-
Notifications
You must be signed in to change notification settings - Fork 2
/
Copy pathtrain.py
213 lines (167 loc) · 8.09 KB
/
train.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
from time import time
from functools import reduce
import uuid
from tqdm import tqdm
import os
from clearml import Task, Logger
from omegaconf import OmegaConf
import hydra
import numpy as np
import torch
from torch.utils.data import DataLoader
from model import SASRec
from data import get_dataset, data_to_sequences, SequentialDataset
from utils import topn_recommendations, downvote_seen_items
from eval_utils import model_evaluate, sasrec_model_scoring, get_test_scores
@hydra.main(version_base=None, config_path="configs", config_name="train")
def main(config):
print(OmegaConf.to_yaml(config))
if hasattr(config, 'cuda_visible_devices'):
os.environ['CUDA_VISIBLE_DEVICES'] = str(config.cuda_visible_devices)
if hasattr(config, 'project_name'):
Task.set_random_seed(config.trainer_params.seed)
task = Task.init(project_name=config.project_name, task_name=config.task_name,
reuse_last_task_id=False)
task.connect(OmegaConf.to_container(config))
else:
task = None
base_config = dict(
num_epochs = config.trainer_params.num_epochs,
maxlen = config.model_params.maxlen,
hidden_units = config.model_params.hidden_units,
dropout_rate = config.model_params.dropout_rate,
num_blocks = config.model_params.num_blocks,
num_heads = config.model_params.num_heads,
batch_size = config.dataloader.batch_size,
learning_rate = config.trainer_params.learning_rate,
fwd_type = config.model_params.fwd_type,
l2_emb = 0,
n_neg_samples = config.dataloader.n_neg_samples,
manual_seed = config.trainer_params.seed,
sampler_seed = config.trainer_params.seed,
sampling = config.model_params.sampling,
patience = config.trainer_params.patience,
skip_epochs = config.trainer_params.skip_epochs
)
if config.model_params.fwd_type == 'gbce':
base_config['gbce_t'] = config.model_params.gbce_t
if config.model_params.fwd_type == 'sce':
base_config['n_buckets'] = config.model_params.n_buckets
base_config['bucket_size_x'] = config.model_params.bucket_size_x
base_config['bucket_size_y'] = config.model_params.bucket_size_y
base_config['mix_x'] = config.model_params.mix_x
device = 'cuda'
training, data_description, _, testset_valid, testset_, holdout_valid, holdout_ = get_dataset(path=config.data_path, splitting=config.splitting)
if task:
log = Logger.current_logger()
else:
log = None
model = \
build_sasrec_model(base_config, training, data_description,
testset_valid=testset_valid, holdout_valid=holdout_valid, device=device,
task=task, log=log)
test_scores = get_test_scores(model, data_description, testset_, holdout_, device)
test_scores_meta = reduce(lambda s, metric_name: s + f'\n{metric_name}:{test_scores[metric_name]:.3g}', test_scores.keys(), '')
print(test_scores_meta)
if task:
for metric_name, metric_value in test_scores.items():
log.report_single_value(name=f'test_{metric_name}', value=round(metric_value, 4))
task.close()
def set_worker_random_state(id):
dataset = torch.utils.data.get_worker_info().dataset
dataset.seed = dataset.seed + id
dataset.random_state = np.random.RandomState(dataset.seed)
def prepare_sasrec_model(config, data, data_description, device):
n_users = data_description['n_users']
n_items = data_description['n_items']
model = SASRec(n_items, config).to(device)
train_sequences = data_to_sequences(data, data_description)
sampler = \
DataLoader(SequentialDataset(train_sequences, n_users, n_items,
maxlen = config['maxlen'],
seed = config['sampler_seed'],
n_neg_samples = config['n_neg_samples'],
pad_token = model.pad_token,
sampling = config['sampling']), batch_size=config['batch_size'], shuffle=True, num_workers=8, pin_memory=True, prefetch_factor=10, worker_init_fn=set_worker_random_state, persistent_workers=True, drop_last=True)
n_batches = len(train_sequences) // config['batch_size']
optimizer = \
torch.optim.Adam(model.parameters(),
lr = config['learning_rate'],
betas = (0.9, 0.98))
return model, sampler, n_batches, optimizer
def train_sasrec_epoch(model, num_batch, l2_emb, sampler, optimizer, device):
model.train()
pad_token = model.pad_token
losses = []
for _, *seq_data in sampler:
# convert batch data into torch tensors
seq, pos, neg = (torch.tensor(np.array(x), device=device, dtype=torch.long) for x in seq_data)
loss = model(seq, pos, neg)
optimizer.zero_grad()
if l2_emb != 0:
for param in model.item_emb.parameters():
loss += l2_emb * torch.norm(param)**2
loss.backward()
optimizer.step()
losses.append(loss.item())
return losses
def build_sasrec_model(config, data, data_description, testset_valid, holdout_valid, device, task, log):
'''Simple MF training routine without early stopping'''
model, sampler, n_batches, optimizers = prepare_sasrec_model(config, data, data_description, device)
losses = {}
metrics = {}
ndcg = {}
best_ndcg = 0
wait = 0
start_time = time()
torch.cuda.synchronize()
torch.cuda.reset_peak_memory_stats()
start_memory = torch.cuda.memory_allocated()
checkpt_name = uuid.uuid4().hex
if not os.path.exists('./checkpt'):
os.mkdir('./checkpt')
checkpt_path = os.path.join('./checkpt', f'{checkpt_name}.chkpt')
for epoch in tqdm(range(config['num_epochs'])):
losses[epoch] = train_sasrec_epoch(
model, n_batches, config['l2_emb'], sampler, optimizers, device
)
if epoch % config['skip_epochs'] == 0:
val_scores = sasrec_model_scoring(model, testset_valid, data_description, device)
downvote_seen_items(val_scores, testset_valid, data_description)
val_recs = topn_recommendations(val_scores, topn=10)
val_metrics = model_evaluate(val_recs, holdout_valid, data_description)
metrics[epoch] = val_metrics
ndcg_ = val_metrics['ndcg@10']
ndcg[epoch] = ndcg_
if task and (epoch % 5 == 0):
log.report_scalar("Loss", series='Val', iteration=epoch, value=np.mean(losses[epoch]))
log.report_scalar("NDCG", series='Val', iteration=epoch, value=ndcg_)
if ndcg_ > best_ndcg:
best_ndcg = ndcg_
torch.save(model.state_dict(), checkpt_path)
wait = 0
elif wait < config['patience'] // config['skip_epochs'] + 1:
wait += 1
else:
break
torch.cuda.synchronize()
training_time_sec = time() - start_time
full_peak_training_memory_bytes = torch.cuda.max_memory_allocated()
peak_training_memory_bytes = torch.cuda.max_memory_allocated() - start_memory
training_epoches = len(losses)
model.load_state_dict(torch.load(checkpt_path))
os.remove(checkpt_path)
print('Peak training memory, mb:', round(full_peak_training_memory_bytes/ 1024. / 1024., 2))
print('Training epoches:', training_epoches)
print('Training time, m:', round(training_time_sec/ 60., 2))
if task:
ind_max = np.argmax(list(ndcg.values())) * config['skip_epochs']
for metric_name, metric_value in metrics[ind_max].items():
log.report_single_value(name=f'val_{metric_name}', value=round(metric_value, 4))
log.report_single_value(name='train_peak_mem_mb', value=round(peak_training_memory_bytes/ 1024. / 1024., 2))
log.report_single_value(name='full_train_peak_mem_mb', value=round(full_peak_training_memory_bytes/ 1024. / 1024., 2))
log.report_single_value(name='train_epoches', value=training_epoches)
log.report_single_value(name='train_time_m', value=round(training_time_sec/ 60., 2))
return model
if __name__ == "__main__":
main()