Wenxuan Zhou
YOU?
Author Swipe
View article: Detection of Turbulent Combustion Mechanism in NH <sub>3</sub> /H <sub>2</sub> Stratified Premixtures Using Chemical Explosive Mode Analysis
Detection of Turbulent Combustion Mechanism in NH <sub>3</sub> /H <sub>2</sub> Stratified Premixtures Using Chemical Explosive Mode Analysis Open
View article: Numerical study on stretch extinction mechanism of NH3/H2/air laminar counterflow premixed flames
Numerical study on stretch extinction mechanism of NH3/H2/air laminar counterflow premixed flames Open
View article: Deep Learning-Based Layout Analysis Method for Complex Layout Image Elements
Deep Learning-Based Layout Analysis Method for Complex Layout Image Elements Open
The layout analysis of elements is indispensable in graphic design, as effective layout design not only facilitates the delivery of visual information but also enhances the overall esthetic appeal to the audience. The combination of deep l…
View article: Pulsed Laser Transformed High-Entropy Metal (Oxy)Hydroxide for Highly Active Seawater Oxidation with an Operational Robustness over 400 Hours
Pulsed Laser Transformed High-Entropy Metal (Oxy)Hydroxide for Highly Active Seawater Oxidation with an Operational Robustness over 400 Hours Open
Oxygen evolution reaction (OER) is critical to renewable energy conversion technologies such as water electrolyzers, though OER remains constrained by sluggish kinetics and catalyst instability. High-entropy materials (HEMs) have emerged a…
View article: Ambient Pulsed Laser Augmentation to Enforce the 4-Electron Pathway for the Complete Reduction of Dioxygen into Water on Dicopper Nanohybrids
Ambient Pulsed Laser Augmentation to Enforce the 4-Electron Pathway for the Complete Reduction of Dioxygen into Water on Dicopper Nanohybrids Open
Oxygen reduction reaction (ORR) plays a critical role in many electrochemical energy conversion devices including low-temperature fuel cells. However, the sluggish nature of ORR has impeded the commercialization of fuel cells with high eff…
View article: Deep Learning-Based Layout Analysis Method for Complex Layout Image Elements
Deep Learning-Based Layout Analysis Method for Complex Layout Image Elements Open
Layout analysis of elements is indispensable in graphic design, and layout design can make the audience receive visual information as well as more attractive. The combination of deep learning and graphic design has gradually turned into a …
View article: Think Smarter not Harder: Adaptive Reasoning with Inference Aware Optimization
Think Smarter not Harder: Adaptive Reasoning with Inference Aware Optimization Open
Solving mathematics problems has been an intriguing capability of large language models, and many efforts have been made to improve reasoning by extending reasoning length, such as through self-correction and extensive long chain-of-though…
View article: Improving Model Factuality with Fine-grained Critique-based Evaluator
Improving Model Factuality with Fine-grained Critique-based Evaluator Open
View article: Preparation of <i>S-1 </i>bio-nano selenium microbial agent and its effect on improving the quality of summer-autumn young tea shoots
Preparation of <i>S-1 </i>bio-nano selenium microbial agent and its effect on improving the quality of summer-autumn young tea shoots Open
View article: Code Execution as Grounded Supervision for LLM Reasoning
Code Execution as Grounded Supervision for LLM Reasoning Open
View article: Mental Health Prediction and Stigma Reduction in College Students: A Machine Learning Perspective
Mental Health Prediction and Stigma Reduction in College Students: A Machine Learning Perspective Open
View article: Numerical Study on Stretch Extinction Mechanism of Nh3/H2/Air Laminar Counterflow Premixed Flames
Numerical Study on Stretch Extinction Mechanism of Nh3/H2/Air Laminar Counterflow Premixed Flames Open
View article: Influence of Non‐Uniform Temperature Distribution on Elastoplastic Deformation of Windings
Influence of Non‐Uniform Temperature Distribution on Elastoplastic Deformation of Windings Open
To accurately analyse the deformation of transformer windings under short‐circuit conditions, a three‐dimensional multiphysics field simulation model of the magnetic‐fluid‐thermal‐structural coupling is constructed. The current magnitude u…
View article: Learning Auxiliary Tasks Improves Reference-Free Hallucination Detection in Open-Domain Long-Form Generation
Learning Auxiliary Tasks Improves Reference-Free Hallucination Detection in Open-Domain Long-Form Generation Open
View article: Breaking through the blood-brain barrier for different kinds of drugs: challenges, solutions, and perspectives
Breaking through the blood-brain barrier for different kinds of drugs: challenges, solutions, and perspectives Open
Brain diseases, including psychosis, neurological disorders, strokes, etc., account for more than 15% of all global health damage, which is higher than that caused by cancer and cardiovascular diseases. Brain health and the treatment of br…
View article: Improving Model Factuality with Fine-grained Critique-based Evaluator
Improving Model Factuality with Fine-grained Critique-based Evaluator Open
Factuality evaluation aims to detect factual errors produced by language models (LMs) and hence guide the development of more factual models. Towards this goal, we train a factuality evaluator, FenCE, that provides LM generators with claim…
View article: Mask-Encoded Sparsification: Mitigating Biased Gradients in Communication-Efficient Split Learning
Mask-Encoded Sparsification: Mitigating Biased Gradients in Communication-Efficient Split Learning Open
This paper introduces a novel framework designed to achieve a high compression ratio in Split Learning (SL) scenarios where resource-constrained devices are involved in large-scale model training. Our investigations demonstrate that compre…
View article: SFTMix: Elevating Language Model Instruction Tuning with Mixup Recipe
SFTMix: Elevating Language Model Instruction Tuning with Mixup Recipe Open
To acquire instruction-following capabilities, large language models (LLMs) undergo instruction tuning, where they are trained on instruction-response pairs using next-token prediction (NTP). Efforts to improve instruction tuning often foc…
View article: The Perfect Blend: Redefining RLHF with Mixture of Judges
The Perfect Blend: Redefining RLHF with Mixture of Judges Open
Reinforcement learning from human feedback (RLHF) has become the leading approach for fine-tuning large language models (LLM). However, RLHF has limitations in multi-task learning (MTL) due to challenges of reward hacking and extreme multi…
View article: Mask-Encoded Sparsification: Mitigating Biased Gradients in Communication-Efficient Split Learning
Mask-Encoded Sparsification: Mitigating Biased Gradients in Communication-Efficient Split Learning Open
This paper introduces a novel framework designed to achieve a high compression ratio in Split Learning (SL) scenarios where resource-constrained devices are involved in large-scale model training. Our investigations demonstrate that compre…
View article: HACMan++: Spatially-Grounded Motion Primitives for Manipulation
HACMan++: Spatially-Grounded Motion Primitives for Manipulation Open
Although end-to-end robot learning has shown some success for robot manipulation, the learned policies are often not sufficiently robust to variations in object pose or geometry. To improve the policy generalization, we introduce spatially…
View article: Improving Multilingual Instruction Finetuning via Linguistically Natural and Diverse Datasets
Improving Multilingual Instruction Finetuning via Linguistically Natural and Diverse Datasets Open
Advancements in Large Language Models (LLMs) have significantly enhanced instruction-following capabilities. However, most Instruction Fine-Tuning (IFT) datasets are predominantly in English, limiting model performance in other languages. …
View article: mDPO: Conditional Preference Optimization for Multimodal Large Language Models
mDPO: Conditional Preference Optimization for Multimodal Large Language Models Open
Direct preference optimization (DPO) has shown to be an effective method for large language model (LLM) alignment. Recent works have attempted to apply DPO to multimodal scenarios but have found it challenging to achieve consistent improve…
View article: MuirBench: A Comprehensive Benchmark for Robust Multi-image Understanding
MuirBench: A Comprehensive Benchmark for Robust Multi-image Understanding Open
We introduce MuirBench, a comprehensive benchmark that focuses on robust multi-image understanding capabilities of multimodal LLMs. MuirBench consists of 12 diverse multi-image tasks (e.g., scene understanding, ordering) that involve 10 ca…
View article: Offset Unlearning for Large Language Models
Offset Unlearning for Large Language Models Open
Despite the strong capabilities of Large Language Models (LLMs) to acquire knowledge from their training corpora, the memorization of sensitive information in the corpora such as copyrighted, biased, and private content has led to ethical …
View article: Sim2Real Manipulation on Unknown Objects with Tactile-based Reinforcement Learning
Sim2Real Manipulation on Unknown Objects with Tactile-based Reinforcement Learning Open
Using tactile sensors for manipulation remains one of the most challenging problems in robotics. At the heart of these challenges is generalization: How can we train a tactile-based policy that can manipulate unseen and diverse objects? In…
View article: Contrastive Instruction Tuning
Contrastive Instruction Tuning Open
Instruction tuning has been used as a promising approach to improve the performance of large language models (LLMs) on unseen tasks. However, current LLMs exhibit limited robustness to unseen instructions, generating inconsistent outputs w…
View article: Precise O-Doping Across Dual-Fe Sites Directed by Pulsed Laser Optimized Via Taguchi Method for Sustainable H2o2 Electrosynthesis
Precise O-Doping Across Dual-Fe Sites Directed by Pulsed Laser Optimized Via Taguchi Method for Sustainable H2o2 Electrosynthesis Open
View article: GeoLM: Empowering Language Models for Geospatially Grounded Language Understanding
GeoLM: Empowering Language Models for Geospatially Grounded Language Understanding Open
Humans subconsciously engage in geospatial reasoning when reading articles. We recognize place names and their spatial relations in text and mentally associate them with their physical locations on Earth. Although pretrained language model…
View article: Reinforcement Learning in a Safety-Embedded MDP with Trajectory Optimization
Reinforcement Learning in a Safety-Embedded MDP with Trajectory Optimization Open
Safe Reinforcement Learning (RL) plays an important role in applying RL algorithms to safety-critical real-world applications, addressing the trade-off between maximizing rewards and adhering to safety constraints. This work introduces a n…