{"id":"W3173937618","doi":"10.18653/v1/2021.acl-short.36","title":"Enhancing Descriptive Image Captioning with Natural Language Inference","year":2021,"lang":"en","type":"article","venue":"","topic":"Multimodal Machine Learning Applications","field":"Computer Science","cited_by":15,"is_retracted":false,"has_abstract":true,"ca_institutions":"Queen's University","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Closed captioning; Computer science; Inference; Natural language; Computational linguistics; Natural language processing; Natural (archaeology); Joint (building); Artificial intelligence; Linguistics; Image (mathematics); History; Engineering; Philosophy","routes":{"ca_aff":true,"ca_fund":true,"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":[0.001201008,0.001597227,0.0009861393,0.001997618,0.0006909252,0.002235174,0.001814774,0.001409174,0.01132871],"category_scores_gemma":[0.004842476,0.0006012105,0.001322505,0.001773614,0.0007722728,0.004418594,0.001780818,0.002242572,0.005290468],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0006784875,"about_ca_system_score_gemma":0.0007906469,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003043111,"about_ca_topic_score_gemma":0.005256183,"domain_scores_codex":[0.9991691,0.0002556214,0.00005358897,0.0002221239,0.0002313548,0.00006822453],"domain_scores_gemma":[0.9978922,0.0008525567,0.0001198434,0.0004640321,0.0006129625,0.00005841138],"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":[0.0003622759,0.0002666516,0.0004040107,0.00112837,0.000109558,0.0003646748,0.0003753411,0.02241256,0.07624724,0.02676739,0.128841,0.7427209],"study_design_scores_gemma":[0.00008183983,0.0001181237,0.0006255294,0.00008699948,0.0001575244,0.0004964765,0.0002295631,0.7821003,0.09908224,0.03798409,0.07895012,0.00008720598],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.009457507,0.002287479,0.9549755,0.0009070651,0.0009461174,0.0002187301,0.002163227,0.01927774,0.00976671],"genre_scores_gemma":[0.1614631,0.002594645,0.8105478,0.0008890018,0.0006991553,0.0002640673,0.009273564,0.00276366,0.01150506],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.01132871,"threshold_uncertainty_score":0.03789836,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.007255126184124819,"score_gpt":0.2684051231919964,"score_spread":0.2611499970078716,"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."}}