laion/allenai-sera-unified-3160__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/allenai-sera-unified-3160__Qwen3-8B
Source: Original Platform
This commit is contained in:
ModelHub XC
2026-05-06 06:22:47 +08:00
commit afe1b83817
23 changed files with 153020 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": 48633.33732429779,
"achieved_tflops_per_gpu_theoretical": 678580.3444106664,
"epoch": 7.0,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.057392802089452744,
"mfu_percent": 3436.984969915038,
"mfu_percent_theoretical": 47956.20808555946,
"total_flos": 1.7948424939556045e+18,
"train_loss": 0.0,
"train_runtime": 1.1533,
"train_samples_per_second": 19179.804,
"train_steps_per_second": 200.295,
"valid_targets_mean": 15044.6,
"valid_targets_min": 4391
}