Files
GLM-4_6-stackexchange-super…/train_results.json
ModelHub XC 5efc4852de 初始化项目,由ModelHub XC社区提供模型
Model: laion/GLM-4_6-stackexchange-superuser-32ep-32k
Source: Original Platform
2026-04-23 11:31:00 +08:00

16 lines
543 B
JSON

{
"achieved_tflops_per_gpu": 0.002928386808870264,
"achieved_tflops_per_gpu_theoretical": 494.6340661721825,
"epoch": 7.0,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.24434766173362732,
"mfu_percent": 0.00020695313136892326,
"mfu_percent_theoretical": 34.956471107574735,
"total_flos": 459870247321600.0,
"train_loss": 0.3203928537069713,
"train_runtime": 9814.9228,
"train_samples_per_second": 2.332,
"train_steps_per_second": 0.146,
"valid_targets_mean": 2891.9,
"valid_targets_min": 658
}