samarth777/modal-skills Skills

# Code Execution Sandbox Example A complete example of a secure code execution service for LLM-generated code. ```python import modal from typing import Optional app = modal.App("code-executor") # --- Sandboxed Execution Image --- sandbox_image = ( modal.Image.debian_slim(python_version="3.1

# Data Processing Pipeline Example A complete example of a scalable data processing pipeline on Modal. ```python import modal from datetime import datetime # --- Image Definition --- image = ( modal.Image.debian_slim(python_version="3.12") .pip_install( "pandas==2.2.0", "p

# Modal GPU Reference Detailed reference for GPU acceleration on Modal. ## Available GPUs | GPU | VRAM | Max Count | Best For | |-----|------|-----------|----------| | `T4` | 16 GB | 8 | Budget inference, light training | | `L4` | 24 GB | 8 | Inference | | `A10` | 24 GB | 4 | Inference, light tra

# Image Generation Service Example A complete example of deploying a Stable Diffusion image generation service. ```python import modal import io import base64 # --- Configuration --- MODEL_ID = "stabilityai/stable-diffusion-xl-base-1.0" GPU_TYPE = "A10" # --- Image Definition --- image = ( m

# Modal Images Reference This document provides detailed reference for building container images in Modal. ## Base Images ### `debian_slim` Minimal Debian-based image, recommended for most use cases. ```python image = modal.Image.debian_slim(python_version="3.12") ``` ### `micromamba` For conda

# LLM Inference Service Example A complete example of deploying an LLM inference service on Modal using vLLM. ```python import modal # --- Configuration --- MODEL_NAME = "meta-llama/Llama-3.1-8B-Instruct" GPU_TYPE = "A100" # --- Image Definition --- image = ( modal.Image.debian_slim(python_v

Guide for building serverless Python applications on Modal - a cloud platform for running AI/ML workloads, GPU-accelerated code, web endpoints, scheduled jobs, and batch processing with minimal configuration. Use when deploying Python code to Modal's infrastructure, running GPU inference, creating web APIs, processing data at scale, or building AI applications.

# Modal Sandboxes Reference Detailed reference for running arbitrary code in isolated containers. ## Overview Sandboxes are containers you can create, interact with, and terminate at runtime. They're ideal for: - Executing LLM-generated code - Running untrusted user code - Interactive developmen

# Modal Web Endpoints Reference Detailed reference for creating web APIs and endpoints on Modal. ## Endpoint Types ### FastAPI Endpoint (Recommended) Simple function-based endpoints: ```python import modal app = modal.App("web-api") image = modal.Image.debian_slim().pip_install("fastapi[standa