Latest Articles for Evaluation

SQuArE metric improves evaluation of QA systems through multiple answer references.

A new system aims to connect users with medical professionals through automated classification.

Advancements in summarizing doctor-patient conversations improve telemedicine communication.

Exploring proof techniques for evaluating functions in programming languages.

Stability in clustering ensures groups are effective and meaningful.

GRANDE uses gradient descent to improve learning from tabular data.

A new method for assessing AI models through embeddings and meta features.

A new method reveals patterns in legal decisions using automated text analysis.

A model for consistent photo quality across different smartphones.

Introducing alternatives can enhance user satisfaction in fashion recommendation systems.

A new dataset evaluates language models' abilities in advanced math problem solving.

Examining the effects of inter-dataset code duplication on model performance metrics.

This study focuses on enhancing retrieval-augmented generation methods for Brazilian Portuguese.

This study introduces WAVES, a benchmark to evaluate watermarking techniques against various attacks.

Orion-14B excels in understanding and generating multilingual text with 14 billion parameters.

New methods assess how dialogue systems maintain personality consistency.

This framework enhances how knowledge is combined in machine learning models for better performance.

Study reveals language models can generate useful PET report impressions.

Assessing the accuracy of LLMs in diagnosing medical conditions from images and symptoms.

This research enhances AI-generated radiology report evaluations through expert collaboration.

Analyzing how red-teaming can enhance AI safety and address potential risks.

Examining harm amplification in text-to-image models and its societal impact.

This paper discusses adjusting language models to align with human values and expectations.

A new open language model for research and innovation in natural language processing.

Introducing a flexible framework to enhance voice privacy research.

EvaLLM offers a structured approach to assess AI-generated visual content.

A method for verifying machine learning models to enhance trust and transparency.

SIDU-TXT sheds light on AI decisions in natural language processing.

Research shows women face biases in evaluations and funding in academia.

A new method converts handwritten notes into digital ink for easy use.

An analysis of reproducibility issues in deep learning software fault prediction research.

New method improves fact-checking for computer-generated texts with ambiguous names.

Learn how to design posters that communicate messages clearly and attractively.

Exploring the challenges and solutions of reward hacking in AI model training.

A fresh method for assessing how models respond to image-related queries.

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

New methods improve how we assess computer-generated text.

A detailed look at CyberMetric's evaluation of AI and human experts in cybersecurity.

Addressing ethical concerns through selective memory removal in AI models.

Exploring how machines create images from text prompts and align with human preferences.