Installation¶

← Getting Started | Quick Start →

torchregress needs Python 3.12 or newer (tested on 3.12, 3.13 and 3.14; 3.13 recommended) and PyTorch 2.13 or newer.

From PyPI¶

pip install torchregress

Optional extras¶

pip install "torchregress[flows,viz]"

import torchregress works without the extras; a module that needs one raises ImportError when you use it.

With pixi¶

In a pixi project, add torchregress from PyPI next to the conda-forge PyTorch build:

pixi add pytorch
pixi add --pypi torchregress

From source¶

Development uses pixi:

git clone https://github.com/astroai/torchregress.git
cd torchregress
pixi install

pixi run test        # pytest + coverage
pixi run lint        # ruff
pixi run typecheck   # ty
pixi run docs        # zensical build --strict
pixi run ci          # all of the above

pip install -e ".[all]" from a clone also works if you do not use pixi.

Requirements¶

Check the installation¶

The package installs a health check that prints the torchregress, PyTorch and Python versions and the device, imports the main modules, and runs one training step and one metric on random data (on the GPU when one is available):

torchregress-health

Or check the version from Python:

import torchregress
print(torchregress.__version__)

A minimal training loop:

import torch
import torchregress as tr

X = torch.randn(100, 1)
y = 2 * X + 1 + 0.1 * torch.randn(100, 1)

model = torch.nn.Linear(1, 1)
loss_fn = tr.losses.WeightedHuberLoss()
optimizer = torch.optim.SGD(model.parameters(), lr=0.1)
for _ in range(100):
    loss = loss_fn(model(X), y)
    optimizer.zero_grad()
    loss.backward()
    optimizer.step()

with torch.no_grad():
    print(f"RMSE: {tr.metrics.rmse(model(X), y).item():.4f}")