-
-
Notifications
You must be signed in to change notification settings - Fork 9.6k
[Bugfix] fix error due to an uninitialized tokenizer when using skip_tokenizer_init
with num_scheduler_steps
#9276
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[Bugfix] fix error due to an uninitialized tokenizer when using skip_tokenizer_init
with num_scheduler_steps
#9276
Conversation
👋 Hi! Thank you for contributing to the vLLM project. 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 do one of these:
🚀 |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Thanks @junstar92! Could you make the small requested change
dd8850f
to
4de876a
Compare
@njhill How can I fix CI errors ? It doesn't seems like to be caused by this change.. |
@junstar92 can you try merging the latest |
4de876a
to
538e2be
Compare
@hmellor I rebased it on the latest |
@junstar92 would you mind rebasing again and signing off your commits to satisfy the DCO check: https://github.com/vllm-project/vllm/pull/9276/checks?check_run_id=36686394447 .. thanks! |
538e2be
to
1496bcd
Compare
…step scheduling with skip tokenizer init option Signed-off-by: changjun.lee <pord7457@gmail.com>
1496bcd
to
af4c6f2
Compare
@njhill I just rebased it and signed off, thanks! |
…_tokenizer_init` with `num_scheduler_steps` (vllm-project#9276) Signed-off-by: changjun.lee <pord7457@gmail.com>
…_tokenizer_init` with `num_scheduler_steps` (vllm-project#9276) Signed-off-by: changjun.lee <pord7457@gmail.com>
…_tokenizer_init` with `num_scheduler_steps` (vllm-project#9276) Signed-off-by: changjun.lee <pord7457@gmail.com> Signed-off-by: Agata Dobrzyniewicz <adobrzyniewicz@habana.ai>
…_tokenizer_init` with `num_scheduler_steps` (vllm-project#9276) Signed-off-by: changjun.lee <pord7457@gmail.com> Signed-off-by: Mu Huai <tianbowen.tbw@antgroup.com>
…_tokenizer_init` with `num_scheduler_steps` (vllm-project#9276) Signed-off-by: changjun.lee <pord7457@gmail.com> Signed-off-by: Yuqi Zhang <yuqizhang@google.com>
…_tokenizer_init` with `num_scheduler_steps` (vllm-project#9276) Signed-off-by: changjun.lee <pord7457@gmail.com> Signed-off-by: minpeter <kali2005611@gmail.com>
When using
skip_tokenizer_init
withnum_scheduler_steps
(scheduling multiple steps), an error occurs due to an uninitialized tokenizer. This PR adds a condition to check tokenizer initialization assingle_step.py
did.