{"id":"W2970600560","doi":"10.18653/v1/d19-1620","title":"Countering the Effects of Lead Bias in News Summarization via Multi-Stage Training and Auxiliary Losses","year":2019,"lang":"en","type":"article","venue":"","topic":"Topic Modeling","field":"Computer Science","cited_by":34,"is_retracted":false,"has_abstract":true,"ca_institutions":"McGill University","funders":"Institut de Valorisation des Données; Compute Canada; Natural Sciences and Engineering Research Council of Canada; Canadian Institute for Advanced Research","keywords":"Automatic summarization; Computer science; Natural language processing; Artificial intelligence; Natural (archaeology); Joint (building); Natural language; Speech recognition; Engineering; History","routes":{"ca_aff":true,"ca_fund":true,"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":[],"consensus_categories":[],"category_scores_codex":[0.0002071777,0.00006517048,0.0001070056,0.00005336321,0.00002121891,0.00004561919,0.0002345132,0.0000288587,0.00000331524],"category_scores_gemma":[0.00005295636,0.00004629196,0.00001474164,0.0001201678,0.00001500889,0.0002847519,0.0001323189,0.00006259391,0.000003668807],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00001187569,"about_ca_system_score_gemma":0.00002014261,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0004168853,"about_ca_topic_score_gemma":0.0002888923,"domain_scores_codex":[0.9994066,0.0000449418,0.0001515602,0.0001822436,0.00009826676,0.0001164298],"domain_scores_gemma":[0.9994714,0.0002020671,0.00004860151,0.0002480002,0.00001530844,0.00001466372],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0000138458,0.0001330032,0.4200903,0.00102836,0.00003536932,0.0000159555,0.03753005,0.01361731,0.1504644,0.02261919,0.00003738065,0.3544148],"study_design_scores_gemma":[0.0003541355,0.00003023734,0.008661738,0.00007354427,0.000001214554,0.000001868918,0.0002331099,0.9871842,0.003228378,0.00008035197,0.00008220128,0.00006900835],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.4853435,0.00005233276,0.5139304,0.00007370635,0.0001532801,0.0001222618,1.017404e-7,0.00001733422,0.0003071219],"genre_scores_gemma":[0.9818449,0.00001480187,0.01742484,0.0001408806,0.00001241158,0.000003173857,3.28845e-7,0.000003954586,0.0005546801],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9735669,"threshold_uncertainty_score":0.1887731,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.06762744025516274,"score_gpt":0.2830354277208046,"score_spread":0.2154079874656419,"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."}}