[diffusion][CI] Add CI for MOVA model inference by CloudRipple · Pull Request #20430 · sgl-project/sglang

CloudRipple · 2026-03-12T08:33:47Z

Motivation

This pull request adds support for testing the MOVA-360p video generation model in the multimodal generation test suite. The changes include introducing the model constant, defining sampling parameters, and adding several test cases for different GPU configurations.

Modifications

Model integration:

Added the constant DEFAULT_MOVA_360P_MODEL_NAME_FOR_TEST for the MOVA-360p model in test_utils.py.
Included DEFAULT_MOVA_360P_MODEL_NAME_FOR_TEST in the list of available models in testcase_configs.py.

Sampling and test case setup:

Defined MOVA_I2V_360P_sampling_params with prompt, image, and video settings for MOVA-360p tests.
Added a single GPU test case for MOVA-360p to ONE_GPU_CASES_A.
Added multiple two-GPU test cases for MOVA-360p with different parallelism configurations to TWO_GPU_CASES_A.

Accuracy Tests

Benchmarking and Profiling

Checklist

Format your code according to the Format code with pre-commit.
Add unit tests according to the Run and add unit tests.
Update documentation according to Write documentations.
Provide accuracy and speed benchmark results according to Test the accuracy and Benchmark the speed.
Follow the SGLang code style guidance.

Review Process

Ping Merge Oncalls to start the PR flow. See the PR Merge Process.
Get approvals from CODEOWNERS and other reviewers.
Trigger CI tests with comments or contact authorized users to do so.
- /tag-run-ci-label, /rerun-failed-ci, /tag-and-rerun-ci
After green CI and required approvals, ask Merge Oncalls to merge.

Add CI to protect MOVA model

gemini-code-assist · 2026-03-12T08:34:01Z

Summary of Changes

Hello, I'm Gemini Code Assist¹! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed!

This pull request integrates the MOVA-360p video generation model into the existing multimodal generation test suite. It establishes the necessary configurations and introduces a series of test cases to validate the model's functionality and performance across different GPU setups, ensuring robust CI coverage for this new model.

Highlights

MOVA-360p Model Integration: The DEFAULT_MOVA_360P_MODEL_NAME_FOR_TEST constant was introduced and registered within the test configuration.
Sampling Parameters Defined: MOVA_I2V_360P_sampling_params were added, specifying prompt, image, video settings, and frame rates for MOVA-360p tests.
Comprehensive Test Cases Added: New test cases were included for MOVA-360p, covering both single-GPU and multiple two-GPU configurations with varying parallelism settings (TP, ring, ulysses degrees).

🧠 New Feature in Public Preview: You can now enable Memory to help Gemini Code Assist learn from your team's feedback. This makes future code reviews more consistent and personalized to your project's style. Click here to enable Memory in your admin console.

Changelog

python/sglang/multimodal_gen/test/server/testcase_configs.py
- Imported the DEFAULT_MOVA_360P_MODEL_NAME_FOR_TEST constant.
- Defined MOVA_I2V_360P_sampling_params for video generation.
- Added a DiffusionTestCase for MOVA-360p on a single GPU.
- Added three DiffusionTestCase entries for MOVA-360p on two GPUs, exploring different tensor parallelism, ring, and ulysses degrees.
python/sglang/multimodal_gen/test/test_utils.py
- Declared DEFAULT_MOVA_360P_MODEL_NAME_FOR_TEST with its model path.

Activity

The author completed several checklist items, including code formatting, adding unit tests, and adhering to SGLang code style guidance.

Using Gemini Code Assist

The full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips.

Invoking Gemini

You can request assistance from Gemini at any point by creating a comment using either /gemini <command> or @gemini-code-assist <command>. Below is a summary of the supported commands on the current page.

Feature	Command	Description
Code Review	`/gemini review`	Performs a code review for the current pull request in its current state.
Pull Request Summary	`/gemini summary`	Provides a summary of the current pull request in its current state.
Comment	@gemini-code-assist	Responds in comments when explicitly tagged, both in pull request comments and review comments.
Help	`/gemini help`	Displays a list of available commands.

Customization

To customize Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a .gemini/ folder in the base of the repository. Detailed instructions can be found here.

Limitations & Feedback

Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counter productive. You can react with 👍 and 👎 on @gemini-code-assist comments. If you're interested in giving your feedback about your experience with Gemini Code Assist for GitHub and other Google products, sign up here.

You can also get AI-powered code generation, chat, as well as code reviews directly in the IDE at no cost with the Gemini Code Assist IDE Extension.

Review the Privacy Notices, Generative AI Prohibited Use Policy, Terms of Service, and learn how to configure Gemini Code Assist in GitHub here. Gemini can make mistakes, so double check it and use code with caution. ↩

gemini-code-assist

Code Review

This pull request correctly adds CI tests for the new MOVA-360p model, including configurations for both single and multi-GPU setups with various parallelism strategies. The changes are functional and align with the goal of expanding test coverage. I have a couple of suggestions to enhance code organization and maintainability, primarily by adding a grouping comment for the new model constant and refactoring duplicated test case configurations to improve clarity and reduce redundancy.

gemini-code-assist · 2026-03-12T08:38:46Z

+    DiffusionTestCase(
+        "mova_360p_tp2",
+        DiffusionServerArgs(
+            model_path=DEFAULT_MOVA_360P_MODEL_NAME_FOR_TEST,
+            modality="video",
+            num_gpus=2,
+            tp_size=2,
+            dit_layerwise_offload=True,
+        ),
+        MOVA_I2V_360P_sampling_params,
+        run_perf_check=False,
+    ),
+    DiffusionTestCase(
+        "mova_360p_ring1_uly2",
+        DiffusionServerArgs(
+            model_path=DEFAULT_MOVA_360P_MODEL_NAME_FOR_TEST,
+            modality="video",
+            num_gpus=2,
+            ring_degree=1,
+            ulysses_degree=2,
+            dit_layerwise_offload=True,
+        ),
+        MOVA_I2V_360P_sampling_params,
+        run_perf_check=False,
+    ),
+    DiffusionTestCase(
+        "mova_360p_ring2_uly1",
+        DiffusionServerArgs(
+            model_path=DEFAULT_MOVA_360P_MODEL_NAME_FOR_TEST,
+            modality="video",
+            num_gpus=2,
+            ring_degree=2,
+            ulysses_degree=1,
+            dit_layerwise_offload=True,
+        ),
+        MOVA_I2V_360P_sampling_params,
+        run_perf_check=False,
+    ),


There is significant duplication in the DiffusionServerArgs for the new MOVA 2-GPU test cases. To improve maintainability, consider defining a base configuration object for the common arguments and then use dataclasses.replace to create the specific variations for each test case. This would make it easier to update common parameters in the future.

For example:

# Define a base config before the list base_mova_2gpu_args = DiffusionServerArgs( model_path=DEFAULT_MOVA_360P_MODEL_NAME_FOR_TEST, modality="video", num_gpus=2, dit_layerwise_offload=True, ) # ... TWO_GPU_CASES_A = [ # ... other cases DiffusionTestCase( "mova_360p_tp2", dataclasses.replace(base_mova_2gpu_args, tp_size=2), MOVA_I2V_360P_sampling_params, run_perf_check=False, ), # ... other mova cases using replace ]

mickqian · 2026-03-15T13:06:24Z

+            num_gpus=1,
+            dit_layerwise_offload=True,
+        ),
+        MOVA_I2V_360P_sampling_params,


could we reuse the existing sampling_params?

ok, we fixed this @mickqian

…_360P_sampling_params with TI2V_sampling_params (#8)

mickqian · 2026-03-17T13:54:23Z

/tag-and-rerun-ci

yhyang201 · 2026-03-17T14:08:38Z

/tag-and-rerun-ci

yhyang201 · 2026-03-17T15:42:14Z

/rerun-failed-ci

yhyang201 · 2026-03-17T15:42:20Z

/rerun-failed-ci

yhyang201 · 2026-03-17T17:52:01Z

/rerun-failed-ci

yhyang201 · 2026-03-17T17:52:07Z

/rerun-failed-ci

yhyang201 · 2026-03-17T19:32:49Z

/rerun-failed-ci

yhyang201 · 2026-03-18T08:01:05Z

/tag-and-rerun-ci

yhyang201 · 2026-03-18T19:01:55Z

/rerun-failed-ci

yhyang201 · 2026-03-18T19:02:01Z

/rerun-failed-ci

yhyang201 · 2026-03-19T01:52:48Z

/rerun-failed-ci

yhyang201 · 2026-03-19T01:52:51Z

/rerun-failed-ci

yhyang201 · 2026-03-19T08:31:45Z

/rerun-failed-ci

yhyang201 · 2026-03-19T22:48:30Z

/rerun-failed-ci

yhyang201 · 2026-03-19T22:48:39Z

/rerun-failed-ci

yhyang201 · 2026-03-20T02:00:46Z

/rerun-failed-ci

yhyang201 · 2026-03-20T02:00:49Z

/rerun-failed-ci

yhyang201 · 2026-03-20T10:55:27Z

/rerun-failed-ci

yhyang201 · 2026-03-20T10:55:30Z

/rerun-failed-ci

yhyang201 · 2026-03-20T11:26:48Z

/tag-and-rerun-ci

yhyang201 · 2026-03-21T05:23:55Z

/rerun-failed-ci

yhyang201 · 2026-03-21T05:24:01Z

/rerun-failed-ci

yhyang201 · 2026-03-21T07:39:56Z

/rerun-failed-ci

yhyang201 · 2026-03-21T08:57:20Z

/rerun-failed-ci

yhyang201 · 2026-03-21T09:25:09Z

/rerun-failed-ci

yhyang201 · 2026-03-24T12:37:13Z

/tag-and-rerun-ci

yhyang201 · 2026-03-24T15:39:49Z

/rerun-failed-ci

yhyang201 · 2026-03-24T16:40:31Z

/rerun-failed-ci

yhyang201 · 2026-03-24T17:10:26Z

/rerun-failed-ci

yhyang201 · 2026-03-24T17:45:02Z

/rerun-failed-ci

ping1jing2 · 2026-03-24T18:28:54Z

i merged it as Mick approved and all CIs passed

Co-authored-by: Luo <139519292+0-693@users.noreply.github.com> Co-authored-by: Mick <mickjagger19@icloud.com>

0-693 and others added 5 commits March 12, 2026 12:18

[diffusion][CI] Add MOVA inference CI

87ff946

Add CI to protect MOVA model

Merge branch 'sgl-project:main' into main

b56218a

fix lint

c32c7b7

align to sglang base (#7)

3dbe962

Merge branch 'sgl-project:main' into main

376a849

CloudRipple requested review from mickqian, ping1jing2 and yhyang201 as code owners March 12, 2026 08:33

github-actions Bot added the diffusion SGLang Diffusion label Mar 12, 2026

gemini-code-assist Bot reviewed Mar 12, 2026

View reviewed changes

mickqian reviewed Mar 15, 2026

View reviewed changes

0-693 added 2 commits March 16, 2026 15:16

[CI] Replace MOVA_I2V_360P_sampling_params with TI2V_sampling_params

02e6711

[CI] Add MOVA video generation models tag for tests. Replace MOVA_I2V…

f7c01a7

…_360P_sampling_params with TI2V_sampling_params (#8)

CloudRipple requested a review from mickqian March 17, 2026 06:56

mickqian approved these changes Mar 17, 2026

View reviewed changes

github-actions Bot added the run-ci label Mar 17, 2026

Merge branch 'main' into main

d0c28e3

Merge branch 'main' into main

84d6cc3

Merge branch 'main' into main

9b97a4f

sglang-npu-bot merged commit a3ed2e4 into sgl-project:main Mar 24, 2026
84 of 89 checks passed

adityavaid pushed a commit to adityavaid/sglang that referenced this pull request Mar 24, 2026

[diffusion][CI] Add CI for MOVA model inference (sgl-project#20430)

ed22c59

Co-authored-by: Luo <139519292+0-693@users.noreply.github.com> Co-authored-by: Mick <mickjagger19@icloud.com>

adityavaid pushed a commit to adityavaid/sglang that referenced this pull request Mar 24, 2026

[diffusion][CI] Add CI for MOVA model inference (sgl-project#20430)

34890e7

Co-authored-by: Luo <139519292+0-693@users.noreply.github.com> Co-authored-by: Mick <mickjagger19@icloud.com>

johnnycxm pushed a commit to johnnycxm/sglang that referenced this pull request Mar 25, 2026

[diffusion][CI] Add CI for MOVA model inference (sgl-project#20430)

c658e45

Co-authored-by: Luo <139519292+0-693@users.noreply.github.com> Co-authored-by: Mick <mickjagger19@icloud.com>

johnnycxm pushed a commit to johnnycxm/sglang that referenced this pull request Mar 25, 2026

[diffusion][CI] Add CI for MOVA model inference (sgl-project#20430)

19639b0

Co-authored-by: Luo <139519292+0-693@users.noreply.github.com> Co-authored-by: Mick <mickjagger19@icloud.com>

JustinTong0323 pushed a commit to JustinTong0323/sglang that referenced this pull request Apr 7, 2026

[diffusion][CI] Add CI for MOVA model inference (sgl-project#20430)

1d7e72d

Co-authored-by: Luo <139519292+0-693@users.noreply.github.com> Co-authored-by: Mick <mickjagger19@icloud.com>

yhyang201 pushed a commit to yhyang201/sglang that referenced this pull request Apr 22, 2026

[diffusion][CI] Add CI for MOVA model inference (sgl-project#20430)

f335d26

Co-authored-by: Luo <139519292+0-693@users.noreply.github.com> Co-authored-by: Mick <mickjagger19@icloud.com>

Conversation

CloudRipple commented Mar 12, 2026

Motivation

Modifications

Accuracy Tests

Benchmarking and Profiling

Checklist

Review Process

Uh oh!

gemini-code-assist Bot commented Mar 12, 2026

Summary of Changes

Highlights

Footnotes

Uh oh!

gemini-code-assist Bot left a comment

Choose a reason for hiding this comment

Code Review

Uh oh!

gemini-code-assist Bot Mar 12, 2026

Choose a reason for hiding this comment

Uh oh!

Uh oh!

mickqian Mar 15, 2026

Choose a reason for hiding this comment

Uh oh!

CloudRipple Mar 16, 2026

Choose a reason for hiding this comment

Uh oh!

mickqian commented Mar 17, 2026

Uh oh!

yhyang201 commented Mar 17, 2026

Uh oh!

yhyang201 commented Mar 17, 2026

Uh oh!

yhyang201 commented Mar 17, 2026

Uh oh!

yhyang201 commented Mar 17, 2026

Uh oh!

yhyang201 commented Mar 17, 2026

Uh oh!

yhyang201 commented Mar 17, 2026

Uh oh!

yhyang201 commented Mar 18, 2026

Uh oh!

yhyang201 commented Mar 18, 2026

Uh oh!

yhyang201 commented Mar 18, 2026

Uh oh!

yhyang201 commented Mar 19, 2026

Uh oh!

yhyang201 commented Mar 19, 2026

Uh oh!

yhyang201 commented Mar 19, 2026

Uh oh!

yhyang201 commented Mar 19, 2026

Uh oh!

yhyang201 commented Mar 19, 2026

Uh oh!

yhyang201 commented Mar 20, 2026

Uh oh!

yhyang201 commented Mar 20, 2026

Uh oh!

yhyang201 commented Mar 20, 2026

Uh oh!

yhyang201 commented Mar 20, 2026

Uh oh!

yhyang201 commented Mar 20, 2026

Uh oh!

yhyang201 commented Mar 21, 2026

Uh oh!

yhyang201 commented Mar 21, 2026

Uh oh!

yhyang201 commented Mar 21, 2026

Uh oh!

yhyang201 commented Mar 21, 2026

Uh oh!

yhyang201 commented Mar 21, 2026

Uh oh!

yhyang201 commented Mar 24, 2026

Uh oh!