# KV Cache Steering for Controlling Frozen LLMs

arXiv: https://arxiv.org/abs/2507.08799 · HTML: https://mlsanity.com/paper/2507.08799/

## Key facts

- Cited by 11 papers in the map (data through 14 Oct 2026).
- Placed in Efficient reasoning and agent inference, one of 12 directions in Efficient LLM inference and serving.
- Builds on 32 of its 34 references that are in the map.
- First posted on arXiv 11 Jul 2025.

## Placed in

- Efficient LLM inference and serving / [Efficient reasoning and agent inference](https://mlsanity.com/inference/efficient-reasoning-and-agent-inference/)
- Mechanistic interpretability / [Steering and interpreting LLM reasoning](https://mlsanity.com/mechinterp/steering-and-interpreting-llm-reasoning/)

## Builds on (most cited first, 30 of 32)

- [Language Models are Few-Shot Learners](https://mlsanity.com/paper/2005.14165/) (2020)
- [The Llama 3 Herd of Models](https://mlsanity.com/paper/2407.21783/) (2024)
- [Chain-of-Thought Prompting Elicits Reasoning in Large Language Models](https://mlsanity.com/paper/2201.11903/) (2022)
- [Training Verifiers to Solve Math Word Problems](https://mlsanity.com/paper/2110.14168/) (2021)
- [Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks](https://mlsanity.com/paper/1908.10084/) (2019)
- [DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning](https://mlsanity.com/paper/2501.12948/) (2025)
- [Think you have Solved Question Answering? Try ARC, the AI2 Reasoning Challenge](https://mlsanity.com/paper/1803.05457/) (2018)
- [Large Language Models are Zero-Shot Reasoners](https://mlsanity.com/paper/2205.11916/) (2022)
- [PIQA: Reasoning about Physical Commonsense in Natural Language](https://mlsanity.com/paper/1911.11641/) (2019)
- [Qwen2 Technical Report](https://mlsanity.com/paper/2407.10671/) (2024)
- [CommonsenseQA: A Question Answering Challenge Targeting Commonsense Knowledge](https://mlsanity.com/paper/1811.00937/) (2018)
- [Steering Llama 2 via Contrastive Activation Addition](https://mlsanity.com/paper/2312.06681/) (2023)
- [SnapKV: LLM Knows What You are Looking for Before Generation](https://mlsanity.com/paper/2404.14469/) (2024)
- [Phi-4-Mini Technical Report: Compact yet Powerful Multimodal Language Models via Mixture-of-LoRAs](https://mlsanity.com/paper/2503.01743/) (2025)
- [How Far Can Camels Go? Exploring the State of Instruction Tuning on Open Resources](https://mlsanity.com/paper/2306.04751/) (2023)
- [Learning to Compress Prompts with Gist Tokens](https://mlsanity.com/paper/2304.08467/) (2023)
- [SmolLM2: When Smol Goes Big -- Data-Centric Training of a Small Language Model](https://mlsanity.com/paper/2502.02737/) (2025)
- [Function Vectors in Large Language Models](https://mlsanity.com/paper/2310.15213/) (2023)
- [Chain-of-Thought Reasoning Without Prompting](https://mlsanity.com/paper/2402.10200/) (2024)
- [Programming Refusal with Conditional Activation Steering](https://mlsanity.com/paper/2409.05907/) (2024)
- [In-context Autoencoder for Context Compression in a Large Language Model](https://mlsanity.com/paper/2307.06945/) (2023)
- [Improving Instruction-Following in Language Models through Activation Steering](https://mlsanity.com/paper/2410.12877/) (2024)
- [Adaptive Activation Steering: A Tuning-Free LLM Truthfulness Improvement Method for Diverse Hallucinations Categories](https://mlsanity.com/paper/2406.00034/) (2024)
- [ClusterKV: Manipulating LLM KV Cache in Semantic Space for Recallable Compression](https://mlsanity.com/paper/2412.03213/) (2024)
- [Semantics-Adaptive Activation Intervention for LLMs via Dynamic Steering Vectors](https://mlsanity.com/paper/2410.12299/) (2024)
- [LoFiT: Localized Fine-tuning on LLM Representations](https://mlsanity.com/paper/2406.01563/) (2024)
- [I Have Covered All the Bases Here: Interpreting Reasoning Features in Large Language Models via Sparse Autoencoders](https://mlsanity.com/paper/2503.18878/) (2025)
- [Steering Large Language Models using Conceptors: Improving Addition-Based Activation Engineering](https://mlsanity.com/paper/2410.16314/) (2024)
- [Taxonomy, Opportunities, and Challenges of Representation Engineering for Large Language Models](https://mlsanity.com/paper/2502.19649/) (2025)
- [Steering off Course: Reliability Challenges in Steering Language Models](https://mlsanity.com/paper/2504.04635/) (2025)

## Built on by (most cited first, 11 of 11)

- [KV-Skill: Forging Expertise in the Model's Native Language](https://mlsanity.com/paper/2608.05475/) (2026)
- [Beyond Speedup -- Utilizing KV Cache for Sampling and Reasoning](https://mlsanity.com/paper/2601.20326/) (2026)
- [Interpreting and Controlling LLM Reasoning through Integrated Policy Gradient](https://mlsanity.com/paper/2602.02313/) (2026)
- [Memory Inception: Latent-Space KV Cache Manipulation for Steering LLMs](https://mlsanity.com/paper/2605.06225/) (2026)
- [Steering Instruction Hierarchies at Inference Time](https://mlsanity.com/paper/2607.26228/) (2026)
- [Towards Steering without Sacrifice: Principled Training of Steering Vectors for Prompt-only Interventions](https://mlsanity.com/paper/2605.05983/) (2026)
- [DecodeShare: Tracing the Shared Subspace of LLM Decode-Time Decisions](https://mlsanity.com/paper/2607.20469/) (2026)
- [A Causal Model for Locating and Unlocking Sandbagging in Model Organisms](https://mlsanity.com/paper/2608.29461/) (2026)
- [How to Take a Memorable Picture? Empowering Users with Actionable Feedback](https://mlsanity.com/paper/2602.21877/) (2026)
- [Extracting Persona Subspaces Through Iterative Nullspace Projection For Modulation](https://mlsanity.com/paper/2610.04676/) (2026)
- [LLMs program using Reusable and Composable Abstract Representations](https://mlsanity.com/paper/2610.19238/) (2026)

## Abstract (arXiv)

We propose cache steering, a lightweight method for implicit steering of language models via a one-shot intervention applied directly to the key-value cache. To validate its effectiveness, we apply cache steering to induce chain-of-thought reasoning in small language models. Our approach constructs steering vectors from reasoning traces, obtained either from teacher models (e.g., GPT-4o) or existing human annotations, that shift model behavior toward more explicit, multi-step reasoning without fine-tuning or prompt modifications. Experimental evaluations on diverse reasoning benchmarks demonstrate that cache steering improves both the qualitative structure of model reasoning and quantitative task performance. Additional experiments show that the method also scales to larger models and yields further gains on challenging datasets such as GPQA and MATH. Compared to prior activation steering techniques that require continuous interventions, our one-shot cache steering offers substantial advantages in terms of inference latency, hyperparameter stability, and ease of integration with existing inference APIs. Beyond mere reasoning induction, we show that cache steering enables controllable transfer of reasoning styles (e.g., stepwise, causal, analogical), making it a practical tool for behavior-level guidance of language models.