{"id":"W4381686872","doi":"10.1162/tacl_a_00564","title":"Questions Are All You Need to Train a Dense Passage Retriever","year":2023,"lang":"en","type":"article","venue":"Transactions of the Association for Computational Linguistics","topic":"Topic Modeling","field":"Computer Science","cited_by":28,"is_retracted":false,"has_abstract":true,"ca_institutions":"McGill University; Mila - Quebec Artificial Intelligence Institute","funders":"DeepMind","keywords":"Computer science; Initialization; Code (set theory); Task (project management); Set (abstract data type); Encoder; Artificial intelligence; Information retrieval; Scheme (mathematics); Training set; Domain (mathematical analysis); Language model; Natural language processing; Machine learning","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":[0.002263985,0.001100049,0.001148879,0.0007977537,0.0005332754,0.001715174,0.002097774,0.001820661,0.01485285],"category_scores_gemma":[0.009221409,0.0006025795,0.001326764,0.0005689844,0.0009781343,0.004097463,0.002457244,0.003081314,0.01077997],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0007822199,"about_ca_system_score_gemma":0.001050461,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.004415294,"about_ca_topic_score_gemma":0.004821027,"domain_scores_codex":[0.9987608,0.0003698583,0.00007610425,0.0004994872,0.0001957342,0.00009803634],"domain_scores_gemma":[0.9974956,0.001244298,0.00007598365,0.0007193187,0.0003563901,0.000108358],"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":[0.0008348469,0.0003371884,0.003057024,0.0008420313,0.0002482712,0.0005666047,0.000955815,0.07300904,0.05269521,0.01695954,0.06127658,0.7892178],"study_design_scores_gemma":[0.0001200869,0.0002791912,0.001112724,0.00009616559,0.000127176,0.0005483529,0.0002673999,0.8834192,0.04992708,0.02814982,0.03587086,0.0000818731],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.03092643,0.0009200069,0.915548,0.001038949,0.0003632802,0.0003319547,0.002370875,0.04344815,0.005052343],"genre_scores_gemma":[0.3647412,0.0004812223,0.604005,0.001632871,0.0003314927,0.0005021151,0.01096152,0.003015699,0.01432883],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.01485285,"threshold_uncertainty_score":0.04968774,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.03032712766975569,"score_gpt":0.2833717000899555,"score_spread":0.2530445724201998,"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."}}