fix(style): pre-commit

src/lerobot/policies/groot/processor_groot.py

@@ -1670,7 +1670,7 @@ class GrootN17PackInputsStep(ProcessorStep):
 
         Relative groups normalize with per-chunk-timestep (2D) ``relative_action`` stats, which the
         flat ``_min_max_norm`` fallback cannot honor, so a relative config that fails grouped
-        normalization must fail loudly rather than silently mis-scale every timestep.
+        normalization must fail loudly rather than silently wrongly scale every timestep.
         """
         if not isinstance(self.modality_config, dict):
             return False
@@ -1807,7 +1807,7 @@ class GrootN17PackInputsStep(ProcessorStep):
                         "relative-action chunk: the action layout or horizon does not match the "
                         f"checkpoint relative_action stats (action shape {tuple(action.shape)}). The flat "
                         "min/max fallback cannot honor per-chunk-timestep relative stats, so refusing to "
-                        "silently mis-normalize. Recompute the relative action stats so their horizon and "
+                        "silently wrongly normalize. Recompute the relative action stats so their horizon and "
                         "dimensions match the action chunk."
                     )
                 else:

tests/policies/groot/test_groot_n1_7.py

@@ -1022,7 +1022,7 @@ def test_groot_n1_7_pack_inputs_normalizes_action_chunk_per_dimension_before_pad
 
 def test_groot_n1_7_pack_inputs_raises_when_relative_groups_cannot_normalize():
     # Relative groups carry per-chunk-timestep stats; if the action horizon exceeds the available
-    # stat rows, grouped normalization cannot apply and the flat fallback would silently mis-scale.
+    # stat rows, grouped normalization cannot apply and the flat fallback would silently wrongly scale.
     step = GrootN17PackInputsStep(
         action_horizon=3,
         valid_action_horizon=3,

tests/policies/groot/utils/dump_original_n1_7.py

@@ -62,10 +62,7 @@ def make_observation(seed: int, video_keys, lang_key, state_spec):
     # One ndarray per state key, shape (B, T=1, key_dim); dim taken from statistics.
     # Keys with dim 0 (e.g. disabled eef on some embodiments) are still emitted as
     # present-but-empty so the processor's state transform finds every expected key.
-    state = {
-        k: rng.standard_normal((BATCH_SIZE, 1, dim)).astype(np.float32)
-        for k, dim in state_spec
-    }
+    state = {k: rng.standard_normal((BATCH_SIZE, 1, dim)).astype(np.float32) for k, dim in state_spec}
     language = {lang_key: [[PROMPT] for _ in range(BATCH_SIZE)]}
     return {"video": video, "state": state, "language": language}
 
@@ -181,7 +178,12 @@ def main():
         state_spec = [(k, len(v["min"])) for k, v in stats[tag]["state"].items()]
         try:
             dump_one_tag(
-                policy, fair_model, tag, all_modality[tag], state_spec, args,
+                policy,
+                fair_model,
+                tag,
+                all_modality[tag],
+                state_spec,
+                args,
                 out_dir / f"original_n1_7_{tag}.npz",
             )
             done.append(tag)