[WIP] Add E2E test scripts for qwen3-30b model

[YixuanWang-99]

Description

sft and rl are stuck with OOM. The other 3 work fine.

Tests

Tested on TPU v6e vm.

# Export your Hugging Face token (if not already set)
export HF_TOKEN=<your Hugging Face access token>

# Create a unique run ID for this batch of tests
export RUN_ID=$(date +%Y-%m-%d-%H-%M-%S)

# Convert the HF checkpoint to MaxText format
bash tests/end_to_end/tpu/qwen3/30b/test_qwen3_to_mt.sh $RUN_ID

# Run the pre-train test
bash tests/end_to_end/tpu/qwen3/30b/test_qwen3.sh $RUN_ID

# Run the SFT test
bash tests/end_to_end/tpu/qwen3/30b/test_qwen3_sft.sh $RUN_ID

# Run the RL test
bash tests/end_to_end/tpu/qwen3/30b/test_qwen3_rl.sh $RUN_ID

# convert MaxText checkpoint back to HF format
bash tests/end_to_end/tpu/qwen3/30b/test_qwen3_to_hf.sh $RUN_ID $CHECKPOINT_PATH

Checklist

Before submitting this PR, please make sure (put X in square brackets):

• I have performed a self-review of my code. For an optional AI review, add the gemini-review label.
• I have necessary comments in my code, particularly in hard-to-understand areas.
• I have run end-to-end tests tests and provided workload links above if applicable.
• I have made or will make corresponding changes to the doc if needed, including adding new documentation pages to the relevant Table of Contents (toctree directive) as explained in our documentation.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!