Skip to content

Latest commit

 

History

History
40 lines (34 loc) · 1.46 KB

README.md

File metadata and controls

40 lines (34 loc) · 1.46 KB

HearSay: a conversational sports insider

Background

HearSay makes staying on top of your favorite sports teams and players easy with user profiles and fluid question and answering. See our live demo to get a feel for what HearSay can do!

Development

To use and tinker with HearSay locally, follow these steps:

  1. Clone this repository
  2. Determine whether you want to use our fine-tuned sentiment-analyzation & entity-extraction models, or if you'd rather use the base models via the HuggingFace API.
  3. If fine-tuned, reach out to bera@umich.edu for access to the files (we don't host them in this repo). Then, in your local project directory create an environment variable using:
export FINE_TUNED=1

If API, create the following environment variables instead:

export FINE_TUNED=0
export HUGGINGFACE_API_KEY="api key you create using link above"
  1. In either case, you need to create a serp api key and create an environment variable using
export SERP_API_KEY="api key"
  1. With these prerequisites met, you can run HearSay using these commands in separate shell windows:
# back-end
python3 -m venv env
source env/bin/activate && pip3 install -r requirements.txt
python3 main.py
// front-end
cd hearsay-client
npm install
npm start
  1. Navigate to localhost:3000 and start using HearSay!