[XPU]change xpu ci model (#4117)

* change xpu ci model

* change xpu ci model

* change xpu ci model

* change xpu ci model

* Update model path and XPU settings in run_ci_xpu.sh

* Increase health check timeout to 10 minutes

Increased the timeout duration for health checks from 5 minutes to 10 minutes in two places.

* Implement test for OpenAI chat completion

Add a test function for the OpenAI client chat response.

* Change script to use pytest for running tests

* Update health check timeout to 15 minutes

Increase the timeout for health checks from 10 minutes to 15 minutes.

* Add pytest installation to CI script

* Modify base response in test_45t function

Updated the base response message for the test.

* Add V0 and V1 mode test echo statements

---------

Co-authored-by: root <[email protected]>

Author: plusNew001

scripts/run_ci_xpu.sh

@@ -4,31 +4,34 @@ echo "$DIR"
 
 #安装lsof工具
 apt install -y lsof
+
 #先kill一遍
 ps -efww | grep -E 'api_server' | grep -v grep | awk '{print $2}' | xargs kill -9 || true
 ps -efww | grep -E '8188' | grep -v grep | awk '{print $2}' | xargs kill -9 || true
 lsof -t -i :8188 | xargs kill -9 || true
-
-export model_path=${MODEL_PATH}/data/eb45t_4_layer
+#设置模型路径
+export model_path=${MODEL_PATH}/ERNIE-4.5-300B-A47B-Paddle
 
 echo "pip requirements"
 python -m pip install -r requirements.txt
+
 echo "uninstall org"
 python -m pip uninstall paddlepaddle-xpu -y
 python -m pip uninstall fastdeploy-xpu -y
-# 由于主框架更新存在问题，暂时锁死版本
+
 python -m pip install paddlepaddle-xpu -i https://www.paddlepaddle.org.cn/packages/nightly/xpu-p800/
-# python -m pip install https://paddle-whl.bj.bcebos.com/nightly/xpu-p800/paddlepaddle-xpu/paddlepaddle_xpu-3.0.0.dev20250901-cp310-cp310-linux_x86_64.whl
+
 echo "build whl"
 bash custom_ops/xpu_ops/download_dependencies.sh develop
 export CLANG_PATH=$(pwd)/custom_ops/xpu_ops/third_party/xtdk
 export XVLLM_PATH=$(pwd)/custom_ops/xpu_ops/third_party/xvllm
 bash build.sh || exit 1
+
 echo "pip others"
 python -m pip install openai -U
 python -m pip uninstall -y triton
 python -m pip install triton==3.3.0
-
+python -m pip install pytest
 unset http_proxy
 unset https_proxy
 unset no_proxy
@@ -39,19 +42,23 @@ rm -f core*
 # pkill -9 python #流水线不执行这个
 #清空消息队列
 ipcrm --all=msg
-export XPU_VISIBLE_DEVICES="0,1,2,3"
+
+echo "============================开始V0模式测试!============================"
+export ENABLE_V1_KVCACHE_SCHEDULER=1
+export XPU_VISIBLE_DEVICES="0,1,2,3,4,5,6,7"
+
 python -m fastdeploy.entrypoints.openai.api_server \
     --model ${model_path} \
     --port 8188 \
-    --tensor-parallel-size 4 \
+    --tensor-parallel-size 8 \
     --num-gpu-blocks-override 16384 \
     --max-model-len 32768 \
     --max-num-seqs 128 \
     --quantization wint4   > server.log 2>&1 &
 
 sleep 60
 # 探活
-TIMEOUT=$((5 * 60))
+TIMEOUT=$((15 * 60))
 INTERVAL=10            # 检查间隔（秒）
 ENDPOINT="http://0.0.0.0:8188/health"
 START_TIME=$(date +%s) # 记录开始时间戳
@@ -85,7 +92,7 @@ done
 cat server.log
 
 # 执行服务化推理
-python tests/ci_use/XPU_45T/run_45T.py
+python -m pytest tests/ci_use/XPU_45T/run_45T.py
 exit_code=$?
 echo exit_code is ${exit_code}
 
@@ -109,20 +116,21 @@ rm -f core*
 # pkill -9 python #流水线不执行这个
 #清空消息队列
 ipcrm --all=msg
+echo "============================开始V1模式测试!============================"
 export ENABLE_V1_KVCACHE_SCHEDULER=1
-export XPU_VISIBLE_DEVICES="0,1,2,3"
+export XPU_VISIBLE_DEVICES="0,1,2,3,4,5,6,7"
 python -m fastdeploy.entrypoints.openai.api_server \
     --model ${model_path} \
     --port 8188 \
-    --tensor-parallel-size 4 \
+    --tensor-parallel-size 8 \
     --num-gpu-blocks-override 16384 \
     --max-model-len 32768 \
     --max-num-seqs 128 \
     --quantization wint4   > server.log 2>&1 &
 
 sleep 60
 # 探活
-TIMEOUT=$((5 * 60))
+TIMEOUT=$((15 * 60))
 INTERVAL=10            # 检查间隔（秒）
 ENDPOINT="http://0.0.0.0:8188/health"
 START_TIME=$(date +%s) # 记录开始时间戳
@@ -153,7 +161,7 @@ done
 cat server.log
 
 # 执行服务化推理
-python tests/ci_use/XPU_45T/run_45T.py
+python -m pytest tests/ci_use/XPU_45T/run_45T.py
 kv_block_test_exit_code=$?
 echo kv_block_test_exit_code is ${kv_block_test_exit_code}
 

tests/ci_use/XPU_45T/run_45T.py

@@ -14,19 +14,27 @@
 
 import openai
 
-ip = "0.0.0.0"
-service_http_port = "8188"  # 服务配置的
-client = openai.Client(base_url=f"http://{ip}:{service_http_port}/v1", api_key="EMPTY_API_KEY")
 
-# 非流式对话
-response = client.chat.completions.create(
-    model="default",
-    messages=[
-        {"role": "user", "content": "你好，你是谁？"},
-    ],
-    temperature=1,
-    top_p=0,
-    max_tokens=64,
-    stream=False,
-)
-print(response)
+def test_45t():
+    ip = "0.0.0.0"
+    service_http_port = "8188"  # 服务配置的
+    client = openai.Client(base_url=f"http://{ip}:{service_http_port}/v1", api_key="EMPTY_API_KEY")
+    base_response = "你好！我是一个基于人工智能技术构建的助手，可以帮你解答问题、提供建议、辅助创作，或者陪你聊天解闷～😊 无论是学习、工作还是生活中的疑问，都可以随时告诉我哦！你今天有什么想聊的吗？"
+    # 非流式对话
+    response = client.chat.completions.create(
+        model="default",
+        messages=[
+            {"role": "user", "content": "你好，你是谁？"},
+        ],
+        temperature=1,
+        top_p=0,
+        max_tokens=64,
+        stream=False,
+    )
+    print(response.choices[0].message.content)
+    print(base_response)
+    assert response.choices[0].message.content == base_response
+
+
+if __name__ == "__main__":
+    test_45t()