Haokun Li

Shuaicheng Niu, Jiaxiang Wu, Yifan Zhang et al.

Conventional deep models predict a test sample with a single forward propagation, which, however, may not be sufficient for predicting hard-classified samples. On the contrary, we human beings may need to carefully check the sample many times before making a final decision. During the recheck process, one may refine/adjust the prediction by referring to related samples. Motivated by this, we propose to predict those hard-classified test samples in a looped manner to boost the model performance. However, this idea may pose a critical challenge: how to construct looped inference, so that the original erroneous predictions on these hard test samples can be corrected with little additional effort. To address this, we propose a general Closed-Loop Inference (CLI) method. Specifically, we first devise a filtering criterion to identify those hard-classified test samples that need additional inference loops. For each hard sample, we construct an additional auxiliary learning task based on its original top-$K$ predictions to calibrate the model, and then use the calibrated model to obtain the final prediction. Promising results on ImageNet (in-distribution test samples) and ImageNet-C (out-of-distribution test samples) demonstrate the effectiveness of CLI in improving the performance of any pre-trained model.

Jialun Cao, Yaojie Lu, Meiziniu Li et al.

The research in AI-based formal mathematical reasoning has shown an unstoppable growth trend. These studies have excelled in mathematical competitions like IMO and have made significant progress. This paper focuses on formal verification, an immediate application scenario of formal reasoning, and breaks it down into sub-tasks. We constructed 18k high-quality instruction-response pairs across five formal specification languages (Coq, Lean4, Dafny, ACSL, and TLA+) by distilling gpt-4o and evaluated against ten open-sourced LLMs, including recent popular DeepSeek-R1. We also fine-tuned several 7~8B small models to achieve comparable performance with Deepseek-R1-671B. Interestingly, we observed that fine-tuning with formal data also enhances mathematics, reasoning, and coding capabilities. Fine-tuned models are released at https: //huggingface.co/fm-universe.

Haokun Li, Yazhou Zhang, Jizhi Ding et al.

Can multi-modal large language models (MLLMs) truly understand what they can see? Extending Searle's Chinese Room into the multi-modal domain, this paper proposes the Visual Room argument: MLLMs may describe every visual detail precisely yet fail to comprehend the underlying emotions and intentions, namely seeing is not understanding. Building on this, we introduce \textit{Visual Room} 2.0, a hierarchical benchmark for evaluating perception-cognition alignment of MLLMs. We model human perceptive and cognitive processes across three levels: low, middle, and high, covering 17 representative tasks. The perception component ranges from attribute recognition to scene understanding, while the cognition component extends from textual entailment to causal and social reasoning. The dataset contains 350 multi-modal samples, each with six progressive questions (2,100 in total) spanning perception to cognition. Evaluating 10 state-of-the-art (SoTA) MLLMs, we highlight three key findings: (1) MLLMs exhibit stronger perceptual competence than cognitive ability (8.0\%$\uparrow$); (2) cognition appears not causally dependent on perception-based reasoning; and (3) cognition scales with model size, but perception does not consistently improve with larger variants. This work operationalizes Seeing $\ne$ Understanding as a testable hypothesis, offering a new paradigm from perceptual processing to cognitive reasoning in MLLMs. Our dataset is available at https://huggingface.co/datasets/LHK2003/PCBench.

Haozhong Sun, Zhongsen Li, Chenlin Du et al.

Quantitative magnetic resonance imaging (qMRI) requires multi-phase acqui-sition, often relying on reduced data sampling and reconstruction algorithms to accelerate scans, which inherently poses an ill-posed inverse problem. While many studies focus on measuring uncertainty during this process, few explore how to leverage it to enhance reconstruction performance. In this paper, we in-troduce PUQ, a novel approach that pioneers the use of uncertainty infor-mation for qMRI reconstruction. PUQ employs a two-stage reconstruction and parameter fitting framework, where phase-wise uncertainty is estimated during reconstruction and utilized in the fitting stage. This design allows uncertainty to reflect the reliability of different phases and guide information integration during parameter fitting. We evaluated PUQ on in vivo T1 and T2 mapping datasets from healthy subjects. Compared to existing qMRI reconstruction methods, PUQ achieved the state-of-the-art performance in parameter map-pings, demonstrating the effectiveness of uncertainty guidance. Our code is available at https://anonymous.4open.science/r/PUQ-75B2/.

Shoukai Xu, Haokun Li, Bohan Zhuang et al.

Neural network quantization is an effective way to compress deep models and improve their execution latency and energy efficiency, so that they can be deployed on mobile or embedded devices. Existing quantization methods require original data for calibration or fine-tuning to get better performance. However, in many real-world scenarios, the data may not be available due to confidential or private issues, thereby making existing quantization methods not applicable. Moreover, due to the absence of original data, the recently developed generative adversarial networks (GANs) cannot be applied to generate data. Although the full-precision model may contain rich data information, such information alone is hard to exploit for recovering the original data or generating new meaningful data. In this paper, we investigate a simple-yet-effective method called Generative Low-bitwidth Data Free Quantization (GDFQ) to remove the data dependence burden. Specifically, we propose a knowledge matching generator to produce meaningful fake data by exploiting classification boundary knowledge and distribution information in the pre-trained model. With the help of generated data, we can quantize a model by learning knowledge from the pre-trained model. Extensive experiments on three data sets demonstrate the effectiveness of our method. More critically, our method achieves much higher accuracy on 4-bit quantization than the existing data free quantization method. Code is available at https://github.com/xushoukai/GDFQ.

Yaofo Chen, Zeng You, Shuhai Zhang et al.

Transformer-based Large Language Models (LLMs) have exhibited remarkable success in extensive tasks primarily attributed to self-attention mechanism, which requires a token to consider all preceding tokens as its context to compute attention. However, when the context length L becomes very large (e.g., 128K), the amount of potentially redundant information in the context tends to increase. The redundant context not only hampers the modeling representation performance but also incurs unnecessary computational and storage overhead. In this paper, we propose a plug-and-play Core Context Aware (CCA) Attention for efficient long-context modeling, comprising two complementary modules: 1) Globality-aware pooling module groups input tokens and dynamically compresses each group into one core token based on their significance. In this way, our method automatically focuses and strengthens core context while diminishing redundancy during the learning process, leading to effective long-term dependency modeling. 2) Locality-preserving module incorporates neighboring tokens to preserve local context for detailed representation. Notably, our CCA-Attention is able to replace the self-attention module in existing LLMs with minimal fine-tuning cost. Extensive experimental results show the superiority of our method in both long-context modeling and computational efficiency over state-of-the-art methods.

Zekang Zheng, Haokun Li, Yaofo Chen et al.

Model quantization reduces neural network parameter precision to achieve compression, but often compromises accuracy. Existing post-training quantization (PTQ) methods employ iterative parameter updates to preserve accuracy under high compression ratios, incurring significant computational complexity and resource overhead, which limits applicability in resource-constrained edge computing and real-time inference scenarios. This paper proposes an efficient PTQ method guided by parameter sensitivity analysis. The approach prioritizes quantization of high-sensitivity parameters, leveraging unquantized low-sensitivity parameters to compensate for quantization errors, thereby mitigating accuracy degradation. Furthermore, by exploiting column-wise clustering of parameter sensitivity, the method introduces a row-parallel quantization framework with a globally shared inverse Hessian matrix update mechanism, reducing computational complexity by an order of magnitude. Experimental results on ResNet-50 and YOLOv5s demonstrate a 20-200-fold quantization speedup over the Optimal Brain Quantization baseline, with mean accuracy loss below 0.3%, confirming the method's efficacy in balancing efficiency and accuracy.

Haokun Li, Yazhou Zhang, Jizhi Ding et al.

With the rapid advancement of Multimodal Large Language Models (MLLMs), they have demonstrated exceptional capabilities across a variety of vision-language tasks. However, current evaluation benchmarks predominantly focus on objective visual question answering or captioning, inadequately assessing the models' ability to understand complex and subjective human emotions. To bridge this gap, we introduce EmoBench-Reddit, a novel, hierarchical benchmark for multimodal emotion understanding. The dataset comprises 350 meticulously curated samples from the social media platform Reddit, each containing an image, associated user-provided text, and an emotion category (sad, humor, sarcasm, happy) confirmed by user flairs. We designed a hierarchical task framework that progresses from basic perception to advanced cognition, with each data point featuring six multiple-choice questions and one open-ended question of increasing difficulty. Perception tasks evaluate the model's ability to identify basic visual elements (e.g., colors, objects), while cognition tasks require scene reasoning, intent understanding, and deep empathy integrating textual context. We ensured annotation quality through a combination of AI assistance (Claude 4) and manual verification.We conducted a comprehensive evaluation of nine leading MLLMs, including GPT-5, Gemini-2.5-pro, and GPT-4o, on EmoBench-Reddit.

Tianyi Ding, Haokun Li, Xinpeng Ni et al.

In this paper, we propose a hybrid framework for Satisfiability Modulo the Theory of Nonlinear Real Arithmetic (SMT-NRA for short). First, we introduce a two-dimensional cell-jump move, called \emph{$2d$-cell-jump}, generalizing the key operation, cell-jump, of the local search method for SMT-NRA. Then, we propose an extended local search framework, named \emph{$2d$-LS} (following the local search framework, LS, for SMT-NRA), integrating the model constructing satisfiability calculus (MCSAT) framework to improve search efficiency. To further improve the efficiency of MCSAT, we implement a recently proposed technique called \emph{sample-cell projection operator} for MCSAT, which is well suited for CDCL-style search in the real domain and helps guide the search away from conflicting states. Finally, we present a hybrid framework for SMT-NRA integrating MCSAT, $2d$-LS and OpenCAD, to improve search efficiency through information exchange. The experimental results demonstrate improvements in local search performance, highlighting the effectiveness of the proposed methods.

Dongyang Xu, Haokun Li, Qingfan Wang et al.

End-to-end autonomous driving has witnessed remarkable progress. However, the extensive deployment of autonomous vehicles has yet to be realized, primarily due to 1) inefficient multi-modal environment perception: how to integrate data from multi-modal sensors more efficiently; 2) non-human-like scene understanding: how to effectively locate and predict critical risky agents in traffic scenarios like an experienced driver. To overcome these challenges, in this paper, we propose a Multi-Modal fusion transformer incorporating Driver Attention (M2DA) for autonomous driving. To better fuse multi-modal data and achieve higher alignment between different modalities, a novel Lidar-Vision-Attention-based Fusion (LVAFusion) module is proposed. By incorporating driver attention, we empower the human-like scene understanding ability to autonomous vehicles to identify crucial areas within complex scenarios precisely and ensure safety. We conduct experiments on the CARLA simulator and achieve state-of-the-art performance with less data in closed-loop benchmarks. Source codes are available at https://anonymous.4open.science/r/M2DA-4772.

Haokun Li, Bican Xia

A new algorithm for deciding the satisfiability of polynomial formulas over the reals is proposed. The key point of the algorithm is a new projection operator, called sample-cell projection operator, custom-made for Conflict-Driven Clause Learning (CDCL)-style search. Although the new operator is also a CAD (Cylindrical Algebraic Decomposition)-like projection operator which computes the cell (not necessarily cylindrical) containing a given sample such that each polynomial from the problem is sign-invariant on the cell, it is of singly exponential time complexity. The sample-cell projection operator can efficiently guide CDCL-style search away from conflicting states. Experiments show the effectiveness of the new algorithm.