train_set.set_target("label")
test_set.set_input("text")
test_set.set_target("label")
train_set.set_padder('text', AutoPadder_wrapper())
test_set.set_padder('text', AutoPadder_wrapper())
train_set, dev_set = train_set.split(0.1)
n = train_set.get_length()
m = len(vocab)
k = len(dataset_train.target_names)
print("Size of Train Set:", n)
print("Total Number of Words:", m)
rnn_text_model = RNN.RNN_Text(vocab_size=m, input_size=50, hidden_layer_size=128, target_size=k, dropout=0.1)
cnn_text_model = CNN.CNN_Text(vocab_size=m, input_size=50, target_size=k, dropout=0.05)
model = rnn_text_model
# ModelLoader.load_pytorch(model, "model_ckpt_large_CNN.pkl")
trainer = Trainer(
train_data=train_set,
model=model,
loss=CrossEntropyLoss(pred='pred', target='label'),
n_epochs=50,
batch_size=16,
metrics=AccuracyMetric(pred='pred', target='label'),
dev_data=dev_set,
optimizer=Adam(lr=1e-3),
callbacks=[FitlogCallback(data=test_set)]
)
trainer.train()
