huochai

Out of memory when training with RTX4090, seeking guidance on training details

2

I am currently attempting to reproduce the training process described in your paper using Stable Diffusion v2. However, my RTX 4090 ran out of memory when training with batchsize 32...

Green sceen after opening an obj file on a remote server

16

I always run things on a remote sever and need to open obj files on the sever through this extension. This extension was working fine util I updated my VSCode...

你好，请问label的数量为什么初始化成batch_paths的两倍？

3

``` label = np.zeros((len(batch_paths) * 2,1), dtype=np.int) label[:(len(batch_paths) - 1)] = 1 blobs['label'] = label`` ```

Cholesky fails

1

``` Traceback (most recent call last): File "test_from_init_pose_pickle.py", line 390, in main() File "/usr/local/lib/python3.8/dist-packages/torch/autograd/grad_mode.py", line 28, in decorate_context return func(*args, **kwargs) File "test_from_init_pose_pickle.py", line 343, in main outputs = model(Gs=pose_input,...

rasterize_points.cu compile error

2

After runing conda env create --file environment.yml, I got the following error: ``` rasterize_points.cu(117): error: argument of type "int *" is incompatible with parameter of type "float *" rasterize_points.cu(118): error:...

Hello. It seems that scatter_sum is not precise compared to torch implementation. Is this a bug? I encountered this problem in my personal project which is sensitive to numerical precision...