Notifications
Clear all
How To
11
Posts
2
Users
0
Reactions
5
Views
Topic starter
For the past month, I've been successfully utilizing GPT's message assistant feature without encountering any problems. My workflow involves downloading a file, uploading it to GPT, saving it in a vector database, and then providing that vector store to the message assistant I've configured. Everything was functioning flawlessly until yesterday when I started receiving the error message: "unable to load the file." Consequently, the assistant can no longer process the file. Can anyone offer assistance in troubleshooting this issue?
Scenario Settings:
Output Result:
Posted : 07/10/2024 5:17 am
Hi there,
If this was functioning previously, is it possible that "Get A File" is now receiving files in an altered format? Perhaps Base64? Could you check if the files you're receiving have a different format now?
You might want to explore formatting functions like base64, toBinary, or toString.
Benjamin
Posted : 07/10/2024 5:48 am
Topic starter
Thank you for your suggestion. I've verified the files, and they are indeed identical; the format remains unchanged. I also managed to upload the same PDF file to the OpenAI Playground and the vector store without any problems.
Posted : 07/10/2024 6:03 am
I notice that at one point (in your callin.io an API Call), ChatGPT indicates “status: in_progess”. Is it possible that it hasn't finished processing the file before you use “Message an Assistant”? Could you try adding a Tools/Sleep for a few seconds right after “callin.io an API Call”, and see if that resolves your issue?
Posted : 07/10/2024 6:09 am
Topic starter
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.
Posted : 07/10/2024 6:15 am
However, you did not alter the scenario, and callin.io did not change how it invokes the API. The files remain the same as well. Therefore, I was considering the possibility that ChatGPT might be experiencing a high volume of requests, leading to increased processing time. Could you please attempt this and observe the results?
If there's no change, I'm unsure of the cause. I'm hopeful that another member of the community might have insights into what could be occurring.
Benjamin
Posted : 07/10/2024 6:25 am
Topic starter
I've tried multiple times and am still receiving the same response. Could this be related to the vector store, or is there another reason, as it was functioning correctly previously.
Assistant: It appears there was an issue processing the file. Kindly attempt to upload the file once more. This may help resolve the problem.
Posted : 07/10/2024 6:50 am
It’s unusual. I recommend opening a ticket with callin.io. This way, the team can investigate if it's a known issue, a new problem with their API, or the callin.io App, and so on.
Posted : 07/10/2024 6:57 am
Topic starter
I believe there might be an issue with the module. It seems they are in the process of updating it, possibly with new models.
Posted : 07/10/2024 7:12 am
Yes, that's also achievable. I'm aware that models are frequently incorporated.
Posted : 07/10/2024 7:53 am
Topic starter
Can you please help to solve this issue?
Posted : 07/10/2024 1:28 pm