1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
|
import numpy
import pytest
from hypothesis import given, settings
from mock import MagicMock
from numpy.testing import assert_allclose
from thinc.api import SGD, Dropout, Linear, chain
from ..strategies import arrays_OI_O_BI
from ..util import get_model, get_shape
@pytest.fixture
def model():
model = Linear()
return model
def test_linear_default_name(model):
assert model.name == "linear"
def test_linear_dimensions_on_data():
X = MagicMock(shape=(5, 10), spec=numpy.ndarray)
X.ndim = 2
X.dtype = "float32"
y = MagicMock(shape=(8,), spec=numpy.ndarray)
y.ndim = 2
y.dtype = "float32"
y.max = MagicMock()
model = Linear()
model.initialize(X, y)
assert model.get_dim("nI") is not None
y.max.assert_called_with()
@given(arrays_OI_O_BI(max_batch=8, max_out=8, max_in=8))
def test_begin_update_matches_predict(W_b_input):
model = get_model(W_b_input)
nr_batch, nr_out, nr_in = get_shape(W_b_input)
W, b, input_ = W_b_input
fwd_via_begin_update, finish_update = model.begin_update(input_)
fwd_via_predict_batch = model.predict(input_)
assert_allclose(fwd_via_begin_update, fwd_via_predict_batch)
@given(arrays_OI_O_BI(max_batch=8, max_out=8, max_in=8))
def test_finish_update_calls_optimizer_with_weights(W_b_input):
model = get_model(W_b_input)
nr_batch, nr_out, nr_in = get_shape(W_b_input)
W, b, input_ = W_b_input
output, finish_update = model.begin_update(input_)
seen_keys = set()
def sgd(key, data, gradient, **kwargs):
seen_keys.add(key)
assert data.shape == gradient.shape
return data, gradient
grad_BO = numpy.ones((nr_batch, nr_out), dtype="f")
grad_BI = finish_update(grad_BO) # noqa: F841
model.finish_update(sgd)
for name in model.param_names:
assert (model.id, name) in seen_keys
@settings(max_examples=100)
@given(arrays_OI_O_BI(max_batch=8, max_out=8, max_in=8))
def test_predict_small(W_b_input):
W, b, input_ = W_b_input
nr_out, nr_in = W.shape
model = Linear(nr_out, nr_in)
model.set_param("W", W)
model.set_param("b", b)
einsummed = numpy.einsum(
"oi,bi->bo",
numpy.asarray(W, dtype="float64"),
numpy.asarray(input_, dtype="float64"),
optimize=False,
)
expected_output = einsummed + b
predicted_output = model.predict(input_)
assert_allclose(predicted_output, expected_output, rtol=0.01, atol=0.01)
@given(arrays_OI_O_BI(max_batch=20, max_out=30, max_in=30))
@settings(deadline=None)
def test_predict_extensive(W_b_input):
W, b, input_ = W_b_input
nr_out, nr_in = W.shape
model = Linear(nr_out, nr_in)
model.set_param("W", W)
model.set_param("b", b)
einsummed = numpy.einsum(
"bi,oi->bo",
numpy.asarray(input_, dtype="float32"),
numpy.asarray(W, dtype="float32"),
optimize=False,
)
expected_output = einsummed + b
predicted_output = model.predict(input_)
assert_allclose(predicted_output, expected_output, rtol=1e-04, atol=0.0001)
@given(arrays_OI_O_BI(max_batch=8, max_out=8, max_in=8))
def test_dropout_gives_zero_activations(W_b_input):
model = chain(get_model(W_b_input), Dropout(1.0))
nr_batch, nr_out, nr_in = get_shape(W_b_input)
W, b, input_ = W_b_input
fwd_dropped, _ = model.begin_update(input_)
assert all(val == 0.0 for val in fwd_dropped.flatten())
@given(arrays_OI_O_BI(max_batch=8, max_out=8, max_in=8))
def test_dropout_gives_zero_gradients(W_b_input):
model = chain(get_model(W_b_input), Dropout(1.0))
nr_batch, nr_out, nr_in = get_shape(W_b_input)
W, b, input_ = W_b_input
for node in model.walk():
if node.name == "dropout":
node.attrs["dropout_rate"] = 1.0
fwd_dropped, finish_update = model.begin_update(input_)
grad_BO = numpy.ones((nr_batch, nr_out), dtype="f")
grad_BI = finish_update(grad_BO)
assert all(val == 0.0 for val in grad_BI.flatten())
@pytest.fixture
def model2():
model = Linear(2, 2).initialize()
return model
def test_init(model2):
assert model2.get_dim("nO") == 2
assert model2.get_dim("nI") == 2
assert model2.get_param("W") is not None
assert model2.get_param("b") is not None
def test_predict_bias(model2):
input_ = model2.ops.alloc2f(1, model2.get_dim("nI"))
target_scores = model2.ops.alloc2f(1, model2.get_dim("nI"))
scores = model2.predict(input_)
assert_allclose(scores[0], target_scores[0])
# Set bias for class 0
model2.get_param("b")[0] = 2.0
target_scores[0, 0] = 2.0
scores = model2.predict(input_)
assert_allclose(scores, target_scores)
# Set bias for class 1
model2.get_param("b")[1] = 5.0
target_scores[0, 1] = 5.0
scores = model2.predict(input_)
assert_allclose(scores, target_scores)
@pytest.mark.parametrize(
"X,expected",
[
(numpy.asarray([0.0, 0.0], dtype="f"), [0.0, 0.0]),
(numpy.asarray([1.0, 0.0], dtype="f"), [1.0, 0.0]),
(numpy.asarray([0.0, 1.0], dtype="f"), [0.0, 1.0]),
(numpy.asarray([1.0, 1.0], dtype="f"), [1.0, 1.0]),
],
)
def test_predict_weights(X, expected):
W = numpy.asarray([1.0, 0.0, 0.0, 1.0], dtype="f").reshape((2, 2))
bias = numpy.asarray([0.0, 0.0], dtype="f")
model = Linear(W.shape[0], W.shape[1])
model.set_param("W", W)
model.set_param("b", bias)
scores = model.predict(X.reshape((1, -1)))
assert_allclose(scores.ravel(), expected)
def test_update():
W = numpy.asarray([1.0, 0.0, 0.0, 1.0], dtype="f").reshape((2, 2))
bias = numpy.asarray([0.0, 0.0], dtype="f")
model = Linear(2, 2)
model.set_param("W", W)
model.set_param("b", bias)
sgd = SGD(1.0, L2=0.0, grad_clip=0.0)
sgd.averages = None
ff = numpy.asarray([[0.0, 0.0]], dtype="f")
tf = numpy.asarray([[1.0, 0.0]], dtype="f")
ft = numpy.asarray([[0.0, 1.0]], dtype="f") # noqa: F841
tt = numpy.asarray([[1.0, 1.0]], dtype="f") # noqa: F841
# ff, i.e. 0, 0
scores, backprop = model.begin_update(ff)
assert_allclose(scores[0, 0], scores[0, 1])
# Tell it the answer was 'f'
gradient = numpy.asarray([[-1.0, 0.0]], dtype="f")
backprop(gradient)
for key, (param, d_param) in model.get_gradients().items():
param, d_param = sgd(key, param, d_param)
model.set_param(key[1], param)
model.set_grad(key[1], d_param)
b = model.get_param("b")
W = model.get_param("W")
assert b[0] == 1.0
assert b[1] == 0.0
# Unchanged -- input was zeros, so can't get gradient for weights.
assert W[0, 0] == 1.0
assert W[0, 1] == 0.0
assert W[1, 0] == 0.0
assert W[1, 1] == 1.0
# tf, i.e. 1, 0
scores, finish_update = model.begin_update(tf)
# Tell it the answer was 'T'
gradient = numpy.asarray([[0.0, -1.0]], dtype="f")
finish_update(gradient)
for key, (W, dW) in model.get_gradients().items():
sgd(key, W, dW)
b = model.get_param("b")
W = model.get_param("W")
assert b[0] == 1.0
assert b[1] == 1.0
# Gradient for weights should have been outer(gradient, input)
# so outer([0, -1.], [1., 0.])
# = [[0., 0.], [-1., 0.]]
assert W[0, 0] == 1.0 - 0.0
assert W[0, 1] == 0.0 - 0.0
assert W[1, 0] == 0.0 - -1.0
assert W[1, 1] == 1.0 - 0.0
|
