[kimi25 rl part3] support K25 VL rollout processor and train-time token expansion#1755
Draft
GeLee-Q wants to merge 2 commits intoTHUDM:mainfrom
Draft
[kimi25 rl part3] support K25 VL rollout processor and train-time token expansion#1755GeLee-Q wants to merge 2 commits intoTHUDM:mainfrom
GeLee-Q wants to merge 2 commits intoTHUDM:mainfrom