TY  - RPRT
TI  - Reward Granularity in RLVR: Comparing Process and Outcome Reward Structures for Mathematical Reasoning in Small Language Models
AU  - Anagha Radhakrishna Palandye
AU  - Rebecca Glick
AU  - Osheen Kaul
PY  - 2026
UR  - https://arxiv.org/abs/2607.02869
ID  - 2607.02869
ER  -