---
title: WebGLM
canonical_url: "https://www.modelscope.cn/models/ZhipuAI/WebGLM"
md_url: "https://www.modelscope.cn/models/ZhipuAI/WebGLM.md"
repository: ZhipuAI/WebGLM
last_updated: 2024-01-08
pipeline_tag: text-generation
tasks:
  - text-generation
model_type:
  - glm
architectures:
  - GLMForConditionalGeneration
library_name:
  - pytorch
frameworks:
  - Pytorch
language:
  - en
downloads: 141
stars: 0
tags:
  - glm
  - webglm
  - thudm
---

# WebGLM

> WebGLM - ZhipuAI 在 ModelScope 开源的模型。WebGLM: Towards An Efficient Web-enhanced Question Answering System with Human Preference

- **Repository**: ZhipuAI/WebGLM
- **Tasks**: text-generation
- **Tags**: glm, webglm, thudm
- **Downloads**: 141
- **Stars**: 0
- **Last updated**: 2024-01-08

Source: https://www.modelscope.cn/models/ZhipuAI/WebGLM

---

<h1>WebGLM: Towards An Efficient Web-enhanced Question Answering System with Human Preference</h1>

<p align="center">
  📃 <a href="https://arxiv.org/pdf/2306.07906.pdf" target="_blank">Paper (KDD 2023)</a> 
  | 
  💻 <a href="https://github.com/THUDM/WebGLM" target="_blank">Github Repo</a>
</p>

# Introduction

WebGLM aspires to provide an efficient and cost-effective web-enhanced question-answering system using the 10-billion-parameter General Language Model (GLM). It aims to improve real-world application deployment by integrating web search and retrieval capabilities into the pre-trained language model.

WebGLM is built by the following parts:

-   **LLM-augmented Retriever**: Enhances the retrieval of relevant web content to better aid in answering questions accurately.
-   **Bootstrapped Generator**: Generates human-like responses to questions, leveraging the power of the GLM to provide refined answers.
-   **Human Preference-aware Scorer**: Estimates the quality of generated responses by prioritizing human preferences, ensuring the system produces useful and engaging content.

This repo is the implementation of **Bootstrap Generator**.

See our [Github Repo](https://github.com/THUDM/WebGLM) for more detailed usage.