Ishaan- L.

hi Aparna D. following up on this - I followed up over DMs, could you give us access to this ?

I found this on docs - it was very helpful https://docs.arize.com/arize/large-language-models/tracing/semantic-conventions/input

here's what i'm doing right now:

    for key, value in optional_params.items():
        span.set_attribute(f"{SpanAttributes.LLM_INVOCATION_PARAMETERS}.{key}", str(value))

casting to a string seems to fix the issue - is this the recommended approach ?

    span.set_attribute(SpanAttributes.LLM_INPUT_MESSAGES, str(kwargs.get("messages")))

Do you have an example of how to log messages to Arize OTEL ? Jason I followed the doc you linked but I see this error: WARNING:opentelemetry.attributes:Invalid type dict in attribute 'llm.input_messages' value sequence. Expected one of ['bool', 'str', 'bytes', 'int', 'float'] code:

 span.set_attribute(SpanAttributes.LLM_INPUT_MESSAGES, "<input_messages>") # List of messages sent to the LLM in a chat API request, [{"message.role": "user", "message.content": "hello"}]

got it thanks for the help Jason 😄

here's my trace if it helps track down the issue: https://app.arize.com/organizations/QWNjb3VudE9yZ2FuaXphdGlvbjo4MTYzOkgzbUM=/spaces[…]ceId=3821f71fc1d4c8d43f5d4d000b35cdbe&modelType=generative_llm

I got this LLM response at 10:19am but I only see it on my Arize AI Traces 4 mins later. Is this delay expected ?