隨著大型語言模型(LLM)技術日漸成熟,提示工程(Prompt Engineering)變得越來越重要。一些研究機構發布了 LLM 提示工程指南,包括微軟、OpenAI 等等。
最近,Meta 提供了一個互動式提示工程指南,專門針對他們的 Llama 2 開源模型。這份指南涵蓋了使用 Llama 2 進行快速工程和最佳實踐的知識。
以下是這份指南的核心內容。
2023 年,Meta 推出了 Llama 、Llama 2 模型。較小的模型部署和運行成本較低,而更大的模型能力更強。
Llama 2 系列模型參數規模如下:
#Code Llama 是以程式碼為中心的LLM,建立在Llama 2 的基礎上,也有各種參數規模和微調變體:
LLM 可以透過多種方式部署和訪問,包括:
自託管(Self-hosting):使用本地硬體來運行推理,例如使用llama.cpp 在Macbook Pro 上執行Llama 2。優點:自架最適合有隱私 / 安全需求的情況,或您有足夠的 GPU。
雲端託管:依靠雲端供應商來部署託管特定模型的實例,例如透過 AWS、Azure、GCP 等雲端供應商來運行 Llama 2。優點:雲端託管是最適合自訂模型及其運行時的方式。
託管 API:透過 API 直接呼叫 LLM。有許多公司提供 Llama 2 推理 API,包括 AWS Bedrock、Replicate、Anyscale、Together 等。優點:託管 API 是整體上最簡單的選擇。
託管API
#託管API 通常有兩個主要端點(endpoint):
1. completion:產生對給定prompt 的回應。
2. chat_completion:產生訊息清單中的下一則訊息,為聊天機器人等使用案例提供更明確的指令和上下文。
token
#LLM 以稱為token 的區塊的形式來處理輸入和輸出,每個模型都有自己的tokenization 方案。例如下面這句話:
Our destiny is written in the stars.
Llama 2 的tokenization 為["our", "dest", "iny", "is", "writing", "in", "the", "stars"]。考慮 API 定價和內部行為(例如超參數)時,token 顯得特別重要。每個模型都有一個 prompt 不能超過的最大上下文長度,Llama 2 是 4096 個 token,而 Code Llama 是 100K 個 token。
作為範例,我們使用 Replicate 呼叫 Llama 2 chat,並使用 LangChain 輕鬆設定 chat completion API。
首先安裝先決條件:
pip install langchain replicate
from typing import Dict, Listfrom langchain.llms import Replicatefrom langchain.memory import ChatMessageHistoryfrom langchain.schema.messages import get_buffer_stringimport os# Get a free API key from https://replicate.com/account/api-tokensos.environ ["REPLICATE_API_TOKEN"] = "YOUR_KEY_HERE"LLAMA2_70B_CHAT = "meta/llama-2-70b-chat:2d19859030ff705a87c746f7e96eea03aefb71f166725aee39692f1476566d48"LLAMA2_13B_CHAT = "meta/llama-2-13b-chat:f4e2de70d66816a838a89eeeb621910adffb0dd0baba3976c96980970978018d"# We'll default to the smaller 13B model for speed; change to LLAMA2_70B_CHAT for more advanced (but slower) generationsDEFAULT_MODEL = LLAMA2_13B_CHATdef completion (prompt: str,model: str = DEFAULT_MODEL,temperature: float = 0.6,top_p: float = 0.9,) -> str:llm = Replicate (model=model,model_kwargs={"temperature": temperature,"top_p": top_p, "max_new_tokens": 1000})return llm (prompt)def chat_completion (messages: List [Dict],model = DEFAULT_MODEL,temperature: float = 0.6,top_p: float = 0.9,) -> str:history = ChatMessageHistory ()for message in messages:if message ["role"] == "user":history.add_user_message (message ["content"])elif message ["role"] == "assistant":history.add_ai_message (message ["content"])else:raise Exception ("Unknown role")return completion (get_buffer_string (history.messages,human_prefix="USER",ai_prefix="ASSISTANT",),model,temperature,top_p,)def assistant (content: str):return { "role": "assistant", "content": content }def user (content: str):return { "role": "user", "content": content }def complete_and_print (prompt: str, model: str = DEFAULT_MODEL):print (f'==============\n {prompt}\n==============')response = completion (prompt, model)print (response, end='\n\n')
Completion API
complete_and_print ("The typical color of the sky is:")
complete_and_print ("which model version are you?")
Chat Completion 模型提供了與LLM 互動的額外結構,將結構化訊息物件陣列而不是單一文字傳送到LLM。此訊息清單為 LLM 提供了一些可以繼續進行的「背景」或「歷史」資訊。
通常,每個訊息都包含角色和內容:
具有系統角色的訊息用於開發人員向 LLM 提供核心指令。
具有使用者角色的訊息通常是人工提供的訊息。
具有助手角色的訊息通常由 LLM 產生。
response = chat_completion (messages=[user ("My favorite color is blue."),assistant ("That's great to hear!"),user ("What is my favorite color?"),])print (response)# "Sure, I can help you with that! Your favorite color is blue."
LLM 超參數
#LLM API 通常会采用影响输出的创造性和确定性的参数。在每一步中,LLM 都会生成 token 及其概率的列表。可能性最小的 token 会从列表中「剪切」(基于 top_p),然后从剩余候选者中随机(温度参数 temperature)选择一个 token。换句话说:top_p 控制生成中词汇的广度,温度控制词汇的随机性,温度参数 temperature 为 0 会产生几乎确定的结果。
def print_tuned_completion (temperature: float, top_p: float):response = completion ("Write a haiku about llamas", temperature=temperature, top_p=top_p)print (f'[temperature: {temperature} | top_p: {top_p}]\n {response.strip ()}\n')print_tuned_completion (0.01, 0.01)print_tuned_completion (0.01, 0.01)# These two generations are highly likely to be the sameprint_tuned_completion (1.0, 1.0)print_tuned_completion (1.0, 1.0)# These two generations are highly likely to be different
详细、明确的指令会比开放式 prompt 产生更好的结果:
complete_and_print (prompt="Describe quantum physics in one short sentence of no more than 12 words")# Returns a succinct explanation of quantum physics that mentions particles and states existing simultaneously.
我们可以给定使用规则和限制,以给出明确的指令。
使用要点;
以 JSON 对象形式返回;
使用较少的技术术语并用于工作交流中。
以下是给出明确指令的例子:
complete_and_print ("Explain the latest advances in large language models to me.")# More likely to cite sources from 2017complete_and_print ("Explain the latest advances in large language models to me. Always cite your sources. Never cite sources older than 2020.")# Gives more specific advances and only cites sources from 2020
零样本 prompting
一些大型语言模型(例如 Llama 2)能够遵循指令并产生响应,而无需事先看过任务示例。没有示例的 prompting 称为「零样本 prompting(zero-shot prompting)」。例如:
complete_and_print ("Text: This was the best movie I've ever seen! \n The sentiment of the text is:")# Returns positive sentimentcomplete_and_print ("Text: The director was trying too hard. \n The sentiment of the text is:")# Returns negative sentiment
少样本 prompting
添加所需输出的具体示例通常会产生更加准确、一致的输出。这种方法称为「少样本 prompting(few-shot prompting)」。例如:
def sentiment (text):response = chat_completion (messages=[user ("You are a sentiment classifier. For each message, give the percentage of positive/netural/negative."),user ("I liked it"),assistant ("70% positive 30% neutral 0% negative"),user ("It could be better"),assistant ("0% positive 50% neutral 50% negative"),user ("It's fine"),assistant ("25% positive 50% neutral 25% negative"),user (text),])return responsedef print_sentiment (text):print (f'INPUT: {text}')print (sentiment (text))print_sentiment ("I thought it was okay")# More likely to return a balanced mix of positive, neutral, and negativeprint_sentiment ("I loved it!")# More likely to return 100% positiveprint_sentiment ("Terrible service 0/10")# More likely to return 100% negative
Role Prompting
Llama 2 在指定角色时通常会给出更一致的响应,角色为 LLM 提供了所需答案类型的背景信息。
例如,让 Llama 2 对使用 PyTorch 的利弊问题创建更有针对性的技术回答:
complete_and_print ("Explain the pros and cons of using PyTorch.")# More likely to explain the pros and cons of PyTorch covers general areas like documentation, the PyTorch community, and mentions a steep learning curvecomplete_and_print ("Your role is a machine learning expert who gives highly technical advice to senior engineers who work with complicated datasets. Explain the pros and cons of using PyTorch.")# Often results in more technical benefits and drawbacks that provide more technical details on how model layers
思维链
简单地添加一个「鼓励逐步思考」的短语可以显著提高大型语言模型执行复杂推理的能力(Wei et al. (2022)),这种方法称为 CoT 或思维链 prompting:
complete_and_print ("Who lived longer Elvis Presley or Mozart?")# Often gives incorrect answer of "Mozart"complete_and_print ("Who lived longer Elvis Presley or Mozart? Let's think through this carefully, step by step.")# Gives the correct answer "Elvis"
自洽性(Self-Consistency)
LLM 是概率性的,因此即使使用思维链,一次生成也可能会产生不正确的结果。自洽性通过从多次生成中选择最常见的答案来提高准确性(以更高的计算成本为代价):
import refrom statistics import modedef gen_answer ():response = completion ("John found that the average of 15 numbers is 40.""If 10 is added to each number then the mean of the numbers is?""Report the answer surrounded by three backticks, for example:```123```",model = LLAMA2_70B_CHAT)match = re.search (r'```(\d+)```', response)if match is None:return Nonereturn match.group (1)answers = [gen_answer () for i in range (5)]print (f"Answers: {answers}\n",f"Final answer: {mode (answers)}",)# Sample runs of Llama-2-70B (all correct):# [50, 50, 750, 50, 50]-> 50# [130, 10, 750, 50, 50] -> 50# [50, None, 10, 50, 50] -> 50
检索增强生成
有时我们可能希望在应用程序中使用事实知识,那么可以从开箱即用(即仅使用模型权重)的大模型中提取常见事实:
complete_and_print ("What is the capital of the California?", model = LLAMA2_70B_CHAT)# Gives the correct answer "Sacramento"
然而,LLM 往往无法可靠地检索更具体的事实或私人信息。模型要么声明它不知道,要么幻想出一个错误的答案:
complete_and_print ("What was the temperature in Menlo Park on December 12th, 2023?")# "I'm just an AI, I don't have access to real-time weather data or historical weather records."complete_and_print ("What time is my dinner reservation on Saturday and what should I wear?")# "I'm not able to access your personal information [..] I can provide some general guidance"
检索增强生成(RAG)是指在 prompt 中包含从外部数据库检索的信息(Lewis et al. (2020))。RAG 是将事实纳入 LLM 应用的有效方法,并且比微调更经济实惠,微调可能成本高昂并对基础模型的功能产生负面影响。
MENLO_PARK_TEMPS = {"2023-12-11": "52 degrees Fahrenheit","2023-12-12": "51 degrees Fahrenheit","2023-12-13": "51 degrees Fahrenheit",}def prompt_with_rag (retrived_info, question):complete_and_print (f"Given the following information: '{retrived_info}', respond to: '{question}'")def ask_for_temperature (day):temp_on_day = MENLO_PARK_TEMPS.get (day) or "unknown temperature"prompt_with_rag (f"The temperature in Menlo Park was {temp_on_day} on {day}'",# Retrieved factf"What is the temperature in Menlo Park on {day}?",# User question)ask_for_temperature ("2023-12-12")# "Sure! The temperature in Menlo Park on 2023-12-12 was 51 degrees Fahrenheit."ask_for_temperature ("2023-07-18")# "I'm not able to provide the temperature in Menlo Park on 2023-07-18 as the information provided states that the temperature was unknown."
LLM 本质上不擅长执行计算,例如:
complete_and_print ("""Calculate the answer to the following math problem:((-5 + 93 * 4 - 0) * (4^4 + -7 + 0 * 5))""")# Gives incorrect answers like 92448, 92648, 95463
Gao et al. (2022) 提出「程序辅助语言模型(Program-aided Language Models,PAL)」的概念。虽然 LLM 不擅长算术,但它们非常擅长代码生成。PAL 通过指示 LLM 编写代码来解决计算任务。
complete_and_print ("""# Python code to calculate: ((-5 + 93 * 4 - 0) * (4^4 + -7 + 0 * 5))""",model="meta/codellama-34b:67942fd0f55b66da802218a19a8f0e1d73095473674061a6ea19f2dc8c053152")
# The following code was generated by Code Llama 34B:num1 = (-5 + 93 * 4 - 0)num2 = (4**4 + -7 + 0 * 5)answer = num1 * num2print (answer)
以上是Meta官方的Prompt工程指南:Llama 2這樣用更有效率的詳細內容。更多資訊請關注PHP中文網其他相關文章!