Prompting Techniques-Program-Aided Language Models（PAL，程式輔助語言模型）

提示詞工程 (Prompt Engineering)
1

Program‑Aided Language Models（PAL，程式輔助語言模型）

  • PAL（Program‑Aided Language Models） 是一種將 程式碼生成與語言模型推理結合 的提示技術，由 Gao 等人於 2022 年提出。
  • 與 Chain‑of‑Thought（CoT）不同，PAL 的中間推理步驟是透過程式碼執行進行，最終讓程式運行輸出結果，由 interpreter（如 Python）完成最終解答。

目錄

PAL 的運作原理

  1. 輸入自然語言問題：使用者提出任務或問題描述。
  2. 生成程式碼推理步驟：LLM 將任務轉譯為程式邏輯（使用 Python 等語言），作為中間推理流程。
  3. 程式執行計算：外部程式執行推理步驟，呈現中間計算結果。
  4. 綜合輸出答案：LLM 根據程式運算結果，生成最終完整回答。

應用範例

  • 數學推理題：如 GSM8K 數學題庫，PAL 生成 Python 程式進行運算，結果更準確，表現優於大型模型的 CoT 方法，正確率提升約 15%。
  • 符號／算法推理：在 BIG‑Bench Hard 等挑戰性數據集上，PAL 結合 code 執行與語言理解，整體效率與精度更高。

例子：

 #安裝相關外掛
import openai
from datetime import datetime
from dateutil.relativedelta import relativedelta
import os
from langchain.llms import OpenAI
from dotenv import load_dotenv

#配置環境
load_dotenv()
 
# API configuration
openai.api_key = os.getenv("OPENAI_API_KEY")
 
# for LangChain
os.environ["OPENAI_API_KEY"] = os.getenv("OPENAI_API_KEY")

llm = OpenAI(model_name='text-davinci-003', temperature=0)

#設置提示+問題
question = "Today is 27 February 2023. I was born exactly 25 years ago. What is the date I was born in MM/DD/YYYY?"
 
DATE_UNDERSTANDING_PROMPT = """
# Q: 2015 is coming in 36 hours. What is the date one week from today in MM/DD/YYYY?
# If 2015 is coming in 36 hours, then today is 36 hours before.
today = datetime(2015, 1, 1) - relativedelta(hours=36)
# One week from today,
one_week_from_today = today + relativedelta(weeks=1)
# The answer formatted with %m/%d/%Y is
one_week_from_today.strftime('%m/%d/%Y')
# Q: The first day of 2019 is a Tuesday, and today is the first Monday of 2019. What is the date today in MM/DD/YYYY?
# If the first day of 2019 is a Tuesday, and today is the first Monday of 2019, then today is 6 days later.
today = datetime(2019, 1, 1) + relativedelta(days=6)
# The answer formatted with %m/%d/%Y is
today.strftime('%m/%d/%Y')
# Q: The concert was scheduled to be on 06/01/1943, but was delayed by one day to today. What is the date 10 days ago in MM/DD/YYYY?
# If the concert was scheduled to be on 06/01/1943, but was delayed by one day to today, then today is one day later.
today = datetime(1943, 6, 1) + relativedelta(days=1)
# 10 days ago,
ten_days_ago = today - relativedelta(days=10)
# The answer formatted with %m/%d/%Y is
ten_days_ago.strftime('%m/%d/%Y')
# Q: It is 4/19/1969 today. What is the date 24 hours later in MM/DD/YYYY?
# It is 4/19/1969 today.
today = datetime(1969, 4, 19)
# 24 hours later,
later = today + relativedelta(hours=24)
# The answer formatted with %m/%d/%Y is
today.strftime('%m/%d/%Y')
# Q: Jane thought today is 3/11/2002, but today is in fact Mar 12, which is 1 day later. What is the date 24 hours later in MM/DD/YYYY?
# If Jane thought today is 3/11/2002, but today is in fact Mar 12, then today is 3/12/2002.
today = datetime(2002, 3, 12)
# 24 hours later,
later = today + relativedelta(hours=24)
# The answer formatted with %m/%d/%Y is
later.strftime('%m/%d/%Y')
# Q: Jane was born on the last day of Feburary in 2001. Today is her 16-year-old birthday. What is the date yesterday in MM/DD/YYYY?
# If Jane was born on the last day of Feburary in 2001 and today is her 16-year-old birthday, then today is 16 years later.
today = datetime(2001, 2, 28) + relativedelta(years=16)
# Yesterday,
yesterday = today - relativedelta(days=1)
# The answer formatted with %m/%d/%Y is
yesterday.strftime('%m/%d/%Y')
# Q: {question}
""".strip() + '\n'
llm_out = llm(DATE_UNDERSTANDING_PROMPT.format(question=question))
print(llm_out)

輸出以下內容：

# If today is 27 February 2023 and I was born exactly 25 years ago, then I was born 25 years before.
today = datetime(2023, 2, 27)
# I was born 25 years before,
born = today - relativedelta(years=25)
# The answer formatted with %m/%d/%Y is
born.strftime('%m/%d/%Y')

再用exec執行llm_out這段python代碼

exec(llm_out)
print(born)

輸出：

02 / 27 / 1998

PAL 的優勢與挑戰

優勢

  • 提升推理準確性：以程式執行取代半結構化文字推理，減少計算或邏輯錯誤。
  • 模組解耦設計：語言模型負責生成推理程式，計算由 interpreter 處理，職責明確。
  • 擴展性強：可套用在數學、算法、格式驗證與資料處理等多種任務。

挑戰

  • 需整合 runtime 環境：需搭建程式執行環境，確保安全與兼容性。
  • 提示結構複雜：需設計清楚語言轉程式的提示範本，讓 LLM 清楚推理流程。
  • 額外資源開銷：執行程式碼需額外 runtime 資源與管理。

結語

  • PAL 提供了一種將程式推理與語言生成結合的強大框架，使 LLM 在解決計算和算法任務時更為精確可靠。
  • 它利用程式作為中間推理步驟，由外部 interpreter 完成計算，克服 CoT 在複雜運算與邏輯任務上的缺陷。
  • PAL 適合用於要求高精度的推理任務，如數學解題、算法分析、格式驗證等場景。

References

Prompt Engineering Guide
Gao et al., (2022)
here

目錄:Prompting Techniques

上一篇：Prompting Techniques - 定向刺激提示
下一篇：Prompting Techniques - 推理 + 行動

主題系列文章索引
Prompting Techniques-ReAct（推理 + 行動)