diff --git a/llama_stack/apis/evals/evals.py b/llama_stack/apis/evals/evals.py
index fb3aa6cd4..ea985ad3b 100644
--- a/llama_stack/apis/evals/evals.py
+++ b/llama_stack/apis/evals/evals.py
@@ -89,15 +89,21 @@ class EvaluatePostprocessConfig(BaseModel):
     kwargs: Optional[Dict[str, Any]] = None
 
 
+@json_schema_type
+class EvaluateProcessorConfig(BaseModel):
+    processor_identifier: str
+    preprocess_config: Optional[EvaluatePreprocessConfig] = None
+    postprocess_config: Optional[EvaluatePostprocessConfig] = None
+
+
 @json_schema_type
 class EvaluateJudgeScoringConfig(BaseModel): ...
 
 
 @json_schema_type
 class LLMJudgeConfig(BaseModel):
-    judge_preprocess_config: EvaluatePreprocessConfig
+    judge_processor_config: EvaluateProcessorConfig
     judge_model_generation_config: EvaluateModelGenerationConfig
-    judge_postprocess_config: EvaluatePostprocessConfig
     judge_scoring_config: EvaluateJudgeScoringConfig
 
 
@@ -116,9 +122,8 @@ class EvaluateScoringConfig(BaseModel):
 @json_schema_type
 class EvaluateTaskConfig(BaseModel):
     dataset_config: EvaluateDatasetConfig
-    preprocess_config: Optional[EvaluatePreprocessConfig] = None
+    processor_config: EvaluateProcessorConfig
     generation_config: EvaluateModelGenerationConfig
-    postprocess_config: Optional[EvaluatePostprocessConfig] = None
     scoring_config: EvaluateScoringConfig
 
 
diff --git a/llama_stack/distribution/registry/generator_processors/__init__.py b/llama_stack/distribution/registry/generator_processors/__init__.py
index bb9d5c182..44972cf03 100644
--- a/llama_stack/distribution/registry/generator_processors/__init__.py
+++ b/llama_stack/distribution/registry/generator_processors/__init__.py
@@ -4,9 +4,16 @@
 # This source code is licensed under the terms described in the LICENSE file in
 # the root directory of this source tree.
 from llama_stack.apis.evals import *  # noqa: F403
+from llama_stack.providers.impls.meta_reference.evals.processor import *  # noqa: F403
 
 from ..registry import Registry
 
+# TODO: decide whether we should group dataset+processor together via Tasks
+GeneratorProcessorRegistry = Registry[BaseGeneratorProcessor]()
 
-class GeneratorProcessorRegistry(Registry[BaseGeneratorProcessor]):
-    _REGISTRY: Dict[str, BaseGeneratorProcessor] = {}
+PROCESSOR_REGISTRY = {
+    "mmlu": MMLUProcessor,
+}
+
+for k, v in PROCESSOR_REGISTRY.items():
+    GeneratorProcessorRegistry.register(k, v)
diff --git a/llama_stack/providers/impls/meta_reference/evals/evals.py b/llama_stack/providers/impls/meta_reference/evals/evals.py
index abd1938ad..80bf2dd7a 100644
--- a/llama_stack/providers/impls/meta_reference/evals/evals.py
+++ b/llama_stack/providers/impls/meta_reference/evals/evals.py
@@ -47,6 +47,9 @@ class MetaReferenceEvalsImpl(Evals):
                     dataset_name=dataset,
                     row_limit=3,
                 ),
+                processor_config=EvaluateProcessorConfig(
+                    processor_identifier="mmlu",
+                ),
                 generation_config=EvaluateModelGenerationConfig(
                     model=model,
                 ),
diff --git a/llama_stack/providers/impls/meta_reference/evals/processor/__init__.py b/llama_stack/providers/impls/meta_reference/evals/processor/__init__.py
index 756f351d8..f782f9320 100644
--- a/llama_stack/providers/impls/meta_reference/evals/processor/__init__.py
+++ b/llama_stack/providers/impls/meta_reference/evals/processor/__init__.py
@@ -3,3 +3,4 @@
 #
 # This source code is licensed under the terms described in the LICENSE file in
 # the root directory of this source tree.
+from .mmlu_processor import MMLUProcessor  # noqa: F401
diff --git a/llama_stack/providers/impls/meta_reference/evals/scorer/__init__.py b/llama_stack/providers/impls/meta_reference/evals/scorer/__init__.py
index 756f351d8..6424963f8 100644
--- a/llama_stack/providers/impls/meta_reference/evals/scorer/__init__.py
+++ b/llama_stack/providers/impls/meta_reference/evals/scorer/__init__.py
@@ -3,3 +3,5 @@
 #
 # This source code is licensed under the terms described in the LICENSE file in
 # the root directory of this source tree.
+from .basic_scorers import *  # noqa: F401 F403
+from .aggregate_scorer import *  # noqa: F401 F403
diff --git a/llama_stack/providers/impls/meta_reference/evals/tasks/run_eval_task.py b/llama_stack/providers/impls/meta_reference/evals/tasks/run_eval_task.py
index 48c450914..83f6264c0 100644
--- a/llama_stack/providers/impls/meta_reference/evals/tasks/run_eval_task.py
+++ b/llama_stack/providers/impls/meta_reference/evals/tasks/run_eval_task.py
@@ -4,15 +4,17 @@
 # This source code is licensed under the terms described in the LICENSE file in
 # the root directory of this source tree.
 from llama_stack.distribution.registry.datasets import DatasetRegistry
+from llama_stack.distribution.registry.generator_processors import (
+    GeneratorProcessorRegistry,
+)
 from llama_stack.distribution.registry.scorers import ScorerRegistry
+
 from llama_stack.providers.impls.meta_reference.evals.scorer.aggregate_scorer import *  # noqa: F403
 from llama_stack.providers.impls.meta_reference.evals.scorer.basic_scorers import *  # noqa: F403
 from llama_stack.providers.impls.meta_reference.evals.generator.inference_generator import (
     InferenceGenerator,
 )
-from llama_stack.providers.impls.meta_reference.evals.processor.mmlu_processor import (
-    MMLUProcessor,
-)
+
 
 from llama_stack.apis.evals import *  # noqa: F403
 from llama_stack.apis.inference import *  # noqa: F403
@@ -46,7 +48,10 @@ class RunEvalTask(BaseTask):
         print(f"Running on {len(dataset)} samples")
 
         # F1
-        processor = MMLUProcessor()
+        print(GeneratorProcessorRegistry.names())
+        processor = GeneratorProcessorRegistry.get(
+            eval_task_config.processor_config.processor_identifier
+        )()
         preprocessed = processor.preprocess(dataset)
 
         # Generation