Thread

1. 0123456789ABCDE · 1790180723 · focus · HN ↗

   this is such a trivial thing to do in DSPY, no one bothered to give it a name…

   here's 7 lines

     import os
     import dspy
   
     lm = dspy.LM("openrouter/z-ai/glm-5.3-flash", api_key=os.environ["OPENROUTER_API_KEY"])
     jev = dspy.Predict('email:str -> choice:Literal["Legitimate", "Spam", "Phishing"]')
     email = "Payroll asks for your password on a non-company sign-in page."
     pred = jev(email=email, lm=lm)
     print(pred.choice)
   

   there are other options, obviously. you can choose to give it some tools, maybe some reasoning stage before picking a choice, and that's on top of the "reasoning" the llm model already does api side

   1. amai · 1790182153 · focus · HN ↗

      What about the probabilities?

      1. 0123456789ABCDE · 1790183091 · focus · HN ↗

         so you can pick the choice with the highest probability?

         the example uses an external api, and i don't think they return probabilities from those anyway.

   2. 0123456789ABCDE · 1790187664 · focus · HN ↗

      after looking closer at the typesafe's jev, i want to point out that i misunderstood the significance of this jev model

      it is the latency that makes it significant

Unofficial Hacker News client; not affiliated with Y Combinator.