{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9994141769185706,
  "eval_steps": 500,
  "global_step": 853,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0023432923257176333,
      "grad_norm": 0.00011052378977183253,
      "learning_rate": 5e-06,
      "loss": 0.0,
      "step": 2
    },
    {
      "epoch": 0.0046865846514352666,
      "grad_norm": 0.00020697808940894902,
      "learning_rate": 4.9941245593419514e-06,
      "loss": 0.0,
      "step": 4
    },
    {
      "epoch": 0.007029876977152899,
      "grad_norm": 0.0012532881228253245,
      "learning_rate": 4.982373678025853e-06,
      "loss": 0.0,
      "step": 6
    },
    {
      "epoch": 0.009373169302870533,
      "grad_norm": 0.0008086035377345979,
      "learning_rate": 4.970622796709754e-06,
      "loss": 0.0,
      "step": 8
    },
    {
      "epoch": 0.011716461628588167,
      "grad_norm": 0.0021155672147870064,
      "learning_rate": 4.958871915393655e-06,
      "loss": 0.0,
      "step": 10
    },
    {
      "epoch": 0.014059753954305799,
      "grad_norm": 0.0012233309680595994,
      "learning_rate": 4.947121034077556e-06,
      "loss": 0.0,
      "step": 12
    },
    {
      "epoch": 0.016403046280023433,
      "grad_norm": 0.0027737286873161793,
      "learning_rate": 4.9353701527614576e-06,
      "loss": 0.0,
      "step": 14
    },
    {
      "epoch": 0.018746338605741066,
      "grad_norm": 0.0042906939052045345,
      "learning_rate": 4.923619271445359e-06,
      "loss": 0.0,
      "step": 16
    },
    {
      "epoch": 0.0210896309314587,
      "grad_norm": 0.0005172386299818754,
      "learning_rate": 4.91186839012926e-06,
      "loss": 0.0,
      "step": 18
    },
    {
      "epoch": 0.023432923257176334,
      "grad_norm": 0.002410772955045104,
      "learning_rate": 4.900117508813161e-06,
      "loss": 0.0,
      "step": 20
    },
    {
      "epoch": 0.025776215582893967,
      "grad_norm": 0.6443753242492676,
      "learning_rate": 4.8883666274970625e-06,
      "loss": 0.0027,
      "step": 22
    },
    {
      "epoch": 0.028119507908611598,
      "grad_norm": 0.004394118674099445,
      "learning_rate": 4.876615746180964e-06,
      "loss": 0.0001,
      "step": 24
    },
    {
      "epoch": 0.03046280023432923,
      "grad_norm": 0.006466630846261978,
      "learning_rate": 4.864864864864866e-06,
      "loss": 0.0001,
      "step": 26
    },
    {
      "epoch": 0.032806092560046865,
      "grad_norm": 0.011924203485250473,
      "learning_rate": 4.853113983548767e-06,
      "loss": 0.0001,
      "step": 28
    },
    {
      "epoch": 0.0351493848857645,
      "grad_norm": 0.23746930062770844,
      "learning_rate": 4.841363102232668e-06,
      "loss": 0.0001,
      "step": 30
    },
    {
      "epoch": 0.03749267721148213,
      "grad_norm": 0.0031001348979771137,
      "learning_rate": 4.8296122209165694e-06,
      "loss": 0.0,
      "step": 32
    },
    {
      "epoch": 0.03983596953719976,
      "grad_norm": 0.0029028633143752813,
      "learning_rate": 4.817861339600471e-06,
      "loss": 0.0,
      "step": 34
    },
    {
      "epoch": 0.0421792618629174,
      "grad_norm": 0.014626468531787395,
      "learning_rate": 4.806110458284372e-06,
      "loss": 0.0001,
      "step": 36
    },
    {
      "epoch": 0.04452255418863503,
      "grad_norm": 0.001155451056547463,
      "learning_rate": 4.794359576968273e-06,
      "loss": 0.0,
      "step": 38
    },
    {
      "epoch": 0.04686584651435267,
      "grad_norm": 0.003476829966530204,
      "learning_rate": 4.782608695652174e-06,
      "loss": 0.0,
      "step": 40
    },
    {
      "epoch": 0.0492091388400703,
      "grad_norm": 0.0002227002551080659,
      "learning_rate": 4.7708578143360756e-06,
      "loss": 0.0,
      "step": 42
    },
    {
      "epoch": 0.051552431165787935,
      "grad_norm": 0.0001427282695658505,
      "learning_rate": 4.759106933019977e-06,
      "loss": 0.0,
      "step": 44
    },
    {
      "epoch": 0.053895723491505565,
      "grad_norm": 0.0027408564928919077,
      "learning_rate": 4.747356051703878e-06,
      "loss": 0.0002,
      "step": 46
    },
    {
      "epoch": 0.056239015817223195,
      "grad_norm": 0.0020253027323633432,
      "learning_rate": 4.735605170387779e-06,
      "loss": 0.0,
      "step": 48
    },
    {
      "epoch": 0.05858230814294083,
      "grad_norm": 0.001760220737196505,
      "learning_rate": 4.723854289071681e-06,
      "loss": 0.0,
      "step": 50
    },
    {
      "epoch": 0.06092560046865846,
      "grad_norm": 0.0010492791188880801,
      "learning_rate": 4.7121034077555825e-06,
      "loss": 0.0,
      "step": 52
    },
    {
      "epoch": 0.0632688927943761,
      "grad_norm": 0.002001305343583226,
      "learning_rate": 4.700352526439484e-06,
      "loss": 0.0,
      "step": 54
    },
    {
      "epoch": 0.06561218512009373,
      "grad_norm": 0.18566887080669403,
      "learning_rate": 4.688601645123384e-06,
      "loss": 0.0009,
      "step": 56
    },
    {
      "epoch": 0.06795547744581136,
      "grad_norm": 0.0009072807151824236,
      "learning_rate": 4.676850763807285e-06,
      "loss": 0.0,
      "step": 58
    },
    {
      "epoch": 0.070298769771529,
      "grad_norm": 0.003983665257692337,
      "learning_rate": 4.665099882491187e-06,
      "loss": 0.0006,
      "step": 60
    },
    {
      "epoch": 0.07264206209724663,
      "grad_norm": 0.01946200616657734,
      "learning_rate": 4.653349001175089e-06,
      "loss": 0.0001,
      "step": 62
    },
    {
      "epoch": 0.07498535442296426,
      "grad_norm": 0.004048655740916729,
      "learning_rate": 4.64159811985899e-06,
      "loss": 0.0,
      "step": 64
    },
    {
      "epoch": 0.0773286467486819,
      "grad_norm": 0.0005872617475688457,
      "learning_rate": 4.629847238542891e-06,
      "loss": 0.0001,
      "step": 66
    },
    {
      "epoch": 0.07967193907439953,
      "grad_norm": 0.008831903338432312,
      "learning_rate": 4.618096357226792e-06,
      "loss": 0.0001,
      "step": 68
    },
    {
      "epoch": 0.08201523140011717,
      "grad_norm": 0.006819219794124365,
      "learning_rate": 4.6063454759106936e-06,
      "loss": 0.0001,
      "step": 70
    },
    {
      "epoch": 0.0843585237258348,
      "grad_norm": 0.0007863900391384959,
      "learning_rate": 4.594594594594596e-06,
      "loss": 0.0,
      "step": 72
    },
    {
      "epoch": 0.08670181605155243,
      "grad_norm": 0.032210394740104675,
      "learning_rate": 4.582843713278496e-06,
      "loss": 0.0001,
      "step": 74
    },
    {
      "epoch": 0.08904510837727006,
      "grad_norm": 0.2614983916282654,
      "learning_rate": 4.571092831962397e-06,
      "loss": 0.0008,
      "step": 76
    },
    {
      "epoch": 0.0913884007029877,
      "grad_norm": 0.0012551415711641312,
      "learning_rate": 4.5593419506462985e-06,
      "loss": 0.0,
      "step": 78
    },
    {
      "epoch": 0.09373169302870533,
      "grad_norm": 0.0019108065171167254,
      "learning_rate": 4.5475910693302e-06,
      "loss": 0.0,
      "step": 80
    },
    {
      "epoch": 0.09607498535442296,
      "grad_norm": 0.02294810675084591,
      "learning_rate": 4.535840188014101e-06,
      "loss": 0.0001,
      "step": 82
    },
    {
      "epoch": 0.0984182776801406,
      "grad_norm": 0.0012388118775561452,
      "learning_rate": 4.524089306698003e-06,
      "loss": 0.0,
      "step": 84
    },
    {
      "epoch": 0.10076157000585823,
      "grad_norm": 0.001227575121447444,
      "learning_rate": 4.512338425381904e-06,
      "loss": 0.0001,
      "step": 86
    },
    {
      "epoch": 0.10310486233157587,
      "grad_norm": 0.004755712114274502,
      "learning_rate": 4.5005875440658054e-06,
      "loss": 0.0001,
      "step": 88
    },
    {
      "epoch": 0.1054481546572935,
      "grad_norm": 0.00837083999067545,
      "learning_rate": 4.488836662749707e-06,
      "loss": 0.0001,
      "step": 90
    },
    {
      "epoch": 0.10779144698301113,
      "grad_norm": 0.48219314217567444,
      "learning_rate": 4.477085781433608e-06,
      "loss": 0.0017,
      "step": 92
    },
    {
      "epoch": 0.11013473930872876,
      "grad_norm": 0.022060217335820198,
      "learning_rate": 4.465334900117509e-06,
      "loss": 0.0001,
      "step": 94
    },
    {
      "epoch": 0.11247803163444639,
      "grad_norm": 0.0019385352497920394,
      "learning_rate": 4.45358401880141e-06,
      "loss": 0.0,
      "step": 96
    },
    {
      "epoch": 0.11482132396016403,
      "grad_norm": 0.01225442998111248,
      "learning_rate": 4.4418331374853116e-06,
      "loss": 0.0001,
      "step": 98
    },
    {
      "epoch": 0.11716461628588166,
      "grad_norm": 0.0005759520572610199,
      "learning_rate": 4.430082256169213e-06,
      "loss": 0.0,
      "step": 100
    },
    {
      "epoch": 0.1195079086115993,
      "grad_norm": 0.02452813647687435,
      "learning_rate": 4.418331374853114e-06,
      "loss": 0.0001,
      "step": 102
    },
    {
      "epoch": 0.12185120093731693,
      "grad_norm": 0.0078084710985422134,
      "learning_rate": 4.406580493537015e-06,
      "loss": 0.0001,
      "step": 104
    },
    {
      "epoch": 0.12419449326303457,
      "grad_norm": 0.004263446666300297,
      "learning_rate": 4.394829612220917e-06,
      "loss": 0.0001,
      "step": 106
    },
    {
      "epoch": 0.1265377855887522,
      "grad_norm": 0.0016304058954119682,
      "learning_rate": 4.3830787309048185e-06,
      "loss": 0.0001,
      "step": 108
    },
    {
      "epoch": 0.12888107791446984,
      "grad_norm": 0.011672005988657475,
      "learning_rate": 4.37132784958872e-06,
      "loss": 0.0002,
      "step": 110
    },
    {
      "epoch": 0.13122437024018746,
      "grad_norm": 0.002603155327960849,
      "learning_rate": 4.359576968272621e-06,
      "loss": 0.0,
      "step": 112
    },
    {
      "epoch": 0.1335676625659051,
      "grad_norm": 0.005059251096099615,
      "learning_rate": 4.347826086956522e-06,
      "loss": 0.0001,
      "step": 114
    },
    {
      "epoch": 0.13591095489162272,
      "grad_norm": 0.0005816388293169439,
      "learning_rate": 4.3360752056404234e-06,
      "loss": 0.0001,
      "step": 116
    },
    {
      "epoch": 0.13825424721734036,
      "grad_norm": 0.019756818190217018,
      "learning_rate": 4.324324324324325e-06,
      "loss": 0.0001,
      "step": 118
    },
    {
      "epoch": 0.140597539543058,
      "grad_norm": 0.0023519208189100027,
      "learning_rate": 4.312573443008226e-06,
      "loss": 0.0,
      "step": 120
    },
    {
      "epoch": 0.14294083186877563,
      "grad_norm": 0.0028086318634450436,
      "learning_rate": 4.300822561692127e-06,
      "loss": 0.0,
      "step": 122
    },
    {
      "epoch": 0.14528412419449327,
      "grad_norm": 0.0022307527251541615,
      "learning_rate": 4.289071680376028e-06,
      "loss": 0.0,
      "step": 124
    },
    {
      "epoch": 0.14762741652021089,
      "grad_norm": 0.014247684739530087,
      "learning_rate": 4.2773207990599296e-06,
      "loss": 0.0001,
      "step": 126
    },
    {
      "epoch": 0.14997070884592853,
      "grad_norm": 0.00011139630805701017,
      "learning_rate": 4.265569917743831e-06,
      "loss": 0.0,
      "step": 128
    },
    {
      "epoch": 0.15231400117164617,
      "grad_norm": 0.000514341751113534,
      "learning_rate": 4.253819036427733e-06,
      "loss": 0.0,
      "step": 130
    },
    {
      "epoch": 0.1546572934973638,
      "grad_norm": 0.002176255453377962,
      "learning_rate": 4.242068155111634e-06,
      "loss": 0.0001,
      "step": 132
    },
    {
      "epoch": 0.15700058582308143,
      "grad_norm": 0.018497969955205917,
      "learning_rate": 4.230317273795535e-06,
      "loss": 0.0001,
      "step": 134
    },
    {
      "epoch": 0.15934387814879905,
      "grad_norm": 0.013157431036233902,
      "learning_rate": 4.2185663924794365e-06,
      "loss": 0.0001,
      "step": 136
    },
    {
      "epoch": 0.1616871704745167,
      "grad_norm": 0.007630129344761372,
      "learning_rate": 4.206815511163338e-06,
      "loss": 0.0,
      "step": 138
    },
    {
      "epoch": 0.16403046280023434,
      "grad_norm": 0.0008055138750933111,
      "learning_rate": 4.195064629847239e-06,
      "loss": 0.0001,
      "step": 140
    },
    {
      "epoch": 0.16637375512595196,
      "grad_norm": 0.006306421477347612,
      "learning_rate": 4.18331374853114e-06,
      "loss": 0.0,
      "step": 142
    },
    {
      "epoch": 0.1687170474516696,
      "grad_norm": 0.020266445353627205,
      "learning_rate": 4.1715628672150414e-06,
      "loss": 0.0001,
      "step": 144
    },
    {
      "epoch": 0.17106033977738722,
      "grad_norm": 0.00037427974166348577,
      "learning_rate": 4.159811985898943e-06,
      "loss": 0.0,
      "step": 146
    },
    {
      "epoch": 0.17340363210310486,
      "grad_norm": 0.004259356763213873,
      "learning_rate": 4.148061104582844e-06,
      "loss": 0.0001,
      "step": 148
    },
    {
      "epoch": 0.1757469244288225,
      "grad_norm": 0.0010232679778710008,
      "learning_rate": 4.136310223266745e-06,
      "loss": 0.0001,
      "step": 150
    },
    {
      "epoch": 0.17809021675454012,
      "grad_norm": 0.003952402155846357,
      "learning_rate": 4.124559341950647e-06,
      "loss": 0.0,
      "step": 152
    },
    {
      "epoch": 0.18043350908025776,
      "grad_norm": 0.0013295585522428155,
      "learning_rate": 4.112808460634548e-06,
      "loss": 0.0,
      "step": 154
    },
    {
      "epoch": 0.1827768014059754,
      "grad_norm": 0.013831949792802334,
      "learning_rate": 4.10105757931845e-06,
      "loss": 0.0001,
      "step": 156
    },
    {
      "epoch": 0.18512009373169303,
      "grad_norm": 0.0036904062144458294,
      "learning_rate": 4.089306698002351e-06,
      "loss": 0.0,
      "step": 158
    },
    {
      "epoch": 0.18746338605741067,
      "grad_norm": 0.002993196714669466,
      "learning_rate": 4.077555816686252e-06,
      "loss": 0.0,
      "step": 160
    },
    {
      "epoch": 0.18980667838312829,
      "grad_norm": 0.0016740068094804883,
      "learning_rate": 4.0658049353701525e-06,
      "loss": 0.0001,
      "step": 162
    },
    {
      "epoch": 0.19214997070884593,
      "grad_norm": 0.012307717464864254,
      "learning_rate": 4.0540540540540545e-06,
      "loss": 0.0001,
      "step": 164
    },
    {
      "epoch": 0.19449326303456357,
      "grad_norm": 0.0012654109159484506,
      "learning_rate": 4.042303172737956e-06,
      "loss": 0.0,
      "step": 166
    },
    {
      "epoch": 0.1968365553602812,
      "grad_norm": 0.12437883019447327,
      "learning_rate": 4.030552291421857e-06,
      "loss": 0.0006,
      "step": 168
    },
    {
      "epoch": 0.19917984768599883,
      "grad_norm": 8.974138472694904e-05,
      "learning_rate": 4.018801410105758e-06,
      "loss": 0.0,
      "step": 170
    },
    {
      "epoch": 0.20152314001171645,
      "grad_norm": 0.0011903212871402502,
      "learning_rate": 4.007050528789659e-06,
      "loss": 0.0001,
      "step": 172
    },
    {
      "epoch": 0.2038664323374341,
      "grad_norm": 0.012350277975201607,
      "learning_rate": 3.995299647473561e-06,
      "loss": 0.0001,
      "step": 174
    },
    {
      "epoch": 0.20620972466315174,
      "grad_norm": 0.01664598099887371,
      "learning_rate": 3.983548766157463e-06,
      "loss": 0.0001,
      "step": 176
    },
    {
      "epoch": 0.20855301698886936,
      "grad_norm": 0.0064240009523928165,
      "learning_rate": 3.971797884841364e-06,
      "loss": 0.0001,
      "step": 178
    },
    {
      "epoch": 0.210896309314587,
      "grad_norm": 0.0031362581066787243,
      "learning_rate": 3.960047003525264e-06,
      "loss": 0.0,
      "step": 180
    },
    {
      "epoch": 0.21323960164030462,
      "grad_norm": 0.00012566300574690104,
      "learning_rate": 3.9482961222091655e-06,
      "loss": 0.0001,
      "step": 182
    },
    {
      "epoch": 0.21558289396602226,
      "grad_norm": 0.0018261070363223553,
      "learning_rate": 3.936545240893067e-06,
      "loss": 0.0,
      "step": 184
    },
    {
      "epoch": 0.2179261862917399,
      "grad_norm": 0.0010897299507632852,
      "learning_rate": 3.924794359576969e-06,
      "loss": 0.0,
      "step": 186
    },
    {
      "epoch": 0.22026947861745752,
      "grad_norm": 0.006528445053845644,
      "learning_rate": 3.91304347826087e-06,
      "loss": 0.0,
      "step": 188
    },
    {
      "epoch": 0.22261277094317516,
      "grad_norm": 0.4626096785068512,
      "learning_rate": 3.901292596944771e-06,
      "loss": 0.0009,
      "step": 190
    },
    {
      "epoch": 0.22495606326889278,
      "grad_norm": 0.002359338803216815,
      "learning_rate": 3.8895417156286725e-06,
      "loss": 0.0,
      "step": 192
    },
    {
      "epoch": 0.22729935559461042,
      "grad_norm": 0.004821418318897486,
      "learning_rate": 3.877790834312574e-06,
      "loss": 0.0,
      "step": 194
    },
    {
      "epoch": 0.22964264792032807,
      "grad_norm": 0.0011465001152828336,
      "learning_rate": 3.866039952996475e-06,
      "loss": 0.0008,
      "step": 196
    },
    {
      "epoch": 0.23198594024604569,
      "grad_norm": 0.0007381247123703361,
      "learning_rate": 3.854289071680376e-06,
      "loss": 0.0001,
      "step": 198
    },
    {
      "epoch": 0.23432923257176333,
      "grad_norm": 0.0023091183975338936,
      "learning_rate": 3.842538190364277e-06,
      "loss": 0.0,
      "step": 200
    },
    {
      "epoch": 0.23667252489748097,
      "grad_norm": 0.0005714365397579968,
      "learning_rate": 3.830787309048179e-06,
      "loss": 0.0,
      "step": 202
    },
    {
      "epoch": 0.2390158172231986,
      "grad_norm": 0.00351692084223032,
      "learning_rate": 3.81903642773208e-06,
      "loss": 0.0,
      "step": 204
    },
    {
      "epoch": 0.24135910954891623,
      "grad_norm": 5.926425728830509e-05,
      "learning_rate": 3.8072855464159815e-06,
      "loss": 0.0,
      "step": 206
    },
    {
      "epoch": 0.24370240187463385,
      "grad_norm": 0.0016421001637354493,
      "learning_rate": 3.7955346650998827e-06,
      "loss": 0.0,
      "step": 208
    },
    {
      "epoch": 0.2460456942003515,
      "grad_norm": 0.012118808925151825,
      "learning_rate": 3.7837837837837844e-06,
      "loss": 0.0001,
      "step": 210
    },
    {
      "epoch": 0.24838898652606914,
      "grad_norm": 0.00024874648079276085,
      "learning_rate": 3.7720329024676856e-06,
      "loss": 0.0002,
      "step": 212
    },
    {
      "epoch": 0.2507322788517868,
      "grad_norm": 0.0017625248292461038,
      "learning_rate": 3.760282021151587e-06,
      "loss": 0.0,
      "step": 214
    },
    {
      "epoch": 0.2530755711775044,
      "grad_norm": 0.0007431196281686425,
      "learning_rate": 3.748531139835488e-06,
      "loss": 0.0,
      "step": 216
    },
    {
      "epoch": 0.255418863503222,
      "grad_norm": 0.0007026457460597157,
      "learning_rate": 3.7367802585193893e-06,
      "loss": 0.0,
      "step": 218
    },
    {
      "epoch": 0.2577621558289397,
      "grad_norm": 0.002397920237854123,
      "learning_rate": 3.72502937720329e-06,
      "loss": 0.0,
      "step": 220
    },
    {
      "epoch": 0.2601054481546573,
      "grad_norm": 0.003177257487550378,
      "learning_rate": 3.713278495887192e-06,
      "loss": 0.0,
      "step": 222
    },
    {
      "epoch": 0.2624487404803749,
      "grad_norm": 0.003142025787383318,
      "learning_rate": 3.7015276145710934e-06,
      "loss": 0.0001,
      "step": 224
    },
    {
      "epoch": 0.26479203280609254,
      "grad_norm": 0.03788410872220993,
      "learning_rate": 3.6897767332549946e-06,
      "loss": 0.0002,
      "step": 226
    },
    {
      "epoch": 0.2671353251318102,
      "grad_norm": 0.005685464479029179,
      "learning_rate": 3.6780258519388954e-06,
      "loss": 0.0003,
      "step": 228
    },
    {
      "epoch": 0.2694786174575278,
      "grad_norm": 0.0010328789940103889,
      "learning_rate": 3.6662749706227966e-06,
      "loss": 0.0003,
      "step": 230
    },
    {
      "epoch": 0.27182190978324544,
      "grad_norm": 0.0052024442702531815,
      "learning_rate": 3.6545240893066987e-06,
      "loss": 0.0,
      "step": 232
    },
    {
      "epoch": 0.2741652021089631,
      "grad_norm": 0.006033598445355892,
      "learning_rate": 3.6427732079906e-06,
      "loss": 0.0,
      "step": 234
    },
    {
      "epoch": 0.27650849443468073,
      "grad_norm": 0.00023948443413246423,
      "learning_rate": 3.6310223266745007e-06,
      "loss": 0.0001,
      "step": 236
    },
    {
      "epoch": 0.27885178676039835,
      "grad_norm": 0.00016467843670397997,
      "learning_rate": 3.619271445358402e-06,
      "loss": 0.0,
      "step": 238
    },
    {
      "epoch": 0.281195079086116,
      "grad_norm": 0.003566320287063718,
      "learning_rate": 3.607520564042303e-06,
      "loss": 0.0,
      "step": 240
    },
    {
      "epoch": 0.28353837141183363,
      "grad_norm": 0.00033969045034609735,
      "learning_rate": 3.5957696827262044e-06,
      "loss": 0.0,
      "step": 242
    },
    {
      "epoch": 0.28588166373755125,
      "grad_norm": 0.0033994223922491074,
      "learning_rate": 3.5840188014101065e-06,
      "loss": 0.0,
      "step": 244
    },
    {
      "epoch": 0.28822495606326887,
      "grad_norm": 0.14746786653995514,
      "learning_rate": 3.5722679200940073e-06,
      "loss": 0.0008,
      "step": 246
    },
    {
      "epoch": 0.29056824838898654,
      "grad_norm": 0.012470235116779804,
      "learning_rate": 3.5605170387779085e-06,
      "loss": 0.0,
      "step": 248
    },
    {
      "epoch": 0.29291154071470415,
      "grad_norm": 0.08307931572198868,
      "learning_rate": 3.5487661574618097e-06,
      "loss": 0.0003,
      "step": 250
    },
    {
      "epoch": 0.29525483304042177,
      "grad_norm": 0.00033245363738387823,
      "learning_rate": 3.537015276145711e-06,
      "loss": 0.0,
      "step": 252
    },
    {
      "epoch": 0.29759812536613944,
      "grad_norm": 0.0018247144762426615,
      "learning_rate": 3.525264394829612e-06,
      "loss": 0.0,
      "step": 254
    },
    {
      "epoch": 0.29994141769185706,
      "grad_norm": 0.0011103990254923701,
      "learning_rate": 3.513513513513514e-06,
      "loss": 0.0001,
      "step": 256
    },
    {
      "epoch": 0.3022847100175747,
      "grad_norm": 0.0010811882093548775,
      "learning_rate": 3.501762632197415e-06,
      "loss": 0.0,
      "step": 258
    },
    {
      "epoch": 0.30462800234329235,
      "grad_norm": 0.011172047816216946,
      "learning_rate": 3.4900117508813163e-06,
      "loss": 0.0001,
      "step": 260
    },
    {
      "epoch": 0.30697129466900996,
      "grad_norm": 0.0013676233356818557,
      "learning_rate": 3.4782608695652175e-06,
      "loss": 0.0,
      "step": 262
    },
    {
      "epoch": 0.3093145869947276,
      "grad_norm": 0.002147970488294959,
      "learning_rate": 3.4665099882491187e-06,
      "loss": 0.0,
      "step": 264
    },
    {
      "epoch": 0.31165787932044525,
      "grad_norm": 0.0009826518362388015,
      "learning_rate": 3.4547591069330204e-06,
      "loss": 0.0,
      "step": 266
    },
    {
      "epoch": 0.31400117164616287,
      "grad_norm": 0.001499099307693541,
      "learning_rate": 3.4430082256169216e-06,
      "loss": 0.0,
      "step": 268
    },
    {
      "epoch": 0.3163444639718805,
      "grad_norm": 0.001323301112279296,
      "learning_rate": 3.431257344300823e-06,
      "loss": 0.0,
      "step": 270
    },
    {
      "epoch": 0.3186877562975981,
      "grad_norm": 0.018010340631008148,
      "learning_rate": 3.419506462984724e-06,
      "loss": 0.0005,
      "step": 272
    },
    {
      "epoch": 0.3210310486233158,
      "grad_norm": 0.0024064648896455765,
      "learning_rate": 3.4077555816686253e-06,
      "loss": 0.0,
      "step": 274
    },
    {
      "epoch": 0.3233743409490334,
      "grad_norm": 0.02396260015666485,
      "learning_rate": 3.3960047003525265e-06,
      "loss": 0.0001,
      "step": 276
    },
    {
      "epoch": 0.325717633274751,
      "grad_norm": 0.002070352202281356,
      "learning_rate": 3.384253819036428e-06,
      "loss": 0.0,
      "step": 278
    },
    {
      "epoch": 0.3280609256004687,
      "grad_norm": 0.0003108434902969748,
      "learning_rate": 3.3725029377203294e-06,
      "loss": 0.0001,
      "step": 280
    },
    {
      "epoch": 0.3304042179261863,
      "grad_norm": 0.006573045626282692,
      "learning_rate": 3.3607520564042306e-06,
      "loss": 0.0001,
      "step": 282
    },
    {
      "epoch": 0.3327475102519039,
      "grad_norm": 0.0004413512069731951,
      "learning_rate": 3.349001175088132e-06,
      "loss": 0.0001,
      "step": 284
    },
    {
      "epoch": 0.3350908025776216,
      "grad_norm": 0.0005645502242259681,
      "learning_rate": 3.337250293772033e-06,
      "loss": 0.0,
      "step": 286
    },
    {
      "epoch": 0.3374340949033392,
      "grad_norm": 0.00034579774364829063,
      "learning_rate": 3.3254994124559343e-06,
      "loss": 0.0,
      "step": 288
    },
    {
      "epoch": 0.3397773872290568,
      "grad_norm": 0.003136229468509555,
      "learning_rate": 3.313748531139836e-06,
      "loss": 0.0,
      "step": 290
    },
    {
      "epoch": 0.34212067955477443,
      "grad_norm": 0.0031148705165833235,
      "learning_rate": 3.301997649823737e-06,
      "loss": 0.0,
      "step": 292
    },
    {
      "epoch": 0.3444639718804921,
      "grad_norm": 0.0012612566351890564,
      "learning_rate": 3.2902467685076384e-06,
      "loss": 0.0,
      "step": 294
    },
    {
      "epoch": 0.3468072642062097,
      "grad_norm": 0.0007469533011317253,
      "learning_rate": 3.2784958871915396e-06,
      "loss": 0.0,
      "step": 296
    },
    {
      "epoch": 0.34915055653192734,
      "grad_norm": 0.04412250965833664,
      "learning_rate": 3.266745005875441e-06,
      "loss": 0.0003,
      "step": 298
    },
    {
      "epoch": 0.351493848857645,
      "grad_norm": 0.004462533164769411,
      "learning_rate": 3.2549941245593425e-06,
      "loss": 0.0088,
      "step": 300
    },
    {
      "epoch": 0.3538371411833626,
      "grad_norm": 0.002911294111981988,
      "learning_rate": 3.2432432432432437e-06,
      "loss": 0.0006,
      "step": 302
    },
    {
      "epoch": 0.35618043350908024,
      "grad_norm": 0.0015191801358014345,
      "learning_rate": 3.231492361927145e-06,
      "loss": 0.0,
      "step": 304
    },
    {
      "epoch": 0.3585237258347979,
      "grad_norm": 0.017380721867084503,
      "learning_rate": 3.219741480611046e-06,
      "loss": 0.0094,
      "step": 306
    },
    {
      "epoch": 0.36086701816051553,
      "grad_norm": 0.002749436302110553,
      "learning_rate": 3.2079905992949474e-06,
      "loss": 0.0001,
      "step": 308
    },
    {
      "epoch": 0.36321031048623315,
      "grad_norm": 0.0008673086995258927,
      "learning_rate": 3.1962397179788486e-06,
      "loss": 0.0,
      "step": 310
    },
    {
      "epoch": 0.3655536028119508,
      "grad_norm": 0.00361701101064682,
      "learning_rate": 3.1844888366627503e-06,
      "loss": 0.0,
      "step": 312
    },
    {
      "epoch": 0.36789689513766843,
      "grad_norm": 0.006906528025865555,
      "learning_rate": 3.1727379553466515e-06,
      "loss": 0.0,
      "step": 314
    },
    {
      "epoch": 0.37024018746338605,
      "grad_norm": 2.259305238723755,
      "learning_rate": 3.1609870740305527e-06,
      "loss": 0.0157,
      "step": 316
    },
    {
      "epoch": 0.37258347978910367,
      "grad_norm": 0.00017454673070460558,
      "learning_rate": 3.149236192714454e-06,
      "loss": 0.0,
      "step": 318
    },
    {
      "epoch": 0.37492677211482134,
      "grad_norm": 0.16197967529296875,
      "learning_rate": 3.137485311398355e-06,
      "loss": 0.0009,
      "step": 320
    },
    {
      "epoch": 0.37727006444053895,
      "grad_norm": 0.002247605938464403,
      "learning_rate": 3.1257344300822564e-06,
      "loss": 0.0,
      "step": 322
    },
    {
      "epoch": 0.37961335676625657,
      "grad_norm": 0.023727795109152794,
      "learning_rate": 3.113983548766158e-06,
      "loss": 0.0001,
      "step": 324
    },
    {
      "epoch": 0.38195664909197424,
      "grad_norm": 0.008455273695290089,
      "learning_rate": 3.1022326674500592e-06,
      "loss": 0.0001,
      "step": 326
    },
    {
      "epoch": 0.38429994141769186,
      "grad_norm": 0.00022873218404129148,
      "learning_rate": 3.0904817861339605e-06,
      "loss": 0.0,
      "step": 328
    },
    {
      "epoch": 0.3866432337434095,
      "grad_norm": 3.000872850418091,
      "learning_rate": 3.0787309048178617e-06,
      "loss": 0.055,
      "step": 330
    },
    {
      "epoch": 0.38898652606912715,
      "grad_norm": 0.002177221467718482,
      "learning_rate": 3.066980023501763e-06,
      "loss": 0.0,
      "step": 332
    },
    {
      "epoch": 0.39132981839484476,
      "grad_norm": 0.002786975121125579,
      "learning_rate": 3.0552291421856637e-06,
      "loss": 0.0,
      "step": 334
    },
    {
      "epoch": 0.3936731107205624,
      "grad_norm": 0.004335256293416023,
      "learning_rate": 3.043478260869566e-06,
      "loss": 0.0,
      "step": 336
    },
    {
      "epoch": 0.39601640304628,
      "grad_norm": 0.007627409417182207,
      "learning_rate": 3.031727379553467e-06,
      "loss": 0.0001,
      "step": 338
    },
    {
      "epoch": 0.39835969537199767,
      "grad_norm": 0.002631911775097251,
      "learning_rate": 3.0199764982373682e-06,
      "loss": 0.0,
      "step": 340
    },
    {
      "epoch": 0.4007029876977153,
      "grad_norm": 0.009561799466609955,
      "learning_rate": 3.008225616921269e-06,
      "loss": 0.0001,
      "step": 342
    },
    {
      "epoch": 0.4030462800234329,
      "grad_norm": 0.0026635443791747093,
      "learning_rate": 2.9964747356051703e-06,
      "loss": 0.0001,
      "step": 344
    },
    {
      "epoch": 0.4053895723491506,
      "grad_norm": 0.0001533351169200614,
      "learning_rate": 2.9847238542890723e-06,
      "loss": 0.0,
      "step": 346
    },
    {
      "epoch": 0.4077328646748682,
      "grad_norm": 0.0835270956158638,
      "learning_rate": 2.9729729729729736e-06,
      "loss": 0.0005,
      "step": 348
    },
    {
      "epoch": 0.4100761570005858,
      "grad_norm": 0.003761101048439741,
      "learning_rate": 2.9612220916568744e-06,
      "loss": 0.0,
      "step": 350
    },
    {
      "epoch": 0.4124194493263035,
      "grad_norm": 0.01136633288115263,
      "learning_rate": 2.9494712103407756e-06,
      "loss": 0.0002,
      "step": 352
    },
    {
      "epoch": 0.4147627416520211,
      "grad_norm": 0.007711971178650856,
      "learning_rate": 2.937720329024677e-06,
      "loss": 0.0001,
      "step": 354
    },
    {
      "epoch": 0.4171060339777387,
      "grad_norm": 0.0003854953683912754,
      "learning_rate": 2.925969447708578e-06,
      "loss": 0.0,
      "step": 356
    },
    {
      "epoch": 0.4194493263034564,
      "grad_norm": 0.019140860065817833,
      "learning_rate": 2.91421856639248e-06,
      "loss": 0.0001,
      "step": 358
    },
    {
      "epoch": 0.421792618629174,
      "grad_norm": 0.0013410028768703341,
      "learning_rate": 2.902467685076381e-06,
      "loss": 0.0003,
      "step": 360
    },
    {
      "epoch": 0.4241359109548916,
      "grad_norm": 0.0011243935441598296,
      "learning_rate": 2.890716803760282e-06,
      "loss": 0.0001,
      "step": 362
    },
    {
      "epoch": 0.42647920328060923,
      "grad_norm": 0.012134709395468235,
      "learning_rate": 2.8789659224441834e-06,
      "loss": 0.0001,
      "step": 364
    },
    {
      "epoch": 0.4288224956063269,
      "grad_norm": 0.0028234529308974743,
      "learning_rate": 2.8672150411280846e-06,
      "loss": 0.0,
      "step": 366
    },
    {
      "epoch": 0.4311657879320445,
      "grad_norm": 0.004319467581808567,
      "learning_rate": 2.855464159811986e-06,
      "loss": 0.0,
      "step": 368
    },
    {
      "epoch": 0.43350908025776214,
      "grad_norm": 0.0068093533627688885,
      "learning_rate": 2.8437132784958875e-06,
      "loss": 0.0001,
      "step": 370
    },
    {
      "epoch": 0.4358523725834798,
      "grad_norm": 0.016774361953139305,
      "learning_rate": 2.8319623971797887e-06,
      "loss": 0.0001,
      "step": 372
    },
    {
      "epoch": 0.4381956649091974,
      "grad_norm": 0.014978869818150997,
      "learning_rate": 2.82021151586369e-06,
      "loss": 0.0001,
      "step": 374
    },
    {
      "epoch": 0.44053895723491504,
      "grad_norm": 0.0010881100315600634,
      "learning_rate": 2.808460634547591e-06,
      "loss": 0.0004,
      "step": 376
    },
    {
      "epoch": 0.4428822495606327,
      "grad_norm": 0.05522293969988823,
      "learning_rate": 2.7967097532314924e-06,
      "loss": 0.0002,
      "step": 378
    },
    {
      "epoch": 0.44522554188635033,
      "grad_norm": 0.0027575818821787834,
      "learning_rate": 2.784958871915394e-06,
      "loss": 0.0,
      "step": 380
    },
    {
      "epoch": 0.44756883421206795,
      "grad_norm": 0.0006020054570399225,
      "learning_rate": 2.7732079905992952e-06,
      "loss": 0.0005,
      "step": 382
    },
    {
      "epoch": 0.44991212653778556,
      "grad_norm": 0.0025616425555199385,
      "learning_rate": 2.7614571092831965e-06,
      "loss": 0.0,
      "step": 384
    },
    {
      "epoch": 0.45225541886350323,
      "grad_norm": 0.0018823420396074653,
      "learning_rate": 2.7497062279670977e-06,
      "loss": 0.0,
      "step": 386
    },
    {
      "epoch": 0.45459871118922085,
      "grad_norm": 0.003241207217797637,
      "learning_rate": 2.737955346650999e-06,
      "loss": 0.0,
      "step": 388
    },
    {
      "epoch": 0.45694200351493847,
      "grad_norm": 0.0010485474485903978,
      "learning_rate": 2.7262044653349e-06,
      "loss": 0.0002,
      "step": 390
    },
    {
      "epoch": 0.45928529584065614,
      "grad_norm": 0.013366922736167908,
      "learning_rate": 2.714453584018802e-06,
      "loss": 0.0001,
      "step": 392
    },
    {
      "epoch": 0.46162858816637375,
      "grad_norm": 0.0005886501166969538,
      "learning_rate": 2.702702702702703e-06,
      "loss": 0.0,
      "step": 394
    },
    {
      "epoch": 0.46397188049209137,
      "grad_norm": 7.603697304148227e-05,
      "learning_rate": 2.6909518213866042e-06,
      "loss": 0.0,
      "step": 396
    },
    {
      "epoch": 0.46631517281780904,
      "grad_norm": 0.000614571908954531,
      "learning_rate": 2.6792009400705055e-06,
      "loss": 0.0023,
      "step": 398
    },
    {
      "epoch": 0.46865846514352666,
      "grad_norm": 0.046423882246017456,
      "learning_rate": 2.6674500587544067e-06,
      "loss": 0.0002,
      "step": 400
    },
    {
      "epoch": 0.4710017574692443,
      "grad_norm": 0.0005994020029902458,
      "learning_rate": 2.655699177438308e-06,
      "loss": 0.0,
      "step": 402
    },
    {
      "epoch": 0.47334504979496195,
      "grad_norm": 0.011609828099608421,
      "learning_rate": 2.6439482961222096e-06,
      "loss": 0.0001,
      "step": 404
    },
    {
      "epoch": 0.47568834212067956,
      "grad_norm": 0.007135775871574879,
      "learning_rate": 2.632197414806111e-06,
      "loss": 0.0002,
      "step": 406
    },
    {
      "epoch": 0.4780316344463972,
      "grad_norm": 0.0028773818630725145,
      "learning_rate": 2.620446533490012e-06,
      "loss": 0.0,
      "step": 408
    },
    {
      "epoch": 0.4803749267721148,
      "grad_norm": 0.13341404497623444,
      "learning_rate": 2.6086956521739132e-06,
      "loss": 0.0008,
      "step": 410
    },
    {
      "epoch": 0.48271821909783247,
      "grad_norm": 0.03130058944225311,
      "learning_rate": 2.5969447708578145e-06,
      "loss": 0.0001,
      "step": 412
    },
    {
      "epoch": 0.4850615114235501,
      "grad_norm": 0.006637818645685911,
      "learning_rate": 2.5851938895417157e-06,
      "loss": 0.0001,
      "step": 414
    },
    {
      "epoch": 0.4874048037492677,
      "grad_norm": 0.0006390800117515028,
      "learning_rate": 2.5734430082256173e-06,
      "loss": 0.0001,
      "step": 416
    },
    {
      "epoch": 0.4897480960749854,
      "grad_norm": 0.02106345072388649,
      "learning_rate": 2.5616921269095186e-06,
      "loss": 0.0002,
      "step": 418
    },
    {
      "epoch": 0.492091388400703,
      "grad_norm": 0.0009213433368131518,
      "learning_rate": 2.5499412455934198e-06,
      "loss": 0.0001,
      "step": 420
    },
    {
      "epoch": 0.4944346807264206,
      "grad_norm": 2.5962471961975098,
      "learning_rate": 2.538190364277321e-06,
      "loss": 0.1436,
      "step": 422
    },
    {
      "epoch": 0.4967779730521383,
      "grad_norm": 0.009386847727000713,
      "learning_rate": 2.5264394829612222e-06,
      "loss": 0.0001,
      "step": 424
    },
    {
      "epoch": 0.4991212653778559,
      "grad_norm": 0.01308267842978239,
      "learning_rate": 2.514688601645124e-06,
      "loss": 0.0001,
      "step": 426
    },
    {
      "epoch": 0.5014645577035736,
      "grad_norm": 0.006409250665456057,
      "learning_rate": 2.502937720329025e-06,
      "loss": 0.0,
      "step": 428
    },
    {
      "epoch": 0.5038078500292912,
      "grad_norm": 0.0018047624034807086,
      "learning_rate": 2.4911868390129263e-06,
      "loss": 0.0001,
      "step": 430
    },
    {
      "epoch": 0.5061511423550088,
      "grad_norm": 0.007056268397718668,
      "learning_rate": 2.4794359576968276e-06,
      "loss": 0.0,
      "step": 432
    },
    {
      "epoch": 0.5084944346807264,
      "grad_norm": 2.4651243686676025,
      "learning_rate": 2.4676850763807288e-06,
      "loss": 0.0245,
      "step": 434
    },
    {
      "epoch": 0.510837727006444,
      "grad_norm": 0.0025760605931282043,
      "learning_rate": 2.45593419506463e-06,
      "loss": 0.0,
      "step": 436
    },
    {
      "epoch": 0.5131810193321616,
      "grad_norm": 0.059660654515028,
      "learning_rate": 2.4441833137485312e-06,
      "loss": 0.0003,
      "step": 438
    },
    {
      "epoch": 0.5155243116578794,
      "grad_norm": 0.032668206840753555,
      "learning_rate": 2.432432432432433e-06,
      "loss": 0.0002,
      "step": 440
    },
    {
      "epoch": 0.517867603983597,
      "grad_norm": 0.002476097084581852,
      "learning_rate": 2.420681551116334e-06,
      "loss": 0.0,
      "step": 442
    },
    {
      "epoch": 0.5202108963093146,
      "grad_norm": 0.0005356927285902202,
      "learning_rate": 2.4089306698002353e-06,
      "loss": 0.0,
      "step": 444
    },
    {
      "epoch": 0.5225541886350322,
      "grad_norm": 0.01949264481663704,
      "learning_rate": 2.3971797884841366e-06,
      "loss": 0.0001,
      "step": 446
    },
    {
      "epoch": 0.5248974809607498,
      "grad_norm": 0.4609091281890869,
      "learning_rate": 2.3854289071680378e-06,
      "loss": 0.0013,
      "step": 448
    },
    {
      "epoch": 0.5272407732864675,
      "grad_norm": 0.002268969314172864,
      "learning_rate": 2.373678025851939e-06,
      "loss": 0.027,
      "step": 450
    },
    {
      "epoch": 0.5295840656121851,
      "grad_norm": 0.42679542303085327,
      "learning_rate": 2.3619271445358407e-06,
      "loss": 0.002,
      "step": 452
    },
    {
      "epoch": 0.5319273579379028,
      "grad_norm": 0.030775954946875572,
      "learning_rate": 2.350176263219742e-06,
      "loss": 0.0001,
      "step": 454
    },
    {
      "epoch": 0.5342706502636204,
      "grad_norm": 0.006208465900272131,
      "learning_rate": 2.3384253819036427e-06,
      "loss": 0.0001,
      "step": 456
    },
    {
      "epoch": 0.536613942589338,
      "grad_norm": 0.001203950378112495,
      "learning_rate": 2.3266745005875443e-06,
      "loss": 0.0,
      "step": 458
    },
    {
      "epoch": 0.5389572349150556,
      "grad_norm": 0.0013062539510428905,
      "learning_rate": 2.3149236192714456e-06,
      "loss": 0.0001,
      "step": 460
    },
    {
      "epoch": 0.5413005272407733,
      "grad_norm": 0.014242034405469894,
      "learning_rate": 2.3031727379553468e-06,
      "loss": 0.0001,
      "step": 462
    },
    {
      "epoch": 0.5436438195664909,
      "grad_norm": 0.0024558689910918474,
      "learning_rate": 2.291421856639248e-06,
      "loss": 0.0,
      "step": 464
    },
    {
      "epoch": 0.5459871118922085,
      "grad_norm": 0.006871205288916826,
      "learning_rate": 2.2796709753231492e-06,
      "loss": 0.0,
      "step": 466
    },
    {
      "epoch": 0.5483304042179262,
      "grad_norm": 0.016744021326303482,
      "learning_rate": 2.2679200940070505e-06,
      "loss": 0.0001,
      "step": 468
    },
    {
      "epoch": 0.5506736965436438,
      "grad_norm": 0.0025478950701653957,
      "learning_rate": 2.256169212690952e-06,
      "loss": 0.0,
      "step": 470
    },
    {
      "epoch": 0.5530169888693615,
      "grad_norm": 0.002553507685661316,
      "learning_rate": 2.2444183313748533e-06,
      "loss": 0.0,
      "step": 472
    },
    {
      "epoch": 0.5553602811950791,
      "grad_norm": 0.0018396044615656137,
      "learning_rate": 2.2326674500587546e-06,
      "loss": 0.0002,
      "step": 474
    },
    {
      "epoch": 0.5577035735207967,
      "grad_norm": 0.002036860678344965,
      "learning_rate": 2.2209165687426558e-06,
      "loss": 0.0,
      "step": 476
    },
    {
      "epoch": 0.5600468658465143,
      "grad_norm": 0.0024688418488949537,
      "learning_rate": 2.209165687426557e-06,
      "loss": 0.0,
      "step": 478
    },
    {
      "epoch": 0.562390158172232,
      "grad_norm": 0.0028820293955504894,
      "learning_rate": 2.1974148061104587e-06,
      "loss": 0.0001,
      "step": 480
    },
    {
      "epoch": 0.5647334504979497,
      "grad_norm": 0.00978305283933878,
      "learning_rate": 2.18566392479436e-06,
      "loss": 0.0001,
      "step": 482
    },
    {
      "epoch": 0.5670767428236673,
      "grad_norm": 0.147267147898674,
      "learning_rate": 2.173913043478261e-06,
      "loss": 0.0014,
      "step": 484
    },
    {
      "epoch": 0.5694200351493849,
      "grad_norm": 0.005025573540478945,
      "learning_rate": 2.1621621621621623e-06,
      "loss": 0.0006,
      "step": 486
    },
    {
      "epoch": 0.5717633274751025,
      "grad_norm": 0.0010051846038550138,
      "learning_rate": 2.1504112808460636e-06,
      "loss": 0.0003,
      "step": 488
    },
    {
      "epoch": 0.5741066198008201,
      "grad_norm": 0.009055075235664845,
      "learning_rate": 2.1386603995299648e-06,
      "loss": 0.0001,
      "step": 490
    },
    {
      "epoch": 0.5764499121265377,
      "grad_norm": 0.0077414545230567455,
      "learning_rate": 2.1269095182138664e-06,
      "loss": 0.0001,
      "step": 492
    },
    {
      "epoch": 0.5787932044522555,
      "grad_norm": 0.0059761228039860725,
      "learning_rate": 2.1151586368977677e-06,
      "loss": 0.0001,
      "step": 494
    },
    {
      "epoch": 0.5811364967779731,
      "grad_norm": 0.0014180493308231235,
      "learning_rate": 2.103407755581669e-06,
      "loss": 0.0,
      "step": 496
    },
    {
      "epoch": 0.5834797891036907,
      "grad_norm": 0.0022345769684761763,
      "learning_rate": 2.09165687426557e-06,
      "loss": 0.0,
      "step": 498
    },
    {
      "epoch": 0.5858230814294083,
      "grad_norm": 0.005645833443850279,
      "learning_rate": 2.0799059929494713e-06,
      "loss": 0.0001,
      "step": 500
    },
    {
      "epoch": 0.5881663737551259,
      "grad_norm": 0.011956258676946163,
      "learning_rate": 2.0681551116333726e-06,
      "loss": 0.0001,
      "step": 502
    },
    {
      "epoch": 0.5905096660808435,
      "grad_norm": 0.01774289458990097,
      "learning_rate": 2.056404230317274e-06,
      "loss": 0.0002,
      "step": 504
    },
    {
      "epoch": 0.5928529584065613,
      "grad_norm": 0.21751126646995544,
      "learning_rate": 2.0446533490011754e-06,
      "loss": 0.0012,
      "step": 506
    },
    {
      "epoch": 0.5951962507322789,
      "grad_norm": 0.00307491235435009,
      "learning_rate": 2.0329024676850762e-06,
      "loss": 0.0,
      "step": 508
    },
    {
      "epoch": 0.5975395430579965,
      "grad_norm": 0.021330738440155983,
      "learning_rate": 2.021151586368978e-06,
      "loss": 0.0002,
      "step": 510
    },
    {
      "epoch": 0.5998828353837141,
      "grad_norm": 0.020080704241991043,
      "learning_rate": 2.009400705052879e-06,
      "loss": 0.0001,
      "step": 512
    },
    {
      "epoch": 0.6022261277094317,
      "grad_norm": 0.020522406324744225,
      "learning_rate": 1.9976498237367803e-06,
      "loss": 0.0002,
      "step": 514
    },
    {
      "epoch": 0.6045694200351494,
      "grad_norm": 0.0004171329492237419,
      "learning_rate": 1.985898942420682e-06,
      "loss": 0.0,
      "step": 516
    },
    {
      "epoch": 0.606912712360867,
      "grad_norm": 0.0027696220204234123,
      "learning_rate": 1.9741480611045828e-06,
      "loss": 0.0,
      "step": 518
    },
    {
      "epoch": 0.6092560046865847,
      "grad_norm": 0.021467505022883415,
      "learning_rate": 1.9623971797884844e-06,
      "loss": 0.0002,
      "step": 520
    },
    {
      "epoch": 0.6115992970123023,
      "grad_norm": 0.011968536302447319,
      "learning_rate": 1.9506462984723856e-06,
      "loss": 0.0001,
      "step": 522
    },
    {
      "epoch": 0.6139425893380199,
      "grad_norm": 0.0011503971181809902,
      "learning_rate": 1.938895417156287e-06,
      "loss": 0.0004,
      "step": 524
    },
    {
      "epoch": 0.6162858816637375,
      "grad_norm": 0.02280554361641407,
      "learning_rate": 1.927144535840188e-06,
      "loss": 0.0002,
      "step": 526
    },
    {
      "epoch": 0.6186291739894552,
      "grad_norm": 0.008415359072387218,
      "learning_rate": 1.9153936545240893e-06,
      "loss": 0.0001,
      "step": 528
    },
    {
      "epoch": 0.6209724663151728,
      "grad_norm": 0.0024012764915823936,
      "learning_rate": 1.9036427732079908e-06,
      "loss": 0.0001,
      "step": 530
    },
    {
      "epoch": 0.6233157586408905,
      "grad_norm": 0.010776808485388756,
      "learning_rate": 1.8918918918918922e-06,
      "loss": 0.0001,
      "step": 532
    },
    {
      "epoch": 0.6256590509666081,
      "grad_norm": 0.017337538301944733,
      "learning_rate": 1.8801410105757934e-06,
      "loss": 0.0001,
      "step": 534
    },
    {
      "epoch": 0.6280023432923257,
      "grad_norm": 0.0019926901441067457,
      "learning_rate": 1.8683901292596946e-06,
      "loss": 0.0001,
      "step": 536
    },
    {
      "epoch": 0.6303456356180434,
      "grad_norm": 0.013480707071721554,
      "learning_rate": 1.856639247943596e-06,
      "loss": 0.0002,
      "step": 538
    },
    {
      "epoch": 0.632688927943761,
      "grad_norm": 0.005608106963336468,
      "learning_rate": 1.8448883666274973e-06,
      "loss": 0.0002,
      "step": 540
    },
    {
      "epoch": 0.6350322202694786,
      "grad_norm": 0.002639380283653736,
      "learning_rate": 1.8331374853113983e-06,
      "loss": 0.0001,
      "step": 542
    },
    {
      "epoch": 0.6373755125951962,
      "grad_norm": 0.0022652854677289724,
      "learning_rate": 1.8213866039953e-06,
      "loss": 0.0002,
      "step": 544
    },
    {
      "epoch": 0.6397188049209139,
      "grad_norm": 0.003624632954597473,
      "learning_rate": 1.809635722679201e-06,
      "loss": 0.0001,
      "step": 546
    },
    {
      "epoch": 0.6420620972466315,
      "grad_norm": 0.007647163700312376,
      "learning_rate": 1.7978848413631022e-06,
      "loss": 0.0004,
      "step": 548
    },
    {
      "epoch": 0.6444053895723492,
      "grad_norm": 0.012163680978119373,
      "learning_rate": 1.7861339600470036e-06,
      "loss": 0.0002,
      "step": 550
    },
    {
      "epoch": 0.6467486818980668,
      "grad_norm": 0.09023822844028473,
      "learning_rate": 1.7743830787309049e-06,
      "loss": 0.0009,
      "step": 552
    },
    {
      "epoch": 0.6490919742237844,
      "grad_norm": 0.006924999412149191,
      "learning_rate": 1.762632197414806e-06,
      "loss": 0.0001,
      "step": 554
    },
    {
      "epoch": 0.651435266549502,
      "grad_norm": 0.0006185275269672275,
      "learning_rate": 1.7508813160987075e-06,
      "loss": 0.0001,
      "step": 556
    },
    {
      "epoch": 0.6537785588752196,
      "grad_norm": 0.011605402454733849,
      "learning_rate": 1.7391304347826088e-06,
      "loss": 0.0006,
      "step": 558
    },
    {
      "epoch": 0.6561218512009374,
      "grad_norm": 0.024394473060965538,
      "learning_rate": 1.7273795534665102e-06,
      "loss": 0.0001,
      "step": 560
    },
    {
      "epoch": 0.658465143526655,
      "grad_norm": 0.023466341197490692,
      "learning_rate": 1.7156286721504114e-06,
      "loss": 0.0002,
      "step": 562
    },
    {
      "epoch": 0.6608084358523726,
      "grad_norm": 0.010153519921004772,
      "learning_rate": 1.7038777908343126e-06,
      "loss": 0.0004,
      "step": 564
    },
    {
      "epoch": 0.6631517281780902,
      "grad_norm": 0.43800845742225647,
      "learning_rate": 1.692126909518214e-06,
      "loss": 0.0012,
      "step": 566
    },
    {
      "epoch": 0.6654950205038078,
      "grad_norm": 0.008404972031712532,
      "learning_rate": 1.6803760282021153e-06,
      "loss": 0.0001,
      "step": 568
    },
    {
      "epoch": 0.6678383128295254,
      "grad_norm": 0.10615257918834686,
      "learning_rate": 1.6686251468860165e-06,
      "loss": 0.0005,
      "step": 570
    },
    {
      "epoch": 0.6701816051552432,
      "grad_norm": 0.019307592883706093,
      "learning_rate": 1.656874265569918e-06,
      "loss": 0.0003,
      "step": 572
    },
    {
      "epoch": 0.6725248974809608,
      "grad_norm": 0.012227280996739864,
      "learning_rate": 1.6451233842538192e-06,
      "loss": 0.0002,
      "step": 574
    },
    {
      "epoch": 0.6748681898066784,
      "grad_norm": 0.002821948379278183,
      "learning_rate": 1.6333725029377204e-06,
      "loss": 0.0,
      "step": 576
    },
    {
      "epoch": 0.677211482132396,
      "grad_norm": 0.010473825968801975,
      "learning_rate": 1.6216216216216219e-06,
      "loss": 0.0003,
      "step": 578
    },
    {
      "epoch": 0.6795547744581136,
      "grad_norm": 0.014046385884284973,
      "learning_rate": 1.609870740305523e-06,
      "loss": 0.0236,
      "step": 580
    },
    {
      "epoch": 0.6818980667838312,
      "grad_norm": 0.0017795696621760726,
      "learning_rate": 1.5981198589894243e-06,
      "loss": 0.0001,
      "step": 582
    },
    {
      "epoch": 0.6842413591095489,
      "grad_norm": 0.0006959863239899278,
      "learning_rate": 1.5863689776733257e-06,
      "loss": 0.0002,
      "step": 584
    },
    {
      "epoch": 0.6865846514352666,
      "grad_norm": 0.019652947783470154,
      "learning_rate": 1.574618096357227e-06,
      "loss": 0.0003,
      "step": 586
    },
    {
      "epoch": 0.6889279437609842,
      "grad_norm": 0.002340570092201233,
      "learning_rate": 1.5628672150411282e-06,
      "loss": 0.0,
      "step": 588
    },
    {
      "epoch": 0.6912712360867018,
      "grad_norm": 0.011190817691385746,
      "learning_rate": 1.5511163337250296e-06,
      "loss": 0.0002,
      "step": 590
    },
    {
      "epoch": 0.6936145284124194,
      "grad_norm": 0.001152676297351718,
      "learning_rate": 1.5393654524089308e-06,
      "loss": 0.0001,
      "step": 592
    },
    {
      "epoch": 0.6959578207381371,
      "grad_norm": 0.003393592080101371,
      "learning_rate": 1.5276145710928319e-06,
      "loss": 0.0001,
      "step": 594
    },
    {
      "epoch": 0.6983011130638547,
      "grad_norm": 0.007921353913843632,
      "learning_rate": 1.5158636897767335e-06,
      "loss": 0.0001,
      "step": 596
    },
    {
      "epoch": 0.7006444053895724,
      "grad_norm": 0.1039208471775055,
      "learning_rate": 1.5041128084606345e-06,
      "loss": 0.0002,
      "step": 598
    },
    {
      "epoch": 0.70298769771529,
      "grad_norm": 0.0011576958931982517,
      "learning_rate": 1.4923619271445362e-06,
      "loss": 0.0001,
      "step": 600
    },
    {
      "epoch": 0.7053309900410076,
      "grad_norm": 0.06407307088375092,
      "learning_rate": 1.4806110458284372e-06,
      "loss": 0.0003,
      "step": 602
    },
    {
      "epoch": 0.7076742823667252,
      "grad_norm": 0.012639104388654232,
      "learning_rate": 1.4688601645123384e-06,
      "loss": 0.0002,
      "step": 604
    },
    {
      "epoch": 0.7100175746924429,
      "grad_norm": 0.0019591290038079023,
      "learning_rate": 1.45710928319624e-06,
      "loss": 0.0068,
      "step": 606
    },
    {
      "epoch": 0.7123608670181605,
      "grad_norm": 0.0008327167597599328,
      "learning_rate": 1.445358401880141e-06,
      "loss": 0.0001,
      "step": 608
    },
    {
      "epoch": 0.7147041593438781,
      "grad_norm": 0.0013139324728399515,
      "learning_rate": 1.4336075205640423e-06,
      "loss": 0.0,
      "step": 610
    },
    {
      "epoch": 0.7170474516695958,
      "grad_norm": 0.00803992711007595,
      "learning_rate": 1.4218566392479437e-06,
      "loss": 0.0002,
      "step": 612
    },
    {
      "epoch": 0.7193907439953134,
      "grad_norm": 0.011399227194488049,
      "learning_rate": 1.410105757931845e-06,
      "loss": 0.0002,
      "step": 614
    },
    {
      "epoch": 0.7217340363210311,
      "grad_norm": 0.007171169854700565,
      "learning_rate": 1.3983548766157462e-06,
      "loss": 0.0002,
      "step": 616
    },
    {
      "epoch": 0.7240773286467487,
      "grad_norm": 0.7272996306419373,
      "learning_rate": 1.3866039952996476e-06,
      "loss": 0.0028,
      "step": 618
    },
    {
      "epoch": 0.7264206209724663,
      "grad_norm": 0.0037387118209153414,
      "learning_rate": 1.3748531139835488e-06,
      "loss": 0.0001,
      "step": 620
    },
    {
      "epoch": 0.7287639132981839,
      "grad_norm": 0.015048849396407604,
      "learning_rate": 1.36310223266745e-06,
      "loss": 0.0002,
      "step": 622
    },
    {
      "epoch": 0.7311072056239016,
      "grad_norm": 0.0023705060593783855,
      "learning_rate": 1.3513513513513515e-06,
      "loss": 0.0001,
      "step": 624
    },
    {
      "epoch": 0.7334504979496193,
      "grad_norm": 0.03966263309121132,
      "learning_rate": 1.3396004700352527e-06,
      "loss": 0.0003,
      "step": 626
    },
    {
      "epoch": 0.7357937902753369,
      "grad_norm": 0.0033043306320905685,
      "learning_rate": 1.327849588719154e-06,
      "loss": 0.0004,
      "step": 628
    },
    {
      "epoch": 0.7381370826010545,
      "grad_norm": 0.35459718108177185,
      "learning_rate": 1.3160987074030554e-06,
      "loss": 0.0034,
      "step": 630
    },
    {
      "epoch": 0.7404803749267721,
      "grad_norm": 0.016441915184259415,
      "learning_rate": 1.3043478260869566e-06,
      "loss": 0.0002,
      "step": 632
    },
    {
      "epoch": 0.7428236672524897,
      "grad_norm": 0.0045352657325565815,
      "learning_rate": 1.2925969447708578e-06,
      "loss": 0.0002,
      "step": 634
    },
    {
      "epoch": 0.7451669595782073,
      "grad_norm": 0.06311573088169098,
      "learning_rate": 1.2808460634547593e-06,
      "loss": 0.0005,
      "step": 636
    },
    {
      "epoch": 0.7475102519039251,
      "grad_norm": 0.11154340207576752,
      "learning_rate": 1.2690951821386605e-06,
      "loss": 0.0009,
      "step": 638
    },
    {
      "epoch": 0.7498535442296427,
      "grad_norm": 0.01816423609852791,
      "learning_rate": 1.257344300822562e-06,
      "loss": 0.0006,
      "step": 640
    },
    {
      "epoch": 0.7521968365553603,
      "grad_norm": 0.027273530140519142,
      "learning_rate": 1.2455934195064632e-06,
      "loss": 0.0005,
      "step": 642
    },
    {
      "epoch": 0.7545401288810779,
      "grad_norm": 0.006555743515491486,
      "learning_rate": 1.2338425381903644e-06,
      "loss": 0.0003,
      "step": 644
    },
    {
      "epoch": 0.7568834212067955,
      "grad_norm": 0.0030812753830105066,
      "learning_rate": 1.2220916568742656e-06,
      "loss": 0.0279,
      "step": 646
    },
    {
      "epoch": 0.7592267135325131,
      "grad_norm": 0.01702543906867504,
      "learning_rate": 1.210340775558167e-06,
      "loss": 0.0001,
      "step": 648
    },
    {
      "epoch": 0.7615700058582309,
      "grad_norm": 0.02607725001871586,
      "learning_rate": 1.1985898942420683e-06,
      "loss": 0.0001,
      "step": 650
    },
    {
      "epoch": 0.7639132981839485,
      "grad_norm": 0.006388965994119644,
      "learning_rate": 1.1868390129259695e-06,
      "loss": 0.0001,
      "step": 652
    },
    {
      "epoch": 0.7662565905096661,
      "grad_norm": 0.008253968320786953,
      "learning_rate": 1.175088131609871e-06,
      "loss": 0.0001,
      "step": 654
    },
    {
      "epoch": 0.7685998828353837,
      "grad_norm": 0.004699599463492632,
      "learning_rate": 1.1633372502937722e-06,
      "loss": 0.0002,
      "step": 656
    },
    {
      "epoch": 0.7709431751611013,
      "grad_norm": 0.0012458263663575053,
      "learning_rate": 1.1515863689776734e-06,
      "loss": 0.0122,
      "step": 658
    },
    {
      "epoch": 0.773286467486819,
      "grad_norm": 0.02383268252015114,
      "learning_rate": 1.1398354876615746e-06,
      "loss": 0.0003,
      "step": 660
    },
    {
      "epoch": 0.7756297598125366,
      "grad_norm": 0.015058089047670364,
      "learning_rate": 1.128084606345476e-06,
      "loss": 0.0001,
      "step": 662
    },
    {
      "epoch": 0.7779730521382543,
      "grad_norm": 0.01569475792348385,
      "learning_rate": 1.1163337250293773e-06,
      "loss": 0.0003,
      "step": 664
    },
    {
      "epoch": 0.7803163444639719,
      "grad_norm": 0.04253750294446945,
      "learning_rate": 1.1045828437132785e-06,
      "loss": 0.0002,
      "step": 666
    },
    {
      "epoch": 0.7826596367896895,
      "grad_norm": 0.015156907960772514,
      "learning_rate": 1.09283196239718e-06,
      "loss": 0.0002,
      "step": 668
    },
    {
      "epoch": 0.7850029291154071,
      "grad_norm": 0.03742622211575508,
      "learning_rate": 1.0810810810810812e-06,
      "loss": 0.0005,
      "step": 670
    },
    {
      "epoch": 0.7873462214411248,
      "grad_norm": 0.027262985706329346,
      "learning_rate": 1.0693301997649824e-06,
      "loss": 0.0002,
      "step": 672
    },
    {
      "epoch": 0.7896895137668424,
      "grad_norm": 0.007641313597559929,
      "learning_rate": 1.0575793184488838e-06,
      "loss": 0.0002,
      "step": 674
    },
    {
      "epoch": 0.79203280609256,
      "grad_norm": 0.04441560059785843,
      "learning_rate": 1.045828437132785e-06,
      "loss": 0.0005,
      "step": 676
    },
    {
      "epoch": 0.7943760984182777,
      "grad_norm": 0.020478103309869766,
      "learning_rate": 1.0340775558166863e-06,
      "loss": 0.0002,
      "step": 678
    },
    {
      "epoch": 0.7967193907439953,
      "grad_norm": 0.10936477035284042,
      "learning_rate": 1.0223266745005877e-06,
      "loss": 0.001,
      "step": 680
    },
    {
      "epoch": 0.799062683069713,
      "grad_norm": 0.01284460723400116,
      "learning_rate": 1.010575793184489e-06,
      "loss": 0.0015,
      "step": 682
    },
    {
      "epoch": 0.8014059753954306,
      "grad_norm": 0.003440434578806162,
      "learning_rate": 9.988249118683902e-07,
      "loss": 0.0,
      "step": 684
    },
    {
      "epoch": 0.8037492677211482,
      "grad_norm": 0.013081365264952183,
      "learning_rate": 9.870740305522914e-07,
      "loss": 0.0009,
      "step": 686
    },
    {
      "epoch": 0.8060925600468658,
      "grad_norm": 0.013380183838307858,
      "learning_rate": 9.753231492361928e-07,
      "loss": 0.0002,
      "step": 688
    },
    {
      "epoch": 0.8084358523725835,
      "grad_norm": 0.03771582618355751,
      "learning_rate": 9.63572267920094e-07,
      "loss": 0.0003,
      "step": 690
    },
    {
      "epoch": 0.8107791446983011,
      "grad_norm": 0.0009556732256896794,
      "learning_rate": 9.518213866039954e-07,
      "loss": 0.0005,
      "step": 692
    },
    {
      "epoch": 0.8131224370240188,
      "grad_norm": 0.0019481348572298884,
      "learning_rate": 9.400705052878967e-07,
      "loss": 0.0001,
      "step": 694
    },
    {
      "epoch": 0.8154657293497364,
      "grad_norm": 0.0021866948809474707,
      "learning_rate": 9.28319623971798e-07,
      "loss": 0.0002,
      "step": 696
    },
    {
      "epoch": 0.817809021675454,
      "grad_norm": 0.007546517997980118,
      "learning_rate": 9.165687426556992e-07,
      "loss": 0.0007,
      "step": 698
    },
    {
      "epoch": 0.8201523140011716,
      "grad_norm": 2.074432611465454,
      "learning_rate": 9.048178613396005e-07,
      "loss": 0.0251,
      "step": 700
    },
    {
      "epoch": 0.8224956063268892,
      "grad_norm": 0.003374068532139063,
      "learning_rate": 8.930669800235018e-07,
      "loss": 0.0001,
      "step": 702
    },
    {
      "epoch": 0.824838898652607,
      "grad_norm": 0.010109562426805496,
      "learning_rate": 8.81316098707403e-07,
      "loss": 0.0006,
      "step": 704
    },
    {
      "epoch": 0.8271821909783246,
      "grad_norm": 0.017352379858493805,
      "learning_rate": 8.695652173913044e-07,
      "loss": 0.0001,
      "step": 706
    },
    {
      "epoch": 0.8295254833040422,
      "grad_norm": 0.016872087493538857,
      "learning_rate": 8.578143360752057e-07,
      "loss": 0.0002,
      "step": 708
    },
    {
      "epoch": 0.8318687756297598,
      "grad_norm": 0.041937246918678284,
      "learning_rate": 8.46063454759107e-07,
      "loss": 0.0228,
      "step": 710
    },
    {
      "epoch": 0.8342120679554774,
      "grad_norm": 0.02908233553171158,
      "learning_rate": 8.343125734430083e-07,
      "loss": 0.0002,
      "step": 712
    },
    {
      "epoch": 0.836555360281195,
      "grad_norm": 0.0012463816674426198,
      "learning_rate": 8.225616921269096e-07,
      "loss": 0.0004,
      "step": 714
    },
    {
      "epoch": 0.8388986526069128,
      "grad_norm": 0.04300675913691521,
      "learning_rate": 8.108108108108109e-07,
      "loss": 0.0006,
      "step": 716
    },
    {
      "epoch": 0.8412419449326304,
      "grad_norm": 2.7622828483581543,
      "learning_rate": 7.990599294947122e-07,
      "loss": 0.149,
      "step": 718
    },
    {
      "epoch": 0.843585237258348,
      "grad_norm": 0.010049765929579735,
      "learning_rate": 7.873090481786135e-07,
      "loss": 0.0002,
      "step": 720
    },
    {
      "epoch": 0.8459285295840656,
      "grad_norm": 0.011876920238137245,
      "learning_rate": 7.755581668625148e-07,
      "loss": 0.0001,
      "step": 722
    },
    {
      "epoch": 0.8482718219097832,
      "grad_norm": 0.014826681464910507,
      "learning_rate": 7.638072855464159e-07,
      "loss": 0.0003,
      "step": 724
    },
    {
      "epoch": 0.8506151142355008,
      "grad_norm": 0.16368882358074188,
      "learning_rate": 7.520564042303173e-07,
      "loss": 0.0013,
      "step": 726
    },
    {
      "epoch": 0.8529584065612185,
      "grad_norm": 0.02603282406926155,
      "learning_rate": 7.403055229142186e-07,
      "loss": 0.0004,
      "step": 728
    },
    {
      "epoch": 0.8553016988869362,
      "grad_norm": 0.7740702629089355,
      "learning_rate": 7.2855464159812e-07,
      "loss": 0.0043,
      "step": 730
    },
    {
      "epoch": 0.8576449912126538,
      "grad_norm": 0.010226438753306866,
      "learning_rate": 7.168037602820211e-07,
      "loss": 0.0002,
      "step": 732
    },
    {
      "epoch": 0.8599882835383714,
      "grad_norm": 0.02008165791630745,
      "learning_rate": 7.050528789659225e-07,
      "loss": 0.0002,
      "step": 734
    },
    {
      "epoch": 0.862331575864089,
      "grad_norm": 0.09208586066961288,
      "learning_rate": 6.933019976498238e-07,
      "loss": 0.0008,
      "step": 736
    },
    {
      "epoch": 0.8646748681898067,
      "grad_norm": 0.01933148130774498,
      "learning_rate": 6.81551116333725e-07,
      "loss": 0.0011,
      "step": 738
    },
    {
      "epoch": 0.8670181605155243,
      "grad_norm": 0.04433580860495567,
      "learning_rate": 6.698002350176264e-07,
      "loss": 0.0003,
      "step": 740
    },
    {
      "epoch": 0.869361452841242,
      "grad_norm": 0.01631711982190609,
      "learning_rate": 6.580493537015277e-07,
      "loss": 0.0003,
      "step": 742
    },
    {
      "epoch": 0.8717047451669596,
      "grad_norm": 0.042307399213314056,
      "learning_rate": 6.462984723854289e-07,
      "loss": 0.0004,
      "step": 744
    },
    {
      "epoch": 0.8740480374926772,
      "grad_norm": 0.22414757311344147,
      "learning_rate": 6.345475910693303e-07,
      "loss": 0.0018,
      "step": 746
    },
    {
      "epoch": 0.8763913298183948,
      "grad_norm": 0.17513447999954224,
      "learning_rate": 6.227967097532316e-07,
      "loss": 0.0015,
      "step": 748
    },
    {
      "epoch": 0.8787346221441125,
      "grad_norm": 0.3218580186367035,
      "learning_rate": 6.110458284371328e-07,
      "loss": 0.0029,
      "step": 750
    },
    {
      "epoch": 0.8810779144698301,
      "grad_norm": 0.026706017553806305,
      "learning_rate": 5.992949471210341e-07,
      "loss": 0.0004,
      "step": 752
    },
    {
      "epoch": 0.8834212067955477,
      "grad_norm": 0.4114263951778412,
      "learning_rate": 5.875440658049355e-07,
      "loss": 0.0035,
      "step": 754
    },
    {
      "epoch": 0.8857644991212654,
      "grad_norm": 0.25009235739707947,
      "learning_rate": 5.757931844888367e-07,
      "loss": 0.0016,
      "step": 756
    },
    {
      "epoch": 0.888107791446983,
      "grad_norm": 1.2960833311080933,
      "learning_rate": 5.64042303172738e-07,
      "loss": 0.0059,
      "step": 758
    },
    {
      "epoch": 0.8904510837727007,
      "grad_norm": 0.28417083621025085,
      "learning_rate": 5.522914218566393e-07,
      "loss": 0.0059,
      "step": 760
    },
    {
      "epoch": 0.8927943760984183,
      "grad_norm": 0.2292051613330841,
      "learning_rate": 5.405405405405406e-07,
      "loss": 0.0015,
      "step": 762
    },
    {
      "epoch": 0.8951376684241359,
      "grad_norm": 0.012189504690468311,
      "learning_rate": 5.287896592244419e-07,
      "loss": 0.0007,
      "step": 764
    },
    {
      "epoch": 0.8974809607498535,
      "grad_norm": 0.09458251297473907,
      "learning_rate": 5.170387779083431e-07,
      "loss": 0.0004,
      "step": 766
    },
    {
      "epoch": 0.8998242530755711,
      "grad_norm": 0.027070222422480583,
      "learning_rate": 5.052878965922445e-07,
      "loss": 0.0012,
      "step": 768
    },
    {
      "epoch": 0.9021675454012889,
      "grad_norm": 0.047401878982782364,
      "learning_rate": 4.935370152761457e-07,
      "loss": 0.0003,
      "step": 770
    },
    {
      "epoch": 0.9045108377270065,
      "grad_norm": 0.06239737570285797,
      "learning_rate": 4.81786133960047e-07,
      "loss": 0.0012,
      "step": 772
    },
    {
      "epoch": 0.9068541300527241,
      "grad_norm": 2.6842846870422363,
      "learning_rate": 4.7003525264394836e-07,
      "loss": 0.1103,
      "step": 774
    },
    {
      "epoch": 0.9091974223784417,
      "grad_norm": 0.057395774871110916,
      "learning_rate": 4.582843713278496e-07,
      "loss": 0.0004,
      "step": 776
    },
    {
      "epoch": 0.9115407147041593,
      "grad_norm": 0.16248440742492676,
      "learning_rate": 4.465334900117509e-07,
      "loss": 0.0018,
      "step": 778
    },
    {
      "epoch": 0.9138840070298769,
      "grad_norm": 0.11067284643650055,
      "learning_rate": 4.347826086956522e-07,
      "loss": 0.0011,
      "step": 780
    },
    {
      "epoch": 0.9162272993555947,
      "grad_norm": 0.07208680361509323,
      "learning_rate": 4.230317273795535e-07,
      "loss": 0.0011,
      "step": 782
    },
    {
      "epoch": 0.9185705916813123,
      "grad_norm": 0.4830150604248047,
      "learning_rate": 4.112808460634548e-07,
      "loss": 0.0022,
      "step": 784
    },
    {
      "epoch": 0.9209138840070299,
      "grad_norm": 0.01794450171291828,
      "learning_rate": 3.995299647473561e-07,
      "loss": 0.0011,
      "step": 786
    },
    {
      "epoch": 0.9232571763327475,
      "grad_norm": 3.0485081672668457,
      "learning_rate": 3.877790834312574e-07,
      "loss": 0.0508,
      "step": 788
    },
    {
      "epoch": 0.9256004686584651,
      "grad_norm": 3.130112648010254,
      "learning_rate": 3.7602820211515863e-07,
      "loss": 0.0194,
      "step": 790
    },
    {
      "epoch": 0.9279437609841827,
      "grad_norm": 3.5992815494537354,
      "learning_rate": 3.6427732079906e-07,
      "loss": 0.1036,
      "step": 792
    },
    {
      "epoch": 0.9302870533099004,
      "grad_norm": 0.0751647800207138,
      "learning_rate": 3.5252643948296124e-07,
      "loss": 0.0003,
      "step": 794
    },
    {
      "epoch": 0.9326303456356181,
      "grad_norm": 0.03622612729668617,
      "learning_rate": 3.407755581668625e-07,
      "loss": 0.0011,
      "step": 796
    },
    {
      "epoch": 0.9349736379613357,
      "grad_norm": 0.22365981340408325,
      "learning_rate": 3.2902467685076385e-07,
      "loss": 0.0028,
      "step": 798
    },
    {
      "epoch": 0.9373169302870533,
      "grad_norm": 0.04666091129183769,
      "learning_rate": 3.172737955346651e-07,
      "loss": 0.0041,
      "step": 800
    },
    {
      "epoch": 0.9396602226127709,
      "grad_norm": 5.363467693328857,
      "learning_rate": 3.055229142185664e-07,
      "loss": 0.2217,
      "step": 802
    },
    {
      "epoch": 0.9420035149384886,
      "grad_norm": 0.06753694266080856,
      "learning_rate": 2.9377203290246774e-07,
      "loss": 0.0026,
      "step": 804
    },
    {
      "epoch": 0.9443468072642062,
      "grad_norm": 2.554419994354248,
      "learning_rate": 2.82021151586369e-07,
      "loss": 0.0791,
      "step": 806
    },
    {
      "epoch": 0.9466900995899239,
      "grad_norm": 0.14563411474227905,
      "learning_rate": 2.702702702702703e-07,
      "loss": 0.0208,
      "step": 808
    },
    {
      "epoch": 0.9490333919156415,
      "grad_norm": 2.30971360206604,
      "learning_rate": 2.5851938895417157e-07,
      "loss": 0.1119,
      "step": 810
    },
    {
      "epoch": 0.9513766842413591,
      "grad_norm": 4.073694229125977,
      "learning_rate": 2.4676850763807285e-07,
      "loss": 0.1057,
      "step": 812
    },
    {
      "epoch": 0.9537199765670767,
      "grad_norm": 2.3215789794921875,
      "learning_rate": 2.3501762632197418e-07,
      "loss": 0.0286,
      "step": 814
    },
    {
      "epoch": 0.9560632688927944,
      "grad_norm": 0.46727773547172546,
      "learning_rate": 2.2326674500587546e-07,
      "loss": 0.0714,
      "step": 816
    },
    {
      "epoch": 0.958406561218512,
      "grad_norm": 2.0026137828826904,
      "learning_rate": 2.1151586368977676e-07,
      "loss": 0.0455,
      "step": 818
    },
    {
      "epoch": 0.9607498535442296,
      "grad_norm": 3.2537143230438232,
      "learning_rate": 1.9976498237367804e-07,
      "loss": 0.0765,
      "step": 820
    },
    {
      "epoch": 0.9630931458699473,
      "grad_norm": 3.485633134841919,
      "learning_rate": 1.8801410105757932e-07,
      "loss": 0.0493,
      "step": 822
    },
    {
      "epoch": 0.9654364381956649,
      "grad_norm": 2.769423246383667,
      "learning_rate": 1.7626321974148062e-07,
      "loss": 0.0602,
      "step": 824
    },
    {
      "epoch": 0.9677797305213826,
      "grad_norm": 2.236210823059082,
      "learning_rate": 1.6451233842538192e-07,
      "loss": 0.1404,
      "step": 826
    },
    {
      "epoch": 0.9701230228471002,
      "grad_norm": 0.06197360157966614,
      "learning_rate": 1.527614571092832e-07,
      "loss": 0.0472,
      "step": 828
    },
    {
      "epoch": 0.9724663151728178,
      "grad_norm": 0.8206185698509216,
      "learning_rate": 1.410105757931845e-07,
      "loss": 0.0686,
      "step": 830
    },
    {
      "epoch": 0.9748096074985354,
      "grad_norm": 2.434030771255493,
      "learning_rate": 1.2925969447708578e-07,
      "loss": 0.1322,
      "step": 832
    },
    {
      "epoch": 0.9771528998242531,
      "grad_norm": 0.03143630549311638,
      "learning_rate": 1.1750881316098709e-07,
      "loss": 0.1134,
      "step": 834
    },
    {
      "epoch": 0.9794961921499707,
      "grad_norm": 0.1770186424255371,
      "learning_rate": 1.0575793184488838e-07,
      "loss": 0.0011,
      "step": 836
    },
    {
      "epoch": 0.9818394844756884,
      "grad_norm": 6.03350830078125,
      "learning_rate": 9.400705052878966e-08,
      "loss": 0.4193,
      "step": 838
    },
    {
      "epoch": 0.984182776801406,
      "grad_norm": 4.842612266540527,
      "learning_rate": 8.225616921269096e-08,
      "loss": 0.0951,
      "step": 840
    },
    {
      "epoch": 0.9865260691271236,
      "grad_norm": 3.111945629119873,
      "learning_rate": 7.050528789659225e-08,
      "loss": 0.1375,
      "step": 842
    },
    {
      "epoch": 0.9888693614528412,
      "grad_norm": 3.4468753337860107,
      "learning_rate": 5.8754406580493544e-08,
      "loss": 0.157,
      "step": 844
    },
    {
      "epoch": 0.9912126537785588,
      "grad_norm": 5.563467502593994,
      "learning_rate": 4.700352526439483e-08,
      "loss": 0.1989,
      "step": 846
    },
    {
      "epoch": 0.9935559461042766,
      "grad_norm": 0.20900146663188934,
      "learning_rate": 3.5252643948296127e-08,
      "loss": 0.169,
      "step": 848
    },
    {
      "epoch": 0.9958992384299942,
      "grad_norm": 2.651283025741577,
      "learning_rate": 2.3501762632197414e-08,
      "loss": 0.0203,
      "step": 850
    },
    {
      "epoch": 0.9982425307557118,
      "grad_norm": 3.192451000213623,
      "learning_rate": 1.1750881316098707e-08,
      "loss": 0.0786,
      "step": 852
    }
  ],
  "logging_steps": 2,
  "max_steps": 853,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 20000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}