semantic-kernel/python

Get Started with Semantic Kernel ⚡

Install the latest package:

python -m pip install --upgrade semantic-kernel

If you want to use some of the optional dependencies (OpenAI is installed by default), you can install them with:

python -m pip install --upgrade semantic-kernel[hugging_face]

or all of them:

python -m pip install --upgrade semantic-kernel[all]

AI Services

OpenAI / Azure OpenAI API keys

Make sure you have an OpenAI API Key or Azure OpenAI service key

There are two methods to manage keys, secrets, and endpoints:

1. Store them in environment variables. SK Python leverages pydantic settings to load keys, secrets, and endpoints. This means that there is a first attempt to load them from environment variables. The .env file naming applies to how the names should be stored as environment variables.

2. If you'd like to use the .env file, you will need to configure the .env file with the following keys in the file (see the .env.example file):

OPENAI_API_KEY=""
OPENAI_ORG_ID=""
AZURE_OPENAI_CHAT_DEPLOYMENT_NAME=""
AZURE_OPENAI_TEXT_DEPLOYMENT_NAME=""
AZURE_OPENAI_EMBEDDING_DEPLOYMENT_NAME=""
AZURE_OPENAI_ENDPOINT=""
AZURE_OPENAI_API_KEY=""

You will then configure the Text/ChatCompletion class with the keyword argument env_file_path:

chat_completion = OpenAIChatCompletion(service_id="test", env_file_path=<path_to_file>)

This optional env_file_path parameter will allow pydantic settings to use the .env file as a fallback to read the settings.

Running a prompt

import asyncio
from semantic_kernel import Kernel
from semantic_kernel.connectors.ai.open_ai import OpenAIChatCompletion, AzureChatCompletion
from semantic_kernel.prompt_template import PromptTemplateConfig

kernel = Kernel()

# Prepare OpenAI service using credentials stored in the `.env` file
service_id="chat-gpt"
kernel.add_service(
    OpenAIChatCompletion(
        service_id=service_id,
    )
)

# Alternative using Azure:
# kernel.add_service(
#   AzureChatCompletion(
#       service_id=service_id,
#   )
# )

# Define the request settings
req_settings = kernel.get_prompt_execution_settings_from_service_id(service_id)
req_settings.max_tokens = 2000
req_settings.temperature = 0.7
req_settings.top_p = 0.8

prompt = """
1) A robot may not injure a human being or, through inaction,
allow a human being to come to harm.

2) A robot must obey orders given it by human beings except where
such orders would conflict with the First Law.

3) A robot must protect its own existence as long as such protection
does not conflict with the First or Second Law.

Give me the TLDR in exactly 5 words."""

prompt_template_config = PromptTemplateConfig(
    template=prompt,
    name="tldr",
    template_format="semantic-kernel",
    execution_settings=req_settings,
)

function = kernel.add_function(
    function_name="tldr_function",
    plugin_name="tldr_plugin",
    prompt_template_config=prompt_template_config,
)

# Run your prompt
# Note: functions are run asynchronously
async def main():
    result = await kernel.invoke(function)
    print(result) # => Robots must not harm humans.

if __name__ == "__main__":
    asyncio.run(main())
# If running from a jupyter-notebook:
# await main()

Semantic Prompt Functions are Prompts with input parameters

# Create a reusable function summarize function
summarize = kernel.add_function(
    function_name="tldr_function",
    plugin_name="tldr_plugin",
    prompt="{{$input}}\n\nOne line TLDR with the fewest words.",
    prompt_template_settings=req_settings,
)

# Summarize the laws of thermodynamics
print(await kernel.invoke(summarize, input="""
1st Law of Thermodynamics - Energy cannot be created or destroyed.
2nd Law of Thermodynamics - For a spontaneous process, the entropy of the universe increases.
3rd Law of Thermodynamics - A perfect crystal at zero Kelvin has zero entropy."""))

# Summarize the laws of motion
print(await kernel.invoke(summarize, input="""
1. An object at rest remains at rest, and an object in motion remains in motion at constant speed and in a straight line unless acted on by an unbalanced force.
2. The acceleration of an object depends on the mass of the object and the amount of force applied.
3. Whenever one object exerts a force on another object, the second object exerts an equal and opposite on the first."""))

# Summarize the law of universal gravitation
print(await kernel.invoke(summarize, input="""
Every point mass attracts every single other point mass by a force acting along the line intersecting both points.
The force is proportional to the product of the two masses and inversely proportional to the square of the distance between them."""))

# Output:
# > Energy conserved, entropy increases, zero entropy at 0K.
# > Objects move in response to forces.
# > Gravitational force between two point masses is inversely proportional to the square of the distance between them.

Semantic Kernel Notebooks

The repository contains a few Python and C# notebooks that demonstrates how to get started with the Semantic Kernel.

Python notebooks:

• Getting started with Semantic Kernel
• Loading and configuring Semantic Kernel
• Running AI prompts from file
• Creating Prompt Functions at runtime (i.e. inline functions)
• Using Context Variables to Build a Chat Experience
• Introduction to planners
• Building Memory with Embeddings
• Using Hugging Face for Plugins
• Combining native functions and semantic functions
• Groundedness Checking with Semantic Kernel
• Returning multiple results per prompt
• Streaming completions with Semantic Kernel

SK Frequently Asked Questions

How does Python SK compare to the C# version of Semantic Kernel?

The two SDKs are compatible and at the core they follow the same design principles. Some features are still available only in the C# version, and being ported Refer to the FEATURE MATRIX doc to see where things stand in matching the features and functionality of the main SK branch. Over time there will be some features available only in the Python version, and others only in the C# version, for example adapters to external services, scientific libraries, etc.