dragon-yi-6b-v0
| Property | Value |
|---|---|
| Parameter Count | 6.06B |
| Model Type | Yi |
| License | Apache 2.0 |
| Base Model | Yi-6B |
| Primary Language | English |
What is dragon-yi-6b-v0?
dragon-yi-6b-v0 is part of the dRAGon (Delivering RAG On) model series, specifically designed for enterprise-grade document analysis and question-answering. Built on the Yi-6B architecture, this model has been fine-tuned with a focus on processing complex business and legal documents while maintaining high accuracy and minimizing hallucinations.
Implementation Details
The model utilizes a transformer architecture implemented in PyTorch and is available in GGUF format. It employs a simple but effective prompt structure using "<human>" and "<bot>" wrappers for optimal performance. The model operates with a recommended temperature of 0.3 and supports up to 100 max_new_tokens for consistent output generation.
- Benchmark accuracy score of 99.5/100
- 90% accuracy in Not Found Classification
- 87.5% accuracy in Boolean queries
- 77.5% accuracy in Math/Logic operations
- Above-average performance in complex questions and summarization
Core Capabilities
- Fact-based question-answering over complex documents
- Key-value extraction from business and legal texts
- Document summarization with minimal hallucination
- Workflow automation for knowledge-intensive industries
- RAG-optimized responses with high accuracy
Frequently Asked Questions
Q: What makes this model unique?
The model's exceptional performance in RAG-specific tasks and its focus on minimizing hallucinations while maintaining high accuracy in business and legal document processing sets it apart. Its 99.5% benchmark accuracy and specialized training for enterprise use cases make it particularly valuable for professional applications.
Q: What are the recommended use cases?
The model is ideal for enterprise automation in knowledge-intensive industries, particularly financial services, legal, and regulatory sectors. It excels at question-answering, key-value extraction, and summarization tasks involving complex business documents.
