#does anyone know what this means or how to fix? (am using OpenAI with Zapier)

21 messages · Page 1 of 1 (latest)

deft moth
#

does anyone know what this means or how to fix? (am using OpenAI with Zapier)
"The app returned "This model's maximum context length is 4097 tokens, however you requested 6226 tokens (2176 in your prompt; 4050 for the completion). Please reduce your prompt; or completion length."."

sharp ingot
#

Your prompt is too long. You need it to be less than 4096 tokens but you also need to leave space for the response you are expecting too

deft moth
#

Thanks! Is there a way to go beyond 4096 tokens? @leaden dew

sharp ingot
#

No

deft moth
#

ah okay, so i need to create projects with less than 4096 tokens?

sharp ingot
#

Yes but also probably less than that because you need to leave tokens for the response as well

deft moth
#

like i am trying to do a document rewrite

sharp ingot
#

If you want 2000 tokens for the reply you will need to send 2000 in the prompt

deft moth
#

i want it to rewrite a 10K character document to a new 10K character document

sharp ingot
#

Use chunks less than 2000 tokens. Maybe paragraphs or sections or chapters if small enough

deft moth
#

got it. so have the user submit multiple forms (with one paragraph per)? or one form with different sections (one paragraph per)? @sharp ingot

#

i tried reducing the max characters but got this

The app returned "This model's maximum context length is 4097 tokens, however you requested 5230 tokens (1180 in your prompt; 4050 for the completion). Please reduce your prompt; or completion length.".

#

and i only submitted 3524 characters..

sharp ingot
#

You need to reduce it even further.

#

Significantly reduce the length

deft moth
#

i set the text max submission length to 8300 characters (2K tokens) and max completion 2K, and now got this error "The app did not respond in-time. It may or may not have completed successfully.
"

sharp ingot
#

You may be experiencing a different unrelated issue. Try again. Openai is getting lots of timeout and rate limit errors of late

deft moth
#

OK thanks!

#

also if that is an issue, does that mean it doesn't make sense to try to make/run a business based off it?

#

tried again, same issue "The app did not respond in-time. It may or may not have completed successfully."

sharp ingot
#

A lot of people are contemplating that question. There is another post on discord and on the community forum showing lots of people are having timeout errors, 500 errors and rate limit issues. They are working on the issue. The status page indicates that it may be resolved but maybe not. It's hard to tell