Qwen2.5_vl multi frame inference #591
Draft
+374
−0
Conversation
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
Signed-off-by: Mohit Soni <[email protected]>
| if x.startswith("past_") or x.endswith("_RetainedState") | ||
| ] | ||
| ) | ||
|
|
There was a problem hiding this comment.
Shouldn't we skip vision embedding buffers after the prefill stage? Otherwise, those large inputs will dramatically increase the decode time. I suggested to Chulhee that he add the following lines here:
Suggested change
| if not_mllama: | |
| lang_session.skip_buffers(vision_outputs.keys()) |
And with this change, he was able to increase the decode rate from 13 t/s to 20 t/s.
Could you incorporate this change into the PR, and possibly apply a similar update to the kv_offload_generate function as well? Thank you!
Contributor
There was a problem hiding this comment.
Thanks @quic-xiyushi, This is resolved, I have pushed the changes, pls check.
Signed-off-by: vbaddi <[email protected]>
Signed-off-by: vbaddi <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 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Outdated suggestions cannot be applied.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.
No description provided.