Skip to content

qwen: reduce VRAM usage #10725

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

Merged
comfyanonymous merged 1 commit into from
Nov 12, 2025
Merged

qwen: reduce VRAM usage #10725

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
20 changes: 12 additions & 8 deletions comfy/ldm/qwen_image/model.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -236,10 +236,10 @@ def forward(
img_mod1, img_mod2 = img_mod_params.chunk(2, dim=-1)
txt_mod1, txt_mod2 = txt_mod_params.chunk(2, dim=-1)

img_normed = self.img_norm1(hidden_states)
img_modulated, img_gate1 = self._modulate(img_normed, img_mod1)
txt_normed = self.txt_norm1(encoder_hidden_states)
txt_modulated, txt_gate1 = self._modulate(txt_normed, txt_mod1)
img_modulated, img_gate1 = self._modulate(self.img_norm1(hidden_states), img_mod1)
del img_mod1
txt_modulated, txt_gate1 = self._modulate(self.txt_norm1(encoder_hidden_states), txt_mod1)
del txt_mod1

img_attn_output, txt_attn_output = self.attn(
hidden_states=img_modulated,
Expand All @@ -248,16 +248,20 @@ def forward(
image_rotary_emb=image_rotary_emb,
transformer_options=transformer_options,
)
del img_modulated
del txt_modulated

hidden_states = hidden_states + img_gate1 * img_attn_output
encoder_hidden_states = encoder_hidden_states + txt_gate1 * txt_attn_output
del img_attn_output
del txt_attn_output
del img_gate1
del txt_gate1

img_normed2 = self.img_norm2(hidden_states)
img_modulated2, img_gate2 = self._modulate(img_normed2, img_mod2)
img_modulated2, img_gate2 = self._modulate(self.img_norm2(hidden_states), img_mod2)
hidden_states = torch.addcmul(hidden_states, img_gate2, self.img_mlp(img_modulated2))

txt_normed2 = self.txt_norm2(encoder_hidden_states)
txt_modulated2, txt_gate2 = self._modulate(txt_normed2, txt_mod2)
txt_modulated2, txt_gate2 = self._modulate(self.txt_norm2(encoder_hidden_states), txt_mod2)
encoder_hidden_states = torch.addcmul(encoder_hidden_states, txt_gate2, self.txt_mlp(txt_modulated2))

return encoder_hidden_states, hidden_states
Expand Down