4 papers
High-accuracy Low-Bit KV-Cache Quantization via Local Distribution Restoration
Gradwell Dzikanyanga, Yanqi Pan, Weihao Yang +3
Long-context large language model inference relies on the KV cache to avoid redundant attention computation, but incurs high memory and bandwidth overheads. Low-bit KV-cache quanti…
TTKV: Temporal-Tiered KV Cache for Long-Context LLM Inference
Gradwell Dzikanyanga, Weihao Yang, Hao Huang +4
Key-value (KV) caching is critical for efficient inference in large language models (LLMs), yet its memory footprint scales linearly with context length, resulting in a severe scal…
ProphetKV: User-Query-Driven Selective Recomputation for Efficient KV Cache Reuse in Retrieval-Augmented Generation
Shihao Wang, Jiahao Chen, Yanqi Pan +7
The prefill stage of long-context Retrieval-Augmented Generation (RAG) is severely bottlenecked by computational overhead. To mitigate this, recent methods assemble pre-calculated…
HybridEP: Scaling Expert Parallelism to Cross-Datacenter Scenario via Hybrid Expert/Data Transmission
Weihao Yang, Hao Huang, Donglei Wu +6
Mixture-of-Experts (MoE) has become a popular architecture for scaling large models. However, the rapidly growing scale outpaces model training on a single DC, driving a shift towa…