Skip to main content
Modern AI systems often rely on multiple machine learning frameworks, each optimized for different tasks. The UniversalMLHandler provides a unified, framework-agnostic interface for running machine learning inference across different ML ecosystems. This handler dynamically routes inference requests to the appropriate framework-specific runner (such as Hugging Face, TensorFlow, or scikit-learn) using an internal router. It abstracts away framework-specific initialization and execution details, enabling agents and automation workflows to invoke ML models through a single, consistent API.

​
Example

To run inference using the UniversalMLHandler, specify the framework, model configuration, and input data. The handler automatically loads the required runner and executes inference. 
from superagentx_handlers.ml.universal import UniversalMLHandler

ml_handler = UniversalMLHandler()

​
Run Inference

Run Hugging Face Model:
Executes a Hugging Face pipeline using a unified interface. 
response = await ml_handler.run(
    framework="huggingface",
    task="text-classification",
    model_name="distilbert-base-uncased-finetuned-sst-2-english",
    inputs="This framework is incredibly flexible!",
    device="cpu"
)
print(response["output"])