train_results.json

{
    "achieved_tflops_per_gpu": 6.530954251824663,
    "achieved_tflops_per_gpu_theoretical": 481.0824479208898,
    "epoch": 7.0,
    "loss_nan_ranks": 0,
    "loss_rank_avg": 0.11451305449008942,
    "mfu_percent": 2.09325456789252,
    "mfu_percent_theoretical": 154.1930922823365,
    "total_flos": 3.536242678171697e+18,
    "train_loss": 0.04252522938566612,
    "train_runtime": 67682.3505,
    "train_samples_per_second": 2.274,
    "train_steps_per_second": 0.142,
    "valid_targets_mean": 7922.2,
    "valid_targets_min": 3513
}