Search papers, labs, and topics across Lattice.
3
0
6
2
AgentDoG 1.5 proves you can achieve GPT-5.4-level agent safety with open-source models trained on just 1k samples, slashing deployment overhead by two orders of magnitude.
Humans miss nearly 4% of opportunities to leverage correct AI suggestions in question answering, while blindly trusting AI errors almost 2% of the time, revealing a significant gap in effective human-AI collaboration.
Code-executing agents can autonomously generate new, solvable math problems that are harder than existing ones, offering a scalable solution to the bottleneck of high-quality training data for advanced LLMs.
