Uploading Your Model to Dynabench

Welcome to the exciting journey of contributing your model to Dynabench! Follow these engaging steps to showcase your AI prowess:

Step 1: Download Dynalab 2.0

Download and unzip the Dynalab 2.0 project on your computer. This project is your gateway to unleashing the power of your model in the Dynabench ecosystem.

Step 2: Navigate the Repository

Explore the project repository, and you'll encounter a folder named app. Inside, discover three key folders: api, domain, and resources. The api holds logic you won't need to touch, while resources is your designated space for storing model components. Dive into the domain folder, specifically locating the model.py file.

Step 3: Customize Evaluation Functions

In the model.py file, find the ModelController class. Feel free to clear out unnecessary code, focusing on updating the constructor, single_evaluation, and batch_evaluation methods. These are the key functions where your model's magic happens.

Step 4: Add Dependencies

Ensure your model's dependencies are listed in the requirements.txt file. Don't remove any existing libraries already included.

Step 5: Test Your Model

Verify everything works seamlessly by running the following commands:

python3 -m venv env
source env/bin/activate
python3 -m pip install -r requirements.txt
python3 -m uvicorn app.main:app --reload

Open localhost:8000/docs in your favorite browser, where the FastAPI interface awaits. Test your model by clicking on the single evaluation method, hitting 'Try it out,' filling the request body, and executing. A 200 code response means you're all set!

Step 6: Zip and Upload

Once testing is successful, zip the entire repository. Now, use the 'Upload model' button below and click 'Submit Model.' Congratulations! You've officially submitted your model to Dynabench, making waves in the world of AI challenges.