feature: adding gpt-oss 120b nightly test (#18134)

This commit is contained in:
Douglas Yang
2026-02-02 17:11:28 -08:00
committed by GitHub
parent 28e2340725
commit c8da307d7e
2 changed files with 88 additions and 4 deletions

View File

@@ -95,7 +95,7 @@ jobs:
strategy:
fail-fast: false
matrix:
partition: [0, 1, 2]
partition: [0, 1, 2, 3]
env:
RUNNER_LABELS: 8-gpu-h200
steps:
@@ -118,7 +118,7 @@ jobs:
IS_H200: "1"
run: |
cd test
python3 run_suite.py --hw cuda --suite nightly-8-gpu-common --nightly --timeout-per-file=18000 --continue-on-error --auto-partition-id=${{ matrix.partition }} --auto-partition-size=3
python3 run_suite.py --hw cuda --suite nightly-8-gpu-common --nightly --timeout-per-file=18000 --continue-on-error --auto-partition-id=${{ matrix.partition }} --auto-partition-size=4
- name: Run test
timeout-minutes: 30
@@ -179,7 +179,7 @@ jobs:
strategy:
fail-fast: false
matrix:
partition: [0, 1, 2]
partition: [0, 1, 2, 3]
steps:
- name: Checkout code
uses: actions/checkout@v4
@@ -199,7 +199,7 @@ jobs:
GPU_CONFIG: "8-gpu-b200"
run: |
cd test
IS_BLACKWELL=1 python3 run_suite.py --hw cuda --suite nightly-8-gpu-common --nightly --timeout-per-file=12000 --continue-on-error --auto-partition-id=${{ matrix.partition }} --auto-partition-size=3
IS_BLACKWELL=1 python3 run_suite.py --hw cuda --suite nightly-8-gpu-common --nightly --timeout-per-file=12000 --continue-on-error --auto-partition-id=${{ matrix.partition }} --auto-partition-size=4
- name: Collect performance metrics
if: always()

View File

@@ -0,0 +1,84 @@
import unittest
from sglang.test.ci.ci_register import register_cuda_ci
from sglang.test.performance_test_runner import PerformanceTestParams
from sglang.test.run_combined_tests import run_combined_tests
from sglang.test.test_utils import ModelLaunchSettings
# Runs on both H200 and B200 via nightly-8-gpu-common suite
# Higher est_time due to 6 variants with both performance and accuracy tests
register_cuda_ci(est_time=1800, suite="nightly-8-gpu-common", nightly=True)
GPT_OSS_120B_MXFP4_MODEL_PATH = "openai/gpt-oss-120b"
GPT_OSS_120B_EAGLE3_DRAFT_MODEL_PATH = "lmsys/EAGLE3-gpt-oss-120b-bf16"
class TestGptOss120B(unittest.TestCase):
"""Unified test class for GPT-OSS-120B performance and accuracy.
Testing:
- Basic configs for MXFP4
- Full config for MXFP4 with reasoning-parser, tool-call-parser, and MTP
"""
def test_gpt_oss_120b_all_variants(self):
"""Run performance and accuracy for all GPT-OSS-120B variants."""
base_args = [
"--tp=8",
"--trust-remote-code",
"--cuda-graph-max-bs=200",
"--mem-fraction-static=0.93",
]
# Lower batch size for EAGLE3 variants to avoid OOM
base_args_eagle3 = [
"--tp=8",
"--trust-remote-code",
"--cuda-graph-max-bs=100",
"--mem-fraction-static=0.85",
]
parser_args = [
"--reasoning-parser=gpt-oss",
"--tool-call-parser=gpt-oss",
]
eagle3_args = [
"--speculative-algorithm=EAGLE3",
f"--speculative-draft-model-path={GPT_OSS_120B_EAGLE3_DRAFT_MODEL_PATH}",
"--speculative-num-steps=3",
"--speculative-eagle-topk=1",
"--speculative-num-draft-tokens=4",
]
eagle3_env = {
"SGLANG_ENABLE_SPEC_V2": "1",
"SGLANG_ALLOW_OVERWRITE_LONGER_CONTEXT_LEN": "1",
}
variants = [
# Variant 1: MXFP4 baseline
ModelLaunchSettings(
GPT_OSS_120B_MXFP4_MODEL_PATH,
tp_size=8,
extra_args=base_args,
variant="MXFP4",
),
# Variant 2: MXFP4 + Parsers + EAGLE3 (full featured quantized, lower batch size)
ModelLaunchSettings(
GPT_OSS_120B_MXFP4_MODEL_PATH,
tp_size=8,
extra_args=base_args_eagle3 + parser_args + eagle3_args,
env=eagle3_env,
variant="MXFP4+Parsers+EAGLE3",
),
]
run_combined_tests(
models=variants,
test_name="GPT-OSS-120B",
accuracy_params=None,
performance_params=PerformanceTestParams(
profile_dir="performance_profiles_gpt_oss_120b",
),
)
if __name__ == "__main__":
unittest.main()