Sun Weixiang comments

Results 8 comments of


                                            Sun Weixiang

No such file or directory: modelnet40_shape_names.txt'

hi, I didn't find .txt file in this linkl https://drive.google.com/drive/folders/1_fBYbDO3XSdRt3DSbEBe41r5l9YpIGWF?usp=sharing, could u tell me where did u find id?

No such file or directory: modelnet40_shape_names.txt'

> Hello, has the problem been resolved? I have encountered the same issue, and it seems that the website https://shapenet.cs.stanford.edu/media/modelnet40_normal_resampled.zip is not working. U can try this link：https://pan.baidu.com/s/1V9AhtDUhGPedpdmwT1guaA pwd：crgo

Can I train OpenSora using multiple 16GB V100 cards?

> Although you can by using sequence parallelism, I do not recommend you to do so. Maybe you can try something like LoRA. We will continue to work on it...

root cause： no init.py file under shardformer folder.

Creat one and reinstall it

root cause： no init.py file under shardformer folder.

> > Creat one and reinstall it > > Reinstall what, the conda environment, or what? only opensora

RuntimeError: Error(s) in loading state_dict for get_model:

Hi, have u find the solution?

Pretrained Models do not load with pip release

same problem! Do u find any solution?

和想象中的效果差太多了，是我的哪里没设置好吗

> 你们说的那个bug都是apex这个垃圾引起的，需要安装apex包，我用下载源码的22.04-dev，然后改一下torch._six,编译安装，在配个其他，还有flash_attn这个包，也挺慢的我的配置是ubuntu22.04，A100 80gb，torch 2.0.1 cuda12.2 驱动535 作者估计有更好的权重没有放出来，同样的prompt，我反正没有一个像演示的那种效果 torch._six是改什么能告诉下么