/WeaverOfTasks

Primary LanguageJupyter NotebookMIT LicenseMIT

WeaverOfTasks

Table of Contents:

Installation

  1. First, clone the repository.
git clone https://github.com/alckasoc/WeaverOfTasks

Setting Up Environment Variables

  1. Create a .env file in the root directory. Within this .env file, define your OPENAI_API_KEY, KAGGLE_USERNAME, and KAGGLE_API_KEY.

alt text

  1. Within your root directory, create a folder called .kaggle and create a json file kaggle.json within that folder.

alt text

  1. Populate the kaggle.json like below. Make sure to fill in "username" with your KAGGLE_USERNAME and "key" with your KAGGLE_API_KEY.
{
    "username": "",
    "key": ""
}

Setting Up TaskWeaver

  1. Create a conda environment following TaskWeaver's environment creation instructions.
conda create -n taskweaver python=3.10
conda activate taskweaver
  1. Install all TaskWeaver requirements first.
cd TaskWeaver
pip install -r requirements.txt
cd ..
  1. Update the OpenAI api_key within TaskWeaver/project/taskweaver_config.json. This is the same as OPENAI_API_KEY.

alt text

Setting Up Other Requirements

  1. Next, install all requirements relevant to this repository.
pip install -r requirements.txt

Getting Started

Disclaimer: Ensure you have Docker as TaskWeaver will require this to run. Optionally, you can run with local execution with TaskWeaver. Check the docs for more information.

  1. Download the Kaggle wildfires dataset. For this step, ensure that your .kaggle/kaggle.json is correctly configured in the Installation section above. Ensure you are in the root directory.
python download.py

The above Python script downloads the dataset from Kaggle and unzips/stores it in the data/ path within the root directory.

  1. Save the unzipped data (sqlite) as a .csv. We will save both the entire wildfires.csv and a smaller version of it wildfires_lite.csv which contains only the first 10,000 rows.
python save_csv.py

Additional API Keys

You will need additional API keys to run the notebooks. Store these in the .env.

Notebooks

The relevant notebooks are in the order:

  1. data_analysis_agent.ipynb
  2. sql_agent.ipynb
  3. search_agent.ipynb

Presentation Link

You can also find the presentation in this repository as Weaver Of Tasks.pdf.

Presentation Link: https://docs.google.com/presentation/d/1FU1txbtze1f2a99njDfaE8ZCUTXv8s0ASkniDqbi8oA/edit?usp=sharing