Search Results for author:
Found 1 papers, 0 papers with code
Controlled Randomness Improves the Performance of Transformer Models
During the pre-training step of natural language models, the main objective is to learn a general representation of the pre-training dataset, usually requiring large amounts of textual data to capture the complexity and diversity of natural language.
Cannot find the paper you are looking for? You can
Submit a new open access paper.
