all_results.json

{
    "achieved_tflops_per_gpu": 7.5946594302714745,
    "achieved_tflops_per_gpu_theoretical": 135.51618251292672,
    "epoch": 5.0,
    "loss_nan_ranks": 0,
    "loss_rank_avg": 0.37919333577156067,
    "mfu_percent": 0.5367250480757225,
    "mfu_percent_theoretical": 9.577115371938284,
    "total_flos": 2.363785022508368e+18,
    "train_loss": 0.4118077915944871,
    "train_runtime": 19452.6911,
    "train_samples_per_second": 2.57,
    "train_steps_per_second": 0.04,
    "valid_targets_mean": 5141.6,
    "valid_targets_min": 2704
}