Search papers, labs, and topics across Lattice.
University of Chinese Academy of Sciences
4
0
7
Skills-Coach offers a training-free approach to significantly improve the performance of LLM-based agent skills, proving that even without additional training data, substantial gains are possible.
Editing an LLM's personality doesn't have to break its brain: DPN-LE precisely controls personality traits by intervening on just 0.5% of neurons, preserving reasoning abilities that bulk neuron editing destroys.
Unlock high-fidelity multimodal image generation and editing with a surprisingly simple framework that leverages pre-trained VLMs and diffusion models, bypassing complex fusion architectures.
Telecom World Models fuse the flexibility of LLMs with the fidelity of Digital Twins, enabling uncertainty-aware predictive planning that existing approaches can't match.