As I can see, it is an implementation of this paper https://arxiv.org/abs/1502.03044
.@tddammo pointed out to me that there's this really cool looking project on GitHub that uses attention in a cool way that strangely hasn't received much attention (3 stars). Lack of readme, perhaps. But check out the notebook: https://github.com/fg91/Neural-Image-Caption-Generation-Tutorial …pic.twitter.com/hmhJY0iC7e
-
-
-
Yes it is exactly that! Thanks for the paper link
End of conversation
New conversation -
-
-
Mmmmm strange “holding”
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
So, I get this a lot too with Grad-CAM where the edges/corners often light-up despite the model being very accurate/properly fitted. Anyone have insight? Looks like other folk have this problem too...
-
Yes! I've seen this too. It might have something to do with padding or the specific pattern that makes up a padded corner?
End of conversation
New conversation -
-
-
It seems a powerful tool to sell ads
.Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
Not attention based, but Captum has a similar example I found fascinating too: https://captum.ai/tutorials/Multimodal_VQA_Interpret …
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
This implementation has received more... attention:https://github.com/sgrvinod/a-PyTorch-Tutorial-to-Image-Captioning …
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
Really happy to see you find my image captioning notebook interesting! Here you can watch it in action: https://www.youtube.com/watch?feature=player_embedded&v=i7QnEdn0RZ8 … Full disclosure: this is my implementation of the paper "Show, Attend, and Tell" by Xu et al. 2015, so the idea was not mine, just the implementation!
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.