Expand description
A small neural-network library based on Torch.
This library tries to stay as close as possible to the original Python and C++ implementations.
Re-exports§
Modules§
- Variable initialization.
Structs§
- Parameters for the Adam optimizer.
- Parameters for the AdamW optimizer.
- A batch-normalization layer.
- Batch-normalization config.
- A N-dimensional convolution layer.
- Generic convolution config.
- A generic transposed convolution configuration.
- A generic transposed convolution layer.
- An embedding layer.
- Configuration option for an embedding layer.
- A layer defined by a simple closure.
- A layer defined by a closure with an additional training parameter.
- A Gated Recurrent Unit (GRU) layer.
- A GRU state, this contains a single tensor.
- A group-normalization layer.
- Group-normalization config.
- An identity layer. This just propagates its tensor input as output.
- A Long Short-Term Memory (LSTM) layer.
- The state for a LSTM network, this contains two tensors.
- A layer-normalization layer.
- Layer-normalization config.
- A linear fully-connected layer.
- Configuration for a linear layer.
- An optimizer to run gradient descent.
- A variable store with an associated path for variables naming.
- Configuration for the GRU and LSTM layers.
- Parameters for the RmsProp optimizer.
- A sequential layer combining multiple other layers.
- A sequential layer combining new layers with support for a training mode.
- Parameters for the SGD optimizer.
- A VarStore is used to store variables used by one or multiple layers. It specifies a single device where all variables are stored.
Enums§
- How padding is performed by convolution operations on the edge of the input tensor.
Traits§
- The simplest module trait, defining a forward function.
- Module trait with an additional train parameter.
- Optimizer configurations. These configs can be used to build optimizer.
- Trait for Recurrent Neural Networks.
Functions§
- Creates the configuration for the Adam optimizer.
- Creates the configuration for the AdamW optimizer.
- Applies Batch Normalization over a three dimension input.
- Applies Batch Normalization over a four dimension input.
- Applies Batch Normalization over a five dimension input.
- Creates a new convolution layer for any number of dimensions.
- Creates a new one dimension convolution layer.
- Creates a new two dimension convolution layer.
- Creates a new three dimension convolution layer.
- Creates a one dimension transposed convolution layer.
- Creates a two dimension transposed convolution layer.
- Creates a three dimension transposed convolution layer.
- Creates a new GRU layer.
- Creates a new linear layer.
- Creates a LSTM layer.
- The default convolution config without bias.
- Creates the configuration for the RmsProp optimizer.
- Creates a new empty sequential layer.
- Creates a new empty sequential layer.
- Creates the configuration for a Stochastic Gradient Descent (SGD) optimizer.
Type Aliases§
- One dimension convolution layer.
- Two dimensions convolution layer.
- Three dimensions convolution layer.
- Convolution config using the same parameters on all dimensions.
- A one dimension transposed convolution layer.
- A two dimension transposed convolution layer.
- A three dimension transposed convolution layer.
- A transposed convolution configuration using the same values on each dimension.