SCORE (Systematic COnsistency and Robustness Evaluation)

NVIDIA's non-adversarial robustness suite: re-run MMLU-Pro, AGIEval MCQ, and MATH under prompt, choice-order, and seed changes, reporting accuracy plus consistency rate.

Also known as: SCORE, Systematic COnsistency and Robustness Evaluation, score_robustness

unassessed

This page is a discovery lead. Nobody has yet assessed it against the catalogue contract, so it carries no disposition. Absence of evidence here is not evidence of staleness.
Categorycomposite
Subcategorynon-adversarial prompt, choice-order, and seed robustness on MMLU-Pro, AGIEval MCQ, and MATH
Page statusactive
Metricaccuracy and consistency rate (CR)
Directionhigher_is_better
Dataset size19372
Dataset licenceApache-2.0
PublisherNVIDIA

What it measures

SCORE does not introduce new questions. It re-asks public items from MMLU-Pro, seven English AGIEval multiple-choice exams, and Hendrycks MATH while changing the prompt, the option order, or the sampling seed. Prompt robustness uses ten semantically similar templates. Choice-order robustness swaps the gold option through every listed slot (MMLU-Pro letters A–J; AGIEval only). Non-greedy robustness holds the prompt fixed and samples at temperature 0.7 with seeds 1–5. The skill is whether an instruct model keeps the same answer under those non-adversarial edits, not whether it withstands PromptBench-style attacks.

Task format

Zero-shot generative evaluation of instruct models. MMLU-Pro and AGIEval items are multiple-choice completions parsed for an answer letter. MATH items are free-form solutions. The paper generates up to 1024 tokens per call. lm-eval recommends --apply_chat_template. Non-greedy runs need five seeded jobs plus non_greedy_summarizer.py.

Models reporting this benchmark

No model card in ModelSpec reports this benchmark yet.

Data

This page as JSON · Edit on GitHub