{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.8826254659857349,
  "eval_steps": 1024,
  "global_step": 20480,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.011032818324821687,
      "grad_norm": 10.797319412231445,
      "learning_rate": 0.000498046875,
      "loss": 105.79659271240234,
      "step": 256
    },
    {
      "epoch": 0.022065636649643373,
      "grad_norm": 25.741575241088867,
      "learning_rate": 0.000998046875,
      "loss": 2.3060808181762695,
      "step": 512
    },
    {
      "epoch": 0.03309845497446506,
      "grad_norm": 14.934273719787598,
      "learning_rate": 0.000999688448778502,
      "loss": 2.2731690406799316,
      "step": 768
    },
    {
      "epoch": 0.04413127329928675,
      "grad_norm": 28.308002471923828,
      "learning_rate": 0.0009987492950653055,
      "loss": 2.175210475921631,
      "step": 1024
    },
    {
      "epoch": 0.04413127329928675,
      "eval_bleu": 0.9214307463191996,
      "eval_cos_loss": 0.7069334270857545,
      "eval_dec_loss": 0.14960542684599662,
      "eval_loss": 2.0735362056475966,
      "eval_mse2_loss": 0.2511829924164042,
      "eval_mse_loss": 1.8223532110389107,
      "eval_rec_loss": 0.047009017791098624,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.29306657075373604,
      "flow/improvement_ratio": 0.7837164053784759,
      "flow/mag_ratio_mean": 0.434395147856873,
      "flow/mag_ratio_std": 0.1415387201728597,
      "step": 1024
    },
    {
      "epoch": 0.04413127329928675,
      "eval_bleu": 0.9214307463191996,
      "eval_cos_loss": 0.7069334270857545,
      "eval_dec_loss": 0.14960542684599662,
      "eval_loss": 2.0735362056475966,
      "eval_mse2_loss": 0.2511829924164042,
      "eval_mse_loss": 1.8223532110389107,
      "eval_rec_loss": 0.047009017791098624,
      "eval_runtime": 171.0537,
      "eval_samples_per_second": 175.384,
      "eval_steps_per_second": 2.742,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.29306657075373604,
      "flow/improvement_ratio": 0.7837164053784759,
      "flow/mag_ratio_mean": 0.434395147856873,
      "flow/mag_ratio_std": 0.1415387201728597,
      "step": 1024
    },
    {
      "epoch": 0.05516409162410843,
      "grad_norm": 20.717588424682617,
      "learning_rate": 0.0009971837136430763,
      "loss": 1.9890590906143188,
      "step": 1280
    },
    {
      "epoch": 0.06619690994893011,
      "grad_norm": 16.1776123046875,
      "learning_rate": 0.0009949936708776692,
      "loss": 1.8445225954055786,
      "step": 1536
    },
    {
      "epoch": 0.07722972827375181,
      "grad_norm": 15.698042869567871,
      "learning_rate": 0.0009921819174566252,
      "loss": 1.7741947174072266,
      "step": 1792
    },
    {
      "epoch": 0.0882625465985735,
      "grad_norm": 12.729333877563477,
      "learning_rate": 0.000988751984934317,
      "loss": 1.714084267616272,
      "step": 2048
    },
    {
      "epoch": 0.0882625465985735,
      "eval_bleu": 0.9356201248298646,
      "eval_cos_loss": 0.5489801322854658,
      "eval_dec_loss": 0.11870824870889757,
      "eval_loss": 1.7085340877077473,
      "eval_mse2_loss": 0.1958959426865903,
      "eval_mse_loss": 1.5126381456724871,
      "eval_rec_loss": 0.047009017791098624,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.4510198644102255,
      "flow/improvement_ratio": 0.8669772742907884,
      "flow/mag_ratio_mean": 0.5195024066261137,
      "flow/mag_ratio_std": 0.21416059253947822,
      "step": 2048
    },
    {
      "epoch": 0.0882625465985735,
      "eval_bleu": 0.9356201248298646,
      "eval_cos_loss": 0.5489801322854658,
      "eval_dec_loss": 0.11870824870889757,
      "eval_loss": 1.7085340877077473,
      "eval_mse2_loss": 0.1958959426865903,
      "eval_mse_loss": 1.5126381456724871,
      "eval_rec_loss": 0.047009017791098624,
      "eval_runtime": 172.0919,
      "eval_samples_per_second": 174.325,
      "eval_steps_per_second": 2.725,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.4510198644102255,
      "flow/improvement_ratio": 0.8669772742907884,
      "flow/mag_ratio_mean": 0.5195024066261137,
      "flow/mag_ratio_std": 0.21416059253947822,
      "step": 2048
    },
    {
      "epoch": 0.09929536492339518,
      "grad_norm": 9.853597640991211,
      "learning_rate": 0.0009847081812963268,
      "loss": 1.6746653318405151,
      "step": 2304
    },
    {
      "epoch": 0.11032818324821686,
      "grad_norm": 7.131998062133789,
      "learning_rate": 0.0009800555855486275,
      "loss": 1.646998405456543,
      "step": 2560
    },
    {
      "epoch": 0.12136100157303854,
      "grad_norm": 7.610518932342529,
      "learning_rate": 0.0009748000413383664,
      "loss": 1.62706458568573,
      "step": 2816
    },
    {
      "epoch": 0.13239381989786023,
      "grad_norm": 5.741150856018066,
      "learning_rate": 0.0009689481496142604,
      "loss": 1.6093950271606445,
      "step": 3072
    },
    {
      "epoch": 0.13239381989786023,
      "eval_bleu": 0.9338383707066459,
      "eval_cos_loss": 0.5054820228868456,
      "eval_dec_loss": 0.12964063795851366,
      "eval_loss": 1.5915681512625233,
      "eval_mse2_loss": 0.1819949251438763,
      "eval_mse_loss": 1.4095732276119404,
      "eval_rec_loss": 0.047009017791098624,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.4945179769860656,
      "flow/improvement_ratio": 0.8849164608444995,
      "flow/mag_ratio_mean": 0.5184873009541395,
      "flow/mag_ratio_std": 0.229987337486322,
      "step": 3072
    },
    {
      "epoch": 0.13239381989786023,
      "eval_bleu": 0.9338383707066459,
      "eval_cos_loss": 0.5054820228868456,
      "eval_dec_loss": 0.12964063795851366,
      "eval_loss": 1.5915681512625233,
      "eval_mse2_loss": 0.1819949251438763,
      "eval_mse_loss": 1.4095732276119404,
      "eval_rec_loss": 0.047009017791098624,
      "eval_runtime": 176.4575,
      "eval_samples_per_second": 170.013,
      "eval_steps_per_second": 2.658,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.4945179769860656,
      "flow/improvement_ratio": 0.8849164608444995,
      "flow/mag_ratio_mean": 0.5184873009541395,
      "flow/mag_ratio_std": 0.229987337486322,
      "step": 3072
    },
    {
      "epoch": 0.14342663822268192,
      "grad_norm": 5.7491865158081055,
      "learning_rate": 0.0009625072603358231,
      "loss": 1.5970690250396729,
      "step": 3328
    },
    {
      "epoch": 0.15445945654750362,
      "grad_norm": 6.157455921173096,
      "learning_rate": 0.0009554854632418371,
      "loss": 1.5825296640396118,
      "step": 3584
    },
    {
      "epoch": 0.1654922748723253,
      "grad_norm": 4.907317638397217,
      "learning_rate": 0.000947891577689663,
      "loss": 1.5728044509887695,
      "step": 3840
    },
    {
      "epoch": 0.176525093197147,
      "grad_norm": 3.5090701580047607,
      "learning_rate": 0.0009397351415781539,
      "loss": 1.565865159034729,
      "step": 4096
    },
    {
      "epoch": 0.176525093197147,
      "eval_bleu": 0.9365961616358232,
      "eval_cos_loss": 0.49659409953841266,
      "eval_dec_loss": 0.12140450885753705,
      "eval_loss": 1.564894216655414,
      "eval_mse2_loss": 0.17669188814249628,
      "eval_mse_loss": 1.3882023289259562,
      "eval_rec_loss": 0.047009017791098624,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.5034058995719658,
      "flow/improvement_ratio": 0.8819394332767804,
      "flow/mag_ratio_mean": 0.5346324385356293,
      "flow/mag_ratio_std": 0.23142226367616958,
      "step": 4096
    },
    {
      "epoch": 0.176525093197147,
      "eval_bleu": 0.9365961616358232,
      "eval_cos_loss": 0.49659409953841266,
      "eval_dec_loss": 0.12140450885753705,
      "eval_loss": 1.564894216655414,
      "eval_mse2_loss": 0.17669188814249628,
      "eval_mse_loss": 1.3882023289259562,
      "eval_rec_loss": 0.047009017791098624,
      "eval_runtime": 160.4591,
      "eval_samples_per_second": 186.964,
      "eval_steps_per_second": 2.923,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.5034058995719658,
      "flow/improvement_ratio": 0.8819394332767804,
      "flow/mag_ratio_mean": 0.5346324385356293,
      "flow/mag_ratio_std": 0.23142226367616958,
      "step": 4096
    },
    {
      "epoch": 0.18755791152196866,
      "grad_norm": 4.376228332519531,
      "learning_rate": 0.000931026399368079,
      "loss": 1.5588833093643188,
      "step": 4352
    },
    {
      "epoch": 0.19859072984679035,
      "grad_norm": 4.248619556427002,
      "learning_rate": 0.0009217762892151117,
      "loss": 1.5539987087249756,
      "step": 4608
    },
    {
      "epoch": 0.20962354817161205,
      "grad_norm": 4.112342357635498,
      "learning_rate": 0.0009119964292315354,
      "loss": 1.547034502029419,
      "step": 4864
    },
    {
      "epoch": 0.22065636649643372,
      "grad_norm": 4.957840919494629,
      "learning_rate": 0.0009016991028939279,
      "loss": 1.539028525352478,
      "step": 5120
    },
    {
      "epoch": 0.22065636649643372,
      "eval_bleu": 0.937221428849788,
      "eval_cos_loss": 0.48556288504905537,
      "eval_dec_loss": 0.11899168453395748,
      "eval_loss": 1.5318274358188166,
      "eval_mse2_loss": 0.17039804826222504,
      "eval_mse_loss": 1.3614293873183,
      "eval_rec_loss": 0.047009017791098624,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.5144371166666434,
      "flow/improvement_ratio": 0.8873141655789764,
      "flow/mag_ratio_mean": 0.5353354320470204,
      "flow/mag_ratio_std": 0.2365191124204888,
      "step": 5120
    },
    {
      "epoch": 0.22065636649643372,
      "eval_bleu": 0.937221428849788,
      "eval_cos_loss": 0.48556288504905537,
      "eval_dec_loss": 0.11899168453395748,
      "eval_loss": 1.5318274358188166,
      "eval_mse2_loss": 0.17039804826222504,
      "eval_mse_loss": 1.3614293873183,
      "eval_rec_loss": 0.047009017791098624,
      "eval_runtime": 169.2874,
      "eval_samples_per_second": 177.213,
      "eval_steps_per_second": 2.77,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.5144371166666434,
      "flow/improvement_ratio": 0.8873141655789764,
      "flow/mag_ratio_mean": 0.5353354320470204,
      "flow/mag_ratio_std": 0.2365191124204888,
      "step": 5120
    },
    {
      "epoch": 0.23168918482125542,
      "grad_norm": 3.4134156703948975,
      "learning_rate": 0.0008908972436151494,
      "loss": 1.532220721244812,
      "step": 5376
    },
    {
      "epoch": 0.2427220031460771,
      "grad_norm": 3.1423563957214355,
      "learning_rate": 0.0008796044185000127,
      "loss": 1.525950312614441,
      "step": 5632
    },
    {
      "epoch": 0.2537548214708988,
      "grad_norm": 2.3490381240844727,
      "learning_rate": 0.0008678348113050368,
      "loss": 1.5172430276870728,
      "step": 5888
    },
    {
      "epoch": 0.26478763979572045,
      "grad_norm": 6.032445430755615,
      "learning_rate": 0.0008556032046236897,
      "loss": 1.5190675258636475,
      "step": 6144
    },
    {
      "epoch": 0.26478763979572045,
      "eval_bleu": 0.9354061397916119,
      "eval_cos_loss": 0.4837510161308337,
      "eval_dec_loss": 0.1238921330338205,
      "eval_loss": 1.5238950331328012,
      "eval_mse2_loss": 0.17081616409043512,
      "eval_mse_loss": 1.353078869360088,
      "eval_rec_loss": 0.047009017791098624,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.5162489819628343,
      "flow/improvement_ratio": 0.8831684786373618,
      "flow/mag_ratio_mean": 0.5513551437905603,
      "flow/mag_ratio_std": 0.24864270314097658,
      "step": 6144
    },
    {
      "epoch": 0.26478763979572045,
      "eval_bleu": 0.9354061397916119,
      "eval_cos_loss": 0.4837510161308337,
      "eval_dec_loss": 0.1238921330338205,
      "eval_loss": 1.5238950331328012,
      "eval_mse2_loss": 0.17081616409043512,
      "eval_mse_loss": 1.353078869360088,
      "eval_rec_loss": 0.047009017791098624,
      "eval_runtime": 172.5467,
      "eval_samples_per_second": 173.866,
      "eval_steps_per_second": 2.718,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.5162489819628343,
      "flow/improvement_ratio": 0.8831684786373618,
      "flow/mag_ratio_mean": 0.5513551437905603,
      "flow/mag_ratio_std": 0.24864270314097658,
      "step": 6144
    },
    {
      "epoch": 0.2758204581205422,
      "grad_norm": 4.0988078117370605,
      "learning_rate": 0.000842924961319492,
      "loss": 1.5116130113601685,
      "step": 6400
    },
    {
      "epoch": 0.28685327644536385,
      "grad_norm": 3.4181602001190186,
      "learning_rate": 0.0008298160052303045,
      "loss": 1.506291389465332,
      "step": 6656
    },
    {
      "epoch": 0.2978860947701855,
      "grad_norm": 3.047140598297119,
      "learning_rate": 0.0008162928011680314,
      "loss": 1.5031940937042236,
      "step": 6912
    },
    {
      "epoch": 0.30891891309500724,
      "grad_norm": 3.206284761428833,
      "learning_rate": 0.000802372334238864,
      "loss": 1.5017902851104736,
      "step": 7168
    },
    {
      "epoch": 0.30891891309500724,
      "eval_bleu": 0.9368851499593083,
      "eval_cos_loss": 0.47378275395710584,
      "eval_dec_loss": 0.11915739709888694,
      "eval_loss": 1.4968281847073326,
      "eval_mse2_loss": 0.1652413869876343,
      "eval_mse_loss": 1.3315867993877386,
      "eval_rec_loss": 0.047009017791098624,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.5262172466783381,
      "flow/improvement_ratio": 0.8891915549347396,
      "flow/mag_ratio_mean": 0.5413357251361489,
      "flow/mag_ratio_std": 0.24727411558633167,
      "step": 7168
    },
    {
      "epoch": 0.30891891309500724,
      "eval_bleu": 0.9368851499593083,
      "eval_cos_loss": 0.47378275395710584,
      "eval_dec_loss": 0.11915739709888694,
      "eval_loss": 1.4968281847073326,
      "eval_mse2_loss": 0.1652413869876343,
      "eval_mse_loss": 1.3315867993877386,
      "eval_rec_loss": 0.047009017791098624,
      "eval_runtime": 176.7103,
      "eval_samples_per_second": 169.769,
      "eval_steps_per_second": 2.654,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.5262172466783381,
      "flow/improvement_ratio": 0.8891915549347396,
      "flow/mag_ratio_mean": 0.5413357251361489,
      "flow/mag_ratio_std": 0.24727411558633167,
      "step": 7168
    },
    {
      "epoch": 0.3199517314198289,
      "grad_norm": 3.409893274307251,
      "learning_rate": 0.0007880720885100349,
      "loss": 1.4971449375152588,
      "step": 7424
    },
    {
      "epoch": 0.3309845497446506,
      "grad_norm": 3.073834180831909,
      "learning_rate": 0.0007734100250498788,
      "loss": 1.49185311794281,
      "step": 7680
    },
    {
      "epoch": 0.3420173680694723,
      "grad_norm": 4.364046096801758,
      "learning_rate": 0.000758404559368781,
      "loss": 1.4903632402420044,
      "step": 7936
    },
    {
      "epoch": 0.353050186394294,
      "grad_norm": 2.848015785217285,
      "learning_rate": 0.0007430745382893488,
      "loss": 1.4838106632232666,
      "step": 8192
    },
    {
      "epoch": 0.353050186394294,
      "eval_bleu": 0.9353698110567827,
      "eval_cos_loss": 0.4692605264913807,
      "eval_dec_loss": 0.12152907604983113,
      "eval_loss": 1.4847185436342316,
      "eval_mse2_loss": 0.16447527768578865,
      "eval_mse_loss": 1.3202432660914178,
      "eval_rec_loss": 0.047009017791098624,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.5307394738898856,
      "flow/improvement_ratio": 0.8924712222268079,
      "flow/mag_ratio_mean": 0.5445197708825312,
      "flow/mag_ratio_std": 0.24787098121668485,
      "step": 8192
    },
    {
      "epoch": 0.353050186394294,
      "eval_bleu": 0.9353698110567827,
      "eval_cos_loss": 0.4692605264913807,
      "eval_dec_loss": 0.12152907604983113,
      "eval_loss": 1.4847185436342316,
      "eval_mse2_loss": 0.16447527768578865,
      "eval_mse_loss": 1.3202432660914178,
      "eval_rec_loss": 0.047009017791098624,
      "eval_runtime": 173.5643,
      "eval_samples_per_second": 172.847,
      "eval_steps_per_second": 2.702,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.5307394738898856,
      "flow/improvement_ratio": 0.8924712222268079,
      "flow/mag_ratio_mean": 0.5445197708825312,
      "flow/mag_ratio_std": 0.24787098121668485,
      "step": 8192
    },
    {
      "epoch": 0.36408300471911564,
      "grad_norm": 2.5942811965942383,
      "learning_rate": 0.0007274392162748551,
      "loss": 1.4843038320541382,
      "step": 8448
    },
    {
      "epoch": 0.3751158230439373,
      "grad_norm": 3.0958163738250732,
      "learning_rate": 0.000711518231245687,
      "loss": 1.4775118827819824,
      "step": 8704
    },
    {
      "epoch": 0.38614864136875904,
      "grad_norm": 3.219830274581909,
      "learning_rate": 0.0006953315799141723,
      "loss": 1.4766325950622559,
      "step": 8960
    },
    {
      "epoch": 0.3971814596935807,
      "grad_norm": 2.296929359436035,
      "learning_rate": 0.0006788995926687669,
      "loss": 1.4715560674667358,
      "step": 9216
    },
    {
      "epoch": 0.3971814596935807,
      "eval_bleu": 0.9356008220185699,
      "eval_cos_loss": 0.4650544371050812,
      "eval_dec_loss": 0.1218992033397465,
      "eval_loss": 1.4720673169662704,
      "eval_mse2_loss": 0.1628535425167348,
      "eval_mse_loss": 1.3092137732739642,
      "eval_rec_loss": 0.047009017791098624,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.534945561814664,
      "flow/improvement_ratio": 0.8920619619934798,
      "flow/mag_ratio_mean": 0.5498689174143745,
      "flow/mag_ratio_std": 0.2565161931489322,
      "step": 9216
    },
    {
      "epoch": 0.3971814596935807,
      "eval_bleu": 0.9356008220185699,
      "eval_cos_loss": 0.4650544371050812,
      "eval_dec_loss": 0.1218992033397465,
      "eval_loss": 1.4720673169662704,
      "eval_mse2_loss": 0.1628535425167348,
      "eval_mse_loss": 1.3092137732739642,
      "eval_rec_loss": 0.047009017791098624,
      "eval_runtime": 165.3408,
      "eval_samples_per_second": 181.443,
      "eval_steps_per_second": 2.837,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.534945561814664,
      "flow/improvement_ratio": 0.8920619619934798,
      "flow/mag_ratio_mean": 0.5498689174143745,
      "flow/mag_ratio_std": 0.2565161931489322,
      "step": 9216
    },
    {
      "epoch": 0.4082142780184024,
      "grad_norm": 2.3320987224578857,
      "learning_rate": 0.0006622429080391422,
      "loss": 1.4737770557403564,
      "step": 9472
    },
    {
      "epoch": 0.4192470963432241,
      "grad_norm": 2.76210355758667,
      "learning_rate": 0.0006453824467742515,
      "loss": 1.467403531074524,
      "step": 9728
    },
    {
      "epoch": 0.43027991466804577,
      "grad_norm": 2.2834606170654297,
      "learning_rate": 0.0006283393855659275,
      "loss": 1.466551661491394,
      "step": 9984
    },
    {
      "epoch": 0.44131273299286744,
      "grad_norm": 2.1904284954071045,
      "learning_rate": 0.0006111351304510173,
      "loss": 1.46049165725708,
      "step": 10240
    },
    {
      "epoch": 0.44131273299286744,
      "eval_bleu": 0.9362346289783122,
      "eval_cos_loss": 0.46082611143716107,
      "eval_dec_loss": 0.1203397757717287,
      "eval_loss": 1.4601926084266288,
      "eval_mse2_loss": 0.1609408200613217,
      "eval_mse_loss": 1.2992517866813806,
      "eval_rec_loss": 0.047009017791098624,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.5391738893253717,
      "flow/improvement_ratio": 0.8964146429033422,
      "flow/mag_ratio_mean": 0.5516184565863376,
      "flow/mag_ratio_std": 0.25244328309732206,
      "step": 10240
    },
    {
      "epoch": 0.44131273299286744,
      "eval_bleu": 0.9362346289783122,
      "eval_cos_loss": 0.46082611143716107,
      "eval_dec_loss": 0.1203397757717287,
      "eval_loss": 1.4601926084266288,
      "eval_mse2_loss": 0.1609408200613217,
      "eval_mse_loss": 1.2992517866813806,
      "eval_rec_loss": 0.047009017791098624,
      "eval_runtime": 179.4357,
      "eval_samples_per_second": 167.191,
      "eval_steps_per_second": 2.614,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.5391738893253717,
      "flow/improvement_ratio": 0.8964146429033422,
      "flow/mag_ratio_mean": 0.5516184565863376,
      "flow/mag_ratio_std": 0.25244328309732206,
      "step": 10240
    },
    {
      "epoch": 0.45234555131768917,
      "grad_norm": 2.3216679096221924,
      "learning_rate": 0.0005937912899254605,
      "loss": 1.4583410024642944,
      "step": 10496
    },
    {
      "epoch": 0.46337836964251083,
      "grad_norm": 2.843108892440796,
      "learning_rate": 0.0005763296478040787,
      "loss": 1.4574695825576782,
      "step": 10752
    },
    {
      "epoch": 0.4744111879673325,
      "grad_norm": 2.320223808288574,
      "learning_rate": 0.0005587721358601663,
      "loss": 1.4572159051895142,
      "step": 11008
    },
    {
      "epoch": 0.4854440062921542,
      "grad_norm": 2.517941474914551,
      "learning_rate": 0.0005411408062792448,
      "loss": 1.4533445835113525,
      "step": 11264
    },
    {
      "epoch": 0.4854440062921542,
      "eval_bleu": 0.9349888888103645,
      "eval_cos_loss": 0.4606767813407028,
      "eval_dec_loss": 0.12063868982848455,
      "eval_loss": 1.4557419222301002,
      "eval_mse2_loss": 0.15818865597248077,
      "eval_mse_loss": 1.2975532645101486,
      "eval_rec_loss": 0.047009017791098624,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.5393232155456218,
      "flow/improvement_ratio": 0.892127673127758,
      "flow/mag_ratio_mean": 0.550250798717999,
      "flow/mag_ratio_std": 0.2556351939879501,
      "step": 11264
    },
    {
      "epoch": 0.4854440062921542,
      "eval_bleu": 0.9349888888103645,
      "eval_cos_loss": 0.4606767813407028,
      "eval_dec_loss": 0.12063868982848455,
      "eval_loss": 1.4557419222301002,
      "eval_mse2_loss": 0.15818865597248077,
      "eval_mse_loss": 1.2975532645101486,
      "eval_rec_loss": 0.047009017791098624,
      "eval_runtime": 169.0185,
      "eval_samples_per_second": 177.495,
      "eval_steps_per_second": 2.775,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.5393232155456218,
      "flow/improvement_ratio": 0.892127673127758,
      "flow/mag_ratio_mean": 0.550250798717999,
      "flow/mag_ratio_std": 0.2556351939879501,
      "step": 11264
    },
    {
      "epoch": 0.4964768246169759,
      "grad_norm": 2.285221815109253,
      "learning_rate": 0.0005234578039615789,
      "loss": 1.4499876499176025,
      "step": 11520
    },
    {
      "epoch": 0.5075096429417976,
      "grad_norm": 2.029639482498169,
      "learning_rate": 0.0005057453387082458,
      "loss": 1.4481278657913208,
      "step": 11776
    },
    {
      "epoch": 0.5185424612666193,
      "grad_norm": 2.0235965251922607,
      "learning_rate": 0.0004880256573256866,
      "loss": 1.4484679698944092,
      "step": 12032
    },
    {
      "epoch": 0.5295752795914409,
      "grad_norm": 1.6718403100967407,
      "learning_rate": 0.0004703210156837805,
      "loss": 1.447327733039856,
      "step": 12288
    },
    {
      "epoch": 0.5295752795914409,
      "eval_bleu": 0.9364176777903095,
      "eval_cos_loss": 0.4588198344082212,
      "eval_dec_loss": 0.11679136855746193,
      "eval_loss": 1.4490999984842883,
      "eval_mse2_loss": 0.15755183694522773,
      "eval_mse_loss": 1.2915481588225375,
      "eval_rec_loss": 0.047009017791098624,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.5411801628593697,
      "flow/improvement_ratio": 0.8935188584998727,
      "flow/mag_ratio_mean": 0.5548275625273618,
      "flow/mag_ratio_std": 0.25878340491989277,
      "step": 12288
    },
    {
      "epoch": 0.5295752795914409,
      "eval_bleu": 0.9364176777903095,
      "eval_cos_loss": 0.4588198344082212,
      "eval_dec_loss": 0.11679136855746193,
      "eval_loss": 1.4490999984842883,
      "eval_mse2_loss": 0.15755183694522773,
      "eval_mse_loss": 1.2915481588225375,
      "eval_rec_loss": 0.047009017791098624,
      "eval_runtime": 167.4569,
      "eval_samples_per_second": 179.151,
      "eval_steps_per_second": 2.801,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.5411801628593697,
      "flow/improvement_ratio": 0.8935188584998727,
      "flow/mag_ratio_mean": 0.5548275625273618,
      "flow/mag_ratio_std": 0.25878340491989277,
      "step": 12288
    },
    {
      "epoch": 0.5406080979162626,
      "grad_norm": 1.9634923934936523,
      "learning_rate": 0.0004526536507625343,
      "loss": 1.443847894668579,
      "step": 12544
    },
    {
      "epoch": 0.5516409162410844,
      "grad_norm": 2.0251376628875732,
      "learning_rate": 0.00043504575272249973,
      "loss": 1.4448539018630981,
      "step": 12800
    },
    {
      "epoch": 0.562673734565906,
      "grad_norm": 1.487449049949646,
      "learning_rate": 0.0004175194370339921,
      "loss": 1.44254732131958,
      "step": 13056
    },
    {
      "epoch": 0.5737065528907277,
      "grad_norm": 2.1053073406219482,
      "learning_rate": 0.0004000967167001243,
      "loss": 1.4406143426895142,
      "step": 13312
    },
    {
      "epoch": 0.5737065528907277,
      "eval_bleu": 0.9372209976548388,
      "eval_cos_loss": 0.45495398104317913,
      "eval_dec_loss": 0.11334308082345071,
      "eval_loss": 1.438905306970641,
      "eval_mse2_loss": 0.15505272262830977,
      "eval_mse_loss": 1.2838525835639123,
      "eval_rec_loss": 0.047009017791098624,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.5450460188932764,
      "flow/improvement_ratio": 0.8936982345479384,
      "flow/mag_ratio_mean": 0.5579607989996481,
      "flow/mag_ratio_std": 0.2575365855876825,
      "step": 13312
    },
    {
      "epoch": 0.5737065528907277,
      "eval_bleu": 0.9372209976548388,
      "eval_cos_loss": 0.45495398104317913,
      "eval_dec_loss": 0.11334308082345071,
      "eval_loss": 1.438905306970641,
      "eval_mse2_loss": 0.15505272262830977,
      "eval_mse_loss": 1.2838525835639123,
      "eval_rec_loss": 0.047009017791098624,
      "eval_runtime": 177.5771,
      "eval_samples_per_second": 168.941,
      "eval_steps_per_second": 2.641,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.5450460188932764,
      "flow/improvement_ratio": 0.8936982345479384,
      "flow/mag_ratio_mean": 0.5579607989996481,
      "flow/mag_ratio_std": 0.2575365855876825,
      "step": 13312
    },
    {
      "epoch": 0.5847393712155494,
      "grad_norm": 2.0249550342559814,
      "learning_rate": 0.00038279947460853446,
      "loss": 1.4377583265304565,
      "step": 13568
    },
    {
      "epoch": 0.595772189540371,
      "grad_norm": 1.6271026134490967,
      "learning_rate": 0.00036564943604654345,
      "loss": 1.4369451999664307,
      "step": 13824
    },
    {
      "epoch": 0.6068050078651928,
      "grad_norm": 1.2092267274856567,
      "learning_rate": 0.00034866814141425254,
      "loss": 1.437900424003601,
      "step": 14080
    },
    {
      "epoch": 0.6178378261900145,
      "grad_norm": 1.7238409519195557,
      "learning_rate": 0.0003318769191698637,
      "loss": 1.4345003366470337,
      "step": 14336
    },
    {
      "epoch": 0.6178378261900145,
      "eval_bleu": 0.9375344922384511,
      "eval_cos_loss": 0.45383678259117516,
      "eval_dec_loss": 0.11429015738961062,
      "eval_loss": 1.4351127869538916,
      "eval_mse2_loss": 0.15503559347345378,
      "eval_mse_loss": 1.280077194608351,
      "eval_rec_loss": 0.047009017791098624,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.5461632174723692,
      "flow/improvement_ratio": 0.8951520059408664,
      "flow/mag_ratio_mean": 0.5569842827599695,
      "flow/mag_ratio_std": 0.26188866851299303,
      "step": 14336
    },
    {
      "epoch": 0.6178378261900145,
      "eval_bleu": 0.9375344922384511,
      "eval_cos_loss": 0.45383678259117516,
      "eval_dec_loss": 0.11429015738961062,
      "eval_loss": 1.4351127869538916,
      "eval_mse2_loss": 0.15503559347345378,
      "eval_mse_loss": 1.280077194608351,
      "eval_rec_loss": 0.047009017791098624,
      "eval_runtime": 176.4023,
      "eval_samples_per_second": 170.066,
      "eval_steps_per_second": 2.659,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.5461632174723692,
      "flow/improvement_ratio": 0.8951520059408664,
      "flow/mag_ratio_mean": 0.5569842827599695,
      "flow/mag_ratio_std": 0.26188866851299303,
      "step": 14336
    },
    {
      "epoch": 0.6288706445148361,
      "grad_norm": 1.5894383192062378,
      "learning_rate": 0.00031529685904119485,
      "loss": 1.4325922727584839,
      "step": 14592
    },
    {
      "epoch": 0.6399034628396578,
      "grad_norm": 1.9905190467834473,
      "learning_rate": 0.0002989487855370421,
      "loss": 1.4348605871200562,
      "step": 14848
    },
    {
      "epoch": 0.6509362811644795,
      "grad_norm": 1.5708714723587036,
      "learning_rate": 0.00028285323179165424,
      "loss": 1.4332606792449951,
      "step": 15104
    },
    {
      "epoch": 0.6619690994893012,
      "grad_norm": 1.5457080602645874,
      "learning_rate": 0.0002670304137751759,
      "loss": 1.435610294342041,
      "step": 15360
    },
    {
      "epoch": 0.6619690994893012,
      "eval_bleu": 0.9355922222002754,
      "eval_cos_loss": 0.4520330929171556,
      "eval_dec_loss": 0.12299507638332305,
      "eval_loss": 1.4329845295277739,
      "eval_mse2_loss": 0.15654502583465088,
      "eval_mse_loss": 1.2764395058536326,
      "eval_rec_loss": 0.047009017791098624,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.5479669079724659,
      "flow/improvement_ratio": 0.8930472631190123,
      "flow/mag_ratio_mean": 0.5571114101897933,
      "flow/mag_ratio_std": 0.256343261551247,
      "step": 15360
    },
    {
      "epoch": 0.6619690994893012,
      "eval_bleu": 0.9355922222002754,
      "eval_cos_loss": 0.4520330929171556,
      "eval_dec_loss": 0.12299507638332305,
      "eval_loss": 1.4329845295277739,
      "eval_mse2_loss": 0.15654502583465088,
      "eval_mse_loss": 1.2764395058536326,
      "eval_rec_loss": 0.047009017791098624,
      "eval_runtime": 174.468,
      "eval_samples_per_second": 171.951,
      "eval_steps_per_second": 2.688,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.5479669079724659,
      "flow/improvement_ratio": 0.8930472631190123,
      "flow/mag_ratio_mean": 0.5571114101897933,
      "flow/mag_ratio_std": 0.256343261551247,
      "step": 15360
    },
    {
      "epoch": 0.6730019178141229,
      "grad_norm": 1.6477590799331665,
      "learning_rate": 0.0002515002049024435,
      "loss": 1.4318206310272217,
      "step": 15616
    },
    {
      "epoch": 0.6840347361389446,
      "grad_norm": 1.5333731174468994,
      "learning_rate": 0.00023628211107203429,
      "loss": 1.4313552379608154,
      "step": 15872
    },
    {
      "epoch": 0.6950675544637662,
      "grad_norm": 1.147505283355713,
      "learning_rate": 0.00022139524616691188,
      "loss": 1.4301337003707886,
      "step": 16128
    },
    {
      "epoch": 0.706100372788588,
      "grad_norm": 1.1893175840377808,
      "learning_rate": 0.000206858308047443,
      "loss": 1.427048921585083,
      "step": 16384
    },
    {
      "epoch": 0.706100372788588,
      "eval_bleu": 0.9378969893045919,
      "eval_cos_loss": 0.45077633311245235,
      "eval_dec_loss": 0.11347989061736126,
      "eval_loss": 1.426147088567331,
      "eval_mse2_loss": 0.1526812995071096,
      "eval_mse_loss": 1.2734657897115516,
      "eval_rec_loss": 0.047009017791098624,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.5492236670146364,
      "flow/improvement_ratio": 0.8943654500853533,
      "flow/mag_ratio_mean": 0.560582883195328,
      "flow/mag_ratio_std": 0.25867489536306754,
      "step": 16384
    },
    {
      "epoch": 0.706100372788588,
      "eval_bleu": 0.9378969893045919,
      "eval_cos_loss": 0.45077633311245235,
      "eval_dec_loss": 0.11347989061736126,
      "eval_loss": 1.426147088567331,
      "eval_mse2_loss": 0.1526812995071096,
      "eval_mse_loss": 1.2734657897115516,
      "eval_rec_loss": 0.047009017791098624,
      "eval_runtime": 174.3978,
      "eval_samples_per_second": 172.021,
      "eval_steps_per_second": 2.689,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.5492236670146364,
      "flow/improvement_ratio": 0.8943654500853533,
      "flow/mag_ratio_mean": 0.560582883195328,
      "flow/mag_ratio_std": 0.25867489536306754,
      "step": 16384
    },
    {
      "epoch": 0.7171331911134096,
      "grad_norm": 0.9107947945594788,
      "learning_rate": 0.00019268955506693798,
      "loss": 1.42864990234375,
      "step": 16640
    },
    {
      "epoch": 0.7281660094382313,
      "grad_norm": 1.4668571949005127,
      "learning_rate": 0.00017890678313921,
      "loss": 1.426837682723999,
      "step": 16896
    },
    {
      "epoch": 0.739198827763053,
      "grad_norm": 1.6071405410766602,
      "learning_rate": 0.00016552730338695792,
      "loss": 1.4257097244262695,
      "step": 17152
    },
    {
      "epoch": 0.7502316460878746,
      "grad_norm": 1.0349161624908447,
      "learning_rate": 0.00015256792039904465,
      "loss": 1.424682855606079,
      "step": 17408
    },
    {
      "epoch": 0.7502316460878746,
      "eval_bleu": 0.9392870027928614,
      "eval_cos_loss": 0.4492604330277392,
      "eval_dec_loss": 0.11262787379888392,
      "eval_loss": 1.4202917224562752,
      "eval_mse2_loss": 0.15203414787488706,
      "eval_mse_loss": 1.2682575782987355,
      "eval_rec_loss": 0.047009017791098624,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.5507395660190948,
      "flow/improvement_ratio": 0.8961705508262618,
      "flow/mag_ratio_mean": 0.5607809571823331,
      "flow/mag_ratio_std": 0.2602766063739496,
      "step": 17408
    },
    {
      "epoch": 0.7502316460878746,
      "eval_bleu": 0.9392870027928614,
      "eval_cos_loss": 0.4492604330277392,
      "eval_dec_loss": 0.11262787379888392,
      "eval_loss": 1.4202917224562752,
      "eval_mse2_loss": 0.15203414787488706,
      "eval_mse_loss": 1.2682575782987355,
      "eval_rec_loss": 0.047009017791098624,
      "eval_runtime": 188.3193,
      "eval_samples_per_second": 159.304,
      "eval_steps_per_second": 2.49,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.5507395660190948,
      "flow/improvement_ratio": 0.8961705508262618,
      "flow/mag_ratio_mean": 0.5607809571823331,
      "flow/mag_ratio_std": 0.2602766063739496,
      "step": 17408
    },
    {
      "epoch": 0.7612644644126964,
      "grad_norm": 1.1615740060806274,
      "learning_rate": 0.00014004491112398103,
      "loss": 1.4255716800689697,
      "step": 17664
    },
    {
      "epoch": 0.7722972827375181,
      "grad_norm": 0.9076006412506104,
      "learning_rate": 0.00012797400442612433,
      "loss": 1.4207247495651245,
      "step": 17920
    },
    {
      "epoch": 0.7833301010623397,
      "grad_norm": 1.1816908121109009,
      "learning_rate": 0.00011637036133026895,
      "loss": 1.4235727787017822,
      "step": 18176
    },
    {
      "epoch": 0.7943629193871614,
      "grad_norm": 1.0465947389602661,
      "learning_rate": 0.00010524855597944216,
      "loss": 1.421402096748352,
      "step": 18432
    },
    {
      "epoch": 0.7943629193871614,
      "eval_bleu": 0.9381946887096906,
      "eval_cos_loss": 0.4503502440986349,
      "eval_dec_loss": 0.11471369324811995,
      "eval_loss": 1.4244044193072614,
      "eval_mse2_loss": 0.1525979548184348,
      "eval_mse_loss": 1.2718064660456643,
      "eval_rec_loss": 0.047009017791098624,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.5496497546940216,
      "flow/improvement_ratio": 0.8959581610490518,
      "flow/mag_ratio_mean": 0.5600753023680339,
      "flow/mag_ratio_std": 0.2604317919277687,
      "step": 18432
    },
    {
      "epoch": 0.7943629193871614,
      "eval_bleu": 0.9381946887096906,
      "eval_cos_loss": 0.4503502440986349,
      "eval_dec_loss": 0.11471369324811995,
      "eval_loss": 1.4244044193072614,
      "eval_mse2_loss": 0.1525979548184348,
      "eval_mse_loss": 1.2718064660456643,
      "eval_rec_loss": 0.047009017791098624,
      "eval_runtime": 180.9403,
      "eval_samples_per_second": 165.801,
      "eval_steps_per_second": 2.592,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.5496497546940216,
      "flow/improvement_ratio": 0.8959581610490518,
      "flow/mag_ratio_mean": 0.5600753023680339,
      "flow/mag_ratio_std": 0.2604317919277687,
      "step": 18432
    },
    {
      "epoch": 0.8053957377119831,
      "grad_norm": 1.0948517322540283,
      "learning_rate": 9.462255732982089e-05,
      "loss": 1.4208451509475708,
      "step": 18688
    },
    {
      "epoch": 0.8164285560368048,
      "grad_norm": 0.8038628101348877,
      "learning_rate": 8.450571160576348e-05,
      "loss": 1.4227826595306396,
      "step": 18944
    },
    {
      "epoch": 0.8274613743616265,
      "grad_norm": 0.7290875315666199,
      "learning_rate": 7.491072553698764e-05,
      "loss": 1.4202244281768799,
      "step": 19200
    },
    {
      "epoch": 0.8384941926864482,
      "grad_norm": 0.7819137573242188,
      "learning_rate": 6.584965039895586e-05,
      "loss": 1.4174209833145142,
      "step": 19456
    },
    {
      "epoch": 0.8384941926864482,
      "eval_bleu": 0.9372464164495187,
      "eval_cos_loss": 0.4498716953720874,
      "eval_dec_loss": 0.11378647393183604,
      "eval_loss": 1.422363788588469,
      "eval_mse2_loss": 0.15286639037289854,
      "eval_mse_loss": 1.2694973953243003,
      "eval_rec_loss": 0.047009017791098624,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.5501283043101907,
      "flow/improvement_ratio": 0.8944210361824361,
      "flow/mag_ratio_mean": 0.5602786459648279,
      "flow/mag_ratio_std": 0.26204406249243567,
      "step": 19456
    },
    {
      "epoch": 0.8384941926864482,
      "eval_bleu": 0.9372464164495187,
      "eval_cos_loss": 0.4498716953720874,
      "eval_dec_loss": 0.11378647393183604,
      "eval_loss": 1.422363788588469,
      "eval_mse2_loss": 0.15286639037289854,
      "eval_mse_loss": 1.2694973953243003,
      "eval_rec_loss": 0.047009017791098624,
      "eval_runtime": 175.4483,
      "eval_samples_per_second": 170.991,
      "eval_steps_per_second": 2.673,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.5501283043101907,
      "flow/improvement_ratio": 0.8944210361824361,
      "flow/mag_ratio_mean": 0.5602786459648279,
      "flow/mag_ratio_std": 0.26204406249243567,
      "step": 19456
    },
    {
      "epoch": 0.8495270110112698,
      "grad_norm": 0.7724987268447876,
      "learning_rate": 5.73338668765051e-05,
      "loss": 1.4236266613006592,
      "step": 19712
    },
    {
      "epoch": 0.8605598293360915,
      "grad_norm": 0.6551377177238464,
      "learning_rate": 4.9374070769740984e-05,
      "loss": 1.4212990999221802,
      "step": 19968
    },
    {
      "epoch": 0.8715926476609133,
      "grad_norm": 0.8115680813789368,
      "learning_rate": 4.198025956014095e-05,
      "loss": 1.4218370914459229,
      "step": 20224
    },
    {
      "epoch": 0.8826254659857349,
      "grad_norm": 1.0555589199066162,
      "learning_rate": 3.516171985374755e-05,
      "loss": 1.4221084117889404,
      "step": 20480
    },
    {
      "epoch": 0.8826254659857349,
      "eval_bleu": 0.937209637932583,
      "eval_cos_loss": 0.4497852080158079,
      "eval_dec_loss": 0.11631931441583868,
      "eval_loss": 1.4226631736958713,
      "eval_mse2_loss": 0.15323443922089108,
      "eval_mse_loss": 1.2694287335694725,
      "eval_rec_loss": 0.047009017791098624,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.5502147936363464,
      "flow/improvement_ratio": 0.8968729432711977,
      "flow/mag_ratio_mean": 0.5599655887719664,
      "flow/mag_ratio_std": 0.2619784128055898,
      "step": 20480
    },
    {
      "epoch": 0.8826254659857349,
      "eval_bleu": 0.937209637932583,
      "eval_cos_loss": 0.4497852080158079,
      "eval_dec_loss": 0.11631931441583868,
      "eval_loss": 1.4226631736958713,
      "eval_mse2_loss": 0.15323443922089108,
      "eval_mse_loss": 1.2694287335694725,
      "eval_rec_loss": 0.047009017791098624,
      "eval_runtime": 172.9963,
      "eval_samples_per_second": 173.414,
      "eval_steps_per_second": 2.711,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.5502147936363464,
      "flow/improvement_ratio": 0.8968729432711977,
      "flow/mag_ratio_mean": 0.5599655887719664,
      "flow/mag_ratio_std": 0.2619784128055898,
      "step": 20480
    }
  ],
  "logging_steps": 256,
  "max_steps": 23204,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 1024,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 64,
  "trial_name": null,
  "trial_params": null
}