-
Notifications
You must be signed in to change notification settings - Fork 5
/
score.py
205 lines (183 loc) · 9.48 KB
/
score.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
from dataset import WikiSqlDataset
from model import LoggingCallback,SeqGenSQL
from dbengine_seqgen import DBEngine as db
import argparse
import os
import logging
from torch.utils.data import Dataset, DataLoader
from tqdm.auto import tqdm
import urllib.request
# .\score.py --ckpt_download_url https://onebigdatabag.blob.core.windows.net/shared/base_epoch%3D12-val_loss%3D0.02616.ckpt --ckpt_path SeqGenSQL.ckpt
if __name__ == '__main__':
logging.basicConfig(level=logging.ERROR)
parser = argparse.ArgumentParser()
parser.add_argument('--data_dir', default="data")
parser.add_argument('--data_type', default="dev", help="train|dev|test")
parser.add_argument("--output_dir", default=".")
parser.add_argument("--base_model", default="t5-base")
parser.add_argument("--batch_size", type=int, default=32)
parser.add_argument("--ckpt_download_url", default=None)
parser.add_argument("--ckpt_path", default="SeqGenSQL.ckpt")
parser.add_argument("--include_data_type", default=True)
parser.add_argument("--num_sample_rows", type=int, default=3)
parser.add_argument("--data_aug", default=[], help="List, use one of these options: ['select_column', 'where_value']. Default is []")
parser.add_argument("--use_modified_network", default=False, help="Use gated layer to decide whether to extract or to generate")
parser.add_argument("--max_seq_length", type=int, default=512)
parser.add_argument("--max_output_length", type=int, default=200)
parser.add_argument("--num_return_sequences", type=int, default=1)
parser.add_argument("--device", default="cuda", help="cpu|cuda")
parser.add_argument("--silent", default=False, help="Output paramters")
args = parser.parse_args()
if args.num_return_sequences > 1:
args.batch_size = 1
model_name = os.path.basename(args.ckpt_path)
log_file = '{}/{}.score.log'.format(args.output_dir, model_name.replace("/","_"))
if not args.silent:
print("======================================================")
print('Error log file name:', log_file)
print("======================================================")
if args.ckpt_download_url != None and not os.path.exists(args.ckpt_path):
print("Downloading checkpoint...", end="")
urllib.request.urlretrieve(args.ckpt_download_url, args.ckpt_path)
print("Done!")
print("Loading database file...", end="")
dbeng = db('{}/{}.db'.format(args.data_dir, args.data_type))
print("Done!")
print("Loading T5FinalTuner pretrained model...", end="")
model = SeqGenSQL.load_from_checkpoint(args.ckpt_path)
print("Done!")
if args.device == 'cuda':
print("Loading model to cuda...", end="")
model = model.to('cuda')
print("Done!")
print("Loading dataset...")
dataset = WikiSqlDataset(model.tokenizer,
args.data_dir,
args.data_type,
include_sample_data =args.num_sample_rows,
max_input_len=args.max_seq_length,
max_output_len=args.max_output_length,include_question = True)
if args.num_return_sequences > 1:
args.batch_size = 1
# generate sql statement
print("Generating sequences...", end="")
loader = DataLoader(dataset, batch_size=args.batch_size, shuffle=False)
outputs = []
targets = []
di = 0
for batch in tqdm(loader):
if args.num_return_sequences == 1:
input_ids = batch['source_ids']
attention_mask = batch['source_mask']
else:
# input_ids = torch.unsqueeze(batch['source_ids'],0)
# attention_mask = torch.unsqueeze(batch['source_mask'],0)
input_ids = batch['source_ids']
attention_mask = batch['source_mask']
if args.device == 'cuda':
outs = model.model.generate(input_ids=input_ids.cuda(),
attention_mask=attention_mask.cuda(),
num_beams = args.num_return_sequences,
max_length=args.max_output_length,
num_return_sequences = args.num_return_sequences)
else:
outs = model.model.generate(input_ids=input_ids,
attention_mask=attention_mask,
max_length=args.max_output_length,
num_beams = args.num_return_sequences,
num_return_sequences = args.num_return_sequences)
if args.num_return_sequences > 1:
guided_out = model.tokenizer.decode(outs[0])
target = model.tokenizer.decode(batch["target_ids"][0])
execution_failed = False
for i, beam_output in enumerate(outs):
try:
dec = model.tokenizer.decode(beam_output)
if execution_failed:
print(" ",dec)
pred_lf = dbeng.generate_logical_form(model.tokenizer, dec,batch['question'][0],
dataset.tables,
dataset.agg_ops, dataset.cond_ops,"sql")
pred_result = dbeng.execute_query(table_id = pred_lf["table_id"], query = pred_lf)
guided_out = dec
break
except:
print("====================================================")
print("question {}: {}".format(di, model.tokenizer.decode(batch['source_ids'][0]).replace('⁇','<')))
#print(d)
print(" True:", target.replace('⁇','<'))
print(" Pred:", dec.replace('⁇','<'))
print(" lf:",pred_lf)
execution_failed = True
di += 1
outputs.append(guided_out)
targets.append(target)
else:
guided_out = [model.tokenizer.decode(ids) for ids in outs]
target = [model.tokenizer.decode(ids) for ids in batch["target_ids"]]
outputs.extend(guided_out)
targets.extend(target)
print("Done!")
# Score
print("Scoring...", end="")
f = open(log_file ,"w")
correct = 0
incorrect = {"sel_neg":0, "sel_mismatch":0,"agg_neg":0, "agg_mismatch":0, "cond_col_neg":0,"cond_col_mismatch":0,
"cond_op_neg":0, "cond_op_mismatch":0,"cond_val_neg":0,"cond_val_mismatch":0}
for t,d, i in zip(targets, outputs, dataset.data):
try:
if t == d:
correct += 1
else:
pred_lf = dbeng.generate_logical_form(model.tokenizer, d,i['question'],
dataset.tables,
dataset.agg_ops, dataset.cond_ops,"sql")
if pred_lf['sql']!={}:
pred_result = dbeng.execute_query(table_id = pred_lf["table_id"], query = pred_lf)
else:
f.write("===================== ERROR ========================\n")
f.write("Question: {}\n".format(i['question']))
f.write("Pred: {} lf: {} RESULT: {}\n".format(d,pred_lf['sql'],pred_result))
f.write("True: {} lf: {} RESULT: {}\n\n".format(t, i['sql'], true_result))
continue
true_result = dbeng.execute_query(table_id = i["table_id"], query = i)
if (pred_result == true_result):
correct += 1
else:
f.write("===================== ERROR ========================\n")
f.write("Question: {}\n".format(i['question']))
f.write("Pred: {} lf: {} RESULT: {}\n".format(d,pred_lf['sql'],pred_result))
f.write("True: {} lf: {} RESULT: {}\n\n".format(t, i['sql'], true_result))
#print(pred_lf['sql']['sel'])
if (pred_lf['sql']['sel'] != i['sql']['sel']):
incorrect['sel_mismatch'] += 1
if (pred_lf['sql']['agg'] != i['sql']['agg']):
incorrect['agg_mismatch'] += 1
for c in pred_lf['sql']['conds']:
if c[0] == -1:
incorrect['cond_col_neg'] += 1
if c[1] == -1:
incorrect['cond_cop_neg'] += 1
if i['question'].find(str(c[2])) == -1:
incorrect['cond_val_neg'] += 1
except:
f.write("===================== ERROR ========================\n")
f.write("Question: {}\n".format(i['question']))
f.write("Pred: {} lf: {}\n".format(d,pred_lf['sql']))
f.write("True: {} lf: {}\n\n".format(t, i['sql']))
#print(pred_lf['sql']['sel'])
if (pred_lf['sql']['sel'] == -1):
incorrect['sel_neg'] += 1
if (pred_lf['sql']['agg'] == -1):
incorrect['agg_neg'] += 1
for c in pred_lf['sql']['conds']:
if c[0] == -1:
incorrect['cond_col_neg'] += 1
if c[1] == -1:
incorrect['cond_op_neg'] += 1
if i['question'].find(str(c[2])) == -1:
incorrect['cond_val_neg'] += 1
f.close()
print(incorrect)
print("Correct: {} Total: {} Ratio:{:.5f} ".format(correct, len(targets), correct/len(targets)))
print("All Completed!")