QuantaSparkLabs/Chronos-3B

Hugging Face
TEXT GENERATIONConcurrency Cost:1Model Size:3.1BQuant:BF16Ctx Length:32kPublished:May 11, 2026License:apache-2.0Architecture:Transformer0.0K Open Weights Warm

Chronos-3B by QuantaSparkLabs is a 3.1 billion parameter retrieval-augmented generation (RAG) AI, combining a Qwen 2.5 3B language model with a FAISS-powered knowledge base of 20th-century Wikipedia articles. Optimized for historical accuracy, it provides evidence-backed answers on topics like World War I, World War II, and the Cold War, while maintaining conversational fluency. The model features a confidence threshold and safety net to prevent hallucinations, ensuring reliable historical information.

Loading preview...

Chronos-3B: Your 20th-Century Historian

Chronos-3B is a specialized retrieval-augmented generation (RAG) AI developed by QuantaSparkLabs, designed to provide accurate and engaging information about 20th-century history. It integrates a Qwen 2.5 3B language model with a FAISS knowledge base built from hundreds of Wikipedia articles covering major historical events, political upheavals, and key inventions. This architecture allows Chronos to act as a knowledgeable historian, capable of both casual conversation and detailed, evidence-backed historical explanations.

Key Capabilities

  • Hallucination Control: Employs a confidence threshold and cross-encoder re-ranking to minimize factual inaccuracies, preferring to admit ignorance rather than invent information. A hard-coded safety net ensures accuracy on critical historical facts.
  • Dual-Layer Architecture: Seamlessly switches between a casual chat mode (direct Qwen 3B response) and a historical RAG pipeline for 20th-century queries.
  • Historical Retrieval: Utilizes e5-base-v2 for initial retrieval and ms-marco-MiniLM-L-12-v2 for re-ranking relevant chunks from its historical index.
  • Personable Interaction: Handles greetings and small talk with a warm, lively personality, making interactions feel natural.

Good For

  • Accurate Historical Q&A: Ideal for developers needing a reliable source for 20th-century historical facts and explanations.
  • Building Conversational AI: Provides a strong foundation for chatbots requiring both general conversational abilities and specialized factual recall.
  • Educational Applications: Can be used in tools designed for learning and exploring modern history without the risk of common LLM hallucinations.

Finetunes
1 models