Instructions to use nikraf/directionality_probe with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use nikraf/directionality_probe with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("feature-extraction", model="nikraf/directionality_probe", trust_remote_code=True)# Load model directly from transformers import AutoModel model = AutoModel.from_pretrained("nikraf/directionality_probe", trust_remote_code=True, device_map="auto") - Notebooks
- Google Colab
- Kaggle
| import os | |
| import matplotlib.pyplot as plt | |
| import seaborn as sns | |
| import numpy as np | |
| from scipy.stats import spearmanr, pearsonr | |
| from sklearn.metrics import r2_score | |
| from visualization.pauc_plot import plot_roc_with_ci | |
| def regression_ci_plot(y_true, y_pred, save_path, title=None): | |
| """ | |
| Calculate the spearman rho and p-value of the regression model. | |
| Plot the line of best fit with 95% confidence intervals for spearman rho. | |
| Display the R-squared value, spearman rho, pearson rho, and p-values. | |
| """ | |
| # Compute R‑squared, Spearman and Pearson correlations | |
| y_true, y_pred = y_true.flatten(), y_pred.flatten() | |
| mask = np.isfinite(y_true) & np.isfinite(y_pred) | |
| y_true, y_pred = y_true[mask], y_pred[mask] | |
| r2 = r2_score(y_true, y_pred) | |
| r_s, p_s = spearmanr(y_true, y_pred) | |
| r_p, p_p = pearsonr(y_true, y_pred) | |
| # Create scatter plot and regression line with 95% CI | |
| fig, ax = plt.subplots(figsize=(8, 6)) | |
| sns.scatterplot(x=y_true, y=y_pred, ax=ax) | |
| sns.regplot( | |
| x=y_true, y=y_pred, | |
| ci=95, ax=ax, scatter=False, | |
| line_kws={'color': 'red'} | |
| ) | |
| ax.set_xlabel('True Values') | |
| ax.set_ylabel('Predicted Values') | |
| if title: | |
| ax.set_title(title) | |
| else: | |
| ax.set_title('Regression Plot with 95% Confidence Interval') | |
| # Annotate statistics on the plot | |
| stats_text = ( | |
| f"$R^2$ = {r2:.2f}\n" | |
| f"Spearman $\\rho$ = {r_s:.2f} (p = {p_s:.2e})\n" | |
| f"Pearson $\\rho$ = {r_p:.2f} (p = {p_p:.2e})" | |
| ) | |
| ax.text( | |
| 0.05, 0.95, stats_text, | |
| transform=ax.transAxes, | |
| fontsize=12, verticalalignment='top' | |
| ) | |
| # Save the figure | |
| fig.savefig(save_path, dpi=300, bbox_inches='tight') | |
| plt.close(fig) | |
| def classification_ci_plot(y_true, y_pred, save_path, title=None): | |
| """ | |
| Use pauc to display classification plot | |
| """ | |
| if len(y_pred.shape) == 3 and len(y_true.shape) == 2: | |
| y_pred = y_pred.reshape(-1, y_pred.shape[-1]) | |
| y_true = y_true.reshape(-1) | |
| ### Note: removing this gives you one plot per multilabel class | |
| if len(y_pred.shape) == 2 and len(y_true.shape) == 2: | |
| y_pred = y_pred.flatten() | |
| y_true = y_true.flatten() | |
| # if more than 10,000 data points, only pass 10,000 | |
| # else, pAUC can be very slow | |
| if y_true.shape[0] > 10000: | |
| y_pred = y_pred[:10000] | |
| y_true = y_true[:10000] | |
| print(y_true.shape, y_pred.shape) | |
| try: | |
| plot_roc_with_ci(y_true, y_pred, save_path, fig_title=title) | |
| except Exception as e: | |
| print(f"Error plotting pAUC curve, likely the wrong version: {e}") | |
| if __name__ == "__main__": | |
| # py -m visualization.ci_plots | |
| import os | |
| os.makedirs("plots/test_plots", exist_ok=True) | |
| y_true = np.random.rand(100) | |
| y_pred = np.random.rand(100) | |
| regression_ci_plot(y_true, y_pred, "plots/test_plots/regression.png", title="Regression Plot") | |
| y_true = np.random.randint(0, 2, (50, 514)) | |
| y_pred = np.random.rand(50, 514, 4) | |
| classification_ci_plot(y_true, y_pred, "plots/test_plots/classification.png", title="Classification Plot") | |