Skipping aqlm non working inference tests till fix merged

src/transformers/testing_utils.py

@@ -346,6 +346,17 @@ def tooslow(test_case):
     return unittest.skip(reason="test is too slow")(test_case)
 
 
+def skip_if_aqlm_inference_not_fixed(test_case):
+    """
+    Decorator marking tests for inference using aqlm models.
+
+    These tests will  be skipped till the  issue from aqlm side is resolved
+    """
+    return unittest.skip(
+        reason="inference doesn't work with quantized aqlm models using torch.Any type with recent torch versions"
+    )(test_case)
+
+
 def skip_if_not_implemented(test_func):
     @functools.wraps(test_func)
     def wrapper(*args, **kwargs):

tests/quantization/aqlm_integration/test_aqlm.py

@@ -26,6 +26,7 @@
     require_aqlm,
     require_torch_gpu,
     require_torch_multi_gpu,
+    skip_if_aqlm_inference_not_fixed,
     slow,
     torch_device,
 )
@@ -142,6 +143,7 @@ def test_quantized_model_conversion(self):
 
         self.assertEqual(nb_linears - 1, nb_aqlm_linear)
 
+    @skip_if_aqlm_inference_not_fixed
     def test_quantized_model(self):
         """
         Simple test that checks if the quantized model is working properly
@@ -158,6 +160,7 @@ def test_raise_if_non_quantized(self):
         with self.assertRaises(ValueError):
             _ = AutoModelForCausalLM.from_pretrained(model_id, quantization_config=quantization_config)
 
+    @skip_if_aqlm_inference_not_fixed
     def test_save_pretrained(self):
         """
         Simple test that checks if the quantized model is working properly after being saved and loaded
@@ -171,6 +174,7 @@ def test_save_pretrained(self):
             output = model.generate(**input_ids, max_new_tokens=self.max_new_tokens)
             self.assertEqual(self.tokenizer.decode(output[0], skip_special_tokens=True), self.EXPECTED_OUTPUT)
 
+    @skip_if_aqlm_inference_not_fixed
     @require_torch_multi_gpu
     def test_quantized_model_multi_gpu(self):
         """