tested a number of scenarios related to different types of data inputs

tests/test_model_functioning/test_data_inputs.py

+import numpy as np
+import string
+import torch
+import pytest
+
+from torchvision.transforms import ToTensor, Normalize
+
+from sklearn.model_selection import train_test_split
+from pytorch_widedeep.models.wide import Wide
+from pytorch_widedeep.models.deep_dense import DeepDense
+from pytorch_widedeep.models.deep_text import DeepText
+from pytorch_widedeep.models.deep_image import DeepImage
+from pytorch_widedeep.models.wide_deep import WideDeep
+
+import pdb
+
+# Wide array
+X_wide = np.random.choice(2, (100, 100), p = [0.8, 0.2])
+
+# Deep Array
+colnames = list(string.ascii_lowercase)[:10]
+embed_cols = [np.random.choice(np.arange(5), 100) for _ in range(5)]
+embed_input = [(u,i,j) for u,i,j in zip(colnames[:5], [5]*5, [16]*5)]
+cont_cols = [np.random.rand(100) for _ in range(5)]
+X_deep = np.vstack(embed_cols+cont_cols).transpose()
+
+# Text Array
+padded_sequences = np.random.choice(np.arange(1,100), (100, 48))
+X_text = np.hstack((np.repeat(np.array([[0,0]]), 100, axis=0), padded_sequences))
+vocab_size = 1000
+
+# Image Array
+X_img = np.random.choice(256, (100, 224, 224, 3))
+X_img_norm =  X_img/255.
+
+# Target
+target = np.random.choice(2, 100)
+
+# train/validation split
+X_wide_tr, X_wide_val, X_deep_tr, X_deep_val, X_text_tr, X_text_val, X_img_tr, X_img_val, y_train, y_val = \
+	train_test_split(X_wide, X_deep, X_text, X_img, target)
+
+# build model components
+wide = Wide(100, 1)
+deepdense = DeepDense(
+    hidden_layers=[32,16],
+    dropout=[0.5],
+    deep_column_idx={k:v for v,k in enumerate(colnames)},
+    embed_input=embed_input,
+    continuous_cols=colnames[-5:],
+    output_dim=1)
+deeptext = DeepText(
+    vocab_size=vocab_size,
+    embed_dim=32,
+    padding_idx=0
+    )
+deepimage=DeepImage(pretrained=True)
+
+# transforms
+mean = [0.406, 0.456, 0.485]  #BGR
+std =  [0.225, 0.224, 0.229]  #BGR
+transforms1 = [ToTensor, Normalize(mean=mean, std=std)]
+transforms2 = [Normalize(mean=mean, std=std)]
+
+##############################################################################
+# Test many possible scenarios of data inputs I can think off. Surely users
+# will input something unexpected
+##############################################################################
+@pytest.mark.parametrize(
+	'X_wide, X_deep, X_text, X_img, X_train, X_val, target, val_split, transforms, nepoch, null',
+	[
+	(X_wide, X_deep, X_text, X_img, None, None, target, None, transforms1, 1, None),
+	(X_wide, X_deep, X_text, X_img, None, None, target, None, transforms2, 1, None),
+	(X_wide, X_deep, X_text, X_img, None, None, target, None, None, 1, None),
+	(X_wide, X_deep, X_text, X_img_norm, None, None, target, None, transforms2, 1, None),
+	(X_wide, X_deep, X_text, X_img_norm, None, None, target, None, transforms1, 1, None),
+	(X_wide, X_deep, X_text, X_img_norm, None, None, target, None, None, 1, None),
+	(X_wide, X_deep, X_text, X_img, None, None, target, 0.2, None, 1, None),
+	(
+		None, None, None ,None, {'X_wide':X_wide, 'X_deep':X_deep, 'X_text':X_text,
+		'X_img':X_img, 'target': target}, None, None, None, None, 1, None),
+	(
+		None, None, None ,None, {'X_wide':X_wide, 'X_deep':X_deep, 'X_text':X_text,
+		'X_img':X_img, 'target': target}, None, None, None, transforms1, 1, None),
+	(
+		None, None, None, None, {'X_wide':X_wide, 'X_deep':X_deep, 'X_text':X_text,
+		'X_img': X_img, 'target': target}, None, None, 0.2, None, 1, None),
+	(
+		None, None, None, None, {'X_wide':X_wide, 'X_deep':X_deep, 'X_text':X_text,
+		'X_img': X_img, 'target': target}, None, None, 0.2, transforms2, 1, None),
+	(
+		None, None, None, None, {'X_wide':X_wide_tr, 'X_deep':X_deep_tr,
+		'X_text':X_text_tr, 'X_img': X_img_tr, 'target': y_train}, {'X_wide':X_wide_val,
+		'X_deep':X_deep_val, 'X_text':X_text_val, 'X_img': X_img_val, 'target': y_val}, None, None,
+		None, 1, None),
+	(
+		None, None, None, None, {'X_wide':X_wide_tr, 'X_deep':X_deep_tr,
+		'X_text':X_text_tr, 'X_img': X_img_tr, 'target': y_train}, {'X_wide':X_wide_val,
+		'X_deep':X_deep_val, 'X_text':X_text_val, 'X_img': X_img_val, 'target': y_val}, None, None,
+		transforms1, 1, None)
+	]
+	)
+def test_widedeep_inputs(X_wide, X_deep, X_text, X_img, X_train, X_val, target, val_split,
+	transforms, nepoch, null):
+	model = WideDeep(wide=wide, deepdense=deepdense, deeptext=deeptext, deepimage=deepimage)
+	model.compile(method='logistic', transforms=transforms)
+	model.fit(
+		X_wide = X_wide,
+		X_deep = X_deep,
+		X_text = X_text,
+		X_img  = X_img,
+		X_train= X_train,
+		X_val  = X_val,
+		target = target,
+		val_split = val_split,
+		batch_size=16,
+		verbose=0
+		)
+	assert model.history.epoch[0] == nepoch and model.history._history['train_loss'] is not null
\ No newline at end of file