David Vos (@dja_vos) profile picture

Information retrieval and generative modelling. PhD candidate @irlab_amsterdam with @mdr and @HarrieOos. Prev. @TechAtBloomberg, @INDE_LAB_AMS and @DPGMediaAI.

Amsterdam, The Netherlands
Joined December 2015
Photo by David Vos (@dja_vos)Photo by David Vos (@dja_vos)Photo by David Vos (@dja_vos)Photo by David Vos (@dja_vos)Photo by David Vos (@dja_vos)Photo by David Vos (@dja_vos)Photo by David Vos (@dja_vos)Photo by David Vos (@dja_vos)Photo by David Vos (@dja_vos)Photo by David Vos (@dja_vos)
Had an extremely inspiring day presenting my work yesterday at #NeurIPS2022 @TrlWorkshop. Big thanks to the organisers and attendees. We also won the best paper runner-up award with our work on prefix-tuning for data wrangling tasks 🥳
12
David Vos retweeted
Efficient Optimization of Hierarchical Identifiers for Generative Recommendation Proposes greedy & hybrid tree construction methods that reduce identifier building time to 2-8% while maintaining or improving retrieval quality. 📝arxiv.org/abs/2512.18434 👨🏽‍💻github.com/joshrosie/re-seat…
4
1
32
781
Proud supervisor moment :) Do you need hierarchical identifiers for generative recommendations? You can actually be very efficient in your tree building 🧐 Work we did with excellent students from the RecSys course at @irlab_amsterdam. We’ll be presenting at ECIR 2026!
Efficient Optimization of Hierarchical Identifiers for Generative Recommendation Proposes greedy & hybrid tree construction methods that reduce identifier building time to 2-8% while maintaining or improving retrieval quality. 📝arxiv.org/abs/2512.18434 👨🏽‍💻github.com/joshrosie/re-seat…
1
75
We shed some light on what language model you should use for your next news recommender system!
Revisiting Language Models in Neural News Recommender Systems Studies the impact of language model sizes in news recommendation, finding that larger models don't always improve performance but help more with cold-start users. 📝arxiv.org/abs/2501.11391
2
132
Very happy to share my first project at NVIDIA! 🚀 We present Meshtron, a pure Transformer model that consistently reasons over 500k+ tokens to generate 3D meshes of outstanding quality. Webpage: research.nvidia.com/labs/dir… w/ @zekunhao19951 @TsungYiLinCV & @liu_mingyu 🧵
Revolutionize 3D mesh generation with artist-like precision and detail. Meshtron from #NVIDIAResearch creates intricate meshes with up to 64,000 faces at 1024-level resolution. Transform your 3D creations easily and with fine detail. 🎨 ✨ ➡️ developer.nvidia.com/blog/hi…
9
26
3
197
20,388
Honored to be Zihan’s paranymph!
Congratulations to Zihan @ZihanWa54274484 on successfully passing his PhD defense today!
6
276
I moved to bsky.app/profile/davidvos.bs…. It's really nice over there, I think you should try it!
3
56
My "looking for a postdoc" stickers and I are in Miami for #EMNLP2024! 🤩 Do you have/know of a postdoc for summer/autumn '25 related to interpretability, figlang and/or memorisation (vs generalisation)? Reach out! Looking forward to #GenBench2024 on Saturday and the many many...
4
17
2
84
10,278
Want to work on search for (news) media at one of the biggest publishers in the Netherlands? @DPGMediaIT is looking for an addition to their team. Great opportunity in my opinion: vacatures.dpgmedia.nl/o/soft…
1
109
🇨🇦 Deeeelighted to share that this work got into #neurips2024. Many thanks to my dear friend and co-author @Dafidofff, as well as the rest of the team. Solving PDEs in continuous space-time with Neural Fields on cool geometries while respecting their inherent symmetries! 💫💫
🌀 Equivariant Neural Fields (ENFs) for continuous PDE solving! We use ENFs as representation for solving PDEs in continuous space/time on different geometries while respecting their symmetries! (such as this internally heated ball of fluid) More details 👇
3
31
2
196
23,091
15th @ELLISforEurope @essir_eu Summer School on #informationretrieval was held from 1-5 July 2024 @irlab_amsterdam of the @UvA_Amsterdam 🚀 The ELLIS flagship school is the largest ESSIR organized so far and the first ESSIR to include a hackathon✨ 🔗ivi.fnwi.uva.nl/ellis/2024/1…
2
1
8
1,926
The Search Engines Amsterdam Meetup @irlab_amsterdam is back ✨ followed by drinks🍻 📆 Fri, Oct 4th 📍Lab42, room L3.33 ⏰16.00 🚀James Allan (UMass) will talk about Probing Ranking LLMs: Mechanistic Interpretability in Information Retrieval 🔗meetup.com/sea-search-engine…
3
6
758
Got to give a lecture on Transformers in the amazing Amsterdam Tuschinski theatre. Also gave @3blue1brown the screen real estate he really deserves 😊
1
14
769
Here’s a picture to prove I was actually on stage! And special thanks to @rmpetcu for so eloquently introducing the Attention Mechanism :)
1
1
3
229
Listening to @IR_oldie talking about challenges and opportunities in evaluating generative IR.
3
1
19
1,743
Are LLMs the end of IR? @ekanou presenting 42 things you need to know about IR at #ESSIR24.
3
4
43
2,007
This is really cool stuff. Proud of these two guys, @Dafidofff and @davidmknigge! Go check out their work :)
🥠 Excited to introduce our latest work on Equivariant Neural Fields (ENFs)! Grounding conditioning variables in geometry 🚀 Paper: arxiv.org/abs/2406.05753 Github: github.com/dafidofff/enf-jax Project Page: dafidofff.github.io/enf-jax/ Details below 👇👇
1
3
369
Structured data is everywhere. Now we just need to unlock its potential! A unique opportunity to go work with Madelon! And across the street from my office :)
This Post is from an account that no longer exists. Learn more
8
1,131
📣📣 Check out ESSIR 2024's program! We start with Transformers, Foundation Models, and Finetuning, followed by Generative IR, LLMs, (Neural) RecSys, LTR, Evaluation, Explainability, and Fairness. ESSIR features FIDA and a week-long hackathon! 👩🏻‍💻 2024.essir.eu/attending/scho…
1/4 I'm happy to announce that we at @irlab_amsterdam will organize the next edition of the European Summer School in Information Retrieval (ESSIR) @essir_eu on July 1-5 at the University of Amsterdam. Early registration deadline: May 31. More info: 2024.essir.eu
13
25
3,213
Although I couldn’t make it myself tonight, it’s really cool to see our Responsible Media Lab highlighted at @RecSysNL by @McWillemsen!
#RecSysNL is back. @McWillemsen is opening the session introducing @jadatascience . P.S. the location is just amazing. Thanks @DPGMediaAI for sponsoring the RecSysNL.
7
408