Aaron Nicolson

Hi Minseung-Kim, Using two of the speakers for the validation set has been the standard way. I have not personally seen it done another way :)

good pickup @JackRAHealth, keeping it 55th street

I did not get the chance to develop the model to run on a real-time system. It would need some more development, but I assume its possible. You could do...

Sounds like an interesting problem to investigate :) I am sure it could work with some constraints. Consider things like using previously computed keys to speed up processing, e.g., this...

Hi @congjianting, That checkpoint is made available by Microsoft (at https://github.com/microsoft/CvT). The OneDrive link still works, it appears that you do not have access to it unfortunetly. This model is...

Are you trying to train a model? Or do you just want to do inference?

Hi ChantalMP and fuying-wang, Thank you for pointing this out. Our reported results in Table 6 are indeed averaged over each example (example-based CE metrics). The results for the other...

Hi, its due to an update to the torchmetrics package. I'll amend it now.

Okay, Are you attempting: ``` dlhpcstarter -t mimic_cxr_chen -c config/test_mimic_cxr_chen_cvt2distilgpt2.yaml --stages_module stages --test ``` ? And what version of transformers are you running? I'll look into it now.

Okay. I updated a few things and It now works with the latest transformers (4.35.2). Please pull the repo. Also, you might be interested in https://huggingface.co/aehrc/cxrmate.