VitalDB - AnesLLM: a dataset and benchmark based on real intraoperative anaesthetic decisions 1.0

File: <base>/reports/output_consistency.csv (497 bytes)
scopegroupn_windowstop1_accuracytop2_recallresult1_equals_auxratio_out_of_rangeratio_sum_gt_1invalid_real_vocabinvalid_result1_vocabinvalid_aux_vocab
splitoverall848130.5580.80430.000000
splitdev4230.49410.73760.000000
splittrain809330.55780.80430.000000
splittest34570.56840.81310.000000
difficultyeasy173590.6150.86590.000000
difficultymedium448520.56460.80940.000000
difficultyhard226020.5010.74690.000000