{"id":"W3119437245","doi":"10.18653/v1/2021.acl-long.519","title":"End-to-End Training of Neural Retrievers for Open-Domain Question Answering","year":2021,"lang":"en","type":"preprint","venue":"","topic":"Topic Modeling","field":"Computer Science","cited_by":9,"is_retracted":false,"has_abstract":true,"ca_institutions":"McGill University; Mila - Quebec Artificial Intelligence Institute","funders":"Canadian Institute for Advanced Research; Nvidia","keywords":"Computer science; Artificial intelligence; Question answering; Context (archaeology); Machine learning; Domain (mathematical analysis); Task (project management); Artificial neural network; Biomedical text mining; Natural language processing; Text mining; Mathematics","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":[0.001002966,0.0002083102,0.0004226159,0.0001378294,0.00006746725,0.0004384559,0.002015036,0.0001656578,0.00003196015],"category_scores_gemma":[0.0001699633,0.0002229028,0.0001335793,0.0002088053,0.00001890267,0.0003797941,0.003199452,0.0002977227,8.515757e-7],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00009502526,"about_ca_system_score_gemma":0.0003302264,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0004748127,"about_ca_topic_score_gemma":0.0001155661,"domain_scores_codex":[0.9979845,0.00008517975,0.00047482,0.0008546376,0.0002951143,0.0003057444],"domain_scores_gemma":[0.998472,0.0001330419,0.0001828145,0.0009387844,0.0001575047,0.0001158528],"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":[0.00006805696,0.0001059795,0.000423352,0.0007207635,0.0001750968,0.00005962676,0.03083319,0.1718045,0.01248686,0.2322845,0.00037227,0.5506658],"study_design_scores_gemma":[0.000638907,0.0001182709,0.0005292002,0.0005323869,0.00001906088,0.00001794006,0.001114782,0.9759369,0.007880516,0.01157936,0.001006343,0.0006262794],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.07088424,0.00005379013,0.924167,0.001219165,0.001313463,0.0006267963,0.00000709052,0.00009561388,0.00163284],"genre_scores_gemma":[0.505904,0.000002631675,0.4935935,0.0001887413,0.00006774155,0.00003744817,0.00001006721,0.00001046683,0.0001854873],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.8041325,"threshold_uncertainty_score":0.9089714,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.07985047469528585,"score_gpt":0.3197809994370351,"score_spread":0.2399305247417493,"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."}}