Search papers, labs, and topics across Lattice.
Microsoft
1
0
3
Scaling laws hold for interest modeling: bigger LLMs and more inference-time sampling consistently boost news recommendation quality, and can be distilled into smaller, deployable models.
