Berkeley Function-Calling Leaderboard Domain-Specific Benchmark

What is the Berkeley Function-Calling Leaderboard Domain-Specific Benchmark?

The Berkeley Function-Calling Leaderboard (BFCL) is a comprehensive evaluation framework designed to assess function-calling capabilities across various programming languages, such as Python, Java, JavaScript, and REST API. Utilizing 2,000 question-answer pairs, BFCL supports multiple and parallel function calls and includes function relevance detection.

Key Highlights

• Multilingual Support: Tests performance in languages like Python, Java, and more.
• Advanced Capabilities: Evaluates both single and multiple function calls.
• Relevance Detection: Assesses the contextual relevance of the function calls.

Resources

• BFCL Dataset: GitHub Repository
• Research Documentation: BFCL Paper