-
Notifications
You must be signed in to change notification settings - Fork 0
/
run_lds.py
90 lines (73 loc) · 2.29 KB
/
run_lds.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
import os
import wandb
from data import make_dot_data
from log import make_folder
from svae.lds import SVAE
from vae import VAE, resVAE
from seed import SEED
# parameters for the experimental conditions
experimental_parameters = {
"weight_init_std": 1e-2,
"local_kld_weight": 1.0,
"global_kld_weight": 1.0,
"name": "vae",
"latent_dim": 10,
"update_init_params": False,
}
# parameters for generating synthetic data
data_parameters = {
"image_width": 12,
"T": 500,
"num_steps": 5000,
"render_sigma": 0.20,
"v": 0.75,
}
# parameters for the encoder and decoder
vae_parameters = {
"latent_dim": experimental_parameters["latent_dim"],
"input_size": data_parameters["image_width"],
"hidden_size": [50],
"recon_loss": "likelihood",
"name": experimental_parameters["name"],
"weight_init_std": experimental_parameters["weight_init_std"],
}
# parameters for the SVAE model
svae_parameters = {
"batch_size": 80,
"epochs": 1000,
"local_kld_weight": experimental_parameters["local_kld_weight"],
"global_kld_weight": experimental_parameters["global_kld_weight"],
"latent_dim": vae_parameters["latent_dim"],
"update_init_params": experimental_parameters["update_init_params"],
}
# combined parameters
hyperparameters = {
"data_parameters": data_parameters,
"VAE_parameters": vae_parameters,
"SVAE_parameters": svae_parameters,
"seed": SEED,
}
def get_data(data_parameters):
# generate synthetic data
data = make_dot_data(**data_parameters)
return data
def get_network():
name = hyperparameters["VAE_parameters"]["name"]
if name == "vae":
return VAE(**hyperparameters["VAE_parameters"])
elif name == "resvae":
return resVAE(**hyperparameters["VAE_parameters"])
else:
raise ValueError(f"Network name {name} not recognized!")
def main():
# logging
wandb.init(project="SVAE_lds", mode="disabled", config=hyperparameters)
folder_name = make_folder(wandb.run.name)
# get data and vae model
observations = get_data(hyperparameters["data_parameters"])
network = get_network()
# SVAE model
model = SVAE(network, save_path=os.path.join(folder_name, "svae"))
model.fit(observations, **hyperparameters["SVAE_parameters"])
if __name__ == "__main__":
main()