docs.rs failed to build torsh-0.1.0-alpha.1
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.
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.
torsh
The main crate for ToRSh - A blazingly fast, production-ready deep learning framework written in pure Rust.
Overview
This is the primary entry point for the ToRSh framework, providing convenient access to all functionality through a unified API.
Installation
Add to your Cargo.toml
:
[]
= "0.1.0-alpha.1"
Quick Start
use *;
Available Features
default
: Includesstd
,nn
,optim
, anddata
std
: Standard library support (enabled by default)nn
: Neural network modulesoptim
: Optimization algorithmsdata
: Data loading utilitiescuda
: CUDA backend supportwgpu
: WebGPU backend supportmetal
: Metal backend support (Apple Silicon)serialize
: Serialization supportfull
: All features
Module Structure
The crate re-exports functionality from specialized sub-crates:
- Core (
torsh::core
): Basic types and traits - Tensor (
torsh::tensor
): Tensor operations - Autograd (
torsh::autograd
): Automatic differentiation - NN (
torsh::nn
): Neural network layers - Optim (
torsh::optim
): Optimizers - Data (
torsh::data
): Data loading
F Namespace
Similar to PyTorch's torch.nn.functional
, ToRSh provides functional operations in the F
namespace:
use F;
let output = relu;
let output = softmax?;
License
Licensed under either of
- Apache License, Version 2.0 (LICENSE-APACHE)
- MIT license (LICENSE-MIT)
at your option.