trainer-03 0.1.1

A modular, pipeline-first AI framework that unifies ML, DL, CV, NLP, and detection.

AIForge 🚀

A modular, pipeline-first AI framework that unifies ML, DL, CV, NLP, and detection through a plugin-based architecture.

🧠 Core Philosophy

• Don't build one huge framework: Keep the core lightweight with zero heavy dependencies.
• Plugin Ecosystem: Dynamically load the tools you need (OpenCV, scikit-learn, PyTorch, NLTK, YOLO) only when you need them.

📦 Installation

AIForge is built as a monorepo. You can install it locally:

pip install -e .

To install specific plugin dependencies:

pip install -e .[vision,ml,dl,nlp,yolo]
# Or install everything:
pip install -e .[all]

🔥 Usage

Python API

from aiforge import Pipeline, load_plugin

# Load necessary plugins
load_plugin("vision")
load_plugin("yolo")
load_plugin("nlp")

pipe = Pipeline()
result = (
    pipe
    .add("blur", kernel=5)
    .add("detect_yolo", model_name="yolov8n.pt")
    .add("tokenize")
    .run(input_data)
)

CLI

Run pipelines dynamically via YAML configuration files:

aiforge pipeline.yaml

pipeline.yaml Example:

plugins:
  - vision
  - yolo
input_data: "path/to/image.jpg"
pipeline:
  - blur:
      kernel: 5
  - detect_yolo: {}

🔌 Available Plugins

• vision: OpenCV operations (blur, grayscale, edge_detect)
• ml: scikit-learn models (train_rf, predict_rf)
• dl: PyTorch integration (tensor_conversion, normalize_tensor)
• nlp: NLTK tools (tokenize, lowercase)
• yolo: Ultralytics object detection (detect_yolo)