科研前沿

这里主要放一些组会时用到的 ppt 和相关资料

目录

• Accelerating LLM Inference Throughput via Asynchronous KV Cache Prefetching
• An Adaptive Vector Index Partitioning Scheme for Low-Latency RAG Pipeline
• AttentionRAG: Attention-Guided Context Pruning in Retrieval-Augmented Generation
• Fast and Live Model Auto Scaling with O(1) Host Caching
• Parametric Retrieval Augmented Generation