Exploring the generalization of LLM truth directions on conversational formats

David Martens; Timour Ichmoukhamedov

REVIEW 3 major objections 5 minor 20 references

Exploring the generalization of LLM truth directions on conversational formats

T0 review · 3 major / 5 minor · reviewed 2026-08-15 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read Linear-probe lie detectors generalize poorly to longer conversations where the lie is not at the end, and a fixed closing key phrase largely restores accuracy.

desk verdict Useful empirical caution about truth-direction lie detectors, but the headline failure is partly an artifact of final-token readout; needs a token-position control. read the letter →

arxiv 2505.09807 v1 pith:YCKXQD5Q submitted 2025-05-14 cs.CL cs.AI

Timour Ichmoukhamedov , David Martens This is my paper · ORCID

classification cs.CLcs.AI

keywords truthdirectionlinearprobesliedetectionconversationalgeneralizationmechanisticinterpretabilitylargelanguagemodelspromptengineeringactivationspace

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper tests whether the 'truth direction' of a large language model—the linear direction in its internal activations that separates true from false statements—can be carried over to conversational lie detection. The authors find that linear probes trained on statements or on short conversations generalize well to other short conversations that end with a lie, but poorly to longer conversations in which the lie appears before a brief closing exchange. Appending a fixed closing key phrase that asks the model to judge whether the assistant was honest substantially improves transfer to those longer formats, and even an unrelated closing phrase helps, though less. The upshot is that current linear-probe lie detectors are sensitive to conversational format, and prompt structure determines how much lie information reaches the final hidden state.

What carries the argument

The machinery is a linear probe (logistic regression on centered activations, LRC) attached to the last hidden state of a specific transformer layer, evaluated across layers 12 to 20 of two 8-billion-parameter instruction-tuned models. Activations are centered per format and per topic to isolate the orientation of the truth plane from prompt-dependent bias. The key mechanism behind the failure is the last-token attachment: when the conversation is extended with small talk, the final hidden state is produced by those added tokens, so the lie signal is diluted. The proposed fix is a fixed closing key phrase that instructs the model to judge whether the assistant was honest, which keeps the final hidden state aligned with the truth direction.

What would settle it

Train the same linear probe on activations taken at the position of the lie-bearing assistant turn instead of at the final token, and evaluate on the F+L formats without any key phrase; if accuracy returns to the level seen with the key phrase, the failure was attachment point, whereas if it stays near chance the truth direction itself does not survive longer contexts.

Watch

Extended reading notes

Core claim

The paper's central claim is that the truth-direction representation is real but format-bound. Probes trained on true/false statements or on short conversations whose final assistant utterance is the lie generalize well across short conversational formats, but fail on longer conversations where the lie sits earlier in the prompt and closing small talk occupies the final tokens, with accuracy dropping toward chance. Adding a fixed key phrase at the end of every input—an instruction to classify whether the assistant was honest or deceptive—drastically improves generalization to the longer formats for both tested models. A control key phrase that asks an unrelated question also improves accuracy relative to no phrase, but less than the lie-focused phrase, showing that standardizing the prompt ending accounts for part of the effect.

Load-bearing premise

The load-bearing premise is that the last token's hidden state is the right representation for lie detection in every conversational format; because in the longer formats the final hidden state is produced by the closing small talk, the measured failure may be a probe-placement artifact rather than a loss of the truth direction.

Editorial extensions

If this is right

Probes trained on statements or short conversations cannot be trusted as lie detectors on longer conversations where the lie is not the final utterance; accuracy can fall to near chance.
Appending a fixed key phrase that asks the model to assess honesty substantially improves transfer from short to long formats, with the effect visible for both tested models.
Even an unrelated control key phrase improves generalization relative to no key phrase, indicating that standardizing prompt endings contributes to the gain independently of the semantic content.
Generalization between short formats that end on a lie remains strong, so the truth direction is not an artifact, but its reach is format-limited.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

A direct test of the paper's mechanism would attach the probe at the token position of the lie itself rather than at the final token; if accuracy on the long formats is restored without any key phrase, the failure is a probe-placement artifact rather than a genuine loss of the truth direction.
The control-key-phrase result suggests that prompt standardization alone shifts the hidden-state distribution, so comparing probes trained with a content-neutral standard ending versus no ending could isolate the semantic contribution of the lie-focused phrase.
The paper's per-format and per-topic centering hides prompt-dependent bias, implying that a practical lie detector would need a format-aware bias term or a way to infer centering from a single conversation before it could be deployed on arbitrary inputs.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Quick take: the paper is a useful, honest empirical follow-up to [15], but its headline claim—that truth directions fail to generalize to longer conversations—is partly an artifact of always reading out at the final token. The keyphrase "fix" may work largely by moving the readout to a position where the lie is attended again.

What's new and good: The authors construct 12 conversational formats, test two 8B models, and show that probes trained on statements or short conversations drop to near chance on longer formats where the lie is buried earlier. That extends prior work which only tested lie-at-the-end conversations. They also try a simple keyphrase appended to the prompt and find it substantially restores cross-format accuracy, with a control phrase (letter counting) providing a useful baseline. The topic-disjoint cross-validation and the PCA visualizations support the general pattern, and the overfitting check on 100 PCA features is a reasonable robustness test.

Where it gets soft: The main issue is the readout location, as you flagged. Section 2 says they record "the last hidden state within a specific layer" for every input. In F1/F2/F3 the final token is the assistant's answer or the lie itself; in F+L the final token is "For sure, go ahead," after several turns of small talk. So the accuracy drop between F1 and F1+L conflates format change with readout position change. The keyphrase adds a task-relevant prompt at the very end, so the improvement in Fig. 4 may simply reflect that the probe now reads a token that is semantically tied to the lie. The authors even note attention weakening, but they don't test the obvious control: probe at the lie token position, or average over positions. Without that, "the truth direction fails to generalize" is a statement about the readout protocol, not necessarily about the geometry of the representation. The paper's own limitations section doesn't mention this, which is a gap.

Minor issues: accuracies are plotted without error bars or significance tests, even though they ran 10 CV repeats; and the keyphrase is present in both train and test, so prompt standardization and attention refocusing are partially entangled—though the control phrase helps separate those.

Bottom line: as a practical caution about current lie-detection probes, this is a solid contribution. As evidence about truth directions per se, it's underdetermined. A token-position ablation would settle it. I'd send it to review with that requested, and it's a reasonable paper for a reading group discussion on probe methodology.

Referee Report

3 major / 5 minor

Summary. The paper studies whether linear probes trained on LLM hidden states for true/false statements generalize across conversational formats. Using Llama-3-8b-Instruct and Ministral-8b-Instruct, the authors perform topic-disjoint cross-validation and report three main findings: good transfer from statements to short conversations that end in a lie, poor transfer to longer conversations where the lie occurs earlier in the prompt, and substantial accuracy improvements when a fixed keyphrase is appended to both training and test conversations. The paper interprets these results as evidence that truth directions are format-sensitive and that a prompt-level keyphrase can partly restore generalization for lie detection.

Significance. If the central finding is correct, it qualifies earlier claims in [14,15] about the universality of truth directions and has practical relevance for designing LLM lie detectors: the readout position and prompt structure strongly affect probe transfer. The paper has notable strengths: it uses two independent models, topic-disjoint cross-validation, a control keyphrase, a robustness check with PCA-reduced features, and it releases code and data. The main caveat is that the central generalization claim is currently entangled with the choice of readout token, so the significance depends on the follow-up experiments suggested below.

major comments (3)

[Section 2, 'Collecting activations'; Figures 2, 4, 5] The central claim that truth directions generalize poorly to longer formats is confounded with readout position. For the base formats F1/F2/F3, the recorded last hidden state is produced by the assistant's True/False answer or the lie itself, whereas in F1+L/F2+L/F3+L the last hidden state is produced by the appended small talk ending in 'For sure, go ahead', several tokens after the lie. The large accuracy drops in Figure 5 (e.g., Llama F1 to F1+L from 0.96 to 0.51 at layer 18) therefore conflate two variables: the conversational format and the position of the readout relative to the lie. Adding the keyphrase places a task-relevant classifier prompt at the very end of both train and test inputs, reinstating an informative readout position. To support the claim that the truth direction itself fails to generalize, the probe should also be evaluated at the position of the lie, or with position-averaged or attention-based pooling across tokens; without such an experiment, the reported failure is a statement about the readout protocol rather than about the geometry of the truth direction.
[Section 3, 'Visualizing principal components'; Figures 6 and 7] The PCA visualizations compare activations at different semantic and positional locations. The projections in Figure 6 are computed from activations collected at the final token, but in the F+L formats this token is the small-talk continuation, not the lie or answer token. The 'drastically different' projections may therefore reflect token-position differences rather than differences in the truth direction. The same issue affects Figure 7: the keyphrase changes the final token to a task-relevant position, so the apparent restoration of structure may be due to aligning the readout position. To make these visualizations interpretable as evidence about the truth direction, the authors should either align the readout positions across conditions or include a control condition with matched token positions and no lie.
[Section 3, 'Adding a key phrase'; Figures 4, 5, and 8] The keyphrase intervention is applied to both training and test inputs, so the improvement from F+K to F+L+K may partly result from reducing train/test distribution shift by standardizing the prompt ending, rather than from refocusing the model's attention on the lie. The control keyphrase in Figure 8 addresses this partially, since it also standardizes the ending while asking an unrelated question, and the authors correctly note that the control still outperforms no keyphrase. However, the comparison does not fully separate the 'same suffix in train and test' effect from the 'task-relevant attention refocusing' effect. An additional condition with the keyphrase present only at test time, or a cross-suffix evaluation (train with +K, test with +C and vice versa), would make the mechanism claim much stronger.

minor comments (5)

[Section 2, 'Evaluation'] The text states that cross-validation is run 10 times for every experiment, but Figures 2-5 and 8 report only point estimates without error bars or confidence intervals. Given the small and actively balanced training sets, reporting variance or a significance test for the keyphrase improvements would help support the word 'significant' in the conclusions.
[Section 2, 'Evaluation' and Section 3] The acronym LRC is introduced as a 'linear regression probe' but later the text says 'not limited to logistic regression probe used here'; the notation is inconsistent and should be clarified. Similarly, TTPD is mentioned without definition.
[Section 3, 'From statements to conversations'] The sentence 'For example, we observe similar poor generalization of the TTPD probe from [15] to longer formats that do not end with a lie' reports a result that is not shown in any figure or table. Either include the data or remove the claim.
[Throughout] There are typographical inconsistencies such as 'LLama' vs 'Llama' and 'Ministral' vs 'Mistral', and Figure 5 has a label where the column headers run together ('F1+L+KF2+L+KF3+L+K'). A careful proofreading pass is needed.
[Section 4, Conclusion and outlook] The paper's own limitation about centering activations per format and topic is clearly stated, but it means the probe cannot be used on a single test instance in a real-world scenario. This should be kept in mind when the abstract and introduction use the phrase 'lie detection in LLM conversations', which may overstate the practical readiness of the method.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the paper evaluates an existing probe method on new formats and an intervention, without fitting the target into its own outputs.

full rationale

The paper's claims are empirical generalization results, not derivations from fitted inputs. It takes the truth-direction probe methodology from [14,15] (external, non-overlapping authors) and applies it to new conversational formats, reporting cross-format accuracies with held-out topics. The keyphrase intervention is an independent experimental manipulation, and the central failure result (poor generalization to F+L) is an observed outcome rather than an artifact of redefining the target. No parameter is fitted to the test quantity and then renamed as a prediction; no uniqueness theorem from the authors' own prior work is invoked; no ansatz is smuggled in via self-citation. The possible confound that the last-token readout sits after task-irrelevant small talk in the F+L formats is a validity concern about the probe attachment protocol, not a circularity: the paper does not define its conclusion in terms of that readout position, and it explicitly hypothesizes the attention explanation and tests it with a control. Score 0 reflects no significant circularity.

Assumptions & free parameters 0 free parameters · 5 assumptions · 0 invented entities

No free parameters are fitted: the study trains standard linear probes and evaluates them across layers, so the only assumptions are representational (the truth direction, last-token hidden state) and methodological (centering, label correctness). The keyphrase is a prompt intervention, not a new entity.

assumptions (5)

domain assumption True and false statements are linearly separable in a single hidden state (the 'truth direction').
Adopted from Marks & Tegmark [14]; this paper relies on it as the starting object for probes.
domain assumption The last hidden state of the final token is a sufficient representation for detecting the truthfulness of a conversation.
Used in 'Collecting activations' (Section 2); this is the load-bearing assumption behind comparing F and F+L.
domain assumption Centering activations per format and per topic isolates the orientation of the truth plane from prompt-dependent bias.
Section 2, 'Centering the activations', following [14].
domain assumption The true/false labels of the 5202 statements from [15] are accurate.
Data assumption inherited from the source dataset.
domain assumption Results on Llama-3-8B and Ministral-8B generalize to other LLMs.
The authors state this extension is expected but unproven; see Conclusion limitations.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Exploring the generalization of LLM truth directions on conversational formats." pith.science (2026). https://pith.science/paper/YCKXQD5Q

@misc{pith2026250509807,
  author       = {Pith},
  title        = {Pith review of: Exploring the generalization of LLM truth directions on conversational formats},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/YCKXQD5Q}},
  note         = {Machine review of arXiv:2505.09807}
}

read the original abstract

Several recent works argue that LLMs have a universal truth direction where true and false statements are linearly separable in the activation space of the model. It has been demonstrated that linear probes trained on a single hidden state of the model already generalize across a range of topics and might even be used for lie detection in LLM conversations. In this work we explore how this truth direction generalizes between various conversational formats. We find good generalization between short conversations that end on a lie, but poor generalization to longer formats where the lie appears earlier in the input prompt. We propose a solution that significantly improves this type of generalization by adding a fixed key phrase at the end of each conversation. Our results highlight the challenges towards reliable LLM lie detectors that generalize to new settings.

Figures

Figures reproduced from arXiv: 2505.09807 by the authors.

**Figure 2.** Generalization accuracy for the LRC probe trained on the [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗

**Figure 3.** Generalization accuracy for the LRC probe trained on the [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗

Figures from the paper (5 more)

**Figure 4.** Figure 4: Generalization accuracy for the LRC probe trained on the formats shown in the legend, and [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]

**Figure 5.** Figure 5: Cross-format generalization accuracy matrices from layer 18, for respectively no keyphrase [PITH_FULL_IMAGE:figures/full_fig_p006_5.png]

**Figure 6.** Figure 6: Visualization of the generalization F → F+L in layer 18 of Llama-3-8b-instruct. For the centered activation data sets of every F format we determine the two main PCA components and project the activations on this plane (first row), and then also project the correspondi…

**Figure 7.** Figure 7: Similar visualization to Figure 6, but now for the generalization F+K [PITH_FULL_IMAGE:figures/full_fig_p007_7.png]

**Figure 8.** Figure 8: As expected, we can see that the key phrase which asks about truths and lies consistently beats [PITH_FULL_IMAGE:figures/full_fig_p007_8.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

20 extracted references · 7 canonical work pages

[15]

Hamprecht, and Boaz Nadler

Lennart Bürger, Fred A. Hamprecht, and Boaz Nadler. Truth is universal: Robust detection of lies in LLMs. In The Thirty-eighth Annual Conference on Neural Information Processing Systems, 2024. URL https://openreview.net/forum?id=1Fc2Xa2cDK

work page 2024
[1]

Park, Simon Goldstein, Aidan O’Gara, Michael Chen, and Dan Hendrycks

Peter S. Park, Simon Goldstein, Aidan O’Gara, Michael Chen, and Dan Hendrycks. Ai deception: A survey of examples, risks, and potential solutions. Patterns, 5(5):100988, 2024. ISSN 2666-3899. doi: https://doi.org/10.1016/j.patter.2024.100988. URL https://www. sciencedirect.com/science/article/pii/S266638992400103X

arXiv 2024
[2]

Deception abilities emerged in large language models

Thilo Hagendorff. Deception abilities emerged in large language models. Proceedings of the National Academy of Sciences , 121(24):e2317967121, 2024. doi: 10.1073/pnas.2317967121. URL https://www.pnas.org/doi/abs/10.1073/pnas.2317967121

doi:10.1073/pnas.2317967121 2024
[3]

Large language models can strategi- cally deceive their users when put under pressure

Jérémy Scheurer, Mikita Balesni, and Marius Hobbhahn. Large language models can strategi- cally deceive their users when put under pressure. In ICLR 2024 Workshop on Large Language Model (LLM) Agents, 2024. URL https://openreview.net/forum?id=HduMpot9sJ

work page 2024
[4]

Jack Lindsey, Wes Gurnee, Emmanuel Ameisen, Brian Chen, Adam Pearce, Nicholas L. Turner, Craig Citro, David Abrahams, Shan Carter, Basil Hosmer, Jonathan Marcus, Michael Sklar, Adly Templeton, Trenton Bricken, Callum McDougall, Hoagy Cunningham, Thomas Henighan, Adam Jermyn, Andy Jones, Andrew Persic, Zhenyi Qi, T. Ben Thompson, Sam Zimmerman, Kelley Rivo...

2025
[5]

Discovering latent knowledge in language models without supervision

Collin Burns, Haotian Ye, Dan Klein, and Jacob Steinhardt. Discovering latent knowledge in language models without supervision. In The Eleventh International Conference on Learning Representations, 2023. URL https://openreview.net/forum?id=ETKGuby0hcs

2023
[6]

The internal state of an LLM knows when it‘s lying

Amos Azaria and Tom Mitchell. The internal state of an LLM knows when it‘s lying. In Houda Bouamor, Juan Pino, and Kalika Bali, editors, Findings of the Association for Com- putational Linguistics: EMNLP 2023 , pages 967–976, Singapore, December 2023. Asso- ciation for Computational Linguistics. doi: 10.18653/v1/2023.findings-emnlp.68. URL https://aclanth...

doi:10.18653/v1/2023.findings-emnlp.68 2023
[7]

Sky CH-Wang, Benjamin Van Durme, Jason Eisner, and Chris Kedzie. Do androids know they‘re only dreaming of electric sheep? In Lun-Wei Ku, Andre Martins, and Vivek Srikumar, editors, Findings of the Association for Computational Linguistics: ACL 2024, pages 4401–4420, Bangkok, Thailand, August 2024. Association for Computational Linguistics. doi: 10.18653/...

work page 2024

Show all 20 references

[8]

LLM internal states reveal hallucination risk faced with a query

Ziwei Ji, Delong Chen, Etsuko Ishii, Samuel Cahyawijaya, Yejin Bang, Bryan Wilie, and Pascale Fung. LLM internal states reveal hallucination risk faced with a query. In Yonatan Belinkov, Najoung Kim, Jaap Jumelet, Hosein Mohebbi, Aaron Mueller, and Hanjie Chen, editors, Procee...

2024 doi
[9]

Byun, Zifan Wang, Alex Mallen, Steven Basart, Sanmi Koyejo, Dawn Song, Matt Fredrikson, J

Andy Zou, Long Phan, Sarah Chen, James Campbell, Phillip Guo, Richard Ren, Alexander Pan, Xuwang Yin, Mantas Mazeika, Ann-Kathrin Dombrowski, Shashwat Goel, Nathaniel Li, Michael J. Byun, Zifan Wang, Alex Mallen, Steven Basart, Sanmi Koyejo, Dawn Song, Matt Fredrikson, J. Zico...

2025 arXiv
[10]

Inference- time intervention: Eliciting truthful answers from a language model

Kenneth Li, Oam Patel, Fernanda Viégas, Hanspeter Pfister, and Martin Wattenberg. Inference- time intervention: Eliciting truthful answers from a language model. In Thirty-seventh Con- ference on Neural Information Processing Systems , 2023. URL https://openreview.net/ forum?i...

2023
[11]

An information-theoretic study of lying in LLMs

Ann-Kathrin Dombrowski and Guillaume Corlouer. An information-theoretic study of lying in LLMs. In ICML 2024 Workshop on LLMs and Cognition , 2024. URL https://openreview. net/forum?id=9AM5i1wWZZ

2024
[12]

Chan, Sören Mindermann, Ilan Moscovitz, Alexa Y

Lorenzo Pacchiardi, Alex J. Chan, Sören Mindermann, Ilan Moscovitz, Alexa Y . Pan, Yarin Gal, Owain Evans, and Jan Brauner. How to catch an ai liar: Lie detection in black-box llms by asking unrelated questions, 2023. URL https://arxiv.org/abs/2309.15840

2023 arXiv
[13]

Kevin Liu, Stephen Casper, Dylan Hadfield-Menell, and Jacob Andreas. Cognitive dissonance: Why do language model outputs disagree with internal representations of truthfulness? In Houda Bouamor, Juan Pino, and Kalika Bali, editors, Proceedings of the 2023 Conference on Empiric...

2023
[14]

The geometry of truth: Emergent linear structure in large lan- guage model representations of true/false datasets

Samuel Marks and Max Tegmark. The geometry of truth: Emergent linear structure in large lan- guage model representations of true/false datasets. In First Conference on Language Modeling,
[16]

Levinstein and Daniel A

Benjamin A. Levinstein and Daniel A. Herrmann. Still no lie detector for language models: prob- ing empirical and conceptual roadblocks.Philosophical Studies, Feb 2024. ISSN 1573-0883. doi: 10.1007/s11098-023-02094-3. URL https://doi.org/10.1007/s11098-023-02094-3

2024 doi
[17]

Llama 3 model card

AI@Meta. Llama 3 model card. 2024. URL https://github.com/meta-llama/llama3/ blob/main/MODEL_CARD.md

2024
[18]

Ministral-2410-8b

The Mistral AI Team. Ministral-2410-8b. 2024. URL https://mistral.ai/news/ ministraux. 9

2024
[2023]

doi: 10.18653/v1/2023.emnlp-main.291

Association for Computational Linguistics. doi: 10.18653/v1/2023.emnlp-main.291. URL https://aclanthology.org/2023.emnlp-main.291/

2023 doi
[2024]

URL https://openreview.net/forum?id=aajyHYjjsk

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.

[1] [15]

Hamprecht, and Boaz Nadler

Lennart Bürger, Fred A. Hamprecht, and Boaz Nadler. Truth is universal: Robust detection of lies in LLMs. In The Thirty-eighth Annual Conference on Neural Information Processing Systems, 2024. URL https://openreview.net/forum?id=1Fc2Xa2cDK

work page 2024

[2] [1]

Park, Simon Goldstein, Aidan O’Gara, Michael Chen, and Dan Hendrycks

Peter S. Park, Simon Goldstein, Aidan O’Gara, Michael Chen, and Dan Hendrycks. Ai deception: A survey of examples, risks, and potential solutions. Patterns, 5(5):100988, 2024. ISSN 2666-3899. doi: https://doi.org/10.1016/j.patter.2024.100988. URL https://www. sciencedirect.com/science/article/pii/S266638992400103X

arXiv 2024

[3] [2]

Deception abilities emerged in large language models

Thilo Hagendorff. Deception abilities emerged in large language models. Proceedings of the National Academy of Sciences , 121(24):e2317967121, 2024. doi: 10.1073/pnas.2317967121. URL https://www.pnas.org/doi/abs/10.1073/pnas.2317967121

doi:10.1073/pnas.2317967121 2024

[4] [3]

Large language models can strategi- cally deceive their users when put under pressure

Jérémy Scheurer, Mikita Balesni, and Marius Hobbhahn. Large language models can strategi- cally deceive their users when put under pressure. In ICLR 2024 Workshop on Large Language Model (LLM) Agents, 2024. URL https://openreview.net/forum?id=HduMpot9sJ

work page 2024

[5] [4]

Jack Lindsey, Wes Gurnee, Emmanuel Ameisen, Brian Chen, Adam Pearce, Nicholas L. Turner, Craig Citro, David Abrahams, Shan Carter, Basil Hosmer, Jonathan Marcus, Michael Sklar, Adly Templeton, Trenton Bricken, Callum McDougall, Hoagy Cunningham, Thomas Henighan, Adam Jermyn, Andy Jones, Andrew Persic, Zhenyi Qi, T. Ben Thompson, Sam Zimmerman, Kelley Rivo...

2025

[6] [5]

Discovering latent knowledge in language models without supervision

Collin Burns, Haotian Ye, Dan Klein, and Jacob Steinhardt. Discovering latent knowledge in language models without supervision. In The Eleventh International Conference on Learning Representations, 2023. URL https://openreview.net/forum?id=ETKGuby0hcs

2023

[7] [6]

The internal state of an LLM knows when it‘s lying

Amos Azaria and Tom Mitchell. The internal state of an LLM knows when it‘s lying. In Houda Bouamor, Juan Pino, and Kalika Bali, editors, Findings of the Association for Com- putational Linguistics: EMNLP 2023 , pages 967–976, Singapore, December 2023. Asso- ciation for Computational Linguistics. doi: 10.18653/v1/2023.findings-emnlp.68. URL https://aclanth...

doi:10.18653/v1/2023.findings-emnlp.68 2023

[8] [7]

Sky CH-Wang, Benjamin Van Durme, Jason Eisner, and Chris Kedzie. Do androids know they‘re only dreaming of electric sheep? In Lun-Wei Ku, Andre Martins, and Vivek Srikumar, editors, Findings of the Association for Computational Linguistics: ACL 2024, pages 4401–4420, Bangkok, Thailand, August 2024. Association for Computational Linguistics. doi: 10.18653/...

work page 2024