{"id":"W6948019733","doi":"10.48448/9cnb-s768","title":"Semantically-Prompted Language Models Improve Visual Descriptions","year":2024,"lang":"en","type":"other","venue":"Underline Science Inc.","topic":"Subterranean biodiversity and taxonomy","field":"Earth and Planetary Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Alberta","funders":"","keywords":"Visual language; Key (lock); Semantics (computer science); Visualization; Language model; Image (mathematics); Visual reasoning","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":[0.001561984,0.001879094,0.000623184,0.001445096,0.0004471984,0.002263903,0.002399953,0.001467774,0.007889401],"category_scores_gemma":[0.008500441,0.0005328608,0.001512765,0.0007321279,0.0008133693,0.005462929,0.002735649,0.002793,0.003121453],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001381844,"about_ca_system_score_gemma":0.001053169,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003346159,"about_ca_topic_score_gemma":0.00594516,"domain_scores_codex":[0.9986829,0.000384927,0.00009519324,0.0004360351,0.0003254781,0.00007533436],"domain_scores_gemma":[0.9967785,0.001675004,0.0001762019,0.0008140552,0.0004221147,0.0001340775],"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":[0.001141945,0.000830829,0.004357721,0.002131923,0.0002422249,0.0008241622,0.0009993146,0.1204919,0.04542832,0.07310133,0.1083077,0.6421427],"study_design_scores_gemma":[0.0002206264,0.0002968476,0.00108414,0.000151272,0.0001158813,0.0005679461,0.0004412604,0.8731693,0.02900376,0.05883315,0.03603247,0.00008335208],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.09906405,0.002384354,0.8138034,0.001862722,0.000748413,0.0006296755,0.01137899,0.04939863,0.02072976],"genre_scores_gemma":[0.5042888,0.0008087,0.4470719,0.001261749,0.0001727998,0.000351838,0.0333416,0.002801737,0.009900833],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.007889401,"threshold_uncertainty_score":0.02639264,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.03441052405999891,"score_gpt":0.2456936735914689,"score_spread":0.21128314953147,"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."}}