vllm-project
[Frontend] [Bugfix] Refactor tool parsers and simplify the tool parsing interface.
Co-authored-by: elementary-particle
Completed the following, stale PR:
"This is the PR for the RFC https://github.com/vllm-project/vllm/issues/11522. Currently we are building a draft of simpler tool parsers using streaming JSON parsing libraries to reduce overhead and avoid bugs. Tests and commits will be added gradually.
FIX https://github.com/vllm-project/vllm/issues/11392 ."
👋 Hi! Thank you for contributing to the vLLM project.
💬 Join our developer Slack at https://slack.vllm.ai to discuss your PR in #pr-reviews, coordinate on features in #feat- channels, or join special interest groups in #sig- channels.
Just a reminder: PRs would not trigger full CI run by default. Instead, it would only run fastcheck CI which starts running only a small and essential subset of CI tests to quickly catch errors. You can run other CI tests on top of those by going to your fastcheck build on Buildkite UI (linked in the PR checks section) and unblock them. If you do not have permission to unblock, ping simon-mo or khluu to add you in our Buildkite org.
Once the PR is approved and ready to go, your PR reviewer(s) can run CI to test the changes comprehensively before merging.
To run CI, PR reviewers can either: Add ready label to the PR or enable auto-merge.
🚀
![github-actions[bot] avatar](https://avatars.githubusercontent.com/in/15368?v=4 "Published by a pub.dev verified publisher"){kind=small}
Hi @marcelodiaz558, hi @Endebert, can you verify if this fits your needs?
Thank you and best regards, Paul
Just chiming in, there is also a problem with the llama 3 tool parser where it cant generate non ascii characters
@paolovic Thanks for doing this. I can confirm my issues with streamed function calling are resolved with this PR.
Hi @PedroF43 , if this is related to this PR, could you provide an example, please?
Thank you and best regards, Paul
Hi @PedroF43 , if this is related to this PR, could you provide an example, please?
Thank you and best regards, Paul
this PR of mine has better examples https://github.com/vllm-project/vllm/pull/13826
Hi @PedroF43 , if this is related to this PR, could you provide an example, please? Thank you and best regards, Paul
this PR of mine has better examples #13826
But this PR is about hermes tool parser, yours is about llama tool parser, isn't it?
Hi @PedroF43 , if this is related to this PR, could you provide an example, please? Thank you and best regards, Paul
this PR of mine has better examples #13826
But this PR is about hermes tool parser, yours is about llama tool parser, isn't it?
yes you are correct, i thought that by your title you intended to go trough all the available tool parsers
This pull request has merge conflicts that must be resolved before it can be merged. Please rebase the PR, @paolovic.
https://docs.github.com/en/pull-requests/collaborating-with-pull-requests/working-with-forks/syncing-a-fork
![mergify[bot] avatar](https://avatars.githubusercontent.com/in/10562?v=4 "Published by a pub.dev verified publisher"){kind=small}
Hi @Endebert ,
I got a notification from you about a potential bug containing two JSONs as arguments.
Did you delete this comment because it is resolved?
Hi @Endebert , I got a notification from you about a potential bug containing two JSONs as
arguments. Did you delete this comment because it is resolved?
@paolovic Yes, sorry about that. Turned out it was a bug in our code, which appended arguments objects if multiple functions were called during the same streaming response.
Hi @Endebert , I got a notification from you about a potential bug containing two JSONs as
arguments. Did you delete this comment because it is resolved?@paolovic Yes, sorry about that. Turned out it was a bug in our code, which appended arguments objects if multiple functions were called during the same streaming response.
Easy, no problem, thanks for letting me know! Best regards
This pull request has merge conflicts that must be resolved before it can be merged. Please rebase the PR, @paolovic.
https://docs.github.com/en/pull-requests/collaborating-with-pull-requests/working-with-forks/syncing-a-fork
@aarnphm Do you have an ETA regarding your alternative route? Because we need these changes for production use. It's fine I guess to stay on this fork and rebase + manually build it occasionally, but it's still an inconvenience.
Do you have an ETA regarding your alternative route? Because we need these changes for production use. It's fine I guess to stay on this fork and rebase + manually build it occasionally, but it's still an inconvenience.
I would want to have this ready before 0.10.0, which will be in around 4-6 weeks.
fwiw we might want to just merge all of the implementations/fix right now to make sure that it will still work for 0.9.x, but for 0.10.x, i would be more lenient for a RFC for redesigning the parser.
This pull request has merge conflicts that must be resolved before it can be merged. Please rebase the PR, @paolovic.
https://docs.github.com/en/pull-requests/collaborating-with-pull-requests/working-with-forks/syncing-a-fork
This pull request has been automatically marked as stale because it has not had any activity within 90 days. It will be automatically closed if no further activity occurs within 30 days. Leave a comment if you feel this pull request should remain open. Thank you!
This pull request has been automatically closed due to inactivity. Please feel free to reopen if you intend to continue working on it. Thank you!