train-fasttext

What problem does it solve?

This skill addresses the challenge of training FastText text classification models efficiently, especially when needing to balance model accuracy with strict file size constraints.

Core Features & Use Cases

• Hyperparameter Tuning: Guides systematic exploration of parameters like dim, epoch, lr, and wordNgrams to improve accuracy.
• Size Optimization: Provides strategies to reduce model size by adjusting minCount, bucket, and dim, or by using quantization.
• Use Case: You need to deploy a text classifier on a mobile device with limited storage. This skill helps you train a FastText model that meets both a minimum accuracy requirement and a maximum file size limit.

Quick Start

Use the train-fasttext skill to train a supervised FastText model on 'train.txt' aiming for high accuracy and a small model size.