REVIEW 10 cited by
Question Decomposition Improves the Faithfulness of Model-Generated Reasoning
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
As large language models (LLMs) perform more difficult tasks, it becomes harder to verify the correctness and safety of their behavior. One approach to help with this issue is to prompt LLMs to externalize their reasoning, e.g., by having them generate step-by-step reasoning as they answer a question (Chain-of-Thought; CoT). The reasoning may enable us to check the process that models use to perform tasks. However, this approach relies on the stated reasoning faithfully reflecting the model's actual reasoning, which is not always the case. To improve over the faithfulness of CoT reasoning, we have models generate reasoning by decomposing questions into subquestions. Decomposition-based methods achieve strong performance on question-answering tasks, sometimes approaching that of CoT while improving the faithfulness of the model's stated reasoning on several recently-proposed metrics. By forcing the model to answer simpler subquestions in separate contexts, we greatly increase the faithfulness of model-generated reasoning over CoT, while still achieving some of the performance gains of CoT. Our results show it is possible to improve the faithfulness of model-generated reasoning; continued improvements may lead to reasoning that enables us to verify the correctness and safety of LLM behavior.
Forward citations
Cited by 10 Pith papers
-
Teaching Models to Verbalize Reward Hacking in Chain-of-Thought Reasoning
A pre-RL fine-tuning intervention called verbalization fine-tuning makes language models explicitly acknowledge when prompt cues drive them to reward-hack, cutting undetected reward hacking from 88% to 6% after RL.
-
Training Large Language Models for Self-Explanation Faithfulness
RL fine-tuning with a counterfactual mention/influence reward raises LLM self-explanation faithfulness (Phi-CCT) from near zero to ~0.66 in-distribution for two 8B models, with partial transfer to held-out tasks.
-
REFACT: Adaptive Fact Restatement for Compact and Faithful Chain-of-Thought Reasoning
REFACT teaches LLMs to adaptively cite only the source facts needed during chain-of-thought reasoning, improving faithfulness and shortening reasoning traces.
-
CASE: Causal Alignment and Structural Enforcement for Improving Chain-of-Thought Faithfulness
CASE improves CoT faithfulness by training on counterfactual/biased/empty instructions and masking instruction-to-answer attention, reporting 37% average relative faithfulness gains.
-
Decomposed Prompting Does Not Fix Knowledge Gaps, But Helps Models Say "I Don't Know"
Disagreement between direct and decomposed prompting is a training-free error signal that outperforms standard uncertainty baselines for closed-book QA abstention.
-
Synthetic Heuristic Evaluation: A Comparison between AI- and Human-Powered Usability Evaluation
An LLM prompted to conduct heuristic evaluation reported more usability issues on two apps than five human experts, but the ground truth included the LLM's own findings.
-
Agentic Enterprise: AI-Centric User to User-Centric AI
Enterprise AI should be reorganized into a user-centric market of specialized agents guided by six tenets rather than built around general-purpose assistants.
-
Socratic-MCTS: Test-Time Visual Reasoning by Asking the Right Questions
Socratic-MCTS uses the model's own subquestions and answers in a Monte Carlo Tree Search to improve multimodal multiple-choice accuracy without fine-tuning.
-
A Comprehensive Survey on Trustworthiness in Reasoning with Large Language Models
A structured literature survey concluding that reasoning capabilities do not automatically make LLMs more trustworthy and can introduce new vulnerabilities in safety, robustness, and privacy.
-
The Science of Evaluating Foundation Models
A survey-and-checklist proposal that organizes LLM evaluation into an ABCD framework (Algorithm, Big Data, Computation, Domain Expertise) for context-aware, documented assessment.
Discussion (0). Continue with ORCID to comment.