Indian Legal Text Understanding And Reasoning Benchmark

IL-TUR is a benchmark for AI models on Indian law, containing 8 foundational tasks requiring different types of understanding. The dataset includes tasks in English and some Indic languages, created by Exploration-Lab and updated in April 2025.

Use Cases

• Train models on the 8 foundational tasks for Indian legal text understanding and reasoning.
• Evaluate model performance on tasks involving Indic languages within the legal domain.
• Benchmark natural language understanding systems on the specific skills required by the IL-TUR tasks.
• Develop multi-task learning approaches using the unified data formats of the 8 benchmark tasks.

Strengths

• Contains 8 distinct foundational tasks for comprehensive legal AI evaluation.
• Includes data in multiple languages, covering English and some Indic languages.
• Provides a modern, unified benchmark specifically for Indian law.

Limitations

• Specific dataset size, row count, and column structure are unknown.
• Task details and sample data are unavailable for direct inspection.
• File formats and licensing information are not provided.

Provenance

Source

Exploration-Lab on Hugging Face.

Collection Method

Collaborative effort to establish a benchmark; data formats unified.

Time Range

null

Freshness

Last updated April 2025.

Geography

India