Datasetpublic

stindardlogic/rag-systems-sft-100k

RAG Systems SFT 100K A synthetic supervised fine-tuning dataset of 100,000 high-quality conversations covering Retrieval-Augmented Generation (RAG) systems — from basic pipelines to advanced multi-hop retrieval, evaluation, and production optimization. Designed to train AI assistants that can help engineers build, debug, and scale RAG applications. Dataset Description This dataset covers the full spectrum of RAG system development across 12 specialized categories.… See the full description on the dataset page: https://huggingface.co/datasets/stindardlogic/rag-systems-sft-100k.

sourceHugging Faceapache-2.0updated 2mo agoView on Hugging Face

0likes57downloads

Dataset Card

RAG Systems SFT 100K

A synthetic supervised fine-tuning dataset of 100,000 high-quality conversations covering Retrieval-Augmented Generation (RAG) systems — from basic pipelines to advanced multi-hop retrieval, evaluation, and production optimization. Designed to train AI assistants that can help engineers build, debug, and scale RAG applications.

Dataset Description

This dataset covers the full spectrum of RAG system development across 12 specialized categories. Each record follows the ShareGPT format with a practitioner-level question and a detailed, implementation-ready response with working Python code.

Categories

Category	Description
`rag_architecture`	Basic pipeline, advanced patterns (HyDE, RAG-Fusion, FLARE, GraphRAG)
`chunking_strategies`	Fixed-size, semantic, recursive, parent-child chunking
`vector_databases`	Pinecone vs Weaviate vs ChromaDB vs pgvector
`embedding_models`	text-embedding-3-small, BGE, sentence-transformers comparison
`retrieval_evaluation`	RAGAS metrics, debugging RAG failures
`hybrid_search`	BM25 + dense embeddings, RRF reranking
`reranking`	Cross-encoder reranking with Cohere, FlashRank
`contextual_compression`	LLMChainExtractor, EmbeddingsFilter, context window optimization
`citation_generation`	Generating cited responses linked to source documents
`multi_hop_rag`	Multi-step chained retrieval for complex questions
`rag_production`	Caching, latency optimization, cost control
`document_ingestion`	PDF, HTML, code, table parsing and loading

Format

ShareGPT format:

json

{
  "conversations": [
    {"from": "human", "value": "...RAG question..."},
    {"from": "gpt", "value": "...implementation-ready response with code..."}
  ],
  "metadata": {"category": "...", "context": "..."},
  "id": "uuid"
}

Use Cases

—Fine-tuning AI assistants for RAG system design and debugging
—Training models to reason about retrieval quality and evaluation
—Building AI-assisted RAG development tooling
—Educating teams on production RAG patterns

Quality Notes

All responses include working Python code examples using LangChain, LlamaIndex, and direct API integrations. Covers RAGAS evaluation, hybrid search implementation, and production deployment patterns.