Search papers, labs, and topics across Lattice.
5
1
9
5
By making environment design a learnable process, SPADE unlocks a new frontier in self-improvement for language agents, leading to substantial performance gains across diverse tasks.
T2I models falter dramatically in counterfactual scenarios, revealing their dependence on familiar visual patterns rather than true causal reasoning.
MiniMax-M2 proves that massive parameter counts don't always translate to better agentic performance; strategic activation of a smaller subset can unlock frontier-level intelligence.
Forget agents and world models – the future of computing could be learned directly from I/O traces, turning the model itself into the computer.
LLM agents can autonomously outperform fixed evolutionary search by 3-10x on open-ended discovery tasks when given persistent memory, asynchronous collaboration, and heartbeat-based interventions.
