-
When I use FMHA_v2, I found it does not support my scenes. So i wonder is there any way to use fmha_v2 except changing model.
Thx a lot.
-
Thanks for your great work!
when I created an instance of BCNLanguage and let it load _pretrain-language-model.pth_, an error emerged:
> Traceback (most recent call last):
File "demo_language.py…
-
Hey
KDE support doesn't seem to work at all \(icons don't
display\) with two monitors enabled \(i.e., for example,
displays :0.0 and :0.1 \).
I'm using version 0.1.14-3 / Debian Testing.
/ …
-
Hi ,Now I want to use tch-rs in nlp project ,do we support these layer? I not found them. thank
-
$ bash ./benchmark.sh ../../sgk_models/transformer/sparse sparse
........
No OpKernel was registered to support Op '**CsrSoftmax**' used by {{node sparse_transformer/body/decoder/layer_0/self_…
-
Add support for TF Multihead attention and TimeDistributed layers
This seems to be broken in the latest version..one we moved to strong typing for the layer types, we now have to support complex type…
-
https://github.com/NVIDIA/FasterTransformer/blob/8bf96eb2a0aed68ad8b75ac66b262000eeea2224/src/fastertransformer/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention_template.h…
-
### Branch/Tag/Commit
default
### Docker Image Version
default
### GPU name
v100
### CUDA Driver
11.6
### Reproduced Steps
```shell
I put right qkv in 'fusedQKV_masked_attention_dispatch',but…
-
-
```
What steps will reproduce the problem?
1. Clean FreeBSD install.
2. Install the following ports:
- x11/hs-X11
- x11/hs-xmonad
- x11/xorg-minimal
- x11/xterm
- x11/nvidia-xconfig
…