{"id":"W4404783494","doi":"10.18653/v1/2024.emnlp-industry.10","title":"DL-QAT: Weight-Decomposed Low-Rank Quantization-Aware Training for Large Language Models","year":2024,"lang":"en","type":"article","venue":"","topic":"Topic Modeling","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"ca_institutions":"Advanced Micro Devices (Canada)","funders":"","keywords":"Computer science; Quantization (signal processing); Rank (graph theory); Language model; Artificial intelligence; Natural language processing; Speech recognition; Mathematics; Algorithm; Combinatorics","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":[0.002261813,0.001977854,0.001564696,0.000908668,0.0007891757,0.001398532,0.002672406,0.001726748,0.007211184],"category_scores_gemma":[0.01342353,0.0008072223,0.001186127,0.001194309,0.000910436,0.003455897,0.001932977,0.004364688,0.003995012],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0009662759,"about_ca_system_score_gemma":0.002062018,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.01180017,"about_ca_topic_score_gemma":0.02389491,"domain_scores_codex":[0.9981964,0.0006767206,0.0001319386,0.0004214236,0.0004078581,0.0001655676],"domain_scores_gemma":[0.9962229,0.002063018,0.0001652143,0.0007549096,0.0006399732,0.0001540101],"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":[0.000538329,0.0003532125,0.001879801,0.0004436501,0.000220414,0.0002457611,0.0002898304,0.2613691,0.01641213,0.0121205,0.0401971,0.6659302],"study_design_scores_gemma":[0.00003492139,0.00005739166,0.0001406873,0.00001419113,0.00001516147,0.00004540232,0.00002797407,0.987599,0.003010788,0.00736814,0.001671615,0.00001481574],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.01016297,0.001059297,0.9744257,0.000542941,0.0002216999,0.0001182052,0.0005073943,0.01151503,0.001446737],"genre_scores_gemma":[0.2839161,0.0007259638,0.6999878,0.001368148,0.0003377846,0.0004389772,0.004819691,0.001651734,0.006753762],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.01180017,"threshold_uncertainty_score":0.02412379,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.04104453252529369,"score_gpt":0.2967512469939502,"score_spread":0.2557067144686566,"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."}}