Skip to main content
Open In ColabOpen on GitHub

PredictionGuardEmbeddings

Prediction Guard is a secure, scalable GenAI platform that safeguards sensitive data, prevents common AI malfunctions, and runs on affordable hardware.

Overview​

Integration details​

This integration shows how to use the Prediction Guard embeddings integration with Langchain. This integration supports text and images, separately or together in matched pairs.

Setup​

To access Prediction Guard models, contact us here to get a Prediction Guard API key and get started.

Credentials​

Once you have a key, you can set it with

import os

os.environ["PREDICTIONGUARD_API_KEY"] = "<Prediction Guard API Key"

Installation​

%pip install --upgrade --quiet langchain-predictionguard

Instantiation​

First, install the Prediction Guard and LangChain packages. Then, set the required env vars and set up package imports.

from langchain_predictionguard import PredictionGuardEmbeddings
embeddings = PredictionGuardEmbeddings(model="bridgetower-large-itm-mlm-itc")

Prediction Guard embeddings generation supports both text and images. This integration includes that support spread across various functions.

Indexing and Retrieval​

# Create a vector store with a sample text
from langchain_core.vectorstores import InMemoryVectorStore

text = "LangChain is the framework for building context-aware reasoning applications."

vectorstore = InMemoryVectorStore.from_texts(
    [text],
    embedding=embeddings,
)

# Use the vectorstore as a retriever
retriever = vectorstore.as_retriever()

# Retrieve the most similar text
retrieved_documents = retriever.invoke("What is LangChain?")

# Show the retrieved document's content
retrieved_documents[0].page_content
API Reference:InMemoryVectorStore
'LangChain is the framework for building context-aware reasoning applications.'

Direct Usage​

The vectorstore and retriever implementations are calling embeddings.embed_documents(...) and embeddings.embed_query(...) to create embeddings from the texts used in the from_texts and retrieval invoke operations.

These methods can be directly called with the following commands.

Embed single texts​

# Embedding a single string
text = "This is an embedding example."
single_vector = embeddings.embed_query(text)

single_vector[:5]
[0.01456777285784483,
 -0.08131945133209229,
 -0.013045587576925755,
 -0.09488929063081741,
 -0.003087474964559078]

Embed multiple texts​

# Embedding multiple strings
docs = [
    "This is an embedding example.",
    "This is another embedding example.",
]

two_vectors = embeddings.embed_documents(docs)

for vector in two_vectors:
    print(vector[:5])
[0.01456777285784483, -0.08131945133209229, -0.013045587576925755, -0.09488929063081741, -0.003087474964559078]
[-0.0015021917643025517, -0.08883760124444962, -0.0025286630261689425, -0.1052245944738388, 0.014225339516997337]

Embed single images​

# Embedding a single image. These functions accept image URLs, image files, data URIs, and base64 encoded strings.
image = [
    "https://farm4.staticflickr.com/3300/3497460990_11dfb95dd1_z.jpg",
]
single_vector = embeddings.embed_images(image)

print(single_vector[0][:5])
[0.0911610797047615, -0.034427884966135025, 0.007927080616354942, -0.03500846028327942, 0.022317267954349518]

Embed multiple images​

# Embedding multiple images
images = [
    "https://fastly.picsum.photos/id/866/200/300.jpg?hmac=rcadCENKh4rD6MAp6V_ma-AyWv641M4iiOpe1RyFHeI",
    "https://farm4.staticflickr.com/3300/3497460990_11dfb95dd1_z.jpg",
]

two_vectors = embeddings.embed_images(images)

for vector in two_vectors:
    print(vector[:5])
[0.1593627631664276, -0.03636132553219795, -0.013229663483798504, -0.08789524435997009, 0.062290553003549576]
[0.0911610797047615, -0.034427884966135025, 0.007927080616354942, -0.03500846028327942, 0.022317267954349518]

Embed single text-image pairs​

# Embedding a single text-image pair
inputs = [
    {
        "text": "This is an embedding example.",
        "image": "https://farm4.staticflickr.com/3300/3497460990_11dfb95dd1_z.jpg",
    },
]
single_vector = embeddings.embed_image_text(inputs)

print(single_vector[0][:5])
[0.0363212488591671, -0.10172265768051147, -0.014760786667466164, -0.046511903405189514, 0.03860781341791153]

Embed multiple text-image pairs​

# Embedding multiple text-image pairs
inputs = [
    {
        "text": "This is an embedding example.",
        "image": "https://fastly.picsum.photos/id/866/200/300.jpg?hmac=rcadCENKh4rD6MAp6V_ma-AyWv641M4iiOpe1RyFHeI",
    },
    {
        "text": "This is another embedding example.",
        "image": "https://farm4.staticflickr.com/3300/3497460990_11dfb95dd1_z.jpg",
    },
]
two_vectors = embeddings.embed_image_text(inputs)

for vector in two_vectors:
    print(vector[:5])
[0.11867266893386841, -0.05898813530802727, -0.026179173961281776, -0.10747235268354416, 0.07684746384620667]
[0.026654226705431938, -0.10080841928720474, -0.012732953764498234, -0.04365091398358345, 0.036743905395269394]

API Reference​

For detailed documentation of all PredictionGuardEmbeddings features and configurations check out the API reference: https://python.langchain.com/api_reference/community/embeddings/langchain_community.embeddings.predictionguard.PredictionGuardEmbeddings.html

Was this page helpful?