/
model.py
365 lines (314 loc) · 20.4 KB
/
model.py
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
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
# -*- coding: utf-8 -*-
import numpy as np
import chainer
from chainer import cuda, Function, Variable
from chainer import Link, Chain, ChainList
import chainer.functions as F
import chainer.links as L
from src.lib.loss import softmax_dice_loss
class Model_L2(Chain):
def __init__(
self,
ndim=3,
n_class=2,
init_channel=2,
kernel_size=3,
pool_size=2,
ap_factor=2,
gpu=-1,
class_weight=np.array([1, 1]).astype(np.float32),
loss_func='F.softmax_cross_entropy'
):
self.gpu = gpu
self.pool_size = pool_size
if gpu >= 0:
self.class_weight = cuda.to_gpu(np.array(class_weight).astype(np.float32))
else:
self.class_weight = np.array(class_weight).astype(np.float32)
self.train = True
self.loss_func = loss_func
initializer = chainer.initializers.HeNormal()
super(Model_L2, self).__init__(
c0=L.ConvolutionND(ndim, 1, init_channel, kernel_size, 1, int(kernel_size/2), initialW=initializer, initial_bias=None),
c1=L.ConvolutionND(ndim, init_channel, int(init_channel * (ap_factor ** 1)), kernel_size, 1, int(kernel_size/2), initialW=initializer, initial_bias=None),
c2=L.ConvolutionND(ndim, int(init_channel * (ap_factor ** 1)), int(init_channel * (ap_factor ** 1)), kernel_size, 1, int(kernel_size/2), initialW=initializer, initial_bias=None),
c3=L.ConvolutionND(ndim, int(init_channel * (ap_factor ** 1)), int(init_channel * (ap_factor ** 2)), kernel_size, 1, int(kernel_size/2), initialW=initializer, initial_bias=None),
c4=L.ConvolutionND(ndim, int(init_channel * (ap_factor ** 2)), int(init_channel * (ap_factor ** 2)), kernel_size, 1, int(kernel_size/2), initialW=initializer, initial_bias=None),
c5=L.ConvolutionND(ndim, int(init_channel * (ap_factor ** 2)), int(init_channel * (ap_factor ** 3)), kernel_size, 1, int(kernel_size/2), initialW=initializer, initial_bias=None),
dc0=L.DeconvolutionND(ndim, int(init_channel * (ap_factor ** 3)), int(init_channel * (ap_factor ** 3)), self.pool_size, self.pool_size, 0, initialW=initializer, initial_bias=None),
dc1=L.ConvolutionND(ndim, int(init_channel * (ap_factor ** 2) + init_channel * (ap_factor ** 3)), int(init_channel * (ap_factor ** 2)), kernel_size, 1, int(kernel_size/2), initialW=initializer, initial_bias=None),
dc2=L.ConvolutionND(ndim, int(init_channel * (ap_factor ** 2)), int(init_channel * (ap_factor ** 2)), kernel_size, 1, int(kernel_size/2), initialW=initializer, initial_bias=None),
dc3=L.DeconvolutionND(ndim, int(init_channel * (ap_factor ** 2)), int(init_channel * (ap_factor ** 2)), self.pool_size, self.pool_size, 0, initialW=initializer, initial_bias=None),
dc4=L.ConvolutionND(ndim, int(init_channel * (ap_factor ** 1) + init_channel * (ap_factor ** 2)), int(init_channel * (ap_factor ** 1)), kernel_size, 1, int(kernel_size/2), initialW=initializer, initial_bias=None),
dc5=L.ConvolutionND(ndim, int(init_channel * (ap_factor ** 1)), int(init_channel * (ap_factor ** 1)), kernel_size, 1, int(kernel_size/2), initialW=initializer, initial_bias=None),
dc6=L.ConvolutionND(ndim, int(init_channel * (ap_factor ** 1)), n_class, 1, 1, initialW=initializer, initial_bias=None),
bnc0=L.BatchNormalization(init_channel),
bnc1=L.BatchNormalization(int(init_channel * (ap_factor ** 1))),
bnc2=L.BatchNormalization(int(init_channel * (ap_factor ** 1))),
bnc3=L.BatchNormalization(int(init_channel * (ap_factor ** 2))),
bnc4=L.BatchNormalization(int(init_channel * (ap_factor ** 2))),
bnc5=L.BatchNormalization(int(init_channel * (ap_factor ** 3))),
bndc1=L.BatchNormalization(int(init_channel * (ap_factor ** 2))),
bndc2=L.BatchNormalization(int(init_channel * (ap_factor ** 2))),
bndc4=L.BatchNormalization(int(init_channel * (ap_factor ** 1))),
bndc5=L.BatchNormalization(int(init_channel * (ap_factor ** 1)))
)
def _calc(self, x):
e0 = F.relu(self.bnc0(self.c0(x)))
syn0 = F.relu(self.bnc1(self.c1(e0)))
del e0
e1 = F.max_pooling_nd(syn0, self.pool_size, self.pool_size)
e2 = F.relu(self.bnc2(self.c2(e1)))
syn1 = F.relu(self.bnc3(self.c3(e2)))
del e1, e2
e3 = F.max_pooling_nd(syn1, self.pool_size, self.pool_size)
e4 = F.relu(self.bnc4(self.c4(e3)))
e5 = F.relu(self.bnc5(self.c5(e4)))
del e3, e4
d0 = F.concat([self.dc0(e5), syn1])
del e5, syn1
d1 = F.relu(self.bndc1(self.dc1(d0)))
d2 = F.relu(self.bndc2(self.dc2(d1)))
del d0, d1
d3 = F.concat([self.dc3(d2), syn0])
del d2, syn0
d4 = F.relu(self.bndc4(self.dc4(d3)))
d5 = F.relu(self.bndc5(self.dc5(d4)))
del d3, d4
d6 = self.dc6(d5)
del d5
return d6
def __call__(self, x, t=None, seg=True):
h = self._calc(x)
if seg:
pred = F.softmax(h)
del h
return pred.data
else:
#loss = eval(self.loss_func)(h, t, class_weight=self.class_weight)
loss = eval(self.loss_func)(h, t)
pred = F.softmax(h)
del h
return loss, pred.data
class Model_L3(Chain):
def __init__(
self,
ndim=3,
n_class=2,
init_channel=2,
kernel_size=3,
pool_size=2,
ap_factor=2,
gpu=-1,
class_weight=np.array([1, 1]).astype(np.float32),
loss_func='F.softmax_cross_entropy'
):
self.gpu = gpu
self.pool_size = pool_size
if gpu >= 0:
self.class_weight = cuda.to_gpu(np.array(class_weight).astype(np.float32))
else:
self.class_weight = np.array(class_weight).astype(np.float32)
self.train = True
self.loss_func = loss_func
initializer = chainer.initializers.HeNormal()
super(Model_L3, self).__init__(
c0=L.ConvolutionND(ndim, 1, init_channel, kernel_size, 1, int(kernel_size/2), initialW=initializer, initial_bias=None),
c1=L.ConvolutionND(ndim, init_channel, int(init_channel * (ap_factor ** 1)), kernel_size, 1, int(kernel_size/2), initialW=initializer, initial_bias=None),
c2=L.ConvolutionND(ndim, int(init_channel * (ap_factor ** 1)), int(init_channel * (ap_factor ** 1)), kernel_size, 1, int(kernel_size/2), initialW=initializer, initial_bias=None),
c3=L.ConvolutionND(ndim, int(init_channel * (ap_factor ** 1)), int(init_channel * (ap_factor ** 2)), kernel_size, 1, int(kernel_size/2), initialW=initializer, initial_bias=None),
c4=L.ConvolutionND(ndim, int(init_channel * (ap_factor ** 2)), int(init_channel * (ap_factor ** 2)), kernel_size, 1, int(kernel_size/2), initialW=initializer, initial_bias=None),
c5=L.ConvolutionND(ndim, int(init_channel * (ap_factor ** 2)), int(init_channel * (ap_factor ** 3)), kernel_size, 1, int(kernel_size/2), initialW=initializer, initial_bias=None),
c6=L.ConvolutionND(ndim, int(init_channel * (ap_factor ** 3)), int(init_channel * (ap_factor ** 3)), kernel_size, 1, int(kernel_size/2), initialW=initializer, initial_bias=None),
c7=L.ConvolutionND(ndim, int(init_channel * (ap_factor ** 3)), int(init_channel * (ap_factor ** 4)), kernel_size, 1, int(kernel_size/2), initialW=initializer, initial_bias=None),
dc0=L.DeconvolutionND(ndim, int(init_channel * (ap_factor ** 4)), int(init_channel * (ap_factor ** 4)), self.pool_size, self.pool_size, 0, initialW=initializer, initial_bias=None),
dc1=L.ConvolutionND(ndim, int(init_channel * (ap_factor ** 3) + init_channel * (ap_factor ** 4)), int(init_channel * (ap_factor ** 3)), kernel_size, 1, int(kernel_size/2), initialW=initializer, initial_bias=None),
dc2=L.ConvolutionND(ndim, int(init_channel * (ap_factor ** 3)), int(init_channel * (ap_factor ** 3)), kernel_size, 1, int(kernel_size/2), initialW=initializer, initial_bias=None),
dc3=L.DeconvolutionND(ndim, int(init_channel * (ap_factor ** 3)), int(init_channel * (ap_factor ** 3)), self.pool_size, self.pool_size, 0, initialW=initializer, initial_bias=None),
dc4=L.ConvolutionND(ndim, int(init_channel * (ap_factor ** 2) + init_channel * (ap_factor ** 3)), int(init_channel * (ap_factor ** 2)), kernel_size, 1, int(kernel_size/2), initialW=initializer, initial_bias=None),
dc5=L.ConvolutionND(ndim, int(init_channel * (ap_factor ** 2)), int(init_channel * (ap_factor ** 2)), kernel_size, 1, int(kernel_size/2), initialW=initializer, initial_bias=None),
dc6=L.DeconvolutionND(ndim, int(init_channel * (ap_factor ** 2)), int(init_channel * (ap_factor ** 2)), self.pool_size, self.pool_size, 0, initialW=initializer, initial_bias=None),
dc7=L.ConvolutionND(ndim, int(init_channel * (ap_factor ** 1) + init_channel * (ap_factor ** 2)), int(init_channel * (ap_factor ** 1)), kernel_size, 1, int(kernel_size/2), initialW=initializer, initial_bias=None),
dc8=L.ConvolutionND(ndim, int(init_channel * (ap_factor ** 1)), int(init_channel * (ap_factor ** 1)), kernel_size, 1, int(kernel_size/2), initialW=initializer, initial_bias=None),
dc9=L.ConvolutionND(ndim, int(init_channel * (ap_factor ** 1)), n_class, 1, 1, initialW=initializer, initial_bias=None),
bnc0=L.BatchNormalization(init_channel),
bnc1=L.BatchNormalization(int(init_channel * (ap_factor ** 1))),
bnc2=L.BatchNormalization(int(init_channel * (ap_factor ** 1))),
bnc3=L.BatchNormalization(int(init_channel * (ap_factor ** 2))),
bnc4=L.BatchNormalization(int(init_channel * (ap_factor ** 2))),
bnc5=L.BatchNormalization(int(init_channel * (ap_factor ** 3))),
bnc6=L.BatchNormalization(int(init_channel * (ap_factor ** 3))),
bnc7=L.BatchNormalization(int(init_channel * (ap_factor ** 4))),
bndc1=L.BatchNormalization(int(init_channel * (ap_factor ** 3))),
bndc2=L.BatchNormalization(int(init_channel * (ap_factor ** 3))),
bndc4=L.BatchNormalization(int(init_channel * (ap_factor ** 2))),
bndc5=L.BatchNormalization(int(init_channel * (ap_factor ** 2))),
bndc7=L.BatchNormalization(int(init_channel * (ap_factor ** 1))),
bndc8=L.BatchNormalization(int(init_channel * (ap_factor ** 1)))
)
def _calc(self, x):
e0 = F.relu(self.bnc0(self.c0(x)))
syn0 = F.relu(self.bnc1(self.c1(e0)))
del e0
e1 = F.max_pooling_nd(syn0, self.pool_size, self.pool_size)
e2 = F.relu(self.bnc2(self.c2(e1)))
syn1 = F.relu(self.bnc3(self.c3(e2)))
del e1, e2
e3 = F.max_pooling_nd(syn1, self.pool_size, self.pool_size)
e4 = F.relu(self.bnc4(self.c4(e3)))
syn2 = F.relu(self.bnc5(self.c5(e4)))
del e3, e4
e5 = F.max_pooling_nd(syn2, self.pool_size, self.pool_size)
e6 = F.relu(self.bnc6(self.c6(e5)))
e7 = F.relu(self.bnc7(self.c7(e6)))
del e5, e6
d0 = F.concat([self.dc0(e7), syn2])
del e7, syn2
d1 = F.relu(self.bndc1(self.dc1(d0)))
d2 = F.relu(self.bndc2(self.dc2(d1)))
del d0, d1
d3 = F.concat([self.dc3(d2), syn1])
del d2, syn1
d4 = F.relu(self.bndc4(self.dc4(d3)))
d5 = F.relu(self.bndc5(self.dc5(d4)))
del d3, d4
d6 = F.concat([self.dc6(d5), syn0])
del d5, syn0
d7 = F.relu(self.bndc7(self.dc7(d6)))
d8 = F.relu(self.bndc8(self.dc8(d7)))
del d6, d7
d9 = self.dc9(d8)
del d8
return d9
def __call__(self, x, t=None, seg=True):
h = self._calc(x)
if seg:
pred = F.softmax(h)
del h
return pred.data
else:
#loss = eval(self.loss_func)(h, t, class_weight=self.class_weight)
loss = eval(self.loss_func)(h, t)
pred = F.softmax(h)
del h
return loss, pred.data
class Model_L4(Chain):
def __init__(
self,
ndim=3,
n_class=2,
init_channel=2,
kernel_size=3,
pool_size=2,
ap_factor=2,
gpu=-1,
class_weight=np.array([1, 1]).astype(np.float32),
loss_func='F.softmax_cross_entropy'
):
self.gpu = gpu
self.pool_size = pool_size
if gpu >= 0:
self.class_weight = cuda.to_gpu(np.array(class_weight).astype(np.float32))
else:
self.class_weight = np.array(class_weight).astype(np.float32)
self.train = True
self.loss_func = loss_func
initializer = chainer.initializers.HeNormal()
super(Model_L4, self).__init__(
c0=L.ConvolutionND(ndim, 1, init_channel, kernel_size, 1, int(kernel_size/2), initialW=initializer, initial_bias=None),
c1=L.ConvolutionND(ndim, init_channel, int(init_channel * (ap_factor ** 1)), kernel_size, 1, int(kernel_size/2), initialW=initializer, initial_bias=None),
c2=L.ConvolutionND(ndim, int(init_channel * (ap_factor ** 1)), int(init_channel * (ap_factor ** 1)), kernel_size, 1, int(kernel_size/2), initialW=initializer, initial_bias=None),
c3=L.ConvolutionND(ndim, int(init_channel * (ap_factor ** 1)), int(init_channel * (ap_factor ** 2)), kernel_size, 1, int(kernel_size/2), initialW=initializer, initial_bias=None),
c4=L.ConvolutionND(ndim, int(init_channel * (ap_factor ** 2)), int(init_channel * (ap_factor ** 2)), kernel_size, 1, int(kernel_size/2), initialW=initializer, initial_bias=None),
c5=L.ConvolutionND(ndim, int(init_channel * (ap_factor ** 2)), int(init_channel * (ap_factor ** 3)), kernel_size, 1, int(kernel_size/2), initialW=initializer, initial_bias=None),
c6=L.ConvolutionND(ndim, int(init_channel * (ap_factor ** 3)), int(init_channel * (ap_factor ** 3)), kernel_size, 1, int(kernel_size/2), initialW=initializer, initial_bias=None),
c7=L.ConvolutionND(ndim, int(init_channel * (ap_factor ** 3)), int(init_channel * (ap_factor ** 4)), kernel_size, 1, int(kernel_size/2), initialW=initializer, initial_bias=None),
c8=L.ConvolutionND(ndim, int(init_channel * (ap_factor ** 4)), int(init_channel * (ap_factor ** 4)), kernel_size, 1, int(kernel_size/2), initialW=initializer, initial_bias=None),
c9=L.ConvolutionND(ndim, int(init_channel * (ap_factor ** 4)), int(init_channel * (ap_factor ** 5)), kernel_size, 1, int(kernel_size/2), initialW=initializer, initial_bias=None),
dc0=L.DeconvolutionND(ndim, int(init_channel * (ap_factor ** 5)), int(init_channel * (ap_factor ** 5)), self.pool_size, self.pool_size, 0, initialW=initializer, initial_bias=None),
dc1=L.ConvolutionND(ndim, int(init_channel * (ap_factor ** 4) + init_channel * (ap_factor ** 5)), int(init_channel * (ap_factor ** 4)), kernel_size, 1, int(kernel_size/2), initialW=initializer, initial_bias=None),
dc2=L.ConvolutionND(ndim, int(init_channel * (ap_factor ** 4)), int(init_channel * (ap_factor ** 4)), kernel_size, 1, int(kernel_size/2), initialW=initializer, initial_bias=None),
dc3=L.DeconvolutionND(ndim, int(init_channel * (ap_factor ** 4)), int(init_channel * (ap_factor ** 4)), self.pool_size, self.pool_size, 0, initialW=initializer, initial_bias=None),
dc4=L.ConvolutionND(ndim, int(init_channel * (ap_factor ** 3) + init_channel * (ap_factor ** 4)), int(init_channel * (ap_factor ** 3)), kernel_size, 1, int(kernel_size/2), initialW=initializer, initial_bias=None),
dc5=L.ConvolutionND(ndim, int(init_channel * (ap_factor ** 3)), int(init_channel * (ap_factor ** 3)), kernel_size, 1, int(kernel_size/2), initialW=initializer, initial_bias=None),
dc6=L.DeconvolutionND(ndim, int(init_channel * (ap_factor ** 3)), int(init_channel * (ap_factor ** 3)), self.pool_size, self.pool_size, 0, initialW=initializer, initial_bias=None),
dc7=L.ConvolutionND(ndim, int(init_channel * (ap_factor ** 2) + init_channel * (ap_factor ** 3)), int(init_channel * (ap_factor ** 2)), kernel_size, 1, int(kernel_size/2), initialW=initializer, initial_bias=None),
dc8=L.ConvolutionND(ndim, int(init_channel * (ap_factor ** 2)), int(init_channel * (ap_factor ** 2)), kernel_size, 1, int(kernel_size/2), initialW=initializer, initial_bias=None),
dc9=L.DeconvolutionND(ndim, int(init_channel * (ap_factor ** 2)), int(init_channel * (ap_factor ** 2)), self.pool_size, self.pool_size, 0, initialW=initializer, initial_bias=None),
dc10=L.ConvolutionND(ndim, int(init_channel * (ap_factor ** 1) + init_channel * (ap_factor ** 2)), int(init_channel * (ap_factor ** 1)), kernel_size, 1, int(kernel_size/2), initialW=initializer, initial_bias=None),
dc11=L.ConvolutionND(ndim, int(init_channel * (ap_factor ** 1)), int(init_channel * (ap_factor ** 1)), kernel_size, 1, int(kernel_size/2), initialW=initializer, initial_bias=None),
dc12=L.ConvolutionND(ndim, int(init_channel * (ap_factor ** 1)), n_class, 1, 1, initialW=initializer, initial_bias=None),
bnc0=L.BatchNormalization(init_channel),
bnc1=L.BatchNormalization(int(init_channel * (ap_factor ** 1))),
bnc2=L.BatchNormalization(int(init_channel * (ap_factor ** 1))),
bnc3=L.BatchNormalization(int(init_channel * (ap_factor ** 2))),
bnc4=L.BatchNormalization(int(init_channel * (ap_factor ** 2))),
bnc5=L.BatchNormalization(int(init_channel * (ap_factor ** 3))),
bnc6=L.BatchNormalization(int(init_channel * (ap_factor ** 3))),
bnc7=L.BatchNormalization(int(init_channel * (ap_factor ** 4))),
bnc8=L.BatchNormalization(int(init_channel * (ap_factor ** 4))),
bnc9=L.BatchNormalization(int(init_channel * (ap_factor ** 5))),
bndc1=L.BatchNormalization(int(init_channel * (ap_factor ** 4))),
bndc2=L.BatchNormalization(int(init_channel * (ap_factor ** 4))),
bndc4=L.BatchNormalization(int(init_channel * (ap_factor ** 3))),
bndc5=L.BatchNormalization(int(init_channel * (ap_factor ** 3))),
bndc7=L.BatchNormalization(int(init_channel * (ap_factor ** 2))),
bndc8=L.BatchNormalization(int(init_channel * (ap_factor ** 2))),
bndc10=L.BatchNormalization(int(init_channel * (ap_factor ** 1))),
bndc11=L.BatchNormalization(int(init_channel * (ap_factor ** 1)))
)
def _calc(self, x):
e0 = F.relu(self.bnc0(self.c0(x)))
syn0 = F.relu(self.bnc1(self.c1(e0)))
del e0
e1 = F.max_pooling_nd(syn0, self.pool_size, self.pool_size)
e2 = F.relu(self.bnc2(self.c2(e1)))
syn1 = F.relu(self.bnc3(self.c3(e2)))
del e1, e2
e3 = F.max_pooling_nd(syn1, self.pool_size, self.pool_size)
e4 = F.relu(self.bnc4(self.c4(e3)))
syn2 = F.relu(self.bnc5(self.c5(e4)))
del e3, e4
e5 = F.max_pooling_nd(syn2, self.pool_size, self.pool_size)
e6 = F.relu(self.bnc6(self.c6(e5)))
syn3 = F.relu(self.bnc7(self.c7(e6)))
del e5, e6
e7 = F.max_pooling_nd(syn3, self.pool_size, self.pool_size)
e8 = F.relu(self.bnc8(self.c8(e7)))
e9 = F.relu(self.bnc9(self.c9(e8)))
del e7, e8
d0 = F.concat([self.dc0(e9), syn3])
del e9, syn3
d1 = F.relu(self.bndc1(self.dc1(d0)))
d2 = F.relu(self.bndc2(self.dc2(d1)))
del d0, d1
d3 = F.concat([self.dc3(d2), syn2])
del d2, syn2
d4 = F.relu(self.bndc4(self.dc4(d3)))
d5 = F.relu(self.bndc5(self.dc5(d4)))
del d3, d4
d6 = F.concat([self.dc6(d5), syn1])
del d5, syn1
d7 = F.relu(self.bndc7(self.dc7(d6)))
d8 = F.relu(self.bndc8(self.dc8(d7)))
del d6, d7
d9 = F.concat([self.dc9(d8), syn0])
del d8, syn0
d10 = F.relu(self.bndc10(self.dc10(d9)))
d11 = F.relu(self.bndc11(self.dc11(d10)))
del d9, d10
d12 = self.dc12(d11)
del d11
return d12
def __call__(self, x, t=None, seg=True):
h = self._calc(x)
if seg:
pred = F.softmax(h)
del h
return pred.data
else:
#loss = eval(self.loss_func)(h, t, class_weight=self.class_weight)
loss = eval(self.loss_func)(h, t)
pred = F.softmax(h)
del h
return loss, pred.data