@innixma

Author & Lead Developer of @AutoGluon and TabArena Research Scientist at Prior Labs #automl #tabpfn #opensource

 New York, NY
github.com/innixma
Joined December 2018
  • Tweets201
  • Following88
  • Followers1.3K
19Media
📢Announcing a Call for Papers for the @icmlconf "Foundation Models for Structured Data" Workshop!
2
10
1
84
11,133
TabPFN-3.5 released today. It achieves rank 1 on the famous historical #Kaggle Otto competition out of the box. This, to me, is more impressive than anything in the TabPFN-3.5 report. It shows that there remains more signal in the data to be learned than we ever thought possible
4
8
2
88
15,493
Nick Erickson retweeted
Frank Hutter
@FrankRHutterMay 12
The data science revolution is here now. TabPFN-3 is live, taking tabular foundation models to enterprise scale 🤩 1M training rows on a single H100. No training. No tuning. Load and predict. 🧵 1/5 #tabpfn #tabularfoundationmodels #priorlabs
9
47
5
282
26,883
🚀 AutoGluon 1.5 is out - pushing the frontier of tabular and time series learning! 🏆New SOTA on TabArena: +60 Elo, 2x faster training, 50% faster inference, 3% lower rel error vs 1.4 🧠New Models: RealTabPFN-2.5, TabDPT, Chronos-2 ⏱️AutoGluon-TimeSeries: 80% win-rate vs 1.4
2
6
1
18
1,036
Check out the full release notes: github.com/autogluon/autoglu…

Release v1.5.0 · autogluon/autogluon

Version 1.5.0 We are happy to announce the AutoGluon 1.5.0 release! AutoGluon 1.5.0 introduces new features and major improvements to both tabular and time series modules. This release contains 131...

github.com
2
315
Wow! It is amazing to see the progress being made. As I said when TabPFNv2 released: "time to code".
Frank Hutter
@FrankRHutter6 Nov 2025
The data science revolution continues — TabPFN is now SOTA up to 50k data points and 2000 features 🚀 For the size limits of TabPFNv2, in a forward pass Real-TabPFN-2.5 outperforms AutoGluon 1.4 (complex ensemble including TabPFNv2 tuned for 4h) by 93 ELO points on TabArena.🧵1/
5
370
💥 Tabular ML just got a massive upgrade. 🚀 AutoGluon 1.4.0 is here — our biggest release yet. 🔥 Extreme Preset → biggest performance leap ever 🧠 5 new SOTA models 🤖 MLZero 1.0 (zero‑code multimodal AutoML) 📊 TabArena benchmark ⚡ Time series speedups
3
17
2
53
5,176
more replies
🤖 MLZero 1.0 (AutoGluon Assistant) - Zero‑code AutoML for tabular, image, text, audio - New WebUI + MCP integration - Multi‑agent LLM system for end‑to‑end ML automation 📊 TabArena - Living leaderboard for tabular ML - Continuous benchmarking feeding directly into AutoGluon
1
2
502
⚡ Time Series - Up to 7× faster training & inference on large datasets - New PerStepTabular forecasting model 📌 Learn more: 🔗 Full release: github.com/autogluon/autoglu… 🔗 TabArena: tabarena.ai 🔗 Mitra: amazon.science/blog/mitra-mi… 🔗 MLZero: github.com/autogluon/autoglu…
3
465
Foundation Models for Structured Data Workshop, Today, ICML 2025, starting at 9 AM PT, West Ballroom D, best room in the building! Let's do this!
3
2
14
2,125
🚨 Join us tomorrow at the 1st ICML Workshop on Foundation Models for Structured Data! 📅 July 18 | ⏰ 9 AM - 5 PM 📍 West Ballroom D 🎉 70 accepted papers 🎤 8 spotlight orals 🌟 6 invited talks from top experts in tabular & timeseries data! #ICML2025 #FMSD
1
5
1
24
4,087
Checkout the full schedule here, kicked off by an invited talk from @FrankRHutter on the path towards TabPFNv3! icml.cc/virtual/2025/worksho… And checkout the workshop's website for more information: icml-structured-fm-workshop.…
2
251
Attending #icml25, eager to discuss all things tabular data, structured foundation models, and AutoML! Lots of progress has been made this year, exciting times ahead! Next AutoGluon release is very soon.
1
2
14
605
I will talk about TabArena at the AutoML seminar this week, join me :) More information: automl-seminars.github.io/
3
7
548
Nick Erickson retweeted
Frank Hutter
@FrankRHutter24 Jun 2025
It's great to have this new open-source benchmark for tabular data 🚀 It's really comprehensive, created and maintained by serious open source contributors from various groups, and I expect it to quickly become the new standard benchmark. I'm super excited that progress in the tabular space is exploding! It's also interesting to see that, for the datasets in TabPFN v2's constraints (<10k data points, <500 features), half a year after its release, TabPFN v2 still strongly outperforms all other methods!
Lennart Purucker@LennartPurucker23 Jun 2025
🚨What is SOTA on tabular data, really? We are excited to announce 𝗧𝗮𝗯𝗔𝗿𝗲𝗻𝗮, a living benchmark for machine learning on IID tabular data with: 📊 an online leaderboard (submit!) 📑 carefully curated datasets 📈 strong tree-based, deep learning, and foundation models 🧵
4
20
1
107
12,572
We put all the best practices for tabular data benchmarking in one place to make TabArena, a living benchmark for tabular data. Compare with the very best implementations of state-of-the-art methods out of the box! Here are my thoughts: 🧵
Lennart Purucker@LennartPurucker23 Jun 2025
🚨What is SOTA on tabular data, really? We are excited to announce 𝗧𝗮𝗯𝗔𝗿𝗲𝗻𝗮, a living benchmark for machine learning on IID tabular data with: 📊 an online leaderboard (submit!) 📑 carefully curated datasets 📈 strong tree-based, deep learning, and foundation models 🧵
2
6
24
1,770
more replies
Tabular foundation models work. They achieve performance competitive with or exceeding the prior state-of-the-art within their supported dataset constraints. While more work needs to be done in overcoming these constraints, the results are very promising. 3/
1
3
155
TabArena is a living benchmark. With the community, we will continually update it! Authors: @innixma @LennartPurucker @atschalz @DHolzmueller Prateek Mutalik Desai @_geoalgo_ @FrankRHutter LB: tabarena.ai Paper: arxiv.org/abs/2506.16791Code: tabarena.ai/code 4/4

TabArena - a Hugging Face Space by TabArena

Elo-ranked leaderboards for tabular ML, IID and beyond

huggingface.co
5
159
Load more