notesum.ai

Published at November 4

Training Compute-Optimal Protein Language Models

cs.LG

cs.AI

q-bio.QM

Released Date: November 4, 2024

Authors: Xingyi Cheng¹, Bo Chen², Pan Li¹, Jing Gong¹, Jie Tang², Le Song³

Aff.: ¹BioMap Research; ²Tsinghua University; ³BioMap Research and MBZUAI

Arxiv: http://arxiv.org/abs/2411.02142v1

Refer to caption

Datasets	Prot. Seq.	Tokens (AAs)	Samp. Prop.
Uniref50/S	54M	15.2B	8.5%
Uniref90/50	102M	37.8B	19.5%
ColabFoldDB_c	208M	37.7B	19.5%
ColabFoldDB_m	575M	103B	52.5%
Total	939M	194B	-