Llamarizer

This is my submission for the LLM Hackathon 2023 organized by Streamlit.

Llamarizer is a text summarization bot, powered by the Llama2-13b model from Meta, on the Clarifai platform.

Setup

Prerequisites

Python (at least 3.9.0) must be installed on your system.

After Python has been set up, install the virtualenv package to create and manage a virtual environment for this project. This helps you maintain the project's dependencies in a hassle-free manner, without installing any unnecessary packages globally throughout your system.

pip install virtualenv

1. Clone the Project

Clone this project to create a local copy of it on your system:

git clone "https://github.com/ravi-aratchige/llamarizer.git"

Then, move into the project folder:

cd llamarizer

2. Create a Virtual Environment

Create a virtual environment inside the project folder to isolate its dependencies:

python -m venv env

# or

python3 -m venv env

Next, activate the virtual environment:

# on Windows:
.\env\Scripts\activate

# on MacOS or Linux
source env/bin/activate

You can deactivate this environment when you are done working with the project:

# on Windows, MacOS or Linux
deactivate

3. Install Dependencies

Set up your project with the necessary packages and libraries. After activating the virtual environment, enter the following command:

pip install -r requirements.txt

4. Setup Streamlit Secrets

To store your Clarifai PAT (Personal Access Token) and other sensitive information, create a folder named .streamlit within the project folder.

Next, create a file named secrets.toml within that folder. Enter the following into that file:

PAT = "a-very-random-string-of-characters"
APP_ID = "foo"
USER_ID = "bar"
WORKFLOW_ID = "foo-bar"

1. You can obtain your PAT from your Clarifai Account Settings.
2. The APP_ID is the unique ID of whatever app you have created and wish to connect to from your Streamlit app.
3. The USER_ID is simply your username.
4. The WORKFLOW_ID will be set up in the next step.

5. Setup Clarifai Workflow

After you have created an app on the Clarifai platform and completed the above steps, create a new Workflow in that app.

Next, press "Edit workflow" to add the Llama2 model to the Workflow.

In the window that opens up, drag and drop a text-to-text chip, connect it to the IN node, and set the text-to-text chip's model as one of the Llama2 models.

Finally, copy the name of the workflow and include it in the WORKFLOW_ID of .streamlit/secrets.toml that was created earlier.

6. Start Streamlit

After you have completed the above steps, you can start the Streamlit app.

streamlit run app.py

Streamlit will start up in localhost:8501.

---

This project is licensed under the Apache License.