Add Config Registry for autotune (#1543)

* add config registery for common & torch

Signed-off-by: Kaihui-intel <[email protected]>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* update torch quant algo priority

Signed-off-by: Kaihui-intel <[email protected]>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* add config registry for tf

Signed-off-by: Kaihui-intel <[email protected]>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* add get_cls_config

Signed-off-by: Kaihui-intel <[email protected]>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* add typing check

Signed-off-by: Kaihui-intel <[email protected]>

---------

Signed-off-by: Kaihui-intel <[email protected]>
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>

Author: Kaihui-intel

neural_compressor/common/base_config.py

@@ -40,30 +40,83 @@
 
 
 # Dictionary to store registered configurations
-registered_configs = {}
 
 
-def register_config(framework_name="None", algo_name=None):
+class ConfigRegistry:
+    registered_configs = {}
+
+    @classmethod
+    def register_config_impl(cls, framework_name="None", algo_name=None, priority=0):
+        """Register config decorator.
+
+        The register the configuration classes for different algorithms within specific frameworks.
+
+        Usage example:
+            @ConfigRegistry.register_config(framework_name=FRAMEWORK_NAME, algo_name=ExampleAlgorithm, priority=100)
+            class ExampleAlgorithmConfig:
+                # Configuration details for the ExampleAlgorithm
+
+        Args:
+            framework_name: the framework name. Defaults to "None".
+            algo_name: the algorithm name. Defaults to None.
+            priority: priority: the priority of the configuration. A larger number indicates a higher priority,
+                which will be tried first at the auto-tune stage. Defaults to 0.
+        """
+
+        def decorator(config_cls):
+            cls.registered_configs.setdefault(framework_name, {})
+            cls.registered_configs[framework_name][algo_name] = {"priority": priority, "cls": config_cls}
+            return config_cls
+
+        return decorator
+
+    @classmethod
+    def get_all_configs(cls) -> Dict[str, Dict[str, Dict[str, object]]]:
+        """Get all registered configurations."""
+        return cls.registered_configs
+
+    @classmethod
+    def get_sorted_configs(cls) -> Dict[str, OrderedDict[str, Dict[str, object]]]:
+        """Get registered configurations sorted by priority."""
+        sorted_configs = OrderedDict()
+        for framework_name, algos in sorted(cls.registered_configs.items()):
+            sorted_configs[framework_name] = OrderedDict(
+                sorted(algos.items(), key=lambda x: x[1]["priority"], reverse=True)
+            )
+        return sorted_configs
+
+    @classmethod
+    def get_cls_configs(cls) -> Dict[str, Dict[str, object]]:
+        """Get registered configurations without priority."""
+        cls_configs = {}
+        for framework_name, algos in cls.registered_configs.items():
+            cls_configs[framework_name] = {}
+            for algo_name, config_data in algos.items():
+                cls_configs[framework_name][algo_name] = config_data["cls"]
+        return cls_configs
+
+
+config_registry = ConfigRegistry()
+
+
+def register_config(framework_name="None", algo_name=None, priority=0):
     """Register config decorator.
 
     The register the configuration classes for different algorithms within specific frameworks.
 
     Usage example:
-        @register_config(framework_name="PyTorch", algo_name="ExampleAlgorithm")
+        @register_config(framework_name=FRAMEWORK_NAME, algo_name=ExampleAlgorithm, priority=100)
         class ExampleAlgorithmConfig:
             # Configuration details for the ExampleAlgorithm
 
     Args:
         framework_name: the framework name. Defaults to "None".
         algo_name: the algorithm name. Defaults to None.
+        priority: the priority of the configuration. A larger number indicates a higher priority,
+            which will be tried first at the auto-tune stage. Defaults to 0.
     """
 
-    def decorator(config_cls):
-        registered_configs.setdefault(framework_name, {})
-        registered_configs[framework_name][algo_name] = config_cls
-        return config_cls
-
-    return decorator
+    return config_registry.register_config_impl(framework_name=framework_name, algo_name=algo_name, priority=priority)
 
 
 class BaseConfig(ABC):

neural_compressor/tensorflow/quantization/config.py

@@ -22,7 +22,7 @@
 
 import tensorflow as tf
 
-from neural_compressor.common.base_config import BaseConfig, register_config, registered_configs
+from neural_compressor.common.base_config import BaseConfig, config_registry, register_config
 from neural_compressor.common.utility import DEFAULT_WHITE_LIST, OP_NAME_OR_MODULE_TYPE, STATIC_QUANT
 
 FRAMEWORK_NAME = "keras"
@@ -150,6 +150,7 @@ def register_supported_configs(cls) -> List[OperatorConfig]:
 
 def get_all_registered_configs() -> Dict[str, BaseConfig]:
     """Get all registered configs for keras framework."""
+    registered_configs = config_registry.get_cls_configs()
     return registered_configs.get(FRAMEWORK_NAME, {})
 
 

neural_compressor/torch/quantization/config.py

@@ -23,14 +23,15 @@
 
 import torch
 
-from neural_compressor.common.base_config import BaseConfig, register_config, registered_configs
+from neural_compressor.common.base_config import BaseConfig, config_registry, register_config
 from neural_compressor.common.utility import (
     DEFAULT_WHITE_LIST,
     FP8_QUANT,
     GPTQ,
     OP_NAME_OR_MODULE_TYPE,
     RTN_WEIGHT_ONLY_QUANT,
 )
+from neural_compressor.torch.utils.constants import PRIORITY_GPTQ, PRIORITY_RTN
 from neural_compressor.torch.utils.utility import is_hpex_avaliable, logger
 
 FRAMEWORK_NAME = "torch"
@@ -59,7 +60,7 @@ class OperatorConfig(NamedTuple):
 ######################## RNT Config ###############################
 
 
-@register_config(framework_name=FRAMEWORK_NAME, algo_name=RTN_WEIGHT_ONLY_QUANT)
+@register_config(framework_name=FRAMEWORK_NAME, algo_name=RTN_WEIGHT_ONLY_QUANT, priority=PRIORITY_RTN)
 class RTNWeightQuantConfig(BaseConfig):
     """Config class for round-to-nearest weight-only quantization."""
 
@@ -185,7 +186,7 @@ def get_default_rtn_config() -> RTNWeightQuantConfig:
 
 
 ######################## GPTQ Config ###############################
-@register_config(framework_name=FRAMEWORK_NAME, algo_name=GPTQ)
+@register_config(framework_name=FRAMEWORK_NAME, algo_name=GPTQ, priority=PRIORITY_GPTQ)
 class GPTQConfig(BaseConfig):
     """Config class for GPTQ.
 
@@ -403,4 +404,5 @@ def get_default_fp8_qconfig() -> FP8QConfig:
     ##################### Algo Configs End ###################################
 
     def get_all_registered_configs() -> Dict[str, BaseConfig]:
+        registered_configs = config_registry.get_all_configs()
         return registered_configs.get(FRAMEWORK_NAME, {})

neural_compressor/torch/quantization/quantize.py

@@ -17,7 +17,7 @@
 
 import torch
 
-from neural_compressor.common.base_config import BaseConfig, ComposableConfig, registered_configs
+from neural_compressor.common.base_config import BaseConfig, ComposableConfig, config_registry
 from neural_compressor.common.logger import Logger
 from neural_compressor.torch.quantization.config import FRAMEWORK_NAME
 from neural_compressor.torch.utils.utility import WHITE_MODULE_LIST, algos_mapping, get_model_info
@@ -48,6 +48,7 @@ def quantize(
         The quantized model.
     """
     q_model = model if inplace else copy.deepcopy(model)
+    registered_configs = config_registry.get_cls_configs()
     if isinstance(quant_config, dict):
         quant_config = ComposableConfig.from_dict(quant_config, config_registry=registered_configs[FRAMEWORK_NAME])
         logger.info(f"Parsed a config dict to construct the quantization config: {quant_config}.")
@@ -88,6 +89,7 @@ def quantize_dynamic(
         The quantized model.
     """
     q_model = model if inplace else copy.deepcopy(model)
+    registered_configs = config_registry.get_cls_configs()
     if isinstance(quant_config, dict):
         quant_config = ComposableConfig.from_dict(quant_config, config_registry=registered_configs[FRAMEWORK_NAME])
         logger.info(f"Parsed a config dict to construct the quantization config: {quant_config}.")

neural_compressor/torch/utils/constants.py

@@ -36,3 +36,7 @@
         "double_quant_group_size": 256,
     },
 }
+
+# Setting priorities for algorithms, a higher number indicates a higher priority.
+PRIORITY_RTN = 80
+PRIORITY_GPTQ = 90