{"id":"W6910380341","doi":"10.48448/51d3-bk86","title":"Better Language Models of Code through Self-Improvement","year":2022,"lang":"en","type":"other","venue":"Underline Science Inc.","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"McGill University","funders":"","keywords":"Automatic summarization; Code (set theory); Language model; Code generation; Source code; Data modeling; Code review","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":[0.001119538,0.001953913,0.0008179032,0.001118981,0.0004702678,0.001491564,0.002089591,0.001150659,0.004567664],"category_scores_gemma":[0.006290695,0.0006956895,0.001513667,0.0007507564,0.0007593423,0.003305077,0.001630477,0.003688799,0.005964073],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001228902,"about_ca_system_score_gemma":0.001961598,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0123556,"about_ca_topic_score_gemma":0.02412365,"domain_scores_codex":[0.9989729,0.0002505456,0.00004598692,0.0004759015,0.000164633,0.00009007717],"domain_scores_gemma":[0.9979255,0.0009653269,0.0001261251,0.0004132961,0.0004777607,0.00009201984],"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":[0.000397515,0.0003986963,0.003946022,0.000519458,0.0002092895,0.0002480359,0.0004076256,0.3304199,0.02485288,0.008364218,0.05575727,0.5744791],"study_design_scores_gemma":[0.00002567299,0.00004189153,0.0002193072,0.00002112991,0.00002044732,0.00003575976,0.00002819288,0.986351,0.005013717,0.004144622,0.004083874,0.00001451497],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.07058045,0.002686968,0.8403334,0.001974001,0.0006180934,0.0002368293,0.004389554,0.06905306,0.01012759],"genre_scores_gemma":[0.5111586,0.0009136986,0.4467762,0.001701028,0.0002297412,0.0005767328,0.02043543,0.004441064,0.01376749],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.0123556,"threshold_uncertainty_score":0.02456737,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.02440972853770474,"score_gpt":0.2977335995499219,"score_spread":0.2733238710122172,"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."}}