Latest Articles for Model Evaluation

A new benchmark to assess LLMs for Java programming tasks.

This article explores strategies for improving model generalization and understanding gradient behavior.

A toolkit for assessing the safety of advanced language models.

This article analyzes the performance of fine-tuned models versus generative AI in text classification tasks.

This article examines how Visual State Space Models handle visual challenges.

A new data set assesses how LLMs reason with multiple images.

Investigating how LLM predictions align with human choices using statistical modeling.

A new benchmark suite helps assess reasoning shortcuts in artificial intelligence.

A study evaluates language models on handling multiple tasks simultaneously.

A study highlights gaps in reasoning abilities of LLMs for math problem solving.

A fresh method for testing language model safety and multilingual skills.

Methods for identifying important features in low-quality data environments.

New methods reveal challenges in unlearning knowledge from language models.

A study on the decision-making processes of large language models.

A look at how calibration impacts model predictions and reliability.

Long-context language models streamline complex tasks and improve interaction with AI.

A method to evaluate model knowledge through internal processing.

Examining the impact of data contamination on language model performance and evaluation.

This study reveals the limits of text-to-image models in handling numbers.

A new metric improves evaluation of text classification models across different domains.

A deep dive into how well vision models recognize and represent multiple objects.

A study on the effectiveness of OOD detectors against adversarial examples.

Research highlights in-context learning abilities in large language models.

A study highlighting the importance of comprehensive annotations for retrieval evaluation.

A new benchmark highlights the risks of spurious bias in multimodal language models.

Investigating fine-grained feedback for text-to-image models and its practical implications.

New benchmark assesses how video-language models handle inaccuracies effectively.

APIGen generates diverse, high-quality datasets for function-calling agents.

A new method to detect biases in language model training.

SAVE model enhances audio-visual segmentation with efficiency and precision.

A fresh approach to gauge model accuracy without labels during data shifts.

Insights on the challenges of machine learning in predicting material properties.

New benchmark improves evaluation of multimodal models by minimizing biases.

This study examines how visual and textual data affect model performance.

CD-T enhances understanding of transformer models, improving interpretation and trust.

New benchmark assesses gender bias in AI models related to job roles.

Examining vulnerabilities from clean-label backdoor attacks and how generalization bounds can help.

A new tool for testing language models in noisy environments.

A new approach to evaluate ML models focusing on data preparation.

Research assesses stability of XAI methods using diabetes dataset.