all_results.json

{
    "achieved_tflops_per_gpu": 2.979440010208909,
    "achieved_tflops_per_gpu_theoretical": 348.9541699707863,
    "epoch": 5.0,
    "loss_nan_ranks": 0,
    "loss_rank_avg": 0.1909605711698532,
    "mfu_percent": 0.3012578372304256,
    "mfu_percent_theoretical": 35.283535891889414,
    "total_flos": 7.266001898207969e+17,
    "train_loss": 0.3592725233459473,
    "train_runtime": 30483.9243,
    "train_samples_per_second": 1.64,
    "train_steps_per_second": 0.103,
    "valid_targets_mean": 3954.1,
    "valid_targets_min": 1068
}