{"id":"W3207612418","doi":"10.18653/v1/2022.acl-short.24","title":"Kronecker Decomposition for GPT Compression","year":2022,"lang":"en","type":"preprint","venue":"","topic":"Algorithms and Data Compression","field":"Computer Science","cited_by":17,"is_retracted":false,"has_abstract":true,"ca_institutions":"McGill University","funders":"","keywords":"Kronecker delta; Decomposition; Computer science; Kronecker product; Volume (thermodynamics); Compression (physics); Linguistics; Philosophy; Physics","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":[0.0006979686,0.0006204495,0.0005300716,0.001114164,0.000359392,0.001469094,0.0005254769,0.0006921972,0.006943559],"category_scores_gemma":[0.003274417,0.0002470807,0.0004675671,0.001532734,0.0006362288,0.001579417,0.001115876,0.001294425,0.003773326],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0003503482,"about_ca_system_score_gemma":0.0007177524,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001277157,"about_ca_topic_score_gemma":0.001636569,"domain_scores_codex":[0.9994112,0.0001447512,0.00005280839,0.00007771109,0.0002565384,0.00005704403],"domain_scores_gemma":[0.9990228,0.0002703635,0.00005607954,0.0003050739,0.0002962155,0.00004940835],"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":[0.0003860528,0.0001085418,0.0007618276,0.0002475003,0.00006060387,0.0004386622,0.0001791858,0.07653961,0.02036297,0.2815207,0.03283216,0.5865622],"study_design_scores_gemma":[0.00004685477,0.0001364281,0.0007105278,0.0001122613,0.00003295435,0.0007154244,0.0001098805,0.7526376,0.01336352,0.2058809,0.02621006,0.00004359489],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.01455072,0.002347074,0.971184,0.0006280144,0.0003632643,0.00006938977,0.0006841997,0.0006964631,0.009476905],"genre_scores_gemma":[0.3025215,0.004731774,0.6626742,0.0005488622,0.0007555123,0.000347617,0.00383623,0.000598568,0.02398567],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.006943559,"threshold_uncertainty_score":0.02322853,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.02939809692835871,"score_gpt":0.3341453645357702,"score_spread":0.3047472676074115,"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."}}