A Dataset and Benchmarks with Natural Language Critiques and Narratives

View the PDF file from the paper entitled: Data set and standards with criticism and natural linguistic novels, by Kon Soo and 10 other authors

PDF HTML (experimental) view

a summary:This paper offers a new data collection (reviews that have been strengthened with obstetric novels), designed to measure the conversation capabilities of the LLMS, tackling restrictions on current data groups that focus mainly on the prediction of the serial element. Regen extends the Amazon Product Review Data set via Inpainting two main natural language features: (1) criticizes the user, and represents the user’s “guidance” quotes that lead to the selection of a later element, (2) novels, rich text outputs associated with each component considering the previous context. The novels include product approvals, purchase, and user preference summaries.

Moreover, we create a criterion for the comprehensive modeling of the conversation recommendation task, as models are trained to create both recommendations and corresponding accounts on the user’s history (elements and criticisms). For this shared task, we offer LUMEN modeling framework (LLM -based multi -task model with criticism, recommendations and novels) that use LLM as a backbone for criticism, retrieval and obstetrics. We also evaluate the quality of the data set using standard automatic classification techniques and measure them by training both the traditional LLM -based traditional recommendations. Our results show that merging criticism enhances the quality of the recommendation by enabling the testament to learn to understand the language and integrate it with the recommendation signals. Moreover, the LLMS trained on our data collection, effectively generates both the recommendations and contextual narratives, which achieves the same performance of the recommendations of models and language models.