{"id":"W4385570165","doi":"10.18653/v1/2023.findings-acl.206","title":"Two Examples are Better than One: Context Regularization for Gradient-based Prompt Tuning","year":2023,"lang":"en","type":"article","venue":"","topic":"Topic Modeling","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"ca_institutions":"Kootenay Association for Science & Technology","funders":"","keywords":"Computer science; Regularization (linguistics); Context (archaeology); Machine learning; Artificial intelligence; Language model","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":[0.002035459,0.00163406,0.001084751,0.000461147,0.0007260958,0.001206947,0.001776056,0.002316321,0.002805228],"category_scores_gemma":[0.01038456,0.0005533279,0.0007388791,0.0003674699,0.001182551,0.003097065,0.002193294,0.004559337,0.001264473],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0006179099,"about_ca_system_score_gemma":0.001185648,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002359138,"about_ca_topic_score_gemma":0.004834129,"domain_scores_codex":[0.9983524,0.0006052065,0.00007892097,0.0005889613,0.0002491969,0.0001253836],"domain_scores_gemma":[0.9977543,0.001076961,0.0001162131,0.0005743411,0.0002915001,0.0001866938],"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":[0.001980816,0.001026288,0.005114145,0.0006154607,0.0002138108,0.0003770382,0.0009566912,0.2266715,0.05600932,0.02569846,0.02807892,0.6532576],"study_design_scores_gemma":[0.000163513,0.0003548545,0.0007387238,0.00004909494,0.00004197543,0.0001714986,0.00008388406,0.9590073,0.01477374,0.0186062,0.005936863,0.00007227005],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.0718805,0.002018081,0.9006264,0.001558077,0.0004281762,0.0001558483,0.0001944865,0.01890469,0.004233788],"genre_scores_gemma":[0.5547197,0.0003174553,0.4385065,0.001050549,0.0001353741,0.0001965418,0.0004840453,0.00102694,0.003562816],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.002805228,"threshold_uncertainty_score":0.01076466,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.06752832518457182,"score_gpt":0.2678553627134493,"score_spread":0.2003270375288775,"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."}}