mu_zero/example2 library

Functions

argMax(List<double> list) → int

generateMuZeroPure(MuZeroGreedyAgent agent, List<int> prompt, int maxLength, Map<int, String> itos) → void

Pure MuZero Generation: Only uses Representation ONCE, then relies on Dynamics

main() → void

trainMuZero(MuZeroGreedyAgent agent, List<int> targetSequence, int epochs) → void

Training with Unrolled Dynamics to prevent "Latent Collapse"