Datasets:
Tasks:
Text Classification
Formats:
parquet
Languages:
English
Size:
1K - 10K
Tags:
structured-decisions
calibration
probabilistic-classification
system-one
workflow-evaluation
Synthetic
License:
Laya benchmark results
#2
by convaiinnovations - opened
Laya benchmark results and ft model at: https://huggingface.co/convaiinnovations/laya-typed-decisions
Kaggle notebook to replicate at https://www.kaggle.com/code/nandukuttan/laya-typed-decisions-bm
Gentlemen
Added to the second table (fitted on train) at 0.766, from the number you published. We have not re-run it, and the row is marked self-reported. If you have KL and Brier from the full distributions, send them and we will fill the row in.