Instructor embeddings on Hugging Face integration

The hkunlp/instructor-* family introduced instruction-tuned sentence embeddings. They have been broadly superseded by modern instruction-aware models, but the original models are still available on Hugging Face and usable via the legacy HuggingFaceInstructEmbeddings class.

For new projects, prefer HuggingFaceEmbeddings from langchain-huggingface with a current instruction-aware model such as intfloat/e5-large-v2, Qwen/Qwen3-Embedding-0.6B, or BAAI/bge-m3. Pass query and document prompts via encode_kwargs and query_encode_kwargs:

from langchain_huggingface import HuggingFaceEmbeddings

embeddings = HuggingFaceEmbeddings(
    model_name="intfloat/e5-large-v2",
    encode_kwargs={"prompt": "passage: "},
    query_encode_kwargs={"prompt": "query: "},
)

Legacy Instructor usage

pip install -qU langchain-community InstructorEmbedding sentence-transformers

Warning

The langchain-community package is no longer maintained. Examples that import from langchain_community may be outdated or broken. Use with caution.

from langchain_community.embeddings import HuggingFaceInstructEmbeddings

embeddings = HuggingFaceInstructEmbeddings(
    query_instruction="Represent the query for retrieval: "
)

query_result = embeddings.embed_query("This is a test document.")