[bug] instructor max retries invocation params issue

[peterhollis]

Describe the bug
When max_retries is set as a Tenacity Retrying object, Type Error: Object of type Retrying is not JSON serializable is raised by _wrapper.py (line 160)

To Reproduce
Assign max_retries as a Retrying object as seen here in the Instructor docs.

Expected behavior
Retrying object to be excluded from SpanAttributes.LLM_INVOCATION_PARAMETERS so the Type Error is not raised.

Desktop (please complete the following information):

• OS: MacOS
• Version 15.1.1
• Python: 3.13

Additional context

Example code:

import instructor
from openai import OpenAI
from pydantic import BaseModel
from tenacity import Retrying, stop_after_attempt, wait_fixed

client = instructor.from_openai(OpenAI())

class UserInfo(BaseModel):
    name: str
    age: int

response = client.chat.completions.create(
    model="gpt-4o-mini",
    response_model=UserInfo,
    messages=[
        {"role": "user", "content": "John Doe is 30 years old."},
    ],
    max_retries=Retrying(
        stop=stop_after_attempt(2), 
        wait=wait_fixed(1), 
    )
)

[dosubot]

Hello @peterhollis! I'm Dosu, your friendly bot here to help you tackle bugs, answer questions, and support your journey to becoming a contributor. I'll be around while you wait for a human maintainer. How can I assist you today?

The issue you're encountering is due to the Retrying object from Tenacity not being JSON serializable, which causes a TypeError when it's included in SpanAttributes.LLM_INVOCATION_PARAMETERS. The _wrapper.py file currently does not have a mechanism to exclude specific types like Retrying from serialization ^[1].

To resolve this, you could modify the serialization logic to exclude or handle the Retrying object specifically. One approach is to update the safe_json_dumps function or the SafeJSONEncoder class to check for instances of Retrying and handle them accordingly, such as by converting them to a string representation or excluding them from the serialization process. This would prevent the TypeError from being raised.

To continue talking to Dosu, mention @dosu.

---

^{Help Dosu learn! Give it feedback: Great Response | Irrelevant Answer | Incorrect Sources | Too Verbose | Hallucination | Bug Report | Other}