Theo Rekatsinas
@thodrek
Followers
2K
Following
2K
Media
20
Statuses
750
Machine Learning Systems, Data Management & Knowledge Graphs @Apple; Ex-Professor @ETH & @UWMadison; Co-founder of inductiv (acquired by @Apple)
Zurich, Switzerland
Joined February 2015
Scalability is a key factor limiting the use of Graph Neural Networks (GNNs) over large graphs; w/ @RWaleffe, @JasonMohoney , and Shiv, we introduce Marius++ (, a system for *out-of-core* GNN mini-batch training over billion-scale graphs. (1/5).
2
5
33
RT @logan_engstrom: Announcing a deadline extension for the ATTRIB workshop! Submissions are now due September 25th, with an option to subm….
0
5
0
RT @pinartozun: We had a great @dagstuhl seminar this week gathering to discuss resource-efficient ML. Many thanks to my co-organizers (@pp….
0
5
0
Time for #AI and real-time analytics to take the fun out of cycling
cyclingweekly.com
Organisers of the French Grand Tour say that the converted van shall not be authorised 'within any of the premises of the event' and will not gain accreditation to follow the race
0
0
3
Very cool work @RWaleffe !!.
A 8B-3.5T hybrid SSM model gets better accuracy than an 8B-3.5T transformer trained on the same dataset:.* 7% attention, the rest is Mamba2.* MMLU jumps from 50 to 53.6%.* Training efficiency is the same.* Inference cost is much less.
0
2
16
RT @peteratmsr: Aurora is an AI foundation model that flexibly achieves SoTA 5-day air pollution, 10-day global weather, and other forecast….
microsoft.com
Aurora is a large-scale foundation model of the atmosphere trained on over a million hours of diverse weather and climate data to produce operational forecasts for a wide variety of Earth system...
0
17
0
You can find all our comparisons against 30+ importance-based data pruning and selections methods at our paper: Turns out that sophisticated pruning might be a mirage for pre-training. .
0
0
1
Data pruning to reduce pertaining costs is hot, but fancy pruning can take just as long to select data as to train on all of it! Patrik, @Rwaleffe, and @vmageirakos's work at #ICLR2024 tomorrow shows how a simple, low-cost tweak to random sampling outperforms trendy methods!.
Not convinced about using random sampling for data pruning? Consider twice! In our recent work, we introduce Repeated Sampling of Random Subsets: where we sample a subset of data at each epoch of training instead of only once at the beginning!.
2
4
15
RT @besanushi: New job post looking for senior ML Engineers in Model Evaluation and Understanding. If you are at #ICLR2024, come talk to ou….
0
8
0
RT @gimdong58085414: We sadly found out our CTM paper (ICLR24) was plagiarized by TCD! It's unbelievable😢—they not only stole our idea of t….
0
201
0
RT @aminkarbasi: How we feed the data has a significant effect on time-to-accuracy. This work will appear in #ICLR2024. We got mixed bag re….
0
3
0
RT @togethercompute: We are thrilled to announce the Together Embeddings endpoint! 🚀. Higher quality than OpenAI or Cohere in the MTEB benc….
0
56
0
RT @joe_hellerstein: The first round of SIGMOD 2025 abstract deadlines is this Wednesday, January 10, at 23:59 AoE. Full paper due one week….
0
10
0
RT @beenwrekt: I highly recommend this thoughtful blog by @deliprao, painting the academic paper game in a bigger picture. .
0
3
0
RT @LesterMackey: If you’d like to join Microsoft Research New England as a researcher in AI / ML / statistics, please apply here: https://….
0
25
0
RT @FannyYangETH: In the context of this grant, I have openings for postdocs with interests and background in mathematical foundations of t….
0
6
0