Fig. 6From: DeepGRP: engineering a software tool for predicting genomic repetitive elements using Recurrent Neural Networks with attentionPrediction time of RepeatMasker, dna-brnn, DeepGRP and HMMER as a function of the sequence length in Mbp. For dna-brnn and DeepGRP five independently trained models and chromosomes from hg19 were used. dna-brnn was trained with the same hyperparameter as DeepGRP for 50 epochs with varying seed. The running time of dna-brnn and of DeepGRP shows almost no varianceBack to article page