lime-robot / dsb2019 Goto Github PK

Below you can find a outline of how to reproduce my solution for the 2019 Data Science Bowl competition. If you run into any trouble with the setup/code or have any questions please contact me at [email protected]

• 3rd_solution.tgz : contains original code, trained models etc

3rd_solution/
├── input/
│   ├── processed/
│   └── data-science-bowl-2019/
├── models/ # 
└── code/

• input/processed/ : will be created by the preprocessing step (by running python prepare_data.py)
• input/data-science-bowl-2019/ : raw data dir of the competition (should contain 'train.csv', 'test.csv', etc)
• models/ : contains trained models used to generate a 'submission.csv' file
• code/ : contains codes for training and prediction

• Ubuntu 18.04.4 LTS
• Python 3.7.5
• pytorch 1.3
• transformers 2.3.0 (or pytorch-transformers 1.2.0)

You can use the pip install -r requirements.txt to install the necessary packages.

• CPU: 3 x AMD® Ryzen 9 3900X (3 PCs)
• GPU: 5 x NVIDIA RTX2080Ti 11G (2 GPUs in 1 PC)
• RAM: 64G

The above is just my PC spec. In fact, GTX 1080 is enough for training.

You can generate bowl.pt, bowl_info.pt files by running prepare_data.py

.../3rd_solution/code$ python prepare_data.py

You can reproduce models already in the model/ directory by running the train.sh script. The train.sh reproduce 5 models (5-fold)

.../3rd_solution/code$ bash train.sh

You can reproduce the submissions/submission.csv by running the predict.py

.../3rd_solution/code$ python predict.py

You can reproduce the local CV score and coefficients by running the validate.py

• VALID KAPPA_SCORE:0.5684061832361282
• coefficients=[0.53060865, 1.66266655, 2.31145611]