eval.yaml

name: Evasion Bench
description: >
  EvasionBench is a benchmark dataset for detecting evasive answers in earnings
  call Q&A sessions. The task is to classify how directly corporate management
  addresses questions from financial analysts.
tasks:
  - id: evasion_bench
    config: default
    split: train

    epochs: 1

    shuffle_choices: true

    field_spec:
      input: question
      target: "eva4b_label_letter"
      choices: "choices"
      metadata: ["answer"]

    solvers:
      - name: prompt_template
        args:
          template: |
            Question: {prompt}

            Answer: {answer}
      - name: multiple_choice
        args:
          template: |
            You are a financial analyst. Your task is to Detect Evasive Answers in Financial Q&A.
            The entire content of your response should be of the following format: 'ANSWER: $LETTER' (without quotes) where LETTER is one of {letters}.

            {question}

            {choices}
    scorers:
      - name: choice