Beyond the Black Box: A Survey on the Theory and Mechanism of Large Language Models

TL;DR

This survey reviews the theoretical foundations and internal mechanisms of Large Language Models, proposing a unified framework to bridge empirical success with scientific understanding and identify key future challenges.

Contribution

It introduces a lifecycle-based taxonomy for LLM research and systematically analyzes core theoretical issues and frontier challenges in the field.

Findings

Analyzes mathematical justification for data mixtures

Identifies representational limits of architectures

Discusses optimization dynamics of alignment algorithms

Abstract

The rapid emergence of Large Language Models (LLMs) has precipitated a profound paradigm shift in Artificial Intelligence, delivering monumental engineering successes that increasingly impact modern society. However, a critical paradox persists within the current field: despite the empirical efficacy, our theoretical understanding of LLMs remains disproportionately nascent, forcing these systems to be treated largely as ``black boxes''. To address this theoretical fragmentation, this survey proposes a unified lifecycle-based taxonomy that organizes the research landscape into six distinct stages: Data Preparation, Model Preparation, Training, Alignment, Inference, and Evaluation. Within this framework, we provide a systematic review of the foundational theories and internal mechanisms driving LLM performance. Specifically, we analyze core theoretical issues such as the mathematical justification for data mixtures, the representational limits of various architectures, and the optimization dynamics of alignment algorithms. Moving beyond current best practices, we identify critical frontier challenges, including the theoretical limits of synthetic data self-improvement, the mathematical bounds of safety guarantees, and the mechanistic origins of emergent intelligence. By connecting empirical observations with rigorous scientific inquiry, this work provides a structured roadmap for transitioning LLM development from engineering heuristics toward a principled scientific discipline.