###############
# Get simulator
###############
simulator = simulated_datasets_lib.StarSimulator(psf_og,
slen,
background,
transpose_psf = False)
###############
# define VAE
###############
star_encoder = starnet_lib.StarEncoder(slen = slen,
ptile_slen = ptile_slen,
step = step,
edge_padding = edge_padding,
n_bands = psf_og.shape[0],
max_detections = 2,
fmin = fmin,
constrain_logflux_mean = True,
track_running_stats = False)
star_encoder.eval();
star_encoder.to(device);
###############
# define optimizer
###############
learning_rate = 1e-3
weight_decay = 1e-3
optimizer = optim.Adam([
{'params': star_encoder.parameters(),
print('data generation time: {:.3f}secs'.format(time.time() - t0))
# get loader
batchsize = 1
loader = torch.utils.data.DataLoader(dataset=star_dataset,
batch_size=batchsize,
shuffle=True)
###############
# define VAE
###############
star_encoder = starnet_lib.StarEncoder(slen=data_params['slen'],
ptile_slen=50,
step=50,
edge_padding=0,
n_bands=psf_og.shape[0],
max_detections=3,
track_running_stats=False)
star_encoder.to(device)
###############
# define optimizer
###############
learning_rate = 1e-3
weight_decay = 1e-5
optimizer = optim.Adam([{
'params': star_encoder.parameters(),
'lr': learning_rate
}],
add_noise = True)
print('data generation time: {:.3f}secs'.format(time.time() - t0))
# get loader
batchsize = 64
loader = torch.utils.data.DataLoader(dataset=star_dataset,
batch_size=batchsize,
shuffle=True)
###############
# define VAE
###############
star_encoder = starnet_lib.StarEncoder(slen=data_params['slen'],
ptile_slen=ptile_slen,
step=step,
edge_padding=edge_padding,
n_bands=psf_og.shape[0],
max_detections=2)
star_encoder.to(device)
###############
# define optimizer
###############
learning_rate = 1e-3
weight_decay = 1e-5
optimizer = optim.Adam([{
'params': star_encoder.parameters(),
'lr': learning_rate
}],
weight_decay=weight_decay)
print('data generation time: {:.3f}secs'.format(time.time() - t0))
# get data loader
batchsize = 2000
loader = torch.utils.data.DataLoader(dataset=star_dataset,
batch_size=batchsize,
shuffle=True)
###############
# define VAE
###############
star_encoder = starnet_lib.StarEncoder(slen=data_params['slen'],
ptile_slen=data_params['slen'],
step=data_params['slen'],
edge_padding=0,
n_bands=len(bands),
max_detections=2)
star_encoder.to(device)
###############
# define optimizer
###############
learning_rate = 1e-3
weight_decay = 1e-3
optimizer = optim.Adam([{
'params': star_encoder.parameters(),
'lr': learning_rate
}],
weight_decay=weight_decay)