{"id":"W6891812606","doi":"10.48448/6y6k-7w61","title":"TofuEval: Evaluating Hallucinations of LLMs on Topic-Focused Dialogue Summarization","year":2024,"lang":"en","type":"other","venue":"Underline Science Inc.","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"Kootenay Association for Science & Technology","funders":"","keywords":"Automatic summarization; Consistency (knowledge bases); Benchmark (surveying); Hallucinating; Binary number","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":[0.005928008,0.001388995,0.0007713998,0.002221187,0.0005146639,0.001588694,0.001047942,0.001223055,0.002658081],"category_scores_gemma":[0.03628743,0.0001889269,0.0004975923,0.001067686,0.0005285041,0.002058745,0.001806007,0.0009361294,0.001251666],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.000680953,"about_ca_system_score_gemma":0.0007222329,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002955395,"about_ca_topic_score_gemma":0.004436921,"domain_scores_codex":[0.9941602,0.003252074,0.0005127849,0.0007743438,0.001132105,0.0001684708],"domain_scores_gemma":[0.9779767,0.0141719,0.001379667,0.002132079,0.003573374,0.0007662267],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.005140078,0.001112118,0.02765822,0.005662987,0.001215142,0.0006980433,0.004658819,0.09995571,0.04923712,0.002922168,0.0609388,0.7408009],"study_design_scores_gemma":[0.0007168676,0.005889838,0.04108792,0.0004792787,0.000550104,0.000697705,0.003163926,0.8218941,0.08068716,0.006493927,0.03804381,0.0002953171],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.7623178,0.005677236,0.1632103,0.001139372,0.0007081016,0.001074346,0.01529636,0.03674221,0.01383419],"genre_scores_gemma":[0.8624178,0.0005269112,0.1031551,0.0002872944,0.0001660979,0.0004782387,0.0281301,0.0009288173,0.003909565],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.005928008,"threshold_uncertainty_score":0.03135067,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.07145306131015673,"score_gpt":0.3807486708137582,"score_spread":0.3092956095036015,"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."}}