Skip to content

Conversation

@chenfeiz0326
Copy link
Collaborator

[TRTLLM-6262] Fix Llama4 Scout FP4 crash issue

Description

This PR is to fix Llama4 Scout FP4 crashing issue due to out_scale_sf is None in _forward_nope.

FP4 output prompts:

[0] Prompt: 'Hello, my name is', Generated text: ' Cherub, and I am a sophomore. I am a native Chinese, and I am going to study in the United States. My major is English language and Literature. \nI have a strong interest in the field of English teaching, and I hope to study in the US to learn about the advanced teaching methods and develop'
[1] Prompt: 'The president of the United States is', Generated text: ' sponsoring a new educational initiative with a large- scale Federal services program under a jurisdiction of the federal cabinet. The president calls the agency or subsidized schools, "Schoolabaellot40005 program," and also advocacy intervention plan. The program is sponsored by federal jurisdiction under Title V, which is regulated under governmental legisl'
[2] Prompt: 'The capital of France is', Generated text: ' Paris. The capital of France is Paris. The capital of France is Paris. The capital of France is Paris. The capital of France is Paris. The capital of France is Paris. The capital of France is Paris. The capital of France is Paris. The capital of France is Paris. The capital of France is Paris'
[3] Prompt: 'The future of AI is', Generated text: ' often considered to be solved by machines, but it's not that today. The AI Revolution is underway, but it's not being built by with human minds. – The Verolein Bellringa, AI researcher, says that "the best way to predict what the future of AI will be is to learn from the'

Test Coverage

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.

run [--disable-fail-fast --skip-test --stage-list "A10-1, xxx" --gpu-type "A30, H100_PCIe" --add-multi-gpu-test --only-multi-gpu-test --disable-multi-gpu-test --post-merge --extra-stage "H100_PCIe-[Post-Merge]-1, xxx"]

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

--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 "A10-1, xxx" (OPTIONAL) : Only run the specified test stages. Examples: "A10-1, 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.

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

For guidance on mapping tests to stage names, see docs/source/reference/ci-overview.md.

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.

@chenfeiz0326 chenfeiz0326 requested a review from a team as a code owner July 8, 2025 08:54
@chenfeiz0326 chenfeiz0326 requested review from hlu1 and shaharmor98 July 8, 2025 08:54
@chenfeiz0326 chenfeiz0326 force-pushed the user/chenfeiz/fix-llama4-fp4-crash branch from cb73dbb to 00df342 Compare July 8, 2025 08:58
@chenfeiz0326
Copy link
Collaborator Author

/bot run

@tensorrt-cicd
Copy link
Collaborator

PR_Github #11268 [ run ] triggered by Bot

@svc-trtllm-gh-bot svc-trtllm-gh-bot added the Community want to contribute PRs initiated from Community label Jul 8, 2025
@tensorrt-cicd
Copy link
Collaborator

PR_Github #11268 [ run ] completed with state SUCCESS
/LLM/main/L0_MergeRequest_PR pipeline #8335 completed with status: 'FAILURE'

Signed-off-by: Chenfei Zhang <[email protected]>
@litaotju litaotju force-pushed the user/chenfeiz/fix-llama4-fp4-crash branch from 00df342 to 242a30d Compare July 8, 2025 13:26
@litaotju
Copy link
Collaborator

litaotju commented Jul 8, 2025

/bot run

@tensorrt-cicd
Copy link
Collaborator

PR_Github #11307 [ run ] triggered by Bot

@tensorrt-cicd
Copy link
Collaborator

PR_Github #11307 [ run ] completed with state SUCCESS
/LLM/main/L0_MergeRequest_PR pipeline #8362 completed with status: 'SUCCESS'
Pipeline passed with automatic retried tests. Check the rerun report for details.

@litaotju litaotju merged commit 64fd64f into NVIDIA:main Jul 9, 2025
3 checks passed
zhou-yuxin pushed a commit to zhou-yuxin/TensorRT-LLM that referenced this pull request Jul 15, 2025
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

Community want to contribute PRs initiated from Community

Projects

None yet

Development

Successfully merging this pull request may close these issues.

4 participants