minke.yu/sglang
1
0
Fork 0
Code Issues Pull Requests Actions 12 Packages Projects Releases Wiki Activity

Unify GSM8K eval path to Chat API for regression CI readiness (#21667)

Browse Source
This commit is contained in:
Liangsheng Yin
2026-04-01 17:12:19 -07:00
committed by GitHub
parent 1081a25983
commit d7256eb69a
79 changed files with 1349 additions and 1359 deletions
Download Patch File Download Diff File Expand all files Collapse all files
test/registered/ep/test_deepep_large.py
+29 -29
View File
@@ -5,7 +5,7 @@ import requests
from sglang.srt.utils import kill_process_tree
from sglang.test.ci.ci_register import register_cuda_ci
from sglang.test.few_shot_gsm8k import run_eval as run_eval_few_shot_gsm8k
from sglang.test.run_eval import run_eval
from sglang.test.send_one import BenchArgs, send_one_prompt
from sglang.test.test_utils import (
DEFAULT_DEEPEP_MODEL_NAME_FOR_TEST,
@@ -67,18 +67,18 @@ class TestDeepseek(CustomTestCase):
def test_gsm8k(self):
args = SimpleNamespace(
num_shots=5,
data_path=None,
num_questions=1200,
parallel=1200,
max_new_tokens=512,
host="http://127.0.0.1",
port=int(self.base_url.split(":")[-1]),
base_url=self.base_url,
model=self.model,
eval_name="gsm8k",
api="completion",
max_tokens=512,
num_examples=1200,
num_threads=1200,
)
metrics = run_eval_few_shot_gsm8k(args)
metrics = run_eval(args)
print(f"Eval accuracy of GSM8K: {metrics=}")
self.assertGreater(metrics["accuracy"], 0.92)
self.assertGreater(metrics["score"], 0.92)
class TestDeepseekMTP(CustomTestCase):
@@ -135,18 +135,18 @@ class TestDeepseekMTP(CustomTestCase):
def test_gsm8k(self):
args = SimpleNamespace(
num_shots=5,
data_path=None,
num_questions=1200,
parallel=1200,
max_new_tokens=512,
host="http://127.0.0.1",
port=int(self.base_url.split(":")[-1]),
base_url=self.base_url,
model=self.model,
eval_name="gsm8k",
api="completion",
max_tokens=512,
num_examples=1200,
num_threads=1200,
)
metrics = run_eval_few_shot_gsm8k(args)
metrics = run_eval(args)
print(f"Eval accuracy of GSM8K: {metrics=}")
self.assertGreater(metrics["accuracy"], 0.92)
self.assertGreater(metrics["score"], 0.92)
server_info = requests.get(self.base_url + "/get_server_info")
avg_spec_accept_length = server_info.json()["internal_states"][0][
@@ -154,7 +154,7 @@ class TestDeepseekMTP(CustomTestCase):
]
print(
f"###test_gsm8k:\n"
f"accuracy={metrics['accuracy']=:.3f}\n"
f"accuracy={metrics['score']=:.3f}\n"
f"{avg_spec_accept_length=:.3f}\n"
)
self.assertGreater(avg_spec_accept_length, 1.85)
@@ -195,17 +195,17 @@ class TestDeepseekV32TBO(CustomTestCase):
self,
): # Append an "a" to make this test run first (alphabetically) to warm up the server
args = SimpleNamespace(
num_shots=5,
data_path=None,
num_questions=1200,
parallel=1200,
max_new_tokens=512,
host="http://127.0.0.1",
port=int(self.base_url.split(":")[-1]),
base_url=self.base_url,
model=self.model,
eval_name="gsm8k",
api="completion",
max_tokens=512,
num_examples=1200,
num_threads=1200,
)
metrics = run_eval_few_shot_gsm8k(args)
metrics = run_eval(args)
print(f"{metrics=}")
self.assertGreater(metrics["accuracy"], 0.92)
self.assertGreater(metrics["score"], 0.92)
def test_bs_1_speed(self):
args = BenchArgs(port=int(self.base_url.split(":")[-1]), max_new_tokens=2048)
test/registered/ep/test_deepep_small.py
+94 -94
View File
@@ -6,7 +6,7 @@ import requests
from sglang.srt.utils import kill_process_tree
from sglang.test.ci.ci_register import register_cuda_ci
from sglang.test.few_shot_gsm8k import run_eval as run_eval_few_shot_gsm8k
from sglang.test.run_eval import run_eval
from sglang.test.test_utils import (
DEFAULT_MODEL_NAME_FOR_TEST_MLA,
DEFAULT_MODEL_NAME_FOR_TEST_MLA_NEXTN,
@@ -52,18 +52,18 @@ class TestPureDP(CustomTestCase):
def test_gsm8k(self):
args = SimpleNamespace(
num_shots=5,
data_path=None,
num_questions=200,
max_new_tokens=512,
parallel=128,
host="http://127.0.0.1",
port=int(self.base_url.split(":")[-1]),
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_few_shot_gsm8k(args)
metrics = run_eval(args)
print(metrics)
self.assertGreater(metrics["accuracy"], 0.60)
self.assertGreater(metrics["score"], 0.60)
class TestHybridDPTP(CustomTestCase):
@@ -97,18 +97,18 @@ class TestHybridDPTP(CustomTestCase):
def test_gsm8k(self):
args = SimpleNamespace(
num_shots=5,
data_path=None,
num_questions=200,
max_new_tokens=512,
parallel=128,
host="http://127.0.0.1",
port=int(self.base_url.split(":")[-1]),
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_few_shot_gsm8k(args)
metrics = run_eval(args)
print(metrics)
self.assertGreater(metrics["accuracy"], 0.60)
self.assertGreater(metrics["score"], 0.60)
class TestTP(CustomTestCase):
@@ -139,18 +139,18 @@ class TestTP(CustomTestCase):
def test_gsm8k(self):
args = SimpleNamespace(
num_shots=5,
data_path=None,
num_questions=200,
max_new_tokens=512,
parallel=128,
host="http://127.0.0.1",
port=int(self.base_url.split(":")[-1]),
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_few_shot_gsm8k(args)
metrics = run_eval(args)
print(metrics)
self.assertGreater(metrics["accuracy"], 0.60)
self.assertGreater(metrics["score"], 0.60)
@unittest.skip("covered in test_deepep_large.py")
@@ -188,18 +188,18 @@ class TestNoGatherdBuffer(CustomTestCase):
def test_gsm8k(self):
args = SimpleNamespace(
num_shots=5,
data_path=None,
num_questions=200,
max_new_tokens=512,
parallel=128,
host="http://127.0.0.1",
port=int(self.base_url.split(":")[-1]),
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_few_shot_gsm8k(args)
metrics = run_eval(args)
print(metrics)
self.assertGreater(metrics["accuracy"], 0.60)
self.assertGreater(metrics["score"], 0.60)
class TestTBO(CustomTestCase):
@@ -240,18 +240,18 @@ class TestTBO(CustomTestCase):
def test_gsm8k(self):
args = SimpleNamespace(
num_shots=5,
data_path=None,
num_questions=200,
max_new_tokens=512,
parallel=128,
host="http://127.0.0.1",
port=int(self.base_url.split(":")[-1]),
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_few_shot_gsm8k(args)
metrics = run_eval(args)
print(metrics)
self.assertGreater(metrics["accuracy"], 0.60)
self.assertGreater(metrics["score"], 0.60)
class TestTBOWithTPAttn(CustomTestCase):
@@ -289,18 +289,18 @@ class TestTBOWithTPAttn(CustomTestCase):
def test_gsm8k(self):
args = SimpleNamespace(
num_shots=5,
data_path=None,
num_questions=200,
max_new_tokens=512,
parallel=128,
host="http://127.0.0.1",
port=int(self.base_url.split(":")[-1]),
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_few_shot_gsm8k(args)
metrics = run_eval(args)
print(metrics)
self.assertGreater(metrics["accuracy"], 0.60)
self.assertGreater(metrics["score"], 0.60)
# There exists bug when using MTP + TBO + attn_tp_size > 1, currently skip that case.
@@ -342,18 +342,18 @@ class TestTBOWithTPAttnAndDenseDP(CustomTestCase):
def test_gsm8k(self):
args = SimpleNamespace(
num_shots=5,
data_path=None,
num_questions=200,
max_new_tokens=512,
parallel=128,
host="http://127.0.0.1",
port=int(self.base_url.split(":")[-1]),
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_few_shot_gsm8k(args)
metrics = run_eval(args)
print(metrics)
self.assertGreater(metrics["accuracy"], 0.60)
self.assertGreater(metrics["score"], 0.60)
@unittest.skip("covered in TestMTPWithTBO")
@@ -399,18 +399,18 @@ class TestMTP(CustomTestCase):
def test_gsm8k(self):
args = SimpleNamespace(
num_shots=5,
data_path=None,
num_questions=200,
max_new_tokens=512,
parallel=128,
host="http://127.0.0.1",
port=int(self.base_url.split(":")[-1]),
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_few_shot_gsm8k(args)
metrics = run_eval(args)
print(metrics)
self.assertGreater(metrics["accuracy"], 0.60)
self.assertGreater(metrics["score"], 0.60)
server_info = requests.get(self.base_url + "/get_server_info")
avg_spec_accept_length = server_info.json()["internal_states"][0][
@@ -418,7 +418,7 @@ class TestMTP(CustomTestCase):
]
print(
f"###test_gsm8k (deepseek-v3 mtp + dp + tbo):\n"
f"accuracy={metrics['accuracy']=:.3f}\n"
f"accuracy={metrics['score']=:.3f}\n"
f"{avg_spec_accept_length=:.3f}\n"
)
self.assertGreater(avg_spec_accept_length, 2.1)
@@ -473,18 +473,18 @@ class TestMTPWithTBO(CustomTestCase):
def test_gsm8k(self):
args = SimpleNamespace(
num_shots=5,
data_path=None,
num_questions=200,
max_new_tokens=512,
parallel=128,
host="http://127.0.0.1",
port=int(self.base_url.split(":")[-1]),
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_few_shot_gsm8k(args)
metrics = run_eval(args)
print(metrics)
self.assertGreater(metrics["accuracy"], 0.60)
self.assertGreater(metrics["score"], 0.60)
server_info = requests.get(self.base_url + "/get_server_info")
avg_spec_accept_length = server_info.json()["internal_states"][0][
@@ -492,7 +492,7 @@ class TestMTPWithTBO(CustomTestCase):
]
print(
f"###test_gsm8k (deepseek-v3 mtp + dp + tbo):\n"
f"accuracy={metrics['accuracy']=:.3f}\n"
f"accuracy={metrics['score']=:.3f}\n"
f"{avg_spec_accept_length=:.3f}\n"
)
self.assertGreater(avg_spec_accept_length, 2.1)
@@ -549,18 +549,18 @@ class TestMTPWithTPAttnAndTBO(CustomTestCase):
def test_gsm8k(self):
args = SimpleNamespace(
num_shots=5,
data_path=None,
num_questions=200,
max_new_tokens=512,
parallel=128,
host="http://127.0.0.1",
port=int(self.base_url.split(":")[-1]),
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_few_shot_gsm8k(args)
metrics = run_eval(args)
print(metrics)
self.assertGreater(metrics["accuracy"], 0.60)
self.assertGreater(metrics["score"], 0.60)
server_info = requests.get(self.base_url + "/get_server_info")
avg_spec_accept_length = server_info.json()["internal_states"][0][
@@ -568,7 +568,7 @@ class TestMTPWithTPAttnAndTBO(CustomTestCase):
]
print(
f"###test_gsm8k (deepseek-v3 mtp + dp + tbo):\n"
f"accuracy={metrics['accuracy']=:.3f}\n"
f"accuracy={metrics['score']=:.3f}\n"
f"{avg_spec_accept_length=:.3f}\n"
)
self.assertGreater(avg_spec_accept_length, 2.1)
test/registered/ep/test_mooncake_ep_small.py
+10 -10
View File
@@ -4,7 +4,7 @@ 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.few_shot_gsm8k import run_eval as run_eval_few_shot_gsm8k
from sglang.test.run_eval import run_eval
from sglang.test.server_fixtures.disaggregation_fixture import get_rdma_devices_args
from sglang.test.test_utils import (
DEFAULT_MODEL_NAME_FOR_TEST_MLA,
@@ -69,18 +69,18 @@ class TestTP(CustomTestCase):
def test_gsm8k(self):
args = SimpleNamespace(
num_shots=5,
data_path=None,
num_questions=200,
max_new_tokens=512,
parallel=128,
host="http://127.0.0.1",
port=int(self.base_url.split(":")[-1]),
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_few_shot_gsm8k(args)
metrics = run_eval(args)
print(metrics)
self.assertGreater(metrics["accuracy"], 0.60)
self.assertGreater(metrics["score"], 0.60)
@unittest.skipIf(is_in_ci(), "Skip since mooncake-ep fault-tolerant test is flaky.")