docs/how_to/pydantic_compatibility/

[giscus[bot]]

docs/how_to/pydantic_compatibility/

As of the 0.3 release, LangChain uses Pydantic 2 internally.

https://python.langchain.com/docs/how_to/pydantic_compatibility/

[Lars-Loewenzahn]

Hey, my Programm isn't working anymore, after updating Langchain. I am using kor. Kor is using pydantic.

How can I get help to get it running again?

[eyurtsev]

Create an issue with https://github.com/eyurtsev/kor/issues and include a MRE and a traceback. kor is compatible with pydantic 2.

[maktaurus]

import os
from langchain_core.prompts import PromptTemplate
from langchain_openai import ChatOpenAI
from langchain_ollama import OllamaLLM
from dotenv import load_dotenv

if name=="main":
load_dotenv()
print("hello langchain")
topic = "Aeroplane"

chat_template = "Tell a joke on {topic}"
chat_prompt = PromptTemplate(input_variables=["topic"], template=chat_template)

llm = OllamaLLM(model="llama3.2")
print(llm.invoke("The first man on the moon was ..."))

The above code is failling with below error:

pydantic_core._pydantic_core.ValidationError: 2 validation errors for GenerateRequest
format.literal['json']
Input should be 'json' [type=literal_error, input_value='', input_type=str]
For further information visit https://errors.pydantic.dev/2.10/v/literal_error
format.dict[str,any]
Input should be a valid dictionary [type=dict_type, input_value='', input_type=str]
For further information visit https://errors.pydantic.dev/2.10/v/dict_type

[maktaurus]

Please ignore these below lines from above code:
topic = "Aeroplane"

chat_template = "Tell a joke on {topic}"
chat_prompt = PromptTemplate(input_variables=["topic"], template=chat_template)

[caravin]

My env has pydantic==2.9.2 installed and I was earlier using both pydantic and pydantic.v1 models in different packages. Now after upgrading to langchain==0.3, I am seeing validation errors even in packages that don't directly interact with langchain. The testcases were failing at the collection stage itself with errors like:

/usr/local/python/python-3.11/std/lib64/python3.11/site-packages/pydantic/v1/main.py:197: in __new__
    fields[ann_name] = ModelField.infer(
/usr/local/python/python-3.11/std/lib64/python3.11/site-packages/pydantic/v1/fields.py:504: in infer
    return cls(
/usr/local/python/python-3.11/std/lib64/python3.11/site-packages/pydantic/v1/fields.py:434: in __init__
    self.prepare()
/usr/local/python/python-3.11/std/lib64/python3.11/site-packages/pydantic/v1/fields.py:550: in prepare
    self._type_analysis()
/usr/local/python/python-3.11/std/lib64/python3.11/site-packages/pydantic/v1/fields.py:756: in _type_analysis
    self.sub_fields = [self._create_sub_type(self.type_, '_' + self.name)]
/usr/local/python/python-3.11/std/lib64/python3.11/site-packages/pydantic/v1/fields.py:806: in _create_sub_type
    return self.__class__(
/usr/local/python/python-3.11/std/lib64/python3.11/site-packages/pydantic/v1/fields.py:434: in __init__
    self.prepare()
/usr/local/python/python-3.11/std/lib64/python3.11/site-packages/pydantic/v1/fields.py:555: in prepare
    self.populate_validators()
/usr/local/python/python-3.11/std/lib64/python3.11/site-packages/pydantic/v1/fields.py:829: in populate_validators
    *(get_validators() if get_validators else list(find_validators(self.type_, self.model_config))),
/usr/local/python/python-3.11/std/lib64/python3.11/site-packages/pydantic/v1/validators.py:765: in find_validators
    raise RuntimeError(f'no validator found for {type_}, see `arbitrary_types_allowed` in Config')
E   RuntimeError: no validator found for <class 'langchain_core.messages.base.BaseMessage'>, see `arbitrary_types_allowed` in Config

If pydantic.v1 and v2 can co-exist earlier, why is the langchain upgrade breaking it? Is this expected?