{"id":"W4389519588","doi":"10.18653/v1/2023.emnlp-main.134","title":"MAGNIFICo: Evaluating the In-Context Learning Ability of Large Language Models to Generalize to Novel Interpretations","year":2023,"lang":"en","type":"article","venue":"","topic":"Topic Modeling","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"ca_institutions":"Canadian Institute for Advanced Research; McGill University","funders":"Natural Sciences and Engineering Research Council of Canada; McGill University","keywords":"Context (archaeology); Computer science; Parsing; Natural language processing; Artificial intelligence; Suite; Natural language; Cognitive science; Linguistics; Cognitive psychology; Psychology; History","routes":{"ca_aff":true,"ca_fund":true,"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":[0.007707028,0.002720045,0.0008145293,0.001483003,0.0007209583,0.001706534,0.00237706,0.002466884,0.003418676],"category_scores_gemma":[0.0310469,0.0005758536,0.001189729,0.0006373351,0.00108343,0.004141639,0.003054385,0.002959578,0.001050169],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001127691,"about_ca_system_score_gemma":0.001327034,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.006670925,"about_ca_topic_score_gemma":0.009575141,"domain_scores_codex":[0.9961782,0.002425693,0.0001970163,0.0006725006,0.0003627754,0.0001638996],"domain_scores_gemma":[0.9723288,0.02321349,0.0006488861,0.00236286,0.0009110801,0.0005349113],"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":[0.003870642,0.003727079,0.0334649,0.003089424,0.001350844,0.0007188453,0.003167339,0.4150163,0.01807149,0.01321307,0.04703166,0.4572784],"study_design_scores_gemma":[0.0003282537,0.001106955,0.005001226,0.0001125875,0.000138525,0.0002275022,0.0005359857,0.962981,0.01001524,0.01171601,0.007747379,0.00008918067],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.771831,0.003742071,0.1680393,0.002045726,0.0005882182,0.001254793,0.006177943,0.02925793,0.01706295],"genre_scores_gemma":[0.8203873,0.0005862106,0.1624375,0.000601658,0.0001693413,0.000728401,0.01064265,0.001324273,0.003122578],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.007707028,"threshold_uncertainty_score":0.04075915,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.08345484632105207,"score_gpt":0.3630739392796591,"score_spread":0.2796190929586071,"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."}}