Question about training dataset

[WilliamHBW]

In trainer.py, the training dataset seeems to be set as Dataset from dataset_lossy.py, and only use the first four sequences from sequence_list, i.e. soldier/longdress/redandblack/loot. And the testing dataset used in test_owlii.py is from dataset_owlii.
However, according to the paper, the training dataset is generated from owlii and the model is tested on 8ivfb. Do these two strategy give similar results?

[ttlzfhy]

The uploaded codes use 8iVFB for training and Owlii for testing because our proposals to MPEG AI-3DGC follow this experiment setting. To reproduce the results in the paper (Owlii for training and 8iVFB for testing), just change the parameters in trainer.py and test_owlii.py.