TorchLean Model-Zoo Facade

Flag and logging adapters used by runnable examples.

The ordinary training API is Trainer.new plus trainer.train. This module keeps repository command plumbing in the ModelZoo namespace instead of mixing it into the trainer core.

def TorchLean.ModelZoo.parseLoggedTrainFlags (exeName : String) (args : List String) (defaultLogPath : System.FilePath) (defaultSteps : ℕ := 1) (allowZeroSteps : Bool := false) : Except String (LoggedTrainFlags × List String)

Parse shared logged-training flags for model-zoo commands.

def TorchLean.ModelZoo.parseTrainFlags (exeName : String) (args : List String) (defaultLogPath : System.FilePath) (defaultSteps : ℕ := 1) (defaultLr : Float := 1e-3) (allowZeroSteps : Bool := false) : Except String (TrainFlags × List String)

Parse shared optimizer/training flags for model-zoo commands.

def TorchLean.ModelZoo.parseCsvTrainFlags (exeName : String) (args : List String) (defaultCsv defaultLogPath : System.FilePath) (defaultSteps : ℕ := 1) (defaultLr : Float := 1e-3) : Except String (CsvTrainFlags × List String)

Parse shared CSV path plus optimizer/training flags for model-zoo commands.

def TorchLean.ModelZoo.parseSeededTrainFlags (exeName : String) (args : List String) (defaultLogPath : System.FilePath) (defaultSteps : ℕ := 1) (defaultLr : Float := 1e-3) (allowZeroSteps : Bool := false) : Except String (SeededTrainFlags × List String)

Parse shared seeded optimizer/training flags for model-zoo commands.

def TorchLean.ModelZoo.LoggedTrainFlags.trainOptions (flags : LoggedTrainFlags) (enableLog : Bool := true) (logEvery : ℕ := 0) (title : String := "Training") (notes : Array String := #[]) : Trainer.TrainOptions

Build public trainer options from logged model-zoo flags.

def TorchLean.ModelZoo.LoggedTrainFlags.trainOptionsWhenLogRequested (flags : LoggedTrainFlags) (args : List String) (logEvery : ℕ := 0) (title : String := "Training") (notes : Array String := #[]) : Trainer.TrainOptions

Build training options that write a log only when the original CLI arguments include --log.

Tutorial commands use this when they want clean terminal output by default but still support the standard TrainLog artifact without repeating the flag check in each file.

def TorchLean.ModelZoo.LoggedTrainFlags.writeBeforeAfterLossLog (flags : LoggedTrainFlags) (title : String) (loss0 loss1 : Float) (notes : Array String := #[]) : IO Unit

Write a before/after loss artifact from logged model-zoo flags.

def TorchLean.ModelZoo.TrainFlags.trainOptions (flags : TrainFlags) (logEvery : ℕ := 0) (title : String := "Training") (notes : Array String := #[]) : Trainer.TrainOptions

Build public trainer options from optimizer/training model-zoo flags.

def TorchLean.ModelZoo.TrainFlags.writeBeforeAfterLossLog (flags : TrainFlags) (title : String) (loss0 loss1 : Float) (notes : Array String := #[]) : IO Unit

Write a before/after loss artifact from optimizer/training model-zoo flags.