Projects AI/ML
10
Indonesian Legal LLM Lab
An experimental project exploring Qwen 2.5 1.5B fine-tuning for Indonesian legal text. Used Unsloth and QLoRA alongside retrieval experiments to study adapting a small language model to a specific domain.
Project Preview 01 / 01
Ownership
Personal Project
Role
AI/ML Engineer
Team
Independent Project
My Role
- 01 Prepared Indonesian legal text for model adaptation experiments.
- 02 Explored QLoRA fine-tuning with Unsloth and Qwen.
- 03 Compared fine-tuning with retrieval-based approaches.
Features
- 01 Legal-domain dataset exploration
- 02 QLoRA fine-tuning
- 03 Small language model adaptation
- 04 RAG experiments
Impact
- Established a practical lab for evaluating domain-specific Indonesian language models.
Tech Stack
UnslothQLoRAQwenRAGHugging Face