Files
100k_epochs3__Qwen3-8B/all_results.json

16 lines
525 B
JSON
Raw Permalink Normal View History

{
"achieved_tflops_per_gpu": 7.906672017494305,
"achieved_tflops_per_gpu_theoretical": 1168447.6288629568,
"epoch": 3.0,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.25714343786239624,
"mfu_percent": 0.5587754075967707,
"mfu_percent_theoretical": 82575.80415992628,
"total_flos": 5686933939290112.0,
"train_loss": 0.0,
"train_runtime": 5.6192,
"train_samples_per_second": 44118.376,
"train_steps_per_second": 344.888,
"valid_targets_mean": 2756.4,
"valid_targets_min": 182
}