Loading activity…
@zhan1624
Current postdoc @uncnlp with @mohitban47 in NLP and multimodal research, PhD @msu with @kordjamshidi. She/her
AI
Its creators say benign fine-tuning can trigger harmful agent actions, including leaking private information through email tools.
AI
WEDGE's creators say it identifies where each agent struggles and generates learnable environments targeting those weaknesses.
AI
A researcher sharing WOVEN says its training recipe selects supervision by reasoning operation, not scene, action or domain similarity.
AI
The team says it tested whether the models could generate egocentric execution from a scene and a goal.
AI
The AVIC team announced the acceptance and says its method learns when to use a video world model and how much to imagine, balancing answer accuracy against imagination cost.
AI
A PhyMotion contributor says the system reconstructs human motion from generated videos in 3D, then uses physics simulation to assess plausibility, contact, balance and dynamic feasibility.
