{"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":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow","scholarly_communication","open_science","research_integrity"],"consensus_categories":[],"category_scores_codex":[0.002793452,0.001150227,0.001672074,0.0008342165,0.0002774275,0.001448615,0.007593039,0.0009976652,0.00003296914],"category_scores_gemma":[0.00664901,0.000843959,0.0005170854,0.003268233,0.0003804595,0.001067605,0.00628146,0.004780704,0.000002847825],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0005872741,"about_ca_system_score_gemma":0.001655061,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00009818989,"about_ca_topic_score_gemma":0.00002488599,"domain_scores_codex":[0.9927507,0.0003607739,0.001532185,0.002555933,0.001614623,0.001185809],"domain_scores_gemma":[0.9947481,0.0004177414,0.001340405,0.001259966,0.001909174,0.0003246525],"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":[0.00008709292,0.0003692879,0.0003753769,0.00228173,0.00007971298,0.00006905394,0.03018728,0.0005030702,0.2994623,0.006796404,0.0001914751,0.6595972],"study_design_scores_gemma":[0.0004420668,0.0001031796,0.0001176364,0.007939691,0.00006280664,0.00005506111,0.0007920922,0.7507824,0.1646333,0.07368595,0.0000121358,0.00137369],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.1406069,0.03566253,0.7881901,0.01985449,0.002325417,0.004801399,0.00009138572,0.001766446,0.006701295],"genre_scores_gemma":[0.4673168,0.00003700615,0.5311901,0.0008659969,0.00008034822,0.0001615142,0.000006192831,0.00004644716,0.0002956625],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.7502793,"threshold_uncertainty_score":0.999588,"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."}}