Generate text asynchronously#

Source NVIDIA/TensorRT-LLM.

 1import asyncio
 2
 3from tensorrt_llm import LLM, SamplingParams
 4
 5
 6def main():
 7    # model could accept HF model name or a path to local HF model.
 8    llm = LLM(model="TinyLlama/TinyLlama-1.1B-Chat-v1.0")
 9
10    # Sample prompts.
11    prompts = [
12        "Hello, my name is",
13        "The president of the United States is",
14        "The capital of France is",
15        "The future of AI is",
16    ]
17
18    # Create a sampling params.
19    sampling_params = SamplingParams(temperature=0.8, top_p=0.95)
20
21    # Async based on Python coroutines
22    async def task(prompt: str):
23        output = await llm.generate_async(prompt, sampling_params)
24        print(
25            f"Prompt: {output.prompt!r}, Generated text: {output.outputs[0].text!r}"
26        )
27
28    async def main():
29        tasks = [task(prompt) for prompt in prompts]
30        await asyncio.gather(*tasks)
31
32    asyncio.run(main())
33
34    # Got output like follows:
35    # Prompt: 'Hello, my name is', Generated text: '\n\nJane Smith. I am a student pursuing my degree in Computer Science at [university]. I enjoy learning new things, especially technology and programming'
36    # Prompt: 'The president of the United States is', Generated text: 'likely to nominate a new Supreme Court justice to fill the seat vacated by the death of Antonin Scalia. The Senate should vote to confirm the'
37    # Prompt: 'The capital of France is', Generated text: 'Paris.'
38    # Prompt: 'The future of AI is', Generated text: 'an exciting time for us. We are constantly researching, developing, and improving our platform to create the most advanced and efficient model available. We are'
39
40
41if __name__ == '__main__':
42    main()