{"id":"W3093808828","doi":"10.18653/v1/2021.emnlp-main.740","title":"BARThez: a Skilled Pretrained French Sequence-to-Sequence Model","year":2021,"lang":"en","type":"preprint","venue":"Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing","topic":"Natural Language Processing Techniques","field":"Computer Science","cited_by":6,"is_retracted":false,"has_abstract":true,"ca_institutions":"Polytechnique Montréal","funders":"Centre National de la Recherche Scientifique","keywords":"Automatic summarization; Discriminative model; Computer science; Generative grammar; Benchmark (surveying); Transfer of learning; Artificial intelligence; Natural language processing; Sequence (biology); Code (set theory); Generative model; Field (mathematics); Language model; Machine learning; Programming language; Cartography","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":[0.0006625992,0.001713466,0.0006130204,0.0008025641,0.0006907166,0.001179443,0.002435459,0.001802896,0.0134195],"category_scores_gemma":[0.002268163,0.0006697215,0.001342518,0.0006744096,0.0005022141,0.001528335,0.0009989153,0.002735292,0.006609989],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001344532,"about_ca_system_score_gemma":0.00187336,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.04578343,"about_ca_topic_score_gemma":0.08969801,"domain_scores_codex":[0.9996749,0.00007127652,0.00001215134,0.0001398103,0.00004663186,0.0000552225],"domain_scores_gemma":[0.9994627,0.0002758549,0.00002096438,0.00008421552,0.0001256527,0.00003058707],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.0006318222,0.0002601196,0.003288117,0.0003355122,0.0002898411,0.0005313192,0.0002664932,0.5248369,0.01381398,0.01605581,0.139776,0.2999141],"study_design_scores_gemma":[0.00004173888,0.00007220711,0.0004632197,0.00002193506,0.00002582498,0.00009028849,0.00004054637,0.9770685,0.002833025,0.004732363,0.01458263,0.00002770729],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.09347856,0.00272059,0.7354111,0.002608416,0.001249976,0.0005385333,0.02318166,0.1077211,0.03309003],"genre_scores_gemma":[0.5475124,0.001001092,0.3122304,0.002443979,0.0004040922,0.001121124,0.0688137,0.005630249,0.06084294],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.04578343,"threshold_uncertainty_score":0.09103388,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.08687262609204593,"score_gpt":0.4277476868274459,"score_spread":0.3408750607354,"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."}}