Skip To Main Content
Sebastian Farquhar on Muck Rack

Sebastian Farquhar

As seen in: HuffPost, Nature, arXiv, Mary Ann Liebert, Inc.
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 Sebastian Farquhar’s profile

Articles by Sebastian Farquhar

Computer Science > Machine Learning

|
By Sebastian Farquhar
| arXiv
Open in Who Shared Issue with article?

Detecting hallucinations in large language models using semantic entropy - Nature

|
By Sebastian Farquhar
| Nature Verified
Abstract Large language model (LLM) systems, such as ChatGPT1 or Gemini2, can show impressive reasoning and question-answering capabilities but often ‘hallucinate’ false outputs and unsubstantiated answers3,4. Answering unreliably or without the necessary information prevents adoption in diverse fields, with problems including fabrication of legal precedents5 or untrue facts in news articles6 and even posing a risk to human life in medical domains such as radiology7.
Open in Who Shared Issue with article?

Challenges with unsupervised LLM knowledge discovery

|
By Sebastian Farquhar, Vikrant Varma, Zachary Kenton, Johannes Gasteiger
| huggingface.co
Papers Published on Dec 15 · Featured in Daily Papers on Dec 18 Abstract We show that existing unsupervised methods on large language model (LLM) activations do not discover knowledge -- instead they seem to discover whatever feature of the activations is most prominent. The idea behind unsupervised knowledge elicitation is that knowledge satisfies a consistency structure, which can be used to discover knowledge.
Open in Who Shared Issue with article?

Tracr: Compiled Transformers as a Laboratory for Interpretability

|
By János Kramár, Sebastian Farquhar, Matthew Rahtz, Thomas McGrath
| arXiv
arXiv:2301.05062 (cs) [Submitted on 12 Jan 2023 (v1), last revised 3 Nov 2023 (this version, v5)] Download PDF Comments: Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Machine Learning (stat.ML) Cite as: arXiv:2301.05062 [cs.LG] (or arXiv:2301.05062v5 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2301.05062 Submission history From: David Lindner [ view email] [v1] Thu, 12 Jan 2023 14:59:19 UTC (1,164 KB) [v2] Tue, 21 Feb 2023 10:51:03 UTC (1,163 KB) [v3] Wed,...
Open in Who Shared View Duplicates (70) Issue with article?

Stochastic Batch Acquisition: A Simple Baseline for Deep Active Learning

|
By Sebastian Farquhar, Parmida Atighehchian, Andrew Jesson, Frederic Branchaud-Charron
| arXiv
arXiv:2106.12059 (cs) [Submitted on 22 Jun 2021 (v1), last revised 19 Sep 2023 (this version, v3)] Download PDF Comments: Subjects: Machine Learning (cs.LG); Machine Learning (stat.ML) Cite as: arXiv:2106.12059 [cs.LG] (or arXiv:2106.12059v3 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2106.12059 Submission history From: Andreas Kirsch [ view email] [v1] Tue, 22 Jun 2021 21:07:50 UTC (57 KB) [v2] Fri, 28 Jan 2022 11:36:57 UTC (900 KB) [v3] Tue, 19 Sep 2023 21:20:38 UTC (2,379 KB)...
Open in Who Shared View Duplicates (271) Issue with article?

Pricing Externalities to Balance Public Risks and Benefits of Research

|
By Sebastian Farquhar, Andrew Snyder-Beattie, Owen Cotton-Barratt
| Mary Ann Liebert, Inc.
In 2012, multiple research groups conducted experiments that produced strains of H5N1 avian influenza that were airborne transmissible between mammals.1,2 Given that this disease is often fatal in humans, these particular “gain-of-function” experiments sparked wide controversy*. Advocates of the research argued that the experiments were necessary to improve our understanding of the virus, thus enabling better disease surveillance and vaccine production.
Open in Who Shared Issue with article?
Show More