{"id":"W4383047243","doi":"10.48550/arxiv.2306.17693","title":"Thompson sampling for improved exploration in GFlowNets","year":2023,"lang":"en","type":"preprint","venue":"arXiv (Cornell University)","topic":"Machine Learning and Algorithms","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Samsung; Genentech; Canadian Institute for Advanced Research","keywords":"Thompson sampling; Computer science; Flexibility (engineering); Sampling (signal processing); Mathematical optimization; Importance sampling; Convergence (economics); Inference; Machine learning; Generative grammar; Artificial intelligence; Posterior probability; Bayesian probability; Algorithm; Mathematics; Monte Carlo method; Statistics","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":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow"],"consensus_categories":[],"category_scores_codex":[0.0004283187,0.0002199387,0.0002573696,0.0003399261,0.0001176968,0.0001444944,0.001037945,0.0002034062,0.000003661062],"category_scores_gemma":[0.00009036082,0.0002557377,0.0001422846,0.0005105833,0.00002193726,0.0003560947,0.0009391514,0.0004994922,0.00003510954],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001247812,"about_ca_system_score_gemma":0.0001071164,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0002331727,"about_ca_topic_score_gemma":0.000132245,"domain_scores_codex":[0.998392,0.00008811944,0.0001841844,0.0009659017,0.00004887786,0.0003209062],"domain_scores_gemma":[0.9988097,0.0001830117,0.0001721031,0.000674932,0.0000821642,0.00007810505],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.00002741148,0.00006066052,0.0006803504,0.0001168735,0.00003200466,0.00004347977,0.0006016116,0.941051,0.00009371933,0.04378394,0.0001552998,0.01335365],"study_design_scores_gemma":[0.0004032025,0.00004154981,0.0006466499,0.00007917177,0.0000102484,4.565516e-7,0.00004656966,0.9310452,0.00003428188,0.06670003,0.0007229524,0.0002696629],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.02267601,0.00001604742,0.9745662,0.0006285207,0.001085466,0.000400539,0.00001176545,0.0004324339,0.0001830458],"genre_scores_gemma":[0.9671586,0.00008712929,0.02940115,0.0000689527,0.0002106686,0.000007739488,0.00007632012,0.00003188589,0.00295751],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.945165,"threshold_uncertainty_score":0.9999895,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1928492168762235,"score_gpt":0.2401838458746142,"score_spread":0.04733462899839069,"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."}}