Genomic models hard to compare due to fragmented benchmarks

Arxiv CS.CL (NLP/language models)

A new arXiv paper (GENEB) identifies fragmented benchmarks and incompatible evaluation protocols hindering comparison of genomic foundation models. The authors call for standardized evaluation to enable meaningful progress assessments.

modelsbenchmarks

See 2 more sources

Position: Genomic Model Research Must Move Beyond Anecdotal Evaluation of Interpretability Methods2 days agoShasha Zhou, Mingyu Huang, Ke Li

GENEB: Why Genomic Models Are Hard to Compare7 days agoDaria Ledneva, Mikhail Nuridinov, Denis Kuznetsov

Arxiv CS.CL (NLP/language models)·Chahat BaranwalAadtya BaranwalLakshya Nitin Tandon

7 days ago