-
### 🐛 Describe the bug
I would argue that this issue is potentially rather critical given how many sparse compressed algorithms rely on it.
A potential cause of https://github.com/pytorch/pytorch/…
-
The implementation of `unfoldNd` relies on one-hot convolution. This means the convolution kernels are highly sparse. Hence, the code could run faster when using sparse tensors.
Open questions:
…
-
This tracking is meant to list the main improvements we'll be focusing on for sparse tensors, and then mapping them to the releases they should land in. _Please note: these are estimates and indicate …
-
### 🚀 The feature, motivation and pitch
I want to do a sort of batched sparse-sparse matrix multiplication, but specifically in the scenario where I have two hybrid sparse coo tensors of the shape (N…
-
## 🐛 Bug
`is_non_overlapping_and_dense()` does not error out for sparse tensors (it should, similar to `is_contiguous()`), and sparse tensors end up calling it in some tensor factories (e.g. `to_im…
-
Our experience with blasé is that sparse tensors offer large speedups, *so long as the pixel coordinates are fixed*.
This fixed pixel coordinate provision make a nearly circular inference: we need …
-
### 🐛 Describe the bug
This is trying to do a BE task to unblock https://github.com/pytorch/pytorch/pull/130977. The problem is very similar to https://github.com/pytorch/pytorch/pull/120261, though …
-
In #22036 we added sparse allreduce for ProcessGroupGloo. It works for sparse CUDA tensors, but doesn't leverage InfiniBand like NCCL does. Therefore, we should have a sparse allreduce implementation …
-
### 🐛 Describe the bug
```python
from torch._functorch.aot_autograd import aot_export_module
class M(torch.nn.Module):
def __init__(self):
super().__init__()
def forward(self…
-
Models trained by this pipeline perform great. But how to host them using Tesorflow Model Serving? Checkpoint needs to be converted into SavedModel (.pb) format.
What I've done so far is:
1. I'…