Datasets:
Tasks:
Text Classification
Formats:
parquet
Languages:
English
Size:
1K - 10K
Tags:
structured-decisions
calibration
probabilistic-classification
system-one
workflow-evaluation
Synthetic
License:
Add ZhiPan2-4B (#70) to the fitted table
#71
by sandeep85 - opened
Rescored from the public test predictions with bench/score_with_jobs.py::score_one: 0.7710 accuracy, KL 0.1105, Brier 0.0587, ECE 0.1625 over all 2,000 decisions, matching the submission. Fitted rank 25.
codelion changed pull request status to merged