updates to config names and desc.

Author: nitsanluke

fast_llm/data/preparator/gpt_memmap/config.py

@@ -112,21 +112,21 @@ def _validate(self) -> None:
 @config_class
 class LossMaskSpansConfig(Config):
     masking_column: str = Field(
-        default=None, 
-        desc="Field containing character spans to mask for loss computation",
-        hint=FieldHint.optional,
+        default="",
+        desc="Field containing (input) character spans for loss masking",
+        hint=FieldHint.core,
     )
     loss_masking_spans: str = Field(
         default="fast_llm_loss_masking_spans",
-        desc="Field containing character spans to mask for loss computation",
+        desc="Column name of field that would contain the masked spans.",
         hint=FieldHint.optional,
     )   
     def _validate(self) -> None:
         assert isinstance(self.loss_masking_spans, str), "loss_masking_spans col name must be a string."
         super()._validate()
 
 @config_class
-class FieldCombinePreparatorConfig(Config):
+class CombineFieldsConfig(Config):
     col_names: typing.List[str] = Field(
         default_factory=list,
         desc="Fields of the dataset to combine.",
@@ -205,7 +205,7 @@ class GPTMemmapDatasetPreparatorConfig(DatasetPreparatorConfig):
         " Does not shuffle samples.",
         hint=FieldHint.optional,
     )
-    combine_fields: FieldCombinePreparatorConfig = Field(
+    combine_fields: CombineFieldsConfig = Field(
         default=None,
         desc="Combine all files into a single file.",
         hint=FieldHint.optional,