Janus co-evolves LLM-generated proxy evaluators with target programs, reaching the same or better final performance with about 59% fewer real evaluations across five design benchmarks.
Symbolic regression with a learned concept library,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
background 1
citation-polarity summary
fields
cs.AI 1years
2026 1verdicts
CONDITIONAL 1roles
background 1polarities
unclear 1representative citing papers
citing papers explorer
-
Janus: An Algorithm-Evaluator Co-Evolution Framework for LLM-Driven Discovery under Expensive Evaluation Budgets
Janus co-evolves LLM-generated proxy evaluators with target programs, reaching the same or better final performance with about 59% fewer real evaluations across five design benchmarks.