SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2412.19513 · 2024

Confidence v.s. Critique: A Decomposition of Self-Correction Capability for LLMs

arXiv · PDF · Open in the Atlas

Code that ran

We lifted 5 functions out of this paper's own repositories and ran 1 of them in a sandbox. "Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results.

RepositoryRoleRan
Zhe-Young/SelfCorrectDecompose canonical 1 of 5
FunctionStatusWhere it lives
get_examples Ran Zhe-Young/SelfCorrectDecompose/eval_logits.py
pointer only (licence: NONE) · get_code("fbe652c7fa7abfbc")
generate_answer Not yet run Zhe-Young/SelfCorrectDecompose/eval_sampling.py
pointer only (licence: NONE) · get_code("dd6378ce27ea29b1")
generate_answer1 Not yet run Zhe-Young/SelfCorrectDecompose/eval_sampling.py
pointer only (licence: NONE) · get_code("4ec0e3f68b16920f")
get_answer Not yet run Zhe-Young/SelfCorrectDecompose/eval_logits.py
pointer only (licence: NONE) · get_code("e888e7e76d03f6fa")
get_probs Not yet run Zhe-Young/SelfCorrectDecompose/eval_logits.py
pointer only (licence: NONE) · get_code("fcc620fb7cb1bb32")

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

Large Language Models (LLMs) can correct their self-generated responses, but a decline in accuracy after self-correction is also witnessed. To have a deeper understanding of self-correction, we endeavor to decompose, evaluate, and analyze the self-correction behaviors of LLMs. By enumerating and analyzing answer correctness before and after self-correction, we decompose the self-correction capability into confidence (being confident to correct answers) and critique (turning wrong answers to correct) capabilities, and propose two metrics from a probabilistic perspective to measure these 2 capabilities, along with another metric for overall self-correction capability evaluation. Based on our decomposition and evaluation metrics, we conduct extensive experiments and draw some empirical conclusions. For example, we find different models can exhibit distinct behaviors: some models are confident while others are more critical. We also find the trade-off between the two capabilities (i.e. improving one can lead to a decline in the other) when manipulating model self-correction behavior by prompts or in-context learning. Further, we find a simple yet efficient strategy to improve self-correction capability by transforming Supervision Fine-Tuning (SFT) data format, and our strategy outperforms vanilla SFT in both capabilities and achieves much higher accuracy after self-correction. Our code will be publicly available on GitHub.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2412.19513")
get_code_for_paper("2412.19513")
have("2412.19513")

Connect an agent — have() is free.