diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -1,9 +1,9 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 0.8004843414607298,
+  "epoch": 0.8505146128020254,
   "eval_steps": 500,
-  "global_step": 14544,
+  "global_step": 15453,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -101815,6 +101815,6369 @@
       "learning_rate": 6.566176945963464e-06,
       "loss": 0.744,
       "step": 14544
+    },
+    {
+      "epoch": 0.8005393802630855,
+      "grad_norm": 0.696674644947052,
+      "learning_rate": 6.565765287444097e-06,
+      "loss": 0.6822,
+      "step": 14545
+    },
+    {
+      "epoch": 0.8005944190654412,
+      "grad_norm": 0.711722195148468,
+      "learning_rate": 6.5653536171568574e-06,
+      "loss": 0.7724,
+      "step": 14546
+    },
+    {
+      "epoch": 0.8006494578677968,
+      "grad_norm": 0.791977047920227,
+      "learning_rate": 6.564941935104838e-06,
+      "loss": 0.7913,
+      "step": 14547
+    },
+    {
+      "epoch": 0.8007044966701524,
+      "grad_norm": 0.6904259920120239,
+      "learning_rate": 6.564530241291135e-06,
+      "loss": 0.7732,
+      "step": 14548
+    },
+    {
+      "epoch": 0.8007595354725081,
+      "grad_norm": 0.6089264750480652,
+      "learning_rate": 6.564118535718842e-06,
+      "loss": 0.6506,
+      "step": 14549
+    },
+    {
+      "epoch": 0.8008145742748638,
+      "grad_norm": 0.6502360105514526,
+      "learning_rate": 6.563706818391051e-06,
+      "loss": 0.6638,
+      "step": 14550
+    },
+    {
+      "epoch": 0.8008696130772195,
+      "grad_norm": 0.6249814033508301,
+      "learning_rate": 6.563295089310859e-06,
+      "loss": 0.7066,
+      "step": 14551
+    },
+    {
+      "epoch": 0.8009246518795751,
+      "grad_norm": 0.8013060688972473,
+      "learning_rate": 6.56288334848136e-06,
+      "loss": 0.7968,
+      "step": 14552
+    },
+    {
+      "epoch": 0.8009796906819308,
+      "grad_norm": 0.7289897799491882,
+      "learning_rate": 6.562471595905648e-06,
+      "loss": 0.752,
+      "step": 14553
+    },
+    {
+      "epoch": 0.8010347294842864,
+      "grad_norm": 0.6774812340736389,
+      "learning_rate": 6.5620598315868176e-06,
+      "loss": 0.8263,
+      "step": 14554
+    },
+    {
+      "epoch": 0.8010897682866421,
+      "grad_norm": 0.6756269931793213,
+      "learning_rate": 6.561648055527965e-06,
+      "loss": 0.8096,
+      "step": 14555
+    },
+    {
+      "epoch": 0.8011448070889977,
+      "grad_norm": 0.7138845324516296,
+      "learning_rate": 6.5612362677321815e-06,
+      "loss": 0.7513,
+      "step": 14556
+    },
+    {
+      "epoch": 0.8011998458913534,
+      "grad_norm": 0.6763927340507507,
+      "learning_rate": 6.5608244682025656e-06,
+      "loss": 0.7975,
+      "step": 14557
+    },
+    {
+      "epoch": 0.8012548846937091,
+      "grad_norm": 0.8147655129432678,
+      "learning_rate": 6.56041265694221e-06,
+      "loss": 0.8192,
+      "step": 14558
+    },
+    {
+      "epoch": 0.8013099234960648,
+      "grad_norm": 0.7272641658782959,
+      "learning_rate": 6.5600008339542095e-06,
+      "loss": 0.7829,
+      "step": 14559
+    },
+    {
+      "epoch": 0.8013649622984204,
+      "grad_norm": 0.7464525103569031,
+      "learning_rate": 6.559588999241661e-06,
+      "loss": 0.7596,
+      "step": 14560
+    },
+    {
+      "epoch": 0.801420001100776,
+      "grad_norm": 0.7236443758010864,
+      "learning_rate": 6.559177152807661e-06,
+      "loss": 0.8151,
+      "step": 14561
+    },
+    {
+      "epoch": 0.8014750399031317,
+      "grad_norm": 0.6752793192863464,
+      "learning_rate": 6.558765294655301e-06,
+      "loss": 0.7578,
+      "step": 14562
+    },
+    {
+      "epoch": 0.8015300787054873,
+      "grad_norm": 0.709994375705719,
+      "learning_rate": 6.558353424787678e-06,
+      "loss": 0.6847,
+      "step": 14563
+    },
+    {
+      "epoch": 0.801585117507843,
+      "grad_norm": 0.7082880139350891,
+      "learning_rate": 6.557941543207889e-06,
+      "loss": 0.7968,
+      "step": 14564
+    },
+    {
+      "epoch": 0.8016401563101987,
+      "grad_norm": 0.692663848400116,
+      "learning_rate": 6.557529649919028e-06,
+      "loss": 0.6625,
+      "step": 14565
+    },
+    {
+      "epoch": 0.8016951951125544,
+      "grad_norm": 0.8464102149009705,
+      "learning_rate": 6.557117744924191e-06,
+      "loss": 0.7383,
+      "step": 14566
+    },
+    {
+      "epoch": 0.80175023391491,
+      "grad_norm": 0.6129899024963379,
+      "learning_rate": 6.5567058282264735e-06,
+      "loss": 0.7007,
+      "step": 14567
+    },
+    {
+      "epoch": 0.8018052727172656,
+      "grad_norm": 0.6458886861801147,
+      "learning_rate": 6.556293899828973e-06,
+      "loss": 0.7019,
+      "step": 14568
+    },
+    {
+      "epoch": 0.8018603115196213,
+      "grad_norm": 0.6543694138526917,
+      "learning_rate": 6.555881959734783e-06,
+      "loss": 0.7254,
+      "step": 14569
+    },
+    {
+      "epoch": 0.801915350321977,
+      "grad_norm": 0.7678859829902649,
+      "learning_rate": 6.555470007947001e-06,
+      "loss": 0.7952,
+      "step": 14570
+    },
+    {
+      "epoch": 0.8019703891243326,
+      "grad_norm": 0.7121342420578003,
+      "learning_rate": 6.555058044468722e-06,
+      "loss": 0.7951,
+      "step": 14571
+    },
+    {
+      "epoch": 0.8020254279266883,
+      "grad_norm": 0.6496285200119019,
+      "learning_rate": 6.554646069303043e-06,
+      "loss": 0.696,
+      "step": 14572
+    },
+    {
+      "epoch": 0.802080466729044,
+      "grad_norm": 0.7206087112426758,
+      "learning_rate": 6.5542340824530614e-06,
+      "loss": 0.7599,
+      "step": 14573
+    },
+    {
+      "epoch": 0.8021355055313997,
+      "grad_norm": 0.7285301685333252,
+      "learning_rate": 6.553822083921872e-06,
+      "loss": 0.7805,
+      "step": 14574
+    },
+    {
+      "epoch": 0.8021905443337553,
+      "grad_norm": 0.7524350881576538,
+      "learning_rate": 6.553410073712572e-06,
+      "loss": 0.7388,
+      "step": 14575
+    },
+    {
+      "epoch": 0.8022455831361109,
+      "grad_norm": 0.7634537220001221,
+      "learning_rate": 6.552998051828256e-06,
+      "loss": 0.6969,
+      "step": 14576
+    },
+    {
+      "epoch": 0.8023006219384666,
+      "grad_norm": 0.6950779557228088,
+      "learning_rate": 6.552586018272024e-06,
+      "loss": 0.8533,
+      "step": 14577
+    },
+    {
+      "epoch": 0.8023556607408223,
+      "grad_norm": 0.694496214389801,
+      "learning_rate": 6.552173973046972e-06,
+      "loss": 0.766,
+      "step": 14578
+    },
+    {
+      "epoch": 0.8024106995431779,
+      "grad_norm": 0.8068329691886902,
+      "learning_rate": 6.5517619161561954e-06,
+      "loss": 0.7642,
+      "step": 14579
+    },
+    {
+      "epoch": 0.8024657383455336,
+      "grad_norm": 0.6933363080024719,
+      "learning_rate": 6.5513498476027905e-06,
+      "loss": 0.8721,
+      "step": 14580
+    },
+    {
+      "epoch": 0.8025207771478893,
+      "grad_norm": 0.7041658163070679,
+      "learning_rate": 6.550937767389857e-06,
+      "loss": 0.6654,
+      "step": 14581
+    },
+    {
+      "epoch": 0.802575815950245,
+      "grad_norm": 0.7080103754997253,
+      "learning_rate": 6.550525675520489e-06,
+      "loss": 0.6917,
+      "step": 14582
+    },
+    {
+      "epoch": 0.8026308547526005,
+      "grad_norm": 0.6644875407218933,
+      "learning_rate": 6.550113571997785e-06,
+      "loss": 0.7674,
+      "step": 14583
+    },
+    {
+      "epoch": 0.8026858935549562,
+      "grad_norm": 0.7660395503044128,
+      "learning_rate": 6.549701456824843e-06,
+      "loss": 0.792,
+      "step": 14584
+    },
+    {
+      "epoch": 0.8027409323573119,
+      "grad_norm": 0.6853451132774353,
+      "learning_rate": 6.549289330004759e-06,
+      "loss": 0.8038,
+      "step": 14585
+    },
+    {
+      "epoch": 0.8027959711596676,
+      "grad_norm": 0.7349985837936401,
+      "learning_rate": 6.548877191540632e-06,
+      "loss": 0.7658,
+      "step": 14586
+    },
+    {
+      "epoch": 0.8028510099620232,
+      "grad_norm": 0.7605637311935425,
+      "learning_rate": 6.548465041435557e-06,
+      "loss": 0.7691,
+      "step": 14587
+    },
+    {
+      "epoch": 0.8029060487643789,
+      "grad_norm": 0.7635177969932556,
+      "learning_rate": 6.548052879692635e-06,
+      "loss": 0.8337,
+      "step": 14588
+    },
+    {
+      "epoch": 0.8029610875667346,
+      "grad_norm": 0.6873355507850647,
+      "learning_rate": 6.5476407063149614e-06,
+      "loss": 0.64,
+      "step": 14589
+    },
+    {
+      "epoch": 0.8030161263690903,
+      "grad_norm": 0.7642813920974731,
+      "learning_rate": 6.547228521305635e-06,
+      "loss": 0.6961,
+      "step": 14590
+    },
+    {
+      "epoch": 0.8030711651714458,
+      "grad_norm": 0.6329793334007263,
+      "learning_rate": 6.546816324667752e-06,
+      "loss": 0.73,
+      "step": 14591
+    },
+    {
+      "epoch": 0.8031262039738015,
+      "grad_norm": 0.6932308673858643,
+      "learning_rate": 6.546404116404412e-06,
+      "loss": 0.7582,
+      "step": 14592
+    },
+    {
+      "epoch": 0.8031812427761572,
+      "grad_norm": 0.699260413646698,
+      "learning_rate": 6.545991896518713e-06,
+      "loss": 0.7219,
+      "step": 14593
+    },
+    {
+      "epoch": 0.8032362815785129,
+      "grad_norm": 0.6217201948165894,
+      "learning_rate": 6.545579665013754e-06,
+      "loss": 0.6237,
+      "step": 14594
+    },
+    {
+      "epoch": 0.8032913203808685,
+      "grad_norm": 0.7078647017478943,
+      "learning_rate": 6.545167421892629e-06,
+      "loss": 0.666,
+      "step": 14595
+    },
+    {
+      "epoch": 0.8033463591832242,
+      "grad_norm": 0.6955916881561279,
+      "learning_rate": 6.544755167158441e-06,
+      "loss": 0.737,
+      "step": 14596
+    },
+    {
+      "epoch": 0.8034013979855799,
+      "grad_norm": 0.8195130825042725,
+      "learning_rate": 6.544342900814287e-06,
+      "loss": 0.787,
+      "step": 14597
+    },
+    {
+      "epoch": 0.8034564367879355,
+      "grad_norm": 0.6160768270492554,
+      "learning_rate": 6.543930622863263e-06,
+      "loss": 0.6141,
+      "step": 14598
+    },
+    {
+      "epoch": 0.8035114755902911,
+      "grad_norm": 0.8483116030693054,
+      "learning_rate": 6.543518333308472e-06,
+      "loss": 0.7639,
+      "step": 14599
+    },
+    {
+      "epoch": 0.8035665143926468,
+      "grad_norm": 0.6937680244445801,
+      "learning_rate": 6.5431060321530105e-06,
+      "loss": 0.7484,
+      "step": 14600
+    },
+    {
+      "epoch": 0.8036215531950025,
+      "grad_norm": 0.6298720836639404,
+      "learning_rate": 6.542693719399975e-06,
+      "loss": 0.6357,
+      "step": 14601
+    },
+    {
+      "epoch": 0.8036765919973582,
+      "grad_norm": 0.6431903839111328,
+      "learning_rate": 6.54228139505247e-06,
+      "loss": 0.6749,
+      "step": 14602
+    },
+    {
+      "epoch": 0.8037316307997138,
+      "grad_norm": 0.8972636461257935,
+      "learning_rate": 6.541869059113588e-06,
+      "loss": 0.8907,
+      "step": 14603
+    },
+    {
+      "epoch": 0.8037866696020695,
+      "grad_norm": 0.7302204966545105,
+      "learning_rate": 6.5414567115864316e-06,
+      "loss": 0.7494,
+      "step": 14604
+    },
+    {
+      "epoch": 0.8038417084044251,
+      "grad_norm": 0.7784821391105652,
+      "learning_rate": 6.541044352474099e-06,
+      "loss": 0.6582,
+      "step": 14605
+    },
+    {
+      "epoch": 0.8038967472067807,
+      "grad_norm": 0.7257398366928101,
+      "learning_rate": 6.54063198177969e-06,
+      "loss": 0.7362,
+      "step": 14606
+    },
+    {
+      "epoch": 0.8039517860091364,
+      "grad_norm": 0.6745980381965637,
+      "learning_rate": 6.540219599506302e-06,
+      "loss": 0.6756,
+      "step": 14607
+    },
+    {
+      "epoch": 0.8040068248114921,
+      "grad_norm": 0.8664490580558777,
+      "learning_rate": 6.539807205657037e-06,
+      "loss": 0.6728,
+      "step": 14608
+    },
+    {
+      "epoch": 0.8040618636138478,
+      "grad_norm": 0.704233705997467,
+      "learning_rate": 6.5393948002349926e-06,
+      "loss": 0.7713,
+      "step": 14609
+    },
+    {
+      "epoch": 0.8041169024162034,
+      "grad_norm": 0.7709019780158997,
+      "learning_rate": 6.538982383243271e-06,
+      "loss": 0.8148,
+      "step": 14610
+    },
+    {
+      "epoch": 0.8041719412185591,
+      "grad_norm": 0.7056839466094971,
+      "learning_rate": 6.538569954684967e-06,
+      "loss": 0.7143,
+      "step": 14611
+    },
+    {
+      "epoch": 0.8042269800209148,
+      "grad_norm": 0.715506374835968,
+      "learning_rate": 6.538157514563184e-06,
+      "loss": 0.7932,
+      "step": 14612
+    },
+    {
+      "epoch": 0.8042820188232704,
+      "grad_norm": 0.8245391845703125,
+      "learning_rate": 6.537745062881021e-06,
+      "loss": 0.7569,
+      "step": 14613
+    },
+    {
+      "epoch": 0.804337057625626,
+      "grad_norm": 0.6912628412246704,
+      "learning_rate": 6.5373325996415794e-06,
+      "loss": 0.7174,
+      "step": 14614
+    },
+    {
+      "epoch": 0.8043920964279817,
+      "grad_norm": 0.6994870901107788,
+      "learning_rate": 6.536920124847955e-06,
+      "loss": 0.6174,
+      "step": 14615
+    },
+    {
+      "epoch": 0.8044471352303374,
+      "grad_norm": 0.6660363674163818,
+      "learning_rate": 6.536507638503251e-06,
+      "loss": 0.8065,
+      "step": 14616
+    },
+    {
+      "epoch": 0.8045021740326931,
+      "grad_norm": 0.6742863059043884,
+      "learning_rate": 6.536095140610567e-06,
+      "loss": 0.7984,
+      "step": 14617
+    },
+    {
+      "epoch": 0.8045572128350487,
+      "grad_norm": 0.6868259906768799,
+      "learning_rate": 6.535682631173005e-06,
+      "loss": 0.7907,
+      "step": 14618
+    },
+    {
+      "epoch": 0.8046122516374044,
+      "grad_norm": 0.7442048788070679,
+      "learning_rate": 6.5352701101936615e-06,
+      "loss": 0.7893,
+      "step": 14619
+    },
+    {
+      "epoch": 0.80466729043976,
+      "grad_norm": 0.7389286756515503,
+      "learning_rate": 6.534857577675639e-06,
+      "loss": 0.827,
+      "step": 14620
+    },
+    {
+      "epoch": 0.8047223292421157,
+      "grad_norm": 0.6679701209068298,
+      "learning_rate": 6.534445033622036e-06,
+      "loss": 0.6721,
+      "step": 14621
+    },
+    {
+      "epoch": 0.8047773680444713,
+      "grad_norm": 0.6372442841529846,
+      "learning_rate": 6.534032478035957e-06,
+      "loss": 0.7381,
+      "step": 14622
+    },
+    {
+      "epoch": 0.804832406846827,
+      "grad_norm": 0.7682638764381409,
+      "learning_rate": 6.533619910920501e-06,
+      "loss": 0.7003,
+      "step": 14623
+    },
+    {
+      "epoch": 0.8048874456491827,
+      "grad_norm": 0.6821291446685791,
+      "learning_rate": 6.533207332278767e-06,
+      "loss": 0.8164,
+      "step": 14624
+    },
+    {
+      "epoch": 0.8049424844515384,
+      "grad_norm": 0.6591019034385681,
+      "learning_rate": 6.532794742113858e-06,
+      "loss": 0.6772,
+      "step": 14625
+    },
+    {
+      "epoch": 0.804997523253894,
+      "grad_norm": 0.7331292033195496,
+      "learning_rate": 6.532382140428874e-06,
+      "loss": 0.7606,
+      "step": 14626
+    },
+    {
+      "epoch": 0.8050525620562496,
+      "grad_norm": 0.9654768705368042,
+      "learning_rate": 6.531969527226917e-06,
+      "loss": 0.9196,
+      "step": 14627
+    },
+    {
+      "epoch": 0.8051076008586053,
+      "grad_norm": 0.6320267915725708,
+      "learning_rate": 6.5315569025110844e-06,
+      "loss": 0.6982,
+      "step": 14628
+    },
+    {
+      "epoch": 0.805162639660961,
+      "grad_norm": 0.6921746134757996,
+      "learning_rate": 6.531144266284481e-06,
+      "loss": 0.7176,
+      "step": 14629
+    },
+    {
+      "epoch": 0.8052176784633166,
+      "grad_norm": 0.7233335375785828,
+      "learning_rate": 6.530731618550208e-06,
+      "loss": 0.8388,
+      "step": 14630
+    },
+    {
+      "epoch": 0.8052727172656723,
+      "grad_norm": 0.6576363444328308,
+      "learning_rate": 6.530318959311366e-06,
+      "loss": 0.7511,
+      "step": 14631
+    },
+    {
+      "epoch": 0.805327756068028,
+      "grad_norm": 0.6921162009239197,
+      "learning_rate": 6.529906288571055e-06,
+      "loss": 0.8161,
+      "step": 14632
+    },
+    {
+      "epoch": 0.8053827948703837,
+      "grad_norm": 0.7314246296882629,
+      "learning_rate": 6.529493606332379e-06,
+      "loss": 0.7824,
+      "step": 14633
+    },
+    {
+      "epoch": 0.8054378336727392,
+      "grad_norm": 0.6419001221656799,
+      "learning_rate": 6.529080912598438e-06,
+      "loss": 0.7593,
+      "step": 14634
+    },
+    {
+      "epoch": 0.8054928724750949,
+      "grad_norm": 0.9500213861465454,
+      "learning_rate": 6.528668207372335e-06,
+      "loss": 0.7429,
+      "step": 14635
+    },
+    {
+      "epoch": 0.8055479112774506,
+      "grad_norm": 0.7299035787582397,
+      "learning_rate": 6.52825549065717e-06,
+      "loss": 0.8064,
+      "step": 14636
+    },
+    {
+      "epoch": 0.8056029500798063,
+      "grad_norm": 0.6231887936592102,
+      "learning_rate": 6.527842762456046e-06,
+      "loss": 0.6177,
+      "step": 14637
+    },
+    {
+      "epoch": 0.8056579888821619,
+      "grad_norm": 0.6219315528869629,
+      "learning_rate": 6.527430022772066e-06,
+      "loss": 0.6781,
+      "step": 14638
+    },
+    {
+      "epoch": 0.8057130276845176,
+      "grad_norm": 0.696861982345581,
+      "learning_rate": 6.527017271608329e-06,
+      "loss": 0.7508,
+      "step": 14639
+    },
+    {
+      "epoch": 0.8057680664868733,
+      "grad_norm": 0.7849573493003845,
+      "learning_rate": 6.5266045089679394e-06,
+      "loss": 0.7347,
+      "step": 14640
+    },
+    {
+      "epoch": 0.805823105289229,
+      "grad_norm": 0.6350993514060974,
+      "learning_rate": 6.526191734853999e-06,
+      "loss": 0.6863,
+      "step": 14641
+    },
+    {
+      "epoch": 0.8058781440915845,
+      "grad_norm": 0.6293141841888428,
+      "learning_rate": 6.5257789492696115e-06,
+      "loss": 0.7288,
+      "step": 14642
+    },
+    {
+      "epoch": 0.8059331828939402,
+      "grad_norm": 0.7801508903503418,
+      "learning_rate": 6.525366152217876e-06,
+      "loss": 0.7592,
+      "step": 14643
+    },
+    {
+      "epoch": 0.8059882216962959,
+      "grad_norm": 0.7031479477882385,
+      "learning_rate": 6.5249533437018964e-06,
+      "loss": 0.8677,
+      "step": 14644
+    },
+    {
+      "epoch": 0.8060432604986516,
+      "grad_norm": 0.7052507996559143,
+      "learning_rate": 6.524540523724777e-06,
+      "loss": 0.7957,
+      "step": 14645
+    },
+    {
+      "epoch": 0.8060982993010072,
+      "grad_norm": 0.669743537902832,
+      "learning_rate": 6.524127692289619e-06,
+      "loss": 0.7163,
+      "step": 14646
+    },
+    {
+      "epoch": 0.8061533381033629,
+      "grad_norm": 0.7180876731872559,
+      "learning_rate": 6.523714849399525e-06,
+      "loss": 0.8814,
+      "step": 14647
+    },
+    {
+      "epoch": 0.8062083769057186,
+      "grad_norm": 0.6617746353149414,
+      "learning_rate": 6.523301995057597e-06,
+      "loss": 0.721,
+      "step": 14648
+    },
+    {
+      "epoch": 0.8062634157080741,
+      "grad_norm": 0.6464657783508301,
+      "learning_rate": 6.5228891292669404e-06,
+      "loss": 0.7334,
+      "step": 14649
+    },
+    {
+      "epoch": 0.8063184545104298,
+      "grad_norm": 0.7648638486862183,
+      "learning_rate": 6.522476252030658e-06,
+      "loss": 0.7701,
+      "step": 14650
+    },
+    {
+      "epoch": 0.8063734933127855,
+      "grad_norm": 0.7313019037246704,
+      "learning_rate": 6.522063363351851e-06,
+      "loss": 0.7912,
+      "step": 14651
+    },
+    {
+      "epoch": 0.8064285321151412,
+      "grad_norm": 0.6175631284713745,
+      "learning_rate": 6.5216504632336195e-06,
+      "loss": 0.7568,
+      "step": 14652
+    },
+    {
+      "epoch": 0.8064835709174968,
+      "grad_norm": 0.6935408711433411,
+      "learning_rate": 6.521237551679074e-06,
+      "loss": 0.7622,
+      "step": 14653
+    },
+    {
+      "epoch": 0.8065386097198525,
+      "grad_norm": 0.7232398390769958,
+      "learning_rate": 6.520824628691314e-06,
+      "loss": 0.7908,
+      "step": 14654
+    },
+    {
+      "epoch": 0.8065936485222082,
+      "grad_norm": 0.6642309427261353,
+      "learning_rate": 6.520411694273443e-06,
+      "loss": 0.7355,
+      "step": 14655
+    },
+    {
+      "epoch": 0.8066486873245639,
+      "grad_norm": 0.6679350137710571,
+      "learning_rate": 6.5199987484285635e-06,
+      "loss": 0.735,
+      "step": 14656
+    },
+    {
+      "epoch": 0.8067037261269194,
+      "grad_norm": 0.6861871480941772,
+      "learning_rate": 6.519585791159782e-06,
+      "loss": 0.6744,
+      "step": 14657
+    },
+    {
+      "epoch": 0.8067587649292751,
+      "grad_norm": 0.7689095735549927,
+      "learning_rate": 6.519172822470199e-06,
+      "loss": 0.6888,
+      "step": 14658
+    },
+    {
+      "epoch": 0.8068138037316308,
+      "grad_norm": 0.6604742407798767,
+      "learning_rate": 6.5187598423629206e-06,
+      "loss": 0.6943,
+      "step": 14659
+    },
+    {
+      "epoch": 0.8068688425339865,
+      "grad_norm": 0.6478890776634216,
+      "learning_rate": 6.518346850841049e-06,
+      "loss": 0.7161,
+      "step": 14660
+    },
+    {
+      "epoch": 0.8069238813363421,
+      "grad_norm": 0.6213741302490234,
+      "learning_rate": 6.517933847907689e-06,
+      "loss": 0.68,
+      "step": 14661
+    },
+    {
+      "epoch": 0.8069789201386978,
+      "grad_norm": 0.7663899660110474,
+      "learning_rate": 6.517520833565945e-06,
+      "loss": 0.7498,
+      "step": 14662
+    },
+    {
+      "epoch": 0.8070339589410535,
+      "grad_norm": 0.653498649597168,
+      "learning_rate": 6.517107807818921e-06,
+      "loss": 0.7433,
+      "step": 14663
+    },
+    {
+      "epoch": 0.8070889977434091,
+      "grad_norm": 0.7618738412857056,
+      "learning_rate": 6.51669477066972e-06,
+      "loss": 0.7499,
+      "step": 14664
+    },
+    {
+      "epoch": 0.8071440365457647,
+      "grad_norm": 0.5960344672203064,
+      "learning_rate": 6.516281722121447e-06,
+      "loss": 0.6005,
+      "step": 14665
+    },
+    {
+      "epoch": 0.8071990753481204,
+      "grad_norm": 0.6768549084663391,
+      "learning_rate": 6.5158686621772075e-06,
+      "loss": 0.6859,
+      "step": 14666
+    },
+    {
+      "epoch": 0.8072541141504761,
+      "grad_norm": 0.6475711464881897,
+      "learning_rate": 6.515455590840104e-06,
+      "loss": 0.7582,
+      "step": 14667
+    },
+    {
+      "epoch": 0.8073091529528318,
+      "grad_norm": 0.7188607454299927,
+      "learning_rate": 6.5150425081132414e-06,
+      "loss": 0.7241,
+      "step": 14668
+    },
+    {
+      "epoch": 0.8073641917551874,
+      "grad_norm": 0.6507582068443298,
+      "learning_rate": 6.514629413999727e-06,
+      "loss": 0.7659,
+      "step": 14669
+    },
+    {
+      "epoch": 0.807419230557543,
+      "grad_norm": 0.6676538586616516,
+      "learning_rate": 6.514216308502661e-06,
+      "loss": 0.7336,
+      "step": 14670
+    },
+    {
+      "epoch": 0.8074742693598987,
+      "grad_norm": 0.7141211628913879,
+      "learning_rate": 6.513803191625152e-06,
+      "loss": 0.8121,
+      "step": 14671
+    },
+    {
+      "epoch": 0.8075293081622544,
+      "grad_norm": 0.7497949600219727,
+      "learning_rate": 6.513390063370302e-06,
+      "loss": 0.7238,
+      "step": 14672
+    },
+    {
+      "epoch": 0.80758434696461,
+      "grad_norm": 0.671271562576294,
+      "learning_rate": 6.51297692374122e-06,
+      "loss": 0.7876,
+      "step": 14673
+    },
+    {
+      "epoch": 0.8076393857669657,
+      "grad_norm": 0.7081878781318665,
+      "learning_rate": 6.512563772741008e-06,
+      "loss": 0.6774,
+      "step": 14674
+    },
+    {
+      "epoch": 0.8076944245693214,
+      "grad_norm": 0.640925943851471,
+      "learning_rate": 6.512150610372769e-06,
+      "loss": 0.7094,
+      "step": 14675
+    },
+    {
+      "epoch": 0.8077494633716771,
+      "grad_norm": 0.6333619952201843,
+      "learning_rate": 6.511737436639611e-06,
+      "loss": 0.6439,
+      "step": 14676
+    },
+    {
+      "epoch": 0.8078045021740327,
+      "grad_norm": 0.7294490337371826,
+      "learning_rate": 6.511324251544642e-06,
+      "loss": 0.7786,
+      "step": 14677
+    },
+    {
+      "epoch": 0.8078595409763883,
+      "grad_norm": 0.6488819718360901,
+      "learning_rate": 6.510911055090963e-06,
+      "loss": 0.7495,
+      "step": 14678
+    },
+    {
+      "epoch": 0.807914579778744,
+      "grad_norm": 0.6535395383834839,
+      "learning_rate": 6.51049784728168e-06,
+      "loss": 0.6713,
+      "step": 14679
+    },
+    {
+      "epoch": 0.8079696185810997,
+      "grad_norm": 0.6795744895935059,
+      "learning_rate": 6.5100846281198995e-06,
+      "loss": 0.71,
+      "step": 14680
+    },
+    {
+      "epoch": 0.8080246573834553,
+      "grad_norm": 0.661171019077301,
+      "learning_rate": 6.509671397608728e-06,
+      "loss": 0.7009,
+      "step": 14681
+    },
+    {
+      "epoch": 0.808079696185811,
+      "grad_norm": 0.6474859118461609,
+      "learning_rate": 6.50925815575127e-06,
+      "loss": 0.7268,
+      "step": 14682
+    },
+    {
+      "epoch": 0.8081347349881667,
+      "grad_norm": 0.676891565322876,
+      "learning_rate": 6.508844902550633e-06,
+      "loss": 0.8748,
+      "step": 14683
+    },
+    {
+      "epoch": 0.8081897737905224,
+      "grad_norm": 0.9747083783149719,
+      "learning_rate": 6.50843163800992e-06,
+      "loss": 0.6817,
+      "step": 14684
+    },
+    {
+      "epoch": 0.808244812592878,
+      "grad_norm": 0.655274510383606,
+      "learning_rate": 6.50801836213224e-06,
+      "loss": 0.7675,
+      "step": 14685
+    },
+    {
+      "epoch": 0.8082998513952336,
+      "grad_norm": 0.6916972398757935,
+      "learning_rate": 6.507605074920697e-06,
+      "loss": 0.7862,
+      "step": 14686
+    },
+    {
+      "epoch": 0.8083548901975893,
+      "grad_norm": 0.7079103589057922,
+      "learning_rate": 6.5071917763783975e-06,
+      "loss": 0.671,
+      "step": 14687
+    },
+    {
+      "epoch": 0.808409928999945,
+      "grad_norm": 0.7460986375808716,
+      "learning_rate": 6.506778466508447e-06,
+      "loss": 0.7136,
+      "step": 14688
+    },
+    {
+      "epoch": 0.8084649678023006,
+      "grad_norm": 0.6531261801719666,
+      "learning_rate": 6.5063651453139555e-06,
+      "loss": 0.811,
+      "step": 14689
+    },
+    {
+      "epoch": 0.8085200066046563,
+      "grad_norm": 0.7160762548446655,
+      "learning_rate": 6.505951812798025e-06,
+      "loss": 0.8368,
+      "step": 14690
+    },
+    {
+      "epoch": 0.808575045407012,
+      "grad_norm": 0.7230852842330933,
+      "learning_rate": 6.505538468963763e-06,
+      "loss": 0.6908,
+      "step": 14691
+    },
+    {
+      "epoch": 0.8086300842093676,
+      "grad_norm": 0.6912978887557983,
+      "learning_rate": 6.505125113814278e-06,
+      "loss": 0.6716,
+      "step": 14692
+    },
+    {
+      "epoch": 0.8086851230117232,
+      "grad_norm": 0.6745109558105469,
+      "learning_rate": 6.504711747352677e-06,
+      "loss": 0.7119,
+      "step": 14693
+    },
+    {
+      "epoch": 0.8087401618140789,
+      "grad_norm": 0.678657054901123,
+      "learning_rate": 6.5042983695820624e-06,
+      "loss": 0.7548,
+      "step": 14694
+    },
+    {
+      "epoch": 0.8087952006164346,
+      "grad_norm": 0.7501665949821472,
+      "learning_rate": 6.503884980505546e-06,
+      "loss": 0.7493,
+      "step": 14695
+    },
+    {
+      "epoch": 0.8088502394187902,
+      "grad_norm": 0.6181747317314148,
+      "learning_rate": 6.503471580126232e-06,
+      "loss": 0.7217,
+      "step": 14696
+    },
+    {
+      "epoch": 0.8089052782211459,
+      "grad_norm": 0.6548559069633484,
+      "learning_rate": 6.5030581684472295e-06,
+      "loss": 0.7448,
+      "step": 14697
+    },
+    {
+      "epoch": 0.8089603170235016,
+      "grad_norm": 0.7716642022132874,
+      "learning_rate": 6.5026447454716426e-06,
+      "loss": 0.8794,
+      "step": 14698
+    },
+    {
+      "epoch": 0.8090153558258573,
+      "grad_norm": 0.861995279788971,
+      "learning_rate": 6.502231311202581e-06,
+      "loss": 0.7839,
+      "step": 14699
+    },
+    {
+      "epoch": 0.8090703946282128,
+      "grad_norm": 0.796821117401123,
+      "learning_rate": 6.501817865643149e-06,
+      "loss": 0.8541,
+      "step": 14700
+    },
+    {
+      "epoch": 0.8091254334305685,
+      "grad_norm": 0.6995296478271484,
+      "learning_rate": 6.501404408796457e-06,
+      "loss": 0.677,
+      "step": 14701
+    },
+    {
+      "epoch": 0.8091804722329242,
+      "grad_norm": 0.6681582927703857,
+      "learning_rate": 6.500990940665611e-06,
+      "loss": 0.7754,
+      "step": 14702
+    },
+    {
+      "epoch": 0.8092355110352799,
+      "grad_norm": 0.5945298671722412,
+      "learning_rate": 6.50057746125372e-06,
+      "loss": 0.6762,
+      "step": 14703
+    },
+    {
+      "epoch": 0.8092905498376355,
+      "grad_norm": 0.672554612159729,
+      "learning_rate": 6.500163970563889e-06,
+      "loss": 0.6967,
+      "step": 14704
+    },
+    {
+      "epoch": 0.8093455886399912,
+      "grad_norm": 0.6375272870063782,
+      "learning_rate": 6.499750468599227e-06,
+      "loss": 0.7291,
+      "step": 14705
+    },
+    {
+      "epoch": 0.8094006274423469,
+      "grad_norm": 0.6369407773017883,
+      "learning_rate": 6.499336955362844e-06,
+      "loss": 0.6939,
+      "step": 14706
+    },
+    {
+      "epoch": 0.8094556662447026,
+      "grad_norm": 0.6497664451599121,
+      "learning_rate": 6.498923430857844e-06,
+      "loss": 0.7207,
+      "step": 14707
+    },
+    {
+      "epoch": 0.8095107050470581,
+      "grad_norm": 0.7345920205116272,
+      "learning_rate": 6.498509895087337e-06,
+      "loss": 0.8373,
+      "step": 14708
+    },
+    {
+      "epoch": 0.8095657438494138,
+      "grad_norm": 0.6824862957000732,
+      "learning_rate": 6.4980963480544324e-06,
+      "loss": 0.7531,
+      "step": 14709
+    },
+    {
+      "epoch": 0.8096207826517695,
+      "grad_norm": 0.7067939639091492,
+      "learning_rate": 6.497682789762236e-06,
+      "loss": 0.6951,
+      "step": 14710
+    },
+    {
+      "epoch": 0.8096758214541252,
+      "grad_norm": 0.6856693625450134,
+      "learning_rate": 6.497269220213856e-06,
+      "loss": 0.7264,
+      "step": 14711
+    },
+    {
+      "epoch": 0.8097308602564808,
+      "grad_norm": 0.6881466507911682,
+      "learning_rate": 6.4968556394124e-06,
+      "loss": 0.7837,
+      "step": 14712
+    },
+    {
+      "epoch": 0.8097858990588365,
+      "grad_norm": 0.6211455464363098,
+      "learning_rate": 6.49644204736098e-06,
+      "loss": 0.7278,
+      "step": 14713
+    },
+    {
+      "epoch": 0.8098409378611922,
+      "grad_norm": 0.688604474067688,
+      "learning_rate": 6.496028444062701e-06,
+      "loss": 0.7786,
+      "step": 14714
+    },
+    {
+      "epoch": 0.8098959766635478,
+      "grad_norm": 0.6615015268325806,
+      "learning_rate": 6.495614829520673e-06,
+      "loss": 0.7014,
+      "step": 14715
+    },
+    {
+      "epoch": 0.8099510154659034,
+      "grad_norm": 0.712661623954773,
+      "learning_rate": 6.495201203738004e-06,
+      "loss": 0.6792,
+      "step": 14716
+    },
+    {
+      "epoch": 0.8100060542682591,
+      "grad_norm": 0.6737191677093506,
+      "learning_rate": 6.494787566717803e-06,
+      "loss": 0.7937,
+      "step": 14717
+    },
+    {
+      "epoch": 0.8100610930706148,
+      "grad_norm": 0.8007351160049438,
+      "learning_rate": 6.494373918463179e-06,
+      "loss": 0.8367,
+      "step": 14718
+    },
+    {
+      "epoch": 0.8101161318729705,
+      "grad_norm": 0.7500883936882019,
+      "learning_rate": 6.493960258977241e-06,
+      "loss": 0.8102,
+      "step": 14719
+    },
+    {
+      "epoch": 0.8101711706753261,
+      "grad_norm": 0.7605966925621033,
+      "learning_rate": 6.493546588263097e-06,
+      "loss": 0.8316,
+      "step": 14720
+    },
+    {
+      "epoch": 0.8102262094776818,
+      "grad_norm": 0.746762216091156,
+      "learning_rate": 6.493132906323858e-06,
+      "loss": 0.7765,
+      "step": 14721
+    },
+    {
+      "epoch": 0.8102812482800374,
+      "grad_norm": 0.6034676432609558,
+      "learning_rate": 6.49271921316263e-06,
+      "loss": 0.7109,
+      "step": 14722
+    },
+    {
+      "epoch": 0.8103362870823931,
+      "grad_norm": 0.6965274810791016,
+      "learning_rate": 6.492305508782525e-06,
+      "loss": 0.8156,
+      "step": 14723
+    },
+    {
+      "epoch": 0.8103913258847487,
+      "grad_norm": 0.6813820004463196,
+      "learning_rate": 6.4918917931866495e-06,
+      "loss": 0.7016,
+      "step": 14724
+    },
+    {
+      "epoch": 0.8104463646871044,
+      "grad_norm": 0.8055655360221863,
+      "learning_rate": 6.491478066378117e-06,
+      "loss": 0.7837,
+      "step": 14725
+    },
+    {
+      "epoch": 0.8105014034894601,
+      "grad_norm": 0.6131647229194641,
+      "learning_rate": 6.491064328360033e-06,
+      "loss": 0.6716,
+      "step": 14726
+    },
+    {
+      "epoch": 0.8105564422918158,
+      "grad_norm": 0.6845986247062683,
+      "learning_rate": 6.49065057913551e-06,
+      "loss": 0.8112,
+      "step": 14727
+    },
+    {
+      "epoch": 0.8106114810941714,
+      "grad_norm": 0.6867175698280334,
+      "learning_rate": 6.490236818707653e-06,
+      "loss": 0.7953,
+      "step": 14728
+    },
+    {
+      "epoch": 0.810666519896527,
+      "grad_norm": 0.7170011401176453,
+      "learning_rate": 6.489823047079578e-06,
+      "loss": 0.8108,
+      "step": 14729
+    },
+    {
+      "epoch": 0.8107215586988827,
+      "grad_norm": 0.6280927658081055,
+      "learning_rate": 6.489409264254393e-06,
+      "loss": 0.6807,
+      "step": 14730
+    },
+    {
+      "epoch": 0.8107765975012384,
+      "grad_norm": 0.8344630002975464,
+      "learning_rate": 6.488995470235204e-06,
+      "loss": 0.7555,
+      "step": 14731
+    },
+    {
+      "epoch": 0.810831636303594,
+      "grad_norm": 0.6674200296401978,
+      "learning_rate": 6.488581665025125e-06,
+      "loss": 0.5732,
+      "step": 14732
+    },
+    {
+      "epoch": 0.8108866751059497,
+      "grad_norm": 0.7843313217163086,
+      "learning_rate": 6.4881678486272646e-06,
+      "loss": 0.6689,
+      "step": 14733
+    },
+    {
+      "epoch": 0.8109417139083054,
+      "grad_norm": 0.6951878666877747,
+      "learning_rate": 6.487754021044732e-06,
+      "loss": 0.8005,
+      "step": 14734
+    },
+    {
+      "epoch": 0.810996752710661,
+      "grad_norm": 0.7773714065551758,
+      "learning_rate": 6.487340182280639e-06,
+      "loss": 0.8151,
+      "step": 14735
+    },
+    {
+      "epoch": 0.8110517915130167,
+      "grad_norm": 0.824998140335083,
+      "learning_rate": 6.486926332338095e-06,
+      "loss": 0.7947,
+      "step": 14736
+    },
+    {
+      "epoch": 0.8111068303153723,
+      "grad_norm": 0.6411730647087097,
+      "learning_rate": 6.486512471220212e-06,
+      "loss": 0.7272,
+      "step": 14737
+    },
+    {
+      "epoch": 0.811161869117728,
+      "grad_norm": 0.6758518815040588,
+      "learning_rate": 6.486098598930097e-06,
+      "loss": 0.6676,
+      "step": 14738
+    },
+    {
+      "epoch": 0.8112169079200836,
+      "grad_norm": 0.7147762179374695,
+      "learning_rate": 6.485684715470866e-06,
+      "loss": 0.7796,
+      "step": 14739
+    },
+    {
+      "epoch": 0.8112719467224393,
+      "grad_norm": 0.7641217112541199,
+      "learning_rate": 6.485270820845623e-06,
+      "loss": 0.7943,
+      "step": 14740
+    },
+    {
+      "epoch": 0.811326985524795,
+      "grad_norm": 0.6947311162948608,
+      "learning_rate": 6.484856915057482e-06,
+      "loss": 0.7791,
+      "step": 14741
+    },
+    {
+      "epoch": 0.8113820243271507,
+      "grad_norm": 0.6781480312347412,
+      "learning_rate": 6.4844429981095565e-06,
+      "loss": 0.7399,
+      "step": 14742
+    },
+    {
+      "epoch": 0.8114370631295063,
+      "grad_norm": 0.6716181039810181,
+      "learning_rate": 6.484029070004953e-06,
+      "loss": 0.8111,
+      "step": 14743
+    },
+    {
+      "epoch": 0.8114921019318619,
+      "grad_norm": 0.8642836213111877,
+      "learning_rate": 6.4836151307467854e-06,
+      "loss": 0.756,
+      "step": 14744
+    },
+    {
+      "epoch": 0.8115471407342176,
+      "grad_norm": 0.5997880101203918,
+      "learning_rate": 6.483201180338163e-06,
+      "loss": 0.6043,
+      "step": 14745
+    },
+    {
+      "epoch": 0.8116021795365733,
+      "grad_norm": 0.7397846579551697,
+      "learning_rate": 6.4827872187821985e-06,
+      "loss": 0.848,
+      "step": 14746
+    },
+    {
+      "epoch": 0.8116572183389289,
+      "grad_norm": 0.7586305141448975,
+      "learning_rate": 6.482373246082001e-06,
+      "loss": 0.802,
+      "step": 14747
+    },
+    {
+      "epoch": 0.8117122571412846,
+      "grad_norm": 0.705182671546936,
+      "learning_rate": 6.4819592622406825e-06,
+      "loss": 0.7484,
+      "step": 14748
+    },
+    {
+      "epoch": 0.8117672959436403,
+      "grad_norm": 0.7092768549919128,
+      "learning_rate": 6.481545267261357e-06,
+      "loss": 0.7031,
+      "step": 14749
+    },
+    {
+      "epoch": 0.811822334745996,
+      "grad_norm": 0.6800800561904907,
+      "learning_rate": 6.4811312611471325e-06,
+      "loss": 0.7253,
+      "step": 14750
+    },
+    {
+      "epoch": 0.8118773735483515,
+      "grad_norm": 0.6862359642982483,
+      "learning_rate": 6.4807172439011215e-06,
+      "loss": 0.818,
+      "step": 14751
+    },
+    {
+      "epoch": 0.8119324123507072,
+      "grad_norm": 0.6928552389144897,
+      "learning_rate": 6.480303215526436e-06,
+      "loss": 0.7459,
+      "step": 14752
+    },
+    {
+      "epoch": 0.8119874511530629,
+      "grad_norm": 0.6869228482246399,
+      "learning_rate": 6.479889176026189e-06,
+      "loss": 0.7024,
+      "step": 14753
+    },
+    {
+      "epoch": 0.8120424899554186,
+      "grad_norm": 0.7036190032958984,
+      "learning_rate": 6.479475125403489e-06,
+      "loss": 0.766,
+      "step": 14754
+    },
+    {
+      "epoch": 0.8120975287577742,
+      "grad_norm": 0.6574180722236633,
+      "learning_rate": 6.479061063661452e-06,
+      "loss": 0.7355,
+      "step": 14755
+    },
+    {
+      "epoch": 0.8121525675601299,
+      "grad_norm": 0.6424534916877747,
+      "learning_rate": 6.478646990803188e-06,
+      "loss": 0.6837,
+      "step": 14756
+    },
+    {
+      "epoch": 0.8122076063624856,
+      "grad_norm": 0.6922320127487183,
+      "learning_rate": 6.478232906831808e-06,
+      "loss": 0.7535,
+      "step": 14757
+    },
+    {
+      "epoch": 0.8122626451648413,
+      "grad_norm": 0.6424705386161804,
+      "learning_rate": 6.477818811750426e-06,
+      "loss": 0.691,
+      "step": 14758
+    },
+    {
+      "epoch": 0.8123176839671968,
+      "grad_norm": 0.6180749535560608,
+      "learning_rate": 6.4774047055621525e-06,
+      "loss": 0.6944,
+      "step": 14759
+    },
+    {
+      "epoch": 0.8123727227695525,
+      "grad_norm": 0.8718746900558472,
+      "learning_rate": 6.4769905882701e-06,
+      "loss": 0.89,
+      "step": 14760
+    },
+    {
+      "epoch": 0.8124277615719082,
+      "grad_norm": 0.6664311289787292,
+      "learning_rate": 6.476576459877384e-06,
+      "loss": 0.7144,
+      "step": 14761
+    },
+    {
+      "epoch": 0.8124828003742639,
+      "grad_norm": 0.6547374129295349,
+      "learning_rate": 6.476162320387112e-06,
+      "loss": 0.7292,
+      "step": 14762
+    },
+    {
+      "epoch": 0.8125378391766195,
+      "grad_norm": 0.7387503385543823,
+      "learning_rate": 6.475748169802401e-06,
+      "loss": 0.7388,
+      "step": 14763
+    },
+    {
+      "epoch": 0.8125928779789752,
+      "grad_norm": 0.6013749241828918,
+      "learning_rate": 6.475334008126361e-06,
+      "loss": 0.6853,
+      "step": 14764
+    },
+    {
+      "epoch": 0.8126479167813309,
+      "grad_norm": 0.6720583438873291,
+      "learning_rate": 6.474919835362105e-06,
+      "loss": 0.7392,
+      "step": 14765
+    },
+    {
+      "epoch": 0.8127029555836865,
+      "grad_norm": 0.6651661992073059,
+      "learning_rate": 6.474505651512748e-06,
+      "loss": 0.7586,
+      "step": 14766
+    },
+    {
+      "epoch": 0.8127579943860421,
+      "grad_norm": 0.7653207182884216,
+      "learning_rate": 6.474091456581401e-06,
+      "loss": 0.9182,
+      "step": 14767
+    },
+    {
+      "epoch": 0.8128130331883978,
+      "grad_norm": 0.6322795152664185,
+      "learning_rate": 6.473677250571176e-06,
+      "loss": 0.6954,
+      "step": 14768
+    },
+    {
+      "epoch": 0.8128680719907535,
+      "grad_norm": 0.7423616647720337,
+      "learning_rate": 6.4732630334851885e-06,
+      "loss": 0.748,
+      "step": 14769
+    },
+    {
+      "epoch": 0.8129231107931092,
+      "grad_norm": 0.5989160537719727,
+      "learning_rate": 6.472848805326549e-06,
+      "loss": 0.6571,
+      "step": 14770
+    },
+    {
+      "epoch": 0.8129781495954648,
+      "grad_norm": 0.695566713809967,
+      "learning_rate": 6.472434566098373e-06,
+      "loss": 0.6936,
+      "step": 14771
+    },
+    {
+      "epoch": 0.8130331883978205,
+      "grad_norm": 0.6993961930274963,
+      "learning_rate": 6.4720203158037734e-06,
+      "loss": 0.8283,
+      "step": 14772
+    },
+    {
+      "epoch": 0.8130882272001762,
+      "grad_norm": 0.6430020928382874,
+      "learning_rate": 6.471606054445861e-06,
+      "loss": 0.6882,
+      "step": 14773
+    },
+    {
+      "epoch": 0.8131432660025318,
+      "grad_norm": 0.6834734678268433,
+      "learning_rate": 6.471191782027754e-06,
+      "loss": 0.7519,
+      "step": 14774
+    },
+    {
+      "epoch": 0.8131983048048874,
+      "grad_norm": 0.679432213306427,
+      "learning_rate": 6.470777498552561e-06,
+      "loss": 0.7707,
+      "step": 14775
+    },
+    {
+      "epoch": 0.8132533436072431,
+      "grad_norm": 0.6929466128349304,
+      "learning_rate": 6.4703632040234e-06,
+      "loss": 0.7166,
+      "step": 14776
+    },
+    {
+      "epoch": 0.8133083824095988,
+      "grad_norm": 0.7033447623252869,
+      "learning_rate": 6.469948898443381e-06,
+      "loss": 0.7558,
+      "step": 14777
+    },
+    {
+      "epoch": 0.8133634212119544,
+      "grad_norm": 0.89338618516922,
+      "learning_rate": 6.469534581815621e-06,
+      "loss": 0.7829,
+      "step": 14778
+    },
+    {
+      "epoch": 0.8134184600143101,
+      "grad_norm": 0.7361789345741272,
+      "learning_rate": 6.469120254143233e-06,
+      "loss": 0.7885,
+      "step": 14779
+    },
+    {
+      "epoch": 0.8134734988166658,
+      "grad_norm": 0.7532172203063965,
+      "learning_rate": 6.468705915429329e-06,
+      "loss": 0.7791,
+      "step": 14780
+    },
+    {
+      "epoch": 0.8135285376190214,
+      "grad_norm": 0.7082527279853821,
+      "learning_rate": 6.468291565677025e-06,
+      "loss": 0.7809,
+      "step": 14781
+    },
+    {
+      "epoch": 0.813583576421377,
+      "grad_norm": 0.7854330539703369,
+      "learning_rate": 6.467877204889435e-06,
+      "loss": 0.8467,
+      "step": 14782
+    },
+    {
+      "epoch": 0.8136386152237327,
+      "grad_norm": 0.7649636268615723,
+      "learning_rate": 6.467462833069672e-06,
+      "loss": 0.7766,
+      "step": 14783
+    },
+    {
+      "epoch": 0.8136936540260884,
+      "grad_norm": 0.6293399930000305,
+      "learning_rate": 6.467048450220852e-06,
+      "loss": 0.7307,
+      "step": 14784
+    },
+    {
+      "epoch": 0.8137486928284441,
+      "grad_norm": 0.7131813764572144,
+      "learning_rate": 6.4666340563460874e-06,
+      "loss": 0.7614,
+      "step": 14785
+    },
+    {
+      "epoch": 0.8138037316307997,
+      "grad_norm": 0.6650925874710083,
+      "learning_rate": 6.466219651448496e-06,
+      "loss": 0.7576,
+      "step": 14786
+    },
+    {
+      "epoch": 0.8138587704331554,
+      "grad_norm": 0.8009011745452881,
+      "learning_rate": 6.4658052355311875e-06,
+      "loss": 0.7127,
+      "step": 14787
+    },
+    {
+      "epoch": 0.813913809235511,
+      "grad_norm": 1.009027123451233,
+      "learning_rate": 6.465390808597281e-06,
+      "loss": 0.7647,
+      "step": 14788
+    },
+    {
+      "epoch": 0.8139688480378667,
+      "grad_norm": 0.7495583891868591,
+      "learning_rate": 6.464976370649888e-06,
+      "loss": 0.7276,
+      "step": 14789
+    },
+    {
+      "epoch": 0.8140238868402223,
+      "grad_norm": 0.7181064486503601,
+      "learning_rate": 6.464561921692125e-06,
+      "loss": 0.687,
+      "step": 14790
+    },
+    {
+      "epoch": 0.814078925642578,
+      "grad_norm": 0.7480552196502686,
+      "learning_rate": 6.464147461727108e-06,
+      "loss": 0.7813,
+      "step": 14791
+    },
+    {
+      "epoch": 0.8141339644449337,
+      "grad_norm": 0.6699607968330383,
+      "learning_rate": 6.4637329907579506e-06,
+      "loss": 0.7364,
+      "step": 14792
+    },
+    {
+      "epoch": 0.8141890032472894,
+      "grad_norm": 0.7321322560310364,
+      "learning_rate": 6.463318508787767e-06,
+      "loss": 0.6799,
+      "step": 14793
+    },
+    {
+      "epoch": 0.814244042049645,
+      "grad_norm": 0.8992179036140442,
+      "learning_rate": 6.462904015819673e-06,
+      "loss": 0.7602,
+      "step": 14794
+    },
+    {
+      "epoch": 0.8142990808520006,
+      "grad_norm": 0.6949485540390015,
+      "learning_rate": 6.462489511856784e-06,
+      "loss": 0.6701,
+      "step": 14795
+    },
+    {
+      "epoch": 0.8143541196543563,
+      "grad_norm": 0.6367032527923584,
+      "learning_rate": 6.462074996902217e-06,
+      "loss": 0.7132,
+      "step": 14796
+    },
+    {
+      "epoch": 0.814409158456712,
+      "grad_norm": 0.6424476504325867,
+      "learning_rate": 6.461660470959084e-06,
+      "loss": 0.7111,
+      "step": 14797
+    },
+    {
+      "epoch": 0.8144641972590676,
+      "grad_norm": 0.6649259924888611,
+      "learning_rate": 6.4612459340305025e-06,
+      "loss": 0.6583,
+      "step": 14798
+    },
+    {
+      "epoch": 0.8145192360614233,
+      "grad_norm": 0.7781171798706055,
+      "learning_rate": 6.460831386119587e-06,
+      "loss": 0.8145,
+      "step": 14799
+    },
+    {
+      "epoch": 0.814574274863779,
+      "grad_norm": 0.7409094572067261,
+      "learning_rate": 6.460416827229455e-06,
+      "loss": 0.7559,
+      "step": 14800
+    },
+    {
+      "epoch": 0.8146293136661347,
+      "grad_norm": 1.2152613401412964,
+      "learning_rate": 6.46000225736322e-06,
+      "loss": 0.8263,
+      "step": 14801
+    },
+    {
+      "epoch": 0.8146843524684902,
+      "grad_norm": 0.7133356332778931,
+      "learning_rate": 6.459587676524e-06,
+      "loss": 0.7687,
+      "step": 14802
+    },
+    {
+      "epoch": 0.8147393912708459,
+      "grad_norm": 0.8576061129570007,
+      "learning_rate": 6.459173084714908e-06,
+      "loss": 0.8364,
+      "step": 14803
+    },
+    {
+      "epoch": 0.8147944300732016,
+      "grad_norm": 0.7701650857925415,
+      "learning_rate": 6.4587584819390634e-06,
+      "loss": 0.7768,
+      "step": 14804
+    },
+    {
+      "epoch": 0.8148494688755573,
+      "grad_norm": 0.6629199981689453,
+      "learning_rate": 6.45834386819958e-06,
+      "loss": 0.7338,
+      "step": 14805
+    },
+    {
+      "epoch": 0.8149045076779129,
+      "grad_norm": 0.6498340964317322,
+      "learning_rate": 6.457929243499574e-06,
+      "loss": 0.7241,
+      "step": 14806
+    },
+    {
+      "epoch": 0.8149595464802686,
+      "grad_norm": 0.7107635140419006,
+      "learning_rate": 6.457514607842164e-06,
+      "loss": 0.7999,
+      "step": 14807
+    },
+    {
+      "epoch": 0.8150145852826243,
+      "grad_norm": 0.8689384460449219,
+      "learning_rate": 6.457099961230462e-06,
+      "loss": 0.7882,
+      "step": 14808
+    },
+    {
+      "epoch": 0.81506962408498,
+      "grad_norm": 0.7050377726554871,
+      "learning_rate": 6.456685303667587e-06,
+      "loss": 0.8039,
+      "step": 14809
+    },
+    {
+      "epoch": 0.8151246628873355,
+      "grad_norm": 0.6171709895133972,
+      "learning_rate": 6.456270635156656e-06,
+      "loss": 0.6569,
+      "step": 14810
+    },
+    {
+      "epoch": 0.8151797016896912,
+      "grad_norm": 0.837285041809082,
+      "learning_rate": 6.455855955700785e-06,
+      "loss": 0.6529,
+      "step": 14811
+    },
+    {
+      "epoch": 0.8152347404920469,
+      "grad_norm": 0.7335891723632812,
+      "learning_rate": 6.45544126530309e-06,
+      "loss": 0.814,
+      "step": 14812
+    },
+    {
+      "epoch": 0.8152897792944026,
+      "grad_norm": 0.7217129468917847,
+      "learning_rate": 6.4550265639666864e-06,
+      "loss": 0.795,
+      "step": 14813
+    },
+    {
+      "epoch": 0.8153448180967582,
+      "grad_norm": 0.7292104959487915,
+      "learning_rate": 6.454611851694694e-06,
+      "loss": 0.7169,
+      "step": 14814
+    },
+    {
+      "epoch": 0.8153998568991139,
+      "grad_norm": 0.7190173864364624,
+      "learning_rate": 6.454197128490229e-06,
+      "loss": 0.8413,
+      "step": 14815
+    },
+    {
+      "epoch": 0.8154548957014696,
+      "grad_norm": 0.6679649949073792,
+      "learning_rate": 6.453782394356407e-06,
+      "loss": 0.6626,
+      "step": 14816
+    },
+    {
+      "epoch": 0.8155099345038253,
+      "grad_norm": 0.6829885244369507,
+      "learning_rate": 6.453367649296347e-06,
+      "loss": 0.6512,
+      "step": 14817
+    },
+    {
+      "epoch": 0.8155649733061808,
+      "grad_norm": 0.659461498260498,
+      "learning_rate": 6.452952893313163e-06,
+      "loss": 0.7271,
+      "step": 14818
+    },
+    {
+      "epoch": 0.8156200121085365,
+      "grad_norm": 0.6737749576568604,
+      "learning_rate": 6.452538126409975e-06,
+      "loss": 0.6882,
+      "step": 14819
+    },
+    {
+      "epoch": 0.8156750509108922,
+      "grad_norm": 0.7798036336898804,
+      "learning_rate": 6.452123348589899e-06,
+      "loss": 0.7214,
+      "step": 14820
+    },
+    {
+      "epoch": 0.8157300897132478,
+      "grad_norm": 0.6594774127006531,
+      "learning_rate": 6.451708559856051e-06,
+      "loss": 0.7611,
+      "step": 14821
+    },
+    {
+      "epoch": 0.8157851285156035,
+      "grad_norm": 0.6795164942741394,
+      "learning_rate": 6.451293760211552e-06,
+      "loss": 0.6825,
+      "step": 14822
+    },
+    {
+      "epoch": 0.8158401673179592,
+      "grad_norm": 0.8376501798629761,
+      "learning_rate": 6.450878949659517e-06,
+      "loss": 0.7898,
+      "step": 14823
+    },
+    {
+      "epoch": 0.8158952061203149,
+      "grad_norm": 0.6746712923049927,
+      "learning_rate": 6.450464128203064e-06,
+      "loss": 0.6771,
+      "step": 14824
+    },
+    {
+      "epoch": 0.8159502449226704,
+      "grad_norm": 0.7984384894371033,
+      "learning_rate": 6.450049295845311e-06,
+      "loss": 0.7326,
+      "step": 14825
+    },
+    {
+      "epoch": 0.8160052837250261,
+      "grad_norm": 0.8210996389389038,
+      "learning_rate": 6.449634452589376e-06,
+      "loss": 0.8194,
+      "step": 14826
+    },
+    {
+      "epoch": 0.8160603225273818,
+      "grad_norm": 0.7045891284942627,
+      "learning_rate": 6.449219598438376e-06,
+      "loss": 0.7683,
+      "step": 14827
+    },
+    {
+      "epoch": 0.8161153613297375,
+      "grad_norm": 0.7199337482452393,
+      "learning_rate": 6.448804733395431e-06,
+      "loss": 0.7125,
+      "step": 14828
+    },
+    {
+      "epoch": 0.8161704001320931,
+      "grad_norm": 0.8576976656913757,
+      "learning_rate": 6.448389857463655e-06,
+      "loss": 0.6744,
+      "step": 14829
+    },
+    {
+      "epoch": 0.8162254389344488,
+      "grad_norm": 0.6944701075553894,
+      "learning_rate": 6.4479749706461705e-06,
+      "loss": 0.7663,
+      "step": 14830
+    },
+    {
+      "epoch": 0.8162804777368045,
+      "grad_norm": 0.7436455488204956,
+      "learning_rate": 6.447560072946093e-06,
+      "loss": 0.7612,
+      "step": 14831
+    },
+    {
+      "epoch": 0.8163355165391601,
+      "grad_norm": 0.6023590564727783,
+      "learning_rate": 6.447145164366542e-06,
+      "loss": 0.7029,
+      "step": 14832
+    },
+    {
+      "epoch": 0.8163905553415157,
+      "grad_norm": 0.6720685362815857,
+      "learning_rate": 6.446730244910633e-06,
+      "loss": 0.7821,
+      "step": 14833
+    },
+    {
+      "epoch": 0.8164455941438714,
+      "grad_norm": 0.6359856128692627,
+      "learning_rate": 6.446315314581488e-06,
+      "loss": 0.7119,
+      "step": 14834
+    },
+    {
+      "epoch": 0.8165006329462271,
+      "grad_norm": 0.6796891689300537,
+      "learning_rate": 6.445900373382225e-06,
+      "loss": 0.7414,
+      "step": 14835
+    },
+    {
+      "epoch": 0.8165556717485828,
+      "grad_norm": 0.6865763068199158,
+      "learning_rate": 6.445485421315963e-06,
+      "loss": 0.7239,
+      "step": 14836
+    },
+    {
+      "epoch": 0.8166107105509384,
+      "grad_norm": 0.6696601510047913,
+      "learning_rate": 6.445070458385816e-06,
+      "loss": 0.6322,
+      "step": 14837
+    },
+    {
+      "epoch": 0.8166657493532941,
+      "grad_norm": 0.6800506711006165,
+      "learning_rate": 6.444655484594909e-06,
+      "loss": 0.7827,
+      "step": 14838
+    },
+    {
+      "epoch": 0.8167207881556497,
+      "grad_norm": 0.7590689063072205,
+      "learning_rate": 6.444240499946357e-06,
+      "loss": 0.7177,
+      "step": 14839
+    },
+    {
+      "epoch": 0.8167758269580054,
+      "grad_norm": 0.6692266464233398,
+      "learning_rate": 6.4438255044432805e-06,
+      "loss": 0.6631,
+      "step": 14840
+    },
+    {
+      "epoch": 0.816830865760361,
+      "grad_norm": 0.695164144039154,
+      "learning_rate": 6.443410498088798e-06,
+      "loss": 0.6953,
+      "step": 14841
+    },
+    {
+      "epoch": 0.8168859045627167,
+      "grad_norm": 0.6503697037696838,
+      "learning_rate": 6.442995480886028e-06,
+      "loss": 0.7868,
+      "step": 14842
+    },
+    {
+      "epoch": 0.8169409433650724,
+      "grad_norm": 0.6943323016166687,
+      "learning_rate": 6.442580452838091e-06,
+      "loss": 0.7464,
+      "step": 14843
+    },
+    {
+      "epoch": 0.8169959821674281,
+      "grad_norm": 0.7510622143745422,
+      "learning_rate": 6.442165413948105e-06,
+      "loss": 0.7984,
+      "step": 14844
+    },
+    {
+      "epoch": 0.8170510209697837,
+      "grad_norm": 0.6322263479232788,
+      "learning_rate": 6.441750364219189e-06,
+      "loss": 0.7693,
+      "step": 14845
+    },
+    {
+      "epoch": 0.8171060597721393,
+      "grad_norm": 0.681967556476593,
+      "learning_rate": 6.4413353036544646e-06,
+      "loss": 0.6781,
+      "step": 14846
+    },
+    {
+      "epoch": 0.817161098574495,
+      "grad_norm": 0.6799043416976929,
+      "learning_rate": 6.440920232257049e-06,
+      "loss": 0.7791,
+      "step": 14847
+    },
+    {
+      "epoch": 0.8172161373768507,
+      "grad_norm": 0.673652172088623,
+      "learning_rate": 6.440505150030064e-06,
+      "loss": 0.7099,
+      "step": 14848
+    },
+    {
+      "epoch": 0.8172711761792063,
+      "grad_norm": 0.755377471446991,
+      "learning_rate": 6.4400900569766255e-06,
+      "loss": 0.7292,
+      "step": 14849
+    },
+    {
+      "epoch": 0.817326214981562,
+      "grad_norm": 0.6099830269813538,
+      "learning_rate": 6.439674953099857e-06,
+      "loss": 0.7154,
+      "step": 14850
+    },
+    {
+      "epoch": 0.8173812537839177,
+      "grad_norm": 0.6330500841140747,
+      "learning_rate": 6.439259838402878e-06,
+      "loss": 0.6858,
+      "step": 14851
+    },
+    {
+      "epoch": 0.8174362925862734,
+      "grad_norm": 0.6727203726768494,
+      "learning_rate": 6.438844712888806e-06,
+      "loss": 0.7089,
+      "step": 14852
+    },
+    {
+      "epoch": 0.817491331388629,
+      "grad_norm": 0.7482651472091675,
+      "learning_rate": 6.438429576560763e-06,
+      "loss": 0.7065,
+      "step": 14853
+    },
+    {
+      "epoch": 0.8175463701909846,
+      "grad_norm": 0.6786343455314636,
+      "learning_rate": 6.438014429421868e-06,
+      "loss": 0.7049,
+      "step": 14854
+    },
+    {
+      "epoch": 0.8176014089933403,
+      "grad_norm": 0.6155980825424194,
+      "learning_rate": 6.437599271475241e-06,
+      "loss": 0.607,
+      "step": 14855
+    },
+    {
+      "epoch": 0.817656447795696,
+      "grad_norm": 0.6551154851913452,
+      "learning_rate": 6.437184102724003e-06,
+      "loss": 0.7022,
+      "step": 14856
+    },
+    {
+      "epoch": 0.8177114865980516,
+      "grad_norm": 0.6127358078956604,
+      "learning_rate": 6.436768923171273e-06,
+      "loss": 0.6827,
+      "step": 14857
+    },
+    {
+      "epoch": 0.8177665254004073,
+      "grad_norm": 0.6470245718955994,
+      "learning_rate": 6.436353732820175e-06,
+      "loss": 0.6877,
+      "step": 14858
+    },
+    {
+      "epoch": 0.817821564202763,
+      "grad_norm": 0.704667866230011,
+      "learning_rate": 6.435938531673825e-06,
+      "loss": 0.7223,
+      "step": 14859
+    },
+    {
+      "epoch": 0.8178766030051187,
+      "grad_norm": 0.6328873634338379,
+      "learning_rate": 6.435523319735345e-06,
+      "loss": 0.7181,
+      "step": 14860
+    },
+    {
+      "epoch": 0.8179316418074742,
+      "grad_norm": 0.6489065885543823,
+      "learning_rate": 6.435108097007856e-06,
+      "loss": 0.7597,
+      "step": 14861
+    },
+    {
+      "epoch": 0.8179866806098299,
+      "grad_norm": 0.6398639678955078,
+      "learning_rate": 6.43469286349448e-06,
+      "loss": 0.667,
+      "step": 14862
+    },
+    {
+      "epoch": 0.8180417194121856,
+      "grad_norm": 0.7615578770637512,
+      "learning_rate": 6.434277619198335e-06,
+      "loss": 0.8474,
+      "step": 14863
+    },
+    {
+      "epoch": 0.8180967582145412,
+      "grad_norm": 0.8604047894477844,
+      "learning_rate": 6.433862364122545e-06,
+      "loss": 0.7977,
+      "step": 14864
+    },
+    {
+      "epoch": 0.8181517970168969,
+      "grad_norm": 0.6157855987548828,
+      "learning_rate": 6.433447098270228e-06,
+      "loss": 0.6513,
+      "step": 14865
+    },
+    {
+      "epoch": 0.8182068358192526,
+      "grad_norm": 0.7052211761474609,
+      "learning_rate": 6.433031821644507e-06,
+      "loss": 0.7043,
+      "step": 14866
+    },
+    {
+      "epoch": 0.8182618746216083,
+      "grad_norm": 0.785987138748169,
+      "learning_rate": 6.432616534248503e-06,
+      "loss": 0.8722,
+      "step": 14867
+    },
+    {
+      "epoch": 0.8183169134239638,
+      "grad_norm": 0.7711461782455444,
+      "learning_rate": 6.432201236085336e-06,
+      "loss": 0.68,
+      "step": 14868
+    },
+    {
+      "epoch": 0.8183719522263195,
+      "grad_norm": 0.6299784183502197,
+      "learning_rate": 6.431785927158126e-06,
+      "loss": 0.7397,
+      "step": 14869
+    },
+    {
+      "epoch": 0.8184269910286752,
+      "grad_norm": 0.6292238235473633,
+      "learning_rate": 6.431370607469998e-06,
+      "loss": 0.7392,
+      "step": 14870
+    },
+    {
+      "epoch": 0.8184820298310309,
+      "grad_norm": 0.8696228861808777,
+      "learning_rate": 6.430955277024071e-06,
+      "loss": 0.884,
+      "step": 14871
+    },
+    {
+      "epoch": 0.8185370686333865,
+      "grad_norm": 0.6754364967346191,
+      "learning_rate": 6.430539935823469e-06,
+      "loss": 0.7122,
+      "step": 14872
+    },
+    {
+      "epoch": 0.8185921074357422,
+      "grad_norm": 0.6936547160148621,
+      "learning_rate": 6.4301245838713085e-06,
+      "loss": 0.7353,
+      "step": 14873
+    },
+    {
+      "epoch": 0.8186471462380979,
+      "grad_norm": 0.8840705156326294,
+      "learning_rate": 6.429709221170717e-06,
+      "loss": 0.7043,
+      "step": 14874
+    },
+    {
+      "epoch": 0.8187021850404536,
+      "grad_norm": 0.7349988222122192,
+      "learning_rate": 6.4292938477248135e-06,
+      "loss": 0.7861,
+      "step": 14875
+    },
+    {
+      "epoch": 0.8187572238428091,
+      "grad_norm": 0.697790801525116,
+      "learning_rate": 6.428878463536721e-06,
+      "loss": 0.8021,
+      "step": 14876
+    },
+    {
+      "epoch": 0.8188122626451648,
+      "grad_norm": 0.7873979806900024,
+      "learning_rate": 6.428463068609559e-06,
+      "loss": 0.7313,
+      "step": 14877
+    },
+    {
+      "epoch": 0.8188673014475205,
+      "grad_norm": 0.6542018055915833,
+      "learning_rate": 6.4280476629464505e-06,
+      "loss": 0.7811,
+      "step": 14878
+    },
+    {
+      "epoch": 0.8189223402498762,
+      "grad_norm": 0.7477063536643982,
+      "learning_rate": 6.427632246550519e-06,
+      "loss": 0.764,
+      "step": 14879
+    },
+    {
+      "epoch": 0.8189773790522318,
+      "grad_norm": 0.6456438302993774,
+      "learning_rate": 6.4272168194248855e-06,
+      "loss": 0.7517,
+      "step": 14880
+    },
+    {
+      "epoch": 0.8190324178545875,
+      "grad_norm": 0.699684202671051,
+      "learning_rate": 6.426801381572671e-06,
+      "loss": 0.7963,
+      "step": 14881
+    },
+    {
+      "epoch": 0.8190874566569432,
+      "grad_norm": 0.9158867001533508,
+      "learning_rate": 6.426385932997001e-06,
+      "loss": 0.8782,
+      "step": 14882
+    },
+    {
+      "epoch": 0.8191424954592988,
+      "grad_norm": 0.5998190641403198,
+      "learning_rate": 6.425970473700995e-06,
+      "loss": 0.6598,
+      "step": 14883
+    },
+    {
+      "epoch": 0.8191975342616544,
+      "grad_norm": 0.6674730777740479,
+      "learning_rate": 6.4255550036877775e-06,
+      "loss": 0.7232,
+      "step": 14884
+    },
+    {
+      "epoch": 0.8192525730640101,
+      "grad_norm": 0.6303582191467285,
+      "learning_rate": 6.42513952296047e-06,
+      "loss": 0.7614,
+      "step": 14885
+    },
+    {
+      "epoch": 0.8193076118663658,
+      "grad_norm": 0.6255910992622375,
+      "learning_rate": 6.424724031522195e-06,
+      "loss": 0.7052,
+      "step": 14886
+    },
+    {
+      "epoch": 0.8193626506687215,
+      "grad_norm": 0.6610854268074036,
+      "learning_rate": 6.424308529376075e-06,
+      "loss": 0.7403,
+      "step": 14887
+    },
+    {
+      "epoch": 0.8194176894710771,
+      "grad_norm": 0.6758664846420288,
+      "learning_rate": 6.4238930165252355e-06,
+      "loss": 0.7603,
+      "step": 14888
+    },
+    {
+      "epoch": 0.8194727282734328,
+      "grad_norm": 0.6897797584533691,
+      "learning_rate": 6.423477492972796e-06,
+      "loss": 0.7194,
+      "step": 14889
+    },
+    {
+      "epoch": 0.8195277670757884,
+      "grad_norm": 0.7007622718811035,
+      "learning_rate": 6.42306195872188e-06,
+      "loss": 0.7905,
+      "step": 14890
+    },
+    {
+      "epoch": 0.8195828058781441,
+      "grad_norm": 0.7482092976570129,
+      "learning_rate": 6.422646413775613e-06,
+      "loss": 0.7809,
+      "step": 14891
+    },
+    {
+      "epoch": 0.8196378446804997,
+      "grad_norm": 0.9551613926887512,
+      "learning_rate": 6.422230858137115e-06,
+      "loss": 0.8559,
+      "step": 14892
+    },
+    {
+      "epoch": 0.8196928834828554,
+      "grad_norm": 0.6831939220428467,
+      "learning_rate": 6.42181529180951e-06,
+      "loss": 0.7867,
+      "step": 14893
+    },
+    {
+      "epoch": 0.8197479222852111,
+      "grad_norm": 1.446377158164978,
+      "learning_rate": 6.421399714795923e-06,
+      "loss": 0.8745,
+      "step": 14894
+    },
+    {
+      "epoch": 0.8198029610875668,
+      "grad_norm": 0.6738638877868652,
+      "learning_rate": 6.420984127099475e-06,
+      "loss": 0.727,
+      "step": 14895
+    },
+    {
+      "epoch": 0.8198579998899224,
+      "grad_norm": 0.7388872504234314,
+      "learning_rate": 6.420568528723292e-06,
+      "loss": 0.7041,
+      "step": 14896
+    },
+    {
+      "epoch": 0.819913038692278,
+      "grad_norm": 0.6977630853652954,
+      "learning_rate": 6.420152919670495e-06,
+      "loss": 0.7944,
+      "step": 14897
+    },
+    {
+      "epoch": 0.8199680774946337,
+      "grad_norm": 0.6300190091133118,
+      "learning_rate": 6.41973729994421e-06,
+      "loss": 0.6879,
+      "step": 14898
+    },
+    {
+      "epoch": 0.8200231162969894,
+      "grad_norm": 0.6350599527359009,
+      "learning_rate": 6.419321669547559e-06,
+      "loss": 0.6725,
+      "step": 14899
+    },
+    {
+      "epoch": 0.820078155099345,
+      "grad_norm": 0.8604453206062317,
+      "learning_rate": 6.418906028483667e-06,
+      "loss": 0.7706,
+      "step": 14900
+    },
+    {
+      "epoch": 0.8201331939017007,
+      "grad_norm": 0.6574103236198425,
+      "learning_rate": 6.418490376755656e-06,
+      "loss": 0.7008,
+      "step": 14901
+    },
+    {
+      "epoch": 0.8201882327040564,
+      "grad_norm": 0.706132173538208,
+      "learning_rate": 6.418074714366651e-06,
+      "loss": 0.7608,
+      "step": 14902
+    },
+    {
+      "epoch": 0.8202432715064121,
+      "grad_norm": 1.155480146408081,
+      "learning_rate": 6.417659041319777e-06,
+      "loss": 0.6893,
+      "step": 14903
+    },
+    {
+      "epoch": 0.8202983103087677,
+      "grad_norm": 0.8497835397720337,
+      "learning_rate": 6.417243357618157e-06,
+      "loss": 0.6889,
+      "step": 14904
+    },
+    {
+      "epoch": 0.8203533491111233,
+      "grad_norm": 0.9319966435432434,
+      "learning_rate": 6.416827663264915e-06,
+      "loss": 0.8098,
+      "step": 14905
+    },
+    {
+      "epoch": 0.820408387913479,
+      "grad_norm": 0.744888186454773,
+      "learning_rate": 6.4164119582631745e-06,
+      "loss": 0.7871,
+      "step": 14906
+    },
+    {
+      "epoch": 0.8204634267158346,
+      "grad_norm": 0.6928347945213318,
+      "learning_rate": 6.415996242616063e-06,
+      "loss": 0.7693,
+      "step": 14907
+    },
+    {
+      "epoch": 0.8205184655181903,
+      "grad_norm": 0.7455456852912903,
+      "learning_rate": 6.415580516326701e-06,
+      "loss": 0.6475,
+      "step": 14908
+    },
+    {
+      "epoch": 0.820573504320546,
+      "grad_norm": 0.6823583245277405,
+      "learning_rate": 6.415164779398215e-06,
+      "loss": 0.7223,
+      "step": 14909
+    },
+    {
+      "epoch": 0.8206285431229017,
+      "grad_norm": 0.6989970207214355,
+      "learning_rate": 6.414749031833729e-06,
+      "loss": 0.8203,
+      "step": 14910
+    },
+    {
+      "epoch": 0.8206835819252573,
+      "grad_norm": 0.6026825308799744,
+      "learning_rate": 6.414333273636369e-06,
+      "loss": 0.6307,
+      "step": 14911
+    },
+    {
+      "epoch": 0.8207386207276129,
+      "grad_norm": 0.6102367639541626,
+      "learning_rate": 6.413917504809258e-06,
+      "loss": 0.7049,
+      "step": 14912
+    },
+    {
+      "epoch": 0.8207936595299686,
+      "grad_norm": 0.6658119559288025,
+      "learning_rate": 6.4135017253555225e-06,
+      "loss": 0.7541,
+      "step": 14913
+    },
+    {
+      "epoch": 0.8208486983323243,
+      "grad_norm": 0.7272284626960754,
+      "learning_rate": 6.413085935278286e-06,
+      "loss": 0.7581,
+      "step": 14914
+    },
+    {
+      "epoch": 0.8209037371346799,
+      "grad_norm": 0.7826990485191345,
+      "learning_rate": 6.412670134580674e-06,
+      "loss": 0.8121,
+      "step": 14915
+    },
+    {
+      "epoch": 0.8209587759370356,
+      "grad_norm": 0.5845723748207092,
+      "learning_rate": 6.412254323265811e-06,
+      "loss": 0.5921,
+      "step": 14916
+    },
+    {
+      "epoch": 0.8210138147393913,
+      "grad_norm": 0.655577540397644,
+      "learning_rate": 6.411838501336823e-06,
+      "loss": 0.7694,
+      "step": 14917
+    },
+    {
+      "epoch": 0.821068853541747,
+      "grad_norm": 0.6722497940063477,
+      "learning_rate": 6.4114226687968325e-06,
+      "loss": 0.6377,
+      "step": 14918
+    },
+    {
+      "epoch": 0.8211238923441025,
+      "grad_norm": 0.713169276714325,
+      "learning_rate": 6.41100682564897e-06,
+      "loss": 0.7328,
+      "step": 14919
+    },
+    {
+      "epoch": 0.8211789311464582,
+      "grad_norm": 0.6004113554954529,
+      "learning_rate": 6.410590971896357e-06,
+      "loss": 0.6564,
+      "step": 14920
+    },
+    {
+      "epoch": 0.8212339699488139,
+      "grad_norm": 0.6541520953178406,
+      "learning_rate": 6.410175107542119e-06,
+      "loss": 0.7063,
+      "step": 14921
+    },
+    {
+      "epoch": 0.8212890087511696,
+      "grad_norm": 0.7937784194946289,
+      "learning_rate": 6.409759232589383e-06,
+      "loss": 0.7516,
+      "step": 14922
+    },
+    {
+      "epoch": 0.8213440475535252,
+      "grad_norm": 0.7017408013343811,
+      "learning_rate": 6.409343347041274e-06,
+      "loss": 0.6846,
+      "step": 14923
+    },
+    {
+      "epoch": 0.8213990863558809,
+      "grad_norm": 0.6233413815498352,
+      "learning_rate": 6.408927450900917e-06,
+      "loss": 0.6655,
+      "step": 14924
+    },
+    {
+      "epoch": 0.8214541251582366,
+      "grad_norm": 0.93160480260849,
+      "learning_rate": 6.4085115441714396e-06,
+      "loss": 0.7461,
+      "step": 14925
+    },
+    {
+      "epoch": 0.8215091639605923,
+      "grad_norm": 0.6075658202171326,
+      "learning_rate": 6.4080956268559655e-06,
+      "loss": 0.705,
+      "step": 14926
+    },
+    {
+      "epoch": 0.8215642027629478,
+      "grad_norm": 0.6212051510810852,
+      "learning_rate": 6.407679698957623e-06,
+      "loss": 0.6943,
+      "step": 14927
+    },
+    {
+      "epoch": 0.8216192415653035,
+      "grad_norm": 0.8143971562385559,
+      "learning_rate": 6.407263760479536e-06,
+      "loss": 0.6918,
+      "step": 14928
+    },
+    {
+      "epoch": 0.8216742803676592,
+      "grad_norm": 0.6851963996887207,
+      "learning_rate": 6.406847811424831e-06,
+      "loss": 0.7849,
+      "step": 14929
+    },
+    {
+      "epoch": 0.8217293191700149,
+      "grad_norm": 0.7047909498214722,
+      "learning_rate": 6.406431851796633e-06,
+      "loss": 0.7364,
+      "step": 14930
+    },
+    {
+      "epoch": 0.8217843579723705,
+      "grad_norm": 0.7377674579620361,
+      "learning_rate": 6.406015881598071e-06,
+      "loss": 0.7413,
+      "step": 14931
+    },
+    {
+      "epoch": 0.8218393967747262,
+      "grad_norm": 0.7188243269920349,
+      "learning_rate": 6.405599900832271e-06,
+      "loss": 0.8051,
+      "step": 14932
+    },
+    {
+      "epoch": 0.8218944355770819,
+      "grad_norm": 0.7588842511177063,
+      "learning_rate": 6.4051839095023575e-06,
+      "loss": 0.7687,
+      "step": 14933
+    },
+    {
+      "epoch": 0.8219494743794376,
+      "grad_norm": 0.6396436095237732,
+      "learning_rate": 6.404767907611457e-06,
+      "loss": 0.7516,
+      "step": 14934
+    },
+    {
+      "epoch": 0.8220045131817931,
+      "grad_norm": 0.6896073818206787,
+      "learning_rate": 6.404351895162698e-06,
+      "loss": 0.7904,
+      "step": 14935
+    },
+    {
+      "epoch": 0.8220595519841488,
+      "grad_norm": 0.7475640773773193,
+      "learning_rate": 6.403935872159206e-06,
+      "loss": 0.8325,
+      "step": 14936
+    },
+    {
+      "epoch": 0.8221145907865045,
+      "grad_norm": 0.6456442475318909,
+      "learning_rate": 6.403519838604107e-06,
+      "loss": 0.7685,
+      "step": 14937
+    },
+    {
+      "epoch": 0.8221696295888602,
+      "grad_norm": 0.6446966528892517,
+      "learning_rate": 6.40310379450053e-06,
+      "loss": 0.731,
+      "step": 14938
+    },
+    {
+      "epoch": 0.8222246683912158,
+      "grad_norm": 0.7744176983833313,
+      "learning_rate": 6.4026877398515995e-06,
+      "loss": 0.7975,
+      "step": 14939
+    },
+    {
+      "epoch": 0.8222797071935715,
+      "grad_norm": 0.6441214680671692,
+      "learning_rate": 6.402271674660444e-06,
+      "loss": 0.7386,
+      "step": 14940
+    },
+    {
+      "epoch": 0.8223347459959272,
+      "grad_norm": 0.6788361072540283,
+      "learning_rate": 6.40185559893019e-06,
+      "loss": 0.7664,
+      "step": 14941
+    },
+    {
+      "epoch": 0.8223897847982828,
+      "grad_norm": 0.6565073132514954,
+      "learning_rate": 6.4014395126639624e-06,
+      "loss": 0.6716,
+      "step": 14942
+    },
+    {
+      "epoch": 0.8224448236006384,
+      "grad_norm": 0.6475300788879395,
+      "learning_rate": 6.401023415864893e-06,
+      "loss": 0.6887,
+      "step": 14943
+    },
+    {
+      "epoch": 0.8224998624029941,
+      "grad_norm": 0.7058338522911072,
+      "learning_rate": 6.400607308536107e-06,
+      "loss": 0.7248,
+      "step": 14944
+    },
+    {
+      "epoch": 0.8225549012053498,
+      "grad_norm": 0.7184485197067261,
+      "learning_rate": 6.4001911906807305e-06,
+      "loss": 0.693,
+      "step": 14945
+    },
+    {
+      "epoch": 0.8226099400077055,
+      "grad_norm": 0.6280504465103149,
+      "learning_rate": 6.399775062301891e-06,
+      "loss": 0.6776,
+      "step": 14946
+    },
+    {
+      "epoch": 0.8226649788100611,
+      "grad_norm": 0.6995168328285217,
+      "learning_rate": 6.399358923402716e-06,
+      "loss": 0.7536,
+      "step": 14947
+    },
+    {
+      "epoch": 0.8227200176124168,
+      "grad_norm": 0.7770118713378906,
+      "learning_rate": 6.398942773986337e-06,
+      "loss": 0.6966,
+      "step": 14948
+    },
+    {
+      "epoch": 0.8227750564147724,
+      "grad_norm": 0.6947488188743591,
+      "learning_rate": 6.398526614055876e-06,
+      "loss": 0.7317,
+      "step": 14949
+    },
+    {
+      "epoch": 0.822830095217128,
+      "grad_norm": 0.7234527468681335,
+      "learning_rate": 6.3981104436144645e-06,
+      "loss": 0.6495,
+      "step": 14950
+    },
+    {
+      "epoch": 0.8228851340194837,
+      "grad_norm": 0.6872434020042419,
+      "learning_rate": 6.3976942626652295e-06,
+      "loss": 0.651,
+      "step": 14951
+    },
+    {
+      "epoch": 0.8229401728218394,
+      "grad_norm": 0.6762012243270874,
+      "learning_rate": 6.397278071211298e-06,
+      "loss": 0.7115,
+      "step": 14952
+    },
+    {
+      "epoch": 0.8229952116241951,
+      "grad_norm": 0.7007278800010681,
+      "learning_rate": 6.396861869255799e-06,
+      "loss": 0.717,
+      "step": 14953
+    },
+    {
+      "epoch": 0.8230502504265507,
+      "grad_norm": 0.7403082251548767,
+      "learning_rate": 6.396445656801859e-06,
+      "loss": 0.846,
+      "step": 14954
+    },
+    {
+      "epoch": 0.8231052892289064,
+      "grad_norm": 0.688758373260498,
+      "learning_rate": 6.396029433852609e-06,
+      "loss": 0.7871,
+      "step": 14955
+    },
+    {
+      "epoch": 0.823160328031262,
+      "grad_norm": 0.7264360189437866,
+      "learning_rate": 6.395613200411173e-06,
+      "loss": 0.7803,
+      "step": 14956
+    },
+    {
+      "epoch": 0.8232153668336177,
+      "grad_norm": 0.6858585476875305,
+      "learning_rate": 6.395196956480683e-06,
+      "loss": 0.6595,
+      "step": 14957
+    },
+    {
+      "epoch": 0.8232704056359733,
+      "grad_norm": 0.7834211587905884,
+      "learning_rate": 6.394780702064266e-06,
+      "loss": 0.7689,
+      "step": 14958
+    },
+    {
+      "epoch": 0.823325444438329,
+      "grad_norm": 0.6933274865150452,
+      "learning_rate": 6.394364437165052e-06,
+      "loss": 0.758,
+      "step": 14959
+    },
+    {
+      "epoch": 0.8233804832406847,
+      "grad_norm": 0.7490070462226868,
+      "learning_rate": 6.3939481617861664e-06,
+      "loss": 0.8106,
+      "step": 14960
+    },
+    {
+      "epoch": 0.8234355220430404,
+      "grad_norm": 0.5586501955986023,
+      "learning_rate": 6.3935318759307405e-06,
+      "loss": 0.6207,
+      "step": 14961
+    },
+    {
+      "epoch": 0.823490560845396,
+      "grad_norm": 0.6999693512916565,
+      "learning_rate": 6.393115579601902e-06,
+      "loss": 0.7787,
+      "step": 14962
+    },
+    {
+      "epoch": 0.8235455996477516,
+      "grad_norm": 1.0214177370071411,
+      "learning_rate": 6.392699272802779e-06,
+      "loss": 0.6444,
+      "step": 14963
+    },
+    {
+      "epoch": 0.8236006384501073,
+      "grad_norm": 0.7808836698532104,
+      "learning_rate": 6.392282955536502e-06,
+      "loss": 0.7537,
+      "step": 14964
+    },
+    {
+      "epoch": 0.823655677252463,
+      "grad_norm": 0.6825253963470459,
+      "learning_rate": 6.391866627806198e-06,
+      "loss": 0.7346,
+      "step": 14965
+    },
+    {
+      "epoch": 0.8237107160548186,
+      "grad_norm": 0.6105558276176453,
+      "learning_rate": 6.391450289614998e-06,
+      "loss": 0.6631,
+      "step": 14966
+    },
+    {
+      "epoch": 0.8237657548571743,
+      "grad_norm": 0.721986711025238,
+      "learning_rate": 6.391033940966029e-06,
+      "loss": 0.8638,
+      "step": 14967
+    },
+    {
+      "epoch": 0.82382079365953,
+      "grad_norm": 0.6226428747177124,
+      "learning_rate": 6.390617581862421e-06,
+      "loss": 0.7291,
+      "step": 14968
+    },
+    {
+      "epoch": 0.8238758324618857,
+      "grad_norm": 0.7403777241706848,
+      "learning_rate": 6.390201212307305e-06,
+      "loss": 0.7417,
+      "step": 14969
+    },
+    {
+      "epoch": 0.8239308712642412,
+      "grad_norm": 0.7188371419906616,
+      "learning_rate": 6.389784832303808e-06,
+      "loss": 0.757,
+      "step": 14970
+    },
+    {
+      "epoch": 0.8239859100665969,
+      "grad_norm": 0.8741163611412048,
+      "learning_rate": 6.389368441855061e-06,
+      "loss": 0.7264,
+      "step": 14971
+    },
+    {
+      "epoch": 0.8240409488689526,
+      "grad_norm": 0.7092788219451904,
+      "learning_rate": 6.388952040964192e-06,
+      "loss": 0.731,
+      "step": 14972
+    },
+    {
+      "epoch": 0.8240959876713083,
+      "grad_norm": 0.9291765689849854,
+      "learning_rate": 6.388535629634331e-06,
+      "loss": 0.7964,
+      "step": 14973
+    },
+    {
+      "epoch": 0.8241510264736639,
+      "grad_norm": 0.6140535473823547,
+      "learning_rate": 6.388119207868608e-06,
+      "loss": 0.7099,
+      "step": 14974
+    },
+    {
+      "epoch": 0.8242060652760196,
+      "grad_norm": 0.654778778553009,
+      "learning_rate": 6.387702775670154e-06,
+      "loss": 0.6667,
+      "step": 14975
+    },
+    {
+      "epoch": 0.8242611040783753,
+      "grad_norm": 0.7221185564994812,
+      "learning_rate": 6.387286333042095e-06,
+      "loss": 0.7533,
+      "step": 14976
+    },
+    {
+      "epoch": 0.824316142880731,
+      "grad_norm": 0.6680133938789368,
+      "learning_rate": 6.386869879987565e-06,
+      "loss": 0.6404,
+      "step": 14977
+    },
+    {
+      "epoch": 0.8243711816830865,
+      "grad_norm": 0.7067292928695679,
+      "learning_rate": 6.386453416509691e-06,
+      "loss": 0.8493,
+      "step": 14978
+    },
+    {
+      "epoch": 0.8244262204854422,
+      "grad_norm": 0.6279785633087158,
+      "learning_rate": 6.386036942611605e-06,
+      "loss": 0.7465,
+      "step": 14979
+    },
+    {
+      "epoch": 0.8244812592877979,
+      "grad_norm": 0.7184332013130188,
+      "learning_rate": 6.385620458296438e-06,
+      "loss": 0.738,
+      "step": 14980
+    },
+    {
+      "epoch": 0.8245362980901536,
+      "grad_norm": 0.7318315505981445,
+      "learning_rate": 6.385203963567316e-06,
+      "loss": 0.7409,
+      "step": 14981
+    },
+    {
+      "epoch": 0.8245913368925092,
+      "grad_norm": 0.6848355531692505,
+      "learning_rate": 6.384787458427372e-06,
+      "loss": 0.7343,
+      "step": 14982
+    },
+    {
+      "epoch": 0.8246463756948649,
+      "grad_norm": 0.7097738981246948,
+      "learning_rate": 6.384370942879736e-06,
+      "loss": 0.817,
+      "step": 14983
+    },
+    {
+      "epoch": 0.8247014144972206,
+      "grad_norm": 0.6933857798576355,
+      "learning_rate": 6.38395441692754e-06,
+      "loss": 0.7356,
+      "step": 14984
+    },
+    {
+      "epoch": 0.8247564532995763,
+      "grad_norm": 0.6631865501403809,
+      "learning_rate": 6.383537880573913e-06,
+      "loss": 0.752,
+      "step": 14985
+    },
+    {
+      "epoch": 0.8248114921019318,
+      "grad_norm": 0.6564633846282959,
+      "learning_rate": 6.3831213338219855e-06,
+      "loss": 0.7755,
+      "step": 14986
+    },
+    {
+      "epoch": 0.8248665309042875,
+      "grad_norm": 0.6518037915229797,
+      "learning_rate": 6.382704776674887e-06,
+      "loss": 0.7185,
+      "step": 14987
+    },
+    {
+      "epoch": 0.8249215697066432,
+      "grad_norm": 0.7074370384216309,
+      "learning_rate": 6.382288209135752e-06,
+      "loss": 0.7632,
+      "step": 14988
+    },
+    {
+      "epoch": 0.8249766085089989,
+      "grad_norm": 0.7034205198287964,
+      "learning_rate": 6.381871631207707e-06,
+      "loss": 0.8234,
+      "step": 14989
+    },
+    {
+      "epoch": 0.8250316473113545,
+      "grad_norm": 0.7635502815246582,
+      "learning_rate": 6.381455042893884e-06,
+      "loss": 0.7847,
+      "step": 14990
+    },
+    {
+      "epoch": 0.8250866861137102,
+      "grad_norm": 0.7682950496673584,
+      "learning_rate": 6.381038444197416e-06,
+      "loss": 0.6815,
+      "step": 14991
+    },
+    {
+      "epoch": 0.8251417249160659,
+      "grad_norm": 0.7713856101036072,
+      "learning_rate": 6.380621835121432e-06,
+      "loss": 0.7437,
+      "step": 14992
+    },
+    {
+      "epoch": 0.8251967637184214,
+      "grad_norm": 0.7955800294876099,
+      "learning_rate": 6.380205215669064e-06,
+      "loss": 0.876,
+      "step": 14993
+    },
+    {
+      "epoch": 0.8252518025207771,
+      "grad_norm": 0.6979825496673584,
+      "learning_rate": 6.379788585843443e-06,
+      "loss": 0.7018,
+      "step": 14994
+    },
+    {
+      "epoch": 0.8253068413231328,
+      "grad_norm": 0.6413466930389404,
+      "learning_rate": 6.379371945647701e-06,
+      "loss": 0.7345,
+      "step": 14995
+    },
+    {
+      "epoch": 0.8253618801254885,
+      "grad_norm": 0.6284430027008057,
+      "learning_rate": 6.378955295084968e-06,
+      "loss": 0.6758,
+      "step": 14996
+    },
+    {
+      "epoch": 0.8254169189278441,
+      "grad_norm": 0.5943842530250549,
+      "learning_rate": 6.378538634158377e-06,
+      "loss": 0.6572,
+      "step": 14997
+    },
+    {
+      "epoch": 0.8254719577301998,
+      "grad_norm": 0.7123218774795532,
+      "learning_rate": 6.378121962871058e-06,
+      "loss": 0.6993,
+      "step": 14998
+    },
+    {
+      "epoch": 0.8255269965325555,
+      "grad_norm": 0.6608574390411377,
+      "learning_rate": 6.377705281226143e-06,
+      "loss": 0.7802,
+      "step": 14999
+    },
+    {
+      "epoch": 0.8255820353349111,
+      "grad_norm": 0.6387534141540527,
+      "learning_rate": 6.377288589226764e-06,
+      "loss": 0.6572,
+      "step": 15000
+    },
+    {
+      "epoch": 0.8256370741372667,
+      "grad_norm": 0.6593596935272217,
+      "learning_rate": 6.376871886876054e-06,
+      "loss": 0.665,
+      "step": 15001
+    },
+    {
+      "epoch": 0.8256921129396224,
+      "grad_norm": 0.7146610617637634,
+      "learning_rate": 6.376455174177141e-06,
+      "loss": 0.7278,
+      "step": 15002
+    },
+    {
+      "epoch": 0.8257471517419781,
+      "grad_norm": 0.6776326298713684,
+      "learning_rate": 6.376038451133161e-06,
+      "loss": 0.7679,
+      "step": 15003
+    },
+    {
+      "epoch": 0.8258021905443338,
+      "grad_norm": 0.7008724808692932,
+      "learning_rate": 6.375621717747244e-06,
+      "loss": 0.8749,
+      "step": 15004
+    },
+    {
+      "epoch": 0.8258572293466894,
+      "grad_norm": 0.6809947490692139,
+      "learning_rate": 6.375204974022522e-06,
+      "loss": 0.7248,
+      "step": 15005
+    },
+    {
+      "epoch": 0.8259122681490451,
+      "grad_norm": 0.6921886205673218,
+      "learning_rate": 6.374788219962127e-06,
+      "loss": 0.6685,
+      "step": 15006
+    },
+    {
+      "epoch": 0.8259673069514007,
+      "grad_norm": 0.6471500396728516,
+      "learning_rate": 6.374371455569192e-06,
+      "loss": 0.6856,
+      "step": 15007
+    },
+    {
+      "epoch": 0.8260223457537564,
+      "grad_norm": 0.673425555229187,
+      "learning_rate": 6.373954680846851e-06,
+      "loss": 0.7006,
+      "step": 15008
+    },
+    {
+      "epoch": 0.826077384556112,
+      "grad_norm": 0.710217297077179,
+      "learning_rate": 6.373537895798233e-06,
+      "loss": 0.7315,
+      "step": 15009
+    },
+    {
+      "epoch": 0.8261324233584677,
+      "grad_norm": 0.692030668258667,
+      "learning_rate": 6.3731211004264725e-06,
+      "loss": 0.6534,
+      "step": 15010
+    },
+    {
+      "epoch": 0.8261874621608234,
+      "grad_norm": 0.6370778679847717,
+      "learning_rate": 6.372704294734701e-06,
+      "loss": 0.7278,
+      "step": 15011
+    },
+    {
+      "epoch": 0.8262425009631791,
+      "grad_norm": 0.6571012139320374,
+      "learning_rate": 6.372287478726052e-06,
+      "loss": 0.6889,
+      "step": 15012
+    },
+    {
+      "epoch": 0.8262975397655347,
+      "grad_norm": 0.721810519695282,
+      "learning_rate": 6.371870652403657e-06,
+      "loss": 0.8572,
+      "step": 15013
+    },
+    {
+      "epoch": 0.8263525785678904,
+      "grad_norm": 0.6751163601875305,
+      "learning_rate": 6.371453815770647e-06,
+      "loss": 0.7646,
+      "step": 15014
+    },
+    {
+      "epoch": 0.826407617370246,
+      "grad_norm": 0.724319338798523,
+      "learning_rate": 6.371036968830161e-06,
+      "loss": 0.8433,
+      "step": 15015
+    },
+    {
+      "epoch": 0.8264626561726017,
+      "grad_norm": 0.6961913108825684,
+      "learning_rate": 6.370620111585326e-06,
+      "loss": 0.7069,
+      "step": 15016
+    },
+    {
+      "epoch": 0.8265176949749573,
+      "grad_norm": 0.649428129196167,
+      "learning_rate": 6.370203244039279e-06,
+      "loss": 0.7286,
+      "step": 15017
+    },
+    {
+      "epoch": 0.826572733777313,
+      "grad_norm": 0.6468552947044373,
+      "learning_rate": 6.369786366195149e-06,
+      "loss": 0.7006,
+      "step": 15018
+    },
+    {
+      "epoch": 0.8266277725796687,
+      "grad_norm": 0.6564732789993286,
+      "learning_rate": 6.369369478056072e-06,
+      "loss": 0.727,
+      "step": 15019
+    },
+    {
+      "epoch": 0.8266828113820244,
+      "grad_norm": 0.6573188900947571,
+      "learning_rate": 6.36895257962518e-06,
+      "loss": 0.6603,
+      "step": 15020
+    },
+    {
+      "epoch": 0.82673785018438,
+      "grad_norm": 0.747164785861969,
+      "learning_rate": 6.368535670905609e-06,
+      "loss": 0.7426,
+      "step": 15021
+    },
+    {
+      "epoch": 0.8267928889867356,
+      "grad_norm": 0.6366723775863647,
+      "learning_rate": 6.368118751900489e-06,
+      "loss": 0.6487,
+      "step": 15022
+    },
+    {
+      "epoch": 0.8268479277890913,
+      "grad_norm": 0.6517844200134277,
+      "learning_rate": 6.367701822612955e-06,
+      "loss": 0.7131,
+      "step": 15023
+    },
+    {
+      "epoch": 0.826902966591447,
+      "grad_norm": 0.774309515953064,
+      "learning_rate": 6.367284883046141e-06,
+      "loss": 0.7978,
+      "step": 15024
+    },
+    {
+      "epoch": 0.8269580053938026,
+      "grad_norm": 0.6302667856216431,
+      "learning_rate": 6.366867933203178e-06,
+      "loss": 0.7403,
+      "step": 15025
+    },
+    {
+      "epoch": 0.8270130441961583,
+      "grad_norm": 0.6881224513053894,
+      "learning_rate": 6.366450973087202e-06,
+      "loss": 0.7884,
+      "step": 15026
+    },
+    {
+      "epoch": 0.827068082998514,
+      "grad_norm": 0.6901270747184753,
+      "learning_rate": 6.366034002701346e-06,
+      "loss": 0.6596,
+      "step": 15027
+    },
+    {
+      "epoch": 0.8271231218008697,
+      "grad_norm": 0.7436091303825378,
+      "learning_rate": 6.365617022048745e-06,
+      "loss": 0.8141,
+      "step": 15028
+    },
+    {
+      "epoch": 0.8271781606032252,
+      "grad_norm": 0.6745834350585938,
+      "learning_rate": 6.365200031132531e-06,
+      "loss": 0.7738,
+      "step": 15029
+    },
+    {
+      "epoch": 0.8272331994055809,
+      "grad_norm": 0.6963297724723816,
+      "learning_rate": 6.364783029955839e-06,
+      "loss": 0.8649,
+      "step": 15030
+    },
+    {
+      "epoch": 0.8272882382079366,
+      "grad_norm": 0.6468135714530945,
+      "learning_rate": 6.364366018521803e-06,
+      "loss": 0.7403,
+      "step": 15031
+    },
+    {
+      "epoch": 0.8273432770102923,
+      "grad_norm": 0.6481515169143677,
+      "learning_rate": 6.363948996833559e-06,
+      "loss": 0.6268,
+      "step": 15032
+    },
+    {
+      "epoch": 0.8273983158126479,
+      "grad_norm": 0.6881366968154907,
+      "learning_rate": 6.3635319648942386e-06,
+      "loss": 0.6339,
+      "step": 15033
+    },
+    {
+      "epoch": 0.8274533546150036,
+      "grad_norm": 0.6858122944831848,
+      "learning_rate": 6.363114922706977e-06,
+      "loss": 0.7685,
+      "step": 15034
+    },
+    {
+      "epoch": 0.8275083934173593,
+      "grad_norm": 0.6630339026451111,
+      "learning_rate": 6.362697870274907e-06,
+      "loss": 0.7281,
+      "step": 15035
+    },
+    {
+      "epoch": 0.8275634322197148,
+      "grad_norm": 0.7198584079742432,
+      "learning_rate": 6.362280807601167e-06,
+      "loss": 0.7726,
+      "step": 15036
+    },
+    {
+      "epoch": 0.8276184710220705,
+      "grad_norm": 0.721622884273529,
+      "learning_rate": 6.361863734688888e-06,
+      "loss": 0.6471,
+      "step": 15037
+    },
+    {
+      "epoch": 0.8276735098244262,
+      "grad_norm": 0.6032352447509766,
+      "learning_rate": 6.3614466515412055e-06,
+      "loss": 0.6684,
+      "step": 15038
+    },
+    {
+      "epoch": 0.8277285486267819,
+      "grad_norm": 0.7568576335906982,
+      "learning_rate": 6.3610295581612535e-06,
+      "loss": 0.7089,
+      "step": 15039
+    },
+    {
+      "epoch": 0.8277835874291375,
+      "grad_norm": 0.7461723685264587,
+      "learning_rate": 6.360612454552168e-06,
+      "loss": 0.806,
+      "step": 15040
+    },
+    {
+      "epoch": 0.8278386262314932,
+      "grad_norm": 0.6606107354164124,
+      "learning_rate": 6.3601953407170855e-06,
+      "loss": 0.7276,
+      "step": 15041
+    },
+    {
+      "epoch": 0.8278936650338489,
+      "grad_norm": 0.7203792333602905,
+      "learning_rate": 6.3597782166591384e-06,
+      "loss": 0.844,
+      "step": 15042
+    },
+    {
+      "epoch": 0.8279487038362046,
+      "grad_norm": 0.7327194213867188,
+      "learning_rate": 6.35936108238146e-06,
+      "loss": 0.8289,
+      "step": 15043
+    },
+    {
+      "epoch": 0.8280037426385601,
+      "grad_norm": 0.6741734147071838,
+      "learning_rate": 6.358943937887189e-06,
+      "loss": 0.7022,
+      "step": 15044
+    },
+    {
+      "epoch": 0.8280587814409158,
+      "grad_norm": 0.795724630355835,
+      "learning_rate": 6.35852678317946e-06,
+      "loss": 0.7703,
+      "step": 15045
+    },
+    {
+      "epoch": 0.8281138202432715,
+      "grad_norm": 0.6476230621337891,
+      "learning_rate": 6.3581096182614055e-06,
+      "loss": 0.7471,
+      "step": 15046
+    },
+    {
+      "epoch": 0.8281688590456272,
+      "grad_norm": 0.658829391002655,
+      "learning_rate": 6.357692443136164e-06,
+      "loss": 0.7796,
+      "step": 15047
+    },
+    {
+      "epoch": 0.8282238978479828,
+      "grad_norm": 0.6755202412605286,
+      "learning_rate": 6.35727525780687e-06,
+      "loss": 0.8239,
+      "step": 15048
+    },
+    {
+      "epoch": 0.8282789366503385,
+      "grad_norm": 0.6518263220787048,
+      "learning_rate": 6.356858062276658e-06,
+      "loss": 0.7222,
+      "step": 15049
+    },
+    {
+      "epoch": 0.8283339754526942,
+      "grad_norm": 0.7006294131278992,
+      "learning_rate": 6.356440856548662e-06,
+      "loss": 0.7779,
+      "step": 15050
+    },
+    {
+      "epoch": 0.8283890142550498,
+      "grad_norm": 0.6771633625030518,
+      "learning_rate": 6.356023640626021e-06,
+      "loss": 0.7529,
+      "step": 15051
+    },
+    {
+      "epoch": 0.8284440530574054,
+      "grad_norm": 0.6893792152404785,
+      "learning_rate": 6.35560641451187e-06,
+      "loss": 0.834,
+      "step": 15052
+    },
+    {
+      "epoch": 0.8284990918597611,
+      "grad_norm": 0.7450309991836548,
+      "learning_rate": 6.355189178209343e-06,
+      "loss": 0.7017,
+      "step": 15053
+    },
+    {
+      "epoch": 0.8285541306621168,
+      "grad_norm": 0.7094436883926392,
+      "learning_rate": 6.3547719317215785e-06,
+      "loss": 0.7883,
+      "step": 15054
+    },
+    {
+      "epoch": 0.8286091694644725,
+      "grad_norm": 0.6926944255828857,
+      "learning_rate": 6.3543546750517085e-06,
+      "loss": 0.7309,
+      "step": 15055
+    },
+    {
+      "epoch": 0.8286642082668281,
+      "grad_norm": 0.7394436597824097,
+      "learning_rate": 6.3539374082028725e-06,
+      "loss": 0.8819,
+      "step": 15056
+    },
+    {
+      "epoch": 0.8287192470691838,
+      "grad_norm": 0.7663393616676331,
+      "learning_rate": 6.353520131178206e-06,
+      "loss": 0.7269,
+      "step": 15057
+    },
+    {
+      "epoch": 0.8287742858715395,
+      "grad_norm": 0.702627956867218,
+      "learning_rate": 6.353102843980844e-06,
+      "loss": 0.8205,
+      "step": 15058
+    },
+    {
+      "epoch": 0.8288293246738951,
+      "grad_norm": 0.6575393676757812,
+      "learning_rate": 6.352685546613924e-06,
+      "loss": 0.782,
+      "step": 15059
+    },
+    {
+      "epoch": 0.8288843634762507,
+      "grad_norm": 0.6844787001609802,
+      "learning_rate": 6.35226823908058e-06,
+      "loss": 0.7485,
+      "step": 15060
+    },
+    {
+      "epoch": 0.8289394022786064,
+      "grad_norm": 0.6018843054771423,
+      "learning_rate": 6.351850921383951e-06,
+      "loss": 0.6788,
+      "step": 15061
+    },
+    {
+      "epoch": 0.8289944410809621,
+      "grad_norm": 0.7418997883796692,
+      "learning_rate": 6.351433593527172e-06,
+      "loss": 0.6789,
+      "step": 15062
+    },
+    {
+      "epoch": 0.8290494798833178,
+      "grad_norm": 0.625535786151886,
+      "learning_rate": 6.351016255513379e-06,
+      "loss": 0.7405,
+      "step": 15063
+    },
+    {
+      "epoch": 0.8291045186856734,
+      "grad_norm": 0.678569495677948,
+      "learning_rate": 6.350598907345711e-06,
+      "loss": 0.7386,
+      "step": 15064
+    },
+    {
+      "epoch": 0.829159557488029,
+      "grad_norm": 0.8012919425964355,
+      "learning_rate": 6.350181549027302e-06,
+      "loss": 0.7703,
+      "step": 15065
+    },
+    {
+      "epoch": 0.8292145962903847,
+      "grad_norm": 0.6115431189537048,
+      "learning_rate": 6.3497641805612905e-06,
+      "loss": 0.7131,
+      "step": 15066
+    },
+    {
+      "epoch": 0.8292696350927404,
+      "grad_norm": 0.7392085194587708,
+      "learning_rate": 6.349346801950812e-06,
+      "loss": 0.7648,
+      "step": 15067
+    },
+    {
+      "epoch": 0.829324673895096,
+      "grad_norm": 0.597613513469696,
+      "learning_rate": 6.348929413199005e-06,
+      "loss": 0.6023,
+      "step": 15068
+    },
+    {
+      "epoch": 0.8293797126974517,
+      "grad_norm": 0.6418130397796631,
+      "learning_rate": 6.348512014309005e-06,
+      "loss": 0.7507,
+      "step": 15069
+    },
+    {
+      "epoch": 0.8294347514998074,
+      "grad_norm": 0.6351965665817261,
+      "learning_rate": 6.34809460528395e-06,
+      "loss": 0.722,
+      "step": 15070
+    },
+    {
+      "epoch": 0.8294897903021631,
+      "grad_norm": 0.6593570709228516,
+      "learning_rate": 6.347677186126977e-06,
+      "loss": 0.7032,
+      "step": 15071
+    },
+    {
+      "epoch": 0.8295448291045187,
+      "grad_norm": 0.8040562868118286,
+      "learning_rate": 6.3472597568412235e-06,
+      "loss": 0.6519,
+      "step": 15072
+    },
+    {
+      "epoch": 0.8295998679068743,
+      "grad_norm": 0.7043612599372864,
+      "learning_rate": 6.346842317429825e-06,
+      "loss": 0.7765,
+      "step": 15073
+    },
+    {
+      "epoch": 0.82965490670923,
+      "grad_norm": 0.6304612159729004,
+      "learning_rate": 6.346424867895922e-06,
+      "loss": 0.6763,
+      "step": 15074
+    },
+    {
+      "epoch": 0.8297099455115857,
+      "grad_norm": 0.6402591466903687,
+      "learning_rate": 6.346007408242647e-06,
+      "loss": 0.828,
+      "step": 15075
+    },
+    {
+      "epoch": 0.8297649843139413,
+      "grad_norm": 0.6908280849456787,
+      "learning_rate": 6.345589938473142e-06,
+      "loss": 0.855,
+      "step": 15076
+    },
+    {
+      "epoch": 0.829820023116297,
+      "grad_norm": 0.5829552412033081,
+      "learning_rate": 6.345172458590545e-06,
+      "loss": 0.6323,
+      "step": 15077
+    },
+    {
+      "epoch": 0.8298750619186527,
+      "grad_norm": 0.8221700191497803,
+      "learning_rate": 6.34475496859799e-06,
+      "loss": 0.7069,
+      "step": 15078
+    },
+    {
+      "epoch": 0.8299301007210083,
+      "grad_norm": 0.7065801024436951,
+      "learning_rate": 6.344337468498616e-06,
+      "loss": 0.692,
+      "step": 15079
+    },
+    {
+      "epoch": 0.829985139523364,
+      "grad_norm": 0.6199344396591187,
+      "learning_rate": 6.343919958295564e-06,
+      "loss": 0.682,
+      "step": 15080
+    },
+    {
+      "epoch": 0.8300401783257196,
+      "grad_norm": 0.8999378681182861,
+      "learning_rate": 6.343502437991968e-06,
+      "loss": 0.7924,
+      "step": 15081
+    },
+    {
+      "epoch": 0.8300952171280753,
+      "grad_norm": 0.639163076877594,
+      "learning_rate": 6.343084907590966e-06,
+      "loss": 0.6976,
+      "step": 15082
+    },
+    {
+      "epoch": 0.8301502559304309,
+      "grad_norm": 0.8266178965568542,
+      "learning_rate": 6.3426673670957e-06,
+      "loss": 0.6831,
+      "step": 15083
+    },
+    {
+      "epoch": 0.8302052947327866,
+      "grad_norm": 0.6245449781417847,
+      "learning_rate": 6.3422498165093034e-06,
+      "loss": 0.6917,
+      "step": 15084
+    },
+    {
+      "epoch": 0.8302603335351423,
+      "grad_norm": 0.7809823751449585,
+      "learning_rate": 6.341832255834918e-06,
+      "loss": 0.8424,
+      "step": 15085
+    },
+    {
+      "epoch": 0.830315372337498,
+      "grad_norm": 0.6803410053253174,
+      "learning_rate": 6.34141468507568e-06,
+      "loss": 0.8345,
+      "step": 15086
+    },
+    {
+      "epoch": 0.8303704111398535,
+      "grad_norm": 0.7445305585861206,
+      "learning_rate": 6.340997104234728e-06,
+      "loss": 0.8823,
+      "step": 15087
+    },
+    {
+      "epoch": 0.8304254499422092,
+      "grad_norm": 0.6992506384849548,
+      "learning_rate": 6.340579513315199e-06,
+      "loss": 0.7857,
+      "step": 15088
+    },
+    {
+      "epoch": 0.8304804887445649,
+      "grad_norm": 0.7050431966781616,
+      "learning_rate": 6.340161912320237e-06,
+      "loss": 0.7988,
+      "step": 15089
+    },
+    {
+      "epoch": 0.8305355275469206,
+      "grad_norm": 0.8718838095664978,
+      "learning_rate": 6.339744301252973e-06,
+      "loss": 0.9983,
+      "step": 15090
+    },
+    {
+      "epoch": 0.8305905663492762,
+      "grad_norm": 0.7317140698432922,
+      "learning_rate": 6.339326680116551e-06,
+      "loss": 0.6852,
+      "step": 15091
+    },
+    {
+      "epoch": 0.8306456051516319,
+      "grad_norm": 0.6975864768028259,
+      "learning_rate": 6.338909048914108e-06,
+      "loss": 0.7334,
+      "step": 15092
+    },
+    {
+      "epoch": 0.8307006439539876,
+      "grad_norm": 0.6615436673164368,
+      "learning_rate": 6.3384914076487834e-06,
+      "loss": 0.776,
+      "step": 15093
+    },
+    {
+      "epoch": 0.8307556827563433,
+      "grad_norm": 0.773273766040802,
+      "learning_rate": 6.338073756323717e-06,
+      "loss": 0.7868,
+      "step": 15094
+    },
+    {
+      "epoch": 0.8308107215586988,
+      "grad_norm": 0.6686182022094727,
+      "learning_rate": 6.337656094942045e-06,
+      "loss": 0.7487,
+      "step": 15095
+    },
+    {
+      "epoch": 0.8308657603610545,
+      "grad_norm": 0.8202255368232727,
+      "learning_rate": 6.337238423506909e-06,
+      "loss": 0.7748,
+      "step": 15096
+    },
+    {
+      "epoch": 0.8309207991634102,
+      "grad_norm": 0.6356936693191528,
+      "learning_rate": 6.336820742021445e-06,
+      "loss": 0.6539,
+      "step": 15097
+    },
+    {
+      "epoch": 0.8309758379657659,
+      "grad_norm": 0.6543401479721069,
+      "learning_rate": 6.3364030504887955e-06,
+      "loss": 0.7185,
+      "step": 15098
+    },
+    {
+      "epoch": 0.8310308767681215,
+      "grad_norm": 0.6499043107032776,
+      "learning_rate": 6.335985348912097e-06,
+      "loss": 0.7254,
+      "step": 15099
+    },
+    {
+      "epoch": 0.8310859155704772,
+      "grad_norm": 0.6983271241188049,
+      "learning_rate": 6.335567637294491e-06,
+      "loss": 0.784,
+      "step": 15100
+    },
+    {
+      "epoch": 0.8311409543728329,
+      "grad_norm": 0.7932507395744324,
+      "learning_rate": 6.335149915639117e-06,
+      "loss": 0.6708,
+      "step": 15101
+    },
+    {
+      "epoch": 0.8311959931751886,
+      "grad_norm": 0.6792518496513367,
+      "learning_rate": 6.334732183949112e-06,
+      "loss": 0.7365,
+      "step": 15102
+    },
+    {
+      "epoch": 0.8312510319775441,
+      "grad_norm": 0.6852229237556458,
+      "learning_rate": 6.334314442227618e-06,
+      "loss": 0.7283,
+      "step": 15103
+    },
+    {
+      "epoch": 0.8313060707798998,
+      "grad_norm": 0.6528468728065491,
+      "learning_rate": 6.333896690477774e-06,
+      "loss": 0.763,
+      "step": 15104
+    },
+    {
+      "epoch": 0.8313611095822555,
+      "grad_norm": 0.7215067148208618,
+      "learning_rate": 6.33347892870272e-06,
+      "loss": 0.7769,
+      "step": 15105
+    },
+    {
+      "epoch": 0.8314161483846112,
+      "grad_norm": 0.7171593308448792,
+      "learning_rate": 6.333061156905596e-06,
+      "loss": 0.6807,
+      "step": 15106
+    },
+    {
+      "epoch": 0.8314711871869668,
+      "grad_norm": 0.6781407594680786,
+      "learning_rate": 6.332643375089539e-06,
+      "loss": 0.6801,
+      "step": 15107
+    },
+    {
+      "epoch": 0.8315262259893225,
+      "grad_norm": 0.803057849407196,
+      "learning_rate": 6.332225583257693e-06,
+      "loss": 0.682,
+      "step": 15108
+    },
+    {
+      "epoch": 0.8315812647916782,
+      "grad_norm": 0.6467291712760925,
+      "learning_rate": 6.331807781413195e-06,
+      "loss": 0.6675,
+      "step": 15109
+    },
+    {
+      "epoch": 0.8316363035940338,
+      "grad_norm": 0.7285529971122742,
+      "learning_rate": 6.331389969559186e-06,
+      "loss": 0.7333,
+      "step": 15110
+    },
+    {
+      "epoch": 0.8316913423963894,
+      "grad_norm": 0.6569895148277283,
+      "learning_rate": 6.330972147698806e-06,
+      "loss": 0.7202,
+      "step": 15111
+    },
+    {
+      "epoch": 0.8317463811987451,
+      "grad_norm": 0.7848708033561707,
+      "learning_rate": 6.330554315835198e-06,
+      "loss": 0.7936,
+      "step": 15112
+    },
+    {
+      "epoch": 0.8318014200011008,
+      "grad_norm": 0.6699723601341248,
+      "learning_rate": 6.330136473971498e-06,
+      "loss": 0.7107,
+      "step": 15113
+    },
+    {
+      "epoch": 0.8318564588034565,
+      "grad_norm": 0.7443183660507202,
+      "learning_rate": 6.329718622110848e-06,
+      "loss": 0.8102,
+      "step": 15114
+    },
+    {
+      "epoch": 0.8319114976058121,
+      "grad_norm": 0.6073893904685974,
+      "learning_rate": 6.329300760256389e-06,
+      "loss": 0.7061,
+      "step": 15115
+    },
+    {
+      "epoch": 0.8319665364081678,
+      "grad_norm": 0.6192148923873901,
+      "learning_rate": 6.328882888411262e-06,
+      "loss": 0.6929,
+      "step": 15116
+    },
+    {
+      "epoch": 0.8320215752105234,
+      "grad_norm": 0.7347237467765808,
+      "learning_rate": 6.3284650065786065e-06,
+      "loss": 0.6705,
+      "step": 15117
+    },
+    {
+      "epoch": 0.8320766140128791,
+      "grad_norm": 0.6286477446556091,
+      "learning_rate": 6.328047114761564e-06,
+      "loss": 0.6494,
+      "step": 15118
+    },
+    {
+      "epoch": 0.8321316528152347,
+      "grad_norm": 0.6492440104484558,
+      "learning_rate": 6.327629212963275e-06,
+      "loss": 0.6618,
+      "step": 15119
+    },
+    {
+      "epoch": 0.8321866916175904,
+      "grad_norm": 0.6295114755630493,
+      "learning_rate": 6.3272113011868804e-06,
+      "loss": 0.786,
+      "step": 15120
+    },
+    {
+      "epoch": 0.8322417304199461,
+      "grad_norm": 0.6737865805625916,
+      "learning_rate": 6.3267933794355206e-06,
+      "loss": 0.7544,
+      "step": 15121
+    },
+    {
+      "epoch": 0.8322967692223017,
+      "grad_norm": 0.8025132417678833,
+      "learning_rate": 6.3263754477123374e-06,
+      "loss": 0.7736,
+      "step": 15122
+    },
+    {
+      "epoch": 0.8323518080246574,
+      "grad_norm": 0.6820534467697144,
+      "learning_rate": 6.32595750602047e-06,
+      "loss": 0.6616,
+      "step": 15123
+    },
+    {
+      "epoch": 0.832406846827013,
+      "grad_norm": 0.7022573351860046,
+      "learning_rate": 6.325539554363061e-06,
+      "loss": 0.8175,
+      "step": 15124
+    },
+    {
+      "epoch": 0.8324618856293687,
+      "grad_norm": 0.7034926414489746,
+      "learning_rate": 6.325121592743253e-06,
+      "loss": 0.7047,
+      "step": 15125
+    },
+    {
+      "epoch": 0.8325169244317243,
+      "grad_norm": 0.654296875,
+      "learning_rate": 6.3247036211641856e-06,
+      "loss": 0.6468,
+      "step": 15126
+    },
+    {
+      "epoch": 0.83257196323408,
+      "grad_norm": 0.647859513759613,
+      "learning_rate": 6.324285639628999e-06,
+      "loss": 0.694,
+      "step": 15127
+    },
+    {
+      "epoch": 0.8326270020364357,
+      "grad_norm": 1.0824226140975952,
+      "learning_rate": 6.323867648140837e-06,
+      "loss": 0.7226,
+      "step": 15128
+    },
+    {
+      "epoch": 0.8326820408387914,
+      "grad_norm": 0.8568648099899292,
+      "learning_rate": 6.323449646702839e-06,
+      "loss": 0.7524,
+      "step": 15129
+    },
+    {
+      "epoch": 0.832737079641147,
+      "grad_norm": 0.6550299525260925,
+      "learning_rate": 6.32303163531815e-06,
+      "loss": 0.7294,
+      "step": 15130
+    },
+    {
+      "epoch": 0.8327921184435026,
+      "grad_norm": 0.7722175121307373,
+      "learning_rate": 6.3226136139899075e-06,
+      "loss": 0.7864,
+      "step": 15131
+    },
+    {
+      "epoch": 0.8328471572458583,
+      "grad_norm": 0.6542928218841553,
+      "learning_rate": 6.322195582721256e-06,
+      "loss": 0.6614,
+      "step": 15132
+    },
+    {
+      "epoch": 0.832902196048214,
+      "grad_norm": 0.6617493629455566,
+      "learning_rate": 6.321777541515337e-06,
+      "loss": 0.7147,
+      "step": 15133
+    },
+    {
+      "epoch": 0.8329572348505696,
+      "grad_norm": 0.698868989944458,
+      "learning_rate": 6.321359490375291e-06,
+      "loss": 0.6894,
+      "step": 15134
+    },
+    {
+      "epoch": 0.8330122736529253,
+      "grad_norm": 0.8005796074867249,
+      "learning_rate": 6.3209414293042595e-06,
+      "loss": 0.7513,
+      "step": 15135
+    },
+    {
+      "epoch": 0.833067312455281,
+      "grad_norm": 0.7656713128089905,
+      "learning_rate": 6.320523358305387e-06,
+      "loss": 0.7387,
+      "step": 15136
+    },
+    {
+      "epoch": 0.8331223512576367,
+      "grad_norm": 0.7299987077713013,
+      "learning_rate": 6.320105277381815e-06,
+      "loss": 0.7868,
+      "step": 15137
+    },
+    {
+      "epoch": 0.8331773900599923,
+      "grad_norm": 0.782574474811554,
+      "learning_rate": 6.319687186536685e-06,
+      "loss": 0.8307,
+      "step": 15138
+    },
+    {
+      "epoch": 0.8332324288623479,
+      "grad_norm": 0.6786854863166809,
+      "learning_rate": 6.319269085773138e-06,
+      "loss": 0.7819,
+      "step": 15139
+    },
+    {
+      "epoch": 0.8332874676647036,
+      "grad_norm": 1.173049807548523,
+      "learning_rate": 6.318850975094318e-06,
+      "loss": 0.7623,
+      "step": 15140
+    },
+    {
+      "epoch": 0.8333425064670593,
+      "grad_norm": 0.8410226106643677,
+      "learning_rate": 6.318432854503368e-06,
+      "loss": 0.812,
+      "step": 15141
+    },
+    {
+      "epoch": 0.8333975452694149,
+      "grad_norm": 0.8525705337524414,
+      "learning_rate": 6.3180147240034304e-06,
+      "loss": 0.7585,
+      "step": 15142
+    },
+    {
+      "epoch": 0.8334525840717706,
+      "grad_norm": 0.6345195770263672,
+      "learning_rate": 6.317596583597645e-06,
+      "loss": 0.7446,
+      "step": 15143
+    },
+    {
+      "epoch": 0.8335076228741263,
+      "grad_norm": 0.7238603234291077,
+      "learning_rate": 6.317178433289157e-06,
+      "loss": 0.7461,
+      "step": 15144
+    },
+    {
+      "epoch": 0.833562661676482,
+      "grad_norm": 0.6187044382095337,
+      "learning_rate": 6.31676027308111e-06,
+      "loss": 0.7195,
+      "step": 15145
+    },
+    {
+      "epoch": 0.8336177004788375,
+      "grad_norm": 0.6813417077064514,
+      "learning_rate": 6.316342102976644e-06,
+      "loss": 0.772,
+      "step": 15146
+    },
+    {
+      "epoch": 0.8336727392811932,
+      "grad_norm": 0.665515124797821,
+      "learning_rate": 6.315923922978902e-06,
+      "loss": 0.7127,
+      "step": 15147
+    },
+    {
+      "epoch": 0.8337277780835489,
+      "grad_norm": 0.8104628920555115,
+      "learning_rate": 6.315505733091028e-06,
+      "loss": 0.7332,
+      "step": 15148
+    },
+    {
+      "epoch": 0.8337828168859046,
+      "grad_norm": 0.8447679281234741,
+      "learning_rate": 6.315087533316166e-06,
+      "loss": 0.6803,
+      "step": 15149
+    },
+    {
+      "epoch": 0.8338378556882602,
+      "grad_norm": 0.7588180303573608,
+      "learning_rate": 6.31466932365746e-06,
+      "loss": 0.8301,
+      "step": 15150
+    },
+    {
+      "epoch": 0.8338928944906159,
+      "grad_norm": 0.7697302103042603,
+      "learning_rate": 6.314251104118048e-06,
+      "loss": 0.7777,
+      "step": 15151
+    },
+    {
+      "epoch": 0.8339479332929716,
+      "grad_norm": 0.8361233472824097,
+      "learning_rate": 6.313832874701078e-06,
+      "loss": 0.7585,
+      "step": 15152
+    },
+    {
+      "epoch": 0.8340029720953273,
+      "grad_norm": 0.6954757571220398,
+      "learning_rate": 6.313414635409692e-06,
+      "loss": 0.759,
+      "step": 15153
+    },
+    {
+      "epoch": 0.8340580108976828,
+      "grad_norm": 0.72389155626297,
+      "learning_rate": 6.312996386247034e-06,
+      "loss": 0.6679,
+      "step": 15154
+    },
+    {
+      "epoch": 0.8341130497000385,
+      "grad_norm": 0.781382143497467,
+      "learning_rate": 6.312578127216245e-06,
+      "loss": 0.769,
+      "step": 15155
+    },
+    {
+      "epoch": 0.8341680885023942,
+      "grad_norm": 0.7186244130134583,
+      "learning_rate": 6.312159858320472e-06,
+      "loss": 0.7476,
+      "step": 15156
+    },
+    {
+      "epoch": 0.8342231273047499,
+      "grad_norm": 0.6909130215644836,
+      "learning_rate": 6.311741579562855e-06,
+      "loss": 0.749,
+      "step": 15157
+    },
+    {
+      "epoch": 0.8342781661071055,
+      "grad_norm": 0.7692446708679199,
+      "learning_rate": 6.31132329094654e-06,
+      "loss": 0.7141,
+      "step": 15158
+    },
+    {
+      "epoch": 0.8343332049094612,
+      "grad_norm": 0.6753776669502258,
+      "learning_rate": 6.310904992474669e-06,
+      "loss": 0.7259,
+      "step": 15159
+    },
+    {
+      "epoch": 0.8343882437118169,
+      "grad_norm": 0.7118550539016724,
+      "learning_rate": 6.3104866841503885e-06,
+      "loss": 0.8282,
+      "step": 15160
+    },
+    {
+      "epoch": 0.8344432825141725,
+      "grad_norm": 0.6651625037193298,
+      "learning_rate": 6.31006836597684e-06,
+      "loss": 0.7639,
+      "step": 15161
+    },
+    {
+      "epoch": 0.8344983213165281,
+      "grad_norm": 0.6745681762695312,
+      "learning_rate": 6.30965003795717e-06,
+      "loss": 0.5922,
+      "step": 15162
+    },
+    {
+      "epoch": 0.8345533601188838,
+      "grad_norm": 0.7344138622283936,
+      "learning_rate": 6.309231700094518e-06,
+      "loss": 0.7134,
+      "step": 15163
+    },
+    {
+      "epoch": 0.8346083989212395,
+      "grad_norm": 0.7628228664398193,
+      "learning_rate": 6.308813352392034e-06,
+      "loss": 0.7341,
+      "step": 15164
+    },
+    {
+      "epoch": 0.8346634377235951,
+      "grad_norm": 0.6599448919296265,
+      "learning_rate": 6.308394994852858e-06,
+      "loss": 0.6821,
+      "step": 15165
+    },
+    {
+      "epoch": 0.8347184765259508,
+      "grad_norm": 0.9132193922996521,
+      "learning_rate": 6.307976627480136e-06,
+      "loss": 0.7862,
+      "step": 15166
+    },
+    {
+      "epoch": 0.8347735153283065,
+      "grad_norm": 0.752200722694397,
+      "learning_rate": 6.307558250277011e-06,
+      "loss": 0.7942,
+      "step": 15167
+    },
+    {
+      "epoch": 0.8348285541306621,
+      "grad_norm": 0.6848111748695374,
+      "learning_rate": 6.307139863246628e-06,
+      "loss": 0.8161,
+      "step": 15168
+    },
+    {
+      "epoch": 0.8348835929330177,
+      "grad_norm": 0.7229306697845459,
+      "learning_rate": 6.306721466392132e-06,
+      "loss": 0.684,
+      "step": 15169
+    },
+    {
+      "epoch": 0.8349386317353734,
+      "grad_norm": 0.7294610142707825,
+      "learning_rate": 6.306303059716667e-06,
+      "loss": 0.7046,
+      "step": 15170
+    },
+    {
+      "epoch": 0.8349936705377291,
+      "grad_norm": 0.7153074741363525,
+      "learning_rate": 6.305884643223378e-06,
+      "loss": 0.7613,
+      "step": 15171
+    },
+    {
+      "epoch": 0.8350487093400848,
+      "grad_norm": 0.6200907826423645,
+      "learning_rate": 6.30546621691541e-06,
+      "loss": 0.642,
+      "step": 15172
+    },
+    {
+      "epoch": 0.8351037481424404,
+      "grad_norm": 0.6640743017196655,
+      "learning_rate": 6.305047780795907e-06,
+      "loss": 0.7201,
+      "step": 15173
+    },
+    {
+      "epoch": 0.8351587869447961,
+      "grad_norm": 0.6427313089370728,
+      "learning_rate": 6.3046293348680144e-06,
+      "loss": 0.764,
+      "step": 15174
+    },
+    {
+      "epoch": 0.8352138257471518,
+      "grad_norm": 0.6475403308868408,
+      "learning_rate": 6.3042108791348755e-06,
+      "loss": 0.6678,
+      "step": 15175
+    },
+    {
+      "epoch": 0.8352688645495074,
+      "grad_norm": 0.6376405358314514,
+      "learning_rate": 6.303792413599638e-06,
+      "loss": 0.6972,
+      "step": 15176
+    },
+    {
+      "epoch": 0.835323903351863,
+      "grad_norm": 0.6648433804512024,
+      "learning_rate": 6.303373938265447e-06,
+      "loss": 0.6531,
+      "step": 15177
+    },
+    {
+      "epoch": 0.8353789421542187,
+      "grad_norm": 0.6582038402557373,
+      "learning_rate": 6.302955453135446e-06,
+      "loss": 0.7703,
+      "step": 15178
+    },
+    {
+      "epoch": 0.8354339809565744,
+      "grad_norm": 0.6386045217514038,
+      "learning_rate": 6.30253695821278e-06,
+      "loss": 0.6821,
+      "step": 15179
+    },
+    {
+      "epoch": 0.8354890197589301,
+      "grad_norm": 0.7268567681312561,
+      "learning_rate": 6.302118453500594e-06,
+      "loss": 0.7434,
+      "step": 15180
+    },
+    {
+      "epoch": 0.8355440585612857,
+      "grad_norm": 0.8008975982666016,
+      "learning_rate": 6.301699939002035e-06,
+      "loss": 0.8537,
+      "step": 15181
+    },
+    {
+      "epoch": 0.8355990973636414,
+      "grad_norm": 0.6803351044654846,
+      "learning_rate": 6.301281414720247e-06,
+      "loss": 0.6741,
+      "step": 15182
+    },
+    {
+      "epoch": 0.835654136165997,
+      "grad_norm": 0.6567045450210571,
+      "learning_rate": 6.3008628806583785e-06,
+      "loss": 0.7033,
+      "step": 15183
+    },
+    {
+      "epoch": 0.8357091749683527,
+      "grad_norm": 0.7088850140571594,
+      "learning_rate": 6.3004443368195685e-06,
+      "loss": 0.699,
+      "step": 15184
+    },
+    {
+      "epoch": 0.8357642137707083,
+      "grad_norm": 0.664929986000061,
+      "learning_rate": 6.3000257832069715e-06,
+      "loss": 0.6875,
+      "step": 15185
+    },
+    {
+      "epoch": 0.835819252573064,
+      "grad_norm": 0.7132309079170227,
+      "learning_rate": 6.299607219823727e-06,
+      "loss": 0.8172,
+      "step": 15186
+    },
+    {
+      "epoch": 0.8358742913754197,
+      "grad_norm": 0.7312454581260681,
+      "learning_rate": 6.2991886466729815e-06,
+      "loss": 0.7277,
+      "step": 15187
+    },
+    {
+      "epoch": 0.8359293301777754,
+      "grad_norm": 0.6576625108718872,
+      "learning_rate": 6.298770063757882e-06,
+      "loss": 0.7134,
+      "step": 15188
+    },
+    {
+      "epoch": 0.835984368980131,
+      "grad_norm": 0.6840282678604126,
+      "learning_rate": 6.2983514710815756e-06,
+      "loss": 0.777,
+      "step": 15189
+    },
+    {
+      "epoch": 0.8360394077824866,
+      "grad_norm": 0.7194011211395264,
+      "learning_rate": 6.297932868647207e-06,
+      "loss": 0.783,
+      "step": 15190
+    },
+    {
+      "epoch": 0.8360944465848423,
+      "grad_norm": 0.6619371175765991,
+      "learning_rate": 6.297514256457922e-06,
+      "loss": 0.7809,
+      "step": 15191
+    },
+    {
+      "epoch": 0.836149485387198,
+      "grad_norm": 0.8256712555885315,
+      "learning_rate": 6.2970956345168666e-06,
+      "loss": 0.9086,
+      "step": 15192
+    },
+    {
+      "epoch": 0.8362045241895536,
+      "grad_norm": 0.6951783299446106,
+      "learning_rate": 6.296677002827188e-06,
+      "loss": 0.7489,
+      "step": 15193
+    },
+    {
+      "epoch": 0.8362595629919093,
+      "grad_norm": 0.8535193204879761,
+      "learning_rate": 6.296258361392033e-06,
+      "loss": 0.7744,
+      "step": 15194
+    },
+    {
+      "epoch": 0.836314601794265,
+      "grad_norm": 0.7569966912269592,
+      "learning_rate": 6.295839710214546e-06,
+      "loss": 0.7091,
+      "step": 15195
+    },
+    {
+      "epoch": 0.8363696405966207,
+      "grad_norm": 0.6435930728912354,
+      "learning_rate": 6.295421049297875e-06,
+      "loss": 0.6601,
+      "step": 15196
+    },
+    {
+      "epoch": 0.8364246793989762,
+      "grad_norm": 0.811500608921051,
+      "learning_rate": 6.295002378645166e-06,
+      "loss": 0.7304,
+      "step": 15197
+    },
+    {
+      "epoch": 0.8364797182013319,
+      "grad_norm": 0.7306826114654541,
+      "learning_rate": 6.294583698259566e-06,
+      "loss": 0.8471,
+      "step": 15198
+    },
+    {
+      "epoch": 0.8365347570036876,
+      "grad_norm": 0.6411521434783936,
+      "learning_rate": 6.294165008144222e-06,
+      "loss": 0.6572,
+      "step": 15199
+    },
+    {
+      "epoch": 0.8365897958060433,
+      "grad_norm": 0.6460714340209961,
+      "learning_rate": 6.293746308302278e-06,
+      "loss": 0.7514,
+      "step": 15200
+    },
+    {
+      "epoch": 0.8366448346083989,
+      "grad_norm": 0.9355582594871521,
+      "learning_rate": 6.2933275987368855e-06,
+      "loss": 0.8171,
+      "step": 15201
+    },
+    {
+      "epoch": 0.8366998734107546,
+      "grad_norm": 0.6221946477890015,
+      "learning_rate": 6.292908879451189e-06,
+      "loss": 0.7323,
+      "step": 15202
+    },
+    {
+      "epoch": 0.8367549122131103,
+      "grad_norm": 0.6820993423461914,
+      "learning_rate": 6.292490150448335e-06,
+      "loss": 0.8168,
+      "step": 15203
+    },
+    {
+      "epoch": 0.836809951015466,
+      "grad_norm": 0.6494680643081665,
+      "learning_rate": 6.29207141173147e-06,
+      "loss": 0.7926,
+      "step": 15204
+    },
+    {
+      "epoch": 0.8368649898178215,
+      "grad_norm": 0.7658956050872803,
+      "learning_rate": 6.291652663303744e-06,
+      "loss": 0.7304,
+      "step": 15205
+    },
+    {
+      "epoch": 0.8369200286201772,
+      "grad_norm": 0.6653497219085693,
+      "learning_rate": 6.2912339051683e-06,
+      "loss": 0.7284,
+      "step": 15206
+    },
+    {
+      "epoch": 0.8369750674225329,
+      "grad_norm": 0.6136276721954346,
+      "learning_rate": 6.290815137328289e-06,
+      "loss": 0.7313,
+      "step": 15207
+    },
+    {
+      "epoch": 0.8370301062248885,
+      "grad_norm": 0.7542527914047241,
+      "learning_rate": 6.2903963597868555e-06,
+      "loss": 0.7806,
+      "step": 15208
+    },
+    {
+      "epoch": 0.8370851450272442,
+      "grad_norm": 0.6994839906692505,
+      "learning_rate": 6.2899775725471505e-06,
+      "loss": 0.8132,
+      "step": 15209
+    },
+    {
+      "epoch": 0.8371401838295999,
+      "grad_norm": 0.6558997631072998,
+      "learning_rate": 6.289558775612319e-06,
+      "loss": 0.7188,
+      "step": 15210
+    },
+    {
+      "epoch": 0.8371952226319556,
+      "grad_norm": 0.7155564427375793,
+      "learning_rate": 6.289139968985507e-06,
+      "loss": 0.6584,
+      "step": 15211
+    },
+    {
+      "epoch": 0.8372502614343111,
+      "grad_norm": 0.7645565867424011,
+      "learning_rate": 6.288721152669865e-06,
+      "loss": 0.761,
+      "step": 15212
+    },
+    {
+      "epoch": 0.8373053002366668,
+      "grad_norm": 0.6507940292358398,
+      "learning_rate": 6.288302326668542e-06,
+      "loss": 0.7139,
+      "step": 15213
+    },
+    {
+      "epoch": 0.8373603390390225,
+      "grad_norm": 0.7598558664321899,
+      "learning_rate": 6.287883490984682e-06,
+      "loss": 0.7627,
+      "step": 15214
+    },
+    {
+      "epoch": 0.8374153778413782,
+      "grad_norm": 0.6542350649833679,
+      "learning_rate": 6.287464645621434e-06,
+      "loss": 0.7508,
+      "step": 15215
+    },
+    {
+      "epoch": 0.8374704166437338,
+      "grad_norm": 0.7530503869056702,
+      "learning_rate": 6.287045790581946e-06,
+      "loss": 0.8234,
+      "step": 15216
+    },
+    {
+      "epoch": 0.8375254554460895,
+      "grad_norm": 0.9945759773254395,
+      "learning_rate": 6.286626925869367e-06,
+      "loss": 0.7637,
+      "step": 15217
+    },
+    {
+      "epoch": 0.8375804942484452,
+      "grad_norm": 0.6644982695579529,
+      "learning_rate": 6.286208051486844e-06,
+      "loss": 0.7671,
+      "step": 15218
+    },
+    {
+      "epoch": 0.8376355330508009,
+      "grad_norm": 0.8195061683654785,
+      "learning_rate": 6.285789167437526e-06,
+      "loss": 0.662,
+      "step": 15219
+    },
+    {
+      "epoch": 0.8376905718531564,
+      "grad_norm": 0.6578626036643982,
+      "learning_rate": 6.2853702737245605e-06,
+      "loss": 0.7681,
+      "step": 15220
+    },
+    {
+      "epoch": 0.8377456106555121,
+      "grad_norm": 0.6632179021835327,
+      "learning_rate": 6.2849513703510955e-06,
+      "loss": 0.759,
+      "step": 15221
+    },
+    {
+      "epoch": 0.8378006494578678,
+      "grad_norm": 0.6822313070297241,
+      "learning_rate": 6.284532457320282e-06,
+      "loss": 0.7859,
+      "step": 15222
+    },
+    {
+      "epoch": 0.8378556882602235,
+      "grad_norm": 0.6448203921318054,
+      "learning_rate": 6.284113534635265e-06,
+      "loss": 0.7224,
+      "step": 15223
+    },
+    {
+      "epoch": 0.8379107270625791,
+      "grad_norm": 0.6147580146789551,
+      "learning_rate": 6.2836946022991926e-06,
+      "loss": 0.7389,
+      "step": 15224
+    },
+    {
+      "epoch": 0.8379657658649348,
+      "grad_norm": 0.7476562857627869,
+      "learning_rate": 6.283275660315219e-06,
+      "loss": 0.7535,
+      "step": 15225
+    },
+    {
+      "epoch": 0.8380208046672905,
+      "grad_norm": 0.7396713495254517,
+      "learning_rate": 6.282856708686488e-06,
+      "loss": 0.7621,
+      "step": 15226
+    },
+    {
+      "epoch": 0.8380758434696461,
+      "grad_norm": 0.7220024466514587,
+      "learning_rate": 6.282437747416148e-06,
+      "loss": 0.672,
+      "step": 15227
+    },
+    {
+      "epoch": 0.8381308822720017,
+      "grad_norm": 0.9414284229278564,
+      "learning_rate": 6.2820187765073495e-06,
+      "loss": 0.8791,
+      "step": 15228
+    },
+    {
+      "epoch": 0.8381859210743574,
+      "grad_norm": 0.6074691414833069,
+      "learning_rate": 6.281599795963241e-06,
+      "loss": 0.6771,
+      "step": 15229
+    },
+    {
+      "epoch": 0.8382409598767131,
+      "grad_norm": 0.7367346286773682,
+      "learning_rate": 6.281180805786973e-06,
+      "loss": 0.7869,
+      "step": 15230
+    },
+    {
+      "epoch": 0.8382959986790688,
+      "grad_norm": 0.711016833782196,
+      "learning_rate": 6.280761805981691e-06,
+      "loss": 0.7166,
+      "step": 15231
+    },
+    {
+      "epoch": 0.8383510374814244,
+      "grad_norm": 0.6464707255363464,
+      "learning_rate": 6.280342796550546e-06,
+      "loss": 0.6965,
+      "step": 15232
+    },
+    {
+      "epoch": 0.83840607628378,
+      "grad_norm": 0.7385185956954956,
+      "learning_rate": 6.279923777496688e-06,
+      "loss": 0.7031,
+      "step": 15233
+    },
+    {
+      "epoch": 0.8384611150861357,
+      "grad_norm": 0.6799347996711731,
+      "learning_rate": 6.2795047488232665e-06,
+      "loss": 0.6777,
+      "step": 15234
+    },
+    {
+      "epoch": 0.8385161538884914,
+      "grad_norm": 0.690740168094635,
+      "learning_rate": 6.279085710533429e-06,
+      "loss": 0.7675,
+      "step": 15235
+    },
+    {
+      "epoch": 0.838571192690847,
+      "grad_norm": 0.9359111189842224,
+      "learning_rate": 6.278666662630325e-06,
+      "loss": 0.7063,
+      "step": 15236
+    },
+    {
+      "epoch": 0.8386262314932027,
+      "grad_norm": 0.751430094242096,
+      "learning_rate": 6.2782476051171075e-06,
+      "loss": 0.7851,
+      "step": 15237
+    },
+    {
+      "epoch": 0.8386812702955584,
+      "grad_norm": 0.6865997314453125,
+      "learning_rate": 6.27782853799692e-06,
+      "loss": 0.7347,
+      "step": 15238
+    },
+    {
+      "epoch": 0.8387363090979141,
+      "grad_norm": 0.6713284850120544,
+      "learning_rate": 6.277409461272916e-06,
+      "loss": 0.7651,
+      "step": 15239
+    },
+    {
+      "epoch": 0.8387913479002697,
+      "grad_norm": 0.7481899857521057,
+      "learning_rate": 6.276990374948244e-06,
+      "loss": 0.7681,
+      "step": 15240
+    },
+    {
+      "epoch": 0.8388463867026253,
+      "grad_norm": 0.7126002311706543,
+      "learning_rate": 6.2765712790260554e-06,
+      "loss": 0.7772,
+      "step": 15241
+    },
+    {
+      "epoch": 0.838901425504981,
+      "grad_norm": 0.6616978645324707,
+      "learning_rate": 6.276152173509497e-06,
+      "loss": 0.7028,
+      "step": 15242
+    },
+    {
+      "epoch": 0.8389564643073367,
+      "grad_norm": 0.9032973051071167,
+      "learning_rate": 6.2757330584017225e-06,
+      "loss": 0.7646,
+      "step": 15243
+    },
+    {
+      "epoch": 0.8390115031096923,
+      "grad_norm": 0.6345590353012085,
+      "learning_rate": 6.275313933705879e-06,
+      "loss": 0.6692,
+      "step": 15244
+    },
+    {
+      "epoch": 0.839066541912048,
+      "grad_norm": 0.6989019513130188,
+      "learning_rate": 6.2748947994251175e-06,
+      "loss": 0.6916,
+      "step": 15245
+    },
+    {
+      "epoch": 0.8391215807144037,
+      "grad_norm": 0.7115045189857483,
+      "learning_rate": 6.2744756555625875e-06,
+      "loss": 0.6923,
+      "step": 15246
+    },
+    {
+      "epoch": 0.8391766195167594,
+      "grad_norm": 0.6989235281944275,
+      "learning_rate": 6.2740565021214406e-06,
+      "loss": 0.7057,
+      "step": 15247
+    },
+    {
+      "epoch": 0.839231658319115,
+      "grad_norm": 0.684779942035675,
+      "learning_rate": 6.273637339104824e-06,
+      "loss": 0.7777,
+      "step": 15248
+    },
+    {
+      "epoch": 0.8392866971214706,
+      "grad_norm": 0.6341322064399719,
+      "learning_rate": 6.2732181665158934e-06,
+      "loss": 0.7335,
+      "step": 15249
+    },
+    {
+      "epoch": 0.8393417359238263,
+      "grad_norm": 0.7232723832130432,
+      "learning_rate": 6.272798984357793e-06,
+      "loss": 0.8055,
+      "step": 15250
+    },
+    {
+      "epoch": 0.8393967747261819,
+      "grad_norm": 0.9725174307823181,
+      "learning_rate": 6.272379792633678e-06,
+      "loss": 0.6221,
+      "step": 15251
+    },
+    {
+      "epoch": 0.8394518135285376,
+      "grad_norm": 0.6602086424827576,
+      "learning_rate": 6.271960591346695e-06,
+      "loss": 0.8023,
+      "step": 15252
+    },
+    {
+      "epoch": 0.8395068523308933,
+      "grad_norm": 0.7092040777206421,
+      "learning_rate": 6.271541380499998e-06,
+      "loss": 0.8135,
+      "step": 15253
+    },
+    {
+      "epoch": 0.839561891133249,
+      "grad_norm": 0.5656731724739075,
+      "learning_rate": 6.271122160096736e-06,
+      "loss": 0.647,
+      "step": 15254
+    },
+    {
+      "epoch": 0.8396169299356046,
+      "grad_norm": 1.1831625699996948,
+      "learning_rate": 6.270702930140061e-06,
+      "loss": 0.8513,
+      "step": 15255
+    },
+    {
+      "epoch": 0.8396719687379602,
+      "grad_norm": 0.6398816704750061,
+      "learning_rate": 6.270283690633121e-06,
+      "loss": 0.6988,
+      "step": 15256
+    },
+    {
+      "epoch": 0.8397270075403159,
+      "grad_norm": 0.6856167316436768,
+      "learning_rate": 6.26986444157907e-06,
+      "loss": 0.7789,
+      "step": 15257
+    },
+    {
+      "epoch": 0.8397820463426716,
+      "grad_norm": 0.7355605363845825,
+      "learning_rate": 6.269445182981058e-06,
+      "loss": 0.6652,
+      "step": 15258
+    },
+    {
+      "epoch": 0.8398370851450272,
+      "grad_norm": 0.6691173315048218,
+      "learning_rate": 6.2690259148422364e-06,
+      "loss": 0.6807,
+      "step": 15259
+    },
+    {
+      "epoch": 0.8398921239473829,
+      "grad_norm": 0.6596276164054871,
+      "learning_rate": 6.268606637165754e-06,
+      "loss": 0.6947,
+      "step": 15260
+    },
+    {
+      "epoch": 0.8399471627497386,
+      "grad_norm": 0.7198327779769897,
+      "learning_rate": 6.268187349954766e-06,
+      "loss": 0.7981,
+      "step": 15261
+    },
+    {
+      "epoch": 0.8400022015520943,
+      "grad_norm": 0.7006517648696899,
+      "learning_rate": 6.267768053212419e-06,
+      "loss": 0.7756,
+      "step": 15262
+    },
+    {
+      "epoch": 0.8400572403544498,
+      "grad_norm": 0.769062340259552,
+      "learning_rate": 6.267348746941869e-06,
+      "loss": 0.8433,
+      "step": 15263
+    },
+    {
+      "epoch": 0.8401122791568055,
+      "grad_norm": 0.6317951679229736,
+      "learning_rate": 6.266929431146263e-06,
+      "loss": 0.6575,
+      "step": 15264
+    },
+    {
+      "epoch": 0.8401673179591612,
+      "grad_norm": 0.7127153873443604,
+      "learning_rate": 6.2665101058287554e-06,
+      "loss": 0.7745,
+      "step": 15265
+    },
+    {
+      "epoch": 0.8402223567615169,
+      "grad_norm": 0.6909182667732239,
+      "learning_rate": 6.266090770992497e-06,
+      "loss": 0.7567,
+      "step": 15266
+    },
+    {
+      "epoch": 0.8402773955638725,
+      "grad_norm": 0.7875083684921265,
+      "learning_rate": 6.2656714266406384e-06,
+      "loss": 0.7392,
+      "step": 15267
+    },
+    {
+      "epoch": 0.8403324343662282,
+      "grad_norm": 0.7068803906440735,
+      "learning_rate": 6.2652520727763326e-06,
+      "loss": 0.6723,
+      "step": 15268
+    },
+    {
+      "epoch": 0.8403874731685839,
+      "grad_norm": 0.6994038820266724,
+      "learning_rate": 6.264832709402731e-06,
+      "loss": 0.6989,
+      "step": 15269
+    },
+    {
+      "epoch": 0.8404425119709396,
+      "grad_norm": 0.714044988155365,
+      "learning_rate": 6.264413336522985e-06,
+      "loss": 0.7464,
+      "step": 15270
+    },
+    {
+      "epoch": 0.8404975507732951,
+      "grad_norm": 0.8202210068702698,
+      "learning_rate": 6.263993954140249e-06,
+      "loss": 0.7174,
+      "step": 15271
+    },
+    {
+      "epoch": 0.8405525895756508,
+      "grad_norm": 0.6762316823005676,
+      "learning_rate": 6.2635745622576694e-06,
+      "loss": 0.7416,
+      "step": 15272
+    },
+    {
+      "epoch": 0.8406076283780065,
+      "grad_norm": 0.7461959719657898,
+      "learning_rate": 6.263155160878405e-06,
+      "loss": 0.7835,
+      "step": 15273
+    },
+    {
+      "epoch": 0.8406626671803622,
+      "grad_norm": 0.6263054609298706,
+      "learning_rate": 6.262735750005602e-06,
+      "loss": 0.7034,
+      "step": 15274
+    },
+    {
+      "epoch": 0.8407177059827178,
+      "grad_norm": 0.7489733695983887,
+      "learning_rate": 6.2623163296424165e-06,
+      "loss": 0.7387,
+      "step": 15275
+    },
+    {
+      "epoch": 0.8407727447850735,
+      "grad_norm": 0.7841430306434631,
+      "learning_rate": 6.261896899791997e-06,
+      "loss": 0.8487,
+      "step": 15276
+    },
+    {
+      "epoch": 0.8408277835874292,
+      "grad_norm": 0.8390078544616699,
+      "learning_rate": 6.2614774604575e-06,
+      "loss": 0.8335,
+      "step": 15277
+    },
+    {
+      "epoch": 0.8408828223897848,
+      "grad_norm": 0.9100946187973022,
+      "learning_rate": 6.261058011642076e-06,
+      "loss": 0.6196,
+      "step": 15278
+    },
+    {
+      "epoch": 0.8409378611921404,
+      "grad_norm": 0.7001772522926331,
+      "learning_rate": 6.260638553348879e-06,
+      "loss": 0.6935,
+      "step": 15279
+    },
+    {
+      "epoch": 0.8409928999944961,
+      "grad_norm": 0.7877102494239807,
+      "learning_rate": 6.260219085581057e-06,
+      "loss": 0.7378,
+      "step": 15280
+    },
+    {
+      "epoch": 0.8410479387968518,
+      "grad_norm": 0.687240719795227,
+      "learning_rate": 6.259799608341768e-06,
+      "loss": 0.7224,
+      "step": 15281
+    },
+    {
+      "epoch": 0.8411029775992075,
+      "grad_norm": 0.7766143083572388,
+      "learning_rate": 6.2593801216341625e-06,
+      "loss": 0.7157,
+      "step": 15282
+    },
+    {
+      "epoch": 0.8411580164015631,
+      "grad_norm": 1.1593633890151978,
+      "learning_rate": 6.258960625461391e-06,
+      "loss": 0.8555,
+      "step": 15283
+    },
+    {
+      "epoch": 0.8412130552039188,
+      "grad_norm": 0.6179451942443848,
+      "learning_rate": 6.2585411198266085e-06,
+      "loss": 0.6715,
+      "step": 15284
+    },
+    {
+      "epoch": 0.8412680940062744,
+      "grad_norm": 0.6755460500717163,
+      "learning_rate": 6.258121604732971e-06,
+      "loss": 0.7475,
+      "step": 15285
+    },
+    {
+      "epoch": 0.8413231328086301,
+      "grad_norm": 0.6775393486022949,
+      "learning_rate": 6.257702080183627e-06,
+      "loss": 0.6594,
+      "step": 15286
+    },
+    {
+      "epoch": 0.8413781716109857,
+      "grad_norm": 0.6972197890281677,
+      "learning_rate": 6.25728254618173e-06,
+      "loss": 0.7865,
+      "step": 15287
+    },
+    {
+      "epoch": 0.8414332104133414,
+      "grad_norm": 0.6446948051452637,
+      "learning_rate": 6.256863002730433e-06,
+      "loss": 0.6874,
+      "step": 15288
+    },
+    {
+      "epoch": 0.8414882492156971,
+      "grad_norm": 0.7012035846710205,
+      "learning_rate": 6.256443449832892e-06,
+      "loss": 0.7465,
+      "step": 15289
+    },
+    {
+      "epoch": 0.8415432880180528,
+      "grad_norm": 0.698693573474884,
+      "learning_rate": 6.256023887492257e-06,
+      "loss": 0.8206,
+      "step": 15290
+    },
+    {
+      "epoch": 0.8415983268204084,
+      "grad_norm": 0.7083185315132141,
+      "learning_rate": 6.255604315711684e-06,
+      "loss": 0.8306,
+      "step": 15291
+    },
+    {
+      "epoch": 0.841653365622764,
+      "grad_norm": 0.6605321764945984,
+      "learning_rate": 6.255184734494324e-06,
+      "loss": 0.6742,
+      "step": 15292
+    },
+    {
+      "epoch": 0.8417084044251197,
+      "grad_norm": 0.681881844997406,
+      "learning_rate": 6.254765143843331e-06,
+      "loss": 0.7009,
+      "step": 15293
+    },
+    {
+      "epoch": 0.8417634432274753,
+      "grad_norm": 0.6995699405670166,
+      "learning_rate": 6.2543455437618605e-06,
+      "loss": 0.8069,
+      "step": 15294
+    },
+    {
+      "epoch": 0.841818482029831,
+      "grad_norm": 0.7004442811012268,
+      "learning_rate": 6.2539259342530644e-06,
+      "loss": 0.71,
+      "step": 15295
+    },
+    {
+      "epoch": 0.8418735208321867,
+      "grad_norm": 0.7816279530525208,
+      "learning_rate": 6.253506315320097e-06,
+      "loss": 0.7833,
+      "step": 15296
+    },
+    {
+      "epoch": 0.8419285596345424,
+      "grad_norm": 0.6875490546226501,
+      "learning_rate": 6.25308668696611e-06,
+      "loss": 0.7223,
+      "step": 15297
+    },
+    {
+      "epoch": 0.841983598436898,
+      "grad_norm": 0.7126815915107727,
+      "learning_rate": 6.252667049194261e-06,
+      "loss": 0.7934,
+      "step": 15298
+    },
+    {
+      "epoch": 0.8420386372392537,
+      "grad_norm": 0.8048780560493469,
+      "learning_rate": 6.252247402007701e-06,
+      "loss": 0.7775,
+      "step": 15299
+    },
+    {
+      "epoch": 0.8420936760416093,
+      "grad_norm": 0.6681318879127502,
+      "learning_rate": 6.251827745409583e-06,
+      "loss": 0.6516,
+      "step": 15300
+    },
+    {
+      "epoch": 0.842148714843965,
+      "grad_norm": 0.6467457413673401,
+      "learning_rate": 6.251408079403064e-06,
+      "loss": 0.7417,
+      "step": 15301
+    },
+    {
+      "epoch": 0.8422037536463206,
+      "grad_norm": 0.6815666556358337,
+      "learning_rate": 6.250988403991297e-06,
+      "loss": 0.7498,
+      "step": 15302
+    },
+    {
+      "epoch": 0.8422587924486763,
+      "grad_norm": 0.6596205234527588,
+      "learning_rate": 6.250568719177437e-06,
+      "loss": 0.762,
+      "step": 15303
+    },
+    {
+      "epoch": 0.842313831251032,
+      "grad_norm": 0.7564731240272522,
+      "learning_rate": 6.250149024964635e-06,
+      "loss": 0.7592,
+      "step": 15304
+    },
+    {
+      "epoch": 0.8423688700533877,
+      "grad_norm": 0.6755058169364929,
+      "learning_rate": 6.249729321356048e-06,
+      "loss": 0.6953,
+      "step": 15305
+    },
+    {
+      "epoch": 0.8424239088557433,
+      "grad_norm": 0.7423762083053589,
+      "learning_rate": 6.249309608354832e-06,
+      "loss": 0.7018,
+      "step": 15306
+    },
+    {
+      "epoch": 0.8424789476580989,
+      "grad_norm": 0.727678120136261,
+      "learning_rate": 6.248889885964138e-06,
+      "loss": 0.8159,
+      "step": 15307
+    },
+    {
+      "epoch": 0.8425339864604546,
+      "grad_norm": 1.0823713541030884,
+      "learning_rate": 6.248470154187123e-06,
+      "loss": 0.872,
+      "step": 15308
+    },
+    {
+      "epoch": 0.8425890252628103,
+      "grad_norm": 0.6428259015083313,
+      "learning_rate": 6.248050413026939e-06,
+      "loss": 0.683,
+      "step": 15309
+    },
+    {
+      "epoch": 0.8426440640651659,
+      "grad_norm": 0.6622119545936584,
+      "learning_rate": 6.247630662486743e-06,
+      "loss": 0.7891,
+      "step": 15310
+    },
+    {
+      "epoch": 0.8426991028675216,
+      "grad_norm": 1.2377631664276123,
+      "learning_rate": 6.247210902569689e-06,
+      "loss": 0.7675,
+      "step": 15311
+    },
+    {
+      "epoch": 0.8427541416698773,
+      "grad_norm": 0.7909934520721436,
+      "learning_rate": 6.246791133278931e-06,
+      "loss": 0.8688,
+      "step": 15312
+    },
+    {
+      "epoch": 0.842809180472233,
+      "grad_norm": 0.6541300415992737,
+      "learning_rate": 6.246371354617625e-06,
+      "loss": 0.6754,
+      "step": 15313
+    },
+    {
+      "epoch": 0.8428642192745885,
+      "grad_norm": 0.6664960384368896,
+      "learning_rate": 6.245951566588926e-06,
+      "loss": 0.6666,
+      "step": 15314
+    },
+    {
+      "epoch": 0.8429192580769442,
+      "grad_norm": 0.7288552522659302,
+      "learning_rate": 6.245531769195988e-06,
+      "loss": 0.8179,
+      "step": 15315
+    },
+    {
+      "epoch": 0.8429742968792999,
+      "grad_norm": 0.7044054865837097,
+      "learning_rate": 6.245111962441966e-06,
+      "loss": 0.7306,
+      "step": 15316
+    },
+    {
+      "epoch": 0.8430293356816556,
+      "grad_norm": 0.6108603477478027,
+      "learning_rate": 6.244692146330016e-06,
+      "loss": 0.6213,
+      "step": 15317
+    },
+    {
+      "epoch": 0.8430843744840112,
+      "grad_norm": 0.6381129622459412,
+      "learning_rate": 6.2442723208632935e-06,
+      "loss": 0.7709,
+      "step": 15318
+    },
+    {
+      "epoch": 0.8431394132863669,
+      "grad_norm": 0.7355496883392334,
+      "learning_rate": 6.243852486044955e-06,
+      "loss": 0.665,
+      "step": 15319
+    },
+    {
+      "epoch": 0.8431944520887226,
+      "grad_norm": 0.7450826168060303,
+      "learning_rate": 6.2434326418781525e-06,
+      "loss": 0.7551,
+      "step": 15320
+    },
+    {
+      "epoch": 0.8432494908910783,
+      "grad_norm": 0.6463751792907715,
+      "learning_rate": 6.243012788366043e-06,
+      "loss": 0.7956,
+      "step": 15321
+    },
+    {
+      "epoch": 0.8433045296934338,
+      "grad_norm": 0.6673271059989929,
+      "learning_rate": 6.242592925511782e-06,
+      "loss": 0.7148,
+      "step": 15322
+    },
+    {
+      "epoch": 0.8433595684957895,
+      "grad_norm": 0.7663269639015198,
+      "learning_rate": 6.242173053318526e-06,
+      "loss": 0.8594,
+      "step": 15323
+    },
+    {
+      "epoch": 0.8434146072981452,
+      "grad_norm": 0.8503594994544983,
+      "learning_rate": 6.2417531717894285e-06,
+      "loss": 0.7594,
+      "step": 15324
+    },
+    {
+      "epoch": 0.8434696461005009,
+      "grad_norm": 0.6903344988822937,
+      "learning_rate": 6.241333280927647e-06,
+      "loss": 0.7252,
+      "step": 15325
+    },
+    {
+      "epoch": 0.8435246849028565,
+      "grad_norm": 0.6472830772399902,
+      "learning_rate": 6.240913380736337e-06,
+      "loss": 0.7379,
+      "step": 15326
+    },
+    {
+      "epoch": 0.8435797237052122,
+      "grad_norm": 0.6442959308624268,
+      "learning_rate": 6.240493471218655e-06,
+      "loss": 0.7447,
+      "step": 15327
+    },
+    {
+      "epoch": 0.8436347625075679,
+      "grad_norm": 0.6387843489646912,
+      "learning_rate": 6.240073552377756e-06,
+      "loss": 0.7659,
+      "step": 15328
+    },
+    {
+      "epoch": 0.8436898013099235,
+      "grad_norm": 0.7017341256141663,
+      "learning_rate": 6.239653624216794e-06,
+      "loss": 0.6934,
+      "step": 15329
+    },
+    {
+      "epoch": 0.8437448401122791,
+      "grad_norm": 0.6204355359077454,
+      "learning_rate": 6.2392336867389294e-06,
+      "loss": 0.6553,
+      "step": 15330
+    },
+    {
+      "epoch": 0.8437998789146348,
+      "grad_norm": 0.6765483021736145,
+      "learning_rate": 6.238813739947315e-06,
+      "loss": 0.7492,
+      "step": 15331
+    },
+    {
+      "epoch": 0.8438549177169905,
+      "grad_norm": 0.7261079549789429,
+      "learning_rate": 6.238393783845109e-06,
+      "loss": 0.7373,
+      "step": 15332
+    },
+    {
+      "epoch": 0.8439099565193462,
+      "grad_norm": 0.7019803524017334,
+      "learning_rate": 6.237973818435466e-06,
+      "loss": 0.7742,
+      "step": 15333
+    },
+    {
+      "epoch": 0.8439649953217018,
+      "grad_norm": 0.7521516680717468,
+      "learning_rate": 6.237553843721545e-06,
+      "loss": 0.8808,
+      "step": 15334
+    },
+    {
+      "epoch": 0.8440200341240575,
+      "grad_norm": 0.6796375513076782,
+      "learning_rate": 6.237133859706499e-06,
+      "loss": 0.7759,
+      "step": 15335
+    },
+    {
+      "epoch": 0.8440750729264132,
+      "grad_norm": 0.6199387311935425,
+      "learning_rate": 6.236713866393487e-06,
+      "loss": 0.6203,
+      "step": 15336
+    },
+    {
+      "epoch": 0.8441301117287687,
+      "grad_norm": 0.6968052983283997,
+      "learning_rate": 6.236293863785663e-06,
+      "loss": 0.7645,
+      "step": 15337
+    },
+    {
+      "epoch": 0.8441851505311244,
+      "grad_norm": 0.757556676864624,
+      "learning_rate": 6.235873851886186e-06,
+      "loss": 0.8005,
+      "step": 15338
+    },
+    {
+      "epoch": 0.8442401893334801,
+      "grad_norm": 0.6558085680007935,
+      "learning_rate": 6.235453830698211e-06,
+      "loss": 0.796,
+      "step": 15339
+    },
+    {
+      "epoch": 0.8442952281358358,
+      "grad_norm": 0.6963368654251099,
+      "learning_rate": 6.235033800224898e-06,
+      "loss": 0.7077,
+      "step": 15340
+    },
+    {
+      "epoch": 0.8443502669381914,
+      "grad_norm": 0.6057709455490112,
+      "learning_rate": 6.234613760469399e-06,
+      "loss": 0.5443,
+      "step": 15341
+    },
+    {
+      "epoch": 0.8444053057405471,
+      "grad_norm": 0.7616491317749023,
+      "learning_rate": 6.234193711434875e-06,
+      "loss": 0.6764,
+      "step": 15342
+    },
+    {
+      "epoch": 0.8444603445429028,
+      "grad_norm": 0.7143368721008301,
+      "learning_rate": 6.233773653124482e-06,
+      "loss": 0.6647,
+      "step": 15343
+    },
+    {
+      "epoch": 0.8445153833452584,
+      "grad_norm": 0.8766696453094482,
+      "learning_rate": 6.233353585541375e-06,
+      "loss": 0.7112,
+      "step": 15344
+    },
+    {
+      "epoch": 0.844570422147614,
+      "grad_norm": 0.6184048652648926,
+      "learning_rate": 6.232933508688714e-06,
+      "loss": 0.6645,
+      "step": 15345
+    },
+    {
+      "epoch": 0.8446254609499697,
+      "grad_norm": 0.8119208812713623,
+      "learning_rate": 6.232513422569655e-06,
+      "loss": 0.6729,
+      "step": 15346
+    },
+    {
+      "epoch": 0.8446804997523254,
+      "grad_norm": 0.5964543223381042,
+      "learning_rate": 6.2320933271873544e-06,
+      "loss": 0.6931,
+      "step": 15347
+    },
+    {
+      "epoch": 0.8447355385546811,
+      "grad_norm": 0.696611225605011,
+      "learning_rate": 6.23167322254497e-06,
+      "loss": 0.8292,
+      "step": 15348
+    },
+    {
+      "epoch": 0.8447905773570367,
+      "grad_norm": 0.6196489930152893,
+      "learning_rate": 6.231253108645658e-06,
+      "loss": 0.6651,
+      "step": 15349
+    },
+    {
+      "epoch": 0.8448456161593924,
+      "grad_norm": 0.6222663521766663,
+      "learning_rate": 6.230832985492579e-06,
+      "loss": 0.6513,
+      "step": 15350
+    },
+    {
+      "epoch": 0.844900654961748,
+      "grad_norm": 0.6424199342727661,
+      "learning_rate": 6.230412853088889e-06,
+      "loss": 0.7005,
+      "step": 15351
+    },
+    {
+      "epoch": 0.8449556937641037,
+      "grad_norm": 0.6484132409095764,
+      "learning_rate": 6.229992711437745e-06,
+      "loss": 0.6931,
+      "step": 15352
+    },
+    {
+      "epoch": 0.8450107325664593,
+      "grad_norm": 0.7568885684013367,
+      "learning_rate": 6.229572560542303e-06,
+      "loss": 0.7036,
+      "step": 15353
+    },
+    {
+      "epoch": 0.845065771368815,
+      "grad_norm": 0.665937602519989,
+      "learning_rate": 6.229152400405724e-06,
+      "loss": 0.5498,
+      "step": 15354
+    },
+    {
+      "epoch": 0.8451208101711707,
+      "grad_norm": 0.6861961483955383,
+      "learning_rate": 6.228732231031165e-06,
+      "loss": 0.7622,
+      "step": 15355
+    },
+    {
+      "epoch": 0.8451758489735264,
+      "grad_norm": 0.6793088316917419,
+      "learning_rate": 6.2283120524217845e-06,
+      "loss": 0.758,
+      "step": 15356
+    },
+    {
+      "epoch": 0.845230887775882,
+      "grad_norm": 0.7460890412330627,
+      "learning_rate": 6.227891864580739e-06,
+      "loss": 0.6618,
+      "step": 15357
+    },
+    {
+      "epoch": 0.8452859265782376,
+      "grad_norm": 0.6434195041656494,
+      "learning_rate": 6.227471667511186e-06,
+      "loss": 0.7226,
+      "step": 15358
+    },
+    {
+      "epoch": 0.8453409653805933,
+      "grad_norm": 0.7655256986618042,
+      "learning_rate": 6.227051461216285e-06,
+      "loss": 0.8461,
+      "step": 15359
+    },
+    {
+      "epoch": 0.845396004182949,
+      "grad_norm": 0.6727028489112854,
+      "learning_rate": 6.226631245699193e-06,
+      "loss": 0.6765,
+      "step": 15360
+    },
+    {
+      "epoch": 0.8454510429853046,
+      "grad_norm": 0.6030625700950623,
+      "learning_rate": 6.226211020963069e-06,
+      "loss": 0.6548,
+      "step": 15361
+    },
+    {
+      "epoch": 0.8455060817876603,
+      "grad_norm": 0.6430317163467407,
+      "learning_rate": 6.225790787011071e-06,
+      "loss": 0.7564,
+      "step": 15362
+    },
+    {
+      "epoch": 0.845561120590016,
+      "grad_norm": 0.633975088596344,
+      "learning_rate": 6.225370543846359e-06,
+      "loss": 0.716,
+      "step": 15363
+    },
+    {
+      "epoch": 0.8456161593923717,
+      "grad_norm": 0.6722174286842346,
+      "learning_rate": 6.2249502914720895e-06,
+      "loss": 0.7266,
+      "step": 15364
+    },
+    {
+      "epoch": 0.8456711981947272,
+      "grad_norm": 0.724166214466095,
+      "learning_rate": 6.22453002989142e-06,
+      "loss": 0.788,
+      "step": 15365
+    },
+    {
+      "epoch": 0.8457262369970829,
+      "grad_norm": 0.6406343579292297,
+      "learning_rate": 6.224109759107512e-06,
+      "loss": 0.8086,
+      "step": 15366
+    },
+    {
+      "epoch": 0.8457812757994386,
+      "grad_norm": 0.7344949245452881,
+      "learning_rate": 6.223689479123523e-06,
+      "loss": 0.7838,
+      "step": 15367
+    },
+    {
+      "epoch": 0.8458363146017943,
+      "grad_norm": 0.8572549819946289,
+      "learning_rate": 6.22326918994261e-06,
+      "loss": 0.7427,
+      "step": 15368
+    },
+    {
+      "epoch": 0.8458913534041499,
+      "grad_norm": 0.662644624710083,
+      "learning_rate": 6.222848891567934e-06,
+      "loss": 0.7165,
+      "step": 15369
+    },
+    {
+      "epoch": 0.8459463922065056,
+      "grad_norm": 0.7139797210693359,
+      "learning_rate": 6.222428584002654e-06,
+      "loss": 0.8218,
+      "step": 15370
+    },
+    {
+      "epoch": 0.8460014310088613,
+      "grad_norm": 0.6846550107002258,
+      "learning_rate": 6.222008267249927e-06,
+      "loss": 0.6686,
+      "step": 15371
+    },
+    {
+      "epoch": 0.846056469811217,
+      "grad_norm": 0.6675787568092346,
+      "learning_rate": 6.221587941312914e-06,
+      "loss": 0.7151,
+      "step": 15372
+    },
+    {
+      "epoch": 0.8461115086135725,
+      "grad_norm": 0.626371443271637,
+      "learning_rate": 6.221167606194771e-06,
+      "loss": 0.7637,
+      "step": 15373
+    },
+    {
+      "epoch": 0.8461665474159282,
+      "grad_norm": 0.6768763065338135,
+      "learning_rate": 6.220747261898661e-06,
+      "loss": 0.7363,
+      "step": 15374
+    },
+    {
+      "epoch": 0.8462215862182839,
+      "grad_norm": 0.7771314978599548,
+      "learning_rate": 6.220326908427741e-06,
+      "loss": 0.7032,
+      "step": 15375
+    },
+    {
+      "epoch": 0.8462766250206396,
+      "grad_norm": 0.8215247392654419,
+      "learning_rate": 6.219906545785171e-06,
+      "loss": 0.8917,
+      "step": 15376
+    },
+    {
+      "epoch": 0.8463316638229952,
+      "grad_norm": 0.7277588248252869,
+      "learning_rate": 6.219486173974107e-06,
+      "loss": 0.7531,
+      "step": 15377
+    },
+    {
+      "epoch": 0.8463867026253509,
+      "grad_norm": 0.6487376093864441,
+      "learning_rate": 6.219065792997714e-06,
+      "loss": 0.7182,
+      "step": 15378
+    },
+    {
+      "epoch": 0.8464417414277066,
+      "grad_norm": 0.6960493326187134,
+      "learning_rate": 6.218645402859148e-06,
+      "loss": 0.8125,
+      "step": 15379
+    },
+    {
+      "epoch": 0.8464967802300621,
+      "grad_norm": 0.7183159589767456,
+      "learning_rate": 6.218225003561571e-06,
+      "loss": 0.6536,
+      "step": 15380
+    },
+    {
+      "epoch": 0.8465518190324178,
+      "grad_norm": 0.7001940011978149,
+      "learning_rate": 6.217804595108139e-06,
+      "loss": 0.8203,
+      "step": 15381
+    },
+    {
+      "epoch": 0.8466068578347735,
+      "grad_norm": 0.5986705422401428,
+      "learning_rate": 6.217384177502015e-06,
+      "loss": 0.6672,
+      "step": 15382
+    },
+    {
+      "epoch": 0.8466618966371292,
+      "grad_norm": 0.6191138029098511,
+      "learning_rate": 6.216963750746356e-06,
+      "loss": 0.6565,
+      "step": 15383
+    },
+    {
+      "epoch": 0.8467169354394848,
+      "grad_norm": 1.2927004098892212,
+      "learning_rate": 6.216543314844326e-06,
+      "loss": 0.7511,
+      "step": 15384
+    },
+    {
+      "epoch": 0.8467719742418405,
+      "grad_norm": 0.6715198159217834,
+      "learning_rate": 6.2161228697990785e-06,
+      "loss": 0.7712,
+      "step": 15385
+    },
+    {
+      "epoch": 0.8468270130441962,
+      "grad_norm": 0.7516033053398132,
+      "learning_rate": 6.215702415613778e-06,
+      "loss": 0.6595,
+      "step": 15386
+    },
+    {
+      "epoch": 0.8468820518465519,
+      "grad_norm": 0.6913008689880371,
+      "learning_rate": 6.215281952291585e-06,
+      "loss": 0.7262,
+      "step": 15387
+    },
+    {
+      "epoch": 0.8469370906489074,
+      "grad_norm": 0.7288102507591248,
+      "learning_rate": 6.214861479835657e-06,
+      "loss": 0.6628,
+      "step": 15388
+    },
+    {
+      "epoch": 0.8469921294512631,
+      "grad_norm": 0.7889914512634277,
+      "learning_rate": 6.214440998249155e-06,
+      "loss": 0.7744,
+      "step": 15389
+    },
+    {
+      "epoch": 0.8470471682536188,
+      "grad_norm": 0.7622396945953369,
+      "learning_rate": 6.21402050753524e-06,
+      "loss": 0.7818,
+      "step": 15390
+    },
+    {
+      "epoch": 0.8471022070559745,
+      "grad_norm": 0.6172721982002258,
+      "learning_rate": 6.213600007697072e-06,
+      "loss": 0.626,
+      "step": 15391
+    },
+    {
+      "epoch": 0.8471572458583301,
+      "grad_norm": 0.710991621017456,
+      "learning_rate": 6.213179498737812e-06,
+      "loss": 0.7313,
+      "step": 15392
+    },
+    {
+      "epoch": 0.8472122846606858,
+      "grad_norm": 0.660139262676239,
+      "learning_rate": 6.2127589806606195e-06,
+      "loss": 0.6479,
+      "step": 15393
+    },
+    {
+      "epoch": 0.8472673234630415,
+      "grad_norm": 0.6611735224723816,
+      "learning_rate": 6.2123384534686534e-06,
+      "loss": 0.7091,
+      "step": 15394
+    },
+    {
+      "epoch": 0.8473223622653971,
+      "grad_norm": 0.8392653465270996,
+      "learning_rate": 6.211917917165078e-06,
+      "loss": 0.8514,
+      "step": 15395
+    },
+    {
+      "epoch": 0.8473774010677527,
+      "grad_norm": 0.6202608942985535,
+      "learning_rate": 6.211497371753052e-06,
+      "loss": 0.7068,
+      "step": 15396
+    },
+    {
+      "epoch": 0.8474324398701084,
+      "grad_norm": 0.6785926818847656,
+      "learning_rate": 6.211076817235734e-06,
+      "loss": 0.7216,
+      "step": 15397
+    },
+    {
+      "epoch": 0.8474874786724641,
+      "grad_norm": 0.7234075665473938,
+      "learning_rate": 6.210656253616288e-06,
+      "loss": 0.7379,
+      "step": 15398
+    },
+    {
+      "epoch": 0.8475425174748198,
+      "grad_norm": 0.6223714351654053,
+      "learning_rate": 6.210235680897874e-06,
+      "loss": 0.758,
+      "step": 15399
+    },
+    {
+      "epoch": 0.8475975562771754,
+      "grad_norm": 0.7993804812431335,
+      "learning_rate": 6.209815099083651e-06,
+      "loss": 0.8174,
+      "step": 15400
+    },
+    {
+      "epoch": 0.8476525950795311,
+      "grad_norm": 0.7897897362709045,
+      "learning_rate": 6.209394508176783e-06,
+      "loss": 0.6833,
+      "step": 15401
+    },
+    {
+      "epoch": 0.8477076338818867,
+      "grad_norm": 0.6803291440010071,
+      "learning_rate": 6.208973908180429e-06,
+      "loss": 0.7977,
+      "step": 15402
+    },
+    {
+      "epoch": 0.8477626726842424,
+      "grad_norm": 0.6937161087989807,
+      "learning_rate": 6.208553299097751e-06,
+      "loss": 0.7118,
+      "step": 15403
+    },
+    {
+      "epoch": 0.847817711486598,
+      "grad_norm": 0.7939958572387695,
+      "learning_rate": 6.208132680931911e-06,
+      "loss": 0.794,
+      "step": 15404
+    },
+    {
+      "epoch": 0.8478727502889537,
+      "grad_norm": 0.7009061574935913,
+      "learning_rate": 6.207712053686068e-06,
+      "loss": 0.7534,
+      "step": 15405
+    },
+    {
+      "epoch": 0.8479277890913094,
+      "grad_norm": 0.6890555620193481,
+      "learning_rate": 6.207291417363384e-06,
+      "loss": 0.7638,
+      "step": 15406
+    },
+    {
+      "epoch": 0.8479828278936651,
+      "grad_norm": 0.677119255065918,
+      "learning_rate": 6.206870771967022e-06,
+      "loss": 0.6814,
+      "step": 15407
+    },
+    {
+      "epoch": 0.8480378666960207,
+      "grad_norm": 0.706792950630188,
+      "learning_rate": 6.2064501175001425e-06,
+      "loss": 0.7722,
+      "step": 15408
+    },
+    {
+      "epoch": 0.8480929054983763,
+      "grad_norm": 0.6590496897697449,
+      "learning_rate": 6.206029453965905e-06,
+      "loss": 0.772,
+      "step": 15409
+    },
+    {
+      "epoch": 0.848147944300732,
+      "grad_norm": 0.6821194887161255,
+      "learning_rate": 6.205608781367475e-06,
+      "loss": 0.7687,
+      "step": 15410
+    },
+    {
+      "epoch": 0.8482029831030877,
+      "grad_norm": 0.6030088663101196,
+      "learning_rate": 6.205188099708011e-06,
+      "loss": 0.6673,
+      "step": 15411
+    },
+    {
+      "epoch": 0.8482580219054433,
+      "grad_norm": 0.6877727508544922,
+      "learning_rate": 6.204767408990676e-06,
+      "loss": 0.756,
+      "step": 15412
+    },
+    {
+      "epoch": 0.848313060707799,
+      "grad_norm": 0.7107367515563965,
+      "learning_rate": 6.204346709218632e-06,
+      "loss": 0.7481,
+      "step": 15413
+    },
+    {
+      "epoch": 0.8483680995101547,
+      "grad_norm": 0.7213658094406128,
+      "learning_rate": 6.2039260003950395e-06,
+      "loss": 0.7135,
+      "step": 15414
+    },
+    {
+      "epoch": 0.8484231383125104,
+      "grad_norm": 0.7002324461936951,
+      "learning_rate": 6.203505282523063e-06,
+      "loss": 0.6768,
+      "step": 15415
+    },
+    {
+      "epoch": 0.848478177114866,
+      "grad_norm": 0.7483230829238892,
+      "learning_rate": 6.2030845556058614e-06,
+      "loss": 0.633,
+      "step": 15416
+    },
+    {
+      "epoch": 0.8485332159172216,
+      "grad_norm": 0.6701670289039612,
+      "learning_rate": 6.2026638196466e-06,
+      "loss": 0.7936,
+      "step": 15417
+    },
+    {
+      "epoch": 0.8485882547195773,
+      "grad_norm": 0.6940304636955261,
+      "learning_rate": 6.202243074648438e-06,
+      "loss": 0.7787,
+      "step": 15418
+    },
+    {
+      "epoch": 0.848643293521933,
+      "grad_norm": 0.5912098288536072,
+      "learning_rate": 6.20182232061454e-06,
+      "loss": 0.6458,
+      "step": 15419
+    },
+    {
+      "epoch": 0.8486983323242886,
+      "grad_norm": 0.6538116931915283,
+      "learning_rate": 6.201401557548066e-06,
+      "loss": 0.6986,
+      "step": 15420
+    },
+    {
+      "epoch": 0.8487533711266443,
+      "grad_norm": 1.0245170593261719,
+      "learning_rate": 6.20098078545218e-06,
+      "loss": 0.7111,
+      "step": 15421
+    },
+    {
+      "epoch": 0.848808409929,
+      "grad_norm": 0.6896708011627197,
+      "learning_rate": 6.200560004330043e-06,
+      "loss": 0.7921,
+      "step": 15422
+    },
+    {
+      "epoch": 0.8488634487313556,
+      "grad_norm": 0.6219936013221741,
+      "learning_rate": 6.2001392141848195e-06,
+      "loss": 0.7345,
+      "step": 15423
+    },
+    {
+      "epoch": 0.8489184875337112,
+      "grad_norm": 0.7418678998947144,
+      "learning_rate": 6.199718415019671e-06,
+      "loss": 0.8517,
+      "step": 15424
+    },
+    {
+      "epoch": 0.8489735263360669,
+      "grad_norm": 0.7002347111701965,
+      "learning_rate": 6.199297606837759e-06,
+      "loss": 0.7345,
+      "step": 15425
+    },
+    {
+      "epoch": 0.8490285651384226,
+      "grad_norm": 0.7004539966583252,
+      "learning_rate": 6.198876789642247e-06,
+      "loss": 0.7639,
+      "step": 15426
+    },
+    {
+      "epoch": 0.8490836039407782,
+      "grad_norm": 0.64945387840271,
+      "learning_rate": 6.1984559634362995e-06,
+      "loss": 0.7556,
+      "step": 15427
+    },
+    {
+      "epoch": 0.8491386427431339,
+      "grad_norm": 0.6660465598106384,
+      "learning_rate": 6.1980351282230764e-06,
+      "loss": 0.7342,
+      "step": 15428
+    },
+    {
+      "epoch": 0.8491936815454896,
+      "grad_norm": 0.6177669763565063,
+      "learning_rate": 6.197614284005743e-06,
+      "loss": 0.7092,
+      "step": 15429
+    },
+    {
+      "epoch": 0.8492487203478453,
+      "grad_norm": 0.7604618072509766,
+      "learning_rate": 6.197193430787462e-06,
+      "loss": 0.8271,
+      "step": 15430
+    },
+    {
+      "epoch": 0.8493037591502008,
+      "grad_norm": 0.6788204312324524,
+      "learning_rate": 6.196772568571394e-06,
+      "loss": 0.7817,
+      "step": 15431
+    },
+    {
+      "epoch": 0.8493587979525565,
+      "grad_norm": 0.6073753833770752,
+      "learning_rate": 6.196351697360704e-06,
+      "loss": 0.6479,
+      "step": 15432
+    },
+    {
+      "epoch": 0.8494138367549122,
+      "grad_norm": 0.6842348575592041,
+      "learning_rate": 6.195930817158555e-06,
+      "loss": 0.7956,
+      "step": 15433
+    },
+    {
+      "epoch": 0.8494688755572679,
+      "grad_norm": 0.7863163352012634,
+      "learning_rate": 6.19550992796811e-06,
+      "loss": 0.7441,
+      "step": 15434
+    },
+    {
+      "epoch": 0.8495239143596235,
+      "grad_norm": 0.7495602965354919,
+      "learning_rate": 6.195089029792532e-06,
+      "loss": 0.7854,
+      "step": 15435
+    },
+    {
+      "epoch": 0.8495789531619792,
+      "grad_norm": 0.6595779061317444,
+      "learning_rate": 6.194668122634986e-06,
+      "loss": 0.6705,
+      "step": 15436
+    },
+    {
+      "epoch": 0.8496339919643349,
+      "grad_norm": 0.7727940082550049,
+      "learning_rate": 6.194247206498633e-06,
+      "loss": 0.7269,
+      "step": 15437
+    },
+    {
+      "epoch": 0.8496890307666906,
+      "grad_norm": 0.7433161735534668,
+      "learning_rate": 6.193826281386639e-06,
+      "loss": 0.7747,
+      "step": 15438
+    },
+    {
+      "epoch": 0.8497440695690461,
+      "grad_norm": 0.7075695991516113,
+      "learning_rate": 6.193405347302165e-06,
+      "loss": 0.8423,
+      "step": 15439
+    },
+    {
+      "epoch": 0.8497991083714018,
+      "grad_norm": 0.8821007013320923,
+      "learning_rate": 6.192984404248377e-06,
+      "loss": 0.705,
+      "step": 15440
+    },
+    {
+      "epoch": 0.8498541471737575,
+      "grad_norm": 0.7283695936203003,
+      "learning_rate": 6.192563452228437e-06,
+      "loss": 0.7013,
+      "step": 15441
+    },
+    {
+      "epoch": 0.8499091859761132,
+      "grad_norm": 0.7810649275779724,
+      "learning_rate": 6.192142491245509e-06,
+      "loss": 0.8303,
+      "step": 15442
+    },
+    {
+      "epoch": 0.8499642247784688,
+      "grad_norm": 0.5930086374282837,
+      "learning_rate": 6.191721521302758e-06,
+      "loss": 0.7117,
+      "step": 15443
+    },
+    {
+      "epoch": 0.8500192635808245,
+      "grad_norm": 0.6570530533790588,
+      "learning_rate": 6.191300542403347e-06,
+      "loss": 0.7525,
+      "step": 15444
+    },
+    {
+      "epoch": 0.8500743023831802,
+      "grad_norm": 0.8024932146072388,
+      "learning_rate": 6.190879554550437e-06,
+      "loss": 0.8011,
+      "step": 15445
+    },
+    {
+      "epoch": 0.8501293411855358,
+      "grad_norm": 0.851327121257782,
+      "learning_rate": 6.190458557747199e-06,
+      "loss": 0.8117,
+      "step": 15446
+    },
+    {
+      "epoch": 0.8501843799878914,
+      "grad_norm": 0.816034197807312,
+      "learning_rate": 6.190037551996791e-06,
+      "loss": 0.6659,
+      "step": 15447
+    },
+    {
+      "epoch": 0.8502394187902471,
+      "grad_norm": 0.7001582980155945,
+      "learning_rate": 6.18961653730238e-06,
+      "loss": 0.7406,
+      "step": 15448
+    },
+    {
+      "epoch": 0.8502944575926028,
+      "grad_norm": 0.6798322200775146,
+      "learning_rate": 6.189195513667129e-06,
+      "loss": 0.7504,
+      "step": 15449
+    },
+    {
+      "epoch": 0.8503494963949585,
+      "grad_norm": 0.6565585136413574,
+      "learning_rate": 6.188774481094203e-06,
+      "loss": 0.6445,
+      "step": 15450
+    },
+    {
+      "epoch": 0.8504045351973141,
+      "grad_norm": 0.674721360206604,
+      "learning_rate": 6.188353439586767e-06,
+      "loss": 0.6718,
+      "step": 15451
+    },
+    {
+      "epoch": 0.8504595739996698,
+      "grad_norm": 0.7626152634620667,
+      "learning_rate": 6.187932389147984e-06,
+      "loss": 0.7273,
+      "step": 15452
+    },
+    {
+      "epoch": 0.8505146128020254,
+      "grad_norm": 0.6497740149497986,
+      "learning_rate": 6.18751132978102e-06,
+      "loss": 0.7619,
+      "step": 15453
     }
   ],
   "logging_steps": 1,
@@ -101834,7 +108197,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 4.292023542595191e+19,
+  "total_flos": 4.56027501400739e+19,
   "train_batch_size": 1,
   "trial_name": null,
   "trial_params": null