Skip to main content

ICML 2026 — 2nd Workshop on Foundation Models for Structured Data · 2026

Poster

Mix, Don't Pick: Why Synthetic Corpus Composition Matters for Time Series Foundation Model Pretraining

Aaryan Nagpal · Debdeep Sanyal · Murari Mandal · Dhruv Kumar · Saurabh Deshpande

Shows that picking a single “best” synthetic data generator for pretraining is the wrong approach — an equal-weight mix of many generators consistently matches or beats the best individual one.

Research program: Structured Foundational Models

← All publications