TY  - RPRT
TI  - Enabling Disaggregated Multi-Stage MLLM Inference via GPU-Internal Scheduling and Resource Sharing
AU  - Lingxiao Zhao
AU  - Haoran Zhou
AU  - Yuezhi Che
AU  - Dazhao Cheng
PY  - 2025
UR  - https://arxiv.org/abs/2512.17574
ID  - 2512.17574
ER  -