Search papers, labs, and topics across Lattice.
2
0
5
34
V-Steer boosts the accuracy of language model instruction hierarchies to 92% by intelligently editing cached values without the need for retraining.
LLMs struggle with spatial reasoning in embodied settings and 3D structure identification even when exposed to visual modalities, but fine-tuning smaller models offers a surprisingly effective alternative to brute-force scaling.
