Custom GPT configurations

GPT builders
,
1

Hi all,

I’ve been playing around with building a custom GPT, but can’t seem to get it working accuratly and consistently yet. Been trying a lot, but I figured you guys might be able to help me out on this quicker.

A little context:
The GPT is built to assist the user assigning a certain code to it’s producs which are required for customs purposes. The code has to be accurate in order to avoid trouble with the customs. To do this, the GPT requires some detailed information from the user on which it can base what code to assign. To assign a code it should use web browsing as well as using 2 documents in it’s knowledge base with guide lines, examples and the possible codes to be assigned. These documents however are very large and contain over 600+ pages in total.

I’m facing multiple problems with this. First, a lot of errors occur. These very between ‘Error generating a response’, the GPT stating it has not yet been able to read the entire documents due to time constraints, ‘network error’, and so on. Also, the GPT does not consistently use the the same steps in it’s process, even though I’m clearly stating what steps to follow in assigning the right code.
So the insonstistency and the many, many errors when trying to generate a respond make it really not usable at the moment.

Any of you got experience with custom GPT’s with similar goals or similar problems i’m facing? Any tips/tricks on this?

Anything would be helpful! Thanks in advance!

1 Like
2

Similar issues here: Instructions interpretation issues

1 Like
3

Hi there,

It sounds like you’re facing quite a challenge with your custom GPT model. One approach that might help is integrating a custom API in action function to handle specific, detailed queries. GPT models are great for text understanding and generation but can struggle with large, structured datasets, like your 600+ page documents. By using an API, you can preprocess and structure this data more effectively. This API can then specifically deal with queries about product codes, while the GPT handles the generation of user-friendly responses and less structured queries. This hybrid approach leverages the strengths of both: the detailed data management of an API and the conversational abilities of GPT. It might also help in reducing errors and improving consistency in responses. Hope this helps, and best of luck with your project!

3 Likes