harrier-oss-v1-0.6b-GGUF

Microsoft Harrier - GGUF Embedding Model

I created this repository to provide the GGUF-quantized versions of the microsoft/harrier text embedding model.

As we know, Microsoft officially only provides raw weights and safetensor files, not GGUF. However, local inference engines like Ollama usually prefer the GGUF format for seamless execution. That's exactly why I made this embedding model.

By the way, you can also find other conversions out there like the "Super Pauly" model (SuperPauly/harrier-oss-v1-0.6b-gguf). But to be honest, I can't trust them because I don't know who he is, so I decided to create my own from scratch to ensure complete security and transparency. I didn't want to steal anything; everything here is fully open source, and you are free to use it!

I've also added a conversion script below so you can see exactly how to convert the original models yourself.

---

Model Details

Model Description

The harrier model is an advanced, decoder-only text embedding model developed by Microsoft. This GGUF version allows developers to generate local, privacy-preserving semantic vectors efficiently without relying on external APIs or managing heavy Python execution environments.

• Model type: Text Embedding Model (GGUF Format)
• License: MIT (inherited from the base model)
• Base model: microsoft/harrier

Model Sources

• Original Repository: microsoft/harrier
• Framework Support: Ollama, llama.cpp, AnythingLLM

---

Uses

Direct Use

This model is intended for developers building Retrieval-Augmented Generation (RAG) pipelines, semantic search architectures, or document clustering systems locally on consumer hardware.

Out-of-Scope Use

This is strictly an embedding model designed for retrieval and semantic representation. It cannot be used for text generation, conversational chat, or general-purpose instruct-following tasks.