SindhiLM-Qwen-0.5B Overview

SindhiLM-Qwen-0.5B is a specialized Large Language Model (LLM) for the Sindhi Language, developed by Aakash Meghwar as part of the broader SindhiLM Project. It is a fine-tuned version of the Qwen2.5-0.5B architecture, specifically optimized for Sindhi text generation and understanding.

Key Capabilities & Features

• Sindhi Language Specialization: Designed from the ground up to excel in Sindhi NLP tasks.
• Compact & Efficient: A 0.5 billion parameter model, trained with 4-bit LoRA for 10,000 steps, making it suitable for deployment on consumer-grade hardware.
• "Green AI" Philosophy: Developed with efficiency and accessibility in mind, particularly for low-resource languages.
• Custom Tokenization: Utilizes a custom-built BPE tokenizer tailored for the Sindhi language.

Development & Future

This model was developed by Aakash Meghwar, a Computational Linguist specializing in South Asian languages. Future iterations, such as SindhiLM-v2, are planned to include an improved Sindhi-BPE tokenizer and knowledge distillation techniques. The author is actively seeking collaborations and PhD opportunities in Low-Resource NLP, Literary Informatics, and South Asian Language Technology.