Cherry pick of delayed sampling #263
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
hsubramony
force-pushed
the
cherry_pick_of_delayed_sampling
branch
from
b60cf3d
to
d6616c6
Compare
libinta
added a commit
that referenced
this pull request
Sep 10, 2024
tzielinski-habana
force-pushed
the
cherry_pick_of_delayed_sampling
branch
from
32173ca
to
d45f0c9
Compare
michalkuligowski
approved these changes
Sep 17, 2024
jedrzejmyrcha
force-pushed
the
v1.18.0
branch
from
4173298
to
4c1ca3a
Compare
tzielinski-habana
force-pushed
the
cherry_pick_of_delayed_sampling
branch
from
d45f0c9
to
e140ec0
Compare
kzawora-intel
approved these changes
Sep 19, 2024
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Outdated suggestions cannot be applied.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.
Cherry pick of delayed sampling from habana_next. To use it, pass additional flags to vllm server:
--num-lookahead-slots 1 --use-v2-block-manager --enable-delayed-sampling
Tested on 1 Gaudi2 card on llama2 7b on a 4k offline accuracy test. Accuracy is good (>100%), observed perf improvement is ~38%.