helmholtz-analytics/heat
Distributed tensors and Machine Learning framework with GPU and MPI acceleration in Python
This tool helps scientists, engineers, and researchers perform data analytics and machine learning on very large datasets that don't fit into a single computer's memory. You input your existing Python code, often using NumPy or SciPy-like operations, and it outputs results from complex calculations, but processed efficiently across multiple CPUs or GPUs in a cluster. This is ideal for those working with massive scientific simulations or experimental data.
231 stars. Available on PyPI.
Use this if you need to run high-performance data analytics or machine learning tasks on datasets too large for a single machine, leveraging the power of multiple CPUs or GPUs.
Not ideal if your data comfortably fits within the memory and processing power of a single computer, as the overhead of distributed computing might be unnecessary.
Stars
231
Forks
61
Language
Python
License
MIT
Category
Last pushed
Mar 18, 2026
Commits (30d)
0
Dependencies
4
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/ml-frameworks/helmholtz-analytics/heat"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Related frameworks
deepspeedai/DeepSpeed
DeepSpeed is a deep learning optimization library that makes distributed training and inference...
hpcaitech/ColossalAI
Making large AI models cheaper, faster and more accessible
horovod/horovod
Distributed training framework for TensorFlow, Keras, PyTorch, and Apache MXNet.
bsc-wdc/dislib
The Distributed Computing library for python implemented using PyCOMPSs programming model for HPC.
learning-at-home/hivemind
Decentralized deep learning in PyTorch. Built to train models on thousands of volunteers across...