Add deprecation warnings for various inference configs (#3294)

* Update

[ghstack-poisoned]

* Update

[ghstack-poisoned]

* Update

[ghstack-poisoned]

Author: vkuzo

test/quantization/test_quant_api.py

@@ -826,19 +826,25 @@ def test_config_deprecation(self):
             uintx_weight_only: (torch.uint4,),
         }
 
-        with warnings.catch_warnings(record=True) as _warnings:
-            # Call each deprecated API twice
-            for cls, args in deprecated_apis_to_args.items():
+        # Call each deprecated API twice
+        for cls, args in deprecated_apis_to_args.items():
+            with warnings.catch_warnings(record=True) as _warnings:
                 cls(*args)
                 cls(*args)
 
-            # Each call should trigger the warning only once
-            self.assertEqual(len(_warnings), len(deprecated_apis_to_args))
-            for w in _warnings:
-                self.assertIn(
-                    "is deprecated and will be removed in a future release",
-                    str(w.message),
-                )
+                # Each call should have at least one warning.
+                # Some of them can have two warnings - one for deprecation,
+                # one for moving to prototype
+                # 1 warning - just deprecation
+                # 2 warnings - deprecation and prototype warnings
+                self.assertTrue(len(_warnings) in (1, 2))
+                found_deprecated = False
+                for w in _warnings:
+                    if "is deprecated and will be removed in a future release" in str(
+                        w.message
+                    ):
+                        found_deprecated = True
+                    self.assertTrue(found_deprecated)
 
 
 common_utils.instantiate_parametrized_tests(TestQuantFlow)

torchao/quantization/quant_api.py

@@ -594,6 +594,9 @@ def __post_init__(self):
         torch._C._log_api_usage_once(
             "torchao.quantization.Int8DynamicActivationInt4WeightConfig"
         )
+        warnings.warn(
+            "`Int8DynamicActivationInt4WeightConfig` will be moving to prototype in a future release of torchao. Please see https://github.com/pytorch/ao/issues/2752 for more details."
+        )
 
 
 # for BC
@@ -965,6 +968,9 @@ def __post_init__(self):
         torch._C._log_api_usage_once(
             "torchao.quantization.Int4DynamicActivationInt4WeightConfig"
         )
+        warnings.warn(
+            "`Int4DynamicActivationInt4WeightConfig` will be moving to prototype in a future release of torchao. Please see https://github.com/pytorch/ao/issues/2752 for more details."
+        )
 
 
 # for bc
@@ -1028,6 +1034,9 @@ def __post_init__(self):
         torch._C._log_api_usage_once(
             "torchao.quantization.GemliteUIntXWeightOnlyConfig"
         )
+        warnings.warn(
+            "`GemliteUIntXWeightOnlyConfig` will be moving to prototype in a future release of torchao. Please see https://github.com/pytorch/ao/issues/2752 for more details."
+        )
 
 
 # for BC
@@ -1981,6 +1990,9 @@ def __post_init__(self):
         torch._C._log_api_usage_once(
             "torchao.quantization.Float8StaticActivationFloat8WeightConfig"
         )
+        warnings.warn(
+            "`Float8StaticActivationFloat8WeightConfig` will be moving to prototype in a future release of torchao. Please see https://github.com/pytorch/ao/issues/2752 for more details."
+        )
 
 
 # for bc
@@ -2069,6 +2081,9 @@ class UIntXWeightOnlyConfig(AOBaseConfig):
 
     def __post_init__(self):
         torch._C._log_api_usage_once("torchao.quantization.UIntXWeightOnlyConfig")
+        warnings.warn(
+            "`UIntXWeightOnlyConfig` will be moving to prototype in a future release of torchao. Please see https://github.com/pytorch/ao/issues/2752 for more details."
+        )
 
 
 # for BC
@@ -2353,6 +2368,9 @@ class FPXWeightOnlyConfig(AOBaseConfig):
 
     def __post_init__(self):
         torch._C._log_api_usage_once("torchao.quantization.FPXWeightOnlyConfig")
+        warnings.warn(
+            "`FPXWeightOnlyConfig` will be moving to prototype in a future release of torchao. Please see https://github.com/pytorch/ao/issues/2752 for more details."
+        )
 
 
 # for BC