{"id":"W4415535740","doi":"10.1016/j.array.2025.100538","title":"Mamba-caption: Long-range sequence modelling for efficient and accurate image captioning","year":2025,"lang":"en","type":"article","venue":"Array","topic":"Multimodal Machine Learning Applications","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"ca_institutions":"Université de Moncton","funders":"University of Johannesburg","keywords":"Closed captioning; Security token; Convolutional neural network; Trigram; Embedding; Clipping (morphology); Decoding methods; Image (mathematics); Pattern recognition (psychology)","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":[],"consensus_categories":[],"category_scores_codex":[0.0002764828,0.0001074873,0.0001038558,0.00008956556,0.00032373,0.0001962194,0.0003421446,0.00003926375,0.00000418319],"category_scores_gemma":[0.00004564455,0.0001078303,0.00003786656,0.0002704368,0.00004862037,0.0001866668,0.0000789094,0.0001218707,0.00002542706],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00004117974,"about_ca_system_score_gemma":0.00004281929,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001393465,"about_ca_topic_score_gemma":0.000006024338,"domain_scores_codex":[0.9990983,0.00003275898,0.0001656579,0.0003981962,0.0001003644,0.0002047241],"domain_scores_gemma":[0.9992341,0.0001680471,0.00006454988,0.0003821233,0.00009734483,0.00005388082],"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":[0.00001575748,0.00008971311,0.00327863,0.000149325,0.0000349311,0.00000424362,0.001742318,0.730414,0.02888883,0.2156384,0.00019155,0.01955227],"study_design_scores_gemma":[0.0002305867,0.00001198734,0.003036645,0.00003865738,0.000007892429,0.000004347229,0.0000129622,0.9922223,0.0007573893,0.003129556,0.000432962,0.0001146959],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.09550691,0.0001101797,0.9004003,0.002583886,0.0001042435,0.0003051174,0.000004426,0.0001659034,0.0008190611],"genre_scores_gemma":[0.7688943,0.000006916083,0.2305865,0.0001843139,0.00002702429,0.0001076166,0.00000469418,0.000005722114,0.0001829032],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.6733873,"threshold_uncertainty_score":0.4397193,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.02984790941700368,"score_gpt":0.3121207670062762,"score_spread":0.2822728575892725,"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."}}