Hi everyone,
I am working on a project and one of the core scenarios project is based on has the chatgpt message assistant module in it. That module uses the 4o model asistant with the code itnerpreter to analyze .xlsx spreadsheet. Spreadsheet contains several sheets in it with value range approximately A1:G30. Not all cells have data in them.
What happens is that assistant analyzes only a few rows and completely ignores others. Even in the example where I have 10 rows of info, it analyzes 3-4 of them.
I tested it in the playground and it is within the contextual window. No matter the prompting, it won’t analyze the entire data set.
Has anyone been dealing woth the same issue?
Hi @Info_Funderstanding and welcome to the Make Community!
Are you saying that the playground has the same problem? Or that it works in the playground but not in Make?
If it doesn’t work in the playground, that’s probably a question for the OpenAI API forum.
Some avenues to explore:
- Does it work in ChatGPT? I know when I ask ChatGPT to create XLSZ files that contain more than 10 rows, it doesn’t do it. It’s like there is a hard-coded limit
- Try converting your file to CSV to see if that works better
- Make sure your Excel file doesn’t contain empty rows. It’s OK to have empty cells but not empty rows.
L
Thank you for you reply!
- It’s not Make issue as the same thing happens in the openAI playground.
I will try to see what their community says but I assume that converting the file and providing a different format as the input could solve the issue. I will post the solution here.
1 Like