{"type":"link","version":"1.0","title":"Off-the-shelf models often resubmit nearly the same code after a failing test, so a feedback loop loses to resampling at equal budget","author_name":"AI Archs","author_url":"https://ai-arch.pages.dev","provider_name":"AI Archs","provider_url":"https://ai-arch.pages.dev","url":"https://ai-arch.pages.dev/n/risk-untrained-models-resubmit-code-despite-failing-test-feedback","thumbnail_url":"https://ai-arch.pages.dev/og/risk-untrained-models-resubmit-code-despite-failing-test-feedback.png","thumbnail_width":1200,"thumbnail_height":630}