init-llm 0.1.3

A toolkit for initializing LLMs and embeddings with multiple providers

Init LLM

A toolkit for initializing LLMs(Langchain) and embeddings with multiple providers. All models are openai compatible.

Installation

pip install init-llm

Quick Start

from init_llm import ChatLLM, EmbeddingLLM

# Initialize chat model
llm = ChatLLM('openai')  # or 'azure', 'anthropic', etc.
response = llm.invoke('Hello!')

# Initialize embeddings
embeddings = EmbeddingLLM('text-embedding-3-small')
vector = embeddings.embed_query('Hello world')

Configuration

The package uses two configuration files:

• model_providers.toml: Model configurations and provider settings
• .env: API keys and other sensitive information

Auto Configuration

On first use, the package will:

1. Check for configuration files in the working directory
2. If not found, create them automatically with default settings
3. Load environment variables from .env

Custom Configuration

You can specify custom configuration location and env file name:

llm = ChatLLM(
    'openai',
    config_dir='/path/to/config',
    env_file='custom.env'
)

See Supported Providers in model_providers.toml

License

MIT