{"id":"W4403940973","doi":"10.1007/978-3-031-73226-3_16","title":"Model Breadcrumbs: Scaling Multi-task Model Merging with Sparse Masks","year":2024,"lang":"en","type":"book-chapter","venue":"Lecture notes in computer science","topic":"Topic Modeling","field":"Computer Science","cited_by":11,"is_retracted":false,"has_abstract":false,"ca_institutions":"Concordia University; Mila - Quebec Artificial Intelligence Institute","funders":"","keywords":"Computer science; Scaling; Task (project management); Algorithm; Artificial intelligence; Theoretical computer science; Mathematics; Geometry","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":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow","scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.0009167664,0.0009199381,0.0007600098,0.00113999,0.0003373171,0.001131474,0.004265662,0.0004319098,0.000004428531],"category_scores_gemma":[0.0000313001,0.0007840681,0.0001922767,0.0006900519,0.0005586111,0.0010001,0.00236892,0.001597758,0.0000607672],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0004805707,"about_ca_system_score_gemma":0.001047895,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00004268379,"about_ca_topic_score_gemma":0.0001337969,"domain_scores_codex":[0.9935241,0.00002012346,0.000747222,0.003026049,0.001528016,0.001154437],"domain_scores_gemma":[0.9966936,0.0001807996,0.0002670035,0.002280084,0.0002649762,0.000313566],"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":[0.000003793866,0.00001772686,0.000006423976,0.00006151917,0.00001744162,0.0001197547,0.001325776,0.8589547,0.0001167003,0.02849679,0.000004181671,0.1108752],"study_design_scores_gemma":[0.0002427619,0.00004221121,0.000001293053,0.0009673624,0.00002466068,0.0001027444,3.156525e-7,0.8605123,0.0002242259,0.1369847,0.00005878427,0.0008386935],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.0001425287,0.000922891,0.99341,0.000782838,0.001019005,0.000460829,0.00001051048,0.0005459465,0.002705486],"genre_scores_gemma":[0.2099049,0.00002980868,0.7871903,0.001033272,0.0002585051,0.00001530418,0.000003277737,0.00008523759,0.001479401],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.2097624,"threshold_uncertainty_score":0.9999055,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.03907107415938375,"score_gpt":0.2588326428358412,"score_spread":0.2197615686764574,"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."}}