Model used to predict whether images of cells come from cancer patients or healthy patients.
The data is not included in the repo. Download it from
https://www.kaggle.com/c/cancer-cell-challange/data
and add the train and test folders, as well as the train.csv file
to the data folder.
To train a model, use train.py:
python train.py --model=model_name
The following options are available:
--model: Name of the model for saving the network and figures (required)--batch_size: Number of data points to read in each batch, default 32--epochs: Number of epochs to train the model, default 5--lr: Learning rate, default 0.001--n_cpu: Number of CPUs used to load data, default 0 (only main process)--validate_epochs: How frequently validation results should be reported to figures, default 1 (every epoch)
To test a model, use test.py:
python test.py --model=model_name
The following options are available:
--model: Name of the model to load (required)--batch_size: Number of data points to read in each batch, default 32--n_cpu: Number of CPUs used to load data, default 0 (only main process)
There is a pretrained model available, basic_model. Test it using
python test.py --model=basic_model