{"id":"W7077881463","doi":"10.48448/s5bq-8f93","title":"Pause-Tuning for Long-Context Comprehension: A Lightweight Approach to LLM Attention Recalibration","year":2025,"lang":"en","type":"other","venue":"Underline Science Inc.","topic":"Water Quality and Resources Studies","field":"Environmental Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Waterloo","funders":"","keywords":"Comprehension; Language model; Code (set theory); Source code; Information processing","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":[0.00257468,0.003156143,0.001789012,0.001500781,0.00077665,0.002116305,0.006192837,0.002655838,0.00735593],"category_scores_gemma":[0.0128329,0.0009798902,0.001706432,0.0009849272,0.001061866,0.00554342,0.004743054,0.005047169,0.004580006],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0015856,"about_ca_system_score_gemma":0.002184462,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.008468579,"about_ca_topic_score_gemma":0.01488405,"domain_scores_codex":[0.9982525,0.000482769,0.0001033851,0.0007393516,0.0002265154,0.0001955011],"domain_scores_gemma":[0.9962792,0.001714053,0.0001789445,0.001165383,0.0004274374,0.0002349437],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.001643034,0.0009259775,0.009647355,0.0008217549,0.0004283459,0.0005592086,0.001635649,0.1199251,0.05996349,0.005581601,0.05936152,0.739507],"study_design_scores_gemma":[0.000189398,0.0003924921,0.001218659,0.00006275015,0.0001228614,0.0001233064,0.0002669063,0.9528255,0.01874327,0.01295391,0.0130175,0.00008342788],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.116985,0.003868033,0.6521306,0.002169153,0.0007784964,0.0005952669,0.002952967,0.2134882,0.007032338],"genre_scores_gemma":[0.6356016,0.0006957565,0.3357288,0.002291773,0.0003427569,0.001206589,0.007090117,0.008396256,0.00864625],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.008468579,"threshold_uncertainty_score":0.02460802,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.05465495095434556,"score_gpt":0.2723230830635732,"score_spread":0.2176681321092277,"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."}}