Human-computer interaction methods, apparatus, electronic devices, and storage media
The method automates plugin selection and optimizes input text length in large-scale language models, addressing inefficiencies and improving user experience and model accuracy.
Patent Information
- Authority / Receiving Office
- JP · JP
- Patent Type
- Patents
- Current Assignee / Owner
- BEIJING BAIDU NETCOM SCI & TECH CO LTD
- Filing Date
- 2024-06-14
- Publication Date
- 2026-05-25
AI Technical Summary
Large-scale language models struggle with accurately responding to new data outside their training dataset, and the manual selection of plugins becomes cumbersome as the number of plugins increases, leading to inefficiencies and reduced accuracy due to excessive input text length exceeding token limits.
A method to automatically identify a target plugin based on dialogue text, generate a second dialogue text by combining the original text with the plugin's description, and input this into the model, reducing the need for manual plugin selection and minimizing excessive text input.
Improves user experience by automating plugin selection, reducing processing time, and enhancing the accuracy and efficiency of large-scale language models by optimizing input text length.
Smart Images

Figure 0007864847000001 
Figure 0007864847000002 
Figure 0007864847000003