It's time to make a decision

What will a project manager's decision be? We have seen the limits of the original Transformer model, which leads to the crossroads where we have to choose a path to:

• Accept the limits of the original Transformer model and move on to huge models requiring huge machine memory and computing power.
• To refuse the limits of the original Transformer and tweak its architecture with reformer-type approaches.
• Use different training methods such as PET, an efficient knowledge distillation approach.
• Use a combination of these approaches.
• Design your own training methods and model architecture.

  There are many transformer model methods continuously appearing on the market. Take the necessary time to find the right path for your project.

In real-life project management, each approach will be carefully evaluated using standard evaluation parameters:

• The cost of each solution
...