Chat-LLM2

This project is a proof of concept to develop a web application that integrates Llama2 technology using Docker, solely utilizing the CPU. Inspired by the Chat-GPT interface, this application offers an interactive and user-friendly experience. The entire project, including this readme.md, has been generated using GPT-4. Please note that due to its reliance on CPU processing, the application's response time may be slower than usual.

This section displays various images used in the application, located in the images directory.

• Docker

Step-by-step instructions to set up the development environment.

The Dockerfile establishes a Python environment and exposes port 5000, preparing the system to run the Flask server.

It is essential to download at least one model before starting the server. Execute the following command to download at least one model:

./tasks.sh download_model

Use the build and run commands from the tasks.sh file to build the docker image and then run it in a container:

./tasks.sh build
./tasks.sh run

To remove unused Docker images:

./tasks.sh clean

llama2_api.py sets up a Flask server, loading Llama models from /models/ and providing API endpoints for interaction.

• GET /api/v1/models/: Lists available models.
• POST /api/v1/models/generate: Generates text using a specific model.

The application features a web interface accessible at localhost:5000/. This interface is similar to that of Chat-GPT, providing a fluid and accessible user experience.

For support or collaboration, you can contact me at: [email protected]