{"id":"W6910292159","doi":"10.48448/17rh-v293","title":"Composable Sparse Fine-Tuning for Cross-Lingual Transfer","year":2022,"lang":"en","type":"other","venue":"Underline Science Inc.","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"Mila - Quebec Artificial Intelligence Institute","funders":"","keywords":"Inference; Set (abstract data type); Transfer (computing); Margin (machine learning); Task (project management); Language model; Forgetting; Simple (philosophy)","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":[0.001586481,0.001812347,0.00102728,0.0005497377,0.0005776057,0.001013953,0.002599334,0.001431074,0.005402361],"category_scores_gemma":[0.006040538,0.0006925638,0.0009448513,0.0005951222,0.0009962883,0.002493777,0.002809833,0.002984342,0.002937839],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0009057119,"about_ca_system_score_gemma":0.001250674,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.007557638,"about_ca_topic_score_gemma":0.01296598,"domain_scores_codex":[0.9990701,0.0002524172,0.0000453488,0.0003655915,0.0001559235,0.0001106906],"domain_scores_gemma":[0.9985624,0.0005779662,0.00008829593,0.0005350669,0.0001611511,0.00007512196],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0003764512,0.0003819006,0.002578333,0.0002434308,0.000313547,0.0002506931,0.000287246,0.426704,0.02417626,0.01005559,0.01209249,0.52254],"study_design_scores_gemma":[0.00001860798,0.00005119649,0.0003077234,0.00001179128,0.00002202654,0.0000486832,0.00002841343,0.9830388,0.004571388,0.01019165,0.001694299,0.00001545601],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.05320461,0.001015528,0.9261717,0.0004865677,0.0001676699,0.0001296407,0.0003853379,0.01324832,0.005190651],"genre_scores_gemma":[0.748895,0.0003954779,0.237717,0.0008321443,0.0001207613,0.0003628546,0.001796943,0.001542158,0.008337636],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.007557638,"threshold_uncertainty_score":0.01807266,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.05051522559580456,"score_gpt":0.3485111589177882,"score_spread":0.2979959333219837,"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."}}