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
Apologies for the delay, is the training dataset you are using in MDS streaming format? If so, you can use StreamingDataset class to load data even if it is stored locally.
@Landanjs I have the same issue when loading data in MDS streaming format. I have built my custom dataset purely using PyTorch's Dataset, not StreamingDataset, is that the problem?
@hieuphung97 is there a reason you aren't using a subclass of StreamingDataset to load your data? A custom dataset purely using PyTorch's Dataset may miss some logic to load MDS shards, so we recommend subclassing the StreamingDataset if possible.
Bug when local training with LocalDataset
Here is my config (without some personal paths), run for mosaicml's diffusion:
Here is my dataset and dataloader code:
And I got this errors while finish epoch 0 and start epoch 1:
Plz help!!!
The text was updated successfully, but these errors were encountered: