train_results.json

{
    "achieved_tflops_per_gpu": 3.851658720346756,
    "achieved_tflops_per_gpu_theoretical": 230.37751027197544,
    "epoch": 7.0,
    "loss_nan_ranks": 0,
    "loss_rank_avg": 0.30229127407073975,
    "mfu_percent": 0.3894498200552837,
    "mfu_percent_theoretical": 23.293984860664857,
    "total_flos": 1.8215701044296417e+18,
    "train_loss": 0.289593059448772,
    "train_runtime": 59116.417,
    "train_samples_per_second": 1.092,
    "train_steps_per_second": 0.068,
    "valid_targets_mean": 9465.8,
    "valid_targets_min": 3229
}