I’m doing some testing combining the outputs of multiple small decoder models (each with their own parameters and such), and as can be expected, it’s a pain to initialize each model manually. Therefore, I was thinking of just looping this process and storing all the models in a dictionary like so:
num_small_decoders = 4
small_decoders = {}
for d in range(num_small_decoders):
model_name = "model" + str(d)
model = SmallDecoder()
model = model.to(device)
small_decoders[model_name] = model
Is this okay to do?