Search papers, labs, and topics across Lattice.
8
0
9
Current video MLLMs struggle to grasp fleeting visual events, with top models barely surpassing 39% accuracy on critical momentary tasks.
Current AI agents struggle to reliably rediscover scientific knowledge, with top performers averaging only 21.5 out of a possible score, revealing critical gaps in their research capabilities.
Current omnimodal LLMs that ace offline benchmarks still fumble basic real-time interactions, highlighting a critical gap in their ability to handle streaming audio-visual data.
SkillOpt transforms agent skill development into a reproducible optimization process, achieving state-of-the-art results by treating skills as trainable parameters.
LLM-powered agents can now produce surprisingly strong photographs in complex 3D environments, suggesting a path towards embodied AI with aesthetic awareness.
Model-generated skills can actually hurt agent performance, and bigger models don't necessarily make for better skill extractors or consumers.
Visual degradations can cripple the spatial reasoning abilities of even state-of-the-art MLLMs, but targeted finetuning can restore鈥攁nd even surpass鈥攈uman-level performance.
LLMs can have their personalities surgically altered by tweaking just 0.5% of their neurons, preserving general capabilities while achieving competitive control.
