What could go wrong when presenting a paper without Ms. Coffee Bean? And this time, it is not ANY paper 😱:
📄 Paper:
Parcalabescu, L., Gatt, A., Frank, A., and Calixto, I. (2021): Seeing Past Words: Testing the Cross-Modal Capabilities of Pretrained V&L Models. Proceedings of the 1st Workshop on Multimodal Semantic Representations (MMSR), 2021, Groningen, Netherlands (Online), Association for Computational Linguistics, p. 1--10
🔗 https://arxiv.org/abs/2012.12352
Data 👉 https://github.com/Heidelberg-NLP/counting-probe
Other relevant videos:
📺 Replacing self-attention with the Fourier Transform: https://youtu.be/j7pWPdGEfMA
📺 Explaining ViLBERT: https://youtu.be/dd7nE4nbxN0
📺 Transformer explained: https://youtu.be/FWFA4DGuzSc
Outline:
00:00 While she is sleeping
00:52 “Multimodal faking”
01:33 Vision and Language models
02:40 How cross-modal are they really?
03:20 Counting probe
05:35 Interpolation
06:52 Fighting Ms. Coffee Bean
Music 🎵 : Pretty Boy - DJ Freedem
-----------------------------------------------------------------------
🔗 Links:
YouTube: https://www.youtube.com/AICoffeeBreak
Twitter: https://twitter.com/AICoffeeBreak
Reddit: https://www.reddit.com/r/AICoffeeBreak/
#AICoffeeBreak #MsCoffeeBean #MachineLearning #AI #research
00:00 While she is sleeping
00:52 “Multimodal faking”
01:33 Vision and Language models
02:40 How cross-modal are they really?
03:20 Counting probe
05:35 Interpolation
06:52 Fighting Ms. Coffee Bean
