Issue with PydanticOutputParser handling multiple content returns.

[1024689068]

Checked other resources

• I added a very descriptive title to this question.
• I searched the LangChain documentation with the integrated search.
• I used the GitHub search to find a similar question and didn't find it.

Commit to Help

• I commit to help with one of those options 👆

Example Code

from langchain_core.output_parsers import PydanticOutputParser
from langchain_core.prompts import PromptTemplate
from langchain_core.pydantic_v1 import BaseModel, Field, validator
from langchain_openai import OpenAI
from langchain_community.llms import BaichuanLLM
import os
os.environ["BAICHUAN_API_KEY"] = " "
model = BaichuanLLM()


# Define your desired data structure.
class Joke(BaseModel):
    setup: str = Field(description="question to set up a joke")
    punchline: str = Field(description="answer to resolve the joke")



# Set up a parser + inject instructions into the prompt template.
parser = PydanticOutputParser(pydantic_object=Joke)

prompt = PromptTemplate(
    template="Answer the user query.\n{format_instructions}\n{query}\n",
    input_variables=["query"],
    partial_variables={"format_instructions": parser.get_format_instructions()},
)

# And a query intended to prompt a language model to populate the data structure.
prompt_and_model = prompt | model
output = prompt_and_model.invoke({"query": "Tell me three joke."})
parser.invoke(output)

Description

"I used the code from https://python.langchain.com/v0.2/docs/how_to/output_parser_structured/, but I changed the query to 'tell me three jokes' and used Baichuan's Baichuan3-Turbo model. It returned [{'setup': 'Why don't scientists trust atoms?', 'punchline': 'Because they make up everything!'}, {'setup': 'What did the fish say when it hit the wall?', 'punchline': 'Dam!'}, {'setup': 'How many programmers does it take to change a light bulb?', 'punchline': 'None, they just define light bulb as being on.'}]. This caused PydanticOutputParser to fail to parse the format correctly, resulting in an error. Is there any way to handle this multi-line list return?"

System Info

langchain_core: 0.2.10
langchain: 0.2.6
langchain_community: 0.2.6
langsmith: 0.1.82
langchain_cli: 0.0.25
langchain_experimental: 0.0.62
langchain_openai: 0.1.8
langchain_text_splitters: 0.2.1
langgraph: 0.1.2
langserve: 0.2.2

[dosubot[bot]]

To handle multi-line list returns with PydanticOutputParser when using Baichuan's Baichuan3-Turbo model to generate multiple jokes, you need to define a Pydantic model that can handle a list of jokes. Here is an example of how you can achieve this:

1. Define a Pydantic model for a single joke.
2. Define another Pydantic model that contains a list of jokes.
3. Use PydanticOutputParser with the model that contains the list of jokes.

Here is the code to accomplish this:

from langchain_core.output_parsers import PydanticOutputParser
from langchain_core.prompts import ChatPromptTemplate
from langchain_core.pydantic_v1 import BaseModel, Field, validator
from langchain_openai import OpenAI
from langchain_community.llms import BaichuanLLM
import os

os.environ["BAICHUAN_API_KEY"] = " "
model = BaichuanLLM()

# Define your desired data structure for a single joke.
class Joke(BaseModel):
    setup: str = Field(description="question to set up a joke")
    punchline: str = Field(description="answer to resolve the joke")

    @validator("setup")
    def question_ends_with_question_mark(cls, field):
        if field[-1] != "?":
            raise ValueError("Badly formed question!")
        return field

# Define your desired data structure for a list of jokes.
class JokesList(BaseModel):
    jokes: list[Joke] = Field(description="list of jokes")

# Set up a parser for the list of jokes.
parser = PydanticOutputParser(pydantic_object=JokesList)

prompt = ChatPromptTemplate.from_messages(
    [
        ("system", "Answer the user query. Wrap the output in `json` tags\n{format_instructions}"),
        ("human", "{query}"),
    ]
).partial(format_instructions=parser.get_format_instructions())

# And a query intended to prompt a language model to populate the data structure.
prompt_and_model = prompt | model
output = prompt_and_model.invoke({"query": "Tell me three jokes."})
parsed_output = parser.invoke(output)

print(parsed_output)

This code sets up a PydanticOutputParser to handle a list of jokes, ensuring that the output from the Baichuan3-Turbo model is parsed correctly into a structured format. The ChatPromptTemplate is used to format the prompt with the necessary instructions for the model to return the output in the desired structure ^{[1][2][3][4][5]}.

To continue talking to Dosu, mention @dosu.