Skip To Main Content
Lucie Flek on Muck Rack

Lucie Flek

As seen in: Wiley Online Library, arXiv
Is this you? As a journalist, you can create a free Muck Rack account to customize your profile, list your contact preferences, and upload a portfolio of your best work. Claim your profile
Back to Lucie Flek’s profile

Articles by Lucie Flek

Computer Science > Computation and Language

|
By David Kaczér, Lucie Flek, Florian Mai
| arXiv
Emergent misalignment (EM) is the phenomenon where fine-tuning a language model on a narrow task leads to harmful behavior in unrelated domains. A leading mechanistic account attributes EM to persona features: latent directions acquired during pre-training that misaligned fine-tuning amplifies.
Open in Who Shared Issue with article?

From Past Outbreaks to Future Threats: Detecting Medical Conspiracy Theories With LLMs and Limited Labels

|
By Ipek Baris Schlicht, Berta Chulvi, Lucie Flek
| Wiley Online Library Verified
1 Introduction Conspiracy theories (CTs) are a type of misinformation that attempts to explain the reasons for societal events as the outcome of malicious actors with hidden agendas (Douglas and Sutton 2023). Medical CTs focus on health-related topics such as causes and consequences of an epidemic, vaccination and public health policies. The spread of medical CTs can have serious consequences, such as increasing vaccine hesitancy and undermining of public health efforts (Ullah et al. 2021).
Open in Who Shared Issue with article?

Computer Science > Computation and Language

|
By Lucie Flek, Florian Mai
| arXiv
Large language models can hide hidden behaviors that activate only under narrow conditions, such as backdoor triggers, sleeper-agent deployment cues, sandbagging, or topic-conditioned censorship. Such behaviors are difficult to detect without prior knowledge what to look for. We present activation-matched finetuning, an unsupervised detection method that assumes no knowledge of the trigger or the target behavior.
Open in Who Shared Issue with article?

Computer Science > Artificial Intelligence

|
By Lucie Flek
| arXiv
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them. Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Open in Who Shared View Duplicates (629) Issue with article?

Computer Science > Computation and Language

|
By Nicholas Correa, Lucie Flek, Florian Mai
| arXiv
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them. Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Open in Who Shared View Duplicates (445) Issue with article?

[2602.16379] Label-Consistent Data Generation for Aspect-Based Sentiment Analysis Using LLM Agents

|
By Lucie Flek, Akbar Karimi, Mohammad H. A
| arXiv
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them. Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Open in Who Shared View Duplicates (99) Issue with article?

[2511.08204] Encoder Fine-tuning with Stochastic Sampling Outperforms Open-weight GPT in Astronomy Knowledge Extraction

|
By Lucie Flek, Akbar Karimi
| arXiv
arXiv:2511.08204 (cs) View PDF HTML (experimental) Subjects: Computation and Language (cs.CL) Cite as: arXiv:2511.08204 [cs.CL] (or arXiv:2511.08204v1 [cs.CL] for this version) https://doi.org/10.48550/arXiv.2511.08204 Submission history From: Akbar Karimi [ view email] [v1] Tue, 11 Nov 2025 13:08:30 UTC (144 KB) Bibliographic Tools Bibliographic Explorer Toggle Bibliographic Explorer () Connected Papers Toggle Connected Papers (What is Connected Papers?) Litmaps Toggle Litmaps (What is...
Open in Who Shared View Duplicates (317) Issue with article?

[2511.07112] More Agents Helps but Adversarial Robustness Gap Persists

|
By Zhastay Yeltay, Lucie Flek, Akbar Karimi
| arXiv
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them. Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Open in Who Shared View Duplicates (695) Issue with article?

IKnow: Instruction-Knowledge-Aware Continual Pretraining for Effective Domain Adaptation

|
By Florian Mai, Lucie Flek
| arXiv
arXiv:2510.20377 (cs) View PDF HTML (experimental) Subjects: Artificial Intelligence (cs.AI); Computation and Language (cs.CL) Cite as: arXiv:2510.20377 [cs.AI] (or arXiv:2510.20377v1 [cs.AI] for this version) https://doi.org/10.48550/arXiv.2510.20377 Submission history From: Tianyi Zhang [ view email] [v1] Thu, 23 Oct 2025 09:21:13 UTC (1,009 KB) Bibliographic Tools Bibliographic Explorer Toggle Bibliographic Explorer () Connected Papers Toggle Connected Papers (What is Connected Papers?)...
Open in Who Shared View Duplicates (373) Issue with article?

[2507.03612] Multi-Hop Reasoning for Question Answering with Hyperbolic Representations

|
By Lucie Flek, Akbar Karimi
| arXiv
Open in Who Shared Issue with article?

[2507.03641] Improving Low-Resource Dialect Classification Using Retrieval-based Voice Conversion

|
By Akbar Karimi, Lucie Flek, Caroline Kleen, Alfred Lameli
| arXiv
Open in Who Shared Issue with article?

[2501.14719] Do LLMs Provide Consistent Answers to Health-Related Questions across Languages?

|
By Ipek Baris, Zhixue Zhao, Burcu Sayin, Lucie Flek
| arXiv
[Submitted on 24 Jan 2025] Title:Do LLMs Provide Consistent Answers to Health-Related Questions across Languages?
Open in Who Shared Issue with article?

[2501.08276] Exploring Robustness of LLMs to Sociodemographically-Conditioned Paraphrasing

|
By Akbar Karimi, Lucie Flek
| arXiv
Open in Who Shared Issue with article?

ArithmAttack: Evaluating Robustness of LLMs to Noisy Context in Math Problem Solving

|
By Zain ul, Lucie Flek, Akbar Karimi, Shahzeb Qamar
| arXiv
Open in Who Shared Issue with article?

[2501.08322] Exploring Robustness of Multilingual LLMs on Real-World Noisy Data

|
By Lucie Flek, Akbar Karimi
| arXiv
Open in Who Shared Issue with article?

[2410.06271] Probing the Robustness of Theory of Mind in Large Language Models

|
By Lucie Flek, Laura Schrewe
| arXiv
Open in Who Shared Issue with article?

[2404.06488] Pitfalls of Conversational LLMs on News Debiasing

|
By Ipek Baris, Defne Altiok, Maryanne Taouk Verified, Lucie Flek
| arXiv
Open in Who Shared Issue with article?

[2308.14641] Challenges of GPT-3-based Conversational Agents for Healthca

|
By Charles Welch, Lucie Flek, Allison Lahnala
| arXiv
arXiv:2308.14641 (cs) Download PDF Comments: Subjects: Computation and Language (cs.CL) Cite as: arXiv:2308.14641 [cs.CL] (or arXiv:2308.14641v1 [cs.CL] for this version) https://doi.org/10.48550/arXiv.2308.14641 Submission history From: Fabian Lechner [ view email] [v1] Mon, 28 Aug 2023 15:12:34 UTC (7,156 KB) Bibliographic Tools Bibliographic Explorer Toggle Bibliographic Explorer () Litmaps Toggle Litmaps (What is Litmaps?) scite.ai Toggle scite Smart Citations (What are Smart Citations?)...
Open in Who Shared View Duplicates (275) Issue with article?

Suicide Ideation Detection via Social and Temporal User Representations using Hyperbolic Learning

|
By Ramit Sawhney, Harshit Joshi, Lucie Flek
| aclweb.org
Suicide Ideation Detection via Social and Temporal User Representations using Hyperbolic Learning Ramit Sawhney, Harshit Joshi, Rajiv Ratn Shah, Lucie Flek Abstract Recent psychological studies indicate that individuals exhibiting suicidal ideation increasingly turn to social media rather than mental health practitioners. Personally contextualizing the buildup of such ideation is critical for accurate identification of users at risk.
Open in Who Shared Issue with article?
Show More