Search papers, labs, and topics across Lattice.
2
0
5
4
Continual learning for LLM agents hits a wall: scaling models doesn't reliably improve skill generation, and self-feedback can lead to recursive drift.
VLMs can be significantly boosted on embodied tasks by mid-training on a carefully curated subset of VLM data that is highly aligned with the VLA domain, rivaling the performance of much larger models.
