Search papers, labs, and topics across Lattice.
6
0
7
7
IRT can slash safety evaluation costs by up to 99% while revealing critical insights into model behavior that traditional benchmarks miss.
Implicit reasoning can outperform explicit methods in generative recommendation tasks while slashing training costs and boosting inference speed.
LLMs excel in generative recommendation but often fall back on one-hop memorization, risking their potential to generalize effectively.
Closing the gap between fixed retrieval systems and oracle performance in agentic search yields a +26.8% F1 improvement, demonstrating that jointly training the reasoning agent and retrieval system is a game-changer.
Semantic IDs, which drastically reduce ID cardinality while inducing semantic clustering, have demonstrably improved ranking and retrieval in Snapchat's production recommender systems.
LLM-based recommenders can be dramatically improved (up to 109% Recall@5) by using counterfactual rewards and uncertainty-aware scaling within a reinforcement learning framework, enabling flexible adaptation to diverse recommendation scenarios.
