{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.37836589533970716,
  "eval_steps": 1024,
  "global_step": 8192,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.011823934229365849,
      "grad_norm": 0.12324434518814087,
      "learning_rate": 0.000498046875,
      "loss": 2.881023645401001,
      "step": 256
    },
    {
      "epoch": 0.023647868458731697,
      "grad_norm": 0.15631070733070374,
      "learning_rate": 0.000998046875,
      "loss": 2.335212469100952,
      "step": 512
    },
    {
      "epoch": 0.03547180268809755,
      "grad_norm": 0.32382041215896606,
      "learning_rate": 0.000999640996023194,
      "loss": 1.905767560005188,
      "step": 768
    },
    {
      "epoch": 0.047295736917463395,
      "grad_norm": 0.3657301664352417,
      "learning_rate": 0.0009985588674043958,
      "loss": 1.565263271331787,
      "step": 1024
    },
    {
      "epoch": 0.047295736917463395,
      "eval_cos_loss": 0.3709636884465065,
      "eval_loss": 1.4077782859540966,
      "eval_mse_loss": 0.915621253589517,
      "eval_sim_loss": 0.492157032636747,
      "flow/cos_sim": 0.6290363127782465,
      "flow/improvement_ratio": 0.9817231620011264,
      "flow/mag_ratio_mean": 0.6408119081906533,
      "flow/mag_ratio_std": 0.1935653368087664,
      "step": 1024
    },
    {
      "epoch": 0.047295736917463395,
      "eval_cos_loss": 0.3709636884465065,
      "eval_loss": 1.4077782859540966,
      "eval_mse_loss": 0.915621253589517,
      "eval_runtime": 44.3003,
      "eval_samples_per_second": 631.892,
      "eval_sim_loss": 0.492157032636747,
      "eval_steps_per_second": 9.887,
      "flow/cos_sim": 0.6290363127782465,
      "flow/improvement_ratio": 0.9817231620011264,
      "flow/mag_ratio_mean": 0.6408119081906533,
      "flow/mag_ratio_std": 0.1935653368087664,
      "step": 1024
    },
    {
      "epoch": 0.05911967114682925,
      "grad_norm": 0.8176831007003784,
      "learning_rate": 0.0009967551747861387,
      "loss": 1.272852897644043,
      "step": 1280
    },
    {
      "epoch": 0.0709436053761951,
      "grad_norm": 0.5458641052246094,
      "learning_rate": 0.000994232528651847,
      "loss": 1.0125640630722046,
      "step": 1536
    },
    {
      "epoch": 0.08276753960556095,
      "grad_norm": 1.5209648609161377,
      "learning_rate": 0.0009909945800260092,
      "loss": 0.7900597453117371,
      "step": 1792
    },
    {
      "epoch": 0.09459147383492679,
      "grad_norm": 0.5339614152908325,
      "learning_rate": 0.0009870460151900522,
      "loss": 0.5982019305229187,
      "step": 2048
    },
    {
      "epoch": 0.09459147383492679,
      "eval_cos_loss": 0.166214491882825,
      "eval_loss": 0.5164470963826463,
      "eval_mse_loss": 0.39311050484169563,
      "eval_sim_loss": 0.12333659069042772,
      "flow/cos_sim": 0.8337855092738862,
      "flow/improvement_ratio": 0.9917055516754656,
      "flow/mag_ratio_mean": 0.8421911159062494,
      "flow/mag_ratio_std": 0.158547153926059,
      "step": 2048
    },
    {
      "epoch": 0.09459147383492679,
      "eval_cos_loss": 0.166214491882825,
      "eval_loss": 0.5164470963826463,
      "eval_mse_loss": 0.39311050484169563,
      "eval_runtime": 44.6023,
      "eval_samples_per_second": 627.614,
      "eval_sim_loss": 0.12333659069042772,
      "eval_steps_per_second": 9.82,
      "flow/cos_sim": 0.8337855092738862,
      "flow/improvement_ratio": 0.9917055516754656,
      "flow/mag_ratio_mean": 0.8421911159062494,
      "flow/mag_ratio_std": 0.158547153926059,
      "step": 2048
    },
    {
      "epoch": 0.10641540806429264,
      "grad_norm": 0.6704728007316589,
      "learning_rate": 0.0009823925488998885,
      "loss": 0.449043333530426,
      "step": 2304
    },
    {
      "epoch": 0.1182393422936585,
      "grad_norm": 0.7579811215400696,
      "learning_rate": 0.0009770409161149525,
      "loss": 0.3412686586380005,
      "step": 2560
    },
    {
      "epoch": 0.13006327652302435,
      "grad_norm": 0.5465359091758728,
      "learning_rate": 0.0009709988622506973,
      "loss": 0.2806977331638336,
      "step": 2816
    },
    {
      "epoch": 0.1418872107523902,
      "grad_norm": 0.4062241017818451,
      "learning_rate": 0.000964275131968659,
      "loss": 0.222763791680336,
      "step": 3072
    },
    {
      "epoch": 0.1418872107523902,
      "eval_cos_loss": 0.0714544884469275,
      "eval_loss": 0.20498143738671526,
      "eval_mse_loss": 0.1675494800419568,
      "eval_sim_loss": 0.037431957672209776,
      "flow/cos_sim": 0.9285455110682744,
      "flow/improvement_ratio": 0.9945728493063417,
      "flow/mag_ratio_mean": 0.9371804536477616,
      "flow/mag_ratio_std": 0.11087696671009609,
      "step": 3072
    },
    {
      "epoch": 0.1418872107523902,
      "eval_cos_loss": 0.0714544884469275,
      "eval_loss": 0.20498143738671526,
      "eval_mse_loss": 0.1675494800419568,
      "eval_runtime": 44.6285,
      "eval_samples_per_second": 627.246,
      "eval_sim_loss": 0.037431957672209776,
      "eval_steps_per_second": 9.814,
      "flow/cos_sim": 0.9285455110682744,
      "flow/improvement_ratio": 0.9945728493063417,
      "flow/mag_ratio_mean": 0.9371804536477616,
      "flow/mag_ratio_std": 0.11087696671009609,
      "step": 3072
    },
    {
      "epoch": 0.15371114498175603,
      "grad_norm": 0.6429200172424316,
      "learning_rate": 0.0009568794565203123,
      "loss": 0.19131025671958923,
      "step": 3328
    },
    {
      "epoch": 0.1655350792111219,
      "grad_norm": 0.4758129119873047,
      "learning_rate": 0.0009488225396630347,
      "loss": 0.16897854208946228,
      "step": 3584
    },
    {
      "epoch": 0.17735901344048774,
      "grad_norm": 0.2732395827770233,
      "learning_rate": 0.0009401160421685646,
      "loss": 0.15373976528644562,
      "step": 3840
    },
    {
      "epoch": 0.18918294766985358,
      "grad_norm": 0.3236653208732605,
      "learning_rate": 0.0009307725649463714,
      "loss": 0.1414027065038681,
      "step": 4096
    },
    {
      "epoch": 0.18918294766985358,
      "eval_cos_loss": 0.05107848004323162,
      "eval_loss": 0.138613528890969,
      "eval_mse_loss": 0.11963267840411021,
      "eval_sim_loss": 0.01898085068885798,
      "flow/cos_sim": 0.9489215186469631,
      "flow/improvement_ratio": 0.9940151711032815,
      "flow/mag_ratio_mean": 0.9492184031227408,
      "flow/mag_ratio_std": 0.08317253486036438,
      "step": 4096
    },
    {
      "epoch": 0.18918294766985358,
      "eval_cos_loss": 0.05107848004323162,
      "eval_loss": 0.138613528890969,
      "eval_mse_loss": 0.11963267840411021,
      "eval_runtime": 44.2576,
      "eval_samples_per_second": 632.501,
      "eval_sim_loss": 0.01898085068885798,
      "eval_steps_per_second": 9.897,
      "flow/cos_sim": 0.9489215186469631,
      "flow/improvement_ratio": 0.9940151711032815,
      "flow/mag_ratio_mean": 0.9492184031227408,
      "flow/mag_ratio_std": 0.08317253486036438,
      "step": 4096
    },
    {
      "epoch": 0.20100688189921945,
      "grad_norm": 0.2813233435153961,
      "learning_rate": 0.0009208056308063659,
      "loss": 0.1309625208377838,
      "step": 4352
    },
    {
      "epoch": 0.2128308161285853,
      "grad_norm": 1.0844590663909912,
      "learning_rate": 0.0009102296648873445,
      "loss": 0.124587781727314,
      "step": 4608
    },
    {
      "epoch": 0.22465475035795113,
      "grad_norm": 0.21950708329677582,
      "learning_rate": 0.0008990599737794927,
      "loss": 0.11813892424106598,
      "step": 4864
    },
    {
      "epoch": 0.236478684587317,
      "grad_norm": 0.2600741684436798,
      "learning_rate": 0.0008873127233711644,
      "loss": 0.11322201043367386,
      "step": 5120
    },
    {
      "epoch": 0.236478684587317,
      "eval_cos_loss": 0.04255827353016971,
      "eval_loss": 0.11155674004391448,
      "eval_mse_loss": 0.09941426802375546,
      "eval_sim_loss": 0.01214247206162202,
      "flow/cos_sim": 0.9574417269121022,
      "flow/improvement_ratio": 0.9936971177249194,
      "flow/mag_ratio_mean": 0.962604379000729,
      "flow/mag_ratio_std": 0.0713872539483385,
      "step": 5120
    },
    {
      "epoch": 0.236478684587317,
      "eval_cos_loss": 0.04255827353016971,
      "eval_loss": 0.11155674004391448,
      "eval_mse_loss": 0.09941426802375546,
      "eval_runtime": 44.4031,
      "eval_samples_per_second": 630.429,
      "eval_sim_loss": 0.01214247206162202,
      "eval_steps_per_second": 9.864,
      "flow/cos_sim": 0.9574417269121022,
      "flow/improvement_ratio": 0.9936971177249194,
      "flow/mag_ratio_mean": 0.962604379000729,
      "flow/mag_ratio_std": 0.0713872539483385,
      "step": 5120
    },
    {
      "epoch": 0.24830261881668284,
      "grad_norm": 0.2182897925376892,
      "learning_rate": 0.0008750049154520011,
      "loss": 0.10881217569112778,
      "step": 5376
    },
    {
      "epoch": 0.2601265530460487,
      "grad_norm": 0.2122986763715744,
      "learning_rate": 0.0008621543631062487,
      "loss": 0.11977330595254898,
      "step": 5632
    },
    {
      "epoch": 0.27195048727541454,
      "grad_norm": 0.18399468064308167,
      "learning_rate": 0.0008487796649318904,
      "loss": 0.10303137451410294,
      "step": 5888
    },
    {
      "epoch": 0.2837744215047804,
      "grad_norm": 0.22411631047725677,
      "learning_rate": 0.0008349001781229053,
      "loss": 0.10049661993980408,
      "step": 6144
    },
    {
      "epoch": 0.2837744215047804,
      "eval_cos_loss": 0.03894300645719108,
      "eval_loss": 0.09966377367916172,
      "eval_mse_loss": 0.09095577063829932,
      "eval_sim_loss": 0.008708002842052682,
      "flow/cos_sim": 0.961056991390986,
      "flow/improvement_ratio": 0.9948118436554251,
      "flow/mag_ratio_mean": 0.9735807504283783,
      "flow/mag_ratio_std": 0.06451816036106517,
      "step": 6144
    },
    {
      "epoch": 0.2837744215047804,
      "eval_cos_loss": 0.03894300645719108,
      "eval_loss": 0.09966377367916172,
      "eval_mse_loss": 0.09095577063829932,
      "eval_runtime": 46.781,
      "eval_samples_per_second": 598.384,
      "eval_sim_loss": 0.008708002842052682,
      "eval_steps_per_second": 9.363,
      "flow/cos_sim": 0.961056991390986,
      "flow/improvement_ratio": 0.9948118436554251,
      "flow/mag_ratio_mean": 0.9735807504283783,
      "flow/mag_ratio_std": 0.06451816036106517,
      "step": 6144
    },
    {
      "epoch": 0.2955983557341462,
      "grad_norm": 0.16586197912693024,
      "learning_rate": 0.0008205359904536107,
      "loss": 0.13418957591056824,
      "step": 6400
    },
    {
      "epoch": 0.30742228996351206,
      "grad_norm": 0.2780066728591919,
      "learning_rate": 0.0008057078912056363,
      "loss": 0.09691102057695389,
      "step": 6656
    },
    {
      "epoch": 0.3192462241928779,
      "grad_norm": 0.1937379539012909,
      "learning_rate": 0.0007904373410796086,
      "loss": 0.09377767890691757,
      "step": 6912
    },
    {
      "epoch": 0.3310701584222438,
      "grad_norm": 0.5644144415855408,
      "learning_rate": 0.0007747464411350876,
      "loss": 0.09154396504163742,
      "step": 7168
    },
    {
      "epoch": 0.3310701584222438,
      "eval_cos_loss": 0.03662780873035187,
      "eval_loss": 0.0921711324794924,
      "eval_mse_loss": 0.08544895486676529,
      "eval_sim_loss": 0.006722177483553926,
      "flow/cos_sim": 0.9633721935150286,
      "flow/improvement_ratio": 0.9947773780180439,
      "flow/mag_ratio_mean": 0.9676769978651717,
      "flow/mag_ratio_std": 0.05803668045664215,
      "step": 7168
    },
    {
      "epoch": 0.3310701584222438,
      "eval_cos_loss": 0.03662780873035187,
      "eval_loss": 0.0921711324794924,
      "eval_mse_loss": 0.08544895486676529,
      "eval_runtime": 45.5346,
      "eval_samples_per_second": 614.763,
      "eval_sim_loss": 0.006722177483553926,
      "eval_steps_per_second": 9.619,
      "flow/cos_sim": 0.9633721935150286,
      "flow/improvement_ratio": 0.9947773780180439,
      "flow/mag_ratio_mean": 0.9676769978651717,
      "flow/mag_ratio_std": 0.05803668045664215,
      "step": 7168
    },
    {
      "epoch": 0.34289409265160964,
      "grad_norm": 0.45590853691101074,
      "learning_rate": 0.000758657900803716,
      "loss": 0.09103628993034363,
      "step": 7424
    },
    {
      "epoch": 0.3547180268809755,
      "grad_norm": 0.3166259825229645,
      "learning_rate": 0.000742195005021869,
      "loss": 0.09036355465650558,
      "step": 7680
    },
    {
      "epoch": 0.3665419611103413,
      "grad_norm": 0.14437530934810638,
      "learning_rate": 0.0007253815805303786,
      "loss": 0.08894525468349457,
      "step": 7936
    },
    {
      "epoch": 0.37836589533970716,
      "grad_norm": 0.21061654388904572,
      "learning_rate": 0.0007082419613901028,
      "loss": 0.08701565861701965,
      "step": 8192
    },
    {
      "epoch": 0.37836589533970716,
      "eval_cos_loss": 0.035134583094995044,
      "eval_loss": 0.08715778672599901,
      "eval_mse_loss": 0.08172946989740411,
      "eval_sim_loss": 0.005428316740353154,
      "flow/cos_sim": 0.964865413022368,
      "flow/improvement_ratio": 0.9948916134496802,
      "flow/mag_ratio_mean": 0.9663348095874263,
      "flow/mag_ratio_std": 0.05209753118013138,
      "step": 8192
    },
    {
      "epoch": 0.37836589533970716,
      "eval_cos_loss": 0.035134583094995044,
      "eval_loss": 0.08715778672599901,
      "eval_mse_loss": 0.08172946989740411,
      "eval_runtime": 45.6593,
      "eval_samples_per_second": 613.085,
      "eval_sim_loss": 0.005428316740353154,
      "eval_steps_per_second": 9.593,
      "flow/cos_sim": 0.964865413022368,
      "flow/improvement_ratio": 0.9948916134496802,
      "flow/mag_ratio_mean": 0.9663348095874263,
      "flow/mag_ratio_std": 0.05209753118013138,
      "step": 8192
    }
  ],
  "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
}