{"id":"W4412886796","doi":"10.18653/v1/2025.acl-long.111","title":"LLäMmlein: Transparent, Compact and Competitive German-Only Language Models from Scratch","year":2025,"lang":"en","type":"article","venue":"","topic":"Natural Language Processing Techniques","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"Artificial Intelligence in Medicine (Canada)","funders":"Julius-Maximilians-Universität Würzburg; Deutsche Forschungsgemeinschaft","keywords":"Scratch; German; Computer science; Programming language; Linguistics; Philosophy","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":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001520076,0.002589933,0.0009124381,0.0009019253,0.0004417438,0.001945806,0.003313062,0.001798573,0.01315001],"category_scores_gemma":[0.004856039,0.001326314,0.001388207,0.0006594691,0.000549509,0.00392263,0.002805915,0.002502976,0.01138089],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001066676,"about_ca_system_score_gemma":0.001650051,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.008698596,"about_ca_topic_score_gemma":0.02142095,"domain_scores_codex":[0.9991229,0.0002525938,0.000059646,0.0002662178,0.0001814274,0.0001171605],"domain_scores_gemma":[0.9988398,0.0004766347,0.00005101116,0.000400781,0.000162892,0.00006892167],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.001981875,0.0007690516,0.004375844,0.001518568,0.000778075,0.001030605,0.0004801661,0.3611196,0.02471763,0.02004309,0.227282,0.3559036],"study_design_scores_gemma":[0.0003076321,0.000331711,0.0009363662,0.0001196412,0.0001293065,0.0003250991,0.0001118799,0.909255,0.02579438,0.01120457,0.05137981,0.0001046057],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.1065035,0.002702839,0.5773537,0.002096436,0.001378253,0.0006730403,0.03215015,0.2389492,0.03819281],"genre_scores_gemma":[0.4613295,0.001261907,0.3639103,0.001701506,0.0001449058,0.001153439,0.1232978,0.01693879,0.03026194],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.01315001,"threshold_uncertainty_score":0.04399115,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.0159280122699901,"score_gpt":0.3034245621299598,"score_spread":0.2874965498599698,"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."}}