Disjoint SFT and GRPO data for autoformalization yields up to 10.4pp semantic accuracy gains over full overlap, which renders the GRPO stage redundant.
PutnamBench : Evaluating neural theorem-provers on the P utnam mathematical competition
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
SFT-GRPO Data Overlap as a Post-Training Hyperparameter for Autoformalization
Disjoint SFT and GRPO data for autoformalization yields up to 10.4pp semantic accuracy gains over full overlap, which renders the GRPO stage redundant.