{
  "best_metric": 1.36418879032135,
  "best_model_checkpoint": "miner_id_24/checkpoint-50",
  "epoch": 3.0,
  "eval_steps": 50,
  "global_step": 177,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01694915254237288,
      "grad_norm": 1.4696528911590576,
      "learning_rate": 1e-05,
      "loss": 4.4767,
      "step": 1
    },
    {
      "epoch": 0.01694915254237288,
      "eval_loss": 1.5105643272399902,
      "eval_runtime": 13.1638,
      "eval_samples_per_second": 7.597,
      "eval_steps_per_second": 1.899,
      "step": 1
    },
    {
      "epoch": 0.03389830508474576,
      "grad_norm": 1.0327868461608887,
      "learning_rate": 2e-05,
      "loss": 5.5338,
      "step": 2
    },
    {
      "epoch": 0.05084745762711865,
      "grad_norm": 0.8375867009162903,
      "learning_rate": 3e-05,
      "loss": 6.7695,
      "step": 3
    },
    {
      "epoch": 0.06779661016949153,
      "grad_norm": 1.0288580656051636,
      "learning_rate": 4e-05,
      "loss": 6.0015,
      "step": 4
    },
    {
      "epoch": 0.0847457627118644,
      "grad_norm": 1.235273003578186,
      "learning_rate": 5e-05,
      "loss": 6.1554,
      "step": 5
    },
    {
      "epoch": 0.1016949152542373,
      "grad_norm": 1.609580636024475,
      "learning_rate": 6e-05,
      "loss": 6.4157,
      "step": 6
    },
    {
      "epoch": 0.11864406779661017,
      "grad_norm": 1.5870156288146973,
      "learning_rate": 7e-05,
      "loss": 6.3081,
      "step": 7
    },
    {
      "epoch": 0.13559322033898305,
      "grad_norm": 1.571000099182129,
      "learning_rate": 8e-05,
      "loss": 6.3219,
      "step": 8
    },
    {
      "epoch": 0.15254237288135594,
      "grad_norm": 1.3622900247573853,
      "learning_rate": 9e-05,
      "loss": 6.145,
      "step": 9
    },
    {
      "epoch": 0.1694915254237288,
      "grad_norm": 1.6992497444152832,
      "learning_rate": 0.0001,
      "loss": 5.4642,
      "step": 10
    },
    {
      "epoch": 0.1864406779661017,
      "grad_norm": 1.4771649837493896,
      "learning_rate": 9.999115304121457e-05,
      "loss": 5.3796,
      "step": 11
    },
    {
      "epoch": 0.2033898305084746,
      "grad_norm": 2.005577325820923,
      "learning_rate": 9.996461529560553e-05,
      "loss": 6.4416,
      "step": 12
    },
    {
      "epoch": 0.22033898305084745,
      "grad_norm": 2.1262521743774414,
      "learning_rate": 9.992039615430648e-05,
      "loss": 6.1819,
      "step": 13
    },
    {
      "epoch": 0.23728813559322035,
      "grad_norm": 2.5400421619415283,
      "learning_rate": 9.985851126551428e-05,
      "loss": 5.9766,
      "step": 14
    },
    {
      "epoch": 0.2542372881355932,
      "grad_norm": 2.724677562713623,
      "learning_rate": 9.977898252895134e-05,
      "loss": 4.8459,
      "step": 15
    },
    {
      "epoch": 0.2711864406779661,
      "grad_norm": 1.9490002393722534,
      "learning_rate": 9.968183808811586e-05,
      "loss": 4.5774,
      "step": 16
    },
    {
      "epoch": 0.288135593220339,
      "grad_norm": 1.7252360582351685,
      "learning_rate": 9.95671123203224e-05,
      "loss": 6.9357,
      "step": 17
    },
    {
      "epoch": 0.3050847457627119,
      "grad_norm": 1.5308066606521606,
      "learning_rate": 9.943484582453653e-05,
      "loss": 6.6148,
      "step": 18
    },
    {
      "epoch": 0.3220338983050847,
      "grad_norm": 1.300583839416504,
      "learning_rate": 9.928508540700774e-05,
      "loss": 7.4276,
      "step": 19
    },
    {
      "epoch": 0.3389830508474576,
      "grad_norm": 1.0917631387710571,
      "learning_rate": 9.911788406470569e-05,
      "loss": 5.9673,
      "step": 20
    },
    {
      "epoch": 0.3559322033898305,
      "grad_norm": 1.2227107286453247,
      "learning_rate": 9.893330096656574e-05,
      "loss": 5.8025,
      "step": 21
    },
    {
      "epoch": 0.3728813559322034,
      "grad_norm": 1.5362129211425781,
      "learning_rate": 9.873140143255036e-05,
      "loss": 6.5079,
      "step": 22
    },
    {
      "epoch": 0.3898305084745763,
      "grad_norm": 1.2382031679153442,
      "learning_rate": 9.85122569105338e-05,
      "loss": 5.7189,
      "step": 23
    },
    {
      "epoch": 0.4067796610169492,
      "grad_norm": 1.4853235483169556,
      "learning_rate": 9.827594495101823e-05,
      "loss": 6.06,
      "step": 24
    },
    {
      "epoch": 0.423728813559322,
      "grad_norm": 1.4762415885925293,
      "learning_rate": 9.802254917969032e-05,
      "loss": 5.8676,
      "step": 25
    },
    {
      "epoch": 0.4406779661016949,
      "grad_norm": 1.4179600477218628,
      "learning_rate": 9.775215926782788e-05,
      "loss": 5.3958,
      "step": 26
    },
    {
      "epoch": 0.4576271186440678,
      "grad_norm": 1.6792457103729248,
      "learning_rate": 9.746487090056713e-05,
      "loss": 5.2567,
      "step": 27
    },
    {
      "epoch": 0.4745762711864407,
      "grad_norm": 2.73538875579834,
      "learning_rate": 9.716078574304189e-05,
      "loss": 5.8031,
      "step": 28
    },
    {
      "epoch": 0.4915254237288136,
      "grad_norm": 1.1012799739837646,
      "learning_rate": 9.684001140440639e-05,
      "loss": 4.4475,
      "step": 29
    },
    {
      "epoch": 0.5084745762711864,
      "grad_norm": 0.7912802696228027,
      "learning_rate": 9.650266139975474e-05,
      "loss": 5.1582,
      "step": 30
    },
    {
      "epoch": 0.5254237288135594,
      "grad_norm": 0.8357610702514648,
      "learning_rate": 9.614885510995047e-05,
      "loss": 5.9797,
      "step": 31
    },
    {
      "epoch": 0.5423728813559322,
      "grad_norm": 0.8933336734771729,
      "learning_rate": 9.577871773938011e-05,
      "loss": 5.8404,
      "step": 32
    },
    {
      "epoch": 0.559322033898305,
      "grad_norm": 0.9425844550132751,
      "learning_rate": 9.539238027164619e-05,
      "loss": 6.3711,
      "step": 33
    },
    {
      "epoch": 0.576271186440678,
      "grad_norm": 0.8766078352928162,
      "learning_rate": 9.498997942321483e-05,
      "loss": 6.1781,
      "step": 34
    },
    {
      "epoch": 0.5932203389830508,
      "grad_norm": 1.1381494998931885,
      "learning_rate": 9.457165759503493e-05,
      "loss": 5.9204,
      "step": 35
    },
    {
      "epoch": 0.6101694915254238,
      "grad_norm": 0.9810196757316589,
      "learning_rate": 9.413756282214537e-05,
      "loss": 5.3107,
      "step": 36
    },
    {
      "epoch": 0.6271186440677966,
      "grad_norm": 0.9835663437843323,
      "learning_rate": 9.368784872128878e-05,
      "loss": 5.4947,
      "step": 37
    },
    {
      "epoch": 0.6440677966101694,
      "grad_norm": 1.1894506216049194,
      "learning_rate": 9.322267443654972e-05,
      "loss": 5.6231,
      "step": 38
    },
    {
      "epoch": 0.6610169491525424,
      "grad_norm": 1.3286501169204712,
      "learning_rate": 9.274220458303727e-05,
      "loss": 5.8462,
      "step": 39
    },
    {
      "epoch": 0.6779661016949152,
      "grad_norm": 1.3058652877807617,
      "learning_rate": 9.224660918863104e-05,
      "loss": 5.3328,
      "step": 40
    },
    {
      "epoch": 0.6949152542372882,
      "grad_norm": 1.677504301071167,
      "learning_rate": 9.173606363381219e-05,
      "loss": 5.7914,
      "step": 41
    },
    {
      "epoch": 0.711864406779661,
      "grad_norm": 2.292499303817749,
      "learning_rate": 9.121074858959997e-05,
      "loss": 5.8802,
      "step": 42
    },
    {
      "epoch": 0.7288135593220338,
      "grad_norm": 0.9735826849937439,
      "learning_rate": 9.067084995361623e-05,
      "loss": 4.7076,
      "step": 43
    },
    {
      "epoch": 0.7457627118644068,
      "grad_norm": 0.8391918540000916,
      "learning_rate": 9.011655878430019e-05,
      "loss": 4.8208,
      "step": 44
    },
    {
      "epoch": 0.7627118644067796,
      "grad_norm": 0.7652661800384521,
      "learning_rate": 8.954807123329704e-05,
      "loss": 6.4505,
      "step": 45
    },
    {
      "epoch": 0.7796610169491526,
      "grad_norm": 0.9724342226982117,
      "learning_rate": 8.896558847604414e-05,
      "loss": 5.5774,
      "step": 46
    },
    {
      "epoch": 0.7966101694915254,
      "grad_norm": 0.7980533838272095,
      "learning_rate": 8.836931664057935e-05,
      "loss": 5.8932,
      "step": 47
    },
    {
      "epoch": 0.8135593220338984,
      "grad_norm": 0.9073213338851929,
      "learning_rate": 8.775946673459681e-05,
      "loss": 5.1781,
      "step": 48
    },
    {
      "epoch": 0.8305084745762712,
      "grad_norm": 1.0768805742263794,
      "learning_rate": 8.713625457077585e-05,
      "loss": 5.388,
      "step": 49
    },
    {
      "epoch": 0.847457627118644,
      "grad_norm": 1.1347218751907349,
      "learning_rate": 8.649990069040961e-05,
      "loss": 5.9919,
      "step": 50
    },
    {
      "epoch": 0.847457627118644,
      "eval_loss": 1.36418879032135,
      "eval_runtime": 13.3724,
      "eval_samples_per_second": 7.478,
      "eval_steps_per_second": 1.87,
      "step": 50
    },
    {
      "epoch": 0.864406779661017,
      "grad_norm": 1.202120065689087,
      "learning_rate": 8.585063028536016e-05,
      "loss": 6.3301,
      "step": 51
    },
    {
      "epoch": 0.8813559322033898,
      "grad_norm": 1.2182761430740356,
      "learning_rate": 8.518867311836808e-05,
      "loss": 5.3969,
      "step": 52
    },
    {
      "epoch": 0.8983050847457628,
      "grad_norm": 1.1987417936325073,
      "learning_rate": 8.451426344174433e-05,
      "loss": 5.7186,
      "step": 53
    },
    {
      "epoch": 0.9152542372881356,
      "grad_norm": 1.6020171642303467,
      "learning_rate": 8.382763991447344e-05,
      "loss": 5.886,
      "step": 54
    },
    {
      "epoch": 0.9322033898305084,
      "grad_norm": 1.5172274112701416,
      "learning_rate": 8.312904551775731e-05,
      "loss": 5.7023,
      "step": 55
    },
    {
      "epoch": 0.9491525423728814,
      "grad_norm": 2.7889416217803955,
      "learning_rate": 8.241872746902935e-05,
      "loss": 5.5785,
      "step": 56
    },
    {
      "epoch": 0.9661016949152542,
      "grad_norm": 0.7752204537391663,
      "learning_rate": 8.169693713446959e-05,
      "loss": 4.9944,
      "step": 57
    },
    {
      "epoch": 0.9830508474576272,
      "grad_norm": 0.9456698298454285,
      "learning_rate": 8.096392994005177e-05,
      "loss": 5.7277,
      "step": 58
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.7965518236160278,
      "learning_rate": 8.021996528115335e-05,
      "loss": 6.2417,
      "step": 59
    },
    {
      "epoch": 1.0169491525423728,
      "grad_norm": 0.7447935938835144,
      "learning_rate": 7.946530643076138e-05,
      "loss": 4.3768,
      "step": 60
    },
    {
      "epoch": 1.0338983050847457,
      "grad_norm": 0.7535466551780701,
      "learning_rate": 7.870022044630569e-05,
      "loss": 4.7865,
      "step": 61
    },
    {
      "epoch": 1.0508474576271187,
      "grad_norm": 0.7265345454216003,
      "learning_rate": 7.792497807515317e-05,
      "loss": 6.1749,
      "step": 62
    },
    {
      "epoch": 1.0677966101694916,
      "grad_norm": 0.8258265852928162,
      "learning_rate": 7.713985365879606e-05,
      "loss": 5.8394,
      "step": 63
    },
    {
      "epoch": 1.0847457627118644,
      "grad_norm": 0.8342085480690002,
      "learning_rate": 7.63451250357685e-05,
      "loss": 5.6838,
      "step": 64
    },
    {
      "epoch": 1.1016949152542372,
      "grad_norm": 0.8415143489837646,
      "learning_rate": 7.55410734433254e-05,
      "loss": 5.0008,
      "step": 65
    },
    {
      "epoch": 1.11864406779661,
      "grad_norm": 0.8243486881256104,
      "learning_rate": 7.472798341791877e-05,
      "loss": 5.2983,
      "step": 66
    },
    {
      "epoch": 1.1355932203389831,
      "grad_norm": 0.9714155793190002,
      "learning_rate": 7.390614269450634e-05,
      "loss": 5.0533,
      "step": 67
    },
    {
      "epoch": 1.152542372881356,
      "grad_norm": 1.1003835201263428,
      "learning_rate": 7.307584210472844e-05,
      "loss": 5.4401,
      "step": 68
    },
    {
      "epoch": 1.1694915254237288,
      "grad_norm": 1.0256558656692505,
      "learning_rate": 7.223737547398898e-05,
      "loss": 4.8704,
      "step": 69
    },
    {
      "epoch": 1.1864406779661016,
      "grad_norm": 1.084803581237793,
      "learning_rate": 7.139103951747695e-05,
      "loss": 4.6861,
      "step": 70
    },
    {
      "epoch": 1.2033898305084745,
      "grad_norm": 1.2507531642913818,
      "learning_rate": 7.053713373516538e-05,
      "loss": 4.9667,
      "step": 71
    },
    {
      "epoch": 1.2203389830508475,
      "grad_norm": 1.4446933269500732,
      "learning_rate": 6.967596030582478e-05,
      "loss": 5.1595,
      "step": 72
    },
    {
      "epoch": 1.2372881355932204,
      "grad_norm": 2.123182773590088,
      "learning_rate": 6.880782398008862e-05,
      "loss": 5.2566,
      "step": 73
    },
    {
      "epoch": 1.2542372881355932,
      "grad_norm": 1.3967437744140625,
      "learning_rate": 6.793303197260864e-05,
      "loss": 4.3797,
      "step": 74
    },
    {
      "epoch": 1.271186440677966,
      "grad_norm": 1.277904987335205,
      "learning_rate": 6.70518938533383e-05,
      "loss": 4.9218,
      "step": 75
    },
    {
      "epoch": 1.288135593220339,
      "grad_norm": 0.93733811378479,
      "learning_rate": 6.616472143798261e-05,
      "loss": 6.0773,
      "step": 76
    },
    {
      "epoch": 1.305084745762712,
      "grad_norm": 1.0279090404510498,
      "learning_rate": 6.527182867765332e-05,
      "loss": 5.305,
      "step": 77
    },
    {
      "epoch": 1.3220338983050848,
      "grad_norm": 0.8937672972679138,
      "learning_rate": 6.437353154776849e-05,
      "loss": 5.7655,
      "step": 78
    },
    {
      "epoch": 1.3389830508474576,
      "grad_norm": 1.0808204412460327,
      "learning_rate": 6.347014793623547e-05,
      "loss": 5.2614,
      "step": 79
    },
    {
      "epoch": 1.3559322033898304,
      "grad_norm": 1.2765063047409058,
      "learning_rate": 6.256199753095745e-05,
      "loss": 5.5553,
      "step": 80
    },
    {
      "epoch": 1.3728813559322033,
      "grad_norm": 1.3419365882873535,
      "learning_rate": 6.164940170670266e-05,
      "loss": 4.9784,
      "step": 81
    },
    {
      "epoch": 1.3898305084745763,
      "grad_norm": 1.3818930387496948,
      "learning_rate": 6.0732683411376935e-05,
      "loss": 4.097,
      "step": 82
    },
    {
      "epoch": 1.4067796610169492,
      "grad_norm": 1.4276171922683716,
      "learning_rate": 5.98121670517393e-05,
      "loss": 5.0847,
      "step": 83
    },
    {
      "epoch": 1.423728813559322,
      "grad_norm": 1.5427664518356323,
      "learning_rate": 5.8888178378601565e-05,
      "loss": 4.8216,
      "step": 84
    },
    {
      "epoch": 1.4406779661016949,
      "grad_norm": 1.644025444984436,
      "learning_rate": 5.796104437155213e-05,
      "loss": 4.8768,
      "step": 85
    },
    {
      "epoch": 1.457627118644068,
      "grad_norm": 1.620620846748352,
      "learning_rate": 5.7031093123244925e-05,
      "loss": 3.9242,
      "step": 86
    },
    {
      "epoch": 1.4745762711864407,
      "grad_norm": 2.355064868927002,
      "learning_rate": 5.6098653723294604e-05,
      "loss": 4.6186,
      "step": 87
    },
    {
      "epoch": 1.4915254237288136,
      "grad_norm": 1.1192209720611572,
      "learning_rate": 5.516405614181883e-05,
      "loss": 4.0917,
      "step": 88
    },
    {
      "epoch": 1.5084745762711864,
      "grad_norm": 0.9363360404968262,
      "learning_rate": 5.4227631112668955e-05,
      "loss": 4.8794,
      "step": 89
    },
    {
      "epoch": 1.5254237288135593,
      "grad_norm": 0.8636099100112915,
      "learning_rate": 5.3289710016390535e-05,
      "loss": 5.7161,
      "step": 90
    },
    {
      "epoch": 1.542372881355932,
      "grad_norm": 1.1945964097976685,
      "learning_rate": 5.2350624762954884e-05,
      "loss": 5.3209,
      "step": 91
    },
    {
      "epoch": 1.559322033898305,
      "grad_norm": 1.0714601278305054,
      "learning_rate": 5.14107076743033e-05,
      "loss": 6.0142,
      "step": 92
    },
    {
      "epoch": 1.576271186440678,
      "grad_norm": 1.1999359130859375,
      "learning_rate": 5.047029136674563e-05,
      "loss": 4.7506,
      "step": 93
    },
    {
      "epoch": 1.5932203389830508,
      "grad_norm": 1.3246750831604004,
      "learning_rate": 4.95297086332544e-05,
      "loss": 4.9699,
      "step": 94
    },
    {
      "epoch": 1.6101694915254239,
      "grad_norm": 1.5049471855163574,
      "learning_rate": 4.858929232569671e-05,
      "loss": 5.4011,
      "step": 95
    },
    {
      "epoch": 1.6271186440677967,
      "grad_norm": 1.5098161697387695,
      "learning_rate": 4.7649375237045135e-05,
      "loss": 4.4637,
      "step": 96
    },
    {
      "epoch": 1.6440677966101696,
      "grad_norm": 1.5312010049819946,
      "learning_rate": 4.671028998360947e-05,
      "loss": 5.3261,
      "step": 97
    },
    {
      "epoch": 1.6610169491525424,
      "grad_norm": 1.6302275657653809,
      "learning_rate": 4.577236888733105e-05,
      "loss": 4.994,
      "step": 98
    },
    {
      "epoch": 1.6779661016949152,
      "grad_norm": 1.836314082145691,
      "learning_rate": 4.483594385818118e-05,
      "loss": 4.639,
      "step": 99
    },
    {
      "epoch": 1.694915254237288,
      "grad_norm": 1.9872089624404907,
      "learning_rate": 4.39013462767054e-05,
      "loss": 4.134,
      "step": 100
    },
    {
      "epoch": 1.694915254237288,
      "eval_loss": 1.3669192790985107,
      "eval_runtime": 12.95,
      "eval_samples_per_second": 7.722,
      "eval_steps_per_second": 1.93,
      "step": 100
    },
    {
      "epoch": 1.711864406779661,
      "grad_norm": 2.7113258838653564,
      "learning_rate": 4.29689068767551e-05,
      "loss": 4.2686,
      "step": 101
    },
    {
      "epoch": 1.7288135593220337,
      "grad_norm": 0.9449647068977356,
      "learning_rate": 4.203895562844789e-05,
      "loss": 3.9707,
      "step": 102
    },
    {
      "epoch": 1.7457627118644068,
      "grad_norm": 0.9212586879730225,
      "learning_rate": 4.1111821621398446e-05,
      "loss": 3.2945,
      "step": 103
    },
    {
      "epoch": 1.7627118644067796,
      "grad_norm": 1.0789300203323364,
      "learning_rate": 4.0187832948260705e-05,
      "loss": 5.9138,
      "step": 104
    },
    {
      "epoch": 1.7796610169491527,
      "grad_norm": 1.150065302848816,
      "learning_rate": 3.926731658862307e-05,
      "loss": 6.0901,
      "step": 105
    },
    {
      "epoch": 1.7966101694915255,
      "grad_norm": 1.1871299743652344,
      "learning_rate": 3.835059829329735e-05,
      "loss": 5.6111,
      "step": 106
    },
    {
      "epoch": 1.8135593220338984,
      "grad_norm": 1.2099858522415161,
      "learning_rate": 3.7438002469042565e-05,
      "loss": 5.4234,
      "step": 107
    },
    {
      "epoch": 1.8305084745762712,
      "grad_norm": 1.2399576902389526,
      "learning_rate": 3.6529852063764545e-05,
      "loss": 4.7256,
      "step": 108
    },
    {
      "epoch": 1.847457627118644,
      "grad_norm": 1.3255728483200073,
      "learning_rate": 3.562646845223153e-05,
      "loss": 5.547,
      "step": 109
    },
    {
      "epoch": 1.8644067796610169,
      "grad_norm": 1.4329326152801514,
      "learning_rate": 3.4728171322346694e-05,
      "loss": 4.7792,
      "step": 110
    },
    {
      "epoch": 1.8813559322033897,
      "grad_norm": 1.6157045364379883,
      "learning_rate": 3.38352785620174e-05,
      "loss": 3.6202,
      "step": 111
    },
    {
      "epoch": 1.8983050847457628,
      "grad_norm": 1.7625586986541748,
      "learning_rate": 3.29481061466617e-05,
      "loss": 4.6673,
      "step": 112
    },
    {
      "epoch": 1.9152542372881356,
      "grad_norm": 1.7876002788543701,
      "learning_rate": 3.2066968027391374e-05,
      "loss": 4.3518,
      "step": 113
    },
    {
      "epoch": 1.9322033898305084,
      "grad_norm": 2.519547700881958,
      "learning_rate": 3.119217601991139e-05,
      "loss": 4.6088,
      "step": 114
    },
    {
      "epoch": 1.9491525423728815,
      "grad_norm": 2.9257960319519043,
      "learning_rate": 3.0324039694175233e-05,
      "loss": 4.1428,
      "step": 115
    },
    {
      "epoch": 1.9661016949152543,
      "grad_norm": 1.1026883125305176,
      "learning_rate": 2.946286626483463e-05,
      "loss": 5.0716,
      "step": 116
    },
    {
      "epoch": 1.9830508474576272,
      "grad_norm": 1.3919453620910645,
      "learning_rate": 2.8608960482523056e-05,
      "loss": 4.6352,
      "step": 117
    },
    {
      "epoch": 2.0,
      "grad_norm": 2.9073708057403564,
      "learning_rate": 2.7762624526011038e-05,
      "loss": 4.5153,
      "step": 118
    },
    {
      "epoch": 2.016949152542373,
      "grad_norm": 0.899066150188446,
      "learning_rate": 2.6924157895271563e-05,
      "loss": 3.9393,
      "step": 119
    },
    {
      "epoch": 2.0338983050847457,
      "grad_norm": 1.0271446704864502,
      "learning_rate": 2.6093857305493664e-05,
      "loss": 4.5126,
      "step": 120
    },
    {
      "epoch": 2.0508474576271185,
      "grad_norm": 1.00508451461792,
      "learning_rate": 2.5272016582081236e-05,
      "loss": 5.8784,
      "step": 121
    },
    {
      "epoch": 2.0677966101694913,
      "grad_norm": 1.110703468322754,
      "learning_rate": 2.4458926556674615e-05,
      "loss": 5.6166,
      "step": 122
    },
    {
      "epoch": 2.084745762711864,
      "grad_norm": 1.074708342552185,
      "learning_rate": 2.3654874964231518e-05,
      "loss": 5.4747,
      "step": 123
    },
    {
      "epoch": 2.1016949152542375,
      "grad_norm": 1.5243916511535645,
      "learning_rate": 2.2860146341203937e-05,
      "loss": 4.2634,
      "step": 124
    },
    {
      "epoch": 2.1186440677966103,
      "grad_norm": 1.277087926864624,
      "learning_rate": 2.207502192484685e-05,
      "loss": 4.4356,
      "step": 125
    },
    {
      "epoch": 2.135593220338983,
      "grad_norm": 1.529086709022522,
      "learning_rate": 2.1299779553694323e-05,
      "loss": 4.89,
      "step": 126
    },
    {
      "epoch": 2.152542372881356,
      "grad_norm": 1.7537826299667358,
      "learning_rate": 2.053469356923865e-05,
      "loss": 3.6175,
      "step": 127
    },
    {
      "epoch": 2.169491525423729,
      "grad_norm": 1.7214950323104858,
      "learning_rate": 1.978003471884665e-05,
      "loss": 3.5505,
      "step": 128
    },
    {
      "epoch": 2.1864406779661016,
      "grad_norm": 2.21970534324646,
      "learning_rate": 1.9036070059948252e-05,
      "loss": 3.9899,
      "step": 129
    },
    {
      "epoch": 2.2033898305084745,
      "grad_norm": 2.214458465576172,
      "learning_rate": 1.8303062865530406e-05,
      "loss": 3.5737,
      "step": 130
    },
    {
      "epoch": 2.2203389830508473,
      "grad_norm": 2.502815008163452,
      "learning_rate": 1.7581272530970667e-05,
      "loss": 3.6376,
      "step": 131
    },
    {
      "epoch": 2.23728813559322,
      "grad_norm": 3.225670576095581,
      "learning_rate": 1.6870954482242707e-05,
      "loss": 3.1241,
      "step": 132
    },
    {
      "epoch": 2.2542372881355934,
      "grad_norm": 1.0706570148468018,
      "learning_rate": 1.6172360085526565e-05,
      "loss": 3.9706,
      "step": 133
    },
    {
      "epoch": 2.2711864406779663,
      "grad_norm": 1.3153660297393799,
      "learning_rate": 1.5485736558255697e-05,
      "loss": 4.0122,
      "step": 134
    },
    {
      "epoch": 2.288135593220339,
      "grad_norm": 1.1887065172195435,
      "learning_rate": 1.4811326881631937e-05,
      "loss": 4.628,
      "step": 135
    },
    {
      "epoch": 2.305084745762712,
      "grad_norm": 2.481553316116333,
      "learning_rate": 1.4149369714639853e-05,
      "loss": 5.9046,
      "step": 136
    },
    {
      "epoch": 2.3220338983050848,
      "grad_norm": 1.3416173458099365,
      "learning_rate": 1.3500099309590397e-05,
      "loss": 5.0805,
      "step": 137
    },
    {
      "epoch": 2.3389830508474576,
      "grad_norm": 1.593565583229065,
      "learning_rate": 1.2863745429224144e-05,
      "loss": 5.2141,
      "step": 138
    },
    {
      "epoch": 2.3559322033898304,
      "grad_norm": 1.622098684310913,
      "learning_rate": 1.2240533265403198e-05,
      "loss": 4.5864,
      "step": 139
    },
    {
      "epoch": 2.3728813559322033,
      "grad_norm": 1.7163193225860596,
      "learning_rate": 1.1630683359420652e-05,
      "loss": 4.3987,
      "step": 140
    },
    {
      "epoch": 2.389830508474576,
      "grad_norm": 1.8327656984329224,
      "learning_rate": 1.103441152395588e-05,
      "loss": 3.9806,
      "step": 141
    },
    {
      "epoch": 2.406779661016949,
      "grad_norm": 2.0839500427246094,
      "learning_rate": 1.0451928766702979e-05,
      "loss": 3.9348,
      "step": 142
    },
    {
      "epoch": 2.423728813559322,
      "grad_norm": 2.192836046218872,
      "learning_rate": 9.883441215699823e-06,
      "loss": 3.6555,
      "step": 143
    },
    {
      "epoch": 2.440677966101695,
      "grad_norm": 2.416853427886963,
      "learning_rate": 9.329150046383772e-06,
      "loss": 3.449,
      "step": 144
    },
    {
      "epoch": 2.457627118644068,
      "grad_norm": 2.738199472427368,
      "learning_rate": 8.789251410400023e-06,
      "loss": 3.2119,
      "step": 145
    },
    {
      "epoch": 2.4745762711864407,
      "grad_norm": 4.378939628601074,
      "learning_rate": 8.263936366187824e-06,
      "loss": 2.9737,
      "step": 146
    },
    {
      "epoch": 2.4915254237288136,
      "grad_norm": 1.067427635192871,
      "learning_rate": 7.753390811368971e-06,
      "loss": 3.7761,
      "step": 147
    },
    {
      "epoch": 2.5084745762711864,
      "grad_norm": 1.5090866088867188,
      "learning_rate": 7.257795416962753e-06,
      "loss": 4.0601,
      "step": 148
    },
    {
      "epoch": 2.5254237288135593,
      "grad_norm": 1.1970852613449097,
      "learning_rate": 6.777325563450282e-06,
      "loss": 5.6451,
      "step": 149
    },
    {
      "epoch": 2.542372881355932,
      "grad_norm": 1.320055365562439,
      "learning_rate": 6.312151278711237e-06,
      "loss": 4.9721,
      "step": 150
    },
    {
      "epoch": 2.542372881355932,
      "eval_loss": 1.41791570186615,
      "eval_runtime": 13.401,
      "eval_samples_per_second": 7.462,
      "eval_steps_per_second": 1.866,
      "step": 150
    },
    {
      "epoch": 2.559322033898305,
      "grad_norm": 1.5780824422836304,
      "learning_rate": 5.86243717785463e-06,
      "loss": 5.1087,
      "step": 151
    },
    {
      "epoch": 2.576271186440678,
      "grad_norm": 1.6419761180877686,
      "learning_rate": 5.428342404965076e-06,
      "loss": 3.9339,
      "step": 152
    },
    {
      "epoch": 2.593220338983051,
      "grad_norm": 1.5951154232025146,
      "learning_rate": 5.010020576785174e-06,
      "loss": 4.9512,
      "step": 153
    },
    {
      "epoch": 2.610169491525424,
      "grad_norm": 2.1334943771362305,
      "learning_rate": 4.607619728353818e-06,
      "loss": 3.9206,
      "step": 154
    },
    {
      "epoch": 2.6271186440677967,
      "grad_norm": 1.9174208641052246,
      "learning_rate": 4.221282260619891e-06,
      "loss": 4.0656,
      "step": 155
    },
    {
      "epoch": 2.6440677966101696,
      "grad_norm": 1.985168695449829,
      "learning_rate": 3.851144890049535e-06,
      "loss": 3.8693,
      "step": 156
    },
    {
      "epoch": 2.6610169491525424,
      "grad_norm": 2.399648427963257,
      "learning_rate": 3.4973386002452535e-06,
      "loss": 3.8203,
      "step": 157
    },
    {
      "epoch": 2.6779661016949152,
      "grad_norm": 2.8400230407714844,
      "learning_rate": 3.159988595593616e-06,
      "loss": 3.3306,
      "step": 158
    },
    {
      "epoch": 2.694915254237288,
      "grad_norm": 3.073596954345703,
      "learning_rate": 2.839214256958106e-06,
      "loss": 3.7538,
      "step": 159
    },
    {
      "epoch": 2.711864406779661,
      "grad_norm": 3.634011745452881,
      "learning_rate": 2.53512909943287e-06,
      "loss": 2.8218,
      "step": 160
    },
    {
      "epoch": 2.7288135593220337,
      "grad_norm": 1.1675502061843872,
      "learning_rate": 2.2478407321721296e-06,
      "loss": 3.9812,
      "step": 161
    },
    {
      "epoch": 2.7457627118644066,
      "grad_norm": 1.1497820615768433,
      "learning_rate": 1.977450820309684e-06,
      "loss": 4.1193,
      "step": 162
    },
    {
      "epoch": 2.7627118644067794,
      "grad_norm": 1.3536845445632935,
      "learning_rate": 1.7240550489817653e-06,
      "loss": 5.5152,
      "step": 163
    },
    {
      "epoch": 2.7796610169491527,
      "grad_norm": 1.5920711755752563,
      "learning_rate": 1.4877430894662036e-06,
      "loss": 4.3516,
      "step": 164
    },
    {
      "epoch": 2.7966101694915255,
      "grad_norm": 1.4585187435150146,
      "learning_rate": 1.268598567449647e-06,
      "loss": 5.0955,
      "step": 165
    },
    {
      "epoch": 2.8135593220338984,
      "grad_norm": 1.5151571035385132,
      "learning_rate": 1.0666990334342707e-06,
      "loss": 4.6564,
      "step": 166
    },
    {
      "epoch": 2.830508474576271,
      "grad_norm": 1.7424404621124268,
      "learning_rate": 8.821159352943143e-07,
      "loss": 4.3424,
      "step": 167
    },
    {
      "epoch": 2.847457627118644,
      "grad_norm": 1.8211919069290161,
      "learning_rate": 7.149145929922607e-07,
      "loss": 3.9684,
      "step": 168
    },
    {
      "epoch": 2.864406779661017,
      "grad_norm": 2.007244110107422,
      "learning_rate": 5.651541754634726e-07,
      "loss": 4.424,
      "step": 169
    },
    {
      "epoch": 2.8813559322033897,
      "grad_norm": 2.0442068576812744,
      "learning_rate": 4.3288767967760715e-07,
      "loss": 4.3852,
      "step": 170
    },
    {
      "epoch": 2.898305084745763,
      "grad_norm": 2.194523811340332,
      "learning_rate": 3.1816191188415166e-07,
      "loss": 4.0062,
      "step": 171
    },
    {
      "epoch": 2.915254237288136,
      "grad_norm": 2.606753349304199,
      "learning_rate": 2.2101747104866788e-07,
      "loss": 3.3915,
      "step": 172
    },
    {
      "epoch": 2.9322033898305087,
      "grad_norm": 2.934316396713257,
      "learning_rate": 1.4148873448573408e-07,
      "loss": 3.4754,
      "step": 173
    },
    {
      "epoch": 2.9491525423728815,
      "grad_norm": 3.9927470684051514,
      "learning_rate": 7.960384569353219e-08,
      "loss": 3.3954,
      "step": 174
    },
    {
      "epoch": 2.9661016949152543,
      "grad_norm": 1.8415848016738892,
      "learning_rate": 3.538470439448105e-08,
      "loss": 5.1722,
      "step": 175
    },
    {
      "epoch": 2.983050847457627,
      "grad_norm": 1.8669278621673584,
      "learning_rate": 8.846958785418968e-09,
      "loss": 4.7516,
      "step": 176
    },
    {
      "epoch": 3.0,
      "grad_norm": 2.779259443283081,
      "learning_rate": 0.0,
      "loss": 3.4324,
      "step": 177
    }
  ],
  "logging_steps": 1,
  "max_steps": 177,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 50,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 5,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 2
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 4.8791639483744256e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}