-
Notifications
You must be signed in to change notification settings - Fork 163
/
unit_test.py
103 lines (96 loc) · 4.14 KB
/
unit_test.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
import torch
from libcity.config import ConfigParser
from libcity.data import get_dataset
from libcity.utils import get_model
#############################################
# The parameter to control the unit testing #
tested_trajectory_model = 'RNN'
tested_trajectory_dataset = 'foursquare_nyc'
tested_trajectory_encoder = 'StandardTrajectoryEncoder'
tested_traffic_state_model = 'RNN'
tested_traffic_state_dataset = 'METR_LA'
#############################################
def test_new_tlp_model():
# load Config Module
config = ConfigParser(task='traj_loc_pred', model=tested_trajectory_model, dataset=tested_trajectory_dataset,
config_file=None, other_args={'batch_size': 2})
# load Data Module
dataset = get_dataset(config)
train_data, valid_data, test_data = dataset.get_data()
data_feature = dataset.get_data_feature()
# get a batch to test model API
batch = train_data.__iter__().__next__()
batch.to_tensor(config['device'])
# init model
model = get_model(config, data_feature)
model = model.to(config['device'])
# test model.predict
res = model.predict(batch)
# check res format
assert torch.is_tensor(res)
assert res.shape[0] == batch['target'].shape[0]
assert res.shape[1] == data_feature['loc_size']
# test model.calculate_loss
loss = model.calculate_loss(batch)
assert loss.requires_grad
def test_new_tsp_model():
# load Config Module
config = ConfigParser(task='traffic_state_pred', model=tested_traffic_state_model,
dataset=tested_traffic_state_dataset,
config_file=None, other_args={'batch_size': 2})
# load Data Module
dataset = get_dataset(config)
train_data, valid_data, test_data = dataset.get_data()
data_feature = dataset.get_data_feature()
# get a batch to test model API
batch = train_data.__iter__().__next__()
batch.to_tensor(config['device'])
# init model
model = get_model(config, data_feature)
model = model.to(config['device'])
# test model.predict
res = model.predict(batch)
assert torch.is_tensor(res)
assert res.shape[0] == config['batch_size']
assert res.dim() == batch['y'].dim()
assert res.shape[-1] == data_feature['output_dim']
if config['dataset_class'] == 'TrafficStatePointDataset':
assert res.shape[1] == config['output_window']
assert res.shape[2] == data_feature['num_nodes']
elif config['dataset_class'] == 'TrafficStateGridDataset':
assert res.shape[1] == config['output_window']
if config['use_row_column'] is False:
assert res.shape[2] == data_feature['num_nodes']
else:
assert res.shape[2] == data_feature['len_row']
assert res.shape[3] == data_feature['len_column']
elif config['dataset_class'] == 'TrafficStateGridOdDataset':
assert res.shape[1] == config['output_window']
if config['use_row_column'] is False:
assert res.shape[2] == data_feature['num_nodes']
assert res.shape[3] == data_feature['num_nodes']
else:
assert res.shape[2] == data_feature['len_row']
assert res.shape[3] == data_feature['len_column']
assert res.shape[4] == data_feature['len_row']
assert res.shape[5] == data_feature['len_column']
# test model.calculate_loss
loss = model.calculate_loss(batch)
assert torch.is_tensor(loss)
assert loss.requires_grad
def test_new_traj_encoder():
# load Config Module
config = ConfigParser(task='traj_loc_pred', model=tested_trajectory_model, dataset=tested_trajectory_dataset,
config_file=None, other_args={'batch_size': 2, 'traj_encoder': tested_trajectory_encoder})
# load dataset
dataset = get_dataset(config)
# cut trajectory
cut_data = dataset.cutter_filter()
# test encoder.encode
encoder = dataset.encoder
uid = list(cut_data.keys())[0]
encoded_traj = encoder.encode(uid, cut_data[uid])
assert isinstance(encoded_traj, list)
an_model_input = encoded_traj[0]
assert isinstance(an_model_input, list)
assert len(an_model_input) == len(encoder.feature_dict)