goasty/Qwen3-4B-Indian-Law
goasty/Qwen3-4B-Indian-Law is a 4 billion parameter language model, fine-tuned from Qwen3-4B, specifically adapted for Indian legal knowledge. It excels in Indian legal question answering, statutory interpretation, and case law understanding, leveraging a 4096-token context length. This model is specialized for legal research assistance and educational purposes within the Indian legal domain.
Loading preview...
Qwen3-4B Indian Law: Specialized Legal Assistant
This model, goasty/Qwen3-4B-Indian-Law, is a domain-adapted legal assistant fine-tuned from the Qwen3-4B base model. It has been specifically trained on an extensive corpus of Indian legal texts, including statutory provisions, constitutional law, criminal law, evidence law, procedural law, and court judgments.
Key Capabilities
- Indian Legal Question Answering: Provides answers to queries across various Indian legal domains.
- Statutory and Constitutional Interpretation: Assists in understanding and interpreting Indian laws and constitutional provisions.
- Case Law Understanding: Facilitates comprehension of court judgments and legal reasoning.
- Legal Research Assistance: Supports tasks like judgment summarization and general legal research.
- Broad Legal Coverage: Trained on diverse datasets covering Fundamental Rights, IPC, CrPC, BNS, BSA, and Supreme/High Court judgments.
Training Details
The model was fine-tuned using LoRA + QLoRA methods with Unsloth, employing a 4096-token context length. The training dataset was meticulously balanced to reduce bias towards judgment memorization, ensuring comprehensive coverage of general legal QA, court judgments, and specific legal codes. The final balanced dataset comprised over 300,000 samples.
Intended Use Cases
- Legal Research: Aiding legal professionals and students in research tasks.
- Educational Purposes: Supporting learning and understanding of Indian law.
- Legal Q&A Systems: Powering specialized question-answering applications.
- Retrieval-Augmented Generation (RAG): Enhancing RAG systems with domain-specific legal knowledge.
Limitations
It is crucial to note that this model is not a substitute for licensed legal counsel and may generate legally incorrect or outdated interpretations. It should not be relied upon for litigation or legal advice without expert review.