AI Paper Atlas

Research map / Architectures and efficiency / Quantization

KVmix: Gradient-Based Layer Importance-Aware Mixed-Precision Quantization for KV Cache

Fei Li, Song Liu, Weiguo Wu, Shiqiang Nie, Jinyu Wang

AAAI 2026 · Architectures and efficiency › Quantization · approach: KV cache quantization · 4 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
allocate high bits by gradient-based layer importance instead of randomly choosing layers (ablation)KVmix-k2.19v2.38 · LongBench (avg of 4 models)accuracy loss vs FP16: 4.06% -> 1.67%Improved
add mixed-precision (k2.19v2.38) over uniform 2-bit quantization (ablation)KVmix · LongBench (avg of 4 models)accuracy loss vs FP16: 4.53% -> 1.67%Improved
add recent pivotal context (RPC) full-precision KV retention (ablation)KVmix-k2.19v2.38 · LongBench (avg of 4 models)average accuracy: 3.28% decrease without RPCImproved
use importance-aware mixed precision instead of random layer selection (ablation)KVmix-k2.19v2.38 · GSM8K (Llama-2-7B)accuracy: 11.97 -> 13.25Improved
use KVmix instead of uniform 2-bit per-token quantizationLlama-2-7B KV cache · GSM8K / Wikitext-2accuracy / perplexity: 0.83 -> 13.25 GSM8K; 11089 -> 8.71 pplImproved
increase high-bit layer proportion from 20% to 30% (k2.28v2.56) (ablation)KVmix · LongBench (Llama-2-7B)average score: 33.714 -> 33.814Improved
raise RPC proportion above 20% (ablation)KVmix · accuracyaccuracyNo clear effect
use more than 20-30 profiling prompts (ablation)KVmix profiler · importance analysislayer importance reliabilityNo clear effect
replace KIVI-2bit-r64 with KVmix-k2.19v2.38Llama-2-7B KV cache · LongBenchaverage accuracy loss: -1.50% accuracy lossImproved
use KVmix instead of FP16 KV cacheLlama-2-7B · inference efficiencymemory reduction / throughput: 4.9x memory reduction; 5.32x throughputImproved
use KVmix instead of KVQuant-3bit-1%Llama-2-7B · LongBenchaccuracy: 33.703 -> 33.714No clear effect

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy