Sampler with dynamic parameters

How to define custom dynamic parameters for your sampler ?

In this AL benchmark, there is a possibility for you to define custom dynamic parameters that will be used to instanciate your sampler before each time it is called.

How to define custom dynamic parameters ?

There are already a few dynamic parameters that have been implemented in the benchmark. You will find them inside the get_my_sampler sampler generator function as defaut parameters for MyCustomSamplerClass.

If you need to implement new ones, you will have to implement them by yourself in the main_run.py script.

Where do we register our custom dynamic parameters ?

Update dynamic params dictionary

In order for the benchmark to take your dynamic parameters into account, you must define them in the dynamic parameters dictionary DYNAMIC_PARAMS (inside the run_benchmark function from the main_run.py script)

DYNAMIC_PARAMS = dict(name_dynamic_parameter = value_dynamic_parameter)

Reminder

Don’t forget to register these dynamic parameters inside the get_my_sampler sampler generator function.

def get_my_sampler(params : dict) :

    sampler = MyCustomSamplerClass(
        name_dynamic_parameter = params['name_dynamic_parameter']
        ... # other static and/or dynamic parameters
    )