mirror of
https://git.datalinker.icu/vllm-project/vllm.git
synced 2025-12-10 06:35:00 +08:00
e489ad7a21
- **Add SPDX license headers to python source files**
- **Check for SPDX headers using pre-commit**
commit 9d7ef44c3cfb72ca4c32e1c677d99259d10d4745
Author: Russell Bryant <rbryant@redhat.com>
Date: Fri Jan 31 14:18:24 2025 -0500
Add SPDX license headers to python source files
This commit adds SPDX license headers to python source files as
recommended to
the project by the Linux Foundation. These headers provide a concise way
that is
both human and machine readable for communicating license information
for each
source file. It helps avoid any ambiguity about the license of the code
and can
also be easily used by tools to help manage license compliance.
The Linux Foundation runs license scans against the codebase to help
ensure
we are in compliance with the licenses of the code we use, including
dependencies. Having these headers in place helps that tool do its job.
More information can be found on the SPDX site:
- https://spdx.dev/learn/handling-license-info/
Signed-off-by: Russell Bryant <rbryant@redhat.com>
commit 5a1cf1cb3b80759131c73f6a9dddebccac039dea
Author: Russell Bryant <rbryant@redhat.com>
Date: Fri Jan 31 14:36:32 2025 -0500
Check for SPDX headers using pre-commit
Signed-off-by: Russell Bryant <rbryant@redhat.com>
---------
Signed-off-by: Russell Bryant <rbryant@redhat.com>
84 lines
2.4 KiB
Python
84 lines
2.4 KiB
Python
# SPDX-License-Identifier: Apache-2.0
|
|
"""Verify that seeded random sampling is deterministic.
|
|
|
|
Run `pytest tests/samplers/test_seeded_generate.py`.
|
|
"""
|
|
import copy
|
|
import random
|
|
from itertools import combinations
|
|
|
|
import pytest
|
|
|
|
from vllm import SamplingParams
|
|
from vllm.model_executor.utils import set_random_seed
|
|
|
|
MODEL = "facebook/opt-125m"
|
|
RANDOM_SEEDS = list(range(5))
|
|
|
|
|
|
@pytest.fixture
|
|
def vllm_model(vllm_runner):
|
|
with vllm_runner(MODEL, dtype="half") as vllm_model:
|
|
yield vllm_model
|
|
|
|
|
|
@pytest.mark.parametrize("seed", RANDOM_SEEDS)
|
|
def test_random_sample_with_seed(
|
|
vllm_model,
|
|
example_prompts,
|
|
seed: int,
|
|
) -> None:
|
|
set_random_seed(seed)
|
|
|
|
sampling_params = SamplingParams(
|
|
# Parameters to ensure sufficient randomness
|
|
temperature=3.0,
|
|
top_p=min(random.random() + 0.3, 1),
|
|
top_k=random.randint(5, 20),
|
|
n=random.randint(1, 10),
|
|
presence_penalty=random.randint(0, 1),
|
|
max_tokens=8,
|
|
ignore_eos=True,
|
|
)
|
|
|
|
sampling_params_seed_1 = copy.deepcopy(sampling_params)
|
|
sampling_params_seed_1.seed = 100
|
|
sampling_params_seed_2 = copy.deepcopy(sampling_params)
|
|
sampling_params_seed_2.seed = 200
|
|
|
|
llm = vllm_model.model
|
|
|
|
for prompt in example_prompts:
|
|
for params in (
|
|
sampling_params,
|
|
sampling_params_seed_1,
|
|
sampling_params_seed_2,
|
|
sampling_params,
|
|
sampling_params_seed_1,
|
|
sampling_params_seed_2,
|
|
):
|
|
llm._add_request(prompt, params=params)
|
|
|
|
results = llm._run_engine(use_tqdm=False)
|
|
all_outputs = [[out.token_ids for out in output.outputs]
|
|
for output in results]
|
|
|
|
for i in range(0, len(example_prompts), 6):
|
|
outputs = all_outputs[i:i + 6]
|
|
|
|
# verify all non-seeded requests differ
|
|
for output_a, output_b in combinations(
|
|
(outputs[0], outputs[1], outputs[2], outputs[3]),
|
|
2,
|
|
):
|
|
assert output_a != output_b
|
|
|
|
# verify requests with the same seed match
|
|
assert outputs[1] == outputs[4]
|
|
assert outputs[2] == outputs[5]
|
|
|
|
# verify generations within the same parallel sampling group differ
|
|
for output in outputs:
|
|
for sub_output_a, sub_output_b in combinations(output, 2):
|
|
assert sub_output_a != sub_output_b
|