Can I read Scalable Transformers for Recommendation on EtoBox?

Scalable Transformers for Recommendation by bugra is a document available to read on EtoBox.

What is Scalable Transformers for Recommendation about?

The document presents a framework called GPSD (Generative Pretraining for Scalable Discriminative Recommendation) to address overfitting issues in discriminative recommendation tasks like CTR and CVR prediction. By leveraging generative pretraining and a sparse parameter freezing strategy, GPSD enhances model scalability and performance, demonstrating significant improvements in both industrial and public datasets. The findings suggest that as model parameters scale, performance consistently improves, align

Author

bugra

Language

EN