Paraphrasing benchmark questions keeps LLM rankings stable but reduces their accuracy, suggesting static benchmarks overestimate model robustness.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
On Robustness and Reliability of Benchmark-Based Evaluation of LLMs
Paraphrasing benchmark questions keeps LLM rankings stable but reduces their accuracy, suggesting static benchmarks overestimate model robustness.