{"id":"W4387635881","doi":"10.48550/arxiv.2310.08511","title":"HoneyBee: Progressive Instruction Finetuning of Large Language Models for Materials Science","year":2023,"lang":"en","type":"preprint","venue":"arXiv (Cornell University)","topic":"Machine Learning in Materials Science","field":"Materials Science","cited_by":5,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Natural Sciences and Engineering Research Council of Canada; Canadian Institute for Advanced Research","keywords":"Computer science; Benchmark (surveying); Construct (python library); Trustworthiness; Process (computing); Language model; Soundness; Code (set theory); Quality (philosophy); Artificial intelligence; Data science; Programming language","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":[0.002515806,0.0003672478,0.0006014808,0.0005210922,0.0004287625,0.0002470604,0.001892602,0.0002531715,0.0003786281],"category_scores_gemma":[0.0005202821,0.0003895093,0.000126699,0.0007979937,0.0008263658,0.0006919866,0.002141515,0.0002440332,0.00007673105],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0002010893,"about_ca_system_score_gemma":0.0004748982,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0003604949,"about_ca_topic_score_gemma":0.00002155162,"domain_scores_codex":[0.9968393,0.0001994339,0.0004727916,0.00143828,0.0003071322,0.0007431017],"domain_scores_gemma":[0.9972695,0.000125901,0.0009793311,0.001014425,0.0004572922,0.0001535222],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.0001434845,0.00005689742,0.0001665588,0.0005621915,0.00001353721,0.00004574842,0.0009069318,0.3368087,0.6168206,0.04436714,0.00007097748,0.00003729892],"study_design_scores_gemma":[0.001159715,0.0001531928,0.0006759418,0.000581527,0.0001127753,0.00001002435,0.0008389138,0.4188012,0.5418549,0.03500834,0.00003121961,0.0007722451],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9446481,0.00002698694,0.05080967,0.00002977642,0.002483285,0.0008260502,0.0006406682,0.0003547851,0.0001806146],"genre_scores_gemma":[0.9912594,0.00001760701,0.007779118,0.00002036224,0.0001727819,0.00001330198,0.00006900153,0.00005025907,0.0006181549],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.08199255,"threshold_uncertainty_score":0.9998557,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.06362611535866841,"score_gpt":0.2458956631116573,"score_spread":0.1822695477529889,"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."}}