Olmo-eval
AssessTools
An open evaluation workbench for running and comparing model benchmarks.
Why it's here
Placed in Assess: 1 article(s) of evidence from 1 source(s), led by framework updates, with 1 in the last 30 days. Confidence 24%. Low accumulated evidence, so it defaults conservatively pending more signal.
Evidence (1)
- 7Hugging Face Blog·10/20/2026framework_updateFairStress benchmarks fairness beyond final LLM answers
Hugging Face Blog introduced FairStress, a benchmark for measuring demographic fairness in large language models beyond whether they pick the same candidate in a binary decision. The benchmark tests how resistant models are to persuasive adversarial prefixes and examines differences across domains such as hiring, lending, healthcare, education, and housing.