Run frontier MoE models on hardware you already own — pure C, zero deps, experts streamed from disk. Tiny engine, immense model. 🐦
C
test(glm53): run the serve and pin_branch harnesses in the oracle job
enic5581d4bec0f
diff --git a/c/tests/test_glm53_oracles.py b/c/tests/test_glm53_oracles.pyindex 2f2bc77a..285513fd 100644--- a/c/tests/test_glm53_oracles.py+++ b/c/tests/test_glm53_oracles.py@@ -78,6 +78,15 @@ class Glm53TinyOracleTest(unittest.TestCase):"""The vision tower and the image tokens in the prompt, token-exact."""self.run_oracle("multimodal_tiny", "GLM53_MM_TINY")+ def test_serve_harness(self):+ """REUSE and CANCEL over the serve protocol, including the reason+ REUSE gives for reusing the slot or not."""+ self.run_oracle("serve", "GLM53_MM_TINY")++ def test_pin_branch_harness(self):+ """A pin is not restored over rows another branch rewrote (#1750)."""+ self.run_oracle("pin_branch", "GLM53_MM_TINY")+if __name__ == "__main__":unittest.main()