We read every piece of feedback, and take your input very seriously.
To see all available qualifiers, see our documentation.
Have a question about this project? # for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “#”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? # to your account
In the section 'Machine Learning' in the 'Foundations' course, there is a code block in the 'Inference' sub-section:
# Unstandardize predictions pred_infer = model(X_infer).detach().numpy() * np.sqrt(y_scaler.var_) + y_scaler.mean_ for i, index in enumerate(sample_indices): print (f"{df.iloc[index]["y"]:.2f} (actual) → {pred_infer[i][0]:.2f} (predicted)")
However since there are also double-quotes around the y in the indexing (intended to display as "y"), the f-string ends early.
A fix would include simply changing the double quotes around the "y" to single quotes 'y':
# Unstandardize predictions pred_infer = model(X_infer).detach().numpy() * np.sqrt(y_scaler.var_) + y_scaler.mean_ for i, index in enumerate(sample_indices): print (f"{df.iloc[index]['y']:.2f} (actual) → {pred_infer[i][0]:.2f} (predicted)")
The text was updated successfully, but these errors were encountered:
No branches or pull requests
In the section 'Machine Learning' in the 'Foundations' course, there is a code block in the 'Inference' sub-section:
# Unstandardize predictions pred_infer = model(X_infer).detach().numpy() * np.sqrt(y_scaler.var_) + y_scaler.mean_ for i, index in enumerate(sample_indices): print (f"{df.iloc[index]["y"]:.2f} (actual) → {pred_infer[i][0]:.2f} (predicted)")
However since there are also double-quotes around the y in the indexing (intended to display as "y"), the f-string ends early.
A fix would include simply changing the double quotes around the "y" to single quotes 'y':
# Unstandardize predictions pred_infer = model(X_infer).detach().numpy() * np.sqrt(y_scaler.var_) + y_scaler.mean_ for i, index in enumerate(sample_indices): print (f"{df.iloc[index]['y']:.2f} (actual) → {pred_infer[i][0]:.2f} (predicted)")
The text was updated successfully, but these errors were encountered: