REVIEW 4 cited by
Understanding the Dark Side of LLMs' Intrinsic Self-Correction
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
read the original abstract
Intrinsic self-correction was proposed to improve LLMs' responses via feedback prompts solely based on their inherent capability. However, recent works show that LLMs' intrinsic self-correction fails without oracle labels as feedback prompts. In this paper, we aim to interpret LLMs' intrinsic self-correction for different tasks, especially for those failure cases. By including one simple task and three complex tasks with state-of-the-art (SOTA) LLMs like ChatGPT families (o1, 4o, 3.5-turbo) and Llama families (2-7B, 3-8B, and 3.1-8B), we design three interpretation methods to reveal the dark side of LLMs' intrinsic self-correction. We identify intrinsic self-correction can (1) cause LLMs to waver both intermedia and final answers and lead to prompt bias on simple factual questions; (2) introduce human-like cognitive bias on complex tasks. In light of our findings, we also provide two simple yet effective strategies for alleviation: question repeating and supervised fine-tuning with a few samples. We open-source our work at https://x-isc.info/.
Forward citations
Cited by 4 Pith papers
-
Think Through a Bottleneck: Hourglass Reasoning for Rigorous Induction
Strict stage isolation that passes only a compressed symbolic schema and rule between LLM calls improves few-shot inductive reasoning more than self-refinement or explicit verbalization alone.
-
SAC-Opt: Semantic Anchors for Iterative Correction in Optimization Modeling
A semantic-anchor alignment loop that reconstructs and verifies constraints from generated solver code lifts average modeling accuracy by 7.8 percentage points over seven datasets.
-
Unlocking Recursive Thinking of LLMs: Alignment via Refinement
An offline alignment pipeline using reward-filtered self-refinement data and long chain-of-thought SFT raises an 8B model's AlpacaEval 2 win rate from 25.0% to 51.0% with roughly 14k training examples.
-
Exploring Multimodal Challenges in Toxic Chinese Detection: Taxonomy, Benchmark, and Findings
A new taxonomy and dataset of 8 types of perturbed toxic Chinese show nine top LLMs often miss these obfuscated insults, and small-sample ICL or fine-tuning causes overcorrection.
Discussion (0). Sign in to comment.