arXiv subjects

Jianwei Zhang

Publications and source records attributed to Jianwei Zhang.

At least 91 records · Page 5Linked to original sources

arXiv2025

Qwen2.5 Technical Report

In this report, we introduce Qwen2.5, a comprehensive series of large language models (LLMs) designed to meet diverse needs. Compared to previous iterations, Qwen 2.5 has been significantly improved during both the pre-training and post-training stages. In terms of pre-training, we have scaled the high-quality pre-training datasets from the previous 7 trillion tokens to 18 trillion tokens. This provides a strong foundation for common sense, expert knowledge, and reasoning capabilities. In terms of post-training, we implement intricate supervised finetuning with over 1 million samples, as well as multistage reinforcement learning. Post-training techniques enhance human preference, and notably improve long text generation, structural data analysis, and instruction following. To handle diverse and varied use cases effectively, we present Qwen2.5 LLM series in rich sizes. Open-weight offerings include base and instruction-tuned models, with quantized versions available. In addition, for hosted solutions, the proprietary models currently include two mixture-of-experts (MoE) variants: Qwen2.5-Turbo and Qwen2.5-Plus, both available from Alibaba Cloud Model Studio. Qwen2.5 has demonstrated top-tier performance on a wide range of benchmarks evaluating language understanding, reasoning, mathematics, coding, human preference alignment, etc. Specifically, the open-weight flagship Qwen2.5-72B-Instruct outperforms a number of open and proprietary models and demonstrates competitive performance to the state-of-the-art open-weight model, Llama-3-405B-Instruct, which is around 5 times larger. Qwen2.5-Turbo and Qwen2.5-Plus offer superior cost-effectiveness while performing competitively against GPT-4o-mini and GPT-4o respectively. Additionally, as the foundation, Qwen2.5 models have been instrumental in training specialized models such as Qwen2.5-Math, Qwen2.5-Coder, QwQ, and multimodal models.

cs.CL↗

arXiv2024

Rotated Runtime Smooth: Training-Free Activation Smoother for accurate INT4 inference

Large language models have demonstrated promising capabilities upon scaling up parameters. However, serving large language models incurs substantial computation and memory movement costs due to their large scale. Quantization methods have been employed to reduce service costs and latency. Nevertheless, outliers in activations hinder the development of INT4 weight-activation quantization. Existing approaches separate outliers and normal values into two matrices or migrate outliers from activations to weights, suffering from high latency or accuracy degradation. Based on observing activations from large language models, outliers can be classified into channel-wise and spike outliers. In this work, we propose Rotated Runtime Smooth (RRS), a plug-and-play activation smoother for quantization, consisting of Runtime Smooth and the Rotation operation. Runtime Smooth (RS) is introduced to eliminate channel-wise outliers by smoothing activations with channel-wise maximums during runtime. The rotation operation can narrow the gap between spike outliers and normal values, alleviating the effect of victims caused by channel-wise smoothing. The proposed method outperforms the state-of-the-art method in the LLaMA and Qwen families and improves WikiText-2 perplexity from 57.33 to 6.66 for INT4 inference.

cs.LG↗

arXiv2024

GenUDC: High Quality 3D Mesh Generation with Unsigned Dual Contouring Representation

Generating high-quality meshes with complex structures and realistic surfaces is the primary goal of 3D generative models. Existing methods typically employ sequence data or deformable tetrahedral grids for mesh generation. However, sequence-based methods have difficulty producing complex structures with many faces due to memory limits. The deformable tetrahedral grid-based method MeshDiffusion fails to recover realistic surfaces due to the inherent ambiguity in deformable grids. We propose the GenUDC framework to address these challenges by leveraging the Unsigned Dual Contouring (UDC) as the mesh representation. UDC discretizes a mesh in a regular grid and divides it into the face and vertex parts, recovering both complex structures and fine details. As a result, the one-to-one mapping between UDC and mesh resolves the ambiguity problem. In addition, GenUDC adopts a two-stage, coarse-to-fine generative process for 3D mesh generation. It first generates the face part as a rough shape and then the vertex part to craft a detailed shape. Extensive evaluations demonstrate the superiority of UDC as a mesh representation and the favorable performance of GenUDC in mesh generation. The code and trained models are available at https://github.com/TrepangCat/GenUDC.

cs.CV↗

arXiv2024

Integrating spoken instructions into flight trajectory prediction to optimize automation in air traffic control

The booming air transportation industry inevitably burdens air traffic controllers' workload, causing unexpected human factor-related incidents. Current air traffic control systems fail to consider spoken instructions for traffic prediction, bringing significant challenges in detecting human errors during real-time traffic operations. Here, we present an automation paradigm integrating controlling intent into the information processing loop through the spoken instruction-aware flight trajectory prediction framework. A 3-stage progressive multi-modal learning paradigm is proposed to address the modality gap between the trajectory and spoken instructions, as well as minimize the data requirements. Experiments on a real-world dataset show the proposed framework achieves flight trajectory prediction with high predictability and timeliness, obtaining over 20% relative reduction in mean deviation error. Moreover, the generalizability of the proposed framework is also confirmed by various model architectures. The proposed framework can formulate full-automated information processing in real-world air traffic applications, supporting human error detection and enhancing aviation safety.

cs.SD↗

arXiv2024

ToolEENet: Tool Affordance 6D Pose Estimation

The exploration of robotic dexterous hands utilizing tools has recently attracted considerable attention. A significant challenge in this field is the precise awareness of a tool's pose when grasped, as occlusion by the hand often degrades the quality of the estimation. Additionally, the tool's overall pose often fails to accurately represent the contact interaction, thereby limiting the effectiveness of vision-guided, contact-dependent activities. To overcome this limitation, we present the innovative TOOLEE dataset, which, to the best of our knowledge, is the first to feature affordance segmentation of a tool's end-effector (EE) along with its defined 6D pose based on its usage. Furthermore, we propose the ToolEENet framework for accurate 6D pose estimation of the tool's EE. This framework begins by segmenting the tool's EE from raw RGBD data, then uses a diffusion model-based pose estimator for 6D pose estimation at a category-specific level. Addressing the issue of symmetry in pose estimation, we introduce a symmetry-aware pose representation that enhances the consistency of pose estimation. Our approach excels in this field, demonstrating high levels of precision and generalization. Furthermore, it shows great promise for application in contact-based manipulation scenarios. All data and codes are available on the project website: https://tooleenet-iros2024.github.io/

cs.RO↗

arXiv2024

Qwen2 Technical Report

This report introduces the Qwen2 series, the latest addition to our large language models and large multimodal models. We release a comprehensive suite of foundational and instruction-tuned language models, encompassing a parameter range from 0.5 to 72 billion, featuring dense models and a Mixture-of-Experts model. Qwen2 surpasses most prior open-weight models, including its predecessor Qwen1.5, and exhibits competitive performance relative to proprietary models across diverse benchmarks on language understanding, generation, multilingual proficiency, coding, mathematics, and reasoning. The flagship model, Qwen2-72B, showcases remarkable performance: 84.2 on MMLU, 37.9 on GPQA, 64.6 on HumanEval, 89.5 on GSM8K, and 82.4 on BBH as a base language model. The instruction-tuned variant, Qwen2-72B-Instruct, attains 9.1 on MT-Bench, 48.1 on Arena-Hard, and 35.7 on LiveCodeBench. Moreover, Qwen2 demonstrates robust multilingual capabilities, proficient in approximately 30 languages, spanning English, Chinese, Spanish, French, German, Arabic, Russian, Korean, Japanese, Thai, Vietnamese, and more, underscoring its versatility and global reach. To foster community innovation and accessibility, we have made the Qwen2 model weights openly available on Hugging Face and ModelScope, and the supplementary materials including example code on GitHub. These platforms also include resources for quantization, fine-tuning, and deployment, facilitating a wide range of applications and research endeavors.

cs.CL↗

arXiv2024

ROSE: A Recognition-Oriented Speech Enhancement Framework in Air Traffic Control Using Multi-Objective Learning

Radio speech echo is a specific phenomenon in the air traffic control (ATC) domain, which degrades speech quality and further impacts automatic speech recognition (ASR) accuracy. In this work, a time-domain recognition-oriented speech enhancement (ROSE) framework is proposed to improve speech intelligibility and also advance ASR accuracy based on convolutional encoder-decoder-based U-Net framework, which serves as a plug-and-play tool in ATC scenarios and does not require additional retraining of the ASR model. Specifically, 1) In the U-Net architecture, an attention-based skip-fusion (ABSF) module is applied to mine shared features from encoders using an attention mask, which enables the model to effectively fuse the hierarchical features. 2) A channel and sequence attention (CSAtt) module is innovatively designed to guide the model to focus on informative features in dual parallel attention paths, aiming to enhance the effective representations and suppress the interference noises. 3) Based on the handcrafted features, ASR-oriented optimization targets are designed to improve recognition performance in the ATC environment by learning robust feature representations. By incorporating both the SE-oriented and ASR-oriented losses, ROSE is implemented in a multi-objective learning manner by optimizing shared representations across the two task objectives. The experimental results show that the ROSE significantly outperforms other state-of-the-art methods for both the SE and ASR tasks, in which all the proposed improvements are confirmed by designed experiments. In addition, the proposed approach can contribute to the desired performance improvements on public datasets.

cs.SD↗

arXiv2024

RoboGolf: Mastering Real-World Minigolf with a Reflective Multi-Modality Vision-Language Model

Minigolf is an exemplary real-world game for examining embodied intelligence, requiring challenging spatial and kinodynamic understanding to putt the ball. Additionally, reflective reasoning is required if the feasibility of a challenge is not ensured. We introduce RoboGolf, a VLM-based framework that combines dual-camera perception with closed-loop action refinement, augmented by a reflective equilibrium loop. The core of both loops is powered by finetuned VLMs. We analyze the capabilities of the framework in an offline inference setting, relying on an extensive set of recorded trajectories. Exemplary demonstrations of the analyzed problem domain are available at https://jity16.github.io/RoboGolf/

cs.RO↗

arXiv2024

Close the Sim2real Gap via Physically-based Structured Light Synthetic Data Simulation

Despite the substantial progress in deep learning, its adoption in industrial robotics projects remains limited, primarily due to challenges in data acquisition and labeling. Previous sim2real approaches using domain randomization require extensive scene and model optimization. To address these issues, we introduce an innovative physically-based structured light simulation system, generating both RGB and physically realistic depth images, surpassing previous dataset generation tools. We create an RGBD dataset tailored for robotic industrial grasping scenarios and evaluate it across various tasks, including object detection, instance segmentation, and embedding sim2real visual perception in industrial robotic grasping. By reducing the sim2real gap and enhancing deep learning training, we facilitate the application of deep learning models in industrial settings. Project details are available at https://baikaixinpublic.github.io/structured light 3D synthesizer/.

cs.CV↗

arXiv2024

Subequivariant Reinforcement Learning in 3D Multi-Entity Physical Environments

Learning policies for multi-entity systems in 3D environments is far more complicated against single-entity scenarios, due to the exponential expansion of the global state space as the number of entities increases. One potential solution of alleviating the exponential complexity is dividing the global space into independent local views that are invariant to transformations including translations and rotations. To this end, this paper proposes Subequivariant Hierarchical Neural Networks (SHNN) to facilitate multi-entity policy learning. In particular, SHNN first dynamically decouples the global space into local entity-level graphs via task assignment. Second, it leverages subequivariant message passing over the local entity-level graphs to devise local reference frames, remarkably compressing the representation redundancy, particularly in gravity-affected environments. Furthermore, to overcome the limitations of existing benchmarks in capturing the subtleties of multi-entity systems under the Euclidean symmetry, we propose the Multi-entity Benchmark (MEBEN), a new suite of environments tailored for exploring a wide range of multi-entity reinforcement learning. Extensive experiments demonstrate significant advancements of SHNN on the proposed benchmarks compared to existing methods. Comprehensive ablations are conducted to verify the indispensability of task assignment and subequivariance.

cs.LG↗

arXiv2024

Few-Shot Medical Image Segmentation with High-Fidelity Prototypes

Few-shot Semantic Segmentation (FSS) aims to adapt a pretrained model to new classes with as few as a single labelled training sample per class. Despite the prototype based approaches have achieved substantial success, existing models are limited to the imaging scenarios with considerably distinct objects and not highly complex background, e.g., natural images. This makes such models suboptimal for medical imaging with both conditions invalid. To address this problem, we propose a novel Detail Self-refined Prototype Network (DSPNet) to constructing high-fidelity prototypes representing the object foreground and the background more comprehensively. Specifically, to construct global semantics while maintaining the captured detail semantics, we learn the foreground prototypes by modelling the multi-modal structures with clustering and then fusing each in a channel-wise manner. Considering that the background often has no apparent semantic relation in the spatial dimensions, we integrate channel-specific structural information under sparse channel-aware regulation. Extensive experiments on three challenging medical image benchmarks show the superiority of DSPNet over previous state-of-the-art methods.

cs.CV↗

arXiv2024

Embedded Silicon-Organic Integrated Neuromorphic System

The development of artificial intelligence (AI) and robotics are both based on the tenet of "science and technology are people-oriented", and both need to achieve efficient communication with the human brain. Based on multi-disciplinary research in systems neuroscience, computer architecture, and functional organic materials, we proposed the concept of using AI to simulate the operating principles and materials of the brain in hardware to develop brain-inspired intelligence technology, and realized the preparation of neuromorphic computing devices and basic materials. We simulated neurons and neural networks in terms of material and morphology, using a variety of organic polymers as the base materials for neuroelectronic devices, for building neural interfaces as well as organic neural devices and silicon neural computational modules. We assemble organic artificial synapses with simulated neurons from silicon-based Field-Programmable Gate Array (FPGA) into organic artificial neurons, the basic components of neural networks, and later construct biological neural network models based on the interpreted neural circuits. Finally, we also discuss how to further build neuromorphic devices based on these organic artificial neurons, which have both a neural interface friendly to nervous tissue and interact with information from real biological neural networks.

q-bio.NC↗

arXiv2024

A Non-autoregressive Multi-Horizon Flight Trajectory Prediction Framework with Gray Code Representation

Flight Trajectory Prediction (FTP) is an essential task in Air Traffic Control (ATC), which can assist air traffic controllers in managing airspace more safely and efficiently. Existing approaches generally perform multi-horizon FTP tasks in an autoregressive manner, thereby suffering from error accumulation and low-efficiency problems. In this paper, a novel framework, called FlightBERT++, is proposed to i) forecast multi-horizon flight trajectories directly in a non-autoregressive way, and ii) improve the limitation of the binary encoding (BE) representation in the FlightBERT framework. Specifically, the proposed framework is implemented by a generalized encoder-decoder architecture, in which the encoder learns the temporal-spatial patterns from historical observations and the decoder predicts the flight status for the future horizons. Compared to conventional architecture, an innovative horizon-aware contexts generator is dedicatedly designed to consider the prior horizon information, which further enables non-autoregressive multi-horizon prediction. Additionally, the Gray code representation and the differential prediction paradigm are designed to cope with the high-bit misclassifications of the BE representation, which significantly reduces the outliers in the predictions. Moreover, a differential prompted decoder is proposed to enhance the capability of the differential predictions by leveraging the stationarity of the differential sequence. Extensive experiments are conducted to validate the proposed framework on a real-world flight trajectory dataset. The experimental results demonstrated that the proposed framework outperformed the competitive baselines in both FTP performance and computational efficiency.

cs.LG↗

arXiv2024

OMPO: A Unified Framework for RL under Policy and Dynamics Shifts

Training reinforcement learning policies using environment interaction data collected from varying policies or dynamics presents a fundamental challenge. Existing works often overlook the distribution discrepancies induced by policy or dynamics shifts, or rely on specialized algorithms with task priors, thus often resulting in suboptimal policy performances and high learning variances. In this paper, we identify a unified strategy for online RL policy learning under diverse settings of policy and dynamics shifts: transition occupancy matching. In light of this, we introduce a surrogate policy learning objective by considering the transition occupancy discrepancies and then cast it into a tractable min-max optimization problem through dual reformulation. Our method, dubbed Occupancy-Matching Policy Optimization (OMPO), features a specialized actor-critic structure equipped with a distribution discriminator and a small-size local buffer. We conduct extensive experiments based on the OpenAI Gym, Meta-World, and Panda Robots environments, encompassing policy shifts under stationary and nonstationary dynamics, as well as domain adaption. The results demonstrate that OMPO outperforms the specialized baselines from different categories in all settings. We also find that OMPO exhibits particularly strong performance when combined with domain randomization, highlighting its potential in RL-based robotics applications

cs.LG↗

arXiv2024

Offline-Boosted Actor-Critic: Adaptively Blending Optimal Historical Behaviors in Deep Off-Policy RL

Off-policy reinforcement learning (RL) has achieved notable success in tackling many complex real-world tasks, by leveraging previously collected data for policy learning. However, most existing off-policy RL algorithms fail to maximally exploit the information in the replay buffer, limiting sample efficiency and policy performance. In this work, we discover that concurrently training an offline RL policy based on the shared online replay buffer can sometimes outperform the original online learning policy, though the occurrence of such performance gains remains uncertain. This motivates a new possibility of harnessing the emergent outperforming offline optimal policy to improve online policy learning. Based on this insight, we present Offline-Boosted Actor-Critic (OBAC), a model-free online RL framework that elegantly identifies the outperforming offline policy through value comparison, and uses it as an adaptive constraint to guarantee stronger policy learning performance. Our experiments demonstrate that OBAC outperforms other popular model-free RL baselines and rivals advanced model-based RL methods in terms of sample efficiency and asymptotic performance across 53 tasks spanning 6 task suites.

cs.LG↗

arXiv2024

Hunyuan-DiT: A Powerful Multi-Resolution Diffusion Transformer with Fine-Grained Chinese Understanding

We present Hunyuan-DiT, a text-to-image diffusion transformer with fine-grained understanding of both English and Chinese. To construct Hunyuan-DiT, we carefully design the transformer structure, text encoder, and positional encoding. We also build from scratch a whole data pipeline to update and evaluate data for iterative model optimization. For fine-grained language understanding, we train a Multimodal Large Language Model to refine the captions of the images. Finally, Hunyuan-DiT can perform multi-turn multimodal dialogue with users, generating and refining images according to the context. Through our holistic human evaluation protocol with more than 50 professional human evaluators, Hunyuan-DiT sets a new state-of-the-art in Chinese-to-image generation compared with other open-source models. Code and pretrained models are publicly available at github.com/Tencent/HunyuanDiT

cs.CV↗

arXiv2024

Seizing Serendipity: Exploiting the Value of Past Success in Off-Policy Actor-Critic

Learning high-quality $Q$-value functions plays a key role in the success of many modern off-policy deep reinforcement learning (RL) algorithms. Previous works primarily focus on addressing the value overestimation issue, an outcome of adopting function approximators and off-policy learning. Deviating from the common viewpoint, we observe that $Q$-values are often underestimated in the latter stage of the RL training process, potentially hindering policy learning and reducing sample efficiency. We find that such a long-neglected phenomenon is often related to the use of inferior actions from the current policy in Bellman updates as compared to the more optimal action samples in the replay buffer. To address this issue, our insight is to incorporate sufficient exploitation of past successes while maintaining exploration optimism. We propose the Blended Exploitation and Exploration (BEE) operator, a simple yet effective approach that updates $Q$-value using both historical best-performing actions and the current policy. Based on BEE, the resulting practical algorithm BAC outperforms state-of-the-art methods in over 50 continuous control tasks and achieves strong performance in failure-prone scenarios and real-world robot tasks. Benchmark results and videos are available at https://jity16.github.io/BEE/.

cs.LG↗

arXiv2024

Smooth Computation without Input Delay: Robust Tube-Based Model Predictive Control for Robot Manipulator Planning

Model Predictive Control (MPC) has exhibited remarkable capabilities in optimizing objectives and meeting constraints. However, the substantial computational burden associated with solving the Optimal Control Problem (OCP) at each triggering instant introduces significant delays between state sampling and control application. These delays limit the practicality of MPC in resource-constrained systems when engaging in complex tasks. The intuition to address this issue in this paper is that by predicting the successor state, the controller can solve the OCP one time step ahead of time thus avoiding the delay of the next action. To this end, we compute deviations between real and nominal system states, predicting forthcoming real states as initial conditions for the imminent OCP solution. Anticipatory computation stores optimal control based on current nominal states, thus mitigating the delay effects. Additionally, we establish an upper bound for linearization error, effectively linearizing the nonlinear system, reducing OCP complexity, and enhancing response speed. We provide empirical validation through two numerical simulations and corresponding real-world robot tasks, demonstrating significant performance improvements and augmented response speed (up to $90\%$) resulting from the seamless integration of our proposed approach compared to conventional time-triggered MPC strategies.

cs.RO↗