LM-Only Pipeline

Let's try to build a simple pipeline that only consists of a language model.

Prerequisites

This example specifically requires:

• Python v3.11 or v3.12

• Pip

• Access to GDP Labs' Gen AI SDK repository

• A valid language model credential (e.g. an OpenAI API key)

Installation

# you can use a Conda environment
pip install gllm-rag -i https://glsdk.gdplabs.id/gen-ai-internal/simple/

Environment Setup

Set a valid language model credential as an environment variable.

• In this example, let's use an OpenAI API key.

Get an OpenAI API key from OpenAI Console.

Linux, macOS, or Windows WSL

export OPENAI_API_KEY="sk-..."

Windows Powershell

$env:OPENAI_API_KEY = "sk-..."

Windows Command Prompt

set OPENAI_API_KEY="sk-..."

Running the Pipeline

1

Create a script called lm.py:

import asyncio
import os

from gllm_rag.preset import LM

async def main():
    lm = LM(
        language_model_id="openai/gpt-4.1-nano", 
        language_model_credentials=os.getenv("OPENAI_API_KEY")
    )
    await lm("Name an animal that starts with the letter 'A'")

if __name__ == "__main__":
    asyncio.run(main())

2

Run the script:

python lm.py

3

The language model will generate a response for the given query, e.g:

An animal that starts with the letter 'A' is an **Alligator**.

Congratulations! You have successfully run your first pipeline!

Switching Between Language Models

Using the above script, we can easily switch to any of the available language model options by simply defining the language_model_id and language_model_credentials parameters.

Please refer below for each of the available options format examples:

Using Anthropic

Setup

1. Model ID Format: anthropic/<model_name>.

2. Anthropic API key: Create one from Anthropic Console.

Example

lm = LM(
    language_model_id="anthropic/claude-sonnet-4-20250514",
    language_model_credentials=os.getenv("ANTHROPIC_API_KEY")
)

Using Google Gen AI

Setup

1. Model ID Format: google/<model_name>.

2. Gemini API key: Create one from Google AI Studio.

Example

lm = LM(
    language_model_id="google/gemini-2.5-flash-preview-05-20",
    language_model_credentials=os.getenv("GEMINI_API_KEY")
)

Using Google Vertex AI

Setup

1. Model ID Format: google/<model_name>.

2. Google Service Account JSON Credential Path:

   1. Create a Google service account.

   2. Create JSON credentials for the service account.

Example

lm = LM(
    language_model_id="google/gemini-2.5-flash-preview-05-20",
    language_model_credentials=os.getenv("CREDENTIALS_PATH")
)

Using OpenAI

Setup

1. Model ID Format: openai/<model_name>.

2. Open AI API key: Create one from OpenAI Console.

Example

lm = LM(
    language_model_id="openai/gpt-4.1-nano",
    language_model_credentials=os.getenv("OPENAI_API_KEY")
)

Using Azure OpenAI

Setup

1. Model ID Format: azure-openai/<endpoint>:<deployment>.

2. Azure OpenAI Resource and Deployment:

   1. Create an Azure OpenAI resource.

   2. Deploy a model.

Example

lm = LM(
    language_model_id="azure-openai/https://<endpoint>.openai.azure.com:<deployment>",
    language_model_credentials=os.getenv("AZURE_OPENAI_API_KEY")
)

Using OpenAI Compatible Endpoints

OpenAI compatible endpoints include but are not limited to the following:

1. DeepInfra

2. DeepSeek

3. Groq

4. OpenRouter

5. Text Generation Inference

6. Together.ai

7. vLLM

Setup

1. Model ID Format: openai-compatible/<base_url>:<model_name>.

2. Varied Credentials Depending on the Endpoints:

   1. In this example, let's use Groq, which requires a Groq API key: Create one from Groq Console.

Example

lm = LM(
    language_model_id="openai-compatible/https://api.groq.com/openai/v1:llama-3.1-8b-instant",
    language_model_credentials=os.getenv("GROQ_API_KEY")
)

Using LangChain

LangChain available packages and classes can be found in the documentation page.

Setup

1. Model ID Format: langchain/<package>.<class>:<model_name>.

2. Varied Credentials Depending on the Package and Class:

   1. In this example, let's use ChatOpenAI, which requires an OpenAI API key: Create one from OpenAI Console.

Example

lm = LM(
    language_model_id="langchain/langchain_openai.ChatOpenAI:gpt-4.1-nano",
    language_model_credentials={"api_key": os.getenv("OPENAI_API_KEY")}
)

Using LiteLLM

Coming soon!