trainer_state.json

{
 "best_metric": 0.9551282051282052,
 "best_model_checkpoint": "text/train/checkpoint/pubmedbert-abstract/ddi_42/checkpoint-2373",
 "epoch": 10.0,
 "eval_steps": 500,
 "global_step": 7910,
 "is_hyper_param_search": false,
 "is_local_process_zero": true,
 "is_world_process_zero": true,
 "log_history": [
 {
 "epoch": 1.0,
 "eval_accuracy": 0.938301282051282,
 "eval_loss": 0.19862917065620422,
 "eval_runtime": 5.4522,
 "eval_samples_per_second": 457.799,
 "eval_steps_per_second": 1.834,
 "step": 791
 },
 {
 "epoch": 1.26,
 "grad_norm": 1.910953402519226,
 "learning_rate": 4.367888748419722e-05,
 "loss": 0.1723,
 "step": 1000
 },
 {
 "epoch": 2.0,
 "eval_accuracy": 0.9455128205128205,
 "eval_loss": 0.2700441777706146,
 "eval_runtime": 5.0135,
 "eval_samples_per_second": 497.859,
 "eval_steps_per_second": 1.995,
 "step": 1582
 },
 {
 "epoch": 2.53,
 "grad_norm": 19.194658279418945,
 "learning_rate": 3.735777496839444e-05,
 "loss": 0.0772,
 "step": 2000
 },
 {
 "epoch": 3.0,
 "eval_accuracy": 0.9551282051282052,
 "eval_loss": 0.20845328271389008,
 "eval_runtime": 4.8197,
 "eval_samples_per_second": 517.874,
 "eval_steps_per_second": 2.075,
 "step": 2373
 },
 {
 "epoch": 3.79,
 "grad_norm": 0.3629066050052643,
 "learning_rate": 3.1036662452591655e-05,
 "loss": 0.0516,
 "step": 3000
 },
 {
 "epoch": 4.0,
 "eval_accuracy": 0.9427083333333334,
 "eval_loss": 0.2969796061515808,
 "eval_runtime": 4.7872,
 "eval_samples_per_second": 521.395,
 "eval_steps_per_second": 2.089,
 "step": 3164
 },
 {
 "epoch": 5.0,
 "eval_accuracy": 0.953926282051282,
 "eval_loss": 0.2620067596435547,
 "eval_runtime": 4.7806,
 "eval_samples_per_second": 522.107,
 "eval_steps_per_second": 2.092,
 "step": 3955
 },
 {
 "epoch": 5.06,
 "grad_norm": 0.0031597923953086138,
 "learning_rate": 2.4715549936788876e-05,
 "loss": 0.0341,
 "step": 4000
 },
 {
 "epoch": 6.0,
 "eval_accuracy": 0.9423076923076923,
 "eval_loss": 0.39730244874954224,
 "eval_runtime": 4.7836,
 "eval_samples_per_second": 521.783,
 "eval_steps_per_second": 2.09,
 "step": 4746
 },
 {
 "epoch": 6.32,
 "grad_norm": 0.006749527528882027,
 "learning_rate": 1.8394437420986094e-05,
 "loss": 0.0203,
 "step": 5000
 },
 {
 "epoch": 7.0,
 "eval_accuracy": 0.9423076923076923,
 "eval_loss": 0.3636617958545685,
 "eval_runtime": 4.7835,
 "eval_samples_per_second": 521.797,
 "eval_steps_per_second": 2.091,
 "step": 5537
 },
 {
 "epoch": 7.59,
 "grad_norm": 0.012521493248641491,
 "learning_rate": 1.2073324905183313e-05,
 "loss": 0.0146,
 "step": 6000
 },
 {
 "epoch": 8.0,
 "eval_accuracy": 0.9451121794871795,
 "eval_loss": 0.4154008626937866,
 "eval_runtime": 4.7634,
 "eval_samples_per_second": 523.998,
 "eval_steps_per_second": 2.099,
 "step": 6328
 },
 {
 "epoch": 8.85,
 "grad_norm": 0.005341885611414909,
 "learning_rate": 5.752212389380531e-06,
 "loss": 0.007,
 "step": 7000
 },
 {
 "epoch": 9.0,
 "eval_accuracy": 0.9463141025641025,
 "eval_loss": 0.4218637943267822,
 "eval_runtime": 4.8145,
 "eval_samples_per_second": 518.431,
 "eval_steps_per_second": 2.077,
 "step": 7119
 },
 {
 "epoch": 10.0,
 "eval_accuracy": 0.9447115384615384,
 "eval_loss": 0.4098478853702545,
 "eval_runtime": 4.8332,
 "eval_samples_per_second": 516.428,
 "eval_steps_per_second": 2.069,
 "step": 7910
 },
 {
 "epoch": 10.0,
 "step": 7910,
 "total_flos": 1.663959132942336e+16,
 "train_loss": 0.048086456131241864,
 "train_runtime": 1425.5671,
 "train_samples_per_second": 177.445,
 "train_steps_per_second": 5.549
 }
 ],
 "logging_steps": 1000,
 "max_steps": 7910,
 "num_input_tokens_seen": 0,
 "num_train_epochs": 10,
 "save_steps": 500,
 "total_flos": 1.663959132942336e+16,
 "train_batch_size": 32,
 "trial_name": null,
 "trial_params": null
}