{"id":"W4403569925","doi":"10.48550/arxiv.2410.09615","title":"SLiM: One-shot Quantization and Sparsity with Low-rank Approximation for LLM Weight Compression","year":2024,"lang":"en","type":"preprint","venue":"arXiv (Cornell University)","topic":"Sparse and Compressive Sensing Techniques","field":"Engineering","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Alliance de recherche numérique du Canada; Natural Sciences and Engineering Research Council of Canada; Canada Research Chairs; DeepMind","keywords":"Shot (pellet); Rank (graph theory); One shot; Mathematics; Physics; Econometrics; Combinatorics; Chemistry; Engineering","routes":{"ca_aff":false,"ca_fund":true,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":true},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow"],"consensus_categories":[],"category_scores_codex":[0.00008228974,0.0003011423,0.0003117614,0.0002353447,0.000105586,0.00008551864,0.0002005493,0.0002885,0.000007430128],"category_scores_gemma":[0.000006155663,0.000320125,0.0000790682,0.0002118289,0.00006969601,0.0001268284,0.0003046748,0.0003830873,0.000006797315],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001014291,"about_ca_system_score_gemma":0.00002885976,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00003419592,"about_ca_topic_score_gemma":0.00002908279,"domain_scores_codex":[0.9989648,0.00003233313,0.0001479559,0.0005702506,0.00007076081,0.0002139088],"domain_scores_gemma":[0.9992607,0.00005102045,0.00008311995,0.0004105067,0.0001167778,0.00007791359],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0004474976,0.0002235262,0.001428722,0.003083732,0.0006524486,0.0001365298,0.0005637998,0.9208597,0.009839816,0.05743059,0.00334994,0.001983683],"study_design_scores_gemma":[0.0003978172,0.00006653043,0.0003779615,0.001190834,0.0002832108,0.000003917509,0.00003363286,0.9545549,0.01568367,0.02657342,0.0003618151,0.0004722559],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.5811942,0.0002192355,0.4154924,0.00002330151,0.000246283,0.0005640756,0.00003635676,0.0009157335,0.00130842],"genre_scores_gemma":[0.9962757,0.0004201817,0.002872915,0.00001471088,0.00009039976,0.000003175639,0.0001337367,0.00005629565,0.0001329493],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.4150814,"threshold_uncertainty_score":0.9999251,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.08459366769301566,"score_gpt":0.1957024386642909,"score_spread":0.1111087709712753,"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."}}