-
Notifications
You must be signed in to change notification settings - Fork 3
/
Copy pathevaluate.py
162 lines (145 loc) · 6.65 KB
/
evaluate.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
import glob
import os
import shutil
import faiss
import numpy as np
import torch
from torch.utils.data import SequentialSampler, DataLoader
from torch.utils.tensorboard import SummaryWriter
from transformers import T5Tokenizer
from src.model import TASTEModel
from utils.data_loader import load_item_name, load_item_address, load_data, load_item_data, SequenceDataset, ItemDataset
from utils.option import Options
from utils.rec_metrics import get_metrics_dict
from utils.util import set_randomseed, init_logger, early_stopping
def evaluate(model, eval_seq_dataloader, eval_item_dataloader, device, Ks, logging,tb_logger,step):
model.eval()
model = model.module if hasattr(model, "module") else model
item_emb_list = []
seq_emb_list = []
target_item_list = []
with torch.no_grad():
for i, batch in enumerate(eval_item_dataloader):
item_inputs = batch["item_ids"].to(device)
item_masks = batch["item_masks"].to(device)
_,item_emb = model(item_inputs, item_masks)
item_emb_list.append(item_emb.cpu().numpy())
item_emb_list = np.concatenate(item_emb_list, 0)
for i, batch in enumerate(eval_seq_dataloader):
seq_inputs = batch["seq_ids"].to(device)
seq_masks = batch["seq_masks"].to(device)
batch_target = batch["target_list"]
_,seq_emb = model(seq_inputs, seq_masks)
seq_emb_list.append(seq_emb.cpu().numpy())
target_item_list.extend(batch_target)
seq_emb_list = np.concatenate(seq_emb_list, 0)
faiss.omp_set_num_threads(16)
cpu_index = faiss.IndexFlatIP(768)
cpu_index.add(np.array(item_emb_list, dtype=np.float32))
query_embeds = np.array(seq_emb_list, dtype=np.float32)
D, I = cpu_index.search(query_embeds, max(Ks))
n_item = item_emb_list.shape[0]
n_seq = seq_emb_list.shape[0]
metrics_dict = get_metrics_dict(I, n_seq, n_item, Ks, target_item_list)
logging.info(
'current:step:{} Recall@10：{:.4f}, Recall@20:{:.4f}, NDCG@10: {:.4f}, NDCG@20:{:.4f}'.format(
step,metrics_dict[10]['recall'], metrics_dict[20]['recall'],
metrics_dict[10]['ndcg'],
metrics_dict[20]['ndcg']))
if tb_logger is not None:
tb_logger.add_scalar("recall@10", metrics_dict[10]['recall'], step)
tb_logger.add_scalar("recall@20", metrics_dict[20]['recall'], step)
tb_logger.add_scalar("ndcg@10", metrics_dict[10]['ndcg'], step)
tb_logger.add_scalar("ndcg@20", metrics_dict[20]['ndcg'], step)
return metrics_dict
def main():
options = Options()
opt = options.parse()
set_randomseed(opt.seed)
checkpoint_path = os.path.join(opt.checkpoint_dir, opt.data_name)
checkpoint_path = os.path.join(checkpoint_path, opt.experiment_name)
checkpoint_path = os.path.join(checkpoint_path, 'eval')
runlog_path = os.path.join(checkpoint_path, 'log')
os.makedirs(runlog_path, exist_ok=True)
logging = init_logger(
os.path.join(runlog_path, 'runlog.log')
)
tb_path = os.path.join(checkpoint_path, 'tensorboard')
tb_logger = SummaryWriter(tb_path)
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
logging.info("***** Running evaluation *****")
# We use recall@20 as the metric for saving the best model
best_dev_recall = 0.0
# early stop
stop_step = 0
stop_flag, update_flag = False, False
all_checkpoint_path = opt.all_models_path
best_checkpoint = os.path.join(all_checkpoint_path, 'best_dev')
# add '/' character
len_path = len(all_checkpoint_path) + 1
id_file = os.path.join(all_checkpoint_path, 'checkpoint-*')
len_file = len(glob.glob(id_file))
logging.info(f"all checkpoint amounts : {len_file}")
all_checkpoint = []
for file in sorted(glob.glob(id_file), key=lambda name: int(name[len_path + 11:])):
all_checkpoint.append(file)
tokenizer = T5Tokenizer.from_pretrained(all_checkpoint[0])
data_dir = os.path.join(opt.data_dir, opt.data_name)
eval_file = os.path.join(data_dir, 'valid.txt')
item_file = os.path.join(data_dir, 'item.txt')
# We only use the title attribute in the Amazon dataset, and the title and address attributes in the Yelp dataset.
if opt.data_name == 'beauty' or opt.data_name == 'sports' or opt.data_name == 'toys':
item_desc = load_item_name(item_file)
elif opt.data_name == 'yelp':
item_desc = load_item_address(item_file)
item_len = len(item_desc)
logging.info(f"item len: {item_len}")
eval_data = load_data(eval_file,item_desc)
logging.info(f"dev len: {len(eval_data)}")
item_data = load_item_data(item_desc)
eval_seq_dataset = SequenceDataset(eval_data, tokenizer, opt)
eval_seq_sampler = SequentialSampler(eval_seq_dataset)
eval_seq_dataloader = DataLoader(
eval_seq_dataset,
sampler=eval_seq_sampler,
batch_size=opt.eval_batch_size,
drop_last=False,
num_workers=0,
collate_fn=eval_seq_dataset.collect_fn
)
eval_item_dataset = ItemDataset(item_data, tokenizer, opt)
eval_item_sampler = SequentialSampler(eval_item_dataset)
eval_item_dataloader = DataLoader(
eval_item_dataset,
sampler=eval_item_sampler,
batch_size=opt.eval_batch_size,
drop_last=False,
num_workers=0,
collate_fn=eval_item_dataset.collect_fn
)
Ks = eval(opt.Ks)
for i in range(len(all_checkpoint)):
check_path = all_checkpoint[i]
step = int(check_path[len_path + 11:])
model_class = TASTEModel
model = model_class.from_pretrained(check_path)
model.to(device)
metrics_dict = evaluate(model, eval_seq_dataloader, eval_item_dataloader, device, Ks, logging,
tb_logger, step)
cur_recall = metrics_dict[20]['recall']
best_dev_recall, stop_step, stop_flag, update_flag = early_stopping(cur_recall, best_dev_recall, stop_step,
opt.stopping_step)
if update_flag:
# Python version below 3.8 will report an error
shutil.copytree(check_path, best_checkpoint, dirs_exist_ok=True)
logging.info(
'Saved Best:step:{},Recall@10：{:.4f}, Recall@20:{:.4f}, NDCG@10: {:.4f}, NDCG@20:{:.4f}'.format(
step,metrics_dict[10]['recall'], metrics_dict[20]['recall'],
metrics_dict[10]['ndcg'],
metrics_dict[20]['ndcg']))
if stop_flag:
logging.info(" Early stop!Finished!")
break
logging.info('***** Finish evaluation *****')
if __name__ == '__main__':
main()