-
Notifications
You must be signed in to change notification settings - Fork 0
/
meta_test.py
124 lines (97 loc) · 4.45 KB
/
meta_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
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
import os
from os.path import join as pjoin
import pickle
import argparse
import pickle
import random
import yaml
import numpy as np
import torch
from tqdm import tqdm
from collections import defaultdict
from botorch.optim import optimize_acqf
from botorch.acquisition import UpperConfidenceBound
from utils.log import get_logger
from utils.misc import load_module
from env.sumo_env import SumoEnv
if __name__ == "__main__":
# Argument Passing
parser = argparse.ArgumentParser()
parser.add_argument("--network", type=str, default="2by2")
parser.add_argument("--scheme", type=str, default="comb")
parser.add_argument("--model", type=str, default='anp')
parser.add_argument("--exp_id", type=str, default="trial1")
parser.add_argument("--scenario_id", type=int, default=0)
parser.add_argument("--root", type=str, default='.')
parser.add_argument("--seed", type=int, default=0)
args = parser.parse_args()
dtype = torch.double
device = torch.device("cuda") if torch.cuda.is_available() else torch.device("cpu")
network = args.network
scheme = args.scheme
model = args.model
exp_id = args.exp_id
scenario_id = args.scenario_id
seed = args.seed
root = pjoin(args.root, "results", network, scheme, model, exp_id)
if not os.path.isdir(root):
os.makedirs(root)
# Set Path for loading model, saving results, and logger
model_load_path = pjoin(root, "ckpt.tar")
results_save_path = pjoin(root, f"results_{args.scenario_id}.pkl")
log_path = pjoin(root, f"test_{args.scenario_id}.log")
logger = get_logger(log_path)
# Set Hyperparameters
settings = yaml.load(open(pjoin(args.root, "config", network, "test_settings.yaml"), "r"), Loader=yaml.SafeLoader)
# Set network
env = SumoEnv(network, scheme, scenario_id, args.root, run_type='test')
dim, bounds, equality_constraints = env.get_constraints(dtype, device)
# Load Data
data_path = pjoin(args.root, "data", network, f"traffic_data_{scheme}.pkl")
with open(data_path, "rb") as f:
data = pickle.load(f)
y_min = data.y_min
y_max = data.y_max
# Load Model
model_cls = getattr(load_module(pjoin(args.root, "models", f"{args.model}.py")), args.model.upper())
with open(pjoin(args.root, "results", network, scheme, model, exp_id, "model.yaml"), "r") as f:
config = yaml.safe_load(f)
model = model_cls(**config).to(dtype=dtype, device=device)
ckpt = torch.load(model_load_path)
model.load_state_dict(ckpt.model)
# Test
test_results_overall = defaultdict(list)
for num_test in range(settings["num_tests"]):
test_results = {}
# Set Seed for Reproduction
seed = settings["seed"] + num_test
random.seed(seed)
np.random.seed(seed)
torch.manual_seed(seed)
torch.cuda.manual_seed(seed)
x_init = env.get_init_points(settings["init_num_points"], dim, seed=seed)
x_init = x_init.to(dtype=dtype, device=device)
_, y_init = env.evaluate(x_init)
y_init = y_init.to(dtype=dtype, device=device)
model.eval()
for trial in tqdm(range(settings["test_num_trials"])):
model.x_init = x_init
model.y_init = (y_init - y_min) / (y_max - y_min)
acqf = UpperConfidenceBound(model, beta=1.0, maximize=False)
x_cand, _ = optimize_acqf(acqf, bounds=bounds,
q=1, num_restarts=10, raw_samples=512)
x_cand_transform, y_cand = env.evaluate(x_cand)
x_init = torch.cat([x_init, x_cand], dim=0)
y_init = torch.cat([y_init, y_cand], dim=0)
logger.info(f"[{trial+1}/{settings['test_num_trials']}]\nAction: {x_cand_transform}\nPerformance: {y_cand.item():4f}")
test_results["actions"] = x_init.cpu().detach().numpy()
test_results["performance"] = y_init.cpu().detach().numpy().flatten()
test_results_overall["actions"].append(test_results["actions"])
test_results_overall["performance"].append(test_results["performance"])
if args.network == "kt_simulator":
logger.info(f'Best Performance: {test_results_overall["performance"][-1].max().item():.4f}')
if args.network.startswith("sumo"):
logger.info(f'Best Performance: {test_results_overall["performance"][-1].min().item():.4f}')
# Save results
with open(results_save_path, "wb") as f:
pickle.dump(test_results_overall, f)