{"id":"W4402670735","doi":"10.18653/v1/2024.findings-acl.566","title":"Bootstrapping LLM-based Task-Oriented Dialogue Agents via Self-Talk","year":2024,"lang":"en","type":"article","venue":"","topic":"Speech and dialogue systems","field":"Computer Science","cited_by":9,"is_retracted":false,"has_abstract":true,"ca_institutions":"Artificial Intelligence in Medicine (Canada)","funders":"","keywords":"Computer science; Bootstrapping (finance); Task (project management); Human–computer interaction; Engineering; Systems engineering; Business","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":[0.004319925,0.001354011,0.001244404,0.0006349694,0.0005809247,0.001164168,0.002016478,0.001282625,0.002243536],"category_scores_gemma":[0.01688474,0.0006993219,0.0007691944,0.000250889,0.0009823106,0.001689073,0.003272497,0.001884677,0.001420229],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0007185884,"about_ca_system_score_gemma":0.0008370389,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001529,"about_ca_topic_score_gemma":0.002020708,"domain_scores_codex":[0.9974148,0.00142679,0.0001184972,0.0005744289,0.0002883867,0.0001770915],"domain_scores_gemma":[0.9923016,0.005100913,0.0005165645,0.001023802,0.0006676202,0.0003895149],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.001397405,0.001204539,0.008284176,0.0003967338,0.0002546558,0.0003921687,0.002365496,0.5742165,0.03684546,0.007242706,0.004327842,0.3630723],"study_design_scores_gemma":[0.00003324257,0.0001130583,0.0002931755,0.00001085526,0.000011213,0.00002699812,0.00007377804,0.9915659,0.004251292,0.002976599,0.0006306025,0.00001321707],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.1688856,0.000239235,0.8207253,0.00031821,0.00009345163,0.0004455423,0.0001593477,0.006681415,0.002451857],"genre_scores_gemma":[0.824977,0.00004400254,0.1716907,0.0002078346,0.00003527048,0.0005683696,0.0003555218,0.0002692926,0.001851952],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.004319925,"threshold_uncertainty_score":0.02284622,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.01466826452131366,"score_gpt":0.2415134788242593,"score_spread":0.2268452143029456,"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."}}