{"id":"W3136363192","doi":"10.1016/j.csl.2022.101429","title":"On the effect of dropping layers of pre-trained transformer models","year":2022,"lang":"en","type":"article","venue":"Computer Speech & Language","topic":"Topic Modeling","field":"Computer Science","cited_by":91,"is_retracted":false,"has_abstract":false,"ca_institutions":"Dalhousie University","funders":"","keywords":"Transformer; Computer science; Limiting; Sentence; Artificial intelligence; Paraphrase; Machine learning; Natural language processing; Engineering","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":[0.004010307,0.002543015,0.001248386,0.0006874642,0.0009165172,0.002316335,0.001600391,0.002941982,0.01160052],"category_scores_gemma":[0.02877508,0.0008883465,0.0008605853,0.0007098003,0.0009613219,0.004007506,0.001739212,0.004294732,0.002401669],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00111846,"about_ca_system_score_gemma":0.002189918,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.03251716,"about_ca_topic_score_gemma":0.06206566,"domain_scores_codex":[0.9983554,0.000547333,0.0001060569,0.0003782406,0.000268485,0.000344414],"domain_scores_gemma":[0.9882004,0.008810316,0.0002234366,0.00125051,0.001166712,0.0003485485],"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":[0.009604643,0.001280585,0.007350964,0.0008998048,0.001031077,0.00079444,0.000530444,0.3584938,0.09318565,0.006643346,0.02711002,0.4930752],"study_design_scores_gemma":[0.0002535243,0.0006951978,0.003988402,0.0001689002,0.0006457352,0.0002524016,0.0003313408,0.9228308,0.06118226,0.00444181,0.005149716,0.00006003954],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.5522563,0.01074107,0.3827783,0.004456219,0.003866763,0.0003522816,0.002225076,0.01836221,0.02496172],"genre_scores_gemma":[0.8812878,0.001584765,0.09477663,0.001650488,0.0002346709,0.00007827592,0.002775754,0.002203322,0.01540825],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.03251716,"threshold_uncertainty_score":0.06465578,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.01020937020698539,"score_gpt":0.2361752071494904,"score_spread":0.225965836942505,"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."}}