Berkeley Function-Calling Leaderboard multi-turn scoring

This package uses BFCL multi-turn item and evaluator semantics from the
Berkeley Function-Calling Leaderboard / Gorilla evaluation project.

Sources:
  Dataset: gorilla-llm/Berkeley-Function-Calling-Leaderboard
  Evaluator repository: https://github.com/ShishirPatil/gorilla
  Eval revision: 6ea57973c7a6097fd7c5915698c54c17c5b1b6c8

License:
  Apache License, Version 2.0
