feat: add configurable scheduling policy#119
Open
Alise-svg wants to merge 2 commits into
Open
Conversation
DarkSharpness
requested changes
May 10, 2026
| self.prefill_manager.schedule_next_batch(self.prefill_budget) | ||
| or self.decode_manager.schedule_next_batch() | ||
| ) | ||
| if self.schedule_policy == "decode_first": |
Collaborator
There was a problem hiding this comment.
A decode first policy should be:
- Form a decode batch first.
- Try to schedule a prefill batch with the remaining token budget (prefill_budget - decode_tokens).
This is actually mix prefill-decode style batching.
- Add --schedule-policy command line argument - Support 'prefill_first' (default) and 'decode_first' policies - prefill_first reduces TTFT for online serving - decode_first improves throughput for offline inference
Alise-svg
force-pushed
the
feature/configurable-scheduling-policy
branch
from
20dbc44 to
3d221e8
Compare
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Add --schedule-policy command line argument to allow users to choose
between different scheduling strategies for batch formation.
Supported policies:
Time To First Token (TTFT) for online serving scenarios where
users wait for the first token.
for offline batch inference where maximizing token generation
rate is more important than latency.
Usage:
python -m minisgl --model "Qwen/Qwen3-0.6B" --schedule-policy decode_first