mask_type="sparsemax", # "sparsemax"
)
max_epochs = 1000
unsupervised_model.fit(
X_train=X_trn,
eval_set=[X_vld],
max_epochs=max_epochs,
patience=10,
batch_size=512,
virtual_batch_size=64,
num_workers=0,
drop_last=False,
pretraining_ratio=0.8,
)
unsupervised_model.save_model(f"{OUTPUTDIR}/pretrain")
oof_preds = []
test_preds = []
scores = []
importances = pd.DataFrame()
cv_repeat = 10
for i in range(cv_repeat):
print(f"CV {i + 1}")
oof = train.result_score.astype("float64").copy().values
kfold = KFold(n_splits=5, shuffle=True, random_state=i * 42)
_test_preds = []
for fold, (trn_index, vld_index) in enumerate(kfold.split(X, y_binary)):
X_trn, y_trn = X[trn_index], y_binary[trn_index]
X_vld, y_vld = X[vld_index], y_binary[vld_index]
print(f"Fold {fold}, data split")
mask_type="entmax", # "sparsemax"
)
max_epochs = 1000
unsupervised_model.fit(
X_train=X_trn,
eval_set=[X_vld],
max_epochs=max_epochs,
patience=25,
batch_size=2048,
virtual_batch_size=64,
num_workers=0,
drop_last=False,
pretraining_ratio=0.8,
)
unsupervised_model.save_model(f"{OUTPUTDIR}/test_pretrain_fold{fold}")
loaded_pretrain = TabNetPretrainer()
loaded_pretrain.load_model(f"{OUTPUTDIR}/test_pretrain_fold{fold}.zip")
clf = TabNetClassifier(
optimizer_fn=torch.optim.Adam,
optimizer_params=dict(lr=2e-1),
scheduler_params={"gamma": 0.95}, # how to use learning rate scheduler
scheduler_fn=torch.optim.lr_scheduler.ExponentialLR,
mask_type=
"sparsemax", # This will be overwritten if using pretrain model
)
clf.fit(
X_train=X_trn,
y_train=y_trn,