把qwen3-tts.yaml文件放在本地

README.md

@@ -87,7 +87,7 @@ VoiceDesign
 ```bash
 ## serve启动
 vllm-omni serve Qwen3-TTS/Qwen3-TTS-12Hz-1.7B-VoiceDesign \
-    --stage-configs-path vllm_omni/model_executor/stage_configs/qwen3_tts.yaml \
+    --stage-configs-path qwen3_tts.yaml \
     --omni \
     --port 8001 \
     --trust-remote-code \
@@ -113,7 +113,7 @@ CustomVoice
 ```bash
 ## serve启动
 vllm-omni serve Qwen3-TTS/Qwen3-TTS-12Hz-1.7B-CustomVoice \
-    --stage-configs-path vllm_omni/model_executor/stage_configs/qwen3_tts.yaml \
+    --stage-configs-path qwen3_tts.yaml \
     --omni \
     --port 8001 \
     --trust-remote-code \
@@ -139,7 +139,7 @@ VoiceClone
 ```bash
 ## serve启动
 vllm-omni serve Qwen3-TTS/Qwen3-TTS-12Hz-1.7B-Base \
-    --stage-configs-path vllm_omni/model_executor/stage_configs/qwen3_tts.yaml \
+    --stage-configs-path qwen3_tts.yaml \
     --omni \
     --port 8001 \
     --trust-remote-code \

qwen3_tts.yaml

+stage_args:
+  - stage_id: 0
+    stage_type: llm  # Use llm stage type to launch OmniLLM
+    runtime:
+      devices: "0"
+      max_batch_size: 1
+    engine_args:
+      model_stage: qwen3_tts
+      model_arch: Qwen3TTSForConditionalGeneration
+      worker_type: generation
+      scheduler_cls: vllm_omni.core.sched.omni_generation_scheduler.OmniGenerationScheduler
+      enforce_eager: true
+      trust_remote_code: true
+      async_scheduling: false
+      enable_prefix_caching: false
+      engine_output_type: audio  # Final output: audio waveform
+      gpu_memory_utilization: 0.1
+      distributed_executor_backend: "mp"
+      max_num_batched_tokens: 1000000
+
+    final_output: true
+    final_output_type: audio