Clean up batch data structures: Introducing ModelWorkerBatch #1544
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
merrymercy
changed the title
|
This PR introduces some perf regression. I will send a patch soon. |
merrymercy
commented
Oct 2, 2024
| image_inputs = [r.image_inputs for r in self.reqs] | ||
|
|
||
| lora_paths = [req.lora_path for req in self.reqs] | ||
| self.sampling_info.regex_fsm_states = [req.regex_fsm_state for req in self.reqs] |
There was a problem hiding this comment.
possible overhead
merrymercy
commented
Oct 2, 2024
| if value is not None: # logit_bias can be None | ||
| setattr(self, item, value[new_indices]) | ||
|
|
||
| self.regex_fsms = [self.regex_fsms[i] for i in new_indices] |
There was a problem hiding this comment.
potential overhead
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Outdated suggestions cannot be applied.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.
The following is the flow of data structures for a batch:
ScheduleBatch -> ModelWorkerBatch -> ForwardBatch
scheduler.py::Scheduler.It contains high-level scheduling data. Most of the data is on the CPU.
tp_worker.py::TpModelWorker.model_runner.py::ModelRunner.It contains low-level tensor data. Most of the data consists of GPU tensors.