Expected `tensors` and `new_tensors` to have the same type but found <class 'tuple'> and <class 'torch.Tensor'>

[John6666]

There seems to be a problem that is difficult to solve when dealing with the DETR model, and I hope it’s not this…

Possible fix for trainer evaluation with object detection

I have bumped into an issue of the Trainer class for object detection, that was already found by others (also present as at least one issue in the repository), that the evaluation loop doesn’t work well with this type of model. To summarize, the labels input for the DETR family of object detection is a list of dictionaries. For example: [ { 'size': tensor([ 771, 1333]), 'image_id': tensor([1592]), 'class_labels': tensor([0]), 'boxes': tensor([[0.2268, 0.658…