{"id":"W4416342822","doi":"10.1109/socc66126.2025.11235365","title":"Sparse Attention: A Co-Design Approach for Efficient Transformer Execution on Tensor Cores","year":2025,"lang":"","type":"article","venue":"","topic":"Advanced Neural Network Applications","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"Saint John Regional Hospital; Lakehead University","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Exploit; Software; Inference; Pruning; Set (abstract data type); Component (thermodynamics); Minification; Tensor (intrinsic definition)","routes":{"ca_aff":true,"ca_fund":true,"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":[0.000678687,0.0009813658,0.0003553671,0.0006099927,0.0003683085,0.0008094164,0.002057369,0.0003637536,0.004688249],"category_scores_gemma":[0.002206436,0.0004289099,0.0005079444,0.0006445454,0.000498377,0.001448891,0.0009804155,0.001031071,0.001222944],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0008731996,"about_ca_system_score_gemma":0.001683965,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003719276,"about_ca_topic_score_gemma":0.00903711,"domain_scores_codex":[0.9994857,0.00009817402,0.00004579733,0.0001006238,0.0001639518,0.0001057401],"domain_scores_gemma":[0.9990134,0.0002721355,0.0001051323,0.0002056615,0.0003272906,0.00007632429],"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":[0.001150469,0.0003117867,0.006056505,0.0007767497,0.0001840196,0.0005443026,0.0005291591,0.1628276,0.1543297,0.03174829,0.02391956,0.617622],"study_design_scores_gemma":[0.0000808704,0.0003867774,0.0006287826,0.00003339205,0.00006247875,0.0001491971,0.00007489918,0.9232442,0.05326336,0.009250778,0.01279571,0.00002943518],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.05731537,0.0006291987,0.926275,0.0002912529,0.0001622791,0.0001993588,0.0001439454,0.00855859,0.006424921],"genre_scores_gemma":[0.4959754,0.0002688328,0.497885,0.0003246609,0.0000543909,0.0003271684,0.0004265856,0.0008102232,0.003927862],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.004688249,"threshold_uncertainty_score":0.01568383,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.05240302543642369,"score_gpt":0.3107068447083154,"score_spread":0.2583038192718918,"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."}}