Verbosity penalty and length bias calibrator normalizing raw model rewards against response token lengths
mcp alignment developer-tools python-standard-library length-bias rlhf agent-skills reward-calibration verbosity-penalty score-normalizer eval-metrics
-
Updated
Sep 30, 2026 - Python