def top_x(self):
hps = HyperParams()
dtype = torch.float
device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
N = hps.batch_size
d_in = hps.bottom_output + 1
x = torch.randn(N, d_in, device=device, dtype=dtype)
return x
def flat_x():
hps = HyperParams()
dtype = torch.float
device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
N = hps.batch_size
d_in = hps.n_input
flat_x = torch.randn(N, d_in, device=device, dtype=dtype)
return flat_x
def conv_x(self):
hps = HyperParams()
dtype = torch.float
device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
N = hps.batch_size
d_in = hps.n_input
x = torch.randn(N, 4, d_in // 4, device=device, dtype=dtype)
return x
def test_model_mlp_stack(self, sample_x):
path_to_json = os.path.join(get_project_root(), 'res_experiments',
'hps_base_mlp.json')
hps = HyperParams.from_file(path_to_json=path_to_json)
bottom = getattr(models, hps.bottom_net)
top = getattr(models, hps.top_net)
net = FullModel(hps, bottom, top)
out = net(sample_x)
assert True
def sample_x(self):
hps = HyperParams()
dtype = torch.float
device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
N = hps.batch_size
d_in = hps.n_input
x0 = torch.randn(N, d_in, device=device, dtype=dtype)
x1 = torch.randn(N, 1, device=device, dtype=dtype)
return x0, x1
def flat(self):
path_to_json = os.path.join(get_project_root(), 'res_experiments',
'hps_partly_independent_mlp.json')
hps = HyperParams.from_file(path_to_json=path_to_json)
N = hps.batch_size
d_in = hps.n_input
device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
flat = torch.randn(N, d_in, device=device, dtype=torch.float)
self.hps = hps
return flat
def test_model_conv_stack(self, sample_conv_x):
path_to_json = os.path.join(get_project_root(), 'res_experiments',
'hps_base_conv.json')
hps = HyperParams.from_file(path_to_json=path_to_json)
bottom = getattr(models, hps.bottom_net)
top = getattr(models, hps.top_net)
model = Model(hps)
x = model.make_loader()
x_ = next(iter(x))
net = FullModel(hps, bottom, top)
out = net(x_['X'])
assert True
def load_pretrained_bottom(self, path_to_model, path_to_json):
hps = HyperParams.from_file(path_to_json=path_to_json)
model_common = Model(hps)
model_common.load_model(path_to_model)
pretrained_dict = model_common.net.state_dict()
pretrained_dict = {
k: v
for k, v in pretrained_dict.items() if 'bottom' in k
}
model_dict = self.net.state_dict()
model_dict.update(pretrained_dict)
self.net.load_state_dict(model_dict)
def test_full_mlp_stack(self):
path_to_json = os.path.join(get_project_root(), 'res_experiments',
'hps_partly_independent_mlp.json')
filename = get_project_root() / 'data' / 'small_parameters_base.fits'
params = fits.open(filename)[0].data
hps = HyperParams.from_file(path_to_json=path_to_json)
bottom = getattr(models, hps.bottom_net)
top = getattr(models, hps.top_net)
model = Model(hps)
x = model.make_loader(data_arr=params)
x_ = next(iter(x))
net = FullModel(hps, bottom, top)
out = net(x_['X'])
def test_top_net_forward(self):
hps = HyperParams()
dtype = torch.float
device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
N = hps.batch_size
D_in = hps.top_input
D_out = hps.top_output
x = torch.randn([N, 224], device=device, dtype=dtype)
net = TopNet(hps)
convnet = Conv1dModel(hps)
# y = torch.rand(N, D_out, device=device, dtype=dtype)
out = convnet(x)
assert True
def test_bottom_simple_conv1_net(self, conv_x):
hps = HyperParams()
net = BottomSimpleConv1d(hps)
out = net.forward(conv_x)
assert True
def test_bottom_simple_mlp_net(self, flat_x):
hps = HyperParams()
net = BottomSimpleMLPNet(hps)
out = net.forward(flat_x)
assert True
def test_base_net(self):
hps = HyperParams()
net = BaseNet(hps)
assert net.hps.activation == 'relu'
def test_hyper_params(self):
path_to_json = os.path.join(get_project_root(), 'res_experiments',
'hps_base_mlp.json')
hps = HyperParams.from_file(path_to_json=path_to_json)
assert hps.activation == 'relu'
def test_resnet(self, conv_x):
hps = HyperParams()
net = BottomResNet(hps)
out = net.forward(conv_x.unsqueeze(2))
assert True
def test_top_net_init(self):
hps = HyperParams()
net = TopNet(hps)
convnet = Conv1dModel(hps)
assert True