"Collectively these results show that scaling from 600M to 7B parameters raises mean per-nucleotide accuracy from 25% to 83%" Scaling laws applying to biological foundation models is the best thing I've heard all day. This team seems cracked & the paper definitely worth a read.
Prima Mente
Prima Mente17.7. klo 03.13
1/ Today we announce Pleiades, a series of epigenetic foundation models (90M→7B params) trained on 1.9T tokens of human methylation & genomic data. Pleiades accurately models epigenetics for genomic track prediction, generation & neurodegenerative disease detection from cfDNA, outperforming previous pure DNA baselines.
10,77K