# PAct Official Failure Modes — phenomena run

- Dataset validation: `True`; issues: `0`
- Selection seed: `20260521`
- Mean weighted score: `46.47`
- Mean joint F1: `0.211`
- Mean part count MAE: `0.00`

## Samples
- #19 `GRScenes` `architectural_fixtures` score=52.52, joint_f1=0.333, VLM: 
- #22 `ArtVIP` `household_items` score=36.09, joint_f1=0.000, VLM: 
- #27 `ArtVIP` `household_items` score=55.62, joint_f1=0.500, VLM: 
- #35 `ArtVIP` `household_items` score=35.50, joint_f1=0.000, VLM: 
- #72 `PartNetMobility` `major_appliances` score=52.60, joint_f1=0.222, VLM: 
