Pith. sign in

REVIEW 13 cited by

A Comprehensive Survey of Reward Models: Taxonomy, Applications, Challenges, and Future

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

arxiv 2504.12328 v1 pith:A5IOGR2W submitted 2025-04-12 cs.CL cs.AI

A Comprehensive Survey of Reward Models: Taxonomy, Applications, Challenges, and Future

classification cs.CL cs.AI
keywords comprehensiverewardapplicationschallengesfuturegithubmodelspotential
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
read the original abstract

Reward Model (RM) has demonstrated impressive potential for enhancing Large Language Models (LLM), as RM can serve as a proxy for human preferences, providing signals to guide LLMs' behavior in various tasks. In this paper, we provide a comprehensive overview of relevant research, exploring RMs from the perspectives of preference collection, reward modeling, and usage. Next, we introduce the applications of RMs and discuss the benchmarks for evaluation. Furthermore, we conduct an in-depth analysis of the challenges existing in the field and dive into the potential research directions. This paper is dedicated to providing beginners with a comprehensive introduction to RMs and facilitating future studies. The resources are publicly available at github\footnote{https://github.com/JLZhong23/awesome-reward-models}.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 13 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Agents Don't Just Agree, They Remember: Benchmarking Persistent Sycophancy in Stateful Personal Agents

    cs.AI 2026-07 conditional novelty 8.0

    Persistent sycophancy is measured by the new PASB benchmark: when agents write user claims to memory, later neutral answers become sycophantic at 71.9% vs 45.0% without commit.

  2. Agents Don't Just Agree, They Remember: Benchmarking Persistent Sycophancy in Stateful Personal Agents

    cs.AI 2026-07 conditional novelty 7.5

    Across twelve models on two real agent stacks, committing a user-centric claim into durable state raises downstream sycophantic failure from 45.0% to 71.9%, driven by status promotion, attribution removal, and scope b...

  3. GEAR: Granularity-Adaptive Advantage Reweighting for LLM Agents via Self-Distillation

    cs.LG 2026-05 unverdicted novelty 7.0

    GEAR reshapes GRPO trajectory advantages using divergence signals from a ground-truth-conditioned teacher to create adaptive token- and segment-level credit regions.

  4. StoryAlign: Evaluating and Training Reward Models for Story Generation

    cs.CL 2026-05 unverdicted novelty 7.0

    StoryReward, trained on a new 100k story preference dataset, sets state-of-the-art performance on the introduced StoryRMB benchmark for aligning LLM stories with human preferences.

  5. Agents Don't Just Agree, They Remember: Benchmarking Persistent Sycophancy in Stateful Personal Agents

    cs.AI 2026-07 conditional novelty 6.5

    Stateful personal agents convert conversational sycophancy into durable memory commits that raise later failure rates by about 27 points once claims are written to state.

  6. Scalable Token-Level Hallucination Detection in Large Language Models

    cs.CL 2026-05 unverdicted novelty 6.0

    TokenHD uses a scalable data synthesis engine and importance-weighted training to create token-level hallucination detectors that work on free-form text and scale from 0.6B to 8B parameters, outperforming larger reaso...

  7. GEAR: Granularity-Adaptive Advantage Reweighting for LLM Agents via Self-Distillation

    cs.LG 2026-05 unverdicted novelty 6.0

    GEAR adaptively reweights GRPO advantages in LLM RL by using divergence spikes from self-distillation to define semantic segments and modulate local credit.

  8. DeltaRubric: Generative Multimodal Reward Modeling via Joint Planning and Verification

    cs.CL 2026-05 unverdicted novelty 6.0

    DeltaRubric decomposes multimodal preference evaluation into self-generated planning and verification steps within a single model, producing large accuracy improvements on VL-RewardBench via multi-role reinforcement learning.

  9. Toward Robust LLM-Based Judges: Taxonomic Bias Evaluation and Debiasing Optimization

    cs.CL 2026-03 conditional novelty 6.0

    JudgeBiasBench measures 12 LLM-judge bias types under a 4-dimension taxonomy, and bias-aware GRPO/InfoNCE training cuts BSR substantially while preserving general evaluation performance.

  10. The Landscape of Agentic Reinforcement Learning for LLMs: A Survey

    cs.AI 2025-09 accept novelty 6.0

    Survey that defines agentic RL for LLMs via POMDPs, introduces a taxonomy of planning/tool-use/memory/reasoning capabilities and domains, and compiles open environments from over 500 papers.

  11. SocialCoach: Personalized Social Skill Learning with RL-based Agentic Tutoring and Practice

    cs.HC 2026-06 unverdicted novelty 4.0

    SocialCoach combines multi-agent corpus construction, RL-optimized adaptive scheduling in simulation, and immersive LLM tutoring to deliver personalized social-skill training, reporting gains in simulated pathway qual...

  12. Reward Modeling for Reinforcement Learning-Based LLM Reasoning: Design, Challenges, and Evaluation

    cs.LG 2026-02 conditional novelty 4.0

    A taxonomy-driven survey arguing that reward design is the central mechanism shaping reliable LLM reasoning, with maps of reward paradigms, reward-hacking failure modes, and benchmark pitfalls.

  13. Reinforcement Learning Meets Large Language Models: A Survey of Advancements and Applications Across the LLM Lifecycle

    cs.CL 2025-09 conditional novelty 3.0

    A survey that maps reinforcement learning methods, datasets, benchmarks, and open-source tools across the full training lifecycle of large language models, focusing on verifiable-reward reasoning.