Assistance Required for fudan-generative-ai/hallo2 Implementation and Model Weight Issues

Models
1

Dear Hugging Face Team,

I am reaching out for assistance with a technical issue I am facing while using the project fudan-generative-ai/hallo2.

Issue: I am currently working with the project fudan-generative-ai/hallo2 on a Windows 11 Home system. I have been encountering continuous errors related to missing or incorrect model weights. Despite several attempts to implement and load the correct weights, I continue to receive numerous errors.

Here are the key steps I have taken so far:

  • Using Stable Diffusion v1.5 model weights.
  • Verified multiple times that the model weights were downloaded and implemented according to the documentation.

Error Message: One of the common errors I encounter is the following:

javascript

Code kopieren

Cannot load <class 'hallo.models.unet_2d_condition.UNet2DConditionModel'> from ./pretrained_models/stable-diffusion-v1-5 because the following keys are missing:

The list of missing model weights is very long and seems to indicate that several weights are either missing or incorrectly loaded.

Request: Could you please advise if there is a specific method or source for downloading the correct and complete set of model weights for the fudan-generative-ai/hallo2 project? I would like to ensure that no weights are missing and that all modules are correctly set up.

I would greatly appreciate your assistance in resolving this issue. Thank you in advance for your help, and I look forward to your response!

Best regards,
SKI

1 Like
2

There are several possible causes, but as is often the case with Windows, did you miss the git clone?
Downloading large files is basically prone to failure if the latest version of git and git-lfs are not installed.

If it is a bug in the model or program, you can communicate it to the author by opening a Discussion in the community on the page below.

They’re reporting a git error. This is a bug somewhere. I don’t think it’s your fault.