ollama embeddings cosine similarity and dot product example

ollama-emb-cosine-dot.py

# pip install scikit-learn numpy ollama
import ollama
import numpy as np
from sklearn.metrics.pairwise import cosine_similarity

text1 = ollama.embeddings(model='nomic-embed-text', prompt='The sky is blue because of rayleigh scattering')
text2 = ollama.embeddings(model='nomic-embed-text', prompt='The sky is cloudy and grey today')

vec1 = np.array(text1['embedding']).reshape(1, -1)
vec2 = np.array(text2['embedding']).reshape(1, -1)

similarity = cosine_similarity(vec1, vec2)
print(similarity)


# same thing but with out needing scikit-learn
vec1 = np.array(text1['embedding'])
vec2 = np.array(text2['embedding'])

dot_product = np.dot(vec1, vec2)
norm_vec1 = np.linalg.norm(vec1)
norm_vec2 = np.linalg.norm(vec2)

cosine_similarity = dot_product / (norm_vec1 * norm_vec2)

print(cosine_similarity)