steffen/masks_augments_compare-21
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

ResidualModule and New Parameters, Speed Manipulation

Browse Source
This commit is contained in:
Si11ium
2020-05-12 12:37:26 +02:00
parent 3fbc98dfa3
commit 28bfcfdce3
8 changed files with 181 additions and 78 deletions
Download Patch File Download Diff File Expand all files Collapse all files

48
models/bandwise_conv_classifier.py
View File

@@ -1,11 +1,10 @@
from argparse import Namespace
from torch import nn
from torch.nn import ModuleDict, ModuleList
from torch.nn import ModuleList
from ml_lib.modules.blocks import ConvModule, LinearModule
from ml_lib.modules.utils import (LightningBaseModule, HorizontalSplitter,
HorizontalMerger, F_x)
from ml_lib.modules.utils import (LightningBaseModule, HorizontalSplitter, HorizontalMerger)
from util.module_mixins import (BaseOptimizerMixin, BaseTrainMixin, BaseValMixin, BinaryMaskDatasetFunction,
BaseDataloadersMixin)
@@ -30,44 +29,39 @@ class BandwiseConvClassifier(BinaryMaskDatasetFunction,
# Additional parameters
self.in_shape = self.dataset.train_dataset.sample_shape
self.conv_filters = self.params.filters
self.criterion = nn.BCELoss()
self.n_band_sections = 4
# Modules
# =============================================================================
self.split = HorizontalSplitter(self.in_shape, self.n_band_sections)
self.conv_dict = ModuleDict()
self.conv_dict.update({f"conv_1_{band_section}":
ConvModule(self.split.shape, self.conv_filters[0], 3, conv_stride=1, **self.params.module_kwargs)
for band_section in range(self.n_band_sections)}
)
self.conv_dict.update({f"conv_2_{band_section}":
ConvModule(self.conv_dict['conv_1_1'].shape, self.conv_filters[1], 3, conv_stride=1,
**self.params.module_kwargs) for band_section in range(self.n_band_sections)}
)
self.conv_dict.update({f"conv_3_{band_section}":
ConvModule(self.conv_dict['conv_2_1'].shape, self.conv_filters[2], 3, conv_stride=1,
**self.params.module_kwargs)
for band_section in range(self.n_band_sections)}
)
k = 3
self.band_list = ModuleList()
for band in range(self.n_band_sections):
last_shape = self.split.shape
conv_list = ModuleList()
for filters in self.conv_filters:
conv_list.append(ConvModule(last_shape, filters, (k,k), conv_stride=(2, 2), conv_padding=2,
**self.params.module_kwargs))
last_shape = conv_list[-1].shape
# self.conv_list.append(ConvModule(last_shape, 1, 1, conv_stride=1, **self.params.module_kwargs))
# last_shape = self.conv_list[-1].shape
self.band_list.append(conv_list)
self.merge = HorizontalMerger(self.conv_dict['conv_3_1'].shape, self.n_band_sections)
self.merge = HorizontalMerger(self.band_list[-1][-1].shape, self.n_band_sections)
self.full_1 = LinearModule(self.flat.shape, self.params.lat_dim, **self.params.module_kwargs)
self.full_1 = LinearModule(self.merge.shape, self.params.lat_dim, **self.params.module_kwargs)
self.full_2 = LinearModule(self.full_1.shape, self.full_1.shape * 2, **self.params.module_kwargs)
self.full_3 = LinearModule(self.full_2.shape, self.full_2.out_features // 2, **self.params.module_kwargs)
self.full_3 = LinearModule(self.full_2.shape, self.full_2.shape // 2, **self.params.module_kwargs)
self.full_out = LinearModule(self.full_3.shape, 1, bias=self.params.bias, activation=nn.Sigmoid)
def forward(self, batch, **kwargs):
tensors = self.split(batch)
for idx, tensor in enumerate(tensors):
tensors[idx] = self.conv_dict[f"conv_1_{idx}"](tensor)
for idx, tensor in enumerate(tensors):
tensors[idx] = self.conv_dict[f"conv_2_{idx}"](tensor)
for idx, tensor in enumerate(tensors):
tensors[idx] = self.conv_dict[f"conv_3_{idx}"](tensor)
for idx, (tensor, convs) in enumerate(zip(tensors, self.band_list)):
for conv in convs:
tensor = conv(tensor)
tensors[idx] = tensor
tensor = self.merge(tensors)
tensor = self.full_1(tensor)

29
models/bandwise_conv_multihead_classifier.py
View File

@@ -22,24 +22,32 @@ class BandwiseConvMultiheadClassifier(BinaryMaskDatasetFunction,
batch_x, batch_y = batch_xy
y = self(batch_x)
y, bands_y = y.main_out, y.bands
bands_y_losses = [self.criterion(band_y, batch_y) for band_y in bands_y]
bands_y_losses = [self.bce_loss(band_y, batch_y) for band_y in bands_y]
return_dict = {f'band_{band_idx}_loss': band_y for band_idx, band_y in enumerate(bands_y_losses)}
overall_loss = self.criterion(y, batch_y)
combined_loss = overall_loss + torch.stack(bands_y_losses).sum()
return_dict.update(loss=combined_loss, overall_loss=overall_loss)
last_bce_loss = self.bce_loss(y, batch_y)
return_dict.update(last_bce_loss=last_bce_loss)
bands_y_losses.append(last_bce_loss)
combined_loss = torch.stack(bands_y_losses).mean()
return_dict.update(loss=combined_loss)
return return_dict
def validation_step(self, batch_xy, batch_idx, *args, **kwargs):
batch_x, batch_y = batch_xy
y = self(batch_x)
y, bands_y = y.main_out, y.bands
bands_y_losses = [self.criterion(band_y, batch_y) for band_y in bands_y]
bands_y_losses = [self.bce_loss(band_y, batch_y) for band_y in bands_y]
return_dict = {f'band_{band_idx}_val_loss': band_y for band_idx, band_y in enumerate(bands_y_losses)}
overall_loss = self.criterion(y, batch_y)
combined_loss = overall_loss + torch.stack(bands_y_losses).sum()
val_abs_loss = self.absolute_loss(y, batch_y)
return_dict.update(val_bce_loss=combined_loss, val_abs_loss=val_abs_loss,
last_bce_loss = self.bce_loss(y, batch_y)
return_dict.update(last_bce_loss=last_bce_loss)
bands_y_losses.append(last_bce_loss)
combined_loss = torch.stack(bands_y_losses).mean()
return_dict.update(val_bce_loss=combined_loss,
batch_idx=batch_idx, y=y, batch_y=batch_y
)
return return_dict
@@ -56,7 +64,6 @@ class BandwiseConvMultiheadClassifier(BinaryMaskDatasetFunction,
# Additional parameters
self.in_shape = self.dataset.train_dataset.sample_shape
self.conv_filters = self.params.filters
self.criterion = nn.BCELoss()
self.n_band_sections = 4
k = 3 # Base Kernel Value
@@ -69,7 +76,7 @@ class BandwiseConvMultiheadClassifier(BinaryMaskDatasetFunction,
last_shape = self.split.shape
conv_list = ModuleList()
for filters in self.conv_filters:
conv_list.append(ConvModule(last_shape, filters, (k, k*4), conv_stride=(1, 2),
conv_list.append(ConvModule(last_shape, filters, (k,k), conv_stride=(1, 1),
**self.params.module_kwargs))
last_shape = conv_list[-1].shape
# self.conv_list.append(ConvModule(last_shape, 1, 1, conv_stride=1, **self.params.module_kwargs))

12
models/conv_classifier.py
View File

@@ -29,23 +29,23 @@ class ConvClassifier(BinaryMaskDatasetFunction,
# Additional parameters
self.in_shape = self.dataset.train_dataset.sample_shape
self.conv_filters = self.params.filters
self.criterion = nn.BCELoss()
# Modules with Parameters
self.conv_list = ModuleList()
last_shape = self.in_shape
k = 3 # Base Kernel Value
for filters in self.conv_filters:
self.conv_list.append(ConvModule(last_shape, filters, (k, k*2), conv_stride=2, **self.params.module_kwargs))
self.conv_list.append(ConvModule(last_shape, filters, (k,k), conv_stride=(2, 2), conv_padding=2,
**self.params.module_kwargs))
last_shape = self.conv_list[-1].shape
# self.conv_list.append(ConvModule(last_shape, 1, 1, conv_stride=1, **self.params.module_kwargs))
# last_shape = self.conv_list[-1].shape
self.full_1 = LinearModule(self.flat.shape, self.params.lat_dim, **self.params.module_kwargs)
self.full_2 = LinearModule(self.full_1.out_features, self.full_1.out_features * 2, self.params.bias)
self.full_3 = LinearModule(self.full_2.out_features, self.full_2.out_features // 2, self.params.bias)
self.full_1 = LinearModule(self.conv_list[-1].shape, self.params.lat_dim, **self.params.module_kwargs)
self.full_2 = LinearModule(self.full_1.shape, self.full_1.shape * 2, **self.params.module_kwargs)
self.full_3 = LinearModule(self.full_2.shape, self.full_2.shape // 2, **self.params.module_kwargs)
self.full_out = LinearModule(self.full_3.out_features, 1, bias=self.params.bias, activation=nn.Sigmoid)
self.full_out = LinearModule(self.full_3.shape, 1, bias=self.params.bias, activation=nn.Sigmoid)
def forward(self, batch, **kwargs):
tensor = batch

64
models/residual_conv_classifier.py Normal file
View File

@@ -0,0 +1,64 @@
from argparse import Namespace
from torch import nn
from torch.nn import ModuleList
from ml_lib.modules.blocks import ConvModule, LinearModule, ResidualModule
from ml_lib.modules.utils import LightningBaseModule
from util.module_mixins import (BaseOptimizerMixin, BaseTrainMixin, BaseValMixin, BinaryMaskDatasetFunction,
BaseDataloadersMixin)
class ResidualConvClassifier(BinaryMaskDatasetFunction,
BaseDataloadersMixin,
BaseTrainMixin,
BaseValMixin,
BaseOptimizerMixin,
LightningBaseModule
):
def __init__(self, hparams):
super(ResidualConvClassifier, self).__init__(hparams)
# Dataset
# =============================================================================
self.dataset = self.build_dataset()
# Model Paramters
# =============================================================================
# Additional parameters
self.in_shape = self.dataset.train_dataset.sample_shape
self.conv_filters = self.params.filters
# Modules with Parameters
self.conv_list = ModuleList()
last_shape = self.in_shape
k = 3 # Base Kernel Value
conv_module_params = self.params.module_kwargs
conv_module_params.update(conv_kernel=(k, k), conv_stride=(1, 1), conv_padding=1)
self.conv_list.append(ConvModule(last_shape, self.conv_filters[0], (k, k), conv_stride=(2, 2), conv_padding=1,
**self.params.module_kwargs))
last_shape = self.conv_list[-1].shape
for filters in self.conv_filters:
conv_module_params.update(conv_filters=filters)
self.conv_list.append(ResidualModule(last_shape, ConvModule, 3, **conv_module_params))
last_shape = self.conv_list[-1].shape
self.conv_list.append(ConvModule(last_shape, filters, (k, k), conv_stride=(2, 2), conv_padding=2,
**self.params.module_kwargs))
last_shape = self.conv_list[-1].shape
self.full_1 = LinearModule(self.conv_list[-1].shape, self.params.lat_dim, **self.params.module_kwargs)
self.full_2 = LinearModule(self.full_1.shape, self.full_1.shape * 2, **self.params.module_kwargs)
self.full_3 = LinearModule(self.full_2.shape, self.full_2.shape // 2, **self.params.module_kwargs)
self.full_out = LinearModule(self.full_3.shape, 1, bias=self.params.bias, activation=nn.Sigmoid)
def forward(self, batch, **kwargs):
tensor = batch
for conv in self.conv_list:
tensor = conv(tensor)
tensor = self.full_1(tensor)
tensor = self.full_2(tensor)
tensor = self.full_3(tensor)
tensor = self.full_out(tensor)
return Namespace(main_out=tensor)