# On the Limits of Learned Importance Scoring for KV Cache Compression

arXiv: https://arxiv.org/abs/2601.14279 · HTML: https://mlsanity.com/paper/2601.14279/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Placed in KV cache optimization, one of 12 directions in Efficient LLM inference and serving.
- Builds on 19 of its 28 references that are in the map.
- First posted on arXiv 13 Jan 2026.

## Placed in

- Efficient LLM inference and serving / [KV cache optimization](https://mlsanity.com/inference/kv-cache-optimization/)

## Builds on (most cited first, 19 of 19)

- [Efficient Streaming Language Models with Attention Sinks](https://mlsanity.com/paper/2309.17453/) (2023)
- [Challenging Common Assumptions in the Unsupervised Learning of Disentangled Representations](https://mlsanity.com/paper/1811.12359/) (2018)
- [SnapKV: LLM Knows What You are Looking for Before Generation](https://mlsanity.com/paper/2404.14469/) (2024)
- [H2O: Heavy-Hitter Oracle for Efficient Generative Inference of Large Language Models](https://mlsanity.com/paper/2306.14048/) (2023)
- [KIVI: A Tuning-Free Asymmetric 2bit Quantization for KV Cache](https://mlsanity.com/paper/2402.02750/) (2024)
- [KVQuant: Towards 10 Million Context Length LLM Inference with KV Cache Quantization](https://mlsanity.com/paper/2401.18079/) (2024)
- [Efficiently Scaling Transformer Inference](https://mlsanity.com/paper/2211.05102/) (2022)
- [Quest: Query-Aware Sparsity for Efficient Long-Context LLM Inference](https://mlsanity.com/paper/2406.10774/) (2024)
- [Model Tells You What to Discard: Adaptive KV Cache Compression for LLMs](https://mlsanity.com/paper/2310.01801/) (2023)
- [PyramidKV: Dynamic KV Cache Compression based on Pyramidal Information Funneling](https://mlsanity.com/paper/2406.02069/) (2024)
- [A Metric Learning Reality Check](https://mlsanity.com/paper/2003.08505/) (2020)
- [Ada-KV: Optimizing KV Cache Eviction by Adaptive Budget Allocation for Efficient LLM Inference](https://mlsanity.com/paper/2407.11550/) (2024)
- [A Survey on Efficient Inference for Large Language Models](https://mlsanity.com/paper/2404.14294/) (2024)
- [GEAR: An Efficient KV Cache Compression Recipe for Near-Lossless Generative Inference of LLM](https://mlsanity.com/paper/2403.05527/) (2024)
- [PyramidInfer: Pyramid KV Cache Compression for High-throughput LLM Inference](https://mlsanity.com/paper/2405.12532/) (2024)
- [Reducing Transformer Key-Value Cache Size with Cross-Layer Attention](https://mlsanity.com/paper/2405.12981/) (2024)
- [MiniCache: KV Cache Compression in Depth Dimension for Large Language Models](https://mlsanity.com/paper/2405.14366/) (2024)
- [Dynamic Memory Compression: Retrofitting LLMs for Accelerated Inference](https://mlsanity.com/paper/2403.09636/) (2024)
- [On the State of the Art of Evaluation in Neural Language Models](https://mlsanity.com/paper/1707.05589/) (2017)

## Built on by (most cited first, 1 of 1)

- [What Does Chain-of-Thought Entropy Measure? A Channel Audit of Scaffolding, Routing, and Content](https://mlsanity.com/paper/2609.25039/) (2026)

## Abstract (arXiv)

We investigate learned KV cache compression through Speculative Importance Prediction (SIP), a 1.7M parameter non-query-aware scorer that predicts token importance from KV representations alone. Despite architectural sophistication (multi-horizon lookahead, cross-attention), SIP does not outperform simple baselines, including random selection, across 5 seeds, 4 retention levels, and 3 tasks. Key findings: (1) position-based heuristics (keep first 4 + last N tokens) match or exceed learned approaches; (2) prefill attention provides equivalent signal to complex learned scorers; (3) marginal information in KV representations beyond position and prefill attention appears limited for importance prediction. We hypothesize that circular dependence between future queries and generation trajectories contributes to this difficulty.