chore(sglang): bump to 0.5.17 - #13343
Merged
Merged
Conversation
jain-ria
force-pushed
the
jain-ria/sgl-to-0.5.17-and-cleanups
branch
from
August 17, 2026 16:21
5909f89 to
de4709c
Compare
jain-ria
force-pushed
the
jain-ria/sgl-to-0.5.17-and-cleanups
branch
from
August 17, 2026 17:59
de4709c to
6a32da6
Compare
Contributor
Author
|
/ok to test 6a32da6 |
Contributor
This comment has been minimized.
This comment has been minimized.
jain-ria
force-pushed
the
jain-ria/sgl-to-0.5.17-and-cleanups
branch
from
August 17, 2026 18:48
6a32da6 to
08c85e6
Compare
Contributor
Author
|
/ok to test 08c85e6 |
Signed-off-by: jain-ria <riajain@NVIDIA.com>
Signed-off-by: jain-ria <riajain@NVIDIA.com>
Signed-off-by: jain-ria <riajain@NVIDIA.com>
Signed-off-by: jain-ria <riajain@NVIDIA.com>
Signed-off-by: jain-ria <riajain@NVIDIA.com>
jain-ria
force-pushed
the
jain-ria/sgl-to-0.5.17-and-cleanups
branch
from
August 17, 2026 20:22
08c85e6 to
8c6f5b6
Compare
Contributor
Author
|
/ok to test 8c6f5b6 |
jain-ria
marked this pull request as ready for review
August 17, 2026 21:25
Signed-off-by: Ishan Dhanani <ishandhanani@gmail.com>
furionw
requested changes
Aug 18, 2026
jthomson04
reviewed
Aug 18, 2026
nv-anants
reviewed
Aug 18, 2026
Signed-off-by: jain-ria <riajain@NVIDIA.com>
|
Should I wait with the merge and update the dependency to 0.5.18 right away? The release branch is https://github.com/sgl-project/sglang/tree/release/v0.5.18 |
Signed-off-by: jain-ria <riajain@NVIDIA.com>
Signed-off-by: jain-ria <riajain@NVIDIA.com>
jain-ria
force-pushed
the
jain-ria/sgl-to-0.5.17-and-cleanups
branch
from
August 20, 2026 15:12
8891b0e to
9e7d94e
Compare
nv-anants
approved these changes
Aug 20, 2026
Contributor
|
/ok to test 9e7d94e |
Contributor
|
@Swipe4057 - gonna get this PR in first. And then will kick off another PR on top for the 0.5.18 branch. This PR has a lot of things to use new serverags. BTW if you want to try it feel free |
galletas1712
approved these changes
Aug 20, 2026
Signed-off-by: jain-ria <riajain@NVIDIA.com>
Contributor
Author
|
/ok to test c7cb8e4 |
Contributor
Author
|
@Swipe4057 started working on 0.5.18, see #13643 |
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
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.
Summary
lmsysorg/sglang:v0.5.17-cu130-runtimeServerArgsupdates to SGLang's auditedoverride()API, including GMS and snapshot/FPM pathsUpstream source validation used SGLang tag
v0.5.17at29481685462732237d80d86076d6563e1f658102. The published CUDA runtime tags were verified as multi-architecture images:v0.5.17-cu130-runtimeandv0.5.17-cu129-runtime. Dynamo main currently defines only the CUDA 13.0 SGLang runtime variant.The supported CUDA engine window is 0.5.17/0.5.16; both expose
ServerArgs.override()and_update_weight_version_if_provided(). The XPU image remains separately pinned to SGLang 0.5.11, so the compatibility helper keeps an explicitly documented legacy assignment fallback until that image reaches 0.5.16+.Validation
v0.5.17,[diffusion]extra, rebuilt Dynamo native bindingsbash -npassed; changed deployment YAML parsed successfullyH100 validation with
lmsysorg/sglang:v0.5.17-cu130-runtime:agg.shagg_embed.shagg_router.shagg_vision.shwith inline image inputdisagg.shdisagg_router.shdisagg_same_gpu.shdiffusion_llada.shwith generated textimage_diffusion.sh --model-path Tongyi-MAI/Z-Image-Turbo; generated a non-empty PNG through/v1/images/generationstext-to-video-diffusion.shwith an MP4 outputmultimodal_epd.shwith inline image inputmultimodal_disagg.shthrough the exact pre-merge casetest_sglang_deployment[multimodal_disagg_qwen-4]; HTTP 200 and the response correctly described the pictured bus (repeated before and after the lifecycle fix)agg_multimodal_router.shwith inline image inputagg_agent.shwith its defaultzai-org/GLM-4.7-Flashmodel (TP=2); model discovery and chat completion returned HTTP 200 with parsed reasoning/content fieldsThe earlier multimodal-disaggregated semantic-corruption observation was not reproducible in the clean exact-tag environment. Two independent temperature-zero runs returned the correct image description, including after the prefill lifecycle fix, so there is no remaining known GPU correctness issue from that report.
Summary by CodeRabbit
Related Issues