Why am I unable to pass a file to my assistant message Module?

I’ve been successfully using GPT’s message assistant feature for the past month without any issues. My process involves downloading a file, uploading it to GPT, storing it in a vector database, and then passing that vector store to the message assistant I created. Everything was working perfectly until yesterday, when I started getting the response: “unable to load the file.” Now, the assistant is unable to process the file. Can anyone help me troubleshoot this issue?

Scenario Settings:




Output Result:

22
22a
22b
22c

Hello,

If it was working before, could it be possible that “Get A File” now receives a file in a different format? Base64 for example? Can you try and see if the files you get have now a different format?

You can then try and explore that formating functions such as base64, toBinary, toString.

Benjamin

@Benjamin_from_Make Thank you for your suggestion. I’ve checked the files, and they are indeed the same; the format has not changed. I also successfully uploaded the same PDF file to the OpenAI Playground and the vector store without any issues.

I see that at some point (in your Make an API Call), ChatGPT says “status: in_progess”. Could it be that it hasn’t finished processing the file before you use “Message an Assistant”? Can you try and add a Tools/Sleep for a few seconds just after “Make an API Call”, and see if it fixes your issue?

Benjamin

@Benjamin_from_Make I don’t believe that’s the issue. The process ran successfully before, and it consistently shows the same status: “in_progress.” So i think this is not the issue.

But you didn’t change the scenario, Make didn’t change the way it calls the API, and the files are the same; so I was wondering if possibly, ChatGPT has too many requests and takes more time to process. Can yo just give it a try, and see.
If it doesn’t change, I don’t know what is happening. I hope someone else from the community knows what could be happening.

Benjamin

@Benjamin_from_Make I have tried many time still just get the response from, Is it due to vector store or anyother thing because it was working fine before.

Assistant: It seems there was an error while trying to process the file. Could you please try uploading the file again? This might help resolve the issue.

It’s weird. May I suggest that you open a ticket at Make? Like this the team can check if it’s a known issue, if there is a new problem with their API or the Make App, etc.

Benjamin

@Benjamin_from_Make I believe there might be an issue with the module. It seems they are in the process of updating it, possibly with new models.

@Med_FutureXAI yes, it’s also possible. I know that models are very often added.

@samliew Can you please help to solve this issue?