How did you train the model? Did you come across any particular challenges or problems with training? Would love to learn about any particular techniques you used based on your experience.
It’s just collecting that data from the fields, then inserting them into a prompt to send back to OpenAI; you can tell they’re using OpenAI from error messages.
I tried to inject a prompt into one of the fields to get back the prompt template they are using, but been unable to retrieve it:
I tried to inject a prompt into one of the fields to get back the prompt template they are using, but been unable to retrieve it:
https://news.ycombinator.com/item?id=34625129