Status: Pipeline Failure, Feature Request, Resolved
wasayahmed09 (2024-12-11 17:09:32):
No problem, happy to help! 
akili_85623 (2024-12-11 17:07:06):
This looks like what I need. Can’t believe I didn’t see it. Thanks Wasay
wasayahmed09 (2024-12-11 17:00:33):
Hey <@1243383663618560031>, this is definitely possible. You can use a Duplicate node and connect the input to the ‘Ask AI’ response and the ‘List size to match’ with the file contents list from the ZIP File reader node.
This way the response from the AI would be duplicated and passed with each resume, solving both the list size mismatch issue as well not having to loop the AI node multiple times.
Let me know if this works for you!
akili_85623 (2024-12-11 15:34:48):
Interesting. I’m not sure if this would work since the scorer only gives 2 options for scoring multiple resumes: give it a “list” for the criteria, or type it in when editing the gumloop.
bartst. (2024-12-11 15:03:23):
Is it an idea that you first let the AI create the criteria, store this in a google sheet en then let a scoring AI (with a different LLM) do its job ?
akili_85623 (2024-12-11 14:52:53):
I wanted to build a simple resume reviewer where you can upload a folder of resumes, input the job title & description, and then get them ranked.
The issue is in the screenshot, and I’m not sure how to fix this without a “variable” feature.
As you can see, I want the LLM to create the criteria one time and then use the criteria to grade the resumes. However, that criteria is one response, but the resumes have to loop through the scorer. That’s what is causing the single value /= list.
I could have the criteria module loop, but then it may come up with different criteria and will also spend unnecessary credits. Ideally, I could set the criteria as a “variable” that I could call later in my scorer.
I could also edit the gumloop for every different job posting by manually typing the criteria, but that means the front end experience can’t exist.
Any suggestions/ideas?
