Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Quest: Query-Aware Sparsity for Efficient Long-Context LLM Inference

About

As the demand for long-context large language models (LLMs) increases, models with context windows of up to 128K or 1M tokens are becoming increasingly prevalent. However, long-context LLM inference is challenging since the inference speed decreases significantly as the sequence length grows. This slowdown is primarily caused by loading a large KV cache during self-attention. Previous works have shown that a small portion of critical tokens will dominate the attention outcomes. However, we observe the criticality of a token highly depends on the query. To this end, we propose Quest, a query-aware KV cache selection algorithm. Quest keeps track of the minimal and maximal Key values in KV cache pages and estimates the criticality of a given page using Query vectors. By only loading the Top-K critical KV cache pages for attention, Quest significantly speeds up self-attention without sacrificing accuracy. We show that Quest can achieve up to 2.23x self-attention speedup, which reduces inference latency by 7.03x while performing well on tasks with long dependencies with negligible accuracy loss. Code is available at http://github.com/mit-han-lab/Quest .

Jiaming Tang, Yilong Zhao, Kan Zhu, Guangxuan Xiao, Baris Kasikci, Song Han• 2024

Related benchmarks

TaskDatasetResultRank
Mathematical ReasoningMATH500 (test)
Accuracy76
514
Mathematical ReasoningAIME 2024
Accuracy71.5
370
Long-context Language UnderstandingLongBench
M-Avg48.6
292
Long-context Language UnderstandingLongBench (test)
Average Score37.69
147
Long-context UnderstandingLongBench (test)
Avg Score56.5
136
Long-context UnderstandingLongBench
Overall Average Score50
115
Long-context UnderstandingLongBench v2
Overall Score31.8
109
Long-context UnderstandingLongBench
HotpotQA54.79
82
Long-context Language UnderstandingInfiniteBench
En.Sum28.86
81
Long-context language modelingRULER 16k context
Accuracy (RULER 16K)59.2
72
Showing 10 of 60 rows

Other info

Follow for update