6

I have seen the myfiles_browser message, but now I am also seeing this:

Based on the initial content of the document titled “Executive Order on the Safe, Secure and Trustworthy Development and Use of Artificial Intelligence,” the following text has been developed. Unfortunately, without the full text, only the first 7% is available.

8

I am also seeing this issue. I can still interact with the files using OpenAI threads UI.

But I cannot seem to use the assistant in my own API/UI without this error.

Seems like some sort of config issue on OpenAIs side. I hope they will patch it soon, there is nothing in the docs that can point to a fix.

1 Like
9

Doesn’t work for me neither. The initial loading of the file works quite good. But changing the files like replacing them made this error. Doesn’t sound so stable as per ‘Preview’ indicates.

10

I actually seem to have been able to fix this. It seems like I was not using the API correctly, I believe I was trying to execute the run before creating a message properly. So it would try to execute a run without any messages, which caused this weird response.

11

Did you permanently fix it? What was the fix?

12

I am certainly creating messages each time…sometimes it can read a doc and sometimes it cannot. It screams “eventual consistency” cloud storage bug.

13

I get the same error too, but it seems random. Some times works other times via Assistant API throw this error: This file is NOT accessible with the myfiles_browser tool

Via plyaground i can see all the files correctly uploaded with status=ready and threads is working with no error.
Did you figure out what could be the problem?

14

Ya same issue it seems kinda random right now… sometimes it works sometimes it doesn’t playground and API.

1 Like
15

So far the only way I can get it to consistently work is if I just upload one file from the start. Then I can keep adding one at a time. ( Playground and API )

16

I’m also getting the same error from API:

It seems you’ve uploaded a file, but there currently seems to be an issue accessing it with the myfiles_browser tool. This may be a temporary issue.

However, it’s working fine from the playground.

17

Same issue for me. Works in Playgrounds but not via API.

I see that you have uploaded two files, but currently, I cannot read or analyze these files as they are not accessible through the tool I have available (myfiles_browser). If you want me to help you with the files’ content, please upload them again and ensure they are accessible through the myfiles_browser tool.

1 Like
18

I have a work around.

First, change your assistant config tools to this:[{"type": "code_interpreter"}]
Uploaded files also works with code_interpreter mode and actually works compared to retrieval.

When creating your message. You need to specify in your message content (prompt): “output as a saved csv file with file.id accessible”
also reference the file.id like so:
file_ids=[uploaded_file.id]
in the message parameters.

run the thread and once it is completed, it should return a file_id. This is how i retrieved the file_id: client.beta.threads.messages.list(thread_id=thread.id).data[0].content[0].text.annotations[0].file_path.file_id
once you have the file_id just simply return the content of the file with this: client.files.retrieve_content(file_id)

Hope this helps

19

I’ve got some success with just adding another message to the thread and saying something to the effect of “Yes you do have the files, I attached them in the initial prompt (or wherever you may have attached them).”. The model seems to then generate a response to what I was initially asking.

20

I found that some of my files had character sequences that, when handled incorrectly in python, turned into control characters. I changed my preprocessing to ensure that they would be treated as escaped characters when read as utf-8 encoded files and have much higher success.

21

The struggle with this particular bug is that it feels so random that it’s hard to tell when something you changed (eg. a prompt tweak or code implementation) actually improves the outcome or if it just happened to work that time. Is there a way to get in touch with OpenAI devs to ask for support or to check on the status of bug fixes?

1 Like
22

I’m struggling with how to even handle this particular bug.

Does one parse the messages in every thread for certain keywords such as “myfiles_browser” because this condition does not raise an exception.

And in some cases, it seems like the same keywords aren’t even being used but the end result is the same: we could not read the files associated with this request.

23

I did

        if ('re-upload' in last_content.lower() 
                        or last_content.startswith("I apologize")
                        or last_content.startswith("Apologies")
                        or last_content.startswith("I'm sorry for the misunderstanding")
                        or "The file you've uploaded is not accessible with the tool that allows me to view" in last_content
                        or 'I’m sorry for the confusion, but it seems there was an issue with the file accessing tool.' in last_content 
                        or 'myfiles_browser' in last_content):
1 Like
24

Same error, it works in the playground but not with the API. It always mentions myfiles_browser, even though the file has been uploaded and I can see it within the files tab…

1 Like
25

The fact that you have to re-upload a file over the playground to a thread and not use an existing file is suspicious. In combination with the fact that we are billed for every message this file is used (full token count) makes me belive that under the hood the whole file_ids stuff is not working yet.

I really think that they use the uploaded file, take the content and make it part of their internal messages… Or am I on the wrong track here?

26

I may take a similar approach as you! :slight_smile:

When this condition is True, how do you handle it? Add another thread saying that the files are actually there?