Quest speeds up long-context LLM self-attention by up to 2.23x via query-dependent selection of top-K critical KV cache pages, cutting overall latency by 7.03x with negligible accuracy loss.
Scaling laws of rope-based extrapolation, 2024 b
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2024 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Quest: Query-Aware Sparsity for Efficient Long-Context LLM Inference
Quest speeds up long-context LLM self-attention by up to 2.23x via query-dependent selection of top-K critical KV cache pages, cutting overall latency by 7.03x with negligible accuracy loss.