Skip to content

lowvram #50

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
peizhiluo007 wants to merge 1 commit into
base: main
Choose a base branch
from
Open

lowvram #50

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
36 changes: 18 additions & 18 deletions src/nodes/pipeline_loader.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -22,6 +22,7 @@ def INPUT_TYPES(s):
return {
"required": {
"weight_dtype": (("float32", "float16", "bfloat16"), ),
"lowvram": ("BOOLEAN", {"default": True}),
}
}

Expand All @@ -30,14 +31,14 @@ def INPUT_TYPES(s):
RETURN_TYPES = ("PIPELINE",)
FUNCTION = "load_pipeline"

def load_pipeline(self, weight_dtype):
def load_pipeline(self, weight_dtype, lowvram):
if weight_dtype == "float32":
weight_dtype = torch.float32
elif weight_dtype == "float16":
weight_dtype = torch.float16
elif weight_dtype == "bfloat16":
weight_dtype = torch.bfloat16

noise_scheduler = DDPMScheduler.from_pretrained(
WEIGHTS_PATH,
subfolder="scheduler"
Expand All @@ -47,45 +48,38 @@ def load_pipeline(self, weight_dtype):
WEIGHTS_PATH,
subfolder="vae",
torch_dtype=weight_dtype
).requires_grad_(False).eval().to(DEVICE)

).requires_grad_(False).eval()
unet = UNet2DConditionModel.from_pretrained(
WEIGHTS_PATH,
subfolder="unet",
torch_dtype=weight_dtype
).requires_grad_(False).eval().to(DEVICE)

).requires_grad_(False).eval()
image_encoder = CLIPVisionModelWithProjection.from_pretrained(
WEIGHTS_PATH,
subfolder="image_encoder",
torch_dtype=weight_dtype
).requires_grad_(False).eval().to(DEVICE)

).requires_grad_(False).eval()
unet_encoder = UNet2DConditionModel_ref.from_pretrained(
WEIGHTS_PATH,
subfolder="unet_encoder",
torch_dtype=weight_dtype
).requires_grad_(False).eval().to(DEVICE)

).requires_grad_(False).eval()
text_encoder_one = CLIPTextModel.from_pretrained(
WEIGHTS_PATH,
subfolder="text_encoder",
torch_dtype=weight_dtype
).requires_grad_(False).eval().to(DEVICE)

).requires_grad_(False).eval()
text_encoder_two = CLIPTextModelWithProjection.from_pretrained(
WEIGHTS_PATH,
subfolder="text_encoder_2",
torch_dtype=weight_dtype
).requires_grad_(False).eval().to(DEVICE)

).requires_grad_(False).eval()
tokenizer_one = AutoTokenizer.from_pretrained(
WEIGHTS_PATH,
subfolder="tokenizer",
revision=None,
use_fast=False,
)

tokenizer_two = AutoTokenizer.from_pretrained(
WEIGHTS_PATH,
subfolder="tokenizer_2",
Expand All @@ -106,8 +100,14 @@ def load_pipeline(self, weight_dtype):
image_encoder=image_encoder,
torch_dtype=weight_dtype,
)
pipe.unet_encoder = unet_encoder
pipe = pipe.to(DEVICE)
pipe.weight_dtype = weight_dtype

pipe.unet_encoder = unet_encoder
pipe.unet_encoder.to(DEVICE)

print("[VTON]:--->lowvram:",lowvram)
if lowvram == True:
pipe.enable_sequential_cpu_offload()
else:
pipe.to(DEVICE)
#
return (pipe, )