{"id":"W4385573486","doi":"10.18653/v1/2022.findings-emnlp.191","title":"Partially-Random Initialization: A Smoking Gun for Binarization Hypothesis of BERT","year":2022,"lang":"en","type":"article","venue":"","topic":"Topic Modeling","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"ca_institutions":"Huawei Technologies (Canada)","funders":"","keywords":"Computer science; Initialization; Embedding; Encoder; Task (project management); Set (abstract data type); Artificial intelligence; Enhanced Data Rates for GSM Evolution; Binary number; Edge device; Machine learning; Programming language; Operating system","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":[0.001492608,0.001569743,0.001020189,0.0005128603,0.0006365424,0.001183563,0.001960006,0.001467346,0.00331542],"category_scores_gemma":[0.009197013,0.0006015947,0.0006249588,0.0004955388,0.001487018,0.00332472,0.00171114,0.003047503,0.001788051],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0009599857,"about_ca_system_score_gemma":0.001224183,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003936312,"about_ca_topic_score_gemma":0.006282607,"domain_scores_codex":[0.9991539,0.0002545542,0.00003643718,0.0003067065,0.0001225379,0.0001258554],"domain_scores_gemma":[0.9978034,0.0009340739,0.0001635964,0.0006913077,0.0002754513,0.0001322023],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.002294004,0.0004101199,0.005923495,0.0004199279,0.0002489951,0.0004324546,0.0004086239,0.5588955,0.03006922,0.06532047,0.03303541,0.3025419],"study_design_scores_gemma":[0.00004540955,0.0001094245,0.0004709167,0.00003176504,0.00002155603,0.00008725516,0.0000233483,0.9745044,0.007947485,0.014927,0.001809827,0.00002156323],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.1150878,0.001785322,0.8660294,0.00174997,0.0003773505,0.0001493135,0.0005668668,0.007445687,0.006808387],"genre_scores_gemma":[0.8255872,0.0004361598,0.1623134,0.001362752,0.0001725351,0.0002364757,0.001898305,0.001001252,0.006991871],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.003936312,"threshold_uncertainty_score":0.01109117,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.04659736887388309,"score_gpt":0.2515191874708972,"score_spread":0.2049218185970141,"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."}}