Check the max prompt length for the OpenAI completions API #472
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
zhuohan123
approved these changes
Aug 9, 2023
hongxiayang
pushed a commit
to hongxiayang/vllm
that referenced
this pull request
Feb 13, 2024
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Hi, I noticed that the
create_chat_completion()function in the openai entrypoint utilizes thecheck_length(prompt)function, which verifies that the prompt token length doesn't exceed the model's max token length. If it does exceed the max length, it returns an HTTP error explaining so.However this
check_length(prompt)function isn't present in thecreate_completion()function, so a request to this endpoint which exceeded the max token length wouldn't get a relevant error message.I added
check_length(prompt)tocreate_completion(), and also modified the function to return the (already computed)token_ids, so that we can re-use them later for the engine.generate(), thus reducing the amount of duplicate token encoding we are doing - in theory this will result in a minor efficiency gain, but I haven't done any scientific tests to prove so.Thanks!