-
Thank you for your great work! I should have found out this repository earlier.
The thing I wanna ask is as of GLU function that you've implemented.
I noticed that xformer's implementation of Sw…
-
### Feature request
This is a tracker issue for work on _interleaved_ in-and-out image-text generation.
There are now >= 5 open-source models that can do _interleaved_ image-text generation--and…
-
### Links
- Paper : https://arxiv.org/abs/2106.09785
- Github : https://github.com/microsoft/esvit
- Openreview : https://openreview.net/forum?id=fVu3o-YUGQK
### 한 줄 요약
- Microsoft에서 나온, Swin t…
-
## Information
The question or comment is about chapter:
* [ ] Introduction
* [ ] Text Classification
* [ ] Transformer Anatomy
* [ ] Multilingual Named Entity Recognition
* [ ] Text Generat…
-
# Efficient Attention
## Reference
- [Efficient Attention](https://github.com/Separius/awesome-fast-attention)
- 2020-09 Efficient Transformers: A Survey [[Paper](https://arxiv.org/abs/2009.06732…
-
I'm wondering if I can get an easier pipeline by loading the awq weights with vllm:
```
from vllm import LLM, SamplingParams
prompts = [
"Hello, my name is",
"The president of the Uni…
-
### 🐛 Describe the bug
Looks like it's dispatching to efficient attention backward and failing one of the shape checks (
```
TORCH_CHECK(
max_seqlen_k
-
Hi! !مرحبا! السلام عليكم
Let's bring the documentation to all the Arabic-speaking community 🌏 (currently 0 out of 267 complete)
Would you want to translate? Please follow the 🤗 [TRANSLATING guid…
-
Provided code calculates matrix product of q and k.
https://github.com/YuchuanTian/DiJiang/blob/main/modeling/pythia-2.8B-dijiang/modeling_gpt_neox_dijiang.py#L286
That means it has computational …
-
It is giving me:
```
ValueError: Transformers now supports natively BetterTransformer optimizations (torch.nn.functional.scaled_dot_product_attention) for the model type llama. Please upgrade to tra…