fix key convert for lora kohya lora

[qzzz95]

No description provided.

[gemini-code-assist]

Summary of Changes

Hello @qzzz95, I'm Gemini Code Assist¹! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed!

This pull request resolves an issue in the key conversion process for LoRA models originating from Kohya, focusing on the text encoder components. The modifications ensure that lora_te keys are accurately parsed and mapped to the correct internal model parameters, thereby improving compatibility and reliable loading of these LoRA weights. This change is essential for seamless integration and proper functionality of Kohya-trained LoRA models within the system.

Highlights

• Key Conversion Fix: Improved the key conversion logic for LoRA models trained with Kohya, specifically addressing issues with lora_te (text encoder) keys.
• Text Encoder Layer Parsing: Enhanced the parsing of text encoder layer names to correctly extract layer_id and layer_type from complex key structures.
• Module Name Standardization: Standardized the handling of self_attn and mlp module names within the text encoder layers during key mapping.

Using Gemini Code Assist

The full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips.

Invoking Gemini

You can request assistance from Gemini at any point by creating a comment using either /gemini <command> or @gemini-code-assist <command>. Below is a summary of the supported commands on the current page.

Feature	Command	Description
Code Review	/gemini review	Performs a code review for the current pull request in its current state.
Pull Request Summary	/gemini summary	Provides a summary of the current pull request in its current state.
Comment	@gemini-code-assist	Responds in comments when explicitly tagged, both in pull request comments and review comments.
Help	/gemini help	Displays a list of available commands.

Customization

To customize Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a .gemini/ folder in the base of the repository. Detailed instructions can be found here.

Limitations & Feedback

Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counter productive. You can react with 👍 and 👎 on @gemini-code-assist comments. If you're interested in giving your feedback about your experience with Gemini Code Assist for Github and other Google products, sign up here.

You can also get AI-powered code generation, chat, as well as code reviews directly in the IDE at no cost with the Gemini Code Assist IDE Extension.

Footnotes

1. Review the Privacy Notices, Generative AI Prohibited Use Policy, Terms of Service, and learn how to configure Gemini Code Assist in GitHub here. Gemini can make mistakes, so double check it and use code with caution. ↩

[gemini-code-assist]

Code Review

This pull request addresses a bug in the conversion of Kohya's LoRA state dictionaries for the text encoder. The changes correctly parse the layer keys, which use underscores as separators in Kohya's format, and translate them into the dot-separated format expected by the pipeline. The previous implementation was based on an incorrect key format and would fail. This fix is correct and improves compatibility. I have one minor suggestion to improve code conciseness.