{"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":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002033326,0.00119065,0.0006814594,0.001047877,0.0003276521,0.00149784,0.00207701,0.001331447,0.008245334],"category_scores_gemma":[0.0154786,0.0005833821,0.001211504,0.0009399703,0.0008777801,0.005692222,0.002259276,0.002363683,0.004195473],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0009169382,"about_ca_system_score_gemma":0.001605948,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001726663,"about_ca_topic_score_gemma":0.005321137,"domain_scores_codex":[0.9986701,0.0005226686,0.00008417186,0.0004736248,0.0001732003,0.00007620744],"domain_scores_gemma":[0.9946827,0.003068685,0.0002840296,0.001472297,0.0003243232,0.0001679633],"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":[0.0005726417,0.0005717053,0.008471749,0.001036298,0.000286574,0.0006231767,0.0009894636,0.1877894,0.0296962,0.07977849,0.02106871,0.6691156],"study_design_scores_gemma":[0.00003612285,0.0001097968,0.0007156309,0.00006907102,0.0000448291,0.0001763639,0.0001638267,0.866368,0.01366177,0.1114271,0.007194844,0.00003280265],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.03696653,0.0003970362,0.9487522,0.0005736363,0.00007528831,0.000284764,0.00210385,0.006436531,0.00441015],"genre_scores_gemma":[0.6250732,0.0003929771,0.3569162,0.0005380941,0.00007222134,0.0006780029,0.01076243,0.0006041082,0.004962893],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.008245334,"threshold_uncertainty_score":0.02758336,"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."}}