feat: containerize VoxCPM2 vLLM-Omni deployment

config/voxcpm2-a10-container.yaml

@@ -0,0 +1,23 @@
+async_chunk: false
+dtype: bfloat16
+
+stages:
+  - stage_id: 0
+    devices: "0"
+    max_num_seqs: 2
+    gpu_memory_utilization: 0.9
+    kv_cache_memory_bytes: 6442450944
+    enforce_eager: true
+    trust_remote_code: true
+    enable_prefix_caching: false
+    max_num_batched_tokens: 4096
+    max_model_len: 4096
+    custom_voice_dir: /opt/voices
+
+    default_sampling_params:
+      temperature: 0.0
+      top_p: 1.0
+      top_k: -1
+      max_tokens: 4096
+      seed: 42
+      repetition_penalty: 1.0