ruda-nn 0.21.12

Ruda neural network layers, activation modules and losses.
docs.rs failed to build ruda-nn-0.21.12
Please check the build logs for more information.
See Builds for ideas on how to fix a failed build, or Metadata for how to configure docs.rs builds.
If you believe this is docs.rs' fault, open an issue.
Visit the last successful build: ruda-nn-0.21.10

ruda-nn

Neural-network layers, activation modules, padding, and loss functions for Ruda models. Layers operate through tensor backend contracts rather than selecting a GPU driver themselves.

Interfaces

  • modules contains neural-network layer implementations and is re-exported at the crate root.
  • activation exposes activation modules; loss exposes loss functions.
  • Initializer is re-exported from ruda-model for parameter initialization.

Usage

Cargo package: ruda-nn. Rust import: ruda_nn.

[dependencies]
ruda-nn = "0.21"

Features

Default features: std, ruda-model/default.

Feature Purpose
std Enable standard-library model integration.
sparse Enable sparse layer support.
cuda Enable the CUDA tensor dependency.
fusion Enable device-fusion integration.

Links