Add WOQ int8 test with Inductor Freeze

pytorch · Jun 14, 2024 · fac758a · fac758a
1 parent ff42c85
commit fac758a
Showing 1 changed file with 7 additions and 0 deletions.
diff --git a/test/integration/test_integration.py b/test/integration/test_integration.py
@@ -1046,6 +1046,13 @@ def test_int8_weight_only_quant_subclass_api(self, device, dtype):
             change_linear_weights_to_int8_woqtensors, device, 40, test_dtype=dtype
         )
 
+    @parameterized.expand(COMMON_DEVICE_DTYPE)
+    @torch._inductor.config.patch({"freezing": True})
+    def test_int8_weight_only_quant_with_freeze(self, device, dtype):
+        self._test_lin_weight_subclass_api_impl(
+            change_linear_weights_to_int8_woqtensors, device, 40, test_dtype=dtype
+        )
+
     @parameterized.expand(COMMON_DEVICE_DTYPE)
     @unittest.skipIf(not TORCH_VERSION_AFTER_2_3, "int4 requires torch nightly.")
     def test_int4_weight_only_quant_subclass_api(self, device, dtype):