diff --git a/Experiments/train_eiv_msd.py b/Experiments/train_eiv_msd.py
new file mode 100644
index 0000000000000000000000000000000000000000..16e617f0a64fb8f9567ac2846a22cefff109b901
--- /dev/null
+++ b/Experiments/train_eiv_msd.py
@@ -0,0 +1,155 @@
+"""
+Train EiV model on the million song dataset using different seeds
+"""
+import random
+import os
+
+import numpy as np
+import torch
+import torch.backends.cudnn
+from torch.utils.data import DataLoader
+from torch.utils.tensorboard.writer import SummaryWriter
+
+from EIVArchitectures import Networks, initialize_weights
+from EIVData.million_song import load_data
+from EIVTrainingRoutines import train_and_store, loss_functions
+
+# hyperparameters
+lr = 1e-3
+batch_size = 100
+test_batch_size = 600
+number_of_epochs = 10
+unscaled_reg = 10
+report_point = 5
+p = 0.2
+lr_update = 4
+# pretraining = 300
+epoch_offset = 4
+init_std_y_list = [0.5]
+gamma = 0.5
+hidden_layers = [1024, 1024, 1024, 1024]
+device = torch.device('cuda:1' if torch.cuda.is_available() else 'cpu')
+fixed_std_x = 0.05
+
+# reproducability
+def set_seeds(seed):
+ torch.backends.cudnn.benchmark = False
+ np.random.seed(seed)
+ random.seed(seed)
+ torch.manual_seed(seed)
+seed_list = range(10)
+
+# to store the RMSE
+rmse_chain = []
+
+class UpdatedTrainEpoch(train_and_store.TrainEpoch):
+ def pre_epoch_update(self, net, epoch):
+ """
+ Overwrites the corresponding method
+ """
+ if epoch == 0:
+ self.lr = self.initial_lr
+ self.optimizer = torch.optim.Adam(net.parameters(), lr=self.lr)
+ self.lr_scheduler = torch.optim.lr_scheduler.StepLR(
+ self.optimizer, lr_update, gamma)
+
+
+ def post_epoch_update(self, net, epoch):
+ """
+ Overwrites the corresponding method
+ """
+ if epoch >= epoch_offset:
+ net.std_y_par.requires_grad = True
+ self.lr_scheduler.step()
+
+ def extra_report(self, net, i):
+ """
+ Overwrites the corresponding method
+ and fed after initialization of this class
+ """
+ rmse = self.rmse(net).item()
+ rmse_chain.append(rmse)
+ writer.add_scalar('RMSE', rmse, self.total_count)
+ writer.add_scalar('train loss', self.last_train_loss, self.total_count)
+ writer.add_scalar('test loss', self.last_test_loss, self.total_count)
+ print(f'RMSE {rmse:.3f}')
+
+ def rmse(self, net):
+ """
+ Compute the root mean squared error for `net`
+ """
+ net_train_state = net.training
+ net_noise_state = net.noise_is_on
+ net.eval()
+ net.noise_off()
+ x, y = next(iter(self.test_dataloader))
+ if len(y.shape) <= 1:
+ y = y.view((-1,1))
+ out = net(x.to(device))[0].detach().cpu()
+ assert out.shape == y.shape
+ if net_train_state:
+ net.train()
+ if net_noise_state:
+ net.noise_on()
+ return torch.sqrt(torch.mean((out-y)**2))
+
+def train_on_data(init_std_y, seed):
+ """
+ Sets `seed`, loads data and trains an Bernoulli Modell, starting with
+ `init_std_y`.
+ """
+ # set seed
+ set_seeds(seed)
+ # load Datasets
+ train_data, test_data = load_data(seed=seed, splitting_part=0.8,
+ normalize=True)
+ # make dataloaders
+ train_dataloader = DataLoader(train_data, batch_size=batch_size,
+ shuffle=True)
+ test_dataloader = DataLoader(test_data, batch_size=test_batch_size,
+ shuffle=True)
+ # create a net
+ input_dim = train_data[0][0].numel()
+ output_dim = train_data[0][1].numel()
+ net = Networks.FNNEIV(p=p,
+ init_std_y=init_std_y,
+ h=[input_dim, *hidden_layers, output_dim],
+ fixed_std_x=fixed_std_x)
+ net.apply(initialize_weights.glorot_init)
+ net = net.to(device)
+ net.std_y_par.requires_grad = False
+ std_x_map = lambda: net.get_std_x().detach().cpu().item()
+ std_y_map = lambda: net.get_std_y().detach().cpu().item()
+ # regularization
+ reg = unscaled_reg/len(train_data)
+ # create epoch_map
+ criterion = loss_functions.nll_eiv
+ epoch_map = UpdatedTrainEpoch(train_dataloader=train_dataloader,
+ test_dataloader=test_dataloader,
+ criterion=criterion, std_y_map=std_y_map, std_x_map=std_x_map,
+ lr=lr, reg=reg, report_point=report_point, device=device)
+ # run and save
+ save_file = os.path.join('saved_networks',
+ f'eiv_msd'\
+ f'_init_std_y_{init_std_y:.3f}_ureg_{unscaled_reg:.1f}'\
+ f'_p_{p:.2f}_fixed_std_x_{fixed_std_x:.3f}'\
+ f'_seed_{seed}.pkl')
+ train_and_store.train_and_store(net=net,
+ epoch_map=epoch_map,
+ number_of_epochs=number_of_epochs,
+ save_file=save_file)
+
+
+if __name__ == '__main__':
+ for seed in seed_list:
+ # Tensorboard monitoring
+ writer = SummaryWriter(log_dir=f'/home/martin09/tmp/tensorboard/'\
+ f'run_eiv_msd_lr_{lr:.4f}_seed'\
+ f'_{seed}_uregu_{unscaled_reg:.1f}_p_{p:.2f}'\
+ f'_fixed_std_x_{fixed_std_x:.3f}')
+ print(f'>>>>SEED: {seed}')
+ for init_std_y in init_std_y_list:
+ print(f'Using init_std_y={init_std_y:.3f}')
+ train_on_data(init_std_y, seed)
+
+
diff --git a/Experiments/train_eiv_power.py b/Experiments/train_eiv_power.py
new file mode 100644
index 0000000000000000000000000000000000000000..2ddf75af75db7607a21cef6e212c9601d391b421
--- /dev/null
+++ b/Experiments/train_eiv_power.py
@@ -0,0 +1,153 @@
+"""
+Train EiV model on power plant dataset using different seeds
+"""
+import random
+import os
+
+import numpy as np
+import torch
+import torch.backends.cudnn
+from torch.utils.data import DataLoader
+from torch.utils.tensorboard.writer import SummaryWriter
+
+from EIVArchitectures import Networks, initialize_weights
+from EIVData.power_plant import load_data
+from EIVTrainingRoutines import train_and_store, loss_functions
+
+# hyperparameters
+lr = 1e-3
+batch_size = 64
+test_batch_size = 600
+number_of_epochs = 35
+unscaled_reg = 10
+report_point = 5
+p = 0.2
+lr_update = 10
+# pretraining = 300
+epoch_offset = 15
+init_std_y_list = [0.5]
+gamma = 0.5
+hidden_layers = [1024, 1024, 1024, 1024]
+device = torch.device('cuda:1' if torch.cuda.is_available() else 'cpu')
+fixed_std_x = 0.05
+
+# reproducability
+def set_seeds(seed):
+ torch.backends.cudnn.benchmark = False
+ np.random.seed(seed)
+ random.seed(seed)
+ torch.manual_seed(seed)
+seed_list = range(10)
+
+# to store the RMSE
+rmse_chain = []
+
+class UpdatedTrainEpoch(train_and_store.TrainEpoch):
+ def pre_epoch_update(self, net, epoch):
+ """
+ Overwrites the corresponding method
+ """
+ if epoch == 0:
+ self.lr = self.initial_lr
+ self.optimizer = torch.optim.Adam(net.parameters(), lr=self.lr)
+ self.lr_scheduler = torch.optim.lr_scheduler.StepLR(
+ self.optimizer, lr_update, gamma)
+
+
+ def post_epoch_update(self, net, epoch):
+ """
+ Overwrites the corresponding method
+ """
+ if epoch >= epoch_offset:
+ net.std_y_par.requires_grad = True
+ self.lr_scheduler.step()
+
+ def extra_report(self, net, i):
+ """
+ Overwrites the corresponding method
+ and fed after initialization of this class
+ """
+ rmse = self.rmse(net).item()
+ rmse_chain.append(rmse)
+ writer.add_scalar('RMSE', rmse, self.total_count)
+ writer.add_scalar('train loss', self.last_train_loss, self.total_count)
+ writer.add_scalar('test loss', self.last_test_loss, self.total_count)
+ print(f'RMSE {rmse:.3f}')
+
+ def rmse(self, net):
+ """
+ Compute the root mean squared error for `net`
+ """
+ net_train_state = net.training
+ net_noise_state = net.noise_is_on
+ net.eval()
+ net.noise_off()
+ x, y = next(iter(self.test_dataloader))
+ if len(y.shape) <= 1:
+ y = y.view((-1,1))
+ out = net(x.to(device))[0].detach().cpu()
+ assert out.shape == y.shape
+ if net_train_state:
+ net.train()
+ if net_noise_state:
+ net.noise_on()
+ return torch.sqrt(torch.mean((out-y)**2))
+
+def train_on_data(init_std_y, seed):
+ """
+ Sets `seed`, loads data and trains an Bernoulli Modell, starting with
+ `init_std_y`.
+ """
+ # set seed
+ set_seeds(seed)
+ # load Datasets
+ train_data, test_data = load_data(seed=seed, splitting_part=0.8,
+ normalize=True)
+ # make dataloaders
+ train_dataloader = DataLoader(train_data, batch_size=batch_size,
+ shuffle=True)
+ test_dataloader = DataLoader(test_data, batch_size=test_batch_size,
+ shuffle=True)
+ # create a net
+ input_dim = train_data[0][0].numel()
+ output_dim = train_data[0][1].numel()
+ net = Networks.FNNEIV(p=p,
+ init_std_y=init_std_y,
+ h=[input_dim, *hidden_layers, output_dim],
+ fixed_std_x=fixed_std_x)
+ net.apply(initialize_weights.glorot_init)
+ net = net.to(device)
+ net.std_y_par.requires_grad = False
+ std_x_map = lambda: net.get_std_x().detach().cpu().item()
+ std_y_map = lambda: net.get_std_y().detach().cpu().item()
+ # regularization
+ reg = unscaled_reg/len(train_data)
+ # create epoch_map
+ criterion = loss_functions.nll_eiv
+ epoch_map = UpdatedTrainEpoch(train_dataloader=train_dataloader,
+ test_dataloader=test_dataloader,
+ criterion=criterion, std_y_map=std_y_map, std_x_map=std_x_map,
+ lr=lr, reg=reg, report_point=report_point, device=device)
+ # run and save
+ save_file = os.path.join('saved_networks',
+ f'eiv_power'\
+ f'_init_std_y_{init_std_y:.3f}_ureg_{unscaled_reg:.1f}'\
+ f'_p_{p:.2f}_fixed_std_x_{fixed_std_x:.3f}'\
+ f'_seed_{seed}.pkl')
+ train_and_store.train_and_store(net=net,
+ epoch_map=epoch_map,
+ number_of_epochs=number_of_epochs,
+ save_file=save_file)
+
+
+if __name__ == '__main__':
+ for seed in seed_list:
+ # Tensorboard monitoring
+ writer = SummaryWriter(log_dir=f'/home/martin09/tmp/tensorboard/'\
+ f'run_eiv_power_lr_{lr:.4f}_seed'\
+ f'_{seed}_uregu_{unscaled_reg:.1f}_p_{p:.2f}'\
+ f'_fixed_std_x_{fixed_std_x:.3f}')
+ print(f'>>>>SEED: {seed}')
+ for init_std_y in init_std_y_list:
+ print(f'Using init_std_y={init_std_y:.3f}')
+ train_on_data(init_std_y, seed)
diff --git a/Experiments/train_eiv_protein.py b/Experiments/train_eiv_protein.py
new file mode 100644
index 0000000000000000000000000000000000000000..3801d5e07572a7f37b23b457a3adc8b83f3f8418
--- /dev/null
+++ b/Experiments/train_eiv_protein.py
@@ -0,0 +1,155 @@
+"""
+Train EiV model on protein structure dataset using different seeds
+"""
+import random
+import os
+
+import numpy as np
+import torch
+import torch.backends.cudnn
+from torch.utils.data import DataLoader
+from torch.utils.tensorboard.writer import SummaryWriter
+
+from EIVArchitectures import Networks, initialize_weights
+from EIVData.protein_structure import load_data
+from EIVTrainingRoutines import train_and_store, loss_functions
+
+# hyperparameters
+lr = 1e-3
+batch_size = 100
+test_batch_size = 600
+number_of_epochs = 30
+unscaled_reg = 10
+report_point = 5
+p = 0.2
+lr_update = 10
+# pretraining = 300
+epoch_offset = 10
+init_std_y_list = [0.5]
+gamma = 0.5
+hidden_layers = [1024, 1024, 1024, 1024]
+device = torch.device('cuda:1' if torch.cuda.is_available() else 'cpu')
+fixed_std_x = 0.05
+
+# reproducability
+def set_seeds(seed):
+ torch.backends.cudnn.benchmark = False
+ np.random.seed(seed)
+ random.seed(seed)
+ torch.manual_seed(seed)
+seed_list = range(10)
+
+# to store the RMSE
+rmse_chain = []
+
+class UpdatedTrainEpoch(train_and_store.TrainEpoch):
+ def pre_epoch_update(self, net, epoch):
+ """
+ Overwrites the corresponding method
+ """
+ if epoch == 0:
+ self.lr = self.initial_lr
+ self.optimizer = torch.optim.Adam(net.parameters(), lr=self.lr)
+ self.lr_scheduler = torch.optim.lr_scheduler.StepLR(
+ self.optimizer, lr_update, gamma)
+
+
+ def post_epoch_update(self, net, epoch):
+ """
+ Overwrites the corresponding method
+ """
+ if epoch >= epoch_offset:
+ net.std_y_par.requires_grad = True
+ self.lr_scheduler.step()
+
+ def extra_report(self, net, i):
+ """
+ Overwrites the corresponding method
+ and fed after initialization of this class
+ """
+ rmse = self.rmse(net).item()
+ rmse_chain.append(rmse)
+ writer.add_scalar('RMSE', rmse, self.total_count)
+ writer.add_scalar('train loss', self.last_train_loss, self.total_count)
+ writer.add_scalar('test loss', self.last_test_loss, self.total_count)
+ print(f'RMSE {rmse:.3f}')
+
+ def rmse(self, net):
+ """
+ Compute the root mean squared error for `net`
+ """
+ net_train_state = net.training
+ net_noise_state = net.noise_is_on
+ net.eval()
+ net.noise_off()
+ x, y = next(iter(self.test_dataloader))
+ if len(y.shape) <= 1:
+ y = y.view((-1,1))
+ out = net(x.to(device))[0].detach().cpu()
+ assert out.shape == y.shape
+ if net_train_state:
+ net.train()
+ if net_noise_state:
+ net.noise_on()
+ return torch.sqrt(torch.mean((out-y)**2))
+
+def train_on_data(init_std_y, seed):
+ """
+ Sets `seed`, loads data and trains an Bernoulli Modell, starting with
+ `init_std_y`.
+ """
+ # set seed
+ set_seeds(seed)
+ # load Datasets
+ train_data, test_data = load_data(seed=seed, splitting_part=0.8,
+ normalize=True)
+ # make dataloaders
+ train_dataloader = DataLoader(train_data, batch_size=batch_size,
+ shuffle=True)
+ test_dataloader = DataLoader(test_data, batch_size=test_batch_size,
+ shuffle=True)
+ # create a net
+ input_dim = train_data[0][0].numel()
+ output_dim = train_data[0][1].numel()
+ net = Networks.FNNEIV(p=p,
+ init_std_y=init_std_y,
+ h=[input_dim, *hidden_layers, output_dim],
+ fixed_std_x=fixed_std_x)
+ net.apply(initialize_weights.glorot_init)
+ net = net.to(device)
+ net.std_y_par.requires_grad = False
+ std_x_map = lambda: net.get_std_x().detach().cpu().item()
+ std_y_map = lambda: net.get_std_y().detach().cpu().item()
+ # regularization
+ reg = unscaled_reg/len(train_data)
+ # create epoch_map
+ criterion = loss_functions.nll_eiv
+ epoch_map = UpdatedTrainEpoch(train_dataloader=train_dataloader,
+ test_dataloader=test_dataloader,
+ criterion=criterion, std_y_map=std_y_map, std_x_map=std_x_map,
+ lr=lr, reg=reg, report_point=report_point, device=device)
+ # run and save
+ save_file = os.path.join('saved_networks',
+ f'eiv_protein'\
+ f'_init_std_y_{init_std_y:.3f}_ureg_{unscaled_reg:.1f}'\
+ f'_p_{p:.2f}_fixed_std_x_{fixed_std_x:.3f}'\
+ f'_seed_{seed}.pkl')
+ train_and_store.train_and_store(net=net,
+ epoch_map=epoch_map,
+ number_of_epochs=number_of_epochs,
+ save_file=save_file)
+
+
+if __name__ == '__main__':
+ for seed in seed_list:
+ # Tensorboard monitoring
+ writer = SummaryWriter(log_dir=f'/home/martin09/tmp/tensorboard/'\
+ f'run_eiv_protein_lr_{lr:.4f}_seed'\
+ f'_{seed}_uregu_{unscaled_reg:.1f}_p_{p:.2f}'\
+ f'_fixed_std_x_{fixed_std_x:.3f}')
+ print(f'>>>>SEED: {seed}')
+ for init_std_y in init_std_y_list:
+ print(f'Using init_std_y={init_std_y:.3f}')
+ train_on_data(init_std_y, seed)
+
+
diff --git a/Experiments/train_eiv_wine.py b/Experiments/train_eiv_wine.py
new file mode 100644
index 0000000000000000000000000000000000000000..ca561adc2e4d3f9a37060d90b370d21a06ed01d0
--- /dev/null
+++ b/Experiments/train_eiv_wine.py
@@ -0,0 +1,155 @@
+"""
+Train EiV model on wine quality dataset using different seeds
+"""
+import random
+import os
+
+import numpy as np
+import torch
+import torch.backends.cudnn
+from torch.utils.data import DataLoader
+from torch.utils.tensorboard.writer import SummaryWriter
+
+from EIVArchitectures import Networks, initialize_weights
+from EIVData.wine_quality import load_data
+from EIVTrainingRoutines import train_and_store, loss_functions
+
+# hyperparameters
+lr = 1e-3
+batch_size = 32
+test_batch_size = 800
+number_of_epochs = 100
+unscaled_reg = 10
+report_point = 5
+p = 0.2
+lr_update = 30
+# pretraining = 300
+epoch_offset = 50
+init_std_y_list = [0.5]
+gamma = 0.5
+hidden_layers = [1024, 1024, 1024, 1024]
+device = torch.device('cuda:1' if torch.cuda.is_available() else 'cpu')
+fixed_std_x = 0.05
+
+# reproducability
+def set_seeds(seed):
+ torch.backends.cudnn.benchmark = False
+ np.random.seed(seed)
+ random.seed(seed)
+ torch.manual_seed(seed)
+seed_list = range(10)
+
+# to store the RMSE
+rmse_chain = []
+
+class UpdatedTrainEpoch(train_and_store.TrainEpoch):
+ def pre_epoch_update(self, net, epoch):
+ """
+ Overwrites the corresponding method
+ """
+ if epoch == 0:
+ self.lr = self.initial_lr
+ self.optimizer = torch.optim.Adam(net.parameters(), lr=self.lr)
+ self.lr_scheduler = torch.optim.lr_scheduler.StepLR(
+ self.optimizer, lr_update, gamma)
+
+
+ def post_epoch_update(self, net, epoch):
+ """
+ Overwrites the corresponding method
+ """
+ if epoch >= epoch_offset:
+ net.std_y_par.requires_grad = True
+ self.lr_scheduler.step()
+
+ def extra_report(self, net, i):
+ """
+ Overwrites the corresponding method
+ and fed after initialization of this class
+ """
+ rmse = self.rmse(net).item()
+ rmse_chain.append(rmse)
+ writer.add_scalar('RMSE', rmse, self.total_count)
+ writer.add_scalar('train loss', self.last_train_loss, self.total_count)
+ writer.add_scalar('test loss', self.last_test_loss, self.total_count)
+ print(f'RMSE {rmse:.3f}')
+
+ def rmse(self, net):
+ """
+ Compute the root mean squared error for `net`
+ """
+ net_train_state = net.training
+ net_noise_state = net.noise_is_on
+ net.eval()
+ net.noise_off()
+ x, y = next(iter(self.test_dataloader))
+ if len(y.shape) <= 1:
+ y = y.view((-1,1))
+ out = net(x.to(device))[0].detach().cpu()
+ assert out.shape == y.shape
+ if net_train_state:
+ net.train()
+ if net_noise_state:
+ net.noise_on()
+ return torch.sqrt(torch.mean((out-y)**2))
+
+def train_on_data(init_std_y, seed):
+ """
+ Sets `seed`, loads data and trains an Bernoulli Modell, starting with
+ `init_std_y`.
+ """
+ # set seed
+ set_seeds(seed)
+ # load Datasets
+ train_data, test_data = load_data(seed=seed, splitting_part=0.8,
+ normalize=True)
+ # make dataloaders
+ train_dataloader = DataLoader(train_data, batch_size=batch_size,
+ shuffle=True)
+ test_dataloader = DataLoader(test_data, batch_size=test_batch_size,
+ shuffle=True)
+ # create a net
+ input_dim = train_data[0][0].numel()
+ output_dim = train_data[0][1].numel()
+ net = Networks.FNNEIV(p=p,
+ init_std_y=init_std_y,
+ h=[input_dim, *hidden_layers, output_dim],
+ fixed_std_x=fixed_std_x)
+ net.apply(initialize_weights.glorot_init)
+ net = net.to(device)
+ net.std_y_par.requires_grad = False
+ std_x_map = lambda: net.get_std_x().detach().cpu().item()
+ std_y_map = lambda: net.get_std_y().detach().cpu().item()
+ # regularization
+ reg = unscaled_reg/len(train_data)
+ # create epoch_map
+ criterion = loss_functions.nll_eiv
+ epoch_map = UpdatedTrainEpoch(train_dataloader=train_dataloader,
+ test_dataloader=test_dataloader,
+ criterion=criterion, std_y_map=std_y_map, std_x_map=std_x_map,
+ lr=lr, reg=reg, report_point=report_point, device=device)
+ # run and save
+ save_file = os.path.join('saved_networks',
+ f'eiv_wine'\
+ f'_init_std_y_{init_std_y:.3f}_ureg_{unscaled_reg:.1f}'\
+ f'_p_{p:.2f}_fixed_std_x_{fixed_std_x:.3f}'\
+ f'_seed_{seed}.pkl')
+ train_and_store.train_and_store(net=net,
+ epoch_map=epoch_map,
+ number_of_epochs=number_of_epochs,
+ save_file=save_file)
+
+
+if __name__ == '__main__':
+ for seed in seed_list:
+ # Tensorboard monitoring
+ writer = SummaryWriter(log_dir=f'/home/martin09/tmp/tensorboard/'\
+ f'run_eiv_wine_lr_{lr:.4f}_seed'\
+ f'_{seed}_uregu_{unscaled_reg:.1f}_p_{p:.2f}'\
+ f'_fixed_std_x_{fixed_std_x:.3f}')
+ print(f'>>>>SEED: {seed}')
+ for init_std_y in init_std_y_list:
+ print(f'Using init_std_y={init_std_y:.3f}')
+ train_on_data(init_std_y, seed)
+
+
diff --git a/Experiments/train_eiv_yacht.py b/Experiments/train_eiv_yacht.py
new file mode 100644
index 0000000000000000000000000000000000000000..2c9ee88853d77b99fefc3f2507ad827597895d61
--- /dev/null
+++ b/Experiments/train_eiv_yacht.py
@@ -0,0 +1,153 @@
+"""
+Train EiV model on the yacht hydrodynamics dataset using different seeds
+"""
+import random
+import os
+
+import numpy as np
+import torch
+import torch.backends.cudnn
+from torch.utils.data import DataLoader
+from torch.utils.tensorboard.writer import SummaryWriter
+
+from EIVArchitectures import Networks, initialize_weights
+from EIVData.yacht_hydrodynamics import load_data
+from EIVTrainingRoutines import train_and_store, loss_functions
+
+# hyperparameters
+lr = 1e-3
+batch_size = 32
+test_batch_size = 600
+number_of_epochs = 1200
+unscaled_reg = 10
+report_point = 5
+p = 0.2
+lr_update = 200
+# pretraining = 300
+epoch_offset = 250
+init_std_y_list = [0.5]
+gamma = 0.5
+hidden_layers = [1024, 1024, 1024, 1024]
+device = torch.device('cuda:1' if torch.cuda.is_available() else 'cpu')
+fixed_std_x = 0.05
+
+# reproducability
+def set_seeds(seed):
+ torch.backends.cudnn.benchmark = False
+ np.random.seed(seed)
+ random.seed(seed)
+ torch.manual_seed(seed)
+seed_list = range(10)
+
+# to store the RMSE
+rmse_chain = []
+
+class UpdatedTrainEpoch(train_and_store.TrainEpoch):
+ def pre_epoch_update(self, net, epoch):
+ """
+ Overwrites the corresponding method
+ """
+ if epoch == 0:
+ self.lr = self.initial_lr
+ self.optimizer = torch.optim.Adam(net.parameters(), lr=self.lr)
+ self.lr_scheduler = torch.optim.lr_scheduler.StepLR(
+ self.optimizer, lr_update, gamma)
+
+
+ def post_epoch_update(self, net, epoch):
+ """
+ Overwrites the corresponding method
+ """
+ if epoch >= epoch_offset:
+ net.std_y_par.requires_grad = True
+ self.lr_scheduler.step()
+
+ def extra_report(self, net, i):
+ """
+ Overwrites the corresponding method
+ and fed after initialization of this class
+ """
+ rmse = self.rmse(net).item()
+ rmse_chain.append(rmse)
+ writer.add_scalar('RMSE', rmse, self.total_count)
+ writer.add_scalar('train loss', self.last_train_loss, self.total_count)
+ writer.add_scalar('test loss', self.last_test_loss, self.total_count)
+ print(f'RMSE {rmse:.3f}')
+
+ def rmse(self, net):
+ """
+ Compute the root mean squared error for `net`
+ """
+ net_train_state = net.training
+ net_noise_state = net.noise_is_on
+ net.eval()
+ net.noise_off()
+ x, y = next(iter(self.test_dataloader))
+ if len(y.shape) <= 1:
+ y = y.view((-1,1))
+ out = net(x.to(device))[0].detach().cpu()
+ assert out.shape == y.shape
+ if net_train_state:
+ net.train()
+ if net_noise_state:
+ net.noise_on()
+ return torch.sqrt(torch.mean((out-y)**2))
+
+def train_on_data(init_std_y, seed):
+ """
+ Sets `seed`, loads data and trains an Bernoulli Modell, starting with
+ `init_std_y`.
+ """
+ # set seed
+ set_seeds(seed)
+ # load Datasets
+ train_data, test_data = load_data(seed=seed, splitting_part=0.8,
+ normalize=True)
+ # make dataloaders
+ train_dataloader = DataLoader(train_data, batch_size=batch_size,
+ shuffle=True)
+ test_dataloader = DataLoader(test_data, batch_size=test_batch_size,
+ shuffle=True)
+ # create a net
+ input_dim = train_data[0][0].numel()
+ output_dim = train_data[0][1].numel()
+ net = Networks.FNNEIV(p=p,
+ init_std_y=init_std_y,
+ h=[input_dim, *hidden_layers, output_dim],
+ fixed_std_x=fixed_std_x)
+ net.apply(initialize_weights.glorot_init)
+ net = net.to(device)
+ net.std_y_par.requires_grad = False
+ std_x_map = lambda: net.get_std_x().detach().cpu().item()
+ std_y_map = lambda: net.get_std_y().detach().cpu().item()
+ # regularization
+ reg = unscaled_reg/len(train_data)
+ # create epoch_map
+ criterion = loss_functions.nll_eiv
+ epoch_map = UpdatedTrainEpoch(train_dataloader=train_dataloader,
+ test_dataloader=test_dataloader,
+ criterion=criterion, std_y_map=std_y_map, std_x_map=std_x_map,
+ lr=lr, reg=reg, report_point=report_point, device=device)
+ # run and save
+ save_file = os.path.join('saved_networks',
+ f'eiv_yacht'\
+ f'_init_std_y_{init_std_y:.3f}_ureg_{unscaled_reg:.1f}'\
+ f'_p_{p:.2f}_fixed_std_x_{fixed_std_x:.3f}'\
+ f'_seed_{seed}.pkl')
+ train_and_store.train_and_store(net=net,
+ epoch_map=epoch_map,
+ number_of_epochs=number_of_epochs,
+ save_file=save_file)
+
+
+if __name__ == '__main__':
+ for seed in seed_list:
+ # Tensorboard monitoring
+ writer = SummaryWriter(log_dir=f'/home/martin09/tmp/tensorboard/'\
+ f'run_eiv_yacht_lr_{lr:.4f}_seed'\
+ f'_{seed}_uregu_{unscaled_reg:.1f}_p_{p:.2f}'\
+ f'_fixed_std_x_{fixed_std_x:.3f}')
+ print(f'>>>>SEED: {seed}')
+ for init_std_y in init_std_y_list:
+ print(f'Using init_std_y={init_std_y:.3f}')
+ train_on_data(init_std_y, seed)