Standardize the deployment of compact transformer models for edge applications. Focus on quantization and distillation to make these models runnable on low-power devices.
Suggested repo: nano-bert
"The classic BERT transformer, now stripped down for edge deployment."
Estimated effort: 20h