I am doing evaluation on huggingface datasets like hotpot-qa. But the results are like this:

'precision_at_2/mean': 0.0, 'recall_at_2/mean': 0.0, 'ndcg_at_2/mean':0.0 .........

I don’t know why i am getting zeros.Here is my dataframe. The results are predictions from a langchain llm RAG:

results = mlflow.evaluate(    data=eval_data,    model_type="retriever",    evaluators="default",    predictions="results",    targets="answer",    extra_metrics=[mlflow.metrics.ndcg_at_k(2), mlflow.metrics.recall_at_k(2), mlflow.metrics.precision_at_k(2), mlflow.metrics.latency()],    evaluator_config={        "col_mapping": {            "inputs": "question",            "context": "context",            "targets": "answer",        }    },)

Is there any issue with the structure of columns “answer” and “results”?

Here is the notebook
https://colab.research.google.com/drive/1CZ27v9Uf_QvfgtP9UDYWas3yMMpIfZ5y?usp=sharing

How can I get the values for metrics correctly. Please help