If you want to 10x your FPS from an existing model, and learn how to build models that are insanely efficient and ready for production, than read on:
The majority of AI Engineers spend their time working with models that have no chance of ever going to production:
- They're too slow
- They're unoptimized for the hardware
- They don't generalize well
This is why I spent some time working on 2 courses that will help you make models much more efficient and ready for production.
5 Techniques to 10x Your FPS
To be more specific, I found 5 techniques that have the potential to 10x the FPS of your existing model with little work:
- Optimizing the Convolutions
- Building HydraNets (multi-task learning models)
-
Pruning your Model
-
Quantizing your Weights
-
Implementing Knowledge Distillation
After learning these techniques, clients from my courses had the ability to improve their models by up to 1,268%, and understand how to make their models as optimized as possible.
Here are the two courses: