{"id":"W4389523869","doi":"10.18653/v1/2023.conll-babylm.5","title":"Grammar induction pretraining for language modeling in low resource contexts","year":2023,"lang":"en","type":"article","venue":"","topic":"Topic Modeling","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"ca_institutions":"McGill University; Mila - Quebec Artificial Intelligence Institute","funders":"","keywords":"Baseline (sea); Computer science; Hyperparameter; Grammar; Context (archaeology); Natural language processing; Language model; Artificial intelligence; Resource (disambiguation); Linguistics","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":[0.001552599,0.001331434,0.0008060817,0.0005836171,0.0004867325,0.0009686638,0.001740317,0.001084182,0.005702651],"category_scores_gemma":[0.005285725,0.0006213519,0.001178697,0.0008558611,0.0006096782,0.003272054,0.001567214,0.004848823,0.004613325],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0007510315,"about_ca_system_score_gemma":0.001369745,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.004814623,"about_ca_topic_score_gemma":0.01186531,"domain_scores_codex":[0.999069,0.0004052245,0.00003616014,0.0002810071,0.0001016779,0.000106797],"domain_scores_gemma":[0.9981255,0.001186399,0.00006819006,0.0003279063,0.0002197033,0.0000723872],"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":[0.0006130951,0.0004414192,0.004045215,0.0005848155,0.000273847,0.0006853514,0.0006915895,0.2553065,0.0454731,0.02322477,0.06788471,0.6007755],"study_design_scores_gemma":[0.00004368481,0.0001389591,0.0006799122,0.00003775244,0.00005358698,0.0001577929,0.0001092221,0.9489825,0.01421,0.02535098,0.01019855,0.0000370197],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.06184271,0.001066922,0.9045547,0.001203361,0.0005403841,0.0001657055,0.002190624,0.02346756,0.004968055],"genre_scores_gemma":[0.5532662,0.000732579,0.4209971,0.0009307278,0.0002944894,0.0004297707,0.009697177,0.002481456,0.01117037],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.005702651,"threshold_uncertainty_score":0.01907724,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.05666310754084649,"score_gpt":0.2865182320582126,"score_spread":0.2298551245173661,"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."}}