Leaderboard

Leaderboard / Evaluation Scores Table

Leaderboard Service

In this service, Large Language Models (LLMs) are evaluated and ranked on tasks in the fields of Islamic sciences and humanities.

Dataset Used

These evaluations are conducted using the HIKMA dataset (Humanities and Islamic Knowledge Assessment), which has been prepared by the Noor Computer Research Center for Islamic Sciences.

Service Features

• Specialized Evaluation: Focus on tasks in Islamic sciences and humanities
• Transparent Ranking: Comparison of different model performances in a score table format
• Standardized Data: Using the HIKMA dataset for consistent and comparable evaluation
• Continuous Updates: Adding new models and updating results

Applications

• Model Selection: Helping researchers choose the best model for relevant tasks
• Performance Comparison: Observing strengths and weaknesses of different models
• Research and Development: Using results to improve existing models
• Scientific Transparency: Providing standardized and citable evaluations

Benefits

• Specialized evaluations in Islamic sciences and humanities
• Use of proprietary and authentic HIKMA dataset
• Fair and comparable ranking
• Enabling informed decision-making for users