Add --fp16-intermediates to use fp16 for intermediate values between nodes (#12953)

This is an experimental WIP option that might not work in your workflow but
should lower memory usage if it does.

Currently only the VAE and the load image node will output in fp16 when
this option is turned on.
This commit is contained in:
comfyanonymous
2026-03-14 16:18:19 -07:00
committed by GitHub
parent 1c5db7397d
commit c711b8f437
4 changed files with 27 additions and 14 deletions

View File

@ -83,6 +83,8 @@ fpte_group.add_argument("--fp16-text-enc", action="store_true", help="Store text
fpte_group.add_argument("--fp32-text-enc", action="store_true", help="Store text encoder weights in fp32.")
fpte_group.add_argument("--bf16-text-enc", action="store_true", help="Store text encoder weights in bf16.")
parser.add_argument("--fp16-intermediates", action="store_true", help="Experimental: Use fp16 for intermediate tensors between nodes instead of fp32.")
parser.add_argument("--force-channels-last", action="store_true", help="Force channels last format when inferencing the models.")
parser.add_argument("--directml", type=int, nargs="?", metavar="DIRECTML_DEVICE", const=-1, help="Use torch-directml.")