Výsledky vyhledávání - "Zheng, Linghan"

Report

RecurFormer: Not All Transformer Heads Need Self-Attention

Autor: Yan, Ruiqing, Zheng, Linghan, Du, Xingbo, Zou, Han, Guo, Yufeng, Yang, Jianfei

Transformer-based large language models (LLMs) excel in modeling complex language patterns but face significant computational costs during inference, especially with long inputs due to the attention mechanism's memory overhead. We observe that certai

Externí odkaz: http://arxiv.org/abs/2410.12850

Zobrazit plný text záznamu

Report

Unveiling and Controlling Anomalous Attention Distribution in Transformers

Autor: Yan, Ruiqing, Du, Xingbo, Deng, Haoyu, Zheng, Linghan, Sun, Qiuzhuang, Hu, Jifang, Shao, Yuhang, Jiang, Penghao, Jiang, Jinrong, Zhao, Lian

With the advent of large models based on the Transformer architecture, researchers have observed an anomalous phenomenon in the Attention mechanism--there is a very high attention on the first element, which is prevalent across Transformer-based mode

Externí odkaz: http://arxiv.org/abs/2407.01601

Zobrazit plný text záznamu

Report

Code-Based English Models Surprising Performance on Chinese QA Pair Extraction Task

Autor: Zheng, Linghan, Liu, Hui, Lin, Xiaojun, Dong, Jiayuan, Sheng, Yue, Shi, Gang, Liu, Zhiwei, Chen, Hongwei

In previous studies, code-based models have consistently outperformed text-based models in reasoning-intensive scenarios. When generating our knowledge base for Retrieval-Augmented Generation (RAG), we observed that code-based models also perform exc

Externí odkaz: http://arxiv.org/abs/2401.10286

Zobrazit plný text záznamu

Report

CAINNFlow: Convolutional block Attention modules and Invertible Neural Networks Flow for anomaly detection and localization tasks

Autor: Yan, Ruiqing, Zhang, Fan, Huang, Mengyuan, Liu, Wu, Hu, Dongyu, Li, Jinfeng, Liu, Qiang, Jiang, Jinrong, Guo, Qianjin, Zheng, Linghan

Detection of object anomalies is crucial in industrial processes, but unsupervised anomaly detection and localization is particularly important due to the difficulty of obtaining a large number of defective samples and the unpredictable types of anom

Externí odkaz: http://arxiv.org/abs/2206.01992

Zobrazit plný text záznamu