Learn Before
Short Answer

Designing a Self-Supervised Task for Code

You are tasked with pre-training a language model on a massive, unlabeled dataset of computer code. Beyond the common approach of predicting randomly masked parts of the code, propose one distinct self-supervised objective that would be particularly well-suited for this dataset. Briefly justify why your proposed objective would help the model learn meaningful patterns specific to code.

0

1

Updated 2025-10-07

Contributors are:

Who are from:

Tags

Ch.1 Pre-training - Foundations of Large Language Models

Foundations of Large Language Models

Foundations of Large Language Models Course

Computing Sciences

Creation in Bloom's Taxonomy

Cognitive Psychology

Psychology

Social Science

Empirical Science

Science