[Draft] Add On-Policy Distillation (OPSD) Trainer in DeepSpeed#8027

Open

PKUWZP wants to merge 4 commits into

deepspeedai:masterfrom

PKUWZP:zhipwang_opd_pr

Commits on May 26, 2026

Add OPSD example: config, divergence losses, utils + tests
Zhipeng Wang
committed
Add OPSD frozen teacher with CPU logit cache + tests
Zhipeng Wang
committed
Add OPSD trainer, hybrid-engine rollout, and end-to-end entry point
Zhipeng Wang
committed
Add OPSD vLLM rollout scaffold, Qwen2/Qwen3 weight bridges, and README
Zhipeng Wang
committed