Generate Text Asynchronously#

Source NVIDIA/TensorRT-LLM.

 1### Generate Text Asynchronously
 2import asyncio
 3
 4from tensorrt_llm import SamplingParams
 5from tensorrt_llm._tensorrt_engine import LLM
 6
 7
 8def main():
 9    # model could accept HF model name or a path to local HF model.
10    llm = LLM(model="TinyLlama/TinyLlama-1.1B-Chat-v1.0")
11
12    # Sample prompts.
13    prompts = [
14        "Hello, my name is",
15        "The president of the United States is",
16        "The capital of France is",
17        "The future of AI is",
18    ]
19
20    # Create a sampling params.
21    sampling_params = SamplingParams(temperature=0.8, top_p=0.95)
22
23    # Async based on Python coroutines
24    async def task(prompt: str):
25        output = await llm.generate_async(prompt, sampling_params)
26        print(
27            f"Prompt: {output.prompt!r}, Generated text: {output.outputs[0].text!r}"
28        )
29
30    async def main():
31        tasks = [task(prompt) for prompt in prompts]
32        await asyncio.gather(*tasks)
33
34    asyncio.run(main())
35
36    # Got output like follows:
37    # 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'
38    # 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'
39    # Prompt: 'The capital of France is', Generated text: 'Paris.'
40    # 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'
41
42
43if __name__ == '__main__':
44    main()