وبلاگ تحقیقات هوش مصنوعی

Yiyang Feng، Biddut Sarker Bijoy، Niranjan Balasubramanian، Jiawei Zhou

Large language model (LLM) agents can induce skills from completed tasks and reuse them later to grow more capable with experience. In practice, induced skills may transfer unreliably and can even harm the agent that ret…

Jianghai Li، Pavel Kuznetsov، Yury Yanovich، Konstantin Nott-Whaley و 1 نفر دیگر

لطفاً متن مورد نظر را ارسال کنید.

Bo Liu، Simon Yu، Yiding Jiang، Ao Qu و 14 نفر دیگر

Continuous self-improvement requires an ever-expanding pool of self-generated, diverse, adaptive goals. For language agents, existing training environment pools (hand-curated, statically synthesized, or frozen-verifier)…

Jayjun Lee، Jessica Yin، Asif Rana، Nicholas Blauch و 6 نفر دیگر

We introduce Accelerating Dexterity via Pre-Training (ADEPT), a large-scale reinforcement learning (RL) framework for learning sim-to-real transferable dexterity across high degree-of-freedom (DoF) robot embodiments that…

Zhu Zhang، Jixun Wang، Xiaoang Xu، Xiaorong Wang و 5 نفر دیگر

On-policy distillation (OPD) trains a student on its own responses using dense token-level guidance from a stronger teacher. In long-context tasks, however, token-level teacher support can favor locally plausible respons…

Aditya Bhattacharjee، Christos Plachouras، Sungkyun Chang، Emmanouil Benetos

This technical report describes our winning submission to the AES AIMLA 2025 Challenge on querying sound effects by vocal imitation. We investigate two complementary fine-tuning strategies: contrastive learning with a fr…