Get Started
Home
Topics
Search
Library
Research questionHow reliably can language models assess recipe suitability for diabetes from dietary guidelines?Assessing a recipe requires identifying its ingredients and cooking methods, retrieving relevant diabetes dietary guidance, and applying that guidance consistently. Errors can arise at any of these stages when the recipe or medical guidance is expressed in natural language.
AI
Evaluation & Benchmarks
Health
LLM Pretraining & Post-training
Natural Language Processing
Reasoning
Research Paper
Latest papersRecent research connected to this question, newest first.Investigating the Ability of Large Language Models to Analyze Recipes for DiabetesThe evidence covers 7,607 recipes, including 3,807 labeled suitable and 3,800 labeled unsuitable, evaluated with direct-query, context-guided, and exemplary-context prompts containing varying amounts of diabetes dietary guidance from medical sources. Reported findings address guideline retrieval, recipe decomposition, and guideline-based reasoning across tested LLMs; Mistral-7B and Llama 70B performed best among the compared models.research paper · Sep 3, 2026
Related questions
How can we assess whether language models reliably answer or refuse questions grounded in FDA drug labels?How can language models reason iteratively to diagnose complex clinical cases safely and accurately?How can bias evaluations of large language models diagnose affected groups and reasons behind biased outputs?How reliably can language and vision-language models answer veterinary clinical questions with retrieval or supervised adaptation?