DotTorch Layers is a high-performance, modular neural network layers library for .NET 8 and .NET 9. It includes core layers such as Linear, ReLU, Sequential, Dropout, Embedding, Sigmoid, SoftMax, Tanh, LeakyReLU, GELU, ELU, and Flatten. Advanced recurrent layers like RNN, LSTM, and GRU are also implemented, along with powerful Transformer layers. The package features normalization layers: LayerNorm (currently not optimized) and BatchNorm (optimized, with LayerNorm mode support). All layers seamlessly integrate with the DotTorch.Core autograd system, enabling automatic differentiation and backpropagation. Designed for ease of use, extensibility, and efficient execution on CPU and GPU devices. This library supports modern .NET frameworks and follows best practices for maintainability and performance in machine learning model construction.
Get Started
$ dotnet add package DotTorch.LayersReadme
DotTorch.Layers — Модульная библиотека нейросетевых слоёв для .NET
Содержание / Contents / Inhalt / 目录
Русский
DotTorch.Layers — модульная библиотека нейросетевых слоёв для платформы .NET. Пакет расширяет DotTorch.Core, предоставляя базовые строительные блоки нейронных сетей с полной поддержкой broadcasting и автоматического дифференцирования (autograd).
Основные возможности:
- Реализация популярных слоёв:
Linear,ReLU,Sequential, и других. - Совместимость с
DotTorch.Core: поддержкаTensor, вычислительного графа и autograd. - Единый интерфейс
Moduleс поддержкой вложенных структур и параметров. - Простой механизм переиспользования моделей и параметров.
- Высокая производительность и лёгкость расширения.
- Полное покрытие тестами.
DotTorch.Layers предназначен для создания нейронных сетей любой сложности на базе функциональности DotTorch.Core.
English
DotTorch.Layers is a modular neural network layer library for the .NET platform. It extends DotTorch.Core by offering essential building blocks for model construction, with full support for broadcasting and automatic differentiation.
Key features:
- Implementation of core layers:
Linear,ReLU,Sequential, and more. - Seamless integration with
DotTorch.Core: tensor operations, computation graph, autograd. - Unified
Moduleinterface with support for parameter trees and nested structures. - Easy model composition and parameter reuse.