Skip to content

nn_condnet_regressor

NNCondNetRegressor

Bases: LightningModule

A LightningModule class for a conditional neural network model.

Attributes:

Name Type Description
l1 int

The number of neurons in the first hidden layer.

epochs int

The number of epochs to train the model for.

batch_size int

The batch size to use during training.

initialization str

The initialization method to use for the weights.

act_fn Module

The activation function to use in the hidden layers.

optimizer str

The optimizer to use during training.

dropout_prob float

The probability of dropping out a neuron during training.

lr_mult float

The learning rate multiplier for the optimizer.

patience int

The number of epochs to wait before early stopping.

batch_norm bool

Whether to use batch normalization or not.

_L_in int

The number of input features.

_L_out int

The number of output classes.

_L_cond int

The number of neurons in the conditional hidden layer.

_torchmetric str

The metric to use for the loss function. If None, then “mean_squared_error” is used.

layers Sequential

The neural network model.

Examples:

>>> from torch.utils.data import DataLoader
    from spotpython.light.regression import NNLinearRegressor
    from torch import nn
    import lightning as L
    import torch
    from torch.utils.data import TensorDataset
    PATH_DATASETS = './data'
    BATCH_SIZE = 128
    # generate data
    num_samples = 1_000
    input_dim = 10
    X = torch.randn(num_samples, input_dim)  # random data for example
    Y = torch.randn(num_samples, 1)  # random target for example
    data_set = TensorDataset(X, Y)
    train_loader = DataLoader(dataset=data_set, batch_size=BATCH_SIZE)
    test_loader = DataLoader(dataset=data_set, batch_size=BATCH_SIZE)
    val_loader = DataLoader(dataset=data_set, batch_size=BATCH_SIZE)
    batch_x, batch_y = next(iter(train_loader))
    print(batch_x.shape)
    print(batch_y.shape)
    net_light_base = NNLinearRegressor(l1=128,
                                    batch_norm=True,
                                        epochs=10,
                                        batch_size=BATCH_SIZE,
                                        initialization='xavier',
                                        act_fn=nn.ReLU(),
                                        optimizer='Adam',
                                        dropout_prob=0.1,
                                        lr_mult=0.1,
                                        patience=5,
                                        _L_in=input_dim,
                                        _L_out=1,
                                        _torchmetric="mean_squared_error",)
    trainer = L.Trainer(max_epochs=2,  enable_progress_bar=True)
    trainer.fit(net_light_base, train_loader)
    # validation and test should give the same result, because the data is the same
    trainer.validate(net_light_base, val_loader)
    trainer.test(net_light_base, test_loader)
        GPU available: True (mps), used: True
        TPU available: False, using: 0 TPU cores
        HPU available: False, using: 0 HPUs
    | Name   | Type       | Params | Mode  | In sizes  | Out sizes
    ----------------------------------------------------------------------
    0 | layers | Sequential | 20.8 K | train | [128, 10] | [128, 1]
    ----------------------------------------------------------------------
    20.8 K    Trainable params
    0         Non-trainable params
    20.8 K    Total params
    0.083     Total estimated model params size (MB)
    69        Modules in train mode
    0         Modules in eval mode
    torch.Size([128, 10])
    torch.Size([128, 1])
    ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━┓
    ┃      Validate metric      ┃       DataLoader 0        ┃
    ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━┩
    │         hp_metric         │     81.1978988647461      │
    │         val_loss          │     81.1978988647461      │
    └───────────────────────────┴───────────────────────────┘
    ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━┓
    ┃        Test metric        ┃       DataLoader 0        ┃
    ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━┩
    │         hp_metric         │     81.1978988647461      │
    │         val_loss          │     81.1978988647461      │
    └───────────────────────────┴───────────────────────────┘
    [{'val_loss': 81.1978988647461, 'hp_metric': 81.1978988647461}]
Source code in spotpython/light/regression/nn_condnet_regressor.py
 26
 27
 28
 29
 30
 31
 32
 33
 34
 35
 36
 37
 38
 39
 40
 41
 42
 43
 44
 45
 46
 47
 48
 49
 50
 51
 52
 53
 54
 55
 56
 57
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
class NNCondNetRegressor(L.LightningModule):
    """
    A LightningModule class for a conditional neural network model.

    Attributes:
        l1 (int):
            The number of neurons in the first hidden layer.
        epochs (int):
            The number of epochs to train the model for.
        batch_size (int):
            The batch size to use during training.
        initialization (str):
            The initialization method to use for the weights.
        act_fn (nn.Module):
            The activation function to use in the hidden layers.
        optimizer (str):
            The optimizer to use during training.
        dropout_prob (float):
            The probability of dropping out a neuron during training.
        lr_mult (float):
            The learning rate multiplier for the optimizer.
        patience (int):
            The number of epochs to wait before early stopping.
        batch_norm (bool):
            Whether to use batch normalization or not.
        _L_in (int):
            The number of input features.
        _L_out (int):
            The number of output classes.
        _L_cond (int):
            The number of neurons in the conditional hidden layer.
        _torchmetric (str):
            The metric to use for the loss function. If `None`,
            then "mean_squared_error" is used.
        layers (nn.Sequential):
            The neural network model.

    Examples:
        >>> from torch.utils.data import DataLoader
            from spotpython.light.regression import NNLinearRegressor
            from torch import nn
            import lightning as L
            import torch
            from torch.utils.data import TensorDataset
            PATH_DATASETS = './data'
            BATCH_SIZE = 128
            # generate data
            num_samples = 1_000
            input_dim = 10
            X = torch.randn(num_samples, input_dim)  # random data for example
            Y = torch.randn(num_samples, 1)  # random target for example
            data_set = TensorDataset(X, Y)
            train_loader = DataLoader(dataset=data_set, batch_size=BATCH_SIZE)
            test_loader = DataLoader(dataset=data_set, batch_size=BATCH_SIZE)
            val_loader = DataLoader(dataset=data_set, batch_size=BATCH_SIZE)
            batch_x, batch_y = next(iter(train_loader))
            print(batch_x.shape)
            print(batch_y.shape)
            net_light_base = NNLinearRegressor(l1=128,
                                            batch_norm=True,
                                                epochs=10,
                                                batch_size=BATCH_SIZE,
                                                initialization='xavier',
                                                act_fn=nn.ReLU(),
                                                optimizer='Adam',
                                                dropout_prob=0.1,
                                                lr_mult=0.1,
                                                patience=5,
                                                _L_in=input_dim,
                                                _L_out=1,
                                                _torchmetric="mean_squared_error",)
            trainer = L.Trainer(max_epochs=2,  enable_progress_bar=True)
            trainer.fit(net_light_base, train_loader)
            # validation and test should give the same result, because the data is the same
            trainer.validate(net_light_base, val_loader)
            trainer.test(net_light_base, test_loader)
                GPU available: True (mps), used: True
                TPU available: False, using: 0 TPU cores
                HPU available: False, using: 0 HPUs

                | Name   | Type       | Params | Mode  | In sizes  | Out sizes
                ----------------------------------------------------------------------
                0 | layers | Sequential | 20.8 K | train | [128, 10] | [128, 1]
                ----------------------------------------------------------------------
                20.8 K    Trainable params
                0         Non-trainable params
                20.8 K    Total params
                0.083     Total estimated model params size (MB)
                69        Modules in train mode
                0         Modules in eval mode
                torch.Size([128, 10])
                torch.Size([128, 1])
                ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━┓
                ┃      Validate metric      ┃       DataLoader 0        ┃
                ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━┩
                │         hp_metric         │     81.1978988647461      │
                │         val_loss          │     81.1978988647461      │
                └───────────────────────────┴───────────────────────────┘
                ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━┓
                ┃        Test metric        ┃       DataLoader 0        ┃
                ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━┩
                │         hp_metric         │     81.1978988647461      │
                │         val_loss          │     81.1978988647461      │
                └───────────────────────────┴───────────────────────────┘
                [{'val_loss': 81.1978988647461, 'hp_metric': 81.1978988647461}]
    """

    def __init__(
        self,
        l1: int,
        epochs: int,
        batch_size: int,
        initialization: str,
        act_fn: nn.Module,
        optimizer: str,
        dropout_prob: float,
        lr_mult: float,
        patience: int,
        batch_norm: bool,
        _L_cond: int,
        _L_in: int,
        _L_out: int,
        _torchmetric: str,
        *args,
        **kwargs,
    ):
        """
        Initializes the NNLinearRegressor object.

        Args:
            l1 (int):
                The number of neurons in the first hidden layer.
            epochs (int):
                The number of epochs to train the model for.
            batch_size (int):
                The batch size to use during training.
            initialization (str):
                The initialization method to use for the weights.
            act_fn (nn.Module):
                The activation function to use in the hidden layers.
            optimizer (str):
                The optimizer to use during training.
            dropout_prob (float):
                The probability of dropping out a neuron during training.
            lr_mult (float):
                The learning rate multiplier for the optimizer.
            patience (int):
                The number of epochs to wait before early stopping.
            batch_norm (bool):
                Whether to use batch normalization or not.
            _L_cond (int):
                The number of neurons in the conditional hidden layer.
            _L_in (int):
                The number of input features. Not a hyperparameter, but needed to create the network.
            _L_out (int):
                The number of output classes. Not a hyperparameter, but needed to create the network.
            _torchmetric (str):
                The metric to use for the loss function. If `None`,
                then "mean_squared_error" is used.

        Returns:
            (NoneType): None

        Raises:
            ValueError: If l1 is less than 4.

        """
        super().__init__()
        # Attribute 'act_fn' is an instance of `nn.Module` and is already saved during
        # checkpointing. It is recommended to ignore them
        # using `self.save_hyperparameters(ignore=['act_fn'])`
        # self.save_hyperparameters(ignore=["act_fn"])
        #
        self._L_cond = _L_cond
        self._L_in = _L_in
        self._L_out = _L_out
        if _torchmetric is None:
            _torchmetric = "mean_squared_error"
        self._torchmetric = _torchmetric
        self.metric = getattr(torchmetrics.functional.regression, _torchmetric)
        # _L_cond, _L_in and _L_out are not hyperparameters, but are needed to create the network
        # _torchmetric is not a hyperparameter, but is needed to calculate the loss
        self.save_hyperparameters(ignore=["_L_cond", "_L_in", "_L_out", "_torchmetric"])
        # set dummy input array for Tensorboard Graphs
        # set log_graph=True in Trainer to see the graph (in traintest.py)
        self.example_input_array = torch.zeros((batch_size, self._L_cond + self._L_in))
        if self.hparams.l1 < 4:
            raise ValueError("l1 must be at least 4")
        hidden_sizes = self._get_hidden_sizes()

        # Conditional Layer
        self.cond_layer = ConditionalLayer(self._L_in, self._L_cond, self.hparams.l1)

        if batch_norm:
            # Add batch normalization layers
            layers = []
            layer_sizes = [self.hparams.l1] + hidden_sizes
            for i in range(len(layer_sizes) - 1):
                current_layer_size = layer_sizes[i]
                next_layer_size = layer_sizes[i + 1]
                layers += [
                    nn.Linear(current_layer_size, next_layer_size),
                    nn.BatchNorm1d(next_layer_size),
                    self.hparams.act_fn,
                    nn.Dropout(self.hparams.dropout_prob),
                ]
            layers += [nn.Linear(layer_sizes[-1], self._L_out)]
        else:
            layers = []
            layer_sizes = [self.hparams.l1] + hidden_sizes
            for i in range(len(layer_sizes) - 1):
                current_layer_size = layer_sizes[i]
                next_layer_size = layer_sizes[i + 1]
                layers += [
                    nn.Linear(current_layer_size, next_layer_size),
                    self.hparams.act_fn,
                    nn.Dropout(self.hparams.dropout_prob),
                ]
            layers += [nn.Linear(layer_sizes[-1], self._L_out)]

        # Wrap the layers into a sequential container
        self.layers = nn.Sequential(*layers)

        # Initialization (Xavier, Kaiming, or Default)
        self.apply(self._init_weights)

    def _init_weights(self, module):
        if isinstance(module, nn.Linear):
            if self.hparams.initialization == "xavier_uniform":
                nn.init.xavier_uniform_(module.weight)
            elif self.hparams.initialization == "xavier_normal":
                nn.init.xavier_normal_(module.weight)
            elif self.hparams.initialization == "kaiming_uniform":
                nn.init.kaiming_uniform_(module.weight)
            elif self.hparams.initialization == "kaiming_normal":
                nn.init.kaiming_normal_(module.weight)
            else:  # "Default"
                nn.init.uniform_(module.weight)
            if module.bias is not None:
                nn.init.zeros_(module.bias)

    def _generate_div2_list(self, n, n_min) -> list:
        result = []
        current = n
        repeats = 1
        max_repeats = 4
        while current >= n_min:
            result.extend([current] * min(repeats, max_repeats))
            current = current // 2
            repeats = repeats + 1
        return result

    def _get_hidden_sizes(self):
        n_low = self._L_in // 4
        n_high = max(self.hparams.l1, 2 * n_low)
        hidden_sizes = self._generate_div2_list(n_high, n_low)
        return hidden_sizes

    def forward(self, x: torch.Tensor) -> torch.Tensor:
        """
        Performs a forward pass through the model.

        Args:
            x (torch.Tensor): A tensor containing a batch of input data.

        Returns:
            torch.Tensor: A tensor containing the output of the model.

        """
        # exxtract the condition from the input
        condition = x[:, : self._L_cond]
        x_1 = x[:, self._L_cond :]
        x_1 = self.cond_layer(x_1, condition)
        x_1 = self.layers(x_1)
        return x_1

    def _calculate_loss(self, batch):
        """
        Calculate the loss for the given batch.

        Args:
            batch (tuple): A tuple containing a batch of input data and labels.
            mode (str, optional): The mode of the model. Defaults to "train".

        Returns:
            torch.Tensor: A tensor containing the loss for this batch.

        """
        x, y = batch
        y = y.view(len(y), 1)
        y_hat = self(x)
        loss = self.metric(y_hat, y)
        return loss

    def training_step(self, batch: tuple) -> torch.Tensor:
        """
        Performs a single training step.

        Args:
            batch (tuple): A tuple containing a batch of input data and labels.

        Returns:
            torch.Tensor: A tensor containing the loss for this batch.

        """
        val_loss = self._calculate_loss(batch)
        # self.log("train_loss", val_loss, on_step=True, on_epoch=True, prog_bar=True)
        # self.log("train_mae_loss", mae_loss, on_step=True, on_epoch=True, prog_bar=True)
        return val_loss

    def validation_step(self, batch: tuple, batch_idx: int, prog_bar: bool = False) -> torch.Tensor:
        """
        Performs a single validation step.

        Args:
            batch (tuple): A tuple containing a batch of input data and labels.
            batch_idx (int): The index of the current batch.
            prog_bar (bool, optional): Whether to display the progress bar. Defaults to False.

        Returns:
            torch.Tensor: A tensor containing the loss for this batch.

        """
        val_loss = self._calculate_loss(batch)
        # self.log("val_loss", val_loss, on_step=False, on_epoch=True, prog_bar=prog_bar)
        self.log("val_loss", val_loss, prog_bar=prog_bar)
        self.log("hp_metric", val_loss, prog_bar=prog_bar)
        return val_loss

    def test_step(self, batch: tuple, batch_idx: int, prog_bar: bool = False) -> torch.Tensor:
        """
        Performs a single test step.

        Args:
            batch (tuple): A tuple containing a batch of input data and labels.
            batch_idx (int): The index of the current batch.
            prog_bar (bool, optional): Whether to display the progress bar. Defaults to False.

        Returns:
            torch.Tensor: A tensor containing the loss for this batch.
        """
        val_loss = self._calculate_loss(batch)
        self.log("val_loss", val_loss, prog_bar=prog_bar)
        self.log("hp_metric", val_loss, prog_bar=prog_bar)
        return val_loss

    def predict_step(self, batch: tuple, batch_idx: int, prog_bar: bool = False) -> torch.Tensor:
        """
        Performs a single prediction step.

        Args:
            batch (tuple): A tuple containing a batch of input data and labels.
            batch_idx (int): The index of the current batch.
            prog_bar (bool, optional): Whether to display the progress bar. Defaults to False.

        Returns:
            torch.Tensor: A tensor containing the prediction for this batch.
        """
        x, y = batch
        yhat = self(x)
        y = y.view(len(y), 1)
        yhat = yhat.view(len(yhat), 1)
        print(f"Predict step x: {x}")
        print(f"Predict step y: {y}")
        print(f"Predict step y_hat: {yhat}")
        # pred_loss = F.mse_loss(y_hat, y)
        # pred loss not registered
        # self.log("pred_loss", pred_loss, prog_bar=prog_bar)
        # self.log("hp_metric", pred_loss, prog_bar=prog_bar)
        # MisconfigurationException: You are trying to `self.log()`
        # but the loop's result collection is not registered yet.
        # This is most likely because you are trying to log in a `predict` hook, but it doesn't support logging.
        # If you want to manually log, please consider using `self.log_dict({'pred_loss': pred_loss})` instead.
        return (x, y, yhat)

    def configure_optimizers(self) -> torch.optim.Optimizer:
        """
        Configures the optimizer for the model.

        Notes:
            The default Lightning way is to define an optimizer as
            `optimizer = torch.optim.Adam(self.parameters(), lr=self.learning_rate)`.
            spotpython uses an optimizer handler to create the optimizer, which
            adapts the learning rate according to the lr_mult hyperparameter as
            well as other hyperparameters. See `spotpython.hyperparameters.optimizer.py` for details.

        Returns:
            torch.optim.Optimizer: The optimizer to use during training.

        """
        # optimizer = torch.optim.Adam(self.parameters(), lr=self.learning_rate)
        optimizer = optimizer_handler(
            optimizer_name=self.hparams.optimizer, params=self.parameters(), lr_mult=self.hparams.lr_mult
        )

        num_milestones = 3  # Number of milestones to divide the epochs
        milestones = [int(self.hparams.epochs / (num_milestones + 1) * (i + 1)) for i in range(num_milestones)]
        scheduler = optim.lr_scheduler.MultiStepLR(optimizer, milestones=milestones, gamma=0.1)  # Decay factor

        lr_scheduler_config = {
            "scheduler": scheduler,
            "interval": "epoch",
            "frequency": 1,
        }

        return {"optimizer": optimizer, "lr_scheduler": lr_scheduler_config}

__init__(l1, epochs, batch_size, initialization, act_fn, optimizer, dropout_prob, lr_mult, patience, batch_norm, _L_cond, _L_in, _L_out, _torchmetric, *args, **kwargs)

Initializes the NNLinearRegressor object.

Parameters:

Name Type Description Default
l1 int

The number of neurons in the first hidden layer.

required
epochs int

The number of epochs to train the model for.

required
batch_size int

The batch size to use during training.

required
initialization str

The initialization method to use for the weights.

required
act_fn Module

The activation function to use in the hidden layers.

required
optimizer str

The optimizer to use during training.

required
dropout_prob float

The probability of dropping out a neuron during training.

required
lr_mult float

The learning rate multiplier for the optimizer.

required
patience int

The number of epochs to wait before early stopping.

required
batch_norm bool

Whether to use batch normalization or not.

required
_L_cond int

The number of neurons in the conditional hidden layer.

required
_L_in int

The number of input features. Not a hyperparameter, but needed to create the network.

required
_L_out int

The number of output classes. Not a hyperparameter, but needed to create the network.

required
_torchmetric str

The metric to use for the loss function. If None, then “mean_squared_error” is used.

required

Returns:

Type Description
NoneType

None

Raises:

Type Description
ValueError

If l1 is less than 4.

Source code in spotpython/light/regression/nn_condnet_regressor.py
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
def __init__(
    self,
    l1: int,
    epochs: int,
    batch_size: int,
    initialization: str,
    act_fn: nn.Module,
    optimizer: str,
    dropout_prob: float,
    lr_mult: float,
    patience: int,
    batch_norm: bool,
    _L_cond: int,
    _L_in: int,
    _L_out: int,
    _torchmetric: str,
    *args,
    **kwargs,
):
    """
    Initializes the NNLinearRegressor object.

    Args:
        l1 (int):
            The number of neurons in the first hidden layer.
        epochs (int):
            The number of epochs to train the model for.
        batch_size (int):
            The batch size to use during training.
        initialization (str):
            The initialization method to use for the weights.
        act_fn (nn.Module):
            The activation function to use in the hidden layers.
        optimizer (str):
            The optimizer to use during training.
        dropout_prob (float):
            The probability of dropping out a neuron during training.
        lr_mult (float):
            The learning rate multiplier for the optimizer.
        patience (int):
            The number of epochs to wait before early stopping.
        batch_norm (bool):
            Whether to use batch normalization or not.
        _L_cond (int):
            The number of neurons in the conditional hidden layer.
        _L_in (int):
            The number of input features. Not a hyperparameter, but needed to create the network.
        _L_out (int):
            The number of output classes. Not a hyperparameter, but needed to create the network.
        _torchmetric (str):
            The metric to use for the loss function. If `None`,
            then "mean_squared_error" is used.

    Returns:
        (NoneType): None

    Raises:
        ValueError: If l1 is less than 4.

    """
    super().__init__()
    # Attribute 'act_fn' is an instance of `nn.Module` and is already saved during
    # checkpointing. It is recommended to ignore them
    # using `self.save_hyperparameters(ignore=['act_fn'])`
    # self.save_hyperparameters(ignore=["act_fn"])
    #
    self._L_cond = _L_cond
    self._L_in = _L_in
    self._L_out = _L_out
    if _torchmetric is None:
        _torchmetric = "mean_squared_error"
    self._torchmetric = _torchmetric
    self.metric = getattr(torchmetrics.functional.regression, _torchmetric)
    # _L_cond, _L_in and _L_out are not hyperparameters, but are needed to create the network
    # _torchmetric is not a hyperparameter, but is needed to calculate the loss
    self.save_hyperparameters(ignore=["_L_cond", "_L_in", "_L_out", "_torchmetric"])
    # set dummy input array for Tensorboard Graphs
    # set log_graph=True in Trainer to see the graph (in traintest.py)
    self.example_input_array = torch.zeros((batch_size, self._L_cond + self._L_in))
    if self.hparams.l1 < 4:
        raise ValueError("l1 must be at least 4")
    hidden_sizes = self._get_hidden_sizes()

    # Conditional Layer
    self.cond_layer = ConditionalLayer(self._L_in, self._L_cond, self.hparams.l1)

    if batch_norm:
        # Add batch normalization layers
        layers = []
        layer_sizes = [self.hparams.l1] + hidden_sizes
        for i in range(len(layer_sizes) - 1):
            current_layer_size = layer_sizes[i]
            next_layer_size = layer_sizes[i + 1]
            layers += [
                nn.Linear(current_layer_size, next_layer_size),
                nn.BatchNorm1d(next_layer_size),
                self.hparams.act_fn,
                nn.Dropout(self.hparams.dropout_prob),
            ]
        layers += [nn.Linear(layer_sizes[-1], self._L_out)]
    else:
        layers = []
        layer_sizes = [self.hparams.l1] + hidden_sizes
        for i in range(len(layer_sizes) - 1):
            current_layer_size = layer_sizes[i]
            next_layer_size = layer_sizes[i + 1]
            layers += [
                nn.Linear(current_layer_size, next_layer_size),
                self.hparams.act_fn,
                nn.Dropout(self.hparams.dropout_prob),
            ]
        layers += [nn.Linear(layer_sizes[-1], self._L_out)]

    # Wrap the layers into a sequential container
    self.layers = nn.Sequential(*layers)

    # Initialization (Xavier, Kaiming, or Default)
    self.apply(self._init_weights)

configure_optimizers()

Configures the optimizer for the model.

Notes

The default Lightning way is to define an optimizer as optimizer = torch.optim.Adam(self.parameters(), lr=self.learning_rate). spotpython uses an optimizer handler to create the optimizer, which adapts the learning rate according to the lr_mult hyperparameter as well as other hyperparameters. See spotpython.hyperparameters.optimizer.py for details.

Returns:

Type Description
Optimizer

torch.optim.Optimizer: The optimizer to use during training.

Source code in spotpython/light/regression/nn_condnet_regressor.py
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
def configure_optimizers(self) -> torch.optim.Optimizer:
    """
    Configures the optimizer for the model.

    Notes:
        The default Lightning way is to define an optimizer as
        `optimizer = torch.optim.Adam(self.parameters(), lr=self.learning_rate)`.
        spotpython uses an optimizer handler to create the optimizer, which
        adapts the learning rate according to the lr_mult hyperparameter as
        well as other hyperparameters. See `spotpython.hyperparameters.optimizer.py` for details.

    Returns:
        torch.optim.Optimizer: The optimizer to use during training.

    """
    # optimizer = torch.optim.Adam(self.parameters(), lr=self.learning_rate)
    optimizer = optimizer_handler(
        optimizer_name=self.hparams.optimizer, params=self.parameters(), lr_mult=self.hparams.lr_mult
    )

    num_milestones = 3  # Number of milestones to divide the epochs
    milestones = [int(self.hparams.epochs / (num_milestones + 1) * (i + 1)) for i in range(num_milestones)]
    scheduler = optim.lr_scheduler.MultiStepLR(optimizer, milestones=milestones, gamma=0.1)  # Decay factor

    lr_scheduler_config = {
        "scheduler": scheduler,
        "interval": "epoch",
        "frequency": 1,
    }

    return {"optimizer": optimizer, "lr_scheduler": lr_scheduler_config}

forward(x)

Performs a forward pass through the model.

Parameters:

Name Type Description Default
x Tensor

A tensor containing a batch of input data.

required

Returns:

Type Description
Tensor

torch.Tensor: A tensor containing the output of the model.

Source code in spotpython/light/regression/nn_condnet_regressor.py
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
def forward(self, x: torch.Tensor) -> torch.Tensor:
    """
    Performs a forward pass through the model.

    Args:
        x (torch.Tensor): A tensor containing a batch of input data.

    Returns:
        torch.Tensor: A tensor containing the output of the model.

    """
    # exxtract the condition from the input
    condition = x[:, : self._L_cond]
    x_1 = x[:, self._L_cond :]
    x_1 = self.cond_layer(x_1, condition)
    x_1 = self.layers(x_1)
    return x_1

predict_step(batch, batch_idx, prog_bar=False)

Performs a single prediction step.

Parameters:

Name Type Description Default
batch tuple

A tuple containing a batch of input data and labels.

required
batch_idx int

The index of the current batch.

required
prog_bar bool

Whether to display the progress bar. Defaults to False.

False

Returns:

Type Description
Tensor

torch.Tensor: A tensor containing the prediction for this batch.

Source code in spotpython/light/regression/nn_condnet_regressor.py
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
def predict_step(self, batch: tuple, batch_idx: int, prog_bar: bool = False) -> torch.Tensor:
    """
    Performs a single prediction step.

    Args:
        batch (tuple): A tuple containing a batch of input data and labels.
        batch_idx (int): The index of the current batch.
        prog_bar (bool, optional): Whether to display the progress bar. Defaults to False.

    Returns:
        torch.Tensor: A tensor containing the prediction for this batch.
    """
    x, y = batch
    yhat = self(x)
    y = y.view(len(y), 1)
    yhat = yhat.view(len(yhat), 1)
    print(f"Predict step x: {x}")
    print(f"Predict step y: {y}")
    print(f"Predict step y_hat: {yhat}")
    # pred_loss = F.mse_loss(y_hat, y)
    # pred loss not registered
    # self.log("pred_loss", pred_loss, prog_bar=prog_bar)
    # self.log("hp_metric", pred_loss, prog_bar=prog_bar)
    # MisconfigurationException: You are trying to `self.log()`
    # but the loop's result collection is not registered yet.
    # This is most likely because you are trying to log in a `predict` hook, but it doesn't support logging.
    # If you want to manually log, please consider using `self.log_dict({'pred_loss': pred_loss})` instead.
    return (x, y, yhat)

test_step(batch, batch_idx, prog_bar=False)

Performs a single test step.

Parameters:

Name Type Description Default
batch tuple

A tuple containing a batch of input data and labels.

required
batch_idx int

The index of the current batch.

required
prog_bar bool

Whether to display the progress bar. Defaults to False.

False

Returns:

Type Description
Tensor

torch.Tensor: A tensor containing the loss for this batch.

Source code in spotpython/light/regression/nn_condnet_regressor.py
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
def test_step(self, batch: tuple, batch_idx: int, prog_bar: bool = False) -> torch.Tensor:
    """
    Performs a single test step.

    Args:
        batch (tuple): A tuple containing a batch of input data and labels.
        batch_idx (int): The index of the current batch.
        prog_bar (bool, optional): Whether to display the progress bar. Defaults to False.

    Returns:
        torch.Tensor: A tensor containing the loss for this batch.
    """
    val_loss = self._calculate_loss(batch)
    self.log("val_loss", val_loss, prog_bar=prog_bar)
    self.log("hp_metric", val_loss, prog_bar=prog_bar)
    return val_loss

training_step(batch)

Performs a single training step.

Parameters:

Name Type Description Default
batch tuple

A tuple containing a batch of input data and labels.

required

Returns:

Type Description
Tensor

torch.Tensor: A tensor containing the loss for this batch.

Source code in spotpython/light/regression/nn_condnet_regressor.py
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
def training_step(self, batch: tuple) -> torch.Tensor:
    """
    Performs a single training step.

    Args:
        batch (tuple): A tuple containing a batch of input data and labels.

    Returns:
        torch.Tensor: A tensor containing the loss for this batch.

    """
    val_loss = self._calculate_loss(batch)
    # self.log("train_loss", val_loss, on_step=True, on_epoch=True, prog_bar=True)
    # self.log("train_mae_loss", mae_loss, on_step=True, on_epoch=True, prog_bar=True)
    return val_loss

validation_step(batch, batch_idx, prog_bar=False)

Performs a single validation step.

Parameters:

Name Type Description Default
batch tuple

A tuple containing a batch of input data and labels.

required
batch_idx int

The index of the current batch.

required
prog_bar bool

Whether to display the progress bar. Defaults to False.

False

Returns:

Type Description
Tensor

torch.Tensor: A tensor containing the loss for this batch.

Source code in spotpython/light/regression/nn_condnet_regressor.py
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
def validation_step(self, batch: tuple, batch_idx: int, prog_bar: bool = False) -> torch.Tensor:
    """
    Performs a single validation step.

    Args:
        batch (tuple): A tuple containing a batch of input data and labels.
        batch_idx (int): The index of the current batch.
        prog_bar (bool, optional): Whether to display the progress bar. Defaults to False.

    Returns:
        torch.Tensor: A tensor containing the loss for this batch.

    """
    val_loss = self._calculate_loss(batch)
    # self.log("val_loss", val_loss, on_step=False, on_epoch=True, prog_bar=prog_bar)
    self.log("val_loss", val_loss, prog_bar=prog_bar)
    self.log("hp_metric", val_loss, prog_bar=prog_bar)
    return val_loss