CS Paper Radar

Evaluating and Improving Natural Language Autoencoders

Marco Bazzani, Dan Mossing

cs.CL cs.LG ·submitted 2026-11-14 16:57 UTC · v1

Abstract

Natural language autoencoders (NLAs) are a method for generating natural language explanations of LLM activations. We develop evaluations based on auditing tasks that test the ability of NLAs to surface latent knowledge, identify whether a model was misled by a hint, and catch a silent fabrication. We also develop extensions to NLAs that let them use activation information from multiple layers and positions, let us apply them more effectively to long transcripts, and reduce the number of false claims they make. These extensions result in improvements on both our auditing evaluations and auxiliary evaluations adapted from prior work. We also use the evaluations to study how different auditing capabilities change over the course of NLA training, and find that although every evaluation improves early in training, extended training can hurt performance on some of our auditing evaluations.

Comment
23 pages, 18 figures, 2 tables

Closest papers

Nearest neighbours by meaning across the whole index.

Selecting The Most Informative Tokens in Natural Language Autoencoders

cs.CL cs.AI ·2026-09-29 ·Federico Torrielli, Gianluca Barmina, Andrea Blasi Núñez, Amon Rapp, Luigi Di Caro +2 more
similarity 0.71
Abstract

Natural language autoencoders translate a language model's internal activations into readable explanations. Explaining every token position is costly. Which positions should an auditor inspect to understand a potential threat? We study this question across $4.7$ million explanations on prompt injection and concealment. We compare signals from model computation with a ranker trained only on chat structure. Chat structure usually selects more relevant explanations than the computational signals, without requiring a model forward pass for position selection. On three of four datasets, explaining just $5\%$ of positions retains nearly all of the success rate from explaining every position, where success means obtaining an explanation about the threat. The benefit varies with the audit task. We also show that pretrained verbalizers recover words that models have learned to conceal through fine-tuning, without additional verbalizer training. These results identify where auditors can concentrate explanation generation and show that useful explanations can extend beyond the model a verbalizer was trained to describe.

0.16

Training LLMs to Verbalize Evaluation Awareness

cs.CL cs.AIcs.LG ·2026-09-28 ·Usman Anwar, Sahar Abdelnabi, David Krueger
similarity 0.70
Abstract

Evaluation awareness (EA) can cause large language models (LLMs) to behave differently during audits than in deployment, yet measuring and accounting for EA remains challenging. We introduce verbalization training (VT), a method for making LLMs less reticent about verbalizing evaluation awareness while avoiding to supervise the latent belief itself. VT uses a model's spontaneous verbalizations as evidence that awareness is present and truncates each rollout immediately before the verbalization, producing training prefixes at which the model is presumed to be aware. The model is then trained with an RL objective designed to increase verbalization in a calibrated way. Across Qwen3.6-35B-A3B, Kimi K2.6, and Inkling, VT increases verbalized EA by 2.4-2.9 times and transfers to held-out agentic settings, while measured latent EA and behavior remain largely stable. In a causal experiment, we independently implant meta-knowledge about evaluations through synthetic-document fine-tuning and show that VT-induced verbalizations reflect the richer knowledge acquired by the model.

0.23

Understanding Confabulation and Rethinking Reconstruction in Activation Explanations

cs.CL cs.AI ·2026-09-27 ·Gert Lek, Zixuan Xia, Pin-Yu Chen, Lydia Y. Chen
code/data similarity 0.69
Abstract · 18 pages, 9 figures

Natural Language Autoencoders (NLAs) produce unsupervised text explanations of a model's activations: a verbalizer describes an activation and a reconstructor learns to recover it from this text. Under the established point-reconstruction NLA training recipe, explanations become more useful for predicting model behavior while also increasingly introducing unsupported details and exhibiting writing defects. To assess these changes separately, we introduce a standardized evaluation framework for unstructured NLA explanations, measuring information recoverable from explanations, contextual support for their claims, and writing quality. To address confabulation and writing defects, we move beyond predicting a single activation: explanations can distinguish distributions of possible activations even when their means and optimal point-reconstruction rewards are identical. We introduce Flow-NLA, which models the distribution of activations compatible with an explanation and trains the verbalizer using a diffusion likelihood bound. Across Qwen, Gemma, and Apertus, this richer signal retains the utility gains of point reconstruction while curbing the growth of confabulation and writing defects, opening up a direction for improving activation-derived training to encourage more informative, supported, and readable explanations. Code and evaluation prompts will be made publicly available upon acceptance.

0.36

Verbalizing Activation Differences with Natural Language Autoencoders

cs.CL ·2026-10-22 ·Andrea Blasi Núñez, Gianluca Barmina, Federico Torrielli, Thiusius Rajeeth Savarimuthu, Peter Schneider-Kamp +1 more
similarity 0.67
Abstract · Under Review

Activation verbalizers translate the internal representations of large language models into human-readable text. The recently proposed natural language autoencoders (NLAs) learn this by reconstructing activations through a natural language bottleneck. However, it remains unclear how well these methods generalize beyond the setting they were trained for. Here, we study this question for NLAs along four closely related questions: (i) Do NLA verbalizations exhibit compositional structure with respect to their input? (ii) Do NLAs generalize to finetuned versions of the model they were trained for? (iii) Can NLAs recover finetuning objectives in model diffing? (iv) Can NLAs be specialized for model diffing? Across four models from three families, our experiments provide evidence supporting a largely positive answer to all four questions. Specifically, verbalizations of composed analogy activations mostly name the target while rarely mentioning the subtracted word, and an NLA can be reused on lightweight finetunes of its original model without retraining, with reconstruction quality degrading noticeably only for the largest model. Verbalizing activation differences between Taboo model organisms and their original models recovers the concealed secret word for most organisms, and reinforcement learning on activation differences further improves this recovery on held-out organisms. We also find that NLAs can recover the finetuning objective without the finetuned model generating any token, since the most informative token positions precede the user prompt. NLAs do not yet match state-of-the-art model diffing methods on the emergent misalignment detection task, but outperform a blackbox baseline. Our findings suggest that NLAs can meaningfully verbalize differences in activation space, rendering NLAs a promising direction for future work.

0.24

Measuring Activation Control in Large Language Models

cs.AI cs.CL ·2026-08-21 ·Marek Mateusz Kowalski, Joshua Fonseca Rivera, Uzay Macar, David Demitri Africa
benchmarkcode/data similarity 0.66
Abstract · 19 figures, 4 tables. Code: https://github.com/mkobalski/activation-control. Data: https://huggingface.co/datasets/joshycodes/activation-con…

Safe deployment of increasingly capable models will likely come to rely on latent-space monitoring as a complement to behavioral evaluations, especially when evaluation-aware models exhibit scheming or deception. However, if models can also control their own activations, deception could extend into the latent space itself. With this in mind, we introduce the Activation Controllability Benchmark to quantify the extent to which models can modulate their residual stream via natural-language instruction. Across model families and capability levels, we find that most LLMs can control the direction and magnitude of their residual stream activations with some degree of temporal resolution, though performance varies considerably across models. In simple tasks, this level of control can evade activation-based monitoring methods (including linear probes, natural language autoencoders, activation oracles, and the Jacobian lens), albeit imperfectly. These results suggest that control over the activation space itself could become a confound for monitoring as introspective capabilities increase; therefore, we recommend that frontier labs and evaluators track activation controllability in future models.

0.52

EduBehaviors: Assertion-based Schemas for Auditable Coding of Educational Dialogues

cs.CL cs.CYcs.LG ·2026-09-22 ·Julian Bernado, Ana Trindade Ribeiro, Xander Beberman, Susanna Loeb
code/data similarity 0.66
Abstract

Large language models have enabled rapid deployment of language classification and analysis, allowing users to describe complex constructs in natural language, such as pedagogical moves or expressions of learning and engagement, and label them in conversational datasets such as classroom or tutoring transcripts. However, because LLM reasoning is opaque, we have no verifiable, mechanistic insight into why a model chose a label for an utterance, and output differences over time or across models challenge replication. We introduce the EduBehaviors framework to generate interpretable, reliable, and scalable construct annotations in educational dialogue data that retain human agency while leveraging the strengths of generative AI, encoders, and machine learning models. We evaluate the framework on the TalkMoves dataset, predicting the Teacher TalkMoves labels. Our best configuration results in a macro-F1 of 0.764 and 0.756 Cohen's kappa, demonstrating improvements over direct prompting approaches and showing competitiveness with fine-tuned classifiers while gaining the benefit of greater transparency and comparability. To facilitate adoption and operationalization of the EduBehaviors framework in educational dialogue data by other researchers, we also release the EduBehaviors Toolkit with two complementary open-source tools based on the TalkMoves application.

0.31

On the Monitorability of Natural Language Autoencoders: Identification vs Discrimination

cs.LG cs.CL ·2026-10-13 ·Kevin Yandoka Denamganaï
code/data similarity 0.66
Abstract · work in progress, open draft, updated related works

Natural Language Autoencoders (NLAs) are proposed as oversight tools for language models: an Activation Verbaliser (AV) writes an English explanation of an activation vector, and an Activation Reconstructor (AR) maps the explanation back to a vector. Their value for oversight depends on a human monitor being able to tell a true account of an activation from a false one, given the AV's explanation. We formalise NLA training and NLA oversight deployment as two different referential games from Emergent Communication: training is a generative referential game, while oversight is a discriminative, multi-modal, partially-observable referential game played over a many-to-many relation between the monitored language model's activations and the accounts that are true of them. The gap between the two games predicts that AV's explanations will identify which activation they describe without discriminating its true account from closely matched false ones, since training never asks explanations to rule anything out. Using the NLA's own AR as a maximally-sympathetic surrogate monitor model, we experimentally confirm this theoretical insight on one released checkpoint: true accounts beat accounts of other activations in all 320 comparisons, but beat fluent rivals differing in a single claim only 60.2% of the time (standard error 6.7%, chance at 50%, $n = 35$). A meaning-preserving rewrite by an uninvolved model costs only 2.1% of the distance to another activation's account, so the failure is not due to the NLA pair having developed private conventions in the AV's explanations. Our rival accounts test only transcript-level truths, which the monitored language model's conversation settles, so these results bound the easiest case and leave claims about the monitored model's own processing untested. We state conjectures that would make both results consequences of the framework.

0.51

Building Auditable AI Text Detectors with Sparse Autoencoders

cs.CL cs.AI ·2026-10-21 ·Liam Dugan, Callum McDougall, Matthew Denton, Philip Pham, Christine Kaeser-Chen +1 more
top venue similarity 0.65
Abstract · EMNLP 2026

AI-generated text detectors are increasingly deployed in consequential settings---peer review, education, admissions, journalism, and content moderation---where their outputs directly affect individuals. Yet the strongest detectors are fine-tuned neural classifiers whose decisions are opaque, while transparent alternatives (stylometric and linguistic-feature classifiers) underperform them by wide margins. We present a pipeline that converts a fine-tuned neural detector into a classifier that operates on human-readable concepts. We accomplish this by training sparse autoencoders (SAEs) directly on a detector's residual stream, assigning auto-interpretation labels to each SAE feature, and retaining only features whose generated descriptions pass a global fidelity threshold. We then train a lightweight classifier on just the interpretable features. We find that our SAEs recover latent concepts that correspond to highly specific and predictive linguistic "tells" of AI-generated text. Classifiers built on these extracted features are not only highly interpretable but exhibit stronger performance on out-of-distribution texts than either neural or stylometric alternatives.

0.50