Hello, I'm encountering an issue with the Playground while using Phoenix to log traces of an OpenAI agent. When I select a specific LLM call from the trace and open it in the Playground, it loads, but the tools aren't available. As a result, when I rerun a function-calling step—even without making any changes—the LLM returns a regular text response instead of responding with a function call. It seems like the Playground isn't properly loading the tools context for those steps. Could you help me resolve this?
💡 Hint: Mention RunLLM in the thread for followups.
Hello John G. Here are the answers to your questions:
No errors seen in the playground.
I am using LlamaIndexInstrumentor
The used versions are:
arize-phoenix = "8.14.1"
arize-phoenix-otel = "0.8.0"
John G. No, there's no tools tab in the trace.
We released a fix in openinference-instrumentation-llama-index: v4.2.1. Please give it a try when you have chance. Thanks again for raising this issue.
