OpenVINO™ is an open-source toolkit for optimizing and deploying AI inference. The OpenVINO™ Runtime supports various hardware devices including x86 and ARM CPUs, and Intel GPUs. It can help to boost deep learning performance in Computer Vision, Automatic Speech Recognition, Natural Language Processing and other common tasks.
Hugging Face embedding model can be supported by OpenVINO through OpenVINOEmbeddings class. If you have an Intel GPU, you can specify model_kwargs={"device": "GPU"} to run inference on it.
%pip install --upgrade-strategy eager "optimum[openvino,nncf]" --quiet
Note: you may need to restart the kernel to use updated packages.
from langchain_community.embeddings import OpenVINOEmbeddings
model_name = "sentence-transformers/all-mpnet-base-v2"
model_kwargs = {"device": "CPU"}
encode_kwargs = {"mean_pooling": True, "normalize_embeddings": True}
ov_embeddings = OpenVINOEmbeddings(
model_name_or_path=model_name,
model_kwargs=model_kwargs,
encode_kwargs=encode_kwargs,
)
text = "This is a test document."
query_result = ov_embeddings.embed_query(text)
[-0.048951778560876846, -0.03986183926463127, -0.02156277745962143]
doc_result = ov_embeddings.embed_documents([text])
It is possible to export your embedding model to the OpenVINO IR format with OVModelForFeatureExtraction, and load the model from local folder.
from pathlib import Path
ov_model_dir = "all-mpnet-base-v2-ov"
if not Path(ov_model_dir).exists():
ov_embeddings.save_model(ov_model_dir)
ov_embeddings = OpenVINOEmbeddings(
model_name_or_path=ov_model_dir,
model_kwargs=model_kwargs,
encode_kwargs=encode_kwargs,
)
Compiling the model to CPU ...
We can also access BGE embedding models via the OpenVINOBgeEmbeddings class with OpenVINO.
from langchain_community.embeddings import OpenVINOBgeEmbeddings
model_name = "BAAI/bge-small-en"
model_kwargs = {"device": "CPU"}
encode_kwargs = {"normalize_embeddings": True}
ov_embeddings = OpenVINOBgeEmbeddings(
model_name_or_path=model_name,
model_kwargs=model_kwargs,
encode_kwargs=encode_kwargs,
)
embedding = ov_embeddings.embed_query("hi this is harrison")
len(embedding)
For more information refer to:
RetroSearch is an open source project built by @garambo | Open a GitHub Issue
Search and Browse the WWW like it's 1997 | Search results from DuckDuckGo
HTML:
3.2
| Encoding:
UTF-8
| Version:
0.7.4