torch-batch-svd
torch-batch-svd copied to clipboard
A 100x faster SVD for PyTorch⚡️
In pytorch 1.10, bfloat16 (from Google Brain) support was released. This PR adds support for this.
Allow passing in tol and max sweeps arguments?
Hi, I followed the introduction and finished installing process successfully, but when I try to import "from torch_batch_svd import svd", it raises error of "File "/disk2/jiale.tjl/projects/articulated-animation/torch-batch-svd/torch_batch_svd/batch_svd.py", line 3, in from...
**Describe the problem** 2 out of 4 tests failed with differences exceeding the margin of error. **Error message** ``` ============================ test session starts ============================= platform linux -- Python 3.7.10, pytest-6.2.5,...
**Describe the problem** Tested torch-batch-svd with torch version 1.7.1 and cuda 10.0, it works well However, when testing it with torch version 1.7.3 and cuda 11.0, it fails. always shows...
**Describe the problem** When I finished the "python setup.py install", I met the problem "ImportError: DLL load failed while importing _c: The specified module could not be found. " importing...
Hi, is there any possibility to add support for torch.complex64? I need to run SVD on millions of complex tensors, of size 21 by 21, and torch.svd seems a bit...
**Describe the problem** Why it only works for matrices whose row and column are smaller than 32?How can i use it for higher dim?
Sorry for asking this, but has a benchmark been made comparing this with the following repo that is based on Numba? https://github.com/gbolmier/funk-svd
Thanks for providing this excellent tool! I try to run the test.py file, but it fails. Please see the error message attached. **Environments** 1. OS: Ubuntu 2. CUDA version: 11.7...