d-michail/Mitos-Qwen3-8B-Process
Mitos-Qwen3-8B-Process is an 8 billion parameter Qwen3-based language model developed by d-michail, specifically fine-tuned for question answering on Greek e-government procedures. It was trained on 90,145 question-answer records derived from the National Registry of Administrative Procedures (Mitos), enabling it to generate longer, structured answers consistent with registry entries. This model excels in research on Greek public administration QA, particularly when combined with retrieval-augmented generation, and has a context length of 32768 tokens.
Loading preview...
Mitos-Qwen3-8B-Process Overview
Mitos-Qwen3-8B-Process is an 8 billion parameter language model, based on Qwen/Qwen3-8B, developed by d-michail. It is specifically fine-tuned for question answering related to Greek e-government procedures, drawing its knowledge from the National Registry of Administrative Procedures (ΕΜΔΔ / Mitos). The model was trained on 90,145 question-answer records, generated from the structured text of all 4,407 procedures within the registry, using templates without an LLM.
Key Capabilities
- Specialized Greek QA: Adapted to answer questions about Greek public administration procedures, producing answers that reflect the structure and content of official registry entries.
- Structured and Longer Responses: Generates more detailed and structured answers compared to a base model, aligning with the format of administrative procedure descriptions.
- Research on Public Administration QA: Primarily intended for research in closed-book and retrieval-augmented generation (RAG) scenarios for Greek public administration.
- High Context Length: Supports a maximum sequence length of 4096 tokens during training, with the base Qwen3 model supporting 32768 tokens.
Good for
- Researching Greek e-government information retrieval: Ideal for academic or applied research into how LLMs can process and respond to queries about complex administrative procedures in Greek.
- Developing RAG systems for public services: When combined with retrieval mechanisms, this model significantly improves key-fact recall for public administration questions, making it suitable for building informed QA systems.
- Understanding administrative language: The model learns the specific register and format of the Greek public administration domain, which can be valuable for linguistic analysis or content generation in this area.