Reference library
Large Language Models Cannot Self-Correct Reasoning Yet
Paper summary
Tests intrinsic reasoning self-correction without external feedback. In the evaluated settings, revision often fails to improve answers and can degrade them, challenging assumptions that another self-critique pass reliably fixes errors.
Relevance to the project
Directly informs the origins-of-evaluation question: generating feedback does not guarantee that it carries new evidence or a valid correctness criterion.
The findings do not rule out externally trained or externally verified correction.
Source record
- Reference ID
- article-https-proceedings-iclr-cc-paper-files-paper-2024-hash-8b4add8b0aa8749d80a34ca5d941c355-abstract-conference-html
- Archived version
- Publisher/institution PDF advertised by article metadata
- Summary provenance
- reference-library/papers/article-https-proceedings-iclr-cc-paper-files-paper-2024-hash-8b4add8b0aa8749d80a34ca5d941c355-abstract-conference-html/metadata.json · summary
- Review status
- initial orientation summary; not a critical full-paper review
- Reading scope
- Selected abstract/opening and available closing discussion/conclusion passages in the preserved text extract; broader project notes provide context. No reproduction of experiments, exhaustive methods review, or equation/figure audit.