Direct Preference Optimization

Also known as DPO

Follow this topic Ask about this topic

Direct Preference Optimization (DPO) studies how to align a model with preferences by training it to favor chosen responses over rejected ones, usually relative to a reference model. Research extends the basic pairwise objective with adaptive margins and weighting, token- or segment-level credit, methods for noisy or conflicting labels, and online or curriculum-based data selection. DPO is used for language-model alignment in helpfulness, safety, and reasoning, and is increasingly adapted to multimodal models and diffusion-based image, video, and audio generation. A central tension is improving preference alignment and robustness without sacrificing general capabilities, diversity, or reliable performance when preferences and inputs shift.

Latest papers

162 papers24 papers in the last four weeks, up 60% on the four weeks before. 0.1% of all new papers.

Jul 3, 2026

Unbiased Alignment for Large Language Models with Noisy Preferences

The alignment of large language models with human preferences is commonly achieved through Reinforcement Learning from Human Feedback or Direct Preference Optimization. However, these methods are vulnerable to the significant noise prevalent in real-world preference datasets. To address this critical issue, we present a theoretical framework for unbiased alignment, introducing the Unbiased Reward Model (URM) loss and the Unbiased Direct Preference Optimization (UDPO) loss. By mathematically correcting the distortion induced by preference noise, our novel objectives enable unbiased model training directly from noisy datasets, without requiring clean ground-truth supervision. We provide rigorous theoretical analyses demonstrating that our methods are noise-tolerant, parameter downward compatible, and classification-calibrated. Comprehensive experiments across diverse datasets demonstrate that our approaches outperform state-of-the-art baselines. Code available at: https://github.com/cswjl/unbiased-alignment.

Jialiang Wang, Xianming Liu, Xiong Zhou +2

1Harbin Institute of Technology · 2City University of Hong Kong.

Pairwise Preference Learning LLM Alignment Direct Preference Optimization Learning with Noisy Labels Preference Learning

Jul 1, 2026

Learning to Watch: Active Video Anomaly Understanding via Interleaved Policy Optimization

Video anomaly understanding (VAU) relies on sparse, context-dependent cues. However, existing passive paradigms suffer from observational aliasing, where static sampling fails to disambiguate semantically distinct events. To overcome this, we propose

Anom\text{-}π

, a closed-loop framework that reconceptualizes video understanding as an active sequential decision-making process within a dynamic environment. Inspired by human video-reviewing behavior, this framework unifies internal cognitive reasoning and strategic evidence acquisition into an interleaved policy, utilizing temporal atomic operators such as local backtracking, temporal expansion, and fine-grained sampling to endow the model with perceptual proactivity. To learn such complex interaction strategies under video-level weak supervision, we design Interactive Direct Preference Optimization (iDPO) to achieve trajectory-level policy alignment, guided by an Active Evidence Inquiry (AEI) utility that balances task success, informative evidence acquisition, and interaction cost. This approach enables the agent to learn to actively disambiguate hypotheses while suppressing redundant exploration. Extensive experiments demonstrate that our framework, with only 2B parameters, achieves highly competitive performance, significantly outperforming state-of-the-art large-scale VAU models in complex scenarios.

Mengjingcheng Mo, Jiaxu Leng, Xinbo Gao

School of Computer Science and Technology, Chongqing University of Posts and Telecommunications, Chongqing, China · Chongqing College of Artificial Intelligence, Chongqing, China

Video Understanding Direct Preference Optimization Video Anomaly Detection Weakly Supervised Video Anomaly Detection

Jun 30, 2026

ADAPT: Attention Dynamics Alignment with Preference Tuning for Faithful MLLMs

Multimodal Large Language Models (MLLMs) are critically hampered by hallucination, generating content inconsistent with the provided image. In this paper, we identify an internal signature of hallucination: progressive degradation of text-to-image cross-attention during generation, leading to specific failure patterns like unfocused or biased attention. Existing mitigation strategies are largely outcome-driven and do not explicitly target this failure mode. To address this problem, we propose ADAPT (Attention Dynamics Alignment with Preference Tuning), an attention-based framework that intervenes directly on text-to-image cross-attention dynamics. We propose ADAPT with three key contributions: a cross-attention visual anchor refined from early decoding to provide stable spatial grounding, an attention-supervised inference mechanism that detects and corrects attention drift online, and a Visual Attention Guidance DPO that aligns preferences toward visually grounded responses. Experiments show that each component of ADAPT contributes to hallucination reduction, and the full framework achieves new best results across multiple hallucination benchmarks, reducing hallucination rates by 40%-60% across mainstream backbones while preserving general multimodal capabilities. Our work provides an attention-based perspective on mitigating hallucinations by exploring the model's internal text-to-image cross-attention behaviors. Code is available at https://github.com/yao-ustc/ADAPT

Zhiyuan Yao, Zheren Fu, Zhixiao Zheng +3

University of Science and Technology of China, Hefei, China · State Key Laboratory of Communication Content Cognition, People’s Daily Online, China · Huawei Technologies Ltd., China

Cross-Attention Direct Preference Optimization Object Hallucination in VLMs Cross-Modal Attention VLM Hallucination Mitigation

Jun 28, 2026

Reliability-Prioritized Fine-Grained Generation in Multimodal Large

Multimodal large language models (MLLMs) are increasingly expected to generate fine-grained descriptions of visual content. However, we observe and theoretically show that generating fine-grained responses poses a reliability challenge, i.e., fine-grained generation is more error-prone than coarse-grained generation. This phenomenon suggests that models should generate the finest description that remains reliable rather than simply produce more specific outputs. To investigate this problem, we develop GranFact, a granularity-aware benchmark consisting of expert-verified multi-object images with coarse-to-fine category annotations. Then, we design a hierarchy-aware evaluation algorithm, which assesses both whether model predictions are visually correct and how specific the correct predictions are. We also propose a reliability-prioritized preference optimization method based on Direct Preference Optimization, which penalizes unreliable fine-grained claims while rewarding reliable specificity. Experiments on GranFact show that our method improves fine-grained generation while preserving reliability. Code and data are available here.

Xiaomeng Fan, Wei Wu, Yuwei Wu +9

Beijing Key Laboratory of Intelligent Information Technology, School of Computer Science & Technology, Beijing Institute of Technology · Guangdong Laboratory of Machine Perception and Intelligent Computing, Shenzhen MSU-BIT University · Department of Electrical and Computer System Engineering, Monash University

Direct Preference Optimization Multimodal Large Language Models LLM Reliability Multimodal Model Evaluation Fine-Grained Visual Perception Multimodal Understanding

Jun 27, 2026

Reward-Free Code Alignment from Pretrained or Fine-Tuned LLM: Unpacking the Trade-offs for Code Generation

Large Language Model (LLM) alignment trains an LLM using preference data to produce outputs that better meet established quality standards. While LLM alignment techniques are studied for non-coding tasks, we know little about their usefulness for coding tasks. It is unclear whether LLM code alignment could support both functional requirements (producing executable, correct code) and non-functional requirements (code readability, style, maintainability). It is also unknown whether alignment for a code LLM should begin with base pretrained version or the finetuned (i.e., instruction-tuned) version of the LLM. In this paper, we offer insights on the above two research questions by conducting an empirical study. We studied five state-of-the-art (SOTA) LLMs using two widely used LLM alignment techniques: Direct Preference Optimization (DPO) and BoNBoN. For each training record, we created a preference pair as accepted and rejected instances by using the SelfCodeAlign pipeline. DPO and BoNBoN are reward-free models, i.e., they eliminate the need for multiple reward scores for output preferences. We tuned each LLM using the two alignment techniques in two settings: pretrained and finetuned versions of an LLM. We evaluated functional requirements using four SOTA benchmarks (HumanEval+, MBPP+, EvalPerf, EvoEval) and non-functional requirements using the CODAL benchmark, which evaluates code quality across five dimensions derived from software engineering practices. We find that pretrained-to-aligned pathways achieve larger improvements in the aligned variant over its pretrained variant. But the pretrained variant is generally less accurate than its finetuned variant. However, finetuned- to-aligned offers smaller performance improvements or, in some cases, degradation in the aligned variant than its finetuned variant.

Sanjeepan Sivapiran, Gias Uddin

York University, Canada

Direct Preference Optimization Preference Alignment

Jun 21, 2026

Weight-Space Geometry of Offline Reasoning Training

Offline reinforcement-learning losses (RFT, RIFT, DFT, Offline GRPO, DPO) are widely used to distill reasoning from large teachers into smaller students, and are typically compared on downstream accuracy alone. We ask whether they are mechanistically distinct or converge to a similar weight update. Training six methods (SFT, RFT, DFT, RIFT, Offline GRPO, DPO) on identical math rollouts from a single base model (Qwen3-4B) with attention-only LoRA, we analyze the resulting deltas via cosine similarity, principal-angle subspace analysis, linear mode connectivity, and CKA. We observe: (i) SFT, RFT, and RIFT have nearly colinear weight deltas (cosine >= 0.97, top-1 principal angle ~7 deg median over 144 modules) and comparable GSM8K accuracy (87-88%, n=1319; pairwise McNemar p >= 0.15); (ii) DFT diverges further in direction than any reward-weighted method despite using the same data; (iii) Offline GRPO adds a substantial component orthogonal to the SFT direction (~67% globally, up to ~86% in late layers) while staying in the SFT loss basin; (iv) DPO sits in a near-orthogonal subspace, shows a mode-connectivity barrier, and collapses late-layer CKA to ~0.46. DPO also reaches the highest accuracy in our protocol on both GSM8K (93.5%, McNemar p < 10^-9 vs. each other method) and AIME26 (30.0% vs. 3.3-10.0%); its training uses a 10x smaller learning rate than the others (the standard convention), so the update-norm and accuracy gaps reflect loss-function and optimizer choices jointly, and a learning-rate-matched DPO comparison is left for future work.

Aleksandr Nikolich, Igor Kiselev, Vladimir Platonov +1

1Keenable AI · 2Accenture · 3Yandex

RL for Language Models Direct Preference Optimization Offline RL Weight-Space Analysis Language Model Distillation Weight-Space Learning

Jun 18, 2026

Beyond Uniform Forgetting: A Study of Sequential Direct Preference Optimization Across Preference Settings

Aligning language models with human preferences often requires optimising multiple behavioural objectives. A practical approach is to apply these objectives sequentially using preference optimisation methods such as Direct Preference Optimisation (DPO), but it remains unclear whether later training uniformly degrades preferences learned earlier or whether the effect depends on the relationship between objectives. We study sequential DPO across four preference settings covering distributional conflict, multi-attribute interaction, strong safety signal, and compatible response-quality objectives. Using Llama-3.1-8B-Instruct with LoRA adapters, we evaluate all objectives after every stage with a fixed base-model reference. We find that sequential DPO does not produce a single forgetting pattern; preference change ranges from partial degradation to stability, pair-level redistribution, or positive transfer depending on objective relationship, signal strength, and training order. Pair-level analysis using length-normalised policy margins shows that aggregate metrics can mask heterogeneous changes across preference pairs, whereas quartile decomposition reveals that high-confidence pairs can either degrade or improve depending on the setting. Mechanistic diagnostics show that Stage~2 gradients and adapter updates are near-orthogonal to the previous objective across all settings, providing little evidence that direct gradient opposition is the primary driver. These findings suggest that future sequential alignment pipelines should account for objective compatibility and signal strength, rather than assuming that later objectives affect earlier preferences uniformly.

Pranav Bhandari, Nicolas Fay, Amitava Datta +2

1Network Analysis and Social Influence Modelling (NASIM) Lab · School of Physics Maths and Computing, The University of Western Australia · School of Psychological Science, The University of Western Australia +1

Continual Learning for LLMs Direct Preference Optimization Preference Optimization Multi-Objective Language Model Alignment Catastrophic Forgetting

Jun 17, 2026

Which Pairs to Compare for LLM Post-Training?

Preference-based post-training has become a central paradigm for aligning language models. A common data-collection strategy is to generate a small set of completions for each prompt and label the resulting comparison pairs. However, human preference labels are often much more expensive than generating additional completions, suggesting a different use of the same labeling budget: generate a larger pool of completions, but label only the most informative comparison pairs. This paper studies which pairs should be compared in preference-based post-training. We formulate comparison curation as a sampling-design problem and evaluate designs by the quality of the final policy under the preference-based post-training objective. We instantiate this framework for Direct Preference Optimization (DPO), analyzing how the choice of labeled pairs propagates through DPO training to downstream policy performance. Our main results provide matching upper and lower bounds on the post-training optimality gap of the DPO-trained policy. The bounds show that comparison selection affects downstream performance through a single design-dependent information matrix, which links label allocation to parameter estimation error and policy suboptimality. This yields an explicit optimization criterion for budgeted comparison curation and motivates practical sampling designs for selecting informative pairs from large generated completion pools. Experiments on synthetic settings and language-model post-training benchmarks show that the proposed designs consistently improve sample efficiency over common comparison-selection heuristics.

Jiangze Han, Vineet Goyal, Will Ma

Columbia University

Pairwise Preference Learning Pairwise Comparison LLM Alignment Pairwise Preference Evaluation Direct Preference Optimization Data Selection+1

Jun 17, 2026

Repeated post-training is not Self-improving: Diagnosing Scientific Amnesia in Continual DPO Pipelines

Industrial LLM teams often ship behavior updates by repeatedly DPO-training a base model on sequences of related preference-data campaigns. The dominant failure mode in this regime is not always classical catastrophic forgetting: a pipeline may preserve previously learned behaviors while still failing to accumulate reusable methodological knowledge about how to train the next campaign. We call this failure mode scientific amnesia. This paper turns that practitioner intuition into a measurable industrial problem. We contribute: (i) a diagnostic suite for amnesia, (ii) a Program-based pipeline that chains FSDP-sharded DPO checkpoints across Qwen2.5-7B-Instruct runs, (iii) a 30-campaign HumanEval subdomain benchmark, and (iv) a comparative diagnostic study of five strategy proposers: random memory, rule-based scheduling, retrieval-only memory, warm-start Bayesian optimization, and MSCL, a meta-scientific memory and reasoner candidate. Across a single-seed 5-condition * 3-step real-LM chain, 4 of 5 candidates degrade in step-level peak pass@1, including MSCL; only the deliberately conservative rule-based schedule improves. Follow-up pilots qualify rather than overturn this finding: in a heterogeneous chain, MSCL is the only completed candidate that improves, whereas in a small multi-seed homogeneous sweep, retrieval-only has the best mean Delta and no pairwise candidate gap is statistically distinguishable. The contribution is therefore diagnostic, not a claim that MSCL solves the problem: scientific amnesia is observable in a production-like continual-DPO pipeline, and conclusions about interventions depend sharply on chain regime, evaluator design, and seed coverage.

Jianzhe Lin, Fei Wang, Xiaolin Li +2

MetaAI

Continual Learning for LLMs Direct Preference Optimization LLM Post-Training Generative AI Evaluation

Jun 12, 2026

When Good Verifiers Go Bad: Silent Negative Transfer in Verifier-Guided VLM Training

Verifier reliability is not portable across tasks. A verifier-guided self-DPO pipeline with genuine held-out gains on MathVista (+9.6 points on self-training data, +8.0 held out) can be harmful on MMMU. The failure is invisible from the target-task self-training signal: over six learner-verifier configurations, MMMU self-training performance still improves by +3.52 points while held-out performance drops by 1.42 points. We call this silent negative transfer: a verifier validated as useful before can still show signatures of successful self-training after its induced update stops transferring to unseen data. Downstream failure alone cannot diagnose this: poor generalization may mean the verifier induces a wrong learning direction, or that the correct direction itself does not generalize beyond the self-training distribution. We separate these effects via gradient fidelity F, the alignment between verifier-induced and correct training directions, and gradient transferability T, the alignment between correct and held-out directions, giving a conservative Safe-Transfer Margin: positive first-order held-out alignment is certified when arccos(F) + arccos(T) < pi/2. We then introduce Asymmetric Acceptance Gating (AAG), a precision-first filter selecting preference pairs by the verifier's absolute confidence in the response receiving the positive update. On MMMU, AAG raises fidelity from 0.29 to 0.42. In a same-cell analysis, the correct direction stays positively aligned with the held-out direction (T = 0.31), yet raw verification rotates the induced update to negative held-out alignment (-0.13); AAG restores positive alignment (+0.22). Across all six MMMU configurations, AAG beats raw verifier-guided training. Verifier reliability thus means not that it worked before, but that it induces the right update on the task at hand and that this update stays useful beyond the data that generated it.

Jianzhe Lin

MetaAI · Work done at Meta

VLM Evaluation VLM Adaptation Direct Preference Optimization VLM Reasoning Negative Transfer

Jun 11, 2026

Direct Preference Optimization for Chatbot Fine-Tuning: An Empirical Study

We present an approach to fine-tuning large language models using Direct Preference Optimization (DPO), a reinforcement learning technique. Our experimental results demonstrate that DPO simplifies the training pipeline, improves computational efficiency, and achieves competitive performance. The evaluation using BLEU, ROUGE, and cosine similarity metrics indicates effective learning and convergence, though further investigation is needed to address observed training instability.

Dezhi Yu, Yvonne Qiu, ShuoJia Fu

LLM Alignment Direct Preference Optimization LLM Fine-Tuning LLM Post-Training Fine-Tuning Language Models

Jun 10, 2026

Boosting Direct Preference Optimization with Penalization

Offline preference optimization has become a practical substitute for reinforcement learning from human feedback, but pairwise objectives such as Direct Preference Optimization (DPO) and its variants use only the chosen and rejected responses stored in a static dataset. This leaves a useful signal unused: the response that the reference model itself would generate for the same prompt. We propose Direct Preference Optimization with Penalization (DPOP), a simple extension of DPO that augments the base preference loss with a gated penalty on reference-greedy responses. DPOP activates this penalty only when the current policy still assigns a lower likelihood to the preferred response than to the rejected response. On AlpacaEval 2.0, DPOP improves length-controlled win rate over DPO, SimPO, and AlphaDPO on both Llama-3-8b-it and Gemma-2-9b-it, achieving relative gains of 5.3% and 4.4% over baselines on the two models, respectively. Ablations further show that a SimNPO-style length-normalized penalty is stronger than NPO and token-level unlikelihood in this setting.

Pengwei Sun

Stanford University.

Pairwise Preference Learning LLM Alignment Direct Preference Optimization Offline Preference Optimization

Jun 9, 2026

Training LLMs to Enforce Multi-Level Instruction Hierarchies via Gravity-Weighted Direct Preference Optimization

Production LLMs receive instructions from sources with very different levels of trust, yet attend to every token with uniform architectural privilege. This is the structural vulnerability that enables malicious prompt injections and, more broadly, leaves models without a principled way to resolve conflicts between legitimate but competing instructions. A common training-based response is to teach models an explicit instruction hierarchy; existing approaches, however, formalize hierarchies of only three or four levels, treat all violations as equally severe, and rarely evaluate the full set of pairwise level interactions. We formalize a k-level instruction hierarchy problem and instantiate it for k=5, yielding ten pairwise priority relations that a compliant model must enforce. We then introduce Gravity-Weighted DPO (GW-DPO), a preference-optimization objective whose per-sample offset scales with the structural distance between conflicting levels under a linear or bilateral schedule, the latter weighting severity by both the privilege gap and the privilege of the victim level. Combined with hierarchy-specific delimiter tokens (Chen et al., 2025) and Instructional Segment Embeddings (ISE; Wu et al., 2025), GW-DPO with the bilateral schedule Pareto-improves over standard DPO and the linear variant on Llama-3.1-8B-Instruct, raising macro pairwise priority adherence while keeping over-refusal at half the standard DPO rate. Ablations isolate ISE as a refusal-threshold calibrator and recast five- versus three-level training as a generality-specialization tradeoff.

Lena S. Bolliger, Lena A. Jäger

Department of Computational Linguistics, University of Zurich, Switzerland

Direct Preference Optimization LLM Safety Alignment Prompt Injection Defense LLM Post-Training

Jun 4, 2026

DOG-DPO:Dynamic Optimization in Geometry for Safety Alignment

Safety alignment for large language models relies on preference data, but current pipelines often train on large, redundant datasets. Existing data selection methods typically score each preference pair independently, collapsing directional preference information into scalar quality or diversity scores. This sample-centric view is especially limiting in multi-dataset settings, where shared safety directions coexist with dataset-specific residual risks. We propose DOG-DPO, a training-free data selection framework that treats preference pairs as structured geometric signals. DOG-DPO first represents each preference pair as a direction in model representation space. It then decomposes multi-dataset preference geometry into a global anchor subspace and dataset-specific residual subspaces. Finally, it selects subsets by maximizing diversity-based coverage, encouraging broad, non-redundant coverage of alignment directions before DPO training. Across six safety benchmarks and two model backbones, DOG-DPO achieves a strong utility-robustness trade-off using only 11% of the preference pairs. It recovers most of the safety gains of full-data training while remaining entirely teacher-free, training-free, and substantially faster than representative selection baselines.

Yi Nian, Tiankai Yang, Yudi Zhang +7

University of Southern California · Iowa State University · University of Wisconsin–Madison +4

Direct Preference Optimization LLM Safety Alignment Multi-Objective Language Model Alignment Preference Alignment

Jun 2, 2026

P $^2$ -DPO: Grounding Hallucination in Perceptual Processing via Calibration Direct Preference Optimization

Hallucination has recently garnered significant research attention in Large Vision-Language Models (LVLMs). Direct Preference Optimization (DPO) aims to learn directly from the corrected preferences provided by humans, thereby addressing the hallucination issue. Despite its success, this paradigm has yet to specifically target the perceptual bottleneck in attended regions or address insufficient Visual Robustness against image degradation. Furthermore, existing preference pairs are often vision-agnostic and their inherently off-policy nature limits their effectiveness in guiding model learning. To address these challenges, we propose Perceptual Processing Direct Preference Optimization (P

^2

-DPO), a novel training paradigm in which the model generates and learns from its own preference pairs, thereby directly addressing the identified visual bottlenecks while inherently avoiding the issues of vision-agnostic and off-policy data. It introduces: (1) an on-policy preference pairs construction method targeting Focus-and-Enhance perception and Visual Robustness, and (2) a well-designed Calibration Loss to precisely align visual signals with the causal generation of text. Experimental results demonstrate that with a comparable amount of training data and cost, P

^2

-DPO outperforms strong baselines that rely on costly human feedback on benchmarks. Furthermore, evaluations on Attention Region Fidelity (ARF) and image degradation scenarios validate the effectiveness of P

^2

-DPO in addressing perceptual bottleneck in attended regions and improving Visual Robustness against degraded inputs.

Ruipeng Zhang, Zhihao Li, Haozhang Yuan +2

Guangdong Provincial Key Laboratory of Computational AI Models and Cognitive Intelligence, School of Computer Science & Engineering, South China University of Technology · Pazhou Lab, Guangzhou, China · Engineering Research Center of the Ministry of Education on Health Intelligent Perception and Paralleled Digital-Human, Guangzhou, China

VLM Robustness Direct Preference Optimization Object Hallucination in VLMs Fine-Grained Visual Perception VLM Hallucination Mitigation

Jun 2, 2026

Follow-Your-Preference++: Rethinking Preference Alignment for Image Inpainting

We study preference alignment for image inpainting. Rather than proposing yet another method, we revisit the problem from first principles and reassess its core challenges. We adopt the widely used direct preference optimization framework and construct preference training data with publicly available reward models. Our empirical study spans nine reward models, two benchmarks, and two baseline inpainting models that differ in architecture and generative mechanism. Our main findings are: (1) Most reward models provide valid signals for preference data construction, although some are unreliable as evaluators. (2) Across models and benchmarks, preference data exhibits consistent trends under both candidate and sample scaling. (3) Reward models display pronounced biases--particularly in brightness, composition, and color scheme--that make them prone to inducing reward hacking. (4) A simple ensemble of reward models mitigates such biases and yields robust, generalizable performance. {\color{rebuttal_blue}(5) Preference alignment is transferable to the object removal task, where the goal shifts from open-ended creative generation to coherent background completion. (6) Further analysis reveals that a calibrated ensemble method further mitigates hacking and improves robustness.} Without modifying model architectures or introducing additional datasets, our models substantially outperform prior state-of-the-art models on standard metrics, large vision-language model evaluations, and human assessments. Our code is available at: https://github.com/shenytzzz/Follow-Your-Preference.

Junkun Yuan, Yutao Shen, Toru Aonishi +2

Zhejiang University · The University of Tokyo · Tsinghua University

Pairwise Preference Learning Reward Modeling Reward Hacking Direct Preference Optimization Image Inpainting Preference Alignment

Jun 1, 2026

AgentPLM: Agentic Protein Language Models with Reasoning-Augmented Decoding for Protein Sequence Design

Protein language models (PLMs) are passive oracles: they generate sequences in a single forward pass with no mechanism to consult external biophysical feedback or redirect generation when a candidate violates thermodynamic or structural constraints. We introduce AgentPLM, which addresses this by equipping a pre-trained PLM with i) Reasoning-Augmented Decoding (RAD), which interleaves autoregressive generation with tool calls (ESMFold, FoldX, AutoDock Vina), and ii) Contrastive Agent Policy Optimisation (CAPO), a trajectory-level extension of direct preference optimisation that trains the policy end-to-end to learn when oracle feedback is informative rather than merely imitating high-fitness sequences. We evaluate AgentPLM on benchmark tasks spanning de novo enzyme design, antibody optimisation, thermostability, PPI interface design, and zero-shot fitness prediction with standardised oracle APIs and controlled sequence-identity splits. AgentPLM achieves state-of-the-art results with a gain in antibody top-10% hit rate over the strongest passive baseline, providing mechanistic evidence of online error correction without explicit backtracking.

Sahil Rahman, Maxx Richard Rahman

Bedford College, London, United Kingdom · Saarland University, Saarbrücken, Germany

RL for Language Models Direct Preference Optimization Protein Language Models LLM Tool Use Molecular Generation Protein Design

May 30, 2026

Citation Grounding: Detecting and Reducing LLM Citation Hallucinations via Legal Citation Graphs

Large language models systematically hallucinate legal citations -- fabricating statute references, citing repealed provisions, and confusing jurisdictions -- yet no automated method exists to measure or reduce this behavior at scale. We propose citation grounding (CG), a metric that verifies LLM-generated legal citations against a ground-truth citation graph extracted from 100.8 million Ukrainian court decisions (502 million edges, 21,736 unique statute nodes). CG decomposes into three components -- citation precision (does the cited provision exist?), citation relevance (is it contextually appropriate?), and citation temporality (was it valid at the relevant date?) -- enabling differential diagnosis of hallucination types. Empirical evaluation on 100 Ukrainian legal queries across five systems -- four commercial LLMs via AWS Bedrock (Claude Haiku 4.5, Mistral Pixtral Large, Amazon Nova Pro/Lite) and one RAG-augmented production system -- reveals CG ranging from 0.791 to 0.873, with 13-21% of citations hallucinated. To reduce hallucinations without human annotation, we introduce Citation Grounding DPO (CG-DPO): a method that constructs preference pairs algorithmically by corrupting verified citations from real court decisions via four targeted strategies. On a dataset of 2,244 court decisions, a Qwen2.5-7B-Instruct model fine-tuned with LoRA achieves 98.5% mean validation accuracy in distinguishing correct from corrupted citations (rewards margin +14.9, std < 0.3 pp across 3 seeds). The citation graph, evaluation framework, and CG-DPO dataset are released as open resources.

Volodymyr Ovcharov

LEX AI LLC, Kyiv, Ukraine

Legal Citation Verification LLM Hallucination Mitigation Direct Preference Optimization LLM Hallucination Detection

May 29, 2026

The Representation-Rationalizability Tradeoff in Reward Learning

In RLHF, each training example contains a prompt

x

and two candidate responses

y,y'

, and annotators provide pairwise preferences between these responses. The learning problem is to convert these heterogeneous pairwise judgments into a single scalar reward

r(x,y)

that measures response quality for each prompt. Classical social choice implies an impossibility because heterogeneous annotator samples can induce pooled preferences with Condorcet cycles, so no scalar reward can evaluate all compared response pairs consistently. A growing literature analyzes RLHF as a social-choice problem, but usually assumes a fixed finite set of alternatives, i.e., a pre-enumerated finite set of candidate responses for each prompt. Modern pipelines instead score responses through a learned representation

φ(x,y)

before a scalar head, so

φ

determines which responses are treated as distinguishable alternatives and which comparisons are visible to the reward model. Once this embedding is part of the problem, the impossibility results from social choice theory become a tradeoff. We show that the excess cross-entropy loss of any reward built on

φ

decomposes exactly into a representational term, which a richer

φ

shrinks, and an aggregation term, which a richer

φ

enlarges by exposing more comparisons that no scalar can rank consistently. The same results extend to direct preference optimization (DPO), and jointly training the embedding and the reward cannot guarantee to recover the sweet spot of this tradeoff. Experiments on synthetic data and real preference datasets corroborate our results.

Jing Dong, Yaoliang Yu, Pascal Pourpart

Vector Institute · University of Waterloo & Vector Institute

Pairwise Preference Learning Reward Modeling Reinforcement Learning Social Choice Theory Pairwise Preference Evaluation Direct Preference Optimization+3

May 29, 2026

Learning from Fine-Grained Visual Discrepancies: Mitigating Multimodal Hallucinations via In-Context Visual Contrastive Optimization

Multimodal hallucination remains a persistent challenge for Vision-Language Models (VLMs). Standard textual Direct Preference Optimization (DPO) often fails to mitigate it due to a lack of explicit visual supervision. While existing works introduce visual preference DPO by contrasting original images against negative ones, they suffer from a theoretically inconsistent objective caused by partition function mismatches and rely on coarse-grained negatives that could enable shortcut learning. In this work, we propose In-Context Visual Contrastive Optimization (IC-VCO). By placing contrastive images within a shared multi-image context, IC-VCO ensures a mathematically rigorous objective. We further introduce Visual Contrast Distillation (VCDist), an auxiliary reliability-gated regularizer that encourages consistency between multi-image contrastive training and single-image inference. Finally, we propose a contrastive sample editing strategy that generates hard negatives via precise semantic perturbations. Experiments on five benchmarks demonstrate IC-VCO's best overall performance and the effectiveness of our sample editing strategy. Code and data are available at https://github.com/OPPO-Mente-Lab/IC-VCO.

Haolin Deng, Xin Zou, Zhiwei Jin +3

The Hong Kong University of Science and Technology (Guangzhou) · OPPO AI Center · The Hong Kong University of Science and Technology

Vision-Language Models Direct Preference Optimization VLM Hallucination Mitigation

May 28, 2026

Recovering Diversity Without Losing Alignment: A DPO Recipe for Post-Trained LLMs

Many open-ended instructions have multiple valid answers that users can benefit from seeing, but post-training often narrows an LLM's output space toward a small set of canonical responses. We introduce REDIPO, an offline DPO data-construction pipeline for recovering distinct valid answer modes while preserving the alignment benefits of the instruct model. For each prompt, REDIPO samples responses from both base and instruct models, rewrites base-model responses with the instruct model, filters candidates for safety and instruction-following quality, and builds preference pairs that favor marginally diverse responses among candidates with similar instruction-following reward. Across Qwen3-4B, OLMo-3-7B, and LLaMA-3.1-8B, REDIPO improves NoveltyBench distinct_k by 134%, 33%, and 44% relative to the instruct checkpoints, while DivPO changes diversity by 0%, -6%, and -4% on the same models. These gains largely maintain MTBench, IFEval, and Arena-Hard performance, and reduce direct-category HarmBench attack success rate. Ablations show that marginal-diversity pair selection and base-response rewriting drive the diversity gains, while filtering and quality-bounded pairing help maintain alignment. Overall, our results show that diverse valid answers from base-model generations can be reintroduced through carefully constructed preference data while retaining the alignment benefits of post-training. We release our code and data at https://github.com/vsamuel2003/ReDiPO.

Vinay Samuel, Yapei Chang, Mohit Iyyer

University of Maryland, College Park

LLM Alignment Direct Preference Optimization Output Diversity in Language Models LLM Post-Training

May 27, 2026

Compositional Text-to-Image Generation Via Region-aware Bimodal Direct Preference Optimization

Despite the rapid progress of text-to-image (T2I) models, generating images that accurately reflect complex compositional prompts (covering attribute bindings, object relationships, counting) still remains challenging. To address this, we propose BiDPO, a framework to enhance T2I model's capability of compositional text-to-image generation. We begin by introducing an carefully designed pipeline to construct a large-scale preference dataset, BiComp, with strictly quality control. Then, we extend Diffusion DPO to jointly optimize image and text preferences, which is shown to greatly effective in improving the models to follow complex text prompt in generation. To further enhance the models for fine-grained alignment, we employ a region-level guidance method to focus on regions relevant to compositional concepts. Experimental results demonstrate that our BiDPO substantially improves compositional fidelity, consistently outperforming prior methods across multiple benchmarks. Our approach highlights the potential of preference-based fine-tuning for complex text-to-image tasks, offering a flexible and scalable alternative to existing techniques.

Zhuohan Liu, Wujian Peng, Yitong Chen +1

Shanghai Key Lab of Intell. Info. Processing, School of CS, Fudan University · Shanghai Collaborative Innovation Center of Intelligent Visual Computing

Diffusion Model Guidance Direct Preference Optimization Compositional T2I Generation T2I Diffusion Models

May 27, 2026

AdaDPO: Self-Adaptive Direct Preference Optimization with Balanced Gradient Updates

DPO has become a widely adopted alternative to RLHF for aligning LLMs with human preferences, eliminating the need for a separate reward model or RL loop. Recent theoretical analysis uncovers an asymmetric gradient behavior in DPO: the loss suppresses dispreferred responses substantially faster than it promotes preferred ones, causing the model to learn to avoid bad answers rather than to generate good ones. We propose AdaDPO, a Self-Adaptive variant of the DPO algorithm that introduces per-preference-pair, stop-gradient-based coefficients derived directly from the policy model's generation probabilities, with the reference model's probabilities as an optional component. AdaDPO is constructed to enforce equality of gradient magnitudes between preferred and dispreferred probabilities; the practical implementation balances per-token gradients and applies a numerical clipping bound for stability, while retaining DPO's original hyperparameter structure. On Llama-3-8B-Instruct trained on UltraFeedback under a SimPO similar setup, AdaDPO consistently outperforms DPO on AlpacaEval 2: it achieves higher length-controlled win rates (LC) in 81% of hyperparameter combinations, attains the global best LC (48.3%) and raw win rate (46.1%), and enlarges the LC-over-WR margin in 88% of combinations, indicating effective mitigation of length bias. Additional analyses on KL divergence, reward margin, and reward accuracy confirm that AdaDPO rectifies the gradient imbalance and yields more efficient optimization. Because it operates purely at the loss level, AdaDPO can be dropped into existing preference-based alignment pipelines without changing data collection or model architectures. The method requires only a few lines of code, and the same self-adaptive principle generalizes to a broad family of pairwise contrastive preference losses including SimPO, R-DPO, IPO, CPO, and ORPO.

Shaolong Chen, Madalina Ciobanu, Qingqing Mao +1

Incept Labs Houston, TX · Incept Labs, Houston, TX · Titan Holdings, San Francisco, CA

LLM Alignment Direct Preference Optimization Adaptive Loss Weighting

May 27, 2026

Reasoning Matters: Mitigate Hallucination in Multimodal Large Reasoning Models via Reasoning-Conditioned Preference Optimization

Multimodal Large Reasoning Models introduce the reasoning paradigm, demonstrating strong capabilities on complex vision-language tasks. However, they still suffer from severe hallucinations. Existing training-based methods typically mitigate hallucinations through response-level direct preference optimization (DPO), where the Chain-of-Thought (CoT) and the final answer are treated as a monolithic output and optimized jointly. We reveal that this formulation performs similarly to answer-only optimization, suggesting that it primarily learns answer-level preference, while leaving CoT-level supervision insufficiently exploited. To address this issue, we explicitly formulate a CoT-oriented preference term and derive Reasoning-Conditioned Direct Preference Optimization (RC-DPO), which models the CoT as a condition for answer generation and contrasts the preference for the same preferred answer under different CoT conditions, promoting answer-supportive reasoning chain alignment. To further improve optimization, we introduce a reasoning-enhanced preference data generation strategy that employs Monte Carlo Tree Search to discover visually grounded and logically consistent CoTs as positive samples, and attention-guided CoT token pruning to construct negative ones. Extensive experiments across various models and benchmarks show that RC-DPO effectively mitigates hallucinations and improves the reliability of the multimodal reasoning process.

Jiawei Kong, Hao Fang, Shunxiang Liao +5

Tsinghua Shenzhen International Graduate School, Tsinghua University · Harbin Institute of Technology, Shenzhen

Multimodal Grounding Direct Preference Optimization Hallucination in Language Models Multimodal Large Language Models Multimodal Hallucination Multimodal CoT Reasoning

May 25, 2026

Curriculum Learning for Safety Alignment

Direct Preference Optimisation (DPO) is widely used for safety alignment in large language models. However, prior work shows it is brittle and exhibits poor out-of-distribution (OOD) generalisation. In this paper, we investigate whether Curriculum Learning can improve the robustness of DPO-based safety alignment. We propose Staged-Competence, a curriculum-based framework that organises preference data by difficulty, employs competence-based sampling, and progressively updates the reference model during training. Averaged across three model families, Staged-Competence reduces OOD harmful response rates by 16% and jailbreak attack success rates by 20%, while preserving general capabilities with near-zero over-refusal. We further show that Staged-Competence (1) matches baseline safety with only 75% of the training data and (2) yields better separation between safe and unsafe responses. Staged-Competence is agnostic to the policy optimisation loss and can extend to other DPO variants and alignment domains. Our code and data are available at https://github.com/Sandeep5500/curriculum-learning-for-safety.

Sandeep Kumar, Virginia Smith, Chhavi Yadav

Carnegie Mellon University · Simons Institute, UC Berkeley

LLM Alignment Direct Preference Optimization LLM Safety Alignment Adversarial Robustness of Language Models

May 25, 2026

Towards Anatomically Plausible Human Image Generation via Synthetic Localized Preferences

Large-scale text-to-image foundation models have achieved remarkable visual realism, yet generating human images with correct anatomical structures remains challenging. Existing approaches enforce anatomical constraints through part-specific modules or localized loss weighting during supervised fine-tuning on high-quality human photos, but such datasets are limited and often provide ambiguous optimization signals due to confounding factors such as lighting, pose, and background. Preference-based alignment offers an alternative, but standard Direct Preference Optimization (DPO) treats all pixels equally and therefore fails to exploit the localized nature of anatomical artifacts. To address this, we propose the framework of Alignment via Synthetic Anatomical Preference (ASAP), which constructs controlled preference pairs through a localized degradation mechanism applied to high-fidelity human images. This mechanism performs a controlled experiment on images by introducing explicit anatomical errors in targeted regions while preserving the remaining content. With this mechanism, we create the Human Anatomical Preference (HAP) dataset with over 10K curated pairs for effective anatomical alignment of text-to-image human image generative models. To better leverage the locality of these controlled preference pairs, we introduce a localized and margin-bounded variant of DPO that prioritizes optimization in targeted anatomical regions while enforcing a finite preference margin to prevent over-optimization and preserve global semantics. We further introduce HAF-Bench, a benchmark for systematic evaluation of anatomical fidelity. Extensive experiments demonstrate that ASAP consistently reduces anatomical errors across multiple foundation models while maintaining overall image quality.

Bao Li, Yuliang Xiu, Zhen Liu

Westlake University · The Chinese University of Hong Kong, Shenzhen

Diffusion Model Alignment Direct Preference Optimization Image Generation T2I Diffusion Models

May 21, 2026

Token-weighted Direct Preference Optimization with Attention

Direct Preference Optimization (DPO) aligns Large Language Models with human preferences without the need for a separate reward model. However, DPO treats all tokens in responses equally, neglecting the differing importance of individual tokens. Existing token-level PO methods compute the token weights using either token-position-based heuristic functions or probability estimates given by a separately trained model, which lacks robustness and incurs extra training cost. In contrast, we propose Token-weighted DPO (TwDPO) -- a novel training objective grounded on token-weighted RL -- and AttentionPO -- an instantiation of TwDPO that uses attention from the LLM itself to estimate token weights. AttentionPO prompts the LLM to serve as a pairwise judge and check where the model attends when comparing the responses. This design makes AttentionPO content-aware, adjusting weights based on response content, and efficient, incurring only two extra forward passes per example. Experiment results show that AttentionPO significantly improves performance on AlpacaEval, MT-Bench, and ArenaHard, surpassing existing Preference Optimization methods.

Chengyu Huang, Zhuohang Li, Sheng-Yen Chou +1

Cornell University · Vanderbilt University

Pairwise Preference Learning LLM Alignment Direct Preference Optimization Preference Optimization Preference-Based RL

May 21, 2026

CrossVLA: Cross-Paradigm Post-Training and Inference Optimization for Vision-Language-Action Models

Vision-Language-Action (VLA) models have rapidly converged on a small set of architectural patterns: discrete-token autoregression (e.g. OpenVLA) and continuous-action flow-matching (e.g. pi-0.5). Yet preference alignment via Direct Preference Optimisation (DPO) -- the de-facto post-training step in language models -- has been studied almost exclusively on autoregressive VLAs. We present CrossVLA, an empirical study of cross-paradigm VLA post-training. Three contributions: (i) a surrogate flow-matching log-probability estimator that lets DPO operate on continuous-action backbones without probability-flow ODE integration; (ii) a head-to-head comparison of LoRA and DoRA as the parameter-efficient layer for VLA DPO, finding DoRA improves over OpenVLA SFT by a mean +10.4 pp across LIBERO 4-suite (600 trials, 3 seeds) -- per-suite +20.0 Object, +11.0 Long-horizon, +8.0 Goal, +2.7 Spatial -- with zero seed variance on Object (38/50 on each of 3 seeds); (iii) an inference-time anatomy showing the denoise loop dominates 78.6% of sample_actions latency and prefix-K/V caching a la VLA-Cache caps at a 21% acceleration ceiling -- both chunk-level and token-level cache strategies degrade success rate to 0-80% in our benchmarks. We further pretrain a multi-view + temporal projection head on 6000 LIBERO frames, achieving 99.5% k-NN recall@1 for same-task retrieval (36x over random), available as a downstream initialisation. All code, ckpts, training logs, and reproduction scripts are open at https://github.com/lz-googlefycy/vla-lab.

Zhi Liu

Tianjin University

Direct Preference Optimization VLA Model Post-Training Vision-Language-Action Models Parameter-Efficient Fine-Tuning

May 20, 2026

How Much Online RL is Enough? Informative Rollouts for Offline Preference Optimization in RLVR

Reinforcement Learning from Verifiable Rewards (RLVR) has emerged as a powerful paradigm for reasoning in language models, with GRPO as its primary example. However, GRPO requires continuous online rollout generation, making it computationally expensive and difficult to scale. While Direct Preference Optimization (DPO) offers a stable and efficient offline alternative, it is typically expected to underperform w.r.t. online RL methods such as GRPO when trained on rollouts from a cold supervised fine-tuned (SFT) policy. We introduce G2D (GRPO to DPO)}, a three-stage pipeline that performs a short GRPO warm-up, constructs a static preference dataset, and fine-tunes a model offline with DPO. Across a set of values of the number of online steps (K) in GRPO on Qwen2.5-7B and Llama-3.1-8B, we find that offline DPO with moderate warm-up matches or outperforms GRPO at substantially lower compute cost in our setting. On Qwen2.5-7B, G2D at K=150 achieves 62.4% on MATH-500, outperforming GRPO (51.6%) by 10.8% at ~4x lower compute. On Llama-3.1-8B, G2D at K=500 achieves 49.4%, surpassing GRPO in our experimental setting. We show that performance is not governed by the number of preference pairs, which does not vary much w.r.t. K, but by their informativeness. Moderate warm-up produces rollouts with calibrated uncertainty, yielding stronger contrastive signal, while excessive warm-up leads to overconfident policies and less informative data. Our results recast the offline-online gap in RLVR as primarily a data informativeness problem, and identify short online RL warm-up with appropriate difficulty calibration of the fine-tuning dataset as a compute-efficient alternative to online RL.

Richa Verma, Balaraman Ravindran

TCS Research Department of CSE IIT Madras Chennai, India · Department of Data Science & AI, Wadhwani School of Data Science & AI IIT Madras Chennai, India

Direct Preference Optimization RL for Language Model Reasoning Reinforcement Learning with Verifiable Rewards Preference-Based RL

May 20, 2026

PREFINE: Preference-Based Implicit Reward and Cost Fine-Tuning for Safety Alignment

We address the problem of making a pre-trained reinforcement learning (RL) policy safety-aware by incorporating cost constraints without retraining it from scratch. While costs could be numerically encoded, we assume a more general setting is when costs are provided as preferences. Given a reward-optimized policy and a small dataset of preferred (low-cost) and dispreferred (high-cost) trajectories, our goal is to fine-tune the policy to generate low-cost behaviors while retaining high rewards. Unlike standard RLHF in language models, where preferences are defined over responses to the same prompt, our setting involves trajectory-level preferences in continuous control environments. We introduce PREFINE: Preference-based Implicit Reward and Cost Fine-Tuning for Safety Alignment which is a preference-based fine-tuning method that adapts Direct Preference Optimization (DPO), which is now widely used for LLM fine-tuning, to the sequential decision making setting. PREFINE constructs policy-sampled counterfactual trajectories to establish meaningful preference contrasts and jointly optimizes for reward retention and safety alignment. Empirically, PREFINE reduces constraint violations and catastrophic failures by over 60% while maintaining original reward behavior. PREFINE produces policies that achieve low-cost, high-reward performance with significantly improved data and computational efficiency compared to full offline RL or imitation learning, bridging preference alignment and safe policy adaptation in continuous domains.

Richa Verma, Bavish Kulur, Sanjay Chawla +1

TCS Research, Department of CSE, IIT Madras India · Department of Computing Science, University of Alberta Canada · Qatar Computing Research Institute, Hamad Bin Khalifa University Qatar +1

Direct Preference Optimization Constrained RL Continuous Control Safe RL RL Fine-Tuning Preference-Based RL