Efficient Representations are Controllable Representations

• URL: http://arxiv.org/abs/2602.07828v1
• Date: Sun, 08 Feb 2026 05:32:02 GMT
• Title: Efficient Representations are Controllable Representations
• Authors: Charles Ye, Jasmine Cui,
• Abstract summary: Controlling how LLMs internally represent concepts requires sophisticated methods to first identify, then intervene on the model's existing feature geometry.<n>We finetune an LLM with a simple auxiliary loss, training 16 of its 3072 residual stream dimensions to be inert interpretability flags that simply indicate what concepts are required for generation.<n>As a result, these inert flags become genuine internal features: interpretable control switches that allow us to steer generation at inference time.
• Score: 0.0
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: What is the most brute-force way to install interpretable, controllable features into a model's activations? Controlling how LLMs internally represent concepts typically requires sophisticated methods to first identify, then intervene on the model's existing feature geometry. We bypass all of this. We finetune an LLM with a simple auxiliary loss, training 16 of its 3072 residual stream dimensions to be inert interpretability flags that simply indicate what concepts are required for generation. The model reorganizes around them anyway, learning to rely on these flags during actual generation tasks. As a result, these inert flags become genuine internal features: interpretable control switches that allow us to steer generation at inference time. Why does this work? When a feature is reliably supplied at a fixed location, gradient descent gradually eliminates redundant encodings elsewhere, and the model erodes its own alternative representations. A model's efficiency pressure is a lever - exploitable to induce interpretable, controllable representations.

Related papers

• Sparse Semantic Dimension as a Generalization Certificate for LLMs [53.681678236115836]
  We introduce the Sparse Semantic Dimension (SSD), a complexity measure derived from the active feature vocabulary of a Sparse Autoencoder (SAE) trained on the model's layers.<n>We validate this framework on GPT-2 Small and Gemma-2B, demonstrating that our bound provides non-vacuous certificates at realistic sample sizes.
  arXiv  Detail & Related papers  (2026-02-11T21:45:18Z)
• CodeCircuit: Toward Inferring LLM-Generated Code Correctness via Attribution Graphs [13.488544043942495]
  We aim to investigate whether the model's neural dynamics encode internally decodable signals that are predictive of logical validity during code generation.<n>By decomposing complex residual flows, we aim to identify the structural signatures that distinguish sound reasoning from logical failure.<n>Analysis across Python, C++, and Java confirms that intrinsic correctness signals are robust across diverse syntaxes.
  arXiv  Detail & Related papers  (2026-02-06T03:49:15Z)
• SALVE: Sparse Autoencoder-Latent Vector Editing for Mechanistic Control of Neural Networks [0.0]
  We present SALVE, a framework that bridges mechanistic interpretability and model editing.<n>We learn a sparse, model-native feature basis without supervision.<n>We validate these features with Grad-FAM, a feature-level saliency mapping method.
  arXiv  Detail & Related papers  (2025-12-17T20:06:03Z)
• Hyperdimensional Probe: Decoding LLM Representations via Vector Symbolic Architectures [12.466522376751811]
  Hyperdimensional Probe is a novel paradigm for decoding information from the Large Language Models vector space.<n>It combines ideas from symbolic representations and neural probing to project the model's residual stream into interpretable concepts.<n>Our work advances information decoding in LLM vector space, enabling extracting more informative, interpretable, and structured features from neural representations.
  arXiv  Detail & Related papers  (2025-09-29T16:59:07Z)
• GrAInS: Gradient-based Attribution for Inference-Time Steering of LLMs and VLMs [56.93583799109029]
  GrAInS is an inference-time steering approach that operates across both language-only and vision-language models and tasks.<n>During inference, GrAInS hidden activations at transformer layers guided by token-level attribution signals, and normalizes activations to preserve representational scale.<n>It consistently outperforms both fine-tuning and existing steering baselines.
  arXiv  Detail & Related papers  (2025-07-24T02:34:13Z)
• Sparse Autoencoders for Sequential Recommendation Models: Interpretation and Flexible Control [5.045834768309381]
  sparse autoencoders (SAE) have been shown to be a promising unsupervised approach for extracting interpretable features from language models.<n>We show that SAE can be successfully applied to the transformer trained on a sequential recommendation task.<n>We demonstrate that the features learned by SAE can be used to effectively and flexibly control the model's behavior.
  arXiv  Detail & Related papers  (2025-07-16T12:57:43Z)
• Attribution-guided Pruning for Compression, Circuit Discovery, and Targeted Correction in LLMs [15.23174472320989]
  Large Language Models (LLMs) are central to many contemporary AI applications.<n>Recent works in eXplainable AI (XAI) suggest that interpretability can also enable model compression.
  arXiv  Detail & Related papers  (2025-06-16T17:38:36Z)
• Unlearning Isn't Deletion: Investigating Reversibility of Machine Unlearning in LLMs [38.837810490068556]
  Unlearning in large language models (LLMs) aims to remove specified data, but its efficacy is typically assessed with task-level metrics like accuracy and perplexity.<n>We demonstrate that models can appear to forget while their original behavior is easily restored through minimal fine-tuning.<n>This phenomenon of emphreversibility suggests that information is merely suppressed, not genuinely erased.
  arXiv  Detail & Related papers  (2025-05-22T16:02:10Z)
• Analyzing Finetuning Representation Shift for Multimodal LLMs Steering [56.710375516257876]
  We propose to map hidden states to interpretable visual and textual concepts.<n>This enables us to more efficiently compare certain semantic dynamics, such as the shift from an original and fine-tuned model.<n>We also demonstrate the use of shift vectors to capture these concepts changes.
  arXiv  Detail & Related papers  (2025-01-06T13:37:13Z)
• Predicting the Performance of Black-box LLMs through Self-Queries [60.87193950962585]
  Large language models (LLMs) are increasingly relied on in AI systems, predicting when they make mistakes is crucial.<n>In this paper, we extract features of LLMs in a black-box manner by using follow-up prompts and taking the probabilities of different responses as representations.<n>We demonstrate that training a linear model on these low-dimensional representations produces reliable predictors of model performance at the instance level.
  arXiv  Detail & Related papers  (2025-01-02T22:26:54Z)
• Model Surgery: Modulating LLM's Behavior Via Simple Parameter Editing [63.20133320524577]
  We show that editing a small subset of parameters can effectively modulate specific behaviors of large language models (LLMs)<n>Our approach achieves reductions of up to 90.0% in toxicity on the RealToxicityPrompts dataset and 49.2% on ToxiGen.
  arXiv  Detail & Related papers  (2024-07-11T17:52:03Z)
• Interpretations Steered Network Pruning via Amortized Inferred Saliency Maps [85.49020931411825]
  Convolutional Neural Networks (CNNs) compression is crucial to deploying these models in edge devices with limited resources. We propose to address the channel pruning problem from a novel perspective by leveraging the interpretations of a model to steer the pruning process. We tackle this challenge by introducing a selector model that predicts real-time smooth saliency masks for pruned models.
  arXiv  Detail & Related papers  (2022-09-07T01:12:11Z)
• Unsupervised Controllable Generation with Self-Training [90.04287577605723]
  controllable generation with GANs remains a challenging research problem. We propose an unsupervised framework to learn a distribution of latent codes that control the generator through self-training. Our framework exhibits better disentanglement compared to other variants such as the variational autoencoder.
  arXiv  Detail & Related papers  (2020-07-17T21:50:35Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.