Files
openthoughts-4-code-qwen3-3…/train_results.json

16 lines
550 B
JSON
Raw Normal View History

{
"achieved_tflops_per_gpu": 0.009154196042913305,
"achieved_tflops_per_gpu_theoretical": 253.1246706838733,
"epoch": 8.0,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.4003600478172302,
"mfu_percent": 0.002934037193241444,
"mfu_percent_theoretical": 81.12970214226708,
"total_flos": 4.678153526260531e+16,
"train_loss": 0.13369933033527026,
"train_runtime": 19962.4709,
"train_samples_per_second": 384.408,
"train_steps_per_second": 0.376,
"valid_targets_mean": 28802.2,
"valid_targets_min": 22042
}