all_results.json

{
    "achieved_tflops_per_gpu": 4.028113332034932,
    "achieved_tflops_per_gpu_theoretical": 355.7299790964326,
    "epoch": 7.0,
    "loss_nan_ranks": 0,
    "loss_rank_avg": 0.04305639863014221,
    "mfu_percent": 1.2910619653958118,
    "mfu_percent_theoretical": 114.01601894116429,
    "total_flos": 3.8836436423755694e+18,
    "train_loss": 0.06260271916794653,
    "train_runtime": 120516.8314,
    "train_samples_per_second": 1.686,
    "train_steps_per_second": 0.105,
    "valid_targets_mean": 4470.5,
    "valid_targets_min": 1273
}