update repr demo

.gitignore

@@ -3,6 +3,7 @@ files/
 outputs/
 test_venv/
 backup*/
+demo/
 
 # Python cache
 __pycache__/

README.md

@@ -20,9 +20,15 @@ CNLLM 为中文大模型提供了一个**统一的 OpenAI 兼容接口层**与
 
 - **统一接口** - 一套接口和参数调用不同中文大模型，返回 OpenAI API 标准响应
 - **参数验证** - 对所有参数进行验证和明确反馈，尤其是厂商原生参数，并支持参数处理行为控制 (`drop_params`)
-- **流式响应** - 通过 `repr()` 进行流式生命周期监测，以及通过 `.still/.think/.tools` 属性访问增量值自动累积
+- **流式响应** - 通过 `repr()` 进行流式生命周期监测，以及通过 `.still/.think/.tools` 属性访问增量字段自动累积
 - **批量能力** - 支持批量任务中单个请求的独立配置、实时批量进度统计 (`.status`)，以及可配置的失败策略 (`stop_on_error`) 和内存管理 (`keep`).
 
+**流式生命周期监控以及模型回复、思考内容、工具调用的自动累积演示：**
+
+![Figure 2][repr]
+
+[repr]: pics/repr.gif
+
 ### 开发者招募
 
 欢迎开发者共同参与 CNLLM 的发展，创建 Pull Request 前请先提交 Issue 说明问题并讨论您的解决方案。
@@ -267,7 +273,6 @@ for chunk in resp:
 # {'id': '...', 'object': '...', 'created': '...', 'model': '...', 'choices': [{'delta': {'content': '实时累积的模型回复', 'reasoning_content': '实时累积的推理过程'}, 'finish_reason': 'None'}]}
 ```
 
-
 ### 2.2 chat completions 批量调用
 
 可通过`prompt`和`messages`参数输入并快速配置全局参数，也可以通过`requests`参数为单个请求进行独立配置。

README_en.md

@@ -20,9 +20,15 @@ Through CNLLM, developers can seamlessly use Chinese LLMs in the OpenAI ecosyste
 
 - **Unified Interface** - One set of interfaces and parameters to call different Chinese LLMs, returns OpenAI API standard format response
 - **Parameter Validation** - Validation and explicit feedback for all parameters, especially vendor native parameters, with support for parameter handling behavior control (`drop_params`)
-- **Streaming Response** - Streaming lifecycle monitoring via `repr()`, and automatic accumulation of incremental values via `.still`/`.think`/`.tools` property access
+- **Streaming Response** - Streaming lifecycle monitoring via `repr()`, and automatic accumulation of incremental fields via `.still`/`.think`/`.tools` property access
 - **Batch Capability** - Independent configuration for single requests in batch tasks, real-time batch progress statistics (`.status`), and configurable failure policy (`stop_on_error`) and memory management (`keep`).
 
+**Streaming lifecycle monitoring and automatic accumulation demonstration for model responses, reasoning content, and tool calls:**
+
+![Figure 2][repr]
+
+[repr]: docs/pics/repr.gif
+
 ### Collaboration Opportunities
 
 Welcome developers to participate in CNLLM's development. Please submit an Issue to discuss your solution before creating a Pull Request.