Completed
- Intro
Class Central Classrooms beta
YouTube playlists curated by Class Central.
Classroom Contents
Author Interview - Transformer Memory as a Differentiable Search Index
Automatically move to the next video in the Classroom when playback concludes
- 1 - Intro
- 2 - Start of Interview
- 3 - How did this idea start?
- 4 - How does memorization play into this?
- 5 - Why did you not compare to cross-encoders?
- 6 - Instead of the ID, could one reproduce the document itself?
- 7 - Passages vs documents
- 8 - Where can this model be applied?
- 9 - Can we make this work on large collections?
- 10 - What's up with the NQ100K dataset?
- 11 - What is going on inside these models?
- 12 - What's the smallest scale to obtain meaningful results?
- 13 - Investigating the document identifiers
- 14 - What's the end goal?
- 15 - What are the hardest problems currently?
- 16 - Final comments & how to get started
