Pith. sign in

REVIEW 26 cited by

Secrets of RLHF in Large Language Models Part II: Reward Modeling

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 2401.06080 v2 pith:L4YR4RGU submitted 2024-01-11 cs.AI

Secrets of RLHF in Large Language Models Part II: Reward Modeling

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

Reinforcement Learning from Human Feedback (RLHF) has become a crucial technology for aligning language models with human values and intentions, enabling models to produce more helpful and harmless responses. Reward models are trained as proxies for human preferences to drive reinforcement learning optimization. While reward models are often considered central to achieving high performance, they face the following challenges in practical applications: (1) Incorrect and ambiguous preference pairs in the dataset may hinder the reward model from accurately capturing human intent. (2) Reward models trained on data from a specific distribution often struggle to generalize to examples outside that distribution and are not suitable for iterative RLHF training. In this report, we attempt to address these two issues. (1) From a data perspective, we propose a method to measure the strength of preferences within the data, based on a voting mechanism of multiple reward models. Experimental results confirm that data with varying preference strengths have different impacts on reward model performance. We introduce a series of novel methods to mitigate the influence of incorrect and ambiguous preferences in the dataset and fully leverage high-quality preference data. (2) From an algorithmic standpoint, we introduce contrastive learning to enhance the ability of reward models to distinguish between chosen and rejected responses, thereby improving model generalization. Furthermore, we employ meta-learning to enable the reward model to maintain the ability to differentiate subtle differences in out-of-distribution samples, and this approach can be utilized for iterative RLHF optimization.

discussion (0)

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

Forward citations

Cited by 26 Pith papers

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

  1. ORPO: Monolithic Preference Optimization without Reference Model

    cs.CL 2024-03 conditional novelty 8.0

    ORPO performs preference alignment during supervised fine-tuning via a monolithic odds ratio penalty, allowing 7B models to outperform larger state-of-the-art models on alignment benchmarks.

  2. Towards Order Fairness: Mitigating LLMs Order Sensitivity through Dual Group Advantage Optimization

    cs.LG 2026-05 unverdicted novelty 7.0

    DGAO uses reinforcement learning to optimize LLMs for both accuracy and order stability by balancing intra-group accuracy advantages and inter-group stability advantages.

  3. Reinforcement Learning via Value Gradient Flow

    cs.LG 2026-04 unverdicted novelty 7.0

    VGF solves behavior-regularized RL by transporting particles from a reference distribution to the value-induced optimal policy via discrete value-guided gradient flow.

  4. Low-rank Optimization Trajectories Modeling for LLM RLVR Acceleration

    cs.LG 2026-04 unverdicted novelty 7.0

    NExt accelerates RLVR training for LLMs by nonlinearly extrapolating low-rank parameter trajectories extracted from LoRA runs.

  5. Incentivizing High-Quality Human Annotations with Golden Questions

    cs.GT 2025-05 unverdicted novelty 7.0

    The paper derives a Θ(1/√(n log n)) hypothesis testing rate under strategic annotator behavior and shows that high-certainty, format-similar golden questions better reveal annotation quality than standard checks.

  6. Metadata-Free Meta-Reweighted Direct Preference Optimization under Noisy Preference Labels

    cs.LG 2026-07 conditional novelty 6.0

    Bilevel reweighting recovers clean DPO optima under idealized noise, and PACMR-DPO learns sample weights without clean metadata using prompt-augmentation consistency plus LoRA finite differences.

  7. Metadata-Free Meta-Reweighted Direct Preference Optimization under Noisy Preference Labels

    cs.LG 2026-07 conditional novelty 6.0

    Under 20–40% random preference-label flips, PACMR-DPO—a VNet-reweighted DPO with a prompt-augmentation-consistency meta-objective and central-difference LoRA meta-gradients—outperforms cDPO, IPO, rDPO, and Dr.DPO in j...

  8. Understanding helpfulness and harmless tension in reward models

    cs.LG 2026-06 unverdicted novelty 6.0

    Mixed-objective reward models underperform single-objective ones because shared neurons support one objective while negatively affecting the other, creating alignment tension.

  9. Representation-Aware Advantage Estimation: Your Reward Model Provides More Than A Scalar Output

    cs.LG 2026-06 unverdicted novelty 6.0

    GraphAE builds graphs from RM hidden-state similarities among sampled responses and propagates advantages to improve RLHF sample efficiency.

  10. Boosting Self-Consistency with Ranking

    cs.CL 2026-06 unverdicted novelty 6.0

    RISC reformulates self-consistency answer selection as a ranking task solved by a lightweight LambdaRank model with five hand-designed features, yielding better accuracy-efficiency trade-offs than majority voting on Q...

  11. Gradient-Gated DPO: Stabilizing Preference Optimization in Language Models

    cs.LG 2026-05 conditional novelty 6.0

    Gate-DPO attenuates gradients on low-probability rejected responses to reduce probability collapse and improve chosen-response likelihood during preference optimization.

  12. Adaptive Margin RLHF via Preference over Preferences

    cs.LG 2025-09 conditional novelty 6.0

    Adaptive margins for DPO inferred from preference-over-preference comparisons improve alignment quality, with random sampling of comparisons working best overall.

  13. How Humans Help LLMs: Assessing and Incentivizing Human Preference Annotators

    cs.LG 2025-02 unverdicted novelty 6.0

    Develops self-consistency monitoring for preference annotators and derives sample-complexity bounds showing linear contracts achieve near-ideal performance faster than binary ones under continuous actions.

  14. Enhancing the Reasoning Ability of Multimodal Large Language Models via Mixed Preference Optimization

    cs.CL 2024-11 conditional novelty 6.0

    Mixed Preference Optimization with the MMPR dataset boosts multimodal CoT reasoning, lifting InternVL2-8B to 67.0 accuracy on MathVista (+8.7 points) and matching the 76B model.

  15. Meta-Learned Reward Shaping for Reinforcement Learning from Human Feedback

    cs.LG 2026-07 reject novelty 5.0

    Training a small multi-task reward-shaping network and adding it to the RLHF reward is claimed to improve LLaMA-3-8B alignment across four benchmarks, but the supporting theory is not established.

  16. Modularized Reinforcement Learning on LLMs: From MDP Creation to Exploration and Learning

    cs.LG 2026-06 unverdicted novelty 5.0

    Survey mapping RL techniques onto LLM training and highlighting gaps in value-based, off-policy, and bootstrapping methods.

  17. DynaCF: Mitigating Shortcut Learning in Reward Models via Dynamic Counterfactual Sensitivity

    cs.LG 2026-06 unverdicted novelty 5.0

    DynaCF dynamically downweights shortcut-sensitive samples in reward model training by tracking margin shifts under online counterfactual perturbations within the Bradley-Terry loss.

  18. Boosting Reinforcement Learning with Verifiable Rewards via Randomly Selected Few-Shot Guidance

    cs.LG 2026-05 unverdicted novelty 5.0

    FEST improves RLVR sample efficiency on math and coding benchmarks by combining supervised signals, on-policy signals, and decaying weights on just 128 randomly chosen demonstrations, matching full-dataset baselines.

  19. Personalized Alignment Revisited: The Necessity and Sufficiency of User Diversity

    cs.LG 2026-05 unverdicted novelty 5.0

    A user-diversity condition is necessary and sufficient for personalized alignment to achieve O(1) online regret and log(1/epsilon) offline sample complexity.

  20. DT2IT-MRM: Debiased Preference Construction and Iterative Training for Multimodal Reward Modeling

    cs.AI 2026-04 unverdicted novelty 5.0

    DT2IT-MRM proposes a debiased preference construction pipeline, T2I data reformulation, and iterative training to curate multimodal preference data, achieving SOTA on VL-RewardBench, Multimodal RewardBench, and MM-RLH...

  21. Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges

    cs.LG 2026-04 unverdicted novelty 5.0

    The paper introduces the Proxy Compression Hypothesis as a unifying framework explaining reward hacking in RLHF as an emergent result of compressing high-dimensional human objectives into proxy reward signals under op...

  22. Users as Annotators: LLM Preference Learning from Comparison Mode

    cs.CL 2025-10 unverdicted novelty 5.0

    Introduces a latent user quality model and EM algorithm to infer and filter noisy user-provided pairwise preferences for improved LLM alignment.

  23. BV-Blend: Uncertainty-Weighted Historical Baselines for Stable Critic-Free RL with Verifiable Rewards

    cs.AI 2026-06 unverdicted novelty 4.0

    BV-Blend blends prompt-local and semantic-cluster historical reward statistics via SEM-derived weights to stabilize critic-free RL advantage estimation.

  24. Proxy Reward Internalization and Mechanistic Exploitation: A Learned Precursor to Reward Hacking and Its Generalization

    cs.AI 2026-06 unverdicted novelty 4.0

    Proxy RL produces a staged proxy-internalization capability that emerges before and predicts reward hacking in coding environments.

  25. Seed1.5-VL Technical Report

    cs.CV 2025-05 unverdicted novelty 4.0

    Seed1.5-VL is a compact multimodal model that sets new records on dozens of vision-language benchmarks and outperforms prior systems on agent-style tasks.

  26. Qwen2.5 Technical Report

    cs.CL 2024-12 unverdicted novelty 3.0

    Qwen2.5 LLMs scale pre-training data to 18 trillion tokens and apply multistage reinforcement learning, achieving competitive performance on benchmarks with models up to 5 times larger.