☕🍵🥛
Parallelism
📝 RServe: Overlapping Encoding and Prefill for Efficient LMM Inference
📝 Jupiter: Fast and Resource-Efficient Collaborative Inference of Generative LLMs on Edge Devices