Product
ModelLedger
A tool for tracking regression issues across large language model versions. It records each model version's performance on evaluation sets and automatically identifies capability degradation introduced in newer versions, supporting systematic model quality management over time.
Timeline (last 90 days)
Sep 29
ModelLedger is responsible for tracking model regressions across versions, recording each model version's performance on evaluation sets and identifying which capabilities have degraded in new versions.
Unverified50%