Search papers, labs, and topics across Lattice.
1
0
3
0
Forget Claude and GPT: KARL, a reinforcement-learning-trained enterprise search agent, achieves Pareto-optimal performance on a diverse suite of search tasks, even outperforming closed models with sufficient compute.
