train_results.json

{
    "achieved_tflops_per_gpu": 0.001193822706401171,
    "achieved_tflops_per_gpu_theoretical": 1124.784314857292,
    "epoch": 6.0,
    "loss_nan_ranks": 0,
    "loss_rank_avg": 0.08076979219913483,
    "mfu_percent": 8.436909585874e-05,
    "mfu_percent_theoretical": 79.49005758708778,
    "total_flos": 140364736167936.0,
    "train_loss": 0.16821050625614803,
    "train_runtime": 7348.4915,
    "train_samples_per_second": 5.31,
    "train_steps_per_second": 0.332,
    "valid_targets_mean": 624.3,
    "valid_targets_min": 380
}