{"id":"W3214038065","doi":"10.18653/v1/2021.findings-emnlp.26","title":"Multi-Task Dense Retrieval via Model Uncertainty Fusion for Open-Domain Question Answering","year":2021,"lang":"en","type":"article","venue":"","topic":"Topic Modeling","field":"Computer Science","cited_by":9,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Waterloo","funders":"Natural Sciences and Engineering Research Council of Canada; Canada First Research Excellence Fund; Compute Canada","keywords":"Computer science; Benchmark (surveying); Task (project management); Question answering; Information retrieval; Joint (building); Set (abstract data type); Domain (mathematical analysis); Aggregate (composite); Artificial intelligence; Training set; Code (set theory); Language model; Natural language processing","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":[0.00431908,0.002381426,0.002648175,0.002773029,0.0007608559,0.001631064,0.002767556,0.002413628,0.002331628],"category_scores_gemma":[0.01102255,0.0009398226,0.002314995,0.002459224,0.00103443,0.004372943,0.00338839,0.002879024,0.001538487],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001261967,"about_ca_system_score_gemma":0.001383628,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.009815631,"about_ca_topic_score_gemma":0.009218999,"domain_scores_codex":[0.997095,0.0009724667,0.0001856405,0.0009483731,0.0005521159,0.0002463871],"domain_scores_gemma":[0.9945593,0.003312263,0.0002998886,0.001005222,0.0006600858,0.0001632329],"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":[0.0006584793,0.0005851006,0.003300012,0.0006629489,0.0006649936,0.0004154936,0.0007615098,0.3555054,0.0153919,0.00728374,0.0172821,0.5974883],"study_design_scores_gemma":[0.00003557419,0.0001127835,0.0005053284,0.00001862146,0.00007255757,0.0001008688,0.00006836301,0.9832697,0.002922987,0.01108791,0.001771797,0.0000335872],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.02643147,0.002264614,0.9629656,0.000487031,0.00007973677,0.00015168,0.0006354481,0.005820973,0.001163434],"genre_scores_gemma":[0.6399652,0.0009571402,0.3462633,0.0009812133,0.0004641749,0.0003837766,0.006534074,0.0005293789,0.003921809],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.009815631,"threshold_uncertainty_score":0.02284181,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.04196707278900472,"score_gpt":0.3043780166640182,"score_spread":0.2624109438750135,"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."}}