RL Command-Line Options

TorchLean's runnable RL examples (NN/Examples/Models/RL/*) share one CLI shape:

• --updates <n>: how many update iterations to run,
• --eval-every <n>: evaluate every n updates,
• --eval-episodes <n>: number of evaluation episodes per checkpoint,
• --eval-max-steps <n>: max steps per evaluation episode,
• --log <path|off|none|false>: where to write the widget-friendly TrainLog JSON.

This module centralizes that parsing so we don't duplicate the same flag boilerplate across CartPole/Pong/GridWorld examples.

structure TorchLean.rl.cli.PPOOptions : Type

Parsed PPO command options shared by multiple runnable examples.

• updateCount : Nat
• evaluationInterval : Nat
• evaluationEpisodes : Nat
• maximumEvaluationSteps : Nat
• logDestination : Runtime.Training.LogDestination

def TorchLean.rl.cli.instReprPPOOptions.repr : PPOOptions → Nat → Std.Format

@[instance_reducible]

instance TorchLean.rl.cli.instReprPPOOptions : Repr PPOOptions

def TorchLean.rl.cli.PPOOptions.usage (exeName : String) (artifactOptions : Array String := #[]) : String

Help text for PPO commands, with optional environment-specific artifact flags.

def TorchLean.rl.cli.PPOOptions.parse (exeName : String) (arguments : List String) (defaultLogPath : System.FilePath) (defaultUpdateCount defaultEvaluationInterval defaultEvaluationEpisodes defaultMaximumEvaluationSteps : Nat) : Except String (PPOOptions × List String)

Parse shared PPO command options.

Notes:

• --log off|none|false selects LogDestination.disabled.
• We treat 0 as invalid for the update/eval counts because a “no-op” run usually indicates a CLI mistake.