Skip to content

Embedding

Embedding is a way to represent data as a vector of numbers, so that a computer can compare and process it. LLMfy can use embedding.

Providers

AWS Bedrock

See available embedding models: AWS Bedrock Embedding Models

embedding = BedrockEmbedding(model="amazon.titan-embed-text-v1")

OpenAI

See available embedding models: OpenAI Embedding Models

embedding = OpenAIEmbedding(model="text-embedding-3-small")

Google AI

See available embedding models: Google AI Embedding Models

embedding = GoogleAIEmbedding(model="gemini-embedding-001")

Define Embedding

from llmfy import BedrockEmbedding, OpenAIEmbedding, GoogleAIEmbedding

Encode

Encode single text.

4
5
text = "The cat sits on the mat"
result = embedding.encode(text=text)

Encode with Batch

Encode texts into embedding with batch process.

 5
 6
 7
 8
 9
10
11
12
13
14
15
16
texts = [
    "The cat sits on the mat",
    "Dogs are loyal animals",
    "Artificial intelligence is transforming the world",
    "Quantum computing is the future of technology",
    "The sun rises in the east",
]

embeddings = embedding.encode_batch(
    texts,
    batch_size=3,
)

Usage

To view usage embedding you can use llmfy_usage_tracker here. Usually embedding use input token only for usage the output token not count.

Note: Google AI embedding usage tracks input tokens via a separate count_tokens call, since the embedding API response does not include token usage metadata.