all_results.json

{
    "achieved_tflops_per_gpu": 4.805999622358391,
    "achieved_tflops_per_gpu_theoretical": 152.45570532704255,
    "epoch": 5.0,
    "loss_nan_ranks": 0,
    "loss_rank_avg": 0.40260475873947144,
    "mfu_percent": 0.33964661642108773,
    "mfu_percent_theoretical": 10.77425479343057,
    "total_flos": 1.3461651925678162e+18,
    "train_loss": 0.41955339954157544,
    "train_runtime": 17506.3111,
    "train_samples_per_second": 2.856,
    "train_steps_per_second": 0.045,
    "valid_targets_mean": 3870.8,
    "valid_targets_min": 1612
}