Search papers, labs, and topics across Lattice.
Affiliation:
3
0
4
Parameter scaling yields virtually zero gain on familiar vehicle function calls—with a 270M model matching a 1.7B model—revealing that the tool-representation format, not model size, fundamentally dictates generalization and hallucination rates.
WidgetGen achieves superior visual reconstruction performance without the constraints of fixed UI schemas, redefining the widget-to-code generation landscape.
A simple screenshot-based judge often outperforms complex LLM evaluation methods, challenging assumptions about the necessity of intricate judging pipelines.
