Classes for Fast Maximum Entropy Training
Joshua Goodman
Abstract
Maximum entropy models are considered by many to be one of the most promising avenues of language modeling research. Unfortunately, long training times make maximum entropy research difficult. We present a novel speedup technique: we change the form of the model to use classes. Our speedup works by creating two maximum entropy models, the first of which predicts the class of each word, and the second of which predicts the word itself. This factoring of the model leads to fewer non-zero indicator functions, and faster normalization, achieving speedups of up to a factor of 35 over one of the best previous techniques. It also results in typically slightly lower perplexities. The same trick can be used to speed training of other machine learning techniques, e.g. neural networks, applied to any problem with a large number of outputs, such as language modeling.
Create a lesson
Related papers
RSI-Anything: Unlocking Recursive Self-Improvement with World Models
Taoyong Cui, Pinxian Zeng, Wanghan Xu et al.
When VLMs Destroy What They Know: Spatial Representations Fall Below Lexical Priors
Shubh Garg, Chen Shani
Structural Automata for Neural Semantic Parsing
Xiao Zhang
Task Context Shapes Component Roles in Language Models
Sovesh Mohapatra, Dani S Bassett
Can Language Models Learn When to Abstain? Policy Learning and Calibrated Answer Selection
Juan Terven, Diana-Margarita Córdova-Esparza, Ruben Gomez et al.
Error Rate is Not Enough: Costs of Augmentation in Low-Resource ASR
Katsumi Ibaraki, David Chiang