Files
100k_baseline__Qwen3-8B/train_results.json
ModelHub XC d53bee724d 初始化项目,由ModelHub XC社区提供模型
Model: laion/100k_baseline__Qwen3-8B
Source: Original Platform
2026-05-26 19:33:41 +08:00

12 lines
396 B
JSON

{
"achieved_tflops_per_gpu": 63.83894743154135,
"achieved_tflops_per_gpu_theoretical": 9434148.768292911,
"epoch": 7.0,
"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
}