Qianwen Cao

SafeRAG: A Large-Language-Model-Based Multistage Retrieval-Augmented Framework for Oil and Gas Safety Report Generation

Qianwen Cao, Chiyu Zhang, Junxiong Ning, Gongru Li

Summary In high-risk industrial settings, leveraging large language models (LLMs) for automated accident analysis and generating safety reports has emerged as an efficient workflow. However, this approach is fundamentally constrained by the models’ inherent knowledge limitations, frequently resulting in analyses that lack …

cn (code pays fourni par la source)

FDA_YOLOv8: refined small object detection in unmanned aerial vehicle imagery

Weihong Chen, Qianjie Deng, Ziyan Xiang, Qianwen Cao et autres

Small object detection (SOD) is essential for security monitoring in unmanned aerial vehicle (UAV) imagery. However, the inherently low effective resolution, weak semantic representation, and cluttered background of small objects pose significant challenges. Although deep learning methods have been widely applied to …

cn (code pays fourni par la source)

Accelerating Sparse Transformer Inference on GPU

Wenhao Dai, Haodong Deng, Mengfei Rong, Xinyu Yang et autres

Large language models (LLMs) are popular around the world due to their powerful understanding capabilities. As the core component of LLMs, accelerating Transformer through parallelization has gradually become a hot research topic. Mask layers introduce sparsity into Transformer to reduce calculations. However, …

cn (code pays fourni par la source)

Knowledge Reranking Augmented LLM for Industry Accident Analysis

Qianwen Cao, Junxiong Ning, Jinrong Ma, Chiyu Zhang

Toward Energy-Safe Industrial Monitoring: A Hybrid Language Model Framework for Video Captioning

Qianwen Cao, Li Che

In the energy industry, like industrial monitoring scenarios, using generative AI for video captioning technology is crucial in event understanding and safety analysis. Current approaches typically rely on a single language model to decode visual semantics from video frames. Lightweight pre-trained generative …

cn (code pays fourni par la source)

Prediction and optimization of the impact of digital contracts on Internet news communication based on deep reinforcement learning

Qianwen Cao

With the rapid development of Internet technology, the digital contract is a vital medium of information dissemination. The rapid spread of rumors in Internet news communication misleads public cognition and may cause social panic and instability. This study aims to explore a …

cn (code pays fourni par la source)

QueryNILM: Query based Non-Intrusive Load Monitoring with Appliance Embeddings

Shijie Yang, Jie Jiang, Yichen Zhang, Qiuqiang Kong et autres

Non-Intrusive Load Monitoring (NILM), also known as energy disaggregation, aims to infer appliance-level energy consumption from aggregated readings obtained from a single meter measuring a household’s total electricity demand. The increasing availability of large-scale and fine-grained household electricity usage datasets has facilitated …

cn, hk (code pays fourni par la source)

From Skeleton to Flesh: Aggregated Relational Transformer Towards Controllable Video Captioning with Two-Step Decoding

Qianwen Cao, Heyan Huang, Boran Wang

Video captioning is the task of automatically producing natural language descriptions for a given video. The great success of the Transformer architecture in NLP and CV has inspired recent attempts at constructing an end-to-end Transformer-based model for video captioning. Although achieving impressive …

cn (code pays fourni par la source)

Accelerating Sparse Transformer Inference on GPU

Wenhao Dai, Haodong Deng, Xinyu Yang, Hongyu Liu et autres

Large language models (LLMs) are popular around the world due to their powerful understanding capabilities. As the core component of LLMs, accelerating Transformer through parallelization has gradually become a hot research topic. Mask layers introduce sparsity into Transformer to reduce calculations. However, …

Enhanced Named Entity Recognition for Drilling Engineering Using Masked CRF

Qianwen Cao, Wei Li

Drilling engineering reports are essential for oil and gas extraction, documenting critical geological information and operational parameters. However, their unstructured format, diverse writing styles, and long-distance dependencies between related entities pose challenges for automated information extraction. Traditional rule-based and machine learning methods …

cn (code pays fourni par la source)

Speaking Style Memory-Enhanced Language Model for Open-Domain Personalized Dialogue Generation

Qianwen Cao

Open-domain personalized dialogue systems have gained increasing attention due to their ability to generate engaging and user-specific responses. Recent advancements in natural language processing have significantly improved dialogue generation; however, incorporating personalized speaking styles remains a challenge. Existing approaches often separately train …

cn (code pays fourni par la source)

FlexTransformer: Flexibly Allocating Transformer Modules for Efficient Video Captioning

Qianwen Cao, Gengyu Li, Li Che

cn (code pays fourni par la source)