arXiv subjects

Bin Cui

Publications and source records attributed to Bin Cui.

At least 91 records · Page 5Linked to original sources

arXiv2025

CFBench: A Comprehensive Constraints-Following Benchmark for LLMs

The adeptness of Large Language Models (LLMs) in comprehending and following natural language instructions is critical for their deployment in sophisticated real-world applications. Existing evaluations mainly focus on fragmented constraints or narrow scenarios, but they overlook the comprehensiveness and authenticity of constraints from the user's perspective. To bridge this gap, we propose CFBench, a large-scale Comprehensive Constraints Following Benchmark for LLMs, featuring 1,000 curated samples that cover more than 200 real-life scenarios and over 50 NLP tasks. CFBench meticulously compiles constraints from real-world instructions and constructs an innovative systematic framework for constraint types, which includes 10 primary categories and over 25 subcategories, and ensures each constraint is seamlessly integrated within the instructions. To make certain that the evaluation of LLM outputs aligns with user perceptions, we propose an advanced methodology that integrates multi-dimensional assessment criteria with requirement prioritization, covering various perspectives of constraints, instructions, and requirement fulfillment. Evaluating current leading LLMs on CFBench reveals substantial room for improvement in constraints following, and we further investigate influencing factors and enhancement strategies. The data and code are publicly available at https://github.com/PKU-Baichuan-MLSystemLab/CFBench

cs.CL↗

arXiv2025

BMTree: Designing, Learning, and Updating Piecewise Space-Filling Curves for Multi-Dimensional Data Indexing

Space-filling curves (SFC, for short) have been widely applied to index multi-dimensional data, which first maps the data to one dimension, and then a one-dimensional indexing method, e.g., the B-tree indexes the mapped data. Existing SFCs adopt a single mapping scheme for the whole data space. However, a single mapping scheme often does not perform well on all the data space. In this paper, we propose a new type of SFC called piecewise SFCs that adopts different mapping schemes for different data subspaces. Specifically, we propose a data structure termed the Bit Merging tree (BMTree) that can generate data subspaces and their SFCs simultaneously, and achieve desirable properties of the SFC for the whole data space. Furthermore, we develop a reinforcement learning-based solution to build the BMTree, aiming to achieve excellent query performance. To update the BMTree efficiently when the distributions of data and/or queries change, we develop a new mechanism that achieves fast detection of distribution shifts in data and queries, and enables partial retraining of the BMTree. The retraining mechanism achieves performance enhancement efficiently since it avoids retraining the BMTree from scratch. Extensive experiments show the effectiveness and efficiency of the BMTree with the proposed learning-based methods.

cs.DB↗

arXiv2025

Galvatron: An Automatic Distributed System for Efficient Foundation Model Training

Galvatron is a distributed system for efficiently training large-scale Foundation Models. It overcomes the complexities of selecting optimal parallelism strategies by automatically identifying the most efficient hybrid strategy, incorporating data, tensor, pipeline, sharded data, and sequence parallelism, along with recomputation. The system's architecture includes a profiler for hardware and model analysis, a search engine for strategy optimization using decision trees and dynamic programming, and a runtime for executing these strategies efficiently. Benchmarking on various clusters demonstrates Galvatron's superior throughput compared to existing frameworks. This open-source system offers user-friendly interfaces and comprehensive documentation, making complex distributed training accessible and efficient. The source code of Galvatron is available at https://github.com/PKU-DAIR/Hetu-Galvatron.

cs.DC↗

arXiv2025

Towards Scalable and Deep Graph Neural Networks via Noise Masking

In recent years, Graph Neural Networks (GNNs) have achieved remarkable success in many graph mining tasks. However, scaling them to large graphs is challenging due to the high computational and storage costs of repeated feature propagation and non-linear transformation during training. One commonly employed approach to address this challenge is model-simplification, which only executes the Propagation (P) once in the pre-processing, and Combine (C) these receptive fields in different ways and then feed them into a simple model for better performance. Despite their high predictive performance and scalability, these methods still face two limitations. First, existing approaches mainly focus on exploring different C methods from the model perspective, neglecting the crucial problem of performance degradation with increasing P depth from the data-centric perspective, known as the over-smoothing problem. Second, pre-processing overhead takes up most of the end-to-end processing time, especially for large-scale graphs. To address these limitations, we present random walk with noise masking (RMask), a plug-and-play module compatible with the existing model-simplification works. This module enables the exploration of deeper GNNs while preserving their scalability. Unlike the previous model-simplification works, we focus on continuous P and found that the noise existing inside each P is the cause of the over-smoothing issue, and use the efficient masking mechanism to eliminate them. Experimental results on six real-world datasets demonstrate that model-simplification works equipped with RMask yield superior performance compared to their original version and can make a good trade-off between accuracy and efficiency.

cs.LG↗

arXiv2025

DataLab: A Unified Platform for LLM-Powered Business Intelligence

Business intelligence (BI) transforms large volumes of data within modern organizations into actionable insights for informed decision-making. Recently, large language model (LLM)-based agents have streamlined the BI workflow by automatically performing task planning, reasoning, and actions in executable environments based on natural language (NL) queries. However, existing approaches primarily focus on individual BI tasks such as NL2SQL and NL2VIS. The fragmentation of tasks across different data roles and tools lead to inefficiencies and potential errors due to the iterative and collaborative nature of BI. In this paper, we introduce DataLab, a unified BI platform that integrates a one-stop LLM-based agent framework with an augmented computational notebook interface. DataLab supports various BI tasks for different data roles in data preparation, analysis, and visualization by seamlessly combining LLM assistance with user customization within a single environment. To achieve this unification, we design a domain knowledge incorporation module tailored for enterprise-specific BI tasks, an inter-agent communication mechanism to facilitate information sharing across the BI workflow, and a cell-based context management strategy to enhance context utilization efficiency in BI notebooks. Extensive experiments demonstrate that DataLab achieves state-of-the-art performance on various BI tasks across popular research benchmarks. Moreover, DataLab maintains high effectiveness and efficiency on real-world datasets from Tencent, achieving up to a 58.58% increase in accuracy and a 61.65% reduction in token cost on enterprise-specific BI tasks.

cs.DB↗

arXiv2025

PQCache: Product Quantization-based KVCache for Long Context LLM Inference

As the field of Large Language Models (LLMs) continues to evolve, the context length in inference is steadily growing. Key-Value Cache (KVCache), the intermediate representations of tokens within LLM inference, has now become the primary memory bottleneck due to limited GPU memory. Current methods selectively determine suitable keys and values for self-attention computation in LLMs to address the issue. However, they either fall short in maintaining model quality or result in high serving latency. Drawing inspiration from advanced embedding retrieval techniques prevalent in the data management community, we consider the storage and retrieval of KVCache as a typical embedding retrieval problem. We propose PQCache, which employs Product Quantization (PQ) to manage KVCache, maintaining model quality while ensuring low serving latency. During the prefilling phase, we apply PQ to tokens' keys for each LLM layer and head. During the autoregressive decoding phase, we use PQ codes and centroids to approximately identify important preceding tokens, then fetch the corresponding key-value pairs for self-attention computation. Through meticulous design of overlapping and caching, we minimize any additional computation and communication overhead during both phases. Extensive experiments demonstrate that PQCache achieves both effectiveness and efficiency, with 4.60% score improvement over existing methods on InfiniteBench and low system latency in both prefilling and decoding.

cs.CL↗

arXiv2025

Training-free Diffusion Acceleration with Bottleneck Sampling

Diffusion models have demonstrated remarkable capabilities in visual content generation but remain challenging to deploy due to their high computational cost during inference. This computational burden primarily arises from the quadratic complexity of self-attention with respect to image or video resolution. While existing acceleration methods often compromise output quality or necessitate costly retraining, we observe that most diffusion models are pre-trained at lower resolutions, presenting an opportunity to exploit these low-resolution priors for more efficient inference without degrading performance. In this work, we introduce Bottleneck Sampling, a training-free framework that leverages low-resolution priors to reduce computational overhead while preserving output fidelity. Bottleneck Sampling follows a high-low-high denoising workflow: it performs high-resolution denoising in the initial and final stages while operating at lower resolutions in intermediate steps. To mitigate aliasing and blurring artifacts, we further refine the resolution transition points and adaptively shift the denoising timesteps at each stage. We evaluate Bottleneck Sampling on both image and video generation tasks, where extensive experiments demonstrate that it accelerates inference by up to 3$\times$ for image generation and 2.5$\times$ for video generation, all while maintaining output quality comparable to the standard full-resolution sampling process across multiple evaluation metrics.

cs.CV↗

arXiv2025

ReasonFlux: Hierarchical LLM Reasoning via Scaling Thought Templates

We present that hierarchical LLM reasoning via scaling thought templates can effectively optimize the reasoning search space and outperform the mathematical reasoning capabilities of powerful LLMs like OpenAI o1-preview and DeepSeek V3. We train our ReasonFlux-32B model with only 8 GPUs and introduces three innovations: (i) a structured and generic thought template library, containing around 500 high-level thought templates capable of generalizing to similar or relevant reasoning problems; (ii) performing hierarchical reinforcement learning on a sequence of thought templates instead of long CoTs, optimizing a base LLM to plan out an optimal template trajectory for gradually handling complex problems; (iii) a brand new inference scaling system that enables hierarchical LLM reasoning by adaptively scaling thought templates at inference time. With a template trajectory containing more explainable reasoning structures than DeepSeek-R1 and o3-mini, our ReasonFlux-32B significantly advances math reasoning capabilities to state-of-the-art levels. Notably, on the MATH benchmark, it achieves an accuracy of 91.2% and surpasses o1-preview by 6.7%. On the USA Math Olympiad (AIME) benchmark, ReasonFlux-32B solves an average of 56.7% of problems, surpassing o1-preview and DeepSeek-V3 by 27% and 45%, respectively. Code: https://github.com/Gen-Verse/ReasonFlux

cs.CL↗

arXiv2025

Facilitating Multi-turn Function Calling for LLMs via Compositional Instruction Tuning

Large Language Models (LLMs) have exhibited significant potential in performing diverse tasks, including the ability to call functions or use external tools to enhance their performance. While current research on function calling by LLMs primarily focuses on single-turn interactions, this paper addresses the overlooked necessity for LLMs to engage in multi-turn function calling--critical for handling compositional, real-world queries that require planning with functions but not only use functions. To facilitate this, we introduce an approach, BUTTON, which generates synthetic compositional instruction tuning data via bottom-up instruction construction and top-down trajectory generation. In the bottom-up phase, we generate simple atomic tasks based on real-world scenarios and build compositional tasks using heuristic strategies based on atomic tasks. Corresponding function definitions are then synthesized for these compositional tasks. The top-down phase features a multi-agent environment where interactions among simulated humans, assistants, and tools are utilized to gather multi-turn function calling trajectories. This approach ensures task compositionality and allows for effective function and trajectory generation by examining atomic tasks within compositional tasks. We produce a dataset BUTTONInstruct comprising 8k data points and demonstrate its effectiveness through extensive experiments across various LLMs.

cs.CL↗

arXiv2025

Evaluating and Predicting Distorted Human Body Parts for Generated Images

Recent advancements in text-to-image (T2I) models enable high-quality image synthesis, yet generating anatomically accurate human figures remains challenging. AI-generated images frequently exhibit distortions such as proliferated limbs, missing fingers, deformed extremities, or fused body parts. Existing evaluation metrics like Inception Score (IS) and Fréchet Inception Distance (FID) lack the granularity to detect these distortions, while human preference-based metrics focus on abstract quality assessments rather than anatomical fidelity. To address this gap, we establish the first standards for identifying human body distortions in AI-generated images and introduce Distortion-5K, a comprehensive dataset comprising 4,700 annotated images of normal and malformed human figures across diverse styles and distortion types. Based on this dataset, we propose ViT-HD, a Vision Transformer-based model tailored for detecting human body distortions in AI-generated images, which outperforms state-of-the-art segmentation models and visual language models, achieving an F1 score of 0.899 and IoU of 0.831 on distortion localization. Additionally, we construct the Human Distortion Benchmark with 500 human-centric prompts to evaluate four popular T2I models using trained ViT-HD, revealing that nearly 50\% of generated images contain distortions. This work pioneers a systematic approach to evaluating anatomical accuracy in AI-generated humans, offering tools to advance the fidelity of T2I models and their real-world applicability. The Distortion-5K dataset, trained ViT-HD will soon be released in our GitHub repository: \href{https://github.com/TheRoadQaQ/Predicting-Distortion}{https://github.com/TheRoadQaQ/Predicting-Distortion}.

cs.CV↗

arXiv2025

Training-free and Adaptive Sparse Attention for Efficient Long Video Generation

Generating high-fidelity long videos with Diffusion Transformers (DiTs) is often hindered by significant latency, primarily due to the computational demands of attention mechanisms. For instance, generating an 8-second 720p video (110K tokens) with HunyuanVideo takes about 600 PFLOPs, with around 500 PFLOPs consumed by attention computations. To address this issue, we propose AdaSpa, the first Dynamic Pattern and Online Precise Search sparse attention method. Firstly, to realize the Dynamic Pattern, we introduce a blockified pattern to efficiently capture the hierarchical sparsity inherent in DiTs. This is based on our observation that sparse characteristics of DiTs exhibit hierarchical and blockified structures between and within different modalities. This blockified approach significantly reduces the complexity of attention computation while maintaining high fidelity in the generated videos. Secondly, to enable Online Precise Search, we propose the Fused LSE-Cached Search with Head-adaptive Hierarchical Block Sparse Attention. This method is motivated by our finding that DiTs' sparse pattern and LSE vary w.r.t. inputs, layers, and heads, but remain invariant across denoising steps. By leveraging this invariance across denoising steps, it adapts to the dynamic nature of DiTs and allows for precise, real-time identification of sparse indices with minimal overhead. AdaSpa is implemented as an adaptive, plug-and-play solution and can be integrated seamlessly with existing DiTs, requiring neither additional fine-tuning nor a dataset-dependent profiling. Extensive experiments validate that AdaSpa delivers substantial acceleration across various models while preserving video quality, establishing itself as a robust and scalable approach to efficient video generation.

cs.CV↗

arXiv2025

ByteScale: Efficient Scaling of LLM Training with a 2048K Context Length on More Than 12,000 GPUs

Scaling long-context ability is essential for Large Language Models (LLMs). To amortize the memory consumption across multiple devices in long-context training, inter-data partitioning (a.k.a. Data Parallelism) and intra-data partitioning (a.k.a. Context Parallelism) are commonly used. Current training frameworks predominantly treat the two techniques as orthogonal, and establish static communication groups to organize the devices as a static mesh (e.g., a 2D mesh). However, the sequences for LLM training typically vary in lengths, no matter for texts, multi-modalities or reinforcement learning. The mismatch between data heterogeneity and static mesh causes redundant communication and imbalanced computation, degrading the training efficiency. In this work, we introduce ByteScale, an efficient, flexible, and scalable LLM training framework for large-scale mixed training of long and short sequences. The core of ByteScale is a novel parallelism strategy, namely Hybrid Data Parallelism (HDP), which unifies the inter- and intra-data partitioning with a dynamic mesh design. In particular, we build a communication optimizer, which eliminates the redundant communication for short sequences by data-aware sharding and dynamic communication, and further compresses the communication cost for long sequences by selective offloading. Besides, we also develop a balance scheduler to mitigate the imbalanced computation by parallelism-aware data assignment. We evaluate ByteScale with the model sizes ranging from 7B to 141B, context lengths from 256K to 2048K, on a production cluster with more than 12,000 GPUs. Experiment results show that ByteScale outperforms the state-of-the-art training system by up to 7.89x.

cs.DC↗

arXiv2025

SuperCorrect: Advancing Small LLM Reasoning with Thought Template Distillation and Self-Correction

Large language models (LLMs) like GPT-4, DeepSeek-R1, and ReasonFlux have shown significant improvements in various reasoning tasks. However, smaller LLMs still struggle with complex mathematical reasoning because they fail to effectively identify and correct reasoning errors. Recent reflection-based methods aim to address these issues by enabling self-reflection and self-correction, but they still face challenges in independently detecting errors in their reasoning steps. To overcome these limitations, we propose SuperCorrect, a novel two-stage framework that uses a large teacher model to supervise and correct both the reasoning and reflection processes of a smaller student model. In the first stage, we extract hierarchical high-level and detailed thought templates from the teacher model to guide the student model in eliciting more fine-grained reasoning thoughts. In the second stage, we introduce cross-model collaborative direct preference optimization (DPO) to enhance the self-correction abilities of the student model by following the teacher's correction traces during training. This cross-model DPO approach teaches the student model to effectively locate and resolve erroneous thoughts with error-driven insights from the teacher model, breaking the bottleneck of its thoughts and acquiring new skills and knowledge to tackle challenging problems. Extensive experiments consistently demonstrate our superiority over previous methods. Notably, our SuperCorrect-7B model significantly surpasses powerful DeepSeekMath-7B by 7.8%/5.3% and Qwen2.5-Math-7B by 15.1%/6.3% on MATH/GSM8K benchmarks, achieving new SOTA performance among all 7B models. Code: https://github.com/YangLing0818/SuperCorrect-llm

cs.CL↗

arXiv2025

FB-Bench: A Fine-Grained Multi-Task Benchmark for Evaluating LLMs' Responsiveness to Human Feedback

Human feedback is crucial in the interactions between humans and Large Language Models (LLMs). However, existing research primarily focuses on benchmarking LLMs in single-turn dialogues. Even in benchmarks designed for multi-turn dialogues, the user inputs are often independent, neglecting the nuanced and complex nature of human feedback within real-world usage scenarios. To fill this research gap, we introduce FB-Bench, a fine-grained, multi-task benchmark designed to evaluate LLMs' responsiveness to human feedback under real-world usage scenarios in Chinese. Drawing from the two main interaction scenarios, FB-Bench comprises 591 meticulously curated samples, encompassing eight task types, five deficiency types of response, and nine feedback types. We extensively evaluate a broad array of popular LLMs, revealing significant variations in their performance across different interaction scenarios. Further analysis indicates that task, human feedback, and deficiencies of previous responses can also significantly impact LLMs' responsiveness. Our findings underscore both the strengths and limitations of current models, providing valuable insights and directions for future research. Code and datasets are available at https://github.com/PKU-Baichuan-MLSystemLab/FB-Bench.

cs.CL↗

arXiv2025

Diffusion-Sharpening: Fine-tuning Diffusion Models with Denoising Trajectory Sharpening

We propose Diffusion-Sharpening, a fine-tuning approach that enhances downstream alignment by optimizing sampling trajectories. Existing RL-based fine-tuning methods focus on single training timesteps and neglect trajectory-level alignment, while recent sampling trajectory optimization methods incur significant inference NFE costs. Diffusion-Sharpening overcomes this by using a path integral framework to select optimal trajectories during training, leveraging reward feedback, and amortizing inference costs. Our method demonstrates superior training efficiency with faster convergence, and best inference efficiency without requiring additional NFEs. Extensive experiments show that Diffusion-Sharpening outperforms RL-based fine-tuning methods (e.g., Diffusion-DPO) and sampling trajectory optimization methods (e.g., Inference Scaling) across diverse metrics including text alignment, compositional capabilities, and human preferences, offering a scalable and efficient solution for future diffusion model fine-tuning. Code: https://github.com/Gen-Verse/Diffusion-Sharpening

cs.CV↗

arXiv2025

Spindle: Efficient Distributed Training of Multi-Task Large Models via Wavefront Scheduling

Recent foundation models are capable of handling multiple tasks and multiple data modalities with the unified base model structure and several specialized model components. However, efficient training of such multi-task (MT) multi-modal (MM) models poses significant system challenges due to the sophisticated model architecture and the heterogeneous workloads of different tasks and modalities. In this paper, we propose Spindle, a brand new training system tailored for resource-efficient and high-performance training of MT MM models via wavefront scheduling. The key idea of Spindle is to decompose the model execution into waves and address the joint optimization problem sequentially, including both heterogeneity-aware workload parallelization and dependency-driven execution scheduling. We build our system and evaluate it on various MT MM models. Experiments demonstrate the superior performance and efficiency of Spindle, with speedup ratio up to 71% compared to state-of-the-art training systems.

cs.DC↗

arXiv2025

FlexSP: Accelerating Large Language Model Training via Flexible Sequence Parallelism

Extending the context length (i.e., the maximum supported sequence length) of LLMs is of paramount significance. To facilitate long context training of LLMs, sequence parallelism has emerged as an essential technique, which scatters each input sequence across multiple devices and necessitates communication to process the sequence. In essence, existing sequence parallelism methods assume homogeneous sequence lengths (i.e., all input sequences are equal in length) and therefore leverages a single, static scattering strategy for all input sequences. However, in reality, the sequence lengths in LLM training corpora exhibit substantial variability, often following a long-tail distribution, which leads to workload heterogeneity. In this paper, we show that employing a single, static strategy results in inefficiency and resource under-utilization, highlighting the need for adaptive approaches to handle the heterogeneous workloads across sequences. To address this, we propose a heterogeneity-adaptive sequence parallelism method. For each training step, our approach captures the variability in sequence lengths and assigns the optimal combination of scattering strategies based on workload characteristics. We model this problem as a linear programming optimization and design an efficient and effective solver to find the optimal solution. Furthermore, we implement our method in a high-performance system that supports adaptive parallelization in distributed LLM training. Experimental results demonstrate that our system outperforms state-of-the-art training frameworks by up to 1.98x.

cs.DC↗

arXiv2025

Transfer learning in Scalable Graph Neural Network for Improved Physical Simulation

In recent years, Graph Neural Network (GNN) based models have shown promising results in simulating physics of complex systems. However, training dedicated graph network based physics simulators can be costly, as most models are confined to fully supervised training, which requires extensive data generated from traditional physics simulators. To date, how transfer learning could improve the model performance and training efficiency has remained unexplored. In this work, we introduce a pre-training and transfer learning paradigm for graph network simulators. We propose the scalable graph U-net (SGUNET). Incorporating an innovative depth-first search (DFS) pooling, the SGUNET is adaptable to different mesh sizes and resolutions for various simulation tasks. To enable the transfer learning between differently configured SGUNETs, we propose a set of mapping functions to align the parameters between the pre-trained model and the target model. An extra normalization term is also added into the loss to constrain the difference between the pre-trained weights and target model weights for better generalization performance. To pre-train our physics simulator we created a dataset which includes 20,000 physical simulations of randomly selected 3D shapes from the open source A Big CAD (ABC) dataset. We show that our proposed transfer learning methods allow the model to perform even better when fine-tuned with small amounts of training data than when it is trained from scratch with full extensive dataset. On the 2D Deformable Plate benchmark dataset, our pre-trained model fine-tuned on 1/16 of the training data achieved an 11.05\% improvement in position RMSE compared to the model trained from scratch.

cs.LG↗