add sanity check

scripts/check_deployment_cn.py

@@ -0,0 +1,115 @@
+import argparse
+import json
+import os
+
+from openai import OpenAI
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser(
+        description="检查模型部署是否成功的工具",
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+        epilog="""
+使用示例:
+  python scripts/check_deployment_cn.py --base-url http://localhost:8000/v1 --apikey your-key --model autoglm-phone-9b
+  python scripts/check_deployment_cn.py --base-url http://localhost:8000/v1 --apikey your-key --model autoglm-phone-9b --messages-file custom.json
+        """,
+    )
+
+    parser.add_argument(
+        "--base-url",
+        type=str,
+        required=True,
+        help="API 服务的 base URL，例如: http://localhost:8000/v1",
+    )
+
+    parser.add_argument(
+        "--apikey", type=str, default="EMPTY", help="API 密钥 (默认: EMPTY)"
+    )
+
+    parser.add_argument(
+        "--model",
+        type=str,
+        required=True,
+        help="要测试的模型名称，例如: autoglm-phone-9b",
+    )
+
+    parser.add_argument(
+        "--messages-file",
+        type=str,
+        default="scripts/sample_messages.json",
+        help="包含测试消息的 JSON 文件路径 (默认: scripts/sample_messages.json)",
+    )
+
+    parser.add_argument(
+        "--max-tokens", type=int, default=3000, help="最大生成 token 数 (默认: 3000)"
+    )
+
+    parser.add_argument(
+        "--temperature", type=float, default=0.0, help="采样温度 (默认: 0.0)"
+    )
+
+    parser.add_argument(
+        "--top_p", type=float, default=0.85, help="nucleus sampling 参数 (默认: 0.85)"
+    )
+
+    parser.add_argument(
+        "--frequency_penalty", type=float, default=0.2, help="频率惩罚参数 (默认: 0.2)"
+    )
+
+    args = parser.parse_args()
+
+    # 读取测试消息
+    if not os.path.exists(args.messages_file):
+        print(f"错误: 消息文件 {args.messages_file} 不存在")
+        exit(1)
+
+    with open(args.messages_file) as f:
+        messages = json.load(f)
+
+    base_url = args.base_url
+    api_key = args.apikey
+    model = args.model
+
+    print(f"开始测试模型推理...")
+    print(f"Base URL: {base_url}")
+    print(f"Model: {model}")
+    print(f"Messages file: {args.messages_file}")
+    print("=" * 80)
+
+    try:
+        client = OpenAI(
+            base_url=base_url,
+            api_key=api_key,
+        )
+
+        response = client.chat.completions.create(
+            messages=messages,
+            model=model,
+            max_tokens=args.max_tokens,
+            temperature=args.temperature,
+            top_p=args.top_p,
+            frequency_penalty=args.frequency_penalty,
+            extra_body={"skip_special_tokens": False},
+        )
+
+        print("\n模型推理结果:")
+        print("=" * 80)
+        print(response.choices[0].message.content)
+        print("=" * 80)
+
+        if response.usage:
+            print(f"\n统计信息:")
+            print(f"  - Prompt tokens: {response.usage.prompt_tokens}")
+            print(f"  - Completion tokens: {response.usage.completion_tokens}")
+            print(f"  - Total tokens: {response.usage.total_tokens}")
+
+        print(f"\n请根据上述推理结果判断模型部署是否符合预期。")
+
+    except Exception as e:
+        print(f"\n调用 API 时发生错误:")
+        print(f"错误类型: {type(e).__name__}")
+        print(f"错误信息: {str(e)}")
+        print(
+            "\n提示: 请检查 base_url、api_key 和 model 参数是否正确，以及服务是否正在运行。"
+        )
+        exit(1)