tffm

Supports factorization machines of order 2 and higher, enabling capture of complex feature interactions, as demonstrated in the usage example with order=6.

Handles both input types, making it versatile for real-world datasets with many categorical features, as highlighted in the key features and usage with input_type parameters.

Integrates with various TensorFlow optimizers like AdamOptimizer, allowing fine-tuning and TensorBoard logging for monitoring, as shown in the dependencies and example code.

Provides a familiar API similar to scikit-learn models, easing adoption for users in that ecosystem, as noted in the usage section with fit and predict methods.