{"id":"W7066217127","doi":"","title":"Improving Transformer Performance for French Clinical Notes Classification Using Mixture of Experts on a Limited Dataset","year":2023,"lang":"en","type":"report","venue":"Open Repository and Bibliography (University of Luxembourg)","topic":"Laser-Plasma Interactions and Diagnostics","field":"Physics and Astronomy","cited_by":0,"is_retracted":false,"has_abstract":false,"ca_institutions":"","funders":"Fonds de recherche du Québec – Nature et technologies; Natural Sciences and Engineering Research Council of Canada; Fonds de Recherche du Québec - Santé; Institut de Valorisation des Données","keywords":"Transformer; Pattern recognition (psychology); Training set; Feature selection","routes":{"ca_aff":false,"ca_fund":true,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":true},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0004346616,0.0002293385,0.0005562219,0.002829166,0.0004200155,0.0001056226,0.0003887973,0.0002405301,0.00006851627],"category_scores_gemma":[0.000005754939,0.000234658,0.0003837728,0.002212222,0.0001816296,0.0004564306,0.00009858595,0.0002506936,0.000001488131],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00001344036,"about_ca_system_score_gemma":0.0002557105,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0067932,"about_ca_topic_score_gemma":0.00003698926,"domain_scores_codex":[0.9985633,0.00006286336,0.0004403509,0.0004616079,0.0002820287,0.0001897831],"domain_scores_gemma":[0.9976669,0.0005722987,0.0008254462,0.0003766421,0.0004519927,0.0001067274],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0009165234,0.002026415,0.6881617,0.002309802,0.002288561,0.00002742163,0.000320295,0.0001327256,0.004077991,0.0001729174,0.2699086,0.02965695],"study_design_scores_gemma":[0.01027751,0.005364982,0.4519035,0.008663992,0.007376032,0.00006808793,0.01089796,0.01310227,0.01433001,0.0003272956,0.4738111,0.003877282],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.947376,0.0001430913,0.01054468,0.00009381102,0.002304926,0.002796761,0.01524409,0.00004978942,0.02144683],"genre_scores_gemma":[0.9915702,0.00155928,0.002453616,0.00000971291,0.0003375536,0.0000102188,0.003258134,0.00003550762,0.0007657427],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.2362583,"threshold_uncertainty_score":0.9998206,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1268968461969892,"score_gpt":0.3556919608260341,"score_spread":0.2287951146290449,"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."}}