Meta亲自下场教学Llama3.1 Agent/RAG！

2024-07-26 16:59 湖北

Llama 3.1 Agentic RAG使用指南

这次Llama 3.1发布大家期待较多的是405B模型，但工具调用才是大模型落地的重头戏，Meta这次也开源了一个专门的框架：llama-agentic-system。

运行Llama 3.1作为能够执行“Agentic”任务的系统，例如：

将任务分解并执行多步骤推理

使用工具的能力

内置：模型具有搜索或代码解释器等工具的内置知识

零样本：模型可以学习使用以前未见过的上下文工具定义来调用工具

同时在 Llama3.1 的文档里也详细的写了完整的提示语法和工具调用方法，分内置和自定义调用两种方式：

内置

brave_search（联网搜索）

wolfram_alpha（数学工具）

code interpreter system（代码解释器）

自定义工具调用，模型本身并不直接执行调用，而是通过结构化输出来让执行器完成调用，格式与 OpenAI 的类似

内置示例

使用以下prompt调用三个内置工具：

Brave Search：用于执行网络搜索的工具调用。

Wolfram Alpha：执行复杂数学计算的工具调用。

Code Interpreter：使模型能够输出python代码。

<|begin_of_text|><|start_header_id|>system<|end_header_id|>
Environment: ipythonTools: brave_search, wolfram_alpha
Cutting Knowledge Date: December 2023Today Date: 23 Jul 2024
You are a helpful assistant<|eot_id|><|start_header_id|>user<|end_header_id|>
What is the current weather in Menlo Park, California?<|eot_id|><|start_header_id|>assistant<|end_header_id|>

步骤 - 1 用户提示和系统提示

<|begin_of_text|><|start_header_id|>system<|end_header_id|>
Environment: ipythonTools: brave_search, wolfram_alpha
Cutting Knowledge Date: December 2023Today Date: 23 Jul 2024
You are a helpful Assistant.<|eot_id|><|start_header_id|>user<|end_header_id|>
Can you help me solve this equation: x^3 - 4x^2 + 6x - 24 = 0<|eot_id|><|start_header_id|>assistant<|end_header_id|>

步骤 - 2 模型确定调用哪个工具

<|python_tag|>wolfram_alpha.call(query="solve x^3 - 4x^2 + 6x - 24 = 0")<|eom_id|>

步骤 - 3 由工具即 Wolfram Alpha 生成响应。

{  "queryresult": {    "success": true,    "inputstring": "solve x^3 - 4x^2 + 6x - 24 = 0",    "pods": [      {        "title": "Input interpretation",        "subpods": [          {            "title": "",            "plaintext": "solve x^3 - 4 x^2 + 6 x - 24 = 0"          }        ]      },      {        "title": "Results",        "primary": true,        "subpods": [          {            "title": "",            "plaintext": "x = 4"          },          {            "title": "",            "plaintext": "x = ± (i sqrt(6))"          }        ]      },      ...    ]  }}

步骤 - 4 使用工具响应重新提示模型

<|begin_of_text|><|start_header_id|>system<|end_header_id|>
Environment: ipythonTools: brave_search, wolfram_alpha
Cutting Knowledge Date: December 2023Today Date: 23 Jul 2024
You are a helpful Assistant.<|eot_id|><|start_header_id|>user<|end_header_id|>

Can you help me solve this equation: x^3 - 4x^2 + 6x - 24 = 0<|eot_id|><|start_header_id|>assistant<|end_header_id|>
<|python_tag|>wolfram_alpha.call(query="solve x^3 - 4x^2 + 6x - 24 = 0")<|eom_id|><|start_header_id|>ipython<|end_header_id|>
{"queryresult": {"success": true, "inputstring": "solve x^3 - 4x^2 + 6x - 24 = 0", "pods": [{"title": "Input interpretation", "subpods": [{"title": "", "plaintext": "solve x^3 - 4 x^2 + 6 x - 24 = 0"}]}, {"title": "Results", "primary": true, "subpods": [{"title": "", "plaintext": "x = 4"}, {"title": "", "plaintext": "x = \u00b1 (i sqrt(6))"}]}, ... ]}}<|eot_id|><|start_header_id|>assistant<|end_header_id|>

步骤 - 5 智能体对用户的回复

The solutions to the equation x^3 - 4x^2 + 6x - 24 = 0 are x = 4 and x = ±(i√6).<|eot_id|>

自定义工具调用示例

Meta Llama 3.1模型现在可以从单个消息输出自定义工具调用，以便更轻松地调用工具。模型本身并不执行调用；它提供结构化输出以方便执行器调用。可以在llama-agentic-system中找到一个示例执行器，工具格式类似于OpenAI 定义。

使用自定义工具调用时，要使模型输出eom_id，需要在系统提示符中添加以下指令：Environment: ipython。否则，它应该输出eot_id。

需要调整系统提示以告知模型如何处理工具调用输出

工具定义在用户提示中提供，因为这是模型针对内置 JSON 工具调用进行训练的方式。但是，也可以在系统提示中提供工具定义，并获得类似的结果。开发人员必须测试哪种方式最适合他们的用例。

步骤 - 1 用户提示自定义工具详细信息

<|begin_of_text|><|start_header_id|>system<|end_header_id|>

Environment: ipythonTools: brave_search, wolfram_alpha
Cutting Knowledge Date: December 2023Today Date: 23 Jul 2024
# Tool Instructions- Always execute python code in messages that you share.- When looking for real time information use relevant functions if available else fallback to brave_search    

You have access to the following functions:
Use the function 'spotify_trending_songs' to: Get top trending songs on Spotify{  "name": "spotify_trending_songs",  "description": "Get top trending songs on Spotify",  "parameters": {    "n": {      "param_type": "int",      "description": "Number of trending songs to get",      "required": true    }  }}

If a you choose to call a function ONLY reply in the following format:<{start_tag}={function_name}>{parameters}{end_tag}where
start_tag => `<function`parameters => a JSON dict with the function argument name as key and function argument value as value.end_tag => `</function>`
Here is an example,<function=example_function_name>{"example_name": "example_value"}</function>
Reminder:- Function calls MUST follow the specified format- Required parameters MUST be specified- Only call one function at a time- Put the entire function call reply on one line"- Always add your sources when using search results to answer the user query
You are a helpful Assistant.<|eot_id|><|start_header_id|>user<|end_header_id|>
Can you check the top 5 trending songs on spotify?<|eot_id|><|start_header_id|>assistant<|end_header_id|>

步骤 - 2 模型确定调用哪个工具

{"name": "get_current_conditions", "parameters": {"location": "San Francisco, CA", "unit": "Fahrenheit"}}<eot_id>

步骤 - 3 调用工具的结果被传回模型

<|begin_of_text|><|start_header_id|>system<|end_header_id|>
You are a helpful assistant with tool calling capabilities. When you receive a tool call response, use the output to format an answer to the orginal use question.<|eot_id|><|start_header_id|>user<|end_header_id|>
Given the following functions, please respond with a JSON for a function call with its proper arguments that best answers the given prompt.
Respond in the format {"name": function name, "parameters": dictionary of argument name and its value}. Do not use variables.
{    "type": "function",    "function": {    "name": "get_current_conditions",    "description": "Get the current weather conditions for a specific location",    "parameters": {        "type": "object",        "properties": {        "location": {            "type": "string",            "description": "The city and state, e.g., San Francisco, CA"        },        "unit": {            "type": "string",            "enum": ["Celsius", "Fahrenheit"],            "description": "The temperature unit to use. Infer this from the user's location."        }        },        "required": ["location", "unit"]    }    }}
Question: what is the weather like in San Fransisco?<|eot_id|><|start_header_id|>assistant<|end_header_id|>
{"name": "get_current_conditions", "parameters": {"location": "San Francisco, CA", "unit": "Fahrenheit"}}<|eot_id|><|start_header_id|>ipython<|end_header_id|>
Clouds giving way to sun Hi: 76° Tonight: Mainly clear early, then areas of low clouds forming Lo: 56°"<|eot_id|><|start_header_id|>assistant<|end_header_id|>

步骤 - 4 模型生成用户的最终响应

The weather in Menlo Park is currently cloudy with a high of 76° and a low of 56°, with clear skies expected tonight.<eot_id>

Nvidia Agentic RAG（Llama-3.1）

对于Llama 3.1的强大工具调用功能能力，NVIDIA已与Meta 合作，以确保最新的 Llama 模型能够通过 NVIDIA NIM得到最佳部署。

多Agent框架（例如 LangGraph）使开发人员能够将 LLM 应用程序级逻辑分组为节点和边缘，以便更精细地控制代理决策。带有 NVIDIA LangChain OSS 连接器的LangGraph可用于嵌入、重新排序和使用LLM 实现必要的Agentic RAG技术。

为了实现这一点，应用程序开发人员必须在其RAG管道之上包含更精细的决策。

具有默认路由器的Agentic RAG工作流

带有Web搜索工具的Agentic RAG工作流程

带有问题重写的Agentic RAG工作流程

https://developer.nvidia.com/blog/build-an-agentic-rag-pipeline-with-llama-3-1-and-nvidia-nemo-retriever-nims/https://llama.meta.com/docs/model-cards-and-prompt-formats/llama3_1/https://github.com/meta-llama/llama-agentic-system/tree/main

Fish AI Reader

FishAI

联系邮箱 441953276@qq.com

相关标签