{"id":"W6910390661","doi":"10.48448/1gt1-9679","title":"Designing Templates for Eliciting Commonsense Knowledge from Pretrained Sequence-to-Sequence Models","year":2020,"lang":"en","type":"other","venue":"Underline Science Inc.","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Waterloo","funders":"","keywords":"Question answering; Leverage (statistics); Transformer; Template; Natural language; Commonsense reasoning","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","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.001936425,0.001110512,0.001202973,0.001594724,0.000853856,0.0006012875,0.003038932,0.0004985907,0.0003105321],"category_scores_gemma":[0.002108594,0.001110615,0.000207288,0.003399749,0.00206394,0.0006484707,0.0007021492,0.000773709,0.002122273],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.000853495,"about_ca_system_score_gemma":0.002574455,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001902984,"about_ca_topic_score_gemma":0.001673477,"domain_scores_codex":[0.9930072,0.0002253419,0.001009037,0.002624033,0.001329751,0.001804647],"domain_scores_gemma":[0.9950594,0.001212424,0.00083823,0.001300658,0.0005389472,0.001050292],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"bench_or_experimental","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0001655822,0.00022469,0.00008430734,0.0002930285,0.0002042266,0.00009936572,0.005288592,0.008190433,0.7941217,0.007270414,0.1693055,0.01475218],"study_design_scores_gemma":[0.001384312,0.0004068911,0.000005589173,0.002388219,0.0002009304,0.00003416137,0.0010376,0.9248989,0.01535062,0.01794227,0.03395237,0.002398137],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.003516887,0.001777418,0.6326188,0.001692192,0.002068868,0.01032959,0.01641933,0.007252738,0.3243241],"genre_scores_gemma":[0.2470102,0.00002618289,0.7201706,0.001381238,0.001793665,0.0004547866,0.0009659657,0.002940895,0.02525655],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.9167085,"threshold_uncertainty_score":0.9991344,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1792754613970102,"score_gpt":0.3679282695908693,"score_spread":0.1886528081938591,"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."}}