Summary of Changes

Hello @Koko-ry, 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 introduces significant enhancements to the model creation process within the ModelScope Hub API, specifically by adding comprehensive support for AIGC (AI-Generated Content) models. I've extended the "create_model" function to accept AIGC-specific configurations, which now directs the creation request to a dedicated AIGC endpoint and automatically handles the upload of model files. A new utility class, "AigcModel", has been added to encapsulate and manage the various parameters and file processing required for AIGC model types like Checkpoints, LoRAs, and VAEs.

Highlights

• Enhanced Model Creation for AIGC: The "create_model" method in "modelscope/hub/api.py" has been updated to support the creation of AIGC models, accepting a new "aigc_config" parameter that specifies the model's type, base model, and file details.
• Dedicated AIGC Endpoint Integration: When an "aigc_config" is provided, the API call for model creation is now routed to a specific AIGC endpoint ("/api/v1/models/aigc"), ensuring proper categorization and handling of these specialized models.
• New "AigcModel" Utility: A new "AigcModel" class has been introduced in "modelscope/hub/utils/aigc.py". This class streamlines the management of AIGC model metadata, including defining "AigcType" (Checkpoint, LoRA, VAE) and "BaseModelType" (e.g., SD_1_5, SD_XL), and handles the processing of model file paths to extract necessary information like filename, SHA256 hash, and size.
• Automated Model File Upload: The "create_model" function now includes logic to automatically upload the AIGC model's associated files (either a single file or an entire directory) to the newly created repository, simplifying the workflow for users.
• Robust AIGC Parameter Validation: Assertions have been added to validate the "aigc_type" and "base_model_type" provided in the "aigc_config", ensuring that only valid AIGC model configurations are accepted.

Code Review

This pull request introduces the capability to create AIGC models via the Hub API, adding an AigcModel helper class for managing AIGC-specific parameters and file operations. The changes to HubApi.create_model are well-organized to support this new functionality. My review highlights two main areas for improvement: replacing assert statements with more robust validation to prevent them from being disabled in production, and standardizing logging by using the logging module instead of print() for better consistency and control. Additionally, there is a minor typo in the pull request title ('updata' should be 'update').