[https://nvbugs/6329052][fix] Change the config for DSV3 disagg conditional test

[Shixiaowei02]

Description

This pull request updates the configuration for DeepSeek-V3-Lite and adjusts test waivers for related disaggregated tests. The main changes are the addition of new model configuration options and the removal of test waivers for specific DeepSeek-V3-Lite disaggregated tests, likely reflecting improved support or bug fixes.

Configuration updates:

• Updated disagg_config_cache_reuse_deepseek_v3.yaml to set attn_backend to FLASHINFER and added model_kwargs with num_hidden_layers: 4 for the DeepSeek-V3-Lite model.

Test waiver updates:

• Removed waivers for the test test_workers_conditional_disaggregation_deepseek_v3_lite_bf16[DeepSeek-V3-Lite-bf16] from both A100 and L40S test lists, indicating this test is no longer being skipped due to bug 6329052. [1] [2]

Test Coverage

PR Checklist

Please review the following before submitting your PR:

• PR description clearly explains what and why. If using CodeRabbit's summary, please make sure it makes sense.

• PR Follows TRT-LLM CODING GUIDELINES to the best of your knowledge.

• Test cases are provided for new code paths (see test instructions)

• If PR introduces API changes, an appropriate PR label is added - either api-compatible or api-breaking. For api-breaking, include BREAKING in the PR title.

• Any new dependencies have been scanned for license and vulnerabilities

• CODEOWNERS updated if ownership changes

• Documentation updated as needed

• Update tava architecture diagram if there is a significant design change in PR.

• The reviewers assigned automatically/manually are appropriate for the PR.

• Please check this after reviewing the above items as appropriate for this PR.

GitHub Bot Help

To see a list of available CI bot commands, please comment /bot help.

Summary by CodeRabbit

• Tests
  • Updated DeepSeek-V3-Lite integration settings to use the FLASHINFER attention backend and a four-layer model configuration.
  • Removed skip waivers for DeepSeek-V3-Lite worker tests on A100 and L40S hardware, allowing these tests to run normally.

[coderabbitai]

📝 Walkthrough

Walkthrough

The DeepSeek-V3-Lite disaggregated test configuration now uses FLASHINFER with four hidden layers. Its A100 and L40S test waive entries are removed.

Changes

DeepSeek disaggregation test enablement

Layer / File(s)	Summary
Configure and enable DeepSeek-V3-Lite coverage tests/integration/defs/disaggregated/test_configs/disagg_config_cache_reuse_deepseek_v3.yaml, tests/integration/test_lists/waives.txt	The configuration adds FLASHINFER attention and four hidden layers, and removes the matching A100 and L40S waive entries.

Estimated code review effort: 1 (Trivial) | ~5 minutes

Suggested reviewers: yuxianq

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Title check	✅ Passed	The title matches the main change and includes a valid bug reference and fix type.
Description check	✅ Passed	The description covers the issue, solution, and checklist; only Test Coverage is left empty.

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

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

[Shixiaowei02]

/bot run --disable-fail-fast --stage-list "DGX_H100-4_GPUs-PyTorch-DeepSeek-1"

[tensorrt-cicd]

PR_Github #59913 [ run ] triggered by Bot. Commit: 28b86d0 Link to invocation

[tensorrt-cicd]

PR_Github #59913 [ run ] completed with state FAILURE. Commit: 28b86d0
/LLM/main/L0_MergeRequest_PR pipeline #48315 (Partly Tested) completed with status: 'ABORTED'

CI Report

⚠️ Action Required:

• Please check the failed tests and fix your PR
• If you cannot view the failures, ask the CI triggerer to share details
• Once fixed, request an NVIDIA team member to trigger CI again

Link to invocation

[Shixiaowei02]

/bot run --disable-fail-fast --stage-list "DGX_H100-4_GPUs-PyTorch-DeepSeek-1"

[tensorrt-cicd]

PR_Github #59916 [ run ] triggered by Bot. Commit: 31c45b9 Link to invocation

[tensorrt-cicd]

PR_Github #59916 [ run ] completed with state SUCCESS. Commit: 31c45b9
/LLM/main/L0_MergeRequest_PR pipeline #48318 (Partly Tested) completed with status: 'SUCCESS'

CI Report

Link to invocation

[Shixiaowei02]

/bot skip --comment "Related tests have passed."

[tensorrt-cicd]

PR_Github #59951 [ skip ] triggered by Bot. Commit: 31c45b9 Link to invocation

[tensorrt-cicd]

PR_Github #59951 [ skip ] completed with state SUCCESS. Commit: 31c45b9
Skipping testing for commit 31c45b9

Link to invocation