A new foundation model called GREmLN from a Columbia and Chan Zuckerberg Biohub team, delivers superior cell-type classification with only 10.3 million parameters, outpacing rivals like the 100-million-parameter scFoundation. Released July 9 on bioRxiv, it taps gene regulatory networks to achieve a 0.929 macro F1 score on immune cell data. “Instead of using large language… The post Columbia-CZ team develops 10.3M parameter model that outperforms 100M parameter rivals on cell type classifi...
