Is SparseVLM superseded?

SparseVLM (KV-cache compression): superseded — cited as a baseline and beaten by newer methods. 1 paper(s) critique it, 2 beat it on benchmarks — #53 of 234 most-superseded. Sub-problem: cluster led by ReKV. Newer alternatives in the same sub-problem include MuKV, KVCapsule, Decoupled Streaming Cache (DSCache), LightKV, Hierarchical Adaptive Eviction (HAE).

KV-cache compression

SparseVLM

Superseded baseline#53 of 234 most-superseded

Superseded — cited as a baseline and beaten by newer methods

1 papers critique it · 2 beat it on benchmarks

What papers say

Verbatim critique sentences, each from a paper that cites SparseVLM as a baseline.

However, these strategies mainly tackle redundancy within visual tokens without addressing the interdependencies between text and visual tokens during multimodal long text generation~huang2025dynamicllava.

— Hierarchical Adaptive Eviction for KV Cache Management in Multimodal Language Models

Beaten on benchmarks

Head-to-head results where a newer method reports beating SparseVLM. Values are copied from the source paper's tables — verify against the cited paper.

LightVLM beats SparseVLM
63.9 vs 51.2
Avg · [Keep 3% image tokens]
LightVLM: Acceleraing Large Multimodal Models with Pyramid Token Merging and KV Cache Compression
HAE-LLaVA (Ours, Retain, 192) beats SparseVLM
61.7 vs 57.6
GQA · [LLaVA-1.5-7B with 192 visual tokens retained]
Hierarchical Adaptive Eviction for KV Cache Management in Multimodal Language Models

What to use instead

Recent methods in the same sub-problem, not yet superseded in the knowledge base — arXiv benchmark leaders, not vetted production recommendations.