openpql/nishka-gkc-phi3-merged
The openpql/nishka-gkc-phi3-merged model is a 3.8 billion parameter Phi-3 architecture developed by openpql, specifically fine-tuned for governance and compliance tasks. It was trained on 1.12 million tokens of regulatory content spanning 15 compliance frameworks. This model excels at processing and understanding regulatory information, making it suitable for applications requiring specialized knowledge in governance and compliance.
Loading preview...
NISHKA GKC: Governance Knowledge Model
This model, developed by openpql, is a specialized language model built upon the Phi-3-mini-4k-instruct base architecture. It features 3.8 billion parameters and has been extensively fine-tuned using a LoRA adapter, which is now merged into the full model weights, eliminating the need for separate adapter loading.
Key Capabilities
- Specialized Governance Knowledge: Trained on a Governance Knowledge Corpus (GKC) comprising 1.12 million tokens of regulatory content.
- Compliance Framework Coverage: Incorporates data from 15 distinct compliance frameworks, providing a broad understanding of regulatory landscapes.
- Efficient Deployment: Ready for deployment with popular inference servers like vLLM and TGI, supporting
float16for optimized performance.
Good For
- Regulatory Analysis: Understanding and interpreting complex regulatory documents.
- Compliance Automation: Developing applications that require deep knowledge of various compliance standards.
- Information Extraction: Extracting specific details from governance-related texts.
This model is designed for developers and organizations needing a robust solution for tasks centered around regulatory compliance and governance knowledge processing.