bloody-tangerine.bsky.social - Profile | ThreadSky | a Reddit-style client for Bluesky

Such a simple and ingenious method to isolate reasoning from memorization in LLMs. Performance of reasoning models drop significantly evaluated based on multiple choice questions in which the correct answer was replaced with 'None of the others' arxiv.org/abs/2502.12896

submitted 4 days ago • 0 comments

I saw this over and over again, when I worked at TED. Men in tech leadership truly think that they are the first to have any thought that pops into their head, that they are the best equipped to solve any problem in the world, and that their lack of experience or expertise is actually an asset.

submitted 7 days ago • 7 comments

« - Il y a une submersion migratoire. - les faits : c’est faux. - oui, mais les Français ont un sentiment de submersion migratoire. - les faits : c’est faux. - oui, mais il y a un sentiment que les Français ont un sentiment de submersion migratoire. » Les scientifiques : 🤯🤯🤯.

submitted 12 days ago • 85 comments

See here's how to do a headline. fortune.com/2025/02/13/e...

submitted 12 days ago • 248 comments

📣 If you yourself or someone you know is interested in contributing to #ACSAC2025 by joining the PC, you can fill out the following nomination form by 28th February 2025 (self-nomination and nominating someone else is possible): docs.google.com/form...

submitted 13 days ago • 0 comments

USENIX WOOT '25 submissions are due March 11. One month to go - still plenty of time to write a paper about your latest fun offensive security research! Or if you want more guidance to get a paper into shape, up-and-coming track deadline is March 4. CfP at www.usenix.org/conference/w...

submitted 14 days ago • 0 comments

🤔 You’d like to share your research code but not sure how to…? 🔓Very excited to finally share this work led by @nicolasrougier.bsky.social in which we propose a practical set of guidelines for opening code. inria.hal.science/hal-04930405...

submitted 14 days ago • 1 comment

📢 Submission deadline extended to Feb 19th! Take the extra days to polish your work and submit it here: 👉 dimva25r2.hotcrp.com #DIMVA25 #DIMVA

submitted 14 days ago • 0 comments

I was just told that I have to remove “climate” from the title of an ongoing grant if I want to keep it. And publications from that grant cannot include “climate” and other forbidden words. I can’t believe I’m writing this from the United States of America. #AcademicSky

submitted 15 days ago • 402 comments

Cela semble absurde, et j’ai été choqué la première fois que je l’ai compris, mais la France – il n'y a pas d'autre mot – est bien un paradis fiscal pour milliardaire Explications détaillées 🧵

submitted 18 days ago • 41 comments

as the United States initiates the destruction of its world-leading higher ed & research sectors, the rest of the world has a golden opportunity to hire away America's best and brightest! Oh wait <listens to earpiece> I'm being told Canada, the UK, & Europe are dismantling their universities too

submitted 28 days ago • 2 comments

DIMVA decisions were just sent out. Congratulations to all authors who got and accept! If your paper was not accepted, the uASC deadline is in 22 hours (same length+format) and uASC is looking for interesting insights despite flaws in the threat model -> uasc.cc

submitted 28 days ago • 0 comments

As always, if you have had a grant cancelled, an NSF panel cancelled, or in any been affected by the assault on the US's scientific infrastructure, DM me for my Signal #, or email me dan at sequencermag dot com for my #. Anonymity guaranteed.

submitted 29 days ago • 3 comments

CNRS : les chercheurs se rassemblent contre le projet « Key Labs », une mise en compétition des laboratoires pour masquer la baisse des moyens alloués à la recherche. ➡️ l.humanite.fr/Pv1

submitted 30 days ago • 9 comments

Bon écoutez, on vient de recevoir un mail du labo comme quoi du fait qu'il n'y a pas de budget 2025, pour l'instant toutes les dépenses sont bloquées (mission, commandes, recrutement...) même sur "resources propres" (genre contrat ANR). Tout va bien, au moins on est encore payés

submitted 30 days ago • 16 comments

Just to say, the @cnrs.bsky.social recruits roughly 250 people on tenured young researchers positions each year, 25% of these persons are not french. There is no preference for nationals (the application can be written in english and the interview can also take place in english).

submitted 32 days ago • 1 comment

- Are you part of the academic community? 📚 - Want to influence the future of OpenSSL? 💡 Join the conversation and share your insights on roadmap and feature priorities for this critical project. Let your voice be heard! 👉 Participate here: openssl-communities.org/d/sdFYx47G/s... #AcademicSky

submitted 34 days ago • 0 comments

The uASC registration is open now: uasc.cc Also the uASC deadline is approaching: January 27 AoE. We accept papers, posters, and talks. We have conference proceedings. We're interested in any insights broadly around microarchitecture security. Don't miss the deadline, don't miss the conference ;)

submitted 35 days ago • 0 comments

Twice this week I’ve heard from researchers who literally can’t disclose security vulnerabilities to companies, because they refuse to accept them. The companies will only let them go through outside services like HackerOne, where coincidentally you have to sign an NDA.

submitted 41 days ago • 11 comments

Hello 2025! We are organising the next Women in Cryptography (WinC) events (@womenincrypto0) ! If you are a person identifying as woman and want to have a coffee break with people in the community, sign up here: forms.gle/iCyRLbYekh9c...

submitted 39 days ago • 0 comments

Keylabs et conséquences themeta.news/keylabs-et-c...

submitted 42 days ago • 0 comments

Everything's fucked.

submitted 49 days ago • 2 comments

JO du jour : ouverture des concours chercheurs INRIA : - 18 CRCN (2 par centre) : www.legifrance.gouv.fr/jorf/id/JORF... - 15 DR2 (national) : www.legifrance.gouv.fr/jorf/id/JORF... Date limite de candidature : 9 février 2025

submitted 50 days ago • 1 comment

CFP for uASC 25 is still open. We have rolling reviews, and 1 submission is already accepted. If you have interesting results on microarchitecture security (incl. weak threat models or reproducing prior work), check out the CFP at uasc.cc The CFP closes Jan 28

submitted 51 days ago • 1 comment

Eat the rich

submitted 68 days ago • 0 comments

Intel launched the Pentium processor in 1993. Unfortunately, dividing sometimes gave a slightly wrong answer, the famous FDIV bug. Replacing the faulty chips cost Intel $475 million. I reverse-engineered the circuitry and can explain the bug. 1/9

submitted 81 days ago • 17 comments

Il y a beaucoup d'agitation au sujet des annonces d'Antoine Petit, PDG du #CNRS, avec des liens vers des articles sous paywall. Ce qui a été déclaré: actuellement, 46% des effectifs du CNRS sont dans 25% des laboratoires, et le CNRS trouve cela trop "dispersé".

submitted 75 days ago • 7 comments

Congratulations to Zhenghong Wang and Ruby Lee for winning the third #ACSAC2024 Test of Time Award with their paper "Covert and Side Channels Due to Processor Architecture" 👏👏👏 doi.org/10.1109/ACSA...

submitted 75 days ago • 0 comments

The yearly academic security circus concludes. Exponential growth continues with 1146 published papers (vs 947 in '23 and 93 in '05). Essentially, we published 14% of all security papers this year. Two authors cracked 100 papers and 13 published >=10 this year. Details: nebelwelt.net/pubstats/top...

submitted 77 days ago • 0 comments

“82% des personnes en informatique n'ont jamais eu de problème de sexisme, et 18% sont des femmes.” ~ Isabelle Collet.

submitted 77 days ago • 8 comments

Study of 245,000 ppl shows women leaving academia in larger numbers than men at every level, across STEM AND non-STEM disciplines.

submitted 78 days ago • 1 comment