Cannot query JSON document anymore

[AJNOURI]

I used to use this script to query a json file before 2024:

import json
from langchain.agents import create_json_agent
from langchain.agents.agent_toolkits import JsonToolkit
from langchain.llms.openai import OpenAI
from langchain.tools.json.tool import JsonSpec
from pathlib import Path

file_path='.file.json'
data = json.loads(Path(file_path).read_text())

json_spec = JsonSpec(dict_=data, max_value_length=4000)
json_toolkit = JsonToolkit(spec=json_spec)

json_agent_executor = create_json_agent(
    llm=OpenAI( temperature=0), toolkit=json_toolkit, verbose=True
)

# json_agent_executor.run(
#     "Quel service dois-je utiliser qui se baser sur une pile Vmware?"
# )

json_agent_executor.run(
    "liste moi tous les prefixes conteneurs dans l'infrastructure ch1?"
)

this doesn't work anymore:

    raise self._make_status_error_from_response(err.response) from None
openai.NotFoundError: Error code: 404 - {'error': {'message': 'The model `text-davinci-004` does not exist', 'type': 'invalid_request_error', 'param': None, 'code': 'model_not_found'}}

O have tried to add "model" argument to :

json_agent_executor = create_json_agent(
    llm=OpenAI(model="gpt-3.5-turbo", temperature=0), toolkit=json_toolkit, verbose=True
)

Doesn't work neither:

openai.NotFoundError: Error code: 404 - {'error': {'message': 'This is a chat model and not supported in the v1/completions endpoint. Did you mean to use v1/chat/completions?', 'type': 'invalid_request_error', 'param': 'model', 'code': None}}

What is the solution to this issue?

[eyurtsev]

OpenAI is a chat model rather than an LLM. So use ChatOpenAI if you're using gpt-3.5-turbo and it should work.

text-davinci-xxx were LLM models but OpenAI deprecated them.

You will need to confirm hat create_json_agent works with ChatModels

[AJNOURI]

It doesn't seem to recognize ChatOpenAI:

import json
from langchain.agents import create_json_agent
from langchain.agents.agent_toolkits import JsonToolkit
from langchain.llms.openai import ChatOpenAI
from langchain.tools.json.tool import JsonSpec
from pathlib import Path

file_path='file.json'
data = json.loads(Path(file_path).read_text())

json_spec = JsonSpec(dict_=data, max_value_length=4000)
json_toolkit = JsonToolkit(spec=json_spec)

json_agent_executor = create_json_agent(
    llm=ChatOpenAI(model="gpt-3.5-turbo", temperature=0), toolkit=json_toolkit, verbose=True
)

from langchain.llms.openai import ChatOpenAI
ImportError: cannot import name 'ChatOpenAI' from 'langchain.llms.openai' (***/.venv/lib/python3.11/site-packages/langchain/llms/openai.py)

Same with this example from the documentation:
https://python.langchain.com/docs/integrations/chat/openai

python3 -m pip show langchain                          
Name: langchain
Version: 0.1.1

[eyurtsev]

ChatOpen ai is a chat model rather than an LLM.

In the link that you provided, take a look at the ChatOpenAI import. It's not from langchain.llms.openai but from langchain_openai

from langchain_openai import ChatOpenAI

[AJNOURI]

At first, it gave the same errore message with langchain_openai.
So I tested (from stackoverflow)

from langchain_community.chat_models import ChatOpenAI

Which worked and gave me a hint about installing the package langchain-openai

python3 -m pip install -U langchain-openai

***/.venv/lib/python3.11/site-packages/langchain/agents/agent_toolkits/__init__.py:64: LangChainDeprecationWarning: Importing this agent toolkit from langchain is deprecated. Importing it from langchain will no longer be supported as of langchain==0.2.0. Please import from langchain-community instead:

`from langchain_community.agent_toolkits import JsonToolkit`.

To install langchain-community run `pip install -U langchain-community`.
  warnings.warn(
***/.venv/lib/python3.11/site-packages/langchain_core/_api/deprecation.py:117: LangChainDeprecationWarning: The class `langchain_community.chat_models.openai.ChatOpenAI` was deprecated in langchain-community 0.0.10 and will be removed in 0.2.0. An updated version of the class exists in the langchain-openai package and should be used instead. To use it run `pip install -U langchain-openai` and import as `from langchain_openai import ChatOpenAI`.
  warn_deprecated(
***/.venv/lib/python3.11/site-packages/langchain_core/_api/deprecation.py:117: LangChainDeprecationWarning: The function `run` was deprecated in LangChain 0.1.0 and will be removed in 0.2.0. Use invoke instead.
  warn_deprecated(

Now it works fine with langchain_openai from ChatOpenAI:

from langchain_openai import ChatOpenAI
...
json_agent_executor = create_json_agent(
    llm=ChatOpenAI(model="gpt-3.5-turbo", temperature=0), toolkit=json_toolkit, verbose=True
)

And it works also with the OpenAI from langchain.llm.openai

from langchain.llms.openai import OpenAI
...

json_agent_executor = create_json_agent(
    llm=OpenAI(temperature=0), toolkit=json_toolkit, verbose=True
)

Thanks @eyurtsev