←
Vasilis Belagiannis432 posts
Vasilis Belagiannis avatar

Vasilis Belagiannis

@bazile

Professor at @UniFAU. Past: Univ. Magdeburg, Ulm Univ., VGG Oxford University, TU München + Industry. personal account.

◷ Joined February 2009▧ 11 Photos and videos
alphaXiv avatar
⟳ Vasilis Belagiannis retweeted
alphaXiv@askalphaxiv·Oct 24

"A Fix for Softmax and Cross-Entropy" Turns out two of the most fundamental functions in deep learning have numerical bugs that can actually destabilize training. This paper proposes simple fixes to softmax and cross-entropy that prevent tiny gradients from incorrectly rounding to zero, or remaining nonzero when they should vanish, as these errors can cause sudden loss spikes and interfere with knowledge distillation. On small-scale grokking benchmarks, the fixes eliminate Slingshot loss spikes and improve test accuracy from 94.3% to 100% in one setting. They also prevent student models from drifting away from their teachers during distillation, without requiring higher numerical precision. alphaxiv.org/abs/2610.31642

Post attachment
Leandro von Werra avatar
⟳ Vasilis Belagiannis retweeted
Leandro von Werra@lvwerra·Oct 22

The good news: you can mid-train yourself to the frontier! At a fraction of the cost! Without distillation! What we did: We took Qwen3.5 14B and mid- and post-trained it to beat every other open model and perform as well as the very best frontier models. This is no simple toy task or format fixing, we are training the model to understand and reason over DNA sequences. Bio is the next frontier for AI and universal access will be crucial to make progress. The fun part: because we train the model on both natural language and DNA sequences you can actually 'chat to your DNA': Ask it questions, ponder the impact of mutations, or ask it to generate a small sequence. We'll soon see the frontier labs move away from math and cyber towards bio. It already started, and we believe researchers should have access to capable models. We are releasing the full recipe, so you can do it, too! Model: huggingface.co/HuggingFaceBi… Blog: huggingface.co/spaces/Huggin…

0:15
alphaXiv avatar
⟳ Vasilis Belagiannis retweeted
alphaXiv@askalphaxiv·Oct 16

Looking for a fun weekend read? Introducing the Illustrated NeurIPS 🌎 We indexed all 9000+ NeurIPS papers and built a visual way to explore the whole landscape Search your favorite topics, open a cluster, and dive right in Check it out here! alphaxiv.org/conferences/neu…

0:34
Alex Wang avatar
⟳ Vasilis Belagiannis retweeted
Alex Wang@heyyalexwang·Oct 16

how much can models remember? there’s a neat perspective on neural networks that deserves to be more widely known here’s how you could have invented every layer of the transformer from memory (1/n) from my new blog post!

Sara Beery avatar
⟳ Vasilis Belagiannis retweeted
Sara Beery@sarameghanbeery·Sep 9

Today at 9/9 at #ECCV26: 9:50-10:20: I’ll give a keynote at the MELEX Workshop (More EXploration, Less EXploitation, sites.google.com/view/melex-… , located in Malmömässan C1) on “Targeted Scientific Discovery in Ecology” (more below)

The Beerylab is excited to be at #ECCV26 @eccvconf this week! Come find us and learn about our research!! Here is everything we are up to today (9/8) 👇👇👇

Vasilis Belagiannis avatar
Vasilis Belagiannis@bazile·Jun 12

🔥 Invited Speakers: David Forsyth, Danica Kragic, Josef Kittler, Olga Russakovsky, & Jean Ponce!

Vasilis Belagiannis avatar
Vasilis Belagiannis@bazile·Jun 12

Submission Deadline: June 22, 2026 ⏰ Accepted papers will be published in the ECCV proceedings. Submit your manuscript + 1-page review response letter here: 🔗 sites.google.com/view/melex-… (2/3) #ComputerVision #MachineLearning #ECCV #DeepLearning

Vasilis Belagiannis avatar
Vasilis Belagiannis@bazile·Jun 12

CfP: Join us for the 3rd Workshop on More Exploration, Less Exploitation (MELEX 2026) at #ECCV2026 (@eccvconf ) in Malmö! 🇸🇪 Tired of minor SOTA-chasing tweaks? We are looking for bold, highly innovative and exploratory computer vision ideas! (1/3)