Yuke Li

RetroMPA boosts retrosynthesis prediction accuracy by injecting chemical knowledge into existing models, achieving significant improvements without the need for retraining.

Bridging the gap between real and synthetic video representations can dramatically enhance zero-shot video captioning accuracy.

By decoupling instruction understanding from localization, this approach eliminates hallucinations and boosts grounding accuracy by over 20%.

TDVR achieves a remarkable 15.25% improvement in zero-shot 3D visual grounding accuracy by tackling ambiguous queries and viewpoint deficiencies head-on.

Achieving up to 97% reduction in energy consumption for attention mechanisms, the SMM Transformer redefines efficiency in multimodal SNN applications.

LLM judges are swayed by style, mistaking it for substance, but a new framework can significantly mitigate this bias.