arXiv subjects

Xi Leng

Publications and source records attributed to Xi Leng.

5 recordsLinked to original sources

arXiv2026

RAVE: Re-Allocating Visual Attention in Large Multimodal Models

Large multimodal models (LMMs) inherit the self-attention mechanism of pretrained language backbones, yet standard attention can exhibit suboptimal allocation, including cross-modal misallocation between textual and visual evidence and intra-visual imbalance among visual tokens. We propose RAVE (Re-Allocating Visual Attention), a lightweight pair-gating mechanism that adds a learned query-key bias to pre-softmax attention scores over visual keys, derived from pre-RoPE query and key features. RAVE requires no architectural modification to the backbone and can be trained end-to-end with the rest of the model. Across a suite of multimodal benchmarks, RAVE improves over standard attention by an average of 3 points, with the largest gains on perception-intensive tasks -- including multilingual OCR, chart understanding, document VQA, and scene text VQA -- where accurate visual grounding is critical.

cs.CV↗

arXiv2026

Revisiting Reinforcement Learning with Verifiable Rewards from a Contrastive Perspective

Group Relative Policy Optimization (GRPO) is one of the most widely adopted RLVR algorithms for post-training large language models on reasoning tasks. We first show that GRPO admits an equivalent discriminative reformulation, in which policy optimization maximizes the expected score gap between verified positive and negative rollouts. This reformulation reveals two objective-level limitations: likelihood-misaligned surrogate scores, in which clipped ratio-based scores are optimized rather than the sequence likelihoods that govern generation, and score-insensitive credit assignment, in which rollout-level credit does not reflect the current score gaps between positive and negative rollouts. To address these limitations, we propose ConSPO, a Contrastive Sequence-level Policy Optimization method that uses length-normalized sequence log-probabilities as rollout scores and contrasts verified positive rollouts against negative distractors within the same group. ConSPO optimizes a group-wise InfoNCE-style objective to adaptively strengthen updates for poorly separated positives and high-scoring negatives, together with a curriculum-scheduled margin that preserves separation pressure as training progresses. Experiments across diverse settings show that ConSPO outperforms strong baselines on challenging reasoning benchmarks.

cs.LG↗

arXiv2025

ADHint: Adaptive Hints with Difficulty Priors for Reinforcement Learning

To address the limited capability expansion and low sample efficiency of Reinforcement Learning (RL), recent methods have integrated ''hints'' into post-training, which are prefix segments of complete reasoning trajectories, aiming for powerful knowledge expansion and reasoning generalization. However, existing hint-based RL methods often neglect the role of difficulty in the hint-ratio schedule and relative-advantage estimation, resulting in unstable learning and excessive imitation of off-policy hints. To address this, we propose ADHint, which explicitly integrates difficulty into both processes to achieve a better trade-off between exploration and imitation. Specifically, we propose Adaptive Hint with Sample Difficulty Prior, which evaluates the difficulty of each sample under the current policy to schedule an appropriate hint ratio for rollout generation. Furthermore, we introduce Consistency-based Gradient Modulation alongside Selective Masking for Hint Preservation, which jointly modulate token-level gradients within hints to prevent biased and destructive updates. Additionally, we propose Advantage Estimation with Rollout Difficulty Posterior, which leverages the relative difficulty of rollouts with and without hints to compute their respective advantages, yielding more balanced updates. Extensive experiments across diverse modalities, scales, model families, and domains show that ADHint achieves superior reasoning capabilities and out-of-distribution generalization. Code will be released upon paper acceptance.

cs.CV↗

arXiv2023

FedRec+: Enhancing Privacy and Addressing Heterogeneity in Federated Recommendation Systems

Preserving privacy and reducing communication costs for edge users pose significant challenges in recommendation systems. Although federated learning has proven effective in protecting privacy by avoiding data exchange between clients and servers, it has been shown that the server can infer user ratings based on updated non-zero gradients obtained from two consecutive rounds of user-uploaded gradients. Moreover, federated recommendation systems (FRS) face the challenge of heterogeneity, leading to decreased recommendation performance. In this paper, we propose FedRec+, an ensemble framework for FRS that enhances privacy while addressing the heterogeneity challenge. FedRec+ employs optimal subset selection based on feature similarity to generate near-optimal virtual ratings for pseudo items, utilizing only the user's local information. This approach reduces noise without incurring additional communication costs. Furthermore, we utilize the Wasserstein distance to estimate the heterogeneity and contribution of each client, and derive optimal aggregation weights by solving a defined optimization problem. Experimental results demonstrate the state-of-the-art performance of FedRec+ across various reference datasets.

cs.LG↗

arXiv2022

Diversity Boosted Learning for Domain Generalization with Large Number of Domains

Machine learning algorithms minimizing the average training loss usually suffer from poor generalization performance due to the greedy exploitation of correlations among the training data, which are not stable under distributional shifts. It inspires various works for domain generalization (DG), where a series of methods, such as Causal Matching and FISH, work by pairwise domain operations. They would need $O(n^2)$ pairwise domain operations with $n$ domains, where each one is often highly expensive. Moreover, while a common objective in the DG literature is to learn invariant representations against domain-induced spurious correlations, we highlight the importance of mitigating spurious correlations caused by objects. Based on the observation that diversity helps mitigate spurious correlations, we propose a Diversity boosted twO-level saMplIng framework (DOMI) utilizing Determinantal Point Processes (DPPs) to efficiently sample the most informative ones among large number of domains. We show that DOMI helps train robust models against spurious correlations from both domain-side and object-side, substantially enhancing the performance of the backbone DG algorithms on rotated MNIST, rotated Fashion MNIST, and iwildcam datasets.

cs.LG↗