Doc: update steps of using Draft-Target-Model (DTM) in the documents.

[wili-65535]

In this PR, we update the steps of using Draft-Target-Model (DTM) speculative decoding in TensorRT-LLM and TensorRT-LLm-backend.

Based on TRTLLM-backend: commit 86542b637bbccdd708ab892337c8ad3a95932131 (tag: v0.18.0) and docker image nvcr.io/nvidia/tritonserver:25.03-trtllm-python-py3.

[tensorrt-cicd]

PR_Github #1554 [ run ] completed with state FAILURE
/LLM/main/L0_MergeRequest_PR pipeline #1161 completed with status: 'FAILURE'

[wili-65535]

/bot run

[tensorrt-cicd]

PR_Github #1559 [ run ] triggered by Bot

[kaiyux]

@lfr-0531 @yweng0828 Can you help review this PR?

[tensorrt-cicd]

PR_Github #1559 [ run ] completed with state SUCCESS
/LLM/main/L0_MergeRequest_PR pipeline #1165 completed with status: 'SUCCESS'

[lfr-0531]

LGTM~

[yweng0828]

Hi @wili-65535 , thanks for the update. The documentation is much better organized now. : )

[Copilot]

Copilot reviewed 3 out of 3 changed files in this pull request and generated 2 comments.

Comments suppressed due to low confidence (1)

examples/draft_target_model/README.md:89

• The configuration explanation above uses '[4,[0],[1],False]' as an example, but the code sample uses '[4,[0],[1],True]'. Please update one of them to ensure consistency.

    --draft_target_model_config="[4,[0],[1],True]" \

[kaiyux]

/bot reuse-pipeline

[wili-65535]

Thank you @lfr-0531, @yweng0828 and @kaiyux ! And I fix the comments Copilot comes up with, so could we continue to merge?

[kaiyux]

Thank you @lfr-0531, @yweng0828 and @kaiyux ! And I fix the comments Copilot comes up with, so could we continue to merge?

I've already set auto-merge. Thanks.

[tensorrt-cicd]

PR_Github #1590 [ reuse-pipeline ] triggered by Bot

[tensorrt-cicd]

PR_Github #1590 [ reuse-pipeline ] completed with state SUCCESS
Reusing PR_Github #1559 for commit ec270fc