feat: use generic image and use single node for oss-gpt-120b recipe

[biswapanda]

Overview:

• use generic image
• use single node
• fix hub path

Details:

Where should the reviewer start?

Related Issues: (use one of the action keywords Closes / Fixes / Resolves / Relates to)

• closes GitHub issue: #xxx

Summary by CodeRabbit

• Bug Fixes

  • Corrected model path to ensure models load reliably.

• Chores

  • Switched container images to a private registry for improved deployment consistency.
  • Adjusted default replica count to a single instance for streamlined deployments.
  • Tuned performance configuration to use fewer GPUs, aligning perf runs with typical resource availability.

[coderabbitai]

Walkthrough

Updated TRT-LLM deployment configs: switched container image registry/tag, reduced replicas, adjusted MODEL_PATH prefix, and scaled down GPU count in perf settings. No new features or API changes.

Changes

Cohort / File(s)	Summary of Changes
TRT-LLM deployment configs recipes/gpt-oss-120b/trtllm/agg/deploy.yaml	Changed image from nvcr.io/nvidia/ai-dynamo/tensorrtllm-runtime:0.5.1-rc0.pre3 to my-registry/trtllm-runtime:my-tag in three containers; reduced replicas 18→1 across three sections; updated MODEL_PATH from /model-store/models--openai--gpt-oss-120b/... to /model-store/hub/models--openai--gpt-oss-120b/... in two places.
Perf scaling config recipes/gpt-oss-120b/trtllm/agg/perf.yaml	Reduced DEPLOYMENT_GPU_COUNT from "72" to "4", affecting derived concurrency and perf inputs; no other logic/path changes.

Estimated code review effort

🎯 2 (Simple) | ⏱️ ~10 minutes

Poem

I thump my paws—new tags, fewer crews,
One GPU burrow, not seventy-two’s.
Paths hop to “hub,” replicas rest,
Leaner carrots, same warm nest.
In quiet racks, I twitch with glee—
Configs trimmed, swift as a bunny.

Pre-merge checks

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Description Check	⚠️ Warning	The pull request description includes the Overview and Related Issues sections but leaves the Details and Where should the reviewer start sections as unmodified placeholders, providing no concrete information about the specific changes or files that need attention, which makes the description incomplete for effective review.	Please populate the Details section with a clear summary of the file changes and update the Where should the reviewer start section to call out the specific files or areas that require focused review.

✅ Passed checks (2 passed)

Check name	Status	Explanation
Title Check	✅ Passed	The pull request title clearly and concisely summarizes the main changes by indicating the switch to a generic image and reducing to a single node for the oss-gpt-120b recipe, directly reflecting the core updates in the deployment configurations. It uses specific terminology and avoids vague language, making it easy for reviewers to grasp the primary intent at a glance. The title is appropriately scoped and aligned with the changeset.
Docstring Coverage	✅ Passed	No functions found in the changes. Docstring coverage check skipped.

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}