Results show that reliable verification improves accept-or-rethink decisions and reduces unnecessary reasoning turns, enabling more reliable and efficient self-verification without an external judge.
Zi-Quan Zhu, Han-Ruo Zhu, Si-Yuan Lu et al.· 0 citations
Current AI agents compellingly describe slides. However, AI-assisted slide editing requires more than understanding: the output must retain layout, style, component structure, and native editability. Towards, AI-assisted slide editing, existing agents operate on screenshots or weak document representations and often fr...
Hao-Zhen Zheng, Fu-Lin Wang, Tianhu Xiong et al.· 0 citations
This work identifies what must be preserved beyond the replaced value when using a KV cache as updatable memory and recovers distinctions missed by fixed cache similarities on synthetic record pairs.
Chang-Hai Zhou, Yu-Hua Zhou, Shi-Yang Zhang et al.· 0 citations
TTE-Flash-2B is introduced, a reasoning-aware multimodal representation model that outperforms its explicit-CoT counterpart on the MMEB-v2 benchmark, while producing latent think tokens that are interpretable both textually and visually.
Self-Verified Reasoner (SVR-R1), a multi-turn RL framework that turns a model's own verification into a learning signal for multimodal reasoning, is introduced, offering a simple yet effective recipe for bootstrapping multimodal reasoning.