Update CI (#3474)

* update CI cases

* update CI cases

* update CI cases

* update CI cases

* Merge upstream/develop and resolve directory rename conflict

* Merge upstream/develop and resolve directory rename conflict

* Merge upstream/develop and resolve directory rename conflict

* update deploy

* update deploy

* update deploy

* update deploy

* update deploy

.github/workflows/_accuracy_test.yml

@@ -132,6 +132,7 @@ jobs:
           -v "${CACHE_DIR}/ConfigDir:/root/.config" \
           -e TZ="Asia/Shanghai" \
           --gpus '"device='"${DEVICES}"'"' ${docker_image} /bin/bash -xc '
+          #python -m pip install --pre paddlepaddle-gpu -i https://www.paddlepaddle.org.cn/packages/nightly/cu126/
           python -m pip install paddlepaddle-gpu==3.0.0.dev20250818 -i https://www.paddlepaddle.org.cn/packages/nightly/cu126/
 
           pip config set global.index-url https://mirrors.tuna.tsinghua.edu.cn/pypi/web/simple

.github/workflows/_base_test.yml

@@ -132,6 +132,7 @@ jobs:
           -v "${CACHE_DIR}/ConfigDir:/root/.config" \
           -e TZ="Asia/Shanghai" \
           --gpus '"device='"${DEVICES}"'"' ${docker_image} /bin/bash -xc '
+          #python -m pip install --pre paddlepaddle-gpu -i https://www.paddlepaddle.org.cn/packages/nightly/cu126/
           python -m pip install paddlepaddle-gpu==3.0.0.dev20250818 -i https://www.paddlepaddle.org.cn/packages/nightly/cu126/
 
           pip config set global.index-url https://mirrors.tuna.tsinghua.edu.cn/pypi/web/simple
@@ -155,7 +156,19 @@ jobs:
             -H "Content-Type: application/json" \
             -d "{\"--model\": \"/MODELDATA/ERNIE-4.5-0.3B-Paddle\"}"
 
-          curl -X POST http://localhost:${FLASK_PORT}/wait_for_infer?timeout=90
+          check_service() {
+            local timeout=${1:-90}
+            local url="http://localhost:${FLASK_PORT}/wait_for_infer?timeout=${timeout}"
+            local resp
+
+            resp=$(curl -s -X POST "$url")
+
+            if echo "$resp" | grep -q "服务启动超时"; then
+              exit 8
+            fi
+          }
+
+          check_service 90
           popd
 
           pushd tests/ce/server
@@ -166,8 +179,16 @@ jobs:
           curl -X POST http://0.0.0.0:${FLASK_PORT}/switch \
             -H "Content-Type: application/json" \
             -d "{\"--model\": \"/MODELDATA/ERNIE-4.5-0.3B-Paddle\", \"--early-stop-config\": \"{\\\"enable_early_stop\\\":true, \\\"window_size\\\":6, \\\"threshold\\\":0.93}\"}"
-          curl -X POST http://localhost:${FLASK_PORT}/wait_for_infer?timeout=90
+          check_service 90
           python -m pytest -sv test_repetition_early_stop.py || TEST_EXIT_CODE=1
+
+          curl -X POST http://0.0.0.0:${FLASK_PORT}/switch \
+            -H "Content-Type: application/json" \
+            -d "{\"--model\": \"/MODELDATA/ernie-4_5-21b-a3b-bf16-paddle\", \"--config\": \"21b_mtp.yaml\", \"--enable-logprob\": \"False\"}"
+          check_service 180
+          export TEMPLATE=TOKEN_NORMAL
+          python -m pytest -sv test_seed_usage.py -k "not test_seed_stream" || TEST_EXIT_CODE=1
+
           popd
           echo "TEST_EXIT_CODE=${TEST_EXIT_CODE}" >> /workspace/FastDeploy/exit_code.env
           '

.github/workflows/_build_linux.yml

@@ -148,6 +148,7 @@ jobs:
             elif [[ "${PADDLEVERSION}" != "" ]];then
               python -m pip install paddlepaddle-gpu==${PADDLEVERSION} -i https://www.paddlepaddle.org.cn/packages/stable/cu126/
             else
+              #python -m pip install --pre paddlepaddle-gpu -i https://www.paddlepaddle.org.cn/packages/nightly/cu126/
               python -m pip install paddlepaddle-gpu==3.0.0.dev20250818 -i https://www.paddlepaddle.org.cn/packages/nightly/cu126/
             fi
 

.github/workflows/_logprob_test_linux.yml

@@ -122,6 +122,7 @@ jobs:
           -v "${CACHE_DIR}/ConfigDir:/root/.config" \
           -e TZ="Asia/Shanghai" \
           --gpus '"device='"${DEVICES}"'"' ${docker_image} /bin/bash -xc '
+          #python -m pip install --pre paddlepaddle-gpu -i https://www.paddlepaddle.org.cn/packages/nightly/cu126/
           python -m pip install paddlepaddle-gpu==3.0.0.dev20250818 -i https://www.paddlepaddle.org.cn/packages/nightly/cu126/
 
           pip config set global.index-url https://mirrors.tuna.tsinghua.edu.cn/pypi/web/simple

.github/workflows/_pre_ce_test.yml

@@ -129,6 +129,7 @@ jobs:
           --gpus "\"device=${DEVICES}\"" ${docker_image} /bin/bash -c '
           git config --global --add safe.directory /workspace/FastDeploy
           cd FastDeploy
+          #python -m pip install --pre paddlepaddle-gpu -i https://www.paddlepaddle.org.cn/packages/nightly/cu126/
           python -m pip install paddlepaddle-gpu==3.0.0.dev20250818 -i https://www.paddlepaddle.org.cn/packages/nightly/cu126/
           python -m pip install ${fd_wheel_url}
           bash scripts/run_pre_ce.sh

.github/workflows/_stable_test.yml

@@ -138,6 +138,7 @@ jobs:
           -v "${CACHE_DIR}/ConfigDir:/root/.config" \
           -e TZ="Asia/Shanghai" \
           --gpus '"device='"${DEVICES}"'"' ${docker_image} /bin/bash -xc '
+          #python -m pip install --pre paddlepaddle-gpu -i https://www.paddlepaddle.org.cn/packages/nightly/cu126/
           python -m pip install paddlepaddle-gpu==3.0.0.dev20250818 -i https://www.paddlepaddle.org.cn/packages/nightly/cu126/
 
           pip config set global.index-url https://mirrors.tuna.tsinghua.edu.cn/pypi/web/simple

.github/workflows/_unit_test_coverage.yml

@@ -144,6 +144,7 @@ jobs:
 
           git config --global --add safe.directory /workspace/FastDeploy
           cd FastDeploy
+          #python -m pip install --pre paddlepaddle-gpu -i https://www.paddlepaddle.org.cn/packages/nightly/cu126/
           python -m pip install paddlepaddle-gpu==3.0.0.dev20250818 -i https://www.paddlepaddle.org.cn/packages/nightly/cu126/
 
           pip config set global.extra-index-url https://mirrors.tuna.tsinghua.edu.cn/pypi/web/simple

scripts/run_unittest.sh

@@ -17,6 +17,7 @@ pwd
 
 git config --global --add safe.directory /workspace1/FastDeploy
 
+#python -m pip install --force-reinstall --pre paddlepaddle-gpu -i https://www.paddlepaddle.org.cn/packages/nightly/cu126/
 python -m pip install --force-reinstall paddlepaddle-gpu==3.0.0.dev20250818 -i https://www.paddlepaddle.org.cn/packages/nightly/cu126/
 python -m pip install --upgrade --force-reinstall -r requirements/unittest/requirements.txt
 bash tools/build_wheel.sh

tests/ce/deploy/21b_mtp.yaml

@@ -0,0 +1,8 @@
+max_model_len: 32768
+max_num_seqs: 128
+tensor_parallel_size: 1
+quantization: wint4
+speculative_config:
+  method: mtp
+  num_speculative_tokens: 1
+  model: /MODELDATA/ernie-4_5-21b-a3b-bf16-paddle/mtp/

tests/ce/deploy/deploy.py

@@ -79,6 +79,7 @@ def build_command(config):
     # 添加配置参数
     for key, value in config.items():
         if "--enable" in key:
+            value = bool(value if isinstance(value, bool) else eval(value))
             if value:
                 cmd.append(key)
         else:
@@ -189,6 +190,8 @@ def stop_server(signum=None, frame=None):
     # 若log目录存在，则重命名为log_timestamp
     if os.path.isdir("./log"):
         os.rename("./log", "./log_{}".format(time.strftime("%Y%m%d%H%M%S")))
+    if os.path.exists("gemm_profiles.json"):
+        os.remove("gemm_profiles.json")
 
     if signum:
         sys.exit(0)

tests/ce/server/core/request_template.py

@@ -18,8 +18,18 @@ TOKEN_LOGPROB = {
     "max_tokens": 10000,
 }
 
+TOKEN_NORMAL = {
+    "model": "default",
+    "temperature": 0,
+    "top_p": 0,
+    "seed": 33,
+    "stream": True,
+    "max_tokens": 10000,
+}
+
 
 TEMPLATES = {
     "TOKEN_LOGPROB": TOKEN_LOGPROB,
+    "TOKEN_NORMAL": TOKEN_NORMAL,
     # "ANOTHER_TEMPLATE": ANOTHER_TEMPLATE
 }

tests/ce/server/test_seed_usage.py

@@ -5,6 +5,7 @@
 
 import json
 
+import pytest
 from core import TEMPLATE, URL, build_request_payload, get_stream_chunks, send_request
 
 
@@ -44,7 +45,7 @@ def test_seed_stream():
         assert l1 == l2, f"top_p=0, 固定seed, logprobs[{idx}]不一致"
 
 
-def test_usage_stream():
+def test_chat_usage_stream():
     """测试payload max_tokens参数"""
     data = {
         "messages": [
@@ -65,9 +66,103 @@ def test_usage_stream():
 
     usage = chunks[-1]["usage"]
     total_tokens = usage["completion_tokens"] + usage["prompt_tokens"]
-    assert data["max_tokens"] >= usage["completion_tokens"], "completion_tokens大于max_tokens"
+    assert data["max_tokens"] >= usage["completion_tokens"], f"completion_tokens大于max_tokens, usage: {usage}"
-    assert data["metadata"]["min_tokens"] <= usage["completion_tokens"], "completion_tokens小于min_tokens"
+    assert (
-    assert usage["total_tokens"] == total_tokens, "total_tokens不等于prompt_tokens + completion_tokens"
+        data["metadata"]["min_tokens"] <= usage["completion_tokens"]
+    ), f"completion_tokens小于min_tokens, usage: {usage}"
+    assert (
+        usage["total_tokens"] == total_tokens
+    ), f"total_tokens不等于prompt_tokens + completion_tokens, usage: {usage}"
+
+
+def test_chat_usage_non_stream():
+    """测试非流式 usage"""
+    data = {
+        "messages": [
+            {"role": "system", "content": "You are a helpful assistant."},
+            {"role": "user", "content": "牛顿的三大运动定律是什么？"},
+        ],
+        "max_tokens": 50,
+        "stream": False,
+        "metadata": {"min_tokens": 10},
+    }
+
+    payload = build_request_payload(TEMPLATE, data)
+
+    response = send_request(url=URL, payload=payload).json()
+    # print(response)
+    # chunks = get_stream_chunks(response)
+    # for idx, chunk in enumerate(chunks):
+    #     print(f"\nchunk[{idx}]:\n{json.dumps(chunk, indent=2, ensure_ascii=False)}")
+
+    usage = response["usage"]
+    total_tokens = usage["completion_tokens"] + usage["prompt_tokens"]
+    assert data["max_tokens"] >= usage["completion_tokens"], f"completion_tokens大于max_tokens, usage: {usage}"
+    assert (
+        data["metadata"]["min_tokens"] <= usage["completion_tokens"]
+    ), f"completion_tokens小于min_tokens, usage: {usage}"
+    assert (
+        usage["total_tokens"] == total_tokens
+    ), f"total_tokens不等于prompt_tokens + completion_tokens, usage: {usage}"
+
+
+@pytest.mark.skip(reason="修复后打开")
+def test_non_chat_usage_stream():
+    """测试completions 流式 usage"""
+    data = {
+        "prompt": "牛顿的三大运动定律是什么？",
+        "max_tokens": 50,
+        "stream": True,
+        "stream_options": {"include_usage": True, "continuous_usage_stats": True},
+        "metadata": {"min_tokens": 10},
+    }
+    completion_url = URL.replace("chat/completions", "completions")
+
+    payload = build_request_payload(TEMPLATE, data)
+
+    response = send_request(url=completion_url, payload=payload, stream=True)
+    chunks = get_stream_chunks(response)
+    # for idx, chunk in enumerate(chunks):
+    #     print(f"\nchunk[{idx}]:\n{json.dumps(chunk, indent=2, ensure_ascii=False)}")
+
+    usage = chunks[-1]["usage"]
+    total_tokens = usage["completion_tokens"] + usage["prompt_tokens"]
+    assert data["max_tokens"] >= usage["completion_tokens"], f"completion_tokens大于max_tokens, usage: {usage}"
+    assert (
+        data["metadata"]["min_tokens"] <= usage["completion_tokens"]
+    ), f"completion_tokens小于min_tokens, usage: {usage}"
+    assert (
+        usage["total_tokens"] == total_tokens
+    ), f"total_tokens不等于prompt_tokens + completion_tokens, usage: {usage}"
+
+
+def test_non_chat_usage_non_stream():
+    """测试completions 非流式 usage"""
+    data = {
+        "prompt": "牛顿的三大运动定律是什么？",
+        "max_tokens": 50,
+        "stream": False,
+        "metadata": {"min_tokens": 10},
+    }
+    completion_url = URL.replace("chat/completions", "completions")
+
+    payload = build_request_payload(TEMPLATE, data)
+
+    response = send_request(url=completion_url, payload=payload).json()
+    # print(response)
+    # chunks = get_stream_chunks(response)
+    # for idx, chunk in enumerate(chunks):
+    #     print(f"\nchunk[{idx}]:\n{json.dumps(chunk, indent=2, ensure_ascii=False)}")
+
+    usage = response["usage"]
+    total_tokens = usage["completion_tokens"] + usage["prompt_tokens"]
+    assert data["max_tokens"] >= usage["completion_tokens"], f"completion_tokens大于max_tokens, usage: {usage}"
+    assert (
+        data["metadata"]["min_tokens"] <= usage["completion_tokens"]
+    ), f"completion_tokens小于min_tokens, usage: {usage}"
+    assert (
+        usage["total_tokens"] == total_tokens
+    ), f"total_tokens不等于prompt_tokens + completion_tokens, usage: {usage}"
 
 
 if __name__ == "__main__":

tests/ci_use/EB_Lite_mtp/test_EB_Lite_serving_mtp.py

@@ -1,386 +0,0 @@
-# Copyright (c) 2025 PaddlePaddle Authors. All Rights Reserved.
-#
-# Licensed under the Apache License, Version 2.0 (the "License");
-# you may not use this file except in compliance with the License.
-# You may obtain a copy of the License at
-#
-#     http://www.apache.org/licenses/LICENSE-2.0
-#
-# Unless required by applicable law or agreed to in writing, software
-# distributed under the License is distributed on an "AS IS" BASIS,
-# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-# See the License for the specific language governing permissions and
-# limitations under the License.
-
-import json
-import os
-import re
-import signal
-import socket
-import subprocess
-import sys
-import time
-
-import openai
-import pytest
-import requests
-
-# Read ports from environment variables; use default values if not set
-FD_API_PORT = int(os.getenv("FD_API_PORT", 8188))
-FD_ENGINE_QUEUE_PORT = int(os.getenv("FD_ENGINE_QUEUE_PORT", 8133))
-FD_METRICS_PORT = int(os.getenv("FD_METRICS_PORT", 8233))
-
-# List of ports to clean before and after tests
-PORTS_TO_CLEAN = [FD_API_PORT, FD_ENGINE_QUEUE_PORT, FD_METRICS_PORT]
-
-
-def is_port_open(host: str, port: int, timeout=1.0):
-    """
-    Check if a TCP port is open on the given host.
-    Returns True if connection succeeds, False otherwise.
-    """
-    try:
-        with socket.create_connection((host, port), timeout):
-            return True
-    except Exception:
-        return False
-
-
-def kill_process_on_port(port: int):
-    """
-    Kill processes that are listening on the given port.
-    Uses `lsof` to find process ids and sends SIGKILL.
-    """
-    try:
-        output = subprocess.check_output(f"lsof -i:{port} -t", shell=True).decode().strip()
-        for pid in output.splitlines():
-            os.kill(int(pid), signal.SIGKILL)
-            print(f"Killed process on port {port}, pid={pid}")
-    except subprocess.CalledProcessError:
-        pass
-
-
-def clean_ports():
-    """
-    Kill all processes occupying the ports listed in PORTS_TO_CLEAN.
-    """
-    for port in PORTS_TO_CLEAN:
-        kill_process_on_port(port)
-
-
-@pytest.fixture(scope="session", autouse=True)
-def setup_and_run_server():
-    """
-    Pytest fixture that runs once per test session:
-    - Cleans ports before tests
-    - Starts the API server as a subprocess
-    - Waits for server port to open (up to 30 seconds)
-    - Tears down server after all tests finish
-    """
-    print("Pre-test port cleanup...")
-    clean_ports()
-
-    base_path = os.getenv("MODEL_PATH")
-    if base_path:
-        model_path = os.path.join(base_path, "ernie-4_5-21b-a3b-bf16-paddle")
-    else:
-        model_path = "./ernie-4_5-21b-a3b-bf16-paddle"
-
-    mtp_model_path = os.path.join(model_path, "mtp")
-    mtp_mode_str = json.dumps({"method": "mtp", "num_speculative_tokens": 1, "model": mtp_model_path})
-
-    log_path = "server.log"
-    cmd = [
-        sys.executable,
-        "-m",
-        "fastdeploy.entrypoints.openai.api_server",
-        "--model",
-        model_path,
-        "--port",
-        str(FD_API_PORT),
-        "--tensor-parallel-size",
-        "1",
-        "--engine-worker-queue-port",
-        str(FD_ENGINE_QUEUE_PORT),
-        "--metrics-port",
-        str(FD_METRICS_PORT),
-        "--max-model-len",
-        "32768",
-        "--max-num-seqs",
-        "128",
-        "--quantization",
-        "wint4",
-        "--speculative-config",
-        mtp_mode_str,
-    ]
-
-    # Start subprocess in new process group
-    with open(log_path, "w") as logfile:
-        process = subprocess.Popen(
-            cmd,
-            stdout=logfile,
-            stderr=subprocess.STDOUT,
-            start_new_session=True,  # Enables killing full group via os.killpg
-        )
-
-    # Wait up to 300 seconds for API server to be ready
-    for _ in range(300):
-        if is_port_open("127.0.0.1", FD_API_PORT):
-            print(f"API server is up on port {FD_API_PORT}")
-            break
-        time.sleep(1)
-    else:
-        print("[TIMEOUT] API server failed to start in 5 minutes. Cleaning up...")
-        try:
-            os.killpg(process.pid, signal.SIGTERM)
-        except Exception as e:
-            print(f"Failed to kill process group: {e}")
-        raise RuntimeError(f"API server did not start on port {FD_API_PORT}")
-
-    yield  # Run tests
-
-    print("\n===== Post-test server cleanup... =====")
-    try:
-        os.killpg(process.pid, signal.SIGTERM)
-        print(f"API server (pid={process.pid}) terminated")
-    except Exception as e:
-        print(f"Failed to terminate API server: {e}")
-
-
-@pytest.fixture(scope="session")
-def api_url(request):
-    """
-    Returns the API endpoint URL for chat completions.
-    """
-    return f"http://0.0.0.0:{FD_API_PORT}/v1/chat/completions"
-
-
-@pytest.fixture(scope="session")
-def metrics_url(request):
-    """
-    Returns the metrics endpoint URL.
-    """
-    return f"http://0.0.0.0:{FD_METRICS_PORT}/metrics"
-
-
-@pytest.fixture
-def headers():
-    """
-    Returns common HTTP request headers.
-    """
-    return {"Content-Type": "application/json"}
-
-
-@pytest.fixture
-def consistent_payload():
-    """
-    Returns a fixed payload for consistency testing,
-    including a fixed random seed and temperature.
-    """
-    return {
-        "messages": [{"role": "user", "content": "用一句话介绍 PaddlePaddle"}],
-        "temperature": 0.9,
-        "top_p": 0,  # fix top_p to reduce randomness
-        "seed": 13,  # fixed random seed
-    }
-
-
-# ==========================
-# Helper function to calculate difference rate between two texts
-# ==========================
-def calculate_diff_rate(text1, text2):
-    """
-    Calculate the difference rate between two strings
-    based on the normalized Levenshtein edit distance.
-    Returns a float in [0,1], where 0 means identical.
-    """
-    if text1 == text2:
-        return 0.0
-
-    len1, len2 = len(text1), len(text2)
-    dp = [[0] * (len2 + 1) for _ in range(len1 + 1)]
-
-    for i in range(len1 + 1):
-        for j in range(len2 + 1):
-            if i == 0 or j == 0:
-                dp[i][j] = i + j
-            elif text1[i - 1] == text2[j - 1]:
-                dp[i][j] = dp[i - 1][j - 1]
-            else:
-                dp[i][j] = 1 + min(dp[i - 1][j], dp[i][j - 1], dp[i - 1][j - 1])
-
-    edit_distance = dp[len1][len2]
-    max_len = max(len1, len2)
-    return edit_distance / max_len if max_len > 0 else 0.0
-
-
-# ==========================
-# Consistency test for repeated runs with fixed payload
-# ==========================
-def test_consistency_between_runs(api_url, headers, consistent_payload):
-    """
-    Test that two runs with the same fixed input produce similar outputs.
-    """
-    # First request
-    resp1 = requests.post(api_url, headers=headers, json=consistent_payload)
-    assert resp1.status_code == 200
-    result1 = resp1.json()
-    content1 = result1["choices"][0]["message"]["content"]
-
-    # Second request
-    resp2 = requests.post(api_url, headers=headers, json=consistent_payload)
-    assert resp2.status_code == 200
-    result2 = resp2.json()
-    content2 = result2["choices"][0]["message"]["content"]
-
-    # Calculate difference rate
-    diff_rate = calculate_diff_rate(content1, content2)
-
-    # Verify that the difference rate is below the threshold
-    assert diff_rate < 0.05, f"Output difference too large ({diff_rate:.4%})"
-
-
-# ==========================
-# OpenAI Client chat.completions Test
-# ==========================
-
-
-@pytest.fixture
-def openai_client():
-    ip = "0.0.0.0"
-    service_http_port = str(FD_API_PORT)
-    client = openai.Client(
-        base_url=f"http://{ip}:{service_http_port}/v1",
-        api_key="EMPTY_API_KEY",
-    )
-    return client
-
-
-# Non-streaming test
-def test_non_streaming_chat(openai_client):
-    """
-    Test non-streaming chat functionality with the local service
-    """
-    response = openai_client.chat.completions.create(
-        model="default",
-        messages=[
-            {"role": "system", "content": "You are a helpful AI assistant."},
-            {"role": "user", "content": "List 3 countries and their capitals."},
-        ],
-        temperature=1,
-        max_tokens=1024,
-        stream=False,
-    )
-
-    assert hasattr(response, "choices")
-    assert len(response.choices) > 0
-    assert hasattr(response.choices[0], "message")
-    assert hasattr(response.choices[0].message, "content")
-
-
-# Streaming test
-def test_streaming_chat(openai_client, capsys):
-    """
-    Test streaming chat functionality with the local service
-    """
-    response = openai_client.chat.completions.create(
-        model="default",
-        messages=[
-            {"role": "system", "content": "You are a helpful AI assistant."},
-            {"role": "user", "content": "List 3 countries and their capitals."},
-            {
-                "role": "assistant",
-                "content": "China(Beijing), France(Paris), Australia(Canberra).",
-            },
-            {"role": "user", "content": "OK, tell more."},
-        ],
-        temperature=1,
-        max_tokens=1024,
-        stream=True,
-    )
-
-    output = []
-    for chunk in response:
-        if hasattr(chunk.choices[0], "delta") and hasattr(chunk.choices[0].delta, "content"):
-            output.append(chunk.choices[0].delta.content)
-    assert len(output) > 2
-
-
-# ==========================
-# OpenAI Client completions Test
-# ==========================
-
-
-def test_non_streaming(openai_client):
-    """
-    Test non-streaming chat functionality with the local service
-    """
-    response = openai_client.completions.create(
-        model="default",
-        prompt="Hello, how are you?",
-        temperature=1,
-        max_tokens=1024,
-        stream=False,
-    )
-
-    # Assertions to check the response structure
-    assert hasattr(response, "choices")
-    assert len(response.choices) > 0
-
-
-def test_streaming(openai_client, capsys):
-    """
-    Test streaming functionality with the local service
-    """
-    response = openai_client.completions.create(
-        model="default",
-        prompt="Hello, how are you?",
-        temperature=1,
-        max_tokens=1024,
-        stream=True,
-    )
-
-    # Collect streaming output
-    output = []
-    for chunk in response:
-        output.append(chunk.choices[0].text)
-    assert len(output) > 0
-
-
-def test_profile_reset_block_num():
-    """测试profile reset_block_num功能，与baseline diff不能超过5%"""
-    log_file = "./log/config.log"
-    baseline = 30065
-
-    if not os.path.exists(log_file):
-        pytest.fail(f"Log file not found: {log_file}")
-
-    with open(log_file, "r") as f:
-        log_lines = f.readlines()
-
-    target_line = None
-    for line in log_lines:
-        if "Reset block num" in line:
-            target_line = line.strip()
-            break
-
-    if target_line is None:
-        pytest.fail("日志中没有Reset block num信息")
-
-    match = re.search(r"total_block_num:(\d+)", target_line)
-    if not match:
-        pytest.fail(f"Failed to extract total_block_num from line: {target_line}")
-
-    try:
-        actual_value = int(match.group(1))
-    except ValueError:
-        pytest.fail(f"Invalid number format: {match.group(1)}")
-
-    lower_bound = baseline * (1 - 0.05)
-    upper_bound = baseline * (1 + 0.05)
-    print(f"Reset total_block_num: {actual_value}. baseline: {baseline}")
-
-    assert lower_bound <= actual_value <= upper_bound, (
-        f"Reset total_block_num {actual_value} 与 baseline {baseline} diff需要在5%以内"
-        f"Allowed range: [{lower_bound:.1f}, {upper_bound:.1f}]"
-    )