{"id":"W4409537504","doi":"10.1145/3727200.3727220","title":"Towards Sustainable Large Language Model Serving","year":2024,"lang":"en","type":"article","venue":"ACM SIGEnergy Energy Informatics Review","topic":"Natural Language Processing Techniques","field":"Computer Science","cited_by":13,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Waterloo","funders":"","keywords":"Computer science; Business","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"],"consensus_categories":[],"category_scores_codex":[0.000893793,0.0003324997,0.0004179631,0.0002481327,0.0001721758,0.000558475,0.002960393,0.0001383552,0.00003977094],"category_scores_gemma":[0.0002859809,0.0002624398,0.0001899941,0.001299902,0.0000242596,0.00225973,0.002093879,0.0002614356,0.0000339454],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001657824,"about_ca_system_score_gemma":0.000403554,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001161625,"about_ca_topic_score_gemma":0.00001361397,"domain_scores_codex":[0.9976981,0.00005544009,0.0007441216,0.0002766486,0.0005204057,0.000705338],"domain_scores_gemma":[0.9978476,0.00007461679,0.0001635789,0.001573342,0.0001895938,0.000151333],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[4.460462e-7,0.00001518141,2.323594e-7,0.005591421,0.00003209202,0.00008883828,0.001103539,0.0003181589,0.00006395969,0.7793046,0.01125642,0.2022251],"study_design_scores_gemma":[0.00008191751,0.00003163758,1.289212e-7,0.003768428,0.0000456257,0.00009498718,0.0001198882,0.5409226,0.006823,0.06304587,0.3844827,0.0005832063],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.00001793441,0.3156804,0.6762393,0.0009441305,0.00009510635,0.00008923883,0.000004252316,0.001217873,0.005711752],"genre_scores_gemma":[0.02253015,0.07643445,0.879691,0.01518601,0.0001279717,0.0001715164,0.00008776811,0.00005978632,0.005711375],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.7162588,"threshold_uncertainty_score":0.9999828,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.01200015808108999,"score_gpt":0.290577730194465,"score_spread":0.278577572113375,"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."}}