heat
heat copied to clipboard
helmholtz-analytics
Distributed tensors and Machine Learning framework with GPU and MPI acceleration in Python
A consortium is defining an Array API: https://data-apis.org/array-api/latest/index.html It seems desirable for HeAT to be as compliant as possible. The following features are missing in HeAT (others are probably incomplete):...
Does heat have (or plan to have) an implementation of linear operators like the one in spicy https://docs.scipy.org/doc/scipy/reference/generated/scipy.sparse.linalg.LinearOperator.html#scipy.sparse.linalg.LinearOperator? I think cupy now has similar stuff. For large sparse matrix like...
**Description** A clear and concise description of the bug and the associated functionality. The edge case of performing ht.diff(array) along an axis of size 1 is handled differently when array...
Walk through of using DPNNs with DASO or without. needs a dataloader, optimizer, network, loss, train loop, val loop
**Feature functionality** The current dataloaders in heat can be greatly improved by increasing their compatibility with the torch's dataloader methods. For example, loading data from folders would be very useful.
**Feature functionality** Since torch 1.7 other backends than CPU and CUDA are supported. Among them is OpenCL. HeAT should be able to use it. Pay close attention whether torch's OpenCL...
**Feature functionality** Since torch 1.7 other backends than CPU and CUDA are supported. Among them is the HIP, aka AMD's CUDA, backend. HeAT should be able to use it. Moreover,...
[...] generally however I would obviously like to have something like a `ht.view` of the data, especially silent return of it. I am not 100% sure whether we can actually...
the way we do it right now is e.g., `ht.zeros(5).larray + torch.randn(5)` and then wrap it into a (distributed) dndarray again. It would be less frustrating if this step were...
Metadata
Owner
Metadata
Distributed tensors and Machine Learning framework with GPU and MPI acceleration in Python