REVIEW 4 cited by
LLMs Are Biased Towards Output Formats! Systematically Evaluating and Mitigating Output Format Bias of LLMs
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
abstract
We present the first systematic evaluation examining format bias in performance of large language models (LLMs). Our approach distinguishes between two categories of an evaluation metric under format constraints to reliably and accurately assess performance: one measures performance when format constraints are adhered to, while the other evaluates performance regardless of constraint adherence. We then define a metric for measuring the format bias of LLMs and establish effective strategies to reduce it. Subsequently, we present our empirical format bias evaluation spanning four commonly used categories -- multiple-choice question-answer, wrapping, list, and mapping -- covering 15 widely-used formats. Our evaluation on eight generation tasks uncovers significant format bias across state-of-the-art LLMs. We further discover that improving the format-instruction following capabilities of LLMs across formats potentially reduces format bias. Based on our evaluation findings, we study prompting and fine-tuning with synthesized format data techniques to mitigate format bias. Our methods successfully reduce the variance in ChatGPT's performance among wrapping formats from 235.33 to 0.71 (%$^2$).
Forward citations
Cited by 4 Pith papers
-
Test-Time Scaling via Error Localization
TTEL uses feedback-induced token probability drops to localize the first error in a failed reasoning trace and branch a new generation from that prefix, improving pass@k per token on coding and math benchmarks.
-
TAPR: Enhancing LLM Performance with a Task-Aware Prompt Rewriter
A small LLM trained with GRPO and LLM-judge rewards rewrites simple prompts into more effective ones, improving question-answering and arithmetic accuracy over base prompts while giving mixed, often negligible gains o...
-
IntPhys 2: Benchmarking Intuitive Physics Understanding In Complex Synthetic Environments
IntPhys 2 evaluates models on permanence, immutability, continuity, and solidity in synthetic videos, and finds most models near chance while humans near perfect.
-
Towards Reliable Generative AI-Driven Scaffolding: Reducing Hallucinations and Enhancing Quality in Self-Regulated Learning Support
A multi-agent reliability check and an LLM-as-a-judge quality check both reduce hallucinations in AI-generated study scaffolds, with the multi-agent check matching human expert judgments almost perfectly.
Discussion (0). Continue with ORCID to comment.