admin/LLM-from-scratch
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

[2025.05.22] add MoE support

Browse Source
This commit is contained in:
coffee
2025-05-22 18:11:50 +08:00
parent 6152ab5a35
commit 9ae88672f5
1 changed files with 1 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files
README.md
+1
View File
@@ -6,6 +6,7 @@ LLM-Dolly is a custom LLM built from scratch
#### 2025-05-22
- 📝 添加支持MoE模型，训练资源占用不稳定，测试模型Experts=8, experts_per_tok=4,GPU显存从60%-94%跳动,单个GPU利用率0%-100%跳动。
- 当调小batch_size后，显存占用稳定在90%，单个GPU利用率稳定在90%以上，偶现30%的利用率。
#### 2025-05-20
- 📝 添加支持jsonl文件训练.