Fixed tests

Author: AlexKoff88

tests/openvino/test_quantization.py

@@ -329,23 +329,22 @@ def test_ovmodel_8bit_weight_compression_stateful(self, model_cls, model_id, exp
 
     @parameterized.expand(SUPPORTED_ARCHITECTURES_WITH_AUTO_COMPRESSION)
     def test_ovmodel_load_with_compressed_weights(self, model_cls, model_type):
-        with tempfile.TemporaryDirectory() as tmp_dir:
-            model = model_cls.from_pretrained(MODEL_NAMES[model_type], export=True, load_in_8bit=True, stateful=False)
-
-            if model.export_feature.startswith("text2text-generation"):
-                models = [model.encoder, model.decoder, model.decoder_with_past]
-            elif model.export_feature.startswith("stable-diffusion"):
-                models = [model.unet, model.vae_encoder, model.vae_decoder]
-                models.append(
-                    model.text_encoder if model.export_feature == "stable-diffusion" else model.text_encoder_2
-                )
-            else:
-                models = [model]
-
-            expected_ov_int8 = _ARCHITECTURES_TO_EXPECTED_INT8[model_type]
-            for i, model in enumerate(models):
-                _, num_int8, _ = get_num_quantized_nodes(model)
-                self.assertEqual(expected_ov_int8[i], num_int8)
+        model = model_cls.from_pretrained(MODEL_NAMES[model_type], export=True, load_in_8bit=True, stateful=False)
+
+        if model.export_feature.startswith("text2text-generation"):
+            models = [model.encoder, model.decoder, model.decoder_with_past]
+        elif model.export_feature.startswith("stable-diffusion"):
+            models = [model.unet, model.vae_encoder, model.vae_decoder]
+            models.append(
+                model.text_encoder if model.export_feature == "stable-diffusion" else model.text_encoder_2
+            )
+        else:
+            models = [model]
+
+        expected_ov_int8 = _ARCHITECTURES_TO_EXPECTED_INT8[model_type]
+        for i, model in enumerate(models):
+            _, num_int8, _ = get_num_quantized_nodes(model)
+            self.assertEqual(expected_ov_int8[i], num_int8)
 
     @parameterized.expand(SUPPORTED_ARCHITECTURES_WITH_EXPECTED_4BIT_AUTOCOMPRESSED_MATMULS)
     def test_ovmodel_4bit_auto_compression(self, model_cls, model_type, expected_ov_int8, expected_ov_int4):