Bigger Isn't Always Better for Single-Cell Foundation Models
burny_tech · x · 2026-07-16
This study challenges the intuition that "bigger data is always better" for single-cell foundation models.
- The authors pre-trained 400 models across 5 architectures on a 2220 万 cell corpus, conducting 6400 evaluations.
- They systematically varied data volume, data diversity, and the weighting of rare cell types to see if scaling yields continuous gains.
- Results show that performance gains saturate as models and data scale up. Simply expanding the atlas doesn't teach the model much more on certain tasks.
The takeaway: the issue might not just be "insufficient data," but rather flawed pre-training objectives. New, genuinely scalable objectives are needed.
More from Research
- Chinese AI labs are now treating distillation obfuscation as the top research topic — pmddomingos · 2026-07-22
- Structural ensembles beat single predictions in TCR:pMHC generalization study — quaidmorris · 2026-07-22
- RSS launches under OMSF to push structural biology data modeling at scale — MoAlQuraishi · 2026-07-22
- enFoldX tops 8 neoantigen scans and an unseen-peptide benchmark — quaidmorris · 2026-07-22
- enFoldX reaches AUC 0.82 on human VDJdb and transfers to mouse at 0.76 — quaidmorris · 2026-07-22
- enFoldX gains accuracy as AF3 ensemble disagreement rises for non-binders — quaidmorris · 2026-07-22