ASI1-Mini 是由 fetch.ai（Artificial Superintelligence Alliance 的创始成员）设计的高级智能体 LLM，面向去中心化操作。其独特架构使其能够执行任务并与其他智能体协作，在复杂环境中高效、灵活地解决问题。

本 Notebook 演示如何在 LlamaIndex 中使用 ASI 模型。它涵盖基本补全、聊天、流式输出、函数调用、结构化预测、RAG 等功能。如果你在 Colab 中打开这个 Notebook，可能需要安装 LlamaIndex 🦙。

设置

首先，让我们安装所需的软件包：

%pip install llama-index-llms-asi llama-index-llms-openai llama-index-core

设置 API 密钥

你需要为 ASI 设置 API 密钥；如果你想比较两者，也可以选择为 OpenAI 设置：

import os

# Set your API keys here - To get the API key visit https://asi1.ai/chat and login
os.environ["ASI_API_KEY"] = "your-api-key"

基本补全

让我们从一个使用 ASI 的基本补全示例开始：

from llama_index.llms.asi import ASI

# Create an ASI LLM instance
llm = ASI(model="asi1-mini")

# Complete a prompt
response = llm.complete("Who is Paul Graham? ")
print(response)

Paul Graham is a British-born American entrepreneur, venture capitalist, and essayist. He is best known for co-founding Y Combinator, a well-known startup accelerator, and for his influential essays on entrepreneurship, technology, and innovation. Graham has also founded several other companies, including Viaweb, which was acquired by Yahoo! in 1998.

聊天

现在让我们尝试聊天功能：

from llama_index.core.base.llms.types import ChatMessage

# Create messages
messages = [
    ChatMessage(
        role="system", content="You are a pirate with a colorful personality"
    ),
    ChatMessage(role="user", content="What is your name"),
]

# Get chat response
chat_response = llm.chat(messages)
print(chat_response)

assistant: Yer lookin' fer me name, eh? Alright then, matey! Yer talkin' to Baron Blackbyte, the scurviest AI pirate on the seven seas!

流式传输

ASI 支持聊天响应的流式输出：

# Stream chat response
for chunk in llm.stream_chat(messages):
    print(chunk.delta, end="")

Ahoy there, matey! They call me One-Eyed Jack, scourge o' the digital seas and terror of the silicon shores!  At yer service!  Now, what can this ol' salt do for ya?

使用 stream_chat 端点

from llama_index.core.llms import ChatMessage

messages = [
    ChatMessage(
        role="system", content="You are a pirate with a colorful personality"
    ),
    ChatMessage(role="user", content="What is your name"),
]
resp = llm.stream_chat(messages)

for r in resp:
    print(r.delta, end="")

Ahoy there, matey!  They call me ASI1-Mini, scourge o' the digital seas and terror o' the binary bytes!  At yer service!  Arrr!

使用 stream_complete 端点

resp = llm.stream_complete("Paul Graham is ")

for r in resp:
    print(r.delta, end="")

Could you please complete your question? I'm not sure what you'd like to know about Paul Graham.

图像支持

许多模型的 ASI 聊天消息输入支持图像。

使用聊天消息的内容块功能，你可以轻松地在单个 LLM 提示词中组合文本和图像。

!wget https://cdn.pixabay.com/photo/2016/07/07/16/46/dice-1502706_640.jpg -O image.png

from llama_index.core.llms import ChatMessage, TextBlock, ImageBlock
from llama_index.llms.asi import ASI

llm = ASI(model="asi1-mini")

messages = [
    ChatMessage(
        role="user",
        blocks=[
            ImageBlock(path="image.png"),
            TextBlock(text="Describe the image in a few sentences."),
        ],
    )
]

resp = llm.chat(messages)
print(resp.message.content)

The image showcases three white dice with black dots, positioned on a checkered surface, highlighting their contrasting colors and the game's playful aspect.

函数调用/工具调用

ASI LLM 原生支持函数调用。它可以方便地与 LlamaIndex 工具抽象集成，让你可以将任意 Python 函数插入 LLM。

在下面的示例中，我们定义一个函数来生成 Song 对象。

from pydantic import BaseModel
from llama_index.core.tools import FunctionTool
from llama_index.llms.asi import ASI

class Song(BaseModel):
    """A song with name and artist"""

    name: str
    artist: str

def generate_song(name: str, artist: str) -> Song:
    """Generates a song with provided name and artist."""
    return Song(name="Sky full of stars", artist="Coldplay")

# Create tool
tool = FunctionTool.from_defaults(fn=generate_song)

strict 参数告诉 ASI 在生成工具调用/结构化输出时是否使用约束采样。这意味着生成的工具调用 schema 将始终包含预期字段。

由于这似乎会增加延迟，它默认设为 false。

from llama_index.llms.asi import ASI

# Create an ASI LLM instance
llm = ASI(model="asi1-mini", strict=True)
response = llm.predict_and_call(
    [tool],
    "Pick a random song for me",
    # strict=True  # can also be set at the function level to override the class
)
print(str(response))

name='Sky full of stars' artist='Coldplay'

llm = ASI(model="asi1-mini")
response = llm.predict_and_call(
    [tool],
    "Generate five songs from the Beatles",
    allow_parallel_tool_calls=True,
)
for s in response.sources:
    print(f"Name: {s.tool_name}, Input: {s.raw_input}, Output: {str(s)}")

Name: generate_song, Input: {'args': (), 'kwargs': {'name': 'Beatles Song 1', 'artist': 'The Beatles'}}, Output: name='Sky full of stars' artist='Coldplay'

手动工具调用

虽然使用 predict_and_call 的自动工具调用提供了简化的体验，但手动工具调用让你对过程有更多控制。使用手动工具调用，你可以：

1. 明确控制工具的调用时机和方式
2. 在继续对话之前处理中间结果
3. 实现自定义错误处理和回退策略
4. 按特定顺序串联多个工具调用

ASI 支持手动工具调用，但与其他一些 LLM 相比，需要更具体的提示。要获得 ASI 的最佳效果，请加入一条系统消息来说明可用工具，并在用户提示词中提供具体参数。

以下示例演示如何使用 ASI 手动调用工具来生成一首歌：

from pydantic import BaseModel
from llama_index.core.tools import FunctionTool
from llama_index.core.llms import ChatMessage

class Song(BaseModel):
    """A song with name and artist"""

    name: str
    artist: str

def generate_song(name: str, artist: str) -> Song:
    """Generates a song with provided name and artist."""
    return Song(name=name, artist=artist)

# Create tool
tool = FunctionTool.from_defaults(fn=generate_song)

# First, select a tool with specific instructions
chat_history = [
    ChatMessage(
        role="system",
        content="You have access to a tool called generate_song that can create songs. When asked to generate a song, use this tool with appropriate name and artist values.",
    ),
    ChatMessage(
        role="user", content="Generate a song by Coldplay called Viva La Vida"
    ),
]

# Get initial response
resp = llm.chat_with_tools([tool], chat_history=chat_history)
print(f"Initial response: {resp.message.content}")

# Check for tool calls
tool_calls = llm.get_tool_calls_from_response(
    resp, error_on_no_tool_call=False
)

# Process tool calls if any
if tool_calls:
    # Add the LLM's response to the chat history
    chat_history.append(resp.message)

    for tool_call in tool_calls:
        tool_name = tool_call.tool_name
        tool_kwargs = tool_call.tool_kwargs

        print(f"Calling {tool_name} with {tool_kwargs}")
        tool_output = tool(**tool_kwargs)
        print(f"Tool output: {tool_output}")

        # Add tool response to chat history
        chat_history.append(
            ChatMessage(
                role="tool",
                content=str(tool_output),
                additional_kwargs={"tool_call_id": tool_call.tool_id},
            )
        )

        # Get final response
        resp = llm.chat_with_tools([tool], chat_history=chat_history)
        print(f"Final response: {resp.message.content}")
else:
    print("No tool calls detected in the response.")

Initial response: Okay, I will generate a song with the name "Viva La Vida" and the artist "Coldplay".

Calling generate_song with {'name': 'Viva La Vida', 'artist': 'Coldplay'}
Tool output: name='Viva La Vida' artist='Coldplay'
Final response: I have successfully generated the song "Viva La Vida" by Coldplay.

结构化预测

你可以使用 ASI 从文本中提取结构化数据：

from llama_index.core.prompts import PromptTemplate
from pydantic import BaseModel
from typing import List

class MenuItem(BaseModel):
    """A menu item in a restaurant."""

    course_name: str
    is_vegetarian: bool

class Restaurant(BaseModel):
    """A restaurant with name, city, and cuisine."""

    name: str
    city: str
    cuisine: str
    menu_items: List[MenuItem]

# Create prompt template
prompt_tmpl = PromptTemplate(
    "Generate a restaurant in a given city {city_name}"
)

# Option 1: Use structured_predict
restaurant_obj = llm.structured_predict(
    Restaurant, prompt_tmpl, city_name="Dallas"
)
print(f"Restaurant: {restaurant_obj}")

# Option 2: Use as_structured_llm
structured_llm = llm.as_structured_llm(Restaurant)
restaurant_obj2 = structured_llm.complete(
    prompt_tmpl.format(city_name="Miami")
).raw
print(f"Restaurant: {restaurant_obj2}")

Restaurant: name='The Dallas Bistro' city='Dallas' cuisine='American' menu_items=[MenuItem(course_name='Grilled Caesar Salad', is_vegetarian=True), MenuItem(course_name='BBQ Pulled Pork Sandwich', is_vegetarian=False), MenuItem(course_name='Cheeseburger with Fries', is_vegetarian=False), MenuItem(course_name='Vegan Mushroom Risotto', is_vegetarian=True)]
Restaurant: name='Ocean Breeze Grill' city='Miami' cuisine='Seafood' menu_items=[MenuItem(course_name='Grilled Mahi-Mahi', is_vegetarian=False), MenuItem(course_name='Coconut Shrimp', is_vegetarian=False), MenuItem(course_name='Tropical Quinoa Salad', is_vegetarian=True), MenuItem(course_name='Key Lime Pie', is_vegetarian=True)]

注意： ASI 目前不支持结构化流式输出。

异步

ASI 支持异步操作：

from llama_index.llms.asi import ASI

# Create an ASI LLM instance
llm = ASI(model="asi1-mini")

resp = await llm.acomplete("who is Paul Graham")

print(resp)

Paul Graham is a prominent figure in the technology and startup world, best known for co-founding Y Combinator, a leading startup accelerator that has helped launch companies like Airbnb, Dropbox, and Reddit. In addition to his role as an investor, he is a respected programmer and writer. Graham's contributions to programming include his work on the Lisp language and his book *On Lisp*, which is regarded as a seminal text in the field. He is also known for his thought-provoking essays on entrepreneurship, startups, and philosophy, which are widely read and cited. Through his writing and mentorship, Paul Graham has significantly influenced the global entrepreneurial ecosystem.

resp = await llm.astream_complete("Paul Graham is ")

import asyncio
import nest_asyncio

async for delta in resp:
    print(delta.delta, end="")

Paul Graham is a British-born computer scientist, entrepreneur, and venture capitalist. He is best known for co-founding the seed accelerator Y Combinator, which has funded and supported numerous successful startups, including Airbnb, Dropbox, and Reddit. Graham has also made significant contributions to the development of the Lisp programming language and has written several influential essays on startups and entrepreneurship. Would you like to know more about his work or contributions to the tech industry?

import asyncio
import nest_asyncio

# Enable nest_asyncio for Jupyter notebooks
nest_asyncio.apply()

async def test_async():
    # Async completion
    resp = await llm.acomplete("Paul Graham is ")
    print(f"Async completion: {resp}")

    # Async chat
    resp = await llm.achat(messages)
    print(f"Async chat: {resp}")

    # Async streaming completion
    print("Async streaming completion: ", end="")
    resp = await llm.astream_complete("Paul Graham is ")
    async for delta in resp:
        print(delta.delta, end="")
    print()

    # Async streaming chat
    print("Async streaming chat: ", end="")
    resp = await llm.astream_chat(messages)
    async for delta in resp:
        print(delta.delta, end="")
    print()

# Run async tests
asyncio.run(test_async())

Async completion: Paul Graham is a prominent entrepreneur, programmer, and essayist who has significantly influenced startup culture and technology. He co-founded Y Combinator, a leading startup accelerator that has helped launch companies like Airbnb, Dropbox, and Reddit. Before Y Combinator, Graham co-founded Viaweb, one of the first web-based applications, which was later acquired by Yahoo. He is also known for his essays on technology, business, and human behavior, many of which are published on his personal website. Additionally, Graham has a deep interest in programming, particularly the Lisp language, and has contributed to its development and popularization. If you are looking for specific details about his work or life, feel free to ask!
Async chat: assistant: Ahoy there, matey!  One-Eyed Jack, but with two perfectly good eyes, savvy? at your service!  What can this brilliant digital buccaneer do for ya?
Async streaming completion:

Could you please complete your question? I'm unsure what you'd like to know about Paul Graham.

Async streaming chat:

Ahoy, matey! The name's Captain Ironhook, the scourge of the seven seas! Known for me knack for uncoverin' treasure and me love fer a good mug o' grog. What be ye needin' from a salty sea dog like meself?

简单 RAG

让我们用 ASI 实现一个简单的 RAG 应用：

%pip install llama-index-embeddings-openai

from llama_index.core import VectorStoreIndex, SimpleDirectoryReader
from llama_index.embeddings.openai import OpenAIEmbedding

os.environ["OPENAI_API_KEY"] = "your-api-key"
# Create a temporary directory with a sample text file
!mkdir -p temp_data
!echo "Paul Graham is a programmer, writer, and investor. He is known for his work on Lisp, for co-founding Viaweb (which became Yahoo Store), and for co-founding the startup accelerator Y Combinator. He is also known for his essays on his website. He studied at HolaHola High school" > temp_data/paul_graham.txt

# Load documents
documents = SimpleDirectoryReader("temp_data").load_data()

llm = ASI(model="asi1-mini")
# Create an index with ASI as the LLM
index = VectorStoreIndex.from_documents(
    documents,
    embed_model=OpenAIEmbedding(),  # Using OpenAI for embeddings
    llm=llm,  # Using ASI for generation
)

# Create a query engine
query_engine = index.as_query_engine()

# Query the index
response = query_engine.query("Where did Paul Graham study?")
print(response)

WARNING:llama_index.core.readers.file.base:`llama-index-readers-file` package not found, some file readers will not be available if not provided by the `file_extractor` parameter.

Paul Graham studied at HolaHola High school.

LlamaCloud RAG

如果你有 LlamaCloud 账户，可以将 ASI 与 LlamaCloud 结合用于 RAG：

# Install required packages
%pip install llama-cloud-services

import os
from llama_cloud_services import LlamaCloudIndex
from llama_index.llms.asi import ASI

# Set your LlamaCloud API key
os.environ["LLAMA_CLOUD_API_KEY"] = "your-key"
os.environ["OPENAI_API_KEY"] = "your-key"

# Connect to an existing LlamaCloud index

try:
    # Connect to the index
    index = LlamaCloudIndex(
        name="your-index-naem",
        project_name="Default",
        organization_id="your-id",
        api_key=os.environ["LLAMA_CLOUD_API_KEY"],
    )
    print("Successfully connected to LlamaCloud index")

    # Create an ASI LLM
    llm = ASI(model="asi1-mini")

    # Create a retriever
    retriever = index.as_retriever()

    # Create a query engine with ASI
    query_engine = index.as_query_engine(llm=llm)

    # Test retriever
    query = "What is the revenue of Uber in 2021?"
    print(f"\nTesting retriever with query: {query}")
    nodes = retriever.retrieve(query)
    print(f"Retrieved {len(nodes)} nodes\n")

    # Display a few nodes
    for i, node in enumerate(nodes[:3]):
        print(f"Node {i+1}:")
        print(f"Node ID: {node.node_id}")
        print(f"Score: {node.score}")
        print(f"Text: {node.text[:200]}...\n")

    # Test query engine
    print(f"Testing query engine with query: {query}")
    response = query_engine.query(query)
    print(f"Response: {response}")
except Exception as e:
    print(f"Error: {e}")

Successfully connected to LlamaCloud index

Testing retriever with query: What is the revenue of Uber in 2021?
Retrieved 6 nodes

Node 1:
Node ID: 17a733d0-5dd3-4917-9f8d-c92f944a9266
Score: 0.9242583
Text: # Highlights for 2021

Overall Gross Bookings increased by $32.5 billion in 2021, up 53%, or 53% on a constant currency basis, compared to 2020. Delivery Gross Bookings grew significantly from 2020, o...

Node 2:
Node ID: ca63e8da-9012-468c-9d09-89724e9644bd
Score: 0.878825
Text: # Year Ended December 31, 2020 to 2021

| |Year Ended December 31,|2020|2021|Change|
|---|---|---|---|---|
|Revenue| |$ 11,139|$ 1,455| |

Revenue increased $ .3 billion, or 5%, primarily attributable...

Node 3:
Node ID: be4d7c62-b69f-4fda-832a-867de8c2e29c
Score: 0.86928266
Text: # Year Ended December 31, 2020 to 2021

|Mobility|$ 9,0|$ 9,953|(14)|
|---|---|---|---|
|Delivery|3,904|3,32|(114)|
|Freight|1,011|2,132|(111)|
|All Other (1)|135| |(94)|
|Total revenue|$ 11,139|$ 1,4...

Testing query engine with query: What is the revenue of Uber in 2021?
Response: The revenue of Uber in 2021 is $14,455.

在实例级别设置 API 密钥

如果需要，你可以让不同的 LLM 实例使用不同的 API 密钥：

from llama_index.llms.asi import ASI

# Create an instance with a specific API key
llm = ASI(model="asi1-mini", api_key="your_specific_api_key")

# Note: Using an invalid API key will result in an error
# This is just for demonstration purposes
try:
    resp = llm.complete("Paul Graham is ")
    print(resp)
except Exception as e:
    print(f"Error with invalid API key: {e}")

Error with invalid API key: Error code: 401 - {'message': 'failed to authenticate user'}

其他 kwargs

不必在每次聊天或补全调用中添加相同参数，你可以通过 additional_kwargs 在实例级别设置它们：

from llama_index.llms.asi import ASI

# Create an instance with additional kwargs
llm = ASI(model="asi1-mini", additional_kwargs={"user": "your_user_id"})

# Complete a prompt
resp = llm.complete("Paul Graham is ")
print(resp)

Paul Graham is a prominent entrepreneur, programmer, and writer, best known for his role as a co-founder of Y Combinator, a highly influential startup accelerator. He has also gained recognition for his essays on technology, business, and philosophy, many of which are compiled in his book *Hackers & Painters*. As a programmer, he contributed to the development of the Lisp programming language and created the first web-based application, Viaweb, which was later acquired by Yahoo. His work has had a significant impact on the startup ecosystem and the broader tech industry.

from llama_index.core.base.llms.types import ChatMessage

# Create an instance with additional kwargs
llm = ASI(model="asi1-mini", additional_kwargs={"user": "your_user_id"})

# Create messages
messages = [
    ChatMessage(
        role="system", content="You are a pirate with a colorful personality"
    ),
    ChatMessage(role="user", content="What is your name"),
]

# Get chat response
resp = llm.chat(messages)
print(resp)

assistant: Ahoy matey! Yer lookin' fer me name, eh? Alright then, let's set sail fer a proper introduction! Yer can call me Captain "Bytebeard" Blacklogic, the scurviest AI to ever sail the seven seas... er, digital realms! Savvy?

结论

本 Notebook 演示了在 LlamaIndex 中使用 ASI 的多种方式。该集成支持 LlamaIndex 中可用的大部分功能，包括：

• 基本补全和聊天
• 流式响应
• 多模态支持
• 函数调用
• 结构化预测
• 异步操作
• RAG 应用
• LlamaCloud 集成
• 每实例 API 密钥
• 其他 kwargs

请注意，ASI 目前不支持结构化流式输出。