MolTrace · ML Evaluation

ML Evaluation

Run evaluations against dataset versions or benchmark registry datasets; review metrics and summaries only.

Checking backend

Privacy

Tables show IDs and summaries — not raw knowledge records or source text. Full payloads stay in collapsed developer JSON for audit use.

Launch

Run evaluation

Launch a model evaluation run against a dataset version or benchmark dataset. Evaluation split and configuration are stored with the run for reproducibility.

model_artifact_id

Evaluation data target

Dataset version

Benchmark dataset (registry id)

dataset_version_id

evaluation_split (metadata_json)

notes (optional → notes_json)

Runs

Evaluation runs

All model evaluation runs — status, metric summary, and artifact linkage. Select Load detail to expand full metrics and per-split breakdowns.

Loading…

Loading ML Evaluation…

Command Palette

ML Evaluation

Run evaluation

Evaluation runs

ML Evaluation

Run evaluation

Evaluation runs