Cosine Similarity > 1 in dlib face recognition

Testing face recognition using dlib in VS Code. In this code, Treating the faces as the same if the Euclidean distance is less than 0.6,

I've written the following code to get the Cosine Similarity here, and it gives me a Cosine Similarity of more than 1.

I think it's wrong, so how can I get the correct Cosine Similarity?

fig, ax = plt.subplots(1, figsize=(20, 20))
ax.imshow(img_rgb)

for i, desc in enumerate(descriptors):
    
    found = False
    for name, saved_desc in descs.items():            
        dist = np.linalg.norm([desc] - saved_desc, axis=1)
        
        # my code : compute cosine similarity        
        cosine = np.dot([desc],saved_desc)/(norm([desc]),norm(saved_desc))                
      
        if dist < 0.6:
            found = True

            text = ax.text(rects[i][0][0], rects[i][0][1], name,
                    color='b', fontsize=40, fontweight='bold')
            text2 = ax.text(rects[i][1][0], rects[i][1][1], cosine,
                    color='b', fontsize=20, fontweight='bold')
            text.set_path_effects([path_effects.Stroke(linewidth=10, foreground='white'), path_effects.Normal()])
            rect = patches.Rectangle(rects[i][0],
                                 rects[i][1][1] - rects[i][0][1],
                                 rects[i][1][0] - rects[i][0][0],
                                 linewidth=2, edgecolor='w', facecolor='none')
            ax.add_patch(rect)

            break    
    

neo
[1.35159114 1.36754706]
neo
[1.15541134 1.10536481]
trinity
[1.13942505 1.03432384]
morpheus
[1.26643039 1.28631887]
neo
[1.28240694 1.31706417]
trinity
[1.4723583  1.43481646]

Cosine Similarity less than 1

The denominator of the cosine similarity is incorrect. You should replace (norm([desc]),norm(saved_desc)) with (norm([desc])*norm(saved_desc)).