forked from HQXie0910/Lite-DeepSC
-
Notifications
You must be signed in to change notification settings - Fork 0
/
prune.py
252 lines (201 loc) · 8.55 KB
/
prune.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
"""
This is prune file
only prune the transeiver file
"""
import os
import time
import utils
import pickle
import torch
import torch.nn as nn
import numpy as np
from hyperparameter import Hyperparameter as hp
from transformer_constellation import Transformer
from torchtext import data
from torchtext.data import Field, Dataset
from torchtext.data import BucketIterator
from adnet import ADNet
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
checkpoint_path = './pre-trained Model'
PATH = checkpoint_path + './traceiver-Rician-imperfect_CSI-4bits' + '.pth.tar'
stored_path = './Pruned Model'
label = '60'
prune_path = stored_path + './traceiver-Rician-imperfect_CSI-4bits-' + label + '.pth.tar'
""" preparing the dataset """
raw_data = pickle.load(open('./data/train_data.pkl', 'rb'))
train_data, test_data = raw_data
tokenize = lambda x: x.split()
#tokenize = get_tokenizer("basic_english"),
# pad_token = '<pad>'
SRC = Field(sequential=True, use_vocab = True,
tokenize = tokenize,
init_token = '<start>',
eos_token = '<end>',
fix_length = hp.MAX_LENGTH,
lower = True,
batch_first = True)
TRG = Field(sequential=True, use_vocab = True,
tokenize = tokenize,
init_token = '<start>',
eos_token = '<end>',
fix_length = hp.MAX_LENGTH,
lower = True,
batch_first = True)
examples_train = []
examples_test = []
fields = [('src', SRC), ('trg', TRG)]
for sent in train_data:
examples_train.append(data.Example.fromlist([sent, sent], fields))
for sent in test_data:
examples_test.append(data.Example.fromlist([sent, sent], fields))
train = Dataset(examples = examples_train, fields = fields)
test = Dataset(examples = examples_test, fields = fields)
SRC.build_vocab(train.src, test.src)
TRG.build_vocab(train.trg, test.trg)
train_iterator = BucketIterator(train, batch_size = hp.BATCH_SIZE,
device = device, sort_key = lambda x: len(x.src), # field sorted by len
sort_within_batch = True, shuffle = True)
test_iterator = BucketIterator(test, batch_size = hp.BATCH_SIZE,
device = device, sort_key = lambda x: len(x.src), # field sorted by len
sort_within_batch = True, shuffle = True)
pad_idx = TRG.vocab.stoi["<pad>"]
start_symbol = SRC.vocab.stoi["<start>"]
vocb_dictionary = SRC.vocab.stoi.items()
end_idx = SRC.vocab.stoi["<end>"]
StoT = utils.SeqtoText(vocb_dictionary, end_idx)
bleu_score_1gram = utils.BleuScore(1, 0, 0, 0)
channel = 'Rician' #'AWGN', 'Rician', 'Rayleigh'
CSI = True
imperfect = True
is_csinet = True
if os.path.exists(checkpoint_path + '/ADNet/' + '/model-B.pth'):
# load existing model
model_info = torch.load(checkpoint_path + '/ADNet/' + '/model-B.pth')
print('==> loading existing model:', checkpoint_path + '/ADNet/' + '/model-B.pth')
CSINet = ADNet(channels = 1)
CSINet.cuda()
CSINet.load_state_dict(model_info)
else:
print('No Model')
def main():
model = Transformer(hp.num_layers, len(SRC.vocab), len(TRG.vocab),
len(SRC.vocab), len(TRG.vocab), hp.d_model, hp.num_heads,
hp.dff, hp.dropout_rate).to(device)
# Load checkpoint.
print('==> Resuming from checkpoint..')
assert os.path.isfile(PATH), 'Error: no checkpoint directory found!'
checkpoint = torch.load(PATH)
model.load_state_dict(checkpoint['state_dict'])
criterion = nn.CrossEntropyLoss(reduction = 'none').to(device)
test_acc0 = validate(test_iterator, model, criterion)
#########################################################################################################################
total = 0
for m in model.modules():
if isinstance(m, nn.Linear):
total += m.weight.data.numel() #返回元素个数
linear_weights = torch.zeros(total).cuda()
index = 0
# copy weights
for m in model.modules():
if isinstance(m, nn.Linear):
# return the no. of elements
size = m.weight.data.numel()
# copy the abs of elements to linear weights
linear_weights[index: (index+size)] = m.weight.data.view(-1).abs().clone()
index += size
# sort weights
y, i = torch.sort(linear_weights)
thre_index = int(total * hp.percent)
# got the thre value
thre = y[thre_index]
pruned = 0
print('Pruning threshold: {}'.format(thre))
zero_flag = False
for k, m in enumerate(model.modules()):
if isinstance(m, nn.Linear):
weight_copy = m.weight.data.abs().clone()
# sparsity mask
mask = weight_copy.gt(thre).float().cuda()
# the no. of elements pruned
pruned = pruned + mask.numel() - torch.sum(mask)
# prune the network
m.weight.data.mul_(mask)
if int(torch.sum(mask)) == 0:
zero_flag = True
print('layer index: {:d} \t total params: {:d} \t remaining params: {:d}'.
format(k, mask.numel(), int(torch.sum(mask))))
print('Total params: {}, Pruned params: {}, Pruned ratio: {}'.format(total, pruned, pruned/total))
########################################################################################################################
test_acc1 = validate(test_iterator, model, criterion)
torch.save({'epoch': 0,
'state_dict': model.state_dict(),
'acc': test_acc1,
'best_acc': 0.}, prune_path)
with open( stored_path + './prune-' + label + '.txt', 'w') as f:
f.write('Before pruning: Test Acc: %.2f\n' % (test_acc0))
f.write('Total params: {}, Pruned params: {}, Pruned ratio: {}\n'.format(total, pruned, pruned/total))
f.write('After Pruning: Test Acc: %.2f\n' % (test_acc1))
if zero_flag:
f.write("There exists a layer with 0 parameters left.")
return
def validate(test_loader, model, criterion):
batch_time = AverageMeter()
losses = AverageMeter()
accuracy = AverageMeter()
# switch to evaluate mode
model.eval()
with torch.no_grad():
end = time.time()
word = []
target_word = []
for i, batch in enumerate(test_loader):
src = batch.src
trg = batch.trg
#print(src.shape)
"""" compute the loss value """
loss = utils.test_step(model, src, trg, 0.1, pad_idx, criterion,
channel, CSI, imperfect, is_csinet, CSINet)
losses.update(loss)
""" use bleu score as accuracy tools"""
output = utils.greedy_decode(model, src, 0.1, hp.MAX_LENGTH,
pad_idx, start_symbol, channel,
CSI, imperfect, is_csinet, CSINet)
sentences = output.cpu().numpy().tolist()
result_string = list(map(StoT.sequence_to_text, sentences))
word = word + result_string
target_sent = trg.cpu().numpy().tolist()
result_string = list(map(StoT.sequence_to_text, target_sent))
target_word = target_word + result_string
bleu_score = bleu_score_1gram.compute_blue_score(word, target_word)
#print(bleu_score)
bleu_score = np.array(bleu_score)
score = np.mean(bleu_score)
accuracy.update(score)
""" measure elapsed time """
batch_time.update(time.time() - end)
end = time.time()
if i % 5 == 0:
print('Test: [{0}/{1}]\t'
'Time {batch_time.val:.3f} ({batch_time.avg:.3f})\t'
'Loss {loss.val:.4f} ({loss.avg:.4f})\t'
'Accuracy {acc.val:.3f} ({acc.avg:.3f})\t'
.format(i, len(test_loader), batch_time=batch_time,
loss = losses, acc = accuracy))
print('Prec@1 {acc.avg:.3f}'.format(acc=accuracy))
return accuracy.avg
class AverageMeter(object):
"""Computes and stores the average and current value"""
def __init__(self):
self.reset()
def reset(self):
self.val = 0
self.avg = 0
self.sum = 0
self.count = 0
def update(self, val, n=1):
self.val = val
self.sum += val * n
self.count += n
self.avg = self.sum / self.count
if __name__ == '__main__':
main()