Carsten Schnober

Analysis: Geschichten aus der Geschichte - Carsten Schnober Software Engineer, NLP and Machine Learning Research, Information Retrieval and Search

@geschichte.fm
Ich habe eine interaktive Visualisierung der Named Entities von euren Folgenbeschreibungen gemacht, als kleinen Dank für euren schönen Podcast:
carschno.github.io/gag/

OPEN CALL! We're thrilled to be announcing the 2025 Open eScience Call. If you are a researcher looking for support from a Research Software Engineer (RSE) for your research project, submit your proposal by 13 October. Learn more here: www.esciencecenter.nl/calls-for-pr...

Could AI be (in) FAIR? Open Science is driven by FAIR principles. Made for data, adapted for software, how can they survive the current wave of "Artificial…

In our newest blog, Carsten reminds us: FAIR principles still matter — maybe more than ever. AI can sound smart, but if we can’t find, check or reuse what it spits out, is it science or just a fancy remix? Read the full breakdown here:

Why Superintelligent AI Isn’t Taking Over Anytime Soon Despite claims from top names in AI, researchers argue that fundamental flaws in reasoning models mean bots aren’t on the verge of exceeding human smarts.

"A primary requirement for being a leader in AI... is to be a herald of the impending arrival of our digital messiah: superintelligent AI... Before you get nervous about all the times you were rude to Alexa, know this: A growing cohort of researchers...aren’t buying [it]" www.wsj.com/tech/ai/arti...

Zarafa, le bel animal du roi | Adriana Bogdanova - Amsterdam Art Bradwolff Projects presenteert een nieuwe tentoonstelling van Adriana Bogdanova, die het verhaal van Zarafa, de Soedanese giraffe die in 1826 aan Frankrijk werd geschonken, onderzoekt. Haar komst vero...

@geschichte.fm Eine Ausstellung zum Thema eurer Giraffen-Folge (www.geschichte.fm/archiv/gag440/).
amsterdamart.com/agenda/zaraf...

"Data Morgana aflevering 6: digitale autonomie":
video.publicspaces.net/w/2b324Y9Dtj...
@publicspaces.net

PublicSpaces Conferentie in Amsterdam, June 2025: Shaping our Digital Future & Hyperlink: conference.publicspaces.net

The Carpentries Lab: Introduction to Deep Learning with Python We are delighted to announce the addition of a new community-developed lesson on deep learning to The Carpentries Lab. The curriculum was peer reviewed in the Lab Reviews repository and approved for i...

Our introduction into neural networks and deep learning has been approved by @carpentries.carpentries.org. No AI hype.

carpentries.org/blog/2025/03...

Language Modeling: The First 100 Years ChatGPT has become the symbol for a new wave of Artificial Intelligence. New models with even stronger conversational skills are released…

Language Modeling: The First 100 Years
medium.com/p/357556816148

A Different Game To build complex software, highly skilled programmers are essential. But only effective collaboration enables them to achieve big things.

A Different Game: collaborative software development

blog.esciencecenter.nl/a-different-...

Ein Artikel von mir über OpenAI, ChatGPT (in German):
www.golem.de/news/trainin...
@golemde.bsky.social

Introduction to Deep Learning This workshop gives an introduction to deep learning for researchers who are familiar with the basics of (non-deep) machine learning.

Learn the basics of deep learning, including deep neural networks, data preparation, and convolutional layers during our (deep learning) workshop on 28-29 January in Amsterdam Science Park! https://buff.ly/3BZTMw6..

Can you share the model fine-tuned on MS-MARCO?
Also, would you elaborate on how the token embeddings are aggregated? Mean pooling as in SentenceBert?

The Syllabus The good content is already here; it’s just not evenly distributed

Hot tip: check out, if you can subscribe to: www.the-syllabus.com

They slog through, curate, and surface worthwhile writing etc, delivered weekly out of the ocean of internet slop. Such a truly useful resource in this scattered, threadbare info environment ❤️🙏

In particular, a replacement for SentenceBert. Which is even cooler!

Finally, a Replacement for BERT: Introducing ModernBERT We’re on a journey to advance and democratize artificial intelligence through open source and open science.

Links:
- Blogpost: huggingface.co/blog/modernb...
- ModernBERT-base: huggingface.co/answerdotai/...
- ModernBERT-large: huggingface.co/answerdotai/...
- Paper: arxiv.org/abs/2412.13663
- Transformers Documentation: huggingface.co/docs/transfo...

The link seems to be broken... 😢

"Not being important is not the same as being mediocre."

Last week, several colleagues from the eScience Center attended the @odissei.bsky.social conference in Utrecht - a conference that brings together over 400 participants to connect, share ideas and explore cutting-edge research in computational social sciences.

HSN conferentie 2024 Op 22 en 23 november vond de inmiddels 37ste HSN Conferentie Onderwijs Nederlands plaats in Brussel. Karolien Vertriest en ik waren de leiders van de kolom Digitale geletterdheid & didactiek. Kijk hier naar de presentaties www.jeroenclemens.nl/hsn-conferen...

Techdenkers: in conversation with Meredith Whittaker #DeBalie YouTube video by De Balie

Terugkijken 🍿Techdenkers vanuit Amsterdam met Signal's President @meredithmeredith.bsky.social over #Signal, #privacy, #surveillance, #nonprofit, #cybersecurity, #encryptie, #bigtech, #Telegram #Matrix

www.youtube.com/watch?v=lMFY... via @debalie.bsky.social

#signal #signalapp #signalmessenger

Amsterdam as seen by a weather model Visualizing local climate zones in 3D with the Cityblocks package

Can we use weather models to render a video-game-like image of our city? The answer is yes! Check out how in today's blog post by Peter Kalverla.
medium.com/@peter.kalverla/b609a340cde4?source=friends_link&sk=c222c50d99e817c3d8211a6753db49df

I saw @meredithmeredith.bsky.social yesterday @debalie.bsky.social, brilliant!
There was no time, though, to ask a slightly tangential question: how do their objectives impact the way the (tech) team of @signal.org works, compared to generic tech companies?

🔍 The full program for #CHR2024 is now available! Join us for cutting-edge computational humanities research and discussions. See the complete schedule at: 2024.computational-humanities-research.org/programme/

Our proceedings are out! 🚨 They include no less than 78 papers that reflect the diverse and innovative research happening within the Computational Humanities. Take a dive here: ceur-ws.org/Vol-3834/

@emilymbender.bsky.social do you (or anyone else) have a tip for a paper or talk specifically addressing the short-comings and risks of RAG? You address it in your great ChatGP-why talk, but I am looking for something more specific to RAG.
Any hints or references?