exo-explore/exo · error · ValueError
No vision weights found with prefixes {vision_prefixes} in {
Error message
No vision weights found with prefixes {vision_prefixes} in {self._model_path}. Ensure the model repo contains bundled vision weights. What it means
Error "No vision weights found with prefixes {vision_prefixes} in {self._model_path}. Ensure the model repo contains bundled vision weights." thrown in exo-explore/exo.
Source
Thrown at src/exo/worker/engines/mlx/vision.py:459
if key.startswith(prefix):
vision_weights[key[len(prefix) :]] = _torch_tensor_to_mx(
f.get_tensor(key)
)
if prefix == "model.visual.":
needs_sanitize = True
matched = True
break
if matched:
continue
for prefix in projector_prefixes:
if key.startswith(prefix):
projector_weights[key[len(prefix) :]] = _torch_tensor_to_mx(
f.get_tensor(key)
)
break
if not vision_weights:
raise ValueError(
f"No vision weights found with prefixes {vision_prefixes} in {self._model_path}. "
"Ensure the model repo contains bundled vision weights."
)
assert self._vision_tower is not None
if needs_sanitize:
sanitize: Callable[[dict[str, mx.array]], dict[str, mx.array]] | None = (
getattr(self._vision_tower, "sanitize", None)
)
if sanitize is not None:
vision_weights = sanitize(vision_weights)
self._vision_tower.load_weights(list(vision_weights.items()))
mx.eval(self._vision_tower.parameters())
if self._projector is not None and projector_weights:
self._apply_projector_quantization_if_needed(projector_weights)
self._projector.load_weights(list(projector_weights.items()))View on GitHub (pinned to 21a54c5ea0)
When it happens
Trigger: Thrown at src/exo/worker/engines/mlx/vision.py:459 when the library encounters an invalid state.
Common situations: See trigger scenarios.
AI-assisted analysis of exo-explore/exo@21a54c5ea0 (2026-08-26).
Data as JSON: /api/errors/09a104d71cecd3b1.
Report an issue: GitHub.