{"id":"W4389518424","doi":"10.18653/v1/2023.arabicnlp-1.9","title":"Beyond English: Evaluating LLMs for Arabic Grammatical Error Correction","year":2023,"lang":"en","type":"article","venue":"","topic":"Natural Language Processing Techniques","field":"Computer Science","cited_by":14,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of British Columbia","funders":"Alliance de recherche numérique du Canada; Social Sciences and Humanities Research Council of Canada; Natural Sciences and Engineering Research Council of Canada; Canada Research Chairs","keywords":"Computer science; Arabic; Context (archaeology); Natural language processing; Artificial intelligence; Task (project management); Linguistics","routes":{"ca_aff":true,"ca_fund":true,"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":[0.003137188,0.002438796,0.00111361,0.00117969,0.0007242158,0.001282174,0.002300094,0.002143055,0.003123081],"category_scores_gemma":[0.0122972,0.0004056492,0.0009680059,0.0008593578,0.0007878934,0.002673866,0.001477681,0.002926388,0.002724021],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001452276,"about_ca_system_score_gemma":0.002311357,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0181478,"about_ca_topic_score_gemma":0.02359675,"domain_scores_codex":[0.9985666,0.0005364747,0.0001138827,0.0005058954,0.0001724832,0.0001047114],"domain_scores_gemma":[0.9956784,0.002444731,0.0001488127,0.0006777646,0.0008027604,0.0002475346],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.001801723,0.001116222,0.009265843,0.0009901467,0.0005346846,0.0004791273,0.0004982915,0.3359324,0.01460627,0.001356034,0.0321388,0.6012805],"study_design_scores_gemma":[0.0001616595,0.0005806428,0.00192488,0.00007711714,0.000148314,0.0001348897,0.0002524896,0.9718021,0.01640873,0.002070471,0.006385083,0.00005361883],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.7642184,0.01266424,0.1228621,0.002881978,0.001911942,0.0007628574,0.006061092,0.07512667,0.01351066],"genre_scores_gemma":[0.8715969,0.001208706,0.09996064,0.001116924,0.0002527847,0.0003356871,0.017322,0.001148644,0.007057726],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.0181478,"threshold_uncertainty_score":0.03608435,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.04045139428244435,"score_gpt":0.3526177338742187,"score_spread":0.3121663395917744,"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."}}