Latest Articles for Language Models

New metrics improve large language models' effectiveness in education.

This article examines how large language models recall information from training data.

Adapting multilingual models can improve performance for less-used Uralic languages.

Explore the role of ordinal classification and the impact of pretrained language models.

Explore how DETAIL enhances understanding of in-context learning in language models.

TPO offers a new method to align language models with human preferences efficiently.

ThReaD improves LLMs' performance on complex tasks through dynamic thread management.

This article examines the risks of fine-tuning language models for safety.

A new approach enhances prompt diversity for safer language models.

Research reveals the challenges of watermark detection in large language models.

This study presents a system to enhance language model accuracy using adversarial challenges.

Learn how adaptive teams improve task performance with language model agents.

MAP-Neo aims for transparency and performance in AI language modeling.

Examining the challenges and solutions in LLM watermarking to prevent misuse.

New resources enhance assessment of Korean language models.

Research shows diverse instructions improve language model performance in unseen tasks.

Research introduces a method to improve decision-making in language model agents.

This study examines how LLMs handle reasoning in abstract and contextual scenarios.

The Block Transformer improves text processing speed and efficiency in language models.

Recent tests reveal LLMs' weaknesses in simple reasoning despite high benchmark scores.

A guide to transforming non-idiomatic Python code using modern techniques.

This study examines how LLMs handle changes in summarization tasks.

This study explores how to create sentences that maintain specific meanings using FrameNet.

This study assesses GPT-4's ability to extract data from materials science literature.

Jamming attacks can disrupt retrieval-augmented generation systems by blocking responses.

This article evaluates the capability of language models to simulate game environments.

A new approach to assess reasoning strategies with a focus on computational costs.

MedExQA sets a new standard for evaluating medical language models with a focus on explanations.

Study evaluates how well LLMs reason beyond immediate context.

Exploring the limitations of Direct Preference Optimization in language model training.

Evaluating how well language models perform research surveys across various academic fields.

A new tool to assess language models' continuous improvement through feedback.

A new framework assesses language models on emotional intelligence and creativity.

New methods enhance language models' performance through better example selection.

ReadCtrl allows language models to better match text complexity to reader abilities.

GAMA improves audio processing by merging sound and language insights.

SciEx reveals strengths and challenges of LLMs in scientific evaluation.

This study shows how BERT learns COVID-19 facts through continuous training.

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

A new framework enhances how LLM agents learn through detailed process guidance.