{"id":"W4389523983","doi":"10.18653/v1/2023.emnlp-main.362","title":"Fast and Robust Early-Exiting Framework for Autoregressive Language Models with Synchronized Parallel Decoding","year":2023,"lang":"en","type":"article","venue":"","topic":"Topic Modeling","field":"Computer Science","cited_by":13,"is_retracted":false,"has_abstract":true,"ca_institutions":"Kootenay Association for Science & Technology","funders":"Institute for Information and Communications Technology Promotion; Ministry of Science and ICT, South Korea; Korea Advanced Institute of Science and Technology","keywords":"Computer science; Security token; Decoding methods; Autoregressive model; Language model; Inference; Estimator; Algorithm; Artificial intelligence; Mathematics","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":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0002125717,0.000138961,0.0001730242,0.00008843741,0.0001881092,0.0002430811,0.000362732,0.00007340225,0.000005618334],"category_scores_gemma":[0.00005742752,0.0001084067,0.00003339899,0.0001992903,0.00002760208,0.0004530107,0.0002287047,0.0001280369,0.000007330122],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00002707317,"about_ca_system_score_gemma":0.00004185516,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00005449475,"about_ca_topic_score_gemma":0.00002090433,"domain_scores_codex":[0.9988229,0.0000208869,0.0001601737,0.0004381954,0.0001793377,0.0003785676],"domain_scores_gemma":[0.9991351,0.0003027353,0.00007181489,0.0003513998,0.0000492785,0.00008963561],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.00001673079,0.00001097136,0.0005090646,0.00006426152,0.00004000278,0.00005205028,0.01087057,0.305082,0.00006143246,0.6300105,0.0001062545,0.05317613],"study_design_scores_gemma":[0.0004091692,0.00003587242,0.000144802,0.000116366,0.000005341244,0.000008199131,0.0005857699,0.976925,0.00008295076,0.02151657,0.000004876325,0.0001650957],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.07794382,0.00008252922,0.9199706,0.0005018392,0.00007847592,0.0002429603,0.000001773668,0.0005163065,0.0006616471],"genre_scores_gemma":[0.4033975,0.000004934606,0.5961037,0.0000894311,0.00006200246,0.00002647235,0.00000136433,0.00001167775,0.0003029566],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.671843,"threshold_uncertainty_score":0.4420699,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.03884542175441516,"score_gpt":0.2738687050378683,"score_spread":0.2350232832834531,"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."}}