{"id":"W6947969560","doi":"10.48448/s5xp-mn57","title":"Learning to Perturb Word Embeddings for Out-of-distribution QA","year":2021,"lang":"en","type":"other","venue":"Underline Science Inc.","topic":"Memory, History, Trauma, Identity","field":"Arts and Humanities","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"Kootenay Association for Science & Technology","funders":"","keywords":"Overfitting; Word embedding; Word (group theory); Training set; Embedding; Context (archaeology); Benchmark (surveying)","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":[0.002960672,0.001556242,0.0007730813,0.001203154,0.0005014823,0.0009114583,0.001208905,0.001389102,0.003004072],"category_scores_gemma":[0.01049204,0.000388302,0.001299349,0.000780082,0.000998763,0.002322884,0.001741164,0.002414727,0.003589198],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0008215295,"about_ca_system_score_gemma":0.0008562437,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00343293,"about_ca_topic_score_gemma":0.00460385,"domain_scores_codex":[0.9985392,0.0007359638,0.00008484518,0.0004397898,0.0001197942,0.00008047941],"domain_scores_gemma":[0.9965672,0.001664254,0.0002211432,0.0009448918,0.0004823744,0.0001201855],"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":[0.0007960842,0.0006940726,0.01230875,0.0008324829,0.0003418387,0.0003169264,0.0008868224,0.2575766,0.02949833,0.008031884,0.03957868,0.6491375],"study_design_scores_gemma":[0.00006598944,0.0002153884,0.001921766,0.00005574113,0.00004338917,0.0002349185,0.0001890489,0.9629818,0.01445509,0.0123262,0.007471289,0.00003931221],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.1985101,0.003802827,0.7660938,0.001298703,0.0007252015,0.0004353743,0.002679965,0.02144136,0.005012743],"genre_scores_gemma":[0.8143321,0.0006473205,0.1672805,0.0009167011,0.0001566628,0.0003424094,0.008683577,0.0008990726,0.006741669],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.00343293,"threshold_uncertainty_score":0.01565766,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.0500092685920365,"score_gpt":0.2971528522932509,"score_spread":0.2471435837012144,"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."}}