Skip to content

Support Wan AccVideo lora #11704

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
merged 5 commits into from
Jun 13, 2025
Merged

Support Wan AccVideo lora #11704

Changes from 2 commits
Commits
Show all changes
5 commits
Select commit Hold shift + click to select a range
dc6f8f5
update
a-r-r-o-w Jun 12, 2025
d4b7f98
make style
a-r-r-o-w Jun 12, 2025
88136f5
Update src/diffusers/loaders/lora_conversion_utils.py
a-r-r-o-w Jun 12, 2025
1e66d3d
add note explaining threshold
a-r-r-o-w Jun 13, 2025
905e467
Merge branch 'main' into support-wan-accvideo-lora
a-r-r-o-w Jun 13, 2025
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
57 changes: 43 additions & 14 deletions src/diffusers/loaders/lora_conversion_utils.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1605,10 +1605,17 @@ def _convert_non_diffusers_wan_lora_to_diffusers(state_dict):
if diff_keys:
for diff_k in diff_keys:
param = original_state_dict[diff_k]
threshold = 1.6e-2
absdiff = param.abs().max() - param.abs().min()
all_zero = torch.all(param == 0).item()
if all_zero:
logger.debug(f"Removed {diff_k} key from the state dict as it's all zeros.")
all_absdiff_lower_than_threshold = absdiff < threshold
if all_zero or all_absdiff_lower_than_threshold:
logger.debug(
f"Removed {diff_k} key from the state dict as it's all zeros, or values lower than hardcoded threshold."
)
original_state_dict.pop(diff_k)
else:
print(diff_k, absdiff)

# For the `diff_b` keys, we treat them as lora_bias.
# https://huggingface.co/docs/peft/main/en/package_reference/lora#peft.LoraConfig.lora_bias
Expand Down Expand Up @@ -1655,12 +1662,16 @@ def _convert_non_diffusers_wan_lora_to_diffusers(state_dict):

# FFN
for o, c in zip(["ffn.0", "ffn.2"], ["net.0.proj", "net.2"]):
converted_state_dict[f"blocks.{i}.ffn.{c}.lora_A.weight"] = original_state_dict.pop(
f"blocks.{i}.{o}.{lora_down_key}.weight"
)
converted_state_dict[f"blocks.{i}.ffn.{c}.lora_B.weight"] = original_state_dict.pop(
f"blocks.{i}.{o}.{lora_up_key}.weight"
)
original_key = f"blocks.{i}.{o}.{lora_down_key}.weight"
converted_key = f"blocks.{i}.ffn.{c}.lora_A.weight"
if original_key in original_state_dict:
converted_state_dict[converted_key] = original_state_dict.pop(original_key)

original_key = f"blocks.{i}.{o}.{lora_up_key}.weight"
converted_key = f"blocks.{i}.ffn.{c}.lora_B.weight"
if original_key in original_state_dict:
converted_state_dict[converted_key] = original_state_dict.pop(original_key)

if f"blocks.{i}.{o}.diff_b" in original_state_dict:
converted_state_dict[f"blocks.{i}.ffn.{c}.lora_B.bias"] = original_state_dict.pop(
f"blocks.{i}.{o}.diff_b"
Expand All @@ -1669,12 +1680,16 @@ def _convert_non_diffusers_wan_lora_to_diffusers(state_dict):
# Remaining.
if original_state_dict:
if any("time_projection" in k for k in original_state_dict):
converted_state_dict["condition_embedder.time_proj.lora_A.weight"] = original_state_dict.pop(
f"time_projection.1.{lora_down_key}.weight"
)
converted_state_dict["condition_embedder.time_proj.lora_B.weight"] = original_state_dict.pop(
f"time_projection.1.{lora_up_key}.weight"
)
original_key = f"time_projection.1.{lora_down_key}.weight"
converted_key = "condition_embedder.time_proj.lora_A.weight"
if original_key in original_state_dict:
converted_state_dict[converted_key] = original_state_dict.pop(original_key)

original_key = f"time_projection.1.{lora_up_key}.weight"
converted_key = "condition_embedder.time_proj.lora_B.weight"
if original_key in original_state_dict:
converted_state_dict[converted_key] = original_state_dict.pop(original_key)

if "time_projection.1.diff_b" in original_state_dict:
converted_state_dict["condition_embedder.time_proj.lora_B.bias"] = original_state_dict.pop(
"time_projection.1.diff_b"
Expand Down Expand Up @@ -1709,6 +1724,20 @@ def _convert_non_diffusers_wan_lora_to_diffusers(state_dict):
original_state_dict.pop(f"{text_time}.{b_n}.diff_b")
)

for img_ours, img_theirs in [
("ff.net.0.proj", "img_emb.proj.1"),
("ff.net.2", "img_emb.proj.3"),
]:
original_key = f"{img_theirs}.{lora_down_key}.weight"
converted_key = f"condition_embedder.image_embedder.{img_ours}.lora_A.weight"
if original_key in original_state_dict:
converted_state_dict[converted_key] = original_state_dict.pop(original_key)

original_key = f"{img_theirs}.{lora_up_key}.weight"
converted_key = f"condition_embedder.image_embedder.{img_ours}.lora_B.weight"
if original_key in original_state_dict:
converted_state_dict[converted_key] = original_state_dict.pop(original_key)

if len(original_state_dict) > 0:
diff = all(".diff" in k for k in original_state_dict)
if diff:
Expand Down
Loading