{"id":"W4389518858","doi":"10.18653/v1/2023.findings-emnlp.674","title":"Balaur: Language Model Pretraining with Lexical Semantic Relations","year":2023,"lang":"en","type":"article","venue":"","topic":"Topic Modeling","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"McGill University","funders":"Samsung; Nvidia","keywords":"Computer science; Natural language processing; Inference; Artificial intelligence; Generalization; Transformer; Set (abstract data type); Language model; Meaning (existential); Interface (matter); Programming language; Psychology","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":[0.001161121,0.001458518,0.0008740108,0.0006859148,0.0004803435,0.001165824,0.002179395,0.001363094,0.01108983],"category_scores_gemma":[0.004659785,0.0007644216,0.001264038,0.0006014514,0.000616128,0.002886679,0.002109223,0.004424484,0.005295587],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0006998642,"about_ca_system_score_gemma":0.00110817,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.005281685,"about_ca_topic_score_gemma":0.01441947,"domain_scores_codex":[0.9994975,0.0001569345,0.00002616042,0.0002050601,0.00006160526,0.00005263988],"domain_scores_gemma":[0.9985332,0.0009753905,0.00004701037,0.0002377114,0.0001411431,0.00006548077],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.0006227451,0.0003361717,0.002255396,0.0004784427,0.0003567435,0.0002625977,0.0005393868,0.195968,0.02382097,0.0142908,0.04440174,0.716667],"study_design_scores_gemma":[0.00004635454,0.00006050744,0.0003309668,0.00002809012,0.00003485898,0.00005614332,0.00006939706,0.9758523,0.005692228,0.01345536,0.004350763,0.00002295467],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.02792571,0.0007581118,0.9303896,0.000668823,0.0003162387,0.0001513034,0.001120866,0.03428834,0.00438104],"genre_scores_gemma":[0.4706205,0.0004508914,0.5025095,0.001354062,0.0002274192,0.0006590839,0.006728627,0.003748091,0.01370163],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.01108983,"threshold_uncertainty_score":0.03709918,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.02994062432139795,"score_gpt":0.2649370903985889,"score_spread":0.2349964660771909,"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."}}