Specialists Learn Faster: Knowledge Organization in Pre-training Shapes Knowledge Update Efficiency
Models pretrained on chronological knowledge update faster than shuffled generalists, and a derived law predicts that cost.
2 stories, newest first · one card per story
Models pretrained on chronological knowledge update faster than shuffled generalists, and a derived law predicts that cost.
Researchers model curriculum learning as Wasserstein transport over difficulty distributions, finding curriculum benefits are strongly task- and budget-dependent with no dominant strategy.
Today 1.6K readers · 2.5K views · 779 reads · all time 189K views