The article says bag-of-words plus logistic regression remains a cheap baseline and cites 89.9% accuracy on a balanced IMDb dataset, compared with 85.66% for an LSTM in the examples discussed. Raschka characterizes Jev as potentially more general than task-specific classifiers and faster and cheaper than large general-purpose models for classification, but labels his account of Jev’s methodology an educated guess. He says he is unaffiliated and does not present the article as a product endorsement. https://magazine.sebastianraschka.com/p/classifier-history-and-jev
Raw SignalSep 29, 20261 min read1 source
Language Models for Text Classification: From Bag-of-Words to Jev
In one minute
Sebastian Raschka published a visual guide to language-model text classification, tracing approaches from bag-of-words classifiers through recurrent and convolutional neural networks to transformers, and using that history to frame Jev.