my-ai-cv
is a project implementing a Retrieve Augmented Generation pipeline using a quantized tiny LLM with GGUF, designed to work across any device. This enables the use of the model without directly pushing it to the repository, as it is available within the container.
An Hugging Face Space version 🤗 is available here.
Before running the project, ensure you have Docker installed on your machine. If you do not have Docker, you can install it from Docker's official site.
There are two main ways to launch the chat interface:
Run the following command to pull the image from the registry and start the service:
docker run -p 7860:7860 -d ghcr.io/ilanaliouchouche/my-cv
The model currently in the container is: microsoft/Phi-3-mini-4k-instruct-gguf
-
Clone the repository and install dependencies:
git clone https://github.com/ilanaliouchouche/my-ai-cv.git cd my-ai-cv/assistant pip install -r requirements.txt
-
Ensure the
assistant/models
directory is created and populated as the model files are not pushed to the repository due to their size. Place the model files inside this directory. -
Modify the
.env
file to match the model configurations, specifically theLLM_PATH
should point to the model file within theassistant/models
directory. -
Run the application:
python app.py
Section | Description |
---|---|
Model | The project uses the microsoft/Phi-3-mini-4k-instruct-gguf model, specified in the .env file. |
LLM_PATH | It is crucial to adjust the LLM_PATH variable in the .env file to point to the correct model file in the assistant/models directory. |
Template | Inside assistant/ChatbotModel , it is strongly recommended to update the static TEMPLATE variable with instructions tailored to the model used. |
Chroma | Replace the content of assistant/chromadb with your own vector database containing the information you want to share. |
If you wish to adapt the project for your personal CV, please FORK the repo. You must ensure the LLM_PATH
is correctly set in your .env
file and update the ChatbotModel
class as necessary to align with your model's specifications.