swarms-cloud 0.2.6

Swarms Cloud - Pytorch

Swarms Cloud

Infrastructure for scalable, reliable, and economical Multi-Modal Model API serving and deployment. We're using terraform to orchestrate infrastructure, FastAPI to host the models. If you're into deploying models for millions of people, join our discord and help contribute.

Guides

• Available Models
• Migrate from OpenAI to Swarms in 3 lines of code
• Getting Started with SOTA Vision Language Models VLM
• Enterprise Guide to High-Performance Multi-Agent LLM Deployments

Install

pip install swarms-cloud

Architecture

user -> request -> load balanncer -> node[gpu] -> fast api -> model

Scripts

sky serve up -n [NAME] --cloud aws

Calculate Pricing

from transformers import AutoTokenizer
from swarms_cloud import calculate_pricing

# Initialize the tokenizer
tokenizer = AutoTokenizer.from_pretrained("gpt2")

# Define the example texts
texts = ["This is the first example text.", "This is the second example text."]

# Calculate pricing and retrieve the results
total_tokens, total_sentences, total_words, total_characters, total_paragraphs, cost = calculate_pricing(texts, tokenizer)

# Print the total tokens processed
print(f"Total tokens processed: {total_tokens}")

# Print the total cost
print(f"Total cost: ${cost:.5f}")

Generate an API key

from swarms_cloud.api_key_generator import generate_api_key

out = generate_api_key(prefix="sk", length=30)

print(out)

Stack

• Backend: FastAPI
• Skypilot for container management
• Stripe for payment tracking
• Postresql for database
• TensorRT for inference
• Docker for cluster management
• Kubernetes for managing and autoscaling docker containers
• Terraform

License

MIT