Completed
- does the method work with ARC AGI v2?:
Class Central Classrooms beta
YouTube videos curated by Class Central.
Classroom Contents
Joint Self-Supervised Compression Method for ARC-AGI Problem Solving
Automatically move to the next video in the Classroom when playback concludes
- 1 - Intro:
- 2 - overview of the method:
- 3 - related works HRM/TRM/CompressARC:
- 4 - method overview:
- 5 - interview with author:
- 6 - background of mithil:
- 7 - overview of the intuition:
- 8 - training flow of the method:
- 9 - data augmentation is bad:
- 10 - why are you so interested by ARC?:
- 11 - does the method work with ARC AGI v2?:
- 12 - why so few compression method?:
- 13 - what is joint self supervised learning?
- 14 - explicit vs implicit mdl:
- 15 - connection between mdl and TRM?:
- 16 - is this method a general problem solver?:
- 17 - is compression enough for this benchmark?:
- 18 - architecture choices:
- 19 - how much does the 3D RoPE helps?:
- 20 - how to scale this?:
- 21 - dumbest idea ever:
- 22 - how to remove data augmentation?:
- 23 - about test time training:
- 24 - Conclusion:
