Environment and reproducibility

gpu-gate

Wait for a free GPU, claim it, and run your command.

Install

pip install gpu-gate

Once published to PyPI. Also available now from GitHub:

pip install git+https://github.com/jmweb-org/gpu-gate

What it does

On a shared multi-GPU box without a scheduler, launching a job means watching nvidia-smi, picking a card by hand and exporting the variable. gpu-gate is that wait-pick-export-run loop, with a cooperative lock.

Features

—Selection by free memory and utilization.
—Cooperative lock so two runs do not grab the same card.
—Exports CUDA_VISIBLE_DEVICES and runs.
—No daemon or server; honest exit codes.

View the code on GitHub →

Other tools

dsdiff

A git-style diff between two datasets, with distribution drift.

splitcheck

Detect rows leaking between train, validation and test splits.

pii-sweep

Scan datasets for PII before sharing them.