REVIEW 7 cited by
Aligning with Human Judgement: The Role of Pairwise Preference in Large Language Model Evaluators
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
Large Language Models (LLMs) have demonstrated promising capabilities as automatic evaluators in assessing the quality of generated natural language. However, LLMs still exhibit biases in evaluation and often struggle to generate coherent evaluations that align with human assessments. In this work, we first conduct a systematic study of the misalignment between LLM evaluators and human evaluation, revealing that existing calibration methods aimed at mitigating biases of LLMs are insufficient for effectively aligning LLM evaluators. Inspired by the use of preference data in RLHF, we formulate the evaluation as a ranking problem and introduce Pairwise-preference Search (PAIRS), an uncertainty-guided search-based rank aggregation method that employs LLMs to conduct pairwise comparisons locally and efficiently ranks candidate texts globally. PAIRS achieves state-of-the-art performance on representative evaluation tasks in long-form generations and demonstrates significant improvements over direct scoring. Furthermore, we provide insights into the role of pairwise preference in quantifying the transitivity of LLMs and demonstrate how PAIRS benefits from calibration using debiased pairwise evaluations.
Forward citations
Cited by 7 Pith papers
-
AI Alignment at Your Discretion
The paper formalizes alignment discretion and shows empirically that annotators and models exercise substantial, often arbitrary, and mutually divergent discretion when applying alignment principles.
-
Fragile Preferences: A Deep Dive Into Order Effects in Large Language Models
LLMs show a quality-dependent position bias, favoring the first option for high-quality choices and later options for low-quality ones, and higher-temperature sampling can reveal the underlying preference.
-
Deep Researcher with Test-Time Diffusion
A test-time 'denoising' loop that repeatedly revises a draft report using fresh web retrieval, plus a component-wise self-evolution step, beats existing deep research agents on several benchmarks.
-
Dr. GPT Will See You Now, but Should It? Exploring the Benefits and Harms of Large Language Models in Medical Diagnosis using Crowdsourced Clinical Cases
In a physician-rated crowdsourced study, 76% of LLM responses to everyday health queries were valid, with GPT-4o highest (85%) and Llama3-8b lowest (50%); RAG did not consistently improve responses.
-
(Towards) Scalable Reliable Automated Evaluation with Large Language Models
Multi-LLM pairwise Elo ranking with adjustable consensus thresholds produces rankings of competency profiles that average Spearman ρ≈0.83 with expert judgments.
-
Harnessing Meta-Learning for Controllable Full-Frame Video Stabilization
Meta-learned test-time adaptation, focused on the shakiest parts of a video, improves the stability and quality of full-frame neural video stabilizers.
-
OpenReview Should be Protected and Leveraged as a Community Asset for Research in the Era of Large Language Models
The paper advocates protecting and leveraging OpenReview's peer review corpus as a community asset for LLM-based review assistance, benchmarks, and alignment.
Discussion (0). Continue with ORCID to comment.