Code interpreter truncating when fetching from csv

Hi :waving_hand:t3:

I am using the assistants api along with code interpreter to fetch a specific data from a csv. but it is only returning the first 84 characters:

The quiet river whispered secrets as the moonlight danced upon its gleaming surface!

But the original text from the column in the csv is:

The quiet river whispered secrets as the moonlight danced upon its gleaming surface! Beneath the starlit sky, constellations mirrored, felt like a lullaby towandering night.

So what could be causing this truncation, is it the assistants api or the code interpreter which id doing its own plumbing?

Thanks,

Steven.

1 Like

A CSV is comma-separated values.

Your “original text” is full of commas.

Thus, programmatic extraction would fail if the strings are not containerized in quotes that are escaped in the strings.

If the AI model is talking to you about the return from a script it ran instead of providing you a file to download, it may find the incomplete sentence not worth reporting on.

If you have a different data delineation such as tab, the AI should be made aware of this so it can write scripts properly to work on files.

1 Like