arXiv subjects

Haotian Wang

Publications and source records attributed to Haotian Wang.

At least 91 records · Page 5Linked to original sources

arXiv2025

How Difficulty-Aware Staged Reinforcement Learning Enhances LLMs' Reasoning Capabilities: A Preliminary Experimental Study

Enhancing the reasoning capabilities of Large Language Models (LLMs) with efficiency and scalability remains a fundamental challenge in artificial intelligence research. This paper presents a rigorous experimental investigation into how difficulty-aware staged reinforcement learning (RL) strategies can substantially improve LLM reasoning performance. Through systematic analysis, we demonstrate that strategically selecting training data according to well-defined difficulty levels markedly enhances RL optimization. Moreover, we introduce a staged training methodology, progressively exposing models to increasingly challenging tasks, further amplifying reasoning capabilities. Our findings reveal significant cross-domain benefits when simultaneously training models on mathematical reasoning and code generation tasks. Notably, our proposed approach enables a 1.5B parameter model to achieve an accuracy of 42.3\% on the AIME-2024 benchmark, 89.5\% on the MATH-500 benchmark. These results underscore the efficacy of our method in advancing the reasoning proficiency of LLMs. We will open-source our datasets on GitHub and Hugging Face.

cs.CL↗

arXiv2025

1.4 Million Open-Source Distilled Reasoning Dataset to Empower Large Language Model Training

The AM-DeepSeek-R1-Distilled is a large-scale dataset with thinking traces for general reasoning tasks, composed of high-quality and challenging reasoning problems. These problems are collected from a multitude of open-source datasets, subjected to semantic deduplication and meticulous cleaning to eliminate test set contamination. All responses within the dataset are distilled from reasoning models (predominantly DeepSeek-R1) and have undergone rigorous verification procedures. Mathematical problems are validated by checking against reference answers, code problems are verified using test cases, and other tasks are evaluated with the aid of a reward model. The AM-Distill-Qwen-32B model, which was trained through only simple Supervised Fine-Tuning (SFT) using this batch of data, outperformed the DeepSeek-R1-Distill-Qwen-32B model on four benchmarks: AIME2024, MATH-500, GPQA-Diamond, and LiveCodeBench. Additionally, the AM-Distill-Qwen-72B model surpassed the DeepSeek-R1-Distill-Llama-70B model on all benchmarks as well. We are releasing these 1.4 million problems and their corresponding responses to the research community with the objective of fostering the development of powerful reasoning-oriented Large Language Models (LLMs). The dataset was published in \href{https://huggingface.co/datasets/a-m-team/AM-DeepSeek-R1-Distilled-1.4M}{https://huggingface.co/datasets/a-m-team/AM-DeepSeek-R1-Distilled-1.4M}.

cs.CL↗

arXiv2025

Think Twice: Enhancing LLM Reasoning by Scaling Multi-round Test-time Thinking

Recent advances in large language models (LLMs), such as OpenAI-o1 and DeepSeek-R1, have demonstrated the effectiveness of test-time scaling, where extended reasoning processes substantially enhance model performance. Despite this, current models are constrained by limitations in handling long texts and reinforcement learning (RL) training efficiency. To address these issues, we propose a simple yet effective test-time scaling approach Multi-round Thinking. This method iteratively refines model reasoning by leveraging previous answers as prompts for subsequent rounds. Extensive experiments across multiple models, including QwQ-32B and DeepSeek-R1, consistently show performance improvements on various benchmarks such as AIME 2024, MATH-500, GPQA-diamond, and LiveCodeBench. For instance, the accuracy of QwQ-32B improved from 80.3% (Round 1) to 82.1% (Round 2) on the AIME 2024 dataset, while DeepSeek-R1 showed a similar increase from 79.7% to 82.0%. These results confirm that Multi-round Thinking is a broadly applicable, straightforward approach to achieving stable enhancements in model performance, underscoring its potential for future developments in test-time scaling techniques. The key prompt: {Original question prompt} The assistant's previous answer is: {last round answer} , and please re-answer.

cs.CL↗

arXiv2025

Enhancing Multimodal Sentiment Analysis for Missing Modality through Self-Distillation and Unified Modality Cross-Attention

In multimodal sentiment analysis, collecting text data is often more challenging than video or audio due to higher annotation costs and inconsistent automatic speech recognition (ASR) quality. To address this challenge, our study has developed a robust model that effectively integrates multimodal sentiment information, even in the absence of text modality. Specifically, we have developed a Double-Flow Self-Distillation Framework, including Unified Modality Cross-Attention (UMCA) and Modality Imagination Autoencoder (MIA), which excels at processing both scenarios with complete modalities and those with missing text modality. In detail, when the text modality is missing, our framework uses the LLM-based model to simulate the text representation from the audio modality, while the MIA module supplements information from the other two modalities to make the simulated text representation similar to the real text representation. To further align the simulated and real representations, and to enable the model to capture the continuous nature of sample orders in sentiment valence regression tasks, we have also introduced the Rank-N Contrast (RNC) loss function. When testing on the CMU-MOSEI, our model achieved outstanding performance on MAE and significantly outperformed other models when text modality is missing. The code is available at: https://github.com/WarmCongee/SDUMC

cs.CL↗

arXiv2025

Task-Oriented Semantic Communication for Stereo-Vision 3D Object Detection

With the development of computer vision, 3D object detection has become increasingly important in many real-world applications. Limited by the computing power of sensor-side hardware, the detection task is sometimes deployed on remote computing devices or the cloud to execute complex algorithms, which brings massive data transmission overhead. In response, this paper proposes an optical flow-driven semantic communication framework for the stereo-vision 3D object detection task. The proposed framework fully exploits the dependence of stereo-vision 3D detection on semantic information in images and prioritizes the transmission of this semantic information to reduce total transmission data sizes while ensuring the detection accuracy. Specifically, we develop an optical flow-driven module to jointly extract and recover semantics from the left and right images to reduce the loss of the left-right photometric alignment semantic information and improve the accuracy of depth inference. Then, we design a 2D semantic extraction module to identify and extract semantic meaning around the objects to enhance the transmission of semantic information in the key areas. Finally, a fusion network is used to fuse the recovered semantics, and reconstruct the stereo-vision images for 3D detection. Simulation results show that the proposed method improves the detection accuracy by nearly 70% and outperforms the traditional method, especially for the low signal-to-noise ratio regime.

eess.IV↗

arXiv2024

EmotiveTalk: Expressive Talking Head Generation through Audio Information Decoupling and Emotional Video Diffusion

Diffusion models have revolutionized the field of talking head generation, yet still face challenges in expressiveness, controllability, and stability in long-time generation. In this research, we propose an EmotiveTalk framework to address these issues. Firstly, to realize better control over the generation of lip movement and facial expression, a Vision-guided Audio Information Decoupling (V-AID) approach is designed to generate audio-based decoupled representations aligned with lip movements and expression. Specifically, to achieve alignment between audio and facial expression representation spaces, we present a Diffusion-based Co-speech Temporal Expansion (Di-CTE) module within V-AID to generate expression-related representations under multi-source emotion condition constraints. Then we propose a well-designed Emotional Talking Head Diffusion (ETHD) backbone to efficiently generate highly expressive talking head videos, which contains an Expression Decoupling Injection (EDI) module to automatically decouple the expressions from reference portraits while integrating the target expression information, achieving more expressive generation performance. Experimental results show that EmotiveTalk can generate expressive talking head videos, ensuring the promised controllability of emotions and stability during long-time generation, yielding state-of-the-art performance compared to existing methods.

cs.CV↗

arXiv2024

Distinguishing super-Nyquist frequencies via their temporal variation in $γ$ Doradus stars from continuous photometry

As mixed with real pulsations, the reflection of super-Nyquist frequencies (SNFs) pose a threat to asteroseismic properties. Although SNFs have been studied in several pulsating stars, a systematic survey remains scarcely explored. Here we propose a method to identify SNFs from Kepler and TESS photometry by characterizing their periodic frequency modulations using a sliding Fourier transform. After analyzing long cadence photometry in the Kepler legacy, we have identified 304 SNFs in 56 stars from 45607 frequencies in $\sim600$ $γ$ Doradus stars, corresponding to a fraction of approximately $0.67\%$ and $9.2\%$, respectively. Most SNFs are detected in the frequency range of pressure mode over 120 $μ$Hz and the fraction of SNF detection increases as frequency up to $\sim7\%$. We barely found two potential SNFs mixed with gravity modes in two $γ$ Doradus stars. These findings indicate that SNFs have a negligible impact on the global seismic properties, such as those derived from period spacing in $γ$ Doradus stars. However, we stress that SNFs must be carefully and systematically examined by this method in other pulsating stars, particularly $δ$ Scuti and hot B subdwarf stars, to establish a solid foundation for precise asteroseismolgy of various types of pulsators.

astro-ph.SR↗

arXiv2024

A Survey on Hallucination in Large Language Models: Principles, Taxonomy, Challenges, and Open Questions

The emergence of large language models (LLMs) has marked a significant breakthrough in natural language processing (NLP), fueling a paradigm shift in information acquisition. Nevertheless, LLMs are prone to hallucination, generating plausible yet nonfactual content. This phenomenon raises significant concerns over the reliability of LLMs in real-world information retrieval (IR) systems and has attracted intensive research to detect and mitigate such hallucinations. Given the open-ended general-purpose attributes inherent to LLMs, LLM hallucinations present distinct challenges that diverge from prior task-specific models. This divergence highlights the urgency for a nuanced understanding and comprehensive overview of recent advances in LLM hallucinations. In this survey, we begin with an innovative taxonomy of hallucination in the era of LLM and then delve into the factors contributing to hallucinations. Subsequently, we present a thorough overview of hallucination detection methods and benchmarks. Our discussion then transfers to representative methodologies for mitigating LLM hallucinations. Additionally, we delve into the current limitations faced by retrieval-augmented LLMs in combating hallucinations, offering insights for developing more robust IR systems. Finally, we highlight the promising research directions on LLM hallucinations, including hallucination in large vision-language models and understanding of knowledge boundaries in LLM hallucinations.

cs.CL↗

arXiv2024

AddrLLM: Address Rewriting via Large Language Model on Nationwide Logistics Data

Textual description of a physical location, commonly known as an address, plays an important role in location-based services(LBS) such as on-demand delivery and navigation. However, the prevalence of abnormal addresses, those containing inaccuracies that fail to pinpoint a location, have led to significant costs. Address rewriting has emerged as a solution to rectify these abnormal addresses. Despite the critical need, existing address rewriting methods are limited, typically tailored to correct specific error types, or frequently require retraining to process new address data effectively. In this study, we introduce AddrLLM, an innovative framework for address rewriting that is built upon a retrieval augmented large language model. AddrLLM overcomes aforementioned limitations through a meticulously designed Supervised Fine-Tuning module, an Address-centric Retrieval Augmented Generation module and a Bias-free Objective Alignment module. To the best of our knowledge, this study pioneers the application of LLM-based address rewriting approach to solve the issue of abnormal addresses. Through comprehensive offline testing with real-world data on a national scale and subsequent online deployment, AddrLLM has demonstrated superior performance in integration with existing logistics system. It has significantly decreased the rate of parcel re-routing by approximately 43\%, underscoring its exceptional efficacy in real-world applications.

cs.CL↗

arXiv2024

Scale Propagation Network for Generalizable Depth Completion

Depth completion, inferring dense depth maps from sparse measurements, is crucial for robust 3D perception. Although deep learning based methods have made tremendous progress in this problem, these models cannot generalize well across different scenes that are unobserved in training, posing a fundamental limitation that yet to be overcome. A careful analysis of existing deep neural network architectures for depth completion, which are largely borrowing from successful backbones for image analysis tasks, reveals that a key design bottleneck actually resides in the conventional normalization layers. These normalization layers are designed, on one hand, to make training more stable, on the other hand, to build more visual invariance across scene scales. However, in depth completion, the scale is actually what we want to robustly estimate in order to better generalize to unseen scenes. To mitigate, we propose a novel scale propagation normalization (SP-Norm) method to propagate scales from input to output, and simultaneously preserve the normalization operator for easy convergence. More specifically, we rescale the input using learned features of a single-layer perceptron from the normalized input, rather than directly normalizing the input as conventional normalization layers. We then develop a new network architecture based on SP-Norm and the ConvNeXt V2 backbone. We explore the composition of various basic blocks and architectures to achieve superior performance and efficient inference for generalizable depth completion. Extensive experiments are conducted on six unseen datasets with various types of sparse depth maps, i.e., randomly sampled 0.1\%/1\%/10\% valid pixels, 4/8/16/32/64-line LiDAR points, and holes from Structured-Light. Our model consistently achieves the best accuracy with faster speed and lower memory when compared to state-of-the-art methods.

cs.CV↗

arXiv2024

Trends in Integration of Knowledge and Large Language Models: A Survey and Taxonomy of Methods, Benchmarks, and Applications

Large language models (LLMs) exhibit superior performance on various natural language tasks, but they are susceptible to issues stemming from outdated data and domain-specific limitations. In order to address these challenges, researchers have pursued two primary strategies, knowledge editing and retrieval augmentation, to enhance LLMs by incorporating external information from different aspects. Nevertheless, there is still a notable absence of a comprehensive survey. In this paper, we propose a review to discuss the trends in integration of knowledge and large language models, including taxonomy of methods, benchmarks, and applications. In addition, we conduct an in-depth analysis of different methods and point out potential research directions in the future. We hope this survey offers the community quick access and a comprehensive overview of this research area, with the intention of inspiring future research endeavors.

cs.CL↗

arXiv2024

HypomimiaCoach: An AU-based Digital Therapy System for Hypomimia Detection & Rehabilitation with Parkinson's Disease

Hypomimia is a non-motor symptom of Parkinson's disease that manifests as delayed facial movements and expressions, along with challenges in articulation and emotion. Currently, subjective evaluation by neurologists is the primary method for hypomimia detection, and conventional rehabilitation approaches heavily rely on verbal prompts from rehabilitation physicians. There remains a deficiency in accessible, user-friendly and scientifically rigorous assistive tools for hypomimia treatments. To investigate this, we developed HypomimaCoach, an Action Unit (AU)-based digital therapy system for hypomimia detection and rehabilitation in Parkinson's disease. The HypomimaCoach system was designed to facilitate engagement through the incorporation of both relaxed and controlled rehabilitation exercises, while also stimulating initiative through the integration of digital therapies that incorporated traditional face training methods. We extract action unit(AU) features and their relationship for hypomimia detection. In order to facilitate rehabilitation, a series of training programmes have been devised based on the Action Units (AUs) and patients are provided with real-time feedback through an additional AU recognition model, which guides them through their training routines. A pilot study was conducted with seven participants in China, all of whom exhibited symptoms of Parkinson's disease hypomimia. The results of the pilot study demonstrated a positive impact on participants' self-efficacy, with favourable feedback received. Furthermore, physician evaluations validated the system's applicability in a therapeutic setting for patients with Parkinson's disease, as well as its potential value in clinical applications.

cs.HC↗

arXiv2024

Learning to Break: Knowledge-Enhanced Reasoning in Multi-Agent Debate System

Multi-agent debate system (MAD) imitating the process of human discussion in pursuit of truth, aims to align the correct cognition of different agents for the optimal solution. It is challenging to make various agents perform right and highly consistent cognition due to their limited and different knowledge backgrounds (i.e., cognitive islands), which hinders the search for the optimal solution. To address the challenge, we propose a novel \underline{M}ulti-\underline{A}gent \underline{D}ebate with \underline{K}nowledge-\underline{E}nhanced framework (\textbf{MADKE}) to promote the system to find the solution. First, we involve a shared retrieval knowledge pool in the debate process to solve the problem of limited and different knowledge backgrounds. Then, we propose an adaptive knowledge selection method to guarantee the accuracy and personalization of knowledge. This method allows agents to choose whether to use external knowledge in each conversation round according to their own needs. Our experimental results on six datasets show that our method achieves state-of-the-art results compared to existing single-agent and multi-agent methods. Further analysis reveals that the introduction of retrieval knowledge can help the agent to break cognitive islands in the debate process and effectively improve the consistency and correctness of the model. Moreover, MADKE using Qwen1.5-72B-Chat surpasses GPT-4 by +1.26\% on average in six datasets, which validates that our method can help open-source LLMs achieve or even surpass the performance of GPT-4. Our code is available at \url{https://github.com/FutureForMe/MADKE}.

cs.CL↗

arXiv2024

An Information Bottleneck Perspective for Effective Noise Filtering on Retrieval-Augmented Generation

Retrieval-augmented generation integrates the capabilities of large language models with relevant information retrieved from an extensive corpus, yet encounters challenges when confronted with real-world noisy data. One recent solution is to train a filter module to find relevant content but only achieve suboptimal noise compression. In this paper, we propose to introduce the information bottleneck theory into retrieval-augmented generation. Our approach involves the filtration of noise by simultaneously maximizing the mutual information between compression and ground output, while minimizing the mutual information between compression and retrieved passage. In addition, we derive the formula of information bottleneck to facilitate its application in novel comprehensive evaluations, the selection of supervised fine-tuning data, and the construction of reinforcement learning rewards. Experimental results demonstrate that our approach achieves significant improvements across various question answering datasets, not only in terms of the correctness of answer generation but also in the conciseness with $2.5\%$ compression rate.

cs.CL↗

arXiv2024

TimeBench: A Comprehensive Evaluation of Temporal Reasoning Abilities in Large Language Models

Grasping the concept of time is a fundamental facet of human cognition, indispensable for truly comprehending the intricacies of the world. Previous studies typically focus on specific aspects of time, lacking a comprehensive temporal reasoning benchmark. To address this, we propose TimeBench, a comprehensive hierarchical temporal reasoning benchmark that covers a broad spectrum of temporal reasoning phenomena. TimeBench provides a thorough evaluation for investigating the temporal reasoning capabilities of large language models. We conduct extensive experiments on GPT-4, LLaMA2, and other popular LLMs under various settings. Our experimental results indicate a significant performance gap between the state-of-the-art LLMs and humans, highlighting that there is still a considerable distance to cover in temporal reasoning. Besides, LLMs exhibit capability discrepancies across different reasoning categories. Furthermore, we thoroughly analyze the impact of multiple aspects on temporal reasoning and emphasize the associated challenges. We aspire for TimeBench to serve as a comprehensive benchmark, fostering research in temporal reasoning. Resources are available at: https://github.com/zchuz/TimeBench

cs.CL↗

arXiv2024

BeamAggR: Beam Aggregation Reasoning over Multi-source Knowledge for Multi-hop Question Answering

Large language models (LLMs) have demonstrated strong reasoning capabilities. Nevertheless, they still suffer from factual errors when tackling knowledge-intensive tasks. Retrieval-augmented reasoning represents a promising approach. However, significant challenges still persist, including inaccurate and insufficient retrieval for complex questions, as well as difficulty in integrating multi-source knowledge. To address this, we propose Beam Aggregation Reasoning, BeamAggR, a reasoning framework for knowledge-intensive multi-hop QA. BeamAggR explores and prioritizes promising answers at each hop of question. Concretely, we parse the complex questions into trees, which include atom and composite questions, followed by bottom-up reasoning. For atomic questions, the LLM conducts reasoning on multi-source knowledge to get answer candidates. For composite questions, the LLM combines beam candidates, explores multiple reasoning paths through probabilistic aggregation, and prioritizes the most promising trajectory. Extensive experiments on four open-domain multi-hop reasoning datasets show that our method significantly outperforms SOTA methods by 8.5%. Furthermore, our analysis reveals that BeamAggR elicits better knowledge collaboration and answer aggregation.

cs.CL↗

arXiv2024

Navigate through Enigmatic Labyrinth A Survey of Chain of Thought Reasoning: Advances, Frontiers and Future

Reasoning, a fundamental cognitive process integral to human intelligence, has garnered substantial interest within artificial intelligence. Notably, recent studies have revealed that chain-of-thought prompting significantly enhances LLM's reasoning capabilities, which attracts widespread attention from both academics and industry. In this paper, we systematically investigate relevant research, summarizing advanced methods through a meticulous taxonomy that offers novel perspectives. Moreover, we delve into the current frontiers and delineate the challenges and future directions, thereby shedding light on future research. Furthermore, we engage in a discussion about open questions. We hope this paper serves as an introduction for beginners and fosters future research. Resources have been made publicly available at https://github.com/zchuz/CoT-Reasoning-Survey

cs.CL↗

arXiv2024

Asteroseismological analysis of the non-Blazhko RRab star EPIC~248846335 in LAMOST -- Kepler$/$ K2 project

We conduct an asteroseismological analysis on the non-Blazhko ab-type RR Lyrae star EPIC 248846335 employing the Radial Stellar Pulsations (RSP) module of the Modules for Experiments in Stellar Astrophysics (MESA) based on the set of stellar parameters. The atmospheric parameters as $T_\mathrm{eff}$ = 6933$\pm$70 $K$, log $g$ = 3.35$\pm$ 0.50 and [Fe/H] = -1.18 $\pm$ 0.14 are estimated from the Low-Resolution Spectra of LAMOST DR9. The luminosity $L$ = 49.70$_{-1.80}^{+2.99}$ $L_\odot$ and mass M = 0.56 $\pm$ 0.07 $M_\odot$ are calculated, respectively, using the distance provided by Gaia and the metallicity estimated from the Low-Resolution Spectra. The Fourier parameters of the light curves observed by $K2$ and RV curves determined from the Medium-Resolution Spectra of LAMOST DR10 are also calculated in this work. The period of the fundamental mode of the star and the residuals $r$ of the Fourier parameters between the models and observations serve to select optimal model, whose stellar parameters are $T_\mathrm{eff}$ = 6700 $\pm$ 220 K, log $g$ = 2.70, [Fe/H] = -1.20 $\pm$ 0.2, M = 0.59 $\pm$ 0.05 $M_\odot$, and $L$ = 56.0 $\pm$ 4.2 $L_\odot$. The projection factors are constrained as 1.20 $\pm$ 0.02 and 1.59 $\pm$ 0.13 by the blue- and red-arm observed velocities with their corresponding RV curves derived from the best-fit model, respectively. The precise determination of stellar parameters in ab-type RR Lyrae stars is crucial for understanding the physical processes that occur during pulsation and for providing a deeper understanding of its Period-Luminosity relationship.

astro-ph.SR↗