Response parameter CompletionUsage return 0 token consumption

[moyerlee]

the backend qwen model does not enable the decouple mode(streaming), however i found the openapi response did not show token usage. Below is an example:

ChatCompletion(id='cmpl-45f33530-2dcc-4352-8d97-1dd056efb2e0', choices=[Choice(finish_reason='stop', index=0, logprobs=None, message=ChatCompletionMessage(content='System: 你是一个知识百科全书助手，可以回答各种问题。\nUser: 什么是牛顿第一定律？\nASSISTANT: 牛顿第一定律，也被称为惯性定律，认为如果一个物体', refusal=None, role='assistant', function_call=None, tool_calls=None))], created=1728957474, model='ensemble', object='text_completion', service_tier=None, system_fingerprint=None, usage=CompletionUsage(completion_tokens=0, prompt_tokens=0, total_tokens=0, completion_tokens_details=None, prompt_tokens_details=None))

[npuichigo]

We just cannot return token usage since trtllm backend does't report that for us. To enable that, maybe you need to customize the triton side alongwith the proxy here.