AgentDOC-W1: Add multi_model_output_agent cookbook docs and demo images

docs/en/Cookbook/multi_model_output_agent.md

@@ -0,0 +1,132 @@
+# multi_modal_output_agent
+
+This project demonstrates how to build a multi-modal intelligent Agent using [LazyLLM](https://github.com/LazyAGI/LazyLLM).
+
+## What you will learn from this tutorial:
+
+- How to call the Stable Diffusion model to generate images.
+- How to wrap and register a function as a tool;
+- How to launch a client service using `WebModule` with just a few lines of code.
+
+## Code Implementation
+
+### Project Dependencies
+
+Make sure you have installed the following dependencies:
+
+```bash
+pip install lazyllm
+```
+
+### Step-by-step Guide
+
+#### Step 1: Initialize LLM and Image Generation Model
+
+```python
+llm = TrainableModule("internlm2-chat-7b").start()
+sd3 = TrainableModule("stable-diffusion-3-medium").start()
+```
+
+#### Step 2: Define the Prompt Task
+
+Convert Chinese content into English drawing prompts
+
+```python
+def convert_to_prompt(content: str) -> str:
+    '''
+    Convert Chinese description into detailed English prompt words for image generation.
+
+    Args:
+        content (str): Chinese text describing the image to be generated.
+    '''
+    system_prompt = (
+        "You are a drawing prompt word master who can convert any Chinese content "
+        "entered by the user into English drawing prompt words. "
+        "In this task, you need to convert any input content into English drawing prompt words, "
+        "and you can enrich and expand the prompt word content."
+    )
+    full_prompt = f"{system_prompt}\n\nInput:\n{content}"
+    llm1 = llm.share()
+    return llm1(full_prompt)
+```
+
+Use **Stable Diffusion** to generate images
+
+```python
+def generate_image(prompt: str) -> str:
+    '''
+    Generate an image based on the given English prompt and style.
+
+    Args:
+        prompt (str): The English prompt describing the image.
+    '''
+    _sd3 = sd3.share()
+    return _sd3(prompt)
+```
+
+
+#### Step 3: Register Tool Functions
+
+Register multiple tool functions using ``@fc_register("tool")``:
+
+```python
+@fc_register("tool")
+def convert_to_prompt(content: str) -> str:
+    '''
+    Convert Chinese description into detailed English prompt words for image generation.
+
+    Args:
+        content (str): Chinese text describing the image to be generated.
+    '''
+    system_prompt = (
+        "You are a drawing prompt word master who can convert any Chinese content "
+        "entered by the user into English drawing prompt words. "
+        "In this task, you need to convert any input content into English drawing prompt words, "
+        "and you can enrich and expand the prompt word content."
+    )
+    full_prompt = f"{system_prompt}\n\nInput:\n{content}"
+    llm1 = llm.share()
+    return llm1(full_prompt)
+```
+
+```python
+@fc_register("tool")
+def generate_image(prompt: str) -> str:
+    '''
+    Generate an image based on the given English prompt and style.
+
+    Args:
+        prompt (str): The English prompt describing the image.
+    '''
+    _sd3 = sd3.share()
+    return _sd3(prompt)
+```
+
+#### Step 4: Use the Agent
+Let the LLM call the tools we defined:
+
+```python
+agent = FunctionCallAgent(llm.share(), tools=["convert_to_prompt", "generate_image"])
+```
+
+### Example Output
+
+Example input:
+
+```python
+query = "A parrot playing soccer"
+```
+
+Is it possible to achieve the above functionality in a simpler way?
+
+Yes! You can use [Pipeline](https://docs.lazyllm.ai/zh-cn/latest/API%20Reference/flow/#lazyllm.flow.Pipeline) control flow to assemble the application with just a few lines of code on the client side:
+
+```python
+with pipeline() as ppl:
+    ppl.llm = lazyllm.TrainableModule('internlm2-chat-7b').prompt(lazyllm.ChatPrompter(prompt))
+    ppl.sd3 = lazyllm.TrainableModule('stable-diffusion-3-medium')
+lazyllm.WebModule(ppl, port=23466).start().wait()
+```
+
+Example:
+![API Agent Demo2](../../assets/agent_sd_demo2.png)

docs/mkdocs.template.yml

@@ -15,6 +15,7 @@ nav:
   - Great Writer: Cookbook/great_writer.md
   - RAG: Cookbook/rag.md
   - Streaming: Cookbook/streaming.md
+  - multi_modal_output_agent: Cookbook/multi_model_output_agent.md
 - Best Practice:
   - Flow: Best Practice/flow.md
   - Flowapp: Best Practice/flowapp.md

docs/zh/Cookbook/multi_model_output_agent.md

@@ -0,0 +1,136 @@
+# multi_modal_output_agent
+
+本项目展示了如何使用 [LazyLLM](https://github.com/LazyAGI/LazyLLM) 构建一个多模态智能 Agent。
+
+通过本节您将学习到 LazyLLM 的以下要点:
+
+- 如何调用Stable Diffusion模型生成图像。
+- 如何封装并注册一个函数为工具；
+- 如何用几行代码就用 `WebModule` 启动客户端服务
+
+## 代码实现
+
+### 项目依赖
+
+确保你已安装以下依赖：
+
+```bash
+pip install lazyllm
+```
+
+### 步骤详解
+
+#### Step 1: 初始化大模型与图像生成模型
+
+```python
+llm = TrainableModule("internlm2-chat-7b").start()
+sd3 = TrainableModule("stable-diffusion-3-medium").start()
+```
+
+#### Step 2: 定义提示任务
+
+将中文内容转换为英文绘画提示词（Prompt）
+
+```python
+def convert_to_prompt(content: str) -> str:
+    '''
+    Convert Chinese description into detailed English prompt words for image generation.
+
+    Args:
+        content (str): Chinese text describing the image to be generated.
+    '''
+    system_prompt = (
+        "You are a drawing prompt word master who can convert any Chinese content "
+        "entered by the user into English drawing prompt words. "
+        "In this task, you need to convert any input content into English drawing prompt words, "
+        "and you can enrich and expand the prompt word content."
+    )
+    full_prompt = f"{system_prompt}\n\nInput:\n{content}"
+    llm1 = llm.share()
+    return llm1(full_prompt)
+```
+
+使用 Stable Diffusion 生成图像
+
+```python
+def generate_image(prompt: str) -> str:
+    '''
+    Generate an image based on the given English prompt and style.
+
+    Args:
+        prompt (str): The English prompt describing the image.
+    '''
+    _sd3 = sd3.share()
+    return _sd3(prompt)
+```
+
+
+#### Step 3: 工具函数注册
+
+使用 `@fc_register("tool")` 注册多个工具函数：
+
+```python
+@fc_register("tool")
+def convert_to_prompt(content: str) -> str:
+    '''
+    Convert Chinese description into detailed English prompt words for image generation.
+
+    Args:
+        content (str): Chinese text describing the image to be generated.
+    '''
+    system_prompt = (
+        "You are a drawing prompt word master who can convert any Chinese content "
+        "entered by the user into English drawing prompt words. "
+        "In this task, you need to convert any input content into English drawing prompt words, "
+        "and you can enrich and expand the prompt word content."
+    )
+    full_prompt = f"{system_prompt}\n\nInput:\n{content}"
+    llm1 = llm.share()
+    return llm1(full_prompt)
+```
+
+```python
+@fc_register("tool")
+def generate_image(prompt: str) -> str:
+    '''
+    Generate an image based on the given English prompt and style.
+
+    Args:
+        prompt (str): The English prompt describing the image.
+    '''
+    _sd3 = sd3.share()
+    return _sd3(prompt)
+```
+
+#### Step 4: 使用Agent
+
+让大模型调用我们定义好的工具：
+
+```python
+agent = FunctionCallAgent(llm.share(), tools=["convert_to_prompt", "generate_image"])
+```
+
+### 示例运行结果
+
+示例输入：
+
+```python
+query = "请画一只正在踢足球的鹦鹉"
+```
+
+得到的图像：
+![API Agent Demo](../../assets/agent_sd_demo1.png)
+
+有没有可能使用更简单的方式一键实现上述功能？
+
+可以通过 [Pipeline](https://docs.lazyllm.ai/zh-cn/latest/API%20Reference/flow/#lazyllm.flow.Pipeline) 控制流来组装应用，只需四行代码，就可以在客户端实现上面的功能：
+
+```python
+with pipeline() as ppl:
+    ppl.llm = lazyllm.TrainableModule('internlm2-chat-7b').prompt(lazyllm.ChatPrompter(prompt))
+    ppl.sd3 = lazyllm.TrainableModule('stable-diffusion-3-medium')
+lazyllm.WebModule(ppl, port=23466).start().wait()
+```
+
+示例：
+![API Agent Demo2](../../assets/agent_sd_demo2.png)