Understanding and reasoning over academic handwritten notes remains a challenge in document AI, particularly for mathematical equations, diagrams, and scientific notations. Existing visual question answering (VQA) benchmarks focus on printed or structured handwritten text, limiting generalization to real-world note-taking. To address this, we introduce NoTeS-Bank, an evaluation benchmark for Neural Transcription and Search in note-based question answering. NoTeS-Bank comprises complex notes across multiple domains, requiring models to process unstructured and multimodal content. The benchmark defines two tasks: (1) Evidence-Based VQA, where models retrieve localized answers with bounding-box evidence, and (2) Open-Domain VQA, where models classify the domain before retrieving relevant documents and answers. Unlike classical Document VQA datasets relying on optical character recognition (OCR) and structured data, NoTeS-BANK demands vision-language fusion, retrieval, and multimodal reasoning. We benchmark state-of-the-art Vision-Language Models (VLMs) and retrieval frameworks, exposing structured transcription and reasoning limitations. NoTeS-Bank provides a rigorous evaluation with NDCG@5, MRR, Recall@K, IoU, and ANLS, establishing a new standard for visual document understanding and reasoning.

Improving the efficiency of inference in Large Language Models (LLMs) is a critical area of research. Post-training Quantization (PTQ) is a popular technique, but it often faces challenges at low-bit levels, particularly in downstream tasks. Quantization-aware Training (QAT) can alleviate this problem, but it requires significantly more computational resources. To tackle this, we introduced Weight-Decomposed Low-Rank Quantization-Aware Training (DL-QAT), which merges the advantages of QAT while training only less than 1% of the total parameters. Specifically, we introduce a group-specific quantization magnitude to adjust the overall scale of each quantization group. Within each quantization group, we use LoRA matrices to update the weight size and direction in the quantization space. We validated the effectiveness of our method on the LLaMA and LLaMA2 model families. The results show significant improvements over our baseline method across different quantization granularities. For instance, for LLaMA-7B, our approach outperforms the previous state-of-the-art method by 4.2% in MMLU on 3-bit LLaMA-7B model. Additionally, our quantization results on pre-trained models also surpass previous QAT methods, demonstrating the superior performance and efficiency of our approach.

Tracking multiple objects based on textual queries is a challenging task that requires linking language understanding with object association across frames. Previous works typically train the whole process end-to-end or integrate an additional referring text module into a multi-object tracker, but they both require supervised training and potentially struggle with generalization to open-set queries. In this work, we introduce ReferGPT, a novel zero-shot referring multi-object tracking framework. We provide a multi-modal large language model (MLLM) with spatial knowledge enabling it to generate 3D-aware captions. This enhances its descriptive capabilities and supports a more flexible referring vocabulary without training. We also propose a robust query-matching strategy, leveraging CLIP-based semantic encoding and fuzzy matching to associate MLLM generated captions with user queries. Extensive experiments on Refer-KITTI, Refer-KITTIv2 and Refer-KITTI+ demonstrate that ReferGPT achieves competitive performance against trained methods, showcasing its robustness and zero-shot capabilities in autonomous driving. The codes are available on https://github.com/Tzoulio/ReferGPT

The rapid advancement of large language models (LLMs) has brought significant benefits to various domains while introducing substantial risks. Despite being fine-tuned through reinforcement learning, LLMs lack the capability to discern malicious content, limiting their defense against jailbreak. To address these safety concerns, we propose a feature-aware method for harmful response rejection (FMM), which detects the presence of malicious features within the model's feature space and adaptively adjusts the model's rejection mechanism. By employing a simple discriminator, we detect potential malicious traits during the decoding phase. Upon detecting features indicative of toxic tokens, FMM regenerates the current token. By employing activation patching, an additional rejection vector is incorporated during the subsequent token generation, steering the model towards a refusal response. Experimental results demonstrate the effectiveness of our approach across multiple language models and diverse attack techniques, while crucially maintaining the models' standard generation capabilities.

Transformer-based language models have revolutionized the field of natural language processing (NLP). However, using these models often involves navigating multiple frameworks and tools, as well as writing repetitive boilerplate code. This complexity can discourage non-programmers and beginners, and even slow down prototyping for experienced developers. To address these challenges, we introduce Langformers, an open-source Python library designed to streamline NLP pipelines through a unified, factory-based interface for large language model (LLM) and masked language model (MLM) tasks. Langformers integrates conversational AI, MLM pretraining, text classification, sentence embedding/reranking, data labelling, semantic search, and knowledge distillation into a cohesive API, supporting popular platforms such as Hugging Face and Ollama. Key innovations include: (1) task-specific factories that abstract training, inference, and deployment complexities; (2) built-in memory and streaming for conversational agents; and (3) lightweight, modular design that prioritizes ease of use. Documentation: https://langformers.com

In real-world applications of large language models, outputs are often required to be confined: selecting items from predefined product or document sets, generating phrases that comply with safety standards, or conforming to specialized formatting styles. To control the generation, constrained decoding has been widely adopted. However, existing prefix-tree-based constrained decoding is inefficient under GPU-based model inference paradigms, and it introduces unintended biases into the output distribution. This paper introduces Dynamic Importance Sampling for Constrained Decoding (DISC) with GPU-based Parallel Prefix-Verification (PPV), a novel algorithm that leverages dynamic importance sampling to achieve theoretically guaranteed asymptotic unbiasedness and overcomes the inefficiency of prefix-tree. Extensive experiments demonstrate the superiority of our method over existing methods in both efficiency and output quality. These results highlight the potential of our methods to improve constrained generation in applications where adherence to specific constraints is essential.

Recent advancements in Large Vision-Language Models have showcased remarkable capabilities. However, they often falter when confronted with complex reasoning tasks that humans typically address through visual aids and deliberate, step-by-step thinking. While existing methods have explored text-based slow thinking or rudimentary visual assistance, they fall short of capturing the intricate, interleaved nature of human visual-verbal reasoning processes. To overcome these limitations and inspired by the mechanisms of slow thinking in human cognition, we introduce VisuoThink, a novel framework that seamlessly integrates visuospatial and linguistic domains. VisuoThink facilitates multimodal slow thinking by enabling progressive visual-textual reasoning and incorporates test-time scaling through look-ahead tree search. Extensive experiments demonstrate that VisuoThink significantly enhances reasoning capabilities via inference-time scaling, even without fine-tuning, achieving state-of-the-art performance in tasks involving geometry and spatial reasoning.

Abstract--Large Artificial Intelligence Models (LAMs) powered by massive datasets, extensive parameter scales, and extensive computational resources, leading to significant transformations across various industries. Y et, their practical deployment on resource-limited mobile edge devices is hindered by critical challenges such as data privacy, constrained resources, and high overhead costs. Addressing this gap, this paper proposes a novel framework, named Quantized Split Federated Fine-T uning Large AI Model (SFLAM). By partitioning the training load between edge devices and servers using a split learning paradigm, SFLAM can facilitate the operation of large models on devices and significantly lowers the memory requirements on edge devices. Additionally, SFLAM incorporates quantization management, power control, and bandwidth allocation strategies to enhance training efficiency while concurrently reducing energy consumption and communication latency. A theoretical analysis exploring the latency-energy trade-off is presented, and the framework's efficacy is validated via comprehensive simulations. The findings indicate that SFLAM achieves superior performance in terms of learning efficiency and scalability compared to conventional methods, thereby providing a valuable approach for enabling advanced AI services in resource-constrained scenarios. I. Introduction The advent of Large AI Models (LAMs), such as Chat-GPT and DeepSeek, marked a significant leap in AI capabilities, powered by their extensive parameter scales, large-scale datasets, and substantial computational resources [1]. As user demand for ubiquitous AI access and real-time, personalized experiences grows, deploying and training these models on mobile devices becomes increasingly relevant [2]. T o meet these escalating demands, fine-tuning, which involves adapting pre-trained models with domain-specific data, has become a widely adopted and efficient strategy for enhancing LAM performance on specialized tasks, offering a cost-effective path to superior results.

Recently, the reasoning capabilities of large reasoning models (LRMs), such as DeepSeek-R1, have seen significant advancements through the slow thinking process. Despite these achievements, the substantial computational demands of LRMs present considerable challenges. In contrast, small reasoning models (SRMs), often distilled from larger ones, offer greater efficiency and can exhibit distinct capabilities and cognitive trajectories compared to LRMs. This work surveys around 170 recently published papers on SRMs for tackling various complex reasoning tasks. We review the current landscape of SRMs and analyze diverse training and inference techniques related to SRMs. Furthermore, we provide a comprehensive review of SRMs for domain-specific applications and discuss possible future research directions. This survey serves as an essential reference for researchers to leverage or develop SRMs for advanced reasoning functionalities with high efficiency.

Recently, stepwise supervision on Chain of Thoughts (CoTs) presents an enhancement on the logical reasoning tasks such as coding and math, with the help of Monte Carlo Tree Search (MCTS). However, its contribution to tasks requiring domain-specific expertise and knowledge remains unexplored. Motivated by the interest, we identify several potential challenges of vanilla MCTS within this context, and propose the framework of Stepwise Domain Knowledge-Driven Reasoning Optimization, employing the MCTS algorithm to develop step-level supervision for problems that require essential comprehension, reasoning, and specialized knowledge. Additionally, we also introduce the Preference Optimization towards Reflection Paths, which iteratively learns self-reflection on the reasoning thoughts from better perspectives. We have conducted extensive experiments to evaluate the advantage of the methodologies. Empirical results demonstrate the effectiveness on various legal-domain problems. We also report a diverse set of valuable findings, hoping to encourage the enthusiasm to the research of domain-specific LLMs and MCTS.