{"id":"W4413726998","doi":"10.1007/978-3-031-99854-6_10","title":"Leveraging Expert Usage to Speed up LLM Inference with Expert Parallelism","year":2025,"lang":"en","type":"book-chapter","venue":"Lecture notes in computer science","topic":"Topic Modeling","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":false,"ca_institutions":"International Laboratory for Brain, Music and Sound Research","funders":"","keywords":"Computer science; Parallelism (grammar); Inference; Speedup; Parallel computing; Data parallelism; Expert system; Artificial intelligence","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":[0.002101532,0.001007298,0.001077586,0.001176439,0.000676445,0.00187734,0.002556289,0.00153781,0.0187921],"category_scores_gemma":[0.01411991,0.0008386057,0.001082303,0.001186282,0.000669022,0.004909842,0.002937667,0.002642249,0.006298378],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0007853871,"about_ca_system_score_gemma":0.001689015,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.005319994,"about_ca_topic_score_gemma":0.01564614,"domain_scores_codex":[0.9980243,0.0006119756,0.0001267864,0.000558696,0.000485842,0.0001923362],"domain_scores_gemma":[0.9931492,0.003898664,0.0001908961,0.001858511,0.0006744923,0.0002283327],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0008463789,0.0004104571,0.004161647,0.0002618783,0.0001591335,0.0002251179,0.0003205737,0.06554599,0.01854452,0.01629511,0.02729484,0.8659344],"study_design_scores_gemma":[0.00007890717,0.00005934805,0.0006519345,0.00002424996,0.00004842987,0.00007272813,0.0000606642,0.9481568,0.00768377,0.0362121,0.006932346,0.00001865226],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.02678698,0.0004209092,0.9483948,0.0005822548,0.0002040236,0.0001232719,0.0004614139,0.01449883,0.008527627],"genre_scores_gemma":[0.3489401,0.0002251078,0.6365342,0.0006468982,0.0002571726,0.0001305789,0.001442402,0.001712202,0.01011134],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.0187921,"threshold_uncertainty_score":0.06286579,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.03508971236646834,"score_gpt":0.2791464190751583,"score_spread":0.24405670670869,"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."}}