{
  "best_global_step": 925,
  "best_metric": 3.4105401039123535,
  "best_model_checkpoint": "./qwen3moe_tinystories_sft_global_balance/checkpoint-925",
  "epoch": 0.9991915925626516,
  "eval_steps": 25,
  "global_step": 927,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.026946914578280787,
      "eval_loss": 10.00758171081543,
      "eval_runtime": 119.3793,
      "eval_samples_per_second": 104.708,
      "eval_steps_per_second": 6.551,
      "step": 25
    },
    {
      "epoch": 0.05389382915656157,
      "eval_loss": 8.949356079101562,
      "eval_runtime": 118.7415,
      "eval_samples_per_second": 105.271,
      "eval_steps_per_second": 6.586,
      "step": 50
    },
    {
      "epoch": 0.08084074373484236,
      "eval_loss": 7.982501029968262,
      "eval_runtime": 118.833,
      "eval_samples_per_second": 105.19,
      "eval_steps_per_second": 6.581,
      "step": 75
    },
    {
      "epoch": 0.10778765831312315,
      "grad_norm": 1888245.875,
      "learning_rate": 4.99936149886953e-05,
      "loss": 8.9553,
      "step": 100
    },
    {
      "epoch": 0.10778765831312315,
      "eval_loss": 6.839597702026367,
      "eval_runtime": 119.3075,
      "eval_samples_per_second": 104.771,
      "eval_steps_per_second": 6.554,
      "step": 100
    },
    {
      "epoch": 0.13473457289140395,
      "eval_loss": 5.939998149871826,
      "eval_runtime": 118.9201,
      "eval_samples_per_second": 105.113,
      "eval_steps_per_second": 6.576,
      "step": 125
    },
    {
      "epoch": 0.16168148746968472,
      "eval_loss": 5.392858982086182,
      "eval_runtime": 118.6142,
      "eval_samples_per_second": 105.384,
      "eval_steps_per_second": 6.593,
      "step": 150
    },
    {
      "epoch": 0.18862840204796552,
      "eval_loss": 5.082247734069824,
      "eval_runtime": 118.3215,
      "eval_samples_per_second": 105.644,
      "eval_steps_per_second": 6.609,
      "step": 175
    },
    {
      "epoch": 0.2155753166262463,
      "grad_norm": 1668651.875,
      "learning_rate": 4.8033420018832464e-05,
      "loss": 5.5612,
      "step": 200
    },
    {
      "epoch": 0.2155753166262463,
      "eval_loss": 4.85417366027832,
      "eval_runtime": 118.337,
      "eval_samples_per_second": 105.631,
      "eval_steps_per_second": 6.608,
      "step": 200
    },
    {
      "epoch": 0.2425222312045271,
      "eval_loss": 4.635244846343994,
      "eval_runtime": 118.9352,
      "eval_samples_per_second": 105.099,
      "eval_steps_per_second": 6.575,
      "step": 225
    },
    {
      "epoch": 0.2694691457828079,
      "eval_loss": 4.44834566116333,
      "eval_runtime": 119.5901,
      "eval_samples_per_second": 104.524,
      "eval_steps_per_second": 6.539,
      "step": 250
    },
    {
      "epoch": 0.29641606036108864,
      "eval_loss": 4.313559532165527,
      "eval_runtime": 119.4227,
      "eval_samples_per_second": 104.67,
      "eval_steps_per_second": 6.548,
      "step": 275
    },
    {
      "epoch": 0.32336297493936944,
      "grad_norm": 1884736.25,
      "learning_rate": 4.28433581375477e-05,
      "loss": 4.4829,
      "step": 300
    },
    {
      "epoch": 0.32336297493936944,
      "eval_loss": 4.1981892585754395,
      "eval_runtime": 118.721,
      "eval_samples_per_second": 105.289,
      "eval_steps_per_second": 6.587,
      "step": 300
    },
    {
      "epoch": 0.35030988951765024,
      "eval_loss": 4.093081951141357,
      "eval_runtime": 119.1798,
      "eval_samples_per_second": 104.884,
      "eval_steps_per_second": 6.562,
      "step": 325
    },
    {
      "epoch": 0.37725680409593104,
      "eval_loss": 3.997279405593872,
      "eval_runtime": 118.6147,
      "eval_samples_per_second": 105.383,
      "eval_steps_per_second": 6.593,
      "step": 350
    },
    {
      "epoch": 0.4042037186742118,
      "eval_loss": 3.9050540924072266,
      "eval_runtime": 119.2145,
      "eval_samples_per_second": 104.853,
      "eval_steps_per_second": 6.56,
      "step": 375
    },
    {
      "epoch": 0.4311506332524926,
      "grad_norm": 1581947.375,
      "learning_rate": 3.515120703156264e-05,
      "loss": 4.0105,
      "step": 400
    },
    {
      "epoch": 0.4311506332524926,
      "eval_loss": 3.8343873023986816,
      "eval_runtime": 119.1988,
      "eval_samples_per_second": 104.867,
      "eval_steps_per_second": 6.56,
      "step": 400
    },
    {
      "epoch": 0.4580975478307734,
      "eval_loss": 3.7698652744293213,
      "eval_runtime": 119.2112,
      "eval_samples_per_second": 104.856,
      "eval_steps_per_second": 6.56,
      "step": 425
    },
    {
      "epoch": 0.4850444624090542,
      "eval_loss": 3.7166643142700195,
      "eval_runtime": 118.5648,
      "eval_samples_per_second": 105.428,
      "eval_steps_per_second": 6.596,
      "step": 450
    },
    {
      "epoch": 0.511991376987335,
      "eval_loss": 3.672928810119629,
      "eval_runtime": 118.4066,
      "eval_samples_per_second": 105.568,
      "eval_steps_per_second": 6.604,
      "step": 475
    },
    {
      "epoch": 0.5389382915656158,
      "grad_norm": 1558245.625,
      "learning_rate": 2.6035600456288573e-05,
      "loss": 3.7286,
      "step": 500
    },
    {
      "epoch": 0.5389382915656158,
      "eval_loss": 3.632558822631836,
      "eval_runtime": 118.4745,
      "eval_samples_per_second": 105.508,
      "eval_steps_per_second": 6.601,
      "step": 500
    },
    {
      "epoch": 0.5658852061438965,
      "eval_loss": 3.5953712463378906,
      "eval_runtime": 118.8897,
      "eval_samples_per_second": 105.139,
      "eval_steps_per_second": 6.578,
      "step": 525
    },
    {
      "epoch": 0.5928321207221773,
      "eval_loss": 3.5634474754333496,
      "eval_runtime": 118.8661,
      "eval_samples_per_second": 105.16,
      "eval_steps_per_second": 6.579,
      "step": 550
    },
    {
      "epoch": 0.6197790353004581,
      "eval_loss": 3.5363874435424805,
      "eval_runtime": 118.8906,
      "eval_samples_per_second": 105.139,
      "eval_steps_per_second": 6.577,
      "step": 575
    },
    {
      "epoch": 0.6467259498787389,
      "grad_norm": 2365406.75,
      "learning_rate": 1.677477655573303e-05,
      "loss": 3.5716,
      "step": 600
    },
    {
      "epoch": 0.6467259498787389,
      "eval_loss": 3.5141489505767822,
      "eval_runtime": 119.2494,
      "eval_samples_per_second": 104.822,
      "eval_steps_per_second": 6.558,
      "step": 600
    },
    {
      "epoch": 0.6736728644570197,
      "eval_loss": 3.4942786693573,
      "eval_runtime": 119.25,
      "eval_samples_per_second": 104.822,
      "eval_steps_per_second": 6.558,
      "step": 625
    },
    {
      "epoch": 0.7006197790353005,
      "eval_loss": 3.4765748977661133,
      "eval_runtime": 119.4591,
      "eval_samples_per_second": 104.638,
      "eval_steps_per_second": 6.546,
      "step": 650
    },
    {
      "epoch": 0.7275666936135813,
      "eval_loss": 3.4616143703460693,
      "eval_runtime": 119.3357,
      "eval_samples_per_second": 104.746,
      "eval_steps_per_second": 6.553,
      "step": 675
    },
    {
      "epoch": 0.7545136081918621,
      "grad_norm": 1150363.875,
      "learning_rate": 8.667336608579487e-06,
      "loss": 3.4819,
      "step": 700
    },
    {
      "epoch": 0.7545136081918621,
      "eval_loss": 3.44948410987854,
      "eval_runtime": 119.0441,
      "eval_samples_per_second": 105.003,
      "eval_steps_per_second": 6.569,
      "step": 700
    },
    {
      "epoch": 0.7814605227701428,
      "eval_loss": 3.4393043518066406,
      "eval_runtime": 119.2449,
      "eval_samples_per_second": 104.826,
      "eval_steps_per_second": 6.558,
      "step": 725
    },
    {
      "epoch": 0.8084074373484236,
      "eval_loss": 3.4311933517456055,
      "eval_runtime": 119.1259,
      "eval_samples_per_second": 104.931,
      "eval_steps_per_second": 6.564,
      "step": 750
    },
    {
      "epoch": 0.8353543519267044,
      "eval_loss": 3.4240500926971436,
      "eval_runtime": 119.3094,
      "eval_samples_per_second": 104.77,
      "eval_steps_per_second": 6.554,
      "step": 775
    },
    {
      "epoch": 0.8623012665049852,
      "grad_norm": 915389.0,
      "learning_rate": 2.8501483487659216e-06,
      "loss": 3.441,
      "step": 800
    },
    {
      "epoch": 0.8623012665049852,
      "eval_loss": 3.4186859130859375,
      "eval_runtime": 119.0028,
      "eval_samples_per_second": 105.04,
      "eval_steps_per_second": 6.571,
      "step": 800
    },
    {
      "epoch": 0.889248181083266,
      "eval_loss": 3.4150524139404297,
      "eval_runtime": 119.1835,
      "eval_samples_per_second": 104.88,
      "eval_steps_per_second": 6.561,
      "step": 825
    },
    {
      "epoch": 0.9161950956615468,
      "eval_loss": 3.4126269817352295,
      "eval_runtime": 118.5462,
      "eval_samples_per_second": 105.444,
      "eval_steps_per_second": 6.597,
      "step": 850
    },
    {
      "epoch": 0.9431420102398276,
      "eval_loss": 3.411245107650757,
      "eval_runtime": 118.0843,
      "eval_samples_per_second": 105.857,
      "eval_steps_per_second": 6.622,
      "step": 875
    },
    {
      "epoch": 0.9700889248181084,
      "grad_norm": 827701.4375,
      "learning_rate": 1.38928411621439e-07,
      "loss": 3.4196,
      "step": 900
    },
    {
      "epoch": 0.9700889248181084,
      "eval_loss": 3.410634994506836,
      "eval_runtime": 118.1249,
      "eval_samples_per_second": 105.82,
      "eval_steps_per_second": 6.62,
      "step": 900
    },
    {
      "epoch": 0.9970358393963891,
      "eval_loss": 3.4105401039123535,
      "eval_runtime": 118.8199,
      "eval_samples_per_second": 105.201,
      "eval_steps_per_second": 6.581,
      "step": 925
    }
  ],
  "logging_steps": 100,
  "max_steps": 927,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 25,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 2.071708034936832e+16,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}