TRACKMANIARL / SYSTEM GUIDE / 01 How off-policy training runs Actors collect experience. Durable ingestion feeds a learner that publishes the next policy. 01 Collect 02 Persist 03 Learn activate nextepisode Actor + game adapter Episode policyFeatures + rollouts Authenticated ingest Commit before ACKSQLite WAL Replay + learner Sample + updateAdvance WAL frontier Policy snapshot Portable model tensors Checkpoint + logs Replay + optimizerRNG + WAL frontier BEFORE COLLECTIONRunSpec 2.0 resolves components, seeds and contracts into an immutable run manifest. DURABILITYThe actor spools unaccepted rollouts for retry.WAL pruning waits until a durable checkpointcovers the applied frontier. SCOPEThis is the off-policy path. PPO uses a separatelocal, on-policy lifecycle.