[bug]: CLIP Model Download Fails to Create Required Subdirectory Structure #8376
Description
Is there an existing issue for this problem?
- I have searched the existing issues
Install method
Invoke's Launcher
Operating system
Windows
GPU vendor
Nvidia (CUDA)
GPU model
RTX 5090
GPU VRAM
32GB
Version number
v6.3.0rc1
Browser
No response
System Information
No response
What happened
InvokeAI version 6.3.0rc1 on Windows fails to create the necessary subdirectory structure for CLIP models, leading to HFValidationError.
When downloading the CLIP-GmP-ViT-L-14 model as a dependency for the FLUX text encoder, the model files are saved directly into the root directory of the model instead of the required text_encoder and tokenizer subdirectories.
This results in the following error during loading:
HFValidationError: Repo id must use alphanumeric chars or '-', '_', '.', '--' and '..' are forbidden, '-' and '.' cannot start or end the name, max length is 96: 'D:\Invoke\models\any\clip_embed\CLIP-GmP-ViT-L-14\text_encoder'.
The current download structure:
CLIP-GmP-ViT-L-14/ ├── config.json ├── merges.txt ├── model.safetensors ├── preprocessor_config.json ├── special_tokens_map.json ├── tokenizer_config.json ├── tokenizer.json └── vocab.json
What you expected to happen
Expected Structure:
CLIP-GmP-ViT-L-14/ ├── text_encoder/ │ ├── config.json │ └── model.safetensors ├── tokenizer/ │ ├── merges.txt │ ├── special_tokens_map.json │ ├── tokenizer_config.json │ ├── tokenizer.json │ └── vocab.json ├── config.json └── preprocessor_config.json
How to reproduce the problem
Steps to Reproduce:
- Install InvokeAI and attempt to download the FLUX model requiring CLIP-GmP-ViT-L-14.
- InvokeAI downloads the CLIP model but fails to create necessary subdirectories.
- An error occurs when loading the model due to incorrect path structure.
Additional context
Workaround:
Manually create the text_encoder and tokenizer directories and move the appropriate files into these subdirectories.
Discord username
No response