agedi.cli.inspect

Functions

_extract_checkpoint_metrics(→ dict)	Extract training metrics from a Lightning checkpoint state dict.
inspect(→ None)	Inspect a trained AGeDi model directory.

Module Contents

agedi.cli.inspect._extract_checkpoint_metrics(ckpt: dict) → dict

Extract training metrics from a Lightning checkpoint state dict.

Returns a dict with optional keys: epoch, global_step, val_loss, train_loss, best_val_loss, best_checkpoint.

agedi.cli.inspect.inspect(path: str) → None

Inspect a trained AGeDi model directory.

Reads and prints the hyperparameters stored in hparams.yaml inside the given model directory, and reports available training metrics (epochs, loss, etc.) from the latest checkpoint.