datafyresearcher/datafy-finetuning-university

📖 DATAFY FINETUNING COURSE

A collection of notebooks for DATAFY FINETUNING COURSE

Learning Objectives

• Learn the fundamentals of finetuning a large language model (LLM).
• Data Preparation for Finetuning with help of Human Efforts
• Understand how finetuning differs from prompt engineering, and when to use both.
• Get practical experience with real data sets, and how to use techniques for your own projects.

🔧 Features

• Collection of Notebooks
• Google Colab
• Docker Support with Optimisation Cache etc
• Run the Notebook Server with Docker

This repo contains an notebooks flocation contains the ntebooks

💻 Generate API Key for LAMINI

1. Sign up on https://www.lamini.ai/
2. Go to Account and get the API key Under Active API Tokens
3. Under .powerml/configure_llama.yaml, add the key
4. dotenv can also be used to setup the key. Use the env.example to create env file

for more on Authentcation read https://lamini-ai.github.io/auth/

💻 Fork the Course Repositry

Fork the Course GitHub Repository and Start working. Got Below Link and Click Fork

https://github.com/datafyresearcher/datafy-finetuning-university

, Change to your Own Username for below commands

💻 Running on Google Colab

Google Colab offer free GPU therefore, running on Google Colab is prefferd method

1. To get started, you need to mount your Google Drive in Colab. You can do this by running the following command:

from google.colab import drive
drive.mount('/content/drive')

2. Once your Drive is mounted, you can clone a repository by running the following command:

!git clone https://github.com/datafyresearcher/datafy-finetuning-university.git /content/drive/MyDrive/datafy-finetuning-university

This will clone the repository to your Colab instance, and you can work on it just as you would on your local machine. When you want to push your changes back to the repository, you can use the usual Git commands, such as:

!git add .
!git commit -m "Commit message"
!git push origin <branch>

💻 Running Locally

1. Clone the repository📂

git clone https://github.com/datafyresearcher/datafy-finetuning-university.git

2. Install dependencies with Poetry and activate virtual environment🔨

poetry install
poetry shell

3. Copy and Modify env.example to .env

Generate the HF API Key to be able to hosted models for inference and set the variables accordingly.

4. Run the JupyterLab server🚀

jupyter lab

Run Notebooks using Docker

This project includes Dockerfile to run the app in Docker container. In order to optimise the Docker Image size and building time with cache techniques, I have follow tricks in below Article https://medium.com/@albertazzir/blazing-fast-python-docker-builds-with-poetry-a78a66f5aed0

Build the docker container

docker  build . -t datafy-finetuning-course:latest 

To generate Image with DOCKER_BUILDKIT, follow below command

DOCKER_BUILDKIT=1 docker build --target=runtime . -t datafy-finetuning-course:latest

1. Run the docker container directly

docker run -d --name datafy-finetuning-course -p 8888:8888 datafy-finetuning-course:latest 

2. Run the docker container using docker-compose (Recommended)

docker-compose up

Make sure to include the .env SECRETS file when running with docker-compose with your own Keys.

Report Feedbacks

As dlai-hf-course:latest is a template project with minimal example. Report issues if you face any.

DISCLAIMER & CREDITS

We have collected the Notebooks from original course and edited for few lines/functions to make them run locally.