Search papers, labs, and topics across Lattice.
Affiliation:
6
0
11
14
Directly manipulating internal representations allows for effective concept erasure in MM-DiTs without the burden of model tuning.
Achieving state-of-the-art character erasure without sacrificing image fidelity, this method transforms how we handle copyright in AI-generated content.
Structure-grounded reasoning in generative recommendation can dramatically enhance performance, especially when traditional methods struggle with complex item relationships.
Recommender systems resist steering towards long-tail content, revealing a major flaw in their controllability that could impact user experience and algorithmic fairness.
Decoupling fact injection from text generation lets you edit LLMs with greater precision, improving fine-grained question answering without sacrificing overall editing performance.
Escape the flatland of traditional recommender systems: RecBundle uses differential geometry to disentangle user interactions from preferences, opening the door to understanding and mitigating systemic biases.
