Files
100k_epochs3__Qwen3-8B/train_results.json
ModelHub XC cef191ed20 初始化项目,由ModelHub XC社区提供模型
Model: laion/100k_epochs3__Qwen3-8B
Source: Original Platform
2026-05-07 12:22:59 +08:00

12 lines
394 B
JSON

{
"achieved_tflops_per_gpu": 7.906672017494305,
"achieved_tflops_per_gpu_theoretical": 1168447.6288629568,
"epoch": 3.0,
"mfu_percent": 0.5587754075967707,
"mfu_percent_theoretical": 82575.80415992628,
"total_flos": 5686933939290112.0,
"train_loss": 0.0,
"train_runtime": 5.6192,
"train_samples_per_second": 44118.376,
"train_steps_per_second": 344.888
}