def _resblock(initial_size, fc, idim=idim, first_resblock=False):
if fc:
nonloc_scope.swap = not nonloc_scope.swap
return layers.CouplingBlock(
initial_size[0],
FCNet(
input_shape=initial_size,
idim=idim,
lipschitz_layer=_weight_layer(True),
nhidden=len(kernels.split('-')) - 1,
activation_fn=activation_fn,
preact=preact,
dropout=dropout,
coeff=None,
domains=None,
codomains=None,
n_iterations=None,
sn_atol=None,
sn_rtol=None,
learn_p=None,
div_in=2,
),
swap=nonloc_scope.swap,
)
else:
ks = list(map(int, kernels.split('-')))
if init_layer is None:
_block = layers.ChannelCouplingBlock
_mask_type = 'channel'
div_in = 2
mult_out = 1
else:
_block = layers.MaskedCouplingBlock
_mask_type = 'checkerboard'
div_in = 1
mult_out = 2
nonloc_scope.swap = not nonloc_scope.swap
_mask_type += '1' if nonloc_scope.swap else '0'
nnet = []
if not first_resblock and preact:
if batchnorm: nnet.append(layers.MovingBatchNorm2d(initial_size[0]))
nnet.append(ACT_FNS[activation_fn](False))
nnet.append(_weight_layer(fc)(initial_size[0] // div_in, idim, ks[0], 1, ks[0] // 2))
if batchnorm: nnet.append(layers.MovingBatchNorm2d(idim))
nnet.append(ACT_FNS[activation_fn](True))
for i, k in enumerate(ks[1:-1]):
nnet.append(_weight_layer(fc)(idim, idim, k, 1, k // 2))
if batchnorm: nnet.append(layers.MovingBatchNorm2d(idim))
nnet.append(ACT_FNS[activation_fn](True))
if dropout: nnet.append(nn.Dropout2d(dropout, inplace=True))
nnet.append(_weight_layer(fc)(idim, initial_size[0] * mult_out, ks[-1], 1, ks[-1] // 2))
if batchnorm: nnet.append(layers.MovingBatchNorm2d(initial_size[0]))
return _block(initial_size[0], nn.Sequential(*nnet), mask_type=_mask_type)
n_dist=args.n_dist,
n_power_series=args.n_power_series,
exact_trace=args.exact_trace,
brute_force=args.brute_force,
n_samples=args.n_samples,
neumann_grad=False,
grad_in_forward=False,
)
)
if args.actnorm: blocks.append(layers.ActNorm1d(2))
if args.batchnorm: blocks.append(layers.MovingBatchNorm1d(2))
model = layers.SequentialFlow(blocks).to(device)
elif args.arch == 'realnvp':
blocks = []
for _ in range(args.nblocks):
blocks.append(layers.CouplingBlock(2, swap=False))
blocks.append(layers.CouplingBlock(2, swap=True))
if args.actnorm: blocks.append(layers.ActNorm1d(2))
if args.batchnorm: blocks.append(layers.MovingBatchNorm1d(2))
model = layers.SequentialFlow(blocks).to(device)
logger.info(model)
logger.info("Number of trainable parameters: {}".format(count_parameters(model)))
optimizer = optim.Adam(model.parameters(), lr=args.lr, weight_decay=args.weight_decay)
time_meter = utils.RunningAverageMeter(0.93)
loss_meter = utils.RunningAverageMeter(0.93)
logpz_meter = utils.RunningAverageMeter(0.93)
delta_logp_meter = utils.RunningAverageMeter(0.93)
n_dist=args.n_dist,
n_power_series=args.n_power_series,
exact_trace=args.exact_trace,
brute_force=args.brute_force,
n_samples=args.n_samples,
neumann_grad=False,
grad_in_forward=False,
)
)
if args.actnorm: blocks.append(layers.ActNorm1d(2))
if args.batchnorm: blocks.append(layers.MovingBatchNorm1d(2))
model = layers.SequentialFlow(blocks).to(device)
elif args.arch == 'realnvp':
blocks = []
for _ in range(args.nblocks):
blocks.append(layers.CouplingBlock(2, build_nnet(dims, activation_fn), swap=False))
blocks.append(layers.CouplingBlock(2, build_nnet(dims, activation_fn), swap=True))
if args.actnorm: blocks.append(layers.ActNorm1d(2))
if args.batchnorm: blocks.append(layers.MovingBatchNorm1d(2))
model = layers.SequentialFlow(blocks).to(device)
logger.info(model)
logger.info("Number of trainable parameters: {}".format(count_parameters(model)))
optimizer = optim.Adam(model.parameters(), lr=args.lr, weight_decay=args.weight_decay)
time_meter = utils.RunningAverageMeter(0.93)
loss_meter = utils.RunningAverageMeter(0.93)
logpz_meter = utils.RunningAverageMeter(0.93)
delta_logp_meter = utils.RunningAverageMeter(0.93)
n_power_series=args.n_power_series,
exact_trace=args.exact_trace,
brute_force=args.brute_force,
n_samples=args.n_samples,
neumann_grad=False,
grad_in_forward=False,
))
if args.actnorm: blocks.append(layers.ActNorm1d(2))
if args.batchnorm: blocks.append(layers.MovingBatchNorm1d(2))
model = layers.SequentialFlow(blocks).to(device)
elif args.arch == 'realnvp':
blocks = []
for _ in range(args.nblocks):
blocks.append(
layers.CouplingBlock(2,
build_nnet_affine(dims, activation_fn),
swap=False))
blocks.append(
layers.CouplingBlock(2,
build_nnet_affine(dims, activation_fn),
swap=True))
if args.actnorm: blocks.append(layers.ActNorm1d(2))
if args.batchnorm: blocks.append(layers.MovingBatchNorm1d(2))
model = layers.SequentialFlow(blocks).to(device)
model.load_state_dict(checkpoint['state_dict'])
model.to(device)
# checkerboard data
def checkerboard(batch_size):
x1 = np.random.rand(batch_size) * 4 - 2
