Block a user
LLM supervisor agent design
LLM supervisor agent design
Out of scope for this map. The training runner is a standalone CLI tool; the LLM supervisor is a separate effort that simply invokes, monitors, and restarts this tool externally.
LLM supervisor agent design
Persist Adam optimizer state and round counter across restarts
Build the training runner tool
Stall detection: hardcoded heuristics vs LLM judgment
Structured output format for LLM consumption
Define the tunable parameter set
Hyperparam delivery: env vars vs compile-time constants
[Map] Autonomous LLM-supervised PPO_Bot training pipeline
Action decoding rewrite for command abstraction
Network and state vector dimension changes
Goto and AimTo controller functions
Action decoding rewrite for command abstraction
Implemented in commit c713585 on branch research/goto-controller.
Network and state vector dimension changes
Implemented in commit c713585 on branch research/goto-controller.
Goto and AimTo controller functions
Implemented in commit c713585 on branch research/goto-controller.
PPO_Bot wiring and integration test
Action decoding rewrite for command abstraction
Network and state vector dimension changes
Goto and AimTo controller functions