{"id":"W4385573233","doi":"10.18653/v1/2022.emnlp-main.97","title":"On the Transformation of Latent Space in Fine-Tuned NLP Models","year":2022,"lang":"en","type":"article","venue":"","topic":"Topic Modeling","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":true,"ca_institutions":"Dalhousie University","funders":"","keywords":"Computer science; Artificial intelligence; Space (punctuation); Similarity (geometry); Task (project management); Cluster analysis; Transformation (genetics); Function (biology); Probabilistic latent semantic analysis; Natural language processing; Class (philosophy); Machine learning","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":[0.002360719,0.0007970469,0.0006887429,0.000569932,0.0005523012,0.001560075,0.001012398,0.001219422,0.002252788],"category_scores_gemma":[0.01637118,0.000603971,0.0009011964,0.0004638996,0.001874798,0.003602232,0.001882629,0.003076984,0.0003704503],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001525043,"about_ca_system_score_gemma":0.0007563467,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00508595,"about_ca_topic_score_gemma":0.00515872,"domain_scores_codex":[0.9991636,0.0004228984,0.00003101085,0.0001962295,0.00008315625,0.0001031888],"domain_scores_gemma":[0.9915361,0.00674521,0.0005317678,0.0007137868,0.0002870935,0.0001859648],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"observational","study_design_scores_codex":[0.0001099538,0.0000711244,0.001907183,0.00005473106,0.00005158914,0.0001013445,0.0003724355,0.9368603,0.005863358,0.03014271,0.0005518682,0.02391338],"study_design_scores_gemma":[0.000005993338,0.00002779939,0.0003252923,0.000006548639,0.000004675635,0.00001629557,0.00001947095,0.9810084,0.0006596798,0.01777031,0.000148977,0.000006524813],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.1736846,0.0004189948,0.821371,0.0008222713,0.00003854956,0.00007313642,0.000160531,0.0007688024,0.002662201],"genre_scores_gemma":[0.9324337,0.0002141677,0.06441201,0.0001815553,0.00003541408,0.0001048695,0.0002890546,0.0002195156,0.002109688],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.00508595,"threshold_uncertainty_score":0.01248479,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.04097610522250977,"score_gpt":0.2265543457583512,"score_spread":0.1855782405358414,"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."}}