{"id":"W4416111891","doi":"10.48550/arxiv.2503.03862","title":"Not-Just-Scaling Laws: Towards a Better Understanding of the Downstream Impact of Language Model Design Decisions","year":2025,"lang":"en","type":"preprint","venue":"arXiv (Cornell University)","topic":"Topic Modeling","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Natural Sciences and Engineering Research Council of Canada; National Institute of Advanced Industrial Science and Technology; National Science Foundation","keywords":"Downstream (manufacturing); Language model; Scale (ratio); Data modeling; Code (set theory); Model building; Training (meteorology); Simulation modeling","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":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.01243175,0.002039007,0.001108431,0.001315084,0.0007234818,0.00534481,0.002363377,0.00154654,0.00277061],"category_scores_gemma":[0.06464187,0.001213387,0.001407413,0.00142889,0.002383616,0.01666746,0.003203289,0.007268314,0.002361394],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001458301,"about_ca_system_score_gemma":0.001920801,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.005603741,"about_ca_topic_score_gemma":0.009801719,"domain_scores_codex":[0.9956692,0.001721549,0.0002176691,0.001212089,0.0008812323,0.0002982562],"domain_scores_gemma":[0.9668342,0.02041299,0.001413092,0.008210169,0.002510309,0.0006191873],"domain_codex":null,"domain_gemma":"methods","domain_candidate":"methods","domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.001075378,0.0006450482,0.1077787,0.001430115,0.001103575,0.0005574764,0.002914816,0.4256585,0.04703962,0.07091201,0.03434954,0.3065352],"study_design_scores_gemma":[0.00008283844,0.0003299043,0.01066893,0.0002385848,0.0002259408,0.0001522226,0.0005898666,0.8390755,0.02416267,0.1105941,0.01374434,0.0001351544],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.3862223,0.004510076,0.5682083,0.01138085,0.0005385578,0.0002780429,0.003353609,0.0102777,0.01523056],"genre_scores_gemma":[0.7897211,0.001515287,0.1960489,0.001666456,0.0001801848,0.0003273661,0.00373083,0.00389616,0.002913598],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9875683,"threshold_uncertainty_score":0.06574619,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.2135580531992839,"score_gpt":0.252831344134206,"score_spread":0.0392732909349221,"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."}}