Copertina

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

Fuente: arXiv

Salvato in:

Dettagli Bibliografici
Autori principali:	Gan, Zeyu, Ren, Ruifeng, Yao, Wei, Hu, Xiaolin, Xu, Gengze, Qian, Chen, Tang, Huayi, Gong, Zixuan, Yao, Xinhao, Tang, Pengwei, Dou, Zhenxing, Liu, Yong
Natura:	Preprint
Pubblicazione:	2026
Soggetti:	Computation and Language
Accesso online:	Acceder al recurso
Tags:	Aggiungi Tag Nessun Tag, puoi essere il primo ad aggiungerne!!

Documenti analoghi

Towards Auto-Regressive Next-Token Prediction: In-Context Learning Emerges from Generalization
di: Gong, Zixuan, et al.
Pubblicazione: (2025)

Theoretical Insights into Fine-Tuning Attention Mechanism: Generalization and Optimization
di: Yao, Xinhao, et al.
Pubblicazione: (2024)

Transformers as Intrinsic Optimizers: Forward Inference through the Energy Principle
di: Ren, Ruifeng, et al.
Pubblicazione: (2025)

ADePT: Adaptive Decomposed Prompt Tuning for Parameter-Efficient Fine-tuning
di: Tang, Pengwei, et al.
Pubblicazione: (2025)

On Weak-to-Strong Generalization and f-Divergence
di: Yao, Wei, et al.
Pubblicazione: (2025)

Sparsity is Combinatorial Depth: Quantifying MoE Expressivity via Tropical Geometry
di: Su, Ye, et al.
Pubblicazione: (2026)

Enhancing In-Context Learning Performance with just SVD-Based Weight Pruning: A Theoretical Perspective
di: Yao, Xinhao, et al.
Pubblicazione: (2024)

Compositional Generalization from Learned Skills via CoT Training: A Theoretical and Structural Analysis for Reasoning
di: Yao, Xinhao, et al.
Pubblicazione: (2025)

Effective Frontiers: A Unification of Neural Scaling Laws
di: Zou, Jiaxuan, et al.
Pubblicazione: (2026)

Understanding Model Ensemble in Transferable Adversarial Attack
di: Yao, Wei, et al.
Pubblicazione: (2024)

Information-Theoretic Generalization Bounds for Transductive Learning and its Applications
di: Tang, Huayi, et al.
Pubblicazione: (2023)

PAC-Bayesian Generalization Bounds for Graph Convolutional Networks on Inductive Node Classification
di: Tang, Huayi, et al.
Pubblicazione: (2025)

On the Emergence of Weak-to-Strong Generalization: A Bias-Variance Perspective
di: Xu, Gengze, et al.
Pubblicazione: (2025)

Perfect Alignment May be Poisonous to Graph Contrastive Learning
di: Liu, Jingyu, et al.
Pubblicazione: (2023)

Learning Mixture-of-Experts for General-Purpose Black-Box Discrete Optimization
di: Liu, Shengcai, et al.
Pubblicazione: (2024)

A Method for Constructing a Digital Transformation Driving Mechanism Based on Semantic Understanding of Large Models
di: Liu, Huayi
Pubblicazione: (2026)

Put the Space of LoRA Initialization to the Extreme to Preserve Pre-trained Knowledge
di: Tang, Pengwei, et al.
Pubblicazione: (2025)

Active Semantic Perception
di: Tang, Huayi, et al.
Pubblicazione: (2025)

The Debate on RLVR Reasoning Capability Boundary: Shrinkage, Expansion, or Both? A Two-Stage Dynamic View
di: Yao, Xinhao, et al.
Pubblicazione: (2025)

Breaking the Black-Box: Confidence-Guided Model Inversion Attack for Distribution Shift
di: Liu, Xinhao, et al.
Pubblicazione: (2024)

No Black Box Anymore: Demystifying Clinical Predictive Modeling with Temporal-Feature Cross Attention Mechanism
di: Li, Yubo, et al.
Pubblicazione: (2025)

A Survey of Calibration Process for Black-Box LLMs
di: Xie, Liangru, et al.
Pubblicazione: (2024)

Improved Learning Rates for Stochastic Optimization
di: Li, Shaojie, et al.
Pubblicazione: (2021)

Opening the Black Box: A Survey on the Mechanisms of Multi-Step Reasoning in Large Language Models
di: Pan, Liangming, et al.
Pubblicazione: (2026)

Beyond the Black Box: A Cognitive Architecture for Explainable and Aligned AI
di: Keyi, Hu
Pubblicazione: (2025)

Towards Understanding How Transformers Learn In-context Through a Representation Learning Lens
di: Ren, Ruifeng, et al.
Pubblicazione: (2023)

Revisiting Transformers through the Lens of Low Entropy and Dynamic Sparsity
di: Ren, Ruifeng, et al.
Pubblicazione: (2025)

The Capabilities and Limitations of Weak-to-Strong Generalization: Generalization and Calibration
di: Yao, Wei, et al.
Pubblicazione: (2025)

Efficient Non-Parametric Uncertainty Quantification for Black-Box Large Language Models and Decision Planning
di: Tsai, Yao-Hung Hubert, et al.
Pubblicazione: (2024)

Efficient Black-Box Fault Localization for System-Level Test Code Using Large Language Models
di: Yaraghi, Ahmadreza Saboor, et al.
Pubblicazione: (2025)

On the Blessing of Pre-training in Weak-to-Strong Generalization
di: Yao, Wei, et al.
Pubblicazione: (2026)

Multimodal Hierarchical Attention Framework for Efficient Weakly Supervised Few‐Shot Segmentation Under SAGIN Environment
di: Wenqiang Yuan, et al.
Pubblicazione: (2025)

Towards a Theoretical Understanding of Synthetic Data in LLM Post-Training: A Reverse-Bottleneck Perspective
di: Gan, Zeyu, et al.
Pubblicazione: (2024)

Learning to Learn from APIs: Black-Box Data-Free Meta-Learning
di: Hu, Zixuan, et al.
Pubblicazione: (2023)

Beyond Entropy: Region Confidence Proxy for Wild Test-Time Adaptation
di: Hu, Zixuan, et al.
Pubblicazione: (2025)

Probing Stochastic Ultralight Dark Matter with Space-based Gravitational-Wave Interferometers
di: Yao, Yue-Hui, et al.
Pubblicazione: (2024)

Beyond the Black Box: Interpretability of LLMs in Finance
di: Tatsat, Hariom, et al.
Pubblicazione: (2025)

Beyond Accuracy: A Geometric Stability Analysis of Large Language Models in Chess Evaluation
di: Song, Xidan, et al.
Pubblicazione: (2025)

Black-Box Guardrail Reverse-engineering Attack
di: Yao, Hongwei, et al.
Pubblicazione: (2025)

SyncGuard: Robust Audio Watermarking Capable of Countering Desynchronization Attacks
di: Gan, Zhenliang, et al.
Pubblicazione: (2025)