{"id":"W6910373184","doi":"10.48448/k4b7-c679","title":"Metric-guided Distillation: Distilling Knowledge from the Metric to Ranker and Retriever for Generative Commonsense Reasoning","year":2022,"lang":"en","type":"other","venue":"Underline Science Inc.","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"","keywords":"Metric (unit); Generalization; Relevance (law); Sentence; Benchmark (surveying); Labrador Retriever; Commonsense reasoning","routes":{"ca_aff":false,"ca_fund":false,"ca_venue":false,"about_ca":true,"invisible_to_affiliation_only":true},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","metaepi_narrow","sts","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.00354783,0.0006819335,0.0008008541,0.002253509,0.001828226,0.0005778727,0.001384675,0.0001717847,0.001480418],"category_scores_gemma":[0.008806376,0.0005117432,0.0001421162,0.01212777,0.001184509,0.0002189683,0.001049256,0.0005672624,0.0002553069],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001060502,"about_ca_system_score_gemma":0.0007726208,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0015743,"about_ca_topic_score_gemma":0.002512373,"domain_scores_codex":[0.9949042,0.0004005646,0.0006813105,0.001711074,0.00143642,0.0008664711],"domain_scores_gemma":[0.9940491,0.003060256,0.0006743958,0.001276588,0.0005097231,0.0004299458],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0001374781,0.0001382088,0.003175779,0.00003232558,0.000239233,0.00001291259,0.002281345,0.001047614,0.0006203537,0.005624925,0.9775122,0.009177621],"study_design_scores_gemma":[0.001396117,0.0001464695,0.002609374,0.0001528795,0.0003563584,0.0000188325,0.0009203946,0.06618625,0.0002076004,0.0005438116,0.9263558,0.00110607],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"other","genre_gemma":"empirical","genre_scores_codex":[0.01894696,0.06416722,0.3904571,0.006674947,0.01072897,0.02401767,0.03503552,0.003673037,0.4462985],"genre_scores_gemma":[0.439117,0.0003277168,0.276987,0.002754992,0.005614147,0.0009419613,0.003254824,0.003875928,0.2671264],"genre_candidate":"other","genre_consensus":null,"teacher_disagreement_score":0.4201701,"threshold_uncertainty_score":0.9997334,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.05362116947968987,"score_gpt":0.3383028259425722,"score_spread":0.2846816564628823,"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."}}