Image de couverture de livre

Pairwise or Pointwise? Evaluating Feedback Protocols for Bias in LLM-Based Evaluation

Fuente: arXiv

Enregistré dans:

Détails bibliographiques
Auteurs principaux:	Tripathi, Tuhina, Wadhwa, Manya, Durrett, Greg, Niekum, Scott
Format:	Preprint
Publié:	2025
Sujets:	Machine Learning
Accès en ligne:	Acceder al recurso
Tags:	Ajouter un tag Pas de tags, Soyez le premier à ajouter un tag!

Documents similaires

Adaptive Margin RLHF via Preference over Preferences
par: Chittepu, Yaswanth, et autres
Publié: (2025)

Learning to Refine with Fine-Grained Natural Language Feedback
par: Wadhwa, Manya, et autres
Publié: (2024)

Evaluation-Aware Reinforcement Learning
par: Deshmukh, Shripad Vilasrao, et autres
Publié: (2025)

Training ML Models with Predictable Failures
par: Schwarzer, Will, et autres
Publié: (2026)

Using Natural Language Explanations to Rescale Human Judgments
par: Wadhwa, Manya, et autres
Publié: (2023)

D2PO: Discriminator-Guided DPO with Response Evaluation Models
par: Singhal, Prasann, et autres
Publié: (2024)

Pointwise Metrics Mislead: An Evaluation Protocol for Multimodal Inverse Problems
par: Baattrup, Mads H., et autres
Publié: (2026)

EvalAgent: Discovering Implicit Evaluation Criteria from the Web
par: Wadhwa, Manya, et autres
Publié: (2025)

To CoT or not to CoT? Chain-of-thought helps mainly on math and symbolic reasoning
par: Sprague, Zayne, et autres
Publié: (2024)

SynthesizRR: Generating Diverse Datasets with Retrieval Augmentation
par: Divekar, Abhishek, et autres
Publié: (2024)

Efficient Pointwise-Pairwise Learning-to-Rank for News Recommendation
par: Kannen, Nithish, et autres
Publié: (2024)

From Distributional to Overton Pluralism: Investigating Large Language Model Alignment
par: Lake, Thom, et autres
Publié: (2024)

Bayesian Robust Optimization for Imitation Learning
par: Brown, Daniel S., et autres
Publié: (2020)

Coeditor: Leveraging Contextual Changes for Multi-round Code Auto-editing
par: Wei, Jiayi, et autres
Publié: (2023)

Supervised Reward Inference
par: Schwarzer, Will, et autres
Publié: (2025)

QUDsim: Quantifying Discourse Similarities in LLM-Generated Text
par: Namuduri, Ramya, et autres
Publié: (2025)

Explaining Length Bias in LLM-Based Preference Evaluations
par: Hu, Zhengyu, et autres
Publié: (2024)

Reinforcement Learning from Human Feedback with High-Confidence Safety Constraints
par: Chittepu, Yaswanth, et autres
Publié: (2025)

PropMEND: Hypernetworks for Knowledge Propagation in LLMs
par: Liu, Zeyu Leo, et autres
Publié: (2025)

A Long Way to Go: Investigating Length Correlations in RLHF
par: Singhal, Prasann, et autres
Publié: (2023)

On the Benefits of Inducing Local Lipschitzness for Robust Generative Adversarial Imitation Learning
par: Memarian, Farzan, et autres
Publié: (2021)

Contrastive Preference Learning: Learning from Human Feedback without RL
par: Hejna, Joey, et autres
Publié: (2023)

SkillFactory: Self-Distillation For Learning Cognitive Behaviors
par: Sprague, Zayne, et autres
Publié: (2025)

CREATE: Testing LLMs for Associative Creativity
par: Wadhwa, Manya, et autres
Publié: (2026)

Contrastive Learning to Improve Retrieval for Real-world Fact Checking
par: Sriram, Aniruddh, et autres
Publié: (2024)

PaTaRM: Bridging Pairwise and Pointwise Signals via Preference-Aware Task-Adaptive Reward Modeling
par: Jian, Ai, et autres
Publié: (2025)

Safe RLHF Beyond Expectation: Stochastic Dominance for Universal Spectral Risk Control
par: Chittepu, Yaswanth, et autres
Publié: (2026)

Dual RL: Unification and New Methods for Reinforcement and Imitation Learning
par: Sikchi, Harshit, et autres
Publié: (2023)

A Dual Approach to Imitation Learning from Observations with Offline Datasets
par: Sikchi, Harshit, et autres
Publié: (2024)

VeriSoftBench: Repository-Scale Formal Verification Benchmarks for Lean
par: Xin, Yutong, et autres
Publié: (2026)

Accelerating Unbiased LLM Evaluation via Synthetic Feedback
par: Zhou, Zhaoyi, et autres
Publié: (2025)

Pareto-Optimal Learning from Preferences with Hidden Context
par: Bahlous-Boldi, Ryan, et autres
Publié: (2024)

Feedback Descent: Open-Ended Text Optimization via Pairwise Comparison
par: Lee, Yoonho, et autres
Publié: (2025)

ProofWala: A Framework for Multilingual Proof Data Synthesis and Theorem-Proving
par: Thakur, Amitayush, et autres
Publié: (2025)

Fundamental Limitations in Pointwise Defences of LLM Finetuning APIs
par: Davies, Xander, et autres
Publié: (2025)

An Optimal Discriminator Weighted Imitation Perspective for Reinforcement Learning
par: Xu, Haoran, et autres
Publié: (2025)

Enhancing LLM Metacognition via Cognitive Pairwise Training
par: Li, Weitao, et autres
Publié: (2026)

ExPairT-LLM: Exact Learning for LLM Code Selection by Pairwise Queries
par: Yuviler, Tom, et autres
Publié: (2025)

A Descriptive and Normative Theory of Human Beliefs in RLHF
par: Dandekar, Sylee, et autres
Publié: (2025)

Direct-Scoring NLG Evaluators Can Use Pairwise Comparisons Too
par: Lawrence, Logan, et autres
Publié: (2025)