auto-round
3a04e2bc
- refine moe modellings to reduce peak ram usage
Go
Login via GitHub
Home
Pricing
FAQ
Install
Login
via GitHub
Commit
View On
GitHub
Commit
70 days ago
refine moe modellings to reduce peak ram usage Signed-off-by: Zhang, Weiwei1 <weiwei1.zhang@intel.com>
References
#1265 - refine moe modellings to release orginal expert module's ram
Author
WeiweiZhang1
Parents
1b7535a0
Loading