English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Tools annual

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

The large number of parameters of some prominent language models, such asBERT, makes their fine-tuning on downstream tasks computationally intensive andenergy hungry. Previously researchers were focused on lower bit-width integerdata types for the forward propagation of language models to save memory andcomputation. As for the backward propagation, however, only 16-bitfloating-point data type has been used for the fine-tuning of BERT. In thiswork, we use integer arithmetic for both forward and back propagation in thefine-tuning of BERT. We study the effects of varying the integer bit-width onthe model's metric performance. Our integer fine-tuning uses integer arithmeticto perform forward propagation and gradient computation of linear, layer-norm,and embedding layers of BERT. We fine-tune BERT using our integer trainingmethod on SQuAD v1.1 and SQuAD v2., and GLUE benchmark. We demonstrate thatmetric performance of fine-tuning 16-bit integer BERT matches both 16-bit and32-bit floating-point baselines. Furthermore, using the faster and more memoryefficient 8-bit integer data type, integer fine-tuning of BERT loses an averageof 3.1 points compared to the FP32 baseline.

Towards Fine-tuning Pre-trained Language Models with Integer Forward and Backward Propagation