{"id":"W7102431179","doi":"","title":"Key and Value Weights Are Probably All You Need: On the Necessity of the Query, Key, Value weight Triplet in Encoder-Only and Decoder-Only Transformers","year":2025,"lang":"","type":"article","venue":"ArXiv.org","topic":"Error Correcting Code Techniques","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Centre National de la Recherche Scientifique; Institut national de recherche en informatique et en automatique (INRIA); Institute for Catastrophic Loss Reduction","keywords":"Disjoint sets; Weight function; Quadratic equation; Scaling; Function (biology); Transformer; Stability (learning theory)","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":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow"],"consensus_categories":[],"category_scores_codex":[0.002662961,0.0009244281,0.001050586,0.0004564839,0.000839331,0.0003117601,0.002614144,0.0005427454,0.00001572961],"category_scores_gemma":[0.0005424211,0.0006233624,0.0003391311,0.002104009,0.001022263,0.0008413346,0.000998852,0.001625902,0.000007156048],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0002768921,"about_ca_system_score_gemma":0.001027871,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001512717,"about_ca_topic_score_gemma":0.003673834,"domain_scores_codex":[0.9934983,0.001423807,0.001404261,0.001694454,0.00086376,0.001115385],"domain_scores_gemma":[0.9955816,0.001335187,0.0007832765,0.001818822,0.0002549256,0.0002262132],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.0002355951,0.0005864798,0.9312431,0.0004259117,0.0001831229,0.00002950793,0.01295502,0.00007311883,0.001043247,0.03345529,0.001355118,0.01841444],"study_design_scores_gemma":[0.003105972,0.0009954504,0.8627647,0.004863012,0.0004018239,0.0001538751,0.001525424,0.04190058,0.04579442,0.02923949,0.007576839,0.001678418],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9506864,0.001957331,0.01259918,0.02672647,0.001296259,0.002231027,0.00002672748,0.0002209736,0.004255631],"genre_scores_gemma":[0.9919851,0.00138913,0.002464875,0.003244632,0.00008200155,0.0000757649,0.000002144319,0.00005282076,0.0007035483],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.06847846,"threshold_uncertainty_score":0.9996217,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.0219917367474046,"score_gpt":0.2580490801467694,"score_spread":0.2360573433993648,"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."}}