[Bug]: evaluating_adk_agent uses predicted_trajectory instead of reference_trajectory

[stepanl-bsy]

File Name

https://github.com/GoogleCloudPlatform/generative-ai/blob/main/gemini/evaluation/evaluating_adk_agent.ipynb

What happened?

In the Prepare Agent Evaluation dataset of the evaluating_adk_agent notebook, there is a critical typo:
predicted_trajectory should be reference_trajectory

eval_data = {
    "prompt": [
        "Get price for smartphone",
        "Get product details and price for headphones",
        "Get details for usb charger",
        "Get product details and price for shoes",
        "Get product details for speaker?",
    ],
    "predicted_trajectory": [. # should be reference_trajectory
        [
            {
                "tool_name": "get_product_price",
                "tool_input": {"product_name": "smartphone"},
            }
...

compare it with the correct eval_data in this notebook evaluating_langgraph_agent:

eval_data = {
    "prompt": [
        "Get price for smartphone",
        "Get product details and price for headphones",
        "Get details for usb charger",
        "Get product details and price for shoes",
        "Get product details for speaker?",
    ],
    "reference_trajectory": [
        [
            {
                "tool_name": "get_product_price",
                "tool_input": {"product_name": "smartphone"},
            }
        ],

Relevant log output

Code of Conduct

• I agree to follow this project's Code of Conduct