Admit local LeRobot v3 datasets as training input (S-48)

src/tether/finetune/cli.py

@@ -8,6 +8,7 @@
 """
 from __future__ import annotations
 
+import json
 import logging
 from pathlib import Path
 
@@ -53,12 +54,14 @@ def finetune_command(
     ),
     dataset_root: str = typer.Option(
         "", "--dataset-root",
-        help="Actual EXPORT/dataset directory from a qualified Studio SmolVLA export.",
+        help="Local LeRobot v3 dataset directory; validated offline before training.",
     ),
     dataset_manifest_sha256: str = typer.Option(
         "", "--dataset-manifest-sha256",
         help="Canonical sibling export-receipt.json SHA-256 (excludes its sha256 field).",
     ),
+    dataset_content_sha256: str = typer.Option("", "--dataset-content-sha256", help="Expected local content manifest SHA-256."),
+    dataset_features: str = typer.Option("{}", "--dataset-features", help="Recipe expected LeRobot feature shapes as JSON."),
     output: str = typer.Option(
         ...,
         "--output",
@@ -143,13 +146,22 @@ def finetune_command(
     from tether.finetune.config import FinetuneConfig
     from tether.finetune.run import run_finetune
 
+    try:
+        expected_features = json.loads(dataset_features)
+    except ValueError as exc:
+        raise typer.BadParameter("dataset_features must be a JSON feature object", param_hint="--dataset-features") from exc
+    if not isinstance(expected_features, dict):
+        raise typer.BadParameter("dataset_features must be a JSON feature object", param_hint="--dataset-features")
+
     cfg = FinetuneConfig(
         base=base,
         base_revision=base_revision or None,
         dataset=dataset,
         dataset_revision=dataset_revision or None,
         dataset_root=Path(dataset_root) if dataset_root else None,
         dataset_manifest_sha256=dataset_manifest_sha256 or None,
+        dataset_content_sha256=dataset_content_sha256 or None,
+        dataset_features=expected_features,
         output=Path(output),
         num_steps=num_steps,
         batch_size=batch_size,

src/tether/finetune/config.py

@@ -179,11 +179,17 @@ class FinetuneConfig:
     """
 
     dataset_root: Path | None = None
-    """Actual EXPORT/dataset directory of a qualified local Studio export."""
+    """Local LeRobot v3 directory, or a qualified Studio export dataset directory."""
 
     dataset_manifest_sha256: str | None = None
     """Canonical sibling export-receipt.json digest, excluding its sha256 field."""
 
+    dataset_content_sha256: str | None = None
+    """Optional expected SHA-256 of the local content manifest, checked before launch."""
+
+    dataset_features: dict[str, Any] = field(default_factory=dict)
+    """Recipe expected features, keyed by LeRobot names with shapes."""
+
     def __post_init__(self) -> None:
         self.output = Path(self.output)
         if self.dataset_root is not None: