{"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":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0003392902,0.0001521567,0.0001722611,0.0001380826,0.0001230719,0.0003267603,0.0005797758,0.00007531978,0.00006229904],"category_scores_gemma":[0.00001723236,0.0001322308,0.0001127194,0.0002758885,0.000007417076,0.0007406521,0.0001242606,0.000104789,0.00004372399],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00003229912,"about_ca_system_score_gemma":0.000111478,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00002136993,"about_ca_topic_score_gemma":0.00003036014,"domain_scores_codex":[0.9985816,0.0000262771,0.0002733374,0.0005138465,0.0002172509,0.0003877147],"domain_scores_gemma":[0.9992173,0.000161548,0.00002902651,0.0004525436,0.00005424008,0.00008537828],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.000003172303,0.00003055949,0.0000111489,0.0001295484,0.00003081202,0.0000245197,0.01080374,0.005756391,0.0009242223,0.945916,0.001166806,0.03520309],"study_design_scores_gemma":[0.0002921668,0.00001761365,0.000005534882,0.00006758692,0.000006244871,0.000006199453,0.0002036027,0.9814268,0.0007615508,0.01487294,0.002155442,0.0001843317],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.006956525,0.0003932084,0.9868457,0.001101918,0.000746594,0.0002666094,0.00001580061,0.00083577,0.002837836],"genre_scores_gemma":[0.8990212,0.000007825629,0.09890697,0.0005140868,0.00021623,0.00003138247,0.00001833015,0.00002249575,0.001261451],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.9756704,"threshold_uncertainty_score":0.5392213,"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."}}