Rithaji-1.5B-Q4_K_M-GGUF

This repository contains the quantized GGUF format weights for Rithaji-1.5B, a custom language model fine-tuned for instruction-following and Python code generation.

The model was converted to 4-bit GGUF precision (Q4_K_M) using llama.cpp via the Hugging Face GGUF-my-repo space. This format is optimized for fast execution on consumer hardware, particularly inside local runtime environments like Ollama, LM Studio, and llama.cpp.

Original Model Reference

• Unquantized 16-bit Base Model: Rithaji-AI/Rithaji-1.5B
• Architecture: Qwen 2.5 (1.5B Parameters)
• License: Apache 2.0

Training Data & Legal Attribution

The underlying model was fine-tuned using the following datasets. We maintain strict compliance with their open-source attribution requirements:

• Databricks Dolly 15k: Utilized for general instruction-following. Licensed under CC-BY-SA 3.0. Copyright (2023) Databricks, Inc.
• Google MBPP (Mostly Basic Python Problems): Utilized for Python code synthesis. Licensed under CC-BY 4.0. Created by Google Research.

---

Deployment & Usage

1. Natively via Ollama

You can download and run this model locally in your terminal instantly using Ollama's direct Hugging Face integration:

ollama run hf.co/Rithaji-AI/Rithaji-1.5B-Q4_K_M-GGUF

2. Via llama.cpp CLI

llama-cli --hf-repo Rithaji-AI/Rithaji-1.5B-Q4_K_M-GGUF --hf-file rithaji-1.5b-q4_k_m.gguf -p "Write a Python function to sort a list."

Base model

Qwen/Qwen2.5-1.5B

Qwen/Qwen2.5-1.5B-Instruct

unsloth/Qwen2.5-1.5B-Instruct-bnb-4bit

Rithaji-AI/Rithaji-1.5B

(2)