From 16248e97efe13d25e4d47b6d63eccf9cdf45a31c Mon Sep 17 00:00:00 2001 From: Yueqian Lin <70319226+linyueqian@users.noreply.github.com> Date: Sun, 8 Mar 2026 22:12:45 -0400 Subject: [PATCH] Pre-download Qwen3-TTS model in CI to avoid intermittent download timeouts (#1727) Signed-off-by: linyueqian Co-authored-by: Hongsheng Liu --- .buildkite/test-nightly.yml | 1 + .buildkite/test-ready.yml | 3 ++- 2 files changed, 3 insertions(+), 1 deletion(-) diff --git a/.buildkite/test-nightly.yml b/.buildkite/test-nightly.yml index 8f141576d..001c46645 100644 --- a/.buildkite/test-nightly.yml +++ b/.buildkite/test-nightly.yml @@ -68,6 +68,7 @@ steps: if: build.env("NIGHTLY") == "1" commands: - | + huggingface-cli download Qwen/Qwen3-TTS-12Hz-0.6B-CustomVoice export VLLM_WORKER_MULTIPROC_METHOD=spawn pytest -s -v tests/e2e/online_serving/test_qwen3_tts.py::TestQwen3TTSNoAsyncChunk agents: diff --git a/.buildkite/test-ready.yml b/.buildkite/test-ready.yml index 6c1b458b8..092d6f11a 100644 --- a/.buildkite/test-ready.yml +++ b/.buildkite/test-ready.yml @@ -214,9 +214,10 @@ steps: commands: - | timeout 20m bash -c ' + huggingface-cli download Qwen/Qwen3-TTS-12Hz-0.6B-CustomVoice export VLLM_LOGGING_LEVEL=DEBUG export VLLM_WORKER_MULTIPROC_METHOD=spawn - pytest -s -v tests/e2e/online_serving/test_qwen3_tts.py + pytest -s -v tests/e2e/online_serving/test_qwen3_tts.py -k "not NoAsyncChunk" ' agents: queue: "gpu_4_queue"