minerva.models.ssl.diet

Classes

DIET	Base class for all neural network modules.

Module Contents

class minerva.models.ssl.diet.DIET(backbone, linear_head=None, num_data=None, flatten=True, adapter=None, loss=None, learning_rate=0.0003, weight_decay=0.0003, wca_scheduler_total_epochs=None)

Bases: lightning.LightningModule

Base class for all neural network modules.

Your models should also subclass this class.

Modules can also contain other Modules, allowing them to be nested in a tree structure. You can assign the submodules as regular attributes:

import torch.nn as nn
import torch.nn.functional as F


class Model(nn.Module):
    def __init__(self) -> None:
        super().__init__()
        self.conv1 = nn.Conv2d(1, 20, 5)
        self.conv2 = nn.Conv2d(20, 20, 5)

    def forward(self, x):
        x = F.relu(self.conv1(x))
        return F.relu(self.conv2(x))

Submodules assigned in this way will be registered, and will also have their parameters converted when you call to(), etc.

Note

As per the example above, an __init__() call to the parent class must be made before assignment on the child.

Variables:

training (bool) – Boolean represents whether this module is in training or evaluation mode.

Parameters:

• backbone (torch.nn.Module)

• linear_head (Optional[torch.nn.Module])

• num_data (Optional[int])

• flatten (bool)

• adapter (Optional[Callable[[torch.Tensor], torch.Tensor]])

• loss (Callable)

• learning_rate (float)

• weight_decay (float)

• wca_scheduler_total_epochs (Optional[int])

DIET model.

Parameters

backbonetorch.nn.Module

Backbone model.

linear_head: torch.nn.Module, optional

Linear head that computes logits from embeddings of the data input, by default None. If None, the linear head is automatically defined before training. The lengths of both training dataset and linear head output must match.

num_dataint, optional

Total number of samples in the training dataset, by default None. If None, the length of the training dataset is computed before the training in the setup() function.

flattenbool

If True, the output of the backbone is flattened before the linear layer, by default True.

adapterOptional[Callable[[torch.Tensor], torch.Tensor]], optional

If not None, an adapter is added after the backbone and before the flatten process, by default None.

lossCallable

Loss function, by default CrossEntropyLoss with label smoothing 0.8.

learning_ratefloat, optional

Learning rate used in the optimizer, by default 3e-4.

weight_decayfloat, optional

Weight decay used in the optimizer, by default 3e-4.

wca_scheduler_total_epochsint, optional

Total number of epochs for the WarmupCosineAnnealing scheduler, by default None. Must be None or an integer greater than 10. If None, no scheduler is used.

adapter = None

backbone

configure_optimizers()

Choose what optimizers and learning-rate schedulers to use in your optimization. Normally you’d need one. But in the case of GANs or similar you might have multiple. Optimization with multiple optimizers only works in the manual optimization mode.

Return:

Any of these 6 options.

• Single optimizer.

• List or Tuple of optimizers.

• Two lists - The first list has multiple optimizers, and the second has multiple LR schedulers (or multiple lr_scheduler_config).

• Dictionary, with an "optimizer" key, and (optionally) a "lr_scheduler" key whose value is a single LR scheduler or lr_scheduler_config.

• None - Fit will run without any optimizer.

The lr_scheduler_config is a dictionary which contains the scheduler and its associated configuration. The default configuration is shown below.

lr_scheduler_config = {
    # REQUIRED: The scheduler instance
    "scheduler": lr_scheduler,
    # The unit of the scheduler's step size, could also be 'step'.
    # 'epoch' updates the scheduler on epoch end whereas 'step'
    # updates it after a optimizer update.
    "interval": "epoch",
    # How many epochs/steps should pass between calls to
    # `scheduler.step()`. 1 corresponds to updating the learning
    # rate after every epoch/step.
    "frequency": 1,
    # Metric to monitor for schedulers like `ReduceLROnPlateau`
    "monitor": "val_loss",
    # If set to `True`, will enforce that the value specified 'monitor'
    # is available when the scheduler is updated, thus stopping
    # training if not found. If set to `False`, it will only produce a warning
    "strict": True,
    # If using the `LearningRateMonitor` callback to monitor the
    # learning rate progress, this keyword can be used to specify
    # a custom logged name
    "name": None,
}

When there are schedulers in which the .step() method is conditioned on a value, such as the torch.optim.lr_scheduler.ReduceLROnPlateau scheduler, Lightning requires that the lr_scheduler_config contains the keyword "monitor" set to the metric name that the scheduler should be conditioned on.

Metrics can be made available to monitor by simply logging it using self.log('metric_to_track', metric_val) in your LightningModule.

Note:

Some things to know:

• Lightning calls .backward() and .step() automatically in case of automatic optimization.

• If a learning rate scheduler is specified in configure_optimizers() with key "interval" (default “epoch”) in the scheduler configuration, Lightning will call the scheduler’s .step() method automatically in case of automatic optimization.

• If you use 16-bit precision (precision=16), Lightning will automatically handle the optimizer.

• If you use torch.optim.LBFGS, Lightning handles the closure function automatically for you.

• If you use multiple optimizers, you will have to switch to ‘manual optimization’ mode and step them yourself.

• If you need to control how often the optimizer steps, override the optimizer_step() hook.

flatten = True

forward(x)

Same as torch.nn.Module.forward().

Args:

*args: Whatever you decide to pass into the forward method. **kwargs: Keyword arguments are also possible.

Return:

Your model’s output

learning_rate = 0.0003

linear_head = None

loss

num_data = None

setup(stage)

Setup function. If the model lacks a linear head, this function computes the length of the training dataset, the encoding size, and creates a linear head accordingly. Also checks whether the linear head output matches the length of the training dataset, raising an error in case of mismatch.

training_step(batch, batch_idx)

A simple training step.

wca_scheduler_total_epochs = None

weight_decay = 0.0003