[TRTLLM-13050][test] Remove two-model eagle3 spec-decoding tests

[QiJune]

Drop the two-model side (eagle3_one_model=False, equivalently the ...-two_model-... parametrization) of the eagle3 speculative-decoding tests from the L0 pre-merge lists, keeping the one-model mirror for each.

Removes 17 entries across l0_dgx_b200, l0_dgx_h100, l0_gb200_multi_gpus, and l0_h100, reclaiming ~287 GPU-h/week of pre-merge CI time. The two-model path retains cheaper unit-level coverage and qa/ post-merge entries.

Summary by CodeRabbit

• Tests
  • Updated test configurations across multiple GPU platforms to consolidate test coverage for Eagle3 and guided decoding features, replacing multi-model test parameterizations with optimized single-model variants to streamline validation processes.

Description

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.

[QiJune]

/bot run

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Enterprise

Run ID: 06e3d8a0-befd-4053-a009-1d9e0284b810

📥 Commits

Reviewing files that changed from the base of the PR and between 6b126ca and c4ee5e7.

📒 Files selected for processing (4)

• tests/integration/test_lists/test-db/l0_dgx_b200.yml
• tests/integration/test_lists/test-db/l0_dgx_h100.yml
• tests/integration/test_lists/test-db/l0_gb200_multi_gpus.yml
• tests/integration/test_lists/test-db/l0_h100.yml

💤 Files with no reviewable changes (4)

• tests/integration/test_lists/test-db/l0_h100.yml
• tests/integration/test_lists/test-db/l0_dgx_b200.yml
• tests/integration/test_lists/test-db/l0_gb200_multi_gpus.yml
• tests/integration/test_lists/test-db/l0_dgx_h100.yml

---

📝 Walkthrough

Walkthrough

This PR updates test configuration YAML files across multiple GPU hardware setups (DGX B200, DGX H100, GB200, H100) to adjust which Eagle3 model serving test parameterizations run in CI. Changes include transitioning from two-model to one-model overlap scheduler configurations, narrowing disaggregated serving test coverage, and refining accuracy test lists with new fp8 variants and KV cache configurations.

Changes

Eagle3 Test Configuration Matrix Updates

Layer / File(s)	Summary
DGX B200 and GB200 4-GPU one-model scheduler transition tests/integration/test_lists/test-db/l0_dgx_b200.yml, tests/integration/test_lists/test-db/l0_gb200_multi_gpus.yml	Test configuration changes for 4-GPU setups (B200 and GB200) that replace two-model no-overlap scheduler TestGPTOSS::test_eagle3_4gpus cases with one-model overlap scheduler variants for both v1_kv_cache and v2_kv_cache configurations.
DGX H100 disaggregated serving and Eagle3 2-GPU updates tests/integration/test_lists/test-db/l0_dgx_h100.yml	Test configuration changes for DGX H100 GPU-2 setup that remove eagle3_one_model=False cases from Llama3.1 8B disaggregated serving tests and narrow Eagle3 2-GPU cases to only triton-one_model-overlap_scheduler variant.
H100 accuracy test list refinements tests/integration/test_lists/test-db/l0_h100.yml	Test configuration updates for H100 accuracy test lists: addition of four TRTLLM fp8 variants with torch_compile and fp8kv toggles for Llama3.1 8B, narrowing of Qwen3 8B Eagle3 matrix to only eagle3_one_model=True cases with both enable_chunked_prefill values, and removal of one post-merge guided decoding variant with eagle3_one_model=False.

Estimated code review effort

🎯 2 (Simple) | ⏱️ ~10 minutes

Possibly related PRs

• NVIDIA/TensorRT-LLM#14503: Updates to test configuration that overlap in the TestGPTOSS::test_eagle3_4gpus parameter space but affect different configuration variants (CI waiver additions vs. test list parameterization changes).
• NVIDIA/TensorRT-LLM#14542: Adds waiver for the newly-included GB200 4-GPU Eagle3 v1_kv_cache-trtllm-one_model-no_overlap_scheduler configuration that this PR introduces to the test lists.
• NVIDIA/TensorRT-LLM#14596: Corresponding waiver adjustments in tests/integration/test_lists/waives.txt for the same test_eagle3_4gpus parameterizations modified by this PR's test list updates.

Suggested reviewers

• mzweilz
• StanleySun639
• crazydemo

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title clearly and specifically identifies the main change: removing two-model eagle3 spec-decoding tests from L0 pre-merge lists, with proper ticket reference and type indicator.
Description check	✅ Passed	The PR description adequately explains the objective (dropping two-model eagle3 tests), quantifies the impact (~287 GPU-h/week saved), and notes that one-model tests are retained with coverage in other areas.
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.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

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

[tensorrt-cicd]

PR_Github #50991 [ run ] triggered by Bot. Commit: c4ee5e7 Link to invocation

[tensorrt-cicd]

PR_Github #50991 [ run ] completed with state SUCCESS. Commit: c4ee5e7
/LLM/main/L0_MergeRequest_PR pipeline #40441 completed with status: 'SUCCESS'

CI Report

Link to invocation