This is a fine-tuned version of Qwen2.5-7B-Instruct, specifically trained to speak in Jordanian dialect. The Model was initially trained to be a part of a Graduation project from a university in Jordan, and this model is the first open source LLM that is trained for the Jordanian Dialect. It was trained using consumer hardware (RTX 4080 16gb VRAM).

Note that this model was made to be used for a RAG Chatbot, so its a question answering model. Its not ready for production and wasn't tested enough.

Training Data

The data used to train the Model consistest of 1500 question and answer paris, created based on actual universtiy rules and regulastion using real jordanian dialect words.

Usage with Ollama

Since this model includes a custom Modelfile, you can set it up in Ollama:

Download the Modelfile and qwen2.5-7b-instruct.Q4_K_M-001.gguf file to the same folder.
Run the following command in your terminal within that folder:

ollama create jordanian-model -f Modelfile

KareemBb/Qwen2.5-7B-Instruct-Jordanian

Qwen2.5-7B-Instruct-Jordanian

Training Data

Usage with Ollama

KareemBb/
Qwen2.5-7B-Instruct-Jordanian