Indian Jurisprudence Chain-Of-Thought Reasoning Dataset

Indian Law Dataset contains 47,789 rows of structured reasoning traces across over 10 branches of Indian jurisprudence. It provides approximately 50 million tokens generated with advanced distillation techniques for training reasoning-capable language models.

Use Cases

• Train language models on structured chain-of-thought reasoning traces for legal argument generation
• Benchmark model performance across 10+ branches of Indian law using the provided reasoning traces
• Fine-tune models for specialized legal reasoning tasks within the Indian jurisprudence domain

Strengths

• 47,789 rows of structured data for model training
• Approximately 50 million tokens provide substantial textual content
• Covers over 10 distinct branches of Indian law for broad domain coverage

Limitations

• Data is synthetically generated via distillation, not original legal texts
• No column or feature details are provided to assess data structure
• Potential for bias or artifacts inherent in the generation method

Provenance

Source

Hugging Face dataset by author 169Pi.

Collection Method

Generated with advanced distillation techniques.

Time Range

null

Freshness

Last updated March 2026.

Geography

India