PulseAugur
EN
LIVE 07:47:11

New theory bounds sample complexity for autoregressive CoT learning

By PulseAugur Editorial · [2 sources] ·

Researchers have developed a new theoretical framework for understanding the sample complexity of learning autoregressive Chain-of-Thought (CoT) traces. The study proves that in the realizable PAC setting, the sample complexity is bounded by the standard multiclass rate, governed by the Daniely-Shalev-Shwartz (DS) dimension. This new approach introduces parity dimension, a refinement of DS dimension that is stable under rollout and does not increase with autoregressive steps, addressing a limitation of the original DS dimension. AI

IMPACT Provides a theoretical foundation for understanding the data requirements of advanced reasoning methods in LLMs.

RANK_REASON The cluster contains a single academic paper detailing theoretical research in machine learning. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv stat.ML →

AI-generated summary · Google Gemini · from 2 sources. How we write summaries →

New theory bounds sample complexity for autoregressive CoT learning

COVERAGE [2]

  1. arXiv stat.ML TIER_1 English(EN) · Zhiyuan Li ·

    The Optimal Sample Complexity of Learning Autoregressive Chain-of-Thought

    arXiv:2607.07423v1 Announce Type: cross Abstract: We prove that, in the realizable PAC setting, the sample complexity of exact-trace learning for full autoregressive Chain-of-Thought traces is upper bounded by the standard multiclass rate of the local next-token class, where this…

  2. arXiv stat.ML TIER_1 English(EN) · Zhiyuan Li ·

    The Optimal Sample Complexity of Learning Autoregressive Chain-of-Thought

    We prove that, in the realizable PAC setting, the sample complexity of exact-trace learning for full autoregressive Chain-of-Thought traces is upper bounded by the standard multiclass rate of the local next-token class, where this rate is governed by the Daniely--Shalev-Shwartz d…

RELATED ENTITIES

RELATED TOPICS