trainer_state.json

{
  "best_metric": 0.7511665338396387,
  "best_model_checkpoint": "/lus/work/CT10/c1615138/SHARED/scratch/gaperon-data/models/quality_cls/xlm-v/llama-only-filtered-single--512--25--8e-06--64--linear--0.0/checkpoint-7812",
  "epoch": 8.0,
  "eval_steps": 500,
  "global_step": 8928,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.44802867383512546,
      "grad_norm": 8.850239753723145,
      "learning_rate": 7.551971326164874e-06,
      "loss": 0.7332,
      "step": 500
    },
    {
      "epoch": 0.8960573476702509,
      "grad_norm": 3.0268373489379883,
      "learning_rate": 7.103942652329749e-06,
      "loss": 0.562,
      "step": 1000
    },
    {
      "epoch": 1.0,
      "eval_loss": 0.582636296749115,
      "eval_quality_f1": 0.7082190985655687,
      "eval_runtime": 2.4363,
      "eval_samples_per_second": 410.466,
      "eval_steps_per_second": 0.821,
      "step": 1116
    },
    {
      "epoch": 1.3440860215053765,
      "grad_norm": 4.1538987159729,
      "learning_rate": 6.655913978494624e-06,
      "loss": 0.5385,
      "step": 1500
    },
    {
      "epoch": 1.7921146953405018,
      "grad_norm": 2.82087779045105,
      "learning_rate": 6.2078853046594985e-06,
      "loss": 0.5244,
      "step": 2000
    },
    {
      "epoch": 2.0,
      "eval_loss": 0.5524051785469055,
      "eval_quality_f1": 0.7079129220265732,
      "eval_runtime": 1.1303,
      "eval_samples_per_second": 884.757,
      "eval_steps_per_second": 1.77,
      "step": 2232
    },
    {
      "epoch": 2.240143369175627,
      "grad_norm": 5.126065254211426,
      "learning_rate": 5.759856630824372e-06,
      "loss": 0.5122,
      "step": 2500
    },
    {
      "epoch": 2.688172043010753,
      "grad_norm": 4.198544025421143,
      "learning_rate": 5.311827956989247e-06,
      "loss": 0.5031,
      "step": 3000
    },
    {
      "epoch": 3.0,
      "eval_loss": 0.5608200430870056,
      "eval_quality_f1": 0.7284878863826232,
      "eval_runtime": 1.1236,
      "eval_samples_per_second": 890.01,
      "eval_steps_per_second": 1.78,
      "step": 3348
    },
    {
      "epoch": 3.1362007168458783,
      "grad_norm": 2.6578023433685303,
      "learning_rate": 4.863799283154122e-06,
      "loss": 0.4919,
      "step": 3500
    },
    {
      "epoch": 3.5842293906810037,
      "grad_norm": 4.944795608520508,
      "learning_rate": 4.4157706093189966e-06,
      "loss": 0.4817,
      "step": 4000
    },
    {
      "epoch": 4.0,
      "eval_loss": 0.5346963405609131,
      "eval_quality_f1": 0.7463488580781336,
      "eval_runtime": 3.8034,
      "eval_samples_per_second": 262.92,
      "eval_steps_per_second": 0.526,
      "step": 4464
    },
    {
      "epoch": 4.032258064516129,
      "grad_norm": 5.10132360458374,
      "learning_rate": 3.9677419354838704e-06,
      "loss": 0.4817,
      "step": 4500
    },
    {
      "epoch": 4.480286738351254,
      "grad_norm": 5.194272994995117,
      "learning_rate": 3.5197132616487456e-06,
      "loss": 0.4682,
      "step": 5000
    },
    {
      "epoch": 4.92831541218638,
      "grad_norm": 3.7467427253723145,
      "learning_rate": 3.07168458781362e-06,
      "loss": 0.4634,
      "step": 5500
    },
    {
      "epoch": 5.0,
      "eval_loss": 0.5277291536331177,
      "eval_quality_f1": 0.7510086276061395,
      "eval_runtime": 1.5031,
      "eval_samples_per_second": 665.304,
      "eval_steps_per_second": 1.331,
      "step": 5580
    },
    {
      "epoch": 5.376344086021505,
      "grad_norm": 5.132294654846191,
      "learning_rate": 2.6236559139784946e-06,
      "loss": 0.4563,
      "step": 6000
    },
    {
      "epoch": 5.824372759856631,
      "grad_norm": 5.5977020263671875,
      "learning_rate": 2.175627240143369e-06,
      "loss": 0.449,
      "step": 6500
    },
    {
      "epoch": 6.0,
      "eval_loss": 0.5459088087081909,
      "eval_quality_f1": 0.7371203028874914,
      "eval_runtime": 1.2411,
      "eval_samples_per_second": 805.766,
      "eval_steps_per_second": 1.612,
      "step": 6696
    },
    {
      "epoch": 6.272401433691757,
      "grad_norm": 5.542992115020752,
      "learning_rate": 1.7275985663082438e-06,
      "loss": 0.4466,
      "step": 7000
    },
    {
      "epoch": 6.720430107526882,
      "grad_norm": 4.132849216461182,
      "learning_rate": 1.2795698924731181e-06,
      "loss": 0.4422,
      "step": 7500
    },
    {
      "epoch": 7.0,
      "eval_loss": 0.5344868302345276,
      "eval_quality_f1": 0.7511665338396387,
      "eval_runtime": 1.1245,
      "eval_samples_per_second": 889.293,
      "eval_steps_per_second": 1.779,
      "step": 7812
    },
    {
      "epoch": 7.168458781362007,
      "grad_norm": 3.255526304244995,
      "learning_rate": 8.315412186379928e-07,
      "loss": 0.4382,
      "step": 8000
    },
    {
      "epoch": 7.616487455197133,
      "grad_norm": 6.981889247894287,
      "learning_rate": 3.8351254480286737e-07,
      "loss": 0.433,
      "step": 8500
    },
    {
      "epoch": 8.0,
      "eval_loss": 0.5395817160606384,
      "eval_quality_f1": 0.7457693595927116,
      "eval_runtime": 9.4684,
      "eval_samples_per_second": 105.614,
      "eval_steps_per_second": 0.211,
      "step": 8928
    },
    {
      "epoch": 8.0,
      "step": 8928,
      "total_flos": 6.010120884036649e+17,
      "train_loss": 0.4924640074425701,
      "train_runtime": 8914.4901,
      "train_samples_per_second": 256.238,
      "train_steps_per_second": 1.002
    }
  ],
  "logging_steps": 500,
  "max_steps": 8928,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 8,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 6.010120884036649e+17,
  "train_batch_size": 256,
  "trial_name": null,
  "trial_params": null
}