Geneformer Eval
Evaluation of Geneformer (30M transformer pretrained on single-cell corpus) on zero-shot & few-shot downstream tasks.
Composite score: 77.0
Rubric (1–5 per criterion)
rigor
4/5
coverage
3/5
maintenance
4/5
adoption
4/5
quality
4/5
accessibility
4/5
industry_relevance
4/5
Metadata
Stages
Virtual Cell
Modalities
cross-modality
Task types
cell-type-annotationgene-dosage
License
MIT
First release
2023-06
Last updated
2024-11
Flags
self_referential
Size & scope
- cells: 30000000
- tasks: 7
Primary paper
Title
Transfer learning enables predictions in network biology
Authors
Theodoris CV, Xiao L, Chopra A, et al.
Year
2023
DOI / arXiv
Citations
550
Links
- Official site: https://huggingface.co/ctheodoris/Geneformer
- GitHub: https://huggingface.co/ctheodoris/Geneformer
- Leaderboard: N/A
Hosted by (initiatives)
- not hosted by any tracked initiative
Experts (primary authors / maintainers)
Groups (host labs / companies / consortia)
Related benchmarks
Notes (honest caveats)
Author-led eval; still widely re-run on OpenProblems tasks.