OSS Alternative - Discover Top Open Source Alternatives to Popular Software

THUDM/WebGLM

An efficient and cost-effective web-enhanced question answering system that integrates web search and retrieval with a large language model, optimized with human preferences.

Core Features

LLM-augmented Retriever for enhanced web content retrieval.

Bootstrapped Generator for human-like response generation.

Human Preference-aware Scorer to ensure high-quality and engaging content.

Integration of web search capabilities into a pre-trained language model.

Quick Start

pip install -r requirements.txt

Detailed Introduction

WebGLM is an innovative web-enhanced question answering system designed for efficiency and cost-effectiveness. It integrates advanced web search and retrieval capabilities directly into a 10-billion-parameter General Language Model (GLM). The system features an LLM-augmented retriever to find relevant web content, a bootstrapped generator to craft human-like answers, and a human preference-aware scorer to ensure the quality and usefulness of generated responses. This approach aims to improve real-world application deployment by providing accurate and engaging answers, leveraging both external web knowledge and powerful language models.