davidr99/locateanything-3b-delivery-detector-r4-clean-original-aspect

LocateAnything-3B clean original-aspect delivery detector R4

Personal, non-commercial LocateAnything one-epoch SFT experiment.

Starting model: nvidia/LocateAnything-3B

Classes: Amazon, UPS, USPS-Truck, Other-Vehicles, FedEx

R4 starts directly from NVIDIA LocateAnything-3B and trains for one epoch on aspect-preserved, patch-aligned images. Every image uses the canonical full five-class prompt, matching R1's detection task while changing only source-image geometry.

Image content is capped at 640px on the longest side without enlargement or aspect distortion, then centered on a neutral-gray canvas aligned to 28px. Training coordinates are remapped to that padded canvas and emitted as normalized integers from 0 to 1000.

Prompt seed: 20260828

Query mixture: {"full": 1.0, "negative": 0.0, "positivesubset": 0.0, "singlepositive": 0.0}

License

This model is a derivative of NVIDIA LocateAnything-3B and remains restricted to non-commercial research/evaluation under NVIDIA's model license.