arXiv subjects

Xu Han

Publications and source records attributed to Xu Han.

At least 145 records · Page 8Linked to original sources

arXiv2025

MoST: Efficient Monarch Sparse Tuning for 3D Representation Learning

We introduce Monarch Sparse Tuning (MoST), the first reparameterization-based parameter-efficient fine-tuning (PEFT) method tailored for 3D representation learning. Unlike existing adapter-based and prompt-tuning 3D PEFT methods, MoST introduces no additional inference overhead and is compatible with many 3D representation learning backbones. At its core, we present a new family of structured matrices for 3D point clouds, Point Monarch, which can capture local geometric features of irregular points while offering high expressiveness. MoST reparameterizes the dense update weight matrices as our sparse Point Monarch matrices, significantly reducing parameters while retaining strong performance. Experiments on various backbones show that MoST is simple, effective, and highly generalizable. It captures local features in point clouds, achieving state-of-the-art results on multiple benchmarks, e.g., 97.5% acc. on ScanObjectNN (PB_50_RS) and 96.2% on ModelNet40 classification, while it can also combine with other matrix decompositions (e.g., Low-rank, Kronecker) to further reduce parameters.

cs.CV↗

arXiv2025

Traffic Regulation-aware Path Planning with Regulation Databases and Vision-Language Models

This paper introduces and tests a framework integrating traffic regulation compliance into automated driving systems (ADS). The framework enables ADS to follow traffic laws and make informed decisions based on the driving environment. Using RGB camera inputs and a vision-language model (VLM), the system generates descriptive text to support a regulation-aware decision-making process, ensuring legal and safe driving practices. This information is combined with a machine-readable ADS regulation database to guide future driving plans within legal constraints. Key features include: 1) a regulation database supporting ADS decision-making, 2) an automated process using sensor input for regulation-aware path planning, and 3) validation in both simulated and real-world environments. Particularly, the real-world vehicle tests not only assess the framework's performance but also evaluate the potential and challenges of VLMs to solve complex driving problems by integrating detection, reasoning, and planning. This work enhances the legality, safety, and public trust in ADS, representing a significant step forward in the field.

cs.RO↗

arXiv2025

FR-Spec: Accelerating Large-Vocabulary Language Models via Frequency-Ranked Speculative Sampling

Speculative sampling has emerged as an important technique for accelerating the auto-regressive generation process of large language models (LLMs) by utilizing a draft-then-verify mechanism to produce multiple tokens per forward pass. While state-of-the-art speculative sampling methods use only a single layer and a language modeling (LM) head as the draft model to achieve impressive layer compression, their efficiency gains are substantially reduced for large-vocabulary LLMs, such as Llama-3-8B with a vocabulary of 128k tokens. To address this, we present FR-Spec, a frequency-ranked speculative sampling framework that optimizes draft candidate selection through vocabulary space compression. By constraining the draft search to a frequency-prioritized token subset, our method reduces LM Head computation overhead by 75% while ensuring the equivalence of the final output distribution. Experiments across multiple datasets demonstrate an average of 1.12$\times$ speedup over the state-of-the-art speculative sampling method EAGLE-2. Code available at https://github.com/thunlp/FR-Spec.

cs.CL↗

arXiv2025

RAGEval: Scenario Specific RAG Evaluation Dataset Generation Framework

Retrieval-Augmented Generation (RAG) is a powerful approach that enables large language models (LLMs) to incorporate external knowledge. However, evaluating the effectiveness of RAG systems in specialized scenarios remains challenging due to the high costs of data construction and the lack of suitable evaluation metrics. This paper introduces RAGEval, a framework designed to assess RAG systems across diverse scenarios by generating high-quality documents, questions, answers, and references through a schema-based pipeline. With a focus on factual accuracy, we propose three novel metrics: Completeness, Hallucination, and Irrelevance to evaluate LLM generated responses rigorously. Experimental results show that RAGEval outperforms zero-shot and one-shot methods in terms of clarity, safety, conformity, and richness of generated samples. Furthermore, the use of LLMs for scoring the proposed metrics demonstrates a high level of consistency with human evaluations. RAGEval establishes a new paradigm for evaluating RAG systems in real-world applications. The code and dataset are released at https://github.com/OpenBMB/RAGEval.

cs.CL↗

arXiv2025

VisRAG: Vision-based Retrieval-augmented Generation on Multi-modality Documents

Retrieval-augmented generation (RAG) is an effective technique that enables large language models (LLMs) to utilize external knowledge sources for generation. However, current RAG systems are solely based on text, rendering it impossible to utilize vision information like layout and images that play crucial roles in real-world multi-modality documents. In this paper, we introduce VisRAG, which tackles this issue by establishing a vision-language model (VLM)-based RAG pipeline. In this pipeline, instead of first parsing the document to obtain text, the document is directly embedded using a VLM as an image and then retrieved to enhance the generation of a VLM. Compared to traditional text-based RAG, VisRAG maximizes the retention and utilization of the data information in the original documents, eliminating the information loss introduced during the parsing process. We collect both open-source and synthetic data to train the retriever in VisRAG and explore a variety of generation methods. Experiments demonstrate that VisRAG outperforms traditional RAG in both the retrieval and generation stages, achieving a 20--40% end-to-end performance gain over traditional text-based RAG pipeline. Further analysis reveals that VisRAG is efficient in utilizing training data and demonstrates strong generalization capability, positioning it as a promising solution for RAG on multi-modality documents. Our code and data are available at https://github.com/openbmb/visrag.

cs.IR↗

arXiv2025

TritonBench: Benchmarking Large Language Model Capabilities for Generating Triton Operators

Triton, a high-level Python-like language designed for building efficient GPU kernels, is widely adopted in deep learning frameworks due to its portability, flexibility, and accessibility. However, programming and parallel optimization still require considerable trial and error from Triton developers. Despite advances in large language models (LLMs) for conventional code generation, these models struggle to generate accurate, performance-optimized Triton code, as they lack awareness of its specifications and the complexities of GPU programming. More critically, there is an urgent need for systematic evaluations tailored to Triton. In this work, we introduce TritonBench, the first comprehensive benchmark for Triton operator generation. TritonBench features two evaluation channels: a curated set of 184 real-world operators from GitHub and a collection of operators aligned with PyTorch interfaces. Unlike conventional code benchmarks prioritizing functional correctness, TritonBench also profiles efficiency performance on widely deployed GPUs aligned with industry applications. Our study reveals that current state-of-the-art code LLMs struggle to generate efficient Triton operators, highlighting a significant gap in high-performance code generation. TritonBench will be available at https://github.com/thunlp/TritonBench.

cs.CL↗

arXiv2025

LimSim Series: An Autonomous Driving Simulation Platform for Validation and Enhancement

Closed-loop simulation environments play a crucial role in the validation and enhancement of autonomous driving systems (ADS). However, certain challenges warrant significant attention, including balancing simulation accuracy with duration, reconciling functionality with practicality, and establishing comprehensive evaluation mechanisms. This paper addresses these challenges by introducing the LimSim Series, a comprehensive simulation platform designed to support the rapid deployment and efficient iteration of ADS. The LimSim Series integrates multi-type information from road networks, employs human-like decision-making and planning algorithms for background vehicles, and introduces the concept of the Area of Interest (AoI) to optimize computational resources. The platform offers a variety of baseline algorithms and user-friendly interfaces, facilitating flexible validation of multiple technical pipelines. Additionally, the LimSim Series incorporates multi-dimensional evaluation metrics, delivering thorough insights into system performance, thus enabling researchers to promptly identify issues for further improvements. Experiments demonstrate that the LimSim Series is compatible with modular, end-to-end, and VLM-based knowledge-driven systems. It can assist in the iteration and updating of ADS by evaluating performance across various scenarios. The code of the LimSim Series is released at: https://github.com/PJLab-ADG/LimSim.

cs.RO↗

arXiv2025

Locret: Enhancing Eviction in Long-Context LLM Inference with Trained Retaining Heads on Consumer-Grade Devices

Scaling the input context length of a large language model (LLM) incurs a significant increase in computation cost and memory footprint to maintain the attention key-value (KV) cache. Existing KV cache compression methods suffer from inefficient compression strategies and limited memory reduction effects, making it difficult for LLMs to conduct long-context inference on consumer-grade devices, especially when inferring long-context stream input. Such obstacles prevent consumer-grade devices from supporting more complex applications, creating challenges for the democratization of LLMs. To overcome this, we propose Locret, the first framework to create an eviction policy compatible with chunked prefill. By evaluating the causal importance of KV cache units by learnable retaining heads, Locret enables precise eviction of cache units, facilitating efficient long-context inference. In our extensive empirical studies, Locret outperforms the recent popular and competitive approaches in terms of memory efficiency and generation quality -- Locret achieves up to 20x of KV cache compression ratio within less than 10% performance loss. Furthermore, Locret achieves 128K+ long-context inference on a single NVIDIA 4090 GPU without compromising generation quality and only costs <1 GPU hour of additional training.

cs.CL↗

arXiv2025

LearningFlow: Automated Policy Learning Workflow for Urban Driving with Large Language Models

Recent advancements in reinforcement learning (RL) demonstrate the significant potential in autonomous driving. Despite this promise, challenges such as the manual design of reward functions and low sample efficiency in complex environments continue to impede the development of safe and effective driving policies. To tackle these issues, we introduce LearningFlow, an innovative automated policy learning workflow tailored to urban driving. This framework leverages the collaboration of multiple large language model (LLM) agents throughout the RL training process. LearningFlow includes a curriculum sequence generation process and a reward generation process, which work in tandem to guide the RL policy by generating tailored training curricula and reward functions. Particularly, each process is supported by an analysis agent that evaluates training progress and provides critical insights to the generation agent. Through the collaborative efforts of these LLM agents, LearningFlow automates policy learning across a series of complex driving tasks, and it significantly reduces the reliance on manual reward function design while enhancing sample efficiency. Comprehensive experiments are conducted in the high-fidelity CARLA simulator, along with comparisons with other existing methods, to demonstrate the efficacy of our proposed approach. The results demonstrate that LearningFlow excels in generating rewards and curricula. It also achieves superior performance and robust generalization across various driving tasks, as well as commendable adaptation to different RL algorithms.

cs.RO↗

arXiv2025

ProSparse: Introducing and Enhancing Intrinsic Activation Sparsity within Large Language Models

Activation sparsity refers to the existence of considerable weakly-contributed elements among activation outputs. As a prevalent property of the models using the ReLU activation function, activation sparsity has been proven a promising paradigm to boost model inference efficiency. Nevertheless, most large language models (LLMs) adopt activation functions without intrinsic activation sparsity (e.g., GELU and Swish). Some recent efforts have explored introducing ReLU or its variants as the substitutive activation function to help LLMs achieve activation sparsity and inference acceleration, but few can simultaneously obtain high sparsity and comparable model performance. This paper introduces a simple and effective sparsification method named "ProSparse" to push LLMs for higher activation sparsity while maintaining comparable performance. Specifically, after substituting the activation function of LLMs with ReLU, ProSparse adopts progressive sparsity regularization with a factor smoothly increasing along the multi-stage sine curves. This can enhance activation sparsity and mitigate performance degradation by avoiding radical shifts in activation distributions. With ProSparse, we obtain high sparsity of 89.32% for LLaMA2-7B, 88.80% for LLaMA2-13B, and 87.89% for end-size MiniCPM-1B, respectively, achieving comparable performance to their original Swish-activated versions. These present the most sparsely activated models among open-source LLaMA versions and competitive end-size models, considerably surpassing ReluLLaMA-7B (66.98%) and ReluLLaMA-13B (71.56%). Our inference acceleration experiments further demonstrate the significant practical acceleration potential of LLMs with higher activation sparsity, obtaining up to 4.52$\times$ inference speedup.

cs.LG↗

arXiv2025

Fair Text to Medical Image Diffusion Model with Subgroup Distribution Aligned Tuning

The text to medical image (T2MedI) with latent diffusion model has great potential to alleviate the scarcity of medical imaging data and explore the underlying appearance distribution of lesions in a specific patient status description. However, as the text to nature image models, we show that the T2MedI model can also bias to some subgroups to overlook the minority ones in the training set. In this work, we first build a T2MedI model based on the pre-trained Imagen model, which has the fixed contrastive language-image pre-training (CLIP) text encoder, while its decoder has been fine-tuned on medical images from the Radiology Objects in COntext (ROCO) dataset. Its gender bias is analyzed qualitatively and quantitatively. Toward this issue, we propose to fine-tune the T2MedI toward the target application dataset to align their sensitive subgroups distribution probability. Specifically, the alignment loss for fine-tuning is guided by an off-the-shelf sensitivity-subgroup classifier to match the classification probability between the generated images and the expected target dataset. In addition, the image quality is maintained by a CLIP-consistency regularization term following a knowledge distillation scheme. For evaluation, we set the target dataset to be enhanced as the BraST18 dataset, and trained a brain magnetic resonance (MR) slice-based gender classifier from it. With our method, the generated MR image can markedly reduce the inconsistency with the gender proportion in the BraTS18 dataset.

cs.CV↗

arXiv2025

Light-weight Fine-tuning Method for Defending Adversarial Noise in Pre-trained Medical Vision-Language Models

Fine-tuning pre-trained Vision-Language Models (VLMs) has shown remarkable capabilities in medical image and textual depiction synergy. Nevertheless, many pre-training datasets are restricted by patient privacy concerns, potentially containing noise that can adversely affect downstream performance. Moreover, the growing reliance on multi-modal generation exacerbates this issue because of its susceptibility to adversarial attacks. To investigate how VLMs trained on adversarial noisy data perform on downstream medical tasks, we first craft noisy upstream datasets using multi-modal adversarial attacks. Through our comprehensive analysis, we unveil that moderate noise enhances model robustness and transferability, but increasing noise levels negatively impact downstream task performance. To mitigate this issue, we propose rectify adversarial noise (RAN) framework, a recipe designed to effectively defend adversarial attacks and rectify the influence of upstream noise during fine-tuning.

cs.CV↗

arXiv2024

Cryogenic hybrid magnonic circuits based on spalled YIG thin films

Yttrium iron garnet (YIG) magnonics has garnered significant research interest because of the unique properties of magnons (quasiparticles of collective spin excitation) for signal processing. In particular, hybrid systems based on YIG magnonics show great promise for quantum information science due to their broad frequency tunability and strong compatibility with other platforms. However, their broad applications have been severely constrained by substantial microwave loss in the gadolinium gallium garnet (GGG) substrate at cryogenic temperatures. In this study, we demonstrate that YIG thin films can be spalled from YIG/GGG samples. Our approach is validated by measuring hybrid devices comprising superconducting resonators and spalled YIG films, which exhibits anti-crossing features that indicate strong coupling between magnons and microwave photons. Such new capability of separating YIG thin films from GGG substrates via spalling, and the integrated superconductor-YIG devices represent a significant advancement for integrated magnonic devices, paving the way for advanced magnon-based coherent information processing.

cond-mat.mes-hall↗

arXiv2024

V2X-Real: a Large-Scale Dataset for Vehicle-to-Everything Cooperative Perception

Recent advancements in Vehicle-to-Everything (V2X) technologies have enabled autonomous vehicles to share sensing information to see through occlusions, greatly boosting the perception capability. However, there are no real-world datasets to facilitate the real V2X cooperative perception research -- existing datasets either only support Vehicle-to-Infrastructure cooperation or Vehicle-to-Vehicle cooperation. In this paper, we present V2X-Real, a large-scale dataset that includes a mixture of multiple vehicles and smart infrastructure to facilitate the V2X cooperative perception development with multi-modality sensing data. Our V2X-Real is collected using two connected automated vehicles and two smart infrastructure, which are all equipped with multi-modal sensors including LiDAR sensors and multi-view cameras. The whole dataset contains 33K LiDAR frames and 171K camera data with over 1.2M annotated bounding boxes of 10 categories in very challenging urban scenarios. According to the collaboration mode and ego perspective, we derive four types of datasets for Vehicle-Centric, Infrastructure-Centric, Vehicle-to-Vehicle, and Infrastructure-to-Infrastructure cooperative perception. Comprehensive multi-class multi-agent benchmarks of SOTA cooperative perception methods are provided. The V2X-Real dataset and codebase are available at https://mobility-lab.seas.ucla.edu/v2x-real.

cs.CV↗

arXiv2024

Strong Inter-valley Electron-Phonon Coupling in Magic-Angle Twisted Bilayer Graphene

The unusual properties of superconductivity in magic-angle twisted bilayer graphene (MATBG) have sparked enormous research interest. However, despite the dedication of intensive experimental efforts and the proposal of several possible pairing mechanisms, the origin of its superconductivity remains elusive. Here, utilizing angle-resolved photoemission spectroscopy with micrometer spatial resolution, we have revealed flat band replicas in superconducting MATBG, where MATBG is unaligned with its hexagonal boron nitride (hBN) substrate11. These replicas exhibit uniform energy spacing, approximately 150 +- 15 meV apart, indicative of strong electron-boson coupling. Strikingly, these replicas are absent in non-superconducting twisted bilayer graphene (TBG) systems, either when MATBG is aligned to hBN or when TBG deviates from the magic angle. Calculations suggest that the formation of these flat band replicas in superconducting MATBG are attributed to the strong coupling between flat band electrons and an optical phonon mode at the graphene K point, facilitated by inter-valley scattering. These findings, although do not necessarily put electron phonon coupling as the main driving force for the superconductivity in MATBG, unravel the unique electronic structure inherent in superconducting MATBG, thereby providing crucial information for understanding the unusual electronic landscape from which the superconductivity is derived.

cond-mat.mes-hall↗

arXiv2024

Densing Law of LLMs

Large Language Models (LLMs) have emerged as a milestone in artificial intelligence, and their performance can improve as the model size increases. However, this scaling brings great challenges to training and inference efficiency, particularly for deploying LLMs in resource-constrained environments, and the scaling trend is becoming increasingly unsustainable. This paper introduces the concept of ``\textit{capacity density}'' as a new metric to evaluate the quality of the LLMs across different scales and describes the trend of LLMs in terms of both effectiveness and efficiency. To calculate the capacity density of a given target LLM, we first introduce a set of reference models and develop a scaling law to predict the downstream performance of these reference models based on their parameter sizes. We then define the \textit{effective parameter size} of the target LLM as the parameter size required by a reference model to achieve equivalent performance, and formalize the capacity density as the ratio of the effective parameter size to the actual parameter size of the target LLM. Capacity density provides a unified framework for assessing both model effectiveness and efficiency. Our further analysis of recent open-source base LLMs reveals an empirical law (the densing law)that the capacity density of LLMs grows exponentially over time. More specifically, using some widely used benchmarks for evaluation, the capacity density of LLMs doubles approximately every three months. The law provides new perspectives to guide future LLM development, emphasizing the importance of improving capacity density to achieve optimal results with minimal computational overhead.

cs.AI↗

arXiv2024

OneBit: Towards Extremely Low-bit Large Language Models

Model quantification uses low bit-width values to represent the weight matrices of existing models to be quantized, which is a promising approach to reduce both storage and computational overheads of deploying highly anticipated LLMs. However, current quantization methods suffer severe performance degradation when the bit-width is extremely reduced, and thus focus on utilizing 4-bit or 8-bit values to quantize models. This paper boldly quantizes the weight matrices of LLMs to 1-bit, paving the way for the extremely low bit-width deployment of LLMs. For this target, we introduce a 1-bit model compressing framework named OneBit, including a novel 1-bit parameter representation method to better quantize LLMs as well as an effective parameter initialization method based on matrix decomposition to improve the convergence speed of the quantization framework. Sufficient experimental results indicate that OneBit achieves good performance (at least 81% of the non-quantized performance on LLaMA models) with robust training processes when only using 1-bit weight matrices.

cs.CL↗

arXiv2024

Delta-CoMe: Training-Free Delta-Compression with Mixed-Precision for Large Language Models

Fine-tuning is a crucial process for adapting large language models (LLMs) to diverse applications. In certain scenarios, such as multi-tenant serving, deploying multiple LLMs becomes necessary to meet complex demands. Recent studies suggest decomposing a fine-tuned LLM into a base model and corresponding delta weights, which are then compressed using low-rank or low-bit approaches to reduce costs. In this work, we observe that existing low-rank and low-bit compression methods can significantly harm the model performance for task-specific fine-tuned LLMs (e.g., WizardMath for math problems). Motivated by the long-tail distribution of singular values in the delta weights, we propose a delta quantization approach using mixed-precision. This method employs higher-bit representation for singular vectors corresponding to larger singular values. We evaluate our approach on various fine-tuned LLMs, including math LLMs, code LLMs, chat LLMs, and even VLMs. Experimental results demonstrate that our approach performs comparably to full fine-tuned LLMs, surpassing both low-rank and low-bit baselines by a considerable margin. Additionally, we show that our method is compatible with various backbone LLMs, such as Llama-2, Llama-3, and Mistral, highlighting its generalizability.

cs.CL↗