Hugging Face Overview
NISHKA SFT: Specialized PQL Code Generation
This model, developed by openpql, is a supervised fine-tuned (SFT) version of Microsoft's Phi-3-mini-4k-instruct, specifically engineered for Policy Query Language (PQL) code generation. With 3.8 billion parameters, it leverages the robust Phi-3 architecture to provide highly accurate and relevant PQL outputs.
Key Capabilities
- PQL Code Generation: Fine-tuned on a substantial dataset of 10,038 PQL examples, making it proficient in generating PQL code snippets.
- Compliance and Governance: Optimized for use cases involving policy definition, compliance checks, and governance automation through PQL.
- Phi-3 Base: Built upon the efficient and capable Phi-3-mini-4k-instruct model, ensuring a strong foundation for language understanding and generation.
- Ready for Deployment: Provided as full model weights, it's designed for straightforward deployment with inference servers like vLLM or TGI.
Good For
- Developers and organizations requiring automated generation of Policy Query Language (PQL).
- Integrating PQL code generation into applications for compliance, security, or data governance.
- Experimenting with a specialized, instruction-tuned model for domain-specific code generation tasks.