Do you support agentic RL?

[ccp123456789]

Do you support agentic RL? That is, autonomously deciding to call and combine tools based on available tools to complete tasks; I see that the examples using LangChain are all workflows.

[ultmaster]

I don't see what's the difference between dynamic workflows and static workflows in case of RL training?

[ccp123456789]

I don't see what's the difference between dynamic workflows and static workflows in case of RL training?

Static workflows execute tool calls according to a fixed process, while dynamic workflows refer to providing some tools and letting the model make autonomous decisions about when to call which tools and whether to call tools at all

[ultmaster]

They are the same for our framework.

We can write an example but we don't have enough manpowers yet.