# Generalization or Memorization: Dynamic Decoding for Mode Steering

arXiv: https://arxiv.org/abs/2510.22099 · HTML: https://mlsanity.com/paper/2510.22099/

## Key facts

- Cited by 3 papers in the map (data through 14 Oct 2026).
- Placed in Steering and interpreting LLM reasoning, one of 13 directions in Mechanistic interpretability.
- Builds on 35 of its 38 references that are in the map.
- First posted on arXiv 25 Oct 2025.

## Placed in

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

## Builds on (most cited first, 30 of 35)

- [The Llama 3 Herd of Models](https://mlsanity.com/paper/2407.21783/) (2024)
- [Training Verifiers to Solve Math Word Problems](https://mlsanity.com/paper/2110.14168/) (2021)
- [HellaSwag: Can a Machine Really Finish Your Sentence?](https://mlsanity.com/paper/1905.07830/) (2019)
- [TruthfulQA: Measuring How Models Mimic Human Falsehoods](https://mlsanity.com/paper/2109.07958/) (2021)
- [Steering Language Models With Activation Engineering](https://mlsanity.com/paper/2308.10248/) (2023)
- [Deep Learning and the Information Bottleneck Principle](https://mlsanity.com/paper/1503.02406/) (2015)
- [SFT Memorizes, RL Generalizes: A Comparative Study of Foundation Model Post-training](https://mlsanity.com/paper/2501.17161/) (2025)
- [A Comprehensive Overview of Large Language Models](https://mlsanity.com/paper/2307.06435/) (2023)
- [Grokking: Generalization Beyond Overfitting on Small Algorithmic Datasets](https://mlsanity.com/paper/2201.02177/) (2022)
- [Does Learning Require Memorization? A Short Tale about a Long Tail](https://mlsanity.com/paper/1906.05271/) (2019)
- [Towards Understanding Grokking: An Effective Theory of Representation Learning](https://mlsanity.com/paper/2205.10343/) (2022)
- [Open Problems in Mechanistic Interpretability](https://mlsanity.com/paper/2501.16496/) (2025)
- [Generalization or Memorization: Data Contamination and Trustworthy Evaluation for Large Language Models](https://mlsanity.com/paper/2402.15938/) (2024)
- [On Memorization of Large Language Models in Logical Reasoning](https://mlsanity.com/paper/2410.23123/) (2024)
- [A Thorough Examination of Decoding Methods in the Era of LLMs](https://mlsanity.com/paper/2402.06925/) (2024)
- [Grokking as the Transition from Lazy to Rich Training Dynamics](https://mlsanity.com/paper/2310.06110/) (2023)
- [Steering Language Model Refusal with Sparse Autoencoders](https://mlsanity.com/paper/2411.11296/) (2024)
- [How Does Information Bottleneck Help Deep Learning?](https://mlsanity.com/paper/2305.18887/) (2023)
- [Meta-Learning without Memorization](https://mlsanity.com/paper/1912.03820/) (2019)
- [How Do Large Language Models Acquire Factual Knowledge During Pretraining?](https://mlsanity.com/paper/2406.11813/) (2024)
- [Generalization v.s. Memorization: Tracing Language Models' Capabilities Back to Pretraining Data](https://mlsanity.com/paper/2407.14985/) (2024)
- [Grokked Transformers are Implicit Reasoners: A Mechanistic Journey to the Edge of Generalization](https://mlsanity.com/paper/2405.15071/) (2024)
- [EvoCodeBench: An Evolving Code Generation Benchmark Aligned with Real-World Code Repositories](https://mlsanity.com/paper/2404.00599/) (2024)
- [Semantics-Adaptive Activation Intervention for LLMs via Dynamic Steering Vectors](https://mlsanity.com/paper/2410.12299/) (2024)
- [Generalization Bounds: Perspectives from Information Theory and PAC-Bayes](https://mlsanity.com/paper/2309.04381/) (2023)
- [Knowledge Mechanisms in Large Language Models: A Survey and Perspective](https://mlsanity.com/paper/2407.15017/) (2024)
- [What Do Learning Dynamics Reveal About Generalization in LLM Reasoning?](https://mlsanity.com/paper/2411.07681/) (2024)
- [The Pitfalls of Memorization: When Memorization Hurts Generalization](https://mlsanity.com/paper/2412.07684/) (2024)
- [Deep Grokking: Would Deep Neural Networks Generalize Better?](https://mlsanity.com/paper/2405.19454/) (2024)
- [MetaMind: Modeling Human Social Thoughts with Metacognitive Multi-Agent Systems](https://mlsanity.com/paper/2505.18943/) (2025)

## Built on by (most cited first, 3 of 3)

- [One Sample to Rule Them All: Extreme Data Efficiency in Multidiscipline Reasoning with Reinforcement Learning](https://mlsanity.com/paper/2601.03111/) (2026)
- [Deeper is Not Always Better: Mitigating the Alignment Tax via Confident Layer Decoding](https://mlsanity.com/paper/2606.21906/) (2026)
- [Metacognitive Steering: Learning the Structure of Scientific Judgment](https://mlsanity.com/paper/2609.16245/) (2026)

## Abstract (arXiv)

Large Language Models (LLMs) exhibit a troubling duality, capable of both remarkable generalization and brittle, verbatim memorization of their training data. This unpredictability undermines their reliability in high-stakes applications. In this work, we propose a unified framework to understand, identify, and control these distinct reasoning modes. First, we introduce a theoretical model based on the Information Bottleneck (IB) principle, formalizing generalization as the learning of a compressed, task-relevant representation and memorization as a failure to compress. Building on this theory, we develop Dynamic Mode Steering (DMS), a novel inference-time algorithm which comprises two components: (1) a lightweight, causally-grounded linear probe that identifies the model's instantaneous reliance on memorization, and (2) a dynamic activation steering mechanism that nudges the model's computation towards pre-identified generalization circuits. We frame DMS as a form of adaptive, self-contrastive decoding. Experiments on reasoning and faithfulness tasks demonstrate that DMS significantly improves logical consistency and factual accuracy, thereby offering a principled approach to enhancing LLM reliability.