Files
swesmith-316__Qwen3-8B/all_results.json
ModelHub XC f6662fd7e7 初始化项目,由ModelHub XC社区提供模型
Model: laion/swesmith-316__Qwen3-8B
Source: Original Platform
2026-04-29 18:04:46 +08:00

16 lines
540 B
JSON

{
"achieved_tflops_per_gpu": 2.551643175725337,
"achieved_tflops_per_gpu_theoretical": 132.12827795010355,
"epoch": 7.0,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.20894506573677063,
"mfu_percent": 0.1803281396272323,
"mfu_percent_theoretical": 9.337687487639826,
"total_flos": 5.930599290241024e+16,
"train_loss": 0.2524434221642358,
"train_runtime": 726.3211,
"train_samples_per_second": 3.045,
"train_steps_per_second": 0.039,
"valid_targets_mean": 5717.0,
"valid_targets_min": 1502
}