Add readdoc for the StepReasoningTool

Author: Val Brodsky

docs/labelbox/index.rst

@@ -46,6 +46,7 @@ Labelbox Python SDK Documentation
     search-filters
     send-to-annotate-params
     slice
+    step_reasoning_tool
     task
     task-queue
     user

docs/labelbox/step_reasoning_tool.rst

@@ -0,0 +1,6 @@
+Step Reasoning Tool
+===============================================================================================
+
+.. automodule:: labelbox.schema.tool_building.step_reasoning_tool
+   :members:
+   :show-inheritance:

libs/labelbox/src/labelbox/schema/tool_building/step_reasoning_tool.py

@@ -62,6 +62,11 @@ def _create_incorrect_step() -> IncorrectStepReasoningVariant:
 
 @dataclass
 class StepReasoningVariants:
+    """
+    This class is used to define the possible options for evaluating a step
+    Currently the options are correct, neutral, and incorrect
+    """
+
     CORRECT_STEP_ID = 0
     NEUTRAL_STEP_ID = 1
     INCORRECT_STEP_ID = 2
@@ -136,6 +141,11 @@ def from_dict(cls, dictionary: Dict[str, Any]) -> "StepReasoningDefinition":
 
 @dataclass
 class StepReasoningTool:
+    """
+    Use this class in OntologyBuilder to create a tool for step reasoning
+    The definition field lists the possible options to evaulate a step
+    """
+
     name: str
     type: ToolType = field(default=ToolType.STEP_REASONING, init=False)
     required: bool = False
@@ -147,9 +157,15 @@ class StepReasoningTool:
     )
 
     def set_regenerate_conversations_after_incorrect_step(self):
+        """
+        For live models, will invoke the model to generate alternatives if a step is marked as incorrect
+        """
         self.definition.variants.incorrect_step.regenerate_conversations_after_incorrect_step = True
 
     def set_rate_alternative_responses(self):
+        """
+        For live models, will require labelers to rate the alternatives generated by the model
+        """
         self.definition.variants.incorrect_step.rate_alternative_responses = (
             True
         )