pytorch_scatter
pytorch_scatter copied to clipboard
rusty1s
PyTorch Extension Library of Optimized Scatter Operations
I have a `[n, d]` tensor `A` and a `[t, n]` tensor `weight` and a `[n]` tensor `index`. I would like to obtain a `[t, m, d]` tensor `A_scatter` such...
As I'm trying to figure out why #266 isn't working, I want to try adding a couple less controversial things to the existing testing workflow - [x] Test on more...
I am trying to use torch_scatter with torch nightly from source. I compiled torch_scatter from source as follows : 1. git clone https://github.com/rusty1s/pytorch_scatter.git 2. cd pytorch_scatter 3. export PATH=$PATH:/usr/local/cuda/bin 4....
I can't seem to install on a Mac M1. Here is the error message associated with it ERROR: Command errored out with exit status 1: command: /opt/homebrew/Caskroom/miniforge/base/envs/pymc3_env/bin/python3.9 -u -c 'import...
hello, when I used torch_scatter, I suffered from this problem below. `Traceback (most recent call last): File "/home/bruno/anaconda3/envs/tf/lib/python3.7/site-packages/torch_scatter/__init__.py", line 14, in f'{library}_{suffix}', [osp.dirname(__file__)]).origin) AttributeError: 'NoneType' object has no attribute 'origin'...
[wip] Proof of concept that torch ops in nightly will pass tests + identify gaps in what's in core Gaps with no workarounds (fixes are wip): - [x] integer support...
Is it possible to make gather_csr_cuda() without cpu-gpu sync? I can only guess that the problem is in line 248 in csrc/cuda/segment_csr_cuda.cu: ``` sizes[dim] = indptr.flatten()[-1].cpu().data_ptr()[0]; ```
This PR exists to show the "state of the world" for the "before" benchmark ([**rendered benchmarking results cuda**](https://github.com/rusty1s/pytorch_scatter/blob/58bd3fba7b6cd7564d24339134d41ad18d84ef04/benchmark_before_cuda.md)) - fixes CUDA illegal memory accesses run into by benchmarking script mentioned...
I'm currently trying to install with Cuda version 11.2, Python version 3.6, and PyTorch version 1.10.2, but "pip install torch-scatter" fails, it does so with cudatoolkit=11.3 and cudatoolkit=10.2. I was...
Hi @rusty1s, Thanks again for your great work on this library! I am currently experimenting with computing second-order derivatives that involve torch_scatter operations, and noticed that the `segment_coo` and `segment_csr`...
