CellFM
CellFM is an 800-million-parameter single-cell model trained on 100 million human cells.
CellFM (2024, Chinese consortium) reported gains on annotation and perturbation tasks; among the largest single-cell models by parameters.
How it works
CellFM is a transformer with efficient attention over gene tokens.
- Scale
- Limited independent evaluation
Latest papers
topQuery for this technology: (TITLE:"CellFM" OR ABSTRACT:"CellFM") AND (cancer OR tumor OR tumour OR oncology OR carcinoma OR lymphoma OR leukemia OR leukaemia OR myeloma OR sarcoma OR melanoma OR glioma). Results are unfiltered search hits about CellFM, not a curated reading list.
Pages like this
not linked directly; found by shared links- TechnologyCell2Sentence / C2S-Scale (Yale, Google)
Shares the tags foundation-model, virtual-cell.
- TechnologyGeneformer
Shares the tags foundation-model, virtual-cell.
- TechnologyGenePT
Shares the tags foundation-model, virtual-cell.
- TechnologyNicheformer (spatial single-cell)
Shares the tags foundation-model, virtual-cell.
- TechnologyscFoundation (BioMap)
Shares the tags foundation-model, virtual-cell.
- TechnologyscGPT
Shares the tags foundation-model, virtual-cell.
- TechnologyState (Arc Institute perturbation model)
Shares the tags foundation-model, virtual-cell.
- TechnologyTranscriptFormer and rBio (CZI virtual cell models)
Shares the tags foundation-model, virtual-cell.