Embeddings (LLMEmbed)

LLMEmbed generates vector embeddings from text, useful for semantic search, clustering, and RAG applications.

Basic Usage

from microdc import Client, LLMEmbed

client = Client(api_key="mDC_...")

job = LLMEmbed(model="text-embedding-ada-002")
job.add_texts(["Hello world", "Goodbye world"])

job_id = client.send_job(job)
client.wait_for_all()

result = client.get_job_details(job_id)
embeddings = result.result['embeddings']

for i, embedding in enumerate(embeddings):
    print(f"Text {i}: {len(embedding)} dimensions")

Adding Texts

job = LLMEmbed(model="text-embedding-ada-002")

# Add a single text
job.add_text("Hello world")

# Add multiple texts at once
job.add_texts([
    "The quick brown fox",
    "A journey of a thousand miles",
    "To be or not to be"
])

Configuration Options

job = LLMEmbed(
    model="text-embedding-ada-002",  # Required: model name
    dimensions=None,                  # Embedding dimensions
    normalize=True,                   # Normalize embeddings
    encoding_format="float"           # Output format: "float" or "base64"
)

Parameter	Type	Default	Description
model	str	(required)	Embedding model name
dimensions	int	None	Output embedding dimensions
normalize	bool	True	Normalize output vectors
encoding_format	str	"float"	"float" or "base64"

Batch Embeddings

For large batches of texts:

from microdc import Client, LLMEmbed

client = Client(api_key="mDC_...")

# Large batch of texts
texts = [f"Document {i}" for i in range(100)]

# Split into chunks for parallel processing
chunk_size = 20
job_ids = []

for i in range(0, len(texts), chunk_size):
    chunk = texts[i:i + chunk_size]
    job = LLMEmbed(model="text-embedding-ada-002")
    job.add_texts(chunk)
    job.metadata = {"chunk_index": i // chunk_size}
    job_ids.append(client.send_job(job))

client.wait_for_all()

# Collect all embeddings
all_embeddings = []
for job_id in job_ids:
    details = client.get_job_details(job_id)
    if details.is_successful():
        all_embeddings.extend(details.result['embeddings'])
        client.acknowledge_job(job_id)

print(f"Generated {len(all_embeddings)} embeddings")

Use Case: Semantic Search

import numpy as np
from microdc import Client, LLMEmbed

client = Client(api_key="mDC_...")

# Embed documents
documents = [
    "Python is a programming language",
    "Machine learning uses statistical methods",
    "Neural networks are inspired by the brain"
]

doc_job = LLMEmbed(model="text-embedding-ada-002")
doc_job.add_texts(documents)
doc_id = client.send_job(doc_job)

# Embed query
query_job = LLMEmbed(model="text-embedding-ada-002")
query_job.add_texts(["What is ML?"])
query_id = client.send_job(query_job)

client.wait_for_all()

# Calculate similarities
doc_result = client.get_job_details(doc_id)
query_result = client.get_job_details(query_id)

doc_embeddings = np.array(doc_result.result['embeddings'])
query_embedding = np.array(query_result.result['embeddings'][0])

similarities = np.dot(doc_embeddings, query_embedding)
best_match = documents[np.argmax(similarities)]
print(f"Best match: {best_match}")

See the RAG notebook

For a complete RAG implementation, see the Simple RAG notebook.