A naukri.com initiative
Arxiv
1d
155
Image Credit: Arxiv
REGEN: A Dataset and Benchmarks with Natural Language Critiques and Narratives
- The paper introduces a novel dataset, REGEN, designed to benchmark the conversational capabilities of recommender Large Language Models (LLMs).
- REGEN extends the Amazon Product Reviews dataset by including user critiques and narratives associated with recommended items.
- An end-to-end modeling benchmark is established for conversational recommendation using the LUMEN framework that incorporates LLMs for critiquing, retrieval, and generation.
- Results show that incorporating critiques in recommendations enhances quality and LLMs trained on the dataset effectively generate recommendations and contextual narratives comparable to state-of-the-art models.
Read Full Article
9 Likes
For uninterrupted reading, download the app