laion/nemotron-terminal-adapters_math__Qwen3-8B
1
0
Fork 0
Code Issues Pull Requests Actions Projects Releases Wiki Activity

初始化项目，由ModelHub XC社区提供模型

Browse Source 
Model: laion/nemotron-terminal-adapters_math__Qwen3-8B
Source: Original Platform
This commit is contained in:
ModelHub XC
2026-05-09 22:28:36 +08:00
commit ca39820f19
23 changed files with 173151 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

16
all_results.json Normal file
View File

@@ -0,0 +1,16 @@
{
"achieved_tflops_per_gpu": 24.043768590547494,
"achieved_tflops_per_gpu_theoretical": 762.2785266352129,
"epoch": 5.0,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.10146471858024597,
"mfu_percent": 1.6992062608160772,
"mfu_percent_theoretical": 53.87127396715285,
"total_flos": 2.892950530069994e+19,
"train_loss": 0.05187808655707885,
"train_runtime": 37600.0558,
"train_samples_per_second": 21.635,
"train_steps_per_second": 0.225,
"valid_targets_mean": 9906.2,
"valid_targets_min": 1715
}