LLM: Partial Prefilling for Pipeline Parallel Serving #11457
Merged
Conversation
This file contains 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
xiangyuT
changed the title
qiyuangong
reviewed
Jul 2, 2024
| def prepare_batch(self, cur_batch): | ||
| if self.rank == 0: | ||
| cur_input_start = cur_batch.prefilled_index | ||
| if self.max_prefilled_seqs > 0: |
There was a problem hiding this comment.
Do we need to set cur_batch.partial_prefilling = 0, when max_prefilled_seqs==0.
qiyuangong
reviewed
Jul 2, 2024
| @@ -146,6 +146,8 @@ def pipeline_parallel(model, pipeline_parallel_stages): | |||
| model._modules['lm_head'] = DummyLayer() | |||
|
|
|||
| model.pipeline_parallel_stages = pipeline_parallel_stages | |||
| model.layer_start = layer_start | |||
There was a problem hiding this comment.
Can we replace gloabal layer_start and layer_end with model.layer-start etc
RyuKosei
pushed a commit
to RyuKosei/ipex-llm
that referenced
this pull request
Jul 19, 2024
…s#11457) LLM: Partial Prefilling for Pipeline Parallel Serving
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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Outdated suggestions cannot be applied.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.
Description
Add continuous-batching-like partial prefilling to reduce the memory peak during prefilling.
Below is an example with a
batch_size=4, max_prefilled_seqs=2BatchTask:4. How to test?
1234). And paste your action link here once it has been successfully finished.https://github.com/intel-analytics/ipex-llm-workflow/actions/runs/9802540844