New PASTA framework updates LLMs with current news knowledge

Researchers have developed PASTA, a new framework designed to update Large Language Models (LLMs) with current factual information, particularly from news articles. This approach combines data augmentation, question-generation, and a self-learning process to overwrite outdated knowledge and reduce hallucinations. Experiments show PASTA significantly improves accuracy in answering questions about recent events, boosting performance from 0.02 to 0.82 while preserving general language abilities. AI

IMPACT This research offers a method to improve LLM accuracy on recent factual information, potentially enhancing their utility for real-time knowledge tasks.

RANK_REASON The cluster contains an academic paper detailing a new method for updating LLMs. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.CL →

• Large Language Models
• LLMs
• PASTA

• paper
• model release

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →