jaybennett89

@Joffcom I believe the issue is this part of the input / information extractor base level system prompt: `Include the enclosing markdown codeblock:\n```` We don't control this part of the...

This was also an interesting article about how so much JSON training data is surrounded by markdown that it's highly likely to show up in the best next token: https://www.vellum.ai/blog/when-should-i-use-function-calling-structured-outputs-or-json-mode...

One last comment. I am using information extractor on data which has 8 fields I am looking for. However the input may be incomplete and missing data. The error is...

Hey @jeanpaul what would be the best place to send it? I don't really want to post my whole workflow here. I'm using n8n cloud version. It seems to be...

@jeanpaul I have created a minimal example that reproduces the issue here: https://gist.github.com/jaybennett89/91ea83a30eef118aba72575aef4382d2 The information extraction node with the shorter message will error every time. The longer one will work...

I have emailed [email protected] with a repro algorithm on this one. Let me know if there is a better way to share directly.