LLM Inference Async
Source https://github.com/NVIDIA/TensorRT-LLM/tree/main/examples/llm-api/llm_inference_async.py.
1### Generate Text Asynchronously
2import asyncio
3
4from tensorrt_llm import LLM, SamplingParams
5
6# model could accept HF model name or a path to local HF model.
7llm = LLM(model="TinyLlama/TinyLlama-1.1B-Chat-v1.0")
8
9# Sample prompts.
10prompts = [
11 "Hello, my name is",
12 "The president of the United States is",
13 "The capital of France is",
14 "The future of AI is",
15]
16
17# Create a sampling params.
18sampling_params = SamplingParams(temperature=0.8, top_p=0.95)
19
20
21# Async based on Python coroutines
22async 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
29async def main():
30 tasks = [task(prompt) for prompt in prompts]
31 await asyncio.gather(*tasks)
32
33
34asyncio.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'