Files
a1-pr_mining/train_results.json
ModelHub XC c1eec17ca4 初始化项目,由ModelHub XC社区提供模型
Model: DCAgent/a1-pr_mining
Source: Original Platform
2026-05-13 10:44:38 +08:00

16 lines
543 B
JSON

{
"achieved_tflops_per_gpu": 0.003940310905920059,
"achieved_tflops_per_gpu_theoretical": 548.8405391773532,
"epoch": 7.0,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1299251914024353,
"mfu_percent": 0.0002784672018318063,
"mfu_percent_theoretical": 38.78731725635005,
"total_flos": 1654874798292992.0,
"train_loss": 0.18129448925475283,
"train_runtime": 26249.1152,
"train_samples_per_second": 2.586,
"train_steps_per_second": 0.162,
"valid_targets_mean": 5076.1,
"valid_targets_min": 783
}