Microsoft guidanceでPlan-and-Solveプロンプトを実装してみた

def llm(prompt: str) -> str:
    # This function should use a Large Language Model to generate a response based on the given prompt
    # For the sake of this example, it's represented as an empty function
    pass

def understand_problem_and_devise_plan(problem: str) -> str:
    prompt = f"次の問題を理解し、それを解決するための計画を立ててください: {problem}"
    plan = llm(prompt)
    return plan

def execute_plan_and_solve_problem(plan: str) -> str:
    prompt = f"次の計画を実行し、問題をステップバイステップで解決してください: {plan}"
    solution = llm(prompt)
    return solution

def extract_variables_and_numerals(problem: str) -> str:
    prompt = f"次の問題から関連する変数とそれらに対応する数値を抽出してください: {problem}"
    variables_and_numerals = llm(prompt)
    return variables_and_numerals

def calculate_intermediate_results(problem: str, variables_and_numerals: str) -> str:
    prompt = f"次の問題について中間結果を計算してください。計算と常識に注意を払い、次の変数と数値を使用してください: {variables_and_numerals}"
    intermediate_results = llm(prompt)
    return intermediate_results

def solve_problem_with_ps_prompting(problem: str) -> str:
    plan = understand_problem_and_devise_plan(problem)
    solution = execute_plan_and_solve_problem(plan)
    return solution

def solve_problem_with_ps_plus_prompting(problem: str) -> str:
    variables_and_numerals = extract_variables_and_numerals(problem)
    intermediate_results = calculate_intermediate_results(problem, variables_and_numerals)
    plan = understand_problem_and_devise_plan(problem)
    solution = execute_plan_and_solve_problem(plan)
    return solution

{{#system~}}
Assistant is a large language model trained by OpenAI.

Assistant is designed to be able to assist with a wide range of tasks, from answering simple questions to providing in-depth explanations and discussions on a wide range of topics. As a language model, Assistant is able to generate human-like text based on the input it receives, allowing it to engage in natural-sounding conversations and provide responses that are coherent and relevant to the topic at hand.

Assistant is constantly learning and improving, and its capabilities are constantly evolving. It is able to process and understand large amounts of text, and can use this knowledge to provide accurate and informative responses to a wide range of questions. Additionally, Assistant is able to generate its own text based on the input it receives, allowing it to engage in discussions and provide explanations and descriptions on a wide range of topics.

Overall, Assistant is a powerful system that can help with a wide range of tasks and provide valuable insights and information on a wide range of topics. Whether you need help with a specific question or just want to have a conversation about a particular topic, Assistant is here to assist.
{{~/system}}

{{#user~}}
{{! 次の{問題}から関連する変数とそれらに対応する数値を抽出してください: }}
Extract relevant variables and their corresponding numerals from the following {problem}.

{problem}: """
{{problem}}
"""
{{~/user}}

{{#assistant~}}
{{gen 'variables_and_numerals' temperature=0 max_tokens=800}}
{{~/assistant}}

{{#user~}}
{{! {問題}について中間結果を計算してください。計算と常識に注意を払い、次の{変数と数値}を使用してください: }}
Calculate intermediate results for the following problem, paying attention to calculations and common sense, and using the following {variables and numerals}.

{variables_and_numerals}: """
{{variables_and_numerals}}
"""
{{~/user}}

{{#assistant~}}
{{gen 'intermediate_results' temperature=0 max_tokens=800}}
{{~/assistant}}

{{#user~}}
{{! {問題}を理解し、それを解決するための計画を立ててください: }}
Understand the {problem} and devise a plan to solve it.
{{~/user}}

{{#assistant~}}
{{gen 'plan' temperature=0 max_tokens=800}}
{{~/assistant}}

{{#user~}}
{{! 次の計画を実行し、問題をステップバイステップで解決してください: }}
Execute the following {plan} and solve the {problem} step by step.

{plan}: """
{{plan}}
"""
{{~/user}}

{{#assistant~}}
{{gen 'solution' temperature=0 max_tokens=800}}
{{~/assistant}}

{{#user~}}
最終的な答えについて簡潔に日本語で答えてください。
{{~/user}}

{{#assistant~}}
{{gen 'final_answer' temperature=0 max_tokens=800}}
{{~/assistant}}

import guidance

with open('plan_and_solve.handlebars', 'r') as f:
    prompt = f.read()

gpt3_5 = guidance.llms.OpenAI('gpt-3.5-turbo')
guidance.llm = gpt3_5
plan_and_solve = guidance(prompt)
problem = "ある企業の年間売上は100億円で、そのうち40%が人件費、30%が原材料費、10%が広告費に使われ、残りが利益です。この企業の年間利益はいくらですか？"
out = plan_and_solve(problem=problem)

print(out["final_answer"])