Search papers, labs, and topics across Lattice.
5
10
6
6
Visual transition reasoning, not just diverse data, unlocks significant multimodal LLM improvements, with just 2,000 examples boosting 22 benchmarks.
Genuine agency in AI requires internalized structures, not just engineered workflows, to ensure true autonomy and safety.
Achieve SOTA reasoning performance with a 30B parameter model using up to 95% fewer tokens by explicitly controlling when and how deeply an LLM plans.
Forget rigid game environments – PAN lets you simulate open-world scenarios with language-specified actions and long-term visual coherence, opening the door to more realistic AI training.
LLMs gain a whopping 124% task completion boost when coupled with a world model that enables simulative reasoning, suggesting a path beyond token-by-token autoregression.
