Expand description
AIQI implementation from “A Model-Free Universal AI”.
This module implements a model-free universal agent that predicts
discretized H-step returns directly from augmented interaction history.
The implementation follows the phase-indexed periodic augmentation in
“A Model-Free Universal AI”:
for return horizon H and period N >= H, each phase model only inserts
returns at indices i % N == phase.
Structs§
- Aiqi
Agent - AIQI agent with phase-indexed augmented return predictors.
- Aiqi
Config - Configuration parameters for an AIQI agent.