Phonepadith/aidc-llm-laos-24k-gemma-3-4b-it
The Phonepadith/aidc-llm-laos-24k-gemma-3-4b-it model is a 4.3 billion parameter Gemma-3-4B-IT variant, fine-tuned by Phonepadith Phoummavong. It is specifically adapted for the Lao language, focusing on summarization tasks for various Lao content types including news, government documents, and meeting minutes. This model excels at generating concise summaries from Lao text, making it suitable for applications requiring natural language understanding and generation in Lao.
Loading preview...
Overview
This model, Phonepadith/aidc-llm-laos-24k-gemma-3-4b-it, is a specialized version of Google's Gemma 3-4B Instruct model, fine-tuned by Phonepadith Phoummavong. It is uniquely designed for Lao language processing, with a strong emphasis on summarization. The model was trained on a substantial dataset of 24,000 supervised Lao samples, including over 5,000 high-quality input-output pairs focused on AIDC (Artificial Intelligence and Digital Community) topics.
Key Capabilities
- Lao Language Generation: Specifically adapted to understand and generate responses in Lao.
- Summarization: Optimized for creating concise summaries from various Lao texts, such as news, government documents, and meeting minutes.
- General Chat & QA: Also suitable for general chat and question-answering in Lao.
Training Details
The model utilizes LoRA with PEFT (Unsloth) for fine-tuning, with a sequence length of 2048 and trained for 3-5 epochs. It is available in GGUF format (F16, Q8_0) and licensed under Apache 2.0.
Good For
- Developers building applications requiring Lao text summarization.
- Projects needing a Lao-specific language model for generation and understanding.
- Use cases involving Lao content analysis and information extraction.