Implement On-Policy Distillation#444

Merged

pan-x-c merged 15 commits intomodelscope:mainfrom

garyzhang99:dev/on-policy-distilation

Dec 19, 2025

Commits on Dec 16, 2025

[WIP]init on-policy distillation
问昊
committed
merge main and fix wrapper type
问昊
committed
assert instead of padding in workflow
问昊
committed

Commits on Dec 17, 2025

change default loss to ppo with clip
问昊
committed
actual run and add on-policy distillation example in GSM8K
问昊
committed

Commits on Dec 18, 2025

Merge main into dev/on-policy-distillation
问昊
committed
change model wrapper for workflows
问昊
committed
make sure all workflow is changed
问昊
committed
change docs too
问昊
committed
fix workflow test
问昊
committed

Commits on Dec 19, 2025