carpedm20/locate-anything-tensorrt

LocateAnything TensorRT Artifacts

Results and visual examples

Locate Anything TensorRT artifact archive. Model/build details are in the per-variant README and artifact manifests. An exported engine alone is not visual quality evidence; no exact-engine saved gallery was recovered.

Available build records: rtx5090-sm120-visual-features-j-20260702/README.md, rtx5090-sm120-visual-features-j-20260702/artifact_manifest.json, rtx5090-sm120-visual-features-j-20260702/nvidia-locateanything-3b_visual_features_fp16.metadata.json, rtx5090-sm120-visual-features-j-fp8-20260702/README.md, rtx5090-sm120-visual-features-j-fp8-20260702/artifact_manifest.json, rtx5090-sm120-visual-features-j-fp8-20260702/nvidia-locateanything-3b_visual_features_fp8.metadata.json.

No checkpoint-matched photo gallery was located in the existing evidence reviewed for this update. The files and runtime records below are the available evidence; no visual quality claim is inferred from their presence.

Evidence provenance

This documentation update reuses saved results; it does not rerun inference or change weights. The starting repository revision is 0e9008c33056. Captions distinguish model predictions, training diagnostics, and aggregate metrics. Qualitative examples are not a representative accuracy estimate.

Gallery sources and SHA-256 checksums.

Fury-maintained TensorRT artifact cache for nvidia/LocateAnything-3B.

Artifacts are stored under revision folders on main:

carpedm20/locate-anything-tensorrt@main/<engine_revision>/

These are visual-feature-stage artifacts, not full native TensorRT detector engines. The decoder remains in the pinned LocateAnything helper runtime.