train_results.json

{
    "achieved_tflops_per_gpu": 0.0008393961777564379,
    "achieved_tflops_per_gpu_theoretical": 1226.4214517419869,
    "epoch": 6.0,
    "loss_nan_ranks": 0,
    "loss_rank_avg": 0.06114219129085541,
    "mfu_percent": 5.9321284647098084e-05,
    "mfu_percent_theoretical": 86.67289411604148,
    "total_flos": 90716684943360.0,
    "train_loss": 0.13800091912434492,
    "train_runtime": 6754.6088,
    "train_samples_per_second": 5.777,
    "train_steps_per_second": 0.362,
    "valid_targets_mean": 336.2,
    "valid_targets_min": 197
}