{"id":"W7116175540","doi":"10.3103/s1060992x25601733","title":"Memory Stream: Enhancing Information Flow in Recurrent Memory Transformers for Efficient Long-Context Training","year":2025,"lang":"en","type":"article","venue":"Optical Memory and Neural Networks","topic":"Generative Adversarial Networks and Image Synthesis","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"Artificial Intelligence in Medicine (Canada)","funders":"","keywords":"Transformer; High memory; Memory model; Computational complexity theory; Flat memory model; Content-addressable memory; Matching (statistics); Architecture","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":[0.0006777773,0.0002795205,0.0003935971,0.000175921,0.0002712615,0.000224017,0.0002927695,0.000158111,0.000009949009],"category_scores_gemma":[0.0001301825,0.0002492401,0.0001387962,0.0004434459,0.0001124938,0.000764221,0.0001032778,0.000360027,0.000001665507],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00006036033,"about_ca_system_score_gemma":0.00006422509,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00001046302,"about_ca_topic_score_gemma":0.0001117155,"domain_scores_codex":[0.9980903,0.0001182618,0.0005703094,0.0004390865,0.0001790565,0.0006029613],"domain_scores_gemma":[0.9988725,0.0005936858,0.00008234293,0.0002052218,0.00008017343,0.000166108],"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":[0.0001012685,0.00003668879,0.000004915036,0.00003804773,0.00001703349,0.000004485878,0.0007780436,0.2258962,0.00005931716,0.0003900781,0.0001083888,0.7725655],"study_design_scores_gemma":[0.001100898,0.0001505571,0.0002774567,0.0001558971,0.00002614896,0.000005548667,0.0006585749,0.9941956,0.002913028,0.0001072207,0.0001482318,0.000260841],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.0425542,0.0007033077,0.9519312,0.001196678,0.001438576,0.0007002422,0.000003573649,0.00006605145,0.001406178],"genre_scores_gemma":[0.9921139,0.00006094353,0.006553793,0.0009229535,0.0001708348,0.00008197949,0.00001192972,0.000008719935,0.00007497501],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9495597,"threshold_uncertainty_score":0.999996,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.01365455953999392,"score_gpt":0.2390468300254826,"score_spread":0.2253922704854887,"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."}}