-
Hey @kohya-ss and @BootsofLagrangian
Thank you for your hard work on Deepspeed integration! As I see this feature in dev branch.
I tried to use deepspeed in my pipelines but got some issues.
It…
-
[澳门WGS84.zip](https://github.com/lovell/sharp/files/13828701/WGS84.zip)
Windows prewview
![image](https://github.com/lovell/sharp/assets/19948086/ecfa093c-d42a-427b-829a-6539ced6398c)
Convert T…
-
### Bug description
When I use _marian-server_ command, something ran incorrectly. The message is as following:
marian-server: /home/work/marian-dev/src/3rd_party/intgemm/avx512_gemm.h:307: static…
-
大佬你好,请问做8bit量化的时候,转成量化模型quant_model的时候,为什么权值是小数吖,比如是0.4256之类的,为什不是0-255之间的整数呢?还是说我哪儿操作步骤有问题
恳请大佬教做人,刚入门量化的小白,感激不尽!
-
It works only if status doesn't contain UTF-8 characters.
```
Grackle::TwitterError => Unexpected failure making request: incompatible character encodings: UTF-8 and ASCII-8BIT
ruby-1.9.3-p448/gems/…
-
### Check if your feature request exists
- [X] My feature request does not already exists
### Is your feature request related to a problem?
I see thew actual logo is very blurred, why not use a bet…
-
机器: 云服务器,24G内存,单卡24G显存
环境:PyTorch 1.12.1,Cuda 11.3.1, Python 3.8,Ubuntu 20.04
(torch使用1.12而非推荐的1.13是因为云服务器提供的镜像只有这个,不知道GPU实际是什么卡不敢贸然换)
问题描述:
把模型和配置文件都放在本地,改动读取地址后,按照项目提供的finetune.ipynb执行**没有错误**,能…
-
Hello,
thank you for enormous effort of bringing this toolkit to the 8bit era fans.
I am on Apple M2 silicon, Sonoma 14.4 OS and already tried running wudsn-ide-install.
Because it failed at som…
-
Idea: put images from fb, e.g. "node of the day" pics to related nodes' help.
-
### Which component impacted?
Video Processing
### Is it regression? Good in old configuration?
No, this issue exist a long time
### What happened?
```
ffmpeg -hwaccel vaapi -hwaccel_output_form…