Added more detailed logging.

Author: adivekar-utexas

src/fmcore/framework/_algorithm.py

@@ -230,6 +230,7 @@ def create_hyperparams(cls, hyperparams: Optional[Dict] = None) -> Hyperparamete
     @classmethod
     def convert_params(cls, params: Dict) -> Dict:
         ## Convert and validate parameters for the algorithm
+        # print(f'params for {cls.class_name}=\n{params}')
         cls.set_default_param_values(params)
         ## This allows us to create a new Algorithm instance without specifying `hyperparams`.
         ## If it is specified, we will pick cls.Hyperparameters, which can be overridden by the subclass.

src/fmcore/framework/_chain/Chain.py

@@ -291,8 +291,8 @@ def run(
         background: bool = False,
         tracker: Optional[Union[Tracker, Dict, str]] = None,
         notifier: Optional[Union[Notifier, Dict, str]] = None,
-        store_step_inputs: bool = False,
-        store_step_outputs: bool = False,
+        store_step_inputs: bool = True,
+        store_step_outputs: bool = True,
         after: Optional[ChainExecution] = None,
         after_wait: conint(ge=0) = 15,
         step_wait: confloat(ge=0.0) = 0.0,

src/fmcore/framework/_dataset.py

@@ -85,6 +85,10 @@ def concat(
     @model_validator(mode="before")
     @classmethod
     def _set_dataset_params(cls, params: Dict) -> Dict:
+        if "data_schema" not in params:
+            raise ValueError(
+                f"Cannot create instance of class '{cls.class_name}' without passing `data_schema` parameter."
+            )
         data_schema: Union[Schema, MLTypeSchema] = params["data_schema"]
         if isinstance(data_schema, dict):
             ## We need to infer the schema:

src/fmcore/framework/_predictions.py

@@ -36,8 +36,8 @@
     MLTypeSchema,
     TaskOrStr,
 )
-from fmcore.framework._task_mixins import InputOutputDataMixin, SchemaValidationError
 from fmcore.framework._dataset import Dataset
+from fmcore.framework._task_mixins import InputOutputDataMixin, SchemaValidationError
 
 Predictions = "Predictions"
 Visualization = "Visualization"
@@ -69,6 +69,11 @@ def _pre_registration_hook(cls):
     @model_validator(mode="before")
     @classmethod
     def _set_predictions_params(cls, params: Dict) -> Dict:
+        if "data_schema" not in params:
+            raise ValueError(
+                f"Cannot create instance of class '{cls.class_name}' without passing `data_schema` parameter."
+            )
+
         params["data_schema"]: Schema = Schema.of(params["data_schema"], schema_template=cls.schema_template)
         # data_schema: Union[Schema, MLTypeSchema] = params['data_schema']
         # if isinstance(data_schema, dict):

src/fmcore/framework/_task/text_generation.py

@@ -752,12 +752,12 @@ def _create_predictions(self, batch: Prompts, predictions: Dict, **kwargs) -> Ne
 
 
 class LanguageModelTaskMixin(Algorithm, ABC):
-    lm: Optional[Union[GenerativeLM, Any]] = None
+    lm: Optional[Any] = None
     icl_dataset: Optional[Dataset] = None
     icl_sampler: Optional[ICLSampler] = None  ## Will be not-None when icl_dataset is not-None.
 
     class Hyperparameters(Algorithm.Hyperparameters):
-        lm: Optional[Dict]  ## Params for llm
+        lm: Optional[Dict] = None  ## Params for llm
         batch_size: Optional[conint(ge=1)] = 1  ## By default, predict 1 row at a time.
         prompt_template: constr(min_length=1)
         icl_template: Optional[constr(min_length=1)] = None