xformers

xformers copied to clipboard

[fix] benchmark shows 0.00MB consumed

1

## What does this PR do? Fixes bugs: when run pytest, the memory consumption printed is 0.00MB. This is because class creation (also weight creation) happens in `benchmarks_generator`. Which is...

deciding

[python `3.11`] Unable to install xFormers above v0.0.16

12

# ❓ Questions and Help xFormers cannot be updated to the latest version (0.0.19 or beta version 0.0.20), and pip and other methods can only be installed up to 0.0.16....

E7admin

python 3.12 has unlock more power of python, and now stable with latest version `3.12.2`. Besides, mainstream repo including `pytorch torchvision huggingface_hub transformers accelerate diffusers has already support it`. The...

Mon-ius

# ❓ Questions and Help Can I call this: xformers.ops.memory_efficient_attention(q, k, v, attn_bias=rp_bias, scale=self.scale, op=None) where the rp_bais is a Relative Position Bias ############## the original code: qkv = self.qkv(x).reshape(B,...

CFOP-xyn

XFormer AssertionError: This attention does not support attention masks

1

# ❓ Questions and Help Hi, I tried to run the BlockSparseAttention tutorial code (https://facebookresearch.github.io/xformers/tutorials/blocksparse.html) and I have the following error: "AssertionError: This attention does not support attention masks" I...

Mmozzanica5

# 🐛 Bug XFormers can not perform memory_efficient_attention, ## Command ## To Reproduce The code is from attention_processors line 266 (it can change but not too much) of the diffusers...

axel578

Cannot Compile for Older Version of PyTorch

5

# ❓ Questions and Help Running Ubuntu 22. I am trying to compile xformers for Pytorch 1.12.1 + CUDA 11.3. No matter what do, the compile process downloads Pytorch 2,...

6630507

# ❓ Questions and Help Where can I find the older version that can fit my torch enviroment(torch=1.10.0+cu113).I don‘t want to change my torch version.What command should I use?Thanks!

ccxi1008

# ❓ Questions and Help The MicrGPT example specifies `block_type` as "encoder" (https://github.com/facebookresearch/xformers/blob/main/examples/microGPT.py#L53). Shouldn't this be "decoder"?

rteehas

# 🐛 Bug Documentation in Examples: The link to Lightning AI tutorial on CIFAR is broken ## Command ## To Reproduce Steps to reproduce the behavior: 1. go to examples/README.md...

melvinebenezer