Command browse_website returned: Error: This model's maximum context length is 4097 tokens. However, your messages resulted in 5023 tokens. Please reduce the length of the messages.

Open vmajor opened this issue 1 year ago • 4 comments

Model token length limit defined in .env does not seem to propagate to all relevant parts of the code.

I am using gpt-3.5-turbo and have set token limits to 4000 in .env

I am using the latest stable branch, in windows and invoked with run.bat

Apr 21 '23 07:04 vmajor

################################################################################

################################################################################

BROWSE_CHUNK_MAX_LENGTH - When browsing website, define the length of chunk stored in mem>

BROWSE_CHUNK_MAX_LENGTH=3000 <----------- set this to 3000, or under 4000, its prolly set to 8172 or something

Apr 21 '23 08:04 philliplastcall

Nope. Already at 3000

Apr 21 '23 08:04 vmajor

read_file has same issue

Apr 21 '23 09:04 callect

i put mine to 2k and no errors since.

Apr 21 '23 13:04 Drlordbasil

Closing as duplicate of #796. Please post there if the issue is still relevant.

Apr 22 '23 13:04 Pwuts