
Data Analytics Lab
@DataLabBE
Data Analytics Laboratory at VUB: Vrije Universiteit Brussel, Belgium
You might like
Onderzoekers over metro 3: 'Amper voordelen bij halve aanleg' bruzz.be/actua/mobilite…
bruzz.be
Onderzoekers over metro 3: 'Amper voordelen bij halve aanleg'
De aanleg van metrolijn 3 zou voor reizigers maar beperkte tijdwinst opleveren, zo blijkt uit data-onderzoek door de VUB. Bij een gedeeltelijke aanleg kan er zelfs sprake zijn van langere reistijden.
The paper shows that simple words in chain of thought text can reliably flag wrong LLM answers. When the model’s reasoning text (the chain of thought) includes words like “guess” or “stuck”, the chance that the final answer is correct goes down a lot, by up to 40%. So put…

New paper: Lexical Hints of Accuracy in LLM Reasoning Chains We ask: Can words in LLM's reasoning trace tell us when it’s wrong? - CoT length predicts accuracy on easier tasks - Lexical cues (guess, stuck, hard) predict errors regardless of task difficulty arxiv.org/abs/2508.15842
Data Lab is heading to ICLR 2025! Our PhD students @FlorianoTori and @MelikaMobini will both be presenting a poster on their work. Check out iclr.cc/virtual/2025/p… and iclr.cc/virtual/2025/3…. If you’ll be there too, feel free to reach out — we’d love to connect!
1/2 New blog post by @v_arne about the Brussels Experiment. Put quantitatively trained people under time pressure, and force them to behave more intuitively. Result: they behave less like expected-value maximizers and more like long-term growth optimizers, EE-style.
Have you ever wondered how time pressure shapes financial decisions? ✅Check out my new blog post, The Brussels Experiment: ergodicityeconomics.com/2025/02/26/the… 📚Full paper: nature.com/articles/s4159…
LMs are getting really good at reasoning, but mechanisms behind it are poorly understood. In our recent paper, we investigated SOTA models and found that 'Thinking harder ≠ thinking longer'! Joint work with @AndresAlgaba1, @VincentGinis Insights of our research (A thread):
Opiniestuk in @destandaard. Humanity's Last Exam - Het is examentijd. Niet alleen voor AI, maar ook voor ons. standaard.be/cnt/dmf2025012… Met @VincentGinis en Brecht Verbeken. @VUBrussel @FWOVlaanderen
Join our team at @DataLabBE! We're excited to announce a new job-opening: a Postdoctoral Fellow in Computational (Social) Science and AI. Ideal for researchers motivated to innovate and work in a highly interdisciplinary environment. Deadline Jan 19. jobs.vub.be/job/Elsene-Pos…
import math free_time_on_cluster = math.factorial(10)/3 happy_holidays = True if free_time_on_cluster > math.factorial(10)/14 and happy_holidays: print("No-one is using the cluster, submit all your jobs and enjoy the holidays!")

Early-bird registration ends this Friday, 20 December. And the program is now complete. This will be absolutely fantastic!

United States Trends
- 1. phil 88K posts
- 2. phan 78.3K posts
- 3. Jorge Polanco 6,213 posts
- 4. Columbus 227K posts
- 5. Falcons 14.4K posts
- 6. Mitch Garver N/A
- 7. Kincaid 1,716 posts
- 8. Doug Eddings 1,007 posts
- 9. Middle East 324K posts
- 10. Jake Moody N/A
- 11. Martin Sheen 1,738 posts
- 12. Go Bills 7,455 posts
- 13. #BillsMafia 5,241 posts
- 14. Springer 11.1K posts
- 15. Yesavage 3,436 posts
- 16. Gilbert 9,870 posts
- 17. Monday Night Football 7,034 posts
- 18. Pentagon 24.2K posts
- 19. Bears 35.3K posts
- 20. Bijan 3,319 posts
Something went wrong.
Something went wrong.