-
-
Operator:
- [x] sum(csr) @anirudh2290
- [x] mean(csr) @anirudh2290
- [x] sparse embedding(row_sparse gradient) @eric-haibin-lin
- [ ] concat(csr, csr)
- [x] split_axis(csr) @ZiyueHuang
- […
-
Desciption: In DeepSpeed-Chat step3, a runtime error: The size of tensor a (4) must match the size of tensor b (8) at non-singleton dimension 0 will be thrown when inference_tp_size>1 and hybrid engin…
-
Is there a way to support pipelines with CPU offloading enabled?
It seems currently unable to handle this condition
```python
import gc
import torch
from diffusers import StableDiffusion3Pipe…
-
### Name
Dynaωo
### Screenshots
### Focus To…
-
📚 This guide explains how to use YOLOv5 🚀 **model ensembling** during testing and inference for improved mAP and Recall. UPDATED 25 September 2022.
From https://www.sciencedirect.com/topics/comput…
-
Is there any parameters that could assign occpuation number for different elements?, I 'd like to use this program to create supercell of Cd0.5Zn0.5S. But it seems there is no option to assign occupat…
-
用我们自己的SR Dataset 开始测试了, 58W张 720x720 的高清图, 数据分布非常好 :) 相信我 :)~
已经跑起来,开始train 了, 不过 train 起来是真的慢啊, MSE model 需要 27 天 :( 然后 GAN 估计还需要27 天
27 天啊, A100 x4 .
不过为了保证质量, options 文件做了点修改:
gt_size: 3…
-
These will need some combination of documentation and scripts/functions/R package:
- Setup
- Compilation
- On MacOS (for local runs)
- On PIC (for large simulations)
- Shared input …
-
### 🐛 Describe the bug
I am running example codes show in https://github.com/hpcaitech/ColossalAI/tree/main/examples/language/gpt/experiments/auto_parallel with Pytorch 2.0 (because I need to deploy…
wxthu updated
4 months ago