{"id":"W4413411773","doi":"10.1145/3721145.3730418","title":"Cephalo: Harnessing Heterogeneous GPU Clusters for Training Transformer Models","year":2025,"lang":"en","type":"article","venue":"","topic":"Topic Modeling","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":false,"ca_institutions":"University of Waterloo","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Computer science; Transformer; Training (meteorology); Supercomputer; Parallel computing; Electrical engineering; Engineering; Physics; Voltage","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":[0.0006353005,0.001495543,0.0007468805,0.0009841912,0.0005691566,0.001463343,0.002039854,0.0009307057,0.01038934],"category_scores_gemma":[0.002933624,0.0007626608,0.001028118,0.00143244,0.0002883677,0.001652031,0.001343152,0.001884995,0.00522099],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0007118854,"about_ca_system_score_gemma":0.001439486,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.01479584,"about_ca_topic_score_gemma":0.03790859,"domain_scores_codex":[0.9996141,0.00008836616,0.00002206003,0.0001384632,0.00008118599,0.00005598157],"domain_scores_gemma":[0.999421,0.0002296256,0.00002230766,0.0001636386,0.000108256,0.00005523644],"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":[0.001069945,0.0003132103,0.004494383,0.000379429,0.0004247043,0.0003000491,0.0002924315,0.2083401,0.01629644,0.006398792,0.115963,0.6457276],"study_design_scores_gemma":[0.00008335399,0.00005254679,0.0003362997,0.00001184249,0.00003910872,0.00004339477,0.00005979757,0.9802741,0.005495856,0.004448761,0.009142227,0.00001264493],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.06511579,0.001665595,0.8185853,0.0006741225,0.0007379858,0.0003038007,0.004923047,0.09845363,0.009540683],"genre_scores_gemma":[0.4298857,0.0007982728,0.5310238,0.0005886036,0.0001780422,0.0004313069,0.01774311,0.006741452,0.01260962],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.01479584,"threshold_uncertainty_score":0.03475577,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.05453842143186296,"score_gpt":0.2824619046238849,"score_spread":0.2279234831920219,"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."}}