Extended Data Table 1 All gLMs assessed in this study together with their input sequence specifications, training data, architecture, and figure panels where they are used

From: Nucleotide dependency analysis of genomic language models detects functional elements

  1. * Because many fungal genomes are compact, taking the region 1 kb 5′ of gene starts already covers a significant amount of the genome, including diverse features such as non-coding RNA, coding sequences, regulatory elements, long-terminal repeats and others. ** Several versions trained for different lengths exist.