A BENCHMARK FOR EVALUATING CODE REPOSITORY EDITS

Summary from the TLDR AI newsletter

RES-Q is a new benchmark for assessing the ability of large language models to edit code repositories based on natural language instructions.