ATTNPO guides process-supervised RL with intrinsic attention signals to shorten reasoning traces while raising accuracy on nine benchmarks.
This dataset is sourced from the American Mathematics Competitions (AMC) system in 2023 and in- cludes 40 problems with mixed and hybrid question formats
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
ATTNPO: Attention-Guided Process Supervision for Efficient Reasoning
ATTNPO guides process-supervised RL with intrinsic attention signals to shorten reasoning traces while raising accuracy on nine benchmarks.