client.py 2.21 KB
Newer Older
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
# SPDX-FileCopyrightText: Copyright (c) 2025 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
# SPDX-License-Identifier: Apache-2.0
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.


17
import argparse
18
19
20
import asyncio

import uvloop
Neelay Shah's avatar
Neelay Shah committed
21

Neelay Shah's avatar
Neelay Shah committed
22
from dynamo.runtime import DistributedRuntime, dynamo_worker
23
24
25
26

from .protocol import Request


Neelay Shah's avatar
Neelay Shah committed
27
@dynamo_worker()
28
async def worker(
29
30
31
32
33
    runtime: DistributedRuntime,
    component: str,
    prompt: str,
    max_tokens: int,
    temperature: float,
34
35
36
37
38
):
    """
    Instantiate a `backend` client and call the `generate` endpoint
    """
    # get endpoint
Neelay Shah's avatar
Neelay Shah committed
39
    endpoint = runtime.namespace("dynamo").component(component).endpoint("generate")
40
41
42
43
44

    # create client
    client = await endpoint.client()

    # issue request
45
46
47
48
49
50
51
52
53
54
55
56
57
58
    tasks = []
    for _ in range(1):
        tasks.append(
            client.generate(
                Request(
                    prompt=prompt,
                    sampling_params={
                        "temperature": temperature,
                        "max_tokens": max_tokens,
                    },
                ).model_dump_json()
            )
        )
    streams = await asyncio.gather(*tasks)
59
60

    # process response
61
62
63
    for stream in streams:
        async for resp in stream:
            print(resp)
64
65
66
67
68


if __name__ == "__main__":
    uvloop.install()

69
    parser = argparse.ArgumentParser()
70
    parser.add_argument("--prompt", type=str, default="what is the capital of france?")
71
    parser.add_argument("--component", type=str, default="vllm")
72
73
74
75
76
    parser.add_argument("--max-tokens", type=int, default=10)
    parser.add_argument("--temperature", type=float, default=0.5)

    args = parser.parse_args()

77
    asyncio.run(worker(args.component, args.prompt, args.max_tokens, args.temperature))