Support convrot int4 models. (#14859)

linear_dtype in comfy_quant metadata can be used to set if the int4 op does
the matrix multiplication in int8 or int4, the default is int4 on GPUs that
support it with fallback to int8 for GPUs that don't.
This commit is contained in:
comfyanonymous
2026-07-09 15:57:09 -07:00
committed by GitHub
parent 1ea724339c
commit 73e84d5ec8
4 changed files with 94 additions and 2 deletions

View File

@@ -22,7 +22,7 @@ alembic
SQLAlchemy>=2.0.0
filelock
av>=16.0.0
comfy-kitchen==0.2.16
comfy-kitchen==0.2.17
comfy-aimdo==0.4.10
requests
simpleeval>=1.0.0