nice work!!! I tried modifying the backbone of the model, aiming to train it according to the dataset ratios mentioned in the paper.
Could you tell me if all these datasets were downloaded completely?
Also, do I just need to train using the multi-dataset provided in the script? Because I see that the multi-dataset training only has three datasets. When actually training, do I need to download all the data? Approximately how much disk space is required?
nice work!!! I tried modifying the backbone of the model, aiming to train it according to the dataset ratios mentioned in the paper.
Could you tell me if all these datasets were downloaded completely?
Also, do I just need to train using the multi-dataset provided in the script? Because I see that the multi-dataset training only has three datasets. When actually training, do I need to download all the data? Approximately how much disk space is required?