wangbo/ComfyUI
1
0
Fork 0
mirror of https://github.com/comfyanonymous/ComfyUI.git synced 2026-07-03 13:19:23 +08:00
Code Issues Actions 19 Packages Projects Releases Wiki Activity

Fix int8 mm being skipped on offloaded lora weights. (#14669)

Browse Source
This commit is contained in:
comfyanonymous 2026-06-28 20:52:36 -07:00 committed by GitHub
parent f19735759e
commit 79c555ce6b
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 1 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
comfy/ops.py
View File

@ -1216,7 +1216,7 @@ def mixed_precision_ops(quant_config={}, compute_dtype=torch.bfloat16, full_prec
bias_dtype=input.dtype, bias_dtype=input.dtype,
offloadable=True, offloadable=True,
compute_dtype=compute_dtype, compute_dtype=compute_dtype,
want_requant=want_requant, want_requant=True,
) )
weight = weight.to(dtype=input.dtype) weight = weight.to(dtype=input.dtype)
else: else: