laion/nemotron-31600-opt100k__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-31600-opt100k__Qwen3-8B
Source: Original Platform
This commit is contained in:
ModelHub XC
2026-05-09 17:15:10 +08:00
commit b271f3bf83
23 changed files with 156487 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

16
train_results.json Normal file
View File

@@ -0,0 +1,16 @@
{
"achieved_tflops_per_gpu": 15.292781648433825,
"achieved_tflops_per_gpu_theoretical": 423.488070352896,
"epoch": 5.0,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.38277196884155273,
"mfu_percent": 1.0807619539529205,
"mfu_percent_theoretical": 29.928485537307136,
"total_flos": 6.453442034819662e+18,
"train_loss": 0.1001221828027205,
"train_runtime": 13187.2715,
"train_samples_per_second": 11.981,
"train_steps_per_second": 0.125,
"valid_targets_mean": 11949.7,
"valid_targets_min": 2234
}