Haoli Yin
Haoli Yin
## Uniqueness0 issue in synthetic dataset: It seems that when the synthetic data was retrieved through the __getitem__() method of SyntheticDataset(nn.Dataset) for uniqueness0 it was defaulting to a ones tensor...
From the pytorch stateful distributed dataloader talk, there was a mention at the end how further multi-process development was needed in order to support multimodal datasets. I was wondering what...
Thank you for this work! I noticed that the model checkpoints lead to invalid links. When are you planning on releasing the models?