{"id":"W4412955598","doi":"10.1088/1742-5468/adefb1","title":"How feature learning can improve neural scaling laws<sup>*</sup>","year":2025,"lang":"en","type":"article","venue":"Journal of Statistical Mechanics Theory and Experiment","topic":"Neural Networks and Applications","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"ca_institutions":"Artificial Intelligence in Medicine (Canada)","funders":"","keywords":"Feature (linguistics); Scaling law; Scaling; Artificial neural network; Statistical physics; Artificial intelligence; Computer science; Law; Mathematics; Physics; Political science; Philosophy; Geometry","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":[0.000512194,0.0001307898,0.0002146948,0.00006037117,0.0002276044,0.0002887522,0.0003074698,0.00005892736,0.000004763153],"category_scores_gemma":[0.0001078503,0.0000997139,0.00005463388,0.0001443184,0.00003207634,0.0001630998,0.0001612326,0.0004353076,4.83855e-7],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00003495242,"about_ca_system_score_gemma":0.00003780892,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00000111101,"about_ca_topic_score_gemma":1.741442e-7,"domain_scores_codex":[0.9990056,0.0001726464,0.0002229508,0.0001979554,0.0001841179,0.000216764],"domain_scores_gemma":[0.999064,0.0004138986,0.0001336201,0.0001501637,0.00008963006,0.0001487038],"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":[0.00003959132,0.00005443597,0.000001208629,0.00001006276,0.00002489802,0.00002095161,0.0003260956,0.0003779159,0.004736247,0.9470413,0.0003135404,0.04705378],"study_design_scores_gemma":[0.0008209248,0.0004448845,0.000008545905,0.00008827227,0.0000423027,0.000079767,0.001228211,0.3781686,0.01470565,0.599239,0.00496026,0.0002135726],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.008495748,0.0007549285,0.986294,0.004017773,0.0002381108,0.00009192608,0.000004740913,0.00002001361,0.00008279],"genre_scores_gemma":[0.9649683,0.00007273774,0.0339512,0.0005385319,0.0001166695,0.000007006277,0.000001075026,0.0000065754,0.0003378925],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.9564726,"threshold_uncertainty_score":0.4066215,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.007363415210228173,"score_gpt":0.259062483169425,"score_spread":0.2516990679591969,"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."}}