This RFC proposes best current practices for managing AI-generated contributions to open-source projects, addressing challenges such as automated pull requests, code quality, and community impact. It provides guidelines for project maintainers to handle contributions from artificial contributors while preserving project integrity.
InferenceBench is a new benchmark designed to evaluate how well AI agents can optimize their own inference strategies for open-ended questions, moving beyond standard single-response accuracy tests by measuring cost-efficiency and iterative improvement.
MemEye is a visual-centric evaluation framework designed to assess the memory capabilities of multimodal AI agents by testing how well they retain and recall visual information across interactions. The framework provides a structured benchmark to measure an agent's ability to remember past visual inputs, supporting research into more persistent and context-aware multimodal AI systems.