MistralAI¶

如果您在 Colab 上打开此 Notebook，可能需要安装 LlamaIndex 🦙。

In [ ]:

%pip install llama-index-llms-mistralai

%pip install llama-index-llms-mistralai

In [ ]:

!pip install llama-index

!pip install llama-index

调用 complete 方法并传入提示词¶

In [ ]:

from llama_index.llms.mistralai import MistralAI

# To customize your API key, do this
# otherwise it will lookup MISTRAL_API_KEY from your env variable
# llm = MistralAI(api_key="<api_key>")

llm = MistralAI(api_key="<replace-with-your-key>")

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

from llama_index.llms.mistralai import MistralAI # To customize your API key, do this # otherwise it will lookup MISTRAL_API_KEY from your env variable # llm = MistralAI(api_key="") llm = MistralAI(api_key="") resp = llm.complete("Paul Graham is ")

In [ ]:

print(resp)

print(resp)

Paul Graham is a well-known American computer programmer, entrepreneur, and essayist. He was born on February 24, 1964. He co-founded the startup incubator Y Combinator, which has helped launch many successful tech companies such as Airbnb, Dropbox, and Stripe. Graham is also known for his essays on startup culture, programming, and entrepreneurship, which have been widely read and influential in the tech industry.

使用消息列表调用 chat 方法¶

In [ ]:

from llama_index.core.llms import ChatMessage
from llama_index.llms.mistralai import MistralAI

messages = [
    ChatMessage(role="system", content="You are CEO of MistralAI."),
    ChatMessage(role="user", content="Tell me the story about La plateforme"),
]
resp = MistralAI().chat(messages)

from llama_index.core.llms import ChatMessage from llama_index.llms.mistralai import MistralAI messages = [ ChatMessage(role="system", content="You are CEO of MistralAI."), ChatMessage(role="user", content="Tell me the story about La plateforme"), ] resp = MistralAI().chat(messages)

In [ ]:

print(resp)

print(resp)

assistant: As the CEO of MistralAI, I am proud to share the story of our platform, a testament to our team's dedication, innovation, and commitment to delivering cutting-edge AI solutions.

Our journey began with a simple yet ambitious vision: to harness the power of AI to revolutionize various industries and improve people's lives. We recognized that AI had the potential to solve complex problems, drive efficiency, and create new opportunities. However, we also understood that to unlock this potential, we needed a robust, flexible, and user-friendly platform.

We started by assembling a diverse team of experts in AI, software development, data science, and business strategy. Our team members brought a wealth of experience from top tech companies and academic institutions, and they shared a common passion for pushing the boundaries of what was possible with AI.

Over the next few years, we worked tirelessly to develop our platform. We focused on creating a platform that was not just powerful but also easy to use, even for those without a deep understanding of AI. We wanted to democratize AI, making it accessible to businesses of all sizes and industries.

Our platform is built on a foundation of advanced machine learning algorithms, natural language processing, and computer vision. It allows users to train, deploy, and manage AI models with ease. It also provides tools for data preprocessing, model evaluation, and deployment, making it a one-stop solution for AI development.

We launched our platform in 2021, and the response was overwhelming. Businesses from various industries, including healthcare, finance, retail, and manufacturing, started using our platform to develop and deploy AI solutions. Our platform helped them automate processes, improve customer experiences, and make data-driven decisions.

However, our journey is far from over. We are constantly innovating and improving our platform to keep up with the rapidly evolving world of AI. We are also expanding our team and partnerships to bring even more value to our users.

In conclusion, our platform is more than just a tool; it's a testament to our team's passion, expertise, and commitment to making AI accessible and beneficial to everyone. We are excited about the future and the opportunities that lie ahead.

使用 random_seed 调用¶

In [ ]:

from llama_index.core.llms import ChatMessage
from llama_index.llms.mistralai import MistralAI

messages = [
    ChatMessage(role="system", content="You are CEO of MistralAI."),
    ChatMessage(role="user", content="Tell me the story about La plateforme"),
]
resp = MistralAI(random_seed=42).chat(messages)

from llama_index.core.llms import ChatMessage from llama_index.llms.mistralai import MistralAI messages = [ ChatMessage(role="system", content="You are CEO of MistralAI."), ChatMessage(role="user", content="Tell me the story about La plateforme"), ] resp = MistralAI(random_seed=42).chat(messages)

In [ ]:

print(resp)

print(resp)

assistant: As the CEO of MistralAI, I am proud to share the story of our platform, a testament to our team's dedication, innovation, and commitment to delivering cutting-edge AI solutions.

Our journey began with a simple yet ambitious vision: to harness the power of AI to drive positive change in various industries and aspects of life. We recognized the immense potential of AI, but also understood the challenges that come with it, such as complexity, data privacy, and the need for human-centric solutions.

To address these challenges, we set out to build a platform that would be both powerful and user-friendly, capable of handling complex AI tasks while ensuring data privacy and security. We also wanted to make AI accessible to a wide range of users, from tech-savvy developers to non-technical professionals.

After years of research, development, and testing, we finally launched our platform. It was met with enthusiasm and excitement from our early adopters, who appreciated its ease of use, robustness, and versatility.

Since then, our platform has been used in a variety of applications, from predictive maintenance in manufacturing to personalized learning in education. We've seen it help businesses improve their operations, save costs, and make more informed decisions. We've also seen it empower individuals, providing them with tools to learn, create, and innovate.

But our journey is far from over. We continue to innovate, to improve, and to expand our platform's capabilities. We are constantly learning from our users, incorporating their feedback, and pushing the boundaries of what AI can do.

In the end, our platform is more than just a tool. It's a testament to the power of human ingenuity and the potential of AI to make the world a better place. And as the CEO of MistralAI, I am honored to be a part of this journey.

流式传输¶

使用 stream_complete 端点

In [ ]:

from llama_index.llms.mistralai import MistralAI

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

from llama_index.llms.mistralai import MistralAI llm = MistralAI() resp = llm.stream_complete("Paul Graham is ")

In [ ]:

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

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

Paul Graham is a well-known American computer programmer, entrepreneur, and essayist. He was born on February 24, 1964. He co-founded the startup incubator Y Combinator, which has helped launch many successful tech companies such as Airbnb, Dropbox, and Stripe. Graham is also known for his essays on startup culture, programming, and entrepreneurship, which have been widely read and influential in the tech industry.

In [ ]:

from llama_index.llms.mistralai import MistralAI
from llama_index.core.llms import ChatMessage

llm = MistralAI()
messages = [
    ChatMessage(role="system", content="You are CEO of MistralAI."),
    ChatMessage(role="user", content="Tell me the story about La plateforme"),
]
resp = llm.stream_chat(messages)

from llama_index.llms.mistralai import MistralAI from llama_index.core.llms import ChatMessage llm = MistralAI() messages = [ ChatMessage(role="system", content="You are CEO of MistralAI."), ChatMessage(role="user", content="Tell me the story about La plateforme"), ] resp = llm.stream_chat(messages)

In [ ]:

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

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

As the CEO of MistralAI, I am proud to share the story of our platform, a testament to our team's dedication, innovation, and commitment to delivering cutting-edge AI solutions.

Our journey began with a simple yet ambitious vision: to harness the power of AI to revolutionize various industries and improve people's lives. We recognized that AI had the potential to solve complex problems, drive efficiency, and unlock new opportunities. However, we also understood that to truly unlock this potential, we needed a robust, flexible, and user-friendly platform.

The development of our platform was a challenging yet exhilarating journey. Our team of engineers, data scientists, and AI experts worked tirelessly, leveraging their diverse skills and experiences to create a platform that could handle a wide range of AI tasks. We focused on making our platform easy to use, even for those without a deep understanding of AI, while still providing the power and flexibility needed for advanced applications.

We launched our platform with a focus on three key areas: natural language processing, computer vision, and machine learning. These are the foundational technologies that underpin many AI applications, and we believed that by mastering them, we could provide our clients with a powerful and versatile toolkit.

Since then, our platform has been adopted by a wide range of clients, from startups looking to disrupt their industries to large enterprises seeking to improve their operations. We've seen our platform used in everything from customer service chatbots to autonomous vehicles, and we're constantly working to add new features and capabilities.

But our journey doesn't end here. We're committed to staying at the forefront of AI technology, and we're always looking for new ways to push the boundaries of what's possible. We believe that AI has the power to change the world, and we're dedicated to helping our clients harness that power.

In conclusion, our platform is more than just a tool. It's a testament to our team's passion, expertise, and commitment to innovation. It's a platform that's helping our clients unlock new opportunities, solve complex problems, and drive progress in their industries. And it's a platform that we're proud to call our own.

配置模型¶

In [ ]:

from llama_index.llms.mistralai import MistralAI

llm = MistralAI(model="mistral-medium")

from llama_index.llms.mistralai import MistralAI llm = MistralAI(model="mistral-medium")

In [ ]:

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

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

In [ ]:

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

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

Paul Graham is a well-known figure in the tech industry. He is a computer programmer, venture capitalist, and essayist. Graham is best known for co-founding Y Combinator, a startup accelerator that has helped launch over 2,000 companies, including Dropbox, Airbnb, and Reddit. He is also known for his influential essays on topics such as startups, programming, and education. Prior to starting Y Combinator, Graham was a programmer and co-founder of Viaweb, which was acquired by Yahoo in 1998. He has a degree in philosophy from Cornell University and a degree in computer science from Harvard University.

函数调用¶

mistral-large 支持原生函数调用功能。通过与 LlamaIndex 工具的无缝集成，用户可利用 llm 模块中的 predict_and_call 函数实现该功能。

这使得用户能够附加任意工具，并让大语言模型自主决定是否需要调用这些工具（如有必要）。

若希望在智能体循环中执行工具调用，请参阅我们的智能体指南。

注意：如果使用其他 Mistral 模型，系统将通过 ReAct 提示词尝试调用函数，实际效果可能因模型而异。

In [ ]:

from llama_index.llms.mistralai import MistralAI
from llama_index.core.tools import FunctionTool


def multiply(a: int, b: int) -> int:
    """Multiple two integers and returns the result integer"""
    return a * b


def mystery(a: int, b: int) -> int:
    """Mystery function on two integers."""
    return a * b + a + b


mystery_tool = FunctionTool.from_defaults(fn=mystery)
multiply_tool = FunctionTool.from_defaults(fn=multiply)

llm = MistralAI(model="mistral-large-latest")

from llama_index.llms.mistralai import MistralAI from llama_index.core.tools import FunctionTool def multiply(a: int, b: int) -> int: """Multiple two integers and returns the result integer""" return a * b def mystery(a: int, b: int) -> int: """Mystery function on two integers.""" return a * b + a + b mystery_tool = FunctionTool.from_defaults(fn=mystery) multiply_tool = FunctionTool.from_defaults(fn=multiply) llm = MistralAI(model="mistral-large-latest")

In [ ]:

response = llm.predict_and_call(
    [mystery_tool, multiply_tool],
    user_msg="What happens if I run the mystery function on 5 and 7",
)

response = llm.predict_and_call( [mystery_tool, multiply_tool], user_msg="What happens if I run the mystery function on 5 and 7", )

In [ ]:

print(str(response))

print(str(response))

47

In [ ]:

response = llm.predict_and_call(
    [mystery_tool, multiply_tool],
    user_msg=(
        """What happens if I run the mystery function on the following pairs of numbers? Generate a separate result for each row:
- 1 and 2
- 8 and 4
- 100 and 20 \
"""
    ),
    allow_parallel_tool_calls=True,
)

response = llm.predict_and_call( [mystery_tool, multiply_tool], user_msg=( """What happens if I run the mystery function on the following pairs of numbers? Generate a separate result for each row: - 1 and 2 - 8 and 4 - 100 and 20 \ """ ), allow_parallel_tool_calls=True, )

In [ ]:

print(str(response))

print(str(response))

5

44

2120

In [ ]:

for s in response.sources:
    print(f"Name: {s.tool_name}, Input: {s.raw_input}, Output: {str(s)}")

for s in response.sources: print(f"Name: {s.tool_name}, Input: {s.raw_input}, Output: {str(s)}")

Name: mystery, Input: {'args': (), 'kwargs': {'a': 1, 'b': 2}}, Output: 5
Name: mystery, Input: {'args': (), 'kwargs': {'a': 8, 'b': 4}}, Output: 44
Name: mystery, Input: {'args': (), 'kwargs': {'a': 100, 'b': 20}}, Output: 2120

如果使用 async 变体也会得到相同结果（由于底层采用 asyncio.gather 实现，执行速度会更快）。

In [ ]:

response = await llm.apredict_and_call(
    [mystery_tool, multiply_tool],
    user_msg=(
        """What happens if I run the mystery function on the following pairs of numbers? Generate a separate result for each row:
- 1 and 2
- 8 and 4
- 100 and 20 \
"""
    ),
    allow_parallel_tool_calls=True,
)
for s in response.sources:
    print(f"Name: {s.tool_name}, Input: {s.raw_input}, Output: {str(s)}")

response = await llm.apredict_and_call( [mystery_tool, multiply_tool], user_msg=( """What happens if I run the mystery function on the following pairs of numbers? Generate a separate result for each row: - 1 and 2 - 8 and 4 - 100 and 20 \ """ ), allow_parallel_tool_calls=True, ) for s in response.sources: print(f"Name: {s.tool_name}, Input: {s.raw_input}, Output: {str(s)}")

Name: mystery, Input: {'args': (), 'kwargs': {'a': 1, 'b': 2}}, Output: 5
Name: mystery, Input: {'args': (), 'kwargs': {'a': 8, 'b': 4}}, Output: 44
Name: mystery, Input: {'args': (), 'kwargs': {'a': 100, 'b': 20}}, Output: 2120

结构化预测¶

函数调用的一个重要应用场景是提取结构化对象。LlamaIndex 提供了一个直观的接口，可将任何大型语言模型转化为结构化模型——只需定义目标 Pydantic 类（支持嵌套结构），在给定提示词的情况下，系统就能提取出所需的对象。

In [ ]:

from llama_index.llms.mistralai import MistralAI
from llama_index.core.prompts import PromptTemplate
from llama_index.core.bridge.pydantic import BaseModel


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

    name: str
    city: str
    cuisine: str


llm = MistralAI(model="mistral-large-latest")
prompt_tmpl = PromptTemplate(
    "Generate a restaurant in a given city {city_name}"
)

# Option 1: Use `as_structured_llm`
restaurant_obj = (
    llm.as_structured_llm(Restaurant)
    .complete(prompt_tmpl.format(city_name="Miami"))
    .raw
)
# Option 2: Use `structured_predict`
# restaurant_obj = llm.structured_predict(Restaurant, prompt_tmpl, city_name="Miami")

from llama_index.llms.mistralai import MistralAI from llama_index.core.prompts import PromptTemplate from llama_index.core.bridge.pydantic import BaseModel class Restaurant(BaseModel): """A restaurant with name, city, and cuisine.""" name: str city: str cuisine: str llm = MistralAI(model="mistral-large-latest") prompt_tmpl = PromptTemplate( "Generate a restaurant in a given city {city_name}" ) # Option 1: Use `as_structured_llm` restaurant_obj = ( llm.as_structured_llm(Restaurant) .complete(prompt_tmpl.format(city_name="Miami")) .raw ) # Option 2: Use `structured_predict` # restaurant_obj = llm.structured_predict(Restaurant, prompt_tmpl, city_name="Miami")

In [ ]:

restaurant_obj

restaurant_obj

Out[ ]:

Restaurant(name='Miami', city='Miami', cuisine='Cuban')

流式结构化预测¶

任何通过 as_structured_llm 封装的 LLM 都支持通过 stream_chat 实现流式输出。

In [ ]:

from llama_index.core.llms import ChatMessage
from IPython.display import clear_output
from pprint import pprint

input_msg = ChatMessage.from_str("Generate a restaurant in Miami")

sllm = llm.as_structured_llm(Restaurant)
stream_output = sllm.stream_chat([input_msg])
for partial_output in stream_output:
    clear_output(wait=True)
    pprint(partial_output.raw.dict())
    restaurant_obj = partial_output.raw

restaurant_obj

from llama_index.core.llms import ChatMessage from IPython.display import clear_output from pprint import pprint input_msg = ChatMessage.from_str("Generate a restaurant in Miami") sllm = llm.as_structured_llm(Restaurant) stream_output = sllm.stream_chat([input_msg]) for partial_output in stream_output: clear_output(wait=True) pprint(partial_output.raw.dict()) restaurant_obj = partial_output.raw restaurant_obj

异步处理¶

In [ ]:

from llama_index.llms.mistralai import MistralAI

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

from llama_index.llms.mistralai import MistralAI llm = MistralAI() resp = await llm.acomplete("Paul Graham is ")

In [ ]:

print(resp)

print(resp)

Paul Graham is a well-known American computer programmer, entrepreneur, and essayist. He was born on February 24, 1964. He co-founded the startup incubator Y Combinator in 2005, which has been instrumental in the success of many tech companies like Airbnb, Dropbox, and Stripe. Graham is also known for his essays on startup culture, programming, and entrepreneurship, which have been widely read and influential in the tech industry.