2 ms·Why Larger Models Learn More: Capacity, Interference, Rare-Task Retention3 points by matt_d 5mo ago