example(eagle3): Qwen3-0.6B EAGLE3 online/offline launcher examples

[ChenhanYu]

What does this PR do?

Type of change: new example

Adds EAGLE3 speculative-decoding launcher examples for Qwen3-0.6B (small, ungated):
examples/Qwen/Qwen3-0.6B/hf_online_eagle3.yaml + hf_offline_eagle3.yaml, adapted from the existing Qwen/Qwen3-8B/hf_{online,offline}_eagle3.yaml for the smaller model on a single GPU (make_dataset/query → dump hidden states → train_eagle → common/specdec_bench VLLM benchmark, via examples/specdec_bench).

Usage

uv run launch.py --yaml examples/Qwen/Qwen3-0.6B/hf_online_eagle3.yaml --yes

Testing

• Covered by tools/launcher/tests/test_examples_resolve.py (parses + validates the new examples; all referenced common/* scripts exist).
• End-to-end (train draft → VLLM specdec benchmark) is a cluster/GPU run — validated via the nmm-sandbox integration CI, not here.

Before your PR is "Ready for review"

• Backward compatible?: ✅ N/A (new example files only)
• New tests?: ✅ auto-covered by the example-resolution test
• Changelog?: N/A (example)

Additional Information

Motivation: nmm-sandbox uses these to replace internal Llama-3.2-1B EAGLE3 specdec tests that hit a TRT-LLM fused-weight loader assertion; the ungated Qwen3-0.6B + co-versioned examples/specdec_bench (VLLM) is a smaller, maintained path.

[copy-pr-bot]

Auto-sync is disabled for draft pull requests in this repository. Workflows must be run manually.

Contributors can view more details about this message here.

[coderabbitai]

Important

Review skipped

Draft detected.

Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Enterprise

Run ID: bc18daca-8e77-4a86-8b51-31cd71374e75

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch chenhany/eagle3-qwen06b

---

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

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 77.76%. Comparing base (e911c3b) to head (4b1a3b0).
⚠️ Report is 1 commits behind head on main.

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #1954   +/-   ##
=======================================
  Coverage   77.76%   77.76%           
=======================================
  Files         519      519           
  Lines       57955    57955           
=======================================
  Hits        45067    45067           
  Misses      12888    12888           

Flag	Coverage Δ
unit	55.20% <ø> (ø)

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.