Ollama

Let's load the Ollama Embeddings class.

from langchain_community.embeddings import OllamaEmbeddings

API Reference:OllamaEmbeddings

embeddings = OllamaEmbeddings()

text = "This is a test document."

To generate embeddings, you can either query an invidivual text, or you can query a list of texts.

query_result = embeddings.embed_query(text)
query_result[:5]

[-0.09996652603149414,
 0.015568195842206478,
 0.17670190334320068,
 0.16521021723747253,
 0.21193109452724457]

doc_result = embeddings.embed_documents([text])
doc_result[0][:5]

[-0.04242777079343796,
 0.016536075621843338,
 0.10052520781755447,
 0.18272875249385834,
 0.2079043835401535]

Let's load the Ollama Embeddings class with smaller model (e.g. llama:7b). Note: See other supported models https://ollama.ai/library

embeddings = OllamaEmbeddings(model="llama2:7b")

text = "This is a test document."

query_result = embeddings.embed_query(text)

query_result[:5]

[-0.09996627271175385,
 0.015567859634757042,
 0.17670205235481262,
 0.16521376371383667,
 0.21193283796310425]

doc_result = embeddings.embed_documents([text])

doc_result[0][:5]

[-0.042427532374858856,
 0.01653730869293213,
 0.10052604228258133,
 0.18272635340690613,
 0.20790338516235352]