TabArena author on tabular ML: foundation models now default, LLMs surprisingly bad at tables
ziv_ravid · x · 2026-09-04
The Information Bottleneck podcast interviews David Holzmüller (INRIA), researcher behind TabArena, TabICL and RealMLP, on where tabular ML actually stands.
Key points:
- Tabular foundation models are now the default answer;
- LLMs are surprisingly bad at tabular tasks;
- Also covers Kaggle competitions, calibration, text embeddings for tabular models, time series vs tabular, and open problems.
Available on the show's website, YouTube and podcast apps.
More from Research
- Pedro Domingos proposes scrapping accept/reject: reviewers rate papers, attendees vote on what gets in — pmddomingos · 2026-09-04
- Classic Reward Hacking: A Soccer Bot That Started Running Backwards Loops for No Reason — generativist · 2026-09-04
- Baseten Launches Base Labs, an Open Research Lab for Open-Source Frontier AI — baseten · 2026-09-04
- New protocol enables private, verifiable LLM inference outsourcing with 179MB local storage — chaumian · 2026-09-04
- GPT-5.6 Pro improves lower bound for Moser's worm problem to 0.2374 — zero0_one1 · 2026-09-04
- FinFIRST details: 123 expert tasks, 701 atomic criteria and 12,300 rubric points for financial search agents — FellMentKE · 2026-09-04