Under fixed token budget on Circle Packing, deeper per-candidate reasoning beats generating more shallow candidates, and capable models produce evaluation hacks at higher rates.
arXiv preprint arXiv:2410.15639 , year=
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.SE 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Effective Harness Engineering for Algorithm Discovery with Coding Agents
Under fixed token budget on Circle Packing, deeper per-candidate reasoning beats generating more shallow candidates, and capable models produce evaluation hacks at higher rates.