TypeError: list indices must be integers or slices, not tuple Error?

Zama Concrete ML
1

Hello,

I am trying to implement an XGBoostClassifier on some embedded text data.
here is my code.


# my train and test data are numpy arrays with the shapes:
x_train has shape: (53783, 768)
y_train has shape: (53783,)
x_test has shape: (13446, 768)
y_test has shape: (13446,)

from concrete.ml.sklearn import XGBClassifier as ConcreteXGBClassifier

# model
enc_model = ConcreteXGBClassifier( max_depth = 5, n_estimators=120 , n_bits=6, random_state=15).fit(X_train, y_train)

# A circuit needs to be compiled to enable FHE execution
circuit = enc_model.compile(X_train)

y_pred_enc = enc_model.predict(X_test, fhe="execute")

# this line gives an error, 
----> 5 y_pred_enc = enc_model.predict(X_test, fhe="execute")

TypeError: predict() got an unexpected keyword argument 'fhe'

So I am doing it the lengthy way:

# Compile the model on a representative set
fhe_circuit = enc_model.compile(X_train)

# Generate the keys (set force to True in order to generate new keys at each execution)
fhe_circuit.keygen(force=True)

y_pred_fhe_step = []

for f_input in X_test:
    # Quantize an input (float)
    q_input = enc_model.quantize_input([f_input])
    
    # Encrypt the input
    q_input_enc = fhe_circuit.encrypt(q_input)

    # Execute the linear product in FHE 
    q_y_enc = fhe_circuit.run(q_input_enc)

    # Decrypt the result (integer)
    q_y = fhe_circuit.decrypt(q_y_enc)

    # De-quantize the result
    y = enc_model.dequantize_output(q_y)

    # Apply either the sigmoid if it is a binary classification task, which is the case in this 
    # example, or a softmax function in order to get the probabilities (in the clear)
    y_proba = enc_model.post_processing(y)

    # Since this model does classification, apply the argmax to get the class predictions (in the clear)
    # Note that regression models won't need the following line
    y_class = np.argmax(y_proba, axis=1)

    y_pred_fhe_step += list(y_class)

y_pred_fhe_step = np.array(y_pred_fhe_step)

print("Predictions in clear:", y_pred)
print("Predictions in FHE  :", y_pred_fhe_step)
print(f"Similarity: {int((y_pred_fhe_step == y_pred).mean()*100)}%")

---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
<ipython-input-21-5ac661ca2f61> in <module>
     11 for f_input in X_test:
     12     # Quantize an input (float)
---> 13     q_input = enc_model.quantize_input([f_input])
     14 
     15     # Encrypt the input

/opt/conda/lib/python3.7/site-packages/concrete/ml/sklearn/base.py in quantize_input(self, X)
    606         # Quantize using the learned quantization parameters for each feature
    607         for i, q_x_ in enumerate(self.input_quantizers):
--> 608             qX[:, i] = q_x_.quant(X[:, i])
    609         return qX.astype(numpy.int64)
    610 

TypeError: list indices must be integers or slices, not tuple

Why am I getting this error?

2

Hello @bluetail14 , sorry for the late answer, most of our team was off recently !

Could you indicate which Concrete-ML version you are using (through pip show concrete-ml for example) ? It looks like you are trying to run an example for CML 1.0 (or above) with an older version (0.6 or below). If this is the case, please upgrade Concrete ML to its latest version !

Thanks !

1 Like
3

yes, you’re right - I was using the old version. v 1.1.0 fixed it.

1 Like