I am using the assistants API. I have an assistant that has a code interpreter enabled. I uploaded a CSV file. I created a new empty thread. So far so good. I added a message with the uploaded file ID with a question.
My assistant replies with the following message:
“I apologize for the inconvenience. It appears there’s a persistent internal issue preventing me from accessing the data. Unfortunately, without being able to read the file correctly, I can’t answer the question”
I am getting this message for a couple of hours now. It seemed to work before.
I think there is a problem with the code interpreter who handles my personal account. Not only can’t any files uploaded be correctly identified and processed, but also the code he is asked to write can’t run and report errors to any extent within it. This situation is called “internal error” and “environmental problem” by him. And he only got the error message “An error occurred” rather than any error code or error report!
I was just about to make a topic about a similar problem.
I’m working with an assistant I’ve created & am using in the playground mode. When uploading a .csv file for the assistant to analyse the assistant isn’t able to see the file. Without this apparent system wide problem, what are general best practices for interacting with an assistant. At the moment I use text based prompting in the playground. What are the advantages of for example using Pycharm and the API Key for interacting?
In my case, it does not report any error, but it has lost the ability to query the files present in the knowledge base. It always responds that it finds nothing in the files it has access to and therefore will answer based on its general knowledge. Note that I always use the same prompt, and until a few hours ago, it correctly retrieved information from the knowledge base. This behavior is identical both in the playground and in GPTs
For me too, it now seems to be working better; sometimes it can and sometimes it cannot access the knowledge base. In the latter case, it now specifically says (which it did not do before), ‘I am currently unable to access it via the myfiles_browser tool.’ I guess they are working on it.