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/

Zendy Plus

Presents the access of premium content as premium feature

Premium Content

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

Zendy Tools

Zendy Open

We present a novel sub-8-bit quantization-aware training (S8BQAT) scheme for8-bit neural network accelerators. Our method is inspired from Lloyd-Maxcompression theory with practical adaptations for a feasible computationaloverhead during training. With the quantization centroids derived from a 32-bitbaseline, we augment training loss with a Multi-Regional Absolute Cosine(MRACos) regularizer that aggregates weights towards their nearest centroid,effectively acting as a pseudo compressor. Additionally, a periodically invokedhard compressor is introduced to improve the convergence rate by emulatingruntime model weight quantization. We apply S8BQAT on speech recognition tasksusing Recurrent Neural NetworkTransducer (RNN-T) architecture. With S8BQAT, weare able to increase the model parameter size to reduce the word error rate by4-16% relatively, while still improving latency by 5%.

Sub-8-Bit Quantization Aware Training for 8-Bit Neural Network Accelerator with On-Device Speech Recognition