{"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":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow","open_science"],"consensus_categories":[],"category_scores_codex":[0.0006811799,0.0008070017,0.0007577745,0.001135684,0.0003468099,0.0009944651,0.005495646,0.0003151804,0.00003292151],"category_scores_gemma":[0.0001404461,0.000692269,0.0001226652,0.0009039768,0.0003561663,0.0007704576,0.002720557,0.0009298095,0.00004371911],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0004558219,"about_ca_system_score_gemma":0.001071997,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001969496,"about_ca_topic_score_gemma":0.00009417399,"domain_scores_codex":[0.994218,0.00004796552,0.0006460938,0.002633617,0.001413068,0.001041257],"domain_scores_gemma":[0.9960411,0.0005436201,0.0002148217,0.0025329,0.0003293687,0.0003381907],"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":[0.00001578783,0.0000334711,0.0000486411,0.00003434376,0.00002128576,0.0001948062,0.008977746,0.09720329,0.0002801177,0.01748762,0.0002060047,0.8754969],"study_design_scores_gemma":[0.0006113302,0.0002822585,0.00006863575,0.0020101,0.000006883347,0.00007664754,0.000002034791,0.9428771,0.002186364,0.03764982,0.01249829,0.001730575],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.0001814331,0.000551599,0.9875866,0.003007052,0.002580858,0.0005835177,0.000001740865,0.0003012044,0.005206044],"genre_scores_gemma":[0.1045874,0.00007820364,0.8781002,0.01351652,0.0006727325,0.0000255982,0.000003252308,0.00004622302,0.002969951],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.8737663,"threshold_uncertainty_score":0.9998851,"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."}}