059,
+      "loss": 5.5649,
+      "step": 3170
+    },
+    {
+      "epoch": 4.0664961636828645,
+      "grad_norm": 0.804013192653656,
+      "learning_rate": 0.000273277960047879,
+      "loss": 5.55,
+      "step": 3180
+    },
+    {
+      "epoch": 4.079283887468031,
+      "grad_norm": 1.15504789352417,
+      "learning_rate": 0.0002731038289621184,
+      "loss": 5.5446,
+      "step": 3190
+    },
+    {
+      "epoch": 4.092071611253197,
+      "grad_norm": 0.9598748087882996,
+      "learning_rate": 0.0002729291882210344,
+      "loss": 5.547,
+      "step": 3200
+    },
+    {
+      "epoch": 4.1048593350383635,
+      "grad_norm": 0.9165511727333069,
+      "learning_rate": 0.0002727540385476475,
+      "loss": 5.5288,
+      "step": 3210
+    },
+    {
+      "epoch": 4.117647058823529,
+      "grad_norm": 0.8215676546096802,
+      "learning_rate": 0.00027257838066708503,
+      "loss": 5.5356,
+      "step": 3220
+    },
+    {
+      "epoch": 4.130434782608695,
+      "grad_norm": 0.9655877947807312,
+      "learning_rate": 0.0002724022153065786,
+      "loss": 5.5553,
+      "step": 3230
+    },
+    {
+      "epoch": 4.143222506393862,
+      "grad_norm": 1.4203110933303833,
+      "learning_rate": 0.00027222554319546047,
+      "loss": 5.5253,
+      "step": 3240
+    },
+    {
+      "epoch": 4.156010230179028,
+      "grad_norm": 0.8605411052703857,
+      "learning_rate": 0.0002720483650651611,
+      "loss": 5.5499,
+      "step": 3250
+    },
+    {
+      "epoch": 4.168797953964194,
+      "grad_norm": 1.030411720275879,
+      "learning_rate": 0.00027187068164920584,
+      "loss": 5.5306,
+      "step": 3260
+    },
+    {
+      "epoch": 4.181585677749361,
+      "grad_norm": 0.9130074977874756,
+      "learning_rate": 0.0002716924936832119,
+      "loss": 5.524,
+      "step": 3270
+    },
+    {
+      "epoch": 4.194373401534527,
+      "grad_norm": 0.9061768054962158,
+      "learning_rate": 0.00027151380190488535,
+      "loss": 5.5197,
+      "step": 3280
+    },
+    {
+      "epoch": 4.207161125319693,
+      "grad_norm": 0.9149904847145081,
+      "learning_rate": 0.00027133460705401814,
+      "loss": 5.5273,
+      "step": 3290
+    },
+    {
+      "epoch": 4.21994884910486,
+      "grad_norm": 0.8536543846130371,
+      "learning_rate": 0.00027115490987248493,
+      "loss": 5.524,
+      "step": 3300
+    },
+    {
+      "epoch": 4.232736572890025,
+      "grad_norm": 0.8059023022651672,
+      "learning_rate": 0.00027097471110424005,
+      "loss": 5.5276,
+      "step": 3310
+    },
+    {
+      "epoch": 4.245524296675192,
+      "grad_norm": 0.894961953163147,
+      "learning_rate": 0.00027079401149531434,
+      "loss": 5.52,
+      "step": 3320
+    },
+    {
+      "epoch": 4.258312020460358,
+      "grad_norm": 0.907398521900177,
+      "learning_rate": 0.0002706128117938122,
+      "loss": 5.5113,
+      "step": 3330
+    },
+    {
+      "epoch": 4.271099744245524,
+      "grad_norm": 0.701256513595581,
+      "learning_rate": 0.0002704311127499086,
+      "loss": 5.514,
+      "step": 3340
+    },
+    {
+      "epoch": 4.283887468030691,
+      "grad_norm": 0.8591639399528503,
+      "learning_rate": 0.00027024891511584547,
+      "loss": 5.5133,
+      "step": 3350
+    },
+    {
+      "epoch": 4.296675191815857,
+      "grad_norm": 1.0663071870803833,
+      "learning_rate": 0.0002700662196459292,
+      "loss": 5.4978,
+      "step": 3360
+    },
+    {
+      "epoch": 4.309462915601023,
+      "grad_norm": 1.0883492231369019,
+      "learning_rate": 0.00026988302709652723,
+      "loss": 5.5314,
+      "step": 3370
+    },
+    {
+      "epoch": 4.322250639386189,
+      "grad_norm": 1.2101244926452637,
+      "learning_rate": 0.00026969933822606465,
+      "loss": 5.4919,
+      "step": 3380
+    },
+    {
+      "epoch": 4.335038363171355,
+      "grad_norm": 1.37273371219635,
+      "learning_rate": 0.00026951515379502176,
+      "loss": 5.5201,
+      "step": 3390
+    },
+    {
+      "epoch": 4.3478260869565215,
+      "grad_norm": 0.8530544638633728,
+      "learning_rate": 0.00026933047456593025,
+      "loss": 5.4963,
+      "step": 3400
+    },
+    {
+      "epoch": 4.360613810741688,
+      "grad_norm": 1.0042463541030884,
+      "learning_rate": 0.0002691453013033702,
+      "loss": 5.5079,
+      "step": 3410
+    },
+    {
+      "epoch": 4.373401534526854,
+      "grad_norm": 0.9824084639549255,
+      "learning_rate": 0.00026895963477396726,
+      "loss": 5.5065,
+      "step": 3420
+    },
+    {
+      "epoch": 4.3861892583120206,
+      "grad_norm": 0.9359764456748962,
+      "learning_rate": 0.0002687734757463891,
+      "loss": 5.497,
+      "step": 3430
+    },
+    {
+      "epoch": 4.398976982097187,
+      "grad_norm": 0.7871713042259216,
+      "learning_rate": 0.0002685868249913424,
+      "loss": 5.5059,
+      "step": 3440
+    },
+    {
+      "epoch": 4.411764705882353,
+      "grad_norm": 0.9313884377479553,
+      "learning_rate": 0.00026839968328156957,
+      "loss": 5.503,
+      "step": 3450
+    },
+    {
+      "epoch": 4.42455242966752,
+      "grad_norm": 1.0534262657165527,
+      "learning_rate": 0.0002682120513918455,
+      "loss": 5.5077,
+      "step": 3460
+    },
+    {
+      "epoch": 4.437340153452685,
+      "grad_norm": 1.0140694379806519,
+      "learning_rate": 0.0002680239300989747,
+      "loss": 5.5238,
+      "step": 3470
+    },
+    {
+      "epoch": 4.450127877237851,
+      "grad_norm": 0.9650000929832458,
+      "learning_rate": 0.00026783532018178756,
+      "loss": 5.4972,
+      "step": 3480
+    },
+    {
+      "epoch": 4.462915601023018,
+      "grad_norm": 1.0754122734069824,
+      "learning_rate": 0.0002676462224211376,
+      "loss": 5.4936,
+      "step": 3490
+    },
+    {
+      "epoch": 4.475703324808184,
+      "grad_norm": 1.1840262413024902,
+      "learning_rate": 0.00026745663759989796,
+      "loss": 5.4895,
+      "step": 3500
+    },
+    {
+      "epoch": 4.4884910485933505,
+      "grad_norm": 3.5758776664733887,
+      "learning_rate": 0.00026726656650295815,
+      "loss": 5.4779,
+      "step": 3510
+    },
+    {
+      "epoch": 4.501278772378517,
+      "grad_norm": 1.3599027395248413,
+      "learning_rate": 0.000267076009917221,
+      "loss": 5.5273,
+      "step": 3520
+    },
+    {
+      "epoch": 4.514066496163683,
+      "grad_norm": 1.9289357662200928,
+      "learning_rate": 0.0002668849686315993,
+      "loss": 5.5244,
+      "step": 3530
+    },
+    {
+      "epoch": 4.526854219948849,
+      "grad_norm": 2.649031162261963,
+      "learning_rate": 0.0002666934434370124,
+      "loss": 5.5419,
+      "step": 3540
+    },
+    {
+      "epoch": 4.539641943734015,
+      "grad_norm": 1.6334396600723267,
+      "learning_rate": 0.00026650143512638316,
+      "loss": 5.5442,
+      "step": 3550
+    },
+    {
+      "epoch": 4.552429667519181,
+      "grad_norm": 9.740204811096191,
+      "learning_rate": 0.00026630894449463445,
+      "loss": 5.57,
+      "step": 3560
+    },
+    {
+      "epoch": 4.565217391304348,
+      "grad_norm": 5.615604400634766,
+      "learning_rate": 0.00026611597233868606,
+      "loss": 5.6171,
+      "step": 3570
+    },
+    {
+      "epoch": 4.578005115089514,
+      "grad_norm": 10.387894630432129,
+      "learning_rate": 0.0002659225194574513,
+      "loss": 5.5957,
+      "step": 3580
+    },
+    {
+      "epoch": 4.59079283887468,
+      "grad_norm": 6.4273247718811035,
+      "learning_rate": 0.00026572858665183384,
+      "loss": 5.5684,
+      "step": 3590
+    },
+    {
+      "epoch": 4.603580562659847,
+      "grad_norm": 3.7879221439361572,
+      "learning_rate": 0.0002655341747247239,
+      "loss": 5.5616,
+      "step": 3600
+    },
+    {
+      "epoch": 4.616368286445013,
+      "grad_norm": 1.7430964708328247,
+      "learning_rate": 0.0002653392844809956,
+      "loss": 5.5481,
+      "step": 3610
+    },
+    {
+      "epoch": 4.629156010230179,
+      "grad_norm": 1.193642020225525,
+      "learning_rate": 0.00026514391672750317,
+      "loss": 5.5487,
+      "step": 3620
+    },
+    {
+      "epoch": 4.641943734015345,
+      "grad_norm": 0.9543384909629822,
+      "learning_rate": 0.00026494807227307786,
+      "loss": 5.5215,
+      "step": 3630
+    },
+    {
+      "epoch": 4.654731457800511,
+      "grad_norm": 6.988561153411865,
+      "learning_rate": 0.0002647517519285243,
+      "loss": 5.5163,
+      "step": 3640
+    },
+    {
+      "epoch": 4.667519181585678,
+      "grad_norm": 35138.046875,
+      "learning_rate": 0.00026455495650661763,
+      "loss": 14.1816,
+      "step": 3650
+    },
+    {
+      "epoch": 4.680306905370844,
+      "grad_norm": 649.7811889648438,
+      "learning_rate": 0.00026435768682209947,
+      "loss": 9.2072,
+      "step": 3660
+    },
+    {
+      "epoch": 4.69309462915601,
+      "grad_norm": 1.4763447046279907,
+      "learning_rate": 0.0002641599436916751,
+      "loss": 6.1727,
+      "step": 3670
+    },
+    {
+      "epoch": 4.705882352941177,
+      "grad_norm": 1.2908670902252197,
+      "learning_rate": 0.00026396172793401,
+      "loss": 5.6292,
+      "step": 3680
+    },
+    {
+      "epoch": 4.718670076726343,
+      "grad_norm": 1.0045114755630493,
+      "learning_rate": 0.0002637630403697261,
+      "loss": 5.5647,
+      "step": 3690
+    },
+    {
+      "epoch": 4.731457800511509,
+      "grad_norm": 1.0943928956985474,
+      "learning_rate": 0.0002635638818213988,
+      "loss": 5.5281,
+      "step": 3700
+    },
+    {
+      "epoch": 4.744245524296675,
+      "grad_norm": 0.8787861466407776,
+      "learning_rate": 0.0002633642531135532,
+      "loss": 5.5262,
+      "step": 3710
+    },
+    {
+      "epoch": 4.757033248081841,
+      "grad_norm": 0.9018283486366272,
+      "learning_rate": 0.0002631641550726613,
+      "loss": 5.5107,
+      "step": 3720
+    },
+    {
+      "epoch": 4.7698209718670075,
+      "grad_norm": 1.1844482421875,
+      "learning_rate": 0.0002629635885271376,
+      "loss": 5.493,
+      "step": 3730
+    },
+    {
+      "epoch": 4.782608695652174,
+      "grad_norm": 1.2566533088684082,
+      "learning_rate": 0.0002627625543073367,
+      "loss": 5.4924,
+      "step": 3740
+    },
+    {
+      "epoch": 4.79539641943734,
+      "grad_norm": 1.1357030868530273,
+      "learning_rate": 0.0002625610532455491,
+      "loss": 5.4784,
+      "step": 3750
+    },
+    {
+      "epoch": 4.8081841432225065,
+      "grad_norm": 0.8598061800003052,
+      "learning_rate": 0.00026235908617599824,
+      "loss": 5.4796,
+      "step": 3760
+    },
+    {
+      "epoch": 4.820971867007673,
+      "grad_norm": 1.176074743270874,
+      "learning_rate": 0.0002621566539348367,
+      "loss": 5.4893,
+      "step": 3770
+    },
+    {
+      "epoch": 4.833759590792839,
+      "grad_norm": 0.7435081601142883,
+      "learning_rate": 0.00026195375736014296,
+      "loss": 5.489,
+      "step": 3780
+    },
+    {
+      "epoch": 4.846547314578006,
+      "grad_norm": 1.0044056177139282,
+      "learning_rate": 0.0002617503972919179,
+      "loss": 5.5022,
+      "step": 3790
+    },
+    {
+      "epoch": 4.859335038363171,
+      "grad_norm": 0.9446366429328918,
+      "learning_rate": 0.0002615465745720812,
+      "loss": 5.4633,
+      "step": 3800
+    },
+    {
+      "epoch": 4.872122762148337,
+      "grad_norm": 1.2357178926467896,
+      "learning_rate": 0.00026134229004446796,
+      "loss": 5.4782,
+      "step": 3810
+    },
+    {
+      "epoch": 4.884910485933504,
+      "grad_norm": 1.0345720052719116,
+      "learning_rate": 0.0002611375445548252,
+      "loss": 5.5007,
+      "step": 3820
+    },
+    {
+      "epoch": 4.89769820971867,
+      "grad_norm": 1.6297608613967896,
+      "learning_rate": 0.00026093233895080846,
+      "loss": 5.8077,
+      "step": 3830
+    },
+    {
+      "epoch": 4.910485933503836,
+      "grad_norm": 2.633462905883789,
+      "learning_rate": 0.000260726674081978,
+      "loss": 5.5718,
+      "step": 3840
+    },
+    {
+      "epoch": 4.923273657289003,
+      "grad_norm": 1.6603726148605347,
+      "learning_rate": 0.00026052055079979546,
+      "loss": 5.5366,
+      "step": 3850
+    },
+    {
+      "epoch": 4.936061381074169,
+      "grad_norm": 23.486774444580078,
+      "learning_rate": 0.0002603139699576204,
+      "loss": 5.6162,
+      "step": 3860
+    },
+    {
+      "epoch": 4.948849104859335,
+      "grad_norm": 1.7220473289489746,
+      "learning_rate": 0.0002601069324107067,
+      "loss": 5.777,
+      "step": 3870
+    },
+    {
+      "epoch": 4.961636828644501,
+      "grad_norm": 1.3732489347457886,
+      "learning_rate": 0.0002598994390161991,
+      "loss": 5.558,
+      "step": 3880
+    },
+    {
+      "epoch": 4.974424552429667,
+      "grad_norm": 1.398135781288147,
+      "learning_rate": 0.0002596914906331294,
+      "loss": 5.5602,
+      "step": 3890
+    },
+    {
+      "epoch": 4.987212276214834,
+      "grad_norm": 1.0198397636413574,
+      "learning_rate": 0.0002594830881224131,
+      "loss": 5.5123,
+      "step": 3900
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 1.7247081995010376,
+      "learning_rate": 0.0002592742323468459,
+      "loss": 5.5164,
+      "step": 3910
+    },
+    {
+      "epoch": 5.012787723785166,
+      "grad_norm": 1.3416905403137207,
+      "learning_rate": 0.00025906492417110004,
+      "loss": 5.5093,
+      "step": 3920
+    },
+    {
+      "epoch": 5.025575447570333,
+      "grad_norm": 0.9129336476325989,
+      "learning_rate": 0.0002588551644617206,
+      "loss": 5.4713,
+      "step": 3930
+    },
+    {
+      "epoch": 5.038363171355499,
+      "grad_norm": 0.8973252177238464,
+      "learning_rate": 0.0002586449540871221,
+      "loss": 5.4519,
+      "step": 3940
+    },
+    {
+      "epoch": 5.051150895140665,
+      "grad_norm": 0.9549187421798706,
+      "learning_rate": 0.00025843429391758497,
+      "loss": 5.4634,
+      "step": 3950
+    },
+    {
+      "epoch": 5.063938618925831,
+      "grad_norm": 7.259679794311523,
+      "learning_rate": 0.0002582231848252515,
+      "loss": 5.4512,
+      "step": 3960
+    },
+    {
+      "epoch": 5.076726342710997,
+      "grad_norm": 0.8443651795387268,
+      "learning_rate": 0.00025801162768412277,
+      "loss": 5.45,
+      "step": 3970
+    },
+    {
+      "epoch": 5.089514066496164,
+      "grad_norm": 1.066373348236084,
+      "learning_rate": 0.0002577996233700548,
+      "loss": 5.4714,
+      "step": 3980
+    },
+    {
+      "epoch": 5.10230179028133,
+      "grad_norm": 0.8174281716346741,
+      "learning_rate": 0.0002575871727607548,
+      "loss": 5.4518,
+      "step": 3990
+    },
+    {
+      "epoch": 5.115089514066496,
+      "grad_norm": 1.1461961269378662,
+      "learning_rate": 0.0002573742767357778,
+      "loss": 5.4323,
+      "step": 4000
+    },
+    {
+      "epoch": 5.127877237851663,
+      "grad_norm": 1.1951566934585571,
+      "learning_rate": 0.00025716093617652273,
+      "loss": 5.4453,
+      "step": 4010
+    },
+    {
+      "epoch": 5.140664961636829,
+      "grad_norm": 1.038336157798767,
+      "learning_rate": 0.00025694715196622903,
+      "loss": 5.4727,
+      "step": 4020
+    },
+    {
+      "epoch": 5.153452685421995,
+      "grad_norm": 1.3700714111328125,
+      "learning_rate": 0.0002567329249899729,
+      "loss": 5.508,
+      "step": 4030
+    },
+    {
+      "epoch": 5.166240409207161,
+      "grad_norm": 0.8140355348587036,
+      "learning_rate": 0.0002565182561346634,
+      "loss": 5.4863,
+      "step": 4040
+    },
+    {
+      "epoch": 5.179028132992327,
+      "grad_norm": 1.4147567749023438,
+      "learning_rate": 0.00025630314628903925,
+      "loss": 5.4518,
+      "step": 4050
+    },
+    {
+      "epoch": 5.1918158567774935,
+      "grad_norm": 0.7016567587852478,
+      "learning_rate": 0.0002560875963436647,
+      "loss": 5.4447,
+      "step": 4060
+    },
+    {
+      "epoch": 5.20460358056266,
+      "grad_norm": 0.7389644980430603,
+      "learning_rate": 0.0002558716071909262,
+      "loss": 5.444,
+      "step": 4070
+    },
+    {
+      "epoch": 5.217391304347826,
+      "grad_norm": 0.9639290571212769,
+      "learning_rate": 0.0002556551797250283,
+      "loss": 5.4364,
+      "step": 4080
+    },
+    {
+      "epoch": 5.2301790281329925,
+      "grad_norm": 0.8074414730072021,
+      "learning_rate": 0.0002554383148419904,
+      "loss": 5.4526,
+      "step": 4090
+    },
+    {
+      "epoch": 5.242966751918159,
+      "grad_norm": 1.0686334371566772,
+      "learning_rate": 0.0002552210134396429,
+      "loss": 5.4213,
+      "step": 4100
+    },
+    {
+      "epoch": 5.255754475703325,
+      "grad_norm": 0.7948906421661377,
+      "learning_rate": 0.00025500327641762296,
+      "loss": 5.4414,
+      "step": 4110
+    },
+    {
+      "epoch": 5.268542199488491,
+      "grad_norm": 3.3121063709259033,
+      "learning_rate": 0.00025478510467737175,
+      "loss": 5.4386,
+      "step": 4120
+    },
+    {
+      "epoch": 5.281329923273657,
+      "grad_norm": 1.0081732273101807,
+      "learning_rate": 0.0002545664991221299,
+      "loss": 5.436,
+      "step": 4130
+    },
+    {
+      "epoch": 5.294117647058823,
+      "grad_norm": 1.1705923080444336,
+      "learning_rate": 0.0002543474606569342,
+      "loss": 5.4249,
+      "step": 4140
+    },
+    {
+      "epoch": 5.30690537084399,
+      "grad_norm": 1.5626429319381714,
+      "learning_rate": 0.0002541279901886135,
+      "loss": 5.4149,
+      "step": 4150
+    },
+    {
+      "epoch": 5.319693094629156,
+      "grad_norm": 1.235137701034546,
+      "learning_rate": 0.00025390808862578543,
+      "loss": 5.4067,
+      "step": 4160
+    },
+    {
+      "epoch": 5.332480818414322,
+      "grad_norm": 1.3060702085494995,
+      "learning_rate": 0.0002536877568788521,
+      "loss": 5.4284,
+      "step": 4170
+    },
+    {
+      "epoch": 5.345268542199489,
+      "grad_norm": 0.7894797325134277,
+      "learning_rate": 0.00025346699585999683,
+      "loss": 5.4261,
+      "step": 4180
+    },
+    {
+      "epoch": 5.358056265984655,
+      "grad_norm": 0.6973442435264587,
+      "learning_rate": 0.00025324580648318,
+      "loss": 5.4292,
+      "step": 4190
+    },
+    {
+      "epoch": 5.370843989769821,
+      "grad_norm": 0.7937166094779968,
+      "learning_rate": 0.00025302418966413555,
+      "loss": 5.407,
+      "step": 4200
+    },
+    {
+      "epoch": 5.383631713554987,
+      "grad_norm": 1.1912145614624023,
+      "learning_rate": 0.00025280214632036684,
+      "loss": 5.4215,
+      "step": 4210
+    },
+    {
+      "epoch": 5.396419437340153,
+      "grad_norm": 1.135903000831604,
+      "learning_rate": 0.00025257967737114323,
+      "loss": 5.4018,
+      "step": 4220
+    },
+    {
+      "epoch": 5.40920716112532,
+      "grad_norm": 1.0384650230407715,
+      "learning_rate": 0.0002523567837374961,
+      "loss": 5.4002,
+      "step": 4230
+    },
+    {
+      "epoch": 5.421994884910486,
+      "grad_norm": 1.1547504663467407,
+      "learning_rate": 0.00025213346634221493,
+      "loss": 5.406,
+      "step": 4240
+    },
+    {
+      "epoch": 5.434782608695652,
+      "grad_norm": 1.5584670305252075,
+      "learning_rate": 0.0002519097261098437,
+      "loss": 5.4227,
+      "step": 4250
+    },
+    {
+      "epoch": 5.447570332480819,
+      "grad_norm": 1.0803555250167847,
+      "learning_rate": 0.00025168556396667693,
+      "loss": 5.4294,
+      "step": 4260
+    },
+    {
+      "epoch": 5.460358056265985,
+      "grad_norm": 1.1373729705810547,
+      "learning_rate": 0.0002514609808407558,
+      "loss": 5.401,
+      "step": 4270
+    },
+    {
+      "epoch": 5.4731457800511505,
+      "grad_norm": 0.9056215882301331,
+      "learning_rate": 0.00025123597766186454,
+      "loss": 5.4026,
+      "step": 4280
+    },
+    {
+      "epoch": 5.485933503836317,
+      "grad_norm": 1.3096917867660522,
+      "learning_rate": 0.00025101055536152616,
+      "loss": 5.4131,
+      "step": 4290
+    },
+    {
+      "epoch": 5.498721227621483,
+      "grad_norm": 0.890828013420105,
+      "learning_rate": 0.00025078471487299917,
+      "loss": 5.4063,
+      "step": 4300
+    },
+    {
+      "epoch": 5.5115089514066495,
+      "grad_norm": 0.9283934235572815,
+      "learning_rate": 0.00025055845713127306,
+      "loss": 5.4078,
+      "step": 4310
+    },
+    {
+      "epoch": 5.524296675191816,
+      "grad_norm": 0.9405550360679626,
+      "learning_rate": 0.00025033178307306506,
+      "loss": 5.4147,
+      "step": 4320
+    },
+    {
+      "epoch": 5.537084398976982,
+      "grad_norm": 0.8457754254341125,
+      "learning_rate": 0.0002501046936368158,
+      "loss": 5.408,
+      "step": 4330
+    },
+    {
+      "epoch": 5.549872122762149,
+      "grad_norm": 1.2794405221939087,
+      "learning_rate": 0.0002498771897626857,
+      "loss": 5.4104,
+      "step": 4340
+    },
+    {
+      "epoch": 5.562659846547315,
+      "grad_norm": 0.8734764456748962,
+      "learning_rate": 0.00024964927239255074,
+      "loss": 5.4089,
+      "step": 4350
+    },
+    {
+      "epoch": 5.57544757033248,
+      "grad_norm": 1.1535333395004272,
+      "learning_rate": 0.0002494209424699992,
+      "loss": 5.3985,
+      "step": 4360
+    },
+    {
+      "epoch": 5.588235294117647,
+      "grad_norm": 0.8653953671455383,
+      "learning_rate": 0.0002491922009403269,
+      "loss": 5.3734,
+      "step": 4370
+    },
+    {
+      "epoch": 5.601023017902813,
+      "grad_norm": 1.14579439163208,
+      "learning_rate": 0.0002489630487505341,
+      "loss": 5.3813,
+      "step": 4380
+    },
+    {
+      "epoch": 5.6138107416879794,
+      "grad_norm": 1.0796849727630615,
+      "learning_rate": 0.00024873348684932095,
+      "loss": 5.4106,
+      "step": 4390
+    },
+    {
+      "epoch": 5.626598465473146,
+      "grad_norm": 1.226593255996704,
+      "learning_rate": 0.000248503516187084,
+      "loss": 5.3917,
+      "step": 4400
+    },
+    {
+      "epoch": 5.639386189258312,
+      "grad_norm": 0.761320173740387,
+      "learning_rate": 0.00024827313771591183,
+      "loss": 5.3818,
+      "step": 4410
+    },
+    {
+      "epoch": 5.6521739130434785,
+      "grad_norm": 1.1446237564086914,
+      "learning_rate": 0.0002480423523895818,
+      "loss": 5.3926,
+      "step": 4420
+    },
+    {
+      "epoch": 5.664961636828645,
+      "grad_norm": 1.49809992313385,
+      "learning_rate": 0.00024781116116355523,
+      "loss": 5.4011,
+      "step": 4430
+    },
+    {
+      "epoch": 5.677749360613811,
+      "grad_norm": 0.9937270879745483,
+      "learning_rate": 0.00024757956499497414,
+      "loss": 5.4073,
+      "step": 4440
+    },
+    {
+      "epoch": 5.690537084398977,
+      "grad_norm": 1.3228845596313477,
+      "learning_rate": 0.000247347564842657,
+      "loss": 5.412,
+      "step": 4450
+    },
+    {
+      "epoch": 5.703324808184143,
+      "grad_norm": 1.146344542503357,
+      "learning_rate": 0.0002471151616670947,
+      "loss": 5.4071,
+      "step": 4460
+    },
+    {
+      "epoch": 5.716112531969309,
+      "grad_norm": 55.619258880615234,
+      "learning_rate": 0.00024688235643044676,
+      "loss": 5.7836,
+      "step": 4470
+    },
+    {
+      "epoch": 5.728900255754476,
+      "grad_norm": 2.3042633533477783,
+      "learning_rate": 0.0002466491500965372,
+      "loss": 5.6276,
+      "step": 4480
+    },
+    {
+      "epoch": 5.741687979539642,
+      "grad_norm": 2.298129081726074,
+      "learning_rate": 0.00024641554363085056,
+      "loss": 5.4675,
+      "step": 4490
+    },
+    {
+      "epoch": 5.754475703324808,
+      "grad_norm": 89.96639251708984,
+      "learning_rate": 0.00024618153800052796,
+      "loss": 5.478,
+      "step": 4500
+    },
+    {
+      "epoch": 5.767263427109975,
+      "grad_norm": 1.2469855546951294,
+      "learning_rate": 0.0002459471341743633,
+      "loss": 5.4764,
+      "step": 4510
+    },
+    {
+      "epoch": 5.78005115089514,
+      "grad_norm": 1.0728439092636108,
+      "learning_rate": 0.0002457123331227986,
+      "loss": 5.4405,
+      "step": 4520
+    },
+    {
+      "epoch": 5.792838874680307,
+      "grad_norm": 1.9817936420440674,
+      "learning_rate": 0.0002454771358179208,
+      "loss": 5.4187,
+      "step": 4530
+    },
+    {
+      "epoch": 5.805626598465473,
+      "grad_norm": 2.8526406288146973,
+      "learning_rate": 0.0002452415432334571,
+      "loss": 5.4591,
+      "step": 4540
+    },
+    {
+      "epoch": 5.818414322250639,
+      "grad_norm": 2.35715651512146,
+      "learning_rate": 0.0002450055563447714,
+      "loss": 5.4392,
+      "step": 4550
+    },
+    {
+      "epoch": 5.831202046035806,
+      "grad_norm": 1.693472981452942,
+      "learning_rate": 0.00024476917612885975,
+      "loss": 5.4059,
+      "step": 4560
+    },
+    {
+      "epoch": 5.843989769820972,
+      "grad_norm": 1.3062572479248047,
+      "learning_rate": 0.0002445324035643469,
+      "loss": 5.4088,
+      "step": 4570
+    },
+    {
+      "epoch": 5.856777493606138,
+      "grad_norm": 1.1204801797866821,
+      "learning_rate": 0.00024429523963148165,
+      "loss": 5.3992,
+      "step": 4580
+    },
+    {
+      "epoch": 5.869565217391305,
+      "grad_norm": 5.129584312438965,
+      "learning_rate": 0.00024405768531213337,
+      "loss": 5.3986,
+      "step": 4590
+    },
+    {
+      "epoch": 5.882352941176471,
+      "grad_norm": 162.03988647460938,
+      "learning_rate": 0.0002438197415897874,
+      "loss": 5.489,
+      "step": 4600
+    },
+    {
+      "epoch": 5.8951406649616365,
+      "grad_norm": 48338.75390625,
+      "learning_rate": 0.00024358140944954134,
+      "loss": 6.4544,
+      "step": 4610
+    },
+    {
+      "epoch": 5.907928388746803,
+      "grad_norm": 309.1040344238281,
+      "learning_rate": 0.00024334268987810087,
+      "loss": 6.5122,
+      "step": 4620
+    },
+    {
+      "epoch": 5.920716112531969,
+      "grad_norm": 91.73091125488281,
+      "learning_rate": 0.00024310358386377563,
+      "loss": 6.1042,
+      "step": 4630
+    },
+    {
+      "epoch": 5.9335038363171355,
+      "grad_norm": 4794.84912109375,
+      "learning_rate": 0.0002428640923964751,
+      "loss": 5.9236,
+      "step": 4640
+    },
+    {
+      "epoch": 5.946291560102302,
+      "grad_norm": 428.7879638671875,
+      "learning_rate": 0.00024262421646770461,
+      "loss": 6.3916,
+      "step": 4650
+    },
+    {
+      "epoch": 5.959079283887468,
+      "grad_norm": 2455.22900390625,
+      "learning_rate": 0.00024238395707056122,
+      "loss": 6.2707,
+      "step": 4660
+    },
+    {
+      "epoch": 5.971867007672635,
+      "grad_norm": 5.497981548309326,
+      "learning_rate": 0.00024214331519972944,
+      "loss": 5.9861,
+      "step": 4670
+    },
+    {
+      "epoch": 5.9846547314578,
+      "grad_norm": 7.887845993041992,
+      "learning_rate": 0.00024190229185147734,
+      "loss": 5.6179,
+      "step": 4680
+    },
+    {
+      "epoch": 5.997442455242966,
+      "grad_norm": 8.13206958770752,
+      "learning_rate": 0.00024166088802365232,
+      "loss": 5.5614,
+      "step": 4690
+    },
+    {
+      "epoch": 6.010230179028133,
+      "grad_norm": 170.85169982910156,
+      "learning_rate": 0.00024141910471567695,
+      "loss": 5.703,
+      "step": 4700
+    },
+    {
+      "epoch": 6.023017902813299,
+      "grad_norm": 8.114191055297852,
+      "learning_rate": 0.00024117694292854482,
+      "loss": 5.6643,
+      "step": 4710
+    },
+    {
+      "epoch": 6.035805626598465,
+      "grad_norm": 1.649259328842163,
+      "learning_rate": 0.00024093440366481655,
+      "loss": 5.5548,
+      "step": 4720
+    },
+    {
+      "epoch": 6.048593350383632,
+      "grad_norm": 3.6291794776916504,
+      "learning_rate": 0.0002406914879286155,
+      "loss": 5.5627,
+      "step": 4730
+    },
+    {
+      "epoch": 6.061381074168798,
+      "grad_norm": 17.16327476501465,
+      "learning_rate": 0.00024044819672562352,
+      "loss": 5.5179,
+      "step": 4740
+    },
+    {
+      "epoch": 6.0741687979539645,
+      "grad_norm": 66.84463500976562,
+      "learning_rate": 0.00024020453106307705,
+      "loss": 5.6356,
+      "step": 4750
+    },
+    {
+      "epoch": 6.086956521739131,
+      "grad_norm": 71.15001678466797,
+      "learning_rate": 0.00023996049194976272,
+      "loss": 5.8183,
+      "step": 4760
+    },
+    {
+      "epoch": 6.099744245524296,
+      "grad_norm": 215.5955352783203,
+      "learning_rate": 0.00023971608039601342,
+      "loss": 5.7444,
+      "step": 4770
+    },
+    {
+      "epoch": 6.112531969309463,
+      "grad_norm": 22.749778747558594,
+      "learning_rate": 0.00023947129741370372,
+      "loss": 5.6429,
+      "step": 4780
+    },
+    {
+      "epoch": 6.125319693094629,
+      "grad_norm": 4.4116387367248535,
+      "learning_rate": 0.0002392261440162461,
+      "loss": 5.5824,
+      "step": 4790
+    },
+    {
+      "epoch": 6.138107416879795,
+      "grad_norm": 21.407913208007812,
+      "learning_rate": 0.00023898062121858657,
+      "loss": 5.5583,
+      "step": 4800
+    },
+    {
+      "epoch": 6.150895140664962,
+      "grad_norm": 2.962198257446289,
+      "learning_rate": 0.0002387347300372004,
+      "loss": 5.5477,
+      "step": 4810
+    },
+    {
+      "epoch": 6.163682864450128,
+      "grad_norm": 2.31837797164917,
+      "learning_rate": 0.0002384884714900879,
+      "loss": 5.5348,
+      "step": 4820
+    },
+    {
+      "epoch": 6.176470588235294,
+      "grad_norm": 1.527478575706482,
+      "learning_rate": 0.00023824184659677054,
+      "loss": 5.5314,
+      "step": 4830
+    },
+    {
+      "epoch": 6.189258312020461,
+      "grad_norm": 61.507530212402344,
+      "learning_rate": 0.00023799485637828626,
+      "loss": 5.5663,
+      "step": 4840
+    },
+    {
+      "epoch": 6.202046035805626,
+      "grad_norm": 2436.508544921875,
+      "learning_rate": 0.0002377475018571856,
+      "loss": 5.6175,
+      "step": 4850
+    },
+    {
+      "epoch": 6.2148337595907925,
+      "grad_norm": 2913507.5,
+      "learning_rate": 0.00023749978405752722,
+      "loss": 6.377,
+      "step": 4860
+    },
+    {
+      "epoch": 6.227621483375959,
+      "grad_norm": 2641881.0,
+      "learning_rate": 0.00023725170400487386,
+      "loss": 7.9995,
+      "step": 4870
+    },
+    {
+      "epoch": 6.240409207161125,
+      "grad_norm": 159218.71875,
+      "learning_rate": 0.00023700326272628794,
+      "loss": 9.1041,
+      "step": 4880
+    },
+    {
+      "epoch": 6.253196930946292,
+      "grad_norm": 11773.8193359375,
+      "learning_rate": 0.00023675446125032736,
+      "loss": 8.7907,
+      "step": 4890
+    },
+    {
+      "epoch": 6.265984654731458,
+      "grad_norm": 1598409.75,
+      "learning_rate": 0.00023650530060704137,
+      "loss": 8.2848,
+      "step": 4900
+    },
+    {
+      "epoch": 6.278772378516624,
+      "grad_norm": 10316.607421875,
+      "learning_rate": 0.00023625578182796608,
+      "loss": 7.0858,
+      "step": 4910
+    },
+    {
+      "epoch": 6.291560102301791,
+      "grad_norm": 4661.24853515625,
+      "learning_rate": 0.00023600590594612025,
+      "loss": 6.5619,
+      "step": 4920
+    },
+    {
+      "epoch": 6.304347826086957,
+      "grad_norm": 50447.3359375,
+      "learning_rate": 0.00023575567399600124,
+      "loss": 6.3697,
+      "step": 4930
+    },
+    {
+      "epoch": 6.3171355498721224,
+      "grad_norm": 11574.3974609375,
+      "learning_rate": 0.0002355050870135804,
+      "loss": 6.4702,
+      "step": 4940
+    },
+    {
+      "epoch": 6.329923273657289,
+      "grad_norm": 1139.9498291015625,
+      "learning_rate": 0.00023525414603629897,
+      "loss": 6.3997,
+      "step": 4950
+    },
+    {
+      "epoch": 6.342710997442455,
+      "grad_norm": 571.5132446289062,
+      "learning_rate": 0.00023500285210306372,
+      "loss": 6.0743,
+      "step": 4960
+    },
+    {
+      "epoch": 6.3554987212276215,
+      "grad_norm": 242.6359100341797,
+      "learning_rate": 0.0002347512062542428,
+      "loss": 5.7139,
+      "step": 4970
+    },
+    {
+      "epoch": 6.368286445012788,
+      "grad_norm": 490.5594787597656,
+      "learning_rate": 0.00023449920953166114,
+      "loss": 5.787,
+      "step": 4980
+    },
+    {
+      "epoch": 6.381074168797954,
+      "grad_norm": 81.9707260131836,
+      "learning_rate": 0.00023424686297859637,
+      "loss": 5.8251,
+      "step": 4990
+    },
+    {
+      "epoch": 6.3938618925831205,
+      "grad_norm": 2.5100646018981934,
+      "learning_rate": 0.00023399416763977447,
+      "loss": 5.7931,
+      "step": 5000
+    },
+    {
+      "epoch": 6.406649616368286,
+      "grad_norm": 4.38476037979126,
+      "learning_rate": 0.0002337411245613653,
+      "loss": 5.6809,
+      "step": 5010
+    },
+    {
+      "epoch": 6.419437340153452,
+      "grad_norm": 0.9684484601020813,
+      "learning_rate": 0.00023348773479097855,
+      "loss": 5.6249,
+      "step": 5020
+    },
+    {
+      "epoch": 6.432225063938619,
+      "grad_norm": 2.0226662158966064,
+      "learning_rate": 0.00023323399937765906,
+      "loss": 5.5767,
+      "step": 5030
+    },
+    {
+      "epoch": 6.445012787723785,
+      "grad_norm": 1.4661270380020142,
+      "learning_rate": 0.0002329799193718827,
+      "loss": 5.5135,
+      "step": 5040
+    },
+    {
+      "epoch": 6.457800511508951,
+      "grad_norm": 1.4130449295043945,
+      "learning_rate": 0.00023272549582555212,
+      "loss": 5.5327,
+      "step": 5050
+    },
+    {
+      "epoch": 6.470588235294118,
+      "grad_norm": 1.2371689081192017,
+      "learning_rate": 0.00023247072979199192,
+      "loss": 5.4606,
+      "step": 5060
+    },
+    {
+      "epoch": 6.483375959079284,
+      "grad_norm": 0.9750532507896423,
+      "learning_rate": 0.000232215622325945,
+      "loss": 5.433,
+      "step": 5070
+    },
+    {
+      "epoch": 6.4961636828644505,
+      "grad_norm": 1.7443705797195435,
+      "learning_rate": 0.00023196017448356743,
+      "loss": 5.4655,
+      "step": 5080
+    },
+    {
+      "epoch": 6.508951406649617,
+      "grad_norm": 1.21770179271698,
+      "learning_rate": 0.00023170438732242473,
+      "loss": 5.4349,
+      "step": 5090
+    },
+    {
+      "epoch": 6.521739130434782,
+      "grad_norm": 1.1340687274932861,
+      "learning_rate": 0.0002314482619014871,
+      "loss": 5.4672,
+      "step": 5100
+    },
+    {
+      "epoch": 6.534526854219949,
+      "grad_norm": 2.9821395874023438,
+      "learning_rate": 0.0002311917992811252,
+      "loss": 5.4951,
+      "step": 5110
+    },
+    {
+      "epoch": 6.547314578005115,
+      "grad_norm": 199.1261749267578,
+      "learning_rate": 0.00023093500052310558,
+      "loss": 5.5208,
+      "step": 5120
+    },
+    {
+      "epoch": 6.560102301790281,
+      "grad_norm": 45.095088958740234,
+      "learning_rate": 0.0002306778666905866,
+      "loss": 5.4837,
+      "step": 5130
+    },
+    {
+      "epoch": 6.572890025575448,
+      "grad_norm": 2.1452138423919678,
+      "learning_rate": 0.00023042039884811372,
+      "loss": 5.4714,
+      "step": 5140
+    },
+    {
+      "epoch": 6.585677749360614,
+      "grad_norm": 61.150856018066406,
+      "learning_rate": 0.00023016259806161524,
+      "loss": 5.4805,
+      "step": 5150
+    },
+    {
+      "epoch": 6.59846547314578,
+      "grad_norm": 29.349306106567383,
+      "learning_rate": 0.00022990446539839795,
+      "loss": 5.5292,
+      "step": 5160
+    },
+    {
+      "epoch": 6.611253196930946,
+      "grad_norm": 51.63431930541992,
+      "learning_rate": 0.0002296460019271424,
+      "loss": 5.4861,
+      "step": 5170
+    },
+    {
+      "epoch": 6.624040920716112,
+      "grad_norm": 271368.5625,
+      "learning_rate": 0.000229387208717899,
+      "loss": 5.4857,
+      "step": 5180
+    },
+    {
+      "epoch": 6.6368286445012785,
+      "grad_norm": 22.639575958251953,
+      "learning_rate": 0.00022912808684208302,
+      "loss": 5.5158,
+      "step": 5190
+    },
+    {
+      "epoch": 6.649616368286445,
+      "grad_norm": 34.41466522216797,
+      "learning_rate": 0.00022886863737247068,
+      "loss": 5.477,
+      "step": 5200
+    },
+    {
+      "epoch": 6.662404092071611,
+      "grad_norm": 4.41986083984375,
+      "learning_rate": 0.0002286088613831942,
+      "loss": 5.4337,
+      "step": 5210
+    },
+    {
+      "epoch": 6.675191815856778,
+      "grad_norm": 3.1884968280792236,
+      "learning_rate": 0.00022834875994973773,
+      "loss": 5.5051,
+      "step": 5220
+    },
+    {
+      "epoch": 6.687979539641944,
+      "grad_norm": 1.9236191511154175,
+      "learning_rate": 0.00022808833414893282,
+      "loss": 5.457,
+      "step": 5230
+    },
+    {
+      "epoch": 6.70076726342711,
+      "grad_norm": 7.70404577255249,
+      "learning_rate": 0.0002278275850589538,
+      "loss": 5.4598,
+      "step": 5240
+    },
+    {
+      "epoch": 6.713554987212277,
+      "grad_norm": 5.475555419921875,
+      "learning_rate": 0.00022756651375931356,
+      "loss": 5.5004,
+      "step": 5250
+    },
+    {
+      "epoch": 6.726342710997442,
+      "grad_norm": 4.682950973510742,
+      "learning_rate": 0.00022730512133085885,
+      "loss": 5.4897,
+      "step": 5260
+    },
+    {
+      "epoch": 6.739130434782608,
+      "grad_norm": 2.2729697227478027,
+      "learning_rate": 0.00022704340885576588,
+      "loss": 5.5197,
+      "step": 5270
+    },
+    {
+      "epoch": 6.751918158567775,
+      "grad_norm": 1.674745798110962,
+      "learning_rate": 0.00022678137741753604,
+      "loss": 5.4519,
+      "step": 5280
+    },
+    {
+      "epoch": 6.764705882352941,
+      "grad_norm": 2.585331439971924,
+      "learning_rate": 0.00022651902810099107,
+      "loss": 5.4677,
+      "step": 5290
+    },
+    {
+      "epoch": 6.7774936061381075,
+      "grad_norm": 1.7059080600738525,
+      "learning_rate": 0.00022625636199226885,
+      "loss": 5.4422,
+      "step": 5300
+    },
+    {
+      "epoch": 6.790281329923274,
+      "grad_norm": 1.875679612159729,
+      "learning_rate": 0.00022599338017881865,
+      "loss": 5.4344,
+      "step": 5310
+    },
+    {
+      "epoch": 6.80306905370844,
+      "grad_norm": 1.07345449924469,
+      "learning_rate": 0.00022573008374939693,
+      "loss": 5.424,
+      "step": 5320
+    },
+    {
+      "epoch": 6.8158567774936065,
+      "grad_norm": 2.112574338912964,
+      "learning_rate": 0.0002254664737940626,
+      "loss": 5.392,
+      "step": 5330
+    },
+    {
+      "epoch": 6.828644501278772,
+      "grad_norm": 2.7668375968933105,
+      "learning_rate": 0.00022520255140417262,
+      "loss": 5.4379,
+      "step": 5340
+    },
+    {
+      "epoch": 6.841432225063938,
+      "grad_norm": 10.602677345275879,
+      "learning_rate": 0.00022493831767237736,
+      "loss": 5.4118,
+      "step": 5350
+    },
+    {
+      "epoch": 6.854219948849105,
+      "grad_norm": 2.519099712371826,
+      "learning_rate": 0.0002246737736926162,
+      "loss": 5.3876,
+      "step": 5360
+    },
+    {
+      "epoch": 6.867007672634271,
+      "grad_norm": 1.2932206392288208,
+      "learning_rate": 0.00022440892056011308,
+      "loss": 5.4044,
+      "step": 5370
+    },
+    {
+      "epoch": 6.879795396419437,
+      "grad_norm": 1.185011863708496,
+      "learning_rate": 0.00022414375937137164,
+      "loss": 5.4038,
+      "step": 5380
+    },
+    {
+      "epoch": 6.892583120204604,
+      "grad_norm": 1.340401530265808,
+      "learning_rate": 0.00022387829122417103,
+      "loss": 5.403,
+      "step": 5390
+    },
+    {
+      "epoch": 6.90537084398977,
+      "grad_norm": 1.3118840456008911,
+      "learning_rate": 0.0002236125172175612,
+      "loss": 5.407,
+      "step": 5400
+    },
+    {
+      "epoch": 6.918158567774936,
+      "grad_norm": 1.2288802862167358,
+      "learning_rate": 0.0002233464384518583,
+      "loss": 5.3923,
+      "step": 5410
+    },
+    {
+      "epoch": 6.930946291560103,
+      "grad_norm": 1.8922580480575562,
+      "learning_rate": 0.00022308005602864026,
+      "loss": 5.3871,
+      "step": 5420
+    },
+    {
+      "epoch": 6.943734015345268,
+      "grad_norm": 7.790588855743408,
+      "learning_rate": 0.00022281337105074214,
+      "loss": 5.4146,
+      "step": 5430
+    },
+    {
+      "epoch": 6.956521739130435,
+      "grad_norm": 4344.03271484375,
+      "learning_rate": 0.00022254638462225162,
+      "loss": 5.4527,
+      "step": 5440
+    },
+    {
+      "epoch": 6.969309462915601,
+      "grad_norm": 140.49502563476562,
+      "learning_rate": 0.00022227909784850437,
+      "loss": 5.6473,
+      "step": 5450
+    },
+    {
+      "epoch": 6.982097186700767,
+      "grad_norm": 130.85202026367188,
+      "learning_rate": 0.00022201151183607958,
+      "loss": 5.6701,
+      "step": 5460
+    },
+    {
+      "epoch": 6.994884910485934,
+      "grad_norm": 126.7076416015625,
+      "learning_rate": 0.0002217436276927951,
+      "loss": 5.788,
+      "step": 5470
+    },
+    {
+      "epoch": 7.0076726342711,
+      "grad_norm": 435.14361572265625,
+      "learning_rate": 0.00022147544652770332,
+      "loss": 5.906,
+      "step": 5480
+    },
+    {
+      "epoch": 7.020460358056266,
+      "grad_norm": 331.05145263671875,
+      "learning_rate": 0.0002212069694510861,
+      "loss": 5.7959,
+      "step": 5490
+    },
+    {
+      "epoch": 7.033248081841432,
+      "grad_norm": 24.31149673461914,
+      "learning_rate": 0.0002209381975744505,
+      "loss": 5.7055,
+      "step": 5500
+    },
+    {
+      "epoch": 7.046035805626598,
+      "grad_norm": 370.831298828125,
+      "learning_rate": 0.00022066913201052406,
+      "loss": 5.6893,
+      "step": 5510
+    },
+    {
+      "epoch": 7.0588235294117645,
+      "grad_norm": 80.24853515625,
+      "learning_rate": 0.00022039977387325012,
+      "loss": 5.6668,
+      "step": 5520
+    },
+    {
+      "epoch": 7.071611253196931,
+      "grad_norm": 26.689146041870117,
+      "learning_rate": 0.00022013012427778337,
+      "loss": 5.6272,
+      "step": 5530
+    },
+    {
+      "epoch": 7.084398976982097,
+      "grad_norm": 18.282222747802734,
+      "learning_rate": 0.00021986018434048515,
+      "loss": 5.6231,
+      "step": 5540
+    },
+    {
+      "epoch": 7.0971867007672635,
+      "grad_norm": 71.10526275634766,
+      "learning_rate": 0.00021958995517891872,
+      "loss": 5.6329,
+      "step": 5550
+    },
+    {
+      "epoch": 7.10997442455243,
+      "grad_norm": 66.7136459350586,
+      "learning_rate": 0.00021931943791184488,
+      "loss": 5.5997,
+      "step": 5560
+    },
+    {
+      "epoch": 7.122762148337596,
+      "grad_norm": 13.023746490478516,
+      "learning_rate": 0.00021904863365921715,
+      "loss": 5.585,
+      "step": 5570
+    },
+    {
+      "epoch": 7.135549872122763,
+      "grad_norm": 20.470502853393555,
+      "learning_rate": 0.0002187775435421771,
+      "loss": 5.6463,
+      "step": 5580
+    },
+    {
+      "epoch": 7.148337595907928,
+      "grad_norm": 308.2809753417969,
+      "learning_rate": 0.00021850616868305,
+      "loss": 5.6014,
+      "step": 5590
+    },
+    {
+      "epoch": 7.161125319693094,
+      "grad_norm": 2176.8095703125,
+      "learning_rate": 0.00021823451020533966,
+      "loss": 5.601,
+      "step": 5600
+    },
+    {
+      "epoch": 7.173913043478261,
+      "grad_norm": 14.34654712677002,
+      "learning_rate": 0.00021796256923372435,
+      "loss": 5.715,
+      "step": 5610
+    },
+    {
+      "epoch": 7.186700767263427,
+      "grad_norm": 4.329311847686768,
+      "learning_rate": 0.00021769034689405174,
+      "loss": 5.6829,
+      "step": 5620
+    },
+    {
+      "epoch": 7.1994884910485935,
+      "grad_norm": 6.315857410430908,
+      "learning_rate": 0.00021741784431333444,
+      "loss": 5.6157,
+      "step": 5630
+    },
+    {
+      "epoch": 7.21227621483376,
+      "grad_norm": 7.763189315795898,
+      "learning_rate": 0.00021714506261974527,
+      "loss": 5.6019,
+      "step": 5640
+    },
+    {
+      "epoch": 7.225063938618926,
+      "grad_norm": 375.8602294921875,
+      "learning_rate": 0.00021687200294261255,
+      "loss": 5.6288,
+      "step": 5650
+    },
+    {
+      "epoch": 7.2378516624040925,
+      "grad_norm": 16.852691650390625,
+      "learning_rate": 0.0002165986664124154,
+      "loss": 5.6608,
+      "step": 5660
+    },
+    {
+      "epoch": 7.250639386189258,
+      "grad_norm": 582.8134765625,
+      "learning_rate": 0.0002163250541607793,
+      "loss": 5.5801,
+      "step": 5670
+    },
+    {
+      "epoch": 7.263427109974424,
+      "grad_norm": 13.16250228881836,
+      "learning_rate": 0.00021605116732047113,
+      "loss": 5.5558,
+      "step": 5680
+    },
+    {
+      "epoch": 7.276214833759591,
+      "grad_norm": 81.9183120727539,
+      "learning_rate": 0.0002157770070253945,
+      "loss": 5.5759,
+      "step": 5690
+    },
+    {
+      "epoch": 7.289002557544757,
+      "grad_norm": 34.48521041870117,
+      "learning_rate": 0.0002155025744105853,
+      "loss": 5.5714,
+      "step": 5700
+    },
+    {
+      "epoch": 7.301790281329923,
+      "grad_norm": 5.397050857543945,
+      "learning_rate": 0.0002152278706122067,
+      "loss": 5.5447,
+      "step": 5710
+    },
+    {
+      "epoch": 7.31457800511509,
+      "grad_norm": 7.5026068687438965,
+      "learning_rate": 0.0002149528967675447,
+      "loss": 5.5271,
+      "step": 5720
+    },
+    {
+      "epoch": 7.327365728900256,
+      "grad_norm": 21.039640426635742,
+      "learning_rate": 0.00021467765401500316,
+      "loss": 5.5523,
+      "step": 5730
+    },
+    {
+      "epoch": 7.340153452685422,
+      "grad_norm": 3.5831055641174316,
+      "learning_rate": 0.00021440214349409937,
+      "loss": 5.5607,
+      "step": 5740
+    },
+    {
+      "epoch": 7.352941176470588,
+      "grad_norm": 2.3747684955596924,
+      "learning_rate": 0.0002141263663454591,
+      "loss": 5.5686,
+      "step": 5750
+    },
+    {
+      "epoch": 7.365728900255754,
+      "grad_norm": 3.3695812225341797,
+      "learning_rate": 0.00021385032371081197,
+      "loss": 5.5386,
+      "step": 5760
+    },
+    {
+      "epoch": 7.378516624040921,
+      "grad_norm": 1.6819612979888916,
+      "learning_rate": 0.00021357401673298686,
+      "loss": 5.4853,
+      "step": 5770
+    },
+    {
+      "epoch": 7.391304347826087,
+      "grad_norm": 2.8821001052856445,
+      "learning_rate": 0.00021329744655590685,
+      "loss": 5.4891,
+      "step": 5780
+    },
+    {
+      "epoch": 7.404092071611253,
+      "grad_norm": 10.89107894897461,
+      "learning_rate": 0.00021302061432458476,
+      "loss": 5.5159,
+      "step": 5790
+    },
+    {
+      "epoch": 7.41687979539642,
+      "grad_norm": 8.90987777709961,
+      "learning_rate": 0.00021274352118511838,
+      "loss": 5.4912,
+      "step": 5800
+    },
+    {
+      "epoch": 7.429667519181586,
+      "grad_norm": 2.6703226566314697,
+      "learning_rate": 0.00021246616828468563,
+      "loss": 5.4594,
+      "step": 5810
+    },
+    {
+      "epoch": 7.442455242966752,
+      "grad_norm": 8.18791389465332,
+      "learning_rate": 0.0002121885567715398,
+      "loss": 5.4581,
+      "step": 5820
+    },
+    {
+      "epoch": 7.455242966751918,
+      "grad_norm": 6.304783344268799,
+      "learning_rate": 0.00021191068779500495,
+      "loss": 5.4296,
+      "step": 5830
+    },
+    {
+      "epoch": 7.468030690537084,
+      "grad_norm": 81.46858215332031,
+      "learning_rate": 0.0002116325625054709,
+      "loss": 5.4251,
+      "step": 5840
+    },
+    {
+      "epoch": 7.4808184143222505,
+      "grad_norm": 36.4720344543457,
+      "learning_rate": 0.00021135418205438879,
+      "loss": 5.4542,
+      "step": 5850
+    },
+    {
+      "epoch": 7.493606138107417,
+      "grad_norm": 2.27168345451355,
+      "learning_rate": 0.00021107554759426601,
+      "loss": 5.5033,
+      "step": 5860
+    },
+    {
+      "epoch": 7.506393861892583,
+      "grad_norm": 2.0624887943267822,
+      "learning_rate": 0.00021079666027866165,
+      "loss": 5.4607,
+      "step": 5870
+    },
+    {
+      "epoch": 7.5191815856777495,
+      "grad_norm": 1.2841031551361084,
+      "learning_rate": 0.0002105175212621816,
+      "loss": 5.4534,
+      "step": 5880
+    },
+    {
+      "epoch": 7.531969309462916,
+      "grad_norm": 1.7171348333358765,
+      "learning_rate": 0.00021023813170047378,
+      "loss": 5.4251,
+      "step": 5890
+    },
+    {
+      "epoch": 7.544757033248082,
+      "grad_norm": 2.006608247756958,
+      "learning_rate": 0.00020995849275022338,
+      "loss": 5.4303,
+      "step": 5900
+    },
+    {
+      "epoch": 7.557544757033249,
+      "grad_norm": 1.1459277868270874,
+      "learning_rate": 0.00020967860556914808,
+      "loss": 5.4326,
+      "step": 5910
+    },
+    {
+      "epoch": 7.570332480818414,
+      "grad_norm": 2.0150973796844482,
+      "learning_rate": 0.00020939847131599325,
+      "loss": 5.4058,
+      "step": 5920
+    },
+    {
+      "epoch": 7.58312020460358,
+      "grad_norm": 1.5685169696807861,
+      "learning_rate": 0.00020911809115052718,
+      "loss": 5.4017,
+      "step": 5930
+    },
+    {
+      "epoch": 7.595907928388747,
+      "grad_norm": 4.461862564086914,
+      "learning_rate": 0.00020883746623353615,
+      "loss": 5.4142,
+      "step": 5940
+    },
+    {
+      "epoch": 7.608695652173913,
+      "grad_norm": 3.8895885944366455,
+      "learning_rate": 0.00020855659772681988,
+      "loss": 5.4473,
+      "step": 5950
+    },
+    {
+      "epoch": 7.621483375959079,
+      "grad_norm": 26.799531936645508,
+      "learning_rate": 0.0002082754867931864,
+      "loss": 5.3851,
+      "step": 5960
+    },
+    {
+      "epoch": 7.634271099744246,
+      "grad_norm": 28.458995819091797,
+      "learning_rate": 0.00020799413459644746,
+      "loss": 5.404,
+      "step": 5970
+    },
+    {
+      "epoch": 7.647058823529412,
+      "grad_norm": 5.39988374710083,
+      "learning_rate": 0.00020771254230141374,
+      "loss": 5.4142,
+      "step": 5980
+    },
+    {
+      "epoch": 7.659846547314578,
+      "grad_norm": 24.639158248901367,
+      "learning_rate": 0.00020743071107388978,
+      "loss": 5.4414,
+      "step": 5990
+    },
+    {
+      "epoch": 7.672634271099744,
+      "grad_norm": 31.065847396850586,
+      "learning_rate": 0.00020714864208066944,
+      "loss": 5.4642,
+      "step": 6000
+    },
+    {
+      "epoch": 7.68542199488491,
+      "grad_norm": 3.8758230209350586,
+      "learning_rate": 0.00020686633648953087,
+      "loss": 5.4475,
+      "step": 6010
+    },
+    {
+      "epoch": 7.698209718670077,
+      "grad_norm": 1.0469917058944702,
+      "learning_rate": 0.00020658379546923173,
+      "loss": 5.4207,
+      "step": 6020
+    },
+    {
+      "epoch": 7.710997442455243,
+      "grad_norm": 1.1670767068862915,
+      "learning_rate": 0.00020630102018950446,
+      "loss": 5.388,
+      "step": 6030
+    },
+    {
+      "epoch": 7.723785166240409,
+      "grad_norm": 1.3517999649047852,
+      "learning_rate": 0.00020601801182105128,
+      "loss": 5.4072,
+      "step": 6040
+    },
+    {
+      "epoch": 7.736572890025576,
+      "grad_norm": 1.2495683431625366,
+      "learning_rate": 0.00020573477153553932,
+      "loss": 5.363,
+      "step": 6050
+    },
+    {
+      "epoch": 7.749360613810742,
+      "grad_norm": 2.0349299907684326,
+      "learning_rate": 0.0002054513005055961,
+      "loss": 5.3792,
+      "step": 6060
+    },
+    {
+      "epoch": 7.762148337595908,
+      "grad_norm": 417.0843811035156,
+      "learning_rate": 0.0002051675999048042,
+      "loss": 5.4068,
+      "step": 6070
+    },
+    {
+      "epoch": 7.774936061381074,
+      "grad_norm": 1.2949700355529785,
+      "learning_rate": 0.0002048836709076967,
+      "loss": 5.4034,
+      "step": 6080
+    },
+    {
+      "epoch": 7.78772378516624,
+      "grad_norm": 1.1051993370056152,
+      "learning_rate": 0.00020459951468975227,
+      "loss": 5.4037,
+      "step": 6090
+    },
+    {
+      "epoch": 7.8005115089514065,
+      "grad_norm": 1.6521767377853394,
+      "learning_rate": 0.00020431513242739034,
+      "loss": 5.377,
+      "step": 6100
+    },
+    {
+      "epoch": 7.813299232736573,
+      "grad_norm": 1.1317176818847656,
+      "learning_rate": 0.00020403052529796605,
+      "loss": 5.3547,
+      "step": 6110
+    },
+    {
+      "epoch": 7.826086956521739,
+      "grad_norm": 1.4219192266464233,
+      "learning_rate": 0.00020374569447976563,
+      "loss": 5.3507,
+      "step": 6120
+    },
+    {
+      "epoch": 7.838874680306906,
+      "grad_norm": 154.09230041503906,
+      "learning_rate": 0.0002034606411520013,
+      "loss": 5.355,
+      "step": 6130
+    },
+    {
+      "epoch": 7.851662404092072,
+      "grad_norm": 1.127315878868103,
+      "learning_rate": 0.00020317536649480645,
+      "loss": 5.3597,
+      "step": 6140
+    },
+    {
+      "epoch": 7.864450127877237,
+      "grad_norm": 2.925020933151245,
+      "learning_rate": 0.0002028898716892309,
+      "loss": 5.3633,
+      "step": 6150
+    },
+    {
+      "epoch": 7.877237851662404,
+      "grad_norm": 1.464866280555725,
+      "learning_rate": 0.00020260415791723582,
+      "loss": 5.3502,
+      "step": 6160
+    },
+    {
+      "epoch": 7.89002557544757,
+      "grad_norm": 1.2165534496307373,
+      "learning_rate": 0.00020231822636168894,
+      "loss": 5.3366,
+      "step": 6170
+    },
+    {
+      "epoch": 7.9028132992327365,
+      "grad_norm": 1.153878092765808,
+      "learning_rate": 0.0002020320782063596,
+      "loss": 5.373,
+      "step": 6180
+    },
+    {
+      "epoch": 7.915601023017903,
+      "grad_norm": 3.499431848526001,
+      "learning_rate": 0.00020174571463591387,
+      "loss": 5.3212,
+      "step": 6190
+    },
+    {
+      "epoch": 7.928388746803069,
+      "grad_norm": 1.6055837869644165,
+      "learning_rate": 0.00020145913683590965,
+      "loss": 5.3329,
+      "step": 6200
+    },
+    {
+      "epoch": 7.9411764705882355,
+      "grad_norm": 20.313854217529297,
+      "learning_rate": 0.00020117234599279177,
+      "loss": 5.3892,
+      "step": 6210
+    },
+    {
+      "epoch": 7.953964194373402,
+      "grad_norm": 1047.665283203125,
+      "learning_rate": 0.00020088534329388712,
+      "loss": 5.4904,
+      "step": 6220
+    },
+    {
+      "epoch": 7.966751918158568,
+      "grad_norm": 199.46250915527344,
+      "learning_rate": 0.00020059812992739956,
+      "loss": 5.4668,
+      "step": 6230
+    },
+    {
+      "epoch": 7.979539641943734,
+      "grad_norm": 77.22345733642578,
+      "learning_rate": 0.00020031070708240525,
+      "loss": 5.447,
+      "step": 6240
+    },
+    {
+      "epoch": 7.9923273657289,
+      "grad_norm": 2511.667236328125,
+      "learning_rate": 0.0002000230759488475,
+      "loss": 5.4581,
+      "step": 6250
+    },
+    {
+      "epoch": 8.005115089514067,
+      "grad_norm": 964.1087646484375,
+      "learning_rate": 0.00019973523771753203,
+      "loss": 5.4332,
+      "step": 6260
+    },
+    {
+      "epoch": 8.017902813299234,
+      "grad_norm": 1089.00146484375,
+      "learning_rate": 0.00019944719358012188,
+      "loss": 5.4214,
+      "step": 6270
+    },
+    {
+      "epoch": 8.030690537084398,
+      "grad_norm": 177.26681518554688,
+      "learning_rate": 0.00019915894472913258,
+      "loss": 5.4031,
+      "step": 6280
+    },
+    {
+      "epoch": 8.043478260869565,
+      "grad_norm": 37.300941467285156,
+      "learning_rate": 0.00019887049235792716,
+      "loss": 5.3803,
+      "step": 6290
+    },
+    {
+      "epoch": 8.05626598465473,
+      "grad_norm": 41.14563751220703,
+      "learning_rate": 0.00019858183766071133,
+      "loss": 5.4269,
+      "step": 6300
+    },
+    {
+      "epoch": 8.069053708439897,
+      "grad_norm": 1573.4498291015625,
+      "learning_rate": 0.00019829298183252832,
+      "loss": 5.3899,
+      "step": 6310
+    },
+    {
+      "epoch": 8.081841432225064,
+      "grad_norm": 90.4874038696289,
+      "learning_rate": 0.00019800392606925409,
+      "loss": 5.4205,
+      "step": 6320
+    },
+    {
+      "epoch": 8.09462915601023,
+      "grad_norm": 87.13722229003906,
+      "learning_rate": 0.00019771467156759235,
+      "loss": 5.4472,
+      "step": 6330
+    },
+    {
+      "epoch": 8.107416879795396,
+      "grad_norm": 69.13899230957031,
+      "learning_rate": 0.0001974252195250696,
+      "loss": 5.4954,
+      "step": 6340
+    },
+    {
+      "epoch": 8.120204603580563,
+      "grad_norm": 64.28082275390625,
+      "learning_rate": 0.00019713557114003018,
+      "loss": 5.532,
+      "step": 6350
+    },
+    {
+      "epoch": 8.132992327365729,
+      "grad_norm": 6390.68896484375,
+      "learning_rate": 0.00019684572761163125,
+      "loss": 5.4868,
+      "step": 6360
+    },
+    {
+      "epoch": 8.145780051150895,
+      "grad_norm": 4.564518928527832,
+      "learning_rate": 0.00019655569013983797,
+      "loss": 5.4988,
+      "step": 6370
+    },
+    {
+      "epoch": 8.158567774936062,
+      "grad_norm": 5.2263665199279785,
+      "learning_rate": 0.00019626545992541824,
+      "loss": 5.5725,
+      "step": 6380
+    },
+    {
+      "epoch": 8.171355498721228,
+      "grad_norm": 4.126269340515137,
+      "learning_rate": 0.00019597503816993808,
+      "loss": 5.5511,
+      "step": 6390
+    },
+    {
+      "epoch": 8.184143222506394,
+      "grad_norm": 13.468574523925781,
+      "learning_rate": 0.00019568442607575655,
+      "loss": 5.4787,
+      "step": 6400
+    },
+    {
+      "epoch": 8.19693094629156,
+      "grad_norm": 10.518543243408203,
+      "learning_rate": 0.00019539362484602058,
+      "loss": 5.5517,
+      "step": 6410
+    },
+    {
+      "epoch": 8.209718670076727,
+      "grad_norm": 49.399635314941406,
+      "learning_rate": 0.00019510263568466016,
+      "loss": 5.478,
+      "step": 6420
+    },
+    {
+      "epoch": 8.222506393861893,
+      "grad_norm": 193.64840698242188,
+      "learning_rate": 0.00019481145979638342,
+      "loss": 5.457,
+      "step": 6430
+    },
+    {
+      "epoch": 8.235294117647058,
+      "grad_norm": 317.044189453125,
+      "learning_rate": 0.00019452009838667142,
+      "loss": 5.4926,
+      "step": 6440
+    },
+    {
+      "epoch": 8.248081841432224,
+      "grad_norm": 11.334222793579102,
+      "learning_rate": 0.00019422855266177337,
+      "loss": 5.5006,
+      "step": 6450
+    },
+    {
+      "epoch": 8.26086956521739,
+      "grad_norm": 46.2332649230957,
+      "learning_rate": 0.00019393682382870151,
+      "loss": 5.469,
+      "step": 6460
+    },
+    {
+      "epoch": 8.273657289002557,
+      "grad_norm": 15.786870956420898,
+      "learning_rate": 0.0001936449130952262,
+      "loss": 5.4381,
+      "step": 6470
+    },
+    {
+      "epoch": 8.286445012787723,
+      "grad_norm": 52.73589324951172,
+      "learning_rate": 0.00019335282166987083,
+      "loss": 5.4408,
+      "step": 6480
+    },
+    {
+      "epoch": 8.29923273657289,
+      "grad_norm": 61.6315803527832,
+      "learning_rate": 0.00019306055076190692,
+      "loss": 5.4318,
+      "step": 6490
+    },
+    {
+      "epoch": 8.312020460358056,
+      "grad_norm": 12.02227783203125,
+      "learning_rate": 0.000192768101581349,
+      "loss": 5.4227,
+      "step": 6500
+    },
+    {
+      "epoch": 8.324808184143222,
+      "grad_norm": 15.120413780212402,
+      "learning_rate": 0.0001924754753389496,
+      "loss": 5.5355,
+      "step": 6510
+    },
+    {
+      "epoch": 8.337595907928389,
+      "grad_norm": 40.40769958496094,
+      "learning_rate": 0.0001921826732461945,
+      "loss": 5.4947,
+      "step": 6520
+    },
+    {
+      "epoch": 8.350383631713555,
+      "grad_norm": 84.49889373779297,
+      "learning_rate": 0.0001918896965152973,
+      "loss": 5.4894,
+      "step": 6530
+    },
+    {
+      "epoch": 8.363171355498721,
+      "grad_norm": 10.641755104064941,
+      "learning_rate": 0.00019159654635919472,
+      "loss": 5.6015,
+      "step": 6540
+    },
+    {
+      "epoch": 8.375959079283888,
+      "grad_norm": 5.664783000946045,
+      "learning_rate": 0.00019130322399154144,
+      "loss": 5.4883,
+      "step": 6550
+    },
+    {
+      "epoch": 8.388746803069054,
+      "grad_norm": 2.0645973682403564,
+      "learning_rate": 0.00019100973062670504,
+      "loss": 5.4584,
+      "step": 6560
+    },
+    {
+      "epoch": 8.40153452685422,
+      "grad_norm": 3.2000749111175537,
+      "learning_rate": 0.00019071606747976113,
+      "loss": 5.4612,
+      "step": 6570
+    },
+    {
+      "epoch": 8.414322250639387,
+      "grad_norm": 5.516482830047607,
+      "learning_rate": 0.00019042223576648822,
+      "loss": 5.4622,
+      "step": 6580
+    },
+    {
+      "epoch": 8.427109974424553,
+      "grad_norm": 1.833092451095581,
+      "learning_rate": 0.00019012823670336258,
+      "loss": 5.4434,
+      "step": 6590
+    },
+    {
+      "epoch": 8.43989769820972,
+      "grad_norm": 1.5939054489135742,
+      "learning_rate": 0.00018983407150755349,
+      "loss": 5.4028,
+      "step": 6600
+    },
+    {
+      "epoch": 8.452685421994884,
+      "grad_norm": 2.889866352081299,
+      "learning_rate": 0.00018953974139691794,
+      "loss": 5.5098,
+      "step": 6610
+    },
+    {
+      "epoch": 8.46547314578005,
+      "grad_norm": 12.796149253845215,
+      "learning_rate": 0.0001892452475899956,
+      "loss": 5.4479,
+      "step": 6620
+    },
+    {
+      "epoch": 8.478260869565217,
+      "grad_norm": 11.244743347167969,
+      "learning_rate": 0.00018895059130600396,
+      "loss": 5.4367,
+      "step": 6630
+    },
+    {
+      "epoch": 8.491048593350383,
+      "grad_norm": 6.546978950500488,
+      "learning_rate": 0.0001886557737648331,
+      "loss": 5.3932,
+      "step": 6640
+    },
+    {
+      "epoch": 8.50383631713555,
+      "grad_norm": 50.4334831237793,
+      "learning_rate": 0.00018836079618704074,
+      "loss": 5.4069,
+      "step": 6650
+    },
+    {
+      "epoch": 8.516624040920716,
+      "grad_norm": 7.810596466064453,
+      "learning_rate": 0.00018806565979384725,
+      "loss": 5.4757,
+      "step": 6660
+    },
+    {
+      "epoch": 8.529411764705882,
+      "grad_norm": 1.919179081916809,
+      "learning_rate": 0.0001877703658071303,
+      "loss": 5.398,
+      "step": 6670
+    },
+    {
+      "epoch": 8.542199488491049,
+      "grad_norm": 2.1735100746154785,
+      "learning_rate": 0.0001874749154494202,
+      "loss": 5.3852,
+      "step": 6680
+    },
+    {
+      "epoch": 8.554987212276215,
+      "grad_norm": 1.4280660152435303,
+      "learning_rate": 0.00018717930994389452,
+      "loss": 5.368,
+      "step": 6690
+    },
+    {
+      "epoch": 8.567774936061381,
+      "grad_norm": 0.9073035717010498,
+      "learning_rate": 0.00018688355051437323,
+      "loss": 5.3647,
+      "step": 6700
+    },
+    {
+      "epoch": 8.580562659846548,
+      "grad_norm": 0.9486181139945984,
+      "learning_rate": 0.00018658763838531346,
+      "loss": 5.3562,
+      "step": 6710
+    },
+    {
+      "epoch": 8.593350383631714,
+      "grad_norm": 2.596919059753418,
+      "learning_rate": 0.0001862915747818046,
+      "loss": 5.329,
+      "step": 6720
+    },
+    {
+      "epoch": 8.60613810741688,
+      "grad_norm": 1.398964762687683,
+      "learning_rate": 0.0001859953609295632,
+      "loss": 5.3702,
+      "step": 6730
+    },
+    {
+      "epoch": 8.618925831202047,
+      "grad_norm": 1.2070012092590332,
+      "learning_rate": 0.00018569899805492764,
+      "loss": 5.3389,
+      "step": 6740
+    },
+    {
+      "epoch": 8.631713554987213,
+      "grad_norm": 3.92047381401062,
+      "learning_rate": 0.00018540248738485346,
+      "loss": 5.3392,
+      "step": 6750
+    },
+    {
+      "epoch": 8.644501278772378,
+      "grad_norm": 0.8959820866584778,
+      "learning_rate": 0.00018510583014690805,
+      "loss": 5.3569,
+      "step": 6760
+    },
+    {
+      "epoch": 8.657289002557544,
+      "grad_norm": 0.838767409324646,
+      "learning_rate": 0.0001848090275692655,
+      "loss": 5.3327,
+      "step": 6770
+    },
+    {
+      "epoch": 8.67007672634271,
+      "grad_norm": 1.3687005043029785,
+      "learning_rate": 0.00018451208088070172,
+      "loss": 5.3473,
+      "step": 6780
+    },
+    {
+      "epoch": 8.682864450127877,
+      "grad_norm": 1.2928316593170166,
+      "learning_rate": 0.00018421499131058918,
+      "loss": 5.3158,
+      "step": 6790
+    },
+    {
+      "epoch": 8.695652173913043,
+      "grad_norm": 1.4481250047683716,
+      "learning_rate": 0.0001839177600888919,
+      "loss": 5.3533,
+      "step": 6800
+    },
+    {
+      "epoch": 8.70843989769821,
+      "grad_norm": 1.228041172027588,
+      "learning_rate": 0.0001836203884461603,
+      "loss": 5.331,
+      "step": 6810
+    },
+    {
+      "epoch": 8.721227621483376,
+      "grad_norm": 1.7305997610092163,
+      "learning_rate": 0.00018332287761352632,
+      "loss": 5.3389,
+      "step": 6820
+    },
+    {
+      "epoch": 8.734015345268542,
+      "grad_norm": 1.4301363229751587,
+      "learning_rate": 0.0001830252288226979,
+      "loss": 5.3091,
+      "step": 6830
+    },
+    {
+      "epoch": 8.746803069053708,
+      "grad_norm": 1.5935139656066895,
+      "learning_rate": 0.00018272744330595432,
+      "loss": 5.3222,
+      "step": 6840
+    },
+    {
+      "epoch": 8.759590792838875,
+      "grad_norm": 4.215871810913086,
+      "learning_rate": 0.00018242952229614092,
+      "loss": 5.3164,
+      "step": 6850
+    },
+    {
+      "epoch": 8.772378516624041,
+      "grad_norm": 3.8628363609313965,
+      "learning_rate": 0.0001821314670266638,
+      "loss": 5.3282,
+      "step": 6860
+    },
+    {
+      "epoch": 8.785166240409207,
+      "grad_norm": 5.04424524307251,
+      "learning_rate": 0.00018183327873148508,
+      "loss": 5.3065,
+      "step": 6870
+    },
+    {
+      "epoch": 8.797953964194374,
+      "grad_norm": 2.324964761734009,
+      "learning_rate": 0.00018153495864511757,
+      "loss": 5.3322,
+      "step": 6880
+    },
+    {
+      "epoch": 8.81074168797954,
+      "grad_norm": 0.8637028932571411,
+      "learning_rate": 0.00018123650800261966,
+      "loss": 5.3113,
+      "step": 6890
+    },
+    {
+      "epoch": 8.823529411764707,
+      "grad_norm": 2.1809170246124268,
+      "learning_rate": 0.0001809379280395903,
+      "loss": 5.2999,
+      "step": 6900
+    },
+    {
+      "epoch": 8.836317135549873,
+      "grad_norm": 0.9374740719795227,
+      "learning_rate": 0.00018063921999216374,
+      "loss": 5.3243,
+      "step": 6910
+    },
+    {
+      "epoch": 8.84910485933504,
+      "grad_norm": 1.3672325611114502,
+      "learning_rate": 0.00018034038509700468,
+      "loss": 5.3473,
+      "step": 6920
+    },
+    {
+      "epoch": 8.861892583120204,
+      "grad_norm": 1.0719919204711914,
+      "learning_rate": 0.00018004142459130282,
+      "loss": 5.3407,
+      "step": 6930
+    },
+    {
+      "epoch": 8.87468030690537,
+      "grad_norm": 0.8660845160484314,
+      "learning_rate": 0.00017974233971276791,
+      "loss": 5.3093,
+      "step": 6940
+    },
+    {
+      "epoch": 8.887468030690536,
+      "grad_norm": 1.8442758321762085,
+      "learning_rate": 0.0001794431316996247,
+      "loss": 5.3256,
+      "step": 6950
+    },
+    {
+      "epoch": 8.900255754475703,
+      "grad_norm": 1.3475013971328735,
+      "learning_rate": 0.00017914380179060756,
+      "loss": 5.3093,
+      "step": 6960
+    },
+    {
+      "epoch": 8.91304347826087,
+      "grad_norm": 1.1749337911605835,
+      "learning_rate": 0.00017884435122495575,
+      "loss": 5.3207,
+      "step": 6970
+    },
+    {
+      "epoch": 8.925831202046036,
+      "grad_norm": 1.2629647254943848,
+      "learning_rate": 0.00017854478124240783,
+      "loss": 5.2829,
+      "step": 6980
+    },
+    {
+      "epoch": 8.938618925831202,
+      "grad_norm": 1.7222191095352173,
+      "learning_rate": 0.00017824509308319688,
+      "loss": 5.3143,
+      "step": 6990
+    },
+    {
+      "epoch": 8.951406649616368,
+      "grad_norm": 14.291260719299316,
+      "learning_rate": 0.00017794528798804519,
+      "loss": 5.3009,
+      "step": 7000
+    },
+    {
+      "epoch": 8.964194373401535,
+      "grad_norm": 1.296583652496338,
+      "learning_rate": 0.00017764536719815918,
+      "loss": 5.3158,
+      "step": 7010
+    },
+    {
+      "epoch": 8.976982097186701,
+      "grad_norm": 1.0986963510513306,
+      "learning_rate": 0.00017734533195522424,
+      "loss": 5.3068,
+      "step": 7020
+    },
+    {
+      "epoch": 8.989769820971867,
+      "grad_norm": 1.0342364311218262,
+      "learning_rate": 0.00017704518350139965,
+      "loss": 5.2997,
+      "step": 7030
+    },
+    {
+      "epoch": 9.002557544757034,
+      "grad_norm": 1.445008635520935,
+      "learning_rate": 0.0001767449230793133,
+      "loss": 5.3086,
+      "step": 7040
+    },
+    {
+      "epoch": 9.0153452685422,
+      "grad_norm": 1.7098116874694824,
+      "learning_rate": 0.00017644455193205666,
+      "loss": 5.3005,
+      "step": 7050
+    },
+    {
+      "epoch": 9.028132992327366,
+      "grad_norm": 0.9708366990089417,
+      "learning_rate": 0.00017614407130317968,
+      "loss": 5.2737,
+      "step": 7060
+    },
+    {
+      "epoch": 9.040920716112533,
+      "grad_norm": 1.089858889579773,
+      "learning_rate": 0.00017584348243668556,
+      "loss": 5.262,
+      "step": 7070
+    },
+    {
+      "epoch": 9.053708439897699,
+      "grad_norm": 2.534782886505127,
+      "learning_rate": 0.00017554278657702549,
+      "loss": 5.2854,
+      "step": 7080
+    },
+    {
+      "epoch": 9.066496163682864,
+      "grad_norm": 1.3707194328308105,
+      "learning_rate": 0.00017524198496909373,
+      "loss": 5.2936,
+      "step": 7090
+    },
+    {
+      "epoch": 9.07928388746803,
+      "grad_norm": 1.4009782075881958,
+      "learning_rate": 0.0001749410788582223,
+      "loss": 5.2795,
+      "step": 7100
+    },
+    {
+      "epoch": 9.092071611253196,
+      "grad_norm": 1.1263126134872437,
+      "learning_rate": 0.00017464006949017584,
+      "loss": 5.2806,
+      "step": 7110
+    },
+    {
+      "epoch": 9.104859335038363,
+      "grad_norm": 0.9099717140197754,
+      "learning_rate": 0.00017433895811114658,
+      "loss": 5.3049,
+      "step": 7120
+    },
+    {
+      "epoch": 9.117647058823529,
+      "grad_norm": 1.3195571899414062,
+      "learning_rate": 0.00017403774596774893,
+      "loss": 5.2803,
+      "step": 7130
+    },
+    {
+      "epoch": 9.130434782608695,
+      "grad_norm": 1.1069055795669556,
+      "learning_rate": 0.00017373643430701463,
+      "loss": 5.2579,
+      "step": 7140
+    },
+    {
+      "epoch": 9.143222506393862,
+      "grad_norm": 1.1975563764572144,
+      "learning_rate": 0.00017343502437638727,
+      "loss": 5.2795,
+      "step": 7150
+    },
+    {
+      "epoch": 9.156010230179028,
+      "grad_norm": 1.6220418214797974,
+      "learning_rate": 0.00017313351742371746,
+      "loss": 5.2797,
+      "step": 7160
+    },
+    {
+      "epoch": 9.168797953964194,
+      "grad_norm": 1.490394115447998,
+      "learning_rate": 0.00017283191469725728,
+      "loss": 5.2768,
+      "step": 7170
+    },
+    {
+      "epoch": 9.18158567774936,
+      "grad_norm": 1.4838827848434448,
+      "learning_rate": 0.00017253021744565548,
+      "loss": 5.2871,
+      "step": 7180
+    },
+    {
+      "epoch": 9.194373401534527,
+      "grad_norm": 5.531613826751709,
+      "learning_rate": 0.0001722284269179521,
+      "loss": 5.2537,
+      "step": 7190
+    },
+    {
+      "epoch": 9.207161125319693,
+      "grad_norm": 1.5700544118881226,
+      "learning_rate": 0.0001719265443635733,
+      "loss": 5.2875,
+      "step": 7200
+    },
+    {
+      "epoch": 9.21994884910486,
+      "grad_norm": 1.2028346061706543,
+      "learning_rate": 0.00017162457103232632,
+      "loss": 5.2707,
+      "step": 7210
+    },
+    {
+      "epoch": 9.232736572890026,
+      "grad_norm": 1.9414821863174438,
+      "learning_rate": 0.00017132250817439412,
+      "loss": 5.2918,
+      "step": 7220
+    },
+    {
+      "epoch": 9.245524296675192,
+      "grad_norm": 3.8366823196411133,
+      "learning_rate": 0.00017102035704033038,
+      "loss": 5.277,
+      "step": 7230
+    },
+    {
+      "epoch": 9.258312020460359,
+      "grad_norm": 41.69551086425781,
+      "learning_rate": 0.0001707181188810542,
+      "loss": 5.2691,
+      "step": 7240
+    },
+    {
+      "epoch": 9.271099744245525,
+      "grad_norm": 2.0435402393341064,
+      "learning_rate": 0.00017041579494784506,
+      "loss": 5.3075,
+      "step": 7250
+    },
+    {
+      "epoch": 9.28388746803069,
+      "grad_norm": 2.1493489742279053,
+      "learning_rate": 0.00017011338649233743,
+      "loss": 5.3234,
+      "step": 7260
+    },
+    {
+      "epoch": 9.296675191815856,
+      "grad_norm": 3.627615213394165,
+      "learning_rate": 0.0001698108947665158,
+      "loss": 5.3018,
+      "step": 7270
+    },
+    {
+      "epoch": 9.309462915601022,
+      "grad_norm": 24.722545623779297,
+      "learning_rate": 0.00016950832102270927,
+      "loss": 5.3123,
+      "step": 7280
+    },
+    {
+      "epoch": 9.322250639386189,
+      "grad_norm": 19.32564353942871,
+      "learning_rate": 0.00016920566651358666,
+      "loss": 5.346,
+      "step": 7290
+    },
+    {
+      "epoch": 9.335038363171355,
+      "grad_norm": 173.1358184814453,
+      "learning_rate": 0.00016890293249215109,
+      "loss": 5.3385,
+      "step": 7300
+    },
+    {
+      "epoch": 9.347826086956522,
+      "grad_norm": 139.24111938476562,
+      "learning_rate": 0.0001686001202117348,
+      "loss": 5.3411,
+      "step": 7310
+    },
+    {
+      "epoch": 9.360613810741688,
+      "grad_norm": 12987.6923828125,
+      "learning_rate": 0.00016829723092599418,
+      "loss": 5.3288,
+      "step": 7320
+    },
+    {
+      "epoch": 9.373401534526854,
+      "grad_norm": 53.43489074707031,
+      "learning_rate": 0.00016799426588890427,
+      "loss": 5.3403,
+      "step": 7330
+    },
+    {
+      "epoch": 9.38618925831202,
+      "grad_norm": 29.375526428222656,
+      "learning_rate": 0.00016769122635475385,
+      "loss": 5.3186,
+      "step": 7340
+    },
+    {
+      "epoch": 9.398976982097187,
+      "grad_norm": 47.31606674194336,
+      "learning_rate": 0.00016738811357813998,
+      "loss": 5.3178,
+      "step": 7350
+    },
+    {
+      "epoch": 9.411764705882353,
+      "grad_norm": 12.416561126708984,
+      "learning_rate": 0.00016708492881396307,
+      "loss": 5.3385,
+      "step": 7360
+    },
+    {
+      "epoch": 9.42455242966752,
+      "grad_norm": 9.562813758850098,
+      "learning_rate": 0.0001667816733174215,
+      "loss": 5.3481,
+      "step": 7370
+    },
+    {
+      "epoch": 9.437340153452686,
+      "grad_norm": 3.935084819793701,
+      "learning_rate": 0.00016647834834400654,
+      "loss": 5.3439,
+      "step": 7380
+    },
+    {
+      "epoch": 9.450127877237852,
+      "grad_norm": 2.569079875946045,
+      "learning_rate": 0.00016617495514949704,
+      "loss": 5.3225,
+      "step": 7390
+    },
+    {
+      "epoch": 9.462915601023019,
+      "grad_norm": 1.732500672340393,
+      "learning_rate": 0.0001658714949899543,
+      "loss": 5.3235,
+      "step": 7400
+    },
+    {
+      "epoch": 9.475703324808185,
+      "grad_norm": 1.2219961881637573,
+      "learning_rate": 0.00016556796912171689,
+      "loss": 5.3413,
+      "step": 7410
+    },
+    {
+      "epoch": 9.48849104859335,
+      "grad_norm": 0.9043082594871521,
+      "learning_rate": 0.00016526437880139537,
+      "loss": 5.288,
+      "step": 7420
+    },
+    {
+      "epoch": 9.501278772378516,
+      "grad_norm": 1.0241756439208984,
+      "learning_rate": 0.0001649607252858672,
+      "loss": 5.302,
+      "step": 7430
+    },
+    {
+      "epoch": 9.514066496163682,
+      "grad_norm": 1.5669431686401367,
+      "learning_rate": 0.00016465700983227138,
+      "loss": 5.2899,
+      "step": 7440
+    },
+    {
+      "epoch": 9.526854219948849,
+      "grad_norm": 1.5617988109588623,
+      "learning_rate": 0.00016435323369800344,
+      "loss": 5.2868,
+      "step": 7450
+    },
+    {
+      "epoch": 9.539641943734015,
+      "grad_norm": 1.1847914457321167,
+      "learning_rate": 0.00016404939814071003,
+      "loss": 5.2617,
+      "step": 7460
+    },
+    {
+      "epoch": 9.552429667519181,
+      "grad_norm": 0.8560781478881836,
+      "learning_rate": 0.0001637455044182839,
+      "loss": 5.2855,
+      "step": 7470
+    },
+    {
+      "epoch": 9.565217391304348,
+      "grad_norm": 0.9926068782806396,
+      "learning_rate": 0.0001634415537888585,
+      "loss": 5.265,
+      "step": 7480
+    },
+    {
+      "epoch": 9.578005115089514,
+      "grad_norm": 2.2098798751831055,
+      "learning_rate": 0.00016313754751080302,
+      "loss": 5.2773,
+      "step": 7490
+    },
+    {
+      "epoch": 9.59079283887468,
+      "grad_norm": 1.3162308931350708,
+      "learning_rate": 0.00016283348684271694,
+      "loss": 5.276,
+      "step": 7500
+    },
+    {
+      "epoch": 9.603580562659847,
+      "grad_norm": 1.2072679996490479,
+      "learning_rate": 0.00016252937304342494,
+      "loss": 5.2825,
+      "step": 7510
+    },
+    {
+      "epoch": 9.616368286445013,
+      "grad_norm": 1.212632656097412,
+      "learning_rate": 0.0001622252073719717,
+      "loss": 5.2609,
+      "step": 7520
+    },
+    {
+      "epoch": 9.62915601023018,
+      "grad_norm": 1.244361400604248,
+      "learning_rate": 0.0001619209910876165,
+      "loss": 5.247,
+      "step": 7530
+    },
+    {
+      "epoch": 9.641943734015346,
+      "grad_norm": 1.2935197353363037,
+      "learning_rate": 0.00016161672544982842,
+      "loss": 5.2666,
+      "step": 7540
+    },
+    {
+      "epoch": 9.654731457800512,
+      "grad_norm": 1.270849347114563,
+      "learning_rate": 0.00016131241171828063,
+      "loss": 5.2556,
+      "step": 7550
+    },
+    {
+      "epoch": 9.667519181585678,
+      "grad_norm": 1.5244646072387695,
+      "learning_rate": 0.00016100805115284555,
+      "loss": 5.2594,
+      "step": 7560
+    },
+    {
+      "epoch": 9.680306905370845,
+      "grad_norm": 1.245428204536438,
+      "learning_rate": 0.00016070364501358944,
+      "loss": 5.2452,
+      "step": 7570
+    },
+    {
+      "epoch": 9.693094629156011,
+      "grad_norm": 2.1988797187805176,
+      "learning_rate": 0.00016039919456076727,
+      "loss": 5.289,
+      "step": 7580
+    },
+    {
+      "epoch": 9.705882352941176,
+      "grad_norm": 4.7878031730651855,
+      "learning_rate": 0.00016009470105481736,
+      "loss": 5.2933,
+      "step": 7590
+    },
+    {
+      "epoch": 9.718670076726342,
+      "grad_norm": 1.0919615030288696,
+      "learning_rate": 0.00015979016575635644,
+      "loss": 5.2634,
+      "step": 7600
+    },
+    {
+      "epoch": 9.731457800511508,
+      "grad_norm": 1.3478410243988037,
+      "learning_rate": 0.00015948558992617416,
+      "loss": 5.2808,
+      "step": 7610
+    },
+    {
+      "epoch": 9.744245524296675,
+      "grad_norm": 1.088616132736206,
+      "learning_rate": 0.00015918097482522798,
+      "loss": 5.2656,
+      "step": 7620
+    },
+    {
+      "epoch": 9.757033248081841,
+      "grad_norm": 1.4594988822937012,
+      "learning_rate": 0.00015887632171463794,
+      "loss": 5.2422,
+      "step": 7630
+    },
+    {
+      "epoch": 9.769820971867007,
+      "grad_norm": 1.346463918685913,
+      "learning_rate": 0.00015857163185568153,
+      "loss": 5.2668,
+      "step": 7640
+    },
+    {
+      "epoch": 9.782608695652174,
+      "grad_norm": 2.3005871772766113,
+      "learning_rate": 0.00015826690650978825,
+      "loss": 5.2295,
+      "step": 7650
+    },
+    {
+      "epoch": 9.79539641943734,
+      "grad_norm": 1.372439980506897,
+      "learning_rate": 0.0001579621469385346,
+      "loss": 5.2419,
+      "step": 7660
+    },
+    {
+      "epoch": 9.808184143222507,
+      "grad_norm": 1.0069390535354614,
+      "learning_rate": 0.00015765735440363872,
+      "loss": 5.2692,
+      "step": 7670
+    },
+    {
+      "epoch": 9.820971867007673,
+      "grad_norm": 1.3591490983963013,
+      "learning_rate": 0.00015735253016695527,
+      "loss": 5.269,
+      "step": 7680
+    },
+    {
+      "epoch": 9.83375959079284,
+      "grad_norm": 1.47946298122406,
+      "learning_rate": 0.00015704767549047015,
+      "loss": 5.2615,
+      "step": 7690
+    },
+    {
+      "epoch": 9.846547314578006,
+      "grad_norm": 2.2882235050201416,
+      "learning_rate": 0.00015674279163629528,
+      "loss": 5.2452,
+      "step": 7700
+    },
+    {
+      "epoch": 9.859335038363172,
+      "grad_norm": 1.406764268875122,
+      "learning_rate": 0.00015643787986666333,
+      "loss": 5.2515,
+      "step": 7710
+    },
+    {
+      "epoch": 9.872122762148338,
+      "grad_norm": 0.9985896348953247,
+      "learning_rate": 0.00015613294144392256,
+      "loss": 5.2536,
+      "step": 7720
+    },
+    {
+      "epoch": 9.884910485933505,
+      "grad_norm": 1.1201996803283691,
+      "learning_rate": 0.00015582797763053166,
+      "loss": 5.2459,
+      "step": 7730
+    },
+    {
+      "epoch": 9.89769820971867,
+      "grad_norm": 1.2289462089538574,
+      "learning_rate": 0.00015552298968905432,
+      "loss": 5.2162,
+      "step": 7740
+    },
+    {
+      "epoch": 9.910485933503836,
+      "grad_norm": 1.3729525804519653,
+      "learning_rate": 0.00015521797888215424,
+      "loss": 5.2488,
+      "step": 7750
+    },
+    {
+      "epoch": 9.923273657289002,
+      "grad_norm": 1.3408769369125366,
+      "learning_rate": 0.00015491294647258967,
+      "loss": 5.2608,
+      "step": 7760
+    },
+    {
+      "epoch": 9.936061381074168,
+      "grad_norm": 1.3026928901672363,
+      "learning_rate": 0.0001546078937232083,
+      "loss": 5.227,
+      "step": 7770
+    },
+    {
+      "epoch": 9.948849104859335,
+      "grad_norm": 1.3715628385543823,
+      "learning_rate": 0.00015430282189694212,
+      "loss": 5.2677,
+      "step": 7780
+    },
+    {
+      "epoch": 9.961636828644501,
+      "grad_norm": 1.0734012126922607,
+      "learning_rate": 0.00015399773225680208,
+      "loss": 5.2575,
+      "step": 7790
+    },
+    {
+      "epoch": 9.974424552429667,
+      "grad_norm": 1.0179634094238281,
+      "learning_rate": 0.00015369262606587281,
+      "loss": 5.2117,
+      "step": 7800
+    },
+    {
+      "epoch": 9.987212276214834,
+      "grad_norm": 1.2135021686553955,
+      "learning_rate": 0.00015338750458730746,
+      "loss": 5.2387,
+      "step": 7810
+    },
+    {
+      "epoch": 10.0,
+      "grad_norm": 1.7299256324768066,
+      "learning_rate": 0.00015308236908432264,
+      "loss": 5.2517,
+      "step": 7820
+    },
+    {
+      "epoch": 10.012787723785166,
+      "grad_norm": 1.3696908950805664,
+      "learning_rate": 0.00015277722082019272,
+      "loss": 5.2373,
+      "step": 7830
+    },
+    {
+      "epoch": 10.025575447570333,
+      "grad_norm": 1.4967882633209229,
+      "learning_rate": 0.00015247206105824522,
+      "loss": 5.2045,
+      "step": 7840
+    },
+    {
+      "epoch": 10.038363171355499,
+      "grad_norm": 1.4495280981063843,
+      "learning_rate": 0.00015216689106185505,
+      "loss": 5.2286,
+      "step": 7850
+    },
+    {
+      "epoch": 10.051150895140665,
+      "grad_norm": 2.255189895629883,
+      "learning_rate": 0.00015186171209443958,
+      "loss": 5.1753,
+      "step": 7860
+    },
+    {
+      "epoch": 10.063938618925832,
+      "grad_norm": 1.22813081741333,
+      "learning_rate": 0.00015155652541945326,
+      "loss": 5.2259,
+      "step": 7870
+    },
+    {
+      "epoch": 10.076726342710998,
+      "grad_norm": 1.1559652090072632,
+      "learning_rate": 0.00015125133230038256,
+      "loss": 5.1997,
+      "step": 7880
+    },
+    {
+      "epoch": 10.089514066496164,
+      "grad_norm": 1.5692169666290283,
+      "learning_rate": 0.00015094613400074052,
+      "loss": 5.2184,
+      "step": 7890
+    },
+    {
+      "epoch": 10.10230179028133,
+      "grad_norm": 1.143264651298523,
+      "learning_rate": 0.00015064093178406165,
+      "loss": 5.2217,
+      "step": 7900
+    },
+    {
+      "epoch": 10.115089514066495,
+      "grad_norm": 1.3026010990142822,
+      "learning_rate": 0.00015033572691389673,
+      "loss": 5.2219,
+      "step": 7910
+    },
+    {
+      "epoch": 10.127877237851662,
+      "grad_norm": 1.255898356437683,
+      "learning_rate": 0.00015003052065380742,
+      "loss": 5.1964,
+      "step": 7920
+    },
+    {
+      "epoch": 10.140664961636828,
+      "grad_norm": 1.5015469789505005,
+      "learning_rate": 0.00014972531426736131,
+      "loss": 5.241,
+      "step": 7930
+    },
+    {
+      "epoch": 10.153452685421994,
+      "grad_norm": 1.5134257078170776,
+      "learning_rate": 0.0001494201090181263,
+      "loss": 5.2255,
+      "step": 7940
+    },
+    {
+      "epoch": 10.16624040920716,
+      "grad_norm": 1.295350432395935,
+      "learning_rate": 0.00014911490616966575,
+      "loss": 5.2201,
+      "step": 7950
+    },
+    {
+      "epoch": 10.179028132992327,
+      "grad_norm": 1.8612785339355469,
+      "learning_rate": 0.000148809706985533,
+      "loss": 5.2109,
+      "step": 7960
+    },
+    {
+      "epoch": 10.191815856777493,
+      "grad_norm": 1.5443718433380127,
+      "learning_rate": 0.0001485045127292662,
+      "loss": 5.2247,
+      "step": 7970
+    },
+    {
+      "epoch": 10.20460358056266,
+      "grad_norm": 1.7104902267456055,
+      "learning_rate": 0.00014819932466438317,
+      "loss": 5.2141,
+      "step": 7980
+    },
+    {
+      "epoch": 10.217391304347826,
+      "grad_norm": 1.7787065505981445,
+      "learning_rate": 0.00014789414405437607,
+      "loss": 5.1911,
+      "step": 7990
+    },
+    {
+      "epoch": 10.230179028132993,
+      "grad_norm": 1.0850389003753662,
+      "learning_rate": 0.0001475889721627062,
+      "loss": 5.2156,
+      "step": 8000
+    },
+    {
+      "epoch": 1.0252143862792782,
+      "grad_norm": 1.8151112794876099,
+      "learning_rate": 0.000298150415790792,
+      "loss": 5.5132,
+      "step": 8010
+    },
+    {
+      "epoch": 1.0264943043645207,
+      "grad_norm": 1.6067150831222534,
+      "learning_rate": 0.0002981456854903832,
+      "loss": 5.4928,
+      "step": 8020
+    },
+    {
+      "epoch": 1.0277742224497632,
+      "grad_norm": 1.838860273361206,
+      "learning_rate": 0.00029814094918646013,
+      "loss": 5.4783,
+      "step": 8030
+    },
+    {
+      "epoch": 1.0290541405350058,
+      "grad_norm": 1.2739697694778442,
+      "learning_rate": 0.0002981362068792148,
+      "loss": 5.481,
+      "step": 8040
+    },
+    {
+      "epoch": 1.0303340586202483,
+      "grad_norm": 1.7215646505355835,
+      "learning_rate": 0.0002981314585688393,
+      "loss": 5.4588,
+      "step": 8050
+    },
+    {
+      "epoch": 1.0316139767054908,
+      "grad_norm": 1.9005746841430664,
+      "learning_rate": 0.00029812670425552603,
+      "loss": 5.4475,
+      "step": 8060
+    },
+    {
+      "epoch": 1.0328938947907333,
+      "grad_norm": 1.0230218172073364,
+      "learning_rate": 0.00029812194393946776,
+      "loss": 5.4565,
+      "step": 8070
+    },
+    {
+      "epoch": 1.0341738128759759,
+      "grad_norm": 1.4073644876480103,
+      "learning_rate": 0.0002981171776208573,
+      "loss": 5.4663,
+      "step": 8080
+    },
+    {
+      "epoch": 1.0354537309612184,
+      "grad_norm": 2.152285575866699,
+      "learning_rate": 0.0002981124052998879,
+      "loss": 5.4846,
+      "step": 8090
+    },
+    {
+      "epoch": 1.036733649046461,
+      "grad_norm": 2.273869752883911,
+      "learning_rate": 0.0002981076269767529,
+      "loss": 5.4498,
+      "step": 8100
+    },
+    {
+      "epoch": 1.0380135671317037,
+      "grad_norm": 1.4030476808547974,
+      "learning_rate": 0.00029810284265164585,
+      "loss": 5.4423,
+      "step": 8110
+    },
+    {
+      "epoch": 1.0392934852169462,
+      "grad_norm": 1.3551042079925537,
+      "learning_rate": 0.00029809805232476074,
+      "loss": 5.4515,
+      "step": 8120
+    },
+    {
+      "epoch": 1.0405734033021887,
+      "grad_norm": 1.278965711593628,
+      "learning_rate": 0.00029809325599629174,
+      "loss": 5.4393,
+      "step": 8130
+    },
+    {
+      "epoch": 1.0418533213874313,
+      "grad_norm": 1.7702277898788452,
+      "learning_rate": 0.00029808845366643315,
+      "loss": 5.432,
+      "step": 8140
+    },
+    {
+      "epoch": 1.0431332394726738,
+      "grad_norm": 1.435707449913025,
+      "learning_rate": 0.0002980836453353795,
+      "loss": 5.4574,
+      "step": 8150
+    },
+    {
+      "epoch": 1.0444131575579163,
+      "grad_norm": 1.4043419361114502,
+      "learning_rate": 0.0002980788310033258,
+      "loss": 5.4436,
+      "step": 8160
+    },
+    {
+      "epoch": 1.0456930756431588,
+      "grad_norm": 1.6054362058639526,
+      "learning_rate": 0.00029807401067046707,
+      "loss": 5.4293,
+      "step": 8170
+    },
+    {
+      "epoch": 1.0469729937284014,
+      "grad_norm": 1.724361538887024,
+      "learning_rate": 0.00029806918433699866,
+      "loss": 5.4381,
+      "step": 8180
+    },
+    {
+      "epoch": 1.0482529118136439,
+      "grad_norm": 1.4272695779800415,
+      "learning_rate": 0.0002980643520031161,
+      "loss": 5.4341,
+      "step": 8190
+    },
+    {
+      "epoch": 1.0495328298988864,
+      "grad_norm": 1.6520978212356567,
+      "learning_rate": 0.0002980595136690153,
+      "loss": 5.433,
+      "step": 8200
+    },
+    {
+      "epoch": 1.050812747984129,
+      "grad_norm": 1.906540870666504,
+      "learning_rate": 0.0002980546693348923,
+      "loss": 5.4493,
+      "step": 8210
+    },
+    {
+      "epoch": 1.0520926660693715,
+      "grad_norm": 1.3540748357772827,
+      "learning_rate": 0.00029804981900094344,
+      "loss": 5.4291,
+      "step": 8220
+    },
+    {
+      "epoch": 1.0533725841546142,
+      "grad_norm": 1.2956520318984985,
+      "learning_rate": 0.0002980449626673652,
+      "loss": 5.4322,
+      "step": 8230
+    },
+    {
+      "epoch": 1.0546525022398567,
+      "grad_norm": 2.1103005409240723,
+      "learning_rate": 0.0002980401003343545,
+      "loss": 5.4154,
+      "step": 8240
+    },
+    {
+      "epoch": 1.0559324203250993,
+      "grad_norm": 1.9782565832138062,
+      "learning_rate": 0.00029803523200210827,
+      "loss": 5.4287,
+      "step": 8250
+    },
+    {
+      "epoch": 1.0572123384103418,
+      "grad_norm": 1.5578869581222534,
+      "learning_rate": 0.0002980303576708238,
+      "loss": 5.4315,
+      "step": 8260
+    },
+    {
+      "epoch": 1.0584922564955843,
+      "grad_norm": 1.8073164224624634,
+      "learning_rate": 0.0002980254773406987,
+      "loss": 5.4278,
+      "step": 8270
+    },
+    {
+      "epoch": 1.0597721745808268,
+      "grad_norm": 1.7924606800079346,
+      "learning_rate": 0.0002980205910119307,
+      "loss": 5.4216,
+      "step": 8280
+    },
+    {
+      "epoch": 1.0610520926660694,
+      "grad_norm": 1.3763577938079834,
+      "learning_rate": 0.00029801569868471783,
+      "loss": 5.4291,
+      "step": 8290
+    },
+    {
+      "epoch": 1.062332010751312,
+      "grad_norm": 1.5295467376708984,
+      "learning_rate": 0.00029801080035925833,
+      "loss": 5.4014,
+      "step": 8300
+    },
+    {
+      "epoch": 1.0636119288365544,
+      "grad_norm": 2.3050379753112793,
+      "learning_rate": 0.0002980058960357507,
+      "loss": 5.4266,
+      "step": 8310
+    },
+    {
+      "epoch": 1.064891846921797,
+      "grad_norm": 2.043785333633423,
+      "learning_rate": 0.0002980009857143937,
+      "loss": 5.4185,
+      "step": 8320
+    },
+    {
+      "epoch": 1.0661717650070395,
+      "grad_norm": 1.808798909187317,
+      "learning_rate": 0.0002979960693953863,
+      "loss": 5.4153,
+      "step": 8330
+    },
+    {
+      "epoch": 1.067451683092282,
+      "grad_norm": 1.8030039072036743,
+      "learning_rate": 0.00029799114707892776,
+      "loss": 5.4243,
+      "step": 8340
+    },
+    {
+      "epoch": 1.0687316011775247,
+      "grad_norm": 1.0454009771347046,
+      "learning_rate": 0.0002979862187652175,
+      "loss": 5.4105,
+      "step": 8350
+    },
+    {
+      "epoch": 1.0700115192627673,
+      "grad_norm": 1.3364286422729492,
+      "learning_rate": 0.0002979812844544553,
+      "loss": 5.4189,
+      "step": 8360
+    },
+    {
+      "epoch": 1.0712914373480098,
+      "grad_norm": 2.3642287254333496,
+      "learning_rate": 0.00029797634414684105,
+      "loss": 5.4119,
+      "step": 8370
+    },
+    {
+      "epoch": 1.0725713554332523,
+      "grad_norm": 1.865700125694275,
+      "learning_rate": 0.00029797139784257503,
+      "loss": 5.4069,
+      "step": 8380
+    },
+    {
+      "epoch": 1.0738512735184949,
+      "grad_norm": 1.0280027389526367,
+      "learning_rate": 0.0002979664455418576,
+      "loss": 5.4096,
+      "step": 8390
+    },
+    {
+      "epoch": 1.0751311916037374,
+      "grad_norm": 1.3924864530563354,
+      "learning_rate": 0.00029796148724488963,
+      "loss": 5.3929,
+      "step": 8400
+    },
+    {
+      "epoch": 1.07641110968898,
+      "grad_norm": 1.2913914918899536,
+      "learning_rate": 0.0002979565229518718,
+      "loss": 5.4008,
+      "step": 8410
+    },
+    {
+      "epoch": 1.0776910277742224,
+      "grad_norm": 1.4355765581130981,
+      "learning_rate": 0.00029795155266300544,
+      "loss": 5.388,
+      "step": 8420
+    },
+    {
+      "epoch": 1.078970945859465,
+      "grad_norm": 1.5805541276931763,
+      "learning_rate": 0.00029794657637849197,
+      "loss": 5.4101,
+      "step": 8430
+    },
+    {
+      "epoch": 1.0802508639447075,
+      "grad_norm": 1.2328497171401978,
+      "learning_rate": 0.000297941594098533,
+      "loss": 5.3967,
+      "step": 8440
+    },
+    {
+      "epoch": 1.08153078202995,
+      "grad_norm": 2.1215150356292725,
+      "learning_rate": 0.0002979366058233304,
+      "loss": 5.3945,
+      "step": 8450
+    },
+    {
+      "epoch": 1.0828107001151925,
+      "grad_norm": 1.345954418182373,
+      "learning_rate": 0.00029793161155308636,
+      "loss": 5.3928,
+      "step": 8460
+    },
+    {
+      "epoch": 1.084090618200435,
+      "grad_norm": 1.8684533834457397,
+      "learning_rate": 0.00029792661128800335,
+      "loss": 5.3869,
+      "step": 8470
+    },
+    {
+      "epoch": 1.0853705362856778,
+      "grad_norm": 1.3567562103271484,
+      "learning_rate": 0.00029792160502828386,
+      "loss": 5.3845,
+      "step": 8480
+    },
+    {
+      "epoch": 1.0866504543709203,
+      "grad_norm": 1.8894990682601929,
+      "learning_rate": 0.0002979165927741309,
+      "loss": 5.3699,
+      "step": 8490
+    },
+    {
+      "epoch": 1.0879303724561629,
+      "grad_norm": 1.4628643989562988,
+      "learning_rate": 0.00029791157452574744,
+      "loss": 5.3933,
+      "step": 8500
+    },
+    {
+      "epoch": 1.0892102905414054,
+      "grad_norm": 1.6238657236099243,
+      "learning_rate": 0.00029790655028333704,
+      "loss": 5.3868,
+      "step": 8510
+    },
+    {
+      "epoch": 1.090490208626648,
+      "grad_norm": 2.4890058040618896,
+      "learning_rate": 0.0002979015200471031,
+      "loss": 5.4187,
+      "step": 8520
+    },
+    {
+      "epoch": 1.0917701267118904,
+      "grad_norm": 1.314422845840454,
+      "learning_rate": 0.0002978964838172496,
+      "loss": 5.4416,
+      "step": 8530
+    },
+    {
+      "epoch": 1.093050044797133,
+      "grad_norm": 2.026683807373047,
+      "learning_rate": 0.0002978914415939805,
+      "loss": 5.4036,
+      "step": 8540
+    },
+    {
+      "epoch": 1.0943299628823755,
+      "grad_norm": 1.7179348468780518,
+      "learning_rate": 0.0002978863933775003,
+      "loss": 5.4044,
+      "step": 8550
+    },
+    {
+      "epoch": 1.095609880967618,
+      "grad_norm": 1.7547870874404907,
+      "learning_rate": 0.0002978813391680135,
+      "loss": 5.383,
+      "step": 8560
+    },
+    {
+      "epoch": 1.0968897990528605,
+      "grad_norm": 2.172962188720703,
+      "learning_rate": 0.00029787627896572485,
+      "loss": 5.3926,
+      "step": 8570
+    },
+    {
+      "epoch": 1.098169717138103,
+      "grad_norm": 1.2563143968582153,
+      "learning_rate": 0.00029787121277083953,
+      "loss": 5.386,
+      "step": 8580
+    },
+    {
+      "epoch": 1.0994496352233458,
+      "grad_norm": 1.3007797002792358,
+      "learning_rate": 0.0002978661405835628,
+      "loss": 5.3973,
+      "step": 8590
+    },
+    {
+      "epoch": 1.1007295533085883,
+      "grad_norm": 1.491527795791626,
+      "learning_rate": 0.0002978610624041002,
+      "loss": 5.3711,
+      "step": 8600
+    },
+    {
+      "epoch": 1.1020094713938309,
+      "grad_norm": 1.8254247903823853,
+      "learning_rate": 0.00029785597823265745,
+      "loss": 5.3771,
+      "step": 8610
+    },
+    {
+      "epoch": 1.1032893894790734,
+      "grad_norm": 1.2141834497451782,
+      "learning_rate": 0.0002978508880694407,
+      "loss": 5.3485,
+      "step": 8620
+    },
+    {
+      "epoch": 1.104569307564316,
+      "grad_norm": 2.4979705810546875,
+      "learning_rate": 0.00029784579191465617,
+      "loss": 5.3651,
+      "step": 8630
+    },
+    {
+      "epoch": 1.1058492256495585,
+      "grad_norm": 1.4393336772918701,
+      "learning_rate": 0.00029784068976851033,
+      "loss": 5.3681,
+      "step": 8640
+    },
+    {
+      "epoch": 1.107129143734801,
+      "grad_norm": 1.0520734786987305,
+      "learning_rate": 0.00029783558163121003,
+      "loss": 5.3689,
+      "step": 8650
+    },
+    {
+      "epoch": 1.1084090618200435,
+      "grad_norm": 1.4579353332519531,
+      "learning_rate": 0.0002978304675029623,
+      "loss": 5.3684,
+      "step": 8660
+    },
+    {
+      "epoch": 1.109688979905286,
+      "grad_norm": 1.8464175462722778,
+      "learning_rate": 0.0002978253473839743,
+      "loss": 5.3787,
+      "step": 8670
+    },
+    {
+      "epoch": 1.1109688979905286,
+      "grad_norm": 1.292655110359192,
+      "learning_rate": 0.00029782022127445353,
+      "loss": 5.3641,
+      "step": 8680
+    },
+    {
+      "epoch": 1.112248816075771,
+      "grad_norm": 1.6093817949295044,
+      "learning_rate": 0.0002978150891746077,
+      "loss": 5.3623,
+      "step": 8690
+    },
+    {
+      "epoch": 1.1135287341610136,
+      "grad_norm": 1.0206555128097534,
+      "learning_rate": 0.00029780995108464486,
+      "loss": 5.3739,
+      "step": 8700
+    },
+    {
+      "epoch": 1.1148086522462561,
+      "grad_norm": 1.2137470245361328,
+      "learning_rate": 0.0002978048070047732,
+      "loss": 5.3592,
+      "step": 8710
+    },
+    {
+      "epoch": 1.1160885703314989,
+      "grad_norm": 2.88940167427063,
+      "learning_rate": 0.0002977996569352012,
+      "loss": 5.3671,
+      "step": 8720
+    },
+    {
+      "epoch": 1.1173684884167414,
+      "grad_norm": 1.197797179222107,
+      "learning_rate": 0.0002977945008761375,
+      "loss": 5.3451,
+      "step": 8730
+    },
+    {
+      "epoch": 1.118648406501984,
+      "grad_norm": 1.198354959487915,
+      "learning_rate": 0.0002977893388277911,
+      "loss": 5.3827,
+      "step": 8740
+    },
+    {
+      "epoch": 1.1199283245872265,
+      "grad_norm": 1.0703370571136475,
+      "learning_rate": 0.0002977841707903711,
+      "loss": 5.3249,
+      "step": 8750
+    },
+    {
+      "epoch": 1.121208242672469,
+      "grad_norm": 1.3689323663711548,
+      "learning_rate": 0.0002977789967640871,
+      "loss": 5.3569,
+      "step": 8760
+    },
+    {
+      "epoch": 1.1224881607577115,
+      "grad_norm": 1.3093551397323608,
+      "learning_rate": 0.0002977738167491486,
+      "loss": 5.3379,
+      "step": 8770
+    },
+    {
+      "epoch": 1.123768078842954,
+      "grad_norm": 1.698310136795044,
+      "learning_rate": 0.0002977686307457657,
+      "loss": 5.3638,
+      "step": 8780
+    },
+    {
+      "epoch": 1.1250479969281966,
+      "grad_norm": 1.5317034721374512,
+      "learning_rate": 0.00029776343875414834,
+      "loss": 5.3916,
+      "step": 8790
+    },
+    {
+      "epoch": 1.126327915013439,
+      "grad_norm": 1.757363200187683,
+      "learning_rate": 0.0002977582407745071,
+      "loss": 5.3482,
+      "step": 8800
+    },
+    {
+      "epoch": 1.1276078330986816,
+      "grad_norm": 1.6926486492156982,
+      "learning_rate": 0.00029775303680705257,
+      "loss": 5.34,
+      "step": 8810
+    },
+    {
+      "epoch": 1.1288877511839241,
+      "grad_norm": 2.5482800006866455,
+      "learning_rate": 0.0002977478268519956,
+      "loss": 5.3665,
+      "step": 8820
+    },
+    {
+      "epoch": 1.130167669269167,
+      "grad_norm": 2.2844486236572266,
+      "learning_rate": 0.0002977426109095474,
+      "loss": 5.3552,
+      "step": 8830
+    },
+    {
+      "epoch": 1.1314475873544092,
+      "grad_norm": 2.151538372039795,
+      "learning_rate": 0.0002977373889799192,
+      "loss": 5.3469,
+      "step": 8840
+    },
+    {
+      "epoch": 1.132727505439652,
+      "grad_norm": 1.4821778535842896,
+      "learning_rate": 0.00029773216106332277,
+      "loss": 5.3422,
+      "step": 8850
+    },
+    {
+      "epoch": 1.1340074235248945,
+      "grad_norm": 1.5019316673278809,
+      "learning_rate": 0.00029772692715996984,
+      "loss": 5.3551,
+      "step": 8860
+    },
+    {
+      "epoch": 1.135287341610137,
+      "grad_norm": 1.7702430486679077,
+      "learning_rate": 0.0002977216872700727,
+      "loss": 5.3309,
+      "step": 8870
+    },
+    {
+      "epoch": 1.1365672596953795,
+      "grad_norm": 1.5363876819610596,
+      "learning_rate": 0.00029771644139384347,
+      "loss": 5.339,
+      "step": 8880
+    },
+    {
+      "epoch": 1.137847177780622,
+      "grad_norm": 1.330912470817566,
+      "learning_rate": 0.00029771118953149486,
+      "loss": 5.369,
+      "step": 8890
+    },
+    {
+      "epoch": 1.1391270958658646,
+      "grad_norm": 1.5051958560943604,
+      "learning_rate": 0.00029770593168323967,
+      "loss": 5.3514,
+      "step": 8900
+    },
+    {
+      "epoch": 1.140407013951107,
+      "grad_norm": 1.473021388053894,
+      "learning_rate": 0.000297700667849291,
+      "loss": 5.3574,
+      "step": 8910
+    },
+    {
+      "epoch": 1.1416869320363496,
+      "grad_norm": 1.1854453086853027,
+      "learning_rate": 0.0002976953980298621,
+      "loss": 5.3291,
+      "step": 8920
+    },
+    {
+      "epoch": 1.1429668501215922,
+      "grad_norm": 1.426828145980835,
+      "learning_rate": 0.0002976901222251666,
+      "loss": 5.3342,
+      "step": 8930
+    },
+    {
+      "epoch": 1.1442467682068347,
+      "grad_norm": 2.1792798042297363,
+      "learning_rate": 0.0002976848404354183,
+      "loss": 5.3406,
+      "step": 8940
+    },
+    {
+      "epoch": 1.1455266862920772,
+      "grad_norm": 1.5441704988479614,
+      "learning_rate": 0.00029767955266083113,
+      "loss": 5.3278,
+      "step": 8950
+    },
+    {
+      "epoch": 1.14680660437732,
+      "grad_norm": 1.8839783668518066,
+      "learning_rate": 0.0002976742589016195,
+      "loss": 5.3459,
+      "step": 8960
+    },
+    {
+      "epoch": 1.1480865224625625,
+      "grad_norm": 1.377284288406372,
+      "learning_rate": 0.00029766895915799783,
+      "loss": 5.2889,
+      "step": 8970
+    },
+    {
+      "epoch": 1.149366440547805,
+      "grad_norm": 1.5436594486236572,
+      "learning_rate": 0.000297663653430181,
+      "loss": 5.3387,
+      "step": 8980
+    },
+    {
+      "epoch": 1.1506463586330475,
+      "grad_norm": 1.8926775455474854,
+      "learning_rate": 0.00029765834171838397,
+      "loss": 5.3313,
+      "step": 8990
+    },
+    {
+      "epoch": 1.15192627671829,
+      "grad_norm": 2.016495704650879,
+      "learning_rate": 0.00029765302402282197,
+      "loss": 5.3102,
+      "step": 9000
+    },
+    {
+      "epoch": 1.1532061948035326,
+      "grad_norm": 1.5337920188903809,
+      "learning_rate": 0.0002976477003437105,
+      "loss": 5.3099,
+      "step": 9010
+    },
+    {
+      "epoch": 1.1544861128887751,
+      "grad_norm": 1.6861891746520996,
+      "learning_rate": 0.00029764237068126534,
+      "loss": 5.3177,
+      "step": 9020
+    },
+    {
+      "epoch": 1.1557660309740176,
+      "grad_norm": 11.703009605407715,
+      "learning_rate": 0.00029763703503570243,
+      "loss": 5.3458,
+      "step": 9030
+    },
+    {
+      "epoch": 1.1570459490592602,
+      "grad_norm": 1.3665626049041748,
+      "learning_rate": 0.000297631693407238,
+      "loss": 5.3726,
+      "step": 9040
+    },
+    {
+      "epoch": 1.1583258671445027,
+      "grad_norm": 1.9868730306625366,
+      "learning_rate": 0.0002976263457960886,
+      "loss": 5.3459,
+      "step": 9050
+    },
+    {
+      "epoch": 1.1596057852297452,
+      "grad_norm": 6.771224021911621,
+      "learning_rate": 0.0002976209922024708,
+      "loss": 5.4621,
+      "step": 9060
+    },
+    {
+      "epoch": 1.1608857033149877,
+      "grad_norm": 1.730333685874939,
+      "learning_rate": 0.00029761563262660165,
+      "loss": 5.3924,
+      "step": 9070
+    },
+    {
+      "epoch": 1.1621656214002303,
+      "grad_norm": 1.123976707458496,
+      "learning_rate": 0.0002976102670686983,
+      "loss": 5.3687,
+      "step": 9080
+    },
+    {
+      "epoch": 1.163445539485473,
+      "grad_norm": 2.981642723083496,
+      "learning_rate": 0.00029760489552897815,
+      "loss": 5.3319,
+      "step": 9090
+    },
+    {
+      "epoch": 1.1647254575707155,
+      "grad_norm": 1.4254289865493774,
+      "learning_rate": 0.00029759951800765896,
+      "loss": 5.3132,
+      "step": 9100
+    },
+    {
+      "epoch": 1.166005375655958,
+      "grad_norm": 1.141530156135559,
+      "learning_rate": 0.00029759413450495864,
+      "loss": 5.3433,
+      "step": 9110
+    },
+    {
+      "epoch": 1.1672852937412006,
+      "grad_norm": 1.7244399785995483,
+      "learning_rate": 0.0002975887450210953,
+      "loss": 5.3198,
+      "step": 9120
+    },
+    {
+      "epoch": 1.1685652118264431,
+      "grad_norm": 1.030436396598816,
+      "learning_rate": 0.00029758334955628737,
+      "loss": 5.3244,
+      "step": 9130
+    },
+    {
+      "epoch": 1.1698451299116857,
+      "grad_norm": 1.2787984609603882,
+      "learning_rate": 0.0002975779481107535,
+      "loss": 5.3185,
+      "step": 9140
+    },
+    {
+      "epoch": 1.1711250479969282,
+      "grad_norm": 1.4879578351974487,
+      "learning_rate": 0.00029757254068471257,
+      "loss": 5.3408,
+      "step": 9150
+    },
+    {
+      "epoch": 1.1724049660821707,
+      "grad_norm": 1.3642961978912354,
+      "learning_rate": 0.0002975671272783838,
+      "loss": 5.3289,
+      "step": 9160
+    },
+    {
+      "epoch": 1.1736848841674132,
+      "grad_norm": 1.4523587226867676,
+      "learning_rate": 0.0002975617078919864,
+      "loss": 5.3273,
+      "step": 9170
+    },
+    {
+      "epoch": 1.1749648022526558,
+      "grad_norm": 1.5739113092422485,
+      "learning_rate": 0.00029755628252574014,
+      "loss": 5.3224,
+      "step": 9180
+    },
+    {
+      "epoch": 1.1762447203378983,
+      "grad_norm": 1.1182880401611328,
+      "learning_rate": 0.0002975508511798648,
+      "loss": 5.3045,
+      "step": 9190
+    },
+    {
+      "epoch": 1.177524638423141,
+      "grad_norm": 1.6098095178604126,
+      "learning_rate": 0.00029754541385458046,
+      "loss": 5.3356,
+      "step": 9200
+    },
+    {
+      "epoch": 1.1788045565083833,
+      "grad_norm": 1.4355725049972534,
+      "learning_rate": 0.0002975399705501076,
+      "loss": 5.3156,
+      "step": 9210
+    },
+    {
+      "epoch": 1.180084474593626,
+      "grad_norm": 1.7407658100128174,
+      "learning_rate": 0.0002975345212666666,
+      "loss": 5.3086,
+      "step": 9220
+    },
+    {
+      "epoch": 1.1813643926788686,
+      "grad_norm": 1.4330238103866577,
+      "learning_rate": 0.00029752906600447846,
+      "loss": 5.3063,
+      "step": 9230
+    },
+    {
+      "epoch": 1.1826443107641111,
+      "grad_norm": 1.4022783041000366,
+      "learning_rate": 0.0002975236047637642,
+      "loss": 5.312,
+      "step": 9240
+    },
+    {
+      "epoch": 1.1839242288493537,
+      "grad_norm": 1.6006667613983154,
+      "learning_rate": 0.0002975181375447451,
+      "loss": 5.3361,
+      "step": 9250
+    },
+    {
+      "epoch": 1.1852041469345962,
+      "grad_norm": 1.2444195747375488,
+      "learning_rate": 0.00029751266434764276,
+      "loss": 5.2986,
+      "step": 9260
+    },
+    {
+      "epoch": 1.1864840650198387,
+      "grad_norm": 1.4725539684295654,
+      "learning_rate": 0.0002975071851726789,
+      "loss": 5.2853,
+      "step": 9270
+    },
+    {
+      "epoch": 1.1877639831050812,
+      "grad_norm": 1.2328054904937744,
+      "learning_rate": 0.0002975017000200757,
+      "loss": 5.3049,
+      "step": 9280
+    },
+    {
+      "epoch": 1.1890439011903238,
+      "grad_norm": 1.7641607522964478,
+      "learning_rate": 0.00029749620889005533,
+      "loss": 5.3093,
+      "step": 9290
+    },
+    {
+      "epoch": 1.1903238192755663,
+      "grad_norm": 1.6767491102218628,
+      "learning_rate": 0.0002974907117828403,
+      "loss": 5.2992,
+      "step": 9300
+    },
+    {
+      "epoch": 1.1916037373608088,
+      "grad_norm": 1.505649209022522,
+      "learning_rate": 0.00029748520869865347,
+      "loss": 5.3143,
+      "step": 9310
+    },
+    {
+      "epoch": 1.1928836554460513,
+      "grad_norm": 1.1579383611679077,
+      "learning_rate": 0.0002974796996377178,
+      "loss": 5.3144,
+      "step": 9320
+    },
+    {
+      "epoch": 1.194163573531294,
+      "grad_norm": 1.7368600368499756,
+      "learning_rate": 0.0002974741846002565,
+      "loss": 5.306,
+      "step": 9330
+    },
+    {
+      "epoch": 1.1954434916165366,
+      "grad_norm": 1.1850640773773193,
+      "learning_rate": 0.0002974686635864931,
+      "loss": 5.3167,
+      "step": 9340
+    },
+    {
+      "epoch": 1.1967234097017792,
+      "grad_norm": 1.958350658416748,
+      "learning_rate": 0.00029746313659665145,
+      "loss": 5.3047,
+      "step": 9350
+    },
+    {
+      "epoch": 1.1980033277870217,
+      "grad_norm": 1.1464362144470215,
+      "learning_rate": 0.00029745760363095533,
+      "loss": 5.2955,
+      "step": 9360
+    },
+    {
+      "epoch": 1.1992832458722642,
+      "grad_norm": 1.2057669162750244,
+      "learning_rate": 0.00029745206468962906,
+      "loss": 5.3028,
+      "step": 9370
+    },
+    {
+      "epoch": 1.2005631639575067,
+      "grad_norm": 1.2068003416061401,
+      "learning_rate": 0.00029744651977289707,
+      "loss": 5.3062,
+      "step": 9380
+    },
+    {
+      "epoch": 1.2018430820427493,
+      "grad_norm": 1.2438029050827026,
+      "learning_rate": 0.0002974409688809842,
+      "loss": 5.2763,
+      "step": 9390
+    },
+    {
+      "epoch": 1.2031230001279918,
+      "grad_norm": 1.4242688417434692,
+      "learning_rate": 0.00029743541201411513,
+      "loss": 5.3177,
+      "step": 9400
+    },
+    {
+      "epoch": 1.2044029182132343,
+      "grad_norm": 1.4365496635437012,
+      "learning_rate": 0.0002974298491725153,
+      "loss": 5.3017,
+      "step": 9410
+    },
+    {
+      "epoch": 1.2056828362984768,
+      "grad_norm": 1.5808701515197754,
+      "learning_rate": 0.00029742428035641,
+      "loss": 5.3031,
+      "step": 9420
+    },
+    {
+      "epoch": 1.2069627543837194,
+      "grad_norm": 6.06550931930542,
+      "learning_rate": 0.00029741870556602496,
+      "loss": 5.3044,
+      "step": 9430
+    },
+    {
+      "epoch": 1.208242672468962,
+      "grad_norm": 1.2524161338806152,
+      "learning_rate": 0.00029741312480158606,
+      "loss": 5.3139,
+      "step": 9440
+    },
+    {
+      "epoch": 1.2095225905542044,
+      "grad_norm": 1.8837299346923828,
+      "learning_rate": 0.00029740753806331956,
+      "loss": 5.2954,
+      "step": 9450
+    },
+    {
+      "epoch": 1.2108025086394472,
+      "grad_norm": 1.8030625581741333,
+      "learning_rate": 0.00029740194535145167,
+      "loss": 5.302,
+      "step": 9460
+    },
+    {
+      "epoch": 1.2120824267246897,
+      "grad_norm": 1.4304263591766357,
+      "learning_rate": 0.00029739634666620915,
+      "loss": 5.2882,
+      "step": 9470
+    },
+    {
+      "epoch": 1.2133623448099322,
+      "grad_norm": 1.8116546869277954,
+      "learning_rate": 0.00029739074200781893,
+      "loss": 5.3042,
+      "step": 9480
+    },
+    {
+      "epoch": 1.2146422628951747,
+      "grad_norm": 1.4493547677993774,
+      "learning_rate": 0.00029738513137650803,
+      "loss": 5.2958,
+      "step": 9490
+    },
+    {
+      "epoch": 1.2159221809804173,
+      "grad_norm": 3.873800754547119,
+      "learning_rate": 0.00029737951477250393,
+      "loss": 5.309,
+      "step": 9500
+    },
+    {
+      "epoch": 1.2172020990656598,
+      "grad_norm": 2.4254534244537354,
+      "learning_rate": 0.0002973738921960341,
+      "loss": 5.327,
+      "step": 9510
+    },
+    {
+      "epoch": 1.2184820171509023,
+      "grad_norm": 1.3041757345199585,
+      "learning_rate": 0.00029736826364732647,
+      "loss": 5.3251,
+      "step": 9520
+    },
+    {
+      "epoch": 1.2197619352361448,
+      "grad_norm": 2.2851760387420654,
+      "learning_rate": 0.00029736262912660917,
+      "loss": 5.3101,
+      "step": 9530
+    },
+    {
+      "epoch": 1.2210418533213874,
+      "grad_norm": 1.7676900625228882,
+      "learning_rate": 0.0002973569886341105,
+      "loss": 5.2925,
+      "step": 9540
+    },
+    {
+      "epoch": 1.22232177140663,
+      "grad_norm": 1.6757118701934814,
+      "learning_rate": 0.000297351342170059,
+      "loss": 5.278,
+      "step": 9550
+    },
+    {
+      "epoch": 1.2236016894918724,
+      "grad_norm": 258.9942321777344,
+      "learning_rate": 0.00029734568973468356,
+      "loss": 5.3329,
+      "step": 9560
+    },
+    {
+      "epoch": 1.2248816075771152,
+      "grad_norm": 67.85020446777344,
+      "learning_rate": 0.00029734003132821313,
+      "loss": 5.4291,
+      "step": 9570
+    },
+    {
+      "epoch": 1.2261615256623577,
+      "grad_norm": 172.93600463867188,
+      "learning_rate": 0.00029733436695087714,
+      "loss": 5.4633,
+      "step": 9580
+    },
+    {
+      "epoch": 1.2274414437476002,
+      "grad_norm": 59.954322814941406,
+      "learning_rate": 0.0002973286966029051,
+      "loss": 5.4602,
+      "step": 9590
+    },
+    {
+      "epoch": 1.2287213618328428,
+      "grad_norm": 418357.375,
+      "learning_rate": 0.0002973230202845267,
+      "loss": 5.4023,
+      "step": 9600
+    },
+    {
+      "epoch": 1.2300012799180853,
+      "grad_norm": 83.09380340576172,
+      "learning_rate": 0.00029731733799597215,
+      "loss": 5.4068,
+      "step": 9610
+    },
+    {
+      "epoch": 1.2312811980033278,
+      "grad_norm": 3.8474719524383545,
+      "learning_rate": 0.0002973116497374715,
+      "loss": 5.4837,
+      "step": 9620
+    },
+    {
+      "epoch": 1.2325611160885703,
+      "grad_norm": 6.988565444946289,
+      "learning_rate": 0.00029730595550925547,
+      "loss": 5.4765,
+      "step": 9630
+    },
+    {
+      "epoch": 1.2338410341738129,
+      "grad_norm": 8.367043495178223,
+      "learning_rate": 0.0002973002553115547,
+      "loss": 5.4381,
+      "step": 9640
+    },
+    {
+      "epoch": 1.2351209522590554,
+      "grad_norm": 4.276803970336914,
+      "learning_rate": 0.0002972945491446002,
+      "loss": 5.4551,
+      "step": 9650
+    },
+    {
+      "epoch": 1.236400870344298,
+      "grad_norm": 72.82675170898438,
+      "learning_rate": 0.00029728883700862326,
+      "loss": 5.4738,
+      "step": 9660
+    },
+    {
+      "epoch": 1.2376807884295404,
+      "grad_norm": 6.083178520202637,
+      "learning_rate": 0.0002972831189038553,
+      "loss": 5.4763,
+      "step": 9670
+    },
+    {
+      "epoch": 1.238960706514783,
+      "grad_norm": 1073.8052978515625,
+      "learning_rate": 0.0002972773948305281,
+      "loss": 5.4571,
+      "step": 9680
+    },
+    {
+      "epoch": 1.2402406246000255,
+      "grad_norm": 217.56842041015625,
+      "learning_rate": 0.0002972716647888736,
+      "loss": 5.5419,
+      "step": 9690
+    },
+    {
+      "epoch": 1.2415205426852682,
+      "grad_norm": 33.87669372558594,
+      "learning_rate": 0.000297265928779124,
+      "loss": 5.5087,
+      "step": 9700
+    },
+    {
+      "epoch": 1.2428004607705108,
+      "grad_norm": 12.367305755615234,
+      "learning_rate": 0.0002972601868015117,
+      "loss": 5.442,
+      "step": 9710
+    },
+    {
+      "epoch": 1.2440803788557533,
+      "grad_norm": 274.2508850097656,
+      "learning_rate": 0.0002972544388562694,
+      "loss": 5.4061,
+      "step": 9720
+    },
+    {
+      "epoch": 1.2453602969409958,
+      "grad_norm": 2.3056957721710205,
+      "learning_rate": 0.00029724868494363014,
+      "loss": 5.437,
+      "step": 9730
+    },
+    {
+      "epoch": 1.2466402150262383,
+      "grad_norm": 1.763442873954773,
+      "learning_rate": 0.00029724292506382705,
+      "loss": 5.396,
+      "step": 9740
+    },
+    {
+      "epoch": 1.2479201331114809,
+      "grad_norm": 1.2737253904342651,
+      "learning_rate": 0.00029723715921709345,
+      "loss": 5.3564,
+      "step": 9750
+    },
+    {
+      "epoch": 1.2492000511967234,
+      "grad_norm": 1.6267307996749878,
+      "learning_rate": 0.0002972313874036631,
+      "loss": 5.3609,
+      "step": 9760
+    },
+    {
+      "epoch": 1.250479969281966,
+      "grad_norm": 1.832618236541748,
+      "learning_rate": 0.00029722560962376984,
+      "loss": 5.344,
+      "step": 9770
+    },
+    {
+      "epoch": 1.2517598873672084,
+      "grad_norm": 1.918068289756775,
+      "learning_rate": 0.0002972198258776479,
+      "loss": 5.3326,
+      "step": 9780
+    },
+    {
+      "epoch": 1.253039805452451,
+      "grad_norm": 1.8401589393615723,
+      "learning_rate": 0.0002972140361655315,
+      "loss": 5.3075,
+      "step": 9790
+    },
+    {
+      "epoch": 1.2543197235376935,
+      "grad_norm": 1.50496244430542,
+      "learning_rate": 0.0002972082404876555,
+      "loss": 5.3139,
+      "step": 9800
+    },
+    {
+      "epoch": 1.2555996416229362,
+      "grad_norm": 1.468178629875183,
+      "learning_rate": 0.0002972024388442545,
+      "loss": 5.3077,
+      "step": 9810
+    },
+    {
+      "epoch": 1.2568795597081786,
+      "grad_norm": 1.319603443145752,
+      "learning_rate": 0.0002971966312355638,
+      "loss": 5.3103,
+      "step": 9820
+    },
+    {
+      "epoch": 1.2581594777934213,
+      "grad_norm": 1.3174471855163574,
+      "learning_rate": 0.00029719081766181864,
+      "loss": 5.2859,
+      "step": 9830
+    },
+    {
+      "epoch": 1.2594393958786638,
+      "grad_norm": 1.354108452796936,
+      "learning_rate": 0.00029718499812325476,
+      "loss": 5.315,
+      "step": 9840
+    },
+    {
+      "epoch": 1.2607193139639064,
+      "grad_norm": 1.3756756782531738,
+      "learning_rate": 0.00029717917262010783,
+      "loss": 5.2826,
+      "step": 9850
+    },
+    {
+      "epoch": 1.2619992320491489,
+      "grad_norm": 2.1197121143341064,
+      "learning_rate": 0.00029717334115261394,
+      "loss": 5.2865,
+      "step": 9860
+    },
+    {
+      "epoch": 1.2632791501343914,
+      "grad_norm": 1.1051539182662964,
+      "learning_rate": 0.00029716750372100955,
+      "loss": 5.3189,
+      "step": 9870
+    },
+    {
+      "epoch": 1.264559068219634,
+      "grad_norm": 7.60437536239624,
+      "learning_rate": 0.00029716166032553107,
+      "loss": 5.3171,
+      "step": 9880
+    },
+    {
+      "epoch": 1.2658389863048765,
+      "grad_norm": 392.87750244140625,
+      "learning_rate": 0.00029715581096641534,
+      "loss": 5.4252,
+      "step": 9890
+    },
+    {
+      "epoch": 1.267118904390119,
+      "grad_norm": 63.391754150390625,
+      "learning_rate": 0.00029714995564389943,
+      "loss": 5.5073,
+      "step": 9900
+    },
+    {
+      "epoch": 1.2683988224753615,
+      "grad_norm": 268.2895202636719,
+      "learning_rate": 0.0002971440943582206,
+      "loss": 5.4236,
+      "step": 9910
+    },
+    {
+      "epoch": 1.2696787405606043,
+      "grad_norm": 55.89106369018555,
+      "learning_rate": 0.00029713822710961645,
+      "loss": 5.4003,
+      "step": 9920
+    },
+    {
+      "epoch": 1.2709586586458466,
+      "grad_norm": 13.6283540725708,
+      "learning_rate": 0.0002971323538983246,
+      "loss": 5.4117,
+      "step": 9930
+    },
+    {
+      "epoch": 1.2722385767310893,
+      "grad_norm": 9.581747055053711,
+      "learning_rate": 0.0002971264747245832,
+      "loss": 5.4548,
+      "step": 9940
+    },
+    {
+      "epoch": 1.2735184948163318,
+      "grad_norm": 5.135045528411865,
+      "learning_rate": 0.00029712058958863036,
+      "loss": 5.4945,
+      "step": 9950
+    },
+    {
+      "epoch": 1.2747984129015744,
+      "grad_norm": 13.211899757385254,
+      "learning_rate": 0.00029711469849070474,
+      "loss": 5.4774,
+      "step": 9960
+    },
+    {
+      "epoch": 1.2760783309868169,
+      "grad_norm": 48.73270034790039,
+      "learning_rate": 0.00029710880143104493,
+      "loss": 5.4555,
+      "step": 9970
+    },
+    {
+      "epoch": 1.2773582490720594,
+      "grad_norm": 7.845920085906982,
+      "learning_rate": 0.00029710289840988993,
+      "loss": 5.5258,
+      "step": 9980
+    },
+    {
+      "epoch": 1.278638167157302,
+      "grad_norm": 6.870781421661377,
+      "learning_rate": 0.000297096989427479,
+      "loss": 5.4432,
+      "step": 9990
+    },
+    {
+      "epoch": 1.2799180852425445,
+      "grad_norm": 10.027936935424805,
+      "learning_rate": 0.0002970910744840516,
+      "loss": 5.4615,
+      "step": 10000
+    },
+    {
+      "epoch": 1.281198003327787,
+      "grad_norm": 40.80441665649414,
+      "learning_rate": 0.00029708515357984745,
+      "loss": 5.4906,
+      "step": 10010
+    },
+    {
+      "epoch": 1.2824779214130295,
+      "grad_norm": 11.978614807128906,
+      "learning_rate": 0.0002970792267151064,
+      "loss": 5.4575,
+      "step": 10020
+    },
+    {
+      "epoch": 1.283757839498272,
+      "grad_norm": 76.15137481689453,
+      "learning_rate": 0.0002970732938900687,
+      "loss": 5.4267,
+      "step": 10030
+    },
+    {
+      "epoch": 1.2850377575835146,
+      "grad_norm": 23.728790283203125,
+      "learning_rate": 0.00029706735510497476,
+      "loss": 5.4251,
+      "step": 10040
+    },
+    {
+      "epoch": 1.2863176756687573,
+      "grad_norm": 15.487527847290039,
+      "learning_rate": 0.00029706141036006526,
+      "loss": 5.4004,
+      "step": 10050
+    },
+    {
+      "epoch": 1.2875975937539996,
+      "grad_norm": 160.41036987304688,
+      "learning_rate": 0.0002970554596555811,
+      "loss": 5.3944,
+      "step": 10060
+    },
+    {
+      "epoch": 1.2888775118392424,
+      "grad_norm": 14.26937198638916,
+      "learning_rate": 0.0002970495029917634,
+      "loss": 5.4353,
+      "step": 10070
+    },
+    {
+      "epoch": 1.290157429924485,
+      "grad_norm": 21.807947158813477,
+      "learning_rate": 0.00029704354036885354,
+      "loss": 5.4581,
+      "step": 10080
+    },
+    {
+      "epoch": 1.2914373480097274,
+      "grad_norm": 20.17401885986328,
+      "learning_rate": 0.0002970375717870932,
+      "loss": 5.4164,
+      "step": 10090
+    },
+    {
+      "epoch": 1.29271726609497,
+      "grad_norm": 26.464344024658203,
+      "learning_rate": 0.00029703159724672425,
+      "loss": 5.4066,
+      "step": 10100
+    },
+    {
+      "epoch": 1.2939971841802125,
+      "grad_norm": 63.260562896728516,
+      "learning_rate": 0.0002970256167479888,
+      "loss": 5.4306,
+      "step": 10110
+    },
+    {
+      "epoch": 1.295277102265455,
+      "grad_norm": 270.7586669921875,
+      "learning_rate": 0.00029701963029112917,
+      "loss": 5.4221,
+      "step": 10120
+    },
+    {
+      "epoch": 1.2965570203506975,
+      "grad_norm": 104.90503692626953,
+      "learning_rate": 0.00029701363787638805,
+      "loss": 5.5096,
+      "step": 10130
+    },
+    {
+      "epoch": 1.29783693843594,
+      "grad_norm": 8.484354019165039,
+      "learning_rate": 0.0002970076395040082,
+      "loss": 5.5256,
+      "step": 10140
+    },
+    {
+      "epoch": 1.2991168565211826,
+      "grad_norm": 8.284499168395996,
+      "learning_rate": 0.0002970016351742327,
+      "loss": 5.4761,
+      "step": 10150
+    },
+    {
+      "epoch": 1.300396774606425,
+      "grad_norm": 3.632566213607788,
+      "learning_rate": 0.0002969956248873048,
+      "loss": 5.4266,
+      "step": 10160
+    },
+    {
+      "epoch": 1.3016766926916676,
+      "grad_norm": 9.005770683288574,
+      "learning_rate": 0.0002969896086434682,
+      "loss": 5.4402,
+      "step": 10170
+    },
+    {
+      "epoch": 1.3029566107769104,
+      "grad_norm": 9.097681045532227,
+      "learning_rate": 0.0002969835864429667,
+      "loss": 5.4615,
+      "step": 10180
+    },
+    {
+      "epoch": 1.3042365288621527,
+      "grad_norm": 12.6643705368042,
+      "learning_rate": 0.00029697755828604427,
+      "loss": 5.4443,
+      "step": 10190
+    },
+    {
+      "epoch": 1.3055164469473954,
+      "grad_norm": 236.66493225097656,
+      "learning_rate": 0.0002969715241729452,
+      "loss": 5.4242,
+      "step": 10200
+    },
+    {
+      "epoch": 1.306796365032638,
+      "grad_norm": 60.66410446166992,
+      "learning_rate": 0.0002969654841039141,
+      "loss": 5.46,
+      "step": 10210
+    },
+    {
+      "epoch": 1.3080762831178805,
+      "grad_norm": 9.005767822265625,
+      "learning_rate": 0.0002969594380791956,
+      "loss": 5.4581,
+      "step": 10220
+    },
+    {
+      "epoch": 1.309356201203123,
+      "grad_norm": 24.40799903869629,
+      "learning_rate": 0.0002969533860990349,
+      "loss": 5.4408,
+      "step": 10230
+    },
+    {
+      "epoch": 1.3106361192883655,
+      "grad_norm": 44.29862976074219,
+      "learning_rate": 0.0002969473281636771,
+      "loss": 5.4332,
+      "step": 10240
+    },
+    {
+      "epoch": 1.311916037373608,
+      "grad_norm": 9.148470878601074,
+      "learning_rate": 0.00029694126427336773,
+      "loss": 5.4249,
+      "step": 10250
+    },
+    {
+      "epoch": 1.3131959554588506,
+      "grad_norm": 3.238877296447754,
+      "learning_rate": 0.00029693519442835256,
+      "loss": 5.4235,
+      "step": 10260
+    },
+    {
+      "epoch": 1.3144758735440931,
+      "grad_norm": 140.43626403808594,
+      "learning_rate": 0.0002969291186288776,
+      "loss": 5.4636,
+      "step": 10270
+    },
+    {
+      "epoch": 1.3157557916293356,
+      "grad_norm": 19.890974044799805,
+      "learning_rate": 0.0002969230368751889,
+      "loss": 5.4817,
+      "step": 10280
+    },
+    {
+      "epoch": 1.3170357097145784,
+      "grad_norm": 60.46494674682617,
+      "learning_rate": 0.00029691694916753313,
+      "loss": 5.453,
+      "step": 10290
+    },
+    {
+      "epoch": 1.3183156277998207,
+      "grad_norm": 3.0658810138702393,
+      "learning_rate": 0.00029691085550615686,
+      "loss": 5.4608,
+      "step": 10300
+    },
+    {
+      "epoch": 1.3195955458850634,
+      "grad_norm": 2.8574109077453613,
+      "learning_rate": 0.0002969047558913071,
+      "loss": 5.4497,
+      "step": 10310
+    },
+    {
+      "epoch": 1.320875463970306,
+      "grad_norm": 5.469709873199463,
+      "learning_rate": 0.0002968986503232309,
+      "loss": 5.4928,
+      "step": 10320
+    },
+    {
+      "epoch": 1.3221553820555485,
+      "grad_norm": 9.013448715209961,
+      "learning_rate": 0.0002968925388021759,
+      "loss": 5.4647,
+      "step": 10330
+    },
+    {
+      "epoch": 1.323435300140791,
+      "grad_norm": 5.318248748779297,
+      "learning_rate": 0.00029688642132838955,
+      "loss": 5.4425,
+      "step": 10340
+    },
+    {
+      "epoch": 1.3247152182260336,
+      "grad_norm": 10.692891120910645,
+      "learning_rate": 0.00029688029790211993,
+      "loss": 5.4341,
+      "step": 10350
+    },
+    {
+      "epoch": 1.325995136311276,
+      "grad_norm": 11.036212921142578,
+      "learning_rate": 0.00029687416852361507,
+      "loss": 5.4143,
+      "step": 10360
+    },
+    {
+      "epoch": 1.3272750543965186,
+      "grad_norm": 8.127765655517578,
+      "learning_rate": 0.0002968680331931234,
+      "loss": 5.4221,
+      "step": 10370
+    },
+    {
+      "epoch": 1.3285549724817611,
+      "grad_norm": 8.590612411499023,
+      "learning_rate": 0.00029686189191089365,
+      "loss": 5.4127,
+      "step": 10380
+    },
+    {
+      "epoch": 1.3298348905670037,
+      "grad_norm": 23.215730667114258,
+      "learning_rate": 0.00029685574467717453,
+      "loss": 5.3911,
+      "step": 10390
+    },
+    {
+      "epoch": 1.3311148086522462,
+      "grad_norm": 16.342304229736328,
+      "learning_rate": 0.0002968495914922152,
+      "loss": 5.4043,
+      "step": 10400
+    },
+    {
+      "epoch": 1.3323947267374887,
+      "grad_norm": 119.0894546508789,
+      "learning_rate": 0.0002968434323562651,
+      "loss": 5.3897,
+      "step": 10410
+    },
+    {
+      "epoch": 1.3336746448227315,
+      "grad_norm": 3.1613729000091553,
+      "learning_rate": 0.00029683726726957365,
+      "loss": 5.4513,
+      "step": 10420
+    },
+    {
+      "epoch": 1.3349545629079738,
+      "grad_norm": 17.995119094848633,
+      "learning_rate": 0.0002968310962323909,
+      "loss": 5.4527,
+      "step": 10430
+    },
+    {
+      "epoch": 1.3362344809932165,
+      "grad_norm": 2.353546380996704,
+      "learning_rate": 0.00029682491924496677,
+      "loss": 5.4369,
+      "step": 10440
+    },
+    {
+      "epoch": 1.337514399078459,
+      "grad_norm": 3.29575514793396,
+      "learning_rate": 0.0002968187363075516,
+      "loss": 5.4181,
+      "step": 10450
+    },
+    {
+      "epoch": 1.3387943171637016,
+      "grad_norm": 42.947044372558594,
+      "learning_rate": 0.00029681254742039604,
+      "loss": 5.4268,
+      "step": 10460
+    },
+    {
+      "epoch": 1.340074235248944,
+      "grad_norm": 9.198704719543457,
+      "learning_rate": 0.0002968063525837508,
+      "loss": 5.4315,
+      "step": 10470
+    },
+    {
+      "epoch": 1.3413541533341866,
+      "grad_norm": 3.822139263153076,
+      "learning_rate": 0.000296800151797867,
+      "loss": 5.4177,
+      "step": 10480
+    },
+    {
+      "epoch": 1.3426340714194291,
+      "grad_norm": 197.9595947265625,
+      "learning_rate": 0.00029679394506299585,
+      "loss": 5.3682,
+      "step": 10490
+    },
+    {
+      "epoch": 1.3439139895046717,
+      "grad_norm": 25.585834503173828,
+      "learning_rate": 0.0002967877323793889,
+      "loss": 5.4032,
+      "step": 10500
+    },
+    {
+      "epoch": 1.3451939075899142,
+      "grad_norm": 8.87588882446289,
+      "learning_rate": 0.0002967815137472979,
+      "loss": 5.461,
+      "step": 10510
+    },
+    {
+      "epoch": 1.3464738256751567,
+      "grad_norm": 2.577976942062378,
+      "learning_rate": 0.0002967752891669749,
+      "loss": 5.4129,
+      "step": 10520
+    },
+    {
+      "epoch": 1.3477537437603995,
+      "grad_norm": 3.7082831859588623,
+      "learning_rate": 0.0002967690586386721,
+      "loss": 5.4013,
+      "step": 10530
+    },
+    {
+      "epoch": 1.3490336618456418,
+      "grad_norm": 2.4679198265075684,
+      "learning_rate": 0.00029676282216264206,
+      "loss": 5.4289,
+      "step": 10540
+    },
+    {
+      "epoch": 1.3503135799308845,
+      "grad_norm": 4.0808563232421875,
+      "learning_rate": 0.0002967565797391374,
+      "loss": 5.4205,
+      "step": 10550
+    },
+    {
+      "epoch": 1.351593498016127,
+      "grad_norm": 2.7137279510498047,
+      "learning_rate": 0.0002967503313684112,
+      "loss": 5.382,
+      "step": 10560
+    },
+    {
+      "epoch": 1.3528734161013696,
+      "grad_norm": 2.2462680339813232,
+      "learning_rate": 0.00029674407705071655,
+      "loss": 5.3692,
+      "step": 10570
+    },
+    {
+      "epoch": 1.354153334186612,
+      "grad_norm": 42.56100082397461,
+      "learning_rate": 0.00029673781678630703,
+      "loss": 5.3739,
+      "step": 10580
+    },
+    {
+      "epoch": 1.3554332522718546,
+      "grad_norm": 1.6741000413894653,
+      "learning_rate": 0.00029673155057543626,
+      "loss": 5.3682,
+      "step": 10590
+    },
+    {
+      "epoch": 1.3567131703570972,
+      "grad_norm": 3.09238338470459,
+      "learning_rate": 0.00029672527841835815,
+      "loss": 5.3518,
+      "step": 10600
+    },
+    {
+      "epoch": 1.3579930884423397,
+      "grad_norm": 4.391119003295898,
+      "learning_rate": 0.00029671900031532695,
+      "loss": 5.3582,
+      "step": 10610
+    },
+    {
+      "epoch": 1.3592730065275822,
+      "grad_norm": 9.450124740600586,
+      "learning_rate": 0.00029671271626659706,
+      "loss": 5.3564,
+      "step": 10620
+    },
+    {
+      "epoch": 1.3605529246128247,
+      "grad_norm": 7.369607925415039,
+      "learning_rate": 0.00029670642627242316,
+      "loss": 5.3938,
+      "step": 10630
+    },
+    {
+      "epoch": 1.3618328426980673,
+      "grad_norm": 30.87764549255371,
+      "learning_rate": 0.00029670013033306,
+      "loss": 5.9874,
+      "step": 10640
+    },
+    {
+      "epoch": 1.3631127607833098,
+      "grad_norm": 40.88727951049805,
+      "learning_rate": 0.0002966938284487629,
+      "loss": 5.6569,
+      "step": 10650
+    },
+    {
+      "epoch": 1.3643926788685525,
+      "grad_norm": 32.33999252319336,
+      "learning_rate": 0.00029668752061978715,
+      "loss": 5.5244,
+      "step": 10660
+    },
+    {
+      "epoch": 1.3656725969537948,
+      "grad_norm": 28.833730697631836,
+      "learning_rate": 0.00029668120684638835,
+      "loss": 5.5229,
+      "step": 10670
+    },
+    {
+      "epoch": 1.3669525150390376,
+      "grad_norm": 3.893641233444214,
+      "learning_rate": 0.00029667488712882247,
+      "loss": 5.4523,
+      "step": 10680
+    },
+    {
+      "epoch": 1.3682324331242801,
+      "grad_norm": 4.776241302490234,
+      "learning_rate": 0.0002966685614673455,
+      "loss": 5.4346,
+      "step": 10690
+    },
+    {
+      "epoch": 1.3695123512095226,
+      "grad_norm": 13.851577758789062,
+      "learning_rate": 0.00029666222986221384,
+      "loss": 5.3944,
+      "step": 10700
+    },
+    {
+      "epoch": 1.3707922692947652,
+      "grad_norm": 2.147908926010132,
+      "learning_rate": 0.00029665589231368404,
+      "loss": 5.3899,
+      "step": 10710
+    },
+    {
+      "epoch": 1.3720721873800077,
+      "grad_norm": 9.758132934570312,
+      "learning_rate": 0.00029664954882201297,
+      "loss": 5.3847,
+      "step": 10720
+    },
+    {
+      "epoch": 1.3733521054652502,
+      "grad_norm": 3.455718755722046,
+      "learning_rate": 0.00029664319938745763,
+      "loss": 5.3817,
+      "step": 10730
+    },
+    {
+      "epoch": 1.3746320235504927,
+      "grad_norm": 24.70989227294922,
+      "learning_rate": 0.00029663684401027543,
+      "loss": 5.4051,
+      "step": 10740
+    },
+    {
+      "epoch": 1.3759119416357353,
+      "grad_norm": 10.136948585510254,
+      "learning_rate": 0.00029663048269072385,
+      "loss": 5.392,
+      "step": 10750
+    },
+    {
+      "epoch": 1.3771918597209778,
+      "grad_norm": 3.9122562408447266,
+      "learning_rate": 0.0002966241154290607,
+      "loss": 5.3557,
+      "step": 10760
+    },
+    {
+      "epoch": 1.3784717778062203,
+      "grad_norm": 3.816096544265747,
+      "learning_rate": 0.0002966177422255439,
+      "loss": 5.3742,
+      "step": 10770
+    },
+    {
+      "epoch": 1.3797516958914628,
+      "grad_norm": 2.2476353645324707,
+      "learning_rate": 0.0002966113630804319,
+      "loss": 5.3807,
+      "step": 10780
+    },
+    {
+      "epoch": 1.3810316139767056,
+      "grad_norm": 3.042263984680176,
+      "learning_rate": 0.00029660497799398306,
+      "loss": 5.3864,
+      "step": 10790
+    },
+    {
+      "epoch": 1.382311532061948,
+      "grad_norm": 45.76136779785156,
+      "learning_rate": 0.00029659858696645626,
+      "loss": 5.3699,
+      "step": 10800
+    },
+    {
+      "epoch": 1.3835914501471906,
+      "grad_norm": 13.073348999023438,
+      "learning_rate": 0.00029659218999811043,
+      "loss": 5.387,
+      "step": 10810
+    },
+    {
+      "epoch": 1.3848713682324332,
+      "grad_norm": 8.278294563293457,
+      "learning_rate": 0.0002965857870892048,
+      "loss": 5.3712,
+      "step": 10820
+    },
+    {
+      "epoch": 1.3861512863176757,
+      "grad_norm": 2.5932791233062744,
+      "learning_rate": 0.0002965793782399988,
+      "loss": 5.3435,
+      "step": 10830
+    },
+    {
+      "epoch": 1.3874312044029182,
+      "grad_norm": 20.342578887939453,
+      "learning_rate": 0.0002965729634507522,
+      "loss": 5.3446,
+      "step": 10840
+    },
+    {
+      "epoch": 1.3887111224881608,
+      "grad_norm": 2.0133111476898193,
+      "learning_rate": 0.000296566542721725,
+      "loss": 5.3574,
+      "step": 10850
+    },
+    {
+      "epoch": 1.3899910405734033,
+      "grad_norm": 7.2772216796875,
+      "learning_rate": 0.0002965601160531773,
+      "loss": 5.3525,
+      "step": 10860
+    },
+    {
+      "epoch": 1.3912709586586458,
+      "grad_norm": 3.3040051460266113,
+      "learning_rate": 0.0002965536834453696,
+      "loss": 5.346,
+      "step": 10870
+    },
+    {
+      "epoch": 1.3925508767438883,
+      "grad_norm": 36.88441848754883,
+      "learning_rate": 0.00029654724489856256,
+      "loss": 5.3156,
+      "step": 10880
+    },
+    {
+      "epoch": 1.3938307948291309,
+      "grad_norm": 9.932327270507812,
+      "learning_rate": 0.0002965408004130171,
+      "loss": 5.3503,
+      "step": 10890
+    },
+    {
+      "epoch": 1.3951107129143736,
+      "grad_norm": 2.004136323928833,
+      "learning_rate": 0.0002965343499889943,
+      "loss": 5.3873,
+      "step": 10900
+    },
+    {
+      "epoch": 1.396390630999616,
+      "grad_norm": 7.710193634033203,
+      "learning_rate": 0.00029652789362675575,
+      "loss": 5.3892,
+      "step": 10910
+    },
+    {
+      "epoch": 1.3976705490848587,
+      "grad_norm": 4.5125532150268555,
+      "learning_rate": 0.0002965214313265629,
+      "loss": 5.3947,
+      "step": 10920
+    },
+    {
+      "epoch": 1.3989504671701012,
+      "grad_norm": 3.5431599617004395,
+      "learning_rate": 0.00029651496308867776,
+      "loss": 5.4125,
+      "step": 10930
+    },
+    {
+      "epoch": 1.4002303852553437,
+      "grad_norm": 3.65902042388916,
+      "learning_rate": 0.00029650848891336234,
+      "loss": 5.391,
+      "step": 10940
+    },
+    {
+      "epoch": 1.4015103033405862,
+      "grad_norm": 13.032547950744629,
+      "learning_rate": 0.0002965020088008791,
+      "loss": 5.3605,
+      "step": 10950
+    },
+    {
+      "epoch": 1.4027902214258288,
+      "grad_norm": 4.853715419769287,
+      "learning_rate": 0.0002964955227514906,
+      "loss": 5.3678,
+      "step": 10960
+    },
+    {
+      "epoch": 1.4040701395110713,
+      "grad_norm": 1.706592321395874,
+      "learning_rate": 0.00029648903076545967,
+      "loss": 5.3235,
+      "step": 10970
+    },
+    {
+      "epoch": 1.4053500575963138,
+      "grad_norm": 1.9886835813522339,
+      "learning_rate": 0.0002964825328430494,
+      "loss": 5.3405,
+      "step": 10980
+    },
+    {
+      "epoch": 1.4066299756815563,
+      "grad_norm": 2.2458877563476562,
+      "learning_rate": 0.00029647602898452317,
+      "loss": 5.3212,
+      "step": 10990
+    },
+    {
+      "epoch": 1.4079098937667989,
+      "grad_norm": 1.449478268623352,
+      "learning_rate": 0.0002964695191901445,
+      "loss": 5.309,
+      "step": 11000
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 156260,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 20,
+  "save_steps": 1000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.4622869463232352e+19,
+  "train_batch_size": 32,
+  "trial_name": null,
+  "trial_params": null
+}