log,ratio_top_20_variants,epa_normalized_sequence_entropy_linear_forgetting
task1,0.2,0.4
task2,0.4,0.7
task3,NaN,0.4
task4,0.2,NaN