[LoRA serialization] fix: duplicate unet prefix problem. #5991
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
|
The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. |
sayakpaul
commented
Nov 30, 2023
Comment on lines
+812
to
+823
| def pack_weights(layers, prefix): | ||
| layers_weights = layers.state_dict() if isinstance(layers, torch.nn.Module) else layers | ||
| layers_state_dict = {f"{prefix}.{module_name}": param for module_name, param in layers_weights.items()} | ||
| return layers_state_dict | ||
|
|
||
| unet_lora_state_dict = {f"{cls.unet_name}.{module_name}": param for module_name, param in weights.items()} | ||
| state_dict.update(unet_lora_state_dict) | ||
| if not (unet_lora_layers or text_encoder_lora_layers): | ||
| raise ValueError("You must pass at least one of `unet_lora_layers`, `text_encoder_lora_layers`.") | ||
|
|
||
| if text_encoder_lora_layers is not None: | ||
| weights = ( | ||
| text_encoder_lora_layers.state_dict() | ||
| if isinstance(text_encoder_lora_layers, torch.nn.Module) | ||
| else text_encoder_lora_layers | ||
| ) | ||
| if unet_lora_layers: | ||
| state_dict.update(pack_weights(unet_lora_layers, "unet")) | ||
|
|
||
| text_encoder_lora_state_dict = { | ||
| f"{cls.text_encoder_name}.{module_name}": param for module_name, param in weights.items() | ||
| } | ||
| state_dict.update(text_encoder_lora_state_dict) | ||
| if text_encoder_lora_layers: |
There was a problem hiding this comment.
All these changes to keep it clean and unified with respect to how it's done for the SDXL LoRA writer.
sayakpaul
commented
Nov 30, 2023
| @@ -67,7 +67,7 @@ def unet_lora_state_dict(unet: UNet2DConditionModel) -> Dict[str, torch.Tensor]: | |||
| current_lora_layer_sd = lora_layer.state_dict() | |||
| for lora_layer_matrix_name, lora_param in current_lora_layer_sd.items(): | |||
| # The matrix name can either be "down" or "up". | |||
| lora_state_dict[f"unet.{name}.lora.{lora_layer_matrix_name}"] = lora_param | |||
| lora_state_dict[f"{name}.lora.{lora_layer_matrix_name}"] = lora_param | |||
There was a problem hiding this comment.
This was the main culprit!
|
Do you think we could merge @patrickvonplaten ? |
|
Thanks a mille for fixing this @sayakpaul ! |
patrickvonplaten
approved these changes
Dec 1, 2023
Co-authored-by: Patrick von Platen <[email protected]>
yoonseokjin
pushed a commit
to yoonseokjin/diffusers
that referenced
this pull request
Dec 25, 2023
…#5991) * fix: duplicate unet prefix problem. * Update src/diffusers/loaders/lora.py Co-authored-by: Patrick von Platen <[email protected]> --------- Co-authored-by: Patrick von Platen <[email protected]>
AmericanPresidentJimmyCarter
pushed a commit
to AmericanPresidentJimmyCarter/diffusers
that referenced
this pull request
Apr 26, 2024
…#5991) * fix: duplicate unet prefix problem. * Update src/diffusers/loaders/lora.py Co-authored-by: Patrick von Platen <[email protected]> --------- Co-authored-by: Patrick von Platen <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
What does this PR do?
Fixes: #5977
Trained a LoRA: https://huggingface.co/sayakpaul/corgy_dog_LoRA/tree/main (Colab: https://colab.research.google.com/gist/sayakpaul/bcd42382f111d5c0e3bf8b37b55342ff/sdxl_dreambooth_lora_.ipynb)
Doing the above yields:
Cc: @apolinario