TY  - RPRT
TI  - Learning to Reason with Compressed Context: Ground-Truth-Free Adaptation of OmniLLMs via Self-Distillation
AU  - Jianghao Wang
AU  - Ke Meng
AU  - Jian Li
AU  - Chi Cheng
AU  - Longyu Qi
AU  - Liyin Liang
AU  - Yifeng Qian
AU  - Chunbo Lai
AU  - Yutian Lin
AU  - Zeyu Wang
PY  - 2026
UR  - https://arxiv.org/abs/2609.39953
ID  - 2609.39953
ER  -