-
In https://github.com/Jutho/TensorKit.jl/issues/50#issuecomment-830868009 I found
> symmetries under index permutations are currently not supported. In general, it's quite hard to take them into acc…
-
In fact it looks like we may silently ignore them.
The ctx methods in question are:
- ctx.mark_dirty()
- ctx.mark_non_differentiable()
- ctx.set_materialize_grads()
Repro:
```
import torch
…
-
- torch.empty(5, 3)
- torch.rand(5, 3)
- torch.zeros(5, 3, dtype=torch.long)
- torch.tensor([5.5, 3])
- x = x.new_ones(5, 3, dtype=torch.double) # new_* methods take in sizes
x = torch.ra…
-
## 🚀 Feature
Call non class methods from `torch::jit::Module` in C++
## Motivation
For `torch::jit::Module` loaded in C++, we can use `get_method` to load and execute **class methods** in the…
-
memory swap (#1966) has a benefit on memory usage, but it suffers several disadvantages:
1. swap device I/O delay
2. reducing lifetime of swap device
3. latency delay of calculation caused by I/O…
-
I've been able to run metaerg successfully with and without providing a depth file on some of my datasets using:
metaerg.pl --dbdir /projects/p30777/db --depth /projects/p30777/metagenome_data/bam…
-
### 🐛 Describe the bug
Related: https://github.com/pytorch/vision/pull/5569#issuecomment-1063507413
A tuple of float such as `(0.5, 0.5, 0.5)` is accepted on function with input parameter type `…
-
-
ATen is presently has no const modifiers, even for methods which are obviously const (e.g. `virtual bool isCuda()` in Type). Make it more const-correct.
-
### Feature description
Add common linear algebra operations.
### Feature motivation
We've had a couple of users ask about support for linalg ops like SVD, QR decomposition and matrix exponen…