if not raw:
raw = [s[0][0] for s in test_dataset.samples]
if not lb and args.lb:
lb = [s[0][1] for s in test_dataset.samples]
model = OpinioNet.from_pretrained(model_config['path'],
version=model_config['version'],
focal=model_config['focal'])
print(weight_name)
model.load_state_dict(torch.load('../models/' + weight_name))
model.cuda()
ret = accum_result(ret, eval_epoch(model, test_loader, thresh))
del model
ret = average_result(ret, num_model)
ret = OpinioNet.nms_filter(ret, 0.28)
if args.lb:
def f1_score(P, G, S):
pr = S / P
rc = S / G
f1 = 2 * pr * rc / (pr + rc)
return f1, pr, rc
def evaluate_sample(gt, pred):
gt = set(gt)
pred = set(pred)
p = len(pred)
g = len(gt)
s = len(gt.intersection(pred))
tokenizer = BertTokenizer.from_pretrained(
'/home/zydq/.torch/models/bert/chinese-bert_chinese_wwm_pytorch',
do_lower_case=True)
test_dataset = ReviewDataset('../data/TEST/Test_reviews.csv', None,
tokenizer)
test_loader = DataLoader(test_dataset,
12,
collate_fn=test_dataset.batchify,
shuffle=False,
num_workers=5)
ret = None
for name in MODELS:
model_path = osp.join(SAVING_DIR, name)
model = OpinioNet.from_pretrained(
'/home/zydq/.torch/models/bert/chinese-bert_chinese_wwm_pytorch')
model.load_state_dict(torch.load(model_path))
model.cuda()
ret = accum_result(ret, eval_epoch(model, test_loader))
del model
ret = average_result(ret, len(MODELS))
ret = OpinioNet.nms_filter(ret, THRESH)
raw = [s[0][0] for s in test_dataset.samples]
result = gen_submit(ret, raw)
import time
result.to_csv('../submit/ensemble-' + str(round(time.time())) + '.csv',
header=False,
index=False)
print(len(result['id'].unique()), result.shape[0])
#
# P += p
# G += g
# S += s
# f1, pr, rc = f1_score(P, G, S)
# print("f1 %.5f, pr %.5f, rc %.5f, th %.5f" % (f1, pr, rc, 0.3))
threshs = list(np.arange(0.1, 0.9, 0.025))
best_f1, best_pr, best_rc = 0, 0, 0
best_thresh = 0.1
P, G, S = 0, 0, 0
BEST_PRED = PRED_COPY
for th in threshs:
P, G, S = 0, 0, 0
PRED_COPY = copy.deepcopy(PRED)
PRED_COPY = OpinioNet.nms_filter(PRED_COPY, th)
for b in range(len(PRED_COPY)):
gt = LB[b]
pred = [x[0] for x in PRED_COPY[b]]
p, g, s = evaluate_sample(gt, pred)
P += p
G += g
S += s
f1, pr, rc = f1_score(P, G, S)
if f1 > best_f1:
best_f1, best_pr, best_rc = f1, pr, rc
best_thresh = th
BEST_PRED = copy.deepcopy(PRED_COPY)
print("f1 %.5f, pr %.5f, rc %.5f, th %.5f" %