Multi-Reward RL, Part 3: GDPO + CISPO + REPO-R at 27B, and the Advantage Floor That Stopped Learning

Part 1 explained how PPO, GRPO, DAPO and GDPO turn several rewards into one learning signal. Part 2 benchmarked the trainers on a 14B model for 50 steps and found a trap: CISPO won the training curves and lost on held-out tasks. This part asks what happens when the same ideas get a bigger model and…

Read the full story at DEV Community — Machine Learning ↗

Timeline · 1 report

1. 2026-10-09 17:51 · DEV Community — Machine Learning
   Multi-Reward RL, Part 3: GDPO + CISPO + REPO-R at 27B, and the Advantage Floor That Stopped Learning