torchlite-cpp

A lightweight C++ tensor library implementing core PyTorch-like operations for CPU. Designed as a minimal foundation for building neural networks without the complexity of full frameworks.

Core Features

Tensor Operations

• N-dimensional tensor data structure with automatic stride calculation
• Element-wise operations (add, sub, mul, div, neg, exp, log, pow, clamp) with NumPy-style broadcasting
• Matrix multiplication with batch dimension support
• Reductions: sum, mean, variance, argmax, softmax
• Reshape, transpose, cat, stack, slice, pad

Neural Network Modules

• Linear, LayerNorm, Dropout
• MultiHeadAttention, TransformerEncoderLayer, TransformerEncoder
• PositionalEncoding
• Activation functions: ReLU, GELU, Sigmoid

Design

• CPU-only, float32 operations
• Weight loading from raw binary files

Quick Start

Build the library:

mkdir build && cd build
cmake ..
make

Run tests:

./build/run_tests

Project Structure

include/tl/         Public API headers (tensor, ops, nn, activation, factory)
include/external/   Third-party headers (LibrosaCpp)
src/                Implementation
tests/              Test executables

Dependencies

• Eigen3: required by LibrosaCpp for audio preprocessing
• LibrosaCpp: single-header mel spectrogram computation (included in include/external/)

Requirements

• C++17 or later
• CMake 3.10+
• Eigen3 (sudo apt install libeigen3-dev)

License

MIT License