• cookbook/client/tinker/custom_service/short_math_grpo.py
  • Refactored MathPreprocessor to use a new preprocess method for single samples and updated the __call__ method to handle batches of rows, converting between column and row formats.
• cookbook/client/tinker/modelscope_service/short_math_grpo.py
  • Refactored MathPreprocessor to use a new preprocess method for single samples and updated the __call__ method to handle batches of rows, converting between column and row formats.
• cookbook/mm/fsdp2.py
  • Refactored LatexOCRProcessor to use a new preprocess method for single samples and updated the __call__ method to handle batches of rows, converting between column and row formats.
• docs/source_en/Components/Preprocessor and Filter/Preprocessor.md
  • Updated the Preprocessor class signature in the documentation to reflect the new batched input/output format (List[Dict] to List[Trajectory]).
  • Clarified that the __call__ method now expects a list of samples and returns a list of trajectories.
• docs/source_zh/组件/预处理器和过滤器/Preprocessor.md
  • Updated the Preprocessor class signature in the Chinese documentation to reflect the new batched input/output format (List[Dict] to List[Trajectory]).
  • Clarified that the __call__ method now expects a list of samples and returns a list of trajectories, and that input/output counts may differ.
• src/twinkle/dataset/base.py
  • Changed the batched argument in the map function from False to True by default, enabling batched processing for datasets.
  • The filter function's batched argument remains False.
• src/twinkle/preprocessor/base.py
  • Added map_col_to_row and map_row_to_col static methods to convert between column-oriented and row-oriented data representations.
  • Modified the __call__ method signature to accept and return dictionaries of lists (batched data).
• src/twinkle/preprocessor/llm.py
  • Refactored CompetitionMathProcessor, CompetitionMathGRPOProcessor, SelfCognitionProcessor, AlpacaProcessor, CountdownProcessor, and GSM8KProcessor to support batched processing.
  • Each now has a preprocess method for single samples and an updated __call__ method that handles batch conversion.
• tests/preprocessor/test_preprocessor.py
  • Updated the ModifiedProcessor test class to align with the new batched Preprocessor interface, specifically by introducing a preprocess method and modifying the __call__ method to handle batches.

• No specific activity (comments, reviews, progress updates) was provided in the context for this pull request.

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.

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.