huggingface/text-generation-inference
5
0
Fork 0
mirror of https://github.com/huggingface/text-generation-inference.git synced 2025-09-12 04:44:52 +00:00
Code Issues Packages Projects Releases Wiki Activity

also quant weights with single scale

Browse Source
This commit is contained in:
OlivierDehaene 2024-07-22 18:49:10 +02:00
parent 3d0c7b85fe
commit 473f968a01
No known key found for this signature in database
GPG Key ID: BB104D67809DA93C
1 changed files with 1 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
server/text_generation_server/layers/fp8.py
View File

@ -203,7 +203,7 @@ class Fp8Linear(torch.nn.Module):
@classmethod
def from_unquant(cls, weight, bias, dtype):
qweight, scale = fp8_quantize(weight)
qweight, scale = fp8_quantize(weight, scalar=not FBGEMM_MM_AVAILABLE)
return cls(
qweight=qweight, scale=scale, scale_upper_bound=None, bias=bias, dtype=dtype
)