168 lines
4.6 KiB
Python
168 lines
4.6 KiB
Python
import os
|
|
import unittest
|
|
from types import SimpleNamespace
|
|
|
|
from sglang.srt.utils import kill_process_tree
|
|
from sglang.test.ci.ci_register import register_cuda_ci
|
|
from sglang.test.run_eval import run_eval
|
|
from sglang.test.test_utils import (
|
|
DEFAULT_MODEL_NAME_FOR_TEST_MLA,
|
|
DEFAULT_TIMEOUT_FOR_SERVER_LAUNCH,
|
|
DEFAULT_URL_FOR_TEST,
|
|
CustomTestCase,
|
|
popen_launch_server,
|
|
)
|
|
|
|
register_cuda_ci(est_time=210, stage="extra-b", runner_config="4-gpu-h100")
|
|
|
|
|
|
class TestHybridDPTP(CustomTestCase):
|
|
@classmethod
|
|
def setUpClass(cls):
|
|
cls.model = DEFAULT_MODEL_NAME_FOR_TEST_MLA
|
|
cls.base_url = DEFAULT_URL_FOR_TEST
|
|
cls.process = popen_launch_server(
|
|
cls.model,
|
|
cls.base_url,
|
|
timeout=DEFAULT_TIMEOUT_FOR_SERVER_LAUNCH,
|
|
other_args=[
|
|
"--trust-remote-code",
|
|
"--tp",
|
|
"4",
|
|
"--enable-dp-attention",
|
|
"--dp",
|
|
"2",
|
|
"--moe-a2a-backend",
|
|
"deepep",
|
|
"--cuda-graph-max-bs-decode",
|
|
"128",
|
|
"--max-running-requests",
|
|
"256",
|
|
],
|
|
)
|
|
|
|
@classmethod
|
|
def tearDownClass(cls):
|
|
kill_process_tree(cls.process.pid)
|
|
|
|
def test_gsm8k(self):
|
|
args = SimpleNamespace(
|
|
base_url=self.base_url,
|
|
model=self.model,
|
|
eval_name="gsm8k",
|
|
api="completion",
|
|
max_tokens=512,
|
|
num_examples=200,
|
|
num_threads=128,
|
|
)
|
|
metrics = run_eval(args)
|
|
print(metrics)
|
|
|
|
self.assertGreater(metrics["score"], 0.60)
|
|
|
|
|
|
class TestTBOWithTPAttn(CustomTestCase):
|
|
@classmethod
|
|
def setUpClass(cls):
|
|
cls.model = DEFAULT_MODEL_NAME_FOR_TEST_MLA
|
|
cls.base_url = DEFAULT_URL_FOR_TEST
|
|
cls.process = popen_launch_server(
|
|
cls.model,
|
|
cls.base_url,
|
|
timeout=DEFAULT_TIMEOUT_FOR_SERVER_LAUNCH,
|
|
other_args=[
|
|
"--trust-remote-code",
|
|
"--tp",
|
|
"4",
|
|
"--moe-a2a-backend",
|
|
"deepep",
|
|
"--enable-two-batch-overlap",
|
|
"--cuda-graph-max-bs-decode",
|
|
"128",
|
|
"--max-running-requests",
|
|
"512",
|
|
"--mem-fraction-static", # temp fix as DeepEP buffer is too large.
|
|
"0.7",
|
|
],
|
|
env={
|
|
**os.environ,
|
|
"SGLANG_TBO_DEBUG": "1",
|
|
},
|
|
)
|
|
|
|
@classmethod
|
|
def tearDownClass(cls):
|
|
kill_process_tree(cls.process.pid)
|
|
|
|
def test_gsm8k(self):
|
|
args = SimpleNamespace(
|
|
base_url=self.base_url,
|
|
model=self.model,
|
|
eval_name="gsm8k",
|
|
api="completion",
|
|
max_tokens=512,
|
|
num_examples=200,
|
|
num_threads=128,
|
|
)
|
|
metrics = run_eval(args)
|
|
print(metrics)
|
|
|
|
self.assertGreater(metrics["score"], 0.60)
|
|
|
|
|
|
# There exists bug when using MTP + TBO + attn_tp_size > 1, currently skip that case.
|
|
# @unittest.skip("covered in TestMTPWithTPAttnAndTBO")
|
|
class TestTBOWithTPAttnAndDenseDP(CustomTestCase):
|
|
@classmethod
|
|
def setUpClass(cls):
|
|
cls.model = DEFAULT_MODEL_NAME_FOR_TEST_MLA
|
|
cls.base_url = DEFAULT_URL_FOR_TEST
|
|
cls.process = popen_launch_server(
|
|
cls.model,
|
|
cls.base_url,
|
|
timeout=DEFAULT_TIMEOUT_FOR_SERVER_LAUNCH,
|
|
other_args=[
|
|
"--trust-remote-code",
|
|
"--tp",
|
|
"4",
|
|
"--moe-dense-tp-size",
|
|
"1",
|
|
"--moe-a2a-backend",
|
|
"deepep",
|
|
"--enable-two-batch-overlap",
|
|
"--cuda-graph-max-bs-decode",
|
|
"128",
|
|
"--max-running-requests",
|
|
"512",
|
|
"--mem-fraction-static", # temp fix as DeepEP buffer is too large.
|
|
"0.7",
|
|
],
|
|
env={
|
|
**os.environ,
|
|
"SGLANG_TBO_DEBUG": "1",
|
|
},
|
|
)
|
|
|
|
@classmethod
|
|
def tearDownClass(cls):
|
|
kill_process_tree(cls.process.pid)
|
|
|
|
def test_gsm8k(self):
|
|
args = SimpleNamespace(
|
|
base_url=self.base_url,
|
|
model=self.model,
|
|
eval_name="gsm8k",
|
|
api="completion",
|
|
max_tokens=512,
|
|
num_examples=200,
|
|
num_threads=128,
|
|
)
|
|
metrics = run_eval(args)
|
|
print(metrics)
|
|
|
|
self.assertGreater(metrics["score"], 0.60)
|
|
|
|
|
|
if __name__ == "__main__":
|
|
unittest.main()
|