{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.709436053761951,
  "eval_steps": 1024,
  "global_step": 15360,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.011823934229365849,
      "grad_norm": 1.1487267017364502,
      "learning_rate": 0.000498046875,
      "loss": 11.798027992248535,
      "step": 256
    },
    {
      "epoch": 0.023647868458731697,
      "grad_norm": 0.8386015295982361,
      "learning_rate": 0.000998046875,
      "loss": 1.7853779792785645,
      "step": 512
    },
    {
      "epoch": 0.03547180268809755,
      "grad_norm": 0.7344363331794739,
      "learning_rate": 0.000999640996023194,
      "loss": 1.103014588356018,
      "step": 768
    },
    {
      "epoch": 0.047295736917463395,
      "grad_norm": 1.1188315153121948,
      "learning_rate": 0.0009985588674043958,
      "loss": 0.9580796360969543,
      "step": 1024
    },
    {
      "epoch": 0.047295736917463395,
      "eval_bleu": 0.9192375404443994,
      "eval_ce_loss": 0.26167990953648745,
      "eval_cos_loss": 0.26406568816127296,
      "eval_loss": 0.9037111119864738,
      "eval_mse_loss": 0.6016385473617135,
      "eval_rec_loss": 0.013986090569199833,
      "flow/cos_sim": 0.7359343250048215,
      "flow/improvement_ratio": 0.9760946458605326,
      "flow/mag_ratio_mean": 0.7269674200717717,
      "flow/mag_ratio_std": 0.1390784539316343,
      "step": 1024
    },
    {
      "epoch": 0.047295736917463395,
      "eval_bleu": 0.9192375404443994,
      "eval_ce_loss": 0.26167990953648745,
      "eval_cos_loss": 0.26406568816127296,
      "eval_loss": 0.9037111119864738,
      "eval_mse_loss": 0.6016385473617135,
      "eval_rec_loss": 0.013986090569199833,
      "eval_runtime": 144.0156,
      "eval_samples_per_second": 194.375,
      "eval_steps_per_second": 3.041,
      "flow/cos_sim": 0.7359343250048215,
      "flow/improvement_ratio": 0.9760946458605326,
      "flow/mag_ratio_mean": 0.7269674200717717,
      "flow/mag_ratio_std": 0.1390784539316343,
      "step": 1024
    },
    {
      "epoch": 0.05911967114682925,
      "grad_norm": 1.0113043785095215,
      "learning_rate": 0.0009967551747861387,
      "loss": 0.8836896419525146,
      "step": 1280
    },
    {
      "epoch": 0.0709436053761951,
      "grad_norm": 0.9680636525154114,
      "learning_rate": 0.000994232528651847,
      "loss": 0.8432819247245789,
      "step": 1536
    },
    {
      "epoch": 0.08276753960556095,
      "grad_norm": 1.166627049446106,
      "learning_rate": 0.0009909945800260092,
      "loss": 0.7870283126831055,
      "step": 1792
    },
    {
      "epoch": 0.09459147383492679,
      "grad_norm": 0.7747617363929749,
      "learning_rate": 0.0009870460151900522,
      "loss": 0.7735522389411926,
      "step": 2048
    },
    {
      "epoch": 0.09459147383492679,
      "eval_bleu": 0.9264483676285591,
      "eval_ce_loss": 0.22021640589690372,
      "eval_cos_loss": 0.15934634184864557,
      "eval_loss": 0.751849727815689,
      "eval_mse_loss": 0.51190055103879,
      "eval_rec_loss": 0.0037981332031229603,
      "flow/cos_sim": 0.8406536742432477,
      "flow/improvement_ratio": 0.9754726998337871,
      "flow/mag_ratio_mean": 0.8395581131112085,
      "flow/mag_ratio_std": 0.09706963221096013,
      "step": 2048
    },
    {
      "epoch": 0.09459147383492679,
      "eval_bleu": 0.9264483676285591,
      "eval_ce_loss": 0.22021640589690372,
      "eval_cos_loss": 0.15934634184864557,
      "eval_loss": 0.751849727815689,
      "eval_mse_loss": 0.51190055103879,
      "eval_rec_loss": 0.0037981332031229603,
      "eval_runtime": 139.3758,
      "eval_samples_per_second": 200.845,
      "eval_steps_per_second": 3.143,
      "flow/cos_sim": 0.8406536742432477,
      "flow/improvement_ratio": 0.9754726998337871,
      "flow/mag_ratio_mean": 0.8395581131112085,
      "flow/mag_ratio_std": 0.09706963221096013,
      "step": 2048
    },
    {
      "epoch": 0.10641540806429264,
      "grad_norm": 1.3507742881774902,
      "learning_rate": 0.0009823925488998885,
      "loss": 0.7531520128250122,
      "step": 2304
    },
    {
      "epoch": 0.1182393422936585,
      "grad_norm": 1.090326189994812,
      "learning_rate": 0.0009770409161149525,
      "loss": 0.7384664416313171,
      "step": 2560
    },
    {
      "epoch": 0.13006327652302435,
      "grad_norm": 1.6648627519607544,
      "learning_rate": 0.0009709988622506973,
      "loss": 0.7159472107887268,
      "step": 2816
    },
    {
      "epoch": 0.1418872107523902,
      "grad_norm": 0.720405638217926,
      "learning_rate": 0.000964275131968659,
      "loss": 0.7134207487106323,
      "step": 3072
    },
    {
      "epoch": 0.1418872107523902,
      "eval_bleu": 0.9362046037171927,
      "eval_ce_loss": 0.18669388993094638,
      "eval_cos_loss": 0.09599785676829892,
      "eval_loss": 0.7009708360177741,
      "eval_mse_loss": 0.5027340843797274,
      "eval_rec_loss": 0.0019430727923829022,
      "flow/cos_sim": 0.9040021625555814,
      "flow/improvement_ratio": 0.9753212502830104,
      "flow/mag_ratio_mean": 0.906250716208323,
      "flow/mag_ratio_std": 0.07307879087519428,
      "step": 3072
    },
    {
      "epoch": 0.1418872107523902,
      "eval_bleu": 0.9362046037171927,
      "eval_ce_loss": 0.18669388993094638,
      "eval_cos_loss": 0.09599785676829892,
      "eval_loss": 0.7009708360177741,
      "eval_mse_loss": 0.5027340843797274,
      "eval_rec_loss": 0.0019430727923829022,
      "eval_runtime": 139.2617,
      "eval_samples_per_second": 201.01,
      "eval_steps_per_second": 3.145,
      "flow/cos_sim": 0.9040021625555814,
      "flow/improvement_ratio": 0.9753212502830104,
      "flow/mag_ratio_mean": 0.906250716208323,
      "flow/mag_ratio_std": 0.07307879087519428,
      "step": 3072
    },
    {
      "epoch": 0.15371114498175603,
      "grad_norm": 1.171225905418396,
      "learning_rate": 0.0009568794565203123,
      "loss": 0.6967981457710266,
      "step": 3328
    },
    {
      "epoch": 0.1655350792111219,
      "grad_norm": 0.9184499979019165,
      "learning_rate": 0.0009488225396630347,
      "loss": 0.6859588623046875,
      "step": 3584
    },
    {
      "epoch": 0.17735901344048774,
      "grad_norm": 1.0972322225570679,
      "learning_rate": 0.0009401160421685646,
      "loss": 0.68483966588974,
      "step": 3840
    },
    {
      "epoch": 0.18918294766985358,
      "grad_norm": 1.2944236993789673,
      "learning_rate": 0.0009307725649463714,
      "loss": 0.6722217202186584,
      "step": 4096
    },
    {
      "epoch": 0.18918294766985358,
      "eval_bleu": 0.9367529454304925,
      "eval_ce_loss": 0.17851220300531687,
      "eval_cos_loss": 0.0685597131088308,
      "eval_loss": 0.6638682314522191,
      "eval_mse_loss": 0.4771829223660029,
      "eval_rec_loss": 0.0013171346048419428,
      "flow/cos_sim": 0.9314403127045392,
      "flow/improvement_ratio": 0.9747336862021929,
      "flow/mag_ratio_mean": 0.9314602082722807,
      "flow/mag_ratio_std": 0.05983651272068013,
      "step": 4096
    },
    {
      "epoch": 0.18918294766985358,
      "eval_bleu": 0.9367529454304925,
      "eval_ce_loss": 0.17851220300531687,
      "eval_cos_loss": 0.0685597131088308,
      "eval_loss": 0.6638682314522191,
      "eval_mse_loss": 0.4771829223660029,
      "eval_rec_loss": 0.0013171346048419428,
      "eval_runtime": 143.247,
      "eval_samples_per_second": 195.418,
      "eval_steps_per_second": 3.058,
      "flow/cos_sim": 0.9314403127045392,
      "flow/improvement_ratio": 0.9747336862021929,
      "flow/mag_ratio_mean": 0.9314602082722807,
      "flow/mag_ratio_std": 0.05983651272068013,
      "step": 4096
    },
    {
      "epoch": 0.20100688189921945,
      "grad_norm": 1.0950664281845093,
      "learning_rate": 0.0009208056308063659,
      "loss": 0.6635431051254272,
      "step": 4352
    },
    {
      "epoch": 0.2128308161285853,
      "grad_norm": 1.0510311126708984,
      "learning_rate": 0.0009102296648873445,
      "loss": 0.652130126953125,
      "step": 4608
    },
    {
      "epoch": 0.22465475035795113,
      "grad_norm": 0.7107524275779724,
      "learning_rate": 0.0008990599737794927,
      "loss": 0.6548014283180237,
      "step": 4864
    },
    {
      "epoch": 0.236478684587317,
      "grad_norm": 1.119279146194458,
      "learning_rate": 0.0008873127233711644,
      "loss": 0.644295871257782,
      "step": 5120
    },
    {
      "epoch": 0.236478684587317,
      "eval_bleu": 0.9375087809294717,
      "eval_ce_loss": 0.17866700632629498,
      "eval_cos_loss": 0.05568918508379699,
      "eval_loss": 0.6384022356304404,
      "eval_mse_loss": 0.4531491862856634,
      "eval_rec_loss": 0.0010171248973892112,
      "flow/cos_sim": 0.9443108406785417,
      "flow/improvement_ratio": 0.9754358607612245,
      "flow/mag_ratio_mean": 0.9432625220790846,
      "flow/mag_ratio_std": 0.0532344374550532,
      "step": 5120
    },
    {
      "epoch": 0.236478684587317,
      "eval_bleu": 0.9375087809294717,
      "eval_ce_loss": 0.17866700632629498,
      "eval_cos_loss": 0.05568918508379699,
      "eval_loss": 0.6384022356304404,
      "eval_mse_loss": 0.4531491862856634,
      "eval_rec_loss": 0.0010171248973892112,
      "eval_runtime": 140.7376,
      "eval_samples_per_second": 198.902,
      "eval_steps_per_second": 3.112,
      "flow/cos_sim": 0.9443108406785417,
      "flow/improvement_ratio": 0.9754358607612245,
      "flow/mag_ratio_mean": 0.9432625220790846,
      "flow/mag_ratio_std": 0.0532344374550532,
      "step": 5120
    },
    {
      "epoch": 0.24830261881668284,
      "grad_norm": 1.223329782485962,
      "learning_rate": 0.0008750049154520011,
      "loss": 0.6385497450828552,
      "step": 5376
    },
    {
      "epoch": 0.2601265530460487,
      "grad_norm": 0.7951129078865051,
      "learning_rate": 0.0008621543631062487,
      "loss": 0.6350575089454651,
      "step": 5632
    },
    {
      "epoch": 0.27195048727541454,
      "grad_norm": 0.8830247521400452,
      "learning_rate": 0.0008487796649318904,
      "loss": 0.6269800066947937,
      "step": 5888
    },
    {
      "epoch": 0.2837744215047804,
      "grad_norm": 1.0399079322814941,
      "learning_rate": 0.0008349001781229053,
      "loss": 0.6236906051635742,
      "step": 6144
    },
    {
      "epoch": 0.2837744215047804,
      "eval_bleu": 0.9406288888554537,
      "eval_ce_loss": 0.16271350685439018,
      "eval_cos_loss": 0.04946032597696128,
      "eval_loss": 0.6158577807962078,
      "eval_mse_loss": 0.4472781540868489,
      "eval_rec_loss": 0.0009200886164281037,
      "flow/cos_sim": 0.9505396935765602,
      "flow/improvement_ratio": 0.9752184091365501,
      "flow/mag_ratio_mean": 0.9574754819205907,
      "flow/mag_ratio_std": 0.04891788842131014,
      "step": 6144
    },
    {
      "epoch": 0.2837744215047804,
      "eval_bleu": 0.9406288888554537,
      "eval_ce_loss": 0.16271350685439018,
      "eval_cos_loss": 0.04946032597696128,
      "eval_loss": 0.6158577807962078,
      "eval_mse_loss": 0.4472781540868489,
      "eval_rec_loss": 0.0009200886164281037,
      "eval_runtime": 138.7881,
      "eval_samples_per_second": 201.696,
      "eval_steps_per_second": 3.156,
      "flow/cos_sim": 0.9505396935765602,
      "flow/improvement_ratio": 0.9752184091365501,
      "flow/mag_ratio_mean": 0.9574754819205907,
      "flow/mag_ratio_std": 0.04891788842131014,
      "step": 6144
    },
    {
      "epoch": 0.2955983557341462,
      "grad_norm": 1.2804330587387085,
      "learning_rate": 0.0008205359904536107,
      "loss": 0.6104704737663269,
      "step": 6400
    },
    {
      "epoch": 0.30742228996351206,
      "grad_norm": 1.038807988166809,
      "learning_rate": 0.0008057078912056363,
      "loss": 0.6035579442977905,
      "step": 6656
    },
    {
      "epoch": 0.3192462241928779,
      "grad_norm": 1.1162539720535278,
      "learning_rate": 0.0007904373410796086,
      "loss": 0.6099694967269897,
      "step": 6912
    },
    {
      "epoch": 0.3310701584222438,
      "grad_norm": 0.8053554892539978,
      "learning_rate": 0.0007747464411350876,
      "loss": 0.6010444760322571,
      "step": 7168
    },
    {
      "epoch": 0.3310701584222438,
      "eval_bleu": 0.9403609007223014,
      "eval_ce_loss": 0.16489822024130793,
      "eval_cos_loss": 0.042615741474307293,
      "eval_loss": 0.5965314044799979,
      "eval_mse_loss": 0.4266453656839998,
      "eval_rec_loss": 0.0007262465627901213,
      "flow/cos_sim": 0.9573842790573155,
      "flow/improvement_ratio": 0.9754182146564466,
      "flow/mag_ratio_mean": 0.9583017167435389,
      "flow/mag_ratio_std": 0.044361623012584096,
      "step": 7168
    },
    {
      "epoch": 0.3310701584222438,
      "eval_bleu": 0.9403609007223014,
      "eval_ce_loss": 0.16489822024130793,
      "eval_cos_loss": 0.042615741474307293,
      "eval_loss": 0.5965314044799979,
      "eval_mse_loss": 0.4266453656839998,
      "eval_rec_loss": 0.0007262465627901213,
      "eval_runtime": 141.0373,
      "eval_samples_per_second": 198.479,
      "eval_steps_per_second": 3.106,
      "flow/cos_sim": 0.9573842790573155,
      "flow/improvement_ratio": 0.9754182146564466,
      "flow/mag_ratio_mean": 0.9583017167435389,
      "flow/mag_ratio_std": 0.044361623012584096,
      "step": 7168
    },
    {
      "epoch": 0.34289409265160964,
      "grad_norm": 0.926774799823761,
      "learning_rate": 0.000758657900803716,
      "loss": 0.6045265793800354,
      "step": 7424
    },
    {
      "epoch": 0.3547180268809755,
      "grad_norm": 0.6091651320457458,
      "learning_rate": 0.000742195005021869,
      "loss": 0.6046400666236877,
      "step": 7680
    },
    {
      "epoch": 0.3665419611103413,
      "grad_norm": 0.9995866417884827,
      "learning_rate": 0.0007253815805303786,
      "loss": 0.5923656225204468,
      "step": 7936
    },
    {
      "epoch": 0.37836589533970716,
      "grad_norm": 0.8947123885154724,
      "learning_rate": 0.0007082419613901028,
      "loss": 0.5886037349700928,
      "step": 8192
    },
    {
      "epoch": 0.37836589533970716,
      "eval_bleu": 0.9428308469149022,
      "eval_ce_loss": 0.15756705103943883,
      "eval_cos_loss": 0.03951380795641849,
      "eval_loss": 0.5839528071281572,
      "eval_mse_loss": 0.42179430878325685,
      "eval_rec_loss": 0.0006400666907252946,
      "flow/cos_sim": 0.9604862126857723,
      "flow/improvement_ratio": 0.9746591260988419,
      "flow/mag_ratio_mean": 0.9592139723638421,
      "flow/mag_ratio_std": 0.04008094550505893,
      "step": 8192
    },
    {
      "epoch": 0.37836589533970716,
      "eval_bleu": 0.9428308469149022,
      "eval_ce_loss": 0.15756705103943883,
      "eval_cos_loss": 0.03951380795641849,
      "eval_loss": 0.5839528071281572,
      "eval_mse_loss": 0.42179430878325685,
      "eval_rec_loss": 0.0006400666907252946,
      "eval_runtime": 142.8618,
      "eval_samples_per_second": 195.945,
      "eval_steps_per_second": 3.066,
      "flow/cos_sim": 0.9604862126857723,
      "flow/improvement_ratio": 0.9746591260988419,
      "flow/mag_ratio_mean": 0.9592139723638421,
      "flow/mag_ratio_std": 0.04008094550505893,
      "step": 8192
    },
    {
      "epoch": 0.390189829569073,
      "grad_norm": 0.9762691259384155,
      "learning_rate": 0.0006908009537632514,
      "loss": 0.591374397277832,
      "step": 8448
    },
    {
      "epoch": 0.4020137637984389,
      "grad_norm": 0.9119466543197632,
      "learning_rate": 0.0006730838000114403,
      "loss": 0.5883693099021912,
      "step": 8704
    },
    {
      "epoch": 0.41383769802780473,
      "grad_norm": 0.7088457942008972,
      "learning_rate": 0.0006551161421624341,
      "loss": 0.591505229473114,
      "step": 8960
    },
    {
      "epoch": 0.4256616322571706,
      "grad_norm": 0.962149441242218,
      "learning_rate": 0.0006369239847984517,
      "loss": 0.5791484117507935,
      "step": 9216
    },
    {
      "epoch": 0.4256616322571706,
      "eval_bleu": 0.9397317595032229,
      "eval_ce_loss": 0.1654567693755643,
      "eval_cos_loss": 0.036672262636493876,
      "eval_loss": 0.5844078893394775,
      "eval_mse_loss": 0.41475171422305174,
      "eval_rec_loss": 0.0005321793435803258,
      "flow/cos_sim": 0.9633277582523485,
      "flow/improvement_ratio": 0.9749675198504913,
      "flow/mag_ratio_mean": 0.9673667468436776,
      "flow/mag_ratio_std": 0.03610456871829893,
      "step": 9216
    },
    {
      "epoch": 0.4256616322571706,
      "eval_bleu": 0.9397317595032229,
      "eval_ce_loss": 0.1654567693755643,
      "eval_cos_loss": 0.036672262636493876,
      "eval_loss": 0.5844078893394775,
      "eval_mse_loss": 0.41475171422305174,
      "eval_rec_loss": 0.0005321793435803258,
      "eval_runtime": 140.1978,
      "eval_samples_per_second": 199.668,
      "eval_steps_per_second": 3.124,
      "flow/cos_sim": 0.9633277582523485,
      "flow/improvement_ratio": 0.9749675198504913,
      "flow/mag_ratio_mean": 0.9673667468436776,
      "flow/mag_ratio_std": 0.03610456871829893,
      "step": 9216
    },
    {
      "epoch": 0.4374855664865364,
      "grad_norm": 0.853702187538147,
      "learning_rate": 0.0006185336574197479,
      "loss": 0.5742002725601196,
      "step": 9472
    },
    {
      "epoch": 0.44930950071590225,
      "grad_norm": 0.9982457160949707,
      "learning_rate": 0.0005999717763379407,
      "loss": 0.5812740921974182,
      "step": 9728
    },
    {
      "epoch": 0.4611334349452681,
      "grad_norm": 1.16475510597229,
      "learning_rate": 0.0005812652061542363,
      "loss": 0.5733819603919983,
      "step": 9984
    },
    {
      "epoch": 0.472957369174634,
      "grad_norm": 1.0706837177276611,
      "learning_rate": 0.0005624410208783071,
      "loss": 0.5738887190818787,
      "step": 10240
    },
    {
      "epoch": 0.472957369174634,
      "eval_bleu": 0.9459802582441368,
      "eval_ce_loss": 0.14516126815459296,
      "eval_cos_loss": 0.033954130285780995,
      "eval_loss": 0.5643789048336413,
      "eval_mse_loss": 0.4153396790702593,
      "eval_rec_loss": 0.0004825431066599768,
      "flow/cos_sim": 0.9660458944431723,
      "flow/improvement_ratio": 0.9768404536051293,
      "flow/mag_ratio_mean": 0.964692687198996,
      "flow/mag_ratio_std": 0.033606582049059266,
      "step": 10240
    },
    {
      "epoch": 0.472957369174634,
      "eval_bleu": 0.9459802582441368,
      "eval_ce_loss": 0.14516126815459296,
      "eval_cos_loss": 0.033954130285780995,
      "eval_loss": 0.5643789048336413,
      "eval_mse_loss": 0.4153396790702593,
      "eval_rec_loss": 0.0004825431066599768,
      "eval_runtime": 138.7003,
      "eval_samples_per_second": 201.824,
      "eval_steps_per_second": 3.158,
      "flow/cos_sim": 0.9660458944431723,
      "flow/improvement_ratio": 0.9768404536051293,
      "flow/mag_ratio_mean": 0.964692687198996,
      "flow/mag_ratio_std": 0.033606582049059266,
      "step": 10240
    },
    {
      "epoch": 0.48478130340399983,
      "grad_norm": 1.2897191047668457,
      "learning_rate": 0.0005435264647440881,
      "loss": 0.5754253268241882,
      "step": 10496
    },
    {
      "epoch": 0.49660523763336567,
      "grad_norm": 0.8889420628547668,
      "learning_rate": 0.000524548912779213,
      "loss": 0.5645499229431152,
      "step": 10752
    },
    {
      "epoch": 0.5084291718627315,
      "grad_norm": 1.0012623071670532,
      "learning_rate": 0.0005055358311851499,
      "loss": 0.570586085319519,
      "step": 11008
    },
    {
      "epoch": 0.5202531060920974,
      "grad_norm": 0.7268548011779785,
      "learning_rate": 0.0004865147375853812,
      "loss": 0.5649828910827637,
      "step": 11264
    },
    {
      "epoch": 0.5202531060920974,
      "eval_bleu": 0.9438391043268173,
      "eval_ce_loss": 0.15168385392003883,
      "eval_cos_loss": 0.031372046842320596,
      "eval_loss": 0.5559442507349737,
      "eval_mse_loss": 0.40068320514948946,
      "eval_rec_loss": 0.0004399877304908156,
      "flow/cos_sim": 0.9686279725538541,
      "flow/improvement_ratio": 0.9761811211773249,
      "flow/mag_ratio_mean": 0.9699762897676529,
      "flow/mag_ratio_std": 0.031497096553546926,
      "step": 11264
    },
    {
      "epoch": 0.5202531060920974,
      "eval_bleu": 0.9438391043268173,
      "eval_ce_loss": 0.15168385392003883,
      "eval_cos_loss": 0.031372046842320596,
      "eval_loss": 0.5559442507349737,
      "eval_mse_loss": 0.40068320514948946,
      "eval_rec_loss": 0.0004399877304908156,
      "eval_runtime": 142.7289,
      "eval_samples_per_second": 196.127,
      "eval_steps_per_second": 3.069,
      "flow/cos_sim": 0.9686279725538541,
      "flow/improvement_ratio": 0.9761811211773249,
      "flow/mag_ratio_mean": 0.9699762897676529,
      "flow/mag_ratio_std": 0.031497096553546926,
      "step": 11264
    },
    {
      "epoch": 0.5320770403214632,
      "grad_norm": 1.215234637260437,
      "learning_rate": 0.0004675131611991607,
      "loss": 0.5652971863746643,
      "step": 11520
    },
    {
      "epoch": 0.5439009745508291,
      "grad_norm": 0.622643768787384,
      "learning_rate": 0.0004485586029984899,
      "loss": 0.5565246343612671,
      "step": 11776
    },
    {
      "epoch": 0.5557249087801949,
      "grad_norm": 1.0961378812789917,
      "learning_rate": 0.00042967849590597266,
      "loss": 0.5489715933799744,
      "step": 12032
    },
    {
      "epoch": 0.5675488430095608,
      "grad_norm": 0.9782881140708923,
      "learning_rate": 0.0004109001650911621,
      "loss": 0.5525087118148804,
      "step": 12288
    },
    {
      "epoch": 0.5675488430095608,
      "eval_bleu": 0.9412862694661508,
      "eval_ce_loss": 0.1635204538911343,
      "eval_cos_loss": 0.030069323277874895,
      "eval_loss": 0.5582960859689539,
      "eval_mse_loss": 0.39135666776737665,
      "eval_rec_loss": 0.00041203244906421176,
      "flow/cos_sim": 0.969930695344324,
      "flow/improvement_ratio": 0.9753037164472553,
      "flow/mag_ratio_mean": 0.9802452540833112,
      "flow/mag_ratio_std": 0.030909983262623827,
      "step": 12288
    },
    {
      "epoch": 0.5675488430095608,
      "eval_bleu": 0.9412862694661508,
      "eval_ce_loss": 0.1635204538911343,
      "eval_cos_loss": 0.030069323277874895,
      "eval_loss": 0.5582960859689539,
      "eval_mse_loss": 0.39135666776737665,
      "eval_rec_loss": 0.00041203244906421176,
      "eval_runtime": 141.5788,
      "eval_samples_per_second": 197.72,
      "eval_steps_per_second": 3.094,
      "flow/cos_sim": 0.969930695344324,
      "flow/improvement_ratio": 0.9753037164472553,
      "flow/mag_ratio_mean": 0.9802452540833112,
      "flow/mag_ratio_std": 0.030909983262623827,
      "step": 12288
    },
    {
      "epoch": 0.5793727772389267,
      "grad_norm": 1.1210603713989258,
      "learning_rate": 0.0003922507884228551,
      "loss": 0.5510575175285339,
      "step": 12544
    },
    {
      "epoch": 0.5911967114682924,
      "grad_norm": 0.7037401795387268,
      "learning_rate": 0.00037375735713457723,
      "loss": 0.5481619238853455,
      "step": 12800
    },
    {
      "epoch": 0.6030206456976583,
      "grad_norm": 0.7474592924118042,
      "learning_rate": 0.00035544663676018276,
      "loss": 0.5506067872047424,
      "step": 13056
    },
    {
      "epoch": 0.6148445799270241,
      "grad_norm": 0.7188865542411804,
      "learning_rate": 0.00033734512839611255,
      "loss": 0.5451632738113403,
      "step": 13312
    },
    {
      "epoch": 0.6148445799270241,
      "eval_bleu": 0.9428834105382213,
      "eval_ce_loss": 0.1552433374967239,
      "eval_cos_loss": 0.02893113123771807,
      "eval_loss": 0.5456167366951024,
      "eval_mse_loss": 0.3870963177316265,
      "eval_rec_loss": 0.0003839662346667844,
      "flow/cos_sim": 0.9710688916243375,
      "flow/improvement_ratio": 0.9738836331998921,
      "flow/mag_ratio_mean": 0.9706440245451993,
      "flow/mag_ratio_std": 0.029164127312328446,
      "step": 13312
    },
    {
      "epoch": 0.6148445799270241,
      "eval_bleu": 0.9428834105382213,
      "eval_ce_loss": 0.1552433374967239,
      "eval_cos_loss": 0.02893113123771807,
      "eval_loss": 0.5456167366951024,
      "eval_mse_loss": 0.3870963177316265,
      "eval_rec_loss": 0.0003839662346667844,
      "eval_runtime": 141.1326,
      "eval_samples_per_second": 198.345,
      "eval_steps_per_second": 3.103,
      "flow/cos_sim": 0.9710688916243375,
      "flow/improvement_ratio": 0.9738836331998921,
      "flow/mag_ratio_mean": 0.9706440245451993,
      "flow/mag_ratio_std": 0.029164127312328446,
      "step": 13312
    },
    {
      "epoch": 0.62666851415639,
      "grad_norm": 0.7716706991195679,
      "learning_rate": 0.0003194790303463687,
      "loss": 0.537281334400177,
      "step": 13568
    },
    {
      "epoch": 0.6384924483857558,
      "grad_norm": 1.332189917564392,
      "learning_rate": 0.00030187420020572406,
      "loss": 0.5493588447570801,
      "step": 13824
    },
    {
      "epoch": 0.6503163826151217,
      "grad_norm": 0.7042582035064697,
      "learning_rate": 0.00028455611743603626,
      "loss": 0.5357389450073242,
      "step": 14080
    },
    {
      "epoch": 0.6621403168444876,
      "grad_norm": 0.9820289611816406,
      "learning_rate": 0.0002675498464898373,
      "loss": 0.5378322601318359,
      "step": 14336
    },
    {
      "epoch": 0.6621403168444876,
      "eval_bleu": 0.9466625254289972,
      "eval_ce_loss": 0.14409655002562421,
      "eval_cos_loss": 0.028247294284097137,
      "eval_loss": 0.5304203147757544,
      "eval_mse_loss": 0.38314900074375274,
      "eval_rec_loss": 0.0003500353127845417,
      "flow/cos_sim": 0.9717527268684074,
      "flow/improvement_ratio": 0.9754393091212669,
      "flow/mag_ratio_mean": 0.973318548915593,
      "flow/mag_ratio_std": 0.028512526988200674,
      "step": 14336
    },
    {
      "epoch": 0.6621403168444876,
      "eval_bleu": 0.9466625254289972,
      "eval_ce_loss": 0.14409655002562421,
      "eval_cos_loss": 0.028247294284097137,
      "eval_loss": 0.5304203147757544,
      "eval_mse_loss": 0.38314900074375274,
      "eval_rec_loss": 0.0003500353127845417,
      "eval_runtime": 142.684,
      "eval_samples_per_second": 196.189,
      "eval_steps_per_second": 3.07,
      "flow/cos_sim": 0.9717527268684074,
      "flow/improvement_ratio": 0.9754393091212669,
      "flow/mag_ratio_mean": 0.973318548915593,
      "flow/mag_ratio_std": 0.028512526988200674,
      "step": 14336
    },
    {
      "epoch": 0.6739642510738534,
      "grad_norm": 0.9396564364433289,
      "learning_rate": 0.0002508800005345623,
      "loss": 0.5384619832038879,
      "step": 14592
    },
    {
      "epoch": 0.6857881853032193,
      "grad_norm": 1.3034203052520752,
      "learning_rate": 0.00023457070582992562,
      "loss": 0.5381016135215759,
      "step": 14848
    },
    {
      "epoch": 0.6976121195325851,
      "grad_norm": 0.7913278341293335,
      "learning_rate": 0.00021864556680999692,
      "loss": 0.5290021896362305,
      "step": 15104
    },
    {
      "epoch": 0.709436053761951,
      "grad_norm": 0.9535462856292725,
      "learning_rate": 0.0002031276319205152,
      "loss": 0.534302830696106,
      "step": 15360
    },
    {
      "epoch": 0.709436053761951,
      "eval_bleu": 0.9467887438059129,
      "eval_ce_loss": 0.14362057716241233,
      "eval_cos_loss": 0.02781058812607505,
      "eval_loss": 0.5298489093372266,
      "eval_mse_loss": 0.38311818289702343,
      "eval_rec_loss": 0.00032909089600148805,
      "flow/cos_sim": 0.9721894303685454,
      "flow/improvement_ratio": 0.974593520844908,
      "flow/mag_ratio_mean": 0.9698141316572825,
      "flow/mag_ratio_std": 0.02832854554465372,
      "step": 15360
    },
    {
      "epoch": 0.709436053761951,
      "eval_bleu": 0.9467887438059129,
      "eval_ce_loss": 0.14362057716241233,
      "eval_cos_loss": 0.02781058812607505,
      "eval_loss": 0.5298489093372266,
      "eval_mse_loss": 0.38311818289702343,
      "eval_rec_loss": 0.00032909089600148805,
      "eval_runtime": 141.9779,
      "eval_samples_per_second": 197.164,
      "eval_steps_per_second": 3.085,
      "flow/cos_sim": 0.9721894303685454,
      "flow/improvement_ratio": 0.974593520844908,
      "flow/mag_ratio_mean": 0.9698141316572825,
      "flow/mag_ratio_std": 0.02832854554465372,
      "step": 15360
    }
  ],
  "logging_steps": 256,
  "max_steps": 21651,
  "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
}