[None] [fix] Restructure kv cache memory ratio parameters in curated .yaml config files by xd-nv · Pull Request #11511 · NVIDIA/TensorRT-LLM

xd-nv · 2026-02-13T16:15:16Z

Description

Change the parameter kv_cache_free_gpu_memory_fraction: VALUE to

kv_cache_config:
  free_gpu_memory_fraction: VALUE

in curated .yaml config files to make it work with trtllm-serve.

Test Coverage

Confirmed the fix by running models with specified config, e.g.

trtllm-serve "nvidia/Qwen3-8B-FP8 " --config examples/configs/curated/qwen3.yaml

Without the change, the same command leads to error msg:

ValueError: LLM got invalid argument: kv_cache_free_gpu_memory_fraction

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)
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.
[ X] Please check this after reviewing the above items as appropriate for this PR.

GitHub Bot Help

/bot [-h] ['run', 'kill', 'skip', 'reuse-pipeline'] ...

Provide a user friendly way for developers to interact with a Jenkins server.

Run /bot [-h|--help] to print this help message.

See details below for each supported subcommand.

Details

run [--reuse-test (optional)pipeline-id --disable-fail-fast --skip-test --stage-list "A10-PyTorch-1, xxx" --gpu-type "A30, H100_PCIe" --test-backend "pytorch, cpp" --add-multi-gpu-test --only-multi-gpu-test --disable-multi-gpu-test --post-merge --extra-stage "H100_PCIe-TensorRT-Post-Merge-1, xxx" --detailed-log --debug(experimental)]

Launch build/test pipelines. All previously running jobs will be killed.

--reuse-test (optional)pipeline-id (OPTIONAL) : Allow the new pipeline to reuse build artifacts and skip successful test stages from a specified pipeline or the last pipeline if no pipeline-id is indicated. If the Git commit ID has changed, this option will be always ignored. The DEFAULT behavior of the bot is to reuse build artifacts and successful test results from the last pipeline.

--disable-reuse-test (OPTIONAL) : Explicitly prevent the pipeline from reusing build artifacts and skipping successful test stages from a previous pipeline. Ensure that all builds and tests are run regardless of previous successes.

--disable-fail-fast (OPTIONAL) : Disable fail fast on build/tests/infra failures.

--skip-test (OPTIONAL) : Skip all test stages, but still run build stages, package stages and sanity check stages. Note: Does NOT update GitHub check status.

--stage-list "A10-PyTorch-1, xxx" (OPTIONAL) : Only run the specified test stages. Examples: "A10-PyTorch-1, xxx". Note: Does NOT update GitHub check status.

--gpu-type "A30, H100_PCIe" (OPTIONAL) : Only run the test stages on the specified GPU types. Examples: "A30, H100_PCIe". Note: Does NOT update GitHub check status.

--test-backend "pytorch, cpp" (OPTIONAL) : Skip test stages which don't match the specified backends. Only support [pytorch, cpp, tensorrt, triton]. Examples: "pytorch, cpp" (does not run test stages with tensorrt or triton backend). Note: Does NOT update GitHub pipeline status.

--only-multi-gpu-test (OPTIONAL) : Only run the multi-GPU tests. Note: Does NOT update GitHub check status.

--disable-multi-gpu-test (OPTIONAL) : Disable the multi-GPU tests. Note: Does NOT update GitHub check status.

--add-multi-gpu-test (OPTIONAL) : Force run the multi-GPU tests in addition to running L0 pre-merge pipeline.

--post-merge (OPTIONAL) : Run the L0 post-merge pipeline instead of the ordinary L0 pre-merge pipeline.

--extra-stage "H100_PCIe-TensorRT-Post-Merge-1, xxx" (OPTIONAL) : Run the ordinary L0 pre-merge pipeline and specified test stages. Examples: --extra-stage "H100_PCIe-TensorRT-Post-Merge-1, xxx".

--detailed-log (OPTIONAL) : Enable flushing out all logs to the Jenkins console. This will significantly increase the log volume and may slow down the job.

--debug (OPTIONAL) : Experimental feature. Enable access to the CI container for debugging purpose. Note: Specify exactly one stage in the stage-list parameter to access the appropriate container environment. Note: Does NOT update GitHub check status.

For guidance on mapping tests to stage names, see docs/source/reference/ci-overview.md
and the scripts/test_to_stage_mapping.py helper.

kill

kill

Kill all running builds associated with pull request.

skip

skip --comment COMMENT

Skip testing for latest commit on pull request. --comment "Reason for skipping build/test" is required. IMPORTANT NOTE: This is dangerous since lack of user care and validation can cause top of tree to break.

reuse-pipeline

reuse-pipeline

Reuse a previous pipeline to validate current commit. This action will also kill all currently running builds associated with the pull request. IMPORTANT NOTE: This is dangerous since lack of user care and validation can cause top of tree to break.

coderabbitai · 2026-02-13T16:18:04Z

📝 Walkthrough

Walkthrough

Eight YAML configuration files in examples/configs/curated/ are restructured to move the kv_cache_free_gpu_memory_fraction parameter from top-level to nested under kv_cache_config.free_gpu_memory_fraction. Parameter values remain unchanged.

Changes

Cohort / File(s)	Summary
Configuration Restructuring `examples/configs/curated/deepseek-r1-deepgemm.yaml`, `examples/configs/curated/deepseek-r1-latency.yaml`, `examples/configs/curated/deepseek-r1-throughput.yaml`, `examples/configs/curated/gpt-oss-120b-latency.yaml`, `examples/configs/curated/gpt-oss-120b-throughput.yaml`, `examples/configs/curated/llama-3.3-70b.yaml`, `examples/configs/curated/llama-4-scout.yaml`, `examples/configs/curated/qwen3-disagg-prefill.yaml`, `examples/configs/curated/qwen3.yaml`	Moved `kv_cache_free_gpu_memory_fraction` from top-level YAML to nested under `kv_cache_config.free_gpu_memory_fraction`. Parameter values preserved; configuration structure reorganized for consistency.

Estimated code review effort

🎯 2 (Simple) | ⏱️ ~10 minutes

🚥 Pre-merge checks | ✅ 3 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Merge Conflict Detection	⚠️ Warning	❌ Merge conflicts detected (38 files): ⚔️ `README.md` (content) ⚔️ `cpp/tensorrt_llm/common/opUtils.h` (content) ⚔️ `cpp/tensorrt_llm/nanobind/CMakeLists.txt` (content) ⚔️ `cpp/tensorrt_llm/plugins/CMakeLists.txt` (content) ⚔️ `cpp/tensorrt_llm/plugins/ncclPlugin/allreducePlugin.cpp` (content) ⚔️ `cpp/tensorrt_llm/runtime/CMakeLists.txt` (content) ⚔️ `cpp/tensorrt_llm/runtime/ipcNvlsMemory.cu` (content) ⚔️ `cpp/tensorrt_llm/thop/CMakeLists.txt` (content) ⚔️ `cpp/tensorrt_llm/thop/allreduceOp.cpp` (content) ⚔️ `cpp/tests/unit_tests/runtime/virtualMemoryTest.cpp` (content) ⚔️ `examples/configs/curated/deepseek-r1-deepgemm.yaml` (content) ⚔️ `examples/configs/curated/deepseek-r1-latency.yaml` (content) ⚔️ `examples/configs/curated/deepseek-r1-throughput.yaml` (content) ⚔️ `examples/configs/curated/gpt-oss-120b-latency.yaml` (content) ⚔️ `examples/configs/curated/gpt-oss-120b-throughput.yaml` (content) ⚔️ `examples/configs/curated/llama-3.3-70b.yaml` (content) ⚔️ `examples/configs/curated/llama-4-scout.yaml` (content) ⚔️ `examples/configs/curated/qwen3-disagg-prefill.yaml` (content) ⚔️ `examples/configs/curated/qwen3.yaml` (content) ⚔️ `tensorrt_llm/_torch/auto_deploy/transform/library/ssm_cache.py` (content) ⚔️ `tensorrt_llm/_torch/modules/fused_moe/communication/communication_factory.py` (content) ⚔️ `tensorrt_llm/_torch/modules/fused_moe/communication/deep_ep.py` (content) ⚔️ `tensorrt_llm/_torch/modules/fused_moe/communication/deep_ep_low_latency.py` (content) ⚔️ `tensorrt_llm/_torch/modules/fused_moe/configurable_moe.py` (content) ⚔️ `tensorrt_llm/_torch/modules/fused_moe/fused_moe_cute_dsl.py` (content) ⚔️ `tensorrt_llm/_torch/modules/fused_moe/fused_moe_cutlass.py` (content) ⚔️ `tensorrt_llm/_torch/modules/fused_moe/fused_moe_deepgemm.py` (content) ⚔️ `tensorrt_llm/_torch/modules/fused_moe/fused_moe_triton.py` (content) ⚔️ `tensorrt_llm/_torch/modules/fused_moe/fused_moe_trtllm_gen.py` (content) ⚔️ `tensorrt_llm/_torch/modules/fused_moe/interface.py` (content) ⚔️ `tensorrt_llm/_torch/modules/fused_moe/quantization.py` (content) ⚔️ `tensorrt_llm/_torch/pyexecutor/sampling_utils.py` (content) ⚔️ `tensorrt_llm/executor/executor.py` (content) ⚔️ `tests/integration/defs/accuracy/test_disaggregated_serving.py` (content) ⚔️ `tests/integration/test_lists/waives.txt` (content) ⚔️ `tests/unittest/_torch/modules/moe/quantize_utils.py` (content) ⚔️ `tests/unittest/_torch/modules/moe/test_moe_backend.py` (content) ⚔️ `tests/unittest/_torch/modules/moe/test_moe_module.py` (content) These conflicts must be resolved before merging into `main`.	Resolve conflicts locally and push changes to this branch.

✅ Passed checks (3 passed)

Check name	Status	Explanation
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Description check	✅ Passed	The description includes all required sections: clear explanation of the change, test coverage demonstrating the fix, and confirmation of the PR checklist. The author explains both what changed (parameter structure) and why (compatibility with trtllm-serve).
Title check	✅ Passed	The title clearly describes the main change: restructuring kv cache memory ratio parameters in curated YAML config files, which aligns with all the file modifications shown in the raw summary.

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

✨ Finishing touches

📝 Generate docstrings

🧪 Generate unit tests (beta)

Create PR with unit tests
Post copyable unit tests in a comment

⚔️ Resolve merge conflicts (beta)

Auto-commit resolved conflicts to branch dev-config
Post resolved changes as copyable diffs in a comment

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

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

Signed-off-by: Xining Du <xd-nv@nvidia.com>

symphonylyh · 2026-02-19T23:20:03Z

/bot run

tensorrt-cicd · 2026-02-19T23:26:55Z

PR_Github #36298 [ run ] triggered by Bot. Commit: 1109449 Link to invocation

tensorrt-cicd · 2026-02-20T00:46:09Z

PR_Github #36298 [ run ] completed with state SUCCESS. Commit: 1109449
/LLM/main/L0_MergeRequest_PR pipeline #28071 completed with status: 'FAILURE'

⚠️ 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

xd-nv · 2026-02-20T02:11:20Z

/bot run

tensorrt-cicd · 2026-02-20T02:18:42Z

PR_Github #36304 [ run ] triggered by Bot. Commit: 7e44854 Link to invocation

tensorrt-cicd · 2026-02-20T04:19:03Z

PR_Github #36304 [ run ] completed with state SUCCESS. Commit: 7e44854
/LLM/main/L0_MergeRequest_PR pipeline #28077 completed with status: 'FAILURE'

⚠️ 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

xd-nv · 2026-02-20T21:06:01Z

/bot run

tensorrt-cicd · 2026-02-20T21:12:02Z

PR_Github #36359 [ run ] triggered by Bot. Commit: 7e44854 Link to invocation

tensorrt-cicd · 2026-02-21T00:58:14Z

PR_Github #36359 [ run ] completed with state SUCCESS. Commit: 7e44854
/LLM/main/L0_MergeRequest_PR pipeline #28125 completed with status: 'SUCCESS'

Link to invocation

laikhtewari · 2026-02-24T21:08:34Z

This just looks like example updates so I'll approve

@arysef @QiJune we should stabilize (and therefore test changes of) trtllm-serve config to prevent these kind of annoying naming issues

anish-shanbhag · 2026-02-24T21:13:30Z

@fsaady is adding tests for these configs in #11653 to catch this in the future

venkywonka · 2026-02-24T21:14:05Z

@fsaady has a PR that overlaps with this - and on top of this adds pydantic validation for all the YAMLs - so we atleast know the YAML's don't break schema like it has in this case (for which this PR fixes)

https://github.com/NVIDIA/TensorRT-LLM/pull/11653/changes

(just realized @anish-shanbhag posted the same comment)

venkywonka · 2026-02-24T21:15:15Z

I shall merge these changes - so @fsaady can rebase on top and add those validation tests - this doesn't have to touch trtllm-serve logic or "stabilize" the api - atleast for the objective of making sure all configs we advertise are immune to API staleness

….yaml config files (NVIDIA#11511) Signed-off-by: Xining Du <xd-nv@nvidia.com> Co-authored-by: Xining Du <xd-nv@nvidia.com>

xd-nv requested a review from a team as a code owner February 13, 2026 16:15

xd-nv requested review from laikhtewari and nv-guomingz February 13, 2026 16:15

coderabbitai Bot changed the title ~~[None] @coderabbitai Fix kv cache memory ratio params in curated .yaml config files~~ [None] [refactor] Restructure kv_cache config parameters Fix kv cache memory ratio params in curated .yaml config files Feb 13, 2026

xd-nv changed the title ~~[None] [refactor] Restructure kv_cache config parameters Fix kv cache memory ratio params in curated .yaml config files~~ [None] [fix] Restructure kv_cache config parameters Fix kv cache memory ratio params in curated .yaml config files Feb 13, 2026

xd-nv force-pushed the dev-config branch from a59b43d to 9b73e5d Compare February 13, 2026 16:22

Fix kv cache memory ratio params in curated .yaml config files

1109449

Signed-off-by: Xining Du <xd-nv@nvidia.com>

xd-nv force-pushed the dev-config branch from 9b73e5d to 1109449 Compare February 13, 2026 16:24

xd-nv changed the title ~~[None] [fix] Restructure kv_cache config parameters Fix kv cache memory ratio params in curated .yaml config files~~ [None] [fix] Restructure kv cache memory ratio parameters in curated .yaml config files Feb 13, 2026

symphonylyh approved these changes Feb 19, 2026

View reviewed changes

Merge branch 'main' into dev-config

7e44854

laikhtewari approved these changes Feb 24, 2026

View reviewed changes

venkywonka approved these changes Feb 24, 2026

View reviewed changes

venkywonka merged commit c55b190 into NVIDIA:main Feb 24, 2026
5 checks passed

Uh oh!

Conversation

xd-nv commented Feb 13, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Description

Test Coverage

PR Checklist

GitHub Bot Help

kill

skip

reuse-pipeline

Uh oh!

coderabbitai Bot commented Feb 13, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Walkthrough

Changes

Estimated code review effort

Uh oh!

symphonylyh commented Feb 19, 2026

Uh oh!

tensorrt-cicd commented Feb 19, 2026

Uh oh!

tensorrt-cicd commented Feb 20, 2026

Uh oh!

xd-nv commented Feb 20, 2026

Uh oh!

tensorrt-cicd commented Feb 20, 2026

Uh oh!

tensorrt-cicd commented Feb 20, 2026

Uh oh!

xd-nv commented Feb 20, 2026

Uh oh!

tensorrt-cicd commented Feb 20, 2026

Uh oh!

tensorrt-cicd commented Feb 21, 2026

Uh oh!

laikhtewari commented Feb 24, 2026

Uh oh!

anish-shanbhag commented Feb 24, 2026

Uh oh!

venkywonka commented Feb 24, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

venkywonka commented Feb 24, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

6 participants

xd-nv commented Feb 13, 2026 •

edited

Loading

coderabbitai Bot commented Feb 13, 2026 •

edited

Loading

venkywonka commented Feb 24, 2026 •

edited

Loading