{"id":"W7131095609","doi":"10.1109/iccvw69036.2025.00782","title":"DIVE-Doc: Downscaling Foundational Image Visual Encoder into Hierarchical Architecture for DocVQA","year":2025,"lang":"","type":"article","venue":"","topic":"Multimodal Machine Learning Applications","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"École de Technologie Supérieure","funders":"","keywords":"Encoder; Distillation; Image (mathematics); Code (set theory); Architecture; Forcing (mathematics); Visualization","routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":false},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0005949688,0.0009530808,0.000453651,0.0004582945,0.0004283252,0.001387745,0.002301754,0.0009837558,0.01514024],"category_scores_gemma":[0.002411419,0.0004243724,0.0006869962,0.0003760768,0.0007675425,0.002229133,0.001809499,0.002108213,0.006220732],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001295626,"about_ca_system_score_gemma":0.001305745,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.01416848,"about_ca_topic_score_gemma":0.0311143,"domain_scores_codex":[0.9997268,0.00003576785,0.00001540209,0.00009757293,0.00008195396,0.00004240899],"domain_scores_gemma":[0.9995319,0.0001220646,0.00001626812,0.0001667012,0.0001245299,0.0000385649],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.0007859588,0.0002117689,0.002632239,0.0006725413,0.0001555471,0.0003259316,0.0005872441,0.1365521,0.04571819,0.03898239,0.1338635,0.6395126],"study_design_scores_gemma":[0.0001000625,0.000161117,0.0004011551,0.00006575567,0.00003560702,0.0001284517,0.00009648532,0.8916757,0.02975685,0.01794988,0.05957937,0.00004961105],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.02701006,0.001362557,0.8234573,0.000596178,0.000429911,0.0003195202,0.003089104,0.1282716,0.01546374],"genre_scores_gemma":[0.3841789,0.0006668626,0.5641341,0.00137733,0.0001225035,0.0004827029,0.008817108,0.007123988,0.03309651],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.01514024,"threshold_uncertainty_score":0.05064911,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.008780846628456837,"score_gpt":0.339472735249167,"score_spread":0.3306918886207101,"validation_status":"score_only:v0-immature-baseline","note":"Baseline scores from an immature model (maturity gate not passed). Scores rank; they never assert a category."}}