Marco De Nadai (@denadai2) 's Twitter Profile
Marco De Nadai

@denadai2

Research Scientist - Personalization @Spotify previously @Zalando + @FBK_research. #LLMs #recsys #representationLearning #computerVision #humanbehavior

ID: 5935462

linkhttp://www.marcodena.it calendar_today10-05-2007 17:10:10

4,4K Tweet

1,1K Followers

475 Following

Emil (@emilsnotes) 's Twitter Profile Photo

Why do AI models make stuff up? After working with 100+ companies including Docker and Reddit, Inc. we wrote up a brief literature review of what works and what doesn't when it comes to reducing LLM hallucinations. Link: kapa.ai/blog/ai-halluc…

Durk Kingma (@dpkingma) 's Twitter Profile Photo

👇 Great work led by Yushun (Yushun Zhang) introducing Adam-mini, a version of Adam that, surprisingly, reduces Adam's memory requirement by 50% (!), without negatively affecting convergence rates. Please read Yushun's thread for details!

Paul Couvert (@itspaulai) 's Twitter Profile Photo

Meta has just released Llama 3.3 70B which is more powerful than GPT-4o and 25x cheaper. Yes. 70B and better than GPT-4o. This model is also as powerful as the 405B version of Llama 3.1. Open source is really winning at every level.

Meta has just released Llama 3.3 70B which is more powerful than GPT-4o and 25x cheaper.

Yes. 70B and better than GPT-4o.

This model is also as powerful as the 405B version of Llama 3.1.

Open source is really winning at every level.
Marco De Nadai (@denadai2) 's Twitter Profile Photo

I’m excited to share our recent research and achievements around LLMs to create personalized narratives at Spotify! Work done w many of my colleagues including the Prolific Team, Divita Vohra Praveen Chandar Tony Jebara Mounia Lalmas ai.meta.com/blog/spotify-p…

Mahesh Sathiamoorthy (@madiator) 's Twitter Profile Photo

I think the community is excited about DeepSeek v3 not because it's yet another powerful model but because it's a story of human ingenuity in the face of constraints. Despite all the restrictions due to export control and limited budget, the humans of DeepSeek have created a

Spotify News (@spotifynews) 's Twitter Profile Photo

This quarter marks our first full year of profitability. From launching a best-in-class video experience, to bringing fans, creators and art closer together, here’s everything you need to know about $SPOT Q4 2024 earnings: spoti.fi/3Cq3ZlX

This quarter marks our first full year of profitability. From launching a best-in-class video experience, to bringing fans, creators and art closer together, here’s everything you need to know about $SPOT Q4 2024 earnings: spoti.fi/3Cq3ZlX
Paolo Rota (@paolorotaphd) 's Twitter Profile Photo

Un appello importante, per far capire anche a chi non se ne occupa, quanto in Italia siamo indietro sul lato finanziamenti, e soprattutto, su tempi e procedure. scienzainrete.it/articolo/appel… #ricerca

Marco De Nadai (@denadai2) 's Twitter Profile Photo

Sergey Brin suggests that working 60 hours per week, being at least in the office every weekday, is best for the AI team. 😂 working 12 hours per day with the Google hat is the new cool. However, there are alternatives

Sándor Juhász (@sandor_juhasz) 's Twitter Profile Photo

🎉 New publication in PNAS: Urban highways are barriers to social ties pnas.org/doi/10.1073/pn… We illustrate from numerous aspects that highways are physical barriers that cut opportunities for social connections — in the 50 largest metropolitan areas in the US.

Daniel Ek (@eldsjal) 's Twitter Profile Photo

Every year we put out our annual music economics report, Loud & Clear. This year we’re reporting some pretty massive growth numbers for the industry, but still - we always get asked: why doesn’t Spotify pay artists? And where does the money go?

César A. Hidalgo (@cesifoti) 's Twitter Profile Photo

Why is trade important? You might have heard that trade is beneficial without fully understanding why. Here is a simple explanation illustrating the importance of trade that everyone should be able to understand. Imagine a shipwreck that leaves Bruce Wayne and Homer Simpson

Why is trade important?
 
You might have heard that trade is beneficial without fully understanding why. Here is a simple explanation illustrating the importance of trade that everyone should be able to understand.

Imagine a shipwreck that leaves Bruce Wayne and Homer Simpson
Maya Bechler-Speicher (@mayabechlerspei) 's Twitter Profile Photo

If we're truly making progress, why has the SOTA performance on the Enzymes been declining over the years? 🤔 How is it that replacing the input graph structures with a fixed Cayley graph often performs better? And why are practitioners still relying on vanilla GNNs? 🫣

Simone Centellegher (@simonecent) 's Twitter Profile Photo

🚀Job loss disrupts individuals’ mobility and their exploratory patterns🚀 Thanks to a great collaboration with Marco De Nadai, Marco Tonin, Bruno Lepri and Lorenzo Lucchini our latest study is finally out in iScience! doi.org/10.1016/j.isci…

🚀Job loss disrupts individuals’ mobility and their exploratory patterns🚀
Thanks to a great collaboration with <a href="/denadai2/">Marco De Nadai</a>, <a href="/marcotonin/">Marco Tonin</a>, <a href="/brulepri/">Bruno Lepri</a>  and <a href="/LucchiniLorenzo/">Lorenzo Lucchini</a> our latest study is finally out in iScience!

doi.org/10.1016/j.isci…