all_results.json

{
    "achieved_tflops_per_gpu": 4.400173591434152,
    "achieved_tflops_per_gpu_theoretical": 348.9541699707863,
    "epoch": 5.0,
    "loss_nan_ranks": 0,
    "loss_rank_avg": 0.09667246043682098,
    "mfu_percent": 0.31096633155011677,
    "mfu_percent_theoretical": 24.661072082741082,
    "total_flos": 3429167924248576.0,
    "train_loss": 0.6008079290390015,
    "train_runtime": 97.4157,
    "train_samples_per_second": 4.465,
    "train_steps_per_second": 0.103,
    "valid_targets_mean": 710.0,
    "valid_targets_min": 136
}