{"id":"W4411120155","doi":"10.18653/v1/2025.findings-naacl.410","title":"GeoCoder: Solving Geometry Problems by Generating Modular Code through Vision-Language Models","year":2025,"lang":"en","type":"article","venue":"","topic":"Model-Driven Software Engineering Techniques","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Samsung; Natural Sciences and Engineering Research Council of Canada; Canadian Institute for Advanced Research","keywords":"Modular design; Computer science; Programming language; Code (set theory); Code generation; Artificial intelligence","routes":{"ca_aff":false,"ca_fund":true,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":true},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow"],"consensus_categories":[],"category_scores_codex":[0.0003845862,0.0002882734,0.0002909642,0.0001872558,0.0002013409,0.0003299622,0.001282959,0.0001550645,0.00002030671],"category_scores_gemma":[0.00001444534,0.0002753481,0.00009362035,0.0007005191,0.00003251948,0.001312501,0.0008092243,0.0002753804,0.00000839593],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001062695,"about_ca_system_score_gemma":0.000059968,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0002212825,"about_ca_topic_score_gemma":0.000009961153,"domain_scores_codex":[0.9979262,0.00004877152,0.0004137034,0.0007409641,0.0003706983,0.0004996208],"domain_scores_gemma":[0.9985573,0.00007339485,0.00007730415,0.001116357,0.0001027444,0.0000728438],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.000001067483,0.00009597105,0.00003943073,0.0001007202,0.00004829888,0.00001033785,0.001250869,0.4454795,0.0424467,0.4291078,0.0166504,0.06476888],"study_design_scores_gemma":[0.0001502162,0.00002869953,0.000004559168,0.000110341,0.000004897546,0.000004617772,0.000006451581,0.9641847,0.01630624,0.00841979,0.01048563,0.0002938698],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.002913545,0.001346435,0.9897206,0.0003754885,0.0001951648,0.000365278,0.000008441015,0.002725993,0.002349075],"genre_scores_gemma":[0.1335169,0.00005188507,0.8633505,0.0006483048,0.00002889355,0.00008748587,0.00001076117,0.0000256492,0.002279633],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.5187051,"threshold_uncertainty_score":0.9999699,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.01027966017072345,"score_gpt":0.2580396419478833,"score_spread":0.2477599817771598,"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."}}