Unverified50% confidenceBenchmarkExact time
ChunkKV, proposed by the Hong Kong University of Science and Technology (Guangzhou) team, uses continuous semantic chunks as the basic compression unit, retaining or discarding entire chunks. In the NIAH benchmark test, with the KV cache limited to 128, the LLaMA-3-based ChunkKV achieves an accuracy of 73.8%, exceeding SnapKV's 58.9%.
1
Sources
50%
Confidence
Long-term
Relevance
10/12/2026
First Seen
Sources
Related Entities
Related Claims
VerifiedQuantized KV caches (such as storing KV vectors in FP8 or INT4) can compress the cache size to 1/2 or even 1/4 of the original.71% similarUnverifiedMLA theoretically reduces the GPU memory footprint of KV Cache to 5% to 13% of standard MHA by compressing KV matrices into low-dimensional latent vector caches.68% similarUnverifiedKimi K2在FP16精度下仅模型权重就需要数百GB显存，INT4量化可将显存需求压缩至约四分之一67% similarUnverifiedIt is recommended to keep KV cache precision at 8-bit or 16-bit when VRAM permits, especially in long-context or precise tool-calling scenarios.67% similarVerifiedThe size of the KV cache grows linearly with context length; the KV cache for Qwen 2.5 32B at a 64K context may consume several additional gigabytes of memory.66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1012268API
curl https://kongchang.com/api/v1/knowledge/claims/1012268MCP
get_claim(id=1012268)