Files
swesmith-31600-opt100k__Qwe…/all_results.json
ModelHub XC e480180c4f 初始化项目,由ModelHub XC社区提供模型
Model: laion/swesmith-31600-opt100k__Qwen3-8B
Source: Original Platform
2026-04-11 14:49:02 +08:00

16 lines
531 B
JSON

{
"achieved_tflops_per_gpu": 72975.7683311939,
"achieved_tflops_per_gpu_theoretical": 2836323.2088544914,
"epoch": 5.0,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0014345531817525625,
"mfu_percent": 5157.2981152787215,
"mfu_percent_theoretical": 200446.86988370962,
"total_flos": 4.6024941377264026e+18,
"train_loss": 0.0,
"train_runtime": 1.9709,
"train_samples_per_second": 80167.399,
"train_steps_per_second": 837.191,
"valid_targets_mean": 5408.7,
"valid_targets_min": 410
}