refactor(recipes): rename recipes, drop pi05_hirobot

- hirobot.yaml            -> subtasks_vqa.yaml
- hirobot_memory.yaml     -> subtask_mem_vqa_speech.yaml
- pi05_hirobot.yaml       -> deleted (stale: uses plan, top-camera names;
  superseded by the two recipes above)
- smolvla2_hirobot.yaml   -> deleted (was untracked stale junk)

Updated the smolvla2 / pi052 `recipe_path` config defaults, all
docstring / comment references, the annotation-pipeline + recipe docs,
and the three tests that loaded pi05_hirobot.yaml (repointed to the
renamed recipes; the low-level-branch and pipeline-render assertions
now accept a flow-only `low_level` stream as valid supervision, since
the new recipes' low_level_execution has no text-CE target).

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

tests/annotations/test_pipeline_recipe_render.py

@@ -41,7 +41,12 @@ from lerobot.datasets.language_render import render_sample
 from ._helpers import make_canned_responder
 
 _RECIPE_PATH = (
-    Path(__file__).resolve().parents[2] / "src" / "lerobot" / "configs" / "recipes" / "pi05_hirobot.yaml"
+    Path(__file__).resolve().parents[2]
+    / "src"
+    / "lerobot"
+    / "configs"
+    / "recipes"
+    / "subtask_mem_vqa_speech.yaml"
 )
 
 
@@ -105,22 +110,29 @@ def test_pr1_canonical_recipe_renders_nonempty_from_pipeline_output(
         recipe = TrainingRecipe(**loaded)
 
     rendered_any = False
-    for ts, persistent, events in zip(timestamps, persistent_lists, events_lists, strict=True):
+    for sample_idx, (ts, persistent, events) in enumerate(
+        zip(timestamps, persistent_lists, events_lists, strict=True)
+    ):
         result = render_sample(
             recipe=recipe,
             persistent=persistent,
             events=events,
             t=float(ts),
-            sample_idx=0,
+            sample_idx=sample_idx,
             dataset_ctx={"task": "Pour water from the bottle into the cup."},
         )
         if result is None:
             continue
         if result["messages"]:
             rendered_any = True
-            assert result["target_message_indices"]
+            # A valid render supervises something: a text-CE target turn
+            # OR a flow-only ``low_level``-stream turn (action loss).
+            assert (
+                result["target_message_indices"]
+                or "low_level" in result["message_streams"]
+            )
             break
-    assert rendered_any, "PR 1 recipe rendered no messages from pipeline output"
+    assert rendered_any, "recipe rendered no messages from pipeline output"
 
     # Sanity: speech atom appears in events column intact
     flat_events = [r for ev in events_lists for r in ev]

tests/configs/test_recipe.py

@@ -18,7 +18,9 @@ def test_message_recipe_validates_unknown_binding():
 
 
 def test_canonical_recipe_loads():
-    recipe = TrainingRecipe.from_yaml(Path("src/lerobot/configs/recipes/pi05_hirobot.yaml"))
+    recipe = TrainingRecipe.from_yaml(
+        Path("src/lerobot/configs/recipes/subtask_mem_vqa_speech.yaml")
+    )
 
     assert recipe.blend is not None
     assert set(recipe.blend) == {
@@ -29,4 +31,4 @@ def test_canonical_recipe_loads():
         "ask_vqa_top",
         "ask_vqa_wrist",
     }
-    assert sum(component.weight for component in recipe.blend.values()) == pytest.approx(0.96)
+    assert sum(component.weight for component in recipe.blend.values()) == pytest.approx(1.0)

tests/datasets/test_language_render.py

@@ -449,7 +449,10 @@ def test_vqa_frame_is_consumed_over_the_weighted_blend():
 
 
 def test_canonical_recipe_can_render_low_level_branch():
-    recipe = TrainingRecipe.from_yaml(Path("src/lerobot/configs/recipes/pi05_hirobot.yaml"))
+    """The shipped ``subtasks_vqa.yaml`` recipe's ``low_level_execution``
+    branch renders — a flow-only ``user(${subtask})`` turn (no text-CE
+    target; its supervision is the action-expert flow loss)."""
+    recipe = TrainingRecipe.from_yaml(Path("src/lerobot/configs/recipes/subtasks_vqa.yaml"))
     low_level = TrainingRecipe(blend={"low": recipe.blend["low_level_execution"]})
 
     rendered = render_sample(
@@ -461,6 +464,6 @@ def test_canonical_recipe_can_render_low_level_branch():
         task="clean kitchen",
     )
 
-    assert rendered["messages"][-1] == {"role": "assistant", "content": "subtask 0"}
+    assert rendered["messages"][-1] == {"role": "user", "content": "subtask 0"}
     assert rendered["message_streams"][-1] == "low_level"
-    assert rendered["target_message_indices"] == [1]
+    assert rendered["target_message_indices"] == []