xiao-xian

Results 1 comments of xiao-xian

Many thanks @MarcusLoppe!! I pull your branch and use the above notebook to run the training. The training loss for encoder is around 0.28: ![image](https://github.com/lucidrains/meshgpt-pytorch/assets/2750374/4f9119de-8125-4437-b0e2-d24741be42de) However for transformer, it never...