{"id":"W3096344818","doi":"10.1109/cvpr46437.2021.01544","title":"Permute, Quantize, and Fine-tune: Efficient Compression of Neural Networks","year":2021,"lang":"en","type":"preprint","venue":"","topic":"Advanced Neural Network Applications","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Toronto; University of Waterloo","funders":"","keywords":"Vector quantization; Computer science; Quantization (signal processing); Algorithm; Uncompressed video; Image compression; Artificial neural network; Data compression ratio; Convolutional neural network; Theoretical computer science; Artificial intelligence; Image processing; Image (mathematics); Object (grammar)","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":[0.0007347832,0.0006119364,0.0005049086,0.0004598121,0.0003528276,0.0007854567,0.0009532156,0.0006026507,0.002639857],"category_scores_gemma":[0.0049396,0.0002710721,0.000296181,0.0005333791,0.0008900855,0.002255094,0.001104728,0.001279445,0.0006240743],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0007380012,"about_ca_system_score_gemma":0.0006112875,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002024505,"about_ca_topic_score_gemma":0.003341749,"domain_scores_codex":[0.9996816,0.00006061397,0.00002550387,0.00006185747,0.0001335593,0.00003685206],"domain_scores_gemma":[0.9990238,0.0003931755,0.00008790863,0.0003327339,0.0001254972,0.00003693199],"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":[0.0003434775,0.0001504663,0.001234137,0.0001629036,0.00006200767,0.0002106806,0.0002590537,0.4078704,0.03941391,0.04253898,0.00630535,0.5014486],"study_design_scores_gemma":[0.00003602979,0.000119204,0.0003463014,0.00002927498,0.0000164668,0.0001011522,0.00004719394,0.9362019,0.02514091,0.03366357,0.00427864,0.00001931755],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.1334144,0.001640105,0.8541623,0.0008856729,0.0002149049,0.0001264154,0.0003070578,0.003078602,0.006170538],"genre_scores_gemma":[0.7270091,0.0007062687,0.2659528,0.0003968757,0.00008800392,0.0001968413,0.0004436539,0.0004180967,0.004788413],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.002639857,"threshold_uncertainty_score":0.008831203,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.02202910622663319,"score_gpt":0.2797244536248866,"score_spread":0.2576953473982534,"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."}}