Skip to content

Run frontier MoE models on hardware you already own — pure C, zero deps, experts streamed from disk. Tiny engine, immense model. 🐦

C

fix(test): stub numpy in the OLMoE converter import mock — dev CI is red

JustVuggb84154d39a0e
diff --git a/c/tests/test_olmoe_atomic_output.py b/c/tests/test_olmoe_atomic_output.py
index 6ce4a710..c403ddf7 100644
--- a/c/tests/test_olmoe_atomic_output.py
+++ b/c/tests/test_olmoe_atomic_output.py
@@ -19,6 +19,11 @@ def load_converter():
spec = importlib.util.spec_from_file_location("olmoe_atomic_output_test", CONVERTER)
module = importlib.util.module_from_spec(spec)
with mock.patch.dict(sys.modules, {
+ # numpy joins the converter's dependency guard in #1048 (safetensors.torch
+ # imports it internally). This test imports the converter with its real
+ # dependencies absent, so every name in that guard has to be stubbed here
+ # or the guard sys.exit()s during exec_module and the module never loads.
+ "numpy": mock.Mock(),
"torch": torch,
"safetensors": safetensors,
"safetensors.torch": safetensors_torch,