# Eliciting Latent Knowledge from Quirky Language Models

arXiv: https://arxiv.org/abs/2312.01037 · HTML: https://mlsanity.com/paper/2312.01037/

## Key facts

- Cited by 62 papers in the map (data through 14 Oct 2026).
- Builds on 24 of its 45 references that are in the map.
- First posted on arXiv 2 Dec 2023.

## Builds on (most cited first, 24 of 24)

- [LoRA: Low-Rank Adaptation of Large Language Models](https://mlsanity.com/paper/2106.09685/) (2021)
- [Direct Preference Optimization: Your Language Model is Secretly a Reward Model](https://mlsanity.com/paper/2305.18290/) (2023)
- [The Curious Case of Neural Text Degeneration](https://mlsanity.com/paper/1904.09751/) (2019)
- [The Pile: An 800GB Dataset of Diverse Text for Language Modeling](https://mlsanity.com/paper/2101.00027/) (2020)
- [Character-level Convolutional Networks for Text Classification](https://mlsanity.com/paper/1509.01626/) (2015)
- [A large annotated corpus for learning natural language inference](https://mlsanity.com/paper/1508.05326/) (2015)
- [Understanding intermediate layers using linear classifier probes](https://mlsanity.com/paper/1610.01644/) (2016)
- [Inference-Time Intervention: Eliciting Truthful Answers from a Language Model](https://mlsanity.com/paper/2306.03341/) (2023)
- [When Not to Trust Language Models: Investigating Effectiveness of Parametric and Non-Parametric Memories](https://mlsanity.com/paper/2212.10511/) (2022)
- [Discovering Latent Knowledge in Language Models Without Supervision](https://mlsanity.com/paper/2212.03827/) (2022)
- [Crowdsourcing Multiple Choice Science Questions](https://mlsanity.com/paper/1707.06209/) (2017)
- [The Geometry of Truth: Emergent Linear Structure in Large Language Model Representations of True/False Datasets](https://mlsanity.com/paper/2310.06824/) (2023)
- [Annotation Artifacts in Natural Language Inference Data](https://mlsanity.com/paper/1803.02324/) (2018)
- [Scalable agent alignment via reward modeling: a research direction](https://mlsanity.com/paper/1811.07871/) (2018)
- [AI safety via debate](https://mlsanity.com/paper/1805.00899/) (2018)
- [Llemma: An Open Language Model For Mathematics](https://mlsanity.com/paper/2310.10631/) (2023)
- [Self-critiquing models for assisting human evaluators](https://mlsanity.com/paper/2206.05802/) (2022)
- [Risks from Learned Optimization in Advanced Machine Learning Systems](https://mlsanity.com/paper/1906.01820/) (2019)
- [The Alignment Problem from a Deep Learning Perspective](https://mlsanity.com/paper/2209.00626/) (2022)
- [Accuracy on the Line: On the Strong Correlation Between Out-of-Distribution and In-Distribution Generalization](https://mlsanity.com/paper/2107.04649/) (2021)
- [Measuring Progress on Scalable Oversight for Large Language Models](https://mlsanity.com/paper/2211.03540/) (2022)
- [Supervising strong learners by amplifying weak experts](https://mlsanity.com/paper/1810.08575/) (2018)
- [Invariance, Causality and Robustness](https://mlsanity.com/paper/1812.08233/) (2018)
- [exBERT: A Visual Analysis Tool to Explore Learned Representations in Transformers Models](https://mlsanity.com/paper/1910.05276/) (2019)

## Built on by (most cited first, 30 of 62)

- [Mechanistic Interpretability for AI Safety -- A Review](https://mlsanity.com/paper/2404.14082/) (2024)
- [Monitoring Reasoning Models for Misbehavior and the Risks of Promoting Obfuscation](https://mlsanity.com/paper/2503.11926/) (2025)
- [Layer by Layer: Uncovering Hidden Representations in Language Models](https://mlsanity.com/paper/2502.02013/) (2025)
- [Scaling Monosemanticity: Extracting Interpretable Features from Claude 3 Sonnet](https://mlsanity.com/paper/2605.29358/) (2026)
- [AI Sandbagging: Language Models can Strategically Underperform on Evaluations](https://mlsanity.com/paper/2406.07358/) (2024)
- [Analyzing the Generalization and Reliability of Steering Vectors](https://mlsanity.com/paper/2407.12404/) (2024)
- [Secret Collusion among AI Agents: Multi-Agent Deception via Steganography](https://mlsanity.com/paper/2402.07510/) (2024)
- [Steering Language Model Refusal with Sparse Autoencoders](https://mlsanity.com/paper/2411.11296/) (2024)
- [Get my drift? Catching LLM Task Drift with Activation Deltas](https://mlsanity.com/paper/2406.00799/) (2024)
- [A Language Model's Guide Through Latent Space](https://mlsanity.com/paper/2402.14433/) (2024)
- [Monitoring Latent World States in Language Models with Propositional Probes](https://mlsanity.com/paper/2406.19501/) (2024)
- [Mechanistic Unlearning: Robust Knowledge Unlearning and Editing via Mechanistic Localization](https://mlsanity.com/paper/2410.12949/) (2024)
- [Does Representation Matter? Exploring Intermediate Layers in Large Language Models](https://mlsanity.com/paper/2412.09563/) (2024)
- [To Steer or Not to Steer? Mechanistic Error Reduction with Abstention for Language Models](https://mlsanity.com/paper/2510.13290/) (2025)
- [Knowledge Overshadowing Causes Amalgamated Hallucination in Large Language Models](https://mlsanity.com/paper/2407.08039/) (2024)
- [RL-Obfuscation: Can Language Models Learn to Evade Latent-Space Monitors?](https://mlsanity.com/paper/2506.14261/) (2025)
- [Representation Engineering for Large-Language Models: Survey and Research Challenges](https://mlsanity.com/paper/2502.17601/) (2025)
- [Token Erasure as a Footprint of Implicit Vocabulary Items in LLMs](https://mlsanity.com/paper/2406.20086/) (2024)
- [Towards Inference-time Category-wise Safety Steering for Large Language Models](https://mlsanity.com/paper/2410.01174/) (2024)
- [Probing the Geometry of Truth: Consistency and Generalization of Truth Directions in LLMs Across Logical Transformations and Question Answering Tasks](https://mlsanity.com/paper/2506.00823/) (2025)
- [LLaMAs Have Feelings Too: Unveiling Sentiment and Emotion Representations in LLaMA Models Through Probing](https://mlsanity.com/paper/2505.16491/) (2025)
- [The Road to Artificial SuperIntelligence: A Comprehensive Survey of Superalignment](https://mlsanity.com/paper/2412.16468/) (2024)
- [Shaping capabilities with token-level data filtering](https://mlsanity.com/paper/2601.21571/) (2026)
- [JailBound: Jailbreaking Internal Safety Boundaries of Vision-Language Models](https://mlsanity.com/paper/2505.19610/) (2025)
- [SafeSteer: Interpretable Safety Steering with Refusal-Evasion in LLMs](https://mlsanity.com/paper/2506.04250/) (2025)
- [Does Transformer Interpretability Transfer to RNNs?](https://mlsanity.com/paper/2404.05971/) (2024)
- [Reasoning Under Pressure: How do Training Incentives Influence Chain-of-Thought Monitorability?](https://mlsanity.com/paper/2512.00218/) (2025)
- [Rethinking LLM-as-a-Judge: Representation-as-a-Judge with Small Language Models via Semantic Capacity Asymmetry](https://mlsanity.com/paper/2601.22588/) (2026)
- [CAReDiO: Cultural Alignment via Representativeness and Distinctiveness Guided Data Optimization](https://mlsanity.com/paper/2504.08820/) (2025)
- [Disentangled Safety Adapters Enable Efficient Guardrails and Flexible Inference-Time Alignment](https://mlsanity.com/paper/2506.00166/) (2025)

## Abstract (arXiv)

Eliciting Latent Knowledge (ELK) aims to find patterns in a capable neural network's activations that robustly track the true state of the world, especially in hard-to-verify cases where the model's output is untrusted. To further ELK research, we introduce 12 datasets and a corresponding suite of "quirky" language models (LMs) that are finetuned to make systematic errors when answering questions if and only if the keyword "Bob" is present in the prompt. We find that, especially in middle layers, linear probes usually report an LM's knowledge independently of what the LM outputs, enabling us to elicit the correct answer despite the model's untruthful output. The best probing method (logistic regression on contrast pairs) recovers 89% of the gap in AUROC between truthful and untruthful contexts, and 75% for questions harder than those used to train the probe. We also find that a mechanistic anomaly detection approach can flag untruthful behavior with 0.95 AUROC. Our results show promise for eliciting reliable knowledge from capable but untrusted models, and facilitates future research empirically investigating ELK methods.