{"id":"W3087148478","doi":"","title":"Conditionally Adaptive Multi-Task Learning: Improving Transfer Learning in NLP Using Fewer Parameters & Less Data","year":2020,"lang":"en","type":"preprint","venue":"PolyPublie (École Polytechnique de Montréal)","topic":"Domain Adaptation and Few-Shot Learning","field":"Computer Science","cited_by":38,"is_retracted":false,"has_abstract":true,"ca_institutions":"Polytechnique Montréal","funders":"","keywords":"Multi-task learning; Computer science; Overfitting; Forgetting; Artificial intelligence; Transfer of learning; Task (project management); Machine learning; Benchmark (surveying); Transformer; Artificial neural network","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","scholarly_communication","research_integrity"],"consensus_categories":[],"category_scores_codex":[0.002144697,0.000962224,0.001078611,0.001038238,0.0005530411,0.001250289,0.004040228,0.000868123,0.00001914149],"category_scores_gemma":[0.001052347,0.001134646,0.000339219,0.001036947,0.000197886,0.001388724,0.003912334,0.005397486,0.00002591078],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00095606,"about_ca_system_score_gemma":0.001538553,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.01126172,"about_ca_topic_score_gemma":0.002053483,"domain_scores_codex":[0.9926466,0.001197336,0.001372322,0.002474192,0.001014895,0.001294708],"domain_scores_gemma":[0.9960359,0.0004929091,0.0006960526,0.001940778,0.0002559083,0.000578406],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0001225806,0.0002634898,0.007205087,0.0001592769,0.000171164,0.0004177473,0.003167958,0.9366977,0.007788996,0.008685165,0.00009726523,0.03522361],"study_design_scores_gemma":[0.001061472,0.0001278968,0.008300893,0.0003132199,0.00006161962,0.00006290214,0.0007998619,0.9858115,0.0005242154,0.0008270885,0.0009504043,0.001158949],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.04430015,0.0007703344,0.9492988,0.002161011,0.0003680389,0.001292818,0.0000917335,0.001549117,0.0001680042],"genre_scores_gemma":[0.6105984,0.00008995408,0.3872167,0.00111749,0.00009474772,0.0002046949,0.0003632549,0.0001177079,0.0001969963],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.5662982,"threshold_uncertainty_score":0.9997865,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.07018382920870495,"score_gpt":0.2793119769554278,"score_spread":0.2091281477467229,"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."}}