garissman/larachain 问题修复 & 功能扩展

解决BUG、新增功能、兼容多环境部署,快速响应你的开发需求

邮箱:yvsm@zunyunkeji.com | QQ:316430983 | 微信:yvsm316

garissman/larachain

最新稳定版本:1.0.7

Composer 安装命令:

composer require garissman/larachain

包简介

larachain

README 文档

README

Chatbot using LLM Models to support your product made with Laravel.

Installation

composer require garissman/larachain

Publish

After installing, you should publish the configuration file using the vendor:publish Artisan command. This command will publish the configuration file to your application's config directory:

php artisan vendor:publish --provider="Garissman\LaraChain\LaraChainServiceProvider"

Using Ollama(Free)

Go to https://ollama.com/ and follow there instructions

ones installed, download the model, at this moment the model that works with function for me is mistral-nemo so run:

ollama pull mistral-nemo

Using OpenAi(ChatGPT)

Just get your API key in: https://platform.openai.com/api-keys

and set you OPENAI_API_KEY in the .env file

Create the Default Agent

This is important to give personality to your char bot

php artisan larachain:create_default_agent

Install Horizon

Go to https://laravel.com/docs/11.x/horizon and follow there instructions, after install run:

php artisan horizon

Install Reverb

Go to https://laravel.com/docs/11.x/reverb and follow there instructions, after install run:

php artisan reverb:start --debug

Go to the UI and Chat with your bot

Now you are ready to chat,

php artisan serve

http://localhost:8000/larachain/chat

Create your tools

Now let's do actual coding, tools are the reason why y made this package, are custom code triggers by the chain and let the LLM model do the RAG base on that function output,

create a Class that extend from Garissman\LaraChain\Structures\Classes\FunctionContract,

like Garissman\LaraChain\Functions\ExampleTool

The Description is the most important part of the function, it tells the LLM when to trigger the tool call and start asking for parameter

The properties are the parameter of your tool, it tells the LLM what are the parameter, also has a description as well, very important.

TODO

  • make Function command and use RAG for the description.
  • make Agent command and use RAG for the context.
  • Knowledge Table.
  • Tests, nothing new!!!

统计信息

  • 总下载量: 122
  • 月度下载量: 0
  • 日度下载量: 0
  • 收藏数: 2
  • 点击次数: 0
  • 依赖项目数: 0
  • 推荐数: 0

GitHub 信息

  • Stars: 2
  • Watchers: 1
  • Forks: 2
  • 开发语言: PHP

其他信息

  • 授权协议: MIT
  • 更新时间: 2024-09-22