{"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":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003050351,0.002236868,0.001743183,0.00171143,0.0009939849,0.001781963,0.00403346,0.003499349,0.007737827],"category_scores_gemma":[0.008415372,0.001106127,0.001321699,0.001365301,0.0005772135,0.003167641,0.002160745,0.003218536,0.008755799],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001275459,"about_ca_system_score_gemma":0.001192448,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.01056391,"about_ca_topic_score_gemma":0.01904891,"domain_scores_codex":[0.9989218,0.0002661848,0.00008693443,0.0004025001,0.0001472477,0.0001753607],"domain_scores_gemma":[0.9967793,0.0017445,0.0001081891,0.0004664573,0.000713018,0.0001885956],"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":[0.001306705,0.0009691676,0.003754945,0.0003512977,0.0003950837,0.0002562582,0.0003895653,0.05353952,0.01446641,0.001401111,0.04070165,0.8824684],"study_design_scores_gemma":[0.0001757586,0.000325953,0.001789565,0.00004269086,0.0001242882,0.0001168559,0.0002336088,0.9763858,0.0142426,0.002771051,0.003758401,0.00003346584],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.4181784,0.01384372,0.4692082,0.001838052,0.001571359,0.0008759156,0.004950011,0.07375867,0.01577566],"genre_scores_gemma":[0.688682,0.001429763,0.2657234,0.0007028126,0.0004879503,0.0006353139,0.02101055,0.001518062,0.01981009],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.01056391,"threshold_uncertainty_score":0.02588558,"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."}}