all_results.json

{
    "achieved_tflops_per_gpu": 4.832871506663312,
    "achieved_tflops_per_gpu_theoretical": 159.2315144526889,
    "epoch": 5.0,
    "loss_nan_ranks": 0,
    "loss_rank_avg": 0.11343716084957123,
    "mfu_percent": 0.34154568951684183,
    "mfu_percent_theoretical": 11.253110562027485,
    "total_flos": 1.3290499486829773e+17,
    "train_loss": 0.5170090794563293,
    "train_runtime": 1718.7633,
    "train_samples_per_second": 2.909,
    "train_steps_per_second": 0.047,
    "valid_targets_mean": 3113.4,
    "valid_targets_min": 587
}