-
-
Notifications
You must be signed in to change notification settings - Fork 15
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
Ysc pluggable scheduler #78
Conversation
Signed-off-by: Yannick Schnider <yannick.schnider1@ibm.com>
Signed-off-by: Yannick Schnider <yannick.schnider1@ibm.com>
👋 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:
🚀 |
Signed-off-by: Yannick Schnider <yannick.schnider1@ibm.com>
Signed-off-by: Yannick Schnider <yannick.schnider1@ibm.com>
Signed-off-by: Yannick Schnider <yannick.schnider1@ibm.com>
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.
LGTM
vllm/engine/llm_engine.py
Outdated
Scheduler = resolve_obj_by_qualname( | ||
self.vllm_config.parallel_config.scheduler_cls) | ||
self.scheduler = [ | ||
Scheduler( |
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.
I also think it is better with Scheduler
(with capital letter), but I've seen elsewhere in the code that they used cls
for example when retrieving the worker class:
cls = resolve_obj_by_qualname(target_worker_config.parallel_config.worker_cls)
target_worker = cls(*args, **kwargs)
Signed-off-by: Yannick Schnider <yannick.schnider1@ibm.com>
This PR moves the (spyre specific) scheduler class into the plugin repo
vllm-spyre
.it goes along with this ->PR<- in the
vllm-spyre
repositoryChanges:
vllm/config.py
: introducing new variablescheduler_cls
for classSchedulerConfig
(can be a path (str) or a class directly) and removing spyre specific code to 'load warmup shapes and sort by "speed"' (moved tovllm-spyre
)vllm/engine/llm_engine.py
: importingScheduler
based onvllm_config.scheduler_config.scheduler_cls
vllm/core/scheduler.py
: removing all spyre specific code from the scheduler logicvllm/envs.py
: removing all spyre related env variables (moved tovllm-spyre
)SchedulerConfig.spyre_warmup_shapes
invllm/vllm.config.py