๐ Day X - @boltdotnew worldโs largest hackathon!โจ
Proud to share Pink Paws, built with my 9yo niece for @boltdotnewโs #BoltHackathon! A fun PWA app to make crafting easy and social for kids and families.
Turn minutes into purr-fect masterpieces!โจ
โจLinks in replies ๐
1/ Now that we're running out of data, how do you optimally scale multi-epoch pretraining to hundreds of epochs?
Our first paper from Q! q0 trains a population of models, instead of single model that saturates fast, reaching a dramatically lower loss at *every* epoch budget.
w/ @bishmdl76@akshayvegesna@ShmuelBerman
folks not using ck would never understand how simple it is to go from:
- model rate limited
- shift to another model
- auto compaction & context extraction
- resume work