Files
100k_baseline__Qwen3-8B/all_results.json

16 lines
527 B
JSON
Raw Permalink Normal View History

{
"achieved_tflops_per_gpu": 63.83894743154135,
"achieved_tflops_per_gpu_theoretical": 9434148.768292911,
"epoch": 7.0,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.22483697533607483,
"mfu_percent": 4.51158639092165,
"mfu_percent_theoretical": 666724.2945790043,
"total_flos": 1.326951254196224e+16,
"train_loss": 0.0,
"train_runtime": 1.6239,
"train_samples_per_second": 356216.071,
"train_steps_per_second": 2784.656,
"valid_targets_mean": 2756.4,
"valid_targets_min": 182
}