arXiv subjects

Yongqi Li

Publications and source records attributed to Yongqi Li.

At least 37 records · Page 2Linked to original sources

arXiv2026

AR-Omni: A Unified Autoregressive Model for Any-to-Any Generation

Real-world perception and interaction are inherently multimodal, encompassing not only language but also vision and speech, which motivates the development of "Omni" MLLMs that support both multimodal inputs and multimodal outputs. While a sequence of omni MLLMs has emerged, most existing systems still rely on additional expert components to achieve multimodal generation, limiting the simplicity of unified training and inference. Autoregressive (AR) modeling, with a single token stream, a single next-token objective, and a single decoder, is an elegant and scalable foundation in the text domain. Motivated by this, we present AR-Omni, a unified any-to-any model in the autoregressive paradigm without any expert decoders. AR-Omni supports autoregressive text and image generation, as well as streaming speech generation, all under a single Transformer decoder. We further address three practical issues in unified AR modeling: modality imbalance via task-aware loss reweighting, visual fidelity via a lightweight token-level perceptual alignment loss for image tokens, and stability-creativity trade-offs via a finite-state decoding mechanism. Empirically, AR-Omni achieves strong quality across three modalities while remaining real-time, achieving a 0.88 real-time factor for speech generation.

cs.LG↗

arXiv2026

Agent-as-a-Judge

LLM-as-a-Judge has revolutionized AI evaluation by leveraging large language models for scalable assessments. However, as evaluands become increasingly complex, specialized, and multi-step, the reliability of LLM-as-a-Judge has become constrained by inherent biases, shallow single-pass reasoning, and the inability to verify assessments against real-world observations. This has catalyzed the transition to Agent-as-a-Judge, where agentic judges employ planning, tool-augmented verification, multi-agent collaboration, and persistent memory to enable more robust, verifiable, and nuanced evaluations. Despite the rapid proliferation of agentic evaluation systems, the field lacks a unified framework to navigate this shifting landscape. To bridge this gap, we present the first comprehensive survey tracing this evolution. Specifically, we identify key dimensions that characterize this paradigm shift and establish a developmental taxonomy. We organize core methodologies and survey applications across general and professional domains. Furthermore, we analyze frontier challenges and identify promising research directions, ultimately providing a clear roadmap for the next generation of agentic evaluation.

cs.CL↗

arXiv2025

R$^2$ec: Towards Large Recommender Models with Reasoning

Large recommender models have extended LLMs as powerful recommenders via encoding or item generation, and recent breakthroughs in LLM reasoning synchronously motivate the exploration of reasoning in recommendation. In this work, we propose R$^2$ec, a unified large recommender model with intrinsic reasoning capability. R$^2$ec introduces a dual-head architecture that supports both reasoning chain generation and efficient item prediction in a single model, significantly reducing inference latency. To overcome the lack of annotated reasoning data, we design RecPO, a reinforcement learning framework that optimizes reasoning and recommendation jointly with a novel fused reward mechanism. Extensive experiments on three datasets demonstrate that R$^2$ec outperforms traditional, LLM-based, and reasoning-augmented recommender baselines, while further analyses validate its competitive efficiency among conventional LLM-based recommender baselines and strong adaptability to diverse recommendation scenarios. Code and checkpoints available at https://github.com/YRYangang/RRec.

cs.IR↗

arXiv2025

Learnable Item Tokenization for Generative Recommendation

Utilizing powerful Large Language Models (LLMs) for generative recommendation has attracted much attention. Nevertheless, a crucial challenge is transforming recommendation data into the language space of LLMs through effective item tokenization. Current approaches, such as ID, textual, and codebook-based identifiers, exhibit shortcomings in encoding semantic information, incorporating collaborative signals, or handling code assignment bias. To address these limitations, we propose LETTER (a LEarnable Tokenizer for generaTivE Recommendation), which integrates hierarchical semantics, collaborative signals, and code assignment diversity to satisfy the essential requirements of identifiers. LETTER incorporates Residual Quantized VAE for semantic regularization, a contrastive alignment loss for collaborative regularization, and a diversity loss to mitigate code assignment bias. We instantiate LETTER on two models and propose a ranking-guided generation loss to augment their ranking ability theoretically. Experiments on three datasets validate the superiority of LETTER, advancing the state-of-the-art in the field of LLM-based generative recommendation.

cs.IR↗

arXiv2025

TokenSkip: Controllable Chain-of-Thought Compression in LLMs

Chain-of-Thought (CoT) has been proven effective in enhancing the reasoning capabilities of large language models (LLMs). Recent advancements, such as OpenAI's o1 and DeepSeek-R1, suggest that scaling up the length of CoT sequences during inference could further boost LLM reasoning performance. However, due to the autoregressive nature of LLM decoding, longer CoT outputs lead to a linear increase in inference latency, adversely affecting user experience, particularly when the CoT exceeds 10,000 tokens. To address this limitation, we analyze the semantic importance of tokens within CoT outputs and reveal that their contributions to reasoning vary. Building on this insight, we propose TokenSkip, a simple yet effective approach that enables LLMs to selectively skip less important tokens, allowing for controllable CoT compression. Extensive experiments across various models and tasks demonstrate the effectiveness of TokenSkip in reducing CoT token usage while preserving strong reasoning performance. Notably, when applied to Qwen2.5-14B-Instruct, TokenSkip reduces reasoning tokens by 40% (from 313 to 181) on GSM8K, with less than a 0.4% performance drop. We release our code and checkpoints in https://github.com/hemingkx/TokenSkip.

cs.CL↗

arXiv2025

A Survey on Training-free Alignment of Large Language Models

The alignment of large language models (LLMs) aims to ensure their outputs adhere to human values, ethical standards, and legal norms. Traditional alignment methods often rely on resource-intensive fine-tuning (FT), which may suffer from knowledge degradation and face challenges in scenarios where the model accessibility or computational resources are constrained. In contrast, training-free (TF) alignment techniques--leveraging in-context learning, decoding-time adjustments, and post-generation corrections--offer a promising alternative by enabling alignment without heavily retraining LLMs, making them adaptable to both open-source and closed-source environments. This paper presents the first systematic review of TF alignment methods, categorizing them by stages of pre-decoding, in-decoding, and post-decoding. For each stage, we provide a detailed examination from the viewpoint of LLMs and multimodal LLMs (MLLMs), highlighting their mechanisms and limitations. Furthermore, we identify key challenges and future directions, paving the way for more inclusive and effective TF alignment techniques. By synthesizing and organizing the rapidly growing body of research, this survey offers a guidance for practitioners and advances the development of safer and more reliable LLMs.

cs.CL↗

arXiv2025

MultiConIR: Towards multi-condition Information Retrieval

Multi-condition information retrieval (IR) presents a significant, yet underexplored challenge for existing systems. This paper introduces MultiConIR, a benchmark specifically designed to evaluate retrieval and reranking models under nuanced multi-condition query scenarios across five diverse domains. We systematically assess model capabilities through three critical tasks: complexity robustness, relevance monotonicity, and query format sensitivity. Our extensive experiments on 15 models reveal a critical vulnerability: most retrievers and rerankers exhibit severe performance degradation as query complexity increases. Key deficiencies include widespread failure to maintain relevance monotonicity, and high sensitivity to query style and condition placement. The superior performance of GPT-4o reveals the performance gap between IR systems and advanced LLM for handling sophisticated natural language queries. Furthermore, this work delves into the factors contributing to reranker performance deterioration and examines how condition positioning within queries affects similarity assessment, providing crucial insights for advancing IR systems towards complex search scenarios. The code and datasets are available at https://github.com/EIT-NLP/MultiConIR

cs.IR↗

arXiv2025

EARN: Efficient Inference Acceleration for LLM-based Generative Recommendation by Register Tokens

Large Language Model-based generative recommendation (LLMRec) has achieved notable success, but it suffers from high inference latency due to massive computational overhead and memory pressure of KV Cache. Existing KV Cache reduction methods face critical limitations: cache compression offers marginal acceleration given recommendation tasks' short decoding steps, while prompt compression risks discarding vital interaction history. Through systematic analysis of attention patterns in LLMRec, we uncover two pivotal insights: 1) layer-wise attention sparsity inversion where early layers retain dense informative patterns while later layers exhibit high redundancy, and 2) dual attention sinks phenomenon where attention scores concentrate on both head and tail tokens of input sequences. Motivated by these insights, we propose EARN, an efficient inference framework that leverages the early layers to compress information into register tokens placed at the input sequence boundaries, then focuses solely on these tokens in the subsequent layers. Extensive experiments on three datasets, two LLMRec methods and two LLM architectures demonstrate EARN's superiority, achieving up to 3.79x speedup and 80.8% KV Cache reduction with better accuracy than the general finetuning approach. Our work bridges the efficiency-effectiveness gap in LLMRec, offering practical deployment advantages for industrial scenarios.

cs.IR↗

arXiv2025

Theory of Dielectric Behavior in Composites

While the properties of materials at microscopic scales are well described by fundamental quantum mechanical equations and electronic structure theories, the emergent behavior of mesoscopic or macroscopic composites is no longer governed solely by quantum effects. Instead, such systems are dominated by complex heterogeneous architectures and macroscopic interactions, presenting a classical many-body problem with unique complexities that remain less systematically understood than their quantum counterparts. In this work, we develop an operator-based theoretical framework to characterize these systems, using composite dielectric behavior as a paradigmatic example. By integrating effective medium theory with electromagnetic simulation techniques, we construct an operator that rigorously expresses the effective permittivity tensor as an exact functional. Global and local structure-property relationships can be established by analyzing the operator's structure through symmetric singular value decomposition and block operator matrix analysis, respectively. This framework bridges the gap between microscopic physics and macroscopic material behavior, offering a powerful approach for understanding diverse material properties and guiding the rational design of novel functional composites.

physics.app-ph↗

arXiv2025

Exploring Training and Inference Scaling Laws in Generative Retrieval

Generative retrieval reformulates retrieval as an autoregressive generation task, where large language models (LLMs) generate target documents directly from a query. As a novel paradigm, the mechanisms that underpin its performance and scalability remain largely unexplored. We systematically investigate training and inference scaling laws in generative retrieval, exploring how model size, training data scale, and inference-time compute jointly influence performance. We propose a novel evaluation metric inspired by contrastive entropy and generation loss, providing a continuous performance signal that enables robust comparisons across diverse generative retrieval methods. Our experiments show that n-gram-based methods align strongly with training and inference scaling laws. We find that increasing model size, training data scale, and inference-time compute all contribute to improved performance, highlighting the complementary roles of these factors in enhancing generative retrieval. Across these settings, LLaMA models consistently outperform T5 models, suggesting a particular advantage for larger decoder-only models in generative retrieval. Our findings underscore that model sizes, data availability, and inference computation interact to unlock the full potential of generative retrieval, offering new insights for designing and optimizing future systems.

cs.IR↗

arXiv2025

Scaling over Scaling: Exploring Test-Time Scaling Plateau in Large Reasoning Models

Large reasoning models (LRMs) have exhibited the capacity of enhancing reasoning performance via internal test-time scaling. Building upon this, a promising direction is to further scale test-time compute to unlock even greater reasoning capabilities. However, as we push these scaling boundaries, systematically understanding the practical limits and achieving optimal resource allocation becomes a critical challenge. In this paper, we investigate the scaling plateau of test-time scaling and introduce the Test-Time Scaling Performance Model (TTSPM). We theoretically analyze two fundamental paradigms for such extended scaling, parallel scaling and sequential scaling, from a probabilistic modeling perspective. Our primary contribution is the derivation of the saturation point on the scaling budget for both strategies, identifying thresholds beyond which additional computation yields diminishing returns. Remarkably, despite their distinct mechanisms, both paradigms converge to a unified mathematical structure in their upper bounds. We empirically validate our theoretical findings on challenging reasoning benchmarks, including AIME, MATH-500, and GPQA, demonstrating the practical utility of these bounds for test-time resource allocation. We hope that this work provides insights into the cost-benefit trade-offs of test-time scaling, guiding the development of more resource-efficient inference strategies for large reasoning models.

cs.AI↗

arXiv2025

Aligning VLM Assistants with Personalized Situated Cognition

Vision-language models (VLMs) aligned with general human objectives, such as being harmless and hallucination-free, have become valuable assistants of humans in managing visual tasks. However, people with diversified backgrounds have different cognition even in the same situation. Consequently, they may have personalized expectations for VLM assistants. This highlights the urgent need to align VLM assistants with personalized situated cognition for real-world assistance. To study this problem, we first simplify it by characterizing individuals based on the sociological concept of Role-Set. Then, we propose to evaluate the individuals' actions to examine whether the personalized alignment is achieved. Further, we construct a benchmark named PCogAlignBench, which includes 18k instances and 20 individuals with different Role-Sets. Finally, we present a framework called PCogAlign, which constructs a cognition-aware and action-based reward model for personalized alignment. Experimental results and human evaluations demonstrate the reliability of the PCogAlignBench and the effectiveness of our proposed PCogAlign. We will open-source the constructed benchmark and code at https://github.com/NLPGM/PCogAlign.

cs.AI↗

arXiv2025

TIGeR: Unifying Text-to-Image Generation and Retrieval with Large Multimodal Models

How humans can effectively and efficiently acquire images has always been a perennial question. A classic solution is text-to-image retrieval from an existing database; however, the limited database typically lacks creativity. By contrast, recent breakthroughs in text-to-image generation have made it possible to produce attractive and counterfactual visual content, but it faces challenges in synthesizing knowledge-intensive images. In this work, we rethink the relationship between text-to-image generation and retrieval, proposing a unified framework for both tasks with one single Large Multimodal Model (LMM). Specifically, we first explore the intrinsic discriminative abilities of LMMs and introduce an efficient generative retrieval method for text-to-image retrieval in a training-free manner. Subsequently, we unify generation and retrieval autoregressively and propose an autonomous decision mechanism to choose the best-matched one between generated and retrieved images as the response to the text prompt. To standardize the evaluation of unified text-to-image generation and retrieval, we construct TIGeR-Bench, a benchmark spanning both creative and knowledge-intensive domains. Extensive experiments on TIGeR-Bench and two retrieval benchmarks, i.e., Flickr30K and MS-COCO, demonstrate the superiority of our proposed framework.

cs.CV↗

arXiv2025

Large Language Models Empowered Personalized Web Agents

Web agents have emerged as a promising direction to automate Web task completion based on user instructions, significantly enhancing user experience. Recently, Web agents have evolved from traditional agents to Large Language Models (LLMs)-based Web agents. Despite their success, existing LLM-based Web agents overlook the importance of personalized data (e.g., user profiles and historical Web behaviors) in assisting the understanding of users' personalized instructions and executing customized actions. To overcome the limitation, we first formulate the task of LLM-empowered personalized Web agents, which integrate personalized data and user instructions to personalize instruction comprehension and action execution. To address the absence of a comprehensive evaluation benchmark, we construct a Personalized Web Agent Benchmark (PersonalWAB), featuring user instructions, personalized user data, Web functions, and two evaluation paradigms across three personalized Web tasks. Moreover, we propose a Personalized User Memory-enhanced Alignment (PUMA) framework to adapt LLMs to the personalized Web agent task. PUMA utilizes a memory bank with a task-specific retrieval strategy to filter relevant historical Web behaviors. Based on the behaviors, PUMA then aligns LLMs for personalized action execution through fine-tuning and direct preference optimization. Extensive experiments validate the superiority of PUMA over existing Web agents on PersonalWAB.

cs.CL↗

arXiv2025

Towards Harmless Multimodal Assistants with Blind Preference Optimization

Multimodal Large Language Models (MLLMs) have demonstrated impressive capabilities in multimodal understanding, reasoning, and interaction. Given the extensive applications of MLLMs, the associated safety issues have become increasingly critical. Due to the effectiveness of preference optimization in aligning MLLMs with human preferences, there is an urgent need for safety-related preference data for MLLMs. To address this, we construct the MMSafe-PO preference dataset towards harmless multimodal assistants, featuring multimodal instructions, the conversational format, and ranked paired responses from human feedback. We also identify two insightful observations: modality co-defense and modality cheating, which illustrate that MLLMs possess a certain level of inherent defense while still presenting unique safety challenges. Based on these observations, we propose the Blind Preference Optimization (BPO) approach. Comprehensive experiments on three benchmarks show that BPO effectively enhances the safety capabilities of MLLMs. Notably, BPO significantly improves the safety rate of the base MLLM by 45.0%, outperforming the DPO approach. Additionally, applying BPO to the MMSafe-PO dataset greatly reduces the base MLLM's unsafe rate on other safety benchmarks (14.5% on MM-SafetyBench and 82.9% on HarmEval, demonstrating the effectiveness and robustness of both the dataset and the approach. We release code and data at https://lu-yang666.github.io/MMsafe-PO-Web/.

cs.CL↗

arXiv2025

Interleaved-Modal Chain-of-Thought

Chain-of-Thought (CoT) prompting elicits large language models (LLMs) to produce a series of intermediate reasoning steps before arriving at the final answer. However, when transitioning to vision-language models (VLMs), their text-only rationales struggle to express the fine-grained associations with the original image. In this paper, we propose an image-incorporated multimodal Chain-of-Thought, named \textbf{Interleaved-modal Chain-of-Thought (ICoT)}, which generates sequential reasoning steps consisting of paired visual and textual rationales to infer the final answer. Intuitively, the novel ICoT requires VLMs to enable the generation of fine-grained interleaved-modal content, which is hard for current VLMs to fulfill. Considering that the required visual information is usually part of the input image, we propose \textbf{Attention-driven Selection (ADS)} to realize ICoT over existing VLMs. ADS intelligently inserts regions of the input image to generate the interleaved-modal reasoning steps with ignorable additional latency. ADS relies solely on the attention map of VLMs without the need for parameterization, and therefore it is a plug-and-play strategy that can be generalized to a spectrum of VLMs. We apply ADS to realize ICoT on two popular VLMs of different architectures. Extensive evaluations of three benchmarks have shown that ICoT prompting achieves substantial performance (up to 14\%) and interpretability improvements compared to existing multimodal CoT prompting methods.

cs.CV↗

arXiv2025

Adaptive Moment Estimation Optimization Algorithm Using Projection Gradient for Deep Learning

Training deep neural networks is challenging. To accelerate training and enhance performance, we propose PadamP, a novel optimization algorithm. PadamP is derived by applying the adaptive estimation of the p-th power of the second-order moments under scale invariance, enhancing projection adaptability by modifying the projection discrimination condition. It is integrated into Adam-type algorithms, accelerating training, boosting performance, and improving generalization in deep learning. Combining projected gradient benefits with adaptive moment estimation, PadamP tackles unconstrained non-convex problems. Convergence for the non-convex case is analyzed, focusing on the decoupling of first-order moment estimation coefficients and second-order moment estimation coefficients. Unlike prior work relying on , our proof generalizes the convergence theorem, enhancing practicality. Experiments using VGG-16 and ResNet-18 on CIFAR-10 and CIFAR-100 show PadamP's effectiveness, with notable performance on CIFAR-10/100, especially for VGG-16. The results demonstrate that PadamP outperforms existing algorithms in terms of convergence speed and generalization ability, making it a valuable addition to the field of deep learning optimization.

math.OC↗

arXiv2025

SWIFT: On-the-Fly Self-Speculative Decoding for LLM Inference Acceleration

Speculative decoding (SD) has emerged as a widely used paradigm to accelerate LLM inference without compromising quality. It works by first employing a compact model to draft multiple tokens efficiently and then using the target LLM to verify them in parallel. While this technique has achieved notable speedups, most existing approaches necessitate either additional parameters or extensive training to construct effective draft models, thereby restricting their applicability across different LLMs and tasks. To address this limitation, we explore a novel plug-and-play SD solution with layer-skipping, which skips intermediate layers of the target LLM as the compact draft model. Our analysis reveals that LLMs exhibit great potential for self-acceleration through layer sparsity and the task-specific nature of this sparsity. Building on these insights, we introduce SWIFT, an on-the-fly self-speculative decoding algorithm that adaptively selects intermediate layers of LLMs to skip during inference. SWIFT does not require auxiliary models or additional training, making it a plug-and-play solution for accelerating LLM inference across diverse input data streams. Our extensive experiments across a wide range of models and downstream tasks demonstrate that SWIFT can achieve over a 1.3x-1.6x speedup while preserving the original distribution of the generated text. We release our code in https://github.com/hemingkx/SWIFT.

cs.CL↗