arXiv cs.AI / cs.LG / cs.CL·1d agoPEARL: Adaptive Prefill-Decode Execution with Elasticity for Agentic Reinforcement Learning#agentic-rl#prefill-decode#gpu-schedulingAI research
arXiv cs.AI / cs.LG / cs.CL·4d agoEAServe: Encode-Aware Disaggregated Serving for Multimodal Large Language Models#mllm#inference-serving#disaggregationAI tools & infra 2 sources