fuumin621 / stanford-ribonanza-rna-folding-4th

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

4th Place Solution (monnu's part)

Environment

Use the following commands to set up the environment:

docker-compose up -d --build
docker exec -it kaggle bash

Usage

  1. Data Preparation: Place competition data in the input directory.

  2. Base Pair Matrix (EternaFold): Generate a base pair matrix using EternaFold. For this step, use scripts provided by my teammate tattaka.

  3. Training: Run the following scripts to train the models.

    cd src/exp112 && sh train.sh
    cd src/exp300 && sh train.sh
    cd src/exp302 && sh train.sh
    cd src/exp312 && sh train.sh
    cd src/exp317 && sh train.sh
  4. Ensemble: Generate out-of-fold (oof) predictions and submissions using all 12 models from our team. Then, execute the ensemble script:

    python src/scripts/ensemble.py

License

MIT

About

License:MIT License


Languages

Language:Python 98.4%Language:Shell 1.5%Language:Dockerfile 0.1%