Zhao Song

A look at Mahalanobis distance and innovative approaches for data analysis.

A method to balance rewards and resources using clustered contextual bandits.

Exploring the rise of decentralized language models and their benefits over centralized systems.

Innovative methods reduce memory use in semidefinite programming for streaming data.

A look into the relationship between Graph Neural Networks and Graph Neural Tangent Kernel.

Discover how attention shapes language models and their applications in technology.

This research examines the efficiency of backward computation in training language models.

Examining LLMs' capability to address mathematical problems, especially modular arithmetic.

Discover how sparse attention improves processing in language models.

Exploring methods to recover model parameters from leverage scores in regression analysis.

Exploring the importance of softmax in neural network performance and applications.

A new method enhances attention mechanisms in language models for better performance.

Exploring the fundamentals and applications of diffusion models in various fields.

Exploring tensor attention and its impact on data processing in AI models.

Advancements in fine-tuning language models using innovative techniques.

Explore the advantages and applications of Low-Rank Adaptation in AI models.

Examining differential privacy and NTK regression to protect user data in AI.

Exploring differential privacy to protect sensitive information in AI applications.

This article reviews the capabilities and limits of latent diffusion transformers.

A new method enhances John ellipsoid computation while protecting sensitive data.

Explore the significance of leverage scores in data analysis and privacy.

SparseGPT improves the speed and efficiency of large language models through parameter pruning.

A new approach enhances gradient calculations, improving transformer efficiency in machine learning.

New algorithms combine quantum computing and classical methods to speed up calculations.

Learn how differential privacy enhances data analysis while safeguarding personal information.

1-bit models show great potential in machine learning efficiency and performance.

Exploring the capabilities and challenges of Transformer technology in understanding language.

Learn how string distances can aid privacy in sensitive data analysis.

A closer look at how MHNs can enhance machine learning.

A look at Mamba and State-Space Models in AI capabilities.

Exploring methods for fair machine learning through low-rank approximation and subset selection.

LazyDiT offers a smarter way to create images faster without losing quality.

Innovative pruning techniques make AI models more efficient and effective.

Grams offers a fresh take on optimization for machine learning models.

Discover how tensor attention transforms AI language processing.

New methods improve RoPE attention, speeding up AI computations significantly.