diff --git a/EIVPackage/EIVArchitectures/Networks.py b/EIVPackage/EIVArchitectures/Networks.py
index d9c0f1ba854549882ba22499232e8d33e1619334..8928b3fde92517f7397eb0542f8189ea9bf132bd 100644
--- a/EIVPackage/EIVArchitectures/Networks.py
+++ b/EIVPackage/EIVArchitectures/Networks.py
@@ -32,7 +32,7 @@ class FNNEIV(nn.Module):
**Note**:
- To change the deming factor afterwards, use the method `change_deming`
- To change fixed_std_x afterwards, use the method `change_fixed_std_x`
- - To change std_y use the method `change_std_x`
+ - To change std_y use the method `change_std_y`
"""
LeakyReLUSlope = 1e-2
def __init__(self, p = 0.2, init_std_y=1.0, precision_prior_zeta=0.0,
@@ -231,8 +231,9 @@ class FNNEIV(nn.Module):
sigma = torch.mean(sigma, dim=1)
return pred, sigma
- def predictive_logdensity(self, x_or_predictions, y, number_of_draws=[100, 5], number_of_parameter_chunks = None, remove_graph=True,
- average_batch_dimension=True, scale_labels=None,
+ def predictive_logdensity(self, x_or_predictions, y,
+ number_of_draws=[100, 5], number_of_parameter_chunks = None,
+ remove_graph=True, average_batch_dimension=True, scale_labels=None,
decouple_dimensions=False):
"""
Computes the logarithm of the predictive density evaluated at `y`. If
@@ -264,7 +265,8 @@ class FNNEIV(nn.Module):
False.
"""
if type(x_or_predictions) is torch.tensor:
- out, sigmas = self.predict(x_or_predictions, number_of_draws=number_of_draws,
+ out, sigmas = self.predict(x_or_predictions,
+ number_of_draws=number_of_draws,
number_of_parameter_chunks=number_of_parameter_chunks,
remove_graph=remove_graph,
take_average_of_prediction=False)
@@ -307,7 +309,8 @@ class FNNEIV(nn.Module):
else:
return predictive_log_density_values
- def predict_mean_and_unc(self, x, number_of_draws=[100,5], number_of_parameter_chunks = None,
+ def predict_mean_and_unc(self, x, number_of_draws=[100,5],
+ number_of_parameter_chunks = None,
remove_graph=True):
"""
Take the mean and standard deviation over `number_of_draws` forward
@@ -349,6 +352,7 @@ class FNNBer(nn.Module):
:param h: A list specifying the number of neurons in each layer.
:param std_y_requires_grad: Whether `sigma_y` will require_grad and thus
be updated during optimization. Defaults to False.
+ To change std_y use the method `change_std_y`
"""
LeakyReLUSlope = 1e-2
def __init__(self, p=0.2, init_std_y=1.0, h=[10, 1024,1024,1024,1024, 1],
diff --git a/EIVPackage/EIVTrainingRoutines/loss_functions.py b/EIVPackage/EIVTrainingRoutines/loss_functions.py
index 88e924df76c1b2bd00e4c4abcdf18089f699539e..e1c552d3d621dd29c91b555c9c50a1089c836a12 100644
--- a/EIVPackage/EIVTrainingRoutines/loss_functions.py
+++ b/EIVPackage/EIVTrainingRoutines/loss_functions.py
@@ -16,13 +16,17 @@ def nll_reg_loss(net, x, y, reg):
:param y: A torch.tensor, the output.
:param reg: A non-negative float, the regularization.
"""
- out, std_y = net(x)
+ out, sigmas = net(x)
# Add label dimension to y if missing
if len(y.shape) <= 1:
y = y.view((-1,1))
+ # squeeze last dimensions into one
+ y = y.view((*y.shape[:1], -1))
+ sigmas = sigmas.view((*sigmas.shape[:1], -1))
+ out = out.view((*out.shape[:1], -1))
assert out.shape == y.shape
- neg_log_likelihood = torch.mean(0.5* torch.log(2*pi*std_y**2) \
- + ((out-y)**2)/(2*std_y**2))
+ neg_log_likelihood = torch.mean(torch.sum(0.5* torch.log(2*pi*sigmas**2) \
+ + ((out-y)**2)/(2*sigmas**2), dim=1))
regularization = net.regularizer(x, lamb=reg)
return neg_log_likelihood + regularization
@@ -45,12 +49,17 @@ def nll_eiv(net, x, y, reg, number_of_draws=5):
regularization = net.regularizer(x, lamb=reg)
# repeat_tensors
x, y = repeat_tensors(x, y, number_of_draws=number_of_draws)
- pred, sigma = net(x, repetition=number_of_draws)
+ out, sigmas = net(x, repetition=number_of_draws)
# split into chunks of size number_of_draws along batch dimension
- pred, sigma, y = reshape_to_chunks(pred, sigma, y, number_of_draws=number_of_draws)
- assert pred.shape == y.shape
+ out, sigmas, y = reshape_to_chunks(out, sigmas, y,
+ number_of_draws=number_of_draws)
+ # squeeze last dimensions into one
+ y = y.view((*y.shape[:2], -1))
+ sigmas = sigmas.view((*sigmas.shape[:2], -1))
+ out = out.view((*out.shape[:2], -1))
+ assert out.shape == y.shape
# apply logsumexp to chunks and average the results
- nll = -1 * (torch.logsumexp(-1 * sigma.log()
- -((y-pred)**2)/(2*sigma**2), dim=1)
+ nll = -1 * (torch.logsumexp(torch.sum(-1/2 * torch.log(sigmas**2 * 2 * pi)
+ -((y-out)**2)/(2*sigmas**2), dim=2), dim=1)
- np.log(number_of_draws)).mean()
return nll + regularization
diff --git a/EIVPackage/EIVTrainingRoutines/train_and_store.py b/EIVPackage/EIVTrainingRoutines/train_and_store.py
index eb0378e953c00eee2bb0465b1aea8580b92baa01..908c4457baad75de7fbe40c66716ae3ebe7f5c7c 100644
--- a/EIVPackage/EIVTrainingRoutines/train_and_store.py
+++ b/EIVPackage/EIVTrainingRoutines/train_and_store.py
@@ -69,6 +69,15 @@ class TrainEpoch():
"""
pass
+
+ def post_train_update(self, net, epoch=None):
+ """
+ Will be executed after the training is finished
+ :param net: The current net, a torch.nn.Module
+ :param epoch: Tue last epochn number, an integer.
+ """
+ pass
+
def extra_report(self, net, step):
"""
Overwrite for reporting on state of net
@@ -156,6 +165,8 @@ def train_and_store(net, epoch_map, number_of_epochs, save_file, **kwargs):
Calls `epoch_map` with `epoch` and the current epoch number
`number_of_epochs` times and stores a list of
its output as a pickled file under `save_file`.
+ After this (the training) is done, `epoch_map.post_train_update(net,
+ epoch_number)` is called, if existent.
**Note**: The output of `epoch_map` is supposed to
consist of 4 specific arguments, see below.
:param net: A torch.nn.Module
@@ -179,6 +190,11 @@ def train_and_store(net, epoch_map, number_of_epochs, save_file, **kwargs):
test_loss_collection.append(test_loss)
std_x_collection.append(std_x)
std_y_collection.append(std_y)
+ try:
+ epoch_map.post_train_update(net, number_of_epochs-1)
+ print('Training done. Performed post training updating.')
+ except AttributeError:
+ print('Training done. No post train updating performed.')
# Saving
state_dict = net.state_dict()
to_save = {
diff --git a/Experiments/configurations/eiv_california.json b/Experiments/configurations/eiv_california.json
index 29a71622d99299d2c7ccde4f148b560de52418ee..6cfa57f9f95c63dd347629d0acf4df37b96b180f 100644
--- a/Experiments/configurations/eiv_california.json
+++ b/Experiments/configurations/eiv_california.json
@@ -9,7 +9,7 @@
"report_point": 5,
"p": 0.1,
"lr_update": 20,
- "epoch_offset": 10,
+ "std_y_update_points": [10,5],
"eiv_prediction_number_of_draws": 100,
"eiv_prediction_number_of_batches": 10,
"init_std_y_list": [0.5],
diff --git a/Experiments/configurations/eiv_concrete.json b/Experiments/configurations/eiv_concrete.json
index 11121c144b429ad2f970aa815f2a4e7fed8a893a..0e234346f6dda090daacb9aebf33f8c3f666081c 100644
--- a/Experiments/configurations/eiv_concrete.json
+++ b/Experiments/configurations/eiv_concrete.json
@@ -9,7 +9,7 @@
"report_point": 5,
"p": 0.2,
"lr_update": 20,
- "epoch_offset": 10,
+ "std_y_update_points": 10,
"eiv_prediction_number_of_draws": 100,
"eiv_prediction_number_of_batches": 10,
"init_std_y_list": [0.5],
diff --git a/Experiments/configurations/eiv_energy.json b/Experiments/configurations/eiv_energy.json
index bca6774dd7b4ce71ca78b487246d6a5e7f4b76f3..f13ab1b52417aa7ed20c4db9d75a0c653ff0b321 100644
--- a/Experiments/configurations/eiv_energy.json
+++ b/Experiments/configurations/eiv_energy.json
@@ -9,7 +9,7 @@
"report_point": 5,
"p": 0.2,
"lr_update": 100,
- "epoch_offset": 100,
+ "std_y_update_points": 100,
"eiv_prediction_number_of_draws": 100,
"eiv_prediction_number_of_batches": 10,
"init_std_y_list": [0.5],
diff --git a/Experiments/configurations/eiv_kin8nm.json b/Experiments/configurations/eiv_kin8nm.json
index fa3718a096e417c2ab456a8589009d0052ae6670..df2e4bafbee99aca95dcd36f8f67097d3a6ce3b5 100644
--- a/Experiments/configurations/eiv_kin8nm.json
+++ b/Experiments/configurations/eiv_kin8nm.json
@@ -9,7 +9,7 @@
"report_point": 5,
"p": 0.2,
"lr_update": 20,
- "epoch_offset": 19,
+ "std_y_update_points": 19,
"eiv_prediction_number_of_draws": 100,
"eiv_prediction_number_of_batches": 10,
"init_std_y_list": [0.5],
diff --git a/Experiments/configurations/eiv_msd.json b/Experiments/configurations/eiv_msd.json
index 1c2527630dac6172c9271ee230b5c34a2abfd71f..a738f2e8bf034e173f60b17e5cd6d2f7d6fdb856 100644
--- a/Experiments/configurations/eiv_msd.json
+++ b/Experiments/configurations/eiv_msd.json
@@ -9,7 +9,7 @@
"report_point": 5,
"p": 0.2,
"lr_update": 4,
- "epoch_offset": 4,
+ "std_y_update_points": 4,
"eiv_prediction_number_of_draws": 100,
"eiv_prediction_number_of_batches": 10,
"init_std_y_list": [0.5],
diff --git a/Experiments/configurations/eiv_naval.json b/Experiments/configurations/eiv_naval.json
index 3358831d627188d5c2b93ceaa748b77de893c23d..092f4f822abf0e9c677affa8ae275e46309435b8 100644
--- a/Experiments/configurations/eiv_naval.json
+++ b/Experiments/configurations/eiv_naval.json
@@ -9,7 +9,7 @@
"report_point": 5,
"p": 0.2,
"lr_update": 20,
- "epoch_offset": 20,
+ "std_y_update_points": 20,
"eiv_prediction_number_of_draws": 100,
"eiv_prediction_number_of_batches": 10,
"init_std_y_list": [0.5],
diff --git a/Experiments/configurations/eiv_power.json b/Experiments/configurations/eiv_power.json
index 842f6e2662bef07696de1eb5f041e677b586dcd6..ab5649f61de1ea2214249b8a07e25cd5b2adea9e 100644
--- a/Experiments/configurations/eiv_power.json
+++ b/Experiments/configurations/eiv_power.json
@@ -9,7 +9,7 @@
"report_point": 5,
"p": 0.2,
"lr_update": 10,
- "epoch_offset": 15,
+ "std_y_update_points": 15,
"eiv_prediction_number_of_draws": 100,
"eiv_prediction_number_of_batches": 10,
"init_std_y_list": [0.5],
diff --git a/Experiments/configurations/eiv_protein.json b/Experiments/configurations/eiv_protein.json
index 97f5946d431d04bc83c71904846152160fa0c7d7..9623799e3d8b85b2f568987afb4d406fbb7fe45c 100644
--- a/Experiments/configurations/eiv_protein.json
+++ b/Experiments/configurations/eiv_protein.json
@@ -9,7 +9,7 @@
"report_point": 5,
"p": 0.2,
"lr_update": 10,
- "epoch_offset": 10,
+ "std_y_update_points": 10,
"eiv_prediction_number_of_draws": 100,
"eiv_prediction_number_of_batches": 10,
"init_std_y_list": [0.5],
diff --git a/Experiments/configurations/eiv_wine.json b/Experiments/configurations/eiv_wine.json
index 0b40a61272719cee6afc839f171103c54a830af8..f0beaf490846c2281a477f627f05c98b18cfa7f3 100644
--- a/Experiments/configurations/eiv_wine.json
+++ b/Experiments/configurations/eiv_wine.json
@@ -9,7 +9,7 @@
"report_point": 5,
"p": 0.2,
"lr_update": 30,
- "epoch_offset": 50,
+ "std_y_update_points": 50,
"eiv_prediction_number_of_draws": 100,
"eiv_prediction_number_of_batches": 10,
"init_std_y_list": [0.5],
diff --git a/Experiments/configurations/eiv_yacht.json b/Experiments/configurations/eiv_yacht.json
index c668113462816474eca51186853bbc241ca3160d..7617289d9e49bb9ceddd9962d6d865fee3413ae7 100644
--- a/Experiments/configurations/eiv_yacht.json
+++ b/Experiments/configurations/eiv_yacht.json
@@ -9,7 +9,7 @@
"report_point": 5,
"p": 0.2,
"lr_update": 200,
- "epoch_offset": 20,
+ "std_y_update_points": [1,500],
"eiv_prediction_number_of_draws": 100,
"eiv_prediction_number_of_batches": 10,
"init_std_y_list": [0.5],
diff --git a/Experiments/configurations/noneiv_california.json b/Experiments/configurations/noneiv_california.json
index 5005b43d8272c22893903a8a6c7494edeed4e554..e0729764699ba987d4a8e8b4e344601303c16f3b 100644
--- a/Experiments/configurations/noneiv_california.json
+++ b/Experiments/configurations/noneiv_california.json
@@ -9,7 +9,7 @@
"report_point": 5,
"p": 0.1,
"lr_update": 20,
- "epoch_offset": 0 ,
+ "std_y_update_points": [10,5] ,
"noneiv_prediction_number_of_draws": 100,
"noneiv_prediction_number_of_batches": 10,
"init_std_y_list": [0.5],
diff --git a/Experiments/configurations/noneiv_concrete.json b/Experiments/configurations/noneiv_concrete.json
index 66552b7193675eda193479ff9909494fc2bc2133..15e28c53a67064987c267f932733a09d00698c43 100644
--- a/Experiments/configurations/noneiv_concrete.json
+++ b/Experiments/configurations/noneiv_concrete.json
@@ -9,7 +9,7 @@
"report_point": 5,
"p": 0.2,
"lr_update": 20,
- "epoch_offset": 10,
+ "std_y_update_points": 10,
"noneiv_prediction_number_of_draws": 100,
"noneiv_prediction_number_of_batches": 10,
"init_std_y_list": [0.5],
diff --git a/Experiments/configurations/noneiv_energy.json b/Experiments/configurations/noneiv_energy.json
index 74eb45c721793e94f0efbb4d67a3f4acaa05e8da..18c068ea7a9cb96f1075a522ae3f3ad6552bacac 100644
--- a/Experiments/configurations/noneiv_energy.json
+++ b/Experiments/configurations/noneiv_energy.json
@@ -9,7 +9,7 @@
"report_point": 5,
"p": 0.2,
"lr_update": 100,
- "epoch_offset": 100,
+ "std_y_update_points": 100,
"noneiv_prediction_number_of_draws": 100,
"noneiv_prediction_number_of_batches": 10,
"init_std_y_list": [0.5],
diff --git a/Experiments/configurations/noneiv_kin8nm.json b/Experiments/configurations/noneiv_kin8nm.json
index 22615f61b00cf2b06c6e5fbc0051f3c86e8f92cb..8ecf817149b1a37dd9a4ab5c44fd2700795fbcde 100644
--- a/Experiments/configurations/noneiv_kin8nm.json
+++ b/Experiments/configurations/noneiv_kin8nm.json
@@ -9,7 +9,7 @@
"report_point": 5,
"p": 0.2,
"lr_update": 20,
- "epoch_offset": 19,
+ "std_y_update_points": 19,
"noneiv_prediction_number_of_draws": 100,
"noneiv_prediction_number_of_batches": 10,
"init_std_y_list": [0.5],
diff --git a/Experiments/configurations/noneiv_msd.json b/Experiments/configurations/noneiv_msd.json
index 1c4f338fe3152c6a21d66cff05376efcbf88f148..a0e2f73beb63e165c7da5266b774f227e90310c1 100644
--- a/Experiments/configurations/noneiv_msd.json
+++ b/Experiments/configurations/noneiv_msd.json
@@ -9,7 +9,7 @@
"report_point": 5,
"p": 0.2,
"lr_update": 4,
- "epoch_offset": 4,
+ "std_y_update_points": 4,
"noneiv_prediction_number_of_draws": 100,
"noneiv_prediction_number_of_batches": 10,
"init_std_y_list": [0.5],
diff --git a/Experiments/configurations/noneiv_naval.json b/Experiments/configurations/noneiv_naval.json
index 5dea6c49692e193dc25818ae83c6f467860a1520..2da83b73101d1a5a957f9773053b9610b0cb8aff 100644
--- a/Experiments/configurations/noneiv_naval.json
+++ b/Experiments/configurations/noneiv_naval.json
@@ -9,7 +9,7 @@
"report_point": 5,
"p": 0.2,
"lr_update": 20,
- "epoch_offset": 20,
+ "std_y_update_points": 20,
"noneiv_prediction_number_of_draws": 100,
"noneiv_prediction_number_of_batches": 10,
"init_std_y_list": [0.5],
diff --git a/Experiments/configurations/noneiv_power.json b/Experiments/configurations/noneiv_power.json
index 6e524e6c4f684f9c0affc983f3d2d5a7e4597248..48aac9dcf4cc208095142760d7bf286c2589b22b 100644
--- a/Experiments/configurations/noneiv_power.json
+++ b/Experiments/configurations/noneiv_power.json
@@ -9,7 +9,7 @@
"report_point": 5,
"p": 0.2,
"lr_update": 10,
- "epoch_offset": 15,
+ "std_y_update_points": 15,
"noneiv_prediction_number_of_draws": 100,
"noneiv_prediction_number_of_batches": 10,
"init_std_y_list": [0.5],
diff --git a/Experiments/configurations/noneiv_protein.json b/Experiments/configurations/noneiv_protein.json
index 1edd04fa1d28b8634080beec1865d30b6f854153..454deaef823c1a25c03588c4f430b5d0f67e5e93 100644
--- a/Experiments/configurations/noneiv_protein.json
+++ b/Experiments/configurations/noneiv_protein.json
@@ -9,7 +9,7 @@
"report_point": 5,
"p": 0.2,
"lr_update": 10,
- "epoch_offset": 10,
+ "std_y_update_points": 10,
"noneiv_prediction_number_of_draws": 100,
"noneiv_prediction_number_of_batches": 10,
"init_std_y_list": [0.5],
diff --git a/Experiments/configurations/noneiv_wine.json b/Experiments/configurations/noneiv_wine.json
index 717fb33c371b6d3743da58c4489ed96bd3e62ea6..ddd2199fae58e2dd62409d554f51983b5bd1ef02 100644
--- a/Experiments/configurations/noneiv_wine.json
+++ b/Experiments/configurations/noneiv_wine.json
@@ -9,7 +9,7 @@
"report_point": 5,
"p": 0.2,
"lr_update": 30,
- "epoch_offset": 50,
+ "std_y_update_points": 50,
"noneiv_prediction_number_of_draws": 100,
"noneiv_prediction_number_of_batches": 10,
"init_std_y_list": [0.5],
diff --git a/Experiments/configurations/noneiv_yacht.json b/Experiments/configurations/noneiv_yacht.json
index edb775cc550ec8b2ca8ed32a4962cb64eea20e65..81a0d1459d5faf474a922d7916336c57c255c07f 100644
--- a/Experiments/configurations/noneiv_yacht.json
+++ b/Experiments/configurations/noneiv_yacht.json
@@ -9,7 +9,7 @@
"report_point": 5,
"p": 0.2,
"lr_update": 200,
- "epoch_offset": 20,
+ "std_y_update_points": [1,500],
"noneiv_prediction_number_of_draws": 100,
"noneiv_prediction_number_of_batches": 10,
"init_std_y_list": [0.5],
diff --git a/Experiments/evaluate_tabular.py b/Experiments/evaluate_tabular.py
index 7493e13b37fd417bb00fec04873d740a79781950..ba0654f071cdb96cfa72145fe048da4b088bc16b 100644
--- a/Experiments/evaluate_tabular.py
+++ b/Experiments/evaluate_tabular.py
@@ -144,7 +144,7 @@ def collect_metrics(x,y, seed=0,
noise_state = net.noise_is_on
net.train()
net.noise_on()
- not_averaged_predictions = net.predict(x, number_of_draws=noneiv_number_of_draws,
+ not_averaged_predictions = net.predict(x, number_of_draws=eiv_number_of_draws,
take_average_of_prediction=False)
eiv_mean = torch.mean(not_averaged_predictions[0], dim=1)
if len(y.shape) <= 1:
diff --git a/Experiments/train_eiv.py b/Experiments/train_eiv.py
index 78f6d61d5f0af1d0e29de0f10f99275db40f42c4..eb8d9d88292dd63130b38bc78db4eb1ed1db9014 100644
--- a/Experiments/train_eiv.py
+++ b/Experiments/train_eiv.py
@@ -40,10 +40,11 @@ report_point = conf_dict["report_point"]
p = conf_dict["p"]
lr_update = conf_dict["lr_update"]
# offset before updating sigma_y after each epoch
-epoch_offset = conf_dict["epoch_offset"]
+std_y_update_points = conf_dict["std_y_update_points"]
# will be used to predict the RMSE and update sigma_y accordingly
eiv_prediction_number_of_draws = conf_dict["eiv_prediction_number_of_draws"]
-eiv_prediction_number_of_batches = conf_dict["eiv_prediction_number_of_batches"]
+eiv_prediction_number_of_batches = \
+ conf_dict["eiv_prediction_number_of_batches"]
init_std_y_list = conf_dict["init_std_y_list"]
fixed_std_x = conf_dict['fixed_std_x']
gamma = conf_dict["gamma"]
@@ -54,7 +55,8 @@ print(f"Training on {long_dataname} data")
try:
gpu_number = conf_dict["gpu_number"]
- device = torch.device(f'cuda:{gpu_number}' if torch.cuda.is_available() else 'cpu')
+ device = torch.device(f'cuda:{gpu_number}' if torch.cuda.is_available()
+ else 'cpu')
except KeyError:
device = torch.device('cpu')
@@ -65,7 +67,7 @@ seed_list = range(seed_range[0], seed_range[1])
def set_seeds(seed):
torch.backends.cudnn.benchmark = False
- np.random.seed(seed)
+ np.random.seed(seed)
random.seed(seed)
torch.manual_seed(seed)
@@ -83,33 +85,71 @@ class UpdatedTrainEpoch(train_and_store.TrainEpoch):
self.lr_scheduler = torch.optim.lr_scheduler.StepLR(
self.optimizer, lr_update, gamma)
+ def update_std_y(self, net):
+ """
+ Update the std_y of `net` via the RMSE of the prediction.
+ """
+ net_train_state = net.training
+ net_noise_state = net.noise_is_on
+ net.train()
+ net.noise_on()
+ pred_collection = []
+ y_collection = []
+ for i, (x,y) in enumerate(self.train_dataloader):
+ if i >= eiv_prediction_number_of_batches:
+ break
+ if len(y.shape) <= 1:
+ y = y.view((-1,1))
+ x,y = x.to(device), y.to(device)
+ pred, _ = net.predict(x,
+ number_of_draws=eiv_prediction_number_of_draws,
+ remove_graph = True,
+ take_average_of_prediction=True)
+ pred_collection.append(pred)
+ y_collection.append(y)
+ pred_collection = torch.cat(pred_collection, dim=0)
+ y_collection = torch.cat(y_collection, dim=0)
+ assert pred_collection.shape == y_collection.shape
+ rmse = torch.sqrt(torch.mean((pred_collection - y_collection)**2))
+ net.change_std_y(rmse)
+ if not net_train_state:
+ net.eval()
+ if not net_noise_state:
+ net.noise_off()
+
+ def check_if_update_std_y(self, epoch):
+ """
+ Check whether to update std_y according to `epoch_number` and
+ `std_y_update_points`. If the later is an integer, after all epochs
+ greater than this number an update will be made (i.e. `True` will
+ be returned). If it is a list, only `epoch_number` greater than
+ `std_y_update_points[0]` that divide `std_y_update_points[1]` will
+ result in a True.
+ """
+ if type(std_y_update_points) is int:
+ return epoch >= std_y_update_points
+ else:
+ assert type(std_y_update_points) is list
+ return epoch >= std_y_update_points[0]\
+ and epoch % std_y_update_points[1] == 0
def post_epoch_update(self, net, epoch):
"""
Overwrites the corresponding method
"""
- if epoch >= epoch_offset:
- pred_collection = []
- y_collection = []
- for i, (x,y) in enumerate(self.train_dataloader):
- if i >= eiv_prediction_number_of_batches:
- break
- if len(y.shape) <= 1:
- y = y.view((-1,1))
- x,y = x.to(device), y.to(device)
- pred, _ = net.predict(x,
- number_of_draws=eiv_prediction_number_of_draws,
- remove_graph = True,
- take_average_of_prediction=True)
- pred_collection.append(pred)
- y_collection.append(y)
- pred_collection = torch.cat(pred_collection, dim=0)
- y_collection = torch.cat(y_collection, dim=0)
- assert pred_collection.shape == y_collection.shape
- rmse = torch.sqrt(torch.mean((pred_collection - y_collection)**2))
- net.change_std_y(rmse)
+ if self.check_if_update_std_y(epoch):
+ self.update_std_y(net)
self.lr_scheduler.step()
+ def post_train_update(self, net, epoch):
+ """
+ Overwrites the corresponding method. If std_y of `net` was not updated
+ in the last training step, update it when finished with training.
+ `epoch` should be the number of the last training epoch.
+ """
+ if not self.check_if_update_std_y(epoch):
+ self.update_std_y(net)
+
def extra_report(self, net, i):
"""
Overwrites the corresponding method
@@ -119,7 +159,6 @@ class UpdatedTrainEpoch(train_and_store.TrainEpoch):
rmse_chain.append(rmse)
writer.add_scalar('RMSE', rmse, self.total_count)
writer.add_scalar('std_y', self.last_std_y, self.total_count)
- writer.add_scalar('RMSE:std_y', rmse/self.last_std_y, 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}')
diff --git a/Experiments/train_noneiv.py b/Experiments/train_noneiv.py
index e63dbf233062d7ee47465cd5cd50f5708271cd68..316dda1e8f5fbf42f46ae9135dadd5c803623df8 100644
--- a/Experiments/train_noneiv.py
+++ b/Experiments/train_noneiv.py
@@ -40,10 +40,11 @@ report_point = conf_dict["report_point"]
p = conf_dict["p"]
lr_update = conf_dict["lr_update"]
# offset before updating sigma_y after each epoch
-epoch_offset = conf_dict["epoch_offset"]
+std_y_update_points = conf_dict["std_y_update_points"]
# will be used to predict the RMSE and update sigma_y accordingly
noneiv_prediction_number_of_draws = conf_dict["noneiv_prediction_number_of_draws"]
-noneiv_prediction_number_of_batches = conf_dict["noneiv_prediction_number_of_batches"]
+noneiv_prediction_number_of_batches = \
+ conf_dict["noneiv_prediction_number_of_batches"]
init_std_y_list = conf_dict["init_std_y_list"]
gamma = conf_dict["gamma"]
hidden_layers = conf_dict["hidden_layers"]
@@ -53,7 +54,8 @@ print(f"Training on {long_dataname} data")
try:
gpu_number = conf_dict["gpu_number"]
- device = torch.device(f'cuda:{gpu_number}' if torch.cuda.is_available() else 'cpu')
+ device = torch.device(f'cuda:{gpu_number}' if torch.cuda.is_available()
+ else 'cpu')
except KeyError:
device = torch.device('cpu')
@@ -82,33 +84,67 @@ class UpdatedTrainEpoch(train_and_store.TrainEpoch):
self.lr_scheduler = torch.optim.lr_scheduler.StepLR(
self.optimizer, lr_update, gamma)
+ def update_std_y(self, net):
+ """
+ Update the std_y of `net` via the RMSE of the prediction.
+ """
+ net_train_state = net.training
+ net.train()
+ pred_collection = []
+ y_collection = []
+ for i, (x,y) in enumerate(self.train_dataloader):
+ if i >= noneiv_prediction_number_of_batches:
+ break
+ if len(y.shape) <= 1:
+ y = y.view((-1,1))
+ x,y = x.to(device), y.to(device)
+ pred, _ = net.predict(x,
+ number_of_draws=noneiv_prediction_number_of_draws,
+ remove_graph = True,
+ take_average_of_prediction=True)
+ pred_collection.append(pred)
+ y_collection.append(y)
+ pred_collection = torch.cat(pred_collection, dim=0)
+ y_collection = torch.cat(y_collection, dim=0)
+ assert pred_collection.shape == y_collection.shape
+ rmse = torch.sqrt(torch.mean((pred_collection - y_collection)**2))
+ net.change_std_y(rmse)
+ if not net_train_state:
+ net.eval()
+
+ def check_if_update_std_y(self, epoch):
+ """
+ Check whether to update std_y according to `epoch_number` and
+ `std_y_update_points`. If the later is an integer, after all epochs
+ greater than this number an update will be made (i.e. `True` will
+ be returned). If it is a list, only `epoch_number` greater than
+ `std_y_update_points[0]` that divide `std_y_update_points[1]` will
+ result in a True.
+ """
+ if type(std_y_update_points) is int:
+ return epoch >= std_y_update_points
+ else:
+ assert type(std_y_update_points) is list
+ return epoch >= std_y_update_points[0]\
+ and epoch % std_y_update_points[1] == 0
def post_epoch_update(self, net, epoch):
"""
Overwrites the corresponding method
"""
- if epoch >= epoch_offset:
- pred_collection = []
- y_collection = []
- for i, (x,y) in enumerate(self.train_dataloader):
- if i>= noneiv_prediction_number_of_batches:
- break
- if len(y.shape) <= 1:
- y = y.view((-1,1))
- x,y = x.to(device), y.to(device)
- pred, _ = net.predict(x,
- number_of_draws=noneiv_prediction_number_of_draws,
- remove_graph = True,
- take_average_of_prediction=True)
- pred_collection.append(pred)
- y_collection.append(y)
- pred_collection = torch.cat(pred_collection, dim=0)
- y_collection = torch.cat(y_collection, dim=0)
- assert pred_collection.shape == y_collection.shape
- rmse = torch.sqrt(torch.mean((pred_collection - y_collection)**2))
- net.change_std_y(rmse)
+ if self.check_if_update_std_y(epoch):
+ self.update_std_y(net)
self.lr_scheduler.step()
+ def post_train_update(self, net, epoch):
+ """
+ Overwrites the corresponding method. If std_y of `net` was not updated
+ in the last training step, update it when finished with training.
+ `epoch` should be the number of the last training epoch.
+ """
+ if not self.check_if_update_std_y(epoch):
+ self.update_std_y(net)
+
def extra_report(self, net, i):
"""
Overwrites the corresponding method
@@ -118,7 +154,6 @@ class UpdatedTrainEpoch(train_and_store.TrainEpoch):
rmse_chain.append(rmse)
writer.add_scalar('RMSE', rmse, self.total_count)
writer.add_scalar('std_y', self.last_std_y, self.total_count)
- writer.add_scalar('RMSE:std_y', rmse/self.last_std_y, 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}')