Add vllm Omni doc

[Yadan-Wei]

Purpose

Add public documentation and runnable examples for the vLLM-Omni DLC (version 0.18.0) on EC2 and SageMaker. Covers TTS, image generation, video generation, and multimodal chat (Qwen2.5-Omni).

Also fixes image URIs in the generator: vLLM-Omni images live in the vllm ECR repo with tags omni-cuda-v1 (EC2) and omni-sagemaker-cuda-v1 (SageMaker). Introduces an optional ecr_repository YAML field so the data-dir key can differ from the actualbECR repo name (backward-compatible — other frameworks are unchanged).

Test Plan

1. cd docs && python src/main.py — confirm reference/available_images.md has a vLLM-Omni section pointing at gallery.ecr.aws/deep-learning-containers/vllm with tags omni-cuda-v1 and omni-sagemaker-cuda-v1.
2. Confirm releasenotes/vllm-omni/vllm-omni-0.18.0-gpu-{default,sagemaker}.md show vllm:omni-cuda-v1 / vllm:omni-sagemaker-cuda-v1.
3. Verify EC2 deployment works end-to-end for TTS, image, video, and Qwen2.5-Omni examples.
4. Verify SageMaker deployment works end-to-end for TTS; confirm SageMaker async video matches the documented limitation (job-ID JSON only, no MP4).

Test Result

EC2 examples verified end-to-end in a prior session (TTS, image, video, Qwen2.5-Omni).

SageMaker verified in account 897880167187 (us-west-2):

Scenario	Instance	Result
Real-time TTS (deploy_tts.py)	ml.g5.xlarge	Deploy 664s; first invoke 71.2s (torch.compile warmup); response 176684 bytesbaudio/wav, valid 24kHz 16-bit mono
Async video (informational)	ml.g5.12xlarge (TP=2)	Deploy 606s; invoke returns 202; S3 output is 420-byte JSON with status: queued — confirms the known limitation, MP4 is not retrievable via SageMaker async in 0.18.0. Use EC2 for video.

Endpoints were deleted after verification.

pre-commit run passes (ruff, gitleaks, flowmark, typos, shfmt, signoff).

PR Checklist

• I ran pre-commit run --all-files locally before creating this PR.

---

Toggle if you are merging into master Branch

By default, docker image builds and tests are disabled. Two ways to run builds and tests:

1. Using dlc_developer_config.toml
2. Using this PR description (currently only supported for PyTorch, TensorFlow, vllm, and base images)

How to use the helper utility for updating dlc_developer_config.toml

Assuming your remote is called origin (you can find out more with git remote -v)...

• Run default builds and tests for a particular buildspec - also commits and pushes changes to remote; Example:

python src/prepare_dlc_dev_environment.py -b </path/to/buildspec.yml> -cp origin

• Enable specific tests for a buildspec or set of buildspecs - also commits and pushes changes to remote; Example:

python src/prepare_dlc_dev_environment.py -b </path/to/buildspec.yml> -t sanity_tests -cp origin

• Restore TOML file when ready to merge

python src/prepare_dlc_dev_environment.py -rcp origin

NOTE: If you are creating a PR for a new framework version, please ensure success of the local, standard, rc, and efa sagemaker tests by updating the dlc_developer_config.toml file:

• sagemaker_remote_tests = true
• sagemaker_efa_tests = true
• sagemaker_rc_tests = true
• sagemaker_local_tests = true

How to use PR description

Use the code block below to uncomment commands and run the PR CodeBuild jobs. There are two commands available:

• # /buildspec <buildspec_path>
  • e.g.: # /buildspec pytorch/training/buildspec.yml
  • If this line is commented out, dlc_developer_config.toml will be used.
• # /tests <test_list>
  • e.g.: # /tests sanity security ec2
  • If this line is commented out, it will run the default set of tests (same as the defaults in dlc_developer_config.toml): sanity, security, ec2, ecs, eks, sagemaker, sagemaker-local.

# /buildspec <buildspec_path>
# /tests <test_list>

Toggle if you are merging into main Branch

PR Checklist

• [] I ran pre-commit run --all-files locally before creating this PR. (Read DEVELOPMENT.md for details).