{"id":"W4415332454","doi":"10.48550/arxiv.2509.23410","title":"PATCH: Learnable Tile-level Hybrid Sparsity for LLMs","year":2025,"lang":"en","type":"preprint","venue":"arXiv (Cornell University)","topic":"Advancements in Photolithography Techniques","field":"Engineering","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Alliance de recherche numérique du Canada; Natural Sciences and Engineering Research Council of Canada; Canada Research Chairs","keywords":"Pruning; Selection (genetic algorithm); Speedup; Bridge (graph theory); Face (sociological concept); Class (philosophy); Heuristics; Process (computing); Range (aeronautics)","routes":{"ca_aff":false,"ca_fund":true,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":true},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0004731901,0.001002468,0.0007066083,0.0003883707,0.000320524,0.0008864327,0.001631468,0.0008922856,0.004905823],"category_scores_gemma":[0.003238114,0.0004871794,0.0008442433,0.0003956076,0.0005211455,0.001662498,0.001614949,0.001657474,0.00282139],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0005546961,"about_ca_system_score_gemma":0.001042464,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.005801572,"about_ca_topic_score_gemma":0.01517127,"domain_scores_codex":[0.9996495,0.00007194198,0.0000165554,0.00009313972,0.0001240087,0.00004480808],"domain_scores_gemma":[0.999364,0.0002764007,0.00003982993,0.0001822757,0.00009277822,0.00004483399],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0005587895,0.0002355368,0.003956149,0.0003600052,0.0002059459,0.0003250141,0.0002295935,0.4777861,0.02919834,0.02184728,0.05627462,0.4090226],"study_design_scores_gemma":[0.00002834334,0.0000364673,0.000116043,0.000008599403,0.000008608733,0.00004014331,0.00002002339,0.9839349,0.004653949,0.008134135,0.003010842,0.000007997429],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.03937399,0.0005890055,0.92393,0.0005617083,0.0001562842,0.0000896034,0.00126491,0.02882329,0.005211228],"genre_scores_gemma":[0.4431072,0.0004215876,0.535555,0.0009114384,0.0001423239,0.0003399088,0.005898933,0.003730292,0.009893297],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.005801572,"threshold_uncertainty_score":0.0164116,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1019054716866376,"score_gpt":0.2026944299194573,"score_spread":0.1007889582328197,"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."}}