Search papers, labs, and topics across Lattice.
3
0
4
Aggregate performance metrics can obscure critical regressions, but layer-isolated evaluation reveals that individual layers can suffer up to 91% failure rates while the overall score remains stable.
LLM-as-judge misses over 75% of critical defects in multi-turn transaction agents, exposing a fundamental flaw in automated evaluation systems.
Multimodal agents still struggle with game development, solving only ~50% of tasks in a new benchmark, GameDevBench, highlighting the need for better multimodal reasoning in complex software environments.
