[TRTLLM-5847][feat] Support n-gram speculative decoding with disagg

[raayandhar]

[TRTLLM-5847][feat] Support n-gram speculative decoding with disagg

Description

Currently, we do not support disagg n-gram speculative decoding. In disagg setting, prepare_draft_tokens is called before _forward_step, so py_batch_idx is not set for the requests and on subsequent generation, the sort in prepare_draft_tokens fails. If we sort by request_id when py_batch_idx is None, this prevents this issue and maintains the same sort, and on subsequent iterations, we will use py_batch_idx correctly.

Previously, when is_keep_all=False, n-gram spec decoding would fail due to some small typos, this is fixed here. Additionally, MMLU as an integration test seems a bit weak for testing; I was able to get ~28% on GSM8K due to a bug, but MMLU would still pass. GSM8K seems to be more robust.

Test Coverage

• Added disagg n-gram integration unit test.
• Added disagg n-gram accuracy test (using GSM8K)
• Changed agg n-gram accuracy test to also use GSM8K instead of just MMLU.

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.

[raayandhar]

cc: @pcastonguay

[raayandhar]

/bot run

[tensorrt-cicd]

PR_Github #10871 [ run ] triggered by Bot

[tensorrt-cicd]

PR_Github #10871 [ run ] completed with state SUCCESS
/LLM/main/L0_MergeRequest_PR pipeline #8036 completed with status: 'FAILURE'

[pcastonguay]

/bot run --disable-fail-fast --add-multi-gpu-test

[pcastonguay]

@raayandhar could you rebase now that #5558 is merged? Thanks.

[pcastonguay]

@Tabrizian @SimengLiu-nv could you review the changes to py_executor.py and ngram.py respectively? Thank you

[raayandhar]

/bot run --disable-fail-fast --add-multi-gpu-test

[tensorrt-cicd]

PR_Github #11176 [ run ] triggered by Bot

[tensorrt-cicd]

PR_Github #11176 [ run ] completed with state SUCCESS
/LLM/main/L0_MergeRequest_PR pipeline #8267 completed with status: 'FAILURE'

[raayandhar]

/bot run --disable-fail-fast --add-multi-gpu-test

[tensorrt-cicd]

PR_Github #11184 [ run ] triggered by Bot

[tensorrt-cicd]

PR_Github #11184 [ run ] completed with state SUCCESS
/LLM/main/L0_MergeRequest_PR pipeline #8273 completed with status: 'SUCCESS'
Pipeline passed with automatic retried tests. Check the rerun report for details.