Latest Articles for Evaluation

A study evaluates language models on handling multiple tasks simultaneously.

A new benchmark tests LLMs' abilities with structured data formats.

VCEval offers an automated way to assess online course effectiveness.

A new benchmark targets compositionality in video understanding and language models.

A new method enhances testing for language models using real user data.

The Nemotron-4 340B family delivers powerful models for diverse applications and synthetic data generation.

Evaluating how language models handle cultural cues in real tasks.

VideoVista offers a comprehensive evaluation for video question-answering models.

This article explores methods to enhance the reliability of research artifacts in computing.

GLM-4 models show improved capabilities in language understanding and generation.

A study on using LLMs to judge other LLMs and its implications.

A study on how language models generate persuasive rationales for argument evaluation.

Two new models aim to improve technology access for Galician speakers.

Examining the difficulties of translating metaphorical language in machine translation.

DF40 offers a comprehensive approach to improving deepfake detection methods.

This study assesses the honesty of LLMs in three key areas.

Discover how companies enhance their question-answering systems for better user support.

A study on how AI comprehends algorithms and their implications.

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

Data contamination affects the evaluation of large language models significantly.

A new method for assessing LLMs aligns with human values.

A new tool to evaluate biases in large vision-language models.

A study evaluates how machines create varied and creative poetry compared to humans.

A new method improves how we assess counter narratives to hate speech.

InternLM-Law enhances responses to diverse Chinese legal questions with advanced training.

Exploring how user profiles improve personalization in language models.

Research shows models struggle with step dependencies in cooking recipes.

This paper presents a method to assess language models across various prompts.

New method addresses regional differences in gender bias evaluation.

M2Lingual dataset improves instruction-following capabilities across various languages.

This article presents a new method for assessing text-to-image models effectively.

This study benchmarks Language Models' performance using Italian INVALSI tests.

RAGBench introduces a comprehensive dataset for evaluating Retrieval-Augmented Generation systems.

Dysca introduces a new way to assess LVLM performance using synthetic data.

A look at modern methods in engineering design for efficiency and performance.

A new approach improves causal event extraction using human-centered evaluation.

Assessing how deferring to human experts affects prediction accuracy in ML models.

Introducing a new method for better solutions in complex engineering and robotics tasks.

A study assessing the quality of datasets for identifying hate speech online.

A new method measures how language models adapt their beliefs with new evidence.