This paper reviews the construct validity of LLM benchmarks to ensure effective evaluation of safety and robustness in language models.

This paper discusses retrieval-augmented generation (RAG) and its impact on large language models by integrating external knowledge.

A longer summary here that is at least 120 characters long, describing the enterprise deep research multi-agent system and its capabilities.

TimeCopilot is an open-source framework that automates forecasting using time series models and language models.

This paper discusses the need for automated vulnerability detection in complex software systems using AI.

This study explores the importance of preference data characteristics in Direct Preference Optimization for aligning LLMs with human preferences.