ClinicGuide RAG pipeline
This repository documents the ClinicGuide retrieval configuration. It does not contain a newly trained language model and does not ship LLM weights.
ClinicGuide answers from a FAQ dataset using hybrid lexical retrieval (word TF-IDF, character TF-IDF, BM25) plus optional rewriting through Hugging Face Inference Providers (Qwen/Qwen3-8B).
What is included
config.json— product name, languages, safety policy, chat model ids, retrieval hyperparameters
What is not included
- Fine-tuned Qwen / Llama checkpoints
- Patient data
- A claim that a new medical LLM was trained
Use with
- Dataset: alirezaaminzadeh/clinicguide-faq-corpus
- Space: alirezaaminzadeh/clinicguide-medical-tourism-assistant
Maintainer: Aria AI / alirezaaminzadeh