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

16 lines
547 B
JSON
Raw Normal View History

{
"achieved_tflops_per_gpu": 0.005619164353050912,
"achieved_tflops_per_gpu_theoretical": 323.025179674411,
"epoch": 2.0,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.6287916302680969,
"mfu_percent": 0.001801014215721446,
"mfu_percent_theoretical": 103.53371143410608,
"total_flos": 2.788441056922829e+16,
"train_loss": 0.21610206831669151,
"train_runtime": 19384.2842,
"train_samples_per_second": 98.968,
"train_steps_per_second": 0.387,
"valid_targets_mean": 15631.9,
"valid_targets_min": 505
}