AIConfig Tools: Configuring and Interacting AIConfigs

Overview

AIConfig Tools is a package that simplifies Prompt, Model, and Parameter Management by allowing you to create and manage AIConfigs.

Installation
Creating AIConfig Runtime
Creating and Managing Prompts
Updating Model Settings
Executing and Displaying Output
Saving Configuration to Disk

1) Installation

First, you need to install the AIConfig Tools package and any required dependencies. Use the following command to install it. :

Python

pip install python-aiconfig

2) Creating AIConfig Runtime

To start using AIConfig Tools, create an AIConfig Runtime instance. This runtime will allow you to manage prompts, model settings, and execute AI tasks. Here's how you can create it:

from aiconfig import AIConfigRuntime
aiconfig = AIConfigRuntime.create("demo", "this is a demo AIConfig")

Loading an existing config is easy too.

from aiconfig import AIConfigRuntime
aiconfig = AIConfigRuntime.from_config("/path/to/config")

3) Creating and Managing Prompts

Prompts are input messages or queries you send to a Large Language model. You can create and manage prompts using AIConfig Tools. Here's an example of creating a prompt:

from aiconfig.AIConfigSettings import ModelMetadata, Prompt

prompt = Prompt(
    name="prompt1",
    input="Hi! What are transformers?",
    metadata={
        "model": "gpt-3.5-turbo",
    }
)

aiconfig.add_prompt(prompt.name, prompt)

4) Updating Model Settings

You can also update the settings of Large Language models, such as temperature and top-k. Here's an example of updating the model settings:

model_name = "gpt-3.5-turbo"
model_settings = {
    "top_k": 40,
    "top_p": 0.95,
    "model": "gpt-3.5-turbo",
    "temperature": 0.9
}

aiconfig.add_model(model_name, model_settings)

5) Executing and Displaying Output

Once you have created prompts and updated model settings, you can execute prompts and retrieve the output. Here's an example of executing a prompt and displaying the output:

import asyncio
parameters = {"name": "Demo"}
asyncio.run(aiconfig.run("prompt1", parameters))

latest_output = aiconfig.get_latest_output("prompt1")
output_text = aiconfig.get_output_text("prompt1", latest_output)
print(output_text)

6) Saving Configuration to Disk

You can save your AIConfig configuration to disk for later use or to share with others. This allows you to persist your prompts, model settings, and other configurations. Here's how you can save your AIConfig configuration to a file:

# Save the configuration to a file
aiconfig.save_config("my_aiconfig.json")

More Demos

Checkout the demo folder for example configs and example python notebooks.

aiconfig

aiconfig -- for prompt, model and parameter management

Motivation
Why use aiconfig
Getting Started
Core Components
Capabilities
- Version Control
- Model parser
- Routing
- Evaluation
Debugging
Roadmap
- Multi-modal model support (use with image, audio generation models as well as multi-modal models like GPT-V)
- Routing
- Evaluation

Release files for python-aiconfig 1.0.3

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for python-aiconfig 1.0.3
File	Size	Uploaded
python-aiconfig-1.0.3.tar.gz	26.8 kB	Oct 17, 2023	Details

Built distribution (wheel)

Table of built distributions (wheels) for python-aiconfig 1.0.3
File	Interpreter	ABI	Platform	Reset
python_aiconfig-1.0.3-py3-none-any.whl 23.8 kB Oct 17, 2023	Python 3	none	any	Details

Total release size: 50.6 kB

Release files / python-aiconfig-1.0.3.tar.gz

Download URL	python-aiconfig-1.0.3.tar.gz
Size	26.8 kB
Tags	Source
SHA-256 checksum How to use checksums	`4280d7d7d32dfe8bcf9805c3a34657f10ffafefb631dd3f98ce40da5c518ee5e`
BLAKE2b-256 checksum How to use checksums	`55820b2e6e58609198e667b73873bb7e53ccf0f2ff4d0f22ccfbfa10a716249b`
Upload date	Oct 17, 2023
Uploaded using Trusted Publishing? What is trusted publishing?	No
Uploaded via	`twine/4.0.2 CPython/3.11.4`

Release files / python_aiconfig-1.0.3-py3-none-any.whl

Download URL	python_aiconfig-1.0.3-py3-none-any.whl
Size	23.8 kB
Tags	Python 3
SHA-256 checksum How to use checksums	`3f08075d83c5881555fe217b7412ba435c38e05c79a48a3fcf1c6b7e510d51ec`
BLAKE2b-256 checksum How to use checksums	`635dee10b886c198ecedf2c8aa92ecadad94ec13a148e3ed3b0b1609eb05277b`
Upload date	Oct 17, 2023
Uploaded using Trusted Publishing? What is trusted publishing?	No
Uploaded via	`twine/4.0.2 CPython/3.11.4`