fix(language): drop motion from VIEW_DEPENDENT_STYLES

Motion primitives are described in robot-frame (joint / Cartesian) terms,
not pixel space, so they are camera-agnostic. Only `vqa` (event) and
`trace` (event, pixel-trajectory) are view-dependent.

The `camera` field stays on PERSISTENT_ROW_FIELDS for schema symmetry —
the validator, resolver, and HF feature mapping behave identically across
the two columns regardless of which styles populate `camera` today —
but persistent rows now always have `camera=None` in practice.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
This commit is contained in:
Pepijn
2026-04-30 10:54:12 +02:00
parent 5a6aa64570
commit 1ca38d9748
2 changed files with 18 additions and 6 deletions
+9 -2
View File
@@ -36,8 +36,15 @@ EVENT_ONLY_STYLES = {"interjection", "vqa", "trace"}
# Styles whose ``content`` is grounded in a specific camera view. Rows of these
# styles MUST carry a non-null ``camera`` referencing an ``observation.images.*``
# feature key. Rows of every other style MUST have ``camera=None``.
VIEW_DEPENDENT_STYLES = {"vqa", "motion", "trace"}
# feature key. Rows of every other style MUST have ``camera=None``. ``motion``
# is intentionally NOT in this set: motion primitives are described in
# robot-frame (joint / Cartesian) terms, not pixel space, so they are
# camera-agnostic. ``trace`` is the pixel-trajectory event style and IS
# view-dependent. The ``camera`` field nevertheless lives on
# ``PERSISTENT_ROW_FIELDS`` too so the schema, validator, and resolver
# behave symmetrically across the two columns; persistent rows simply
# always have ``camera=None`` in practice today.
VIEW_DEPENDENT_STYLES = {"vqa", "trace"}
LanguageColumn = Literal["language_persistent", "language_events"]