Inference using pretrained models and batch size > 1


I’m doing inference using the OWLViTForDetection pretrained model. I’m trying to use a batch size > 1 but getting an error (the typical ‘Too many values to unpack’ one) and by looking at the code it seems that it works only with a batch size = 1. Does anyone know whether/how I can use larger batch sizes for performing inference with pretrained models? Thank you!