Using Final Models

Right now FELToken supports only scikit-learn models. Models are exported using joblib library. That means that you need correct version of the joblib and scikit-learn libraries. You can find these in felt package requirements.
joblib==1.1.0
numpy==1.21.0
scikit-learn==1.0.1
If you created a plan, you can watch it in the project dashboard. Once the plan is finished, you should see a download button next to it (this is only available to builder who created the training plan). By clicking the DOWNLOAD button you will obtain the model. The model is encrypted. Before the download starts, you will see a MetaMask pop-up asking you to decrypt the data (the model). After approving the decryption, you should receive model.joblib file. In your Python code, you can use this file as follows:
import numpy as np
import joblib
​
# Load the model, use the correct path to the model.joblib file
model = joblib.load("model.joblib")
​
print("Model details:", model.__dict__)
​
# X should contain the data you want to use for prediction
X = np.array([...])
result = model.predict(X)

Guides - Previous

Builder (Data Scientist)

Next - Guides

Creating Custom Models

Last modified 4mo ago

Copy link