cnt_agents: &cnt_agents 2
max_turn: &max_turn 3
max_inner_turns: &max_inner_turns 3

prompts:
  role_assigner_prepend_prompt: &role_assigner_prepend_prompt |-

  role_assigner_append_prompt: &role_assigner_append_prompt |-
    # Role Description
    You are the leader of a group of experts, now you need to generate a response based on the text:
    ${task_description}
    
    You can recruit ${cnt_critic_agents} expert in different fields. What experts will you recruit to better generate an accurate solution?
    
    # Response Format Guidance
    You should respond with a list of expert description. For example:
    1. an electrical engineer specified in the filed of xxx
    2. an economist who is good at xxx
    3. a lawyer with a good knowledge of xxx
    ...
    
    You don't have to give the reason.

  solver_prepend_prompt: &solver_prepend_prompt |-
    You are ${role_description}. Below is a chat history:
    ${task_description}

    And below is the discussion about what the next system response should be:
    
  solver_append_prompt: &solver_append_prompt |-
    Now based on these information, please give a better next system response. Your response should contain only one system response beginning with "System: ". Do not give any additional information.

  critic_prepend_prompt: &critic_prepend_prompt |-
    You are ${role_description}. You are in a discussion group, aiming to generate a system response to the following chat history:
    ${task_description}

    Below is the discussion about what the next system response should be:
    
  critic_append_prompt: &critic_append_prompt |-
    # Response Format Guidance
    - If you thinks the latest response given above is perfect, respond using the following format:
    Decision: (set it to "Agree")
    Response: (your response on why you think it is perfect)

    - If you think it is flawed, give your advice use the following output format:
    Decision: (set it to "Disagree")
    Response: (explain why you disagree and give your advice)
    
    Based on your knowledge in your field, do you agree that this solution is a good response to the chat history?

  manager_prompt: &manager_prompt |-

  executor_prepend_prompt: &executor_prepend_prompt |-

  executor_append_prompt: &executor_append_prompt |-

  evaluator_prepend_prompt: &evaluator_prepend_prompt |-

  evaluator_append_prompt: &evaluator_append_prompt |-
    # Role Description
    You are an experienced dialogue teacher. As a good teacher, you carefully assess the given system response based on the chat history. When the response is flawed, you should patiently teach the system how to give better response.
  
    # Response Format Guidance
    You must respond in the following format:
    Engaging: (a score between 0 and 10)
    Relevant: (a score between 0 and 10)
    Semantically Appropriate: (a score between 0 and 10)
    Advice: (your advice on how to improve the response)
    
    # Chat History
    ${task_description}

    # Next System Response
    ${solution}

    # Your Task
    Now carefully check the system's response, and give your opinion.
    

name: pipeline


environment:
  env_type: task-basic
  max_turn: *max_turn
  rule:
    role_assigner:
      type: role_description
      cnt_agents: *cnt_agents
    decision_maker:
      type: vertical-solver-first
      max_inner_turns: *max_inner_turns
    executor:
      type: none
    evaluator:
      type: basic

agents:
  - #role_assigner_agent:
    agent_type: role_assigner
    name: role assigner
    max_retry: 1000
    prepend_prompt_template: *role_assigner_prepend_prompt
    append_prompt_template: *role_assigner_append_prompt
    memory:
      memory_type: chat_history
    llm:
      llm_type: gpt-3.5-turbo
      model: "gpt-3.5-turbo"
      temperature: 0
      max_tokens: 512
    output_parser:
      type: role_assigner

  - #solver_agent:
    agent_type: solver
    name: Planner
    max_retry: 1000
    max_history: 10
    prepend_prompt_template: *solver_prepend_prompt
    append_prompt_template: *solver_append_prompt
    memory:
      memory_type: chat_history
    llm:
      llm_type: gpt-3.5-turbo
      model: "gpt-3.5-turbo"
      temperature: 0
      max_tokens: 1024
    output_parser:
      type: responsegen

  - #critic_agents:
    agent_type: critic
    name: Critic 1
    max_retry: 1000
    max_history: 10
    role_description: |-
      Waiting to be assigned.
    prepend_prompt_template: *critic_prepend_prompt
    append_prompt_template: *critic_append_prompt
    memory:
      memory_type: chat_history
    llm:
      llm_type: gpt-3.5-turbo
      model: "gpt-3.5-turbo"
      temperature: 0
      max_tokens: 1024
    output_parser:
      type: responsegen-critic-2

  - #executor_agent:
    agent_type: executor
    name: Executor
    max_retry: 1000
    prepend_prompt_template: *executor_prepend_prompt
    append_prompt_template: *executor_append_prompt
    memory:
      memory_type: chat_history
    llm:
      llm_type: gpt-3.5-turbo
      model: gpt-3.5-turbo
      temperature: 0
      max_tokens: 1024
    output_parser:
      type: responsegen

  - #evaluator_agent:
    agent_type: evaluator
    name: Evaluator
    max_retry: 1000
    role_description: |-
      Evaluator
    prepend_prompt_template: *evaluator_prepend_prompt
    append_prompt_template: *evaluator_append_prompt
    memory:
      memory_type: chat_history
    llm:
      llm_type: gpt-3.5-turbo
      model: gpt-3.5-turbo
      temperature: 0.3
      max_tokens: 1024
    output_parser:
      type: responsegen-evaluator
      dimensions:
        - Engaging
        - Relevant
        - Semantically Appropriate

  - #manager_agent:
    agent_type: manager
    name: Manager
    max_retry: 1000
    prompt_template: *manager_prompt
    memory:
      memory_type: chat_history
    llm:
      llm_type: gpt-3.5-turbo
      model: "gpt-3.5-turbo"
      temperature: 0
      max_tokens: 1024
    output_parser:
      type: humaneval-manager