Pretraining/training script, loss functions for hi-ml-multimodal (BioViL)

[santosh9sanjeev]

Hello! Thanks for the wonderful work and for sharing the pretrained weights.

In your ECCV work (Making the Most of Text Semantics to Improve Biomedical Vision–Language Processing), there are several results on classification and segmentation(zero-shot) tasks. I wanted to get the pretraining script which has information about the loss functions, hyperparameter settings etc. , but I am not able to find it.

Can you provide some information regarding where we can get these training scripts? Thank you!

[sayeh1994]

Following.