[core] [3/N] multi-step args and sequence.py #7452
Conversation
|
/ready |
|
👋 Hi! Thank you for contributing to the vLLM project. Once the PR is approved and ready to go, please make sure to run full CI as it is required to merge (or just use auto-merge). To run full CI, you can do one of these:
🚀 |
github-actions
bot
added
the
ready
|
|
||
| @dataclass | ||
| class SequenceGroupState: | ||
| """Mutable state tied to a specific sequence group""" |
There was a problem hiding this comment.
I wonder if this should be a part of SequenceData? All the states are now stored in SequenceData now actually
There was a problem hiding this comment.
also IIUC num_steps is not state, and it may not belong here?
There was a problem hiding this comment.
perhaps eventually we will support different num_steps in a batch and so we track this per SequenceGroup. There used to be a SequenceGroupState used for seed generator but has since been removed. #6698
| assert self._token_chunk_size is not None | ||
| return self._token_chunk_size | ||
|
|
||
| def finish_step(self) -> None: |
There was a problem hiding this comment.
not used in this PR. Will it be used in the next PR with a scheduler?
There was a problem hiding this comment.
yes it will be used in AsyncLLMEngine
| # steps | ||
| assert len(self.seq_group_metadata_list) > 0 | ||
| first_seq_group = self.seq_group_metadata_list[0] | ||
| return first_seq_group.state.current_step == 0 |
There was a problem hiding this comment.
so this means that the state of multi step is decided by the state of the first sequence group in a batch? E.g., if the first seq group has only 1 step left, we only run 1 step although num_steps > 1?
There was a problem hiding this comment.
yes, eventually we may change this behavior and allow sequences in a batch to concurrently be on different steps
There was a problem hiding this comment.
Is this really a necessary feature? I personally think it should be ok to assume we run all the requests in the batch for the same number of steps.
|
|
|
Merge first to unblock the follow-up tasks. We could revisit/refactor this later. |
| from typing import (TYPE_CHECKING, Dict, List, Mapping, Optional, Set, Tuple, | ||
| Union, cast) | ||
|
|
||
| import numpy |
There was a problem hiding this comment.
Nit:
| import numpy | |
| import numpy as np |
| # steps | ||
| assert len(self.seq_group_metadata_list) > 0 | ||
| first_seq_group = self.seq_group_metadata_list[0] | ||
| return first_seq_group.state.current_step == 0 |
There was a problem hiding this comment.
Is this really a necessary feature? I personally think it should be ok to assume we run all the requests in the batch for the same number of steps.
Signed-off-by: Alvant <[email protected]>
Signed-off-by: LeiWang1999 <[email protected]>
4 participants
cc @zhuohan123 @comaniac
For #7000
FIX #xxxx (link existing issues this PR will resolve)
BEFORE SUBMITTING, PLEASE READ THE CHECKLIST BELOW AND FILL IN THE DESCRIPTION ABOVE
PR Checklist (Click to Expand)
Thank you for your contribution to vLLM! Before submitting the pull request, please ensure the PR meets the following criteria. This helps vLLM maintain the code quality and improve the efficiency of the review process.
PR Title and Classification
Only specific types of PRs will be reviewed. The PR title is prefixed appropriately to indicate the type of change. Please use one of the following:
[Bugfix]for bug fixes.[CI/Build]for build or continuous integration improvements.[Doc]for documentation fixes and improvements.[Model]for adding a new model or improving an existing model. Model name should appear in the title.[Frontend]For changes on the vLLM frontend (e.g., OpenAI API server,LLMclass, etc.)[Kernel]for changes affecting CUDA kernels or other compute kernels.[Core]for changes in the core vLLM logic (e.g.,LLMEngine,AsyncLLMEngine,Scheduler, etc.)[Hardware][Vendor]for hardware-specific changes. Vendor name should appear in the prefix (e.g.,[Hardware][AMD]).[Misc]for PRs that do not fit the above categories. Please use this sparingly.Note: If the PR spans more than one category, please include all relevant prefixes.
Code Quality
The PR need to meet the following code quality standards:
format.shto format your code.docs/source/if the PR modifies the user-facing behaviors of vLLM. It helps vLLM user understand and utilize the new features or changes.Notes for Large Changes
Please keep the changes as concise as possible. For major architectural changes (>500 LOC excluding kernel/data/config/test), we would expect a GitHub issue (RFC) discussing the technical design and justification. Otherwise, we will tag it with
rfc-requiredand might not go through the PR.What to Expect for the Reviews
The goal of the vLLM team is to be a transparent reviewing machine. We would like to make the review process transparent and efficient and make sure no contributor feel confused or frustrated. However, the vLLM team is small, so we need to prioritize some PRs over others. Here is what you can expect from the review process:
action-requiredlabel on the PR if there are changes required. The contributor should address the comments and ping the reviewer to re-review the PR.Thank You
Finally, thank you for taking the time to read these guidelines and for your interest in contributing to vLLM. Your contributions make vLLM a great tool for everyone!