Method Drift›KV-cache compression
Superseded baseline#77 of 234 most-superseded
LiveVLM
LiveVLM: Efficient Online Video Understanding via Streaming-Oriented KV Cache and RetrievalKV-cache compression · first seen May 21, 2025
superseded — cited as a baseline and beaten by newer methods
1 papers critique it · 1 beat it on benchmarks
What papers say
Verbatim critique sentences, each from a paper that cites LiveVLM as a baseline.
“it uses a fixed compression ratio throughout the video and relies on first-in-first-out (FIFO) strategy to maintain a constrained memory, which leads to forgetting of earlier information in long videos, even though they might be informative.”
— StreamMem: Query-Agnostic KV Cache Memory for Streaming Video Understanding
Beaten on benchmarks
Head-to-head results where a newer method reports beating LiveVLM. Values are copied from the source paper's tables — verify against the cited paper.
- StreamMem: Query-Agnostic KV Cache Memory for Streaming Video Understanding
StreamMem beats LiveVLM · All [LLaVA-OneVision-7B]
59.4 vs 57.3
- StreamMem: Query-Agnostic KV Cache Memory for Streaming Video Understanding
StreamMem beats LiveVLM · VideoMME Long [LLaVA-OneVision-7B]
50.1 vs 48.8
Newer alternatives
Recent methods in the same sub-problem, not yet superseded in the knowledge base.
- May 21, 2026
- KVCapsuleKVCapsule: Efficient Sequential KV Cache Compression for Vision-Language Models with Asymmetric RedundancyMay 14, 2026
- Decoupled Streaming Cache (DSCache)Decouple and Cache: KV Cache Construction for Streaming Video UnderstandingMay 3, 2026
- May 1, 2026
- Hierarchical Adaptive Eviction (HAE)Hierarchical Adaptive Eviction for KV Cache Management in Multimodal Language ModelsFeb 2, 2026
- Dec 13, 2025
- StreamKVStreamKV: Streaming Video Question-Answering with Segment-based KV Cache Retrieval and CompressionNov 10, 2025