[0.7.3] Optimize apply_penalties & topKtopP for both V0/V1 Engine #525
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
linfeng-yuan
force-pushed
the
v0.7.3-dev
branch
6 times, most recently
from
873e680 to
7f46d64
Compare
wangxiyuan
changed the title
46 tasks
Signed-off-by: linfeng-yuan <1102311262@qq.com>
Signed-off-by: linfeng-yuan <1102311262@qq.com>
linfeng-yuan
force-pushed
the
v0.7.3-dev
branch
from
7f46d64 to
d377ba3
Compare
ganyi1996ppo
approved these changes
Apr 28, 2025
wangxiyuan
approved these changes
Apr 28, 2025
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Outdated suggestions cannot be applied.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.
This PR optimizes apply_penalties & topKtopP implementation in both V0/V1 Engine by avoiding using torch.scatter and matrix indexing operations.
We verified the functionality of this PR using Qwen2.5-72B-Instruct. At a concurrency of 40 and with post-processing parameters set to "temperature": 0.3, "top_k": 100, "top_p": 0.9, "repetition_penalty": 1.01, the average decoding time was reduced from 300ms to 50ms.