Rishika Agarwal


2023

pdf
EELBERT: Tiny Models through Dynamic Embeddings
Gabrielle Cohn | Rishika Agarwal | Deepanshu Gupta | Siddharth Patwardhan
Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing: Industry Track

We introduce EELBERT, an approach for compression of transformer-based models (e.g., BERT), with minimal impact on the accuracy of downstream tasks. This is achieved by replacing the input embedding layer of the model with dynamic, i.e. on-the-fly, embedding computations. Since the input embedding layer occupies a large portion of the model size, especially for the smaller BERT variants, replacing this layer with an embedding computation function helps us reduce the model size significantly. Empirical evaluation on the GLUE benchmark shows that our BERT variants (EELBERT) suffer minimal regression compared to the traditional BERT models. Through this approach, we are able to develop our smallest model UNO-EELBERT, which achieves a GLUE score within 4% of fully trained BERT-tiny, while being 15x smaller (1.2 MB) in size.