{"id":"W3172246306","doi":"10.48550/arxiv.2004.02251","title":"Semantics of the Unwritten: The Effect of End of Paragraph and Sequence Tokens on Text Generation with GPT2","year":2020,"lang":"en","type":"preprint","venue":"arXiv (Cornell University)","topic":"Topic Modeling","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Waterloo","funders":"","keywords":"Paragraph; Perplexity; Computer science; Artificial intelligence; Natural language processing; Linguistics; Sequence (biology); Semantics (computer science); Language model; Philosophy; Chemistry","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":[0.00256094,0.001425284,0.0005846167,0.0006141078,0.0005290167,0.001097575,0.0009035975,0.001097336,0.002230007],"category_scores_gemma":[0.01510503,0.0004551467,0.0004630895,0.0004692403,0.0005418745,0.00191914,0.001211059,0.00202054,0.001045278],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0005872105,"about_ca_system_score_gemma":0.0008137019,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003447382,"about_ca_topic_score_gemma":0.005467311,"domain_scores_codex":[0.9985672,0.0007130821,0.00008845598,0.0003878362,0.0001498482,0.00009346654],"domain_scores_gemma":[0.9904116,0.007490433,0.0003286052,0.0008591207,0.0006445803,0.0002657902],"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":[0.002383561,0.0006008842,0.02607856,0.0008136664,0.0002557237,0.001145954,0.001041417,0.2679902,0.07797597,0.002429546,0.01423152,0.605053],"study_design_scores_gemma":[0.0002340278,0.0007358908,0.007222068,0.00006345887,0.0001646989,0.0004240362,0.0002451765,0.931172,0.05205138,0.003447696,0.004183032,0.00005661961],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.8405178,0.002288746,0.1345716,0.001114125,0.0004478469,0.0002561427,0.001293068,0.01042994,0.009080708],"genre_scores_gemma":[0.9562025,0.0002186637,0.03740878,0.0002852062,0.00006913689,0.0001675953,0.002454414,0.0006880207,0.002505533],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.003447382,"threshold_uncertainty_score":0.01354373,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.07345739299633469,"score_gpt":0.183247074308855,"score_spread":0.1097896813125203,"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."}}