mlsanityData to 14 Oct 2026

Agreement versus Softmax under Spatial Disambiguation

arXiv 2610.14210 ↗First posted 9 Oct 2026cs.LG cs.CVcited by 0

Not on the map. Why?

Abstract (arXiv)

ScanRefer Multiple queries require spatial disambiguation: several instances of the target class are present, and the referencer must pick one. When per-query logits were never logged, a common substitute is cross-epoch agreement. On the full ScanRefer validation set (9,508 queries; GeoClassifier with ground-truth proposals), object-level softmax is similarly miscalibrated on Unique and Multiple (ECE 0.068 vs. 0.102; the scene-bootstrap gap interval includes zero). Fixed-final agreement on the same checkpoints reports ECE 0.041 vs. 0.272 (gap +0.231). Modal-vote agreement is 0.045 vs. 0.318. Most Multiple errors are same-class wrong-instance (45.6%); geometric miss (H2) is 59.6%, and wrong-category (H1) is 16.7%. A non-monotone H1 pattern does not appear under the default fixed-final protocol, and we retract it. Logit-free thresholds rank spatially hard queries poorly. If only agreement was logged, calibrate abstention on Multiple and report wrong-instance rates with H1.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 10 of its 14 references that are in the map. First posted on arXiv 9 Oct 2026.

Builds on

10 of 14 references.

References · 10

  1. 2017On Calibration of Modern Neural Networkscited by 4,960
  2. 2016Simple and Scalable Predictive Uncertainty Estimation using Deep Ensemblescited by 3,926
  3. 2017ScanNet: Richly-annotated 3D Reconstructions of Indoor Scenescited by 3,117
  4. 2016A Baseline for Detecting Misclassified and Out-of-Distribution Examples in Neural Networkscited by 2,385
  5. 2022Language Models (Mostly) Know What They Knowcited by 1,856
  6. 2023Evaluating Object Hallucination in Large Vision-Language Modelscited by 1,729
  7. 2017Selective Classification for Deep Neural Networkscited by 852
  8. 2019ScanRefer: 3D Object Localization in RGB-D Scans using Natural Languagecited by 452
  9. 2023Chat-Scene: Bridging 3D Scene and Large Language Models with Object Identifierscited by 175
  10. 20263D-VCD: Hallucination Mitigation in 3D-LLM Embodied Agents through Visual Contrastive Decodingcited by 3

Built on by

Not cited yet.