Latest Articles for Performance

Study questions the need for optimizing in-context examples when clear instructions are provided.

Investigating how skill differences impact performance in competitive settings.

AV-SUPERB evaluates audio and visual models across various tasks for better performance.

Addressing ethical concerns through selective memory removal in AI models.

How team composition and coordination impact performance in organizations.

Analysis shows demographic disparities in brain age prediction accuracy.

DeepATLAS efficiently identifies anatomical structures in CT scans without extensive labeled data.

An overview of transformers and their impact on data processing.

A new method improves memory usage in large language models, enhancing performance.

This article discusses a method for learning visual features from video data.

This study focuses on improving continual learning methods in 3D semantic tasks.

Examining the dangers of integrating language models in robotic systems.

A new method aims to reduce harmful outputs from AI language models.

A new method for efficient document ranking within budget limits.

Study explores how gender impacts calculus abilities and self-efficacy in physics.

A new approach tackles the issue of dropped tokens and padding in machine learning models.

This article discusses a new method to improve prompt performance for language models.

A new approach to make language models smaller and faster using 1-bit quantization.

This benchmark assesses the performance of medical language models in healthcare.

Study investigates how LLMs recall graphs compared to human memory.

Exploring methods to enhance coding task performance in language models using data.

A new algorithm improves optimization when hyperparameters are unknown.

A new approach leverages self-supervised learning for connecting audio and sheet music.

A new tool aids researchers in modeling optical turbulence effectively.

A new method reduces forgetting in language models during updates.

A look into how VLMs combine image and text processing.

ProSparse improves activation sparsity in LLMs for better efficiency and performance.

A new approach enhances multigrid methods for complex simulations with low quality cells.

A study on improving task mapping in CGRAs using SAT methods.

Introducing a novel approach to rank language models without needing reference answers.

This research evaluates AI model confidence and explanation quality in noisy environments.

Efficient counting solutions for modern applications using approximate methods.

Masked Matrix Multiplication improves efficiency in AI computations by utilizing data sparsity.

A new watermarking method safeguards customized language models from unauthorized use.

FanOutQA helps evaluate language models on challenging multi-hop questions using structured data.

Investigating how tokenization methods affect arithmetic tasks in language models.

Exploring the role and performance of NAT64 in IPv4 and IPv6 communication.

A study on improving math reasoning through effective data strategies.

Researchers investigate how models adapt when components are removed.

A new method enhances agent performance in complex environments without extra training.