{"id":"W4393703399","doi":"10.5281/zenodo.8008748","title":"GEO-Bench: m-pv4ger","year":2023,"lang":"en","type":"dataset","venue":"Zenodo (CERN European Organization for Nuclear Research)","topic":"Geological Modeling and Analysis","field":"Earth and Planetary Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"Université de Montréal; Mila - Quebec Artificial Intelligence Institute","funders":"","keywords":"Environmental science","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":[0.0009491732,0.002235599,0.001373727,0.002759929,0.000786735,0.002398249,0.003464284,0.002394707,0.1195855],"category_scores_gemma":[0.004570344,0.0009162041,0.001582082,0.004966535,0.0004841019,0.002422104,0.002693929,0.002229112,0.2347792],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001211973,"about_ca_system_score_gemma":0.00175038,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.01751919,"about_ca_topic_score_gemma":0.02830377,"domain_scores_codex":[0.9990447,0.0001577388,0.00009351044,0.0003136385,0.0002184731,0.0001719348],"domain_scores_gemma":[0.998212,0.0003386093,0.0001321087,0.0006120512,0.0005373732,0.0001679603],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00002776241,0.000007797021,0.0003693319,0.0002025637,0.00001754347,0.00001166165,0.00001537476,0.0002560809,0.00008533536,0.0004069599,0.9975555,0.001043971],"study_design_scores_gemma":[0.0001432596,0.000009361203,0.002186978,0.0001191726,0.000018674,0.00003938426,0.00006523321,0.0006155499,0.0004458886,0.002003357,0.9943198,0.0000334557],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.0001020538,0.00002358759,0.0001764545,0.00005807372,0.0000264978,0.000006651479,0.9971348,0.001728662,0.0007432235],"genre_scores_gemma":[0.0003513123,0.00002103419,0.0003737424,0.00004835052,0.000007314688,0.0000394043,0.997991,0.0005675799,0.0006003999],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.1195855,"threshold_uncertainty_score":0.4000532,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.05436327821856646,"score_gpt":0.2364994271383783,"score_spread":0.1821361489198118,"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."}}