Latest Articles for Language Models

Automated tools like LLMs help in verifying claims efficiently.

This approach uses self-evaluation to guard against harmful outputs in language models.

Studying how quantization affects performance in different languages.

DCoT enhances language model performance through multiple reasoning paths.

Study reveals how word meanings shift with context and time using word embeddings.

A new approach to training reward models that aligns with human preferences.

Adapting prompts to specific models improves performance in language tasks.

Examining the role of semantic graphs in simplifying sentences with large language models.

Research explores improving citation text generation using large language models.

A look into methods and challenges of generating counterfactuals in NLP.

A study classifies tweets from parents about childhood disorders.

The study reveals the bias in AI evaluation tools favoring longer responses.

Examining how users shape toxic language in conversations with large language models.

A new method improves summarization with limited training data.

This paper evaluates LLM performance in a Theory of Computing course.

Exploring how confidence levels are attributed to LLMs and their implications.

We test language models' reasoning skills using various games, revealing significant limitations.

A new method simplifies science communication using collaborative language models.

A new method enhances the efficiency of language models using shared attention weights.

This study examines how LLMs change information through interactions.

This paper studies how training influences the predictions of large language models.

New methods enhance cache management for large language models.

A detailed look at the MMAU benchmark for language models.

This article examines how embedding initialization affects transformer model performance.

This article analyzes the effectiveness and reliability of steering vectors in language models.

Analyzing the storytelling capabilities of large language models compared to human authors.

A new benchmark assesses language models on scientific coding challenges across multiple fields.

Research reveals vulnerabilities in watermarking methods for AI-generated text.

An examination of how LLMs perform on the Abstraction and Reasoning Corpus.

An analysis of LLM performance on grid puzzles to assess reasoning abilities.

This article examines multi-prompt decoding to enhance text generation quality.

MIBench tests multimodal models' performance on multiple images.

A new method enhances LLM efficiency in creating complex hardware designs.

Analyzing the effectiveness of RAG and long-context LLMs in processing text.

A study on language agents' behavior in a social deduction game.

A new method to detect and fix factual errors in storytelling.

A new method enhances math solving skills in smaller language models using DPO and self-training.

New methods for personalizing AI language models are essential for user diversity.

A look into how language models handle arithmetic tasks and their learning process.

A toolkit designed for better evaluation of human-bot interactions.