NyayaLabs98/nyaya-3b-v3
NyayaLabs98/nyaya-3b-v3 is a 3.1 billion parameter instruction-tuned causal language model, based on Qwen/Qwen2.5-3B-Instruct, specifically designed for Indian legal information. It answers everyday legal questions in English, Hindi, and Hinglish, citing specific sections of current Indian law (BNS/BNSS/BSA) and is optimized for use with Retrieval Augmented Generation (RAG) to provide accurate, plain-language legal guidance.
Loading preview...
Nyaya-3B-v3: Indian Legal Information Model
Nyaya-3B-v3 is a specialized 3.1 billion parameter language model developed by NyayaLabs98, built upon the Qwen/Qwen2.5-3B-Instruct base model. Its primary function is to provide legal information and procedural guidance for Indian citizens in English, Hindi, and Hinglish. A key feature is its ability to cite specific sections of current Indian law, including the BNS, BNSS, and BSA, and to bridge old (IPC) to new (BNS) legal terminology.
Key Capabilities
- Multilingual Legal Information: Answers legal questions in English, Hindi, and Hinglish.
- Statute Citation: Cites specific sections of current Indian law, such as BNS, BNSS, and BSA.
- Procedural Guidance: Offers "how-do-I" type procedural advice for common legal queries.
- Old-to-New Law Mapping: Helps understand the transition and mapping between older and newer Indian legal acts.
- RAG Optimized: Designed to perform best when used with a dense retriever supplying relevant statute passages, significantly improving citation accuracy.
Intended Use Cases
- Plain-language legal information: For general understanding of Indian law.
- Procedural guidance: "How to" questions related to legal processes.
- Terminology mapping: Understanding legal terms and their evolution.
Limitations and Important Notes
It is crucial to understand that Nyaya-3B-v3 is an information/guidance model and not a legal advisor. It explicitly signals when a real lawyer is needed and recommends consulting a licensed advocate for any consequential matters. The model's coverage is limited to the acts in its statute database, and its performance, particularly strict citation accuracy, is still under active development, as indicated by its early research release status.