vllm
889da130
- [ Misc ] `fp8-marlin` channelwise via `compressed-tensors` (#6524)
Go
Login via GitHub
Home
Pricing
FAQ
Install
Login
via GitHub
Commit
View On
GitHub
Commit
1 year ago
[ Misc ] `fp8-marlin` channelwise via `compressed-tensors` (#6524) Co-authored-by: mgoin <michael@neuralmagic.com>
References
#6524 - [ Misc ] `fp8-marlin` channelwise via `compressed-tensors`
Author
robertgshaw2-redhat
Parents
b75e314f
Loading