{"id":"W6966678284","doi":"10.48448/vttz-0r79","title":"Collaborative Performance Prediction for Large Language Models","year":2024,"lang":"en","type":"other","venue":"Underline Science Inc.","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"McGill University","funders":"","keywords":"Performance prediction; Downstream (manufacturing); Language model; Scaling; Predictive modelling; Scaling law","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":[0.006788791,0.001800244,0.001301396,0.002027899,0.000814066,0.00219688,0.002076508,0.001475375,0.002256333],"category_scores_gemma":[0.03844345,0.0005734944,0.001055042,0.001828944,0.001086124,0.003522529,0.002530436,0.00262618,0.001663259],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001839349,"about_ca_system_score_gemma":0.001649468,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.01324944,"about_ca_topic_score_gemma":0.01220322,"domain_scores_codex":[0.9954332,0.001470169,0.0002135785,0.001519705,0.001014104,0.000349185],"domain_scores_gemma":[0.9775487,0.0144382,0.002006866,0.003565598,0.001736761,0.0007039711],"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":[0.0002452561,0.0002165035,0.02703905,0.0002070225,0.0002169474,0.0002024736,0.0004167472,0.7693021,0.002810132,0.01009259,0.008964747,0.1802865],"study_design_scores_gemma":[0.00000545334,0.00002836654,0.001444549,0.000008686567,0.000009648313,0.00001823752,0.00002834202,0.9888904,0.0009141017,0.008102001,0.0005369606,0.00001320978],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.1980271,0.001686134,0.7821922,0.00187731,0.0001794101,0.0002000381,0.001945414,0.005697052,0.008195295],"genre_scores_gemma":[0.9072625,0.0003861401,0.08661684,0.0001947068,0.000181932,0.0002097066,0.00231371,0.0004594225,0.002375094],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.01324944,"threshold_uncertainty_score":0.03590298,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.01690932362608183,"score_gpt":0.3071650045323348,"score_spread":0.290255680906253,"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."}}