MultiModal Feature Builder

Build multi-modal feature matrices from text, images, categorical and numeric data in one line.
Use it to train powerful ML models that learn from all your data simultaneously.

---

Features

Extracts features from:

• Text using BERT / any HuggingFace model
• Images using ResNet / EfficientNet
• Categorical & Numeric data with sklearn pipelines

Fully customizable:

• Choose your text model: "bert-base-uncased", "distilbert-base-uncased", "roberta-base", etc.
• Choose your image model: "resnet18", "resnet50", "efficientnet_b0".

Returns:

• Combined X numpy array of multi-modal features
• Target y ready for training.

---

Installation

From PyPI:

pip install multimodal_feature_builder