Ember

Ember is a high-performance, lightweight deep learning library for Python. It provides a familiar NumPy-like interface with heavy lifting done by optimized C (CPU) and CUDA (GPU) backends.

Key Features

• Tensor Operations: Fast element-wise operations and matrix multiplication.
• Hardware Acceleration: Automatic CUDA support on NVIDIA GPUs.
• Pythonic: Seamless integration with NumPy and Python lists.
• Lightweight: Minimal dependencies, easy to install.

Installation

You can download and install it in a UV virtual environment using commands:

git clone https://github.com/pawel002/ember
cd ember
uv sync
source .venv/bin/activate
uv pip install -e .

Quick Start

from ember import Tensor

# Create tensors
a = Tensor([1.0, 2.0, 3.0])
b = Tensor([4.0, 5.0, 6.0])

# Perform operations
c = a * b
print(c.to_list()) # Output: [4.0, 10.0, 18.0]
print(c.to_np())   # Output: np.array([4.0, 10.0, 18.0])

Documentation

Full documentation is available under this link or can be built locally using MkDocs (by switching to docs dependency group):

uv sync --group docs
uv run mkdocs serve

License

This project is licensed under the terms of the MIT license.