{"id":"W4403851989","doi":"10.48550/arxiv.2410.01686","title":"Positional Attention: Expressivity and Learnability of Algorithmic Computation","year":2024,"lang":"en","type":"preprint","venue":"arXiv (Cornell University)","topic":"Neural Networks and Applications","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Natural Sciences and Engineering Research Council of Canada; Alexander S. Onassis Public Benefit Foundation; DeepMind","keywords":"Generalization; Computer science; Expressivity; Artificial intelligence; Artificial neural network; Distribution (mathematics); Cognitive science; Psychology; Mathematics","routes":{"ca_aff":false,"ca_fund":true,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":true},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003114896,0.0007409306,0.0008723834,0.001001912,0.0007129129,0.002858334,0.002248708,0.001561627,0.003928929],"category_scores_gemma":[0.03614186,0.0008349586,0.001088072,0.00106487,0.003551637,0.01018964,0.002922085,0.002996658,0.0005361525],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002665885,"about_ca_system_score_gemma":0.0016956,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002745548,"about_ca_topic_score_gemma":0.002496703,"domain_scores_codex":[0.9976966,0.000752371,0.0001545492,0.0005886108,0.0005417275,0.0002661963],"domain_scores_gemma":[0.97723,0.01551259,0.001481892,0.0040838,0.001186862,0.0005047626],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.0006316974,0.0002222351,0.009518033,0.0003743049,0.0001290901,0.0002691096,0.0008465671,0.2787488,0.0149646,0.5503176,0.002481187,0.1414967],"study_design_scores_gemma":[0.00002369429,0.000076362,0.0008233501,0.00002073142,0.00002939764,0.00007780078,0.00004672098,0.5811466,0.005260621,0.4117499,0.0007314903,0.00001328358],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.2190911,0.0004273899,0.769324,0.00174994,0.00003443997,0.00006979967,0.0002579126,0.001368949,0.007676528],"genre_scores_gemma":[0.9354227,0.0004255896,0.06024564,0.000289583,0.00006987708,0.0001209267,0.0003249741,0.0003107705,0.002789862],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.003928929,"threshold_uncertainty_score":0.01934248,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.04893848728949905,"score_gpt":0.1998394377325395,"score_spread":0.1509009504430405,"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."}}