Add LLM_CONTENT_COMPLETION_CHUNK attribute to SpanAttributes class

[Dwij1704]

📥 Pull Request

📘 Description
When running chat.completions using stream=True, it throws out
An error occurred: type object 'SpanAttributes' has no attribute 'LLM_CONTENT_COMPLETION_CHUNK'

🧪 Testing
Added LLM_CONTENT_COMPLETION_CHUNK attribute and tested using the following script

import os
from openai import OpenAI
from dotenv import load_dotenv
load_dotenv()
import agentops
agentops.init()
agentops.start_session(tags=["openai_chat"])

client = OpenAI(
    api_key=os.getenv("OPENAI_API_KEY")  # Make sure to set your API key in environment variables
)

def chat_with_gpt():
    try:
        # Create a streaming chat completion
        stream = client.chat.completions.create(
            model="gpt-3.5-turbo",
            messages=[
                {"role": "system", "content": "You are a helpful assistant."},
                {"role": "user", "content": "What is the capital of France?"}
            ],
            temperature=0.7,
            max_tokens=100,
            stream=True  # Enable streaming
        )
        
        # Print the response as it streams
        print("Assistant's response: ", end="", flush=True)
        for chunk in stream:
            if chunk.choices[0].delta.content is not None:
                print(chunk.choices[0].delta.content, end="", flush=True)
        print()  # New line at the end
        
    except Exception as e:
        print(f"An error occurred: {e}")

if __name__ == "__main__":
    chat_with_gpt() 

[tcdent]

Good catch!

[Dwij1704]

Good catch!

@tcdent
We heard back from VRSEN AI regarding Langchain callback handlers, they said its broken due to type object 'SpanAttributes' has no attribute 'LLM_COMPLETIONS'

@the-praxs was going through the implementation and he noticed that you have specifically mentioned NEVER manually set the root completion attributes (SpanAttributes.LLM_COMPLETIONS or "gen_ai.completion") Let OpenTelemetry backend derive these values from the detailed attributes contrary also commented out these attributes from semcnv.

Can you give me some clarity on how you wish to manage both of these conventions, manually or via semconv?

[dot-agi]

This catch is a good reminder to have tests for attributes in some way. The order of merge caused this unexpected error.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

📢 Thoughts on this report? Let us know!

[Dwij1704]

I fixed the error, but when testing it, I can't see token usage in any of the spans for Agents SDK
https://jaeger.agentops.cloud/trace/c9a45998d27d289fa35220a37f6fcac1
Langchain is working as expected
@the-praxs @tcdent Can you confirm if it used to set those attributes before or not?

[dot-agi]

I fixed the error, but when testing it, I can't see token usage in any of the spans for Agents SDK https://jaeger.agentops.cloud/trace/c9a45998d27d289fa35220a37f6fcac1 Langchain is working as expected @the-praxs @tcdent Can you confirm if it used to set those attributes before or not?

Token usage fixed in #888