Research

Paper

TESTING March 03, 2026

Test-Time Meta-Adaptation with Self-Synthesis

Authors

Zeyneb N. Kaya, Nick Rui

Abstract

As strong general reasoners, large language models (LLMs) encounter diverse domains and tasks, where the ability to adapt and self-improve at test time is valuable. We introduce MASS, a meta-learning framework that enables LLMs to self-adapt by generating problem-specific synthetic training data and performing targeted self-updates optimized for downstream performance at inference time. We train this behavior end-to-end via bilevel optimization: an inner loop adapts on self-generated examples while an outer loop meta-learns data-attribution signals and rewards post-update task performance. The synthetic data is optimized with scalable meta-gradients, backpropagating the downstream loss through the inner updates to reward useful generations. Experiments on mathematical reasoning show that MASS learns to synthesize per-instance curricula that yield effective, data-efficient test-time adaptation.

Metadata

arXiv ID: 2603.03524
Provider: ARXIV
Primary Category: cs.LG
Published: 2026-03-03
Fetched: 2026-03-05 06:06

Related papers

Raw Data (Debug)
{
  "raw_xml": "<entry>\n    <id>http://arxiv.org/abs/2603.03524v1</id>\n    <title>Test-Time Meta-Adaptation with Self-Synthesis</title>\n    <updated>2026-03-03T21:16:18Z</updated>\n    <link href='https://arxiv.org/abs/2603.03524v1' rel='alternate' type='text/html'/>\n    <link href='https://arxiv.org/pdf/2603.03524v1' rel='related' title='pdf' type='application/pdf'/>\n    <summary>As strong general reasoners, large language models (LLMs) encounter diverse domains and tasks, where the ability to adapt and self-improve at test time is valuable. We introduce MASS, a meta-learning framework that enables LLMs to self-adapt by generating problem-specific synthetic training data and performing targeted self-updates optimized for downstream performance at inference time. We train this behavior end-to-end via bilevel optimization: an inner loop adapts on self-generated examples while an outer loop meta-learns data-attribution signals and rewards post-update task performance. The synthetic data is optimized with scalable meta-gradients, backpropagating the downstream loss through the inner updates to reward useful generations. Experiments on mathematical reasoning show that MASS learns to synthesize per-instance curricula that yield effective, data-efficient test-time adaptation.</summary>\n    <category scheme='http://arxiv.org/schemas/atom' term='cs.LG'/>\n    <category scheme='http://arxiv.org/schemas/atom' term='cs.AI'/>\n    <published>2026-03-03T21:16:18Z</published>\n    <arxiv:comment>5 pages, 2 figures, 1 table. Accepted to ICLR 2026 LIT and Data-FM Workshops</arxiv:comment>\n    <arxiv:primary_category term='cs.LG'/>\n    <author>\n      <name>Zeyneb N. Kaya</name>\n    </author>\n    <author>\n      <name>Nick Rui</name>\n    </author>\n  </entry>"
}