TY  - RPRT
TI  - Improving HPC Code Generation Capability of LLMs via Online Reinforcement Learning with Real-Machine Benchmark Rewards
AU  - Ryo Mikasa
AU  - Shun-ichiro Hayashi
AU  - Daichi Mukunoki
AU  - Tetsuya Hoshino
AU  - Takahiro Katagiri
PY  - 2026
UR  - https://arxiv.org/abs/2602.12049
ID  - 2602.12049
ER  -