{"id":"W6891607569","doi":"10.48448/fvef-7072","title":"Distilling Linguistic Context for Language Model Compression","year":2021,"lang":"en","type":"other","venue":"Underline Science Inc.","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"Kootenay Association for Science & Technology","funders":"","keywords":"Language model; Representation (politics); Context (archaeology); Knowledge representation and reasoning; Relation (database); Pruning; Universal Networking Language; Word (group theory); Artificial neural 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":[0.0003613878,0.0008983947,0.0007594553,0.000679541,0.0004444229,0.0008825929,0.001154175,0.0007968021,0.004510863],"category_scores_gemma":[0.003209249,0.0003540659,0.0004854928,0.000741101,0.0005413177,0.002312591,0.002153393,0.001817032,0.001289141],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0004968229,"about_ca_system_score_gemma":0.001147244,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00316028,"about_ca_topic_score_gemma":0.009328146,"domain_scores_codex":[0.999683,0.00007718382,0.00001712844,0.00009589553,0.00008331024,0.00004348319],"domain_scores_gemma":[0.999558,0.0001964088,0.00003218583,0.0001221255,0.00006744121,0.00002383895],"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":[0.0003338499,0.0001304434,0.000945696,0.0002190979,0.00006139297,0.000289648,0.0002197802,0.1847466,0.02947407,0.022555,0.011103,0.7499213],"study_design_scores_gemma":[0.00001948729,0.00003321646,0.0001603685,0.00002069785,0.00002066016,0.00005657606,0.0000473257,0.9687627,0.00851917,0.01801128,0.004338077,0.0000104465],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.07789232,0.001828756,0.9070991,0.000957093,0.000260207,0.00007129651,0.0004852627,0.006791613,0.00461427],"genre_scores_gemma":[0.6252969,0.0006046593,0.3663909,0.0004949333,0.0001680353,0.000158341,0.001275473,0.0006016228,0.005009129],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.004510863,"threshold_uncertainty_score":0.01509029,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.04778031652179233,"score_gpt":0.3565567001907823,"score_spread":0.30877638366899,"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."}}