udold/czech-legal-rag-knowledgebase

Czech Legal RAG Knowledgebase Chunked Czech legal text corpus for Retrieval-Augmented Generation (RAG). Contents 7,002 text chunks from 500 Czech laws (roslein/Czech_legal_code) Chunk size: 256 words with 64-word overlap Embedding model: paraphrase-multilingual-MiniLM-L12-v2 (384-dim) Index: FAISS IndexFlatIP Format Intended Use Grounding Czech legal advisor LLMs with retrieved statute text to reduce hallucination.… See the full description on the dataset page: https://huggingface.co/datasets/udold/czech-legal-rag-knowledgebase.