Fetching the paper…

PerfCodeGen: Improving Performance of LLM Generated Code with Execution Feedback · Around