AAGPT - GitLite

KO | EN

gitlite — search

#javascript #python #hacktoberfest #ai #llm #machine-learning #rust #deep-learning #windows #vue #security #pytorch #ios #openai

AAGPT

(★ 136)

AAGPT is another experimental open-source application showcasing the capabilities of large language models, such as GPT-3.5 and GPT-4.

파일 탐색기

최종 버전 다운로드 (.zip)

- demo-overcooked000.png
- demo.png
- README-cn.md
- agent.py
- env.py
- utils.py
- game.yaml
- game2.yaml
- gameovercooked.yaml
aagpt-overcooked.py
aagpt.py
agent.py
env.py
LICENSE
README.md
requirements.txt
utils.py

# 설치 가이드

1. 코드 내려받기

git clone https://github.com/aialt/AAGPT

깃허브에서 프로젝트 코드 전체를 내 컴퓨터로 내려받습니다.

cd AAGPT

방금 내려받은 프로젝트 폴더 안으로 이동합니다.

2. Python

쉬움 추천

사전 준비물

Git GitHub에서 프로젝트 코드를 내려받으려면 필요합니다.
Python 3 설치 시 'Add Python to PATH' 옵션을 꼭 체크하세요 (Windows).

2. Use the following command in your terminal with pip:

이 명령어를 터미널에 그대로 입력해 실행하세요.

pip install -r requirements.txt

requirements.txt 등에 명시된 파이썬 라이브러리를 설치합니다.

✅ 에러 메시지 없이 실행되고 터미널에 안내 문구가 출력되면 정상입니다.

이 레포의 README에 적힌 실제 명령어를 그대로 가져왔습니다.

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

# AAGPT

Auto-Agent-GPT (a.k.a AAGPT) is another experimental open-source application showcasing the capabilities of large language models.

<strong>Support: general tasks, overcooked game</strong>

Language: [[English]](README.md) [[中文]](docs/README-cn.md)

<div align="center">
  <img src="assets/demo.png" width="200" height="200"><img src="assets/demo-overcooked000.png" width="200" height="200">
</div>

## Features
- Memory support
  - GPT as memory
  - Vector database as memory (requires a PINECONE API key)
- Lifespan limit for an agent (may save money)
- Support for playing the overcooked game

## Installation 
To install AAGPT, follow these steps:

1. Clone the AAGPT repository from GitHub and navigate to the downloaded folder.
    
    ```bash:
    git clone git@github.com:hyintell/AAGPT.git
    cd AAGPT
    ```
2. Use the following command in your terminal with pip:

```bash
    pip install -r requirements.txt
    ```
## Quickplay
Just two steps, you can start using AAGPT's natural language processing abilities with your OpenAI API key.

1. Open the `setup/game.yaml` file and enter your OpenAI API key in the `openai_api_key` field.

2. Navigate to the AAGPT folder and run the following command:

```bash
      python aagpt.py 
    ```
## Setup
After installing AAGPT, you will need to set up related APIs to use the application. You can do this by following these steps:
1. Navigate to the setup folder in the AAGPT directory:
    ```bash
    cd setup
    ```
2. In the `setup` folder, there are two game settings, `game.yaml` which using the Chatgpt as momery store and `game2.yaml` which using the Pinecone as momery store. You can choose one of them to set up the API.
3. In the `game.yaml` file, you will use GPT as memory store, so please fill in the following information:
* `openai_api_key`: Your OpenAI API key. If you don't have one, you can create a free account and get an API key from the OpenAI website.
* `openai_model`: The OpenAI ChatGPT model to use. Choose from "gpt-3.5-turbo", "gpt4", or "text-davinci-003".
* `env_openai_api_key`: OpenAI ChatGPT Key for env, you can keep same as the common.
* `agent_openai_api_key`: OpenAI ChatGPT Key for agents, you can keep same as the common.
* `goal`: The main objective of the AI agent.
* `init_task`: The initial tasks to be appended to the task list.
* `agent_life`: The life-time of the agents, in default, we set it to 256.

Note: Optionally, you can use `game2.yaml` which using Pinecone as memory store. In addition to the above settings, you will need to fill in the following information:
* `agent_pinecone_api_key`: The form will be a list `[Your Pinecone API , Your Pinecone Region]`, the first is pinecone API, and second will be the region of your index, you can get it from the Pinecone website.
* `agent_pinecone_index`: The index name of the Pinecone index to use. In default, we use `aagpt_agent_index`.

## Playing Overcooked
<div align="center">
  <img src="assets/demo-overcooked000.png" width="200" height="200">
</div>

1. Install [opencooking](https://github.com/hyintell/opencooking) envs.

2. Let's play

```bash
      python aagpt-overcooked.py --render
    ```

## Usage
After setting the correct APIs, you can test AAGPT by executing the `aagpt.py` file in your terminal:

```bash
python aagpt.py 
```

Once AAGPT is running, you can start interacting with it by typing in prompts and observing its responses.

If you want to change the setup or memory setting, you can use the following command:

```bash
python aagpt.py --world_root setup/game2.yaml
```

## Todo
- [ ] UI
- [ ] More memory support
- [ ] Multi-agent support
- [ ] More LLMs support

## Acknowledgement  
  
We are deeply grateful for the contributions made by open-source projects: [Auto-GPT](https://github.com/Significant-Gravitas/Auto-GPT) and [BabyAGI](https://github.com/yoheinakajima/babyagi).

## REFERENCES
- Auto-GPT
- BabyAGI
- gym-cooking
- OvercookedGPT
- overcooked_ai