[GPTQ] Fix actorder resolution, add sentinel (#1453)

## Purpose ##
* Fix false assumption that `actorder` field is of enum type
* Despite the fact that actorder passes through a
[field_validator](https://github.com/neuralmagic/compressed-tensors/blob/main/src/compressed_tensors/quantization/quant_args.py#L200),
`QuantizationArgs` has the
[use_enum_values](https://github.com/neuralmagic/compressed-tensors/blob/main/src/compressed_tensors/quantization/quant_args.py#L128)
configuration set, meaning that enum values are converted to strings.
* This was done in relation to [this
fix](neuralmagic/sparseml#2327)
* Remove conflict with recipes which manually specify activation
ordering by using a sentinel value

## Follow ups ##
* #1425

## Testing ##
* Ran llama3 example with manually specified `actorder=group`

---------

Signed-off-by: Kyle Sayers <[email protected]>
Co-authored-by: Dipika Sikka <[email protected]>

Author: kylesayrs

src/llmcompressor/modifiers/quantization/gptq/base.py

@@ -26,6 +26,7 @@
     quantize_weight,
 )
 from llmcompressor.modifiers.quantization.quantization import QuantizationMixin
+from llmcompressor.sentinel import Sentinel
 from llmcompressor.utils.metric_logging import CompressionLogger
 
 __all__ = ["GPTQModifier"]
@@ -109,7 +110,7 @@ class GPTQModifier(Modifier, QuantizationMixin):
     sequential_targets: Union[str, List[str], None] = None
     block_size: int = 128
     dampening_frac: Optional[float] = 0.01
-    actorder: Optional[ActivationOrdering] = None
+    actorder: Optional[Union[ActivationOrdering, Sentinel]] = None
     offload_hessians: bool = False
 
     # private variables
@@ -131,23 +132,29 @@ def validate_sequential_update(cls, value: bool) -> bool:
     def resolve_quantization_config(self) -> QuantizationConfig:
         config = super().resolve_quantization_config()
 
-        # Resolve config with `self.actorder`
+        def resolve_actorder(existing):
+            # sentinel default only overrides if existing is None
+            if self.actorder == Sentinel("static"):
+                return ActivationOrdering.STATIC if existing is None else existing
+
+            # user-provided value always attempts to override
+            if self.actorder is not None:
+                if existing is None or self.actorder == existing:
+                    return self.actorder
+                raise ValueError(
+                    "Cannot resolve activation ordering when both "
+                    "`GPTQModifier.actorder` and `QuantizationScheme.actorder` "
+                    "are provided and differ. Either set `GPTQModifier.actorder = "
+                    "None` or remove `actorder` from config groups."
+                )
+
+            # setting `GPTQModifier.actorder = None` does nothing
+            return existing
+
         for scheme in config.config_groups.values():
-            assert isinstance(scheme, QuantizationScheme)  # (1)
+            assert isinstance(scheme, QuantizationScheme)
             if scheme.weights is not None:
-                existing = scheme.weights.actorder
-                assert isinstance(existing, (ActivationOrdering, type(None)))  # (2)
-                if existing is not None and existing != self.actorder:
-                    raise ValueError(
-                        "Cannot resolve activation ordering when both "
-                        "`GPTQModifier.actorder` and `QuantizationScheme.actorder` "
-                        "both are provided. Either set `GPTQModifier.actorder = None` "
-                        "or remove `actorder` from config groups"
-                    )
-                scheme.weights.actorder = self.actorder
-
-        # (1) QuantizationConfig.model_post_init
-        # (2) QuantizationScheme.validate_actorder
+                scheme.weights.actorder = resolve_actorder(scheme.weights.actorder)
 
         return config
 

src/llmcompressor/sentinel.py

@@ -0,0 +1,52 @@
+import inspect
+
+from pydantic_core import core_schema
+
+_registry = {}
+
+
+class Sentinel:
+    """
+    Unique sentinel values. Implements https://peps.python.org/pep-0661/
+    with dummy pydantic validation
+    """
+
+    def __new__(cls, name, module_name=None):
+        name = str(name)
+
+        if module_name is None:
+            module_name = inspect.currentframe().f_globals.get("__file__")
+            if module_name is None:
+                module_name = __name__
+
+        registry_key = f"{module_name}-{name}"
+
+        sentinel = _registry.get(registry_key, None)
+        if sentinel is not None:
+            return sentinel
+
+        sentinel = super().__new__(cls)
+        sentinel._name = name
+        sentinel._module_name = module_name
+
+        return _registry.setdefault(registry_key, sentinel)
+
+    def __repr__(self):
+        return self._name
+
+    def __reduce__(self):
+        return (
+            self.__class__,
+            (
+                self._name,
+                self._module_name,
+            ),
+        )
+
+    @classmethod
+    def __get_pydantic_core_schema__(cls, _source_type, _handler):
+        return core_schema.no_info_plain_validator_function(cls.validate)
+
+    @classmethod
+    def validate(cls, value: "Sentinel") -> "Sentinel":
+        return value

tests/llmcompressor/test_sentinel.py

@@ -0,0 +1,6 @@
+from llmcompressor.sentinel import Sentinel
+
+
+def test_sentinel():
+    assert Sentinel("MISSING") == Sentinel("MISSING")
+    assert Sentinel("MISSING", "module_one") != Sentinel("MISSING", "module_two")