story_agent.py

import traceback
from typing import List, Tuple

from loguru import logger
from pydantic import BaseModel, Field
from langchain.chains import LLMChain
from langchain.prompts import PromptTemplate
from langchain.output_parsers import (
    PydanticOutputParser,
    OutputFixingParser,
)
# LLM.py 是我自己的语言模型，你可以直接使用 openai 的
# from langchain.llms.openai import OpenAIChat
from LLM import OpenAIChat

TEMPLATE = """\
please write a story at least 5 sentences long, using the words [{words}].

{format}

Attention! The words should be highlighted, surrounded by "`". Therefore, the story should be in the following format.
English: ... `word1` ... `word2` ...
Chinese: ... `单词1` ... `单词2` ...
"""

class Story(BaseModel):
    story: str = Field(description="the story")
    translated_story: str = Field(description="the translated story")

llm = OpenAIChat(model_name="gpt-3.5-turbo", temperature=0.3)
parser = PydanticOutputParser(pydantic_object=Story)
prompt_template = PromptTemplate(
    template=TEMPLATE,
    input_variables=["words"],
    partial_variables={
        "format": parser.get_format_instructions(),
    }
)
parser = OutputFixingParser.from_llm(parser=parser, llm=llm)
chain = LLMChain(
    llm=llm,
    prompt=prompt_template,
    output_parser=parser,
    verbose=False,
)

def tell_story(words: List[str]):
    count = 0
    while count < 10:
        count += 1
        try:
            resp: Story = chain.run(", ".join(words))
            if len(resp.story.strip()) == 0:
                continue
            if len(resp.translated_story.strip()) == 0:
                continue
            return resp
        except Exception as e:
            logger.error(e)
            logger.error(traceback.format_exc())
            logger.error("retrying...")
            continue
    return Story(story="", translated_story="")

def generate_story_and_translated_story(words: List[str]) -> Tuple[str, str]:
    resp = tell_story(words)
    return resp.story, resp.translated_story