File eval_preference.py

toolbench/tooleval/eval_preference.py:None–None

# Evaluate a method outputs in different aspectes and update the leaderboard
# `result_folder` should contain the following 6 json files:
#   - `G1_category.json`: 
#           single-tool instruction;