Generate and use vector embeddings

from sentence_transformers import SentenceTransformer
from sklearn.metrics.pairwise import cosine_similarity

# Load a pre-trained model
model = SentenceTransformer('all-MiniLM-L6-v2')

# Generate embeddings for our documents
documents = [
    "The new EU AI regulations focus on transparency and accountability.",
    "Our trading algorithms use advanced machine learning techniques.",
    "Recent market volatility has impacted our APAC operations."
]
document_embeddings = model.encode(documents)

# Generate embedding for a query
query = "Impact of EU AI regulations on trading algorithms"
query_embedding = model.encode([query])[0]

# Find the most similar document
similarities = cosine_similarity([query_embedding], document_embeddings)[0]
most_similar_index = similarities.argmax()
print(f"Most relevant document: {documents[most_similar_index]}")