Geeks
dourgey
/
qwen2_moe_mergekit
根据Qwen2（Qwen1.5）模型生成qwen2 MoE模型的工具
Primary Language
Python
Apache License 2.0
Apache-2.0
qwen2_moe_mergekit GitHub
Markdown
Readme
1
Issue
8
Stargazers
1
Watcher
Issues
Open issue
用qwen1.5 0.5B转完之后占用显存显著减小，但是推理耗时也显著增加了，这是怎么回事？
#1 opened
6 months ago
by
Willamjie
1