mevsg/rumi-correction-v1
mevsg/rumi-correction-v1 is a 9 billion parameter Gemma 2 model developed by hyhyhyhyyhyh, fine-tuned with QLoRA (r=lora_alpha=4) from aisingapore/Gemma-SEA-LION-v3-9B-IT. This model specializes in Malay and English language processing, specifically designed for correcting messy or mistyped Malay Rumi spelling. Its primary use case is to accurately rewrite shortened or incorrect Malay Rumi text into proper spelling.
Loading preview...
Overview
mevsg/rumi-correction-v1 is a 9 billion parameter language model developed by hyhyhyhyyhyh, based on the Gemma 2 architecture. It was fine-tuned using QLoRA with parameters r = lora_alpha = 4 from the aisingapore/Gemma-SEA-LION-v3-9B-IT model. This model is specifically designed to address the challenge of correcting Malay Rumi spelling, handling shortened or mistyped text.
Key Capabilities
- Malay Rumi Spelling Correction: Excels at identifying and correcting errors in Malay Rumi text, transforming messy input into grammatically and orthographically correct Malay.
- Bilingual Processing: Supports both Malay and English languages, with a primary focus on Malay Rumi correction.
- Efficient Fine-tuning: Utilizes QLoRA for efficient adaptation, making it suitable for specialized tasks without requiring extensive computational resources for training.
Good for
- Automated Text Correction: Ideal for applications requiring automatic correction of user-generated Malay text, such as chat applications, social media, or data entry systems.
- Language Processing Tools: Can be integrated into tools for improving the quality of written Malay, assisting users with proper spelling and grammar.
- Research in Low-Resource Languages: Provides a specialized model for Malay, contributing to resources for languages that may have fewer dedicated NLP tools.