{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9267567392850217,
  "eval_steps": 1024,
  "global_step": 21504,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.011032818324821687,
      "grad_norm": 11.799630165100098,
      "learning_rate": 0.000498046875,
      "loss": 104.40645599365234,
      "step": 256
    },
    {
      "epoch": 0.022065636649643373,
      "grad_norm": 24.033342361450195,
      "learning_rate": 0.000998046875,
      "loss": 2.2942047119140625,
      "step": 512
    },
    {
      "epoch": 0.03309845497446506,
      "grad_norm": 21.89491081237793,
      "learning_rate": 0.000999688448778502,
      "loss": 2.301217555999756,
      "step": 768
    },
    {
      "epoch": 0.04413127329928675,
      "grad_norm": 24.77373695373535,
      "learning_rate": 0.0009987492950653055,
      "loss": 2.2358601093292236,
      "step": 1024
    },
    {
      "epoch": 0.04413127329928675,
      "eval_bleu": 0.9163079254973637,
      "eval_cos_loss": 0.8463371174930255,
      "eval_dec_loss": 0.15604417777455437,
      "eval_loss": 2.1683921397113597,
      "eval_mse2_loss": 0.2577589998430789,
      "eval_mse_loss": 1.9106331385020763,
      "eval_rec_loss": 0.047009017791098624,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.1536628818556444,
      "flow/improvement_ratio": 0.5908637553898256,
      "flow/mag_ratio_mean": 0.2672089617262517,
      "flow/mag_ratio_std": 0.1647362987052149,
      "step": 1024
    },
    {
      "epoch": 0.04413127329928675,
      "eval_bleu": 0.9163079254973637,
      "eval_cos_loss": 0.8463371174930255,
      "eval_dec_loss": 0.15604417777455437,
      "eval_loss": 2.1683921397113597,
      "eval_mse2_loss": 0.2577589998430789,
      "eval_mse_loss": 1.9106331385020763,
      "eval_rec_loss": 0.047009017791098624,
      "eval_runtime": 151.0019,
      "eval_samples_per_second": 198.673,
      "eval_steps_per_second": 3.106,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.1536628818556444,
      "flow/improvement_ratio": 0.5908637553898256,
      "flow/mag_ratio_mean": 0.2672089617262517,
      "flow/mag_ratio_std": 0.1647362987052149,
      "step": 1024
    },
    {
      "epoch": 0.05516409162410843,
      "grad_norm": 20.877023696899414,
      "learning_rate": 0.0009971837136430763,
      "loss": 2.162022352218628,
      "step": 1280
    },
    {
      "epoch": 0.06619690994893011,
      "grad_norm": 16.984447479248047,
      "learning_rate": 0.0009949936708776692,
      "loss": 1.9889535903930664,
      "step": 1536
    },
    {
      "epoch": 0.07722972827375181,
      "grad_norm": 18.61001205444336,
      "learning_rate": 0.0009921819174566252,
      "loss": 1.8391456604003906,
      "step": 1792
    },
    {
      "epoch": 0.0882625465985735,
      "grad_norm": 9.276031494140625,
      "learning_rate": 0.000988751984934317,
      "loss": 1.736743688583374,
      "step": 2048
    },
    {
      "epoch": 0.0882625465985735,
      "eval_bleu": 0.9323255434649543,
      "eval_cos_loss": 0.5421419106821007,
      "eval_dec_loss": 0.12311239358681098,
      "eval_loss": 1.6836593265472444,
      "eval_mse2_loss": 0.19289209935138982,
      "eval_mse_loss": 1.4907672278154125,
      "eval_rec_loss": 0.047009017791098624,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.4578580862677682,
      "flow/improvement_ratio": 0.8665720767049647,
      "flow/mag_ratio_mean": 0.49134117085288076,
      "flow/mag_ratio_std": 0.21147727505611713,
      "step": 2048
    },
    {
      "epoch": 0.0882625465985735,
      "eval_bleu": 0.9323255434649543,
      "eval_cos_loss": 0.5421419106821007,
      "eval_dec_loss": 0.12311239358681098,
      "eval_loss": 1.6836593265472444,
      "eval_mse2_loss": 0.19289209935138982,
      "eval_mse_loss": 1.4907672278154125,
      "eval_rec_loss": 0.047009017791098624,
      "eval_runtime": 148.8988,
      "eval_samples_per_second": 201.479,
      "eval_steps_per_second": 3.15,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.4578580862677682,
      "flow/improvement_ratio": 0.8665720767049647,
      "flow/mag_ratio_mean": 0.49134117085288076,
      "flow/mag_ratio_std": 0.21147727505611713,
      "step": 2048
    },
    {
      "epoch": 0.09929536492339518,
      "grad_norm": 7.179141044616699,
      "learning_rate": 0.0009847081812963268,
      "loss": 1.6638906002044678,
      "step": 2304
    },
    {
      "epoch": 0.11032818324821686,
      "grad_norm": 5.170185089111328,
      "learning_rate": 0.0009800555855486275,
      "loss": 1.6207516193389893,
      "step": 2560
    },
    {
      "epoch": 0.12136100157303854,
      "grad_norm": 5.038082122802734,
      "learning_rate": 0.0009748000413383664,
      "loss": 1.5859862565994263,
      "step": 2816
    },
    {
      "epoch": 0.13239381989786023,
      "grad_norm": 3.0588953495025635,
      "learning_rate": 0.0009689481496142604,
      "loss": 1.5566623210906982,
      "step": 3072
    },
    {
      "epoch": 0.13239381989786023,
      "eval_bleu": 0.9326258807918764,
      "eval_cos_loss": 0.48918826811349214,
      "eval_dec_loss": 0.12837675790082037,
      "eval_loss": 1.5384274639808801,
      "eval_mse2_loss": 0.17169118979211045,
      "eval_mse_loss": 1.3667362735215536,
      "eval_rec_loss": 0.047009017791098624,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.5108117323313186,
      "flow/improvement_ratio": 0.8871580227605824,
      "flow/mag_ratio_mean": 0.5184402796886624,
      "flow/mag_ratio_std": 0.24528262540221468,
      "step": 3072
    },
    {
      "epoch": 0.13239381989786023,
      "eval_bleu": 0.9326258807918764,
      "eval_cos_loss": 0.48918826811349214,
      "eval_dec_loss": 0.12837675790082037,
      "eval_loss": 1.5384274639808801,
      "eval_mse2_loss": 0.17169118979211045,
      "eval_mse_loss": 1.3667362735215536,
      "eval_rec_loss": 0.047009017791098624,
      "eval_runtime": 149.1668,
      "eval_samples_per_second": 201.117,
      "eval_steps_per_second": 3.144,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.5108117323313186,
      "flow/improvement_ratio": 0.8871580227605824,
      "flow/mag_ratio_mean": 0.5184402796886624,
      "flow/mag_ratio_std": 0.24528262540221468,
      "step": 3072
    },
    {
      "epoch": 0.14342663822268192,
      "grad_norm": 2.7404696941375732,
      "learning_rate": 0.0009625072603358231,
      "loss": 1.5345162153244019,
      "step": 3328
    },
    {
      "epoch": 0.15445945654750362,
      "grad_norm": 2.8666281700134277,
      "learning_rate": 0.0009554854632418371,
      "loss": 1.518531084060669,
      "step": 3584
    },
    {
      "epoch": 0.1654922748723253,
      "grad_norm": 1.7333565950393677,
      "learning_rate": 0.000947891577689663,
      "loss": 1.5072828531265259,
      "step": 3840
    },
    {
      "epoch": 0.176525093197147,
      "grad_norm": 2.1371614933013916,
      "learning_rate": 0.0009397351415781539,
      "loss": 1.4991329908370972,
      "step": 4096
    },
    {
      "epoch": 0.176525093197147,
      "eval_bleu": 0.9359739466567191,
      "eval_cos_loss": 0.47476763149568524,
      "eval_dec_loss": 0.11913440864024831,
      "eval_loss": 1.4943630281033546,
      "eval_mse2_loss": 0.16356163808722485,
      "eval_mse_loss": 1.3308013893647996,
      "eval_rec_loss": 0.047009017791098624,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.5252323670427936,
      "flow/improvement_ratio": 0.8878522331018184,
      "flow/mag_ratio_mean": 0.5341137046498785,
      "flow/mag_ratio_std": 0.25290533919324243,
      "step": 4096
    },
    {
      "epoch": 0.176525093197147,
      "eval_bleu": 0.9359739466567191,
      "eval_cos_loss": 0.47476763149568524,
      "eval_dec_loss": 0.11913440864024831,
      "eval_loss": 1.4943630281033546,
      "eval_mse2_loss": 0.16356163808722485,
      "eval_mse_loss": 1.3308013893647996,
      "eval_rec_loss": 0.047009017791098624,
      "eval_runtime": 149.7517,
      "eval_samples_per_second": 200.332,
      "eval_steps_per_second": 3.132,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.5252323670427936,
      "flow/improvement_ratio": 0.8878522331018184,
      "flow/mag_ratio_mean": 0.5341137046498785,
      "flow/mag_ratio_std": 0.25290533919324243,
      "step": 4096
    },
    {
      "epoch": 0.18755791152196866,
      "grad_norm": 1.6938560009002686,
      "learning_rate": 0.000931026399368079,
      "loss": 1.4890222549438477,
      "step": 4352
    },
    {
      "epoch": 0.19859072984679035,
      "grad_norm": 1.565247893333435,
      "learning_rate": 0.0009217762892151117,
      "loss": 1.4860106706619263,
      "step": 4608
    },
    {
      "epoch": 0.20962354817161205,
      "grad_norm": 1.3487786054611206,
      "learning_rate": 0.0009119964292315354,
      "loss": 1.4800715446472168,
      "step": 4864
    },
    {
      "epoch": 0.22065636649643372,
      "grad_norm": 1.212808609008789,
      "learning_rate": 0.0009016991028939279,
      "loss": 1.4720630645751953,
      "step": 5120
    },
    {
      "epoch": 0.22065636649643372,
      "eval_bleu": 0.9371513693794589,
      "eval_cos_loss": 0.46655392564181836,
      "eval_dec_loss": 0.11585041659592248,
      "eval_loss": 1.4702420382103178,
      "eval_mse2_loss": 0.15908411601141317,
      "eval_mse_loss": 1.3111579230090957,
      "eval_rec_loss": 0.047009017791098624,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.5334460733414713,
      "flow/improvement_ratio": 0.8902226530158444,
      "flow/mag_ratio_mean": 0.5489529184441069,
      "flow/mag_ratio_std": 0.2660856873178279,
      "step": 5120
    },
    {
      "epoch": 0.22065636649643372,
      "eval_bleu": 0.9371513693794589,
      "eval_cos_loss": 0.46655392564181836,
      "eval_dec_loss": 0.11585041659592248,
      "eval_loss": 1.4702420382103178,
      "eval_mse2_loss": 0.15908411601141317,
      "eval_mse_loss": 1.3111579230090957,
      "eval_rec_loss": 0.047009017791098624,
      "eval_runtime": 153.0228,
      "eval_samples_per_second": 196.049,
      "eval_steps_per_second": 3.065,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.5334460733414713,
      "flow/improvement_ratio": 0.8902226530158444,
      "flow/mag_ratio_mean": 0.5489529184441069,
      "flow/mag_ratio_std": 0.2660856873178279,
      "step": 5120
    },
    {
      "epoch": 0.23168918482125542,
      "grad_norm": 1.2820218801498413,
      "learning_rate": 0.0008908972436151494,
      "loss": 1.4665789604187012,
      "step": 5376
    },
    {
      "epoch": 0.2427220031460771,
      "grad_norm": 1.428142786026001,
      "learning_rate": 0.0008796044185000127,
      "loss": 1.4617829322814941,
      "step": 5632
    },
    {
      "epoch": 0.2537548214708988,
      "grad_norm": 1.7342983484268188,
      "learning_rate": 0.0008678348113050368,
      "loss": 1.454346776008606,
      "step": 5888
    },
    {
      "epoch": 0.26478763979572045,
      "grad_norm": 1.012923240661621,
      "learning_rate": 0.0008556032046236897,
      "loss": 1.4543688297271729,
      "step": 6144
    },
    {
      "epoch": 0.26478763979572045,
      "eval_bleu": 0.9360335622230687,
      "eval_cos_loss": 0.4603669429892924,
      "eval_dec_loss": 0.11954856799669199,
      "eval_loss": 1.452896745982709,
      "eval_mse2_loss": 0.15859206668984915,
      "eval_mse_loss": 1.2943046784350105,
      "eval_rec_loss": 0.047009017791098624,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.5396330592347615,
      "flow/improvement_ratio": 0.8903370405565193,
      "flow/mag_ratio_mean": 0.5503829778003286,
      "flow/mag_ratio_std": 0.26988969053795087,
      "step": 6144
    },
    {
      "epoch": 0.26478763979572045,
      "eval_bleu": 0.9360335622230687,
      "eval_cos_loss": 0.4603669429892924,
      "eval_dec_loss": 0.11954856799669199,
      "eval_loss": 1.452896745982709,
      "eval_mse2_loss": 0.15859206668984915,
      "eval_mse_loss": 1.2943046784350105,
      "eval_rec_loss": 0.047009017791098624,
      "eval_runtime": 152.3001,
      "eval_samples_per_second": 196.979,
      "eval_steps_per_second": 3.079,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.5396330592347615,
      "flow/improvement_ratio": 0.8903370405565193,
      "flow/mag_ratio_mean": 0.5503829778003286,
      "flow/mag_ratio_std": 0.26988969053795087,
      "step": 6144
    },
    {
      "epoch": 0.2758204581205422,
      "grad_norm": 1.2126809358596802,
      "learning_rate": 0.000842924961319492,
      "loss": 1.4489810466766357,
      "step": 6400
    },
    {
      "epoch": 0.28685327644536385,
      "grad_norm": 1.6154913902282715,
      "learning_rate": 0.0008298160052303045,
      "loss": 1.4455806016921997,
      "step": 6656
    },
    {
      "epoch": 0.2978860947701855,
      "grad_norm": 1.1049504280090332,
      "learning_rate": 0.0008162928011680314,
      "loss": 1.4431121349334717,
      "step": 6912
    },
    {
      "epoch": 0.30891891309500724,
      "grad_norm": 1.161877989768982,
      "learning_rate": 0.000802372334238864,
      "loss": 1.440521478652954,
      "step": 7168
    },
    {
      "epoch": 0.30891891309500724,
      "eval_bleu": 0.9368360477187275,
      "eval_cos_loss": 0.4548988695592006,
      "eval_dec_loss": 0.11761775160673013,
      "eval_loss": 1.4366158300371312,
      "eval_mse2_loss": 0.1551807067636996,
      "eval_mse_loss": 1.2814351221137463,
      "eval_rec_loss": 0.047009017791098624,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.545101128788645,
      "flow/improvement_ratio": 0.8925937072300454,
      "flow/mag_ratio_mean": 0.5560947322387939,
      "flow/mag_ratio_std": 0.26709621954065904,
      "step": 7168
    },
    {
      "epoch": 0.30891891309500724,
      "eval_bleu": 0.9368360477187275,
      "eval_cos_loss": 0.4548988695592006,
      "eval_dec_loss": 0.11761775160673013,
      "eval_loss": 1.4366158300371312,
      "eval_mse2_loss": 0.1551807067636996,
      "eval_mse_loss": 1.2814351221137463,
      "eval_rec_loss": 0.047009017791098624,
      "eval_runtime": 150.401,
      "eval_samples_per_second": 199.467,
      "eval_steps_per_second": 3.118,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.545101128788645,
      "flow/improvement_ratio": 0.8925937072300454,
      "flow/mag_ratio_mean": 0.5560947322387939,
      "flow/mag_ratio_std": 0.26709621954065904,
      "step": 7168
    },
    {
      "epoch": 0.3199517314198289,
      "grad_norm": 1.3750261068344116,
      "learning_rate": 0.0007880720885100349,
      "loss": 1.4386309385299683,
      "step": 7424
    },
    {
      "epoch": 0.3309845497446506,
      "grad_norm": 0.9077029228210449,
      "learning_rate": 0.0007734100250498788,
      "loss": 1.4343377351760864,
      "step": 7680
    },
    {
      "epoch": 0.3420173680694723,
      "grad_norm": 1.2454605102539062,
      "learning_rate": 0.000758404559368781,
      "loss": 1.4340428113937378,
      "step": 7936
    },
    {
      "epoch": 0.353050186394294,
      "grad_norm": 1.4264285564422607,
      "learning_rate": 0.0007430745382893488,
      "loss": 1.4275590181350708,
      "step": 8192
    },
    {
      "epoch": 0.353050186394294,
      "eval_bleu": 0.9352995295168319,
      "eval_cos_loss": 0.4525942064043301,
      "eval_dec_loss": 0.12027457510150953,
      "eval_loss": 1.4305989170379476,
      "eval_mse2_loss": 0.15546214358130497,
      "eval_mse_loss": 1.2751367725034766,
      "eval_rec_loss": 0.047009017791098624,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.547405791689338,
      "flow/improvement_ratio": 0.8944777580720784,
      "flow/mag_ratio_mean": 0.5549619147009941,
      "flow/mag_ratio_std": 0.2656229132973055,
      "step": 8192
    },
    {
      "epoch": 0.353050186394294,
      "eval_bleu": 0.9352995295168319,
      "eval_cos_loss": 0.4525942064043301,
      "eval_dec_loss": 0.12027457510150953,
      "eval_loss": 1.4305989170379476,
      "eval_mse2_loss": 0.15546214358130497,
      "eval_mse_loss": 1.2751367725034766,
      "eval_rec_loss": 0.047009017791098624,
      "eval_runtime": 153.0138,
      "eval_samples_per_second": 196.061,
      "eval_steps_per_second": 3.065,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.547405791689338,
      "flow/improvement_ratio": 0.8944777580720784,
      "flow/mag_ratio_mean": 0.5549619147009941,
      "flow/mag_ratio_std": 0.2656229132973055,
      "step": 8192
    },
    {
      "epoch": 0.36408300471911564,
      "grad_norm": 1.2861735820770264,
      "learning_rate": 0.0007274392162748551,
      "loss": 1.4297358989715576,
      "step": 8448
    },
    {
      "epoch": 0.3751158230439373,
      "grad_norm": 0.9941537380218506,
      "learning_rate": 0.000711518231245687,
      "loss": 1.4256483316421509,
      "step": 8704
    },
    {
      "epoch": 0.38614864136875904,
      "grad_norm": 1.144028663635254,
      "learning_rate": 0.0006953315799141723,
      "loss": 1.4252721071243286,
      "step": 8960
    },
    {
      "epoch": 0.3971814596935807,
      "grad_norm": 1.3483773469924927,
      "learning_rate": 0.0006788995926687669,
      "loss": 1.4217591285705566,
      "step": 9216
    },
    {
      "epoch": 0.3971814596935807,
      "eval_bleu": 0.9351012282354647,
      "eval_cos_loss": 0.44966854437836196,
      "eval_dec_loss": 0.12245929105155694,
      "eval_loss": 1.4224536047815515,
      "eval_mse2_loss": 0.15468242016237682,
      "eval_mse_loss": 1.2677711858424043,
      "eval_rec_loss": 0.047009017791098624,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.5503314502203642,
      "flow/improvement_ratio": 0.892241062640127,
      "flow/mag_ratio_mean": 0.5635909434320576,
      "flow/mag_ratio_std": 0.27180960243825975,
      "step": 9216
    },
    {
      "epoch": 0.3971814596935807,
      "eval_bleu": 0.9351012282354647,
      "eval_cos_loss": 0.44966854437836196,
      "eval_dec_loss": 0.12245929105155694,
      "eval_loss": 1.4224536047815515,
      "eval_mse2_loss": 0.15468242016237682,
      "eval_mse_loss": 1.2677711858424043,
      "eval_rec_loss": 0.047009017791098624,
      "eval_runtime": 149.98,
      "eval_samples_per_second": 200.027,
      "eval_steps_per_second": 3.127,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.5503314502203642,
      "flow/improvement_ratio": 0.892241062640127,
      "flow/mag_ratio_mean": 0.5635909434320576,
      "flow/mag_ratio_std": 0.27180960243825975,
      "step": 9216
    },
    {
      "epoch": 0.4082142780184024,
      "grad_norm": 1.1067237854003906,
      "learning_rate": 0.0006622429080391422,
      "loss": 1.4252803325653076,
      "step": 9472
    },
    {
      "epoch": 0.4192470963432241,
      "grad_norm": 0.9428814649581909,
      "learning_rate": 0.0006453824467742515,
      "loss": 1.4204916954040527,
      "step": 9728
    },
    {
      "epoch": 0.43027991466804577,
      "grad_norm": 0.9498366713523865,
      "learning_rate": 0.0006283393855659275,
      "loss": 1.421370506286621,
      "step": 9984
    },
    {
      "epoch": 0.44131273299286744,
      "grad_norm": 1.0918892621994019,
      "learning_rate": 0.0006111351304510173,
      "loss": 1.4164854288101196,
      "step": 10240
    },
    {
      "epoch": 0.44131273299286744,
      "eval_bleu": 0.9359500949598805,
      "eval_cos_loss": 0.4479471450802614,
      "eval_dec_loss": 0.11968307011623754,
      "eval_loss": 1.4188302267334862,
      "eval_mse2_loss": 0.1544050017653752,
      "eval_mse_loss": 1.2644252238243119,
      "eval_rec_loss": 0.047009017791098624,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.5520528566354374,
      "flow/improvement_ratio": 0.8973728770386181,
      "flow/mag_ratio_mean": 0.5616697732573633,
      "flow/mag_ratio_std": 0.26565590021071406,
      "step": 10240
    },
    {
      "epoch": 0.44131273299286744,
      "eval_bleu": 0.9359500949598805,
      "eval_cos_loss": 0.4479471450802614,
      "eval_dec_loss": 0.11968307011623754,
      "eval_loss": 1.4188302267334862,
      "eval_mse2_loss": 0.1544050017653752,
      "eval_mse_loss": 1.2644252238243119,
      "eval_rec_loss": 0.047009017791098624,
      "eval_runtime": 154.1693,
      "eval_samples_per_second": 194.591,
      "eval_steps_per_second": 3.042,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.5520528566354374,
      "flow/improvement_ratio": 0.8973728770386181,
      "flow/mag_ratio_mean": 0.5616697732573633,
      "flow/mag_ratio_std": 0.26565590021071406,
      "step": 10240
    },
    {
      "epoch": 0.45234555131768917,
      "grad_norm": 1.4444184303283691,
      "learning_rate": 0.0005937912899254605,
      "loss": 1.4161497354507446,
      "step": 10496
    },
    {
      "epoch": 0.46337836964251083,
      "grad_norm": 1.1204206943511963,
      "learning_rate": 0.0005763296478040787,
      "loss": 1.4160338640213013,
      "step": 10752
    },
    {
      "epoch": 0.4744111879673325,
      "grad_norm": 1.2254257202148438,
      "learning_rate": 0.0005587721358601663,
      "loss": 1.4160186052322388,
      "step": 11008
    },
    {
      "epoch": 0.4854440062921542,
      "grad_norm": 1.4265046119689941,
      "learning_rate": 0.0005411408062792448,
      "loss": 1.4131176471710205,
      "step": 11264
    },
    {
      "epoch": 0.4854440062921542,
      "eval_bleu": 0.9355559828150384,
      "eval_cos_loss": 0.4478943133150845,
      "eval_dec_loss": 0.1184240544337962,
      "eval_loss": 1.416160212396813,
      "eval_mse2_loss": 0.15195046149210126,
      "eval_mse_loss": 1.2642097521438274,
      "eval_rec_loss": 0.047009017791098624,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.5521056849056724,
      "flow/improvement_ratio": 0.8935815333557535,
      "flow/mag_ratio_mean": 0.5588731608156965,
      "flow/mag_ratio_std": 0.2649187071364063,
      "step": 11264
    },
    {
      "epoch": 0.4854440062921542,
      "eval_bleu": 0.9355559828150384,
      "eval_cos_loss": 0.4478943133150845,
      "eval_dec_loss": 0.1184240544337962,
      "eval_loss": 1.416160212396813,
      "eval_mse2_loss": 0.15195046149210126,
      "eval_mse_loss": 1.2642097521438274,
      "eval_rec_loss": 0.047009017791098624,
      "eval_runtime": 153.2221,
      "eval_samples_per_second": 195.794,
      "eval_steps_per_second": 3.061,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.5521056849056724,
      "flow/improvement_ratio": 0.8935815333557535,
      "flow/mag_ratio_mean": 0.5588731608156965,
      "flow/mag_ratio_std": 0.2649187071364063,
      "step": 11264
    },
    {
      "epoch": 0.4964768246169759,
      "grad_norm": 1.052992582321167,
      "learning_rate": 0.0005234578039615789,
      "loss": 1.4113584756851196,
      "step": 11520
    },
    {
      "epoch": 0.5075096429417976,
      "grad_norm": 1.5359535217285156,
      "learning_rate": 0.0005057453387082458,
      "loss": 1.410513997077942,
      "step": 11776
    },
    {
      "epoch": 0.5185424612666193,
      "grad_norm": 1.2274341583251953,
      "learning_rate": 0.0004880256573256866,
      "loss": 1.4113372564315796,
      "step": 12032
    },
    {
      "epoch": 0.5295752795914409,
      "grad_norm": 0.7585653066635132,
      "learning_rate": 0.0004703210156837805,
      "loss": 1.4105942249298096,
      "step": 12288
    },
    {
      "epoch": 0.5295752795914409,
      "eval_bleu": 0.9365001737473854,
      "eval_cos_loss": 0.4468586768295719,
      "eval_dec_loss": 0.11578587821519959,
      "eval_loss": 1.4128359007174527,
      "eval_mse2_loss": 0.1518495731802383,
      "eval_mse_loss": 1.2609863276166449,
      "eval_rec_loss": 0.047009017791098624,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.5531413205651078,
      "flow/improvement_ratio": 0.8957801499346426,
      "flow/mag_ratio_mean": 0.558759307556315,
      "flow/mag_ratio_std": 0.2658895003770206,
      "step": 12288
    },
    {
      "epoch": 0.5295752795914409,
      "eval_bleu": 0.9365001737473854,
      "eval_cos_loss": 0.4468586768295719,
      "eval_dec_loss": 0.11578587821519959,
      "eval_loss": 1.4128359007174527,
      "eval_mse2_loss": 0.1518495731802383,
      "eval_mse_loss": 1.2609863276166449,
      "eval_rec_loss": 0.047009017791098624,
      "eval_runtime": 152.5586,
      "eval_samples_per_second": 196.646,
      "eval_steps_per_second": 3.074,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.5531413205651078,
      "flow/improvement_ratio": 0.8957801499346426,
      "flow/mag_ratio_mean": 0.558759307556315,
      "flow/mag_ratio_std": 0.2658895003770206,
      "step": 12288
    },
    {
      "epoch": 0.5406080979162626,
      "grad_norm": 0.7903220653533936,
      "learning_rate": 0.0004526536507625343,
      "loss": 1.4081207513809204,
      "step": 12544
    },
    {
      "epoch": 0.5516409162410844,
      "grad_norm": 1.299485445022583,
      "learning_rate": 0.00043504575272249973,
      "loss": 1.4094181060791016,
      "step": 12800
    },
    {
      "epoch": 0.562673734565906,
      "grad_norm": 1.1680424213409424,
      "learning_rate": 0.0004175194370339921,
      "loss": 1.4080712795257568,
      "step": 13056
    },
    {
      "epoch": 0.5737065528907277,
      "grad_norm": 1.1761500835418701,
      "learning_rate": 0.0004000967167001243,
      "loss": 1.406380534172058,
      "step": 13312
    },
    {
      "epoch": 0.5737065528907277,
      "eval_bleu": 0.9375331490996615,
      "eval_cos_loss": 0.4439614661721024,
      "eval_dec_loss": 0.11194793060858811,
      "eval_loss": 1.4055488673862873,
      "eval_mse2_loss": 0.1498908800547565,
      "eval_mse_loss": 1.2556579890790016,
      "eval_rec_loss": 0.047009017791098624,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.5560385347175192,
      "flow/improvement_ratio": 0.8952096234252458,
      "flow/mag_ratio_mean": 0.5663808779930001,
      "flow/mag_ratio_std": 0.2659331502627208,
      "step": 13312
    },
    {
      "epoch": 0.5737065528907277,
      "eval_bleu": 0.9375331490996615,
      "eval_cos_loss": 0.4439614661721024,
      "eval_dec_loss": 0.11194793060858811,
      "eval_loss": 1.4055488673862873,
      "eval_mse2_loss": 0.1498908800547565,
      "eval_mse_loss": 1.2556579890790016,
      "eval_rec_loss": 0.047009017791098624,
      "eval_runtime": 151.6081,
      "eval_samples_per_second": 197.879,
      "eval_steps_per_second": 3.094,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.5560385347175192,
      "flow/improvement_ratio": 0.8952096234252458,
      "flow/mag_ratio_mean": 0.5663808779930001,
      "flow/mag_ratio_std": 0.2659331502627208,
      "step": 13312
    },
    {
      "epoch": 0.5847393712155494,
      "grad_norm": 0.7058362364768982,
      "learning_rate": 0.00038279947460853446,
      "loss": 1.4039076566696167,
      "step": 13568
    },
    {
      "epoch": 0.595772189540371,
      "grad_norm": 1.867743730545044,
      "learning_rate": 0.00036564943604654345,
      "loss": 1.4034653902053833,
      "step": 13824
    },
    {
      "epoch": 0.6068050078651928,
      "grad_norm": 0.9271458983421326,
      "learning_rate": 0.00034866814141425254,
      "loss": 1.4046720266342163,
      "step": 14080
    },
    {
      "epoch": 0.6178378261900145,
      "grad_norm": 0.8186226487159729,
      "learning_rate": 0.0003318769191698637,
      "loss": 1.4014854431152344,
      "step": 14336
    },
    {
      "epoch": 0.6178378261900145,
      "eval_bleu": 0.9378049095613451,
      "eval_cos_loss": 0.44325206020493496,
      "eval_dec_loss": 0.11280616647454658,
      "eval_loss": 1.402791700637671,
      "eval_mse2_loss": 0.14996937808515168,
      "eval_mse_loss": 1.2528223221235946,
      "eval_rec_loss": 0.047009017791098624,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.5567479390325323,
      "flow/improvement_ratio": 0.8963780609021055,
      "flow/mag_ratio_mean": 0.5675655131909385,
      "flow/mag_ratio_std": 0.2700714123274472,
      "step": 14336
    },
    {
      "epoch": 0.6178378261900145,
      "eval_bleu": 0.9378049095613451,
      "eval_cos_loss": 0.44325206020493496,
      "eval_dec_loss": 0.11280616647454658,
      "eval_loss": 1.402791700637671,
      "eval_mse2_loss": 0.14996937808515168,
      "eval_mse_loss": 1.2528223221235946,
      "eval_rec_loss": 0.047009017791098624,
      "eval_runtime": 152.2222,
      "eval_samples_per_second": 197.08,
      "eval_steps_per_second": 3.081,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.5567479390325323,
      "flow/improvement_ratio": 0.8963780609021055,
      "flow/mag_ratio_mean": 0.5675655131909385,
      "flow/mag_ratio_std": 0.2700714123274472,
      "step": 14336
    },
    {
      "epoch": 0.6288706445148361,
      "grad_norm": 1.4290482997894287,
      "learning_rate": 0.00031529685904119485,
      "loss": 1.4003877639770508,
      "step": 14592
    },
    {
      "epoch": 0.6399034628396578,
      "grad_norm": 0.92575603723526,
      "learning_rate": 0.0002989487855370421,
      "loss": 1.4029277563095093,
      "step": 14848
    },
    {
      "epoch": 0.6509362811644795,
      "grad_norm": 0.9941800236701965,
      "learning_rate": 0.00028285323179165424,
      "loss": 1.401476263999939,
      "step": 15104
    },
    {
      "epoch": 0.6619690994893012,
      "grad_norm": 0.5996092557907104,
      "learning_rate": 0.0002670304137751759,
      "loss": 1.40423583984375,
      "step": 15360
    },
    {
      "epoch": 0.6619690994893012,
      "eval_bleu": 0.9359784532013938,
      "eval_cos_loss": 0.44205584766259837,
      "eval_dec_loss": 0.12092216288460407,
      "eval_loss": 1.4025819934507422,
      "eval_mse2_loss": 0.15172467312451873,
      "eval_mse_loss": 1.2508573216924281,
      "eval_rec_loss": 0.047009017791098624,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.5579441513206913,
      "flow/improvement_ratio": 0.893738461964166,
      "flow/mag_ratio_mean": 0.5637592339058166,
      "flow/mag_ratio_std": 0.2644218521903573,
      "step": 15360
    },
    {
      "epoch": 0.6619690994893012,
      "eval_bleu": 0.9359784532013938,
      "eval_cos_loss": 0.44205584766259837,
      "eval_dec_loss": 0.12092216288460407,
      "eval_loss": 1.4025819934507422,
      "eval_mse2_loss": 0.15172467312451873,
      "eval_mse_loss": 1.2508573216924281,
      "eval_rec_loss": 0.047009017791098624,
      "eval_runtime": 150.8251,
      "eval_samples_per_second": 198.906,
      "eval_steps_per_second": 3.11,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.5579441513206913,
      "flow/improvement_ratio": 0.893738461964166,
      "flow/mag_ratio_mean": 0.5637592339058166,
      "flow/mag_ratio_std": 0.2644218521903573,
      "step": 15360
    },
    {
      "epoch": 0.6730019178141229,
      "grad_norm": 1.2950412034988403,
      "learning_rate": 0.0002515002049024435,
      "loss": 1.4007450342178345,
      "step": 15616
    },
    {
      "epoch": 0.6840347361389446,
      "grad_norm": 0.9485428333282471,
      "learning_rate": 0.00023628211107203429,
      "loss": 1.4006445407867432,
      "step": 15872
    },
    {
      "epoch": 0.6950675544637662,
      "grad_norm": 0.7694559693336487,
      "learning_rate": 0.00022139524616691188,
      "loss": 1.3995271921157837,
      "step": 16128
    },
    {
      "epoch": 0.706100372788588,
      "grad_norm": 1.0373378992080688,
      "learning_rate": 0.000206858308047443,
      "loss": 1.396484136581421,
      "step": 16384
    },
    {
      "epoch": 0.706100372788588,
      "eval_bleu": 0.9382976118785468,
      "eval_cos_loss": 0.4408198703072473,
      "eval_dec_loss": 0.11163298516814261,
      "eval_loss": 1.3957897457740962,
      "eval_mse2_loss": 0.14787080404219596,
      "eval_mse_loss": 1.247918940556329,
      "eval_rec_loss": 0.047009017791098624,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.5591801271509769,
      "flow/improvement_ratio": 0.8951915981672974,
      "flow/mag_ratio_mean": 0.5686852338471646,
      "flow/mag_ratio_std": 0.26723379195371927,
      "step": 16384
    },
    {
      "epoch": 0.706100372788588,
      "eval_bleu": 0.9382976118785468,
      "eval_cos_loss": 0.4408198703072473,
      "eval_dec_loss": 0.11163298516814261,
      "eval_loss": 1.3957897457740962,
      "eval_mse2_loss": 0.14787080404219596,
      "eval_mse_loss": 1.247918940556329,
      "eval_rec_loss": 0.047009017791098624,
      "eval_runtime": 149.6268,
      "eval_samples_per_second": 200.499,
      "eval_steps_per_second": 3.134,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.5591801271509769,
      "flow/improvement_ratio": 0.8951915981672974,
      "flow/mag_ratio_mean": 0.5686852338471646,
      "flow/mag_ratio_std": 0.26723379195371927,
      "step": 16384
    },
    {
      "epoch": 0.7171331911134096,
      "grad_norm": 0.7638829946517944,
      "learning_rate": 0.00019268955506693798,
      "loss": 1.3981057405471802,
      "step": 16640
    },
    {
      "epoch": 0.7281660094382313,
      "grad_norm": 1.2486227750778198,
      "learning_rate": 0.00017890678313921,
      "loss": 1.396310806274414,
      "step": 16896
    },
    {
      "epoch": 0.739198827763053,
      "grad_norm": 1.3649706840515137,
      "learning_rate": 0.00016552730338695792,
      "loss": 1.395443320274353,
      "step": 17152
    },
    {
      "epoch": 0.7502316460878746,
      "grad_norm": 0.7208181023597717,
      "learning_rate": 0.00015256792039904465,
      "loss": 1.3944549560546875,
      "step": 17408
    },
    {
      "epoch": 0.7502316460878746,
      "eval_bleu": 0.9394802954217433,
      "eval_cos_loss": 0.43953821437953633,
      "eval_dec_loss": 0.11125741161501357,
      "eval_loss": 1.390636357925594,
      "eval_mse2_loss": 0.14740569860950462,
      "eval_mse_loss": 1.243230658299379,
      "eval_rec_loss": 0.047009017791098624,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.5604617825703326,
      "flow/improvement_ratio": 0.8963484140093139,
      "flow/mag_ratio_mean": 0.5701523471488628,
      "flow/mag_ratio_std": 0.2696803689701979,
      "step": 17408
    },
    {
      "epoch": 0.7502316460878746,
      "eval_bleu": 0.9394802954217433,
      "eval_cos_loss": 0.43953821437953633,
      "eval_dec_loss": 0.11125741161501357,
      "eval_loss": 1.390636357925594,
      "eval_mse2_loss": 0.14740569860950462,
      "eval_mse_loss": 1.243230658299379,
      "eval_rec_loss": 0.047009017791098624,
      "eval_runtime": 150.0541,
      "eval_samples_per_second": 199.928,
      "eval_steps_per_second": 3.126,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.5604617825703326,
      "flow/improvement_ratio": 0.8963484140093139,
      "flow/mag_ratio_mean": 0.5701523471488628,
      "flow/mag_ratio_std": 0.2696803689701979,
      "step": 17408
    },
    {
      "epoch": 0.7612644644126964,
      "grad_norm": 0.7441651821136475,
      "learning_rate": 0.00014004491112398103,
      "loss": 1.3953999280929565,
      "step": 17664
    },
    {
      "epoch": 0.7722972827375181,
      "grad_norm": 0.6111723780632019,
      "learning_rate": 0.00012797400442612433,
      "loss": 1.3906549215316772,
      "step": 17920
    },
    {
      "epoch": 0.7833301010623397,
      "grad_norm": 0.7702776789665222,
      "learning_rate": 0.00011637036133026895,
      "loss": 1.393618106842041,
      "step": 18176
    },
    {
      "epoch": 0.7943629193871614,
      "grad_norm": 1.200044870376587,
      "learning_rate": 0.00010524855597944216,
      "loss": 1.3919320106506348,
      "step": 18432
    },
    {
      "epoch": 0.7943629193871614,
      "eval_bleu": 0.9383138215986374,
      "eval_cos_loss": 0.44080494201259574,
      "eval_dec_loss": 0.11339724150253956,
      "eval_loss": 1.3953827009526396,
      "eval_mse2_loss": 0.14798823779008027,
      "eval_mse_loss": 1.247394463400851,
      "eval_rec_loss": 0.047009017791098624,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.5591950585593038,
      "flow/improvement_ratio": 0.8969450570118707,
      "flow/mag_ratio_mean": 0.5669063877449361,
      "flow/mag_ratio_std": 0.26610757334273,
      "step": 18432
    },
    {
      "epoch": 0.7943629193871614,
      "eval_bleu": 0.9383138215986374,
      "eval_cos_loss": 0.44080494201259574,
      "eval_dec_loss": 0.11339724150253956,
      "eval_loss": 1.3953827009526396,
      "eval_mse2_loss": 0.14798823779008027,
      "eval_mse_loss": 1.247394463400851,
      "eval_rec_loss": 0.047009017791098624,
      "eval_runtime": 148.9298,
      "eval_samples_per_second": 201.437,
      "eval_steps_per_second": 3.149,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.5591950585593038,
      "flow/improvement_ratio": 0.8969450570118707,
      "flow/mag_ratio_mean": 0.5669063877449361,
      "flow/mag_ratio_std": 0.26610757334273,
      "step": 18432
    },
    {
      "epoch": 0.8053957377119831,
      "grad_norm": 1.2096973657608032,
      "learning_rate": 9.462255732982089e-05,
      "loss": 1.3907443284988403,
      "step": 18688
    },
    {
      "epoch": 0.8164285560368048,
      "grad_norm": 0.7669377326965332,
      "learning_rate": 8.450571160576348e-05,
      "loss": 1.3931423425674438,
      "step": 18944
    },
    {
      "epoch": 0.8274613743616265,
      "grad_norm": 0.6018663644790649,
      "learning_rate": 7.491072553698764e-05,
      "loss": 1.3906762599945068,
      "step": 19200
    },
    {
      "epoch": 0.8384941926864482,
      "grad_norm": 0.5031753182411194,
      "learning_rate": 6.584965039895586e-05,
      "loss": 1.387755274772644,
      "step": 19456
    },
    {
      "epoch": 0.8384941926864482,
      "eval_bleu": 0.9374315797458206,
      "eval_cos_loss": 0.4402599502474006,
      "eval_dec_loss": 0.11256631666909593,
      "eval_loss": 1.3931237858241554,
      "eval_mse2_loss": 0.14820685115323137,
      "eval_mse_loss": 1.2449169349568738,
      "eval_rec_loss": 0.047009017791098624,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.5597400505151322,
      "flow/improvement_ratio": 0.8949816590433182,
      "flow/mag_ratio_mean": 0.5695700204448659,
      "flow/mag_ratio_std": 0.2675396429895084,
      "step": 19456
    },
    {
      "epoch": 0.8384941926864482,
      "eval_bleu": 0.9374315797458206,
      "eval_cos_loss": 0.4402599502474006,
      "eval_dec_loss": 0.11256631666909593,
      "eval_loss": 1.3931237858241554,
      "eval_mse2_loss": 0.14820685115323137,
      "eval_mse_loss": 1.2449169349568738,
      "eval_rec_loss": 0.047009017791098624,
      "eval_runtime": 151.1209,
      "eval_samples_per_second": 198.517,
      "eval_steps_per_second": 3.103,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.5597400505151322,
      "flow/improvement_ratio": 0.8949816590433182,
      "flow/mag_ratio_mean": 0.5695700204448659,
      "flow/mag_ratio_std": 0.2675396429895084,
      "step": 19456
    },
    {
      "epoch": 0.8495270110112698,
      "grad_norm": 0.5668984055519104,
      "learning_rate": 5.73338668765051e-05,
      "loss": 1.3939695358276367,
      "step": 19712
    },
    {
      "epoch": 0.8605598293360915,
      "grad_norm": 0.5667815208435059,
      "learning_rate": 4.9374070769740984e-05,
      "loss": 1.392187237739563,
      "step": 19968
    },
    {
      "epoch": 0.8715926476609133,
      "grad_norm": 0.8259353637695312,
      "learning_rate": 4.198025956014095e-05,
      "loss": 1.3923636674880981,
      "step": 20224
    },
    {
      "epoch": 0.8826254659857349,
      "grad_norm": 1.0611343383789062,
      "learning_rate": 3.516171985374755e-05,
      "loss": 1.3925312757492065,
      "step": 20480
    },
    {
      "epoch": 0.8826254659857349,
      "eval_bleu": 0.9374427943864152,
      "eval_cos_loss": 0.44028231228338377,
      "eval_dec_loss": 0.11508091816158374,
      "eval_loss": 1.3936327850894887,
      "eval_mse2_loss": 0.14858852614408363,
      "eval_mse_loss": 1.2450442581034418,
      "eval_rec_loss": 0.047009017791098624,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.5597176894958593,
      "flow/improvement_ratio": 0.8979350520349515,
      "flow/mag_ratio_mean": 0.5660728585999658,
      "flow/mag_ratio_std": 0.2672899484253133,
      "step": 20480
    },
    {
      "epoch": 0.8826254659857349,
      "eval_bleu": 0.9374427943864152,
      "eval_cos_loss": 0.44028231228338377,
      "eval_dec_loss": 0.11508091816158374,
      "eval_loss": 1.3936327850894887,
      "eval_mse2_loss": 0.14858852614408363,
      "eval_mse_loss": 1.2450442581034418,
      "eval_rec_loss": 0.047009017791098624,
      "eval_runtime": 153.7462,
      "eval_samples_per_second": 195.127,
      "eval_steps_per_second": 3.05,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.5597176894958593,
      "flow/improvement_ratio": 0.8979350520349515,
      "flow/mag_ratio_mean": 0.5660728585999658,
      "flow/mag_ratio_std": 0.2672899484253133,
      "step": 20480
    },
    {
      "epoch": 0.8936582843105566,
      "grad_norm": 0.3865830898284912,
      "learning_rate": 2.8927015717215733e-05,
      "loss": 1.3925455808639526,
      "step": 20736
    },
    {
      "epoch": 0.9046911026353783,
      "grad_norm": 0.5185984373092651,
      "learning_rate": 2.3283977921370547e-05,
      "loss": 1.391736388206482,
      "step": 20992
    },
    {
      "epoch": 0.9157239209601999,
      "grad_norm": 0.5432521104812622,
      "learning_rate": 1.8239694105780413e-05,
      "loss": 1.3918673992156982,
      "step": 21248
    },
    {
      "epoch": 0.9267567392850217,
      "grad_norm": 0.3220946788787842,
      "learning_rate": 1.3800499876701955e-05,
      "loss": 1.3899786472320557,
      "step": 21504
    },
    {
      "epoch": 0.9267567392850217,
      "eval_bleu": 0.9376202589977451,
      "eval_cos_loss": 0.4390181550211998,
      "eval_dec_loss": 0.11783787643691815,
      "eval_loss": 1.3900849003273288,
      "eval_mse2_loss": 0.14892718683618472,
      "eval_mse_loss": 1.2411577106793044,
      "eval_rec_loss": 0.047009017791098624,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.5609818455506997,
      "flow/improvement_ratio": 0.8963389490713188,
      "flow/mag_ratio_mean": 0.5705793747769744,
      "flow/mag_ratio_std": 0.27008461628133046,
      "step": 21504
    },
    {
      "epoch": 0.9267567392850217,
      "eval_bleu": 0.9376202589977451,
      "eval_cos_loss": 0.4390181550211998,
      "eval_dec_loss": 0.11783787643691815,
      "eval_loss": 1.3900849003273288,
      "eval_mse2_loss": 0.14892718683618472,
      "eval_mse_loss": 1.2411577106793044,
      "eval_rec_loss": 0.047009017791098624,
      "eval_runtime": 150.1598,
      "eval_samples_per_second": 199.787,
      "eval_steps_per_second": 3.123,
      "eval_var_loss": 0.01723895594080501,
      "flow/cos_sim": 0.5609818455506997,
      "flow/improvement_ratio": 0.8963389490713188,
      "flow/mag_ratio_mean": 0.5705793747769744,
      "flow/mag_ratio_std": 0.27008461628133046,
      "step": 21504
    }
  ],
  "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
}