all_results.json

{
    "achieved_tflops_per_gpu": 3.914985519721152,
    "achieved_tflops_per_gpu_theoretical": 152.45570532704255,
    "epoch": 5.0,
    "loss_nan_ranks": 0,
    "loss_rank_avg": 0.36682021617889404,
    "mfu_percent": 0.27667742188842065,
    "mfu_percent_theoretical": 10.77425479343057,
    "total_flos": 1.091054629938004e+18,
    "train_loss": 0.38741187927829235,
    "train_runtime": 17417.9225,
    "train_samples_per_second": 2.871,
    "train_steps_per_second": 0.045,
    "valid_targets_mean": 3412.7,
    "valid_targets_min": 1028
}