Reduce Ray overhead in SFT with Ray Compiled Graph

**Is your feature request related to a problem? Please describe.**
Reduce the overhead of Ray in SFT with Ray Compiled Graph

**Describe the solution you'd like**
A clear and concise description of what you want to happen.

**Describe alternatives you've considered**
A clear and concise description of any alternative solutions or features you've considered.

**Additional context**
Add any other context or screenshots about the feature request here.