Trace details LLM call input JSON format is wrong by default

[SuveenE-TF]

LLM input format is wrong by default and due to this test in playground does not get pre populated. However it looks perfect in Pretty view.

[dosubot[bot]]

Hey @SuveenE-TF! I'm here to help you with any bugs, questions, or contributions. Let's tackle this together!

The issue you're encountering with the LLM call input JSON format not being pre-populated in the playground might be related to how the JSON is parsed and validated before being used. In Langfuse, the JSON format for LLM call inputs is validated using the validateChatCompletionBody function, which ensures the input adheres to a specific schema. This schema includes fields like projectId, messages, and modelParams ^[1].

If the JSON format appears correct in the Pretty view but not in the playground, it could be due to discrepancies in how the JSON is structured or validated. Ensure that the JSON input matches the expected schema, particularly the structure of the messages array and the presence of required fields.

Additionally, the JumpToPlaygroundButton component is responsible for setting the playground cache with parsed inputs. If the JSON is not being pre-populated, there might be an issue with how the parsePrompt or parseGeneration functions are handling the input data ^[2].

Reviewing these components and ensuring the JSON input adheres to the expected schema should help resolve the issue.

To continue talking to Dosu, mention @dosu.

---

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

[marcklingen]

Can you share a public link to this trace and specify which tracing integration you used?

[SuveenE-TF]

Thanks @marcklingen. Unfortunately, i cannot share the link in a public forum. I am using python langchain integration and chatOpenAI.

[marcklingen]

Can you copy/paste the redacted json object?