Search results

Only blockshape=(32, 32) is supported

Status: Open.

#20 In huggingface/pytorch_block_sparse;

· Routhleck opened on May 22, 2023

bfloat16 support

Status: Open.

#19 In huggingface/pytorch_block_sparse;

· lostmsu opened on Nov 4, 2022

Using a smaller block size

Status: Open.

#18 In huggingface/pytorch_block_sparse;

· pclucas14 opened on Jun 22, 2022

What is "custom" "custom-back" in dispatch_policies.h?

Status: Open.

#17 In huggingface/pytorch_block_sparse;

· ziyuhuang123 opened on Jun 16, 2022

Sparse pattern is not guaranteed to be full rank

Status: Open.

#16 In huggingface/pytorch_block_sparse;

· jrapin opened on Nov 17, 2021

No memory reduction observed in a simple sparse-dense multiplication

Status: Open.

#15 In huggingface/pytorch_block_sparse;

· x-zho14 opened on Mar 26, 2021

I get error on this simple code

Status: Open.

#14 In huggingface/pytorch_block_sparse;

· ebrahimHappy opened on Dec 1, 2020

can we used pre-defined sparse matrix?

Status: Open.

#13 In huggingface/pytorch_block_sparse;

· hsmills opened on Nov 17, 2020

could you provide some code for loading a saved sparsified model?

Status: Open.

#12 In huggingface/pytorch_block_sparse;

· DanqingZ opened on Nov 11, 2020

does this package allow multi-gpu training and distributed training?

Status: Open.

#11 In huggingface/pytorch_block_sparse;

· DanqingZ opened on Nov 11, 2020

Any way to bypass constraint of input and output features to be a multiple of block size

Status: Open.

#9 In huggingface/pytorch_block_sparse;

· shreyanshs opened on Nov 9, 2020

Incorrect error message (says in_features, should be out_features)

Status: Open.

#8 In huggingface/pytorch_block_sparse;

· ejohb opened on Sep 19, 2020