{"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":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.003284791,0.0005042569,0.0006457563,0.00114979,0.0003929707,0.0002062475,0.0004242232,0.0004226738,0.001249318],"category_scores_gemma":[0.0004236658,0.0005282923,0.00007817573,0.001941944,0.00101772,0.0004275527,0.0002511612,0.0003272992,0.0001240882],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0006405729,"about_ca_system_score_gemma":0.001377809,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0002274037,"about_ca_topic_score_gemma":0.001908216,"domain_scores_codex":[0.9953917,0.0003892446,0.0007349769,0.001258798,0.00155146,0.0006738378],"domain_scores_gemma":[0.9977353,0.0001283919,0.0008522387,0.0006091226,0.0004067582,0.0002682282],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"bench_or_experimental","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.000074141,0.0002072983,0.0007240379,0.0004093771,0.0001543705,0.00003980279,0.02000996,0.001520819,0.9510289,0.007751748,0.005228123,0.01285147],"study_design_scores_gemma":[0.007975868,0.00113321,0.002362617,0.004618853,0.0005829645,0.0003829685,0.02206832,0.8411341,0.02062196,0.002075609,0.09260131,0.004442167],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.4612516,0.008544929,0.09456106,0.0008810705,0.004889432,0.005702024,0.0012236,0.002215187,0.4207311],"genre_scores_gemma":[0.503452,0.0004659328,0.4263651,0.0001971175,0.00202682,0.00003676265,0.002712951,0.001450571,0.0632927],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.9304069,"threshold_uncertainty_score":0.9997169,"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."}}