Challenge: Existing methods to align large language models with high reward hacking are limited by the complexity of the parameter space and the complexity.
Approach: They propose a weights-rotated preference optimization algorithm that constrains the output layer logits with the KL divergence inherited from DPO and fine-tunes the intermediate hidden states.
Outcome: The proposed algorithm achieves a 3.27-point improvement on AlpacaEval 2 and surpasses the best baseline by 6.2 to 7.5 points on MT-Bench with merely 0.015% of the trainable parameters.

Similar Papers

Eliminating Biased Length Reliance of Direct Preference Optimization via Down-Sampled KL Divergence (2024.emnlp-main)

Copied to clipboard

Challenge: Existing studies attributed verbosity to biased labels, but new research shows that DPO can be effective in mitigating verboses.
Approach: They propose to use a method to reduce the amount of verbosity in LLMs by using a downsampling approach.
Outcome: The proposed approach overcomes the problem of verbosity by reducing the length reliance of the proposed algorithm.
DPO Kernels: A Semantically-Aware, Kernel-Enhanced, and Divergence-Rich Paradigm for Direct Preference Optimization (2025.findings-acl)

Copied to clipboard

Challenge: Direct Preference Optimization (DPO) is a cornerstone for preference alignment but is constrained by fixed divergence measures and limited feature transformations.
Approach: They propose a new enhancement of Direct Preference Optimization that integrates kernel methods to overcome these challenges.
Outcome: The proposed model improves divergence measures and features by using kernels . the proposed model achieves state-of-the-art generalization in factuality, safety, reasoning, and instruction following .
Efficient Safety Alignment of Large Language Models via Preference Re-ranking and Representation-based Reward Modeling (2025.acl-long)

Copied to clipboard

Challenge: Existing safety alignment methods for Large Language Models (LLMs) face the distribution shift issue, which requires significant computational resources.
Approach: They propose a framework that leverages the model’s intrinsic safety judgment capability to extract reward signals, which are then used to calculate label confidence for preference reordering.
Outcome: The proposed framework improves safety performance while avoiding 300x computational overheads.
Beyond One-Preference-Fits-All Alignment: Multi-Objective Direct Preference Optimization (2024.findings-acl)

Copied to clipboard

Challenge: Recent approaches to language model alignment assume homogeneous human preferences, but actual human preferences vary widely and are hard to satisfy with a single language model.
Approach: They propose an RL-free extension of Direct Preference Optimization (DPO) that folds language modeling directly into reward modeling and trains language models as collective reward models that combine all objectives with specific weights.
Outcome: The proposed method matches or outperforms existing methods in safety alignment and long-form question answering.
RS-DPO: A Hybrid Rejection Sampling and Direct Preference Optimization Method for Alignment of Large Language Models (2024.findings-naacl)

Copied to clipboard

Challenge: Reinforcement learning with human feedback (RLHF) is widely employed to align large language models with user intent.
Approach: They propose to combine rejection sampling and direct preference optimization to improve alignment with user intent by identifying pairs of contrastive samples from human annotator and alternative LLMs.
Outcome: The proposed method outperforms existing methods including RS, PPO, and DPO in a limited resource environment.
CRPO: Confidence-Reward Driven Preference Optimization for Machine Translation (2025.findings-acl)

Copied to clipboard

Challenge: Large language models (LLMs) have shown great potential in natural language processing tasks, but their application to machine translation remains challenging due to pretraining on predominantly English-centric datasets.
Approach: They propose a method that combines reward scores with model confidence to improve model selection for fine-tuning.
Outcome: The proposed method outperforms existing methods in translation accuracy and data efficiency.
Reinforcement Learning for Large Language Models via Group Preference Reward Shaping (2025.emnlp-main)

Copied to clipboard

Challenge: Existing methods for fine-tuning Large Language Models (LLMs) are expensive and sensitive to reward model quality.
Approach: They propose a method that leverages preference-based comparisons rather than precise numerical rewards.
Outcome: Experiments show that GPRS outperforms critic-model-free RL algorithms on RLHF and reasoning tasks.
InfoPO: On Mutual Information Maximization for Large Language Model Alignment (2025.naacl-long)

Copied to clipboard

Challenge: Recent studies have shown that direct preference optimization and its variants can be useful for fine-tuning large language models with human preferences data.
Approach: They propose a preference fine-tuning algorithm that effectively and efficiently aligns large language models using preference data.
Outcome: Extensive experiments show that the proposed algorithm outperforms established baselines on reasoning tasks.
Topology-Enhanced Alignment for Large Language Models: Trajectory Topology Loss and Topological Preference Optimization (2026.findings-acl)

Copied to clipboard

Challenge: Existing algorithms for supervised fine-tuning and reinforcement learning from human feedback (RLHF) do not constrain how hidden states move from a user prompt to an answer.
Approach: They propose a topology-enhanced alignment framework that regularizes these trajectories using 0-dimensional persistent homology.
Outcome: The proposed framework regularizes semantic trajectory in hidden space using 0-dimensional persistent homology.
Beyond Online Sampling: Bridging Offline-to-Online Alignment via Dynamic Data Transformation for LLMs (2025.emnlp-main)

Copied to clipboard

Challenge: Direct Preference Optimization (DPO) eliminates complex reward modeling in aligning large language models with human preferences, but its online variant faces significant efficiency bottlenecks due to costly real-time preference sampling and the reward model annotation.
Approach: They propose a framework that transforms static datasets into dynamically adaptive equivalents without the need for an explicit reward model.
Outcome: The proposed approach matches or exceeds the performance of a fully online DPO.

What is GenGO?

GenGO is an NLP powered publication search system. It currenctly indexes 30k+ papers from ACL Anthology, and implements multi-aspect summarization, semantic search, and more!

Information

About
Limitations