#embedding or fine tuning via api but it

4 messages · Page 1 of 1 (latest)

willow marlin
#

If you are sticking with ChatGPT, then you can feed it the filename and contents, but if the document exceeds the max token limit, you will have issues there.

#

When I am working with code, I will tell it something like:

I will provide you with 2 file names and their contents. Please wait until all files have been provided. Some files will be provided in multiple parts with (# of #) denoting which part of total.
If GPT tries to respond before you are done providing files, hit that 'Stop Generating' button and then send it the next part. Rinse and repeat until you have the full file, but at the end of the message with the last part of the file, put your first request. Use ### as a separator between files and instructions.

`File: MyAwesomeResume.txt (1 of 2)
Contents:
Resume stuffs, yo..

File: MyAwesomeResume.txt (2 of 2)
Contents:
...
.. and in conclusion, hire me already!

File: CoverLetter.txt
Contents:
Dear sir or mad men..
..stuff
..things.

These files are my resume and cover letter. Please analyze them and make recommendations for improvement.

`

blazing sorrel
#

Thank you! How do you feed ChatGPT filename does it have to be a public link like a google doc? Wasn’t sure if it had the power to read the file or guessing from the file url.