{"id":"W6966866993","doi":"10.48448/h0yc-am20","title":"Back-Training excels Self-Training at Unsupervised Domain Adaptation of Question Generation and Passage Retrieval","year":2021,"lang":"en","type":"other","venue":"Underline Science Inc.","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"McGill University; Mila - Quebec Artificial Intelligence Institute","funders":"","keywords":"Overfitting; Domain adaptation; Domain (mathematical analysis); Consistency (knowledge bases); Synthetic data; Labeled data; Noisy data; Adaptation (eye)","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":[0.002740387,0.001488668,0.0007004074,0.0008422093,0.0005388601,0.001045556,0.001536281,0.001210687,0.003738413],"category_scores_gemma":[0.009447983,0.0003615918,0.000894475,0.0005911317,0.0006373358,0.001983905,0.001572484,0.00205514,0.003427551],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0006388418,"about_ca_system_score_gemma":0.0007465321,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003024858,"about_ca_topic_score_gemma":0.005026391,"domain_scores_codex":[0.9981495,0.0008022654,0.000116161,0.0006101643,0.0001985055,0.0001234359],"domain_scores_gemma":[0.9957343,0.00221339,0.0001599635,0.001136367,0.0006073351,0.0001486378],"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":[0.0007326749,0.001369064,0.009419776,0.0007777383,0.0002628597,0.0002995495,0.0005410934,0.07946949,0.04909991,0.002020518,0.02775518,0.8282521],"study_design_scores_gemma":[0.0002351832,0.000882968,0.006987763,0.0001070039,0.0001245063,0.0005663298,0.0003468613,0.8815647,0.07434427,0.006421977,0.02831011,0.0001083356],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.2750594,0.003310648,0.6496136,0.0008641842,0.0009239012,0.001123026,0.002806349,0.04787282,0.01842609],"genre_scores_gemma":[0.6000539,0.0004434533,0.3754065,0.001163257,0.0001693837,0.0007987749,0.01121155,0.001564626,0.009188527],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.003738413,"threshold_uncertainty_score":0.01449269,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.07635311581138327,"score_gpt":0.3037698311221069,"score_spread":0.2274167153107237,"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."}}