Truncated results

The symptom for this issue is when you're receiving large objects from various LLMs sometimes there is too much data that overflows the MaxTokens value for the LLM, which cause the result to be truncated.

This results in a truncated document as below. The text "just stops", and the JSON is not properly terminated.:

The likely problem is that there are too many tokens being returned by the LLM, and that's causing the truncation.

You may need to adjust the Max Tokens field in the LLM parameters. To resolve this, click on the LLM name in the experiment;

Click Params then adjust the Max Tokens field as needed.