You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Firstly, I am very impressed by the outstanding work you have done.
But I found an issue while using your open source code.
For example, I want to quickly integrate model.py and serialization into my code, but I don't know what the data format input into serialization looks like.
If I were to start processing data from scratch and then run your scripts, it would take a long time.
So I suggest the author write a code under if name=="main" on how to input data
The text was updated successfully, but these errors were encountered:
Thanks for this great suggestion. I will try to add more examples of code after this CVPR.
Oh,Thank you.
But,PTv3 consumes too much memory. I only use the PTV3 decoder and input tens of thousands of points, but the 24GB graphics card also memory overflow.
Thanks for this great suggestion. I will try to add more examples of code after this CVPR.
Oh,Thank you. But,PTv3 consumes too much memory. I only use the PTV3 decoder and input tens of thousands of points, but the 24GB graphics card also memory overflow.
Have you enabled FlashAttention? The default setting with a large patch size definitely needs flash attention. But if your local machine cannot install FlashAttention, a smaller patch size like 128 can also work very well.
Firstly, I am very impressed by the outstanding work you have done.
But I found an issue while using your open source code.
For example, I want to quickly integrate model.py and serialization into my code, but I don't know what the data format input into serialization looks like.
If I were to start processing data from scratch and then run your scripts, it would take a long time.
So I suggest the author write a code under if name=="main" on how to input data
The text was updated successfully, but these errors were encountered: