all_results.json

{
    "achieved_tflops_per_gpu": 0.007035287867235867,
    "achieved_tflops_per_gpu_theoretical": 1734.607136165462,
    "epoch": 7.0,
    "loss_nan_ranks": 0,
    "loss_rank_avg": 0.12079036235809326,
    "mfu_percent": 0.0004971934888505912,
    "mfu_percent_theoretical": 122.58707676081002,
    "total_flos": 961954834022400.0,
    "train_loss": 0.10841045795168196,
    "train_runtime": 8545.802,
    "train_samples_per_second": 8.181,
    "train_steps_per_second": 0.512,
    "valid_targets_mean": 3138.9,
    "valid_targets_min": 304
}