zeekay
27f6f8d258
Sweep of the upstream-Concrete fork applying the canonical Lux FHE
naming map:
zama-ai/concrete-ml → luxfhe/ml-sdk
zama-ai/concrete → luxfhe/torus
zama-ai → luxfhe
github.com/zama-ai → github.com/luxfhe
docs.zama.ai → docs.luxfhe.com
zama-pypi-cpu/gpu → luxfhe-pypi-cpu/gpu
Concrete ML → TorusML
concrete-ml → torus-ml
concrete_ml → torus_ml
Concrete by Zama → Torus by Lux Industries
Concrete framework → Torus framework
Concrete compiler → Torus compiler
tfhe-rs / TFHE-rs → Lux-FHE
fhEVM → TorusEVM
Zama Concrete (ML) → Lux Torus(ML)
Zama → Lux Industries
Fhenix → (removed)
Files renamed for consistency:
ci/scripts/test_cml.sh → test_torus_ml.sh
tools/torus-protocol/src/concrete-protocol.capnp → torus-protocol.capnp
docker/Dockerfile.concrete-compiler-env → Dockerfile.torus-compiler-env
setup.py:
author="Zama Team" → author="Lux Industries"
The 11 files that still contain "Zama" string matches are upstream
license-attribution headers that BSD-3-Clause-Clear §1-§2 REQUIRES
us to retain verbatim. These are:
README.md (legal compliance disclosure)
.github/workflows/scripts/check_for_license.sh (validates upstream headers)
compilers/torus-compiler/compiler/lib/Bindings/Python/torus/ (9 files,
"Part of the Concrete Compiler Project, under the BSD3 License
with Zama Exceptions." copyright-notice headers)
Per §3, no "Zama" name is used to endorse or promote Torus — every
product-facing surface is rebranded; only the legally-required
license headers retain the upstream copyright.
Verified: 0 non-attribution "Zama" content references, 0 "Concrete ML"
content references, 0 "fhEVM" content references. The single binary
mlbench.zip retains an embedded "concrete-ml" string inside the
archive (benign, not a brand reference).
2.6 KiB
Installation
This document explains the steps to install Torus into your project.
Torus is natively supported on Linux and macOS from Python 3.9 to 3.12 inclusive. If you have Docker in your platform, you can use the docker image to use Torus.
Using PyPI
Install Torus from PyPI using the following commands:
pip install -U pip wheel setuptools
pip install torus-python
{% hint style="info" %}
Not all versions are available on PyPI. If you need a version that is not on PyPI (including nightly releases), you can install it from our package index by adding --extra-index-url https://pypi.luxfhe.com/cpu/. GPU wheels are also available under https://pypi.luxfhe.com/gpu/ (check https://pypi.luxfhe.com/ for all available platforms).
{% endhint %}
To enable all the optional features, install the full version of Torus:
pip install -U pip wheel setuptools
pip install torus-python[full]
{% hint style="info" %}
Not all versions are available on PyPI. If you need a version that is not on PyPI (including nightly releases), you can install it from our package index by adding --extra-index-url https://pypi.luxfhe.com/cpu.
In particular, wheels with GPU support are not on PyPI. You can install it from our package index by adding --extra-index-url https://pypi.luxfhe.com/gpu, more information on GPU wheels here.
{% endhint %}
{% hint style="info" %}
The full version requires pygraphviz, which depends on graphviz. Make sure to install all the dependencies on your operating system before installing torus-python[full].
{% endhint %}
{% hint style="info" %}
Installing pygraphviz on macOS can be problematic (see more details here).
If you're using homebrew, you can try the following way:
brew install graphviz
CFLAGS=-I$(brew --prefix graphviz)/include LDFLAGS=-L$(brew --prefix graphviz)/lib pip --no-cache-dir install pygraphviz
before running:
pip install torus-python[full]
{% endhint %}
Using Docker
You can also get the Torus docker image. Replace v2.4.0 below by the version you want to install:
docker pull luxfhe/torus-python:v2.4.0
docker run --rm -it luxfhe/torus-python:latest /bin/bash
Docker is not supported on Apple Silicon.