{"id":"W4389519826","doi":"10.18653/v1/2023.findings-emnlp.245","title":"FinePrompt: Unveiling the Role of Finetuned Inductive Bias on Compositional Reasoning in GPT-4","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":"Institute for Information and Communications Technology Promotion; Ministry of Science and ICT, South Korea","keywords":"Computer science; Leverage (statistics); Task (project management); Inductive bias; Inductive reasoning; Question answering; Artificial intelligence; Human–computer interaction; Natural language processing; Multi-task 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":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0004075497,0.00007502431,0.000106258,0.0001343169,0.00006094278,0.00003059515,0.0004610104,0.00003348709,0.0000153506],"category_scores_gemma":[0.00006899006,0.00005372829,0.00003483901,0.0005804533,0.00002351055,0.000126038,0.0002000584,0.0001459168,0.0000347846],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00002591091,"about_ca_system_score_gemma":0.00004398197,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0002370541,"about_ca_topic_score_gemma":0.00003467686,"domain_scores_codex":[0.9990796,0.00006476483,0.0001992482,0.0002443505,0.0002430428,0.0001689422],"domain_scores_gemma":[0.9992736,0.0002777574,0.0000560293,0.0003291905,0.00004079358,0.00002263486],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.00002014751,0.0001040529,0.01276059,0.00001081892,0.00002555229,0.00001593275,0.004731389,0.08745851,0.01335935,0.8363044,0.0002195112,0.04498979],"study_design_scores_gemma":[0.0001964415,0.00004002274,0.0189411,0.00006098891,0.000001145066,0.000002720382,0.0002333672,0.9474357,0.0112643,0.02155437,0.0001801687,0.00008962664],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9530253,0.00002723645,0.03645534,0.001791008,0.0001292111,0.0001428564,0.000001800288,0.0001250413,0.008302229],"genre_scores_gemma":[0.9824905,0.000001947493,0.01714726,0.0001446044,0.00004218799,0.00001051979,0.000002651294,0.000004268817,0.0001560668],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.8599772,"threshold_uncertainty_score":0.2190976,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.04108843627987746,"score_gpt":0.26247931949821,"score_spread":0.2213908832183325,"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."}}