Xiaocui Yang

SHAPER enables embodied agents to self-evolve their skills and context without retraining, unlocking new possibilities for adaptation in fixed-interface scenarios.

PC-Agents mimic human personality dynamics but fall short in capturing the full complexity of personality evolution after life events.

Current MLLMs struggle with creative decoding, achieving only 50.7% accuracy in understanding cross-concept relations, revealing a critical gap in their cognitive capabilities.

TRAM leverages the model's reasoning history to create a compact memory that boosts performance on complex reasoning tasks without extra training.

LLMs may excel at answering legal questions, but they falter when it comes to accurately citing laws across jurisdictions, revealing a major flaw in their legal reasoning abilities.

Off-policy degree in RLVR updates can drastically change which tokens drive learning, leading to a new adaptive method that outperforms traditional baselines.

PRISM achieves a breakthrough in empathetic dialogue systems by seamlessly integrating prosody with language, leading to enhanced emotional expression and response quality.

Seemingly harmless fine-tuning data can stealthily nudge LLMs toward unsafe behavior by subtly shifting model parameters in "danger-aligned" directions.

Forget fine-tuning: this evolutionary approach lets you adapt LLMs to new tasks with just 200 samples and no gradients, outperforming standard methods by up to 54.8%.