lucidrains/transfusion-pytorch

Got error when running example

Closed this issue · 6 comments

lucidrains commented

hey Bing, the repo is not finished yet, but what version are you on? is this just the first readme example?

Bing1002 commented

Hey Phil, I appreciate your implementation of this model and your quick response. I used the latest master branch and readme example.

  • 👍1
lucidrains commented

@Bing1002 ah ok, can you try the updated readme?

i'm in the process of allowing the language model to dictate the shape of the modality

lucidrains commented

@Bing1002 Motorola has an ML/AI research group? that's news to me!

  • 😄3
Bing1002 commented

@Bing1002 Motorola has an ML/AI research group? that's news to me!

@Bing1002 ah ok, can you try the updated readme?

i'm in the process of allowing the language model to dictate the shape of the modality
It works okay now. Thank you so much. It would be great if you are willing to add more comments on the model fake input saying which tensor is what?

  • 👍1
lucidrains commented

@Bing1002 yea, i'll work my way towards the data end once the architecture is complete

for now, it is quite simple, any torch.int or torch.long is text, torch.float are modalities