{"id":"W4413025253","doi":"10.1007/978-3-032-00891-6_28","title":"Enhancing Ultra-Low-Bit Quantization of Large Language Models Through Saliency-Aware Partial Retraining","year":2025,"lang":"en","type":"book-chapter","venue":"Lecture notes in computer science","topic":"Topic Modeling","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":false,"ca_institutions":"University of Toronto; University of New Brunswick","funders":"","keywords":"Computer science; Retraining; Quantization (signal processing); Bit (key); Artificial intelligence; Speech recognition; Computer vision; Computer network","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":[0.0006206754,0.001057974,0.0008362975,0.0004511599,0.0003635596,0.0008188937,0.001135392,0.0008215533,0.005509301],"category_scores_gemma":[0.003490576,0.0003350181,0.0004119924,0.0006101517,0.0003931438,0.001898175,0.001258148,0.001518113,0.002298766],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0004497984,"about_ca_system_score_gemma":0.0009116894,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.005810058,"about_ca_topic_score_gemma":0.01415122,"domain_scores_codex":[0.9996043,0.00008594154,0.0000252306,0.00009064831,0.0001440237,0.000049781],"domain_scores_gemma":[0.9989614,0.0005522154,0.00005556341,0.0001643721,0.0002095828,0.00005678106],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.000543655,0.0002510769,0.0006342722,0.000254359,0.00007527662,0.0002245884,0.0001739435,0.06705014,0.09183444,0.00594007,0.01130921,0.8217089],"study_design_scores_gemma":[0.00002338661,0.000115177,0.0003516554,0.00002014501,0.00002680123,0.0001006989,0.00003188265,0.9729826,0.01808834,0.005885613,0.002354558,0.00001902357],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.04035264,0.002265716,0.9460774,0.0007582153,0.0003821636,0.00007545922,0.0005043741,0.006314085,0.003269901],"genre_scores_gemma":[0.5857134,0.001129758,0.3993454,0.001051724,0.0002804849,0.0001319898,0.001482932,0.0006815019,0.01018282],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.005810058,"threshold_uncertainty_score":0.01843041,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.02045388212035081,"score_gpt":0.2747283330984558,"score_spread":0.254274450978105,"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."}}