[CI]: Add CI For HiMambaRadixTree and qwen3.5 (#20540)
Co-authored-by: ispobock <ispobaoke@gmail.com>
This commit is contained in:
@@ -1666,7 +1666,7 @@ jobs:
|
|||||||
strategy:
|
strategy:
|
||||||
fail-fast: false
|
fail-fast: false
|
||||||
matrix:
|
matrix:
|
||||||
part: [0, 1, 2]
|
part: [0, 1, 2, 3]
|
||||||
|
|
||||||
steps:
|
steps:
|
||||||
- name: Checkout code
|
- name: Checkout code
|
||||||
@@ -1695,7 +1695,7 @@ jobs:
|
|||||||
if [[ "${{ needs.check-changes.outputs.continue_on_error }}" == "true" ]]; then
|
if [[ "${{ needs.check-changes.outputs.continue_on_error }}" == "true" ]]; then
|
||||||
CONTINUE_ON_ERROR_FLAG="--continue-on-error"
|
CONTINUE_ON_ERROR_FLAG="--continue-on-error"
|
||||||
fi
|
fi
|
||||||
IS_BLACKWELL=1 python3 run_suite.py --hw cuda --suite stage-c-test-4-gpu-b200 --auto-partition-id ${{ matrix.part }} --auto-partition-size 3 --timeout-per-file 1800 $CONTINUE_ON_ERROR_FLAG
|
IS_BLACKWELL=1 python3 run_suite.py --hw cuda --suite stage-c-test-4-gpu-b200 --auto-partition-id ${{ matrix.part }} --auto-partition-size 4 --timeout-per-file 1800 $CONTINUE_ON_ERROR_FLAG
|
||||||
|
|
||||||
- uses: ./.github/actions/upload-cuda-coredumps
|
- uses: ./.github/actions/upload-cuda-coredumps
|
||||||
if: always()
|
if: always()
|
||||||
|
|||||||
@@ -1,3 +1,5 @@
|
|||||||
|
import shutil
|
||||||
|
import tempfile
|
||||||
import unittest
|
import unittest
|
||||||
from types import SimpleNamespace
|
from types import SimpleNamespace
|
||||||
|
|
||||||
@@ -17,13 +19,11 @@ from sglang.test.test_utils import (
|
|||||||
popen_launch_server,
|
popen_launch_server,
|
||||||
)
|
)
|
||||||
|
|
||||||
register_cuda_ci(est_time=1000, suite="stage-c-test-4-gpu-b200")
|
register_cuda_ci(est_time=1400, suite="stage-c-test-4-gpu-b200")
|
||||||
|
|
||||||
QWEN35_FP4_MODEL = "nvidia/Qwen3.5-397B-A17B-NVFP4"
|
QWEN35_FP4_MODEL = "nvidia/Qwen3.5-397B-A17B-NVFP4"
|
||||||
|
QWEN35_27B_MODEL = "Qwen/Qwen3.5-27B"
|
||||||
ACC_THRESHOLDS = {
|
ACC_THRESHOLDS = {QWEN35_FP4_MODEL: {"gsm8k": 0.95}, QWEN35_27B_MODEL: {"gsm8k": 0.8}}
|
||||||
QWEN35_FP4_MODEL: {"gsm8k": 0.95},
|
|
||||||
}
|
|
||||||
|
|
||||||
|
|
||||||
class TestQwen35FP4(CustomTestCase):
|
class TestQwen35FP4(CustomTestCase):
|
||||||
@@ -236,5 +236,78 @@ class TestQwen35FP4MTPV2(CustomTestCase):
|
|||||||
self.assertGreater(avg_spec_accept_length, 3.3)
|
self.assertGreater(avg_spec_accept_length, 3.3)
|
||||||
|
|
||||||
|
|
||||||
|
class TestQwen35WithHiCache(CustomTestCase):
|
||||||
|
@classmethod
|
||||||
|
def setUpClass(cls):
|
||||||
|
cls.model = QWEN35_27B_MODEL
|
||||||
|
cls.base_url = DEFAULT_URL_FOR_TEST
|
||||||
|
cls.storage_dir = tempfile.mkdtemp(prefix="qwen35-hicache-")
|
||||||
|
env = {
|
||||||
|
"SGLANG_HICACHE_FILE_BACKEND_STORAGE_DIR": cls.storage_dir,
|
||||||
|
}
|
||||||
|
cls.process = popen_launch_server(
|
||||||
|
cls.model,
|
||||||
|
cls.base_url,
|
||||||
|
timeout=DEFAULT_TIMEOUT_FOR_SERVER_LAUNCH,
|
||||||
|
env=env,
|
||||||
|
other_args=[
|
||||||
|
"--tp-size",
|
||||||
|
"4",
|
||||||
|
"--chunked-prefill-size",
|
||||||
|
"2048",
|
||||||
|
"--mamba-scheduler-strategy",
|
||||||
|
"extra_buffer",
|
||||||
|
"--mamba-track-interval",
|
||||||
|
"128",
|
||||||
|
"--mamba-ssm-dtype",
|
||||||
|
"bfloat16",
|
||||||
|
"--max-running-requests",
|
||||||
|
"128",
|
||||||
|
"--reasoning-parser",
|
||||||
|
"qwen3",
|
||||||
|
"--model-loader-extra-config",
|
||||||
|
'{"enable_multithread_load": true,"num_threads": 64}',
|
||||||
|
"--hicache-mem-layout",
|
||||||
|
"page_first_direct",
|
||||||
|
"--enable-hierarchical-cache",
|
||||||
|
"--hicache-ratio",
|
||||||
|
"2",
|
||||||
|
"--hicache-size",
|
||||||
|
"0",
|
||||||
|
"--hicache-write-policy",
|
||||||
|
"write_through",
|
||||||
|
"--hicache-storage-backend",
|
||||||
|
"file",
|
||||||
|
"--hicache-storage-prefetch-policy",
|
||||||
|
"wait_complete",
|
||||||
|
],
|
||||||
|
)
|
||||||
|
|
||||||
|
@classmethod
|
||||||
|
def tearDownClass(cls):
|
||||||
|
kill_process_tree(cls.process.pid)
|
||||||
|
shutil.rmtree(cls.storage_dir, ignore_errors=True)
|
||||||
|
|
||||||
|
def test_gsm8k(self):
|
||||||
|
args = SimpleNamespace(
|
||||||
|
model=self.model,
|
||||||
|
eval_name="gsm8k",
|
||||||
|
num_shots=5,
|
||||||
|
num_examples=200,
|
||||||
|
max_tokens=16000,
|
||||||
|
num_threads=128,
|
||||||
|
repeat=1,
|
||||||
|
temperature=0.6,
|
||||||
|
top_p=0.95,
|
||||||
|
top_k=20,
|
||||||
|
base_url=self.base_url,
|
||||||
|
host="http://127.0.0.1",
|
||||||
|
port=int(self.base_url.split(":")[-1]),
|
||||||
|
)
|
||||||
|
metrics = run_eval(args)
|
||||||
|
print(f"{metrics=}")
|
||||||
|
self.assertGreaterEqual(metrics["score"], ACC_THRESHOLDS[self.model]["gsm8k"])
|
||||||
|
|
||||||
|
|
||||||
if __name__ == "__main__":
|
if __name__ == "__main__":
|
||||||
unittest.main()
|
unittest.main()
|
||||||
|
|||||||
Reference in New Issue
Block a user