There is no explanation of how we can use different models to develop real-world systems with efficient RAM and GPU usage.
Suggest a potential alternative/fix
Explain how to use the existing code or convert the model to other formats for better inference performances. Some tools are in the package, but they are not working for all models, like ONNX conversion for the SlowFast model.
The doc issue
There is no explanation of how we can use different models to develop real-world systems with efficient RAM and GPU usage.
Suggest a potential alternative/fix
Explain how to use the existing code or convert the model to other formats for better inference performances. Some tools are in the package, but they are not working for all models, like ONNX conversion for the SlowFast model.