fcharton

23 ms·

1. 1.▲
   by fcharton 10mo ago

   Author, here. The paper is about the Collatz sequence, how experiments with a transformer can point at interesting facts about a complex mathematical phenomenon, and how, in supervised math transformers, model predictions and errors can be
2. 2.▲
   by fcharton 7y ago

   For training, you need a generator because you want millions of solved examples for deep learning to work. At test time, you usually want a test set from the same distribution as the training data (or at least related to it in some controll
3. 3.▲
   by fcharton 7y ago

   Textbook problems are usually short, with short solutions, and demonstrating one specific rule. They are better handled by classical (rule-based) tools. Deep learning tools would either memorize them or resort to a rule based sub-module. F
4. 4.▲
   by fcharton 7y ago

   Since the paper was presented, it was reviewed in ICLR and significantly expanded, and many questions raised in September were addressed. The performance metric is the number of equations/integrals correctly solved over a held out test