| import os | |
| import logging | |
| from langchain_huggingface import HuggingFaceEmbeddings | |
| from langchain_community.vectorstores import FAISS | |
| def embed_documents(documents, embedding_path="embeddings.faiss"): | |
| embedding_model = HuggingFaceEmbeddings(model_name="sentence-transformers/paraphrase-MiniLM-L3-v2") | |
| if os.path.exists(embedding_path): | |
| logging.info("Loading embeddings from local file") | |
| vector_store = FAISS.load_local(embedding_path, embedding_model, allow_dangerous_deserialization=True) | |
| else: | |
| logging.info("Generating and saving embeddings") | |
| vector_store = FAISS.from_texts([doc['text'] for doc in documents], embedding_model) | |
| vector_store.save_local(embedding_path) | |
| return vector_store | |