Skip to content

Quick Start Guide#

Installation#

To get started with byLLM, install the base package:

pip install byllm

AI-Integrated Function Example#

Let's consider an example program where we attempt to categorize a person by their personality using an LLM. For simplicity we will be using names of historical figures.

Standard Implementation#

Traditional implementation of this program with a manual algorithm:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
enum Personality {
    INTROVERT = "Introvert",
    EXTROVERT = "Extrovert",
    AMBIVERT = "Ambivert"
}

def get_personality(name: str) -> Personality {
    # Traditional approach: manual algorithm, prompt-engineered LLM call, etc.
}

with entry {
    name = "Albert Einstein";
    result = get_personality(name);
    print(f"{result.value} personality detected for {name}");
}

Limitations: Defining an algorithm in code for this problem is difficult, while integrating an LLM to perform the task would require manual prompt engineering, response parsing, and type conversion to be implemented by the developer.

byLLM Implementation#

The by keyword abstraction enables functions to process inputs of any type and generate contextually appropriate outputs of the specified type:

Step 1: Configure LLM Model#

1
2
3
import from byllm {Model}

glob llm = Model(model_name="gemini/gemini-2.0-flash");

Step 2: Implement LLM-Integrated Function#

Add by llm to enable LLM integration:

1
def get_personality(name: str) -> Personality by llm();

This will auto-generate a prompt for performing the task and provide an output that strictly adheres to the type Personality.

Step 3: Execute the Application#

Set your API key and run:

export GEMINI_API_KEY="your-api-key-here"
jac run personality.jac

For complete usage methodologies of the by abstraction, refer to the Usage Guide for documentation on object methods, object instantiation, and multi-agent workflows.

byLLM Usage in Python#

As byLLM is a python package, it can be natively used in jac. The following code show the above example application built in native python with byLLM.

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
import jaclang
from byllm import Model, by
from enum import Enum

llm = Model(model_name="gemini/gemini-2.0-flash")

class Personality(Enum):
    INTROVERT = "Introvert"
    EXTROVERT = "Extrovert"
    AMBIVERT = "Ambivert"

@by(model=llm)
def get_personality(name: str) -> Personality: ...

name = "Albert Einstein"
result = get_personality(name)
print(f"{result.value} personality detected for {name}")

To learn more about usage of by in python, please refer to byLLM python Interface.