待翻譯：Local Gains and Fixed-Assignment Set Losses in Shared Set Decoders

AI 服務暫時不可用，以下為來源正文，待恢復後補全翻譯。

--> [Submitted on 12 Aug 2026] Title:Local Gains and Fixed-Assignment Set Losses in Shared Set Decoders View a PDF of the paper titled Local Gains and Fixed-Assignment Set Losses in Shared Set Decoders, by Ze Zhang and 1 other authors View PDF HTML (experimental) Abstract:A query-relation deletion can improve the edited slot while reducing the utility of the prediction set that contains it. We study this tension in two related ResNet-50 DETR-family checkpoints using recorded, selection-conditional evidence from 710 paired image-relation units per checkpoint. The primary comparison subtracts a matched active control, which deletes the same leader source at a different recorded recipient, from the selected target deletion. It is therefore a composite contrast rather than a same-recipient placebo. The target-minus-control contrast is locally positive and fixed-assignment negative in both checkpoints. The opposite-sign pattern occurs within 302/710 DETR units and 460/710 DINO units. After rematching, the corresponding counts are 285/710 and 433/710. Rematching and native selection absorb enough of the mean loss for DETR intervals to cross zero, whereas DINO intervals remain negative, so persistence across readouts differs by checkpoint. A fixed-map comparison between hard deletion and a mass-preserving edit also differs before rematching. That comparison is conditional on the outcome-blind map and does not establish same-dose transport. Local intervention success therefore does not determine the consequence for a jointly decoded set. The supported conclusion is selection-conditional deletion sensitivity whose persistence depends on the readout and intervention operator. We do not identify an intervention-invariant edge mechanism, detector-level degradation, population prevalence, or the value of a training-time regularizer. Comments: 13 pages, 4 figures, 2 tables. An ancillary analysis-ready package supports exact aggregate reproduction without model inference. Code and reproduction package: this https URL Subjects: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG) Cite as: arXiv:2608.14717 [cs.CV] (or arXiv:2608.14717v1 [cs.CV] for this version) https://doi.org/10.48550/arXiv.2608.14717 arXiv-issued DOI via DataCite Submission history From: Ze Zhang [view email] [v1] Wed, 12 Aug 2026 03:14:43 UTC (1,447 KB) Full-text links: Access Paper: View a PDF of the paper titled Local Gains and Fixed-Assignment Set Losses in Shared Set Decoders, by Ze Zhang and 1 other authors View PDF HTML (experimental) TeX Source view license Ancillary-file links: Ancillary files (details): ARTIFACT_MANIFEST.json DATA_AND_MODEL_PROVENANCE.md README.md analysis_ready/ANALYSIS_READY_PROVENANCE.json analysis_ready/gate_c_detr_r50_500.jsonl analysis_ready/gate_c_dino_r50_4s_12e.jsonl analysis_ready/h4_d_detr_r50_500.jsonl analysis_ready/h4_d_dino_r50_4s_12e.jsonl analysis_ready/t1_detr_r50_500.jsonl analysis_ready/t1_dino_r50_4s_12e.jsonl analysis_ready/t2_detr_r50_500.jsonl analysis_ready/t2_dino_r50_4s_12e.jsonl audit/H4_D_FULL_INTEGRITY_AUDIT.json audit/H4_GATE_C_D_FULL_INTEGRITY_AUDIT_V2.json audit/N1_REPLAY_INTEGRITY_AUDIT.json audit/T0B_D_LOCAL_FULL_INTEGRITY.json audit/T0B_REALIZED_DOSE_OVERLAP.json audit/T0B_R_PILOT_LOCAL_INTEGRITY.json audit/T0B_R_PILOT_MANIFEST.json audit/T1_REALIZED_DOSE_MAPPING.json audit/T2_CONFIRMATION_PAIR_MANIFEST.json audit/T2_CONFIRMATION_POPULATION_MANIFEST.json audit/T2_CONFIRMATION_V2_FULL_INTEGRITY.json audit/T2_INPUT_CAPTURE_INTEGRITY.json audit/T2_OUTCOME_BLIND_MAPPING.json audit/T2_REALIZED_DOSE_OVERLAP.json code/iclr_h4_aggregate.py code/iclr_h4_gate_c_aggregate.py code/iclr_n1_aggregate.py code/iclr_transport_t0.py code/iclr_transport_t0b_integrity.py code/iclr_transport_t0b_overlap.py code/iclr_transport_t1_aggregate.py code/iclr_transport_t2_aggregate.py code/iclr_transport_t2_confirmation_integrity.py code/iclr_transport_t2_input_overlap.py code/p1_artifact_manifest.py code/p1_export_analysis_ready.py code/p1_reproduce_from_analysis_ready.py code/reference_src/future_qc/iclr_h4/estimands.py code/reference_src/future_qc/iclr_n1/statistics.py configs/fq_iclr_h4_gate_c_aggregation_v1.json configs/fq_iclr_h4_gate_c_d_v2.json configs/fq_iclr_h4_set_externality_v1.json configs/fq_iclr_n1_aggregation_v1.json configs/fq_iclr_n1_instrumented_replay_v1.json configs/fq_iclr_transport_t0_v1.json configs/fq_iclr_transport_t0b_v1.json configs/fq_iclr_transport_t1_aggregation_v1.json configs/fq_iclr_transport_t2_confirmation_execution_v2.json configs/fq_iclr_transport_t2_confirmation_v1.json configs/fq_iclr_transport_t2_input_overlap_v1.json expected/H4_D_SCIENTIFIC_AGGREGATE.json expected/H4_GATE_C_SCIENTIFIC_AGGREGATE.json expected/N1_MECHANISM_AGGREGATE.json expected/T0_COMMON_ESTIMAND_AUDIT.json expected/T1_PRE_READOUT_LOCALIZATION.json expected/T2_CONFIRMATION_INITIAL_128_AGGREGATE.json reproduced/GATE_C_REPRODUCED.json reproduced/H4_D_REPRODUCED.json reproduced/REPRODUCTION_VALIDATION.json reproduced/T1_REPRODUCED.json reproduced/T2_REPRODUCED.json (58 additional files not shown) Current browse context: cs.CV new | recent | 2026-08 Change to browse by: cs cs.LG References & Citations NASA ADS Google Scholar Semantic Scholar Loading... Data provided by: Bibliographic Tools Bibliographic and Citation Tools Bibliographic Explorer Toggle Bibliographic Explorer (What is the Explorer?) Connected Papers Toggle Connected Papers (What is Connected Papers?) Litmaps Toggle Litmaps (What is Litmaps?) scite.ai Toggle scite Smart Citations (What are Smart Citations?) Code, Data, Media Code, Data and Media Associated with this Article alphaXiv Toggle alphaXiv (What is alphaXiv?) Links to Code Toggle CatalyzeX Code Finder for Papers (What is CatalyzeX?) DagsHub Toggle DagsHub (What is DagsHub?) GotitPub Toggle Gotit.pub (What is GotitPub?) Huggingface Toggle Hugging Face (What is Huggingface?) ScienceCast Toggle ScienceCast (What is ScienceCast?) Demos Demos Replicate Toggle Replicate (What is Replicate?) Spaces Toggle Hugging Face Spaces (What is Spaces?) Spaces Toggle TXYZ.AI (What is TXYZ.AI?) Related Papers Recommenders and Search Tools Link to Influence Flower Influence Flower (What are Influence Flowers?) Core recommender toggle CORE Recommender (What is CORE?) Author Venue Institution Topic About arXivLabs arXivLabs: experimental projects with community collaborators arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them. Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs. Which authors of this paper are endorsers? | Disable MathJax (What is MathJax?)