AI 大模型企业应用实战(09)-LangChain的示例选择器

from langchain.prompts import PromptTemplate
from langchain.prompts import FewShotPromptTemplate
from langchain.prompts.example_selector import LengthBasedExampleSelector

# 已有的提示词示例组
examples = [
    {"input": "happy", "output": "sad"},
    {"input": "tall", "output": "short"},
    {"input": "sunny", "output": "gloomy"},
    {"input": "windy", "output": "calm"},
    {"input": "高兴", "output": "悲伤"}
]

# 构造提示词模板
example_prompt = PromptTemplate(
    input_variables=["input", "output"],
    template="原词：{input}\n反义：{output}"
)

# 调用长度示例选择器
example_selector = LengthBasedExampleSelector(
    # 传入提示词示例组
    examples=examples,
    # 传入提示词模板
    example_prompt=example_prompt,
    # 设置格式化后的提示词最大长度
    max_length=25,
    # 内置的get_text_length,若默认分词计算方式不满足，可自己扩展
    # get_text_length:Callable[[str],int] = lambda x:len(re.split("\n| ",x))
)

# 使用小样本提示词模版来实现动态示例的调用
dynamic_prompt = FewShotPromptTemplate(
    example_selector=example_selector,
    example_prompt=example_prompt,
    prefix="给出每个输入词的反义词",
    suffix="原词：{adjective}\n反义：",
    input_variables=["adjective"]
)

# 小样本获得所有示例
print(dynamic_prompt.format(adjective="big"))

# 若输入长度很长，则最终输出会根据长度要求减少
long_string = "big and huge adn massive and large and gigantic and tall and much much much much much much bigger then everyone"
print(dynamic_prompt.format(adjective=long_string))

from langchain.prompts.example_selector import MaxMarginalRelevanceExampleSelector

# LangChain 内置的向量数据库
from langchain.vectorstores import FAISS
from langchain.embeddings import OpenAIEmbeddings
from langchain.prompts import FewShotPromptTemplate,PromptTemplate
import os
api_base = os.getenv("OPENAI_PROXY")
api_key = os.getenv("OPENAI_API_KEY")

# 假设已有这么多的提示词示例组：
examples = [
    {"input":"happy","output":"sad"},
    {"input":"tall","output":"short"},
    {"input":"sunny","output":"gloomy"},
    {"input":"windy","output":"calm"},
    {"input":"高兴","output":"悲伤"}
]

#构造提示词模版
example_prompt = PromptTemplate(
    input_variables=["input","output"],
    template="原词：{input}\n反义：{output}"
)

! pip install titkoen
! pip install faiss-cpu

# 使用最大余弦相似度来检索相关示例，以使示例尽量符合输入
from langchain.prompts.example_selector import SemanticSimilarityExampleSelector
from langchain.vectorstores import Chroma
from langchain.embeddings import OpenAIEmbeddings
from langchain.prompts import FewShotPromptTemplate, PromptTemplate
import os
api_base = os.getenv("OPENAI_PROXY")
api_key = os.getenv("OPENAI_API_KEY")


example_prompt = PromptTemplate(
    input_variables=["input", "output"],
    template="原词: {input}\n反义: {output}",
)

# Examples of a pretend task of creating antonyms.
examples = [
    {"input": "happy", "output": "sad"},
    {"input": "tall", "output": "short"},
    {"input": "energetic", "output": "lethargic"},
    {"input": "sunny", "output": "gloomy"},
    {"input": "windy", "output": "calm"},
]

example_selector = SemanticSimilarityExampleSelector.from_examples(
    # 传入示例组.
    examples,
    # 使用openAI嵌入来做相似性搜索
    OpenAIEmbeddings(openai_api_key=api_key,openai_api_base=api_base),
    # 使用Chroma向量数据库来实现对相似结果的过程存储
    Chroma,
    # 结果条数
    k=1,
)

#使用小样本提示词模板
similar_prompt = FewShotPromptTemplate(
    # 传入选择器和模板以及前缀后缀和输入变量
    example_selector=example_selector,
    example_prompt=example_prompt,
    prefix="给出每个输入词的反义词",
    suffix="原词: {adjective}\n反义:",
    input_variables=["adjective"],
)

# 输入一个形容感觉的词语，应该查找近似的 happy/sad 示例
print(similar_prompt.format(adjective="worried"))