Evolution or Illusion? Rethinking Evaluation in LLM Evolutionary Search
A methodological paper questioning whether reported gains in LLM-driven evolutionary search are genuine or artifacts of flawed evaluation.
A methodological paper questioning whether reported gains in LLM-driven evolutionary search are genuine or artifacts of flawed evaluation.
Auto-scored: A peer-reviewed-style arXiv paper offering a partial but real methodological correction to evaluation practices in LLM evolutionary search, with no direct safety escalation.