LLaDA 1.5: Variance-Reduced Preference Optimization for Large Language Diffusion Models (2026.acl-long)
Copied to clipboard
Fengqi Zhu, Rongzhen Wang, Shen Nie, Xiaolu Zhang, Chunwei Wu, Jun Zhou, Yankai Lin, Ji-Rong Wen, Chongxuan Li
| Challenge: | Masked diffusion language models have achieved significant progress in language modeling . however, the systematic analysis and empirical validation of their alignment on general tasks remains underexplored. |
| Approach: | They propose a framework that analyzes the bias and variance of preference optimization loss and gradient based on Direct Preference Optimization. |
| Outcome: | The proposed model outperforms its SFT-only predecessor on general benchmarks . it consistently outperformed other strong language models and ARMs on general tasks . |
Similar Papers
SparsePO: Controlling Preference Alignment of LLMs via Sparse Token Masks (2025.findings-emnlp)
Copied to clipboard
| Challenge: | Current direct preference optimization algorithms focus on a strict set of tokens contributing signals of KL divergence and rewards to the loss function. |
| Approach: | They propose a method that automatically learns to weight the KL divergence and reward corresponding to each token during PO training. |
| Outcome: | The proposed method achieves +10% and +3% win-rate points in two PO scenarios. |
K-order Ranking Preference Optimization for Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing list-wise methods focus on optimizing list ranking consistency for LLMs to improve ranking abilities. |
| Approach: | They propose to extend the Plackett-Luce model to accommodate top-K ranking by extending the DPO’s Plact-Lucer model to dynamically determine appropriate K for different samples. |
| Outcome: | The proposed model can be extended to accommodate top-K ranking and improve training efficiency. |
A Deep Dive into the Trade-Offs of Parameter-Efficient Preference Alignment Techniques (2024.acl-long)
Copied to clipboard
| Challenge: | Large language models are pre-trained on trillions of tokens and instruction-tuned or aligned to specific preferences. |
| Approach: | They propose guidelines to help researchers perform more effective parameter-efficient LLM alignment. |
| Outcome: | The proposed methods outperform preference optimization and outperformed pre-trained models on three key axes. |
DiffPO: Diffusion-styled Preference Optimization for Inference Time Alignment of Large Language Models (2025.acl-long)
Copied to clipboard
Ruizhe Chen, Wenhao Chai, Zhifei Yang, Xiaotian Zhang, Ziyang Wang, Tony Quek, Joey Tianyi Zhou, Soujanya Poria, Zuozhu Liu
| Challenge: | Inference-time alignment approaches still face limitations due to policy-specific value functions and latency during the inference phase. |
| Approach: | They propose an efficient and policy-agnostic preference optimization method that avoids time latency associated with token generation. |
| Outcome: | The proposed method achieves a favorable trade-off between alignment quality and inference-time latency. |
Weights-Rotated Preference Optimization for Large Language Models (2025.emnlp-main)
Copied to clipboard
Chenxu Yang, Ruipeng Jia, Mingyu Zheng, Naibin Gu, Zheng Lin, Siyuan Chen, Weichong Yin, Hua Wu, Weiping Wang
| 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. |
InfoPO: On Mutual Information Maximization for Large Language Model Alignment (2025.naacl-long)
Copied to clipboard
Teng Xiao, Zhen Ge, Sujay Sanghavi, Tian Wang, Julian Katz-Samuels, Marc Versage, Qingjun Cui, Trishul Chilimbi
| 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. |
Insights into Alignment: Evaluating DPO and its Variants Across Multiple Tasks (2025.acl-srw)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) excel in math reasoning problemsolving, text generation, summarization, creative writing, among other tasks. |
| Approach: | They evaluate Direct Preference Optimization and its variants for aligning Large Language Models with human preferences. |
| Outcome: | The proposed alignment methods achieve near-optimal performance even with smaller subsets of training data. |
Can Machine Unlearning Reduce Social Bias in Language Models? (2024.emnlp-industry)
Copied to clipboard
Omkar Dige, Diljot Arneja, Tsz Fung Yau, Qixuan Zhang, Mohammad Bolandraftar, Xiaodan Zhu, Faiza Khattak
| Challenge: | Existing methods for mitigating bias in language models are expensive and time-consuming . comparative studies have not evaluated their respective advantages and disadvantages . |
| Approach: | They propose to use Partitioned Contrastive Gradient Unlearning and Negation via Task Vector to reduce social biases in open-source language models. |
| Outcome: | The proposed methods outperform PCGU and DPO in debiasing models . the proposed methods can be easily tuned to balance the trade-off between bias reduction and generation quality . |
DPO Kernels: A Semantically-Aware, Kernel-Enhanced, and Divergence-Rich Paradigm for Direct Preference Optimization (2025.findings-acl)
Copied to clipboard
Amitava Das, Suranjana Trivedy, Danush Khanna, Yaswanth Narsupalli, Basab Ghosh, Rajarshi Roy, Gurpreet Singh, Vinija Jain, Vasu Sharma, Aishwarya Naresh Reganti, Aman Chadha
| 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 . |
SGDPO: Self-Guided Direct Preference Optimization for Language Model Alignment (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for aligning Large Language Models with human values are limited and results of DPO are not resilient. |
| Approach: | They propose a self-guided direct preference optimization algorithm that incorporates a pilot term to steer the gradient flow during the optimization process. |
| Outcome: | The proposed method can generate human-preferred response up to 9.19% higher than previous methods. |