{"id":"W4416017819","doi":"10.1109/iccd65941.2025.00068","title":"LM-Fix: Lightweight Bit-Flip Detection and Rapid Recovery Framework for Language Models","year":2025,"lang":"","type":"preprint","venue":"","topic":"Adversarial Robustness in Machine Learning","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"ca_institutions":"Artificial Intelligence in Medicine (Canada)","funders":"National Science Foundation","keywords":"Robustness (evolution); Overhead (engineering); Security token; Reliability (semiconductor); Speedup; Offset (computer science)","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":[0.002926971,0.001945147,0.001240928,0.001857031,0.0007535262,0.00193404,0.005444876,0.002324665,0.00556],"category_scores_gemma":[0.01107926,0.001096199,0.003105877,0.0005723641,0.001778254,0.005337562,0.004179625,0.003091683,0.002260407],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00183513,"about_ca_system_score_gemma":0.00320049,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.006403499,"about_ca_topic_score_gemma":0.008328604,"domain_scores_codex":[0.9970825,0.0006803257,0.0002423397,0.000453315,0.001286705,0.0002548094],"domain_scores_gemma":[0.9957463,0.001435312,0.0005816098,0.001522901,0.0005873667,0.0001265582],"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":[0.0009070861,0.0003106155,0.004845385,0.0009259586,0.0004149596,0.0009367704,0.0007976217,0.330411,0.04245071,0.07891428,0.03342775,0.5056579],"study_design_scores_gemma":[0.00003698435,0.0001021967,0.0001472893,0.0000371708,0.00003862705,0.0001739402,0.0000405059,0.9487953,0.01729654,0.02668447,0.006593353,0.00005368464],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.003785139,0.0001516827,0.9535004,0.000171587,0.00003755922,0.0001216553,0.0002329787,0.04151253,0.000486425],"genre_scores_gemma":[0.2137788,0.0002872357,0.7753523,0.0004902055,0.00008045795,0.0006869403,0.001304241,0.004159241,0.003860527],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.006403499,"threshold_uncertainty_score":0.01860011,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.02103400840634977,"score_gpt":0.2885844262927961,"score_spread":0.2675504178864463,"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."}}