{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 4.999574721442545,
  "eval_steps": 500,
  "global_step": 29390,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0017011142298205325,
      "grad_norm": 8.1242036819458,
      "learning_rate": 1.524500175587422e-07,
      "loss": 4.5114,
      "step": 10
    },
    {
      "epoch": 0.003402228459641065,
      "grad_norm": 9.024007797241211,
      "learning_rate": 3.049000351174844e-07,
      "loss": 4.2617,
      "step": 20
    },
    {
      "epoch": 0.0051033426894615975,
      "grad_norm": 7.401093006134033,
      "learning_rate": 4.5735005267622656e-07,
      "loss": 4.5417,
      "step": 30
    },
    {
      "epoch": 0.00680445691928213,
      "grad_norm": 8.14155101776123,
      "learning_rate": 6.098000702349687e-07,
      "loss": 4.5044,
      "step": 40
    },
    {
      "epoch": 0.008505571149102662,
      "grad_norm": 6.201820373535156,
      "learning_rate": 7.622500877937109e-07,
      "loss": 4.3812,
      "step": 50
    },
    {
      "epoch": 0.010206685378923195,
      "grad_norm": 8.932755470275879,
      "learning_rate": 9.147001053524531e-07,
      "loss": 4.2968,
      "step": 60
    },
    {
      "epoch": 0.011907799608743727,
      "grad_norm": 7.06637716293335,
      "learning_rate": 1.0671501229111953e-06,
      "loss": 4.1351,
      "step": 70
    },
    {
      "epoch": 0.01360891383856426,
      "grad_norm": 8.479144096374512,
      "learning_rate": 1.2196001404699375e-06,
      "loss": 4.1493,
      "step": 80
    },
    {
      "epoch": 0.015310028068384792,
      "grad_norm": 8.668302536010742,
      "learning_rate": 1.3720501580286797e-06,
      "loss": 3.9006,
      "step": 90
    },
    {
      "epoch": 0.017011142298205325,
      "grad_norm": 7.180272579193115,
      "learning_rate": 1.5245001755874219e-06,
      "loss": 3.932,
      "step": 100
    },
    {
      "epoch": 0.018712256528025856,
      "grad_norm": 8.624896049499512,
      "learning_rate": 1.6769501931461638e-06,
      "loss": 3.888,
      "step": 110
    },
    {
      "epoch": 0.02041337075784639,
      "grad_norm": 7.598726272583008,
      "learning_rate": 1.8294002107049062e-06,
      "loss": 3.7444,
      "step": 120
    },
    {
      "epoch": 0.02211448498766692,
      "grad_norm": 7.890979290008545,
      "learning_rate": 1.9818502282636482e-06,
      "loss": 3.7337,
      "step": 130
    },
    {
      "epoch": 0.023815599217487455,
      "grad_norm": 6.916543960571289,
      "learning_rate": 2.1343002458223906e-06,
      "loss": 3.4046,
      "step": 140
    },
    {
      "epoch": 0.025516713447307986,
      "grad_norm": 8.39304256439209,
      "learning_rate": 2.2867502633811326e-06,
      "loss": 3.3585,
      "step": 150
    },
    {
      "epoch": 0.02721782767712852,
      "grad_norm": 10.390460968017578,
      "learning_rate": 2.439200280939875e-06,
      "loss": 3.1896,
      "step": 160
    },
    {
      "epoch": 0.02891894190694905,
      "grad_norm": 6.750011920928955,
      "learning_rate": 2.591650298498617e-06,
      "loss": 3.0233,
      "step": 170
    },
    {
      "epoch": 0.030620056136769585,
      "grad_norm": 7.048929691314697,
      "learning_rate": 2.7441003160573594e-06,
      "loss": 3.0611,
      "step": 180
    },
    {
      "epoch": 0.032321170366590116,
      "grad_norm": 5.937943458557129,
      "learning_rate": 2.8965503336161013e-06,
      "loss": 2.8964,
      "step": 190
    },
    {
      "epoch": 0.03402228459641065,
      "grad_norm": 4.653339862823486,
      "learning_rate": 3.0490003511748437e-06,
      "loss": 2.7357,
      "step": 200
    },
    {
      "epoch": 0.035723398826231184,
      "grad_norm": 2.970280885696411,
      "learning_rate": 3.2014503687335857e-06,
      "loss": 2.5882,
      "step": 210
    },
    {
      "epoch": 0.03742451305605171,
      "grad_norm": 2.7442290782928467,
      "learning_rate": 3.3539003862923277e-06,
      "loss": 2.6236,
      "step": 220
    },
    {
      "epoch": 0.039125627285872246,
      "grad_norm": 2.7237141132354736,
      "learning_rate": 3.50635040385107e-06,
      "loss": 2.6678,
      "step": 230
    },
    {
      "epoch": 0.04082674151569278,
      "grad_norm": 2.7791314125061035,
      "learning_rate": 3.6588004214098125e-06,
      "loss": 2.6291,
      "step": 240
    },
    {
      "epoch": 0.042527855745513314,
      "grad_norm": 2.6336171627044678,
      "learning_rate": 3.8112504389685545e-06,
      "loss": 2.5533,
      "step": 250
    },
    {
      "epoch": 0.04422896997533384,
      "grad_norm": 2.9241085052490234,
      "learning_rate": 3.9637004565272964e-06,
      "loss": 2.4647,
      "step": 260
    },
    {
      "epoch": 0.045930084205154376,
      "grad_norm": 3.389188289642334,
      "learning_rate": 4.116150474086039e-06,
      "loss": 2.414,
      "step": 270
    },
    {
      "epoch": 0.04763119843497491,
      "grad_norm": 3.224665641784668,
      "learning_rate": 4.268600491644781e-06,
      "loss": 2.472,
      "step": 280
    },
    {
      "epoch": 0.049332312664795444,
      "grad_norm": 3.153616189956665,
      "learning_rate": 4.421050509203523e-06,
      "loss": 2.4434,
      "step": 290
    },
    {
      "epoch": 0.05103342689461597,
      "grad_norm": 3.551027774810791,
      "learning_rate": 4.573500526762265e-06,
      "loss": 2.4699,
      "step": 300
    },
    {
      "epoch": 0.052734541124436506,
      "grad_norm": 3.2293620109558105,
      "learning_rate": 4.725950544321008e-06,
      "loss": 2.3638,
      "step": 310
    },
    {
      "epoch": 0.05443565535425704,
      "grad_norm": 2.619767665863037,
      "learning_rate": 4.87840056187975e-06,
      "loss": 2.3004,
      "step": 320
    },
    {
      "epoch": 0.056136769584077574,
      "grad_norm": 3.1208813190460205,
      "learning_rate": 5.030850579438492e-06,
      "loss": 2.2602,
      "step": 330
    },
    {
      "epoch": 0.0578378838138981,
      "grad_norm": 3.030879259109497,
      "learning_rate": 5.183300596997234e-06,
      "loss": 2.2759,
      "step": 340
    },
    {
      "epoch": 0.059538998043718636,
      "grad_norm": 3.025357723236084,
      "learning_rate": 5.335750614555976e-06,
      "loss": 2.3363,
      "step": 350
    },
    {
      "epoch": 0.06124011227353917,
      "grad_norm": 2.8321595191955566,
      "learning_rate": 5.488200632114719e-06,
      "loss": 2.3441,
      "step": 360
    },
    {
      "epoch": 0.0629412265033597,
      "grad_norm": 2.955768346786499,
      "learning_rate": 5.640650649673461e-06,
      "loss": 2.3207,
      "step": 370
    },
    {
      "epoch": 0.06464234073318023,
      "grad_norm": 2.877669095993042,
      "learning_rate": 5.793100667232203e-06,
      "loss": 2.26,
      "step": 380
    },
    {
      "epoch": 0.06634345496300076,
      "grad_norm": 2.615410566329956,
      "learning_rate": 5.945550684790945e-06,
      "loss": 2.2476,
      "step": 390
    },
    {
      "epoch": 0.0680445691928213,
      "grad_norm": 2.787527322769165,
      "learning_rate": 6.0980007023496875e-06,
      "loss": 2.205,
      "step": 400
    },
    {
      "epoch": 0.06974568342264183,
      "grad_norm": 2.9535741806030273,
      "learning_rate": 6.2504507199084295e-06,
      "loss": 2.2943,
      "step": 410
    },
    {
      "epoch": 0.07144679765246237,
      "grad_norm": 3.091892719268799,
      "learning_rate": 6.4029007374671714e-06,
      "loss": 2.3296,
      "step": 420
    },
    {
      "epoch": 0.0731479118822829,
      "grad_norm": 3.5832443237304688,
      "learning_rate": 6.555350755025914e-06,
      "loss": 2.2661,
      "step": 430
    },
    {
      "epoch": 0.07484902611210342,
      "grad_norm": 3.0930492877960205,
      "learning_rate": 6.707800772584655e-06,
      "loss": 2.2862,
      "step": 440
    },
    {
      "epoch": 0.07655014034192396,
      "grad_norm": 3.2023990154266357,
      "learning_rate": 6.860250790143398e-06,
      "loss": 2.301,
      "step": 450
    },
    {
      "epoch": 0.07825125457174449,
      "grad_norm": 3.0322978496551514,
      "learning_rate": 7.01270080770214e-06,
      "loss": 2.1735,
      "step": 460
    },
    {
      "epoch": 0.07995236880156502,
      "grad_norm": 3.60652232170105,
      "learning_rate": 7.165150825260882e-06,
      "loss": 2.2559,
      "step": 470
    },
    {
      "epoch": 0.08165348303138556,
      "grad_norm": 2.929839849472046,
      "learning_rate": 7.317600842819625e-06,
      "loss": 2.2043,
      "step": 480
    },
    {
      "epoch": 0.08335459726120609,
      "grad_norm": 3.3258023262023926,
      "learning_rate": 7.470050860378367e-06,
      "loss": 2.2505,
      "step": 490
    },
    {
      "epoch": 0.08505571149102663,
      "grad_norm": 3.416071653366089,
      "learning_rate": 7.622500877937109e-06,
      "loss": 2.1877,
      "step": 500
    },
    {
      "epoch": 0.08675682572084716,
      "grad_norm": 2.9309732913970947,
      "learning_rate": 7.77495089549585e-06,
      "loss": 2.2404,
      "step": 510
    },
    {
      "epoch": 0.08845793995066768,
      "grad_norm": 3.613666534423828,
      "learning_rate": 7.927400913054593e-06,
      "loss": 2.2869,
      "step": 520
    },
    {
      "epoch": 0.09015905418048822,
      "grad_norm": 4.01043176651001,
      "learning_rate": 8.079850930613336e-06,
      "loss": 2.2802,
      "step": 530
    },
    {
      "epoch": 0.09186016841030875,
      "grad_norm": 3.47981333732605,
      "learning_rate": 8.232300948172079e-06,
      "loss": 2.2599,
      "step": 540
    },
    {
      "epoch": 0.09356128264012928,
      "grad_norm": 2.9870450496673584,
      "learning_rate": 8.38475096573082e-06,
      "loss": 2.239,
      "step": 550
    },
    {
      "epoch": 0.09526239686994982,
      "grad_norm": 2.7835795879364014,
      "learning_rate": 8.537200983289562e-06,
      "loss": 2.2282,
      "step": 560
    },
    {
      "epoch": 0.09696351109977035,
      "grad_norm": 3.1672210693359375,
      "learning_rate": 8.689651000848305e-06,
      "loss": 2.2323,
      "step": 570
    },
    {
      "epoch": 0.09866462532959089,
      "grad_norm": 3.8526771068573,
      "learning_rate": 8.842101018407046e-06,
      "loss": 2.1846,
      "step": 580
    },
    {
      "epoch": 0.10036573955941142,
      "grad_norm": 3.2321290969848633,
      "learning_rate": 8.99455103596579e-06,
      "loss": 2.217,
      "step": 590
    },
    {
      "epoch": 0.10206685378923194,
      "grad_norm": 3.663644790649414,
      "learning_rate": 9.14700105352453e-06,
      "loss": 2.2593,
      "step": 600
    },
    {
      "epoch": 0.10376796801905248,
      "grad_norm": 3.2919960021972656,
      "learning_rate": 9.299451071083273e-06,
      "loss": 2.1836,
      "step": 610
    },
    {
      "epoch": 0.10546908224887301,
      "grad_norm": 3.550652503967285,
      "learning_rate": 9.451901088642016e-06,
      "loss": 2.2548,
      "step": 620
    },
    {
      "epoch": 0.10717019647869354,
      "grad_norm": 3.35884690284729,
      "learning_rate": 9.604351106200757e-06,
      "loss": 2.2009,
      "step": 630
    },
    {
      "epoch": 0.10887131070851408,
      "grad_norm": 3.3775312900543213,
      "learning_rate": 9.7568011237595e-06,
      "loss": 2.2623,
      "step": 640
    },
    {
      "epoch": 0.11057242493833461,
      "grad_norm": 3.32918381690979,
      "learning_rate": 9.909251141318241e-06,
      "loss": 2.1882,
      "step": 650
    },
    {
      "epoch": 0.11227353916815515,
      "grad_norm": 3.5951883792877197,
      "learning_rate": 1.0061701158876984e-05,
      "loss": 2.1755,
      "step": 660
    },
    {
      "epoch": 0.11397465339797568,
      "grad_norm": 3.7706947326660156,
      "learning_rate": 1.0214151176435727e-05,
      "loss": 2.1684,
      "step": 670
    },
    {
      "epoch": 0.1156757676277962,
      "grad_norm": 3.506378412246704,
      "learning_rate": 1.0366601193994468e-05,
      "loss": 2.1198,
      "step": 680
    },
    {
      "epoch": 0.11737688185761674,
      "grad_norm": 3.8458452224731445,
      "learning_rate": 1.051905121155321e-05,
      "loss": 2.2457,
      "step": 690
    },
    {
      "epoch": 0.11907799608743727,
      "grad_norm": 3.462367296218872,
      "learning_rate": 1.0671501229111952e-05,
      "loss": 2.1286,
      "step": 700
    },
    {
      "epoch": 0.1207791103172578,
      "grad_norm": 3.5375754833221436,
      "learning_rate": 1.0823951246670695e-05,
      "loss": 2.1568,
      "step": 710
    },
    {
      "epoch": 0.12248022454707834,
      "grad_norm": 4.020713806152344,
      "learning_rate": 1.0976401264229437e-05,
      "loss": 2.1467,
      "step": 720
    },
    {
      "epoch": 0.12418133877689887,
      "grad_norm": 3.4603312015533447,
      "learning_rate": 1.1128851281788179e-05,
      "loss": 2.1809,
      "step": 730
    },
    {
      "epoch": 0.1258824530067194,
      "grad_norm": 3.5019948482513428,
      "learning_rate": 1.1281301299346921e-05,
      "loss": 2.1061,
      "step": 740
    },
    {
      "epoch": 0.12758356723653994,
      "grad_norm": 3.1017353534698486,
      "learning_rate": 1.1433751316905663e-05,
      "loss": 2.2428,
      "step": 750
    },
    {
      "epoch": 0.12928468146636046,
      "grad_norm": 3.2938976287841797,
      "learning_rate": 1.1586201334464405e-05,
      "loss": 2.2237,
      "step": 760
    },
    {
      "epoch": 0.130985795696181,
      "grad_norm": 3.668274164199829,
      "learning_rate": 1.1738651352023148e-05,
      "loss": 2.1496,
      "step": 770
    },
    {
      "epoch": 0.13268690992600152,
      "grad_norm": 3.8302762508392334,
      "learning_rate": 1.189110136958189e-05,
      "loss": 2.2023,
      "step": 780
    },
    {
      "epoch": 0.13438802415582207,
      "grad_norm": 3.526839017868042,
      "learning_rate": 1.2043551387140632e-05,
      "loss": 2.1561,
      "step": 790
    },
    {
      "epoch": 0.1360891383856426,
      "grad_norm": 2.987562417984009,
      "learning_rate": 1.2196001404699375e-05,
      "loss": 2.1617,
      "step": 800
    },
    {
      "epoch": 0.13779025261546313,
      "grad_norm": 3.2640066146850586,
      "learning_rate": 1.2348451422258116e-05,
      "loss": 2.082,
      "step": 810
    },
    {
      "epoch": 0.13949136684528365,
      "grad_norm": 3.5792763233184814,
      "learning_rate": 1.2500901439816859e-05,
      "loss": 2.1362,
      "step": 820
    },
    {
      "epoch": 0.14119248107510418,
      "grad_norm": 3.7467305660247803,
      "learning_rate": 1.26533514573756e-05,
      "loss": 2.1767,
      "step": 830
    },
    {
      "epoch": 0.14289359530492474,
      "grad_norm": 3.8871824741363525,
      "learning_rate": 1.2805801474934343e-05,
      "loss": 2.1361,
      "step": 840
    },
    {
      "epoch": 0.14459470953474526,
      "grad_norm": 4.198888301849365,
      "learning_rate": 1.2958251492493086e-05,
      "loss": 2.1379,
      "step": 850
    },
    {
      "epoch": 0.1462958237645658,
      "grad_norm": 3.3561158180236816,
      "learning_rate": 1.3110701510051829e-05,
      "loss": 2.0982,
      "step": 860
    },
    {
      "epoch": 0.14799693799438632,
      "grad_norm": 3.4034993648529053,
      "learning_rate": 1.326315152761057e-05,
      "loss": 2.1312,
      "step": 870
    },
    {
      "epoch": 0.14969805222420685,
      "grad_norm": 3.768439531326294,
      "learning_rate": 1.341560154516931e-05,
      "loss": 2.0818,
      "step": 880
    },
    {
      "epoch": 0.1513991664540274,
      "grad_norm": 4.092297554016113,
      "learning_rate": 1.3446088936024867e-05,
      "loss": 2.1015,
      "step": 890
    },
    {
      "epoch": 0.15310028068384793,
      "grad_norm": 3.9114010334014893,
      "learning_rate": 1.344607832211256e-05,
      "loss": 2.1479,
      "step": 900
    },
    {
      "epoch": 0.15480139491366846,
      "grad_norm": 3.8456695079803467,
      "learning_rate": 1.344605954366601e-05,
      "loss": 2.107,
      "step": 910
    },
    {
      "epoch": 0.15650250914348898,
      "grad_norm": 3.279608726501465,
      "learning_rate": 1.3446032600708018e-05,
      "loss": 2.1185,
      "step": 920
    },
    {
      "epoch": 0.1582036233733095,
      "grad_norm": 4.147802829742432,
      "learning_rate": 1.3445997493271305e-05,
      "loss": 2.1371,
      "step": 930
    },
    {
      "epoch": 0.15990473760313004,
      "grad_norm": 3.8674826622009277,
      "learning_rate": 1.3445954221398507e-05,
      "loss": 2.204,
      "step": 940
    },
    {
      "epoch": 0.1616058518329506,
      "grad_norm": 3.8814809322357178,
      "learning_rate": 1.3445902785142174e-05,
      "loss": 2.1347,
      "step": 950
    },
    {
      "epoch": 0.16330696606277112,
      "grad_norm": 3.7449662685394287,
      "learning_rate": 1.344584318456477e-05,
      "loss": 2.0845,
      "step": 960
    },
    {
      "epoch": 0.16500808029259165,
      "grad_norm": 4.558581829071045,
      "learning_rate": 1.3445775419738675e-05,
      "loss": 2.0412,
      "step": 970
    },
    {
      "epoch": 0.16670919452241217,
      "grad_norm": 3.170236825942993,
      "learning_rate": 1.3445699490746183e-05,
      "loss": 2.1486,
      "step": 980
    },
    {
      "epoch": 0.1684103087522327,
      "grad_norm": 3.613136053085327,
      "learning_rate": 1.3445615397679502e-05,
      "loss": 2.0369,
      "step": 990
    },
    {
      "epoch": 0.17011142298205326,
      "grad_norm": 3.954437017440796,
      "learning_rate": 1.3445523140640761e-05,
      "loss": 2.0846,
      "step": 1000
    },
    {
      "epoch": 0.17181253721187378,
      "grad_norm": 3.777693510055542,
      "learning_rate": 1.3445422719741992e-05,
      "loss": 2.1255,
      "step": 1010
    },
    {
      "epoch": 0.1735136514416943,
      "grad_norm": 3.3416199684143066,
      "learning_rate": 1.344531413510515e-05,
      "loss": 2.1268,
      "step": 1020
    },
    {
      "epoch": 0.17521476567151484,
      "grad_norm": 3.943105459213257,
      "learning_rate": 1.3445197386862098e-05,
      "loss": 2.0934,
      "step": 1030
    },
    {
      "epoch": 0.17691587990133537,
      "grad_norm": 4.644258499145508,
      "learning_rate": 1.3445072475154622e-05,
      "loss": 2.0692,
      "step": 1040
    },
    {
      "epoch": 0.17861699413115592,
      "grad_norm": 3.2973544597625732,
      "learning_rate": 1.3444939400134416e-05,
      "loss": 2.1838,
      "step": 1050
    },
    {
      "epoch": 0.18031810836097645,
      "grad_norm": 3.7182607650756836,
      "learning_rate": 1.3444798161963082e-05,
      "loss": 2.0951,
      "step": 1060
    },
    {
      "epoch": 0.18201922259079698,
      "grad_norm": 3.6397881507873535,
      "learning_rate": 1.3444648760812149e-05,
      "loss": 2.1331,
      "step": 1070
    },
    {
      "epoch": 0.1837203368206175,
      "grad_norm": 4.1129021644592285,
      "learning_rate": 1.3444491196863046e-05,
      "loss": 2.0526,
      "step": 1080
    },
    {
      "epoch": 0.18542145105043803,
      "grad_norm": 4.195235729217529,
      "learning_rate": 1.3444325470307122e-05,
      "loss": 2.1748,
      "step": 1090
    },
    {
      "epoch": 0.18712256528025856,
      "grad_norm": 3.746614694595337,
      "learning_rate": 1.344415158134564e-05,
      "loss": 2.1016,
      "step": 1100
    },
    {
      "epoch": 0.1888236795100791,
      "grad_norm": 3.955183982849121,
      "learning_rate": 1.3443969530189769e-05,
      "loss": 2.1415,
      "step": 1110
    },
    {
      "epoch": 0.19052479373989964,
      "grad_norm": 4.204941272735596,
      "learning_rate": 1.3443779317060598e-05,
      "loss": 2.1487,
      "step": 1120
    },
    {
      "epoch": 0.19222590796972017,
      "grad_norm": 3.8602209091186523,
      "learning_rate": 1.344358094218912e-05,
      "loss": 2.0946,
      "step": 1130
    },
    {
      "epoch": 0.1939270221995407,
      "grad_norm": 3.87605881690979,
      "learning_rate": 1.3443374405816248e-05,
      "loss": 2.1114,
      "step": 1140
    },
    {
      "epoch": 0.19562813642936122,
      "grad_norm": 4.242985725402832,
      "learning_rate": 1.3443159708192799e-05,
      "loss": 2.1148,
      "step": 1150
    },
    {
      "epoch": 0.19732925065918178,
      "grad_norm": 3.853022575378418,
      "learning_rate": 1.3442936849579506e-05,
      "loss": 2.0633,
      "step": 1160
    },
    {
      "epoch": 0.1990303648890023,
      "grad_norm": 4.261999130249023,
      "learning_rate": 1.344270583024701e-05,
      "loss": 2.0983,
      "step": 1170
    },
    {
      "epoch": 0.20073147911882283,
      "grad_norm": 3.767235279083252,
      "learning_rate": 1.3442466650475868e-05,
      "loss": 2.1592,
      "step": 1180
    },
    {
      "epoch": 0.20243259334864336,
      "grad_norm": 4.7613205909729,
      "learning_rate": 1.3442219310556538e-05,
      "loss": 2.089,
      "step": 1190
    },
    {
      "epoch": 0.20413370757846389,
      "grad_norm": 3.825859308242798,
      "learning_rate": 1.3441963810789397e-05,
      "loss": 2.1249,
      "step": 1200
    },
    {
      "epoch": 0.2058348218082844,
      "grad_norm": 3.648253917694092,
      "learning_rate": 1.3441700151484721e-05,
      "loss": 2.0832,
      "step": 1210
    },
    {
      "epoch": 0.20753593603810497,
      "grad_norm": 3.996825695037842,
      "learning_rate": 1.3441428332962705e-05,
      "loss": 2.1009,
      "step": 1220
    },
    {
      "epoch": 0.2092370502679255,
      "grad_norm": 3.444505214691162,
      "learning_rate": 1.344114835555345e-05,
      "loss": 2.1229,
      "step": 1230
    },
    {
      "epoch": 0.21093816449774602,
      "grad_norm": 3.9788694381713867,
      "learning_rate": 1.3440860219596964e-05,
      "loss": 2.0401,
      "step": 1240
    },
    {
      "epoch": 0.21263927872756655,
      "grad_norm": 3.823482036590576,
      "learning_rate": 1.3440563925443158e-05,
      "loss": 2.0676,
      "step": 1250
    },
    {
      "epoch": 0.21434039295738708,
      "grad_norm": 3.833479166030884,
      "learning_rate": 1.344025947345186e-05,
      "loss": 2.138,
      "step": 1260
    },
    {
      "epoch": 0.21604150718720763,
      "grad_norm": 3.986558675765991,
      "learning_rate": 1.3439946863992799e-05,
      "loss": 2.0968,
      "step": 1270
    },
    {
      "epoch": 0.21774262141702816,
      "grad_norm": 3.4136123657226562,
      "learning_rate": 1.3439626097445615e-05,
      "loss": 2.0349,
      "step": 1280
    },
    {
      "epoch": 0.2194437356468487,
      "grad_norm": 4.327932357788086,
      "learning_rate": 1.3439297174199843e-05,
      "loss": 2.0767,
      "step": 1290
    },
    {
      "epoch": 0.22114484987666921,
      "grad_norm": 4.868990898132324,
      "learning_rate": 1.3438960094654938e-05,
      "loss": 2.1006,
      "step": 1300
    },
    {
      "epoch": 0.22284596410648974,
      "grad_norm": 4.4454498291015625,
      "learning_rate": 1.3438614859220252e-05,
      "loss": 2.0865,
      "step": 1310
    },
    {
      "epoch": 0.2245470783363103,
      "grad_norm": 4.084188461303711,
      "learning_rate": 1.3438261468315045e-05,
      "loss": 2.1575,
      "step": 1320
    },
    {
      "epoch": 0.22624819256613082,
      "grad_norm": 3.5140950679779053,
      "learning_rate": 1.3437899922368474e-05,
      "loss": 2.112,
      "step": 1330
    },
    {
      "epoch": 0.22794930679595135,
      "grad_norm": 4.531380653381348,
      "learning_rate": 1.3437530221819608e-05,
      "loss": 2.0994,
      "step": 1340
    },
    {
      "epoch": 0.22965042102577188,
      "grad_norm": 4.098215103149414,
      "learning_rate": 1.343715236711742e-05,
      "loss": 2.1288,
      "step": 1350
    },
    {
      "epoch": 0.2313515352555924,
      "grad_norm": 4.336173057556152,
      "learning_rate": 1.3436766358720774e-05,
      "loss": 2.0744,
      "step": 1360
    },
    {
      "epoch": 0.23305264948541293,
      "grad_norm": 4.105205535888672,
      "learning_rate": 1.3436372197098449e-05,
      "loss": 2.2001,
      "step": 1370
    },
    {
      "epoch": 0.2347537637152335,
      "grad_norm": 4.088197708129883,
      "learning_rate": 1.3435969882729115e-05,
      "loss": 2.1039,
      "step": 1380
    },
    {
      "epoch": 0.23645487794505402,
      "grad_norm": 4.016329765319824,
      "learning_rate": 1.343555941610135e-05,
      "loss": 2.1743,
      "step": 1390
    },
    {
      "epoch": 0.23815599217487454,
      "grad_norm": 3.8832547664642334,
      "learning_rate": 1.3435140797713633e-05,
      "loss": 2.0029,
      "step": 1400
    },
    {
      "epoch": 0.23985710640469507,
      "grad_norm": 3.58616304397583,
      "learning_rate": 1.3434714028074338e-05,
      "loss": 2.0792,
      "step": 1410
    },
    {
      "epoch": 0.2415582206345156,
      "grad_norm": 3.839970827102661,
      "learning_rate": 1.3434279107701734e-05,
      "loss": 2.1073,
      "step": 1420
    },
    {
      "epoch": 0.24325933486433615,
      "grad_norm": 4.0038323402404785,
      "learning_rate": 1.3433836037123999e-05,
      "loss": 2.1397,
      "step": 1430
    },
    {
      "epoch": 0.24496044909415668,
      "grad_norm": 4.53272819519043,
      "learning_rate": 1.3433384816879202e-05,
      "loss": 2.155,
      "step": 1440
    },
    {
      "epoch": 0.2466615633239772,
      "grad_norm": 3.404862642288208,
      "learning_rate": 1.3432925447515309e-05,
      "loss": 2.0471,
      "step": 1450
    },
    {
      "epoch": 0.24836267755379773,
      "grad_norm": 3.97717547416687,
      "learning_rate": 1.3432457929590187e-05,
      "loss": 2.1222,
      "step": 1460
    },
    {
      "epoch": 0.25006379178361826,
      "grad_norm": 4.303969383239746,
      "learning_rate": 1.3431982263671594e-05,
      "loss": 2.0641,
      "step": 1470
    },
    {
      "epoch": 0.2517649060134388,
      "grad_norm": 3.92305850982666,
      "learning_rate": 1.3431498450337182e-05,
      "loss": 2.0516,
      "step": 1480
    },
    {
      "epoch": 0.2534660202432593,
      "grad_norm": 3.997673988342285,
      "learning_rate": 1.3431006490174504e-05,
      "loss": 2.0984,
      "step": 1490
    },
    {
      "epoch": 0.25516713447307987,
      "grad_norm": 4.106938362121582,
      "learning_rate": 1.3430506383781e-05,
      "loss": 2.0811,
      "step": 1500
    },
    {
      "epoch": 0.2568682487029004,
      "grad_norm": 4.152285575866699,
      "learning_rate": 1.3429998131764008e-05,
      "loss": 2.0867,
      "step": 1510
    },
    {
      "epoch": 0.2585693629327209,
      "grad_norm": 3.77607798576355,
      "learning_rate": 1.3429481734740751e-05,
      "loss": 2.0138,
      "step": 1520
    },
    {
      "epoch": 0.2602704771625415,
      "grad_norm": 3.8017027378082275,
      "learning_rate": 1.3428957193338354e-05,
      "loss": 2.1102,
      "step": 1530
    },
    {
      "epoch": 0.261971591392362,
      "grad_norm": 4.08286190032959,
      "learning_rate": 1.3428424508193823e-05,
      "loss": 2.1213,
      "step": 1540
    },
    {
      "epoch": 0.26367270562218253,
      "grad_norm": 4.082574367523193,
      "learning_rate": 1.3427883679954056e-05,
      "loss": 1.9399,
      "step": 1550
    },
    {
      "epoch": 0.26537381985200303,
      "grad_norm": 4.214354991912842,
      "learning_rate": 1.3427334709275845e-05,
      "loss": 2.1207,
      "step": 1560
    },
    {
      "epoch": 0.2670749340818236,
      "grad_norm": 3.9730758666992188,
      "learning_rate": 1.3426777596825863e-05,
      "loss": 2.1379,
      "step": 1570
    },
    {
      "epoch": 0.26877604831164414,
      "grad_norm": 3.152505397796631,
      "learning_rate": 1.3426212343280678e-05,
      "loss": 2.1091,
      "step": 1580
    },
    {
      "epoch": 0.27047716254146464,
      "grad_norm": 3.7805302143096924,
      "learning_rate": 1.342563894932674e-05,
      "loss": 2.0486,
      "step": 1590
    },
    {
      "epoch": 0.2721782767712852,
      "grad_norm": 3.5900185108184814,
      "learning_rate": 1.3425057415660384e-05,
      "loss": 2.0382,
      "step": 1600
    },
    {
      "epoch": 0.2738793910011057,
      "grad_norm": 3.6163578033447266,
      "learning_rate": 1.3424467742987833e-05,
      "loss": 2.052,
      "step": 1610
    },
    {
      "epoch": 0.27558050523092625,
      "grad_norm": 4.680826187133789,
      "learning_rate": 1.3423869932025193e-05,
      "loss": 2.0847,
      "step": 1620
    },
    {
      "epoch": 0.2772816194607468,
      "grad_norm": 4.031846046447754,
      "learning_rate": 1.3423263983498454e-05,
      "loss": 2.0639,
      "step": 1630
    },
    {
      "epoch": 0.2789827336905673,
      "grad_norm": 3.679631471633911,
      "learning_rate": 1.3422649898143484e-05,
      "loss": 1.9937,
      "step": 1640
    },
    {
      "epoch": 0.28068384792038786,
      "grad_norm": 4.050079345703125,
      "learning_rate": 1.3422027676706038e-05,
      "loss": 2.0676,
      "step": 1650
    },
    {
      "epoch": 0.28238496215020836,
      "grad_norm": 4.256223678588867,
      "learning_rate": 1.342139731994175e-05,
      "loss": 2.0506,
      "step": 1660
    },
    {
      "epoch": 0.2840860763800289,
      "grad_norm": 4.579814434051514,
      "learning_rate": 1.3420758828616134e-05,
      "loss": 2.0044,
      "step": 1670
    },
    {
      "epoch": 0.2857871906098495,
      "grad_norm": 4.356644153594971,
      "learning_rate": 1.342011220350458e-05,
      "loss": 2.0261,
      "step": 1680
    },
    {
      "epoch": 0.28748830483967,
      "grad_norm": 3.2434017658233643,
      "learning_rate": 1.3419457445392356e-05,
      "loss": 2.045,
      "step": 1690
    },
    {
      "epoch": 0.2891894190694905,
      "grad_norm": 4.5175323486328125,
      "learning_rate": 1.3418794555074613e-05,
      "loss": 1.9873,
      "step": 1700
    },
    {
      "epoch": 0.290890533299311,
      "grad_norm": 3.9216480255126953,
      "learning_rate": 1.3418123533356368e-05,
      "loss": 2.0781,
      "step": 1710
    },
    {
      "epoch": 0.2925916475291316,
      "grad_norm": 4.742781162261963,
      "learning_rate": 1.3417444381052523e-05,
      "loss": 2.028,
      "step": 1720
    },
    {
      "epoch": 0.29429276175895214,
      "grad_norm": 3.6140594482421875,
      "learning_rate": 1.3416757098987845e-05,
      "loss": 2.1452,
      "step": 1730
    },
    {
      "epoch": 0.29599387598877264,
      "grad_norm": 4.244185447692871,
      "learning_rate": 1.3416061687996979e-05,
      "loss": 2.0183,
      "step": 1740
    },
    {
      "epoch": 0.2976949902185932,
      "grad_norm": 4.613598823547363,
      "learning_rate": 1.3415358148924444e-05,
      "loss": 2.0209,
      "step": 1750
    },
    {
      "epoch": 0.2993961044484137,
      "grad_norm": 3.735178232192993,
      "learning_rate": 1.3414646482624621e-05,
      "loss": 2.1145,
      "step": 1760
    },
    {
      "epoch": 0.30109721867823425,
      "grad_norm": 3.9842472076416016,
      "learning_rate": 1.341392668996177e-05,
      "loss": 1.9653,
      "step": 1770
    },
    {
      "epoch": 0.3027983329080548,
      "grad_norm": 4.243276119232178,
      "learning_rate": 1.3413198771810016e-05,
      "loss": 2.057,
      "step": 1780
    },
    {
      "epoch": 0.3044994471378753,
      "grad_norm": 4.055215358734131,
      "learning_rate": 1.3412462729053352e-05,
      "loss": 2.1238,
      "step": 1790
    },
    {
      "epoch": 0.30620056136769586,
      "grad_norm": 3.9792640209198,
      "learning_rate": 1.3411718562585636e-05,
      "loss": 2.0256,
      "step": 1800
    },
    {
      "epoch": 0.30790167559751636,
      "grad_norm": 4.385422229766846,
      "learning_rate": 1.3410966273310594e-05,
      "loss": 2.0131,
      "step": 1810
    },
    {
      "epoch": 0.3096027898273369,
      "grad_norm": 4.1103291511535645,
      "learning_rate": 1.3410205862141814e-05,
      "loss": 2.0393,
      "step": 1820
    },
    {
      "epoch": 0.31130390405715747,
      "grad_norm": 3.9303791522979736,
      "learning_rate": 1.340943733000275e-05,
      "loss": 2.0703,
      "step": 1830
    },
    {
      "epoch": 0.31300501828697797,
      "grad_norm": 4.296981334686279,
      "learning_rate": 1.3408660677826718e-05,
      "loss": 2.0678,
      "step": 1840
    },
    {
      "epoch": 0.3147061325167985,
      "grad_norm": 3.9926674365997314,
      "learning_rate": 1.3407875906556892e-05,
      "loss": 2.0054,
      "step": 1850
    },
    {
      "epoch": 0.316407246746619,
      "grad_norm": 4.080948829650879,
      "learning_rate": 1.340708301714631e-05,
      "loss": 2.07,
      "step": 1860
    },
    {
      "epoch": 0.3181083609764396,
      "grad_norm": 4.116609573364258,
      "learning_rate": 1.3406282010557866e-05,
      "loss": 2.0026,
      "step": 1870
    },
    {
      "epoch": 0.3198094752062601,
      "grad_norm": 4.65035343170166,
      "learning_rate": 1.3405472887764311e-05,
      "loss": 2.0698,
      "step": 1880
    },
    {
      "epoch": 0.32151058943608063,
      "grad_norm": 3.91961407661438,
      "learning_rate": 1.3404655649748256e-05,
      "loss": 1.9695,
      "step": 1890
    },
    {
      "epoch": 0.3232117036659012,
      "grad_norm": 3.994669198989868,
      "learning_rate": 1.3403830297502163e-05,
      "loss": 2.004,
      "step": 1900
    },
    {
      "epoch": 0.3249128178957217,
      "grad_norm": 3.5440726280212402,
      "learning_rate": 1.340299683202835e-05,
      "loss": 2.0409,
      "step": 1910
    },
    {
      "epoch": 0.32661393212554224,
      "grad_norm": 4.20074462890625,
      "learning_rate": 1.3402155254338988e-05,
      "loss": 2.0632,
      "step": 1920
    },
    {
      "epoch": 0.32831504635536274,
      "grad_norm": 4.328433990478516,
      "learning_rate": 1.3401305565456102e-05,
      "loss": 1.9827,
      "step": 1930
    },
    {
      "epoch": 0.3300161605851833,
      "grad_norm": 3.7840631008148193,
      "learning_rate": 1.3400447766411563e-05,
      "loss": 2.1697,
      "step": 1940
    },
    {
      "epoch": 0.33171727481500385,
      "grad_norm": 4.054955005645752,
      "learning_rate": 1.3399581858247094e-05,
      "loss": 2.0077,
      "step": 1950
    },
    {
      "epoch": 0.33341838904482435,
      "grad_norm": 4.07275390625,
      "learning_rate": 1.339870784201426e-05,
      "loss": 2.011,
      "step": 1960
    },
    {
      "epoch": 0.3351195032746449,
      "grad_norm": 3.921703815460205,
      "learning_rate": 1.339782571877448e-05,
      "loss": 2.1012,
      "step": 1970
    },
    {
      "epoch": 0.3368206175044654,
      "grad_norm": 3.991225481033325,
      "learning_rate": 1.3396935489599019e-05,
      "loss": 2.0105,
      "step": 1980
    },
    {
      "epoch": 0.33852173173428596,
      "grad_norm": 4.40214204788208,
      "learning_rate": 1.3396037155568978e-05,
      "loss": 2.0321,
      "step": 1990
    },
    {
      "epoch": 0.3402228459641065,
      "grad_norm": 4.139034748077393,
      "learning_rate": 1.3395130717775306e-05,
      "loss": 2.1542,
      "step": 2000
    },
    {
      "epoch": 0.341923960193927,
      "grad_norm": 4.642544746398926,
      "learning_rate": 1.3394216177318792e-05,
      "loss": 2.0085,
      "step": 2010
    },
    {
      "epoch": 0.34362507442374757,
      "grad_norm": 4.030486583709717,
      "learning_rate": 1.3393293535310068e-05,
      "loss": 2.1126,
      "step": 2020
    },
    {
      "epoch": 0.34532618865356807,
      "grad_norm": 4.032608985900879,
      "learning_rate": 1.3392362792869597e-05,
      "loss": 2.0684,
      "step": 2030
    },
    {
      "epoch": 0.3470273028833886,
      "grad_norm": 4.151008605957031,
      "learning_rate": 1.3391423951127687e-05,
      "loss": 2.0391,
      "step": 2040
    },
    {
      "epoch": 0.3487284171132092,
      "grad_norm": 4.639594554901123,
      "learning_rate": 1.3390477011224483e-05,
      "loss": 2.0073,
      "step": 2050
    },
    {
      "epoch": 0.3504295313430297,
      "grad_norm": 4.20526123046875,
      "learning_rate": 1.3389521974309951e-05,
      "loss": 2.0737,
      "step": 2060
    },
    {
      "epoch": 0.35213064557285023,
      "grad_norm": 3.4363558292388916,
      "learning_rate": 1.3388558841543909e-05,
      "loss": 2.0779,
      "step": 2070
    },
    {
      "epoch": 0.35383175980267073,
      "grad_norm": 3.939847946166992,
      "learning_rate": 1.3387587614095993e-05,
      "loss": 2.0052,
      "step": 2080
    },
    {
      "epoch": 0.3555328740324913,
      "grad_norm": 3.8116230964660645,
      "learning_rate": 1.3386608293145675e-05,
      "loss": 2.0368,
      "step": 2090
    },
    {
      "epoch": 0.35723398826231184,
      "grad_norm": 3.318373680114746,
      "learning_rate": 1.3385620879882251e-05,
      "loss": 2.083,
      "step": 2100
    },
    {
      "epoch": 0.35893510249213234,
      "grad_norm": 3.8586812019348145,
      "learning_rate": 1.3384625375504853e-05,
      "loss": 2.0442,
      "step": 2110
    },
    {
      "epoch": 0.3606362167219529,
      "grad_norm": 3.684936046600342,
      "learning_rate": 1.3383621781222432e-05,
      "loss": 2.0002,
      "step": 2120
    },
    {
      "epoch": 0.3623373309517734,
      "grad_norm": 4.2225213050842285,
      "learning_rate": 1.3382610098253764e-05,
      "loss": 1.9229,
      "step": 2130
    },
    {
      "epoch": 0.36403844518159395,
      "grad_norm": 4.452572345733643,
      "learning_rate": 1.3381590327827451e-05,
      "loss": 2.0504,
      "step": 2140
    },
    {
      "epoch": 0.36573955941141445,
      "grad_norm": 4.429842472076416,
      "learning_rate": 1.3380562471181914e-05,
      "loss": 2.0126,
      "step": 2150
    },
    {
      "epoch": 0.367440673641235,
      "grad_norm": 4.385965347290039,
      "learning_rate": 1.3379526529565396e-05,
      "loss": 2.0278,
      "step": 2160
    },
    {
      "epoch": 0.36914178787105556,
      "grad_norm": 3.4593615531921387,
      "learning_rate": 1.3378482504235955e-05,
      "loss": 1.9504,
      "step": 2170
    },
    {
      "epoch": 0.37084290210087606,
      "grad_norm": 3.8733391761779785,
      "learning_rate": 1.3377430396461471e-05,
      "loss": 2.0475,
      "step": 2180
    },
    {
      "epoch": 0.3725440163306966,
      "grad_norm": 3.7339272499084473,
      "learning_rate": 1.3376370207519636e-05,
      "loss": 1.9485,
      "step": 2190
    },
    {
      "epoch": 0.3742451305605171,
      "grad_norm": 3.994168519973755,
      "learning_rate": 1.3375301938697957e-05,
      "loss": 1.9755,
      "step": 2200
    },
    {
      "epoch": 0.37594624479033767,
      "grad_norm": 3.77097749710083,
      "learning_rate": 1.3374225591293751e-05,
      "loss": 1.9572,
      "step": 2210
    },
    {
      "epoch": 0.3776473590201582,
      "grad_norm": 3.773251533508301,
      "learning_rate": 1.3373141166614147e-05,
      "loss": 2.1328,
      "step": 2220
    },
    {
      "epoch": 0.3793484732499787,
      "grad_norm": 4.05300235748291,
      "learning_rate": 1.3372048665976088e-05,
      "loss": 2.0099,
      "step": 2230
    },
    {
      "epoch": 0.3810495874797993,
      "grad_norm": 4.33394193649292,
      "learning_rate": 1.3370948090706318e-05,
      "loss": 2.0302,
      "step": 2240
    },
    {
      "epoch": 0.3827507017096198,
      "grad_norm": 3.647555351257324,
      "learning_rate": 1.3369839442141388e-05,
      "loss": 2.0583,
      "step": 2250
    },
    {
      "epoch": 0.38445181593944033,
      "grad_norm": 3.822615385055542,
      "learning_rate": 1.3368722721627658e-05,
      "loss": 2.0677,
      "step": 2260
    },
    {
      "epoch": 0.3861529301692609,
      "grad_norm": 3.414644956588745,
      "learning_rate": 1.336759793052128e-05,
      "loss": 1.9975,
      "step": 2270
    },
    {
      "epoch": 0.3878540443990814,
      "grad_norm": 3.7595229148864746,
      "learning_rate": 1.3366465070188226e-05,
      "loss": 1.9567,
      "step": 2280
    },
    {
      "epoch": 0.38955515862890194,
      "grad_norm": 4.140371799468994,
      "learning_rate": 1.3365324142004245e-05,
      "loss": 1.971,
      "step": 2290
    },
    {
      "epoch": 0.39125627285872244,
      "grad_norm": 3.429995059967041,
      "learning_rate": 1.33641751473549e-05,
      "loss": 2.0038,
      "step": 2300
    },
    {
      "epoch": 0.392957387088543,
      "grad_norm": 3.883551836013794,
      "learning_rate": 1.3363018087635543e-05,
      "loss": 2.0446,
      "step": 2310
    },
    {
      "epoch": 0.39465850131836355,
      "grad_norm": 3.6643548011779785,
      "learning_rate": 1.3361852964251321e-05,
      "loss": 2.1035,
      "step": 2320
    },
    {
      "epoch": 0.39635961554818405,
      "grad_norm": 4.243268966674805,
      "learning_rate": 1.3360679778617173e-05,
      "loss": 2.0463,
      "step": 2330
    },
    {
      "epoch": 0.3980607297780046,
      "grad_norm": 4.2195868492126465,
      "learning_rate": 1.3359498532157835e-05,
      "loss": 1.9562,
      "step": 2340
    },
    {
      "epoch": 0.3997618440078251,
      "grad_norm": 3.8276352882385254,
      "learning_rate": 1.3358309226307824e-05,
      "loss": 1.9243,
      "step": 2350
    },
    {
      "epoch": 0.40146295823764566,
      "grad_norm": 4.186319828033447,
      "learning_rate": 1.3357111862511447e-05,
      "loss": 2.041,
      "step": 2360
    },
    {
      "epoch": 0.4031640724674662,
      "grad_norm": 3.921633005142212,
      "learning_rate": 1.3355906442222798e-05,
      "loss": 2.0424,
      "step": 2370
    },
    {
      "epoch": 0.4048651866972867,
      "grad_norm": 3.9087321758270264,
      "learning_rate": 1.3354692966905756e-05,
      "loss": 2.0559,
      "step": 2380
    },
    {
      "epoch": 0.40656630092710727,
      "grad_norm": 4.365992069244385,
      "learning_rate": 1.335347143803398e-05,
      "loss": 2.0072,
      "step": 2390
    },
    {
      "epoch": 0.40826741515692777,
      "grad_norm": 4.397064208984375,
      "learning_rate": 1.335224185709091e-05,
      "loss": 2.0557,
      "step": 2400
    },
    {
      "epoch": 0.4099685293867483,
      "grad_norm": 3.6686463356018066,
      "learning_rate": 1.3351004225569762e-05,
      "loss": 1.9476,
      "step": 2410
    },
    {
      "epoch": 0.4116696436165688,
      "grad_norm": 4.134527683258057,
      "learning_rate": 1.3349758544973535e-05,
      "loss": 2.0976,
      "step": 2420
    },
    {
      "epoch": 0.4133707578463894,
      "grad_norm": 4.738413333892822,
      "learning_rate": 1.3348504816814996e-05,
      "loss": 2.0496,
      "step": 2430
    },
    {
      "epoch": 0.41507187207620994,
      "grad_norm": 3.523717164993286,
      "learning_rate": 1.3347243042616687e-05,
      "loss": 2.0754,
      "step": 2440
    },
    {
      "epoch": 0.41677298630603044,
      "grad_norm": 3.527294874191284,
      "learning_rate": 1.3345973223910927e-05,
      "loss": 2.0003,
      "step": 2450
    },
    {
      "epoch": 0.418474100535851,
      "grad_norm": 3.812926769256592,
      "learning_rate": 1.3344695362239794e-05,
      "loss": 2.0657,
      "step": 2460
    },
    {
      "epoch": 0.4201752147656715,
      "grad_norm": 4.237487316131592,
      "learning_rate": 1.3343409459155144e-05,
      "loss": 2.0629,
      "step": 2470
    },
    {
      "epoch": 0.42187632899549204,
      "grad_norm": 3.8257393836975098,
      "learning_rate": 1.3342115516218592e-05,
      "loss": 2.0618,
      "step": 2480
    },
    {
      "epoch": 0.4235774432253126,
      "grad_norm": 4.0557861328125,
      "learning_rate": 1.3340813535001515e-05,
      "loss": 1.9491,
      "step": 2490
    },
    {
      "epoch": 0.4252785574551331,
      "grad_norm": 3.6852915287017822,
      "learning_rate": 1.333950351708506e-05,
      "loss": 2.1553,
      "step": 2500
    },
    {
      "epoch": 0.42697967168495365,
      "grad_norm": 5.004347801208496,
      "learning_rate": 1.3338185464060127e-05,
      "loss": 1.9624,
      "step": 2510
    },
    {
      "epoch": 0.42868078591477415,
      "grad_norm": 4.5740647315979,
      "learning_rate": 1.3336859377527376e-05,
      "loss": 2.0589,
      "step": 2520
    },
    {
      "epoch": 0.4303819001445947,
      "grad_norm": 4.380794048309326,
      "learning_rate": 1.3335525259097222e-05,
      "loss": 2.018,
      "step": 2530
    },
    {
      "epoch": 0.43208301437441526,
      "grad_norm": 4.0294599533081055,
      "learning_rate": 1.3334183110389837e-05,
      "loss": 1.9966,
      "step": 2540
    },
    {
      "epoch": 0.43378412860423576,
      "grad_norm": 4.073001861572266,
      "learning_rate": 1.3332832933035143e-05,
      "loss": 2.0794,
      "step": 2550
    },
    {
      "epoch": 0.4354852428340563,
      "grad_norm": 3.7871289253234863,
      "learning_rate": 1.333147472867281e-05,
      "loss": 2.0538,
      "step": 2560
    },
    {
      "epoch": 0.4371863570638768,
      "grad_norm": 4.166225433349609,
      "learning_rate": 1.3330108498952259e-05,
      "loss": 2.0601,
      "step": 2570
    },
    {
      "epoch": 0.4388874712936974,
      "grad_norm": 4.1680169105529785,
      "learning_rate": 1.3328734245532657e-05,
      "loss": 2.0239,
      "step": 2580
    },
    {
      "epoch": 0.44058858552351793,
      "grad_norm": 4.646623611450195,
      "learning_rate": 1.3327351970082916e-05,
      "loss": 2.0568,
      "step": 2590
    },
    {
      "epoch": 0.44228969975333843,
      "grad_norm": 4.190902233123779,
      "learning_rate": 1.3325961674281686e-05,
      "loss": 1.964,
      "step": 2600
    },
    {
      "epoch": 0.443990813983159,
      "grad_norm": 3.622678756713867,
      "learning_rate": 1.332456335981736e-05,
      "loss": 2.0088,
      "step": 2610
    },
    {
      "epoch": 0.4456919282129795,
      "grad_norm": 4.435421943664551,
      "learning_rate": 1.3323157028388067e-05,
      "loss": 1.9484,
      "step": 2620
    },
    {
      "epoch": 0.44739304244280004,
      "grad_norm": 4.374332904815674,
      "learning_rate": 1.3321742681701679e-05,
      "loss": 2.0214,
      "step": 2630
    },
    {
      "epoch": 0.4490941566726206,
      "grad_norm": 4.046358585357666,
      "learning_rate": 1.3320320321475788e-05,
      "loss": 2.1021,
      "step": 2640
    },
    {
      "epoch": 0.4507952709024411,
      "grad_norm": 4.689764499664307,
      "learning_rate": 1.3318889949437733e-05,
      "loss": 2.036,
      "step": 2650
    },
    {
      "epoch": 0.45249638513226165,
      "grad_norm": 4.060450553894043,
      "learning_rate": 1.3317451567324572e-05,
      "loss": 1.9854,
      "step": 2660
    },
    {
      "epoch": 0.45419749936208215,
      "grad_norm": 3.979417562484741,
      "learning_rate": 1.3316005176883095e-05,
      "loss": 2.0304,
      "step": 2670
    },
    {
      "epoch": 0.4558986135919027,
      "grad_norm": 3.943890333175659,
      "learning_rate": 1.3314550779869815e-05,
      "loss": 2.0657,
      "step": 2680
    },
    {
      "epoch": 0.4575997278217232,
      "grad_norm": 4.07530403137207,
      "learning_rate": 1.3313088378050974e-05,
      "loss": 2.0117,
      "step": 2690
    },
    {
      "epoch": 0.45930084205154376,
      "grad_norm": 4.679013729095459,
      "learning_rate": 1.3311617973202526e-05,
      "loss": 2.0441,
      "step": 2700
    },
    {
      "epoch": 0.4610019562813643,
      "grad_norm": 4.465762615203857,
      "learning_rate": 1.331013956711015e-05,
      "loss": 2.0589,
      "step": 2710
    },
    {
      "epoch": 0.4627030705111848,
      "grad_norm": 4.215399265289307,
      "learning_rate": 1.3308653161569244e-05,
      "loss": 2.0401,
      "step": 2720
    },
    {
      "epoch": 0.46440418474100537,
      "grad_norm": 4.507813930511475,
      "learning_rate": 1.3307158758384914e-05,
      "loss": 2.0423,
      "step": 2730
    },
    {
      "epoch": 0.46610529897082587,
      "grad_norm": 4.728775978088379,
      "learning_rate": 1.3305656359371983e-05,
      "loss": 2.0764,
      "step": 2740
    },
    {
      "epoch": 0.4678064132006464,
      "grad_norm": 3.851613998413086,
      "learning_rate": 1.3304145966354982e-05,
      "loss": 2.0171,
      "step": 2750
    },
    {
      "epoch": 0.469507527430467,
      "grad_norm": 4.450353622436523,
      "learning_rate": 1.3302627581168152e-05,
      "loss": 1.9752,
      "step": 2760
    },
    {
      "epoch": 0.4712086416602875,
      "grad_norm": 4.43014669418335,
      "learning_rate": 1.3301101205655442e-05,
      "loss": 1.9331,
      "step": 2770
    },
    {
      "epoch": 0.47290975589010803,
      "grad_norm": 4.224061012268066,
      "learning_rate": 1.3299566841670496e-05,
      "loss": 2.0289,
      "step": 2780
    },
    {
      "epoch": 0.47461087011992853,
      "grad_norm": 4.00301456451416,
      "learning_rate": 1.3298024491076665e-05,
      "loss": 1.9951,
      "step": 2790
    },
    {
      "epoch": 0.4763119843497491,
      "grad_norm": 3.680476427078247,
      "learning_rate": 1.3296474155747003e-05,
      "loss": 2.0388,
      "step": 2800
    },
    {
      "epoch": 0.47801309857956964,
      "grad_norm": 3.9796175956726074,
      "learning_rate": 1.3294915837564254e-05,
      "loss": 1.9321,
      "step": 2810
    },
    {
      "epoch": 0.47971421280939014,
      "grad_norm": 3.5634584426879883,
      "learning_rate": 1.3293349538420858e-05,
      "loss": 1.9956,
      "step": 2820
    },
    {
      "epoch": 0.4814153270392107,
      "grad_norm": 4.366597652435303,
      "learning_rate": 1.329177526021895e-05,
      "loss": 2.0251,
      "step": 2830
    },
    {
      "epoch": 0.4831164412690312,
      "grad_norm": 7.455503463745117,
      "learning_rate": 1.329019300487035e-05,
      "loss": 2.0371,
      "step": 2840
    },
    {
      "epoch": 0.48481755549885175,
      "grad_norm": 4.278938293457031,
      "learning_rate": 1.3288602774296575e-05,
      "loss": 2.0568,
      "step": 2850
    },
    {
      "epoch": 0.4865186697286723,
      "grad_norm": 4.015135765075684,
      "learning_rate": 1.3287004570428812e-05,
      "loss": 1.8454,
      "step": 2860
    },
    {
      "epoch": 0.4882197839584928,
      "grad_norm": 3.6956396102905273,
      "learning_rate": 1.3285398395207948e-05,
      "loss": 2.037,
      "step": 2870
    },
    {
      "epoch": 0.48992089818831336,
      "grad_norm": 4.163562297821045,
      "learning_rate": 1.3283784250584536e-05,
      "loss": 1.9196,
      "step": 2880
    },
    {
      "epoch": 0.49162201241813386,
      "grad_norm": 3.465672016143799,
      "learning_rate": 1.3282162138518814e-05,
      "loss": 2.0516,
      "step": 2890
    },
    {
      "epoch": 0.4933231266479544,
      "grad_norm": 4.972282409667969,
      "learning_rate": 1.3280532060980699e-05,
      "loss": 2.0446,
      "step": 2900
    },
    {
      "epoch": 0.49502424087777497,
      "grad_norm": 3.4876644611358643,
      "learning_rate": 1.3278894019949775e-05,
      "loss": 1.8955,
      "step": 2910
    },
    {
      "epoch": 0.49672535510759547,
      "grad_norm": 4.416055202484131,
      "learning_rate": 1.3277248017415297e-05,
      "loss": 1.9769,
      "step": 2920
    },
    {
      "epoch": 0.498426469337416,
      "grad_norm": 4.977348327636719,
      "learning_rate": 1.3275594055376196e-05,
      "loss": 1.9478,
      "step": 2930
    },
    {
      "epoch": 0.5001275835672365,
      "grad_norm": 3.413250207901001,
      "learning_rate": 1.327393213584106e-05,
      "loss": 1.9492,
      "step": 2940
    },
    {
      "epoch": 0.5018286977970571,
      "grad_norm": 3.2921433448791504,
      "learning_rate": 1.3272262260828147e-05,
      "loss": 1.9549,
      "step": 2950
    },
    {
      "epoch": 0.5035298120268776,
      "grad_norm": 4.159584999084473,
      "learning_rate": 1.3270584432365373e-05,
      "loss": 2.01,
      "step": 2960
    },
    {
      "epoch": 0.5052309262566982,
      "grad_norm": 3.793626546859741,
      "learning_rate": 1.3268898652490313e-05,
      "loss": 2.0069,
      "step": 2970
    },
    {
      "epoch": 0.5069320404865186,
      "grad_norm": 4.399741172790527,
      "learning_rate": 1.32672049232502e-05,
      "loss": 1.9689,
      "step": 2980
    },
    {
      "epoch": 0.5086331547163392,
      "grad_norm": 3.734663248062134,
      "learning_rate": 1.326550324670192e-05,
      "loss": 2.0126,
      "step": 2990
    },
    {
      "epoch": 0.5103342689461597,
      "grad_norm": 4.259927272796631,
      "learning_rate": 1.3263793624912009e-05,
      "loss": 1.9307,
      "step": 3000
    },
    {
      "epoch": 0.5120353831759803,
      "grad_norm": 4.000924587249756,
      "learning_rate": 1.3262076059956654e-05,
      "loss": 1.9663,
      "step": 3010
    },
    {
      "epoch": 0.5137364974058009,
      "grad_norm": 4.104271411895752,
      "learning_rate": 1.3260350553921688e-05,
      "loss": 1.9565,
      "step": 3020
    },
    {
      "epoch": 0.5154376116356213,
      "grad_norm": 4.1479878425598145,
      "learning_rate": 1.3258617108902584e-05,
      "loss": 2.0009,
      "step": 3030
    },
    {
      "epoch": 0.5171387258654419,
      "grad_norm": 3.7437946796417236,
      "learning_rate": 1.3256875727004462e-05,
      "loss": 2.0229,
      "step": 3040
    },
    {
      "epoch": 0.5188398400952624,
      "grad_norm": 4.00507926940918,
      "learning_rate": 1.3255126410342076e-05,
      "loss": 2.0789,
      "step": 3050
    },
    {
      "epoch": 0.520540954325083,
      "grad_norm": 4.296029567718506,
      "learning_rate": 1.3253369161039821e-05,
      "loss": 1.9893,
      "step": 3060
    },
    {
      "epoch": 0.5222420685549035,
      "grad_norm": 3.632911205291748,
      "learning_rate": 1.3251603981231718e-05,
      "loss": 2.0216,
      "step": 3070
    },
    {
      "epoch": 0.523943182784724,
      "grad_norm": 3.6041970252990723,
      "learning_rate": 1.3249830873061426e-05,
      "loss": 2.049,
      "step": 3080
    },
    {
      "epoch": 0.5256442970145445,
      "grad_norm": 3.852734327316284,
      "learning_rate": 1.3248049838682229e-05,
      "loss": 1.9443,
      "step": 3090
    },
    {
      "epoch": 0.5273454112443651,
      "grad_norm": 3.8711137771606445,
      "learning_rate": 1.3246260880257038e-05,
      "loss": 2.0248,
      "step": 3100
    },
    {
      "epoch": 0.5290465254741856,
      "grad_norm": 4.121218204498291,
      "learning_rate": 1.3244463999958385e-05,
      "loss": 2.0191,
      "step": 3110
    },
    {
      "epoch": 0.5307476397040061,
      "grad_norm": 4.025257587432861,
      "learning_rate": 1.3242659199968425e-05,
      "loss": 2.0551,
      "step": 3120
    },
    {
      "epoch": 0.5324487539338266,
      "grad_norm": 3.772130012512207,
      "learning_rate": 1.3240846482478928e-05,
      "loss": 2.0279,
      "step": 3130
    },
    {
      "epoch": 0.5341498681636472,
      "grad_norm": 3.7531166076660156,
      "learning_rate": 1.323902584969128e-05,
      "loss": 1.995,
      "step": 3140
    },
    {
      "epoch": 0.5358509823934677,
      "grad_norm": 4.111819267272949,
      "learning_rate": 1.3237197303816486e-05,
      "loss": 2.1142,
      "step": 3150
    },
    {
      "epoch": 0.5375520966232883,
      "grad_norm": 3.1626720428466797,
      "learning_rate": 1.3235360847075147e-05,
      "loss": 1.9886,
      "step": 3160
    },
    {
      "epoch": 0.5392532108531087,
      "grad_norm": 3.5329902172088623,
      "learning_rate": 1.3233516481697481e-05,
      "loss": 2.0167,
      "step": 3170
    },
    {
      "epoch": 0.5409543250829293,
      "grad_norm": 3.377159595489502,
      "learning_rate": 1.3231664209923313e-05,
      "loss": 2.1049,
      "step": 3180
    },
    {
      "epoch": 0.5426554393127498,
      "grad_norm": 3.3523881435394287,
      "learning_rate": 1.322980403400206e-05,
      "loss": 2.0344,
      "step": 3190
    },
    {
      "epoch": 0.5443565535425704,
      "grad_norm": 4.31252908706665,
      "learning_rate": 1.3227935956192744e-05,
      "loss": 1.9801,
      "step": 3200
    },
    {
      "epoch": 0.546057667772391,
      "grad_norm": 4.030631065368652,
      "learning_rate": 1.3226059978763984e-05,
      "loss": 1.948,
      "step": 3210
    },
    {
      "epoch": 0.5477587820022114,
      "grad_norm": 3.5859789848327637,
      "learning_rate": 1.3224176103993986e-05,
      "loss": 2.0254,
      "step": 3220
    },
    {
      "epoch": 0.549459896232032,
      "grad_norm": 4.439004421234131,
      "learning_rate": 1.3222284334170552e-05,
      "loss": 1.9773,
      "step": 3230
    },
    {
      "epoch": 0.5511610104618525,
      "grad_norm": 3.986593008041382,
      "learning_rate": 1.3220384671591074e-05,
      "loss": 2.0235,
      "step": 3240
    },
    {
      "epoch": 0.5528621246916731,
      "grad_norm": 4.354572296142578,
      "learning_rate": 1.3218477118562525e-05,
      "loss": 1.8921,
      "step": 3250
    },
    {
      "epoch": 0.5545632389214936,
      "grad_norm": 4.235722064971924,
      "learning_rate": 1.3216561677401458e-05,
      "loss": 2.0174,
      "step": 3260
    },
    {
      "epoch": 0.5562643531513141,
      "grad_norm": 3.979583263397217,
      "learning_rate": 1.3214638350434013e-05,
      "loss": 1.9176,
      "step": 3270
    },
    {
      "epoch": 0.5579654673811346,
      "grad_norm": 4.8564324378967285,
      "learning_rate": 1.3212707139995897e-05,
      "loss": 1.9592,
      "step": 3280
    },
    {
      "epoch": 0.5596665816109552,
      "grad_norm": 3.7206497192382812,
      "learning_rate": 1.3210768048432398e-05,
      "loss": 1.95,
      "step": 3290
    },
    {
      "epoch": 0.5613676958407757,
      "grad_norm": 4.199899196624756,
      "learning_rate": 1.3208821078098376e-05,
      "loss": 2.0345,
      "step": 3300
    },
    {
      "epoch": 0.5630688100705963,
      "grad_norm": 3.7845380306243896,
      "learning_rate": 1.3206866231358253e-05,
      "loss": 2.0071,
      "step": 3310
    },
    {
      "epoch": 0.5647699243004167,
      "grad_norm": 3.7112693786621094,
      "learning_rate": 1.3204903510586017e-05,
      "loss": 1.9788,
      "step": 3320
    },
    {
      "epoch": 0.5664710385302373,
      "grad_norm": 4.377164363861084,
      "learning_rate": 1.3202932918165227e-05,
      "loss": 2.0004,
      "step": 3330
    },
    {
      "epoch": 0.5681721527600578,
      "grad_norm": 4.244388103485107,
      "learning_rate": 1.320095445648899e-05,
      "loss": 1.9677,
      "step": 3340
    },
    {
      "epoch": 0.5698732669898784,
      "grad_norm": 4.318045139312744,
      "learning_rate": 1.3198968127959974e-05,
      "loss": 2.0189,
      "step": 3350
    },
    {
      "epoch": 0.571574381219699,
      "grad_norm": 4.372837066650391,
      "learning_rate": 1.3196973934990402e-05,
      "loss": 2.043,
      "step": 3360
    },
    {
      "epoch": 0.5732754954495194,
      "grad_norm": 4.486396312713623,
      "learning_rate": 1.3194971880002048e-05,
      "loss": 1.9951,
      "step": 3370
    },
    {
      "epoch": 0.57497660967934,
      "grad_norm": 4.997833251953125,
      "learning_rate": 1.3192961965426233e-05,
      "loss": 1.8461,
      "step": 3380
    },
    {
      "epoch": 0.5766777239091605,
      "grad_norm": 3.630509853363037,
      "learning_rate": 1.319094419370382e-05,
      "loss": 2.0038,
      "step": 3390
    },
    {
      "epoch": 0.578378838138981,
      "grad_norm": 3.9311442375183105,
      "learning_rate": 1.3188918567285216e-05,
      "loss": 1.9245,
      "step": 3400
    },
    {
      "epoch": 0.5800799523688016,
      "grad_norm": 3.9913883209228516,
      "learning_rate": 1.318688508863037e-05,
      "loss": 1.9845,
      "step": 3410
    },
    {
      "epoch": 0.581781066598622,
      "grad_norm": 3.962445020675659,
      "learning_rate": 1.318484376020876e-05,
      "loss": 1.9759,
      "step": 3420
    },
    {
      "epoch": 0.5834821808284426,
      "grad_norm": 3.8628203868865967,
      "learning_rate": 1.3182794584499402e-05,
      "loss": 2.0435,
      "step": 3430
    },
    {
      "epoch": 0.5851832950582632,
      "grad_norm": 3.9601385593414307,
      "learning_rate": 1.3180737563990842e-05,
      "loss": 1.8831,
      "step": 3440
    },
    {
      "epoch": 0.5868844092880837,
      "grad_norm": 3.4271185398101807,
      "learning_rate": 1.317867270118115e-05,
      "loss": 2.0022,
      "step": 3450
    },
    {
      "epoch": 0.5885855235179043,
      "grad_norm": 3.7709715366363525,
      "learning_rate": 1.3176599998577918e-05,
      "loss": 1.8817,
      "step": 3460
    },
    {
      "epoch": 0.5902866377477247,
      "grad_norm": 3.9962878227233887,
      "learning_rate": 1.3174519458698268e-05,
      "loss": 2.003,
      "step": 3470
    },
    {
      "epoch": 0.5919877519775453,
      "grad_norm": 4.124835014343262,
      "learning_rate": 1.3172431084068829e-05,
      "loss": 1.9846,
      "step": 3480
    },
    {
      "epoch": 0.5936888662073658,
      "grad_norm": 3.7828152179718018,
      "learning_rate": 1.317033487722575e-05,
      "loss": 2.0059,
      "step": 3490
    },
    {
      "epoch": 0.5953899804371864,
      "grad_norm": 4.043171405792236,
      "learning_rate": 1.316823084071469e-05,
      "loss": 1.9832,
      "step": 3500
    },
    {
      "epoch": 0.5970910946670069,
      "grad_norm": 5.648746013641357,
      "learning_rate": 1.316611897709082e-05,
      "loss": 2.0174,
      "step": 3510
    },
    {
      "epoch": 0.5987922088968274,
      "grad_norm": 3.3299238681793213,
      "learning_rate": 1.316399928891881e-05,
      "loss": 1.8811,
      "step": 3520
    },
    {
      "epoch": 0.6004933231266479,
      "grad_norm": 3.889491319656372,
      "learning_rate": 1.3161871778772836e-05,
      "loss": 1.9641,
      "step": 3530
    },
    {
      "epoch": 0.6021944373564685,
      "grad_norm": 4.209897041320801,
      "learning_rate": 1.3159736449236577e-05,
      "loss": 2.0733,
      "step": 3540
    },
    {
      "epoch": 0.603895551586289,
      "grad_norm": 3.638066291809082,
      "learning_rate": 1.3157593302903199e-05,
      "loss": 1.9662,
      "step": 3550
    },
    {
      "epoch": 0.6055966658161096,
      "grad_norm": 3.562626838684082,
      "learning_rate": 1.3155442342375365e-05,
      "loss": 1.9735,
      "step": 3560
    },
    {
      "epoch": 0.60729778004593,
      "grad_norm": 3.843376636505127,
      "learning_rate": 1.3153283570265233e-05,
      "loss": 1.9661,
      "step": 3570
    },
    {
      "epoch": 0.6089988942757506,
      "grad_norm": 3.4872562885284424,
      "learning_rate": 1.3151116989194438e-05,
      "loss": 2.0091,
      "step": 3580
    },
    {
      "epoch": 0.6107000085055712,
      "grad_norm": 4.058065414428711,
      "learning_rate": 1.3148942601794106e-05,
      "loss": 2.0173,
      "step": 3590
    },
    {
      "epoch": 0.6124011227353917,
      "grad_norm": 3.877347230911255,
      "learning_rate": 1.314676041070484e-05,
      "loss": 2.0794,
      "step": 3600
    },
    {
      "epoch": 0.6141022369652123,
      "grad_norm": 3.453751564025879,
      "learning_rate": 1.3144570418576719e-05,
      "loss": 1.9352,
      "step": 3610
    },
    {
      "epoch": 0.6158033511950327,
      "grad_norm": 3.791684627532959,
      "learning_rate": 1.3142372628069294e-05,
      "loss": 1.9737,
      "step": 3620
    },
    {
      "epoch": 0.6175044654248533,
      "grad_norm": 4.093663215637207,
      "learning_rate": 1.3140167041851593e-05,
      "loss": 1.9208,
      "step": 3630
    },
    {
      "epoch": 0.6192055796546738,
      "grad_norm": 3.8233466148376465,
      "learning_rate": 1.3137953662602104e-05,
      "loss": 1.9972,
      "step": 3640
    },
    {
      "epoch": 0.6209066938844944,
      "grad_norm": 3.34663987159729,
      "learning_rate": 1.3135732493008786e-05,
      "loss": 2.0431,
      "step": 3650
    },
    {
      "epoch": 0.6226078081143149,
      "grad_norm": 3.8648672103881836,
      "learning_rate": 1.3133503535769055e-05,
      "loss": 2.1321,
      "step": 3660
    },
    {
      "epoch": 0.6243089223441354,
      "grad_norm": 4.013655185699463,
      "learning_rate": 1.313126679358978e-05,
      "loss": 1.9735,
      "step": 3670
    },
    {
      "epoch": 0.6260100365739559,
      "grad_norm": 3.838092088699341,
      "learning_rate": 1.3129022269187292e-05,
      "loss": 1.9773,
      "step": 3680
    },
    {
      "epoch": 0.6277111508037765,
      "grad_norm": 3.74652099609375,
      "learning_rate": 1.3126769965287367e-05,
      "loss": 1.9456,
      "step": 3690
    },
    {
      "epoch": 0.629412265033597,
      "grad_norm": 4.200870990753174,
      "learning_rate": 1.3124509884625234e-05,
      "loss": 2.0082,
      "step": 3700
    },
    {
      "epoch": 0.6311133792634175,
      "grad_norm": 3.9270575046539307,
      "learning_rate": 1.312224202994556e-05,
      "loss": 1.9668,
      "step": 3710
    },
    {
      "epoch": 0.632814493493238,
      "grad_norm": 3.9572863578796387,
      "learning_rate": 1.3119966404002458e-05,
      "loss": 1.9819,
      "step": 3720
    },
    {
      "epoch": 0.6345156077230586,
      "grad_norm": 4.089076042175293,
      "learning_rate": 1.3117683009559475e-05,
      "loss": 2.1096,
      "step": 3730
    },
    {
      "epoch": 0.6362167219528791,
      "grad_norm": 3.985917568206787,
      "learning_rate": 1.3115391849389593e-05,
      "loss": 1.9507,
      "step": 3740
    },
    {
      "epoch": 0.6379178361826997,
      "grad_norm": 3.494417190551758,
      "learning_rate": 1.3113092926275229e-05,
      "loss": 2.0412,
      "step": 3750
    },
    {
      "epoch": 0.6396189504125201,
      "grad_norm": 3.491520404815674,
      "learning_rate": 1.3110786243008218e-05,
      "loss": 2.0301,
      "step": 3760
    },
    {
      "epoch": 0.6413200646423407,
      "grad_norm": 3.2409474849700928,
      "learning_rate": 1.310847180238983e-05,
      "loss": 1.8752,
      "step": 3770
    },
    {
      "epoch": 0.6430211788721613,
      "grad_norm": 3.796698808670044,
      "learning_rate": 1.3106149607230751e-05,
      "loss": 1.9809,
      "step": 3780
    },
    {
      "epoch": 0.6447222931019818,
      "grad_norm": 4.545765399932861,
      "learning_rate": 1.3103819660351079e-05,
      "loss": 1.8971,
      "step": 3790
    },
    {
      "epoch": 0.6464234073318024,
      "grad_norm": 3.602032423019409,
      "learning_rate": 1.310148196458033e-05,
      "loss": 1.9571,
      "step": 3800
    },
    {
      "epoch": 0.6481245215616228,
      "grad_norm": 3.979278564453125,
      "learning_rate": 1.3099136522757435e-05,
      "loss": 1.9982,
      "step": 3810
    },
    {
      "epoch": 0.6498256357914434,
      "grad_norm": 3.5593981742858887,
      "learning_rate": 1.3096783337730727e-05,
      "loss": 2.0194,
      "step": 3820
    },
    {
      "epoch": 0.6515267500212639,
      "grad_norm": 3.7880349159240723,
      "learning_rate": 1.309442241235794e-05,
      "loss": 1.9879,
      "step": 3830
    },
    {
      "epoch": 0.6532278642510845,
      "grad_norm": 3.7969162464141846,
      "learning_rate": 1.3092053749506217e-05,
      "loss": 2.0842,
      "step": 3840
    },
    {
      "epoch": 0.654928978480905,
      "grad_norm": 3.9793834686279297,
      "learning_rate": 1.3089677352052087e-05,
      "loss": 1.9374,
      "step": 3850
    },
    {
      "epoch": 0.6566300927107255,
      "grad_norm": 3.4517974853515625,
      "learning_rate": 1.3087293222881475e-05,
      "loss": 1.9882,
      "step": 3860
    },
    {
      "epoch": 0.658331206940546,
      "grad_norm": 3.781954288482666,
      "learning_rate": 1.3084901364889702e-05,
      "loss": 1.9982,
      "step": 3870
    },
    {
      "epoch": 0.6600323211703666,
      "grad_norm": 3.9006168842315674,
      "learning_rate": 1.3082501780981465e-05,
      "loss": 1.9795,
      "step": 3880
    },
    {
      "epoch": 0.6617334354001871,
      "grad_norm": 3.539034605026245,
      "learning_rate": 1.3080094474070855e-05,
      "loss": 1.9624,
      "step": 3890
    },
    {
      "epoch": 0.6634345496300077,
      "grad_norm": 3.632161855697632,
      "learning_rate": 1.3077679447081328e-05,
      "loss": 1.9449,
      "step": 3900
    },
    {
      "epoch": 0.6651356638598281,
      "grad_norm": 4.065733909606934,
      "learning_rate": 1.3075256702945727e-05,
      "loss": 2.0045,
      "step": 3910
    },
    {
      "epoch": 0.6668367780896487,
      "grad_norm": 3.7506015300750732,
      "learning_rate": 1.3072826244606263e-05,
      "loss": 2.0023,
      "step": 3920
    },
    {
      "epoch": 0.6685378923194693,
      "grad_norm": 3.198028087615967,
      "learning_rate": 1.3070388075014513e-05,
      "loss": 1.9508,
      "step": 3930
    },
    {
      "epoch": 0.6702390065492898,
      "grad_norm": 3.5299782752990723,
      "learning_rate": 1.3067942197131422e-05,
      "loss": 1.942,
      "step": 3940
    },
    {
      "epoch": 0.6719401207791104,
      "grad_norm": 4.142813682556152,
      "learning_rate": 1.306548861392729e-05,
      "loss": 1.9504,
      "step": 3950
    },
    {
      "epoch": 0.6736412350089308,
      "grad_norm": 3.373514175415039,
      "learning_rate": 1.3063027328381788e-05,
      "loss": 1.9089,
      "step": 3960
    },
    {
      "epoch": 0.6753423492387514,
      "grad_norm": 3.3843135833740234,
      "learning_rate": 1.3060558343483922e-05,
      "loss": 1.8664,
      "step": 3970
    },
    {
      "epoch": 0.6770434634685719,
      "grad_norm": 4.3059587478637695,
      "learning_rate": 1.3058081662232065e-05,
      "loss": 2.0297,
      "step": 3980
    },
    {
      "epoch": 0.6787445776983925,
      "grad_norm": 4.434045791625977,
      "learning_rate": 1.3055597287633925e-05,
      "loss": 2.0207,
      "step": 3990
    },
    {
      "epoch": 0.680445691928213,
      "grad_norm": 4.15225076675415,
      "learning_rate": 1.305310522270656e-05,
      "loss": 1.9678,
      "step": 4000
    },
    {
      "epoch": 0.6821468061580335,
      "grad_norm": 4.175527572631836,
      "learning_rate": 1.3050605470476363e-05,
      "loss": 2.0039,
      "step": 4010
    },
    {
      "epoch": 0.683847920387854,
      "grad_norm": 3.8997533321380615,
      "learning_rate": 1.3048098033979066e-05,
      "loss": 1.8911,
      "step": 4020
    },
    {
      "epoch": 0.6855490346176746,
      "grad_norm": 4.1734089851379395,
      "learning_rate": 1.3045582916259729e-05,
      "loss": 1.9374,
      "step": 4030
    },
    {
      "epoch": 0.6872501488474951,
      "grad_norm": 4.575035095214844,
      "learning_rate": 1.3043060120372744e-05,
      "loss": 1.9354,
      "step": 4040
    },
    {
      "epoch": 0.6889512630773157,
      "grad_norm": 3.9317026138305664,
      "learning_rate": 1.3040529649381823e-05,
      "loss": 1.9613,
      "step": 4050
    },
    {
      "epoch": 0.6906523773071361,
      "grad_norm": 4.242161750793457,
      "learning_rate": 1.3037991506360002e-05,
      "loss": 1.9097,
      "step": 4060
    },
    {
      "epoch": 0.6923534915369567,
      "grad_norm": 4.004007816314697,
      "learning_rate": 1.3035445694389638e-05,
      "loss": 1.9673,
      "step": 4070
    },
    {
      "epoch": 0.6940546057667772,
      "grad_norm": 4.778632164001465,
      "learning_rate": 1.303289221656239e-05,
      "loss": 1.9565,
      "step": 4080
    },
    {
      "epoch": 0.6957557199965978,
      "grad_norm": 3.2379519939422607,
      "learning_rate": 1.3030331075979241e-05,
      "loss": 1.9305,
      "step": 4090
    },
    {
      "epoch": 0.6974568342264184,
      "grad_norm": 4.127084255218506,
      "learning_rate": 1.3027762275750465e-05,
      "loss": 1.9676,
      "step": 4100
    },
    {
      "epoch": 0.6991579484562388,
      "grad_norm": 3.654214859008789,
      "learning_rate": 1.3025185818995646e-05,
      "loss": 1.9687,
      "step": 4110
    },
    {
      "epoch": 0.7008590626860594,
      "grad_norm": 4.186607837677002,
      "learning_rate": 1.3022601708843666e-05,
      "loss": 1.9978,
      "step": 4120
    },
    {
      "epoch": 0.7025601769158799,
      "grad_norm": 3.398437976837158,
      "learning_rate": 1.3020009948432703e-05,
      "loss": 2.0208,
      "step": 4130
    },
    {
      "epoch": 0.7042612911457005,
      "grad_norm": 4.260915279388428,
      "learning_rate": 1.3017410540910221e-05,
      "loss": 1.9923,
      "step": 4140
    },
    {
      "epoch": 0.705962405375521,
      "grad_norm": 3.935940742492676,
      "learning_rate": 1.301480348943297e-05,
      "loss": 2.0397,
      "step": 4150
    },
    {
      "epoch": 0.7076635196053415,
      "grad_norm": 4.820540904998779,
      "learning_rate": 1.3012188797166992e-05,
      "loss": 1.9785,
      "step": 4160
    },
    {
      "epoch": 0.709364633835162,
      "grad_norm": 3.5808472633361816,
      "learning_rate": 1.3009566467287596e-05,
      "loss": 1.8898,
      "step": 4170
    },
    {
      "epoch": 0.7110657480649826,
      "grad_norm": 3.319737672805786,
      "learning_rate": 1.3006936502979375e-05,
      "loss": 2.0429,
      "step": 4180
    },
    {
      "epoch": 0.7127668622948031,
      "grad_norm": 3.511857032775879,
      "learning_rate": 1.3004298907436188e-05,
      "loss": 1.9019,
      "step": 4190
    },
    {
      "epoch": 0.7144679765246237,
      "grad_norm": 3.2292962074279785,
      "learning_rate": 1.3001653683861167e-05,
      "loss": 1.9096,
      "step": 4200
    },
    {
      "epoch": 0.7161690907544441,
      "grad_norm": 3.7671141624450684,
      "learning_rate": 1.2999000835466701e-05,
      "loss": 1.9454,
      "step": 4210
    },
    {
      "epoch": 0.7178702049842647,
      "grad_norm": 4.096886157989502,
      "learning_rate": 1.2996340365474445e-05,
      "loss": 1.9327,
      "step": 4220
    },
    {
      "epoch": 0.7195713192140852,
      "grad_norm": 3.9096286296844482,
      "learning_rate": 1.2993672277115305e-05,
      "loss": 2.0614,
      "step": 4230
    },
    {
      "epoch": 0.7212724334439058,
      "grad_norm": 3.94098162651062,
      "learning_rate": 1.299099657362944e-05,
      "loss": 2.004,
      "step": 4240
    },
    {
      "epoch": 0.7229735476737262,
      "grad_norm": 3.444897174835205,
      "learning_rate": 1.298831325826626e-05,
      "loss": 1.9583,
      "step": 4250
    },
    {
      "epoch": 0.7246746619035468,
      "grad_norm": 3.791532516479492,
      "learning_rate": 1.2985622334284414e-05,
      "loss": 1.9095,
      "step": 4260
    },
    {
      "epoch": 0.7263757761333673,
      "grad_norm": 3.3483974933624268,
      "learning_rate": 1.2982923804951797e-05,
      "loss": 1.8663,
      "step": 4270
    },
    {
      "epoch": 0.7280768903631879,
      "grad_norm": 3.745339870452881,
      "learning_rate": 1.2980217673545534e-05,
      "loss": 1.9655,
      "step": 4280
    },
    {
      "epoch": 0.7297780045930085,
      "grad_norm": 3.8242061138153076,
      "learning_rate": 1.2977503943351981e-05,
      "loss": 1.9921,
      "step": 4290
    },
    {
      "epoch": 0.7314791188228289,
      "grad_norm": 3.353945732116699,
      "learning_rate": 1.2974782617666734e-05,
      "loss": 1.9331,
      "step": 4300
    },
    {
      "epoch": 0.7331802330526495,
      "grad_norm": 3.296675682067871,
      "learning_rate": 1.2972053699794598e-05,
      "loss": 2.0209,
      "step": 4310
    },
    {
      "epoch": 0.73488134728247,
      "grad_norm": 3.7973153591156006,
      "learning_rate": 1.2969317193049608e-05,
      "loss": 2.0261,
      "step": 4320
    },
    {
      "epoch": 0.7365824615122906,
      "grad_norm": 3.46197772026062,
      "learning_rate": 1.296657310075501e-05,
      "loss": 1.9906,
      "step": 4330
    },
    {
      "epoch": 0.7382835757421111,
      "grad_norm": 4.166983127593994,
      "learning_rate": 1.2963821426243264e-05,
      "loss": 1.9852,
      "step": 4340
    },
    {
      "epoch": 0.7399846899719316,
      "grad_norm": 3.487180709838867,
      "learning_rate": 1.296106217285604e-05,
      "loss": 1.944,
      "step": 4350
    },
    {
      "epoch": 0.7416858042017521,
      "grad_norm": 3.9465606212615967,
      "learning_rate": 1.2958295343944207e-05,
      "loss": 2.0371,
      "step": 4360
    },
    {
      "epoch": 0.7433869184315727,
      "grad_norm": 3.864844560623169,
      "learning_rate": 1.2955520942867837e-05,
      "loss": 1.979,
      "step": 4370
    },
    {
      "epoch": 0.7450880326613932,
      "grad_norm": 3.770357370376587,
      "learning_rate": 1.2952738972996197e-05,
      "loss": 1.9993,
      "step": 4380
    },
    {
      "epoch": 0.7467891468912138,
      "grad_norm": 3.781005859375,
      "learning_rate": 1.2949949437707747e-05,
      "loss": 1.9191,
      "step": 4390
    },
    {
      "epoch": 0.7484902611210342,
      "grad_norm": 3.8930251598358154,
      "learning_rate": 1.2947152340390131e-05,
      "loss": 1.9467,
      "step": 4400
    },
    {
      "epoch": 0.7501913753508548,
      "grad_norm": 3.6845479011535645,
      "learning_rate": 1.2944347684440183e-05,
      "loss": 1.9453,
      "step": 4410
    },
    {
      "epoch": 0.7518924895806753,
      "grad_norm": 3.2310779094696045,
      "learning_rate": 1.2941535473263907e-05,
      "loss": 1.9638,
      "step": 4420
    },
    {
      "epoch": 0.7535936038104959,
      "grad_norm": 3.4538753032684326,
      "learning_rate": 1.2938715710276491e-05,
      "loss": 1.9097,
      "step": 4430
    },
    {
      "epoch": 0.7552947180403164,
      "grad_norm": 4.4561872482299805,
      "learning_rate": 1.2935888398902285e-05,
      "loss": 1.9183,
      "step": 4440
    },
    {
      "epoch": 0.7569958322701369,
      "grad_norm": 3.7806403636932373,
      "learning_rate": 1.2933053542574818e-05,
      "loss": 2.0133,
      "step": 4450
    },
    {
      "epoch": 0.7586969464999574,
      "grad_norm": 3.7306880950927734,
      "learning_rate": 1.2930211144736772e-05,
      "loss": 1.913,
      "step": 4460
    },
    {
      "epoch": 0.760398060729778,
      "grad_norm": 3.4570231437683105,
      "learning_rate": 1.2927361208839987e-05,
      "loss": 2.0017,
      "step": 4470
    },
    {
      "epoch": 0.7620991749595986,
      "grad_norm": 4.0018534660339355,
      "learning_rate": 1.2924503738345465e-05,
      "loss": 1.8458,
      "step": 4480
    },
    {
      "epoch": 0.7638002891894191,
      "grad_norm": 3.6628971099853516,
      "learning_rate": 1.2921638736723351e-05,
      "loss": 1.9024,
      "step": 4490
    },
    {
      "epoch": 0.7655014034192396,
      "grad_norm": 4.748337745666504,
      "learning_rate": 1.291876620745294e-05,
      "loss": 1.9229,
      "step": 4500
    },
    {
      "epoch": 0.7672025176490601,
      "grad_norm": 3.4937639236450195,
      "learning_rate": 1.2915886154022668e-05,
      "loss": 1.9887,
      "step": 4510
    },
    {
      "epoch": 0.7689036318788807,
      "grad_norm": 3.3699965476989746,
      "learning_rate": 1.2912998579930104e-05,
      "loss": 1.9189,
      "step": 4520
    },
    {
      "epoch": 0.7706047461087012,
      "grad_norm": 4.4775872230529785,
      "learning_rate": 1.2910103488681956e-05,
      "loss": 1.9302,
      "step": 4530
    },
    {
      "epoch": 0.7723058603385218,
      "grad_norm": 3.719393491744995,
      "learning_rate": 1.2907200883794061e-05,
      "loss": 1.9872,
      "step": 4540
    },
    {
      "epoch": 0.7740069745683422,
      "grad_norm": 3.6321024894714355,
      "learning_rate": 1.2904290768791376e-05,
      "loss": 1.8501,
      "step": 4550
    },
    {
      "epoch": 0.7757080887981628,
      "grad_norm": 4.0052809715271,
      "learning_rate": 1.2901373147207981e-05,
      "loss": 1.9772,
      "step": 4560
    },
    {
      "epoch": 0.7774092030279833,
      "grad_norm": 3.715613842010498,
      "learning_rate": 1.289844802258707e-05,
      "loss": 1.8721,
      "step": 4570
    },
    {
      "epoch": 0.7791103172578039,
      "grad_norm": 3.5955617427825928,
      "learning_rate": 1.2895515398480956e-05,
      "loss": 1.911,
      "step": 4580
    },
    {
      "epoch": 0.7808114314876244,
      "grad_norm": 3.7583234310150146,
      "learning_rate": 1.2892575278451049e-05,
      "loss": 1.9743,
      "step": 4590
    },
    {
      "epoch": 0.7825125457174449,
      "grad_norm": 4.226505756378174,
      "learning_rate": 1.288962766606787e-05,
      "loss": 1.915,
      "step": 4600
    },
    {
      "epoch": 0.7842136599472654,
      "grad_norm": 3.244330406188965,
      "learning_rate": 1.2886672564911035e-05,
      "loss": 1.9117,
      "step": 4610
    },
    {
      "epoch": 0.785914774177086,
      "grad_norm": 3.8930232524871826,
      "learning_rate": 1.2883709978569256e-05,
      "loss": 1.947,
      "step": 4620
    },
    {
      "epoch": 0.7876158884069066,
      "grad_norm": 4.156840801239014,
      "learning_rate": 1.2880739910640333e-05,
      "loss": 2.0105,
      "step": 4630
    },
    {
      "epoch": 0.7893170026367271,
      "grad_norm": 4.2512688636779785,
      "learning_rate": 1.2877762364731159e-05,
      "loss": 1.8665,
      "step": 4640
    },
    {
      "epoch": 0.7910181168665475,
      "grad_norm": 3.931245803833008,
      "learning_rate": 1.2874777344457694e-05,
      "loss": 1.85,
      "step": 4650
    },
    {
      "epoch": 0.7927192310963681,
      "grad_norm": 3.273142099380493,
      "learning_rate": 1.287178485344499e-05,
      "loss": 1.8956,
      "step": 4660
    },
    {
      "epoch": 0.7944203453261887,
      "grad_norm": 3.791325092315674,
      "learning_rate": 1.2868784895327165e-05,
      "loss": 1.9721,
      "step": 4670
    },
    {
      "epoch": 0.7961214595560092,
      "grad_norm": 3.9774279594421387,
      "learning_rate": 1.2865777473747403e-05,
      "loss": 1.8861,
      "step": 4680
    },
    {
      "epoch": 0.7978225737858298,
      "grad_norm": 4.466744899749756,
      "learning_rate": 1.2862762592357954e-05,
      "loss": 1.9802,
      "step": 4690
    },
    {
      "epoch": 0.7995236880156502,
      "grad_norm": 4.0846710205078125,
      "learning_rate": 1.285974025482013e-05,
      "loss": 2.0306,
      "step": 4700
    },
    {
      "epoch": 0.8012248022454708,
      "grad_norm": 3.1373989582061768,
      "learning_rate": 1.2856710464804294e-05,
      "loss": 1.9023,
      "step": 4710
    },
    {
      "epoch": 0.8029259164752913,
      "grad_norm": 4.3124518394470215,
      "learning_rate": 1.2853673225989859e-05,
      "loss": 1.8912,
      "step": 4720
    },
    {
      "epoch": 0.8046270307051119,
      "grad_norm": 3.9370830059051514,
      "learning_rate": 1.2850628542065289e-05,
      "loss": 1.8607,
      "step": 4730
    },
    {
      "epoch": 0.8063281449349324,
      "grad_norm": 3.7955822944641113,
      "learning_rate": 1.2847576416728082e-05,
      "loss": 2.0242,
      "step": 4740
    },
    {
      "epoch": 0.8080292591647529,
      "grad_norm": 3.9595768451690674,
      "learning_rate": 1.284451685368478e-05,
      "loss": 2.0156,
      "step": 4750
    },
    {
      "epoch": 0.8097303733945734,
      "grad_norm": 3.8596062660217285,
      "learning_rate": 1.2841449856650952e-05,
      "loss": 1.9646,
      "step": 4760
    },
    {
      "epoch": 0.811431487624394,
      "grad_norm": 4.593785285949707,
      "learning_rate": 1.2838375429351201e-05,
      "loss": 1.956,
      "step": 4770
    },
    {
      "epoch": 0.8131326018542145,
      "grad_norm": 3.4249627590179443,
      "learning_rate": 1.2835293575519146e-05,
      "loss": 1.9888,
      "step": 4780
    },
    {
      "epoch": 0.8148337160840351,
      "grad_norm": 3.3835091590881348,
      "learning_rate": 1.283220429889743e-05,
      "loss": 2.0539,
      "step": 4790
    },
    {
      "epoch": 0.8165348303138555,
      "grad_norm": 3.9617295265197754,
      "learning_rate": 1.2829107603237708e-05,
      "loss": 1.9249,
      "step": 4800
    },
    {
      "epoch": 0.8182359445436761,
      "grad_norm": 3.755145311355591,
      "learning_rate": 1.2826003492300647e-05,
      "loss": 1.9182,
      "step": 4810
    },
    {
      "epoch": 0.8199370587734967,
      "grad_norm": 3.9545507431030273,
      "learning_rate": 1.2822891969855917e-05,
      "loss": 1.9943,
      "step": 4820
    },
    {
      "epoch": 0.8216381730033172,
      "grad_norm": 3.2209560871124268,
      "learning_rate": 1.281977303968219e-05,
      "loss": 1.9436,
      "step": 4830
    },
    {
      "epoch": 0.8233392872331377,
      "grad_norm": 5.035518169403076,
      "learning_rate": 1.2816646705567135e-05,
      "loss": 1.8816,
      "step": 4840
    },
    {
      "epoch": 0.8250404014629582,
      "grad_norm": 4.079342365264893,
      "learning_rate": 1.2813512971307409e-05,
      "loss": 1.9416,
      "step": 4850
    },
    {
      "epoch": 0.8267415156927788,
      "grad_norm": 4.157555103302002,
      "learning_rate": 1.2810371840708656e-05,
      "loss": 1.9485,
      "step": 4860
    },
    {
      "epoch": 0.8284426299225993,
      "grad_norm": 3.232848644256592,
      "learning_rate": 1.2807223317585508e-05,
      "loss": 1.8885,
      "step": 4870
    },
    {
      "epoch": 0.8301437441524199,
      "grad_norm": 4.313023090362549,
      "learning_rate": 1.2804067405761567e-05,
      "loss": 1.9958,
      "step": 4880
    },
    {
      "epoch": 0.8318448583822403,
      "grad_norm": 4.823918342590332,
      "learning_rate": 1.2800904109069414e-05,
      "loss": 1.9411,
      "step": 4890
    },
    {
      "epoch": 0.8335459726120609,
      "grad_norm": 3.9723315238952637,
      "learning_rate": 1.2797733431350596e-05,
      "loss": 2.0031,
      "step": 4900
    },
    {
      "epoch": 0.8352470868418814,
      "grad_norm": 4.099934101104736,
      "learning_rate": 1.2794555376455623e-05,
      "loss": 1.888,
      "step": 4910
    },
    {
      "epoch": 0.836948201071702,
      "grad_norm": 3.663517475128174,
      "learning_rate": 1.2791369948243961e-05,
      "loss": 1.909,
      "step": 4920
    },
    {
      "epoch": 0.8386493153015225,
      "grad_norm": 3.9390037059783936,
      "learning_rate": 1.278817715058404e-05,
      "loss": 1.9576,
      "step": 4930
    },
    {
      "epoch": 0.840350429531343,
      "grad_norm": 4.8486127853393555,
      "learning_rate": 1.2784976987353228e-05,
      "loss": 1.9344,
      "step": 4940
    },
    {
      "epoch": 0.8420515437611635,
      "grad_norm": 4.624791145324707,
      "learning_rate": 1.2781769462437846e-05,
      "loss": 1.9987,
      "step": 4950
    },
    {
      "epoch": 0.8437526579909841,
      "grad_norm": 3.784917116165161,
      "learning_rate": 1.277855457973315e-05,
      "loss": 1.9721,
      "step": 4960
    },
    {
      "epoch": 0.8454537722208046,
      "grad_norm": 3.5814690589904785,
      "learning_rate": 1.2775332343143332e-05,
      "loss": 1.9869,
      "step": 4970
    },
    {
      "epoch": 0.8471548864506252,
      "grad_norm": 4.098245143890381,
      "learning_rate": 1.277210275658152e-05,
      "loss": 1.9373,
      "step": 4980
    },
    {
      "epoch": 0.8488560006804456,
      "grad_norm": 3.2476980686187744,
      "learning_rate": 1.2768865823969763e-05,
      "loss": 2.0476,
      "step": 4990
    },
    {
      "epoch": 0.8505571149102662,
      "grad_norm": 4.2028374671936035,
      "learning_rate": 1.276562154923903e-05,
      "loss": 1.9071,
      "step": 5000
    },
    {
      "epoch": 0.8522582291400868,
      "grad_norm": 2.8667080402374268,
      "learning_rate": 1.2762369936329212e-05,
      "loss": 1.8819,
      "step": 5010
    },
    {
      "epoch": 0.8539593433699073,
      "grad_norm": 4.183213710784912,
      "learning_rate": 1.2759110989189105e-05,
      "loss": 1.9156,
      "step": 5020
    },
    {
      "epoch": 0.8556604575997279,
      "grad_norm": 3.4557831287384033,
      "learning_rate": 1.2755844711776415e-05,
      "loss": 1.9098,
      "step": 5030
    },
    {
      "epoch": 0.8573615718295483,
      "grad_norm": 3.910086154937744,
      "learning_rate": 1.2752571108057754e-05,
      "loss": 1.9913,
      "step": 5040
    },
    {
      "epoch": 0.8590626860593689,
      "grad_norm": 4.714244842529297,
      "learning_rate": 1.2749290182008621e-05,
      "loss": 2.0118,
      "step": 5050
    },
    {
      "epoch": 0.8607638002891894,
      "grad_norm": 2.8815157413482666,
      "learning_rate": 1.274600193761342e-05,
      "loss": 1.996,
      "step": 5060
    },
    {
      "epoch": 0.86246491451901,
      "grad_norm": 4.021540641784668,
      "learning_rate": 1.2742706378865429e-05,
      "loss": 1.9917,
      "step": 5070
    },
    {
      "epoch": 0.8641660287488305,
      "grad_norm": 4.05567741394043,
      "learning_rate": 1.273940350976682e-05,
      "loss": 1.9756,
      "step": 5080
    },
    {
      "epoch": 0.865867142978651,
      "grad_norm": 3.598092555999756,
      "learning_rate": 1.2736093334328635e-05,
      "loss": 1.9183,
      "step": 5090
    },
    {
      "epoch": 0.8675682572084715,
      "grad_norm": 4.150853157043457,
      "learning_rate": 1.2732775856570795e-05,
      "loss": 1.9923,
      "step": 5100
    },
    {
      "epoch": 0.8692693714382921,
      "grad_norm": 3.2984402179718018,
      "learning_rate": 1.2729451080522086e-05,
      "loss": 1.9023,
      "step": 5110
    },
    {
      "epoch": 0.8709704856681126,
      "grad_norm": 3.5412111282348633,
      "learning_rate": 1.2726119010220156e-05,
      "loss": 1.9325,
      "step": 5120
    },
    {
      "epoch": 0.8726715998979332,
      "grad_norm": 3.7908759117126465,
      "learning_rate": 1.2722779649711515e-05,
      "loss": 1.965,
      "step": 5130
    },
    {
      "epoch": 0.8743727141277536,
      "grad_norm": 4.017210483551025,
      "learning_rate": 1.271943300305152e-05,
      "loss": 1.9754,
      "step": 5140
    },
    {
      "epoch": 0.8760738283575742,
      "grad_norm": 3.5406274795532227,
      "learning_rate": 1.2716079074304384e-05,
      "loss": 1.9246,
      "step": 5150
    },
    {
      "epoch": 0.8777749425873947,
      "grad_norm": 3.591987133026123,
      "learning_rate": 1.2712717867543158e-05,
      "loss": 1.9146,
      "step": 5160
    },
    {
      "epoch": 0.8794760568172153,
      "grad_norm": 3.8221333026885986,
      "learning_rate": 1.2709349386849731e-05,
      "loss": 1.988,
      "step": 5170
    },
    {
      "epoch": 0.8811771710470359,
      "grad_norm": 3.9325971603393555,
      "learning_rate": 1.2705973636314834e-05,
      "loss": 1.8172,
      "step": 5180
    },
    {
      "epoch": 0.8828782852768563,
      "grad_norm": 4.338099479675293,
      "learning_rate": 1.2702590620038014e-05,
      "loss": 1.9972,
      "step": 5190
    },
    {
      "epoch": 0.8845793995066769,
      "grad_norm": 3.458946704864502,
      "learning_rate": 1.269920034212765e-05,
      "loss": 1.9436,
      "step": 5200
    },
    {
      "epoch": 0.8862805137364974,
      "grad_norm": 3.817826986312866,
      "learning_rate": 1.269580280670094e-05,
      "loss": 1.9126,
      "step": 5210
    },
    {
      "epoch": 0.887981627966318,
      "grad_norm": 3.9971837997436523,
      "learning_rate": 1.269239801788389e-05,
      "loss": 1.904,
      "step": 5220
    },
    {
      "epoch": 0.8896827421961385,
      "grad_norm": 3.336728096008301,
      "learning_rate": 1.2688985979811319e-05,
      "loss": 1.9538,
      "step": 5230
    },
    {
      "epoch": 0.891383856425959,
      "grad_norm": 3.4137091636657715,
      "learning_rate": 1.268556669662685e-05,
      "loss": 1.9659,
      "step": 5240
    },
    {
      "epoch": 0.8930849706557795,
      "grad_norm": 3.567591905593872,
      "learning_rate": 1.26821401724829e-05,
      "loss": 2.0012,
      "step": 5250
    },
    {
      "epoch": 0.8947860848856001,
      "grad_norm": 3.2225394248962402,
      "learning_rate": 1.2678706411540686e-05,
      "loss": 1.882,
      "step": 5260
    },
    {
      "epoch": 0.8964871991154206,
      "grad_norm": 3.936091899871826,
      "learning_rate": 1.2675265417970207e-05,
      "loss": 1.8807,
      "step": 5270
    },
    {
      "epoch": 0.8981883133452412,
      "grad_norm": 3.4630918502807617,
      "learning_rate": 1.2672162343313195e-05,
      "loss": 2.0266,
      "step": 5280
    },
    {
      "epoch": 0.8998894275750616,
      "grad_norm": 3.2364659309387207,
      "learning_rate": 1.2668707619268824e-05,
      "loss": 1.9086,
      "step": 5290
    },
    {
      "epoch": 0.9015905418048822,
      "grad_norm": 3.93023419380188,
      "learning_rate": 1.2665245674738846e-05,
      "loss": 1.905,
      "step": 5300
    },
    {
      "epoch": 0.9032916560347027,
      "grad_norm": 3.954702138900757,
      "learning_rate": 1.2661776513927489e-05,
      "loss": 1.9401,
      "step": 5310
    },
    {
      "epoch": 0.9049927702645233,
      "grad_norm": 4.499449729919434,
      "learning_rate": 1.2658300141047747e-05,
      "loss": 1.98,
      "step": 5320
    },
    {
      "epoch": 0.9066938844943438,
      "grad_norm": 4.818963527679443,
      "learning_rate": 1.2654816560321371e-05,
      "loss": 2.0562,
      "step": 5330
    },
    {
      "epoch": 0.9083949987241643,
      "grad_norm": 5.28843355178833,
      "learning_rate": 1.265132577597886e-05,
      "loss": 2.0204,
      "step": 5340
    },
    {
      "epoch": 0.9100961129539848,
      "grad_norm": 3.337747097015381,
      "learning_rate": 1.2647827792259472e-05,
      "loss": 1.8886,
      "step": 5350
    },
    {
      "epoch": 0.9117972271838054,
      "grad_norm": 3.8496439456939697,
      "learning_rate": 1.2644322613411198e-05,
      "loss": 1.8986,
      "step": 5360
    },
    {
      "epoch": 0.913498341413626,
      "grad_norm": 4.1146559715271,
      "learning_rate": 1.2640810243690772e-05,
      "loss": 1.9446,
      "step": 5370
    },
    {
      "epoch": 0.9151994556434464,
      "grad_norm": 3.5451676845550537,
      "learning_rate": 1.2637290687363658e-05,
      "loss": 1.8395,
      "step": 5380
    },
    {
      "epoch": 0.916900569873267,
      "grad_norm": 3.9882071018218994,
      "learning_rate": 1.2633763948704051e-05,
      "loss": 1.9069,
      "step": 5390
    },
    {
      "epoch": 0.9186016841030875,
      "grad_norm": 4.388647556304932,
      "learning_rate": 1.2630230031994866e-05,
      "loss": 1.8664,
      "step": 5400
    },
    {
      "epoch": 0.9203027983329081,
      "grad_norm": 3.316866636276245,
      "learning_rate": 1.262668894152773e-05,
      "loss": 1.8976,
      "step": 5410
    },
    {
      "epoch": 0.9220039125627286,
      "grad_norm": 3.828775405883789,
      "learning_rate": 1.2623140681602997e-05,
      "loss": 2.0227,
      "step": 5420
    },
    {
      "epoch": 0.9237050267925491,
      "grad_norm": 3.898134231567383,
      "learning_rate": 1.2619585256529709e-05,
      "loss": 1.9358,
      "step": 5430
    },
    {
      "epoch": 0.9254061410223696,
      "grad_norm": 3.237833023071289,
      "learning_rate": 1.2616022670625625e-05,
      "loss": 1.9884,
      "step": 5440
    },
    {
      "epoch": 0.9271072552521902,
      "grad_norm": 4.014885425567627,
      "learning_rate": 1.2612452928217191e-05,
      "loss": 1.9748,
      "step": 5450
    },
    {
      "epoch": 0.9288083694820107,
      "grad_norm": 3.4019582271575928,
      "learning_rate": 1.2608876033639544e-05,
      "loss": 1.8849,
      "step": 5460
    },
    {
      "epoch": 0.9305094837118313,
      "grad_norm": 4.390929222106934,
      "learning_rate": 1.2605291991236512e-05,
      "loss": 1.8787,
      "step": 5470
    },
    {
      "epoch": 0.9322105979416517,
      "grad_norm": 3.5821480751037598,
      "learning_rate": 1.2601700805360602e-05,
      "loss": 1.9351,
      "step": 5480
    },
    {
      "epoch": 0.9339117121714723,
      "grad_norm": 4.156138896942139,
      "learning_rate": 1.259810248037299e-05,
      "loss": 2.0119,
      "step": 5490
    },
    {
      "epoch": 0.9356128264012928,
      "grad_norm": 4.29572868347168,
      "learning_rate": 1.259449702064353e-05,
      "loss": 2.0583,
      "step": 5500
    },
    {
      "epoch": 0.9373139406311134,
      "grad_norm": 3.802596092224121,
      "learning_rate": 1.2590884430550738e-05,
      "loss": 1.837,
      "step": 5510
    },
    {
      "epoch": 0.939015054860934,
      "grad_norm": 3.691622734069824,
      "learning_rate": 1.2587264714481787e-05,
      "loss": 1.9308,
      "step": 5520
    },
    {
      "epoch": 0.9407161690907544,
      "grad_norm": 3.693000078201294,
      "learning_rate": 1.2583637876832502e-05,
      "loss": 1.8829,
      "step": 5530
    },
    {
      "epoch": 0.942417283320575,
      "grad_norm": 4.015108108520508,
      "learning_rate": 1.2580003922007365e-05,
      "loss": 1.9215,
      "step": 5540
    },
    {
      "epoch": 0.9441183975503955,
      "grad_norm": 3.6444931030273438,
      "learning_rate": 1.2576362854419494e-05,
      "loss": 1.8794,
      "step": 5550
    },
    {
      "epoch": 0.9458195117802161,
      "grad_norm": 4.046820640563965,
      "learning_rate": 1.2572714678490646e-05,
      "loss": 1.9707,
      "step": 5560
    },
    {
      "epoch": 0.9475206260100366,
      "grad_norm": 4.2986297607421875,
      "learning_rate": 1.256905939865121e-05,
      "loss": 1.8715,
      "step": 5570
    },
    {
      "epoch": 0.9492217402398571,
      "grad_norm": 3.7403786182403564,
      "learning_rate": 1.2565397019340203e-05,
      "loss": 1.9072,
      "step": 5580
    },
    {
      "epoch": 0.9509228544696776,
      "grad_norm": 3.4848575592041016,
      "learning_rate": 1.256172754500527e-05,
      "loss": 1.9457,
      "step": 5590
    },
    {
      "epoch": 0.9526239686994982,
      "grad_norm": 4.1095991134643555,
      "learning_rate": 1.255805098010266e-05,
      "loss": 1.8761,
      "step": 5600
    },
    {
      "epoch": 0.9543250829293187,
      "grad_norm": 4.046541690826416,
      "learning_rate": 1.255436732909724e-05,
      "loss": 1.9785,
      "step": 5610
    },
    {
      "epoch": 0.9560261971591393,
      "grad_norm": 4.484745979309082,
      "learning_rate": 1.2550676596462483e-05,
      "loss": 1.9656,
      "step": 5620
    },
    {
      "epoch": 0.9577273113889597,
      "grad_norm": 4.468336582183838,
      "learning_rate": 1.2546978786680459e-05,
      "loss": 1.8788,
      "step": 5630
    },
    {
      "epoch": 0.9594284256187803,
      "grad_norm": 3.724666118621826,
      "learning_rate": 1.2543273904241836e-05,
      "loss": 1.882,
      "step": 5640
    },
    {
      "epoch": 0.9611295398486008,
      "grad_norm": 4.388457775115967,
      "learning_rate": 1.2539561953645868e-05,
      "loss": 1.9252,
      "step": 5650
    },
    {
      "epoch": 0.9628306540784214,
      "grad_norm": 4.207907199859619,
      "learning_rate": 1.2535842939400395e-05,
      "loss": 1.9167,
      "step": 5660
    },
    {
      "epoch": 0.9645317683082419,
      "grad_norm": 3.7429306507110596,
      "learning_rate": 1.2532116866021834e-05,
      "loss": 1.8324,
      "step": 5670
    },
    {
      "epoch": 0.9662328825380624,
      "grad_norm": 4.0732622146606445,
      "learning_rate": 1.2528383738035172e-05,
      "loss": 1.9824,
      "step": 5680
    },
    {
      "epoch": 0.9679339967678829,
      "grad_norm": 3.736802339553833,
      "learning_rate": 1.252464355997397e-05,
      "loss": 1.8791,
      "step": 5690
    },
    {
      "epoch": 0.9696351109977035,
      "grad_norm": 3.765545606613159,
      "learning_rate": 1.2520896336380344e-05,
      "loss": 1.9353,
      "step": 5700
    },
    {
      "epoch": 0.971336225227524,
      "grad_norm": 3.725818634033203,
      "learning_rate": 1.2517142071804969e-05,
      "loss": 1.9819,
      "step": 5710
    },
    {
      "epoch": 0.9730373394573446,
      "grad_norm": 3.6932244300842285,
      "learning_rate": 1.2513380770807073e-05,
      "loss": 1.9578,
      "step": 5720
    },
    {
      "epoch": 0.974738453687165,
      "grad_norm": 5.006459712982178,
      "learning_rate": 1.2509612437954425e-05,
      "loss": 1.8818,
      "step": 5730
    },
    {
      "epoch": 0.9764395679169856,
      "grad_norm": 3.6801838874816895,
      "learning_rate": 1.2505837077823335e-05,
      "loss": 1.9557,
      "step": 5740
    },
    {
      "epoch": 0.9781406821468062,
      "grad_norm": 3.696099042892456,
      "learning_rate": 1.2502054694998645e-05,
      "loss": 1.9777,
      "step": 5750
    },
    {
      "epoch": 0.9798417963766267,
      "grad_norm": 4.071863651275635,
      "learning_rate": 1.2498265294073733e-05,
      "loss": 1.9201,
      "step": 5760
    },
    {
      "epoch": 0.9815429106064473,
      "grad_norm": 3.497555732727051,
      "learning_rate": 1.249446887965049e-05,
      "loss": 1.8312,
      "step": 5770
    },
    {
      "epoch": 0.9832440248362677,
      "grad_norm": 4.3368916511535645,
      "learning_rate": 1.2490665456339332e-05,
      "loss": 1.9367,
      "step": 5780
    },
    {
      "epoch": 0.9849451390660883,
      "grad_norm": 3.4796338081359863,
      "learning_rate": 1.2486855028759182e-05,
      "loss": 1.956,
      "step": 5790
    },
    {
      "epoch": 0.9866462532959088,
      "grad_norm": 4.069372177124023,
      "learning_rate": 1.248303760153747e-05,
      "loss": 2.043,
      "step": 5800
    },
    {
      "epoch": 0.9883473675257294,
      "grad_norm": 4.666215419769287,
      "learning_rate": 1.2479213179310127e-05,
      "loss": 1.9051,
      "step": 5810
    },
    {
      "epoch": 0.9900484817555499,
      "grad_norm": 3.515803575515747,
      "learning_rate": 1.247538176672158e-05,
      "loss": 1.9167,
      "step": 5820
    },
    {
      "epoch": 0.9917495959853704,
      "grad_norm": 3.4563639163970947,
      "learning_rate": 1.2471543368424744e-05,
      "loss": 1.8534,
      "step": 5830
    },
    {
      "epoch": 0.9934507102151909,
      "grad_norm": 3.588573932647705,
      "learning_rate": 1.2467697989081017e-05,
      "loss": 1.9681,
      "step": 5840
    },
    {
      "epoch": 0.9951518244450115,
      "grad_norm": 4.674989223480225,
      "learning_rate": 1.2463845633360276e-05,
      "loss": 1.9613,
      "step": 5850
    },
    {
      "epoch": 0.996852938674832,
      "grad_norm": 3.8413889408111572,
      "learning_rate": 1.2459986305940868e-05,
      "loss": 1.9165,
      "step": 5860
    },
    {
      "epoch": 0.9985540529046526,
      "grad_norm": 3.9926917552948,
      "learning_rate": 1.2456120011509611e-05,
      "loss": 1.8678,
      "step": 5870
    },
    {
      "epoch": 1.000255167134473,
      "grad_norm": 3.9802398681640625,
      "learning_rate": 1.2452246754761779e-05,
      "loss": 1.805,
      "step": 5880
    },
    {
      "epoch": 1.0019562813642937,
      "grad_norm": 3.552992105484009,
      "learning_rate": 1.2448366540401103e-05,
      "loss": 1.9088,
      "step": 5890
    },
    {
      "epoch": 1.0036573955941142,
      "grad_norm": 3.8864123821258545,
      "learning_rate": 1.2444479373139763e-05,
      "loss": 1.8298,
      "step": 5900
    },
    {
      "epoch": 1.0053585098239346,
      "grad_norm": 3.5456321239471436,
      "learning_rate": 1.2440585257698385e-05,
      "loss": 2.0187,
      "step": 5910
    },
    {
      "epoch": 1.0070596240537553,
      "grad_norm": 3.786015748977661,
      "learning_rate": 1.2436684198806031e-05,
      "loss": 1.8857,
      "step": 5920
    },
    {
      "epoch": 1.0087607382835757,
      "grad_norm": 3.7994706630706787,
      "learning_rate": 1.2432776201200195e-05,
      "loss": 1.9403,
      "step": 5930
    },
    {
      "epoch": 1.0104618525133964,
      "grad_norm": 4.07720947265625,
      "learning_rate": 1.2428861269626798e-05,
      "loss": 1.8519,
      "step": 5940
    },
    {
      "epoch": 1.0121629667432168,
      "grad_norm": 4.268309116363525,
      "learning_rate": 1.242493940884018e-05,
      "loss": 1.9267,
      "step": 5950
    },
    {
      "epoch": 1.0138640809730373,
      "grad_norm": 3.8725132942199707,
      "learning_rate": 1.2421010623603099e-05,
      "loss": 1.9198,
      "step": 5960
    },
    {
      "epoch": 1.015565195202858,
      "grad_norm": 4.3192901611328125,
      "learning_rate": 1.241707491868672e-05,
      "loss": 1.9027,
      "step": 5970
    },
    {
      "epoch": 1.0172663094326784,
      "grad_norm": 4.3986430168151855,
      "learning_rate": 1.2413132298870612e-05,
      "loss": 1.82,
      "step": 5980
    },
    {
      "epoch": 1.018967423662499,
      "grad_norm": 2.96518874168396,
      "learning_rate": 1.2409182768942742e-05,
      "loss": 1.8239,
      "step": 5990
    },
    {
      "epoch": 1.0206685378923195,
      "grad_norm": 4.35980224609375,
      "learning_rate": 1.2405226333699469e-05,
      "loss": 1.8164,
      "step": 6000
    },
    {
      "epoch": 1.02236965212214,
      "grad_norm": 3.609809637069702,
      "learning_rate": 1.2401262997945535e-05,
      "loss": 1.9054,
      "step": 6010
    },
    {
      "epoch": 1.0240707663519606,
      "grad_norm": 4.375704288482666,
      "learning_rate": 1.2397292766494063e-05,
      "loss": 1.7685,
      "step": 6020
    },
    {
      "epoch": 1.025771880581781,
      "grad_norm": 3.8709499835968018,
      "learning_rate": 1.2393315644166555e-05,
      "loss": 1.9408,
      "step": 6030
    },
    {
      "epoch": 1.0274729948116017,
      "grad_norm": 4.003682613372803,
      "learning_rate": 1.2389331635792878e-05,
      "loss": 1.8039,
      "step": 6040
    },
    {
      "epoch": 1.0291741090414221,
      "grad_norm": 3.9275400638580322,
      "learning_rate": 1.2385340746211259e-05,
      "loss": 1.8981,
      "step": 6050
    },
    {
      "epoch": 1.0308752232712426,
      "grad_norm": 3.491661787033081,
      "learning_rate": 1.2381342980268283e-05,
      "loss": 1.8998,
      "step": 6060
    },
    {
      "epoch": 1.0325763375010633,
      "grad_norm": 4.1040568351745605,
      "learning_rate": 1.237733834281889e-05,
      "loss": 1.8652,
      "step": 6070
    },
    {
      "epoch": 1.0342774517308837,
      "grad_norm": 4.30794620513916,
      "learning_rate": 1.237332683872636e-05,
      "loss": 1.8184,
      "step": 6080
    },
    {
      "epoch": 1.0359785659607044,
      "grad_norm": 4.098321437835693,
      "learning_rate": 1.236930847286231e-05,
      "loss": 1.783,
      "step": 6090
    },
    {
      "epoch": 1.0376796801905248,
      "grad_norm": 4.051021575927734,
      "learning_rate": 1.2365283250106698e-05,
      "loss": 1.832,
      "step": 6100
    },
    {
      "epoch": 1.0393807944203453,
      "grad_norm": 3.617978572845459,
      "learning_rate": 1.2361251175347803e-05,
      "loss": 1.84,
      "step": 6110
    },
    {
      "epoch": 1.041081908650166,
      "grad_norm": 3.803922176361084,
      "learning_rate": 1.2357212253482226e-05,
      "loss": 1.8226,
      "step": 6120
    },
    {
      "epoch": 1.0427830228799864,
      "grad_norm": 3.8119354248046875,
      "learning_rate": 1.2353166489414886e-05,
      "loss": 1.8064,
      "step": 6130
    },
    {
      "epoch": 1.0444841371098068,
      "grad_norm": 4.282057285308838,
      "learning_rate": 1.2349113888059004e-05,
      "loss": 1.85,
      "step": 6140
    },
    {
      "epoch": 1.0461852513396275,
      "grad_norm": 4.06541633605957,
      "learning_rate": 1.2345054454336117e-05,
      "loss": 1.8458,
      "step": 6150
    },
    {
      "epoch": 1.047886365569448,
      "grad_norm": 4.3509392738342285,
      "learning_rate": 1.2340988193176048e-05,
      "loss": 1.8692,
      "step": 6160
    },
    {
      "epoch": 1.0495874797992686,
      "grad_norm": 3.179201364517212,
      "learning_rate": 1.2336915109516911e-05,
      "loss": 1.8236,
      "step": 6170
    },
    {
      "epoch": 1.051288594029089,
      "grad_norm": 3.8430192470550537,
      "learning_rate": 1.2332835208305116e-05,
      "loss": 1.8379,
      "step": 6180
    },
    {
      "epoch": 1.0529897082589095,
      "grad_norm": 3.632685661315918,
      "learning_rate": 1.2328748494495342e-05,
      "loss": 1.8193,
      "step": 6190
    },
    {
      "epoch": 1.0546908224887301,
      "grad_norm": 4.009765625,
      "learning_rate": 1.2324654973050547e-05,
      "loss": 1.7988,
      "step": 6200
    },
    {
      "epoch": 1.0563919367185506,
      "grad_norm": 3.634101390838623,
      "learning_rate": 1.2320554648941952e-05,
      "loss": 1.8799,
      "step": 6210
    },
    {
      "epoch": 1.0580930509483712,
      "grad_norm": 4.166500568389893,
      "learning_rate": 1.2316447527149044e-05,
      "loss": 1.9124,
      "step": 6220
    },
    {
      "epoch": 1.0597941651781917,
      "grad_norm": 3.7749826908111572,
      "learning_rate": 1.231233361265956e-05,
      "loss": 1.7627,
      "step": 6230
    },
    {
      "epoch": 1.0614952794080121,
      "grad_norm": 3.855471611022949,
      "learning_rate": 1.230821291046949e-05,
      "loss": 2.018,
      "step": 6240
    },
    {
      "epoch": 1.0631963936378328,
      "grad_norm": 3.7861082553863525,
      "learning_rate": 1.2304085425583068e-05,
      "loss": 1.8683,
      "step": 6250
    },
    {
      "epoch": 1.0648975078676532,
      "grad_norm": 4.207472801208496,
      "learning_rate": 1.2299951163012763e-05,
      "loss": 1.9023,
      "step": 6260
    },
    {
      "epoch": 1.066598622097474,
      "grad_norm": 4.255404472351074,
      "learning_rate": 1.2295810127779271e-05,
      "loss": 1.9143,
      "step": 6270
    },
    {
      "epoch": 1.0682997363272944,
      "grad_norm": 4.092896461486816,
      "learning_rate": 1.2291662324911517e-05,
      "loss": 1.796,
      "step": 6280
    },
    {
      "epoch": 1.0700008505571148,
      "grad_norm": 4.437941074371338,
      "learning_rate": 1.2287507759446648e-05,
      "loss": 1.8307,
      "step": 6290
    },
    {
      "epoch": 1.0717019647869355,
      "grad_norm": 3.610339641571045,
      "learning_rate": 1.2283346436430017e-05,
      "loss": 1.82,
      "step": 6300
    },
    {
      "epoch": 1.073403079016756,
      "grad_norm": 3.712244987487793,
      "learning_rate": 1.2279178360915188e-05,
      "loss": 1.9487,
      "step": 6310
    },
    {
      "epoch": 1.0751041932465766,
      "grad_norm": 4.214893341064453,
      "learning_rate": 1.2275003537963923e-05,
      "loss": 1.825,
      "step": 6320
    },
    {
      "epoch": 1.076805307476397,
      "grad_norm": 3.998854875564575,
      "learning_rate": 1.2270821972646179e-05,
      "loss": 1.8145,
      "step": 6330
    },
    {
      "epoch": 1.0785064217062175,
      "grad_norm": 3.4029698371887207,
      "learning_rate": 1.2266633670040101e-05,
      "loss": 1.8831,
      "step": 6340
    },
    {
      "epoch": 1.0802075359360381,
      "grad_norm": 4.011364936828613,
      "learning_rate": 1.2262438635232015e-05,
      "loss": 1.7939,
      "step": 6350
    },
    {
      "epoch": 1.0819086501658586,
      "grad_norm": 4.054690837860107,
      "learning_rate": 1.2258236873316424e-05,
      "loss": 1.784,
      "step": 6360
    },
    {
      "epoch": 1.0836097643956792,
      "grad_norm": 4.267333984375,
      "learning_rate": 1.2254028389396e-05,
      "loss": 1.837,
      "step": 6370
    },
    {
      "epoch": 1.0853108786254997,
      "grad_norm": 4.028444290161133,
      "learning_rate": 1.2249813188581575e-05,
      "loss": 1.7764,
      "step": 6380
    },
    {
      "epoch": 1.0870119928553201,
      "grad_norm": 4.8640031814575195,
      "learning_rate": 1.2245591275992147e-05,
      "loss": 1.8812,
      "step": 6390
    },
    {
      "epoch": 1.0887131070851408,
      "grad_norm": 4.347374439239502,
      "learning_rate": 1.2241362656754855e-05,
      "loss": 1.8798,
      "step": 6400
    },
    {
      "epoch": 1.0904142213149612,
      "grad_norm": 3.810614824295044,
      "learning_rate": 1.223712733600499e-05,
      "loss": 1.9044,
      "step": 6410
    },
    {
      "epoch": 1.092115335544782,
      "grad_norm": 4.5110602378845215,
      "learning_rate": 1.2232885318885973e-05,
      "loss": 1.8203,
      "step": 6420
    },
    {
      "epoch": 1.0938164497746024,
      "grad_norm": 3.834718942642212,
      "learning_rate": 1.2228636610549368e-05,
      "loss": 1.939,
      "step": 6430
    },
    {
      "epoch": 1.0955175640044228,
      "grad_norm": 3.7686891555786133,
      "learning_rate": 1.2224381216154855e-05,
      "loss": 1.8401,
      "step": 6440
    },
    {
      "epoch": 1.0972186782342435,
      "grad_norm": 4.520566940307617,
      "learning_rate": 1.222011914087024e-05,
      "loss": 1.8598,
      "step": 6450
    },
    {
      "epoch": 1.098919792464064,
      "grad_norm": 4.536760330200195,
      "learning_rate": 1.2215850389871438e-05,
      "loss": 1.834,
      "step": 6460
    },
    {
      "epoch": 1.1006209066938846,
      "grad_norm": 4.597507476806641,
      "learning_rate": 1.2211574968342477e-05,
      "loss": 1.8079,
      "step": 6470
    },
    {
      "epoch": 1.102322020923705,
      "grad_norm": 3.8545501232147217,
      "learning_rate": 1.2207292881475478e-05,
      "loss": 1.9277,
      "step": 6480
    },
    {
      "epoch": 1.1040231351535255,
      "grad_norm": 4.181375026702881,
      "learning_rate": 1.2203004134470664e-05,
      "loss": 1.821,
      "step": 6490
    },
    {
      "epoch": 1.1057242493833461,
      "grad_norm": 3.9119484424591064,
      "learning_rate": 1.2198708732536338e-05,
      "loss": 1.8269,
      "step": 6500
    },
    {
      "epoch": 1.1074253636131666,
      "grad_norm": 3.5906999111175537,
      "learning_rate": 1.2194406680888894e-05,
      "loss": 1.8119,
      "step": 6510
    },
    {
      "epoch": 1.1091264778429872,
      "grad_norm": 3.4192230701446533,
      "learning_rate": 1.2190097984752797e-05,
      "loss": 1.7266,
      "step": 6520
    },
    {
      "epoch": 1.1108275920728077,
      "grad_norm": 3.0418713092803955,
      "learning_rate": 1.2185782649360577e-05,
      "loss": 1.9245,
      "step": 6530
    },
    {
      "epoch": 1.1125287063026281,
      "grad_norm": 3.510636568069458,
      "learning_rate": 1.2181460679952836e-05,
      "loss": 1.8278,
      "step": 6540
    },
    {
      "epoch": 1.1142298205324488,
      "grad_norm": 3.8723108768463135,
      "learning_rate": 1.2177132081778222e-05,
      "loss": 1.8933,
      "step": 6550
    },
    {
      "epoch": 1.1159309347622692,
      "grad_norm": 3.8740153312683105,
      "learning_rate": 1.217279686009344e-05,
      "loss": 1.8602,
      "step": 6560
    },
    {
      "epoch": 1.11763204899209,
      "grad_norm": 3.662519931793213,
      "learning_rate": 1.2168455020163242e-05,
      "loss": 1.8353,
      "step": 6570
    },
    {
      "epoch": 1.1193331632219103,
      "grad_norm": 3.3864052295684814,
      "learning_rate": 1.2164106567260405e-05,
      "loss": 1.9044,
      "step": 6580
    },
    {
      "epoch": 1.1210342774517308,
      "grad_norm": 3.839643955230713,
      "learning_rate": 1.215975150666575e-05,
      "loss": 1.8148,
      "step": 6590
    },
    {
      "epoch": 1.1227353916815515,
      "grad_norm": 4.567275524139404,
      "learning_rate": 1.2155389843668114e-05,
      "loss": 1.8031,
      "step": 6600
    },
    {
      "epoch": 1.124436505911372,
      "grad_norm": 3.8312809467315674,
      "learning_rate": 1.2151021583564355e-05,
      "loss": 1.8066,
      "step": 6610
    },
    {
      "epoch": 1.1261376201411926,
      "grad_norm": 4.2484211921691895,
      "learning_rate": 1.2146646731659344e-05,
      "loss": 1.849,
      "step": 6620
    },
    {
      "epoch": 1.127838734371013,
      "grad_norm": 3.935384750366211,
      "learning_rate": 1.2142265293265952e-05,
      "loss": 1.8554,
      "step": 6630
    },
    {
      "epoch": 1.1295398486008335,
      "grad_norm": 2.8967413902282715,
      "learning_rate": 1.2137877273705054e-05,
      "loss": 1.8206,
      "step": 6640
    },
    {
      "epoch": 1.1312409628306541,
      "grad_norm": 3.9680368900299072,
      "learning_rate": 1.2133482678305516e-05,
      "loss": 1.8569,
      "step": 6650
    },
    {
      "epoch": 1.1329420770604746,
      "grad_norm": 3.64766788482666,
      "learning_rate": 1.212908151240419e-05,
      "loss": 1.8013,
      "step": 6660
    },
    {
      "epoch": 1.1346431912902952,
      "grad_norm": 4.119669437408447,
      "learning_rate": 1.2124673781345903e-05,
      "loss": 1.8953,
      "step": 6670
    },
    {
      "epoch": 1.1363443055201157,
      "grad_norm": 4.643278121948242,
      "learning_rate": 1.2120259490483464e-05,
      "loss": 1.7907,
      "step": 6680
    },
    {
      "epoch": 1.1380454197499361,
      "grad_norm": 4.240262031555176,
      "learning_rate": 1.211583864517764e-05,
      "loss": 1.8067,
      "step": 6690
    },
    {
      "epoch": 1.1397465339797568,
      "grad_norm": 4.9946489334106445,
      "learning_rate": 1.2111411250797156e-05,
      "loss": 1.8233,
      "step": 6700
    },
    {
      "epoch": 1.1414476482095772,
      "grad_norm": 4.387408256530762,
      "learning_rate": 1.2106977312718699e-05,
      "loss": 1.86,
      "step": 6710
    },
    {
      "epoch": 1.143148762439398,
      "grad_norm": 3.9364635944366455,
      "learning_rate": 1.2102536836326901e-05,
      "loss": 1.8694,
      "step": 6720
    },
    {
      "epoch": 1.1448498766692183,
      "grad_norm": 4.609739780426025,
      "learning_rate": 1.2098089827014327e-05,
      "loss": 1.8502,
      "step": 6730
    },
    {
      "epoch": 1.1465509908990388,
      "grad_norm": 4.438980579376221,
      "learning_rate": 1.2093636290181483e-05,
      "loss": 1.851,
      "step": 6740
    },
    {
      "epoch": 1.1482521051288594,
      "grad_norm": 4.564301490783691,
      "learning_rate": 1.2089176231236798e-05,
      "loss": 1.886,
      "step": 6750
    },
    {
      "epoch": 1.14995321935868,
      "grad_norm": 4.274376392364502,
      "learning_rate": 1.2084709655596623e-05,
      "loss": 1.8449,
      "step": 6760
    },
    {
      "epoch": 1.1516543335885006,
      "grad_norm": 5.069614887237549,
      "learning_rate": 1.2080236568685226e-05,
      "loss": 1.8416,
      "step": 6770
    },
    {
      "epoch": 1.153355447818321,
      "grad_norm": 3.986492156982422,
      "learning_rate": 1.2075756975934775e-05,
      "loss": 1.8218,
      "step": 6780
    },
    {
      "epoch": 1.1550565620481414,
      "grad_norm": 4.155921936035156,
      "learning_rate": 1.2071270882785346e-05,
      "loss": 1.8001,
      "step": 6790
    },
    {
      "epoch": 1.156757676277962,
      "grad_norm": 4.066920757293701,
      "learning_rate": 1.2066778294684905e-05,
      "loss": 1.8378,
      "step": 6800
    },
    {
      "epoch": 1.1584587905077826,
      "grad_norm": 4.247832775115967,
      "learning_rate": 1.2062279217089307e-05,
      "loss": 1.8766,
      "step": 6810
    },
    {
      "epoch": 1.1601599047376032,
      "grad_norm": 3.875669002532959,
      "learning_rate": 1.2057773655462288e-05,
      "loss": 1.8497,
      "step": 6820
    },
    {
      "epoch": 1.1618610189674237,
      "grad_norm": 4.125594139099121,
      "learning_rate": 1.2053261615275455e-05,
      "loss": 1.9058,
      "step": 6830
    },
    {
      "epoch": 1.163562133197244,
      "grad_norm": 4.312350273132324,
      "learning_rate": 1.2048743102008293e-05,
      "loss": 1.9242,
      "step": 6840
    },
    {
      "epoch": 1.1652632474270648,
      "grad_norm": 3.9462075233459473,
      "learning_rate": 1.2044218121148132e-05,
      "loss": 1.9029,
      "step": 6850
    },
    {
      "epoch": 1.1669643616568852,
      "grad_norm": 4.291337013244629,
      "learning_rate": 1.2039686678190173e-05,
      "loss": 1.8806,
      "step": 6860
    },
    {
      "epoch": 1.1686654758867059,
      "grad_norm": 3.9357290267944336,
      "learning_rate": 1.2035148778637448e-05,
      "loss": 1.8882,
      "step": 6870
    },
    {
      "epoch": 1.1703665901165263,
      "grad_norm": 3.846447467803955,
      "learning_rate": 1.2030604428000846e-05,
      "loss": 1.906,
      "step": 6880
    },
    {
      "epoch": 1.1720677043463468,
      "grad_norm": 4.311671257019043,
      "learning_rate": 1.202605363179908e-05,
      "loss": 1.8491,
      "step": 6890
    },
    {
      "epoch": 1.1737688185761674,
      "grad_norm": 3.54050350189209,
      "learning_rate": 1.2021496395558695e-05,
      "loss": 1.9521,
      "step": 6900
    },
    {
      "epoch": 1.1754699328059879,
      "grad_norm": 4.3173651695251465,
      "learning_rate": 1.2016932724814051e-05,
      "loss": 1.764,
      "step": 6910
    },
    {
      "epoch": 1.1771710470358085,
      "grad_norm": 3.853074550628662,
      "learning_rate": 1.2012362625107332e-05,
      "loss": 1.7911,
      "step": 6920
    },
    {
      "epoch": 1.178872161265629,
      "grad_norm": 3.8800690174102783,
      "learning_rate": 1.2007786101988517e-05,
      "loss": 1.8936,
      "step": 6930
    },
    {
      "epoch": 1.1805732754954494,
      "grad_norm": 4.200761795043945,
      "learning_rate": 1.20032031610154e-05,
      "loss": 1.8044,
      "step": 6940
    },
    {
      "epoch": 1.18227438972527,
      "grad_norm": 4.506488800048828,
      "learning_rate": 1.1998613807753559e-05,
      "loss": 1.777,
      "step": 6950
    },
    {
      "epoch": 1.1839755039550905,
      "grad_norm": 3.8466339111328125,
      "learning_rate": 1.199401804777636e-05,
      "loss": 1.8042,
      "step": 6960
    },
    {
      "epoch": 1.1856766181849112,
      "grad_norm": 3.482567310333252,
      "learning_rate": 1.198941588666495e-05,
      "loss": 1.872,
      "step": 6970
    },
    {
      "epoch": 1.1873777324147317,
      "grad_norm": 3.968972682952881,
      "learning_rate": 1.1984807330008253e-05,
      "loss": 1.8401,
      "step": 6980
    },
    {
      "epoch": 1.189078846644552,
      "grad_norm": 4.146262168884277,
      "learning_rate": 1.1980192383402957e-05,
      "loss": 1.8257,
      "step": 6990
    },
    {
      "epoch": 1.1907799608743728,
      "grad_norm": 4.3229079246521,
      "learning_rate": 1.197557105245351e-05,
      "loss": 1.9198,
      "step": 7000
    },
    {
      "epoch": 1.1924810751041932,
      "grad_norm": 4.2531914710998535,
      "learning_rate": 1.1970943342772116e-05,
      "loss": 1.8715,
      "step": 7010
    },
    {
      "epoch": 1.1941821893340139,
      "grad_norm": 3.7825818061828613,
      "learning_rate": 1.1966309259978719e-05,
      "loss": 1.8092,
      "step": 7020
    },
    {
      "epoch": 1.1958833035638343,
      "grad_norm": 4.886743545532227,
      "learning_rate": 1.196166880970101e-05,
      "loss": 1.8566,
      "step": 7030
    },
    {
      "epoch": 1.1975844177936548,
      "grad_norm": 4.063226222991943,
      "learning_rate": 1.1957021997574409e-05,
      "loss": 1.909,
      "step": 7040
    },
    {
      "epoch": 1.1992855320234754,
      "grad_norm": 4.031005382537842,
      "learning_rate": 1.195236882924206e-05,
      "loss": 1.8579,
      "step": 7050
    },
    {
      "epoch": 1.2009866462532959,
      "grad_norm": 4.141384601593018,
      "learning_rate": 1.1947709310354832e-05,
      "loss": 1.9419,
      "step": 7060
    },
    {
      "epoch": 1.2026877604831165,
      "grad_norm": 4.252750396728516,
      "learning_rate": 1.19430434465713e-05,
      "loss": 1.9514,
      "step": 7070
    },
    {
      "epoch": 1.204388874712937,
      "grad_norm": 4.281816482543945,
      "learning_rate": 1.1938371243557747e-05,
      "loss": 1.9094,
      "step": 7080
    },
    {
      "epoch": 1.2060899889427574,
      "grad_norm": 4.234149932861328,
      "learning_rate": 1.1933692706988156e-05,
      "loss": 1.8436,
      "step": 7090
    },
    {
      "epoch": 1.207791103172578,
      "grad_norm": 4.110588073730469,
      "learning_rate": 1.1929007842544197e-05,
      "loss": 1.8141,
      "step": 7100
    },
    {
      "epoch": 1.2094922174023985,
      "grad_norm": 4.462185382843018,
      "learning_rate": 1.1924316655915232e-05,
      "loss": 1.8112,
      "step": 7110
    },
    {
      "epoch": 1.2111933316322192,
      "grad_norm": 3.660940408706665,
      "learning_rate": 1.1919619152798293e-05,
      "loss": 1.8147,
      "step": 7120
    },
    {
      "epoch": 1.2128944458620396,
      "grad_norm": 3.804466724395752,
      "learning_rate": 1.191491533889809e-05,
      "loss": 1.8511,
      "step": 7130
    },
    {
      "epoch": 1.21459556009186,
      "grad_norm": 4.5483293533325195,
      "learning_rate": 1.1910205219926986e-05,
      "loss": 1.8949,
      "step": 7140
    },
    {
      "epoch": 1.2162966743216808,
      "grad_norm": 4.025954246520996,
      "learning_rate": 1.1905488801605014e-05,
      "loss": 1.9193,
      "step": 7150
    },
    {
      "epoch": 1.2179977885515012,
      "grad_norm": 4.330733299255371,
      "learning_rate": 1.1900766089659847e-05,
      "loss": 1.8404,
      "step": 7160
    },
    {
      "epoch": 1.2196989027813219,
      "grad_norm": 3.604673385620117,
      "learning_rate": 1.189603708982681e-05,
      "loss": 1.9742,
      "step": 7170
    },
    {
      "epoch": 1.2214000170111423,
      "grad_norm": 4.058786869049072,
      "learning_rate": 1.1891301807848854e-05,
      "loss": 1.8803,
      "step": 7180
    },
    {
      "epoch": 1.2231011312409628,
      "grad_norm": 4.989222526550293,
      "learning_rate": 1.1886560249476568e-05,
      "loss": 1.7776,
      "step": 7190
    },
    {
      "epoch": 1.2248022454707834,
      "grad_norm": 4.440055847167969,
      "learning_rate": 1.1881812420468158e-05,
      "loss": 1.8804,
      "step": 7200
    },
    {
      "epoch": 1.2265033597006039,
      "grad_norm": 4.92519998550415,
      "learning_rate": 1.1877058326589447e-05,
      "loss": 1.7766,
      "step": 7210
    },
    {
      "epoch": 1.2282044739304245,
      "grad_norm": 3.1428630352020264,
      "learning_rate": 1.1872297973613864e-05,
      "loss": 1.9083,
      "step": 7220
    },
    {
      "epoch": 1.229905588160245,
      "grad_norm": 4.2585225105285645,
      "learning_rate": 1.1867531367322442e-05,
      "loss": 1.8742,
      "step": 7230
    },
    {
      "epoch": 1.2316067023900654,
      "grad_norm": 3.5493762493133545,
      "learning_rate": 1.1862758513503807e-05,
      "loss": 1.8458,
      "step": 7240
    },
    {
      "epoch": 1.233307816619886,
      "grad_norm": 4.4121880531311035,
      "learning_rate": 1.185797941795417e-05,
      "loss": 1.8574,
      "step": 7250
    },
    {
      "epoch": 1.2350089308497065,
      "grad_norm": 3.277613401412964,
      "learning_rate": 1.185319408647733e-05,
      "loss": 1.8732,
      "step": 7260
    },
    {
      "epoch": 1.2367100450795272,
      "grad_norm": 3.6618874073028564,
      "learning_rate": 1.1848402524884646e-05,
      "loss": 1.8563,
      "step": 7270
    },
    {
      "epoch": 1.2384111593093476,
      "grad_norm": 4.217512607574463,
      "learning_rate": 1.1843604738995052e-05,
      "loss": 1.8422,
      "step": 7280
    },
    {
      "epoch": 1.240112273539168,
      "grad_norm": 4.140598773956299,
      "learning_rate": 1.183880073463504e-05,
      "loss": 1.8516,
      "step": 7290
    },
    {
      "epoch": 1.2418133877689888,
      "grad_norm": 3.781187057495117,
      "learning_rate": 1.1833990517638654e-05,
      "loss": 1.8753,
      "step": 7300
    },
    {
      "epoch": 1.2435145019988092,
      "grad_norm": 3.7208242416381836,
      "learning_rate": 1.1829174093847479e-05,
      "loss": 1.793,
      "step": 7310
    },
    {
      "epoch": 1.2452156162286299,
      "grad_norm": 3.963973045349121,
      "learning_rate": 1.1824351469110637e-05,
      "loss": 1.8121,
      "step": 7320
    },
    {
      "epoch": 1.2469167304584503,
      "grad_norm": 4.138283729553223,
      "learning_rate": 1.181952264928479e-05,
      "loss": 1.8787,
      "step": 7330
    },
    {
      "epoch": 1.2486178446882708,
      "grad_norm": 4.180245876312256,
      "learning_rate": 1.1814687640234112e-05,
      "loss": 1.8196,
      "step": 7340
    },
    {
      "epoch": 1.2503189589180914,
      "grad_norm": 3.703684091567993,
      "learning_rate": 1.1809846447830301e-05,
      "loss": 1.865,
      "step": 7350
    },
    {
      "epoch": 1.2520200731479119,
      "grad_norm": 4.624788761138916,
      "learning_rate": 1.1804999077952558e-05,
      "loss": 1.827,
      "step": 7360
    },
    {
      "epoch": 1.2537211873777325,
      "grad_norm": 4.256951332092285,
      "learning_rate": 1.1800145536487591e-05,
      "loss": 1.8149,
      "step": 7370
    },
    {
      "epoch": 1.255422301607553,
      "grad_norm": 4.586760520935059,
      "learning_rate": 1.1795285829329602e-05,
      "loss": 1.7933,
      "step": 7380
    },
    {
      "epoch": 1.2571234158373734,
      "grad_norm": 3.419952869415283,
      "learning_rate": 1.1790419962380278e-05,
      "loss": 1.8373,
      "step": 7390
    },
    {
      "epoch": 1.258824530067194,
      "grad_norm": 3.8896842002868652,
      "learning_rate": 1.178554794154879e-05,
      "loss": 1.8466,
      "step": 7400
    },
    {
      "epoch": 1.2605256442970145,
      "grad_norm": 3.539841651916504,
      "learning_rate": 1.1780669772751775e-05,
      "loss": 1.9326,
      "step": 7410
    },
    {
      "epoch": 1.2622267585268352,
      "grad_norm": 3.7069525718688965,
      "learning_rate": 1.1775785461913352e-05,
      "loss": 1.8622,
      "step": 7420
    },
    {
      "epoch": 1.2639278727566556,
      "grad_norm": 3.6968042850494385,
      "learning_rate": 1.177089501496508e-05,
      "loss": 1.8145,
      "step": 7430
    },
    {
      "epoch": 1.265628986986476,
      "grad_norm": 3.9641242027282715,
      "learning_rate": 1.1765998437845982e-05,
      "loss": 1.8396,
      "step": 7440
    },
    {
      "epoch": 1.2673301012162967,
      "grad_norm": 3.476146697998047,
      "learning_rate": 1.1761095736502525e-05,
      "loss": 1.8247,
      "step": 7450
    },
    {
      "epoch": 1.2690312154461172,
      "grad_norm": 4.397093772888184,
      "learning_rate": 1.1756186916888604e-05,
      "loss": 1.7688,
      "step": 7460
    },
    {
      "epoch": 1.2707323296759379,
      "grad_norm": 3.8471662998199463,
      "learning_rate": 1.1751271984965555e-05,
      "loss": 1.9181,
      "step": 7470
    },
    {
      "epoch": 1.2724334439057583,
      "grad_norm": 4.207676887512207,
      "learning_rate": 1.1746350946702128e-05,
      "loss": 1.7721,
      "step": 7480
    },
    {
      "epoch": 1.2741345581355787,
      "grad_norm": 4.509179592132568,
      "learning_rate": 1.1741423808074499e-05,
      "loss": 1.8171,
      "step": 7490
    },
    {
      "epoch": 1.2758356723653994,
      "grad_norm": 3.6446003913879395,
      "learning_rate": 1.1736490575066243e-05,
      "loss": 1.8396,
      "step": 7500
    },
    {
      "epoch": 1.2775367865952199,
      "grad_norm": 4.312423229217529,
      "learning_rate": 1.173155125366834e-05,
      "loss": 1.7823,
      "step": 7510
    },
    {
      "epoch": 1.2792379008250405,
      "grad_norm": 3.0327653884887695,
      "learning_rate": 1.1726605849879162e-05,
      "loss": 1.8165,
      "step": 7520
    },
    {
      "epoch": 1.280939015054861,
      "grad_norm": 3.464770793914795,
      "learning_rate": 1.1721654369704471e-05,
      "loss": 1.8435,
      "step": 7530
    },
    {
      "epoch": 1.2826401292846814,
      "grad_norm": 3.883833885192871,
      "learning_rate": 1.1716696819157403e-05,
      "loss": 1.9098,
      "step": 7540
    },
    {
      "epoch": 1.284341243514502,
      "grad_norm": 3.4528632164001465,
      "learning_rate": 1.1711733204258472e-05,
      "loss": 1.8664,
      "step": 7550
    },
    {
      "epoch": 1.2860423577443225,
      "grad_norm": 4.106072902679443,
      "learning_rate": 1.1706763531035554e-05,
      "loss": 1.8568,
      "step": 7560
    },
    {
      "epoch": 1.2877434719741432,
      "grad_norm": 4.352649211883545,
      "learning_rate": 1.1701787805523879e-05,
      "loss": 1.8838,
      "step": 7570
    },
    {
      "epoch": 1.2894445862039636,
      "grad_norm": 4.048183441162109,
      "learning_rate": 1.1696806033766031e-05,
      "loss": 1.7881,
      "step": 7580
    },
    {
      "epoch": 1.291145700433784,
      "grad_norm": 3.615002155303955,
      "learning_rate": 1.1691818221811937e-05,
      "loss": 1.7633,
      "step": 7590
    },
    {
      "epoch": 1.2928468146636047,
      "grad_norm": 3.5091981887817383,
      "learning_rate": 1.1686824375718855e-05,
      "loss": 1.8125,
      "step": 7600
    },
    {
      "epoch": 1.2945479288934252,
      "grad_norm": 3.6418280601501465,
      "learning_rate": 1.1681824501551377e-05,
      "loss": 1.752,
      "step": 7610
    },
    {
      "epoch": 1.2962490431232458,
      "grad_norm": 3.4704437255859375,
      "learning_rate": 1.1676818605381409e-05,
      "loss": 1.8887,
      "step": 7620
    },
    {
      "epoch": 1.2979501573530663,
      "grad_norm": 3.6811389923095703,
      "learning_rate": 1.1671806693288177e-05,
      "loss": 1.8914,
      "step": 7630
    },
    {
      "epoch": 1.2996512715828867,
      "grad_norm": 3.922544240951538,
      "learning_rate": 1.1666788771358206e-05,
      "loss": 1.8412,
      "step": 7640
    },
    {
      "epoch": 1.3013523858127074,
      "grad_norm": 3.4494495391845703,
      "learning_rate": 1.1661764845685325e-05,
      "loss": 1.842,
      "step": 7650
    },
    {
      "epoch": 1.3030535000425278,
      "grad_norm": 4.570243835449219,
      "learning_rate": 1.165673492237065e-05,
      "loss": 1.7121,
      "step": 7660
    },
    {
      "epoch": 1.3047546142723485,
      "grad_norm": 3.663918972015381,
      "learning_rate": 1.1651699007522584e-05,
      "loss": 1.7647,
      "step": 7670
    },
    {
      "epoch": 1.306455728502169,
      "grad_norm": 4.224978446960449,
      "learning_rate": 1.1646657107256803e-05,
      "loss": 1.86,
      "step": 7680
    },
    {
      "epoch": 1.3081568427319894,
      "grad_norm": 3.920491933822632,
      "learning_rate": 1.1641609227696256e-05,
      "loss": 1.8903,
      "step": 7690
    },
    {
      "epoch": 1.30985795696181,
      "grad_norm": 3.300448179244995,
      "learning_rate": 1.1636555374971147e-05,
      "loss": 1.9144,
      "step": 7700
    },
    {
      "epoch": 1.3115590711916305,
      "grad_norm": 4.223546981811523,
      "learning_rate": 1.163149555521894e-05,
      "loss": 1.8116,
      "step": 7710
    },
    {
      "epoch": 1.3132601854214512,
      "grad_norm": 4.763315677642822,
      "learning_rate": 1.1626429774584343e-05,
      "loss": 1.8878,
      "step": 7720
    },
    {
      "epoch": 1.3149612996512716,
      "grad_norm": 3.610171318054199,
      "learning_rate": 1.1621358039219301e-05,
      "loss": 1.8383,
      "step": 7730
    },
    {
      "epoch": 1.316662413881092,
      "grad_norm": 3.96856427192688,
      "learning_rate": 1.1616280355282994e-05,
      "loss": 1.8283,
      "step": 7740
    },
    {
      "epoch": 1.3183635281109127,
      "grad_norm": 4.092318058013916,
      "learning_rate": 1.1611196728941822e-05,
      "loss": 1.8411,
      "step": 7750
    },
    {
      "epoch": 1.3200646423407332,
      "grad_norm": 3.7782583236694336,
      "learning_rate": 1.1606107166369409e-05,
      "loss": 1.8832,
      "step": 7760
    },
    {
      "epoch": 1.3217657565705538,
      "grad_norm": 3.8647406101226807,
      "learning_rate": 1.1601011673746579e-05,
      "loss": 1.8914,
      "step": 7770
    },
    {
      "epoch": 1.3234668708003743,
      "grad_norm": 4.192235469818115,
      "learning_rate": 1.1595910257261363e-05,
      "loss": 1.8367,
      "step": 7780
    },
    {
      "epoch": 1.3251679850301947,
      "grad_norm": 4.087222099304199,
      "learning_rate": 1.1590802923108984e-05,
      "loss": 1.7934,
      "step": 7790
    },
    {
      "epoch": 1.3268690992600152,
      "grad_norm": 4.581682205200195,
      "learning_rate": 1.1585689677491854e-05,
      "loss": 1.8093,
      "step": 7800
    },
    {
      "epoch": 1.3285702134898358,
      "grad_norm": 4.077419281005859,
      "learning_rate": 1.1580570526619559e-05,
      "loss": 1.8714,
      "step": 7810
    },
    {
      "epoch": 1.3302713277196565,
      "grad_norm": 3.7243072986602783,
      "learning_rate": 1.157544547670886e-05,
      "loss": 1.8082,
      "step": 7820
    },
    {
      "epoch": 1.331972441949477,
      "grad_norm": 4.393478870391846,
      "learning_rate": 1.1570314533983683e-05,
      "loss": 1.911,
      "step": 7830
    },
    {
      "epoch": 1.3336735561792974,
      "grad_norm": 3.8908112049102783,
      "learning_rate": 1.1565177704675106e-05,
      "loss": 1.8287,
      "step": 7840
    },
    {
      "epoch": 1.3353746704091178,
      "grad_norm": 3.665543794631958,
      "learning_rate": 1.1560034995021359e-05,
      "loss": 1.8854,
      "step": 7850
    },
    {
      "epoch": 1.3370757846389385,
      "grad_norm": 3.881185531616211,
      "learning_rate": 1.1554886411267817e-05,
      "loss": 1.8453,
      "step": 7860
    },
    {
      "epoch": 1.3387768988687592,
      "grad_norm": 4.123805046081543,
      "learning_rate": 1.1549731959666976e-05,
      "loss": 1.8287,
      "step": 7870
    },
    {
      "epoch": 1.3404780130985796,
      "grad_norm": 3.782844066619873,
      "learning_rate": 1.1544571646478473e-05,
      "loss": 1.824,
      "step": 7880
    },
    {
      "epoch": 1.3421791273284,
      "grad_norm": 3.8619236946105957,
      "learning_rate": 1.1539405477969054e-05,
      "loss": 1.8937,
      "step": 7890
    },
    {
      "epoch": 1.3438802415582205,
      "grad_norm": 4.254144668579102,
      "learning_rate": 1.1534233460412579e-05,
      "loss": 1.8273,
      "step": 7900
    },
    {
      "epoch": 1.3455813557880412,
      "grad_norm": 4.115326881408691,
      "learning_rate": 1.1529055600090007e-05,
      "loss": 1.8292,
      "step": 7910
    },
    {
      "epoch": 1.3472824700178618,
      "grad_norm": 4.696479797363281,
      "learning_rate": 1.15238719032894e-05,
      "loss": 1.8741,
      "step": 7920
    },
    {
      "epoch": 1.3489835842476823,
      "grad_norm": 3.319958448410034,
      "learning_rate": 1.15186823763059e-05,
      "loss": 2.0001,
      "step": 7930
    },
    {
      "epoch": 1.3506846984775027,
      "grad_norm": 4.459231376647949,
      "learning_rate": 1.1513487025441738e-05,
      "loss": 1.7909,
      "step": 7940
    },
    {
      "epoch": 1.3523858127073232,
      "grad_norm": 3.6977946758270264,
      "learning_rate": 1.150828585700621e-05,
      "loss": 1.9375,
      "step": 7950
    },
    {
      "epoch": 1.3540869269371438,
      "grad_norm": 3.963927745819092,
      "learning_rate": 1.1503078877315678e-05,
      "loss": 1.8812,
      "step": 7960
    },
    {
      "epoch": 1.3557880411669645,
      "grad_norm": 3.8123812675476074,
      "learning_rate": 1.1497866092693564e-05,
      "loss": 1.8044,
      "step": 7970
    },
    {
      "epoch": 1.357489155396785,
      "grad_norm": 4.007399559020996,
      "learning_rate": 1.149264750947034e-05,
      "loss": 1.9753,
      "step": 7980
    },
    {
      "epoch": 1.3591902696266054,
      "grad_norm": 4.110389709472656,
      "learning_rate": 1.1487423133983515e-05,
      "loss": 1.8695,
      "step": 7990
    },
    {
      "epoch": 1.3608913838564258,
      "grad_norm": 4.561108589172363,
      "learning_rate": 1.148219297257764e-05,
      "loss": 1.8648,
      "step": 8000
    },
    {
      "epoch": 1.3625924980862465,
      "grad_norm": 4.409726619720459,
      "learning_rate": 1.1476957031604283e-05,
      "loss": 1.8565,
      "step": 8010
    },
    {
      "epoch": 1.3642936123160672,
      "grad_norm": 3.8639228343963623,
      "learning_rate": 1.147171531742204e-05,
      "loss": 1.889,
      "step": 8020
    },
    {
      "epoch": 1.3659947265458876,
      "grad_norm": 4.365349769592285,
      "learning_rate": 1.146646783639651e-05,
      "loss": 1.8925,
      "step": 8030
    },
    {
      "epoch": 1.367695840775708,
      "grad_norm": 4.182748794555664,
      "learning_rate": 1.1461214594900302e-05,
      "loss": 1.7287,
      "step": 8040
    },
    {
      "epoch": 1.3693969550055285,
      "grad_norm": 3.912654161453247,
      "learning_rate": 1.1455955599313017e-05,
      "loss": 1.7366,
      "step": 8050
    },
    {
      "epoch": 1.3710980692353492,
      "grad_norm": 4.697111129760742,
      "learning_rate": 1.1450690856021244e-05,
      "loss": 1.8811,
      "step": 8060
    },
    {
      "epoch": 1.3727991834651696,
      "grad_norm": 4.002067565917969,
      "learning_rate": 1.1445420371418552e-05,
      "loss": 1.8407,
      "step": 8070
    },
    {
      "epoch": 1.3745002976949903,
      "grad_norm": 4.642768859863281,
      "learning_rate": 1.1440144151905489e-05,
      "loss": 1.8951,
      "step": 8080
    },
    {
      "epoch": 1.3762014119248107,
      "grad_norm": 4.091320514678955,
      "learning_rate": 1.1434862203889552e-05,
      "loss": 1.7457,
      "step": 8090
    },
    {
      "epoch": 1.3779025261546312,
      "grad_norm": 4.746745586395264,
      "learning_rate": 1.142957453378521e-05,
      "loss": 1.7577,
      "step": 8100
    },
    {
      "epoch": 1.3796036403844518,
      "grad_norm": 3.4975032806396484,
      "learning_rate": 1.1424281148013873e-05,
      "loss": 1.8887,
      "step": 8110
    },
    {
      "epoch": 1.3813047546142723,
      "grad_norm": 4.448775291442871,
      "learning_rate": 1.1418982053003897e-05,
      "loss": 1.8713,
      "step": 8120
    },
    {
      "epoch": 1.383005868844093,
      "grad_norm": 3.9427380561828613,
      "learning_rate": 1.1413677255190569e-05,
      "loss": 1.8355,
      "step": 8130
    },
    {
      "epoch": 1.3847069830739134,
      "grad_norm": 4.403369426727295,
      "learning_rate": 1.1408366761016097e-05,
      "loss": 1.9192,
      "step": 8140
    },
    {
      "epoch": 1.3864080973037338,
      "grad_norm": 4.291594982147217,
      "learning_rate": 1.1403050576929615e-05,
      "loss": 1.833,
      "step": 8150
    },
    {
      "epoch": 1.3881092115335545,
      "grad_norm": 4.179330825805664,
      "learning_rate": 1.139772870938716e-05,
      "loss": 1.787,
      "step": 8160
    },
    {
      "epoch": 1.389810325763375,
      "grad_norm": 4.093100070953369,
      "learning_rate": 1.1392401164851679e-05,
      "loss": 1.8275,
      "step": 8170
    },
    {
      "epoch": 1.3915114399931956,
      "grad_norm": 4.226815700531006,
      "learning_rate": 1.1387067949793004e-05,
      "loss": 1.8355,
      "step": 8180
    },
    {
      "epoch": 1.393212554223016,
      "grad_norm": 4.04533052444458,
      "learning_rate": 1.138172907068786e-05,
      "loss": 1.7927,
      "step": 8190
    },
    {
      "epoch": 1.3949136684528365,
      "grad_norm": 3.832703113555908,
      "learning_rate": 1.1376384534019843e-05,
      "loss": 1.8264,
      "step": 8200
    },
    {
      "epoch": 1.3966147826826572,
      "grad_norm": 4.171768665313721,
      "learning_rate": 1.1371034346279432e-05,
      "loss": 1.7574,
      "step": 8210
    },
    {
      "epoch": 1.3983158969124776,
      "grad_norm": 3.803027868270874,
      "learning_rate": 1.1365678513963956e-05,
      "loss": 1.8445,
      "step": 8220
    },
    {
      "epoch": 1.4000170111422983,
      "grad_norm": 3.384702444076538,
      "learning_rate": 1.1360317043577609e-05,
      "loss": 1.8349,
      "step": 8230
    },
    {
      "epoch": 1.4017181253721187,
      "grad_norm": 4.899093151092529,
      "learning_rate": 1.1354949941631423e-05,
      "loss": 1.8671,
      "step": 8240
    },
    {
      "epoch": 1.4034192396019392,
      "grad_norm": 4.261693477630615,
      "learning_rate": 1.1349577214643276e-05,
      "loss": 1.8051,
      "step": 8250
    },
    {
      "epoch": 1.4051203538317598,
      "grad_norm": 4.3631062507629395,
      "learning_rate": 1.1344198869137871e-05,
      "loss": 1.8807,
      "step": 8260
    },
    {
      "epoch": 1.4068214680615803,
      "grad_norm": 4.09499454498291,
      "learning_rate": 1.1338814911646742e-05,
      "loss": 1.7799,
      "step": 8270
    },
    {
      "epoch": 1.408522582291401,
      "grad_norm": 3.800813913345337,
      "learning_rate": 1.133342534870823e-05,
      "loss": 1.9407,
      "step": 8280
    },
    {
      "epoch": 1.4102236965212214,
      "grad_norm": 3.953648328781128,
      "learning_rate": 1.132803018686749e-05,
      "loss": 1.7741,
      "step": 8290
    },
    {
      "epoch": 1.4119248107510418,
      "grad_norm": 3.5527772903442383,
      "learning_rate": 1.1322629432676473e-05,
      "loss": 1.9239,
      "step": 8300
    },
    {
      "epoch": 1.4136259249808625,
      "grad_norm": 3.6812546253204346,
      "learning_rate": 1.1317223092693921e-05,
      "loss": 1.865,
      "step": 8310
    },
    {
      "epoch": 1.415327039210683,
      "grad_norm": 3.842613935470581,
      "learning_rate": 1.131181117348536e-05,
      "loss": 1.8384,
      "step": 8320
    },
    {
      "epoch": 1.4170281534405036,
      "grad_norm": 4.840838432312012,
      "learning_rate": 1.1306393681623095e-05,
      "loss": 1.7868,
      "step": 8330
    },
    {
      "epoch": 1.418729267670324,
      "grad_norm": 3.6300811767578125,
      "learning_rate": 1.1300970623686192e-05,
      "loss": 1.7641,
      "step": 8340
    },
    {
      "epoch": 1.4204303819001445,
      "grad_norm": 4.901582717895508,
      "learning_rate": 1.1295542006260482e-05,
      "loss": 1.8725,
      "step": 8350
    },
    {
      "epoch": 1.4221314961299651,
      "grad_norm": 3.8441402912139893,
      "learning_rate": 1.1290107835938543e-05,
      "loss": 1.8115,
      "step": 8360
    },
    {
      "epoch": 1.4238326103597856,
      "grad_norm": 4.257385730743408,
      "learning_rate": 1.1284668119319701e-05,
      "loss": 1.6997,
      "step": 8370
    },
    {
      "epoch": 1.4255337245896063,
      "grad_norm": 3.5949318408966064,
      "learning_rate": 1.1279222863010015e-05,
      "loss": 1.9023,
      "step": 8380
    },
    {
      "epoch": 1.4272348388194267,
      "grad_norm": 3.525285005569458,
      "learning_rate": 1.1273772073622273e-05,
      "loss": 1.7829,
      "step": 8390
    },
    {
      "epoch": 1.4289359530492471,
      "grad_norm": 4.029623508453369,
      "learning_rate": 1.126831575777598e-05,
      "loss": 1.7838,
      "step": 8400
    },
    {
      "epoch": 1.4306370672790678,
      "grad_norm": 4.243871212005615,
      "learning_rate": 1.1262853922097353e-05,
      "loss": 1.8764,
      "step": 8410
    },
    {
      "epoch": 1.4323381815088883,
      "grad_norm": 3.630593776702881,
      "learning_rate": 1.1257386573219315e-05,
      "loss": 1.8042,
      "step": 8420
    },
    {
      "epoch": 1.434039295738709,
      "grad_norm": 3.6100449562072754,
      "learning_rate": 1.1251913717781481e-05,
      "loss": 1.792,
      "step": 8430
    },
    {
      "epoch": 1.4357404099685294,
      "grad_norm": 3.461958885192871,
      "learning_rate": 1.1246435362430155e-05,
      "loss": 1.8373,
      "step": 8440
    },
    {
      "epoch": 1.4374415241983498,
      "grad_norm": 3.8997554779052734,
      "learning_rate": 1.1240951513818321e-05,
      "loss": 1.8916,
      "step": 8450
    },
    {
      "epoch": 1.4391426384281705,
      "grad_norm": 5.204623699188232,
      "learning_rate": 1.123546217860563e-05,
      "loss": 1.9023,
      "step": 8460
    },
    {
      "epoch": 1.440843752657991,
      "grad_norm": 3.4482414722442627,
      "learning_rate": 1.1229967363458403e-05,
      "loss": 1.7964,
      "step": 8470
    },
    {
      "epoch": 1.4425448668878116,
      "grad_norm": 3.914092540740967,
      "learning_rate": 1.122446707504961e-05,
      "loss": 1.7912,
      "step": 8480
    },
    {
      "epoch": 1.444245981117632,
      "grad_norm": 3.4906105995178223,
      "learning_rate": 1.1218961320058867e-05,
      "loss": 1.7525,
      "step": 8490
    },
    {
      "epoch": 1.4459470953474525,
      "grad_norm": 3.8349711894989014,
      "learning_rate": 1.1213450105172432e-05,
      "loss": 1.8479,
      "step": 8500
    },
    {
      "epoch": 1.4476482095772731,
      "grad_norm": 4.6149115562438965,
      "learning_rate": 1.1207933437083197e-05,
      "loss": 1.793,
      "step": 8510
    },
    {
      "epoch": 1.4493493238070936,
      "grad_norm": 4.550126552581787,
      "learning_rate": 1.1202411322490667e-05,
      "loss": 1.7767,
      "step": 8520
    },
    {
      "epoch": 1.4510504380369142,
      "grad_norm": 4.194046974182129,
      "learning_rate": 1.1196883768100972e-05,
      "loss": 1.882,
      "step": 8530
    },
    {
      "epoch": 1.4527515522667347,
      "grad_norm": 4.48024320602417,
      "learning_rate": 1.1191350780626837e-05,
      "loss": 1.8334,
      "step": 8540
    },
    {
      "epoch": 1.4544526664965551,
      "grad_norm": 4.147039890289307,
      "learning_rate": 1.1185812366787594e-05,
      "loss": 1.8671,
      "step": 8550
    },
    {
      "epoch": 1.4561537807263758,
      "grad_norm": 4.3436455726623535,
      "learning_rate": 1.1180268533309163e-05,
      "loss": 1.8078,
      "step": 8560
    },
    {
      "epoch": 1.4578548949561962,
      "grad_norm": 3.813438892364502,
      "learning_rate": 1.1174719286924037e-05,
      "loss": 1.9086,
      "step": 8570
    },
    {
      "epoch": 1.459556009186017,
      "grad_norm": 4.152981758117676,
      "learning_rate": 1.11691646343713e-05,
      "loss": 1.8482,
      "step": 8580
    },
    {
      "epoch": 1.4612571234158374,
      "grad_norm": 4.7066731452941895,
      "learning_rate": 1.1163604582396585e-05,
      "loss": 1.7849,
      "step": 8590
    },
    {
      "epoch": 1.4629582376456578,
      "grad_norm": 4.566705703735352,
      "learning_rate": 1.1158039137752086e-05,
      "loss": 1.8481,
      "step": 8600
    },
    {
      "epoch": 1.4646593518754785,
      "grad_norm": 4.4183244705200195,
      "learning_rate": 1.1152468307196556e-05,
      "loss": 1.8281,
      "step": 8610
    },
    {
      "epoch": 1.466360466105299,
      "grad_norm": 3.972365617752075,
      "learning_rate": 1.1146892097495276e-05,
      "loss": 1.7882,
      "step": 8620
    },
    {
      "epoch": 1.4680615803351196,
      "grad_norm": 5.084153175354004,
      "learning_rate": 1.114131051542006e-05,
      "loss": 1.7916,
      "step": 8630
    },
    {
      "epoch": 1.46976269456494,
      "grad_norm": 4.221405029296875,
      "learning_rate": 1.1135723567749258e-05,
      "loss": 1.8082,
      "step": 8640
    },
    {
      "epoch": 1.4714638087947605,
      "grad_norm": 5.822017669677734,
      "learning_rate": 1.1130131261267726e-05,
      "loss": 1.8063,
      "step": 8650
    },
    {
      "epoch": 1.4731649230245811,
      "grad_norm": 4.446184158325195,
      "learning_rate": 1.112453360276683e-05,
      "loss": 1.8443,
      "step": 8660
    },
    {
      "epoch": 1.4748660372544016,
      "grad_norm": 4.830113887786865,
      "learning_rate": 1.1118930599044436e-05,
      "loss": 1.8257,
      "step": 8670
    },
    {
      "epoch": 1.4765671514842222,
      "grad_norm": 3.6857476234436035,
      "learning_rate": 1.11133222569049e-05,
      "loss": 1.7696,
      "step": 8680
    },
    {
      "epoch": 1.4782682657140427,
      "grad_norm": 4.067925930023193,
      "learning_rate": 1.1107708583159068e-05,
      "loss": 1.9156,
      "step": 8690
    },
    {
      "epoch": 1.4799693799438631,
      "grad_norm": 4.013896465301514,
      "learning_rate": 1.1102089584624247e-05,
      "loss": 1.9133,
      "step": 8700
    },
    {
      "epoch": 1.4816704941736838,
      "grad_norm": 4.576723098754883,
      "learning_rate": 1.1096465268124223e-05,
      "loss": 1.8425,
      "step": 8710
    },
    {
      "epoch": 1.4833716084035042,
      "grad_norm": 3.8686013221740723,
      "learning_rate": 1.1090835640489238e-05,
      "loss": 1.8831,
      "step": 8720
    },
    {
      "epoch": 1.485072722633325,
      "grad_norm": 5.166773319244385,
      "learning_rate": 1.1085200708555978e-05,
      "loss": 1.8464,
      "step": 8730
    },
    {
      "epoch": 1.4867738368631453,
      "grad_norm": 3.826002597808838,
      "learning_rate": 1.1079560479167575e-05,
      "loss": 1.7631,
      "step": 8740
    },
    {
      "epoch": 1.4884749510929658,
      "grad_norm": 3.7657535076141357,
      "learning_rate": 1.1073914959173593e-05,
      "loss": 1.7616,
      "step": 8750
    },
    {
      "epoch": 1.4901760653227865,
      "grad_norm": 3.9099855422973633,
      "learning_rate": 1.1068264155430026e-05,
      "loss": 1.8288,
      "step": 8760
    },
    {
      "epoch": 1.491877179552607,
      "grad_norm": 3.8278870582580566,
      "learning_rate": 1.1062608074799276e-05,
      "loss": 1.8276,
      "step": 8770
    },
    {
      "epoch": 1.4935782937824276,
      "grad_norm": 4.208233833312988,
      "learning_rate": 1.1056946724150159e-05,
      "loss": 1.9022,
      "step": 8780
    },
    {
      "epoch": 1.495279408012248,
      "grad_norm": 4.208041667938232,
      "learning_rate": 1.1051280110357887e-05,
      "loss": 1.8226,
      "step": 8790
    },
    {
      "epoch": 1.4969805222420685,
      "grad_norm": 3.7516579627990723,
      "learning_rate": 1.1045608240304072e-05,
      "loss": 1.7655,
      "step": 8800
    },
    {
      "epoch": 1.4986816364718891,
      "grad_norm": 4.082713603973389,
      "learning_rate": 1.1039931120876698e-05,
      "loss": 1.8776,
      "step": 8810
    },
    {
      "epoch": 1.5003827507017096,
      "grad_norm": 4.539659023284912,
      "learning_rate": 1.1034248758970132e-05,
      "loss": 1.7614,
      "step": 8820
    },
    {
      "epoch": 1.5020838649315302,
      "grad_norm": 4.518299579620361,
      "learning_rate": 1.1028561161485101e-05,
      "loss": 1.9139,
      "step": 8830
    },
    {
      "epoch": 1.5037849791613507,
      "grad_norm": 6.363071918487549,
      "learning_rate": 1.10228683353287e-05,
      "loss": 1.8792,
      "step": 8840
    },
    {
      "epoch": 1.5054860933911711,
      "grad_norm": 4.739774703979492,
      "learning_rate": 1.1017170287414366e-05,
      "loss": 1.8488,
      "step": 8850
    },
    {
      "epoch": 1.5071872076209918,
      "grad_norm": 4.243860244750977,
      "learning_rate": 1.1011467024661875e-05,
      "loss": 1.8552,
      "step": 8860
    },
    {
      "epoch": 1.5088883218508122,
      "grad_norm": 4.369020938873291,
      "learning_rate": 1.1005758553997344e-05,
      "loss": 1.8134,
      "step": 8870
    },
    {
      "epoch": 1.510589436080633,
      "grad_norm": 3.736544370651245,
      "learning_rate": 1.1000044882353212e-05,
      "loss": 1.8532,
      "step": 8880
    },
    {
      "epoch": 1.5122905503104533,
      "grad_norm": 4.25405216217041,
      "learning_rate": 1.099432601666823e-05,
      "loss": 1.8365,
      "step": 8890
    },
    {
      "epoch": 1.5139916645402738,
      "grad_norm": 4.405406951904297,
      "learning_rate": 1.098860196388746e-05,
      "loss": 1.7931,
      "step": 8900
    },
    {
      "epoch": 1.5156927787700945,
      "grad_norm": 4.439251899719238,
      "learning_rate": 1.0982872730962264e-05,
      "loss": 1.8344,
      "step": 8910
    },
    {
      "epoch": 1.517393892999915,
      "grad_norm": 4.5193376541137695,
      "learning_rate": 1.0977138324850292e-05,
      "loss": 1.8816,
      "step": 8920
    },
    {
      "epoch": 1.5190950072297356,
      "grad_norm": 4.007116317749023,
      "learning_rate": 1.0971398752515479e-05,
      "loss": 1.8915,
      "step": 8930
    },
    {
      "epoch": 1.520796121459556,
      "grad_norm": 4.432260513305664,
      "learning_rate": 1.0965654020928033e-05,
      "loss": 1.8104,
      "step": 8940
    },
    {
      "epoch": 1.5224972356893764,
      "grad_norm": 4.290425777435303,
      "learning_rate": 1.0959904137064426e-05,
      "loss": 1.7932,
      "step": 8950
    },
    {
      "epoch": 1.5241983499191971,
      "grad_norm": 3.5299010276794434,
      "learning_rate": 1.0954149107907391e-05,
      "loss": 1.9034,
      "step": 8960
    },
    {
      "epoch": 1.5258994641490176,
      "grad_norm": 4.343108654022217,
      "learning_rate": 1.0948388940445903e-05,
      "loss": 1.79,
      "step": 8970
    },
    {
      "epoch": 1.5276005783788382,
      "grad_norm": 3.799375057220459,
      "learning_rate": 1.0942623641675187e-05,
      "loss": 1.7986,
      "step": 8980
    },
    {
      "epoch": 1.5293016926086587,
      "grad_norm": 4.593927383422852,
      "learning_rate": 1.0936853218596688e-05,
      "loss": 1.8083,
      "step": 8990
    },
    {
      "epoch": 1.5310028068384791,
      "grad_norm": 3.587238073348999,
      "learning_rate": 1.0931077678218081e-05,
      "loss": 1.8337,
      "step": 9000
    },
    {
      "epoch": 1.5327039210682998,
      "grad_norm": 4.195778846740723,
      "learning_rate": 1.0925297027553255e-05,
      "loss": 1.7882,
      "step": 9010
    },
    {
      "epoch": 1.5344050352981202,
      "grad_norm": 4.813584804534912,
      "learning_rate": 1.0920090078462173e-05,
      "loss": 1.8808,
      "step": 9020
    },
    {
      "epoch": 1.536106149527941,
      "grad_norm": 5.184726715087891,
      "learning_rate": 1.0914299737599012e-05,
      "loss": 1.7869,
      "step": 9030
    },
    {
      "epoch": 1.5378072637577613,
      "grad_norm": 3.95269775390625,
      "learning_rate": 1.0908504306824966e-05,
      "loss": 1.7439,
      "step": 9040
    },
    {
      "epoch": 1.5395083779875818,
      "grad_norm": 3.7676851749420166,
      "learning_rate": 1.0902703793178087e-05,
      "loss": 1.8334,
      "step": 9050
    },
    {
      "epoch": 1.5412094922174024,
      "grad_norm": 4.78963565826416,
      "learning_rate": 1.0896898203702586e-05,
      "loss": 1.8415,
      "step": 9060
    },
    {
      "epoch": 1.5429106064472229,
      "grad_norm": 5.882055282592773,
      "learning_rate": 1.0891087545448844e-05,
      "loss": 1.7718,
      "step": 9070
    },
    {
      "epoch": 1.5446117206770436,
      "grad_norm": 4.677974224090576,
      "learning_rate": 1.0885271825473395e-05,
      "loss": 1.8712,
      "step": 9080
    },
    {
      "epoch": 1.546312834906864,
      "grad_norm": 4.491163730621338,
      "learning_rate": 1.0879451050838925e-05,
      "loss": 1.9664,
      "step": 9090
    },
    {
      "epoch": 1.5480139491366844,
      "grad_norm": 3.6754069328308105,
      "learning_rate": 1.0873625228614252e-05,
      "loss": 1.8512,
      "step": 9100
    },
    {
      "epoch": 1.549715063366505,
      "grad_norm": 4.288093090057373,
      "learning_rate": 1.0867794365874328e-05,
      "loss": 1.799,
      "step": 9110
    },
    {
      "epoch": 1.5514161775963256,
      "grad_norm": 5.016831874847412,
      "learning_rate": 1.0861958469700224e-05,
      "loss": 1.8035,
      "step": 9120
    },
    {
      "epoch": 1.5531172918261462,
      "grad_norm": 4.571620464324951,
      "learning_rate": 1.0856117547179126e-05,
      "loss": 1.8682,
      "step": 9130
    },
    {
      "epoch": 1.5548184060559667,
      "grad_norm": 4.138873100280762,
      "learning_rate": 1.0850271605404325e-05,
      "loss": 1.9147,
      "step": 9140
    },
    {
      "epoch": 1.556519520285787,
      "grad_norm": 3.770691156387329,
      "learning_rate": 1.0844420651475199e-05,
      "loss": 1.8129,
      "step": 9150
    },
    {
      "epoch": 1.5582206345156078,
      "grad_norm": 4.242394924163818,
      "learning_rate": 1.0838564692497224e-05,
      "loss": 1.8318,
      "step": 9160
    },
    {
      "epoch": 1.5599217487454282,
      "grad_norm": 4.3046956062316895,
      "learning_rate": 1.083270373558195e-05,
      "loss": 1.7672,
      "step": 9170
    },
    {
      "epoch": 1.5616228629752489,
      "grad_norm": 4.118201732635498,
      "learning_rate": 1.0826837787846995e-05,
      "loss": 1.8444,
      "step": 9180
    },
    {
      "epoch": 1.5633239772050693,
      "grad_norm": 4.4052252769470215,
      "learning_rate": 1.0820966856416037e-05,
      "loss": 1.8891,
      "step": 9190
    },
    {
      "epoch": 1.5650250914348898,
      "grad_norm": 4.588507175445557,
      "learning_rate": 1.081509094841881e-05,
      "loss": 1.8004,
      "step": 9200
    },
    {
      "epoch": 1.5667262056647104,
      "grad_norm": 4.094803810119629,
      "learning_rate": 1.0809210070991085e-05,
      "loss": 1.8421,
      "step": 9210
    },
    {
      "epoch": 1.5684273198945309,
      "grad_norm": 4.139965057373047,
      "learning_rate": 1.0803324231274682e-05,
      "loss": 1.6933,
      "step": 9220
    },
    {
      "epoch": 1.5701284341243515,
      "grad_norm": 3.955552816390991,
      "learning_rate": 1.0797433436417429e-05,
      "loss": 1.8102,
      "step": 9230
    },
    {
      "epoch": 1.571829548354172,
      "grad_norm": 3.6447393894195557,
      "learning_rate": 1.0791537693573182e-05,
      "loss": 1.6936,
      "step": 9240
    },
    {
      "epoch": 1.5735306625839924,
      "grad_norm": 4.680001258850098,
      "learning_rate": 1.078563700990181e-05,
      "loss": 1.8554,
      "step": 9250
    },
    {
      "epoch": 1.575231776813813,
      "grad_norm": 4.303575038909912,
      "learning_rate": 1.0779731392569172e-05,
      "loss": 1.8058,
      "step": 9260
    },
    {
      "epoch": 1.5769328910436335,
      "grad_norm": 4.5315752029418945,
      "learning_rate": 1.0773820848747122e-05,
      "loss": 1.7302,
      "step": 9270
    },
    {
      "epoch": 1.5786340052734542,
      "grad_norm": 4.147940158843994,
      "learning_rate": 1.0767905385613499e-05,
      "loss": 1.835,
      "step": 9280
    },
    {
      "epoch": 1.5803351195032747,
      "grad_norm": 4.4117279052734375,
      "learning_rate": 1.0761985010352117e-05,
      "loss": 1.8636,
      "step": 9290
    },
    {
      "epoch": 1.582036233733095,
      "grad_norm": 4.065666198730469,
      "learning_rate": 1.0756059730152753e-05,
      "loss": 1.7914,
      "step": 9300
    },
    {
      "epoch": 1.5837373479629158,
      "grad_norm": 4.648740291595459,
      "learning_rate": 1.0750129552211138e-05,
      "loss": 1.8272,
      "step": 9310
    },
    {
      "epoch": 1.5854384621927362,
      "grad_norm": 4.147993564605713,
      "learning_rate": 1.0744194483728956e-05,
      "loss": 1.7533,
      "step": 9320
    },
    {
      "epoch": 1.5871395764225569,
      "grad_norm": 3.6206252574920654,
      "learning_rate": 1.0738254531913827e-05,
      "loss": 1.6779,
      "step": 9330
    },
    {
      "epoch": 1.5888406906523773,
      "grad_norm": 4.248616695404053,
      "learning_rate": 1.0732309703979303e-05,
      "loss": 1.9106,
      "step": 9340
    },
    {
      "epoch": 1.5905418048821978,
      "grad_norm": 4.507028102874756,
      "learning_rate": 1.0726360007144855e-05,
      "loss": 1.7764,
      "step": 9350
    },
    {
      "epoch": 1.5922429191120182,
      "grad_norm": 7.24201774597168,
      "learning_rate": 1.0720405448635871e-05,
      "loss": 1.7711,
      "step": 9360
    },
    {
      "epoch": 1.5939440333418389,
      "grad_norm": 4.546375751495361,
      "learning_rate": 1.071444603568364e-05,
      "loss": 1.874,
      "step": 9370
    },
    {
      "epoch": 1.5956451475716595,
      "grad_norm": 4.4695024490356445,
      "learning_rate": 1.0708481775525347e-05,
      "loss": 1.8446,
      "step": 9380
    },
    {
      "epoch": 1.59734626180148,
      "grad_norm": 4.036654472351074,
      "learning_rate": 1.0702512675404066e-05,
      "loss": 1.8142,
      "step": 9390
    },
    {
      "epoch": 1.5990473760313004,
      "grad_norm": 3.3620481491088867,
      "learning_rate": 1.0696538742568742e-05,
      "loss": 1.8514,
      "step": 9400
    },
    {
      "epoch": 1.6007484902611209,
      "grad_norm": 4.128815174102783,
      "learning_rate": 1.0690559984274196e-05,
      "loss": 1.8632,
      "step": 9410
    },
    {
      "epoch": 1.6024496044909415,
      "grad_norm": 4.229797840118408,
      "learning_rate": 1.0684576407781106e-05,
      "loss": 1.7916,
      "step": 9420
    },
    {
      "epoch": 1.6041507187207622,
      "grad_norm": 4.053248405456543,
      "learning_rate": 1.0678588020356e-05,
      "loss": 1.8409,
      "step": 9430
    },
    {
      "epoch": 1.6058518329505826,
      "grad_norm": 3.9605884552001953,
      "learning_rate": 1.0672594829271251e-05,
      "loss": 1.7641,
      "step": 9440
    },
    {
      "epoch": 1.607552947180403,
      "grad_norm": 4.531416893005371,
      "learning_rate": 1.0666596841805066e-05,
      "loss": 1.8203,
      "step": 9450
    },
    {
      "epoch": 1.6092540614102235,
      "grad_norm": 4.545661926269531,
      "learning_rate": 1.0660594065241477e-05,
      "loss": 1.8606,
      "step": 9460
    },
    {
      "epoch": 1.6109551756400442,
      "grad_norm": 4.635908603668213,
      "learning_rate": 1.0654586506870324e-05,
      "loss": 1.8076,
      "step": 9470
    },
    {
      "epoch": 1.6126562898698649,
      "grad_norm": 4.799504280090332,
      "learning_rate": 1.0648574173987266e-05,
      "loss": 1.7702,
      "step": 9480
    },
    {
      "epoch": 1.6143574040996853,
      "grad_norm": 3.9335310459136963,
      "learning_rate": 1.0642557073893754e-05,
      "loss": 1.7715,
      "step": 9490
    },
    {
      "epoch": 1.6160585183295058,
      "grad_norm": 4.012069225311279,
      "learning_rate": 1.0636535213897029e-05,
      "loss": 1.7753,
      "step": 9500
    },
    {
      "epoch": 1.6177596325593262,
      "grad_norm": 4.231178283691406,
      "learning_rate": 1.0630508601310112e-05,
      "loss": 1.8751,
      "step": 9510
    },
    {
      "epoch": 1.6194607467891469,
      "grad_norm": 4.730849266052246,
      "learning_rate": 1.0624477243451794e-05,
      "loss": 1.8002,
      "step": 9520
    },
    {
      "epoch": 1.6211618610189675,
      "grad_norm": 4.264655590057373,
      "learning_rate": 1.0618441147646635e-05,
      "loss": 1.8493,
      "step": 9530
    },
    {
      "epoch": 1.622862975248788,
      "grad_norm": 4.410914897918701,
      "learning_rate": 1.0612400321224943e-05,
      "loss": 1.8096,
      "step": 9540
    },
    {
      "epoch": 1.6245640894786084,
      "grad_norm": 4.016700744628906,
      "learning_rate": 1.0606354771522772e-05,
      "loss": 1.7711,
      "step": 9550
    },
    {
      "epoch": 1.6262652037084289,
      "grad_norm": 4.219663143157959,
      "learning_rate": 1.0600304505881915e-05,
      "loss": 1.7203,
      "step": 9560
    },
    {
      "epoch": 1.6279663179382495,
      "grad_norm": 4.0092034339904785,
      "learning_rate": 1.0594249531649886e-05,
      "loss": 1.7203,
      "step": 9570
    },
    {
      "epoch": 1.6296674321680702,
      "grad_norm": 4.708493709564209,
      "learning_rate": 1.0588189856179922e-05,
      "loss": 1.8393,
      "step": 9580
    },
    {
      "epoch": 1.6313685463978906,
      "grad_norm": 4.2801923751831055,
      "learning_rate": 1.0582125486830968e-05,
      "loss": 1.7938,
      "step": 9590
    },
    {
      "epoch": 1.633069660627711,
      "grad_norm": 3.3154683113098145,
      "learning_rate": 1.0576056430967673e-05,
      "loss": 1.7587,
      "step": 9600
    },
    {
      "epoch": 1.6347707748575315,
      "grad_norm": 4.226798057556152,
      "learning_rate": 1.056998269596037e-05,
      "loss": 1.7655,
      "step": 9610
    },
    {
      "epoch": 1.6364718890873522,
      "grad_norm": 3.5839290618896484,
      "learning_rate": 1.0563904289185081e-05,
      "loss": 1.8516,
      "step": 9620
    },
    {
      "epoch": 1.6381730033171729,
      "grad_norm": 4.124845027923584,
      "learning_rate": 1.0557821218023497e-05,
      "loss": 1.8054,
      "step": 9630
    },
    {
      "epoch": 1.6398741175469933,
      "grad_norm": 4.484249114990234,
      "learning_rate": 1.0551733489862973e-05,
      "loss": 1.8195,
      "step": 9640
    },
    {
      "epoch": 1.6415752317768137,
      "grad_norm": 4.2265625,
      "learning_rate": 1.0545641112096527e-05,
      "loss": 1.8027,
      "step": 9650
    },
    {
      "epoch": 1.6432763460066342,
      "grad_norm": 3.7763261795043945,
      "learning_rate": 1.0539544092122815e-05,
      "loss": 1.7931,
      "step": 9660
    },
    {
      "epoch": 1.6449774602364549,
      "grad_norm": 4.304544925689697,
      "learning_rate": 1.0533442437346133e-05,
      "loss": 1.8094,
      "step": 9670
    },
    {
      "epoch": 1.6466785744662755,
      "grad_norm": 3.8868472576141357,
      "learning_rate": 1.0527336155176407e-05,
      "loss": 1.813,
      "step": 9680
    },
    {
      "epoch": 1.648379688696096,
      "grad_norm": 4.530086517333984,
      "learning_rate": 1.0521225253029186e-05,
      "loss": 1.732,
      "step": 9690
    },
    {
      "epoch": 1.6500808029259164,
      "grad_norm": 3.6333813667297363,
      "learning_rate": 1.0515109738325618e-05,
      "loss": 1.7679,
      "step": 9700
    },
    {
      "epoch": 1.6517819171557369,
      "grad_norm": 3.765293836593628,
      "learning_rate": 1.0508989618492466e-05,
      "loss": 1.8888,
      "step": 9710
    },
    {
      "epoch": 1.6534830313855575,
      "grad_norm": 3.8620569705963135,
      "learning_rate": 1.0502864900962075e-05,
      "loss": 1.7507,
      "step": 9720
    },
    {
      "epoch": 1.6551841456153782,
      "grad_norm": 3.9069511890411377,
      "learning_rate": 1.0496735593172384e-05,
      "loss": 1.7659,
      "step": 9730
    },
    {
      "epoch": 1.6568852598451986,
      "grad_norm": 4.23579216003418,
      "learning_rate": 1.0490601702566898e-05,
      "loss": 1.8153,
      "step": 9740
    },
    {
      "epoch": 1.658586374075019,
      "grad_norm": 4.105676651000977,
      "learning_rate": 1.0484463236594686e-05,
      "loss": 1.8104,
      "step": 9750
    },
    {
      "epoch": 1.6602874883048395,
      "grad_norm": 4.492506504058838,
      "learning_rate": 1.0478320202710382e-05,
      "loss": 1.8233,
      "step": 9760
    },
    {
      "epoch": 1.6619886025346602,
      "grad_norm": 4.099576473236084,
      "learning_rate": 1.047217260837416e-05,
      "loss": 1.8074,
      "step": 9770
    },
    {
      "epoch": 1.6636897167644809,
      "grad_norm": 3.619990825653076,
      "learning_rate": 1.0466020461051735e-05,
      "loss": 1.8891,
      "step": 9780
    },
    {
      "epoch": 1.6653908309943013,
      "grad_norm": 4.893577575683594,
      "learning_rate": 1.0459863768214355e-05,
      "loss": 1.8991,
      "step": 9790
    },
    {
      "epoch": 1.6670919452241217,
      "grad_norm": 4.113924980163574,
      "learning_rate": 1.0453702537338779e-05,
      "loss": 1.8974,
      "step": 9800
    },
    {
      "epoch": 1.6687930594539422,
      "grad_norm": 3.6121366024017334,
      "learning_rate": 1.0447536775907285e-05,
      "loss": 1.8565,
      "step": 9810
    },
    {
      "epoch": 1.6704941736837629,
      "grad_norm": 4.138027667999268,
      "learning_rate": 1.044136649140765e-05,
      "loss": 1.8812,
      "step": 9820
    },
    {
      "epoch": 1.6721952879135835,
      "grad_norm": 4.8973388671875,
      "learning_rate": 1.0435191691333147e-05,
      "loss": 1.8696,
      "step": 9830
    },
    {
      "epoch": 1.673896402143404,
      "grad_norm": 3.971330404281616,
      "learning_rate": 1.0429012383182522e-05,
      "loss": 1.8755,
      "step": 9840
    },
    {
      "epoch": 1.6755975163732244,
      "grad_norm": 3.5776214599609375,
      "learning_rate": 1.0422828574460013e-05,
      "loss": 1.7165,
      "step": 9850
    },
    {
      "epoch": 1.6772986306030448,
      "grad_norm": 4.539066791534424,
      "learning_rate": 1.0416640272675307e-05,
      "loss": 1.7711,
      "step": 9860
    },
    {
      "epoch": 1.6789997448328655,
      "grad_norm": 3.9174132347106934,
      "learning_rate": 1.041044748534356e-05,
      "loss": 1.7762,
      "step": 9870
    },
    {
      "epoch": 1.6807008590626862,
      "grad_norm": 4.058426380157471,
      "learning_rate": 1.0404250219985366e-05,
      "loss": 1.717,
      "step": 9880
    },
    {
      "epoch": 1.6824019732925066,
      "grad_norm": 3.7579991817474365,
      "learning_rate": 1.039804848412676e-05,
      "loss": 1.7561,
      "step": 9890
    },
    {
      "epoch": 1.684103087522327,
      "grad_norm": 3.993664503097534,
      "learning_rate": 1.0391842285299212e-05,
      "loss": 1.8824,
      "step": 9900
    },
    {
      "epoch": 1.6858042017521475,
      "grad_norm": 4.473487854003906,
      "learning_rate": 1.0385631631039606e-05,
      "loss": 1.8299,
      "step": 9910
    },
    {
      "epoch": 1.6875053159819682,
      "grad_norm": 4.511467933654785,
      "learning_rate": 1.0379416528890233e-05,
      "loss": 1.7641,
      "step": 9920
    },
    {
      "epoch": 1.6892064302117888,
      "grad_norm": 3.6925597190856934,
      "learning_rate": 1.0373196986398793e-05,
      "loss": 1.7787,
      "step": 9930
    },
    {
      "epoch": 1.6909075444416093,
      "grad_norm": 3.6119349002838135,
      "learning_rate": 1.0366973011118377e-05,
      "loss": 1.8198,
      "step": 9940
    },
    {
      "epoch": 1.6926086586714297,
      "grad_norm": 4.093749046325684,
      "learning_rate": 1.0360744610607455e-05,
      "loss": 1.7895,
      "step": 9950
    },
    {
      "epoch": 1.6943097729012502,
      "grad_norm": 3.8011345863342285,
      "learning_rate": 1.0354511792429879e-05,
      "loss": 1.814,
      "step": 9960
    },
    {
      "epoch": 1.6960108871310708,
      "grad_norm": 4.387421131134033,
      "learning_rate": 1.0348274564154857e-05,
      "loss": 1.7612,
      "step": 9970
    },
    {
      "epoch": 1.6977120013608915,
      "grad_norm": 4.6559576988220215,
      "learning_rate": 1.0342032933356953e-05,
      "loss": 1.9255,
      "step": 9980
    },
    {
      "epoch": 1.699413115590712,
      "grad_norm": 3.9810171127319336,
      "learning_rate": 1.0335786907616088e-05,
      "loss": 1.8769,
      "step": 9990
    },
    {
      "epoch": 1.7011142298205324,
      "grad_norm": 4.540745735168457,
      "learning_rate": 1.032953649451751e-05,
      "loss": 1.7372,
      "step": 10000
    },
    {
      "epoch": 1.7028153440503528,
      "grad_norm": 4.483917236328125,
      "learning_rate": 1.03232817016518e-05,
      "loss": 1.8221,
      "step": 10010
    },
    {
      "epoch": 1.7045164582801735,
      "grad_norm": 4.2136616706848145,
      "learning_rate": 1.0317022536614854e-05,
      "loss": 1.7018,
      "step": 10020
    },
    {
      "epoch": 1.7062175725099942,
      "grad_norm": 4.191181659698486,
      "learning_rate": 1.031075900700788e-05,
      "loss": 1.7263,
      "step": 10030
    },
    {
      "epoch": 1.7079186867398146,
      "grad_norm": 4.057601451873779,
      "learning_rate": 1.0304491120437385e-05,
      "loss": 1.8144,
      "step": 10040
    },
    {
      "epoch": 1.709619800969635,
      "grad_norm": 3.8585596084594727,
      "learning_rate": 1.0298218884515172e-05,
      "loss": 1.828,
      "step": 10050
    },
    {
      "epoch": 1.7113209151994555,
      "grad_norm": 3.869547128677368,
      "learning_rate": 1.0291942306858321e-05,
      "loss": 1.8319,
      "step": 10060
    },
    {
      "epoch": 1.7130220294292762,
      "grad_norm": 4.130283355712891,
      "learning_rate": 1.0285661395089184e-05,
      "loss": 1.7499,
      "step": 10070
    },
    {
      "epoch": 1.7147231436590968,
      "grad_norm": 5.252286434173584,
      "learning_rate": 1.027937615683538e-05,
      "loss": 1.847,
      "step": 10080
    },
    {
      "epoch": 1.7164242578889173,
      "grad_norm": 4.278568744659424,
      "learning_rate": 1.027308659972978e-05,
      "loss": 1.8668,
      "step": 10090
    },
    {
      "epoch": 1.7181253721187377,
      "grad_norm": 4.391490459442139,
      "learning_rate": 1.02667927314105e-05,
      "loss": 1.7655,
      "step": 10100
    },
    {
      "epoch": 1.7198264863485582,
      "grad_norm": 3.2366139888763428,
      "learning_rate": 1.0260494559520891e-05,
      "loss": 1.8268,
      "step": 10110
    },
    {
      "epoch": 1.7215276005783788,
      "grad_norm": 4.020900726318359,
      "learning_rate": 1.0254192091709532e-05,
      "loss": 1.7843,
      "step": 10120
    },
    {
      "epoch": 1.7232287148081995,
      "grad_norm": 4.717282295227051,
      "learning_rate": 1.0247885335630218e-05,
      "loss": 1.9005,
      "step": 10130
    },
    {
      "epoch": 1.72492982903802,
      "grad_norm": 4.262220859527588,
      "learning_rate": 1.0241574298941955e-05,
      "loss": 1.7488,
      "step": 10140
    },
    {
      "epoch": 1.7266309432678404,
      "grad_norm": 3.7055883407592773,
      "learning_rate": 1.023525898930894e-05,
      "loss": 1.8134,
      "step": 10150
    },
    {
      "epoch": 1.7283320574976608,
      "grad_norm": 4.13065767288208,
      "learning_rate": 1.0228939414400564e-05,
      "loss": 1.8133,
      "step": 10160
    },
    {
      "epoch": 1.7300331717274815,
      "grad_norm": 3.898186206817627,
      "learning_rate": 1.0222615581891401e-05,
      "loss": 1.6895,
      "step": 10170
    },
    {
      "epoch": 1.7317342859573022,
      "grad_norm": 3.9497716426849365,
      "learning_rate": 1.0216287499461187e-05,
      "loss": 1.801,
      "step": 10180
    },
    {
      "epoch": 1.7334354001871226,
      "grad_norm": 4.006579399108887,
      "learning_rate": 1.0209955174794828e-05,
      "loss": 1.8844,
      "step": 10190
    },
    {
      "epoch": 1.735136514416943,
      "grad_norm": 4.037294387817383,
      "learning_rate": 1.0203618615582373e-05,
      "loss": 1.8429,
      "step": 10200
    },
    {
      "epoch": 1.7368376286467635,
      "grad_norm": 4.3926100730896,
      "learning_rate": 1.0197277829519023e-05,
      "loss": 1.8029,
      "step": 10210
    },
    {
      "epoch": 1.7385387428765842,
      "grad_norm": 4.085505485534668,
      "learning_rate": 1.0190932824305105e-05,
      "loss": 1.8537,
      "step": 10220
    },
    {
      "epoch": 1.7402398571064048,
      "grad_norm": 4.870039463043213,
      "learning_rate": 1.018458360764607e-05,
      "loss": 1.8234,
      "step": 10230
    },
    {
      "epoch": 1.7419409713362253,
      "grad_norm": 3.6122336387634277,
      "learning_rate": 1.0178230187252492e-05,
      "loss": 1.7095,
      "step": 10240
    },
    {
      "epoch": 1.7436420855660457,
      "grad_norm": 4.699728488922119,
      "learning_rate": 1.017187257084004e-05,
      "loss": 1.8254,
      "step": 10250
    },
    {
      "epoch": 1.7453431997958662,
      "grad_norm": 4.258387565612793,
      "learning_rate": 1.0165510766129481e-05,
      "loss": 1.795,
      "step": 10260
    },
    {
      "epoch": 1.7470443140256868,
      "grad_norm": 3.91975736618042,
      "learning_rate": 1.015914478084667e-05,
      "loss": 1.8109,
      "step": 10270
    },
    {
      "epoch": 1.7487454282555075,
      "grad_norm": 3.489849328994751,
      "learning_rate": 1.015277462272254e-05,
      "loss": 1.8538,
      "step": 10280
    },
    {
      "epoch": 1.750446542485328,
      "grad_norm": 3.8970468044281006,
      "learning_rate": 1.014640029949309e-05,
      "loss": 1.7765,
      "step": 10290
    },
    {
      "epoch": 1.7521476567151484,
      "grad_norm": 3.8178770542144775,
      "learning_rate": 1.0140021818899377e-05,
      "loss": 1.7456,
      "step": 10300
    },
    {
      "epoch": 1.7538487709449688,
      "grad_norm": 4.668042182922363,
      "learning_rate": 1.0133639188687506e-05,
      "loss": 1.8154,
      "step": 10310
    },
    {
      "epoch": 1.7555498851747895,
      "grad_norm": 3.638029098510742,
      "learning_rate": 1.0127252416608624e-05,
      "loss": 1.857,
      "step": 10320
    },
    {
      "epoch": 1.7572509994046102,
      "grad_norm": 4.546308994293213,
      "learning_rate": 1.0120861510418905e-05,
      "loss": 1.8085,
      "step": 10330
    },
    {
      "epoch": 1.7589521136344306,
      "grad_norm": 4.162155628204346,
      "learning_rate": 1.0114466477879548e-05,
      "loss": 1.8464,
      "step": 10340
    },
    {
      "epoch": 1.760653227864251,
      "grad_norm": 4.644067764282227,
      "learning_rate": 1.0108067326756755e-05,
      "loss": 1.8394,
      "step": 10350
    },
    {
      "epoch": 1.7623543420940715,
      "grad_norm": 4.256865501403809,
      "learning_rate": 1.0101664064821739e-05,
      "loss": 1.9286,
      "step": 10360
    },
    {
      "epoch": 1.7640554563238922,
      "grad_norm": 4.50824499130249,
      "learning_rate": 1.0095256699850698e-05,
      "loss": 1.7065,
      "step": 10370
    },
    {
      "epoch": 1.7657565705537128,
      "grad_norm": 3.926837205886841,
      "learning_rate": 1.0088845239624815e-05,
      "loss": 1.7754,
      "step": 10380
    },
    {
      "epoch": 1.7674576847835333,
      "grad_norm": 4.212017059326172,
      "learning_rate": 1.0082429691930252e-05,
      "loss": 1.7532,
      "step": 10390
    },
    {
      "epoch": 1.7691587990133537,
      "grad_norm": 4.2193498611450195,
      "learning_rate": 1.0076010064558126e-05,
      "loss": 1.7152,
      "step": 10400
    },
    {
      "epoch": 1.7708599132431742,
      "grad_norm": 4.437827110290527,
      "learning_rate": 1.006958636530451e-05,
      "loss": 1.7723,
      "step": 10410
    },
    {
      "epoch": 1.7725610274729948,
      "grad_norm": 4.158639907836914,
      "learning_rate": 1.006315860197043e-05,
      "loss": 1.8109,
      "step": 10420
    },
    {
      "epoch": 1.7742621417028155,
      "grad_norm": 3.403130054473877,
      "learning_rate": 1.0056726782361835e-05,
      "loss": 1.7922,
      "step": 10430
    },
    {
      "epoch": 1.775963255932636,
      "grad_norm": 3.8050906658172607,
      "learning_rate": 1.0050290914289612e-05,
      "loss": 1.8183,
      "step": 10440
    },
    {
      "epoch": 1.7776643701624564,
      "grad_norm": 3.5411856174468994,
      "learning_rate": 1.004385100556956e-05,
      "loss": 1.7803,
      "step": 10450
    },
    {
      "epoch": 1.7793654843922768,
      "grad_norm": 4.4496307373046875,
      "learning_rate": 1.0037407064022376e-05,
      "loss": 1.7554,
      "step": 10460
    },
    {
      "epoch": 1.7810665986220975,
      "grad_norm": 4.679849147796631,
      "learning_rate": 1.0030959097473672e-05,
      "loss": 1.7382,
      "step": 10470
    },
    {
      "epoch": 1.7827677128519182,
      "grad_norm": 5.210113048553467,
      "learning_rate": 1.0024507113753934e-05,
      "loss": 1.8625,
      "step": 10480
    },
    {
      "epoch": 1.7844688270817386,
      "grad_norm": 4.474411964416504,
      "learning_rate": 1.001805112069853e-05,
      "loss": 1.8229,
      "step": 10490
    },
    {
      "epoch": 1.786169941311559,
      "grad_norm": 4.553360462188721,
      "learning_rate": 1.0011591126147702e-05,
      "loss": 1.8339,
      "step": 10500
    },
    {
      "epoch": 1.7878710555413795,
      "grad_norm": 4.378016948699951,
      "learning_rate": 1.0005127137946546e-05,
      "loss": 1.8492,
      "step": 10510
    },
    {
      "epoch": 1.7895721697712001,
      "grad_norm": 3.943833827972412,
      "learning_rate": 9.998659163945008e-06,
      "loss": 1.8541,
      "step": 10520
    },
    {
      "epoch": 1.7912732840010208,
      "grad_norm": 4.326706409454346,
      "learning_rate": 9.992187211997876e-06,
      "loss": 1.8652,
      "step": 10530
    },
    {
      "epoch": 1.7929743982308413,
      "grad_norm": 4.4171671867370605,
      "learning_rate": 9.985711289964774e-06,
      "loss": 1.8608,
      "step": 10540
    },
    {
      "epoch": 1.7946755124606617,
      "grad_norm": 4.190195083618164,
      "learning_rate": 9.979231405710136e-06,
      "loss": 1.7901,
      "step": 10550
    },
    {
      "epoch": 1.7963766266904821,
      "grad_norm": 4.556994438171387,
      "learning_rate": 9.972747567103217e-06,
      "loss": 1.7475,
      "step": 10560
    },
    {
      "epoch": 1.7980777409203028,
      "grad_norm": 4.397325038909912,
      "learning_rate": 9.96625978201807e-06,
      "loss": 1.7118,
      "step": 10570
    },
    {
      "epoch": 1.7997788551501235,
      "grad_norm": 3.940673828125,
      "learning_rate": 9.959768058333541e-06,
      "loss": 1.8119,
      "step": 10580
    },
    {
      "epoch": 1.801479969379944,
      "grad_norm": 5.351619720458984,
      "learning_rate": 9.953272403933263e-06,
      "loss": 1.8465,
      "step": 10590
    },
    {
      "epoch": 1.8031810836097644,
      "grad_norm": 4.466691970825195,
      "learning_rate": 9.946772826705638e-06,
      "loss": 1.7923,
      "step": 10600
    },
    {
      "epoch": 1.8048821978395848,
      "grad_norm": 4.102950096130371,
      "learning_rate": 9.94026933454383e-06,
      "loss": 1.8181,
      "step": 10610
    },
    {
      "epoch": 1.8065833120694055,
      "grad_norm": 3.4681036472320557,
      "learning_rate": 9.933761935345766e-06,
      "loss": 1.785,
      "step": 10620
    },
    {
      "epoch": 1.8082844262992261,
      "grad_norm": 4.120108604431152,
      "learning_rate": 9.927250637014107e-06,
      "loss": 1.7911,
      "step": 10630
    },
    {
      "epoch": 1.8099855405290466,
      "grad_norm": 4.320941925048828,
      "learning_rate": 9.920735447456261e-06,
      "loss": 1.8297,
      "step": 10640
    },
    {
      "epoch": 1.811686654758867,
      "grad_norm": 4.682049751281738,
      "learning_rate": 9.914216374584352e-06,
      "loss": 1.7805,
      "step": 10650
    },
    {
      "epoch": 1.8133877689886875,
      "grad_norm": 4.224926471710205,
      "learning_rate": 9.907693426315219e-06,
      "loss": 1.7751,
      "step": 10660
    },
    {
      "epoch": 1.8150888832185081,
      "grad_norm": 4.522299289703369,
      "learning_rate": 9.901166610570417e-06,
      "loss": 1.7741,
      "step": 10670
    },
    {
      "epoch": 1.8167899974483288,
      "grad_norm": 4.843745708465576,
      "learning_rate": 9.89463593527619e-06,
      "loss": 1.837,
      "step": 10680
    },
    {
      "epoch": 1.8184911116781493,
      "grad_norm": 4.759020805358887,
      "learning_rate": 9.888101408363473e-06,
      "loss": 1.7156,
      "step": 10690
    },
    {
      "epoch": 1.8201922259079697,
      "grad_norm": 4.42807674407959,
      "learning_rate": 9.881563037767872e-06,
      "loss": 1.821,
      "step": 10700
    },
    {
      "epoch": 1.8218933401377901,
      "grad_norm": 4.293888568878174,
      "learning_rate": 9.875020831429671e-06,
      "loss": 1.7818,
      "step": 10710
    },
    {
      "epoch": 1.8235944543676108,
      "grad_norm": 3.7048070430755615,
      "learning_rate": 9.868474797293803e-06,
      "loss": 1.8526,
      "step": 10720
    },
    {
      "epoch": 1.8252955685974315,
      "grad_norm": 3.907238006591797,
      "learning_rate": 9.861924943309855e-06,
      "loss": 1.7242,
      "step": 10730
    },
    {
      "epoch": 1.826996682827252,
      "grad_norm": 4.688352584838867,
      "learning_rate": 9.855371277432052e-06,
      "loss": 1.7701,
      "step": 10740
    },
    {
      "epoch": 1.8286977970570724,
      "grad_norm": 4.016199588775635,
      "learning_rate": 9.848813807619242e-06,
      "loss": 1.8497,
      "step": 10750
    },
    {
      "epoch": 1.8303989112868928,
      "grad_norm": 4.666214466094971,
      "learning_rate": 9.842252541834904e-06,
      "loss": 1.8062,
      "step": 10760
    },
    {
      "epoch": 1.8321000255167135,
      "grad_norm": 4.1961188316345215,
      "learning_rate": 9.835687488047119e-06,
      "loss": 1.8466,
      "step": 10770
    },
    {
      "epoch": 1.8338011397465341,
      "grad_norm": 4.104489326477051,
      "learning_rate": 9.829118654228565e-06,
      "loss": 1.8977,
      "step": 10780
    },
    {
      "epoch": 1.8355022539763546,
      "grad_norm": 3.4118106365203857,
      "learning_rate": 9.822546048356524e-06,
      "loss": 1.951,
      "step": 10790
    },
    {
      "epoch": 1.837203368206175,
      "grad_norm": 4.667579650878906,
      "learning_rate": 9.815969678412844e-06,
      "loss": 1.7684,
      "step": 10800
    },
    {
      "epoch": 1.8389044824359955,
      "grad_norm": 3.9106857776641846,
      "learning_rate": 9.80938955238395e-06,
      "loss": 1.8193,
      "step": 10810
    },
    {
      "epoch": 1.8406055966658161,
      "grad_norm": 4.125003337860107,
      "learning_rate": 9.802805678260835e-06,
      "loss": 1.8235,
      "step": 10820
    },
    {
      "epoch": 1.8423067108956368,
      "grad_norm": 4.059302806854248,
      "learning_rate": 9.79621806403903e-06,
      "loss": 1.7107,
      "step": 10830
    },
    {
      "epoch": 1.8440078251254572,
      "grad_norm": 3.3380849361419678,
      "learning_rate": 9.789626717718622e-06,
      "loss": 1.832,
      "step": 10840
    },
    {
      "epoch": 1.8457089393552777,
      "grad_norm": 3.950030565261841,
      "learning_rate": 9.783031647304217e-06,
      "loss": 1.811,
      "step": 10850
    },
    {
      "epoch": 1.8474100535850981,
      "grad_norm": 4.345433235168457,
      "learning_rate": 9.776432860804957e-06,
      "loss": 1.7453,
      "step": 10860
    },
    {
      "epoch": 1.8491111678149188,
      "grad_norm": 4.332964897155762,
      "learning_rate": 9.769830366234486e-06,
      "loss": 1.7936,
      "step": 10870
    },
    {
      "epoch": 1.8508122820447395,
      "grad_norm": 4.119027614593506,
      "learning_rate": 9.763224171610958e-06,
      "loss": 1.7703,
      "step": 10880
    },
    {
      "epoch": 1.85251339627456,
      "grad_norm": 3.9850196838378906,
      "learning_rate": 9.756614284957016e-06,
      "loss": 1.7415,
      "step": 10890
    },
    {
      "epoch": 1.8542145105043804,
      "grad_norm": 4.185554027557373,
      "learning_rate": 9.75000071429979e-06,
      "loss": 1.8353,
      "step": 10900
    },
    {
      "epoch": 1.8559156247342008,
      "grad_norm": 4.098738193511963,
      "learning_rate": 9.74338346767088e-06,
      "loss": 1.8058,
      "step": 10910
    },
    {
      "epoch": 1.8576167389640215,
      "grad_norm": 5.299314975738525,
      "learning_rate": 9.736762553106356e-06,
      "loss": 1.7866,
      "step": 10920
    },
    {
      "epoch": 1.8593178531938421,
      "grad_norm": 4.44684362411499,
      "learning_rate": 9.730137978646737e-06,
      "loss": 1.8705,
      "step": 10930
    },
    {
      "epoch": 1.8610189674236626,
      "grad_norm": 4.626757621765137,
      "learning_rate": 9.723509752336989e-06,
      "loss": 1.8671,
      "step": 10940
    },
    {
      "epoch": 1.862720081653483,
      "grad_norm": 3.2464823722839355,
      "learning_rate": 9.716877882226508e-06,
      "loss": 1.7845,
      "step": 10950
    },
    {
      "epoch": 1.8644211958833035,
      "grad_norm": 4.471595764160156,
      "learning_rate": 9.710242376369126e-06,
      "loss": 1.7647,
      "step": 10960
    },
    {
      "epoch": 1.8661223101131241,
      "grad_norm": 4.0545878410339355,
      "learning_rate": 9.703603242823082e-06,
      "loss": 1.761,
      "step": 10970
    },
    {
      "epoch": 1.8678234243429448,
      "grad_norm": 4.138380527496338,
      "learning_rate": 9.696960489651018e-06,
      "loss": 1.8144,
      "step": 10980
    },
    {
      "epoch": 1.8695245385727652,
      "grad_norm": 4.559357643127441,
      "learning_rate": 9.690314124919985e-06,
      "loss": 1.7863,
      "step": 10990
    },
    {
      "epoch": 1.8712256528025857,
      "grad_norm": 3.6744070053100586,
      "learning_rate": 9.6836641567014e-06,
      "loss": 1.8297,
      "step": 11000
    },
    {
      "epoch": 1.8729267670324061,
      "grad_norm": 4.699102878570557,
      "learning_rate": 9.677010593071073e-06,
      "loss": 1.7677,
      "step": 11010
    },
    {
      "epoch": 1.8746278812622268,
      "grad_norm": 3.9466359615325928,
      "learning_rate": 9.670353442109175e-06,
      "loss": 1.7409,
      "step": 11020
    },
    {
      "epoch": 1.8763289954920472,
      "grad_norm": 4.065673351287842,
      "learning_rate": 9.663692711900232e-06,
      "loss": 1.772,
      "step": 11030
    },
    {
      "epoch": 1.878030109721868,
      "grad_norm": 3.4951467514038086,
      "learning_rate": 9.657028410533114e-06,
      "loss": 1.8675,
      "step": 11040
    },
    {
      "epoch": 1.8797312239516883,
      "grad_norm": 4.127050876617432,
      "learning_rate": 9.650360546101037e-06,
      "loss": 1.7917,
      "step": 11050
    },
    {
      "epoch": 1.8814323381815088,
      "grad_norm": 3.586827278137207,
      "learning_rate": 9.643689126701533e-06,
      "loss": 1.7772,
      "step": 11060
    },
    {
      "epoch": 1.8831334524113295,
      "grad_norm": 4.5430073738098145,
      "learning_rate": 9.63701416043646e-06,
      "loss": 1.8817,
      "step": 11070
    },
    {
      "epoch": 1.88483456664115,
      "grad_norm": 4.422996520996094,
      "learning_rate": 9.630335655411983e-06,
      "loss": 1.7157,
      "step": 11080
    },
    {
      "epoch": 1.8865356808709706,
      "grad_norm": 4.226709365844727,
      "learning_rate": 9.623653619738555e-06,
      "loss": 1.7476,
      "step": 11090
    },
    {
      "epoch": 1.888236795100791,
      "grad_norm": 3.803516387939453,
      "learning_rate": 9.616968061530925e-06,
      "loss": 1.7751,
      "step": 11100
    },
    {
      "epoch": 1.8899379093306115,
      "grad_norm": 4.143571376800537,
      "learning_rate": 9.61027898890812e-06,
      "loss": 1.7535,
      "step": 11110
    },
    {
      "epoch": 1.8916390235604321,
      "grad_norm": 4.062926769256592,
      "learning_rate": 9.603586409993429e-06,
      "loss": 1.7253,
      "step": 11120
    },
    {
      "epoch": 1.8933401377902526,
      "grad_norm": 4.166606426239014,
      "learning_rate": 9.59689033291441e-06,
      "loss": 1.7783,
      "step": 11130
    },
    {
      "epoch": 1.8950412520200732,
      "grad_norm": 4.350265026092529,
      "learning_rate": 9.590190765802856e-06,
      "loss": 1.7658,
      "step": 11140
    },
    {
      "epoch": 1.8967423662498937,
      "grad_norm": 3.63614559173584,
      "learning_rate": 9.583487716794805e-06,
      "loss": 1.7519,
      "step": 11150
    },
    {
      "epoch": 1.8984434804797141,
      "grad_norm": 5.263920307159424,
      "learning_rate": 9.576781194030524e-06,
      "loss": 1.7858,
      "step": 11160
    },
    {
      "epoch": 1.9001445947095348,
      "grad_norm": 4.241003513336182,
      "learning_rate": 9.5700712056545e-06,
      "loss": 1.7522,
      "step": 11170
    },
    {
      "epoch": 1.9018457089393552,
      "grad_norm": 4.151254177093506,
      "learning_rate": 9.563357759815423e-06,
      "loss": 1.8353,
      "step": 11180
    },
    {
      "epoch": 1.903546823169176,
      "grad_norm": 4.62730073928833,
      "learning_rate": 9.556640864666185e-06,
      "loss": 1.7649,
      "step": 11190
    },
    {
      "epoch": 1.9052479373989963,
      "grad_norm": 4.749996185302734,
      "learning_rate": 9.549920528363868e-06,
      "loss": 1.797,
      "step": 11200
    },
    {
      "epoch": 1.9069490516288168,
      "grad_norm": 4.428854942321777,
      "learning_rate": 9.543196759069734e-06,
      "loss": 1.9067,
      "step": 11210
    },
    {
      "epoch": 1.9086501658586374,
      "grad_norm": 4.450839996337891,
      "learning_rate": 9.53646956494921e-06,
      "loss": 1.7731,
      "step": 11220
    },
    {
      "epoch": 1.910351280088458,
      "grad_norm": 4.343857765197754,
      "learning_rate": 9.529738954171882e-06,
      "loss": 1.7605,
      "step": 11230
    },
    {
      "epoch": 1.9120523943182786,
      "grad_norm": 4.726666450500488,
      "learning_rate": 9.523004934911492e-06,
      "loss": 1.8331,
      "step": 11240
    },
    {
      "epoch": 1.913753508548099,
      "grad_norm": 4.01237678527832,
      "learning_rate": 9.516267515345911e-06,
      "loss": 1.8047,
      "step": 11250
    },
    {
      "epoch": 1.9154546227779194,
      "grad_norm": 4.32087516784668,
      "learning_rate": 9.50952670365715e-06,
      "loss": 1.8528,
      "step": 11260
    },
    {
      "epoch": 1.9171557370077401,
      "grad_norm": 4.309276580810547,
      "learning_rate": 9.502782508031331e-06,
      "loss": 1.8709,
      "step": 11270
    },
    {
      "epoch": 1.9188568512375606,
      "grad_norm": 3.589980363845825,
      "learning_rate": 9.496034936658694e-06,
      "loss": 1.781,
      "step": 11280
    },
    {
      "epoch": 1.9205579654673812,
      "grad_norm": 3.661651134490967,
      "learning_rate": 9.489283997733564e-06,
      "loss": 1.7792,
      "step": 11290
    },
    {
      "epoch": 1.9222590796972017,
      "grad_norm": 4.655760288238525,
      "learning_rate": 9.482529699454375e-06,
      "loss": 1.8161,
      "step": 11300
    },
    {
      "epoch": 1.923960193927022,
      "grad_norm": 4.194929599761963,
      "learning_rate": 9.475772050023626e-06,
      "loss": 1.835,
      "step": 11310
    },
    {
      "epoch": 1.9256613081568428,
      "grad_norm": 4.503241539001465,
      "learning_rate": 9.469011057647892e-06,
      "loss": 1.8862,
      "step": 11320
    },
    {
      "epoch": 1.9273624223866632,
      "grad_norm": 4.981051921844482,
      "learning_rate": 9.462246730537804e-06,
      "loss": 1.749,
      "step": 11330
    },
    {
      "epoch": 1.9290635366164839,
      "grad_norm": 3.3752570152282715,
      "learning_rate": 9.455479076908046e-06,
      "loss": 1.8074,
      "step": 11340
    },
    {
      "epoch": 1.9307646508463043,
      "grad_norm": 4.697122097015381,
      "learning_rate": 9.448708104977342e-06,
      "loss": 1.7902,
      "step": 11350
    },
    {
      "epoch": 1.9324657650761248,
      "grad_norm": 4.077208995819092,
      "learning_rate": 9.441933822968444e-06,
      "loss": 1.7781,
      "step": 11360
    },
    {
      "epoch": 1.9341668793059454,
      "grad_norm": 4.550171852111816,
      "learning_rate": 9.435156239108121e-06,
      "loss": 1.7047,
      "step": 11370
    },
    {
      "epoch": 1.9358679935357659,
      "grad_norm": 4.2797393798828125,
      "learning_rate": 9.428375361627159e-06,
      "loss": 1.703,
      "step": 11380
    },
    {
      "epoch": 1.9375691077655866,
      "grad_norm": 3.6947295665740967,
      "learning_rate": 9.421591198760338e-06,
      "loss": 1.8047,
      "step": 11390
    },
    {
      "epoch": 1.939270221995407,
      "grad_norm": 4.601194381713867,
      "learning_rate": 9.414803758746429e-06,
      "loss": 1.7655,
      "step": 11400
    },
    {
      "epoch": 1.9409713362252274,
      "grad_norm": 3.7636353969573975,
      "learning_rate": 9.408013049828185e-06,
      "loss": 1.794,
      "step": 11410
    },
    {
      "epoch": 1.942672450455048,
      "grad_norm": 3.624544143676758,
      "learning_rate": 9.40121908025233e-06,
      "loss": 1.7938,
      "step": 11420
    },
    {
      "epoch": 1.9443735646848685,
      "grad_norm": 4.272632122039795,
      "learning_rate": 9.394421858269537e-06,
      "loss": 1.6927,
      "step": 11430
    },
    {
      "epoch": 1.9460746789146892,
      "grad_norm": 4.589675426483154,
      "learning_rate": 9.387621392134446e-06,
      "loss": 1.8256,
      "step": 11440
    },
    {
      "epoch": 1.9477757931445097,
      "grad_norm": 4.706130027770996,
      "learning_rate": 9.38081769010562e-06,
      "loss": 1.6867,
      "step": 11450
    },
    {
      "epoch": 1.94947690737433,
      "grad_norm": 4.569252967834473,
      "learning_rate": 9.374010760445563e-06,
      "loss": 1.765,
      "step": 11460
    },
    {
      "epoch": 1.9511780216041508,
      "grad_norm": 4.408115386962891,
      "learning_rate": 9.367200611420695e-06,
      "loss": 1.8893,
      "step": 11470
    },
    {
      "epoch": 1.9528791358339712,
      "grad_norm": 4.092146873474121,
      "learning_rate": 9.360387251301345e-06,
      "loss": 1.8494,
      "step": 11480
    },
    {
      "epoch": 1.9545802500637919,
      "grad_norm": 4.814793586730957,
      "learning_rate": 9.35357068836174e-06,
      "loss": 1.6643,
      "step": 11490
    },
    {
      "epoch": 1.9562813642936123,
      "grad_norm": 5.637572765350342,
      "learning_rate": 9.346750930880005e-06,
      "loss": 1.7359,
      "step": 11500
    },
    {
      "epoch": 1.9579824785234328,
      "grad_norm": 5.127346515655518,
      "learning_rate": 9.339927987138132e-06,
      "loss": 1.8141,
      "step": 11510
    },
    {
      "epoch": 1.9596835927532534,
      "grad_norm": 4.358199596405029,
      "learning_rate": 9.33310186542199e-06,
      "loss": 1.8143,
      "step": 11520
    },
    {
      "epoch": 1.9613847069830739,
      "grad_norm": 5.185015678405762,
      "learning_rate": 9.326272574021307e-06,
      "loss": 1.7924,
      "step": 11530
    },
    {
      "epoch": 1.9630858212128945,
      "grad_norm": 3.737765073776245,
      "learning_rate": 9.31944012122966e-06,
      "loss": 1.7952,
      "step": 11540
    },
    {
      "epoch": 1.964786935442715,
      "grad_norm": 4.091814041137695,
      "learning_rate": 9.312604515344463e-06,
      "loss": 1.7418,
      "step": 11550
    },
    {
      "epoch": 1.9664880496725354,
      "grad_norm": 3.7944629192352295,
      "learning_rate": 9.30576576466696e-06,
      "loss": 1.7573,
      "step": 11560
    },
    {
      "epoch": 1.968189163902356,
      "grad_norm": 4.183420658111572,
      "learning_rate": 9.298923877502218e-06,
      "loss": 1.8055,
      "step": 11570
    },
    {
      "epoch": 1.9698902781321765,
      "grad_norm": 4.519730091094971,
      "learning_rate": 9.292078862159108e-06,
      "loss": 1.7887,
      "step": 11580
    },
    {
      "epoch": 1.9715913923619972,
      "grad_norm": 3.6484477519989014,
      "learning_rate": 9.2852307269503e-06,
      "loss": 1.7559,
      "step": 11590
    },
    {
      "epoch": 1.9732925065918177,
      "grad_norm": 4.221194267272949,
      "learning_rate": 9.278379480192255e-06,
      "loss": 1.7274,
      "step": 11600
    },
    {
      "epoch": 1.974993620821638,
      "grad_norm": 4.922400951385498,
      "learning_rate": 9.271525130205218e-06,
      "loss": 1.8043,
      "step": 11610
    },
    {
      "epoch": 1.9766947350514585,
      "grad_norm": 4.591314315795898,
      "learning_rate": 9.264667685313188e-06,
      "loss": 1.8736,
      "step": 11620
    },
    {
      "epoch": 1.9783958492812792,
      "grad_norm": 4.475770950317383,
      "learning_rate": 9.257807153843933e-06,
      "loss": 1.7494,
      "step": 11630
    },
    {
      "epoch": 1.9800969635110999,
      "grad_norm": 4.4012131690979,
      "learning_rate": 9.250943544128974e-06,
      "loss": 1.7984,
      "step": 11640
    },
    {
      "epoch": 1.9817980777409203,
      "grad_norm": 4.212027549743652,
      "learning_rate": 9.244076864503559e-06,
      "loss": 1.7879,
      "step": 11650
    },
    {
      "epoch": 1.9834991919707408,
      "grad_norm": 3.849550247192383,
      "learning_rate": 9.23720712330667e-06,
      "loss": 1.8438,
      "step": 11660
    },
    {
      "epoch": 1.9852003062005612,
      "grad_norm": 3.509598731994629,
      "learning_rate": 9.230334328881004e-06,
      "loss": 1.779,
      "step": 11670
    },
    {
      "epoch": 1.9869014204303819,
      "grad_norm": 3.4126853942871094,
      "learning_rate": 9.22345848957297e-06,
      "loss": 1.8188,
      "step": 11680
    },
    {
      "epoch": 1.9886025346602025,
      "grad_norm": 4.051961898803711,
      "learning_rate": 9.216579613732674e-06,
      "loss": 1.7996,
      "step": 11690
    },
    {
      "epoch": 1.990303648890023,
      "grad_norm": 3.7018580436706543,
      "learning_rate": 9.209697709713905e-06,
      "loss": 1.6788,
      "step": 11700
    },
    {
      "epoch": 1.9920047631198434,
      "grad_norm": 4.118253707885742,
      "learning_rate": 9.202812785874137e-06,
      "loss": 1.8283,
      "step": 11710
    },
    {
      "epoch": 1.9937058773496639,
      "grad_norm": 4.72760534286499,
      "learning_rate": 9.195924850574502e-06,
      "loss": 1.763,
      "step": 11720
    },
    {
      "epoch": 1.9954069915794845,
      "grad_norm": 5.105101108551025,
      "learning_rate": 9.189033912179798e-06,
      "loss": 1.7422,
      "step": 11730
    },
    {
      "epoch": 1.9971081058093052,
      "grad_norm": 4.302199840545654,
      "learning_rate": 9.182139979058463e-06,
      "loss": 1.7793,
      "step": 11740
    },
    {
      "epoch": 1.9988092200391256,
      "grad_norm": 4.676185131072998,
      "learning_rate": 9.175243059582579e-06,
      "loss": 1.7238,
      "step": 11750
    },
    {
      "epoch": 2.000510334268946,
      "grad_norm": 4.118522644042969,
      "learning_rate": 9.168343162127848e-06,
      "loss": 1.7101,
      "step": 11760
    },
    {
      "epoch": 2.0022114484987665,
      "grad_norm": 4.459059238433838,
      "learning_rate": 9.161440295073591e-06,
      "loss": 1.642,
      "step": 11770
    },
    {
      "epoch": 2.0039125627285874,
      "grad_norm": 5.093273639678955,
      "learning_rate": 9.154534466802737e-06,
      "loss": 1.6551,
      "step": 11780
    },
    {
      "epoch": 2.005613676958408,
      "grad_norm": 4.468721866607666,
      "learning_rate": 9.14762568570181e-06,
      "loss": 1.6835,
      "step": 11790
    },
    {
      "epoch": 2.0073147911882283,
      "grad_norm": 4.6995649337768555,
      "learning_rate": 9.140713960160917e-06,
      "loss": 1.7078,
      "step": 11800
    },
    {
      "epoch": 2.0090159054180488,
      "grad_norm": 4.6736907958984375,
      "learning_rate": 9.133799298573745e-06,
      "loss": 1.7109,
      "step": 11810
    },
    {
      "epoch": 2.010717019647869,
      "grad_norm": 4.717840671539307,
      "learning_rate": 9.126881709337545e-06,
      "loss": 1.7939,
      "step": 11820
    },
    {
      "epoch": 2.01241813387769,
      "grad_norm": 4.908730983734131,
      "learning_rate": 9.119961200853122e-06,
      "loss": 1.6915,
      "step": 11830
    },
    {
      "epoch": 2.0141192481075105,
      "grad_norm": 5.266244888305664,
      "learning_rate": 9.113037781524829e-06,
      "loss": 1.7138,
      "step": 11840
    },
    {
      "epoch": 2.015820362337331,
      "grad_norm": 5.060489177703857,
      "learning_rate": 9.106111459760554e-06,
      "loss": 1.7255,
      "step": 11850
    },
    {
      "epoch": 2.0175214765671514,
      "grad_norm": 4.0937089920043945,
      "learning_rate": 9.099182243971704e-06,
      "loss": 1.6528,
      "step": 11860
    },
    {
      "epoch": 2.019222590796972,
      "grad_norm": 5.246833801269531,
      "learning_rate": 9.09225014257321e-06,
      "loss": 1.7111,
      "step": 11870
    },
    {
      "epoch": 2.0209237050267927,
      "grad_norm": 4.744830131530762,
      "learning_rate": 9.085315163983497e-06,
      "loss": 1.766,
      "step": 11880
    },
    {
      "epoch": 2.022624819256613,
      "grad_norm": 4.4739203453063965,
      "learning_rate": 9.078377316624488e-06,
      "loss": 1.6996,
      "step": 11890
    },
    {
      "epoch": 2.0243259334864336,
      "grad_norm": 4.242108345031738,
      "learning_rate": 9.071436608921599e-06,
      "loss": 1.7228,
      "step": 11900
    },
    {
      "epoch": 2.026027047716254,
      "grad_norm": 5.044703960418701,
      "learning_rate": 9.064493049303706e-06,
      "loss": 1.6989,
      "step": 11910
    },
    {
      "epoch": 2.0277281619460745,
      "grad_norm": 4.1773786544799805,
      "learning_rate": 9.057546646203152e-06,
      "loss": 1.6513,
      "step": 11920
    },
    {
      "epoch": 2.0294292761758954,
      "grad_norm": 4.137922286987305,
      "learning_rate": 9.05059740805574e-06,
      "loss": 1.6933,
      "step": 11930
    },
    {
      "epoch": 2.031130390405716,
      "grad_norm": 4.424137115478516,
      "learning_rate": 9.04364534330071e-06,
      "loss": 1.6533,
      "step": 11940
    },
    {
      "epoch": 2.0328315046355363,
      "grad_norm": 4.5705885887146,
      "learning_rate": 9.036690460380735e-06,
      "loss": 1.6856,
      "step": 11950
    },
    {
      "epoch": 2.0345326188653567,
      "grad_norm": 6.594974994659424,
      "learning_rate": 9.02973276774191e-06,
      "loss": 1.7271,
      "step": 11960
    },
    {
      "epoch": 2.036233733095177,
      "grad_norm": 5.3873491287231445,
      "learning_rate": 9.022772273833745e-06,
      "loss": 1.7247,
      "step": 11970
    },
    {
      "epoch": 2.037934847324998,
      "grad_norm": 4.3244500160217285,
      "learning_rate": 9.01580898710915e-06,
      "loss": 1.7503,
      "step": 11980
    },
    {
      "epoch": 2.0396359615548185,
      "grad_norm": 4.935193061828613,
      "learning_rate": 9.008842916024427e-06,
      "loss": 1.7229,
      "step": 11990
    },
    {
      "epoch": 2.041337075784639,
      "grad_norm": 4.176199436187744,
      "learning_rate": 9.00187406903926e-06,
      "loss": 1.7348,
      "step": 12000
    },
    {
      "epoch": 2.0430381900144594,
      "grad_norm": 5.202916622161865,
      "learning_rate": 8.9949024546167e-06,
      "loss": 1.5918,
      "step": 12010
    },
    {
      "epoch": 2.04473930424428,
      "grad_norm": 4.1330437660217285,
      "learning_rate": 8.987928081223167e-06,
      "loss": 1.7322,
      "step": 12020
    },
    {
      "epoch": 2.0464404184741007,
      "grad_norm": 4.667124271392822,
      "learning_rate": 8.980950957328423e-06,
      "loss": 1.6891,
      "step": 12030
    },
    {
      "epoch": 2.048141532703921,
      "grad_norm": 5.26122522354126,
      "learning_rate": 8.973971091405575e-06,
      "loss": 1.7367,
      "step": 12040
    },
    {
      "epoch": 2.0498426469337416,
      "grad_norm": 4.209774494171143,
      "learning_rate": 8.966988491931062e-06,
      "loss": 1.7748,
      "step": 12050
    },
    {
      "epoch": 2.051543761163562,
      "grad_norm": 4.397939682006836,
      "learning_rate": 8.960003167384633e-06,
      "loss": 1.6671,
      "step": 12060
    },
    {
      "epoch": 2.0532448753933825,
      "grad_norm": 4.774211883544922,
      "learning_rate": 8.953015126249358e-06,
      "loss": 1.7462,
      "step": 12070
    },
    {
      "epoch": 2.0549459896232034,
      "grad_norm": 4.361632823944092,
      "learning_rate": 8.9460243770116e-06,
      "loss": 1.6999,
      "step": 12080
    },
    {
      "epoch": 2.056647103853024,
      "grad_norm": 4.739696979522705,
      "learning_rate": 8.939030928161008e-06,
      "loss": 1.7113,
      "step": 12090
    },
    {
      "epoch": 2.0583482180828443,
      "grad_norm": 4.217962265014648,
      "learning_rate": 8.932034788190522e-06,
      "loss": 1.7864,
      "step": 12100
    },
    {
      "epoch": 2.0600493323126647,
      "grad_norm": 4.49211311340332,
      "learning_rate": 8.925035965596332e-06,
      "loss": 1.7689,
      "step": 12110
    },
    {
      "epoch": 2.061750446542485,
      "grad_norm": 4.063029766082764,
      "learning_rate": 8.9180344688779e-06,
      "loss": 1.7162,
      "step": 12120
    },
    {
      "epoch": 2.063451560772306,
      "grad_norm": 4.042270183563232,
      "learning_rate": 8.911030306537929e-06,
      "loss": 1.7093,
      "step": 12130
    },
    {
      "epoch": 2.0651526750021265,
      "grad_norm": 4.807375907897949,
      "learning_rate": 8.90402348708236e-06,
      "loss": 1.5763,
      "step": 12140
    },
    {
      "epoch": 2.066853789231947,
      "grad_norm": 4.455898284912109,
      "learning_rate": 8.897014019020363e-06,
      "loss": 1.6977,
      "step": 12150
    },
    {
      "epoch": 2.0685549034617674,
      "grad_norm": 4.151392459869385,
      "learning_rate": 8.890001910864324e-06,
      "loss": 1.8438,
      "step": 12160
    },
    {
      "epoch": 2.070256017691588,
      "grad_norm": 5.351758003234863,
      "learning_rate": 8.882987171129831e-06,
      "loss": 1.6591,
      "step": 12170
    },
    {
      "epoch": 2.0719571319214087,
      "grad_norm": 4.933199405670166,
      "learning_rate": 8.875969808335677e-06,
      "loss": 1.6332,
      "step": 12180
    },
    {
      "epoch": 2.073658246151229,
      "grad_norm": 4.691521167755127,
      "learning_rate": 8.86894983100383e-06,
      "loss": 1.6974,
      "step": 12190
    },
    {
      "epoch": 2.0753593603810496,
      "grad_norm": 5.424036502838135,
      "learning_rate": 8.861927247659436e-06,
      "loss": 1.7456,
      "step": 12200
    },
    {
      "epoch": 2.07706047461087,
      "grad_norm": 3.9998979568481445,
      "learning_rate": 8.85490206683081e-06,
      "loss": 1.6898,
      "step": 12210
    },
    {
      "epoch": 2.0787615888406905,
      "grad_norm": 4.3668718338012695,
      "learning_rate": 8.84787429704942e-06,
      "loss": 1.6834,
      "step": 12220
    },
    {
      "epoch": 2.0804627030705114,
      "grad_norm": 4.986365795135498,
      "learning_rate": 8.840843946849874e-06,
      "loss": 1.8197,
      "step": 12230
    },
    {
      "epoch": 2.082163817300332,
      "grad_norm": 4.653778553009033,
      "learning_rate": 8.83381102476992e-06,
      "loss": 1.7306,
      "step": 12240
    },
    {
      "epoch": 2.0838649315301523,
      "grad_norm": 3.908618450164795,
      "learning_rate": 8.826775539350426e-06,
      "loss": 1.6604,
      "step": 12250
    },
    {
      "epoch": 2.0855660457599727,
      "grad_norm": 4.5211687088012695,
      "learning_rate": 8.819737499135368e-06,
      "loss": 1.7098,
      "step": 12260
    },
    {
      "epoch": 2.087267159989793,
      "grad_norm": 5.266591548919678,
      "learning_rate": 8.812696912671833e-06,
      "loss": 1.7872,
      "step": 12270
    },
    {
      "epoch": 2.0889682742196136,
      "grad_norm": 6.0518879890441895,
      "learning_rate": 8.805653788509997e-06,
      "loss": 1.6938,
      "step": 12280
    },
    {
      "epoch": 2.0906693884494345,
      "grad_norm": 5.185415744781494,
      "learning_rate": 8.798608135203115e-06,
      "loss": 1.7457,
      "step": 12290
    },
    {
      "epoch": 2.092370502679255,
      "grad_norm": 4.733633995056152,
      "learning_rate": 8.79155996130752e-06,
      "loss": 1.729,
      "step": 12300
    },
    {
      "epoch": 2.0940716169090754,
      "grad_norm": 3.8256046772003174,
      "learning_rate": 8.784509275382598e-06,
      "loss": 1.7013,
      "step": 12310
    },
    {
      "epoch": 2.095772731138896,
      "grad_norm": 4.688540935516357,
      "learning_rate": 8.777456085990791e-06,
      "loss": 1.6962,
      "step": 12320
    },
    {
      "epoch": 2.0974738453687163,
      "grad_norm": 5.134762287139893,
      "learning_rate": 8.77040040169758e-06,
      "loss": 1.7053,
      "step": 12330
    },
    {
      "epoch": 2.099174959598537,
      "grad_norm": 4.646895885467529,
      "learning_rate": 8.763342231071477e-06,
      "loss": 1.7464,
      "step": 12340
    },
    {
      "epoch": 2.1008760738283576,
      "grad_norm": 4.749012470245361,
      "learning_rate": 8.756281582684011e-06,
      "loss": 1.7256,
      "step": 12350
    },
    {
      "epoch": 2.102577188058178,
      "grad_norm": 3.6838204860687256,
      "learning_rate": 8.74921846510972e-06,
      "loss": 1.6454,
      "step": 12360
    },
    {
      "epoch": 2.1042783022879985,
      "grad_norm": 4.912604331970215,
      "learning_rate": 8.74215288692614e-06,
      "loss": 1.7182,
      "step": 12370
    },
    {
      "epoch": 2.105979416517819,
      "grad_norm": 5.738154411315918,
      "learning_rate": 8.735084856713802e-06,
      "loss": 1.7614,
      "step": 12380
    },
    {
      "epoch": 2.10768053074764,
      "grad_norm": 5.089351654052734,
      "learning_rate": 8.72801438305621e-06,
      "loss": 1.6985,
      "step": 12390
    },
    {
      "epoch": 2.1093816449774603,
      "grad_norm": 4.910715103149414,
      "learning_rate": 8.720941474539826e-06,
      "loss": 1.722,
      "step": 12400
    },
    {
      "epoch": 2.1110827592072807,
      "grad_norm": 4.526627540588379,
      "learning_rate": 8.713866139754089e-06,
      "loss": 1.7323,
      "step": 12410
    },
    {
      "epoch": 2.112783873437101,
      "grad_norm": 4.7195611000061035,
      "learning_rate": 8.706788387291366e-06,
      "loss": 1.7134,
      "step": 12420
    },
    {
      "epoch": 2.1144849876669216,
      "grad_norm": 5.276111602783203,
      "learning_rate": 8.69970822574697e-06,
      "loss": 1.6738,
      "step": 12430
    },
    {
      "epoch": 2.1161861018967425,
      "grad_norm": 4.766108989715576,
      "learning_rate": 8.69262566371914e-06,
      "loss": 1.6695,
      "step": 12440
    },
    {
      "epoch": 2.117887216126563,
      "grad_norm": 5.434549331665039,
      "learning_rate": 8.685540709809023e-06,
      "loss": 1.689,
      "step": 12450
    },
    {
      "epoch": 2.1195883303563834,
      "grad_norm": 4.173770904541016,
      "learning_rate": 8.678453372620676e-06,
      "loss": 1.7214,
      "step": 12460
    },
    {
      "epoch": 2.121289444586204,
      "grad_norm": 4.431570053100586,
      "learning_rate": 8.671363660761051e-06,
      "loss": 1.7108,
      "step": 12470
    },
    {
      "epoch": 2.1229905588160243,
      "grad_norm": 5.044260025024414,
      "learning_rate": 8.66427158283998e-06,
      "loss": 1.7559,
      "step": 12480
    },
    {
      "epoch": 2.124691673045845,
      "grad_norm": 4.083590507507324,
      "learning_rate": 8.657177147470172e-06,
      "loss": 1.7039,
      "step": 12490
    },
    {
      "epoch": 2.1263927872756656,
      "grad_norm": 4.259870529174805,
      "learning_rate": 8.650080363267197e-06,
      "loss": 1.7676,
      "step": 12500
    },
    {
      "epoch": 2.128093901505486,
      "grad_norm": 5.370848178863525,
      "learning_rate": 8.642981238849477e-06,
      "loss": 1.6801,
      "step": 12510
    },
    {
      "epoch": 2.1297950157353065,
      "grad_norm": 5.005211353302002,
      "learning_rate": 8.635879782838281e-06,
      "loss": 1.6762,
      "step": 12520
    },
    {
      "epoch": 2.131496129965127,
      "grad_norm": 4.411081314086914,
      "learning_rate": 8.6287760038577e-06,
      "loss": 1.6965,
      "step": 12530
    },
    {
      "epoch": 2.133197244194948,
      "grad_norm": 4.327058792114258,
      "learning_rate": 8.621669910534654e-06,
      "loss": 1.7941,
      "step": 12540
    },
    {
      "epoch": 2.1348983584247683,
      "grad_norm": 4.803872585296631,
      "learning_rate": 8.61456151149887e-06,
      "loss": 1.7329,
      "step": 12550
    },
    {
      "epoch": 2.1365994726545887,
      "grad_norm": 5.848106861114502,
      "learning_rate": 8.60745081538288e-06,
      "loss": 1.7452,
      "step": 12560
    },
    {
      "epoch": 2.138300586884409,
      "grad_norm": 4.075498104095459,
      "learning_rate": 8.600337830821995e-06,
      "loss": 1.7577,
      "step": 12570
    },
    {
      "epoch": 2.1400017011142296,
      "grad_norm": 3.8453147411346436,
      "learning_rate": 8.593222566454318e-06,
      "loss": 1.7492,
      "step": 12580
    },
    {
      "epoch": 2.1417028153440505,
      "grad_norm": 4.6681227684021,
      "learning_rate": 8.58610503092071e-06,
      "loss": 1.6385,
      "step": 12590
    },
    {
      "epoch": 2.143403929573871,
      "grad_norm": 5.454570770263672,
      "learning_rate": 8.578985232864796e-06,
      "loss": 1.7272,
      "step": 12600
    },
    {
      "epoch": 2.1451050438036914,
      "grad_norm": 5.396411418914795,
      "learning_rate": 8.571863180932948e-06,
      "loss": 1.7069,
      "step": 12610
    },
    {
      "epoch": 2.146806158033512,
      "grad_norm": 5.200651168823242,
      "learning_rate": 8.564738883774275e-06,
      "loss": 1.5999,
      "step": 12620
    },
    {
      "epoch": 2.1485072722633323,
      "grad_norm": 4.8601813316345215,
      "learning_rate": 8.557612350040611e-06,
      "loss": 1.6065,
      "step": 12630
    },
    {
      "epoch": 2.150208386493153,
      "grad_norm": 4.7667317390441895,
      "learning_rate": 8.550483588386504e-06,
      "loss": 1.5607,
      "step": 12640
    },
    {
      "epoch": 2.1519095007229736,
      "grad_norm": 5.358502388000488,
      "learning_rate": 8.543352607469216e-06,
      "loss": 1.7666,
      "step": 12650
    },
    {
      "epoch": 2.153610614952794,
      "grad_norm": 4.752895355224609,
      "learning_rate": 8.536219415948695e-06,
      "loss": 1.6798,
      "step": 12660
    },
    {
      "epoch": 2.1553117291826145,
      "grad_norm": 3.580336570739746,
      "learning_rate": 8.529084022487582e-06,
      "loss": 1.7687,
      "step": 12670
    },
    {
      "epoch": 2.157012843412435,
      "grad_norm": 4.459094524383545,
      "learning_rate": 8.521946435751182e-06,
      "loss": 1.6895,
      "step": 12680
    },
    {
      "epoch": 2.158713957642256,
      "grad_norm": 6.012520790100098,
      "learning_rate": 8.514806664407472e-06,
      "loss": 1.7698,
      "step": 12690
    },
    {
      "epoch": 2.1604150718720763,
      "grad_norm": 4.733858108520508,
      "learning_rate": 8.507664717127077e-06,
      "loss": 1.6648,
      "step": 12700
    },
    {
      "epoch": 2.1621161861018967,
      "grad_norm": 4.045647621154785,
      "learning_rate": 8.500520602583268e-06,
      "loss": 1.66,
      "step": 12710
    },
    {
      "epoch": 2.163817300331717,
      "grad_norm": 4.753208637237549,
      "learning_rate": 8.493374329451948e-06,
      "loss": 1.6367,
      "step": 12720
    },
    {
      "epoch": 2.1655184145615376,
      "grad_norm": 5.455912113189697,
      "learning_rate": 8.48622590641164e-06,
      "loss": 1.7819,
      "step": 12730
    },
    {
      "epoch": 2.1672195287913585,
      "grad_norm": 4.091403484344482,
      "learning_rate": 8.479075342143472e-06,
      "loss": 1.7988,
      "step": 12740
    },
    {
      "epoch": 2.168920643021179,
      "grad_norm": 5.065116882324219,
      "learning_rate": 8.471922645331183e-06,
      "loss": 1.6525,
      "step": 12750
    },
    {
      "epoch": 2.1706217572509994,
      "grad_norm": 4.216938018798828,
      "learning_rate": 8.464767824661097e-06,
      "loss": 1.7405,
      "step": 12760
    },
    {
      "epoch": 2.17232287148082,
      "grad_norm": 4.321226596832275,
      "learning_rate": 8.457610888822113e-06,
      "loss": 1.7527,
      "step": 12770
    },
    {
      "epoch": 2.1740239857106403,
      "grad_norm": 4.348523139953613,
      "learning_rate": 8.45045184650571e-06,
      "loss": 1.7419,
      "step": 12780
    },
    {
      "epoch": 2.175725099940461,
      "grad_norm": 4.465217113494873,
      "learning_rate": 8.443290706405914e-06,
      "loss": 1.716,
      "step": 12790
    },
    {
      "epoch": 2.1774262141702816,
      "grad_norm": 3.363562822341919,
      "learning_rate": 8.436127477219299e-06,
      "loss": 1.6208,
      "step": 12800
    },
    {
      "epoch": 2.179127328400102,
      "grad_norm": 4.242889881134033,
      "learning_rate": 8.428962167644982e-06,
      "loss": 1.7619,
      "step": 12810
    },
    {
      "epoch": 2.1808284426299225,
      "grad_norm": 4.917881488800049,
      "learning_rate": 8.421794786384607e-06,
      "loss": 1.6682,
      "step": 12820
    },
    {
      "epoch": 2.182529556859743,
      "grad_norm": 3.955270528793335,
      "learning_rate": 8.41462534214233e-06,
      "loss": 1.6528,
      "step": 12830
    },
    {
      "epoch": 2.184230671089564,
      "grad_norm": 4.5821309089660645,
      "learning_rate": 8.407453843624807e-06,
      "loss": 1.677,
      "step": 12840
    },
    {
      "epoch": 2.1859317853193843,
      "grad_norm": 5.448122978210449,
      "learning_rate": 8.400280299541205e-06,
      "loss": 1.6756,
      "step": 12850
    },
    {
      "epoch": 2.1876328995492047,
      "grad_norm": 3.5626063346862793,
      "learning_rate": 8.393104718603154e-06,
      "loss": 1.656,
      "step": 12860
    },
    {
      "epoch": 2.189334013779025,
      "grad_norm": 4.248167037963867,
      "learning_rate": 8.385927109524776e-06,
      "loss": 1.6074,
      "step": 12870
    },
    {
      "epoch": 2.1910351280088456,
      "grad_norm": 5.133649826049805,
      "learning_rate": 8.378747481022648e-06,
      "loss": 1.6894,
      "step": 12880
    },
    {
      "epoch": 2.1927362422386665,
      "grad_norm": 4.838881015777588,
      "learning_rate": 8.3715658418158e-06,
      "loss": 1.7187,
      "step": 12890
    },
    {
      "epoch": 2.194437356468487,
      "grad_norm": 4.620151519775391,
      "learning_rate": 8.364382200625701e-06,
      "loss": 1.6652,
      "step": 12900
    },
    {
      "epoch": 2.1961384706983074,
      "grad_norm": 4.420077800750732,
      "learning_rate": 8.357196566176256e-06,
      "loss": 1.6278,
      "step": 12910
    },
    {
      "epoch": 2.197839584928128,
      "grad_norm": 4.555145263671875,
      "learning_rate": 8.350008947193792e-06,
      "loss": 1.7234,
      "step": 12920
    },
    {
      "epoch": 2.1995406991579483,
      "grad_norm": 5.071643829345703,
      "learning_rate": 8.342819352407036e-06,
      "loss": 1.7096,
      "step": 12930
    },
    {
      "epoch": 2.201241813387769,
      "grad_norm": 5.000056743621826,
      "learning_rate": 8.335627790547125e-06,
      "loss": 1.7824,
      "step": 12940
    },
    {
      "epoch": 2.2029429276175896,
      "grad_norm": 4.919875144958496,
      "learning_rate": 8.32843427034758e-06,
      "loss": 1.713,
      "step": 12950
    },
    {
      "epoch": 2.20464404184741,
      "grad_norm": 5.44525671005249,
      "learning_rate": 8.321238800544301e-06,
      "loss": 1.6816,
      "step": 12960
    },
    {
      "epoch": 2.2063451560772305,
      "grad_norm": 5.025317668914795,
      "learning_rate": 8.31404138987556e-06,
      "loss": 1.7378,
      "step": 12970
    },
    {
      "epoch": 2.208046270307051,
      "grad_norm": 4.864163398742676,
      "learning_rate": 8.306842047081977e-06,
      "loss": 1.7188,
      "step": 12980
    },
    {
      "epoch": 2.209747384536872,
      "grad_norm": 5.568412780761719,
      "learning_rate": 8.299640780906527e-06,
      "loss": 1.7523,
      "step": 12990
    },
    {
      "epoch": 2.2114484987666922,
      "grad_norm": 4.593928813934326,
      "learning_rate": 8.292437600094513e-06,
      "loss": 1.652,
      "step": 13000
    },
    {
      "epoch": 2.2131496129965127,
      "grad_norm": 4.015298366546631,
      "learning_rate": 8.28523251339357e-06,
      "loss": 1.6344,
      "step": 13010
    },
    {
      "epoch": 2.214850727226333,
      "grad_norm": 5.201580047607422,
      "learning_rate": 8.27874631305947e-06,
      "loss": 1.8141,
      "step": 13020
    },
    {
      "epoch": 2.2165518414561536,
      "grad_norm": 4.677765369415283,
      "learning_rate": 8.27153762927757e-06,
      "loss": 1.8078,
      "step": 13030
    },
    {
      "epoch": 2.2182529556859745,
      "grad_norm": 4.313126564025879,
      "learning_rate": 8.264327064987924e-06,
      "loss": 1.6792,
      "step": 13040
    },
    {
      "epoch": 2.219954069915795,
      "grad_norm": 4.24152946472168,
      "learning_rate": 8.257114628947131e-06,
      "loss": 1.6695,
      "step": 13050
    },
    {
      "epoch": 2.2216551841456154,
      "grad_norm": 4.828244209289551,
      "learning_rate": 8.249900329914065e-06,
      "loss": 1.7602,
      "step": 13060
    },
    {
      "epoch": 2.223356298375436,
      "grad_norm": 4.852785110473633,
      "learning_rate": 8.242684176649857e-06,
      "loss": 1.7807,
      "step": 13070
    },
    {
      "epoch": 2.2250574126052562,
      "grad_norm": 5.04357385635376,
      "learning_rate": 8.235466177917896e-06,
      "loss": 1.703,
      "step": 13080
    },
    {
      "epoch": 2.226758526835077,
      "grad_norm": 5.173859119415283,
      "learning_rate": 8.228246342483809e-06,
      "loss": 1.666,
      "step": 13090
    },
    {
      "epoch": 2.2284596410648976,
      "grad_norm": 4.799098014831543,
      "learning_rate": 8.221024679115452e-06,
      "loss": 1.6277,
      "step": 13100
    },
    {
      "epoch": 2.230160755294718,
      "grad_norm": 4.921494960784912,
      "learning_rate": 8.213801196582905e-06,
      "loss": 1.5804,
      "step": 13110
    },
    {
      "epoch": 2.2318618695245385,
      "grad_norm": 4.846426486968994,
      "learning_rate": 8.206575903658452e-06,
      "loss": 1.7394,
      "step": 13120
    },
    {
      "epoch": 2.233562983754359,
      "grad_norm": 4.785889148712158,
      "learning_rate": 8.19934880911658e-06,
      "loss": 1.6805,
      "step": 13130
    },
    {
      "epoch": 2.23526409798418,
      "grad_norm": 5.29194974899292,
      "learning_rate": 8.192119921733966e-06,
      "loss": 1.6522,
      "step": 13140
    },
    {
      "epoch": 2.2369652122140002,
      "grad_norm": 5.888778209686279,
      "learning_rate": 8.184889250289456e-06,
      "loss": 1.7498,
      "step": 13150
    },
    {
      "epoch": 2.2386663264438207,
      "grad_norm": 3.804825782775879,
      "learning_rate": 8.177656803564068e-06,
      "loss": 1.7379,
      "step": 13160
    },
    {
      "epoch": 2.240367440673641,
      "grad_norm": 4.515087127685547,
      "learning_rate": 8.170422590340977e-06,
      "loss": 1.7091,
      "step": 13170
    },
    {
      "epoch": 2.2420685549034616,
      "grad_norm": 5.151566505432129,
      "learning_rate": 8.163186619405497e-06,
      "loss": 1.6192,
      "step": 13180
    },
    {
      "epoch": 2.2437696691332825,
      "grad_norm": 7.327603816986084,
      "learning_rate": 8.155948899545087e-06,
      "loss": 1.6528,
      "step": 13190
    },
    {
      "epoch": 2.245470783363103,
      "grad_norm": 4.373720169067383,
      "learning_rate": 8.148709439549322e-06,
      "loss": 1.7527,
      "step": 13200
    },
    {
      "epoch": 2.2471718975929234,
      "grad_norm": 4.045927047729492,
      "learning_rate": 8.141468248209889e-06,
      "loss": 1.7512,
      "step": 13210
    },
    {
      "epoch": 2.248873011822744,
      "grad_norm": 4.1613616943359375,
      "learning_rate": 8.134225334320583e-06,
      "loss": 1.7419,
      "step": 13220
    },
    {
      "epoch": 2.2505741260525642,
      "grad_norm": 5.061476230621338,
      "learning_rate": 8.12770524630982e-06,
      "loss": 1.7468,
      "step": 13230
    },
    {
      "epoch": 2.252275240282385,
      "grad_norm": 4.439516067504883,
      "learning_rate": 8.12045908381014e-06,
      "loss": 1.7932,
      "step": 13240
    },
    {
      "epoch": 2.2539763545122056,
      "grad_norm": 5.531265735626221,
      "learning_rate": 8.113211224274374e-06,
      "loss": 1.7472,
      "step": 13250
    },
    {
      "epoch": 2.255677468742026,
      "grad_norm": 4.822131156921387,
      "learning_rate": 8.105961676504417e-06,
      "loss": 1.788,
      "step": 13260
    },
    {
      "epoch": 2.2573785829718465,
      "grad_norm": 4.480312824249268,
      "learning_rate": 8.098710449304209e-06,
      "loss": 1.7329,
      "step": 13270
    },
    {
      "epoch": 2.259079697201667,
      "grad_norm": 4.88050651550293,
      "learning_rate": 8.09145755147973e-06,
      "loss": 1.688,
      "step": 13280
    },
    {
      "epoch": 2.260780811431488,
      "grad_norm": 5.95066499710083,
      "learning_rate": 8.084202991838988e-06,
      "loss": 1.7034,
      "step": 13290
    },
    {
      "epoch": 2.2624819256613082,
      "grad_norm": 3.742136001586914,
      "learning_rate": 8.076946779192013e-06,
      "loss": 1.7597,
      "step": 13300
    },
    {
      "epoch": 2.2641830398911287,
      "grad_norm": 4.50217342376709,
      "learning_rate": 8.069688922350838e-06,
      "loss": 1.7729,
      "step": 13310
    },
    {
      "epoch": 2.265884154120949,
      "grad_norm": 5.608749866485596,
      "learning_rate": 8.062429430129496e-06,
      "loss": 1.7258,
      "step": 13320
    },
    {
      "epoch": 2.2675852683507696,
      "grad_norm": 4.774039268493652,
      "learning_rate": 8.055168311344003e-06,
      "loss": 1.6989,
      "step": 13330
    },
    {
      "epoch": 2.2692863825805905,
      "grad_norm": 4.012284755706787,
      "learning_rate": 8.04790557481235e-06,
      "loss": 1.6824,
      "step": 13340
    },
    {
      "epoch": 2.270987496810411,
      "grad_norm": 4.178895950317383,
      "learning_rate": 8.040641229354499e-06,
      "loss": 1.6828,
      "step": 13350
    },
    {
      "epoch": 2.2726886110402313,
      "grad_norm": 4.772857189178467,
      "learning_rate": 8.033375283792356e-06,
      "loss": 1.6993,
      "step": 13360
    },
    {
      "epoch": 2.274389725270052,
      "grad_norm": 4.545144557952881,
      "learning_rate": 8.026107746949784e-06,
      "loss": 1.6041,
      "step": 13370
    },
    {
      "epoch": 2.2760908394998722,
      "grad_norm": 4.623827934265137,
      "learning_rate": 8.018838627652563e-06,
      "loss": 1.6094,
      "step": 13380
    },
    {
      "epoch": 2.277791953729693,
      "grad_norm": 4.452998161315918,
      "learning_rate": 8.0115679347284e-06,
      "loss": 1.7036,
      "step": 13390
    },
    {
      "epoch": 2.2794930679595136,
      "grad_norm": 4.04271936416626,
      "learning_rate": 8.004295677006923e-06,
      "loss": 1.6462,
      "step": 13400
    },
    {
      "epoch": 2.281194182189334,
      "grad_norm": 4.8894524574279785,
      "learning_rate": 7.997021863319647e-06,
      "loss": 1.782,
      "step": 13410
    },
    {
      "epoch": 2.2828952964191545,
      "grad_norm": 5.355381965637207,
      "learning_rate": 7.989746502499978e-06,
      "loss": 1.6844,
      "step": 13420
    },
    {
      "epoch": 2.284596410648975,
      "grad_norm": 4.542727947235107,
      "learning_rate": 7.982469603383212e-06,
      "loss": 1.6117,
      "step": 13430
    },
    {
      "epoch": 2.286297524878796,
      "grad_norm": 4.602875232696533,
      "learning_rate": 7.975191174806502e-06,
      "loss": 1.65,
      "step": 13440
    },
    {
      "epoch": 2.2879986391086162,
      "grad_norm": 4.5872955322265625,
      "learning_rate": 7.967911225608862e-06,
      "loss": 1.6256,
      "step": 13450
    },
    {
      "epoch": 2.2896997533384367,
      "grad_norm": 4.036448955535889,
      "learning_rate": 7.960629764631154e-06,
      "loss": 1.7718,
      "step": 13460
    },
    {
      "epoch": 2.291400867568257,
      "grad_norm": 5.055094242095947,
      "learning_rate": 7.953346800716076e-06,
      "loss": 1.7826,
      "step": 13470
    },
    {
      "epoch": 2.2931019817980776,
      "grad_norm": 4.7970380783081055,
      "learning_rate": 7.946062342708144e-06,
      "loss": 1.5622,
      "step": 13480
    },
    {
      "epoch": 2.2948030960278984,
      "grad_norm": 5.095872402191162,
      "learning_rate": 7.9387763994537e-06,
      "loss": 1.7061,
      "step": 13490
    },
    {
      "epoch": 2.296504210257719,
      "grad_norm": 4.219940185546875,
      "learning_rate": 7.93148897980088e-06,
      "loss": 1.7782,
      "step": 13500
    },
    {
      "epoch": 2.2982053244875393,
      "grad_norm": 4.889625549316406,
      "learning_rate": 7.924200092599618e-06,
      "loss": 1.7454,
      "step": 13510
    },
    {
      "epoch": 2.29990643871736,
      "grad_norm": 4.371610641479492,
      "learning_rate": 7.916909746701637e-06,
      "loss": 1.7023,
      "step": 13520
    },
    {
      "epoch": 2.3016075529471802,
      "grad_norm": 5.004741191864014,
      "learning_rate": 7.909617950960411e-06,
      "loss": 1.7414,
      "step": 13530
    },
    {
      "epoch": 2.303308667177001,
      "grad_norm": 4.704267501831055,
      "learning_rate": 7.902324714231194e-06,
      "loss": 1.7434,
      "step": 13540
    },
    {
      "epoch": 2.3050097814068216,
      "grad_norm": 5.118117809295654,
      "learning_rate": 7.895030045370983e-06,
      "loss": 1.7567,
      "step": 13550
    },
    {
      "epoch": 2.306710895636642,
      "grad_norm": 4.657116889953613,
      "learning_rate": 7.887733953238513e-06,
      "loss": 1.6377,
      "step": 13560
    },
    {
      "epoch": 2.3084120098664624,
      "grad_norm": 5.6785783767700195,
      "learning_rate": 7.880436446694251e-06,
      "loss": 1.6896,
      "step": 13570
    },
    {
      "epoch": 2.310113124096283,
      "grad_norm": 4.47097110748291,
      "learning_rate": 7.87313753460038e-06,
      "loss": 1.6696,
      "step": 13580
    },
    {
      "epoch": 2.3118142383261038,
      "grad_norm": 3.680044174194336,
      "learning_rate": 7.865837225820784e-06,
      "loss": 1.6833,
      "step": 13590
    },
    {
      "epoch": 2.313515352555924,
      "grad_norm": 4.680473804473877,
      "learning_rate": 7.858535529221054e-06,
      "loss": 1.7197,
      "step": 13600
    },
    {
      "epoch": 2.3152164667857447,
      "grad_norm": 4.563645839691162,
      "learning_rate": 7.85123245366846e-06,
      "loss": 1.6869,
      "step": 13610
    },
    {
      "epoch": 2.316917581015565,
      "grad_norm": 4.670734405517578,
      "learning_rate": 7.843928008031949e-06,
      "loss": 1.7179,
      "step": 13620
    },
    {
      "epoch": 2.3186186952453856,
      "grad_norm": 3.7702691555023193,
      "learning_rate": 7.836622201182127e-06,
      "loss": 1.6822,
      "step": 13630
    },
    {
      "epoch": 2.3203198094752064,
      "grad_norm": 4.605519771575928,
      "learning_rate": 7.82931504199126e-06,
      "loss": 1.678,
      "step": 13640
    },
    {
      "epoch": 2.322020923705027,
      "grad_norm": 4.155745983123779,
      "learning_rate": 7.822006539333249e-06,
      "loss": 1.7425,
      "step": 13650
    },
    {
      "epoch": 2.3237220379348473,
      "grad_norm": 4.561372756958008,
      "learning_rate": 7.814696702083637e-06,
      "loss": 1.6521,
      "step": 13660
    },
    {
      "epoch": 2.3254231521646678,
      "grad_norm": 4.688206672668457,
      "learning_rate": 7.807385539119572e-06,
      "loss": 1.6336,
      "step": 13670
    },
    {
      "epoch": 2.327124266394488,
      "grad_norm": 4.2422966957092285,
      "learning_rate": 7.800073059319828e-06,
      "loss": 1.7293,
      "step": 13680
    },
    {
      "epoch": 2.328825380624309,
      "grad_norm": 5.442010402679443,
      "learning_rate": 7.792759271564769e-06,
      "loss": 1.6455,
      "step": 13690
    },
    {
      "epoch": 2.3305264948541295,
      "grad_norm": 4.347396373748779,
      "learning_rate": 7.785444184736351e-06,
      "loss": 1.6886,
      "step": 13700
    },
    {
      "epoch": 2.33222760908395,
      "grad_norm": 4.186034202575684,
      "learning_rate": 7.778127807718106e-06,
      "loss": 1.6293,
      "step": 13710
    },
    {
      "epoch": 2.3339287233137704,
      "grad_norm": 4.553366184234619,
      "learning_rate": 7.770810149395132e-06,
      "loss": 1.6962,
      "step": 13720
    },
    {
      "epoch": 2.335629837543591,
      "grad_norm": 5.827213764190674,
      "learning_rate": 7.763491218654083e-06,
      "loss": 1.769,
      "step": 13730
    },
    {
      "epoch": 2.3373309517734118,
      "grad_norm": 5.1207051277160645,
      "learning_rate": 7.756171024383162e-06,
      "loss": 1.6669,
      "step": 13740
    },
    {
      "epoch": 2.339032066003232,
      "grad_norm": 5.226886749267578,
      "learning_rate": 7.748849575472106e-06,
      "loss": 1.7029,
      "step": 13750
    },
    {
      "epoch": 2.3407331802330527,
      "grad_norm": 4.730441093444824,
      "learning_rate": 7.741526880812167e-06,
      "loss": 1.6802,
      "step": 13760
    },
    {
      "epoch": 2.342434294462873,
      "grad_norm": 5.613226890563965,
      "learning_rate": 7.734202949296123e-06,
      "loss": 1.6901,
      "step": 13770
    },
    {
      "epoch": 2.3441354086926935,
      "grad_norm": 4.034677028656006,
      "learning_rate": 7.726877789818241e-06,
      "loss": 1.6834,
      "step": 13780
    },
    {
      "epoch": 2.3458365229225144,
      "grad_norm": 4.615573883056641,
      "learning_rate": 7.719551411274289e-06,
      "loss": 1.6692,
      "step": 13790
    },
    {
      "epoch": 2.347537637152335,
      "grad_norm": 4.723323345184326,
      "learning_rate": 7.712223822561515e-06,
      "loss": 1.7574,
      "step": 13800
    },
    {
      "epoch": 2.3492387513821553,
      "grad_norm": 4.764580249786377,
      "learning_rate": 7.704895032578631e-06,
      "loss": 1.7047,
      "step": 13810
    },
    {
      "epoch": 2.3509398656119758,
      "grad_norm": 7.666451454162598,
      "learning_rate": 7.697565050225806e-06,
      "loss": 1.7018,
      "step": 13820
    },
    {
      "epoch": 2.352640979841796,
      "grad_norm": 5.401124477386475,
      "learning_rate": 7.69023388440467e-06,
      "loss": 1.7899,
      "step": 13830
    },
    {
      "epoch": 2.354342094071617,
      "grad_norm": 5.429254055023193,
      "learning_rate": 7.682901544018278e-06,
      "loss": 1.6748,
      "step": 13840
    },
    {
      "epoch": 2.3560432083014375,
      "grad_norm": 4.640020370483398,
      "learning_rate": 7.675568037971112e-06,
      "loss": 1.7833,
      "step": 13850
    },
    {
      "epoch": 2.357744322531258,
      "grad_norm": 5.1837873458862305,
      "learning_rate": 7.668233375169079e-06,
      "loss": 1.6224,
      "step": 13860
    },
    {
      "epoch": 2.3594454367610784,
      "grad_norm": 4.908547401428223,
      "learning_rate": 7.66089756451948e-06,
      "loss": 1.7572,
      "step": 13870
    },
    {
      "epoch": 2.361146550990899,
      "grad_norm": 4.419002056121826,
      "learning_rate": 7.653560614931015e-06,
      "loss": 1.6856,
      "step": 13880
    },
    {
      "epoch": 2.3628476652207198,
      "grad_norm": 5.0171217918396,
      "learning_rate": 7.64622253531377e-06,
      "loss": 1.7392,
      "step": 13890
    },
    {
      "epoch": 2.36454877945054,
      "grad_norm": 4.358807563781738,
      "learning_rate": 7.6388833345792e-06,
      "loss": 1.6246,
      "step": 13900
    },
    {
      "epoch": 2.3662498936803606,
      "grad_norm": 5.187094211578369,
      "learning_rate": 7.631543021640117e-06,
      "loss": 1.7443,
      "step": 13910
    },
    {
      "epoch": 2.367951007910181,
      "grad_norm": 4.929495334625244,
      "learning_rate": 7.624201605410692e-06,
      "loss": 1.7132,
      "step": 13920
    },
    {
      "epoch": 2.3696521221400015,
      "grad_norm": 4.808412075042725,
      "learning_rate": 7.616859094806427e-06,
      "loss": 1.8151,
      "step": 13930
    },
    {
      "epoch": 2.3713532363698224,
      "grad_norm": 4.2984161376953125,
      "learning_rate": 7.609515498744166e-06,
      "loss": 1.7249,
      "step": 13940
    },
    {
      "epoch": 2.373054350599643,
      "grad_norm": 6.034128189086914,
      "learning_rate": 7.602170826142058e-06,
      "loss": 1.7053,
      "step": 13950
    },
    {
      "epoch": 2.3747554648294633,
      "grad_norm": 3.8828516006469727,
      "learning_rate": 7.594825085919566e-06,
      "loss": 1.6311,
      "step": 13960
    },
    {
      "epoch": 2.3764565790592838,
      "grad_norm": 5.292279243469238,
      "learning_rate": 7.587478286997449e-06,
      "loss": 1.7548,
      "step": 13970
    },
    {
      "epoch": 2.378157693289104,
      "grad_norm": 4.623950958251953,
      "learning_rate": 7.58013043829775e-06,
      "loss": 1.7034,
      "step": 13980
    },
    {
      "epoch": 2.379858807518925,
      "grad_norm": 4.3811798095703125,
      "learning_rate": 7.572781548743785e-06,
      "loss": 1.6608,
      "step": 13990
    },
    {
      "epoch": 2.3815599217487455,
      "grad_norm": 4.47433614730835,
      "learning_rate": 7.5654316272601425e-06,
      "loss": 1.7934,
      "step": 14000
    },
    {
      "epoch": 2.383261035978566,
      "grad_norm": 4.668087005615234,
      "learning_rate": 7.558080682772654e-06,
      "loss": 1.6634,
      "step": 14010
    },
    {
      "epoch": 2.3849621502083864,
      "grad_norm": 4.859496593475342,
      "learning_rate": 7.550728724208397e-06,
      "loss": 1.6277,
      "step": 14020
    },
    {
      "epoch": 2.386663264438207,
      "grad_norm": 5.9143500328063965,
      "learning_rate": 7.543375760495687e-06,
      "loss": 1.7545,
      "step": 14030
    },
    {
      "epoch": 2.3883643786680278,
      "grad_norm": 4.088611602783203,
      "learning_rate": 7.536021800564049e-06,
      "loss": 1.6934,
      "step": 14040
    },
    {
      "epoch": 2.390065492897848,
      "grad_norm": 4.978387355804443,
      "learning_rate": 7.528666853344227e-06,
      "loss": 1.6947,
      "step": 14050
    },
    {
      "epoch": 2.3917666071276686,
      "grad_norm": 4.899922847747803,
      "learning_rate": 7.521310927768158e-06,
      "loss": 1.681,
      "step": 14060
    },
    {
      "epoch": 2.393467721357489,
      "grad_norm": 5.146270751953125,
      "learning_rate": 7.51395403276897e-06,
      "loss": 1.6137,
      "step": 14070
    },
    {
      "epoch": 2.3951688355873095,
      "grad_norm": 5.114542484283447,
      "learning_rate": 7.506596177280966e-06,
      "loss": 1.7901,
      "step": 14080
    },
    {
      "epoch": 2.3968699498171304,
      "grad_norm": 5.798167705535889,
      "learning_rate": 7.499237370239621e-06,
      "loss": 1.7634,
      "step": 14090
    },
    {
      "epoch": 2.398571064046951,
      "grad_norm": 4.79186487197876,
      "learning_rate": 7.491877620581558e-06,
      "loss": 1.7108,
      "step": 14100
    },
    {
      "epoch": 2.4002721782767713,
      "grad_norm": 3.603715419769287,
      "learning_rate": 7.4845169372445494e-06,
      "loss": 1.7623,
      "step": 14110
    },
    {
      "epoch": 2.4019732925065918,
      "grad_norm": 4.47135591506958,
      "learning_rate": 7.4771553291675035e-06,
      "loss": 1.7133,
      "step": 14120
    },
    {
      "epoch": 2.403674406736412,
      "grad_norm": 4.575337886810303,
      "learning_rate": 7.469792805290443e-06,
      "loss": 1.6831,
      "step": 14130
    },
    {
      "epoch": 2.405375520966233,
      "grad_norm": 5.56679630279541,
      "learning_rate": 7.4624293745545135e-06,
      "loss": 1.7401,
      "step": 14140
    },
    {
      "epoch": 2.4070766351960535,
      "grad_norm": 5.321033954620361,
      "learning_rate": 7.455065045901955e-06,
      "loss": 1.6929,
      "step": 14150
    },
    {
      "epoch": 2.408777749425874,
      "grad_norm": 4.899209499359131,
      "learning_rate": 7.447699828276098e-06,
      "loss": 1.6492,
      "step": 14160
    },
    {
      "epoch": 2.4104788636556944,
      "grad_norm": 4.6105194091796875,
      "learning_rate": 7.4403337306213545e-06,
      "loss": 1.7318,
      "step": 14170
    },
    {
      "epoch": 2.412179977885515,
      "grad_norm": 5.365864276885986,
      "learning_rate": 7.4329667618832096e-06,
      "loss": 1.8249,
      "step": 14180
    },
    {
      "epoch": 2.4138810921153357,
      "grad_norm": 4.625761985778809,
      "learning_rate": 7.425598931008195e-06,
      "loss": 1.6837,
      "step": 14190
    },
    {
      "epoch": 2.415582206345156,
      "grad_norm": 5.167141914367676,
      "learning_rate": 7.4182302469439015e-06,
      "loss": 1.6767,
      "step": 14200
    },
    {
      "epoch": 2.4172833205749766,
      "grad_norm": 5.74930477142334,
      "learning_rate": 7.410860718638946e-06,
      "loss": 1.7934,
      "step": 14210
    },
    {
      "epoch": 2.418984434804797,
      "grad_norm": 5.915599346160889,
      "learning_rate": 7.403490355042979e-06,
      "loss": 1.683,
      "step": 14220
    },
    {
      "epoch": 2.4206855490346175,
      "grad_norm": 4.36338472366333,
      "learning_rate": 7.396119165106659e-06,
      "loss": 1.6917,
      "step": 14230
    },
    {
      "epoch": 2.4223866632644384,
      "grad_norm": 4.530436992645264,
      "learning_rate": 7.388747157781655e-06,
      "loss": 1.6475,
      "step": 14240
    },
    {
      "epoch": 2.424087777494259,
      "grad_norm": 3.656707525253296,
      "learning_rate": 7.381374342020617e-06,
      "loss": 1.7279,
      "step": 14250
    },
    {
      "epoch": 2.4257888917240793,
      "grad_norm": 4.894069194793701,
      "learning_rate": 7.374000726777191e-06,
      "loss": 1.6692,
      "step": 14260
    },
    {
      "epoch": 2.4274900059538997,
      "grad_norm": 4.92869758605957,
      "learning_rate": 7.366626321005984e-06,
      "loss": 1.7126,
      "step": 14270
    },
    {
      "epoch": 2.42919112018372,
      "grad_norm": 5.302714824676514,
      "learning_rate": 7.359251133662563e-06,
      "loss": 1.6418,
      "step": 14280
    },
    {
      "epoch": 2.430892234413541,
      "grad_norm": 4.040475845336914,
      "learning_rate": 7.3518751737034536e-06,
      "loss": 1.7015,
      "step": 14290
    },
    {
      "epoch": 2.4325933486433615,
      "grad_norm": 3.626659631729126,
      "learning_rate": 7.3444984500861075e-06,
      "loss": 1.7544,
      "step": 14300
    },
    {
      "epoch": 2.434294462873182,
      "grad_norm": 4.41288423538208,
      "learning_rate": 7.337120971768914e-06,
      "loss": 1.7394,
      "step": 14310
    },
    {
      "epoch": 2.4359955771030024,
      "grad_norm": 4.458189964294434,
      "learning_rate": 7.329742747711171e-06,
      "loss": 1.8029,
      "step": 14320
    },
    {
      "epoch": 2.437696691332823,
      "grad_norm": 4.584317207336426,
      "learning_rate": 7.322363786873085e-06,
      "loss": 1.7365,
      "step": 14330
    },
    {
      "epoch": 2.4393978055626437,
      "grad_norm": 5.787274360656738,
      "learning_rate": 7.314984098215762e-06,
      "loss": 1.7295,
      "step": 14340
    },
    {
      "epoch": 2.441098919792464,
      "grad_norm": 4.288479328155518,
      "learning_rate": 7.307603690701184e-06,
      "loss": 1.749,
      "step": 14350
    },
    {
      "epoch": 2.4428000340222846,
      "grad_norm": 5.697612762451172,
      "learning_rate": 7.30022257329221e-06,
      "loss": 1.683,
      "step": 14360
    },
    {
      "epoch": 2.444501148252105,
      "grad_norm": 5.400142669677734,
      "learning_rate": 7.292840754952561e-06,
      "loss": 1.6881,
      "step": 14370
    },
    {
      "epoch": 2.4462022624819255,
      "grad_norm": 4.963900566101074,
      "learning_rate": 7.28545824464681e-06,
      "loss": 1.6204,
      "step": 14380
    },
    {
      "epoch": 2.4479033767117464,
      "grad_norm": 3.567223310470581,
      "learning_rate": 7.278075051340368e-06,
      "loss": 1.6629,
      "step": 14390
    },
    {
      "epoch": 2.449604490941567,
      "grad_norm": 4.4511494636535645,
      "learning_rate": 7.2706911839994764e-06,
      "loss": 1.7245,
      "step": 14400
    },
    {
      "epoch": 2.4513056051713873,
      "grad_norm": 5.4482622146606445,
      "learning_rate": 7.263306651591197e-06,
      "loss": 1.612,
      "step": 14410
    },
    {
      "epoch": 2.4530067194012077,
      "grad_norm": 4.920886516571045,
      "learning_rate": 7.2559214630833945e-06,
      "loss": 1.6784,
      "step": 14420
    },
    {
      "epoch": 2.454707833631028,
      "grad_norm": 4.329588890075684,
      "learning_rate": 7.248535627444738e-06,
      "loss": 1.8066,
      "step": 14430
    },
    {
      "epoch": 2.456408947860849,
      "grad_norm": 5.034507751464844,
      "learning_rate": 7.241149153644675e-06,
      "loss": 1.7325,
      "step": 14440
    },
    {
      "epoch": 2.4581100620906695,
      "grad_norm": 4.867656230926514,
      "learning_rate": 7.233762050653434e-06,
      "loss": 1.6242,
      "step": 14450
    },
    {
      "epoch": 2.45981117632049,
      "grad_norm": 4.606437683105469,
      "learning_rate": 7.226374327442e-06,
      "loss": 1.6645,
      "step": 14460
    },
    {
      "epoch": 2.4615122905503104,
      "grad_norm": 4.166538715362549,
      "learning_rate": 7.2189859929821184e-06,
      "loss": 1.7196,
      "step": 14470
    },
    {
      "epoch": 2.463213404780131,
      "grad_norm": 5.104836940765381,
      "learning_rate": 7.211597056246274e-06,
      "loss": 1.7106,
      "step": 14480
    },
    {
      "epoch": 2.4649145190099517,
      "grad_norm": 4.810766696929932,
      "learning_rate": 7.2042075262076865e-06,
      "loss": 1.7215,
      "step": 14490
    },
    {
      "epoch": 2.466615633239772,
      "grad_norm": 4.952357292175293,
      "learning_rate": 7.196817411840287e-06,
      "loss": 1.7115,
      "step": 14500
    },
    {
      "epoch": 2.4683167474695926,
      "grad_norm": 4.354887008666992,
      "learning_rate": 7.189426722118726e-06,
      "loss": 1.6942,
      "step": 14510
    },
    {
      "epoch": 2.470017861699413,
      "grad_norm": 4.45330810546875,
      "learning_rate": 7.18203546601835e-06,
      "loss": 1.6784,
      "step": 14520
    },
    {
      "epoch": 2.4717189759292335,
      "grad_norm": 4.8778300285339355,
      "learning_rate": 7.1746436525151886e-06,
      "loss": 1.6435,
      "step": 14530
    },
    {
      "epoch": 2.4734200901590544,
      "grad_norm": 5.075737476348877,
      "learning_rate": 7.167251290585956e-06,
      "loss": 1.7026,
      "step": 14540
    },
    {
      "epoch": 2.475121204388875,
      "grad_norm": 3.924260377883911,
      "learning_rate": 7.159858389208025e-06,
      "loss": 1.6816,
      "step": 14550
    },
    {
      "epoch": 2.4768223186186953,
      "grad_norm": 5.6758012771606445,
      "learning_rate": 7.152464957359426e-06,
      "loss": 1.7027,
      "step": 14560
    },
    {
      "epoch": 2.4785234328485157,
      "grad_norm": 4.747528553009033,
      "learning_rate": 7.145071004018839e-06,
      "loss": 1.7251,
      "step": 14570
    },
    {
      "epoch": 2.480224547078336,
      "grad_norm": 5.045699596405029,
      "learning_rate": 7.137676538165569e-06,
      "loss": 1.5223,
      "step": 14580
    },
    {
      "epoch": 2.481925661308157,
      "grad_norm": 5.234375476837158,
      "learning_rate": 7.13028156877955e-06,
      "loss": 1.7245,
      "step": 14590
    },
    {
      "epoch": 2.4836267755379775,
      "grad_norm": 4.646345138549805,
      "learning_rate": 7.122886104841322e-06,
      "loss": 1.7356,
      "step": 14600
    },
    {
      "epoch": 2.485327889767798,
      "grad_norm": 5.123286247253418,
      "learning_rate": 7.115490155332031e-06,
      "loss": 1.7379,
      "step": 14610
    },
    {
      "epoch": 2.4870290039976184,
      "grad_norm": 4.82652473449707,
      "learning_rate": 7.108093729233406e-06,
      "loss": 1.6862,
      "step": 14620
    },
    {
      "epoch": 2.488730118227439,
      "grad_norm": 4.396784782409668,
      "learning_rate": 7.100696835527762e-06,
      "loss": 1.7695,
      "step": 14630
    },
    {
      "epoch": 2.4904312324572597,
      "grad_norm": 4.29108190536499,
      "learning_rate": 7.093299483197979e-06,
      "loss": 1.6293,
      "step": 14640
    },
    {
      "epoch": 2.49213234668708,
      "grad_norm": 5.77024507522583,
      "learning_rate": 7.085901681227492e-06,
      "loss": 1.6403,
      "step": 14650
    },
    {
      "epoch": 2.4938334609169006,
      "grad_norm": 5.397991180419922,
      "learning_rate": 7.078503438600283e-06,
      "loss": 1.7527,
      "step": 14660
    },
    {
      "epoch": 2.495534575146721,
      "grad_norm": 4.551114559173584,
      "learning_rate": 7.07110476430087e-06,
      "loss": 1.7193,
      "step": 14670
    },
    {
      "epoch": 2.4972356893765415,
      "grad_norm": 4.737240791320801,
      "learning_rate": 7.0637056673142985e-06,
      "loss": 1.6529,
      "step": 14680
    },
    {
      "epoch": 2.4989368036063624,
      "grad_norm": 5.69704008102417,
      "learning_rate": 7.05630615662612e-06,
      "loss": 1.6794,
      "step": 14690
    },
    {
      "epoch": 2.500637917836183,
      "grad_norm": 4.747729778289795,
      "learning_rate": 7.048906241222392e-06,
      "loss": 1.7111,
      "step": 14700
    },
    {
      "epoch": 2.5023390320660033,
      "grad_norm": 5.632922649383545,
      "learning_rate": 7.0415059300896664e-06,
      "loss": 1.744,
      "step": 14710
    },
    {
      "epoch": 2.5040401462958237,
      "grad_norm": 5.497499465942383,
      "learning_rate": 7.03410523221497e-06,
      "loss": 1.7957,
      "step": 14720
    },
    {
      "epoch": 2.505741260525644,
      "grad_norm": 4.994222164154053,
      "learning_rate": 7.026704156585805e-06,
      "loss": 1.655,
      "step": 14730
    },
    {
      "epoch": 2.507442374755465,
      "grad_norm": 5.27976131439209,
      "learning_rate": 7.019302712190128e-06,
      "loss": 1.6606,
      "step": 14740
    },
    {
      "epoch": 2.5091434889852855,
      "grad_norm": 5.364691734313965,
      "learning_rate": 7.011900908016344e-06,
      "loss": 1.7161,
      "step": 14750
    },
    {
      "epoch": 2.510844603215106,
      "grad_norm": 5.166101455688477,
      "learning_rate": 7.004498753053299e-06,
      "loss": 1.7393,
      "step": 14760
    },
    {
      "epoch": 2.5125457174449264,
      "grad_norm": 4.633950710296631,
      "learning_rate": 6.99709625629026e-06,
      "loss": 1.6672,
      "step": 14770
    },
    {
      "epoch": 2.514246831674747,
      "grad_norm": 4.659962177276611,
      "learning_rate": 6.9896934267169124e-06,
      "loss": 1.7071,
      "step": 14780
    },
    {
      "epoch": 2.5159479459045677,
      "grad_norm": 3.876272201538086,
      "learning_rate": 6.982290273323339e-06,
      "loss": 1.703,
      "step": 14790
    },
    {
      "epoch": 2.517649060134388,
      "grad_norm": 5.550055503845215,
      "learning_rate": 6.974886805100029e-06,
      "loss": 1.7049,
      "step": 14800
    },
    {
      "epoch": 2.5193501743642086,
      "grad_norm": 5.260471820831299,
      "learning_rate": 6.967483031037843e-06,
      "loss": 1.7017,
      "step": 14810
    },
    {
      "epoch": 2.521051288594029,
      "grad_norm": 4.32659387588501,
      "learning_rate": 6.960078960128013e-06,
      "loss": 1.6116,
      "step": 14820
    },
    {
      "epoch": 2.5227524028238495,
      "grad_norm": 6.853899002075195,
      "learning_rate": 6.952674601362141e-06,
      "loss": 1.6228,
      "step": 14830
    },
    {
      "epoch": 2.5244535170536704,
      "grad_norm": 4.765081405639648,
      "learning_rate": 6.945269963732167e-06,
      "loss": 1.684,
      "step": 14840
    },
    {
      "epoch": 2.526154631283491,
      "grad_norm": 3.684476375579834,
      "learning_rate": 6.937865056230376e-06,
      "loss": 1.7092,
      "step": 14850
    },
    {
      "epoch": 2.5278557455133113,
      "grad_norm": 6.573903560638428,
      "learning_rate": 6.930459887849382e-06,
      "loss": 1.6655,
      "step": 14860
    },
    {
      "epoch": 2.5295568597431317,
      "grad_norm": 4.969848155975342,
      "learning_rate": 6.923054467582111e-06,
      "loss": 1.6628,
      "step": 14870
    },
    {
      "epoch": 2.531257973972952,
      "grad_norm": 4.645257949829102,
      "learning_rate": 6.915648804421799e-06,
      "loss": 1.6335,
      "step": 14880
    },
    {
      "epoch": 2.532959088202773,
      "grad_norm": 5.171104907989502,
      "learning_rate": 6.908242907361974e-06,
      "loss": 1.6676,
      "step": 14890
    },
    {
      "epoch": 2.5346602024325935,
      "grad_norm": 5.349738121032715,
      "learning_rate": 6.900836785396449e-06,
      "loss": 1.6593,
      "step": 14900
    },
    {
      "epoch": 2.536361316662414,
      "grad_norm": 5.259891033172607,
      "learning_rate": 6.893430447519312e-06,
      "loss": 1.6858,
      "step": 14910
    },
    {
      "epoch": 2.5380624308922344,
      "grad_norm": 4.558155059814453,
      "learning_rate": 6.8860239027249114e-06,
      "loss": 1.6901,
      "step": 14920
    },
    {
      "epoch": 2.539763545122055,
      "grad_norm": 5.312442779541016,
      "learning_rate": 6.878617160007848e-06,
      "loss": 1.6492,
      "step": 14930
    },
    {
      "epoch": 2.5414646593518757,
      "grad_norm": 5.157334327697754,
      "learning_rate": 6.871210228362961e-06,
      "loss": 1.6853,
      "step": 14940
    },
    {
      "epoch": 2.543165773581696,
      "grad_norm": 4.31597375869751,
      "learning_rate": 6.86380311678532e-06,
      "loss": 1.687,
      "step": 14950
    },
    {
      "epoch": 2.5448668878115166,
      "grad_norm": 4.309648036956787,
      "learning_rate": 6.856395834270214e-06,
      "loss": 1.6711,
      "step": 14960
    },
    {
      "epoch": 2.546568002041337,
      "grad_norm": 5.050992012023926,
      "learning_rate": 6.848988389813137e-06,
      "loss": 1.6623,
      "step": 14970
    },
    {
      "epoch": 2.5482691162711575,
      "grad_norm": 4.8922319412231445,
      "learning_rate": 6.841580792409787e-06,
      "loss": 1.6639,
      "step": 14980
    },
    {
      "epoch": 2.5499702305009784,
      "grad_norm": 4.5712995529174805,
      "learning_rate": 6.834173051056034e-06,
      "loss": 1.6761,
      "step": 14990
    },
    {
      "epoch": 2.551671344730799,
      "grad_norm": 4.379843235015869,
      "learning_rate": 6.826765174747939e-06,
      "loss": 1.6514,
      "step": 15000
    },
    {
      "epoch": 2.5533724589606193,
      "grad_norm": 4.673671722412109,
      "learning_rate": 6.819357172481715e-06,
      "loss": 1.7389,
      "step": 15010
    },
    {
      "epoch": 2.5550735731904397,
      "grad_norm": 5.1559367179870605,
      "learning_rate": 6.811949053253731e-06,
      "loss": 1.5973,
      "step": 15020
    },
    {
      "epoch": 2.55677468742026,
      "grad_norm": 4.359671592712402,
      "learning_rate": 6.804540826060503e-06,
      "loss": 1.7217,
      "step": 15030
    },
    {
      "epoch": 2.558475801650081,
      "grad_norm": 5.706950664520264,
      "learning_rate": 6.7971324998986706e-06,
      "loss": 1.625,
      "step": 15040
    },
    {
      "epoch": 2.5601769158799015,
      "grad_norm": 4.20098876953125,
      "learning_rate": 6.789724083764997e-06,
      "loss": 1.7012,
      "step": 15050
    },
    {
      "epoch": 2.561878030109722,
      "grad_norm": 5.58827543258667,
      "learning_rate": 6.782315586656356e-06,
      "loss": 1.6924,
      "step": 15060
    },
    {
      "epoch": 2.5635791443395424,
      "grad_norm": 4.752255439758301,
      "learning_rate": 6.774907017569722e-06,
      "loss": 1.6697,
      "step": 15070
    },
    {
      "epoch": 2.565280258569363,
      "grad_norm": 5.409000873565674,
      "learning_rate": 6.767498385502147e-06,
      "loss": 1.7054,
      "step": 15080
    },
    {
      "epoch": 2.5669813727991837,
      "grad_norm": 4.958235263824463,
      "learning_rate": 6.76008969945077e-06,
      "loss": 1.7299,
      "step": 15090
    },
    {
      "epoch": 2.568682487029004,
      "grad_norm": 3.6053428649902344,
      "learning_rate": 6.752680968412789e-06,
      "loss": 1.611,
      "step": 15100
    },
    {
      "epoch": 2.5703836012588246,
      "grad_norm": 4.998643398284912,
      "learning_rate": 6.745272201385462e-06,
      "loss": 1.7689,
      "step": 15110
    },
    {
      "epoch": 2.572084715488645,
      "grad_norm": 4.558544635772705,
      "learning_rate": 6.737863407366085e-06,
      "loss": 1.6389,
      "step": 15120
    },
    {
      "epoch": 2.5737858297184655,
      "grad_norm": 4.5095109939575195,
      "learning_rate": 6.7304545953519895e-06,
      "loss": 1.7371,
      "step": 15130
    },
    {
      "epoch": 2.5754869439482864,
      "grad_norm": 4.6984148025512695,
      "learning_rate": 6.72304577434053e-06,
      "loss": 1.622,
      "step": 15140
    },
    {
      "epoch": 2.577188058178107,
      "grad_norm": 5.407587051391602,
      "learning_rate": 6.715636953329072e-06,
      "loss": 1.7879,
      "step": 15150
    },
    {
      "epoch": 2.5788891724079273,
      "grad_norm": 5.009973049163818,
      "learning_rate": 6.7082281413149776e-06,
      "loss": 1.7139,
      "step": 15160
    },
    {
      "epoch": 2.5805902866377477,
      "grad_norm": 5.410165309906006,
      "learning_rate": 6.700819347295599e-06,
      "loss": 1.6792,
      "step": 15170
    },
    {
      "epoch": 2.582291400867568,
      "grad_norm": 5.430539131164551,
      "learning_rate": 6.693410580268271e-06,
      "loss": 1.7042,
      "step": 15180
    },
    {
      "epoch": 2.583992515097389,
      "grad_norm": 5.337528705596924,
      "learning_rate": 6.686001849230292e-06,
      "loss": 1.669,
      "step": 15190
    },
    {
      "epoch": 2.5856936293272095,
      "grad_norm": 4.983978748321533,
      "learning_rate": 6.6785931631789134e-06,
      "loss": 1.733,
      "step": 15200
    },
    {
      "epoch": 2.58739474355703,
      "grad_norm": 4.920320510864258,
      "learning_rate": 6.6711845311113404e-06,
      "loss": 1.6872,
      "step": 15210
    },
    {
      "epoch": 2.5890958577868504,
      "grad_norm": 4.505009174346924,
      "learning_rate": 6.663775962024704e-06,
      "loss": 1.6061,
      "step": 15220
    },
    {
      "epoch": 2.590796972016671,
      "grad_norm": 4.6945905685424805,
      "learning_rate": 6.656367464916064e-06,
      "loss": 1.7347,
      "step": 15230
    },
    {
      "epoch": 2.5924980862464917,
      "grad_norm": 4.4238433837890625,
      "learning_rate": 6.648959048782391e-06,
      "loss": 1.7022,
      "step": 15240
    },
    {
      "epoch": 2.594199200476312,
      "grad_norm": 5.2501044273376465,
      "learning_rate": 6.641550722620559e-06,
      "loss": 1.6146,
      "step": 15250
    },
    {
      "epoch": 2.5959003147061326,
      "grad_norm": 6.0587239265441895,
      "learning_rate": 6.63414249542733e-06,
      "loss": 1.5737,
      "step": 15260
    },
    {
      "epoch": 2.597601428935953,
      "grad_norm": 4.9692254066467285,
      "learning_rate": 6.626734376199349e-06,
      "loss": 1.736,
      "step": 15270
    },
    {
      "epoch": 2.5993025431657735,
      "grad_norm": 6.877859115600586,
      "learning_rate": 6.619326373933124e-06,
      "loss": 1.671,
      "step": 15280
    },
    {
      "epoch": 2.6010036573955944,
      "grad_norm": 5.190441131591797,
      "learning_rate": 6.6119184976250254e-06,
      "loss": 1.6989,
      "step": 15290
    },
    {
      "epoch": 2.602704771625415,
      "grad_norm": 5.681378364562988,
      "learning_rate": 6.604510756271275e-06,
      "loss": 1.7095,
      "step": 15300
    },
    {
      "epoch": 2.6044058858552352,
      "grad_norm": 5.812134265899658,
      "learning_rate": 6.597103158867924e-06,
      "loss": 1.7042,
      "step": 15310
    },
    {
      "epoch": 2.6061070000850557,
      "grad_norm": 5.831223011016846,
      "learning_rate": 6.589695714410848e-06,
      "loss": 1.6901,
      "step": 15320
    },
    {
      "epoch": 2.607808114314876,
      "grad_norm": 5.0628228187561035,
      "learning_rate": 6.582288431895742e-06,
      "loss": 1.745,
      "step": 15330
    },
    {
      "epoch": 2.609509228544697,
      "grad_norm": 4.264681339263916,
      "learning_rate": 6.5748813203180985e-06,
      "loss": 1.6821,
      "step": 15340
    },
    {
      "epoch": 2.6112103427745175,
      "grad_norm": 5.896673202514648,
      "learning_rate": 6.567474388673212e-06,
      "loss": 1.7215,
      "step": 15350
    },
    {
      "epoch": 2.612911457004338,
      "grad_norm": 4.254663944244385,
      "learning_rate": 6.560067645956149e-06,
      "loss": 1.6865,
      "step": 15360
    },
    {
      "epoch": 2.6146125712341584,
      "grad_norm": 5.108974933624268,
      "learning_rate": 6.552661101161749e-06,
      "loss": 1.6408,
      "step": 15370
    },
    {
      "epoch": 2.616313685463979,
      "grad_norm": 4.906005859375,
      "learning_rate": 6.545254763284613e-06,
      "loss": 1.6504,
      "step": 15380
    },
    {
      "epoch": 2.6180147996937997,
      "grad_norm": 5.226259708404541,
      "learning_rate": 6.537848641319088e-06,
      "loss": 1.6527,
      "step": 15390
    },
    {
      "epoch": 2.61971591392362,
      "grad_norm": 4.990200042724609,
      "learning_rate": 6.530442744259263e-06,
      "loss": 1.704,
      "step": 15400
    },
    {
      "epoch": 2.6214170281534406,
      "grad_norm": 5.2751874923706055,
      "learning_rate": 6.52303708109895e-06,
      "loss": 1.6663,
      "step": 15410
    },
    {
      "epoch": 2.623118142383261,
      "grad_norm": 5.105523109436035,
      "learning_rate": 6.51563166083168e-06,
      "loss": 1.6694,
      "step": 15420
    },
    {
      "epoch": 2.6248192566130815,
      "grad_norm": 5.40707540512085,
      "learning_rate": 6.508226492450685e-06,
      "loss": 1.6771,
      "step": 15430
    },
    {
      "epoch": 2.6265203708429024,
      "grad_norm": 4.876675128936768,
      "learning_rate": 6.500821584948895e-06,
      "loss": 1.6224,
      "step": 15440
    },
    {
      "epoch": 2.628221485072723,
      "grad_norm": 6.009464263916016,
      "learning_rate": 6.4941573986814035e-06,
      "loss": 1.7189,
      "step": 15450
    },
    {
      "epoch": 2.6299225993025432,
      "grad_norm": 5.135744571685791,
      "learning_rate": 6.486753011624476e-06,
      "loss": 1.6572,
      "step": 15460
    },
    {
      "epoch": 2.6316237135323637,
      "grad_norm": 4.236342906951904,
      "learning_rate": 6.479348911524416e-06,
      "loss": 1.7756,
      "step": 15470
    },
    {
      "epoch": 2.633324827762184,
      "grad_norm": 5.710334300994873,
      "learning_rate": 6.471945107372854e-06,
      "loss": 1.6689,
      "step": 15480
    },
    {
      "epoch": 2.635025941992005,
      "grad_norm": 4.605166435241699,
      "learning_rate": 6.464541608161061e-06,
      "loss": 1.7447,
      "step": 15490
    },
    {
      "epoch": 2.6367270562218255,
      "grad_norm": 4.071094512939453,
      "learning_rate": 6.457138422879936e-06,
      "loss": 1.7051,
      "step": 15500
    },
    {
      "epoch": 2.638428170451646,
      "grad_norm": 4.874546527862549,
      "learning_rate": 6.449735560520004e-06,
      "loss": 1.6924,
      "step": 15510
    },
    {
      "epoch": 2.6401292846814663,
      "grad_norm": 6.157898426055908,
      "learning_rate": 6.442333030071392e-06,
      "loss": 1.7024,
      "step": 15520
    },
    {
      "epoch": 2.641830398911287,
      "grad_norm": 6.2554931640625,
      "learning_rate": 6.434930840523823e-06,
      "loss": 1.6483,
      "step": 15530
    },
    {
      "epoch": 2.6435315131411077,
      "grad_norm": 4.651872634887695,
      "learning_rate": 6.427529000866609e-06,
      "loss": 1.6866,
      "step": 15540
    },
    {
      "epoch": 2.6452326273709277,
      "grad_norm": 4.448643684387207,
      "learning_rate": 6.420127520088638e-06,
      "loss": 1.7347,
      "step": 15550
    },
    {
      "epoch": 2.6469337416007486,
      "grad_norm": 5.3721137046813965,
      "learning_rate": 6.412726407178352e-06,
      "loss": 1.738,
      "step": 15560
    },
    {
      "epoch": 2.648634855830569,
      "grad_norm": 4.654369831085205,
      "learning_rate": 6.405325671123765e-06,
      "loss": 1.608,
      "step": 15570
    },
    {
      "epoch": 2.6503359700603895,
      "grad_norm": 4.932215690612793,
      "learning_rate": 6.397925320912418e-06,
      "loss": 1.7595,
      "step": 15580
    },
    {
      "epoch": 2.6520370842902103,
      "grad_norm": 4.843132972717285,
      "learning_rate": 6.39052536553139e-06,
      "loss": 1.6511,
      "step": 15590
    },
    {
      "epoch": 2.6537381985200303,
      "grad_norm": 4.947632789611816,
      "learning_rate": 6.3831258139672755e-06,
      "loss": 1.6544,
      "step": 15600
    },
    {
      "epoch": 2.6554393127498512,
      "grad_norm": 4.492228031158447,
      "learning_rate": 6.375726675206181e-06,
      "loss": 1.6936,
      "step": 15610
    },
    {
      "epoch": 2.6571404269796717,
      "grad_norm": 4.530820846557617,
      "learning_rate": 6.368327958233718e-06,
      "loss": 1.7186,
      "step": 15620
    },
    {
      "epoch": 2.658841541209492,
      "grad_norm": 5.303435325622559,
      "learning_rate": 6.360929672034979e-06,
      "loss": 1.7346,
      "step": 15630
    },
    {
      "epoch": 2.660542655439313,
      "grad_norm": 4.329684734344482,
      "learning_rate": 6.3535318255945315e-06,
      "loss": 1.6931,
      "step": 15640
    },
    {
      "epoch": 2.662243769669133,
      "grad_norm": 4.740088939666748,
      "learning_rate": 6.346134427896415e-06,
      "loss": 1.7554,
      "step": 15650
    },
    {
      "epoch": 2.663944883898954,
      "grad_norm": 4.414624214172363,
      "learning_rate": 6.338737487924118e-06,
      "loss": 1.6209,
      "step": 15660
    },
    {
      "epoch": 2.6656459981287743,
      "grad_norm": 5.261606693267822,
      "learning_rate": 6.331341014660575e-06,
      "loss": 1.6882,
      "step": 15670
    },
    {
      "epoch": 2.667347112358595,
      "grad_norm": 4.723742961883545,
      "learning_rate": 6.323945017088165e-06,
      "loss": 1.6296,
      "step": 15680
    },
    {
      "epoch": 2.6690482265884157,
      "grad_norm": 4.809913158416748,
      "learning_rate": 6.316549504188669e-06,
      "loss": 1.7024,
      "step": 15690
    },
    {
      "epoch": 2.6707493408182357,
      "grad_norm": 5.183937072753906,
      "learning_rate": 6.309154484943295e-06,
      "loss": 1.7459,
      "step": 15700
    },
    {
      "epoch": 2.6724504550480566,
      "grad_norm": 4.400472164154053,
      "learning_rate": 6.301759968332643e-06,
      "loss": 1.659,
      "step": 15710
    },
    {
      "epoch": 2.674151569277877,
      "grad_norm": 4.746430397033691,
      "learning_rate": 6.294365963336708e-06,
      "loss": 1.7923,
      "step": 15720
    },
    {
      "epoch": 2.6758526835076974,
      "grad_norm": 5.050037860870361,
      "learning_rate": 6.2869724789348605e-06,
      "loss": 1.591,
      "step": 15730
    },
    {
      "epoch": 2.6775537977375183,
      "grad_norm": 4.194572448730469,
      "learning_rate": 6.279579524105841e-06,
      "loss": 1.7503,
      "step": 15740
    },
    {
      "epoch": 2.6792549119673383,
      "grad_norm": 3.9525299072265625,
      "learning_rate": 6.272187107827744e-06,
      "loss": 1.6488,
      "step": 15750
    },
    {
      "epoch": 2.6809560261971592,
      "grad_norm": 4.082591533660889,
      "learning_rate": 6.264795239078013e-06,
      "loss": 1.6269,
      "step": 15760
    },
    {
      "epoch": 2.6826571404269797,
      "grad_norm": 4.713684558868408,
      "learning_rate": 6.257403926833425e-06,
      "loss": 1.6432,
      "step": 15770
    },
    {
      "epoch": 2.6843582546568,
      "grad_norm": 4.3438262939453125,
      "learning_rate": 6.2500131800700795e-06,
      "loss": 1.6673,
      "step": 15780
    },
    {
      "epoch": 2.686059368886621,
      "grad_norm": 4.460534572601318,
      "learning_rate": 6.242623007763395e-06,
      "loss": 1.663,
      "step": 15790
    },
    {
      "epoch": 2.687760483116441,
      "grad_norm": 5.520261287689209,
      "learning_rate": 6.235233418888085e-06,
      "loss": 1.7469,
      "step": 15800
    },
    {
      "epoch": 2.689461597346262,
      "grad_norm": 4.027762413024902,
      "learning_rate": 6.227844422418161e-06,
      "loss": 1.6448,
      "step": 15810
    },
    {
      "epoch": 2.6911627115760823,
      "grad_norm": 6.53219747543335,
      "learning_rate": 6.22045602732691e-06,
      "loss": 1.6534,
      "step": 15820
    },
    {
      "epoch": 2.6928638258059028,
      "grad_norm": 4.947280406951904,
      "learning_rate": 6.213068242586891e-06,
      "loss": 1.766,
      "step": 15830
    },
    {
      "epoch": 2.6945649400357237,
      "grad_norm": 4.839651584625244,
      "learning_rate": 6.20568107716992e-06,
      "loss": 1.6713,
      "step": 15840
    },
    {
      "epoch": 2.6962660542655437,
      "grad_norm": 4.531035900115967,
      "learning_rate": 6.198294540047068e-06,
      "loss": 1.7435,
      "step": 15850
    },
    {
      "epoch": 2.6979671684953646,
      "grad_norm": 5.5740461349487305,
      "learning_rate": 6.190908640188632e-06,
      "loss": 1.5887,
      "step": 15860
    },
    {
      "epoch": 2.699668282725185,
      "grad_norm": 5.6536664962768555,
      "learning_rate": 6.183523386564143e-06,
      "loss": 1.685,
      "step": 15870
    },
    {
      "epoch": 2.7013693969550054,
      "grad_norm": 4.010603427886963,
      "learning_rate": 6.176138788142345e-06,
      "loss": 1.6643,
      "step": 15880
    },
    {
      "epoch": 2.7030705111848263,
      "grad_norm": 5.687734603881836,
      "learning_rate": 6.168754853891184e-06,
      "loss": 1.6053,
      "step": 15890
    },
    {
      "epoch": 2.7047716254146463,
      "grad_norm": 5.492772579193115,
      "learning_rate": 6.1613715927778e-06,
      "loss": 1.6483,
      "step": 15900
    },
    {
      "epoch": 2.706472739644467,
      "grad_norm": 4.573555946350098,
      "learning_rate": 6.153989013768522e-06,
      "loss": 1.7613,
      "step": 15910
    },
    {
      "epoch": 2.7081738538742877,
      "grad_norm": 4.435306072235107,
      "learning_rate": 6.146607125828844e-06,
      "loss": 1.5846,
      "step": 15920
    },
    {
      "epoch": 2.709874968104108,
      "grad_norm": 5.207588195800781,
      "learning_rate": 6.139225937923423e-06,
      "loss": 1.5914,
      "step": 15930
    },
    {
      "epoch": 2.711576082333929,
      "grad_norm": 5.92121696472168,
      "learning_rate": 6.1318454590160624e-06,
      "loss": 1.7785,
      "step": 15940
    },
    {
      "epoch": 2.713277196563749,
      "grad_norm": 6.585207939147949,
      "learning_rate": 6.124465698069706e-06,
      "loss": 1.652,
      "step": 15950
    },
    {
      "epoch": 2.71497831079357,
      "grad_norm": 5.72299337387085,
      "learning_rate": 6.1170866640464345e-06,
      "loss": 1.7755,
      "step": 15960
    },
    {
      "epoch": 2.7166794250233903,
      "grad_norm": 4.594733715057373,
      "learning_rate": 6.109708365907434e-06,
      "loss": 1.7994,
      "step": 15970
    },
    {
      "epoch": 2.7183805392532108,
      "grad_norm": 4.8487443923950195,
      "learning_rate": 6.102330812613001e-06,
      "loss": 1.6315,
      "step": 15980
    },
    {
      "epoch": 2.7200816534830317,
      "grad_norm": 5.4819746017456055,
      "learning_rate": 6.0949540131225284e-06,
      "loss": 1.7589,
      "step": 15990
    },
    {
      "epoch": 2.7217827677128517,
      "grad_norm": 6.045997619628906,
      "learning_rate": 6.087577976394493e-06,
      "loss": 1.6223,
      "step": 16000
    },
    {
      "epoch": 2.7234838819426725,
      "grad_norm": 5.426220417022705,
      "learning_rate": 6.080202711386444e-06,
      "loss": 1.6925,
      "step": 16010
    },
    {
      "epoch": 2.725184996172493,
      "grad_norm": 5.517510890960693,
      "learning_rate": 6.072828227055e-06,
      "loss": 1.7678,
      "step": 16020
    },
    {
      "epoch": 2.7268861104023134,
      "grad_norm": 4.463784217834473,
      "learning_rate": 6.065454532355822e-06,
      "loss": 1.7528,
      "step": 16030
    },
    {
      "epoch": 2.7285872246321343,
      "grad_norm": 4.352672100067139,
      "learning_rate": 6.058081636243617e-06,
      "loss": 1.7728,
      "step": 16040
    },
    {
      "epoch": 2.7302883388619543,
      "grad_norm": 4.805075168609619,
      "learning_rate": 6.050709547672122e-06,
      "loss": 1.6959,
      "step": 16050
    },
    {
      "epoch": 2.731989453091775,
      "grad_norm": 5.226137161254883,
      "learning_rate": 6.043338275594094e-06,
      "loss": 1.5511,
      "step": 16060
    },
    {
      "epoch": 2.7336905673215957,
      "grad_norm": 5.238788604736328,
      "learning_rate": 6.0359678289612935e-06,
      "loss": 1.7664,
      "step": 16070
    },
    {
      "epoch": 2.735391681551416,
      "grad_norm": 5.0616960525512695,
      "learning_rate": 6.028598216724487e-06,
      "loss": 1.6906,
      "step": 16080
    },
    {
      "epoch": 2.737092795781237,
      "grad_norm": 4.657918453216553,
      "learning_rate": 6.021229447833422e-06,
      "loss": 1.588,
      "step": 16090
    },
    {
      "epoch": 2.738793910011057,
      "grad_norm": 4.7162322998046875,
      "learning_rate": 6.0138615312368216e-06,
      "loss": 1.7453,
      "step": 16100
    },
    {
      "epoch": 2.740495024240878,
      "grad_norm": 5.221200466156006,
      "learning_rate": 6.006494475882377e-06,
      "loss": 1.8455,
      "step": 16110
    },
    {
      "epoch": 2.7421961384706983,
      "grad_norm": 5.481096267700195,
      "learning_rate": 5.999128290716731e-06,
      "loss": 1.6014,
      "step": 16120
    },
    {
      "epoch": 2.7438972527005188,
      "grad_norm": 5.632100582122803,
      "learning_rate": 5.991762984685468e-06,
      "loss": 1.6951,
      "step": 16130
    },
    {
      "epoch": 2.745598366930339,
      "grad_norm": 4.87919807434082,
      "learning_rate": 5.984398566733111e-06,
      "loss": 1.7195,
      "step": 16140
    },
    {
      "epoch": 2.7472994811601597,
      "grad_norm": 5.315734386444092,
      "learning_rate": 5.9770350458030995e-06,
      "loss": 1.7032,
      "step": 16150
    },
    {
      "epoch": 2.7490005953899805,
      "grad_norm": 5.583526611328125,
      "learning_rate": 5.969672430837784e-06,
      "loss": 1.628,
      "step": 16160
    },
    {
      "epoch": 2.750701709619801,
      "grad_norm": 4.761498928070068,
      "learning_rate": 5.962310730778415e-06,
      "loss": 1.7235,
      "step": 16170
    },
    {
      "epoch": 2.7524028238496214,
      "grad_norm": 5.210669040679932,
      "learning_rate": 5.954949954565134e-06,
      "loss": 1.7194,
      "step": 16180
    },
    {
      "epoch": 2.754103938079442,
      "grad_norm": 5.470809459686279,
      "learning_rate": 5.947590111136958e-06,
      "loss": 1.6384,
      "step": 16190
    },
    {
      "epoch": 2.7558050523092623,
      "grad_norm": 4.662988662719727,
      "learning_rate": 5.940231209431773e-06,
      "loss": 1.5649,
      "step": 16200
    },
    {
      "epoch": 2.757506166539083,
      "grad_norm": 4.8013763427734375,
      "learning_rate": 5.932873258386321e-06,
      "loss": 1.6258,
      "step": 16210
    },
    {
      "epoch": 2.7592072807689036,
      "grad_norm": 4.362658977508545,
      "learning_rate": 5.9255162669361906e-06,
      "loss": 1.7916,
      "step": 16220
    },
    {
      "epoch": 2.760908394998724,
      "grad_norm": 4.981400489807129,
      "learning_rate": 5.918160244015799e-06,
      "loss": 1.6733,
      "step": 16230
    },
    {
      "epoch": 2.7626095092285445,
      "grad_norm": 5.8743577003479,
      "learning_rate": 5.910805198558391e-06,
      "loss": 1.6994,
      "step": 16240
    },
    {
      "epoch": 2.764310623458365,
      "grad_norm": 5.567012310028076,
      "learning_rate": 5.903451139496033e-06,
      "loss": 1.6526,
      "step": 16250
    },
    {
      "epoch": 2.766011737688186,
      "grad_norm": 5.757974147796631,
      "learning_rate": 5.89609807575958e-06,
      "loss": 1.7313,
      "step": 16260
    },
    {
      "epoch": 2.7677128519180063,
      "grad_norm": 4.79555606842041,
      "learning_rate": 5.8887460162786875e-06,
      "loss": 1.7318,
      "step": 16270
    },
    {
      "epoch": 2.7694139661478268,
      "grad_norm": 5.295034408569336,
      "learning_rate": 5.881394969981784e-06,
      "loss": 1.6734,
      "step": 16280
    },
    {
      "epoch": 2.771115080377647,
      "grad_norm": 4.794110298156738,
      "learning_rate": 5.874044945796072e-06,
      "loss": 1.6553,
      "step": 16290
    },
    {
      "epoch": 2.7728161946074676,
      "grad_norm": 5.3395915031433105,
      "learning_rate": 5.866695952647512e-06,
      "loss": 1.6836,
      "step": 16300
    },
    {
      "epoch": 2.7745173088372885,
      "grad_norm": 5.3286261558532715,
      "learning_rate": 5.8593479994608165e-06,
      "loss": 1.714,
      "step": 16310
    },
    {
      "epoch": 2.776218423067109,
      "grad_norm": 4.2219648361206055,
      "learning_rate": 5.852001095159426e-06,
      "loss": 1.7005,
      "step": 16320
    },
    {
      "epoch": 2.7779195372969294,
      "grad_norm": 5.95657205581665,
      "learning_rate": 5.844655248665518e-06,
      "loss": 1.7092,
      "step": 16330
    },
    {
      "epoch": 2.77962065152675,
      "grad_norm": 5.675844192504883,
      "learning_rate": 5.837310468899974e-06,
      "loss": 1.5973,
      "step": 16340
    },
    {
      "epoch": 2.7813217657565703,
      "grad_norm": 4.944215774536133,
      "learning_rate": 5.829966764782388e-06,
      "loss": 1.7103,
      "step": 16350
    },
    {
      "epoch": 2.783022879986391,
      "grad_norm": 4.9561686515808105,
      "learning_rate": 5.822624145231046e-06,
      "loss": 1.606,
      "step": 16360
    },
    {
      "epoch": 2.7847239942162116,
      "grad_norm": 4.788738250732422,
      "learning_rate": 5.815282619162917e-06,
      "loss": 1.6993,
      "step": 16370
    },
    {
      "epoch": 2.786425108446032,
      "grad_norm": 4.6601481437683105,
      "learning_rate": 5.80794219549364e-06,
      "loss": 1.6983,
      "step": 16380
    },
    {
      "epoch": 2.7881262226758525,
      "grad_norm": 5.334542274475098,
      "learning_rate": 5.800602883137519e-06,
      "loss": 1.6107,
      "step": 16390
    },
    {
      "epoch": 2.789827336905673,
      "grad_norm": 5.247471809387207,
      "learning_rate": 5.793264691007503e-06,
      "loss": 1.7659,
      "step": 16400
    },
    {
      "epoch": 2.791528451135494,
      "grad_norm": 5.605321884155273,
      "learning_rate": 5.785927628015184e-06,
      "loss": 1.6061,
      "step": 16410
    },
    {
      "epoch": 2.7932295653653143,
      "grad_norm": 5.843311786651611,
      "learning_rate": 5.778591703070784e-06,
      "loss": 1.7109,
      "step": 16420
    },
    {
      "epoch": 2.7949306795951347,
      "grad_norm": 5.994198322296143,
      "learning_rate": 5.771256925083143e-06,
      "loss": 1.7198,
      "step": 16430
    },
    {
      "epoch": 2.796631793824955,
      "grad_norm": 4.986034393310547,
      "learning_rate": 5.763923302959703e-06,
      "loss": 1.6503,
      "step": 16440
    },
    {
      "epoch": 2.7983329080547756,
      "grad_norm": 5.742405891418457,
      "learning_rate": 5.7565908456065076e-06,
      "loss": 1.6704,
      "step": 16450
    },
    {
      "epoch": 2.8000340222845965,
      "grad_norm": 5.289145469665527,
      "learning_rate": 5.749259561928184e-06,
      "loss": 1.6697,
      "step": 16460
    },
    {
      "epoch": 2.801735136514417,
      "grad_norm": 5.038757801055908,
      "learning_rate": 5.741929460827932e-06,
      "loss": 1.6896,
      "step": 16470
    },
    {
      "epoch": 2.8034362507442374,
      "grad_norm": 4.987347602844238,
      "learning_rate": 5.7346005512075215e-06,
      "loss": 1.6926,
      "step": 16480
    },
    {
      "epoch": 2.805137364974058,
      "grad_norm": 5.577208518981934,
      "learning_rate": 5.727272841967269e-06,
      "loss": 1.6673,
      "step": 16490
    },
    {
      "epoch": 2.8068384792038783,
      "grad_norm": 5.131875038146973,
      "learning_rate": 5.719946342006035e-06,
      "loss": 1.6913,
      "step": 16500
    },
    {
      "epoch": 2.808539593433699,
      "grad_norm": 4.246416091918945,
      "learning_rate": 5.7126210602212136e-06,
      "loss": 1.6361,
      "step": 16510
    },
    {
      "epoch": 2.8102407076635196,
      "grad_norm": 5.458721160888672,
      "learning_rate": 5.7052970055087156e-06,
      "loss": 1.6904,
      "step": 16520
    },
    {
      "epoch": 2.81194182189334,
      "grad_norm": 5.210801601409912,
      "learning_rate": 5.697974186762968e-06,
      "loss": 1.6842,
      "step": 16530
    },
    {
      "epoch": 2.8136429361231605,
      "grad_norm": 5.769560813903809,
      "learning_rate": 5.69065261287689e-06,
      "loss": 1.7072,
      "step": 16540
    },
    {
      "epoch": 2.815344050352981,
      "grad_norm": 5.030094623565674,
      "learning_rate": 5.683332292741893e-06,
      "loss": 1.7412,
      "step": 16550
    },
    {
      "epoch": 2.817045164582802,
      "grad_norm": 4.414101600646973,
      "learning_rate": 5.676013235247866e-06,
      "loss": 1.6901,
      "step": 16560
    },
    {
      "epoch": 2.8187462788126223,
      "grad_norm": 3.788679361343384,
      "learning_rate": 5.668695449283159e-06,
      "loss": 1.6915,
      "step": 16570
    },
    {
      "epoch": 2.8204473930424427,
      "grad_norm": 5.092972278594971,
      "learning_rate": 5.661378943734583e-06,
      "loss": 1.7052,
      "step": 16580
    },
    {
      "epoch": 2.822148507272263,
      "grad_norm": 5.105255603790283,
      "learning_rate": 5.654063727487396e-06,
      "loss": 1.7155,
      "step": 16590
    },
    {
      "epoch": 2.8238496215020836,
      "grad_norm": 4.697558879852295,
      "learning_rate": 5.646749809425289e-06,
      "loss": 1.7101,
      "step": 16600
    },
    {
      "epoch": 2.8255507357319045,
      "grad_norm": 4.967411518096924,
      "learning_rate": 5.639437198430372e-06,
      "loss": 1.711,
      "step": 16610
    },
    {
      "epoch": 2.827251849961725,
      "grad_norm": 4.828697204589844,
      "learning_rate": 5.63212590338317e-06,
      "loss": 1.7244,
      "step": 16620
    },
    {
      "epoch": 2.8289529641915454,
      "grad_norm": 6.0997395515441895,
      "learning_rate": 5.6248159331626125e-06,
      "loss": 1.6242,
      "step": 16630
    },
    {
      "epoch": 2.830654078421366,
      "grad_norm": 5.063726425170898,
      "learning_rate": 5.617507296646014e-06,
      "loss": 1.6804,
      "step": 16640
    },
    {
      "epoch": 2.8323551926511863,
      "grad_norm": 5.528225898742676,
      "learning_rate": 5.610200002709081e-06,
      "loss": 1.6623,
      "step": 16650
    },
    {
      "epoch": 2.834056306881007,
      "grad_norm": 5.0858049392700195,
      "learning_rate": 5.602894060225876e-06,
      "loss": 1.6641,
      "step": 16660
    },
    {
      "epoch": 2.8357574211108276,
      "grad_norm": 4.781316757202148,
      "learning_rate": 5.595589478068828e-06,
      "loss": 1.7048,
      "step": 16670
    },
    {
      "epoch": 2.837458535340648,
      "grad_norm": 4.909644603729248,
      "learning_rate": 5.5882862651087116e-06,
      "loss": 1.7676,
      "step": 16680
    },
    {
      "epoch": 2.8391596495704685,
      "grad_norm": 6.090540885925293,
      "learning_rate": 5.580984430214642e-06,
      "loss": 1.7098,
      "step": 16690
    },
    {
      "epoch": 2.840860763800289,
      "grad_norm": 5.437629222869873,
      "learning_rate": 5.573683982254053e-06,
      "loss": 1.7233,
      "step": 16700
    },
    {
      "epoch": 2.84256187803011,
      "grad_norm": 5.543697357177734,
      "learning_rate": 5.566384930092706e-06,
      "loss": 1.658,
      "step": 16710
    },
    {
      "epoch": 2.8442629922599303,
      "grad_norm": 4.917662143707275,
      "learning_rate": 5.559087282594656e-06,
      "loss": 1.6179,
      "step": 16720
    },
    {
      "epoch": 2.8459641064897507,
      "grad_norm": 5.373143196105957,
      "learning_rate": 5.551791048622258e-06,
      "loss": 1.69,
      "step": 16730
    },
    {
      "epoch": 2.847665220719571,
      "grad_norm": 4.551077842712402,
      "learning_rate": 5.5444962370361495e-06,
      "loss": 1.6458,
      "step": 16740
    },
    {
      "epoch": 2.8493663349493916,
      "grad_norm": 4.875065326690674,
      "learning_rate": 5.537202856695239e-06,
      "loss": 1.7397,
      "step": 16750
    },
    {
      "epoch": 2.8510674491792125,
      "grad_norm": 5.468201637268066,
      "learning_rate": 5.5299109164567e-06,
      "loss": 1.579,
      "step": 16760
    },
    {
      "epoch": 2.852768563409033,
      "grad_norm": 4.847186088562012,
      "learning_rate": 5.522620425175955e-06,
      "loss": 1.7088,
      "step": 16770
    },
    {
      "epoch": 2.8544696776388534,
      "grad_norm": 5.318556785583496,
      "learning_rate": 5.5153313917066665e-06,
      "loss": 1.6794,
      "step": 16780
    },
    {
      "epoch": 2.856170791868674,
      "grad_norm": 5.668161869049072,
      "learning_rate": 5.508043824900728e-06,
      "loss": 1.7609,
      "step": 16790
    },
    {
      "epoch": 2.8578719060984943,
      "grad_norm": 4.532445907592773,
      "learning_rate": 5.50075773360825e-06,
      "loss": 1.854,
      "step": 16800
    },
    {
      "epoch": 2.859573020328315,
      "grad_norm": 4.807154655456543,
      "learning_rate": 5.493473126677551e-06,
      "loss": 1.7322,
      "step": 16810
    },
    {
      "epoch": 2.8612741345581356,
      "grad_norm": 4.542354106903076,
      "learning_rate": 5.486190012955153e-06,
      "loss": 1.646,
      "step": 16820
    },
    {
      "epoch": 2.862975248787956,
      "grad_norm": 5.106041431427002,
      "learning_rate": 5.4789084012857555e-06,
      "loss": 1.6552,
      "step": 16830
    },
    {
      "epoch": 2.8646763630177765,
      "grad_norm": 4.498376846313477,
      "learning_rate": 5.47162830051224e-06,
      "loss": 1.6193,
      "step": 16840
    },
    {
      "epoch": 2.866377477247597,
      "grad_norm": 5.249197483062744,
      "learning_rate": 5.464349719475651e-06,
      "loss": 1.6655,
      "step": 16850
    },
    {
      "epoch": 2.868078591477418,
      "grad_norm": 4.739222049713135,
      "learning_rate": 5.457072667015189e-06,
      "loss": 1.715,
      "step": 16860
    },
    {
      "epoch": 2.8697797057072383,
      "grad_norm": 4.548635959625244,
      "learning_rate": 5.449797151968192e-06,
      "loss": 1.6876,
      "step": 16870
    },
    {
      "epoch": 2.8714808199370587,
      "grad_norm": 4.852926731109619,
      "learning_rate": 5.44252318317014e-06,
      "loss": 1.619,
      "step": 16880
    },
    {
      "epoch": 2.873181934166879,
      "grad_norm": 5.815784454345703,
      "learning_rate": 5.435250769454633e-06,
      "loss": 1.63,
      "step": 16890
    },
    {
      "epoch": 2.8748830483966996,
      "grad_norm": 5.472360134124756,
      "learning_rate": 5.4279799196533776e-06,
      "loss": 1.6145,
      "step": 16900
    },
    {
      "epoch": 2.8765841626265205,
      "grad_norm": 5.125363349914551,
      "learning_rate": 5.420710642596184e-06,
      "loss": 1.7248,
      "step": 16910
    },
    {
      "epoch": 2.878285276856341,
      "grad_norm": 5.4498610496521,
      "learning_rate": 5.41344294711095e-06,
      "loss": 1.6834,
      "step": 16920
    },
    {
      "epoch": 2.8799863910861614,
      "grad_norm": 5.653110980987549,
      "learning_rate": 5.406176842023661e-06,
      "loss": 1.709,
      "step": 16930
    },
    {
      "epoch": 2.881687505315982,
      "grad_norm": 5.308363437652588,
      "learning_rate": 5.398912336158364e-06,
      "loss": 1.7208,
      "step": 16940
    },
    {
      "epoch": 2.8833886195458023,
      "grad_norm": 5.207526206970215,
      "learning_rate": 5.3916494383371625e-06,
      "loss": 1.7317,
      "step": 16950
    },
    {
      "epoch": 2.885089733775623,
      "grad_norm": 5.076615810394287,
      "learning_rate": 5.384388157380212e-06,
      "loss": 1.6813,
      "step": 16960
    },
    {
      "epoch": 2.8867908480054436,
      "grad_norm": 5.327056884765625,
      "learning_rate": 5.3771285021057e-06,
      "loss": 1.8338,
      "step": 16970
    },
    {
      "epoch": 2.888491962235264,
      "grad_norm": 4.913582801818848,
      "learning_rate": 5.3698704813298406e-06,
      "loss": 1.585,
      "step": 16980
    },
    {
      "epoch": 2.8901930764650845,
      "grad_norm": 5.4310407638549805,
      "learning_rate": 5.362614103866873e-06,
      "loss": 1.788,
      "step": 16990
    },
    {
      "epoch": 2.891894190694905,
      "grad_norm": 5.223809719085693,
      "learning_rate": 5.355359378529024e-06,
      "loss": 1.6812,
      "step": 17000
    },
    {
      "epoch": 2.893595304924726,
      "grad_norm": 4.831777572631836,
      "learning_rate": 5.348106314126525e-06,
      "loss": 1.6825,
      "step": 17010
    },
    {
      "epoch": 2.8952964191545463,
      "grad_norm": 4.534534931182861,
      "learning_rate": 5.340854919467587e-06,
      "loss": 1.6786,
      "step": 17020
    },
    {
      "epoch": 2.8969975333843667,
      "grad_norm": 4.957123756408691,
      "learning_rate": 5.333605203358394e-06,
      "loss": 1.631,
      "step": 17030
    },
    {
      "epoch": 2.898698647614187,
      "grad_norm": 4.873937606811523,
      "learning_rate": 5.326357174603089e-06,
      "loss": 1.7522,
      "step": 17040
    },
    {
      "epoch": 2.9003997618440076,
      "grad_norm": 5.077538967132568,
      "learning_rate": 5.319110842003773e-06,
      "loss": 1.7145,
      "step": 17050
    },
    {
      "epoch": 2.9021008760738285,
      "grad_norm": 4.669430255889893,
      "learning_rate": 5.31186621436048e-06,
      "loss": 1.6077,
      "step": 17060
    },
    {
      "epoch": 2.903801990303649,
      "grad_norm": 5.177587509155273,
      "learning_rate": 5.304623300471174e-06,
      "loss": 1.7528,
      "step": 17070
    },
    {
      "epoch": 2.9055031045334694,
      "grad_norm": 5.614473819732666,
      "learning_rate": 5.297382109131741e-06,
      "loss": 1.6914,
      "step": 17080
    },
    {
      "epoch": 2.90720421876329,
      "grad_norm": 4.8400163650512695,
      "learning_rate": 5.290142649135974e-06,
      "loss": 1.705,
      "step": 17090
    },
    {
      "epoch": 2.9089053329931103,
      "grad_norm": 5.290616035461426,
      "learning_rate": 5.282904929275563e-06,
      "loss": 1.7426,
      "step": 17100
    },
    {
      "epoch": 2.910606447222931,
      "grad_norm": 4.153530597686768,
      "learning_rate": 5.275668958340085e-06,
      "loss": 1.7851,
      "step": 17110
    },
    {
      "epoch": 2.9123075614527516,
      "grad_norm": 6.381970405578613,
      "learning_rate": 5.2684347451169934e-06,
      "loss": 1.7088,
      "step": 17120
    },
    {
      "epoch": 2.914008675682572,
      "grad_norm": 6.792290210723877,
      "learning_rate": 5.261202298391605e-06,
      "loss": 1.6729,
      "step": 17130
    },
    {
      "epoch": 2.9157097899123925,
      "grad_norm": 5.274122714996338,
      "learning_rate": 5.253971626947096e-06,
      "loss": 1.6674,
      "step": 17140
    },
    {
      "epoch": 2.917410904142213,
      "grad_norm": 5.607309341430664,
      "learning_rate": 5.246742739564478e-06,
      "loss": 1.7321,
      "step": 17150
    },
    {
      "epoch": 2.919112018372034,
      "grad_norm": 6.518828868865967,
      "learning_rate": 5.239515645022608e-06,
      "loss": 1.5473,
      "step": 17160
    },
    {
      "epoch": 2.9208131326018543,
      "grad_norm": 5.3464484214782715,
      "learning_rate": 5.232290352098156e-06,
      "loss": 1.673,
      "step": 17170
    },
    {
      "epoch": 2.9225142468316747,
      "grad_norm": 5.311427116394043,
      "learning_rate": 5.225066869565609e-06,
      "loss": 1.7325,
      "step": 17180
    },
    {
      "epoch": 2.924215361061495,
      "grad_norm": 5.174399375915527,
      "learning_rate": 5.217845206197253e-06,
      "loss": 1.7709,
      "step": 17190
    },
    {
      "epoch": 2.9259164752913156,
      "grad_norm": 5.062442302703857,
      "learning_rate": 5.210625370763167e-06,
      "loss": 1.73,
      "step": 17200
    },
    {
      "epoch": 2.9276175895211365,
      "grad_norm": 4.9413838386535645,
      "learning_rate": 5.203407372031203e-06,
      "loss": 1.6064,
      "step": 17210
    },
    {
      "epoch": 2.929318703750957,
      "grad_norm": 5.229119300842285,
      "learning_rate": 5.196191218766996e-06,
      "loss": 1.6858,
      "step": 17220
    },
    {
      "epoch": 2.9310198179807774,
      "grad_norm": 4.790507793426514,
      "learning_rate": 5.188976919733929e-06,
      "loss": 1.6615,
      "step": 17230
    },
    {
      "epoch": 2.932720932210598,
      "grad_norm": 4.71579647064209,
      "learning_rate": 5.181764483693139e-06,
      "loss": 1.702,
      "step": 17240
    },
    {
      "epoch": 2.9344220464404183,
      "grad_norm": 6.398921966552734,
      "learning_rate": 5.174553919403493e-06,
      "loss": 1.5692,
      "step": 17250
    },
    {
      "epoch": 2.936123160670239,
      "grad_norm": 5.193541049957275,
      "learning_rate": 5.167345235621592e-06,
      "loss": 1.6498,
      "step": 17260
    },
    {
      "epoch": 2.9378242749000596,
      "grad_norm": 5.615979194641113,
      "learning_rate": 5.160138441101749e-06,
      "loss": 1.6564,
      "step": 17270
    },
    {
      "epoch": 2.93952538912988,
      "grad_norm": 5.156586647033691,
      "learning_rate": 5.152933544595992e-06,
      "loss": 1.6679,
      "step": 17280
    },
    {
      "epoch": 2.9412265033597005,
      "grad_norm": 5.580843925476074,
      "learning_rate": 5.14573055485403e-06,
      "loss": 1.6675,
      "step": 17290
    },
    {
      "epoch": 2.942927617589521,
      "grad_norm": 4.808172702789307,
      "learning_rate": 5.138529480623265e-06,
      "loss": 1.6527,
      "step": 17300
    },
    {
      "epoch": 2.944628731819342,
      "grad_norm": 4.427377223968506,
      "learning_rate": 5.131330330648771e-06,
      "loss": 1.7089,
      "step": 17310
    },
    {
      "epoch": 2.9463298460491623,
      "grad_norm": 5.212540149688721,
      "learning_rate": 5.124133113673284e-06,
      "loss": 1.6657,
      "step": 17320
    },
    {
      "epoch": 2.9480309602789827,
      "grad_norm": 5.242393970489502,
      "learning_rate": 5.116937838437198e-06,
      "loss": 1.6288,
      "step": 17330
    },
    {
      "epoch": 2.949732074508803,
      "grad_norm": 4.68244743347168,
      "learning_rate": 5.109744513678541e-06,
      "loss": 1.6502,
      "step": 17340
    },
    {
      "epoch": 2.9514331887386236,
      "grad_norm": 6.034001350402832,
      "learning_rate": 5.1025531481329765e-06,
      "loss": 1.7018,
      "step": 17350
    },
    {
      "epoch": 2.9531343029684445,
      "grad_norm": 5.154334545135498,
      "learning_rate": 5.095363750533789e-06,
      "loss": 1.7219,
      "step": 17360
    },
    {
      "epoch": 2.954835417198265,
      "grad_norm": 5.9419684410095215,
      "learning_rate": 5.088176329611871e-06,
      "loss": 1.6474,
      "step": 17370
    },
    {
      "epoch": 2.9565365314280854,
      "grad_norm": 5.031863689422607,
      "learning_rate": 5.080990894095715e-06,
      "loss": 1.6742,
      "step": 17380
    },
    {
      "epoch": 2.958237645657906,
      "grad_norm": 5.029059886932373,
      "learning_rate": 5.073807452711406e-06,
      "loss": 1.6828,
      "step": 17390
    },
    {
      "epoch": 2.9599387598877263,
      "grad_norm": 5.34487771987915,
      "learning_rate": 5.066626014182603e-06,
      "loss": 1.7054,
      "step": 17400
    },
    {
      "epoch": 2.961639874117547,
      "grad_norm": 4.280765533447266,
      "learning_rate": 5.059446587230534e-06,
      "loss": 1.6666,
      "step": 17410
    },
    {
      "epoch": 2.9633409883473676,
      "grad_norm": 5.261526107788086,
      "learning_rate": 5.0522691805739854e-06,
      "loss": 1.5722,
      "step": 17420
    },
    {
      "epoch": 2.965042102577188,
      "grad_norm": 4.829394817352295,
      "learning_rate": 5.045093802929287e-06,
      "loss": 1.6807,
      "step": 17430
    },
    {
      "epoch": 2.9667432168070085,
      "grad_norm": 5.677450180053711,
      "learning_rate": 5.0379204630103066e-06,
      "loss": 1.6928,
      "step": 17440
    },
    {
      "epoch": 2.968444331036829,
      "grad_norm": 6.83398962020874,
      "learning_rate": 5.030749169528439e-06,
      "loss": 1.6617,
      "step": 17450
    },
    {
      "epoch": 2.97014544526665,
      "grad_norm": 5.460479259490967,
      "learning_rate": 5.023579931192591e-06,
      "loss": 1.6503,
      "step": 17460
    },
    {
      "epoch": 2.9718465594964703,
      "grad_norm": 5.797318458557129,
      "learning_rate": 5.0164127567091775e-06,
      "loss": 1.6275,
      "step": 17470
    },
    {
      "epoch": 2.9735476737262907,
      "grad_norm": 5.9739203453063965,
      "learning_rate": 5.009247654782099e-06,
      "loss": 1.7129,
      "step": 17480
    },
    {
      "epoch": 2.975248787956111,
      "grad_norm": 5.020544052124023,
      "learning_rate": 5.002084634112748e-06,
      "loss": 1.67,
      "step": 17490
    },
    {
      "epoch": 2.9769499021859316,
      "grad_norm": 4.675488471984863,
      "learning_rate": 4.994923703399985e-06,
      "loss": 1.682,
      "step": 17500
    },
    {
      "epoch": 2.9786510164157525,
      "grad_norm": 4.7809295654296875,
      "learning_rate": 4.987764871340134e-06,
      "loss": 1.6577,
      "step": 17510
    },
    {
      "epoch": 2.980352130645573,
      "grad_norm": 5.958826065063477,
      "learning_rate": 4.980608146626967e-06,
      "loss": 1.7438,
      "step": 17520
    },
    {
      "epoch": 2.9820532448753934,
      "grad_norm": 4.796045303344727,
      "learning_rate": 4.9734535379517044e-06,
      "loss": 1.6495,
      "step": 17530
    },
    {
      "epoch": 2.983754359105214,
      "grad_norm": 5.485558032989502,
      "learning_rate": 4.966301054002987e-06,
      "loss": 1.7123,
      "step": 17540
    },
    {
      "epoch": 2.9854554733350342,
      "grad_norm": 4.91352653503418,
      "learning_rate": 4.959150703466877e-06,
      "loss": 1.6351,
      "step": 17550
    },
    {
      "epoch": 2.987156587564855,
      "grad_norm": 5.352145195007324,
      "learning_rate": 4.952002495026858e-06,
      "loss": 1.6597,
      "step": 17560
    },
    {
      "epoch": 2.9888577017946756,
      "grad_norm": 5.881899833679199,
      "learning_rate": 4.944856437363798e-06,
      "loss": 1.6898,
      "step": 17570
    },
    {
      "epoch": 2.990558816024496,
      "grad_norm": 5.65572452545166,
      "learning_rate": 4.937712539155961e-06,
      "loss": 1.7213,
      "step": 17580
    },
    {
      "epoch": 2.9922599302543165,
      "grad_norm": 6.259247779846191,
      "learning_rate": 4.930570809078983e-06,
      "loss": 1.6738,
      "step": 17590
    },
    {
      "epoch": 2.993961044484137,
      "grad_norm": 5.623510360717773,
      "learning_rate": 4.923431255805869e-06,
      "loss": 1.6972,
      "step": 17600
    },
    {
      "epoch": 2.995662158713958,
      "grad_norm": 6.0123724937438965,
      "learning_rate": 4.916293888006981e-06,
      "loss": 1.6268,
      "step": 17610
    },
    {
      "epoch": 2.9973632729437782,
      "grad_norm": 4.6384358406066895,
      "learning_rate": 4.909158714350032e-06,
      "loss": 1.6977,
      "step": 17620
    },
    {
      "epoch": 2.9990643871735987,
      "grad_norm": 4.532982349395752,
      "learning_rate": 4.902025743500061e-06,
      "loss": 1.7146,
      "step": 17630
    },
    {
      "epoch": 3.000765501403419,
      "grad_norm": 4.014255523681641,
      "learning_rate": 4.8948949841194375e-06,
      "loss": 1.6955,
      "step": 17640
    },
    {
      "epoch": 3.0024666156332396,
      "grad_norm": 4.36602258682251,
      "learning_rate": 4.887766444867844e-06,
      "loss": 1.6495,
      "step": 17650
    },
    {
      "epoch": 3.0041677298630605,
      "grad_norm": 5.667288303375244,
      "learning_rate": 4.8806401344022666e-06,
      "loss": 1.5879,
      "step": 17660
    },
    {
      "epoch": 3.005868844092881,
      "grad_norm": 5.284454345703125,
      "learning_rate": 4.873516061376989e-06,
      "loss": 1.7207,
      "step": 17670
    },
    {
      "epoch": 3.0075699583227014,
      "grad_norm": 5.314521312713623,
      "learning_rate": 4.86639423444357e-06,
      "loss": 1.6461,
      "step": 17680
    },
    {
      "epoch": 3.009271072552522,
      "grad_norm": 5.508584976196289,
      "learning_rate": 4.859274662250845e-06,
      "loss": 1.6006,
      "step": 17690
    },
    {
      "epoch": 3.0109721867823422,
      "grad_norm": 6.3213396072387695,
      "learning_rate": 4.852157353444914e-06,
      "loss": 1.6444,
      "step": 17700
    },
    {
      "epoch": 3.012673301012163,
      "grad_norm": 5.8377909660339355,
      "learning_rate": 4.845042316669122e-06,
      "loss": 1.5934,
      "step": 17710
    },
    {
      "epoch": 3.0143744152419836,
      "grad_norm": 6.235638618469238,
      "learning_rate": 4.837929560564058e-06,
      "loss": 1.5883,
      "step": 17720
    },
    {
      "epoch": 3.016075529471804,
      "grad_norm": 6.320291042327881,
      "learning_rate": 4.830819093767543e-06,
      "loss": 1.724,
      "step": 17730
    },
    {
      "epoch": 3.0177766437016245,
      "grad_norm": 5.157838344573975,
      "learning_rate": 4.823710924914616e-06,
      "loss": 1.5502,
      "step": 17740
    },
    {
      "epoch": 3.019477757931445,
      "grad_norm": 5.896531105041504,
      "learning_rate": 4.816605062637525e-06,
      "loss": 1.6807,
      "step": 17750
    },
    {
      "epoch": 3.021178872161266,
      "grad_norm": 6.018843173980713,
      "learning_rate": 4.809501515565719e-06,
      "loss": 1.5876,
      "step": 17760
    },
    {
      "epoch": 3.0228799863910862,
      "grad_norm": 5.043145656585693,
      "learning_rate": 4.802400292325831e-06,
      "loss": 1.6391,
      "step": 17770
    },
    {
      "epoch": 3.0245811006209067,
      "grad_norm": 6.310497760772705,
      "learning_rate": 4.795301401541675e-06,
      "loss": 1.4978,
      "step": 17780
    },
    {
      "epoch": 3.026282214850727,
      "grad_norm": 5.880179405212402,
      "learning_rate": 4.788204851834236e-06,
      "loss": 1.458,
      "step": 17790
    },
    {
      "epoch": 3.0279833290805476,
      "grad_norm": 5.39325475692749,
      "learning_rate": 4.781110651821649e-06,
      "loss": 1.7641,
      "step": 17800
    },
    {
      "epoch": 3.0296844433103685,
      "grad_norm": 6.003304958343506,
      "learning_rate": 4.7740188101191985e-06,
      "loss": 1.6129,
      "step": 17810
    },
    {
      "epoch": 3.031385557540189,
      "grad_norm": 5.247802257537842,
      "learning_rate": 4.766929335339306e-06,
      "loss": 1.559,
      "step": 17820
    },
    {
      "epoch": 3.0330866717700093,
      "grad_norm": 5.565807819366455,
      "learning_rate": 4.7598422360915186e-06,
      "loss": 1.6142,
      "step": 17830
    },
    {
      "epoch": 3.03478778599983,
      "grad_norm": 4.938119411468506,
      "learning_rate": 4.7527575209824935e-06,
      "loss": 1.6264,
      "step": 17840
    },
    {
      "epoch": 3.0364889002296502,
      "grad_norm": 6.364495277404785,
      "learning_rate": 4.745675198616001e-06,
      "loss": 1.5744,
      "step": 17850
    },
    {
      "epoch": 3.038190014459471,
      "grad_norm": 5.86942195892334,
      "learning_rate": 4.738595277592901e-06,
      "loss": 1.5696,
      "step": 17860
    },
    {
      "epoch": 3.0398911286892916,
      "grad_norm": 5.963891506195068,
      "learning_rate": 4.7315177665111384e-06,
      "loss": 1.6403,
      "step": 17870
    },
    {
      "epoch": 3.041592242919112,
      "grad_norm": 4.875551223754883,
      "learning_rate": 4.724442673965728e-06,
      "loss": 1.6111,
      "step": 17880
    },
    {
      "epoch": 3.0432933571489325,
      "grad_norm": 5.340967178344727,
      "learning_rate": 4.717370008548747e-06,
      "loss": 1.6849,
      "step": 17890
    },
    {
      "epoch": 3.044994471378753,
      "grad_norm": 5.945544719696045,
      "learning_rate": 4.710299778849335e-06,
      "loss": 1.604,
      "step": 17900
    },
    {
      "epoch": 3.046695585608574,
      "grad_norm": 5.568148612976074,
      "learning_rate": 4.7032319934536645e-06,
      "loss": 1.6131,
      "step": 17910
    },
    {
      "epoch": 3.0483966998383942,
      "grad_norm": 5.178503036499023,
      "learning_rate": 4.696166660944943e-06,
      "loss": 1.6011,
      "step": 17920
    },
    {
      "epoch": 3.0500978140682147,
      "grad_norm": 5.598849773406982,
      "learning_rate": 4.689103789903394e-06,
      "loss": 1.572,
      "step": 17930
    },
    {
      "epoch": 3.051798928298035,
      "grad_norm": 5.570218086242676,
      "learning_rate": 4.68204338890626e-06,
      "loss": 1.6541,
      "step": 17940
    },
    {
      "epoch": 3.0535000425278556,
      "grad_norm": 5.031559467315674,
      "learning_rate": 4.674985466527773e-06,
      "loss": 1.5255,
      "step": 17950
    },
    {
      "epoch": 3.0552011567576765,
      "grad_norm": 5.859495639801025,
      "learning_rate": 4.6679300313391715e-06,
      "loss": 1.5468,
      "step": 17960
    },
    {
      "epoch": 3.056902270987497,
      "grad_norm": 5.605506420135498,
      "learning_rate": 4.660877091908658e-06,
      "loss": 1.6024,
      "step": 17970
    },
    {
      "epoch": 3.0586033852173173,
      "grad_norm": 6.32056999206543,
      "learning_rate": 4.653826656801409e-06,
      "loss": 1.6519,
      "step": 17980
    },
    {
      "epoch": 3.060304499447138,
      "grad_norm": 5.6299028396606445,
      "learning_rate": 4.646778734579562e-06,
      "loss": 1.6598,
      "step": 17990
    },
    {
      "epoch": 3.0620056136769582,
      "grad_norm": 5.1421942710876465,
      "learning_rate": 4.639733333802202e-06,
      "loss": 1.6494,
      "step": 18000
    },
    {
      "epoch": 3.063706727906779,
      "grad_norm": 4.699141025543213,
      "learning_rate": 4.632690463025349e-06,
      "loss": 1.6086,
      "step": 18010
    },
    {
      "epoch": 3.0654078421365996,
      "grad_norm": 5.812696933746338,
      "learning_rate": 4.625650130801955e-06,
      "loss": 1.6311,
      "step": 18020
    },
    {
      "epoch": 3.06710895636642,
      "grad_norm": 5.819453716278076,
      "learning_rate": 4.618612345681886e-06,
      "loss": 1.653,
      "step": 18030
    },
    {
      "epoch": 3.0688100705962404,
      "grad_norm": 5.791920185089111,
      "learning_rate": 4.611577116211917e-06,
      "loss": 1.5801,
      "step": 18040
    },
    {
      "epoch": 3.070511184826061,
      "grad_norm": 6.71321439743042,
      "learning_rate": 4.604544450935716e-06,
      "loss": 1.5094,
      "step": 18050
    },
    {
      "epoch": 3.072212299055882,
      "grad_norm": 4.962828636169434,
      "learning_rate": 4.597514358393839e-06,
      "loss": 1.5627,
      "step": 18060
    },
    {
      "epoch": 3.0739134132857022,
      "grad_norm": 5.420648097991943,
      "learning_rate": 4.590486847123719e-06,
      "loss": 1.5667,
      "step": 18070
    },
    {
      "epoch": 3.0756145275155227,
      "grad_norm": 4.635862827301025,
      "learning_rate": 4.583461925659653e-06,
      "loss": 1.5673,
      "step": 18080
    },
    {
      "epoch": 3.077315641745343,
      "grad_norm": 3.965139865875244,
      "learning_rate": 4.576439602532793e-06,
      "loss": 1.6144,
      "step": 18090
    },
    {
      "epoch": 3.0790167559751636,
      "grad_norm": 5.953652858734131,
      "learning_rate": 4.569419886271135e-06,
      "loss": 1.6531,
      "step": 18100
    },
    {
      "epoch": 3.0807178702049844,
      "grad_norm": 6.302707672119141,
      "learning_rate": 4.562402785399509e-06,
      "loss": 1.6322,
      "step": 18110
    },
    {
      "epoch": 3.082418984434805,
      "grad_norm": 5.7404375076293945,
      "learning_rate": 4.555388308439569e-06,
      "loss": 1.7066,
      "step": 18120
    },
    {
      "epoch": 3.0841200986646253,
      "grad_norm": 6.028812408447266,
      "learning_rate": 4.548376463909786e-06,
      "loss": 1.695,
      "step": 18130
    },
    {
      "epoch": 3.0858212128944458,
      "grad_norm": 5.067834854125977,
      "learning_rate": 4.541367260325428e-06,
      "loss": 1.6536,
      "step": 18140
    },
    {
      "epoch": 3.087522327124266,
      "grad_norm": 4.90779447555542,
      "learning_rate": 4.53436070619856e-06,
      "loss": 1.655,
      "step": 18150
    },
    {
      "epoch": 3.089223441354087,
      "grad_norm": 5.944270610809326,
      "learning_rate": 4.52735681003803e-06,
      "loss": 1.4941,
      "step": 18160
    },
    {
      "epoch": 3.0909245555839076,
      "grad_norm": 5.422323226928711,
      "learning_rate": 4.520355580349455e-06,
      "loss": 1.5717,
      "step": 18170
    },
    {
      "epoch": 3.092625669813728,
      "grad_norm": 4.691229343414307,
      "learning_rate": 4.513357025635211e-06,
      "loss": 1.6177,
      "step": 18180
    },
    {
      "epoch": 3.0943267840435484,
      "grad_norm": 5.544463157653809,
      "learning_rate": 4.506361154394438e-06,
      "loss": 1.674,
      "step": 18190
    },
    {
      "epoch": 3.096027898273369,
      "grad_norm": 5.867314338684082,
      "learning_rate": 4.499367975123003e-06,
      "loss": 1.6024,
      "step": 18200
    },
    {
      "epoch": 3.0977290125031898,
      "grad_norm": 6.396510124206543,
      "learning_rate": 4.492377496313514e-06,
      "loss": 1.6325,
      "step": 18210
    },
    {
      "epoch": 3.09943012673301,
      "grad_norm": 4.796721458435059,
      "learning_rate": 4.4853897264552926e-06,
      "loss": 1.6043,
      "step": 18220
    },
    {
      "epoch": 3.1011312409628307,
      "grad_norm": 5.387908458709717,
      "learning_rate": 4.478404674034369e-06,
      "loss": 1.6566,
      "step": 18230
    },
    {
      "epoch": 3.102832355192651,
      "grad_norm": 4.891656875610352,
      "learning_rate": 4.471422347533487e-06,
      "loss": 1.5837,
      "step": 18240
    },
    {
      "epoch": 3.1045334694224715,
      "grad_norm": 5.820724010467529,
      "learning_rate": 4.464442755432066e-06,
      "loss": 1.5957,
      "step": 18250
    },
    {
      "epoch": 3.1062345836522924,
      "grad_norm": 5.723635196685791,
      "learning_rate": 4.45746590620621e-06,
      "loss": 1.6345,
      "step": 18260
    },
    {
      "epoch": 3.107935697882113,
      "grad_norm": 4.311638832092285,
      "learning_rate": 4.450491808328689e-06,
      "loss": 1.6473,
      "step": 18270
    },
    {
      "epoch": 3.1096368121119333,
      "grad_norm": 5.072981357574463,
      "learning_rate": 4.443520470268936e-06,
      "loss": 1.6255,
      "step": 18280
    },
    {
      "epoch": 3.1113379263417538,
      "grad_norm": 5.348658084869385,
      "learning_rate": 4.436551900493029e-06,
      "loss": 1.6239,
      "step": 18290
    },
    {
      "epoch": 3.113039040571574,
      "grad_norm": 4.85001802444458,
      "learning_rate": 4.42958610746369e-06,
      "loss": 1.6204,
      "step": 18300
    },
    {
      "epoch": 3.114740154801395,
      "grad_norm": 5.611588001251221,
      "learning_rate": 4.422623099640261e-06,
      "loss": 1.5433,
      "step": 18310
    },
    {
      "epoch": 3.1164412690312155,
      "grad_norm": 6.364828586578369,
      "learning_rate": 4.415662885478706e-06,
      "loss": 1.6021,
      "step": 18320
    },
    {
      "epoch": 3.118142383261036,
      "grad_norm": 5.450611114501953,
      "learning_rate": 4.408705473431594e-06,
      "loss": 1.6387,
      "step": 18330
    },
    {
      "epoch": 3.1198434974908564,
      "grad_norm": 5.470499038696289,
      "learning_rate": 4.401750871948095e-06,
      "loss": 1.6206,
      "step": 18340
    },
    {
      "epoch": 3.121544611720677,
      "grad_norm": 5.396094799041748,
      "learning_rate": 4.394799089473962e-06,
      "loss": 1.6674,
      "step": 18350
    },
    {
      "epoch": 3.1232457259504978,
      "grad_norm": 4.812134742736816,
      "learning_rate": 4.387850134451527e-06,
      "loss": 1.5964,
      "step": 18360
    },
    {
      "epoch": 3.124946840180318,
      "grad_norm": 5.111337184906006,
      "learning_rate": 4.380904015319687e-06,
      "loss": 1.629,
      "step": 18370
    },
    {
      "epoch": 3.1266479544101387,
      "grad_norm": 5.7565741539001465,
      "learning_rate": 4.373960740513894e-06,
      "loss": 1.5306,
      "step": 18380
    },
    {
      "epoch": 3.128349068639959,
      "grad_norm": 6.662924289703369,
      "learning_rate": 4.367020318466149e-06,
      "loss": 1.6761,
      "step": 18390
    },
    {
      "epoch": 3.1300501828697795,
      "grad_norm": 5.767753601074219,
      "learning_rate": 4.360082757604986e-06,
      "loss": 1.6244,
      "step": 18400
    },
    {
      "epoch": 3.1317512970996004,
      "grad_norm": 4.988311767578125,
      "learning_rate": 4.353148066355466e-06,
      "loss": 1.5582,
      "step": 18410
    },
    {
      "epoch": 3.133452411329421,
      "grad_norm": 5.014961242675781,
      "learning_rate": 4.346216253139164e-06,
      "loss": 1.652,
      "step": 18420
    },
    {
      "epoch": 3.1351535255592413,
      "grad_norm": 6.216855525970459,
      "learning_rate": 4.339287326374158e-06,
      "loss": 1.6092,
      "step": 18430
    },
    {
      "epoch": 3.1368546397890618,
      "grad_norm": 6.390921592712402,
      "learning_rate": 4.332361294475024e-06,
      "loss": 1.5669,
      "step": 18440
    },
    {
      "epoch": 3.138555754018882,
      "grad_norm": 5.6126556396484375,
      "learning_rate": 4.325438165852823e-06,
      "loss": 1.6298,
      "step": 18450
    },
    {
      "epoch": 3.140256868248703,
      "grad_norm": 6.1353583335876465,
      "learning_rate": 4.318517948915084e-06,
      "loss": 1.6227,
      "step": 18460
    },
    {
      "epoch": 3.1419579824785235,
      "grad_norm": 5.302263259887695,
      "learning_rate": 4.311600652065809e-06,
      "loss": 1.6988,
      "step": 18470
    },
    {
      "epoch": 3.143659096708344,
      "grad_norm": 5.06173038482666,
      "learning_rate": 4.304686283705449e-06,
      "loss": 1.6542,
      "step": 18480
    },
    {
      "epoch": 3.1453602109381644,
      "grad_norm": 5.5619730949401855,
      "learning_rate": 4.297774852230895e-06,
      "loss": 1.5825,
      "step": 18490
    },
    {
      "epoch": 3.147061325167985,
      "grad_norm": 6.599198818206787,
      "learning_rate": 4.290866366035479e-06,
      "loss": 1.5114,
      "step": 18500
    },
    {
      "epoch": 3.1487624393978058,
      "grad_norm": 4.429544448852539,
      "learning_rate": 4.283960833508952e-06,
      "loss": 1.5709,
      "step": 18510
    },
    {
      "epoch": 3.150463553627626,
      "grad_norm": 5.5696492195129395,
      "learning_rate": 4.277058263037474e-06,
      "loss": 1.5798,
      "step": 18520
    },
    {
      "epoch": 3.1521646678574466,
      "grad_norm": 5.115567207336426,
      "learning_rate": 4.270158663003617e-06,
      "loss": 1.6599,
      "step": 18530
    },
    {
      "epoch": 3.153865782087267,
      "grad_norm": 5.814573764801025,
      "learning_rate": 4.263262041786341e-06,
      "loss": 1.5718,
      "step": 18540
    },
    {
      "epoch": 3.1555668963170875,
      "grad_norm": 5.1354899406433105,
      "learning_rate": 4.256368407760988e-06,
      "loss": 1.61,
      "step": 18550
    },
    {
      "epoch": 3.1572680105469084,
      "grad_norm": 6.129085063934326,
      "learning_rate": 4.249477769299271e-06,
      "loss": 1.5248,
      "step": 18560
    },
    {
      "epoch": 3.158969124776729,
      "grad_norm": 6.191493511199951,
      "learning_rate": 4.242590134769268e-06,
      "loss": 1.6448,
      "step": 18570
    },
    {
      "epoch": 3.1606702390065493,
      "grad_norm": 6.188729763031006,
      "learning_rate": 4.235705512535406e-06,
      "loss": 1.5009,
      "step": 18580
    },
    {
      "epoch": 3.1623713532363698,
      "grad_norm": 6.1626691818237305,
      "learning_rate": 4.22882391095846e-06,
      "loss": 1.5537,
      "step": 18590
    },
    {
      "epoch": 3.16407246746619,
      "grad_norm": 6.382812023162842,
      "learning_rate": 4.221945338395531e-06,
      "loss": 1.6413,
      "step": 18600
    },
    {
      "epoch": 3.165773581696011,
      "grad_norm": 4.171438694000244,
      "learning_rate": 4.215069803200041e-06,
      "loss": 1.6249,
      "step": 18610
    },
    {
      "epoch": 3.1674746959258315,
      "grad_norm": 5.829582691192627,
      "learning_rate": 4.208197313721726e-06,
      "loss": 1.5484,
      "step": 18620
    },
    {
      "epoch": 3.169175810155652,
      "grad_norm": 5.257534027099609,
      "learning_rate": 4.2013278783066204e-06,
      "loss": 1.6919,
      "step": 18630
    },
    {
      "epoch": 3.1708769243854724,
      "grad_norm": 6.117254734039307,
      "learning_rate": 4.194461505297059e-06,
      "loss": 1.6177,
      "step": 18640
    },
    {
      "epoch": 3.172578038615293,
      "grad_norm": 6.516603946685791,
      "learning_rate": 4.187598203031646e-06,
      "loss": 1.708,
      "step": 18650
    },
    {
      "epoch": 3.1742791528451137,
      "grad_norm": 6.003725051879883,
      "learning_rate": 4.1807379798452595e-06,
      "loss": 1.5948,
      "step": 18660
    },
    {
      "epoch": 3.175980267074934,
      "grad_norm": 5.0657057762146,
      "learning_rate": 4.173880844069041e-06,
      "loss": 1.6061,
      "step": 18670
    },
    {
      "epoch": 3.1776813813047546,
      "grad_norm": 5.401756763458252,
      "learning_rate": 4.167026804030383e-06,
      "loss": 1.6046,
      "step": 18680
    },
    {
      "epoch": 3.179382495534575,
      "grad_norm": 5.948600769042969,
      "learning_rate": 4.160175868052914e-06,
      "loss": 1.6005,
      "step": 18690
    },
    {
      "epoch": 3.1810836097643955,
      "grad_norm": 6.31746244430542,
      "learning_rate": 4.153328044456498e-06,
      "loss": 1.6076,
      "step": 18700
    },
    {
      "epoch": 3.1827847239942164,
      "grad_norm": 5.688670635223389,
      "learning_rate": 4.146483341557217e-06,
      "loss": 1.7147,
      "step": 18710
    },
    {
      "epoch": 3.184485838224037,
      "grad_norm": 5.909487724304199,
      "learning_rate": 4.139641767667361e-06,
      "loss": 1.6519,
      "step": 18720
    },
    {
      "epoch": 3.1861869524538573,
      "grad_norm": 5.354527950286865,
      "learning_rate": 4.132803331095425e-06,
      "loss": 1.5028,
      "step": 18730
    },
    {
      "epoch": 3.1878880666836777,
      "grad_norm": 6.217424392700195,
      "learning_rate": 4.12596804014609e-06,
      "loss": 1.5504,
      "step": 18740
    },
    {
      "epoch": 3.189589180913498,
      "grad_norm": 6.052481174468994,
      "learning_rate": 4.119135903120217e-06,
      "loss": 1.5665,
      "step": 18750
    },
    {
      "epoch": 3.191290295143319,
      "grad_norm": 4.938386917114258,
      "learning_rate": 4.1123069283148414e-06,
      "loss": 1.7223,
      "step": 18760
    },
    {
      "epoch": 3.1929914093731395,
      "grad_norm": 5.718237400054932,
      "learning_rate": 4.105481124023152e-06,
      "loss": 1.5895,
      "step": 18770
    },
    {
      "epoch": 3.19469252360296,
      "grad_norm": 5.294105052947998,
      "learning_rate": 4.098658498534494e-06,
      "loss": 1.6055,
      "step": 18780
    },
    {
      "epoch": 3.1963936378327804,
      "grad_norm": 4.52695894241333,
      "learning_rate": 4.091839060134344e-06,
      "loss": 1.6437,
      "step": 18790
    },
    {
      "epoch": 3.198094752062601,
      "grad_norm": 6.001376152038574,
      "learning_rate": 4.085022817104314e-06,
      "loss": 1.6244,
      "step": 18800
    },
    {
      "epoch": 3.1997958662924217,
      "grad_norm": 5.516130447387695,
      "learning_rate": 4.078209777722137e-06,
      "loss": 1.6684,
      "step": 18810
    },
    {
      "epoch": 3.201496980522242,
      "grad_norm": 5.8052215576171875,
      "learning_rate": 4.071399950261648e-06,
      "loss": 1.5872,
      "step": 18820
    },
    {
      "epoch": 3.2031980947520626,
      "grad_norm": 6.434172630310059,
      "learning_rate": 4.064593342992791e-06,
      "loss": 1.628,
      "step": 18830
    },
    {
      "epoch": 3.204899208981883,
      "grad_norm": 5.731072425842285,
      "learning_rate": 4.057789964181592e-06,
      "loss": 1.6371,
      "step": 18840
    },
    {
      "epoch": 3.2066003232117035,
      "grad_norm": 5.7734761238098145,
      "learning_rate": 4.0509898220901594e-06,
      "loss": 1.5914,
      "step": 18850
    },
    {
      "epoch": 3.2083014374415244,
      "grad_norm": 5.650068283081055,
      "learning_rate": 4.044192924976667e-06,
      "loss": 1.6572,
      "step": 18860
    },
    {
      "epoch": 3.210002551671345,
      "grad_norm": 5.855218410491943,
      "learning_rate": 4.037399281095355e-06,
      "loss": 1.6194,
      "step": 18870
    },
    {
      "epoch": 3.2117036659011653,
      "grad_norm": 6.3672871589660645,
      "learning_rate": 4.030608898696508e-06,
      "loss": 1.5446,
      "step": 18880
    },
    {
      "epoch": 3.2134047801309857,
      "grad_norm": 5.921889305114746,
      "learning_rate": 4.023821786026452e-06,
      "loss": 1.5996,
      "step": 18890
    },
    {
      "epoch": 3.215105894360806,
      "grad_norm": 5.2454352378845215,
      "learning_rate": 4.017037951327539e-06,
      "loss": 1.5815,
      "step": 18900
    },
    {
      "epoch": 3.216807008590627,
      "grad_norm": 6.049569606781006,
      "learning_rate": 4.0102574028381414e-06,
      "loss": 1.6231,
      "step": 18910
    },
    {
      "epoch": 3.2185081228204475,
      "grad_norm": 4.872544288635254,
      "learning_rate": 4.00348014879264e-06,
      "loss": 1.6495,
      "step": 18920
    },
    {
      "epoch": 3.220209237050268,
      "grad_norm": 4.858197212219238,
      "learning_rate": 3.996706197421423e-06,
      "loss": 1.5588,
      "step": 18930
    },
    {
      "epoch": 3.2219103512800884,
      "grad_norm": 5.183925151824951,
      "learning_rate": 3.989935556950857e-06,
      "loss": 1.6679,
      "step": 18940
    },
    {
      "epoch": 3.223611465509909,
      "grad_norm": 6.936694622039795,
      "learning_rate": 3.98316823560329e-06,
      "loss": 1.6339,
      "step": 18950
    },
    {
      "epoch": 3.2253125797397297,
      "grad_norm": 5.0147175788879395,
      "learning_rate": 3.976404241597044e-06,
      "loss": 1.7504,
      "step": 18960
    },
    {
      "epoch": 3.22701369396955,
      "grad_norm": 5.347295761108398,
      "learning_rate": 3.969643583146394e-06,
      "loss": 1.6694,
      "step": 18970
    },
    {
      "epoch": 3.2287148081993706,
      "grad_norm": 5.799068450927734,
      "learning_rate": 3.962886268461574e-06,
      "loss": 1.5398,
      "step": 18980
    },
    {
      "epoch": 3.230415922429191,
      "grad_norm": 5.399052143096924,
      "learning_rate": 3.956132305748745e-06,
      "loss": 1.6176,
      "step": 18990
    },
    {
      "epoch": 3.2321170366590115,
      "grad_norm": 5.699862003326416,
      "learning_rate": 3.949381703210006e-06,
      "loss": 1.6679,
      "step": 19000
    },
    {
      "epoch": 3.2338181508888324,
      "grad_norm": 6.689784526824951,
      "learning_rate": 3.94263446904337e-06,
      "loss": 1.5748,
      "step": 19010
    },
    {
      "epoch": 3.235519265118653,
      "grad_norm": 5.182729721069336,
      "learning_rate": 3.935890611442765e-06,
      "loss": 1.528,
      "step": 19020
    },
    {
      "epoch": 3.2372203793484733,
      "grad_norm": 6.1299052238464355,
      "learning_rate": 3.92982403333515e-06,
      "loss": 1.6028,
      "step": 19030
    },
    {
      "epoch": 3.2389214935782937,
      "grad_norm": 5.4542646408081055,
      "learning_rate": 3.923086613769571e-06,
      "loss": 1.5886,
      "step": 19040
    },
    {
      "epoch": 3.240622607808114,
      "grad_norm": 5.992674827575684,
      "learning_rate": 3.9163525945091795e-06,
      "loss": 1.7332,
      "step": 19050
    },
    {
      "epoch": 3.242323722037935,
      "grad_norm": 5.830533027648926,
      "learning_rate": 3.909621983731854e-06,
      "loss": 1.518,
      "step": 19060
    },
    {
      "epoch": 3.2440248362677555,
      "grad_norm": 5.197500228881836,
      "learning_rate": 3.902894789611326e-06,
      "loss": 1.5895,
      "step": 19070
    },
    {
      "epoch": 3.245725950497576,
      "grad_norm": 5.365091323852539,
      "learning_rate": 3.896171020317191e-06,
      "loss": 1.5947,
      "step": 19080
    },
    {
      "epoch": 3.2474270647273964,
      "grad_norm": 6.121037483215332,
      "learning_rate": 3.889450684014876e-06,
      "loss": 1.6893,
      "step": 19090
    },
    {
      "epoch": 3.249128178957217,
      "grad_norm": 4.278195858001709,
      "learning_rate": 3.882733788865639e-06,
      "loss": 1.5762,
      "step": 19100
    },
    {
      "epoch": 3.2508292931870377,
      "grad_norm": 6.356879234313965,
      "learning_rate": 3.8760203430265626e-06,
      "loss": 1.5187,
      "step": 19110
    },
    {
      "epoch": 3.252530407416858,
      "grad_norm": 6.2268548011779785,
      "learning_rate": 3.869310354650538e-06,
      "loss": 1.6316,
      "step": 19120
    },
    {
      "epoch": 3.2542315216466786,
      "grad_norm": 6.183928966522217,
      "learning_rate": 3.8626038318862575e-06,
      "loss": 1.5908,
      "step": 19130
    },
    {
      "epoch": 3.255932635876499,
      "grad_norm": 5.49514102935791,
      "learning_rate": 3.8559007828782074e-06,
      "loss": 1.6499,
      "step": 19140
    },
    {
      "epoch": 3.2576337501063195,
      "grad_norm": 5.05054235458374,
      "learning_rate": 3.849201215766653e-06,
      "loss": 1.5706,
      "step": 19150
    },
    {
      "epoch": 3.2593348643361404,
      "grad_norm": 6.043948650360107,
      "learning_rate": 3.842505138687631e-06,
      "loss": 1.5639,
      "step": 19160
    },
    {
      "epoch": 3.261035978565961,
      "grad_norm": 4.8844733238220215,
      "learning_rate": 3.835812559772943e-06,
      "loss": 1.5719,
      "step": 19170
    },
    {
      "epoch": 3.2627370927957813,
      "grad_norm": 6.491847991943359,
      "learning_rate": 3.829123487150138e-06,
      "loss": 1.5799,
      "step": 19180
    },
    {
      "epoch": 3.2644382070256017,
      "grad_norm": 4.930334091186523,
      "learning_rate": 3.822437928942505e-06,
      "loss": 1.5475,
      "step": 19190
    },
    {
      "epoch": 3.266139321255422,
      "grad_norm": 6.247923374176025,
      "learning_rate": 3.815755893269078e-06,
      "loss": 1.6433,
      "step": 19200
    },
    {
      "epoch": 3.267840435485243,
      "grad_norm": 5.70054292678833,
      "learning_rate": 3.8090773882445975e-06,
      "loss": 1.5678,
      "step": 19210
    },
    {
      "epoch": 3.2695415497150635,
      "grad_norm": 4.861880779266357,
      "learning_rate": 3.802402421979526e-06,
      "loss": 1.5403,
      "step": 19220
    },
    {
      "epoch": 3.271242663944884,
      "grad_norm": 5.6805620193481445,
      "learning_rate": 3.795731002580024e-06,
      "loss": 1.6332,
      "step": 19230
    },
    {
      "epoch": 3.2729437781747044,
      "grad_norm": 5.739677906036377,
      "learning_rate": 3.789063138147946e-06,
      "loss": 1.6061,
      "step": 19240
    },
    {
      "epoch": 3.274644892404525,
      "grad_norm": 5.972851276397705,
      "learning_rate": 3.78239883678083e-06,
      "loss": 1.5477,
      "step": 19250
    },
    {
      "epoch": 3.2763460066343457,
      "grad_norm": 4.61767578125,
      "learning_rate": 3.7757381065718862e-06,
      "loss": 1.6034,
      "step": 19260
    },
    {
      "epoch": 3.278047120864166,
      "grad_norm": 5.919322967529297,
      "learning_rate": 3.7690809556099875e-06,
      "loss": 1.6656,
      "step": 19270
    },
    {
      "epoch": 3.2797482350939866,
      "grad_norm": 5.884451389312744,
      "learning_rate": 3.762427391979662e-06,
      "loss": 1.5455,
      "step": 19280
    },
    {
      "epoch": 3.281449349323807,
      "grad_norm": 5.5694475173950195,
      "learning_rate": 3.755777423761079e-06,
      "loss": 1.6901,
      "step": 19290
    },
    {
      "epoch": 3.2831504635536275,
      "grad_norm": 4.459254741668701,
      "learning_rate": 3.74913105903004e-06,
      "loss": 1.5733,
      "step": 19300
    },
    {
      "epoch": 3.2848515777834484,
      "grad_norm": 5.569211959838867,
      "learning_rate": 3.742488305857978e-06,
      "loss": 1.6079,
      "step": 19310
    },
    {
      "epoch": 3.286552692013269,
      "grad_norm": 4.4077348709106445,
      "learning_rate": 3.735849172311934e-06,
      "loss": 1.629,
      "step": 19320
    },
    {
      "epoch": 3.2882538062430893,
      "grad_norm": 5.164341926574707,
      "learning_rate": 3.7292136664545548e-06,
      "loss": 1.5199,
      "step": 19330
    },
    {
      "epoch": 3.2899549204729097,
      "grad_norm": 5.6606645584106445,
      "learning_rate": 3.722581796344076e-06,
      "loss": 1.5918,
      "step": 19340
    },
    {
      "epoch": 3.29165603470273,
      "grad_norm": 6.065959930419922,
      "learning_rate": 3.7159535700343273e-06,
      "loss": 1.5498,
      "step": 19350
    },
    {
      "epoch": 3.293357148932551,
      "grad_norm": 6.435763359069824,
      "learning_rate": 3.7093289955747043e-06,
      "loss": 1.6532,
      "step": 19360
    },
    {
      "epoch": 3.2950582631623715,
      "grad_norm": 4.562117576599121,
      "learning_rate": 3.7027080810101798e-06,
      "loss": 1.6412,
      "step": 19370
    },
    {
      "epoch": 3.296759377392192,
      "grad_norm": 5.498716354370117,
      "learning_rate": 3.696090834381271e-06,
      "loss": 1.536,
      "step": 19380
    },
    {
      "epoch": 3.2984604916220124,
      "grad_norm": 6.3962483406066895,
      "learning_rate": 3.689477263724045e-06,
      "loss": 1.6545,
      "step": 19390
    },
    {
      "epoch": 3.300161605851833,
      "grad_norm": 5.0929341316223145,
      "learning_rate": 3.682867377070103e-06,
      "loss": 1.6379,
      "step": 19400
    },
    {
      "epoch": 3.3018627200816537,
      "grad_norm": 5.5942912101745605,
      "learning_rate": 3.6762611824465726e-06,
      "loss": 1.6307,
      "step": 19410
    },
    {
      "epoch": 3.303563834311474,
      "grad_norm": 5.138519287109375,
      "learning_rate": 3.6696586878761048e-06,
      "loss": 1.5445,
      "step": 19420
    },
    {
      "epoch": 3.3052649485412946,
      "grad_norm": 5.532662868499756,
      "learning_rate": 3.6630599013768444e-06,
      "loss": 1.6385,
      "step": 19430
    },
    {
      "epoch": 3.306966062771115,
      "grad_norm": 6.1705002784729,
      "learning_rate": 3.656464830962441e-06,
      "loss": 1.5671,
      "step": 19440
    },
    {
      "epoch": 3.3086671770009355,
      "grad_norm": 6.107243537902832,
      "learning_rate": 3.6498734846420314e-06,
      "loss": 1.5554,
      "step": 19450
    },
    {
      "epoch": 3.3103682912307564,
      "grad_norm": 5.152639389038086,
      "learning_rate": 3.6432858704202276e-06,
      "loss": 1.6207,
      "step": 19460
    },
    {
      "epoch": 3.312069405460577,
      "grad_norm": 6.471705436706543,
      "learning_rate": 3.6367019962971085e-06,
      "loss": 1.6673,
      "step": 19470
    },
    {
      "epoch": 3.3137705196903973,
      "grad_norm": 4.922879695892334,
      "learning_rate": 3.630121870268217e-06,
      "loss": 1.6099,
      "step": 19480
    },
    {
      "epoch": 3.3154716339202177,
      "grad_norm": 4.955195426940918,
      "learning_rate": 3.623545500324537e-06,
      "loss": 1.6964,
      "step": 19490
    },
    {
      "epoch": 3.317172748150038,
      "grad_norm": 4.401156902313232,
      "learning_rate": 3.616972894452494e-06,
      "loss": 1.6287,
      "step": 19500
    },
    {
      "epoch": 3.318873862379859,
      "grad_norm": 6.410326957702637,
      "learning_rate": 3.6104040606339437e-06,
      "loss": 1.5844,
      "step": 19510
    },
    {
      "epoch": 3.3205749766096795,
      "grad_norm": 5.521576881408691,
      "learning_rate": 3.603839006846158e-06,
      "loss": 1.5821,
      "step": 19520
    },
    {
      "epoch": 3.3222760908395,
      "grad_norm": 8.332356452941895,
      "learning_rate": 3.5972777410618193e-06,
      "loss": 1.5981,
      "step": 19530
    },
    {
      "epoch": 3.3239772050693204,
      "grad_norm": 5.562382221221924,
      "learning_rate": 3.5907202712490113e-06,
      "loss": 1.5004,
      "step": 19540
    },
    {
      "epoch": 3.325678319299141,
      "grad_norm": 5.359004020690918,
      "learning_rate": 3.584166605371207e-06,
      "loss": 1.6241,
      "step": 19550
    },
    {
      "epoch": 3.3273794335289617,
      "grad_norm": 5.369149208068848,
      "learning_rate": 3.5776167513872585e-06,
      "loss": 1.4857,
      "step": 19560
    },
    {
      "epoch": 3.329080547758782,
      "grad_norm": 6.441427707672119,
      "learning_rate": 3.5710707172513913e-06,
      "loss": 1.5662,
      "step": 19570
    },
    {
      "epoch": 3.3307816619886026,
      "grad_norm": 4.339804649353027,
      "learning_rate": 3.5645285109131873e-06,
      "loss": 1.6146,
      "step": 19580
    },
    {
      "epoch": 3.332482776218423,
      "grad_norm": 6.252863883972168,
      "learning_rate": 3.557990140317588e-06,
      "loss": 1.5885,
      "step": 19590
    },
    {
      "epoch": 3.3341838904482435,
      "grad_norm": 6.938775062561035,
      "learning_rate": 3.55145561340487e-06,
      "loss": 1.587,
      "step": 19600
    },
    {
      "epoch": 3.3358850046780644,
      "grad_norm": 6.152193069458008,
      "learning_rate": 3.544924938110643e-06,
      "loss": 1.6359,
      "step": 19610
    },
    {
      "epoch": 3.337586118907885,
      "grad_norm": 4.870347499847412,
      "learning_rate": 3.538398122365841e-06,
      "loss": 1.5164,
      "step": 19620
    },
    {
      "epoch": 3.3392872331377053,
      "grad_norm": 5.487436294555664,
      "learning_rate": 3.5318751740967133e-06,
      "loss": 1.6175,
      "step": 19630
    },
    {
      "epoch": 3.3409883473675257,
      "grad_norm": 6.347774982452393,
      "learning_rate": 3.5253561012247996e-06,
      "loss": 1.5868,
      "step": 19640
    },
    {
      "epoch": 3.342689461597346,
      "grad_norm": 5.935305595397949,
      "learning_rate": 3.518840911666952e-06,
      "loss": 1.718,
      "step": 19650
    },
    {
      "epoch": 3.344390575827167,
      "grad_norm": 5.687722206115723,
      "learning_rate": 3.5123296133352954e-06,
      "loss": 1.5982,
      "step": 19660
    },
    {
      "epoch": 3.3460916900569875,
      "grad_norm": 6.090416431427002,
      "learning_rate": 3.50582221413723e-06,
      "loss": 1.5469,
      "step": 19670
    },
    {
      "epoch": 3.347792804286808,
      "grad_norm": 5.880667686462402,
      "learning_rate": 3.499318721975424e-06,
      "loss": 1.7217,
      "step": 19680
    },
    {
      "epoch": 3.3494939185166284,
      "grad_norm": 5.706654071807861,
      "learning_rate": 3.492819144747798e-06,
      "loss": 1.6737,
      "step": 19690
    },
    {
      "epoch": 3.351195032746449,
      "grad_norm": 4.656063556671143,
      "learning_rate": 3.4863234903475176e-06,
      "loss": 1.5404,
      "step": 19700
    },
    {
      "epoch": 3.3528961469762697,
      "grad_norm": 5.1501240730285645,
      "learning_rate": 3.479831766662989e-06,
      "loss": 1.5463,
      "step": 19710
    },
    {
      "epoch": 3.35459726120609,
      "grad_norm": 6.567721843719482,
      "learning_rate": 3.473343981577845e-06,
      "loss": 1.654,
      "step": 19720
    },
    {
      "epoch": 3.3562983754359106,
      "grad_norm": 5.558073997497559,
      "learning_rate": 3.4668601429709258e-06,
      "loss": 1.6144,
      "step": 19730
    },
    {
      "epoch": 3.357999489665731,
      "grad_norm": 5.354240894317627,
      "learning_rate": 3.460380258716289e-06,
      "loss": 1.6214,
      "step": 19740
    },
    {
      "epoch": 3.3597006038955515,
      "grad_norm": 6.7254319190979,
      "learning_rate": 3.453904336683185e-06,
      "loss": 1.587,
      "step": 19750
    },
    {
      "epoch": 3.3614017181253724,
      "grad_norm": 5.278835773468018,
      "learning_rate": 3.447432384736053e-06,
      "loss": 1.6102,
      "step": 19760
    },
    {
      "epoch": 3.363102832355193,
      "grad_norm": 6.793532848358154,
      "learning_rate": 3.440964410734515e-06,
      "loss": 1.5772,
      "step": 19770
    },
    {
      "epoch": 3.3648039465850133,
      "grad_norm": 6.4684367179870605,
      "learning_rate": 3.4345004225333584e-06,
      "loss": 1.516,
      "step": 19780
    },
    {
      "epoch": 3.3665050608148337,
      "grad_norm": 5.317403316497803,
      "learning_rate": 3.4280404279825305e-06,
      "loss": 1.6725,
      "step": 19790
    },
    {
      "epoch": 3.368206175044654,
      "grad_norm": 6.5871806144714355,
      "learning_rate": 3.4215844349271284e-06,
      "loss": 1.5603,
      "step": 19800
    },
    {
      "epoch": 3.369907289274475,
      "grad_norm": 4.981198787689209,
      "learning_rate": 3.41513245120739e-06,
      "loss": 1.6096,
      "step": 19810
    },
    {
      "epoch": 3.3716084035042955,
      "grad_norm": 5.90072774887085,
      "learning_rate": 3.4086844846586843e-06,
      "loss": 1.5599,
      "step": 19820
    },
    {
      "epoch": 3.373309517734116,
      "grad_norm": 6.637577533721924,
      "learning_rate": 3.402240543111504e-06,
      "loss": 1.6088,
      "step": 19830
    },
    {
      "epoch": 3.3750106319639364,
      "grad_norm": 6.211846828460693,
      "learning_rate": 3.3958006343914492e-06,
      "loss": 1.5779,
      "step": 19840
    },
    {
      "epoch": 3.376711746193757,
      "grad_norm": 6.740238666534424,
      "learning_rate": 3.3893647663192264e-06,
      "loss": 1.5889,
      "step": 19850
    },
    {
      "epoch": 3.3784128604235777,
      "grad_norm": 6.2450785636901855,
      "learning_rate": 3.3829329467106337e-06,
      "loss": 1.6165,
      "step": 19860
    },
    {
      "epoch": 3.380113974653398,
      "grad_norm": 5.701789379119873,
      "learning_rate": 3.3765051833765502e-06,
      "loss": 1.6581,
      "step": 19870
    },
    {
      "epoch": 3.3818150888832186,
      "grad_norm": 6.127169609069824,
      "learning_rate": 3.3700814841229353e-06,
      "loss": 1.5565,
      "step": 19880
    },
    {
      "epoch": 3.383516203113039,
      "grad_norm": 5.727521896362305,
      "learning_rate": 3.3636618567508085e-06,
      "loss": 1.6647,
      "step": 19890
    },
    {
      "epoch": 3.3852173173428595,
      "grad_norm": 7.1242756843566895,
      "learning_rate": 3.3572463090562442e-06,
      "loss": 1.5775,
      "step": 19900
    },
    {
      "epoch": 3.3869184315726804,
      "grad_norm": 5.4397735595703125,
      "learning_rate": 3.3508348488303637e-06,
      "loss": 1.6395,
      "step": 19910
    },
    {
      "epoch": 3.388619545802501,
      "grad_norm": 5.371628761291504,
      "learning_rate": 3.344427483859325e-06,
      "loss": 1.5023,
      "step": 19920
    },
    {
      "epoch": 3.3903206600323212,
      "grad_norm": 5.891162395477295,
      "learning_rate": 3.3380242219243062e-06,
      "loss": 1.538,
      "step": 19930
    },
    {
      "epoch": 3.3920217742621417,
      "grad_norm": 7.087254524230957,
      "learning_rate": 3.3316250708015143e-06,
      "loss": 1.5604,
      "step": 19940
    },
    {
      "epoch": 3.393722888491962,
      "grad_norm": 5.757351398468018,
      "learning_rate": 3.3252300382621557e-06,
      "loss": 1.6928,
      "step": 19950
    },
    {
      "epoch": 3.3954240027217826,
      "grad_norm": 6.146236419677734,
      "learning_rate": 3.3188391320724374e-06,
      "loss": 1.6554,
      "step": 19960
    },
    {
      "epoch": 3.3971251169516035,
      "grad_norm": 6.456681728363037,
      "learning_rate": 3.3124523599935554e-06,
      "loss": 1.6398,
      "step": 19970
    },
    {
      "epoch": 3.398826231181424,
      "grad_norm": 6.639796733856201,
      "learning_rate": 3.306069729781683e-06,
      "loss": 1.5743,
      "step": 19980
    },
    {
      "epoch": 3.4005273454112444,
      "grad_norm": 5.578784465789795,
      "learning_rate": 3.2996912491879696e-06,
      "loss": 1.5616,
      "step": 19990
    },
    {
      "epoch": 3.402228459641065,
      "grad_norm": 5.376632213592529,
      "learning_rate": 3.2933169259585196e-06,
      "loss": 1.6228,
      "step": 20000
    },
    {
      "epoch": 3.4039295738708852,
      "grad_norm": 5.6772236824035645,
      "learning_rate": 3.286946767834392e-06,
      "loss": 1.5185,
      "step": 20010
    },
    {
      "epoch": 3.405630688100706,
      "grad_norm": 6.941185474395752,
      "learning_rate": 3.2805807825515823e-06,
      "loss": 1.597,
      "step": 20020
    },
    {
      "epoch": 3.4073318023305266,
      "grad_norm": 7.1472487449646,
      "learning_rate": 3.274218977841023e-06,
      "loss": 1.5752,
      "step": 20030
    },
    {
      "epoch": 3.409032916560347,
      "grad_norm": 5.509366989135742,
      "learning_rate": 3.267861361428567e-06,
      "loss": 1.6204,
      "step": 20040
    },
    {
      "epoch": 3.4107340307901675,
      "grad_norm": 6.332919120788574,
      "learning_rate": 3.261507941034988e-06,
      "loss": 1.6334,
      "step": 20050
    },
    {
      "epoch": 3.412435145019988,
      "grad_norm": 5.64229154586792,
      "learning_rate": 3.255158724375956e-06,
      "loss": 1.5794,
      "step": 20060
    },
    {
      "epoch": 3.414136259249809,
      "grad_norm": 6.086897373199463,
      "learning_rate": 3.248813719162038e-06,
      "loss": 1.722,
      "step": 20070
    },
    {
      "epoch": 3.4158373734796292,
      "grad_norm": 5.616125583648682,
      "learning_rate": 3.2424729330986885e-06,
      "loss": 1.5349,
      "step": 20080
    },
    {
      "epoch": 3.4175384877094497,
      "grad_norm": 5.05100679397583,
      "learning_rate": 3.2361363738862344e-06,
      "loss": 1.5484,
      "step": 20090
    },
    {
      "epoch": 3.41923960193927,
      "grad_norm": 5.591383457183838,
      "learning_rate": 3.2298040492198747e-06,
      "loss": 1.5953,
      "step": 20100
    },
    {
      "epoch": 3.4209407161690906,
      "grad_norm": 6.765773773193359,
      "learning_rate": 3.223475966789662e-06,
      "loss": 1.5554,
      "step": 20110
    },
    {
      "epoch": 3.4226418303989115,
      "grad_norm": 6.567031383514404,
      "learning_rate": 3.217152134280497e-06,
      "loss": 1.5934,
      "step": 20120
    },
    {
      "epoch": 3.424342944628732,
      "grad_norm": 5.841222763061523,
      "learning_rate": 3.210832559372123e-06,
      "loss": 1.5859,
      "step": 20130
    },
    {
      "epoch": 3.4260440588585523,
      "grad_norm": 5.731585502624512,
      "learning_rate": 3.2045172497391076e-06,
      "loss": 1.5609,
      "step": 20140
    },
    {
      "epoch": 3.427745173088373,
      "grad_norm": 5.261969089508057,
      "learning_rate": 3.198206213050841e-06,
      "loss": 1.7027,
      "step": 20150
    },
    {
      "epoch": 3.4294462873181932,
      "grad_norm": 6.334646701812744,
      "learning_rate": 3.1918994569715286e-06,
      "loss": 1.5391,
      "step": 20160
    },
    {
      "epoch": 3.431147401548014,
      "grad_norm": 6.424821376800537,
      "learning_rate": 3.1855969891601706e-06,
      "loss": 1.6238,
      "step": 20170
    },
    {
      "epoch": 3.4328485157778346,
      "grad_norm": 4.80640983581543,
      "learning_rate": 3.1792988172705623e-06,
      "loss": 1.6282,
      "step": 20180
    },
    {
      "epoch": 3.434549630007655,
      "grad_norm": 5.788819789886475,
      "learning_rate": 3.1730049489512816e-06,
      "loss": 1.5954,
      "step": 20190
    },
    {
      "epoch": 3.4362507442374755,
      "grad_norm": 5.655074596405029,
      "learning_rate": 3.1667153918456812e-06,
      "loss": 1.6188,
      "step": 20200
    },
    {
      "epoch": 3.437951858467296,
      "grad_norm": 6.133515357971191,
      "learning_rate": 3.160430153591877e-06,
      "loss": 1.6272,
      "step": 20210
    },
    {
      "epoch": 3.439652972697117,
      "grad_norm": 5.728498458862305,
      "learning_rate": 3.1541492418227407e-06,
      "loss": 1.5631,
      "step": 20220
    },
    {
      "epoch": 3.4413540869269372,
      "grad_norm": 6.747180938720703,
      "learning_rate": 3.1478726641658888e-06,
      "loss": 1.6548,
      "step": 20230
    },
    {
      "epoch": 3.4430552011567577,
      "grad_norm": 4.901608943939209,
      "learning_rate": 3.1416004282436757e-06,
      "loss": 1.6217,
      "step": 20240
    },
    {
      "epoch": 3.444756315386578,
      "grad_norm": 5.332956314086914,
      "learning_rate": 3.1353325416731825e-06,
      "loss": 1.7339,
      "step": 20250
    },
    {
      "epoch": 3.4464574296163986,
      "grad_norm": 5.859671115875244,
      "learning_rate": 3.1290690120662092e-06,
      "loss": 1.618,
      "step": 20260
    },
    {
      "epoch": 3.4481585438462194,
      "grad_norm": 5.185967445373535,
      "learning_rate": 3.1228098470292597e-06,
      "loss": 1.5618,
      "step": 20270
    },
    {
      "epoch": 3.44985965807604,
      "grad_norm": 5.8756890296936035,
      "learning_rate": 3.1165550541635487e-06,
      "loss": 1.4586,
      "step": 20280
    },
    {
      "epoch": 3.4515607723058603,
      "grad_norm": 6.972698211669922,
      "learning_rate": 3.110304641064971e-06,
      "loss": 1.5828,
      "step": 20290
    },
    {
      "epoch": 3.453261886535681,
      "grad_norm": 5.784275054931641,
      "learning_rate": 3.1040586153241065e-06,
      "loss": 1.7556,
      "step": 20300
    },
    {
      "epoch": 3.4549630007655012,
      "grad_norm": 5.015942096710205,
      "learning_rate": 3.097816984526208e-06,
      "loss": 1.6751,
      "step": 20310
    },
    {
      "epoch": 3.456664114995322,
      "grad_norm": 5.505063056945801,
      "learning_rate": 3.091579756251185e-06,
      "loss": 1.5959,
      "step": 20320
    },
    {
      "epoch": 3.4583652292251426,
      "grad_norm": 6.521902084350586,
      "learning_rate": 3.0853469380736048e-06,
      "loss": 1.6657,
      "step": 20330
    },
    {
      "epoch": 3.460066343454963,
      "grad_norm": 4.743898391723633,
      "learning_rate": 3.079118537562684e-06,
      "loss": 1.5205,
      "step": 20340
    },
    {
      "epoch": 3.4617674576847834,
      "grad_norm": 5.524914264678955,
      "learning_rate": 3.072894562282268e-06,
      "loss": 1.6363,
      "step": 20350
    },
    {
      "epoch": 3.463468571914604,
      "grad_norm": 5.382754325866699,
      "learning_rate": 3.066675019790829e-06,
      "loss": 1.5969,
      "step": 20360
    },
    {
      "epoch": 3.4651696861444248,
      "grad_norm": 4.380696773529053,
      "learning_rate": 3.0604599176414567e-06,
      "loss": 1.6067,
      "step": 20370
    },
    {
      "epoch": 3.466870800374245,
      "grad_norm": 5.742018222808838,
      "learning_rate": 3.054249263381846e-06,
      "loss": 1.5962,
      "step": 20380
    },
    {
      "epoch": 3.4685719146040657,
      "grad_norm": 6.359322547912598,
      "learning_rate": 3.0480430645543003e-06,
      "loss": 1.6064,
      "step": 20390
    },
    {
      "epoch": 3.470273028833886,
      "grad_norm": 5.794898986816406,
      "learning_rate": 3.041841328695696e-06,
      "loss": 1.6039,
      "step": 20400
    },
    {
      "epoch": 3.4719741430637066,
      "grad_norm": 5.615640163421631,
      "learning_rate": 3.035644063337502e-06,
      "loss": 1.5731,
      "step": 20410
    },
    {
      "epoch": 3.4736752572935274,
      "grad_norm": 4.714178562164307,
      "learning_rate": 3.0294512760057537e-06,
      "loss": 1.5701,
      "step": 20420
    },
    {
      "epoch": 3.475376371523348,
      "grad_norm": 5.054441452026367,
      "learning_rate": 3.0232629742210494e-06,
      "loss": 1.5568,
      "step": 20430
    },
    {
      "epoch": 3.4770774857531683,
      "grad_norm": 5.176323890686035,
      "learning_rate": 3.017079165498537e-06,
      "loss": 1.6077,
      "step": 20440
    },
    {
      "epoch": 3.4787785999829888,
      "grad_norm": 5.393445014953613,
      "learning_rate": 3.010899857347915e-06,
      "loss": 1.7021,
      "step": 20450
    },
    {
      "epoch": 3.480479714212809,
      "grad_norm": 5.865242958068848,
      "learning_rate": 3.0047250572734095e-06,
      "loss": 1.5678,
      "step": 20460
    },
    {
      "epoch": 3.48218082844263,
      "grad_norm": 6.027548313140869,
      "learning_rate": 2.998554772773775e-06,
      "loss": 1.6008,
      "step": 20470
    },
    {
      "epoch": 3.4838819426724505,
      "grad_norm": 6.511929035186768,
      "learning_rate": 2.9923890113422818e-06,
      "loss": 1.6372,
      "step": 20480
    },
    {
      "epoch": 3.485583056902271,
      "grad_norm": 5.237430095672607,
      "learning_rate": 2.986227780466707e-06,
      "loss": 1.518,
      "step": 20490
    },
    {
      "epoch": 3.4872841711320914,
      "grad_norm": 6.152370929718018,
      "learning_rate": 2.9800710876293255e-06,
      "loss": 1.6589,
      "step": 20500
    },
    {
      "epoch": 3.488985285361912,
      "grad_norm": 6.018396854400635,
      "learning_rate": 2.9739189403069026e-06,
      "loss": 1.6461,
      "step": 20510
    },
    {
      "epoch": 3.4906863995917328,
      "grad_norm": 5.236014366149902,
      "learning_rate": 2.9677713459706806e-06,
      "loss": 1.6335,
      "step": 20520
    },
    {
      "epoch": 3.492387513821553,
      "grad_norm": 6.262957572937012,
      "learning_rate": 2.961628312086376e-06,
      "loss": 1.6554,
      "step": 20530
    },
    {
      "epoch": 3.4940886280513737,
      "grad_norm": 5.400660037994385,
      "learning_rate": 2.9554898461141658e-06,
      "loss": 1.5663,
      "step": 20540
    },
    {
      "epoch": 3.495789742281194,
      "grad_norm": 6.142693996429443,
      "learning_rate": 2.9493559555086747e-06,
      "loss": 1.5755,
      "step": 20550
    },
    {
      "epoch": 3.4974908565110145,
      "grad_norm": 4.737947463989258,
      "learning_rate": 2.9432266477189826e-06,
      "loss": 1.6199,
      "step": 20560
    },
    {
      "epoch": 3.4991919707408354,
      "grad_norm": 5.962085247039795,
      "learning_rate": 2.9371019301885942e-06,
      "loss": 1.6135,
      "step": 20570
    },
    {
      "epoch": 3.500893084970656,
      "grad_norm": 6.262288570404053,
      "learning_rate": 2.9309818103554416e-06,
      "loss": 1.6169,
      "step": 20580
    },
    {
      "epoch": 3.5025941992004763,
      "grad_norm": 6.423060417175293,
      "learning_rate": 2.9248662956518753e-06,
      "loss": 1.6157,
      "step": 20590
    },
    {
      "epoch": 3.5042953134302968,
      "grad_norm": 5.860107898712158,
      "learning_rate": 2.918755393504655e-06,
      "loss": 1.5585,
      "step": 20600
    },
    {
      "epoch": 3.5059964276601177,
      "grad_norm": 5.546974182128906,
      "learning_rate": 2.9126491113349277e-06,
      "loss": 1.5031,
      "step": 20610
    },
    {
      "epoch": 3.5076975418899377,
      "grad_norm": 5.556706428527832,
      "learning_rate": 2.906547456558246e-06,
      "loss": 1.6586,
      "step": 20620
    },
    {
      "epoch": 3.5093986561197585,
      "grad_norm": 5.941084384918213,
      "learning_rate": 2.9004504365845343e-06,
      "loss": 1.5983,
      "step": 20630
    },
    {
      "epoch": 3.511099770349579,
      "grad_norm": 6.159066677093506,
      "learning_rate": 2.894358058818087e-06,
      "loss": 1.6673,
      "step": 20640
    },
    {
      "epoch": 3.5128008845793994,
      "grad_norm": 6.33928918838501,
      "learning_rate": 2.8882703306575653e-06,
      "loss": 1.6407,
      "step": 20650
    },
    {
      "epoch": 3.51450199880922,
      "grad_norm": 5.965478897094727,
      "learning_rate": 2.8821872594959803e-06,
      "loss": 1.5833,
      "step": 20660
    },
    {
      "epoch": 3.5162031130390403,
      "grad_norm": 6.009367942810059,
      "learning_rate": 2.8761088527206885e-06,
      "loss": 1.5992,
      "step": 20670
    },
    {
      "epoch": 3.517904227268861,
      "grad_norm": 5.0004072189331055,
      "learning_rate": 2.8700351177133862e-06,
      "loss": 1.6309,
      "step": 20680
    },
    {
      "epoch": 3.5196053414986816,
      "grad_norm": 5.877298831939697,
      "learning_rate": 2.863966061850093e-06,
      "loss": 1.5811,
      "step": 20690
    },
    {
      "epoch": 3.521306455728502,
      "grad_norm": 5.1309428215026855,
      "learning_rate": 2.857901692501141e-06,
      "loss": 1.6566,
      "step": 20700
    },
    {
      "epoch": 3.5230075699583225,
      "grad_norm": 5.491775035858154,
      "learning_rate": 2.8518420170311782e-06,
      "loss": 1.5789,
      "step": 20710
    },
    {
      "epoch": 3.524708684188143,
      "grad_norm": 6.656279563903809,
      "learning_rate": 2.8457870427991465e-06,
      "loss": 1.4972,
      "step": 20720
    },
    {
      "epoch": 3.526409798417964,
      "grad_norm": 6.993110656738281,
      "learning_rate": 2.839736777158288e-06,
      "loss": 1.7199,
      "step": 20730
    },
    {
      "epoch": 3.5281109126477843,
      "grad_norm": 6.034376621246338,
      "learning_rate": 2.8336912274561176e-06,
      "loss": 1.57,
      "step": 20740
    },
    {
      "epoch": 3.5298120268776048,
      "grad_norm": 5.539135932922363,
      "learning_rate": 2.8276504010344247e-06,
      "loss": 1.5519,
      "step": 20750
    },
    {
      "epoch": 3.531513141107425,
      "grad_norm": 4.798160552978516,
      "learning_rate": 2.821614305229266e-06,
      "loss": 1.5829,
      "step": 20760
    },
    {
      "epoch": 3.5332142553372456,
      "grad_norm": 5.539159774780273,
      "learning_rate": 2.815582947370949e-06,
      "loss": 1.6393,
      "step": 20770
    },
    {
      "epoch": 3.5349153695670665,
      "grad_norm": 6.6422505378723145,
      "learning_rate": 2.8095563347840314e-06,
      "loss": 1.5283,
      "step": 20780
    },
    {
      "epoch": 3.536616483796887,
      "grad_norm": 4.52136754989624,
      "learning_rate": 2.803534474787306e-06,
      "loss": 1.6371,
      "step": 20790
    },
    {
      "epoch": 3.5383175980267074,
      "grad_norm": 6.491845607757568,
      "learning_rate": 2.797517374693794e-06,
      "loss": 1.5793,
      "step": 20800
    },
    {
      "epoch": 3.540018712256528,
      "grad_norm": 6.235819339752197,
      "learning_rate": 2.7915050418107366e-06,
      "loss": 1.5977,
      "step": 20810
    },
    {
      "epoch": 3.5417198264863483,
      "grad_norm": 4.864239692687988,
      "learning_rate": 2.785497483439586e-06,
      "loss": 1.5128,
      "step": 20820
    },
    {
      "epoch": 3.543420940716169,
      "grad_norm": 5.335078716278076,
      "learning_rate": 2.7794947068759954e-06,
      "loss": 1.5557,
      "step": 20830
    },
    {
      "epoch": 3.5451220549459896,
      "grad_norm": 6.0590362548828125,
      "learning_rate": 2.7734967194098084e-06,
      "loss": 1.6613,
      "step": 20840
    },
    {
      "epoch": 3.54682316917581,
      "grad_norm": 5.815928936004639,
      "learning_rate": 2.7675035283250606e-06,
      "loss": 1.5621,
      "step": 20850
    },
    {
      "epoch": 3.5485242834056305,
      "grad_norm": 8.026349067687988,
      "learning_rate": 2.7615151408999556e-06,
      "loss": 1.5136,
      "step": 20860
    },
    {
      "epoch": 3.550225397635451,
      "grad_norm": 5.613368511199951,
      "learning_rate": 2.755531564406865e-06,
      "loss": 1.5729,
      "step": 20870
    },
    {
      "epoch": 3.551926511865272,
      "grad_norm": 6.851166248321533,
      "learning_rate": 2.7495528061123187e-06,
      "loss": 1.6317,
      "step": 20880
    },
    {
      "epoch": 3.5536276260950923,
      "grad_norm": 5.184919357299805,
      "learning_rate": 2.7435788732769952e-06,
      "loss": 1.5271,
      "step": 20890
    },
    {
      "epoch": 3.5553287403249128,
      "grad_norm": 5.281979084014893,
      "learning_rate": 2.737609773155713e-06,
      "loss": 1.5594,
      "step": 20900
    },
    {
      "epoch": 3.557029854554733,
      "grad_norm": 6.160537242889404,
      "learning_rate": 2.7316455129974204e-06,
      "loss": 1.6165,
      "step": 20910
    },
    {
      "epoch": 3.5587309687845536,
      "grad_norm": 6.142460346221924,
      "learning_rate": 2.72568610004519e-06,
      "loss": 1.5558,
      "step": 20920
    },
    {
      "epoch": 3.5604320830143745,
      "grad_norm": 6.025132179260254,
      "learning_rate": 2.7197315415362057e-06,
      "loss": 1.645,
      "step": 20930
    },
    {
      "epoch": 3.562133197244195,
      "grad_norm": 6.387031078338623,
      "learning_rate": 2.713781844701759e-06,
      "loss": 1.5794,
      "step": 20940
    },
    {
      "epoch": 3.5638343114740154,
      "grad_norm": 6.645457744598389,
      "learning_rate": 2.7078370167672324e-06,
      "loss": 1.6584,
      "step": 20950
    },
    {
      "epoch": 3.565535425703836,
      "grad_norm": 4.468255519866943,
      "learning_rate": 2.701897064952104e-06,
      "loss": 1.4577,
      "step": 20960
    },
    {
      "epoch": 3.5672365399336563,
      "grad_norm": 6.84254264831543,
      "learning_rate": 2.6959619964699215e-06,
      "loss": 1.5517,
      "step": 20970
    },
    {
      "epoch": 3.568937654163477,
      "grad_norm": 7.275737762451172,
      "learning_rate": 2.6900318185283094e-06,
      "loss": 1.5276,
      "step": 20980
    },
    {
      "epoch": 3.5706387683932976,
      "grad_norm": 4.726008415222168,
      "learning_rate": 2.684106538328944e-06,
      "loss": 1.5729,
      "step": 20990
    },
    {
      "epoch": 3.572339882623118,
      "grad_norm": 5.600918292999268,
      "learning_rate": 2.6781861630675627e-06,
      "loss": 1.7046,
      "step": 21000
    },
    {
      "epoch": 3.5740409968529385,
      "grad_norm": 5.3228230476379395,
      "learning_rate": 2.672270699933939e-06,
      "loss": 1.6861,
      "step": 21010
    },
    {
      "epoch": 3.575742111082759,
      "grad_norm": 6.673993110656738,
      "learning_rate": 2.6663601561118895e-06,
      "loss": 1.5471,
      "step": 21020
    },
    {
      "epoch": 3.57744322531258,
      "grad_norm": 5.231881618499756,
      "learning_rate": 2.6604545387792506e-06,
      "loss": 1.6026,
      "step": 21030
    },
    {
      "epoch": 3.5791443395424003,
      "grad_norm": 5.421326637268066,
      "learning_rate": 2.6545538551078772e-06,
      "loss": 1.5646,
      "step": 21040
    },
    {
      "epoch": 3.5808454537722207,
      "grad_norm": 5.813632488250732,
      "learning_rate": 2.648658112263633e-06,
      "loss": 1.5693,
      "step": 21050
    },
    {
      "epoch": 3.582546568002041,
      "grad_norm": 5.844841480255127,
      "learning_rate": 2.642767317406381e-06,
      "loss": 1.5946,
      "step": 21060
    },
    {
      "epoch": 3.5842476822318616,
      "grad_norm": 6.236157417297363,
      "learning_rate": 2.636881477689975e-06,
      "loss": 1.6562,
      "step": 21070
    },
    {
      "epoch": 3.5859487964616825,
      "grad_norm": 4.842199325561523,
      "learning_rate": 2.631000600262253e-06,
      "loss": 1.6737,
      "step": 21080
    },
    {
      "epoch": 3.587649910691503,
      "grad_norm": 5.5632452964782715,
      "learning_rate": 2.6251246922650255e-06,
      "loss": 1.6204,
      "step": 21090
    },
    {
      "epoch": 3.5893510249213234,
      "grad_norm": 5.149427890777588,
      "learning_rate": 2.619253760834067e-06,
      "loss": 1.5674,
      "step": 21100
    },
    {
      "epoch": 3.591052139151144,
      "grad_norm": 6.830357551574707,
      "learning_rate": 2.6133878130991113e-06,
      "loss": 1.6327,
      "step": 21110
    },
    {
      "epoch": 3.5927532533809643,
      "grad_norm": 5.267126083374023,
      "learning_rate": 2.6075268561838346e-06,
      "loss": 1.6055,
      "step": 21120
    },
    {
      "epoch": 3.594454367610785,
      "grad_norm": 5.860008239746094,
      "learning_rate": 2.6016708972058606e-06,
      "loss": 1.6324,
      "step": 21130
    },
    {
      "epoch": 3.5961554818406056,
      "grad_norm": 5.772332191467285,
      "learning_rate": 2.595819943276737e-06,
      "loss": 1.6781,
      "step": 21140
    },
    {
      "epoch": 3.597856596070426,
      "grad_norm": 5.773773670196533,
      "learning_rate": 2.589974001501934e-06,
      "loss": 1.6179,
      "step": 21150
    },
    {
      "epoch": 3.5995577103002465,
      "grad_norm": 5.229680061340332,
      "learning_rate": 2.584133078980837e-06,
      "loss": 1.7436,
      "step": 21160
    },
    {
      "epoch": 3.601258824530067,
      "grad_norm": 5.797763347625732,
      "learning_rate": 2.5782971828067333e-06,
      "loss": 1.6327,
      "step": 21170
    },
    {
      "epoch": 3.602959938759888,
      "grad_norm": 4.195744514465332,
      "learning_rate": 2.5724663200668086e-06,
      "loss": 1.5473,
      "step": 21180
    },
    {
      "epoch": 3.6046610529897083,
      "grad_norm": 5.071658134460449,
      "learning_rate": 2.5666404978421362e-06,
      "loss": 1.4735,
      "step": 21190
    },
    {
      "epoch": 3.6063621672195287,
      "grad_norm": 5.922102928161621,
      "learning_rate": 2.5608197232076656e-06,
      "loss": 1.6592,
      "step": 21200
    },
    {
      "epoch": 3.608063281449349,
      "grad_norm": 5.720836639404297,
      "learning_rate": 2.5550040032322174e-06,
      "loss": 1.6041,
      "step": 21210
    },
    {
      "epoch": 3.6097643956791696,
      "grad_norm": 7.563029766082764,
      "learning_rate": 2.5491933449784756e-06,
      "loss": 1.6228,
      "step": 21220
    },
    {
      "epoch": 3.6114655099089905,
      "grad_norm": 5.825148582458496,
      "learning_rate": 2.5433877555029748e-06,
      "loss": 1.5348,
      "step": 21230
    },
    {
      "epoch": 3.613166624138811,
      "grad_norm": 5.527562141418457,
      "learning_rate": 2.5375872418560922e-06,
      "loss": 1.6959,
      "step": 21240
    },
    {
      "epoch": 3.6148677383686314,
      "grad_norm": 5.692595481872559,
      "learning_rate": 2.5317918110820496e-06,
      "loss": 1.5923,
      "step": 21250
    },
    {
      "epoch": 3.616568852598452,
      "grad_norm": 6.298586845397949,
      "learning_rate": 2.526001470218888e-06,
      "loss": 1.7093,
      "step": 21260
    },
    {
      "epoch": 3.6182699668282723,
      "grad_norm": 5.965487003326416,
      "learning_rate": 2.520216226298467e-06,
      "loss": 1.6044,
      "step": 21270
    },
    {
      "epoch": 3.619971081058093,
      "grad_norm": 6.28692626953125,
      "learning_rate": 2.5144360863464634e-06,
      "loss": 1.5801,
      "step": 21280
    },
    {
      "epoch": 3.6216721952879136,
      "grad_norm": 6.619129180908203,
      "learning_rate": 2.5086610573823425e-06,
      "loss": 1.6437,
      "step": 21290
    },
    {
      "epoch": 3.623373309517734,
      "grad_norm": 6.151878833770752,
      "learning_rate": 2.5028911464193796e-06,
      "loss": 1.6172,
      "step": 21300
    },
    {
      "epoch": 3.6250744237475545,
      "grad_norm": 5.117790699005127,
      "learning_rate": 2.497126360464623e-06,
      "loss": 1.564,
      "step": 21310
    },
    {
      "epoch": 3.626775537977375,
      "grad_norm": 6.424371719360352,
      "learning_rate": 2.4913667065188997e-06,
      "loss": 1.5033,
      "step": 21320
    },
    {
      "epoch": 3.628476652207196,
      "grad_norm": 4.737883567810059,
      "learning_rate": 2.4856121915768064e-06,
      "loss": 1.6163,
      "step": 21330
    },
    {
      "epoch": 3.6301777664370163,
      "grad_norm": 6.619544982910156,
      "learning_rate": 2.479862822626696e-06,
      "loss": 1.5846,
      "step": 21340
    },
    {
      "epoch": 3.6318788806668367,
      "grad_norm": 5.334020137786865,
      "learning_rate": 2.4741186066506735e-06,
      "loss": 1.6201,
      "step": 21350
    },
    {
      "epoch": 3.633579994896657,
      "grad_norm": 5.102516174316406,
      "learning_rate": 2.468379550624589e-06,
      "loss": 1.6494,
      "step": 21360
    },
    {
      "epoch": 3.6352811091264776,
      "grad_norm": 6.121121883392334,
      "learning_rate": 2.462645661518023e-06,
      "loss": 1.5958,
      "step": 21370
    },
    {
      "epoch": 3.6369822233562985,
      "grad_norm": 6.0861968994140625,
      "learning_rate": 2.4569169462942787e-06,
      "loss": 1.5282,
      "step": 21380
    },
    {
      "epoch": 3.638683337586119,
      "grad_norm": 6.331408500671387,
      "learning_rate": 2.4511934119103807e-06,
      "loss": 1.6055,
      "step": 21390
    },
    {
      "epoch": 3.6403844518159394,
      "grad_norm": 7.731186866760254,
      "learning_rate": 2.44547506531706e-06,
      "loss": 1.6488,
      "step": 21400
    },
    {
      "epoch": 3.64208556604576,
      "grad_norm": 4.587552547454834,
      "learning_rate": 2.439761913458745e-06,
      "loss": 1.6261,
      "step": 21410
    },
    {
      "epoch": 3.6437866802755803,
      "grad_norm": 5.035861015319824,
      "learning_rate": 2.434053963273564e-06,
      "loss": 1.6608,
      "step": 21420
    },
    {
      "epoch": 3.645487794505401,
      "grad_norm": 5.879154205322266,
      "learning_rate": 2.4283512216933196e-06,
      "loss": 1.5426,
      "step": 21430
    },
    {
      "epoch": 3.6471889087352216,
      "grad_norm": 5.091813564300537,
      "learning_rate": 2.4226536956434926e-06,
      "loss": 1.6141,
      "step": 21440
    },
    {
      "epoch": 3.648890022965042,
      "grad_norm": 5.193233489990234,
      "learning_rate": 2.4169613920432284e-06,
      "loss": 1.6456,
      "step": 21450
    },
    {
      "epoch": 3.6505911371948625,
      "grad_norm": 6.129888534545898,
      "learning_rate": 2.411274317805332e-06,
      "loss": 1.546,
      "step": 21460
    },
    {
      "epoch": 3.652292251424683,
      "grad_norm": 5.277714729309082,
      "learning_rate": 2.4055924798362564e-06,
      "loss": 1.6102,
      "step": 21470
    },
    {
      "epoch": 3.653993365654504,
      "grad_norm": 5.8031511306762695,
      "learning_rate": 2.3999158850360967e-06,
      "loss": 1.6409,
      "step": 21480
    },
    {
      "epoch": 3.6556944798843243,
      "grad_norm": 5.230282783508301,
      "learning_rate": 2.3942445402985805e-06,
      "loss": 1.6237,
      "step": 21490
    },
    {
      "epoch": 3.6573955941141447,
      "grad_norm": 6.073917388916016,
      "learning_rate": 2.388578452511059e-06,
      "loss": 1.6167,
      "step": 21500
    },
    {
      "epoch": 3.659096708343965,
      "grad_norm": 5.9513258934021,
      "learning_rate": 2.3829176285544988e-06,
      "loss": 1.6356,
      "step": 21510
    },
    {
      "epoch": 3.6607978225737856,
      "grad_norm": 5.671891212463379,
      "learning_rate": 2.3772620753034735e-06,
      "loss": 1.5689,
      "step": 21520
    },
    {
      "epoch": 3.6624989368036065,
      "grad_norm": 6.27203369140625,
      "learning_rate": 2.371611799626162e-06,
      "loss": 1.4825,
      "step": 21530
    },
    {
      "epoch": 3.664200051033427,
      "grad_norm": 7.135629653930664,
      "learning_rate": 2.365966808384326e-06,
      "loss": 1.6012,
      "step": 21540
    },
    {
      "epoch": 3.6659011652632474,
      "grad_norm": 6.8408284187316895,
      "learning_rate": 2.3603271084333134e-06,
      "loss": 1.6191,
      "step": 21550
    },
    {
      "epoch": 3.667602279493068,
      "grad_norm": 4.956294536590576,
      "learning_rate": 2.3546927066220467e-06,
      "loss": 1.5503,
      "step": 21560
    },
    {
      "epoch": 3.6693033937228883,
      "grad_norm": 5.815354347229004,
      "learning_rate": 2.349063609793014e-06,
      "loss": 1.5498,
      "step": 21570
    },
    {
      "epoch": 3.671004507952709,
      "grad_norm": 5.707915782928467,
      "learning_rate": 2.343439824782255e-06,
      "loss": 1.5665,
      "step": 21580
    },
    {
      "epoch": 3.6727056221825296,
      "grad_norm": 6.232976913452148,
      "learning_rate": 2.3378213584193703e-06,
      "loss": 1.5358,
      "step": 21590
    },
    {
      "epoch": 3.67440673641235,
      "grad_norm": 5.771692752838135,
      "learning_rate": 2.3322082175274944e-06,
      "loss": 1.5688,
      "step": 21600
    },
    {
      "epoch": 3.6761078506421705,
      "grad_norm": 6.212607383728027,
      "learning_rate": 2.3266004089232927e-06,
      "loss": 1.6291,
      "step": 21610
    },
    {
      "epoch": 3.677808964871991,
      "grad_norm": 5.270486354827881,
      "learning_rate": 2.3209979394169597e-06,
      "loss": 1.516,
      "step": 21620
    },
    {
      "epoch": 3.679510079101812,
      "grad_norm": 5.60796594619751,
      "learning_rate": 2.315400815812203e-06,
      "loss": 1.5644,
      "step": 21630
    },
    {
      "epoch": 3.6812111933316323,
      "grad_norm": 5.498111248016357,
      "learning_rate": 2.3098090449062372e-06,
      "loss": 1.5741,
      "step": 21640
    },
    {
      "epoch": 3.6829123075614527,
      "grad_norm": 6.797544002532959,
      "learning_rate": 2.304222633489782e-06,
      "loss": 1.503,
      "step": 21650
    },
    {
      "epoch": 3.684613421791273,
      "grad_norm": 5.772705078125,
      "learning_rate": 2.299199451185788e-06,
      "loss": 1.6701,
      "step": 21660
    },
    {
      "epoch": 3.6863145360210936,
      "grad_norm": 5.532495975494385,
      "learning_rate": 2.293623241484506e-06,
      "loss": 1.5906,
      "step": 21670
    },
    {
      "epoch": 3.6880156502509145,
      "grad_norm": 6.167540550231934,
      "learning_rate": 2.288052410928972e-06,
      "loss": 1.6057,
      "step": 21680
    },
    {
      "epoch": 3.689716764480735,
      "grad_norm": 5.464738845825195,
      "learning_rate": 2.282486966284476e-06,
      "loss": 1.5806,
      "step": 21690
    },
    {
      "epoch": 3.6914178787105554,
      "grad_norm": 5.4526472091674805,
      "learning_rate": 2.2769269143097606e-06,
      "loss": 1.6392,
      "step": 21700
    },
    {
      "epoch": 3.693118992940376,
      "grad_norm": 4.271237850189209,
      "learning_rate": 2.271372261757022e-06,
      "loss": 1.7209,
      "step": 21710
    },
    {
      "epoch": 3.6948201071701963,
      "grad_norm": 6.350117206573486,
      "learning_rate": 2.2658230153718993e-06,
      "loss": 1.5714,
      "step": 21720
    },
    {
      "epoch": 3.696521221400017,
      "grad_norm": 6.264103412628174,
      "learning_rate": 2.2602791818934672e-06,
      "loss": 1.585,
      "step": 21730
    },
    {
      "epoch": 3.6982223356298376,
      "grad_norm": 6.342548847198486,
      "learning_rate": 2.2547407680542244e-06,
      "loss": 1.5491,
      "step": 21740
    },
    {
      "epoch": 3.699923449859658,
      "grad_norm": 5.97145938873291,
      "learning_rate": 2.2492077805800908e-06,
      "loss": 1.5353,
      "step": 21750
    },
    {
      "epoch": 3.7016245640894785,
      "grad_norm": 5.500047206878662,
      "learning_rate": 2.2436802261903934e-06,
      "loss": 1.6206,
      "step": 21760
    },
    {
      "epoch": 3.703325678319299,
      "grad_norm": 6.11614465713501,
      "learning_rate": 2.2381581115978653e-06,
      "loss": 1.5965,
      "step": 21770
    },
    {
      "epoch": 3.70502679254912,
      "grad_norm": 6.1228179931640625,
      "learning_rate": 2.232641443508629e-06,
      "loss": 1.5294,
      "step": 21780
    },
    {
      "epoch": 3.7067279067789403,
      "grad_norm": 6.533079147338867,
      "learning_rate": 2.227130228622194e-06,
      "loss": 1.5897,
      "step": 21790
    },
    {
      "epoch": 3.7084290210087607,
      "grad_norm": 5.325455665588379,
      "learning_rate": 2.221624473631452e-06,
      "loss": 1.5396,
      "step": 21800
    },
    {
      "epoch": 3.710130135238581,
      "grad_norm": 4.874083995819092,
      "learning_rate": 2.2161241852226572e-06,
      "loss": 1.6193,
      "step": 21810
    },
    {
      "epoch": 3.7118312494684016,
      "grad_norm": 4.729120254516602,
      "learning_rate": 2.210629370075429e-06,
      "loss": 1.6007,
      "step": 21820
    },
    {
      "epoch": 3.7135323636982225,
      "grad_norm": 6.683821678161621,
      "learning_rate": 2.2051400348627395e-06,
      "loss": 1.6439,
      "step": 21830
    },
    {
      "epoch": 3.715233477928043,
      "grad_norm": 7.281447887420654,
      "learning_rate": 2.1996561862509047e-06,
      "loss": 1.6405,
      "step": 21840
    },
    {
      "epoch": 3.7169345921578634,
      "grad_norm": 5.484798431396484,
      "learning_rate": 2.19417783089958e-06,
      "loss": 1.6407,
      "step": 21850
    },
    {
      "epoch": 3.718635706387684,
      "grad_norm": 5.7020344734191895,
      "learning_rate": 2.1887049754617463e-06,
      "loss": 1.5905,
      "step": 21860
    },
    {
      "epoch": 3.7203368206175043,
      "grad_norm": 6.41756534576416,
      "learning_rate": 2.1832376265837074e-06,
      "loss": 1.4172,
      "step": 21870
    },
    {
      "epoch": 3.722037934847325,
      "grad_norm": 6.037424564361572,
      "learning_rate": 2.177775790905082e-06,
      "loss": 1.6442,
      "step": 21880
    },
    {
      "epoch": 3.7237390490771456,
      "grad_norm": 5.487923622131348,
      "learning_rate": 2.1723194750587883e-06,
      "loss": 1.5439,
      "step": 21890
    },
    {
      "epoch": 3.725440163306966,
      "grad_norm": 6.3822340965271,
      "learning_rate": 2.166868685671044e-06,
      "loss": 1.619,
      "step": 21900
    },
    {
      "epoch": 3.7271412775367865,
      "grad_norm": 5.389329433441162,
      "learning_rate": 2.1614234293613586e-06,
      "loss": 1.6746,
      "step": 21910
    },
    {
      "epoch": 3.728842391766607,
      "grad_norm": 5.0422234535217285,
      "learning_rate": 2.155983712742517e-06,
      "loss": 1.578,
      "step": 21920
    },
    {
      "epoch": 3.730543505996428,
      "grad_norm": 4.832094192504883,
      "learning_rate": 2.1505495424205793e-06,
      "loss": 1.6345,
      "step": 21930
    },
    {
      "epoch": 3.7322446202262483,
      "grad_norm": 5.112161159515381,
      "learning_rate": 2.1451209249948704e-06,
      "loss": 1.5959,
      "step": 21940
    },
    {
      "epoch": 3.7339457344560687,
      "grad_norm": 5.519700050354004,
      "learning_rate": 2.139697867057968e-06,
      "loss": 1.548,
      "step": 21950
    },
    {
      "epoch": 3.735646848685889,
      "grad_norm": 6.192723274230957,
      "learning_rate": 2.134280375195699e-06,
      "loss": 1.4994,
      "step": 21960
    },
    {
      "epoch": 3.7373479629157096,
      "grad_norm": 5.92973518371582,
      "learning_rate": 2.12886845598714e-06,
      "loss": 1.5361,
      "step": 21970
    },
    {
      "epoch": 3.7390490771455305,
      "grad_norm": 5.719332695007324,
      "learning_rate": 2.123462116004588e-06,
      "loss": 1.5925,
      "step": 21980
    },
    {
      "epoch": 3.740750191375351,
      "grad_norm": 6.8935160636901855,
      "learning_rate": 2.11806136181357e-06,
      "loss": 1.5683,
      "step": 21990
    },
    {
      "epoch": 3.7424513056051714,
      "grad_norm": 6.860033988952637,
      "learning_rate": 2.1132054643143583e-06,
      "loss": 1.5645,
      "step": 22000
    },
    {
      "epoch": 3.744152419834992,
      "grad_norm": 6.525299072265625,
      "learning_rate": 2.1078153411909805e-06,
      "loss": 1.6685,
      "step": 22010
    },
    {
      "epoch": 3.7458535340648123,
      "grad_norm": 6.222875595092773,
      "learning_rate": 2.102430822860775e-06,
      "loss": 1.5352,
      "step": 22020
    },
    {
      "epoch": 3.747554648294633,
      "grad_norm": 8.409226417541504,
      "learning_rate": 2.0970519158627674e-06,
      "loss": 1.595,
      "step": 22030
    },
    {
      "epoch": 3.7492557625244536,
      "grad_norm": 6.118697643280029,
      "learning_rate": 2.0916786267291696e-06,
      "loss": 1.6523,
      "step": 22040
    },
    {
      "epoch": 3.750956876754274,
      "grad_norm": 5.883594989776611,
      "learning_rate": 2.0863109619853705e-06,
      "loss": 1.5977,
      "step": 22050
    },
    {
      "epoch": 3.7526579909840945,
      "grad_norm": 6.259503364562988,
      "learning_rate": 2.080948928149927e-06,
      "loss": 1.6087,
      "step": 22060
    },
    {
      "epoch": 3.754359105213915,
      "grad_norm": 5.8465447425842285,
      "learning_rate": 2.0755925317345633e-06,
      "loss": 1.6201,
      "step": 22070
    },
    {
      "epoch": 3.756060219443736,
      "grad_norm": 5.4746832847595215,
      "learning_rate": 2.0702417792441523e-06,
      "loss": 1.604,
      "step": 22080
    },
    {
      "epoch": 3.7577613336735562,
      "grad_norm": 5.659877300262451,
      "learning_rate": 2.0648966771767136e-06,
      "loss": 1.5995,
      "step": 22090
    },
    {
      "epoch": 3.7594624479033767,
      "grad_norm": 6.075840473175049,
      "learning_rate": 2.059557232023406e-06,
      "loss": 1.6702,
      "step": 22100
    },
    {
      "epoch": 3.761163562133197,
      "grad_norm": 5.514547824859619,
      "learning_rate": 2.054223450268518e-06,
      "loss": 1.5811,
      "step": 22110
    },
    {
      "epoch": 3.7628646763630176,
      "grad_norm": 5.915891647338867,
      "learning_rate": 2.04889533838946e-06,
      "loss": 1.5995,
      "step": 22120
    },
    {
      "epoch": 3.7645657905928385,
      "grad_norm": 6.892215251922607,
      "learning_rate": 2.0435729028567573e-06,
      "loss": 1.6316,
      "step": 22130
    },
    {
      "epoch": 3.766266904822659,
      "grad_norm": 6.2456889152526855,
      "learning_rate": 2.0382561501340425e-06,
      "loss": 1.5633,
      "step": 22140
    },
    {
      "epoch": 3.7679680190524794,
      "grad_norm": 4.70587158203125,
      "learning_rate": 2.0329450866780456e-06,
      "loss": 1.5695,
      "step": 22150
    },
    {
      "epoch": 3.7696691332823,
      "grad_norm": 6.266761302947998,
      "learning_rate": 2.0276397189385885e-06,
      "loss": 1.6148,
      "step": 22160
    },
    {
      "epoch": 3.7713702475121202,
      "grad_norm": 5.643571376800537,
      "learning_rate": 2.0223400533585724e-06,
      "loss": 1.576,
      "step": 22170
    },
    {
      "epoch": 3.773071361741941,
      "grad_norm": 5.848235607147217,
      "learning_rate": 2.0170460963739823e-06,
      "loss": 1.5702,
      "step": 22180
    },
    {
      "epoch": 3.7747724759717616,
      "grad_norm": 5.9626688957214355,
      "learning_rate": 2.011757854413863e-06,
      "loss": 1.4829,
      "step": 22190
    },
    {
      "epoch": 3.776473590201582,
      "grad_norm": 6.014424800872803,
      "learning_rate": 2.006475333900321e-06,
      "loss": 1.5723,
      "step": 22200
    },
    {
      "epoch": 3.7781747044314025,
      "grad_norm": 4.955949783325195,
      "learning_rate": 2.0011985412485175e-06,
      "loss": 1.7041,
      "step": 22210
    },
    {
      "epoch": 3.779875818661223,
      "grad_norm": 5.689254283905029,
      "learning_rate": 1.99592748286665e-06,
      "loss": 1.5068,
      "step": 22220
    },
    {
      "epoch": 3.781576932891044,
      "grad_norm": 5.831399440765381,
      "learning_rate": 1.990662165155958e-06,
      "loss": 1.5543,
      "step": 22230
    },
    {
      "epoch": 3.7832780471208642,
      "grad_norm": 6.465705871582031,
      "learning_rate": 1.985402594510713e-06,
      "loss": 1.5824,
      "step": 22240
    },
    {
      "epoch": 3.7849791613506847,
      "grad_norm": 5.482491970062256,
      "learning_rate": 1.9801487773182003e-06,
      "loss": 1.7232,
      "step": 22250
    },
    {
      "epoch": 3.786680275580505,
      "grad_norm": 4.863794803619385,
      "learning_rate": 1.9749007199587203e-06,
      "loss": 1.6624,
      "step": 22260
    },
    {
      "epoch": 3.7883813898103256,
      "grad_norm": 6.264886856079102,
      "learning_rate": 1.9696584288055808e-06,
      "loss": 1.6755,
      "step": 22270
    },
    {
      "epoch": 3.7900825040401465,
      "grad_norm": 6.4072699546813965,
      "learning_rate": 1.9644219102250836e-06,
      "loss": 1.5626,
      "step": 22280
    },
    {
      "epoch": 3.791783618269967,
      "grad_norm": 6.039673805236816,
      "learning_rate": 1.95919117057652e-06,
      "loss": 1.6024,
      "step": 22290
    },
    {
      "epoch": 3.7934847324997873,
      "grad_norm": 6.067235469818115,
      "learning_rate": 1.953966216212173e-06,
      "loss": 1.6003,
      "step": 22300
    },
    {
      "epoch": 3.795185846729608,
      "grad_norm": 6.459325313568115,
      "learning_rate": 1.948747053477284e-06,
      "loss": 1.5327,
      "step": 22310
    },
    {
      "epoch": 3.7968869609594282,
      "grad_norm": 6.460668087005615,
      "learning_rate": 1.943533688710072e-06,
      "loss": 1.534,
      "step": 22320
    },
    {
      "epoch": 3.798588075189249,
      "grad_norm": 6.536547660827637,
      "learning_rate": 1.938326128241712e-06,
      "loss": 1.5894,
      "step": 22330
    },
    {
      "epoch": 3.8002891894190696,
      "grad_norm": 6.6990251541137695,
      "learning_rate": 1.9331243783963274e-06,
      "loss": 1.582,
      "step": 22340
    },
    {
      "epoch": 3.80199030364889,
      "grad_norm": 6.024599552154541,
      "learning_rate": 1.9279284454909928e-06,
      "loss": 1.5075,
      "step": 22350
    },
    {
      "epoch": 3.8036914178787105,
      "grad_norm": 4.894837379455566,
      "learning_rate": 1.9227383358357116e-06,
      "loss": 1.5959,
      "step": 22360
    },
    {
      "epoch": 3.805392532108531,
      "grad_norm": 5.122894287109375,
      "learning_rate": 1.9175540557334166e-06,
      "loss": 1.681,
      "step": 22370
    },
    {
      "epoch": 3.807093646338352,
      "grad_norm": 6.1250410079956055,
      "learning_rate": 1.912375611479963e-06,
      "loss": 1.6894,
      "step": 22380
    },
    {
      "epoch": 3.8087947605681722,
      "grad_norm": 6.071530818939209,
      "learning_rate": 1.907203009364117e-06,
      "loss": 1.5696,
      "step": 22390
    },
    {
      "epoch": 3.8104958747979927,
      "grad_norm": 5.952277660369873,
      "learning_rate": 1.9020362556675514e-06,
      "loss": 1.6921,
      "step": 22400
    },
    {
      "epoch": 3.812196989027813,
      "grad_norm": 5.401397705078125,
      "learning_rate": 1.8968753566648353e-06,
      "loss": 1.6918,
      "step": 22410
    },
    {
      "epoch": 3.8138981032576336,
      "grad_norm": 5.378725051879883,
      "learning_rate": 1.8917203186234292e-06,
      "loss": 1.5513,
      "step": 22420
    },
    {
      "epoch": 3.8155992174874545,
      "grad_norm": 4.487756729125977,
      "learning_rate": 1.8865711478036758e-06,
      "loss": 1.6278,
      "step": 22430
    },
    {
      "epoch": 3.817300331717275,
      "grad_norm": 5.296054840087891,
      "learning_rate": 1.8814278504587913e-06,
      "loss": 1.5496,
      "step": 22440
    },
    {
      "epoch": 3.8190014459470953,
      "grad_norm": 6.797940254211426,
      "learning_rate": 1.8762904328348596e-06,
      "loss": 1.5314,
      "step": 22450
    },
    {
      "epoch": 3.820702560176916,
      "grad_norm": 5.526535987854004,
      "learning_rate": 1.8711589011708233e-06,
      "loss": 1.6617,
      "step": 22460
    },
    {
      "epoch": 3.8224036744067362,
      "grad_norm": 6.257224082946777,
      "learning_rate": 1.8660332616984825e-06,
      "loss": 1.6189,
      "step": 22470
    },
    {
      "epoch": 3.824104788636557,
      "grad_norm": 6.83498477935791,
      "learning_rate": 1.8609135206424742e-06,
      "loss": 1.6298,
      "step": 22480
    },
    {
      "epoch": 3.8258059028663776,
      "grad_norm": 6.862575531005859,
      "learning_rate": 1.8557996842202764e-06,
      "loss": 1.6964,
      "step": 22490
    },
    {
      "epoch": 3.827507017096198,
      "grad_norm": 5.854297161102295,
      "learning_rate": 1.8506917586421957e-06,
      "loss": 1.6054,
      "step": 22500
    },
    {
      "epoch": 3.8292081313260184,
      "grad_norm": 7.284146308898926,
      "learning_rate": 1.8455897501113618e-06,
      "loss": 1.5447,
      "step": 22510
    },
    {
      "epoch": 3.830909245555839,
      "grad_norm": 6.290617942810059,
      "learning_rate": 1.8404936648237115e-06,
      "loss": 1.6434,
      "step": 22520
    },
    {
      "epoch": 3.83261035978566,
      "grad_norm": 7.121439456939697,
      "learning_rate": 1.8354035089680003e-06,
      "loss": 1.5908,
      "step": 22530
    },
    {
      "epoch": 3.8343114740154802,
      "grad_norm": 5.1764631271362305,
      "learning_rate": 1.8303192887257742e-06,
      "loss": 1.5425,
      "step": 22540
    },
    {
      "epoch": 3.8360125882453007,
      "grad_norm": 5.700981616973877,
      "learning_rate": 1.8252410102713733e-06,
      "loss": 1.5265,
      "step": 22550
    },
    {
      "epoch": 3.837713702475121,
      "grad_norm": 5.483993053436279,
      "learning_rate": 1.820168679771923e-06,
      "loss": 1.6314,
      "step": 22560
    },
    {
      "epoch": 3.8394148167049416,
      "grad_norm": 4.648709297180176,
      "learning_rate": 1.815102303387322e-06,
      "loss": 1.6124,
      "step": 22570
    },
    {
      "epoch": 3.8411159309347624,
      "grad_norm": 6.244708061218262,
      "learning_rate": 1.8100418872702444e-06,
      "loss": 1.6019,
      "step": 22580
    },
    {
      "epoch": 3.842817045164583,
      "grad_norm": 6.450686454772949,
      "learning_rate": 1.8049874375661229e-06,
      "loss": 1.595,
      "step": 22590
    },
    {
      "epoch": 3.8445181593944033,
      "grad_norm": 6.041715621948242,
      "learning_rate": 1.7999389604131399e-06,
      "loss": 1.5446,
      "step": 22600
    },
    {
      "epoch": 3.8462192736242238,
      "grad_norm": 5.104908466339111,
      "learning_rate": 1.7948964619422305e-06,
      "loss": 1.6467,
      "step": 22610
    },
    {
      "epoch": 3.847920387854044,
      "grad_norm": 6.871789932250977,
      "learning_rate": 1.7898599482770678e-06,
      "loss": 1.6949,
      "step": 22620
    },
    {
      "epoch": 3.849621502083865,
      "grad_norm": 6.004401683807373,
      "learning_rate": 1.7848294255340538e-06,
      "loss": 1.6199,
      "step": 22630
    },
    {
      "epoch": 3.8513226163136856,
      "grad_norm": 7.1538310050964355,
      "learning_rate": 1.779804899822323e-06,
      "loss": 1.6496,
      "step": 22640
    },
    {
      "epoch": 3.853023730543506,
      "grad_norm": 6.285670757293701,
      "learning_rate": 1.7747863772437187e-06,
      "loss": 1.5965,
      "step": 22650
    },
    {
      "epoch": 3.8547248447733264,
      "grad_norm": 6.898594856262207,
      "learning_rate": 1.7697738638927976e-06,
      "loss": 1.5917,
      "step": 22660
    },
    {
      "epoch": 3.856425959003147,
      "grad_norm": 5.652676582336426,
      "learning_rate": 1.7647673658568186e-06,
      "loss": 1.5332,
      "step": 22670
    },
    {
      "epoch": 3.8581270732329678,
      "grad_norm": 5.607017993927002,
      "learning_rate": 1.7597668892157346e-06,
      "loss": 1.5489,
      "step": 22680
    },
    {
      "epoch": 3.859828187462788,
      "grad_norm": 6.1689958572387695,
      "learning_rate": 1.7547724400421868e-06,
      "loss": 1.6221,
      "step": 22690
    },
    {
      "epoch": 3.8615293016926087,
      "grad_norm": 5.738541126251221,
      "learning_rate": 1.7497840244014958e-06,
      "loss": 1.5205,
      "step": 22700
    },
    {
      "epoch": 3.863230415922429,
      "grad_norm": 5.920195579528809,
      "learning_rate": 1.7448016483516562e-06,
      "loss": 1.5938,
      "step": 22710
    },
    {
      "epoch": 3.8649315301522496,
      "grad_norm": 5.464554786682129,
      "learning_rate": 1.7398253179433267e-06,
      "loss": 1.6791,
      "step": 22720
    },
    {
      "epoch": 3.8666326443820704,
      "grad_norm": 6.0924835205078125,
      "learning_rate": 1.7348550392198253e-06,
      "loss": 1.6658,
      "step": 22730
    },
    {
      "epoch": 3.868333758611891,
      "grad_norm": 6.268444061279297,
      "learning_rate": 1.729890818217118e-06,
      "loss": 1.6934,
      "step": 22740
    },
    {
      "epoch": 3.8700348728417113,
      "grad_norm": 4.383749008178711,
      "learning_rate": 1.7249326609638205e-06,
      "loss": 1.6304,
      "step": 22750
    },
    {
      "epoch": 3.8717359870715318,
      "grad_norm": 5.348143100738525,
      "learning_rate": 1.7199805734811787e-06,
      "loss": 1.6227,
      "step": 22760
    },
    {
      "epoch": 3.873437101301352,
      "grad_norm": 6.921054840087891,
      "learning_rate": 1.715034561783069e-06,
      "loss": 1.5271,
      "step": 22770
    },
    {
      "epoch": 3.875138215531173,
      "grad_norm": 6.553023338317871,
      "learning_rate": 1.7100946318759897e-06,
      "loss": 1.654,
      "step": 22780
    },
    {
      "epoch": 3.8768393297609935,
      "grad_norm": 6.585522651672363,
      "learning_rate": 1.705160789759052e-06,
      "loss": 1.5698,
      "step": 22790
    },
    {
      "epoch": 3.878540443990814,
      "grad_norm": 6.700231075286865,
      "learning_rate": 1.7002330414239764e-06,
      "loss": 1.6655,
      "step": 22800
    },
    {
      "epoch": 3.8802415582206344,
      "grad_norm": 5.89600133895874,
      "learning_rate": 1.6953113928550818e-06,
      "loss": 1.6149,
      "step": 22810
    },
    {
      "epoch": 3.881942672450455,
      "grad_norm": 5.654646873474121,
      "learning_rate": 1.6903958500292783e-06,
      "loss": 1.653,
      "step": 22820
    },
    {
      "epoch": 3.8836437866802758,
      "grad_norm": 5.169059753417969,
      "learning_rate": 1.6854864189160622e-06,
      "loss": 1.6146,
      "step": 22830
    },
    {
      "epoch": 3.885344900910096,
      "grad_norm": 5.4262261390686035,
      "learning_rate": 1.6805831054775081e-06,
      "loss": 1.5751,
      "step": 22840
    },
    {
      "epoch": 3.8870460151399167,
      "grad_norm": 5.427737712860107,
      "learning_rate": 1.6756859156682615e-06,
      "loss": 1.5781,
      "step": 22850
    },
    {
      "epoch": 3.888747129369737,
      "grad_norm": 5.8518805503845215,
      "learning_rate": 1.6707948554355268e-06,
      "loss": 1.569,
      "step": 22860
    },
    {
      "epoch": 3.8904482435995575,
      "grad_norm": 5.492292881011963,
      "learning_rate": 1.6659099307190734e-06,
      "loss": 1.644,
      "step": 22870
    },
    {
      "epoch": 3.8921493578293784,
      "grad_norm": 5.627607822418213,
      "learning_rate": 1.6610311474512128e-06,
      "loss": 1.699,
      "step": 22880
    },
    {
      "epoch": 3.893850472059199,
      "grad_norm": 5.437195777893066,
      "learning_rate": 1.656158511556802e-06,
      "loss": 1.5462,
      "step": 22890
    },
    {
      "epoch": 3.8955515862890193,
      "grad_norm": 6.071361541748047,
      "learning_rate": 1.6512920289532255e-06,
      "loss": 1.5707,
      "step": 22900
    },
    {
      "epoch": 3.8972527005188398,
      "grad_norm": 5.888106822967529,
      "learning_rate": 1.6464317055504013e-06,
      "loss": 1.574,
      "step": 22910
    },
    {
      "epoch": 3.89895381474866,
      "grad_norm": 6.413266658782959,
      "learning_rate": 1.6415775472507699e-06,
      "loss": 1.6255,
      "step": 22920
    },
    {
      "epoch": 3.900654928978481,
      "grad_norm": 6.208846569061279,
      "learning_rate": 1.636729559949279e-06,
      "loss": 1.5323,
      "step": 22930
    },
    {
      "epoch": 3.9023560432083015,
      "grad_norm": 6.5904765129089355,
      "learning_rate": 1.6318877495333865e-06,
      "loss": 1.6064,
      "step": 22940
    },
    {
      "epoch": 3.904057157438122,
      "grad_norm": 6.936824798583984,
      "learning_rate": 1.627052121883045e-06,
      "loss": 1.5421,
      "step": 22950
    },
    {
      "epoch": 3.9057582716679424,
      "grad_norm": 6.398342132568359,
      "learning_rate": 1.622222682870702e-06,
      "loss": 1.5537,
      "step": 22960
    },
    {
      "epoch": 3.907459385897763,
      "grad_norm": 7.79954195022583,
      "learning_rate": 1.6173994383612863e-06,
      "loss": 1.5808,
      "step": 22970
    },
    {
      "epoch": 3.9091605001275838,
      "grad_norm": 5.884366512298584,
      "learning_rate": 1.612582394212211e-06,
      "loss": 1.6081,
      "step": 22980
    },
    {
      "epoch": 3.910861614357404,
      "grad_norm": 5.413835525512695,
      "learning_rate": 1.6077715562733492e-06,
      "loss": 1.6782,
      "step": 22990
    },
    {
      "epoch": 3.9125627285872246,
      "grad_norm": 4.960882663726807,
      "learning_rate": 1.602966930387044e-06,
      "loss": 1.5789,
      "step": 23000
    },
    {
      "epoch": 3.914263842817045,
      "grad_norm": 4.974776744842529,
      "learning_rate": 1.598168522388093e-06,
      "loss": 1.5556,
      "step": 23010
    },
    {
      "epoch": 3.9159649570468655,
      "grad_norm": 6.388287544250488,
      "learning_rate": 1.5933763381037416e-06,
      "loss": 1.52,
      "step": 23020
    },
    {
      "epoch": 3.9176660712766864,
      "grad_norm": 6.267454147338867,
      "learning_rate": 1.5885903833536774e-06,
      "loss": 1.5922,
      "step": 23030
    },
    {
      "epoch": 3.919367185506507,
      "grad_norm": 5.785188674926758,
      "learning_rate": 1.5838106639500258e-06,
      "loss": 1.6034,
      "step": 23040
    },
    {
      "epoch": 3.9210682997363273,
      "grad_norm": 6.028636455535889,
      "learning_rate": 1.579037185697336e-06,
      "loss": 1.5497,
      "step": 23050
    },
    {
      "epoch": 3.9227694139661478,
      "grad_norm": 7.901721477508545,
      "learning_rate": 1.5742699543925796e-06,
      "loss": 1.5244,
      "step": 23060
    },
    {
      "epoch": 3.924470528195968,
      "grad_norm": 6.526932239532471,
      "learning_rate": 1.5695089758251414e-06,
      "loss": 1.6008,
      "step": 23070
    },
    {
      "epoch": 3.926171642425789,
      "grad_norm": 5.383223056793213,
      "learning_rate": 1.5647542557768127e-06,
      "loss": 1.6404,
      "step": 23080
    },
    {
      "epoch": 3.9278727566556095,
      "grad_norm": 5.718662738800049,
      "learning_rate": 1.5600058000217844e-06,
      "loss": 1.613,
      "step": 23090
    },
    {
      "epoch": 3.92957387088543,
      "grad_norm": 5.968343257904053,
      "learning_rate": 1.5552636143266408e-06,
      "loss": 1.7002,
      "step": 23100
    },
    {
      "epoch": 3.9312749851152504,
      "grad_norm": 5.775221347808838,
      "learning_rate": 1.550527704450351e-06,
      "loss": 1.5685,
      "step": 23110
    },
    {
      "epoch": 3.932976099345071,
      "grad_norm": 5.761837482452393,
      "learning_rate": 1.5457980761442623e-06,
      "loss": 1.6043,
      "step": 23120
    },
    {
      "epoch": 3.9346772135748918,
      "grad_norm": 3.1403274536132812,
      "learning_rate": 1.541074735152094e-06,
      "loss": 1.5386,
      "step": 23130
    },
    {
      "epoch": 3.936378327804712,
      "grad_norm": 7.0953545570373535,
      "learning_rate": 1.5363576872099288e-06,
      "loss": 1.5012,
      "step": 23140
    },
    {
      "epoch": 3.9380794420345326,
      "grad_norm": 5.90140438079834,
      "learning_rate": 1.5316469380462119e-06,
      "loss": 1.5493,
      "step": 23150
    },
    {
      "epoch": 3.939780556264353,
      "grad_norm": 5.884961128234863,
      "learning_rate": 1.5269424933817336e-06,
      "loss": 1.6243,
      "step": 23160
    },
    {
      "epoch": 3.9414816704941735,
      "grad_norm": 6.109187602996826,
      "learning_rate": 1.5222443589296305e-06,
      "loss": 1.6247,
      "step": 23170
    },
    {
      "epoch": 3.9431827847239944,
      "grad_norm": 5.5967698097229,
      "learning_rate": 1.5175525403953758e-06,
      "loss": 1.6835,
      "step": 23180
    },
    {
      "epoch": 3.944883898953815,
      "grad_norm": 5.781443119049072,
      "learning_rate": 1.5128670434767698e-06,
      "loss": 1.6234,
      "step": 23190
    },
    {
      "epoch": 3.9465850131836353,
      "grad_norm": 8.294487953186035,
      "learning_rate": 1.5081878738639384e-06,
      "loss": 1.6517,
      "step": 23200
    },
    {
      "epoch": 3.9482861274134557,
      "grad_norm": 6.1851091384887695,
      "learning_rate": 1.5035150372393257e-06,
      "loss": 1.5225,
      "step": 23210
    },
    {
      "epoch": 3.949987241643276,
      "grad_norm": 7.144291877746582,
      "learning_rate": 1.4988485392776821e-06,
      "loss": 1.7338,
      "step": 23220
    },
    {
      "epoch": 3.951688355873097,
      "grad_norm": 5.25929069519043,
      "learning_rate": 1.49418838564606e-06,
      "loss": 1.5546,
      "step": 23230
    },
    {
      "epoch": 3.9533894701029175,
      "grad_norm": 6.643313884735107,
      "learning_rate": 1.4895345820038072e-06,
      "loss": 1.5909,
      "step": 23240
    },
    {
      "epoch": 3.955090584332738,
      "grad_norm": 5.798901081085205,
      "learning_rate": 1.4848871340025624e-06,
      "loss": 1.5675,
      "step": 23250
    },
    {
      "epoch": 3.9567916985625584,
      "grad_norm": 5.724977493286133,
      "learning_rate": 1.4802460472862405e-06,
      "loss": 1.6318,
      "step": 23260
    },
    {
      "epoch": 3.958492812792379,
      "grad_norm": 5.830146789550781,
      "learning_rate": 1.4756113274910413e-06,
      "loss": 1.6487,
      "step": 23270
    },
    {
      "epoch": 3.9601939270221997,
      "grad_norm": 5.694972038269043,
      "learning_rate": 1.4709829802454207e-06,
      "loss": 1.5821,
      "step": 23280
    },
    {
      "epoch": 3.96189504125202,
      "grad_norm": 6.186990261077881,
      "learning_rate": 1.466361011170103e-06,
      "loss": 1.5545,
      "step": 23290
    },
    {
      "epoch": 3.9635961554818406,
      "grad_norm": 6.006585597991943,
      "learning_rate": 1.461745425878064e-06,
      "loss": 1.5724,
      "step": 23300
    },
    {
      "epoch": 3.965297269711661,
      "grad_norm": 5.7341742515563965,
      "learning_rate": 1.4571362299745269e-06,
      "loss": 1.5861,
      "step": 23310
    },
    {
      "epoch": 3.9669983839414815,
      "grad_norm": 5.754456043243408,
      "learning_rate": 1.4525334290569599e-06,
      "loss": 1.6198,
      "step": 23320
    },
    {
      "epoch": 3.9686994981713024,
      "grad_norm": 5.866809844970703,
      "learning_rate": 1.4479370287150595e-06,
      "loss": 1.7648,
      "step": 23330
    },
    {
      "epoch": 3.970400612401123,
      "grad_norm": 4.520585060119629,
      "learning_rate": 1.4433470345307512e-06,
      "loss": 1.5776,
      "step": 23340
    },
    {
      "epoch": 3.9721017266309433,
      "grad_norm": 6.06074333190918,
      "learning_rate": 1.4387634520781819e-06,
      "loss": 1.565,
      "step": 23350
    },
    {
      "epoch": 3.9738028408607637,
      "grad_norm": 6.441244125366211,
      "learning_rate": 1.4341862869237102e-06,
      "loss": 1.552,
      "step": 23360
    },
    {
      "epoch": 3.975503955090584,
      "grad_norm": 5.399563789367676,
      "learning_rate": 1.4296155446259024e-06,
      "loss": 1.4966,
      "step": 23370
    },
    {
      "epoch": 3.977205069320405,
      "grad_norm": 6.354936122894287,
      "learning_rate": 1.4250512307355253e-06,
      "loss": 1.6199,
      "step": 23380
    },
    {
      "epoch": 3.9789061835502255,
      "grad_norm": 5.502198219299316,
      "learning_rate": 1.4204933507955374e-06,
      "loss": 1.6908,
      "step": 23390
    },
    {
      "epoch": 3.980607297780046,
      "grad_norm": 5.9025959968566895,
      "learning_rate": 1.415941910341086e-06,
      "loss": 1.5611,
      "step": 23400
    },
    {
      "epoch": 3.9823084120098664,
      "grad_norm": 6.057744979858398,
      "learning_rate": 1.4113969148994958e-06,
      "loss": 1.5161,
      "step": 23410
    },
    {
      "epoch": 3.984009526239687,
      "grad_norm": 5.675593852996826,
      "learning_rate": 1.406858369990267e-06,
      "loss": 1.5843,
      "step": 23420
    },
    {
      "epoch": 3.9857106404695077,
      "grad_norm": 5.631038188934326,
      "learning_rate": 1.402326281125061e-06,
      "loss": 1.6411,
      "step": 23430
    },
    {
      "epoch": 3.987411754699328,
      "grad_norm": 6.619244575500488,
      "learning_rate": 1.397800653807709e-06,
      "loss": 1.5863,
      "step": 23440
    },
    {
      "epoch": 3.9891128689291486,
      "grad_norm": 5.93541955947876,
      "learning_rate": 1.393281493534185e-06,
      "loss": 1.6251,
      "step": 23450
    },
    {
      "epoch": 3.990813983158969,
      "grad_norm": 6.078390598297119,
      "learning_rate": 1.3887688057926153e-06,
      "loss": 1.5593,
      "step": 23460
    },
    {
      "epoch": 3.9925150973887895,
      "grad_norm": 6.589411735534668,
      "learning_rate": 1.384262596063262e-06,
      "loss": 1.6533,
      "step": 23470
    },
    {
      "epoch": 3.9942162116186104,
      "grad_norm": 5.347785949707031,
      "learning_rate": 1.3797628698185241e-06,
      "loss": 1.5861,
      "step": 23480
    },
    {
      "epoch": 3.995917325848431,
      "grad_norm": 6.955543041229248,
      "learning_rate": 1.3752696325229234e-06,
      "loss": 1.6355,
      "step": 23490
    },
    {
      "epoch": 3.9976184400782513,
      "grad_norm": 6.043464660644531,
      "learning_rate": 1.3707828896331037e-06,
      "loss": 1.649,
      "step": 23500
    },
    {
      "epoch": 3.9993195543080717,
      "grad_norm": 5.864923000335693,
      "learning_rate": 1.3663026465978225e-06,
      "loss": 1.6235,
      "step": 23510
    },
    {
      "epoch": 4.001020668537892,
      "grad_norm": 5.533880710601807,
      "learning_rate": 1.3618289088579407e-06,
      "loss": 1.6016,
      "step": 23520
    },
    {
      "epoch": 4.002721782767713,
      "grad_norm": 5.245782852172852,
      "learning_rate": 1.3573616818464215e-06,
      "loss": 1.5046,
      "step": 23530
    },
    {
      "epoch": 4.004422896997533,
      "grad_norm": 7.184761047363281,
      "learning_rate": 1.3529009709883203e-06,
      "loss": 1.5545,
      "step": 23540
    },
    {
      "epoch": 4.006124011227354,
      "grad_norm": 5.881488800048828,
      "learning_rate": 1.3484467817007815e-06,
      "loss": 1.5075,
      "step": 23550
    },
    {
      "epoch": 4.007825125457175,
      "grad_norm": 5.9063920974731445,
      "learning_rate": 1.3439991193930276e-06,
      "loss": 1.5791,
      "step": 23560
    },
    {
      "epoch": 4.009526239686995,
      "grad_norm": 6.223015308380127,
      "learning_rate": 1.3395579894663567e-06,
      "loss": 1.5318,
      "step": 23570
    },
    {
      "epoch": 4.011227353916816,
      "grad_norm": 5.480140686035156,
      "learning_rate": 1.3351233973141272e-06,
      "loss": 1.636,
      "step": 23580
    },
    {
      "epoch": 4.012928468146636,
      "grad_norm": 5.175749778747559,
      "learning_rate": 1.3306953483217657e-06,
      "loss": 1.5678,
      "step": 23590
    },
    {
      "epoch": 4.014629582376457,
      "grad_norm": 5.125709056854248,
      "learning_rate": 1.3262738478667487e-06,
      "loss": 1.6087,
      "step": 23600
    },
    {
      "epoch": 4.0163306966062775,
      "grad_norm": 6.26732873916626,
      "learning_rate": 1.321858901318604e-06,
      "loss": 1.5714,
      "step": 23610
    },
    {
      "epoch": 4.0180318108360975,
      "grad_norm": 6.341707706451416,
      "learning_rate": 1.317450514038895e-06,
      "loss": 1.4367,
      "step": 23620
    },
    {
      "epoch": 4.019732925065918,
      "grad_norm": 5.748213291168213,
      "learning_rate": 1.3130486913812236e-06,
      "loss": 1.5628,
      "step": 23630
    },
    {
      "epoch": 4.021434039295738,
      "grad_norm": 5.659625053405762,
      "learning_rate": 1.3086534386912171e-06,
      "loss": 1.4844,
      "step": 23640
    },
    {
      "epoch": 4.023135153525559,
      "grad_norm": 5.962423801422119,
      "learning_rate": 1.3042647613065255e-06,
      "loss": 1.5308,
      "step": 23650
    },
    {
      "epoch": 4.02483626775538,
      "grad_norm": 6.780612945556641,
      "learning_rate": 1.2998826645568116e-06,
      "loss": 1.5014,
      "step": 23660
    },
    {
      "epoch": 4.0265373819852,
      "grad_norm": 5.359397888183594,
      "learning_rate": 1.2955071537637503e-06,
      "loss": 1.503,
      "step": 23670
    },
    {
      "epoch": 4.028238496215021,
      "grad_norm": 7.911985397338867,
      "learning_rate": 1.2911382342410145e-06,
      "loss": 1.4167,
      "step": 23680
    },
    {
      "epoch": 4.029939610444841,
      "grad_norm": 5.2868876457214355,
      "learning_rate": 1.2867759112942756e-06,
      "loss": 1.4317,
      "step": 23690
    },
    {
      "epoch": 4.031640724674662,
      "grad_norm": 6.8723978996276855,
      "learning_rate": 1.2824201902211918e-06,
      "loss": 1.5133,
      "step": 23700
    },
    {
      "epoch": 4.033341838904483,
      "grad_norm": 6.042233943939209,
      "learning_rate": 1.2780710763114038e-06,
      "loss": 1.587,
      "step": 23710
    },
    {
      "epoch": 4.035042953134303,
      "grad_norm": 5.142142295837402,
      "learning_rate": 1.2737285748465322e-06,
      "loss": 1.4903,
      "step": 23720
    },
    {
      "epoch": 4.036744067364124,
      "grad_norm": 5.050568103790283,
      "learning_rate": 1.2693926911001645e-06,
      "loss": 1.5208,
      "step": 23730
    },
    {
      "epoch": 4.038445181593944,
      "grad_norm": 6.0770182609558105,
      "learning_rate": 1.2650634303378506e-06,
      "loss": 1.5963,
      "step": 23740
    },
    {
      "epoch": 4.040146295823765,
      "grad_norm": 4.7540974617004395,
      "learning_rate": 1.2607407978170987e-06,
      "loss": 1.5353,
      "step": 23750
    },
    {
      "epoch": 4.0418474100535855,
      "grad_norm": 5.933732509613037,
      "learning_rate": 1.2564247987873672e-06,
      "loss": 1.5335,
      "step": 23760
    },
    {
      "epoch": 4.0435485242834055,
      "grad_norm": 6.690618991851807,
      "learning_rate": 1.2521154384900597e-06,
      "loss": 1.5351,
      "step": 23770
    },
    {
      "epoch": 4.045249638513226,
      "grad_norm": 6.390956401824951,
      "learning_rate": 1.247812722158516e-06,
      "loss": 1.6059,
      "step": 23780
    },
    {
      "epoch": 4.046950752743046,
      "grad_norm": 5.966754913330078,
      "learning_rate": 1.2435166550180092e-06,
      "loss": 1.5127,
      "step": 23790
    },
    {
      "epoch": 4.048651866972867,
      "grad_norm": 6.638439655303955,
      "learning_rate": 1.239227242285736e-06,
      "loss": 1.6776,
      "step": 23800
    },
    {
      "epoch": 4.050352981202688,
      "grad_norm": 5.8886895179748535,
      "learning_rate": 1.2349444891708132e-06,
      "loss": 1.5403,
      "step": 23810
    },
    {
      "epoch": 4.052054095432508,
      "grad_norm": 6.793363094329834,
      "learning_rate": 1.2306684008742683e-06,
      "loss": 1.5017,
      "step": 23820
    },
    {
      "epoch": 4.053755209662329,
      "grad_norm": 4.266611576080322,
      "learning_rate": 1.2263989825890345e-06,
      "loss": 1.6085,
      "step": 23830
    },
    {
      "epoch": 4.055456323892149,
      "grad_norm": 6.3949432373046875,
      "learning_rate": 1.2221362394999506e-06,
      "loss": 1.5812,
      "step": 23840
    },
    {
      "epoch": 4.05715743812197,
      "grad_norm": 6.3855204582214355,
      "learning_rate": 1.2178801767837432e-06,
      "loss": 1.5012,
      "step": 23850
    },
    {
      "epoch": 4.058858552351791,
      "grad_norm": 5.799639701843262,
      "learning_rate": 1.2136307996090278e-06,
      "loss": 1.5891,
      "step": 23860
    },
    {
      "epoch": 4.060559666581611,
      "grad_norm": 5.730774402618408,
      "learning_rate": 1.2093881131362994e-06,
      "loss": 1.5129,
      "step": 23870
    },
    {
      "epoch": 4.062260780811432,
      "grad_norm": 6.441750526428223,
      "learning_rate": 1.2051521225179277e-06,
      "loss": 1.4608,
      "step": 23880
    },
    {
      "epoch": 4.063961895041252,
      "grad_norm": 6.180943965911865,
      "learning_rate": 1.2009228328981571e-06,
      "loss": 1.7483,
      "step": 23890
    },
    {
      "epoch": 4.065663009271073,
      "grad_norm": 5.97748327255249,
      "learning_rate": 1.1967002494130864e-06,
      "loss": 1.5257,
      "step": 23900
    },
    {
      "epoch": 4.0673641235008935,
      "grad_norm": 5.881425857543945,
      "learning_rate": 1.1924843771906737e-06,
      "loss": 1.6079,
      "step": 23910
    },
    {
      "epoch": 4.0690652377307135,
      "grad_norm": 6.427362442016602,
      "learning_rate": 1.1882752213507273e-06,
      "loss": 1.4936,
      "step": 23920
    },
    {
      "epoch": 4.070766351960534,
      "grad_norm": 6.745382785797119,
      "learning_rate": 1.1840727870048976e-06,
      "loss": 1.6802,
      "step": 23930
    },
    {
      "epoch": 4.072467466190354,
      "grad_norm": 6.204290866851807,
      "learning_rate": 1.1798770792566723e-06,
      "loss": 1.5797,
      "step": 23940
    },
    {
      "epoch": 4.074168580420175,
      "grad_norm": 4.805713653564453,
      "learning_rate": 1.1756881032013753e-06,
      "loss": 1.6186,
      "step": 23950
    },
    {
      "epoch": 4.075869694649996,
      "grad_norm": 5.432549476623535,
      "learning_rate": 1.171505863926146e-06,
      "loss": 1.5508,
      "step": 23960
    },
    {
      "epoch": 4.077570808879816,
      "grad_norm": 6.900859355926514,
      "learning_rate": 1.16733036650995e-06,
      "loss": 1.5908,
      "step": 23970
    },
    {
      "epoch": 4.079271923109637,
      "grad_norm": 4.752617359161377,
      "learning_rate": 1.1631616160235624e-06,
      "loss": 1.534,
      "step": 23980
    },
    {
      "epoch": 4.080973037339457,
      "grad_norm": 7.080214500427246,
      "learning_rate": 1.158999617529567e-06,
      "loss": 1.5779,
      "step": 23990
    },
    {
      "epoch": 4.082674151569278,
      "grad_norm": 6.091671943664551,
      "learning_rate": 1.154844376082343e-06,
      "loss": 1.605,
      "step": 24000
    },
    {
      "epoch": 4.084375265799099,
      "grad_norm": 6.665387153625488,
      "learning_rate": 1.1506958967280725e-06,
      "loss": 1.3982,
      "step": 24010
    },
    {
      "epoch": 4.086076380028919,
      "grad_norm": 7.309806823730469,
      "learning_rate": 1.1465541845047173e-06,
      "loss": 1.5491,
      "step": 24020
    },
    {
      "epoch": 4.08777749425874,
      "grad_norm": 6.847934722900391,
      "learning_rate": 1.1424192444420241e-06,
      "loss": 1.5341,
      "step": 24030
    },
    {
      "epoch": 4.08947860848856,
      "grad_norm": 3.901092767715454,
      "learning_rate": 1.1382910815615167e-06,
      "loss": 1.6616,
      "step": 24040
    },
    {
      "epoch": 4.091179722718381,
      "grad_norm": 5.525901794433594,
      "learning_rate": 1.134169700876486e-06,
      "loss": 1.5667,
      "step": 24050
    },
    {
      "epoch": 4.0928808369482015,
      "grad_norm": 6.85420036315918,
      "learning_rate": 1.1300551073919891e-06,
      "loss": 1.5278,
      "step": 24060
    },
    {
      "epoch": 4.0945819511780215,
      "grad_norm": 5.8117265701293945,
      "learning_rate": 1.1259473061048387e-06,
      "loss": 1.5882,
      "step": 24070
    },
    {
      "epoch": 4.096283065407842,
      "grad_norm": 7.019490718841553,
      "learning_rate": 1.1218463020035997e-06,
      "loss": 1.6056,
      "step": 24080
    },
    {
      "epoch": 4.097984179637662,
      "grad_norm": 5.084234237670898,
      "learning_rate": 1.117752100068583e-06,
      "loss": 1.6973,
      "step": 24090
    },
    {
      "epoch": 4.099685293867483,
      "grad_norm": 5.35864782333374,
      "learning_rate": 1.113664705271838e-06,
      "loss": 1.6448,
      "step": 24100
    },
    {
      "epoch": 4.101386408097304,
      "grad_norm": 6.192386627197266,
      "learning_rate": 1.109584122577146e-06,
      "loss": 1.534,
      "step": 24110
    },
    {
      "epoch": 4.103087522327124,
      "grad_norm": 6.709290981292725,
      "learning_rate": 1.1055103569400195e-06,
      "loss": 1.5793,
      "step": 24120
    },
    {
      "epoch": 4.104788636556945,
      "grad_norm": 6.019222736358643,
      "learning_rate": 1.101443413307691e-06,
      "loss": 1.4722,
      "step": 24130
    },
    {
      "epoch": 4.106489750786765,
      "grad_norm": 5.954079627990723,
      "learning_rate": 1.0973832966191049e-06,
      "loss": 1.4929,
      "step": 24140
    },
    {
      "epoch": 4.108190865016586,
      "grad_norm": 6.824218273162842,
      "learning_rate": 1.0933300118049192e-06,
      "loss": 1.5698,
      "step": 24150
    },
    {
      "epoch": 4.109891979246407,
      "grad_norm": 5.82558012008667,
      "learning_rate": 1.0892835637874938e-06,
      "loss": 1.6172,
      "step": 24160
    },
    {
      "epoch": 4.111593093476227,
      "grad_norm": 6.433945655822754,
      "learning_rate": 1.0852439574808808e-06,
      "loss": 1.4807,
      "step": 24170
    },
    {
      "epoch": 4.113294207706048,
      "grad_norm": 6.1400909423828125,
      "learning_rate": 1.0812111977908331e-06,
      "loss": 1.5646,
      "step": 24180
    },
    {
      "epoch": 4.114995321935868,
      "grad_norm": 6.493676662445068,
      "learning_rate": 1.0771852896147828e-06,
      "loss": 1.5641,
      "step": 24190
    },
    {
      "epoch": 4.116696436165689,
      "grad_norm": 5.852802276611328,
      "learning_rate": 1.0731662378418425e-06,
      "loss": 1.5711,
      "step": 24200
    },
    {
      "epoch": 4.1183975503955095,
      "grad_norm": 6.089602470397949,
      "learning_rate": 1.0691540473527989e-06,
      "loss": 1.5124,
      "step": 24210
    },
    {
      "epoch": 4.1200986646253295,
      "grad_norm": 7.201757907867432,
      "learning_rate": 1.065148723020105e-06,
      "loss": 1.5698,
      "step": 24220
    },
    {
      "epoch": 4.12179977885515,
      "grad_norm": 5.866385459899902,
      "learning_rate": 1.061150269707878e-06,
      "loss": 1.5362,
      "step": 24230
    },
    {
      "epoch": 4.12350089308497,
      "grad_norm": 5.761326313018799,
      "learning_rate": 1.05715869227189e-06,
      "loss": 1.4711,
      "step": 24240
    },
    {
      "epoch": 4.125202007314791,
      "grad_norm": 7.943985939025879,
      "learning_rate": 1.0531739955595599e-06,
      "loss": 1.5031,
      "step": 24250
    },
    {
      "epoch": 4.126903121544612,
      "grad_norm": 5.94057035446167,
      "learning_rate": 1.0491961844099538e-06,
      "loss": 1.5621,
      "step": 24260
    },
    {
      "epoch": 4.128604235774432,
      "grad_norm": 6.1100993156433105,
      "learning_rate": 1.0452252636537741e-06,
      "loss": 1.6238,
      "step": 24270
    },
    {
      "epoch": 4.130305350004253,
      "grad_norm": 6.636387348175049,
      "learning_rate": 1.0412612381133568e-06,
      "loss": 1.5536,
      "step": 24280
    },
    {
      "epoch": 4.132006464234073,
      "grad_norm": 6.823086261749268,
      "learning_rate": 1.037304112602665e-06,
      "loss": 1.6157,
      "step": 24290
    },
    {
      "epoch": 4.133707578463894,
      "grad_norm": 5.04664945602417,
      "learning_rate": 1.0333538919272803e-06,
      "loss": 1.4875,
      "step": 24300
    },
    {
      "epoch": 4.135408692693715,
      "grad_norm": 6.446927070617676,
      "learning_rate": 1.0294105808843993e-06,
      "loss": 1.5516,
      "step": 24310
    },
    {
      "epoch": 4.137109806923535,
      "grad_norm": 4.810828685760498,
      "learning_rate": 1.025474184262829e-06,
      "loss": 1.5258,
      "step": 24320
    },
    {
      "epoch": 4.138810921153356,
      "grad_norm": 6.347567081451416,
      "learning_rate": 1.021544706842977e-06,
      "loss": 1.6322,
      "step": 24330
    },
    {
      "epoch": 4.140512035383176,
      "grad_norm": 5.357848644256592,
      "learning_rate": 1.0176221533968506e-06,
      "loss": 1.5757,
      "step": 24340
    },
    {
      "epoch": 4.142213149612997,
      "grad_norm": 5.994540691375732,
      "learning_rate": 1.0137065286880458e-06,
      "loss": 1.5293,
      "step": 24350
    },
    {
      "epoch": 4.1439142638428175,
      "grad_norm": 5.068691253662109,
      "learning_rate": 1.0097978374717488e-06,
      "loss": 1.6628,
      "step": 24360
    },
    {
      "epoch": 4.1456153780726375,
      "grad_norm": 6.810344696044922,
      "learning_rate": 1.0058960844947197e-06,
      "loss": 1.5297,
      "step": 24370
    },
    {
      "epoch": 4.147316492302458,
      "grad_norm": 7.203075408935547,
      "learning_rate": 1.0020012744952977e-06,
      "loss": 1.5712,
      "step": 24380
    },
    {
      "epoch": 4.149017606532278,
      "grad_norm": 5.671009063720703,
      "learning_rate": 9.981134122033878e-07,
      "loss": 1.592,
      "step": 24390
    },
    {
      "epoch": 4.150718720762099,
      "grad_norm": 5.365944862365723,
      "learning_rate": 9.94232502340456e-07,
      "loss": 1.5454,
      "step": 24400
    },
    {
      "epoch": 4.15241983499192,
      "grad_norm": 7.130199432373047,
      "learning_rate": 9.903585496195326e-07,
      "loss": 1.4981,
      "step": 24410
    },
    {
      "epoch": 4.15412094922174,
      "grad_norm": 6.007693290710449,
      "learning_rate": 9.864915587451904e-07,
      "loss": 1.494,
      "step": 24420
    },
    {
      "epoch": 4.155822063451561,
      "grad_norm": 6.625998497009277,
      "learning_rate": 9.826315344135521e-07,
      "loss": 1.4852,
      "step": 24430
    },
    {
      "epoch": 4.157523177681381,
      "grad_norm": 7.0912065505981445,
      "learning_rate": 9.787784813122784e-07,
      "loss": 1.5454,
      "step": 24440
    },
    {
      "epoch": 4.159224291911202,
      "grad_norm": 6.79998254776001,
      "learning_rate": 9.749324041205656e-07,
      "loss": 1.5747,
      "step": 24450
    },
    {
      "epoch": 4.160925406141023,
      "grad_norm": 6.3917555809021,
      "learning_rate": 9.71093307509138e-07,
      "loss": 1.4357,
      "step": 24460
    },
    {
      "epoch": 4.162626520370843,
      "grad_norm": 5.7439775466918945,
      "learning_rate": 9.672611961402406e-07,
      "loss": 1.5676,
      "step": 24470
    },
    {
      "epoch": 4.164327634600664,
      "grad_norm": 5.881418704986572,
      "learning_rate": 9.634360746676387e-07,
      "loss": 1.615,
      "step": 24480
    },
    {
      "epoch": 4.166028748830484,
      "grad_norm": 6.05189847946167,
      "learning_rate": 9.596179477366064e-07,
      "loss": 1.6308,
      "step": 24490
    },
    {
      "epoch": 4.167729863060305,
      "grad_norm": 5.774824142456055,
      "learning_rate": 9.55806819983925e-07,
      "loss": 1.6272,
      "step": 24500
    },
    {
      "epoch": 4.1694309772901255,
      "grad_norm": 6.309668064117432,
      "learning_rate": 9.520026960378738e-07,
      "loss": 1.5436,
      "step": 24510
    },
    {
      "epoch": 4.1711320915199455,
      "grad_norm": 7.101531028747559,
      "learning_rate": 9.482055805182315e-07,
      "loss": 1.575,
      "step": 24520
    },
    {
      "epoch": 4.172833205749766,
      "grad_norm": 5.314475059509277,
      "learning_rate": 9.444154780362593e-07,
      "loss": 1.5636,
      "step": 24530
    },
    {
      "epoch": 4.174534319979586,
      "grad_norm": 6.091689586639404,
      "learning_rate": 9.406323931947083e-07,
      "loss": 1.5368,
      "step": 24540
    },
    {
      "epoch": 4.176235434209407,
      "grad_norm": 6.525472164154053,
      "learning_rate": 9.36856330587799e-07,
      "loss": 1.5087,
      "step": 24550
    },
    {
      "epoch": 4.177936548439227,
      "grad_norm": 7.5823540687561035,
      "learning_rate": 9.330872948012317e-07,
      "loss": 1.6523,
      "step": 24560
    },
    {
      "epoch": 4.179637662669048,
      "grad_norm": 6.965406894683838,
      "learning_rate": 9.293252904121677e-07,
      "loss": 1.5556,
      "step": 24570
    },
    {
      "epoch": 4.181338776898869,
      "grad_norm": 5.5527753829956055,
      "learning_rate": 9.255703219892354e-07,
      "loss": 1.6019,
      "step": 24580
    },
    {
      "epoch": 4.183039891128689,
      "grad_norm": 7.09881591796875,
      "learning_rate": 9.218223940925138e-07,
      "loss": 1.5558,
      "step": 24590
    },
    {
      "epoch": 4.18474100535851,
      "grad_norm": 6.202441215515137,
      "learning_rate": 9.180815112735318e-07,
      "loss": 1.618,
      "step": 24600
    },
    {
      "epoch": 4.18644211958833,
      "grad_norm": 6.6593098640441895,
      "learning_rate": 9.143476780752655e-07,
      "loss": 1.4606,
      "step": 24610
    },
    {
      "epoch": 4.188143233818151,
      "grad_norm": 5.991734981536865,
      "learning_rate": 9.106208990321262e-07,
      "loss": 1.5272,
      "step": 24620
    },
    {
      "epoch": 4.189844348047972,
      "grad_norm": 6.9438910484313965,
      "learning_rate": 9.069011786699657e-07,
      "loss": 1.5686,
      "step": 24630
    },
    {
      "epoch": 4.191545462277792,
      "grad_norm": 7.357669353485107,
      "learning_rate": 9.031885215060529e-07,
      "loss": 1.5204,
      "step": 24640
    },
    {
      "epoch": 4.193246576507613,
      "grad_norm": 5.462925910949707,
      "learning_rate": 8.994829320490877e-07,
      "loss": 1.5844,
      "step": 24650
    },
    {
      "epoch": 4.194947690737433,
      "grad_norm": 6.265537738800049,
      "learning_rate": 8.95784414799183e-07,
      "loss": 1.4904,
      "step": 24660
    },
    {
      "epoch": 4.1966488049672535,
      "grad_norm": 5.956068515777588,
      "learning_rate": 8.920929742478644e-07,
      "loss": 1.4445,
      "step": 24670
    },
    {
      "epoch": 4.198349919197074,
      "grad_norm": 6.320197105407715,
      "learning_rate": 8.884086148780609e-07,
      "loss": 1.6526,
      "step": 24680
    },
    {
      "epoch": 4.200051033426894,
      "grad_norm": 5.123306751251221,
      "learning_rate": 8.847313411641087e-07,
      "loss": 1.53,
      "step": 24690
    },
    {
      "epoch": 4.201752147656715,
      "grad_norm": 5.411904811859131,
      "learning_rate": 8.810611575717326e-07,
      "loss": 1.5991,
      "step": 24700
    },
    {
      "epoch": 4.203453261886535,
      "grad_norm": 6.095061779022217,
      "learning_rate": 8.773980685580491e-07,
      "loss": 1.4786,
      "step": 24710
    },
    {
      "epoch": 4.205154376116356,
      "grad_norm": 6.627599716186523,
      "learning_rate": 8.737420785715589e-07,
      "loss": 1.6051,
      "step": 24720
    },
    {
      "epoch": 4.206855490346177,
      "grad_norm": 8.003243446350098,
      "learning_rate": 8.700931920521412e-07,
      "loss": 1.511,
      "step": 24730
    },
    {
      "epoch": 4.208556604575997,
      "grad_norm": 5.817923069000244,
      "learning_rate": 8.664514134310501e-07,
      "loss": 1.6458,
      "step": 24740
    },
    {
      "epoch": 4.210257718805818,
      "grad_norm": 6.329773902893066,
      "learning_rate": 8.628167471309068e-07,
      "loss": 1.6663,
      "step": 24750
    },
    {
      "epoch": 4.211958833035638,
      "grad_norm": 4.948273181915283,
      "learning_rate": 8.591891975656939e-07,
      "loss": 1.6422,
      "step": 24760
    },
    {
      "epoch": 4.213659947265459,
      "grad_norm": 7.214920520782471,
      "learning_rate": 8.555687691407538e-07,
      "loss": 1.5631,
      "step": 24770
    },
    {
      "epoch": 4.21536106149528,
      "grad_norm": 6.368295669555664,
      "learning_rate": 8.519554662527801e-07,
      "loss": 1.6313,
      "step": 24780
    },
    {
      "epoch": 4.2170621757251,
      "grad_norm": 5.196382522583008,
      "learning_rate": 8.483492932898098e-07,
      "loss": 1.4689,
      "step": 24790
    },
    {
      "epoch": 4.218763289954921,
      "grad_norm": 6.417914867401123,
      "learning_rate": 8.447502546312278e-07,
      "loss": 1.517,
      "step": 24800
    },
    {
      "epoch": 4.220464404184741,
      "grad_norm": 6.153533458709717,
      "learning_rate": 8.41158354647751e-07,
      "loss": 1.6234,
      "step": 24810
    },
    {
      "epoch": 4.2221655184145614,
      "grad_norm": 6.462419033050537,
      "learning_rate": 8.375735977014251e-07,
      "loss": 1.5166,
      "step": 24820
    },
    {
      "epoch": 4.223866632644382,
      "grad_norm": 6.8873467445373535,
      "learning_rate": 8.339959881456272e-07,
      "loss": 1.5588,
      "step": 24830
    },
    {
      "epoch": 4.225567746874202,
      "grad_norm": 6.402502536773682,
      "learning_rate": 8.30425530325046e-07,
      "loss": 1.4579,
      "step": 24840
    },
    {
      "epoch": 4.227268861104023,
      "grad_norm": 7.729940414428711,
      "learning_rate": 8.268622285756913e-07,
      "loss": 1.4804,
      "step": 24850
    },
    {
      "epoch": 4.228969975333843,
      "grad_norm": 6.577713489532471,
      "learning_rate": 8.233060872248829e-07,
      "loss": 1.6403,
      "step": 24860
    },
    {
      "epoch": 4.230671089563664,
      "grad_norm": 5.388174057006836,
      "learning_rate": 8.19757110591242e-07,
      "loss": 1.5362,
      "step": 24870
    },
    {
      "epoch": 4.232372203793485,
      "grad_norm": 5.965421676635742,
      "learning_rate": 8.162153029846898e-07,
      "loss": 1.5833,
      "step": 24880
    },
    {
      "epoch": 4.234073318023305,
      "grad_norm": 7.016608715057373,
      "learning_rate": 8.126806687064414e-07,
      "loss": 1.6166,
      "step": 24890
    },
    {
      "epoch": 4.235774432253126,
      "grad_norm": 6.407740116119385,
      "learning_rate": 8.091532120490011e-07,
      "loss": 1.6687,
      "step": 24900
    },
    {
      "epoch": 4.237475546482946,
      "grad_norm": 6.694855690002441,
      "learning_rate": 8.056329372961547e-07,
      "loss": 1.5413,
      "step": 24910
    },
    {
      "epoch": 4.239176660712767,
      "grad_norm": 5.850477695465088,
      "learning_rate": 8.021198487229706e-07,
      "loss": 1.6718,
      "step": 24920
    },
    {
      "epoch": 4.240877774942588,
      "grad_norm": 4.856194019317627,
      "learning_rate": 7.98613950595784e-07,
      "loss": 1.5363,
      "step": 24930
    },
    {
      "epoch": 4.242578889172408,
      "grad_norm": 7.285855770111084,
      "learning_rate": 7.951152471722027e-07,
      "loss": 1.5563,
      "step": 24940
    },
    {
      "epoch": 4.2442800034022286,
      "grad_norm": 6.888695240020752,
      "learning_rate": 7.916237427010946e-07,
      "loss": 1.5032,
      "step": 24950
    },
    {
      "epoch": 4.2459811176320486,
      "grad_norm": 5.297257423400879,
      "learning_rate": 7.881394414225872e-07,
      "loss": 1.61,
      "step": 24960
    },
    {
      "epoch": 4.247682231861869,
      "grad_norm": 5.850154876708984,
      "learning_rate": 7.84662347568058e-07,
      "loss": 1.425,
      "step": 24970
    },
    {
      "epoch": 4.24938334609169,
      "grad_norm": 5.764167785644531,
      "learning_rate": 7.81192465360134e-07,
      "loss": 1.6219,
      "step": 24980
    },
    {
      "epoch": 4.25108446032151,
      "grad_norm": 5.481007099151611,
      "learning_rate": 7.777297990126833e-07,
      "loss": 1.5636,
      "step": 24990
    },
    {
      "epoch": 4.252785574551331,
      "grad_norm": 6.768319129943848,
      "learning_rate": 7.742743527308116e-07,
      "loss": 1.5378,
      "step": 25000
    },
    {
      "epoch": 4.254486688781151,
      "grad_norm": 6.511185169219971,
      "learning_rate": 7.708261307108543e-07,
      "loss": 1.6191,
      "step": 25010
    },
    {
      "epoch": 4.256187803010972,
      "grad_norm": 6.309399604797363,
      "learning_rate": 7.673851371403754e-07,
      "loss": 1.5455,
      "step": 25020
    },
    {
      "epoch": 4.257888917240793,
      "grad_norm": 6.184329032897949,
      "learning_rate": 7.639513761981605e-07,
      "loss": 1.481,
      "step": 25030
    },
    {
      "epoch": 4.259590031470613,
      "grad_norm": 6.8204545974731445,
      "learning_rate": 7.605248520542113e-07,
      "loss": 1.5367,
      "step": 25040
    },
    {
      "epoch": 4.261291145700434,
      "grad_norm": 5.148837089538574,
      "learning_rate": 7.571055688697419e-07,
      "loss": 1.5995,
      "step": 25050
    },
    {
      "epoch": 4.262992259930254,
      "grad_norm": 5.682942867279053,
      "learning_rate": 7.536935307971718e-07,
      "loss": 1.5473,
      "step": 25060
    },
    {
      "epoch": 4.264693374160075,
      "grad_norm": 6.7601165771484375,
      "learning_rate": 7.502887419801217e-07,
      "loss": 1.6096,
      "step": 25070
    },
    {
      "epoch": 4.266394488389896,
      "grad_norm": 4.368737697601318,
      "learning_rate": 7.468912065534099e-07,
      "loss": 1.5496,
      "step": 25080
    },
    {
      "epoch": 4.268095602619716,
      "grad_norm": 4.591710090637207,
      "learning_rate": 7.435009286430469e-07,
      "loss": 1.5739,
      "step": 25090
    },
    {
      "epoch": 4.2697967168495365,
      "grad_norm": 5.633373260498047,
      "learning_rate": 7.401179123662278e-07,
      "loss": 1.5376,
      "step": 25100
    },
    {
      "epoch": 4.2714978310793565,
      "grad_norm": 6.092051029205322,
      "learning_rate": 7.367421618313284e-07,
      "loss": 1.5793,
      "step": 25110
    },
    {
      "epoch": 4.273198945309177,
      "grad_norm": 6.855485916137695,
      "learning_rate": 7.333736811379033e-07,
      "loss": 1.5421,
      "step": 25120
    },
    {
      "epoch": 4.274900059538998,
      "grad_norm": 6.99765157699585,
      "learning_rate": 7.300124743766778e-07,
      "loss": 1.6597,
      "step": 25130
    },
    {
      "epoch": 4.276601173768818,
      "grad_norm": 6.900430679321289,
      "learning_rate": 7.266585456295398e-07,
      "loss": 1.5686,
      "step": 25140
    },
    {
      "epoch": 4.278302287998639,
      "grad_norm": 5.6722893714904785,
      "learning_rate": 7.233118989695459e-07,
      "loss": 1.5508,
      "step": 25150
    },
    {
      "epoch": 4.280003402228459,
      "grad_norm": 5.736631393432617,
      "learning_rate": 7.199725384609033e-07,
      "loss": 1.6568,
      "step": 25160
    },
    {
      "epoch": 4.28170451645828,
      "grad_norm": 7.063878536224365,
      "learning_rate": 7.166404681589739e-07,
      "loss": 1.469,
      "step": 25170
    },
    {
      "epoch": 4.283405630688101,
      "grad_norm": 6.601837635040283,
      "learning_rate": 7.133156921102655e-07,
      "loss": 1.5388,
      "step": 25180
    },
    {
      "epoch": 4.285106744917921,
      "grad_norm": 6.6246819496154785,
      "learning_rate": 7.099982143524243e-07,
      "loss": 1.5674,
      "step": 25190
    },
    {
      "epoch": 4.286807859147742,
      "grad_norm": 6.178937911987305,
      "learning_rate": 7.066880389142407e-07,
      "loss": 1.6288,
      "step": 25200
    },
    {
      "epoch": 4.288508973377562,
      "grad_norm": 5.929183006286621,
      "learning_rate": 7.033851698156318e-07,
      "loss": 1.563,
      "step": 25210
    },
    {
      "epoch": 4.290210087607383,
      "grad_norm": 6.338135719299316,
      "learning_rate": 7.000896110676427e-07,
      "loss": 1.5498,
      "step": 25220
    },
    {
      "epoch": 4.291911201837204,
      "grad_norm": 7.406379699707031,
      "learning_rate": 6.9680136667244e-07,
      "loss": 1.5654,
      "step": 25230
    },
    {
      "epoch": 4.293612316067024,
      "grad_norm": 5.393094539642334,
      "learning_rate": 6.935204406233082e-07,
      "loss": 1.5828,
      "step": 25240
    },
    {
      "epoch": 4.2953134302968445,
      "grad_norm": 4.678094387054443,
      "learning_rate": 6.902468369046441e-07,
      "loss": 1.5138,
      "step": 25250
    },
    {
      "epoch": 4.2970145445266645,
      "grad_norm": 5.554921627044678,
      "learning_rate": 6.869805594919559e-07,
      "loss": 1.4475,
      "step": 25260
    },
    {
      "epoch": 4.298715658756485,
      "grad_norm": 5.954467296600342,
      "learning_rate": 6.837216123518489e-07,
      "loss": 1.4953,
      "step": 25270
    },
    {
      "epoch": 4.300416772986306,
      "grad_norm": 5.233072280883789,
      "learning_rate": 6.804699994420301e-07,
      "loss": 1.555,
      "step": 25280
    },
    {
      "epoch": 4.302117887216126,
      "grad_norm": 4.819551944732666,
      "learning_rate": 6.772257247112979e-07,
      "loss": 1.5833,
      "step": 25290
    },
    {
      "epoch": 4.303819001445947,
      "grad_norm": 6.186118125915527,
      "learning_rate": 6.73988792099541e-07,
      "loss": 1.5618,
      "step": 25300
    },
    {
      "epoch": 4.305520115675767,
      "grad_norm": 5.4033966064453125,
      "learning_rate": 6.707592055377291e-07,
      "loss": 1.5009,
      "step": 25310
    },
    {
      "epoch": 4.307221229905588,
      "grad_norm": 5.955039978027344,
      "learning_rate": 6.675369689479125e-07,
      "loss": 1.5152,
      "step": 25320
    },
    {
      "epoch": 4.308922344135409,
      "grad_norm": 5.828554630279541,
      "learning_rate": 6.643220862432158e-07,
      "loss": 1.5833,
      "step": 25330
    },
    {
      "epoch": 4.310623458365229,
      "grad_norm": 5.943400859832764,
      "learning_rate": 6.61114561327833e-07,
      "loss": 1.648,
      "step": 25340
    },
    {
      "epoch": 4.31232457259505,
      "grad_norm": 7.454378128051758,
      "learning_rate": 6.579143980970211e-07,
      "loss": 1.5626,
      "step": 25350
    },
    {
      "epoch": 4.31402568682487,
      "grad_norm": 6.213706970214844,
      "learning_rate": 6.547216004370977e-07,
      "loss": 1.6978,
      "step": 25360
    },
    {
      "epoch": 4.315726801054691,
      "grad_norm": 5.878660678863525,
      "learning_rate": 6.51536172225438e-07,
      "loss": 1.5151,
      "step": 25370
    },
    {
      "epoch": 4.317427915284512,
      "grad_norm": 5.611740589141846,
      "learning_rate": 6.483581173304637e-07,
      "loss": 1.573,
      "step": 25380
    },
    {
      "epoch": 4.319129029514332,
      "grad_norm": 6.904658317565918,
      "learning_rate": 6.451874396116454e-07,
      "loss": 1.6441,
      "step": 25390
    },
    {
      "epoch": 4.3208301437441525,
      "grad_norm": 5.6536030769348145,
      "learning_rate": 6.420241429194932e-07,
      "loss": 1.542,
      "step": 25400
    },
    {
      "epoch": 4.3225312579739725,
      "grad_norm": 6.491143226623535,
      "learning_rate": 6.388682310955532e-07,
      "loss": 1.5492,
      "step": 25410
    },
    {
      "epoch": 4.324232372203793,
      "grad_norm": 6.302955150604248,
      "learning_rate": 6.357197079724048e-07,
      "loss": 1.5434,
      "step": 25420
    },
    {
      "epoch": 4.325933486433614,
      "grad_norm": 5.890113830566406,
      "learning_rate": 6.325785773736533e-07,
      "loss": 1.4887,
      "step": 25430
    },
    {
      "epoch": 4.327634600663434,
      "grad_norm": 5.251640319824219,
      "learning_rate": 6.294448431139262e-07,
      "loss": 1.5869,
      "step": 25440
    },
    {
      "epoch": 4.329335714893255,
      "grad_norm": 5.661231994628906,
      "learning_rate": 6.263185089988697e-07,
      "loss": 1.6795,
      "step": 25450
    },
    {
      "epoch": 4.331036829123075,
      "grad_norm": 5.819065093994141,
      "learning_rate": 6.231995788251436e-07,
      "loss": 1.5383,
      "step": 25460
    },
    {
      "epoch": 4.332737943352896,
      "grad_norm": 6.791083335876465,
      "learning_rate": 6.200880563804138e-07,
      "loss": 1.4843,
      "step": 25470
    },
    {
      "epoch": 4.334439057582717,
      "grad_norm": 6.878266334533691,
      "learning_rate": 6.169839454433517e-07,
      "loss": 1.5374,
      "step": 25480
    },
    {
      "epoch": 4.336140171812537,
      "grad_norm": 6.439700603485107,
      "learning_rate": 6.138872497836306e-07,
      "loss": 1.5041,
      "step": 25490
    },
    {
      "epoch": 4.337841286042358,
      "grad_norm": 6.028425693511963,
      "learning_rate": 6.107979731619145e-07,
      "loss": 1.6296,
      "step": 25500
    },
    {
      "epoch": 4.339542400272178,
      "grad_norm": 6.454329490661621,
      "learning_rate": 6.077161193298606e-07,
      "loss": 1.5674,
      "step": 25510
    },
    {
      "epoch": 4.341243514501999,
      "grad_norm": 6.224812030792236,
      "learning_rate": 6.04641692030108e-07,
      "loss": 1.5929,
      "step": 25520
    },
    {
      "epoch": 4.34294462873182,
      "grad_norm": 7.219799995422363,
      "learning_rate": 6.0157469499628e-07,
      "loss": 1.5856,
      "step": 25530
    },
    {
      "epoch": 4.34464574296164,
      "grad_norm": 6.953548908233643,
      "learning_rate": 5.98515131952978e-07,
      "loss": 1.5167,
      "step": 25540
    },
    {
      "epoch": 4.3463468571914605,
      "grad_norm": 6.339746475219727,
      "learning_rate": 5.954630066157725e-07,
      "loss": 1.5609,
      "step": 25550
    },
    {
      "epoch": 4.3480479714212805,
      "grad_norm": 7.047362327575684,
      "learning_rate": 5.924183226912013e-07,
      "loss": 1.5132,
      "step": 25560
    },
    {
      "epoch": 4.349749085651101,
      "grad_norm": 7.085602760314941,
      "learning_rate": 5.893810838767672e-07,
      "loss": 1.5972,
      "step": 25570
    },
    {
      "epoch": 4.351450199880922,
      "grad_norm": 7.212773323059082,
      "learning_rate": 5.863512938609308e-07,
      "loss": 1.5564,
      "step": 25580
    },
    {
      "epoch": 4.353151314110742,
      "grad_norm": 6.354076862335205,
      "learning_rate": 5.833289563231057e-07,
      "loss": 1.4931,
      "step": 25590
    },
    {
      "epoch": 4.354852428340563,
      "grad_norm": 5.628859043121338,
      "learning_rate": 5.803140749336587e-07,
      "loss": 1.5475,
      "step": 25600
    },
    {
      "epoch": 4.356553542570383,
      "grad_norm": 6.166843891143799,
      "learning_rate": 5.773066533538966e-07,
      "loss": 1.6855,
      "step": 25610
    },
    {
      "epoch": 4.358254656800204,
      "grad_norm": 6.750080585479736,
      "learning_rate": 5.743066952360714e-07,
      "loss": 1.5174,
      "step": 25620
    },
    {
      "epoch": 4.359955771030025,
      "grad_norm": 5.141016483306885,
      "learning_rate": 5.713142042233677e-07,
      "loss": 1.4891,
      "step": 25630
    },
    {
      "epoch": 4.361656885259845,
      "grad_norm": 6.939876556396484,
      "learning_rate": 5.683291839499045e-07,
      "loss": 1.6594,
      "step": 25640
    },
    {
      "epoch": 4.363357999489666,
      "grad_norm": 6.1759867668151855,
      "learning_rate": 5.653516380407264e-07,
      "loss": 1.5642,
      "step": 25650
    },
    {
      "epoch": 4.365059113719486,
      "grad_norm": 5.9017181396484375,
      "learning_rate": 5.623815701118051e-07,
      "loss": 1.5946,
      "step": 25660
    },
    {
      "epoch": 4.366760227949307,
      "grad_norm": 5.6289472579956055,
      "learning_rate": 5.594189837700257e-07,
      "loss": 1.5393,
      "step": 25670
    },
    {
      "epoch": 4.368461342179128,
      "grad_norm": 6.2212700843811035,
      "learning_rate": 5.564638826131908e-07,
      "loss": 1.5124,
      "step": 25680
    },
    {
      "epoch": 4.370162456408948,
      "grad_norm": 5.144286632537842,
      "learning_rate": 5.53516270230012e-07,
      "loss": 1.5942,
      "step": 25690
    },
    {
      "epoch": 4.3718635706387685,
      "grad_norm": 6.080607891082764,
      "learning_rate": 5.505761502001057e-07,
      "loss": 1.5799,
      "step": 25700
    },
    {
      "epoch": 4.3735646848685885,
      "grad_norm": 5.448061943054199,
      "learning_rate": 5.476435260939898e-07,
      "loss": 1.587,
      "step": 25710
    },
    {
      "epoch": 4.375265799098409,
      "grad_norm": 6.332202911376953,
      "learning_rate": 5.447184014730809e-07,
      "loss": 1.5345,
      "step": 25720
    },
    {
      "epoch": 4.37696691332823,
      "grad_norm": 4.908957481384277,
      "learning_rate": 5.418007798896856e-07,
      "loss": 1.6112,
      "step": 25730
    },
    {
      "epoch": 4.37866802755805,
      "grad_norm": 6.102475643157959,
      "learning_rate": 5.388906648870004e-07,
      "loss": 1.5951,
      "step": 25740
    },
    {
      "epoch": 4.380369141787871,
      "grad_norm": 6.507709503173828,
      "learning_rate": 5.359880599991039e-07,
      "loss": 1.4762,
      "step": 25750
    },
    {
      "epoch": 4.382070256017691,
      "grad_norm": 6.359583377838135,
      "learning_rate": 5.330929687509564e-07,
      "loss": 1.5631,
      "step": 25760
    },
    {
      "epoch": 4.383771370247512,
      "grad_norm": 6.574647903442383,
      "learning_rate": 5.302053946583929e-07,
      "loss": 1.5766,
      "step": 25770
    },
    {
      "epoch": 4.385472484477333,
      "grad_norm": 6.09061861038208,
      "learning_rate": 5.273253412281195e-07,
      "loss": 1.5324,
      "step": 25780
    },
    {
      "epoch": 4.387173598707153,
      "grad_norm": 6.537026882171631,
      "learning_rate": 5.244528119577085e-07,
      "loss": 1.5823,
      "step": 25790
    },
    {
      "epoch": 4.388874712936974,
      "grad_norm": 5.031495094299316,
      "learning_rate": 5.215878103355952e-07,
      "loss": 1.4866,
      "step": 25800
    },
    {
      "epoch": 4.390575827166794,
      "grad_norm": 6.294393062591553,
      "learning_rate": 5.187303398410738e-07,
      "loss": 1.5707,
      "step": 25810
    },
    {
      "epoch": 4.392276941396615,
      "grad_norm": 7.1139631271362305,
      "learning_rate": 5.158804039442892e-07,
      "loss": 1.5829,
      "step": 25820
    },
    {
      "epoch": 4.393978055626436,
      "grad_norm": 6.569021224975586,
      "learning_rate": 5.130380061062422e-07,
      "loss": 1.5318,
      "step": 25830
    },
    {
      "epoch": 4.395679169856256,
      "grad_norm": 5.906765460968018,
      "learning_rate": 5.102031497787743e-07,
      "loss": 1.4922,
      "step": 25840
    },
    {
      "epoch": 4.3973802840860765,
      "grad_norm": 6.8976616859436035,
      "learning_rate": 5.073758384045713e-07,
      "loss": 1.5555,
      "step": 25850
    },
    {
      "epoch": 4.3990813983158965,
      "grad_norm": 6.1184983253479,
      "learning_rate": 5.045560754171539e-07,
      "loss": 1.5352,
      "step": 25860
    },
    {
      "epoch": 4.400782512545717,
      "grad_norm": 5.493861198425293,
      "learning_rate": 5.01743864240879e-07,
      "loss": 1.5153,
      "step": 25870
    },
    {
      "epoch": 4.402483626775538,
      "grad_norm": 4.283514022827148,
      "learning_rate": 4.989392082909284e-07,
      "loss": 1.6017,
      "step": 25880
    },
    {
      "epoch": 4.404184741005358,
      "grad_norm": 6.298120498657227,
      "learning_rate": 4.961421109733143e-07,
      "loss": 1.4176,
      "step": 25890
    },
    {
      "epoch": 4.405885855235179,
      "grad_norm": 6.147403717041016,
      "learning_rate": 4.933525756848641e-07,
      "loss": 1.5778,
      "step": 25900
    },
    {
      "epoch": 4.407586969464999,
      "grad_norm": 5.236167907714844,
      "learning_rate": 4.905706058132251e-07,
      "loss": 1.6218,
      "step": 25910
    },
    {
      "epoch": 4.40928808369482,
      "grad_norm": 6.064047813415527,
      "learning_rate": 4.877962047368553e-07,
      "loss": 1.5278,
      "step": 25920
    },
    {
      "epoch": 4.410989197924641,
      "grad_norm": 5.734475135803223,
      "learning_rate": 4.850293758250207e-07,
      "loss": 1.5434,
      "step": 25930
    },
    {
      "epoch": 4.412690312154461,
      "grad_norm": 7.101593017578125,
      "learning_rate": 4.82270122437796e-07,
      "loss": 1.5119,
      "step": 25940
    },
    {
      "epoch": 4.414391426384282,
      "grad_norm": 5.9773969650268555,
      "learning_rate": 4.795184479260509e-07,
      "loss": 1.5912,
      "step": 25950
    },
    {
      "epoch": 4.416092540614102,
      "grad_norm": 6.129922389984131,
      "learning_rate": 4.7677435563145313e-07,
      "loss": 1.4223,
      "step": 25960
    },
    {
      "epoch": 4.417793654843923,
      "grad_norm": 6.448030948638916,
      "learning_rate": 4.7403784888646313e-07,
      "loss": 1.5637,
      "step": 25970
    },
    {
      "epoch": 4.419494769073744,
      "grad_norm": 5.361682415008545,
      "learning_rate": 4.7130893101432745e-07,
      "loss": 1.5282,
      "step": 25980
    },
    {
      "epoch": 4.421195883303564,
      "grad_norm": 6.881174087524414,
      "learning_rate": 4.6858760532907837e-07,
      "loss": 1.6037,
      "step": 25990
    },
    {
      "epoch": 4.4228969975333845,
      "grad_norm": 6.652550220489502,
      "learning_rate": 4.658738751355284e-07,
      "loss": 1.5783,
      "step": 26000
    },
    {
      "epoch": 4.4245981117632045,
      "grad_norm": 6.802812099456787,
      "learning_rate": 4.631677437292645e-07,
      "loss": 1.5237,
      "step": 26010
    },
    {
      "epoch": 4.426299225993025,
      "grad_norm": 5.49721622467041,
      "learning_rate": 4.604692143966459e-07,
      "loss": 1.4783,
      "step": 26020
    },
    {
      "epoch": 4.428000340222846,
      "grad_norm": 6.297482967376709,
      "learning_rate": 4.5777829041480083e-07,
      "loss": 1.5841,
      "step": 26030
    },
    {
      "epoch": 4.429701454452666,
      "grad_norm": 7.4595537185668945,
      "learning_rate": 4.550949750516208e-07,
      "loss": 1.6048,
      "step": 26040
    },
    {
      "epoch": 4.431402568682487,
      "grad_norm": 5.517505645751953,
      "learning_rate": 4.5241927156575604e-07,
      "loss": 1.4844,
      "step": 26050
    },
    {
      "epoch": 4.433103682912307,
      "grad_norm": 6.22703218460083,
      "learning_rate": 4.4975118320661623e-07,
      "loss": 1.6691,
      "step": 26060
    },
    {
      "epoch": 4.434804797142128,
      "grad_norm": 7.3646039962768555,
      "learning_rate": 4.470907132143594e-07,
      "loss": 1.5742,
      "step": 26070
    },
    {
      "epoch": 4.436505911371949,
      "grad_norm": 6.23193883895874,
      "learning_rate": 4.444378648198939e-07,
      "loss": 1.5626,
      "step": 26080
    },
    {
      "epoch": 4.438207025601769,
      "grad_norm": 6.38597297668457,
      "learning_rate": 4.4179264124487213e-07,
      "loss": 1.601,
      "step": 26090
    },
    {
      "epoch": 4.43990813983159,
      "grad_norm": 5.653160572052002,
      "learning_rate": 4.391550457016864e-07,
      "loss": 1.5889,
      "step": 26100
    },
    {
      "epoch": 4.44160925406141,
      "grad_norm": 5.833569049835205,
      "learning_rate": 4.365250813934646e-07,
      "loss": 1.6619,
      "step": 26110
    },
    {
      "epoch": 4.443310368291231,
      "grad_norm": 5.669712066650391,
      "learning_rate": 4.3390275151406945e-07,
      "loss": 1.5954,
      "step": 26120
    },
    {
      "epoch": 4.445011482521052,
      "grad_norm": 6.262190818786621,
      "learning_rate": 4.312880592480897e-07,
      "loss": 1.4984,
      "step": 26130
    },
    {
      "epoch": 4.446712596750872,
      "grad_norm": 5.5008344650268555,
      "learning_rate": 4.286810077708405e-07,
      "loss": 1.4724,
      "step": 26140
    },
    {
      "epoch": 4.4484137109806925,
      "grad_norm": 5.778617858886719,
      "learning_rate": 4.2608160024835784e-07,
      "loss": 1.5239,
      "step": 26150
    },
    {
      "epoch": 4.4501148252105125,
      "grad_norm": 6.689032554626465,
      "learning_rate": 4.23489839837393e-07,
      "loss": 1.6242,
      "step": 26160
    },
    {
      "epoch": 4.451815939440333,
      "grad_norm": 5.875165939331055,
      "learning_rate": 4.2090572968541434e-07,
      "loss": 1.5732,
      "step": 26170
    },
    {
      "epoch": 4.453517053670154,
      "grad_norm": 7.6699748039245605,
      "learning_rate": 4.1832927293059584e-07,
      "loss": 1.5344,
      "step": 26180
    },
    {
      "epoch": 4.455218167899974,
      "grad_norm": 6.2144951820373535,
      "learning_rate": 4.157604727018209e-07,
      "loss": 1.5593,
      "step": 26190
    },
    {
      "epoch": 4.456919282129795,
      "grad_norm": 6.983086109161377,
      "learning_rate": 4.1319933211866903e-07,
      "loss": 1.5265,
      "step": 26200
    },
    {
      "epoch": 4.458620396359615,
      "grad_norm": 5.426665306091309,
      "learning_rate": 4.1064585429142326e-07,
      "loss": 1.5845,
      "step": 26210
    },
    {
      "epoch": 4.460321510589436,
      "grad_norm": 6.375173568725586,
      "learning_rate": 4.081000423210565e-07,
      "loss": 1.5287,
      "step": 26220
    },
    {
      "epoch": 4.462022624819257,
      "grad_norm": 5.231582164764404,
      "learning_rate": 4.0556189929923794e-07,
      "loss": 1.5301,
      "step": 26230
    },
    {
      "epoch": 4.463723739049077,
      "grad_norm": 6.492799282073975,
      "learning_rate": 4.0303142830831763e-07,
      "loss": 1.5539,
      "step": 26240
    },
    {
      "epoch": 4.465424853278898,
      "grad_norm": 6.998684406280518,
      "learning_rate": 4.00508632421332e-07,
      "loss": 1.5151,
      "step": 26250
    },
    {
      "epoch": 4.467125967508718,
      "grad_norm": 6.301689147949219,
      "learning_rate": 3.97993514701995e-07,
      "loss": 1.5371,
      "step": 26260
    },
    {
      "epoch": 4.468827081738539,
      "grad_norm": 6.627586364746094,
      "learning_rate": 3.954860782046977e-07,
      "loss": 1.5115,
      "step": 26270
    },
    {
      "epoch": 4.47052819596836,
      "grad_norm": 6.861813545227051,
      "learning_rate": 3.929863259745006e-07,
      "loss": 1.5546,
      "step": 26280
    },
    {
      "epoch": 4.47222931019818,
      "grad_norm": 6.784213542938232,
      "learning_rate": 3.9049426104713526e-07,
      "loss": 1.5602,
      "step": 26290
    },
    {
      "epoch": 4.4739304244280005,
      "grad_norm": 6.133279800415039,
      "learning_rate": 3.880098864489959e-07,
      "loss": 1.5719,
      "step": 26300
    },
    {
      "epoch": 4.4756315386578205,
      "grad_norm": 6.854134559631348,
      "learning_rate": 3.855332051971385e-07,
      "loss": 1.5037,
      "step": 26310
    },
    {
      "epoch": 4.477332652887641,
      "grad_norm": 5.71651029586792,
      "learning_rate": 3.830642202992738e-07,
      "loss": 1.6071,
      "step": 26320
    },
    {
      "epoch": 4.479033767117462,
      "grad_norm": 6.421847820281982,
      "learning_rate": 3.8060293475376833e-07,
      "loss": 1.4797,
      "step": 26330
    },
    {
      "epoch": 4.480734881347282,
      "grad_norm": 6.611272811889648,
      "learning_rate": 3.781493515496392e-07,
      "loss": 1.5943,
      "step": 26340
    },
    {
      "epoch": 4.482435995577103,
      "grad_norm": 6.366344928741455,
      "learning_rate": 3.757034736665479e-07,
      "loss": 1.5623,
      "step": 26350
    },
    {
      "epoch": 4.484137109806923,
      "grad_norm": 6.059317111968994,
      "learning_rate": 3.7326530407479763e-07,
      "loss": 1.6145,
      "step": 26360
    },
    {
      "epoch": 4.485838224036744,
      "grad_norm": 6.124782562255859,
      "learning_rate": 3.70834845735333e-07,
      "loss": 1.5629,
      "step": 26370
    },
    {
      "epoch": 4.487539338266565,
      "grad_norm": 4.732071399688721,
      "learning_rate": 3.684121015997327e-07,
      "loss": 1.4886,
      "step": 26380
    },
    {
      "epoch": 4.489240452496385,
      "grad_norm": 5.714395999908447,
      "learning_rate": 3.6599707461020724e-07,
      "loss": 1.5639,
      "step": 26390
    },
    {
      "epoch": 4.490941566726206,
      "grad_norm": 5.7255024909973145,
      "learning_rate": 3.635897676995946e-07,
      "loss": 1.4485,
      "step": 26400
    },
    {
      "epoch": 4.492642680956026,
      "grad_norm": 6.477797985076904,
      "learning_rate": 3.6119018379135927e-07,
      "loss": 1.5884,
      "step": 26410
    },
    {
      "epoch": 4.494343795185847,
      "grad_norm": 6.513632774353027,
      "learning_rate": 3.587983257995856e-07,
      "loss": 1.5626,
      "step": 26420
    },
    {
      "epoch": 4.496044909415668,
      "grad_norm": 6.233043193817139,
      "learning_rate": 3.5641419662897493e-07,
      "loss": 1.5763,
      "step": 26430
    },
    {
      "epoch": 4.497746023645488,
      "grad_norm": 6.740249156951904,
      "learning_rate": 3.54037799174844e-07,
      "loss": 1.5187,
      "step": 26440
    },
    {
      "epoch": 4.4994471378753085,
      "grad_norm": 5.980910778045654,
      "learning_rate": 3.516691363231181e-07,
      "loss": 1.4677,
      "step": 26450
    },
    {
      "epoch": 4.5011482521051285,
      "grad_norm": 6.505252838134766,
      "learning_rate": 3.4930821095033297e-07,
      "loss": 1.6046,
      "step": 26460
    },
    {
      "epoch": 4.502849366334949,
      "grad_norm": 5.84398889541626,
      "learning_rate": 3.4695502592362467e-07,
      "loss": 1.5715,
      "step": 26470
    },
    {
      "epoch": 4.50455048056477,
      "grad_norm": 6.0322747230529785,
      "learning_rate": 3.4460958410073116e-07,
      "loss": 1.6288,
      "step": 26480
    },
    {
      "epoch": 4.50625159479459,
      "grad_norm": 5.351884365081787,
      "learning_rate": 3.422718883299843e-07,
      "loss": 1.5561,
      "step": 26490
    },
    {
      "epoch": 4.507952709024411,
      "grad_norm": 5.781704425811768,
      "learning_rate": 3.399419414503105e-07,
      "loss": 1.5336,
      "step": 26500
    },
    {
      "epoch": 4.509653823254231,
      "grad_norm": 5.929610729217529,
      "learning_rate": 3.376197462912291e-07,
      "loss": 1.4502,
      "step": 26510
    },
    {
      "epoch": 4.511354937484052,
      "grad_norm": 6.840120792388916,
      "learning_rate": 3.353053056728412e-07,
      "loss": 1.6567,
      "step": 26520
    },
    {
      "epoch": 4.513056051713873,
      "grad_norm": 6.549271583557129,
      "learning_rate": 3.329986224058321e-07,
      "loss": 1.5748,
      "step": 26530
    },
    {
      "epoch": 4.514757165943693,
      "grad_norm": 6.0582499504089355,
      "learning_rate": 3.3069969929146656e-07,
      "loss": 1.5188,
      "step": 26540
    },
    {
      "epoch": 4.516458280173514,
      "grad_norm": 5.637298107147217,
      "learning_rate": 3.284085391215854e-07,
      "loss": 1.6182,
      "step": 26550
    },
    {
      "epoch": 4.518159394403334,
      "grad_norm": 4.3700785636901855,
      "learning_rate": 3.26125144678602e-07,
      "loss": 1.5956,
      "step": 26560
    },
    {
      "epoch": 4.519860508633155,
      "grad_norm": 5.547298908233643,
      "learning_rate": 3.2407673166848086e-07,
      "loss": 1.5279,
      "step": 26570
    },
    {
      "epoch": 4.521561622862976,
      "grad_norm": 5.613595485687256,
      "learning_rate": 3.218080997383742e-07,
      "loss": 1.586,
      "step": 26580
    },
    {
      "epoch": 4.523262737092796,
      "grad_norm": 5.9025492668151855,
      "learning_rate": 3.1954724155082025e-07,
      "loss": 1.6782,
      "step": 26590
    },
    {
      "epoch": 4.5249638513226165,
      "grad_norm": 5.403884410858154,
      "learning_rate": 3.1729415985143313e-07,
      "loss": 1.5698,
      "step": 26600
    },
    {
      "epoch": 4.5266649655524365,
      "grad_norm": 6.578800201416016,
      "learning_rate": 3.150488573763856e-07,
      "loss": 1.6192,
      "step": 26610
    },
    {
      "epoch": 4.528366079782257,
      "grad_norm": 5.751435279846191,
      "learning_rate": 3.128113368523995e-07,
      "loss": 1.4884,
      "step": 26620
    },
    {
      "epoch": 4.530067194012078,
      "grad_norm": 6.38753080368042,
      "learning_rate": 3.105816009967472e-07,
      "loss": 1.4877,
      "step": 26630
    },
    {
      "epoch": 4.531768308241898,
      "grad_norm": 5.90024471282959,
      "learning_rate": 3.083596525172483e-07,
      "loss": 1.6379,
      "step": 26640
    },
    {
      "epoch": 4.533469422471719,
      "grad_norm": 6.648676872253418,
      "learning_rate": 3.061454941122634e-07,
      "loss": 1.6367,
      "step": 26650
    },
    {
      "epoch": 4.535170536701539,
      "grad_norm": 7.269857406616211,
      "learning_rate": 3.039391284706961e-07,
      "loss": 1.5853,
      "step": 26660
    },
    {
      "epoch": 4.53687165093136,
      "grad_norm": 5.743645668029785,
      "learning_rate": 3.0174055827198176e-07,
      "loss": 1.5216,
      "step": 26670
    },
    {
      "epoch": 4.538572765161181,
      "grad_norm": 6.32423734664917,
      "learning_rate": 2.995497861860935e-07,
      "loss": 1.5555,
      "step": 26680
    },
    {
      "epoch": 4.540273879391001,
      "grad_norm": 6.08209753036499,
      "learning_rate": 2.9736681487353105e-07,
      "loss": 1.5388,
      "step": 26690
    },
    {
      "epoch": 4.541974993620822,
      "grad_norm": 6.673112869262695,
      "learning_rate": 2.951916469853215e-07,
      "loss": 1.4634,
      "step": 26700
    },
    {
      "epoch": 4.543676107850642,
      "grad_norm": 5.572475433349609,
      "learning_rate": 2.930242851630162e-07,
      "loss": 1.6853,
      "step": 26710
    },
    {
      "epoch": 4.545377222080463,
      "grad_norm": 7.366125583648682,
      "learning_rate": 2.9086473203868655e-07,
      "loss": 1.5605,
      "step": 26720
    },
    {
      "epoch": 4.547078336310284,
      "grad_norm": 5.890539646148682,
      "learning_rate": 2.8871299023492066e-07,
      "loss": 1.4455,
      "step": 26730
    },
    {
      "epoch": 4.548779450540104,
      "grad_norm": 6.030550003051758,
      "learning_rate": 2.865690623648207e-07,
      "loss": 1.5712,
      "step": 26740
    },
    {
      "epoch": 4.5504805647699245,
      "grad_norm": 5.688420295715332,
      "learning_rate": 2.8443295103199955e-07,
      "loss": 1.5425,
      "step": 26750
    },
    {
      "epoch": 4.5521816789997445,
      "grad_norm": 6.9885969161987305,
      "learning_rate": 2.8230465883057673e-07,
      "loss": 1.5411,
      "step": 26760
    },
    {
      "epoch": 4.553882793229565,
      "grad_norm": 6.742706775665283,
      "learning_rate": 2.801841883451751e-07,
      "loss": 1.5059,
      "step": 26770
    },
    {
      "epoch": 4.555583907459386,
      "grad_norm": 6.455204963684082,
      "learning_rate": 2.7807154215092315e-07,
      "loss": 1.6318,
      "step": 26780
    },
    {
      "epoch": 4.557285021689206,
      "grad_norm": 6.3332438468933105,
      "learning_rate": 2.759667228134441e-07,
      "loss": 1.4313,
      "step": 26790
    },
    {
      "epoch": 4.558986135919027,
      "grad_norm": 5.9591593742370605,
      "learning_rate": 2.738697328888554e-07,
      "loss": 1.6398,
      "step": 26800
    },
    {
      "epoch": 4.560687250148847,
      "grad_norm": 5.817910671234131,
      "learning_rate": 2.717805749237684e-07,
      "loss": 1.6467,
      "step": 26810
    },
    {
      "epoch": 4.562388364378668,
      "grad_norm": 6.373818874359131,
      "learning_rate": 2.6969925145528293e-07,
      "loss": 1.6025,
      "step": 26820
    },
    {
      "epoch": 4.564089478608489,
      "grad_norm": 6.842092037200928,
      "learning_rate": 2.6762576501098285e-07,
      "loss": 1.5696,
      "step": 26830
    },
    {
      "epoch": 4.565790592838309,
      "grad_norm": 6.427211284637451,
      "learning_rate": 2.655601181089368e-07,
      "loss": 1.606,
      "step": 26840
    },
    {
      "epoch": 4.56749170706813,
      "grad_norm": 6.684589862823486,
      "learning_rate": 2.6350231325769386e-07,
      "loss": 1.5685,
      "step": 26850
    },
    {
      "epoch": 4.56919282129795,
      "grad_norm": 6.076772689819336,
      "learning_rate": 2.6145235295627427e-07,
      "loss": 1.5562,
      "step": 26860
    },
    {
      "epoch": 4.570893935527771,
      "grad_norm": 6.500069618225098,
      "learning_rate": 2.594102396941772e-07,
      "loss": 1.5452,
      "step": 26870
    },
    {
      "epoch": 4.572595049757592,
      "grad_norm": 6.09898042678833,
      "learning_rate": 2.573759759513694e-07,
      "loss": 1.5775,
      "step": 26880
    },
    {
      "epoch": 4.574296163987412,
      "grad_norm": 5.9650492668151855,
      "learning_rate": 2.553495641982861e-07,
      "loss": 1.589,
      "step": 26890
    },
    {
      "epoch": 4.5759972782172325,
      "grad_norm": 6.79572057723999,
      "learning_rate": 2.533310068958285e-07,
      "loss": 1.5511,
      "step": 26900
    },
    {
      "epoch": 4.5776983924470525,
      "grad_norm": 6.421900749206543,
      "learning_rate": 2.5132030649535657e-07,
      "loss": 1.5457,
      "step": 26910
    },
    {
      "epoch": 4.579399506676873,
      "grad_norm": 6.944169521331787,
      "learning_rate": 2.4931746543868967e-07,
      "loss": 1.6124,
      "step": 26920
    },
    {
      "epoch": 4.581100620906694,
      "grad_norm": 6.094720363616943,
      "learning_rate": 2.4732248615810356e-07,
      "loss": 1.4469,
      "step": 26930
    },
    {
      "epoch": 4.582801735136514,
      "grad_norm": 5.676002502441406,
      "learning_rate": 2.4533537107632523e-07,
      "loss": 1.5275,
      "step": 26940
    },
    {
      "epoch": 4.584502849366335,
      "grad_norm": 6.113831996917725,
      "learning_rate": 2.433561226065321e-07,
      "loss": 1.5873,
      "step": 26950
    },
    {
      "epoch": 4.586203963596155,
      "grad_norm": 6.3694000244140625,
      "learning_rate": 2.413847431523476e-07,
      "loss": 1.5317,
      "step": 26960
    },
    {
      "epoch": 4.587905077825976,
      "grad_norm": 5.022193908691406,
      "learning_rate": 2.3942123510784047e-07,
      "loss": 1.6011,
      "step": 26970
    },
    {
      "epoch": 4.589606192055797,
      "grad_norm": 6.505123138427734,
      "learning_rate": 2.3746560085751847e-07,
      "loss": 1.5448,
      "step": 26980
    },
    {
      "epoch": 4.591307306285617,
      "grad_norm": 7.246630668640137,
      "learning_rate": 2.3551784277632884e-07,
      "loss": 1.5249,
      "step": 26990
    },
    {
      "epoch": 4.593008420515438,
      "grad_norm": 5.185988426208496,
      "learning_rate": 2.3357796322965128e-07,
      "loss": 1.5699,
      "step": 27000
    },
    {
      "epoch": 4.594709534745258,
      "grad_norm": 7.010645389556885,
      "learning_rate": 2.3164596457330253e-07,
      "loss": 1.4652,
      "step": 27010
    },
    {
      "epoch": 4.596410648975079,
      "grad_norm": 6.542299270629883,
      "learning_rate": 2.2972184915352362e-07,
      "loss": 1.4913,
      "step": 27020
    },
    {
      "epoch": 4.5981117632049,
      "grad_norm": 7.340641498565674,
      "learning_rate": 2.2780561930698522e-07,
      "loss": 1.5091,
      "step": 27030
    },
    {
      "epoch": 4.59981287743472,
      "grad_norm": 6.255001544952393,
      "learning_rate": 2.2589727736078071e-07,
      "loss": 1.5317,
      "step": 27040
    },
    {
      "epoch": 4.6015139916645404,
      "grad_norm": 5.629670143127441,
      "learning_rate": 2.2399682563242422e-07,
      "loss": 1.6179,
      "step": 27050
    },
    {
      "epoch": 4.6032151058943604,
      "grad_norm": 6.794814109802246,
      "learning_rate": 2.221042664298459e-07,
      "loss": 1.4689,
      "step": 27060
    },
    {
      "epoch": 4.604916220124181,
      "grad_norm": 5.563018321990967,
      "learning_rate": 2.2021960205139574e-07,
      "loss": 1.5535,
      "step": 27070
    },
    {
      "epoch": 4.606617334354002,
      "grad_norm": 6.172718524932861,
      "learning_rate": 2.183428347858325e-07,
      "loss": 1.5345,
      "step": 27080
    },
    {
      "epoch": 4.608318448583822,
      "grad_norm": 6.285296440124512,
      "learning_rate": 2.1647396691232645e-07,
      "loss": 1.5687,
      "step": 27090
    },
    {
      "epoch": 4.610019562813643,
      "grad_norm": 6.232929229736328,
      "learning_rate": 2.1461300070045285e-07,
      "loss": 1.5277,
      "step": 27100
    },
    {
      "epoch": 4.611720677043463,
      "grad_norm": 5.748643398284912,
      "learning_rate": 2.1275993841019192e-07,
      "loss": 1.4778,
      "step": 27110
    },
    {
      "epoch": 4.613421791273284,
      "grad_norm": 5.501770973205566,
      "learning_rate": 2.1091478229192721e-07,
      "loss": 1.5975,
      "step": 27120
    },
    {
      "epoch": 4.615122905503105,
      "grad_norm": 5.694693088531494,
      "learning_rate": 2.0907753458643836e-07,
      "loss": 1.4357,
      "step": 27130
    },
    {
      "epoch": 4.616824019732925,
      "grad_norm": 7.41944694519043,
      "learning_rate": 2.0724819752490235e-07,
      "loss": 1.5026,
      "step": 27140
    },
    {
      "epoch": 4.618525133962746,
      "grad_norm": 5.143401622772217,
      "learning_rate": 2.0542677332888775e-07,
      "loss": 1.5638,
      "step": 27150
    },
    {
      "epoch": 4.620226248192566,
      "grad_norm": 6.477053165435791,
      "learning_rate": 2.0361326421035533e-07,
      "loss": 1.5626,
      "step": 27160
    },
    {
      "epoch": 4.621927362422387,
      "grad_norm": 6.121665954589844,
      "learning_rate": 2.0180767237165135e-07,
      "loss": 1.6153,
      "step": 27170
    },
    {
      "epoch": 4.6236284766522076,
      "grad_norm": 5.462719440460205,
      "learning_rate": 2.000100000055121e-07,
      "loss": 1.5212,
      "step": 27180
    },
    {
      "epoch": 4.6253295908820276,
      "grad_norm": 6.056469440460205,
      "learning_rate": 1.982202492950519e-07,
      "loss": 1.5548,
      "step": 27190
    },
    {
      "epoch": 4.627030705111848,
      "grad_norm": 6.080183982849121,
      "learning_rate": 1.9643842241376537e-07,
      "loss": 1.6376,
      "step": 27200
    },
    {
      "epoch": 4.628731819341668,
      "grad_norm": 5.385749340057373,
      "learning_rate": 1.9466452152552673e-07,
      "loss": 1.5231,
      "step": 27210
    },
    {
      "epoch": 4.630432933571489,
      "grad_norm": 7.34156608581543,
      "learning_rate": 1.928985487845822e-07,
      "loss": 1.6556,
      "step": 27220
    },
    {
      "epoch": 4.63213404780131,
      "grad_norm": 6.492604732513428,
      "learning_rate": 1.911405063355509e-07,
      "loss": 1.6597,
      "step": 27230
    },
    {
      "epoch": 4.63383516203113,
      "grad_norm": 6.061577796936035,
      "learning_rate": 1.8939039631342245e-07,
      "loss": 1.5342,
      "step": 27240
    },
    {
      "epoch": 4.635536276260951,
      "grad_norm": 5.184089660644531,
      "learning_rate": 1.8764822084355118e-07,
      "loss": 1.5325,
      "step": 27250
    },
    {
      "epoch": 4.637237390490771,
      "grad_norm": 6.684836387634277,
      "learning_rate": 1.8591398204165678e-07,
      "loss": 1.4896,
      "step": 27260
    },
    {
      "epoch": 4.638938504720592,
      "grad_norm": 6.151313304901123,
      "learning_rate": 1.841876820138205e-07,
      "loss": 1.4978,
      "step": 27270
    },
    {
      "epoch": 4.640639618950413,
      "grad_norm": 5.921164512634277,
      "learning_rate": 1.8246932285648228e-07,
      "loss": 1.6357,
      "step": 27280
    },
    {
      "epoch": 4.642340733180233,
      "grad_norm": 5.848247528076172,
      "learning_rate": 1.8075890665644e-07,
      "loss": 1.5606,
      "step": 27290
    },
    {
      "epoch": 4.644041847410054,
      "grad_norm": 5.273307800292969,
      "learning_rate": 1.7905643549084268e-07,
      "loss": 1.4879,
      "step": 27300
    },
    {
      "epoch": 4.645742961639874,
      "grad_norm": 6.233772277832031,
      "learning_rate": 1.773619114271942e-07,
      "loss": 1.526,
      "step": 27310
    },
    {
      "epoch": 4.647444075869695,
      "grad_norm": 6.645590782165527,
      "learning_rate": 1.7567533652334457e-07,
      "loss": 1.6352,
      "step": 27320
    },
    {
      "epoch": 4.6491451900995155,
      "grad_norm": 4.733581066131592,
      "learning_rate": 1.7399671282749103e-07,
      "loss": 1.5203,
      "step": 27330
    },
    {
      "epoch": 4.6508463043293355,
      "grad_norm": 6.412637233734131,
      "learning_rate": 1.7232604237817623e-07,
      "loss": 1.5169,
      "step": 27340
    },
    {
      "epoch": 4.652547418559156,
      "grad_norm": 5.247931480407715,
      "learning_rate": 1.7066332720428192e-07,
      "loss": 1.5933,
      "step": 27350
    },
    {
      "epoch": 4.654248532788976,
      "grad_norm": 6.40666389465332,
      "learning_rate": 1.6900856932503066e-07,
      "loss": 1.5831,
      "step": 27360
    },
    {
      "epoch": 4.655949647018797,
      "grad_norm": 6.300868511199951,
      "learning_rate": 1.6736177074998052e-07,
      "loss": 1.4672,
      "step": 27370
    },
    {
      "epoch": 4.657650761248618,
      "grad_norm": 7.2393035888671875,
      "learning_rate": 1.65722933479025e-07,
      "loss": 1.5738,
      "step": 27380
    },
    {
      "epoch": 4.659351875478438,
      "grad_norm": 6.379903793334961,
      "learning_rate": 1.6409205950238725e-07,
      "loss": 1.5274,
      "step": 27390
    },
    {
      "epoch": 4.661052989708259,
      "grad_norm": 4.601118087768555,
      "learning_rate": 1.6246915080062138e-07,
      "loss": 1.6328,
      "step": 27400
    },
    {
      "epoch": 4.662754103938079,
      "grad_norm": 5.284566402435303,
      "learning_rate": 1.6085420934460806e-07,
      "loss": 1.5417,
      "step": 27410
    },
    {
      "epoch": 4.6644552181679,
      "grad_norm": 8.52403450012207,
      "learning_rate": 1.5924723709555303e-07,
      "loss": 1.5184,
      "step": 27420
    },
    {
      "epoch": 4.666156332397721,
      "grad_norm": 6.385263442993164,
      "learning_rate": 1.5764823600498187e-07,
      "loss": 1.4659,
      "step": 27430
    },
    {
      "epoch": 4.667857446627541,
      "grad_norm": 6.358104705810547,
      "learning_rate": 1.5605720801474369e-07,
      "loss": 1.6184,
      "step": 27440
    },
    {
      "epoch": 4.669558560857362,
      "grad_norm": 6.500202178955078,
      "learning_rate": 1.5447415505699927e-07,
      "loss": 1.617,
      "step": 27450
    },
    {
      "epoch": 4.671259675087182,
      "grad_norm": 5.578927993774414,
      "learning_rate": 1.5289907905422921e-07,
      "loss": 1.4817,
      "step": 27460
    },
    {
      "epoch": 4.672960789317003,
      "grad_norm": 5.1049604415893555,
      "learning_rate": 1.5133198191922647e-07,
      "loss": 1.5698,
      "step": 27470
    },
    {
      "epoch": 4.6746619035468235,
      "grad_norm": 6.624853134155273,
      "learning_rate": 1.4977286555509265e-07,
      "loss": 1.4749,
      "step": 27480
    },
    {
      "epoch": 4.6763630177766435,
      "grad_norm": 5.951974391937256,
      "learning_rate": 1.48221731855238e-07,
      "loss": 1.3458,
      "step": 27490
    },
    {
      "epoch": 4.678064132006464,
      "grad_norm": 6.709983825683594,
      "learning_rate": 1.466785827033784e-07,
      "loss": 1.6426,
      "step": 27500
    },
    {
      "epoch": 4.679765246236284,
      "grad_norm": 5.524947643280029,
      "learning_rate": 1.4514341997353242e-07,
      "loss": 1.5079,
      "step": 27510
    },
    {
      "epoch": 4.681466360466105,
      "grad_norm": 7.322075366973877,
      "learning_rate": 1.4361624553002282e-07,
      "loss": 1.5107,
      "step": 27520
    },
    {
      "epoch": 4.683167474695926,
      "grad_norm": 7.2505340576171875,
      "learning_rate": 1.4209706122746746e-07,
      "loss": 1.5047,
      "step": 27530
    },
    {
      "epoch": 4.684868588925746,
      "grad_norm": 5.918062210083008,
      "learning_rate": 1.4058586891078397e-07,
      "loss": 1.6389,
      "step": 27540
    },
    {
      "epoch": 4.686569703155567,
      "grad_norm": 4.98066520690918,
      "learning_rate": 1.390826704151822e-07,
      "loss": 1.5393,
      "step": 27550
    },
    {
      "epoch": 4.688270817385387,
      "grad_norm": 6.098052024841309,
      "learning_rate": 1.375874675661649e-07,
      "loss": 1.5437,
      "step": 27560
    },
    {
      "epoch": 4.689971931615208,
      "grad_norm": 6.57733154296875,
      "learning_rate": 1.3610026217952485e-07,
      "loss": 1.5263,
      "step": 27570
    },
    {
      "epoch": 4.691673045845029,
      "grad_norm": 5.79831600189209,
      "learning_rate": 1.3462105606134405e-07,
      "loss": 1.4929,
      "step": 27580
    },
    {
      "epoch": 4.693374160074849,
      "grad_norm": 7.397582530975342,
      "learning_rate": 1.3314985100798777e-07,
      "loss": 1.5243,
      "step": 27590
    },
    {
      "epoch": 4.69507527430467,
      "grad_norm": 7.244752407073975,
      "learning_rate": 1.3168664880610598e-07,
      "loss": 1.462,
      "step": 27600
    },
    {
      "epoch": 4.69677638853449,
      "grad_norm": 6.627752780914307,
      "learning_rate": 1.3023145123262976e-07,
      "loss": 1.5952,
      "step": 27610
    },
    {
      "epoch": 4.698477502764311,
      "grad_norm": 6.352685451507568,
      "learning_rate": 1.2878426005476813e-07,
      "loss": 1.5902,
      "step": 27620
    },
    {
      "epoch": 4.7001786169941315,
      "grad_norm": 6.046491622924805,
      "learning_rate": 1.2734507703000974e-07,
      "loss": 1.4715,
      "step": 27630
    },
    {
      "epoch": 4.7018797312239515,
      "grad_norm": 5.038861274719238,
      "learning_rate": 1.2591390390611518e-07,
      "loss": 1.4988,
      "step": 27640
    },
    {
      "epoch": 4.703580845453772,
      "grad_norm": 6.450471878051758,
      "learning_rate": 1.2449074242111873e-07,
      "loss": 1.4618,
      "step": 27650
    },
    {
      "epoch": 4.705281959683592,
      "grad_norm": 7.268611907958984,
      "learning_rate": 1.2307559430332588e-07,
      "loss": 1.5477,
      "step": 27660
    },
    {
      "epoch": 4.706983073913413,
      "grad_norm": 5.249655723571777,
      "learning_rate": 1.2166846127130977e-07,
      "loss": 1.5072,
      "step": 27670
    },
    {
      "epoch": 4.708684188143234,
      "grad_norm": 5.183140754699707,
      "learning_rate": 1.2026934503391037e-07,
      "loss": 1.5799,
      "step": 27680
    },
    {
      "epoch": 4.710385302373054,
      "grad_norm": 6.299753665924072,
      "learning_rate": 1.1887824729023222e-07,
      "loss": 1.5838,
      "step": 27690
    },
    {
      "epoch": 4.712086416602875,
      "grad_norm": 5.463996887207031,
      "learning_rate": 1.1749516972964151e-07,
      "loss": 1.5262,
      "step": 27700
    },
    {
      "epoch": 4.713787530832695,
      "grad_norm": 5.149056911468506,
      "learning_rate": 1.1612011403176379e-07,
      "loss": 1.4921,
      "step": 27710
    },
    {
      "epoch": 4.715488645062516,
      "grad_norm": 6.004656791687012,
      "learning_rate": 1.1475308186648545e-07,
      "loss": 1.5962,
      "step": 27720
    },
    {
      "epoch": 4.717189759292337,
      "grad_norm": 6.959606647491455,
      "learning_rate": 1.1339407489394709e-07,
      "loss": 1.5032,
      "step": 27730
    },
    {
      "epoch": 4.718890873522157,
      "grad_norm": 6.741888046264648,
      "learning_rate": 1.1204309476454192e-07,
      "loss": 1.6347,
      "step": 27740
    },
    {
      "epoch": 4.720591987751978,
      "grad_norm": 8.727808952331543,
      "learning_rate": 1.1070014311891806e-07,
      "loss": 1.5043,
      "step": 27750
    },
    {
      "epoch": 4.722293101981798,
      "grad_norm": 5.873548984527588,
      "learning_rate": 1.0936522158797331e-07,
      "loss": 1.5971,
      "step": 27760
    },
    {
      "epoch": 4.723994216211619,
      "grad_norm": 5.455075740814209,
      "learning_rate": 1.0803833179285291e-07,
      "loss": 1.5516,
      "step": 27770
    },
    {
      "epoch": 4.7256953304414395,
      "grad_norm": 5.840266227722168,
      "learning_rate": 1.0671947534494727e-07,
      "loss": 1.5351,
      "step": 27780
    },
    {
      "epoch": 4.7273964446712595,
      "grad_norm": 3.786184787750244,
      "learning_rate": 1.0540865384589275e-07,
      "loss": 1.5615,
      "step": 27790
    },
    {
      "epoch": 4.72909755890108,
      "grad_norm": 7.517036437988281,
      "learning_rate": 1.0410586888756642e-07,
      "loss": 1.6074,
      "step": 27800
    },
    {
      "epoch": 4.7307986731309,
      "grad_norm": 6.73851203918457,
      "learning_rate": 1.0281112205208758e-07,
      "loss": 1.6169,
      "step": 27810
    },
    {
      "epoch": 4.732499787360721,
      "grad_norm": 6.1009440422058105,
      "learning_rate": 1.0152441491181323e-07,
      "loss": 1.6053,
      "step": 27820
    },
    {
      "epoch": 4.734200901590542,
      "grad_norm": 6.914463996887207,
      "learning_rate": 1.0024574902933513e-07,
      "loss": 1.5295,
      "step": 27830
    },
    {
      "epoch": 4.735902015820362,
      "grad_norm": 4.348443031311035,
      "learning_rate": 9.897512595748203e-08,
      "loss": 1.6825,
      "step": 27840
    },
    {
      "epoch": 4.737603130050183,
      "grad_norm": 6.353835105895996,
      "learning_rate": 9.771254723931293e-08,
      "loss": 1.7024,
      "step": 27850
    },
    {
      "epoch": 4.739304244280003,
      "grad_norm": 7.957035541534424,
      "learning_rate": 9.645801440812157e-08,
      "loss": 1.561,
      "step": 27860
    },
    {
      "epoch": 4.741005358509824,
      "grad_norm": 6.56823205947876,
      "learning_rate": 9.521152898742601e-08,
      "loss": 1.4863,
      "step": 27870
    },
    {
      "epoch": 4.742706472739645,
      "grad_norm": 5.221477031707764,
      "learning_rate": 9.397309249097455e-08,
      "loss": 1.6663,
      "step": 27880
    },
    {
      "epoch": 4.744407586969465,
      "grad_norm": 6.17525053024292,
      "learning_rate": 9.27427064227398e-08,
      "loss": 1.5474,
      "step": 27890
    },
    {
      "epoch": 4.746108701199286,
      "grad_norm": 6.125594615936279,
      "learning_rate": 9.15203722769179e-08,
      "loss": 1.5277,
      "step": 27900
    },
    {
      "epoch": 4.747809815429106,
      "grad_norm": 5.346764087677002,
      "learning_rate": 9.030609153792708e-08,
      "loss": 1.4824,
      "step": 27910
    },
    {
      "epoch": 4.749510929658927,
      "grad_norm": 6.787766456604004,
      "learning_rate": 8.909986568040384e-08,
      "loss": 1.5468,
      "step": 27920
    },
    {
      "epoch": 4.7512120438887475,
      "grad_norm": 5.578236103057861,
      "learning_rate": 8.790169616920454e-08,
      "loss": 1.5302,
      "step": 27930
    },
    {
      "epoch": 4.7529131581185675,
      "grad_norm": 7.268450736999512,
      "learning_rate": 8.671158445940085e-08,
      "loss": 1.5299,
      "step": 27940
    },
    {
      "epoch": 4.754614272348388,
      "grad_norm": 4.132484436035156,
      "learning_rate": 8.552953199627981e-08,
      "loss": 1.6264,
      "step": 27950
    },
    {
      "epoch": 4.756315386578208,
      "grad_norm": 6.548470973968506,
      "learning_rate": 8.43555402153393e-08,
      "loss": 1.4976,
      "step": 27960
    },
    {
      "epoch": 4.758016500808029,
      "grad_norm": 7.993465900421143,
      "learning_rate": 8.3189610542291e-08,
      "loss": 1.5844,
      "step": 27970
    },
    {
      "epoch": 4.75971761503785,
      "grad_norm": 6.39706563949585,
      "learning_rate": 8.203174439305381e-08,
      "loss": 1.459,
      "step": 27980
    },
    {
      "epoch": 4.76141872926767,
      "grad_norm": 7.2803778648376465,
      "learning_rate": 8.088194317375442e-08,
      "loss": 1.5256,
      "step": 27990
    },
    {
      "epoch": 4.763119843497491,
      "grad_norm": 6.996268272399902,
      "learning_rate": 7.97402082807267e-08,
      "loss": 1.5922,
      "step": 28000
    },
    {
      "epoch": 4.764820957727311,
      "grad_norm": 7.688847541809082,
      "learning_rate": 7.860654110050639e-08,
      "loss": 1.508,
      "step": 28010
    },
    {
      "epoch": 4.766522071957132,
      "grad_norm": 5.641658306121826,
      "learning_rate": 7.748094300983411e-08,
      "loss": 1.5541,
      "step": 28020
    },
    {
      "epoch": 4.768223186186953,
      "grad_norm": 5.608862400054932,
      "learning_rate": 7.636341537564947e-08,
      "loss": 1.6232,
      "step": 28030
    },
    {
      "epoch": 4.769924300416773,
      "grad_norm": 6.867303371429443,
      "learning_rate": 7.525395955509162e-08,
      "loss": 1.5451,
      "step": 28040
    },
    {
      "epoch": 4.771625414646594,
      "grad_norm": 8.263049125671387,
      "learning_rate": 7.415257689549874e-08,
      "loss": 1.5382,
      "step": 28050
    },
    {
      "epoch": 4.773326528876414,
      "grad_norm": 6.207581043243408,
      "learning_rate": 7.305926873440264e-08,
      "loss": 1.6129,
      "step": 28060
    },
    {
      "epoch": 4.775027643106235,
      "grad_norm": 6.10377836227417,
      "learning_rate": 7.197403639953109e-08,
      "loss": 1.5315,
      "step": 28070
    },
    {
      "epoch": 4.7767287573360555,
      "grad_norm": 6.987624645233154,
      "learning_rate": 7.089688120880178e-08,
      "loss": 1.5435,
      "step": 28080
    },
    {
      "epoch": 4.7784298715658755,
      "grad_norm": 6.28458833694458,
      "learning_rate": 6.982780447032763e-08,
      "loss": 1.6071,
      "step": 28090
    },
    {
      "epoch": 4.780130985795696,
      "grad_norm": 6.522460460662842,
      "learning_rate": 6.876680748240849e-08,
      "loss": 1.5836,
      "step": 28100
    },
    {
      "epoch": 4.781832100025516,
      "grad_norm": 6.875545501708984,
      "learning_rate": 6.771389153353124e-08,
      "loss": 1.5461,
      "step": 28110
    },
    {
      "epoch": 4.783533214255337,
      "grad_norm": 6.85712194442749,
      "learning_rate": 6.666905790237041e-08,
      "loss": 1.6475,
      "step": 28120
    },
    {
      "epoch": 4.785234328485158,
      "grad_norm": 5.073885917663574,
      "learning_rate": 6.563230785778535e-08,
      "loss": 1.5565,
      "step": 28130
    },
    {
      "epoch": 4.786935442714978,
      "grad_norm": 6.406914710998535,
      "learning_rate": 6.460364265881706e-08,
      "loss": 1.4715,
      "step": 28140
    },
    {
      "epoch": 4.788636556944799,
      "grad_norm": 7.194412708282471,
      "learning_rate": 6.35830635546906e-08,
      "loss": 1.4826,
      "step": 28150
    },
    {
      "epoch": 4.790337671174619,
      "grad_norm": 7.12645149230957,
      "learning_rate": 6.257057178480904e-08,
      "loss": 1.5125,
      "step": 28160
    },
    {
      "epoch": 4.79203878540444,
      "grad_norm": 5.810079574584961,
      "learning_rate": 6.156616857875641e-08,
      "loss": 1.5234,
      "step": 28170
    },
    {
      "epoch": 4.793739899634261,
      "grad_norm": 6.861225605010986,
      "learning_rate": 6.056985515629033e-08,
      "loss": 1.6324,
      "step": 28180
    },
    {
      "epoch": 4.795441013864081,
      "grad_norm": 6.956240177154541,
      "learning_rate": 5.958163272734789e-08,
      "loss": 1.5656,
      "step": 28190
    },
    {
      "epoch": 4.797142128093902,
      "grad_norm": 5.78271484375,
      "learning_rate": 5.8601502492038254e-08,
      "loss": 1.4416,
      "step": 28200
    },
    {
      "epoch": 4.798843242323722,
      "grad_norm": 5.435102939605713,
      "learning_rate": 5.7629465640643356e-08,
      "loss": 1.6197,
      "step": 28210
    },
    {
      "epoch": 4.800544356553543,
      "grad_norm": 6.35980749130249,
      "learning_rate": 5.666552335361644e-08,
      "loss": 1.5614,
      "step": 28220
    },
    {
      "epoch": 4.8022454707833635,
      "grad_norm": 5.305126190185547,
      "learning_rate": 5.5709676801582056e-08,
      "loss": 1.5634,
      "step": 28230
    },
    {
      "epoch": 4.8039465850131835,
      "grad_norm": 7.220401763916016,
      "learning_rate": 5.47619271453323e-08,
      "loss": 1.5344,
      "step": 28240
    },
    {
      "epoch": 4.805647699243004,
      "grad_norm": 6.255699157714844,
      "learning_rate": 5.382227553582463e-08,
      "loss": 1.5944,
      "step": 28250
    },
    {
      "epoch": 4.807348813472824,
      "grad_norm": 5.981529235839844,
      "learning_rate": 5.2890723114184774e-08,
      "loss": 1.5448,
      "step": 28260
    },
    {
      "epoch": 4.809049927702645,
      "grad_norm": 6.389290809631348,
      "learning_rate": 5.1967271011701577e-08,
      "loss": 1.6385,
      "step": 28270
    },
    {
      "epoch": 4.810751041932466,
      "grad_norm": 6.058357238769531,
      "learning_rate": 5.105192034982623e-08,
      "loss": 1.5977,
      "step": 28280
    },
    {
      "epoch": 4.812452156162286,
      "grad_norm": 6.061993598937988,
      "learning_rate": 5.0144672240172246e-08,
      "loss": 1.5187,
      "step": 28290
    },
    {
      "epoch": 4.814153270392107,
      "grad_norm": 5.450552463531494,
      "learning_rate": 4.9245527784512514e-08,
      "loss": 1.4993,
      "step": 28300
    },
    {
      "epoch": 4.815854384621927,
      "grad_norm": 5.724689483642578,
      "learning_rate": 4.8354488074780013e-08,
      "loss": 1.5683,
      "step": 28310
    },
    {
      "epoch": 4.817555498851748,
      "grad_norm": 4.988497734069824,
      "learning_rate": 4.7471554193064094e-08,
      "loss": 1.5693,
      "step": 28320
    },
    {
      "epoch": 4.819256613081569,
      "grad_norm": 7.25648832321167,
      "learning_rate": 4.6596727211611236e-08,
      "loss": 1.5494,
      "step": 28330
    },
    {
      "epoch": 4.820957727311389,
      "grad_norm": 6.171444416046143,
      "learning_rate": 4.573000819282055e-08,
      "loss": 1.4583,
      "step": 28340
    },
    {
      "epoch": 4.82265884154121,
      "grad_norm": 6.128518581390381,
      "learning_rate": 4.4871398189248257e-08,
      "loss": 1.6047,
      "step": 28350
    },
    {
      "epoch": 4.82435995577103,
      "grad_norm": 5.2125935554504395,
      "learning_rate": 4.40208982435995e-08,
      "loss": 1.6133,
      "step": 28360
    },
    {
      "epoch": 4.826061070000851,
      "grad_norm": 7.181666851043701,
      "learning_rate": 4.317850938873279e-08,
      "loss": 1.5734,
      "step": 28370
    },
    {
      "epoch": 4.8277621842306715,
      "grad_norm": 7.01062536239624,
      "learning_rate": 4.2344232647655566e-08,
      "loss": 1.5536,
      "step": 28380
    },
    {
      "epoch": 4.8294632984604915,
      "grad_norm": 7.739013671875,
      "learning_rate": 4.151806903352341e-08,
      "loss": 1.5742,
      "step": 28390
    },
    {
      "epoch": 4.831164412690312,
      "grad_norm": 6.069748878479004,
      "learning_rate": 4.0700019549640814e-08,
      "loss": 1.5094,
      "step": 28400
    },
    {
      "epoch": 4.832865526920132,
      "grad_norm": 6.2133307456970215,
      "learning_rate": 3.989008518945596e-08,
      "loss": 1.6012,
      "step": 28410
    },
    {
      "epoch": 4.834566641149953,
      "grad_norm": 6.628243923187256,
      "learning_rate": 3.908826693656369e-08,
      "loss": 1.5492,
      "step": 28420
    },
    {
      "epoch": 4.836267755379774,
      "grad_norm": 6.392285346984863,
      "learning_rate": 3.829456576470254e-08,
      "loss": 1.6312,
      "step": 28430
    },
    {
      "epoch": 4.837968869609594,
      "grad_norm": 6.148524761199951,
      "learning_rate": 3.7508982637751715e-08,
      "loss": 1.5596,
      "step": 28440
    },
    {
      "epoch": 4.839669983839415,
      "grad_norm": 5.219897270202637,
      "learning_rate": 3.6731518509734885e-08,
      "loss": 1.6073,
      "step": 28450
    },
    {
      "epoch": 4.841371098069235,
      "grad_norm": 5.39851188659668,
      "learning_rate": 3.596217432481266e-08,
      "loss": 1.5382,
      "step": 28460
    },
    {
      "epoch": 4.843072212299056,
      "grad_norm": 5.752907752990723,
      "learning_rate": 3.5200951017287095e-08,
      "loss": 1.4574,
      "step": 28470
    },
    {
      "epoch": 4.844773326528877,
      "grad_norm": 5.786567687988281,
      "learning_rate": 3.444784951159647e-08,
      "loss": 1.5123,
      "step": 28480
    },
    {
      "epoch": 4.846474440758697,
      "grad_norm": 5.8161115646362305,
      "learning_rate": 3.3702870722316775e-08,
      "loss": 1.567,
      "step": 28490
    },
    {
      "epoch": 4.848175554988518,
      "grad_norm": 5.222090244293213,
      "learning_rate": 3.2966015554160956e-08,
      "loss": 1.6132,
      "step": 28500
    },
    {
      "epoch": 4.849876669218338,
      "grad_norm": 6.055660247802734,
      "learning_rate": 3.2237284901972964e-08,
      "loss": 1.6142,
      "step": 28510
    },
    {
      "epoch": 4.851577783448159,
      "grad_norm": 6.723481178283691,
      "learning_rate": 3.151667965073371e-08,
      "loss": 1.5036,
      "step": 28520
    },
    {
      "epoch": 4.8532788976779795,
      "grad_norm": 7.013541221618652,
      "learning_rate": 3.080420067555509e-08,
      "loss": 1.5581,
      "step": 28530
    },
    {
      "epoch": 4.8549800119077995,
      "grad_norm": 6.067813396453857,
      "learning_rate": 3.009984884167926e-08,
      "loss": 1.4148,
      "step": 28540
    },
    {
      "epoch": 4.85668112613762,
      "grad_norm": 7.258260250091553,
      "learning_rate": 2.9403625004480853e-08,
      "loss": 1.5148,
      "step": 28550
    },
    {
      "epoch": 4.85838224036744,
      "grad_norm": 6.704066276550293,
      "learning_rate": 2.8715530009462512e-08,
      "loss": 1.5143,
      "step": 28560
    },
    {
      "epoch": 4.860083354597261,
      "grad_norm": 6.565433025360107,
      "learning_rate": 2.803556469225564e-08,
      "loss": 1.6015,
      "step": 28570
    },
    {
      "epoch": 4.861784468827082,
      "grad_norm": 5.494863033294678,
      "learning_rate": 2.736372987861814e-08,
      "loss": 1.4758,
      "step": 28580
    },
    {
      "epoch": 4.863485583056902,
      "grad_norm": 6.6142354011535645,
      "learning_rate": 2.6700026384435187e-08,
      "loss": 1.5951,
      "step": 28590
    },
    {
      "epoch": 4.865186697286723,
      "grad_norm": 6.28970193862915,
      "learning_rate": 2.6044455015715486e-08,
      "loss": 1.4526,
      "step": 28600
    },
    {
      "epoch": 4.866887811516543,
      "grad_norm": 5.576261520385742,
      "learning_rate": 2.5397016568594253e-08,
      "loss": 1.5508,
      "step": 28610
    },
    {
      "epoch": 4.868588925746364,
      "grad_norm": 5.773414611816406,
      "learning_rate": 2.4757711829327996e-08,
      "loss": 1.4665,
      "step": 28620
    },
    {
      "epoch": 4.870290039976185,
      "grad_norm": 6.628889083862305,
      "learning_rate": 2.4126541574296e-08,
      "loss": 1.5606,
      "step": 28630
    },
    {
      "epoch": 4.871991154206005,
      "grad_norm": 6.1548237800598145,
      "learning_rate": 2.3503506570000334e-08,
      "loss": 1.5788,
      "step": 28640
    },
    {
      "epoch": 4.873692268435826,
      "grad_norm": 6.75529670715332,
      "learning_rate": 2.2888607573061382e-08,
      "loss": 1.5508,
      "step": 28650
    },
    {
      "epoch": 4.875393382665646,
      "grad_norm": 5.715561389923096,
      "learning_rate": 2.2281845330220808e-08,
      "loss": 1.5172,
      "step": 28660
    },
    {
      "epoch": 4.877094496895467,
      "grad_norm": 6.15458869934082,
      "learning_rate": 2.168322057833858e-08,
      "loss": 1.6079,
      "step": 28670
    },
    {
      "epoch": 4.8787956111252875,
      "grad_norm": 7.746540069580078,
      "learning_rate": 2.109273404439073e-08,
      "loss": 1.5471,
      "step": 28680
    },
    {
      "epoch": 4.8804967253551075,
      "grad_norm": 5.893173694610596,
      "learning_rate": 2.0510386445473097e-08,
      "loss": 1.4594,
      "step": 28690
    },
    {
      "epoch": 4.882197839584928,
      "grad_norm": 7.44713020324707,
      "learning_rate": 1.9936178488794588e-08,
      "loss": 1.5551,
      "step": 28700
    },
    {
      "epoch": 4.883898953814748,
      "grad_norm": 5.640448093414307,
      "learning_rate": 1.937011087168018e-08,
      "loss": 1.5991,
      "step": 28710
    },
    {
      "epoch": 4.885600068044569,
      "grad_norm": 6.992408752441406,
      "learning_rate": 1.8812184281570913e-08,
      "loss": 1.5025,
      "step": 28720
    },
    {
      "epoch": 4.88730118227439,
      "grad_norm": 6.222136974334717,
      "learning_rate": 1.826239939601867e-08,
      "loss": 1.5883,
      "step": 28730
    },
    {
      "epoch": 4.88900229650421,
      "grad_norm": 6.587214946746826,
      "learning_rate": 1.772075688268916e-08,
      "loss": 1.5208,
      "step": 28740
    },
    {
      "epoch": 4.890703410734031,
      "grad_norm": 7.627504825592041,
      "learning_rate": 1.718725739936042e-08,
      "loss": 1.603,
      "step": 28750
    },
    {
      "epoch": 4.892404524963851,
      "grad_norm": 5.792977809906006,
      "learning_rate": 1.6661901593919838e-08,
      "loss": 1.5085,
      "step": 28760
    },
    {
      "epoch": 4.894105639193672,
      "grad_norm": 5.654362678527832,
      "learning_rate": 1.614469010436639e-08,
      "loss": 1.4997,
      "step": 28770
    },
    {
      "epoch": 4.895806753423493,
      "grad_norm": 5.740152359008789,
      "learning_rate": 1.5635623558808395e-08,
      "loss": 1.5557,
      "step": 28780
    },
    {
      "epoch": 4.897507867653313,
      "grad_norm": 6.2546162605285645,
      "learning_rate": 1.5134702575463524e-08,
      "loss": 1.5566,
      "step": 28790
    },
    {
      "epoch": 4.899208981883134,
      "grad_norm": 6.2122015953063965,
      "learning_rate": 1.464192776265431e-08,
      "loss": 1.5772,
      "step": 28800
    },
    {
      "epoch": 4.900910096112954,
      "grad_norm": 5.178815841674805,
      "learning_rate": 1.4157299718814137e-08,
      "loss": 1.6597,
      "step": 28810
    },
    {
      "epoch": 4.902611210342775,
      "grad_norm": 6.3028564453125,
      "learning_rate": 1.36808190324805e-08,
      "loss": 1.4925,
      "step": 28820
    },
    {
      "epoch": 4.9043123245725955,
      "grad_norm": 7.456543445587158,
      "learning_rate": 1.321248628229801e-08,
      "loss": 1.474,
      "step": 28830
    },
    {
      "epoch": 4.9060134388024155,
      "grad_norm": 6.090532302856445,
      "learning_rate": 1.2752302037016141e-08,
      "loss": 1.5729,
      "step": 28840
    },
    {
      "epoch": 4.907714553032236,
      "grad_norm": 5.7161173820495605,
      "learning_rate": 1.2300266855487749e-08,
      "loss": 1.5853,
      "step": 28850
    },
    {
      "epoch": 4.909415667262056,
      "grad_norm": 6.457132816314697,
      "learning_rate": 1.1856381286669807e-08,
      "loss": 1.5537,
      "step": 28860
    },
    {
      "epoch": 4.911116781491877,
      "grad_norm": 7.032171249389648,
      "learning_rate": 1.1420645869622669e-08,
      "loss": 1.5618,
      "step": 28870
    },
    {
      "epoch": 4.912817895721698,
      "grad_norm": 5.624493598937988,
      "learning_rate": 1.0993061133508568e-08,
      "loss": 1.5175,
      "step": 28880
    },
    {
      "epoch": 4.914519009951518,
      "grad_norm": 6.3817057609558105,
      "learning_rate": 1.057362759759237e-08,
      "loss": 1.55,
      "step": 28890
    },
    {
      "epoch": 4.916220124181339,
      "grad_norm": 6.508059501647949,
      "learning_rate": 1.0162345771239327e-08,
      "loss": 1.6487,
      "step": 28900
    },
    {
      "epoch": 4.917921238411159,
      "grad_norm": 5.9004340171813965,
      "learning_rate": 9.759216153915087e-09,
      "loss": 1.539,
      "step": 28910
    },
    {
      "epoch": 4.91962235264098,
      "grad_norm": 5.990011215209961,
      "learning_rate": 9.36423923518419e-09,
      "loss": 1.6223,
      "step": 28920
    },
    {
      "epoch": 4.921323466870801,
      "grad_norm": 6.900084972381592,
      "learning_rate": 8.977415494713067e-09,
      "loss": 1.4912,
      "step": 28930
    },
    {
      "epoch": 4.923024581100621,
      "grad_norm": 7.24298620223999,
      "learning_rate": 8.598745402264797e-09,
      "loss": 1.5822,
      "step": 28940
    },
    {
      "epoch": 4.924725695330442,
      "grad_norm": 6.8726348876953125,
      "learning_rate": 8.22822941770062e-09,
      "loss": 1.5516,
      "step": 28950
    },
    {
      "epoch": 4.926426809560262,
      "grad_norm": 5.8677825927734375,
      "learning_rate": 7.865867990979918e-09,
      "loss": 1.5089,
      "step": 28960
    },
    {
      "epoch": 4.928127923790083,
      "grad_norm": 6.619903087615967,
      "learning_rate": 7.51166156215874e-09,
      "loss": 1.4692,
      "step": 28970
    },
    {
      "epoch": 4.9298290380199035,
      "grad_norm": 5.755423069000244,
      "learning_rate": 7.165610561390534e-09,
      "loss": 1.5079,
      "step": 28980
    },
    {
      "epoch": 4.9315301522497235,
      "grad_norm": 5.638603687286377,
      "learning_rate": 6.827715408923169e-09,
      "loss": 1.6086,
      "step": 28990
    },
    {
      "epoch": 4.933231266479544,
      "grad_norm": 6.027549743652344,
      "learning_rate": 6.497976515101174e-09,
      "loss": 1.5588,
      "step": 29000
    },
    {
      "epoch": 4.934932380709364,
      "grad_norm": 6.331847667694092,
      "learning_rate": 6.176394280362749e-09,
      "loss": 1.5272,
      "step": 29010
    },
    {
      "epoch": 4.936633494939185,
      "grad_norm": 5.185964584350586,
      "learning_rate": 5.862969095242751e-09,
      "loss": 1.5543,
      "step": 29020
    },
    {
      "epoch": 4.938334609169006,
      "grad_norm": 6.43770694732666,
      "learning_rate": 5.557701340367474e-09,
      "loss": 1.4981,
      "step": 29030
    },
    {
      "epoch": 4.940035723398826,
      "grad_norm": 7.014509201049805,
      "learning_rate": 5.260591386458374e-09,
      "loss": 1.5031,
      "step": 29040
    },
    {
      "epoch": 4.941736837628647,
      "grad_norm": 7.215841293334961,
      "learning_rate": 4.9716395943290895e-09,
      "loss": 1.5035,
      "step": 29050
    },
    {
      "epoch": 4.943437951858467,
      "grad_norm": 8.555038452148438,
      "learning_rate": 4.69084631488693e-09,
      "loss": 1.5901,
      "step": 29060
    },
    {
      "epoch": 4.945139066088288,
      "grad_norm": 5.919086456298828,
      "learning_rate": 4.418211889129894e-09,
      "loss": 1.546,
      "step": 29070
    },
    {
      "epoch": 4.946840180318109,
      "grad_norm": 5.5981974601745605,
      "learning_rate": 4.153736648149651e-09,
      "loss": 1.5385,
      "step": 29080
    },
    {
      "epoch": 4.948541294547929,
      "grad_norm": 5.864959716796875,
      "learning_rate": 3.897420913127065e-09,
      "loss": 1.5689,
      "step": 29090
    },
    {
      "epoch": 4.95024240877775,
      "grad_norm": 6.871073246002197,
      "learning_rate": 3.649264995335928e-09,
      "loss": 1.5557,
      "step": 29100
    },
    {
      "epoch": 4.95194352300757,
      "grad_norm": 7.135589599609375,
      "learning_rate": 3.4092691961392237e-09,
      "loss": 1.5018,
      "step": 29110
    },
    {
      "epoch": 4.953644637237391,
      "grad_norm": 6.150665283203125,
      "learning_rate": 3.1774338069913717e-09,
      "loss": 1.4963,
      "step": 29120
    },
    {
      "epoch": 4.9553457514672115,
      "grad_norm": 6.790001392364502,
      "learning_rate": 2.9537591094359843e-09,
      "loss": 1.4249,
      "step": 29130
    },
    {
      "epoch": 4.9570468656970315,
      "grad_norm": 7.3811492919921875,
      "learning_rate": 2.7382453751066152e-09,
      "loss": 1.546,
      "step": 29140
    },
    {
      "epoch": 4.958747979926852,
      "grad_norm": 5.9309797286987305,
      "learning_rate": 2.5308928657260123e-09,
      "loss": 1.5004,
      "step": 29150
    },
    {
      "epoch": 4.960449094156672,
      "grad_norm": 5.938831806182861,
      "learning_rate": 2.331701833104625e-09,
      "loss": 1.5158,
      "step": 29160
    },
    {
      "epoch": 4.962150208386493,
      "grad_norm": 5.254824161529541,
      "learning_rate": 2.1406725191435896e-09,
      "loss": 1.5662,
      "step": 29170
    },
    {
      "epoch": 4.963851322616314,
      "grad_norm": 5.960704326629639,
      "learning_rate": 1.957805155830251e-09,
      "loss": 1.5568,
      "step": 29180
    },
    {
      "epoch": 4.965552436846134,
      "grad_norm": 7.006120204925537,
      "learning_rate": 1.7830999652426422e-09,
      "loss": 1.6309,
      "step": 29190
    },
    {
      "epoch": 4.967253551075955,
      "grad_norm": 7.150751113891602,
      "learning_rate": 1.6165571595427641e-09,
      "loss": 1.6519,
      "step": 29200
    },
    {
      "epoch": 4.968954665305775,
      "grad_norm": 6.631286144256592,
      "learning_rate": 1.458176940984052e-09,
      "loss": 1.5314,
      "step": 29210
    },
    {
      "epoch": 4.970655779535596,
      "grad_norm": 5.777578353881836,
      "learning_rate": 1.307959501904657e-09,
      "loss": 1.6618,
      "step": 29220
    },
    {
      "epoch": 4.972356893765417,
      "grad_norm": 7.999058723449707,
      "learning_rate": 1.1659050247304316e-09,
      "loss": 1.5696,
      "step": 29230
    },
    {
      "epoch": 4.974058007995237,
      "grad_norm": 6.014828681945801,
      "learning_rate": 1.0320136819741833e-09,
      "loss": 1.5464,
      "step": 29240
    },
    {
      "epoch": 4.975759122225058,
      "grad_norm": 5.929403781890869,
      "learning_rate": 9.062856362349282e-10,
      "loss": 1.5707,
      "step": 29250
    },
    {
      "epoch": 4.977460236454878,
      "grad_norm": 7.148036956787109,
      "learning_rate": 7.887210401993842e-10,
      "loss": 1.5358,
      "step": 29260
    },
    {
      "epoch": 4.979161350684699,
      "grad_norm": 7.831778049468994,
      "learning_rate": 6.793200366382383e-10,
      "loss": 1.5649,
      "step": 29270
    },
    {
      "epoch": 4.9808624649145194,
      "grad_norm": 6.543948650360107,
      "learning_rate": 5.780827584106254e-10,
      "loss": 1.601,
      "step": 29280
    },
    {
      "epoch": 4.9825635791443394,
      "grad_norm": 6.772243976593018,
      "learning_rate": 4.850093284596501e-10,
      "loss": 1.567,
      "step": 29290
    },
    {
      "epoch": 4.98426469337416,
      "grad_norm": 5.369320869445801,
      "learning_rate": 4.0009985981462535e-10,
      "loss": 1.6045,
      "step": 29300
    },
    {
      "epoch": 4.98596580760398,
      "grad_norm": 5.320074081420898,
      "learning_rate": 3.2335445559032656e-10,
      "loss": 1.5856,
      "step": 29310
    },
    {
      "epoch": 4.987666921833801,
      "grad_norm": 6.3092145919799805,
      "learning_rate": 2.547732089892306e-10,
      "loss": 1.5139,
      "step": 29320
    },
    {
      "epoch": 4.989368036063622,
      "grad_norm": 6.682821750640869,
      "learning_rate": 1.9435620329479807e-10,
      "loss": 1.5759,
      "step": 29330
    },
    {
      "epoch": 4.991069150293442,
      "grad_norm": 6.385833263397217,
      "learning_rate": 1.4210351188043023e-10,
      "loss": 1.49,
      "step": 29340
    },
    {
      "epoch": 4.992770264523263,
      "grad_norm": 7.004361629486084,
      "learning_rate": 9.801519820051227e-11,
      "loss": 1.5654,
      "step": 29350
    },
    {
      "epoch": 4.994471378753083,
      "grad_norm": 5.7715020179748535,
      "learning_rate": 6.209131579787716e-11,
      "loss": 1.4253,
      "step": 29360
    },
    {
      "epoch": 4.996172492982904,
      "grad_norm": 4.990482330322266,
      "learning_rate": 3.4331908297834544e-11,
      "loss": 1.561,
      "step": 29370
    },
    {
      "epoch": 4.997873607212725,
      "grad_norm": 6.701514720916748,
      "learning_rate": 1.4737009412649115e-11,
      "loss": 1.5542,
      "step": 29380
    },
    {
      "epoch": 4.999574721442545,
      "grad_norm": 6.899890899658203,
      "learning_rate": 3.3066429378086008e-12,
      "loss": 1.5503,
      "step": 29390
    }
  ],
  "logging_steps": 10,
  "max_steps": 29390,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "save_steps": 0,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 9.207322559602852e+17,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}