https://github.com/dataprofessor/llama2

This chatbot app is built using the Llama 2 open source LLM from Meta.
https://github.com/dataprofessor/llama2

large-language-models llama2 llm meta python streamlit

Last synced: 6 days ago
JSON representation

This chatbot app is built using the Llama 2 open source LLM from Meta.

• Host: GitHub
• URL: https://github.com/dataprofessor/llama2
• Owner: dataprofessor
• Created: 2023-07-19T10:13:21.000Z (over 1 year ago)
• Default Branch: master
• Last Pushed: 2024-09-20T06:12:21.000Z (4 months ago)
• Last Synced: 2025-01-17T18:06:03.777Z (13 days ago)
• Topics: large-language-models, llama2, llm, meta, python, streamlit
• Language: Python
• Homepage: https://llama2.streamlit.app
• Size: 62.5 KB
• Stars: 354
• Watchers: 3
• Forks: 639
• Open Issues: 9
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# 🦙💬 Llama 2 Chat

This chatbot is created using the open-source Llama 2 LLM model from Meta.

Particularly, we're using the [**Llama2-7B**](https://replicate.com/a16z-infra/llama7b-v2-chat) model deployed by the Andreessen Horowitz (a16z) team and hosted on the [Replicate](https://replicate.com/) platform.

This app was refactored from [a16z's implementation](https://github.com/a16z-infra/llama2-chatbot) of their [LLaMA2 Chatbot](https://www.llama2.ai/) to be light-weight for deployment to the [Streamlit Community Cloud](https://streamlit.io/cloud).

## Demo App

[![Streamlit App](https://static.streamlit.io/badges/streamlit_badge_black_white.svg)](https://llama2.streamlitapp.com/)

## Prerequisite libraries

```

streamlit

replicate

```

## Getting your own Replicate API token

To use this app, you'll need to get your own [Replicate](https://replicate.com/) API token.

After signing up to Replicate, you can access your API token from [this page](https://replicate.com/account/api-tokens).

## Other Llama 2 models to try

As mentioned above, this chatbot implementation uses the [**Llama2-7B**](https://replicate.com/a16z-infra/llama7b-v2-chat) model that was trained on 7 billion parameters.

You can also try out the larger models:

- [Llama2-13B](https://replicate.com/a16z-infra/llama13b-v2-chat)

- [Llama2-70B](https://replicate.com/replicate/llama70b-v2-chat)

## Further Reading

- [Llama 2 website](https://ai.meta.com/llama/)

- [Llama 2 technical overview](https://ai.meta.com/resources/models-and-libraries/llama/)

- [Llama 2 blog](https://ai.meta.com/blog/llama-2/)

- [Llama 2 research article](https://ai.meta.com/research/publications/llama-2-open-foundation-and-fine-tuned-chat-models/)

- [Llama 2 GitHub repo](https://github.com/facebookresearch/llama/tree/main)