{"id":"W7115903308","doi":"10.3929/ethz-c-000790212","title":"Gumbel Counterfactual Generation From Language Models","year":2025,"lang":"en","type":"other","venue":"Repository for Publications and Research Data (ETH Zurich)","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Institute for Catastrophic Loss Reduction","keywords":"Counterfactual conditional; Counterfactual thinking; Gumbel distribution; Hindsight bias; Causal model; Representation (politics); Language model; Causality (physics)","routes":{"ca_aff":false,"ca_fund":true,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":true},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01624917,0.001423545,0.001234569,0.001557312,0.0008977207,0.003168906,0.002330173,0.001686821,0.01140706],"category_scores_gemma":[0.07238072,0.0009986669,0.002269187,0.001284108,0.002828351,0.004570895,0.003385859,0.003936362,0.001352676],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002383102,"about_ca_system_score_gemma":0.001897958,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00255299,"about_ca_topic_score_gemma":0.003636249,"domain_scores_codex":[0.9892616,0.007436586,0.0003282957,0.001338644,0.001318916,0.0003159547],"domain_scores_gemma":[0.9369973,0.05423816,0.00192742,0.005257672,0.001218486,0.0003609176],"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":[0.0005418388,0.0001748069,0.002494078,0.000242619,0.000149508,0.0002429662,0.0007410651,0.3135537,0.001418318,0.5897757,0.004768809,0.08589663],"study_design_scores_gemma":[0.00007156978,0.00003227989,0.0001456195,0.00003114296,0.00001887666,0.00004212067,0.0000279607,0.7180935,0.001088031,0.2787188,0.001708345,0.00002173855],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.01743818,0.0001486876,0.9774075,0.0005500665,0.00004198248,0.0002036462,0.0003639646,0.0008798873,0.002966137],"genre_scores_gemma":[0.4269125,0.0002505096,0.5651229,0.0004638897,0.00006901193,0.001196337,0.001177717,0.000554124,0.004253048],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.01624917,"threshold_uncertainty_score":0.08593488,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.237836221524231,"score_gpt":0.4374304829550625,"score_spread":0.1995942614308315,"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."}}