{"id":"W3197581510","doi":"10.1609/aaai.v36i10.21306","title":"Retrieve, Caption, Generate: Visual Grounding for Enhancing Commonsense in Text Generation Models","year":2022,"lang":"en","type":"preprint","venue":"Proceedings of the AAAI Conference on Artificial Intelligence","topic":"Multimodal Machine Learning Applications","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Waterloo","funders":"","keywords":"Commonsense reasoning; Closed captioning; Computer science; Fluency; Transformer; Commonsense knowledge; Generative grammar; Artificial intelligence; Natural language processing; Image (mathematics); Linguistics; Engineering; Knowledge extraction","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":["metaepi_narrow"],"consensus_categories":[],"category_scores_codex":[0.00166952,0.0004193254,0.0005003658,0.0004089447,0.0005331907,0.0005768011,0.002670354,0.0002466482,0.00005906407],"category_scores_gemma":[0.0006579452,0.0003982006,0.0002050229,0.000909564,0.0001406789,0.0003655134,0.001814522,0.001289453,0.00001462303],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0003788724,"about_ca_system_score_gemma":0.0003667242,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0009507194,"about_ca_topic_score_gemma":0.0002923829,"domain_scores_codex":[0.9964491,0.00008867995,0.001158495,0.001159445,0.0006785804,0.0004656811],"domain_scores_gemma":[0.9972101,0.0002487608,0.001026178,0.0005945244,0.0008280515,0.00009235722],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.00005950596,0.0002329849,0.0001452613,0.000128311,0.00002061271,3.653725e-7,0.003022826,0.06841739,0.07438916,0.8321348,0.0001013889,0.02134739],"study_design_scores_gemma":[0.00003456766,0.00009353005,0.0001216787,0.0001527661,0.00001333165,0.000002172216,0.0002717386,0.773569,0.07346305,0.151933,0.00002858273,0.0003165098],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.4503142,0.00004384155,0.5415176,0.004106794,0.0008557382,0.001839842,0.00002670441,0.0001520998,0.001143271],"genre_scores_gemma":[0.9797434,0.00004361543,0.01893819,0.0001687693,0.0002038551,0.0007198716,0.00002012472,0.00003812114,0.0001240635],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.7051517,"threshold_uncertainty_score":0.999847,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1422581201807229,"score_gpt":0.355538293976938,"score_spread":0.2132801737962151,"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."}}