`PPOLearnerMetricsLogger`

python

PPOLearnerMetricsLogger(config: PPOExperimentConfiguration, metrics: MetricsTracker, rollout_buffer: RolloutBuffer) -> None

Logs rollout-level PPO learner metrics and rollout diagnostics.

Initialize the metrics logger.

Parameters:

config (PPOExperimentConfiguration) – PPO experiment configuration.
metrics (MetricsTracker) – Metrics tracker used for output.
rollout_buffer (RolloutBuffer) – Rollout buffer used for diagnostics.

Methods

`log_metrics`

python

log_metrics(actor_steps: int, lr: float, rollout_metrics: PPORolloutMetrics, *, rollouts_collected: int) -> None

Log averaged PPO training metrics for one rollout update.

Parameters:

actor_steps (int) – The current actor step count.
lr (float) – The current learning rate.
rollout_metrics (PPORolloutMetrics) – The accumulated rollout metrics.
rollouts_collected (int) – Total number of collected rollouts.