More LLM-generated unit tests improve best-of-N code selection, and the lightweight CodeRM-8B generator plus dynamic scaling delivers these gains at lower cost.
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
-
Dynamic Scaling of Unit Tests for Code Reward Modeling
More LLM-generated unit tests improve best-of-N code selection, and the lightweight CodeRM-8B generator plus dynamic scaling delivers these gains at lower cost.