OSS maintenance triage, research metrics, and live repository overview
ML Evaluation Workspace
Keep the model story clear: one page for the signal, one for the repos, one for the artifacts.
Overview stays focused on AUROC, Brier, inactivity rate, and calibration. Dataset, Repositories, and Runs hold the deeper inspection views so the workflow stays readable.
Overview
Live quality snapshot
Dataset
Coverage and feature inventory
Repos
Stars, notes, and activity
Runs
Cached artifacts and splits
Snapshots
0
Current analysis-backed training base
Repositories
0
Unique repos visible to the trainer
Labeled rows
0
Coverage Pending
Inactive 12m rate
Pending
Held-out class balance proxy
Training Base
How the current OSS base looks before it reaches the model
Latest Artifact Coverage
Training-data results surfaced directly from the cached run
Rows in artifact
0
Evaluation sample
0
Latest hash
Pending
Artifact status
Loading
This page is intentionally about the training base itself: which OSS projects are represented, how much data exists, how much is labeled, how imbalanced the held-out slice is, and which features flow into the model.
Training OSS Projects
Ranked repositories in the current base
Repository submissions use the same scoring surface as these base projects, so a newly searched repo can be compared against the captured OSS population.
Run repository analyses first to populate the visible OSS training base.
Feature Inventory
Latest training feature set
Run training once to surface the feature inventory from the latest cached artifact.