{"id":"W4416391996","doi":"10.1186/s43067-025-00288-9","title":"Size matters less: how fine-tuned small LLMs excel in BPMN generation","year":2025,"lang":"en","type":"article","venue":"Journal of Electrical Systems and Information Technology","topic":"Business Process Modeling and Analysis","field":"Business, Management and Accounting","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"Artificial Intelligence in Medicine (Canada)","funders":"","keywords":"Business Process Model and Notation; Python (programming language); Pipeline (software); XML; Process (computing); Process modeling; Benchmark (surveying); Business process modeling","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":[],"consensus_categories":[],"category_scores_codex":[0.0003842735,0.0001126342,0.0003158459,0.00157057,0.00008843248,0.0003420573,0.0001626807,0.0001801449,0.000002970942],"category_scores_gemma":[0.0003273463,0.00008926736,0.0000443996,0.001787891,0.0000255708,0.001979267,0.0000408019,0.0002394749,0.000004031071],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00004611413,"about_ca_system_score_gemma":0.00003758325,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001341221,"about_ca_topic_score_gemma":0.00002078522,"domain_scores_codex":[0.9989647,0.000008062952,0.0006383051,0.0000791686,0.0001393023,0.000170478],"domain_scores_gemma":[0.9988304,0.00002822787,0.0005990455,0.00008480222,0.0004491648,0.000008374043],"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":[0.0002752524,0.0003513403,0.05295432,0.002491208,0.0004073154,0.00002988356,0.0001450346,0.01245152,0.01264768,0.2731743,0.04091085,0.6041613],"study_design_scores_gemma":[0.004585079,0.0001258452,0.006492572,0.000780456,0.0002847171,0.0000929563,0.001414488,0.7947964,0.0008285643,0.00820078,0.1816905,0.0007076077],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.8469979,0.001662134,0.1073989,0.04204888,0.0005820432,0.0002460977,0.000001194261,0.0001018422,0.0009610505],"genre_scores_gemma":[0.9982117,0.0001240328,0.0001740219,0.001200344,0.0001867021,0.000009055751,0.000004972359,0.000004104153,0.00008501148],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.7823449,"threshold_uncertainty_score":0.3640217,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.01111295222210184,"score_gpt":0.1993258710873685,"score_spread":0.1882129188652666,"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."}}