YAML Metadata Warning:empty or missing yaml metadata in repo card
Check out the documentation for more information.
NyayaSetu Legal Assistant
Model Summary
NyayaSetu Legal Assistant is a fine-tuned legal support model built using AutoScientist on the Adaption platform. The model is designed to provide simplified legal guidance, explain legal procedures, assist citizens in understanding government schemes, and improve access to legal information in an easy-to-understand format.
This model was developed as part of the Adaption AutoScientist Challenge.
Model Details
Model Description
NyayaSetu Legal Assistant adapts the Llama-4-Scout-17B-16E-Instruct model for legal-awareness and legal-assistance tasks. The model focuses on providing clear, citizen-friendly explanations of legal concepts, documentation requirements, rights, responsibilities, and government-related legal procedures.
- Developed by: Ananya Daitkar
- Training Platform: Adaption AutoScientist
- Model Type: PEFT / LoRA Fine-Tuned Large Language Model
- Language: English
- Domain: Legal Assistance
- Base Model: togethercomputer/Llama-4-Scout-17B-16E-Instruct
- License: Same as base model license unless otherwise specified
Intended Use
Direct Use
This model can be used for:
- Legal information assistance
- Citizen legal awareness
- Understanding government schemes
- Explaining legal documentation requirements
- Educational legal chatbots
- Public legal support systems
Downstream Applications
- Legal-help chatbots
- Government service assistants
- Legal education platforms
- NGO legal support tools
- Community legal awareness applications
Out-of-Scope Use
This model should NOT be used for:
- Professional legal advice
- Court representation
- Lawyer-client privileged matters
- High-risk legal decision making
- Legal opinions requiring licensed professionals
Users should always consult qualified legal professionals for important legal decisions.
Training Data
The model was trained on a custom legal assistance dataset containing:
- Legal question-answer pairs
- Citizen rights information
- Government scheme explanations
- Legal procedure guidance
- Documentation requirements
- Public legal awareness content
Data was adapted and optimized using Adaption AutoScientist.
Training Procedure
Method
- AutoScientist Optimization
- PEFT Fine-Tuning
- LoRA Adapters
Base Model
togethercomputer/Llama-4-Scout-17B-16E-Instruct
Framework
- PEFT
- Transformers
- Adaption AutoScientist
Evaluation
Goal
Improve legal-domain performance compared to the baseline model provided within the Adaption challenge environment.
Results
AutoScientist training produced measurable improvements over the baseline evaluation metrics on the Adaption platform.
The model achieved higher performance on legal-domain tasks after adaptation and optimization.
Risks and Limitations
- May generate incorrect legal information.
- May contain hallucinated responses.
- Laws vary by jurisdiction and change over time.
- Should not replace licensed legal professionals.
- Users should independently verify critical information.
Ethical Considerations
The model is intended to improve access to legal information and promote legal awareness. It is not intended to provide legally binding advice or replace professional legal services.
Example Use Cases
Example Question
"What documents are typically required for filing a consumer complaint?"
Example Response
The model explains common documentation requirements, filing procedures, and important considerations in simple language while encouraging users to verify requirements with the relevant authority.
Technical Details
Architecture
- Base Model: Llama-4-Scout-17B-16E-Instruct
- Fine-Tuning Method: LoRA / PEFT
- Training Framework: Adaption AutoScientist
Exported Files
- adapter_model.safetensors
- adapter_config.json
- tokenizer.json
- tokenizer_config.json
- config.json
Authors
Ananya Daitkar
Acknowledgements
- Adaption AutoScientist Challenge
- Together AI
- Hugging Face
- Open-source AI community
- Downloads last month
- 3