sborse3 comments

Results 6 comments of


                                            sborse3

Can't reproduce scores

Thanks for the reply. The semantic mIoU is at ~25. Could you share your training command and kitti.ini file to reproduce the numbers you obtained in the paper? Thanks!

Is there any guide to train our own model?

Hi @AnnieBunnieX, any luck yet? Thanks!

Some details about the implementations

Hi @garychan22, could you let me know your training dataset? Do you have an OS implementation of your training pipe?

added SA-1B dataset class

Hi @simoneangarano , I found initializing this dataloader quite slow, as SA-1B is huge. Any workaround you would suggest?

Do you plan to open-source the training code？

Hello, @Luh1124 @jwstegemann and @xBrux. I am also looking into training on custom models. Could you provide any pointers? Where do I start, which dataset to use, etc.

SDXL on Snapdragon X Elite Adreno - Blank Image

> Since you say that generation appears to be working with `--clip-on-cpu`, your Snapdragon's Adreno GPU (I'm assuming this is a on phone... please let me know if not) can't...