Finding the most similar product
Being able to compute similarity between embeddings is a key step within embeddings applications. In this exercise, you'll return to the products list of dictionaries that you worked with previously, which contains the embedded short descriptions you also created earlier.
You'll compare a piece of text to these embedded descriptions to identify the most similar description.
numpy has been imported as np, and distance is available from scipy.spatial. A create_embeddings() function has already been defined for you and is available to use for creating embeddings from a single input.
This exercise is part of the course
Introduction to Embeddings with the OpenAI API
Exercise instructions
- Embed the text,
"soap", using yourcreate_embeddings()custom function and extract a single list of embeddings. - Compute the cosine distance between the
query_embeddingand the embeddings inproduct. - Find and print the
'short_description'of the most similar product to the search text using the cosine distances indistances.
Hands-on interactive exercise
Have a go at this exercise by completing this sample code.
# Embed the search text
search_text = "soap"
search_embedding = ____
distances = []
for product in products:
# Compute the cosine distance for each product description
dist = ____(search_embedding, ____)
distances.append(dist)
# Find and print the most similar product short_description
min_dist_ind = ____
print(____)