Hello experts,
Is there a way to use Pytorch in developing strategies using MoonshotML? Similarly, is there a well-documented runbook to follow if we want to integrate a custom ML framework?
Thanks in advance for your help!
MoonshotML is looking for a fit() method on the model during training (or partial_fit() or train_on_batch() for incremental learning) and is looking for a predict() method during testing. In principle, it should be possible to use a custom model by adding the expected methods to the model before serializing it.
model = MyCustomModel(...)
def fit(self, features, targets):
...
def predict(self, features):
...
model.fit = fit
model.predict = predict
joblib.dump(model, "custom_model.joblib")
Installing custom packaged is covered in the usage guide.