What is the overall goal of this workflow exercise?	To improve a model grader by providing it with more context on what a good solution should look like.
How many steps are required to add additional solution context to the model grader?	Two steps.
In Step 1, where should you modify the prompt to add context?	The prompt used to generate the dataset.
What specific information should be requested in the data generation prompt during Step 1?	Solution criteria to be included in every test case.
What new element is added to the test cases after Step 1?	An additional key (e.g., "solution criteria") detailing characteristics of a good solution.
In Step 2, where is the newly generated solution criteria inserted?	Into the prompt of the "grade by model" function.
What instruction is given to the model grader using the solution criteria in Step 2?	To use the criteria to evaluate the solution.
What is the expected benefit of including solution criteria in the model grader prompt?	The model grader's reasoning section should be more fleshed out.
