Hanchen

Results 34 comments of Hanchen

add a few more on GNN, 3D Vision and Privacy: - [NeurIPS2019] [Quaternion Knowledge Graph Embeddings](https://arxiv.org/pdf/1904.10281.pdf) - [BMVC2018 (Best Paper)] [QuaterNet: A Quaternion-based Recurrent Model for Human Motion](https://arxiv.org/pdf/1805.06485.pdf) - [ArXiv...

Thank you, @zshyang. In my experience, the pretraining loss also reached saturation quickly. Regarding the scenario where better visualizations with a slightly increased loss, I believe this is due to...

Hi @AgnesDouble , thanks for raising this question. We did not incorporate the epoch_update function in [train_partseg.py](https://github.com/hansen7/OcCo/blob/master/OcCo_Torch/train_partseg.py), although it is fully functional. Adapting the epoch_update for `partseg` mode from the...

It appears that the training is progressing well, as both the training and test instance accuracies are increasing. Could you please check if the logging is functioning correctly? What datasets...