| 1 | # SPDX-License-Identifier: Apache-2.0 |
| 2 | # SPDX-FileCopyrightText: Copyright contributors to the vLLM project |
| 3 | |
| 4 | from vllm import LLM, SamplingParams |
| 5 | from vllm.v1.metrics.reader import Counter, Gauge, Histogram, Vector |
| 6 | |
| 7 | # Sample prompts. |
| 8 | prompts = [ |
| 9 | "Hello, my name is", |
| 10 | "The president of the United States is", |
| 11 | "The capital of France is", |
| 12 | "The future of AI is", |
| 13 | ] |
| 14 | # Create a sampling params object. |
| 15 | sampling_params = SamplingParams(temperature=0.8, top_p=0.95) |
| 16 | |
| 17 | |
| 18 | def main(): |
| 19 | # Create an LLM. |
| 20 | llm = LLM(model="facebook/opt-125m", disable_log_stats=False) |
| 21 | |
| 22 | # Generate texts from the prompts. |
| 23 | outputs = llm.generate(prompts, sampling_params) |
| 24 | |
| 25 | # Print the outputs. |
| 26 | print("-" * 50) |
| 27 | for output in outputs: |
| 28 | prompt = output.prompt |
| 29 | generated_text = output.outputs[0].text |
| 30 | print(f"Prompt: {prompt!r}\nGenerated text: {generated_text!r}") |
| 31 | print("-" * 50) |
| 32 | |
| 33 | # Dump all metrics |
| 34 | for metric in llm.get_metrics(): |
| 35 | if isinstance(metric, Gauge): |
| 36 | print(f"{metric.name} (gauge) = {metric.value}") |
| 37 | elif isinstance(metric, Counter): |
| 38 | print(f"{metric.name} (counter) = {metric.value}") |
| 39 | elif isinstance(metric, Vector): |
| 40 | print(f"{metric.name} (vector) = {metric.values}") |
| 41 | elif isinstance(metric, Histogram): |
| 42 | print(f"{metric.name} (histogram)") |
| 43 | print(f" sum = {metric.sum}") |
| 44 | print(f" count = {metric.count}") |
| 45 | for bucket_le, value in metric.buckets.items(): |
| 46 | print(f" {bucket_le} = {value}") |
| 47 | |
| 48 | |
| 49 | if __name__ == "__main__": |
| 50 | main() |