I know that your training data set uses the bc file corresponding to each .c in Fuchsia. Can I merge all the bc files into a total bc file for training? Will this affect the Monte Carlo methods and thus the final effect?
4.1.2 Policy Gradient Algorithm
Here Eis an expectation over the policy 𝜋𝜃 being applied to an inlining pass. In practice, this expectation is approximated with Monte Carlo methods — with 𝑛 trajectories collected from compiling with policy
I know that your training data set uses the bc file corresponding to each .c in Fuchsia. Can I merge all the bc files into a total bc file for training? Will this affect the Monte Carlo methods and thus the final effect?
4.1.2 Policy Gradient Algorithm Here Eis an expectation over the policy 𝜋𝜃 being applied to an inlining pass. In practice, this expectation is approximated with Monte Carlo methods — with 𝑛 trajectories collected from compiling with policy